Memorandum

CHARACTERISATION OF THE GAS-PHASE ENVIRONMENT IN A HOT FILAMENT DIAMOND CHEMICAL VAPOUR DEPOSITION CHAMBER USING MOLECULAR BEAM MASS SPECTROMETRY.

Roland S. Tsang

A thesis submitted to the University of Bristol in accordance with the requirements for the degree of Doctor of Philosophy in the Faculty of Science, Department of Physical Chemistry.

August 1997

ABSTRACT

Molecular beam mass spectrometry has proved to be capable of providing quantitative measurements of both stable free radical gas-phase species under conditions typical in the diamond CVD process. Due care has to be taken, however, in the data reduction procedures because the overall system sensitivity is critically dependent on the local temperature, pressure and composition of the gas being sampled. Simple empirically based correction procedures can offset such variations in the sampling efficiency. Armed with a sensitive gas-phase analysis technique and the necessary data reduction procedures for the characterisation of a CVD environment, attention was turned towards the study of the effects that different gas additions to the standard hydrocarbon/H₂ gas mixtures have on the gas-phase reaction mechanism occurring during the CVD growth process.

First, an understanding in the behaviour and applicability of chlorine-assisted CVD was accomplished in a hot filament reactor. Quantitative measurements of the composition of the gas-phase species prevailing during diamond deposition were obtained for a variety of chlorine containing source gases. Two forms of gas mixtures were used: (1) 1% of a chlorinated methane (CH_4-xCl_x, x=1-4) in H₂ and (2) 1% CH₄ in H₂, then introducing chlorine varying from 1-4%. At filament temperatures at and above the optimum for diamond growth (~2300C), the relative concentrations of the various hydrocarbon species (CH₄, C₂H₂, C₂H₄) in the gas phase are insensitive to the choice of Cl precursor used. Furthermore, the product distribution at these temperatures is remarkably similar to that measured when CH₄ is the precursor. For both forms of Cl precursor gases used, chlorine is effectively reduced to HCl at standard growth temperatures, whose concentration is proportional to the chlorine fraction input in the feed gas. The as-grown diamond films were analysed using AES and SEM techniques. No chlorine was detected in bulk structure of the films but ~1-2% was found on the diamond surface. The apparent catalytic activity of Cl atoms in the CVD process is therefore likely to be due to its role in abstracting surface terminating hydrogen or H abstraction of surface C-Cl, at lower substrate temperatures.

Next, the effects of nitrogen on the CVD diamond growth mechanism was examined. The deposition rate at optimum growth conditions depends critically on the choice of C/N precursor used, and the origin of the carbon-containing species. The reactions occurring in the gas-phase seem to lead predominantly to the formation of HCN, (except for CH₄/N₂ gas mixtures). The stability of this species precludes most of the cycling of carbon during the CVD process, resulting in low rates of diamond deposition. Thermodynamic equilibrium calculations confirm that HCN production is highly favoured in H/C/N gas mixtures at high gas processing temperatures. For a 1:1 C:N ratio in the feed gas, continuous films were produced after 6 hour deposition only by CH₄/H₂/N₂ gas mixtures. Incorporation of nitrogen in the grown diamond films was very low, consistent a theoretical calculations. At lower temperatures N₂ simply acts as a spectator to the CVD process, as evidenced by the significant increase in the C₂H₂ concentration and reduction in the HCN concentration in the gas-phase compared to other N source gas additions. At optimum filament temperatures (~2400°C), addition of N₂ to a CH₄/H₂ gas mixture leads to higher deposition rates of poor quality diamond films (determined by LRS). We believe that this can be explained if N₂ is acting as a catalyst for the destruction of H atoms, thereby reducing the etching rate of non-diamond phases on the film surface. Thus addition of a tiny amount of N₂ to the hot filament CVD process will affect not only the gas-phase chemistry, but the growth rate, the morphology and the quality of the resulting diamond films.

The effects of addition of phosphine on the growth behaviour of diamond films have also been investigated. Films were grown using gas mixtures of 1% CH₄ with increasing amounts of PH₃ (1000-5000 ppm). Gas phase species prevalent during the growth process (e.g. CH₄, CH₃, C₂H₂, PH₃ and HCP) have been monitored, quantitatively, and compared with the corresponding growth rates, quality and properties of the resulting films. Addition of up to 2000 ppm PH₃ produced good quality diamond films at the highest growth rates, and changed the crystal morphology in favour of [100]. At higher PH₃ concentrations (3000-5000 ppm) the growth rate decreases again, with predominantly [111] faceted crystals and a compromise in film quality. These observations can be rationalised by the rapidly cycling between methyl radicals and HCP molecules. At low PH₃ input, an unusually high CH₃ mole fraction was observed due to additional [H] produced as a result of rapid decomposition of PH₃ to P and 3H, lending to fast deposition rates. At high PH₃ input, a reversal in growth rates was observed, which we suggest may be due to another, competing reaction, which instead serves to deplete [CH₃].

Finally, as a means to test the validity of the MBMS data we have also employed a highly structured computer package called CHEMKIN, using the SPIN application code to aid in the incorporation of complex gas-phase chemical reaction mechanisms into numerical simulations. For the first time we have performed a simulation on a 1% CH₄ in H₂ mixture as a function of filament temperature. We find that the mole fractions predicted by SPIN for the various hydrocarbon species (CH₄, C₂H₂, C₂H₄) in the gas phase depends critically on the amount of [H] introduced in the initial SPIN input file, for any particular filament/substrate temperature. Filament poisoning effects (which reduce the effective concentration of H atoms produced at the filament) have been taken into account in choosing an appropriate starting [H]. At growth temperatures, Dandy and Coltrin predicted an H atom concentration of 5% at the filament (having corrected for filament poisoning effects) which was the value we used in the input with 1% CH₄ in H₂. Subsequent simulation of this gas mixture under standard growth conditions produced calculated mole fractions of CH₄, C₂H₂, and C₂H₄ that are accurate to within a factor of 1-2 compared to the measured values. Different H atom concentrations were introduced to the input file for different filament temperatures, and the results obtained again were in good agreement with those measured by MBMS.

ACKNOWLEDGEMENTS

Firstly, I would like to thank my supervisor, Professor Mike Ashfold, for his ever-enthusiastic and valued advice during my postgraduate studies.

My very special thanks go to Dr. Paul May, for not only being a constant source of encouragement, inspiration and efficient proof-reading over the past three years, but also for his excellent humour and treasured friendship.

I would also like to thank Dr. Christopher Rego for his support and expertise during my first year PhD, and whose work has really set the foundations not only for my project work, but for our understanding of this area of chemistry.

I am grateful to Mr. Keith Rosser for his technical support and advice throughout my experimental work, to John Cole for successfully setting up the CHEMKIN computer package, and to Dr. Thomas Badgwell of the Chemical Engineering Department at Rice University for advice on the many aspects of the package. My special thanks also goes to Dr. Jim Butler of the Naval Research Laboratory not only for valued discussions on my research, but also a most delicious tub of chilli peanut butter.

Thanks go to Mr. John Dimery for the use of the scanning electron microscope (SEM), to Dr. Charles Younes for all the Auger analyses, and to Mr. David Jones for developing outrageously large quantities of high quality SEM photos presented within this thesis. A special acknowledgement must go to Mr. Tim Davis form the Physics Department. Thanks dude for all your help and advice with these seemingly endless Raman curve fittings and data collections, but mostly for your patience and good humour, especially on those ‘Friday afternoons’.

I am indebted to the legendary “BUDGIES”, namely Paul, Stuart, Tim, Dave, Annette, and Stefan, our temporary German virgin addition, for making the last three years so unforgettable, providing a most sparkling research atmosphere and for many of those ‘it’ll be rude not to’ pub sessions.

To my family and friends, a big thank you for all of your support during my years as a student. Finally an extra special mention goes to Vicks who never ceased to believe in me. Thank you for all your care and support during the so often underestimated task of ‘writing-up’.

“To win them, temples have been profaned, palaces looted, thrones torn to fragments, princes tortured, women strangled, guests poisoned by their hosts, and slaves disembowelled. Some have fallen on battlefields, to be picked up by ignorant freebooters, and sold for a few silver coins. Others have been cast into ditches by thieves or swallowed by guards, or sunk in shipwrecks, or broken into powder in moments of frenzy. No strain of fancy in an Arabian tale has outstripped the marvels of fact in the diamond's history.”

Gardner Williams (General Manager of de Beers ca.1890's)

“The BUDGIES are getting rowdy…”

Jim Butler, NRL, Diamond Conference Meal, Tours, 1996.

“They resist blows to such an extent that the hammer rebounds and the very anvil splits asunder, but this invincible element which defies Natures two most violent forces, iron and fire, can be broken by ram's blood. But it must be steeped in blood that is fresh and warm and even so, many blows are needed.”

Pliny the Elder ca. 1 century AD.

“Je suis vraiment desolé…”

The BUDGIES, Diamond Conference, Tours, 1996.

To mum, dad and sis.

Memorandum

The research described in this thesis was carried out by the author in the School of Chemistry at the University of Bristol under the supervision of Professor MNR Ashfold, Dr. P.W. May and Dr. C.A. Rego.

The work reported herein is original to the author, except where acknowledged by reference or special recognition. No part of this work has been submitted previously for any degree.

CONTENTS

1. INTRODUCTION

1.1 Diamond 1

1.2 Historical overview of the CVD diamond process 3

1.3 The diamond CVD technique 4

1.4 The CVD diamond film 6

1.5 The choice of substrates suitable for growing CVD diamonds 8

1.6 Mechanism of CVD growth 9

1.7 Role of atomic hydrogen in the CVD process 11

1.8 Gas-phase chemistry involved in the CVD process 12

1.9 Characterisation of the CVD process 13

1.10 Summary of the work performed on hot filament CVD systems. 14

1.11 References 17

2. ANALYTICAL TECHNIQUES

2.1 Introduction 20

2.2 Scanning Electron Microscopy (SEM) 21

2.3 Laser Raman Spectroscopy (LRS) 21

2.4 Auger Electron Spectroscopy (AES) 23

3. EXPERIMENTAL

3.1 Introduction 25

3.2 Hot filament CVD reactor 26

(a) The filament 26

(b) The substrate 27

(d) Gas flow system 30

3.3 Molecular beam mass spectrometry of diamond CVD 34

3.4 Molecular beam mass spectrometer design 35

(a) Choice of sampling cone orifice diameters 38

(b) Two stage differential pumping 39

3.5 Characterisation of the mass spectrometer 50

(a) Energy scale calibration 50

(b) MS calibration 51

(d) Temperature dependence of MS sampling efficiency 54

(e) Thermal diffusion effects 56

(f) Dissociation patterns 56

(g) Ionisation cross sections and potentials 57

(h) Detection of radical species 58

(i) Procedure for obtaining quantitative measurements of CH₃ radicals 59

3.6 Step-by-step procedure for converting MBMS raw data into species

mole fractions 62

(a) Stable gas phase species 62

(b) Methyl radicals 66

3.7 References 72

4. RESULTS FOR C/H SYSTEMS

4.1 Introduction 74

4.2 Cracking patterns of CH₄ 75

4.3 Cracking patterns of C₂H₂ 76

4.4 Gas phase composition as a function of filament temperature

for 1% CH₄ in H₂ 77

4.5 Discussion of errors 78

4.6 Gas composition as a function of filament temperature

for a variety of hydrocarbon precursor gases 79

4.7 Analysis of the films grown using 0.5% and 1% CH₄ in H₂ 79

4.8 Appendix: Experimental Data 83

5. RESULTS FOR C/H/Cl SYSTEMS

5.1 Introduction 87

5.2 Experimental Details 91

(a) Deposition experiments 91

(b) Film analysis 91

5.3 Analysis of the diamond films 94

5.4 Gas composition versus filament temperature for a variety of chlorine

containing precursor gases in H₂ 99

(a) Chloromethane (1% CH₃Cl in H₂) 99

(b) Dichloromethane (1% CH₂Cl₂ in H₂) 99

(d) Tetrachloromethane (1% CCl₄ in H₂) 100

5.5 Gas composition versus filament temperature for various CH₄/Cl₂/H₂

containing gas mixtures 104

5.6 Gas composition versus filament temperature using 1% CF₄ in H₂ 105

5.7 Discussion 106

5.8 References 112

5.9 Appendix 114

(I) Ionisation potentials (I.P.) of the various precursor gases used 114

(II) Experimental Data 114

6. RESULTS FOR C/H/N SYSTEMS

6.1 Introduction 122

6.2 Experimental Details 124

(a) Deposition experiments 124

(b) Film analysis 125

6.3 Analysis of the diamond films 126

(a) Methane and Ammonia as precursor gas mixture 126

(b) Methylamine as precursor gas mixture 128

(d) Methane and nitrogen as precursor gas mixture 132

6.4 Gas composition versus filament temperature for a variety

of C-/N-containing precursor gases in H₂ 139

(a) Methane and Ammonia as source gas mixture 139

(b) Methylamine as source gas mixture 144

(d) Methane and nitrogen as source gas mixtures 145

6.5 Conclusions 152

6.6 References 154

6.7 Appendix 156

(I) Synthesis of Hydrogen Cyanide 156

(II) Ionisation potentials and the user selected electron energies

of the various gas-phase species monitored 157

(III) SEM Photo Library 158

(IV) Experimental Data 166

7. RESULTS FOR C/H/P SYSTEMS

7.1 Introduction 175

7.2 Experimental Details 176

(a) Deposition experiments 176

(b) Film analysis 177

(d) Cracking patterns of PH₃ 177

7.3 Analysis of the diamond films 179

7.4 Gas-phase composition measurements 182

7.5 Discussion 185

7.6 References 189

7.7 Appendix 190

(I) Ionisation potentials and the user selected electron energies

of the various gas-phase species monitored 190

(II) SEM Photo Library 191

(III) Laser Raman spectra (uv - 325 nm) of films grown using 1% CH₄

in PH₃/H₂, the amount of phosphine varying from 0.1%-0.5%. 205

(IV) Experimental Data 211

8. CHEMKIN SIMULATIONS

8.1 Introduction 226

8.2 Structure of CHEMKIN 227

8.3 Structure of the Transport Property Fitting Code

and SURFACE CHEMKIN 232

8.4 Application codes 237

(a) SENKIN 238

(b) SPIN 241

8.5 Numerical simulation of the diamond CVD process 246

8.6 Numerical simulation of the gas-phase composition vs. filament temperature

for 1% CH₄ in H₂ at 20 Torr 251

8.7 Filament poisoning effects 265

8.8 References 272