original - max planck societyferroelectric thin films. the growth of non-c-axis oriented epitaxial...

phys. stat. sol. (a) 202, No. 12, 2287–2298 (2005) / DOI 10.1002/pssa.200521176

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Original

Paper

Microstructure of (104)-oriented Bi3.25La0.75Ti3O12 and Bi3.54Nd0.46Ti3O12 ferroelectric thin films

on multiply twinned SrRuO3/Pt(111) electrodes

on YSZ(100)-buffered Si(100)

Dietrich Hesse*, Sung Kyun Lee, and Ulrich Gösele

Max-Planck-Institut für Mikrostrukturphysik, Weinberg 2, 06120 Halle (Saale), Germany

Received 30 May 2005, revised 8 July 2005, accepted 27 July 2005

Published online 5 September 2005

Dedicated to Professor Dr. Johannes Heydenreich on the occasion of his 75th birthday

PACS 68.37.Lp, 68.37.Ps, 68.55.Jk, 77.84.Dy

Uniformly (111)-oriented, multiply twinned SrRuO3-covered platinum electrodes have been grown on

YSZ(100) buffer layers on Si(100) substrates by a combination of r.f. sputtering and pulsed laser deposi-

tion (PLD). They provide a smooth and plane substrate surface for the growth of multiply twinned, uni-

formly (104)-oriented ferroelectric Bi3.25

La0.75

Ti3O

12 (BLT) and Bi3.54Nd0.46Ti3O12 (BNT) thin films grown by

PLD at an optimum substrate temperature of 750 °C. Microstructure, morphology and crystallographic

orientation of the SrRuO3/Pt electrodes and the BLT and BNT films are characterized by XRD, AFM,

TEM, and SAED. In spite of the multiply twinned structure, the entire ferroelectric film has a uniform

component P⊥ of the polarization vector perpendicular to the film plane. The (104)-oriented BLT and

BNT films on electroded Si(100) are shown to have good ferroelectric properties (remanent polarization,

coercive field, fatigue resistance) and are thus suitable for applications in silicon-based technologies.


1 Introduction

Metal-ferroelectric-metal (MFM) thin film structures play an important role in a number of applications, e.g., as tunable capacitors [1], ferroelectric memories [2, 3], or electroceramic-based microelectromecha-nical systems (MEMS) [4]. Their properties depend on nature, stoichiometry, phase contents, morpho-logy and microstructure of both the metal electrodes and the ferroelectric film, as well as on the structu-ral and electronic character of the top and bottom metal/ferroelectric interfaces [5, 6]. A widely used electrode-ferroelectric material combination is the one consisting of platinum electrodes and a simple perovskite-type ferroelectric, like BaTiO3 or Pb(Zr,Ti)O3 (PZT). However, material combinations of this type suffer from problems relevant to practical applications, especially from a considerable deterioration of the switchable polarisation with the number of switching cycles, a failure called fatigue [2, 7]. Two principal solutions to the latter problem are given by either replacing the platinum by an oxide electrode (RuO2, IrO2, SrRuO3) [8–10], or replacing the simple perovskite by a bismuth-layered perovskite (SrBi2Ta2O9, Bi4Ti3O12) [11, 12]. The present work is concerned with a mixed approach, considering the growth of epitaxial Bi4Ti3O12-related thin films on SrRuO3-covered platinum electrodes. Epitaxial ferroelectric films have a number of advantages over polycrystalline films. For example, they permit to study the structural and electrical anisotropies of the ferroelectric material in question avoiding the growth of single crystals. They will also be required for applications involving arrays of

* Corresponding author: e-mail: [email protected], Phone: +49 345 5582 741, Fax: +49 345 5511 223

2288 D. Hesse et al.: Microstructure of (104)-oriented Bi3.25

La0.75

Ti3O

4 and Bi

3.54Nd

0.46Ti

3O

12 ferroelectric films


ferroelectric cells of less than 100 nm lateral size, like nonvolatile ferroelectric random access memories of Gigabit memory density: Cell arrays prepared from polycrystalline films would suffer from cell-to-cell variations of the crystallographic orientation and thus of the memory properties [13, 14]. For thermo-dynamic reasons, epitaxial bismuth-layered perovskite films can easily be grown in c-axis orientation, i.e. with the (001) plane parallel to the film plane. However, these (pseudo-)orthorhombic materials have the vector of their (major) spontaneous polarization along the a-axis. Therefore, c-axis oriented films do not have a usable polarization component along the film normal, as required by most applications. This is the reason why worldwide efforts concentrate on the growth of non-c-axis oriented epitaxial bismuth-layered ferroelectric thin films, e.g. [15–25]. The present work considers epitaxial Bi4Ti3O12-related thin films of a specific non-c-axis orientation. Among Bi4Ti3O12-related materials, rare-earth substituted bismuth titanates are in the focus of inter-national research since 1999, when Park et al. had demonstrated that lanthanum-substituted Bi4Ti3O12 has a number of advantageous properties, like low fatigue, high remanent polarization, and low processing temperature [26]. Since then, numerous authors have studied Bi4Ti3O12 thin films substituting bismuth by different rare-earth elements. Substitution by Nd and La, respectively, has proven to be most effective, and non-c-axis oriented Nd- or La-substituted Bi4Ti3O12 films have been grown [20–24]. The most fa-vourable a-axis orientation has been obtained for BLT films on buffered Si(100) substrates, but it re-quires rather specific growth conditions [21]. The present work is part of our efforts to grow and study uniformly (104)-oriented Bi3.25La0.75Ti3O12 and Bi3.54Nd0.46Ti3O12 ferroelectric thin films. The growth of non-c-axis oriented epitaxial thin films of Bi4Ti3O12-related materials requires high temperatures in excess of 600 or 700 °C, at which the bottom electrode must be stable. In addition, the bottom electrode must serve as an epitaxial template and thus have a proper crystallographic orientation. Platinum has been demonstrated to grow epitaxially on oxide single crystals like MgO or SrTiO3 [25, 27, 28], but most applications require Si(100) wafers as substrates. Many ferroelectric films have been grown on textured Pt(111)/Ti/SiO2/Si(100) substrates, but BLT films grow either with random or with (001) orientation on these substrates [29]. On the other hand, high substrate temperatures used to im-prove the crystallographic orientation of ferroelectric films on Pt(111)/Ti/SiO2/Si(100) have been shown to result in interdiffusion and reaction processes that end up in a deterioration of the ferroelectric proper-ties [30, 31]. Platinum films directly grown on Si(100) substrates may react with the substrate (e.g., at temperatures above 700 °C), forming platinum silicides [32]. To prevent interdiffusion and reaction processes of this type, (100)-oriented yttria-stabilized zirconia (YSZ) buffer layers have proven most suitable. They can easily be grown epitaxially on Si(100) substrates [33]. The present work considers growth, microstructure and morphology of (111)-oriented, SrRuO3-covered platinum films on YSZ(100)-buffered Si(100) substrates; it also studies the possibility to use these well-oriented bottom electrodes as epitaxial templates for the growth of (104)-oriented Bi3.25La0.75Ti3O12 (BLT) and Bi3.54Nd0.46Ti3O12 (BNT) ferroelectric thin films on Si(100) substrates.

2 Experimental

Commercial Si(100) wafers were cut into pieces of 10 × 10 mm2. These pieces were heated to 800 °C in a substrate heater of a high-vacuum deposition stage, at which temperature they were covered by YSZ(100) buffer layers using pulsed laser deposition (PLD) in a flowing oxygen atmosphere of 2.4 × 10–4 mbar. PLD was performed employing a KrF excimer laser at a wavelength of 248 nm, a repeti-tion rate of 5 Hz, and an energy density of 1.7 . . . 3.4 J/cm2, using a stoichiometric ZrO2: 9 at% Y2O3 pressed powder target. Pt electrodes were deposited onto these YSZ(100) buffer layers by r.f. sputtering in an argon atmosphere of 2.4 × 10–3 mbar, at a substrate temperature of 400 °C and a r.f. power of 10 W, employing a platinum sputter target 2 inch in diameter. Epitaxial SrRuO3 intermediate layers, and ferro-electric BLT and BNT films were in turn deposited by PLD at the same laser conditions as for YSZ, but at different substrate temperatures and oxygen pressures (700 °C and 0.14 mbar for SrRuO3; 500 to 825 °C and 0.4 mbar for BLT and BNT). Here we will mostly limit ourselves to the results obtained with the optimum BLT and BNT substrate temperature of 750 °C. Pressed powder targets of the compositions SrRuO3, Bi3.75La0.75Ti3O12 (Bi-rich BLT) and Bi3.54Nd0.46Ti3O12 (BNT) were used for PLD.

phys. stat. sol. (a) 202, No. 12 (2005) / www.pss-a.com 2289


Original

Paper

The grown buffer, electrode/epitaxial template, and intermediate layers, as well as the ferroelectric films and the corresponding interfaces were investigated by X-ray diffractometry (XRD) in θ–2θ scans, φ scans, and φ−ψ pole figures applying a Philips X’Pert MRD four-axis diffractometer with CuK

α radia-

tion, and by transmission electron microscopy (TEM) with selected area electron diffraction (SAED), applying a Philips CM20T electron microscope at 200 keV primary beam energy. Atomic force micros-copy (AFM) was applied to investigate the surface topography of the grown films, applying a Digital Instruments D5000 with an ultrasharp tip (MikroMasch Noncontact Si NSC15/50, radius of curvature less than 10 nm, full tip cone angle less than 20°). The results of piezoresponse-AFM and ferroelectric analyses are not included into the present paper but will be reported separately.

3 Results and discussions

3.1 (111)-oriented SrRuO3/Pt electrode layers on YSZ(100)-buffered Si(100)

The cubic lattice parameters of Si, YSZ, and Pt are aSi = 5.43 Å, aYSZ = 5.14 Å, and aPt = 3.923 Å, respec-tively. SrRuO3 is orthorhombic, but can be considered pseudo-cubic, with a cubic lattice parameter of

pc

SROa = 3.928 Å, which is very close to aPt. Let us first consider the growth of Pt(111) on YSZ(100)-buffered Si(100). We are not aware of previ-ous work having shown that (111)-oriented Pt films can be grown on YSZ(100)-buffered Si(100); (111)-oriented Pt electrodes have been deposited and used as electrodes on YSZ(100) and YSZ(111) single crystals within electrochemical cells [34]. Figure 1a shows a XRD θ–2θ scan of a platinum film on YSZ(100)-buffered Si(100), revealing peaks for (111)-oriented Pt and (100)-oriented YSZ, without any foreign phases. (The very low-intensity

Fig. 1 XRD results obtained from a Pt(111)/YSZ(100)/Si(100) sample, and schematic explanation. (a)

θ–2θ scan; (b) pole figure (center ψ = 0°, rim ψ = 90°) using 2θ = 46.24°, corresponding to Pt(200);

(c) two φ scans at ψ = 55°, corresponding to Pt(200) and YSZ(111); (d) schematic showing the four azi-

muthal platinum domain variants on YSZ(100).


La0.75

Ti3O

4 and Bi

3.54Nd

0.46Ti

3O



peak at around 33° is at the position of the Si(200) reflection. The latter being kinematically forbidden, this peak most probably stems from the Si(400) Bragg reflection of half the wavelength of the CuK

α

radiation, the half wavelength being present in the white radiation background and passing the secondary monochromator.) A pole figure taken at 2θ = 46.24°, corresponding to the Pt(200) reflection, is shown in Fig. 1b. Twelve peaks with a peak-to-peak separation of φ = 30° are observed at ψ ≈ 55°, revealing a perfect out-of-plane orientation of the Pt film. (∠ (200); (111) = 54.7° for cubic platinum.) The good in-plane orientation of both the YSZ(100) buffer layer and the Pt(111) films is shown by two φ scans in Fig. 1c, both taken at ψ = 55°, one taken with the Pt(200) reflection, the other with the YSZ(111) reflection. The presence of twelve peaks in the platinum φ scan can be explained by help of the scheme in Fig. 1d. The fourfold symmetry of the YSZ(100) plane results in four types of azimuthal orientation variants (multiple twins) being present within the Pt film, in-plane rotated by 0°, 90°, 180°, and 270°. The Pt(111) surface in turn being of threefold symmetry, overall twelve reflections with a mutual φ separation of 30° result. The following orientation relationships were deduced from the XRD investiga-tions:

Pt(111) || YSZ(100) || Si(100); Pt[0 11] || YSZ〈001〉 || Si〈001〉 ,

taking into account that the Pt[0 11] direction can be parallel to any of the four [010], [0 10], [001], and [001] directions of YSZ (and Si). A similar orientation relation had been found for Pt(111) on MgO(100)

Fig. 2 TEM images and diffraction pattern of a Pt(111)/YSZ(100)/Si(100) sample. (a) Cross-section

bright field image with diffraction pattern (inset); (b) cross-section platinum dark field image; (c) magni-

fied inset from (a), with indexes and squares indicating reflections from Pt, YSZ and Si; (d) plan-view

bright field image; (e) plan-view platinum dark field image.



Original

Paper

[27, 28]. The lattice mismatch along Pt[2 1 1] || YSZ[001] amounts to –6.5%, that along Pt[0 11] || YSZ [010] to +8%. Figure 2a shows a cross-section bright field image, and Fig. 2b a corresponding platinum dark field image of a 20 nm thick Pt(111) film on a 35 nm thick YSZ(100) buffer layer on Si(100). The well-known columnar growth of YSZ is visible in the bright-field image and also in a corresponding YSZ dark field image (not shown), which shows that the column width is about 5 nm. The Pt dark-field image (Fig. 2b) reveals the Pt film consisting of large grains of about 80 to 150 nm diameter. The Si/YSZ and YSZ/Pt interfaces are sharp, smooth and plane, whereas the Pt surface seems to be somewhat wavy in correspondence to the grain structure. The inset in Fig. 2a shows a cross-section diffraction pattern, well aligned with respect to the micrograph. It is repeated in magnified form in Fig. 2c, together with the indexes of some of the reflections. The exact cube-on-cube epitaxy of YSZ(100) on Si(100) is clearly revealed, visualized by the two squares formed by the corresponding (400) and (040) reflections of YSZ and Si. Only one row of Pt reflections is seen, consisting of (111) and (222) Pt reflections on the vertical axis of the figure, which clearly shows that the Pt film is (111)-oriented. The absence of other Pt reflecti-ons is due to the four specific azimuthal Pt orientations which obviously do not give diffraction spots in the case of the specific beam direction used. Figures 2d, e show a plan-view bright field and Pt-dark field image, respectively, revealing the rather irregular shape of the Pt grains, which in fact are equivalent to azimuthal domains or azimuthal twins. A wide variation of the lateral size of these domains is visible, viz. between about 80 and 150 nm, some grains even reaching a diameter of 200 nm. Interestingly, most of the domain boundaries show a regular arrangement of dislocations which are certainly due to the re-gular character of the grain boundaries: In fact the latter are equivalent to regular twin boundaries between each two of the four azimuthal Pt twins. A more detailed characterization of the twin boundaries has not been performed so far. As will be explained in Section 3.2, a conducting, epitaxial SrRuO3(111) intermediate layer between the Pt(111) electrode and the (104)-oriented ferroelectric BLT or BNT film occured to be most useful with respect to the quality of the BNT and BLT films. Figure 3a shows a cross sectional TEM dark field image, and Fig. 3b a plan-view diffraction pattern of such a thin intermediate SrRuO3 layer on Pt(111) on

Fig. 3 TEM image and diffraction pattern of a

SrRuO3(111)/Pt(111)/YSZ(100)/Si(100) sample. (a) Cross-

section Pt/SrRuO3 dark field image; (b) plan-view diffrac-

tion pattern.


La0.75

Ti3O

4 and Bi

3.54Nd

0.46Ti

3O



Fig. 4 AFM images (3 µm × 3 µm each) of BNT films grown at different substrate temperatures. The

vertical (brightness) scale refers to the 825 °C sample only; other images have scales corresponding to

higher surface roughnesses. The root mean square roughness values were determined as 22.5 nm,

31.0 nm, 37.4 nmn, 40.5 nm, 49.5 nm, 16.1 nm, 27.3 nm, and 23.8 nm for the substrate temperatures of

500 °C, 550 °C, 600 °C, 650 °C, 700 °C, 750 °C, 800 °C, and 825 °C, respectively.

YSZ(100)-buffered Si(100). The TEM image shows that the surface of the SrRuO3 layer is smoother than the previous surface of the Pt(111) electrode. The diffraction pattern shows a quadratic spot pattern resulting from YSZ and Si, as well as two rings consisting of 24 reflections each. The cubic lattice param-eters of Pt (3.923 Å) and SrRuO3(3.928 Å) being so close, the rings may generally be due to both SrRuO3 and Pt. The Pt(110) ring being, however, kinematically forbidden, the inner ring in Fig. 3b is most pro-bably due to SrRuO3 only. (Since SrRuO3 is in fact non-cubic, the extinction rule for f.c.c. metals does not apply for SrRuO3.) A very faint (211) ring is also visible. As a detailed consideration of the non-cubic SrRuO3 crystallography (in analogy to Fig. 1d) shows, the 24 reflections in each of these rings can be explained by the presence of 12 azimuthal domain variants within the SrRuO3 film resulting from the threefold symmetry of each of the four azimuthal Pt(111) domain variants.

3.2 (104)-oriented Bi3.25La0.75Ti3O12 and Bi3.54Nd0.46Ti3O12 ferroelectric films

Bi4Ti3O12 crystallizes in a monoclinic lattice, which for simplicity can, however, be considered pseudo-orthorhombic. BLT and BNT have been reported to be orthorhombic. The corresponding lattice parameters are aBLT = 5.42 Å, bBLT = 5.415 Å, cBLT = 32.89 Å for Bi3.25La0.75Ti3O12 [35, 36], and aBNT = 5.4290 Å,



Original

Paper

bBNT = 5.4058 Å, and cBNT = 32.832 Å for Bi3.6Nd0.4Ti3O12 [37]. (104)-oriented ferroelectric BLT and BNT thin films have previously been shown to grow on uncovered and SrRuO3-covered SrTiO3(111) substrates [20, 22–24]. It could thus be expected that (104)-oriented BLT and BNT films will also grow on Pt(111). Although this expectation was generally confirmed in our experiments, we found out that a thin intermediate layer of SrRuO3(111) deposited between the Pt(111) electrode and the BLT(104) or BNT(104) film improved the degree of orientation, and also the ferroelectric properties. This is most probably due to the above-mentioned better smoothness of the SrRuO3(111) surface compared to the Pt(111) surface, and probably also to the better chemical stability of the SrRuO3/BLT and SrRuO3/BNT interfaces compared to the respective interfaces with platinum. Similar experience has been reported with (103)-oriented SrBi2Nb2O9 films on SrRuO3(111) on fiber-textured Pt/Ti/SiO2/Si substrates [38]. In the following, we will focus on BLT and BNT films grown on SrRuO3(111)-covered Pt(111)/YSZ(100)/ Si(100). We have not determined the exact chemical composition of our BLT and BNT films. However, con-sidering the composition of the targets given above, and the lattice parameters determined from XRD and SAED, we concluded that the composition is close to the nominal formulas Bi3.25La0.75Ti3O12 (BLT) and Bi3.54Nd0.46Ti3O12 (BNT), respectively. BLT and BNT films turned out to be most similar to each other, with respect to their orientation relationship, morphology and microstructure, although somewhat diffe-rent in the ferroelectric properties. The structure-related results of our investigations will thus be illustrated with figures mostly related to only one (or the other) of these films. AFM and XRD rocking curve investigations of BLT and BNT films grown at different substrate tem-peratures between 500 and 825 °C have been performed in order to establish the optimum substrate tem-perature. Out of these detailed investigations, Fig. 4 shows an AFM study of BNT films grown at differ-ent substrate temperatures. At substrate temperatures from 750 °C upwards, plate-like grains with poly-hedral shape are present, indicating a good crystal quality. For both BLT and BNT films, 750 °C has been found to be the optimum substrate temperature involving a minimum of the root mean square roughness (see Fig. 4). At this temperature, the average grain size is about 250 nm × 250 nm for BNT, and about 70 nm × 250 nm for BLT. Figure 5a shows a XRD θ–2θ scan of a BNT film grown at 750 °C. Apart from the Si(h00), YSZ(h00), Pt(hhh), and several CuKβ- and WLα-related subsidiary reflections, the BNT(104) reflection is clearly visible as the only BNT reflection. (The BNT(208) and (4016) reflections are hidden behind the Pt(111) and (222) reflections, respectively.) The BNT film has thus the (104) plane parallel to the substrate plane. Pole figures were taken at different 2θ values, from BNT and BLT films grown at vari-ous substrate temperatures. As two examples, Figs. 5b and c show pole figures of a BNT film grown at 750 °C, taken with (b) 2θ = 30.1° corresponding to the (117) BNT reflection, and (c) 2θ = 23.31° corres-ponding to the (111) BNT reflection. In Fig. 5b, four sets of 12 peaks are seen at ψ ≈ 36° and 84°, cor-responding to the (117)/(117) and (117)/(117 ) reflections, respectively (cf. the angles ∠ (104); (117) = ∠ (104); (117) = 36.4°; ∠ (104); (117) = ∠ (104); (117) = 84.1°). The four peaks at ψ ≈ 55° stem from the YSZ(111) reflection. The schematic Fig. 5d shows the location of four peaks that belong to one and the same azimuthal domain: The azimuthal angle difference between the (117) and (117) reflections is ∼134°. As a consequence, the peaks at ψ ≈ 36° in Fig. 5b have peak-to-peak separation angles of ∆φ ≈ 134° – (4 · 30°) = 14° and ([4 + 1] · 30°) – 134° = 16° between neighbouring peaks. Correspond- ingly the azimuthal angle difference between the (117) and (117) reflections of ∼67° results in peak-to-peak separation angles of ∆φ ≈ 67° – (2 · 30°) = 7° and ([2 + 1] · 30°) – 67° = 23° of the peaks at ψ ≈ 84°. In Fig. 5c, four sets of 12 peaks are seen at ψ = 50° and 59°, corresponding to the (111)/(111) and (111)/(111 ) reflections, respectively (cf. the angles ∠(104); (111) = ∠(104); (111) = 49.7°; ∠ (104); (111) = ∠ (104); (111 ) = 58.8°.) The schematic Fig. 5e shows the location of four peaks that belong to one and the same azimuthal domain: The azimuthal angle difference between the (111) and (111) reflections is ~134°, thus the peaks at ψ = 50° have peak-to-peak separation angles of ∆φ ≈ 134° – (4 · 30°) = 14° and ([4 + 1] · 30°) – 134° = 16° between neighbouring peaks. Correspondingly the azi-muthal angle difference between the (111) and (111) reflections of ~110° results in peak-to-peak separa-tion angles of ∆φ ≈ 110° – (3 · 30°) = 20° and ([3 + 1] · 30°) – 110° = 10° of the peaks at ψ = 59°. A detailed evaluation of these and other pole figures confirms the (104) orientation of the BLT and BNT


La0.75

Ti3O

4 and Bi

3.54Nd

0.46Ti

3O



Fig. 5 XRD results obtained from a BNT(104)/SrRuO3(111)/Pt(111)/YSZ(100)/Si(100) sample, and sche-

matic explanation. (a) θ–2θ scan; (b) pole figure (center ψ = 0°, rim ψ = 90°) using 2θ = 30.1°, corresponding

to BNT(117); (c) pole figure (center ψ = 0°, rim ψ = 90°) using 2θ = 23.31°, corresponding to BNT(111);

(d) schematic referring to (b); (e) schematic referring to (c). For details of the schematics, see text. films and the presence of 12 azimuthal domain variants within each film. The latter obviously are inher-ited from the 12 azimuthal domain variants within the SrRuO3 intermediate layer. Figure 6 shows corresponding φ scans of the entire BNT(104)/SrRuO3(111)/Pt(111)/YSZ(100)/ Si(100) heterostructure, using the reflections – from top to bottom in the figure – BNT(117)/(117) (ψ = 56.4°), BNT(0014) (ψ = 36.4°), Pt(200) (ψ = 54.7°), YSZ(111) (ψ = 54.7°), and Si(111) (ψ = 54.7°). The φ scans confirm the presence of three azimuthal BNT domain variants for each of the four azimuthal platinum domain variants, i.e. overall the presence of 12 azimuthal BNT domain variants. Based on all these XRD characterizations, the following epitaxial orientation relationships have bee established for both BLT and BNT films:

BNT(104) || SrRuO3(111) || Pt(111) || YSZ(100) || Si(100)

BNT[010] || SrRuO3[011] || Pt[011] || YSZ〈001〉 || Si〈001〉 .



Original

Paper

It should be pointed out that in all the 12 azimuthal domain variants the a-axis of BLT or BNT is at an angle of about 56° to the film plane, so that all of the domain variants have the same component P

⊥ of the

polarization vector perpendicular to the film plane1. Figure 7a shows a cross sectional TEM image of a 220 nm thick (104)-oriented BLT film grown at a substrate temperature of 750 °C on SrRuO3(111)/Pt(111)/YSZ(100)/Si(100). The BLT film consists of large elongated grains about 100 to 200 nm in lateral size. The surface morphology of the BLT film is determined by the shapes of the grains, resulting in a rather rough surface. Some voids of about 40 nm lateral size and 20 nm height are visible at the bottom of the Pt layer, which most probably result from the fact that the SrRuO3/Pt/YSZ/Si substrate had been heated to the rather high temperature of 750 °C during BLT deposition. Some recrystallization of the Pt layer may have occurred at this temperature, resulting in the condensation of some free volume into large voids. Figure 7b taken at higher magnifica-tion reveals the Bi2O2 layers or (002) planes within a BLT grain. These planes are at an angle of about 56° with the (104) plane, i.e. with the substrate plane. Planar crystal defects, most probably stacking faults, intergrowth defects, and/or out-of-phase boundaries, are clearly visible in the otherwise rather regular pattern of the (002) planes. For details of such specific crystal defects, which are well known from the bismuth-layered perovskite materials, see, e.g., [18, 39, 40]. Apart from these lattice defects, the well-pronounced parallel structure of the (002) planes confirms the good crystallinity of the BLT films. As a comparison of Figs. 7a and b shows, the (001) plane seems to be a favourable habit plane of the BLT grains, resulting in a tilted by about 55° appearance of the overal grain morphology. The inset of Fig. 7b shows the selected-area electron diffraction pattern of the grain seen in the image and its surroun-dings, revealing the narrow-spaced (00�) row of diffraction spots. For a particular direction of the elec-tron beam (i.e. for a particular sample tilt), this regular type of diffraction pattern showing the (00�) spots, and the corresponding images revealing the (002) Bi2O2 planes, can at best be seen in only 2/12 or 16.7% of the grains, because the other 83.3% of the grains have a different azimuth and thus the beam direction is different with respect to their lattice. Figure 8a shows a corresponding cross sectional image of a 240 nm thick (104)-oriented BNT film grown at 750 °C on SrRuO3(111)/Pt(111)/YSZ(100)/Si(100). The overall characteristics of the BNT film are similar to those of the BLT film: The BNT film, too, consists of large elongated grains of about 100 nm lateral size, and the surface morphology of the BNT film is also determined by the shapes of the grains, resulting in a very rough surface. The BNT grains have, however, a somewhat larger aspect ratio (length-

1 We do not consider the question of 90° a–b twins and especially of a possible a–b switchability in rare-earth substituted

bismuth titanate films, which is still a matter of controversy in literature.

Fig. 6 φ scans of a BNT(104)/SrRuO3(111)/

Pt(111)/YSZ(100)/Si(100) sample. The used reflec-

tions are indicated. For the corresponding ψ angles,

see the text.


La0.75

Ti3O

4 and Bi

3.54Nd

0.46Ti

3O



to-diameter ratio). Like for the BLT film, the (001) plane seems to be a favourable habit plane of the BNT grains; in Fig. 8a, however, two senses of the tilt of about 55° are visible, reflecting two (out of 12) different azimuthal domain variants. These two variants are related to each other as twins. No voids were found in the Pt layer of this sample. Figure 8b taken at higher magnification reveals the Bi2O2 layers or (002) planes within a single BNT grain, similarly to those within a BLT grain of Fig. 7b. Again these planes are at an angle of about 56° with the (104) plane, i.e. with the substrate plane. The inset of Fig. 8b shows the selected-area electron diffraction pattern of the grain seen in the image (and its surroundings), revealing the well-known narrow-spaced (00�) rows of diffraction spots. The difference of the two dif-fraction patterns of Figs. 7b and 8b stems from a slightly different beam direction (sample tilt), which in the case of Fig. 8b resulted in a low-index crystal direction (diffraction pole) of the BNT grain.

Fig. 7 Cross-section images and diffraction pattern

of a BLT(104)/SrRuO3(111)/Pt(111)/YSZ(100)/Si(100)

sample. (a) Overview; (b) magnified detail revealing

the BLT(002) planes, and corresponding diffraction

pattern (inset).

Fig. 8 Cross-section images and diffraction pattern

of a BNT(104)/SrRuO3(111)/Pt(111) /YSZ(100)/Si(100)

sample. (a) Overview; (b) magnified detail revealing

the BNT(002) planes, and corresponding diffraction

pattern (inset).



Original

Paper

The results of detailed characterizations of the ferroelectric properties of these (104)-oriented BLT and BNT films will be published separately. Here we just want to mention that these films show good ferro-electric P–E hysteresis loops with remanent polarization values of 2Pr = 26 µC/cm

2 and 2Pr = 38.5 µC/cm2, as well as coercive fields of 2EC = 160 kV/cm and 2EC = 210 kV/cm, for the BLT and BNT films, respectively. Fatigue measurements performed at a frequency of 1 MHz using a fatiguing electric field of 180 kV/cm revealed a good fatigue endurance up to 109 switching cycles, and a reduction of switched polarization by less than 10% after 1011 cycles. With all these values, the performance of our (104)-oriented BLT and BNT films on Si(100) is only slightly less good than that of (104)-oriented BLT and BNT films grown on SrRuO3-covered SrTiO3(111) (SCS111) single crystal model substrates. For example, the 2Pr value of Bi3.25La0.75Ti3O12 films grown by PLD on SCS111 was 31.9 µC/cm

2 [20], and that of Bi3.44La0.56Ti3O12 films grown by MOCVD on SCS111 was 34 µC/cm

2 [24]. The corresponding values for PLD-grown Bi3.15Nd0.85Ti3O12 films on SCS111 are 2Pr = 40 µC/cm

2 [22] and 2Pr = 50 µC/cm2

for MOCVD-grown Bi3.54Nd0.46Ti3O12 films on SCS111 [24]. Taking into account that SCS111 is a model oxide substrate, whereas films on Si(100) are applicable in a wide variety of silicon-based microtech-nologies, our films can be considered very good candidates for such applications.

4 Conclusions

SrRuO3-covered Pt(111) electrodes can easily be grown on YSZ(100)-buffered Si(100) substrates by r.f. sputtering. Due to the fourfold symmetry of YSZ(100) and the threefold symmetry of the Pt(111) plane these electrodes are multiply twinned. The SrRuO3-covered Pt(111) electrodes provide a smooth and plane surface. (104)-oriented ferroelectric Bi3.25La0.75Ti3O12 and Bi3.54Nd0.46Ti3O12 thin films can easily be grown onto this surface by PLD at an optimum substrate temperature of 750 °C. These ferroelectric films are also multiply twinned, inheriting this property from the electrodes. However, all of the azimuthal do-mains (twins) have the same component P

⊥ of the polarization vector perpendicular to the film plane.

These (104)-oriented BLT and BNT films on Si(100) have good ferroelectric properties and are suitable for applications in a number of silicon-based microtechnologies.

Acknowledgements The authors are thankful to Drs. M. Alexe (Halle), H. N. Lee (Oak Ridge) and S. Senz (Halle)

for numerous valuable discussions. Work in part supported by DFG via the Group of Researchers FOR 404 at Mar-

tin-Luther-Universität Halle-Wittenberg.

References

[1] S. Gevorgian, in: [4], Chap. 7.

[2] J. F. Scott and C. A. Paz de Araujo, Science 246, 1400 (1989).

[3] J. F. Scott, Ferroelectric memories (Springer, Berlin, 2000).

[4] N. Setter (ed.), Electroceramic-based MEMS: Fabrication technology and applications (Springer, New York,

2005).

[5] M. Alexe, St. Senz, A. Pignolet, D. Hesse, and U. Gösele, Ferroelectrics 225, 75 (1999).

[6] R. Waser (ed.), Nanoelectronics and information technology (Wiley-VCH, Weinheim, 2003).

[7] A. K. Tagantsev, I. Stolichnov, E. Colla, and N. Setter, J. Appl. Phys. 90, 1387 (2001).

[8] O. Auciello, K. D. Gifford, and A. I. Kingon, Appl. Phys. Lett. 64, 2873 (1994).

[9] T. Nakamura, Y. Nakao, A. Kamisawa, and H. Takasu, Appl. Phys. Lett. 65, 1522 (1994).

[10] S. Aggarwal, I. G. Jenkins, B. Nagaraj, C. J. Kerr, C. Canedy, R. Ramesh, G. Velasquez, L. Boyer, and J. T.

Evans, Appl. Phys. Lett. 75, 1787 (1999).

[11] C. A. Paz de Araujo, J. D. Cuchiaro, L. D. McMillan, M. C. Scott, and J. F. Scott, Nature 374, 627 (1995).

[12] H. N. Al-Shareef, D. Dimos, T. J. Boyle, W. L. Warren, and B. A. Tuttle, Appl. Phys. Lett. 68, 690 (1996).

[13] A. Gruverman, Appl. Phys. Lett. 75, 1452 (1999).

[14] H. Ishiwara (ed.), Ferroelectric random access memories: Fundamentals and applications (Springer, Berlin,

2004).

[15] J. Lettieri, Y. Jia, M. Urbanik, C. I. Weber, J.-P. Maria, D. G. Schlom, H. Li, R. Ramesh, R. Uecker, and

P. Reiche, Appl. Phys. Lett. 73, 2923 (1998).

[16] K. Ishikawa and H. Funakubo, Appl. Phys. Lett. 75, 1970 (1999).


La0.75

Ti3O

4 and Bi

3.54Nd

0.46Ti

3O



[17] H. N. Lee, S. Senz, N. D. Zakharov, C. Harnagea, A. Pignolet, D. Hesse, and U. Gösele, Appl. Phys. Lett. 77,

3260 (2000).

[18] J. Lettieri, M. A. Zurbuchen, Y. Jia, D. G. Schlom, S. K. Streiffer, and M. E. Hawley, Appl. Phys. Lett. 76,

2937 (2000).

[19] H. N. Lee, S. Senz, A. Pignolet, and D. Hesse, Appl. Phys. Lett. 78, 2922 (2001).

[20] H. N. Lee and D. Hesse, Appl. Phys. Lett. 80, 1040 (2002).

[21] H. N. Lee, D. Hesse, N. Zakharov, and U. Gösele, Science 296, 2006 (2002).

[22] A. Garg, Z. H. Barber, M. Dawber, J. F. Scott, A. Snedden, and P. Lightfood, Appl. Phys. Lett. 83, 2414

(2003).

[23] A. Garg, A. Snedden, P. Lightfood, J. F. Scott, X. Hu, and Z. H. Barber, J. Appl. Phys. 96, 3408 (2004).

[24] T. Kojima, T. Sakai, T. Watanabe, H. Funakubo, K. Saito, and M. Osada, Appl. Phys. Lett. 80, 2746 (2002).

[25] J. R. Duclere, M. Guilloux-Viry, A. Perrin, E. Cattan, C. Soyer, and D. Remiens, Appl. Phys. Lett. 81, 2067 (2002).

[26] B. H. Park, B. S. Kang, S. D. Bu, T. W. Noh, J. Lee, and W. Jo, Nature 401, 682 (1999).

[27] P. C. McIntyre, C. J. Maggiore, and M. Nastasi, J. Appl. Phys. 77, 6201 (1995).

[28] K. H. Ahn, S. Baik, and S. S. Kim, J. Mater. Res. 17, 2334 (2002).

[29] C. H. Kim, J. K. Lee, H. S. Suh, J. Y. Yi, K. S. Hong, and T. S. Hang, Jpn. J. Appl. Phys. 41, 1495 (2002).

[30] L. Goux, M. Gervais, F. Gervais, A. Catherinot, C. Champeaux, and F. Sabary, Mater. Sci. Semicond. Process.

5, 189 (2002).

[31] J. R. Duclere, M. Guilloux-Viry, A. Perrin, A. Dauscher, S. Weber, B. Lenoir, C. Soyer, E. Cattan, and

D. Remiens, J. Phys. 11, 133 (2001).

[32] E.-S. Choi, S.-G. Yoon, W.-Y. Choi, and H.-G. Kim, Appl. Surf. Sci. 141, 77 (1999).

[33] H. Myoren, Y. Nishiyama, H. Fukumoto, H. Nasu, and Y. Osaka, Jpn. J. Appl. Phys. 28, 351 (1989).

[34] H. Fischer and J. Janek (Gießen), private communication (2004).

[35] Y. Shimakawa, Y. Kubo, Y. Tauchi, H. Asano, T. Kamiyama, F. Izumi, and Z. Hiroi, Appl. Phys. Lett. 79,

2791 (2001).

[36] M.-W. Chu, M.-T. Caldes, L. Brohan, M. Ganne, A.-M. Marie, O. Joubert, and Y. Piffard, Chem. Mater. 16, 31

(2004).

[37] H. McMurdie, M. Morris, E. Evans, B. Paretzkin, W. Wong-Ng, and C. Hubbard, Powder Diffr. 1, 84 (1986).

[38] G. Asayama, J. Lettieri, M. A. Zurbuchen, Y. Jia, S. Trolier-McKinstry, D. G. Schlom, S. K. Streiffer, J.-P.

Maria, S. D. Bu, and C. B. Eom, Appl. Phys. Lett. 80, 2371 (2002).

[39] D. Hesse, N. D. Zakharov, A. Pignolet, A. R. James, and S. Senz, Cryst. Res. Technol. 35, 641 (2000).

[40] N. D. Zakharov, A. R. James, A. Pignolet, S. Senz, and D. Hesse, Mater. Res. Soc. Symp. Proc. 623, 173

(2000).

original - max planck societyferroelectric thin films. the growth of non-c-axis oriented epitaxial...

Documents