handbook of nanophase and nano structured materials 4
TRANSCRIPT
Volume IV: Materials Systems and Applications II
I
Content
10. NANOMECHANISM OF THE HEXAGONAL-CUBIC PHASE TRANSITION IN BORON NITRIDE UNDER HIGH PRESSURE AT HIGH TEMPERATURE ---------------------------------------------------------------------------------------------------- 1
10.1 INTRODUCTION------------------------------------------------------------------------------------------------------------------------------ 1 10.2 PROCESSING METHOD TO GET C-BN ----------------------------------------------------------------------------------------------------- 2 10.3 CHARACTERIZATION METHOD ------------------------------------------------------------------------------------------------------------ 3 10.4 PHASE TRANSITION OF BORON NITRIDE ------------------------------------------------------------------------------------------------- 4
10.4.1 Nanostructure of the Starting Material -------------------------------------------------------------------------------------------- 4 10.4.2 Phases and Nanostructures Appearing during the Hexagonal-Cubic Transition--------------------------------------------- 5
11.1 GENERAL OVERVIEW OF BATTERIES AND FUEL CELLS ------------------------------------------------------------------------------- 19 11.1.1 Introduction-------------------------------------------------------------------------------------------------------------------------- 19 11.1.2 An Overview of Batteries----------------------------------------------------------------------------------------------------------- 20 11.1.3 An Overview of Fuel Cells --------------------------------------------------------------------------------------------------------- 24 11.1.4 Importance of Nanomaterials in Batteries and Fuel Cells--------------------------------------------------------------------- 26
10.5 MECHANISM OF HEXAGONAL-CUBIC TRANSITION ----------------------------------------------------------------------------------- 27 10.5.1 Model for the Transition Mechanism--------------------------------------------------------------------------------------------- 27 10.5.2 Atomic Movement during the Conversion from w-to c-BN -------------------------------------------------------------------- 30 10.5.3 Facilitation of Synthesis of c-BN by Mechanochemical Effect---------------------------------------------------------------- 30
10.6 PROSPECT---------------------------------------------------------------------------------------------------------------------------------- 33 10.7 CONCLUSIONS----------------------------------------------------------------------------------------------------------------------------- 34
References------------------------------------------------------------------------------------------------------------------------------------ 35 13.2 SYNTHETIC STRATEGIES FOR VARIOUS NANOTUBE ARCHITECTURES -------------------------------------------------------------- 36
13.2.1 Chemical Vapor Deposition ------------------------------------------------------------------------------------------------------- 36 13.2.2 Growth of Self-oriented Multi-walled Nanotubes------------------------------------------------------------------------------- 37 13.2.3 Enable the Growth of Single-Walled Nanotubes by CVD---------------------------------------------------------------------- 38 13.2.4 Growth Mechanism of SWNT------------------------------------------------------------------------------------------------------ 40 13.2.5 Growth of Isolated Single-Walled Nanotubes on Controlled Surface Sites-------------------------------------------------- 41 13.2.6 Growth of Suspended SWNTs with Directed Orientations --------------------------------------------------------------------- 43
11. NANOMATERIALS FOR ENERGY STORAGE: BATTERIES AND FUEL CELLS---------------------------------------- 46
11.1 GENERAL OVERVIEW OF BATTERIES AND FUEL CELLS ------------------------------------------------------------------------------- 46 11.1.1 Introduction-------------------------------------------------------------------------------------------------------------------------- 46 11.1.2 An Overview of Batteries----------------------------------------------------------------------------------------------------------- 46 11.1.3 An Overview of Fuel Cells --------------------------------------------------------------------------------------------------------- 50 11.1.4 Importance of Nanomaterials in Batteries and Fuel Cells--------------------------------------------------------------------- 52
11.2 BATTERIES AND NANOMATERIALS ------------------------------------------------------------------------------------------------------ 53 11.2.1 Classifications of Advanced Batteries -------------------------------------------------------------------------------------------- 53 11.2.2 Major Components of Batteries --------------------------------------------------------------------------------------------------- 56 11.2.3 Applications of Nanomaterials in Advanced Batteries ------------------------------------------------------------------------- 58 11.2.4 Most Recent Developments -------------------------------------------------------------------------------------------------------- 63
11.3 FUEL CELLS AND NANOMATERIALS----------------------------------------------------------------------------------------------------- 63 11.3.1 Classifications of Fuel Cell Systems---------------------------------------------------------------------------------------------- 64 11.3.2 Major Components and Nanomaterials in Fuel Cells-------------------------------------------------------------------------- 66
II
11.3.3 Applications of Nanomaterials in Fuel Cells ------------------------------------------------------------------------------------ 67 11.3.4 Summary ----------------------------------------------------------------------------------------------------------------------------- 75
11.4 CONCLUSIONS ----------------------------------------------------------------------------------------------------------------------------- 75 References------------------------------------------------------------------------------------------------------------------------------------ 76
12. NANOCOMPOSITES---------------------------------------------------------------------------------------------------------------------- 85
12.1 INTRODUCTION---------------------------------------------------------------------------------------------------------------------------- 85 12.2 GENERAL FEATURES OF NANOCOMPOSITES ------------------------------------------------------------------------------------------- 90
12.2.1 Physical Sensitivity: Three Effects of Nanoparticles on Material Properties ----------------------------------------------- 90 12.2.2 Chemical Reactivity ---------------------------------------------------------------------------------------------------------------- 91 12.2.3 Promising Improvements in Nanocomposites ----------------------------------------------------------------------------------- 92 12.2.4 Origin of Nanophases and Generating Stages ---------------------------------------------------------------------------------- 93
12.3 CERAMIC-BASED NANOCOMPOSITES--------------------------------------------------------------------------------------------------- 95 12.3.1 Strength Improvement of Ceramic-Based Nanocomposites ------------------------------------------------------------------- 96 12.3.2 Toughening Effect of Nanoceramic Composites ------------------------------------------------------------------------------- 101 12.3.3 Improvements of Nanoceramic Composites on Hardness and Wear -------------------------------------------------------- 102 12.3.4 Superplasticity of Ceramic Nanocomposites ----------------------------------------------------------------------------------- 103 12.3.5 Improvement of Nanoceramic Composites on Creep-------------------------------------------------------------------------- 104 12.3.6 Ceramic-Based Nanometallic Composites ------------------------------------------------------------------------------------- 104
12.4 METALLIC-BASED NANOCOMPOSITES------------------------------------------------------------------------------------------------- 105 12.5 POLYMER-BASED NANOCOMPOSITES ------------------------------------------------------------------------------------------------- 106 12.6 SUMMARIES OF NANOCOMPOSITES ---------------------------------------------------------------------------------------------------- 108
References----------------------------------------------------------------------------------------------------------------------------------- 109
13. GROWTH AND PROPERTIES OF SINGLE-WALLED NANOTUBES------------------------------------------------------- 111
13.1 INTRODUCTION--------------------------------------------------------------------------------------------------------------------------- 111 13.2 SYNTHETIC STRATEGIES FOR VARIOUS NANOTUBE ARCHITECTURES ------------------------------------------------------------- 112
13.2.1 Chemical Vapor Deposition ------------------------------------------------------------------------------------------------------ 112 13.2.2 Growth of Self-oriented Multi-walled Nanotubes------------------------------------------------------------------------------ 113 13.2.3 Enable the Growth of Single-Walled Nanotubes by CVD--------------------------------------------------------------------- 114 13.2.4 Growth Mechanism of SWNT----------------------------------------------------------------------------------------------------- 116 13.2.5 Growth of Isolated Single-Walled Nanotubes on Controlled Surface Sites------------------------------------------------- 117 13.2.6 Growth of Suspended SWNTs with Directed Orientations -------------------------------------------------------------------- 119
13.3 PHYSICS IN ATOMICALLY WELL-DEFINED NANOWIRES ----------------------------------------------------------------------------- 121 13.3.1 Integrated Circuits of Individual Single-Walled Nanotubes ------------------------------------------------------------------ 121 13.3.2 Electron Transport Properties of Metallic Nanotubes ------------------------------------------------------------------------ 123 13.3.3 Electron Transport Properties of Semiconducting Nanotubes --------------------------------------------------------------- 125 13.3.4 Electron Transport Properties of Semiconducting Nanotubes with Small Band Gaps------------------------------------ 128
13.4 INTEGRATED NANOTUBE DEVICES----------------------------------------------------------------------------------------------------- 135 13.4.1 Nanotube Molecular Transistors with High Gains ---------------------------------------------------------------------------- 135
13.5 CONCLUSIONS---------------------------------------------------------------------------------------------------------------------------- 137 References----------------------------------------------------------------------------------------------------------------------------------- 138
14. NANOMATERIALS FROM LIGHT-ELEMENT COMPOSITES -------------------------------------------------------------- 142
14.1 INTRODUCTION--------------------------------------------------------------------------------------------------------------------------- 142
III
14.2 THEORETICAL PREDICTION ------------------------------------------------------------------------------------------------------------- 142 14.2.1 Empirical Model ------------------------------------------------------------------------------------------------------------------- 142 14.2.2 First-Principles Study ------------------------------------------------------------------------------------------------------------- 143
14.3 SYNTHESIS BY CHEMICAL VAPOR DEPOSITION (CVD) ------------------------------------------------------------------------------ 144 14.3.1 Bias-Assisted Hot Filament CVD ------------------------------------------------------------------------------------------------ 145 14.3.2 Electron Cyclotron Resonance Microwave Plasma-Assisted CVD (MPCVD) --------------------------------------------- 146
14.4 UNIFORM SIZE-CONTROLLED NANOCRYSTALLINE DIAMOND FILMS -------------------------------------------------------------- 147 14.4.1 Deposition with CN4/N2 Precursor ---------------------------------------------------------------------------------------------- 148 14.4.2 Influence of Additional H2 on Microstructure---------------------------------------------------------------------------------- 151 14.4.3 Nitrogen Incorporation ----------------------------------------------------------------------------------------------------------- 153 14.4.4 Surface Stable Growth Model ---------------------------------------------------------------------------------------------------- 153 14.4.5 Field Electron Emission and Transport Tunneling Mechanism-------------------------------------------------------------- 154
14.5 NANOCRYSTALLINE CARBON NITRIDE FILMS ---------------------------------------------------------------------------------------- 156 14.5.1 α and β Structures------------------------------------------------------------------------------------------------------------------ 156 14.5.2 Tetragonal Structure--------------------------------------------------------------------------------------------------------------- 158 14.5.3 Monoclinic Structure -------------------------------------------------------------------------------------------------------------- 158 14.5.4 Fullerene-like Structure ----------------------------------------------------------------------------------------------------------- 159 14.5.5 Carbon Nitride/Diamond/Silicon Layers --------------------------------------------------------------------------------------- 159 14.5.6 Physical and Chemical Properties----------------------------------------------------------------------------------------------- 160
14.6 NANOCRYSTALLINE SILICON CARBONITRIDE FILMS--------------------------------------------------------------------------------- 161 14.6.1 Deposition with Nitrogen and Methane----------------------------------------------------------------------------------------- 162 14.6.2 Deposition with Nitrogen, Methane and Hydrogen: Influence of Hydrogen Flow Ratio --------------------------------- 164 14.6.3 Lattice-Matched Growth Model-------------------------------------------------------------------------------------------------- 165
14.7 TURBOSTRATIC BORON CARBONITRIDE FILMS --------------------------------------------------------------------------------------- 166 14.7.1 Morphology and Composition --------------------------------------------------------------------------------------------------- 166 14.7.2 Turbostratic Structure ------------------------------------------------------------------------------------------------------------- 167 14.7.3 Raman and Photoluminescence-------------------------------------------------------------------------------------------------- 169 14.7.4 Field Electron Emission----------------------------------------------------------------------------------------------------------- 170
14.8 POLYMERIZED NITROGEN-INCORPORATED CARBON NANOBELLS ----------------------------------------------------------------- 171 14.8.1 Polymerized Nanobell Structure ------------------------------------------------------------------------------------------------- 171 14.8.2 Chemical Separation and Application ------------------------------------------------------------------------------------------ 172 14.8.3 Wall-Side Field Emission Mechanism------------------------------------------------------------------------------------------- 173
14.9 HIGHLY ORIENTED BORON CARBONITRIDE NANOFIBERS--------------------------------------------------------------------------- 175 14.9.1 Microstructure and Composition ------------------------------------------------------------------------------------------------ 175 14.9.2 Field Electron Emission----------------------------------------------------------------------------------------------------------- 176
14.10 CONCLUSIONS -------------------------------------------------------------------------------------------------------------------------- 177 References----------------------------------------------------------------------------------------------------------------------------------- 178
15，SELF ASSEMBLED ORDERED NANOSTRUCTURES ------------------------------------------------------------------------ 183
15.1 ORDERED SELF-ASSEMBLED NANOCRYSTALS --------------------------------------------------------------------------------------- 183 15.1.1 Processing of Nanocrystals for Self-Assembly --------------------------------------------------------------------------------- 185 15.1.2 Technical Aspects of Self-Assembling ------------------------------------------------------------------------------------------- 189 15.1.3 Structure of the Nanocrystal Self-Assembly ------------------------------------------------------------------------------------ 193 15.1.4 Properties of the Nanocrystal Self-Assembly----------------------------------------------------------------------------------- 198
15.2 ORDERED SELF-ASSEMBLY OF MESOPOROUS MATERIALS-------------------------------------------------------------------------- 202
IV
15.2.1 Processing -------------------------------------------------------------------------------------------------------------------------- 203 15.2.2 The Formation Mechanisms------------------------------------------------------------------------------------------------------ 204 15.2.3 Applications ------------------------------------------------------------------------------------------------------------------------ 208 15.2.4 Mesoporous Materials of Transition Metal Oxides---------------------------------------------------------------------------- 212
15.3 HIERARCHICALLY STRUCTURED NANOMATERIALS ---------------------------------------------------------------------------------- 213 15.4 SUMMARY -------------------------------------------------------------------------------------------------------------------------------- 215
References----------------------------------------------------------------------------------------------------------------------------------- 216
16, MOLECULARLY ORGANIZED NANOSTRUCTURAL MATERIALS ------------------------------------------------------ 221
16.1 INTRODUCTION--------------------------------------------------------------------------------------------------------------------------- 221 16.1.1 Nanostructural Materials in Energy Sciences---------------------------------------------------------------------------------- 221 16.1.2 Nanophase Materials in Environmental and Health Sciences --------------------------------------------------------------- 221 16.1.3 Molecularly Organized Nanostructural Materials ---------------------------------------------------------------------------- 222
16.2 MOLECULARLY DIRECTED NUCLEATION AND GROWTH, AND MATRIX MEDIATED NANOCOMPOSITES ------------------------ 222 16.2.1 Molecularly Directed Nanoscale Materials in Nature ------------------------------------------------------------------------ 222 16.2.2 Directed Nucleation and Growth of Thin Films-------------------------------------------------------------------------------- 223 16.2.3 Matrix Mediated Nanocomposites----------------------------------------------------------------------------------------------- 227
16.3 SURFACTANT DIRECTED HYBRID NANOSCALE MATERIALS ------------------------------------------------------------------------ 231 16.3.1 Ordered Nanoporous Materials-------------------------------------------------------------------------------------------------- 232 16.3.2 Hybrid Nanoscale Materials ----------------------------------------------------------------------------------------------------- 237
16.4 SUMMARY AND PROSPECTS------------------------------------------------------------------------------------------------------------- 243 References----------------------------------------------------------------------------------------------------------------------------------- 243
17, NANOSTRUCTURED BIO-INSPIRED MATERIALS ---------------------------------------------------------------------------- 246
17.1 INTRODUCTION--------------------------------------------------------------------------------------------------------------------------- 246 17.2 CASE STUDY I: TEETH------------------------------------------------------------------------------------------------------------------- 248
17.2.1 Control over Mineralization at Nanometer Scale------------------------------------------------------------------------------ 249 17.2.2 Hierarchical Structure in Biological Materials -------------------------------------------------------------------------------- 252
17.3 CASE STUDY II: MESOSCOPIC SILICA FILMS------------------------------------------------------------------------------------------ 254 17.3.1 Hierarchical Film Structure ------------------------------------------------------------------------------------------------------ 256 17.3.2 Towards Control of the Properties ----------------------------------------------------------------------------------------------- 262
17.4 CONCLUSION ----------------------------------------------------------------------------------------------------------------------------- 262 References----------------------------------------------------------------------------------------------------------------------------------- 263
18, NANOPHASE METAL OXIDE MATERIALS FOR ELECTROCHROMIC DISPLAYS---------------------------------- 266
18.1 INTRODUCTION--------------------------------------------------------------------------------------------------------------------------- 266 18.2 BASIC CONCEPTS IN ELECTROCHROMISM--------------------------------------------------------------------------------------------- 267
18.2.1 Electrochromic Display Device -------------------------------------------------------------------------------------------------- 267 18.2.2 Electrochromic Materials--------------------------------------------------------------------------------------------------------- 268 18.2.3 Perceived Color and Contrast Ratio -------------------------------------------------------------------------------------------- 270 18.2.4 Coloration Efficiency and Response Time -------------------------------------------------------------------------------------- 270 18.2.5 Write-Erase Efficiency and Cycle Life ------------------------------------------------------------------------------------------ 270
18.3 NANOPHASE METAL OXIDE ELECTROCHROMIC MATERIALS ----------------------------------------------------------------------- 271 18.3.1 Synthesis of Supported ATO Nanocrystallites ---------------------------------------------------------------------------------- 272 18.3.2 Characterization of Supported ATO Nanocrystallites------------------------------------------------------------------------- 273
V
18.4 CONSTRUCTION OF PRINTED, FLEXIBLE DISPLAYS USING INTERDIGITATED ELECTRODES-------------------------------------- 275 18.4.1 Design Strategy -------------------------------------------------------------------------------------------------------------------- 276 18.4.2 Materials Selection ---------------------------------------------------------------------------------------------------------------- 278 18.4.3 Display Examples------------------------------------------------------------------------------------------------------------------ 279
18.5 CONTRAST OF PRINTED ELECTROCHROMIC DISPLAYS USING ATO NANOPHASE MATERIALS ---------------------------------- 281 18.5.1 Effect of Antimony Doping on Contrast Ratio---------------------------------------------------------------------------------- 282 18.5.2 Effect of Annealing Temperature on Contrast Ratio--------------------------------------------------------------------------- 288 18.5.3 Other Factors That Affect the Contrast Ratio ---------------------------------------------------------------------------------- 292
18.6 SUMMARY -------------------------------------------------------------------------------------------------------------------------------- 297 References----------------------------------------------------------------------------------------------------------------------------------- 297
19, ENGINEERED MICROSTRUCTURES FOR NONLINEAR OPTICS -------------------------------------------------------- 300
19.1 INTRODUCTION--------------------------------------------------------------------------------------------------------------------------- 300 19.2 PREPARATION OF DSLS ----------------------------------------------------------------------------------------------------------------- 300
19.2.1 Preparation of DSLs by Modulation of Ferroelectric Domains ------------------------------------------------------------- 301 19.2.2 Preparation of DSL by Using Photorefractive Effect-------------------------------------------------------------------------- 303
19.3 OUTLINE OF THE NONLINEAR OPTICS ------------------------------------------------------------------------------------------------- 304 19.4 WAVE VECTOR CONSERVATION--------------------------------------------------------------------------------------------------------- 305 19.5 NONLINEAR OPTICAL FREQUENCY CONVERSION IN 1-D PERIODIC DSLS -------------------------------------------------------- 308 19.6 NONLINEAR OPTICAL FREQUENCY CONVERSION IN 1-D QPDSLS ---------------------------------------------------------------- 310
19.6.1 The Construction of QPDSL------------------------------------------------------------------------------------------------------ 310 19.6.2 Theoretical Treatment of the Nonlinear Optical Processes in QPDSLs ---------------------------------------------------- 311 19.6.3 The Effective Nonlinear Optical Coefficients ---------------------------------------------------------------------------------- 315 19.6.4 QPM Multiwavelength SHG (Zhu, et al., 1990, 1997b; Qin, et al., 1999) ------------------------------------------------- 316 19.6.5 Direct THG (Feng, et al., 1990; Zhu, et al., 1997a, 1998; Qin, et al., 1999) ---------------------------------------------- 317
19.7 OPTICAL BISTABILITY IN A 2-D DSL -------------------------------------------------------------------------------------------------- 317 19.7.1 Bloch Wave Approach (Xu and Ming, 1993b, 1994; Wang, et al., 1996c) -------------------------------------------------- 318 19.7.2 Four-Path Switch: Linear Case (Feng and Ming, 1989) --------------------------------------------------------------------- 320 19.7.3 A New Type of Optical Bistability Mechanism: Nonlinear Case with One Incident Wave (Xu and Ming, 1993a, b, 1994; Wang, et al., 1996a, b, c, d, e, 1997)----------------------------------------------------------------------------------------------------- 321 19.7.4 A New Type of Optical Bistability Mechanism: Nonlinear Case with Two Incident Waves (Chen, et al., 1995, 1996b)------------------------------------------------------------------------------------------------------------------------------------------------ 325
19.8 OUTLOOK --------------------------------------------------------------------------------------------------------------------------------- 326 References----------------------------------------------------------------------------------------------------------------------------------- 327
APPENDIX -------------------------------------------------------------------------------------------------------------------------------------- 332
1
10. Nanomechanism of the Hexagonal-Cubic Phase Transition in Boron
Nitride under High Pressure at High Temperature
10.1 Introduction
Boron nitride (BN) is an important industrial material; e.g., cubic-type BN (c-BN) is widely used. Because it is the second hardest material known to man (the hardest being diamond), it is mainly used for grinding and cutting industrial ferrous materials instead of diamond, since it does not react with iron. Another prominent feature is its high thermal conductivity, 6–9 W/(cm • deg), which is the second largest after diamond (Corrigan, 1979). Hexagonal-type BN (h-BN) is chemically stable at high temperature and therefore used as crucibles for single crystal growth. Since it is transparent without any color, it is used as a raw powder for a cosmetic material.
c-BN is synthesized from h-BN under high pressure at high temperature (Bundy and Wentorf, 1963; Wentorf, 1957). It is then rather expensive. The exact mechanism of the hexagonal-cubic transition is not clear, since nanophases are complicatedly involved during the transition. Its elucidation is highly expected from the standpoint of nanostructure analysis.
There are four main phases in boron nitride. h-BN and rhombohedral-type BN (r-BN) are formed under ambient pressure (Pease, 1952; Ishii, et al., 1981). Boron and nitrogen atoms form hexagonal rings, which are linked by sp2 bonding and extend two-dimensionally in sheets. Neighboring sp2 sheets are weakly bound by van der Waals force. Their interplanar spacing is 0.3328 nm and 0.334 nm for h-BN and r-BN, respectively. Their stacking sequences are conventionally represented by ab′ and abc, respectively (Fig. 10.1(a) and (b)).
Figure 10.1 Projections of the crystal structure of h-BN (a), r-NB (b), w-BN (c) and (d); • and means B and N atoms, respectively. (from S. Horiuchi)
2
Wurtzite-type BN (w-BN) is prepared from h-BN under high pressure at room temperature (Bundy and Wentorf, 1963; Wakatsuki, et al., 1986). The interplanar spacing becomes small (d0002 = 0.2211 nm) since B and N atoms are linked by the sp3 bond and the sheets become puckered. Their stacking sequence is represented by AB′ (Fig. 10.1(c)).
For the structure of c-BN, which is another high-pressure phase, the stacking sequence is ABC (d111 = 0.2087 nm) (Fig. 10.1(d)). By using catalysts the pressure and/or temperature for the synthesis is considerably decreased (Endo, et al., 1979; Sei, et al., 1993), but the products suffer from the contamination due to the catalysts.
In the present chapter, structural evolution during the hexagonal-cubic phase transition under high pressure at high temperature is examined in detail using high resolution transmission electron microscopy (HRTEM) and electron energy loss spectroscopy (EELS). It is found that the formation of w-BN plays an important role on the mechanism of the transition. In order to verify the validity of the proposed model, the initial h-BN powders are ball-milled so that w-BN includes many lattice defects. The hexagonal-cubic transition is then prominently facilitated as compared to the non-milled case due to the mechanochemical effect.
10.2 Processing Method to Get c-BN
3
c-BN is usually obtained by processing h-BN under high pressure at high temperature. Besides, it can be also synthesized at low pressure in a similar manner to diamond. Its formation under low pressure has been widely examined and the mechanism is under discussion at present.
In the present study the starting material was h-BN hot-pressed in a cylinder shape, which was commercially obtained from Denkikagaku Co. (Type N-1). It was wrapped in Zr foil and pressed under high pressure (6.5–7.7 GPa) at high temperature (1700–2150°C) for 20–30 min, using a belt-type high-pressure machine (Akaishi, et al., 1994), as shown in Table 10.1 (He, et al., 1998). The press direction was parallel to the cylinder axis.
Table 10.1 Synthetic condition and products*
Sample Pressure (GPa) Temperature (°C) Time (min) Phases
1 — — — h
2 6.5 1730 30 h+c(+m)
3 7.7 1700 30 c+m+h+w
4 7.7 1800 20 c+m+h
5 7.7 2000 20 c+m
6 7.7 2150 20 c
*h: hexagonal-type BN, c: cubic-type BN, w: wurtzite-type BN, m: monoclinic-type BN, (+m): the signal of m-BN is obtained only in HRTEM images and ED patterns. Sample 1 is a starting, hot-pressed material.
10.3 Characterization Method
The recovered specimens were examined by X-ray diffraction (XRD), using Cu-Kα radiation, at room temperature. In order to clarify the effects of the press direction upon microstructure, XRD analyses were carried out in both the longitudinal and transverse cross sections.
Transmission election microscopy (TEM) specimens were also prepared for these two cross sections by using a conventional method; thin plates were cut out, polished, dimpled and ion-milled. They were observed by a high-voltage electron microscope (model H-1500) at an accelerating voltage of 1000 kV with
4
a resolving power of about 0.12 nm. The electron energy-loss spectroscopy (EELS) analysis was carried out using an analytical electron microscope (model HF-2000) with a field emission gun at 200 kV.
A computer simulation of high-resolution transmission electron microscory (HRTEM) images was carried out by using the microscope parameters: the spherical aberration coefficient of 2.1 mm, the defocus spread of 10 nm, and the beam convergence of 0.3 mrad. The software was developed by us (Horiuchi, 1994).
10.4 Phase Transition of Boron Nitride
10.4.1 Nanostructure of the Starting Material
1. h-BN
Figure 10.2 is a typical structure of hot-pressed h-BN. It is noted that the grains have plate-like form. Their thickness fluctuates between 1 and 6 µm. They are closely packed in most regions. Small cracks occur in some of the thick plates, as indicated by arrowheads in Fig. 10.2.
Figure 10.2 TEM image of the starting material of hot-pressed h-BN (sample 1). Plate-like crystals are closely packed. Arrowheads indicate the occurrence of small cracks. (from S. Horiuchi)
5
Figure 10.3 is an HRTEM image with the corresponding electron diffraction (ED) pattern, taken with an incident electron beam parallel to of a h-BN plate. It is found that plates are actually composed of "sub-plates". Most of the subplates are only several nm in thickness. An example of a boundary between subplates is marked by an arrowhead. The formation of subplates must be a result of the weak chemical bond between the sp2 sheets. The small cracks observed in Fig. 10.2 must have been formed at the boundary between subplates when the specimen was hot-pressed. Figure 10.3 also demonstrates that the dangling bonds usually arise at the edge of sp2 sheets, which is shown by an arrow. The situation is schematically depicted by an inset.
2. t-BN
Figure 10.3 HRTEM image of h-BN with the incident electron beam along . The corresponding ED pattern is attached. An inset schematically shows the formation of dangling bonds at the edge of the crystal, marked by an arrow. An arrowhead indicates the boundary between subplates. (from S. Horiuchi)
In some areas a semi-spiral structure is found at the edge of sp2 sheets (Fig. 10.4). It is composed of 5 curved sp2 sheets and one dangling bond at the center. From the morphology and the interplanar distance (d = 0.33–0.35 nm), it is identified to be a turbostratic-type BN (t-BN). The semi-spiral structure seems smaller in size compared to that found under high pressure (7.7 GPa) at high temperature (1800°C) (Horiuchi, et al., 1995).
10.4.2 Phases and Nanostructures Appearing during the Hexagonal-Cubic Transition
6
10.4.2.1 XRD
The results of XRD analysis are shown in Figs. 10.5 and 10.6. Figures 10.5(a)—(f) are from samples 1–6 indicated in Table 10.1. They were obtained from the transverse cross section. Figure 10.6 shows the XRD charts obtained from the longitudinal cross section of sample 1–3. The identified phases are listed in Table 10.1.
Figure 10.4 A semi-spiral structure of t-BN formed locally at the edge of a starting h-BN plate. (from S. Horiuchi)
Figure 10.5 XRD charts obtained from the transverse cross sections of samples 1–6. h, h-BN, c, c-BN, m, m-BN, w, w-BN. (a) is taken from sample 1 (cf. Table 10.1); (b) from sample 2; (c) from sample 3; (d) from sample 4; (e) from sample 5; and (f) from sample 6. (from S. Horiuchi)
7
8
Figure 10.6 XRD charts obtained from the longitudinal cross sections of samples 1–3. h, h-BN, c, c-BN. (a) is taken from sample 1, (b) from sample 2 and (c) from sample 3. (from S. Horiuchi)
In sample 1, the intensity ratio of h ( ) to h (0004) is different between Figs. 10.5(a) and 10.6(a). This means that there is a texture in the hot-pressed h-BN with a preferred distribution of (0002) normal to the hot-press direction. In sample 2, most of h-BN still remains unchanged (Figs. 10.5(b) and 10.6(b)). However, a very small amount of c-BN is formed, as shown by an open arrow. It is noted that h (0, 0, 0, 2n) peaks are slightly higher in Fig. 10.5(b) than in Fig. 10.5(a), while lower in Fig. 10.6(b) than in Fig. 10.6(a). This means that the texture becomes more prominent in this sample. How the texture is formed will be discussed later.
Under 7.7 GPa at 1700°C (sample 3) a large amount of c-BN appears (Figs. 10.5(c) and 10.6(c)). The residual h (0002) peak, marked by a dark arrow, becomes broad. This is due to the formation of m-BN (monoclinic-type BN, see below) (Horiuchi, et al., 1996). It is interesting to note that the initially intense h (0, 0, 0, 2n) peaks strongly decrease in intensity, while the initially weak h ( ) and h ( ) peaks are still observable in Fig. 10.5(c). We can interpret this by the preferential formation of c-BN in the h-BN plates, whose (0002) plane is normal to the press direction. Moreover, we note the appearance of w-BN in Fig. 10.5(c).
9
At 1800°C (sample 4) the m (002) peak becomes slightly sharp (Fig. 10.5(d)). The peak becomes very weak at 2000°C (sample 5, Fig. 10.5(e)) and disappears at 2 150°C (sample 6, Fig. 10.5(f)). These results clearly show that m-BN is an intermediate phase during the phase transition from h-BN to c-BN.
10.4.2.2 HRTEM
1. h-BN
Figure 10.7 shows TEM images of the h-BN in sample 2. The plates are prominently folded locally (Fig. 10.7(a)), while some others keep the plate-form with slight bending (Fig. 10.7(b)). Arrowheads indicate the direction of pressure. As stated above in relation to Fig. 10.6(a) and (b), a strong texture has been developed in sample 2. Some h-BN plates, whose plate plane was parallel to the press direction initially, are folded under high pressure at high temperature. Frequent folding causes a zigzag shape. On the other hand, the plates, whose plane was almost normal to the press direction initially, are slightly bent. Such structural evolution is schematically shown in Fig. 10.8. During folding and bending, most of the h-BN plates tend to be normal to the press direction, while some still remain nearly parallel to the press direction.
Figure 10.7 Characteristic structures formed in h-BN plates of sample 2. Plates are frequently folded in (a) while bending in (b), Arrowheads show the direction of pressure. (from S. Horiuchi)
Figure 10.8 Geometry of h-BN plates before pressing (a) and after pressing (b). (from S. Horiuchi)
10
It has been found by HRTEM observation that the folding is achieved by a mechanical twinning, as will be shown below.
2. m-BN
An X-ray diffraction peak of so-called "compressed h-BN", which shows a smaller interplanar distance (d = 0.31 nm) than that for the normal h-BN mentioned above, was detected in the course of the hexagonal-cubic transition (Corrigan and Bundy, 1975; Corrigan, 1979). An example of the XRD peak is shown in Fig. 10.5(d).
Recently its structure has been analyzed and clarified by HRTEM (Horiuchi, et al., 1996). The contraction of the interplanar spacing occurs when the sp2 sheets are sheared to each other so that the symmetry changes from hexagonal to monoclinic. The lattice parameters of the monoclinic-type BN (m-BN) are α = 0.433 nm, b = 0.250 nm, c = 0.31–0.33 nm, β = 92°C—95°C.
The folding of the initial h-BN plates mentioned above causes the shear. Figure 10.9 is an HRTEM image of sample 2, showing an enlarged region near the folding plane, taken with the incident electron beam parallel to (He, et al., 1998). Near the plane (0002) sheets are sheared with respect to each other and the interplanar distance is slightly decreased. As a result, the symmetry changes from hexagonal to monoclinic. However, in the area slightly far from the plane, i. e., at the rightmost area of Fig. 10.9, h-BN keeps the original symmetry. A computer-simulated image for the h-BN, inserted in Fig. 10.9, fits well to the real image. The optical diffraction patterns taken from small areas of 6.5 nm × 6.5 nm in size proves that the shear of the 0002 atom plane varies continuously in this region, giving the value of β from 92°C (near the folding plane) to 90°C (slightly far from the plane). The inset is the ED pattern taken from a region including the whole area of Fig. 10.9. It is clear that the major spots are from m-BN, and weak spots indicated by arrowheads in the inset are from the conventional h-BN. This means that h-BN still remains in this sample, being in agreement with the result by XRD. We may conclude that the formation of m-BN
11
has been initiated near the folding plane. Since the volume of the m-BN is still very small, no clear XRD peaks of this phase have been obtained in Figs. 10.5(b) and 10.6(b).
Figure 10.9 HRTEM image of sample 2, showing the details of the folding of h-BN plate in sample 2, with a corresponding ED pattern. The electron beam is incident parallel to . m-BN appears near the folding plane, while h-BN is seen in the area at the right-hand side, in which a computer-simulated image is inserted. The folding is due to the mechanical twinning, whose atomic arrangement is schematically depicted. (from S. Horiuchi)
It can be seen also in the ED pattern of Fig. 10.9 that the two areas adjoining at the folding plane are in such an orientation relationship as the rotation of 57°C—71°C about the sheet normal . We may then say that the folding was done by mechanical twinning. Another inset in Fig. 10.9 shows a schematic structure model of the twinning, in which the rotation angle is measured to be 66°C (He, et al., 1997).
In samples 3 and 4, the shear of sp2 sheets became prominent and prevailed in larger areas. The value of β approached what was previously reported (β is 92°—95°), with the increase of the preparation temperature under 7.7 GPa. A typical HRTEM image of m-BN is shown in Fig. 10.10(a), which has been taken from sample 4, with the corresponding ED pattern. The lattice relationship between h- and m-BN is schematically shown in Fig. 10.10(b) (Horiuchi, et al., 1997).
Figure 10.10 (a) HRTEM image of m-BN in sample 4, together with the corresponding ED pattern. (b) Lattice relationship between h- and m-BN. (from S. Horiuchi)
12
Similar effect on the shrinkage of (0002) spacing can be expected from lattice bending. A clear example for this is noted at the nucleation of c-BN, since there is a distinct difference in the compressibility and in the thermal expansion between adjoining h- and c-BN phases, as is shown below (Fig. 10.12).
Figure 10.11 HRTEM image of sample 4, showing a nucleation of c-BN, marked by an asterisk, in a plate-like crystal. Plates of h-BN are sharply folded at the sites of arrows. An ED pattern taken from the corresponding area is inserted. (from S. Horiuchi)
13
Figure 10.12 HRTEM image of the boundary area between the c-BN grain and the surrounding h-BN matrix in Fig. 10.11, in which the lattice fringes are prominently bent. The arrangement of bright spots in areas of downward arrows shows the monoclinic lattice distortion. An upward arrow in the c-BN grain shows the appearance of w-BN. (from S. Horiuchi)
14
In the HRTEM observation of m-BN it was noted that this phase is stable for the long-time irradiation of the highly energetic electron beam, compared to h-BN. This means that the chemical bonding in m-BN is substantially stronger than that in h-BN.
3. c-BN
Figure 10.11 is a typical morphology of sample 4, showing a nucleation stage of c-BN. Some h-BN plates are folded. A small grain of c-BN is found and marked by an asterisk in a h-BN plate. An inserted ED pattern corresponds to the image of Fig. 10.11. The diffraction spots indicate the formation of c-BN. One of them is marked by an arrow. The innermost diffraction ring, marked by an arrowhead, is from h-BN but it does not form a perfect circle. Referring to the ED spots of c-BN, the lattice spacing between sp2 sheets is measured from 0.34 to 0.31 nm. This means that m-BN is formed.
Figure 10.12 is an HRTEM image for the boundary area of the c-BN grain in Fig. 10.11. The electron beam is incident almost parallel to of c-BN. Nanoscale twins of c-BN with (111) boundaries are seen in the grain. It should be noted that the w-BN is partially found in the grain as marked by an upward arrow. From the ED pattern and HRTEM image the orientation relationship among c-BN, w-BN and the surrounding h-BN matrix is known as follows;
In the surrounding matrix the lattice planes of the sp2 sheet are prominently bending. In some areas bright spots are resolved. From their arrangement it is noted that the sheets are slightly sheared along
15
the lattice plane; i.e., the array of spots is not rectangular but slightly distorted in the area marked by downward arrows. The distortion angle is different for different areas but lies in the range of 5°—9°. This shows the formation of m-BN.
Under the high pressure of 7.7 GPa, h- and c-BN are compressed by about 89% and 98%, respectively (Lynch and Drickamer, 1996; Knittle, et al., 1989). However, for the present case, in which the two phases coexist, the volume cannot be completely retrieved when the high pressure is removed because of mutual restraint at the boundaries. This restraint must have caused the lattice bending observed above. Also, the thermal expansion coefficient is significantly different between h- and c-BN (Pease, 1952; Slack and Bartram, 1974). This will induce similar effect on the difference in the compressibility, and is another cause of the mutual lattice restraint. The interplanar spacing of the bending lattice in Fig. 10.12 is measured as 0.31–0.34 nm of d. Here we have used the measured value (d111 = 0.21 nm) of c-BN as a reference.
When temperature increases, the number of c-BN grains increases but the growth of the grains is not prominent for temperatures lower than 2000°C We have found that a large number of thin twins (nanoscale twins) have formed. Figure 10.13 shows an HRTEM image and the corresponding ED pattern taken from sample 5. In this case the thin film for TEM observation was obtained by crushing the retrieved specimen. It is clear that each twin slice contains only several atomic planes. A computer-simulated image for c-BN fits well to the real image. We may then say that a bright spot appears at the center of a pair of B and N columns, which are adjacent in the projection plane. In the calculation the defocus amount is taken to be 80 nm underfocus. The diffraction streaks along 111 are just due to the shape effect of the nanotwins.
Figure 10.13 HRTEM image of a c-BN grain in sample 5, showing the formation of nanoscale twins. The corresponding ED pattern and a computer-simulated image of c-BN are inserted. (from S. Horiuchi)
16
Figure 10.14(a) demonstrates a typical structure of the c-BN grain prepared at the highest temperature (sample 6). Grains have grown to µm size and the thin twin plates have disappeared. Instead, the twinning along the second set of 111 planes appeared. Figure 10.14(b) is the corresponding ED pattern.
4. w-BN
Figure 10.14 (a) TEM image of a c-BN grain in sample 6, including secondary twins. (b) A corresponding ED pattern. (from S. Horiuchi)
17
Small domains of w-BN are found in some c-BN grains of samples 3 and 4, although very locally in the latter sample. This is always the case for small grains of c-BN. Figure 10.15 is another example of HRTEM image taken from sample 4, showing the coexistence of c-BN and w-BN. Typical, narrow areas of c-BN are marked by "c", and those of w-BN by "w". They face each other at the sites marked by circles. A corresponding ED pattern is inserted and verifies the orientation relationship of Eq. (10.1).
10.4.2.3 EELS
Figure 10.16 shows the EELS profiles in core-loss regions for h-BN (a), c-BN (b) and m-BN (c). A fine electron beam of about 1 nm in diameter was used.
Figure 10.15 HRTEM image of sample 4, showing the coexistence of c- and w-BN in a grain. Typical, narrow areas of c-BN are marked by "c", and those of w-BN by "w". They face each other at the sites marked by circles. A corresponding ED pattern is inserted. Leftward arrowheads show diffraction spots from c-BN, rightward ones those from w-BN. (from S. Horiuchi)
18
Figure 10.16 Core-loss EELS profiles from h-BN (a), c-BN (b), and m-BN (c). (from S. Horiuchi)
19
The condition of measurement was kept constant for all of them. In (a) there are two main peaks at 196 and 202.5 eV (marked with π* and σ*, respectively). They are considered to correspond to the transition to π* and σ* bands, respectively. In (b) there is only one main peak, σ* peak. For m-BN (c) there are also two peaks, π* and σ* peaks, as is the case of (a). However, the π* peak is stronger than the σ* peak. This difference implies that the band structure of sp2 sheets is different and the density of states for the π* bond becomes higher in m-BN than in h-BN.
11.1 General Overview of Batteries and Fuel Cells
11.1.1 Introduction
Batteries and fuel cells are important power sources today (Berger, 1997; Georgano, 1996; Ondrey, et al., 1999) and will continue to be used in a wide variety of consumer, industrial and military applications in the 21st century. As the technologies of electronics industry advance, batteries are becoming a critical component (Ruetschi, et al., 1995; Seitz, 1991) for portable electronic devices, lighting, photographic equipment, watches, calculators, memory backup and a wide variety of other applications, giving freedom to utility power. Batteries have many advantages (Linden, 1984) over other power sources; they are usually self-contained, efficient, convenient, reliable and can be easily configured to user requirements.
20
Beyond batteries, fuel cells are highly efficient and less pollutive power-generating systems that produce DC electricity through the combination of fuel and oxidant in an electrochemical reaction (Apple by and Foulkes, 1989). As such, these are becoming more and more important. Traditionally, the electrical supply homes is generated in coal- or gas-fired power stations. Chemical energy from coal is burnt to produce heat energy which changes water into steam. Steam is then used to turn huge turbines and several rolls of wires are connected to these spinning turbines. As these move through a strong magnetic field usually created by arrays of powerful magnets, an electric current is induced to flow through the wires. The electric supply is then connected to the power sockets by running cables throughout a large area. This has not been an efficient way of harvesting energy, because some available energy in coal is lost at every stage of conversion. In fact, coal-fired stations are only around 30% efficient in the conversion of chemical energy from fuel to electrical energy. There is a further loss of energy through transportation of the electrical supply by high voltage cables. This huge consumption of fuel is responsible for the rapid depletion of our nonrenewable energy sources. On top of this, inefficient combustion of fuels usually creates disruptive impacts on the environment including pollution and strong possibility of global climate change. At this end, fuel cells are able to promote energy diversity and provide a transition to renewable energy sources. The most abundant element on the Earth, hydrogen (Pohl, 1995) can be directly used in the functioning of fuel cells. Alternative fuels containing hydrogen, including natural gas, methanol, ethanol and even diesel or gasoline fuel can be utilized. Because fuel cells convert chemical energy directly into electrical energy without the intermediate combustion processes, these are not limited by the Carnot efficiency of thermal engines and are usually 60% efficient, equivalent to about 2–3 times more efficient than the combustion processes. As a result, these have lower emission levels, producing less CO2 associated with more traditional means of power generation.
11.1.2 An Overview of Batteries
The first work on batteries was done by Volta (Applely and Foulkes, 1989) around 1800. Ritter (Vinal, 1950) subsequently constructed what was perhaps the first battery. In 1859, Planté began the foremost studies (Vinal, 1950) which later led to the development of the first practical rechargeable (secondary) battery, the lead-acid battery. Since this early work, a variety of new battery systems have been discovered and developed (McCroy, 1977). The introduction of the nickel metal hydride (NiMH) in the late eighties and the lithium ion (Li-ion) in the early nineties brought more energy to a given cell compared to the earlier generation of batteries including the nickel cadmium (NiCd) batteries. Figure 11.1 compares the gravimetric and volumetric energy densities of rechargeable lithium batteries with those of other systems.
Figure 11.1 The gravimetric and volumetric energy densities of rechargeable lithium batteries as compared with those of other systems.
21
In 1990, an initial prototype Li-ion (AA size) cell made of carbon/lithium cobalt oxide (C/LiCoO2) was introduced to the battery world by Sony Energytec. Subsequently, a joint venture company A & T Battery, was formed by Asahi Chemical Industry and Toshiba Battery to manufacture C/LiCoO2 cells. Carbon/lithium manganese oxide (C/LiMn2 O4) technology was later developed in 1992 by Bell Communications Research. Basically, it was clear that lithium ion rechargeable had became the most promising battery technology, with most manufacturers of small rechargeable batteries in Japan, the United States and Europe engaging in Li-ion research by mid-1994.
11.1.2.1 Operation Principle of Batteries
The operation (Linden, 1984) of a battery cell during charge and discharge is shown schematically in Fig. 11.2. During discharge, electrons will flow from anode through an external load once the cell is connected to it. At the anode, the oxidation takes place, while at the cathode a reduction reaction occurs. The electrical circuit is completed in the electrolyte by the flow of negativeions (anions) and positive ions (cations) to the anode and cathode, respectively. Thus, the electronic current delivered by the cell is matched by the ionic current within the cell. Any leakage of electrons from the anode to the cathode within the cell reduces the current generated by the battery. When a piece of zinc metal is assumed as the anode material and chlorine (Cl2) is the cathode material, the discharge reactions (shown in Table 11.1) take place.
Figure 11.2 The operation of a battery cell during (a) charge and (b) discharge.
22
Table 11.1 Reaction mechanisms for a Zn/Cl2 cell
Zn/Cl2 Cell
Anode (negative electrode) Zn → Zn2+ + 2e-
Cathode (positive electrode) Cl2 + 2e- → Cl-
Overall Zn + Cl2 → Zn2+ + 2Cl- (ZnCl2)
Batteries are the lifeblood of many low-power portable devices which face an increasing demand. Essentially, they can be classified into two broad categories. Depending on the batteries' capability of being electrically rechargeable, they can be identified as primary (non-rechargeable) or secondary (rechargeable) cells. Primary cells are not capable of being easily recharged and are suitable for one-time use only. However, they are usually light, inexpensive, of high energy density and possess good shelf-life. Secondary batteries can be electrically recharged to their original condition after discharge by passing current through them in the opposite direction to that of the discharge current. They are good storage devices for electrical energy and are often charged by a primary energy source, only to deliver energy to the load upon demand. In addition to their ability to be recharged, such batteries are characterized by high energy density, high discharge rates, flat discharge curves and good low-temperature performance. The secondary batteries are used essentially to lower cost since they can be used multiple times. They are also more environmentally friendly as there is less concern for battery disposal compared to primary batteries.
11.1.2.2 Research of High Energy Storage Batteries
23
In the early world market for small rechargeable batteries, where it was basically dominated by two rechargeable battery technologies, nickel cadmium (NiCd) batteries held more than 80% of world markets while sealed lead-acid (SLA) batteries held about 13% (Seitz and Shimosato, 1994). In recent years, three advanced rechargeable battery technologies appeared in the market as well, namely the nickel metal hydride (NiMH), rechargeable alkaline manganese (RAM), and lithium ion (Li Ion). Presently, the huge spate of research into rechargeable lithium batteries has led to the emergence of lithium-solid positive-electrode prototype cells ever since the first Li/MoS2 production cell. Shown in Fig. 11.3 are the market share for the penetration of rechargeable cells in Japan, United States and Western Europe to the year 2000.
Figure 11.3 The market scenarios for the penetration of rechargeable cells in Japan, the United States and western part of Europe by the year 2000. * Li+ = lithium Ion, NiCd = nickel Cadmium, NiMH = nickel metal hydride, RAM = rechargeable alkaline manganese, SLA=sealed lead acid.
24
11.1.3 An Overview of Fuel Cells
The first fuel cell (O'Sullivan, 1972) was built up by Sir William Grove, a Welsh judge and scientist, in 1839. However, the use of fuel cells had been confined to laboratories until recent decades. It was only in the 1960s that fuel cells were engaged to provide power on board for the Gemini and Apollo space missions (Kordesch and Simader, 1996) while providing the astronauts with clean drinking water.
11.1.3.1 Operation Principles of Fuel Cells
Fuel cells have beneficial operating characteristics unmatched by any other technology. As in batteries, silent reactions produce an electric current. Unlike batteries, however, fuel cells can be continuously recharged. In fuel cells run on hydrogen, oxygen from the air reacts with the input hydrogen in such a way that a voltage is generated between the two electrodes; the reactions occur via a chemical mediator known as the electrolyte. The basic design of a fuel cell is illustrated in Fig. 11.4. Two catalyzed carbon electrodes are immersed in an electrolyte while the fuel, hydrogen, is supplied from one surface of the electrode. When the electrodes are electrically connected through an external load, the following events occur:
1. Dissociation of hydrogen occurs on the catalytic surface of the anode, forming hydrogen ions and electrons. 2. Migration of the hydrogen ions occurs through the electrolyte to the catalytic surface of the oxygen electrode
(cathode). 3. Movement of the electrons through the external circuit to the catalytic surface of the oxygen electrode. 4. The formation of water due to combination of the hydrogen ions, oxygen and electrons on the cathode's catalytic
surface.
Figure 11.4 Schematic operation principle of a fuel cell with a solid polymer electrolyte.
25
For different types of fuel cells, the ionic species carrying charge from one electrode to another varies and they are generally distinguished by the type of electrolytes used. The fuel cell depicted in Fig. 11.4 has an acidic electrolyte (such as phosphoric acid or a proton-exchange membrane), relying on H+ ions to carry charge from the anode to the cathode. In various other fuel cells, the charge can move from cathode to anode via O2- ions (solid oxide ceramic fuel cells), OH- ions (alkaline fuel cells) or CO2-
3 ions (molten-carbonate fuel cells). For example, the reaction mechanisms for a hydrogen fuel cell in acidic and alkaline electrolytes are shown in Table 11.2.
Table 11.2 Reaction mechanism for fuel cells
Acidic electrolyte Alkaline electrolyte
Anode H2 → 2H+ + 2e- H2 + 2OH- → 2H2O + 2e-
Cathode O2 + 4H+ + 4e- → 2H2O O2 + 4e- + 2H2O → 2OH-
Overall 2H2 + O2 → 2H2O 2H2 + O2 → 2H2O
11.1.3.2 Research of Fuel Cells
Fuel cells were first considered for applications where noxious emissions or noise would be objectable and where water was unavailable. With the additional advantage of on-site operation, rather than remote fuel cells can also be utilized for portability. Commercialization has been a leading theme for fuel cell technology in recent years. The major use of fuel cells as potential replacements for internal combustion
26
engines in transportation (Lemons, 1990; Prater, 1990; Srinivasan, et al., 1988) is being actively promoted. Early in the next century, fuel cells may start to appear as alternatives to internal combustion engines in automobiles, buses and other vehicles. In a pursuit for "clean" power, several vehicle demonstration projects have already been employed around the world, including those of H Power of Belleville, New Jersey; Ballard Power Systems of Vancouver, British Columbia, and Daimler-Benz of Germany.
In the first major practical use of fuel cells, the Gemini space flight missions, the solid polymer electrolyte fuel cell system (Kordesch and Simader, 1996) of the General Electric Company was used. In 1987, Ballard made a subsequent breakthrough in fuel cell performance through a new sulfonated fluorocarbon polymer membrane from Dow Chemical, similar to the earlier Nafion membrane (Gavach and Pourcelly, 1992) developed by DuPont, but capable of passing four times the current at the same operating voltages. Besides huge investments into the research of fuel cells by the U.S. Department of energy (DOE), the U.S. Government owns and operates 30 fuel cell cogeneration units, the world's largest fleet of fuel cells. Simultaneously, Japan, Canada, and Germany are promoting fuel cell developments with tax credits, low interest loans and grants to support early purchases and drive down costs.
11.1.4 Importance of Nanomaterials in Batteries and Fuel Cells
There are many chemical and physical requirements (Julien and Nazri, 1994) which will limit the choice of materials in the application of batteries and fuel cells. Various factors have contributed to the growth of new battery technologies. These factors, such as the performance requirements and specific environmental constraints of hand-held products will, in turn, determine and mould the material requirements for batteries. The key feature of the application of nanostructured materials in the development of batteries (Graetzel, 1996), particular by for rechargeable lithium ion batteries is the ease of rapid intercalation of Li+ ions (Dutta, 1996) into the lattice structure. For example, improvements in battery performance have been achieved by the use of nanocrystalline oxide films for the anode material. The microstructure of the oxide film is composed of a network of interpenetrating pores, which implies that a huge surface area can be electronically accessed. To increase the energy density in batteries, it is thus realized that the application of nanophase structures is desirable as this would provide the large surface areas required for chemical reactions to occur.
Fuel cells are characterized by the electrolytes used (Appleby, 1995), which determines both their operating temperature and the materials they can employ. The electrolytes used must possess high ionic conductivities and low concentration gradients as the choice of materials used is greatly limited by the effective rates of reactions on the catalytic electrodes. Fuel cells also require porous electrodes showing a microscopic three-dimensional interface between the reactant gas and electrolyte phases. Thus, the application of nanomaterials is essential to the development of practical fuel cells. The current state of fuel cells today revolves around the use of electrocatalysts (Sandstede, 1972; Appleby and Foulkes, 1989) required for the charge transfer reactions which produces the flow of electricity. Nanosized metallic powders are often used for this purpose. Small platinum (Pt) particles of sizes 2–3 nm at monolayer coverage offer the high surface areas needed for improved catalytic activity. This is mandatory for applications in electric vehicles since only nanosized Pt powders can offer the high electrocatalytic activity of the charge transfer reactions while keeping the catalyst loadings low. Besides the use of nanomaterials in fuel cells, it is also the intention of
27
this chapter to cover in detail the use of nanomaterials in secondary batteries rather than their primary counterparts.
10.5 Mechanism of Hexagonal-Cubic Transition
10.5.1 Model for the Transition Mechanism
The fact that w-BN domains are always included in small c-BN grains suggests that the phase transition from h-BN to c-BN occurs not directly but through w-BN. On giving shock compression to h-BN only w-BN has been obtained at medium temperature, whereas together with c-BN is also obtained at high temperature (Sekine, 1989). From these experimental results we may say that h-BN changes initially to w-BN and then to c-BN.
Figure 10.17(a) shows structure units of w- and c-BN. When w-BN converts to c-BN, a rotation of the lower part of the unit by 60° is apparently necessary. The rotation requires the breakdown of chemical bonds. It seems difficult and may be one of the main reasons why c-BN forms only under high pressure at high temperature.
Figure 10.17 A model for the mechanism of the conversion from w-to c-BN. (a) Shows conversion from structure unit of w- (reight) to that of c-BN (left). On the conversion a rotation of the lower part of the units by 60° is apparently necessary. • and mean B and N atoms, respectively. (b) The rotation is achieved by introducing a stacking fault in w-BN at the plane marked by a triangle. At the tip of the stacking fault a Shockley-type partial dislocation occurs. is the Burgers vector. The sequence of the atomic stacking changes from AB′ AB′ A′ to AB′ C′ A′ C′, i.e., to that of the c-BN locally. (c) If the similar stacking faults successively occur at every second sheet, the resultant structure will be that of c-BN. (d) In order to decrease mechanical strain, nano-scale twins are introduced. (from S. Horiuchi)
28
A possible mechanism for the rotation may be as follows: first we introduce a stacking fault at the plane in w-BN, marked by a triangle in Fig. 10.17(b). At the tip of the stacking fault a Shockley-type partial dislocation appears and the atomic array at the dislocation core is severely disturbed. The dislocation is taken to be of edge type. On the formation of the stacking fault the sequence of the atomic stacking changes from AB′ AB′ A′ to AB′ C′ A′ C′, i.e., to that of the c-BN locally. This means that the atoms have rearranged to achieve the rotation of Fig. 10.17(a). When the dislocation moves, the area of c-BN extends. The real movement of atoms at the dislocation core will be discussed later (Fig. 10.18).
Figure 10.18 Atomic movement during the conversion from w-to c-BN. (a) shows atomic positions on the planes marked by a triangle in Fig. 10.17(b). Chemical bonds are broken down at the end of the stacking fault, i.e., along the dislocation line, which runs along the vertical arrow 1. (b) On the movement of the dislocation in the rightward direction the chemical bonds at the positions of the arrow 2 are cut, while those of arrow 1
29
are restored. In the successive stage of the dislocation motion the chemical bonds of arrow 3 are cut and those of the arrow 2 are restored. (from S. Horiuchi)
If similar stacking faults successively occur at every second sheets, the resultant structure will be c-BN (Fig. 10.17(c)). In this case, however, a large mechanical strain is induced because of large deformation in shape.
Other combinations of stacking faults cause twinning (Pirouz and Yang, 1993). An example of a twin structure is shown in Fig. 10.17(d). A twin contains four sheets. In principle, twinning is possible with any thickness, depending on how frequently the stacking faults are introduced. It is important to note that twins can relax the mechanical strain induced on the transition to c-BN. The relaxation could be more effective if they would contain fewer sheets. This is the reason why many nanoscale twins have occurred in c-BN (Figs. 10.12 and 10.13).
The model proposed above can explain the orientation relationship among h-, w- and c-BN of Eq. (10.1). This supports the validity of the model.
30
10.5.2 Atomic Movement during the Conversion from w-to c-BN
Figure 10.18 shows atomic positions on the planes marked by a triangle in Fig. 10.17(b). Chemical bonds are broken down at the end of the stacking fault, i.e., along the dislocation line, which runs along the vertical arrow 1. On the movement of the dislocation in the rightward direction the chemical bonds at the positions of arrow 2 are cut, while those of arrow 1 are restored. In the successive stages of the dislocation motion the chemical bonds of arrow 3 are cut and those of arrow 2 are restored. It should be noted that the movement of atoms is simple and almost along the same direction as that of the Burgers vector of the dislocation, although each unit in the resultant structure has apparently been "rotated". The rate-controlling stage in the atomic rearrangement must be the breakdown of chemical bonds.
When the partial dislocation is of pure screw-type, the atomic rearrangement is also simple to form the stacking faults. Unidirectional movements of atoms are necessary to achieve the apparent "rotation" also in this case.
10.5.3 Facilitation of Synthesis of c-BN by Mechanochemical Effect
An essentially important factor in the transition mechanism proposed above is an introduction of lattice defect like dislocations as well as related stacking faults in w-BN, as shown in Fig. 10.17(b). In order to verify this from the standpoint of actual experiments, we have tried to do ball-milling on the initial h-BN powders under the expectation that the obtained w-BN include more defects. Figure 10.19(a) shows an HRTEM image of the specimen, in which w-BN has just been initiated in an h-BN matrix under high pressure at high temperature (Horiuchi, et al., 1998). In this series of experiments the formation of c-BN has started on the treatment under 7.7 GPa at 1250°C for pre-milled specimens and 1450°C for non-milled ones. That is, a prominent mechanochemical effect is observed, since the chemical reaction in a solid substance is accelerated under the introduction of mechanical strain.
Figure 10.19 (a) HRTEM image of the specimen, which was ball-milled and heated at 1350°C under 7.7 GPa. w-BN has appeared in h-BN matrix. (b) schematic representation of (a). h-and w-BN coexist with an interface nearly parallel to (0002). In h-BN a Frank-type edge dislocation is seen at the part of an arrow. The stacking sequence is ab′ c′ bc′ bc′ a′. In w-BN a similar stacking fault is included so that a stacking sequence of AB′ AB′ C′ BC′ B is created. In fact, such stacking sequences can be observed at the parts indicated by arrowheads in w-BN of (a). Since the dislocation is not brought in w-BN, an anti-phase boundary is formed. (from S. Horiuchi)
31
32
Figure 10.19(b) is a schematic representation of (a). h- and w-BN coexist with an interface nearly parallel to (0002). In n-BN a Frank-type edge dislocation is seen at the part indicated by an arrow, introducing a stacking fault. The stacking sequence is described as ab′ c′ bc′ bc′ a′. In w-BN a similar stacking fault is included so that a stacking sequence of AB′ AB′ C′ BC′ B is created. The stacking sequence for c-BN partly occurs. Since the dislocation is not brought in w-BN, an anti-phase boundary is formed. In fact, such stacking sequences can often be observed at the parts indicated by arrowheads in w-BN of Fig. 10.19(a).
If a Shockley-type partial dislocation moves, as mentioned above in relatio to Fig. 10.17(b), a region with a stacking sequence AB′ AB′ C′ A′ B′ A occurs. This means that a thin region of c-BN is nucleated. A possible source of the partial dislocation may be the anti-phase boundary, mentioned above. This type of reaction, however, requires the movement of dislocations for long distance.
It should be pointed out that, if some small areas with such a stacking sequence as above, arising from the Frank-type dislocations, are tentatively formed and adjoin to each other, thin areas of c-BN will be formed. This may be possible in the area, where the Frank-type dislocations are densely formed. Since the partial dislocation of Shockley type is required to move only a short distance in this case, the reaction becomes more probable.
In fact, according to HRTEM observation, many Frank-type dislocations have been created during ball-milling together with Shockley type ones. The former increased the number under high pressure at high temperature. For example, their density was locally counted to be 2 × 1012 cm-2 after treating the ball-milled h-BN under 7.7 GPa at 1350°C, as shown in Fig. 10.20 (Horiuchi, et al., 1998). We may then say that the formation of the Frank-type dislocation in high density is the essential reason for the occurrence of the present mechanochemical reaction.
Figure 10.20 HRTEM image of the specimen, which was prepared at 1350°C under 7.7 GPa after milling the starting h-BN. There are many dislocations with the Burgers vector (c/2) [0001], as indicated by arrowheads. Some of them are paired to form loops. (from S. Horiuchi)
33
We may assume that the Frank-type dislocations have been created when the amorphous areas, which formed locally in the milled specimen, crystallize to h-BN. In fact, it was noted that h-BN first nucleates in the amorphous BN, which has been formed by a stronger ball-milling (Huang, et al., 1999).
The geometry in Fig. 10.19 shows that the w-BN has been created by a simple compression of h-BN in the 0001 direction. The interface is however slightly tilted against the common (0001) plane of h- and w-BN.
The tilting is accomplished by the presence of dislocations in h-BN. It must be useful to release the lattice misfit of about 2% along the direction in the (0002) plane of h- and w-BN.
Finally, the transition to c-BN preferentially occurred in h-BN, whose (0002) plane is initially normal to the press direction. It is reasonable to consider that h-BN grains in such orientation suffered from severer deformation than those in other orientations so that the mechanochemical effect is more prominent.
10.6 Prospect
As mentioned, the introduction of lattice defects is effective on facilitating the hexagonal-cubic transition. This must be a useful method to synthesize c-BN in an industrial scale.
34
How much it is facilitated may depend on the degree of the mechanical strain introduced. In order to prove this we have performed a much stronger BM (planetary-type BM) and a more prominent mechanochemical effect was observed. The details will be published elsewhere (Huang, et al., 1999).
10.7 Conclusions
1. Structural evolution in BN during the hexagonal-cubic phase transition without any catalysts is examined using XRD, HRTEM and EELS.
2. The starting material was h-BN hot-pressed in a cylinder shape, which was obtained commercially. Grains have plate-like form with a thickness between 1–6 µm (Fig. 10.2). The plate plane is parallel to (0001). Each plate is actually composed of "sub-plates" with thickness of several nano-meters (Fig. 10.3). There is such a texture as the preferred distribution of 0001 normal to the hot-press direction. Besides, t-BN with curved sp2 sheets was locally found (Fig. 10.4).
3. It was pressed under high pressure (6.5–7.7 GPa) at high temperature(1700–2150°C), using a belt-type high-pressure machine (Table 10.1).
4. According to XRD (Table 10.1), a very small amount of c-BN is formed in h-BN in the recovered sample 2 (Figs. 10.5(b) and 10.6(b)), which was pressed under 6.5GPa at 1730 °C. Besides, the texture of h-BN becomes more prominent. Under 7.7 GPa at 1700°C (sample 3) a large amount of c-BN appears (Figs. 10.5(c) and 10.6(c)). The residual h(0002) peak becomes broad. This is due to the formation of m-BN. Moreover, a substantial amount of w-BN appears. At 1800°C (sample 4) the m(002) peak becomes slightly sharp (Fig. 10.5(d)). The peak becomes very weak at 2000°C (sample 5, Fig. 10.5(e)) and disappears at 2150°C (sample 6, Fig. 10.5(f)), showing that m-BN is an intermediate phase during the phase transition from h-BN to c-BN.
5. According to HRTEM, the plates of h-BN are prominently folded locally in sample 2 (Fig. 10.7(a)). During folding and bending, most of the h-BN plates tend to be normal to the press direction so that the texture becomes strong (Fig. 10.8).
6. The folding of the initial h-BN plates causes the shear (Fig. 10.9). As a result, the symmetry changes from hexagonal to monoclinic (Fig. 10.10). The lattice parameters of m-BN are: a is 0.433 nm, b is 0.250 nm, c is 0.31–0.33 nm, β is 92°—95°. The folding was performed mainly by mechanical twinning.
7. Grains of c-BN appear under 6.5 GPa at 1730°C. They occur in h-BN plates (Fig. 10.11). Nanoscale twins with (111) boundaries are seen in the c-BN grain (Fig. 10.12). Besides, small areas of w-BN are partially found in the grain. There is a definite orientation relationship among c-BN, w-BN and surrounding h-BN.
8. As heating temperature increases, the number of c-BN grains increases. They are consisted of nano-twins (Fig. 10.13) and their growth is not prominent as long as the temperature is lower than 2000°C. At 2150°C grains have grown with the formation of secondary twins (Fig. 10.14).
9. In the EELS core-loss patterns of h- and m-BN the π* peak is weaker than the σ* peak for h-BN, but stronger for m-BN, suggesting that their band structures of sp2 sheets are different (Fig. 10.16).
10. Small domains of w-BN are always found in small grains of c-BN (Figs. 10.12 and 10.15). We may then consider that the phase transition from h-BN to c-BN occurs not directly but through w-BN, being in agreement with the previous result of shock compression to h-BN (Sekine, 1990).
11. A model for the mechanism of the hexagonal-cubic transition was proposed (Fig. 10.17). In the model it starts with the conversion from h- to w-BN and is completed by that from w- to c-BN. The most severe point of the transition is in the second conversion, in which a rotation of atom groups is apparently necessary. The "rotation" can be
35
achieved by introducing a stacking fault in w-BN. At the tip of the stacking fault a Shockley-type partial dislocation locates. When the dislocation moves, the local area of c-BN extends. Moreover, in order to release a mechanical strain a number of nanoscale twins are formed in c-BN (Figs. 10.12 and 10.13).
12. The atomic movement during the conversion from w- to c-BN can be explained by a simple atomic rearrangement based on the movement of dislocation (Fig. 10.18).
13. In order to verify the validity of the above model from the standpoint of actual experiments, the initial h-BN powders were ball-milled and many dislocations and related stacking faults were introduced. The resultant w-BN included a number of dislocations (Fig. 10.19) and the formation of c-BN was prominently facilitated as compared to the non-milled case due to the mechanochemical effect; the formation of the Frank-type dislocation in high density (Fig. 10.20) exercises a positively effect so that the distance required for the movement of Shockley type dislocations is shortened.
References
Akaishi M., T. Sato, M. Ishii, T. Taniguchi and S. Yamaoka. J. Mater. Sci. Lett.. 12, 1883 (1994)
Bundy F. P. and R. H. Wentorf, Jr. J. Chem. Phys.. 38, 1144 (1963)
Corrigan F. R. High Pressure Science and Technology. In: 6th AIRAPT Conf.. eds. by Timmenhaus K. D. and M. S. Barber, Plenum Press, New York, vol. 1, pp. 994 (1979)
Corrigan F. R. and F. P. Bundy. J. Chem. Phys.. 63, 3812 (1975)
Endo T., O. Fukunaga, and M. Iwata. J. Mater. Sci.. 14, 1676 (1979)
He L. L., M. Akaishi, and S. Horiuchi. Microsc. Res. Tech.. 40, 243 (1998)
He L. L., T. Taniguchi T. Sato, and S. Horiuchi. J. Appl. Phys.. 82, 4241 (1997)
Horiuchi S. Fundamentals of High- Resolution Transmission Electron Microscopy. North-Holland, Amsterdam, pp. 211 (1994)
Horiuchi S., L. L. He, and M. Akaishi. Jpn. J. Appl. Phys.. 34, L1612 (1995)
Horiuchi S., L. L. He, J. Y. Huang, T. Taniguchi, and M. Akaishi. J. Surf. Anal.. 3, 197 (1997)
Horiuchi S., L. L. He, M. Onoda, and M. Akaishi. Appl. Phys. Lett.. 68, 182 (1996)
Horiuchi S., J. Y. Huang, L. L. He, J. F. Mao, and T. Taniguchi. Phil. Mag. 78, 1065 (1998)
Huang J. Y. T. Taniguchi, and S. Horiuchi. to be published. 1999
Ishii T., T. Sato, Y. Sekikawa and M. Iwata. J. Cryst. Growth. 52, 285 (1981)
Knittle E., R. M. Wentzcovitch, R. Jeanloz, and M. L. Cohen, Nature. 337, 349 (1989)
Lynch R. W. and H. G. Drickamer. J. Chem. Phys.. 44, 181 (1996)
Pease R. S., Acta Crystallog. 5, 356 (1952)
36
Pirouz P. and J. W. Yang. Ultramicrosc. 51, 189 (1993)
Sei H., M. Akaishi, and S. Yamaoka. Diamond Related Mater.. 2, 1160 (1993)
Sekine T. Shock Compression of Condensed Matter-1989. eds. by Schmidt S. C., J. N. Johnson, and L. W. Davison, Elsevier Sci. Pub. B.V., pp. 511
Slack G. A. and S. F. Bartram. J. Appl. Phys.. 46, 89 (1974)
Wakatsuki M., K. J. Takano, and G. Fujita. Physica. B. 139–140, 256 (1986)
Wentorf, Jr. R. H. J. Chem. Phys.. 26, 956 (1957)
13.2 Synthetic Strategies for Various Nanotube Architectures
13.2.1 Chemical Vapor Deposition
Chemical vapor deposition of hydrocarbons over metal catalysts has been a classical method to produce various forms of carbon fibers, filaments and multi-walled nanotubes in the past (Tibbetts, 1990, 1983; Endo, 1988; Snyder, et al., 1989; Baker and Rodriguez, 1994). The typical growth temperature Tg is typically 500 °C ≤ Tg ≤ 1000 °C. The first step in a CVD process involves the absorption and decomposition of hydrocarbon molecules on transition-metal (Fe, Ni, Co, etc.) catalytic particles. The carbon atoms diffuse into the interior of the catalyst to form a metal-carbon solid state solution (Baker, 1989; Tibbetts, et al., 1987; Tibbetts, 1984). Subsequent precipitation of carbon from the supersaturated catalyst particle will then occur and lead to the formation of a carbon tube structure (Fig. 13.1). Typically, two modes of nanotube growth can operate in CVD. The base-growth mode involves the metal catalyst particle pinned on the support substrate, and the nanotube lengthens with a particle-free closed end. Carbon feedstock is supplied from the base where the nanotube interfaces with the catalyst material (Fig. 13.1, left panel). The tip-growth model involves a metal catalyst particle at a nanotube end being carried away as the nanotube lengthens (Fig. 13.1, right panel). The carried-along particle is responsible for supplying carbon feedstock needed for the nanotube growth. For the synthesis of nanotubes, the catalytic metal nanoparticles are typically obtained on high surface area support materials such as Al2 O3 and SiO2. The size of the catalytic particles determines the size of the nanotubes. Multi-walled or single-walled can be synthesized by CVD depending on the particle size, as well as the type of hydrocarbon feedstock and growth conditions.
Figure 13.1 Schematic growth modes of carbon nanotubes in CVD. Single-walled nanotubes are shown as examples. Left panel: base-growth mode. Right panel: tip-growth mode.
37
Notably, a pitfall of CVD synthetic approaches has been that defective tubular carbon materials tend to be formed. Only recently, we have developed a CVD approach to grow nearly perfect SWNTs by using methane as the carbon feedstock (Kong, et al., 1998a, 1998b). This result will be presented later in the chapter.
13.2.2 Growth of Self-oriented Multi-walled Nanotubes
In controlling the orientation of nanotubes during CVD growth, previous methods have relied on growth of nanotubes in confined environments including the pores of mesoporous silica or channels of alumina membranes (Li, et al., 1996; Pan, et al., 1998; Che, et al., 1998; Kyotani, et al., 1996). We have found that nanotubes can self-assemble into aligned structures during CVD growth, and the driving force for self-alignment is the van der Waals interactions between nanotubes (Fan, et al., 1999). Our synthesis approach involves catalyst patterning and rational design of the substrate to enhance catalyst-substrate interactions and control the catalyst particle size. The substrates are porous silicon obtained by electrochemical etching of n-type silicon wafers in HF/methanol solutions. The resulting substrate consists of a thin nanoporous layer (pore size 3 nm) on top of a macroporous layer (with submicron pores) (Vial and Derrien, 1994; Smith and Collins, 1992). Patterned catalyst squares on the porous silicon substrate are obtained by evaporating a 5 nm thick iron film through a shadow mask. CVD growth using the substrate is then carried out in a 2 in. tube furnace at 700°C under an ethylene flow of 1000 sccm/min for 15–60 min. Figure 13.2(a) shows a scanning electron microscope (SEM) image of regularly spaced arrays of nanotube towers grown on top of patterned iron squares on a porous silicon substrate. The nanotube towers exhibit very sharp edges and corners with no nanotubes branching away from the blocks. The high resolution SEM image (Fig. 13.2(b)) reveals that the MWNTs (Fig. 13.2(b), inset) within each block are well aligned along the direction perpendicular to the substrate surface. The length of the nanotubes and thus the height of the
38
towers can be controlled in the range of 10–240 µm by varying the CVD reaction time, and the width of the towers is controlled by the size of the openings in the shallow mask. The smallest self-oriented nanotube towers synthesized by our method are 2 µm × 2 µm wide.
Figure 13.2 (a) Scanning electron microscopy image of arrays of bundled multi-Walled nanotube towers. (b) A high resolution SEM showing aligned MWNTs within a tower. The inset shows a TEM image of the bundled MWNTs.
The mechanism of nanotube self-orientation involves the nanotube base-growth mode (Fan, et al., 1999). Since the nanoporous layer on the porous silicon substrate serves as an excellent catalyst support, the iron catalyst nanoparticles formed on the nanoporous layer interact strongly with the substrate and remain pinned on the surface. During CVD growth, the outmost walls of nanotubes interact with their neighbors via van der Waals forces to form a rigid bundle, which allows the growth of nanotubes perpendicular to the substrate. The porous silicon substrates exhibit important advantages over plain silicon substrates in the synthesis of self-aligned nanotubes. Growth on substrates containing both porous silicon and plain silicon portions find that nanotubes grow at a higher rate (in length/min) on porous silicon than on plain silicon. This result suggests that ethylene molecules can permeate through the macroporous silicon layer and thus efficiently feed the growth of inner and outer nanotubes within the towers. The nanotubes grown on porous silicon substrates exhibit monodispersed diameters since catalyst nanoparticles with a narrow size distribution can be formed on the porous supporting surface, and the strong catalyst-support interactions prevent the catalyst particles from sintering at elevated temperatures during CVD growth.
13.2.3 Enable the Growth of Single-Walled Nanotubes by CVD
Chemical vapor deposition methods have been very successful in synthesizing carbon fibers, filaments and MWNTs (Tibbetts, 1983; 1990; Endo, 1988; Baker and Rodriguez, 1994; Snyder, et al., 1989). However, CVD synthesis of high quality SWNTs is only recent. Structurally perfect SWNTs can now be grown in a CVD process using methane as carbon feedstock and iron-oxide nanoparticles supported on high surface
39
area alumina as the catalyst (Kong, et al., 1998a). High temperature conditions (850–1000°C) are employed in the growth in order to overcome high strain energies in forming small diameter SWNTs (<5 nm), and obtain nearly defect-free tube structures. The choice of methane is critical to the CVD approach to SWNTs. We have found that methane is stable at elevated growth temperatures without appreciable self-pyrolysis. This stability prevents the formation of amorphous carbon that tends to cause catalyst poisoning and overcoating the nanotubes. Catalytic decomposition of methane by the transition-metal catalyst particles is thus the dominant process in SWNT growth (Kong, et al., 1998a, 1998b; Cassell, et al., 1999b).
Within the methane CVD approach, we find that the chemical and textural properties of the catalyst materials dictate the yield and quality of SWNTs (Cassell, et al., 1999b). Bulk quantities of high quality SWNTs can be synthesized by optimizing the catalyst. Thus far, our optimized catalyst consists of Fe/Mo bimetallic species supported on a sol-gel derived alumina-silica multicomponent material (Cassell, et al., 1999b). Shown in Fig. 13.3 is a transmission electron microscopy (TEM) image of SWNTs synthesized in bulk by using this catalyst. The image illustrates remarkable abundance of individual and bundled SWNTs that are free of defects and amorphous carbon coating. The diameters of the SWNTs are dispersed in the range of 0.7–5 nm with a peak at 1.7 nm. Weight gain studies find that the yield of nanotubes can be as high as 45 wt%. Through systematic studies, we have found that a good catalyst material for SWNT synthesis necessarily exhibits strong metal-support interactions, possesses a high surface area and large pore volume, and retains these characteristics at high temperatures without sintering. The strong metal-support interactions allow high metal dispersion and thus a high density of catalytic sites. The open pore structure of a catalyst allows efficient diffusion of reactant and intermediate hydrocarbon species. We believe that the rate-limiting step in SWNT CVD growth involves gas diffusion. This is based on the results that high SWNT yielding catalysts exhibit large pore volumes in the mesopore regime (Cassell, et al., 1999b).
Figure 13.3 A TEM image of SWNTs synthesized in bulk using a catalyst supported on a sol-gel derived alumina-silica hybrid material. Inset: an example of the frequently observed SWNT ends that are closed and free of metal particles.
40
13.2.4 Growth Mechanism of SWNT
The states of nanotube ends contain rich information about nanotube growth mechanisms. Careful high resolution TEM imaging of the SWNTs synthesized by our CVD method frequently observes closed tube ends that are free of attached or encapsulated metal particles (Fig. 13.3 inset). The opposite ends are typically found embedded in the catalyst support particles when imaged along the lengths of the nanotubes. These observations suggest that SWNTs grow in the methane CVD process predominantly via the base-growth process (Fig. 13.1, left panel) (Tibbetts, 1983, 1989, 1990; Tibbetts, et al., 1987; Tibbetts, 1984; Baker, 1989; Amelinckx, et al., 1994; Kong, et al., 1998a, 1998b; Cassell, et al., 1999b). Base-growth operates when strong metal-support interactions exist so that the metal species remain pinned on the support surface. In contrast, the tip-growth mode operates when the metal-support interaction is weak. In the methane CVD method, we find that enhancing metal-support interactions leads to significant improvement to the performance of the catalyst material in producing high yield SWNTs. This is rationalized by the increased catalytic sites and the facilitated base-mode growth processes. On the other hand, catalysts with
41
weak metal-support interactions lead to aggregation of metal species and reduced nanotube yield and purity (Cassell, et al., 1999b). Further understanding of the chemistry of catalysts and nanotube growth will undoubtedly lead to the synthesis of bulk quantities of high quality SWNTs approaching the kilogram scale.
13.2.5 Growth of Isolated Single-Walled Nanotubes on Controlled Surface Sites
The successful CVD synthesis of SWNTs in bulk forms has led to a straightforward synthetic route to addressable individual nanotube wires. By using substrates patterned with 1–5 µm wide catalytic islands, we obtain "nanotube chips" that contain isolated single-walled nanotubes grown from desired locations on the substrates (Kong, et al., 1998b, 1999; Soh, et al., 1999). Atomic force microscopy (AFM) images of SWNTs on a nanotube-chip are shown in Fig. 13.4, where the synthesized nanotubes extending from the catalyst islands are clearly observed. The diameters of the nanotubes are measured to be in the range of 0.7–4.0 nm, which is consistent with TEM results obtained with bulk SWNT materials. Some of the nanotubes have one end attached to a catalyst island and the other end terminated between islands. Nanotubes bridging islands with both ends attached to the opposing islands are also observed. As described in a later section, the bridging SWNTs allow reliable electrical connections to be made from the macroscopic scale to individual SWNTs. Thus, our controlled chemical synthesis opens up a new route to individual nanowire electrical circuits that are needed for fundamental and practical purposes.
Figure 13.4 (a) An AFM image of SWNTs grown from patterned catalyst islands on a silicon oxide substrate. (b) Image of an individual SWNT bridging adjacent islands.
42
43
13.2.6 Growth of Suspended SWNTs with Directed Orientations
Obtaining single-walled carbon nanotube architectures with nanotubes in aligned orientations has been challenging. We have devised a synthetic strategy that leads to suspended SWNTs directed towards controlled orientations parallel to the plane of a silicon substrate. The SWNTs are suspended bridges grown from catalyst material placed on top of regularly patterned silicon tower structures. The synthesis approach begins with developing a series of liquid-phase catalyst precursor materials that allow for uniform film formation and large-scale catalyst patterning. A specific precursor material consists of ethanol (40 mL) and 2-butanol (20 mL) solutions of P-123 block copolymer (1.0 g) (Yang, et al., 1998a, 1998b), AlCl3 · 6H2O (2.4 g), FeCl3 · 6H2O (0.09 g) and MoO2Cl2 (0.004 g). Using contact printing (Xia and Whitesides, 1998; Ferguson, et al., 1991) of a flat PDMS stamp inked with a film of the precursor material, we selectively place the precursor on top of tower arrays pre-made on a silicon substrate (Cassell, et al., 1999a). Calcination at 700°C for 4 h leads to the formation of alumina/silica mixed oxides confined on the tower tops. Subsequent CVD growth using the substrate yields SWNTs emanating from the towers. Directed free-standing SWNT networks are formed by nanotubes growing to adjacent towers and suspended above the surface. When examining with an SEM, we observe that highly directional suspended SWNTs are formed on the synthesized sample. The directions of the suspended tubes are determined by the pattern of the towers. Well-aligned SWNT bridges are obtained in an area of the substrate containing isolated rows of towers as shown in Fig. 13.5(a), where suspended tubes forming a power-line-like structure can be seen. In an area containing towers in a square configuration, a square of suspended nanotube bridges is obtained (Fig. 13..5(b)). Directionality of the suspended tubes is simply a result of the rationally designed substrate. During the CVD growth, nanotubes emanate from the top of the towers. The nanotubes growing towards adjacent towers become suspended, whereas nanotubes directed towards other orientations fall onto the sidewalls of the towers (not easily resolved under SEM). In Fig. 13.5(c), we show a TEM image of a suspended SWNT bridge between silicon towers, and an image (Fig. 13.5(c), inset) showing the high resolution structure of the SWNT.
Figure 13.5 (a) SEM image of a suspended SWNT "power-line-like" structure. (b) SEM image of a square of suspended SWNT bridges. (c) TEM image of a SWNT bridge suspended between silicon towers. Inset: a high magnification TEM image showing the structure of a SWNT.
44
45
The directed growth of suspended SWNTs presented here involves developing a new type of liquid phase catalyst material, contact printing of catalyst onto designed substrates and CVD synthesis. The method should open a new window in characterization and device applications of organized nanowire architectures in suspended states or after being transferred onto flat substrates.
46
11. Nanomaterials for Energy Storage: Batteries and Fuel Cells
11.1 General Overview of Batteries and Fuel Cells
11.1.1 Introduction
Batteries and fuel cells are important power sources today (Berger, 1997; Georgano, 1996; Ondrey, et al., 1999) and will continue to be used in a wide variety of consumer, industrial and military applications in the 21st century. As the technologies of electronics industry advance, batteries are becoming a critical component (Ruetschi, et al., 1995; Seitz, 1991) for portable electronic devices, lighting, photographic equipment, watches, calculators, memory backup and a wide variety of other applications, giving freedom to utility power. Batteries have many advantages (Linden, 1984) over other power sources; they are usually self-contained, efficient, convenient, reliable and can be easily configured to user requirements.
Beyond batteries, fuel cells are highly efficient and less pollutive power-generating systems that produce DC electricity through the combination of fuel and oxidant in an electrochemical reaction (Apple by and Foulkes, 1989). As such, these are becoming more and more important. Traditionally, the electrical supply homes is generated in coal- or gas-fired power stations. Chemical energy from coal is burnt to produce heat energy which changes water into steam. Steam is then used to turn huge turbines and several rolls of wires are connected to these spinning turbines. As these move through a strong magnetic field usually created by arrays of powerful magnets, an electric current is induced to flow through the wires. The electric supply is then connected to the power sockets by running cables throughout a large area. This has not been an efficient way of harvesting energy, because some available energy in coal is lost at every stage of conversion. In fact, coal-fired stations are only around 30% efficient in the conversion of chemical energy from fuel to electrical energy. There is a further loss of energy through transportation of the electrical supply by high voltage cables. This huge consumption of fuel is responsible for the rapid depletion of our nonrenewable energy sources. On top of this, inefficient combustion of fuels usually creates disruptive impacts on the environment including pollution and strong possibility of global climate change. At this end, fuel cells are able to promote energy diversity and provide a transition to renewable energy sources. The most abundant element on the Earth, hydrogen (Pohl, 1995) can be directly used in the functioning of fuel cells. Alternative fuels containing hydrogen, including natural gas, methanol, ethanol and even diesel or gasoline fuel can be utilized. Because fuel cells convert chemical energy directly into electrical energy without the intermediate combustion processes, these are not limited by the Carnot efficiency of thermal engines and are usually 60% efficient, equivalent to about 2–3 times more efficient than the combustion processes. As a result, these have lower emission levels, producing less CO2 associated with more traditional means of power generation.
11.1.2 An Overview of Batteries
The first work on batteries was done by Volta (Applely and Foulkes, 1989) around 1800. Ritter (Vinal, 1950) subsequently constructed what was perhaps the first battery. In 1859, Planté began the foremost studies
47
(Vinal, 1950) which later led to the development of the first practical rechargeable (secondary) battery, the lead-acid battery. Since this early work, a variety of new battery systems have been discovered and developed (McCroy, 1977). The introduction of the nickel metal hydride (NiMH) in the late eighties and the lithium ion (Li-ion) in the early nineties brought more energy to a given cell compared to the earlier generation of batteries including the nickel cadmium (NiCd) batteries. Figure 11.1 compares the gravimetric and volumetric energy densities of rechargeable lithium batteries with those of other systems.
Figure 11.1 The gravimetric and volumetric energy densities of rechargeable lithium batteries as compared with those of other systems.
In 1990, an initial prototype Li-ion (AA size) cell made of carbon/lithium cobalt oxide (C/LiCoO2) was introduced to the battery world by Sony Energytec. Subsequently, a joint venture company A & T Battery, was formed by Asahi Chemical Industry and Toshiba Battery to manufacture C/LiCoO2 cells. Carbon/lithium manganese oxide (C/LiMn2 O4) technology was later developed in 1992 by Bell Communications Research. Basically, it was clear that lithium ion rechargeable had became the most promising battery technology, with most manufacturers of small rechargeable batteries in Japan, the United States and Europe engaging in Li-ion research by mid-1994.
11.1.2.1 Operation Principle of Batteries
The operation (Linden, 1984) of a battery cell during charge and discharge is shown schematically in Fig. 11.2. During discharge, electrons will flow from anode through an external load once the cell is connected to it. At the anode, the oxidation takes place, while at the cathode a reduction reaction occurs. The electrical circuit is completed in the electrolyte by the flow of negativeions (anions) and positive ions (cations) to the anode and cathode, respectively. Thus, the electronic current delivered by the cell is matched by the ionic current within the cell. Any leakage of electrons from the anode to the cathode within the cell reduces the
48
current generated by the battery. When a piece of zinc metal is assumed as the anode material and chlorine (Cl2) is the cathode material, the discharge reactions (shown in Table 11.1) take place.
Figure 11.2 The operation of a battery cell during (a) charge and (b) discharge.
Table 11.1 Reaction mechanisms for a Zn/Cl2 cell
Zn/Cl2 Cell
Anode (negative electrode) Zn → Zn2+ + 2e-
Cathode (positive electrode) Cl2 + 2e- → Cl-
Overall Zn + Cl2 → Zn2+ + 2Cl- (ZnCl2)
Batteries are the lifeblood of many low-power portable devices which face an increasing demand. Essentially, they can be classified into two broad categories. Depending on the batteries' capability of being electrically rechargeable, they can be identified as primary (non-rechargeable) or secondary (rechargeable) cells. Primary cells are not capable of being easily recharged and are suitable for one-time use only. However, they are usually light, inexpensive, of high energy density and possess good shelf-life. Secondary batteries can be electrically recharged to their original condition after discharge by passing current through them in the opposite direction to that of the discharge current. They are good storage devices for electrical energy and are often charged by a primary energy source, only to deliver energy to the load upon demand. In addition to their ability to be recharged, such batteries are characterized by high energy density, high discharge rates, flat discharge curves and good low-temperature performance. The secondary batteries are
49
used essentially to lower cost since they can be used multiple times. They are also more environmentally friendly as there is less concern for battery disposal compared to primary batteries.
11.1.2.2 Research of High Energy Storage Batteries
In the early world market for small rechargeable batteries, where it was basically dominated by two rechargeable battery technologies, nickel cadmium (NiCd) batteries held more than 80% of world markets while sealed lead-acid (SLA) batteries held about 13% (Seitz and Shimosato, 1994). In recent years, three advanced rechargeable battery technologies appeared in the market as well, namely the nickel metal hydride (NiMH), rechargeable alkaline manganese (RAM), and lithium ion (Li Ion). Presently, the huge spate of research into rechargeable lithium batteries has led to the emergence of lithium-solid positive-electrode prototype cells ever since the first Li/MoS2 production cell. Shown in Fig. 11.3 are the market share for the penetration of rechargeable cells in Japan, United States and Western Europe to the year 2000.
Figure 11.3 The market scenarios for the penetration of rechargeable cells in Japan, the United States and western part of Europe by the year 2000. * Li+ = lithium Ion, NiCd = nickel Cadmium, NiMH = nickel metal hydride, RAM = rechargeable alkaline manganese, SLA=sealed lead acid.
50
11.1.3 An Overview of Fuel Cells
The first fuel cell (O'Sullivan, 1972) was built up by Sir William Grove, a Welsh judge and scientist, in 1839. However, the use of fuel cells had been confined to laboratories until recent decades. It was only in the 1960s that fuel cells were engaged to provide power on board for the Gemini and Apollo space missions (Kordesch and Simader, 1996) while providing the astronauts with clean drinking water.
11.1.3.1 Operation Principles of Fuel Cells
Fuel cells have beneficial operating characteristics unmatched by any other technology. As in batteries, silent reactions produce an electric current. Unlike batteries, however, fuel cells can be continuously
51
recharged. In fuel cells run on hydrogen, oxygen from the air reacts with the input hydrogen in such a way that a voltage is generated between the two electrodes; the reactions occur via a chemical mediator known as the electrolyte. The basic design of a fuel cell is illustrated in Fig. 11.4. Two catalyzed carbon electrodes are immersed in an electrolyte while the fuel, hydrogen, is supplied from one surface of the electrode. When the electrodes are electrically connected through an external load, the following events occur:
1. Dissociation of hydrogen occurs on the catalytic surface of the anode, forming hydrogen ions and electrons. 2. Migration of the hydrogen ions occurs through the electrolyte to the catalytic surface of the oxygen electrode
(cathode). 3. Movement of the electrons through the external circuit to the catalytic surface of the oxygen electrode. 4. The formation of water due to combination of the hydrogen ions, oxygen and electrons on the cathode's catalytic
surface.
Figure 11.4 Schematic operation principle of a fuel cell with a solid polymer electrolyte.
For different types of fuel cells, the ionic species carrying charge from one electrode to another varies and they are generally distinguished by the type of electrolytes used. The fuel cell depicted in Fig. 11.4 has an acidic electrolyte (such as phosphoric acid or a proton-exchange membrane), relying on H+ ions to carry charge from the anode to the cathode. In various other fuel cells, the charge can move from cathode to anode via O2- ions (solid oxide ceramic fuel cells), OH- ions (alkaline fuel cells) or CO2-
3 ions (molten-carbonate fuel cells). For example, the reaction mechanisms for a hydrogen fuel cell in acidic and alkaline electrolytes are shown in Table 11.2.
Table 11.2 Reaction mechanism for fuel cells
52
Acidic electrolyte Alkaline electrolyte
Anode H2 → 2H+ + 2e- H2 + 2OH- → 2H2O + 2e-
Cathode O2 + 4H+ + 4e- → 2H2O O2 + 4e- + 2H2O → 2OH-
Overall 2H2 + O2 → 2H2O 2H2 + O2 → 2H2O
11.1.3.2 Research of Fuel Cells
Fuel cells were first considered for applications where noxious emissions or noise would be objectable and where water was unavailable. With the additional advantage of on-site operation, rather than remote fuel cells can also be utilized for portability. Commercialization has been a leading theme for fuel cell technology in recent years. The major use of fuel cells as potential replacements for internal combustion engines in transportation (Lemons, 1990; Prater, 1990; Srinivasan, et al., 1988) is being actively promoted. Early in the next century, fuel cells may start to appear as alternatives to internal combustion engines in automobiles, buses and other vehicles. In a pursuit for "clean" power, several vehicle demonstration projects have already been employed around the world, including those of H Power of Belleville, New Jersey; Ballard Power Systems of Vancouver, British Columbia, and Daimler-Benz of Germany.
In the first major practical use of fuel cells, the Gemini space flight missions, the solid polymer electrolyte fuel cell system (Kordesch and Simader, 1996) of the General Electric Company was used. In 1987, Ballard made a subsequent breakthrough in fuel cell performance through a new sulfonated fluorocarbon polymer membrane from Dow Chemical, similar to the earlier Nafion membrane (Gavach and Pourcelly, 1992) developed by DuPont, but capable of passing four times the current at the same operating voltages. Besides huge investments into the research of fuel cells by the U.S. Department of energy (DOE), the U.S. Government owns and operates 30 fuel cell cogeneration units, the world's largest fleet of fuel cells. Simultaneously, Japan, Canada, and Germany are promoting fuel cell developments with tax credits, low interest loans and grants to support early purchases and drive down costs.
11.1.4 Importance of Nanomaterials in Batteries and Fuel Cells
There are many chemical and physical requirements (Julien and Nazri, 1994) which will limit the choice of materials in the application of batteries and fuel cells. Various factors have contributed to the growth of new battery technologies. These factors, such as the performance requirements and specific environmental constraints of hand-held products will, in turn, determine and mould the material requirements for batteries. The key feature of the application of nanostructured materials in the development of batteries (Graetzel, 1996), particular by for rechargeable lithium ion batteries is the ease of rapid intercalation of Li+ ions (Dutta, 1996) into the lattice structure. For example, improvements in battery performance have been achieved by the use of nanocrystalline oxide films for the anode material. The microstructure of the oxide film is composed of a network of interpenetrating pores, which implies that a huge surface area can be electronically accessed. To increase the energy density in batteries, it is thus realized that the application of
53
nanophase structures is desirable as this would provide the large surface areas required for chemical reactions to occur.
Fuel cells are characterized by the electrolytes used (Appleby, 1995), which determines both their operating temperature and the materials they can employ. The electrolytes used must possess high ionic conductivities and low concentration gradients as the choice of materials used is greatly limited by the effective rates of reactions on the catalytic electrodes. Fuel cells also require porous electrodes showing a microscopic three-dimensional interface between the reactant gas and electrolyte phases. Thus, the application of nanomaterials is essential to the development of practical fuel cells. The current state of fuel cells today revolves around the use of electrocatalysts (Sandstede, 1972; Appleby and Foulkes, 1989) required for the charge transfer reactions which produces the flow of electricity. Nanosized metallic powders are often used for this purpose. Small platinum (Pt) particles of sizes 2–3 nm at monolayer coverage offer the high surface areas needed for improved catalytic activity. This is mandatory for applications in electric vehicles since only nanosized Pt powders can offer the high electrocatalytic activity of the charge transfer reactions while keeping the catalyst loadings low. Besides the use of nanomaterials in fuel cells, it is also the intention of this chapter to cover in detail the use of nanomaterials in secondary batteries rather than their primary counterparts.
11.2 Batteries and Nanomaterials
11.2.1 Classifications of Advanced Batteries
Today, the most promising batteries are based on the lithium ion (Li-ion), lithium ion polymer (Li-ion-P), nickel metal hydride (NiMH), and nickel cadmium (NiCd) batteries, all being of the rechargeable type.
11.2.1.1 Lithium Ion (Li+-Ion) Battery
The lithium ion battery system (Pistoia, 1994; Scrosati, 1995), commonly used in high energy density applications, is the fastest growing technology. Despite the fact that it is the latest to emerge from the laboratory into the real-world applications, this revolution in battery technology is targeted to fulfil the requirements for future applications like electric vehicles and biomedical devices such as an artificial heart (Owens, et al., 1984). The pioneering work for research on lithium batteries was started by G. N. Lewis in 1912, but it was not until the seventies that the first non-rechargeable lithium batteries became commercially available. Lithium (Bach, 1985), being the lightest of all metals, has the greatest electrochemical potential and provides the largest energy content. However, it was found that occasional shorts from lithium dendrites developed in lithium electrodes could cause thermal runaway. The temperature would approach the melting temperature of metallic lithium which could result in violent explosions (Wilkinson, et al., 1990). Due to this inherent instability, especially during the charging process, attempts to develop rechargeable lithium batteries failed and research has since shifted to a non-metallic lithium battery. The concept of shuttling lithium ions between insertion electrodes was first proposed in the 1980s (Lazzari and Scrosati, 1980;
54
Armand, 1980; Mizushima, et al., 1980; Auborn and Barbario, 1987; Tarascon, 1985). As a result, chemicals such as lithium cobalt dioxide (LiCoO2) were used, albeit that they yielded a lower energy density compared to lithium metal (Buchmann, 1997). The lithium ion is much safer and has a reasonably high energy density, which is at least twice that of NiCd batteries. In the present market, there are two basic types of Li+-ion batteries, namely the coke version provided by Sony and the graphite version that has been adapted by most other manufacturers. The new graphite electrode that has emerged provides a relatively flatter discharge voltage curve than the coke electrode and is capable of delivering a higher current while remaining cool during both the discharge and charge processes.
Li+-ion batteries avoid the potential flammability and dendrite-shorting problems (Brown, et al., 1988) characteristic of metallic lithium rechargeable batteries by having lithium ions shuttle back and forth (Venkatasetty, 1984) between the two electrodes (the "rocking chair" technology). The chemical reactions at the anode and cathode of a lithium secondary battery must be reversible. The anode supplies Li+ ions to the Li+ ion electrolyte while directing electrons to the load circuit upon discharge. On the other hand, the cathode is an electronically conducting host into which the Li+ ions are inserted from the electrolyte as guest species and charge-compensated by electrons from the external circuit. On charge, removal of electrons from the cathode by an external field will facilitate the flow of Li+ ions back to the electrolyte and hence restores the parent host structure. The addition of electrons to the anode by the external field attracts charge-compensating Li+ ions back into the anode to restore it to the original composition. In principle, the anode can be elemental lithium itself (Yang, et al., 1996); in practice, it is found necessary to use a reductant host for lithium. For example, the Sony cell was based on this "rocking chair" concept, composing of a carbon anode, LiCoO2 cathode and a non-aqueous electrolyte. A lithium-insertion compound is usually used for the electrodes whereby there is a host matrix into/from which the guest species Li+ can be inserted or extracted reversibly without any change in the arrangement of the host atoms. Such a concept has been demonstrated with the layered TiS2 cathode which consists of a close-packed hexagonal sulfide ion array. An example of the lithium insertion process is illustrated in Fig. 11.5.
Figure 11.5 An example of the lithium insertion process with the layered TiS2 cathode consisting of a close-packed hexagonal sulfide ion array.
55
11.2.1.2 Lithium Ion Polymer (Li-ion-P) Battery
Closest to commercial viability is lithium polymer technology, which seeks to improve the advances made by liquid electrolyte Li-ion batteries in recent years. Currently under development, the Li-polymer battery system is a rechargeable yet potentially cheaper version of the Li-ion battery (Birke, et al., 1999). The projected commercial market is expected to commence by the year 2000. When commercially available, the Li-polymer battery will offer a high energy density with low self-discharge but may only be suitable for low-power applications. The original concept of the Li-polymer battery was based on the use of a solid polymer electrolyte which offers great potential in design flexibility with respect to fabrication, ruggedness, safety and cost. Such a design avoids the presence of any leakage and high flammability of the liquid electrolyte used in Li-ion and lithium batteries. If fully developed, the Li-polymer will provide more than thrice the energy density compared to NiCd batteries.
11.2.1.3 Nickel Metal Hydride (NiMH) Battery
The nickel metal battery cell consists of a positive nickel hydroxide electrode and a negative electrode comprising hydrogen-absorbing alloys put in an alkaline electrolyte. The overall reaction is
When the NiMH battery was first introduced, there was great publicity about its "memory-free" status, since it was believed that the NiMH battery need not be fully discharged before each charging cycle. However, today research has shown that they suffer the same "memory" effects (Davolio and Soragui, 1998;
56
Buchmann, 1996) in much the same way as nickel cadmium (NiCd) batteries. The crystalline formation which causes capacity loss is created by the nickel plate, a metal common to both the NiMH and NiCd chemistries (Thaller and Zimmerman, 1996; Sato, et al., 1996) Besides, cell capacity will be limited by the positive electrode as the generated oxygen at this electrode has to be recombined. This is achieved by the application of an excess negative material in the ratio of 3 : 2 with respect to the positive electrode. Traditionally, NiMH batteries are commonly used in cellular phones, video cameras and laptop computers where high energy is of utmost importance. However, these incremental improvements in higher capacity over nickel cadmium batteries are often at the expense of reduced cycle life and lower load current.
11.2.1.4 Nickel Cadmium (NiCd) Battery
The nickel cadmium battery is one of the most mature and well-understood technologies today. Required in applications where long life, high discharge rates, and cost are important factors, NiCd batteries found lasting uses in portable radios, video cameras, and power tools. The NiCd battery possesses the shortest charge time, delivers the highest load current and offers the lowest cost-per-cycle, however, it is the most demanding on exercise requirements to prevent any "cyclic memory" effects. Nevertheless, this problem was virtually solved with the advancements in battery technology and hence the elimination of this phenomenon. The problem now is not so much on the "memory" effects but on the effects of crystalline formation. In the modern NiCd battery technology, the active materials are present in finely divided crystals (Buchmann, 1997). The crystals should remain small, hence obtaining maximum surface area but this is not the case. When "memory effects" occur, the crystals actually grow up to 150 times the original size and drastically reduce the surface area (Buchmann, 1997). Under certain circumstances, the sharp edges of these crystals may pierce the separator and cause a high self-discharge or an electrical short. Another form of memory that occurs on some cells is the formation of an intermetallic compound consisting of nickel and cadmium which holds some of the needed cadmium and imposes an extra resistance in the cell (Davolio and Soragni, 1998).
11.2.2 Major Components of Batteries
Batteries have had such a long history that major developments have led to more diversified applications, particularly so nowadays with more rudimentary use of new materials. Nevertheless, there are still some basic components common to the structure of every battery: the electrodes and the electrolyte. A battery consists of a group of interconnected electrochemical cells. An electrode is a condensed phase which has the property of electronic conduction: it can be a semi-conductor or a metallic conductor taking on various forms, e. g., an amalgam, a liquid, or a solid metal, graphite or carbon conducting carbides, borides or nitrides, many oxides and even sulfides. Electrodes are in direct contact with the electrolyte which is an ionic conductor and can be a solid or an aqueous phase.
Electrodes in batteries can be broadly categorized into four groups with respect to their mode of operation, which may take a number of differing forms (Barak, 1980).
57
1. A metal electrode or alloy in contact with an electrolyte containing the same ions of the metal. As such, the metal is oxidized and may dissolve freely to provide electron flow as in Zn/air primary batteries. By convention, this is represented by M Mn+
(sol). 2. An inert metal electrode at which an oxidation or reduction may take place with electron transfer. Other than acting
as a source or sink of electrons, the metal does not play any other role in participating in the reactions. 3. A metal in contact with its salt or oxide form which stays in contact with an electrolyte containing the anion of the
salt or hydroxyl ions. 4. An inert conductor in contact with a salt or oxide and totally in contact with an electrolyte containing an ion which
undergoes an oxidation or reduction with electron transfer.
The electrolyte, which is also an ionic conductor, provides the medium for the transfer of charges as ions inside the cell between the anode and cathode. Typically, a liquid such as water or other solvents, the electrolyte contains dissolved salts, acids or alkalis to impart the required conductivity. More recently, certain crystalline and amorphous substances (Julien and Nazri, 1994) have been as solid electrolytes which are ionic conductors at their operating temperature.
Where the electrochemical reaction involves a gas phase across an interface, the construction of a porous, small-particle electrode is necessary as shown in Fig. 11.6. The achievement and retention of a relatively high electrode capacity, i.e., the utilization of a large area of electrode material in the electrochemical reactions, requires good electronic contact between particles over several charge/discharge cycles. Electrochemical performances are strongly influenced by the powder characteristics such as particle size, porosity, surface area and morphology of the electrode properties. For example, lithium ion mobility and rate capability can be increased by keeping the particle size small.
Figure 11.6 The construction of a porous, small-particle electrode in the electrochemical reaction involving the ionic transport across an interface: (a) gas; (b) carbon particle; (c) wet-proofing agent; (d) electrolyte.
For the more popular secondary batteries, some additional difficulties arising from the need to recycle the systems have been reviewed elsewhere (Sequeira and Hooper, 1985). These include relatively low diffusion coefficients for ionic transport within the intercalation materials and the various failure mechanisms.
58
11.2.3 Applications of Nanomaterials in Advanced Batteries
11.2.3.1 Electrodes
11.2.3.1.1 Nickel Cadmium/Nickel Metal Hydride Electrodes
The cathode for the nickel cadmium or nickel metal hydride battery, NiOOH, is composed of a layered oxide with Ni3+ ions residing on every basal plane of the octahedral sites in a close-packed cubic array of oxide ions (Zimmerman, 1994). The protons provide a network of hydrogen bonding between the O—Ni—O sandwich layers with one hydrogen bond per oxygen atom in Ni(OH)2 (Yamada, et al., 1999). On discharge, trivalent nickel hydroxide is reduced to divalent nickel hydroxide under the consumption of water, while the metallic cadmium is oxidized to cadmium hydroxide. On charge, the opposite reactions take place. In this way, the protons may be reversibly inserted into the hydrogen-bonding network of NiOOH if charge compensation can occur via the introduction of electrons into the Ni3+/Ni2+ couple as indicated by the reaction below:
Cobalt metal or oxides at 5–10 wt% are usually deposited on the Ni(OH)2 cathodes to impart conductivity (Ferrando and Lee, 1984; Lee, 1986). Due to its costly nature, lower levels of Co are utilized while, on the other hand, the loss of conductivity associated with it (Reisner, et al., 1997) can be compensated by the addition of nanophase cobalt particles. This is critical in establishing cost effectiveness. However, as with most batteries, several mechanisms such as self-discharging (Ikoma, et al., 1996) and loss of conductivity can cause capacity loss in the NiMH system upon long term storage. Self-discharge is often caused by the decomposition of NiOOH, the possible degradation of the separator and migration of cobalt from the positive electrode. To solve these problems, it was proposed that the capacity recovery of the battery can be improved by the addition of stable conductive powders such as Ni 210 (Singh, et al., 1998); the main reasons being that Ni 210 powders possess very fine morphologies and high surface areas, which are typical of nanopowders.
Several unique nanostructured materials have been developed by U. S. Nanocorp. Inc. (Reisner, et al., 1997) for use in a variety of energy storage applications such as batteries. For instance, nanostructured β-nickel hydroxide has been synthesized, and this hexagonal form of Ni(OH)2 can significantly boost the energy density of the nickel cadmium/nickel metal hydride or even nickel zinc batteries. For over 90 years, the traditional material used for batteries usually has more than 50% pores, of which there were a significant number of large pores not required for electrochemical efficiency. Clearly, under the conditions of superplasticity in nanophase hard ceramics and intermetallic materials, high energy densities close to the theoretical densities are attainable even at ambient temperatures. Thus, U. S. Nanocorp. Inc. initiated several wet chemical synthesis approaches for these active nanostructured battery materials (Reisner, et al., 1996), after recognizing that the metallic oxide Ni(OH)2 can also show similar behavior. The assynthesized β-Ni(OH)2 powder is a mixture of highly nanoporous fibers and equiaxed grain particles with fibers being ca. 2–5 nm in diameter by 15–50 nm long, and the grains having an average particle size of 5 nm. With the nanosized particles capitalizing on the huge surface area, the present cathodic material has a much higher
59
packing density than other conventional materials. Capable of boosting the performance of both the nickel alkaline and lithium rechargeable batteries in a cost-effective manner, these ultrafine particle-size materials must still be agglomerated to become useful battery materials. Favorably, the active materials would be able to attain a homogeneous porosity with a narrow pore distribution.
A number of studies have shown that certain intermetallics such as the nanocrystalline AB, AB2, and AB5 type alloys are suitable materials for use as anodes in rechargeable NiMH batteries due to their high electrochemical capacity and excellent cycle life. Mechanical milling has been introduced by Jung (Jung, et al., 1998) to modify the electrocatalytic activity and activation behavior of the ZrCrNi alloy, thus producing a composite-like structure (ZrCrNi + nickel) and nanocrystalline ZrCrNi. In their study, the new type of metal hydride particle is composed of a composite-like structure with two components (metal hydride + nickel). It was reported that the Ni nanoparticles not only serve as electrocatalytic reaction sites but as hydrogen adsorption sites as well. Easier hydrogenation behavior is believed to be a consequence of ball milling with pure nickel, which supplies the Ni regions surrounding the ZrCrNi alloy particles, hence creating various defects both on the surface and the interior of the powder particles. In addition, electrochemically activated nanocrystalline ZrCrNi powder particles were also formed after the milling process.
11.2.3.1.2 Lithium Rechargeable Electrodes
The earliest rechargeable lithium battery design consisted of a negative electrode fashioned from lithium metal, a metal oxide intercalation compound as the positive electrode and a non-aqueous liquid or solid polymer electrolyte. After much research, the lithium metal anode was replaced by a second intercalation electrode, which was usually carbon-based. Where both the anode and the cathode are hosts for the reversible insertion or removal of the working ion into/from the electrolyte, the electrochemical cell is commonly called a "rocking chair", or a swing battery (Julien and Nazri, 1994). The present day lithium batteries, however, use a solid reductant as the anode and a solid oxidant as the cathode such as LiS2, LiMn2O4 or LiCoO2. In fact, commercially available lithium ion batteries are made with the more expensive LiCoO2 cathodes (Huang, et al., 1998; Kim, et al., 1996; Numata, et al., 1999) and carbon anodes; however, there is much interests in using new oxides solely to lower the cost and reduce environmental concerns. It was also found that mixed-valent transition metal oxides tend to be good electronic conductors, and with few exceptions they are generally stable against disproportionation reactions. Besides, oxides allow the achievement of higher voltages than sulfides because higher cation valence states can be achieved in oxides (Wakihara and Yamamoto, 1998). γ-MnO2 (Davidson, et al., 1995; Bruce, et al., 1999) and spinel Mn2O4 (Molenda, et al., 1999) have been studied more extensively as cathodic materials because these are less expensive, less toxic and can be prepared by easier methods than can the cobalt oxides. Nanofibrous MnO2 (Kordesch and Weissenbacher, 1994; Rossouw, et al., 1992; Kao, et al., 1992; DeGuzman, et al., 1994; Ohzuku, et al., 1991) with a hollandite structure exhibiting an unusual "bird's nest" morphology (Feng, et al., 1994; Benaissa, et al., 1997) is one such example. Each "bird's nest" is about 10 µm in diameter, which constitutes an assemblage of many individual nanofibers with diameters from 5 to 25 nm. This geometry has potential applications as an intercalation material in lithium ion rechargeable batteries as well as in zinc/manganese dioxide rechargeable alkaline cells. Due to the extremely low conductivity observed in manganese oxides, the use of conducting additives such as Co particles is necessary to improve the activities as cathodes.
60
Composites of manganese oxides particles and polypyrrole (PPy) have been attained by the addition of pyrrole into acidic aqueous suspension of the oxide (Gemeay, et al., 1995; Kuwabata, et al., 1994; Yoneyama, et al., 1991), and the prepared composites had enhanced specific capacities compared to the oxides mixed with carbon powder. To further improve on the charge/discharge performance, the electrode structure of the LiMn2O4/PPy composite can be made in nanoscopic scale, it has been demonstrated by Tarascon and co-workers (Tarascon et al., 1991; Shokoohini, et al., 1992) that the morphology of LiMn2O4 plays an important role in its electrode properties. This was shown by the increase in energy capacities of the electrode with decreasing oxide grain size. In addition, the Li+-ion mobility and rate capability can be increased by keeping the particle sizes of the electrode materials small. The synthesis of transition metal oxides is usually done at ambient temperatures in solutions employing reducing agents. Alkali-metal borohydrides ABH4 were frequently engaged to reduce [MO4]n- (M = V, Cr, Mn, Mo, and W) under aqueous conditions to obtain binary or ternary oxides MxOy and AxMyOz (A is Na or K). Such reduction methods (Tsang, et al., 1998) have resulted in the formation of nanocrystalline or amorphous oxides that were generally metastable. The nanocrystalline or amorphous nature of these materials is exceptionally attractive for achieving better lithium ion diffusion.
Cathodes based on the creation of inorganic/organic hybrid structures that exhibit cooperative interaction among their constituents are hopeful candidates for use in Li-ion batteries. Recently, the incorporation of transition metal oxide particles into conductive, redox-active polymers has been proposed (Shouji, et al., 1999). Examples of the resultant materials comprise of PPy/MnO2 composite films (Gemeay, et al., 1995; Kuwabata, et al., 1994) and LiMn2O4 nanotubes coated with PPy (Kuwabata, et al., 1996). Li-insertion studies have shown that these interesting composite materials act as conducting networks as well as active materials which ultimately enhance the electrochemical response. In fact, these represent examples of true organic/inorganic hybrids in "nano" scale as they basically comprise conductive organic polymer chains interleaved between the atomic sheet structure of a layered inorganic oxide or sulfide lattice. In, 1997, electrochemical insertion of Li into a series of "nanocomposites" composed of alternating V2O5 sheets and conductive polymer layers such as polypyrrole (PPy) and polyaniline (PAN1) (Leroux, et al., 1997) was carried out. It was found that in comparison to the pristine V2O5 electrode material, the incorporation of polymer enhanced the reversibility of Li insertion and thus increased the Li capacity in the nanocomposite. Polypyrrole is an electronically conductive polymer and the support of PPy on high surface V2O5 aerogels (ARG) composites (Wong, et al., 1998) can enhance performance. Most importantly, the underlying electrochemical response of these nanocomposites is synergistic in nature, i.e., they show improved performance greater than the sum of the two components (inorganic and organic). It must also be noted that the physical properties of these "nanocomposites" are very different from the "microcomposites" prepared by an alternative route, in which the oxide gel is formed in the presence of preformed micrometer-sized polypyrrole particles.
The choice in the use of carbonaceous materials for anodes matches the current trend towards high performance cathodes such as LiCoO2, LiNO2 and LiMn2O4 (Nishizawa, et al., 1997). The carbon (graphite, petroleum coke, etc.) electrode offers good cyclability owing to the formation of a e-/Li+ conducting passivation layer in organic electrolyte solutions. It is common among many types of carbon electrodes used in batteries (Flandrois and Simon, 1999) that most of their structurelie in the nanoscopic scale. Normally, a composite electrode such as graphite + acetylene black is used. In hard carbons, it is believed that lithium can be adsorbed onto the internal surfaces of nanopores, which are formed by the layers of graphene sheets arranged like a "house of cards". Since safety measurements have led to replacement of pure lithium by
61
heavier carbonaceous materials with the general formula LixC6 as the anodes for lithium batteries, any loss in energy density due to this replacement has to be minimized (Peled, et al., 1996). Many groups have tried to search for materials with larger specific capacities (Yata, et al., 1994; Sato, et al., 1994; Sonobe, et al., 1994a, 1994b; Zheng, et al., 1995a, 1995b, 1995c) than those currently employed as anodes in commercially available balleries. Carbon is a very versatile element; the nature of chemical bonds between carbon atoms themselves or with other elements are varied. New forms of carbon can appear, like fullerenes and nanotubes (Flandrois and Simon, 1999). Lastly, in an effort to make the carbon anode as inexpensive as possible, coals as starting materials have been proposed by Zheng et al. (Zheng, et al., 1996).
Among the various forms of carbon, carbon nanobeads are becoming increasingly important as anodes and catalysts, providing a large surface area for chemical reactions to occur (Megahed and Scrosati, 1995). The extent to which lithium ions can be interled between two consecutive layers of graphite is limited by the repulsive forces created between the positively charged lithium ions. However, if they can be inserted individually into nanosized hollow beads, such repulsive forces among the ions would be avoided. The concentration of lithium ions which can be inserted into the electrode would thus depend on the number of carbon nanobeads per unit area of the electrode. Spherical spongy nanobeads as small as 250 nm have been prepared from camphor and the average surface area determined to be 16 m2/g. It was suggested that this material could be very useful anodes for anodes in secondary lithium batteries. The discovery of carbon nanotubes (Iijima, 1991) led to their applications as support matrices for the preparation of one-dimensional, nanoscale, composite structures such as carbon nanotubes coated with nickel (Li, 1997). Strictly, a 1-D metal cannot exist since the Jahn-Teller (Peierls) distortion would give rise to a periodic change in bond length and a band gap. Graphite, as a 2-D conductor, is metallic, but the conductivity is poor compared with metals. Nanotubes (Calvert, 1997) are expected to fall between the band gaps of 0 to 3 eV, depending on their helical structures and tube diameters. The requirement for a battery electrode is that it must intercalate a cation (Li+) with the release or uptake of an electron rapidly. The large surface area of a mat of nanotubes is favorable to kinetics, but the closed nature of the graphitic sheets may inhibit the process. Since the carbon nanotubes can provide pathways for Li diffusion into the crystal lattice of CuO copper oxide/carbon (CuO/C) composite nanotubes have been prepared (Wu, et al., 1998) for use as cathodes. Studies of disordered carbons containing nanodispersed silicon have also been presented by various groups (Wang, et al., 1998) showing that Si atoms or clusters of Si atoms can be embodied into these poorly stacked regions of the carbon structures for materials to reach working capacities near 500 mAh/g. Besides, a high volumetric energy density is expected because silicon crystals can be intercalated to a maximum of 4.4 lithium atoms per silicon atom. Furthermore, the crumbling rate of alloying materials due to the large volume expansion during lithium insertion and brittleness of the intermetallic Zintl phase (Yang, et al., 1996) is greatly reduced if the silicon host particles are nanometers in size. However, at higher operating temperatures, the passivation layer of the carbon decomposes leading to possible battery failure. Alternative host materials such as nanocrystalline TiO2 (anatase) have also been proposed and studied by Huang (Huang, et al., 1995), and investigations showed that they could be promising battery materials for low drain devices.
Recently, however, several tin-oxide-based compounds have been reported (Idota, et al., 1997; Courtney and Dahn, 1997a, 1997b; Brousse, et al., 1997, 1998; Liu, et al. 1998) to be good candidates for anodes in lithium-ion batteries instead of the carbonaceous electrodes. Among these patent and literature reports, Idota et al. 1997) has described how composite glasses based on tin-oxides can be used as high capacity anodes in Li-ion cells. The key reasons for the success are the finely dispersed tin regions, primarily responsible for the reversible reaction with Li, and also the Li2O plus other oxides playing the role of a "matrixglue"
62
holding the particles together. Unfortunately, the Sn oxide-based anode has a large irreversible capacity resulting from the Li reacting with the oxygen bonded to Sn to form Li2O. In order to reduce the irreversible capacity, oxygen bonded to tin should be avoided in the material. Claimed to possess both greater volumetric and gravimetric capacity over the graphite anode (Retoux, et al., 1999), these tin oxides are also stable in air and standard electrolytes. The electrochemical reaction occurring in these compounds is not the intercalation of lithium into the host structure, but follows a two-step reaction process of Li with the tin oxide-based composite glass. First, the Sn oxide in the composite reacts with Li to form amorphous Li2O and metallic Sn. Depending on the composition of the tin oxide-based composite, the sizes of the Sn grains can be very small. Further reaction then occurs between the newly created metallic Sn with Li and leads to the formation of Li-Sn alloys. Thus the reversible reaction in the anode is essentially the alloying and dealloying of Li with the very small grains of tin. Unlike intercalation, which is structurally benign (Nagaura and Tozawa, 1990; Brandt, 1994), the crystal structures of bulk Li-Sn alloy phases have different lattice constants and different volumes per tin atom. Whenever different bulk alloy phases coexist in the same grains, structural changes such as fracture of the material, loss of contact and poor capacity retention as a function of cycle number (Boukamp, et al., 1981; Wang, et al., 1986) are bound to occur. However, if very small grains of tin (in nanometer sizes) are formed, then the bulk phases of Li-Sn alloys do not occur (Cock, 1993) and the Li: Sn ratio in the cluster can vary continuously, so that the Li-Sn clusters and the composite glass matrix in which these reside are not broken up. Good capacity retention during the charge-discharge cycles can then result. The Sn-Fe-C ternary alloy system has been recently investigated (Mao, et al., 1999a, 1996) by employing mechanical alloying techniques (Cock, 1993), where elemental powders of Sn, Fe and, C are mixed mechanically in high-impact ball mills to give mixtures of coexisting intermetallic phases with very small grain sizes. Nanometer-sized grains of an alloy which would react with lithium (forming the "active" phase) are dispersed in an electrically conductive matrix of nanometer-sized grains of another phase which cannot react with lithium (the "inactive" phase). The inactive grains, hence, serve to act as a matrix to hold the active grains as they repeatedly alloy with lithium during operation.
11.2.3.2 Electrolyte
Today's Li-ion batteries use a polyethylene-polypropylene (PE-PP) film structure to separate the electrodes. However, future versions will focus on polymeric films which will display better stability at higher voltages or temperatures such as polyvinylidene fluoride (PVDF), polyacrylonitrile and polyethylene oxide (Mason, et al., 1999). The correct choice of a polymer matrix is very important in the cell's manufacturability and its longevity under strenuous operating conditions. To refine the morphological and electrochemical properties of polymer electrolytes, addition of ceramic fillers has been suggested. Up to date, various highly conducting ceramic fillers, zeolites, ionites, etc., have been investigated (Nazar, et al., 1995). The addition of ceramics like alumina (Al2O3) (Croce, et al., 1998) facilitates an improvement in conductivity of poly (ethylene) oxide (PEO)-based electrolytes (Lightfoot, 1993) as well as their interfacial behavior (Croce and Scrosati, 1993; Borghini, et al., 1995; Capuano, et al., 1996) when in contact with the lithium electrode. Nanometer-sized ceramic powders can act as plasticizers for PEO and kinetically inhibit crystallization upon annealing from the amorphous state. Excellent mechanical stability (Weston and Steele, 1982) promoted by the network of the fillers into the polymer bulk and high ionic conductivity promoted by the large surface areas of the well-dispersed fillers make the composite electrolyte unique. In all cases, particle size and content of ceramic appear as critical factors. It is also clear that the mechanical and electrochemical properties of electrolytes will improve as the grain size of ceramic particles used is descreased, with the
63
main contributing factor being that the conductivity is closely linked with the processes occurring at the grain boundaries. In these polymeric nanomaterials, the particles added are so fine that a large volume of the solid mixture consists of grain boundaries. Since polymer nanocomposites that consist of just a few percent of ceramic particles can behave like other conventional microcomposites that have 60%—70% ceramics in them, a reduction of the ceramic powder particle size from microns to nanometers should lead to further increase in the conductivity as well as reduced weight. Furthermore, Giannelis and co-workers (Giannelis, 1992; Vaia, et al., 1993) created a new type of layered nanocomposite polymer electrolytes of single conductivity. In their work based on mica-type ceramics, the isomorphous substitution of silicon by aluminium led to immobilized and highly delocalized charge balanced by lithium ions present in the nanosized galleries infused with PEO polymer chains.
The majority of the publications on solid-state battery electrolytes are based on the polymeric electrolytes used. Instability of these electrolytes caused mainly by high temperature or voltage may be detrimental to the operation of the battery. Thus, ceramic electrolytes have also been identified, such as Li-doped BPO4. It is well known that undoped BPO4 has been determined as the high-cristoballite structure (Schulze, 1934; Kosten and Arnold, 1980; Long, 1975) since the late 1930s. According to this structure, all the B and P ions are tetrahedrally coordinated by O ions and each O ion is shared by two tetrahedra. In, 1996, Kelder and co-workers (Kelder, et al., 1996) demonstrated the use of Li-doped BPO4 as a Li-ion conducting electrolyte for Li batteries, and a maximum in total ionic conductivity was shown at doping levels of 7 mol% lithium. The ionic conductivity mechanisms were proposed to take place via the interstitial Li ions as the defect chemistry of the BPO4-xLi2O structure was studied. Subsequent investigations made by Jak and Kelder (Jak, et al., 1998, 1999) reveal that the total ionic conductivity increased with decreasing grain size. This indicates that the nanostructure of the ceramic electrolyte is vital for use in all-solid-state Li-ion batteries, in which its conductivity is comparable to the state-of-art polymer-based electrolytes.
11.2.4 Most Recent Developments
A prototype Li-ion polymer cell based on a solid-state conductor where electrodes and the electrolyte were embedded in a flexible plastic matrix has been developed by Institute for Silicon Technology(ISIT) (in collaboration with the Christian Albrechts University, Germany). Solid-state materials in the form of pastes are applied to the polymer film and the electrolyte made thin enough to compensate for the reduced ionic conductivity of polymer compared to the liquid electrolytes. Currently, a copolymer gel electrolyte based on polyacrylonitrile (PAN) for the development of the Li-ion-P battery is being studied by Sony Corp. The challenging concept in secondary lithium batteries lies in the removal of the dendrite regrowth problem. This has led to many prototypes of carbon-based lithium ion batteries developed since 1990 (specific energy densities as high as 340 Wh/L) Since reversible intercalation, or insertion of Lithium into carbon host lattices avoids the problem of lithium dendrite formation.
11.3 Fuel Cells and Nanomaterials
64
11.3.1 Classifications of Fuel Cell Systems
Fuel cells can take on several different configurations, usually determined by the various combinations of the type of fuel and oxidant, the way fuel is fed to the system (direct or indirect), the type of electrolyte used, the operating temperature, etc. In general, we have the following classification shown in Table 11.3.
Table 11.3 Classification of fuel cells
Fuel Oxidant Temperature Electrolyte
Direct Indirect
Hydrogen
Hydrazine
Ammonia
Hydrocarbon
Methanol
Natural Gas
Coal
Hydride
Ethanol
Ammonia
Hydrocarbon
Methanol
Coal
Oxygen
Oxygen(air)
Hydrogen
Peroxide
Low (120°C)
Intermediate
(120–160°C)
High (260–750°C)
Very high (≥750°C)
Aqueous acid
1. Phosphoric 2. Sulphuric
3. Solid polymer
electrolyte (SPE)
Aqueous alkaline
Molten carbonate
Solid cxide
11.3.1.1 Acid and Proton Exchange Membrane Fuel Cell (PEM-FC)
The acid fuel cell can be characterized by:
1. Hydrogen ions [or by hydronium ions (H3O+)] providing the passage of ionic conduction; 2. Current collectors and gas separators are made of carbon (graphite); 3. Platinum or platinum alloys (in very small quantity) as the active electrocatalysts.
There are essentially two types of acid fuel cells, namely the solid polymer electrolyte system and the phosphoric acid cell. The solid polymer electrolyte (SPE) cell uses an ion exchange membrane as the electrolyte and is thus called a proton exchange membrane fuel cell (PEM-FC). These are suited for low-temperature use, but are prone to poisoning by carbon monoxide (Parsons and VanderNoot, 1988) which will inhibit the fuel cell anode reaction. Through the use of different materials, new approaches are being evaluated that would allow future PEM fuel cells to operate with fuel gases containing CO. These techniques involve the construction of new anode electrocatalysts and the control of external prehumidifying of air, thus permitting the working temperature to be raised. Reformation of natural gas or other fuels containing hydrocarbons can also be accomplished within the generator, thus eliminating the need for a separate fuel processor with its ancilliary equipment. This also gives rise to the direct methanol
65
fuel cell (DMFC) in which methanol is reformed to produce H2 and CO2. Methanol is easily transported and converted to energy from the liquid state.
The phosphoric acid electrolyte system is the most mature fuel cell technology and is commercially available now. This type of fuel cell uses liquid phosphoric acid as the electrolyte with hydrogen gas ionized at the anode to form hydrogen ions and electrons. The electrons travel to the cathode via an external circuit while the hydrogen ions travel to the cathode through the electrolyte. The phosphoric acid electrolyte system operates at 150°C to 220°C since at lower temperatures phosphoric acid is a poor ionic conductor, while at too high temperature material stability of carbon and platinum becomes limiting. The cathode performance is usually sluggish, thus pushing major technology in this area towards the improvement of cathode materials. Successfully operating in Japan for a number of years, the current design possesses an electrical conversion efficiency of 41%.
11.3.1.2 Solid Oxide Fuel Cells (SOFC)
As the name implies, solid oxide electrolyte fuel cells utilize a nonporous solid oxide, usually doped with zirconia, as the electrolyte. It has a very high operating temperature (1000°C) at which the lattice structure of the electrolyte material becomes sufficiently conductive to oxide ions. Oxygen ions from an air electrode (cathode) migrate through the solid electrolyte to a fuel electrode (anode). There, they react with carbon monoxide (CO) and hydrogen (H2) contained in the fuel gas to deliver electrons and hence generate electricity. Several features of SOFC technology make it very attractive for industrial and utility applications; one being high tolerance to fuel contaminants, the other being the ability to work together with steam turbine generators. Because of the high operational temperature, the system does not need costly catalysts and allows direct processing in the fuel cells. The solid oxide electrolyte is very stable. Without the presence of liquid phases in the electrolyte, many of the problems associated with electrode flooding, electrolyte migration and catalyst wetting are avoided.
11.3.1.3 Molten Carbonate Fuel Cells (MCFC)
Molten carbonate fuel cells are a type of direct fuel cell that eliminates external fuel processors. Steam and methane (the main ingredient of natural gas) are converted into a hydrogen-rich gas in the reforming anode or in a reforming chamber, which comprises part of the fuel cell stack. The fuel cell stack is made up of two porous electrodes in contact with a molten salt of lithium-potassium carbonate (LiKCO3), operating at approximately 650°C. At the cathode, oxygen (O2) and carbon dioxide (CO2) are converted into carbonate ions. The electrolyte allows carbonate ions to migrate to the anode. At the anode, hydrogen will react with carbonate ions to form water and CO2, releasing two electrons. By completing the flow of electrons through an external circuit DC electricity is generated. Molten carbonate fuel cell plants can achieve an electric efficiency of 50%.
11.3.1.4 Alkaline Fuel Cells (AFC)
66
The early alkaline fuel cells operated at relatively high temperature (≈250°C) with concentrated (85 wt%) potassium hydroxide, but the use of less concentrated (35 to 50 wt%) potassium hydroxide in recent alkaline fuel cells leads to a much lower operating temperature (<120°C). The O2 reduction process in alkaline electrolytes is more favourable than in acid electrolytes. Due its alkaline nature, no acidic impurities, e.g., carbon dioxide, are tolerated in either of the reactants. In the alkaline medium, the presence of carbon dioxide results in the formation of carbonates which would block electrolyte pathways and electrode pores. This is very detrimental to the stability of the alkaline fuel cells. In practice, the alkaline systems usually feed on highly purified hydrogen gas from electrolysis or ammonia plants. Unlike acidic cells, the choice of catalysts used in alkaline systems is not solely limited to the platinum and tungsten carbides group.
11.3.2 Major Components and Nanomaterials in Fuel Cells
Three important components of fuel cells are the electrodes, electrolytes, and fuel supply/storage systems. The electrodes, comprising both the anode (fuel electrode) and the cathode (oxygen electrode), must provide the common interfaces for the electrolyte with the fuel and oxidant respectively. At the anode, the fuel-electrolyte interface must provide good catalysis of the fuel oxidation reaction and conduct electrons from the reaction site to the external circuit (or to a current collector that, in turn, directs the electrons to the external circuit). At the cathode, the oxidant-electrolyte interface must catalyze the oxidant reduction, and conduct electrons from the external circuit to the oxygen electrode reaction site.
The electrolyte must act as a medium of transport for the ionic species involved in the fuel and oxidant electrode reactions while restricting the flow of electrons (conduction of electrons in the electrolyte causes a short circuit). In practical cells, gas separation is also provided by the electrolyte system. This is accomplished by retaining the electrolyte in the pores of a matrix. The matrix may then act to separate the gases through the presence of capillary forces of the electrolyte within the pores.
The material compatibility imposes a primary constraint (Jensen, 1982) on the construction of fuel cells. It is clear that the use of ultrafine catalytically active material dispersed in an appropriate electrode matrix is required to achieve the maximum current density per unit of projected electrode surface area. The application of nanosized materials is, therefore, an important aspect in fuel cells. Being a necessity in the working of an automotive vehicle, nanosized platinum particulates have been shown to offer substantial advantages concerning activity and lifetime in electrocatalysis.
Today, the state-of-art technology of fuel cells in electric vehicles (EV) involves the incorporation of platinum loadings of no more than 3 µg/cm2, which can only be achieved by Pt metal clusters in the nanometers range. Fuel cells, in the power range of up to 200 W, can be a winning choice of alternative next to batteries for long-term usage, especially with potential reduction in the mass and cost. Easily handled and readily oxidized fuels such as methanol, hydrazine, ammonia, etc., have been used in most fuel cells. The direct-type fuel cells, in which fuel can be introduced without the conversion to hydrogen, are presently attractive options for small fuel-cell systems. Methanol and hydrazine are the main liquid fuels used. Ammonia may also be used in direct fuel cells because the oxidation of ammonia to nitrogen and water occurs readily on noble-metal catalysts. However, the majority of current portable fuel cell developments use a metal hydride as the source of hydrogen fuel. Metal hydrides are popular since they are able to store large amounts of hydrogen in a higher energy density (total equivalents of hydrogen per total mass of
67
hydrogen source and container) than hydrogen stored in pressurized or liquefied form. In the following, the status of fuel cell technology with reference to nanomaterials is listed.
11.3.3 Applications of Nanomaterials in Fuel Cells
11.3.3.1 Electrodes
From the space fuel cell by United Technologies Inc., subsequent technical improvement of fuel cell electrodes was only possible because gas diffusion electrodes were developed with dedicated electrocatalysts having an efficiency of about 60% as compared with the theoretical efficiency of 80% (Wendt, 1994). Previously, high platinum loadings (typically 4 mg/cm2 of platinum) were needed for useful rates of hydrogen oxidation and oxygen-reduction reactions, but this problem was solved by the use of supported platinum catalysts similar to those in liquid-electrolyte fuel cells (2–3 nm diameter Pt particles (Dalmia, et al., 1998) dispersed on the surface of fine carbon electrodes). These low-temperature fuel cell electrodes of the alkaline, solid membrane and phosphoric acid type are composed of the granular, nanosized electrocatalysts, and polytetrafluoroethylene (PTFE) as a hydrophobic binder. Initially, highly porous metals (e.g., Raney nickel for hydrogen anodes, Raney silver for oxygen cathodes in alkaline cells) were employed, but recently, the trend is certainly to use the platinized carbon exclusively as cathodic and anodic catalyst for any type of low-temperature cells. The carbon electrodes consist of soot agglomerates 100 to 500 nm in diameter while Pt or Pt alloys are uniformly dispersed on the internal surfaces of these soot particles (Faubert, et al., 1998). By the use of nanosized platinum metal clusters, the effective surface area will be greatly increased. In the polymer electrolyte fuel cell, the catalyst at the electrodes must have access to the gas and be in contact with both the electrical and protonic conductors simultaneously. A certain procedure of creating effective contact with the protonic conductor is by impregnating the supported-catalyst electrode with the protonic conducting materials, usually achieved by covering the surface with a solution of solubilized membrane material such as Nafion (Yeo, 1983). The primary role of the carbon support is to provide electrical connection between the widely dispersed Pt catalyst particles and the porous current collector (carbon paper or cloth). However, since it does not conduct protons and is impermeable to gases, the incorporation of carbon into the catalyst layer restricts the flow of oxygen, water and proton transport and limits the overall performance of the fuel cell. The ideal catalyst support should be permeable to gas and water as well as capable of conducting protons and electrons. Hence, such a material is liable to fully replace the carbon support and Nafion in the catalyst layer (Appleby, 1996) and achieve enhanced performance and weight advantage. Qi et al. proposed the use of conducting polymer/proton exchange polymer composites (Qi, et al., 1998), such as polypyrrole/polystyrene sulphonate (PPy/PSS), which have been shown to exhibit high electron and proton conductivities. In their work, they also produced large quantities of catalyzed polymer particles which could be incorporated into standard PEM fuel cell electrodes.
In this area of research, focus is essentially on either reducing the Pt loading or replacing it completely by the use of non-noble metals (Scherson, et al., 1983, 1986; Wiesener, 1986; van Veen, et al., 1988; Franke, et al., 1989; Tarasevich and Radyushkina, 1989; Savy, et al., 1990; Widelöv and Larsson, 1992; Biloul, et al., 1992). In the early 1990s, experiments were carried out to replace Pt at the cathode by nanoparticle catalysts (Srinivasan, et al., 1991; Gottesfeld and Wilson, 1992a, 1992b; Taylor, et al., 1992) based on organometallic
68
compounds absorbed on carbon black (Dodelet, et al., 1994) and pyrolyzed at various temperatures. In 1997, Dodelet et al. used carbon-coated fcc cobalt nanocrystallites (Dodelet, et al., 1997) produced by the carbon arc method as the catalysts for electroreduction of oxygen in PEM-FC and activated this material containing 17.0 wt% Co by a pyrolysis step at 1000°C. They found that catalysts obtained by the pyrolysis of the organometallic precursors at high temperatures (≥800°C) were very efficient and stable at the same time.
11.3.3.2 Hydrogen Storage
In order to evaluate the possibilities of achieving hydrogen utilization in a fuel cell (Appleby and Foulkes, 1989), scientific problems involving its storage must be addressed. Hydrogen may be stored as a gas, liquid or in easily dissociated compounds such as metal hydrides. The storage of hydrogen (Riley, 1994) as a gas requires large volumes of space and even under compression, the volume efficiencies cannot match those of liquid hydrogen. Coupled with the extra mass from the storage cylinder, gaseous storage becomes a major disadvantage. The volume efficiency can be improved by liquefied hydrogen, yet the energy required for liquefaction and the need for excellent insulation of the hydrogen containers to prevent loss of H2 over time become great disadvantages. At this end, the storage of hydrogen as metal hydrides becomes one of the most promising alternatives because of its unique feature. The amount of hydrogen that can be stored with respect to volume is much more efficient than liquid and gas. Moreover, metal hydrides have a better reversibility of the formation reaction compared to water (which forms H2 and O2 by electrolysis).
The formation of the hydride is an exothermic, usually spontaneous, reaction, while the reverse hydrogen recovery can also be easily achieved by heating the hydride. Therefore, since hydrogen recovery is endothermic, any leakage of hydrogen will be suppressed through self-cooling. The storage hydrides are customarily safe and stable when under their dissociation temperatures. In the early years, the metal hydrides that received the most attention and showed the highest potential applications were FeTi and LaNi5 (Busch, et al., 1978) hydrides. Still one of the well-known hydrogen storage materials currently, FeTi is an intermetallic Lave phase material. However, its activation is rather difficult and research work is required to make the activation process amenable. Recently, it was reported that catalytic decomposition of hydrogen in FeTi was possible, enabling the activation temperature to be reduced considerably (Singh, et al., 1998a). This was accomplished through the addition of zirconia corresponding to the formation of Fe0.8 Zr0.2 Ti1.3. New hydrogen storage materials for improved hydrogenation characteristics, such as the synthesis of Mg2 Ni (Zaluski, et al., 1995; Keskinen and Ruuskanen, 1996), was accomplished by ball-milling of a correct stoichiometric mixture of magnesium and nickel. The resultant nanoparticles yielded a higher hydrogen storage capacity due to the increased surface area corresponding to smaller particle sizes. The nanocrystalline structure is the key factor for achieving such a remarkable improvement in the hydriding of Mg2 Ni.
Today, due to the large number of available materials and the different requirements for a wide variety of applications in fuel cells, the challenge is to select the most appropriate metal hydride or some other material and to design the best storage vessel for a specific application. An ideal solid for hydrogen storage would be a structure consisting of slit-shaped nanopores that have a width slightly higher than the kinetic
69
diameter of hydrogen, 0.3 nm. As such, metal-intercalated graphite has been considered as a possible candidate for hydrogen storage. In recent years, graphite nanofibers (GNF), a novel type of carbon material, have been intensely studied for hydrogen storage purposes (Rodriguez, et al., 1995; Dillon, et al., 1997; Chambers, et al., 1998; Che, et al., 1998). GNF are produced from the dissociation of carbon-containing gases over selected metal surfaces. The nanofibers consist of very small graphite platelets, being 0.3 to 50 nm in width and are stacked in a perfectly arranged conformation as shown in Fig. 11.7. This unique conformation bestows upon this material the excellent qualities which are very much desired for gas sorption applications. Research has indicated that this material is capable of sorbing and retaining in excess of 20 L of hydrogen per gram of carbon when the nanofibers are exposed to the gas at pressures of 120 atm (1 atm=1.01 × 105 Pa) at 25°C. This storage volume is of a magnitude higher than that with the conventional hydrogen storage systems. Due to the unique crystalline arrangement existing within the graphite nanofiber structure, where platelets generate a system comprised of entirely slit-shaped nanopores and the short diffusion path, the sorption of molecular hydrogen in the interplanar distance of 0.337 nm is facile. In addition, the weak van der Waals bondings of the platelets allow the nonrigid wall nanopores to expand to accommodate molecular hydrogen in a multilayered configuration.
Figure 11.7 (a) Schematic representation of the arrangement of platelets in catalytically grown graphite nanofibers. (b) An enlarged section showing the details of area marked in (a).
11.3.3.3 Reforming
In addition to the practice of fuel cells feeding on pure hydrogen, reformed gases from methanol or natural gas may also be used. Reforming to produce hydrogen gas from natural gas, or in fact, any other carbon-containing fuel creates the possibility of separating carbon dioxide and either reusing or disposing of it. The direct methanol fuel cell (DMFC) (Scott, et al., 1998) uses methanol as fuel and is mainly based on a solid polymer electrolyte. The PEMFC functions at relatively low temperature, e.g., 60–80°C. The anode catalyst is very susceptible to poisoning by the carbon monoxide (CO), which may remain in the reformed gases up to 100 parts per million (ppm). Therefore, it is necessary to develop a simple system which can removed CO to less than 100 ppm (Gottesfeld and Pafford, 1988; Gottesfeld, 1990) in large excess of H2.
70
Oxidative elimination of CO was proposed using Pt catalyst supported on alumina (Pt/Al2O2-3) in the
presence of 2% oxygen or air (Cohn, 1965; Bonacci, 1980; Vanderborgh, et al., 1987, 1988). However, since the O2 content present is close to the lower limit for explosion of mixed gases, the hazard of accidental explosion cannot be ignored. Hence, catalysts comprised of highly dispersed metals such as platinum in A-zeolite were reported by Watanabe and coworkers (Watanabe, 1995). Such a catalyst was able to oxidize CO selectively in reformed gases. The synthesized catalyst consisted of Pt nanocrystallites dispersed on A-zeolites and showed exceptionally high selectively for the oxidative elimination of CO present in the 1% mixed gases with an excess of H2; i.e., 10 times larger than the selectivity of the conventional Pt/Al2O3 reported. Since the problem lies in the poisoning of the catalyst surface by carbon monoxide, in order to avoid this, composite oxides as catalysts have been reported (Zhao, et al., 1999). It was found that composite oxides with the perovskite structure are good catalysts for complete oxidation of methanol because electrons on the d-orbitals are always available for transit and are in the high-spin state. Therefore, nanocrystalline composite oxide SrRuO3 (with a perovskite structure) can partially substitute platinum as a catalyst in methanol oxidation with improved features such as better performance and lower cost.
Other research groups have suggested that alloys of Pt with more oxophilic elements (Koch, 1964) such as Pt-Ru oxide blacks (Gurau, et al., 1998; Ross, 1992; Freelink, et al., 1994, 1995) can be active for direct methanol oxidation. These nanoscale Pt-Ru blacks are widely accepted to be bimetallic alloys based on their X-ray diffraction patterns. However, bulk and surface analyses (Rolison, et al., 1999) recently suggest that these are primarily a mix of Pt metal and the oxides of Ru, plus some Pt oxides with little Ru metal. X-ray photoelectron spectroscopy (XPS) and Mössbauer studies have been conducted and the substantial amounts of hydrous ruthenium oxide, RuO2 · xH2 O or RuOx Hy have been detected in the lilevature. Due to their amorphous structure, they cannot be discerned by X-ray differaction. However, of key importance to the mechanism of methanol oxidation is that hydrous ruthenium oxide is a mixed proton-electron conductor. The presence of the hydrous oxide with RuOx Hy rather than Ru metal or anhydrous RuO2 proves to be an efficient catalyst for oxidation of alcohols and more chemically durable under corrosive conditions.
11.3.3.4 Proton Exchange Membranes
The proton exchange membrane fuel cell utilizes an ion exchange electrolyte which is an excellent conductor of protons, yet is an electronic insulator. Examples of the early membranes tested include the hydrocarbon-type polymers such as crosslinked polystyrene-divinylbenzene sulphonic acids and sulphonated phenolformaldehyde. However, the hydrocarbon-type polymers are easily subjected to oxidation due to the C-H bond cleavage, especially at the reactive α-H sites where the functional groups are attached. Thus, the polystyrene-based sulphonic acids are replaced with fluorine substituted polystyrenes (e.g., polytrifluorostyrene sulphonic acid). Today, the world's leading perfluorosulphonic polymer electrolyte is Nafion produced in 1962 by E. I. Dupont de Nemours. It is commercially available in various forms: homogeneous or reinforced membranes, powders, tubes and solutions. Several studies on the structure and properties of Nafion have been reported in the past two decades (Pineri, et al., 1982; Yeo and Eisenberg, 1977); however, the configuration and geometry of the nanosized ion aggregates within the conducting matrix are still being debated. To say the least, this material consists of a fluorocarbon polymer backbone, similar to Teflon PTFE (polytetrafluoroethylene) to which the sulphonic acid groups have been chemically bonded. The acid molecules are anchored to the polymer and cannot be leached out, but the protons on these acid groups are free to migrate through the electrolyte, showing high conductivity (see Figs.
71
11.8 and 11.9). The Nafion membrane is very stable and strong, has an equivalent weight of 1100 and when saturated contains 25 molecules of water per SO2-
3. At least four grades of Nafion are currently available: 117, 115, 112 and 105 (Kordesch and Simader, 1996). The initial two digits indicate equivalent mass, e.g., 1100. The last digit represents the dry thickness in mils (e.g., 7 mils or 175–180 µm). Similar ion conducting membranes are available from Dow Chemical Company (m = 0, n = 2, equivalent mass 800, normally 125 µm thick) with m and n as denoted in Fig. 11.8. Another homologous series, Aciplex-S from Asahi Chemical Industry Company, in which m is 0–2 and n is 2–5, is also known. These materials are synthesized through slightly different chemistry and possess slightly different properties having equivalent mass of 1100, 800 and 1000.
Figure 11.8 Structure of perfluorocarbon ion exchange polymers.
Figure 11.9 Schematic representation of ion clustering in Nafion
72
One key technological problem in PEM-FC is the water content management and reactant crossover problem (Verbrugge and Hill, 1990; Zawodzinski, et al., 1993; Fuller and Newman, 1992). During operation, polymer electrolyte membranes such as Nafion require water to maintain their protonic conductivities. However, the water content is changed in an extremely complicated manner with the operating conditions and further discussion is beyond the scope of this chapter. In practical use, the water content in PEMs can be indirectly controlled by humidifying either the fuel gas or the fuel and oxygen. By reducing the polymer electrolyte thickness, water management problems due to the water back-diffusion from the cathode (Springer, et al., 1991) are distinctively reduced, due to a decrease ohmic potential drop in the cell while improving cathode performance (Watanabe, et al., 1994a). However, this usually accelerates the crossover of H2 and O2 through the thin electrolyte membrane and leads to lower cell performance and fuel utilization. Attempting to address this problem, new polymer electrolyte membranes with highly dispersed nanometer-size Pt and/or metal oxides such as titanium dioxide powders (Watanabe, et al., 1994b, 1995a) have been proposed. In 1998, a design concept for the Pt and metal oxides dispersed PEM (Watanabe, et al., 1998) shown in Fig. 11.10 was proposed. Platinum nanocrystals (d is 1–2 nm) were highly dispersed in a Nafion 112 film (Pt-PEM, thickness 50 µm) to catalyze the recombination of the crossover H2 with O2, and the water generated was found to humidify the Pt-PEM direct fuel cell. The Pt particles were expected to inhibit the crossover of the gases by the catalytic recombination of O2 and H2, while the oxide particles (with hygroscopic property) were expected to adsorb the water produced at the cathode reaction, releasing it only when required. In this work, the self-humidification behaviour of Pt-PEM, TiO2-PEM, and Pt-TiO2-PEM prepared by the dispersion of small quantities of Pt particulates (1–2 nm dia.) and/or TiO2 particles (5 nm dia.) in Nafion 112 recasted film (50 µm thick) was investigated. In particular, the Pt-TiO2-PEMs show that the crossover H2 and O2 was able to recombine on Pt particles and all the moisture generated inside the PEMs was exhausted from the anode. Being hygroscopic in nature, the TiO2 particles enhanced the back-diffusion of water produced by faradaic reaction at the cathode and resulted in very efficient
73
humidification of the PEM of the anode side dried by the electro-osmotic drag. Overall, the novel PEMs were found to improve the cathodic potential and demonstrated superior performance in which the suppression of the crossover of gases under the unhumidified environment was comparable to that of the conventional (humidified) case. This also suggested the elimination of any short-circuit reaction of the crossover gases in the cathode catalyst layer and thus the non-faradaic consumption of H2 was reduced without any disturbance of O2 diffusion by the produced water vapor.
Figure 11.10 Schematic operation concept of the PEM-FC using self-humidifying Pt-oxide-PEM.
In recent times, as an alternative to ionic conducting polymeric electrolytes, a bicontinuous structure made by the microemulsion technique has been proposed (Chow, et al., 1999). Microemulsion systems are thermodynamically stable, isotropic assemblies of oil and water, separated by an interfacial film of surfactant molecules (Friberg, 1983; Langevin, 1988; Paul, et al., 1997). Depending on the composition of oil, water and surfactant, the formation of microemulsions can be in the form of water-in-oil droplets or oil-in-water droplets. In addition, there may exist bicontinuous microemulsions in which the oil and aqueous conduits (channels) are randomly interconnected forming sponge-like structures. The oil phase can then be polymerized together with the polymerizable surfactant (Gan, et al., 1995), and the water phase be treated to host ionic species. With the copolymerization of ionic monomers, such as 4-vinylbenzene sulfonate (SVBS), in the polymerizable microemulsions, ionic conducting membranes can be produced. The resulting material provides great potential for use in ion exchange polymers as the bicontinuous structures prepared from the microemulsion techniques contain mobile cations which are counterbalanced by the sulfonic groups anchored to the PEO backbone of the polymer. The most probable pore sizes (in diameter) were found to be less than 8 nm. The pore volume distributions also become narrower with increasing ionic content in the microemulsion system. Similar to Nafion, these nanosized aqueous pores found in the styrene-sulfonate microemulsion membranes are crucial to the ionic transport. The conductivities reached 10-3 S/cm which are much better than conventional PEO (doped with salt) type polymer electrolytes. Due to the nature of the microemulsion technique, many parameters can be fine-tuned in the future improvement of the membrane.
74
For example, the oil phase may be modified to increase the thermal, mechanical and chemical stabilities, whereas the aqueous phase can be modified to host various ions and to remove the frozen state.
New technological applications in electrochemical devices, including electrochromic displays, chemical sensors and fuel cells, suggest demand for high temperature protonic conducting membranes. Organic/inorganic nanocomposite materials which show high ionic conductivities at temperatures in excess of 100°C, are good candicates for high temperature membranes. These solid nanocomposite membranes belong to a remarkable family of isotropic, amorphous materials with excellent thermal properties, electrical conductivities, flexibility and mechanical strength. Synthesized by sol-gel processes, these properties including optical density can be carefully controlled through the adjustment of the compositions, nanophase size and chemical bonding between the organic/inorganic composite membranes. A specific example is the organic/inorganic nanocomposite membrane consisting of SiO2/PEO which showed good protonic conductivities and thermal stabilities up to 160°C. It was synthesized via a sol-gel method (Honma, et al., 1999). The composite membrane was obtained by hydrolysis and a condensation reaction of polymer precursors consisting of polyethylene oxides endcapped with trioxysilane and further subjected to doping by acidic surfactant molecules MDP. The hybrid membrane had a structure of an interpenetrated network composed of a nanosized mixture of silica (silsesquioxane, Si2O3) and PEO, with each silica domain restricted to a distance of approximately 3 nm by the chemically bound PEO chain. Because of the presence of inorganic silica as part of the structural elements, the composite membrane showed high temperature tolerance. Thus, the properties of such hybrid membranes can be designed through delicate manipulation of the compositions of the components containing nanosized composite structures among the organic PEO, inorganic silica and functionally doped MDP.
11.3.3.5 Solid Oxide FC Membranes
The task of developing electrolyte membrane for the solid oxide fuel cell (SOFC) (Watanabe, et al., 1997), expected for future electricity generation, will be a demanding one due to the need to simultaneously satisfy electrical, chemical and thermomechanical requirements. Usually, the electrolyte is zirconia (ZrO2) doped with 8 mol%-10 mol% yttria (Y2O3). Pure zirconia is an insulator. However, when doped with yttria, some Y3+ replaces Zr4+ in the fluorite-type crystal structure and results in a number of vacant oxide-ion sites. These sites become available since three O2- replace four O2- when two Zr4+ are replaced by two Y3+ in the lattice structure. As such, the mobile ionic species O2- can move across the electrolyte via the vacant lattice sites at high temperatures and conduct charges. The current carrying O2- species arrive at the electrolyte-electrode interface and react with the gas phase within the porous electrode to generate electricity.
A representative electrolyte material such as 8 mol% Y2O3 stabilized ZrO2 (8YSZ) possesses low electronic and high oxygen ionic conductivity. Although the traditional thin-layer 8YSZ can show excellent electrical properties at 1000°C, this oxide-ion-conducting electrolyte possesses poor mechanical properties, such as low strength, toughness, Young's modulus and thermal conductivity (Minh, 1993; Isaacs, 1981). Hence, to broaden its use at high temperatures, some improvements must be made to 8YSZ. Current studies to improve the mechanical properties of zirconia and other ceramics are directed at using phase transformation as well as composite techniques. The phase transformation of zirconia from tetragonal to monoclinic (Claussen, 1984; Masaki, 1986; Tsukuma, et al., 1988), however, may not be as effective on strengthening
75
and toughening of the material at elevated temperatures (Masaki, et al., 1986). The latter method of strengthening by dispersing particles, whiskers, platelets or fibers into the electrolyte is undoubtedly more effective. In particular, ceramic-based nanocomposites such as the ceramic/ceramic and ceramic/metal composite systems have attracted enormous attention as particle-reinforced composites, in which nanosized second phases are dispersed within the matrix. The random dispersion of the second phase nanosized particulates improves not only the mechanical properties but also functional properties. In fact, attempts have been made to fabricate 8YSZ/silicon carbide (SiC) nanocomposites by a hot-pressing technique (Bamba, et al., 1998). The effects of dispersed nanoparticles of SiC on the microstructures of the material and the resulting ionic conductivities have been investigated. It was found that ultrafine SiC particulates inhibit the densification process and normal/abnormal grain growth of the 8YSZ matrix grains occurred with decreases in grain boundary diffusivity and mobility. As the SiC content was increased, the inhibition phenomenon became more obvious although the nanocomposites still retain a fine and homogeneous microstructure. The total conductivity, however, decreased with increasing SiC content due to the decrease in grain boundary conductivity, which must be related to the increase in grain boundary length and the decrease of effective volume of ionic conductor by SiC particles at the grain boundaries.
Apart from the Y2O3 stabilized ZrO2/SiC nanocompsites, other rare-earth and calcium doped zirconates are also materials of choice for electrolytes used in intermediate temperature solid oxide fuel cells. The dopant, in each case, serves not only to stabilize the cubic structure of zirconia but also to introduce and create anion defects which presumably increase the ionic conductivity. Efforts to develop new oxygen ion conductors with enhanced properties have spurred the development of many potential electrolyte materials for use in solid oxide fuel cells. Alternatives to zirconia, e.g., Bi2O3-and CeO2-based materialss were proposed. In the past, ceria-based (CeO2-x) materials were extensively investigated as catalysts, structural and electronic promoters of heterogeneous catalytic reactions and oxide ion conducting solid-state electrochemical cells. To date, research on ceria solid electrolytes doped with small amounts of Pr or Tb as electron traps to extend the oxygen partial pressure has been carried out (Greenblatt, et al., 1998). Beside, other rare-earth elements such as samarium have also been investigated extensively (Watanabe et al., 1997). Prepared hydrothermally, the average crystallite size of samarium-doped ceria powders (7–68 nm) enables sintering into dense ceramic pellets at 1400°C. Compared with zirconia, the ionic conductivity was high (→ 10-2 S/cm) even at an operational temperature of only 600°C.
11.3.4 Summary
Fuel cells will play a key role in the future world energy scenario. Their most important characteristics, namely high efficiency and extremely low emission and noise levels, will be mandatory in the next generation of power plants. Should hydrogen be the main energy carrier of the 21st century, as many technical studies point out, then fuel cells will have many undisputed advantages over all other energy conversion devices.
11.4 Conclusions
76
The most pressing need for research in energy conversion for the next century will be to meet the different challenges in the ways of generating or storing electric power in a more efficient as well as environmentally benign way (Kartha and Grimes, 1994). Today, new possibilities and opportunities have opened up by timely nanotechnology in the synthesis of materials that can be engineered for the efficient functioning of batteries and fuel cells. Clearly, the ongoing research projects as well as future industrial applications will bring down the costs to manufacture nanophase materials and enhance the importance of nanomaterials in energy storage systems. Inorganic and organic nanocomposites (hybrids) are also great contenders for the next generation battery or fuel cell systems which may offer optimum characteristics. From the use of nanomaterials, it is thus expected that the functional performance of the electrochemical systems will be improved tremendously.
References
Appleby A. J. Materials for Electrochemical Energy Storage and Conversion - Batteries, Capacitors and Fuel Cells. eds. by D. H. Doughty, B. Vyas, T. Takamura, J. R. Huff, Materials Research Society. USA (1995)
Appleby A. J. Philos. Trans. R. Soc. Lond. Ser.. A 354, 1681 (1996)
Appleby A. J. and F. R. Foulkes. Fuel Cell Handbook. Van Nostrand Reinhold, New York, 362 (1989)
Armand M. Materials for Advanced Batteries. eds. by D. W. Murphy, J. Broadhead, and B. C. H. Steele, Plenum, New York, 145 (1980)
Auborn J. J. and Y. L. Barbario. J. Electrochem. Soc.. 134, 638 (1987)
Bach R. O. Lithium, Current Applications in Science, Medicine and Technology. Wiley New York, (1985)
Bamba, N. Y. H. Choa, T. Sekino, K. Niihara. Solid State Ionics. 111, 171 (1998)
Barak M. In: Electrochemical Power Sources: Primary and Secondary Batteries. Peregrinus, New York, (1980)
Benaissa M., M. José-Yacamán, T. D. Xiao, P. R. Strutt. Appl. Phys. Lett.. 70, 2120 (1997)
Berger J. J. Charging Ahead: the Business of Renewable Energy and What It Means for America. Henry Holt & Company, New York, (1997)
Biloul A., O. Contamin., G. Scarbeck, M. Savy, D. van den Ham, J. Riga, and J. J. Vebist. J. Electroanal. Chem.. 335, 163 (1992)
Birke P., F. Salam, S. Döring, and W. Wepper. Solid State Ionics. 118, 149 (1999)
Bonacci J. C., US Patent. 4238468 (1980)
Borghini M. C., M. Mastragostino, S. Passerini, and B. Scrosati, B. J. Electrochem. Soc.. 142, 2118 (1995)
Boukamp B. A., G. C. Lesh, R. A. Huggins. J. Electrochem. Soc.. 128, 725 (1981)
77
Brandt K. Solid State Ionics 69, 173 (1994)
Brousse T., D. Defives, L. Pasquereau, S. M. Lee, U. Herterich, D. M. Schleich. Ionics. 3, 332 (1997)
Brousse T., R. Retoux, U. Herterich, D. M. Schleich. J. Electrochem. Soc.. 145, 1 (1998)
Brown P. J., R. S. Kirk, P. G. Patil. Proc. 23rd Intersoc. Energy Conv. Eng. Conf. Am. Soc. Mech. Eng.. New York, 271 (1988)
Bruce P. G., R. A. Armstrong, and R. L. Gitzendanner. J. Mater. Chem.. 9, 193 (1999)
Buchmann I., In: The Electronics Handbook, ed. by J. C. Whitaker, CRC Press, (1996). Florida, Buchmann, I. Batteries in a Portable World. Cadex Electronics, Canada. (1997)
Busch G., L. Schlapbach, A. Seiler. In: Hydrides for Energy Storage, Proc. Internat. Symp. in Geilo, Norway, 14–19 Aug 1977 eds. by A. E. Andresen, A. J. Maeland, Pergamon Press, New York, (1978)
Calvert, P. In: Carbon Nanotubes. ed. by T. Ebbesen, CRC Press, New York, 286 (1997)
Capuano F., F. Croce, B. Scrosati. US Patent No. 5576115 (1996)
Chambers A., C. Park, R. T. K. Baker, N. M. Rodriguez. J. Phys. Chem. B. 102, 4253 (1998)
Che G., B. B. Lakshmi, E. T. Fisher, C. R. Martin. Nature. 393, 346 (1998)
Chow P. Y., C. H. Chew, C. L. Ong, J. Wang, G. Xu, L. M. Gan. Langmuir. 15, 3202 (1999)
Claussen N. Science and Technology of Zirconia II. In: Advances in Ceramics. Vol 12, eds. by N. Claussen, M. Rühle, A. H. Heuer, American Ceramics Society, Columbus. 325 (1984)
Cock C. C. in Materials Science and Technology 15, VCH Publishing, Weinheim. 193 (1993)
Cohn J. G. E. U. S. Patent 3216783 (1965)
Courtney I. A., J. R. Dahn. J. Electrochem. Soc.. 144, 2045 (1997a)
Courtney I. A. J. R. Dahn. J. Electrochem. Soc.. 144, 2943 (1997b)
Croce F., B. Scrosati. J. Power Sources. 43, 9 (1993)
Croce F., B. Appetecchi, L. Persi, B. Scrosati. Nature. 394, 456 (1998)
Dalmia A., C. L. Lineken, R. F. Savinell. J. Colloid Interf. Sci.. 205, 535 (1998)
Davidson I. J., R. S. McMillan, J. J. Murray, J. E. Greedan. J. Power Sources. 54, 232 (1995)
Davolio G., E. Soragni. J. Appl. Electrochem. 28, 1313 (1998)
DeGuzman R. N., Y. F. Shen, E. J. Neth, S. L. Suib, C. L. O. Young, S. Levine, J. M. Newsan. Chem. Mater.. 6, 815 (1994)
Dillon A. C., K. M. Jones, Bekkedahl, C. H. Kiang, D. S. Bethune, M. J. Heben. Nature. 386, 377 (1997)
78
Dodelet J. P., G. Lalande, D. Guay, L. Dignard-Bailey, M. L. Trudeau, A. Joly, R. Schulz. J. Mater. Res.. 9, 3203 (1994)
Dodelet J. P., G. Lalande, D. Guay. Chem Mater.. 9, 784 (1997)
Dutta J. Analysis. 24, M16 (1996)
Faubert G., D. Guay, J. P. Dodelet. J. Electrochem. Soc.. 145, 2985 (1998)
Feng Q., H. Kanoh, K. Ooi, M. Tani, Y. Nakacho. J. Electrochem. Soc.. 141, L135 (1994)
Ferrando W. A., W. W. Lee. A suspension method for impregnating sintered nickel composite plaque, Proc. 31st Power Sources Symp., Cherry Hill, New Jersey, The Electrochemical Society, New Jersey. (1984)
Flandrois S., B. Simon. Carbon. 37, 165 (1999)
Franke, R., D. Ohms, and K. Wiesener. J. Electroanal. Chem.. 260, 63 (1989)
Freelink T., W. Visscher, J. A. R. van Veen. Electrochim. Acta. 39, 1871 (1994)
Freelink T., W. Visscher, A. P. Cox, J. A. R. van Veen. Electrochim. Acta. 40, 1537 (1995)
Friberg S. E. Prog. Colloid & Polym. Sci.. 68, 41 (1983)
Fuller T. F. and J. Newman. J. Electrochem. Soc.. 139, 1332 (1992)
Gan L. M., T. D. Li, C. H. Chew, W. K. Teo. Langmuir. 11, 3316 (1995)
Gavach C., G. Pourcelly. In: Proton Conductors: Solids, Membranes and Gels-Materials and Devices. ed. by Colomban, P. Cambridge University Press, New York, chap. 32 (1992)
Georgano G. N.. Electric Vehicles. Buckingham Shire Publications, Prince Risborough (1996)
Giannelis E. P. J. Met. March. 28 (1992)
Gemeay A. H., H. Nishiyama, S. Kuwabata, H. Yoneyama. J. ElectroChem. Soc.. 142, 4190 (1995)
Gottesfeld S. and J. Pafford. J. Electrochem Soc.. 135, 2651 (1988)
Gottesfeld. U. S. Patent 4910099 (1990)
Gottesfeld S., M. S. Wilson. J. Appl. Electrochem. 22, 1 (1992a)
Gottesfeld S., M. S. Wilson. J. Electrochem. Soc.. 139, L28 (1992b)
Graetzel M. Analysis 24: 6 M17-M17 (1996)
Greenblatt M., W. Huang, and P. Shuk. In: Mat Res. Symp. Proc.. Vol. 496, eds. by Ginley D. S., D. H. Doughty, B. Scrosati, T. Takamura, Z. Zhang. Materials Research Society, Boston, Massachusetts. 149 (1998)
Gurau B., E. Reddington, S. Sarangapani. J. Phys. Chem.. B 102, 9997 (1998)
Honma I., S. Hirakawa, K. Yamada, J. M. Bae. Solid State Ionics. 118, 29 (1999)
79
Huang S. Y., L. Kavan, I. Exnar, M. Grätzel. J. Electrochem. Soc.. 142, L142 (1995)
Huang B., Y. I. Jang, Y. M. Chiang, D. R. Sadoway. J. Appl. Electrochem. 28, 1365 (1998)
Idota Y., M. Mishima, M. Miyaki, T. Miyasaka. U. S. Patent 5618640 (1997)
Iijima S. Nature. 354, 56 (1991)
Ikoma M., Y. Hoshina, I. Matsumoto. J. Electrochem. Soc.. 143, 1904 (1996)
Isaacs H. S. In: Advances in Ceramics Vol 3, eds. by A. H. Heuer, L. W. Hobbs, American Ceramics Society, Columbus. 406 (1981)
Jak M. J. G., E. M. Kelder, J. Schoonman, N. M. van der Pers, A. J. Weisenburger. Electroceramics 2:2, 127 (1998)
Jak M. J. G., E. M. Kelder, J. Schoonman. J. Solid State Chem.. 142, 74 (1999)
Jensen J. In: Solid State Protonic Conductors I for Fuel Cells and Sensors. eds. by J. Jensen, M. Kleitz, Odense University Press Denmark (1982)
Julien C., G. A. Nazri. Solid State Batteries: Materials Design and Optimization. Kluwer Academic Publishers. London. 609 (1994)
Jung C. B., J. H. Kim, K. S. Lee. J. Alloys Compd. 267, 265 (1998)
Kao W. H., V. J. Weibel, M. J. Root. J. Electrochem. Soc.. 139, 1223 (1992)
Kartha S., P. Grimes. Phys. Today. 47:11, 54 (1994)
Kelder E. M., M. J. G. Jak, F. de Lange, J. Schoonman. Solid State Lonics. 85, 285 (1996)
Keskinen J., P. Ruuskanen. In: Hydrogen Power: Theoretical and Engineering Solutions. ed. by T. O. Saetre, Kluwer Academic Publishers, Netherlands. 377 (1996)
Kim H. S. B. W. Cho, J. T. Kim, K. S. Yun, H. S. Chun. J. Power Sources. 62, 21 (1996)
Koch D. F. A. Australian Patent 46123 (1964)
Kordesch K., M. Weissenbacher. J. Power Sources. 51, 62 (1994)
Kordesch K., G. Simader G. Fuel Cells and Their Applications, VCH Publishers, New York. 86 (1996)
Kosten K., H. Arnold. Z. Kristallogr. 132, 119 (1980)
Kuwabata S., A. Kishimoto, T. Tanaka, H. Yoneyama. J. Electro Chem. Soc.. 141, 10 (1994)
Kuwabata S., K. Mukai, M. Nishizawa, C. R. Martin, H. Yoneyama. The Electrochemical Society Meeting Abstracts 96 (1996): 2, San Antonio, TX, Oct 6–11, Abstract 827. 1015
Langevin D. Acc. Chem. Res.. 21, 255 (1988)
Lazzari M., B. Scrosati. J. Electrochem. Soc.. 127, 773 (1980)
80
Lee W. W. U. S. Patent 4595463 (1986)
Lemons R. A. J Power Sources. 29, 251 (1990)
Leroux F., G. Goward, W. P. Power, L. F. Nazar. J. Electrochem. Soc.. 144, 886 (1997)
Li Q. Jpn. J. Appl. Phys.. Part 2 36, L501 (1997)
Linden D. Handbook of Batteries and Fuel Cells. McGraw-Hill Book Company, New York. 1–9 (1984)
Liu W., X. Huang, Z. Wang, H. Li, L. Chen. J. Electrochem. Soc.. 145, 59 (1998)
Long B. P. Supplement to Mellor's Treatise on Inorganic and TheoreticalChemistry, 5 Section A13. Longman, London. 651 (1975)
Mao O., R. L. Turner, I. A. Courtney, B. D. Fredericksen, M. I. Buckett, L. J. Krause, J. R. Dahn. Electrochem. Solid-State Lett. 2, 3 (1996a)
Mao O., R. A. Dunlap, J. R. Dahn. J. Electrochem. Soc.. 146, 405 (1996b)
Masaki T. J.. Am. Ceram. Soc.. 69, 638 (1986)
Masaki T., K. Nakajima, K. Shinjo. J. Mater. Sci. Lett.. 5, 1115 (1986)
Mason R. N., M. Smith, T. Andrews, D. Teeters. Solid State Ionics. 118, 129 (1999)
McCroy L. R. In: Special Topics in Electrochemistry. ed. by Rock P. A., Elsevier, Amsterdam. chap. 1 (1977)
Megahed S., S. Scrosati. Electrochem. Soc. Interface. 4, 34 (1995)
Minh N. Q. J. Am. Ceram. Soc.. 76, 563 (1993)
Mizushima K., P. C. Jones, P. J. Wiseman, J. B. Goodenough. Mat. Res. Bull. 15, 783 (1980)
Molenda J., W. Kucza. Solid State Ionics. 117, 41 (1999)
Nagaura T., K. Tozawa. Prog. Batteries Sol. Cells. 9, 209 (1990)
Nazar L. F., H. Wu, W. P. Power. J. Mater. Chem.. 5, 1985 (1995)
Nishizawa M., K. Mukai, S. Kuwabata, C. R. Martin, H. Yoneyama. J. Electrochem. Soc.. 144, 1923 (1997)
Numata K., C. Sakaki, S. Yamanaka. Solid State Ionics. 117, 257 (1999)
Ohzuku T., M. Kitagawa, K. Sawai, T. Hirai. J. Electochem. Soc.. 138, 360 (1991)
Ondrey G., C. Crabb, T. Kamiya. Chem Eng. 106:2 47 (1999)
O'Sullivan J. B. Historical Review of Fuel Cell Technology. Proc. 25th Power Sources Symp.. 149 (1972)
Owens B. B., Skarstad P. M., Unterecker, D. F.. In: Handbook of Batteries and Fuel Cells. ed. by D. Linden, McGraw Hill, New York, 12–1 (1984)
81
Parsons R., T. VanderNoot. J. Electroanal. Chem.. 257, 9 (1988)
Paul B. K., S. P. Moulik. J. Dispersion Sci. Technol.. 18, 301 (1997)
Peled E., C. Menachem, D. Bar-Tow, A. Melman. J. Electrochem. Soc.. 143, L4 (1996)
Pineri M., E. Roche, B. Rodmacq, F. Volino. In: Solid State Protonic Conductors I for Fuel Cells and Sensors. eds. by J. Jensen, M. Kleitz, Odense University Press, Denmark (1982)
Pistoia G. Lithium batteries: New Materials, Developments and Perspectives. Elsevier, Amsterdam (1994)
Pohl H. W. Hydrogen and Other Alternative Fuels for Air and Ground Transportation. Wiley, New York, (1995)
Prater K. In: Fuel Cells – Grove Anniversary Symposium. 1989. ed. by D. G. Lovering, Elsevier, London (1990)
Qi Z., M. C. Lefebvre, P. G. Pickup. J. Electroanal. Chem.. 459, 9 (1998)
Reisner D. E., A. J. Salkind, P. R. Strutt, T. D. Xiao. J. Power Sources. 65, 231 (1997)
Reisner D. E., T. D. Xiao, A. J. Salkind, P. R. Strutt. In: Electric and Hybrid Vehicles Technology. 1996, UK and Int. Press, London. 112 (1996)
Retoux R., T. Brousse, D. M. Schleich. J. Electrochem. Soc.. 146, 2472 (1999)
Riley R. Q. Alternative Cars in the 21st Century: A New Personal Transportation Paradigm. Society of Automotive Engineers, Inc, United States of America. chap. 4 (1994)
Rodriguez N. M., A. Chambers, R. T. K. Baker. Langmuir. 11, 3862 (1995)
Rolison D. R., P. L. Hagans, K. E. Swider, J. W. Long. Langmuir. 15, 774 (1999)
Ross P. N. In: Proceedings of the Workshop on Direct Methanol-Air Fuel Cells Vol. 92–14, eds. by Landgrebe, A. R. R. K. Sen, 1992, The Electrochemical Society, Pennington, New Jersey, 51 (1992)
Rossouw M. H., D. C. Liles, M. M. Thackeray, et al. Mater. Res. Bull. 27, 221 (1992)
Ruetschi P., F. Meli, J. Desilvestro. J. Power Sources. 57, 85 (1995)
Sandstede G. From Electrocatalysis to Fuel Cells. University of Washington Press, Seattle (1972)
Sato K., M. Noguchi, A. Demachi, N. Oki, M. Endo. Science. 264, 556 (1994)
Sato, Y., K. Ito, T. Arakawa, K. Kobayakawa. J. Electrochem. Soc.. 143, L225 (1996)
Savy M., F. Coowar, J. Riga, J. J. Verbist, G. Bronoël and S. Bex. J. Appl. Electrochem. 20, 260. (1990)
Scherson D. A., S. L. Gupta, C. Fierro, E. B. Yeager, M. E. Kordesch, J. Eldridge R. W. Hoffman, R. W. Blue. J. Electrochim. Acta. 28, 1205 (1983)
Scherson D. A., A. A. Tanaka, S. L. Gupta, D. Fryk, C. Fierro, R. Holze, E. B. Yeager, R. P. Lattimer. Electrochim. Acta. 31, 1247 (1986)
82
Schokoohi F. K., J. M. Tarascon, B. J. Wilkens, D. Guyomard, C. C. Chang. J. Electrochem. Soc.. 139, 1845 (1992)
Schulze G. E. R. Z. Physik. Chemie. B 25, 215 (1934)
Scott K., W. Taama, J. Cruichkshank. J. Appl. Electrochem. 28, 289 (1998)
Scrosati B. Nature 373, 557 (1995)
Seitz W. C., J. Shimosato. The Promise of Lithium-Ion Batteries: Expanding Markets for Portable Devices. SRI International Business Intelligence Program, California (1994)
Seitz W. C. Advanced Batteries for Mobile Life-Styles. ed. by Harris A., SRI International Business Intelligence Program Report. No. 802, California (1991)
Sequeira C. A. C., A. Hooper. Solid State Batteries. NATO-ASI, Series, Ser. E101, Martinus Nijhoff Publications, Dortercht (1985)
Shouji E., D. A. Buttry. Langmuir. 15, 669 (1999)
Singh B. K., S. S. Sai Raman, A. K. Singh, B. B. Bansal, O. N. Srivastava. In: Hydrogen Power: Theoretical and Engineering Solutions. ed. by Saetre, T. O. Kluwer Academic Publishers, Netherlands (1998)
Singh D., T. Wu, M. Wendling, P. Bendale, J. Ware, D. Ritter, L. Zhang. In: Materials for Electrochemical Storage and Conversion II-Batteries, Capacitors and Fuel Cells. eds. by D. Ginley, D. H. Doughty, B. Scrosati, T. Takamura, J. Zhang, Mat. Res. Soc. Symp. Proc. Vol. 496, Materials Research Society, Pennsylvania, 25 (1986b)
Sonobe N., M. Ishikawa, T. Iwasaki. 35th Battery Symposium in Japan. 14–16 Nov, Nagoya (1994a), paper 2B09, extended abstracts, 47
Sonobe N., M. Ishikawa, T. Iwasaki. Chuman, H. 35th Battery Symposium in Japan. 14–16 Nov, Nagoya, paper 2B10, extended abstracts, 49 (1994b)
Springer T. E., T. A. Zawodzinski S. S. Gottesfeld. Proceedings of the Symposium Modeling of Batteries and Fuel Cells. Vol. 91–10. The Electrochemical Society, New Jersey, 209; J. Electrochem. Soc.. 138, 2334 (1991)
Srinivasan S., E. A. Ticianelli, C. R. Derouin, A. Redondo. J. Power Sources. 22, 359 (1988)
Srinivasan S., O. A. Velev, A. Parthasarathy, D. J. Manko, A. J. Appleby. J. Power Sources. 36, 229 (1991)
Tarascon J. M. J. Electrochem. Soc.. 132, 2089 (1985)
Tarascon J. M., E. Wang, F. K. Shokoohi, W. R. McKinnon, S. Colson. J. Electrochem. Soc.. 138, 2859 (1991)
Tarasevich M. R., K. A. Radyushkina. Mater. Chem. Phys.. 22, 477 (1989)
Taylor E. J., E. G. Anderson, N. R. K. Vilambi. J. Electrochem. Soc.. 139, L45 (1992)
Thaller L. H., A. H. Zimmerman. J. Power Sources. 63, 53 (1996)
83
Tsang C., J. Kim, A. Manthiram. J. Solid State Chem.. 137, 28 (1998)
Tsukuma K., T. Takahata, M. Shiomi. Science and Technology of Zirconia III. In: Advances in Ceramics. Vol 24. eds. by S. Somiya, N. Yamamoto, H. Yanagida, American Ceramics Society, Columbus, 721 (1988)
Vaia R. A., H. Ishii, E. P. Giannelis. Chem. Mater.. 5, 1964 (1993)
Vanderborgh N. E., C. A. Spirio, and J. R. Huff. Extended Abstracts of the International Seminar on Fuel Cell Technology and Applications. The Hague, Netherlands (1987) Oct. 1987, 253
Vanderborgh N. V., J. Guante, R. E. Dean, and R. D. Sutton. Abstracts of the Fuel Cell Seminar, Long Beach, California (1988). Oct. 1988, 52
Van Veen J. A. R., H. A. Colijn, and J. F. van Baar. Electrochim. Acta. 33, 801 (1988)
Venkatasetty H. V. Lithium Battery Technology. Wiley, New York, (1984)
Verbrugge M. W. and R. F. Hill. J. Electrochem. Soc.. 137, 886 (1990)
Vinal G. W. Primary Batteries. Wiley, New York, (1950)
Wakihara M., O. Yamamoto. Lithium Ion Batteries: Fundamentals and Performance. Wiley, Weinheim chap. 4 (1998)
Wang C. S., G. T. Wu, X. B. Zhang, Z. F. Qi, W. Z. Li. J. Electrochem. Soc.. 145, 2751 (1998)
Wang J., I. D. Raistrick, R. A. Huggins. J. Electrochem Soc.. 133, 457 (1986)
Watanabe M., K. Sakairi, M. Inuoue. J. Electroanal. Chem.. 375, 415 (1994a)
Watanabe M., H. Uchida, Y. Seki. Igarashi, M. Emori. Abstract. 606, The Electrochemical Society Extended Abstracts. Vol 94–2, Miami Beach, Florida, Oct. 9–14 (1994b)
Watanabe M. US Patent No. 05472799 (1995a)
Watanabe M., H. Uchida, H. Igarashi, M. Suzuki. Chemistry Letters. 21 (1995b)
Watanabe M., H. Uchida, M. Emori. J. Phys. Chem.. B102, 3129 (1998); J. Electrochem. Soc.. 145, 1137 (1998)
Watanabe M., H. Uchida, M. Yoshida. J. Electrochem. Soc.. 144, 1739 (1997)
Wendt H. Electrochim. Acta. 39, 1749 (1994)
Wenzl H., K. H. Klatt. In: Hydrides for Energy Storage, Proc. of an Internat. Symp. In Geilo, Norway, 14–19 Aug 1977 eds. by Andresen A. E., A. J. Maeland, Pergamon Press, New York, (1978)
Weston J. E., B. C. H. Steele. Solid State Ionics. 7, 75 (1982)
Widelöv A. and R. Larsson. Electrochim. Acta. 37, 187 (1992)
Wiesener K. Electrochim. Acta. 31, 1073 (1986)
84
Wilkinson D. P., J. R. Dahn, U. von Sacken, D. T. Fouchard. J. Electrochem. Soc.. 137, 370C (1990)
Wong H. P., B. C. Dave, B. C. F. Leroux, J. Harreld, B. Dunn, L. F. Nazar. J. Mater. Chem.. 8, 1019 (1998)
Wu G. T., C. S. Wang, X. B. Zhang, H. S. Yang, Z. F. Qi, W. Z. Li. J. Power Sources. 75, 175 (1998)
Yamada O., M. Ishikawa, M. Morita. Electrochim. Acta. 44, 1607 (1999)
Yang J., M. Winter, J. O. Besenhard. Solid State Ionics. 90, 281 (1996)
Yata S., H. Kinoshita, M. Komori, et al. Synth. Met. 62, 153 (1994)
Yeo R. S., A. Eisenberg. J. Appl. Polym. Sci.. 21, 875 (1977)
Yeo R. J. Electrochem. Soc.. 130, 533 (1983)
Yoneyama H., A. Kishimoto, S. Kuwabata. J. Chem. Soc., Chem Commun., 986 (1991)
Zhao Y., Z. Yuan, H. Zhang, X. Cao, S. Li, C. Sun. Mater. Chem. Phys.. 57, 285 (1999)
Zaluski L., A. Zaluski, J. O. Ström-Olsen. J. Alloys Comp.. 217, 245 (1995)
Zawodzinski T. A. Jr., C. Derouin, S. Radzinski, R. J. Sherman, V. T. Smith, T. E. Springer, and S. Gottesfeld. J. Electrochem. Soc.. 140, 1041 (1993)
Zheng T., Y. Liu, E. W. Fuller, S. Tseng, U. von Sacken, J. R. Dahn. J. Electrochem. Soc.. 142, 2581 (1995a)
Zheng T., Q. Zhong, J. R. Dahn. J. Electrochem. Soc.. 142, L211 (1995b)
Zheng T., J. S. Xue, J. R. Dahn. Science. 270, 590 (1995c)
Zheng T., W. Xing, J. R. Dahn. Carbon. 34, 1501 (1996)
Zimmerman. In: Introduction to Nickel Hydroxide Electrode, Electrochemical Society Proceedings. Vol 94–27, 268 (1994)
85
12. Nanocomposites
12.1 Introduction
Materials research is so important that the historical ages of human civilizations have been named after the materials that enabled that age (stone age, bronze age, iron age, etc). Even today, the information age is based on silicon-based semiconducting materials. Materials science and technology is seen as one of the most important disciplines along with information technology and life science that will pave the way to technology in the newly 21st century.
Previously, a great deal of effort worldwide has being directed toward developing new materials. The knowledge and technology of materials have been accumulated that most demands of industry can be met. Accordingly, most elemental materials have been studied thoroughly and many single-phase materials have been successfully manufactured. Recent trends of materials research approach two aspects: compositional refinement and structural integration of existing materials. Structural integration of materials belongs to two or more functionally different materials in monolithic form, such as functional gradient materials, intelligent materials, or smart structures. A composite is one or more distinctive components dispersed in a continuous matrix creating a compositional heterogeneity of the final solid structure. A typical conventional composite is glass-fiber-reinforced plastic (GFRP) that is widely used in aircrafts, large containers, and automotive parts. "Nanocomposite" is defined as a composite in which the distinctive component, or Gibbsian solid phase, is in the nanometer range. The accepted length for the nanophase is less than 100 nm in at least one dimension. The continuous matrices can be ceramic, metallic or organic materials, either in bulk form or as thin films.
One way of classifying nanocomposites is by Newnham's classification of conventional composites (Newnham, et al., 1978). He has proposed a classification for a two-component composite by the numbers 0–1, 0–2, 1–3, etc to describe the connective relationship of the component phases in the composites as shown in Fig. 12.1, where the first number refers to the connectivity of one phase and the second to the connectivity of another phase. The numerals 0, 1, 2 and 3 mean that this phase connects by itself at zero dimensions, one dimensions, two dimensions, and three dimensions, respectively. This classification is convenient for recognizing the structural relationship of constitutional phases. Another classification of composite materials is based on the continuous matrix composition, such as ceramic, metallic or polymeric matrix, and it is also applied to nanocomposites.
Figure 12.1 Classification of two-phases composite based on connectivity. Each phase has zero, one-, two-, or three-dimensional connectivity to itself. In the 3-1 composite, for instance, the shaded phase is three-dimensional connected. Arrows are used to indicate the connected directions. Two views of the 3-3 and 3-2 patterns are given because the two interpenetrating networks are difficult to visualize on paper. The views are related by 90° counterclockwise rotation about z.
86
Composites are expected to exhibit superior properties or better performance than their elemental or monolithic counterparts. Most of the property changes can be estimated by some rule of mixtures. The simplest change of a composite property, Pc, is monotonically increasing or decreasing with the increase of volume fraction, Vi, of the added components; which is represented by the following equation (Kakawa, 1990):
Pi is the added component property and n is an experimental parameter (1 n -1).
In the case of n = 1, n = 0 and n = -1, the property of a fiber-reinforced composite can be evaluated by the following three equations, assuming the properties of fiber and matrix are Pf and Pm, respectively. The volume fraction of reinforcing fiber is Vf:
87
Using a fiber reinforced plastic composite, for example, the strength parallel to the fiber direction mostly follows Eq. (12.2), and the electronic conductivity perpendicular to the fiber direction follows Eq. (12.4). These equations are illustrated in Fig. 12.2, which shows that Eq. (12.2) gives the highest value of property of composites. This is based on the occurrence of minimal chemical interaction between phases and lesser structural defects or flaws.
Figure 12.2 Plot of change in property with constituent volume based on rule-of-mixtures.
Nanocomposites differ from their conventional composites in the strong interaction of grains around grain boundaries. Each crystalline particle consists of inner ordered lattice atoms and outer disordered grain boundary atoms (Fig. 12.3). In nanoparticles and bulk nanocrystalline materials, the grain boundary atoms have a volume fraction much higher than conventional materials, up to 20 vol% in 20 nm materials (Fig. 12.4). These grain boundary atoms are electrically unsaturated and chemically active, which cause some unusual electromagnetic behaviors that would not be expected from Eq. (12.1). These may be nonlinear changes or a rule of products. For example, the strengths of Al2O3 and SiC are 800 MPa and 500 MPa, but a nanocomposite composed of Al2O3 and SiC shows 1500 MPa strength (Niihara and Nakahira, 1990). In this case, grain boundary interaction plays an important role in the strength improvement. Many nanocomposites have been experimentally investigated, and the nonlinear effects of property/performance improvement have been verified in some systems, e.g., mechanical properties, electromagnetic properties, and optoelectronic properties.
Figure 12.3 Two kinds of atoms existing in the bulk nanomaterials, emphasizing a high percentage of the atoms associated with the disrupted bonding of the grain boundary.
88
Figure 12.4 Ratio of unsaturated atoms in nanomaterials.
Scientists are increasingly researching nanocomposites. A recent literature search showed that in the last seven years research articles number n on nanocomposites published in each year have increased approximately 25% each year and to total of 10 times, from 1992 to 1998 (Fig. 12.5). In 1998, we found 261 papers using the SCI database and 150 using EI database, which show the increased interest in nanocomposites.
Figure 12.5 Literature search on nanocomposites using SCI and EI databases, showing rapid increase of the number of the scientific articles published in the recent years.
89
Figure 12.6 shows that the ratio of the top six properties of interest in nanocomposites. The greatest is mechanical properties, followed by magnetic properties, optical properties, electrical properties, quantum dots, and catalysis. Half of the EI database indexed researchers mainly focus on mechanical property improvements, in contrast to the SCI papers that showed focus on mechanical properties, magnetic properties, and optical properties.
Figure 12.6 The ratio of the major topics of interest on nanocomposites. The mechanical properties, magnetic properties, optical properties were given more attention by researchers.
This chapter will review the processing and properties of nanocomposites with ceramic, metallic and organic matrices, and will summarize the general features.
90
12.2 General Features of Nanocomposites
Nanocomposites differ from traditional composites in the smaller size of the component phases. The small size of the phases may cause stronger physical sensitivity of bulk materials to physical or mechanical energy and higher chemical reactivity of grain boundaries.
12.2.1 Physical Sensitivity: Three Effects of Nanoparticles on Material Properties
Nanoparticles and nanoclusters may exhibit unique behaviors due to three effects: the small size effect, the large grain boundary effect, and the quantum confining effect.
(1) Small size effect
A nanophase is defined as a particle or cluster with a size of approximately 100 nm. Many natural basic units have this range of size. Cells, a red blood cell or a bacteria cell, are 200–600 nm. A magnetic domain is several tens of nanometers. Optical wavelengths are several hundreds of nanometers. When the particle sizes in composite materials approach lengths of physical interaction with energy, such as a light wave, electromagnetic wave, or thermal conductivity, the periodic boundary conditions of the coupling interaction with the energy would behave different from its microscopic counterparts, which results in unusual properties. A magnetic nanoparticle usually only contains a single magnetic domain whose domain wall is easy to move and polarize under electromagnetic field. Most of the unique magnetic and optical behaviors of nanocomposites are attributed to the small size effect.
(2) Grain boundary effect
A crystalline particle is composed of two parts: the inner regular crystalline structure and the outer irregular crystalline structure at the grain boundary. An atom on the grain boundary is unsaturated in chemical bonding and partially disordered in the lattice. The big difference between nanomaterials and micrograined materials is the larger volume fraction of grain boundary atoms. Assuming a grain boundary thickness of 1 nm, the grain boundary volume fraction increases approximately 10 times in materials with a particle size of several nanometers compared to those with a particle size of a 100 nm. The number of atoms on the boundary (ns) related to the total number of atoms in the grain (N) and is represented as the fraction of surface atoms F, which is a function of diameter (D) of the nanograin, as shown below:
91
Such a huge quantity of unsaturated surface atoms will significantly enhance the reactivity of nanomaterials. At room temperature, some metallic nanoparticles can burn when exposed in air, and some inorganic nanoparticles exhibit much higher adsorption of gas due to the enhanced reactivity. At elevated temperatures, the activity of the nanoparticles significantly enhances the sintering process and lowers the sintering temperature by 200 degrees for alumina.
Lowering melting points of metallic nanomaterials and the electrical resistance change and optical behavior in metallic and ceramic nanomaterials may be attributed to the much higher activity of unsaturated atoms in nanomaterials.
(3) Quantum confinement effect
When electrons are confined to a small domain, such as a nanoparticle, the electrons behave like "particles in a box" and their resulting new energy levels, are determined by quantum "confinement" effects. These new energy levels give rise to the modification of optoelectronic properties such as spectral "blue shift" light emitting diode (LED). Although quantum dot devices are highly attractive for optoelectronics applications, their processing is still very difficult.
12.2.2 Chemical Reactivity
Nanoparticles exhibit high reactivity during synthesis and processing of composites and in service situations. This refinement-enhanced reactivity is an important factor for property improvements in nanocomposites.
Roy and co-workers (Roy, 1993) confirmed that nanocomposites have much higher reactivity than equivalent microcomposites and emphasized that solid state epitaxy is taking place in nanocomposites while they are heat-treated. They utilized this solid state epitaxy to explain virtually all their observations of enhanced chemical reactivity, such as that shown in Table 12.1.
Table 12.1 Reactivity achievements with nanocomposites: compositional (C) and structural (S)
All reaction and sintering temperatures lowered by 150–300°C (C & S)
Can drive complex ceramics to different final phase, some obviously metastable but very useful (S)
Can crystallize glasses never crystallized before (S)
Can radically refine the microstructure (S & C)
Can tailor morphology (C)
Can achieve properties (such as hygroscopicity) greater than components (C)
The high reactivity of nanomaterials may enhance the interaction of the nanophase with the matrix and may exhibit the following results:
92
1. Higher gas absorption: large specific area of nanomaterials can easily absorb gaseous species. 2. Increased nonstoichiometric phases: nanomaterials easily form chemically unsaturated bonds and
nonstoichiometric compounds. 3. Regrowth: nanomaterials are probably easier to recrystallize and regrow in processing and service conditions than
traditional materials. 4. Rotation and orientation: crystallographic rotation and orientation of nanoparticles have been found in processing
of nanocomposites. 5. Sub-graining: nanoparticles enveloped into larger particles act as dispersed pinholes to divide the large particles
into several parts. 6. Epitaxial growth: epitaxy expanding and regrowth of the crystallites are confirmed in the nanomaterials. 7. Assembly: nanoparticles are easy to aggregate and assemble in lines or other regular features in liquid or gaseous
media.
12.2.3 Promising Improvements in Nanocomposites
Nanocomposites are usually materials in which nanoparticles are dispersed in a continuous matrix. Matrices for nanocomposites can be ceramic, metallic and organic materials. They can be compact bulk materials, thin films, or mesoporous host-guest preforms. Matrices play the roles of structural stability, passivation, and supporting substrates. Some nanophase and matrix systems include mechanically hard crystallites of SiC in high strength ceramic materials, transformation toughened ZrO2 in superplastic ceramics, magnetic metallic phases of Fe and Co in magnetic materials, conducting particles of Ag in quantum confinement materials, insulated ceramic SiO2 particles in luminescent materials, and delaminated clay minerals as filler in polymers.
Possible property Improvements differ for each matrix. Expected nanocomposites are listed below:
1. Ceramic matrix nanocomposites 1. Increase in strength, hardness, and abrasion by refining particle size; 2. Enhance ductility, toughness, formability, superplasticity by nanophases; 3. Change electrical conduction and magnetic properties by increasing the disordered grain boundary
interface. 2. Metallic matrix nanocomposites
1. Increase the hardness and strength; 2. Lower the melting point; 3. Increase the electrical resistivity by increasing the disordered grain boundary area; 4. Increase the solubility for solid solution.
3. Polymeric matrix nanocomposites 1. Use fine filler in organic materials to increase properties such as abrasion, heat retardation, and modulus; 2. Decrease the electrostatic properties by adding TiO2, Cr2 O3, Fe2 O3, ZnO instead of carbon black; 3. Increase absorption of ultraviolet wavelength; 4. Provide magnetic fluid by adding magnetic nanoparticles in fluid polymers; 5. Passivation of quantum dots.
93
The applications of nanocomposites on semiconducting properties and magnetic properties have been studied in detail for the above matrices due to their practical usage. The most important improvements by nanophases are:
4. Semiconducting nanocomposites 1. Add semiconducting particles in the polymer to shield the electrostatic field; 2. Add semiconducting particles in mesoporous silica to absorb light and microwave radiation; 3. Increase blue shift of optical spectra by adding nanoparticles in the polymer to form quantum dots; 4. Increase dielectric constant by adding metal Ag nanoparticles in epoxy resin; 5. Increase unique magnetic resistance by adding magnetic particles such as Fe, Co, and Ni in metals such as
Ag, Cu, and Au. 5. Magnetic nanocomposites
Lower critical grain size to decrease in coercivity, leading to superparamagnetic behavior.
12.2.4 Origin of Nanophases and Generating Stages
A typical argument against nanocomposites is their instability and tendency to grow to micrometer size in the synthesis process. The whole process flow to prepare a nanocomposite includes powder preparation stage, mixing stage, consolidation stage and annealing stage. In general, every stage can produce nanophases such as shown in Fig. 12.7. The most important stages of nanophase formation are in power preparation stages according to a literature search. The aerogels process, coprecipitation and decomposition of polymeric precursor are frequently reported as brief processes to prepare the nanoparticles for the nanocomposites. Mixing stage produces nanophases by mechanical alloying based on powerful ball-milling technology. Consolidation stage sometimes produces nanophases by the reaction of additive with matrix and the re-crystallization of the glassy phases. Figure 12.8 shows an example of Ti-N-C nanophases in the Si3 N4 grain due to the reaction of the added TiC micrometer particles with the matrix Si3 N4 grains at the sintering temperature (1820°C for 90 min). But in the most cases, consolidation dramatically reduces the nanophases due to their surficial activity while the nanophases were explored at elevated temperature. Most of the nanometer particles of an alumina powder tend to grow into micrometer grains when sintered above 1650°C. An annealing and heat treatment after consolidation and sintering may produce nanophases by combining the eutectic crystallization, spinodal discomposition and phase separation. Figure 12.9 shows a nanometer structure of (Ti, Zr) N in which the TiN and ZrN separated in nanometer layers (approximately 30 nm in thickness for one layer) after the micrometer particles of the (Ti, Zr) N solid solution were annealed at 1400–1600°C in N2 atmosphere (Li, et al., 2000). Except for the bulk nanocomposites, part of the nanocomposites was formed in thin film by coating technology based on the chemical vapor deposition (CVD) and physical vapor deposition (PVD) in literatures, which shall be introduced in detail in an other chapters.
Figure 12.7 Schematic showing the origin and generation time of nanophases in the nanocomposite synthesis process.
94
Figure 12.8 Ti(N, C) nanophases in the Si3N4 grains after TiC/Si3N4 composite was sintered at 1820°C /90 min. A: Si3N4 grains in the TiC/Si3N4 composite; B: Ti(N, C) nanophases in the Si3N4 grains.
Figure 12.9 Nanolayers formed in the (Ti, Zr)N grains in TiN/ZrO2 composite after annealing at 1600°C in N2 for 2 h. A: Natural surface of TiN grain with nanometer layers; B: Fracture surface of TiN grains showing the fracture steps.
95
12.3 Ceramic-Based Nanocomposites
Ceramic-based nanocomposites are composites that contain ceramic nanophases either comprising over half of the total volume fraction or with an interconnective relationship. Based on the position of the nanophase particles in the final dense material, Niihara classifies ceramic nanocomposites into four types: intra-type, inter-type, intra/inter type, and nano/nano type as illustrated in Fig. 12.10 (Niihara, et al., 1993). He emphasizes that the intra-type is mostly expected to improve the mechanical properties of ceramic matrix composites.
Figure 12.10 Niihara's classification of ceramic nanocomposites based on the position of the nanophase: of the four types of nanocomposites, intra-type is expected to exhibit improvement on mechanical properties.
96
It has been investigated and confirmed in many compositional systems that ceramic nanocomposites for structural materials application exhibit great improvements in mechanical properties such as strength, toughness and hardness. Ceramic matrices are also used as stabilizers or passive hosts for some functional nanoparticles such as Al2O3 for Au quantum dots, SiO2 for Fe magnetic nanoparticles, and SnO2 for Pt nanocatalyst. A mesoporous ceramic matrix is a common support or substrate for optoluminescent nanomaterials. This section focuses on mechanical improvement of ceramic-based nanocomposites.
12.3.1 Strength Improvement of Ceramic-Based Nanocomposites
Since Niihara's research group reported great improvement in the mechanical properties of ceramic nanocomposites (Niihara, 1991), researchers have attempted to produce most ceramic material systems in the past 10 years. Table 12.2 shows typical results of Niihara's group on ceramic nanocomposites. Of these, the most impressive result is the nanosized SiC particle reinforced Al2O3 matrix composites in which the addition of 5 vol% SiC particles can increase the strength of hot-pressed Al2O3 from 350 MPa to over 1 GPa and fracture toughness from 3.25 to 4.7 MPa m1/2. They also revealed that the post-processing of heat treatment at 1350°C for 2 h of a sintered ceramic body can further improve the strength of SiC/Al2O3 up to 1500 MPa. A detailed observation of microstructure revealed that SiC nanoparticles were partially enveloped into large Al2O3 matrix grains and divided the matrix grains into several areas by forming subgrain boundaries inside the matrix grains. The fracture surface showed transgranular fracture, and the crack deflected around the nanoparticles. The mechanism of annealing enhancement was conjectured to be attributed to either or both the microcrack-healing and microstress-removing at heat treatment temperature. Although several groups have tried to reproduce this result, so far none has reached such high magnitudes of improvement as Niihara's group did.
97
Table 12.2 Some typical systems of ceramic/ceramic nanocomposites showing the mechanical improvements of nanosized SiC particles over micrometer sized SiC and Si3N4
Toughness (MPa · m1/2) Strength (MPa) System
Microcomposites Nanocomposites Microcomposites Nanocomposites
SiC/Al2O3 3.5 4.8 350 1520
Si3N4/Al2O3 3.5 4.7 350 850
SiC/MgO 1.2 4.5 340 700
SiC/Si3N4 4.5 7.5 850 1550
The promising increase in strength is the greatest benefit of ceramic nanocomposites. The improvement of strength varies from several percent up to three times that of the monolithic counterparts. The mechanism of strengthening is probably due to one or several of the following factors depending on the composition and microstructure: grain refinement, flaw reduction, sub-grain formation and residual stress recontribution.
1. Strengthening from grain refinement
The strength of ceramic materials is dependent on the grain size of the component phases. The Hall-Petch equation shows the simple relationship between fracture strength (σ) and grain size (d-1/2) assuming other factors remain unchanged (Borsa, et al., 1994).
It can be understood from the above equation that refinement and size reduction of matrix particles (d) may increase the strength (σ) of ceramic materials. Mathematically, if the added nanophase reduces the grain size to half of the monolithic one, the strength could increase by approximate 40%.
Ceramic matrix grains may increase in size, regrowth, while they are consolidated at elevated temperature. The addition of a small amount of inert nanoparticles such as SiC would block the regrowth process due to the reduction of the diffusion of the constituents. Al2O3 grains originally 200 nm grew to over 1000 nm after sintering, but with the addition of 5vol% SiC the grains grew to less than 500 nm. (Niihara and Nakahira, 1990).
2. Strengthening from reduction of flaw size
Ceramic materials are flaw sensitive. The relationship between the fracture strength (σf) of ceramic materials and the critical flaw size can be evaluated by the well-established Griffith equation:
98
where K1c is a material's resistance to inherent flaws that is usually called the fracture toughness for brittle materials, Y is a geometric factor approximately equal to 1, and α is the length of the flaw. This equation indicates that increase in fracture toughness, K1c, and decrease in flaw size can lead to improvement in strength.
The critical flaw size in ceramic materials is in most cases proportional to the grain size. The refinement of grain size leads to the reduction of flaw size and, therefore, to an increase in the strength compared to equivalent monolithic materials.
Nanophases added to ceramic materials sometimes produce a toughening effect. The mechanism for this effect will be discussed later. It is clear that increased toughness by nanophases should lead to increased strength according to Eq. (12.8).
3. Strengthening from sub-graining
Very small particles can be enveloped into large matrix grains during the regrowth process at the sintering temperature. These small particles can then divide the matrix grain into several sub-grains due to the thermal mismatch. This sub-graining effect causes transgranular fracture and thus improves the strength.
4. Strengthening from residual stress redistribution
The second phase introduces a perturbation into the matrix by introducing a stress field. While the ceramic composite is cooling from the sintering temperature, Ts, the thermal expansion coefficient mismatch (TECM, Δα = α1 - α2) induces a hydrostatic stress in both the matrix and the second phase. The magnitude of this residual stress, P, is equal to
where Δ T is the cooling range during which plasticity is negligible and ν1, ν2 and E1, E2 are Poisson's ratios and Young's moduli of the matrix and the particles, respectively (Fantozzi and Olagnon, 1996).
The magnitude of residual stress also depends on the distance, x, from a point in the matrix to the particle and the particle's radius, r, with the following relationship:
99
It is clear that a composition system with a large TECM and large grain size induces a large residual stress. A reasonable magnitude of residual stress stored in the matrix will absorb elastic energy and contribute to a fracture energy increase, but if the residual stress is too large, it can cause microcracking and result in a decrease in strength.
Sternitzke (Sternikze, 1997) has tried to calculate the stress field distribution around a SiC particle in Al2O3 matrix and found that in a cubic-body-centered arrangement of nine equivalent particles the central particle results in a high tensile stress and the other particles 100 nm away from it generate compressive stresses up to 120 MPa. He concluded that SiC particles within the Al2O3 grain located close to grain boundaries strengthen the grain boundaries due to radial compressive stress and thus exhibit a strengthening effect.
The critical size of the second phase that will not cause microcracking in the matrix has been given by Davidge and Greene (1968):
This equation means the critical size of the second phase which causes the microcracking is dependent on the surface energy, γs, of the matrix, the moduli, and the applied pressure. A second phase smaller than this critical size can absorb the external energy and raise the strength. The critical sizes for SiC particles are estimated to be 30–100 nm and 100–300 nm for matrices of Al2O3 and Si3N4, respectively, based on 5 vol%—15 vol% addition.
Levin et al. (1994) have experimentally measured the residual stress of SiC/Al2O3 using X-ray diffraction and found that a 5 wt% addition of nanophase is the critical amount for SiC/Al2O3. Less than 5 wt% of SiC nanophase causes part of the Al2O3 matrix to experience compression, and greater than 5 wt% causes the entire Al2O3 matrix to experience tension. This explains why most of the effective strengthening results for the SiC/Al2O3 system have been reported with less than 10 wt% addition. The critical size and the critical amount of nanophase addition are important factors to induce the proper compressive stress toward the grain boundaries where cracks that propagate into the compression field would be deflected to regions of tension and result in improved strength and toughness.
In ceramic nanocomposites, Al2O3 and Si3N4 matrices have been greatly studied for the past decade. The nanophase for these matrices is SiC nanoparticles. SiC can absorb a small amount of Al2O3 at the sintering temperature to form a limited solid solution, but the grain boundaries between grains of SiC and Al2O3 are almost clear or glassy. In contrast, SiC nanoparticles within Si3N4 matrix exhibit more reaction and interaction with matrix grain, such as grain rotation to form the small angle boundaries with matrix grain and formation of solid solution within the grain boundary.
The typical processing of ceramic nanocomposites is shown in Fig. 12.11 and the typical result is indexed in Fig. 12.12 (Niihara, et al., 1993). Nanoparticles, such as SiC, can be introduced into the matrix by fine powders or pyrolysis of silicon-containing organic polymer (precursor). Consolidation of nanoceramic
100
composites is usually conducted by hot-pressing sintering but pressureless sintering is also useful for the alumina matrix.
Figure 12.11 Flowchart of preparation processing of a typical ceramic nanocomposite.
Figure 12.12 Typical improvement on properties of ceramic nanocomposites. Annealing process was effective.
101
Other systems that have been studied include SiC/mullite, SiC/MgO, TiO2/Al2O3, ZrO2/mullite, ZrO2/Al2O3, TiC/Al2O3, TiN/Al2O3, TiC/Si3N4, and BN/AlN. Most of them have been confirmed to exhibit improvements of strength and toughness.
12.3.2 Toughening Effect of Nanoceramic Composites
There are a few approaches to improve the toughness of brittle materials: crack deflection, microcracking, transformation toughening, and crack bridging. However, nanophases toughen only by phase transformation and microcracking.
Nanophase and fine particles of tetragonal ZrO2 greatly improve the toughness of bulk ceramic and other brittle materials. The improved toughness for Al2O3 and Si3N4 may be higher than 3 to 6 MPa · m1/2. The toughening effect results from the transformation of ZrO2 from the tetragonal phase to the monoclinic phase, resulting in volume expansion of approximately 5%. This transformation may be induced by any stress such as the machining force in the surface and the crack tip opening force (crack stress field) inside the materials. The crack tip stress induces the transformation of t-ZrO2 when the crack has propagated to the face of t-ZrO2 grain. Expansion of the volume of the t-ZrO2→m-ZrO2 transformation counteracts the crack tip and tends to close the tip opening, which results in an apparent increase of the toughness.
It is believed that the addition of an inert stable nanophase would improve the toughness of brittle materials by the microcracking mechanism. A few researchers reported the increase of fracture energy in nanocomposites using the indentation method (Kodama and Miyoshi, 1990) but direct improvement of toughness of significant magnitude has not been reported so far, though a number of experiments have been conducted by independent researchers for the last decade. Experimental verification of the toughening effect of nanophase particles in brittle materials requires further study.
102
Another possible way for nanophases to toughen ceramic materials is from morphology control. A nanophase may toughen the matrix by influencing the morphology of the matrix grain and thus the microstructure of the bulk materials. Niihara et al. (1989, 1988) reported that a small amount (less than 10 vol%) of SiC particles with diameters of less than 100 nm can enhance the development of an elongated Si3N4 grain. The elongated grain may act as a reinforcement to significantly improve the toughness similar to a short fiber or whisker. The presence of the nanophase may restrict the growth of most matrix grains and cause some abnormal grain growth in some systems. There must exist a proper amount of nanophase to balance between oppressing the growth of most matrix grains and enhancing elongation of other grains. Further composition selection and microstructure design associated with synthesis and processing are still needed to demonstrate the toughening effect of nanophase in nanocomposites.
12.3.3 Improvements of Nanoceramic Composites on Hardness and Wear
S. Veprek et al. (1996), have prepared thin films of the nanoparticles of TiN and W2N mixed into α-Si3N4 deposited by plasma CVD on 500–550°C substrate and found the hardness exceeded 50 GPa. They also found that when the particle diameter decreased from 70 nm to 2.5 nm, the plastic hardness approached that of diamond, as shown in Fig. 12.13.
Figure 12.13 Hardness increases with the decreasing dispersed nanoparticle size, and the highest plastic hardness of nanometer TiN reinforced silicon nitride composite is close to that of diamond.
103
Wear resistance is an important property for alumina materials when used as cutting tools, wear plates, and transportation support plates. A common expression for wear resistance is the volume of the materials removed (V) as a function of applied load P, the hardness H and fracture toughness K of materials, as shown in the following:
One can understand from this equation that increases of either toughness or hardness or both in alumina will improve its wear resistance.
Anya (1998) has studied the effect of SiC nanophase dispersed into Al2O3 on wet wear resistance. He found that 10%—15% addition of SiC nanophase in alumina decreased the wear rate from 3.2 nm/s in the monolithic material to 1.4 nm/s in the nanocomposite, which is an improvement of over 200%. He also found that the wear rate is proportional to the grain size (µm) when the grain size is smaller than 2 µm. The explanation of the wear resistance improvement of the nanocomposite is that nanophases induce compressive stress toward the grain boundary and enhance the bonding strength of the boundaries, thereby, increasing the wear resistance.
12.3.4 Superplasticity of Ceramic Nanocomposites
Refinement of grain in crystalline materials also affects the creep behavior since one of the predominant mechanisms for creep involves atomic transport along the grain boundaries. Nanomaterials and nanocomposites with large quantity of grain boundaries will exhibit a greater creep rate, which may even lead to superplasticity.
Superplasticity along with the refinement of grain size is desirable in ceramic materials. It has become a new challenge to change the fracture behavior of ceramic materials from brittle into ductile. It also has the potential to enhance the formability of ceramic materials.
Karch, et al. (1987) indicated the possibility for brittle materials to have ductile deformation at low temperature when the grain size is on the order of 100 nm or less. Most deformation and creep are related to grain boundary diffusion. A well-known relation in this respect is the Coble creep equation (Coble, 1963):
where dε/dt is the strain rate, A is a constant, k is Boltzmann's constant, σ is the stress, ξ is the grain boundary width, Ω is the atomic volume, Db is the diffusion constant in grain boundaries and T is the temperature. This equation shows that the reduction in grain size may cause a considerable increase of the creep rate, by decreasing d3 and increasing Db. For instance, changing the grain size from 10 µm to 10 nm would increase the creep rate by nine orders of magnitude assuming all else remain constant. The grain refinement, furthermore, can enhance the boundary diffusivity.
104
Karch, et al. (1987) have reported that nanophase TiO2 and CaF2 exhibit extensive ductility at room temperature (tested at 80°C and 180°C, with melting points 1825°C and 1675°C, respectively). Wakei et al. (1986) demonstrated that yttria stabilized tetragonal zirconia polycrystals (Y-TZP) with approximate 300 nm grain size could be elongated over 100% in tension. Utilizing superplasticity has made it possible to forge and machine a prototypical part made of hard brittle ceramic materials into a net-shaped part at low temperature. Figure 12.14 shows an example of a prototypical forged preform (PFP) net-shaped part. This component was plastically forged from a preform ring of nanophase ZrO2 ceramic at 1400°C in less than 15 min. (Edelstein and Cammarata, 1996).
Figure 12.14 A plastically forged ceramic component fabricated from a nanostructured preform ring. The forging operation was performed at 1400°C in less than 15 min.
Weertman and Averback (1989) have summarized the common characteristics of superplastic materials:
1. small grain size to make diffusion easier; 2. equiaxed grains to reduce the sliding resistance of grain boundaries; 3. high energy grain boundaries to enhance diffusion; 4. presence of second phase to restrict grain growth.
12.3.5 Improvement of Nanoceramic Composites on Creep
Thompson et al. (1997) report that the addition of 15 vol% SiC nanoparticles to Al2O3 was sufficient to reduce the tensile creep rate by two to three orders of magnitude, showing the effect of nanosized SiC on reduction of creep and strain.
12.3.6 Ceramic-Based Nanometallic Composites
Ductile metallic nanoparticles usually exhibit obvious strengthening effects in a ceramic matrix. High temperature metals, such as W, Mo, Co, Ni, Ti, Cr and their alloys, are common candidates for the
105
dispersoids. V. D. Krstic has summarized Al2O3-based nanoparticle composites in Table 12.3 (Krstic, 1998). This data shows that the strengthening is approximately three times the microparticle reinforced composite.
Table 12.3 Room temperature fracture toughness and strength of metal nanoparticle reinforced Al2O3 matrix nanocomposites in comparison with micrometer sized composites (metal/ceramic matrix)
Toughness (MPa · m1/2) Strength (MPa) System
Microcomposite Nanocomposite Microcomposite Nanocomposite
W/Al2O3 3.5 4.0 350 1105
Mo/Al2O3 3.5 7.2 350 920
Ni/Al2O3 3.5 4.5 350 1090
Ti/Al2O3 3.5 4.3 350 816
Awano (1997) and Oh et al. (1998) reported that the ductile metallic nanoparticles of Ni-Co alloy not only improved the strength but also introduced a stress sensing function to the alumina material by detecting its magnetic property via the applied stress. Addition of 10 wt% Ni-Co alloy of 100–200 nm grain size dispersant into 0.5–1.0 µm grains size alumina increased the strength to 1000 MPa from 500 MPa and the magnetization changed up to 6% in comparison with micrograined composites. They attribute this improvement in strength to refinement of the matrix grains and saturation magnetization to magnetic nanoparticles.
12.4 Metallic-Based Nanocomposites
Metallic-based nanocomposites are expected to have advantageous mechanical and magnetic properties. The promising improvements of nanocomposites over traditional composites are:
1. Increased hardness, strength and superplasticity; 2. Lowered melting point; 3. Increased electrical resistivity due to increased disordered grain surfaces; 4. Increased miscibility of the non-equilibrium components in alloying and solid solution; 5. Improved magnetic properties such as coercivity, superpara-magnetization, saturation magnetization and
magnetocolatic properties.
The most attractive challenges of metallic nanocomposites are the unique magnetic properties. When the sizes of the magnetic particles are small enough to approximate the size of a single magnetic domain, their magnetic spins are all aligned to produce a magnetic moment in one direction, the domain walls are easy to move, and the exchange coupling is substantially enhanced in the magnetic field, resulting in better
106
magnetic properties. High density magnetic recording materials require a high coercivity and a high magnetization. It is known that a maximum value for coercivity exists in nanocomposites containing magnetic crystallites of an optimum size to a single magnetic domain. Nanocrystallite size larger or smaller than the optimum size leads to a sharp decrease in magnetization. The critical sizes vary a little depending on the component elements. Changing the size of the hard magnetic crystallite or the substitute composition can result in high coercivity that meets the requirements of high density recording materials (magnetic credit cards, keys, and tickets) and low coercivity for power transformers, magnetic recording heads and microwave applications. Iron powder with a crystallite size of 13 µm exhibited 900 Oe in comparison with 10 Oe on bulk Fe (Gangopadhyay, et al., 1992). NdFeB ternary alloy is a common magnetic material but its intrinsic coercivity of less than 9 kOe limits its application in certain areas, such as the micromotor. By controlling the soft magnetic phase (α-Fe) or hard magnetic phase Nd2Fe14B (commonly named 2-14-1 phase) in nanometer size, a desirable coercivity can be obtained.
Giant magnetoresistance (GMR) effect is another attractive property of the magnetic nanocomposite. GMR is the phenomenon of a large decrease in electrical resistivity of certain materials when exposed to a magnetic field. This effect is mostly exhibited in magnetic multilayered nanocomposites (such as Fe/Cr multilayered nanocomposite) and was recently found in magnetic/conductive nanocomposites containing magnetic nanoparticles such as Co, Fe embedded in conducting metallic matrices of Cu, Ag, Au. It is known that a smaller particle size results in a larger GMR effect. The composition producing the highest GMR is commonly near the percolation threshold. The mechanism is alignment of the ferromagnetic domain by the magnetic field reducing the scattering of the electrons and thereby decreasing the electrical resistance. Possible applications are magnetic sensors and new storage devices.
Wide range solid solubility and alloying of immiscible metals have been observed in metallic nanocomposites. Nanometer sized materials of immiscible metals have been deposited, and the metastable powder was annealed to cause the phase separation. This method allows the in situ fabrication of metastable phases embedded in nanocomposite.
One of the well-known features of metallic nanomaterials is the lowered melting point due to the relative increase of the activity of the grain boundary and the unsaturated atom volume (El-shall and Edelstcin, 1996). The interfacial activity and grain boundary volume also affect the self-diffusion coefficient, increase electrical resistance and increase catalytic activity.
Due to the relatively higher electrical attractivity to the nanoparticle's mass, nanoparticles are easily moved and crystallographically rotated in viscous fluids and tend to self-assemble in chains, array and mesoporous materials. It has been found that Cr nanoparticles align in chains and the particles grow in perfect morphology and with the same crystal habit plane. Alumina and AlOOH also have self-assembling single chains.
12.5 Polymer-Based Nanocomposites
107
Certain types of polymer nanocomposites have been produced commercially for more than half a century. During World War II, Germany developed fused silica as a substitute for ultrafine carbon black particles used for rubber reinforcement when petroleum was needed for tanks. During the 1990s, the Japanese manufactured metallic nanoparticles for producing magnetic recording tapes, and NASA in the United States developed a magnetic fluid in which nanometer sized magnetic particles of iron oxide are suspended in the fluid polymer.
Organic nanocomposites have the advantages of common polymeric materials including high toughness, good transparence, easy formability, light weight and low cost. Their matrices may be epoxies, polypropylene, polyesters, fluoropolymers and other plastics. The dispersed nanoparticles may be inorganic or clay fillers, magnetic or conducting metals and oxide ceramics. The promising improvements of the nanoparticles in polymeric matrices are expected to be in structural, mechanical properties and flame-retardant, thermal and barrier properties without significant loss of clarity, toughness or impact strength compared to common organic composites. Some unique optical and electrical properties can also be expected to appear in nanocomposites due to the size effect and high reactivity. Following are some typical applications:
1. Finer filler: Clay is a category of fine layer aluminosilicate mineral and is often used as the filler in the plastics. It can absorb water or other polar ions between the layers and swell the interlayer distance many times similar to smectite. Montmorillonite can absorb 20 to 30 times its own volume in water. The layers of aluminosilicate are approximately 1 nm in thickness and 1 µm in diameter. Many metallic hydroxides and polymers are easy to insert into the layers to form intercalation. Intercalation is an important technique to modify the layer materials. Normal intercalation will space the distances between the aluminosilicate layers to a certain degree, but the polymer intercalated into the layers can separate the layers into nanoplatelets (Kornmann, et al. 1998). Recent approaches on clay-plastics composites are popularly focused on using the structurally separated monolayered clay or nanoplatelets in the plastics. A group of researchers have shown that nanoplatelets greatly improved modulus, hardness, and strength without the loss of toughness and formability compared to conventional fillers for plastics and nylons.
Giannelis (1998) reported that a doubling of the tensile modulus and strength is achieved for nylon-layered silicate nanocomposite with as little as 2 vol% of inorganic content and the heat release rate in the nanocomposite is reduced by up to 63%. They reported the relative permeability of nanocomposite greatly decreased compared to the conventional composite. They (1998) reported that only 1.5 vol% of delaminated montmorillonite enhanced the toughness (fracture energy) of unsaturated polyester from 70 J/m2 up to 138 J/m2.
There is an application of scratch-resistant polymers molded in color for interior and exterior parts of automobiles. They can also be used in barrier packages, food packages, fire-retartant pouches, high-speed printing film, gasoline tanks and fuel-line tubes, and shock absorbers. Commercial Nylon-clay hybrid with 2% clay can be used as a gas and UV barrier as well as for high heat resistance.
Conducting polymer-smectite clay thin films can demonstrate promise as active components in sensing devices because inorganic clay serves as a porous structural framework for polymeric sensing materials, as well as adding greater structural and thermal stability to the composite.
108
2. Nanoparticles can be used as light emitting diodes (LED) with high conversion efficiency by using quantum dots embedded in polymers. Calvin (1994) assembled CdSe nanoparticles on the surface of electroluminescent polymer (PPV) and found green light emitting from the polymer and red light from the nanoparticle layer, which is also color tunable as a function of applied voltage.
3. Gao et al. (1995) found that in the sandwiched structure of polymer layer-conducting nanoparticle layer-polymer layer the conductance increased by a fact or of 107, rendering the film conductance with linear current-voltage characteristics when the voltage is higher than a certain threshold value. This nanocomposite is expected to have a switching application in electronics.
4. Sohn et al. (1998) investigated the magnetic properties of nanocomposite film in which γ-Fe2O3 nanoclusters were dispersed into optically transparent block copolymer and found a hysteresis with the saturated coercivity H oc equal to 530 Oe.
5. Oxide nanoparticles such as TiO2 and Cr2O3 can absorb and scatter UV radiation by dispersing them into polymer or pigment, protect the plastics against color degradation, or can be used for solar protection creams. Colby et al. (1993) prepared Au/polyethylene composites by orienting nanometer sized Au (60 nm) cylinder array on polyethylene polymer and observed the dependence of absorption maximum on the polarization angles of incident light in the composite where the absorption maximum shifts from 550 nm to longer wavelengths (800 nm) when the polarization angles of incident light with 330 nm wavelength are changed from the direction perpendicular to Au cylinder axes to parallel to the axes.
6. Optical application: Semiconductor nanoclusters (referred to as quantum dots) possess chemical and physical properties that differ substantially from those of analogous bulk solid. Nanoclusters are highly reactive and must be passivated by dispersion or confined in insulated matrices or hosts in the practical application. There are many kinds of matrices such as zeolites, porous glass, micelles, membranes and anionic polymers. CdS nanoparticles have promise as photoluminescent properties. Dendrimer is a quasi-spherical organic polymer that has well-defined structures with less dense interiors and densely packed surfaces. CdS/dendrimer nanocomposites that contain CdS nanoclusters of 2–4 nm show significant absorption as UV light, which has a promising application as UV protection film.
12.6 Summaries of Nanocomposites
Most dense nanocomposites are studied with the intent of improving mechanical properties, and achievements in strength enhancement as well as hardness and wear resistance have been reached in ceramic-based nanocomposites including nanoceramic/ceramic composites and nanometallic/ceramic composites. Mechanical property improvements of nanodispersoids are attributed to grain refinement and residual stress optimization. Nanoceramic composites that exhibit good plasticity provide a promising technology to forge the hard ceramic materials into complicated forms. Toughness improvement of nanoparticles in a ceramic matrix has been expected but the practical progression is still underway. Metallic-based nanocomposites are mostly produced by mechanical alloying method (for structural application) and vapor-medium deposition method (for magnetic and related functional application). Improvements in magnetic properties have excelled in recent years. Polymer-based nanocomposites are composed of two groups. One uses transparent/formable polymer as the passive matrix or substrate for optical materials. The other uses finer inorganic powders as fillers to improve the mechanical properties of matrices. Further approaches to nanocomposites probably include assembling nanoparticles and embedding
109
them into heteromatrices to form multiple dimension nanocomposites. The unique electric, magnetic and optical properties of nanocomposites will result in new applications in the coming years.
For more detail the following reviews are recommended:
1. Nanocomposites with improved mechanical properties: Krstic (1998), Sternitzke (1997), Reimanis (1997), Rigueiro et al. (1998), and Niihara (1993);
2. Nanocomposites for functional materials: Bhaduri and Bhaduri (1998), Rittner and Abraham (1998), and Kruis et al. (1998).
References
Anya C. C.. Ceramics International. 24, 533 (1998)
Awano M.. Seramikkusu. 32, 997 (1997)
Bhaduri S. and S. B. Bhaduri. JOM January 44 (1998)
Borsa C. E., S. Jiao, R. I. Todd and R. J. Brook. J. Microscopy. 177, 305 (1994)
Calvin V. L.. Nature. 374, 354 (1994)
Coble K. L.. J. Appl. Phys.. 34, 1679 (1963)
Colby A. F., L. Hornyak Jr., J. A. Stockert and C. R. Martin. MRS Proceedings. 286, 431 (1993)
Davidge R. W. and T. J. Green. J. Materials Sci. 3, 629 (1968)
Edelstein A. S. and R. C. Cammarata. Nanomaterials. Institute of Physics Publishing, 577 (1996)
El-shall M. S. and A. S. Edelstein. Nanomaterials. ed. by. Edelstein A. S., 34 (1996)
Fantozzi G. and C. Olagnon. Material Science and Engineering ed. 11, 185 (1996)
Gangopadhyay S., G. C. Hadjipanayis, B. Dale, C. M. Sorensen, K. J. Klabunde, V. Papaefthymiou and A. Kostilas. Phys Rev.. B 45, 9778 (1992)
Gao H. J., Z. Q. Xue, Q. D. Wu and S. J. Pang. J. Vac. Sci. Tech.. B 13, 1242 (1995)
Giannelis E. P.. Appl. Organometal Chem.. 12, 675 (1998)
Kakawa Y.. Tailoring Ceramic Composites (Japanese). Agune Shyofusha Publishing (1990) 96
Karch J., R. Biringer and H. Gleiter. Nature. 330, 556 (1987)
Kodama H. and T. J. Miyoshi. Am. Ceram. Soc.. 73, 3081 (1990)
Kornmann X., Berglund L. A. and J. Sterte. Polymer Engineering and Science. 38, 1351 (1998)
Krstic V. D.. Materials Science Forum. 282–283, 209 (1998)
110
Kruis F. E., H. Fissan and A. J. Peled. Aerosol Sci. 29, 511 (1998)
Levin I., W. D. Kaldan, D. G. Brandon, T. Wieder. Acta Metall Mater. 42, 1147 (1994)
Li J. B., H. Matsubara, H. Yanagida.. Proceedings of the JFCC International workshop on Fine Ceramics 2000, Nagoya, Japan, 15–17th March (2000)
Newnham R. E., D. P. Skinner, and L. E. Cross.. Materials Research Bull. 13, 525 (1978)
Niihara K. and A. Nakahira. Proceedings of the Third International Symposium on Ceramic Materials and Components for Engines, ed. by V. J. Tenmery. American Ceramic Society (1988)
Niihara K., A. Nakahira, A. Sasaki and M. Hirabayashi. MRS proceedings (1989) 4, 129
Niihara K., A. Nakahira. Advanced Structrural Inorganic Composites, ed. by Vincenzizi P., Elsevier Publisher (1990)
Niihara K. J.. Ceram. Soc of JPN 99, 974 (1991)
Niihara K., A. Nakahira and Sekino T. MRS Proceedings 286, 405 (1993)
Oh S. T., M. Sando and K. J. Niihara. Am. Ceram Soc.. 81, 3013 (1998)
Reimanis I. E.. Mater. Sci. Eng.. A237, 159 (1997)
Rigueiro J. P., J. Y. Pastor, J. Llorca, M. Ellces, P. Miranza, J. S. Moya. Acta Mater.. 46, 5399 (1998)
Rittner M. N. and T. Abraham. JOM, January 36 (1998)
Roy R. MRS Proceedings 286, eds. by Sridhar Komarneri, et al. 242 (1993)
Sohn B. H., R. E. Cohen, G. C. Papaefthyniou. J Magn. & Magn. Mater.. 182, 216 (1998)
Sternitzke M. J.. Eur. Ceram. Soc. 17, 1061 (1997)
Thompson A. M., H. M. Chan, M. P. Harmer and R. F. Cook. J. Am. Ceram. Soc.. 78, 567 (1997)
Veprek S.. Haussmann and Reiprich S. J. Vac. Sci. Tech.. A14, 46 (1996)
Wakei F., S. Sakaguchi and Y. Matsuno. Advanced Ceramic Materials. 1, 259 (1986)
111
13. Growth and Properties of Single-Walled Nanotubes
13.1 Introduction
Carbon nanotubes have attracted much attention in the past several years because of their unique and potentially useful structural, electrical and mechanical properties (Dresselhaus, et al., 1996; Ebbesen, 1996; Yakobson and Smalley, 1997; Dekker, 1999). A nanotube has high Young's modulus and tensile strength, and can be metallic, semiconducting or semimetallic depending on the helicity and diameter (Dresselhaus, et al., 1996). Utilization of these properties with individual or ensembles of nanotubes have led to advanced scanning probes (Dai, et al., 1996, 1998; Wong, et al., 1998b, 1998a; Hafner, et al., 1999) nano-electronic devices (Tans, et al., 1998b; Martel, et al., 1998; Soh, et al., 1999) and electron field emission sources (de Heer, et al., 1995; Bonard, et al., 1998; Collins and Zettl, 1996; Saito, et al., 1998; Wang, et al., 1998).
Arc discharge (Iijima, 1991; Iijima and Ichihashi, 1993; Bethune, et al., 1993; Journet, et al., 1997) and laser ablation (Thess, et al., 1996) have been the principal methods for obtaining high quality nanotube materials. However, there are several key issues concerning both methods. The first is that these methods involve evaporating carbon atoms from solid carbon sources at ≥3000°C. This sets a limitation to the quantity of nanotubes that can be synthesized, and it remains unclear how to scale up nanotube production to the kilogram level using the evaporation approaches. The second issue relates to the fact that evaporation methods grow nanotubes in highly tangled forms mixed with unwanted carbon or metal species. The nanotubes are difficult to purify, manipulate and assemble for building nanotube device structures.
Developing controlled synthesis methods to obtain ordered carbon nanotube architectures is an important and viable route to fundamental characterizations and potential applications of nanotube based molecular wires. The ultimate goal in nanotube synthesis should be gaining control over the locations and orientations of nanotubes, as well as their atomic structures including helicity, diameter, and topological defects. Recently, significant progress has been made in controlling the growth of multi-walled carbon nanotubes (MWNTs) on surfaces using chemical vapor deposition (CVD) methods. Long and well-aligned MWNTs on large-scale substrates have been synthesized (Li, et al., 1996; Pan, et al., 1998; Ren, et al., 1998; Fan, et al., 1999). Nevertheless, the growth of single-walled nanotubes (SWNTs) into ordered architectures has been challenging.
This chapter presents our recent development of controlled CVD synthesis strategies to obtain ordered multi-walled and single-walled nanotube structures. The key approaches include understanding the chemistry of catalyst materials and nanotube growth, and synthesizing nanotubes on rationally designed substrates containing patterned catalyst. The results include MWNTs self-oriented perpendicular to substrates and assembled into regular arrays. By enabling a CVD method for structurally perfect SWNTs, individual SWNTs are grown on specific sites on surfaces. Also, for the first time, suspended SWNT architectures with nanotubes directed towards well-defined orientations are synthesized.
The nanotube architectures derived by controlled chemical synthesis have opened new possibilities in fundamental characterizations and potential applications of novel nanowire materials. In particular, single-walled nanotubes are true molecular wires with their diameters in such a regime (about 1–5 nm) that
112
the electronic structure of a SWNT sensitively depends on its chirality. SWNTs serve as ideal systems to study physics problems in quasi one dimension. To this end, we will show that our growth strategy readily allows SWNTs to be integrated into electrical circuits and addressed individually. Systematic electron transport measurements are carried out to elucidate the intrinsic electrical properties of various classes of nanotubes. Also, functional electronic structures based on individual SWNTs with comparable characteristics as conventional silicon devices are demonstrated.
13.2 Synthetic Strategies for Various Nanotube Architectures
13.2.1 Chemical Vapor Deposition
Chemical vapor deposition of hydrocarbons over metal catalysts has been a classical method to produce various forms of carbon fibers, filaments and multi-walled nanotubes in the past (Tibbetts, 1990, 1983; Endo, 1988; Snyder, et al., 1989; Baker and Rodriguez, 1994). The typical growth temperature Tg is typically 500 °C ≤ Tg ≤ 1000 °C. The first step in a CVD process involves the absorption and decomposition of hydrocarbon molecules on transition-metal (Fe, Ni, Co, etc.) catalytic particles. The carbon atoms diffuse into the interior of the catalyst to form a metal-carbon solid state solution (Baker, 1989; Tibbetts, et al., 1987; Tibbetts, 1984). Subsequent precipitation of carbon from the supersaturated catalyst particle will then occur and lead to the formation of a carbon tube structure (Fig. 13.1). Typically, two modes of nanotube growth can operate in CVD. The base-growth mode involves the metal catalyst particle pinned on the support substrate, and the nanotube lengthens with a particle-free closed end. Carbon feedstock is supplied from the base where the nanotube interfaces with the catalyst material (Fig. 13.1, left panel). The tip-growth model involves a metal catalyst particle at a nanotube end being carried away as the nanotube lengthens (Fig. 13.1, right panel). The carried-along particle is responsible for supplying carbon feedstock needed for the nanotube growth. For the synthesis of nanotubes, the catalytic metal nanoparticles are typically obtained on high surface area support materials such as Al2 O3 and SiO2. The size of the catalytic particles determines the size of the nanotubes. Multi-walled or single-walled can be synthesized by CVD depending on the particle size, as well as the type of hydrocarbon feedstock and growth conditions.
Figure 13.1 Schematic growth modes of carbon nanotubes in CVD. Single-walled nanotubes are shown as examples. Left panel: base-growth mode. Right panel: tip-growth mode.
113
Notably, a pitfall of CVD synthetic approaches has been that defective tubular carbon materials tend to be formed. Only recently, we have developed a CVD approach to grow nearly perfect SWNTs by using methane as the carbon feedstock (Kong, et al., 1998a, 1998b). This result will be presented later in the chapter.
13.2.2 Growth of Self-oriented Multi-walled Nanotubes
In controlling the orientation of nanotubes during CVD growth, previous methods have relied on growth of nanotubes in confined environments including the pores of mesoporous silica or channels of alumina membranes (Li, et al., 1996; Pan, et al., 1998; Che, et al., 1998; Kyotani, et al., 1996). We have found that nanotubes can self-assemble into aligned structures during CVD growth, and the driving force for self-alignment is the van der Waals interactions between nanotubes (Fan, et al., 1999). Our synthesis approach involves catalyst patterning and rational design of the substrate to enhance catalyst-substrate interactions and control the catalyst particle size. The substrates are porous silicon obtained by electrochemical etching of n-type silicon wafers in HF/methanol solutions. The resulting substrate consists of a thin nanoporous layer (pore size 3 nm) on top of a macroporous layer (with submicron pores) (Vial and Derrien, 1994; Smith and Collins, 1992). Patterned catalyst squares on the porous silicon substrate are obtained by evaporating a 5 nm thick iron film through a shadow mask. CVD growth using the substrate is then carried out in a 2 in. tube furnace at 700°C under an ethylene flow of 1000 sccm/min for 15–60 min. Figure 13.2(a) shows a scanning electron microscope (SEM) image of regularly spaced arrays of nanotube towers grown on top of patterned iron squares on a porous silicon substrate. The nanotube towers exhibit very sharp edges and corners with no nanotubes branching away from the blocks. The high resolution SEM image (Fig. 13.2(b)) reveals that the MWNTs (Fig. 13.2(b), inset) within each block are well aligned along the direction perpendicular to the substrate surface. The length of the nanotubes and thus the height of the
114
towers can be controlled in the range of 10–240 µm by varying the CVD reaction time, and the width of the towers is controlled by the size of the openings in the shallow mask. The smallest self-oriented nanotube towers synthesized by our method are 2 µm × 2 µm wide.
Figure 13.2 (a) Scanning electron microscopy image of arrays of bundled multi-Walled nanotube towers. (b) A high resolution SEM showing aligned MWNTs within a tower. The inset shows a TEM image of the bundled MWNTs.
The mechanism of nanotube self-orientation involves the nanotube base-growth mode (Fan, et al., 1999). Since the nanoporous layer on the porous silicon substrate serves as an excellent catalyst support, the iron catalyst nanoparticles formed on the nanoporous layer interact strongly with the substrate and remain pinned on the surface. During CVD growth, the outmost walls of nanotubes interact with their neighbors via van der Waals forces to form a rigid bundle, which allows the growth of nanotubes perpendicular to the substrate. The porous silicon substrates exhibit important advantages over plain silicon substrates in the synthesis of self-aligned nanotubes. Growth on substrates containing both porous silicon and plain silicon portions find that nanotubes grow at a higher rate (in length/min) on porous silicon than on plain silicon. This result suggests that ethylene molecules can permeate through the macroporous silicon layer and thus efficiently feed the growth of inner and outer nanotubes within the towers. The nanotubes grown on porous silicon substrates exhibit monodispersed diameters since catalyst nanoparticles with a narrow size distribution can be formed on the porous supporting surface, and the strong catalyst-support interactions prevent the catalyst particles from sintering at elevated temperatures during CVD growth.
13.2.3 Enable the Growth of Single-Walled Nanotubes by CVD
Chemical vapor deposition methods have been very successful in synthesizing carbon fibers, filaments and MWNTs (Tibbetts, 1983; 1990; Endo, 1988; Baker and Rodriguez, 1994; Snyder, et al., 1989). However, CVD synthesis of high quality SWNTs is only recent. Structurally perfect SWNTs can now be grown in a CVD process using methane as carbon feedstock and iron-oxide nanoparticles supported on high surface
115
area alumina as the catalyst (Kong, et al., 1998a). High temperature conditions (850–1000°C) are employed in the growth in order to overcome high strain energies in forming small diameter SWNTs (<5 nm), and obtain nearly defect-free tube structures. The choice of methane is critical to the CVD approach to SWNTs. We have found that methane is stable at elevated growth temperatures without appreciable self-pyrolysis. This stability prevents the formation of amorphous carbon that tends to cause catalyst poisoning and overcoating the nanotubes. Catalytic decomposition of methane by the transition-metal catalyst particles is thus the dominant process in SWNT growth (Kong, et al., 1998a, 1998b; Cassell, et al., 1999b).
Within the methane CVD approach, we find that the chemical and textural properties of the catalyst materials dictate the yield and quality of SWNTs (Cassell, et al., 1999b). Bulk quantities of high quality SWNTs can be synthesized by optimizing the catalyst. Thus far, our optimized catalyst consists of Fe/Mo bimetallic species supported on a sol-gel derived alumina-silica multicomponent material (Cassell, et al., 1999b). Shown in Fig. 13.3 is a transmission electron microscopy (TEM) image of SWNTs synthesized in bulk by using this catalyst. The image illustrates remarkable abundance of individual and bundled SWNTs that are free of defects and amorphous carbon coating. The diameters of the SWNTs are dispersed in the range of 0.7–5 nm with a peak at 1.7 nm. Weight gain studies find that the yield of nanotubes can be as high as 45 wt%. Through systematic studies, we have found that a good catalyst material for SWNT synthesis necessarily exhibits strong metal-support interactions, possesses a high surface area and large pore volume, and retains these characteristics at high temperatures without sintering. The strong metal-support interactions allow high metal dispersion and thus a high density of catalytic sites. The open pore structure of a catalyst allows efficient diffusion of reactant and intermediate hydrocarbon species. We believe that the rate-limiting step in SWNT CVD growth involves gas diffusion. This is based on the results that high SWNT yielding catalysts exhibit large pore volumes in the mesopore regime (Cassell, et al., 1999b).
Figure 13.3 A TEM image of SWNTs synthesized in bulk using a catalyst supported on a sol-gel derived alumina-silica hybrid material. Inset: an example of the frequently observed SWNT ends that are closed and free of metal particles.
116
13.2.4 Growth Mechanism of SWNT
The states of nanotube ends contain rich information about nanotube growth mechanisms. Careful high resolution TEM imaging of the SWNTs synthesized by our CVD method frequently observes closed tube ends that are free of attached or encapsulated metal particles (Fig. 13.3 inset). The opposite ends are typically found embedded in the catalyst support particles when imaged along the lengths of the nanotubes. These observations suggest that SWNTs grow in the methane CVD process predominantly via the base-growth process (Fig. 13.1, left panel) (Tibbetts, 1983, 1989, 1990; Tibbetts, et al., 1987; Tibbetts, 1984; Baker, 1989; Amelinckx, et al., 1994; Kong, et al., 1998a, 1998b; Cassell, et al., 1999b). Base-growth operates when strong metal-support interactions exist so that the metal species remain pinned on the support surface. In contrast, the tip-growth mode operates when the metal-support interaction is weak. In the methane CVD method, we find that enhancing metal-support interactions leads to significant improvement to the performance of the catalyst material in producing high yield SWNTs. This is rationalized by the increased catalytic sites and the facilitated base-mode growth processes. On the other hand, catalysts with
117
weak metal-support interactions lead to aggregation of metal species and reduced nanotube yield and purity (Cassell, et al., 1999b). Further understanding of the chemistry of catalysts and nanotube growth will undoubtedly lead to the synthesis of bulk quantities of high quality SWNTs approaching the kilogram scale.
13.2.5 Growth of Isolated Single-Walled Nanotubes on Controlled Surface Sites
The successful CVD synthesis of SWNTs in bulk forms has led to a straightforward synthetic route to addressable individual nanotube wires. By using substrates patterned with 1–5 µm wide catalytic islands, we obtain "nanotube chips" that contain isolated single-walled nanotubes grown from desired locations on the substrates (Kong, et al., 1998b, 1999; Soh, et al., 1999). Atomic force microscopy (AFM) images of SWNTs on a nanotube-chip are shown in Fig. 13.4, where the synthesized nanotubes extending from the catalyst islands are clearly observed. The diameters of the nanotubes are measured to be in the range of 0.7–4.0 nm, which is consistent with TEM results obtained with bulk SWNT materials. Some of the nanotubes have one end attached to a catalyst island and the other end terminated between islands. Nanotubes bridging islands with both ends attached to the opposing islands are also observed. As described in a later section, the bridging SWNTs allow reliable electrical connections to be made from the macroscopic scale to individual SWNTs. Thus, our controlled chemical synthesis opens up a new route to individual nanowire electrical circuits that are needed for fundamental and practical purposes.
Figure 13.4 (a) An AFM image of SWNTs grown from patterned catalyst islands on a silicon oxide substrate. (b) Image of an individual SWNT bridging adjacent islands.
118
119
13.2.6 Growth of Suspended SWNTs with Directed Orientations
Obtaining single-walled carbon nanotube architectures with nanotubes in aligned orientations has been challenging. We have devised a synthetic strategy that leads to suspended SWNTs directed towards controlled orientations parallel to the plane of a silicon substrate. The SWNTs are suspended bridges grown from catalyst material placed on top of regularly patterned silicon tower structures. The synthesis approach begins with developing a series of liquid-phase catalyst precursor materials that allow for uniform film formation and large-scale catalyst patterning. A specific precursor material consists of ethanol (40 mL) and 2-butanol (20 mL) solutions of P-123 block copolymer (1.0 g) (Yang, et al., 1998a, 1998b), AlCl3 · 6H2O (2.4 g), FeCl3 · 6H2O (0.09 g) and MoO2Cl2 (0.004 g). Using contact printing (Xia and Whitesides, 1998; Ferguson, et al., 1991) of a flat PDMS stamp inked with a film of the precursor material, we selectively place the precursor on top of tower arrays pre-made on a silicon substrate (Cassell, et al., 1999a). Calcination at 700°C for 4 h leads to the formation of alumina/silica mixed oxides confined on the tower tops. Subsequent CVD growth using the substrate yields SWNTs emanating from the towers. Directed free-standing SWNT networks are formed by nanotubes growing to adjacent towers and suspended above the surface. When examining with an SEM, we observe that highly directional suspended SWNTs are formed on the synthesized sample. The directions of the suspended tubes are determined by the pattern of the towers. Well-aligned SWNT bridges are obtained in an area of the substrate containing isolated rows of towers as shown in Fig. 13.5(a), where suspended tubes forming a power-line-like structure can be seen. In an area containing towers in a square configuration, a square of suspended nanotube bridges is obtained (Fig. 13..5(b)). Directionality of the suspended tubes is simply a result of the rationally designed substrate. During the CVD growth, nanotubes emanate from the top of the towers. The nanotubes growing towards adjacent towers become suspended, whereas nanotubes directed towards other orientations fall onto the sidewalls of the towers (not easily resolved under SEM). In Fig. 13.5(c), we show a TEM image of a suspended SWNT bridge between silicon towers, and an image (Fig. 13.5(c), inset) showing the high resolution structure of the SWNT.
Figure 13.5 (a) SEM image of a suspended SWNT "power-line-like" structure. (b) SEM image of a square of suspended SWNT bridges. (c) TEM image of a SWNT bridge suspended between silicon towers. Inset: a high magnification TEM image showing the structure of a SWNT.
120
121
The directed growth of suspended SWNTs presented here involves developing a new type of liquid phase catalyst material, contact printing of catalyst onto designed substrates and CVD synthesis. The method should open a new window in characterization and device applications of organized nanowire architectures in suspended states or after being transferred onto flat substrates.
13.3 Physics in Atomically Well-Defined Nanowires
It is well recognized that single-walled carbon nanotubes are ideal systems for studying solid-state physics in quasi one dimension (Dresselhaus, et al., 1996). SWNT wires have well-defined atomic structures and can be considered as molecule-like wires (Dekker, 1999). It has been actively pursued to elucidate their band structures (Dresselhaus, et al., 1996; Saito, et al., 1992; Mintmire, et al., 1992; Hamada, et al., 1992; Odom, et al., 1998; Wildoer, et al., 1997), quantum confinement (Tans, et al., 1997; Bockrath, et al., 1997; Cobden, et al., 1998), electron-electron (Tans, et al., 1998a; Bockrath, et al., 1999; Kane, et al., 1997; White and Todorov, 1998), and electron-lattice (Kane, et al., 1998; Kane and Mele, 1997) interaction effects in SWNTs. The remarkable sensitivity of a SWNT electronic structure on tube chirality and diameter requires addressing individual nanotubes in order to understand electron transport in various classes of nanotubes. Previous approaches to addressable SWNT electrical devices include randomly depositing SWNTs from liquid suspensions onto pre-defined electrodes (Tans, et al., 1997) or onto a flat substrate followed by locating nanotubes and patterning electrodes (Bockrath, et al., 1999). Results obtained with individual single-walled tubes and ropes include Coulomb charging (Tans, et al., 1998a; Bockrath, et al., 1999) and Luttinger liquid behavior (Bockrath, et al., 1999) in metallic tubes. Semiconducting SWNTs have been found to exhibit field-effect transistor characteristics at room temperature (Tans, et al., 1998b; Martel, et al., 1998; Soh, et al., 1999; Kong, et al., 1999). Nevertheless, the intrinsic electrical properties of various types of nanotubes remain to be fully understood.
13.3.1 Integrated Circuits of Individual Single-Walled Nanotubes
The synthesis of individual SWNTs at desired surface sites has enabled a controlled route to addressable SWNTs for systematic electrical measurements of nanotubes. With the SWNT chips described earlier, integrated nanotube circuits can be constructed by using a straightforward microfabrication procedure (Soh, et al., 1999; Kong, et al., 1999). Various types of metals are used to contact nanotubes including Ti, Ni, Nb, Cr, Au, Al and Ag. The contacting process places metal electrodes to fully cover the catalyst islands and extend over the edges by 0.5–1 µm. The individual SWNT bridges between islands leads to a significant number of metal-SWNT-metal electrical devices. Optical and AFM images of a representative device are shown in Fig. 13.6 In our samples, the lengths of individual SWNTs between electrodes were ≥3 µm. Degenerately doped silicon wafers with 500 nm thick thermally grown oxide on the surfaces are used as substrates. The heavily doped substrate is conducting at low temperatures and used as a backgate. Figure 13.6(c) shows the schematic structure of our devices.
Figure 13.6 (a) Optical image of SWNT integrated circuits. (b) AFM image of an individual SWNT device. (c) A schematic view of the device structure.
122
123
13.3.2 Electron Transport Properties of Metallic Nanotubes
Single-walled armchair nanotubes with (n, n) indices are true metallic wires with finite density of states at the Fermi level (Dresselhaus, et al., 1996). At low temperatures, a perfect (n, n) armchair tube is conducting, and transport along the length of the tube could be ballistic due to quenched phonon modes and the defect- and impurity-free nature of the nanotube. If ohmic contacts are made between metal electrodes and the tube, a resistance of 6.5 kΩ, i.e., half of the resistance quantum h/e2, can be expected (Chico, et al., 1996). However, it has been challenging to make good electrical contacts to SWNTs. At low temperatures, Coulomb blockade has been observed in individual metallic single-walled tubes or ropes due to high contact resistance on the order of mega-ohms (Tans, et al., 1998a; Bockrath, et al., 1999). Coulomb charging effects have been the main physics studied in these samples (Tans, et al., 1998a; Bockrath, et al., 1999).
Using the approach of controlled SWNT synthesis and contacting, we have obtained a large number of low resistance individual SWNT samples in the range of tens to hundreds of kilo-ohms (length between the edges of contacting electrodes ≥3 µm). These individual SWNT samples remain to be in low resistance states when cooled to low temperatures. The lowest single-tube resistance is 12 kΩ measured at 2 K. Detailed transport data obtained with this sample are shown in Fig. 13.7. The linear resistance of the sample decreases as the temperature is decreased, and a slight upturn is observed below 30 K in the resistance vs. temperature curve (solid line in Fig. 13.7(a), measured under zero gate voltage). Applying a - 10 V gate voltage is found to lower the resistance of the sample (dashed line in Fig. 13.7(a)), and the resistance is 12 kΩ at 2 K. As shown in Fig. 13.7(b), current vs. voltage I-V curves recorded under Vg = - 10 V at room temperature and 4 K exhibit linear characteristics near zero bias voltage. At 2 K, the sample exhibits clear stair-case structures in the I-V (dotted line in Fig. 13.7(b)). However, the origin of these interesting features remains unclear at the present time. In Fig. 13.7(c), current vs. gate voltage curves measured under various bias voltages show fluctuations but the absence of periodic Coulomb oscillations. Sweeping the gate voltage from - 100 to 100 V find that the sample remains conducting in the entire experimentally accessible gate voltage range. These results suggest that the SWNT is of true metallic nature and correspond to a (n, n) armchair type. Coulomb blockade behavior is not observed in this sample because highly transparent contacts have been made to the nanotube.
Figure 13.7 (a) Resistance vs. temperature R(T) for a metallic SWNT contacted by 20 nm/60 nm Ti/Au. Solid line: R(T) measured under zero gate voltage. Dashed line: R(T) measured under -10 V gate voltage. (b) I-V curves recorded at 300 K, 4 K and 2 K respectively. Gate voltage =-10 V. (c) Current vs. gate voltage curve recorded at 2 K under various bias voltages.
124
125
Among the various contacting metals investigated for SWNTs, we find that Ti tends to give rise to the lowest contact resistance. Other types of metal typically lead to higher resistance. The low contact resistance is also attributed to the controlled approach of nanotube growth and integration that allow metal electrodes coupling to the sides and the ends of a nanotube. It is expected that the broken translational symmetry at a nanotube end allows strong electrical coupling between the nanotube and a metal (Tersoff, 1999), which could be a significant factor in the observed low contact resistance of our samples. It remains a challenge, however, to address the precise nature of the contacts and achieve the 6.5 kΩ resistance limit in metallic SWNT systems. It is a common observation with our samples that upon varying gate voltage, the resistance of metallic SWNTs can be changed by a factor of 2–4. This gate dependence is far from negligible although much weaker than semiconducting SWNTs. We attribute this phenomenon to the fact that the transparency of metal-tube contacts (or junctions) is tunable by gate voltages. This is based on recent experiments that have observed superconducting proximity effect in SWNT samples contacted by superconducting Nb electrodes (Morpurgo, et al., 1999). In this case, only under certain gate voltages are pronounced proximity effects observable, suggesting enhanced metal-tube contact transparency by these gate voltages (Morpurgo, et al., 1999). The gate dependence of metal-tube coupling is also consistent with recent numerical calculations that found contact resistance to a SWNT depending on the Fermi energy of the tube (Rochefort, et al., 1999). The origin of non-periodic fluctuations in the current vs. gate voltage data for the metallic SWNT (Fig. 13.7(c)) at low temperatures remains to be fully understood. Some of the variations could be due to metal-tube transparency changes under various gate voltages.
13.3.3 Electron Transport Properties of Semiconducting Nanotubes
Chiral (m, n) SWNTs with m-n ≠ 3 × integer are semiconducting in nature and have primary energy gaps Eg ∞ 1/d, where d is the nanotube diameter (Dresselhaus, et al., 1996). Transport studies of this class of nanotubes have been limited to 300K, and temperature-dependent electrical properties remain unexplored so far (Tans, et al., 1998b; Martel, et al., 1998). Furthermore, fundamental issues such as the origin of carrier type, nature of contact and transport mechanisms in semiconducting SWNTs have not been addressed. Our growth method produces SWNTs with diameters dispersed in the range of about 0.7–3 nm, and the resulting semiconducting tubes have energy gaps about 1–0.25 eV according to band structure calculations (Dresselhaus, et al., 1996). We present in this section temperature-dependent transport data of individual semiconducting SWNTs of various tube diameters. New possible mechanisms of hole-doping to nanotubes and the nature of metal-tube contacts are proposed. Also, transport mechanisms through semiconducting SWNTs at various temperatures are elucidated.
We find that semiconducting SWNT samples exhibit common characteristics but differ quantitatively. The room temperature resistance is typically in the range of 200–500 kΩ for samples with relatively large diameter (d>2.0 nm) tubes. Smaller diameter tube samples with d≤1.5 nm exhibit higher resistance on the order of mega-ohms or higher. The semiconducting tubes appear to be hole-doped and their conductance can be diminished by applying positive gate voltages. At 4 K, the samples are insulating and show gap-like region within ± 10 mV to ± 1000 mV in the I-V curves. The gap region is found to be larger for smaller diameter tubes. Temperature-dependent resistance of the samples shows that transport through the semiconducing tubes involves thermal activation with higher barriers for smaller diameter tubes.
126
Figure 13.8(a), (b) show the room temperature I-V curves obtained with a d = 2.8 nm length 3 µm semiconducting SWNT (sample # S1). The sample exhibits a highly linear I-V with resistance 370 kΩ measured at zero gate voltage. Positive gate voltages progressively reduce the linear conductance of the sample (Fig. 13.8(a)). At Vg>3 V, the conductance is suppressed by four orders of magnitude from that at Vg = 0. These I-V characteristics are signatures of hole-doped semiconducting SWNTs acting as p-type transistors as reported previously (Tans, et al., 1998b; Martel, et al., 1998; Soh, et al., 1999; Kong, et al., 1999). The result of resistance vs. temperature measured by using near zero bias voltages ( V ≤1 mV) under zero gate voltage is shown in the inset of Fig. 13.8(b). The temperature dependence of resistance is weak from 300 K to 25 K, with an increase from 370 kΩ to 1 MΩ. Below 25 K, the resistance increases sharply and R(T) can be approximately fitted to exp(-Ea/KBT) with an activation barrier of Ea 4.6 meV. At 4 K, a gap 20 mV is observed in the I-V curve and the sample is insulating in the bias range V ≤ 10 mV (Fig. 13.8(b)). The insulating region is found to be significantly suppressed by applying a -1.5 V gate voltage. On the other hand, applying a positive gate voltage leads to a larger insulating region in the I-V curve (Fig. 13.8(b)).
Figure 13.8 (a) Room temperature I-V characteristics of a d = 2.8 nm semiconducting SWNT (sample # S1) contacted by 20 nm/60 nm Ni/Au. (b) I-V curves recorded at 4 K for sample # S1 under various gate voltages. Inset: linear resistance vs. temperature. (c) I-V curve for a d = 1.3 nm semiconducting SWNT (sample # S2) with Ni contacts recorded at 4 K. Inset: room temperature I-V characteristics. (d) Linear resistance vs. temperature. Inset: plot of resistance in log scale vs. 1/T.
127
In Fig. 13.8(c), (d), we show the results obtained with a d = 1.3 nm and length 5 µm semiconducting SWNT (sample #S2). The room temperature linear resistance of the sample is 3.4 MΩ under zero gate and is dramatically increased by positive gate voltages (Fig. 13.8(c) inset). Upon cooling, the I-V curve becomes nonlinear and the resistance is infinity below 40 K due to the appearance of a gap-like insulating region in I-V. At 4 K, the insulating region is within ±0.7 V as shown in Fig. 13.8(c). The linear resistance of the sample sharply increases as temperature decreases, and can be well fitted into an exponential form R (T) exp (- Ea/KB T) with an activation barrier Ea 25 meV (Fig. 13.8(d)).
A band diagram model is presented in Fig. 13.9 for a metal-semiconducting tube-metal system to rationalize the obtained transport results. First, we consider that the nanotube is uniformly hole-doped along its entire length. Hole-doping to nanotubes has been previously attributed to electron transfer from nanotubes to metal electrodes due to different work functions (Tans, et al., 1998b). In our samples, it is not plausible that the effect of work function mismatch (=Ni = 5.5 eV vs. =NT 4.5 eV) can extend over the 3 µm long tube. Furthermore, we have used titanium, aluminum, and silver that have similar work functions as graphite, and magnesium and calcium with lower work functions than graphite for the contacting metals. The resulting semiconducting tube samples all appear to be hole-doped and show p-type transistor behavior. Thus, metal/tube work function mismatch does not appear to be the origin of hole-doping in our samples. Nevertheless, a complete understanding of the precise doping mechanism requires further investigations. We tentatively propose that a possible doping source could be charged species on the sample surface or trapped charges in the substrate near the SWNT. Stray or trapped charges are known to exist in SiO2/Si systems and affect the electrical characteristics of conventional semiconductor devices (Sze, 1981). It is also possible that interactions between a semiconducting nanotube and molecular species in the environment (including surface hydroxyl groups and other species on the SiO2) lead to the observed hole-doping effects. Such doping mechanisms act over the entire length of a nanotube, instead of being localized near the contacts. Note that homogeneous hole-doping to SWNTs was also believed to be the case in a previous study (Martel, et al., 1998). Secondly, we propose that each contact consists of a series resistance (black bar in Fig. 13.9) and forms a Schottky-like junction with the p-type nanotube bridge. The junction barrier height approximately equals to the separation between the nanotube Fermi level and the valence band EFV = EF - EV.
Figure 13.9 A proposed band diagram for a metal-semiconducting SWNT-metal system. The junction barrier height Ea is determined by the separation between the nanotube Fermi level and the valence band EFV = EF - EV.
128
The junction barrier is responsible for the observed thermally activated transport through semiconducting SWNTs, and is determined to be Ea EFV about 4.6 meV and 25 meV respectively for samples #S1 and #S2. The expected energy gap for the d = 2.8 nm and 1.3 nm tubes are Eg 0.2 and 0.6 eV respectively according to band structure calculations. Thus, the semiconducting tubes are hole-doped to large degrees. At 300 K, the samples exhibits linear I-V curves (under zero gate voltage) since the junction barriers can be overcome by thermal energy Ea ≤ KB T = 26 meV. Positive gate voltages cause the valence band shifting down away from the Fermi level, leading to higher barriers and thus less conducting states as seen in Fig. 13.8(a), (c). At 4 K where KB T << Ea, thermally activated transport through the system is quenched. The sample is in an insulating state near zero bias as shown in Fig. 13.8(b), (c). Under significantly high bias voltages, the sample is turned into a conducting state. Analyses of the I-V curves (Fig. 13.8(b), (c)) find that current increases by three orders of magnitude after the turn-on, and can be fitted into I exp (- c/V) where c is a constant. These results suggest that electron transport through a semiconducting tube at low temperatures is via a tunneling mechanism. Under a high bias voltage, electron tunneling occurs through the reverse biased Schottky-like junction. The mechanism of thermally activated transport at high temperatures and transport via tunneling at low temperatures is similar to that observed in conventional metal-semiconductor-metal systems by Lepselter and Sze (Lepselter and Sze, 1968)
13.3.4 Electron Transport Properties of Semiconducting Nanotubes with Small Band
Gaps
WNTs with (m, n) indices m - n = 3 × integer are semimetallic with zero band gap within tight-binding calculations based on pπ-orbitals alone. For this type of SWNT, Hamada et al. (1992) and White et al. 1996) pointed out that the curvature of nanotubes leads to non-parallel pπ orbitals interacting with pπ-orbitals, which causes the opening of a small band gap to result in a semiconductor from a semimetal. Louie and co-workers carried out first-principles ab initio calculations and found that the curvatures of small diameter SWNTs can lead to rehybridization of π* and σ* orbitals and thus altered electronic structures of SWNTs from those of flat graphene stripes (Blase, et al., 1994). Kane and Mele categorized SWNT into three types, true metallic armchair SWNTs, semiconducting SWNTs, and curvature induced small gap semiconducting SWNTs (Kane and Mele, 1997). The band gaps for S-SWNTs scale as Eg 1/d about 0.2–1 eV for d = 3–0.7 nm, whereas for SGS-SWNTs, the band gaps scale as Eg 1/d2 about 2–50 meV for d = 3–0.7 nm (Kane and Mele, 1997). These theoretical work clearly pointed out that the curvature of SWNTs has non-trivial consequences to their electrical transport properties (Blase, et al., 1994; Kane and Mele, 1997).
Thus far, experimental work characterized SWNTs into metallic and semiconducting two general categories. In previous electrical transport experiments, SWNTs exhibiting significant conductivity at low temperatures were believed to be metallic. Semiconducting tubes were identified when their conductance can be significantly modulated by gate voltages at room temperature and the nanotubes become insulating at low temperatures (Tans, et al., 1997, 1998b; Bockrath, et al., 1997; Martel, et al., 1998; Soh, et al., 1999; Kong, et al., 1999; Nygard, et al., 1999). In scanning tunneling microscopy (STM) studies, metallic and semiconducting SWNTs were identified by probing the local density of states of nanotubes near the Fermi level (Odom, et al., 1998; Wildoer, et al., 1997). However, no experimental evidence was reported so far to prove the existence of small-gap semiconducting SWNTs predicted by theory. Large metal-SWNT contact resistance tended to cause Coulomb charging effects observed in transport measurements (Tans, et al., 1997;
129
Bockrath, et al., 1997), which has long been realized as an obstacle to elucidating the nature and intrinsic electrical properties of SWNTs (Nygard, et al., 1999). In this section, we present the observation of individual semiconducting SWNTs with band gaps on the order of 10 meV. We find that the electrical characteristics of small-gap semiconducting SWNTs are clearly distinguishable from that of true metallic SWNTs and semiconducting SWNTs with large primary band gaps. Controlled device fabrication for low metal-tube contact resistance is essential to the elucidation of the intrinsic electrical properties of SWNTs.
Figure 13.10 shows a tapping mode atomic force microscopy (AFM) image of an individual SWNTs exhibiting small-gap semiconducting characteristics. The diameter of the SWNT was 1.3 nm, determined by AFM topographic measurements. The current-voltage I-V curves obtained at room temperature are shown in Fig. 13.10(a). The I-V curves are highly linear, showing a resistance of 36 kα under zero gate voltage (Vg). Increasing Vg reduced the linear conductance of the sample and reached a minimum at Vg 5 V. As Vg was further increased, the conductance was observed to recover and increase with Vg. In Fig. 13.10(b), we show the conductance vs. gate voltage (dI/dV - Vg) curve recorded under a constant source drain bias of V = 1 mV. Consistent with the results in Fig. 13.10(a), the conductance of the sample was suppressed by initial increase in Vg and was lowered by 4 times at Vg 5 V, beyond which further increase in Vg recovered the conductance of the SWNT sample. This resulted in a valley in the conductance vs. gate voltage curve.
Figure 13.10 (a) Room temperature I-V characteristics of an SGS-SWNT. (b) dI/dV vs. Vg at room temperature. Inset: AFM image of the d 1.3 nm SWNT.
130
Upon cooling the SWNT sample to 2 K, we recorded 400 I-V curves in the source-drain bias range of V = -40 to 40 mV with an incremental step size of 400 µV, under gate voltages in the range of Vg = 0 to 20 V with an incremental step of 50 mV. A grey-scale 2-D conductance map was obtained by plotting the conductance values at all of the (V, Vg) points, as shown in Fig. 13.11(a). The conductance map exhibits symmetrical structures centered at V = 0 and Vg = V*
g 8.5 V. In the central region of the map, within V -8 mV to 8 mV and Vg 7.5 to 10 V, the sample conductance is highly suppressed and the corresponding resistance 5 MΩ. However, the sample is highly conducting in the corner regions of the map where V
131
> 10–20 mV and Vg 0 or 20 V. The corresponding resistance 20 kΩ, which is more than two orders of magnitude lower than that in the central region. In Fig. 13.11(b), we show conductance vs. bias voltage (dI/dV - V) curves recorded under various gate voltages. These conductance curves shift downwards as Vg increases from 0 to 8 V, then shift upwards upon further increase in Vg. The I-V curves are nonlinear near zero bias, as dips of reduced conductance are seen in dI/dV vs. V. Under Vg 8 V V*
g (i.e., horizontal centerline crossing the suppressed region in Fig. 13.11(a), the conductance is highly suppressed for small biases V < 8 mV in the dI/dV-V curve. The suppression is nearly exponential in V, indicating a gap-like structure in I-V. However, for gate voltages far away from V*
g (e.g., 0 or 20 V), only slight dips are seen in the dI/dV curves near zero bias, and the high bias conductance is about 5 ×10-5 S, corresponding to 20 kΩ resistance for the SWNT. We also measured the zero-bias conductance dI/dV under various gate voltages using the lock-in technique, as shown in Fig. 13.11(c). A gap-like region with highly suppressed conductance is observed between Vg 9 to 12 V in the dI/dV-Vg curve. Outside the gap, the sample exhibits high conductance with fluctuations. These results are consistent with the 2-D conductance data in Fig. 13.11(a).
Figure 13.11 (a) Grey-scale 2-D conductance plot log (dI/dV) at various (V, Vg). The brightest color corresponds to the lowest conductance 1 × 10-7 S. The darkest color corresponds to the highest conductance
4 × 10-5 S. (b) dI/dV vs. V curves recorded at various Vg. (c) Zero-bias dI/dV vs. Vg recorded at V*g 8 V.
132
133
The results presented above clearly show the small-gap semiconducting nature of the SWNT. Under V*g
where the Fermi level of the nanotube is in the middle of the band gap, a barrier exists to transport and the conductance of the sample is suppressed at both room temperature and low temperature. At 2 K, the suppressed region in conductance vs. bias recorded under V*
g (bottom curve in Fig. 13.11(b)) suggests that the SWNT band gap is on the order of 8 meV. Under Vg V*
g, the Fermi level is inside the nanotube valence band, which leads to significant conductance as transport through the valence band can occur (p-type). For Vg V*
g, the Fermi level is shifted into the conduction band, through which electron transport occurs (n-type).
The band gap can also be estimated from the conductance vs. gate voltage data shown in Fig. 13.11(c), where the gap region exhibiting highly suppressed conductance spans ΔVg' 3 V. The gate efficiency factor α for our general sample geometry can be estimated by using Coulomb blockade theory on our high resistance (hundreds of kiloOhms) metallic SWNT samples that exhibit periodic Coulomb oscillations in conductance vs. gate voltage measurements. The typical oscillation period is ΔVg 200 meV. From the previously found Coulomb charging energy U 1.4 eV/L (nm), energy level spacing ΔE 0.5 eV/L (Bockrath, et al., 1997), and L 3 µm, we obtain α (ΔE+U)/ΔVg 2.5 meV/V. This gives an estimated band gap Eg α ΔVg' 7.5 meV.
The low temperature data shown in Fig. 13.11 exhibit no clear signatures of Coulomb blockade. Thus, Coulomb charging effects are not dominating the observed small-gap semiconducting characteristics. However, we do observe significant conductance fluctuations upon gate voltage variations, especially under low bias voltages ( V <10 mV), as streaking lines are seen near the central region in Fig. 13.11(a). These fluctuations could be due to electron interaction effects, but their precise origins are not understood at the present time. At 2 K, the resistance of the SWNT sample away from the suppressed region in Fig. 13.11(a) is 20 kΩ, which is close to the resistance quantum h/2e2. The low resistance points to excellent metal-tube coupling, which is consistent with the fact that Coulomb charging is not the dominant phenomenon observed with the sample.
We also elucidated the temperature-dependent electrical properties of the small-gap semiconducting SWNT. The linear conductance vs. gate voltage curves measured at 290 K, 60 K and 10 K under a bias of V = 1 mV are shown in Fig. 13.12(a). We observed that V*
g (valley position in a dI/dV-Vg curve, corresponding to the gate voltage under which the Fermi level of the nanotube is in the middle of the band gap) drifted as temperature decreased. The drifts can be interpreted as due to changes in the electrostatic charge state of the substrate, as the temperature was lowered. To correct for this unwanted effect, we determined the temperature-dependent resistance of the SWNT sample under conditions with fixed Fermi-level position relative to the bands at all temperatures. Measured at V*
g(T), the valley resistance was found to increase monotonically as temperature decreased. The resistance was found to scale as exp(-Ea/KBT) with Ea 6 meV (Fig. 13.12(b)). This suggests that when the Fermi level resides inside the band gap, transport through the SWNT under small bias voltages is thermally activated across a barrier Eg/2. The small-gap semiconducting nature of the SWNT is thus fully manifested. On the other hand, when measured at gate voltages that are shifted from V*
g(T) by a constant, Vg = V*g(T)-5 V, the resistance of the SWNT was found to
exhibit drastically different dependence on temperature. The resistance decreased from 36 kΩ to 25 kΩ as temperature decreased from 290 K to 80 K. Upon further cooling, the I-V curve became nonlinear with suppressed conductance near zero bias. The resistance measured at V = 1 mV increased with decrease in temperature below 80 K, resulting in an upturn pattern in the resistance vs. temperature curve in Fig.
134
13.12(c). These results clearly show that, when the Fermi level resides within the valence (or conduction band, data not shown), the small-gap semiconducting SWNT behaves like a quasi-metal. In this quasi-metallic state, positive slope in dR(T)/dT can be attributed to reduced phonon (e.g., twiston) scattering (Kane, et al., 1998) as temperature decreases. A possible reason for the upturn at low temperature could be due to small barriers existing near the contacts due to band bending of the heavily doped p- or n-type SWNT.
Figure 13.12 (a) dI/dV vs. Vg curves recorded at 290, 60 and 10 K, respectively. Note the shifts in V*g(T). (b)
Linear resistance vs. 1/T measured under V*g(T). Solid line: fitting of R(T) exp(-Ea/KBT) with Ea 6 meV.
(c) Linear resistance vs. T measured under V*g(T)-5 V
The SGS-SWNT sample described above was highly stable and allowed for reproducible transport data. Nevertheless, upon repeated thermal cycles and loading/unloading the sample, shift in V*
g by several volts
135
was observed. This indicates slightly different environments felt by the nanotube during different cool-downs. Results shown in Fig. 13.12 were recorded during a different thermal cycle from that shown in Figs. 13.10 and 13.11. A shift in V*
g had occurred between these measurements.
We attribute the origin the observed small band gap to non-trivial curvature effects predicted to exist in small-diameter SWNTs (Hamada, et al., 1992; White, et al., 1996; Blase, et al., 1994; Kane and Mele, 1997). The SWNT is not an S-SWNT with large primary band gap, since the observed band gap is 10 meV and is much smaller than 600 meV expected for an S-SWNT with d 1.3 nm. Out of 20 systematically characterized individual SWNT samples, we observed three small-gap semiconducting SWNTs in electrical transport measurements.
The electrical properties of SGS-SWNTs can be clearly distinguished from other types of SWNTs described earlier. Our results provide direct transport evidence for the existence of SWNTs with small band gaps 10 meV. We have also elucidated the intrinsic electrical transport properties of SGS-SWNTs: either semiconducting or quasi-metallic behaviors. This "dual-personality" depends on the Fermi level position relative to the energy bands of the nanotube. Excellent coupling from metal electrodes to various types of SWNTs are reproducibly obtained within our fabrication approach, which leads to low contact resistance on the order of h/2e2 and facilitates the elucidation of intrinsic electrical properties of SWNTs.
13.4 Integrated Nanotube Devices
13.4.1 Nanotube Molecular Transistors with High Gains
Previous single-walled nanotube (SWNT) field-effect transistors (FET) obtained by randomly deposit SWNTs across electrodes exhibit low transconductance and voltage gain (Tans, et al., 1998b; Martel, et al., 1998). Our controlled chemical synthesis and integration approach have led to single-walled nanotube transistors that "mimic" silicon based metal-oxide field-effect transistors (MOSFET) with similar I-V characteristics and normalized transconductance. First, with our semiconducting nanotube samples, we reproducibly observed an interesting feature in the I-V curves when high bias voltages were applied. The I-V curves show marked asymmetry with respect to the polarity of the bias voltage when V > 1 V. I-V curves obtained at room temperature with the d = 2.8 nm tube sample (#S1) over a bias range of 3 to -3 V are presented in Fig. 13.13(a). Under a given gate voltage, the absolute current value is higher under a positive bias than under its negative bias. In the negative bias side, the current initially scales linearly as V but reaches saturation and stays constant at large negative biases. In the positive bias side, the current increases monotonically as the bias voltage increases and does not show any saturation. The asymmetry in I-V becomes increasingly dramatic under higher gate voltages. At Vg = 3 V, the I-V curve essentially resembles that of a rectifying diode.
Figure 13.13 (a) Room temperature I-V curves recorded with sample #S1 for V in the range of 3 to -3 V under various gate voltages. (b) I-V curves recorded after exchanging the source-drain electrodes. (c) Symmetrical I-V curves obtained by scanning V while biasing the two electrodes at -V/2 and V/2, respectively.
136
The asymmetry in I-V is found to be inherent to the metal/tube/metal system. I-V curves recorded after exchanging the source and drain electrodes show nearly unchanged asymmetry. Positive bias voltages still lead to higher absolute current values than negative biases (Fig. 13.13(b)). Furthermore, we find that symmetrical I-V curves can be obtained (Fig. 13.13(c)) by scanning V in the range of -3 to 3 V while biasing the two electrodes at -V/2 and +V/2, respectively. These results suggest that the observed asymmetrical I-V curves are not caused by asymmetrical parameters such as different contact resistances at the two metal-tube
137
interfaces. We conclude that for a significantly high bias voltage V, the current flow I is not solely determined by the absolute value of voltage V across the system and can be influenced by the bias configuration. The [0, V ] bias configuration leads to higher current flow than the [0, - V ] configuration, such asymmetry is introduced into the system by the large applied bias voltage. Note that asymmetry is nearly absent in the I-V curves in the small bias regime ( V <0.1) as can be seen in Fig. 13.8(a).
The bias polarity associated asymmetrical I-V has been consistently observed in all of the p-type semiconducting SWNT samples that we have studied. We propose that the asymmetry in I-V is caused by local gating effects of the biased drain electrode. Under a given gate voltage, the nanotube can be considered to have a constant hole-density along its length. Under a negative bias configuration [0, - V ], the electric field lines can be substantially enhanced at the nanotube section near the drain electrode for V> 1 V. The negative drain bias effectively introduces an increase in the gate voltage localized near the tube section close to the drain, resulting in a reduction of hole density in the section. Equivalently, a higher barrier to transport is produced at the junction at the drain. The I-V curve thus falls below the line extrapolated from the low bias curve. Saturation occurs in the I-V curve for large V because of the competing roles of driving and gating of the drain bias voltage. On the other hand, under the positive bias configuration [0, V ], the electric field lines at the section of the nanotube close to the drain becomes suppressed or even reversed at large V . This leads to local hole enrichment (or equivalently barrier lowering) in the nanotube section near the drain, and the I-V curve deviates to the higher current direction from the linear curve extrapolated from the small bias regime. Overall, a negative or positive bias voltage with absolute value on the order of 1 V locally depletes or increases the hole-carrier density in the nanotube segment near the drain, leading to a lower absolute current level when the system is negatively biased. This phenomenon can be related to local carrier depletion and channel pinch-off by negative drain bias in a conventional p-type MOSFET (Sze, 1981).
Our results are significant in the realization of high performance nanotube based transistors. First, the Ids-Vds curves of our samples exhibit similar characteristics as silicon based devices (Sze, 1981). Secondly, high voltage gain and transconductance are obtainable with our samples. Sample # S1 (Fig. 13.8(a)) exhibits positive voltage gain of ΔVds/ΔVg I=3µA 3 compared to the maximum gain of 0.35 obtained previously (Tans, et al., 1998b). From the linear region of the Ids-Vds curves, we derive a transconductance Ids/ Vg V=100mV 200 nA/V, which is two orders of magnitude higher than previous results with SWNTs (Martel, et al., 1998). The high transconductance is a direct result of low resistance ( 370 kΩ) of the semiconducting SWNT system. Normalized by the diameter of the nanotube ("channel width"), the transconductance is 0.1 mS/µm and is comparable to that of a silicon p-MOSFET.
13.5 Conclusions
This chapter presented synthetic strategies that lead to various nanotube architectures useful for fundamental studies and potential applications of nanotube wires. Undoubtedly, the approach of direct growth of nanowire materials into ordered structures represents a promising direction towards new nanoscale science and devices. This method achieves atomically well-defined nanowires by bottom-up chemical routes, and wired-up architectures that normally require top-down fabrication or assembly techniques. The general
138
approach of chemical vapor deposition has received considerable attention recently because of the unique opportunities in scaling up nanotubes in bulk and organizing nanowires on large-scale surfaces. Ordered multi-walled nanotube architectures can now be synthesized by several CVD methods on various substrates. The structured nanotube materials can be scaled up in straightforward ways and shall find wide applications in electron field emission, flat-panel display, large-scale scanning force probes, electrochemical sensors and electrodes. Single-walled carbon nanotubes are unique because their diameters enter such a regime that their electrical properties is strongly manifested by quantum confinement effects in the other two dimensions and their structures at the atomic scale. It can be envisioned that in a foreseeable future, SWNT architectures is integrated as key components into microchips of next generations of electronic, memory, logic, mechanical, and electromechanical devices.
Significant future work are required in order to meet the continued challenge in controlled nanotube synthesis. It is desired to gain the full ability to grow SWNTs at any desired locations with controlled orientations, either on flat or textured surfaces, or at the ends of scanning probes or at any other locations. It is an ultimate synthetic goal if one can control whether a semiconducting or metallic SWNT is grown at a particular location with directionality. Controlling the atomic structure of nanotubes including diameter and chirality probably will require the development of catalytic particles with molecularly well-defined structures. Thus, one of the main focuses in future research should be developing new catalyst materials and catalyst support materials. Also, deeper understandings of the chemistry of catalyst and nanotube growth are required. We are convinced that further control in nanotube synthesis will continue to open new possibilities in nanoscale science and technology.
References
Amelinckx S., X. B. Zhang, D. Bernaerts, Zhang, X. F., Ivanov, V. and J. B. Nagy. Science. 265, 635–639 (1994)
Baker R. T. K.. Carbon 27, 315–323 (1989)
Baker R. T. K. and N. M. Rodriguez. Symposium of the Materials Research Society. 251–256 (1994)
Bethune D. S., C. H. Kiang, M. DeVries, G. Gorman, R. Savoy, J. Vazquez and R. Beyers. Nature. 363, 605–7 (1993)
Blase X., L. X. Benedict, E. L. Shirley and S. G. Louie. Phys. Rev. Lett.. 72, 1878–1881 (1994)
Bockrath M., D. Cobden, J. Lu, Rinzler, A., Smalley, R., Balents, L. and P. McEuen. Nature. 397, 598–601 (1999)
Bockrath M., D. H. Cobden, P. L. McEuen, Chopra, N. G., Zettl, A., Thess, A. and Smalley, R. E.. Science. 275, 1922–1925 (1997)
Bonard J. M., J. P. Salvetat, T. Stöckli, de Heer, W. A., Forró, L. and Ch telain, A. Appl. Phys. Lett.. 73, 918–920 (1998)
Cassell A., N. Franklin, T. Tombler, Chan, E. Han, J., and Dai, H.. J. Am. Chem. Soc.. 121, 7975–7976 (1999a)
139
Cassell A., J. Raymakers, J. Kong and Dai, H.. J. Phys. Chem.. 103, 6484–6492 (1999b)
Che G., B. Lakshmi, E. Fisher and Martin, C.. Nature. 393, 346–349 (1998)
Chico L., L. X. Benedict, S. G. Louie and Cohen, M. L. Phys. Rev.. B 54, 2600–2606 (1996)
Cobden, Bockrath, Chopra, Zettle, McEuen, Rinzler, and Smalley. Phys. Rev. Lett.. 81, 681–684 (1998)
Collins P. G. and Z. Zettl. Appl. Phys. Lett.. 69 (13), 1969–1971 (1996)
Dai H., N. Franklin and J. Han. Appl. Phys. Lett.. 73, 1508–1510 (1998)
Dai H., J. Hafner, A. Rinzler Colbert, D. and Smalley, R. Nature. 384, 147–150 (1996)
de Heer W., A. Chaelain and D. Ugart. Science. 270, 1179–1180 (1995)
Dekker C. Phys. Today. 52, 22–28 (1999)
Dresselhaus M. S., G. Dresselhaus and P. C. Eklund. Science of Fullerenes and Carbon Nanotubes. Academic Press, San Diego (1996)
Ebbesen T. W. Phys. Today. 49, 26–32 (1996)
Endo M. Chemtech. 568–576 (1988)
Fan S., M. Chapline, N. Franklin, T. Tombler, A. Cassell and H. Dai. Science. 283, 512–514 (1999)
Ferguson G., M. Chaudhury, G. Sigal and G. M. Whitesides. Science. 253, 776–778 (1991)
Hafner J., C. Cheung and C. Lieber. Nature. 398, 761–762 (1999)
Hamada N., S. i. Sawada and A. Oshiyama. Phys. Rev. Lett.. 68 (10), 1579–1581 (1992)
Iijima S. Nature. 354, 56–58 (1991)
Iijima S. and T. Ichihashi. Nature. 363, 603–605 (1993)
Journet C., W. K. Maser, P. Bernier, A. Loiseau, M. L. Delachapelle, S. Lefrant, P. Deniard, R. Lee and J. E. Fischer. Nature. 388, 756–758 (1997)
Kane C., L. Balents and M. P. A. Fisher. Phys. Rev. Lett.. 79, 5086–5089 (1997)
Kane C. L. and E. J. Mele. Phys. Rev. Lett. 78, 1932–1935 (1997)
Kane C. L., E. J. Mele, R. Lee, Fischer, J. E., Petit, P., Dai. H., Thess A., Smalley, R. E., Verschueren, A. R. M., Tans, S. J. and Dekker, C. Euro. Phys. Lett.. 6, 683–688 (1998)
Kong J., A. M. Cassell and H. Dai. Chem. Phys. Lett.. 292, 567–574 (1998a)
Kong J., H. Soh, A. Cassell, Quate, C. F. and Dai, H. Nature. 395, 878 (1998b)
Kong J., C. Zhou, A. Morpurgo, Soh, T., Marcus, C., Quate, C. and Dai. H. Appl. Phys.. A 69, 305–308 (1999)
140
Kyotani T., L. F. Tsai and A. Tomita. Chem. Mater.. 8, 2109–2113 (1996)
Lepselter M. P. and S. M. Sze. Proc. IEEE.. 56, 1400–1402 (1968)
Li W. Z., S. S. Xie, L. X. Qian, Chang, B. H., Zou, B. S., Zhou, W. Y., Zhao, R. A. and Wang, G. Science. 274, 1701–1703 (1996)
Martel R., T. Schmidt, H. R. Shea, Hertel, T. and Avouris, P. Appl. Phys. Lett.. 73, 2447–2449 (1998)
Mintmire J. W., B. I. Dunlap and C. T. White. Phys. Rev. Lett.. 68, 631–634 (1992)
Morpurgo A., J. Kong, C. Marcus and Dai, H. Science. 286, 263–265 (1999)
Nygard J., D. H. Cobden, M. Bockrath, McEuen, P. L. and Lindelof, P. E. Appl Phys. A 69, 297–304 (1999)
Odom T., J. Huang, P. Kim and Lieber, C. M. Nature. 391, 62–64 (1998)
Pan Z., S. S. Xie, B. Chang and Wang, C. Nature. 394, 631–632 (1998)
Ren Z. F., Z. P. Huang, J. W. Xu and Wang, J. H. Science. 282, 1105–1107 (1998)
Rochefort A., F. Lesage, D. Salhub and Avouris, P. xxx.lanl.gov/cond-mat/9904083 (1999)
Saito R., M. Fujita, G. Dresselhaus and Dresselhaus, M. S. Appl. Phys. Lett.. 60, 2204 (1992)
Saito Y., K. Hamaguchi, K. Hata and Tohji, K. Ultramicroscopy. 73, N1–4: 1–6 (1998)
Smith R. L. and S. D. Collins. J. Appl. Phys.. 71, R1–R22 (1992)
Snyder C. E., W. H. Mandeville, H. G. Tennent and Truesdale, L. K.. Int. Patent WO89/07163 (1989)
Soh H., C. Quate, A. Morpurgo, Marcus, C., Kong, J. and Dai, H. Appl. Phys. Lett.. 75, 627–629 (1999)
Sze S. M. Physics of semiconductor devices. Wiley, New York, (1981)
Tans S., M. Devoret, R. Groeneveld and Dekker, C. Nature. 394, 761–764 (1998a)
Tans S., A. Verschueren and C. Dekker. Nature. 393, 49–52 (1998b)
Tans S. J., M. H. Devoret, H. Dai, Thess, A., Smalley, R. E., Geerligs, L. J. and Dekker, C. Nature. 386, 474–477 (1997)
Tersoff J. Appl. Phys. Lett.. 74, 2122–2124 (1999)
Thess A., R. Lee, P. Nikolaev, Dai, H. J., Petit, P., Robert, J., Xu, C. H., Lee, Y. H., Kim, S. G., Rinzler, A. G., Colbert, D. T., Scuseria, G. E., Tomanek, D., Fischer, J. E. and Smalley, R. E. Science. 273, 483–487 (1996)
Tibbetts G. G. Appl. Phys. Lett.. 42, 666–668 (1983)
Tibbetts G. G.. J Cryst. Growth 66, 632–638 (1984)
Tibbetts G. G.. Vapor Grown Carbon Fibers. In: Carbon Fibers, Filaments and Composites. Kluwer Academic, Amsterdam (1990), 73–94
141
Tibbetts G. G., M. G. Devour and E. J. Rodda. Carbon. 25, 367–375 (1987)
Vial J.-C. and J. Derrien. Porous silicon science and technology: winter school. Les Houches, 8 to 12 February, 1994. Centre de physique des Houches. Springer-Verlag, Berlin (1994).
Wang Q., A. Setlur, J. Lauerhaas Dai, J. and Chang, R. H. Appl. Phys. Lett.. 72, 2912–2913 (1998)
White C. T., D. H. Robertson and J. W. Mintmire, In: Clusters and Nanostructured Materials. eds. by P. Jena and S. Behera, Nova, New York (1996), 231
White C. T. and T. N. Todorov. Nature. 393, 240 (1998)
Wildoer J. W. G., L. C. Venema, A. G. Rinzler, Smalley, R. E. and Dekker, C. Nature. 391, 59–62 (1997)
Wong S., J. Harper, P. Lansbury and Lieber, C. M. J. Am. Chem. Soc.. 120, 603–604 (1998a)
Wong S., E. Joselevich, A. Woolley Cheung, C. and Lieber, C. Nature. 394, 52–55 (1998b)
Xia Y. and G. M. Whitesides. Angew. Chemie.. (Int. Ed.) 37, 551–575 (1998)
Yakobson B. I. and R. E. Smalley. Am. Sci.. 85, 324–337 (1997)
Yang P., T. Deng, D. Zhao, Feng, P., Pine, D., Chmelka, B. F., Whitesides, G. M., and Stucky G. D. Science. 282, 2244–2246 (1998a)
Yang P., D. Zhao, D. Margolese Chemelka, B. and Stucky, G. Nature. 396, 152 (1998b)
142
14. Nanomaterials from Light-Element Composites
14.1 Introduction
For over 10 years a large amount of work worldwide has been directed towards obtaining an understanding of the new covalently bonded nanomaterials made from light atomic weight elements from the first row of the periodic table because of the novel microstructures and the extraordinary combination of physical properties (Wang, 1997, 1999; Veprek, 1999). For example, Cohen proposed that carbon nitride should have diamond-like properties with a relatively isotropic arrangement of short ( 0.147 nm in length) and covalent ( 7% ionic) bonds (Cohen, 1985, 1989a, 1989b). Such materials with high hardness and toughness, oxidation resistance, chemical stability, high adherence, and high thermal conductivity, are important in high-performance engineering applications for high-temperature, high-power, or high-frequency devices ranging from microelectronic to spaceflight industries.
In addition to the potential applications, the goal of this effort is also to see if one can design a high-performance material by beginning with theories to select candidates for laboratory synthesis. As one of the computer designed structures, this study provides a test of the effectiveness of first-principles calculations in materials science. It is the purpose of the present chapter to survey the recent work which has been carried out and to detail the level of understanding which has been attained in the research of light-element nanomaterials.
14.2 Theoretical Prediction
A number of workers have published theoretical calculations for light-element compounds, for example of carbon nitride polymorphs. An early empirical model (Cohen, 1985) is directed towards understanding their covalent and ionic nature through a study of scaling arguments based on the Phillips-Van Vechten scheme (Phillips, 1973). Other theoretical efforts have focused on ab initio band structural calculations and ab initio molecular dynamics (MD) simulations. Investigators have also studied the phase transition under pressure and vibrational properties (Widany, et al., 1996).
14.2.1 Empirical Model
An empirical approach was first developed by Cohen (Cohen, 1985) to study the bulk modulus on the basis of the Phillips-Van Vechten scheme (Phillips, 1973) for characterizing the covalent and ionic nature of tetrahedral solids by means of their spectral properties. For purely covalent solids such as carbon and silicon, the bulk modulus scales as the average homopolar energy gap in the reflectivity spectrum divided by the volume of the bond charge. This leads to the bulk modulus varying as d-3.5. For the zinc blendes, an additional empirical term is added to account for the depletion of bond charge with increasing ionicity. The formula is modified further by adding a factor Nc/4, where Nc is the average coordination number. The
143
resulting relationship between the compressivity modulus B (Mbar)(1 bar = 1 × 105 Pa), ionicity λ and length d (0.1 nm) of the interatomic bond can be written as
The ionicity parameter, λ = 0, 1 and 2 for Groups IV, III-V and II-VI semiconductors, respectively, accounts for the reduction in B arising from increased charge transfer. This is in agreement with experimental results showing an increase of ionicity and a loss of covalency in going from Group IV, III-V and II-VI semiconductors. As a semi-empirical method depends only an input parameter d, this expression gives results which are as accurate as first-principles total-energy calculations (Cohen, 1988).
The B values obtained for a number of zinc-blende compounds from Eq.(14.1) are in excellent agreement with experiment and rival the costly first-principles calculations for accuracy. Therefore, it is possible to use the theory to design materials with predictable properties. From Eq.(14.1) we can easily find that the condition for large B is to minimize d and λ. Because of the small atomic radii of carbon (0.077 nm) and nitrogen (0.070 nm) and λ<1, the carbon nitride compounds should have high bulk moduli.
A β-C3N4 structure was first suggested by Cohen based upon knowledge of the known β-Si3N4 (Cohen, 1985). The average coordination number, Nc, is 3.43. As this structure is a Group IV-V material, Liu and Cohen chose to use an ionic parameter, λ, in the range 0 to 0.5 (Liu and Cohen, 1990). By using estimates of d (0.147 nm) and λ, Eq.(14.1) yields B is 410–440 GPa which brackets the calculated value for diamond.
14.2.2 First-Principles Study
Further first-principles studies of carbon nitride compounds are motivated by an early discussion, which indicates that short bond lengths and low ionicity are favorable for achieving large bulk moduli. The most obvious candidate structure to be used as a prototype for a first-principles investigation of the properties of a covalent solid formed between carbon and nitrogen is the simple zinc blende structure. However, this was ruled out by Liu and Cohen because in a hypothetical zinc blende C-N compound with nine valence electrons per unit cell the first antibonding band would be occupied (Liu and Cohen, 1989). This would distribute charge in the antibonding band and hence it is doubtful that such a structure would be stable with respect to other cases.
By evaluating the known β-Si3N4 structure, Liu and Cohen suggested a hypothetical carbon-nitrogen compound wherein carbon is substituted for silicon (Liu and Cohen, 1989). An ab initio variable-cell-shape molecular-dynamics algorithm was used to investigate the stability of three carbon nitrides with composition C3N4 (Liu and Wentzcovitch, 1994). Systematic first-principles calculations of the relative stability, structure, and physical properties of carbon nitride polymorphs, including α-, β-, cubic-, pseudocubic-, and graphitic-C3N4, was given by Teter and Hemley within the local density approximation (LDA) (Teter and Hemley, 1996). The band gap of β-C3N4 was 3.56 eV given by Yao and Ching using first-principles local-density calculations (Yao and Ching, 1994). A quasiparticle electronic band calculation of β-C3N4 was given by Corkill and Cohen (Corkill and Cohen, 1993). From the GW correction, the lowest
144
indirect gap is predicted to be 6.4 eV while the direct gap at is 6.75 eV. These gaps are even larger than the minimum gap in diamond.
Badding and Nesting presented a thermodynamic analysis of the formation of a more carbon-rich C4N3 under pressure (Badding and Nesting, 1996). Their results show that C3N4 and C4N3 form in a similar pressure range. A study on the amorphous state of carbon nitrides for a wide range of stoichiometries and densities was performed by using a molecular-dynamics simulation (Weich, et al., 1997). By the same method, this group also calculated the vibrational property of C-N structures (Widany, et al., 1996).
Most recently, the ternary compounds, such as boron carbonitride (BCN) and silicon carbonitride (SiCN), have also attracted considerable attention. According to theoretical calculations, besides excellent mechanical properties (Hernandez, et al., 1998), it is relatively easier to control the electronic properties of BCN nanotubes since these properties are mainly determined by the atomic composition and atomic configuration (Blasé, et al., 1997; 1995; Hamada, et al., 1992). It is also expected that the band gap of BCN nanotubes decreases for the series of intermediate phases from insulating hexagonal boron nitride (h-BN) to semimetallic graphite (Miyamoto, et al., 1994; Watanabe, et al., 1996). These results imply potential applications of BCN nanotubes or nanofibers in nanosized electronic and photonic devices.
14.3 Synthesis by Chemical Vapor Deposition (CVD)
Ever since the theoretical prediction by Cohen in 1985 (Cohen, 1985), a tremendous amount of experimental effort has been applied to synthesize light-element nanomaterials by a large variety of the more readily available techniques, such as plasma, sputtering, laser ablation, chemical vapor deposition, ion beam deposition, and high-pressure pyrolysis.
In a capacitively coupled rf plasma reactor, amorphous carbon nitride thin films were grown by Han and Feldman using a feedstock of CH4 and N2 (Han and Feldman, 1998). A pulsed high-energy plasma technique was used by Peng et al. (Peng, et al., 1996) and Bursill et al. (1995) to deposit CNx films. Some small crystalline particles embedded in amorphous C-N films were observed by Yu et al. using sputtering techniques (Yu, et al., 1994). Electron diffraction indicates that β-C3N4 is a viable structure for these crystallites. Okada et al. obtained carbon nitride films with N/C value up to 1.20 by relatively high-pressure rf magnetron sputtering (Okada, et al., 1995). Li et al. reported polycrystalline C3N4 films deposited on Si(111) with N/C ratio of 1.33 (Li, et al., 1995). Using dc magnetron sputtering, Sjöström, et al., synthesized CNx thin films with a fullerene-like microstructure (Sjöström, et al., 1994). Niu, Lu, and Lieber deposited covalent solid carbon nitrogen films on Si (100) using low-energy (about 1 eV) atomic beam-assisted laser ablation (Niu, et al., 1993). Recently this group claimed that they had obtained thin-film C-N materials with <50 at % nitrogen (Zhang, et al., 1995).
Fujimoto and Ogata (Fujimoto and Ogata, 1993), and Ogata, Chubaci, and Fujimoto (Ogata, et al., 1994) reported on the formation of nitrogen rich C-N materials by 0.5–10.0 keV N ion beam. Most recently, possible evidence for the stabilization of β carbon nitride was reported by several groups using CVD,
145
high-energy ball milling, sputtering, and other techniques (Chen, et al., 1997a, 1996a; Fahmy, et al., 1999; Jagielski, et al., 1999; Chowdlhury, et al., 1999).
Well-aligned carbon nitride nanotubes were synthesized in anodic alumina (Sung, et al., 1999) and carbon nitride nanofibers in high yield were deposited on silica substrate by pyrolyzing melamine over laser-patterned thin films of Fe and Ni catalyst (Terrones, et al., 1999).
Nanocrystalline diamond, BN, BCN and SiCN with various microstructures have been also reported experimentally. Since the first study on BCN by Badzian et al. (1972), the cubic BCN phase (Weng-Sieh, et al., 1995), turbostratic BCN structure (Yu and Wang, 1999; Yu, et al., 1999), BCN nanotubes (Way, et al., 1992), and BCN microrods (Yu, et al., 1998) have been synthesized. A systematic investigation on SiCN has been performed by varying the growth parameters over a wide range (Gong, et al., 1999). Among the techniques used so far, chemical vapor deposition (CVD) is the most successful for growing light-element composites.
14.3.1 Bias-Assisted Hot Filament CVD
A bias-assisted hot filament CVD (bias-HFCVD) was designed and used to synthesize light-element composites (Chen, et al., 1996b, 1998a; Wang, et al., 1997). The vacuum chamber is a water-cooled stainless steel tube with a size 45 cm in diameter and 60 cm in length (see Fig. 14.1). High pure gases are mixed as the reactive source, which is introduced from the chamber top, and a carbonized tungsten filament (= = 0.3 mm) is heated to dissociate the mixture. Distinct from a conventional apparatus, a tantalum (Ta) mesh 30 mm × 40 mm in area is installed above the filament at a distance of 15–20 mm, and a Ta wafer of dimensions 10 mm × 20 mm × 0.2 mm is used as the substrate holder. A dc bias (both negative and positive) with a power supply (500 V × 1 A) can be used to generate discharge plasma between the filament and the substrate, which is useful to increase the activation of nitrogenous precursors. The negative bias plays an important role, as we have found that carbon-rich films with no perfect crystal grains have been obtained in the experiments without the application of the bias. The general deposition procedure is carried out as below: after the chamber pressure of 1×10-3 Torr (1 Torr = 1.333×102 Pa) is pumped, the gas mixture is introduced into it with desired concentration. Then the filament is heated and the negative substrate bias is applied. The depositions are performed from several minutes to more than 10 hours for different purposes.
Figure 14.1 Schematic of the bias-assisted HFCVD.
146
The main experimental parameters are set: filament temperature, 1800°C–2100°C (measured by an optical pyrometer); substrate temperature 750°C–950°C (measured by a W25Re thermocouple fixed on the back of the substrate); negative bias voltage (Vmesh - Vsub<0), 50–400 V; DC glow discharge current, 50–350 mA; filament-substrate distance, 5–8 mm. The flow ratios of the reactive gases are controlled by mass-flowmeters respectively. By using this system, we have successfully obtained polycrystalline diamond (Chen, et al., 1998), nanocrystalline carbon nitride (Chen, et al., 1996a, 1997a), turbostratic boron carbonitride (Yu and Wang, 1999; Yu, et al., 1999) and well-aligned boron carbonitride nanofibers.
14.3.2 Electron Cyclotron Resonance Microwave Plasma-Assisted CVD (MPCVD)
Electron cyclotron resonance microwave plasma-assisted CVD is already commercialized. The system, shown in Fig. 14.2, consists of an independent vacuum processing chamber equipped with a electron cyclotron resonance (ECR) plasma source. A 2.45 GHz, 1500 W microwave power supply produces microwaves that are guided through a quart window to the ECR plasma processing chamber. Two magnetic coils, which are cooled by a water-circle system, driven in a mirror configuration establish an 875 G magnetic field. The vacuum chamber is usual evacuated to about 10-7 Torr (1 Torr = 1.333 × 102 Pa) prior to deposition using a turbomolecular pump. The ion species generated by the combination of magnetic field and microwave excitation are accelerated from the plasma to the substrate by a negative dc bias, which voltage is in the range of 0–90 V. A radio-frequency (rf) heater with power controlled dynamically by a temperature controller keeps the substrate temperature quite stable during processing. In the experiments,
147
the substrate temperature can be varied from room temperature to over 1500°C. By using this system, we have successfully obtained nitrogen-incorporated nanocrystalline diamond (Wu, et al., 1999), polymerized carbon nitride nanobells (Ma, et al., 1999), and carbon nitride/diamond/silicon multilayers.
Figure 14.2 Schematic of the ECR-MPCVD.
14.4 Uniform Size-Controlled Nanocrystalline Diamond Films
Diamond and related materials have been extensively studied for their novel mechanical, chemical and electrical properties. Particularly, because of the low or negative electron affinity (NEA) of diamond (Himpsel, et al., 1979), special interest is focused on its field electron emission (FEE), which is believed to have potential applications from flat panel display to power transmitters. Cold cathode field emission has been demonstrated in CVD polycrystalline diamond films, of which even a prototype flat panel display was made (Kumar and Schmidt, 1995). Since intrinsic diamond is a wide band gap (5.45 eV) insulator, the NEA is almost useless unless there exists an effective mechanism supplying electrons to the conduction band. It has been reported that defect-related energy states or sub-bands may play important role in supplying electrons and forming conduction channel for field emission (Wisitsora-at, et al., 1997).
In undoped CVD diamond films, the bulk conductivity is dominated by the space charge limited current (SCLC) (May, et al., 1998). As a result, grain boundaries, which are highly disordered or contain
148
co-deposited graphite impurities, have been suggested as the main conduction channel through CVD diamond films. Thus a direct way to improve the FEE properties of diamond is to reduce the film grain size. Because of this and other unique features of nanocrystalline diamond films, such as low friction and transparency, they have attracted more and more attention. Zhu, et al., observed an emission threshold as low as 1 V/µm on nanocrystalline diamond films prepared from industrial diamond nanopowders. Zhou et al. (2000) fabricated nanocrystalline diamond films by plasma enhanced CVD (PECVD) using C60/Ar/H2, or O2/Ar addition to the H2/CH4 precursors, and give an emission threshold of about 3 V/µm. Very recently, uniform size-controlled nanodiamond films were obtained by Wu, et al., using hydrogen addition to N2/CH4 precursor in microwave plasma-assisted CVD (Wu, et al., 2000). The film microstructure, including the diamond grain size and the content of the graphite impurity, can be well controlled by the deposition parameters.
14.4.1 Deposition with CN4/N2 Precursor
Film deposition is carried out using an ASTeX 2115 microwave plasma-assisted CVD system with a maximum microwave output power is 1500 W (Wu, et al., 1999). High pure nitrogen (99.999%) and methane (99.9%) gases are employed as precursors. Polycrystalline molybdenum and (100)-oriented Si wafers are used as the substrates, which are first polished by 1.0 µm diamond paste and then ultrasonically cleaned in acetone, ethanol and deionized water consequently.
All as-deposited films in the experiments present very smooth surface. The first series of films are grown by using different CH4 flow rates from 2.1 to 8.4 sccm, while the N2 is kept at 150 sccm and no H2 gas is used. The as-grown films have a deep black color under visible light. Figure 14.3(a) and (b) shows the transmission electron microscopy (TEM) images of the two films, where the CH4 flow rates are 3.5 sccm for (a) and 2.1 sccm for (b), respectively. Figure 14.3(a) reveals the nanocrystalline nature of the film, which is composed of the diamond crystallites at the size of 8 nm, embedded in amorphous carbon textures. The content of amorphous carbon in this film is relatively high, reaching up to 40% as roughly estimated from the image. The inserted selected-area electron microscopy (SAED) image shows a typical ring pattern of the nanocrystalline diamond. The film in Fig. 14.3(b) shows a larger grain size at about 20–30 nm, and the content of amorphous carbon is much lower, below 5% as estimated from the TEM image.
Figure 14.3 TEM images of the nanocrystalline diamond films grown with N2/CH4, where (a) and (b) are grown with 3.5 and 2.1 sccm, respectively. Inserted is the SAED pattern.
149
XRD spectra show strong diamond (111) and (220) feature peaks, where the full-width-half-maximums (FWHM) of these peaks are significantly broadened due to the confinement effect. The broadening of the diamond peaks provides a way to determine the average diamond grain size in the films. According to the Scherrer equation D = λ/(Bcosθ), where D is the average diamond grain size, λ the wavelength of X-ray, B
150
the integral width of the peak, and θ the Bragg angle. Here we take the strongest diamond (111) peak for analysis, and Gaussian fitting is used. For a series of films grown with different CH4 concentration, the calculated diamond grain sizes are shown in Fig. 14.4. It is found that within a certain range of CH4 concentration (certain sccm CH4), the diamond grain size does not show much deviation. But as the CH4 flow rate is reduced to 3 sccm or below, the diamond grain size becomes larger. This is inconsistent with the TEM results, although the quantitative values of the diamond grain sizes obtained by these two methods are not exactly the same due to the residual stress and defects in the films. On the other hand, since X-ray diffraction (XRD) is insensitive to non-crystalline phase, the change of the amorphous carbon content cannot be shown in the XRD spectra.
Figure 14.4 Influence of CH4 concentration on the diamond grain size of the films grown with N2/CH4.
Raman spectra of the nanocrystalline diamond films grown with different CH4 concentrations are shown in Fig. 14.5. The spectra consist of two wide peaks around 1560 cm-1 and 1350 cm-1, which correspond to the G band and D band of graphite, respectively. This confirms the existence of graphite phase in our films as observed by TEM. A weak diamond peak at 1332 cm-1 is presented, probably because the graphite contents in the films are relatively high. There is an additional peak at about 1150 cm-1, which is caused by the size effect of the nanoscale diamond grain (Walter and Messier, 1990; Talin, et al., 1996). It provides additional evidence for the nanocrystalline nature of the films. In addition, two findings can be derived by the Raman spectra. First, with the decreasing of CH4 concentration in the precursors, the diamond 1332 cm-1 peak appears clear, showing that the graphite content is reduced in these films. Secondly, the 1150 cm-1 peak becomes weak with the decreasing of CH4 concentration, indicating that the diamond grain size gets larger. These are all consistent with the TEM and XRD results. It should be noted that even in the film with lowest graphite content, for example of the film grown with 2.1 sccm CH4 flow rate, the diamond signal at 1332 cm-1 is still some what weak. This is because that the graphite phase has much higher resonance cross section than diamond, and the graphite cannot be completely removed without addition of hydrogen to the precursors.
Figure 14.5 Raman spectra of the nanocrystalline diamond films grown with different CH4 concentrations.
151
To study the influence of CH4 concentration on the film growth rate, the film thickness is measured by a profilemeter, and the film growth rate is calculated by dividing the film thickness by the growth time. It is found that the growth rate decreases nearly linearly with the CH4 concentration. When the CH4 flow rate is reduced to about 1.5 sccm, there is completely no film growth, only etching effect is observed. This is in contrast with the deposition of polycrystalline diamond films with H2/CH4 precursors, where the diamond films can be grown with very low CH4 concentration, such as 0.5%.
14.4.2 Influence of Additional H2 on Microstructure
In order to further control the microstructure of the nanocrystalline diamond films, we then introduce a small amount of hydrogen (0–10 sccm), a new deposition parameter, to the precursors. Figures 14.6(a) and (b) show the TEM images of the films grown with H2 flow rate at 5.0 and 10.0 sccm, respectively, where the CH4 is fixed at 3.5 sccm. By comparing the two images with that in Fig. 14.3(a), where the same growth condition is used except for the addition of H2 gas, apparent morphology evolution can be found. First, with the increasing of H2 flow rate, the diamond grain size increases from 8 to 20 nm (5 sccm H2), and then to 50 nm (10 sccm H2). Secondly, the graphite phase is greatly reduced, nearly pure nanocrystalline diamond films are shown in the images. In fact, it can be observed that with the increasing of H2 gas concentration the films, colors change significantly from deep black to gray and then to transparent under visible light. It means that pure nanocrystalline diamond films with controllable grain size can be obtained by varying the H2 concentration during growth.
152
Figure 14.6 TEM images of the nanocrystalline diamond films grown with H2 flow rate at (a) 5.0 and (b) 10.0 sccm, respectively, where the CH4 flux ratio is fixed at 2.3%.
153
XRD and Raman spectra also confirm these results. There is a very clear evolution of the peak, where the FWHM decreases very quickly with the increasing of H2 flow rate. When the H2 gas flow rate increases from 0 to 10 sccm, the average diamond grain size increases from 8 to 50 nm. This value is well consistent with that estimated by TEM images. The evolution of Raman spectra with the increasing of H2 flow rate is obvious. First, the relative strength of the 1332 cm-1 diamond peak to the graphite bands (G band at 1560 cm-1 and D band at 1350 cm-1) increases significantly with the increasing of H2 flow rate. This means the graphite content is greatly reduced. Secondly, the peak at 1150 cm-1, which is indicative of the nanosize diamond crystallites, disappears in the films grown with higher H2 of 7.5 and 10.0 sccm. It means that the diamond grain size has exceeded the value needed to generate this peak.
Based on the above analysis, we can conclude that: (1) The microstructure of the nanocrystalline diamond films can be varied by either the CH4 concentration or the H2 concentration; (2) By introducing H2 gas, an independent parameter, to the CH4/N2 precursors, the diamond grain size and graphite content can be easily controlled. The results show that the diamond grain size can be adjusted in a wide range (0–50 nm), and the graphite impurity can be almost completely removed from the films. Furthermore, we find that the film growth rate of this series of films decreases with increasing of H2 flow rate. When the H2 flow rate increases to 15 sccm, no film growth but only etching effect is found.
14.4.3 Nitrogen Incorporation
Secondary ion mass spectroscopy (SIMS) has been carried out on a CAMECA ims 4f., where 10 keV positive Cs ion is used. The results show that the two series of the films with or without hydrogen are highly doped with nitrogen irrespective of the CH4 concentration or the H2 flow rate. The dopant concentrations are around 1021 cm-3 for all the films measured.
14.4.4 Surface Stable Growth Model
In order to explain why the nanocrystalline diamond films can be easily obtained in the N2/CH4 environment, whereas by using traditional H2/hydrocarbon precursors polycrystalline diamond films are usually obtained, we propose a surface stable growth model here. In fact, the atomic H plays two key roles in the deposition of CVD diamond films. It is known that H can selectively etch off non-diamond phase to get pure diamond, and at the same time the terminated H on the diamond surface also plays a role to stabilize the diamond growing surface (Walter, et al., 1990). Thus with the help of atomic H, the formation of new diamond can easily take place on the existing diamond surface, making it easy to grow large-size diamond particles. Controversely, in the N2/CH4 environment, we believe that the atomic N does not have the stabilize the diamond growing surface. Hence, the diamond growth would takes place rather from a new diamond seed, and nanocrystalline diamond films can be easily obtained. If we introduce H2 to the gas mixture, the surface stabilizing mechanism is simultaneously introduced. The amount of the additional hydrogen gas determines how stable the growing surface is, and thus determines the diamond grain size. At the same time, the selective etching of non-diamond phase of the atomic H results in the deposition of the pure nanocrystalline diamond films.
154
Atomic H has a selective etching effect of diamond and non-diamond phases, which enables the growth of pure diamond films with low CH4 concentration when we use CH4/H2 as precursors. From our experiments, atomic N is found to have a similar effect, as the films grown with lower CH4 concentration have less non-diamond impurity. But this effect is certainly weaker than atomic H, as when the CH4 concentration is reduced to about 1.0%, there will be no film growth but only an etching effect. It seems that N has a relatively stronger etching effect to diamond phase as compared to the atomic H in a CH4/H2 environment.
14.4.5 Field Electron Emission and Transport Tunneling Mechanism
The FEE test is carried out in a high-vacuum system with a base pressure below 5 × 10-7 Pa. The film acts as the cathode adheres to a metal base. A molybdenum probe with tip area of 1 mm × 1 mm is controlled by a stepper and acts as the anode. In the current test the anode-cathode distance is kept constant at 100 µm. The I-V characteristics of the films are measured by varying the applied voltage, and the emission current is read from a Cathylet 617 nm.
It is found that with the decreasing of the CH4 concentration the FEE property of the films is significantly improved. When the CH4 flow rate decreases from 3.5 to 2.1 sccm, the emission threshold drops from 15 V/µm to 1 V/µm, while the maximum emission current increases from 0.5 to 10 mA/cm2. The results of the films grown with CH4 higher than 4.9 are not shown here, because there is no observable electron emission even when the applied field strength reaches 30 V/µm (which is the upper limit of our high voltage supply). It is worthy to note that the FEE result obtained from the film grown with 2.1 sccm CH4 is among the best up to date (Okano, et al., 1996; Zhu, et al., 1998). In particular, the present results are measured on a surface without any treatment. The corresponding F-N plots well fitted by straight lines indicate that the field emission process can be explained by a tunneling mechanism. A repeatable abrupt change of "on" and "off" states of emission is observed at two corresponding specific fields during circling of both increasing and decreasing applied gap fields (Chen, et al., 1999). A plausible explanation is given to this type of field-induced electron emission phenomenon, in which a two-layer structure consisting of amorphous carbon and diamond is proposed.
As H2 varies from 0 to 10 sccm and CH4 is kept at 3.5 sccm, the emission properties of the films show some improvement at the beginning when the H2 flow rate increases from 0 to 2.5 sccm, but soon turn bad with further increasing of H2. In the film grown with 5.0 and 7.5 sccm H2, the emission threshold becomes rather high, and the maximum emission current decreases significantly. Furthermore, in the film grown with 10 sccm hydrogen flow, there is no observable emission even when the applied field strength reaches 30 V/µm.
In order to explain the FEE mechanism of the nanocrystalline diamond films, a model based on the graphite/nanodiamond mix-phase structure is proposed. In such a structure, graphite works as a conduction channel from the back-contact metal to the film surface, while on the surface the diamond has a relatively low or even negative electron affinity. Thus electrons will first tunnel through the diamond edges and then emerge from the diamond surface. Figure 14.7 shows a schematic of this process. We find that there are two main factors dominating the field emission property in this process: (1) the diamond grain size and (2) the graphite content in the film. As electron from the diamond/graphite interface must tunnel through diamond to vacuum, the diamond grain size is a critical factor determining the tunneling probability. For the large-size diamond grains, electrons can only be emitted from the regions close to the edge, which must be
155
thin enough for an electron to tunnel through. For the small-size diamond grains, the electrons can effectively be emitted from a larger area or even the whole diamond particles. This process will greatly increase the emission site density. The graphite content is another critical factor because the emission takes place from the diamond/graphite interface. When the graphite content is large enough to fill the gaps between the diamond grains, the decreasing of the graphite content will increase the diamond/graphite interface area and thus enhance the electron emission. But when the graphite content is further reduced to a critical value that is not enough to fill the gaps, the diamond/graphite interface area will decrease, and the field emission properties will turn bad. Further decreasing of the graphite phase will even make it not enough to form a conducting channel through the film, and the emission property will rapidly be degraded.
Figure 14.7 Schematic diagram of the field electron emission from diamond/graphite mixed-phase structure.
Based on the above model, our experimental results can be understood by the following discussion. As to the first series of the films grown without H2, the decreasing of CH4 concentration will (a) increase the diamond grain size from about 8 nm to 20–30 nm, and (b) reduce the graphite content from 40% to a much lower value as shown by the TEM images. The increasing of the diamond grain size has a negative effect on the FEE. Although the change is not very significant, this effect is relatively small. The decreasing of the graphite content, on the other hand, has a positive effect on FEE. Combining the two effects, we believe that the positive effect of reducing graphite content can overcome the negative effect of increasing grain size, so an improved FEE property is achieved.
As to the other series of the films grown with addition H2, with the increasing of H2 flow rate from 0 to 10 sccm, the diamond grain size increases from 8 to 50 nm. So the trend is similar to the previous case, but the
156
much large grain size have a more negative effect on FEE. In addition, the graphite content decreases very fast with increasing H2, and the films grown with higher H2 have little graphite phase as revealed by TEM and Raman. At the beginning, i.e., when the graphite content has not been reduced too much, the decreasing of graphite content has a positive effect on FEE. Thus the FEE property of the films may be improved as a combined effect of the changes of grain size and graphite content, as seen in the experiment. But after this point, the field emission property of the films rapidly turns bad, indicating that the graphite content in the films has been reduced too low to result in a strong negative effect on FEE. In this case, graphite plays an important role in the field emission of the nanocrystalline diamond films. By optimizing the deposition parameters, a novel FEE property can be obtained.
14.5 Nanocrystalline Carbon Nitride Films
The investigation and development of carbon nitride has been a subject of intense research for the past years (Wang, 1997, 1999).
By CVD, for instance, small grains ( 0.1 µm) and nanocrystallites were grown in the films and identified to be β-C3N4 by Yen and Chou (Yen and Chou, 1995a, 1995b). Veprek et al. later showed that compact, uniform films of the composition of C3N4 could be prepared by plasma CVD in an intense nitrogen discharge, but they did not obtain crystalline films (Veprek, et al., 1995). Bhusari et al. deposited crystalline C-N films with Si content less than 5 at % on Si (100) substrates by using a microwave plasma-assisted CVD (Bhusari, et al., 1997). Two years ago, our group reported the results of the stabilization of nanocrystalline carbon nitride films (Wang, et al., 1997; Chen, et al., 1996a, 1997a; Guo, et al., 1997a, 1997b). Very recently, some American groups showed new evidence of β phase CN structure (Fahmy, et al., 1999; Jagielski, et al., 1999; Chowdlhury, et al., 1999).
14.5.1 α and β Structures
Figure 14.8 shows an ideal β-C3N4 structure. The carbon atoms are fourfold sp3-coordinated by nitrogen atoms in a regular tetrahedron, which is linked together by other nitrogen atoms in a nearly planar, threefold, sp2 coordination. The simple hexagonal phenacite structure with P3 symmetry is formed by 14 atoms: eight nitrogen and six carbon. The α-C3N4 structure is described as an ABAB… stacking sequence of layers of β-C3N4 (A) and its mirror image (B). The unit cell with space group P31C has 28 atoms.
Figure 14.8 Unit cell of a β-C3N4 structure.
157
The earliest accurate calculation of the structural and electronic properties of β-C3N4 was performed by Liu and Cohen, where they gave the indirect band gap of 3.2 eV (Liu and Cohen, 1989, 1990). By pseudopotential total energy approach they calculated the cohesive energy of β-C3N4 to be 81.5 eV per cell or an average value of 5.8 eV per atom. This moderately large cohesive energy suggests that β-C3N4 is at least a metastable structure. Corrected GW study shows that both α and β carbon nitrides are wide band gap materials (Corkill and Cohen, 1993).
Nearly pure crystalline C-N films, which are composed of α-C3N4, β-C3N4, and other unknown C-N phases were prepared on silicon and nickel substrates (Chen, et al., 1996a, 1997a). Well-faceted nanocrystallites consisting entirely of carbon-nitride network on nickel substrates are obtained from tens of nanometers to several micrometers, as shown in Fig. 14.9. The experimental lattice constants (Chen, et al., 1997a) of α-C3N4 (a = 0.638 nm, c = 0.4648 nm) and β-C3N4 (a = 0.624 nm, c = 0.236 nm) with relative N/C ratios of 1.20–1.60 on nickel are in good agreement with the ab initio calculations (Liu and Cohen, 1990; Teter and Hemley, 1996) by less than 1.3% and 2.5%, respectively. The successful synthesis of mixed-phase C-N films has prompted us to consider whether we can get a specific C3N4 phase film and the necessary experimental parameters. The first attempts to control the methane concentration and to add a fraction of hydrogen were reported by us to see the influence on the growth mechanism (Chen, et al., 1997b, 1997c).
Figure 14.9 SEM of several C3N4 columns on a nickel substrate.
158
For these samples micro-Raman spectroscopy measurements are performed. The spectra show several sharp lines in the spectral range from 500 cm-1 to 1600 cm-1, which are not related to the formation of other compounds of the participating materials, but to the formation of C3N4. The comparison with theoretically calculated vibrational densities of states and calculations utilizing Hooke's law enables the correlation of the measured phonon features with C3N4 phonon modes (Werninghuas, et al., 1998).
14.5.2 Tetragonal Structure
By applying selected-growth parameter in bias-HFCVD, Guo et al. obtained a group of CN films with a relatively higher concentration of the unknown CN phases (Guo, et al., 1997a). Some columns with prismatic tetrahedrons in morphology are observed. In order to identify their structures, more effort is being given to TEM and XRD. It is noticed that not only these tetrahedrons but also some grains with irregular shape present tetragonal symmetry. Further corresponding SAED patterns are taken for the tetragonal phase with different shapes and crystal axes (orientations). On the basis of XRD and TEM results, the lattice parameters for the new tetragonal CN phase are determined as a = 0.556 nm and c = 0.275 nm. The N/C ratio of the new tetrahedral phase is about 0.8–1.0 by energy-dispersive X-ray (EDX).
14.5.3 Monoclinic Structure
A monolinic CN structure with some relatively larger grains was identified in the CVD growth films (Guo, et al., 1997b). A detailed study of the TEM micrographs and SAED patterns of the monoclinic grains with different main crystal axes has been presented. From the scanning electron microscopy (SEM) and TEM
159
results, it is found that most of the monoclinic CN grains exhibit irregular surface morphology, which is quite distinguishable from those of the α, β and tetragonal CN phases in morphology. Combined with TEM, further XRD tests identify the new monoclinic CN structure with lattice parameters of a = 0.5065 nm, b = 0.115 nm, c = 0.2801 nm and β = 96°. The EDX analysis shows that the N/C ratio of this new phase is 0.5–1.0. With this identification, Guo et al. claimed that the nearly pure crystalline CN films on nickel substrate by bias-HFCVD are mainly composed of α, β, tetragonal and monoclinic CN phases. However, it should be noted that the atomic arrangement and the accurate stoichiometry are still unknown at the present time for these two new phases (Guo, et al., 1997b).
14.5.4 Fullerene-like Structure
From quantum-chemical calculations and photoelectron spectroscopy, Sjöström et al. proposed a fullerene-like microstructure to consist of a network of buckled sp3-hybridized CNx planes, crosslinked by sp2-hybridized bonds (Sjöström, et al., 1995). The CN films are deposited on Si (001) substrates by dc magnetron sputtering of a graphite target in pure nitrogen discharge. Nitrogen concentration is below 30 at % by RBS analysis. HRTEM is performed for the samples prepared by mechanical cleaving. The structure can be described as graphite-like with interplanar distances of 0.347, 0.209 and 0.120 nm as obtained from SAED patterns. Sjöström et al. pointed out that this structure is different from both graphite and turbostratic carbon, while is similar to that of fullerene-like buckey-onions. The main difference is the basal planes. In the present structure they are interlocked with covalent bonds of a much shorter bond length than that of the van der Waals bonds between the basal planes in graphite. The planes observed with ring and fingerprint shapes indicate that pentagons are presented in the structure.
14.5.5 Carbon Nitride/Diamond/Silicon Layers
It is known that both diamond and CN are wide band gap materials with high hardness, high wear resistance, and high thermal conductivity. A better approach, therefore, will be to integrate the superior properties of CN films with the mature technology of diamond, since the technology of growing diamond films is now well established and it is very easy to deposit high-quality polycrystalline diamond films on various substrates. The experiments are carried out electron cyclotron resonance microwave plasma-assisted chemical vapor deposition (ECR-MPCVD) apparatus, discussed in section 14.3.2. A two-step growth mode has been adopted in which a diamond layer is first deposited onto the substrate (Si or Mo), and then the carbon nitride films are grown (Wu, et al., 1998).
The SEM is used to study the surface morphology of the film before and after CN deposition (Wu, et al., 1998). It can be seen from the images that the diamond layer is quite dense, which is also proved by our further test that the Si 2p signal is not present in the XPS spectrum. The CN films are usually composed of many hillocklike crystal grains without prismatic morphology as deposited directly on silicon and nickel substrates (Chen, et al., 1996a, 1997a). Figure 14.10 shows the center of a CN grain, which looks like a sunflower composed of many small CN fibers.
Figure 14.10 SEM of the center of a CN cluster on diamond layers.
160
From typical X-ray photoemission spectroscopy (XPS) C 1s spectra of the films before and after CN deposition, it is found that there is no clear difference in the C 1s peaks. The peak shape and bond energy remain almost unchanged after deposition, only the full-width-half-maximum (FWHM) value are slightly enlarged from 1.3 eV to 1.4–1.5 eV. For the N 1s results, there is absolutely no N signal in the film before deposition, showing that the diamond film does not contain nitrogen impurities. But after deposition, a clear N 1s peak appears at 398.2–398.4 eV. This value is different from the binding energy of molecular nitrogen at 404 eV, proving that the N atoms are chemically bonded to C atoms. Further study shows that the nitrogen concentrations in the films are almost kept as a constant when the substrate temperatures vary from 100°C to 700°C, which suggests a stable CN phase has formed.
In addition, defects in this structure are investigated by infrared light scattering tomography (Ma, et al., 1999a). Most defects in the CN/diamond/silicon multilayers are introduced by an extended growth of the original defects in Si substrate determined through layer-by-layer tomography. The defect type is analytically thought to be dislocation clusters agglomerated by an interstitial-type defect.
14.5.6 Physical and Chemical Properties
In most machining applications, hardness is only one of many properties which such a material has to meet (Veprek, 1999). The importance of this property can be illustrated by the fact that today more than 40% of all cutting tools are coated by wear-resistant coatings and the market is growing fast. Fujimoto and Ogata tested the amorphous CN1.94 films deposited onto tungsten carbide substrates by a Knoop type diamond indenter with a 10.0 gf (1 gf = 9.81 × 10-3 N) load (Fujimoto and Ogata, 1993). The reported Knoop
161
hardness of these samples is 6500 kgf/mm2(1 kgf/cm2 = 9.81 × 104 Pa). This should be compared to the Knoop hardness of 5500 kgf/mm2 for c-BN made by the same experimental setup. The superhardness of 60 GPa is obtained from a CNx film with fullerene-like microstructure, which is closest to that for diamond ( 100 GPa) (Sjöström, et al., 1995). In addition, nanoindentation of the film shows an elastic recovery of 85%. These results present the superhard nature of the C-N bonds, but even so there is no direct hardness test for β-C3N4 because no high-quality single crystals are available at present time.
Apart from its extraordinary mechanical property, carbon nitrides show great promise in transport properties. Zhang et al. studied the electrical properties of carbon nitride materials synthesized on quartz substrates by pulsed laser deposition (Zhang, et al., 1996). It is found that the increase of resistivity with increasing nitrogen composition is highly nonlinear, which is opposite to nitrogen implanted diamond-like films. In addition, they found that the thermal conductivity of the C-N film is 0.8 to 1.3 W/(m · k) at 300 K. This value is among the highest observed for any type of amorphous materials. All the results show that carbon nitride is a very good electrical insulator and thermal conductor for high-performance electronics.
An optical band gap of 2.7 eV was determined by Ogata et al. using UV transmission spectrometry for a nitrogen-rich C-N film (Ogata, et al., 1994). Another study presented an optical band gap of 3.81 eV for a crystalline C-N film with small amount of silicon (Lin, et al., 1997). Large area field emission has been observed from thin C-N films (Chen, et al., 1998b). Recently, Wu et al. reported high resistance of C-N films against acid and electrochemical etching (Wu, et al., 1997).
Although the recent spate of success on CN research is impressive, much work remains to be accomplished. The synthesis of high-quality and large CN crystal with interesting predicted properties remains a challenge for the future.
14.6 Nanocrystalline Silicon Carbonitride Films
The SiCN compounds have also motivated much attention because they exhibit some new features other than those of mixtures of crystalline Si3N4and SiC phases. Fabrication of SiCN films was realized by reactive magnetron sputtering (Komateu, et al., 1990), ion or plasma sputtering (Novikov, et al., 1992; He, et al., 1996), ion implantation (Uslu, et al., 1996), RF nitrogen plasma-assisted pulsed laser ablation (Thärigen, et al., 1999), and plasma-assisted (Zhang, et al., 1994; Chen, et al., 1996c) thermal chemical vapor deposition (Bendeddouche, et al., 1997). Recently, the first ternary crystalline SiC2N4 and Si2CN4 have been studied both experimentally (Riedel, et al., 1997) and theorectically (Wang, et al., 1998; Lowther, 1999).
First-principles calculations are performed for the structural properties of hexagonal β-Si3N4, β-C3N4, and two model structures of β-Si2CN4 and β-SiC2N4 (Wang, et al., 1998). Bulk modulus and lattice constants α and c have been obtained by fitting the total energy calculations to the Birch equation of state. As more C atoms are substituted for Si in β-Si3N4, the bulk modulus is observed to progressively increase up to 4.44 Mbar(1 bar = 105 Pa), comparable to that of diamond (4.43 Mbar), and constants α and c are reduced (see Table 14.1). The averages of the calculated lattice constants for the four systems, α = 0.709 nm and c =
162
0.263 nm, appear to be close to the experiments, α = 0.706 nm and c = 0.272 nm, respectively (Chen, et al., 1996a; Wang, et al., 1998). Recently, these results have been confirmed by Lowther (Lowther, 1999).
Table 14.1 Calculated lattice constants and bulk modulus of Si3N4, Si2CN4, SiC2N4 and C3N4
α(Å) c(Å) c/α B0(Mbar) B′0
Si3N4 7.65 2.84 0.372 2.52 3.8
Si2CN4 7.30 2.72 0.372 2.87 5.0
SiC2N4 6.92 2.57 0.371 3.19 3.0
C3N4 6.43 2.39 0.372 4.44 3.2
By using bias-assisted hot filament CVD method, Gong et al. have synthesized SiCN films under a wide variety of deposition conditions (Gong, et al., 1999). The uniqueness lies in the mere reactive Si source totally from Si substrate and combination of thermal decomposition and plasma enhancement. The effect of deposition parameters on the morphologies of the films is demonstrated by SEM images. The relationship between deposition parameters and the chemical environments of the elements, especially carbon and nitrogen, in the films is discussed on the basis of XPS analyses. The chemical composition of the films is studied by Auger electron spectroscopy (AES) depth profile and EDX. For depth profiling with Ar+ ions, though the energy transferring efficiency between Ar and N atoms is much lower than that between Ar and Si, yet N atoms are preferentially lost due to weak binding at the surface and high diffusivity in the sputter damaged solid. This generally results in N concentration lower than the true value. The preferential loss of N atoms has been observed in amorphous CN films when they are annealed or exposed to an electron beam.
14.6.1 Deposition with Nitrogen and Methane
14.6.1.1 Influence of Methane Flow Ratio
In the present experiments, the silicon atoms only come from the substrates. After 30 min deposition, a thin-film layer (about 0.3 µm in thickness) fully covers on the substrate surface, so not enough silicon atoms further diffuse from the substrates. The deposition of SiCN film is saturated. The films with some embedded nanosize crystals are observed by SEM, which demonstrates that some typical clusters are composed of hexagonal columnar crystals. At lower CH4/N2 flow ratio, the facets of the crystals continue to integrate, while the crystals lose surface symmetry when a high ratio of 3.5% is used. Moreover, on a sample grown at a median ratio, the density of the clusters is relatively higher. By EDX the atomic ratio of carbon and nitrogen is about 3:4, while a small amount of oxygen is also found. Besides the embedded clusters, the overall carbon content of the remaining flat film is much lower by AES depth profiling after 2 min sputtering, and there is abundant silicon element as shown in Fig. 14.11. It suggests that the flat part, around the C-N nanocrystals, of the films is mainly composed of Si and N elements. The sputtering rate used here is about 5 nm per minute. As the depth increases, the silicon content increases linearly and nitrogen decreases,
163
while carbon content remains constant. From the slope of nitrogen profile, the thickness of nitride layer is estimated to approach 100 nm. Kubler et al. studied thermal nitridation of Si (100) surface by NH3 from room temperature to 800°C (Kubler, et al., 1988). They found that the nitridation is rapidly saturated at high temperature and limited in thickness. Considering the higher temperature (>800°C), higher gas pressure (3 kPa), and plasma enhancement and etching applied here, nitrogen diffusion and penetration into the substrate could be promoted. Therefore, a thick nitride overlayer is formed, while the nanosize C-N crystals shown are simultaneously grown under suitable local conditions in this layer.
Figure 14.11 AEG depth profile of a SiCN sample with 100 sccm N2 and 3.5 sccm CH4 at the substrate temperature of 850–900°C.
The film compositions and chemical bond stages are analyzed ex situ by XPS with a VG ESCALAB MKII system where Mg Kα line is used and the energy resolution is about 0.5 eV at 800 eV. The XPS C 1s and N 1s spectra of the samples show both carbon and nitrogen have more than one bonding state. The XPS Si 2p spectra of the samples show only one bonding state near 101.5 eV for silicon. It is readily assigned to Si-N bond existing mainly in the nitride overlayer of the Si substrate (Wagner, et al., 1982). There are apparently two XPS peaks around 286.6 eV and 284.8 eV, respectively. Compared with the C 1s binding energy of graphite at 284.0 eV and of silicon carbide at 283.8 eV, the latter reasonably corresponds to C-Si and/or C-C bond. Thus, the former with higher binding energy can be assigned to C-N bond because of the larger ionicity of nitrogen atoms (Ogata, et al., 1994). It is supposed that C-N bonds exist in the nanosize crystals. Due to the low carbon concentration, the main N 1s peak is ascribed to N-Si bond, the binding energy of which is around 398.7 eV (Karcher, et al., 1984). The line broadening could be explained as the presence of
164
complex local environments for N atoms. It is noted that N 1s exhibits another peak at 396.8 eV. On account of the strong peak at 286.5 eV of C 1s, this can be attributed to C-N bond in the sample that contains some C-N particles. This binding energy is a bit lower than the 286.6 eV for CN2 reported previously (Marton, et al., 1994). If the assignment is true, the N atoms should be bonded to sp3-hybridized C rather than sp2-hybridized C (Zheng, et al., 1996). A smaller percentage of C bonds are ascribed to C-N and more to C-Si/C-C in the samples of CH4 with 1.0 sccm and 3.5 sccm. It suggests that a median CH4 concentration of 2.1 sccm is probably more advantageous to C-N bonding. If a perfect crystal morphology reflects an ideal bonding structure, the inference above would be consistent with the changes in the crystal forms.
14.6.1.2 Influence of Substrate Temperature
In order to study the temperature influence, some samples are prepared in three different temperature ranges of 850–900°C, 900–1000°C, and above 1100°C, respectively, with a fixed CH4/N2 ratio of 2.1%. As shown by SEM, higher temperature results in second nucleation on previous crystals and disorder of crystal orientation (Gong, et al., 1999). When the temperature is raised to above 1100 °C, regular crystals disappear and gross tiny particles are left. As the temperature increases, the deposition rate (about 12 nm/min) is found to increase slightly. But the crystalline contents in the film are still very limited. In XRD analysis, only some samples prepared at very high temperature exhibited a few diffraction peaks attached to Si3 N4.
There also exist two C 1s states in the XPS spectra, but their positions shift slightly apart when the temperature increases from 850°C to over 1100°C. The proportion of C-Si/C-C eases quickly as temperature rises. It indicates that C-N bonds could be unstable at high temperature. Consequently, the SiCN films prepared at high temperature could be characterized as SiN/SiC mixture. The Si 2p peaks change little except for a small one at 99.2 eV in the XPS spectrum of the sample grown at a temperature above 1100°C. Considering the abundance of C-Si/C-C in this sample, it may be linked with Si-C bonding, but the possibility of assignment to Si-Si is not precluded (Kaplan, 1984).
14.6.2 Deposition with Nitrogen, Methane and Hydrogen: Influence of Hydrogen Flow
Ratio
When hydrogen gas is introduced into deposition, the growth speed (about 100 nm/min) is elevated by about 10 times. SEM shows the changes in morphologies of the films at increased hydrogen flow ratios. When the hydrogen content is small, there are many polyfaceted crystalline particles with a size of 500 nm. As hydrogen ratio is increased, the particles become smaller and agglomerate in balls in size of 1 µm. Finally, the particles become less than 100 nm and form big cauliflowers. XRD analysis indicates that there is some crystalline Si3 N4 content in those films. However, the XRD peaks are low and broad, which are relative to disorder or small crystal size. Furthermore, AES depth profile exhibits about 10% carbon content in the film. Therefore, the structures and properties of the films differ from those of pure Si3 N4.
The XPS C 1s and N 1s spectra of the samples show the impact of the introduced hydrogen gas on the chemical states in the films. When the hydrogen concentration is small, there is still a little C-N bonding content and most bonds belong to C-Si/C-C. When the hydrogen ratio is increased, more carbon bonding
165
transfers from C-N to C-Si/C-C and finally all become the latter, whose binding energy increases to 285.5 eV when the hydrogen concentration reaches to 30%. Simultaneously, the two peaks are observed to shift apart slightly as the hydrogen ratio increases, while nitrogen appears almost purely N-Si bonds once hydrogen is applied. For the uniformity of the films, that means a microscopic mixture of SiN and SiC compounds occurs in the particles due to the function of hydrogen. It seems that a few N atoms in the SiN network are replaced by C atoms and that C and N atoms are bridged by other atoms. There are also bonding states near 99.0 eV in Si 2p spectra, which is ascribed potentially to Si-Si bond. The broadened peak suggests local disorder of silicon bonds in this sample with 30 sccm hydrogen at 950–1000°C. Previous works in CNx alloys have shown an important dependence of the electron energy core level values on x (Ronning, et al., 1998).
It is reasonable to link the Si-CH4-N2-H2 and Si-NH3 systems (Kubler et al., 1988), for there exist similar reactive species, such as NH, NH2, H, etc. To deposit SiNx thin films, Kubler et al. used electron beam evaporation of Si surface under an NH3 ambient. Here, ion bombardment of negatively biased Si substrate under cathode plasma sheath is essentially applied. Therefore, the solid-gas interface diffusion and reaction in the two comparable systems would apply some common mechanism. Compared with the situation without hydrogen, the hydrogen expedites the interface reaction by rapidly producing larger amount of activated Si atoms near the substrate surface. Thus the nitride overlayer becomes thicker and more uniform. The depth-profile illustrating its chemical composition remains constant in a bulk of at least 50 nm thick. The total film thickness is estimated to be 1 µm or so by SEM cross sectional images. One possible reason to have a distant Si diffusion during the deposition is due to the hydrogen addition and ion bombardment which activate some Si or Si-H ion clusters above the substrate surface with the assistance of the negative bias and high substrate temperature.
14.6.3 Lattice-Matched Growth Model
It should be noticed that some beautiful Si containing CN flowers are observed in Fig. 14.12 after 40 min deposition by HFCVD. It is composed of many columnar crystals with hexagonal facets of 20–200 nm in across radiating outwards. A possible growth mechanism is proposed. It obeys a "lattice-matched selection". Starting from random nuclei only those crystals with matched lattice to the substrate by an angle γ will survive. As growth proceeds, more and more crystals nucleate on the faces of the formed crystals. Then, the same as before, only those new crystals with a matched lattice to the faces by an angle will continually grow, whereas all the others are gradually buried. The crucial property, which determines the probability of survival, is the lattice match.
Figure 14.12 A silicon-containing CN cluster on Si substrate.
166
14.7 Turbostratic Boron Carbonitride Films
Due to the structural similarity between graphite and h-BN, the solid solutions of BCN with graphite-like structure were proposed and prepared (Yu and Wang, 1999). Theoretical calculation and experimental study indicate that the BCN compounds possess semiconducting property between those of semimetallic graphite and insulating h-BN (Liu, et al., 1989; Watanabe, et al., 1996). The importance lies in that the electric property of the BCN compounds can be controlled by varying atomic composition and atomic arrangement (Yu, et al., 1999).
The BCN compounds are prepared in a bias-assisted hot-filament chemical vapor deposition apparatus discussed in Section 14.3.1. High purity N2, H2, CH4, and B2 H6 (diluted in N2 at the concentration of 10%) are used as reactive source gases. The total gas flow rate is 100 sccm and keeps a constant during the deposition. The concentrations of H2 and CH4 in the gas mixture are kept at 37% and 4%, respectively. The film compositions are changed by varying the B2 H6 concentration from 0.5% to 3.0%. The filament temperature is about 2050°C, and the substrate temperature is about 800°C.
14.7.1 Morphology and Composition
167
From SEM study it is manifested that the morphology of the deposited BCN films on Mo substrate depends greatly on the substrate temperature. Three different regions for BCN film growth are divided. They are above 1173 K, around 1073 K, and below 973 K (Yu, et al., 1999).
From XPS spectra of the films grown at temperature above 1073 K some heteropolar bonds between substrate Mo and elements B, C, or N are observed. At temperature below 1073 K, however, the interference of Mo element is dramatically decreased. It is indicated that at all temperatures the binding energies of B 1s, C 1s, and N 1s peaks are centered at 189.7, 284.3, and 397.5 eV, respectively. It shows that there exists a stable phase in the films. The chemical compositions, which are shown in Fig. 14.13 with substrate temperatures, are calculated from areas under Gaussians using atomic sensitivity factors. Generally speaking, the N concentration decreases with increasing substrate temperature, and it reaches the highest at 973 K. The C concentration is its lowest at 1073 K. The N concentration (0.09) is much lower than that of C (0.37) and B (0.54) at 1273 K, and it seems a boron carbide phase formed at this temperature. As the C and N active species are sufficient for the formation of BCN compounds in the process, the BCN compositions are determined mainly by substrate temperature. The formulas of the BCN compounds deposited at different temperatures are as follows: (a) 873 K, B0.83 C0.17 + B0.39 C0.35 N0.26; (b) 973 K, B0.30 C0.34 N0.36; (c) 1073 K, B0.64 C0.36 + B0.51 C0.23 N0.26; (d) 1173 K, B0.51 C0.31 N0.18; and (e) 1273 K, B0.37 C0.54 N0.09, see Fig. 14.13.
Figure 14.13 Chemical composition of B-C-N phase in deposited films versus substrate temperature.
14.7.2 Turbostratic Structure
168
Figure 14.14 shows the X-ray diffraction (XRD) patterns of the BCN films on Si substrate for different chemical compositions. The wide (002) and (100) diffraction peaks reflect the nature of the turbostratic structure, where the hexagonal BCN network sheets are irregularly stacked in the directions of the c axis. Apart from the turbostratic BCN no other phases are found in the spectra. It is shown that the films are composed of small size grains, and the crystallinity does not change much with composition. It is noted that different from most of the early reports the peak positions here do not change with the film compositions. The interlayer spacing of the obtained BCN compounds is determined to be 0.349 nm, which is larger than pure graphite (0.3348 nm), boron carbide (0.3336 nm), and boron nitride (0.333 nm) (May, et al., 1992; Takahashi, et al., 1979). The crystallinity and d-spacing of the BCN films are independent of the film compositions. The same case is also obtained even when the B concentration reaches 70 at. %. It is believed that the three kinds of B, C, and N can be incorporated into the turbostratic structure with unlimited concentrations from graphite to graphite-like BCN to h-BN. Well-crystallized graphite-like structures are only observed at small amount of B or N doped into carbon layers or C doped into h-BN films.
Figure 14.14 XRD results of BCN films with different chemical compositions (a) B0.12 C0.71N0.17, (b) B0.32 C0.21 N0.47, (c) B0.36 C0.30 N0.34, (d) B0.45 C0.17 N0.38, (e) B0.54 C0.13 N0.33, (f) B0.72 C0.13 N0.15.
169
The HRTEM images of the BCN compounds are shown in Fig. 14.15. It clearly shows the turbostratic nature of the BCN films. The crystallinity is inhomogeneous in the films. Mostly, the deposits are the mixture of the turbostratic and amorphous domains, and in some cases the layers are embedded in the amorphous domains. The codeposition of the layered structure and amorphous domains indicates that the kinetic factor as structural fluidity and concentration distribution of active species play a key role in the formation of the BCN structure during the CVD growth. The layer continuity is frequently obstructed by the amorphous domains and defects, and the size of the turbostratic region ranges from several to over 50 nm. The layers are randomly oriented with various angles, and the curved layers with various curvature are formed. It is believed that the crossed layers are bonded with each other at some junctions. Even by adjusting experimental conditions over a wide range, only turbostratic structure is obtained in the present study. Because of the different atomic radius of B (0.076 nm), C (0.067 nm), and N (0.060 nm) the randomly incorporated B and N atoms in the graphitic structure cause local distortions and thus induce sheet curvature. The formation of the turbostratic structure may also be related to the poor structural fluidity at the low growth temperature and the over-high decomposition rate of B2H6 during the CVD process. It is known that the crystallinity of the turbostratic BCN films will be improved by heat treatment at high temperature of 2000°C (Filipozzi, et al., 1995) and the crystallinity of h-BN also improved with increasing growth temperature (Takahashi, et al., 1979).
Figure 14.15 HRTEM image of the turbostratic BCN structure.
14.7.3 Raman and Photoluminescence
170
Two very broad asymmetric bands centered at 1544 cm-1 and 2935 cm-1, respectively, are presented in the Raman spectrum. The band centered at 2935 cm-1 is characterized by second order Raman scattering. Different from other reports, no distinguishable D and G band, which are in general presented in Raman spectra of poor graphitized or amorphous carbon and carbon related materials, are found in the spectra (Yu, et al., 2000). It is found that the band width and band position remain almost unchanged with changing film composition. This implies that the film structures are about the same in this composition range. The change of the band intensity with film composition is probably related to the increase of antisymmetric vibrations with increasing B content. The Raman results here are consistent with that of XRD and HRTEM.
In order to further determine the bonding states of the deposited BCN films, IR measurements are made. All the IR spectra show broad absorption band from about 1150–1670 cm-1. The absorption of this broad band increases with increasing B content in the films. This broad absorption band is believed to correspond to the broad Raman band centered at 1544 cm-1. Due to the incorporation of the B and N atoms into the graphite network and the breaking of network symmetry, the Raman active band becomes IR active. This broad absorption band is formed by the overlapping of a diversity of vibration modes. It is well known that the vibration characteristic absorption frequencies of some cluster or bond in a polyatomic molecule are affected by its atomic environments, so the differences in film composition and structure will cause the shifts of absorption frequency.
The room temperature PL results of the BCN films are obtained for various compositions (Yu, et al., 2000). The broad PL bands show the characteristic of amorphous materials. It is found that with increasing the B content the PL peak energy shifts from about 2.80 to 3.40 eV. This shows that with increasing boron content the band gap of the BCN compounds increases, which is consistent with the prediction that the electronic structure of the BCN compounds can be controlled by changing its composition. It is indicated that the BCN materials emit blue to violet to ultraviolet light for different film compositions. It was reported that the PL peak energy at room temperature BC2N is 2.07 eV (Watanabe, et al., 1996). Because of the exceptionally high energy of emitting light, BCN compounds are interesting and important candidates for blue-light emitter.
14.7.4 Field Electron Emission
Furthermore, we have studied the field emission behavior of the BCN films (Yu, et al., 1999). The films used for field emission tests are deposited on polycrystalline molybednum substrates with film compositions of B 19 at. %, C 60 at. %, and N 21 at. %. The F-N plot almost follows a linear relationship, which indicates a Fowler-Nordheim tunneling mechanism. When assuming that the BCN film is an ideal plane emitter with a field enhancement factor β of 1, then the work function = = 0.044 eV is obtained from the slope of the straight line. Such values have been reported by different authors for α-C (Silva, et al., 1997), diamond-like carbon (Lee, et al., 1996), and h-BN (Sugino, et al., 1997). The emission current of 0.31 nA can be detected at the applied electric field of 4 V/µm, which is regarded as the threshold electric field. The maximum emission current of 0.31 mA is obtained at the electrical field of 13 V/µm. Although even lower threshold fields are reported for nanodiamond and carbon nanotubes, the BCN films have attractive advantage as field emitters for their smooth surface, low cost, easier fabrication, and better mechanical properties. Furthermore, it is expected that the field emitting properties of the BCN films can be improved by adjusting composition and structure from amorphous to turbostratic to crystalline.
171
14.8 Polymerized Nitrogen-Incorporated Carbon Nanobells
Carbon nanotubes were first studied by Iijima. For a practical application, however, a uniform doped nanotube is still a remaining challenge. Polymerized nitrogen-incorporated carbon nanobells are prepared from the source gases of CH4 and N2 by MPCVD discussed in Section 3.2 (Ma, et al., 1999b). Mesoporous silica plates (Novak, 1993) containing catalytic iron and nickel nanoparticles are used as substrates, these nanoparticles being essential for both the initial nucleation and the subsequent growth of carbon nanotubes. The working temperature (500°C) of the substrate is well below the operating temperature (900–1000°C) employed in the MPCVD method reported previously (Kuttel, et al., 1998). The films produced in this way are composed of large quantities, e.g., up to 0.8 g/d from a single chamber, of aligned nanotubes.
The XPS of the specimens indicate a covalently bonded C-N arising from nitrogen doping into the carbon network. The C 1s spectrum presents a main peak at 285.4 eV and a distinct shoulder at 286.9 eV. The former is characteristic of graphitic carbon and the latter indicates that the carbon atoms are bonded by nitrogen. The N 1s peak is located at 400.9 eV, which confirms the presence of nitrogen atoms in a graphitic-like structure (Casanovas, 1996). The presence of nitrogen in individual nanofibres is also detected by spatially resolved electron energy loss spectroscopy (EELS) and the level of nitrogen doping is found to be as high as 10%.
14.8.1 Polymerized Nanobell Structure
The HRTEM images reveal that conventional nanotubes with long cylindrical structures are not present in the samples. Instead, a bamboo-like structure is observed from these nanofibers (Fig. 14.16(a)). A similar morphology was previously observed from pure carbon fibers prepared under catalytic growth conditions (Audier, et al., 1981). However, the latter, which forms during a high-temperature treatment, has a very thin conical top and the surface of the fibers is not smooth. In comparative experiments, carbon nanotubes prepared from pure CH4 without using N2 under exactly the same conditions invariably form elongated cylindrical structures and no bamboo-like morphology is observed. Closer examination of the HRTEM images indicates that the nitrogen-doped carbon nanofibers are actually linearly polymerised short nanotubes with one end sealed and another open, and designated carbon nanobells (CNBs) (Fig. 14.16(b)). The atomic layers in the walls of the CNBs are parallel to each other with an interplanar spacing of about 0.34 nm, which is similar to that of multiwall carbon nanotubes. Unlike carbon nanotubes, however, the diameter of a nanobell increases continuously from the top to the bottom, i.e., no part of the walls is parallel to the axis of the nanobell. The formation of such a conical structure is probably due to the presence of an excess of pentagons near the upper (or closed) end of the bell, but in addition, it seems apparent that all the graphitic sheets stop growing when their open edges extend to a certain distance from the center of the catalytic particles. The detailed mechanism for this is by no means clear, but since the metal nanoparticles serve only as initial nucleation centers on the substrate, and all individual nanobells are completely self-contained throughout the entire length of the nanofiber, it must be dependent on the presence of nitrogen in the gas mixture, as confirmed by experimental findings. As a result of being built up of nanobells, the outer longitudinal surface of the nanofiber is no longer terminated with a single graphitic
172
layer as in the normal carbon nanotubes, although it looks just as smooth in the SEM or the low magnification TEM images. In the nanofibers built up from nanotubes, however, the greater part of the fiber surface consists of apparently open ends of the graphitic sheets! This novel surface structure undoubtedly plays a major part in determining the unusual physical properties of these nanofibers.
Figure 14.16 (a) HRTEM images of a single fiber showing the polymerized carbon nanobells. (b) The detailed structure of the nanobells.
Many nanofibers have larger diameters, i.e., in a range of 100–200 nm. The structures of these nanofibers seem to be much more complicated. However, the HRTEM images from these nanofibers again indicate the same principle of polymerisation of nanotubes. Apart from the differences in diameter between the thin and thick nanofibers, the wall thickness of the latter type is often greatly reduced and the nanobells are apparently much less rigidly arranged, often displaying distortions and various defects. Nevertheless, the surface atomic structure of these thick nanofibers, which is of main interest, is the same as that of the fine ones. Both types of nanofibers are straight and lack the helical morphology observed in the previously reported C-N nanofibers: they can therefore be aligned easily in thin films and are consequently suitable for large area field emission experiments.
14.8.2 Chemical Separation and Application
These nanofibers can be used as precursors to produce isolated nanobells, which may serve as the basis of nanoreactors, allowing catalytic particles to be loaded. In addition, these nanobells can be regarded as unclosed multiwall fullerenes and may be used to store various large molecules or clusters of metal atoms followed by closing the open ends in a suitable treatment (Zhou, et al., 2000).
173
It is known that direct control of the length of carbon nanotubes in a nanometer scale is extremely difficult. Therefore, the nanofibers containing polymerised nanobells are good precursors for producing short nanotubes. The open edge of the nanobells is a highly reactive chemical center. Several methods for the oxidation and opening of carbon nanotubes have now been published (Tsang, et al., 1994), and of the solution methods acidified potassium permanganate seems to offer the best results. We therefore used this method to corrode the open edges of the nanobells, and completely isolated nanobells are eventually obtained (Zhou, et al., 2000). It is found that the reaction time is crucial, as when the time is too short the nanobells are still connected to each other in the nanofibres. Conversely, if the sample is over-reacted (e.g., longer than 4 days), the nanobells are mostly damaged.
14.8.3 Wall-Side Field Emission Mechanism
The typical field electron emission characteristics, i.e., both the spatial distribution of emission sites and the total emission current-voltage (I-V) characteristic of the specimens, are investigated with the transparent anode imaging technique. A quite low threshold field of 1.0 V/mm is observed, which is considerably lower than the value of 1.5 V/mm from pure multiwall carbon nanotubes reported by Kuttel et al. (Kuttel, et al., 1998). The highest current density detected from the specimens is about 200 mA/cm2 for an applied field of 5–6 V/mm. The relative fluctuation from a specimen at a current density of 150 mA/cm2 is 1.3% during a test period of 200 s and no significant degradation of the current density is observed over 100 h.
Figure 14.17(a) shows a typical spatial distribution of the emission sites on a piece of film, which has an irregular shape as shown in Fig. 14.17(b). Apart from the strong emission sites appearing as white pattern in the picture, a number of individual dots corresponding to "weak" sites are also observed. The overall uniformity of emission is demonstrated by the similarity between the image and the real specimen shape.
Figure 14.17 (a) The spatial distribution of emission sites of a nanofiber film. (b) The shape of the film. (c) A multisegments microscopic emission image. (d) A single-segment microscopic emission image.
174
Further studies concentrating on the "weak" sites show that the field emission pattern consisted of a number of bright segments of circles (Fig. 14.17(c)). The images having a single segment, as shown in Fig. 14.17(d), can be obtained at relatively low gap fields. Some moon-like segments rotated during the observation to other parts of circles with the diameters matching the size of the nanofibers. Consequently, the patterns consisting of full circular bright spots can often be observed as time progresses. Saito et al. obtained the field emission patterns from both close-top and open-top multiwall carbon nanotubes (Saito, et al., 1997). Their images show full bright rings due to the electron emission from the circular edges of the graphite layers. In our samples, however, whether before or after undergoing an emission test, neither graphite flakes nor open-top nanofibers are detected. The only possible mechanism of formation of these segment images is therefore that the electrons are emitted from the sides of the nanofibers, which are not standing exactly parallel to the applied field. More precisely, it appears highly probable that the emission sites are the open ends of the nanobells.
175
14.9 Highly Oriented Boron Carbonitride Nanofibers
Different experimental methods, such as electric arc discharge (Stepham, et al., 1994; Weng-Sieh, et al., 1995; Redlich, et al., 1996; Suenaha, et al., 1997), pyrolysis (Terrones, et al., 1996; Sen, et al., 1998), and laser ablation (Zhang, et al., 1997), etc., have been adopted to grow BCN nanotubes. Microsize BCN rods are also reported by HFCVD (Yu, et al., 1998). However, these studies only present some nanotubes/nanofibers in random orientation, tangly distribution and low yields, which hamper both the fundamental and the applied studies of the BCN nanotubes. The large-area highly oriented BCN nanofibers are first synthesized directly on polycrystalline nickel substrates by bias-HFCVD (Bai, et al., 2000).
14.9.1 Microstructure and Composition
Figure 14.18 shows the well-aligned BCN nanofibers on the nickel substrate. The growth time of this sample is 20 min. The BCN nanofibers are obviously perpendicular to the substrate surface and are of more or less similar height. However, the site distribution of the nanofibers is not quite uniform, which is due to the insufficiency of nickel as catalyist along the grain boundaries of polycrystalline nickel substrate (Huang, et al., 1998) and also probably results from uneven ion bombardment during nucleation period. The diameters of BCN tubes are in the range of 200–400 nm, and the average density of the nanofibers is estimated to be about 108/mm2.
Figure 14.18 SEM image of large-scale oriented BCN nanofibers.
176
HRTEM is used to determine the interior and wall structures of the BCN nanofibers. It definitely shows that the nanofiber is a multiwalled centrally hollow tube. The BCN nanofibers in the present work have very similar structures to those of large CN nanofibers discussed in above section. A growth mechanism for pure carbon tubular structures is given on catalytically active Ni110 (Kuang, et al., 2000). However, a distinct feature of the BCN fibers is that many small graphitic spines stand on the surface of the fibers, making the fibers cactus-like.
The EELS measurements of K-edge absorption for B, C, and N are used to determine the chemical composition of nanofibers. Figure 14.19 shows the local EELS spectrum of a nanofiber. The typical ionization edges are observed at about 188, 284, and 401 eV, which correspond to the characteristic K-shell ionization edges of B, C, and N, respectively. Each core edge fine structure consists of a sharp π* peak and a well resolved σ* band characteristic of sp2 hybridization. This attests that the three atomic species are arranged in planes of graphite-like hexagonal rings. Using a method described by Egerton (Egerton, 1986), the chemical composition of this BCN nanotube can be determined quantitatively from EELS. The analysis of the data reveals that the ternary BCN compound is B1.0C0.64N0.60. Furthermore, we show that the composition of the BCN nanofibers can be controlled artificially (Bai, et al., 2000).
Figure 14.19 EELS spectrum of a BCN nanofiber.
14.9.2 Field Electron Emission
Furthermore, we have studied the field emission behavior of the BCN nanofibers. An emission current of 0.26 nA can be detected at the applied electric field intensity of 1.8 V/µm, which is regarded as the threshold electric field (Fig. 14.20). A maximum emission current of 0.92 mA is obtained at an electric field of 8.6 V/µm, and the corresponding maximum emission current density is about 120 mA/cm2. These results show that the BCN nanofibers have lower threshold electric field and higher emission current than that of
177
BCN films reported in Section 14.7.4. The field emission properties can be improved by adjusting the composition of the BCN nanofibers.
Figure 14.20 Field emission current vs. applied electric field of the BCN film.
14.10 Conclusions
The covalently bonded nanomaterials from light-element composites are an interesting, challenging, and technologically important material system, which is not only important for basic research but also has the potential for industrial applications. Over the last decade, the physics/materials science community has been witnessing a flood of studies, which specifically address the novel formation mechanism as well as their physical and chemical properties of these nanomaterials. As elaborated earlier, our ever-increasing understanding of this class of systems has been greatly facilitated by the use of modern growth and analysis technologies and also by a comparison of theoretical calculations.
178
An overview of the kind given in this chapter–and also the attempt to identify general trends–requires omission of interesting details. Many questions could not be addressed, but some of them will be in other chapters in this book.
References
Audier, M., A. Oberlin, M. Oberlin, M. Coulon, and L. Bonnetain. Carbon. 19, 217 (1981)
Badding J. V. and D. C. Nesting. Chem. Mater.. 8, 535 (1996)
Badzian A. R., T. Niemyski, S. Appenheimer, and Olkusnik, E. Proc. of the 3rd Inter. Conf. on Chemical Vapor Deposition. American Nuclear Society, Hinsale. 3, 747 (1972)
Bai X. D., J. D. Guo, J. Yu, E. G. Wang, J. Yuan, and W. Zhuo. Appl. Phys. Lett.. (submitted, 2000)
Bendeddouche A., R. Berjoan, E. Beche, T. Merle-Mejean, S. Schamm, V. Serin, G. Tailades, and R. J. Hillel. Appl. Phys.. 81, 6147 (1997)
Blasé X., A. Rubio, S. G. Louie, and M. L. Cohen. Phys. Rev.. B51, 6868 (1995)
Blasé X., J.-C. -Charlier, A. De Vita, and R. Car. Appl. Phys. Lett.. 70, 197 (1997)
Bhusari D. M., C. K. Chen, K. H. Chen, T. J. Chuang, L. C. Chen, and M. C. Lin. J. Mater. Res.. 12, 322 (1997)
Bursill L. A., Julin Peng, V. N. Gurarie, A. V. Orlov and S. J. Prawer. Mater. Res.. 10, 2277 (1995)
Casanovas J., J. M. Ricart, J. Rubio, F. Illas and J. M. JimenezMateos. J. Am. Chem. Soc.. 118, 8071 (1996)
Chen Y., L. P. Guo, F. Chen and E. G. Wang. J. Phys. Condons. Matter. 8, L685 (1996a)
Chen Y., E. G. Wang and L. P. Guo. Chinese Patent. 961032299.5 (1996b)
Chen L. C., C. Y. Yang, D. M. Bhusari, Chen, K. H. M. C. Lin, J. C. Lin and T. J. Chuang. Diam. Rel. Mater.. 5, 514 (1996c)
Chen, Y., L. P. Guo and E. G. Wang. Phil. Mag. Lett.. 75, 155 (1997a)
Chen Y., L. P. Guo and E. G. Wang. J. Mater. Sci. Lett.. 16, 594 (1997b)
Chen Y., L. P. Guo and E. G. Wang. J. Cryst. Growth. 179, 515 (1997c)
Chen Y., F. Chen and E. G. Wang. J. Mater. Res.. 13, 126 (1998a)
Chen J., N. S. Xu E. G. Wang D. H. Chen and A. X. Wei. Chin. Phys. Lett.. 15, 539 (1998b)
Chen J., S. Z. Deng N. S. Xu K. H. Wu and E. G. Wang. Appl. Phys. Lett.. 75, 1323 (1999)
Chowdhury A. K. M. S., D. C. Cameron and J. M. Gregg. J. Mater. Res.. 14, 2359 (1999)
Cohen M. L.. Phys. Rev.. B32, 1988 (1985)
179
Cohen M. L.. Mater. Sci. Eng.. A105, 11 (1988)
Cohen M. L.. Science. 245, 84 (1989a)
Cohen M. L.. Nature. 338, 291 (1989b)
Cohen M. L.. Mater. Sci. Eng.. A209, 1 (1996)
Corkill J. L. and M. L. Cohen. Phys. Rev.. B48, 17622 (1993)
Egerton R. F.. EELS in the Electron Microscopy. Plenum, New York (1986), p. 262
Fahmy Y., T. D. Shen and C. C. Koch J. Mater. Res.. 14, 2488 (1999)
Filipozzi L., A. Derre, J. Conard Piraux, L. and Marchand, A. Carbon. 33, 1747 (1995)
Fujimoto F. and K. Ogata. Jpn. J. Appl. Phys.. 32, L420 (1993)
Gong Z., E. G. Wang G. C. Xu and Y. Chen. Thin Soild Films. 348, 114 (1999)
Guo L. P., Y. Chen, E. G. Wang Z. X. Zhao and L. Li. Crystal Growth. 178, 639 (1997a)
Guo L. P., Y. Chen, E. G. Wang Z. X. Zhao and L. Li. Chem. Phys. Lett.. 268, 26 (1997b)
Hamada N., S. Sawada, A. Oshiyama. Phys. Rev. Lett.. 68, 1579 (1992)
Han H. X. and B. J. Feldman. Solid. State. Commun. 65, 921 (1998)
He Z., G. Carter, J. S. Colligon. Thin Solid Films. 283, 90 (1996)
Hernandez E., C. Goze, P. Bernier and Rubio, A. Phys. Rev. Lett.. 80, 4502 (1998)
Himpsel F. J., J. A. Knapp J. A. Van Vechten and D. E. Eastman. Phys. Rev.. B15, 624 (1979)
Huang P., J. W. Xu Z. F. Ren J. H. Wang M. P. Siegal and P. N. Provencio. Appl. Phys. Lett.. 73, 3845 (1998)
Jagielski J., N. Monocoffre, G. Marest. J. Mater. Sci.. 34, 2949 (1999)
Kaplan. R. J. Appl. Phys.. 56, 1636 (1984)
Karcher R., L. Ley, R. L. Johnson. Phys. Rev.. B30, 1896 (1984)
Komateu S., Y. Hirohata, S. Fukuda, T. Hino. Yamashina, T. Hata and K. Kusakabe, K. Thin Solid Films. 193/194, 917 (1990)
Kuang M. H., Z. L. Wang X. D. Bai J. D. Guo and E. G. Wang. Appl. Phys. Lett.. (submitted, 2000)
Kubler L., J. L. Bischoff, D. Bolment. Phys. Rev.. B38, 13113 (1988)
Kumar N. and H. Schmidt. Solid State Technol.. 71, 73 (1995)
Kuttel O. M., O. Groening, C. Emmenegger and L. Schlapbach. Appl. Phys. Lett.. 73, 2113 (1998)
180
Lee K. R., K. Y. Eun, S. Lee and Jeon, D. R. Thin Solid Films. 290/291, 171 (1996)
Li Yin-an, Z. B. Zhang, S. S. Xie, and G. Z. Yang. Chem. Phys. Lett.. 247, 253 (1995)
Lin D. Y., C. F. Li Y. S. Huang Y. C. Jong Y. F. Chen C. K. Chen K. H. Chen and D. M. Bhusari. Phys. Rev.. B56, 6498 (1997)
Liu A. Y. and M. L. Cohen. Science. 245, 841 (1989)
Liu A. Y. and M. L. Cohen. Phys. Rev.. B41, 10727 (1990)
Liu A. Y. and R. M. Wentzcovitch, Phys. Rev.. B50, 10362 (1994)
Liu A. Y., R. M. Wentzcovitch M. L. Cohen, Phys. Rev.. B39, 1989 (1989)
Lowther. J. E.. Phys. Rev.. B60, 11943 (1999)
Ma M., T. Tsuru, T. Ogawa, Z. H. Mai C. Y. Wang J. D. Guo X. C. Ma and E. G. Wang. J. Phys.: Condens Matter. 11, L191 (1999a)
Ma X. C., E. G. Wang, W. Zhou, D. A. Jefferson, J. Chen, S. Deng and N. Xu: Appl. Phys. Lett.. 75, 3105 (1999b)
Marton D., K. J. Boyd, A. H. Al-Dayati, S. S. Todorov and J. W. Rabalais. Phys. Rev. Lett.. 73, 118 (1994)
May B. M., J. R. Dahn, T. Tiedje, K. Myrtle and Kasrai, M. Phys. Rev.. B46, 1697 (1992)
May P. W., S. Hohn, M. N. R. Ashfold, N. W. Wang, N. A. Fox, T. J. Davis and J. W. Steeds. J. Appl. Phys.. 84, 1618 (1998)
Miyamoto Y., A. Rubio, M. L. Cohen and S. G. Louie. Phys. Rev.. B50, 4976 (1994)
C. M. Niu, Y. Z. Lu, C. M. Lieber. Science. 261, 334 (1993)
Novak B. M. Adv. Mater.. 5, 422 (1993)
Novikov N., V. M. A. Voronkin. N. I. Zaika. Diam. Relat. Mater.. 1, 80 (1992)
Ogata K., J. F. D. Chubaci, F. Fujimoto. J. Appl. Phys.. 76, 3791 (1994)
Okada T., S. Yamada, Y. Takluchi and T. Wada J. J. Appl. Phys.. 78, 7416 (1995)
Okano K., S. Koizumi, R. P. Silva and A. J. Amaratunga. Nature. 381, 1678 (1996)
Peng J., Y. F. Zhang S. Z. Yang and G. H. Chen. Mater. Lett.. 27, 125 (1996)
Phillips J. C. Bonds and Bands in Semiconductors. Academic Press, New York (1973)
Redlich Ph., J. Loeffler, P. M. Ajayan, J. Bill, F. Aldinger and M. Rühle. Chem. Phys. Lett.. 260, 465 (1996)
Riedel R., A. Greiner, G. Miehe, W. Dressler, H. Fuess, J. Bill and F. Aldinger. Angew. Chem. Int. Ed. Engl.. 36, 603 (1997)
Ronning C., H. Feldermann, R. Merk, H. Hofsass, P. Reimke and J. U. Thiele. Phys. Rev.. B58, 2207 (1998)
181
Saito Y., K. Hamaguchi, K. Hata, K. Uchida, Y. Tasaka, F. Ikazaki, M. Yumura, A. Kasuya and Y. Nishina, Nature. 389, 554 (1997)
Sen R., B. C. Satishkumar, A. Govindaraj, K. R. Harikumar, G. Raina, J. P. Zhang, A. K. Cheetham and C. N. Rao. R. Chem. Phys. Lett.. 287, 671 (1998)
Silva S. R. P., G. A. J. Amaratunga, J. R. Barnes. Appl. Phys. Lett.. 71, 1477 (1997)
Sjöström H., I. Ivanov, M. Johasson, L. Hultman, J. E. Sundgren, S. V. Hainswith, T. F. Page, and L. R. Wallenberg. Thin Solid Film. 246, 103 (1994)
Sjöström H., M. Boman, S. Stafström and Sundgren, J. E. Phys. Rev. Lett.. 75, 1336 (1995)
Stepham O., P. M. Ajyan, C. Colliex, C. Redlich, Lambert. P. Bernier, J. M. and Lefin, P. Science. 266, 1683 (1994)
Suenaga K., C. Colliex, N. Demoncy, A. Loiseau, H. Pascard and F. Willaime, Science. 278, 653 (1997)
Sugino T., S. Kawasaki, K. Tanioka and Shirafuji, J. Appl. Phys. Lett.. 71, 2074 (1997)
Sung S. L., S. H. Tsai, H. C. Shih. Appl. Phys. Lett.. 74, 197 (1999)
Takahashi T., H. Itoh, A. Takeuchi. J. Crystal Growth. 47, 245 (1979)
Talin A. A., L. S. Pan K. F. McCarty and T. E. Felter. Appl. Phys. Lett.. 69, 3842 (1996)
Terrones M., A. M. Beneto, C. Manteca-Diego, W. K. Hsu, O. I. Osman, J. P. Hare, D. G. Reid, H. Terrones, A. K. Cheetham, K. Prassides, H. W. Kroto and D. R. Walton. M. Chem. Phys. Lett.. 257, 576 (1996)
Terrones M.. P. Redlich, D. R. M. Walton. Adv. Mater.. 11, 655 (1999)
Teter D. M. and R. J. Hemley. Science. 271, 53 (1996)
Thärigen T., G. Lippold, V. Riede, M. Lorenz, K. J. Koivusaari. D. Lorenz, S. Mosch, P. Grau, R. Hesse, D. Streubed and R. Szargan. Thin Solid Films. 348, 103 (1999)
Tsang S., C. Y. K. Chen, P. J. F. Harris and M. L. Green. Nature. 372, 159 (1994)
Uslu C., B. Park, D. B. Poker. J. Elec. Mater.. 25, 23 (1996)
Veprek S.. J. Vac. Sci. Technol.. A17, 2401 (1999)
Veprek S., J. Weidmann, F. Glatz. J. Vac. Sci. Technol.. A13, 2914 (1995)
Wagner C. D., D. E. Passoja H. F. Hillery T. G. Kiniski H. A. Six W. T. Jansen and J. A. Taylor. J. Vac. Sci. Technol.. 21, 933 (1982)
Walter A. Y. and R. Messier. Science. 247, 688 (1990)
Wang E. G.. Prog. Mater. Sci.. 41, 241 (1997)
Wang E. G.. Adv. Mater.. 11, 1129 (1999)
182
Wang E. G., Y. Chen and L. P. Guo. Phys. Scr.. T46, 108 (1997)
Watanabe M., O. S. Itoh, T. Sasaki and K. Mizushima. Phys. Rev. Lett.. 77, 187 (1996)
Way B. M., J. R. Dahn, T. Tiedje, K. Myrtle and M. Kasrai. Phys. Rev.. B46, 1697 (1992)
Weich F., J. Widany, Th. Frauenheim. Phys. Rev. Lett.. 78, 3326 (1997)
Weng-Sieh Z., K. Cherrey, G. Chopra, X. Blase, Y. Miyamoto, A. Rubio, M. L. Cohen, S. G. Louie, A. Zettl and R. Gronsky. Phys. Rev.. B51, 11229 (1995)
Werninghuas T., D. R. Zahn E. G. Wang and Y. Chen. Diam. Relat. Mater.. 7, 52 (1998)
Widany J., F. Weich, Th. Köhler, D. Porezag and Frauenheim, Th. Diam. Relat. Mater.. 5, 1031 (1996)
Wisitsora-at A., W. P. Kang, J. L. Davidson and D. V. Kerns. Appl. Phys. Lett.. 71, 3394 (1997)
Wu D., D. Fu, H. Guo, Z. Zhang, X. Meng and X. Fan, Phys. Rev.. B56, 4949 (1997)
Wu K., E. G. Wang, J. Qing and G. C. Xu. J. Appl. Phys.. 83, 1702 (1998)
Wu K., E. G. Wang, J. Chen and N. S. Xu. J. Vac. Sci. Technol.. B17, 1059 (1999)
Wu K., E. G. Wang, Z. L. Wang. Appl. Phys. Lett.. (submitted, 2000)
Yao H. and W. Y. Ching. Phys. Rev.. B50, 11231 (1994)
Yen T. Y. and C. P. Chou. Appl. Phys. Lett.. 67, 2801 (1995a)
Yen T. Y. and C. P. Chou. Solid State Commun. 95, 281 (1995b)
Yu K., M. M. L. Cohen, E. E. Haller, W. L. Wansen, A. Y. Liu and I. C. Wu. Phys. Rev.. B49, 5034 (1994)
Yu J. and E. G. Wang. Appl. Phys. Lett.. 74, 2948 (1999)
Yu J., E. G. Wang, G. C. Xu. Chem. Phys. Lett.. 292, 531 (1998)
Yu J., E. G. Wang, G. C. Xu. J. Mater. Res.. 14, 1137 (1999)
Yu J., E. G. Wang, J. Ahn, S. F. Yoon, Q. Zhang, J. Cui and M. B. Yu. J. Appl. Phys.. (submitted, 2000)
Zhang W., K. Zhang, B. Wang. Mater. Sci. Eng.. B26, 133 (1994)
Zhang Y., H. Gu, K. Suenaga and Iijima, S. Chem. Phys. Lett.. 279, 264 (1997)
Zhang Z. J., S. Fan, C. M. Lieber. Appl. Phys. Lett.. 66, 3582 (1995)
Zhang Z. J., J. L. Huang S. S. Fan and C. M. Lieber. Mater. Sci. Eng.. A209, 5 (1996)
Zheng W. T., H. Sjöström, I. Ivanov, K. Z. Xing, E. Broitman, W. R. Salaneck. J. E. Greene and J. E. Sundgren. J. Vac. Sci. Technol.. A14, 2696 (1996)
Zhou W., R. D. Tilley D. A. Jefferson, Ma, X. C. and E. G. Wang. Adv. Mater.. submitted, (2000)
183
，15 Self Assembled Ordered Nanostructures
15.1 Ordered Self-Assembled Nanocrystals
Self-assembly nanostructure is a new form of material with fundamental interest and potential technological applications (Brus, 1991; Wang, et al., 1991; Fendler, et al., 1995; Alivisatos, 1996a, 1996b; Pileni, 1997; Wang, 1998a; Collier, et al., 1998; Brinker, et al., 1999; Ying, et al., 1999). The rapid expansion of research work in this area is driven by the need to further miniaturize electronic components. Another reason is that, in nanoscale, materials properties are quite different from bulks and they are strongly size-dependent. It is possible to process materials which can be tuned via size control to achieve specific functionality. In nanocrystals, the effect from the surface is comparable in some cases to the chemical composition in influencing the chemical, electronic, magnetic and optical behaviors.
In nanocrystal self-assembled structures, each individual nanocrystal is the fundamental building unit and it serves as an "artificial atom" for constructing the ordered structure. The analogy of a nanocrystal with an atom is obvious. If the size of nanocrystals is below 10 nm, the energy levels of each individual nanocrystal can be discrete, as in the case of atoms. The energy level spacing and other "atomic" properties can be adjusted by changing the size of the nanocrystals, in contrast to those of the atoms. Size and shape controlled nanocrystals can be viewed as molecular matter with specific shape and electronic structure, and a self-assembly of the nanocrystals can form "nanocrystal solids" with translation and even orientation order. Based on this idea, several kinds (semiconductors, metals, and oxides) of nanocrystal molecules have been successfully synthesized recently, including CdSe (Murray, et al., 1993, 1995; Braun, et al., 1996; Shenton, et al., 1997), InP (Guzelian, et al., 1996), CdS (Hu, et al., 1998), Au (Whetten, et al., 1996; Andres, et al., 1996; Kiely, et al., 1998), Ag (Harfenist, et al., 1996, 1997; Collier, et al., 1997), Pt (Provencio, et al., 1998), Co (Sun, et al., 1999; Yin and Wang, 1999; Petit, et al., 1998), Ni (Sun, et al., 1999), TiO2 (Moritz, et al., 1997), CoO (Yin and Wang, 1997a, b, 1999a), and Fe2O3 (Bentzon, et al., 1989), etc..
At nanoscale, not only the physical properties, but also the chemical properties of materials have been profoundly changed. The huge surface area of each bare nanocrystal suggests that nanocrystals are very reactive, thus, surfactant molecules are needed to cap and stabilize the surfaces of nanocrystals. In the evaporation processing method, for instance, if no surfactant is applied, the freshly made gold nanocrystals coalescence with one another to form heavily twinned larger nanocrystals, but if the surfactant molecules are applied into the reaction chamber, gold nanocrystals can keep the size and shape. After an inorganic nanocrystal is coated with a densely packed monolayer of surfactant molecules, the surface of nanocrystals becomes hydrophobic, and this kind of nanocrystal-surfactant combination is soluble in non-polar solvents and suspend in the solution, forming stable colloids. After the solvent is evaporated or removed, the passivated nanocrystals rearrange themselves to form assemblies instead of fusing together, because they are separated by a thin layer of molecules. Due to the existence of the passivation molecules, the aggregates can be re-dissolved in suitable solutions. If the size of nanocrystals is monodispersive, ordered self-assembly of nanocrystals will be formed, just like the crystallization from solution. The schematic illustration of a typical monolayered nanocrystal self-assembly is shown in Fig. 15.1. The ordered self-assembled nanocrystals of silver are shown in Fig. 15.2.
184
Figure 15.1 Schematic illustration of the self-assembled nanocrystals. To form ordered self-assembly, several conditions need to be met: the nanocrystals should be monodispersive; the surfactant molecules should be strong enough to separate the individual nanocrystals; and the drying rate should be so slow that the nanocrystals can move to the suitable positions. The above plot is 2-D assembly of faceted nanocrystals, and the bottom one is assembly of non-faceted nanocrystals.
Figure 15.2 Ordered self-assemblies of silver nanocryastals. The nanocrystals are processed by a vertical aerosol method and later size-selected.
185
In summary, due to the capping monolayer of molecules, nanocrystals are held together mainly by van der Waals forces. Three requirements need to be met to make ordered self-assembly of nanocrystals: (1) building blocks (monodispersive nanocrystals); (2) passivation layers (suitable surfactants and solutions); and (3) a slow drying rate to allow diffusion of the passivated nanocrystals in solution to find their equilibrium position. If the drying rate is too high, the disordered arrangement of nanocrystal self-assembly will form, just like a rapid quenching can lead to amorphous phase during solidification. In the following sections, we will first discuss the processing methods, then the characterization method, and the uniqueness of self-assembled nanostructures. Finally, the future directions are also discussed.
15.1.1 Processing of Nanocrystals for Self-Assembly
The controlled fabrication of very small structures at scales beyond the current limits of lithographic techniques is a technological goal of great practical and fundamental interest. Important progress has been made over the past few years in the preparation of ordered ensembles of metal and semiconductor nanocrystals. Processing of nanocrystals can be dated back to Faraday's time, though at that time gold nanocrystals could not been physically observed. Many methods have been developed to process different kinds of nanocrystals, single phase or compounds. Here only the methods that can produce or have the potential to produce monodispersive nanocrystals or quasi monodispersive nanocrystals will be discussed.
15.1.1.1 Metallic Nanocrystals (Au, Ag, Co, Ni, Pt, etc.)
The processing methods of metal nanocrystals can be roughly divided into two groups: chemical methods and physical methods. The reduction of metal ions in solution is the most popular and economical one. Though evaporation of metals at high temperature can also yield metal nanoclusters, this method is not suitable for compounds because of potential problems such as thermal decomposition and possible oxidation.
Gold nanocrystals can be prepared either by an aerosol method or the chemistry method. In the chemistry route, Farady (1857) first used the two-phase method to prepare stable colloidal metal nanocrystals, in which he reduced an aqueous gold salt with phosphorus in carbon disulfide and obtained a ruby colored aqueous solution of dispersed gold nanoparticles. In a typical process (Whetten, et al., 1996), the gold nanoparticles growing from metal ions AuCl-
4 are reduced at the oil-water interface in the presence of an alkylthiolate surfactant (SR, where R=n-CnH2n+1, n=4, 6, 8, 12,...) and a reducing agent, sodium borohydride. The strategy used here is to finish processing of gold nanocrystals and attaching surfactant molecules in one step. By an extended exposure to excess reducing agent defective structures initially formed can be etched away.
The aerosol method can also be used to process monodispersive metal nanocrystals. From the flow direction of the aerosol in the furnace, the setup can have two different types: one is horizontal (Andres, et al., 1996), the other is vertical (Whetten, et al., 1996). Gold atoms are evaporated first from a carbon crucible in a resistively heated carbon tube, which are entrained in He and induced to condense into nanoclusters by mixing the hot flow from the oven with a room temperature stream of helium. Controlling conditions in the oven and the flow downstream from the oven controls the mean cluster size. The clusters are molten and
186
recrystallized while still in the gas phase. They are scrubbed from the gas phase by contact with a mist of organic solvent containing 1-dodecanethiol and collected as a stable colloidal suspension. In order to ensure that all of the clusters are single crystalline, a dilute aerosol stream of clusters suspended in inert gas is passed through 1 m long tube in which the clusters are first heated above their melting temperature and then cooled to room temperature. Treating small gold clusters in this way transforms them into face-centered-cubic (f.c.c.) single crystals.
Platinum nanocrystals can also be processed by reduction of platinum ions with the presence of capping materials (Ahmadi, et al., 1996). For example, the shapes and sizes of platinum nanoparticles are controlled by changes in the ratio of the concentration of the capping polymer material (sodium polyacrylate) to the concentration of the platinum ions (from K2PtCl4) used in the reductive synthesis of colloidal particles in solution at room temperature. In this method, hydrogen gas is bubbled at a high flow rate through the solution. Tetrahedral, cubic, irregular-prismatic, icosahedral, and cubo-octahedral particle shapes were observed (Wang, et al., 1997), whose distribution was dependent on the concentration ratio of the capping polymer material to the platinum cation.
The magnetic nanocrystals like iron, cobalt, and nickel can be processed by the decomposition of metal carbonyl in organic solution. It was noticed very early that thermal decomposition of metal carbonyl in organic solution (with surfactant) often led to metal nanocrystals with a very narrow size distribution (Thomas, et al., 1966; Papiper, et al., 1983). Thermal decomposition of cobalt carbonyl in different kinds of solution (toluene, xylene, etc.) with different kinds of long C-H chains and strong ionic group (sulfonate) surfactant have been systematically studied. Ordered magnetic nanocrystal self-assembly was processed (Sun, et al., 1999) using the traditional reverse micelle technique. Choosing the cationic surfactant, didodecryld ammonium bromide (DDAB) with toluene as a binary system, DDAB as the reducing agent, the Co2+ from CoCl2 can be reduced to form cobalt nanocrystals with the assistance of PR3 (R=n-C4H9, n-C8H17) as the capping material.
15.1.1.2 Semiconductor Nanocrystals (CdSe, CdTe, etc.)
From the technological point of view, semiconductor nanocrystals are potentially important and useful. The band structure of semiconductor nanocrystals is quite different from that of the bulk material (Fig. 15.3). Nanocrystals have discrete excited electronic states and an increased band-gap in comparison to the bulk semiconductor materials. The smaller is the size, the bigger is the difference. The difference can be easily differentiated by optical absorption spectroscopy. But for indirect bandgap semiconductors like silicon and rock-salt CdSe, the optical spectra are continuous, though the individual valence and conduction band eigenstates are discrete. The bandgap is tunable via controlling nanocrystal size, providing an effective way of adjusting the electronic structure in addition to controlling particle chemistry.
Figure 15.3 Schematic comparison of the band structure and electronic states of semiconductor nanocrystals and bulk semiconductor (Brus, 1991).
187
The most typical example is silicon nanocrystals. The dynamics and spectroscopy of silicon nanocrystals that emit at visible wavelengths were analyzed by Wilson (Wilson, et al., 1993). Size-selective precipitation and size-exclusion chromatography clearly separate the silicon nanocrystals from larger crystallites and aggregates and provide direct evidence for quantum confinement in luminescence. Measured quantum yields are as high as 50% at low temperature, principally as a result of efficient oxide passivation. Despite a 0.9 eV shift of the bandgap to higher energy, the nanocrystals behave fundamentally as indirect bandgap materials with low oscillator strength.
Due to the oxidation on the surface of silicon nanocrystals, more effort as concentrated on the semiconductor compound (VI-II and V-III type) nanocrystals. In 1988, Brus's group at Bell Lab first reported a process for synthesis of pure and stable organic capped CdSe nanocrystals using an inverse micelle method (Steigerwald, et al., 1988). And up to now, CdSe is still the most intensively studied semiconductor nanocrystals (Nirmal, et al., 1996; Bruchez, et al., 1998; Tomaselli, et al., 1999). By separating the nucleation stage from the growth stage and the following precipitation process, the CdSe nanocrystals are nearly monodispersive. Semiconductor nanocrystals have discrete excited electronic states and an increased band gap in comparison with bulk materials (Fig. 15.3). The size-dependent emission peaks collected from several monodispersive semiconductor nanocrystals are shown in Fig. 15.4. For the indirect bandgap semiconductors such as silicon and rock-salt CdSe nanocrystals the optical absorption spectrum is continuous, even though the bandgap increases and the conduction band eigenstates are discrete. This is because many overlapping discrete transitions appear to be present with roughly equal intensity via electron-phonon interaction.
Figure 15.4 (a) The emission spectra of several surfactant-coated monodispersive semiconductor nanocrystals. The solid lines series represents different sizes of CdSe nanocrystals with diameters of 2.1, 2.4, 3.1, 3.6 and
188
4.6 nm (from right to left). The dash lines series is of InP nanocrystals with diameters of 3.0, 3.5 and 4.6 nm. The dotted lines series is of InAs nanocrystals with diameters of 2.8, 3.6, 4.6 and 6.0 nm. (b) A true-color image of a series of silica-coated core (CdSe)-shell (ZnS or CdS) nanocrystal probes in aqueous buffer, all illuminated simultaneously with a handheld ultraviolet lamp (Bruchez, et al., 1998).
The properties of CdSe nanocrystal dispersions have been studied by small-angle X-ray scattering method (Mattoussi, et al., 1998). The study can provide accurate measures of the nanocrystal size and size distribution. The low polydispersity measured confirmed the high quality of the nanocrystals prepared using a high-temperature solution chemistry route, as anticipated from optical data. The study also provides information on the interparticle interactions and their dependence on a few relevant parameters, such as nature of the capping molecules and solvent. Nanocrystal association, e.g., dimers, in dispersions characterized by weakly attractive interactions, and/or aggregation in solutions with strong attractions, can be observed. The study also unveiled other features where interactions are reversed from repulsive stabilizing to attractive as the particle size is decreased. This behavior, unexpected for colloidal dispersions, may be caused by a reduction of the cap density as the size is decreased. The general trend for the interparticle interactions in these dispersions can be understood within the framework of a van der Waals core-to-core attractive potential, to which are superposed effects of cap affinity to the core and to the surrounding solvent. Within these considerations, one can distinguish three types of dispersions: sterically stabilized dispersions, dispersions thermodynamically stable but governed by weak attractions, and unstable dispersions where strong attractions induce macroscopic aggregation.
The self-organization of CdSe nanocrystallites into three-dimensional semiconductor quantum dot superlattices (colloidal crystals) has been demonstrated. The size and spacing of the dots within the superlattice are controlled with near atomic precision. This control is a result of synthetic advances that produce CdSe nanocrystallites exhibiting monodisperse within the limit of atomic roughness. The methodology is not limited to semiconductor quantum dots but provides general procedures for the preparation and characterization of ordered structures of nanocrystallites from a variety of materials. It was shown spectroscopically that electronic energy transfer in close-packed CdSe quantum-dot (QD) solids
189
arises from dipole-dipole interdot interactions between proximal dots. They used time-resolved photoluminescence to study electronic energy transfer in optically thin and clear, close-packed QD solids prepared from CdSe QD samples tunable from 1.7 nm to 15 nm in diameter (σ<4.5%). In mixed QD solids of small and large dots, they measured quenching of the luminescence (lifetime) of the small dots accompanied by enhancement of the luminescence (lifetime) of the large dots, consistent with electronic energy transfer from the small to the large dots. In QD solids of single size dots, a red-shifted and modified emission line shape is consistent with electronic energy transfer within the sample inhomogeneous distribution.
Electronic energy transfer between close-packed quantum dots using time resolved photoluminescence has also been demonstrated (Kagan, et al., 1996). Optically clear and thin, close-packed quantum dot solids were prepared from mixtures of small and large CdSe quantum dots (3.85 and 6.2 nm, σ<4.5%). Quenching of the luminescence (lifetime) of the small dots accompanied by enhancement of the luminescence (lifetime) of the large dots is consistent with long-range resonance transfer of electronic excitations from the more electronically confined states of the small dots to the higher excited states of the large dots.
15.1.1.3 Oxides Nanocrystals (CoO, α-Fe2O3, TiO2, etc.)
Self-assembling of α-Fe2O3 was first achieved by accident. After the ferrofluid of iron nanocrystals was exposed to air for about one month (processed by thermal decomposition of iron carbonyl in the mixture of decalin with oleic acid as the surfactant), it was found that the iron nanocrystals have been transformed into iron oxide. After drying the colloidal solution, a hexagonal close packing of hematite (α-Fe2O3) nanoparticles (antiferromagnet) formed on the carbon substrate (Bentzon, et al., 1989). The nanocrystals had a very narrow size distribution (in the case of mean particle size 6.9 nm, the standard deviation is 0.4 nm). Dispersive ε-Fe3N fine particles synthesized by a vapor-liquid chemical reaction between Fe(CO)5 and ammonia have shown a narrow size distribution and they can form nicely locally ordered monolayer array.
A simple method to control the growth of TiO2 nanocrystallites and the formation of nanostructured TiO2-based materials has also been developed. The method used to form these materials is based on controlling the hydrolysis and polycondensation of titanium alkoxide using organic ligands in order to build and stabilize intermediate building units (slabs). Anatase structured TiO2 particles with different sizes and shapes are obtained simply by changing the titanium/cation ratio. The small clusters agglomerate into condensed "snow-ball" structure, which in turn self-assemble into superlattices.
15.1.2 Technical Aspects of Self-Assembling
15.1.2.1 Selection of Monodispersive Nanocrystals
Whatever processing methods used, the nanocrystal size has a broad distribution even if the processing conditions can be strictly controlled. So the size selection is very important because sometimes the ideal conditions cannot be obtained. Lyophobic colloidal nanocrystals attract each other via the van der Waals force. The attraction is strong because of the near linear additivity of forces between pairs of unit cells in
190
different nanoparticles. The efficiency of the steric stabilization is strongly dependent on the interaction of the alkyl groups with the solvent. Gradual addition of the nonsolvent can produce size-dependent flocculation of the nanocrystal dispersion. The effect of size selection on CdSe nanocrystals is illustrated in Fig. 15.5 (Murray, et al., 1993), in which from (a) to (d) the peak becomes sharper. After the size selection, it is obvious that the color of the solutions of different sized semiconductor nanocrystals is different.
Figure 15.5 An example of size selection of semiconductor ( 3.7 nm CdSe) nanocrystals. (a) Optical absorption spectrum shows a very broad peak before selection. (b) Spectra after one size selective precipitation from the solution with methanol. (c) Spectrum after dispersion in 1-butanol. (d) Spectrum of the final selected nanocrystals. (Murray, et al., 1996).
Several solvent/nonsolvent systems can be used to make size selection, for example, hexane/ethanol, chloroform/methanol, pyridine/hexane, etc.. The addition of the nonsolvent increases the average polarity of the solvent and reduces the energy barrier to flocculation. Large nanoparticles have a higher probability of overcoming the reduced energy barrier and precipitate. With more nonsolvent added, the size distribution becomes narrower and narrower. After the addition of methanol, the light absorption peak becomes sharper and the position shifts from 530 nm to 400 nm. From quantum theory, the smaller the nanocrystals, the
191
wider the energy gap compared to that of the bulk. The shift of the peak means that more and more larger sized nanocrystals have precipitated from the colloidal solution. In addition to CdSe semiconductor nanocrystals, this method has also been successfully applied to gold nanocrystal separation. One point needs to be addressed, that this method can only be applied to nanoparticles with relatively small polydispersity.
Photocorrosion is also a very useful method to narrow the size distribution of some nanocrystal systems. For example, polydispersive CdS nanocrystals can be selected to have an average diameter of 4.2 nm with standard deviation was 1.9 nm by a sequential irradiation with a monochromatic light whose wavelength was changed step by step from 490 nm to 430 nm in air-saturated sodium hexametaphosphate solution. Analysis of the amount of sulfate ions produced by photocorrosion of quantum-CdS colloids revealed that the number of CdS particles in the colloid decreased with promotion of photocorrosion, suggesting that during the course of photocorrosion process photocorroded CdS particles were agglomerated to give larger particles which were further photocorroded. The molar absorption coefficient of CdS particles at the first exciton peak was found to be independent of the particle size (Matsumoto, et al., 1996).
15.1.2.2 Controlling the Formation of Ordered Self-Assembly
Though 2-D ordered self-assembling of nanocrystals can be processed by choosing different processing parameters, in most cases, the formation of self-assembly is basically random. How the self-assembly forms at a specific area or position on the substrate is a very critical problem for applications. It is the key to search techniques that can guide and design the required nanostructures. Self-assembly of silver nanocrystals forming a ring shape with micron scale diameter is an example, which shows the possibility of controlling self-assembling (Ohara, et al., 1997). In the schematic illustration of the process of forming self-assembling nanocrystal rings (Fig. 15.6), the process is comparable to the ordinary bubble nucleation in a superheated liquid. The only difference is that the nanocrystal ring is open in the vertical direction, but the bubble in the liquid is closed in all directions. The 2-D self-assembled nanocrystal wires with width 20 nm to 300 nm was fabricated (Chung, et al., 1998). These wires with a narrow distribution of width (15%—25%) were processed by a Langmuir compression technique.
Figure 15.6 An example of self-assembled silver nanocrystal ring formed by an air-bulb assisted method (Ohara, et al., 1997).
192
Light-directed nanocrystal self-assembly is also an effective method to control the position of self-assembly of nanocrystals (Vossmeyer, et al., 1997). Using the light sensitive substrate and suitable mask, different kinds of self-assembling patterns can be formed on a substrate. Because the surface-bound amino group can be derivatized to give a thiol therminus, the light directed assembly of CdS, CdSe and other semiconductors or metal nanocrystals is possible.
15.1.2.3 Assembling of Nanoparticles with Mixed Sizes/Phases
Can the mixture of nanocrystals with different sizes form self-assembling packing? Several recent papers addressed this question (Ohara, et al., 1995; Kiely, et al., 1998). First take the simplest situation: the mixture of two monodispersive nanocrystals (A and B) with particle sizes RA and RB. On micron-scale, it was
193
reported that when 0.482 < RA/RB < 0.624, the mixture can form stable structure, and when RA/RB 0.58, stable AB2 structure will be formed. For 0.458 < RA/RB<0.482, phase separation occurs. Can these theories be applied to self-assembling of nanocrystals? The self-assembly of nanoscale bimodal packing was observed (Fig. 15.7), which agrees well with the micrometric scale colloidal crystals. So if the size ratio is in a certain range, superlattices can be formed in "binary" systems. Heath et al. also used polydispersive gold nanocrystals to examine the mechanism of self-assembling (Ohara, et al., 1995). It was found that self-assembling is an entropy-driven crystallization process.
Figure 15.7 A monolayer ordered superlattice of thiol-stabilized gold nanocrystals with two distinctive sizes and the particle diameter ratio about 0.58. (Kiely, et al., 1998)
15.1.3 Structure of the Nanocrystal Self-Assembly
Periodic packing of nanocrystals is different from 3-D packing of atoms in several aspects. First, to an excellent approximation atoms are spherical, while nanoparticles can be faceted polyhedra; thus, the 3-D packing of particles can be critically affected by their shapes and sizes. Secondly, the sizes of atoms are
194
fixed, but the sizes of nanoparticles can have a slight variation although their size distribution is very narrow. Finally, atomic bonding is due to the outer shell electrons via ionic, covalent, metallic bonding or the mixture, and in most cases the interatomic distance is fixed, while the bonding between nanoparticles is generated by the passivating thiolate surfactant whose length is controllable; thus, the ratio of particle size to interparticle distance is adjustable. This is a parameter that is likely to determine the 3-D packing of the nanoparticles.
There are many characterization methods can be used to measure the properties of the self-assembly of nanocrystals. After the formation of self-assembling, X-ray diffraction (XRD) and transmission electron microscopy are the most common tools for structure analysis. In the stable colloidal solution, UV light absorption can be used to determine the polydispersity of the nanocrystals dispersed in the solution. This method applies to metal nanocrystals and semiconductor nanocrystals. The interconnecting between the nanocrystals in the self-assembly can be analyzed by impedance spectroscopy.
15.1.3.1 Small-Angle X-ray Diffraction
X-ray diffraction is the first choice to examine the formation of crystalline assembling, provided the amount of the sample is sufficient. The diffraction spectrum at the high-angle range is directly related to the atomic structure of the nanocrystals, while the spectrum at the small angle region is directly associated with the ordered assembling of nanocrystals (Whetten, et al., 1996; Murray, et al., 1995). By examining the diffraction peaks that are extinct in the spectrum one may identify the crystallography of the packing. This analysis is based on an assumption that each particle is identical in size, shape and even orientation (i.e., the same X-ray scattering factor), so that the extinction rules derived from diffraction physics apply. In practice, however, a fluctuation in the size, orientation or shape can easily abolish this assumption. This is the reason that a quantitative analysis of the low-angle diffraction spectrum is rather difficult. Nevertheless, X-ray diffraction is still the most powerful technique to evaluate the average interparticle distance, and it is a unique technique to study the in situ pressure and/or temperature induced phase transformation in nanocrystals (Alivisatos, et al., 1996).
X-ray diffraction is a powerful tool for refining the structure of nanoclusters, particularly those smaller than 2 nm. If the particles are oriented randomly so that the entire assembly can be treated as a "polycrystalline" specimen composed of nanocrystals with identical structure but random orientations, their scattering from each can be treated independently. The structure of nanocrystals can be refined by quantitative comparison of the theoretically calculated diffraction spectra for different nanocrystal models with the experimentally observed ones. This has been shown recently by Cleveland et al. (1997a, 1997b) in determination of decahedral Au nanoclusters with sizes of 1.7–1.9 nm.
Mass spectroscopy is also very powerful for detecting the critical sizes of the smallest nanoclusters that can be synthesized and the magic number of atoms comprising the clusters (Alvarez, et al., 1997). This type of analysis is unique for clusters in which the atoms have not formed a well-defined crystal lattice, prohibiting the access of X-ray diffraction and HRTEM.
15.1.3.2 Crystallography and Orientation Ordering Determined by TEM
195
Transmission electron microscopy is an ideal choice in determination of the symmetry and orientation ordering of the self-assembled nanostructures (Wang, 1998a, 2000). It is particularly useful if the ordering is short range and/or the amount of the specimen is rather small. Studies of atomic structures of nanoparticles have been described in the chapter concerning transmission electron microscopy (TEM). Here we mainly concentrate on the analysis of nanocrystal assemblies.
Self-assembly of nanocrystals can be 2-D or 3-D. Figure 15.8(a) shows a transmission electron microscopy (TEM) image of monolayer assembling of Ag nanocrystals on a carbon substrate, where the 2-D superlattice is apparent. The Ag particles have sizes of (4±0.5) nm and their shapes are dominated by tetrahedral (see the projected triangle shapes in Fig. 15.8(b)) (Wang, et al., 1998b), a four-face polyhedron enclosed by 111 facets.
Figure 15.8 Monolayer self-assembly of Ag nanocrystals whose shape is dominated by tetrahedral. The inset is an HRTEM image showing the [110] projection of the tetrahedral particles.
Three-dimensional assembling of nanocrystals can form large size bulk crystalline materials. Figure 15.9(a) is a TEM image recorded from an Ag NCS deposited on a carbon substrate. The Ag nanocrystals have a truncated octahedral shape and they are oriented along the [110] of the Ag atomic lattice in the image, along which four 111 and two 100 facets are imaged edge-on. The unit cell of the NCS is also oriented along [110]s of fcc. Therefore, the orientational relationship between the Ag particles and the nanocrystal lattice is
196
[110] [110]s and [001] . Accordingly, a structure model for the NCS is built (Fig. 15.9(b)), in which the nanocrystals are oriented following an assembling principle of face-to-face (Harfenist, et al., 1996).
Figure 15.9 (a) [110]s TEM image of a 3-D fcc assembled Ag nanocrystal superlattice. The nanocrystals are turncated octahedra and they have orientational symmetry. (b) The 3-D fcc self-assembling model of the NCS.
197
15.1.3.3 Imaging the Surface Adsorbed Molecules
If the nanocrystals can be taken as the building blocks, their 3-D assembling is unavoidably affected by the particle shape. For the case shown in Fig. 15.9, one may wonder if the surface adsorbed molecules are distributed on the particle surfaces, edges or corners. Here we illustrate the observation of directional, interparticle molecular bonds formed by the thiolates.
An examination of the image shown in Fig. 15.9(a) indicates there are some white spots in the image, corresponding to open channels formed by the thiolate molecules, suggesting that the thiolates are tethered on the faces of the nanocrystals and they are likely to be erected on the surface. The shortest distance between the face-to-face 100 facets of the two adjacent particles is only 1.5 nm to 2 nm, almost equal to the 1.5 nm chain length of the thiolate molecules used for passivating the Ag nanocrystals. Therefore, the thiolate molecules tethered on the facets of the nanocrystals are likely to interpenetrate, forming the interdigitative bonds. This model is supported by the image contrast displayed in Fig. 15.9(a). A direct observation of the thiolates can be provided by the energy-filtered TEM (EF-TEM) (Reimer, 1995).
The EF-TEM relies on the principle of electron energy loss spectroscopy (EELS) and the images (or diffraction patterns) are formed by electrons with specific energy losses. If the electrons which have excited the carbon K ionization edge are selected for forming the image, the image contrast is approximately proportional to the thickness projected carbon atoms in the specimen, providing a direct chemical map of carbon. The thiolate molecules are composed of mainly carbon, thus the EF-TEM of the carbon K edge can give the distribution of the thiolates around the nanocrystals. For this analysis, Ag NCSs are deposited on an amorphous SiOx substrate and the effects from the substrate can be removed by processing the experimental images acquired pre- and post-edge.
The EF-TEM was performed for the Ag NCS oriented along [110]s, which is the optimum orientation for imaging thiolate distribution between the particles (Wang, et al., 1998c). The EF-TEM image acquired using the carbon K edge from a Ag NCS gives an interesting contrast feature (Fig. 15.10(a)). The projected carbon density between the particles shows a contrast pattern that is the strongest between the A and B types of particles, while the contrast is lower between the A and C or B and C types of particles. To interpret this phenomenon, we first construct the [110]s projection of the NCS based on the 3-D model given in Fig. 15.9(b), and the result is shown in Fig. 15.10(b). From the structural point of view, the molecular bonds tend to align parallell on the facets on which they are tethered. For the nanocrystals A and B assembled by facing the 100 faces, in addition to the carbon density contributed by the interdigitated thiolates passivated on the 100 facets (which are edge-on while viewed along [110]s), the thiolates passivated on the four 111 planes (not edge-on) also contribute to the projected carbon density although the 111 faces are at an angle with the projection direction. Therefore, the projected density of the thiolate molecules between particles A and B is expected to be higher than that between A and C (or B and C) if the size of 111 faces is the same as the that of 100 as well as the density of the thiolate passivation is the same on both 111 and 100. With consideration of the resolution of the EF-TEM of 2 nm, the channels formed by the bundled thiolates may not be resolved in this type of image.
Figure 15.10 (a) Energy filtered TEM image of the fcc structured Ag NCS recorded using the electrons after exciting the carbon K ionization edge, showing the distribution of the thiolate molecules around the
198
nanocrystals. The orientation of the NCS is [110]s and the 3-D self-assembling model is given in Fig. 15.9(b). (b) A model of thiolate distribution on the surfaces of Ag nanocrystals, which supports the interdigitative molecular bonding in the NCS.
15.1.4 Properties of the Nanocrystal Self-Assembly
199
As discussed above, self-assembling of nanocrystals can be ordered if the processing conditions are optimized. Due to the unique structure of self-assembling, they are expected to have many interesting properties different from either individual nanocrystals or the bulk materials. In recent years, there have been many research groups trying to explore their special properties. The big obstacle to making full use of the self-assembled nanocrystals is making a defect-free self-assembly. A recent report sheds some light on solving this problem.
Teramac, a massively parallel experimental computer, was built at Hewlett-Packard Laboratories to investigate a wide range of different computational architectures (Heath, et al., 1998). This machine contains about 22000 hardware defects, any one of which could prove fatal to a conventional computer, and yet it operated 100 times faster than a high-end single-processor workstation for some of its configurations. The defect-tolerant architecture of Teramac, which incorporates a high communication bandwidth that enables it to easily route around defects, has significant implications for any future nanometer-scale computational paradigm. It may be feasible to chemically synthesize individual electronic components with less than a 100% yield, assemble them into systems with appreciable uncertainty in their connectivity, and still create a powerful and reliable data communications network. Future nanoscale computers may consist of extremely large-configuration memories that are programmed for specific tasks by a tutor that locates and tags the defects in the system.
Heterostructured diode at nanoscale is also an important application. Metal/self-assembled monolayer heterostructured diodes connected by conjugated molecular wire had been made (Zhou, et al., 1997). Electronic transport measurement showed a distinctive rectifying behavior from the asymmetry of molecular heterostructure. If these conductive conjugated molecular wires can be used to link metal nanocrystals, the dimension of one diode can be decreased below 10 nm. To make such devices successful, it is crucial to understand the bonding between metal nanocrystals and the surfactant molecules, the charging transfer process in and between metal nanocrystals and the relative distance between the nanocrystals.
In 1995, close-packed planar arrays of nanometer-diameter metal clusters that are covalently linked to each other by rigid, double-ended organic molecules were successfully self-assembled (Andres, et al., 1996). Gold nanocrystals, each encapsulated by a monolayer of alkyl thiol molecules, were cast from a colloidal solution onto a flat substrate to form a close-packed cluster monolayer. Organic interconnects (aryl dithiols or aryl di-isonitriles) displaced the alkyl thiol molecules and covalently linked adjacent clusters in the monolayer to form a 2-D superlattice of metal quantum dots coupled by uniform tunnel junctions. Electrical conductance through such a superlattice of 3.7 nm-diameter gold clusters, deposited on a SiO2 substrate in the gap between two gold contacts and linked by an aryl di-isonitrile [1, 4-di(4-isocyanophenylethynyl)-2-ethylbenzene], exhibited nonlinear Coulomb charging behavior. The electrical conductance through gold clusters interconnected by aryl dithiol and aryl di-isonitrile molecules has been measured by a scanning tunneling microscope (STM), which was used to measure the current-voltage characteristics of a bare gold cluster deposited on a dithiol SAM grown on a flat Au(111) surface (Andres, et al., 1996). The results are in good agreement with semiclassical predictions. Clusters with diameters <2 nm exhibited "Coulomb staircase" behavior in conductance at room temperature.
Other significant progress was made in the property measurement by light absorption and impedance spectroscopy in the ordered self-assembled silver monolayer thin film (Collier, et al., 1997; Markovich, et al., 1998). In situ measurements were conducted on both linear and nonlinear optical properties of
200
organically functionalized silver nanocrystal Langmuir monolayers as a continuous function of interparticle separation distance. The results are shown in Fig. 15.11. As the monolayer was compressed from an average separation between the surfaces of the metal cores of 1.2 (± 0.2) nm to 0.5 (± 0.2) nm, the linear and nonlinear optical properties reveal evidence of both classical and quantum interparticle coupling phenomena. Below 0.5 nm, evidence for a sharp insulator-to-metal transition is observed in both optical signals. The nonlinear optical response abruptly decreases to a nearly constant value, and the linear reflectance drops precipitously until it matches that of a continuous metallic film. This transition is reversible: The particles can be re-dissolved back into a colloid, or, if the trough barriers are opened, the film is again characterized by the optical properties of near-isolated silver nanocrystals. The transition between the metal-like and insulator-like can be summarized using a new parameter D/2r, in which D is the interparticle separation and r is the radius of the nanocrystal. For D/2r>1.3, the optical response abides by the classical coupling model; for 1.2 < D/2r < 1.3, quantum coupling dominates the trends in the linear and non-linear optical responses; for D/2r<1.2, a sharp metal/insulator transition was observed. More work is required to verify if this general conclusion applies to other metal nanocrystal systems. Practically, this is also a good method to handle the self-assembled thin films. First, large area self-assembled nanocrystal films can be formed on the surface of a Langmuir trough and then transfer red to a suitable substrate so that different kinds of properties can be measured.
Figure 15.11 (a) The change from insulator connection between thiol capped silver nanocrystals (diameter 3.5 nm) to metallic connection was demonstrated on the plot of frequency dependent dielectric modulus. The Langmuir film response is initially an RC equivalent circuit, as the separation between the particles in the self-assembly became less than 0.6 nm, the film becomes inductive. (b) UV absorption spectra collected in situ as film was compressed. Correlation with the nonlinear optical response is indicated. After excessive compression, the strong plasma peak vanished (Collier, et al., 1997; Markovich, et al., 1998).
201
202
A strategy for the synthesis of "nanocrystal molecules", in which discrete numbers of gold nanocrystals are organized into spatially defined structures based on Watson-Crick base-pairing interactions, were proposed by Alivisatos et al. (1996). Single-stranded DNA oligonucleotides of defined length and sequence were attached to individual nanocrystals, and these assemble into dimers and trimers in addition of a complementary single-stranded DNA template. This approach should allow the construction of more complex 2- and 3-D assemblies between organic and inorganic materials.
The charging effect of metal nanocrystals in the self-assembly is very important to understand the basic properties of self-assembly. A transition from metal-like double-layer capacitive charging to redox-like charging was observed in electrochemical ensemble Coulomb staircase experiments from solutions of gold nanoparticles of varied core sizes (Chen, et al., 1998). The monodisperse gold nanoparticles are stabilized by short-chain alkanethiolate monolayers and have 8 × 103 to 38 × 103 core mass (1.1 nm to 1.9 nm in diameter). Larger cores display Coulomb staircase responses consistent with double-layer charging of metal-electrolyte interfaces, whereas smaller core nanoparticles exhibit redox chemical character, including a large central gap. The change in behavior is consistent with new near-infrared spectroscopic data showing an emerging gap between the highest occupied and lowest unoccupied orbitals of 0.4 eV to 0.9 eV. In addition, because the staircase behavior is closely related to MPC core electronic energy structure, it may aid understanding of other nanophase properties, such as the metal-insulator transition of silver nanoparticles upon compression. Finally, although differences in fundamental properties reside in the metal-like and molecule-like charging behaviors, we anticipate that their electrochemical, thermodynamic, and kinetic properties will, upon further study, prove to fit within a common formal representation.
If the building block in the self-assembly is magnetic nanocrystals, some extra magnetic effects may be brought in. Preliminary results showed that the blocking temperature of self-assembled magnetic nanocrystals have some differences from the bulk materials. It was reported that the blocking temperature of 2-D monolayer of cobalt nanocrystal increased from 58 K (for isolated cobalt nanocrystals with an average size of 5.8 nm) to 63 K. It was predicted that the difference may be even higher in the case of 3-D nanocrystal self-assembly. Larger area 2-D magnetic nanocrystal self-assembly and the corresponding magnetic properties have also been reported (Sun, et al., 1999). The magnetic measurement showed that broad transition from superparamagnetic to ferromagnetic at 165 K is due to the magnetostatic particle interactions in the close-packed nanocrystal network. But for the diluted solution with dispersed magnetic nanocrystals, the transition temperature is 105 K. All these results demonstrate that the interactions between the nanocrystals in a self-assembly influence the magnetic properties.
Quantum-confined Stark effect in quantum wells is very useful for optical modulations. In the case of semiconductors, Stark effect was found very strong at nanoscale. Semiconductor nanocrystals can also be used for labeling materials because the band gap can be adjusted by changing the size in the visible light range (Bruchez, et al., 1998).
15.2 Ordered Self-Assembly of Mesoporous Materials
203
Preparation of porous materials with low density and large surface area is important for many technological applications, such as catalysis and sensors. The low density (about 10%—30% of the bulk density) of the material results in very low dielectric constant (Bruinsma, et al., 1997), with great prospects in electronic devices. The large surface area of the porous materials is ideal for catalysis due to the enhanced surface activities. The controllable pore size and the interconnected channels in the material make it ideal for applications in filtering, separation membranes and electrodes for energy storage. Further, the ordered assembly has a great potential in photonic crystals, in which synthetic structures are required with modulated dielectric constant at a periodicity of a few hundred nanometers. Photonic crystals of tunable frequency band gap are useful for optically activated devices, such as waveguides and optical switches (Wijnhoven and Vos, 1998; Holland, et al., 1998; Zakhidov, et al., 1998).
15.2.1 Processing
There are many methods available for synthesis of porous materials, but most of them are inadequate for processing ordered porous structures. Recently, using stable colloidal monodispersive template spheres to process the ordered porous materials has attracted a great deal of research interest due to the simplicity and flexibility of the method. It is easy to form ordered structure after the colloidal solution is dried, and the size of the ordered pores is adjustable by changing the size of the monodispersive spheres. The most common templates are silica or polystyrene, which can be used to process positive or negative ordered structure. Ordered porous materials of oxides (Velev, et al., 1997), graphite (Zakhidov, et al., 1998), diamond (Zakhidov, et al., 1998), or organic films (Park, et al., 1997) have been prepared by this technique. Using a template-assisted technique, three different kinds of porous materials have been prepared: ordered porous polyurethane filter, porous (La, Sr) MnO3, and double length-scale ordered porous structure of SiO2.
Like ordered self-assembly of nanocrystals, ordered mesoporous materials can also be processed by self-assembling mechanism through hydrogen bonding, van der Waals force, electrostatic forces, etc. The general accepted definition of mesoporous materials refers to inorganic materials with pore sizes between 2–50 nm. It is only very recently that there has been a big step forward in this area. As we all know, it is much easier to make porous media with pore sizes below 1.5 nm, such as zeolites with pore size less than 1.5 nm successfully developed several dozen years ago.
From the geometric arrangement of pores, these materials can be divided into two groups: ordered pore structure and disordered pore structure. The methods used to make these two kinds of mesoporous materials are quite different. For example, the aerogel method can be used to process the disordered mesoporous materials also with high pore volume percentage (Prakash, 1995a, 1995b). But the pore size distribution of the mesoporous materials is very broad, which lowers the connectivity of pores and degrades the mechanical properties. In this thesis, only the ordered mesoporous materials are to be covered. Scientists at Mobil invented a novel and simple method that can be used to process ordered mesoporous materials. This method is called "supramolecular templating" (Beck, et al., 1992; Kresge, et al., 1992).
Normally, surfactant molecules used as supramolecular template have an asymmetric structure, a hydrophilic head group and a long hydrophobic tail group. After the surfactant and material precursor are mixed, the surfactant molecules self-organize into many small micelles in nanoscale into 1-D, 2-D or 3-D material's networks with long range order in the liquid mixture, trying to minimize the interfacial contact
204
between the incompatible molecular ends and achieve low-energy status. The typical self-assembling structural model includes lamellar, simple cubic, hexagonal, face-centered cubic, etc. The solution with ordered micelle structures can be gelled later at temperature higher than room temperature for several hours or longer after the surfactant molecules are removed by heating or washing. If the material does not collapse, the skeleton is preserved and a mesoporous material is formed.
15.2.2 The Formation Mechanisms
The formation mechanisms for the ordered mesoporous materials are still in debate. The most acceptable mechanisms include: liquid crystal mechanism, electrostatic charge balance, etc.. The mechanisms proposed by Mobil scientists are shown in Fig. 15.12(a). Both of the proposed pathways involve the interaction between the positively charged ammonium head group of the surfactant and the negatively charged inorganic precursors. But from Fig. 15.12(b), which is the structure-composition phase diagram of surfactant (CTBA) in water, liquid phase templating mechanism is almost impossible. It is clearly shown that, as the concentration of surfactant increases, the micelle can self-organize into hexagonal liquid crystals and then to lamellae structure. But in the case in which liquid crystal mechanism was proposed, the concentration of surfactant was well below the critical micelle concentration (CMC), so a micelle rod cannot even be formed during the early stage of self-assembling. But in the case of dip-coating, the concentration of surfactant changes during the drying process, from which CMC criteria cannot be ruled out as a mechanism Fig. 15.13. The drying path in the ternary phase diagram is shown in Fig. 15.12(b). The geometry configuration is summarized in Fig. 15.14.
Figure 15.12 (a) The two mechanisms proposed for the formation of ordered mesoporous silica (MCM-41) which was processed by ionic surfactant CTAB or CTAC: liquid-templating mechanism and incorporated mechanism. (Beck, et al., 1992). (b) The temperature dependent phase diagram of CTAB and water. (Brinker, et al., 1999)
205
Figure 15.13 (a) Steady-state film thinning profile during dipcoating of a mixed solution of silica precursor, water, surfactant and alcohol. Initial surfactant concentration is below critical micelle concentration (CMC). (b) An approximate path in the ternary phase diagram during dip coating. Point A is the initial composition, point B is the near drying line, and point C is the composition of dried film. (Brinker, et al., 1999)
206
Figure 15.14 The possible configurations of liquid crystals: (a) Body-centered cubic packing of disperse spheres; (b) Face-centered cubic packing of disperse spheres; (c) Hexagonal structure; (d) Undulated
207
cylinders with tetragonal packing; (e) Gyroid phase; (f) Bicontinous double-diamond structure; (g) Im3m phase; (h) Spongelike structure; (i) Planar lamellar structure; (j) Undulated lamellae with tetragonal symmetry; (k) Tetragonally perforated lamellae; (l) Hexagonally perforated lamellae with rhombohedral symmetry (Antonietti, et al., 1997).
208
Several other mechanisms were proposed, including micelle layer mechanism, charge density matching mechanism, etc. Even though it seems none of them can solve all of the problems, one point is widely accepted that the self-assembly is formed due to the electrostatic interaction between the inorganic precursor and the surfactant head. This is easy to be understood for ionic surfactants. For non-ionic surfactants, there are no positively and negatively charged species. But the surfactant molecules and the inorganic precursor can form hydrogen bonding or covalent bonding. A general form of electrostatic interaction can be used to represent the mechanism: for example, for ionic surfactant, the reaction mechanism can be represented as S+ I-, in which S is surfactant head group and I is inorganic precursor. For non-ionic surfactant, the form is S0 I0.
Selection of surfactants depends on the compatibility of the precursor, the pore size required, the favorable structure, and heating temperature. Compared with ionic surfactants, the polyethylene oxide polymer surfactant has many advantages, including low cost, non-toxicity, larger structure features up to 30 nm by increasing the molecular weight, lower decomposition temperature (compared with cationic ammonium the most common surfactant used), and excellent interfacial stability. For example, mesoporous Nb2 O5 was processed by forming a covalent bond between the Nb2 (C2 H5 O)5 precursor and an amine temping agent.
15.2.3 Applications
The applications of mesoporous materials mainly depend on these two properties: (1) the huge surface area, (2) the porous structure (low dielectric constant). To use the high surface area in catalytic applications, the pores need to be interconnected because the closed internal pores are inaccessible to outside by gas or liquid phases. For some special purposes, the internal surface should be activated for the functional groups to be attached. The surface area measurement of silica indicated that the internal connectivity of ordered mesoporous materials is better than disordered. Functional groups like thiol groups have been successfully introduced onto the pore surface of mesoporous silica as the terminal groups of organic monolayers (Fig. 15.15). The hydrocarbon chains aggregated and formed close-packed arrays on the substrate. The siloxane groups then underwent hydrolysis and ultimately became covalently attached to the substrate and cross-linked to one another. This material can efficiently remove mercury and other heavy metals (such as lead and silver) from contaminated aqueous and organic solutions. The distribution coefficient, Kd, has been measured as high as 340000. [Kd is the amount of adsorbed metal (µg) on 1 g of adsorbing material divided by the metal concentration (µg/L) remaining in the treated waste stream.]
Figure 15.15 Functional monolayers on the internal surface of mesorporous silica support can remove heavy metal contamination such as mercury from aqueous and organic solutions. (a) The TEM image shows the preservation of the porous structure after the attachment of functional monolayer. (b) An energy dispersive spectrum shows the sulfur signal from functional monolayer and mercury signal from the contamination in water. (Feng, et al., 1997)
209
Another useful area of mesoporous materials is for catalytic applications. The accessible internal surface area can be considered approximately proportional to the catalytic activity. Especially, the large pore size can be used to fix large active complexes which are out of reach of zeolites. Recently, crystalline linear polyethylene nanofibers (diameter 30 nm to 50 nm) were grown from the pores of the internal surface of mesoporous silica by polymerization of ethylene (Kageyama, et al., 1999). Another interesting work is using the nanopores of the mesoporous silica to limit the size of semiconductor clusters, thus controlling their properties. For example, a simple method was developed to process silicon nanocrystals with size 1 nm in the nanopores of hexagonal mesoporous silica free-standing films. The whole film displayed strong yellow-orange photoluminescence and nanosecond luminescence lifetimes, in contrast to milli- or microsecond lifetimes for porous silicon and nanocrystalline silicon (Dago, et al., 1999). This process can be easily adapted to make silicon-based light-emitting diodes, optical interconnections, displays and chemical sensors.
210
For low dielectric applications, it is the total volume of pores in the film rather than the internal connectivity that influences the dielectric constant of the porous media. There has been a great deal of research work on the application of so-called "nanoglass" processed by xerogel method. Figure 15.15 shows the application of nanoporous silica as a low-dielectric material and very good gap-filling property.
Continuous, flexible and transparent filters can be processed as well. Polyurethane was infiltrated into the space between polystyrene spheres (PSS). Then the composites of PSS and polyurethane were gelled at 80°C for 24 h. After the gellation, the sample was put into a beaker with pure toluene for 24 h to dissolve the PSS. The sample was put into SEM for observation after coating with a thin layer of gold to minimize charging effect. In Fig. 15.16 (a), the plane view of the pore films showed that the pores are ordered and from the cross section of the torn surface shown in Fig. 15.16 (b). The thickness of the filter can be controlled by the mold and the amount of polyurethane used in the experiment. The film is continuous without crack, and it has extremely high mechanical flexibility and is optically transparent. These are the unique characteristics of the film.
Figure 15.16 Scanning electron microscopy images of an ordered porous polymer showing (a) the top as-prepared surface, with continuous and fairly smooth morphology, and (b) the torn cross section of the film exhibiting porous interior structure.
211
212
15.2.4 Mesoporous Materials of Transition Metal Oxides
Due to several oxidation states of transition metal ions, mesoporous transition metal oxides have many important applications like gas separation membranes, electrodes in fuel cells and mixed valence catalysts. This is the basis of functional and smart materials (Wang and Kang, 1998). For example, the solar cell using mesoporous titania (anatase phase) as electrode has a high conversion efficiency (as high as 33%) from photons to electric current. TiO2 (Antonelli, et al., 1995), ReO2 (Herrmann, et al., 1995; Froba, et al., 1999), ZrO2 (Antonelli, et al., 1996b), HfO2 (Liu, et al., 1997), Nb2O5 (Sun, et al., 1997), SnO2 (Severin, et al., 1998), Ta2O5 (Antonelli, et al., 1996a), TiO2 doped with Co (Yin and Wang, 1999b), and MnxOy (Tian, et al., 1997) have been successfully synthesized.
The synthesis of (La, Sr)MnO3 is an example. The precursor used to prepare porous La0.66Sr0.33MnO3 were 0.03 mol water (with 25% acetic acid) solution of lanthanum acetate hydrate, strontium acetate, and manganese (II) acetate tetrahydrate (Yang, et al., 1998). It was shown that this precursor is a stable, clear solution. After the solution was infiltrated into the template and gelled at 80°C for several days, the sample was put into an oven at 600°C for 5 h to remove the PSS and transform the gellation into mixed transition metal oxides. SEM image clearly reveals the formation of the porous structure (Fig. 15.17(a)). Though the pore structure is visible, there is less ordering possibly due to the large mass change and volume shrinkage during heat treatment. X-ray diffraction showed that the structure was La1-xSrxMnO3 perovskite (Fig. 15.17(b)), and the average grain size calculated from the peak width at half maximum using the Scherr equation is about 12 nm, in agreement with the measurements from a TEM image (Fig. 15.17(c)). Electron diffraction (Fig. 15.17(d)) and EDS microanalysis also proved that the composition and crystal structure agreed well with perovskite La0.66Sr0.33MnO3.
Figure 15.17 (a) SEM image of the porous La1-xSrxMnO3 structure prepared by the template-assisted technique. (b) An X-ray diffraction spectrum of the La1-xSrxMnO3 powder, showing the perovskite structure. (c) A bright-field TEM image showing the average grain size is about 10 nm—20 nm. (d) An electron diffraction pattern from the particles proving the perovskite structure.
213
15.3 Hierarchically Structured Nanomaterials
The last few sections were focused on the self-assembly of ordered structures at one length scale. A hierarchically ordered material is introduced by combining the ordering at different length scales, which is formed in a single process. In fact, hierarchical ordering is very common in co-polymer structures. As discussed above, the ordered assembly of hollow structures of polyurethane and transition metal oxide can be processed by the template-assisted technique. These structures are ordered on the length scale of the template spheres and the pore sizes are in submicron to micron range. Here, we call this a first order porous structure. Alternatively, ordered mesoporous silica with much smaller pore sizes in nanoscale range (< 30 nm), produced deliberately by introducing surfactant, has also been processed, in which the porosity is created by the co-polymer. This is a second order porous structure. We have combined the two types of porosities into a new silica structure that has ordering and porosity on two length scales, one is at the scale
214
of hollow spheres created by a template of polystyrene, and the other is the nanocavities created by self-assembled molecular co-polymers (Yin and Wang, 1999c). The structure is stable to temperatures as high as 500°C.
In a typical synthesis process, 40 mL solution was put in a perpendicular open-end glass tube with inner diameter of 1.5 mm and the drying process in air took about 10 h, forming the ordered template of PS. Secondly, after the template was dried, the silica precursor, tetraethoxysiliane (TEOS), and a surfactant co-polymer were infiltrated into the space between the PSS in the ordered template (Zhao, et al., 1998). It has been demonstrated that the pore brought in by non-ionic co-polymer surfactant is larger than that reported previously by using ionic surfactant (Kresge, et al., 1992). Finally, after the precursor was dried slowly at room temperature, annealing of the template at 450°C for 5 h resulted in the simultaneous formation of the ordered porosity at double length-scales. The material processed without the surfactant co-polymer has porosity only in the length scale of the PSS.
The ordering at the two length scales is revealed by TEM (Fig. 15.18). The size of the hollow spheres is (120±8) nm, smaller than the size of the PS due to volume shrinkage. TEM images and their Fourier transforms have shown that the packing of the hollow spheres has the hexagonal close-packed (hcp) (αh = (120±8) nm) and the face-centered cubic (fcc) (αc = (175±10) nm) structures. The nanocavities formed in the walls of the shells have sizes of 4–5 nm and the interpore distance is (8±1.5) nm.
Figure 15.18 Transmission electron microscopy image of the porous silica, exhibiting ordering in two length scales: close-packed hollow spheres ( 120 nm) and self-organized nanocavities (4–5 nm).
215
15.4 Summary
Self-assembling process is fundamental in biological systems in nature. Synthesis of new materials using self-assembly is a new approach that has the potential of producing high quality, large quantity and chemically and structurally controlled new materials. This chapter reviewed the current status of the ordered self-assembled nanocrystals, ordered mesoporous nanostructured materials, and hierarchically ordered
216
materials. Most of the results are rather exciting, but substantial research is needed to improve and control the synthesis process to attain chemically and structurally well-characterized materials of high quality at a large yield. Growth of large single-crystalline self-assembled nanocrystal structures is crucial for their applications. Template-assisted synthesis techniques are possible solutions. Mesoporous structures are likely to be adequate for some applications, but improving ordering, minimizing defects and eliminating cracks are important issues. The hierarchically structured materials are interesting, but the unique properties of them remain to be investigated.
References
Ahmadi T. S., Z. L. Wang, T. C. Green, A. Henglein and M. A. El-Sayed. Science. 272, 1924 (1996)
Alivisatos A. P. Science. 271, 933 (1996a)
Alivisatos, A. P., J. Phys. Chem.. B100, 13226 (1996b)
Alivisatos A. P., K. P. Johnson, X. G. Peng, T. E. Wilson, C. J. Loweth, M. P. Bruchez, P. G. Schultz. Nature. 382, 609 (1996c)
Alvarez M. M., J. T. Khoury, T. G. Schaaff, M. Shafigullin, I. Vezmar and R. L. Whetten. Chem. Phys. Lett.. 266, 91 (1997)
Andres R. P., J. D. Bielefeld, J. I. Henderson, D. B. Janes, V. R. Kolagunta, C. P. Kubiak, W. J. Mahoney, R. G. Osifchin. Science. 273, 1690 (1996)
Antonelli D. M. and J. Y. Ying. Angew. Chem. Int.. Ed.. 34, 2014 (1995)
Antonelli D. M. and J. Y. Ying. Chem. Mater.. 8, 874 (1996a)
Antonelli D. M. and J. Y. Ying. Chem. Mater.. 8, 461 (1996b)
Antonietti M., C. Goltner. Angew. Chem. Int.. Ed. 36, 910 (1997)
Beck J. S., J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, K. D. Schmitt, C. T. W. Chu, D. H. Olson, E. W. Sheppard, S. B. Mccullen, J. B. Higgins, J. L. Schlenker. J. Am. Chem. Soc.. 114, 10834 (1992)
Bentzon M. D., J. Van Wonterghem, S. MØrup, A. Thölen, C. J. W. Koch. Phil. Mag.. B60, 169 (1989)
Braun P. V., P. Osenar, S. I. Stupp. Nature. 380, 325 (1996)
Brinker C. J., Y. Lu, A. Sellinger, and H. Fan, Adv. Mater.. 11, 579 (1999)
Bruchez M., M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos. Science. 281, 2013 (1998)
Bruinsma P. J., N. J. Hess, J. R. Bontha, J. Liu, and S. Baskaran. Mater. Res. Soc. Symp. Proc.. 443, 105 (1997)
Brus L.. Appl. Phys.. A53, 465 (1991)
217
Chen S. W., R. S. Ingram, M. J. Hostetler, J. J. Pietron, R. W. Murray, T. G. Schaaff, J. T. Khoury, M. M. Alvarez, R. L. Whetten. Science. 280, 2098 (1998)
Chung S. W., G. Markovich, J. R. Heath. J. Phys. Chem.. B102, 6685 (1998)
Cleveland C. L., U. Landman, M. N. Shafigullin, P. W. Stephens, R. L. Whetten. Z. Phys.. D,40, 503 (1997a)
Cleveland C. L., U. Landman, T. G. Schaaff, M. N. Shafigullin, P. W. Stephens, R. L. Whetten. Phys. Rev. Lett.. 79, 1873 (1997b)
Collier C. P., R. J. Saykally, J. J. Shiang, S. E. Henrichs, J. R. Heath. Science. 277, 1978 (1997)
Collier C. P., T. Vossmeyer, J. R. Heath. Auual Rev Phys. Chem.. 49, 371 (1998)
Dagö., G. A. Ozin, H. Yang, C. Reber, and G. Bussiere. Adv. Mater.. 11, 474 (1999)
Farady M.. Philos. Trans. Ro. Soc.. London, 147, 145 (1857)
Fendler J. H., F. C. Meldrum. Adv. Mater.. 7, 607 (1995)
Feng X., G. E. Fryxell, L. Q. Wang, A. Y. Kim, J. Liu, and K. M. Kemner. Science. 276, 923 (1997)
Froba M. and O. Muth. Adv. Mater.. 11, 564 (1999)
Giersig M., P. Mulvaney. J. Phys. Chem.. B97, 6334 (1993)
Grunig U., V. Lehmann, S. Ottow, and K. Busch. Appl. Phys. Lett.. 68, 747 (1996)
Guzelian A. A., J. E. B. Katari, A. V. Kadavanich. J. Phys. Chem.. B 100, 7212 (1996)
Harfenist S. A., Z. L. Wang, R. L. Whetten, I. Vezmar, M. M. Alvarez. J. Phys. Chem.. B 100, 13904 (1996)
Harfenist S. A., Z. L. Wang, R. L. Whetten, I. Vezmar, M. M. Alvarez. Adv. Mater.. 9, 817 (1997)
Hayashi T., S. Hirono, M. Tomita, and S. Umemura. Nature. 381, 772 (1996)
Heath J. R., P. J. Kuekes, G. S. Snider, R. S. Williams. Science. 280, 1716 (1998)
Herrmann W. A., R. W. Fischer. J. Am. Chem. Soc.. 117, 3223 (1995)
Holland B. T., C. F. Blanford, A. Stein. Science. 281, 538 (1998)
Holland B. T., C. F. Blanford, T. Do and A. Stein. Chem. Mater.. 11, 795 (1999)
Hu H., M. Brust, A. J. Bard. Chem. Mater.. 10, 1160 (1998)
Kageyama K., J. Tamazawa, T. Aida. Science. 285, 2113 (1999)
Kiely C. J., J. Fink, M. Brust, D. Bethell, D. J. Schiffrin. Nature. 396, 444 (1998)
Kresge C. T., M. E. Leonowioz, W. J. Roth, J. C. Vartuli and J. S. Beck. Nature. 359, 710 (1992)
218
Liu P., J. Liu, A. Sayari. Chem. Commun.. 577 (1997)
Markovich G., C. P. Collier, J. R. Heath. Phys. Rev. Lett.. 80, 3807 (1998)
Matsumoto H., T. Sakata, H. Mori, H. Yoneyama. J. Phys. Chem.. B 100, 13781 (1996)
Mattoussi H., A. W. Cumming, C. B. Murray, M. G. Bawendi, R. Ober. Phys. Rev.. B58, 7850 (1998)
Moritz T., J. Reiss, K. Diesner, D. Su, A. Chemseddine. J. Phys. Chem.. B 101, 8052 (1997)
Murray C. B., C. R. Kagan, M. G. Bawendi. J. Am. Chem. Soc.. 115, 8706 (1993)
Murray C. B., C. R. Kagan, M. G. Bawendi. Science. 270, 1335 (1995)
Nirmal M., B. O. Dabbousi, M. G. Bawendi, J. J. Macklin, J. K. Trautman, T. D. Harris, L. E. Brus. Nature. 383, 802 (1996)
Ohara P. C., J. R. Heath, W. M. Gelbart. Angew. Chem. Int. Ed. Eng.. 36, 1078 (1997)
Ohara P. C., D. V. Leff, J. R. Heath, W. M. Gelbart. Phys. Rev. Lett.. 75, 3466 (1995)
Papiper E., P. Horny, H. Balard, R. Anthore. C. Petipas and A. Martinet. J. Colloid Interface Sci.. 94, 207 (1983)
Park S. H. and Y. Xia. J. Mater. Res.. 10, 1745 (1997)
Petit C., A. Taleb, M. P. Pileni. Adv. Mater.. 10, 259 (1998)
Pileni M. P. Langmuir. 13, 3266 (1997)
Prakash S. S., C. J. Brinker, A. L. Hurd. J. Non-Crystalline Solids. 190, 264 (1995a)
Prakash S. S., C. J. Brinker, A. L. Hurd and S. M. Rao. Nature. 374, 439 (1995b)
Provencio P. N., J. E. Martin, J. G. Odinek, J. P. Wilcoxon. Microscopy and Microanalysis. 4(suppl. 2), 734 (1998)
Reimer L.. Energy-Filtering Transmission Electron Microscopy. Springer Series in Optical Sciences, vol. 71, Springer, Berlin (1995)
Severin K. G., T. M. Abdel-Fattah, T. J. Pinnavaia. Chem. Commun.. 1471 (1998)
Shenton W., D. Pum, U. B. Sleytr, S. Mann. Nature. 389, 585 (1997)
Steigerwald M. L., A. P. Alivisatos, J. M. Gibson, T. D. Harris, R. Kortan, A.J. Muller, A. M. Thayer, T. M. Duncan, D. C. Douglass, L. E. Brus. J. Am. Chem. Soc.. 110, 3046 (1988)
Sun H. B., S. Matsuo, H. Misawa. Appl. Phys. Lett.. 74, 786 (1999)
Sun T. and J. Y. Ying. Nature. 389, 704 (1997)
Thomas J. R. J. Appl. Phys.. 37, 2914 (1966)
Tian Z. R., W. Tong, J. Y. Yang, N. G. Duan, V. V. Krishnan, S. L. Suib, Science. 276, 926 (1997)
219
Tomaselli M., J. L. Yarger, M. Bruchez, R. H. Havlin, D. deGraw, A. Pines, A. P. Alivisatos. J. Chem. Phys.. B110, 8861 (1999)
Velev O. D., T. A. Jede, R. F. Lobo and A. M. Lenhoff. Nature. 389, 448 (1997)
Velev O. D., T. A. Jede, R. F. Lobo and A. M. Lenhoff. Chem. Mater.. 10, 3597 (1998)
Vossmeyer T., E. Delonno, J. R. Heath. Angew. Chem. Int. Ed. Eng.. 36, 1080 (1997)
Wang Y., N. Herron. J. Phys. Chem.. B95, 525 (1991)
Wang Z. L.. Characterization of Nanophase Materials. Wiley-VCH (2000)
Wang Z. L., T. S. Ahmadi,M. A. El-Sayed. Surface Sci.. 380, 302 (1997)
Wang Z. L. and Z. C. Kang. Functional and Smart Materials—Structural Evolution and Structure Analysis. Plenum, New York (1998)
Wang Z. L. Adv. Mater.. 10, 13 (1998a)
Wang Z. L., S. A. Harfenist. I. Vezmar. R. L. Whetten, J. Bentley, N. D. Evans, K. B. Alexander. Adv. Mater.. 10, 808 (1998b)
Wang Z. L., S. A. Harfenist, R. L. Whetten, J. Bentley, N. D. Evans. J. Phys. Chem.. B102, 3068 (1998c)
Whetten R. L., J. T. Khoury, M. M. Alvarez, S. Murthy, S. I. Vezmar, Z. L. Wang, C. C. Cleveland, W. D. Luedtke, U. Landman. Adv. Mater.. 8, 429 (1996)
Wijnhoven J. E. G. J. and W. L. Vos. Science. 281, 802 (1998)
Wilson W. L., P. F. Szajowski, L. E. Brus. Science. 262, 1242 (1993)
Yang S., C. T. Lin, K. Rogacki, B. Dabrowski, P. M. Adams and D. M. Speckman. Chem. Mater.. 10, 1374 (1998)
Yin J. S., Z. L. Wang. J. Mater. Res.. 14, 503 (1999a)
Yin J. S. and Z. L. Wang. Adv. Mater.. 11, 469 (1999b)
Yin J. S. and Z. L. Wang. Appl. Phys. Lett.. 74, 2629 (1999c)
Yin J. S. and Z. L. Wang. Phys. Rev. Lett.. 79, 2570 (1997a)
Yin J. S. and Z. L. Wang. J. Phys. Chem.. B101, 8979 (1997b)
Ying J. Y., C. P. Mehnert, M. S. Wong. Angew. Chem. Int. Ed. Eng.. 38, 56 (1999)
Zakhidov A. Z., R. H. Baughman, Z. Iqbal, C. Cui, I. Khayrullin, S. O. Dantas, J. Marti and V. G. Ralchenko. Science. 282, 897 (1998)
Zhao D., J. Feng, Q. Huo, N. Melosh, G. H. Fredrickson, B. F. Chmelka, and G. D. Stucky. Science. 279, 548 (1998)
220
Zhou C., M. R. Deshpande, M. A. Reed, L. Jones, J. M. Tour. Appl. Phys. Lett.. 71, 611 (1997)
221
16, Molecularly Organized nanostructural Materials
16.1 Introduction
As discussed in other chapters of this book, nanoscale materials and nanotechnology have generated great interest in the area of microelectronic devices, quantum computation, and nanoscale instrumentation. However, great opportunities in energy storage and conversion, chemical industry, environmental and health sciences have just emerged in the last few years. These opportunities can be illustrated by the following examples.
16.1.1 Nanostructural Materials in Energy Sciences
There has been intensive research in developing alternative energy devices to increase the efficiency of energy storage and delivery, and to reduce environmental pollution caused by energy consumption. In order to improve the energy density and power density for batteries and capacitors, many nanoscale electrode materials have been investigated, including nanoporous carbon and conducting oxide materials (Delnick and Tomkiewicz, 1996; Conway, 1991). The successful design and synthesis of next generation electrode materials depend on the manipulation of the electrochemical potential at the atomic level and on the enhanced charge storage and transfer through nanostructural engineering (Zheng and Jow, 1995). Recently carbon nanotubes have been investigated for efficient hydrogen storage (Chen, et al., 1999). Such materials may have potential use in high-density rechargeable batteries and fuel cells. It was found that at low temperature and high pressure, carbon nanotubes can store up to 67 wt% hydrogen (Dillonet, al., 1997, Chambers, et al., 1998). Alkali metal doped carbon nanotubes can store up to 20 wt% hydrogen under ambient pressure and moderate temperature.
16.1.2 Nanophase Materials in Environmental and Health Sciences
16.1.2.1 Drug Delivery
Nanoscale materials are recently beginning to attract interest as emerging technologies for medical diagnosis and treatment. Nanoparticulate drug delivery systems are among the latest entry into the drug delivery arena (Hnatyszyn, et al., 1994). The nanoparticles are either polymeric or ceramic in nature. The ceramic particles can be made of calcium phosphate, a biodegradable mineral existing in natural biological systems. The ceramic particles are coated with tightly adsorbed phospholipid membranes derived from cell structures and viral membranes, or other cell-specific ligands. The drugs are immobilized on the outer membrane and delivered to the targeted sites which recognize the membrane structures on the ceramic particles.
16.1.2.2 Diagonosis
222
DNA molecules have been attached to colloidal gold particles to form "DNA specific nano-colloids" (Mirkin, et al., 1996). These DNA coated nanoparticles can be used to detect the sequence of unknown DNA molecules and may offer the advantage of high selectivity and sensitivity. Depending upon whether the target DNA molecules are complete complementary, partial complementary, non-complementary to the DNA molecules attached to the particles, the nano-colloids will display a different color change as a function of temperature, thus providing easy means of diagnosis.
16.1.2.3 Functional Selectivity
The DNA modified nanoparticles are among many other examples in which the surfaces of the nanoparticles or nanoscale materials are modified with target specific molecules and functional groups. Many of these materials are already used or investigated for commercial applications. The recently developed ordered nanoporous materials based on templated synthesis (Beck, et al., 1992; Kresge, et al., 1992; Beck and Vartuli, 1996; Liu, et al., 1996; Raman, et al., 1996) provide an ideal platform to manipulate the surface chemistry of the materials on molecular and nanometer scales. Organized monolayers of functional molecules were assembled inside the ordered nanoporosity to form highly functional hybrid materials (Moller and Bein, 1998; Feng, et al., 1997; Liu, et al., 1998; Mercier and Pinnavaia, 1997, 1998). This approach allows for rational design of molecular sites. These materials demonstrated extremely high efficiency in adsorbing a wide range of species, including toxic metals.
16.1.3 Molecularly Organized Nanostructural Materials
There are two general approaches to fabricate the nanoscale materials: gas phase synthesis and solution phase synthesis. One of the new strategies in the solution synthesis is to take advantage of the fundamental molecular interactions between the different species and use these fundamental interactions to direct the scale, the structural ordering, and the functionality in the nanocomposite materials. We call such materials as molecularly organized nanocomposites. This chapter will use several examples to illustrate the different approaches to fabricate both 2-D and 3-D nanoscale materials.
16.2 Molecularly Directed Nucleation and Growth, and Matrix Mediated
Nanocomposites
16.2.1 Molecularly Directed Nanoscale Materials in Nature
There are many examples of sophisticated nanostructural materials that can satisfy the multifunctional needs of biological systems. For instance, the hierarchical structure of calcified bone tissues has been studied for a long time (Katz, 1996). The bone tissue consists mainly of collagen fibers and an inorganic component calcium hydroxyapatite (HAP, or Ca10(PO4)6(OH)2). On the nanometer scale, the HAP crystals (20 nm to 40
223
nm long) are aligned along the collagen fibers. The collagen fibers further organize themselves into helical concentric lamellar structures, which are also aligned along the longitudinal direction of the bone tissue. This sophisticated microstructure offers the optimum mechanical properties for the bone tissue. Another widely studied example is the abalone sea shell, a hybrid composite material made from calcium carbonate and organic materials (Currey, 1987). The outer layer of the shell consists of oriented columnar calcite crystals to provide strength and hardness to sustain the pressure and protect the living organism. The inner shell (nacre) is made of a lamellar composite of aragonite platelets and organic soft tissue. The aragonite crystals and the soft tissue are 0.5 µm and 10 nm in thickness, respectively (Liu, et al., 1992). This combination provides optimum strength and toughness. Some of the most important characteristics of biological composites include (Simkiss and Wilbur, 1989): (1) The materials are highly organized on nanometer scale. (2) The material synthesis is directed and controlled by organized functional molecules. (3) The materials are highly functional.
It is difficult to have a complete understanding of exactly how the biological systems regulate the growth of such sophisticated nanoscale materials of different phases and different orientations, not to mention duplicating these materials with synthetic approach. It has been recognized that the mineral nucleation and growth in the nacreous layer of abalone shells are regulated by an organic template structure. An insoluble organic framework (β-chitin layered between glycine and alanine rich proteins) (Weiner and Traub, 1984; Weiner, et al., 1983) is first formed as lamellar envelopes. Soluble polyanionic proteins (aspartic and glutamic acids) (Nakahara, et al., 1980; Cariolu and Morse, 1988) are then deposited on the insoluble organic template, and these soluble macromolecules alone are responsible for nucleation and growth of the inorganic minerals, whether it is calcite or aragonite (Belcher, et al., 1996; Zaremba, et al., 1996). Analysis of the crystalline structure of the nacreous layer reveals a hierarchical tiling of twin related aragonite platelets oriented along the c-axis (Liu, et al., 1992; Sarikaya, et al., 1995). The limited knowledge on the nucleation and growth of biocrystals provides some insights on how to synthesize tailored nanoscale composites using lessons learned from the biological systems.
16.2.2 Directed Nucleation and Growth of Thin Films
Taking the lessons from nature, functionalized molecules are used to direct the nucleation and growth, as well as the ordering of the crystalline materials. There are two possible pathways to achieve this goal, as illustrated in Fig. 16.1: directed nucleation and growth of thin films, and matrix mediated nanocomposites. These approaches allow the formation of either 2-D nanocrystalline films or 3-D nanocomposites.
Figure 16.1 Two pathways to prepare nanoscale materials utilizing functionalized molecules as directing agents: (a) Formation of functionalized monolayers on a substrate and growth of oriented crystals and films. (b) In situ swelling, functionalization, and growth of nanophase materials in polymer composites.
224
16.2.2.1 Self-assembled Functional Monolayers on Substrate
For 2-D thin films, we have used self-assembled bifunctional monolayers to direct nucleation and growth of ceramic materials on a substrate. Bifunctional molecules containing a hydrophilic head group and a hydrophobic tail group adsorb onto a substrate or an interface as closely packed monolayers. The tail group and the head group can be chemically modified to contain functional groups that stimulate the growth of the desired minerals. Experimentally, molecular monolayers can be prepared from organosilanes, such as Cl3Si(CH2)nX. The chlorosilane end of the molecule is hydrolyzed and then covalently bonded to the oxide surface and cross-linked to adjacent silanes. The hydrocarbon tails provide the driving force (van der Waals interaction) for the self-assembly of the molecules into ordered arrays on the substrate. Different functional groups, such as sulfate, phosphate, and carboxylic acid, as well as protein molecules, can be introduced to the head group via proper chemical treatment. These functional groups will induce the nucleation of minerals dissolved in the solution.
16.2.2.2 Controlled Nucleation and Growth
To illustrate how the crystalline orientation and morphology can be controlled, TEM micrographs of a cross-sectioned samples of iron oxyhydroxide (goethite, FeOOH) films on silicon wafers and polystyrene substrates (Rieke, et al., 1995; Tarasevich, et al., 1996) are shown in Fig. 16.2. Sulfonated monolayers on silicon were prepared by placing wafers into solutions of vinyl-terminated alkyl trichlorosilanes in cyclohexane. The vinyl groups were converted to sulfonate groups by reacting in sulfur trioxide (SO3) vapor. Deposition of the FeOOH film occurred by thermal hydrolysis of the acidified Fe (NO3)3 solutions at 70°C. Similar films were grown on sulfonated polystyrene. Before film growth, the polystyrene substrate was treated in sulfuric acid or in SO3 vapor. What is striking in Fig. 16.2 are the very different microstructures of the same materials under different conditions. In Fig. 16.2(a), the film growth rate is low, and the film orientation was controlled by the sulfonate groups. Selected area electron diffraction and X-ray indicated
225
that the film has a preferred orientation, with [001] direction parallel to the substrate. We can also see uniform fiber-like structures parallel to the substrate surface. On the other hand, the film on polystyrene was grown with [001] direction perpendicular to the substrate. Fiber-like microstructures are vertical with respect to the substrate surfaces. The iron hydroxide films shown in Fig. 16.2 closely resemble the two morphologies in the abalone shells: layered microstructure in the nacreous layer and the column-like structure in the outer layer. The layered structure is directed by the functional molecules, and the column like structures are controlled by growth kinetics. The ability to deposit oriented iron phases may lead to the development of high-density magnetic storage materials. TEM study revealed that in the commercial magnetic recording tape, the crystals assume a parallel orientation with respect to the tape, which is not a very favorable microstructure for high-density storage. The column-like microstructure is more desirable (Fig. 16.2(b)).
Figure 16.2 Two morphologies of FeOOH films on functionalized substrates. (a) Controlled nucleation and growth of layered structures with [001] direction parallel to the substrate. (b) Kinetically oriented films with [001] column-like structures perpendicular to the substrate.
16.2.2.3 Application
This molecular monolayer approach has been used to prepare iron oxide, tin oxide, titanium oxide, and calcium phosphate as coatings on ceramics, metal, and plastic substrates (Bunker, et al., 1994). Such films and coatings have potential applications in magnetic materials, scratch resistive coatings, ultraviolet blockers, and medical implants. The synthesis, performed in aqueous solutions, does not require high-temperature treatment. We also demonstrated that the desired micro- and macrostructural control (including crystalline phase, crystal size, morphology and orientation) can be obtained.
226
Another advantage is the ability to uniformly deposit films on complicated shapes. Figure 16.3(a) shows examples of commercial products on which we have successfully applied the controlled coating techniques. Otherwise it would be difficult to deposit uniform ceramic films on these complex objects. Figure 16.3(b) shows the UV adsorption spectrum of TiO2 films deposited on commercial light fixtures. The TiO2 films protect the fixtures and reduce human UV exposure. As can be seen from the figure, the optical transmission below 300 nm is reduced to less than 5% by a 90 nm TiO2 coating (Baskaran, et al., 1998).
Figure 16.3 (a) Ceramic thin film coatings applied to complicated shapes. (b) Optical transmission of TiO2 films with different thickness (Baskaran, et al., 1998, with permission from American Ceramic Society).
227
16.2.3 Matrix Mediated Nanocomposites
Three-dimensional nanocomposite materials can be developed based on simultaneous swelling and functionalizing of polymer materials (Fig. 16.1, second pathway). In this approach, a polymer material, such
228
as polystyrene, is exposed to SO3 vapor or soaked in a liquid solvent. The polymer swells into a microporous material. At the same time, the active groups on the polystyrene are converted to desired functional groups, such as sulfonate groups. Subsequently, nanophase ceramic materials are nucleated in the functionalized polymer matrix. This is a very simple, yet efficient technique to make polymer-ceramic nanocomposites.
Figure 16.4 shows that when polystyrene was exposed to SO3 vapor for some time, a swollen layer was formed. The actual layer thickness is more than 5 µm. Only part of the layer was observed in TEM due to tearing during sample sectioning. Sulfur can be detected from the sulfonate groups by electron energy dispersive (EDS) spectral compositional analysis. Figure 16.5 shows the time sequences of the growth of tin oxide-polystyrene nanocomposites. In the initial stage (30 s), a functionalized layer is over 5 µm in thickness and contains almost no ceramic material. As the reaction proceeds, SnO2 begins to deposit, and the relative concentration of Sn to S increases. At the same time, nanocrystalline SnO2 begins to appear, and the thickness of the functionalized layer decreases. The decrease in the layer thickness is caused by deswelling due to the collapse of the osmotic pressure through the deposition of the mineral phase. Finally, a SnO2 polystyrene nanocomposite is formed after several days, with the SnO2 crystals less than 50 nm in size. SnO2 is widely used as gas a sensor material (Williams and Coles, 1999). Finely divided SnO2 in the composite material could potentially increase the sensitivity of the sensing devices.
Figure 16.4 TEM image of sulfonated layers in polystyrene.
Figure 16.5 TEM images of the growth process of SnO2 nanocrystals in polystyrene matrix.
229
Longer growth time can lead to the coarsening of the SnO2 particles, as shown in Fig. 16.6. It is also interesting to notice that at the outmost surface a new morphology with large SnO2 crystals develops after a long time. These new particles are generated by significant crystal growth at the free surface. The change in particle size and morphology is also reflected in the selected area diffraction (SAD) patterns, with increased intensity as a function of time. In the outer layer in Fig. 16.6, the diffraction rings become less continuous due to the larger crystalline size.
Figure 16.6 TEM image of SnO2 nanocrystals in polystyrene after long growth time. Large particles are observed, especially on the outer layer.
230
Similar iron hydroxide-polystyrene nanocomposites can also be prepared, as shown in Fig. 16.7. Figure 16.7(a) is the nanocomposite with very fine particles (less than 5 nm), and Fig. 16.7(b) is the composite with long time aging. The crystals have grown into needle-like particles.
Figure 16.7 TEM images of FeOOH nanocrystals in polystyrene matrix. (a) Amorphous nanocrystals. (b) Needle-like particles after long growth time.
231
16.3 Surfactant Directed Hybrid Nanoscale Materials
The examples discussed so far have illustrated how the crystalline size, morphology, and orientation can be controlled through molecular directing. In the last few years it has become possible to further regulate the ordering on the nanometer scale to form more sophisticated nanoporous and nanostructural composite materials. These approaches are discussed in the followingsections.
232
16.3.1 Ordered Nanoporous Materials
16.3.1.1 Mesoporous Materials
In 1992, scientists at Mobil Oil Research successfully synthesized ordered nanoporous mesoporous materials using surfactant liquid crystals as structure directing agents (Beck, et al., 1992; Kresge, et al., 1992). This new class of materials is characterized by an extremely uniform pore size distribution, well-ordered structure, tunable pore size from 2 nm to 40 nm, and simple preparation methods. Since 1992 ordered mesoporous materials have become a very active research area, and there have been many excellent publications related to the fundamental mechanism, the synthesis and the properties of such materials. More recently, the templates were extended to include block copolymers (Yang, et al., 1998). The use of block copolymer surfactants further expanded pore size and the possible compositional range.
Experimentally the ordered mesoporous materials were prepared by mixing various surfactants with a wide range of ceramic precursors, and letting the mixture react under mild hydrothermal conditions. Schematically (not mechanistically), as illustrated in Fig. 16.8, the surfactant molecules aggregate into micelles (only rod- like micellar structure is shown). The rod-like micelles further aggregate into ordered hexagonal liquid crystalline structures. The ceramic precursors bind to the micellar head groups and further condense into a 3-D ceramic phase. Afterwards the surfactant molecules can be removed by thermal or chemical means. This process will give a unique ordered porous material as shown by the TEM micrograph in Fig. 16.8. If a polymeric material is used as the template, the organic phase can be left behind to give a highly ordered nanocomposite material (Gray, et al., 1997). Such ordered nanocomposites are difficult to prepare by other synthetic routes.
Figure 16.8 Schematic of the formation of ordered mesoporous materials using surfactant as the directing templates.
233
16.3.1.2 Mechanisms and Pathways
The formation of the ordered nanostructure is determined by the fundamental interactions between the molecules with surfactant like properties. These molecules include surfactant, lipids, copolymers and proteins (Israelachvili, 1991; Evans and Wennerstrom, 1994; Vinson, et al., 1991). Like the bifunctional molecules discussed earlier, these amphiphilic molecules have two components: a hydrophobic tail group (or segment) and a hydrophilic head group (or segment). The major driving forces for forming well-defined aggregates are the hydrophobic attractions at the hydrocarbon-water interfaces and the hydrophilic ionic or steric repulsion between the head groups. A variety of ordered structures can be formed depending on the solution conditions (pH, temperature, or electrolyte concentrations). The equilibrium structures are determined by the thermodynamics of the self-assembly process and the inter- and intra-aggregate forces. In order to quantitatively describe the molecular interactions, a critical geometric packing parameter, v/aolc, where v is the volume of the hydrocarbon chains, ao is the optimal head-group area, and lc is the critical chain length, is proposed. As shown in Fig. 16.9, a small critical packing parameter (<0.5) favors the formation of a highly curved interface (spherical micelles and rod-like micelles), and a larger critical packing parameter (>0.5) favors the formation of flat interfaces (flexible bilayers and planar bilayers). A critical packing parameter larger than unity will produce inverse micelles. However, the surfactant geometry shown in Fig. 16.9 depends on experimental conditions. For example, the ao of an ionic surfactant can be effectively decreased by adding salt or counter ions to decrease the electrostatic repulsion between the head group and/or by increasing the temperature; v can be effectively increased by adding oil, which mingles with the hydrophobic hydrocarbon tails. Therefore the mixing of a surfactant with ceramic precursors sometimes fundamentally changes the molecular interactions. Even though the phase behavior of a wide range of surfactants has been investigated (Tiddy, 1980), the phase diagrams of the surfactants should only be used as a guideline, not a predictive tool.
Figure 16.9 Packing geometry of surfactant molecules and its relation to the micellar structures (Israelachvili, 1991, with permission from Academic Press).
234
Depending on the solution composition, spherical micelles, rod-like micelles, hexagonally ordered crystals, cubic crystals, lamellar phases, inverse micelles, or inverse micellar liquid crystals can be formed. Using the ordered liquid crystalline structures as the directing templates, a wide range of ordered nanocomposites can
235
be synthesized. Figure 16.10 illustrates several important template structures formed by the surfactants, such as the cubic, hexagonal, and the lamellar phase, and the spherical, rod-like, and inverse micelles. Among these, the cubic, hexagonal, and lamellar structures have been utilized for preparing ordered mesoporous materials, and the spherical and inverse micelles have been used to prepare nanocrystalline particles. Ordered nanostructural composites consisting of many of the phases have been successfully synthesized (Huo, et al., 1994). In addition to ordered mesoporous materials, it is conceivable to prepare other highly ordered nanocomposites based on these ordered liquid crystalline structures. These ordered composites will contain unique network structures and arrays of interpenetrating phases to meet the structural requirements of "smart composites" with proper connectivity patterns, which are desirable in many applications, such as in piezoelectric transducers and pyroelectric detectors (Newnham, 1988).
Figure 16.10 Different microstructures formed in surfactant systems: six typical micellar structures observed in surfactant solutions.
Several possible mechanisms and pathways have been discussed in the literature when surfactants are involved in directing the ordered structures. Figure 16.11 illustrates the three main pathways: (1) pre-assembly, (2) co-assembly, and (3) modified co-assembly. In the pre-assembly process, the ordered surfactant liquid crystals formed first, and subsequently the silicate ions bonded to the surfactant head group (Dubois, et al., 1993; Attard, et al., 1995; Braun, et al., 1996). In the co-assembly process, the surfactants exist as individual micelles or molecules. Adding the silicate species caused the formation of silicate encapsulated surfactant micelles and subsequent ordering (Monnier, et al., 1993; Firouzi, et al., 1995). In the modified co-assembly process, the inorganic species were bonded or integrated to the surfactant molecules to form hybrid inorganic amphiphiles, followed by the ordering process (Antonelli and Ying, 1995).
Figure 16.11 Different mechanisms discussed for the formation of mesoporous materials. (1) Pre-assembly. (2) Co-assembly. (3) Modified co-assembly (Liu, et al., 1996, with permission from Elsevier).
236
16.3.1.3 Binding with Surfactants
The preparation of ordered mesoporous materials can be also classified based on what surfactants are used and how they are bonded to the ceramic phases. The initial Mobil (MCM-41 series) silicate-based materials were prepared using direct ionic bonding with a cationic surfactant. Apparently, mesoporous silicates can be synthesized under a variety of experimental conditions. A more generalized synthesis route was developed to include ion-mediated ionic bonding, in which the surfactants and the ceramics are similarly charged and bind together through an intermediate ionic species of opposite charge (Huo, et al., 1994). A range of ordered phases, including several lamellar, hexagonal (P63/mmc, P6m), and cubic phases (Pm3n, Ia3d), have been observed (Huo, et al., 1995). Neutral surfactants can be also used (Tanev and Pinnavaia, 1995; Bagshaw, et al., 1995). The neutral synthesis route used uncharged (dodecyl amine) or nonionic surfactant (polyethylene oxide). The oxide precursor was bonded to the surfactant through hydrogen bonding. This produced mesoporous materials with thick walls and small grain sizes. The surfactant can be removed by simple solvent extraction. A ligand-assisted templated approach using amine surfactant was also developed
237
to prepare transition metal mesoporous materials (niobium and tantalum oxides). In this approach the ceramics (alkoxide) is covalently bonded to the long-chain amine surfactant before the hydrolysis and condensation of the precursor. This allows the selective hydrolysis and condensation of the oxide precursor on the micellar surface and prevents the uncontrolled precipitation of particulate oxide materials that is commonly observed in non-silicate materials. The various chemical binding conformations involved with surfactants are illustrated in Table 16.1.
Table 16.1 Different approaches according to the surfactant and the interfacial bonding
Surfactant Bonding Mediating ions Inorganics Conformation
Cationic A+ Direct ionic None Anionic M- A+M-
Mediated ionic Anionic X- Cationic M+ A+X-M+
Anionic A- Direct ionic None Cationic M+ A-M+
Mediated ionic Cationic X+ Anionic M- A-X+M-
Neutral A0 or nonionic Hydrogen None Neutral M0 A0M0
Neutral amine Covalent None M(OEt)x –H2N–M(OEt)x
Amphoteric A+B Covalent None Aqueous precursor M A+B–M
16.3.2 Hybrid Nanoscale Materials
Many applications have been considered for the new mesoporous materials, including energy storage, catalysis, adsorption, ion exchange, sensing controlled release, etc. However, most of the applications require the materials to have specific binding sites, stereochemical configuration or charge density, and acidity (Sayari, 1996; Anthony, et al., 1993; Schierbaum, et at., 1994). Most mesoporous materials do not themselves have the appropriate surface properties.
16.3.2.1 Molecular Monolayers in Ordered Nanoscale Materials
A method has been developed to systematically modify the surface chemistry and tailor the molecular recognition process of mesoporous materials toward the targets (Fig. 16.12) (Feng, et al., 1997; Liu, et al., 1998). In this approach, high-quality, oriented molecular monolayers are spontaneously grown on ordered mesoporous ceramic substrates with controlled pore shape and pore size. The functional molecules are closely packed and cross-linked with one another. The terminal functional groups on the monolayer can be easily modified, thereby allowing rational design and layer-by-layer construction of host sites on the nanoporous substrates that match the shape, size, or chemical properties of heavy metals, transition metals,
238
or organic molecules. As such, making them extremely efficient scavengers of these species, or effective catalysts for reactions involving these species.
Figure 16.12 Formation of molecular monolayers in ordered mesoporous supports (Liu, et al., 1998, with permission from Wiley-VCH).
16.3.2.2 Application for Specific Adsorption
One application we studied involves alkyl thiols [tris(methoxy) mercaptopropylsilane, TMMPS] as the functional molecules for heavy metal remediation. Mercury and heavy-metal contamination is a serious problem at waste contaminated sites of the Department of Energy (Klein, 1994). Industrial and civilian sources deposit a large amount of mercury into the environment every year (Mitra, 1986). We selected TMMPS because it has been previously used to make functional monolayers, and the thiol groups have a high affinity for binding heavy metals. The thiol-silica hybrid mesoporous materials thus produced can efficiently remove mercury and other heavy metals (such as lead and silver) from contaminated aqueous and organic solutions. The distribution coefficient, Kd, has been measured as high as 108. (Kd is defined as the amount of adsorbed metal (µg) on 1 g of adsorbing material, divided by metal concentration (µg/mL) remaining in the treated waste stream.)
The exceptional selectivity and capability of hybrid materials to remove mercury and other heavy metals from contaminated solutions have been demonstrated under a wide range of conditions (water, oil, acidic, neutral and basic solvents). Figure 16.13 shows the mercury concentration remaining in the waste solution as a function of treatment time. The hybrid materials remove the Hg in the water much faster and to a much lower concentration. A loading capacity of 600 mg (Hg)/g (absorbing materials) has been obtained. A single treatment of highly contaminated water usually reduced the mercury concentration to well below U.S. Environmental Protection Agency elemental limits for hazardous wastes and even drinking water standards.
239
Similar results have also been obtained for lead and silver, which are major concerns in drinking water. The performance of the materials is not affected by the presence of background electrolytes (ions of barium, zinc, sodium, or nitrate). In addition, the hybrid materials have other advantages, such as stability and recyclability. In situ NMR experiments indicated the bonding between the mercury and thiol group and the structure of the organic monolayers are stable up to 125°C. The mercury-loaded materials were heated in water at 70°C and released little mercury. The mercury-loaded materials can be regenerated by washing in a concentrated HCl (12.1 mol/L) solution. The hybrid materials have also shown high efficiency in treating different species, such as methylmercury. Similar loading capacities have been obtained for mercury ions (Hg2+ in mercury nitrate) and methylmercury, one of the most toxic forms of mercury. Methylmercury exists in the environment through methylation of mercury by methanogenic bacteria that are widely distributed in the sediments of ponds and in the sludge of sewage beds. A very small amount of methylmercury can be fatal to the human body.
Figure 16.13 Hg concentration remaining in the solution as a function of time when the functionalized mesoporous silica was used as the sorbent, and the comparison with commercial materials. (Liu, et al., 1998, with permission from Wiley-VCH)
240
Another application is arsenate removal (Fryxell, et al., 1999). Recent reports of the crisis caused by arsenic contamination of drinking water in Bangladesh and other parts of the world have attracted wide public attention (Nickson, et al., 1998; http, 1999). In Bangladesh alone, health officials estimated 50 to 70 million people could be affected by drinking water contaminated by natural arsenic sources. Arsenic, along with other toxic metals like chromium and selenium, are included in the U.S. Environmental Protection Agency's list of priority pollutants. These contaminating species, unlike many heavy metals and transition metals, can exist in nature as tetrahedral oxyanions (arsenate ions HAsO2-
4, H2AsO1-4, and chromate ions HCrO-
4, CrO2-4)
(Baes and Mesmer, 1976). In many cases, trace amounts of arsenate and chromate need to be removed from waste solutions containing high concentrations of competing anions, sulfate, and chloride in particular.
Currently, the development of effective anion binding materials is an important subject in chemistry, biochemistry and materials and environmental science (Woolson, 1983; Atwood, et al., 1996). Most of the anion treatment technologies are not very effective. We synthesized and used metal chelated ligands immobilized on mesoporous silica as an efficient anion binding material for both arsenate and chromate. The mesoporous silica was functionalized with an ethylenediamine (EDA) terminated silane [(2 aminoethyl)-3-aminopropyl trimethyl silane]. Cu(II) ions were binded to the EDA monolayer with a 3 to 1 EDA to Cu ratio, forming an approximately octahedral Cu(EDA)3 complex structure. This complex structure, generated from computer modeling using the PM3 and PM3™ Hamiltonians (Stewart and stewart, 1989 and 1990), as implemented in PCSpartan 5.1 (Wavefunction, Inc., Von Carman Ave, Irvine, CA), is shown in Fig. 16.14. The cationic octahedral complex contains an electrophilic basket with C3 symmetry that forms an ideal host for a tetrahedral anion. The adsorption isotherms for removing arsenate and chromate from contaminated water are plotted in Fig. 16.15. Nearly complete removal of arsenate and chromate has been achieved in the presence of competing anions for solutions containing up to 100 ppm toxic metal anions under a variety of experimental conditions. Good selectivity between chromate (or arsenate) and sulfate ions can be achieved at high anion concentrations. Anion loading is more than 120 mg (anion)/g of adsorption materials. The anion loading capacity of this material is comparable (on a molar basis) to the heavy. metal loading capacity achieved with the best cation sorbent materials (functionalized mesoporous silica) discussed earlier, when the stoichiometry of binding and the atomic/molecular weight of the target species are taken into consideration. This approach is especially promising considering the rich chemistry that can be explored with monolayers (Whitesides, 1995; Ulman, 1996; Schierbaum, et al., 1994), with mesoporous silica, and the possibility of designing better anion recognition ligands.
Figure 16.14 Cage structures of EDA-Cu complex generated by computer modeling. The C3 symmetry is ideal for tetrahedral anions. (Fryxell, et al., 1999, with permission from ACS)
241
Figure 16.15 Adsorption isotherms of chromate and arsenate using EDA-Cu modified mesoporous materials as the sorbents. (Fryxell, et al., 1999, with permission from ACS)
242
16.3.2.3 Application for Catalysis
Finally, one area of great importance is found in using mesoporous materials as catalysts or catalyst supports. The high surface area and the uniform pore channels make the mesoporous materials ideal catalyst supports. The key is the dispersion and stabilization of the catalyst on the support. We will use the preparation of solid-acid catalyst as an example (Choi, et al., 1999). Traditionally, HF and H2SO4 liquids are used as homogeneous catalysts for the commercial alkylation process (Misono and Okuhara, 1993; Corma and Matinez, 1993). The corrosive and toxic nature of such chemicals presents a problem for the environment and for the operation. One solution to this problem is to replace the liquid acids with solid acid catalysts, such as tungstophosphoric acid (H3PW12O40, TPA) (Misono and Norjiri, 1990; Corma, 1995; Okuhara, et al., 1996), which is a stronger acid than 100% H2SO4. The proton in TPA can be partially substituted with Cs+ to form Cs-TPA, which improves the thermal stability (Okuhara, et al., 1992; Soled, et al., 1997). However, it is difficult to disperse and stabilize TPA on high surface area mesoporous silica. The TEM image in Fig. 16.16(a) shows that simple impregnation technique just produced a material in which the Cs-TPA catalyst was phase segregated from the mesoporous support, therefore defeating the purpose of using a high surface area support. We also used a modified grafting technique, which produced a more homogeneous catalyst material, as shown in Fig. 16.16(b).
Figure 16.16 TEM images of Cs-TPA dispersed on mesoporous silica. (a) Phase segregated Cs-TPA. (b) Dispersed Cs-TPA (Choi, et al., 1999).
The catalytic properties of the two materials corresponding to Fig. 16.16(a) and 16.16(b) are plotted in Fig. 16.17. The catalytic activity was evaluated using alkylation of trimethylbenzene (mesitylene) by cyclohexene as a model reaction. The conversion percentage is plotted as a function of Cs stoichiometry. Clearly, the uniformly dispersed catalyst (Fig. 16.17(a)) gave much higher conversions.
243
Figure 16.17 Conversions of tremethylbenzene by cyclohexene as a function of Cs stoichiometry. (1) Dispersed catalyst; (2) Phase segregated catalyst (Choi, et al., 1999).
16.4 Summary and Prospects
Molecularly directed synthesis of nanophase materials and nanocomposites have shown great promise as a new approach to prepare a wide range of organized nanoscale structures and functional materials. Novel materials, novel applications, and more sophisticated functional nanocomposites, will arrive with a better understanding of the role of the directing molecules, and with the increased ability to take advantage of more complicated organized structures and processes.
References
Anthony R. G., C. V. Philips, and R. G. Dosch. Waste Management. 13, 503, (1993) Antonelli D. M. and J. Y. Ying. Angew. Chem. Int. Ed. Engl.. 34, 2014, 1995 (1995) Attard G. A., J. C. Glyde, and C. G. Goltner. Nature. 378, 366 (1995) Atwood J. L., K. T. Holman, J. W. Steed. Chem. Commun. 1401 (1996) Baes C. F. Jr., R. E. Mesmer. The Hydrolysis of Cations. John Wiley & Sons: New York, (1976) 215 and 366–368 Bagshaw S. A., E. Prouzet, and T. J. Pinnavia. Science. 267, 865 (1995)
244
Baskaran S., L. Song, J. Liu, Y. L. Chen, and G. L. Graff. J. Am. Ceram. Soc.. 81, 401 (1998) Beck J. S., J. C. Vartuli, R. J. Roth, M. E. Leonowicz, C. T. Kresgec. T-W. Chu, and D. H. Olson. J. Am. Chem. Soc.. 114, 10834 (1992) Beck J. S. and J. C. Vartuli. Cur. Opin. Sol. St. Mater. Sci.. 1, 76 (1996) Belcher A. M., X. H. Wu, R. J. Christensen, P. K. Hansma, G. D. Stucky, and D. E. Morse. Nature. 381, 56 (1996) Braun P. V., P. Osenar, and S. I. Stupp. Nature. 380, 325 (1996) Bunker B. C., P. C. Rieke, B. J. Tarasevich, A. A. Campbell, G. E. Fryxell, G. L. Graff, L. Song, J. Liu, J. W. Virden, and G. L. McVay. Science. 264, 48 (1994) Cariolu M., D. E. Morse. J. Comp. Bio.. B 157, 717 (1988) Chambers A., C. Park, R. T. K. Baker. J. Phys. Chem.. B 102, 4253 (1998) Chen P., X. Wu, J. Lin, K. L. Tan. Science. 285, 5424 (1999) Choi S., Y. Wang, Z. Nie, J. Liu, and C. H. F. Peden. Catalyst Today. in press (1999) Conway B. E. J. Electrochem. Soc.. 138, 1539 (1991) Corma A. and A. Matinez. Catal. Rev.-Sci. Eng.. 35, 483 (1993) Corma A. Chem. Rev.. 95, 559 (1995) Currey J. D. J. Mater. Edu.. 9, 118 (1987) Delnick F. M., M. Tomkiewicz. Electrochemical Capacitors. The Electrochemcial Society. Pennington (1996) Dillon A. C., K. M. Jones, T. A. Bekkedahl, C. H. Kiang, D. S. Bethuune, M. J. Heben. Nature. 386, 377 (1997) Dubois M. Th. Gulic-Krzywicki, and B. Cabane. Langmuir. 9, 673 (1993) Evans D. F. and H. Wennerstrom. The Colloidal Domain, Where Physics, Chemistry, Biology, and Technology Meet. VHC Publishers Inc, New York, (1994) Chapter 2 Feng X., G. E. Fryxell, L. Q. Wang, A. Y. Kim, K. Kemner, and J. Liu. Science. 276, 923 (1997) Firouzi A., D. Kumar, L. M. Bull, T. Besier, P. Sieger, Q. Huo, S. A. Walker, J. A. Zasadzinski, C. Glinka, J. Nicol, D. Margolese, G. D. Stucky, and B. F. Chmelka. Science. 267, 1138 (1995) Fryxell G. E., J. Liu, T. A. Hauser, Z. Nie, K. F. Feris, S. Mattigod. Chem. Mater.. 11, 2148 (1999) Gray D. H., S. Hu, E. Juang, and D. L. Gin. Advanced Materials. 9, 731 (1997) Hnatyszyn H. J., N. Kossovsky, A. Gelman, E. Sponsler. PDA J. Pharmaceutical Sciences & Technology. 48, 247 (1994) Huo Q., R. Leon, P. M. Petroff, and G. Stucky. Science. 268, 1324 (1995) Huo Q., D. I. Margolese, U. Ciesla, P. Feng, T. E. Gier, P. Sieger, R. Leon, P. M. Petroff, F. Schuth, and G. D. Stucky. Nature. 368, 317 (1994) http://bicn.com/acic, 1999 Israelachvili J. Intermolecular & Surface Forces. 2nd Edition, Academic Press, San Diego (1991) Chapter 16, 17 and 18 Katz J. L. Biomaterials Science. eds. by B. D. Ratner, A. S. Hoffman, F. J. Schoen, and J. E. Lemons. Academic Press, San Diego (1996), 335 Klein J. E. R&D Needs for Mixed Waste Tritium Pump Oils (U). Westinghouse Savannah River Company Inter-Office Memorandum, SRT-HTS-94–0235, July 11 (1994) Kresge C. T., M. E. Leonowicz, W. J. Roth, J. C. Vartuli, and J. S. Beck, J. S. Nature. 359, 710 (1992) Liu J., A. Y. Kim, L. Q. Wang, B. J. Plamer, Y. L. Chen, P. Bruinsma, B. C. Bunker, G. J. Exarhos, G. L. Graff, P. C. Rieke, G. E. Fryxell, J. W. Virden, B. J. Tarasevich, L. A. Chick. Adv. Colloid. Interface. Sci.. 69, 131 (1996) Liu J., K. E. Gunnison, M. Sarikaya, and I. A. Aksay. Proceedings of 50th Annual Meeting of the Electron Microscopy Society of America. ed. by Bailey G. W., J. Bentley, and J. Small, San Francisco Press, San Francisco (1992), 1024 Liu J., X. Feng, G. E. Fryxell, L. Q. Wang A. Y. Kim, and M. Gong. Advanced Materials. 10, 161 (1998) Mercier L. and T. J. Pinnavaia. Advanced Materials. 9, 500 (1997) Mercier L. and T. J. Pinnavaia. Envir. Sci. Technol.. 32, 2749 (1998)
245
Mirkin C. A., R. L. Letsinger, R. C. Mucic, and J. J. Storhoff. Nature. 382, 607 (1996) Misono M. and T. Okuhara. ChemTech. 11, 23 (1993) Misono M. and N. Norjiri. Appl. Catal.. 64, 1 (1990) Mitra S. Mercury in the Ecosystem. Trans Tech Publications, Lancaster, PA (1986) Moller K. and T. Bein. Chem. Mater.. 10, 2950 (1998) Monnier A., F. Schuth, Q. Huo, D. Kumar, D. Margolese, R. S. Maxwell, G. D. Stucky, M. Krishnamurty, P. Petroff, A. Firouzi, M. Janicke, and B. F. Chmelka. Science. 261, 1299 (1993) Nakahara H., M. Kakei, and G. Bevelander. Jpn. J. Malacol. 39, 167 (1980) Newnham R. E. Adv. Ceram. Mater.. 3, 12 (1988) Nickson R., J. McArthur, W. Burgess, K. M. Ahmed, P. Ravenscroft, M. Rahman. Nature. 395, 338 (1998) Okuhara T., N. Mizuno, and M. Misono. Advances in Catl.. 41, 113 (1996) Okuhara T., T. Nishimura, H. Watanable and M. Misono. J. Mol. Catal 74, 247 (1992) Raman N. K., M. T. Anderson, C. J. Brinker. Chem. Mater.. 8, 1682 (1996) Rieke P. C., B. D. Marsh, L. L. Wood, B. J. Tarasevich, J. Liu, L. Song, and G. E. Fryxell. Langmuir. 11, 318 (1995) Sarikaya M., J. Liu, and I. A. Aksay. Biomimetics: Design and Processing of Materials. eds. by M. Sarikay and I. A. Aksay, AIP Press, Woodbury (1995) Sayari A. Chem. Mater.. 8, 1840 (1996) Schierbaum K. D., T. Weiss, E. U. Thoden van Velzen, J. F. J. Engbersen, D. N. Reinhoudt, W. Gopel, W. Science. 265, 1413 (1994) Simkiss K. and K. M. Wilbur. Biomineralization. Academic Press, New York (1989) Soled S., S. Miseo, G. McVicker, W. E. Gates, A. Gutierrez and J. Paes. Catal. Today. 36, 441 (1997) Stewart J. J., P. J. Stewart. Comp. Chem.. 10, 221 (1989) Stewart J. J., P. J. Stewart. Computer Aided Molecular Design. 6, 69 (1990) Tanev P. T. and T. J. Pinnavaia. Science. 267, 865 (1995) Tarasevich B. J., P. C. Rieke, and J. Liu. Chem. Mater.. 2, 201 (1996) Tiddy G. J. T. Physics Report. 57, 1 (1980) Ulman A. Chem. Rev.. 96, 1533 (1996) Vinson P. K., J. R. Bellare, H. T. Davis, W. G. Miller, and L. E. Scriven. J. Coll. Int. Sci.. 142, 74 (1991) Weiner S., Y. Talmon, and W. Trab. Int. J. Biol. Macromol. 5, 325 (1983) Weiner S. and W. Traub. Philos. Trans. R. Soc. London Ser. B 304, 421 (1984) Whitesides G. M. Sci. Am.. 273, 146 (1995) Williams G. W. and G. S. V. Coles. MRS Bulletin. 24, 25 (1999) Woolson E. A. Biological and Environmental Effects of Arsenic, ed. by B. A. Fowler, Elsevier, New York, (1983), 51–120 Yang P., D. Zhao, D. I. Margolese, B. F. Chmelka, and G. D. Stucky. Nature. 396, 152 (1998) Zaremba M., A. M. Belcher, M. Fritz, Y. Li, S. Mann, P. K. Hansma, D. E. Morse, J. S. Speck, and G. D. Stucky. Chem. Mater.. 8, 679 (1996) Zheng J. P. and T. R. Jow. J. Electrochem. Soc.. 142, L6 (1995)
246
17, Nanostructured Bio-inspired Materials
17.1 Introduction
This chapter examines a relatively new paradigm for tailoring the mechanical properties of composite materials. As has often happened throughout history, biological systems have been the inspiration. Recent studies of biological ceramic composites (bone and teeth, mollusk shells, etc.) have generated a number of new ideas on how structure can affect material properties (Weiner and wagner, 1998; Weiner and Addadi, 1997; Aksay and Weiner, 1998; Wang, et al., 1997; Wang, 1998; Berman, et al., 1993, 1990, 1988; Smith, et al., 1999).
The most important idea to emerge, and the focus of much of this chapter, is the notion of hierarchical structure. Simply stated, hierarchical structure refers to the existence of structural organization at multiple length scales (often nanometer to micrometer, to millimeter) within a given material. Examples of such structures are shown in Fig. 17.1. Of course, all materials possess some degree of organized structure at multiple length scales; the key is that this organization occurs in such a way as to cooperatively enhance a given property. In the case of biological ceramic composites, organic phase controls the structure of the mineral phase at multiple length scales in a manner that leads to dramatic improvements in the mechanical toughness over bulk mineral.
Figure 17.1 Examples of biological structures that have inspired new materials, with (a) abalone, (b) teeth, and (c) radiolaria.
The nacreous layer of mollusk shells is one example of this phenomenon. Mollusk shells are mostly CaCO3 and about 5 wt% protein. Nacre has a fracture toughness of about three orders of magnitude higher than that of monolithic aragonite (a type of CaCO3). Nacre owes its fracture strength and toughness to its composite organic/inorganic microstructure.
Detailed studies of the structure of nacre have provided clues for materials scientists hoping to mimic nacre's microstructure in fracture-resistant synthetic materials. A fractured abalone shell exposes a brick and mortar arrangement. The mortar is made out of polysaccharide and protein fiber saligned orthogonal to each
247
other and parallel to the aragonite crystal axes. The bricks of highly uniform aragonite (thickness 0.5 µm) in interdigitated columns (Fig. 17.2(a)). The cross-sectional transmission electron image (TEM) reveals organic mortar layers, approximately 10 nm thick, which separate the aragonite lamellae (Fig. 17.2(b)). Typical interfaces between two phases are not atomically smooth; the aragonite phase is composed of regular single crystalline CaCO3 structure, and the organic phase is composed of a mixture of protein fibers and nano-aragonite inclusions (Fig. 17.2(c)). A mortar layer, continuous with the interlamella sheets separates laterally adjacent tablets. Furthermore, organic inclusions, roughly round in cross section, can be seen within the mineral tablets. These observations hint at some mechanism for the sequestering of soluble organic material within the aragonitic crystals. This mechanism may be responsible for the nucleation and growth of the crystal.
Figure 17.2 (a) SEM image of abalone nacre exhibiting interdigitated stacks of aragonite tablets; (b) Cross-sectional TEM image of nacre. The curved lateral boundary between coplanar tablets can also be seen. (c) High-resolution TEM mage of interface between brick and mortar phases.
The mineralized collagen tissue is another natural example of improved properties due to hierarchical structure. Although members of this family of materials (i.e., bone, dentin, cementum, and mineralized turkey tendon) are constructed from the same building blocks, the mineralized collagen fibril, they have mechanical properties suited to their specific functions. The variation in mechanical properties is due to differences in the way the collagen fibrils are assembled—in short, due to the various hierarchical structures (Weiner and Wagner, 1998). These materials share one common feature, high toughness, which results from the well-ordered composite nature of mineral and the collagen matrix, which are ordered at the nanometer level.
The development of hierarchical structure is still an area of active research. Although many of the details are still unclear, a few general principles are commonly accepted. One of these is the idea of templating. In biological ceramic composites, the organic phase (typically proteins) serves to direct the structure of (template) the mineral phase. This templating effect occurs at both molecular length scales (where the protein serves as a nucleation site) and higher (the nucleation sites are arranged in a larger framework which imposes organization at larger length scales). Efforts to develop and manipulate artificial systems with
248
hierarchical order have shown that applied fields further impose or augment structural control at multiple length scales in organic-inorganic systems (e.g., mesoscopic silica)(Trau, et al., 1997; Firouzi, et al. 1997; Hillhouse, et al., 1999).
Biological macromolecules (usually acidic) play a key role in the templating process. They can serve to initiate or halt the growth of the crystalline mineral phase. Early in the growth, these macromolecules control the nucleation and orientation of a specific mineral. In later stages, the texture and organization of crystals are controlled by the organic phase. Growth is terminated by the selective adsorption of macromolecules onto crystal surfaces or by impingement of the crystal on a preformed organic framework (Weiner and Addadi, 1997). There is a high degree of specificity in the nucleation and inhibition processes. It was found that acidic macromolecules extracted from the nacre were capable of inducing aragonite crystallization, while macromolecules extracted from the calcitic layer of the same shell only induced calcite formation. In addition to chemical control, the organic phase can also affect the mineral structure by forming a physical framework. A good example of this effect is collagen in bone. Type I collagen forms a 3-D ordered framework for mineral formation. Inside each collagen fibril, mineral growth is guided by aligned grooves between many triple helical molecules. The oriented growth of Ca-P minerals within a collagen framework has also been successfully demonstrated by in vitro experiments (Wang and Weiner, 1998).
The insights gained from the study of natural systems point to two general principles for the design of synthetic high-performance composite materials. First, cooperative structural control at multiple length scales is desirable. Although our understanding of how to specifically structure a material is still embryonic, there is no question that the structure of successful synthetic analogues of biological composites will need to be closely controlled. Methods for simultaneously controlling the structure from atomic to macroscopic length scales still need refinement. Second, the idea of templating through a ternary phase (usually organic) is a flexible and effective way to manipulate structure of materials at multiple length scales.
The remainder of this chapter is devoted to case studies of two hierarchically structured materials grown through a templating process. In the first case study, we take a closer look at a naturally occurring material—–teeth, specifically dentin. Special attention is paid to the hierarchical structure and its resulting improvements in mechanical properties. We will also illustrate how the ideas of templating and directed growth emerged from studies of tooth growth and structure. In the second case study, we will discuss recent efforts for applying some of the insights gained. One synthetic system that has and continues to receive much attention is mesoporous silica films. These hierarchically structured composites are formed using some of the principles described above. Although we are not yet at a point where we can tailor the mechanical properties of these materials, the fact that we are able to apply some of the principles of hierarchical ordering in an artificial system is a promising step in the right direction.
17.2 Case Study I: Teeth
In this case study, we will examine the concepts of hierarchical structure and the role of organic species in templating in the context of dentin. Dentin fulfills a vital function in many vertebrates, as well as
249
invertebrates. Its structure is presumably fine-tuned to specific mechanical requirements—the cutting, crushing and grinding of food. Studies of the structure of teeth and its relationship to mechanical performance at various length scales provide invaluable opportunities to relate function to structure. For detailed information on other biological materials, readers are encouraged to read recent articles and reviews (Weiner and Wagner, 1998; Weiner and Addadi, 1997; Aksay and Weiner, 1998).
Most vertebrate teeth, with the notable exception of fish, have the same basic structural pattern. They are composed of a very dense outer enamel layer, which is present only on the working surface, and a less dense but generally thicker inner layer of dentin (Fig. 17.3) (Ten Cate, 1994; Carlson, 1990). Enamel is hard but brittle; it is designed for cutting and grinding food. The supporting dentin is relatively soft, yet very tough. It effectively transfers stress from the outer enamel layer into the jawbone, while absorbing part of the working energy. Clinically, microcracks are always found on the enamel surface of adult human teeth, but they usually do not penetrate into dentin. This is an indicator of the mechanical function of dentin. Studies on dentin have revealed that its mechanical toughness results from two features: precise control over mineralization at the nanometer scale and the presence of cooperative hierarchical structures.
Figure 17.3 Labio-lingual section of a human incisor.
17.2.1 Control over Mineralization at Nanometer Scale
250
Dentin is a member of a family of materials, including the various types of bone. These materials all have the same structural building block—the mineralized collagen fibril (Weiner, et al., 1998). These fibrils contain about 70 wt% carbonate apatite minerals (Ca5(PO4, CO3)3(OH)), 20 wt% organic materials, and 10 wt% water (Wang, et al., 1998). Collagen (Type I) is one of the main components of the three-dimensional matrix into (or onto) which the mineral crystal forms an ordered structure (Fig. 17.4). The apatite crystals in mineralized collagen fibrils are plate-shaped, with average lengths and widths of 50 nm×25 nm and a thickness of 1.5–4 nm (Weiner and Wagner, et al., 1998; Lowenstam, et al., 1989; Hohansen, et al., 1962). The hexagonal axis is always perpendicular to the plate surface, and the c-axis is parallel to the long axis of the plate. Furthermore, crystals of mineralized collagen fibrils are aligned with the crystallographic c-axis parallel to the collagen fibril axis and the plate surfaces are parallel to each other (Figs. 17.4 and 17.5)
Figure 17.4 3-D organization of apatite minerals inside collagen fibrils at nanometer scale. (a) Collagen molecules (represented by rods) aggregate into a quarter-staggered pattern described by Hodge and Petruska, (1963). (b) 3-D model showing the alignment of the holes to form a channel (Yamauch, et al., 1989); (c) Models showing the proposed location of the platy crystals in the channels. (Traub, et al., 1989)
Figure 17.5 A twisted mineralized turkey tendon fibril in vitrous ice showing the apatite mineral plates both edge-on (B) and face-on (A). (Veis, 1989), (Bar = 0.2 µm).
251
Many non-collageneous proteins (less than 10% of the total proteins) exist between the collagen and crystals as well.
The orthotropic structure of mineralized collagen fibrils shown in Fig. 17.4 directly results in the anisotropic mechanical properties at higher levels of structure. In mineralized turkey leg tendon, the elastic modulus is the highest in the direction parallel to the collagen fibril axis and is the lowest in the orthogonal direction (Lees, et al., 1992). In parallel-fibered bone, microhardness is highest on the plane perpendicular to the collagen fibril axis, where the crystal plates are oriented edge-on, and is lowest on the plane parallel to the collagen fibrils where the plates are face-on (Ziv, et al., 1996).
The above evidence strongly suggests that both the nucleation and growth of the minerals have been well controlled in this biological system, and that the enhanced mechanical properties originate at the nanometer level where the minerals are organized within the matrix frame. The basic mechanisms underlying the genetic control over crystal precipitation are the main focus of biomineralization (Lowenstam, et al., 1989), a research field now closely linking biology to materials science.
252
A number of biomineralization studies have found that biological macro-molecules are the key to controlling mineralization. These macromolecules can be divided into two main groups (Weiner, et al., 1997), "control macromolecules" and "framework macromolecules". Both types of macromolecules are present at the nanometer level. They mediate mineral growth in different ways. Control macromolecules usually contain charged groups (e.g., carboxylate groups, phosphate or sulfate groups) and closely interact with the mineral ions in solution or with the surfaces of the solid phase. Numerous in vitro studies have indicated that mineralization and control over crystal growth are accomplished through these molecules, typically through electrostatic interactions (Veis, 1989; Addadi, et al., 1992). Phosphorated proteins are control macromolecules in the dentin of teeth (Veis, 1989).
The "framework macromolecules" constitute the bulk of the organic matrix in biological materials. They are more hydrophobic, often crosslinked, and relatively insoluble. The main function of these macromolecules is to provide a three-dimensional matrix from which the control proteins interact with the mineral phase (Weiner, et al., 1997; Wang, 1998). In dentin and other mineralized collagen tissues, type I collagen forms a three-dimensional ordered framework for mineral formation. Inside each collagen fibril, many triple helical molecules, which are composed of two α1 polypeptide chains and one α2 chain, are arranged in a staggered array. A gap (or hole) exists between the NH2-terminus of one triple helical molecule and the COOH-terminus of the next (Fig. 17.4 (a)). These holes are aligned to form transversely contiguous grooves, which are one molecule thick ( 1.5 nm), 37 nm high in the direction of the fibril axis, and extend for some unknown distance across the fibril (Fig. 17.4 (b)). These aligned grooves provide the nucleation sites (with the presence of acidic noncollagenous proteins) of minerals and guide the growth of plate-shaped minerals (Fig. 17.4 (c)). The oriented growth of Ca-P minerals within collagen framework has also been successfully demonstrated by in vitro experiments (Iijima, et al., 1997).
17.2.2 Hierarchical Structure in Biological Materials
In much the same way a building is organized at the multiple length scales of bricks, walls, rooms and finally the whole structure, biological materials have distinct levels of organization. Low levels of structure (smaller length scales) determine the fundamental properties, which distinguish one family of materials from another. High levels of structure (larger length scales) provide properties that make each member within a family unique.
Weiner and Wagner have described seven hierarchical levels of organization in the bone family, mainly through illustrating the structure of bone (Weiner, et al., 1998). The corresponding features in dentin are described here. The first two levels (the molecular components and the collagen fibril) have been described above. Those two levels are common in dentin, cementum, bone and the mineralized tendon. The collagen fibrils are a clear example of the composite nature of bone family (Fig. 17.5). They contribute to the well-known toughness of bone-like materials, and at the same time it gives rise to anisotropy, which is overcome at higher levels of organization.
Bundles of collagen fibrils comprise the third level of structure. The collagen fibrils in dentin form small bundles, within which the fibrils are aligned to each other only with their fibril axes. Each mineralized collagen fibril is rotated around this axis in such a way that the mineral plates from adjacent fibrils often have different azimuthal orientation (Wang, et al., 1998). The resulting structure has transversal isotropy.
253
The collagen bundles at the "fourth" level of organization are laid down, surrounding dentinal tubules, in the planes parallel to the surface at which dentin formation takes place (Fig. 17.6 (a)). Within these incremental planes, the fibril bundles are either randomly distributed (coronal dentin) or poorly oriented (root dentin) (Wang, et al., 1998). It is these two levels of organization that differentiate dentin from bone and other mineralized collagen tissues. The layered structure results in much higher cracking resistance in the direction perpendicular to the incremental planes than parallel to them (Wang, et al., 1998). At the same time, the random orientation within layers and the transversal isotropy of the bundles produce isotropy in Vickers microhardness, which is closely related to elasticity (Wang, et al., 1998).
Figure 17.6 Ultrastructure of (a) cow root dentin and (b) coronal dentin.
At the fifth level, the relative content of the mineral and the density of the dentinal tubules are systematically changed from central pulp cavity side of a tooth to the peripheral sides. Dentin immediately underneath enamel has much lower mineral content and is softer than the central part of the dentin (Fig. 17.7). This "design" of mineral gradient could protect the dentin-enamel junction from failure and facilitate the even transfer of stress from enamel to dentin. At the sixth level of organization the difference between two types of dentin (root dentin and coronal dentin) can be seen. A dense, highly mineralized peritubular dentin forms a cylinder surrounding each dentinal tubule in the coronal dentin (Fig. 17.6 (b)). As a result,
254
coronal dentin has higher elastic modulus than root dentin. It is also more isotropic in fracture properties than root dentin, as shown by indentation studies (Wang, et al., 1998).
Figure 17.7 Backscattering scanning electron image of a section from a human premolar. The upper white part is enamel, and the lower gray side is dentin. The dark zone next to enamel showing lower mineral content in that dentin area. (Wang and Weiner, 1998)
The combination of dentin and enamel, their relative thickness, and the size and geometry of a whole tooth construct the highest level of the hierarchical structure (Fig. 17.3). This is the level where different types of teeth (i.e., incisors, premolars, molars, canines, etc.) and teeth of different species come to function.
17.3 Case Study II: Mesoscopic Silica Films
In this case study, we will describe one way in which some of the concepts from biological composites have been used in the design of synthetic materials. A recent class of materials, the mesoporous silicates, in which amphiphilic molecules are used to control the growth and structure of an inorganic matrix, will be examined.
In 1992, researchers at Mobil Research reported the successful synthesis of silica molecular sieves, designated M41S, with variable, but monodisperse, pore sizes (1.5 nm to 10 nm) and morphologies (hexagonal, cubic and lamella) (Kresge, et al., 1992). They prepared this structure using an inorganic/organic templating mechanism, where silica condenses around a surfactant micelle mesophase (Fig. 17.8).
255
Figure 17.8 Organic-inorganic templating: (a) schematic of templating; (b) morphologies: cubic, hexagonal, lamellar; (c) pore size control by inserting organic spacer molecules.
Subsequent studies have shown that pore size and packing can be adjusted by varying the surfactant or introducing additional species in the reacting solution and that the pore morphology of the silicate product is closely linked to the structure of the surfactant mesophase (Vartuli, et al., 1994; Beck, et al., 1994). Researchers at the University of California at Santa Barbara demonstrated that the synthesis of mesoscopic ceramics could be generalized by categorizing them according to the electrostatic interactions between the organic and inorganic molecules (Huo, et al., 1994; 1996). Later, they showed that neutral surfactant species (Attard, et al., 1995), as well as amphiphilic block copolymers, could be used as the structure determining agents (Zhao, et al., 1998a). Under most synthesis conditions, the concentrations of amphiphiles are too low to produce the observed channel packing. Thus, the UCSB group concluded that the inorganic species plays a non-trivial role in determining the final structure. "Co-assembly" is the term applied to this situation to distinguish it from direct templating, where the inorganic infiltrates and solidifies a pre-existing ordered amphiphile array.
The first mesoporous materials prepared were micron-sized particles. A number of research groups have shown that other macroscopic morphologies are possible. Researchers at Princeton, Sandia National Laboratory, UCSB and Toronto were among the first to report thin films, fibers, and other forms of mesoporous silica. Thin films have been grown using direct templating and dip-coating methods (Aksay, et al., 1996; Lu, et al., 1997; Trau, et al., 1997), on materials, such as graphite, mica, silica, gold overlain with patterned organic thin films and the air-water interface (Ryoo, et al., 1997; Yang, et al., 1996; Yang, et al., 1997). Fibers, tubes, spheres, gyroids and other shapes have resulted from modifications of the solution composition (Lin, 1996; Yang, et al., 1997).
Mesoporous silica is currently an extremely active area of research. Some research groups have concentrated their efforts on understanding and extending the templating process, others have opted to move toward commercial applications by derivatizing the molecular sieves. Several comprehensive reviews have been published describing these recent directions and results (Maschmeyer, 1998; Ozin, et al., 1999; Zhao, et al., 1998b). In the remainder of this section, we will describe the hierarchical structure observed in
256
mesoscopic silica thin films grown in our lab under acidic conditions and illustrate how to achieve hierarchical structural control.
17.3.1 Hierarchical Film Structure
The biological principle of organic-directed mineral growth is operative, albeit in a simplified manner, in the growth of mesoscopic silica thin films. Although the manner in which structure is controlled may not be as elegant as protein-directed dentin growth, the success in applying the biologically inspired principle of organically directed mineralization is an encouraging result. The purpose of this section is to document the hierarchical order present in this entirely synthetic system. The exact mechanism by which the surfactant affects structure at lengths larger than the channels is not completely understood. The films described in this section were grown at room temperature under quiescent conditions in a dilute acidic (pH<2) solution containing surfactant (cetyltrimethylammonium chloride, or CTAC) and an alkoxysilane precursor (tetraethylorthosilicate, or TEOS). Films appeared at the air-water interface, along the sides of the container and on any substrates included in the container. Films grown on solid surfaces are described at length elsewhere (Aksay, et al., 1996; Yang, et al., 1996, 1997).
17.3.1.1 Nanometer Level Structure
At the nanometer level, silica thin films formed in this manner possess regularly packed channels. Both XRD and TEM analysis show that mesoscopic materials have channels that are packed along a hexagonal lattice with an average channel spacing of about 4 nm. The channel diameter corresponds to the approximate diameter of a surfactant micelle. This suggests that the surfactant plays a critical role in initiating film growth as well as directing mesoscopic structure. Indeed, if no surfactant is added to the system, the resulting material lacks channels.
Channel size and organization can be controlled in a number of ways (Vartuli, et al., 1994; Huo, et al., 1994; Zhao, et al., 1997, Aksay, et al., 1996). As might be expected, increasing the organic tail of the surfactant leads to larger pore diameters. The same effect can be accomplished by introducing ternary organic components which "swell" the micelles. The use of amphiphilic block copolymers allows materials with even larger pores to be made. The channel organization can be manipulated using similar guiding principles. For example, it has been found that the addition of cosolvents can cause a system of hexagonally packed channels to adopt a cubic or lamellar configuration. Qualitatively, the surfactant system responds in the same way as a liquid crystal; however, the development of a rigorous method of predicting channel organization based on the initial composition of the system is still being actively pursued.
There is also interest in controlling the long-range orientation of the channels. In some cases, the packing organization dictates the local orientation. For example, hexagonally packed channels are spaced according to a regular lattice. This, however, still allows for the meandering of channels along their axes. That is, an entire bundle of channels may bend in unison, maintaining regular packing while at the same time having poor long-range order.
257
One interesting way to influence the long-range channel orientation is by introducing an interface. Typical TEM images of a mesostructured silica film grown on mica are shown in Fig. 17.9. Both images are in a transverse orientation with respect to the film and reveal hexagonal packing of tubules aligned parallel to the substrate. Figure 17.9 (a) reveals a slight elliptical distortion of the tubules suggesting that the films are strained, i. e., compressed in the direction normal to the template. Figure 17.10 shows a TEM planar cross section of a film grown on silica. In contrast to those shown in Fig. 17.9, this image exhibits a spiraling and twisting arrangement of surfactant tubules.
Figure 17.9 Cross-sectional TEM images of the final film interior, perpendicular to the film surface. Side view (a) and head-on (b) images are shown. Inset: SAED patterns indicating regular hexagonal packing and parallel channels. The channels in Fig. 17.7 (a) show a -5% to 10% strain in both the parallel and perpendicular directions. These values suggest that the film is not highly distorted.
Figure 17.10 TEM image of a planar cross section of a film grown on silica. This section was taken through a macroscopic swirl similar to those shown in Fig. 17.12 (c).
Figure 17.11 In situ AFM images of mesostructured films growing on mica, graphite, and amorphous silica substrates, respectively. AFM images of the mica, graphite, and silica substrates used to grow mesoscopic silica films are shown in the insets. (a) and (b) illustrate the periodic mica and graphite atomic lattice onto which CTAC adsorb and orient, and (c) reveals a smooth, amorphous silica substrate. Images of the films
258
were obtained in "non-contact" mode, utilizing the electrical double layer force described (Manne, et al., 1994; Manne and Gaub, 1995; Senden, et al., 1994). (a) reveals meandering surfactant tubules on the mica substrate, 6.2–6.8 nm spacing, oriented parallel to the solid/liquid interface. Tubules are initially aligned along one of the three next-nearest-neighbor directions of the mica oxygen lattice displayed in inset. In the early stages of the reaction (<7 h), this orientation is preserved as tubules continue to assemble and grow away from the interface coupled with silica polymerization. On graphite, tubules align parallel to the substrate, along one of three symmetry axes of the hexagonal carbon lattice shown in inset. Unlike the mica case, these do not meander, however, form rigid parallel stripes. (c) On amorphous silica, periodic dimples are observed rather than stripes, suggesting an orientation of the tubules away from the interface.
259
Figure 17.12 SEM images of mesoscopic silica films grown at (a) mica/water, (b) graphite/water, (c) silica/water interface for 24 h, respectively. Oriented tapes are observed on mica and graphite. The films grown at the silica/water interface are uniform initially (dark background) but spiral like structures (light features) form later.
260
This behavior can be explained by considering the structure of the micelle layer that is adsorbed at each interface (Aksay, et al., 1996). Although the molecular organization and self-assembly of surfactants at
261
interfaces is a widely studied area, little is known about the precise structure of adsorbed surfactant layers. Recent work has shown that 3-D surfactant structures such as cylindrical tubules and spheres can form at solid/liquid interfaces (Manne, et al., 1995). Adsorbed hemi-micellar arrangements are observed on poorly orienting amorphous substrates, such as silica, and aligned tubular structures are observed on more strongly orienting crystalline substrates such as mica and graphite. The latter substrates orient adsorbed surfactants via anisotropic attraction (either van der Waals or electrostatic) between the crystalline substrate and the surfactant molecule. The amorphous silica substrate has no preferential orientation for surfactant adsorption.
This argument is further supported by an elegant investigation at the early stage of film growth. Figure 17.11 shows in situ atomic force microscopy (AFM) images of the atomic lattice of each substrate as well as the structure of the mesoscopic silica overlayer growing on each surface (Senden, et al., 1994). Aksay et al, showed that there is a sequential process of film growth (Aksay, et al., 1996). The first step is the self-assembly of surfactant at the interface to form meandering tubules. Silicon hydroxide monomers (or multimers) polymerize at the micellar surface. As polymerization continues, more surfactant is adsorbed to the freshly formed inorganic surface allowing the templated mesoscopic structure to replicate itself and grow into the bulk solution. Surfactant tubules appear to register the underlying lattice of the substrate. Two directions are preferred for surfactant micelles adsorbed on mica (Fig. 17.12 (a)) owing to its distorted hexagonal structure. On graphite substrates (Fig. 17.12 (b)) the surfactant tubules are also aligned parallel to the surface; however, in this case they are rigid, parallel stripes without the meandering curvature observed on mica. Micelles at silica interfaces (Fig. 17.12 (c)) exhibit dramatically different orientations. The weakly interacting silica surface may be insufficient to impose long-range orientational order on the micelle assemblies. More recently, Yao et al. reported a detailed study of the mesoscopic (nanometer level) and microscopic (micron-level) structural evolution of mesoscopic silica thin films grown at the air-water interface under dilute, acidic (pH<2) conditions. This work clarified the role of the air-water interface in confining film growth to two dimensions during the initial stages and provided a detailed mechanism for the development of mesoscopic order and microscopic features and considered the possibility of a universal growth mechanism for films and particles (Yao, et al., 2000).
17.3.1.2 Micrometer-Level Structure
The principle of hierarchical order directed by organic species also manifests in the mesoscopic silica system. SEM images of the films grown on mica, graphite and silica interfaces are shown in Fig. 17.12. Films grown on mica (Fig. 17.12(a)) and graphite (Fig. 17.12(b)) exhibit preferred orientations at mesoscopic and in-plane confinement at microscopic length scales. By contrast, films grown on silica surfaces are initially uniform, but over time develop 3-D structures, most notably particle-like bundles.
The role of organic species in directing the structure at the micron level is more subtle when compared with the nanometer level. While the nanometer structure is likely the result of individual micelles, the micron structure may be the result of the grains. Micelles adsorbed onto mica and graphite surfaces appear to recognize the asymmetry in the underlying mica lattice. Further, films grown on mica and graphite seem to possess the same orientation constraints. It is remarkable that features at the length scale of atomic dimensions (mica and graphite lattice) can influence the structure at nanometer and micron length scales in such a pronounced manner.
262
17.3.2 Towards Control of the Properties
At present, the mesoscopic silica thin film is a rather simplistic analogue of naturally occurring biominerals; there is much room for improvement. Progress is expected on two fronts. First, the degree of control over the structure at all levels can be improved. Our current understanding of the mechanisms behind the development of order is still rather primitive. Much of our knowledge has come from a phenomenological approach; predictive approaches to mesoscopic silica synthesis must be based on qualitative models. Further study of the growth mechanism is required for the development of such models. Secondly, finer control of the structural features is desired. At the nanometer length scale, this includes the ability to orient the pores in arbitrary directions, produce materials with a single orientation of pores. The ability to impose arbitrary micron sized features may also be important in the development of commercial applications for mesoporous silica.
Steady progress in understanding the mechanism and controlling the growth has been made. Mobil Research and Stucky's group at UCSB developed some of the earliest ideas about templating (Kresge, et al., 1992; Huo, et al., 1994). Mechanistic studies of the system have shown the chemistry to be rather complicated. Several pathways are possible depending on the identity of the amphiphile, relative compositions of the reagents, presence of ternary components and reaction conditions. Although it is not clear at present what exact mechanistic steps are involved, two things are certain. Just as in biological systems, hierarchical order is present. Furthermore, the organic phase plays a role in determining the structure at multiple length scales.
Efforts to engineer more complicated features have generally followed two tracks. One strategy is to follow the example of nature. Orientation of the structural features is imposed through the use of a physical template. The growth of mollusk nacre ultimately occurs on the protein surfaces. Significant strides have been made in controlling the micron level morphology of mesoscopic silica materials using soft lithography, a class of techniques popularized by Whitesides' group at Harvard University (Zhao, et al., 1997).
A second strategy for controlling structure involves applying non-biological methods. The effects of electrical, magnetic and shear fields on fluids has been the subject of much experimental and theoretical work. The use of external fields, which may not be available to biological organisms, may provide additional (and perhaps more effective) ways to influence the structure. All of these approaches have been used to orient the internal pore arrangements as well as micron-sized features with varying degrees of success (Trau, et al., 1997; Firouzi, et al., 1997; Hillhouse, et al., 1999). These approaches represent cases where a synthetic processes may benefit from both bio-inspired and more "conventional" design principles.
17.4 Conclusion
Studies of naturally occurring mineral composites with mechanical properties far superior to pure minerals have revealed that cooperative hierarchical structure is one way to attain improved material properties. The design and construction of these materials is one of the most challenging and promising directions of
263
research today. Nature uses organic molecules to direct the structure and organization of minerals. Preliminary efforts to mimic this process have been successful, but much work remains.
References
Aksay I. A., M. Trau, S. Manne, I. Honma, N. Yao, L. Zhou, P. Fenter, P. M. Eisenberger and S. M. Gruner. Science. 273, 892 (1996)
Addadi L. and S. Weiner. Angrew. Chem. Int. Ed. Engl.. 31, 153 (1992)
Aksay I. A. and S. Weiner. Current Opinion in Solid State & Materials Science. 3, 219 (1998)
Attard G. S., J. C. Glyde, and C. G. Goltner. Nature. 378, 366 (1995)
Berman A., L. Addadi, and S. Weiner. Nature. 331, 546 (1988)
Beck J. S., J. C. Vartuli, G. J. Kennedy, C. T. Kresge, W. J. Roth and S. E. Schramm. Chemistry of Materials. 6, 1816 (1994)
Berman A., L. Addadi, A. Kvick, L. Leiserowitz, M. Nelson and S. Weiner. Science. 250, 664 (1990)
Berman A., J. Hanson, L. Leiserowitz, T. F. Koetzle, S. Weiner and L. Addadi. Science. 259, 776 (1993)
Carlson S. J. Skeletal Biomineralization: Patterns, Process and Evolutionary Trends. ed. by J. G. Carter, Van Nostrand Reinhold, New York, (1990), 531
Firouzi A., D. J. Schaefer, S. H. Tolbert, G. D. Stucky and B. F. Chmelka. Journal of the American Chemical Society. 119, 9466 (1997)
Hillhouse H. W., J. W. van Egmond, M. Tsapatsis. Langmuir. 15, 4544 (1999)
Hodge A. J.. and J. A. Petruska. In: Aspects of Protein Structure. ed. by Ramachandran, G. N. Academic, New York, (1963), 289
Hohansen E. and H. F. Parks. Arch. Oral. Biol.. 7, 185 (1962)
Huo Q., D. I. Margolese, U. Ciesla, P. Feng, T. E. Gier, P. Sieger, R. Leon, P. M. Petroff, F. Schuth and G. D. Stucky. Nature. 368, 317 (1994)
Huo Q., D. I. Margolese, U. Ciesla, D. G. Demuth, P. Feng, T. E. Gier, P. Sieger, A. Firouzi, B. F. Chmelka, F. Schuth and G. D. Stucky. Chemistry of Materials. 8, 1147 (1996)
Kresge C. T., M. E. Leonowicz, W. J. Roth, J. C. Vartuli and J. S. Beck. Nature. 359, 710 (1992)
Lees S. and F. A. Page. A study of some properties of mineralized turkey leg tendon. Conn. Tiss. Res.. 28, 263 (1992)
Iijima M., Y. Moriwaki, and Y. Kuboki. Conn. Tiss. Res.. 36, 51 (1997)
Lin H. P. and C. Y. Mou. Science. 273, 765 (1996)
264
Lu Y., R. Ganguli, C. A. Drewien, M. T. Anderson, C. J. Brinker, W. Gong, Y. Guo, H. Soyez, B. Dunn, M. H. Huang, and J. I. Zink. Nature. 389, 364 (1997)
Lowenstam H. A. and S. Weiner. On Biomineralization. Oxford University Press, New York, (1989)
Manne S., J. P. Cleveland, H. E. Gaub, G. D. Stucky, P. K. Hansma. Langmuir. 10, 4409 (1994); S. Manne and H. Gaub. Science. 270, 1480 (1995)
Maschmeyer T. Current Opinion in Solid State and Materials Science. 3, 71 (1998)
Ozin G. A., E. Chomski, D. Khushalani, M. J. MacLachlan. Current Opinion in Colloid and Interface Science. 3, 181 (1998)
Ozin G. A., H. Yang, I. Sokolov and N. Coombs. Advanced Materials. 9, 662 (1997)
Ryoo R., C. H. Ko, S. J. Cho and J. Kim. Journal of Physical Chemistry. B 101, 10610 (1997)
Senden T. J., C. J. Drummond, and P. Kekicheff. Langmuir. 10, 358 (1994)
Smith B. L., T. E. Schaffer, M. Viani, J. B. Thompson N. A. Frederick, J. Kindt, A. Belcher, G. D. Stucky, D. E. Morse and P. K. Hansma. Nature. 399, 761 (1999)
Ten Cate A. R. Oral Histology: Development, Structure and Function. St. Louis, MO: Mosby (1994)
Traub W., T. Arad, and S. Weiner. Proc. Natl. Acad. Sci.. 86, 9822 (1989)
Trau M., N. Yao, E. Kim, Y. Xia, G. M. Whitesides and I. A. Aksay. Nature. 390, 674 (1997)
Vartuli J. C., K. D. Schmitt, C. T. Kresge, W. J. Roth, M. E. Leonowicz, S. B. McCullen, S. D. Hellring, J. S. Beck, J. L. Schlenker, D. H. Olson and E. W. Sheppard. Chemistry of Materials. 6, 2317 (1994)
Veis A. In: Biomineralization, 1989, Chemical and Biological Perspectives. eds. by S. Mann J. Webb and R. J. P. Williams. VCH, Weinheim. (1989), 189
Wang R. Z., L. Addadi, and S. Weiner. Philos. T Roy Soc.. B 352, 469 (1997)
Wang R. Z. J. Am. Ceram. Soc.. 81, 1037 (1998)
Wang R. and S. Weiner. Conn. Tiss. Res.. 39, 269 (1998)
Wang R. Z. and S. Weiner. J. Biomechanics. 31, 135 (1998)
Weiner S. and L. Addadi. J. Mater. Chem.. 7, 689 (1997)
Weiner S. and H. D. Wagner. Annu. Rev. Mater. Sci.. 28, 271 (1998)
Yamauchi M., E. P. Katz, O. Kazunori, K. Teraoka and G. L. Mechanic. Conn. Tissue Res.. 21, 159 (1989)
Yang H., N. Coombs, and G. A. Ozin. Nature. 386, 692 (1997)
Yang H., N. Coombs, I. Sokolov and G. A. Ozin. Journal of Materials Chemistry. 7, 1285 (1997)
265
Yao N., A. Ku, N. Nakagawa, T. Lee, D. A. Saville, and I. A. Aksay Disorder-Order Transition in Mesoscopic Silica Thin Films. Chem. Mater.. 12, 1536–48 (2000)
Yang H., A. Kuperman, N. Coombs, S. Mamiche-Afara and G. A. Ozin. Nature. 379, 703 (1996)
Zhao X. M., Y. Xia, and G. M. Whitesides. Journal of Materials Chemistry. 7, 1069 (1997)
Zhao D., P. Yang P. Q. Huo, B. F. Chmelka and G. D. Stucky. Current Opinion in Solid State and Materials Science. 3, 121 (1998b)
Zhao D., J. Feng, Q. Huo, N. Melosh, G. H. Fredrickson, B. F. Chmelka and G. D. Stucky. Science. 279, 548 (1998a)
Ziv V., H. D. Wagner, and S. Weiner. Bone. 18, 417 (1996)
266
18, Nanophase metal oxide materials for electrochromic displays
18.1 Introduction
Nanophase materials generally refer to material systems consisting of interconnected nanodomains or other nanocomponents. These nanodomains, by definition, contain only a small number of atoms tightly bound together by an attractive potential; their sizes are usually in the range of 1 nm to 100 nm. Nanophase materials often exhibit novel chemical, electrical, optical, mechanical, or magnetic properties different from their bulk materials and the individual atomic constituents.
A unique feature of nanodomains is that a large fraction of the atoms are located on the surfaces of the domains; nanodomains have a high surface-to-bulk atom ratio. When these nanodomains are joined to form bulk nanophase materials, a large number of interfaces or grain boundaries are created. Atoms located at interfaces or grain boundaries are flexible and dynamic: they interact with and conform to the surrounding environment. Relaxation and reconstruction can relocate interfacial atoms from their bulk-like positions; the nature and the number of atoms surrounding an interfacial atom can be very different from those of atoms located within the nanograins. The interfacial regions or grain boundaries introduce new electronic or crystal structures. Thus, a nanophase material can be treated as a nanocomposite consisting of at least two phases: one phase is related to the atoms located within the nanograins and another phase is related to the atoms located at grain boundaries or interfaces.
Various nanophase materials and their applications have been discussed in the previous chapters. In this chapter, we will discuss electrochromic properties of nanophase metal oxide materials; discuss a novel device design for fabricating flexible, printed electrochromic displays; and elucidate the origin of the enhanced electrochromism of tin oxide nanocrystallites heavily doped with antimony.
Although great efforts have been put into the development of commercial electrochromic products, as demonstrated by the significant increase in the number of patents in the last two decades, large-scale applications of electrochromic devices have not yet been realized. One of the main reasons for this lack of breakthrough is probably due to the limited understanding of the electrical, optical, and electrochemical properties of nanophase materials and the relationship of these properties to the nanostructure of electrochromic materials. In this chapter, we will discuss some recent research efforts in elucidating the relationship between the nanostructure of metal oxide nanophase materials and their electrochromic properties.
Basic knowledge of the properties of electrochromism and electrochromic devices may help the readers appreciate the issues involved in producing commercially successful devices. Section 18.2 is intended to provide basic background on electrochromic materials and the performance of electrochromic displays. In Section 18.3, we will focus our discussion on the synthesis and structural characterization of antimony-tin oxide (ATO) nanophase materials that were recently reported to have enhanced electrochromic properties (Coleman, et al., 1999a). We will discuss, in Section 18.4, a new design strategy for constructing printed, flexible electrochromic displays using interdigitated electrodes. Detailed discussions on the electrical,
267
structural, and electrochromic properties of ATO electrochromic displays will be given in Section 18.5. We will conclude this chapter with a brief summary in Section 18.6.
18.2 Basic Concepts in Electrochromism
During an electrochemical reaction, electron-transfer reactions often change the optical absorption properties of the reacting compounds. Electrochromism refers to the phenomenon of reversible color changes induced in materials by electrochemical processes. Many inorganic and organic compounds exhibit pronounced electrochromic behavior that makes these compounds interesting for developing commercially viable electrochromic displays, windows, shutters, and electro-optical data storage devices (Monk, et al., 1995). Most of the following material in this section is drawn from the extensive literature on electrochromism, electrochromic materials, and electrochromic devices (Baucke, 1991; Dautremont-Smith, 1982; Deb, 1995; Granqvist, 1994; Mastragostino, 1993; Monk, et al., 1995; Mortimer, 1997; Scrosati, 1993; Ziegler and Howard, 1995).
18.2.1 Electrochromic Display Device
Electrochromic devices are able to change their color reversibly under the action of an externally applied voltage pulse. An electrochromic device is essentially a rechargeable battery with the electrochromic electrode separated by a solid, gel, or liquid electrolyte from a charge-balancing counter-electrode.
Figure 18.1 illustrates the essential components of a conventional electrochromic display with a "sandwich" configuration of electrodes. The front panel usually consists of a thin film of optically transparent electrode (OTE) deposited on a glass support. Tin-doped indium oxide (ITO) is the most commonly used material as the OTE. A thin film of electrochromic material is then deposited on the OTE by various deposition techniques. The OTE supplies electrons to the electrochromic layer. An ionically conductive electrolyte is then incorporated between the counter electrode and the electrochromic layer. Ideally, the electrochromic layer should be a good conductor for both small ions (e.g., H+, Li+, Na+, etc.) and electrons while the electrolyte layer should have zero conductivity for electrons but very high conductivity for small ions.
Figure 18.1 Schematic illustrating the conventional "sandwich" design of reflective electrochromic display using optically transparent electrode (OTE) and glass support.
268
When an external potential is applied between the working and counter electrodes, the charge or discharge of this electrochemical cell induces color changes in the electrochromic layer. The modulation of the optical properties of electrochromic devices is largely determined by the capacity of the electrochromic materials for charge injection and expulsion.
Although details of the electrochromic processes depend on the specific materials used, it is generally believed that the coloration occurs as a result of the injection and trapping of electrons in the electrochromic materials. When an appropriate potential is applied between the front OTE and the counter electrode, highly mobile ionic species (e.g., Li+,) in the electrolyte is inserted into the electrochromic material. To ensure charge neutrality, the insertion of one positive ion into the electrochromic material must be accompanied by the injection of a balancing electron into the same material. Thus, the ion-insertion process induces changes in the electronic distribution of the host electrochromic material. The injection of electrons shifts the Fermi energy of the electrochromic material and, thus, changes its optical absorption properties.
18.2.2 Electrochromic Materials
Materials that undergo color changes during an electrochemical redox reaction are called electrochromic materials. Electrochromic materials are currently attracting much attention because of their fascinating fundamental electrochemical properties as well as their potential commercial applications. A comprehensive discussion on inorganic and organic electrochromic materials can be found in a recent book on electrochromism (Monk, et al., 1995) and in several recent review articles (Mastragostino, 1993; Scrosati, 1993; Granqvist, 1994; Mortimer, 1997).
Many chemical species exhibit electrochromism. Based on their mechanism of coloration, however, electrochromic materials can be broadly grouped into two categories: ion-insertion materials change color
269
through reversible insertion of ions and injection of electrons; reversible electrodeposition materials produce color change through the plating of thin films onto and stripping of thin films from an OTE (Ziegler and Howazd, 1995).
Ever since the discovery of the electrochromic behavior of thin films of tungsten trioxide (Deb, 1969), electrochromic metal oxides have been the most intensively studied ion-insertion materials. Organic systems such as bipyridilium and electroactive conducting polymer systems have also received great interest (Mastrogastino, 1993; Monk, et al., 1995; Mortimer, 1997).
Although the details of the electrochromic processes are not known on a nanoscopic or atomic level, the ion-insertion or ion-extraction processes of electrochromic metal oxides can often be schematically represented by
where Me represents a metal, A± represents a singly charged ion, e- is the charge of an electron, α is a fractional insertion coefficient, and n is usually an integer that depends on the particular type of metal oxide of interest.
Many thin films of transition metal oxides can be electrochemically switched to a non-stoichiometric redox state that has an intense electronic absorption band due to optical intervalence chargetransfer (Hush, 1967). The most classical example of this type of ion-insertion metal oxide materials is tungsten trioxide, WO3.
Thin films of tungsten trioxide are usually transparent or pale yellow, depending on how the films are prepared; all the tungsten sites have an oxidation state W6+. Upon electrochemical reduction, some tungsten sites with an oxidation state W5+ can be created. With the assistance of a photon, intervalence electron transfer can occur between W5+ and W6+ sites, thus, giving an intense optical absorption. Although the detailed electrochromism is still not well understood, the injection and extraction of electrons and metal cations play a critical role in determining the electrochromic performance of tungsten trioxide electrochromic devices. With the use of Li+ containing electrolytes, the electrochemical reaction can be written as
This photon-assisted intervalence transition between lattice sites of different oxidation states has been used to explain the electrochromic behavior of most transition metal oxide systems.
An electronic band theory has also been proposed to explain the electrochromic mechanisms of various metal oxide systems (Granqvist, 1994). For heavily disordered tungsten trioxide films, the Fermi energy lies in the gap between the valence and the conduction band; and the bandgap is so wide that the material appears transparent in its pristine state. The insertion of electrons has the effect of lifting the Fermi energy
270
upward into the conduction band, thus making the material more electronically conductive and intensely absorbing of photons by free electrons.
18.2.3 Perceived Color and Contrast Ratio
The color we see from an object is a subjective visual impression involving retinal responses of our eyes to particular wavelengths of the impinging electromagnetic waves. Light comprising all visible electromagnetic wavelengths (420–700 nm) appears white to our eyes. Diffusely reflected colors from a display device result from absorption of visible light with certain wavelengths. Under white light illumination, the perceived color of a display is related to the complementary color of the light that the electrochromic material absorbs, plus the scattering effect of particulate materials.
Absorption of photons by a solid material depends on the band structure of the material and can be measured by a spectrophotometer. Diffuse reflectance spectroscopy is often used to measure the absorption spectrum of reflection displays although complications may arise from light scattering by microparticles or grains.
For electrochromic device's, a contrast ratio is often used to quantitatively measure the degree of color change or photon absorption. The contrast ratio of a display is defined as the ratio of the intensity of light diffusely reflected from its bleached state to that of light diffusely reflected from its colored state. The contrast ratio often depends on the specific wavelength of the illumination system. Therefore, in practice, an integrated (over visible wavelengths) contrast ratio is often used for white light illumination. For successful commercial applications, a contrast ratio with as high a value as possible is desired.
18.2.4 Coloration Efficiency and Response Time
The optical absorption properties of an electrochromic display are related to the injected charge per unit area of the active electrode. The coloration efficiency is defined as the ratio of the change (between the colored and bleached state) of optical absorbance to the charge injected per unit area of the display electrode. The optical absorbance is defined as the log of the contrast ratio. The coloration efficiency is a measure of the performance of the electrochromic material; it depends on the chemistry, microstructure, and the film thickness of the electrochromic layer.
The time required for an electrochromic device to change from its bleached state to its colored state is called the device's response time. Since the electrochromic process involves migration of ions in the electrolyte, diffusion of ions into the electrochromic material, and electron diffusion from the electrode into the electrochromic layer, the response time is, for practical devices, usually a few seconds or longer. However, electrochromic devices exhibiting much faster response times were reported in the literature (Haranahalli and Dove, 1980; Oi, 1986).
18.2.5 Write-Erase Efficiency and Cycle Life
271
The write-erase efficiency of an electrochromic display refers to the percentage of the originally formed color that can be subsequently bleached by a reverse potential; it can be expressed as the ratio of changes of absorbance after one complete redox reaction cycle. The write-erase efficiency is a measure of the reversibility of the electrochemical reaction and, for practical applications, is closely coupled to the response time. Ideally, the physical structure and the chemistry of the electrochromic device should not change after many cycles of electrochemical reactions. In practice, however, the write-erase efficiency slowly degrades because of structural or chemical changes occurring within the electrochromic device. For successful commercial applications, the initial write-erase efficiency should approach 100% or as high a value as possible. Various strategies have been proposed to increase the write-erase efficiency of electrochromic devices (Wrighton and Bookbinder, 1983; Calvert, et al., 1986).
When an electrochromic device is continually cycled between its bleached and colored states, the coloration efficiency progressively deteriorates and device failure eventually occurs as a result of physical changes or chemical reactions. The cycle life is a measure of the long-term stability of an electrochromic device. A major goal for device design and fabrication is to maximize the device's cycle life.
18.3 Nanophase Metal Oxide Electrochromic Materials
Transition metal oxides, such as WO3, MoO3, V2O5, IrO2, Rh2O3, Co2O3, NiOx, etc., exhibit intense optical absorption, presumably due to optical intervalence charge transfer occurring in mixed-valence compounds (Hush, 1967). The coexistence of ions with different oxidation states in a transition metal oxide makes it possible for an electron to be excited from one site to an adjacent site, accompanied by the absorption of a photon. These transition metal oxides possess unique structural configurations, comprising MeO6 octahedral building blocks connected either by corner sharing, edge sharing, or both.
Thin films of metal oxides deposited on transparent, conductive electrodes are usually used in electrochromic devices. These films can be prepared by a variety of techniques including thermal evaporation (Holland, 1956), sputtering (Suzuki and Mizuhashi, 1982), chemical vapor deposition (Donnadieu, et al., 1988), electrodeposition (Bendert and Corrigan, 1989), sol-gel process (Hensch and West, 1990), or dip-coating (Hinokuma, et al., 1992).
Since there is extensive literature on the electrochromism and the preparation methods of these metal oxide materials, we will not review them here; instead we will focus our discussion on some recent development in using nanophase, heavily doped, semiconducting metal oxide materials for electrochromic displays.
Thin films of tin-doped indium oxide (ITO) are widely used as semiconducting, optically transparent coatings or electrodes for the construction of electrochromic devices; their electrical and optical properties have been extensively reviewed (Hamberg and Granqvist, 1986; Granqvist, 1993). Previous studies show that ITO films exhibited little or no optical change in the visible region under typical conditions of operating electrochromic displays (Brotherston, 1995). Because of its inherently weak electrochromism and quasi-reversible ion-insertion processes, ITO films are rarely used as the primary electrochrome.
272
Antimony-doped tin oxide (ATO) materials have many interesting and useful properties for industrial applications: ATO particles or powders are used as oxidation catalysts (McAteer, 1979; Berry, 1981), antistatic pigments, and components for solar cells and gas sensors (Oyabu, 1982; Chopra, et al., 1983). Similar to ITO films, thin films of ATO materials exhibited very weak electrochromism (Orel, et al., 1994).
However, when ATO or ITO nanocrystallites are dispersed onto transparent or light-colored oxide powders such as titania, silica, alumina, etc., ATO and ITO nanophase materials exhibit surprisingly high level electrochromism (Coleman, et al., 1998, 1999a). Although many factors affect the perceived contrast of reflective electrochromic displays, we found that the enhanced electrochromism of supported ATO or ITO nanocrystallites is related to the nanostructure of the synthesized nanophase materials (Liu, et al., 1998).
Since the synthesis processes and the electrochromism of ITO nanophase materials are similar to those of ATO nanophase materials, we will focus our discussion only on ATO systems. The knowledge gained from the study of ATO nanophase materials can be equally applied to ITO or other doped metal oxide nanophase materials.
18.3.1 Synthesis of Supported ATO Nanocrystallites
To better control the synthesis processes of nanophase materials and to keep the cost for mass production as low as possible, we chose to use wet chemical techniques to produce large-scale ATO nanophase materials. The wet chemical methods usually start with molecular precursors in a solution form to make a final condensed phase. The nature of the condensed phase depends on the physicochemical conditions such as the type of precursor, concentration, stoichiometry, reaction time, temperature, pressure, solvent system, the presence of additives or surfactants, the specific method of mixing, and, in some cases, the use of light irradiation.
The advantages of using wet chemistry in making nanophase materials include the following: control the mixing of different precursors at the molecular level; control the reaction kinetics by choosing the right solvent, additives, and temperature; and control the solution homogeneity during reaction, nucleation, growth, and precipitation of amorphous or crystalline nanoclusters.
In the following, we describe a detailed procedure of how to make tin dioxide nanocrystallites heavily doped with antimony (43 mol/% Sb) and finely dispersed onto alumina particulates to make a powdered material consisting of 75 wt% of ATO and 25 wt% of alumina.
Figure 18.2 illustrate the sample preparation procedure. A solution of SnCl4 · 0.5H2O (40.0 g) in deionised water (20 mL) is mixed with SbCl5 (25.0 g) in concentrated HCl (20 mL). Alumina powder (10.0 g) is stirred in deionised water (200 mL) in a beaker heated to 90°C. The mixed Sb/Sn chloride solution is added to the alumina powder solution over a period of about 45 min from a burette with concurrent addition of 15% sodium hydroxide solution from another burette to keep the pH value of the mixture at 0.8–1.2. After the Sb/Sn chloride solution is completely added to the alumina solution and the neutralization is complete, the pH value of the final mixture is adjusted to 2.0; the heater is turned off; and the mixture, still being stirred, is allowed to cool down for 3 h.
273
Figure 18.2 Schematic illustrating the synthesis process of antimony-tin oxide nanophase materials using wet chemistry method.
The white, solid precipitates are then filtered off, washed with deionised water, and dried at 60°C in a vacuum oven for > 3 h. The dried powders are then calcined in an open ceramic crucible at 600°C for 3 h. At the end of the calcination, the sample is removed from the oven and is allowed to cool quickly in air. The final light-gray powders are then mixed with polymer binders to produce inks suitable for printing (see Section 18.4).
Our studies showed that it is critical to simultaneously precipitate Sb and Sn molecular species to optimize the contrast ratio of ATO electrochromic displays. A sequential addition of Sb and Sn solutions to alumina or other light-colored metal oxide supports did not give satisfactory electrochromic performance. ATO nanophase materials prepared by using antimony pentachloride precursors may give better switching properties than do those prepared by using antimony trichloride precursors.
18.3.2 Characterization of Supported ATO Nanocrystallites
274
The synthesis strategy is to uniformly coat ATO nanocrystallites onto light-colored oxide powders. The purpose of using powder supports is similar to that of supported metal catalysts: to increase the total surface area of ATO nanocrystallites that can be easily accessed by ionic species in the electrolyte. Figure 18.3(a) illustrate the spatial arrangement of ATO nanocrystallites and titania particulates. The oxide supports may not be conductive, such as alumina powders or silica shells. Therefore, to ensure a good electrical conductivity of the electrochromic ATO-support materials, the oxide powders should be completely coated with one or, preferably, more layers of conductive ATO nanocrystallites.
Figure 18.3 (a) Coating of antimony-tin oxide nanocrystallites onto light-color oxide supports such as titania, alumina, and silica. High-resolution backscattered electron image. (b) Antimony-tin oxide supported on titania powders reveals coating of ATO nanocrystallites onto titania particles.
Figure 18.3(b) shows a high-resolution backscattered electron image of a cross-sectional sample of ATO-TiO2 powder, clearly revealing the coating of ATO nanocrystallites onto the large TiO2 particulates.
To understand the atomic structure of the ATO nanocrystallites and their interactions with the TiO2 support, Fig. 18.4 shows a high-resolution transmission electron microscopy (HRTEM) image of a cross-sectional
275
sample of ATO-TiO2 powder as that shown in Fig. 18.3(b). The amorphous material in the image represents the structure of the resin used for embedding the ATO-TiO2 powder. Various sizes of ATO nanocrystallites are tightly coated onto the TiO2 microparticles and the ATO nanocrystallites are interconnected. There is no preferential crystallographic orientation relationship among the ATO nanocrystallites or between the ATO nanocrystallites and the crystalline TiO2 microparticles. The coating thickness is in the range of 5 nm to 50 nm and the sizes of the ATO nanocrystallites range from about 2 nm to 10 nm. HRTEM images of ATO nanocrystallites supported on other oxide particulates showed similar results.
Figure 18.4 High-resolution transmission electron microscopy image of ATO nanocrystallites attached to titania microparticles. The sample was prepared by ultramicrotming ATO-titania powders embedded in resin.
To synthesize high-performance ATO electrochromic nanophase materials, we need to optimize the Sn/Sb ratio, the mixing of Sn and Sb molecular precursors, the solution pH value, and the annealing temperature. The effects of these factors on the contrast of electrochromic displays will be discussed in Section 18.5.
18.4 Construction of Printed, Flexible Displays Using Interdigitated
Electrodes
The conventional method of fabricating electrochromic displays utilizes a "sandwich" configuration as illustrated in Fig. 18.1. While such a design can give desired performance, it suffers several drawbacks: high cost, rigid device because of the use of glass support, low conductivity on plastic substrates, difficulty in fabricating sophisticated display designs, and electrochemical instability leading to corrosion, especially in aqueous systems. To overcome these difficulties and to produce large area, flexible electrochromic displays,
276
we developed an interdigitated electrode approach by using low cost printing and coating processes (Coleman, et al., 1999b).
18.4.1 Design Strategy
The design objective is to print the display on flexible polymer films using commercially viable conductive inks in an interdigitated electrode structure. Figure 18.5(a) illustrate the cross-sectional view of a display consisting of several layers of printed materials. An appropriate circuit can be designed and screen-printed onto a polymer film with silver-carbon ink as working and counter electrode contacts. Then, a layer of carbon ink is printed, completely covering the silver-carbon layer to provide corrosion protection.
Figure 18.5 Schematic illustrating (a) "side-by-side" design of a reflective electrochromic display using interdigitated electrode approach (low magnification backscattered electron image); and (b) the cross-sectional view of a printed electrochromic display showing the Ag/C contacts and the ATO-TiO2 electrochromic layer.
277
Without the protection of the conforming carbon layer, the display may deteriorate quickly and might not maintain a long cycle-life. Electromigration of silver microcrystals may occur under prolonged use of the display. When silver microcrystals interact with other components, especially with electrolyte that diffuses into the electrochromic materials, the display device will be degraded.
After printing the conforming carbon layer, the connecting circuits are then covered with an insulating material. The actual electrode surface is then printed with light-colored conductive metal oxide powder dispersed in a polymer binder. The electrochromic material can be mixed into, or printed on top of, the conductive metal oxide materials. A layer of an aqueous gel electrolyte is then put down before the whole system is sealed with a transparent polymer film.
Figure 18.5(b) shows a backscattered electron image of the cross-sectional view of a display device consisting of Mylar support, Ag/C electrodes, and an ATO-TiO2 electrochromic layer. The combined thickness of the electrochromic layer and the conductive metal oxide dispersion layer is less than 50 µm.
When an external voltage is applied between the working and the counter electrodes, electric current moves vertically through the conductive metal oxide dispersion to the interface between the electrochromic material and the gel or liquid electrolyte. Electrochromic process or color change takes place in this interfacial region. Ionic current flow through the electrolyte completes the electrical circuit. If the interelectrode spacing is kept significantly larger than the thickness of the metal oxide coating layer, then the leakage current within the metal oxide dispersion layer can be minimized. This leakage current can be further reduced by printing a thin layer of insulator between the working and counter electrodes as illustrated in Fig. 18.5(a). For this "side-by-side" display design to work efficiently, high ionic conductivity of the electrolyte is required to maximize current flow through the desired path.
Sophisticated patterns can be designed using commercially available graphic design programs; these patterns can be transformed into displays by using specialized screen-printing technology with a high production rate and low cost. Figure 18.6(a) shows a photograph of a printed word display, revealing the uncovered Ag/C working and counter electrodes and the insulator layer. Figure 18.6(b) shows the same area after printing a layer of ATO-TiO2-polymer electrochromic ink, covering the area of active display. Figure 18.6(c) shows a low-voltage, secondary electron image of the ATO-TiO2 layer, revealing the surface morphology, ATO-TiO2 particulates, and the polymer binder. After a gel-electrolyte layer is pressed onto the electrochromic layer and sealed with a transparent polymer film, the display device is ready to operate.
Figure 18.6 Photomicrographs of a printed electrochromic display show the exposed electrodes (a) and the same area after covered with the ATO-TiO2 electrochromic layer (b). Low-voltage field-emission SEM image (c) shows the highly porous ATO-TiO2 surface layer, the ATO coated TiO2 particulates, and the polymer binder.
278
18.4.2 Materials Selection
1. Conducting Electrodes With the "side-by-side" interdigitated electrode configuration, the patterned conducting layer on the polymer support acts as both the working and the counter electrode. We screened a variety of materials including silver and silver-carbon ink, carbon or graphite ink, electrochemically deposited copper, photo-patterned aluminum, electroless nickel, conducting polyaniline, conductive metal oxides, and sputtered gold. To obtain a combination of good conductivity and ease of printing into sophisticated circuitry with fine resolution, commercially available silver or silver-carbon inks give the best performance.
2. Insulating Materials To assure easy printing with acceptable resolution, commercially available UV-curable acrylated polyols can be used as insulating materials.
3. Conducting Metal Oxides The electrode surface layer should be conductive, light in color, and stable both chemically and electrochemically under operating conditions. In addition, the material should be easily configured and formulated into printable inks. Powders of conductive metal oxides dispersed in polymer binder solutions can meet these criteria. Many light-colored conductive pigments, for example, can be used. A typical material may comprise antimony-doped tin oxide crystals dispersed onto inert, light-scattering TiO2. The stability of these pigment coatings towards aqueous electrolyte is acceptable for commercial applications. However, these materials are susceptible to overreduction and loss of conductivity when an excessive potential is applied.
4. Electrochromic Layer Many organic or inorganic electrochromic materials are available (Monk, et al., 1995; Mortimer, 1997). Electrochromic materials that are soluble in aqueous electrolytes can be used. Those materials
279
include the following: methyl viologen, heptyl viologen, and metal ions which can be electroplated such as bismuth (III). Insoluble electrochromic materials such as Prussian Blue, poly (xylyl viologen)-poly (styrenesulfonate), and polyaniline can be either coated onto the metal oxide surface layer or coated onto the conductive pigment powders prior to being formulated into inks.
However, since nanophase ATO-oxide powders display significant electrochromic effect in neutral electrolytes such as aqueous sodium sulfate solution, we can use ATO nanocrystallites dispersed onto light-colored powder supports as the conductive electrode as well as the electrochromic layer.
5. Electrolyte With the "side-by-side" electrode configuration, the path of ion migration in the electrolyte may be as long as, or even longer than, half the combined width of both the working and the counter electrodes. To minimize the effects of field gradients on ionic conduction and to keep the electrolyte resistance as low as possible, liquid electrolytes, such as aqueous salt solutions, are preferred. We typically use an electrolyte that contains a combination of polymeric thickening agent and hygroscopic salt. Hydroxyethyl cellulose, polyacrylic acid, and poly-AMPS (poly [acrylamido methylpropane sulfonic acid]) are often used as thickening agents. The following hygroscopic salts are frequently used: lithium chloride, potassium chloride, potassium formate, lithium bromide, and calcium chloride.
6. Sealing Film Since a liquid or gel electrolyte is usually used in the construction of the reflective electrochromic display, an inert, transparent film is used to seal the display to minimize the evaporation of water present in the electrolytes. Some commercial polyester films (e. g., Mylar) can be used because of their excellent transparency, low cost, flexibility, and good water barrier properties.
7. Ink Formulation Since we intend to fabricate the electrochromic displays by an inexpensive printing technology, it is crucial to formulate the powdered materials into conductive inks with excellent printability. After screening many binder-solvent systems, we selected a fluoroelastomer (Viton) dissolved in 2-butoxyethyl acetate. Pigment-to-binder ratio is an important factor in optimizing the properties of the inks. Higher binder levels give increased electrical resistance while lower binder levels give a less stable, chalky printing layer. It is also possible to incorporate fluorescent pigments into the formulated inks to produce bright-colored displays (Coleman, et al., 1999b).
18.4.3 Display Examples
The "side-by-side" design of electrochromic displays eliminates the use of expensive transparent electrodes, commonly used in the conventional "sandwich" design. A simple printing process, using inks of powdered materials, can be used to fabricate large area, flexible electrochromic displays with interdigitated electrodes. We have demonstrated that many types of devices with sophisticated designs and a range of colors can be fabricated (Coleman, et al., 1998, 1999b). Conventional electrochromic materials (e.g., heptyl viologen and formulations of Prussian Blue with or without incorporated pigments) as well as the newly discovered ATO, ITO, or other nanophase metal oxide materials can be easily formulated into printable inks and give excellent device performances.
Figure 18.7(a) shows a word display (neutral state) printed with ATO-silica nanostructured materials as the electrochromic layer. Figure 18.7(b) and 18.7(c) show the same word display but with the polarity of the applied voltage switched. The externally applied voltage was 1.5 V. With a pulsed external voltage, these flexible displays can be cycled many times with high contrast.
280
Figure 18.7 Printed electrochromic word displays using ATO-silica powders. The applied potential was 0.0 V for (a), +1.5 V for (b), and -1.5 V for (c).
Figure 18.8 shows another example of a printed seven-segment numerical display. Only half of the numerical display was activated to demonstrate the performance and the active control of printed, reflective displays.
Figure 18.8 Seven-segment electrochromic numerical display.
281
The contrast of reflective displays depends on many parameters including the ATO loading, the types of support used, the doping level of antimony, the oxidation state(s) of the incorporated antimony ions, the average size of the ATO nanocrystallites, and the spatial distribution of the antimony containing species. In the next section, we will discuss in detail how these factors affect the perceived contrast of electrochromic displays.
18.5 Contrast of Printed Electrochromic Displays Using ATO Nanophase
Materials
Commercially viable electrochromic devices require electrochromic materials having high contrast-ratio, high coloration efficiency, fast response time, high write-erase efficiency, and long cycle life. The contrast ratio, however, is the most important parameter for evaluating the performance of a commercial electrochromic display. Therefore, to optimize their performance, we need to understand the factors that affect the contrast ratio of reflective, printed electrochromic displays.
The perceived contrast of reflective displays fabricated by the printing technology is influenced by many interacting factors as shown in Fig. 18.9. Both the scattering and the absorption of the incident (white) light determine the contrast ratio. To enhance the performance of the printed electrochromic displays, we need to simultaneously optimize all the interconnecting and interacting factors. Detailed discussions on how to accomplish the multi-parameter optimization process and to select the appropriate parameters are beyond the scope of this chapter. In the following, we will primarily focus on how the nanostructure of the synthesized ATO-support nanophase materials affect the observed contrast ratio of printed electrochromic displays.
282
Figure 18.9 Factors affecting the perceived contrast of reflective, printed electrochromic displays using ATO nanophase materials.
18.5.1 Effect of Antimony Doping on Contrast Ratio
The contrast ratio of printed, reflective displays depends on the dopant level of antimony in tin dioxide nanocrystallites. Figure 18.10 shows the contrast ratio as a function of antimony doping for display devices prepared with materials consisting of ATO nanocrystallites dispersed onto titanium dioxide powders. The ATO-TiO2 powders were annealed at 600°C for 3 h. The contrast ratio initially increases, reaches a maximum value at about 43 mol% of antimony, then decreases. A similar dependence of contrast ratio on antimony doping was also observed for ATO materials dispersed onto powders of other types of oxides.
283
Figure 18.10 Variations of contrast ratio with antimony doping for printed electrochromic displays. The ATO nanophase materials were annealed at 600°C for 3 h.
The electrical properties of tin dioxide nanocrystallites are also drastically changed by doping with antimony. The resistivity of the ATO nanophase materials critically depends on the doping level. Table 18.1 shows the resistivity of ATO-alumina powders at room temperature as a function of antimony doping. The resistivity rapidly decreases with low levels of antimony doping, reaches a minimum at a value below 10 mol% antimony, and then slowly increases with further incorporation of antimony into the tin dioxide nanophase materials.
Table 18.1 Effect of antimony doping on electrical resistivity and color of ATO nanophase materials*
Antimony Doping (mol%) Resistivity (Ω · cm) Color
0 3100000 Pale Yellow
10 Gray
20 2 Gray
43 10 Olive
* 75 wt % ATO-25 wt % alumina; 600°C/3 h
Not only the contrast ratio of the electrochromic displays varies with the doping level of antimony but also the color of the ATO-alumina nanophase materials changes from pale yellow for pure tin dioxide to olive
284
green for ATO nanophase materials with 43 mol% antimony doping. The change in color is related to the increased absorption of photons by the ATO nanophase materials.
The electrical, optical, and structural properties of ATO films and powders have been extensively studied (Carroll and Slack, 1976; Shanthi, et al., 1980; Cox, et al., 1982; Chopra, et al., 1983; Berry and Smith, 1984; Egdell, et al., 1984; Miyata and Kitahata, 1985; Kojima, et al., 1993, 1996; Goyal, et al., 1993; Mishra, et al., 1995; Caldararu, et al., 1996). Tin dioxide is an oxygen-deficient n-type semiconductor (Jarzebski and Morton, 1976). Reduction of SnO2 material can further increase its conductivity because of the increased level of oxygen vacancies. The addition of a small amount of antimony to SnO2 results in a dramatic increase in the electrical conductivity, resulting in a semiconductor with a low temperature dependence of the conductivity (Shanthi, et al., 1980). With the addition of more than about 10 mol% antimony to thin films of SnO2, however, the electrical resistivity increases significantly (Kojima, et al., 1993).
The change in color with the amount of antimony doping has also been observed in thin films of tin dioxide heavily doped with antimony (Kojima, et al., 1993). With low levels of antimony doping, some of the Sn4+ ions are replaced by Sb5+ ions; however, with increasing levels of antimony in SnO2, some of the Sn4+ ions may be replaced by Sb3+ ions. In a material with mixed oxidation states of the same element, a photon-assisted electron transfer between the different oxidation states can occur, resulting in intense photon absorption (Robin and Day, 1967; Kojima, et al., 1993, 1996). The presence of a mixture of Sb5+ ions and Sb3+ ions in the SnO2 lattice is responsible for the blackening of ATO thin films with heavy antimony doping (Kojima, et al., 1993).
For thin films of ATO materials, the increase in conductivity or in carrier concentration with low amount of antimony doping is attributed to the substitutional doping by pentavalent Sb5+ ions (Goyal, et al., 1993; Kojima, et al., 1993). At very high doping concentrations, the decrease in carrier concentration or the increase in resistivity is probably related to the increase of substitutional doping by trivalent Sb3+ ions (Mulla, et al., 1986).
To understand the effect of antimony doping on the contrast ratio of the ATO electrochromic displays, the color change of the ATO nanophase materials, and the change of the electrical resistivity, we examined the nanostructure of the synthesized ATO nanophase materials doped with different levels of antimony.
X-ray diffraction (XRD) patterns of the powdered materials with antimony doping levels up to 43 mol% showed that the ATO nanocrystallites have a single Cassiterite crystal structure similar to that of pure tin dioxide. The average size of the ATO nanocrystallites decreases, however, from about 15 nm for pure tin dioxide to about 3.8 nm for ATO materials with 43 mol% antimony doping (Fig. 18.11). Not only the average size of the ATO nanocrystallites becomes smaller but also the size distribution becomes narrower with increasing antimony doping.
Figure 18.11 Variations of the average size of ATO nanograins with antimony doping.
285
HRTEM images also clearly showed that under the same annealing conditions the average size of ATO nanocrystallites decreases with increasing antimony doping. Figure 18.12(a) and (b), respectively, shows HRTEM images of SnO2 and ATO nanocrystallites prepared by annealing SnO2 and ATO (43 mol% Sb + 57 mol% Sn) precursors at 600°C for 3 h. All the crystallites present in the ATO material have a single rutile-type structure indistinguishable from that of pure SnO2 nanocrystallites.
Figure 18.12 High-resolution transmission electron microscopy images of (a) pure tin dioxide sample and (b) tin dioxide with 43 mol% antimony doping. Antimony inhibits the growth of tin dioxide crystallites. The samples were annealed at 600°C for 3 h.
286
Both the XRD and the HRTEM results clearly show that the presence of antimony in the precursor materials inhibits the growth of the SnO2 nanocrystallites during the annealing process. It is interesting to note that even with a doping level as high as 43 mol% there were no distinctive antimony oxide phases observed in the ATO nanophase materials. However, Fig. 18.12(b) clearly shows that there exist highly disordered materials around the ATO nanocrystallites, especially at the grain boundaries and triple-junction regions. These highly disordered interfacial regions may support a high concentration of randomly distributed antimony cations.
Surface area measurement showed that the heavily doped ATO nanophase materials have a total surface area as high as 40 m2/g. Since HRTEM images showed that almost all the ATO nanocrystallites are aggregated to form large agglomerates with many grain boundaries and triple-junction regions, the high surfacearea of the ATO nanophase materials suggests that the highly disordered grain boundary or triple-junction regions are easily accessible to small ions. The presence of many grain boundaries facilitates a fast diffusion of small ions into and out of the electrochromic ATO layer, resulting in a faster response time. The presence of many grain boundaries also makes more ATO sites accessible to charge injection and expulsion, resulting in a higher coloration efficiency.
The variations of lattice constants of the Cassiterite ATO nanocrystallites with antimony doping are shown in Fig. 18.13. With increasing antimony content, the lattice constants slightly increases, then decreases with further antimony doping. Because the ionic radius of Sn4+ is larger than that of Sb5+ but smaller than that of Sb3+ (Shannon, 1976), the initial increase in the lattice constants seems to suggest that with low levels of antimony doping (<10 mol%) the amount of substitutional doping by Sb3+ also increase with the doping level.
287
Figure 18.13 Variations of lattice constants of antimony-tin oxide nanocrystallites with doping level: a-axis (a) and c-axis (b).
With low levels of antimony doping, both Sb3+ and Sb5+ ions are present within individual ATO nanocrystallites and both Sb3+ and Sb5+ ions are surrounded by six anions in octahedral coordination. However, in a mixed-valence compound such as antimony dioxide, the Sb5+ ions prefer sixfold octahedral coordination while the Sb3+ ions prefer fourfold pyramidal coordination (Rogers and Skapski, 1964; Skapski and Rogers, 1965). Therefore, the trivalent antimony ions substituted on the tin sites in the rutile-type ATO nanocrystallites are not in a preferred coordination.
With increasing levels of antimony doping, the ATO nanocrystallites becomes smaller and smaller. Since Sb3+ prefers threefold or fourfold coordination, when the ATO nanocrystallites become so small that the trivalent antimony ions can be preferentially segregated to the grain boundaries or surfaces of small particles (Pyke, et al., 1978). However, a charge balance on the ATO nanocrystallites requires the presence of both Sb3+ and Sb5+ ions. Therefore, with increasing amount of antimony doping, more and more Sb3+ ions are created and preferentially located at the grain boundaries while an increasing amount of Sb5+ ions are located within the ATO nanocrystallites.
With high levels of antimony doping (>10 mol%), the decrease of the lattice constants can be attributed to the increased amount of Sb5+ ions and the decreased amount of Sb3+ ions within the ATO nanocrystallites. Since grain boundary scattering and ionized impurity scattering play a major role in determining the carrier mobility in doped semiconductor materials, the increase in resistivity with high levels of antimony doping may be attributed to the significant increase in the number of grain boundaries and Sb3+ ions at these boundaries.
The variations in the contrast ratio with antimony doping shown in Fig. 18.10 can now be explained as follows. With increasing levels of antimony doping, the average size of the ATO nanocrystallites becomes smaller; the total number of grain boundaries increases; the total surface area of accessible sites increases;
288
and the total number of Sb5+ sites within individual ATO nanocrystallites also increases. The increase in contrast ratio with increasing antimony doping for doping levels <43 mol% is related to the increasing number of Sb5+ ions in the ATO nanophase materials.
Because of the preferential segregation of Sb3+ ions to the grain boundaries during the annealing process, there are more Sb5+ ions than Sb3+ ions within individual ATO nanocrystallites. The intervalence optical transition between the Sb3+ ions and some Sb5+ ions gives the gray or slightly dark gray color of the ATO electrochromic display in the bleached state. Upon electrochemical reaction, more Sb3+ ions may be generated within the ATO nanocrystallites. The photon assisted intervalence transition between the Sb3+ ions and the Sb5+ ions gives the black color of the ATO electrochromic display in the colored state as shown in Fig. 18.7(b).
The presence of many grain boundaries may also introduce interface states that can be drastically modified by the presence of Sb3+ ions. In the colored state, the injection of electrons will further shift the Fermi energy to the conduction band. All these factors contribute to the observed increase in contrast ratio with increasing antimony doping.
When the antimony content is increased to over 43 mol%, the average size of the ATO nanocrystallites becomes very small; the ATO nanophase materials are highly disordered; more and more antimony ions are located at the grain boundaries; disordered antimony oxide domains may be formed; and the conductivity of the ATO nanophase materials decreases rapidly. When the ATO nanophase materials become highly disordered or amorphous-like, antimony may act as a glass network former in stead of as a dopant in crystalline tin dioxide. Since highly disordered or amorphous materials do not have the constraints imposed by the long-range periodicity of crystalline materials, antimony ions may be incorporated in the glass network with most favored coordination. All these factors contribute to the observed decrease in contrast ratio for antimony doping levels higher than about 43 mol%.
18.5.2 Effect of Annealing Temperature on Contrast Ratio
The contrast ratio of the electrochromic displays also crucially depends on the annealing temperature of the ATO precursor materials. Figure 18.14 shows the variations in contrast ratio with annealing temperature for displays made with ATO nanocrystallites (43 mol% Sb + 57 mol% Sn) supported on alumina powders. The contrast ratio first increases with increasing temperature, reaches a maximum value at about 600°C, and then decreases with further increase in temperature. We found that the ATO materials that gave maximum contrast ratio should contain about 43 mol% antimony and be annealed at 600°C for 3 h.
Figure 18.14 Variations of contrast ratio with annealing temperature for printed electrochromic displays. The precursor materials comprise 75 wt% ATO (43 mol% Sb and 57 mol% Sn) and 25 wt% alumina.
289
The electrical properties of the ATO powders not only change with antimony doping but also vary drastically with the annealing temperature. Figure 18.15 shows the resistivity of the ATO powders as a function of annealing temperature. The resistivity of the precursor material is about 1.45 × 106 Ω · cm. With increasing annealing temperature, the resistivity reaches a maximum value of 16.46 × 106 Ω · cm at about 250°C, rapidly decreases to 123 Ω · cm at 450°C, then slowly decreases.
Figure 18.15 Variations of electrical resistivity with annealing temperature for ATO nanophase materials (43 mol% Sb and 57 mol% Sn).
290
The color of the ATO powders also changes with the annealing temperature: the precursor materials are off-white in color; the ATO powders annealed at 600°C are gray; and the ATO powders annealed at 1000°C are dark gray.
To understand the structural evolution of the ATO nanophase materials during the annealing process, both XRD and HRTEM techniques were used to examine the nanostructure of ATO nanophase materials annealed at different temperatures.
Figure 18.16 shows the average size of the ATO nanocrystallites as a function of the annealing temperature, clearly revealing that the average size does not change significantly at annealing temperatures <800°C. However, when the annealing temperature is higher than about 800°C, the ATO nanocrystallites suddenly grow significantly. XRD data also showed that when the annealing temperature is lower than about 750°C the ATO nanophase materials contain a single crystal Cassiterite structure. When the annealing temperature is higher than about 800°C, however, the ATO powders contain not only crystallites with a Cassiterite structure but also crystallites of separate antimony oxide phases.
Figure 18.16 Variations of average size of ATO nanocrystallites with annealing temperature. The precursor materials comprise 75 wt% ATO (43 mol% Sb and 57 mol% Sn) and 25 wt% alumina.
Figure 18.17 shows a series of HRTEM images of ATO nanophase materials annealed at different temperatures. The precursor material comprised many domains of nanocrystallites joined by highly disordered, amorphous-like materials. The amount of disordered materials present in the ATO nanophase materials decreased with increasing annealing temperature but the ATO nanocrystallites did not grow significantly at annealing temperatures <750°C. At annealing temperatures above 800°C, the ATO crystallites grew significantly (Fig. 18.17(d) and (e)).
291
Figure 18.17 High-resolution transmission electron microscopy images of (a) ATO precursor sample, (b) ATO nanophase materials annealed at 450°C, (c) ATO nanophase materials annealed at 600°C, (d) ATO nanophase materials annealed at 800°C, and (e) ATO nanophase materials annealed at 1000°C.
In the precursor sample, the highly disordered materials may contain antimony and other species that are probably non-conducting. Because the small SnO2 nanocrystallites are separated by the non-conducting amorphous species, the precursor material has a very high resistivity. At low annealing temperatures, the moisture and other organic species present in the powdered ATO materials may evaporate; thus, the resistivity of the dried ATO nanophase materials may increase.
292
During annealing at moderate temperatures, antimony incorporates into the SnO2 nanocrystallites; most of the antimony atoms diffuse to the grain boundaries of the SnO2 nanocrystallites. The high level of antimony present at the interfaces and grain boundaries inhibits the growth of SnO2 nanocrystallites. The reduction of highly disordered materials around the SnO2 nanocrystallites and the incorporation of Sb5+ ions into the SnO2 nanocrystallites increase the conductivity of the ATO nanophase materials.
At high annealing temperatures >750°C, because of the increased mobility, antimony first migrates away from the grain boundaries to form separate and stable antimony oxide phases; and then the lightly doped SnO2 nanocrystallites grow significantly. Even though separate antimony oxide phases were observed in ATO materials annealed at 1000°C, there were still Sb5+ and Sb3+ ions incorporated into the large SnO2 crystallites. Therefore, the high-temperature annealed ATO materials still have a good conductivity; the dark-gray color is due to photon-assisted intervalence transitions between Sb5+ and Sb3+ ions located within the large SnO2 crystallites.
The improved electrochromic performance of ATO nanophase materials annealed at about 600°C can be attributed to the formation of nanocrystallites with many antimony-rich grain boundaries. At a nanoscopic level, these materials provide high surface area that allows easy transport of mobile ions across the electrochrome-electrolyte interfaces. Thus, more ATO sites can be quickly accessed during the charge injection and expulsion process. See Section 18.5.1 for the discussion on the origin of the enhanced contrast ratio of ATO nanophase materials. An increased level of electronic states at the grain boundaries may also be responsible for the color formation of ATO electrochromic displays.
At annealing temperatures <450°C, the incorporation of antimony into the SnO2 nanocrystallites and grain boundaries may not be complete. Incomplete dehydration to form appropriate metal oxide lattices and insufficient oxygen evolution which is necessary for antimony reduction to provide electron charge carriers may also contribute to the observed low contrast ratio. The resistivity of the ATO nanophase materials is also too high for electrochromic processes to give a high contrast ratio.
At very high annealing temperatures, the total number of grain boundaries is reduced; the total surface area of accessible sites is significantly reduced; and the amount of antimony available to provide electron charge carriers is also reduced. Therefore, the contrast ratio decreases with annealing temperature for temperatures >750°C as shown in Fig. 18.14.
18.5.3 Other Factors That Affect the Contrast Ratio
18.5.3.1 Effect of ATO Loading
The contrast ratio of printed electrochromic displays is influenced by many interacting parameters as shown in Fig. 18.9. For example, the contrast ratio varies with the loading of ATO nanocrystallites onto oxide powders. With alumina powders as support, we found that the contrast ratio maximizes with a loading at about 75 wt% ATO (43 mol% Sb + 57 mol% Sn) onto alumina (Fig. 18.18). Beyond this loading level, the substrate does not have sufficient scattering to make the bleached state appear bright. With lower loading level, the ATO coating onto the non-conducting alumina may not be optimized.
293
Figure 18.18 Variations of contrast ratio with ATO (43 mol% Sb and 57 mol% Sn) loading level onto alumina powders.
18.5.3.2 Effect of Substrate
The contrast ratio also depends on the types of substrate for supporting the ATO nanocrystallites. A variety of light-color powder materials can be used as support for ATO nanocrystallites. We have produced electrochromic displays using various powder materials such as titania microparticles, alumina powders, silica shells, barium sulfate powders, aluminum borate powders, and calcium fluoride powders (Coleman, et al., 1999a, 1999b). Figure 18.19 shows the effect of using different substrates on the contrast ratio of ATO electrochromic displays. The loading of ATO nanocrystallites (43 mol% Sb + 57 mol% Sn) onto the various substrates is 75 wt%.
Figure 18.19 Variations of contrast ratio with types of substrate.
294
18.5.3.3 Effect of Precipitation Sequence
The structure of antimony-tin oxide with high levels of doping has been the subject of discussion in the literature. Various upper limits of antimony doping for forming homogeneous solid solutions have been reported (Berry, 1981; Cox, et al., 1982). Most of the discrepancies may originate from the different sample preparation methods used to produce the ATO materials.
We found that coprecipitation of antimony and tin is crucial to maximizing the contrast ratio of ATO electrochromic displays. The optical and electrical properties of three samples prepared using the same ingredients but with different precipitation sequence are shown in Table 18.2. The electrochromic contrast is clearly better for the "homogeneous" sample where the antimony can make the greatest electronic contribution to material properties.
Table 18.2 Effect of order of metal hydroxides deposition on contrast ratio*
Order of deposition Crystallite size (nm) Resistivity (Ω · cm) Contrast ratio
Sb First 5.5 6.1 2.3
Sb First 3.2 133.0 3.0
Simultaneous 5.0 11.1 6.8
* 75 wt% ATO-25 wt% alumina; Sb (43 mol%) Sn (57 mol%); 600°C/3 h
295
Since all three samples clearly show different properties, we can infer that the antimony and tin did not completely interdiffuse under the annealing conditions (600°C for 3 h). With antimony precipitated first, partially doped tin dioxide nanocrystallites coated the outside of large antimony oxide particles as shown in Fig. 18.20. This sample has the lowest electrical resistance of all three samples. Since the tin dioxide nanocrystallites were not fully doped with antimony, the contrast ratio of the electrochromic display should be low as we discussed in Section 18.5.1.
Figure 18.20 High-resolution transmission electron microscopy image of ATO nanophase materials annealed at 600°C for 3 h. The precursor materials were prepared by a sequential precipitation process (antimony first).
When tin is precipitated first, small SnO2 nanocrystallites may be coated with a layer of antimony oxide; thus the smallest crystallites are generated during annealing because of the inhibition of crystal growth by the antimony oxide layer at moderate annealing temperatures. A separate antimony pyrochlore phase was observed which may be responsible for the relatively high electrical resistance.
Homogeneous distribution of high levels of antimony dopant in the ATO nanophase materials is required for optimal electrochromic properties.
18.5.3.4 Effect of pH Value
Because the ATO synthesis starts with molecular precursors in a solution form and hydrolysis and condensation reactions occur during the reaction processes, the pH value of the metal cation solution plays
296
an important role in determining the structure of the ATO precursor precipitates. To understand the effect of solution pH value on the contrast of the electrochromic displays, a series of ATO nanophase materials were prepared with different pH values. Table 18.3 shows the electrical, structural, and electrochromic properties of these ATO materials. As the pH value of the solution is increased from 2 to 7, sodium ions are exchanged into the ATO nanocrystallites, resulting in the formation of a separate, electrically insulating ilmenite phase. With increasing pH value, the sodium content increases; the electrical resistivity increases significantly; the average size of the ATO nanocrystallites decreases; and the contrast ratio of the ATO electrochromic display declines. With increasing pH value, the color of the annealed ATO powders also changes from dark gray to light gray.
Table 18.3 Effect of solution pH value on contrast ratio*
pH
ATO crystallite size
(nm) Resistivity (Ω · cm) Contrast ratio Color
Sodium Ion (wt%)
Chloride (ppm)
Other crystalline
phase
2 5.0 13 4.3 Dark gray 0.5 842
5 4.2 672 3.2 Medium gray 1.4 61 Ilmenite
7 3.8 1.3 × 106 1.6 Light gray 2.5 247 NaSbO3
* 75 wt% ATO-25 wt% alumina; Sb (43 mol%) Sn (57 mol%); 600°C/3 h
18.5.3.5 Effect of Adding Salts
Since the electrochromic performance of the ATO nanophase materials depends on quick access of ions to the metal oxide sites and the nanostructure of the electrolyte-electrochrome interface, adding salts to the metal oxide precursor could help create ionic pathways during annealing. The effect of adding NaCl to the oxide precursor on the contrast ratio and electrical resistivity of the synthesized ATO nanophase materials is shown in Table 18.4. The addition of NaCl to ATO precursors resulted in higher electrical resistance, lighter color of the ATO nanophase materials, and lower contrast ratio of the electrochromic displays. HRTEM images revealed the presence of NaCl nanocrystallites in the agglomerates of ATO nanocrystallites. The preferential aggregation of small NaCl particles along the grain boundaries of the ATO nanophase materials could increase the electrical resistance. The addition of salts could also affect the electronic band structure of the interface states and, thus, adversely modify the electrochromic properties of the ATO nanophase materials.
Table 18.4 Effect of adding NaCl to ATO precursors on contrast ratio*
NaCl added (%) ATO crystallite size
(nm) Resistivity (Ω · cm) Contrast ratio Color
297
0 5.0 12 6.8 Dark gray
1 4.8 33 7.0 Dark gray
5 4.5 290 4.5 Light gray
10 4.5 5680 3.4 Light gray
* 75 wt% ATO-25 wt% alumina; Sb (43 mol%) Sn (57 mol%); 600°C/3 h
18.6 Summary
Nanophase antimony-tin oxide materials have unique electrochromic properties. The enhanced optical absorption and fast switching process can be attributed to the presence of many grain boundaries, the small size of ATO nanocrystallites, and the high surface area accessible to injected charges. The nanostructure of the synthesized ATO nanophase materials determines their optical and electrical properties.
The enhanced contrast of ATO electrochromic displays originates from the presence of Sb5+ ions within individual ATO nanocrystallites: upon electrochemical reaction, photon-assisted intervalence transition between Sb5+ ions and Sb3+ ions within ATO nanocrystallites gives the intense dark color observed in reflective ATO electrochromic displays.
Flexible, printed electrochromic displays can be produced by an inexpensive printing technology. The novel "side-by-side" design, using interdigitated electrodes, permits the production of large-area displays with sophisticated patterns at a much lower cost than that of the conventional "sandwich" design using rigid, optically transparent electrodes.
References
Baucke F. G. K. Materials Science and Engineering. B10, 285 (1991)
Bendert R. M. and D. A. Corrigan, J. Electrochem. Soc.. 136, 723 (1989)
Berry F. J. Advances in Catalysis. 30, 97 (1981)
Berry F. J. and D. J. Smith. J. Catalysis. 88, 107 (1984)
Brotherston I. D., Z. Cao, G. Thomas, P. Weglicki and J. R. Owen. Solar Energy Materials and Solar Cells. 39, 257 (1995)
Caldararu M., A. Ovenston, D. Sprinceana, J. R. Walls and N. I. Ionescu. Applied Catalysis. A141, 31 (1996)
298
Calvert J. M., T. J. Manuccia and R. J. Nowak. J. Electrochem. Soc.. 133, 951 (1986)
Carroll A. F. and L. H. Slack. J. Electrochem. Soc.. 123, 1,889 (1976)
Chopra K. L., S. Major and D. K. Pandya. Thin Solid Films. 102, 1 (1983)
Coleman J. P., A. T. Lynch, P. Madhukar and J. H. Wagenknecht. Solar Energy Materials and Solar Cells. 56, 375 (1999a)
Coleman J. P., A. T. Lynch, P. Madhukar and J. H. Wagenknecht. Solar Energy Materials and Solar Cells. 56, 395 (1999b)
Coleman J. P., J. J. Freeman A. T. Lynch, P. Madhukar and J. H. Wagenknecht. Acta Chemica Scandinavica. 52, 86 (1998)
Cox P. A., R. G. Egdell, C. Harding, W. R. Patterson and P. J. Tavener. Surface Science. 123, 179 (1982)
Dautremont-Smith W. C. Displays. 3, 3 (1982)
Deb S. K. Applied Optics Suppl.. 3, 192 (1969)
Deb S. K. Solar Energy Materials and Solar Cells. 39, 191 (1995)
Donnadieu A., D. Davazoglou, and A. Abdellaoui. Thin Solid Films. 164, 333 (1988)
Egdell R. G., W. R. Flavell, and P. Tavener. J. Solid State Chemistry. 51, 345 (1984)
Goyal D. J., C. Agashe, B. R. Marathe, M. G., Takwale and V. G. Bhide. J. Applied Physics. 73, 7,520 (1993)
Granqvist C. G. Applied Physics. A57, 19 (1993)
Granqvist C. G. Solar Energy Materials and Solar Cells. 32, 369 (1994)
Hamberg I. and C. G. Granqvist. J. Applied Physics. 60, R123 (1986)
Haranahalli A. R. and D. B. Dove. Appl. Phys. Lett.. 36, 791 (1980)
Hensch L. L. and J. K. West. Chem. Rev.. 90, 33 (1990)
Hinokuma K., K. Ogasawara, A. Kishimoto, S. Tanaka and T. Kudo. Solid State Ionics. 53–56, 507 (1992)
Holland L. Vacuum Deposition of Thin Films. (Chapman and Hall, London, 1956)
Hush N. S. Prog. Inorg. Chem.. 8, 357 (1967)
Jarzebski Z. M. and P. J. Morton. J. Electrochem. Soc.. 123, 299C (1976)
Kojima M., H. Kato, and M. Gatto. Philosophical Magazine. B68, 215 (1993)
Kojima M., H. Kato, and M. Gatto. Philosophical Magazine. B73, 277 (1996)
299
Liu J., J. P. Coleman, and P. Madhukar. In: Surface-Controlled Nanoscale Materials for High-Added-Value Applications. eds. by K. E. Gonsalves, M-I. Baraton, R. Singh, H. Hofmann, J. X. Chen, and J. A. Akkara. Mat. Res. Soc. Symp. Proc.. Vol. 501 (1998) pp. 47
Mastragostino M. In: Applications of Electroactive Polymers. ed. by B. Scrosati. Chapman and Hall, London, (1993), 223
McAteer J. J. Chem. Soc. Faraday Trans.. I 75, 2,768 (1979)
Mishra K. C., K. H. Johnson and P. C. Schmidt. Physical Review. B51, 13, 972 (1995)
Miyata N. and H. Kitahata. Thin Solid Films. 125, 33 (1985)
Monk P. M. S., R. J. Mortimer, and D. R. Rosseinsky. Electron Chromism: Fundamentals and Applications. VCH, Weinheim (1995)
Mortimer R. J. Chemical Society Reviews 26, 147 (1997)
Mulla I. S., H. S. Soni V. J. Rao and A. P. B. Sinha. J. Mater. Sci.. 21, 1,280 (1986)
Oi T. Ann. Rev. Mater. Sci. 16, 185 (1986)
Orel B., U. Lavrencic-Stangar, and K. Kalcher. J. Electrochem. Soc.. 141, L27 (1994)
Oyabu T. J. Applied Physics. 53, 2,785 (1982)
Pyke D., R. Reid and R. J. D. Tilley. J. Solid State Chemistry. 25, 231 (1978)
Robin M. B. and P. Day. Adv. Inorg. Chem. Radiochem. 10, 247 (1967)
Rogers D. and A. C. Skapski. Proc. Chem. Soc.. 400 (1964)
Scrosati B. In: Applications of Electroactive Polymers. ed. by B. Scrosati, Chapman and Hall, London, (1993), 250
Shannon R. D. Acta Crystallography. A32, 751 (1976)
Shanthi E., V. Dutta, A. Banerjee and K. L. Chopra. J. Applied Physics. 51, 6,243 (1980)
Skapski A. C. and D. Rogers. Chem. Commun.. 611 (1965)
Suzuki K. and M. Mizuhashi, Thin Solid Films. 97, 119 (1982)
Wrighton M. S. and D. C. Bookbinder. J. Electrochem. Soc.. 130, 1, 081 (1983)
Ziegler J. P. and B. M. Howard. Solar Energy Materials and Solar Cells. 39, 317 (1995)
300
19, Engineered Microstructures for Nonlinear Optics
19.1 Introduction
Artificial superlattices are engineered microstructures. Like bulk materials, they can be divided according to their constituents into several categories: semiconductor, metallic, dielectric, etc. These new materials show many unusual behaviors, which are of fundamental interest in physics and have potential applications in microelectronics. In the past several decades, the most studied are the semiconductor and metallic superlattices. With the development of modern science and technology, research on dielectric superlattices (DSLs) has been possible only recently. As is well known, In dielectric crystals, the most important physical processes are the propagation and excitation of classical waves (optical and acoustic waves). The propagation of classical waves in a DSL (classical system) is similar to the electron motion in a periodic potential of crystal lattice (quantum system). Thus, some ideas in solid state electronics, for example, the reciprocal space, Brillouin zone, dispersion relation, and the like, may be used in classical wave processes. Such is the case for photonic crystal (Yablonovitch, 1987), which is important for applications such as suppressing spontaneous emission, novel laser geometries, etc (Joannopoulos, et al., 1995). Associated with the photonic crystal is the variation of dielectric constants. Another DSL, a periodic elastic composite known as the phononic crystal, has also attracted much interest (Espinosa, et al., 1998). Apart from the band structure, attention has been paid to phenomena such as Anderson localization and possible applications such as acoustic filters and new transducers (Kushwaha, et al., 1994; Sigalas, 1997). The modulation parameters may not be restricted to those mentioned above. For example, the nonlinear optical coefficients can be engineered, leading to quasi-phase-matching or quasi-phase-matched (QPM) frequency conversion more efficient than by a birefringence phase-matching (PM) method (Feng, et al., 1980; Fejer, et al., 1992). In fact, research on DSLs for nonlinear optics may be dated back to 1962, when Bloembergen et al. (Armstrong, et al., 1962) proposed the idea that nonlinear optical phenomena could be enhanced enormously by the modulation of nonlinear optical coefficients in the dielectric material. Great progress has been made along this direction since then.
In this Chapter, an overview will be given of the studies on DSLs for nonlinear optical effects. Since there are reviews of progress in periodic poling methods for second-harmonic generation (SHG) (Houe and Townsend, 1995) and in QPM nonlinear interactions (Byer, 1997), much has been focused on the work done in our Lab, especially on coupled nonlinear optical frequency conversion in quasiperiodic structures and optical bistability. Although periodically poled (PP) something (for example, PPLN, PPKTP, etc.) has been widely used in the literature, here we prefer to use DSL for these structures.
19.2 Preparation of DSLs
DSLs can be realized by the modulation of microstructures, such as by the modulation of domain structures, phase structures, compositions, by the modulation of crystallographic orientations and heteroepitaxy structures. The modern technology of crystal growth and microprocessing (Joannapoulos, et al., 1997; Ming,
301
et al., 1982a, b, 1980; Xue, et al., 1984; Chen, et al., 1989; Zhu, et al., 1993; Ming, 1982; Wang, et al., 1986; Lu, Y. Q., et al., 1996a; Lu, Y. L., et al., 1997; Hadimioglu, et al., 1987; Wong, et al., 1982), the assembly of small dielectric spheres (Wei, et al., 1993a, b; Wen, et al., 1999), the atomic-layer-controlledepitaxy and heteroepitaxy (patterned growth by MBE, MOCVD) (Nishizawa, 1991; Ando, et al., 1991; Byer, 1997), or the use of acousto-optical effect (Feng and Ming, 1989), electro-optical effect (Zhu and Ming, 1992) or photorefractive effect (Xu and Ming, 1993a; Wang, et al., 1996a) make it possible to fabricate 1-D, 2-D and 3-D DSLs. Below we will describe the preparation of DSL by the modulation of ferroelectric domains and by photorefractive effect.
19.2.1 Preparation of DSLs by Modulation of Ferroelectric Domains
The DSL was first realized by Feng (1980) and Ming et al. (1982a) in LiNbO3 (LN) crystals with periodic ferroelectric domains (periodic DSL) by the growth striation method. Since then, much progress has been made towards the fabrication of such a structure either in a bulk material or in an optical waveguide material through a variety of techniques (Feisst and Koidl, 1985; Magel, et al., 1990; Lim, et al., 1986; Webjorn, et al., 1989; Ito, et al., 1991; Harada and Nihei, 1996; Risk and Lau, 1996; Batchko, et al., 1999). Among them, room temperature poling technique (RTPT) (Yamada, et al., 1993; Webjorn, et al., 1994; Zhu, S. N., et al., 1995; Zhu, Y. Y., et al., 1997a; Chen and Risk, 1994; Myers, et al., 1995; Ito, et al., 1995; Karlsson, et al., 1996) has attracted much attention due to its capability to fabricate, when combined with photolithography, domain structures with any pre-designed patterns. Figure 19.1 is a schematic of the structure obtained by RTPT.
Figure 19.1 Schematic diagram of a periodic DSL.
Since the coercive field of bulk LN and LiTaO3 (LT) is very high, it is not easy to reverse their spontaneous polarization at room temperature (Nassau and Levinstein, 1965). Although Camlibel (1969) measured the spontaneous polarization of LN and LT using a pulsed electric field, it was Yamada et al. (1993) who fabricated successfully for the first time a periodically poled LN with thickness of about 0.2 mm by RTPT. Recently, much effort has been paid towards the fabrication of more thicker samples as well as other materials, such as LT, KTiOPO4 (KTP), Sr0.6 Ba0.4 Nb2O6 (SBN), RbTiOAsO4 (RTA), BaTiO3, KNbO3, etc. (Webjorn, et al., 1994; Zhu, et al., 1995, 1997a; Chen and Risk, 1994; Myers, et al., 1995; Ito, et al., 1995; Karlsson, et al., 1996; Meyn, et al., 1999; Setzler, et al., 1999).
302
The applied electric field depends on temperature as well as on stoichiometry. In 1972, Ballman and Brown (1972) experimented with a pulsed field technique to invert single-domain LT crystals at different temperatures. It was found that the field required decreased with increasing temperature. Up to now, most of the reported domain inversion kinetics in these crystals has been for the congruent composition, since it is easy to grow and is available commercially with high quality. Recent progress in growth technique makes it possible to grow LN, LT with stoichiometric content. The fields for the domain reversal in the stoichiometric crystal are much lower than those for the congruent crystal. It was found that the electric field for domain reversal in stoichiometric LN is about one fifth of that in the congruent LN (Gopalan, et al., 1998). Whereas for LT, the field is about one thirteenth (Kitamura, et al., 1998). This may enable the fabrication of bulk devices (thickner than a few millimeters) with better mechanical stability and performance.
It was found that polarization reversal by RTPT is as follows (Zhu, et al., 1995). For a plane electrode sample, the inverted domain first randomly nucleates at the interface between wafer and electrodes and imperfections in the interior in the form of needles, while for a sample with periodic electrode on its surface, the nucleation first occurs under the periodic electrode. Then it grows both forwardly and laterally. If the electric pulse applied in the poling process is too long, gap regions between the electrodes will be filled by the broadening of the inverted domains.
On the other hand, domain reversal in both uniform-electrode and patterned-electrode ferroelectrics is driven by the total electric field. The total electric field is comprised of the sum of an applied external field, a depolarization field due to the unscreened portion of the polarization charge, and an internal field that has been attributed to nonstoichiometric point defects, a surface dielectric gap, and bulk charges. The presence of the internal field results in axial anisotropy of the coercive field, leading to spontaneous backswitching upon abrupt removal of the external field. Batchko et al. (1999) used an improved electric field poling technique, which incorporates this phenomenon, to obtain uniform short-period domain structure essential for SHG of blue and UV light.
The modulated domain structure thus prepared is characterized by the fact that the direction of spontaneous polarization of domains is parallel to the domain boundaries. There is an another type of periodic domain structure where the direction of spontaneous polarization of domains is perpendicular to the domain boundaries, which can only be prepared by the growth striation technique (Feng, et al., 1980, 1986; Ming, et al., 1982a; Feisst and Koidl, 1985; Magel, et al., 1990; Wang and Qi, 1986; Xu, et al., 1992), for example, the Czochralski method. Both types of domain structure show potential applications also in acoustic devices (Xu, et al., 1992; Zhu, Y. Y., et al., 1988a, b, 1989, 1992, 1993, 1995, 1996a, b, c; Cheng, et al., 1995a, b; Chen, et al., 1997).
Growth striation method has been used to grow LN, LT, Ba2NaNb5O15 and TGS (triglycine sulphate) periodic DSL single crystals (Feng, et al., 1980, 1986; Ming, et al., 1982a; Wang and Qi, 1986; Xu, et al., 1992). Taking LN as an example, in the process of crystal growth, the melt is doped with solutes, such as yttrium, indium or chromium, with concentration about 0.1 wt%—0.5 wt%, meanwhile a temperature fluctuation is introduced into the solid-liquid interface either through an eccentric rotation or through the application of an alternating electric current, which results in the growth striations, i. e., the regular variation in dopant concentration of growing crystal. When cooling through the Curie point, crystals with superlattice are formed automatically (Ming, et al., 1982a). The periods of the superlattice may be adjusted by choosing
303
suitable pulling rate and rotation frequency or by changing the period of the modulating electric current. In order to reveal the relationship between solute fluctuation and the ferroelectric domain structures, energy dispersive X-ray analysis has been used to measure the solute fluctuation in growth striations. Experimental result shows that the direction of spontaneous polarization of ferroelectric domain is determined by the gradient of solute concentration in growth striations. The fact may be explained as follows. The nonuniform distribution of solute, which in a crystal is generally ionized, is equivalent to the nonuniform space-charge distribution in the crystal and a nonuniform internal electric field is produced in it. It can induce the metallic ions within the lattice to displace preferentially at a temperature close to the Curie point (Ming, et al., 1982a).
Another advantage of the growth method is that it is easy to dope some active ions into the crystal during growth process for the design of multifunctional laser devices (Lu, et al., 1996a, b; Zheng, et al., 1998).
The domain structure can be observed using several methods. One most commonly used to reveal the structure is to etch the sample. Recently environment scanning electron microscopy (Zhu and Cao, 1997), scanning-tip microwave near-field microscopy (Lu, et al., 1997), scanning force microscopy (Eng, et al., 1998), coherent X-rays (Hu, et al., 1998) and even polarized or unpolarized light microscopes (Gopalan, et al., 1996) have been applied successfully to observe the domain structure nondestructively.
19.2.2 Preparation of DSL by Using Photorefractive Effect
Usually domain inversion does not change the dielectric constants of the material. In order to modulate the dielectric constants of a DSL, other methods should be used, such as by volume holographic record and thermal fixation. In this case, 1-D, 2-D or 3-D DSL (or say, photonic crystals) can be realized. In our experiments, we have used an oxidized Fe-doped LN (0.1 wt% Fe) single domain crystal for holographic record. A beam of the blue line of an argon-ion laser is split into three nearly-equal-intensity ones, S1, S2, and S3, which are recombined and interfered to cause spatial variations of the refractive index. Thus a 2-D square periodic grating can be recorded into the crystal as shown in Fig. 19.2(a). The oxidized Fe: LN sample exhibits high resistance to relax ation and the grating remains stable after the writing beams are removed. If a beam of an argon-ion laser is split into four equal-intensity ones and an S4 is applied parallel to the normal of the sheet of Fig. 19.2(a), a 3-D cubic DSL may be formed (Xu and Ming, 1993a; Wang, et al., 1996a, b).
Figure 19.2 (a) Optical geometry for recording a 2-D refractive index grating into a Fe: LN single domain crystal. (b) Schematic diagram of four-wave diffraction. Iin is the incident wave satisfying the exact Bragg condition of four-wave diffraction.
304
19.3 Outline of the Nonlinear Optics
It is well known for intense fields that the induced polarization P of the material should be expressed as (Yariv and Yeh, 1984)
where Pi is the ith component of the spontaneous polarization (i = x, y, z); ε0 is dielectric constant of vacuum; χij is linear susceptibility; dijk is second-order nonlinear optical coefficients; El is electric fields (l = 1,2,3 refer to fundamental, second harmonic and third harmonic, respectively); χijkl is third-order nonlinear optical susceptibility.
The nonlinear optical response characterized by the parameter tensors dijk and χijkl gives rise to numerous interesting phenomena and applications. The second-order nonlinearity is responsible for second-harmonic generation, for sum- and difference-frequency generation, and for parametric amplification and oscillation. The third-order term figures in such diverse phenomena as third-harmonic generation, Raman and Brillouin scattering, self-focusing, optical phase conjugation and Kerr nonlinearity. The dijk coefficients are measured most often in SHG experiments. In that case, the dijk can be contracted as diJ with i = 1,2,3 and J = 1-6. Thus we can quantify the SHG performance for a particular material or crystal orientation to obtain
305
where diJ are second-order nonlinear optical coefficients.
The matrix can usually be simplified, since symmetry allows cancellation or equalization of certain coefficients. For example, for LN, the matrix is reduced to
The conversion efficiency of the SHG is given by
where η is conversion efficiency; π is modulation ratio of the two-dimensional nonlinear superlattice with π = nx/ny; deff is effective nonlinear coefficient; L is total length of the sample; e is speed of light in vacuum; nl is the refractive indices (l = 1,2,3 refer to fundamental, second harmonic and third harmonic, respectively); λ is wavelength of light; l is light intensity; Δk is wave vector mismatch in a homogeneous media with Δk = k2 - 2k1.
If a modulated structure is introduced into the crystal, then by using the Fourier transformation (Yariv and Yeh, 1984), a relation similar to Eq.(19.4) can be obtained
where ΔK is wave vector mismatch in a superlattice with ΔK = k2 - 2k1 - Gm (or Gm,n).
The only difference is that the phase mismatch Δk is changed into ΔK, in which a reciprocal vector provided by the modulated structure is involved.
19.4 Wave Vector Conservation
Usually wave vector conservation plays an important role in interactions between electromagnetic waves and media, whether the interactions are linear or nonlinear. One best known example is the Bragg condition in X-ray diffraction, where the wave vector conservation between the incident and diffracted X-rays is
306
fulfilled with a reciprocal vector provided by the crystal lattice. In optical range, more interesting phenomena exist related to the wave vector conservation.
In the linear interaction regime, when the light transmission in a 1-D DSL is considered, Bragg condition is important for constructive interference to take place. The treatment can be extended to light transmission in 2-D DSL. In that case, more than one Bragg condition, for example, two Bragg conditions can be fulfilled (Feng and Ming, 1989). It is worth pointing out that an X-ray can hardly satisfy two Bragg conditions at the same time in a crystal because the lattice periods are nonadjustable, whereas in a 2-D DSL the reciprocal vectors can be adjusted to make the incident waves satisfy the two conditions simultaneously.
Among the nonlinear interaction regime, the most studied are optical frequency conversion and Kerr-form nonlinearity. The former one is related to the second-order nonlinear process, the latter to the third-order nonlinear process.
In the optical frequency conversion process, the wave vector conservation involves light waves with different wavelengths. In that case, the wave vector conservation is just the so-called PM condition. In SHG process, the phase mismatch between the fundamental and the second harmonic can be compensated either using the birefringence of the crystal (birefringence PM method) or using the reciprocal vectors in a DSL (QPM method) (Armstrong, et al., 1962). The same concept can be extended to other nonlinear optical frequency conversion processes, such as the sum- and difference-frequency generation.
In the case of SHG, the PM condition can be obtained from Eq.(19.4). Clearly, when
where kl are wave vecitors (l = 1,2,3 refer to fundamental, second harmonic and third harmonic, respectively), the SH conversion efficiency grows quadratically along the propagation direction. This can be realized in birefringent crystals, where the two waves experience the same refractive indices, n2 = n1 and consequently propagate with the same phase velocity. If the PM condition is violated (Δk ≠ 0) then the SH conversion efficiency will oscillate sinusoidally. That is, energy will be transferred back and forth periodically with a coherence length
where lc is coherence length.
In QPM scheme a periodicity is introduced into the nonlinear media, which can be a linear refractive index modulation or a modulation of the nonlinearity.In its most efficient form QPM is obtained when the sign of the nonlinear coefficient is reversed every coherence length. It is the reciprocal vectors provided by the modulated structure that is used to compensate for the phase mismatch. The QPM condition is given by
307
where Gm are reciprocal vectors of one-dimensional periodic superlattices; Λ is period of the periodic superlattice, as can be seen in Eq.(19.5). Figure 19.3 depicts the relationship between the SHG and the length of the sample for the three different cases. The advantage of QPM over the birefringence PM is that it permits access to the highest effective nonlinear coefficients of the materials, thus providing greater conversion efficiency. In LN, QPM with all waves polarized parallel to the z axis yields a gain enhancement over the birefringence phase-matched process of (2d33/πd31)2 ≈ 23. Another significant advantage of QPM is that any interaction within the transparency range of the material can be noncritically phase matched at any temperature, even interactions for which birefringence PM is impossible (for example, LT crystals).
Figure 19.3 The relationship between the SH intensity (relative value) and wave vector mismatch.
When optical bistability is discussed in 1-D and 2-D DSLs with Kerr-form nonlinearity, the situation becomes quite different. In 1-D DSLs, the light transmission in the exact phase-matched condition (the incident wave vector satisfies the exact Bragg condition) corresponds to the forbidden state. It is the change of the phase mismatch between the propagating wave vector and the reciprocal vector provided by the periodic structure that brings the incident wave from a forbidden transmission state to an allowed state or from a low-transmission state to a high-transmission state (Xu and Ming, 1993b, 1994; Agranovich, et al., 1991; He and Cada, 1991; Delyon, et al., 1986; Winful, et al., 1979). The bistability in a 1-D DSL is thus attributed to the phase-mismatch mechanism. Recent research work by present authors (Xu and Ming, 1993a, b, 1994; Wang, et al., 1996a, b, c, d, e, 1997; Chen, et al., 1995, 1996a, b), has revealed that a novel bistable mechanism, that is, the refractive index-modulation (RIM) mechanism, which is characteristically different from the phase-mismatch mechanism, exists in a 2-D DSL containing Kerr-form nonlinearity. In 2-D DSLs, when the transmitting light satisfies the exact Bragg condition of multiwave diffraction, whether the transmission state is located in the forbidden band or in the allowed band, is determined by the value of a set of parameters which are defined as the RIM strengths of the 2-D periodic structure. If the values of these parameters are arranged suitably, the light will propagate in the allowed band. Changes of these values will lead to high-or low-transmission states for each diffracted wave, since in the allowed band in the exact Bragg condition, the intensities of the multidiffracted waves are oscillation functions of these parameters.
308
19.5 Nonlinear Optical Frequency Conversion in 1-D Periodic DSLs
As early as 1980 (Feng, et al., 1980), we prepared for the first time a periodic DSL LN by Czochralski method. Later with the same method, we fabricated the first periodic DSL LT (Wang, et al., 1986). With these periodic DSLs, we verified the QPM theory proposed by Bloembergen et al. in 1962 (Armstrong, et al., 1962). Table 19.1 shows the results. The highest enhancement obtained is 16.9, close to the theoretical one, 23.
Table 19.1 The enhancement factors for DSL LN (QPM for d33 relative to the single domain crystal with PM for d31)
Length of crystal (mm)
Number of domains
SH intensity relative value (PM,
d31)
SH intensity relative value
(QPM, d33) Enhancement
factor
1 0.340 100 29.7 1.75 16.9
2 0.408 120 41 2.5 16.3
3 0.476 140 57 3.4 16.6
4 0.544 160 74.6 4.5 16.6
5 0.612 180 92 5.7 16.1
6 0.680 200 96.8 7 13.8
In 1985, Feisst and Koidl (1985) performed the experiment with an LN DSL of less period, prepared through the application of an alternating electric current during the growth process. Later, Magel et al. (1990) realized the blue light SHG in an LN fiber DSL. In 1990s, the QPM technique, spurred by the need for blue light laser sources for data storage, compact disc players etc., has made great progress (Feng, et al., 1980; Yamada, et al., 1993; Miller, et al., 1997; Mizuuchi, et al., 1997; Zhu, et al., 1995a, b, 1996, 1997a, b, 1998; Webjorn, et al., 1994; Chen and Risk, 1994; Myers, et al., 1995, 1996; Ito, et al., 1995; Karlsson, et al., 1996, 1998; Zheng, et al., 1998; Gupta, et al., 1994; Bierlein, et al., 1990; Arie, et al., 1998; Lu, et al., 1991, 1994a, b, 1996a, b, c, d, e; Englander, et al., 1997; Burr, et al., 1997; Missey, et al., 1998; Kartaloglu, et al., 1998; Reid, et al., 1997; Wang, et al., 1998; Mizuuchi and Yamamoto, 1992; Hadi, et al., 1997; Arbore and Fejer, 1997; Serkland, et al., 1997; Kintaka, et al., 1997; Eger, et al., 1997; Arbore, et al., 1997; Reid, et al., 1998; Chou, et al., 1998; Landry and Maldonado, 1997; Qin, et al., 1998). High-efficiency QPM SHG has been demonstrated in LN DSL and KTP DSL in both cw and pulsed regimes. For example, single-pass cw and quasi-cw SHG with efficiencies high as 42% (Miller, et al., 1997) and 65% (Pruneri, et al., 1996) were realized in LN DSLs, respectively. Internal conversion efficiency of 64% was achieved using a KTP DSL for single-pass SHG of high-repetition-rate, low-energy, diode-pumped lasers (Englander, et al., 1997). By placing a KTP DSL in an external resonant cavity, conversion efficiency of 55% was obtained for a cw Nd:YAG laser (Arie, et al., 1998). When used for optical parametric oscillator (OPO), DSLs show
309
advantages such as high gain, low threshold and engineerability of domain structures, thus making it possible to develop a robust, all solid-state, diode-pumped, miniaturized OPO (Myers, et al., 1996; Burr, et al., 1997; Missey, et al., 1998; Kartaloglu, et al., 1998; Reid, et al., 1997). For example, Batchko et al. (1998) demonstrated a 532 nm cw-pumped single-resonant OPO based on a 5.2 cm long first-order QPM-LN sample with a 64% quantum efficiency and a threshold less than 1 W. With all these achievements, it is possible, by intracavity frequency doubling the outputs (signal and idler waves), to generate blue and red light for display applications. For more detailed discussions on the SHG and OPO with the DSL, referr to the review article by Byer (1997).
Usually, DSL LN and LT are limited to periods not very short and to thicknesses ≤1 mm because of the side growth of inverted domains. This inhibits the use of DSL LN and LT in UV SHG and high-power pulsed OPOs. However, with the improvement of the poling technique and the use of new materials, this situation has been changed gradually. High-quality DSL LT with period as short as 1.7 µm was prepared for UV SHG (Mizuuchi, et al., 1997). DSL LN (Burr, et al., 1997), SBN (Zhu, et al., 1997a, b) and KTP (Wang, et al., 1998) with 1 mm thickness were fabricated successfully. The low coercive field of RTA, which allows poling of thick crystals (3 mm) (Karlsson, et al., 1998), together with its high damage threshold, low temperature dependence, and high resistance to photorefractive damage, makes this material more suitable for high-power applications.
The single pass SH conversion efficiency in bulk nonlinear devices is limited by diffraction spreading of the focussed laser beam. The conversion efficiency can be improved by confining the field to a waveguide. The guide wave interaction allows longer interaction distances at high field intensities by preventing diffraction beam spreading. Because of this, during the same period, waveguide DSLs have also attracted much attention (Lim, et al., 1989; Webjorn, et al., 1989; Mizuuchi and Yamamoto, 1992; Hadi, et al., 1997; Arbore and Fejer, 1997; Serkland, et al., 1997; Kintaka, et al., 1997; Eger, et al., 1997). New ideas such as balanced phase-matching was proposed (Bierlein, et al., 1990). High-efficiency QPM SHG was realized in waveguide DSLs (Mizuuchi and Yamamoto, 1992; Yi, et al., 1996).
Thus far in our lab, efficient pulsed SHG (including ns, ps, and fs) has been obtained with periodic DSL LN, LT, SBN samples prepared by RTPT or Czochralski method (Feng, et al., 1980; Wang, et al., 1986; Zhu, et al., 1995a, 1996, 1997a; Lu, et al., 1991, 1994a, b, 1996a, b, c). Table 19.2 shows the results obtained for several samples by using a picosecond automatic tunable OPO as a fundamental source (Lu, et al., 1996d). Continuous wave (CW) 0.35 mW, 405 nm blue light and CW 1.34 mW, 489 nm blue light generation by directly doubling an 810 nm GaAlAs laser diode and a 978 nm InGaAs laser diode, respectively, have been reported (Lu, et al., 1996a, b). Periodic DSL Nd: MgO: LN has been grown for self-frequency doubling (Lu, et al., 1996c, e). Visible dual-wavelength light generation through upconversion and QPM frequency doubling has been realized in erbium-doped periodic DSL LN (Zheng, et al., 1998).
Table 19.2 Parameters of DSL LN samples and results of SHG experiments by using an OPO as a fundamental source
Sample No. dthick (mm) Period (µm) N (periods) Fundamental
wavelength (µm) Efficiency (%)
310
1 0.62 2.8 220 0.815 3.0
2 0.78 3.4 230 0.860 4.2
3 1.56 5.2 300 0.980 24.0
4 2.20 6.4 310 1.026 17.0
5 1.50 8.3 180 1.130 19.8
19.6 Nonlinear Optical Frequency Conversion in 1-D QPDSLs
Before 1984, research work on superlattices was mainly focused on periodic structures. The discovery of quasicrystal has opened up a new field in condensed matter physics and therefore attracted much attention (Steinhardt and Ostlund, 1997; Janot, 1992). Then a question may be raised: Can this kind of material be of any practical use?
It is well known that the key to QPM is to construct a 1-D periodic structure that provides a reciprocal vector to compensate the mismatch of wave vectors due to dispersive effect of refractive index. This 1-D periodic structure can provide a series of reciprocal vectors, each of which is an integer times a primitive vector. In other words, the reciprocal vectors of a periodic DSL are determined by an integer and one structural parameter. Compared with the periodic structure, a 1-D quasiperiodic dielectric superlattice (QPDSL) has a low space group symmetry; but its symmetry is higher than that of an aperiodic structure. Its reciprocal vectors are governed by two integers and two structural parameters rather than by one integer and one structural parameter as in the case of the periodic one. Therefore, a QPDSL can provide more reciprocal vectors to the QPM optical parametric process. This fact makes the QPDSL more flexible in structure designing. Because of this, not only the QPM multiwavelength SHG but also some coupled parametric processes, such as the third-harmonic generation (THG) and fourth-harmonic generation, can be realized with high efficiency (Feng, et al., 1990; Zhu, et al., 1990, 1997a, b, 1998; Qin, et al., 1999). This is an example of possible applications of quasiperiodic structure materials in nonlinear optics.
19.6.1 The Construction of QPDSL
There are several types of quasiperiodic structures. The most famous ones are those with Fibonacci sequence and its generalizations, which can be generated by the concurrent inflation rule: A → Ap B and B → A (where p is positive integer) (Birch, et al., 1990). Superlattices with different quasiperiodic modulation have been predicted to exhibit different properties. Therefore, it is of interest to be able to fabricate and study such superlattices.
311
A typical generalized Fibonacci ferroelectric QPDSL is constructed from two building blocks A and B (Qin, et al., 1999). The arrangement of the two types of block follow the generalized Fibonacci sequence:
where (pSj-1) Sj-2 means a sequence of p (j - 1) st generation followed by the (j - 2) nd generation. When p = 1, the sequence is just the Fibonacci sequence.
In our case, each block is composed of a pair of oppositely polarized domains. The thickness of the positive ferroelectric domain in blocks A and B are lA1 and lB1, respectively. The thickness of the negative ferroelectric domain in block A is lA2 and that in block B is lB2. Here, lA1 is chosen to be equal to lB1, that is, lA1 = lB1 = l. Figure 19.4 shows an example of QPDSL with p = 1.
Figure 19.4 A QPDSL made from a LT single crystal. (a) Two building blocks: A and B, each composed of one positive and one negative ferroelectric domain. (b) Schematic diagram of a QPDSL with Fibonacci sequence.
19.6.2 Theoretical Treatment of the Nonlinear Optical Processes in QPDSLs
312
Here, we consider a case in which a single laser beam with ω1 = ω is incident from the left onto the surface of a QPDSL made from, for example, LN, propagates along the x axis with its polarization along the z axis in order to use the largest nonlinear optical coefficient d33 as shown in Fig. 19.4. Through the nonlinear optical effect, the SHG and THG exist simultaneously in the QPDSL. Here these three electric fields, E1, E2, and E3 must be taken into account, which satisfy the wave equation:
where E = ∑l = 1, 2, 3 El, P = ∑l = 1, 2, 3 Pl, P = PL + PNL, µ0 is the magnetic permeability of vacuum, ε is dielectric constant, PL: linear polarization, PNL: nonlinear polarization. The parametric SHG and THG processes can be analyzed by solving the coupled nonlinear wave equations that describe the interaction of the three fields in the 1-D QPDSL structure:
where Al2 is proportional to the photon flux at ωl with I = 1, 2, and 3 f (x) is a quasiperiodic
function with
Here f (x) can be Fourier transformed to (Birch, et al., 1990)
313
with the reciprocal vector of one-dimensional quasiperiodic superlattices
where with p an integer, the p th "precious mean", and
Here m and n are two integers and D = γ(p)IA+IB. In fact, only the Fourier component that is phase-matched contributes significantly to the parametric interaction. If the non-phase-matched components are ignored, the coupled Eqs. (19.10) become:
with
Equation (19.14) can be solved analytically for the boundary conditions of A2(0) = 0 and A3(0) = 0 under a
small signal approximation, i.e.,
314
with
Under QPM conditions:
Equation (19.15) can be simplified to
Using the following relation, we can get the harmonic intensity (Shen, 1984):
315
If k2A1L is very small, then we can get the conversion efficiency (Zhu, et al., 1997a):
Usually under the small signal approximation, the second harmonic generated through the second-order nonlinear process and the third harmonic generated through the third-order nonlinear process are proportional to L2 (Armstrong, et al., 1962). Here because of the coupling effect between the SHG and sum-frequency generation processes, the third harmonic generated through the second-order nonlinear process is proportional to L4. Thus it is possible to obtain THG in a QPDSL with high efficiency. In addition, it can be seen clearly that the effective nonlinear optical coefficients play an important role in nonlinear parametric processes. Thus it is helpful to derive the expression for the effective nonlinear optical coefficients.
From Eq. (19.14), if we let A3 = 0 and under the small signal approximation, we can get:
From this result, the second-harmonic spectrum can be obtained and the peak positions of the SHG are determined by ΔK1 = 0, the QPM condition Eq. (19.16).
19.6.3 The Effective Nonlinear Optical Coefficients
For QPM SHG in a periodic DSL, the conversion efficiency is proportional to the square of the effective
nonlinear optical coefficient, which is represented by with I the length of the reverse
domain (Fejer, et al., 1992). n = 1 corresponds to the first-order QPM with the largest conversion efficiency. For QPM THG in a QPDSL, the situation becomes much more complex. In order to see the relationship, it is needed to obtain the expression of the effective nonlinear optical coefficient for QPDSL.
For the QPDSL, the modulated nonlinear coefficient d(x) = d33f(x). According to Eq. (19.13), the effective nonlinear coefficient is
316
where Xm.n = πD-1(1+γ(p))(mIA-nIB). Obviously, the value of dm.n depends strongly on the adjustable structure parameters, such as I and the ratio IA/IB. Note that when p = 1, the result is just the one for Fibonacci sequence. When IA/IB, the QPDSL turns back to a periodic DSL and Eq. (19.21) reduces to
with m = n, the effective nonlinear coefficient for the periodic DSL.
19.6.4 QPM Multiwavelength SHG (Zhu, et al., 1990, 1997b; Qin, et al., 1999)
To verify the theoretical predictions, two types of QPDSL have been fabricated: one with p = 1, the so-called Fibonacci type and the other one a generalized QPDSL with p = 2. The sample was fabricated by poling a z-cut LT single domain wafer at room temperature (Zhu, et al., 1995).
The SHG spectrum of the QPDSL LT was measured with the fundamental tuned in the infrared. With the QPDSL samples, we obtained QPM second harmonic blue, green, red and infrared light output with conversion efficiencies up to about 5%—20%. Table 19.3 shows the results for m = 1. The measured and calculated results are in good agreement.
Table 19.3 Experimental results of SHG spectrum in the QPDSL LT
Reciprocal vectors
Fundamental wavelength (µm)
Harmonic wavelength
(µm) Input energy Output energy FWHM (nm) Efficiency (%)
Gm, n Cal. meas. meas. (µJ) (µJ)
(3,4) 0.9720 0.9726 0.4863 40 3 0.3 7.5
(2,3) 1.0820 1.0846 0.5423 40 7 0.4 17.5
(1,2) 1.2830 1.2834 0.6417 33 3 0.85 9.1
(2,1) 1.3640 1.3650 0.6825 30 2 1.1 6.7
(1,1) 1.5687 1.5699 0.7845 54 11 2.5 20.4
It is theoretically predicted and experimentally observed that the X-ray diffraction and Raman spectra of quasiperiodic superlattice exhibit self-similarity. However, in the SHG spectra of the QPDSL, by careful analysis to the measured spectrum, the self-similarity no longer holds. This is due to the dispersive effect of the refractive index, although the reciprocal vector does in reciprocal space.
As has been pointed out that here the QPM multiwavelength SHGs are wholly determined by the distribution of the reciprocal vectors of the QPDSL. Therefore with the aid of Fourier transformation, arbitrary wavelength-response functions can be obtained by design of appropriate DSL (Chou, et al., 1999).
317
QPM structures with multiple phase-matching wavelengths can be used for wavelength-division-multiplexed wavelength conversion.
Using the SH spectra obtained in QPDSL either with p = 1 or p = 2, the dispersion relationship of the refractive index has been deduced (Zhu, et al., 1997b; Qin, et al., 1999).
19.6.5 Direct THG (Feng, et al., 1990; Zhu, et al., 1997a, 1998; Qin, et al., 1999)
THG has a wide application as a means to extend coherent light sources to short wavelengths. The creation of the third harmonic directly from a third-order nonlinear process is of little practical importance because of the intrinsic low third-order optical nonlinearity. Conventionally, an efficient THG was achieved by a two-step process. Two nonlinear optical crystals are needed: the first one for SHG and the second one for sum-frequency generation (Shen, 1984). In this regard, QPDSL has some advantages over the conventional method. Here only one crystal is needed and the harmonic generation can be realized using the largest nonlinear optical coefficient over the entire transparency range of the material with high efficiency. For the QPDSL, the QPM conditions for THG in a collinear interaction are given by Eq. (19.16).
THG was tested with a tunable optical parametric oscillator. Several THGs have been detected. However, only one TH has a high conversion efficiency (>20%). Others are very weak. Theoretical analysis has been shown that the THG can be generated through the nearly QPM SFG with the SHG either quasiphase matched or mismatched. However, an efficient third harmonic can be generated only when both the SHG and SFG processes are quasiphase matched.
Recently, QPM THG has been demonstrated using a simple silica structure of six modulation periods through cubic nonlinearity (χijkl) (Williams, et al., 1998). Continuous-wave frequency tripling by simultaneous three-wave mixings has been realized in a periodically poled LiNbO3 (LN) crystal (Pfister, et al., 1997). In an LN waveguide, ultraviolet THG of 355 nm has been observed (Kintaka, et al., 1997).
With the aid of RTPT, we have performed a systematic study on QPDSL SBN related to SHG and THG (Zhu, et al., 1997a, b, 1998).
19.7 Optical Bistability in a 2-D DSL
Optical wave propagation in a dielectric structure with spatially periodic refractive index has been a topic for a long time. The study can be divided into two classes: linear regime and nonlinear regime. In linear regime in which the dielectric constant is independent of the optical field intensity, physical phenomena including inhibition of spontaneous emission and energy transfer (Yablonovitch, 1987), strong localization of light (John, 1987), and photon-atom bound states (John and Wang, 1990), can occur in these materials. These materials are known as photonic band gap materials or photonic crystals. However, if nonlinearity is introduced into the structure, novel phenomena such as optical bistability (Xu and Ming, 1993a, b, 1994; Wang, et al., 1996a, b, c, d, e; Agranovich, et al., 1991; He and Cada, 1991; Delyon, et al., 1986; Winful, et
318
al., 1979; Chen, et al., 1995, 1996a, b), gap solitary waves (Chen and Mills, 1987; Mills and Trullingre, 1987; Sterke and Sipe, 1988, 1989), self-induced transparency (Kozhekin and Kurizki, 1995), and ultrashort pulse propagation (Scalora, et al., 1994) will emerge. As such, these materials may be termed nonlinear photonic crystals.
Usually for optical bistability to take place in a medium, two necessary elements are expected to coexist: one is the nonlinear response element, requiring that the transmission light field respond in nonlinear form to parameters that are related to the transmission process; the other is the positive feedback element, requiring that the values of these parameters change with the transmission field. In common dispersive bistability in a Fabry-Perot etalon (Gibbs, et al., 1976) or in a 1-D superlattice (Agranovich, et al., 1991; He and Cada, 1991; Delyon, et al., 1986; Winful, et al., 1979), the related parameter is the mismatch of the propagating wave vector that detunes from the cavity-matching condition or the Bragg condition. The average energy density of the transmission field in the cavity or in the superlattice is an oscillation function of the parameter (Agranovich, et al. 1991; He and Cada 1991). This nonlinear response element, coupled with the feedback element that the wave vector's mismatch is established by the field's average energy density in the medium via the Kerr-form nonlinearity will produce bistability in the system's input/output relation. This kind of bistability is thus attributed to the phase-mismatched mechanism. However, in a 2-D DSL containing Kerr-form nonlinearity, theoretical and experimental work has revealed a novel bistable mechanism, i.e., the RIM mechanism (Xu and Ming, 1993a, b, 1994; Wang, et al., 1996a, b, c, d, e, 1997; Chen, et al., 1995, 1996b). Compared with the phase-mismatched mechanism, the related parameter in the RIM mechanism is not the wave vector's mismatch, but the RIM strength of the 2-D DSL.
Some preliminary work concerning the 2-D DSL has been done by our group since 1989 (Xu and Ming, 1993a, b, 1994; Wang, et al., 1996a, b, c, d, e, 1997; Chen, et al., 1995, 1996b). Light propagation in a 2-D DSL in which the refractive index is sinusoidally modulated in two dimensions was studied by Feng et al. (Feng and Ming, 1989). Taking nonlinearity of the medium into account, Xu et al. (Xu and Ming, 1993a, b, 1994) showed that a 2-D DSL can exhibit optical multistability and instability under exact Bragg conditions. Wang et al. (Wang, et al., 1996a, b, c, d, e, 1997) discussed the influence of the modification of RIM strengths and the effect of non-Bragg incidence on the optical response in these structures. Chen et al. treated the case of two incident waves (Chen, et al., 1995, 1996b).
19.7.1 Bloch Wave Approach (Xu and Ming, 1993b, 1994; Wang, et al., 1996c)
We consider a lossless 2-D DSL with a simple Kerr-form nonlinearity shown in Fig. 19.5(a). We assume that the linear refractive index of a 2-D nonlinear DSL is weakly, sinusoidally modulated in the x and y directions. Thus the refractive index of a 2-D nonlinear DSL is given by
Figure 19.5 Four-wave diffraction in a 2-D DSL. (a) Schematic of four-wave diffraction in real space. The dotted lines denote the incident-dependent RIM along the x direction formed by incident fields Ein1 and Ein2. (b) Bragg condition with four reciprocal points located on the Ewald sphere.
319
Here n0 is the average refractive index, nα is the Kerr coefficient; Gx, Gy are reciprocal vectors of two-dimensional superlattices, nx, ny are refractive index modulation along the x and y directions of two-dimensional superlattices, taken to be rather weak. The 2-D DSL described by Eq. (19.22) can be fabricated by using a holographic recording technique (Xu and Ming, 1993a). Another such structure would be a doubly periodic planar waveguide where Eq. (19.22) describes the distribution of an effective mode index (Zengerle, 1987).
For a 2-D DSL, when an incoming wave Ein1 satisfies two Bragg conditions simultaneously, four diffracted waves can be excited as indicated in Fig. 19.5(a); the wave vectors of the four waves are Ko, Kh, K- o and K- h, respectively. In reciprocal space, these four wave vectors are on the Ewald sphere as shown in Fig. 19.5(b). There are two wave vector conservation relations:
From the above equations, two Bragg conditions can be expressed as:
where kB is the Bragg wave vector and θB is the Bragg angle.
At a specified frequency of incoming wave, the Bragg conditions set an angular range near the Bragg angle, within which multiwave dynamical diffraction may occur.
According to the above discussions, for an incoming wave with wave vector k polarized in the z direction in the vicinity of Bragg incidence the Bloch waves in a 2-D nonlinear DSL can be decomposed into four partial modes,
320
where Eσ: electric fields in two-dimensional superlattices (σ = o, - o, h, - h indicate the different directions).
A useful representation of the dispersion effect caused by multiwave interaction in a 2-D nonlinear DSL is to introduce a new set of axes (ξo, ξh) which are defined through (Feng and Ming, 1989)
with k = k . Then carrying manipulations similar to the ones used in the linear multiwave diffraction dynamics, we obtain a nonlinear matrix equation for Eσ,
The field-dependent RIM strengths in Eq. (19.27) are expressed as
where Mα = 2nα/no. To obtain Eq. (19.27), the approximations 1 - K2o/k2 ≈ - 2ξo, 1 - K2
h/k2 ≈ - 2ξh, 1 - K2-o/k2 ≈
2ξo, and 1 - K2-h/k2 ≈ 2ξh were used, which follow from inequalities nx no, and ny no. The relation
between ξo and ξh can be expressed as, in the vicinity of Bragg incidence,
These equations constitute the basis of our numerical calculations and discussions of the remaining parts of this paper. We will discuss them in detail in the following sections. Since the nonlinear response element of a medium can be found by checking the dynamics of light transmission in the corresponding linear dielectric system, we shall at first present some main results obtained in linear regime.
19.7.2 Four-Path Switch: Linear Case (Feng and Ming, 1989)
321
If the intensity of the incident optical wave is very weak, then the fourth term in Eq. (19.22) can be neglected. In other words, all the ΔMβ's (β = x, y, x + y, x - y, 0) in Eq. (19.27) equal zero. In this way, the medium is a linear medium. Setting the determinant of matrix Eq. (19.27) to zero gives a dispersion relation:
The solution for ξh is readily obtained:
where ρ is angular deviation of the incoming light from the Bragg angle, = is modulation ratio of the two-dimensional nonlinear superlattice with = = nx/ny. For =<1 and ρ in the range ρsin2θB <1/2My (1 - =2)1/2, Eq. (19.31) yields complex ξh values. According to Eqs. (19.26) and (19.29), this leads to complex Ko and Kh values, corresponding to evanescent Bloch waves, or the forbidden band. When =>1, real ξh values are always obtained from Eq. (19.31) for all ρ. That is, for =>1 we obtain propagating Bloch wave and in this case the intensities of the four diffracted waves of a 2-D DSL are oscillation functions of RIM strengths. In other words, although there are four diffracted beams, their strengths can be adjusted by modulating the modulation ratio of the 2-D DSL, =. That is, the incident beam can be diffracted into one of the four propagating directions (Fig. 19.5(a)) by appropriately modulating the value of =. This phenomenon can be used to fabricate a four-path switch in which the modulation can be realized by applying two acoustic waves along two orthogonal directions. The strength ratio can be obtained by controlling the intensity ratio of two acoustic waves.
Numerical calculation shows that all four diffracted intensities are oscillation functions of the RIM strengths when =>1. Such a dynamical behavior revealed in a 2-D DSL provides a new type of nonlinear response element. The related parameters here are the RIM strengths. It is expected that if a Kerr-form nonlinearity is considered, the values of the RIM strengths will be perturbed by the interference of the four diffracted waves in the transmission field. This is the feedback element. Thus it is to be expected that the bistability may be exhibited in the incident-diffracted relations of a 2-D DSL containing Kerr-form nonlinearity.
19.7.3 A New Type of Optical Bistability Mechanism: Nonlinear Case with One
Incident Wave (Xu and Ming, 1993a, b, 1994; Wang, et al., 1996a, b, c, d, e, 1997)
As the input power is increased, the Kerr-form nonlinearity in Eq. (19.22) should be taken into account. In this way, the refractive index of the 2-D DSL will be modulated by the optical waves through the terms ΔMβ in Eq. (19.27).
To investigate possible bistability of the RIM mechanism, we have applied a kind of self-consistent method to achieve the incident-diffracted relations. When the incident wave satisfying the Bragg condition excites four diffracted waves, the interference of the field will give perturbations to the values of the RIM strengths
322
via the Kerr-form nonlinear term. The perturbed RIM strengths will then return to affect the transmission field. Eventually this dynamical interaction between the transmission field and the RIM strengths will reach a stable state, i.e., the two are in a self-consistent manner. Such a stable convergent self-consistent solution can be easily obtained with numerical computation.
Figure 19.6 is one of our results that shows the intensities of four diffracted waves as functions of the incident intensity Iin. The shapes of the incident-diffracted curves are determined by the RIM strengths. There always appear discontinuous jumps of intensity in the bistable region and the jumps can be either from higher values to the lower or from the lower to the higher. Thus a hysteresis loop is traced out. The hysteresis loop width depends on Mx and My. The threshold for bistability here is comparable with that of the dispersive bistability in a Fabry-Perot etalon. This RIM mechanism for bistability is characteristic of multiwave diffraction cases in 2-D DSL. It is not exhibited in a 1-D superlattice because there is only one parameter of RIM strength in a 1-D superlattice; when the incident wave satisfies the Bragg condition, the transmissivity is a monotonous function of the parameter.
Figure 19.6 Relative intensities of four diffracted waves as functions of the incident intensity Iin (calculated).
The experiment was performed in a 2-D nonlinear DSL constructed by recording a 2-D refractive index grating into a photorefractive material, a Fe-doped LiNbO3 single crystal. The nonlinearity induced in this 2-D DSL can be proved to be in a form similar to the Kerr-type (Wang, et al., 1996b). Figure 19.2(b) is a schematic of four-wave diffraction. Figure 19.7 is one of the results recorded in the experiment. Bistable
323
behavior with discontinuous jumps of the diffracted intensities, that is, a hysteresis loop, can be seen clearly and the jumps occur simultaneously for four diffracted waves. The occurrence of bistable behavior can be understood by the following analysis. In the illumination of the incident wave, the trapped electrons in the space charge pattern of the original grating are excited and drift. Because there is an interference field existing in the medium due to the incident wave diffracted into four directions, the space charge pattern that the drifting of the free carriers is constructing coincides with the original one. By the electric field induced modulations on the refractive index, and because of the low illuminating intensity and the low erase sensitivity in this oxidized sample (Orlowski, et al., 1997), this redistribution of the trapped electrons relative to the former actually gives perturbations to the values of the original RIM strengths of the superlattice. The strengths of the perturbation depend on the incident intensity and they may be positive or negative, respectively, with different interference field. This fact provides the feedback element for the RIM mechanism, which is expected to exist in such a 2-D DSL. The change of the values of the perturbed RIM strengths causes a change of the transmission-diffraction field, and this changed field will return to affect the perturbations of the RIM strengths. Diffracted intensities will thus change nonlinearly with the incident intensity. In addition, this system can enter the regions of bistability, instability or chaos by means of either adjusting the incident intensity or changing the values of the RIM strengths (Xu and Ming, 1993a).
Figure 19.7 Intensities of diffracted waves recorded as functions of the incident intensity Iin. The arrows indicate where the switching occurs.
324
When the incident wave deviates from the Bragg condition, then something new appears. Figure 19.8 shows the calculated results under four different angles of incidence. In the figure, we only plot the intensity of one excited wave versus incoming intensity to illustrate the effect of non-Bragg incidence. For comparison, in Fig. 19.8, the results under Bragg incidence are also shown. We see that, a non-Bragg incidence leads to a large increase in the threshold for optical bistability. Moreover, optical bistability may disappear if the angular deviation is too large.
Figure 19.8 Influence of the angle of incidence on the optical response (calculated). Optical bistability disappears as the deviation increases. The angular deviation is ρ = 0 in (a); ρ = 0.29 mrad in (b); ρ = 0.58 mrad in (c); and ρc = 0.755 mrad in (d). No optical switching exists when ρ>ρc.
325
The physical reason for the occurrence of the above phenomenon is that under non-Bragg incidence the coupling of four diffracted waves in the structure is much reduced. In a 2-D nonlinear DSL, a simultaneous excitation of four diffracted waves is responsible for the existence of RIM optical bistability mechanism. This requires a nearly Bragg incidence. The limit in the angular deviation at which the system exhibits simple switching is characterized by a threshold that is called ρc.
The above work can be extended to a heterosuperlattice-junction structure (Chen, et al., 1996a). Optical limiting has been predicted theoretically in such a structure.
19.7.4 A New Type of Optical Bistability Mechanism: Nonlinear Case with Two
Incident Waves (Chen, et al., 1995, 1996b)
The above discussion involves only one incident wave. Chen et al. (1995) considered the case of two incident waves. In that case, two coherent incident waves Ein1 and Ein2, with the same incident Bragg angle, symmetrically fall down the 2-D DSL (see Fig. 19.5(a)). The interference formed by two incident waves in the medium is characterized by a periodic spatial variation of the intensity. When Kerr-form nonlinearity is considered, as shown in Fig. 19.5(a), the nonlinear response in the medium leads to the formation of incident-dependent periodic RIM along the x direction with its periodicity characterized by Gx in reciprocal space and with its strength proportional to the incident intensities. In the figure, the dotted lines denote the
326
incident-dependent RIM along the x direction. This kind of process is similar to that of volume grating formation. When the incident-dependent periodic RIM matches the preconstructed RIM (namely the dotted and the solid lines in Fig. 19.5(a) are overlapped), the total RIM strength along the x direction, and thus the effective π, is enhanced in proportion to the incident intensities. Therefore, in the case of two incident waves, for a 2-D DSL with its π less than 1, a sufficient increase of the two incident intensities will lead to the values of effective π greater than 1. That is to say, the two intense waves can bring their transmission from a forbidden transmission state into an allowed transmission state. Thus optical bistability occurs. For a 2-D DSL whose values of π is near but smaller than 1, a bit of enhancement of RIM strength along the x direction can lead to this kind of transition. Such a transition only requires a very low input power, which might be beneficial for constructing low power cost 2-D optical bistable devices. With two incident waves, optical bistability with very low threshold can also occur when the relative phase between the two waves varies slowly within one cycle.
19.8 Outlook
Superlattices has opened an area to the development of new synthetic materials that do not exist in nature. The flexibility in the choice of superlattice materials allows superlattices to exhibit a wide range of tailorable properties that are of interest for scientific and device purposes.
Recently, much interest has been aroused on new applications of DSLs, such as engineerable compression of ultrashort pulses in chirped-periodic DSL (Arbore, et al., 1997; Reid, et al., 1998), amplitude squeezing (Serkland, et al., 1997), wavelength division multiplexing (WDM) (Chou, et al., 1998) and cascaded nonlinearity (Landry and Maldonado, 1997; Qin, et al., 1998), etc.. Engineered QPM patterns also hold great promise for use in soliton systems. For example, soliton-based signal compression and shaping in QPM structures with longitudinal chirps has been proposed (Torner, et al., 1998). Spatial switching between different output soliton state has been predicted in QPM geometries with dislocations, tilts and wells (Clausen, et al., 1999). More recently, quadratic spatial solitons has been observed in a DSL LN (Bourliaguet, et al., 1999).
QPM also opens the search for better nonlinear media to new classes of materials, such as poled-polymer and fused-silica films, diffusion-bonded stacks of plates, laterally patterned semiconductors and asymmetric quantum wells (Byer, 1997).
Among them, silica and other glasses are particularly attractive, since they are dominant materials in information technology and in the development of fiber laser sources. These glasses offer high transparency, low cost, a high optical damage threshold, and straightforward integrability; moreover, rare-earth doping of glass fibers has allowed the development of important laser devices, such as erbium-doped fiber amplifiers and high-power cw and pulsed fiber lasers. Unfortunately, the inversion symmetry of the glass matrix prevents frequency conversion of coherent radiation through second-order parametric processes. The recent discovery that poling techniques can provide a permanent and large second-order nonlinearity in silica made it possible to implement QPM in glass and glass waveguides and fibers. Periodically poled glass waveguides and fibers are ideal for a wide range of QPM processes, such as frequency conversion of fiber lasers,
327
difference-frequency generation as a means for frequency conversion of telecommunication wavelengths, generation of correlated photon pairs by parametric processes for quantum cryptography, and cascading of second-order nonlinearities to produce equivalent third-order effects (self-and cross-phase modulation) for all-optical switching (Pruneri, et al., 1999; Bonfrate, et al., 1999). Recently, greater than 20%-efficient frequency doubling of 1532 nm nanosecond pulses has been realized in QPM germanosilicate optical fibers (Pruneri, et al., 1999).
The DSL can also be applied to beam control, beam focusing and beam steering (Chiu, et al., 1996; Yamada, et al., 1996). The ability to photolithgraphically pattern the DSL structure allows the consideration of prism and lens arrays for these applications. Beam deflection with prism arrays and focal length control, optical switching have been demonstrated successfully. These results show that the possibilities for extending the use of DSL to control optical beams is open for further exploration and development.
Apart from all of these, even coupling effects between motion of electrons, photons and phonons can be expected to exist in the DSL structures. For example, infrared absorption and polariton excitation is resulted from the coupling between lattice vibrations and electromagnetic waves in an ionic crystal. Recently, we established one-to-one correspondence between the real ionic crystal and the DSL when piezoelectric coefficient modulation is taken into consideration. Infrared absorption and polariton excitation in the DSL has been verified by experiments. From the similarity between the real ionic crystal and the DSL, other long wavelength optical properties, such as the Raman scattering and Brillouin scattering might also be expected. The only different is they occur in different frequencies. For example, Raman scattering appears in the THz region for a real ionic crystal; whereas it might appear in the GHz region for a DSL. Study on these effects is of fundamental interest in physics (Lu, et al., 1999).
References
Agranovich V. M., S. A. Kiselev, D. L. Mills. Phys. Rev.. B44, 10917 (1991) Ando S., S. S. Chang, T. Fukui. J. Cryst. Growth. 115, 69 (1991) Arbore M. A., M. M. Fejer. Opt. Lett.. 22, 151 (1997) Arbore M. A., A. Galvanauskas, D. Harter, M. H. Chou and M. M. Fejer. Opt. Lett.. 22, 1341 (1997) Arie A., G. Rosenman, A. Korenfeld, A. Skliar, M. Oron, M. Katz and D. Eger. Opt. Lett.. 23, 28 (1998) Armstrong J. A., N. Bloembergen, J. Ducuing and P. S. Pershan. Phys. Rev.. 127, 1918 (1962) Ballman A. A., H. Brown. Ferroelectrics 4, 189 (1972) Batchko R. G., M. M. Fejer R. L. Byer, D. Woll, R. Wallenstein, V. Y. Shur and L. Erman. Opt. Lett.. 24, 1293 (1999) Batchko R. G., D. R. Weise, T. Plettner, G. D. Miller, M. M. Fejer, R. L. Byer, Opt. Lett.. 23, 168 (1998) Bierlein J. D., D. B. Laubacher J. B. Brown and C. J. van der Poel. Appl. Phys. Lett.. 56, 1725 (1990) Birch J., M. Severin, U. Wahlstrom, Y. Yamamoto, G. Radnoczi, R. Riklund and L. R. Wallenberg. Phys. Rev.. B41, 10 398 (1990) Bonfrate G., V. Pruneri, P. G. Kazansky, P. Tapster and J. G. Rarity. Appl. Phys. Lett.. 75, 2356 (1999) Bourliaguet B., V. Couderc, Barthelemy, G. W. Ross, P. G. R. Smith, D. C. Hanna and C. de Angelis. Opt. Lett.. 24, 1410 (1999) Burr K. C., C. L. Tang M. A. Arbore and M. M. Fejer. Opt. Lett.. 22, 1458 (1997) Byer R. L. Nonlinear Opt.. 7, 234 (1994) Byer R. L. J. Nonlinear Optical Physics & Materials. 6, 549 (1997)
328
Camlibel I. J. Appl. Phys.. 40, 1690 (1969) Chen J., Q. Zhou, J. F. Hong W. S. Wang N. B. Ming and D. Feng. J. Appl. Phys.. 66, 336 (1989) Chen Q., W. P. Risk. Electron. Lett.. 30, 1516 (1994) Chen X. F., Y. L. Lu Y. Q. Lu and N. B. Ming.. SPIE 2897, 146 (1996a) Chen X. F., Y. L. Lu Y. Q. Lu and N. B. Ming. Chin. J. Phys.. 34, 1145 (1996b) Chen X. F., Y. L. Lu Z. L. Wang and N. B. Ming. Appl. Phys.. Lett. 67, 3538 (1995) Chen Y. F., S. N. Zhu Y. Y. Zhu N. B. Ming B. B. Jin, R. X. Wu. Appl. Phys. Lett.. 70, 592 (1997) Chen W., D. L. Mills. Phys. Rev. Lett.. 58, 160 (1987) Cheng S. D., Y. Y. Zhu, N. B. Ming. Ferroelectrics. 173, 153 (1995a) Cheng S. D., Y. Y. Zhu Y. L. Lu and N. B. Ming. Appl. Phys. Lett.. 66, 291 (1995b) Chiu Y., D. D. Stancil T. E. Schlesinger and W. P. Risk. Appl. Phys. Lett.. 69, 3134 (1996) Chou M. H., K. R. Parameswaran M. A. Arbore, J. Hauden and M. M. Fejer. In Conference on Lasers and Electro-Optics. Vol. 6 of 1998 OSA Technical Digest Series (Optical Society of America, Washington, D. C.) (1998), 475 Chou M. H., K. R. Parameswaran M. M. Fejer and I. Brener. Opt. Lett.. 24, 1157 (1999) Clausen, C. B., L. Torner. Opt. Lett.. 24, 7 (1999) Delyon F., Y. E. Levy, B. Souillard. Phys. Rev. Lett.. 57, 2010 (1986) Eger D., M. Oron, M. Katz, A. Reizman, G. Rosenman and A. Skliar. Electron. Lett.. 33, 1548 (1997) Eng L. M., H. J. Guntherodt, G. Rosenman, A. Skliar, M. Oron, M. Katz and D. Eger. J. Appl. Phys., 83, 5973 (1998) Englander A., R. Lavi, M. Katz, M. Oron, D. Eger, E. Lebiush, G. Rossenman and A. Skliar. Opt. Lett.. 22, 1598 (1997) Espinosa F. R. M. de, E. Jimenez, M. Torres. Phys. Rev. Lett.. 80, 1208 (1998) Feisst A., P. Koidl. Appl. Phys. Lett.. 47, 1125 (1985) Fejer M. M., G. A. Magel D. H. Jundt and R. L. Byer. IEEE J. Quantum Electron. 28, 2631 (1992) Feng D., N. B. Ming J. F. Hong J. S. Zhu, Z. Yang and Y. N. Wang. Appl. Phys. Lett.. 37, 607 (1980) Feng D., W. S. Wang, Q. Zhou and Z. H. Geng. Chinese Phys. Lett.. 3, 181 (1986) Feng J., N. B. Ming. Phys. Rev.. A40, 7047 (1989) Feng J., Y. Y. Zhu, N. B. Ming. Phys. Rev.. B 41, 5578 (1990) Gibbs H. M., S. L. McCall, T. N. C. Venkatesan. Phys. Rev. Lett.. 36, 1135 (1976) Gopalan V., T. E. Mitchell, Y. Furukawa and K. Kitamura. Appl. Phys. Lett.. 72, 1981 (1998) Gopalan V., M. C. Gupta. J. Appl. Phys.. 80, 6099 (1996) Gupta M. C., W. Kozloveski, A. C. G. Nutt. Appl. Phys. Lett.. 64, 3210 (1994) Hadi K. E., M. Sundheimer, P. Aschieri, P. Baldi, M. P. De Micheli, D. B. Ostrowsky and F. Laurell. J. Opt. Soc. Amer.. B 14, 3197 (1997) Hadimioglu B., L. J. La Comb, Jr., D. R. Wright, B. T. Khuri-Yakub and C. F. Quate. Appl. Phys. Lett.. 50, 1642 (1987) Harada A., Y. Nihei. Appl. Phys. Lett.. 69, 2629 (1996) He J., M. Cada. IEEE J. Quantum Electron. 27, 1182 (1991) Houe M., P. D. Townsend. J. Phys. D: Appl. Phys.. 28, 1747 (1995) Hu Z. W., P. A. Thomas, A. Snigirev, I. Snigireva, A. Souvorov, P. G. R. Smith, G. W. Ross and S. Teat. Nature. 392, 690 (1998) Ito H., C. Takyu, H. Inaba. Electron. Lett.. 27, 1221 (1991) Ito H., C. Takyu, M. Ohashi and M. Sato. Nonlinear. 14, 283 (1995) Janot C. Quasicrystals. Clarendon Press, Oxford, UK (1992) Joannopoulos J. D., R. D. Meade, J. Winn. Photonic Crystals. Princeton University, Princeton, NJ (1995) Joannopoulos J. D., P. R. Villeneuve, S. Fan. Nature. 386, 143 (1997) John S. Phys. Rev. Lett.. 58, 2486 (1987)
329
John S., J. Wang. Phys. Rev. Lett.. 64, 2418 (1990) Karlsson H., F. Laurell, P. Henriksson and G. Arvidsson. Electron. Lett.. 32, 556 (1996) Karlsson H., M. Olsson, G. Arvidsson, F. Laurell, U. Bader, A. Borsutzky, R. Wallenstein, S. Wickstrom and M. Gustafsson. In: Conference on Lasers and Electro-Optics. Vol. 6 of 1998 OSA Technical Digest Series. Optical Society of America, Washington, D. C.. (1998), 539 Kartaloglu T., K. G. Koprulu, O. Aytur, M. Sundheimer and W. P. Risk. Opt. Lett.. 23, 61 (1998) Kintaka K., M. Fujimura, T. Suhara, and H. Nishihara. Electron. Lett.. 33, 1459 (1997) Kitamura K., Y. Furukawa, K. Niwa, V. Gopalan and T. E. Mitchell. Appl. Phys. Lett.. 73, 3073 (1998) Kozhekin A., G. Kurizki. Phys. Rev. Lett.. 74, 5020 (1995) Kushwaha M. S., P. Halevi, Appl. Phys. Lett.. 64, 1085 (1994) Landry G. D., T. A. Maldonado. Opt. Lett.. 22, 1400 (1997) Lim E. J., M. M. Fejer R. L. Byer, Electron. Lett.. 25, 174 (1989) Lu Y. Q., Y. L. Lu, Q. Luo, Y. Y. Zhu, X. F. Chen, C. C. Xue and N. B. Ming. J. Phys. : Condens. Matter.. 9, 747 (1997) Lu Y. Q., Y. L. Lu C. C. Xue X. F. Chen J. J. Zheng and N. B. Ming. SPIE. 2897, 152 (1996a) Lu Y. Q., Y. L. Lu C. C. Xue J. J. Zheng X. F. Chen G. P. Luo N. B. Ming B. H. Feng and X. L. Zhang. Appl. Phys. Lett.. 69, 3155 (1996b) Lu Y. Q., Y. L. Lu G. P. Luo X. F. Chen C. C. Xue and N. B. Ming. Electron. Lett.. 32, 336 (1996c) Lu Y. Q., Y. Y. Zhu Y. F. Chen S. N. Zhu N. B. Ming and Y. J. Feng. Science. 284, 1822 (1999) Lu Y. L., Y. Q. Lu X. F. Chen G. P. Luo C. C. Xue and N. B. Ming. Appl. Phys. Lett.. 68, 2642 (1996a) Lu Y. L., Y. Q. Lu C. C. Xue and N. B. Ming. Appl. Phys. Lett.. 68, 1476 (1996b) Lu Y. L., T. Wei, F. Duewer, Y. Q. Lu N. B. Ming P. G. Schultz and X. D. Xiang. Science. 276, 2004 (1997) Lu Y. L., L. Mao, N. B. Ming. Appl. Phys. Lett.. 59, 516 (1991) Lu Y. L., L. Mao, N. B. Ming. Appl. Phys. Lett.. 64, 3092 (1994a) Lu Y. L., L. Mao, N. B. Ming. Opt. Lett.. 19, 1037 (1994b) Lu Y. L., Y. Q. Lu C. C. Xue X. F. Chen J. J. Zheng and N. B. Ming. SPIE. 2897, 357 (1996c) Lu Y. L., Y. Q. Lu X. F. Chen C. C. Xue and N. B. Ming. Appl. Phys. Lett.. 68, 2781 (1996d) Lu Y. L., Y. Q. Lu J. J. Zheng C. C. Xue X. F. Chen G. P. Luo and N. B. Ming. Appl. Phys. Lett.. 69, 1660 (1996e) Magel G. A., M. M. Fejer, R. L. Byer. Appl. Phys. Lett.. 56, 108 (1990) Meyn J. P., M. E. Klein, D. Woll, R. Wallenstein and D. Rytz. Opt. Lett.. 24, 1154 (1999) Miller G. D., R. G. Batchko W. M. Tulloch D. R. Weise M. M. Fejer and R. L. Byer. Opt. Lett.. 22, 1834 (1997) Mills D. L., S. E. Trullingre. Phys. Rev.. B36, 947 (1987) Ming N. B., J. F. Hong, D. Feng. J. Mater. Sci.. 17, 1663 (1982a) Ming N. B., J. F. Hong, D. Feng. Acta Phys. Sin.. (in Chinese) 31, 104 (1982b) Ming N. B., J. F. Hong Z. M. Sun and Y. S. Yang. Acta. Phys. Sin.. (in Chinese) 30, 1672 (1980) Ming N. B. Physical Fundamentals of Crystal Growth. Shanghai Scientific & Technical Publishers, Shanghai (in Chinese) (1982) Missey M. J., V. Dominic, J. Mckay, R. C. Eckardt, M. Vaidyanathan and L. E. Myers. In: Conference on Lasers and Electro-Optics. Vol. 6 of 1998 OSA Technical Digest Series. Optical Society of America, Washington, D.C., (1998), 540 Mizuuchi K., K. Yamamoto, M. Kato. Appl. Phys. Lett.. 70, 1201 (1997) Mizuuchi K., K. Yamamoto. Appl. Phys. Lett.. 60, 1283 (1992) Myers L. E., G. D. Miller R. C. Eckardt M. M. Fejer R. L. Byer and W. R. Bosenberg. Opt. Lett.. 20, 52 (1995) Myers L. E., R. C. Eckardt M. M. Fejer and R. L. Byer. Opt. Lett.. 21, 591 (1996) Nassau K., H. J. Levinstein. Appl. Phys. Lett.. 7, 69 (1965) Nishizawa J. J. Cryst. Growth. 115, 69 (1991)
330
Orlowski R., E. Kratzig, H. Kurz. Opt. Commun. 20, 171 (1997) Pfister O., J. S. Wells, L. Hollberg, L. Zink, D. A. Van Baak, M. D. Levenson and W. R. Bosenberg, Opt. Lett.. 22, 1211 (1997). Pruneri V., S. D. Betterworth, D. C. Hanna. Opt. Lett.. 21, 390 (1996) Pruneri V., G. Bonfrate, P. G. Kazansky D. J. Richardson N. G. Broderick J. P. de Sandro, C. Simonneau, P. Vidakovic and J. A. Levenson. Opt. Lett.. 24, 208 (1999) Qin Y. Q., Y. Y. Zhu S. N. Zhu and N. B. Ming. J. Phys.: Condensed Matter. 10, 8939 (1998) Qin Y. Q., Y. Y. Zhu S. N. Zhu G. P. Lu, J. Ma and N. B. Ming. Appl. Phys. Lett.. 75, 448 (1999) Reid D. T., Z. Penman, M. Ebrahimzadeh, W. Sibbett, H. Karlsson and F. Laurell. Opt. Lett.. 22, 1397 (1997) Reid D. T., P. Loza-Alvarez, M. Ebrahimzadeh, E. U. Rafailov, P. Faller, D. J. Birkin, W. Sibbett, H. Karlsson and F. Laurell. In: Conference on Lasers and Electro-Optics. Vol. 6 of 1998 OSA Technical Digest Series. Optical Society of America, Washington, D.C., (1998), 17 Risk W. P., S. D. Lau. Appl. Phys. Lett.. 69, 3999 (1996) Scalora M., J. P. Dowling C. M. Bowden and M. J. Bloemer. Phys. Rev. Lett.. 73, 1368 (1994) Scalora M., J. P. Dowling C. M. Bowden and M. J. Bloemer. J. Appl. Phys.. 76, 2023 (1994) Serkland D. K., Prem Kumar, M. A. Arbore and M. M. Fejer. Opt. Lett.. 22, 1497 (1997) Setzler S. D., P. G. Schunemann T. M. Pollak L. A. Pomeranz and M. J. Missey. In: Advanced Solid-State Lasers. Fejer, M. M., H. Injeyan and U. Keller. Vol. 26 of OSA Trends in Optics and Photonics Series. Optical Society of America, Washington, D.C., (1999), 676. Shen Y. R. The Principles of Nonlinear Optics. Wiley, New York, (1984) Sigalas M. M., J. Acoust. Soc. Am, 101, 1256 (1997) Steinhardt P. J., S. Ostlund. The physics of Quasicrystals. World Scientific, Singapore (1997) de Sterke C. M., J. E. Sipe. Phys. Rev.. A 38, 5149 (1988) de Sterke C. M., J. E. Sipe. Phys. Rev.. A 39, 5163 (1989) Torner L., C. B. Clausen, M. M. Fejer. Opt. Lett.. 23, 903 (1998) Wang S., H. Karlsson, F. Laurell. In: Conference on Lasers and Electro-Optics. Vol. 6 of 1998 OSA Technical Digest Series. Optical Society of America, Washington, D.C., (1998), 520. Wang W. S., Q. Zhou, Z. H. Geng and D. Feng. J. Cryst. Growth. 79, 706 (1986) Wang W. S., M. Qi. J. Cryst. Growth. 79, 758 (1986) Wang Z. L., J. Wu, Z. J. Yang Y. Y. Zhu and N. B. Ming. Solid State Commun.. 98, 1057 (1996a) Wang Z. L., J. Wu, Z. J. Yang Y. Y. Zhu and N. B. Ming. Chin. Phys. Lett.. 13, 440 (1996b) Wang Z. L., J. Wu, X. J. Liu Y. Y. Zhu and N. B. Ming. Phys. Rev.. B56, 9185 (1997) Wang Z. L., Y. Y. Zhu, N. Xu and N. B. Ming. J. Appl. Phys.. 80, 25 (1996c) Wang Z. L., Y. Y. Zhu Z. J. Yang and N. B. Ming. Phys. Rev.. B53, 6984 (1996d) Wang Z. L., Z. J. Yang Y. Y. Zhu and N. B. Ming. Opt. Commun.. 123, 649 (1996e) Webjorn J., F. Laurell, G. Arvidsson. J. Lightwave Technol.. 7, 1597 (1989) Webjorn J., V. Pruneri, P. St. J. Russell, J. R. M. Barr and D. C. Hanna. Electron. Lett.. 30, 894 (1994) Wei Q. H., X. H. Liu C. H. Zhou and N. B. Ming. Phys. Rev.. E48, 2786 (1993a) Wei Q. H., C. H. Zhou, N. B. Ming. J. Cryst. Growth. 130, 447 (1993b) Wen W. J., N. Wang, H. R. Ma Z. F. Lin W. Y. Tam C. T. Chan and P. Sheng. Phys. Rev. Lett.. 82, 4248 (1999) Williams D. L., D. P. West T. A. King, Opt. Commun. 148, 208 (1998) Winful H. G., J. H. Marburger and E. Garmire. Appl. Phys. Lett.. 35, 379 (1979) Wong H. K., G. K. Wong, J. B. Ketterson. J. Appl. Phys.. 53, 6834 (1982) Xu B., N. B. Ming. Phys. Rev. Lett.. 71, 3959 (1993a)
331
Xu B., N. B. Ming. Phys. Rev. Lett.. 71, 1003 (1993b) Xu B., N. B. Ming. Phys. Rev.. A50, 5197 (1994) Xu H. P., G. Z. Jiang, L. Mao, Y. Y. Zhu, M. Qi, N. B. Ming, J. H. Yin and Y. A. Shui. J. Appl. Phys.. 71, 2480 (1992) Xue Y. H., N. B. Ming J. S. Zhu and D. Feng. Chin. Phys.. 4, 554 (1984) Yablonovitch E. Phys. Rev. Lett.. 58, 2059 (1987) Yamada M., N. Nada, M. Saitoh and K. Watanabe. Appl. Phys. Lett.. 62, 435 (1993) Yamada M., M. Saitoh, H. Ooki. Appl. Phys. Lett.. 69, 3659 (1996) Yariv A., P. Yeh. Optical Waves in Crystals. John Wiley & Sons (1984) Yi S. Y., S. Y. Shin Y. S. Jin and Y. S. Son. Appl. Phys. Lett.. 68, 2493 (1996) Zengerle R. J. Mod. Opt.. 34, 1589 (1987) Zheng J. J., Y. Q. Lu G. P. Luo, J. Ma, Y. L. Lu, N. B. Ming, J. L. He and Z. Y. Xu. Appl. Phys. Lett.. 72, 1808 (1998) Zhu S. N., Y. Y. Zhu, N. B. Ming. Science. 278, 843 (1997a) Zhu S. N., Y. Y. Zhu Z. Y. Zhang, H. Shu, H. F. Wang, J. F. Hong, C. Z. Ge and N. B. Ming. J. Appl. Phys. 77, 5481 (1995a) Zhu S. N., W. W. Cao. Phys. Rev. Lett.. 79, 2558 (1997) Zhu S. N., Y. Y. Zhu Y. Q. Qin H. F. Wang C. Z. Ge and N. B. Ming. Phys. Rev. Lett.. 78, 2752 (1997b) Zhu S. N., Y. Y. Zhu W. Z. Shen H. F. Wang Z. Y. Zhang N. B. Ming, Y. Chang and X. C. Shen. J. Phys.. D29, 76 (1996) Zhu S. N., Y. Y. Zhu Z. J. Yang H. F. Wang Z. Y. Zhang J. F. Hong C. Z. Ge and N. B. Ming. Appl. Phys. Lett.. 67, 320 (1995b) Zhu Y. Y., J. F. Hong, N. B. Ming. Ferroelectrics. 142, 31 (1993) Zhu Y. Y., N. B. Ming. J. Phys.: Condens. Matter.. 4, 8078 (1992) Zhu Y. Y., J. S. Fu R. F. Xiao and G. K. L. Wong. Appl. Phys. Lett.. 70, 1793 (1997a) Zhu Y. Y., N. B. Ming, W. H. Jiang and Y. A. Shui. Appl. Phys. Lett.. 53, 2278 (1988a) Zhu Y. Y., N. B. Ming W. H. Jiang and Y. A. Shui. Appl. Phys. Lett.. 53, 1381 (1988b) Zhu Y. Y., N. B. Ming, W. H. Jiang. Phys. Rev.. B40, 8536 (1989) Zhu Y. Y., N. B. Ming. J. Appl. Phys.. 72, 904 (1992) Zhu Y. Y., N. B. Ming. Ferroelectrics. 142, 231 (1993) Zhu Y. Y., S. D. Cheng, N. B. Ming. Ferroelectrics. 173, 207 (1995) Zhu Y. Y., Y. F. Chen S. N. Zhu Y. Q. Qin and N. B. Ming. Materials Letters. 28, 503 (1996a) Zhu Y. Y., S. N. Zhu, N. B. Ming. J. Phys.. D 29, 185 (1996b) Zhu Y. Y., S. N. Zhu Y. Q. Qin and N. B. Ming. J. Appl. Phys.. 79, 2221 (1996c) Zhu Y. Y., R. F. Xiao J. S. Fu, G. K. L. Wong and N. B. Ming. Optics Letters. 22, 1382 (1997) Zhu Y. Y., R. F. Xiao J. S. Fu, G. K. L. Wong and N. B. Ming. Appl. Phys. Lett.. 73, 432 (1998) Zhu Y. Y., N. B. Ming. Phys. Rev.. B 42, 3676 (1990)
332
Appendix
Figure IV. 1 The most likely application of carbon nanotubes in recent years is field emission. Besides fundamentl studies, several companies have demonstrated prototype displays using carbon nanotubes. Shown in this figure is a dark-field TEM image displaying the field distribution at the tips of carbon nanotubes under an externally applied electric field (Courtesy of Dr. Ruiping Gao and Dr. Z. L. Wang, Georgia Institute of Technology).
Figure IV. 2 Quantum conductance of nanowires is an important and interesting physics phenomenon, and it is usually observed in ultra-thin nanowires at low temperatures. The conductance of a carbon nanotube was measured as a function of the depth with which the tube was inserted into a mercury bath (S. Frank, P. Poncharal, Z. L. Wang, and W. A. de Heer, Science 280 (1998) 1744). In-situ TEM observation confirmed this phenomenon. The most surprising fact is that a 20 nm diameter multi-walled carbon nanotube exhibts quantum conductance at room temperature. No heat dissipation was observed in the nanotube. This is the result of ballistic conductance, and it is believed to be a result of single graphite layer conductance due to the unique structure of graphite. It is necessary to point out that this phenomenon was observed only for carbon nanotubes that are structurally perfect and surface clean (Courtesy of Dr. Z. L. Wang, and Dr. W. A. de Heer, Georgia Institute of Technology).
333
Figure IV. 3 Nanobelts of semiconducting oxides are a group of materials for sensor and optoelectronic applications. Field effect transistors (FETs) can be fabricated using individual nanobelts (Z. W. Pan, Z. R. Dai, and Z. L. Wang, Science, 291 (2001) 1947). Shown here is an AFM image of the FET and its corresponding schematic diagram. The substrate is silicon, on which there is a thin layer of silica insulating that serves as the gate oxide. Gold electrodes were built on the chip using lithographic technique. By carefully placing a single nanobelt into the gold electrodes and ensuring good contact between the nanobelt and the electrode, an FET is built. By controlling the voltage applied between the backgate and the source, the current flowing from the source to the drain is controlled. The nanobelts are semiconductors and their conductance depends strongly on the molecules adsorbed on the surface. Based on this principle, nano-size sensors using single nanobelts have been fabricated (Courtesy of M. Arnold, Dr. P. Avouris, and Dr. Z. L. Wang).
334
Figure IV. 4 A large group of materials have been found to form nanowires, from metals and semiconductors to oxides and even to polymers. This TEM image shows typical morphology of CuS material, which consists of long needles and short rods. The needles have a diameter of 200 nm at their roots and 50 nm at tips. Rods have a diameter about 200 nm (Courtesy of Dr. Shihe Yang and Dr. Z. L. Wang).
Figure IV. 5 It is Known that nanocrystals melt at a much lower temperature than the bulk, but how does the melting start? One may say it starts at the surface. This dynamic process has been recently studied for gold
335
nanorods using femto- and nano-second laser. The as-synthesized nanorods are defect free (Wang et al., Surface Sci., 440 (1999) L809). After being irradiated with femto-second laser, (a, b) TEM shows that point defects have been created inside the volume. After being illuminated by nanosecond laser, (c) the nanorod has changed not only in its shape but also with a stacking fault, which is created to form the 111 surfaces. It is clear that the melting starts simultaneously at the surface and inside the bulk (S. Link, Z. L. Wang, and M. A. El-Sayed, J. Phys. Chem. B, 104 (2000) 7867).
Figure IV. 6 TEM image of silver/silica coaxial nanocables synthesized by directly coating silver nanowires with amorphous silica. The silver nanowires were synthesized using a polyol method that involved the reduction of silver nitrate with ethylene glycol in the presence of poly(vinyl pyrrolidone). The formation of uniform silica shell involved the base-catalyzed hydrolysis of tetraethyl orthosilicate (TEOS) and subsequent condensation of silica on the silver surface (Courtesy of Dr. Y. Xia et al., University of Washington, Nano Lett., 2 (2002) 427).
336
Figure IV. 7 Photonic crystals are meso-scale structures, and self-assembly is an important process for fabrication of photonic crystals. This figure shows an SEM image of an inverse opal fabricated by templating a sol-gel silicate precursor against opaline lattices of 380 nm polystyrene beads. The polymer beads have been selectively removed through etching with toluene (Courtesy of Dr. Y. Xia et al., University of Washington, Adv. Mater., 12 (2001) 206).
Figure IV. 8 Combining self-assembly with lithographic technique is an approach for future nanotechnology. This figure shows an SEM image of two linear chains self-assembled from 150 nm polystyrene beads by
337
dewetting of colloidal dispersions from a surface containing physical templates (a parallel array of trenches that were 150 and 150 nm in width and depth, respectively). The inset shows an SEM image of the templates that were fabricated using near field optical lithography with a binary phase shift mask (Courtesy of Dr. Y. Xia et al., University of Washington, J. Am. Chem. Soc., 123 (2001) 8718).