ultralow energy sims

A STUDY OF THE EFFECTS OF ULTRALOW-ENERGY

SECONDARY ION MASS SPECTROMETRY (SIMS) ON

SURFACE TRANSIENT AND DEPTH RESOLUTION

AB RAZAK CHANBASHA

(M.Sc. University of Strathclyde, UK)

A THESIS SUBMITTED

FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

DEPARTMENT OF PHYSICS

NATIONAL UNIVERSITY OF SINGAPORE

(2007)

Acknowledgements

i

Acknowledgements

This has been one of the most arduous projects that I have undertaken. This

would not have been possible without the opportunity and the trust that has been

given to me by my mentor and supervisor, Professor Andrew Wee Thye Shen. Prof.

Wee has been most encouraging and had never failed to keep up with my progress

throughout the course of this research. To him, I am indebted.

I would also like to thank Mr. Ridzuan Wu Chia Chung, Managing Director at

Omega Scientific Pte Ltd for the use of the ATOMIKA SIMS without which this

research would not have taken place. I am also grateful to my colleague Mr. Amir

Jantan for maintaining the SIMS instrument in excellent condition.

I am also fortunate to have research colleagues Dr. Liu Rong and Dr. Md.

Abdul Kader Zilani, formerly of Surface Science Laboratory (NUS), both of whom I

have had regular discussions with throughout the course of my study. Thank you, Liu

Rong and Zilani.

Finally, my utmost gratitude goes to my family. To my wife, Aishah for her

untiring support, constant encouragement and having to tolerate my spending family-

time away from home. To my children, Irshaad, Muneerah, Shafeeq and Nazheef, I

hope to have inspired them to continuously quest for knowledge. I am also grateful to

my parents for instilling in me the value of education.

Table of contents

ii

Table of contents

Acknowledgments i

Table of contents ii

Summary vi

Abbreviations and symbols viii

Figure captions x

List of Tables xii

Publications xiii

Chapter 1: Introduction 1

1.1 The need for SIMS in the semiconductor industry. 1

1.2 Challenges of ultralow-energy SIMS 3

1.2.1 Surface transient effects 3

1.2.1.1 Surface transient effects with ultralow-energy

O2+ sputtering 4

1.2.1.2 Surface transient effects with ultralow-energy

Cs+ sputtering 6

1.2.2 Sputter rate 9

1.2.3 Depth resolution 10

1.2.3.1 Factors affecting depth resolution 10

1.2.3.2 Methods used to improve depth resolution: O2+

bombardment 13

(a) Lower primary ion energy 13

(b) Varying the incident angle 13

(c) Oxygen flooding and sample rotation 15

1.2.3.3 Methods used to improve depth resolution: Cs+

bombardment 15

1.3 Outline of research project 17

References 19

Chapter 2: SIMS Principles 25

2.1 Introduction 25

2.2 Fundamentals of SIMS 26

2.2.1 Sputtering & collision theory 26

2.2.2 Sputtering yield 30

2.2.3 Sputter rate 31

2.2.4 Secondary ion emission 32

2.2.4.1 Secondary ion yield and ionisation probability 35

2.2.4.2 Ionisation mechanism with O2+ ion beam 35

2.2.4.3 Ionisation mechanism with Cs+ ion beam 36

2.2.4.4 Secondary ion species 38

2.3 Factors affecting depth profiling 38

Table of contents

iii

2.3.1 Primary ion species 39

2.3.2 Primary ion energy 39

2.3.3 Primary ion angle of incidence 40

2.3.4 Crater edge effect 41

2.3.5 Sample charging effect 42

2.4 Quantification of depth profiles 43

2.4.1 Relative sensitivity factors and absolute standards 43

2.4.2 Depth calibration 45

References 47

Chapter 3: Experimental and instrumentation 51

3.1 SIMS instrumentation 51

3.1.1 Introduction 51

3.1.2 Vacuum system and sample handling 52

3.1.2.1 Vacuum system and monitors 52

3.1.2.2 Sample introduction 54

3.1.2.3 Sample manipulator 55

3.1.3 Primary ion gun 56

3.1.3.1 O2+ source 56

3.1.3.2 Cs+ source 57

3.1.3.3 FLIG 58

3.1.4 Secondary Ion Optics 60

3.1.5 Quadrupole mass spectrometry 60

3.1.6 Secondary ion detection 63

3.1.7 Data acquisition and electronics 63

3.2 Atomic force microscopy 64

3.2.1 Principles 64

3.2.2 Contact mode 66

3.2.3 Non-contact mode 66

3.2.4 Tapping mode 67

3.3 Experimental 68

3.3.1 Sample 68

3.3.2 Analysis parameters with O2+ SIMS 69

3.3.3 Analysis parameters with Cs+ SIMS 70

3.3.4 Sputter rate determination 71

References 71

Chapter 4: Effect of ultralow-energy O2+ SIMS on Si surface transient 73

4.1 Introduction 73

4.2 Results & discussion 74

4.2.1 Surface transients 74

4.2.1.1 Surface spikes and incomplete oxidation 75

4.2.1.2 Transient width 79

Table of contents

iv

4.2.2 Sputter rates 82

4.3 Summary 88

References 89

Chapter 5: Effect of ultralow-energy O2+ SIMS on depth resolution 91

5.1 Introduction 91



5.2.1.1 Depth resolution in terms of FWHM 93

5.2.1.2 Depth resolution in terms of exponential decay 100

5.2.1.3 MRI model and evaluation 103

5.2.2 Dynamic range 107

5.3 Summary 108

References 110

Chapter 6: Effect of ultralow-energy Cs+ SIMS on Si surface transient 111

6.1 Introduction 111


6.2.1 Surface transients 112

6.2.1.1 30

Si- profiles 113

6.2.1.2 Transient width 117

6.2.1.3 Steady state intensity of 30

Si- profiles 119

6.2.2 Sputter rates 120

6.3 Summary 125

References 126

Chapter 7: Effect of ultralow-energy Cs+ SIMS on depth resolution 128

7.1 Introduction 128



7.2.1.1 Depth resolution in terms of FWHM 131

7.2.1.2 Depth resolution in terms of exponential decay 137

7.2.1.3 Depth resolution evaluated with MRI model 141

7.2.2 Dynamic range 146

7.3 Summary 148

References 151

Table of contents

v

Chapter 8: Conclusion 152

8.1 Surface transient 152

8.2 Sputter rate 154

8.3 Depth resolution 155

8.4 Dynamic range 156

8.5 Optimum conditions for analysis 157

8.6 Proposed future work 158

Appendix A 159 Depth profiles with O2

+ primary ion beam

Appendix B 171

Depth profiles with Cs+ primary ion beam

Summary

vi

Summary

Ultralow-energy secondary ion mass spectrometry (SIMS) has been

introduced to meet the increasing demand for depth profiling of ultrashallow junctions

(< 20 nm) and ultrathin films following the developments in device miniaturisation in

the semiconductor industry. This challenge for accurate profiling at the near surface

(SIMS transient region) and for achieving high depth resolution is directly influenced

by the probe energy and the incident angle of the primary ion used. However, issues

such as surface roughening, atomic mixing, secondary ion yield, sputter rates and for

ultralow-energy Cs+, a poorly focused beam are important considerations.

The objective of this research is to understand the trends, characteristic

behaviour and processes involved with the use of ultralow-energy SIMS. Obtaining

such information will aid method development and the optimization of parameters for

accurate depth profiling. In this research, an ATOMIKA 4500 SIMS depth profiler

with O2+ and Cs

+ primary ion beams at an ultralow-energy (< 1 keV) and incidence

angles between 0 - 70o without oxygen flooding was used.

The dependence of sputter rate and transient width as a function of primary

ion energy (Ep) and incident angle () is studied. Sputter rate variations with depth are

also evaluated to determine the accuracy of using average sputter rates for depth

calibrations.

The minimum transient width (ztr) is achievable at normal and near normal

incidence with an O2+ primary ion beam. When Ep is reduced to less than 500 eV,

there is no significant reduction in ztr but the range of in which this can be achieved

is greater. This corresponds to achieving steady state sputtering earlier because of

complete oxidation of silicon. The critical angle at which incomplete oxidation begins

is also ascertained.

The narrowest transient widths achieved with Cs+ were at ~ 30-50

o and were

achieved when the Cs concentration stabilizes in the vicinity of the Cs+ penetration

depth. An extended transient effect was observed when profiled at > 50o

suggesting

that ~ 60o, which is widely used for depth profiling with Cs

+, is not suitable for

ultrashallow junction depth profiling. The best achieved detection sensitivity is at ~

30o for all energies investigated.

Summary

vii

Depth resolution is improved by lowering Ep and good depth resolution can be

achieved not only at normal incidence but over a wider range of incident angles with

O2+. The best depth resolution observed is with Ep ~ 250eV and ~ 40

o throughout

the depth profiled (120nm).

High depth resolution with Cs+ is achieved at ~ 50

o. It is also established that

the relationship between improvements in depth resolution is linear and gradual with

increasing . Increasing is noted to be more sensitive in improving depth resolution

than reducing Ep at ultralow-energy.

Based on the mixing-roughness-information model (MRI), it is possible to

differentiate the effects of atomic mixing and surface roughness on the depth

resolution of -layers. MRI is also found to be a more sensitive indicator of surface

roughness compared to another method where an increase in the Si+ matrix signal

were used. The dynamic range is also evaluated.

Abbreviations and symbols

viii

Abbreviations & Symbols

AES Auger Electron Spectroscopy

AFM Atomic Force Microscopy

APCVD Atmospheric Pressure Chemical Vapour Deposition

ASF Absolute Sensitivity Factor

ASTM American Society for Testing and Materials

Standards

CMOS Complementary Metal-Oxide-Semiconductor

DC Direct Current

Ep Primary Ion Energy (probe energy)

FLIGtm

Floating Low Energy Ion Gun

FWHM Full Width at Half Maximum

LDD Lightly Doped Drain

m/e Mass-to-charge Ratio

MQW Multi-Quantum Well

MRI Mixing-Roughness-Information

M-SIMS Double Focussing Magnetic Sector SIMS

PLC Programmable Logic Controller

PVR Peak-to-Valley Ratio

QMS Quadrupole Mass Spectrometer

Q-SIMS Quadrupole SIMS

RF Radio Frequency

Rnorm Penetration Depth

RSF Relative Sensitivity Factor

SIMS Secondary Ion Mass Spectroscopy

SIO Secondary Ion Optics

SOI Silicon-on-Insulator

S/D Source and Drain

SDE Shallow Drain Extension

STM Scanning Tunnelling Microscopy

STS Sample Transfer System

TOF-SIMS Time-of-Flight SIMS

TRIM Transport of Ions in Matter

UHV Ultra High Vacuum

XPS X-ray Photoelectron Spectroscopy

A Screening radius

ao Bohr radius

E Electron charge

I Material of interest

iM Secondary ion yield of species M

M Species

Mn Mass of atom

N Number of atoms in primary ion

T Time

tox Gate oxide thickness

Abbreviations and symbols

ix

X Penetration depth of primary ion

xj Junction depth

Z Depth

Ż Erosion rate (sputter rate)

ztr Transient width

A Area of crater

An Electron affinity

CM Fractional concentration of species M

E Initial energy

Eb Lattice binding energy

EF Fermi level

Ev Vacuum level

FD Density of energy deposited

I Secondary ion intensity

In Ionisation potential

Ip Primary ion current

L Transistor gate length

M Matrix

N Atomic density of target atoms

P± Sputtered ion emission yield

R Reference material

Rc Resolution contrast

Se(E) Electronic stopping cross-section

Sn(E) Nuclear stopping cross-section

T Transferred energy (recoil energy)

U DC voltage

Uo Surface binding energy

V Peak amplitude of the RF voltage

Ws Source depletion depth

Wd Drain depletion depth

Vt Threshold voltage

Y Sputtering yield

Z Atomic number

α± Ionisation probability

λd Decay length

Ρ Density of material

(t) Applied potential

Ω Angular frequency

Angle of incidence

Σ Elastic collision cross-section

Dose

Φ Workfunction

Η Transmission probability

Figure captions

x

Figure captions

Fig. 2.1 Ion-solid interaction processes resulting in the emission of ions.

Fig. 2.2 Relative positive ion yields for 13.5 keV O-, ~ 0

o. ∆ represents that a

compound sample was used and B.D. means barely detectable.

Fig. 2.3 Relative negative ion yields for 16.5 keV Cs+, ~ 0


compound sample was used and N.D. means not detectable.

Fig. 3.1 Front view of the ATOMIKA 4500.

Fig. 3.2 The ATOMIKA 4500 SIMS Depth Profiler.

Fig. 3.3 Schematic of the vacuum system for SIMS 4500.

Fig. 3.4 Schematic of duoplasmatron ion source.

Fig. 3.5 Schematic of cesium surface ionisation source.

Fig. 3.6 Schematics of the FLIG.

Fig. 3.7 Schematic of the quadrupole mass spectrometer connections.

Fig. 3.8 Interatomic force vs. distance curve.

Fig. 3.9 Schematic of an atomic force microscope in tapping mode.

Fig. 4.1 Depth profile obtained with Ep ~ 250eV at ~ 20o. The dips in the matrix

profiles and the peaks correspond to the positions of the Ge delta-layers.

Fig. 4.2 30Si

+ profiles for (a) Ep ~250 eV (b) Ep ~500eV and (c) Ep ~1keV.

Fig. 4.3 The XPS Si 2p spectra of the surface at steady state for 4 keV O2+

bombardment: (a) virgin Si(100) sample; (b) sputtering at 0°; (c)

sputtering at 15°; (d) sputtering at 30°; (e) sputtering at 45°; (f) sputtering

at 55°; (g) sputtering at 70°.

Fig. 4.4 Measured transient width based on both 44

SiO+ and

30Si

+.

Fig. 4.5 Sputter rates vs. angle of incidence. d1 shows the average sputter rate from

the surface to the first Ge delta-layer and the other is the average sputter

rate observed from the surface to the tenth Ge delta-layer.

Fig. 4.6 Normalized sputter rates throughout the depth.

Fig. 5.1 Profiles of 70

Ge+ obtained with different Ep (normalized to the first peak of

Ep ~ 250 eV) at a) ~ 0o and b) ~ 40

o.

Fig. 5.2 Depth resolution of Ge delta-layer as a function of profile depth, measured

by FWHM for a) Ep ~ 250 eV b) Ep ~ 500 eV c) Ep ~ 1 keV at various

incident angles.

Fig. 5.3 Ion image of 70

Ge+ taken at the depth where the cursor position is. a) Ep ~

250 eV, ~ 70o at the 2nd delta-layer b) Ep ~ 250 eV, ~ 70

o at the 4

th

delta-layer c) Ep ~ 1 keV, ~ 40o at the 4

th delta-layer.

Fig. 5.4 Depth resolution of Ge -layer at various primary ion incident angles.

Fig. 5.5 Comparison of the normalized 70

Ge+ profiles of the first -layer at various

Ep.

Fig. 5.6 Plot of decay length (d) versus incident angle ().

Figure captions

xi

Fig. 5.7 The projected ion range of O2+ primary ion beam at various incident

angles calculated from TRIM.

Fig. 5.8 Depth resolution parameters (FWHM and d) dependence on energy for

the first delta-layer profile at ~ 0o.

Fig. 5.9 Dynamic range averaged over the first nine Ge peaks vs incident angle.

Fig. 6.1 Depth profile obtained using Ep ~320eV at ~50o.

Fig. 6.2 30Si

- profiles for (a) Ep ~320 eV (b) Ep ~500 eV and (c) Ep ~1 keV.

Fig. 6.3 AFM images of crater bottoms resulting from Ep ~ 320 eV a) ~ 0o, b)

~ 10o, c) ~ 20

o and d) ~ 60

o taken at 2-3 nm depth. The rms roughness

values for craters from ~ 0o, 10

o and 20

o are ~ 0.18 nm and for ~ 60

o is

~ 0.27 nm, where the onset of roughening is noted.

Fig. 6.4 Measured transient width based on 30

Si- profiles for Ep ~320 eV, 500 eV

and 1 keV. The plot of Rnorm of Cs+ primary ion at various Ep is also

shown. For conditions where extended transient effects is not observed,

the ztr < 2Rnorm.

Fig. 6.5 30Si

- intensity at steady state for various Ep and . Si

- intensity from high

energy of Ep ~ 8 keV is also shown.

Fig. 6.6 Sputter rates vs angle of incidence. For each Ep, d1 represents the average

sputter rate from the surface to the first delta-layer and the other is the

average sputter rate from the surface to the tenth delta-layer.

Fig. 6.7 Normalized sputter rates to the last delta-layer throughout the depth.

Fig. 7.1 Typical depth profiles (a) 98

SiGe- profiles with

30Si

- and

59Si

2- (b)

98SiGe

-

profiles with various Ep at ~ 60o (c)

98SiGe

- profiles with Ep ~ 320 eV at

various .

Fig. 7.2 Depth resolution of Ge delta-layers as a function of profile depth,

measured by FWHM for (a) Ep ~ 320 eV, (b) Ep ~ 500 eV and (c) Ep ~ 1

keV at various incident angles.

Fig. 7.3 Depth resolution of Ge delta-layers at various primary ion incident angles.

Fig. 7.4 (a) Depth resolution of the first Ge delta-layer peak at various Ep. (b)

Linear relationship between depth resolution in terms of FWHM with

when no surface roughening is present.

Fig. 7.5 (a) – (c) Plot of decay length against incident angle for various Ep at

selected depth. (d) Comparison of d at the first -layer for various Ep.

Fig. 7.6 Comparison of the normalized 98

SiGe- profiles of the first delta-layer at

various Ep.

Fig. 7.7 Penetration depth of Cs+ primary ions at various incident angles calculated

from TRIM.

Fig. 7.8 Depth resolution parameters (FWHM and d) dependence on energy for


Fig. 7.9 Dynamic range averaged over the first nine 98

SiGe- peaks vs incident

angle.

List of tables

xii

List of tables

Table 1.1 Summary of published work on surface transients with ultralow-

energy O2+ sputtering

Table 1.2 Summary of published work on surface transients using ultralow-

energy Cs+ sputtering.

Table 1.3 Summary of ripple formation observed in previous work published

using ultralow-energy Cs+ sputtering.

Table 4.1 Summary of surface transient width, sputter rate and the onset of

roughening observed for O2+ sputtering on silicon wafer.

Table 5.1 Summary of depth resolution and dynamic range observed with O2+

sputtering.

Table 6.1 Summary of results from this work using ultralow-energy Cs+

sputtering.

Table 7.1 Primary ion penetration depth and FWHM at with best depth

resolution.

Table 7.2 Summary of results observed with ultralow-energy Cs+ sputtering.

Publications

xiii

PUBLICATIONS

1. A. R. Chanbasha and A. T. S. Wee, “Surface transient effects in ultralow-

energy O2+ sputtering of Si”, Surface and Interface Analysis, 37, 628 (2005).

2. A. R. Chanbasha and A. T. S. Wee, “Depth resolution studies in SiGe delta-

doped multilayers using ultralow-energy O2+ secondary ion mass

spectrometry”, Journal of Vacuum Science and Technology B, 24, 547 (2006).

3. A. R. Chanbasha and A. T. S. Wee, “Narrow surface transient and high depth

resolution SIMS using 250 eV O2+”, Applied Surface Science, 252, 7243

(2006).

4. A. R. Chanbasha and A. T. S. Wee, “Surface transient effects in ultralow-

energy Cs+ sputtering of Si”, Surface and Interface Analysis, 39, 397 (2007).

5. A. R. Chanbasha and A. T. S. Wee, “Depth resolution studies in SiGe delta-

doped multilayers using ultralow-energy Cs+ secondary ion mass

spectrometry”, Journal of Vacuum Science and Technology B, 25, 277 (2007).

Chapter 1: Introduction

1

Chapter One

Introduction

1.1 The need for SIMS in the semiconductor industry

Secondary Ion Mass Spectrometry (SIMS) is widely used as a surface analysis

tool in the semiconductor industry for dopant depth profiling, contamination

monitoring1,2

and determination of dopant concentrations because of its ability to

detect all elements, high sensitivity, large dynamic range, unrivalled depth resolution

and minimal sample preparation. Other techniques such as Auger Electron

Spectroscopy (AES) are commonly used due to its excellent spatial resolution (< 100

nm), and X-ray Photoelectron Spectroscopy (XPS) for its ability to provide surface

chemical information and thin film thickness.

Rapid progress in complementary metal-oxide-semiconductor (CMOS)

integrated circuit technology has been associated with successful device

miniaturization with the objective of increasing packing density and switching speed.

This is best done by reducing the transistor gate length (L). Inevitably, the vertical and

lateral dimensions of doped regions in the silicon substrate must be shortened as well.

However, when L is reduced, short-channel effects arise, such as threshold voltage

(Vt) reduction and high sub-threshold leakage current, hot-electron effects, and

source-to-drain punchthrough breakdown.3 These effects are directly related to

channel length and drain voltage, and all are affected, directly or indirectly, by the

dimensions of the source and drain junction (xj).


2

To reduce the channel length while maintaining acceptable long channel

behaviour, an empirical relationship was developed4

Lmin= A [xj tox (Ws + Wd )2]

1/3 (1.1)

where Lmin is the minimum transistor gate length, A is a constant, xj is the junction

depth where the doping concentration in the diffused profile falls to the concentration

of the background doping, and tox is the gate oxide thickness. Ws and Wd are the

source and drain (S/D) depletion depths, which are functions of the channel doping

and supply voltage.

To reduce the peak electric fields in drain regions, a very shallow, lightly-

doped drain region (LDD)5 may be placed between the drain and channel (also called

drain extension) so that the drain voltage drop is shared between the LDD and the

channel. These shallow drain extensions (SDE) are often called ultrashallow

junctions. The 65 nm technology node, will require drain extensions of 13-30 nm

depth depending on the type of devices.6

Besides the channel length, the gate dielectric thickness has also been scaled

down (1 ~ 3 nm for SiO2/SiON)7,8

to improve performance. There are, however,

concerns that ultrathin SiO2 introduces other problems such as boron penetration and

high leakage currents. Such reliability issues have led to research into alternative

high- dielectric materials.9 In the meantime, oxynitrides and oxide/nitride stacks are

being used. The addition of N reduces boron penetration through the dielectric. Small

amounts of N (~ 0.1 atomic %) at or near the Si channel interface have been shown to

improve device performance10

but larger amounts are undesirable.9

Devices such as semiconductor lasers, photonic and electronic integrated

devices based on multi-quantum well (MQW) structures are gaining much attention.


3

Such periodic structures with well widths of less than 10 nm demand a high depth

resolution to resolve the individual quantum wells.11-13

With the miniaturization of devices, current depth scales required in the

profiling of ultrashallow junctions and ultrathin oxynitride gate dielectrics coincide

with the surface transient depth region, i.e. the depth after which sputtering yield is

constant, of SIMS. This has created a challenge for accurate SIMS profiling at

ultrashallow depths of < 10 nm. This challenge for accurate profiling at the near

surface (SIMS transient region) is directly influenced by the probe energy (Ep), the

angle of incidence (), and the primary ion used.14

These parameters determine the

penetration depth of the primary ion beam and the development of surface roughening

which also has implications on depth resolution.

1.2 Challenges of ultralow-energy SIMS

1.2.1 Surface transient effects

Upon bombardment of the surface by primary ions, the altered surface is

accompanied by a combination of secondary ion yield variations,15

sputter rate

variations,16,17

and under some conditions, surface roughening.18-20

These variable

conditions are experienced mainly at the initial stages of profiling, and are referred to

as the surface transient. The depth up to which the surface transient effect prevails is

referred to as the transient width (ztr). Wittmaack21

and Dowsett et al.22

reported that

in this pre-equilibrium region, the ion yield and sputter rate vary by more than a factor

of 10. After this transient region, the profiling sputter rate and the secondary ion yield

stabilize and only then can a quantitative depth scale and concentration scale be


4

derived. It is therefore imperative that the width of the transient region be known so

that the first reliable data point can be ascertained.

Most work on the surface transient has been done using O2+ beams. However,

owing to instrument geometry and a lack of ultralow-energy ion sources, isolated data

have been published covering limited incident angles from normal to grazing angles

using ultralow-energy (Ep < 1 keV) primary ion beams.

1.2.1.1 Surface transient effects with ultralow-energy O2+ sputtering

The surface transient has been shown to shorten with the use of ultralow-

energy SIMS. To reduce the transient width for O2+

beams, many have attempted

reducing the impact energy and/or increasing the incident angle23,24

with respect to the

surface normal, while others have found normal incidence to be advantageous.21,25

Clegg et al.,26

Wittmaack21

and Dowsett27

pioneered studies on surface transients in Si

and GaAs for normal-incident O2+ beams at energies down to 230 eV. Chu et al.

28

reported a transient width of 0.5 nm using 230 eV O2+ at normal incidence, and

Dowsett et al.27

reported ztr to be < 1 nm at Ep ~ 500 eV, and 0.4 nm at Ep ~ 300 eV at

normal incidence. Jiang and Alkemade19

obtained ztr in the range 3-3.5 nm at Ep ~

600-700 eV and 70o < < 80

o. ztr has been observed to increase with impact energy

27

for Ep > 1 keV and incident angle up to 45o,21

followed by a decrease with increasing

incident angle beyond 45o.19

No equivalent comprehensive data are available for

ultralow-energy SIMS using Ep < 1 keV. A summary of published work is given in

Table 1.1.


5

Table 1.1 Summary of published work on surface transients with ultralow-energy O2+

sputtering

Transient width Ep of O2+ References

ztr ~ 3-3.5 nm 600–700eV 80

o

19

ztr increased with Ep

ztr decreased with Ep

> 1 keV < 45

o

> 45o

21

19

ztr ~ 0.5nm 230 eV ~ 0o

28

ztr ~ 0.4 nm

ztr < 1 nm

300 eV

500 eV ~ 0

o

27

ztr ~1.5-2 nm 600 eV ~ 73o

24

Oxygen flooding has also been used to overcome the surface transient effect.29

The presence of oxygen effectively promotes complete oxidation thus stabilizing the

ion yield and suppressing the onset of roughening. However, the presence of oxygen

alters the erosion rate initially resulting in shallower profiles.30,31

Other techniques

such as silicon capping32

and backside depth profiling33,34

have been successful in

circumventing the surface transient effect.

Ng et al. demonstrated that with an appropriate silicon capping layer

thickness, the sputtering equilibrium is reached before the impurity depth profile

begins.32

This approach ensures transient-free SIMS depth profiling of the

ultrashallow implant. Yeo et al. developed a backside profiling technique using

silicon-on-insulator (SOI) substrates.33,34

The backside of the wafer was carefully

thinned down by a combination of mechanical grinding and chemical etching prior to

SIMS depth profiling. Besides reaching sputtering equilibrium well in advance of the

implant profile, the technique also demonstrated a significant improvement in depth


6

resolution compared to conventional front-side depth profiling. This technique was

successfully used to demonstrate accurate measurement of ultrashallow implants33

and B penetration through de-coupled plasma nitrided thin gate oxides.35

Nevertheless, the sample preparation prior to SIMS depth profiling in both techniques

can be very tedious.

1.2.1.2 Surface transient effects with ultralow-energy Cs+ sputtering

When using a Cs+ primary ion beam, the transient width can be reduced by

decreasing Ep, since Ep is related to the primary ion range in the material being

analysed.36-38

Alternatively, increasing the incident angle has also helped.39

The

transient width has also been observed to increase at ~ 70o even though Ep is

reduced.40

Beyond a certain energy and incident angle as defined by the following

relationship, the increase in the transient width has been observed up to 30 nm:40

(min) = 14 log (Ep) + 65 (1.2)

where is degrees and Ep is in keV.

For instance with Ep ~ 250 eV, the enhanced transient effect was observed at

> 57o, for Ep ~ 500 eV when > 60

o, and for Ep ~ 1 keV when > 65

o. This

phenomenon has been associated with a decrease in the sputter rate (by almost half),

the formation of ripples at the crater bottom15,17,41

and an increasing Cs concentration

on the surface.16

The amount of Cs retained increases while the sputter rate decreases

with depth in the transient region. The ripple formation slows down the sputter rate

which in turn allows for a greater Cs build-up in the near surface region.16,17

The Cs

concentration continues to increase until a steady state sputtering is achieved.15,42,43


7

As the Cs concentration increases, so does the sputtering yield, which is much higher

than at steady state.44

The variations in the concentration of primary ions in the substrate causes

variations in the secondary ion yield15

of up to an order of magnitude higher as

compared to O2+. The presence of Cs is known to induce a significant reduction in the

substrate work function,15,45,46

thus enhancing negative secondary ion yield.

The transient width has been reported to be less than 5 nm with Ep < 1 keV and

~ 60o.44,47

These and other work on the transient width using Cs+ primary ions are

summarized in Table 1.2.

Table 1.2 Summary of published work on surface transients using ultralow-energy

Cs+ sputtering.

Transient width Ep of Cs+ References

ztr decreased with Ep 500 eV - 1 keV

9.5 keV

~ 60o

~ 25o

37

38

ztr decreased as increased. 250 eV – 5 keV > 80o

39

ztr increased when Ep is

decreased

250 eV – 2.5

keV ~ 70

o

40

Enhanced ztr ( ~ 30 nm)

observed

250 eV

500 eV

1 keV

> 57o

> 60o

> 65o

40

ztr < 5 nm 250 eV – 1 keV ~ 60o

44

47


8

Surface transient effects using Cs+ primary ions have also been associated with

crater bottom roughening. No ripple formation was reported at < 50-60o during Si

sputtering with Ep < 1 keV.44,48

Rapid formation of ripples was noted at incident

angles between 50-80o resulting in poorer depth resolution.

48,49 The critical angles

above which ripples form are: Ep ~ 250 eV/ ~ 50o, Ep ~ 500 eV/ ~ 55

o and Ep ~ 1

keV/ ~ 60o.48,50

Ripple formation or its absence in the transient region is

summarized in Table 1.3.

Table 1.3 Summary of ripple formation observed in previous work published using

ultralow-energy Cs+ sputtering.

Ripple formation Ep of Cs+ References

No ripple formation < 1 keV < 50 - 60o

44

48

No ripple growth as observed

by AFM 250 eV ~ 45, 55, 65

o

45

Rapid ripple formation < 1 keV ~ 50 - 80o

48

49

Critical angles for ripple

formation

250 eV

500 eV

1 keV

~ 50o

~ 55o

~ 60o

48

50

Ripple growth increases with 500 eV ~ 50/ 60 - 75o

48

Ripples growth from 2nm deep 500 eV ~ 65o

48

Ripple topography begins after

5 nm

250 eV,

1 keV ~ 75

o

16

17


9

Kataoka et al. recommended operating at ~ 45-50o with Ep ~ 250 eV, ~ 50-55

o

with Ep ~ 500 eV and ~ 55-60o with 1 keV, these being conditions where there is no

ripple formation and minimal profile shift.50

Other suggestions to reduce the transient

width include coating a thin layer of Cs on the surface prior to analysis and

introducing an O2 leak during analysis, which almost completely removes the

transient, but the disadvantage is a loss in sensitivity and depth resolution due to

roughening.51

There is still insufficient consistent experimental data on the transient width for

ultralow-energy Cs+ depth profiling. Cs

+, being of higher mass, has a shallower

penetration depth and reduced ion beam broadening effects. Thus, it is expected to

have a narrower transient width extending to ~ 2-2.5 times the Cs+

penetration depth

(Rnorm), provided there is no roughening.17,51,52

1.2.2 Sputter Rate

The rate of recession of the sample is known as the sputter rate. It is dependent on

the primary ion parameters (Ep, , mass, current density) and the nature of the target

(mass, crystallinity, topography, surface binding energy).53,54

In depth profiling, the

sputter rate is not only important for analysis speed, but more importantly for depth

conversion, where a constant erosion rate is usually assumed. However the erosion

rate at the near surface is not constant due to transient effects21

or the onset of

roughening,18,55

thus producing distorted profiles i.e. a depth scale offset of about 1nm

per keV O2+ beam energy increment, resulting in the profile appearing nearer the

surface. Wittmaack21

however, observed that the effect is negligible at Ep < 500 eV

when using O2+ at normal incidence. Napolitani et al.

55 also confirmed that there is


10

negligible variation (+/-2%) in sputtering yield during the surface transient or from

the eventual onset of roughening working at grazing angles with 1.5 keV O2+.

Alkemade et al.24

observed that the sputter rate decreases as the incident angle is

increased from 60o to 80

o for Ep ~ 600 eV O2

+. They attributed this to an increase in

the fraction of backscattered ions.

1.2.3 Depth resolution

Depth resolution, as defined by the ASTM E-42 committee, is the distance

over which a 16-84% (or 84-16%) change in signal over an interface is measured.56

More recently, depth resolution is also taken to be a measure of the ability to

discriminate between features in adjacent thin layers.57

It is often annotated by the full

width at half maximum (FWHM) of signals from atomic layer delta-doped samples as

well as decay length (d) which is the distance over which the intensity drops by a

factor of e as in:

I2z = I dzz

z e/)( 12

1

, (1.3)

where I is the secondary-ion intensity, z2 and z1 are the depths between which the

decay length is determined.

1.2.3.1 Factors affecting depth resolution

In SIMS, the attainable depth resolution is affected mainly by three factors;

instrument related factors, sample characteristics and ion-solid interactions.18,58

Non-

uniform irradiation from a poorly focused beam and poor rastering quality cause

erosion inhomogeneity (crater edge effects),59

resulting in a deterioration of depth

resolution with depth. Re-deposition from crater walls (crater sidewall effects)29

has


11

been known to affect the quality of the depth profiles. Intrinsic surface roughness on

the sample will affect topography development as depth profiling progresses.

Instrumental factors can be minimized with good ion source design and electronic or

optical gating. Near-atomically flat surfaces are now possible to achieve, thus

minimizing sample roughness.

Amongst the three factors, ion-solid interactions cannot be completely

eliminated as they are part of the sputtering process. Upon surface bombardment, ion-

solid interactions cause atomic mixing, recoil implantation, radiation-enhanced

diffusion and chemically driven segregation.59

Atomic mixing or bombardment-

induced atom relocation is a dominant factor affecting depth resolution; ion

bombardment induced roughness and segregation are other ion–solid interactions that

affect depth resolution. By minimizing the effect of atomic mixing and thereby

maintaining the structure in the altered layer, nanometer scale depth resolution can be

achieved.14

This ability to obtain high depth resolution in the order of nanometers is

directly influenced by the probe energy of the SIMS tool.14,60,61

Besides using a lower

Ep, an improved depth resolution has also been achieved by primary ion bombardment

at oblique incidence.23,60,62

The shorter penetration depth at oblique incidence results

in a shallower altered layer and consequently, a better depth resolution.

Ion bombardment induced roughness has been attributed to a heterogeneous

oxide layer in the case of O2+ bombardment.

63 It has been observed that at lower

energy, the onset of surface roughening begins earlier for oblique angles.31,55

At

normal incidence, the onset of roughening has not been observed, at least for Si and

GaAs for Ep < 2 keV.64

When roughening occurs, the magnitude of the roughness

depends on the incident angle.65


12

During Cs+ primary ion sputtering of Si, the root cause of surface roughening

and ripple formation remains unclear. It has been suggested that ripple formation

appears to occur via competing roughening and smoothing effect on/in amorphous

substrate.17

Though the root cause is unclear, it was thought to be initiated by sputter

induced surface stress which results in microscopic surface curvature. This effect,

together with the agglomeration of Cs on the surface causes a variation in the primary

ion incident angle which effectively alters the sputtering yield. As a result, hillocks

and valleys are propagated over time in the direction of the incident beam.

Conversely, smoothing is also induced via atomic diffusion either thermally or

ballistically by the sputtering process. Another possible mechanism proposed is that

microetch pits grows into ripples due to fluctuations in beam density in a microarea.66

Depth resolution is element dependent; B in silicon differs from Ge in

silicon.23,60,67

Jiang et al.68

for instance, observed 50% better depth resolution

parameters for Ge delta-layers compared to B delta-layers analysed at grazing angle

with O2+ primary ions. However, at normal incidence, B delta-layers gave better depth

resolution. At normal incidence, where the oxygen concentration is greater, SiGe is

oxidized and Ge segregates near the oxide–bulk interface, thus degrading the depth

resolution. At grazing incidence, the oxygen incorporation is low and hence Ge

segregation is minimal. The depth resolution is then determined by atomic mixing

only, which appears to be less for Ge than B in silicon.


13

1.2.3.2 Methods used to improve depth resolution with O2+ bombardment

(a) Lower primary ion energy

In an early work done in the sub-keV range, Ormsby et al.64

observed that the

depth resolution deteriorates as the impact energy is increased. Besides atomic

mixing, they also attributed the deterioration in the depth resolution to the onset of

roughening. At normal incidence, Ormsby et al.64

obtained high depth resolution

across the whole energy range of 230 eV < Ep < 1 keV with FWHM < 3 nm. At

oblique angles, although a higher depth resolution was observed as the energy was

lowered, the onset of roughening occurred earlier.31,65,68,69

Several other works63,68,70

cited that the advantages of lowering the impact energy were compromised by

significant surface roughening, particularly at oblique angles. Another setback is the

decrease in sputtering yield.

At Ep ~ 300 eV, Clegg et al.26

performed SIMS measurements on Ge delta-

layers in Si at normal incidence and reported a FWHM = 1.7 nm and d = 1.0 nm.

Similar results were achieved by Jiang et al.23

using Ep ~ 700 eV and ~ 71o. At

about the same time, Smith et al.71

reported FWHM of better than 1 nm at normal

incidence using a 16-period GeSi superlattice sample. More recently, Jiang et al.68

achieved a FWHM of 1.1 nm using Ep ~ 158 eV and ~ 45o, better than their results

at normal incidence.

(b) Varying the incident angle

Besides lowering the impact energy, oblique incidence angles have been used

when normal incidence is not possible due to instrument geometry. High depth

resolution at sub-keV has been achieved at normal incidence,26,64,68,71,72

and the depth


14

resolution was observed to be constant with depth at Ep ~ 230 eV, Ep ~ 250 eV and Ep

~ 158 eV.68

Besides normal incidence, profiling at ~o and ~

o with Ep ~ 500

eV showed a constant depth resolution throughout the depth range of 190 nm

studied.64

In another work, the optimum depth resolution showed no difference

whether profiling was done at ~ 0o or 34

o. Wittmaack

72 concluded that below some

critical anglec) for oxidation, with measurements done at Ep ~ 1 keV, the depth

resolution should not differ much.

At oblique angles, a severe drop in depth resolution was observed above a

critical incidence ~ 50-60o with Ep ~ 500 eV.

64 This phenomenon was attributed to

micro-roughening of the crater bottom. Even at very low energy of Ep ~ 158 eV, Jiang

et al.68

noticed the onset of roughening and consequently poor depth resolution for Ge

delta-layers at ~ 60o, but at ~ 45

o and 50

o, good depth resolution was observed.

Wittmaack72

also noted ripple growth at ~ 38 - 62

o, Ep ~ 1 keV and extrapolated that

at Ep ~ 500 eV, the ripples may grow as early as at a depth of a few nanometers.

Liu et al.69

working at Ep ~ 500 eV and ~ 46o, 56

o and 69

o, observed that the

onset of roughening occurs earlier at ultralow-energy and concluded that ~ 69o gives

the best depth resolution, with the onset of roughness observed only at 110 nm.

The depth resolution improved with an increasing incident angle for > 60o to

~ 80o using Ep ~ 600 eV.

24 However, there is a limit to the improvement in depth

resolution as the incident angle is further increased. Eventually, at very grazing

incident angle, the lateral straggling will limit the improvement in depth resolution.73


15

(c) Oxygen flooding and sample rotation

There have also been attempts to reduce roughening by rotation and oxygen

flooding, mainly at grazing incidence, with the objective of removing or reducing

roughening. Oxygen flooding, however, has no particular advantage over normal

incidence without oxygen flooding,14,30,31,64,74

at ultralow energies. Nevertheless, for

instruments that cannot operate at normal incidence, oxygen flooding and sample

rotation75

have been demonstrated to be effective in reducing surface roughening70

at

Ep ~ 500 eV and ~56o.

1.2.3.3 Methods used to improve depth resolution with Cs+ bombardment

There are a few reported studies on the influence of Cs+ primary ion energy

and incident angle on depth resolution. Generally, beam induced broadening effects

have been observed to decrease with the use of heavy primary ions59

, lower Ep,76,77

and increasing 78from 0

o to 60

o as it minimizes the width of the atomic mixing zone

in the sample. However, these observations were made at higher beam energies of

greater than 1 keV. At ultralow-energy (Ep < 1 keV), depth resolution improves by

increasing only to a certain critical cbefore the onset of surface roughening.17,79

Another consideration is that by lowering Ep, the beam current density is lower and so

is the sputtering yield. The consequence is a trade-off between depth resolution and

detection limits.78

At ultralow Cs+ primary energy ( ~ 0-75

o), Kataoka et al.

79 observed an

improvement in depth resolution (defined as resolution contrast Rc = (Imax-Imin)/(Imax +

I min) of Sb delta-layers up to a critical angle of ~ 50o with Ep ~ 250 eV, ~ 55

o

with Ep ~ 500 eV and ~ 60o with Ep ~ 1 keV. van der Heide et al.

17 made similar


16

conclusions based on observations of surface roughness while working on B delta-

layers. The deterioration in depth resolution was attributed to rapid ripple formation at

oblique incidence.17,79

Kelly et al.80

observed the best depth resolution on B delta-

layers at ~ 50o (Ep ~ 1 keV). Similar resolution can only be achieved at ~ 60

o

when a trapezoidal scan correction is used. This technique corrects for the inclined

crater bottom caused by projection of the beam raster and spot shape onto the sample

at glancing angles, by creating a rectangular projection on the sample.

Li et al.81

used ultralow-energy Cs+ on B delta-layers in Si at ~ 45-80

o, and

observed that the depth resolution (estimated from Imax/Imin) degrades as is increased

up to ~ 70o. This observation, however, contradicts those made by Kataoka et al.

79

and van der Heide et al.17

cited earlier, where depth resolution improves up to only ~

50-60o. A similar conclusion could have been reached if profiles were done at ~ 50

o.

Li et al. observed the worst resolutions at 70o (Ep ~ 250-500 eV) and ~ 75

o (Ep ~ 1

keV), before improving to the „best‟ depth resolution at ~ 80o. Others have also

observed good depth resolution at glancing angle, ~ 85o using Ep ~ 1 keV.

82 To

obtain the best depth resolution whilst avoiding surface roughening, Kataoka et al.50

recommended the use of ~ 45-50o (Ep ~ 250 eV), ~ 50-55

o (Ep ~ 500 eV) and ~

55-60o (Ep ~ 1 keV).

Techniques such as oxygen flooding used with O2+ sputtering, have the

opposite effect when used with Cs+ sputtering. Ripple topography was observed at the

crater bottom when sputtered at Ep ~ 750 eV / ~ 60o in the presence of oxygen.

83


17

1.3 Outline of Research Project

Given the trends in the industry, there is a need to better understand the

capability and limitation of SIMS as an analytical tool to meet the more stringent

demands. The “push” from the industry has led to the development of ultralow-energy

SIMS by Dowsett et al.84

Though the use of ultralow-energy SIMS was initially

limited only to quadrupole SIMS, the capability is now possible also with magnetic

sector instruments.85

However, due to the design of the sector instrument, the incident

angle for the primary ion beam is limited to oblique angles. With the benefit of

ultralow-energy ion sources and a quadrupole analyzer, the ATOMIKA 4500 SIMS

Depth Profiler can cover a wide range of incident angles by simply tilting the

eucentric sample stage accordingly.

The objective of this work is to comprehensively study the effect of ultralow-

energy O2+ and Cs

+ primary ion beams on the Si transient width and depth resolution

over a wide range of incident angles and to a depth of about 120nm. The depth

resolution was systematically evaluated with established indicators: FWHM and d of

delta layers. Sputter rate variations, sensitivity and the dynamic range achievable were

also investigated. From the observations of this study, we will be able to draw

conclusions on the optimum conditions necessary to minimise the transient width and

to obtain high depth resolution.

The sample used was a Ge delta-doped Si sample with ten delta-layers at

regular intervals. The delta-layers are suitable for depth resolution studies as the effect

on band broadening can be observed. The sample was first subjected to O2+

bombardment at normal incidence and sputtered to beyond the last delta-layer before

the profile was terminated. Subsequently, the sample was tilted at 10o interval to


18

cover a wide range of angles from normal incidence to 70o. The scan sizes chosen

were sufficiently large to avoid crater wall effects. The sample current was also

monitored throughout the profile to observe for beam stability. A similar procedure

was done with the Cs+ primary ion beam.

The depth profiles obtained were analyzed for sputter rates, transient width,

depth resolution and dynamic range. Observations made with O2+ SIMS suggest that a

narrow transient width is possible not only at normal incidence but also at near-

normal incidence. There is no significant improvement in transient width going down

in energy from Ep ~ 500 eV to Ep ~ 250 eV. We also observed that although the

sputter rate during the transient is normally different from that at steady state, only at

Ep ~ 250 eV will the sputter rate remain fairly independent of depth. We confirmed

that depth resolution can be improved by lowering Ep to ultralow-energy. By varying

from 0o to 70

o, we noted that a better depth resolution is achievable not only at

normal incidence but over a wider range of incident angles as the probe energy is

reduced. Contributions from roughening and atomic mixing to the depth resolution of

-layers were discussed using the mixing-roughness-information depth (MRI) model.

Observations made with Cs+ SIMS suggested that the narrowest transient

width can be achieved at 30o < < 50

o. At ultralow energies, reducing Ep does not

have a significant effect in reducing the transient width. An extended transient effect

was observed when profiled at > 50o. We demonstrated that high depth resolution is

achievable with Ep ~ 320 eV and Ep ~ 500 eV at ~ 50o, and with Ep ~ 1 keV at ~

60o, over a significant depth (120 nm). We established that the relationship between

improvements in depth resolution (FWHM) is linear and gradual with increasing .

We also demonstrated that good depth resolution is achievable across a broader range


19

of . By using the MRI model, we were able to differentiate the effect of atomic

mixing and surface roughness on the depth resolution of -layers. The impact of

atomic mixing, surface roughness and instrument conditions (poor focus) on depth

resolution are discussed.

Finally, optimum conditions for depth profiling of ultrashallow junctions and

MQW are suggested.

References

1. C. W. Magee and M. R. Frost, International Journal of Mass Spectrometry

and Ion Processes 143, 29 (1995).

2. K. Tsukamoto, S. Yoshikawa, F. Toujou, H. Morita, Applied Surface Science

203, 404 (2003).

3. S. M. Sze, Physics of Semiconductor Devices (Wiley, New York, 1981), p. 1.

4. J. R. Brews, W. Fichtner, E. H. Nicollian, S. M. Sze, IEEE Electron Device

Lett. 2, (1980).

5. S. Wolf, Silicon Processing for the VLSI Era, Vol. 3 (Lattice Press, California,

1995), p. 591.

6. J. O. Borland, Proc. of the Fourth Int. Workshop on Junction Technology

(2004), p. 8.

7. E. P. Gusev, V. Narayanan, M. M. Frank, IBM Journal of Research &

Development 50, 387 (2006).

8. S. Nakai, M. Kojima, N. Misawa, M. Miyajima, S. Asai, S. Inagaki, Y. Iba, T.

Ohba, M. Kase, H. Kitada, and S. Satoh, IEEE International Electronic

Devices Meeting Technical Digest 11.3.1 (2003).

9. G. D. Wilks, R. M. Wallace, J. M. Anthony, Journal of Applied Physics 89,

5243 (2001).

10. H. Yang and G. Lucovsky, IEEE International Electronic Devices Meeting

Technical Digest 245 (1999).

11. P. Zory, Quantum Well Lasers (Academic Press, New York, 1993).

12. W. K. Choi, Applied Physics Letters 86, 021108 (2005).


20

13. S. Kakimoto, Journal of Applied Physics 92, 6403 (2002).

14. M. G. Dowsett, Applied Surface Science 203, 5 (2003).

15. P. A. W. van der Heide, Surface Science 447, 62 (2000).

16. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. Bennett, Applied Surface

Science 203, 156 (2003).

17. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. Bennett, Nuclear

Instruments and Methods in Physics Research B 201, 413 (2003).

18. P. C. Zalm, Rep. Prog. Phys. 58, 1321 (1995).

19. Z. X. Jiang and P. F. A. Alkemade, Surface and Interface Analysis 27, 125

(1999).

20. H. Tomizuka and A. Ayame, Applied Surface Science 89, 281 (1995).

21. K. Wittmaack, Philosophical Transactions of the Royal Society of London

Series A-Mathematical Physical and Engineering Sciences 354, 2731 (1996).

22. M. G. Dowsett et al., Applied Surface Science 203, 363 (2003).

23. Z. X. Jiang, P. F. A. Alkemade, E. Algra, S. Radelaar, Surface and Interface

Analysis 25, 285 (1997).

24. P. F. A. Alkemade, Z. X. Jiang, C. C. G. Visser, S. Radelaar, W. M.

Arnoldbik, Journal of Vacuum Science & Technology B 16, 373 (1998).

25. M. G. Dowsett, T. J. Ormsby, G. A. Cooke, D. P. Chu, Journal of Vacuum

Science & Technology B 16, 302 (1998).

26. J. B. Clegg, N. S. Smith, M. G. Dowsett, M. J. J. Theunissen, W. B. deBoer,

Journal of Vacuum Science & Technology A 14, 2645 (1996).

27. M. G. Dowsett, T. J. Ormsby, D. I. Ellinet, and G. A. Cooke Secondary Ion

Mass Spectrometry SIMS XI edited by G. Gillen, R. Lareau, J. Bennett, and F.

Stevie (Wiley, Chichester, 1998), p. 371.

28. D. P. Chu, M. G. Dowsett, T. J. Ormsby, and G. A.Cooke, Proceedings of

International Conference on Characterization & Metrology for ULSI

Technology. (1998)

29. R. G. Wilson, F. A. Stevie, and C. W. Magee, Secondary ion Mass

Spectrometry: A Practical Handbook for Depth Profiling and Bulk Impurity

Analysis ( John Wiley and Sons, New York, 1989)

30. K. Wittmaack and S. F. Corcoran, Journal of Vacuum Science & Technology

B 16, 272 (1998).


21

31. Z. X. Jiang and P. F. A. Alkemade, Journal of Vacuum Science & Technology

B 16, 1971 (1998).

32. C. M. Ng, A. T. S. Wee, C. H. A. Huan, C. S. Ho, N. Yakovlev, A. See,

Surface and Interface Analysis 33, 735 (2002).

33. K. L. Yeo, A. T. S. Wee, A. See, R. Liu, C. M. Ng, Applied Surface Science

203, 335 (2003).

34. K. L. Yeo, A. T. S. Wee, R. Liu, C. M. Ng, A. See, Surface and Interface

Analysis 33, 373 (2002).

35. K. L. Yeo, A. T. S. Wee, R. Liu, F. F. Zhou, A. See, Journal of Vacuum


36. T. Hoshi and M. Kudo Secondary Ion Mass Spectrometry SIMS XII edited by

A. Benninghoven, P. Bertrand, H. N. Migneon and H. W. Werner (Elsevier,

Amsterdam, 1998), p. 549.

37. G. R. Mount, C. J. Hitzman and S. P. Smith, Secondary Ion Mass

Spectrometry SIMS XI edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie

(Wiley, Chichester, 1998) p. 317.

38. W. Vandervorst and F. R. Shepherd, Journal of Vacuum Science &

Technology A 5, 313 (1987).

39. S. Yoshikawa, H. Morita, F. Toujou, T. Matsunaga, K. Tsukamoto, Applied

Surface Science 203, 252 (2003).

40. P. A. W. van der Heide and J. Bennett, Applied Surface Science 203, 306

(2003).

41. P. F. A. Alkemade and Z. X. Liu, Surface and Interface Analysis 31, 799

(2001).

42. P. A. W. van der Heide, Nuclear Instruments & Methods in Physics Research

B 194, 489 (2002).

43. H. Gnaser, Physical Review B 54, 16456 (1996).

44. Y. Kataoka, M. Shigeno, Y. Tada, K. Wittmaack, Applied Surface Science

203, 329 (2003).


46. M. L. Yu, Nuclear Instruments & Methods in Physics Research B 15, 151

(1986).

47. T. Eto and K. Shibahara, Japanese Journal of Applied Physics 44, 2433

(2005).


22

48. Y. Kataoka, K. Yamazaki, M. Shigeno, Y. Tada, K. Wittmaack, Applied


49. Y. Kataoka, M. Shigeno, K. Wittmaack, Journal of Vacuum Science &

Technology B 20, 441 (2002).

50. Y. Kataoka, M. Shigeno, Y. Tada, K. Yamazaki, M. Kase, Fujitsu Scientific &

Technical Journal 38, 69 (2002).

51. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. W. Rabalais, Journal of

Chemical Physics 113, 10344 (2000).

52. P. D. Augustus et al., Secondary Ion Mass Spectrometer SIMS VI Proceedings

(Wiley, New York, 1988), p. 485.

53. J. C. Vickerman, A. Brown, and N. M. Reed, Secondary Ion Mass

Spectrometry: Principles and Applications ( Oxford University Press, New

York, 1989).

54. K. Wittmaack, Dynamics SIMS and its Application in Microelectronics, in

Practical Surface Analysis Vol II. edited by D. Briggs and M. Seah (Wiley,

Chichester, 1992).

55. E. Napolitani, A. Carnera, V. Privitera, F. Priolo, Materials Science in

Semiconductor Processing 4, 55 (2001).

56. "ASTM E-42, Standard Terminology Relating to Surface Analysis E 673-

91c" (ASTM, Philadelphia, 1992).

57. M. G. Dowsett, Secondary Ion Mass Spectrometry SIMS X edited by A.

Benninghoven, B. Hagenhoff, H.W. Werner (Wiley, Chichester, 1996) p. 355.

58. H. H. Andersen, Applied Physics 18, 131 (1979).

59. A. Benninghoven, F. G. Rudenauer, H. W. Werner, Secondary Ion Mass

Spectrometry: Basic Concepts, Instrumental Aspects, Applications and Trends

(John Wiley & Sons, New York, 1987).

60. K. Wittmaack, Surface and Interface Analysis 21, 323 (1994).

61. M. G. Dowsett, R. D. Barlow, P. N. Allen, Journal of Vacuum Science &

Technology B 12, 186 (1994).

62. R. Liu, A. T. S. Wee, D. H. Shen, H. Takenaka, Surface and Interface

Analysis 36, 172 (2004).

63. C. M. Ng, A. T. S. Wee, C. H. A. Huan, A. See, Nuclear Instruments &

Methods in Physics Research B 179, 557 (2001).


23

64. T. J. Ormsby, D. P. Chu, M. G. Dowsett, G. A. Cooke, S. B. Patel, Applied


65. P. F. A. Alkemade and Z. X. Jiang, Journal of Vacuum Science & Technology

B 19, 1699 (2001).

66. Y. Matsuura, H. Shichi, Y. Mitsui, Journal of Vacuum Science & Technology

A 12, 2641 (1994).

67. M. G. Dowsett, R. D. Barlow, H. S. Fox, R. A. A. Kubiak, R. Collins, Journal

of Vacuum Science & Technology B 10, 336 (1992).

68. Z. X. Jiang et al., Journal of Vacuum Science & Technology B 22, 630 (2004).

69. R. Liu, C. M. Ng, A. T. S. Wee, Applied Surface Science 203, 256 (2003).

70. R. Liu and A. T. S. Wee, Applied Surface Science 231-2, 653 (2004).

71. N. S. Smith, M. G. Dowsett, B. McGregor and P. Phillips, Secondary Ion

Mass Spectrometry SIMS X edited by A. Benninghoven, B. Hagenhoff, H.W.

Werner (Wiley, Chichester, 1996) p. 363.

72. K. Wittmaack, Journal of Vacuum Science & Technology B 16, 2776 (1998).

73. P. F. A. Alkemade, Z. X. Jiang, J. A. van den Berg, R. Badheka and D. G.

Armour, Secondary Ion Mass Spectrometry SIMS XI edited by G. Gillen, R.

Lareau, J. Bennett, and F. Stevie (Wiley, Chichester, 1998) p. 375.

74. Z. X. Jiang and P. F. A. Alkemade, Secondary Ion Mass Spectrometry SIMS

XI edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie (Wiley, Chichester,

1998) p. 431.

75. C. M. Ng, A. T. S. Wee, C. H. A. Huan, A. See, Journal of Vacuum Science &

Technology B 19, 829 (2001).

76. J. J. Lee et al., Journal of Vacuum Science & Technology A 8, 2287 (1990).

77. J. M. McKinley et al., Journal of Vacuum Science & Technology B 18, 514

(2000).

78. K. Wittmaack, Journal of Vacuum Science & Technology A 3, 1350 (1985).

79. Y. Kataoka, K. Yamazaki, M. Shigeno, Y. Tada, K. Wittmaack, Applied


80. J. H. Kelly, M. G. Dowsett, P. Augustus, R. Beanland, Applied Surface

Science 203, 260 (2003).

81. Z. P. Li, T. Hoshi, R. Oiwa, Applied Surface Science 203, 323 (2003).


24

82. W. R. Morinville, and C. Blackmer, Secondary Ion Mass Spectrometry SIMS

XI edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie (Wiley, Chichester,

1998) p. 297.

83. P. A. W. van der Heide, M. S. Lim, S. S. Perry and J. Kulik, Secondary Ion

Mass Spectrometry SIMS XII edited by A. Benninghoven, P. Bertrand, H. N.

Migneon and H. W. Werner (Elsevier, Amsterdam, 2000) p. 485.

84. M. G. Dowsett, N. S. Smith, R. Bridgeland, D. Richards, A. C. Lovejoy and P.

Pedrick, Secondary Ion Mass Spectrometry SIMS X edited by A.

Benninghoven, B. Hagenhoff, H. W. Werner (John Wiley, Chichester, 1997)

p. 367.

85. E. de Chambost, A. Merkulov, M. Schumacher, P. Peres, Journal of Vacuum


Chapter 2:SIMS Principles

25

Chapter Two

SIMS principles

2.1 Introduction

In the SIMS process, a beam of high energy ions bombards a solid surface

resulting in a range of particles being sputtered from a finite volume at the surface.1

The energy of the primary ions is usually in the range of 1-20 keV, though more

recent instruments are capable of generating 100 eV beams. If the primary ion dose is

limited to less than 1013

cm-2

, then the analysis is restricted to the first few monolayers

of the surface causing minimal damage. This technique is referred to as static SIMS.2

When the primary ion dose exceeds 1017

cm-2

, the surface is continuously being

eroded thus generating a depth profile. This process is often referred to as dynamic

SIMS.3 Only dynamic SIMS will be discussed here, as it is the technique used in this

thesis. The secondary ions emitted are monitored with a mass analyzer as a function

of sputtering time, which corresponds to the depth at which the ions are located. The

depth scale or sputter rate can be calculated by measuring the physical depth of the

sputtered crater, assuming that the sputter rate is constant for a given bombardment

condition in the same matrix. The interaction of the secondary ions with the substrate

may result in neutralization or further ionization. The fraction or yield of secondary

ions produced from this interaction is dependent on the composition of the substrate

and the properties of the secondary ions.4,5

This is referred to as the matrix effect.6

The presence of chemically active species on the surface will also affect the

ionization yield. The implantation of chemically actives species such as O2+ enhances


26

the positive secondary ion yield from electropositive elements, while Cs+ ions

enhance negative secondary ion yields from electronegative elements.4,7

This matrix

effect has been favourably employed for signal enhancement, and O2+

and Cs+ ion

sources are now commonly installed as primary ion sources in SIMS.

When a highly focused primary ion beam such as Ga+ is used, it is possible to

achieve microanalysis and imaging. Lateral imaging of pre-selected elements can be

performed by raster scanning the Ga+ beam.

8 Three dimensional concentration

distributions are also possible if the image is continuously developed as the depth

profile progresses.9

2.2 Fundamentals of SIMS

2.2.1 Sputtering and collision theory

When a solid surface is irradiated with energetic ions, the ions that penetrate

the surface will lose energy by two main energy transfer process: nuclear stopping

when dominated by elastic atomic collisions with target atoms, and electronic

stopping when dominated by a non-collisional inelastic process of energy loss to

electrons in the target.10

Sigmund described the elastic collision process as knock-on

sputtering and the latter as sputtering by electronic excitation.11

Knock-on sputtering occurs mainly in metals and is distinguished by:

i) Single knock-on, where the impact is a direct energy transfer from the

primary ion to a surface atom in a single collision or a small number of

further collisions which results in the emission of a surface particle.

ii) Linear cascade, where the primary ion energy is dissipated via a series of

collisions among the lattice atoms. The recoil atoms involved in the


27

primary collision will in turn collide with other atoms resulting in a

collision cascade. Some target atoms and molecules near the surface may

acquire enough energy to overcome the surface binding energy and leave

the surface. The spatial density of moving atoms in this regime is small.

The linear cascade is the main process which is of interest in SIMS.

iii) Spike regime, where the process is similar to the linear cascade except that

the spatial density for moving atoms is large i.e. a majority of atoms within

a certain volume are in motion. Figure 2.1 illustrates some ion-solid

interactions.

Figure 2.1 Ion-solid interaction processes resulting in the emission of ions.

Emission

zone

(~1 nm)

Implantation

zone

(~10 nm)

Secondary ion

Single knock-on Linear cascade

Primary

ion


28

Sputtering by electronic excitation occurs mainly in insulators as the excited

electronic states have longer lifetimes, sufficient to allow energy to be transferred into

atomic motion.

The elastic binary collision theory as propounded by Sigmund is summarised

here.12

When a primary ion penetrates the surface of the target, it is slowed down by

the nuclear stopping cross-section, Sn(E) described as:

TTEdESn ),()( , (2.1)

where is the elastic collision cross-section, E is the initial energy and T is the

transferred energy (recoil energy) in a single collision. It is assumed that the energy

transferred in an elastic collision between two atoms obeys the conservation laws of

energy and momentum. An atom with mass M1 can transfer at most Tm energy to an

atom with mass M2 in a head-on collision.

E

MM

MM

ETm

2

21

21

)(

4

(2.2)

The nuclear stopping cross-section can be summarized in another form:

)(4)(21

12

21 nn sMM

MeZaZES

, (2.3)

where sn() is a universal function based on the Thomas-Fermi model of atomic

interaction,11

Z is the atomic number, e is the electron charge, a is the screening

radius, ao is the Bohr radius and

21

32

32

)(885.0 21

ZZaa o (2.4)

529.0oa Å (2.5)


29

If the recoil energy exceeds the lattice binding energy of an atom Eb, then the

impacted atom is set in motion and will in turn collide with other atoms in the

material. The resulting collision cascade will cease when T ≤ Eb i.e. the energetic

condition when atomic displacement can no longer occur. During the collision

cascade, each atom is displaced from its original lattice position thus giving rise to

atomic mixing, which is an isotropic process. A small fraction of the these back-

scattered target atoms with momentum towards the surface will remove the outermost

target atoms, if the energy transferred is sufficient to overcome the surface binding

energy of the surface atoms. This whole process is known as sputtering.

We have so far only considered elastic collisions. The non-collisional inelastic

electronic stopping process occurs mainly with higher projectile energies not used in

SIMS. In most sputtering situations in SIMS, the primary ion energies involved are

lower. At such lower velocities, a correction due to Lindhard and Scharf is normally

added to the nuclear stopping cross-section.13

The electronic stopping cross section is:

21

123

32

23

2

1

26

7

183.3)(

M

E

ZZ

ZZESe (2.6)

Sputtering from the surfaces produces mainly neutrals. Secondary ion yields

are 1% or less for most materials. In addition to the emission processes, ion

bombardment also results in changes to the surface zone of the solid. These regions in

the vicinity of the point of impact of the primary ion remain in a deformed or altered

state.


30

2.2.2 Sputtering yield

The erosion in the sputtering process is measured by the sputtering yield, Y

which is defined as the average number of atoms released at the solid surface per

incident ion. The sputtering yield is proportional to the number of displaced or recoil

atoms. In the linear cascade regime, the number of recoil atoms is proportional to the

energy deposited per unit depth in nuclear energy loss. The sputtering yield for

primary ions at normal incidence can be expressed as:14

),,(),,( xEFxEY D (2.7)

where x is the penetration depth of the primary ion, is the angle of incidence, FD is

the density of energy deposited at the surface and represents the target properties.

FD is dependent on the mass, energy and direction of the incident ions as well as the

composition of the target:

)(ESNF nD , (2.8)

where is a correction factor which takes into account the incident angle and

contributions due to large angle scattering events, and N is the atomic density of the

target atoms. The value of varies between 0.2 and 0.4 and increases with the

incident angle due to an increase in the amount of energy deposited at the surface.

The derivation of involves a description of the number of recoil atoms that

can overcome the surface energy and leave the solid. For the linear cascade regime,

Sigmund derived that

oNU

042.0 (2.9)

where Uo is the surface binding energy. Thus,


31

)(168.0

)(042.0

21

12

21

n

o

n

o

sMM

MeZaZ

U

ENSNU

Y

(2.10)

Hence, from equation 2.10 we can see that sputtering yield is dependent on both the

primary ion and the target parameters. The impact energy, atomic number, mass, and

incident angle of the primary ions are critical parameters. For the target, the surface

binding energy, atomic number, mass and screening radius of the target atoms directly

influence the sputtering yield. Other factors include the crystallinity and crystal

orientation of the solid, topography, and preferential sputtering.15

Sigmund‟s theory relates to normally incident ions. However, as the angle of

incidence is varied from the surface normal to some incident angle , the penetration

of the ion into the surface will decrease by cos ( for a constant range), and the whole

scattering cascade will therefore become more concentrated nearer the surface region.

The sputtering yield is then expected to increase11

by a factor 1/cos . The minimum

sputtering yield occurs at normal incidence, increasing gradually to a maximum at 60-

70o before decreasing at grazing angles.

16 The maximum is achieved due to a larger

cascade surface interaction. At grazing angles, less collision cascades occur due to a

significant scattering of primary ions from the surface.

2.2.3 Sputter rate

The sputter or erosion rate (ż) of a sample due to ion bombardment is

eA

nYI p

z , (2.11)


32

where n is the number of atoms in the primary ion (n=2 for O2+), is density of the

material (atoms cm-3

), e is electron charge, Y is sputtering yield, A is the area of the

crater and Ip is the primary ion current.15

Thus, the sputter rate is directly proportional

to sputtering yield and primary ion current density.

2.2.4 Secondary ion emission

In SIMS, only the charged atomic and molecular species are mass analysed.

The process involved in the production of secondary ions is strongly dependent on the

electronic state of the target material (matrix), ionization potential of the atoms for

positive secondary ions or the electron affinity for negative secondary ions and the

primary ion reactivity. Particles sputtered from the target surface are in the excited

metastable state, which may de-excite through photon or electron emission. Positive

ions are formed by electron emission. Alkali and alkaline earth metals have low

ionization potentials, thus these atoms have a high probability to form positive ions

compared to elements with higher ionization potentials. Conversely, negative ions are

formed by electron capture as the sputtering process also emits electrons. Halogens

and group IVA elements forms negative ions preferentially because of their high

electron affinity. These periodic trends for the elements are shown in figures 2.2 and

2.3.17

Besides the dependence on secondary ion generation, the intensity is also

dependent on the survivability of the generated secondary ions as they leave the

surface. In the case of positive ions, the survivability decreases with increasing

availability of electrons. Thus, targets with low workfunction such as gold have low


33

Figure 2.2 Relative positive ion yields for 13.5 keV O-, ~ 0


compound sample was used and B.D. means barely detectable.17

positive secondary ion yields. However, the electronic state of the target materials

such as metals can be modified by the presence of oxygen or the absorption of other

electronegative species on the surface.18

The absorption of such species increases the

work function of the surface and therefore increases the probability of positive metal

ions escaping the surface.

Conversely, the emission of negative secondary ions is enhanced when more

free electrons are made available by using electropositive ions such as Cs+. Thus, to

attain high sensitivity, the choice of primary ion species and the polarity of the


34

secondary ion are paramount. The ionisation probability also increases with mass of

the primary ion, but mainly because of an increase in sputtering yield.

Figure 2.3 Relative negative ion yields for 16.5 keV Cs+, ~ 0


compound sample was used and N.D. means not detectable.17


35

2.2.4.1 Secondary ion yield and ionisation probability

Secondary ion yield is defined as the average number of detected ions of a

selected species and polarity, per primary ion. The secondary ion yield, iM (counts s-1

)

of species M is given by the following equation, also known as the basic SIMS

equation:

MpM CYIi , (2.12)

where Ip is the primary ion beam current, Y is the sputtering yield, is the ionization

probability of M, CM is the fractional concentration of M in the surface layer and is

the transmission probability of the analytical system. Thus, sensitivity is very much a

function of ionization probability and the instrument transmission probability. Even

though can be optimised, it is limited by the geometry and the type of mass analyzer

used; for instance, magnetic sector SIMS has better transmission probability than

quadrupole SIMS. As mentioned in the previous section, the ionization probability

can be enhanced by modifying the surface electronic state to achieve greater

sensitivity.

2.2.4.2 Ionisation mechanism with O2+ ion beam

The presence of electronegative species such as oxygen due to oxygen

flooding or ion bombardment can enhance the positive secondary ion yield.19

The

enhancement has been observed to be as much as three orders of magnitude.20

The

most likely explanation for this enhancement effect is the bond-breaking

mechanism.21-23


36

The ionisation of the sputtered atom results from the breaking of chemical

bond with oxygen.24

The electron transfer is localised between the atomic state of the

sputtered atom and a particular localised substrate electronic state.

The local bond-breaking mechanism is based on the assumption that

secondary ion yield from ionic surfaces can be described using the Landau-Zener-

Stueckelberg theory25

of curve crossing for the ionic dissociation of diatomic

molecules. It also assumes that the ionisation/neutralisation takes place via resonance

charge transfer as the bond between the departing ion/atom and the substrate surface

breaks. The anion leaving a surface site can trap an electron from the outgoing atom

when the covalent and ionic curves for the original molecule cross at a certain

distance from the equilibrium position. Hence, the ion yield is related to the local

breaking of bonds.

2.2.4.3 Ionisation mechanism with Cs+ ion beam

When Cs+ or other electropositive elements such as alkaline and alkaline earth

metals were deposited or implanted onto/into solid surfaces, a strong correlation

between the secondary ion yield and work function of the sample surface was

observed.26,27

Negative ion yields were enhanced while positive ion yields were

suppressed.26-28

The work function Ф, which is the difference between the Fermi level (EF) and

the vacuum level (EV) is the minimum energy that is required to remove an electron

from the solid surface. The ionisation process involves interaction with electrons

coming from atoms beyond the immediate neighbours suggesting a non-local

interaction. The electron tunnelling model describes this phenomenon best.26,26


37

The presence of Cs+ causes a reduction in the surface work function,

effectively enabling electrons to tunnel from the surface into the unoccupied valence

states of sputtered atoms, resulting in ionisation or neutralisation of the departing

atoms. This model assumes the dominance of the resonance charge transfer process

taking place between electronic states within the substrate close to the Fermi edge

(valence electrons) and unoccupied valence states of the departing ion/atom. Only

valence electrons are considered since their wavefunctions extend furthest from the

nucleus.26

Hence, the efficiency of the electron transfer process depends on the

overlap of the wavefunctions of substrate and departing ion/atom.

The model predicts that the sputtered ion emission yield relates to exponential

scaling as:

]/)(exp[ onIP (2.13)

]/)(exp[ onAP (2.14)

where In and An are the ionisation potential and the electron affinity respectively of

the departing ion/atom.29-31

εo is a specific parameter describing the electronic

interaction and its relation with the normal component of the ion emission velocity.

Lang found εo to be 0.5 -1 eV but experimental evidence shows that εo is independent

of emission velocity with values between 0.2-0.4 eV.32

The relationship has been

verified by Yu26,33

for static conditions and Gnaser27

for dynamic conditions. Villegas

et al. recorded a greater work function change, the greater the Cs concentration on the

Si surface.34


38

2.2.4.4 Secondary ion species

The most predominant species observed in SIMS are the singly charged atomic

ions M; cluster ions Mn

, and molecular ions MxAy

. Doubly charged ions are

normally less abundant, though group II elements may be an exception.35

Higher

order molecular ions diminish in abundance as the order increases.36

Molecular ions

are formed by a combination of matrix atoms or by molecular species present on the

surface. Molecular ions have narrower energy distributions than monoatomic ions,

thus the use of energy bandpass can discriminate them. For example, 31

P- in Si can be

profiled without interference from cluster ion 30

SiH- in quadrupole SIMS by use of

energy bandpass or more simply by appropriate sample biasing.37

The cluster ion

MCs+ has been noted to reduce matrix effects during Cs

+ SIMS profiling, and the so-

called “MCs+ method” is now commonly used in semi-quantitative SIMS.

38,39

2.3 Factors affecting depth profiling

There are numerous factors that need to be considered in SIMS depth

profiling. The most critical factors that influence sensitivity and depth resolution are

the primary ion species, primary ion energy and primary ion angle of incidence. In

addition, crater edge effects must also be taken into consideration and avoided. When

insulators are analysed, sample charging needs to be minimized. An understanding of

the effects, influences and techniques developed to overcome such issues is important

for obtaining accurate depth profiles.


39

2.3.1 Primary ion species

The choice of primary ion beam species depends on several factors such as

sputtering yield, secondary ion yield, detection limits, depth resolution, lateral

resolution, crater topography and the most important consideration, availability. Many

primary ion species have been used in SIMS; for instance, O2+, Cs

+, O

-, Ar

+, Xe

+, Ga

+,

SF5+ and more recently C60

+. Ga

+ is widely used for SIMS imaging because of its

small beam diameter giving good lateral resolution (200-20 nm).40

The heavier the

primary ion mass41

such as Xe+, the better the depth resolution since it has a shallower

penetration depth. Moreover, molecular ion species such as C60+

disintegrate upon

reaching the surface thus dissipating their energy in the top few layers.42

Atomic ions

such as Ar+ and Xe

+ and molecular ions such as SF5

+, are used in the analysis of

polymers as they give high secondary ion yields of molecular fragments.43

However,

the most widely used primary ion species are O2+ and Cs

+. These two species provide

a good combination for general dynamic SIMS applications. The use of O2+ and Cs

+

result in respectable positive and negative secondary ion yield enhancement

respectively.

2.3.2 Primary ion energy

The choice of primary ion energy has a significant impact on ion yield and

sputtering yield44-46

as well as depth resolution.47,48

Generally, the lower the primary

ion energy, the narrower the ion-beam mixing depth and therefore the better the depth

resolution. However, lower primary ion energies will also lower the sputtering yield

resulting in longer analysis time. Similarly, the ion yield is also affected. The

secondary ion yield has been reported to increase with increasing primary ion energy


40

with Ar+ and Xe

+ (for Si

+ and Si

-), decrease with Cs

+ (for Si

-) and to remain constant

with O2+ (for Si

+) bombardment of silicon at normal incidence.

46,49,50 However, it has

been observed that at ultralow-energy (< 1 keV), the onset of roughening occurs

earlier i.e. at a shallower depth under conditions when surface roughening prevails.51-

53 This effect often negates any improvement in depth resolution in ultralow-energy

SIMS.

2.3.3 Primary ion angle of incidence

Changing the angle of incidence has an impact on the secondary ion yield,

sputtering yield and depth resolution since increasing the angle of incidence results in

a shallower penetration depth. The sputtering yield has been observed to increase with

increasing angle of incidence to a certain angle, about 60-70o for Si, before

decreasing.49,54

The secondary ion yield decreases as the angle of incidence is increased when

profiled with O2+ and Ar

+. The dependence is greater with O2

+ as the silicon surface is

completely oxidised at normal incidence. Less oxidation occurs with increasing angle

of incidence due to a drop in the primary beam incorporation into the silicon surface

and therefore less oxidation.16

Hence, the decrease in the ion yield since ion yield

increases with oxidation.

Generally, the depth resolution improves with increasing angle of incidence.

The improvement in depth resolution is a result of a narrower atomic mixing length

since the collision cascade occurs nearer the surface with increasing angle of

incidence. However, the onset of surface roughening at certain angle of incidence has

a negative impact on depth resolution as has been described in 1.2.3.1 earlier. Hence


41

the depth resolution improves with increasing angle of incidence as long as there is no

onset of roughening.

2.3.4 Crater edge effect

In a depth profile, the total rastered area cannot be used for detection of

secondary ions. This is because of contributions from the crater edges and side walls.

Re-deposition of ions sputtered from crater walls (crater-sidewall effects) has been

known to affect the depth profiles.41,55

Such sidewall contributions when detected are

seen as an increase in the ion concentration but are not representative of the depth

profile. During raster scanning of the primary ion beam, secondary ions are emitted

from all over the crater. The greatest contribution is at the crater edge when the

primary beam is near the end of the scan in the x or y direction.15

This is made worse

by non-uniform irradiation from a poorly focused beam and poor rastering quality

causing erosion inhomogeneity.1

To eliminate such crater edge effects, electronic and/or optical gating has been

implemented in SIMS instruments. In electronic gating, there will be no detection

during the raster scan except when the beam is scanning over the specified area near

the centre of the crater. Therefore, only secondary ions emitted from the flat centre of

the crater are collected for analysis. Similarly, in optical gating, the ion optics in the

secondary ion optics column determines the field of view of the crater base for

analysis. The overall effect of gating results in an improvement of the dynamic range

over which the secondary ions of interest are monitored.15


42

2.3.5 Sample charging effect

Positive primary ion bombardment emits secondary ions, neutrals and

secondary electrons, giving a net electric charge at the sample surface. Thus, for

insulating surfaces, a positive charge will build up at the surface since there is no

conduction path to ground. The effect of this charging is that the energy distribution

and paths of positive secondary ions are altered, accelerating them beyond the

acceptance energy of the mass analyzer. This is particularly a problem with a

quadrupole analyzer which has a narrow energy acceptance window (typically 5 eV).

In the case of negative secondary ions, their emission will be inhibited by the

presence of a positive surface potential.

To neutralize the charging effect on the surface, low energy electron flooding

(0.1 – 5 keV) has been used effectively.56,57

This has been found to be successful for

almost all matrices attempted.58

However, this electron beam neutralization procedure

is difficult with magnetic sector SIMS and TOF-SIMS instruments because of the

high secondary ion extraction fields used and the restricted geometry of the sample

stage.

In magnetic sector instruments, a normal incidence electron beam

neutralisation technique is used.59

High energy electrons are directed along the surface

normal in the opposite direction to the secondary ion extraction also optimised at

normal to the surface. The high ion extraction field at the sample repels the incoming

electrons such that they arrive at the sample surface with virtually zero energy as

required to neutralise the charge build-up. Recently, a combined electron and

focussed ion-beam system has also been demonstrated to be effective in eliminating

sample charging in sector instruments.60

In TOF-SIMS, charge compensation is done


43

by exposing low-energy electron beam to the sample in between the primary ion beam

pulses.61

Another common method to overcome the surface charging is to coat the

sample with conductive materials like gold or carbon.62,63

2.4 Quantification of depth profiles

The raw data obtained in SIMS is the intensity (counts s-1

) of the profiled

element against the sputtered time (s). To quantify, we will need to convert the

secondary ion counts to atomic concentrations and the sputtering time to a depth

scale. Quantitative analysis in SIMS is complicated because of the varying elemental

ion yields and the various factors that affect secondary ion yields as discussed earlier.

As such, quantification based on fundamental theories has not been very forthcoming.

The use of absolute standards and relative sensitivity factors (RSF) has been more

acceptable and practised.

2.4.1 Relative sensitivity factors and absolute standards

The relative sensitivity factor (RSF) is a factor converting the secondary ion

intensity to atom density. The RSF of an element of a particular isotope is determined

by measuring the isotope‟s ion intensity from a known matrix and recording it as a

ratio to a reference element.41

The matrix element is usually chosen as the reference

as it is the most abundant. Normalising to the reference element effectively removes

possible variations in primary ion current, sputtering yield and instrument

transmission factors. The RSF of the element of interest i in matrix m can thus be

expressed as follows:


44

RSFI

I

m

ii , (2.15)

where I is the secondary ion intensity and ρi is the concentration or atom density of

the element of interest (atoms cm-3

). Rearranging equation 2.15 gives:

i

mi

I

IRSF , (2.16)

For accurate quantification, the instrument and analysis conditions must be

identical for the standards and samples. The conditions include the primary ion energy

and incident angle, primary ion current density, energy bandpass and detection

efficiency of the mass spectrometer.

However, elemental standards with fixed concentration level for various

impurity/matrix combinations are not always possible. This is a major setback of the

RSF method.5 Ion implanted standards in the matrix of interest are an ideal

alternative.64

Such standards are available commercially or can be made by

implanting a specified dose (expressed in atoms cm-2

) of a specific isotope of the

element of interest into an undoped matrix. The unknown element of interest and the

implanted standard in similar matrices must be profiled under identical conditions.

Thus, the absolute sensitivity factor (ASF) is:

stop

start

x

R

x

R

I

ASF

(2.17)

where Rx is the dose (atoms cm

-2) of the reference material for species x and IR

x is the

yield or secondary ion intensity (counts s-1

) of the implant profile for species x of the


45

reference material, summed up from the beginning (start) of the profile to the end

(stop) of the profile.

A conversion of intensity (counts s-1

) to concentration (atoms cm-3

) for the

element of interest involves the erosion rate as:

total

x

i

iz

ASFI

. , (2.18)

or

z

ASFI x

ii

. , (2.19)

where ztotal is the total depth of the implant and ∆z is the depth increment. The

equation involving ASF can also be normalized to a matrix ion signal similar to the

RSF concept.

2.4.2 Depth calibration

To convert the time scale to a depth scale, the resulting crater after sputtering

must be physically measured. The measurement can be done with a stylus

profilometer. The physical depth measured is then converted to a depth scale with the

assumption that the erosion rate is constant throughout the sputtered time. This

assumption is valid provided that the matrix composition and the beam current

remains constant throughout the depth profiled.

The assumption of a constant erosion rate can be erroneous in a number of

circumstances. Firstly, a sample with particles and/or roughness will give a distorted

depth information, characteristic of the surface topography prior to sputtering.

Secondly, at the initial stage of primary ion bombardment, the presence of surface


46

impurities such as native oxide will result in a higher ion yield when bombarded with

O2+. Subsequently, as the concentration of the probe ions build up in the near surface

region, the sample forms an altered layer. As discussed earlier, in the surface transient

region the erosion rate varies until a steady state is achieved at equilibrium. Thirdly,

the incorporation of the probe ions will also cause swelling of the crater base. The

swelling will result in the final crater depth to be slightly underestimated.65

The error in the depth profile is also known as differential shift.66

The size of

the displacement is energy dependent and may amount to 2 nm keV-1

of primary ion

energy. The differential shift in the profile can be calculated and corrected for, by

profiling a referenced delta-layer. By plotting the position of the delta-layer centroid

versus energy, it is possible to extrapolate the position of the feature at 0 keV.67

However, at ultralow-energy it is difficult to ascertain the differential shift effect as

the accuracy of the profilometer does not exceed 3-4%.68

For multilayer samples, the erosion rates will differ from layer to layer. The

erosion rate of each layer will have to be determined separately and computed into the

depth scale calibration.


47

References

1. A. Benninghoven, F. G. Rudenauer, and H. W. Werner, Secondary Ion Mass

Spectrometry: Basic Concepts, Instrumental Aspects, Applications and Trends

(John Wiley & Sons, New York, 1987).

2. M. G. Dowsett and E. A. Clark, Dynamics SIMS and its Application in

Microelectronics, in Practical Surface Analysis Vol II, edited by D. Briggs and

M. Seah (Wiley, Chichester, 1992), p. 229

3. M. G. Dowsett, C. Jeynes, E. A. Clark, R. Webb, and S. M. Newstead,

Secondary Ion Mass Spectrometry SIMS VII, edited by A. Benninghoven, C. A.

Evans, K. D. McKeegan, H. A. Storms and H. W. Werner (Wiley, Chichester,

1990), p. 615.

4. M. L. Yu, in: Sputtering by Particle Bombardment III, edited by R. Behrisch

and K. Wittmaack (Springer-Verlag, Berlin, 1991), p. 92



7. V. R. Deline, C. A. Evans, P. Williams, Applied Physics Letters 33, 578 (1978).

8. S. Seki, H. Tamura, S. Horita, N. Ito, Surface and Interface Analysis 36, 896

(2004).

9. P. A. W. van der Heide, Secondary Ion Mass Spectrometry SIMS XI, edited by

G. Gillen, R. Lareau, J. Bennett, and F. Stevie (Wiley, Chichester, 1998), p.

821.

10. M. Nastasi, J. W. Mayer, and J. K. Hirvonen, Ion-Solids Interactions:

Fundamentals and Applications (Cambridge University Press, Cambridge,

1996), p. 99.

11. P. Sigmund, Sputtering by Ion Bombardment: Theoretical Concepts in

Sputtering by Particle Bombardment I ( Springer - Verlag, Berlin, 1981), p. 9.

12. P. Sigmund, Physical Review 184, 383 (1969).



1996), p. 110.

14. H. H. Andersen and H. L. Bay, in Sputtering by Particle Bombardment I

(Springer-Verlag, Berlin, 1981), Chap. 4.


48


Spectrometry: Principles and Applications (Oxford University Press, New York,

1989).

16. K. Wittmaack, Nuclear Instruments and Methods in Physics Research B 218,

307 (1983).

17. H. A. Storms, K. F. Brown, J. D. Stein, Analytical Chemistry 49, 2023 (1977).

18. J. L. Maul and K. Wittmaack, Surface Science 47, 358 (1975).

19. A. Benninghoven, Surface Science 53, 596 (1975).

20. M. L. Yu and N. D. Lang, Nuclear Instruments & Methods in Physics Research

B 14, 403 (1986).

21. P. Williams, Surface Science 90, 588 (1979).

22. G. Slodzian, Surface Science 48, 161 (1975).

23. M. L. Yu, Nuclear Instruments & Methods in Physics Research B 15, 151

(1986).

24. M. L. Yu and K. Mann, Physical Review Letters 57, 1476 (1986).

25. C. Coudray, and G. Slodzian, Physical Review B 49, 9344 (1994).

26. M. L. Yu, Physical Review Letters 40, 574 (1978).

27. H. Gnaser, Physical Review B 54, 16456 (1996).

28. M. L. Yu, Physical Review B 24, 1147 (1981).

29. N. D.Lang, Physical Review B 27, 2019 (1983).

30. J. K. Norskov and B. I. Lundqvist, Physical Review B 19, 5661 (1979).

31. M. L. Yu, in Sputtering by Particle Bombardment III, edited by R. Behrisch and

K. Wittmaack (Springer-Verlag, Berlin, 1991), p. 98.

32. M. Bernheim and F. Le Bourse, Nuclear Instruments and Methods in Physics

Research B 27, 94 (1987).


34. A. Villegas, Y. Kudriavtsev, A. Godines, R. Asomoza, Applied Surface Science

203-204, 94 (2003).

35. R. Holland and G. W.Blackmore, International Journal of Mass Spectrometry

and Ion Physics 46, 527 (1983).


49

36. R. J. Colton, M. M. Ross, D. A. Kidwell, Nuclear Instruments and Methods in

Physics Research B 13, 259 (1986).

37. K. Wittmaack, Surface Science 53, 626 (1975).

38. Y. Gao, Journal of Applied Physics 64, 3760 (1988).

39. T. Wirtz, H. N. Migeon, H. Scherrer, Applied Surface Science 203-204, 189

(2003).

40. P. Williams and C. A. Evans, Applied Physics Letters 30, 559 (1977).

41. R. G. Wilson, F. A. Stevie, and C. W. Magee, Secondary ion Mass

Spectrometry: A Practical Handbook for Depth Profiling and Bulk Impurity

Analysis (John Wiley and Sons, New York, 1989).

42. R. Hill, R. Blenkinsopp, A. Barber, C. Everest, Applied Surface Science 252,

7304 (2006).

43. F. Kotter and A. Benninghoven, Applied Surface Science 133, 47 (1998).



46. P. C. Zalm, Journal of Applied Physics 54, 2660 (1983).

47. P. Williams and J. E. Baker, Nuclear Instruments & Methods 182, 15 (1981).

48. K. Wittmaack, Vacuum 34, 119 (1984).

49. A. E. Morgan, H. A. M. de Grefte, N. Warmolts, H. W. Werner, H. J. Tolle,

Applied Surface Science 7, 372 (1981).

50. K. Wittmaack, Applied Surface Science 9, 315 (1981).

51. Z. X. Jiang and P. F. A. Alkemade, Applied Physics Letters 73, 315 (1998).




55. C. W. Magee and R. E. Honig, Surface and Interface Analysis 4, 35 (1982).

56. B. Guzman de la Mata, M. G. Dowsett, R. J. H. Morris, Journal of Vacuum

Science & Technology A 24, 953 (2006).

57. I. S. Gilmore, and M. P. Seah, Surface and Interface Analysis 23, 191 (1995).


50

58. K. Wittmaack, Applied Physics 12, 149 (1977).

59. G. Slodzian, M. Chaintreau, R. Dennebouy, Secondary ion Mass Spectrometry

SIMS V, edited by A. Benninghoven, R. J. Colton, D. S. Simmons and H. W.

Werner (Springer, Berlin,1986), p. 158.

60. L. Ji, Q. Ji, K. N. Leung, and R. A. Gough, Applied Physics Letters 89, 164103

(2006).

61. B. Hagenhoff, D. van Leyen, E. Niehuis, and A. Benninghoven, Journal of

Vacuum Science & Technology B 7, 3056 (1989).

62. H. W. Werner and A. E. Morgan, Journal of Applied Physics 47, 1232 (1976).

63. D. S. McPhail, M. G. Dowsett and E. H. C. Parker, Journal of Applied Physics

60, 2573 (1986).

64. D. P. Leta, and G. H. Morrison, Analytical Chemistry 52, 277 (1980).

65. R. D. Barlow, M. G. Dowsett, H. S. Fox, R. A. A. Kubiak, S. M. Newstead,

Nuclear Instruments and Methods in Physics Research B 72, 42 (1992).



67. M. G. Dowsett, R. D. Barlow, P. N. Allen, Journal of Vacuum Science &

Technology B 12, 186 (1994).



Chapter 3: Experimental and instrumentation

51

Chapter 3

Experimental and instrumentation

3.1 SIMS instrumentation

3.1.1 Introduction

There are three main types of SIMS instrumentation that are commercially

available namely; time-of-flight SIMS (TOF-SIMS),1,2

double focusing magnetic

sector SIMS (M-SIMS)3 and quadrupole SIMS (Q-SIMS).

4 Each of these instruments

has their strengths and has been known to excel in certain applications. TOF-SIMS is

best suited for static SIMS and organics analysis, M-SIMS is best known for its high

mass resolution and Q-SIMS for its ultralow-energy capability and charge

compensation with insulators.

The SIMS instrument used for this work is the ATOMIKA SIMS 4500

instrument with a quadrupole mass analyzer.5 The system is equipped with two

Floating Low energy Ion Guns (FLIGtm

) which are mounted perpendicular to each

other on the main analysis chamber.6 Each FLIG can produce a mono-energetic ion

beam in the range of 150 eV to 5 keV. The ion beams are mass filtered by a Wien

filter to ensure ion beam purity.7 The ions produced at source are extracted and

directed to the sample via the ion optics in the ion gun column. The sample

manipulator allows for flexibility in positioning the sample. A fine focus electron

beam gun is also positioned at the analysis chamber for the purpose of sample charge

equilibration that is normally required for non-conducting samples. The secondary

ions produced are extracted with a low field via an acceleration - deceleration type


52

Figure 3.1 Front view of the ATOMIKA 4500

secondary ion optics, designed to maximize the secondary ion collection efficiency.

Finally, the secondary ions are mass analysed with the quadrupole mass analyzer and

detected by a channeltron. A schematic of the SIMS apparatus is shown in figure 3.1

and a photo of the system is shown in figure 3.2.

3.1.2 Vacuum system and sample handling

3.1.2.1 Vacuum system and monitors

An ultra high vacuum (UHV) environment is necessary to maintain

reproducibility and good detection limits especially for species present in the residual

gas. UHV conditions are necessary to minimise collisions in the path of the primary

and secondary particles with other particles in the vacuum and to prevent gas

adsorption onto the clean sputtered surface. Typically, at a vacuum pressure of 10-6

Sample holder

Ion Sources

FLIG Ion Optics

Main Chamber

Quadrupole Mass Spectrometer and Secondary Ion Detector

Sample Transfer System

Cesium Ion Gun Oxygen Ion Gun

Secondary Ion Optics


53

Figure 3.2 The ATOMIKA 4500 SIMS Depth Profiler

mbar, a surface will absorb one monolayer of air molecules in about 2.5 s.8 Thus, to

ensure that a surface can remain clean beyond 250 s, a vacuum pressure of less than

10-8

mbar must prevail. To maintain a good vacuum, the chambers are constructed of

stainless steel and divided into three isolated volumes; the sample transfer system

(STS), the analysis chamber and the differential stage at the primary ion guns.

Stainless steel is used because of its strength, low vapour pressure and resistance to

corrosion. The STS and the differential stage are pumped by individual 70 l s-1

turbomolecular pumps with a diaphragm pump each for roughing and backing. The

analysis chamber is equipped with a 220 l s-1

ion pump and also a titanium

sublimation pump. The base pressure in the main chamber is 3 x 10-10

mbar under


54

Figure 3.3 Schematic of the vacuum system for SIMS 45009

optimal conditions. The working pressure is monitored by ion gauge controllers at

several locations. Figure 3.3 shows a schematic of the vacuum system.

3.1.2.2 Sample introduction

Samples for analysis are mounted onto a sample holder and held down by a

spring. Several samples of between 5-10 mm2 can be accommodated. Only the surface

to be analyzed is exposed through the sample holder window. This ensures that all

samples are at the same height and also acts to shape the local extraction field above

them. This type of holder is preferred to the top-loading design which consists of a

metal plate onto which samples are glued with colloidal silver or carbon tape. The

loaded sample holder is then transferred into the sample transfer system which is

vented with nitrogen prior to opening to the atmosphere. The STS allows fast sample

Forepump Cs+ source

Forepump STS

Titanium sublimation pump

Ion pump

Forepump O2

+ source

Bottle valve

Reducing valve

Bypass valve

Leak valve

Bypass valve

Gate valve

Gate valve

Gate valve Ventilation valve

Ventilation valve

Ventilation valve

Analysis Chamber

TMP

IG-Ion gauge IGC-Ion gauge controller TMP-Turbo molecular pump

IG

IG

IG

IGC

IG

IGC

IGC

IGC

TMP

TMP

Cs+

Source O2

+

Source


55

loading/exchange via the load lock without affecting the analysis chamber vacuum. It

is separated from the main chamber by a gate valve. An interlock is maintained to

prevent the gate valve from opening before the STS reaches a vacuum of 6 x 10-6

mbar to prevent the main analysis chamber from being exposed to poor vacuum. In

the STS chamber, there are two stubs for mounting two sample holders. When the

necessary sample holder has been loaded, the chamber is evacuated again to the

acceptable vacuum pressure before the gate valve to the main analysis chamber can be

opened. The sample holder can now be transferred to the analysis chamber onto the

sample stage.

3.1.2.3 Sample manipulator

The sample manipulator provides three degrees of sample motion. The

primary ion impact angle can be varied in a eucentric manner from 0o to 85

o with

respect to the ion beam by tilting the sample holder.5 Hence, 0

o incident angle is when

the ion beam is normal to the sample surface. The eucentric sample manipulator

ensures that there is no change in the position of ion beam impact on the sample when

the sample holder is tilted. A combination of sample holder rotation and translation is

used to position the area of interest. Sample alignment is also aided by an optical

video camera system which is fitted with a zoom microscope lens allowing visual

selection of the area of interest. External lighting via fibre optics is used to illuminate

the sample. The image of the area of interest can be observed on a monochrome

monitor.


56

3.1.3 Primary ion gun

The system is equipped with two ion guns. One ion gun is equipped with a

duoplasmatron source for oxygen or argon and another is fitted with a thermal

ionisation source for cesium. Typically, an ion gun comprises an ion source and a

series of lenses to extract ions from the source and focus the beam onto the sample.

To force ions to move from the source to the sample, a potential difference is applied

between the source and the sample. The sample is normally grounded or held at a

small bias and the source region is floated to the accelerating voltage. X-Y deflection

plates are used to raster scan. A mass energy filter (Wien filter) is used to remove

contaminant ions and other multiple charged ions of the main beam. Only ions of the

required mass and energy are allowed through whilst other unwanted species are

deflected away from the beam. Neutral particles that are not affected are removed

through a slight bend further down the ion column.

3.1.3.1 O2+ source

The source comprises a cathode, an anode and an intermediate electrode. The

intermediate electrode is mounted between the anode and cathode to create two

discrete plasma chambers. A schematic of the source is shown in figure 3.4. A leak

valve is used to introduce oxygen into the source chamber. The high current density in

the cathode region ionises the gas efficiently producing dense plasma. The gas

pressure in the source is usually greater than 2 x 10-6

mbar. The intermediate electrode

compresses the discharged plasma magnetically. The dense plasma is then allowed to

expand into an expansion cup through a small orifice in the anode which completes


57

Figure 3.4 Schematic of duoplasmatron ion source.

the magnetic circuit. Ions are then extracted from the plasma surface by an electrical

field between the anode and extraction electrode facing the plasma.

3.1.3.2 Cs+ source

The cesium source is of the surface ionisation type. The schematics are shown

in figure 3.5. Cesium contained in the reservoir is evaporated by the reservoir heater

into the ioniser. At the end of the ioniser feed tube is an apertured porous frit made of

tungsten, a high work function metal. The tungsten frit is heated by passing a constant

current to the frit heater. After diffusing through the frit, the cesium atom leaves

behind an electron and becomes ionised. However, this ionisation process is known to

fail if too much cesium collects on the tungsten substrate, hence the amount of cesium

evaporated need to be carefully controlled. The whole source is floated at high voltage

which determines the ion energy. A high extraction field is maintained to prevent a

Gas (O2)

Cathode & holder

magnet anode

Extraction electrode (ground)

Intermediate electrode

O2+, O-


58

Figure 3.5 Schematic of cesium surface ionisation source

build up of ions at the surface of the ioniser. The rate at which the cesium atoms

arrive at the surface of the ioniser is determined by the vapour pressure of the cesium

in the reservoir and the porosity of the frit. The cesium vapour pressure depends on

the temperature of the reservoir which can be controlled.

3.1.3.3 FLIG

The FLIG (Floating Low energy Ion Gun) is designed to produce low energy

ion beams with high beam currents. It is self-aligning on assembly and the main

optical elements are a compound accelerating immersion lens for ion extraction from

source (lens 1), a variable intermediate aperture and a compound retarding immersion

lens (lens 2) as the objective.

Ions are extracted and transported at high energy along the column at high

voltage (up to 5 kV). A float voltage will allow a deceleration at the final lens for a

low energy impact on the sample. Despite the deceleration, the design of the optics is

reservoir heater

cesium reservoir

porous tungsten frit

frit heater

feed tube accelerator electrode

focus electrode

Cs Cs+


59

Figure 3.6 Schematics of the FLIG9

such that a good beam current is maintained. A schematic of the FLIG is shown in

figure 3.6.

The extractor and lens 1 is mounted at the opposite end of the source flange.

Immediately after lens 1 is the Wien filter and the first alignment unit. These

assemblies are in the manifold housing where there is a differential pumping port.

Source anode aperture

Differential pumping port

Extractor & lens 1

Mini gate valve

Wien mass filter & x-align plates

Main isolator ceramic ring

Pressure step aperture

10-pin feedthrough port

Stigmator plates

Scan plates

Neutral dump with x- & y-alignment plates

Lens 2

Micrometer drive with aperture strip


60

This whole section of the ion gun floats at the float voltage (0-5 kV) when in

operation. A mini gate valve is positioned to isolate the lower column from the

extractor and source which is necessary for source replacement or maintenance. It

also serves to block the beam when it is not in use. An adjustable aperture mechanism

with six apertures is also available to control beam current and spot size.

The lower section of the gun is designed with a 2o bend to reject neutrals. The

stigmators and a second alignment unit are mounted in the first segment after the bend

and the retarding lens (lens 2) in the second segment. The lower column assembly

floats at the float voltage. Lens 2 focuses and decelerates the beam. After lens 2 is a

set of X and Y raster plates for high precision scanning.

3.1.4 Secondary ion optics

The secondary ion optics (SIO) is designed to collect as many ions from the

defined sputtered region and transfer to the mass analyzer. It also limits the energy of

these ions, thus enhancing the mass resolution of the quadrupole. Unlike magnetic

sector instruments, the ultralow extraction potential at the SIO does not influence the

settings on the primary ion beam, especially the incident angle.

3.1.5 Quadrupole mass spectrometer

The quadrupole mass spectrometer (QMS) is positioned at the exit of the

secondary ion optics. Ion species that enter the QMS are channelled through to the ion

detector sequentially on the basis of their mass-to-charge (m/e) ratio.

The QMS is constructed of four electrically conducting round electrodes

(rods) held in a square array and parallel to each other as shown in figure 3.7.


61

Figure 3.7 Schematic of the quadrupole mass spectrometer connections

Opposite electrodes are connected together electrically giving a two-dimensional (x-

y) quadrupole field. During operation, one pair (x-z plane) is connected to a positive

DC voltage and a sinusoidal RF voltage is superimposed. Similarly, the other pair of

electrodes (y-z plane) is connected to a negative DC voltage with a sinusoidal RF

voltage superimposed, but 180o out of phase with the RF voltage to the first pair. The

potentials applied are represented by the following expression:

)]cos([)( tVUt ,

where U is the DC voltage, V is the amplitude of the RF voltage applied to either pair

of rods and is the angular frequency (2f) of the RF. Under these circumstances, the


62

equipotential surfaces are symmetric hyperbolic cylinders and the potential on the z-

axis is zero.

If an ion is injected at the end of this assembly with motion generally parallel

to the z-axis, the RF and DC fields perpendicular to the z-axis will cause it to undergo

transverse motion. Light ions (low m/e ratio) will be able to follow the alternating

component of the field. In the x-direction, ions are in phase with the RF drive and

gain energy from the field, oscillating with increasingly large amplitudes until they

collide with one of the rods and are discharged. Therefore, the x-direction acts as a

high-pass mass filter i.e. only high masses will be transmitted to the other end of the

quadrupole without striking the x-electrodes.

On the other hand, heavy ions will be unstable in the y-direction, because of

the defocusing effect of the DC component. However, lighter ions will be stabilized

by the alternating component provided that its magnitude and amplitude are such as to

correct the trajectory whenever its amplitude tends to increase. Therefore, the y-

direction can be considered as a low-pass mass filter, i.e. only low masses will be

transmitted to the opposite end of the quadrupole without striking the y-electrodes.

Thus with proper selection of RF/DC ratio, the two directions together can be

made to discriminate against both high and low mass ions. The RF magnitude and

frequency determine the mass of the ions that can undergo stable trajectories through

the quadrupole in the z-axis. As the RF amplitude increases, heavier ions begin to

oscillate in phase with the RF and collide with the rods. The RF/DC ratio determines

the mass filter selectivity, i.e. the mass resolution.4

Quadrupole mass spectrometers cannot differentiate ion species that have

similar masses, unlike magnetic sector instruments which have better mass resolution.


63

However, molecular ion interference can be removed by setting the ion energy filter

to accept only higher energy ions which are predominantly atomic ions.

3.1.6 Secondary ion detection

Ions emerging from the quadrupole mass analyser are detected with a channel

electron multiplier (channeltron). This is a horn-shaped glass tube coated with a

resistive material on the inside wall which acts as the secondary electron emitting

surface. The large opening provides a higher probability for an incident ion to be

captured. When a potential is applied between the ends of the tube, the resistive

surface forms a continuous dynode. Electrons produced are multiplied down the

curved channel and pulses are counted via the electronics in counts per second. The

pulse count is proportional to the number of ions detected.

3.1.7 Data acquisition and electronics

There are several electronic modules, each controlling a specific part of the

SIMS operations. A programmable logic controller (PLC) is also available to ensure

safety as well as to control the vacuum interlocks. A computer system is used for data

acquisition and data analysis.


64

3.2 Atomic Force Microscopy

3.2.1 Principles

Atomic force microscopy (AFM) is widely used for surface topography

imaging down to atomic resolution and for a wide variety of fundamental surface

science applications.10-12

Unlike scanning tunnelling microscopy (STM) which is

limited to semiconductors and conductive materials, AFM can be used with insulators

too as it does not work on the phenomenon of electron tunneling.13

In the atomic force

microscope, an atomically sharp tip typically made of silicon nitride (Si3N4) or single

crystalline silicon (Si), located at the free end of a cantilever, is scanned over the

surface of the sample.12,13

As the tip moves over the sample (or the sample traverses

under the tip), forces between the tip and the surface cause the cantilever to bend or

deflect depending on the contour of the surface.

A laser and/or a position sensitive detector measure the deflection of the

cantilever and feedback to a control system. The signals received enable the control

system to maintain either a constant force or a constant height above the sample as

well as generating a map of the surface topography.12

Several forces are involved in the deflection of the cantilever, the most

common being the van der Waals force. The dependence of the van der Waals force

upon the distance of the tip from the sample is described in figure 3.8.

As the tip moves closer to the surface, the atoms between the tip and the

surface weakly attract each other by the long range attractive force. This is the

distance in which the AFM operates in the non-contact regime. The attraction

increases until the atoms are in close proximity where their electron clouds will repel


65

Figure 3.8 Interatomic force vs. distance curve.11

each other electrostatically. The electron-electron electrostatic repulsion progressively

weakens the attractive force as the interatomic distance decreases. This force goes to

zero when the interatomic distance is a few angstroms, about the length of a chemical

bond. When the van der Waals force becomes positive (repulsive), the atoms are in

contact. The AFM now operates in the contact mode. When a vibrating tip approaches

the sample, they come into intermittent contact. The atoms between the tip and

sample, alternate between being repelled by the short range repulsive force and

attracted by the attractive force. This mode is also known as the tapping or

(intermittent-contact) mode. These modes are described further in the following.

Force

Distance (tip-to-sample separation)

contact

non-contact

intermittent-contact

attractive force

repulsive force


66

3.2.2 Contact mode

In this mode, the tip is held less than a few angstroms from the sample surface,

and the resultant interatomic force between the cantilever and the surface is repulsive.

The repulsive van der Waals force will counter any force acting against it. This

translates into the bending of the cantilever as the atoms at the tip are close to the

atoms on the sample surface. Other than the van der Waals force and the force exerted

by the cantilever, a capillary force exerted by the presence of a thin film of water is

also present. This capillary force (~ 10-8

N) arises when water around the tip holds the

tip in contact with the surface. This force is constant as long as the tip is in contact

with the surface. Hence, the total force exerted on the sample is the variable cantilever

force plus the constant capillary force which will be countered by the repulsive van

der Waals force. The deflection of the cantilever is sensed and the voltage that the

feedback amplifier applies to the piezoelectric device to maintain constant-height or

constant-force is translated to topographical information. Since the repulsive force is

not unique to any material surface, AFM is applicable to conducting as well as

insulating materials.

However, in the contact mode, the sample surface may be damaged due to

interaction of the hard tip with the surface. This mode is therefore not suitable for soft

or elastic samples.

3.2.3 Non-contact mode

In the non-contact mode, the cantilever is vibrated near the surface of the

sample (~ 50-150Å). Since the distance between the atom at the tip and the surface is

greater, the attractive forces are weaker. To overcome this, a stiff cantilever is


67

oscillated near its resonant frequency by means of a piezo oscillator with an amplitude

of a few tens to hundreds of angstroms. The control system then detects the changes

in the frequency, phase or vibration amplitude in response to the force gradients as the

tip comes near the sample surface. The changes are monitored and the feedback

system will maintain the resonant frequency or amplitude thus keeping the average

tip-to-sample distance constant. A combination of this feedback and the scanning

motion is used to generate the topographical image.

Non-contact AFM is preferred for measuring soft samples. However, if a few

layers of condensed water are on the surface of the sample, then the AFM will image

the water layer instead.

3.2.4 Tapping mode

The tapping mode14,15

or intermittent contact mode is similar to the non-contact mode

except that the vibrating cantilever is now closer to the surface of the sample, lightly

touching or tapping the surface. During scanning, the tapping takes place at a

frequency of between 50,000 to 500,000 Hz, corresponding to the resonant frequency

of the cantilever.10

As the oscillating cantilever intermittently contacts the surface, the

oscillation will be reduced due to energy loss when contacting the surface. The

resultant reduction in oscillation amplitude is used to identify and measure the surface

features. This method gives high resolution images without inducing destructive

lateral forces (frictional or drag). Thus, scratching of the soft surface and removal of

loosely bound surface materials can be avoided. Issues with the fluid contamination

layer which affects the oscillating probe will also be eliminated. A schematic of the

AFM in tapping mode is shown in figure 3.9.


68

Figure 3.9 Schematic of an atomic force microscope in tapping mode.

3.3 Experimental

3.3.1 Sample

The sample used was a Ge delta-doped Si sample comprising ten Si0.7Ge0.3

delta-layers of 0.4 nm thickness (nominally), grown by atmospheric pressure chemical

vapour deposition (APCVD) at 700oC. The first layer is at 12 nm and subsequent

nominal depths of the delta-layers are at multiples of 11 nm.16

The surface of the

sample was smooth with an rms roughness ~ 0.15 nm obtained by measuring with an

AFM. The native oxide, estimated at about 1-2 nm based on the age of the sample,

was not removed prior to sputtering.17

The sample was obtained from Z. X. Jiang,

Delft University of Technology, and the dimensions were not verified as there was

insufficient sample for Transmission Electron Microscope analysis.

laser

position sensing device computer

control system oscillating

tip

sample

Piezoelectric scanner

Cantilever with tip

positioning device


69

Silicon substrate was chosen as it is of technological importance and is widely

used in the semiconductor industry. The delta-layers are used as depth markers for the

study of depth resolution. The Ge delta-layer system has also been noted not to induce

any beam-induced roughening in Si.18

3.3.2 Analysis parameters with O2+ SIMS

The O2+ primary ion energies used were 1 keV, 500 eV and 250 eV at incident

angles from 0o (normal incidence) to 70

o at 10

o intervals. The corresponding beam

currents were 140 nA, 130 nA and 48 nA respectively. The beam diameters were

estimated from the ion image of the Cu aperture on the sample holder to be ~ 30 m

for Ep ~ 1 keV and 500eV, and 60 m for Ep ~ 250 eV. The beam was rastered over

an area of 200 x 200 m for Ep ~ 250 eV and Ep ~ 1 keV, and 180 x 180 m for Ep ~

500 eV. Crater edge effects were minimized by using 6% area electronic gating

positioned at the centre of the crater. The secondary ions monitored were 44

SiO+,

30Si

+

and 70

Ge+. The matrix ions

44SiO

+ and

30Si

+ were monitored for the purpose of

evaluating surface transients.19

30

Si+ and

70Ge

+ isotopes were monitored even though

their relative abundance are smaller, 3% for 30

Si and 20% for 70

Ge as the signal

intensity (cps) obtained were sufficiently high but below the saturation level of the

channeltron. These isotopes were also selected as they are free from mass interference

with other elements or molecular ions.


70

3.3.3 Analysis parameters with Cs+ SIMS

The Cs+ primary ion energies used were 1 keV, 500 eV and 320 eV at incident

angles from 0o (normal incidence) to 70

o at 10

o intervals. Ep ~ 320 eV was the

minimum primary ion energy possible on the SIMS instrument used. The

corresponding beam currents used were 77 nA, 40 nA and 40 nA respectively. The

beam diameters were estimated to be 50 m for Ep ~ 1 keV and 70 m for Ep ~ 500

eV and 320 eV. The values were estimated from the ion image of the Cu aperture on

the sample holder.

The beam was rastered over an area of 300 x 300 m for 1 keV and 400 x 400

m for 500 eV and 320 eV. Larger scan areas were selected compared to that used

with O2+ primary ion beam as the beam diameter is larger. A 6% area electronic

gating together with a sufficiently large scan area will minimise the crater edge

effects. The secondary ions monitored were 30

Si-,

59Si

2- and

98SiGe

-.

30Si

- and

59Si2

-

were chosen as matrix signals as these are the most probable states for Si secondary

ions and their signal intensities were not too high as to saturate the channeltron.

98SiGe

- molecular secondary ion was monitored instead of

70Ge

- as it was found to be

more sensitive.

The samples were analysed with the ATOMIKA 4500 SIMS Depth Profiler

described earlier. The profiles obtained were subsequently analysed and the results are

presented and findings discussed in the next few chapters. The surface topography of

several craters was determined using a Digital Instruments (Veeco) Nanoscope

Multimode D3000 series atomic force microscopy (AFM). All AFM images were

collected using the tapping mode.


71

3.3.4 Sputter rate determination

The crater depths at the completion of sputtering were not measured as the

depth of the delta-layers was used for sputter rate determination. The sputter rates

were calculated for each depth segment i.e. from the surface to the first delta-layer

and subsequently from one delta-layer to the next delta-layer. The depths were taken

at each 70

Ge- peak position and the sputter rate was assumed to be constant from one

delta-layer to the next.

The sputter rate for each depth segment is calculated as:

t

nn 1

z nm min

-1

where n is the peak position at the nth delta-layer and t is the time taken to sputter

from the position n-1 to n.

References

1. E. Niehuis and T. Grehl, Secondary Ion Mass Spectrometry SIMS XII, edited by

A. Benninghoven, P. Bertrand, H. N. Migneon and H. W. Werner, (Amsterdam,

Elsevier, 2000), p. 49.

2. P. Steffens, E. Niehuis, T. Friese, D. Greifendorf, A. Benninghoven, Journal of

Vacuum Science & Technology A 3, 1322 (1985).

3. J. M. Chabala, Secondary Ion Mass Spectrometry SIMS X, edited by A.

Benninghoven, B. Hagenhoff, H. W. Werner, (Wiley, Chichester, 1997), p. 23.

4. A. R. Krauss and D. M. Gruen, Applied Physics 14, 89 (1977).

5. J. L. Maul and S. B. Patel, Secondary Ion Mass Spectrometry SIMS XI, edited by

G. Gillen, R. Lareau, J. Bennett, and F. Stevie, (Wiley, Chichester, 1998), p.

707.

6. M. G. Dowsett, N. S. Smith, R. Bridgeland, D. Richards, A. C. Lovejoy and P.

Pedrick, Secondary Ion Mass Spectrometry SIMS X, edited by A. Benninghoven,

B. Hagenhoff, H. W. Werner, (John Wiley, Chichester, 1997), p. 367.


72

7. C. W. Magee, W. L. Harrington, R. E. Honig, Review of Scientific Instruments

49, 477 (1978).

8. J. F. O'Hanlon, A User's Guide to Vacuum Technology, (Wiley-Interscience,

New Jersey, 2003), p. 14.

9. Atomika SIMS 4500 Operation Manual v. 3.8, (1998).

10. E. Meyer, H. J. Hug, R. Bannewitz, Scanning Probe Microscopy: The Lab on a

Tip, (Springer-Verlag, Berlin Heidelberg, 2004), pp. 45-95.

11. R. Howland and L. Benatar, A Practical Guide to Scanning Probe Microscopy,

(Park Scientific Instruments, 1996).

12. S. N. Margonov and M. H. Whangbo, Surface Analysis with STM and AFM,

(VCH, Weheim, 1996).

13. G. Binnig, C. F. Quate, Ch. Gerber, Physical Review Letters 56, 930 (1986).

14. Q. Zhong, D. Inniss, K. Kjoller, V. B. Elings, Surface Science Letters 290, L688

(1993).

15. P. K. Hansma et al., Applied Physics Letters 64, 1738 (1994).



(1999).

18. Z. X. Jiang and P. F. A. Alkemade, Journal of Vacuum Science & Technology B

16, 1971 (1998).



Chapter 4: O2+ SIMS and surface transient

73

Chapter Four

Effect of ultralow-energy O2+ SIMS on Si surface transient

4.1 Introduction

In this chapter, I present my study using ultralow-energy O2+ primary ion

beams. Three primary ion beam energies were selected and depth profiles were

obtained with a range of incident angles. Since the O2+ primary ion beam enhances the

secondary ion yield of electropositive elements, we focus our study only on positive

secondary ion profiles.

The Si+ matrix profiles were evaluated and surface spikes were observed to

occur under certain conditions. Possible reasons for the existence of surface spikes are

discussed. The matrix secondary ions Si+ and SiO

+ monitored were evaluated for

transient width as a function of incident angle. The ztr values obtained from both Si+

and SiO+ were compared to decide whether it is sufficient to determine the end of

surface transient based on a single matrix profile. From the available data on ztr, the

dependence of ztr on and Ep is discussed. The onset of surface roughening is

ascertained by observing the Si+ matrix profile. The condition whereby this

phenomenon occurs is reported.

The sputter rates calculated show that at ultralow-energy, the maximum rates

are achievable at a lower compared to that at higher energies greater than 1 keV.1

Sputter rate variations with depth are also evaluated to determine the accuracy of

using average sputter rates for depth calibrations. Finally, the conditions under which

minimum ztr and accurate depth calibration can be achieved are recommended.


74

4.2 Results and Discussion

4.2.1 Surface Transient

Figure 4.1 shows the depth profiles of 44

SiO+,

30Si

+ and

70Ge

+ using Ep ~ 250

eV at ~ 20o. Ip is the primary ion current profile, which was monitored throughout

the depth profile. Only profiles with constant current were considered in this study.

This is essential to ensure that there are no variations in sputtering yield with depth

due to the primary ion beam current. The Ge delta-layer peaks are used as depth

markers. The 30

Si+ and

44SiO

+ profiles were used in the determination of the surface

transient width. The 70

Ge+ profile shows ten well-resolved peaks. However, some

degradation in the resolution sets in as the profile goes deeper. This phenomenon is

discussed in the next chapter on depth resolution. The dips in the intensity of the 30

Si+

and 44

SiO+

profiles correspond to the positions of the Ge delta-layers. Similar profiles

were obtained by varying Ep and as mentioned in Chapter 3 are shown in appendix

A.

Figure 4.1 Depth profile obtained with Ep ~ 250eV at ~ 20o. The dips in the matrix

profiles and the peaks correspond to the positions of the Ge delta-layers.

70Ge+

30Si+

44SiO+Ip

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)


75

4.2.1.1 Surface spikes and incomplete oxidation

Figure 4.2(a)-4.2(c) show the 30

Si+

profiles at the near surface region for Ep ~

250 eV, Ep ~ 500 eV and Ep ~ 1 keV respectively, at different primary ion incident

angles. The apparent depth was calculated based on the assumption of a constant

erosion rate from the surface to the first delta-layer.2 At Ep ~ 250 eV there is no

surface spike (yield enhancement) at all incident angles, as is normally observed at

higher energies.3 At Ep ~ 500 eV the surface spikes begin to show from ~ 50-60

o

upwards and at Ep ~ 1 keV, the surface spikes appear at ~ 30-40o. This is somewhat

different from that observed by Dowsett et al.4, who reported surface spikes from Ep ~

900 eV onwards at ~ 0o for Si surfaces with or without native oxides. They

suggested that the spikes cannot be entirely due to yield enhancement in native oxide,

otherwise it would prevail throughout all energy ranges. Based on our data, we expect

the surface spike to appear at incident angles closer to normal as the primary ion

energy is increased to > 1 keV.4

Several explanations have been suggested for the presence or absence of

surface spikes.4 The spikes are non-existent at low energies possibly because the

ionization probabilities of Si+ and SiO

+ at these energies are low and independent of

oxygen concentration. The contribution from electronic stopping may not contribute

significantly to ionization at sub-keV O2+ primary ion energies.

4 At higher energies

(keV), around 10% of neutral sputtering is believed to be due to electronic energy

deposition rather than screened nuclear interaction.5


76

Figure 4.2 30

Si+ profiles for (a) Ep ~250 eV (b) Ep ~500eV and (c) Ep ~1keV.

a) 250 eV

1.E+03

1.E+04

1.E+05

0 1 2 3 4 5

Inte

ns

ity

(c

ps

)

0 10 20 3040 50 60 70

b) 500 eV

1.E+04

1.E+05

1.E+06

0 1 2 3 4 5

Inte

ns

ity

(c

ps

)

c) 1 keV

1.E+04

1.E+05

1.E+06

1.E+07

0 2 4 6 8 10

Apparent Depth (nm)

Inte

ns

ity

(c

ps

)


77

To account for the spike followed by a dip before reaching equilibrium at

oblique incidence with Ep ~ 500 eV, the following explanation is proposed. At oblique

angles above ~ 60o, the total energy loss at the near-surface is greater than that at

normal incidence.6 The energy deposited at the immediate surface results in an

increase in the sputtering yield. At the same time, oxygen is being incorporated into

the surface, gradually forming an oxidized altered layer and thus effectively reducing

the sputter rate.7 This effect must be greater than the increase in sputtering yield with

increase in resulting in a lowering of the Si+ and SiO

+ intensities. After the initial

dip in intensity, further oxygen incorporation will gradually enhance the ionization

yield before reaching equilibrium when the fully altered layer is formed.1,8

Hence, this

proposed mechanism accounts for the spike followed by a dip feature at oblique ion

incidence.

For Ep ~ 1 keV, the 30

Si+ profiles at < 40

o show equilibrium signals at higher

intensities than the initial intensities, but at about > 50o the signal intensities at

equilibrium are lower than the initial intensities [Fig. 4.2(c)]. A similar trend was

observed for Ep ~ 500 eV but the lower intensities at equilibrium were observed only

for angles at about > 60o [Fig. 4.2(b)]. For Ep ~ 250 eV no such trend was observed,

i.e. the intensities at equilibrium are higher for all angles of incidence [Fig. 4.2(a)]. It

appears that at ultra low energies such as Ep ~ 250 eV the sputter rate is lower,

allowing sufficient oxygen mixing to form a saturated oxide-altered layer at all

incident angles, thereby increasing the ion yield at equilibrium. However, this is not

the case at higher angles for Ep ~ 500 eV and Ep ~ 1 keV.

For Ep ~ 500 eV and Ep ~ 1 keV, and at ~ 50-60o, there is incomplete

oxidation and hence a lower ion yield. As the primary ion incident angle increases, the


78

maximum oxygen concentration that can be achieved decreases.9,10

There comes a

critical angle c, where incomplete oxidation due to O2+ bombardment is observed.

From Figures 4.2(b) and 4.2(c), it can be ascertained that c ~ 50o at Ep ~ 1 keV and c

~ 60o at Ep ~ 500 eV. The incomplete oxidation could be due to a loss of implanted

oxygen caused by back-scattered incident primary atoms, because the reflection

coefficient increases with an increase in the incident angle.11

This was also confirmed

by transport of ions in matter (TRIM) simulations.12

These observations are consistent with XPS studies done by Tan et al. with 4

keV O2+ bombardment on silicon.

13 Figure 4.3 shows the Si 2p spectra of the surface

at steady state for silicon craters bombarded at various incident angles. At ~ 0o

sputtering, the XPS chemical state at equilibrium Si intensity shows that Si(IV)

dominates corresponding to a complete oxidation and higher 30

Si+ equilibrium profile

than initially. At ~ 45o sputtering, the concentration of Si(I) and Si(III) chemical

states are greater compared to Si(IV) giving rise to a mixed layer of sub-oxides and

SiO2. At ~ 55o, Si(IV) is almost non-existent with Si(I) and Si(III) dominating, thus

forming the sub-oxides on the crater surface at steady state sputtering. This

corresponds to a lower 30

Si+ profile at equilibrium than initially. This lead to the

conclusion that c is at about 40o incident angle. At > 40

o, the formation of sub-

oxide is favoured.

The XPS data suggest that when the O2+ penetration depth is shallow as in Ep

~ 250 eV, Si(IV) dominates and thus a saturated oxide layer is formed corresponding

to a higher 30

Si+ ion yield. Conversely, beyond c, Si(I) and Si(III) dominates forming

sub-oxides. Hence, the reduced intensities are an indication of incomplete oxidation

of silicon.


79

Figure 4.3 The XPS Si 2p spectra of the surface at steady state for 4 keV O2+

bombardment: (a) virgin Si(100) sample; (b) sputtering at 0°; (c) sputtering at 15°;

(d) sputtering at 30°; (e) sputtering at 45°; (f) sputtering at 55°; (g) sputtering at 70°.13

4.2.1.2 Transient Width

Figure 4.4 shows the measured transient widths based on both 44

SiO+ and

30Si

+

profiles as a function of incident angle. The end of the surface transient was taken to

be at 95% of the equilibrium signal for 30

Si+ (or

44SiO

+), consistent with the criterion

used by Jiang and Alkemade.14

A marginal difference in ztr between 30

Si+ and

44SiO

+

is noted throughout all the Ep monitored, except in the range 30o < < 50

o for Ep ~ 1

keV and 40o < < 60

o for Ep ~ 500 eV. With Ep ~ 1 keV and at ~ 40

o,

30Si

+ intensity


80

reaches equilibrium at a depth of 40 nm whereas 44

SiO+ at 70 nm. With Ep ~ 500 eV

and at ~ 50o,

30Si

+ intensity reaches equilibrium at a depth of 9 nm whereas

44SiO

+

at 14 nm. There is a negligible difference observed at Ep ~ 250 eV, although at larger

incident angles ( > 60o) ztr is slightly larger for

44SiO

+. Previous reports

4,14 suggested

that both measured transient widths (Si+ and SiO

+) are not much different, but these

observations were limited to normal incidence at 300 eV < Ep < 2.5 keV,4 and at ~

60o with Ep ~ 1 keV.

14 We note that the convenience of determining ztr using only Si

+

cannot be extended to all incident angles.

We observe that the variation in ztr values using 44

SiO+ and

30Si

+ yields (Fig.

4.4) is greatest at incident angles where the surface transient coincides with the onset

of surface roughening or ripple formation. A sign of the onset of roughening is when

the matrix profile does not reach a steady state but instead rises gradually.15-17

At 1

keV, this occurs at ~ 40o, and at 500 eV, it occurs at ~ 50

o. It is known that at

oblique angles ripples are formed. The critical angle for ripple formation is the same

as, or marginally larger than the critical angle for complete bombardment-induced

oxidation.18

Incomplete oxidation lead to the formation of sub-oxides13

and

subsequently a heterogeneous oxide layer which is known to initiate surface

roughening.17

SiO+ is a sensitive indicator of oxygen concentration at the sputtered

surface.19

Therefore in the region of c, oxidation of the silicon is incomplete; thus it

takes a longer time (hence, deeper) for the SiO+ signal to stabilize.


81

Figure 4.4 Measured transient width based on both 44

SiO+ and

30Si

+.

Figure 4.4 also shows that the transient width is lowest at near-normal

incidence and increases to near the critical angle where the surface transient and the

onset of roughening overlap; thereafter, it decreases to ztr values that are larger than at

normal incidence. It is also interesting to note that this overlap between the surface

transient and onset of roughening occurs at lower for higher Ep, i.e. ~ 40o for Ep ~

1 keV and ~ 50o for Ep ~ 500 eV. It appears that at higher primary ion energy, the

sputtering process is more dominant than the ion incorporation process at a lower

incident angle than at lower primary ion energy, resulting in incomplete oxidation and

the onset of roughening.

We obtained very low ztr values of < 1 nm at Ep ~ 250 eV over the range of

angles 0o < < 50

o. The lowest ztr achieved with Ep ~ 250 eV was at ~ 0-20

o

incidence with ztr ~ 0.7 nm. For Ep ~ 500 eV ztr is also ~ 0.7 nm for ~ 0-10o. For Ep

Si+ (250 eV)

SiO+ (250 eV)

Si+ (500 eV)

SiO+ (500 eV)

Si+

(1 keV)

SiO+ (1 keV)

Si+(560 eV)

ref.12

Si+ (1 keV) ref 12

0

2

4

6

8

10

12

14

16

0 10 20 30 40 50 60 70 80 90

Angle of Incidence

Tra

ns

ien

t W

idth

(n

m)


82

~ 1 keV, the ztr is about 1.5 nm for ~ 0-10o. At low Ep the sputter rate is lower as

there are fewer „knock-on‟ collisions. The lower sputtering yield allows a higher

concentration of oxygen to build up in the sample surface, forming SiO2 over a wider

range of Further deposition of oxygen atoms results in a rapid outward growth of

ion-beam-synthesized SiO2 similar to that observed at normal incidence.20,21

Hence,

we conclude that the lowest ztr values are achievable over a larger range of incident

angles at lower impact energy, and not only at normal incidence as previously

reported.4,22

It is widely reported that ztr decreases with impact energy, but we observe that

at ultra low energies between Ep ~ 250 eV and 500 eV and ~ 0-10o ztr does not differ

significantly. Although the sputtering yield difference is quite significant (see Fig.

4.5), this does not appear to affect the oxygen concentration on the surface. At such

low impact energies, the primary ion penetration is shallow, in the range of 2-3 nm.

This is in the range of the native oxide and thus explains the levelling-off in the ztr

values even though the Ep is reducing.

4.2.2 Sputter Rates

Figure 4.5 shows the sputter rates at various angles of incidence and primary

ion energies. Two plots are shown at each primary energy: one (labelled d1)

represents the average sputter rate at the near surface, i.e. the average sputter rate from

the surface to the first Ge delta-layer, (i.e. top 12 nm); and the other is the average

sputter rate observed from the surface to the tenth Ge delta-layer. The sputter rate at

the near surface is known to vary because of the surface transients. Therefore, the


83

comparison with the average sputter rate will be useful as the average sputter rate is

commonly used in depth profiling.

As expected, the sputter rate increases with incident angle, reaching a

maximum at ~ 50o

for Ep ~ 250 eV. The lowest sputter rate is about 20 nm min-1

nA-1

cm2 and occurs at normal incidence for Ep ~ 250 eV. The maximum sputter rate

for both Ep ~ 500 eV and 1 keV occurs at ~ 60o, which is consistent with that

observed at higher energies.1

For all energies studied, the sputter rate is lowest at normal incidence. The

bombardment of silicon with oxygen ions at normal incidence and near-normal

incidence generates an altered layer of SiO2. Silicon has a strong affinity with oxygen

Figure 4.5 Sputter rates vs. angle of incidence. d1 shows the average sputter rate from

the surface to the first Ge delta-layer and the other is the average sputter rate observed

from the surface to the tenth Ge delta-layer.

0

1

2

3

0 20 40 60 80

250 eV d 1

250 eV

d 1

500 eVd 1

1 keV

0

5

10

15

20

25

30

35

40

0 20 40 60 80

Angle of Incidence

Sp

utt

er

Ra

te (

nm

/min

)


84

to form SiO2, which sputters more slowly than Si. The oxidation of silicon also

reduces the sputtering yield.6,23

As the incident angle increases, the effective range of

implanted oxygen decreases12

and sputtering yield increases,24

resulting in higher

sputter rates. At > 60o there is little or no effect of implanted oxygen on sputtering

yield.24

Figure 4.5 also shows that the near-surface sputter rate is higher than the

average sputter rates for all angles and energies except for > 50o with Ep ~ 1 keV,

where the near-surface sputter rate is lower than the average sputter rate. This

anomaly can be explained by the fact that beyond > 50o the matrix profiles do not

reach a steady state even at the tenth Ge delta-layer, implying that the onset of

roughening did not reach a steady state throughout the depth range measured. Hence,

the sputter rate remains high throughout and leads to a higher average sputter rate

comparable to that at the near-surface. For all energies up to ~ 40o the difference in

the d1 sputter rate to that of the average sputter rate is quite constant (< 1 nm min-1

);

beyond ~ 40o the difference widens in the case of Ep ~ 250 eV and ~ 500 eV.

Figure 4.6(a)-4.6(c) show plots of relative sputter rates as a function of depth,

normalized to the sputter rate at the ninth to tenth Ge delta-layer. At Ep ~ 250 eV and

0o < < 40

o, sputtering is stable as a function of depth except at the surface. The

difference in sputter rate at the near-surface to the average sputter rate is about 7%.

For Ep ~ 500 eV and 0o < < 40

o the difference is 10%. At Ep ~ 1 keV and 0

o < <

20o the difference is about 11% and for ~ 30

o the difference is about 16%.


85

Figure 4.6 Normalized sputter rates throughout the depth.

a) 250 eV

0.6

0.8

1.0

1.2

1.4

1.6R

ela

tiv

e S

pu

tte

r R

ate

0 10 20 30

40 50 60 70

b) 500 eV

0.6

0.8

1.0

1.2

1.4

1.6

Re

lati

ve

Sp

utt

er

Ra

te

c) 1 keV

0.6

0.8

1.0

1.2

1.4

1.6

0 20 40 60 80 100 120

Depth (nm)

Re

lati

ve

Sp

utt

er

Ra

te


86

We deduce that for Ep ~ 500 eV and Ep ~ 250 eV we may use the average

sputter rate for a reasonably accurate depth conversion at the near surface for 0o < <

40o because the error will be in the range of 5 – 8% for Ep ~ 250 eV and 10% for Ep ~

500 eV. For Ep ~ 1 keV the error at 0o < < 20

o is about 11%. This finding expands

on the observation made by Wittmaack.2 He reported that the difference in erosion

rate at the near surface to the average erosion rate is negligible for Ep ~ 500 eV at

normal incidence. The results of surface transient width, sputter rate and the onset of

roughening observed under ultralow-energy O2+ bombardment on silicon surface are

summarized in table 4.1.


87

Table 4.1 Summary of surface transient width, sputter rate and the onset of

roughening observed for O2+ sputtering on silicon wafer.

250eV 500eV 1keV

Surface Spike Nil > 50-60o >30-40

o

Equilibrium signals

lower than initial Nil > 60

o > 50

o

c (incomplete

oxidation) Nil > 60o 50

o

Different ztr for SiO+

and Si+ > 60

o 40

o < 50

o 30

o < 50

o

Onset of roughening Not obvious ~ 50o ~ 40

o

Range of giving

lowest ztr ~ 0-20o ~ 0–10

o ~ 0–10

o

Lowest ztr 0.7 nm 0.7 nm 1.5 nm

Sputter rate max. at ~ 50o ~ 60

o ~ 60

o

Sputter rate min. at ~ 0o ~ 0

o ~ 0

o

Stable erosion rate at 0o < 40

o 0

o < 40

o 0

o < 20

o

Difference in sputter rate

between near surface

and throughout depth for

with stable erosion

rate


88

4.3 Summary

In this work, we confirm that it is sufficient to use either 30

Si+ or

44SiO

+

profiles to observe the surface transient width, but only in regions where the transient

width and the onset of roughening can be discriminated. We also observe that the

lowest transient width is obtainable at normal and near normal incidence. Although

the transient width decreases with the primary ion energy, there is no significant

improvement going from Ep ~ 500 eV to ~ 250 eV. However, we have a wider range

of incident angles at lower energies where ztr is minimum, i.e. 0o < < 10

o for Ep ~

500 eV and 0o < < 20

o for Ep ~ 250 eV.

The onset of roughening occurs at a smaller at higher energy ( ~ 40o for Ep

~ 1 keV, ~ 50o for Ep ~ 500 eV) and does not seem to appear at all at Ep ~ 250 eV

over the whole incident angle range studied. Higher matrix signals were observed

upon equilibrium at Ep ~ 250 eV for all angles. This is true only for 0o < < 50

o at

500 eV, and for 0o < < 40

o at 1 keV.

For all energies, the sputter rate is lowest at normal incidence, gradually

increasing to a maximum at about 50o < < 60

o. This is similar to that reported at

higher energies4. The sputter rate at the near surface does not differ much from the

sputter rates at other depths for Ep ~ 250 eV. However, this is only true at 0o < < 40

o

for Ep ~ 500 eV and 0o < < 20

o for Ep ~ 1 keV. Within these incident angles, the

average sputter rate can be used reliably for depth conversion.

We conclude that the best working range to achieve a narrow transient width

and accurate depth calibration are at Ep ~ 250 eV, 0o < < 20

o and 500 eV, 0

o < <

10o where the ztr achievable is ~ 0.7 nm.


89

References




3. W. Vandervorst and F. R. Shepherd, Applied Surface Science 21, 230 (1985).

4. M. G. Dowsett, T. J. Ormsby, G. A. Cooke, D. P. Chu, Journal of Vacuum


5. J. P. Biersack, Nuclear Instruments and Methods in Physics Research B 27, 21.

(1987).

6. T. Ishitani and R. Shimizu, Applied Physics 6, 241 (1975).

7. K. Wittmaack and S. F. Corcoran, Journal of Vacuum Science & Technology B

16, 272 (1998).

8. W. Reuter and K. Wittmaack, Applied Surface Science 5, 221 (1980).

9. J. L. Alay, Surface and Interface Analysis 19, 313 (2006).


11. W. Eckstein and J. P. Biersack, Z. Phys. B 63, 109 (1986).

12. J. F. Ziegler and J. P.Biersack, SRIM 2003, http://www.srim.org (2003).

13. S. K. Tan, K. L. Yeo, A. T. S. Wee, Surface and Interface Analysis 36, 640

(2004).


(1999).




17. C. M. Ng, A. T. S. Wee, C. H. A. Huan, A. See, Journal of Vacuum Science &

Technology B 19, 829 (2001).



20. W. Vandervorst, F. R. Shepherd, J. Newman, B. F. Philips, J. Remmerie,



90


22. D. P. Chu, M. G. Dowsett, T. J. Ormsby, G. A. Cooke, Proceedings of

International Conference on Characterization & Metrology for ULSI

Technology (1998).

23. A. E. Morgan, H. A. M. de Grefte, N. Warmolts, H. W. Werner, H. J. Tolle,

Applied Surface Science 7, 372 (1981).


Chapter 5: O2+ SIMS and depth resolution

91

Chapter Five

Effect of ultralow-energy O2+ SIMS on depth resolution

5.1 Introduction

In this chapter, we focus on the influence of ultralow-energy O2+ primary ion

bombardment on depth resolution. The same set of profiles obtained for the study on

surface transients were used but only the 70

Ge+ profiles were analysed. The profiles

from delta-layers are ideal for the study of depth resolution as the slopes and the peak-

broadening are indicative of the deviation from the ideal profile. The depth

resolutions at various Ep and combinations with depth were evaluated by calculating

the FWHM and d of the 70

Ge+ peaks which represent the SIMS depth profiles of the

Si0.7Ge0.3 delta-layers. We confirm that depth resolution can be improved by lowering

the primary ion impact energy at ultralow-energy. By varying the angle from 0o to

70o, we note that a better depth resolution is achievable not only at normal incidence

but over a wider range of incident angles as the probe energy is reduced. The dynamic

range for the detected secondary ions is also evaluated. Contributions from

roughening and atomic mixing to the depth resolution of delta-layers are discussed

using the mixing-roughness-information depth model.


92

Figure 5.1 Profiles of 70

Ge+ obtained with different Ep (normalized to the first peak of

Ep ~ 250 eV) at a) ~ 0o and b) ~ 40

o.

1keV

500eV

a)0o

250eV

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04In

ten

sit

y (

cp

s)

1keV

500eV

250eV

b) 40o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 20 40 60 80 100 120 140

Depth (nm)

Inte

ns

ity

(c

ps

)


93

5.2 Results and discussion

5.2.1 Depth resolution .

Figure 5.1(a) and 5.1(b) show the profiles of 70

Ge+ normalized to the first peak

of the Ep ~ 250 eV profile, at ~ 0oand 40

o respectively using Ep ~ 1 keV, 500 eV

and 250 eV. We observe that at normal incidence, the depth resolution obtained when

using Ep ~ 500 eV is comparable to that at Ep ~ 250 eV. Clearly, the depth resolution

using Ep < 500 eV is better than that at Ep ~ 1 keV. At normal incidence, the peak-to-

valley ratio (PVR) obtained when using Ep ~ 1 keV is smaller compared to that at Ep ~

250 eV and Ep ~ 500 eV at all depths, while the PVR observed at Ep ~ 500 eV and Ep

~ 250 eV are similar. At ~ 40o, the deterioration in depth resolution with depth and

energy is obvious as Ep is increased from 250 eV to 1 keV, i.e. the 70

Ge+ peak

broadens. The PVR also decreases significantly from Ep ~ 250 eV to Ep ~ 1 keV.

Comparing the 70

Ge+ profiles obtained using Ep ~ 1 keV at ~ 0

o and ~ 40

o,

it is clear that there is a broadening of the peaks, i.e. deterioration of depth resolution

due to the onset of roughening1 as is increased. At ~ 40

o, the depth resolution

deteriorates more severely than at ~ 0o as Ep is increased. The deterioration in depth

resolution observed as Ep is increased for a particular is the result of atomic mixing

due to increased ion penetration depth.2 It is more severe at ~ 40

o due to a

combination of both factors of roughening and atomic mixing.

5.2.1.1 Depth resolution in terms of FWHM

Figure 5.2(a)-5.2(c) show the depth resolution measured in terms of FWHM of

the 70

Ge+ peaks against the depth. To minimise errors in the determination of the

resolution parameters, large scan sizes were used to maximise the number of data


94

points. Another source of error is the choice of baselines for the peaks in the

determination of FWHM. The errors were minimised by consistently using the same

method in determining the base of the peaks. The base of the peak was determined by

the valley-to-valley method i.e. by drawing a line from the valley before the peak to

the valley after the peak. The peak height was measured based on the vertical height

from the peak to the baseline.

At Ep ~ 250 eV, the depth resolution is constant with a mean FWHM of 1.5 nm

throughout the analysis depth of 115 nm, when ~ 0-40o. The difference in depth

resolution from the first delta-layer to the last delta-layer is less than 15%. At ~ 50o,

the depth resolution is constant at FWHM of 1.9 nm up to about 50 nm before

deteriorating further to 2.3 nm or by about 20%. At ~ 60o

and 70o, the FWHMs at

the first delta-layer are 2.4 nm and 3.3 nm respectively, but worsen with depth. The

FWHM increase from the first to the last delta-layer is 29% at ~ 60o and 55% at ~

70o. At all investigated, the delta-layers are well resolved. Given the above

observations, a good depth resolution in the order of 1-2 nm can be obtained using ~

0-40o with Ep ~ 250 eV, throughout the depth. Depth resolution degrades at > 50

o

possibly coinciding with the onset of roughening but is not obvious as discussed

earlier in Chapter 4 (4.2.1.2).1

Using Ep ~ 500 eV, the depth resolution is again constant throughout the depth

studied for ~ 0-40o with a mean FWHM of 2.2 nm. At ~ 50

o, the depth resolution

degrades by about one and a half times, and worsens with depth after the fourth delta-

layer i.e. about 50 nm deep. A similar observation was reported by Liu et al.3 using

~ 56o where the onset of roughening was observed from a depth of 60 nm, and


95

Figure 5.2 Depth resolution of Ge delta-layer as a function of profile depth, measured

by FWHM for a) Ep ~ 250 eV b) Ep ~ 500 eV c) Ep ~ 1 keV at various incident angles.

a) 250eV

0o-40

o50

o

60o

70o

0

1

2

3

4

5

6

7

8

FW

HM

(n

m)

0 10 20 30

40 50 60 70

b) 500eV

0o-40

o

50o

60o

70o

0

1

2

3

4

5

6

7

FW

HM

(n

m)

c) 1 KeV

0o-30

o

40o

50o

60o

70o

0

1

2

3

4

5

6

7

0 20 40 60 80 100 120 140Depth (nm)

FW

HM

(n

m)


96

consistent with our earlier finding discussed in Chapter 4 (4.2.1.2) that the onset of

roughening occurs at ~ 50o.1 Such degradation in depth resolution was also observed

by Ormsby et al.2 at 500 eV and ~ 50-60

o. At ~ 60-70

o the FWHM at the first

delta-layer (d1) is about 2.4 nm, but deteriorates significantly beyond this depth.

Looking at the depth profiles of ~ 60-70o (Appendix A, page 166), it can be deduced

that the degradation in resolution with depth was a result of erosion

inhomogeneity which has been attributed to a variation in sputter rate due to

rastering4 and/or poor beam focus (instrument related factors) and not due to

sputtering induced surface roughness. The evidence of this observation is discussed

using Figures 5.3.

Using Ep ~ 1 keV, the depth resolution is constant at 3.5 nm as a function of

depth at ~ 0-20o only. Unlike at lower energies, the depth resolution at Ep ~ 1 keV is

better at ~ 60-70o to a depth of about 50 nm, but degrades thereafter. Good depth

resolution has previously been observed at Ep ~ 1 keV and ~ 71o.5 The poorer depth

resolution at ~ 40-50o could be due to the onset of roughening which was observed

as described earlier in Chapter 4 (4.2.1.2)1 at ~ 40

o. Wittmaack

6 also reported rapid

ripple growth at Ep ~ 1 keV and ~ 38-62o. Jiang and Alkemade

7 observed the lowest

surface roughness using ~ 55-70o at Ep ~ 1 keV in the range of ~ 48-79

o evaluated.

This is consistent with our observation of better resolution at ~ 60-70o compared to

at ~ 40-50o. Comparing across Ep, we observe that a better depth resolution is

generally achieved at lower energy. Using Ep ~ 250 eV or 500 eV, the depth

resolution is good at normal to 40o incidence. At Ep of 1 keV, the depth resolution is

better towards glancing angle, i.e. ~ 60o onwards, but only down to a critical depth


97

of about 50 nm; nevertheless constant depth resolution with depth at 1 keV is

achievable near normal incidence at ~ 0-20o.

Figure 5.3(a)-5.3(c) show the ion intensity images of 70

Ge+ at the depth where

the cursor positions are, respectively. The yellow squares are indicative of a high

concentration of 70

Ge+, while the black squares indicate the absence of

70Ge

+. Figure

5.3(b) shows that 70

Ge+ was detected earlier on the right side of the crater confirming

that the crater base was not flat and parallel to the initial surface. At a shallower depth

as depicted in Figure 5.3(a), the 70

Ge+ ion image shows an even breakthrough as the

crater was sputtered down the depth. Similarly, Figure 5.3(c) shows an even 70

Ge+ ion

image at about the same depth as that in Figure 5.3(b). Looking at the depth profile, it

is obvious that the depth resolution is poor but this was not caused by an inclined

crater base (instrument related factors) as in the profile of Ep ~ 250 eV/ ~ 70o but

rather by sputtering induced surface roughening (microtopography) as discussed

earlier.

Figure 5.4(a)-5.4(c) show the depth resolution (in terms of FWHM) of three

representative Ge delta-layers at the near surface d1 (12.2 nm), intermediate depth d5

(57.8 nm) and deepest at d10 (114.8 nm) for all three energies investigated. Figures

5.4(a)-5.4(c) show clearly that using Ep ~ 250 eV, a good resolution prevails at ~ 0-

40o at all depths, before deteriorating at grazing angles. At Ep ~ 500 eV, good

resolution is maintained similarly at ~ 0-40o. At 1 keV, the best depth resolution

obtained is at ~ 60-70o, but only nearer the surface. The depth resolution is stable

with depth only at ~ 0-20o.


98

Figure 5.3 Ion image of 70

Ge+ taken at the depth where the cursor position is. a) Ep ~

250 eV, ~ 70o at the 2nd delta-layer b) Ep ~ 250 eV, ~ 70

o at the 4

th delta-layer c)

Ep ~ 1 keV, ~ 40o at the 4

th delta-layer.


99

Figure 5.4 Depth resolution of Ge delta-layer at various primary ion incident angles.

a) 250 eV

d 1

d 5

d 10

0

1

2

3

4

5

6

FW

HM

(n

m)

d1 d5 d10

b) 500eV

d 1

d 5

d 10

0

1

2

3

4

5

6

7

8

FW

HM

(n

m)

c) 1keV

d 1

d 5

d 10

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70

Angle of Incidence

FW

HM

(n

m)


100

5.2.1.2 Depth resolution in terms of exponential decay

Figure 5.5 shows the overlay of profiles of the first Ge delta-layer obtained at

various and Ep normalized to the peak obtained at normal incidence. The trailing

edges at all three values of Ep are always gentler than the leading edges due to ion

beam mixing effects. As Ep increases, the leading edge slope becomes gentler. At

higher primary ion energy, the ion projected range is longer and hence the altered

layer is thicker, resulting in a gentler slope.8 Similarly, d increases as Ep increases for

the trailing edge due to an increased knock-on effect of Ge atoms. It can be clearly

seen that using Ep ~ 250 eV, that the depth resolution is best at ~ 40o [Fig. 5.5(a)];

this is not obvious in the FWHM results which only give resolution information at

mid-height of the peak. At Ep ~ 500 eV, the depth resolution is best at ~ 0-30o [Fig

5.5(b)]; and at Ep ~ 1 keV, it is best at ~ 0-20o.

Figure 5.6 shows the decay length for selected -layers using Ep ~ 250 eV, 500

eV and 1 keV against 0-70o. The figure shows that d decreases as Ep is lowered.

At lower Ep, d is small over a greater range of . For Ep ~ 250 eV, d is lowest at ~

0-50o; for Ep ~ 500 eV, it is at ~ 0-30

o and for Ep ~ 1 keV, it is at ~ 0-20

o. Using

Ep ~ 250 eV, d decreases from 0.9 nm to 0.6 nm as increases from 0o to 50

o. Even

though d at ~ 50o is the smallest, the FWHM is higher, i.e. the peak is broader,

resulting in a poorer depth resolution than at ~ 0-40o. This is because d only gives

information about the trailing edge, and not the full peak. Using Ep ~ 500 eV and ~

40o, d is larger even though the FWHM is comparable to ~ 0-30

o, thus giving an

overall depth resolution which is poorer than ~ 0-30o. Using Ep ~ 1 keV, profiling at

~ 60-70oresults in lower d than at near normal. However the depth resolution is


101

Figure 5.5 Comparison of the normalized 70

Ge+ profiles of the first delta-layer at

various Ep.

40o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 10 20Depth (nm)

Inte

nsit

y (

cp

s)

40

50

a) 250 eV

40o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

Inte

ns

ity

(c

ps

)

0

10

20

30

40

b) 500 eV

40o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

Inte

ns

ity

(c

ps

)

c) 1 keV

30o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 5 10 15 20

Depth (nm)

Inte

nsit

y (

cp

s)


102

Figure 5.6 Plot of decay length (d) versus incident angle ().

a) 250eV

d 1

d 5

d 10

0

1

2

3D

ec

ay

Le

ng

th (

nm

/e) d1 d5 d10

b) 500eV

d 1

d 5d 10

0

2

4

6

8

De

ca

y L

en

gth

(n

m/e

)

c) 1 keV

d 1

d 5

d 10

0

1

2

3

4

5

0 10 20 30 40 50 60 70

Angle of Incidence

Decay len

gth

(n

m/e

)


103

better than near normal only to a depth of 50 nm, as the FWHM beyond the 50 nm is

larger, thus giving a poorer depth resolution beyond that depth.

5.2.1.3 MRI model and evaluation

Lau et al.9 have successfully demonstrated with a similar SiGe -doped

sample that the surface roughening behaviour during ion sputtering can be accounted

for by the Mixing-Roughness-Information depth (MRI) model proposed by

Hoffman.10

The model considers the contributions of three fundamental parameters

namely, atomic mixing, roughness and information depth towards depth resolution.

Atomic mixing is described by an exponential function, with a characteristic mixing

length w, as shown in equation 5.1. Roughness is represented by a Gaussian term with

standard deviation which corresponds to the root mean square surface roughness, as

in equation 5.2; and information depth is represented by an exponential term with

characteristic information depth as in equation 5.3. z is the sputtered depth and zo is

the running depth parameter for which the composition is defined. For example, each

monoatomic layer at a location zo gives a normalised contribution at a sputtered depth

z that is described by11

Atomic mixing: ]/)(exp[1

wwzzw

g ow , (5.1)

Roughness: ]2

)(exp[

2

12

2

ozzg

, (5.2)

Information depth: ]/)(exp[1

ozzg . (5.3)


104

In the atomic mixing parameter, mixing is assumed to be instantaneous,12

complete, and it extends to a depth w. Therefore, roughness and atomic mixing are

assumed to be independent of each other. The information depth parameter can be

neglected as it is about one to two monolayers based on the secondary-ion escape

depth in low energy SIMS.10

The roughness parameter consists of three components:

the original interface roughness (i), sputtering induced surface roughness (s) and

straggling of the mixing length (w).10

The total parameter in the model is

21

222

iws . (5.4)

Interface roughness can be neglected as the sample is assumed to have high quality

interfaces. The contribution from mixing length straggling13

can be significant

compared to the contribution of the original interface roughness and sputtering

induced surface roughness but is difficult to determine. The rms value of s can be

obtained from AFM measurements.

Figure 5.7 shows the projected ion range of O2+ primary ions at ultralow-

energies from ~ 0-70o using TRIM calculations. In Figure 5.1, we observe that as Ep

is increased, the depth resolution becomes poorer. At ~ 0o and with increasing Ep,

the primary ion range and hence the atomic mixing length also increases (see Fig.

5.7). Hence, the dominant contribution to depth resolution is atomic mixing. When

peak broadening occurs as is increased from 0o to 40

o for Ep ~ 500 eV and 1 keV,

the major contribution to the peak shape is from roughening since the effective ion

range is decreasing (by a cos factor). Another observation is the increase in peak

broadening as the sputtered depth increases, and this must be due to sputter-induced


105

Figure 5.7 The projected ion range of O2+ primary ion beam at various incident

angles calculated from TRIM8

roughening. We can conclude that as the peak broadens with depth, the overall crater

morphology becomes progressively rougher.10

At Ep ~ 250 eV, the FWHM and d values are constant for ~ 0-40o and the

profiles are Gaussian, indicating that there is no significant onset of roughening.

However, at ~ 50o, the onset of roughening is apparent as the FWHM values are

distinctively higher than that at ~ 0-40o, since the contribution from atomic mixing

does not change very much. The observation is similar for ~ 60o and 70

o.

From Fig. 5.5 we infer that based on MRI modelling, the onset of roughening

at ~ 50o for Ep ~ 250 eV, ~ 40

o for Ep ~ 500 eV and ~ 30

o for Ep ~ 1 keV is

consistent with Chapter 4 where the onset of roughening was determined by the less

250 eV500 eV

1 keV

0

1

2

3

4

5

0 10 20 30 40 50 60 70 80 90Angle of Incidence

Ion

Ra

ng

e (

nm

)

250eV 500eV 1keV


106

sensitive indicator of using the increase in 30

Si+ profile which was not obvious for Ep

~ 250 eV, ~ 50o for Ep ~ 500 eV and ~ 40

o for Ep ~ 1 keV.

Figure 5.8 shows the profile resolution data (FWHM and d) from the first

delta-layer as a function of beam energy when bombarded at normal incident. The

data demonstrates effectively the improvement in depth resolution by the use of lower

Ep. The y-intercept can be interpreted as extrapolation to „zero-energy‟. The value of

the intercept for FWHM is 0.8 nm. This can be interpreted to be the sum of the

nominal width of the delta-layer and the information depth since the contribution from

atomic mixing and roughness is absent. The information depth for SIMS which is the

secondary ion escape depth, is about two to three atomic layers (0.3 – 0.4 nm). Hence,

the width of the delta-layer is about 0.4 nm as specified. The value of the y-intercept

for d is 0.6 nm. At „zero-energy‟ there is no atomic mixing, hence this value is the

limit which is due to the initial surface topography and also the information depth of

the technique.

Figure 5.8 Depth resolution parameters (FWHM and d) dependence on energy for


y = 0.0029x + 0.845

y = 0.0013x + 0.6

0

1

2

3

4

0 250 500 750 1000 1250

Re

so

luti

on

Pa

ram

ete

rs (

nm

)

Primary Ion Energy (eV)

FWHM

d


107

5.2.2 Dynamic range

Figure 5.9 shows the dynamic range averaged over the first nine peaks. It

gives an indication of the range of concentrations that can be detected by SIMS. The

dynamic range was calculated based on the ratio of peak intensity to the background

intensity. The lower of the intensities before the first delta-layer or after the last-delta

layer was taken to be the background intensity. At higher Ep and correspondingly

higher beam current, we observe a better dynamic range. For all three primary ion

energies studied, the dynamic range increases as moves away from normal, reaching

a maximum before decreasing. For Ep ~ 1 keV, the maximum is at ~ 30o; for Ep ~

500 eV, the maximum is at ~ 30-40o; and for Ep ~ 250 eV, the maximum is at ~

40-50o. The results of high depth resolution and best dynamic range are summarized

in Table 5.2.

Figure 5.9 Dynamic range averaged over the first nine Ge peaks vs incident angle.

250eV500eV

1keV

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 10 20 30 40 50 60 70

Angle of Incidence

Dy

na

mic

Ra

ng

e (

Ima

x/I

min

)

250eV 500eV 1keV


108

Table 5.1 Summary of depth resolution and dynamic range observed with O2+

sputtering.

5.3 Summary

At Ep ~ 250 eV, a good depth resolution of about 1.5 nm FWHM and d of

less than 1 nm is obtained at ~ 0-40o throughout the depth range studied. The best

depth resolution is seen at ~ 40o. At Ep ~ 500 eV, we observe a good depth

resolution of about 2.2 nm FWHM and d averaged at 1.2 nm throughout the depth

evaluated at ~ 0-30o. At Ep ~ 1keV, we obtained a good depth resolution of 3.5nm

FWHM and d of 1.8 nm at ~ 0-20o throughout the depth evaluated. The best

dynamic range is achieved at ~ 40-50o for Ep ~ 250 eV, ~ 30-40

o for Ep ~ 500 V,

250eV 500eV 1keV

Lowest FWHM 1.5nm 2.2nm 3.5nm

with lowest FWHM

throughout depth ~ 0 - 40

o ~ 0 – 30

o ~ 0 – 20

o

with lowest FWHM

to a limited depth (nm) Nil

~ 60 – 70o

(12 nm)

~ 60 – 70o

(50 nm)

Lowest d < 1nm ~ 1.2nm ~ 1.8nm

with lowest d

throughout depth ~ 0 -50o

~ 0 - 30o

~ 60 -70o

~ 0 - 20o

~ 60 -70o

with best dynamic

range ~ 50o ~ 40

o ~ 30

o

Highest dynamic

range 6.8 x 103

8.9 x 103 2.3 x 10

4


109

and ~ 30o for Ep ~ 1 keV. We conclude that the best condition for high depth

resolution with a good dynamic range is at Ep ~ 250 eV and ~ 40o.

From our analysis using the MRI model, we confirmed that depth resolution

can be improved by lowering the primary ion impact energy which effectively reduces

atomic mixing. We also note that at ultralow-energy, a better depth resolution is

achievable not only at normal incidence but also over a wider range of incident angle

as Ep is decreased. Over this range, there is complete oxidation resulting in no onset

of roughening.14

As is increased, surface roughening sets in, resulting in a poorer

depth resolution.

From the 70

Ge+ profiles, we can distinguish between contributions from

roughening and atomic mixing to the depth resolution of delta-layers. The Gaussian

broadening is due mainly to the roughening component, and the trailing edge

broadening, i.e. increases in d, is due primarily to the atomic mixing component.

Based on MRI modelling, the onset of roughening is detected earlier as compared to

observing an increase in the 30

Si+ profile

1 suggesting that the MRI model is a more

sensitive indicator of the onset of roughening.


110

References

1. A. R. Chanbasha and A. T. S. Wee, Surface and Interface Analysis 37, 628

(2005).




4. M. Meuris, P. Debisschop, J. F. Leclair, W. Vandervorst, Surface and Interface

Analysis 14, 739 (1989).

5. Z. X. Jiang, P. F. A. Alkemade, E. Algra, S. Radelaar, Surface and Interface

Analysis 25, 285 (1997).




9. G. S. Lau, E. S. Tok, R. Liu, A. T. S. Wee, J. Zhang, Nuclear Instruments &

Methods in Physics Research Section B 215, 76 (2004).

10. S. Hofmann, Surface and Interface Analysis 27, 825 (1999).

11. S. Hofmann, Surface and Interface Analysis 30, 228 (2000).

12. Z. L. Liau, B. Y. Tsaur, J. W. Mayer, Journal of Vacuum Science & Technology

B 16, 121 (1979).

13. A. Rar, D. W. Kojima, D. W. Moon, S. Hofmann, Thin Solid Films 355, 390

(1999).


Chapter 6: Cs+ SIMS and surface transient

111

Chapter Six

Effect of ultralow-energy Cs+ SIMS on Si surface transient

6.1 Introduction

Another widely used primary ion beam in SIMS is Cs+. It complements O2

+

SIMS as it is sensitive to electronegative elements. The presence of Cs+ enhances the

negative secondary ion yield. In this chapter, we will study the dependence of

transient width and sputter rate as a function of Cs+ primary ion energy (Ep ~ 320 eV,

500 eV, 1 keV) and incident angles between 0-70o.

The matrix profiles (30

Si-) were monitored and evaluated for transient width

determination. An extended transient width was observed and the width was

compared to the penetration depth of the incident Cs+ ion. The influence of Cs on the

intensity and shape of the profile is also discussed.

The sputter rates were calculated and showed similar behaviour as that

observed with ultralow-energy O2+ SIMS. It is known that the transient width is

reduced when the primary ion energy is reduced. However, we find that at ultralow-

energies, reducing Ep does not have a significant effect in reducing the transient

width. From the equilibrium intensity of the 30

Si- profile, we draw conclusions on the

detection sensitivity with respect to the incident angle used.


112


6.2.1 Surface transients

Figure 6.1 shows the depth profiles of 30

Si-,

59Si2

- and

98SiGe

- using Ep ~ 320

eV at ~ 50o. Ip is the primary ion current profile, which was monitored throughout

the depth profile. Only profiles with constant current, monitored as a horizontal line

(Ip), were used in our study. The peaks (maximum intensity) are taken to be located at

the centre of the Ge delta-layers and are used as depth markers. However, we note

that the peak shape is asymmetrical depending on the primary beam energy and angle

of incidence. Hence, there may be a slight deviation of the peak from the actual centre

of the Ge delta-layer. The 30

Si- profiles were used in the determination of the surface

transient width. The 98

SiGe- profile shows ten well-resolved peaks and is used to

determine the depth resolution. The depth resolution is observed to deteriorate

progressively with depth. The dips in the intensities of the 30

Si- and

59Si2

- profiles

correspond to the positions of the Ge delta-layer. Similar profiles were obtained for all

three energies and angles of incidence specified earlier are presented in appendix B.

Figure 6.1 Depth profile obtained using Ep ~320eV at ~50o.

30Si

-

59Si

2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Ip


113

6.2.1.1 30Si

- profiles

Figure 6.2(a)-6.2(c) show the 30

Si-

profiles (intensity vs apparent depth)

recorded at the near surface region for Ep ~ 320 eV, Ep ~ 500 eV and Ep ~ 1 keV

respectively, at different primary ion incident angles. The apparent depth was

calculated based on the assumption of a constant erosion rate from the surface to the

first delta-layer.1

For Ep ~ 320 eV, the 30

Si-

surface transient profiles obtained at various can

be grouped according to three distinct profile trends:

(i) 30Si

- intensity rises abruptly with depth before reaching a steady state. This

is observed at ~ 30-50o.

(ii) 30Si

- intensity rises abruptly but midway before arriving at steady state

intensity, it develops a shoulder. This is observed at ~ -20o.

(iii) 30Si

- intensity rises less abruptly, develops a peak before reaching the

steady state intensity. This is observed with profiles from ~ 60-70o.

When Ep ~ 500 eV, 30

Si- surface transient profiles from ~ 30-40

o rise

abruptly with no peaks or shoulders. Profiles from ~ 0-20o do not develop a

shoulder but instead increase gradually to the steady state intensity. Profiles from ~

50-70o develop a peak prior to reaching steady state intensity.

When Ep ~ 1 keV, 30

Si- surface transient profiles from ~ 20-30

o rise abruptly

with no peaks or shoulders. Similar to Ep ~ 500 eV, profiles from ~ 0-10o, do not

develop a shoulder but rise gradually. Finally, the last group of profiles from ~ 40-

70o, develop a peak prior to reaching the steady state intensity.


114

Figure 6.2 30

Si- profiles for (a) Ep ~320 eV (b) Ep ~500 eV and (c) Ep ~1 keV.

a) 320eV

1.E+03

1.E+04

1.E+05

1.E+06

Inte

ns

ity

(c

ps

)

0 10 20 3040 50 60 70

b) 500eV

1.E+03

1.E+04

1.E+05

1.E+06

Inte

ns

ity

(c

ps

)

c) 1keV

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

0 2 4 6 8 10

Apparent Depth (nm)

Inte

ns

ity

(c

ps

)


115

In the optimal profiles i.e. abrupt rise to steady state and therefore with the

shallowest transient (Ep ~ 320 eV/ ~ 30-50o; Ep ~ 500 eV/ ~ -40

o; Ep ~ 1 keV/

~ 20-30o), the rapid increase in the

30Si

- intensity corresponds to a progressive build-

up of Cs concentration at the initial stages of bombardment. This causes a rapid

increase in 30

Si- ionization

2 resulting in a yield enhancement effect.

3 As the Cs

intensity rises, the work function decreases proportionally, thus enhancing the yield of

negative ions.4 Subsequently, the Cs concentration stabilizes and the

30Si

- intensity

reaches a steady state. This equilibrium is achieved when the amount of Cs implanted

is equal to the amount of Cs sputtered. At this stage, the work function also

stabilizes.5

It is interesting to note that profiles with the shoulder appear only at the

ultralow-energy of 320 eV ( ~ -20o) but not at 500 eV and 1 keV. If we compare

this to Si+ profiles from O2

+ bombardment,

6 this profile is indicative of the onset of

roughening. However, ripple formation has been reported to occur only beyond ~

50o with Ep ~ 250 eV.

7 This is also supported by our

98SiGe

- profiles which show little

or no loss in depth resolution with depth implying the absence of roughening effects.

Further investigations were done with AFM. Figure 6.3 shows the AFM images of

1m x 1 m scans done at the crater bottom from ~ 0o, 10

o, 20

o and 60

o samples at a

depth of 2-3 nm. The bright features are very likely to be dust particles. The rms

roughness observed in the areas without the surface particles are ~ 0.18 nm for ~ 0o,

10o, 20

o and ~ 0.27 nm for ~ 60

o, where the onset of roughening is more obvious.

The change in the gradient of the profiles at the point of inflexion, where the

shoulder begins, is at an apparent depth of 0.5 nm. This is typically in the vicinity of

the native oxide layer (estimated at ~ 1-2 nm). The sputter rate at Ep ~ 320 eV / ~ 0-


116

Figure 6.3 AFM images of crater bottoms resulting from Ep ~ 320 eV a) ~ 0o, b)

~ 10o, c) ~ 20

o and d) ~ 60

o taken at 2-3 nm depth. The rms roughness values for

craters from ~ 0o, 10

o and 20

o are ~ 0.18 nm and for ~ 60

o is ~ 0.27 nm, where the

onset of roughening is noted.

20o is the lowest (cf. Fig. 6.6) compared to other incident angles, corresponding to a

greater implanted Cs build-up on the surface.8 The Cs concentration has also been

reported to be higher at the SiO2/Si interface,9 owing to possibly the difference in

stopping power10

or the strong cesium-oxygen bonding11,12

as Cs2O which hinders the

out-diffusion of Cs+.9 Therefore, as the

30Si

- intensity rises, the presence of oxygen

quenches the Cs-induced enhancement effect13

on Si- culminating in the point of

inflexion, possibly in the region of the SiO2/Si interface. On further bombardment

across the interface, the oxygen concentration will diminish and the interaction with

Cs will be negligible resulting in the subsequent gradual increase in the 30

Si– profile.

This does not happen at higher energies with similar , as the deeper Cs+ penetration


117

depth and higher sputter rate may have negated the cesium-oxygen interaction. This is

particularly significant for Ep ~ 1 keV where the ion penetration depth is ~ 1.3 nm.14

Similarly, the development of peaks in the profiles all occur at an apparent

depth of less than 1nm, which is possibly at the SiO2/Si interface. The Cs

concentration at the surface builds up rather slowly as the Cs accumulates at the

SiO2/Si interface coupled with a higher sputtering rate at ~ 40-70o (cf. fig. 6.6).

11,12

The peak corresponds to the high Cs concentration at the SiO2/Si interface. Beyond

the interface, the profile intensity decreases gradually as the oxygen diminishes and

the Cs builds up again with depth. The gradual rise in the 30

Si- intensity stabilizes only

after a considerable depth giving rise to a wider ztr, which will be discussed in Section

6.2.1.2. It is noted that the critical angle at which a transient peak occurs increases as

the primary ion energy is lowered: At Ep ~ 320 eV, c is 60o; at Ep ~ 500 eV, c is 50

o;

and at Ep ~ 1 keV, c is 40o.

6.2.1.2 Transient width

Figure 6.4 shows the measured transient width based on 30

Si- profiles as a

function of incident angle. The end of the surface transient was taken to be at 95% of

the equilibrium signal for 30

Si-, consistent with the criterion used by Jiang and

Alkemade12

for O2+ bombardment. We have also included plots of 2Rnorm of Cs

+ for

each Ep and since it is expected that ztr ~ 2-2.5Rnorm where there is no

roughening.13,15

The values are from TRIM calculations.14


118

Figure 6.4 Measured transient width based on 30

Si- profiles for Ep ~320 eV, 500 eV

and 1 keV. The plot of Rnorm of Cs+ primary ion at various Ep is also shown. For

conditions where extended transient effects is not observed, the ztr < 2Rnorm

We noted that the 30

Si- profiles reach equilibrium in the shortest time, giving

the narrowest ztr in the range of 1.4 nm to 2.7 nm. Using primary ion energy of 320

eV and 500 eV, the minimum ztr is observed at ~ 40o, with ztr ~ 2.0 nm and ztr ~ 1.4

nm respectively. At Ep~ 1 keV, the minimum ztr ~ 1.5 nm is observed at ~ 30o. The

range where the narrowest ztr is observed is greater, i.e. ~ 30-50ofor Ep ~ 320 eV

and 500 eV, but is limited to ~ 20-30o for Ep ~ 1 keV. The ztr at 1 keV/ ~ 40-50

o is

about 3.7 nm. Beyond ~ 50o for all Ep studied, the transient width broadens by 7-8

times compared to the narrowest ztr achievable. This broadening has been referred to

as the enhanced transient effect.10

This is an important observation, as ~ 60o is

widely used with Cs+ bombardment for dopant profiling in SIMS since the beam

30Si-

(320eV)

30Si-

(500eV)

30Si-

(1KeV)

2Rnorm 320 eV

2Rnorm 500 eV

2Rnorm 1 keV

0

5

10

15

20

25

0 10 20 30 40 50 60 70 80 90

Angle of Incidence

Tra

ns

ien

t W

idth

(n

m)


119

induced broadening effects was found to be a minimum16-18

and also due to the

limited incident angle range available in magnetic sector instruments.19-23

However,

with ztr ~ 20 nm, it is obvious that ~ 60o is not suitable for ultrashallow dopant

profiling. We also noted that ztr increases even though Ep decreases when profiled at

~ 60o. Under the conditions where extended ztr is not observed, ztr < 2Rnorm for Ep ~

320 eV and 500eV, while for Ep ~ 1 keV, ztr < 1.5Rnorm. At minimum ztr, Ep ~ 320 eV

/ ~ 30-50o, ztr ~ 1-1.4Rnorm; Ep ~ 500 eV / ~ 30-50

o, ztr ~ 0.7-1.4Rnorm; Ep ~ 1 keV

/ ~ 20-30o, ztr ~ 0.5-0.8Rnorm. This implies that the Cs concentration must have

stabilized in the region of where the majority of Cs+ ions come to rest.

6.2.1.3 Steady-state intensity of 30

Si- profiles

Figure 6.5 shows the steady state intensity (sputter yield + ion yield) of 30

Si- at

various and Ep. At all three primary ion energies investigated, the 30

Si- intensity at

steady state increases from when ~ 0o to a maximum when ~

o, then decreases

to a lower level than at ~ 0o as is increased from 40

o to 70

o at 10

o intervals. This

is consistent with earlier work using Ep ~ 1 keV where the Cs concentration was

found to decrease as increases from 30-80o.24

However, this trend differs from that

at higher energy25

where there is no change in the 30

Si- intensity for ~ 0-20

o. Since

the sputter rate is lowest at ~ 0o, intuitively the steady state Cs concentration, and

hence 30

Si- intensity, should be highest. However, this is not the case. At normal and

near normal incidence, Cs accumulates at the surface and hence less Si is sputtered as

compared to at ~ 30o. Beyond ~ 70

o, the high sputter rate together with an

increasing backscattered primary ion yield with , results in a lower Cs surface


120

Figure 6.5 30

Si- intensity at steady state for various Ep and . Si

- intensity from high

energy of Ep ~ 8 keV is also shown. 25

concentration which is too small to produce a significant enhancement of negative ion

yield.25

6.2.2 Sputter rates

Figure 6.6 shows the sputter rates at various angles of incidence and primary ion

energies. Two plots are shown for each primary energy, one (labelled d1) represents

the average sputter rate at the near surface, i.e. the average sputter rate from the

surface to the first Ge delta-layer(top 12 nm), and the other is the average sputter rate

observed from the surface to the 10th

Ge delta-layer.

As expected, the sputter rate increases with the incident angle, reaching a

maximum at ~ 60o in the case of both Ep ~ 500 eV and 1 keV, consistent with that

observed at higher energies. The maximum sputter rate for Ep ~ 320 eV occurs at ~

320eV

500eV

1keV

8keV1.E+04

1.E+05

1.E+06

0 10 20 30 40 50 60 70 80Angle of Incidence

Inte

ns

ity

(c

ps

)

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

Inte

ns

ity

(c

ou

nts

/nC

)


121

Figure 6.6 Sputter rates vs angle of incidence. For each Ep, d1 represents the average

sputter rate from the surface to the first delta-layer and the other is the average sputter

rate from the surface to the tenth delta-layer.

50o. This is similar to that observed with ultralow-energy O2

+ SIMS discussed earlier

in Chapter 4 (4.2.2).26

The lowest sputter rate is about 9.3 nm min-1

nA-1

cm2 and

occurs at normal and near normal incidence with Ep ~ 320 eV. For all energies

studied, the sputter rate is lowest at normal incidence. This is partly due to the

accumulation of Cs at the surface. As the incident angle increases, the penetration

depth of implanted Cs+ decreases

14 and the sputtering yield increases, resulting in a

higher sputter rate. Beyond ~ 60o, backscattering of the primary ions increases

leading to a subsequent decrease in the sputter rate.14

Figure 6.6 also shows that the near surface sputter rate is higher than the

average sputter rates for all and Ep. For all Ep up to ~ 50o, the difference in the d1

sputter rate to that of the average sputter rate is quite constant; beyond ~ 50o the

difference widens. This is indicative of a greater transient effect or the onset of

roughening.

500eVd1

d1

320eV

1keV

0

2

4

6

8

10

12

14

16

0 20 40 60 80

Angle of Incidence

Sp

utt

er

Ra

te (

nm

/min

)

0

1

2

0 20 40 60 80

320eV

d1


122

Figure 6.7 Normalized sputter rates to the last delta-layer throughout the depth.

a) 320eV

0.6

0.8

1.0

1.2

1.4

1.6

1.8R

ela

tiv

e S

pu

tte

r R

ate

0 10 20 30

40 50 60 70

b) 500eV

0.6

0.8

1.0

1.2

1.4

1.6

Re

lati

ve

Sp

utt

er

Ra

te

c) 1keV

0.6

0.8

1.0

1.2

1.4

1.6

0 20 40 60 80 100 120Depth (nm)

Re

lati

ve

Sp

utt

er

Ra

te


123

The increase in the sputter rates is not proportional to Ep. The variance is

greatest at oblique incidence. The sputter rates when Ep ~ 500 eV compared to when

Ep ~ 320 eV, increases by about 1.3 to 1.5 times at ~ 0-50o to double at oblique

incidence. The sputter rate when Ep ~ 1 keV is much higher compared to Ep ~ 320 eV,

about 6-9 times at ~ 0-50o and 13 times at oblique angles.

Figure 6.7 shows plots of relative sputter rates as a function of depth,

normalized to the sputter rate at the 9th

to 10th

Ge delta-layer. Using Ep ~ 320 eV and

~ 0-10o, the sputter rate is stable as a function of depth. The difference in sputter

rate at the near surface to the average sputtering rate is about 3%. At ~ 20-50o the

difference is about 10-15%. When Ep ~ 500 eV and ~ 0-30o, the difference between

transient and steady-state sputter rates is higher at 20%; but at ~ 40-50o, the

difference is reduced to about 14%. When Ep ~ 1 keV, the difference is 25% at ~

10-30o, and 15% at ~ 50-60

o. From these observations, we can infer that the

variance in the surface sputter rates to the average sputter rate increases with Ep.

Hence, only at very low energies of Ep ~ 320 eV and ~ 0-10o, may we use

the average sputter rate for a reasonably accurate depth conversion at the near surface.

For other combinations reported above, the error in the depth will be about 15%. The

results of the effect of ultralow-energy Cs+ on transient width and sputter rate of Si are

summarized in table 6.1.


124

Table 6.1 Summary of results from this work using ultralow-energy Cs+ sputtering.

320eV 500eV 1keV

30Si

- profile with no

peaks & no shoulders) ~ 30-50

o ~ 30-40

o

~ 20-30o

~ 0-10o

(gradual rise)

30Si

- profiles with

shoulders ~ 0 - 20

o Nil Nil

30Si

- profiles with peak ~ 60-70

o ~ 50-70

o ~ 40-70

o

Enhanced transient

width 11

> 57o

(250 eV)> 60

o > 65

o

Equilibrium signals

maximum intensity at ~ 30

o ~ 30

o ~ 20-30

o

c roughening (evaluated

at 25m) 27

~ 50o

(250eV) ~ 55o

o

Range of giving

lowest ztr ~ 30-50o ~ 30-50

o ~ 20-30

o

Lowest ztr 2.0 nm 1.4 nm 1.5 nm

Sputter rate max. at ~ 50o ~ 60

o ~ 60

o

Sputter rate min. at ~ 0o, 70

o ~ 0

o ~ 0

o

Stable erosion rate

(surface cf depth) at

(i) ~ 0-10o

(ii) ~ 20-50o ~ 40-50

o ~ 50-60

o

Difference in sputter rate

between near surface

and throughout depth for

with stable erosion

rate

(i) ~ 3%

(ii) ~ 15% ~ 15% ~ 15%


125

6.3 Summary

The lowest transient width observed with ultralow-energy Cs+ ion bombardment is

at an apparent depth of between 1.4 to 2.0 nm, observed under the following

conditions: Ep ~ 320, 500 eV / ~ 30-50o and Ep ~ 1 keV / ~ 20-30

o. To the best of

our knowledge, this is the lowest ztr that has been reported with Cs+ primary beams.

We postulate that the onset of roughening occurs in the profiles with peaks and

shoulders since the profiles take a longer time to reach equilibrium. For > 50o,

extended transient effects are observed. For conditions where this effect is not

observed, ztr < 1.5-2Rnorm. Under the conditions where minimum ztr is achieved, ztr ~

Rnorm, implying that the Cs concentration stabilizes in the vicinity of Rnorm.

The detection sensitivity is best when profiled at ~ 30o for all energies used. The

sputter rate is lowest at normal incidence, gradually increasing to a maximum at about

~ 50-60o. This is similar to that reported for O2

+ bombardment at ultralow-energy.

Only for Ep ~ 320 eV / ~ 0-10o can we use the average sputter rate. For other

conditions such as for Ep ~ 320 eV / ~ 20-50o, Ep ~ 500 eV / ~ 40-50

o and Ep ~ 1

keV / ~ 50-60o, the error in the apparent depth at the top 12 nm to the average

sputter rate is better than 15%.

As reducing Ep is not significant in reducing ztr at ultralow-energies, we

conclude that the optimum condition for ultrashallow SIMS profiling is at ~ 30o for

all ultralow-energies (Ep < 1 keV) as it gives the lowest ztr, highest detection

sensitivity and good sputter rate. Transient width and sensitivity in sputter profiling

with Cs+ at ultralow-energy can thus be optimized simultaneously. Another important

characteristic of depth profiling is the depth resolution, which will be presented in the


126

next chapter. Profiling at ~ 60o as is commonly done is less sensitive, has larger

transient width and also roughening effects. Hence although ~ 60o is widely used for

dopant profiling with good depth resolution, it is not suitable for ultrashallow

profiling where the bulk of the implant profile lies in the transient region.

References





4. M. L. Yu, Nuclear Instruments & Methods in Physics Research Section B-Beam

Interactions with Materials and Atoms 15, 151 (1986).




7. Y. Kataoka, K. Yamazaki, M. Shigeno, Y. Tada, K. Wittmaack, Applied Surface

Science 203, 43 (2003).

8. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. Bennett, Applied Surface

Science 203, 156 (2003).

9. M. Anderle and C. M. Loxton, Nuclear Instruments & Methods in Physics

Research B 15, 186 (1986).

10. P. Williams and J. E. Baker, Applied Physics Letters 36, 842 (1980).


12. K. Wittmaack, Nuclear Instruments & Methods in Physics Research B 7-8, 750

(1985).

13. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. W. Rabalais, Journal of

Chemical Physics 113, 10344 (2000).



127

15. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. Bennett, Nuclear Instruments

and Methods in Physics Research B 201, 413 (2003).

16. C. W. Magee, S. A. Cohen, D. E. Voss, D. K. Brice, Nuclear Instruments &

Methods 168, 383 (1980).

17. C. W. Magee and R. E. Honig, Surface and Interface Analysis 4, 35 (1982).

18. J. J. Lee et al., Journal of Vacuum Science & Technology A 8, 2287 (1990).

19. J. Bennett, Surface and Interface Analysis 25, 454 (1997).

20. T. H. Buyuklimanli, J. W. Marino, S. W. Novak, Applied Surface Science 231-

232, 636 (2004).

21. T. Eto and K. Shibahara, Japanese Journal of Applied Physics 44, 2433 (2005).

22. G. R. Mount, C. J. Hitzman and S. P. Smith, Secondary Ion Mass Spectrometry

SIMS XI edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie, (Wiley

Chichester, 1998), p. 317.

23. M. H. Yang, G. Mount, I. Mowat, Journal of Vacuum Science & Technology B

24, 428 (2006).

24. P. A. W. van der Heide, C. Lupu, A. Kutana, J. W. Rabalais, Applied Surface

Science 231-2, 90 (2004).


26. A. R. Chanbasha and A. T. S. Wee, Surface and Interface Analysis 37, 628

(2005).


Science 203, 43 (2003).

Chapter 7: Cs+ SIMS and depth resolution

128

Chapter Seven

Effect of ultralow-energy Cs+ SIMS on depth resolution

7.1 Introduction

It is known that depth resolution can be improved by lowering the primary ion

impact energy and/or increasing the incident angle up to a critical , beyond which,

surface roughening ensues. However, a lower Ep is accompanied by lower secondary

ion yield, and for an ultralow-energy Cs+ primary beam, a poorly focused beam. In

this chapter, we investigate the effect of depth resolution with ultralow-energy (<1

keV) Cs+ bombardment over a wide range of incident angles ( ~ 0-70

o) with the aid

of Ge delta-layers in Si. The 98

SiGe- profiles were systematically evaluated by

calculating the FWHM and exponential decay to a depth of 120 nm. Based on the

results obtained, the optimum conditions demonstrating high depth resolution are

ascertained. It is established that the relationship between improvements in depth

resolution (FWHM) and increasing is linear and gradual. The factors affecting depth

resolution are also examined with the use of the mixing-roughness-information model

(MRI). With this model, it is possible to differentiate the effect of atomic mixing and

surface roughness on the resolution of the -layers. The impact of atomic mixing,

surface roughness and instrument conditions (poor focus) on depth resolution is also

discussed. Finally, the optimum conditions for high depth resolution with good

dynamic range are recommended.


129


7.2.1 Depth resolution

Figure 7.1(a) shows typical depth profiles of 30

Si-,

59Si2

- and

98SiGe

- using Ep ~

320 eV and ~ 50o. The

98SiGe

- profile shows ten well-resolved -layers. However,

before the trailing edge of the first peak reaches the background intensity level, the

leading edge of the next peak begins. This is a result of the atomic mixing depth being

comparable to the inter--layer spacing. We also noted tailing at the base of the

trailing edge, and the baseline appears to drift linearly upward with depth. Similar

profiles were done using different Ep and and the results are further evaluated.

Figure 7.1(b) shows an example of 98

SiGe- profiles when Ep is varied, using Ep

~ 320 eV, 500 eV and 1 keV at ~ 60o, normalized to the first peak of Ep ~ 500eV

profile. The depth resolution is observed to be slightly better with lower Ep at

shallower depths (< 23 nm); but the resolution deteriorates more quickly with depth as

compared to the profile at Ep ~ 1 keV. This can be seen from the rapidly decreasing

peak-to-valley ratio (PVR). The PVR is comparable at all Ep at the first delta-layer but

becomes smaller with depth, the deterioration being more severe at lower Ep.

Figure 7.1(c) shows typical 98

SiGe- profiles when is varied, using Ep ~ 320

eV with ~ 40-70o, and normalized to the first peak at ~ 40

o. We observe that the

~ 50o profile has the narrowest peak, indicating the best depth resolution. The ~ 40

o

profile has a gentler trailing edge whereas the ~ 60o

profile has a gentler leading

edge. Both ~ 60o and ~ 70

o profiles show greater deterioration in depth resolution

with depth compared to ~ 40o and ~ 50

o profiles.


130

Figure 7.1 Typical depth profiles (a) 98

SiGe- profiles with

30Si

- and

59Si

2- (b)

98SiGe

-

profiles with various Ep at ~ 60o (c)

98SiGe

- profiles with Ep ~ 320 eV at various .

a) 320 eV / ~ 50o

30Si

-

59Si

2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

320eV

500 eV

1 keV

b) 60o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

Inte

ns

ity

(c

ps

)

320eV 500eV 1keV

40o

50o

c) 320eV

60o

70o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 20 40 60 80 100 120 140Depth (nm)

Inte

ns

ity

(c

ps

)

40 50 60 70


131

7.2.1.1 Depth Resolution in terms of FWHM

Figures 7.2(a)-7.2(c) show the depth resolution measured in terms of FWHM

of the 98

SiGe- peaks obtained at Ep ~ 320 eV, 500 eV and 1 keV and ~ 0-70

o against

depth. At Ep ~ 320 eV, good depth resolution is obtained at ~ 30-50o with FWHM

of less than 2.4 nm throughout the analysis depth of 120 nm. The best depth

resolution is observed at ~ 50o with a mean FWHM of 1.9 nm. The difference in

depth resolution from the first -layer to the last -layer is about 18%. At ~ 60o, the

depth resolution is stable at FWHM of 2.6 nm up to a depth of 23 nm (second -layer)

before deteriorating linearly to 5.7 nm at the last -layer. The worst depth resolution is

at ~ 70o, beginning with a FWHM of 3.4 nm that degrades with depth.

The trends in depth resolution are similar at Ep ~ 500 eV with good depth

resolution of less than 2.5 nm FWHM observed at ~ 30-50o. The best depth

resolution is at ~ 50o with a mean FWHM of 2.2 nm throughout the analysis depth.

The difference in depth resolution from the first -layer to the last -layer is about

17%. At ~ 60o, the FWHM at the first -layer (depth of 12 nm) is 2.3 nm; but it

deteriorates with depth to two and half times the initial FWHM at the last -layer. The

depth resolution is worst at ~ 70o, with a FWHM of 3.3 nm at the first -layer and

degrading thereafter.

Using Ep ~ 1 keV, we observe a different trend compared to that at lower Cs+

primary ion energy. The depth resolution deteriorates with depth when profiled at ~

70o but not at ~ 60

o. The best depth resolution is observed when profiled at ~ 60

o

with a mean FWHM of 2.5 nm up to a depth of about 80 nm (seventh -layer) before


132

Figure 7.2 Depth resolution of Ge delta-layers as a function of profile depth,

measured by FWHM for (a) Ep ~ 320 eV, (b) Ep ~ 500 eV and (c) Ep ~ 1 keV at

various incident angles.

a) 320eV

0-20o

30-50o

60o

70o

0

1

2

3

4

5

6

7

8F

WH

M (

nm

)

0 10 20 30

40 50 60 70

b) 500eV

0-20o

30-50o

60o

70o

0

1

2

3

4

5

6

7

FW

HM

(n

m)

c) 1 KeV

0-30o

40-60o

70o

0

1

2

3

4

5

6

7

0 20 40 60 80 100 120 140

Depth (nm)

FW

HM

(n

m)


133

it deteriorates gradually by 34% at the last delta-layer. At ~ 50o, the mean FWHM

is 2.6 nm up to a depth of 103 nm (ninth delta-layer), which is slightly worse than at

~ 60o but constant to a greater depth.

Given the above observations, a good depth resolution of about FWHM ~ 2

nm can be obtained at an ultralow energy of 320 – 500 eV with ~ 50o throughout the

depth analysed. At higher primary ion energy of 1 keV, the best depth resolution is

achievable at a higher incident angle of ~ 60o. These observations with Ge delta-

layers are similar to those made with Sb delta-layers and B delta-layers.1,2

Beyond

these incident angles, the depth resolution deteriorates severely, consistent with the

data reported by Li et al.3 Generally, good depth resolution is achievable across a ~

20o range of ; namely at ~ 30-50

o for Ep ~ 320 or 500 eV, and ~ 40-60

o for Ep ~ 1

keV.

Figures 7.3(a) – 7.3(c) show the depth resolution (in terms of FWHM) of three

representative Ge delta-layers at the near surface d1 (12.2 nm), intermediate depth d5

(57.8 nm) and deepest at d10 (114.8 nm) for all three energies investigated at ~ 0-

70o. Generally, the depth resolution improves gradually as increases up to a critical

angle before worsening considerably. The critical angle is ~ 50o at Ep ~ 320 eV and

Ep ~ 500 eV, and at ~ 60o at Ep ~ 1 keV. Beyond the critical angles, the depth

resolution degradation is more severe with depth as can be seen from the increasing

slopes of d1, d5 and d10.


134

Figure 7.3 Depth resolution of Ge delta-layers at various primary ion incident angles.

a) 320 eV

d1

d5

d10

0

1

2

3

4

5

6

7

8

FW

HM

(n

m)

d1 d5 d10

b) 500eV

d1

d5

d10

0

1

2

3

4

5

6

7

8

FW

HM

(n

m)

c) 1keV

d1

d5

d10

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70

Incident Angle

FW

HM

(n

m)


135

Figure 7.4(a) shows depth resolution in terms of FWHM of the first Ge delta-

layer peak for all three Ep at ~ 0-70o. It confirms that depth resolution improves with

decreasing Ep provided that there is no surface roughening. However, the

improvement in depth resolution is marginal, about 1.3 times from 1 keV to 320 eV.

This improvement is not as significant as that observed when using O2+ where the

improvement in depth resolution was more than 2.5 times.4

Figure 7.4(b) shows the plot of depth resolution in terms of FWHM (denoted

as dz) of the Ge delta-layer peak for d1, d5 and d10 against up to the critical angle

for all three impact energies. At each impact energy, a linear relationship is observed

as follows:

For Ep ~ 320 eV dz = -0.019 + 3.0 (2)

For Ep ~ 500 eV dz = -0.019 + 3.2 (3)

For Ep ~ 1 keV dz = -0.019 + 3.8 (4)

The incremental improvement in depth resolution (decreasing FWHM) with

increasing is linear, gradual and noticeably similar across all Ep.


136

Figure 7.4 (a) Depth resolution of the first Ge delta-layer peak at various Ep. (b)

Linear relationship between depth resolution in terms of FWHM with when no

surface roughening is present.

1 keV

a) d1

500 eV

320 eV

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70Incident Angle

FW

HM

(n

m)

1 keV 500 eV 320 eV

b)

y = -0.019x + 3.0

y = -0.019x + 3.2

y = -0.019x + 3.8

0

1

2

3

4

5

0 10 20 30 40 50 60 70Incident Angle

FW

HM

(n

m)

320 eV 500 eV 1 keV


137

7.2.1.2 Depth Resolution in terms of exponential decay

Figures 7.5(a)-7.5(c) show the decay length for selected delta-layers using Ep

~ 320 eV, 500 eV and 1 keV against 0-70o. Generally, for the first delta-layer, the

d decreases as increases, similar to that observed at higher energy (Ep ~ 8 keV).5

For Ep ~ 320 eV, we observe d for the first delta-layer decreasing with increasing

( >10o) from 2.7 nm/e to a minimum at 1.1 nm/e at ~ 50-70

o. d values at the fifth

and tenth delta-layers are correspondingly higher, also decreasing as increases

reaching a minimum at ~ 50o. At Ep ~ 500 eV, d at the first delta-layer decreases

with increasing from 2.5 nm/e to 1.1 nm/e at ~ 60-70o; at 1 keV, d decreases with

increasing from 3.8 nm/e to 1.5 nm/e at ~ 70o.

Figure 7.5(d) compares the exponential decay of the first delta-layer against

depth for various Ep. At Ep ~ 320 eV, increasing beyond 50o does not improve d

significantly. Similarly for Ep ~ 500eV, increasing beyond 60o does not improve d.

However, an improvement is observed with Ep ~ 1 keV when is increased beyond

~ 60o but limited to the near surface regions ( ~ 12 nm) only, similar to that estimated

by Wittmaack5 at Ep ~ 8 keV. This has been observed in the case of depth profiling of

75As implants

6 where using Ep ~ 1 keV / ~ 75

o gives better depth resolution than at

~ 60o, when only the trailing edge is evaluated. It is clear that d decreases with Ep but

at ultralow energies (Ep < 1 keV) improvements in d beyond ~ 50-60o is negligible.


138

Figure 7.5 (a) – (c) Plot of decay length (d) against incident angle () for various Ep

at selected depth. (d) Comparison of d at the first delta-layer for various Ep.

a) 320 eV

d1

d5

d10

0

1

2

3

4

Ex

po

ne

nti

al D

ec

ay

(n

m/e

)

d1 d5 d10

b) 500 eV

d1

d5

d10

0

1

2

3

4

Ex

po

ne

nti

al D

ec

ay

(n

m/e

)

c) 1 keV

d1

d5

d10

0

1

2

3

4

5

Ex

po

ne

nti

al D

ec

ay

(n

m/e

)

320 eV500 eV

d) d1

1 keV

0

1

2

3

4

5

0 10 20 30 40 50 60 70

Incident Angle

Ex

po

ne

nti

al D

ec

ay

(n

m/e

)

320 eV 500 eV 1 keV


139

Figures 7.6(a)–7.6(c) show the overlay of profiles of the first Ge delta-layer

obtained at various and Ep normalized to the peak obtained at ~ 30o. The trailing

edges at all three values of Ep are always gentler than the leading edges due to knock-

on and ion beam mixing effects. It can be clearly seen that at Ep ~ 320 eV, the depth

resolution is best at ~ 50o [Fig. 7.6(a)]. At Ep ~ 500 eV, the depth resolution is

marginally better at ~ 50o compared to ~ 60

o [Fig. 7.6(b)]. Even though d

obtained when using Ep ~ 500 eV / ~ 60o is smaller than that at ~ 50

o, the depth

resolution at ~ 50o is better as the FWHM is smaller, i.e. there is less ion

bombardment induced peak broadening. At Ep ~ 1 keV, it is best at ~ 60o [Fig.

7.6(c)]. Therefore, based on FWHM and d data, we conclude that the best depth

resolution is achievable using Ep ~ 320 eV / ~ 50o, Ep ~ 500 eV / ~ 50

o and Ep ~ 1

keV / ~ 60o.


140

Figure 7.6 Comparison of the normalized 98

SiGe- profiles of the first delta-layer at

various Ep.

30o

40o

50o

a) 320 eV

60o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05In

ten

sit

y (

cp

s)

30 4050 60

30o

40o

50o

b) 500 eV

60o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

Inte

ns

ity

(c

ps

)

c) 1 keV 30o

40o

50o

60o

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 5 10 15 20

Depth (nm)

Inte

ns

ity

(c

ps

)


141

7.2.1.3 Depth resolution evaluated with MRI model

Figure 7.7 shows the penetration depth of Cs+ primary ions at ultralow-

energies from ~ 0-70o using TRIM calculations.

7 In Figure 7.1(b), when Ep is

increased at a constant incident angle ~ 60o, we observe that the depth resolution of

the first delta-layer (at ~ 12 nm depth) is poorer at higher impact energy (Ep ~ 1 keV).

This is expected since Ep scales with the penetration depth and therefore w. At the

same depth, there is no difference in d between the lower energies (Ep ~ 320 eV and

500 eV) suggesting that the small difference in penetration depth does not affect w

and therefore has little or insignificant impact on d. However, as the profile deepens,

lowering Ep results in a worsening of the depth resolution as can be seen by the peak

broadening and a drop in the PVR. Since w decreases with Ep and the penetration

depth cannot explain this worsening, sputter-induced roughening must be the

dominant factor in causing the degradation in the depth resolution. Surface

roughening as evidenced by AFM measurements has been shown to contribute to the

deterioration of depth resolution.1,8,9

By analysing the ~ 60o data in figure 7.2, we

deduce that the onset of roughening for Ep ~ 320 eV occurs after 23 nm (second delta-

layer), and that for Ep ~ 500 eV occurs after 12 nm (first delta-layer). These findings

complement the observations made by Kataoka et al.1 where conclusions on surface

roughening were based on observations made at 35 nm.

In figure 7.1(c), where the impact energy is constant at Ep ~ 320eV and is

varied, the depth resolution at ~ 40o is worse than at ~ 50

o. The peak broadening,

however, is asymmetrical and occurs only at the trailing edge. This is obviously

caused by a larger w at ~ 40o. When the incident angle is increased further to ~70

o,


142

Figure 7.7 Penetration depths of Cs+ primary ions at various incident angles

calculated from TRIM. 7

the depth resolution does not improve but instead worsens when compared to that at

~ 50o. Even though the penetration depth is shallower at such oblique angles, the

narrowing of the atomic mixing length (w) is insignificant when compared to the peak

broadening contributed by surface roughening. We infer that surface roughening

escalates beyond the second delta-layer as the depth resolution deteriorates at a faster

rate at ~ 60o. We also notice that surface roughening corresponds to a drop in the

peak intensity as it reduces the sputter rate and hence the ion yield and intensity. On

the other hand, when the degradation in depth resolution is dominated by atomic

mixing as at ~ 40o, it corresponds to an increase in d and no change in peak

intensity. In both cases a drop in PVR is noticeable.

0

1

2

3

4

5

0 10 20 30 40 50 60 70

Incident Angle

De

pth

(n

m)

1 keV 500 eV 320 eV


143

In figures 7.2 and 7.3, the increase in FWHM observed are a result of the onset

of roughening occurring at ~ 60o for Ep ~ 320 eV and ~ 500 eV, and at ~ 70

o for

Ep ~ 1 keV. In figure 7.4(a), under conditions where surface roughening does not

occur, peak broadening is expected since the penetration depth increases with Ep

causing a wider w and d. This conclusion is also confirmed by analyzing figure 7.6.

In figure 7.5(d), the decrease in d near the surface ( ~ 12 nm) levels off when

increases at ~ 50o with Ep ~ 320 eV, ~ 60

o with Ep ~ 500 eV and Ep ~ 1 keV. We

propose two possibilities for this occurrence. As increases, the primary ion

penetration depth becomes shallower thus reducing atomic mixing and d but as

surface roughening sets in, the decrease in d is offset by the peak broadening brought

about by ripple formation. Alternatively, the decrease in penetration depth at oblique

angles is not significant enough to cause a variation in d. Another observation is that

with Ep ~ 320 eV [Fig. 7.5(a)] and 500 eV [Fig. 7.5(b)], d deteriorates (increases)

with depth only at ~ 60-70o onwards, even though d is decreasing with increasing

at the surface. We infer that the deterioration ind with depth is mainly attributed to

the onset of roughening which begins at a depth of 12-23 nm. Similarly, at Ep ~ 1 keV

[Fig. 7.5(c)], the onset of roughening is experienced at ~ 70o. Figure 7.5(a) shows

that only at Ep ~ 320 eV, we observe a significant difference in d with depth

represented by d1, d5 and d10. At ~ 0-50o where no surface roughening is present,

the wider d can be attributed to poor beam focus which is commonly experienced

using ultralow-energy Cs+ beams. Poor beam focus has been reported to cause erosion

inhomogeneity.10

We studied the profiles obtained at incident angles where surface

roughening does not occur and concluded that the effect of poor beam focus is


144

characterized by a linear increase in baseline with depth but with no change in the

peak intensity [cf figure 7.1(a)]. We noted, however, that an increase in baseline also

occurs when there is surface roughening but it is always accompanied by peak

broadening and a decrease in peak intensity. In both situations, the PVR will decrease

but only in the case of surface roughening where a deterioration in depth resolution is

obvious. Hence, PVR is not a conclusive measure of depth resolution. The

deterioration in depth resolution caused by poor beam focus with depth is better

understood with the following relationship:

dz = M + Uz (9)

where the U term reflects instrumental problems (poor focus in this case) and the M

term represents the peak shape due to ion beam broadening, which like atomic mixing

is independent of depth.11

While the drop in PVR with depth is obvious, the increase

in d with depth does not affect the FWHM significantly (cf Figure 7.2)

Table 7.1 shows the best depth resolution obtained with Cs+ SIMS compared

to that from a similar work done with O2+ primary ion SIMS.

4 The penetration depth

based on TRIM calculations for the with the best depth resolution is also tabled. We

find that with Ep ~ 1 keV Cs+ SIMS gives better depth resolution than O2

+ SIMS. This

is expected as a higher mass primary ion will give a shallower penetration depth and

hence narrower atomic mixing width. Penetration depth is determined by the rate of

energy loss along the path of the ion.12

At constant energy, the bombarding ion with a

higher mass (larger atomic no.) will have a higher nuclear energy-loss rate in an

elastic binary collision due to its higher nuclear cross-section when penetrating the

target. However, at lower Ep, the best depth resolutions are similar. This unexpected

behaviour can be explained by the contribution to depth resolution from a poorly


145

Table 7.1 Primary ion penetration depth and FWHM at with best depth resolution.

focused Cs+ beam. The Cs

+ beam with a diameter larger than the O2

+ beam is more

difficult to focus as it fills up the aperture as it enters the focusing lens in the ion gun

column. Moreover, at ultralow-energy, the beam is retarded as it enters the lens which

spontaneously expands the beam.

Figure 7.8 shows the profile resolution data (FWHM and d) from the first

delta-layer as a function of beam energy when bombarded at normal incident. Similar

to that observed with O2+ primary ion beam, the data demonstrates effectively the

improvement in depth resolution by the use of lower Ep. A linear plot with

extrapolation to „zero- energy‟ has been done though the fit is not as good as that

compared to the data from O2+ primary ion beam. The value of the intercept for

FWHM is 2.2 nm and the corresponding value for d is 1.6 nm. Both values are

higher than that obtained with O2+ primary ion beam. As discussed earlier these limits

are influenced by the poorly focussed Cs+ primary ion beam which caused an increase

in surface roughening due to inhomogeneous sputtering.

Penetration depth (nm) Depth resolution FWHM (nm)

Ep O2+ Cs

+ O2

+ Cs

+

250 eV 1.7 ( ~ 40o) - 1.5 -

320 eV - 1.7 ( ~ 50o) - 1.9

500 eV 2.8 ( ~ 30o) 1.9 ( ~ 50

o) 2.2 2.2

1 keV 4.6 ( ~ 20o) 2.0 ( ~ 60

o) 3.5 2.5


146

Figure 7.8 Depth resolution parameters (FWHM and d) dependence on energy

for the first delta-layer profile at ~ 0o.

7.2.2 Dynamic range

Figure 7.9 shows the dynamic range averaged over the first nine peaks. The

dynamic range gives an indication of the range of concentrations that can be detected

by SIMS. For all three primary ion energies studied, the dynamic range increases as

moves away from normal. A good dynamic range of more than three decades is

obtained at ~ 40-70o with all impact energies studied. The results are summarized in

table 7.2.

y = 0.0015x + 2.185

y = 0.0021x + 1.65

0

1

2

3

4

5

0 250 500 750 1000 1250Primary Ion Energy (eV)

Re

so

luti

on

pa

ram

ete

rs(n

m)

FWHM

d


147

Figure 7.9 Dynamic range averaged over the first nine 98

SiGe- peaks vs incident

angle.

320eV

500eV

1keV

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

0 10 20 30 40 50 60 70

Incident Angle

Dy

na

mic

Ra

ng

e (

Ima

x/I

min

)

320eV 500eV 1keV


148

Table 7.2 Summary of results observed with ultralow-energy Cs+ sputtering.

7.3 Summary

High depth resolution can be achieved with ultralow-energy Cs+ SIMS. The

trends in depth resolution are quite similar when using Ep ~ 320 eV and Ep ~ 500 eV

with high depth resolution of less than 2.5 nm FWHM observed at ~ 30-50o

throughout the depth range studied. The best depth resolution of FWHM 1.9 nm is

observed at ~ 50o with Ep ~ 320 eV and 2.2 nm with Ep ~ 500 eV. With Ep ~ 1 keV,

320 eV 500 eV 1 keV

Lowest FWHM 1.9nm 2.2nm 2.5nm

with lowest FWHM

throughout depth

(120 nm)

~ 50o ~ 50

o

~ 60o

(up to 80 nm)

with low FWHM to

a limited depth

~ 60o

FWHM ~ 2.6 nm

(up to 23 nm)

~ 60o

FWHM ~ 2.3 nm

(up to 12 nm)

~ 50o

FWHM ~ 2.6 nm

(up to 103 nm)

with good depth

resolution

~ 30-50o

(< 2.4 nm)

~ 30-50o

(< 2.5 nm)

~ 40-60o

(< 3.0 nm)

Lowest d ~ 1.1 nm/e ~ 1.1 nm/e ~ 1.5 nm/e

with lowest d

throughout depth ~ 50-70o ~ 60 -70

o ~ 70

o

with best dynamic

range ~ 60o

~ 50o

~ 70o

Highest dynamic

range 1.1 x 103

3.0 x 103 2.8 x 10

3


149

depth resolution of less than 3 nm FWHM is observed at ~ 40-60o. The best is

observed at ~ 60o with a mean FWHM of 2.6 nm.

By considering the MRI model, we confirm that depth resolution can be

improved by reducing atomic mixing and/or eliminating surface roughness. Atomic

mixing is reduced by lowering Ep and/or by increasing up to a critical incident angle

of ~ 50o at Ep ~ 320 eV and ~ 500eV, and at ~ 60

o for Ep ~ 1 keV. Beyond the

critical angle, the depth resolution deteriorates severely with depth due to the onset of

surface roughening. However, a close examination reveals that the onset of surface

roughening is present with Ep ~ 320 eV / ~ 60o only after 23 nm and with Ep ~

500eV / ~ 60o after 12 nm. The relationship between depth resolution (decreasing

FWHM) and is ascertained to be linear and gradual with all Ep evaluated.

The decay length decreases as is increased, with the narrowest d ~ 1.1-1.5

nm/e obtained at oblique angles. We can extrapolate that a narrower d can be

achieved beyond ~ 70o but only at the near surface. However, at this , the FWHM

is broader. Hence smaller decay lengths do not necessary indicates a better resolution.

Under the conditions when d is narrow, it is useful for distinguishing interfaces. d

also decreases with impact energy but does not decrease significantly at ultralow-

energy (Ep < 1 keV). A small change in the penetration depth does not affect the

atomic mixing length and therefore has little or an insignificant impact on d.

Using the MRI model, we can establish whether the broadening is dominated

by surface roughening or atomic mixing. Surface roughening corresponds to a

decrease in peak intensity and peak broadening while atomic mixing corresponds to

an increase in the exponential decay of the trailing edge but not a reduction in peak


150

intensity. Another phenomenon that is observed in Cs+ SIMS is poor beam focus. This

is seen as a linear increase in profile baseline but without any change in peak intensity

provided there is no surface roughening.

Cs+ SIMS gives better depth resolution than O2

+ SIMS at higher impact energy

but not at lower energies below Ep ~ 500 eV due to the poor beam focus that causes

uneven erosion. We suspect that poor beam focus limits the improvement in depth

resolution at ultralow energy.


151

References


Science 203, 43 (2003).

2. P. A. W. van der Heide, M. S. Lim, S. S. Perry, J. Bennett, Nuclear Instruments

and Methods in Physics Research B 201, 413 (2003).

3. Z. P. Li, T. Hoshi, R. Oiwa, Applied Surface Science 203, 323 (2003).

4. A. R. Chanbasha and A. T. S. Wee, Journal of Vacuum Science & Technology B

24, 547 (2006).

5. K. Wittmaack, Journal of Vacuum Science & Technology A-Vacuum Surfaces

and Films 3, 1350 (1985).

6. G. R. Mount, C. J. Hitzman and S. P. Smith, Secondary Ion Mass Spectrometry

SIMS XI, edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie (Wiley,

Chichester, 1998), p. 273.

7. J. F. Ziegler and J. P. Biersack, SRIM 2003, http://www.srim.org (2003).

8. C. M. Ng, A. T. S. Wee, C. H. A. Huan, A. See, Nuclear Instruments & Methods

in Physics Research B 179, 557 (2001).

9. G. S. Lau, E. S. Tok, R. Liu, A. T. S. Wee, J. Zhang, Nuclear Instruments &

Methods in Physics Research B 215, 76 (2004).



Spectrometry: Principles and Applications (Oxford University Press, New York,

1989).



1996), p. 88.

Conclusion

152

Chapter Eight

Conclusion

Ultralow-energy SIMS has been introduced for almost ten years now to meet

the increasing demands to reduce the transient width and to increase the depth

resolution possible for ultrashallow depth profiling and ultrathin films. However,

there are insufficient comprehensive studies done on the effects of ultralow-energy

O2+ and Cs

+ primary ion beams on surface transient and depth resolution over a wide

range of incident angles. Reducing the primary ion energy and increasing the incident

angle has positive effects on reducing the transient width and improving the depth

resolution. It is critical that the effects, the processes involved, the possibilities and

the limitations be understood so that methods can be developed towards achieving

more accurate SIMS depth profiling data to support the needs of the semiconductor

industry.

8.1 Surface transient

One of the motivations of this study is the need to qualitatively and

quantitatively profile ultrashallow junctions consistent with the demands of the

semiconductor industry. A significant part of the doped region in ultrashallow

junctions coincides with the surface transient region of the SIMS technique. The

surface transient cannot be avoided even with the introduction of ultralow-energy

SIMS as the dynamics of ion-surface interactions takes time to reach equilibrium.

Nevertheless, it is imperative that the inaccuracies owing to varying sputter rate and

Conclusion

153

secondary ion yield experienced during the surface transient should be limited to a

very shallow depth (narrow transient width).

Our study with an ultralow-energy O2+ primary ion beam shows that the

minimum transient width is achievable at normal and near normal incidence. To date,

only normal incidence has been reported. The lowest transient width achieved is ~ 0.7

nm (apparent depth) with primary ion energies of less than 500 eV. However, we

notice that at ultralow-energy, reducing the incident energy beyond 500 eV does not

result in a significant reduction in transient width. Nevertheless, the range of incident

angles to which the minimum transient width can be achieved is greater when the

incident energy is lowered. Equilibrium is achieved earlier as a result of complete

oxidation of silicon which is readily achieved at these incident angles. The complete

oxidation of silicon also prevents the onset of roughening with Ep ~ 250 eV.

When an ultralow Cs+ primary ion beam is used, a minimum transient width of

1.4 nm to 2.0 nm (apparent depth) is achieved with Ep ~ 320 eV, 500 eV, ~ 30-50o

and Ep ~ 1 keV, ~ 20-30o. This is the lowest transient width that has been reported.

Unlike O2+

sputtering, where the oxidation of silicon reduces the time taken to reach

equilibrium, Cs+ sputtering proceeds with a gradual build up of Cs concentration on/in

the silicon substrate. Hence, the wider transient width as compared to that achieved

with O2+ sputtering. We have shown that the narrowest transient widths are achieved

when the Cs concentration stabilises in the vicinity of the Cs penetration depth

(Rnorm).

Beyond ~ 50o for all energies studied, an extended transient effect has been

observed. This observation suggests that ~ 60o which has been widely used for

Conclusion

154

depth profiling with Cs+, is not appropriate for profiling ultrashallow junctions i.e.

when narrow transient is essential.

There are techniques such as silicon capping and backside depth profiling that

have been developed and are successful in circumventing the surface transient effect.

Nevertheless, the sample preparation that is necessary prior to SIMS depth profiling

in both techniques are very tedious and are not suitable for routine analysis.

8.2 Sputter rate

In the ultralow-energy regime, the sputter yield decreases correspondingly.

This is an important consideration, as it affects the speed of analysis. For both primary

ion beam species used, the lowest sputter yield is experienced at normal incidence and

the highest at an incident angle of 50-60o which is similar to that at higher energy ion

sputtering. However, it is worthwhile to note that at Ep ~ 250 eV, the maximum

sputter rate is observed at ~ 50o.

Another important consideration for sputter rate is the use of average sputter

rate for depth conversion. We confirmed that the intrinsic surface transient causes the

sputter rate at the near surface to be higher than the average sputter rate. As such, it is

prudent to have an idea of the error in depth scale when the average sputter rate is

used.

With a Cs+ primary ion beam, the range of incident angle where the average

sputter rate can be reliably used is smaller compared to O2+ primary ion beam.

Conclusion

155

8.3 Depth resolution

Another motivation in this study is the need to achieve high depth resolution

with ultralow-energy O2+

and Cs+ primary ion beam. A high depth resolution is

critical to be able to differentiate between thin adjacent layers as in gate oxides and

multi-quantum wells.

With ultralow-energy O2+

primary ion beams, the depth resolution improves as

the primary ion energy is lowered. It is also noted that the range of incident angles

where high depth resolution is achieved is greater with decreasing incident energy.

Thus, high depth resolution is achievable not only at normal incidence but over a

wider range of incidence angle as Ep is reduced. The best depth resolution is observed

at ~ 40o with Ep ~ 250 eV.

Similarly, depth resolution improves when the ultralow-energy Cs+ primary

ion beam incident energy is lowered. The best depth resolution is achieved at ~ 50-

60o. We established that at ultralow-energy the relationship between depth resolution

(decreasing FWHM) and incident angle is linear and gradual with < 50-60o. We also

noticed that when Ep decreases, d does not decrease significantly because the atomic

mixing length do not change very much with a small change in penetration depth

resulting from a decrease in Ep at ultralow-energy. However, when the incident angle

is increased, the decrease in d is obvious. Hence, oblique angles are better at

profiling steeply varying profiles and interfaces. We predict that d may be smaller at

> 70o but only at the near-surface.

Based on the MRI model, we are able to identify the factors that contribute to

poor depth resolution: surface roughening by observing a drop in the peak intensity

and Gaussian peak broadening, and atomic mixing by a broadening of the trailing

Conclusion

156

edge due to an increase in d but without a decrease in peak intensity. We also

confirmed that depth resolution (decrease in FWHM) can be improved by reducing

atomic mixing i.e. by reducing Ep and/or increasing up to 40-60o with Cs

+ primary

ion beam and up to 20-40o with O2

+ primary ion beam. When Ep decreases, it

corresponds to a reduction in atomic mixing. When increases beyond a critical

angle, surface roughening will set in contributing to a poorer depth resolution.

The onset of roughening can also be detected earlier with the MRI model

compared with observing an increase in the matrix 30

Si- profile. Thus, the MRI model

is a more sensitive indicator of the onset of roughening.

A Cs+ primary ion beam gives better depth resolution than O2

+ primary ion

beam at higher Ep but not at ultralow-energy of less than 500 eV due mainly to poor

beam focus. Therefore, poor beam focus is the limiting factor in the improvement of

depth resolution at ultralow energy.

Other methods such as oxygen flooding and sample rotation have been used to

suppress surface roughness and therefore sustain high depth resolution. Oxygen

flooding has been successful only with O2+ primary ion beam. These techniques,

however, have been developed mainly for use with instruments that can operate only

at oblique angles.

8.4 Dynamic range

With an ultralow-energy O2+ primary ion beam, the dynamic range increases

as moves away from the normal with the highest dynamic range achieved at ~ 30-

50o. The dynamic range is about three to four orders of magnitude. Beyond these

incident angles the dynamic range decreases.

Conclusion

157

With a Cs+ primary ion beam, a dynamic range of three orders of magnitude is

possible when profiled at ~ 40-70o.

8.5 Optimum conditions for analysis

From the observed effects of ultralow-energy O2+ and Cs

+ SIMS on silicon, we

recommend different sets of conditions for the analysis of ultrashallow implants and

thin films or multi-quantum well structures which are positioned deeper beneath the

surface. For the profiling of electropositive elements in ultrashallow implants, it is

best to operate with O2+ primary ion beam. To obtain a narrow transient width and

reliable depth conversion by using the average sputter rate, Ep < 500 eV, ~ 0-10o or

Ep ~ 250 eV, ~ 0-20o are recommended.

For electronegative elements, profiling with a Cs+ primary ion beam with Ep <

500 eV, ~ 30-50o and Ep ~ 1 keV, ~ 20-30

o is recommended for a narrow

transient. To achieve a narrow transient and the highest detection sensitivity, it is best

to profile at ~ 30o for Ep < 1 keV. With this combination, the transient width and

sensitivity can be optimized simultaneously.

For applications such as the profiling of multi-quantum well and thin films, a

high depth resolution is essential. Such requirements are best achieved when profiling

at ~ 20o with Ep ~ 1 keV, ~ 30

o with Ep ~ 500 eV and ~ 40

o with Ep ~ 250 eV

O2+ primary ion beam. The dynamic range is almost the maximum achievable. With a

Cs+ primary ion beam, ~ 50

o is recommended with Ep < 500 eV and ~ 60

o with Ep

~ 1 keV without compromising on dynamic range.

Conclusion

158

When both a narrow transient width and a high depth resolution are required,

it is best achieved at ~ 40o with Ep ~ 250 eV O2

+ primary ion beam. The transient

width is marginally worse off than at normal and near normal incidence. This is also

possible with a Cs+ primary ion beam at ~ 50

o with Ep < 500 eV.

8.6 Proposed future work

In the sample used for this study, the first delta-layer is at a depth of 12 nm

and this is the first reference position for an accurate depth determination. The

transient width and sputter rate can be more accurately determined if the first delta-

layer can be nearer the surface. Further work with the first delta-layer nearer the

surface is necessary to determine a more accurate measure of the transient width.

Based on the recommended optimum operating conditions, it is useful to study

the percentage error in the quantification of dose of commonly used ultrashallow

implants such as B, P and As. The same set of samples can also be used to observe the

error in the junction depth compared to the use of alternative methods such as

spreading resistance profilometry (SRP) or carrier illumination method (CI).

Similar studies on other common substrates such as GaN, GaAs and InSb is

also beneficial to observe for similarities or differences when subjected to different

matrices.

The onset of roughening has a deleterious effect on both the surface transient

and depth resolution. More detailed scientific studies with in-situ XPS on ultralow-

energy SIMS will provide a better understanding on the process of surface roughening

especially with Cs+ primary ion beam.

Appendix A

159

Appendix A

Depth profiles with O2+ primary on beam

b) 250 eV, 10o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

a) 250 eV, 0o

70Ge

+

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Ip

Appendix A

160

d) 250 eV, 30o

70Ge

+

Ip30

Si+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

c) 250 eV, 20o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

161

e) 250 eV, 40o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

f) 250 eV, 50o

70Ge

+

Ip30

Si+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

162

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

h) 250 eV, 70o

g) 250 eV, 60o

70Ge

+

Ip30

Si+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

163

j) 500 eV, 10o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

i) 500eV, 0o Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

164

k) 500 eV, 20o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

l) 500 eV, 30o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

165

m) 500 eV, 40o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

n) 500 eV, 50o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

166

o) 500 eV, 60o

70Ge

+

Ip 30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200 250 300Depth (nm)

Inte

ns

ity

(c

ps

)

p) 500 eV, 70o

70Ge

+

Ip

30Si

+44

SiO+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200 250 300Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

167

q) 1 keV, 0o

70Ge

+

Ip

30Si

+44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

r) 1 keV, 10o

70Ge

+

Ip

30Si

+44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

168

s) 1 keV, 20o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

t) 1 keV, 30o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

169

u) 1 keV, 40o

70Ge+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

v) 1 keV, 50o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix A

170

w) 1 keV, 60o

70Ge

+

Ip30

Si+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

x) 1 keV, 70o

70Ge

+

Ip

30Si

+

44SiO

+

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

171

Appendix B

Depth profiles with Cs+ primary ion beam

a)320 eV, 0o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

b) 320 eV,10o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

172

c) 320 eV, 20o

30Si-

Ip

59Si2-

98SiGe-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

d) 320 eV, 30o

30Si

-

Ip59

Si2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

173

e) 320 eV, 40o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

f) 320 eV, 50o

30Si

-

Ip59

Si2-

98SiGe-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

174

g) 320 eV, 60o

30Si

-

Ip59

Si2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

h) 320 eV, 70o

30Si

-

Ip59

Si2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

175

i) 500 eV, 0o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

j) 500 eV, 10o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

176

k) 500 eV, 20o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

l) 500 eV, 30o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

177

m) 500 eV, 40o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

inte

ns

ity

(c

ps

)

n) 500 eV, 50o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

178

o) 500 eV, 60o

30Si

-Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

p) 500 eV, 70o

30Si

-Ip

59Si2

-

98SiGe-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

179

r) 1 keV, 10o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

q) 1 keV, 0o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

180

t) 1 keV, 30o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

s) 1 keV, 20o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

181

u) 1 keV, 40o

30Si

-

Ip

59Si2

-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

v) 1 keV, 50o

30Si

-

Ip59

Si2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

Appendix B

182

w) 1 keV, 60o

30Si

-Ip59

Si2-

98SiGe-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

0 50 100 150 200Depth (nm)

Inte

ns

ity

(c

ps

)

x) 1 keV, 70o

30Si

-

Ip59

Si2-

98SiGe

-

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

0 50 100 150 200

Depth (nm)

Inte

ns

ity

(c

ps

)

ultralow energy sims

Documents