study the structural properties of ge nanoparticles …...study the structural properties of ge...

Study the structural properties ofGe nanoparticles formed by ion

implantation and thermalannealing in nitride-base dielectric

matrices

Sahar Mirzaei

A thesis submitted for the degree ofDoctor Of Philosophy

The Australian National University

March 2016

c© Sahar Mirzaei 2011

Except where otherwise indicated, this thesis is my own original work.

Sahar Mirzaei18 March 2016

To my best friend Dan and my lovely family and to those whom will read thisthesis.

Acknowledgments

It was almost 4 years ago that I decided to leave my country and move to Australiato take up a PhD in Canberra. Despite all the ups and downs of my experience I amdelighted that I did.

I would like to thank Prof. Mark Ridgway (dec.) and Prof. Robert Elliman fortheir contributions to this work. Thanks to my supervisory panel, Dr Felipe Kremerwho helped me to find my tiny nanoparticles in a dark microscopy room, and DrDavid Sprouster for his knowledge on synchrotron data analysis and his great ad-vice.

I would also like to extend my thanks to Dr Chris Glover and Dr Bernt Johan-nessen at the Australian Synchrotron for their wisdom and guidance.

A massive thanks to Dan for his support. I could not have managed this workwithout his friendship.

I am also deeply grateful to my lovely friends who helped me over the past fewyears; David L, Kidane, Thomas, Jess, Jenny, Roisin and Andy. It has been a pleasureto work with you.

Lastly and most importantly, thanks Mum, Dad and my lovely sisters for theirunconditional love throughout my life.

vii

Publications

• S. Mirzaei, F. Kremer, D. Sprouster, L. Araujo, R. Feng, C. Glover, M. Ridg-way, Formation of Ge nanoparticles in SiOxNy by ion implantation and thermalannealing. J. Appl. Phys. 118, 15 (2015).

• R. Feng, F. Kremer, D. J. Sprouster, S. Mirzaei, S. Decoster, C. J. Glover, S. A.Medling, J. L. Hansen, A. Nylandsted-Larsen, S. P. Russo, and M. C. Ridgway,Enhanced electrical activation in In-implanted Si0.35Ge0.65 by C co-doping, sub-mitted to Matt. Res. Lett., (2016).

• R. Feng, F. Kremer, D. J. Sprouster, S. Mirzaei, S. Decoster, C. J. Glover, S. A.Medling, J. L. Hansen, A. Nylandsted-Larsen, S. P. Russo, and M. C. Ridgway,Electrical and structural properties of In-implanted Si1xGex alloys, J. Appl. Phys.119, 025709 (2016).

• R. Feng, F. Kremer, D. J. Sprouster, S. Mirzaei, S. Decoster, C. J. Glover, S.A. Medling, S. P. Russo, and M. C. Ridgway, EXAFS study of the structuralproperties of In and In + C implanted Ge, accepted by J. Phys.: Conf. Ser.,(2016).

• R. Feng, F. Kremer, D. J. Sprouster, S. Mirzaei, S. Decoster, C. Glover, S. Medling,S. Russo, M. Ridgway, Enhanced electrical activation in In-implanted Ge by Cco-doping. Appl. Phys. Lett. 107 212101 (2015).

• R. Feng, F. Kremer, D. J. Sprouster, S. Mirzaei, S. Decoster, C. Glover, S. Medling,S. Russo, M. Ridgway, Structural and electrical properties of In-implanted Ge.J. Appl. Phys. 118 165701 (2015).

ix

Abstract

This thesis investigates the formation of Ge nanoparticles (NPs) in amorphous Si3N4

and SiOxNy by ion implantation and thermal annealing. The structural propertiesof the NPs were determined using a combination of laboratory and synchrotronbased techniques including cross-section transmission electron microscopy (TEM),x-ray diffraction (XRD), Rutherford backscattering spectrometry (RBS), Raman spec-troscopy measurements and x-ray absorption spectroscopy (XAS).

The motivation for this research is that NPs synthesized from Group IV semi-conductors, including Ge, show potential for novel electronic and optoelectronic de-vices. Ge was chosen as the material for this study, mainly due to its very large bulkexciton Bohr radius (Rb ∼10 nm) and consequently small R/Rb ratio (where R isthe NP radius), which increases the quantum confinement regime for optoelectronicapplications. Also, Ge NPs embedded in thin dielectric films have exhibited impres-sive charge storage capabilities, useful in non-volatile memory (NVM) applications.While many studies have focused on the SiO2 matrix, there have been few experi-mental studies on the growth of Ge NPs in a silicon nitride and silicon oxynitridematrices.

The main focus of this project was on the formation of Ge NPs in different Si3N4-based matrices and examining the short range atomic structure of embedded NPs asfunction of NP size using extended x-ray absorption fine structure (EXAFS). Specifi-cally, four different matrices have considered: PECVD Si3N4, LPCVD Si3N4, PECVDSiO1.67N0.14 and PECVD SiO1.12N0.37. Size evolution and structural properties of NPswere examined as function of implantation conditions and the host matrix.

Ge NPs were formed in plasma enhanced chemical vapour deposition (PECVD)Si3N4. Precipitations occurred for Ge concentrations of ≥ 6 at.%, which suggeststhe solubility limits of this order. NP size was influenced by Ge concentration andannealing temperature. NP size increased from 2.4 to 4 nm - for samples annealedat 900 ◦C for 1 hourwhen concentration was increased from 9 to 12 at.%. Moreover,as annealing temperature was increased from 700 to 900 ◦C NP size increased from 3to 4.5 nm (for 12 at.% Ge samples). In general NP diameters were much small com-pared to SiO2 matrix due to the N content of the system, low diffusivity and largeinterfacial energy between Ge atoms and Si3N4-based matrices.

Unlike PECVD Si3N4, ion beam synthesis of Ge in LPCVD Si3N4 layers resultedin the formation of SiGe NPs. NP size altered from 2.8 to 3.2 nm for 12 at.% Geconcentration samples after annealing temperature was increased from 700 to 900

xi

xii

◦C. Also, NP size increased from 2.3 to 3.2 nm for samples annealed at 900 ◦C for 1hour when Ge concentration was increased from 9 to 12 at.%. Complementary XAStechniques (in particular EXAFS), allowed us a precise and better understanding ofthe structure of embedded SiGe NPs. Since the lattice difference between Ge andSiGe NPs (less than 4%) could not be distinguished by TEM.

A phase transition of the matrix was observed for LPCVD and PECVD Si3N4

matrices after annealing at 1100 ◦C. This significantly lowered crystallization tem-perature of un-implanted layers (1600-1800 ◦C). Formation of nanosilicide particles,change of stoichiometry and implantation induced-damage are found to be the mostlikely causes of crystallization.

Ge NPs were formed in SiO1.67N0.14 matrix for different Ge concentrations andannealing temperatures. For Ge concentrations from 9 to 12 at.% annealed at 900 ◦Cfor 1 hour, NP size increased from 3.7 to 4.5 nm. Also, when annealing temperaturewas increased from 700 to 900 ◦C NP size increased from 4.1 to 4.5 for 12 at.% Geconcentration. We observed that slight amount of the N in the system had a sig-nificant effect on NP size similar to PECVD Si3N4. Unlike other examined systemscavities were formed near the implanted region in SiO1.12N0.37 matrix for differentconcentrations and annealing temperatures, which could be correlated to the matrixstructure and composition or vacancy events after ion implantation. The size, originand properties of formed cavities are subjects of future work. No phase transition(or long-range diffusivity of Ge atoms) was observed for SiOxNy matrices after an-nealing at 1100 ◦C.

A brief highlights of this research can be summarized as follow:

• No NPs observed for Ge concentrations less than 6 at.% - suggests solubilitylimit is of this order.

• NP size appreciably changed by increasing Ge concentration - up to 50% in-crease in size for samples annealed at 900 ◦C for 1 hour when concentrationincreased from 9 to 12 at.%.

• TEM results indicate NPs were commonly formed upon annealing and anneal-ing temperature had a slight effect on NP size - up to 20% increase in size forsamples with Ge concentration of 9 at.% annealed when annealing temperatureincreased from 700 to 900 ◦C.

• NP size is much small in Si3N4 and SiOxNy matrices - due to presence of N(provides a means of tuning NP size) .

• For approximately similar Ge concentrations, different type of NPs were formedin LPCVD and PECVD Si3N4 - due to the stoichiometry variation of LPCVDand PECVD layers.

xiii

• The crystallization temperature of LPCVD and PECVD Si3N4 (1600-1800 ◦C) islowered to 1100 ◦C by ion implantation with Ge and other species.

• Voids and cavities are formed near the implanted region of SiO1.12N0.37 layers.

Contents

Acknowledgments vii

Abstract xi

1 Introduction 11.1 Historical overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Production techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Applications of Ge nanoparticles . . . . . . . . . . . . . . . . . . . . . . . 31.4 Applications of GeSi nanoparticles . . . . . . . . . . . . . . . . . . . . . . 51.5 Why embedded nanoparticles : Interaction with a matrix . . . . . . . . . 61.6 Si3N4 growth process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.6.1 LPCVD Si3N4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.6.2 PECVD Si3N4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.6.3 Comparison between PECVD and LPCVD Si3N4 . . . . . . . . . 91.6.4 PECVD SiOxNy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.7 Synthesis of nanoparticles: NPs formed in Si3N4 and SiOxNy . . . . . . 101.8 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.9 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Formation and characterization techniques 152.1 Ion implantation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.1.1 Ion accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.1.2 Nucleation and growth process . . . . . . . . . . . . . . . . . . . . 192.1.3 Thermal annealing . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.2 Laboratory based techniques . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.1 Rutherford backscattering spectrometry . . . . . . . . . . . . . . . 202.2.2 Transmission electron microscopy . . . . . . . . . . . . . . . . . . 21

2.2.2.1 TEM sample preparation method . . . . . . . . . . . . . 212.2.2.2 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.2.3 Raman spectroscopy measurements . . . . . . . . . . . . . . . . . 232.2.4 X-ray diffraction measurements . . . . . . . . . . . . . . . . . . . 23

2.3 Synchrotron-based techniques . . . . . . . . . . . . . . . . . . . . . . . . . 242.3.1 X-ray absorption spectroscopy . . . . . . . . . . . . . . . . . . . . 242.3.2 Basic principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252.3.3 XANES region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252.3.4 EXAFS region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3.5 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

xv

xvi Contents

2.3.6 XAS sample preparation technique . . . . . . . . . . . . . . . . . 312.3.7 XANES/EXAFS data analysis . . . . . . . . . . . . . . . . . . . . . 31

3 Formation of GeSi nanoparticles in LPCVD Si3N4 353.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.2 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.3 Results and discussion for low concentration samples . . . . . . . . . . . 373.4 Summary- Low Ge concentrations (≤4 at.%) . . . . . . . . . . . . . . . . 443.5 High Ge concentration samples (> 4 at.%) . . . . . . . . . . . . . . . . . 44

3.5.1 RBS measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 443.5.2 XRD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.5.3 TEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.5.4 XANES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.5.5 EXAFS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513.5.6 Size-dependent atomic structure . . . . . . . . . . . . . . . . . . . 53

3.6 Summary- high Ge concentrations (>4 at.%) . . . . . . . . . . . . . . . . 54

4 Formation of Ge nanoparticles in PECVD Si3N4 574.1 Experimental procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . 584.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4.2.1 RBS and XRD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584.2.2 TEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624.2.3 Raman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624.2.4 XANES measurements . . . . . . . . . . . . . . . . . . . . . . . . . 664.2.5 EXAFS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674.2.6 Size-dependent atomic structure . . . . . . . . . . . . . . . . . . . 68

4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5 Formation of Ge NPs in SiOxNy matrices of different stoichiometry 715.1 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725.2 Formation of Ge nanoparticles in SiO1.67N0.14 . . . . . . . . . . . . . . . . 73

5.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735.2.2 RBS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735.2.3 TEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745.2.4 XANES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765.2.5 Raman measurements . . . . . . . . . . . . . . . . . . . . . . . . . 765.2.6 EXAFS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 775.2.7 Summary of SiO1.67N0.14 matrix . . . . . . . . . . . . . . . . . . . 79

5.3 Formation of Ge nanoparticles in SiO1.12N0.37 . . . . . . . . . . . . . . . . 805.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805.3.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . 80

5.3.2.1 RBS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805.3.2.2 Raman measurements . . . . . . . . . . . . . . . . . . . . 805.3.2.3 TEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

Contents xvii

5.3.2.4 XANES . . . . . . . . . . . . . . . . . . . . . . . . . . . . 845.3.2.5 EXAFS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855.3.2.6 Structural properties . . . . . . . . . . . . . . . . . . . . 86

5.3.3 Summary of SiO1.12N0.37 matrix . . . . . . . . . . . . . . . . . . . 88

6 Summary and Conclusion 896.1 This work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 906.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

A Crystallization of the matrix 97A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98A.2 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98A.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

A.3.1 Implantation-temperature dependence . . . . . . . . . . . . . . . 99A.3.2 Concentrations dependence . . . . . . . . . . . . . . . . . . . . . . 100A.3.3 Species dependence . . . . . . . . . . . . . . . . . . . . . . . . . . 102A.3.4 Annealing-temperature dependence . . . . . . . . . . . . . . . . . 104A.3.5 Annealing-time dependence . . . . . . . . . . . . . . . . . . . . . 104A.3.6 Ambient dependence . . . . . . . . . . . . . . . . . . . . . . . . . 105

A.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

B H implantation effect 109B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110B.2 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110B.3 Result and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

C XANES spectra of crystalline standards 113

D Quantitative analyse of Raman measurements 115

xviii Contents

List of Figures

1.1 The Lycurgus Cup is a 4th century Roman glass cage cup. The colorchanges when the source of light varies from transmission to reflection. 2

1.2 Room temperature PL from Ge NPs (a) embedded in SiO2 with differ-ent sizes [1], and (b) NPs with different size in different matrices [1].

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.3 Room-temperature EL spectra of (a) 3 nm Ge NPs after annealing at

800 ◦C and (b) 6.1 nm Ge NPs after annealing at 950 ◦C embedded inAl2O3 MIS structure for different injected currents.(adapted from [2]) . 5

1.4 (a) TEM images of a sample after annealing at 475 ◦C, (b) PL spectraof samples after annealing at different temperatures. (adapted from [3]) 11

1.5 STEM images of Ge NPs formed in (a) Si3N4 , (b) SiO2. (adapted from[4]) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1 Electronic and nuclear stopping power of aluminum ions in aluminum,versus particle energy per nucleon. . . . . . . . . . . . . . . . . . . . . . 16

2.2 SRIM simulation for Ge ions profiles in a Si3N4 target. . . . . . . . . . . 172.3 Schematic digram of a 1.7 MV tandem accelerator at the Department

of Electronic Materials Engineering, Australian National University. . . 182.4 Scheme of ion-beam synthesis of nanoparticles [5]. . . . . . . . . . . . . 192.5 Schematic diagram of the RBS principle. . . . . . . . . . . . . . . . . . . 212.6 Schematic of sample preparation produced for cross section transmis-

sion electron microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.7 Schematic diagram of glancing-incidence XRD set up adopted from [6]. 232.8 Schematic of the XAS beamline optics at the Australian Synchrotron. . . 242.9 Different parts of an x-ray absorption spectra as a function of photo-

electron energy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.10 Schematic of a photoelectron multiple scattering. . . . . . . . . . . . . . 272.11 Schematic of the EXAFS process illustrating the origin of EXAFS os-

cillations due to the interference of outgoing and backscattered photo-electron wave. The red line is the background. . . . . . . . . . . . . . . . 28

2.12 Schematic of the experimental setup for EXAFS measurements at theAustralian Synchrotron. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.13 Schematic diagram of EXAFS sample preparation method. . . . . . . . . 312.14 Schematic of the EXAFS data analysing(a) energy calibration (b) back-

ground removal, (c) k3-weighting of χ(k) and (d) Fourier transforming. 32

xix

xx LIST OF FIGURES

3.1 RBS spectra of 4 at.% samples implanted with Ge for as-implanted andannealed at 1100 ◦C. The inset shows the Ge distribution as a functionof Si3N4 depth. Samples were implanted at -196 ◦C. . . . . . . . . . . . . 38

3.2 RBS spectra for all 4 at.% samples as a function of implanting temper-ature. All implanted samples were annealed at 1100 ◦C. . . . . . . . . . 39

3.3 XRD pattern for 4 at.% Ge samples as a function of implanting tem-perature (offset for comparison). Samples were annealed at 1100 ◦C.The diffraction peaks of alpha Si3N4 are illustrated with star symbols. . 40

3.4 TEM images of 4 at.% samples implanted at -196 ◦C after annealing at1100 ◦C: a) Si/Si3N4 interface and b) implanted region. . . . . . . . . . . 41

3.5 Raman spectra of 4 at.% samples after annealing at 1100 ◦C as a func-tion of implanting temperatures, and crystalline GeSi standard. . . . . . 41

3.6 XANES spectra of 4 at.% samples, after annealing at 1100 ◦C, as afunction of implanting temperature plus amorphous and crystallineGeSi standards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.7 EXAFS spectra of 4 at.% samples, after annealing at 1100 ◦C, as a func-tion of implanting temperature plus amorphous and crystalline GeSistandards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

3.8 (a) RBS spectra for 12 at.% sample before and after annealing at 1100◦C and (b) Ge as a function of Si3N4 depth. . . . . . . . . . . . . . . . . 45

3.9 XRD pattern of Ge-implanted samples of different concentrations (af-ter annealing at 1100 ◦C/1 hr) with alpha Si3N4 pattern. . . . . . . . . . 46

3.10 TEM images of NPs: 9 and 12 at.% Ge samples after annealing at 700◦C and representative size distributions. . . . . . . . . . . . . . . . . . . . 48

3.11 TEM images of NPs: 9 and 12 at.% Ge samples after annealing at 900◦C and representative size distributions. . . . . . . . . . . . . . . . . . . 49

3.12 TEM images for 12 at.% samples annealed at 1100 ◦C : (a) the wholelayer, (b) a high magnification image of the implanted region (c) im-planted region with crystalline NPs, and (d) diffraction pattern of theimplanted region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.13 XANES spectra at Ge K-edge for crystalline and amorphous standards,9 and 12 at.% samples annealed at 700 and 900 ◦C. Spectra are shiftedfor clarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.14 EXAFS oscillations obtained after background removal, multiplied byk3 for bulk standard and GeSi NP samples. (a), (c) and (e) k3-weightedEXAFS and corresponding non-phase-corrected FT spectra (b), (d) and(f) for crystalline standards, 9 and 12 at.% samples after annealing at700 and 900 ◦C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4.1 RBS spectra for 9 and 12 at.% Ge samples as a function of energyplus the depth distribution of different concentrations and annealingtemperatures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

LIST OF FIGURES xxi

4.2 TEM images of 12 at.% Ge samples after annealing at (a) 900 ◦C and (b)1100 ◦C, respectively. Ge profiles determined from RBS measurementsare included for reference. . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.3 TEM images for different Ge concentrations after annealing at 1100 ◦Cplus corresponding XRD pattern of examined samples. . . . . . . . . . . 61

4.4 TEM images recorded near the Ge concentration maximum for all at.%Ge samples after annealing at 700 ◦C with respective NP size distribu-tions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.5 TEM images recorded near the Ge concentration maximum for all at.%Ge samples after annealing at 900 ◦C with respective NP size distribu-tions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.6 Raman spectra for 6, 9 and 12 at.% Ge samples as a function of anneal-ing temperature (700, 900 and 1100 ◦C) plus crystalline and amorphousstandards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

4.7 XANES spectra for 6, 9 and 12 at.% Ge samples as a function of an-nealing temperature (700 and 900 ◦C) plus crystalline and amorphousstandards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.8 XANES spectra at the Ge K-edge for 12 at.% samples (a) annealed at700 and (b) annealed at 900 ◦C with linear combination fits to experi-mental data. (c) fitted crystalline fraction as a function of NP size. . . . 67

4.9 EXAFS (c and e) and FT (d and f) spectra for 9 and 12 at.% Ge samplesas a function of annealing temperature (700 and 900 ◦C) plus crys-talline and amorphous standards. . . . . . . . . . . . . . . . . . . . . . . 68

5.1 RBS spectra before and after annealing at 1100 ◦C for samples of 12at.% Ge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.2 TEM images recorded near the Ge concentration maximum for 9 and12 at.% Ge samples annealed at 900 ◦C and their representative sizedistributions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.3 TEM images recorded near the Ge concentration maximum for 9 and12 at.% Ge samples annealed at 1100 ◦C and their representative sizedistributions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

5.4 XANES spectra for 9 and 12 at.% Ge samples as a function of annealingtemperatures (700, 900 and 1100 ◦C) plus crystalline and amorphousstandards. The spectra are shifted for clarity. . . . . . . . . . . . . . . . 77

5.5 Raman spectra for 9 and 12 at.% Ge samples as a function of annealingtemperature (700, 900 and 1100 ◦C) plus crystalline standard. Thespectra are shifted for clarity. . . . . . . . . . . . . . . . . . . . . . . . . . 77

5.6 EXAFS (c and e) and FT (d and f) spectra for 9 and 12 at.% Ge samplesas a function of annealing temperature (700, 900 and 1100 ◦C) pluscrystalline and amorphous standards. The spectra are shifted for clarity. 78

5.7 RBS spectra for the 12 at.% Ge samples as a function of annealingtemperature plus the Ge depth profile. . . . . . . . . . . . . . . . . . . . 81

xxii LIST OF FIGURES

5.8 Raman spectra for 9 and 12 at.% Ge samples as a function of annealingtemperature plus crystalline standard. . . . . . . . . . . . . . . . . . . . . 82

5.9 TEM images of a 12 at.% sample after annealing at 900 ◦C, including(a) the whole layer, (b) end side of implanted region, (c) implantedregion and (d) inside the cavities. . . . . . . . . . . . . . . . . . . . . . . . 83

5.10 EDS spectra of a 12 at.% sample after annealing at 900 ◦C and corre-sponding mapping of different spots. . . . . . . . . . . . . . . . . . . . . 84

5.11 XANES spectra of standards as compared to a selection of Ge NP sam-ples.The spectra are shifted for clarity. . . . . . . . . . . . . . . . . . . . . 85

5.12 x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

5.13 Non-phase-corrected FT EXAFS spectra as a function of radial distancecomparing NP samples to the amorphous and crystalline Ge standards. 87

6.1 Size comparison of NPs for given concentrations and annealing tem-peratures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

A.1 XRD pattern of 4 at.% Ge samples implanted at different temperaturesafter annealing at 1100 ◦C and un-implanted layer of LPCVD Si3N4

after annealing at the same temperature. Spectra are shifted verticallyfor clarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

A.2 TEM images of an un-implanted LPCVD layer and 4 at.% Ge samplesafter annealing at 1100 ◦C as function of implantation temperatures. . 100

A.3 XRD pattern of Ge samples after annealing at 1100 ◦C as a function ofconcentration. Spectra are shifted vertically for clarity. . . . . . . . . . . 101

A.4 XRD pattern of Ge implanted PECVD Si3N4 layers after annealing at1100 ◦C as a function of concentration. Spectra are shifted verticallyfor clarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

A.5 TEM images of (a), (b) whole layer plus (b) and (d) implanted regionof LPCVD and PECVD Si3N4 layers, respectively. . . . . . . . . . . . . . 102

A.6 XRD pattern of different species implanted in LPCVD Si3N4. All sam-ples were annealed at 1100 ◦C for 1 hour. Also, un-implanted Si3N4

(after annealing at 1100 ◦C) is shown. Arrows indicate the silicidepeaks. The star symbols refer to the diffraction data of alpha Si3N4.Spectra are shifted vertically for clarity. . . . . . . . . . . . . . . . . . . . 103

A.7 XRD patterns of LPCVD Si3N4 layers implanted with Ge and annealedat different temperatures and alpha structure presented with Star sym-bols. Spectra are shifted vertically for clarity. . . . . . . . . . . . . . . . . 104

A.8 XRD pattern of LPCVD Si3N4 layers implanted with Ge and annealedat 1100 ◦C in N2 atmosphere for 1 and 10 hrs. Poly-crystallizationof the matrix observed for both cases. The dotted purple and bluelines refer to diffraction structures of alpha and beta Si3N4 respectively.Spectra are shifted vertically for clarity. . . . . . . . . . . . . . . . . . . . 105

LIST OF FIGURES xxiii

A.9 XRD pattern of LPCVD Si3N4 layers implanted with Ge implanted.Samples were annealed at 1100 ◦C in different ambient. The star sym-bols show the alpha structure. Spectra are shifted vertically for clarity.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

B.1 H+Ge depth profiles. Sample was implanted with H first and thenimplanted with Ge and the other sample just implanted with Ge of thesame concentration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

B.2 XANES spectra at the Ge K-edge for Ge and H+Ge implanted LPCVDlayers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

B.3 (a) k3-weighted EXAFS spectra and (b) corresponding non-phase-correctedFT spectra of Ge and H+Ge samples. . . . . . . . . . . . . . . . . . . . . . 112

C.1 XANES spectra of crystalline Ge3N4 standard. . . . . . . . . . . . . . . . 113

xxiv LIST OF FIGURES

List of Tables

1.1 Properties of SiO2 and Si3N4 at 300K . . . . . . . . . . . . . . . . . . . . . 81.2 Properties of SiO2 and Si3N4 at 300 K . . . . . . . . . . . . . . . . . . . . 9

3.1 Summary of Ge implantation parameters for low concentration sam-ples. Fluences for each sample are given at corresponding energies E(keV). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.2 Refined EXAFS fitting parameters of 4 at.% implanted at different tem-peratures, plus crystalline and amorphous standards. . . . . . . . . . . . 43

3.3 Summary of Ge implantation parameters. The fluence of each sampleis given at corresponding energies E (keV). . . . . . . . . . . . . . . . . . 44

3.4 The NP size as a function annealing temperature. . . . . . . . . . . . . . 473.5 Structural parameters -coordination number (CN), bond length R, and

Debye-Waller factors (σ2)- obtained for the first NN shell of NPs. . . . . 533.6 Summary of matrix and NP phase as a function of implantation and

annealing temperatures for LPCVD matrix. . . . . . . . . . . . . . . . . . 54

4.1 NP size as a function of annealing temperature. . . . . . . . . . . . . . . 624.2 Structural parameters of NPs as a function of annealing temperature. . 694.3 Summary of matrix and NP phase as a function of implantation and

annealing temperatures for PECVD matrix. . . . . . . . . . . . . . . . . . 70

5.1 The stoichiometry of SiOxNy matrices determined with spectral reflec-tometry and RBS measurements . . . . . . . . . . . . . . . . . . . . . . . 72

5.2 Ge NP size as a function of implantation and annealing temperature. . 755.3 Refined EXAFS structural parameters as a function of implantation

and annealing temperature. Coordination number (CN), bond length(R), Deby-Waller factor (σ2) and third cumulant (C3) parameters arelisted. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

5.4 NP size and structural parameters as a function of implantation andannealing temperature extracted by first NN shell EXAFS analysis fora SiO1.12N0.37 matrix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

6.1 Summary of NP size for Si3N4, SiOxNy and SiO2 matrix. . . . . . . . . . 94

A.1 Type of implanted species and implanting temperatures. . . . . . . . . . 98A.2 The melting points of implanted species and their silicide form . . . . . 103A.3 Bond energy of metal-silicide and GeSi bonds plus N bonds [7, 8] . . . 107

xxv

xxvi LIST OF TABLES

B.1 Summary of Ge and H implantation parameters. . . . . . . . . . . . . . 110B.2 Rifined EXAFS fitting parameters. . . . . . . . . . . . . . . . . . . . . . . 111

C.1 Bond length of first nearest neighbour derived from EXAFS data forcrystalline standards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

C.2 Bond energies for Si- and Ge- bonds[7] . . . . . . . . . . . . . . . . . . . 114

D.1 Peak frequency for the main Raman-active mode (Ge-Ge) for Ge em-bedded NPs in PECVD Si3N4 and SiO1.67N0.14 matrices. The frequen-cies are in units of cm−1. Samples are referred first by their atomicconcentration (at.%) followed by their annealing temperature (◦C) . . . 115

Chapter 1

Introduction

This chapter begins with a general introduction and discussion of the optical proper-ties of semiconductor nanoparticles. This is followed by a comparison of SiO2, Si3N4

and SiOxNy (of different stoichiometry) as suitable candidates for a host matrix. Fur-thermore, it discusses the existence of techniques to deposit an amorphous Si3N4

matrix. Section 1.7 discusses the properties of metalloid and metal nanoparticles em-bedded in aforementioned matrices. Finally, section 1.8 describes the motivation ofthe present research.

1

2 Introduction

1.1 Historical overview

Although nanotechnology’s chief concepts have developed recently, it has been usedfor over a thousand years, albeit unknowingly, in the process of painting and makingsteel. Nowadays, we are able to manipulate and engineer materials at a nanoscaleand change the properties of the nanostructures.

Figure 1.1 illustrates one of the earliest historical examples of the application ofnanotechnology from the Roman empire. The glass has a dichroic nature which isa result of Au nanoparticles (NPs) inserted in its structure. It can be seen that thecolour of the cup changed, from green (as a result of reflective light) into a glowingtranslucent red (as a result of transitive light), when light is transmitted through thecup. Faraday, in 1856, started to study the size dependence on physical propertiesof materials for the first time. He pressed very small pieces of gold (nanocrystallinegold), and smoothed them over the surface of a rock crystal plano convex lens. Dur-ing this process he observed that the colour changed from violet, or dark tint, to abeautiful green [9]. He discovered that the color of a metal is size dependent belowa certain scale. This critical size, and why it differed between metals, he did notknow and could not have discovered [10]. Many years later, the first experimentsproved that this size dependency of the material properties can also be applied tosemiconductors [2].

Figure 1.1: The Lycurgus Cup is a 4th century Roman glass cage cup. The color changes,from red to green, when the source of light varies from transmission to reflection.

1.2 Production techniques

There is a wide variety of techniques for producing NPs. Some common ones in-clude: condensation from a vapour, chemical synthesis, ultrasonic synthesis, me-chanical process such as milling and implantation techniques. Ion implantation is

§1.3 Applications of Ge nanoparticles 3

an established technique commonly used in the semiconductor industry. It is veryversatile and can be used to incorporate nearly any ionic species into a wide rangeof hosts. It is possible to accurately control the number of implanted ions (to controlthe finla NP size) and the NP formation process is thus highly reproducible. It is thetechnique of choice for this thesis and a more through explanation follows in Chapter2.

1.3 Applications of Ge nanoparticles

Semiconductor NPs have properties that can be controlled by their size and struc-ture [2], and host matrices. Ge NPs embedded in a dielectric matrix are of interestdue to their ability to both emit light [11] and store charge [12]. Ge NPs displaysize-dependent structural, electronic and optical properties, which differ significantlyfrom those of their bulk counterparts. These differences stem from the increase insurface-to-volume ratio and quantum confinement effects. Properties that are in-fluenced by the surface (cohesive energy, melting point and average coordinationnumber) scale linearly with the inverse NP diameter. Also, the quantum confine-ment effect predominantly results in unique electronic and optical properties. Asa consequence, Ge NPs are promising candidates for novel optoelectronic, photonicand non-volatile memory devices. Examples of these applications follows.

Nanopaticle based flash memories

Ge and Si NPs embedded in the gate oxide of a field effect transistor have at-tracted considerable attention due to their applications for memory devices [2]. Thesedevices show excellent memory performance and high scalability [13]. The first flashmemory device using Si nanoparticles (embedded in SiO2) were first proposed byTiwari et al [14]. Using NP floating gate can significantly improve the degree of non-volatility and charge retention time, mainly due to the reduction in charge leakagefrom isolated NPs. Compared with Si, Ge shows significant improvement in termsof charge retention memory performance characteristics [15, 16] due to its smallerband gap. The performance of Ge NP based non-volatile memory devices could beimproved by using a high-k dielectric [17, 18]. Ge NPs embedded in Hf-based di-electrics, one of the promising matrices, have been studied only recently [19]. Resultsshowed that embedded Ge NPs could be an ideal candidate for operation in flashmemory devices.

Photoluminescence properties of Ge

Most studies report the photoluminescence (PL) of Ge NPs embedded in SiO2

matrix[20, 21, 22, 23, 11, 24, 25, 26], which is a size dependent property, in the regionof visible and near-infrared. The light emission mechanism from Ge NPs is attributedeither to radiative recombination via quantum confined states [27, 22, 28, 23, 29, 11]

4 Introduction

or to defects at the NP/matrix interface and in the matrix itself [20, 21, 30]. Figure 1.2shows PL spectra of Ge NPs with different sizes, from 2.5 to 10.1 nm, embedded indifferent dielectric matrices [1]. The appearance of PL peaks at 536.8, 589.3 and 658.5nm can be explained in accordance with the quantum size effect, which is due tothe radiative recombination of excitons in quantum confined Ge NPs. Furthermore,peaks at 480.6 nm and 442.8 nm for NP size of 2.5 nm and 5.3 nm, respectively, isattributed to oxygen vacancies, oxygen-germanium vacancy pairs, and related de-fect centres in the SiO2 matrix region or the interface between the SiO2 and Ge NPs.Also, Figure 1.2 (b) shows the effect of the host matrix on the emission of PL peakfor embedded Ge NPs. The different conduction and valence band of SiO2, Al2O3

and HfO2 matrices and carriers confined in Ge NPs is clearly effect the emission ofthe offset peaks around 589.3 nm, 709.1 nm and 770.3 nm, respectively.

Figure 1.2: Room temperature PL from Ge NPs (a) embedded in SiO2 with different sizes(2.5-10.1 nm) [1], and (b) NPs with different size in different matrices (HfO2, Al2O3 and

SiO2)[1].

Nanoparticle based LEDs

After the first demonstration of photoluminescence from semiconductor NPs em-bedded in a SiO2 matrix, the possibility of fabricating LEDs based on these structuresreceived much interest. The size dependency of electroluminescence (EL) spectra ofGe NPs has been studied [31]. For example, Shen et al [32] has reported on theelectroluminescence peak at 590 nm in metal-insulator-semiconductor (MIS) struc-tures but concluded that the size dependent electroluminescence peak could haveoriginated from a defect-mediated emission. Figure 1.3 (a) and (b) shows room tem-perature EL spectra of Ge NPs embedded in Al2O3 matrix in a MIS structure afterannealing at 800 and 950 ◦C, respectively. Both spectra show three peaks at ∼600,700 and 800 nm. These three peaks is likely to be dependence on the size distributionof Ge NPs or to different recombination mechanisms. Peak at lower frequencies isrelated to defect-related recombination at the Ge NP/Al2O3 interfaces [33, 34]. Also,a blue shift can be observed for smaller Ge NPs (lower annealing temperature) dueto the quantum size effect. Therefore, the NP size and matrix plays an important roleon the optical properties of NPs devices [2].

§1.4 Applications of GeSi nanoparticles 5

Figure 1.3: Room-temperature EL spectra of (a) 3 nm Ge NPs after annealing at 800 ◦C and(b) 6.1 nm Ge NPs after annealing at 950 ◦C embedded in Al2O3 MIS structure for different

injected currents. (adapted from [2])

1.4 Applications of GeSi nanoparticles

GeSi nanostructures attract attention because their properties can be tuned betweenthose of Si and Ge to achieve optimum performance characteristics for particular ap-plications. In order to achieve highly efficient optoelectronic devices it is essential tomanipulate and engineer the size, shape, defects and atomic structure of these NPs.GeSi alloy NPs can be found in forms of a core-shell or a homogeneous structure.Also, there are different ways of synthesizing and fabricating these structures suchas sputtering, chemical vapour deposition and ion implantation. Some of the moresignificant observations are summarized below.

Ortiz et al. [35] studied the properties of amorphous GeSi nanoparicles grownby low pressure chemical vapour deposition, embedded in SiO2 matrix, for nano-electronic device applications. The properties of NPs were examined using multipletechniques namely Rutherford backscattering spectrometry, x-ray reflectrometry andtransmission electron microscopy. They concluded that the NPs size and composi-tion can be controlled by adjusting the deposition time and the precursors flow ratiorespectively. Amorphous GeSi NPs with a controllable composition (Ge fraction inthe 0.2<x<0.45 interval) and uniform size - 2.5 to 5 nmwere obtained. Also, usingrapid thermal annealing, the amorphous NPs were formed to crystalline at tempera-tures between 800 and 900 ◦C.

Zhu et al. [36] reported on the fabrication of group-IV semiconductor materialNPs in SiO2 by ion implantation and thermal annealing. The microstructure of theseNPs were characterized by transmission electron microscopy and showed that an-nealing temperature and implantation fluence have a critical effect on NP size dis-tributions. Ion implantation of Si and Ge (with the same fluences) of (0.6-3)×1017

ions/cm2 was performed into a 1.7 µm SiO2. Samples were then annealed at tem-peratures between 600 to 1100 ◦C for 1 hr under Ar+4%H2 ambient at atmosphericpressure. Spherical GeSi NPs with a size distribution of 10 to 15 were obtained withrandom crystal orientation after annealing at 1100 ◦C.

6 Introduction

Mogaddam et al. [37] studied the phase separation of GeSi NPs embedded in aSiO2 matrix by cosputtering Si, Ge and SiO2 on a Si substrate independently. Thestructure and composition of the NPs were examined with transmission electron mi-croscopy, Raman spectroscopy and x-ray diffraction as function of annealing timeand temperature. Mixtures of nanocrystals of all sizes with the range of 10-15 nmwere observed within the cosputtered layers. Uniform GeSi nanocrystal were formedupon annealing at 900 ◦C for 1 hour. It shows the effect of annealing temperature andtime on the properties of GeSi nanocrystals. For samples with higher Ge content thecompositional uniformity is found to be disturbed as result of increasing annealingtemperature (to 1000 ◦C) and time to 3 hours . This could be due to a phase separa-tion in GeSi nanocrystals resulting in the formation a Si-rich GeSi core covered by arelatively Ge-rich GeSi shell.

1.5 Why embedded nanoparticles : Interaction with a matrix

The properties of embedded NPs are influenced by their interaction with the hostmatrix in which they are confined. The structure, size and shape of NPs are affectedby the matrix, in which they are formed. For example faceted nanoparticles could beformed in a crystalline matrix [38]. Superheating of the nanoparticles is also possiblebecause the interfacial energies can effect the melting point characteristics of embed-ded NPs.

SiO2 is an excellent dielectric matrix for exploiting NP properties and is wellmatched with current optical integrated circuit technology. As a consequence, thestructural properties of Ge NPs embedded in an amorphous SiO2 matrix have beenextensively reported [36, 39, 40, 41, 42, 43, 44, 45]. However, in order to investigatenew devices or improve the efficiency of current optoelectronic devices it is impor-tant to characterize new dielectric matrices. As an example, the difficulties of carriertransport in a SiO2 matrix due to its large band gap make it suitable candidate formemory devices. Although, utilizing smaller band gap matrices could enhance elec-troluminescence properties.

In order to investigate the role of the host matrix on the size and structure ofNPs, we study Ge NP synthesis in different types of Si3N4 and SiOxNy with differentstoichiometries. The following sections focus on the properties of Si3N4 and SiOxNy,with the stoichiometry between Si3N4 and SiO2.

Si3N4

Silicon Nitride (Si3N4) is a well established dielectric with excellent optoelectronicproperties, which makes it a suitable candidate for a wide range of applications inseveral fields of materials engineering [46]. Insulating triple oxide-nitride-oxide layer

§1.6 Si3N4 growth process 7

structures [47] for microelectronics applications can be fabricated using amorphousSi3N4. Due to its high charge traps, amorphous Si3N4 is utilized as charge storagelayer in nonvolatile memory devices [48, 49]. Amorphous Si3N4 has high dielectricconstant, large electronic gap, high temperature stability (up to 1900 ◦C at ambientpressure), high strength and compressibility, and excellent resistivity against oxida-tion and corrosion and is chemically inert.

Si3N4 can be prepared by several processes: sputtering, chemical vapour depo-sition and glow-discharge deposition. The different deposition processes provideconsiderably different film properties. Si3N4 exhibits two crystalline structures, al-pha and beta Si3N4 and the later one is the stable phase.

Comparison of amorphous SiO2 and Si3N4 matrices

The Si3N4 structure is quite different from that of SiO2. Instead of flexible, ad-justable Si-O-Si bridge bonds, the Si-N-Si structure is rendered rigid by the necessityof nitrogen forming three rather than two bonds [50]. Amorphous SiO2 is the mostwidely used dielectric matrix in Si devices at present because it is easily and cheaplyprepared by the oxidation of silicon. However, SiO2 is structurally porous, as in-dicated by its low density (2.2-2.6 g/cm3) [51] hence the use of SiO2 film can causesome inherent disadvantages depending on the device applications. For example, thehigh permeability of SiO2 toward water vapour and other impurities reduce its effec-tiveness for passivation, and the migration of alkali ions in thermally grown SiO2 isresponsible for instability in metal-oxide-semiconductor devices. On the other hand,presence of the porosities makes it a suitable candidate for filtering and sensor deviceapplications.

Amorphous Si3N4 films can retain local Si−N bonding configurations but lacklong-term translational order. Because of their dense film structure, which can ef-fectively prevent the diffusion of water, oxygen, and even small anions such as Na+,stoichiometric amorphous films are widely used in the integrated circuit industry aspassivation layers. Although, the stoichiometry and therefore the film properties aredependent on many parameters such as the growth process. Table 1.1 compares theproperties of Si3N4 and SiO2 matrices at 300K.

1.6 Si3N4 growth process

1.6.1 LPCVD Si3N4

Low Pressure Chemical Vapor Deposition (LPCVD) is one of the processing methodused for the growth of Si3N4. The formation of Si3N4 in LPCVD requires a Si source,typically dichlorosilane (DCS) and ammonia (NH3). The use of dichlorosilane ratherthan silane improves film uniformity and allows closer wafer spacing. The LPCVD

8 Introduction

Table 1.1: Properties of SiO2 and Si3N4 at 300KProperties SiO2 Si3N4

Structure amorphous amorphousMelting Point(◦C) ∼ 1600 . . .Density (g/cm3) 2.2 3.1Refractive Index 1.46 2.05

Dielectric Constant 3.9 7.5Band gap (eV) 9 ∼ 5

Infrared Absorption Band (µm) 9.3 11.5-12.0

Si3N4 films extremely used as dielectrics and membrances. The deposited films haveexcellent purity, uniformity good step coverage, and are close to the stoichiometricwhich leads to good electric properties and high thermal stability and low etch rates.However, the reaction rate is slow and high temperatures are necessary for deposi-tion. Stoichiometric Si3N4 is used as passivation layers [52], oxidation masks [53],gate dielectrics [54], and diffusion barriers [53]. Enriching Si3N4 films with siliconreduces the stress so that thicker films can be used without rupturing; it and alsoreduces the HF etch rate [55].

1.6.2 PECVD Si3N4

Si3N4 thin films fabricated by plasma enhanced chemical vapour deposition (PECVD)have been extensively used in semiconductor industry. Deposition is usually carriedout in showerhead configuration reactors where temperatures of 250 to 400 ◦C aretypically employed. Ammonia is the most common oxidant and at low power den-sity and low ammonia concentrations, the gas phase is dominated by SiH2, SiH3, andpolymerized silanes.

PECVD deposited Si3N4 films are often under large tensile stress, presumablyresulting from the elimination reactions which form the film. Ion bombardment canbe used to densify films and make them more compressed. Ion bombardment isalso easily adjusted by using dual frequency reactors and this approach is often em-ployed for the facile adjustment of stress of deposited nitrides [56]. The properties ofthe PECVD film are controlled by many operating variables, including pressure, gasflow rate, reactant composition, and substrate temperature [57].

PECVD Si3N4 films are useful for a variety of applications, including anti-reflectioncoatings in solar cells [58], passivation layers [59], dielectric layers in metal/insulatorstructures, as capacitor dielectric and diffusion masks. PECVD Si3N4 films are knownto contain hydrogen, this fact enhances their efficiency in polycrystalline silicon solarcells due to defect passivation by hydrogenation. In addition, the PECVD Si3N4 thinfilms tend to be non-stoichiometric and the composition of the film can be varied tobe adjustable for specific purposes. The mechanical properties of the PECVD films

§1.6 Si3N4 growth process 9

are strongly influenced by the specific details of the deposition parameters.

1.6.3 Comparison between PECVD and LPCVD Si3N4

LPCVD has become the preferred method for deposing amorphous films due tohigher throughput, lower particulate contamination, easier film composition control,excellent uniformity, and especially lower hydrogen content in the prepared films[60, 61]. Table 1.2 summarized some of the properties of LPCVD and PECVD films.The most significant difference between the PECVD and LPCVD process is the hy-

Table 1.2: Properties of SiO2 and Si3N4 at 300 KLPCVD PECVD

Deposition temperature (◦C) 700-800 >300

Gases SiCl2H2 + NH3 SiH4+NH3 , SiH4+N3

SiH4+NH3 SiH4+NH3+N2

Typical deposition pressure (Pa) 20-70 60-700

Film composition SiNx(H) SiNxHy

(Si3N4 (H) for SiH2Cl2/NH3 ∼= 1/3-1/4) (Si3N4Hy for SiH4/NH3 ∼= 1/3-1/4)

Atomic % H in SiN 3-8 10-40Density(g cm−3) 2.9-3.1 2.3-3.1

drogen content of the film. The H2 content can reach up to 40 at.% for PECVD whilefor LPCVD layers it is as low as 3 at.%. The total hydrogen content of the Si3N4 filmand the type of hydrogen bonds have a strong effect on the stress of the film. Anothernoticeable difference is the growth temperature required for deposition of the layerwhich result in layers with different properties and constraints. The deposition rateof the LPCVD process is higher than PECVD but require a higher substrate tempera-ture. The LPCVD films are more uniform and more stoichiomtric compare to PECVDfilms. PECVD utilizes silane and ammonia whereas LPCVD uses dichlorosilane andammonia. They are resistant against poor environmental conditions and chemicalsand are therefore used also as etch mask material against KOH and to a lesser degreeagainst diluted HF solutions.

LPCVD Si3N4 films cannot be used for gate dielectrics due to their poor interfacequality and high bulk trap density, but they are widely chosen for applications suchas diffusion mask, nitride spacers, etc. Therefore, PECVD films can be utilized foraforementioned applications. The high refractive indexes and bandgap make thesefilms possible candidates for optical waveguides.

1.6.4 PECVD SiOxNy

SiOxNy layers have attracted lots of attention due to their highly tunable optical andmechanical properties, which make them suitable candidates for gradient-index op-tics [62], antireflection coating and passivation surfaces for crystalline Si solar cells[63], and light emitting devices [64]. Compared to conventional silica materials,

10 Introduction

SiOxNy thin films offer a much larger refractive index for changing the core of opti-cal waveguide and allowing the minimum bending radius for waveguide designs tobe lowered. Furthermore, the high transparency and index of refraction for siliconbased oxynitride thin layers can easily be adjusted over a wide range, between 1.45(SiO2) and 2.0 (Si3N4). SiOxNy thin films can be formed and grown on the surface ofSi by chemical vapour deposition (PECVD, LPCVD and electron cyclotron resonancePECVD). The quality of deposited layers is highly related to processing temperatures.LPCVD methods typically require a high temperature (>400 ◦C) whereas the PECVDmethods require temperatures of <400 ◦C. Also, the precursor gas species are dif-ferent for the aforementioned methods. PECVD is the preferred method due to itslow temperature process (150-350◦C), which prevents film performance degradationduring the fabrication of optical and photonics devices [65].

1.7 Synthesis of nanoparticles: NPs formed in Si3N4 and SiOxNy

Si NPs

Some researches have focused on the formation of Si NPs in a Si3N4 matrix andthe PL emission of these NPs have been extensively reported.

Cen et al. [66] studied the formation of Si NPs in 120 nm LPCVD Si3N4 films,thermally grown on either a quartz substrate or a 30 nm SiO2 thin film, on Si (100)wafers. Si ions were implanted at the energy of 30 keV with a fluence of 3.5 ×1016 atoms/cm2 at room temperature. Subsequently, the implanted films were ther-mally annealed in N2 atmosphere at temperatures ranging from 800 to 1100 ◦C for1 hour. Optical transmission and photoluminescence of Si3N4 thin films were mea-sured. Also, the effects of Si implantation and thermal annealing on the PL wereexamined. The study shows an enhanced red PL band due to the transitions be-tween the Si-band-related states close to the Si3N4 band.

Kwang et al. [67] examined the effect of ion implantation of Ar+1 ions and ther-mal annealing on the PL in a LPCVD amorphous Si3N4. Post-implantation sampleswere annealed in N2 or in vacuum for 20 min at temperatures from 800 to 1000 ◦C.A broad PL band centred at a range of 309-688 nm observed for the as-implantedsamples, can be related to the defects formed by breaking bonds, which act as PL-quenching nonradioactive recombination centres in amorphous Si3N4.

Behar et al. [3] reported PL results of a Si-rich PECVD Si3N4. Si ions were im-planted at the energy of 170 keV with a fluences of 0.5-1 × 1017 atoms/cm2 at roomtemperature, 200 and 400 ◦C. Post-implantation annealing temperatures varied be-tween 350 and 900 ◦C under a N2 atmosphere for 15 to 120 minutes. Samples wereimplanted with different fluences in order to modify the size of the NPs and the PLband. However, no change was observed in the position of the PL band, suggested

§1.7 Synthesis of nanoparticles: NPs formed in Si3N4 and SiOxNy 11

that the emission is related to radiative defects at the interface between the NPs andthe matrix. Figure 1.4 (a) shows the TEM images of a sample annealed at 475 ◦C for1 hour and Figure 1.4 (b) illustrates the PL spectra of a sample annealed at differenttemperatures [3]. Mean NP diameter of Si NPs are from 6 to 8 nm. The study focuses

Figure 1.4: (a) TEM images of a sample after annealing at 475 ◦C, (b) PL spectra of samplesafter annealing at different temperatures. (adapted from [3])

on the origin of the PL bands. Results show that PL spectra of all samples presenttwo bands at ∼ 760 and 900 nm independent of annealing temperature or NP size.These bands are amorphous Si structures and/or nitrogen-related surface states in SiNPs. Keita et al. [68] studied the effect of growth process on the properties of Si NPsgrown by PECVD within a Si3N4 matrix. They found out that amonia flows influencethe film thickness and the NPs physical properties.

Wang et al. [69], investigated the high-efficiency visible PL of Si NPs embedded inSi3N4. A strong room-temperature PL was observed in the whole visible light rangefrom a sample that was annealed at 500 ◦C for 2 min, which makes the system verycompetitive candidate for light-emitting devices.

Ge NPs

The properties of Ge NPs embedded in Si3N4 have been reported to a lesser ex-tent. Lee et al. [70], characterized the structural and optical properties of embeddedGe NPs in a Si3N4 matrix. Ge NPs were fabricated by radio-frequency magnetronsputtering, followed by post-annealing in N2. Ge content was varied between 30-50vol% with variation of annealing temperatures between 600-900 ◦C. Different tech-niques were used to characterise the crystallization properties of a Ge-rich SiN layer.TEM images show disk-shaped Ge crystals being formed in the 50 vol% sample an-nealed at 900 ◦C. For other samples no crystallization was observed. It is shown thatthe crystallization of the 50 vol% samples were initiated at 800◦C and the sphericalcrystals become disk-shaped as a result of increasing annealing temperature.

12 Introduction

Sahu et al. [71], synthesised Ge NPs in SiN films to examine the effect of ionimplantation energy for memory applications. Post-implantation samples were an-nealed at 800◦C for 30 min in N2 atmosphere. Two different energies (3 and 5 keV)were chosen for the evolution of Ge NPs. In the case of low energy, no Ge NPs wereobserved. In contrast, for energy of 5 keV, nearly spherical and well-isolated Ge NPswith an average size 3.5 nm were self-assembled.

Mirabella et al. [4], focused on a comparison of the properties of Ge NPs and therole of the host matrix. Ge ions were implanted into SiO2 and Si3N4 matrices withfluences of 2.9 to 9.6 × 1016 atoms/cm2. Post implantation samples were annealedat up to 900◦C . Comparing the obtained NPs, it can be seen from TEM images thatsmall and amorphous NPs were formed in Si3N4 as it is shown in figure 1.5. This isclear evidence for the role of the host matrix on Ge NP synthesis.

Figure 1.5: STEM images of Ge NPs formed in (a) Si3N4 , (b) SiO2. (adapted from [4])

Si NPs formed in SiOxNy

The formation of NPs in a SiOxNy matrix has not been investigated widely butshow promise for use in memory devices and other engineer applications.

Ehrhardt et al. [72] recently reported that the size of Si NPs in a SiOxNy matrix canbe altered by varying N content. The study reports on the size-dependent structuralproperties of Si NPs formed in an amorphous SiOxNy matrix by ion implantationand thermal annealing.

Perret et al. [73] reported on the controlled fabrication of Si NPs in thin SiOxNy

layers. They investigated how the size of NPs are affected by annealing conditions.The obtained results indicate that this system could be used for nonvolatile memoryapplications.

§1.8 Motivation 13

1.8 Motivation

The synthesis of NPs embedded in dielectric matrices provide a new pathway to fab-ricate structures of interest for new optical, electronics and optoelectronic devices.The growth and characterisation of Ge NPs embedded in a SiO2 matrix, and to alesser extent Si3N4, have been extensively reported. However, the prospect of de-veloping new optoelectronic devices requires a good understanding of NPs and thehost matrix mechanism. In fact, the physical and chemical interaction between theimplanted ions and the host matrix atoms is crucial in determining the final systemconfiguration. The formation of Ge NPs in SiOxNy matrices by ion implantation hasnot been studied broadly. Formation of metalloid NPs (with different techniques) inPECVD and LPCVD Si3N4 have been reported, but no record of Ge NPs formed byion implantation in LPCVD Si3N4 and PECVD SiOxNy has been found.

To explore the potentialities of NPs a thorough understanding of implantationcondition on the structural properties of NPs is required. However, this step couldbe hampered by some peculiarities of NPs, which make their structural characteri-zation a challenge task. To mention just a few, the long-range order characteristicof NPs is difficult due to the reduction in their dimension. Also, NPs are highlydisordered, given a lesser constrain on the position of surface atoms causing reduc-tion of coordination and/or interaction with their surrounding matrix. Therefore, awell established technique is essential for the better understanding of the structureand morphology of NPs. This study highlights the importance of using synchrotron-based techniques (x-ray absorption near-edge structure and extended x-ray absorp-tion fine structure) for characterization of embedded NPs. This suit of techniquesproduces a complete picture of the structural properties of NPs under study.

This thesis is intended to contribute to the understanding of Ge NP synthesis andto compare the role of the host matrix on the structure and size of NPs. LPCVD andPECVD Si3N4 layers were chosen to characterize the influence of stoichiometry andgrowth mechanism of the host matrix on formation of NPs. PECVD SiOxNy layers(with different stoichiometry between SiO2 and Si3N4) were studied to compare thesize evolution and structural properties of embedded NPs.

The variation of the NP size is characterised by cross-section transmission elec-tron microscopy (XTEM), while the phase and structure are characterise by x-rayabsorption spectroscopy (XAS). Synchrotron techniques were specifically used to testthe efficacy of XANES and EXAFS for NP characterization.

1.9 Outline of the thesis

The thesis is organized in the following sequence:

• Chapter I provides an introduction and motivation for the research developed

14 Introduction

for this thesis, with the description of some important applications for whichNPs can be used.

• Chapter II describes the laboratory-based techniques used for the NP fabri-cation and analysis, namely ion implantation, TEM, and synchrotron basedexperimental techniques such as XAS. Also explains the sample preparationand data analyses for the aforementioned techniques.

• Chapter III and Chapter IV Presents results from the ion beam synthesis of Geions in LPCVD and PECVD Si3N4.

• Chapter V presents results from the ion beam synthesis of Ge ions in SiOxNy.

• Chapter VI summarizes the main conclusions of size and structural propertiesof NPs from the results of previous chapters.

Chapter 2

Formation and characterizationtechniques

The following chapter describes the principles and technical aspects of the exper-imental techniques used in this thesis to fabricate and characterise the propertiesof nanoparticles, namely ion implantation, transmission electron microscopy, x-raydiffraction, Raman spectroscopy and Rutherford backscattering spectroscopy, andsynchrotron radiation techniques: x-ray absorption spectroscopy (x-ray absorptionnear-edge structure and extended x-ray absorption fine structure).

15

16 Formation and characterization techniques

2.1 Ion implantation

Ion implantation is a well established doping technique in the semiconductor in-dustry. This versatile technique can also be used for synthesizing nanometer-scaleclusters and crystals embedded in the near-surface region of hosts in order to cre-ate nanocomposite materials [74]. Compared to other techniques - chemical vapourdeposition, sputter deposition and microwave radiation - ion implantation is a goodway to control the size and the depth distribution of nanoparticles. Moreover, im-plantation can be performed at different temperatures. This technique has a longhistory of usage in integrated-circuit technology and microelectronics industry forintroducing both impurities and disorder in a solid layer. The concept of ion implan-tation for doping semiconductors was introduced by William Shockley in 1956. Itis used as a materials engineering process by which implanted ions as a means ofchanging the physical, chemical, or electrical properties of the solid.

Energetic ions are stopped in a material via two processes: nuclear stopping dueto elastic scattering of ions with atoms, and electronic stopping due to interactionof implanted ions with the electrons of the target. Figure 2.1 shows the nuclear andelectronic components of ion stopping power as a function of ion energy. The relativeimportance of the two processes depends on the energy E and the atomic numberZ of the ion. Note that nuclear collisions dominate at low energies and electroniccollisions at high energies.

Figure 2.1: Electronic and nuclear stopping power of aluminum ions in aluminum, versusparticle energy per nucleon.

One of the disadvantages of ion implantation for doping application is the pro-duction of lattice damage (vacancies and interstitials) which may evolve from simplepoint defects into complex dislocations or voids. Displacements result in forma-tion of interstitial atoms and a vacancies. The displaced atoms may have energy

§2.1 Ion implantation 17

high enough to further displace other target atoms along its path. Post implantationthermal annealing is generally required to minimize implantation-induced disorderwithin the implanted layer.

In this study, ion implantation profiles were calculated with the SRIM code[75].As each incoming ion interacts with the target atoms it loses energy and stops at aspecific depth. The depth distribution is a function of the ion species, energy andnature of the target. In this study two profiles were used to form nanoparticles. Forlow concentration samples multiple ion energies/fluences were used to create a near-uniform ion distribution. High concentration samples were implanted with a singleenergy. Figure 2.2 shows the simulations of predicted implantation depth profiles ofsamples implanted with Ge.

0 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 00

5 0 0 0

1 0 0 0 0

1 5 0 0 0

2 0 0 0 0

Atoms

/cm3 /At

oms/c

m2

0 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 00

5 0 0 0

1 0 0 0 0

1 5 0 0 0

2 0 0 0 0 b )

a )

D e p t h ( µ m )Figure 2.2: SRIM simulation for Ge ions profiles in a Si3N4 target.

2.1.1 Ion accelerator

An ion implanter generally consists of discrete sections for ion production, acceler-ation, mass-to-charge filtering, rastering, and implantation of target substrate. IonImplantation in this study was carried out using 1.7 MV NEC tandem acceleratorlocated at Department of Electronic Materials Engineering at the ANU. Figure 2.3shows an schematic of ion implanter. The accelerator is equipped with a source ofnegative ions from Cs+1 sputtering, which produces negative ions from a solid cath-ode target. Vaporized Cs+1 is incident onto the surface of an ionizer and acceleratedtowards a negatively based cathode surface. Negative ions are accelerated away from


the cathode through an aperture in the ionizer by an extraction potential of 15 kV.An additional acceleration section, with bias potential of 65 kV and an Einzel lens,increases the energy and focus the beam respectively. The 90◦ magnet acts as a massseparator of the beam, which contains desired ion species (Ge) and some residualimpurities and molecular components from the cathode holder. The magnetic fieldof the 90◦ magnet is chosen such that only the desired ion of a specific atomic massis deflected through. After the 90◦ magnet, the beam is delivered to the low energyend of the accelerator tank via an Einzel lens and x-y magnet steerers into the accel-erator tank. There is a high voltage terminal (1.7 MV) in the middle of the tank withpressurized SF6 gas for insulator purposes. The beam of negative ions is acceleratedtowards the positively charged terminal where they are partially stripped of electronsby N2 charge exchange. The positive terminal repels the positive ions which receiveadditional acceleration from the high energy end of the tank. A quadrupole lens andfurther steering are then used to bring the focused beam onto a 15◦ analysing magnetwhich is used to select the desired charge state. An electrostatic scanning is designedto obtain two focused bean for scanning purpose. Samples discussed in the presentwork were thin film of Si3N4 and SiOxNy on Si which were mounted onto a sampleholder. The chamber pressure was typically 10−7 Torr. The samples were attached toa stainless-steel movable strikethrough sample holder (80×40 mm) with metal clipsand carbon thermal conductive paste. The sample holder can be heated up to 500◦C using a 200 V lamp or cooled down with liquid nitrogen to -196 ◦C dependingon the required implantation temperature. To enable accurate implantation fluencemeasurement, a Cu radiation shield surrounding the sample holder was maintainedat -300 W to suppress secondary electrons (which are produced by the ions impact-ing the atoms in the sample). Irradiation was performed perpendicular to the samplesurface.

Figure 2.3: Schematic digram of a 1.7 MV tandem accelerator at the Department of ElectronicMaterials Engineering, Australian National University.

§2.1 Ion implantation 19

2.1.2 Nucleation and growth process

The precipitation of nanoparticles, is rather complex and quiet well understood. Thisis because the process of NP formation is highly dependent on the mobility of theimplanted species. For high mobility implanted species, the nucleation and phaseseparation can occur in as-implanted samples. However, post implantation thermalannealing is generally required to induce phase separation and promote the growthof NPs. In addition, a combination of different parameters influence the mean NPsize. Depending on the nature of the implanted element and the matrix, the implan-tation conditions (implantation energy, fluence) and post implantation treatments(thermal annealing, laser annealing) in controlled atmosphere (such as an oxidizingatmosphere like air or oxygen, an inert atmosphere like Ar, a reducing atmospherelike hydrogen or a mixture of hydrogen and argon) can be used to fabricate NPs [74].

If the concentration of the implanted species is above its solubility limit in the ma-trix, a thermodynamically metastable phase is formed. This metastable phase resultsin homogeneous or heterogeneous nucleation and growth. The classical nucleationtheory of supersaturated solutions can be adopted to explain the precipitation ofNPs. The growth of NPs relies on the concept of surface tension or Gibbs free energy(competition between surface and bulk free energy): the additional free energy re-lated to the formation of a crystalline nucleus with radius r in an amorphous matrix,

Figure 2.4: Scheme of ion-beam synthesis of nanoparticles [5].


written as equation 2.1.

∆Gtotal =43

πr3∆Gphase + 4πr2γ (2.1)

where ∆Gphase is the bulk free energy change and γ is the surface or interface freeenergy (surface tension) [76]. If the free energy of equation 2.1 is negative thengrowth at the nucleated phase favoured. A critical radius size is required which canbe obtained by differentiating equation 2.1 as

r∗ =−2γ

∆Gphase(2.2)

If the degree of supersaturation is large then NP can grow without competingeffects. For nuclei lager than r∗ NP growth occurs via diffusion from the surroundingmatrix. As the degree of supersaturation is reduced, Ostwad ripening, which is thegrowth of large NPs at the expense of smaller ones, occurs. A schematic of ion-beamsynthesis of NPs is presented in figure 2.4.

2.1.3 Thermal annealing

Thermal annealing (up to 1100 ◦C) was performed in a conventional horizontalsingle-zone quartz tube furnace in a N2 gas ambient. Samples were locating on aLPCVD or PECVD Si3N4- or SiOxNy-on-Si wafer in a silica glass boat, to minimizethe chance of contamination. This was placed in the tube furnace following stabiliza-tion of the operating temperature.

2.2 Laboratory based techniques

2.2.1 Rutherford backscattering spectrometry

Rutherford backscattering spectrometry is a versatile and non-destructive techniqueto provide information about the elemental composition as a function of depth. Thistechnique is based on classical scattering in a central-force field, which is an elasticcollision between the positively charged nuclei of the projectile and target atom. Itinvolves measuring the number and energy of backscatter ions after colliding withatoms in the near surface region of a sample. Figure 2.5 shows a schematic diagramof the RBS principle. In the scattering process, a small fraction of incident particlesundergo a direct Coulomic repulsive interaction with a nucleus in the target whichcan be modelled as an elastic collision using classical physics. The energy of backscat-tered ions are measured by a surface barrier detector fixed at a scattering angle of168◦. The energy of the particles, measured by the detector, depend on many param-eters including the stopping cross-section of the target materials. The energy transferor kinematics in elastic collision can be solved using the principles of conservationof energy and momentum. The ratio of the projectile energies, commonly refers toas the kinematic factor (K), shows that the energy after scattering is determined only

§2.2 Laboratory based techniques 21

Figure 2.5: Schematic diagram of the RBS principle.

by the masses of the particle and target atom and the scattering angle, which can becalculated for surface scattering approximation as:

K ≡ Es

Ei= [

(1− (M1 sin(θ)M2

)2)12 + M1 cos(θ)

M2

1 + M1M2

]2 (2.3)

where Es and Ei are the energies immediately after and before the scatteringevent, M2 and M1 are the masses of the target ion and incident atom, while θ is thescattering angle.

RBS measurements were performed using a 1.7 MV NEC tandem accelerator,similar to that used for ion implantation, equipped with He source. A Si surface-barrier detector positioned at a scattering angle of 168◦ detects the backscatter ions.RBS spectra were analysed with the RUMP program [77]. The conversion from chan-nel to depth scale was performed using a mean energy approximation [78] with theenergy resolution of 5-7 keV per channel. The depth distribution of implanted Geatoms in the matrix layer was measured before and after annealing with RBS.

2.2.2 Transmission electron microscopy

Transmission electron microscopy was was used to measure the shape, phase andsize distribution of nanoparticles formed in dielectric matrices. TEM images werecollected with a microscopy (JEOL 2100F) operating with 200kV electrons.

2.2.2.1 TEM sample preparation method

Figure 2.6 illustrates the steps in preparing a TEM sample for cross section mi-croscopy.


• a) Samples were cut into 3×5 mm pieces and cleaned in Acetone.

• b) pieces were glued face-to-face to a slightly thicker dummy Si sample usingan G-5 epoxy cured on a hotplate at ∼ 100◦C. An additional stack of dummywafers can be attached for support.

• c) The stack was then cut with a low speed diamond impregnated wheel toa thickness of few mm, and a 3 mm diameter disc is ultrasonically cored andcentred about the interface to be examined.

• d) The core was mechanically ground to ∼ 100 µm, then dimpled further to∼80-90 µm.

• f) Finally, the dimpled region was ion milled until perforation.

Figure 2.6: Schematic of sample preparation produced for cross section transmission elec-tron microscopy.

2.2.2.2 Data analysis

Average NP diameters were calculated by averaging the NP volume through cubicweighting of the extracted NP diameters. This allows for direct comparison with theextended x-ray absorption fine structure (EXAFS) measurements, being a volume-weighted technique. The NP size distributions were determined from micrographscovering the extent of implanted depth range. Size distribution is depth dependent,

§2.2 Laboratory based techniques 23

with longer NPs at peak of Ge distribution and smaller ones at shallower and deeperdepth.

2.2.3 Raman spectroscopy measurements

The Raman spectroscopy experiments in this thesis were carried out with the Ren-ishaw Raman System in the department of Electronic Material Engineering at theAustralian National University. The red laser (632.8 nm) of the system was employed.The scattered light was collected by a 100× objective lens of a confocal optical micro-scope with a spot size of 0.91 µm. A high performance edge filter was used for theRayleigh scattered light rejection, and allowed the measurements to have the lowestRaman shift down to 100 cm−1. The Raman scattered light dispersion was achievedby a grating whose groove density is 1200 lines/mm, and a Peltier cooled (70 ◦C)CCD detector was used for the Raman data recording. The Raman spectra werenormalized after the acquisition, and fitted with a Voigt function to calculate the po-sitions the Raman peaks, which were used to characterize the phase component ofthe ion implanted samples1. Common Raman artifact [79, 80] near ∼300 cm−1 re-sulting from the overlap of the transverse-acoustic second-order peak for Si and thefirst-order optical peak for Ge were completely eliminated for all examined samplesby etching the Si substrate. The removal of the Si substrate (with the method thatwill be explained in details in 2.3.6) negates the need to use special measurementgeometries (crossed polarization) [79] or posterior subtraction of the spectra [81].

2.2.4 X-ray diffraction measurements

Figure 2.7: Schematic diagram of glancing-incidence XRD setup adopted from [6].

X-ray diffraction (XRD) was used for phase identification in bulk crystalline andnanoparticle samples. XRD profiles were collected using an XRD system (Panalytical

1Details can be found in Appendix D


X’Pert Pro Diffractometer) with a glancing incidence geometry (ω=0.8) Cu-kα radia-tion (45 KV 40 mA source). The Glancing-incident XRD configuration was used torecord the diffraction patterns of thin films, with minimum contribution of substrate.This allows a parallel monochromatic x-ray beam to fall on a sample surface at afixed angle of incidence (α) and for a diffraction profiles to be recorded by a detector.Figure 2.7 shows a schematic of the experimental set up. After the measurement thedata were analysed with the PowderCell software.

2.3 Synchrotron-based techniques

Figure 2.8 shows XAS beamline at the Australian Synchrotron facility. All measure-ments were performed at the x-ray absorption spectroscopy beamline. The followingsections present a detailed outline of the sample preparation and data analysis.

Figure 2.8: Schematic of the XAS beamline optics at the Australian Synchrotron.

2.3.1 X-ray absorption spectroscopy

X-ray absorption spectroscopy (XAS) is a powerful technique to characterize thestructural properties of materials in a solid, liquid or gas phase. XAS yields infor-mation on the local atomic structure and electronic states. Because of the versatilityof this technique, including element-specificity and nanometer range, it can be usedwith complementary techniques to provide useful information about the structuralproperties of examined systems.

§2.3 Synchrotron-based techniques 25

2.3.2 Basic principle

XAS measures the energy-dependent fine structure of the x-ray absorption coefficientnear the absorption edge of a particular element. As a result of an interaction of x-rays with a solid or liquid photons may transfer their energy to bound electrons. Theprobability of this energy transition is dependent on both the initial and final statesof the electrons.

The absorption of monochromatic radiation by a material according to Beer’s Lawis as follow:

I = I0e−µx (2.4)

where µ is the x-ray absorption coefficient, and I0 and I are the incident and trans-mitted radiation, respectively. When the photon energy matches the binding energyof inner shell electrons, there is a sudden increase in absorption, which is known asan absorption edge. Figure 2.9 schematically shows three parts in a x-ray absorptionspectrum as a function of photoelectron energy. It can be seen that the absorptionspectrum is composed of three distinct regions. Pre-edge region: when a photo-electron does not have enough energy to completely leave the atom, it transits to anunoccupied valence state. XANES region: if a photoelectron has enough kinetic en-ergy to escape into the continuum, it results in a multiple scattering process. FinallyEXAFS region: a photoelectron with a very high kinetic energy and weak backscat-tering, will result in a single scattering process between only one neighbour atom.These regions are described in details in the following sessions.

2.3.3 XANES region

The x-ray absorption near-edge structure is the part of the absorption spectrum nearan absorption edge, which can be used to determine the valence state and coordi-nation geometry. The XANES region is commonly considered as some tens of eVbeyond the absorption edge. In this region the excited photoelectron has enoughkinetic energy to occupy the continuum. However, the wavelength of excited pho-toelectron is larger than the interatomic distance between the absorbing atom andfirst near neighbour. Therefore, a multiple scattering effect occurs since neighboringatoms are not only close to the center atom, but also relatively close to each other.Hence, the excited photoelectron tends to bounce between the neighboring atomsbefore it return to hit the absorber atom [82]. This scattering can be used to fin-gerprint the chemical bonding environment such as oxidation states and electronicconfiguration and site symmetry around the absorber atom. Due to the complicatednature of the theoretical calculation for this region, a quantitative analysis is not dis-cussed here. A qualitative comparison of the samples implanted with Ge and thestandards are, however, considered in this thesis. Figure 2.10 is a diagram of themultiple scattering in the XANES region.


Figure 2.9: Different parts of an x-ray absorption spectra as a function of photoelectronenergy.


Figure 2.10: Schematic of a photoelectron multiple scattering.

2.3.4 EXAFS region

The extended x-ray absorption fine structure (EXAFS) region is typically ∼30 eVabove the absorption edge, to 1000 eV or higher. EXAFS oscillation is a result of bothsingle and multiple scattering and therefore it contains information about the coordi-nation number, inter atomic distance and structural and thermal disorder around theabsorber atoms. EXAFS oscillations are the interference of outgoing and backscat-tered waves as it is shown in figure 2.11. The maxima corresponds to the in-phasebackscattered and outgoing waves while, the minima is the result of an out-of-phaseinterference of the aforementioned waves. The EXAFS signal can be defined as theoscillatory part of the normalised absorption coefficient:

χ(E) =µ(E)− µ0(E)

µ0(E)(2.5)

where µ(E) is the x-ray absorption coefficien (equation 2.4) and µ0(E) is the smoothlyvarying background. Instead of using χ(E) equation 2.5 can be expressed as a func-

tion of wave number k=√

2me(E−E0)h , where me is the electron mass. If we consider

only the contribution of single scattering from the surrounding atoms then the EX-AFS signal can be completely described by a sum of sine terms in the wavevector k.Each term represents the contribution of a spherical shell of equivalent atoms at adistance Rj from the absorbing atom. The fine structure can be considered as a sumover the scattering contributions of different paths

χ(k) = ∑ S02N j

∣∣ f j(k)∣∣

kR2j

e−2Rjλ(k) e−2σ2

j k2× sin[2kRj + 2σc(k)) + σj(k)] (2.6)


Figure 2.11: Schematic of the EXAFS process illustrating the origin of EXAFS oscillationsdue to the interference of outgoing and backscattered photoelectron wave. The red line is thebackground. The background parameter determines the cutoff frequency below which theoptimization happens. Thus, the spline is varied until the transformed spectrum between 0and background is optimized. In general, this means that portion of the spectrum is simply

minimized.


Equation 2.6 is a direct relation between EXAFS signals and the structural parametersnamely Nj, Rj and σj.

Where Nj is the number of coordinating atoms within a particular shell, for sin-gle scattering. S0

2 is representative of the amplitude reduction factor which typicallyhas a value between 0.7 and 1.0. This factor accounts for the slight relaxation of thesystem in the presence of the core hole.

f (k) =∣∣ f j(k)

∣∣ eiσj(k) represents the complex scattering amplitude, which is theatomic scattering factor used in x-ray diffraction for a single scattering path. Thisterm explains the element sensitivity of EXAFS and since it is dependant on thenumber of electrons, for elements with very close atomic number the scattering am-plitude is a distinguishable factor.

Rj is the mean value of the bond length distribution, the distance between the ab-sorber and a coordination atom for a single-scattering event. This value diminisheswith increasing distance from the absorber.

The sin[2kRj + 2σc(k)) + σj(k)] represents the oscillation part of the EXAFS signaland the term 2kRj + 2σc(k) + σj(k) is the path of the photoelectron multiplied by thewavenumber to determine the phase and phase shift (2σc(k)) + σj(k)) of the photo-electron as a result of its interaction with the nuclei and the absorber atom.

e−2σ2j k2

is the Debye-Waller factor and to a good approximation it can be consid-ered as the σ2 value. This value accounts for the disorder in the interatomic distancesand is the mean-square displacement of the bond length between the absorber atomand the coordination atoms in a shell. This term is a contribution of dynamic (ther-mal) and static (structural heterogeneity) disorder [83].

For systems with significant asymmetry, a Gaussian function may no longer bea good approximation for the distance distribution, hence higher moments must beaccounted. The equation 2.6 yields the distance distribution ρ(R)

χ(k) = S2o N| f (k)|

k

∫ ∞

0

e−2Rλ(k)

R2 sin[2kR + δ(k)]ρ(R)dR

= Re[1jS2

o N| f (k)|

kejδ(k)

∫ ∞

0

e−2Rλ(k)

R2 e2ikRρ(R)dR] (2.7)

Assuming a narrow distribution with the mean value < R >, the integral can beapproximated by

∫ ∞

0

e−2Rλ(k)

R2 e2jkRρ(R)dR ∼ e−2<R>

λ(k)

< R >2

∫ ∞

0e2jkρ(R)dR (2.8)


Equation 2.9 can be expressed in terms of the cumulants Cn of the distance distribu-tion ∫ ∞

0e2jkρ(R)dR = exp

[∞

∑n=0

(2jk)n

n!Cn

](2.9)

where C1 =< R >= Rj, the second cumulant C2 = σ2j represents variance, and the

third and forth cumulants, C3 and C4, denote asymmetric and symmetric deviationsfrom a Gaussian profile, respectively.

2.3.5 Experimental setup

Figure 2.12: Schematic of the experimental setup for EXAFS measurements at the AustralianSynchrotron.

XANES and EXAFS measurements at the Ge K-edge (11.103 keV) were performedat the XAS beamline of the Australian Synchrotron. A schematic of the XAS beam-line is shown in figure 2.12. The first optical component is a a bendable collimatingmirror with 3 stripes - Si, Rh and Pt - on a flat Si substrate. A Si(311) double crystalmonochromator is used to select the energy for this study, providing an x-ray energyregion from 5.5 to 37 keV. The monochromator can be operated in fixed offset, pseudochannel-cut, step-acquire-step and slew-scan modes of operation. A simple harmonicrejection mirror on the experimental table will restrict the harmonic content to < 10−5

over the entire energy range. Further information on the optical design of the beam-line can be find at Ref. [84]. The monochromatic beam then enters the first detector(where the intensity of the beam is measured). The sample is rotated 45◦ with re-spect to the incoming beam. The solid state detector, with a 10×10 Ge pixel-array, isat 90◦ with respect to the beam, which is measuring the x-ray flourescence spectra.The last detector measures the intensity remaining after absorption by the sample.For energy calibration, a crystalline Ge (c-Ge) reference foil was simultaneously mea-sured in transmission mode. Data were collected to a photoelectron wavenumber (k)


space up to a value up to 15 Å−1. In order to improve the signal-to-noise ratio forhigh resolution XAS measurements the Si substrate below the Ge-implanted layerwas removed, as is explained in Section 2.3.6. Samples were normally kept at ∼ 15 Kand mounted inside a He cryostat to minimise thermally induced disorder and hencemaximise the EXAFS amplitudes. More information about the beamline design canbe found in reference [85].

2.3.6 XAS sample preparation technique

Figure 2.13: Schematic diagram of EXAFS sample preparation method.

In order to prepare samples for x-ray absorption spectroscopy (XAS) a uniquetechnique was used. The schematic of this method is shown in figure 2.13. Sampleswere mechanically ground to reduce the thickness of the Si substrate. Then sampleswith thickness of 100 µm were etched in a 12.5 M KOH solution to remove the Sisubstrate [86]. The thin implanted layers were stacked together and mounted on a2×5 mm cavity in a multiple slot sample holder. The cavities were sealed with x-raytransparent Kapton tape. This method eliminates the scattering from Si, which signif-icantly improves the signal-to-nose ratio, and enables high resolution measurements.The amorphous Ge (standard) sample was prepared by ion implantation describedin reference [87].

2.3.7 XANES/EXAFS data analysis

The aim of the analysis of EXAFS data is to determine structural parameters suchas coordination number (CN), bond length (R) and disorder of atoms in the variouscoordination shells. In the process of data analysis a hypothetical model of the datacontaining unknown parameters (mentioned above) are to be determined throughcomparison with experimental data [88]. As described in equation 2.6, EXAFS has astandard parameterization in terms of multiple scattering paths.


Figure 2.14: Schematic of the EXAFS data analysing.(a) energy calibration (b) backgroundremoval, (c) k3 weighting of χ(k) and (d) Fourier transforming.


The open source package IFEFFIT [89] is a flexible data reduction/fitting enginethat can be used for EXAFS. In this study, IFEFFIT [89] was used to analyse XASdata. This software package combines of algorithms of AUTOBK [90] and FEFFIT[91]. The former removes the atomic background from a measured EXAFS spec-trum, while the latter fits the experimental EXAFS to theoretical standard calculationfrom FEFF [92]. IFEFFIT is a command-line program with several graphical user in-terfaces using the IFEFFIT library, including ATHENA and ARTEMIS, which makethe IFEFFIT fitting procedure easily accessible [93]. Background subtraction, spec-tra alignment and normalisation of EXAFS data were performed in ATHENA usingAUTOBK while ARTEMIS used FEFF to fit the spectra.

The absorption spectra were first averaged and energy calibrated to the referencefoil signal with the program AVERAGE [94]. The average data were then processedand analysed by ATHENA and ARTEMIS [93]. It is critical to extract the backgroundsince it can affect the final conclusion of the structural information. A best splinefunction was used to remove the background and eliminate the nonstructural com-ponents. This function is controlled by numbers of knots where two polynomialsegments join. After subtracting the background from the EXAFS oscillation, viathe AUTOBK algorithm implemented in ATHENA [93], EXAFS data were convertedfrom χ(E) into a k-space, χ(κ). The χ amplitude rapidly decays with increasing k,therefore the χ(κ) is k-weighted to emphasize the high-k component of the spec-trum. Data were then Fourier transformed using a Haning window through this the-sis which uses a symmetric window to minimise truncation ripples. The structuralparameters were refined through a non-linear least squares fit to the experimentaldata with multiple k-weights of 1-4 to reduce inter parameter correlation.

Values of the energy shift parameter ∆E0 were defined using ARTEMIS and byfollowing the procedure suggested in reference [95] to align the k scale of the theo-retical standard for all samples, and to avoid distortions in the structural parametersarising from a poor choice of the edge energy position E0. Effective scattering ampli-tudes and phase shifts were calculated ab initio with FEFF8 [92]. Spectra were fittedwith a single-scattering path to probe the first near neighbour (NN) in ARTEMIS[93], to investigate the structural parameters of the Ge-implanted samples the fittingprocedure followed suggested references [96, 44, 97, 98]. Eo was allowed to varyduring the fit for the bulk and then fixed for the NPs. S2

o was determined for thestandard (identical within error) and then fixed for the subsequent fitting of all NPspectra. The CN was set to 4 for the bulk standards but allowed to float for the NPsamples. The first two cumulants (interatomic distance and Debye-Waller factor σ2)were allowed to vary during the fit for all bulk and NP samples. Schematics of theEXAFS data analysis steps are presented in figure 2.14.

Chapter 3

Formation of GeSi nanoparticles inLPCVD Si3N4

This chapter discusses size distribution and the structural properties of GeSi NPsformed in LPCVD Si3N4 by ion implantation and thermal annealing. The evolutionof NP size was determined as a function of concentrations and annealing tempera-tures using TEM. We observed a clear effect of concentration on NP precipitation. Forconcentration below 6 at.%, no NP formed independent of annealing temperature.Foe higher concentration (9 and 12 at.%), crystalline and spherical GeSi NPs withdiamond structural were formed after thermal annealing. XANES analysis was usedto characterise the bonding environment of NPs. The structural properties were char-acterized using EXAFS and finite-size effects were readily apparent. With a decreasein NP size, an increase in structural disorder and bond length and a decrease in CNwas observed, consistent with the change in surface-to-volume fraction expected.

35

36 Formation of GeSi nanoparticles in LPCVD Si3N4

3.1 Introduction

The structural properties of NPs and their size evolution were studied as a function ofthe formation conditions. Properties of the NPs were characterized by using varietyof techniques such as XRD, Raman spectroscopy, TEM and XAS.

3.2 Experimental procedure

An amorphous Si3N4 layer of thickness of 2 µm was deposited on a Si (100) sub-strate by LPCVD technique. Layers were deposited at the University of Minnesotananofabrication center on c-Si(100) wafers. The LPCVD system holds four quartztubes that each runs a separate process. A heating elements heat the tube and itcontents up to the correct temperature and causes the gases to react and a film isdeposited. Samples were brought commercially to work as a bench market, hencefurther details about the structure requires more systematic investigations.

In this study 74Ge+1 ions were implanted into 2 µm thick amorphous LPCVDSi3N4 layers, with different fluences and energies. Samples are categorized as lowconcentration and high concentration according to the implantation conditions. Theresults of low concentration samples are discussed in Section 3.3. Also, Section 3.5provides the results of high concentration samples.

RBS measurements were performed with 4.8 MeV He2+ ions and a Si surface-barrier detector positioned at a scattering angle of 168 ◦ to determine the Ge depthprofile. RBS spectra were analysed with the RUMP program [77]. The use of 4.8MeV energy enables the complete separation of the Ge depth distribution from thesubstrate signal extending from the surface to the Si3N4/substrate interface.

X-ray diffraction (XRD) patterns were collected using an XRD system (PanalyticalX′pert Pro Diffractometer) with glancing incidence geometry (ω=0.8) Cu-Kα radiation(45 kV 40 mA source).

Cross-sectional TEM (XTEM) images were obtained in bright-field mode using aJEOL 2100F microscope operated at 200 kV. Samples were prepared with a standardion-beam-milling protocol. Average NP diameters for different concentrations werecalculated by averaging over NP volume through cubic weighting of the extractedNP diameters.

XANES and EXAFS measurements were performed at the Ge K-edge (11.103 keV)at a temperature of 15 K using the x-ray absorption spectroscopy beamline of the Aus-tralian Synchrotron. Fluorescence spectra were recorded with a 10×10 Ge pixel-arraydetector. Data were collected to a photoelectron wavenumber (k) up to a value of 15Å−1. For energy calibration, a crystalline Ge (c-Ge) reference foil was simultaneouslymeasured in transmission mode. Then data were trimmed following the procedure

§3.3 Results and discussion for low concentration samples 37

explained in Chapter 2, data were then Fourier transformed (FT) using a Hanningwindow of with 3.8 to 12.5 Å−1 in k space. In back transforming from radial distance(r) space, a Hanning window range of 1.3-2.7 was utilized. For the bulk c-Ge thevalues of the S2

0 and ∆E0 were 0.94±0.1 and 3.7±0.9 eV, respectively. Spectra werefitted with multiple k-weightings, ranging from 1 to 4. Spectra of NP samples werefitted with two single-scattering paths of GeSi and GeGe random structure to deter-mine the bond length (R), coordination number (CN) and Deby-Waller DW factors(σ2). The DW factors for Ge-Ge and Ge-Si paths (σ2

Ge−Ge and σ2Si−Ge, respectively)

were restrained such that σ2GeSi = σ2

GeGe as is consistent with previous theoretical andexperimental reports [99].

3.3 Results and discussion for low concentration samples

Table 3.1 refers to low concentration samples. These samples were implanted at -196

Table 3.1: Summary of Ge implantation parameters for low concentration samples.Fluences for each sample are given at corresponding energies E (keV).

E=800 E=1300 E=2000 Ge concentration (at.%) Implantation temperature (◦C) Annealing temperature(◦C)

3.23×1015 4.98×1015 8.31×1015 0.1 -196 800, 950 and 1100

1.11×1016 1.66×1016 2.77×1016 0.5 -196 800, 950 and 1100

3.32×1016 4.9×1016 8.31×1016 1.7 -196 800, 950 and 1100

7.27×1016 1.05×1017 1.91×1017 4 -196, 200 and 400 800, 950 and 1100

◦C with multiple energy/fluence implantations to provide uniform distribution. Thepeak Ge concentrations varied from 0.1 to 4.0 at.%. These samples were annealedat 800, 950 and 1100 ◦C for 1 hr in N2 atmosphere (100 litre/min) to promote Geprecipitation and the growth of the NPs.

The depth distribution of implanted Ge atoms (low concentration samples) wasmeasured before and after annealing. A comparison RBS spectra from as-implantedand annealed samples, for concentrations < 4 at.%, showed no change in the Ge dis-tribution after annealing. No precipitation or NP formation was observed for theselow concentration samples (from RBS results and complementary TEM images).

Figure 3.1 shows RBS spectra of 4 at.% Ge samples, implanted at -196 ◦C, for bothas-implanted and annealed samples. Comparison of the as-implanted and annealedsamples shows a noticeable redistribution of Ge towards the Si/Si3N4 interface afterannealing. The Ge depth distribution, calculated by mean-energy-approximation, ispresented in the inset. It can be seen that a narrow Ge peak is also centred close tothe Si/Si3N4 interface. The concentration profile for the as-implanted sample deter-mined from the RBS analysis was in good agreement with that predicted by SRIMprogram [75].


2 0 0 0 3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4

a s - i m p l a n 1 1 0 0 ° C

Conce

ntratio

ns (at

.%)

Norm

alized

counts

(arb. u

nits)

0 . 0 0 0 . 5 0 1 . 0 0 1 . 5 0 2 . 0 00

2

4

6

E n e r g y ( k e V )

D e p t h ( µm )

Figure 3.1: RBS spectra of 4 at.% samples implanted with Ge for as-implanted and annealedat 1100 ◦C. The inset shows the Ge distribution as a function of Si3N4 depth. samples were

implanted at -196 ◦C.

In order to promote the growth of NPs higher implantation temperature werealso considered. Figure 3.2 (a), (b) and (c) compare the RBS spectra of samples withsimilar fluences (4 at.%) as a function of implantation temperature. All samples wereannealed at 1100 ◦C. As mentioned, a significant difference between as-implantedand annealed samples was observed for implanting temperature of -196 ◦C. Mean-while, no redistribution was observed for higher implanting temperatures (200 and400 ◦C). For these cases, the position of peak concentration for as-implanted sam-ples determined from the RBS analysis were in good agreement with that predictedby SRIM program [75]. Slight difference in fluence (∼ 33%) can be noticed for thesample implanted at 400 ◦C, which could be as a miscalculation of the fluence or dueto the not performance of secondary ion suppressor. In any cases, the conclusion isindependent of this problem similar to implanting temperature of 200 ◦C. Therefore,higher implanting temperature inhibits the long range diffusivity of Ge atoms forexamined concentration. The following section will investigate the reasons behindthis observation.

Figure 3.3 shows the XRD pattern for 4 at.% samples after annealing at 1100 ◦C asa function of implanting temperature, using the diffraction patterns of polycrystallinealpha Si3N4 as a reference (indicated by star symbols). Amorphous phase retainedfor elevated temperature implants (200 and 400 ◦C) but, undergoes an amorphous-to-crystalline transformation for sample implanted at low temperature (-196 ◦C) after


2 0 0 0 3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4

0 . 6 a s - i m p l a n t e d 1 1 0 0 ° C

4 0 0 ° C E n e r g y ( k e V )

Norm

alized

counts

(arb. u

nits)

2 0 0 0 3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4

0 . 6 a s - i m p l a n t e d 1 1 0 0 ° C

a s - i m p l a n t e d 1 1 0 0 ° C

2 0 0 ° C

2 0 0 0 3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4

0 . 6

- 1 9 6 ° C

c )

b )

a )

Figure 3.2: RBS spectra for all 4 at.% samples as a function of implanting temperature. Allimplanted samples were annealed at 1100 ◦C.


2 5 3 0 3 5 4 0 4 5 5 00 . 0

0 . 2

0 . 4

0 . 6

0 . 8

1 . 0

1 . 2

1 . 4 a l p h a - S i 3 N 4

2 0 0 ° C4 0 0 ° C

Norm

alized

intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

- 1 9 6 ° C

Figure 3.3: XRD pattern for 4 at.% Ge samples as a function of implanting temperature(offset for comparison). Samples were annealed at 1100 ◦C. The diffraction peaks of alpha

Si3N4 are illustrated with stars.

thermal annealing at 1100 ◦C. Diffraction peaks of the sample corresponding to poly-crystalline alpha Si3N4. As observed from the RBS spectra, the diffusion of Ge atomsoccurred only for an implanting temperature of -196 ◦C, which coincided with thepoly-crystallization and phase transition of the matrix. Clearly, crystallization of thematrix is correlated with long range Ge diffusion and Ge atoms likely diffusing aftercrystallization of the matrix. The cause of the phase transformation of the matrix isinvestigated in more detail in Appendix A. No crystallization was observed for lowerGe concentrations.

Diffusion of Ge to the Si/Si3N4 interface modified the structure of the interface.Figure 3.4 shows the TEM images of (a) the Si/Si3N4 interface and (b) a high mag-nification image of the Si3N4 layer for sample implanted at -196 ◦C. A thin andcrystalline GeSi alloy layer is observed at the interface, as seen in figure 3.4 (a), andpreviously observed in the RBS depth profile of the sample (figure 3.2). No Ge NPformation was observed in this sample. We note that crystallization of the matrixand the formation of the self assembled GeSi layer (at the interface) only occurredfor an implantation temperature of -196 ◦C.

Raman spectroscopy was utilized to confirm the stoichiometry of the self-assembledGeSi layer. Figure 3.5 shows Raman spectra of a 4 at.% Ge samples as a function of


Figure 3.4: TEM images of 4 at.% samples implanted at -196 ◦C after annealing at 1100 ◦C:a) Si/Si3N4 interface and b) implanted region.

implantation temperature and a crystalline GeSi standard. The featureless spectra for

2 0 0 3 0 0 4 0 0 5 0 0 6 0 00

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

G e - G eG e - S i

4 0 0 ° C2 0 0 ° C

c - S i 8 0 G e 2 0α- S i 3 N 4

Intens

ity (a

rb uni

ts)

W a v e n u m b e r ( c m - 1 )

- 1 9 6 ° C

Figure 3.5: Raman spectra of 4 at.% samples after annealing at 1100 ◦C as a function ofimplanting temperatures, and crystalline GeSi standard.

high implanting temperatures reveals an amorphous environment, as was observedwith XRD measurement. However, the spectrum of the sample implanted at -196 ◦Cclosely matches that of c-Si80Ge20 - the stoichiometry of the standard was determinedwith RBS. The first peak at ∼ 250 cm−1 is attributed to Si-N Raman active bond ofalpha Si3N4 as shown by an arrow in figure 3.5. The other features at ∼ 300, 450 and


500 relate to Ge-Ge, Si-Ge and Si-Si (from the local bond and substrate) Raman activebonds of a GeSi alloy. Implanted samples were examined after removal of the Sisubstrate (with the technique explained in Chapter 2), hence the Si-Si Raman activebond at 500 cm−1 is related to the local Si-Si bond from GeSi alloy. However, for thec-Si80Ge20 standard the Si substrate was not removed and it’s feature is a combina-tion of local bond and substrate vibration. The GeSi signal (showed by an arrow) forthe sample implanted at -196 ◦C and annealed at 1100 ◦C is related to the Ge atomsdissolved in the matrix.

Figure 3.6 (a) and (b) show XANES spectra for 4 at.% samples as a function ofimplantation temperatures. Data for crystalline and amorphous Si80Ge20 standardsare also plotted for comparison. Similar to Raman measurements, the XANES spec-

1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 00 . 0

0 . 4

0 . 8

1 . 2

1 . 6

2 . 0

E n e r g y ( e V )

Absor

ption

(arb. u

nits)

a - S i 8 0 G e 2 0 4 0 0 ° C 2 0 0 ° C a - G e

Absor

ption

(arb. u

nits)

E n e r g y ( e V )

a )

1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 00 . 0

0 . 4

0 . 8

1 . 2

1 . 6

2 . 0

c - S i 8 0 G e 2 0 - 1 9 6 ° C c - G e

b )

Figure 3.6: XANES spectra of 4 at.% samples, after annealing at 1100 ◦C, as a function ofimplanting temperature plus amorphous and crystalline GeSi standards.

trum of the sample implanted at -196 ◦C show characteristic features similar to thec-Si80Ge20 standard. In contrast, an amorphous-dominant environment is evident infigure 3.6 (b) for high implantation temperatures (200 and 400 ◦C). Si-Ge bondingenvironment is clearly apparent for all cases, however, phase transition of the ma-trix resulted in a crystalline bonding environment for implantation at -196 ◦C. TheXANES spectra also confirms formation of a crystalline bonding environment forlow implanting temperature in agreement with Raman measurements.


Figure 3.7 (a) and (c) show representative k3-weighted EXAFS spectra for crys-talline standards and 4 at.% Ge samples, as a function of implantation temperatures.From a comparison of the two standards, the spectra of both samples implanted at

0 2 4 6 8 1 0 1 2 1 4- 5 0- 4 0- 3 0- 2 0- 1 0

01 02 03 0 a - S i 8 0 G e 2 0

2 0 0 o C 4 0 0 o C a - G e

W a v e n u m b e r k ( Å - 1 )

c )

a - S i 8 0 G e 2 0 2 0 0 o C 4 0 0 o C a - G e 0 2 4 60

2

4

R a d i a l d i s t a n b c e ( Å )

FT [k3 *χ(

κ) (Å

-4 )] d )0 2 4 6 8 1 0 1 2 1 4- 5 0

- 4 0- 3 0- 2 0- 1 0

01 02 03 0

k3 *χ(κ)

(Å-3 )

c - S i 8 0 G e 2 0 - 1 9 6 °C c - G e

k3 *χ(κ)

(Å-3 )

c - S i 8 0 G e 2 0 - 1 9 6 °C c - G e

a )

0 2 4 60

2

4

FT [k3 *χ(

κ) (Å

-4 )] b )

3 r d2 n d

Figure 3.7: EXAFS spectra of 4 at.% samples, after annealing at 1100 ◦C, as a function ofimplanting temperature plus amorphous and crystalline GeSi standards.

high temperatures are similar to an amorphous Si80Ge20 standard and not amor-phous Ge. Moreover, the spectrum of samples implanted at -196 ◦C is similar toc-Si80Ge20 standard and not crystalline Ge standard. Figure 3.7 (b) and (d) show thecorresponding non-phase-corrected FT spectra of Ge samples implanted at differenttemperatures and standards. Refined structural parameters (coordination number,bond length (R) and Debye-Waller factor) for Si-Ge and Ge-Ge path of the first NNshell are listed in Table 3.2.

Table 3.2: Refined EXAFS fitting parameters of 4 at.% implanted at different temper-atures, plus crystalline and amorphous standards.

System Implanting temperature (◦C) CNGe (atoms) RGe (Å) CNSi (atoms) RSi (Å) σ2 (10−3)c-Ge 4 2.447±0.001 2.2±0.1c-Si80Ge20 1.4±0.3 2.429±0.008 3.6±0.4 2.408±0.003 2.6±0.34 at.% -196 1.2±0.4 2.438±0.004 3.7±0.3 2.406±0.002 3.3±0.3a-Ge 3.4±0.3 2.463±0.007 3.7±0.5a-Si80Ge20 1± 0.1 2.457±0.007 3.1±0.6 2.42± 0.004 3.6±0.74 at.% 200 1±0.4 2.46±0.2 2.8±0.3 2.418±0.7 4.1±0.24 at.% 400 1±0.1 2.46±0.3 2.8±0.2 2.417±0.5 4.1±0.2


3.4 Summary- Low Ge concentrations (≤4 at.%)

No NPs were formed for Ge concentrations of ≤ 4 at.% independent of implantationtemperature or thermal annealing. A crystalline or amorphous Si-Ge bonding envi-ronment was observed - according to synchrotron-based techniques- for all examinedsamples independent of Ge concentration, annealing and implantation temperatures.

Crystallization and phase transition of the matrix were detected after annealingat 1100 ◦C for 4 at.% Ge samples, implanted at -196 ◦C. However, implanting athigh temperatures resulted in no phase transition of the matrix. This phenomena isdiscussed in more detail in Appendix A.

3.5 High Ge concentration samples (> 4 at.%)

Since no precipitation occurred for Ge concentrations up to 4 at.%, the study pro-ceeded to investigate high Ge concentrations. High concentration refers to sampleswith peak Ge concentrations of 6, 9 and 12 at.%, implanted with single energy ionsas listed in Table 3.3. Implantations were performed at 400 ◦C to promote in-situgrowth of NPs. These samples were then annealed at 700, 900 and 1100 ◦C for 1 hrin N2 gas to promote Ge precipitation and the growth of the NPs. The results ofhigh concentration samples are discussed in Section 3.5. The following sections sum-

Table 3.3: Summary of Ge implantation parameters. The fluence of each sample isgiven at corresponding energies E (keV).

E=1300 Ge concentration (at.%) Implantation temperature (◦C) Annealing temperature(◦C)

2.4×1017 6 400 700, 900 and 1100

3.6×1017 9 400 700, 900 and 1100

4.8×1017 12 400 700, 900 and 1100

marize the impact of concentration on the formation, size evolution, and structuralproperties of NPs. The characterizing techniques used to investigate the propertiesof NP are explained in Section 3.2.

3.5.1 RBS measurements

Figure 3.8 shows RBS spectra for the 12 at.% Ge samples before and after annealingat 1100 ◦C. Comparing the as-implanted and 1100 ◦C sample, redistribution of Geatoms is apparent for the annealed sample. Ge atoms migrated both towards the sur-face and towards the Si/Si3N4 interface. Similar behaviour was observed for otherconcentrations after annealing at 1100 ◦C. Appendix A further discussed this obser-vation. However, no redistribution of Ge atoms was observed for lower annealingtemperatures. A slight shift of Ge atoms from the surface could be resulted from

§3.5 High Ge concentration samples (> 4 at.%) 45

2 0 0 0 3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4

0 . 6

0 . 0 0 . 5 1 . 0 1 . 5 2 . 002468

1 01 21 41 6

A s - i m p l a n t e d 1 1 0 0 ° C

D e p t h ( µm )

Conce

ntratio

n (at.%

)No

rmaliz

ed cou

nts (ar

b. unit

s)


a )

b ) A s - i m p l a n t e d 1 1 0 0 ° C

Figure 3.8: (a) RBS spectra for 12 at.% sample before and after annealing at 1100 ◦C and (b)Ge as a function of Si3N4 depth.


formation of an oxide layer on the surface.

No loss of Ge through evaporation and/or diffusion was detected beyond theSi3N4/Si interface. The lack of diffusivity of Ge at annealing temperatures below1100 ◦C could be attributed to the N content of the film [100, 101], which formsa strong covalent bond with Si as well as higher density of Si3N4 similar to lowconcentration samples.

3.5.2 XRD

The crystallographic phases of the Ge implanted samples annealed at 1100 ◦C werestudied as a function of Ge concentration. Figure 3.9 shows XRD pattern for differentconcentrations after annealing at 1100 ◦C and the alpha structure pattern of Si3N4. It

3 0 3 5 4 0 4 5 5 0

0 . 0 1

0 . 0 2

0 . 0 3

0 . 0 4

0 . 0 5

0 . 0 6

1 2 a t . %

9 a t . %

a l p h a - S i 3 N 4

Norm

alized

Intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

6 a t . %

Figure 3.9: XRD pattern of Ge-implanted samples of different concentrations (after anneal-ing at 1100 ◦C/1 hr) with alpha Si3N4 pattern.

can be seen that, the matrix is poly-crystallized for all concentrations. All samplesrevealed structures resembling an alpha structure. As the concentrations increasedthe intensity of the peaks increased which could be a result of an increased grainsize and/or increase of volume fraction. For a better comparison XRD pattern of as-implanted samples were also measured (not shown here) which revealed no phasetransition for the matrix.


Comparing the RBS spectra (figure 3.8) and XRD results, it can be concludedthat redistribution of Ge atoms coincides with phase transition of the matrix. Poly-crystallization of the matrix eased the diffusivity of Ge atoms after annealing at 1100◦C. More details of the crystallization can be found in Appendix A. Such phasetransition was not detected for lower annealing temperatures.

3.5.3 TEM

The TEM images provided both qualitative and quantitative information of NP sizedistribution and shape. The NPs were spherical and crystalline for all concentrationsand annealing temperatures. The high resolution image of an individual NP con-firmed the cubic structure. For both cases, the size distribution was narrow with aFWHM of ∼ 2 nm. Figure 3.10 shows TEM images of the 9 and 12 at.% samplesannealed at 700 ◦C and respective size distributions, evaluated from the micrographscovering the extent of the implantation range. In addition, figure 3.11 presents TEMimages of the 9 and 12 at.% samples annealed at 900 ◦C, and respective size distribu-tions. Table 3.4 summarizes the average NP diameter extracted from TEM analysis.The NP formation was not observed for 6 at.% concentration after thermal annealing.

Table 3.4: The NP size as a function annealing temperature.System Annealing temp. (◦C) Size (nm)9 at.% 700 1.8±0.2

900 2.3±0.312 at.% 700 2.8±0.5

900 3.2±0.5

In the case of SiO2, NPs of different sizes (from 4-10 nm) were formed for compa-rable concentrations of Ge (3 at.%). Smaller NPs was formed in LPCVD Si3N4, whichis due to larger inter-facial energy of Ge/Si3N4 interface than Ge/SiO2 [102]. Fur-thermore, the mobility and diffusivity of Ge atoms is less in Si3N4 due to strong Si-Ncovalent bonds compared to flexible Si-O bonds. Hence, smaller NP would formedin Si3N4 matrix compare to SiO2.

Annealing at 1100 ◦C resulted in poly-crystallization of the matrix, therefore,these samples were not analysed further. Figure 3.12 (a), (b), (c) and (d) presents TEMimages for 12 at.% samples after annealing at 1100 ◦C. As observed previously, thematrix is poly-crystalline and it is not possible to differentiate between the crystallineseeds and the NPs.

3.5.4 XANES

The evolution of the crystallographic phase of the NPs was studied as a functionof annealing temperature by XANES. XANES spectra of Ge samples as a functionof annealing temperature (for different concentrations) are plotted in figure 3.13.


Figure 3.10: TEM images of NPs: 9 and 12 at.% Ge samples after annealing at 700 ◦C andrepresentative size distributions.


Figure 3.11: TEM images of NPs: 9 and 12 at.% Ge samples after annealing at 900 ◦C andrepresentative size distributions.


Figure 3.12: TEM images for 12 at.% samples annealed at 1100 ◦C : (a) the whole layer, (b) ahigh magnification image of the implanted region (c) implanted region with crystalline NPs,

and (d) diffraction pattern of the implanted region.


Figure 3.13 presents XANES spectra of crystalline Ge and Ge-rich GeSi (c-Si34Ge66)standards. Both spectra have similar characteristic features which can be used to fin-gerprint chemical bonding, electronic configuration, and site symmetry around theabsorber. Figure 3.13 shows results for 9 at.% and 12 at.% Ge annealed at 700 and

1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 0- 0 . 4- 0 . 20 . 00 . 20 . 40 . 60 . 81 . 01 . 21 . 41 . 61 . 82 . 02 . 22 . 4

c - S i 2 2 G e 7 8

c - G e

9 a t . %1 2 a t . %

a - S i 8 0 G e 2 0

7 0 0 ° C 9 0 0 ° C

Absor

ption

(arb. u

nits)

E n e r g y ( e V )

c - S i 3 4 G e 6 6

Figure 3.13: XANES spectra at Ge K-edge for crystalline and amorphous standards, 9 and12 at.% samples annealed at 700 and 900 ◦C. Spectra are shifted for clarity.

900 ◦C, which have characteristic features more similar to the c-Si34Ge66 standardthan c-Ge. In terms of increasing annealing temperature and/or concentration, nosignificant changes can be distinguished from the XANES spectra of NP samples.A combination of the XANES and FT spectra of NP samples revealed that the NPspectra are very similar to the c-Si34Ge66 then c-Ge standard. Also, no indication ofGe-N bonds1 is apparent for any sample. The XANES spectrum of NP samples ap-peared as a combination of amorphous and crystalline structures. However, becauseno amorphous standard was measured, an estimate of the crystallographic portionof NP samples were not achieved.

3.5.5 EXAFS

Figure 3.14 a), c) and e) display k3-weighted spectra as a function of annealing tem-perature and implanted Ge concentration, for standards and 9 and 12 at.% Ge sam-ples, respectively. In general, the EXAFS amplitude increases with increasing anneal-ing temperature, consistent with the process of NP nucleation and growth. As can beseen this trend was also observed for other concentrations. Figure 3.14 b), d) and f)

1A comparison of XANES spectra of crystalline Ge, GeSi, GeO2 and Ge3N4 are presented in Ap-pendix C.


0 2 4 6 8 1 0 1 2- 4 0- 3 0- 2 0- 1 0

01 02 0 a )

S t a n d a r d s

0 2 4 6 8 1 0 1 2- 4 0- 3 0- 2 0- 1 0

01 02 0 c ) 9 a t . %

0 2 4 6 8 1 0 1 2- 4 0- 3 0- 2 0- 1 0

01 02 0

0 1 2 3 4 50

1 0

2 0

3 0

c - G e S i 2 2 G e 7 8 S i 3 4 G e 6 6

b )

0 1 2 3 4 50

1 0

2 0

7 0 0 ° C 9 0 0 ° C

7 0 0 ° C 9 0 0 ° C

d )

0 1 2 3 4 50

1 0

2 0e ) 1 2 a t . %

FT [k

3 *χ(k)

(Å-3 )]

FT [k

3 *χ(k)

(Å-3 )]

f )

FT [k

3 *χ(k)

(Å-3 )]

R a d i a l d i s t a n c e ( Å )

k3 *χ(k)

(Å-3 )

k3 *χ(k)

(Å-3 )

k3 *χ(k)

(Å-3 )

W a v e n u m b e r k ( Å - 1 )Figure 3.14: EXAFS oscillations obtained after background removal, multiplied by k3 forbulk standard and GeSi NP samples. (a), (c) and (e) k3-weighted EXAFS and correspond-ing non-phase-corrected FT spectra (b), (d) and (f) for crystalline standards, 9 and 12 at.%

samples after annealing at 700 and 900 ◦C.


show the corresponding Fourier transforms (FTs) as a function of radial distance (us-ing a k-window of 3.8-12.5). The first nearest neighbour shell is observed at ∼ 2.01 Å.

For all concentrations the intensity of the second and third nearest neighbour arenegligible. The lack of a strong signal from the second and third nearest neighbourcan be attributed to the very small NP size, confirmed by TEM results. Furthermore,small size together with strain configuration different from bulk and possible recon-struction close to the NPs interface could be the cause of the loss of the signal fromthe second and third nearest neighbour. However, at a concentration of 6 at.%, Gesamples showed no sign of second or third nearest neighbours, consistent with noNP of formation. Concentrations below 6 at.% are below the solubility limit of Si3N4.For those concentrations there are no precipitates and only amorphous covalent Si-Ge bonding was observed.

The NP samples exhibited a reduced amplitude with respect to the bulk, due tohigher structural disorder and lower coordination number. As the annealing temper-ature increased the amplitudes increased, which as mentioned, is consistent with apattern of annealing-temperature-dependant NP formation and growth. The ampli-tude of NPs also decreased as the concentration decreased.

Atoms bonding with Ge with a shorter bond length (such as O and N) showedto the FT spectra at lower radial distances. No sign of Ge-O (∼ 1.7 Å) or Ge-N(∼ 1.8 Å) bonds were observed for all examined samples. It was observed that thefirst peak position shifted slightly towards the lower bond length due to the bondcontribution from Si atoms. As the concentrations and/or annealing temperaturesvaried, the position of the first peak also changed, which could be due to a changeof the stoichiometry of NP samples. In summary, a signification fraction of the FTspectra for all examined samples demonstrated Ge atoms bonding with Si atoms,though the stoichiometry of a GeSi bond for NP varies slightly as a function ofconcentration and/or annealing temperature.

3.5.6 Size-dependent atomic structure

The refined structural parameters from the FT spectra of examined samples are listedin Table 3.5. As mentioned, a decrease in the FT amplitude of NP samples is due

Table 3.5: Structural parameters -coordination number (CN), bond length R, andDebye-Waller factors (σ2)- obtained for the first NN shell of NPs.

System Annealing temp. (◦C) Size (nm) CNGe (atoms) RGe (Å) CNSi (atoms) RSi (Å) σ2 (10−3Å2)9 at.% 700 1.8±0.2 0.9±0.1 2.446±0.008 1.6±0.2 2.382±0.01 3.3±0.3

900 2.3±0.3 1.3±0.2 2.445±0.007 1.7±0.2 2.391±0.013 3.2±0.212 at.% 700 2.8±0.5 1.6±0.1 2.447±0.004 1.4±0.1 2.387±0.008 3.5±0.4

900 3.2±0.5 1.8±0.1 2.441±0.005 1.2±0.1 2.389±0.004 3.2±0.5

to a decrease in the average coordination number, therefore an increase in structural


disorder can be observed. In Table 3.5, the three refined first nearest neighbour struc-tural parameters (from EXAFS) are correlated with the mean NP size (from TEM).All structural parameters are clearly size dependent, demonstrating that finite-sizeeffects can be measured for the given range of NP diameters.

The CN of Ge-Ge bond increased as the NP size increased and as the NP sizeincreased the fraction of Si decreased. Increase of the CN as a function of NP size isa clear manifestation of finite-size effects where increase in surface-to-volume ratioyields a relative increase in uncoordinated surface atoms, which are independent ofthe crystallographic phase at the NP surface or interface with the matrix.

No significant changes were observed for the values of the bond length. The bondlength of NPs changed slightly as a function of NP size consistent within experimen-tal error.

The Debye-Waller factor values increased within the error bars as a result of adecrease in NP size due to the relative decreased fraction of surface atoms whichresulted in an increase in structural disorder.

3.6 Summary- high Ge concentrations (>4 at.%)

Table 3.6: Summary of matrix and NP phase as a function of implantation and an-nealing temperatures for LPCVD matrix.

Implantation temperature (◦C) Ge concentration (at.%) Annealing temperature (◦C) Matrix phase NP phase-196, 200 and 400 0.1-4 850-1100 amorphous NO NP (GeSi alloys)400 6 700 amorphous NO NP (GeSi alloys)400 900 amorphous NO NP (GeSi alloys)400 1100 crystalline crystalline GeSi400 9 700 amorphous crystalline GeSi400 900 amorphous crystalline GeSi400 1100 crystalline crystalline GeSi400 12 700 amorphous crystalline GeSi400 900 amorphous crystalline GeSi400 1100 crystalline crystalline GeSi

Table 3.6 summarizes the phase of the matrix and formed NPs as a function ofimplanted conditions.

Formation of GeSi NPs for concentrations of 9 and 12 at.% after annealing at 700and 900 ◦C.

Crystallization of the matrix was observed for concentrations of 9 and 12 at.%after annealing at 1100 ◦C.

The local and structural parameters of NP samples were determined using XANESand EXAFS measurements. XANES spectra prove formation of a Ge-rich GeSi bond-

§3.6 Summary- high Ge concentrations (>4 at.%) 55

ing environment for all NPs. It is readily apparent that the GeSi bond is preferable tothe GeGe bond even for high atomic concentrations. EXAFS results confirm that theCN is readily size dependant. Moreover, structural disorder and asymmetric devia-tion from a Gaussian-like bond length distribution enhanced - within the error bars-as the NPs size decreased. It can be concluded that finite-size effects are evident fromthe structural properties of NPs.

Chapter 4

Formation of Ge nanoparticles inPECVD Si3N4

This chapter discusses the formation of Ge NPs in amorphous PECVD Si3N4 layersas a function of synthesis parameters, including implanted Ge concentration andthermal annealing temperature. Spherical and single crystalline Ge NPs with cubicdiamond structural and narrow size distributions were obtained for concentrationsof ≥ 6 at.% after annealing at temperatures above 700 ◦C. The highest annealingtemperature (1100 ◦C) resulted in crystallization of Si3N4 matrix. XANES analysiswas used to qualify the bonding environment of NP samples. Structural propertieswere characterized using EXAFS, which confirmed finite-size effects. As the size ofthe NP decreased there was a notable increase in structural disorder and bond lengthand decrease in coordination number consistent with the non-negligible surface-to-bulk-ratio of NPs.

57

58 Formation of Ge nanoparticles in PECVD Si3N4

4.1 Experimental procedures

An amorphous Si3N4 layer of thickness of ∼ 1 µm was deposited on a Si (100)substrate by PECVD technique. Layers were deposited at the Australian NationalFabrication Facility (ANFF) on c−Si(100) wafers with an Oxford Plasma Technology100PECVD system. The SiH4:NH3:N2:N2O gases with the ratio of 8:14:980:0 (in sccm)were mixed. The substrate was kept at temperature of 650 ◦C and the growth ratewas 9.9 nm/min.

74Ge ions were implanted into the Si3N4 layer at an energy of 1 MeV to fluencesof 2×1017, 3×1017 and 4×1017 cm−2 corresponding to peak concentrations of 6, 9 and12 at.% respectively. To induce in-situ NP precipitation, the substrate was maintainedat 400 ◦C during implantation. Samples were then annealed at 700, 900 and 1100 ◦Cunder a N2 atmosphere for 1 hour.

RBS measurements were performed with 4.5 MeV with He2+ ions using a Sisurface-barrier detector positioned at a scattering angle of 168◦. The use of this en-ergy enabled the complete separation of the Ge depth distribution from the substratesignal extending from the surface to the Si/Si3N4 substrate interface. RBS spectrawere analysed with the RUMP program [77].

Conventional XRD was utilized to identify the phase transformation of the matrixand the NPs. Measurements were carried out with the sample surface oriented at aglancing incident angle of 0.7◦ to the x-ray beam.

Cross-section TEM images were obtained in bright field mode for quantitativeanalysis of the NP size distribution. Average NP diameters for different concentra-tions were calculated by averaging over NP volume through cubic weighting of theextracted NP diameters.

XANES and EXAFS measurements were performed at the Ge K-edge (11.103 keV)and a temperature of 15 K. Data were collected to a photoelectron wavenumber (k) upto a value of 15 Å−1. Background subtraction, spectra alignment, normalization andstructural refinement of EXAFS data were performed following the methods definedin Chapter 2. Spectra were then fitted with a single-scattering path about the Geabsorber to probe the structural parameters of the NN shell.

4.2 Results and discussion

4.2.1 RBS and XRD

Figure 4.1 shows the Ge concentration profiles measured by RBS for different concen-trations and annealing temperatures. The insets show the Ge signal after subtractionof the Si3N4 background and its conversion from energy to depth scale using the

§4.2 Results and discussion 59

3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4

3 0 0 0 4 0 0 00 . 0

0 . 2

0 . 4 a s - i m p l a n t 7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

a ) 9 a t . %

a s - i m p l a n t 7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 002468

1 01 21 4

D e p t h ( µm )


Conce

ntratio

n (at.%

)a )No

rm. co

unts (a

rb. un

its)

b ) 1 2 a t . %

0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 002468

1 0No

rm. co

unts (a

rb. un

its)

Conce

ntratio

n (at.%

)

D e p t h ( µm )

Figure 4.1: RBS spectra for 9 and 12 at.% Ge samples as a function of energy and depthdistribution of different concentrations and annealing temperatures.

mean-energy-approximation [78] for as-implanted and 1100 ◦C annealed samples.

We note that, the profiles for the 9 and 12 at.% samples show similar behaviour.Up to 900 ◦C the Ge profiles are near Gaussian similar to that predicted by SRIMprogram [75]. In the as-implanted state, the peak concentrations were centred at adepth of ∼0.4 µm with a full width at half maximum (FWHM) of ∼0.3 µm. In com-parison profiles for samples annealed at 1100 ◦C show significant Ge redistribution.Ge atoms have migrated towards both the Si3N4 surface and the Si/Si3N4 interface.There is also slight diffusion of Ge into the Si substrate. However, the total amountof Ge in each sample remained constant after annealing.

Figure 4.2 (a) and (b) show TEM images for 12 at.% samples annealed at 900 and1100 ◦C overlaid with the depth profiles of Ge atoms calculated from RBS. The sam-ple annealed at 900 ◦C figure 4.2 (a), has a Gaussian-like Ge profile, as predicted bySRIM, while at the higher temperature of 1100 ◦C the profile is completely changed.The TEM image of 12 at.% sample annealed at 1100 ◦C (figure 4.2 (b)) shows theimplanted region is noticeably different from other annealing temperatures. Similar


Figure 4.2: TEM images of 12 at.% Ge samples after annealing at (a) 900 ◦C and (b) 1100 ◦C,respectively. Ge profiles determined from RBS measurements are included for reference.

results were observed after annealing at 1100 ◦C for other Ge concentrations. Theoverlaid spectra show limited diffusion of Ge through the Si/Si3N4 interface.

Figure 4.3 show TEM images for the different Ge concentrations after annealingat 1100 ◦C plus representative XRD pattern for each concentration. XRD data revealsan amorphous-to-crystalline phase transition of the implanted matrix after annealingat 1100 ◦C. The XRD pattern of the transformed phase resembles that of alpha Si3N4

ans as the Ge concentration increases the volume fraction of crystallinity increases.


Figure 4.3: TEM images for different Ge concentrations after annealing at 1100 ◦C pluscorresponding XRD pattern of examined samples.


The crystallization of the matrix was only observed in the case of thermal anneal-ing at 1100 ◦C, similar to LPCVD implanted layers, and is clearly correlated with thelong range migration of Ge. We considered possible mechanisms for crystallisationincluding loss of N2, ion implantation damage, and change of the stoichiometry Ap-pendix A. Here we focus on characterisation of Ge NP produced at temperatures of700 and 900 ◦C.

4.2.2 TEM

Figure 4.4 displays TEM images and extracted NP size distributions for different Geconcentrations, after annealing at 700 ◦C. The NPs were spherical and crystalline forall concentrations. Mean NP diameters were calculated by averaging the NP volumethrough cubic weighting of the extracted NP diameters, with results summarizedin Table 4.1. As the implanted concentration increased, the size of the NPs also in-creased. Ge NPs were not observed near the Si3N4/Si interface or the sample surfacein agreement with the RBS measurements 4.2. The NP distribution maintained aGaussian-like profile.

Table 4.1: NP size as a function of annealing temperature.Sample Annealing temp. (◦C) NP size (nm)6 at.% 700 1.88±0.3

900 2.27±0.369 at.% 700 2.37±0.38

900 2.42±0.4412 at.% 700 3.00±0.8

900 4.25±1.05

Figure 4.5 shows TEM images and their representative size distributions for dif-ferent concentrations after annealing at 900 ◦C. The size distributions are Gaussian-like throughout the implanted range. As above, all NPs were spherical and crys-talline for all concentrations. High resolution images are shown in the figure 4.5insets where (111) lattice planes of the Ge diamond cubic structure are discernible. Itcan be seen that the NP size increases as the concentration increased.

The overall TEM results show that NP size slightly increases as annealing tem-perature and/or Ge concentration increases. Comparing the results with LPCVD, wenote that even for a 6 at.% concentration, small NPs were formed after annealing at700 and 900 ◦C.

4.2.3 Raman

Raman measurements for Ge NPs and bulk amorphous and crystalline standardswere performed to study the relationship between experimental parameters and theparticle size. Figure 4.6 shows Raman spectra for Ge NPs and a/c-Ge standards. The


Figure 4.4: TEM images recorded near the Ge concentration maximum for all at.% Gesamples after annealing at 700 ◦C with respective NP size distributions.


Figure 4.5: TEM images recorded near the Ge concentration maximum for all at.% Gesamples after annealing at 900 ◦C with respective NP size distributions.


phase evolution of Ge NP through annealing temperatures and concentrations can beinvestigated quantitatively. In general for each examined annealing temperature, asthe concentration of Ge increased, the main peak position shifted to higher frequen-cies and the FHWM decreased. The spectrum noticeably changed for the annealingtemperature of 1100 ◦C.

For all annealing temperatures, we observe a relatively broad and asymmetricpeak centred around ∼ 291 cm−1, indicative of the quasi amorphous fraction of GeNPs. We also observed a sharp peak at ∼ 300 cm−1 accompanied by a wide shoulderon the low frequency side due to the crystalline fraction of Ge NPs. By removing theSi substrate, interference of the second-order Raman peak from Si is eliminated. Peakbroadening, redshift and asymmetry were observed for all NP samples, relative tothe crystalline standard. The broadening of the NP Ge-Ge mode is the result of struc-tural disorder and size effect, while redshift can be attributed to compressive strainand/or an isotopic effect [103]. Low frequency asymmetric broadening and tailingof the Raman spectra [104, 105] may be further attributed to finite-size effects[106]and an amorphous component. A quantitative analysis was conducted and detailsare presented in Appendix D.

1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 00

5 0 0 0

1 0 0 0 0

1 5 0 0 0

2 0 0 0 0 c ) 1 1 0 0 ° C

W a v e n u m b e r ( c m - 1 )

Intens

ity (ar

b. unit

s) a ) 7 0 0 ° C

1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 00

2 0 0 0

4 0 0 0

6 0 0 0

6 a t . % 9 a t . % 1 2 a t . %

b ) 9 0 0 ° C

1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 002 0 0 04 0 0 06 0 0 08 0 0 0

1 0 0 0 0 c - G e a - G e

Figure 4.6: Raman spectra for 6, 9 and 12 at.% Ge samples as a function of annealingtemperature (700, 900 and 1100 ◦C) plus crystalline and amorphous standards.


At 1100 ◦C, in addition to the sharp feature associated with Ge NPs, the spectrumdisplays a weak but clearly discernible peak at ∼ 400 cm−1 indicating the formationof a SiGe alloy likely at Si/Si3N4 interface.

4.2.4 XANES measurements

Figure 4.7 shows XANES spectra for the Ge NP samples plus crystalline and amor-phous Ge standards. The spectra more closely resembles c-Ge than a-Ge, from which

1 1 0 8 0 1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 0 1 1 2 0 0- 0 . 4- 0 . 20 . 00 . 20 . 40 . 60 . 81 . 01 . 21 . 41 . 61 . 8

7 0 0 ° C 9 0 0 ° CAb

sorpti

on (ar

b. unit

s)

E n e r g y ( e V )

c - G e

a - G e6 a t . %9 a t . %1 2 a t . %

Figure 4.7: XANES spectra for 6, 9 and 12 at.% Ge samples as a function of annealingtemperature (700, 900 and 1100 ◦C) plus crystalline and amorphous standards.

we infer that the phase fraction of an amorphous component is minimal at most.This similarity becomes less noticeable for lower concentration and annealing tem-peratures, due to a decrease in NP size. Equivalently, the c-Ge like features of the NPspectra demonstrate that the phase fraction of a crystalline component is dominant.Indeed, linear combination fitting indicats an amorphous fraction was negligible.Also, no evidence of Ge nitride was apparent from the XANES analysis.

The amorphous and crystalline phase fractions of the Ge NPs were determinedfrom XANES spectra. The Ge NP spectra were fitted with a liner combination1 of

1The linear combination fit method is employed here to quantify the fractions of absorbing atoms ineach environment, with

µcalc = ∑ f jµj (4.1)

Where fj used to fit the experimental data. By minimizing the difference between the linear combinedspectrum and the measured spectrum, the fj values are approximated. The spectrum of the sample of


the two bulk standards. The fitting was performed with ATHENA over an energyrange of -20 to 60 eV above the absorption edge. Figure 4.8 (a) and (b) demonstratesthat the XANES spectra of Ge NPs are accurately reproduced as a combination ofbulklike amorphous and crystalline environments. Also, figure 4.8 (c) shows the

1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 0 1 1 2 0 00 . 00 . 20 . 40 . 60 . 81 . 01 . 21 . 41 . 6

1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 0 1 1 2 0 00 . 00 . 20 . 40 . 60 . 81 . 01 . 21 . 41 . 6

1 . 6 1 . 8 2 . 0 2 . 2 2 . 4 2 . 6 2 . 8 3 . 0 3 . 2 3 . 4 3 . 6 3 . 8 4 . 0 4 . 2 4 . 45 56 06 57 07 58 08 59 0

E x p e r i m e n t a l d a t a c - G e c o m p o n e n t a - G e c o m p o n e n t L i n e a r c o m b o f i t

a - G e = 1 2 . 2 %

c - G e = 8 7 . 8 %

c ) N P c r y s t a l l i n e f r a c t i o n

E x p e r i m e n t a l d a t a c - G e c o m p o n e n t a - G e c o m p o n e n t L i n e a r c o m b o f i t

a ) 3 . 0 0 n m N P s

a - G e = 2 3 %

c - G e = 7 7 %

E n e r g y ( e V )

Phase

fractio

n

N P d i a m e t e r ( n m )

b ) 4 . 2 5 n m N P s

Absor

ption

(arb. u

nits)

E n e r g y ( e V )

Figure 4.8: XANES spectra at the Ge K-edge for 12 at.% samples a) annealed at 700 and (b)annealed at 900 ◦C with linear combination fits to experimental data. (c) fitted crystalline

fraction as a function of NP size.

fraction of Ge atoms in a bulklike crystalline environment as a function of NP size.Clearly, the crystalline fraction decreases as the NPs become smaller or, equivalently,the surface-to-volume ratio increases.

4.2.5 EXAFS

Figure 4.9 (a), (c) and (e) show k3-weighted EXAFS spectra as a function of anneal-ing temperature and implanted Ge concentration. In general, the EXAFS amplitudeincreases for higher temperature or Ge concentration, consistent with NP formationand growth. Figure 4.9 (b), (d) and (f) show corresponding non-phase-corrected

interest is modelled by weighting the spectrum of each known standard (amorphous and crystallineGe) j with factor fj and adding them together. Factors fi represent the fraction of contribution from eachstandard in the sample and ideally they should sum to one.


FT spectra. As mentioned above, the NP samples exhibited a reduced amplitude

0 2 40

1 0

2 0

3 0

k3 *χ(k)

(Å-3 )

FT [k3 *χ

(k) (Å

-4 )]FT

[k3 *χ(k)

(Å-4 )]

FT [k3 *χ

(k) (Å

-4 )]

k3 *χ(k)

(Å-3 )

f )

6 a t . % 9 a t . % 1 2 a t . %7 0 0 ° C

W a v e n u m b e r k ( Å - 1 )

k3 *χ(k)

(Å-3 )

0 2 4 6 8 1 0 1 2 1 4

- 1 0

0

1 0

e )

6 a t . % 9 a t . % 1 2 a t . %

0 2 40

1 0

2 0

3 0

d )

6 a t . % 9 a t . % 1 2 a t . %

0 2 4 6 8 1 0 1 2 1 4

- 1 0

0

1 0 9 0 0 ° C

c )

6 a t . % 9 a t . % 1 2 a t . %

0 1 2 3 4 50

1 0

2 0

3 0 b )

a )

0 2 4 6 8 1 0 1 2 1 4- 3 0- 2 0- 1 0

01 02 03 0


a - G e

S t a n d a r d s

c - G e

Figure 4.9: EXAFS (c and e) and FT (d and f) spectra for 9 and 12 at.% Ge samples as a func-tion of annealing temperature (700 and 900 ◦C) plus crystalline and amorphous standards.

with respect to the bulk, indicating higher structural disorder and/or a lower coor-dination number. The existence of the second and faint third NN in the FT spectracan be observed for all NP samples. As the NP size decreased, the intensity ofthese peaks reduced due to the decreasing crystallinity fraction, consistent with theXANES quantifications presented above. The refined fitting parameters for the firstNN shell, including coordination number and bond length, the Debye-Waller fac-tor and asymmetry (third cumulant) of the NPs were determined from the EXAFSspectra.

4.2.6 Size-dependent atomic structure

Table 4.2 presents the first shell coordination number, bond length, Debye-Wallerfactor and third cumulants, respectively, as a function of NP size. As the NP size in-creases, the relative fraction of under-coordinated surface atoms decreases and thusthe average CN slightly increases. The increase in size, and consequently decrease

§4.3 Summary 69

in surface-to-volume ratio, yields an increase in structural disorder. The increase ofstructural disorder is due to bonding distortions at or near the NP/matrix interface.Bond length slightly varies as the NP size decreases. The bond length decreases as

Table 4.2: Structural parameters of NPs as a function of annealing temperature.Sample Annealing temp. (◦C) NP size (nm) CN (atoms) R (Å) σ2 (10−3Å2) C3(10−4Å3)c-Ge 4 2.445±0.001 2.5±0.2 0.0±0.36 at.% 700 1.88±0.3 2.5±0.1 2.436±0.005 4.1±0.1 0.4±0.6

900 2.27±0.36 3.6±0.3 2.432±0.003 4.4±0.6 0.2±0.59 at.% 700 2.37±0.38 3.1±0.1 2.439±0.002 4.0±0.6 0.2±0.3

900 2.42±0.44 3.7±0.1 2.434±0.002 4.2±0.5 0.1±0.312 at.% 700 3.00±0.8 3.2±0.1 2.441±0.002 3.7±0.7 0.2±0.3

900 4.25±1.05 3.7±0.7 2.438±0.001 3.7±0.4 0.1±0.2a-Ge 3.88±0.16 2.447±0.004 4.5±0.2 -0.3±0.7

the NP diameter increases. Bond length contraction, observed in other studies formetallic NPs [107, 108, 109], were not observed here, consistent with previous resultsfor Ge NPs embedded in SiO2 [44].

A decrease in NP size yields an increase in structural disorder as apparent fromthe results. The increase in Debye-Waller factor is characteristic of so-called finite-sizeeffects as the surface-to-volume ratio increases. For semiconductor NPs embedded inSiO2, simulations [110] and experiments [44] have demonstrated that the interfacialatom environment is highly distorted accommodate material and bonding differenceacross the interface.

Third cumulant (C3) is the asymmetric deviation from a Gaussian interatomicdistance distribution. As the NP size decreases, a slight increase in the deviation isnoticeable within error bars. In principle, a broad NP size distribution may also leadto a non-symmetrical bond length distribution. The effect of a non-symmetrical bondlength distribution is not operative here as the Ge NPs have a small Gaussian sizedistribution as shown in figure 4.4 and figure 4.5.

4.3 Summary

Table 4.3 are summarizing the effects of implantation conditions on the matrix andNP phase. In summary, this chapter has investigated the concentration and annealingtemperature dependent structure of Ge NPs formed in amorphous PECVD Si3N4. ANP size range of 1.8 to 4.2 nm was determined by TEM for different concentrationsand annealing temperatures. Narrow size distributions were obtained for all cases.The NPs were also spherical in shape and crystalline, retaining the cubic phase forall concentration and annealing temperatures. Comparing to LPCVD matrix, NPprecipitation in PECVD matrix occurred at concentration of 6 at.% and with a slightincrease in the size distribution. However, comparing to SiO2 for both matrices theNP size is significantly smaller.


Table 4.3: Summary of matrix and NP phase as a function of implantation and an-nealing temperatures for PECVD matrix.

Implanting temperature (◦C) Ge concentration (at.%) Annealing temperature (◦C) Matrix phase NP phase400 6 700 amorphous crystalline Ge400 900 amorphous crystalline Ge400 1100 crystalline crystalline Ge400 9 700 amorphous crystalline Ge400 900 amorphous crystalline Ge400 1100 crystalline crystalline Ge400 12 700 amorphous crystalline Ge400 900 amorphous crystalline Ge400 1100 crystalline crystalline Ge

In general, the diffusivity of Ge atoms in LPCVD and PECVD Si3N4 is slow andhence the annealing temperature has less influence on the ripening and formation ofcrystalline NPs. Precipitation and NP formation occurred for all examined concen-trations and annealing temperatures.

Formation of a Ge-Ge bonding environment was observed for all concentrationsand annealing temperatures up to 900 ◦C. The crystallinity fraction of NPs was ex-amined. A decrease in NP size and increase in surface-to-volume ratio yielded aslightly bigger amorphous fraction.

We note that all structural parameters were both concentration and annealingtemperature dependent within the error bars. Although, due to small size of NPno significant trend were observed for bond length, DW factor and C3. Increasingconcentration and annealing temperature resulted in an increase in CN significantlywhich revealed the average coordination number of NPs was less than the bulk stan-dard, in agreement with finite size effect.

Chapter 5

Formation of Ge NPs in SiOxNymatrices of different stoichiometry

SiOxNy films have a tunable refractive index, from SiO2 with a refractive index of 1.46through to SixN1−x films of refractive index >2.7. For the present chapter, SiOxNy

layers of different stoichiometry (SiO1.67N0.14 and SiO1.12N0.37) were implanted withGe ions in order to investigate the role of the host matrix on the structural propertiesof embedded NPs. These stoichiometries were chosen as they are very close to thestoichiometry of SiO2 and Si3N4, respectively. The present study seeks to investigatethe effect of N content on the atomic structure of Ge NPs synthesised in SiOxNy byion implantation and thermal annealing, and compare the results with NPs formedin SiO2 and Si3N4.

The crystalline and amorphous fractions were determined with XANES whilethe structural parameters of the first NN shell were determined with EXAFS. Theseresults were supplemented by RBS, TEM and Raman spectrometry measurements.RBS was used to measure the depth distribution of Ge atoms, TEM to determinethe NP size distribution and Raman spectroscopy to support the XANES results.Using this combination of complementary techniques, we achieve a detailed pictureof the size-dependent structural properties of Ge NPs and show they are governedby finite-size effects.

71

72 Formation of Ge NPs in SiOxNy matrices of different stoichiometry

5.1 Experimental procedure

Amorphous SiO1.67N0.14 and SiO1.12N0.37 layers of thickness ∼1 µm were depositedon Si (100) substrates by plasma-enhanced chemical vapour deposition with the sto-ichiometry deduced by RBS. As-deposited, the H content is typically ∼1-5 at.%, de-creasing to a negligible fraction after annealing above 1100 ◦C. Ge ions were im-planted into the SiOxNy layers at an energy of 700 keV to fluences of either 2.7 or3.3×1017cm−2. To promote in-situ NP nucleation and growth, the substrate tempera-ture was maintained at 400 ◦C during implantation. Samples were then annealed for1 hr in a N2 ambient at 700, 900 or 1100 ◦C.

Silicon oxynitride films were deposited at the Australian National FabricationFacility on Si(100) wafers with an Oxford Plasma Technology 100PECVD system.The SiH4:NH3:N2:N2O chemistry was controlled to vary the relative concentrationof Si, O and N [111]. The growth parameters and layer compositions are shown inTable 5.1. Refractive index and thickness were measured by spectral reflectometryusing a SCI Filmtek 4000 with a spectral range of 440 to 1660 nm at normal and 70◦

incidence. Stoichiometry was determined by RBS with 2.4 MeV He+ ions. For thisstudy we considered the values obtained from RBS.

Table 5.1: The stoichiometry of SiOxNy matrices determined with spectral reflectom-etry and RBS measurements

Sample Spectral reflectometry RBSSiOxNy XSiO2 XSi3 N4 x y x ySiO1.67N0.14 0.948 0.052 1.82 0.08 1.67 0.14SiO1.12N0.37 0.608 0.392 1.09 0.61 1.12 0.37

We consider an error of 0.002 for the refractive index, which means an error of 1.4% for the values ofXSiO2 and XSi3 N4 . RBS has a higher error, 2.8% for SiO2 and 3.5% for Si3N4, we have a 3.13% error from

the value of SiO2 and 3.8% from Si3N4 value.

RBS measurements were performed with 3.5 (for SiO1.12N0.37 matrix) and 4.5 (forSiO1.67N0.14 matrix) MeV He2+ ions using a Si surface-barrier detector positioned at ascattering angle of 168◦. The use of these energies enabled the complete separation ofthe Ge depth distribution from the substrate signal extending from the surface to theoxynitride/substrate interface. RBS spectra were analysed with the RUMP program[77].

Cross-sectional TEM images were obtained in bright-field mode using a JEOL2100F microscope operated at 200 kV. Samples were prepared using a standard ion-beam-milling protocol. Average NP diameters for different concentrations were cal-culated by averaging over NP volume through cubic weighting of the extracted NPdiameters.

XANES and EXAFS measurements were performed at the Ge K-edge (11.103 keV)and a temperature of 15 K using the x-ray absorption spectroscopy beamline of the

§5.2 Formation of Ge nanoparticles in SiO1.67N0.14 73

Australian Synchrotron. Fluorescence spectra were recorded with a 10×10 Ge pixel-array detector. Data were collected to a photoelectron wavenumber (k) of 15 Å−1.

5.2 Formation of Ge nanoparticles in SiO1.67N0.14

5.2.1 Introduction

In this section we investigate the formation of Ge NPs in amorphous SiO1.67N0.14 as afunction of implanted atom concentration and thermal annealing temperature. Whilethe composition of SiO1.67N0.14 is close to that of SiO2, the addition of N yields a muchreduced NP size relative to those formed in SiO2 under comparable implantation andannealing conditions [112]. We attribute this difference to reduced diffusivity of Gein the oxynitride matrix.

5.2.2 RBS

The depth distribution of implanted Ge atoms in the amorphous SiO1.67N0.14 layerwas measured by RBS before and after thermal annealing, an example of which isshown in figure 5.1. In the as-implanted state, the peak concentrations for the two

2 5 0 0 3 0 0 0 3 5 0 00 . 0

0 . 2

0 . 4

0 . 6

0 . 8

1 . 0 a s - i m p l a n t 1 1 0 0 ° C


Norm

alized

counts

0 2 0 0 0 4 0 0 0 6 0 0 0 8 0 0 0 1 0 0 0 002468

1 01 21 4

at. %

D e p t h ( Å )

Figure 5.1: RBS spectra before and after annealing at 1100 ◦C for samples of 12 at.% Ge.

implantation fluences were ∼9 and ∼12 at.% and were centred at a depth of ∼0.3µm with a full width at half maximum (FWHM) of ∼0.3 µm. The depth of the


concentration maximum determined by RBS agreed with SRIM predictions [75]. Fig-ure 5.1 shows that thermal annealing does not lead to significant redistribution ofGe atoms. Similar observations have been reported previously for Ge NPs formed inSiO2 [44, 113]. Acquiring the spectra at the energy of 4.5 MeV resulted in the appear-ance of nuclear resonances from different isotopes of N, O and Si. The presence ofmultiple resonances impedes accurate normalization. Both spectra were normalizedat the low-medium energy end of the spectrum, below the energy of the resonances.

5.2.3 TEM

Figure 5.2 shows TEM images and NP size distributions for concentrations of 9 and12 at.% after annealing at 900 ◦C. Spherical, crystalline NPs were observed in all

Figure 5.2: TEM images recorded near the Ge concentration maximum for 9 and 12 at.% Gesamples annealed at 900 ◦C and their representative size distributions.

annealed samples. High resolution images are shown in the insets and (111) lattice


planes of the Ge diamond cubic structure are discernible. The average diametersare listed in Table 5.2 and show that the concentration and annealing temperatureinfluence the average NP size poorly.

Table 5.2: Ge NP size as a function of implantation and annealing temperature.sample Annealing temp.(◦C) NP size (nm)9 at.% 900 3.0±0.4

1100 3.7±0.712 at.% 900 4.1±0.7

1100 4.5±0.8

Representative TEM images and NP size distribution for concentrations of 9 and12 at.% after annealing at 1100 ◦C are shown in figure 5.3. Similar to other annealing

Figure 5.3: TEM images recorded near the Ge concentration maximum for 9 and 12 at.% Gesamples annealed at 1100 ◦C and their representative size distributions.


temperatures, the size distributions are Gaussian-like. Mean NP diameters of NP arelisted in Table 5.2. The NP size clearly increases as the annealing temperature and/orGe concentration increases.

For annealing temperature and time (1100 ◦C, 1 hr), the NP sizes determined inSiO2 (10 at.%: 14 nm, 3 at.%: 6.4 nm) are significantly larger than those determinedhere for SiO1.67N0.14 with higher Ge concentrations (9 at.%: 3.7 nm, 12 at.%: 4.5nm). Ge NPs formed in SiO1.67N0.14 are smaller, indicative of a slower rate of NPgrowth. The presence of N therefore have a significant effect on the developmentof Ge NPs. This is likely due to a reduced Ge diffusivity in SiO1.67N0.14 comparedto SiO2, consistent with the observations of Mirabella et al. for Si3N4 [4]. The twomatrices differ considerably in atomic density (6.82×1022 and 1.03×1023 atoms/cm3,respectively) and the greater number of bonds per unit volume in the nitride mayfurther impede diffusion. The three bonds connected to each N in Si3N4 are alsomore constraining than the two bonds of each O atom in SiO2 [100]. These moreconstrained bonds contribute to the reduced diffusivity in nitrided lattices [100].

5.2.4 XANES

The amorphous and crystalline phase fractions of the Ge NPs were determined withXANES. Representative spectra are shown in figure 5.4 for Ge samples of differentconcentrations as a function of annealing temperature. The spectra more closelyresemble that of c-Ge than a-Ge, from which we infer that the crystalline componentis dominant. Indeed, linear combination fitting indicated the amorphous fraction wasnegligible, consistent with the TEM observations presented above. Also, no evidenceof Ge nitride, oxide or silicide formation was apparent from the XANES analysis.While Si and Ge are completely miscible and readily alloy, clearly Si preferentiallybonds to O and N in this case.

5.2.5 Raman measurements

Raman spectra are shown in figure 5.5 for SiO1.67N0.14 films implanted with differentfluences of Ge samples of different concentrations as a function of annealing temper-ature. Further analysis of Ge-Ge Raman-active mode can be found in Appendix D.As discussed previously, there is no interference from the second-order Raman peakfrom Si due to the sample preparation method. Relative to the crystalline standard,peak broadening, a redshift and asymmetry are observed. The broadening of the NPGe-Ge mode is the result of structural disorder while the redshift can be attributed tocompressive strain and/or an isotopic effect [103]. The asymmetric broadening is theresult of finite-size effects [106]. Scattering contributions from crystalline Ge-Si werenot apparent, consistent with the XANES results. The presence of an a-Ge phasefraction could contribute to the observed peak broadening. However, no evidence ofan amorphous phase component was apparent with XANES, as noted previously.


1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0

- 0 . 4

0 . 0

0 . 4

0 . 8

1 . 2

1 . 6

2 . 0

7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

9 a t . %

a - G e

c - G e

Absor

ption

(arb. u

nits)

E n e r g y ( e V )

1 2 a t . %

Figure 5.4: XANES spectra for 9 and 12 at.% Ge samples as a function of annealing tem-peratures (700, 900 and 1100 ◦C) plus crystalline and amorphous standards. The spectra are

shifted for clarity.

1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0- 5 0 0 0

0

5 0 0 0

1 0 0 0 0

1 5 0 0 0

2 0 0 0 0

1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0

- 5 0 0 0

0

5 0 0 0

1 0 0 0 0 b ) 1 2 a t . %

7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

a ) 9 a t . % 7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C c - G e

Intens

ity (ar

b. unit

s)

W a v e n u m b e r ( c m - 1 )

W a v e n u m b e r ( c m - 1 )Figure 5.5: Raman spectra for 9 and 12 at.% Ge samples as a function of annealing temper-

ature (700, 900 and 1100 ◦C) plus crystalline standard. The spectra are shifted for clarity.

5.2.6 EXAFS

The structural parameters of the first NN shell surrounding a Ge atom were deter-mined with EXAFS. Figures 5.6 (a), (c) and (e) show isolated k3-weighted EXAFS


spectra while figures 5.6 (b), (d) and (f) show FT spectra of the standards and sam-ples with different Ge concentrations and annealing temperatures. For the crystalline

4 6 8 1 0 1 2- 3 0

- 2 0

- 1 0

0

1 0

2 0

e ) 1 2 a t . %

d ) 9 a t . %c ) 9 a t . %

b ) a )

0 2 40

1 0

2 0

3 0 c - G e a - G e


FT[ k3 ∗χ

(k) (Å

-3 )]FT

[ k3 ∗χ(k)

(Å-3 )]

FT[ k3 ∗χ

(k) (Å

-3 )]

W a v e n u m b e r k ( Å - 1 )

s t a n d a r d s

4 6 8 1 0 1 2- 3 0

- 2 0

- 1 0

0

1 0

2 0

4 6 8 1 0 1 2- 3 0

- 2 0

- 1 0

0

1 0

2 0

0 2 40

1 0

2 0

3 0 7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

k3 ∗χ(k)

(Å-3 )

k3 ∗χ(k)

(Å-3 )

k3 ∗χ(k)

(Å-3 )

f ) 1 2 a t . %0 2 40

1 0

2 0

3 0 7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

Figure 5.6: EXAFS (c and e) and FT (d and f) spectra for 9 and 12 at.% Ge samples as afunction of annealing temperature (700, 900 and 1100 ◦C) plus crystalline and amorphous

standards. The spectra are shifted for clarity.

and amorphous bulk standards, the EXAFS spectra shown in figure 5.6 (a) are theresult of scattering from multiple and single atomic shells surrounding a Ge atom,respectively. Indeed, for a-Ge, no scattering contribution from beyond the first NNshell is evident in the FT spectrum of figure 5.6 (b), consistent with the structural dis-order inherent in the amorphous phase. In contrast, scattering contributions from thefirst, second and third NN shells are apparent for c-Ge, consistent with the diamondcubic structure. For the NP samples, the amplitude of the first NN peak decreases asthe annealing temperature decreases. This is the result of the decrease in size, andhence increase in surface-to-volume ratio.

Table 5.3 summaries the refined fitting parameters (coordination number, bondlength, Debye-Waller factors, and third cumulants) of the first shell as a function ofNP diameter, determined with TEM. It can be seen that the coordination number


decreases as the NPs become smaller. This is a clear manifestation of finite-size ef-fects where the increase in surface area-to-volume ratio yields a relative increase inundercoordinated surface atoms.A mean bond length contraction were not observed for NP samples as the NP size

Table 5.3: Refined EXAFS structural parameters as a function of implantation andannealing temperature. Coordination number (CN), bond length (R), Deby-Waller

factor (σ2) and third cumulant (C3) parameters are listed.sample Annealing temp.(◦C) NP size (nm) CN (atoms) R(Å) σ2(10−3Å2) C3 (10−5Å3)c-Ge 4 2.445±0.001 2.5±0.2 0.0±0.39 at.% 900 3.0±0.4 3.3±0.3 2.457±0.004 4.4±0.5 2±3.1

1100 3.7±0.7 3.5±0.1 2.453±0.005 4.1±0.7 2.6±3.312 at.% 900 4.1±0.7 3.5±0.3 2.452±0.008 3.8±0.4 2±5

1100 4.5±0.8 3.6±0.2 2.451±0.005 3.7±0.7 2.1±3.2a-Ge 3.9±0.2 2.447±0.004 4.5±0.2 -0.3±0.7

decreased, as is detected for metallic NPs due to capillary pressure. Also, previousworks reported for embedded NP smaller than 4 nm, the ST-12 phase was preferableto diamond phase [114, 115]. No such results were observed in this study.

While within the error estimate results for Deby-Waller factor suggest a slightincrease in structural disorder with decreasing NP size due to the increased surface-to-volume ratio, yielding both a reduction in average coordination number and anincrease in structural disorder. The latter is expected due to bonding distortions ator near the NP/matrix interface [116].

The near-zero value of third cumulant, C3, is indicative of a lack of asymmetryin the inter-atomic distance distribution, consistent with the random orientation ofindividual NPs.

5.2.7 Summary of SiO1.67N0.14 matrix

Spherical Ge NPs were formed in SiO1.67N0.14 for concentrations of 9 and 12 at.%after thermal annealing at 900 and 1100 ◦C.

No long-term diffusion was observed for Ge atoms after thermal annealing (es-pecially at 1100 ◦C) in SiO1.67N0.14 matrix.

The structural properties of the NPs - CN, bond length and DW factor- quanti-fied here are in agreement with the size dependent trends previously observed forGe NPs in SiO2 [44].

The most striking difference is the decrease in NP size observed in the oxynitridewhich again demonstrates the very significant influence of matrix composition onthe size distribution of Ge NPs. The size of Ge NPs can therefore be controlled bythe very slight addition of N to a nominally oxide matrix.


5.3 Formation of Ge nanoparticles in SiO1.12N0.37

5.3.1 Introduction

This section summarises the results of Ge implantation in a SiO1.12N0.37 matrix. Theconsistent energy, fluences and annealing temperatures for a straightforward com-parison of matrix impacts on the size evolution and structural properties of embed-ded NPs. However, the film morphology was significantly modified by implantationwhen compared to the SiO1.67N0.14 film after implantation. As a consequence NPsize distributions were not measured by TEM. From the extracted refined structuralparameters obtained from EXAFS it can be predicted that the average NP size tobe slightly smaller than that obtained in the SiO1.67N0.14 matrix. The details of thestoichiometry and growth process can be found in Table 5.1.

5.3.2 Results and discussion

5.3.2.1 RBS

Figure 5.7 (a) shows the 12 at.% Ge concentration profiles measured by RBS as afunction of annealing temperature, and figure 5.7 (b) presents the depth profile ofas-implanted Ge distribution for the 12 at.% sample. In the as-implanted state, peakconcentrations for the 12 at.% implanted fluence were centred at a depth of ∼ 0.4µm with a FWHM of ∼ 0.3 µm. The depth of the concentration maximum deter-mined by RBS was 0.1 µm less than that predicted by SRIM calculation. Similar toprevious observations, figure 5.7 shows that thermal annealing did not lead to a sig-nificant loss or long-range redistribution of Ge atoms in either towards the surfaceor Si/SiO1.12N0.37 interface.

5.3.2.2 Raman measurements

Figure 5.8 shows the Raman spectra for Ge NPs and the c-Ge standard as a functionof concentrations and annealing temperatures. Figure 5.8 (a) presents data for the9 at.% Ge NP samples. For all annealing temperatures, a peak appears around 300cm−1 with asymmetrical broadening over the low frequency of the spectrum. Thispeak becomes stronger as the annealing temperature is increased. The annealingtemperature dependence of Raman spectrum also depends on the Ge concentration.Figure 5.8 (b), shows the Ge peak becomes sharper as the Ge concentration increases.As mentioned before, the broadening of the NP Ge-Ge mode is the result of struc-tural disorder, while the redshift can be attributed to compressive strain [103]. Theasymmetric broadening is the result of finite-size effects [106]. For the same concen-tration and annealing temperature, we note that the peak intensity of present NPsslightly decreased comparing with SiO1.67N0.14 matrix.


2 0 0 0 3 0 0 00 . 0

0 . 2

0 . 4

0 . 6

0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 002468

1 01 21 4 b )

a s - i m p l a n t 9 0 0 ° C 1 1 0 0 ° C

Norm

. count

s (arb.

units)


a )Co

ncentr

ation (

at.%)

D e p t h ( µ m )Figure 5.7: RBS spectra for the 12 at.% Ge samples as a function of annealing temperature

plus the Ge depth profile.


2 0 0 4 0 00 . 0

0 . 2

0 . 4

2 0 0 4 0 0

0 . 2

0 . 4

W a v e n u m b e r ( c m - 1 )

b ) 1 2 a t . % 7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C c - G e

Intens

ity (a

rb. un

its)

a ) 9 a t . %

7 0 0 ° C 9 0 0 ° C 1 1 0 0 ° C

W a v e n u m b e r ( c m - 1 )Figure 5.8: Raman spectra for 9 and 12 at.% Ge samples as a function of annealing temper-

ature plus crystalline standard.

5.3.2.3 TEM

Figure 5.9 (a) presents an image of the whole layer, and we note that around theimplanted band, two lines of white feature cavities are noticeable. Figure 5.9 (a)-(d)shows TEM images of the 12 at.% sample after annealing at 900 ◦C. The same obser-vation was evident for other concentrations and annealing temperatures. Figure 5.9(b) is a higher resolution image close to the Si/SiO1.12N0.37 interface. The boundarybetween the end of cavities and amorphous, non-implanted region is clearly distin-guishable. Figure 5.9 (d) presents a magnified image of the tail of the implantedregion, which shows an accumulation of Ge inside cavities and the nucleation ofsmall NPs. Formation of cavities could be due to decrease of the H content of thematrix or implantation-vacancies. Figure 5.9 (c) is a TEM image of the implantedregion close to the surface. A high resolution image of an individual NP reveals theformation of spherical (near the surface) and crystalline NP where (111) lattice planesof the Ge diamond cubic structure are discernible, similar to previous observations.TEM images of samples annealed at 1100 ◦C did not show any phase transition forthe SiO1.12N0.37 matrix. Note that, the TEM images were characterised qualitativelyand no NP size distribution was extracted.

Figure 5.10 presents a high angle annular dark field image of a 12 at.% sampleafter annealing at 900 ◦C for energy dispersive spectroscopy (EDS) analysis. Differentspots were chosen to characterise and map the composition of NPs and cavities. Eachspot refers to a number from 1 to 5 and the representative map is also presented withthe same numbers. For almost all spots, except spot number 5, we can see Ge signalsin addition to Si and N signals. The intensity of related peaks (Kα and Kβ) variesas the spot locations change. Spot number 5 is very close to the surface where noNP formed and only some cavities formed. Spots number 1 and 3 are located inside


Figure 5.9: TEM images of a 12 at.% sample after annealing at 900 ◦C, including (a) thewhole layer, (b) end side of implanted region, (c) implanted region and (d) inside the cavities.


Figure 5.10: EDS spectra of a 12 at.% sample after annealing at 900 ◦C and correspondingmapping of different spots.

cavities and, as predicted, a large concentration of Ge is trapped therefore its signalis significantly increased. However, spots number 2 and 4 present a decrease in theintensity of the Ge peaks due to accumulation of smaller NPs.

5.3.2.4 XANES

Ge K-edge XANES spectra are shown in figure 5.11 for the bulk standards and NPsamples as a function of annealing temperature and concentration. The prominentfeatures of the c-Ge and NP spectra resulted from multiple-scattering resonances,which at the Ge K-edge is the excitation of a 1s photoelectron with features abovethe absorption edge. The a-Ge standard exhibits a less fine structure due to the lossof multiple scattering contributions. XANES analysis reveals that the spectra of NPsamples are a combination of the amorphous and crystalline phase, however, thecrystallinity fraction of NPs decreased as concentration and/or annealing tempera-ture decreased. Similar to previous result (LPCVD and PECVD Si3N4 and SiO1.67N0.14

matrices), there was no evident of oxide, nitride or Ge-Si bonding components.


1 1 0 8 0 1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 0 1 1 2 0 0- 0 . 20 . 00 . 20 . 40 . 60 . 81 . 01 . 21 . 41 . 61 . 8

9 a t . %1 2 a t . %a - G e

9 0 0 ° C 1 1 0 0 ° C

Absor

ption

(arb. u

nits)

E n e r g y ( e V )

c - G e

Figure 5.11: XANES spectra of standards as compared to a selection of Ge NP samples.Thespectra are shifted for clarity.

5.3.2.5 EXAFS

Figure 5.12 displays k3-weighted EXAFS spectra as a function of wave number kfor bulk standards and NP samples. A k-window of 4.4-12.5 −1 was used and it isshown with a red line in the EXAFS spectra. Similar to previous data, a reductionin amplitude of the Ge NP spectra compared with the standards spectra is apparent.Moreover, the EXAFS amplitude increases slightly with an increasing annealing tem-perature, consistent with NP growth.

Figure 5.13 (a), (b) and (c) show the corresponding non-phase-corrected FT spec-tra of bulk standards and NP samples, as a function of radial distance. The amplitudeof the FT spectra is highest for the c-Ge standard and decreases with lower annealingtemperatures or concentrations. Due to an increased average CN and lower structuraldisorder. The first peak is due to a single-scattering events between the absorbingatom and the first NN shell of atoms. The signal at higher distance is the super-position of both single-scattering photoelectron paths from the more distant shellsand the contributions from multiple scattering paths. As the concentration and/orannealing temperature decreases, the second and third NN peak amplitudes alsodecrease and become slightly broader. Consistent with previous observation.


0 2 4 6 8 1 0 1 2- 4 0- 3 0- 2 0- 1 0

01 02 03 04 05 06 0

9 0 0 ° C 1 1 0 0 ° C

9 a t . %

1 2 a t . %

a - G e

k3 *χ(k)

(Å-3 )

W a v e n u m b e r k ( Å - 1 )

c - G e

Figure 5.12: EXAFS spectra of different concentrations and annealing temperatures com-pared to that of bulk standards. The spectra are shifted for clarity.

5.3.2.6 Structural properties

The Ge-Ge analysis for the refined structural parameters of the first NN shell for NPsamples summarized in Table 5.4. The obtained results reveal a concentration and

Table 5.4: NP size and structural parameters as a function of implantation and an-nealing temperature extracted by first NN shell EXAFS analysis for a SiO1.12N0.37

matrix.sample Annealing temp.(◦C) CN (atoms) R(Å) σ2(10−3Å2) C3 (10−4Å3)c-Ge 4 2.445±0.001 2.5±0.2 0.0±0.39 at.% 900 2.7±0.3 2.456±0.006 3.5±0.6 1.3±9

1100 3.2±0.2 2.455±0.003 3.2±0.3 0.8±5.412 at.% 900 3.2±0.2 2.454±0.003 3.2±0.3 0.3±1.6

1100 3.3±0.3 2.450±0.006 2.9±0.4 0.8±7a-Ge 3.9±0.2 2.447±0.004 4.5±0.2 -0.3±0.7

annealing temperature dependence in all structural parameters as observed qualita-tively in the FT spectra of figure 5.13. In general, a suppressed CN and decreasedbond length is apparent together with a slightly increased Debye-Waller factor andC3 term as NP size decreased. A decrease of the average CN of lower concentrationsand annealing temperatures is a result of the significantly lower coordination state ofthe surface atoms. The increase in the Debye-Waller factor is consistent with surface


0 50

5

1 0

1 5

2 0

2 5

3 0

0 50

5

1 0

1 5

2 0

2 5

3 0

0 50

5

1 0

1 5

2 0

2 5

3 0

c - G e a - G e

9 0 0 ° C 1 1 0 0 ° C

b ) 9 a t . %

9 0 0 ° C 1 1 0 0 ° C

c ) 1 2 a t . %

a ) s t a n d a r d s


FT [k3 *χ

(k) (Å

-3 )]FT

[k3 *χ(k)

(Å-3 )]

FT [k3 *χ

(k) (Å

-3 )]

Figure 5.13: Non-phase-corrected FT EXAFS spectra as a function of radial distance com-paring NP samples to the amorphous and crystalline Ge standards.


reconstruction and/or disorder. Furthermore, an increase in bond length and C3 arealso observed as the annealing temperature increased.

5.3.3 Summary of SiO1.12N0.37 matrix

In summary, we have investigated the structural properties of Ge NPs synthesized inSiO1.12N0.37 by ion implantation and thermal annealing. No long-range diffusion wasobserved for Ge atoms after thermal annealing (especially at 1100 ◦C) in SiO1.12N0.37

matrix.

The NP size was not measured, though all NPs were spherical (near the surface)and/or crescent shape and crystalline, retaining the diamond cubic phase.

XANES spectra of NP samples demonstrated the formation of Ge-Ge bonding en-vironment for all concentrations and annealing temperatures, consistent with Ramanmeasurements.

Cavities formed after ion implantation and thermal annealing, independent ofannealing temperature and concentrations. The size of the cavities appeared to varyas a function of concentrations. The mechanism behind cavities formation was notinvestigated here but is clearly of future study.

Chapter 6

Summary and Conclusion

This chapter summarizes the main conclusions and key findings from the researchpresented in this thesis, as well as possible experimental improvements and futuredirections.

89

90 Summary and Conclusion

6.1 This work

Metalloid NPs (Ge and SiGe) were formed in amorphous SiOxNy by ion implantationand thermal annealing. The formation, size evolution and short range atomic struc-ture of NPs were studied as a function of implantation concentration and annealingtemperature. Two types of Si3N4 (LPCVD and PECVD) plus PECVD SiO1.67N0.14 andSiO1.12N0.37 with different stoichiometries were studied. The key findings of whichare summarized below.

Properties of SiGe NPs embedded in a LPCVD Si3N4 matrix

Ge ions were implanted into 2 µm LPCVD Si3N4 to maximum concentrations of0.1, 0.5, 1.7 and 4 at.%. Implantations were performed at -196 ◦C and samples werethen annealed at 850, 925 and 1100 ◦C. It was found that the combination of low Geconcentration and high annealing temperature did not lead to NP formation. Higherconcentrations (6, 9 and 12 at.%) were then considered. For the highest concentra-tions of 9 and 12 at.%, SiGe NPs were formed after annealing at 700 and 900 ◦C for 1hr in a N2 atmosphere. In order to promote the in-situ growth of NPs, these sampleswere implanted at 400 ◦C. Clearly, solubility limits of Ge in LPCVD Si3N4 is of theorder ∼ 6 at.%.

The SiGe NPs were found to be spherical and crystalline, exhibiting the cubicdiamond structure. The NP size was found to be dependent on the implanted Geconcentrations. NP size increased from 2.8 to 3.2 nm for 12 at.% Ge concentrationsamples after annealing temperature was changed from 700 to 900 ◦C for 1 hour.Also, NP size increased from 2.3 to 3.2 nm for samples annealed at 900 ◦C for 1 hourwhen concentration was increased from 9 to 12 at.%.

Finite-size effects were readily apparent in SiGe NPs structure. As the size ofNPs decreased the structural properties changed, including a reduction in CN andenhanced structural disorder.

Properties of Ge NPs embedded in a PECVD Si3N4 matrix

Ge ions were implanted into 1 µm PECVD Si3N4 to maximum concentrations of6, 9 and 12 at.%. The samples were then annealed at 700, 900 and 1100 ◦C for 1 hr ina N2 atmosphere.

The Ge NPs which formed were spherical, ∼ 4 nm in diameter and crystallinewith the cubic diamond structure. NP size was found to be increased as functions ofGe concentration and annealing temperature. NP size varied from 2.4 to 4 nm (forsamples annealed at 900 ◦C for 1 our) when concentration increased form 9 to 12at.%. Moreover, as annealing temperature was increased from 700 to 900 ◦C NP sizeincreased from 3 to 4.5 nm for 12 at.% Ge samples .

§6.1 This work 91

EXAFS measurements revealed the short range atomic structure deviated fromthat of the bulk reference. Four structural parameters of the first NN shell weremeasured: CN, bond length, DW factor and third cumulant. CN of NPs exhibitssize-dependent structural properties.

The vibrational properties of Ge NPs were probed with Raman spectroscopy.Peak broadening, redshift and asymmetry were observed for all NP samples, relativeto the crystalline standard.

Comparison of LPCVD and PECVD Si3N4 matrix

Annealing at 1100 ◦C resulted in long-range diffusion and crystallization of thematrix for Ge concentrations of (≥ 4 at.%) in both LPCVD and PECVD Si3N4 ma-trices. It clearly lowered the crystallization temperature of an um-implanted Si3N4

(1600-1800 ◦C). Formation of nanosilicide particles, change of the matrix stoichiom-etry (due to ion implantation and bond breaking) and induced-damage mechanismsignificantly are found to lower crystallization temperature.

LPCVD Si3N4 layers were implanted at 10 at.% H and then co-implanted witha Ge ion energy and fluence similar to the highest Ge PECVD concentration. Af-ter implantation each sample was annealed at 700 and 900 ◦C. This experimentwas performed to investigate whether H implantation affects the bonding of Ge ions(more details in Appendix B). XANES spectra of examined samples revealed a Si-Ge bonding environment was achieved, independent of H implantation. In fact, Himplantation did not significantly affect the structural properties of embedded NPs.Therefore, one reason for the formation of SiGe NPs in LPCVD matrix could be cor-related to the film stoichiometry (Si-rich film), which requires further precise RBSmeasurements.

Properties of Ge NPs embedded in PECVD SiOxNy matrices

Ge ions were implanted into 1 µm PECVD SiO1.67N0.14 and SiO1.12N0.37 matricesup to maximum concentrations of 9 and 12 at.%. The samples were then annealed at700, 900 and 1100 ◦C for 1 hr in a N2 atmosphere.

For both matrices, it appears that annealing at 1100 ◦C did not result in long-range diffusion of Ge. SiO1.12N0.37 matrix contained cavities near the surface of thelayer and at the end of the implantation range, hence large NPs (with non-sphericalshape) were trapped inside cavities.

The Ge NPs in SiO1.67N0.14 were single-crystalline and diamond cubic in struc-ture with finite-size effects readily measurable from EXAFS parameters. Annealingtemperature and Ge concentration slightly influenced NP size. As Ge concentrations


was changed from 9 to 12 at.% NP size increased from 3.7 to 4.5 nm for samplesannealed at 900 ◦C for 1 hour. When annealing temperature was increased from 700to 900 ◦C NP size increased from 4.1 to 4.5 for 12 at.% Ge concentration.

Raman and XANES measurements both confirmed the formation of Ge NPs.

Summary and discussion

Figure 6.1 summarises the NP sizes in SiOxNy matrices compared to SiO2 [44, 117]for different concentration and annealing temperature. NP size in SiOxNy matrices

2 3 4 5 6 7 8 9 1 0 1 1 1 20

2

4

6

8

1 0

1 2

1 4

NP di

amete

r (nm)

G e c o n c e n t r a t i o n ( a t . % )

L P - 7 0 0 ° C L P - 9 0 0 ° C P E - 7 0 0 ° C P E - 9 0 0 ° C S i O 1 . 6 7 N 0 . 1 4 - 9 0 0 ° C S i O 1 . 6 7 N 0 . 1 4 - 1 1 0 0 ° C S i O 2 - 1 1 0 0 ° C [ R e f . 4 4 , 1 1 6 ]

N O N P si n L P C V D S i 3 N 4

P E C V D S i 3 N 4

Figure 6.1: Size comparison of NPs for given concentrations and annealing temperatures.

are noticeably smaller compared to SiO2. This can be attributed to the presence of N,low diffusivity of Ge atoms in SiOxNy

1 and a larger interfacial energy between Geatoms and SiOxNy [4]. From NP size the partial value of diffusion coefficient can be

1Diffusivity of Si in SiO2 is about 3×10−13 cm2/s, [118] while it decreased to 1×10−24 cm2/s inSi3N4 at similar temperature [101]

§6.1 This work 93

estimated using the following equation [119]

D =9Tr3

4t(6.1)

where t and T are the annealing time and temperature, and D is the diffusion coeffi-cient. By using the NP sizes from figure 6.1, the partial value of diffusion in SiOxNy

and SiO2 is to be close to (2/3)3.

Figure 6.1 shows that no NP were formed for concentrations below 6 at.%, whichsuggests that solubility limit is of this order.

From figure 6.1 we can conclude that, annealing temperature and Ge concentra-tion has significant effect on NP size, through observation of NP diameters from 1.8to 4.5 nm.

In this study, XANES was utilized to investigate the bonding environment of NPs.Different bonding environments were observed for LPCVD and PECVD NPs. Stoi-chiometry of the host matrix may have had an effect on the local bonding of NPs.Furthermore, EXAFS refined parameters from the first NN - including CN, DW fac-tors, bond length and C3- were used to study the size dependent structural propertiesof NPs. In general, a decrease in CN was observed as the NP size decreased.

Ion-implantation with Ge and other species lowered the crystallization temper-ature of Si3N4, with extensive crystallization observed after annealing at 1100 ◦C,compared to un-implanted crystallization temperature of 1600-1800 ◦C. The mainobservations are listed as follow:

• No crystallization was observed for lower Ge concentrations (<4 at.%).

• Crystallization was inhibited for elevated implantation temperatures (200 and400 ◦C) for 4 at.% Ge.

• Crystallization of the matrix for Ge concentrations of ≥6 at.% was observed in-dependent of implantation temperatures. Suggest a radiation-damage thresh-old in order of ≥ 1017 atoms/cm2.

• Crystallization of the matrix was occurred independent of chemical impurities(Co, Ni and Cu).

• Formation of silicide particles.

Phase transition of the matrix could be attributed to combination of mechanisms.In particular, formation of nanosilicide particles, change of the matrix stoichiometry(due to ion implantation and bond breaking) and induced-damage mechanism sig-nificantly reduce crystallization temperature.


Cavities and voids were formed in the samples with a projected range closer tothe surface and to the end of implantation band in SiO1.12N0.37. Future studies maytry to measure the size evolution of the cavities as a function of concentration andannealing temperature. Also, a qualitative model is required to justify the process ofcavity formation.

This thesis enhances the understanding of Ge NPs formation and their atomicand local structural parameters. It also contributes to the pathway of understandingof the interaction between the matrix and NPs. The results reported here couldlead to a significant improvement in the scientific study, technological relevance andapplication of metalloid NPs. A brief summary of NP and matrix phase, NP size andtype is listed in Table 6.1. These results demonstrate the potential for tailoring Ge

Table 6.1: Summary of NP size for Si3N4, SiOxNy and SiO2 matrix.System LPCVD Si3N4 PECVD Si3N4 SiO1.67N0.14 SiO1.12N0.37 SiO2

*

NP precipitation (at.%) >6 at.% ≥6 at.% n/m n/m 0.3 at.%NP type SiGe Ge Ge Ge GeNP size (nm) 1.8-3.2 1.9-4.3 and <2∗∗ 3-4.5 n/m 4-10 and 3-24∗∗

Poly-crystallization at 1100 ◦C YES YES NO NO NOCavities formed NO NO NO YES NO

∗Ref. [44, 45, 96, 97, 98, 117]∗∗ Ref. [4]

NP sizes and structural properties in SiOxNy matrices by controlling the oxynitridestoichiometry.

6.2 Future Work

The findings presented in this thesis can form the basis of a number of future studies,including:

• Vibrational properties of NPs: Temperature dependent EXAFS measurementscould be performed to investigate the vibrational properties of embedded NPs.Temperature dependent EXAFS measurement is a unique technique to studythe thermal characteristics of NPs for comparison with bulk standards.

• Swift heavy ion irradiation induced shape and structural transformations: Pre-vious studies have shown the shape and structure of Ge NPs embedded in SiO2

were changed after heavy ion irradiation. It would be interesting to see how themorphology of NPs embedded in LPCVD, PECVD Si3N4 and PECVD SiOxNy

matrices are modified by ion irradiation.

• Other stoichiometry: In order to get a better idea of the impact of N on thesize and structural properties of NPs it would be useful to examine a variety ofstoichiometries ranging from pure SiO2 to Si3N4.

§6.2 Future Work 95

• Optical and electronic properties: As a continuation of the NP formation studyin this thesis, it would be interesting to correlate optical and electronic proper-ties of NPs as a function of SiGe and matrix composition.

• Amorphous-to-crystalline phase transition: Further studies are required to in-vestigate the causes of poly-crystallization of the LPCVD and PECVD Si3N4

matrices at 1100 ◦C.

Appendix A

Crystallization of the matrix

A unique amorphous-to-crystalline phase transformation was observed for implantedSi3N4 films at a much lower temperature than usually required. Multiple techniqueswere used to study the phase transition of Si3N4 films after ion implantation andthermal annealing. Here, we isolated different mechanisms involved with the crys-tallization process. Crystallization of amorphous Si3N4 can usually be achieved athigh temperatures, 1600-1800 ◦C. However, in this study a phase transformationof LPCVD and PECVD Si3N4 films was observed, post implantation, after thermalannealing at 1100 ◦C. Ion implantation caused a reduction in the temperature ofcrystallization. The crystalline phase, is identified to be alpha Si3N4.

97

98 Crystallization of the matrix

A.1 Introduction

Si3N4 thin films are common structural materials used in the fabrication of variousmicro-electromechanical devices [120, 121, 122, 123]. Chemical vapour deposition(CVD) techniques are frequently used to deposit Si3N4 thin films. The most commontechniques to deposit amorphous Si3N4 thin films are low pressure CVD (LPCVD)and plasma enhanced CVD (PECVD). Crystallization of Si3N4 powder has been ex-tensively studied [124, 125, 126, 127]. Yanhui et al. [128] reported on the crystal-lization of an amorphous Si3N4 powder after Si powder was added. This processsignificantly reduced the time for crystallization and led to the formation of a sub-micrometre/nanosized alpha Si3N4. Post crystallization the powder was a mixture ofalpha Si3N4 and Si2N2O. The crystallization of LPCVD and PECVD Si3N4 thin filmshas been reported to a lesser extent [129, 130, 131].

Here, we focus on the crystallization dynamics of amorphous LPCVD and PECVDSi3N4 layers, implanted with different species after thermal annealing at 1100 ◦C. Weoutline the parameters that resulted in the crystallization of Si3N4.

A.2 Experimental procedure

LPCVD and PECVD Si3N4 grown on Si (100) substrates was implanted with Ge,Co, Ni, Cu and Si ions at temperatures of -196, 200 and 400 ◦C. Post implantationannealing of the different samples was performed at 1100 ◦C in N2, Ar, O2 andforming gas (N2 + H2) ambient for 1 and 10 hours. Table A.1 summarizes the typeof ions, implanting temperature, multiple fluence, energies and atomic concentrationof studied species for LPCVD and PECVD films.

Table A.1: Type of implanted species and implanting temperatures.Ion System Implantation temp (◦C) Energy (MeV) Fluence (ions/cm−2) Concentration (at.%)Ge LPCVD -196 and 400 0.8, 1.3 and 2 7.27, 1.05, 1.9, 2.4, 3.6 and 4.8 ×1017 0.1, 0.5, 1.7, 4, 6, 9 and 12Ge PECVD 400 1 2, 3 and 4 ×1017 6, 9 and 12Cu LPCVD -196, RT, 400 0.9, 1.3, 1.9 0.3, 0.9, 2.7 ×1017 0.3, 1, 3Ni LPCVD -196, RT, 400 0.9, 1.3, 1.9 0.3, 0.9, 2.7 ×1017 0.3, 1, 3Co LPCVD -196, RT, 400 0.9, 1.3, 1.9 0.3, 0.9, 2.7 ×1017 0.3, 1, 3

X-ray diffraction (XRD) patterns were collected using an XRD system (PanalyticalX′pert Pro Diffractometer) with a glancing incidence geometry (ω = 0.8) Cu-Kα ra-diation (45kV 40mA source). Data were then analysed using PowderCell 2.4 software.

Cross-section TEM images were obtained in bright field mode using a JOEL 2100microscope operated at 200 kV. Samples were prepared with a standard ion-beam-milling protocol.

The synchrotron-based technique of x-ray absorption near-edge structure spec-troscopy (XANES) was used to identify the bonding environment around a Ge ab-

§A.3 Results and discussion 99

sorber atom. XANES measurement at the Ge K-edge (11.103 keV) was performed ata temperature of 15 K at the XAS beamline of the Australian Synchrotron. Fluores-cence spectra were recorded with a 10×10 Ge pixel-array detector. Details of samplepreparation and data analysis and described in Chapter 2.

A.3 Results and discussion

A.3.1 Implantation-temperature dependence

In order to examine the effects of implantation temperature on the crystallizationprocess, LPCVD layers were implanted with Ge ions at different temperatures. Fig-ure A.1 presents XRD pattern of 4 at.% Ge samples as function of implantation tem-peratures after annealing at 1100 ◦C. Comparing spectra, crystallization only oc-

3 0 3 5 4 0 4 5 5 00 . 0

0 . 2

0 . 4

0 . 6

0 . 8

1 . 0

1 . 2

1 . 4

u n - i m p l a n t

a l p h a - S i 3 N 4

2 0 0 ° C4 0 0 ° C

Norm

alized

intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

- 1 9 6 ° C

4 a t . % G e

Figure A.1: XRD pattern of 4 at.% Ge samples implanted at different temperatures afterannealing at 1100 ◦C and un-implanted layer of LPCVD Si3N4 after annealing at the same

temperature. Spectra are shifted vertically for clarity.

curred for an implantation temperature of -196 ◦C. The crystalline peaks are similarto alpha phase of Si3N4 [132]. The spectra of un-implanted Si3N4 layer and elevatedGe implanted samples maintained amorphous structure after annealing at 1100 ◦C.

Figure A.2 shows TEM images of 4 at.% Ge samples and the un-implanted Si3N4

layer after annealing at 1100 ◦C for different implantation temperatures. ComparingTEM images, significant changes can only be detected from the images of the samplethat implanted at -196 ◦C. The un-implanted LPCVD Si3N4 is still amorphous afterannealing at 1100 ◦C. Also, higher implantation temperatures inhibit crystallization


Figure A.2: TEM images of an un-implanted LPCVD layer and 4 at.% Ge samples afterannealing at 1100 ◦C as function of implantation temperatures.

of the implanted matrix. It is clear that ion implantation at -196 ◦C results in severchanges near the implanted region. It is likely that elevated implantation temperaturecould compensate implantation damages to some extent. It can be concluded thatimplantation-induced-damages influenced crystallization mechanism.

A.3.2 Concentrations dependence

Figure A.3 presents the XRD pattern of LPCVD layers implanted with Ge, at concen-trations ≥ 6 at.%, after annealing at 1100 ◦C. It can be seen that for all concentrationsthe matrix is crystallized although implantation were performed at 400 ◦C. We notethat, all samples show characteristic peaks similar to the alpha phase of Si3N4 (in-dicated by star symbols). As Ge concentration increases the intensity of peaks alsoincreases. It is clear that as fluences exceeded a certain value (>1017) implantationtemperature no longer affected the crystallization mechanism.

Similar results were obtained for PECVD Si3N4 layers implanted with Ge. Im-plantation were performed at 400 ◦C and then samples were annealed at 1100 ◦C.Figure A.4 shows XRD pattern of PECVD layers as a function of Ge concentrations.All samples reveal a crystalline structure similar to alpha phase of Si3N4. Once again,we notice the diffraction peaks significantly decreases for lower concentration due toa decease of grain size or crystalline volume fraction. Also, a significant difference


3 0 3 5 4 0 4 5 5 0

0 . 0 1

0 . 0 2

0 . 0 3

0 . 0 4

0 . 0 5

0 . 0 6

1 2 a t . %

9 a t . %

a l p h a - S i 3 N 4

Norm

alized

Intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

6 a t . %

Figure A.3: XRD pattern of Ge samples after annealing at 1100 ◦C as a function of concen-tration. Spectra are shifted vertically for clarity.

2 5 3 0 3 5 4 0 4 5

0 . 2

0 . 4

0 . 6

0 . 8

1 . 0

1 . 2

1 . 4 a l p h a - S i 3 N 4 b e t a - S i 3 N 4

1 2 a t . %

9 a t . %

Norm

alized

Intens

ity (ar

b. unit

s)

2 θ ( d e g r e e )

6 a t . %

Figure A.4: XRD pattern of Ge implanted PECVD Si3N4 layers after annealing at 1100 ◦C asa function of concentration. Spectra are shifted vertically for clarity.

between the LPCVD and PECVD layer is, the appearance of beta phase in PECVDsamples -with fraction ∼10%- only for high Ge concentration (12 at.%). Beta structureis the stable crystalline phase of Si3N4. As the concentration increases, the number ofinduced-damages also increases. The fact that for similar Ge concentration PECVD


matrix shows a beta phase can be attributed to the density, stoichiometry and struc-tural difference between LPCVD and PECVD matrix. Also, existence of impurities-such as O2 and H2- could also effect on the crystallization phase. It is subject offurther studies and more details can be found here [133, 134].

Figure A.5: TEM images of (a), (b) whole layer plus (c) and (d) implanted region of LPCVDand PECVD Si3N4 layers, respectively.

Figure A.5 (a) and (b) present TEM images of LPCVD and PECVD layers. Bothsamples are crystallized (as observed in the XRD results) near the implantation band,which shows ion implantation significantly influences the crystallization dynamics.Figure A.5 (b) and (d) shows high magnified images of the implanted regions. In bothcases, defects produced by implantation in addition to voids and nano-crystallineprecipitants can be detected in the implanted region. It is suggested that the precipi-tation of embedded nano-silicide crystals act as nucleation centres for crystallization,hence silicide precipitates can be considered as crystalline nuclei.

A.3.3 Species dependence

Herbert et al. [135] reported on the phase transformation and degradation of LPCVDSi3N4 during cobalt silicide processing in N2 ambient. Similar results on silicideformation and crystallization of the LPCVD matrix were obtained in our study. Dif-ferent types of metallic (Co, Cu and Ni) impurities of different concentrations were


implanted at low or high temperatures into LPCVD Si3N4 layers. This process wasused to compare the effect of metallic and metalloid impurities on the crystalliza-tion mechanism. XRD patterns of un-implanted LPCVD layers, and layers implantedwith a common concentration (∼4 at.%) after annealing at 1100 ◦C are shown inFigure A.6. It can be seen that, independent of the type of impurity, all samples im-

2 5 3 0 3 5 4 0 4 5 5 0 5 5 6 0 6 5 7 0 7 5 8 0- 0 . 0 5- 0 . 0 4- 0 . 0 3- 0 . 0 2- 0 . 0 10 . 0 00 . 0 10 . 0 20 . 0 30 . 0 40 . 0 50 . 0 60 . 0 70 . 0 80 . 0 90 . 1 0

a l p h a - S i 3 N 4

u n - i m p l a n t e d S i 3 N 4

C o

N iC u

Norm

alized

Intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

G e

Figure A.6: XRD pattern of different species implanted in LPCVD Si3N4. All samples wereannealed at 1100 ◦C for 1 hour. Also, un-implanted Si3N4 (after annealing at 1100 ◦C) isshown. Arrows indicate the silicide peaks. Star symbols refer to the diffraction data of alpha

Si3N4. Spectra are shifted vertically for clarity.

planted at low and elevated temperatures revealed characteristic crystalline peaks.For all implanted species the crystalline phase is similar to the common alpha Si3N4

phase (represented by star symbols). For metallic species, a peak related to the sili-cide bond is highlighted by arrows. The un-implanted Si3N4 layer was still found tobe amorphous after annealing at the same temperature and conditions. According toTable A.2, it is clear that the melting points of the implanted ions and their silicideprecipitants are relatively above 1100 ◦C.

Table A.2: The melting points of implanted species and their silicide formSystem Element melting point (◦C) Silicide melting point (◦C) Ref.Ge 937 Si1−xGex T≈[1412-738x+263x2]

Ge0.75Si0.25 = 1056.5Ge0.5Si0.5 = 1176Ge0.25Si0.75 = 1295.5

[136]

Cu 1083 825 [137]Ni 1453 981-1455 [137]Co 1495 1325 [138]


A.3.4 Annealing-temperature dependence

The effect of annealing temperature on crystallization was studied by annealing par-ticular concentrations of Ge samples at different temperatures (950-1100 ◦C) for 1hour. Figure A.7 illustrates the XRD pattern of LPCVD Si3N4 layers implanted withGe. No sign of crystallization or Si-Ge bond formation was observed under 1000 ◦C.

2 5 3 0 3 5 4 0 4 5 5 00 . 0

0 . 5

1 . 0

1 . 5

2 . 0 a l p h a - S i 3 N 4

9 5 0 °C1 0 0 0 ° C1 0 2 5 °C

1 0 5 0 °CNorm

alized

intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

1 0 7 5 ° C

Figure A.7: XRD patterns of LPCVD Si3N4 layers implanted with Ge and annealed at dif-ferent temperatures and alpha structure presented with star symbols. Spectra are shifted

vertically for clarity.

Furthermore, no phase transition of the matrix was detected after thermal annealingat 950 ◦C and 1000 ◦C. However, characteristic peaks related to the alpha phase ofSi3N4 appeared at a temperature of 1025 ◦C. The intensity of the XRD peaks increasesas the annealing temperature is elevated due to an increase of either the grain size orvolume fraction.

A.3.5 Annealing-time dependence

LPCVD Si3N4 layers implanted with a specific Ge concentration were annealed at1100 ◦C for 1 and 10 hours to characterize the effect of annealing time on phasetransformation and the crystallization process. XRD results are displayed in Fig-ure A.8. For both cases, implanted layers were crystallized. A longer annealing timeincreased the intensity of the XRD peaks and equivalent grain sizes and/or volumefraction, as a result the crystalline structure, more closely resembled alpha phase. Aslight phase transition from alpha to beta was also observed as the annealing timewas increased. This can be explained by the fact that crystallization was initiated bythe alpha phase and then altered to the beta phase which is the more stable phase[139].


2 0 3 0 4 0 5 0 6 0 7 0 8 00

1

22.89

068

26.51

037

28.28

049

30.94

703

35.27

023

38.88

991

40.36

502

43.56

487

46.93

492

48.77

313

51.67

795

52.44

955

55.88

768

57.83

936

59.97

259

62.40

084

64.94

257

65.71

416

69.50

405

71.34

226

72.88

545

22.89

068

26.51

037

28.28

049

30.94

703

35.27

023

38.88

991

40.36

502

43.56

487

46.93

492

48.77

313

51.67

795

52.44

955

55.88

768

57.83

936

59.97

259

62.40

084

64.94

257

65.71

416

69.50

405

71.34

226

72.88

545

20.49

3 22.92

26.49

328

.239 31

.032

34.56

635

.367

38.94

340

.272

41.88 43

.504

47.02

548

.387

48.90

950

.652

51.66

152

.747

56.10

357

.768

59.53

61.53

662

.496

64.90

965

.878

69.62

171

.404

72.82

874

.7775

.999

77.64

4

1 0 h r sNo

rmaliz

ed Int

ensity

(arb. u

nits)

2 θ ( d e g r e e s )

1 h r

Figure A.8: XRD pattern of LPCVD Si3N4 layers implanted with Ge and annealed at 1100◦C in N2 atmosphere for 1 and 10 hrs. Poly-crystallization of the matrix observed for bothcases. The dotted purple and blue lines refer to diffraction structures of alpha and beta Si3N4

respectively. Spectra are shifted vertically for clarity.

A.3.6 Ambient dependence

LPCVD Si3N4 layers implanted with Ge (concentration of 9 at.%) were annealed inN2, O2 and Ar atmospheres at 1100 ◦C for 1 hour to investigate the role of annealingatmosphere on the crystallization process. XRD results are presented in Figure A.9. Itcan be observed that, independent of annealing atmosphere, all samples crystallizedand transformed into the alpha phase. However, the intensity of the peaks varied fordifferent ambients. The intensity of peaks are slightly higher in the case of O2 and Aratmospheres. Annealing in an Ar atmosphere yields significantly larger grain sizesthan the other annealing ambients. Also, the O2 ambient resulted in slightly largergrains compared to N2, and smaller ones than Ar. Annealing in an inert atmosphere- Ar- leads to larger grains since no thermodynamic interaction occurs. However,annealing in O2 may enhance the interaction of Ge atoms with the annealing gas andlead to a defect-mediated diffusion. It can be seen that the intensity of peaks areslightly decreased at N2 atmosphere which could be due to the interaction of impu-rities with N2. The effect of the ambient on LPCVD Si3N4 layer implanted with Cowas studied by Tue et al. [135]. Samples were annealed at a different range of tem-peratures (200-1000 ◦C) in Ar and Ar-H2 ambient. Their study showed an interactionbetween Co and Si3N4 after high temperature annealing, which was enhanced by thepresence of H2. Also, in all examined atmospheres, Co formed weak Co-Si and Co-Nbonds [135].


2 5 3 0 3 5 4 0 4 5 5 00 . 0 0

0 . 0 1

0 . 0 2

0 . 0 3

0 . 0 4

0 . 0 5

0 . 0 6

0 . 0 7

0 . 0 8 a l p h a - S i 3 N 4

A r

O 2

Norm

alized

Intens

ity (ar

b. unit

s)

2 θ ( d e g r e e s )

N 2

Figure A.9: XRD pattern of LPCVD Si3N4 layers implanted with Ge implanted. Sampleswere annealed at 1100 ◦C in different ambient. The star symbols show the alpha structure.

Spectra are shifted vertically for clarity.

A.4 Summary

A variety of parameters such as ion species, annealing temperature, time and ambientwere examined to investigate the causes of crystallization. In all cases an amorphousto crystalline phase transition of the matrix occurred after annealing at 1100◦C. Hereis the summary of results:

• No crystallization for Ge concentrations of < 4 at.%.

• No crystallization for 4 at.% Ge samples at high implantation temperatures (200and 400 ◦C).

• Crystallization of the matrix for Ge concentrations of ≥ 6 at.%, independent ofimplantation temperature.

• Crystallization of the matrix at 1100 ◦C independent of chemical impurity (Cu,Co and Ni).

• Crystallization of the matrix independent of annealing time and atmosphere at1100 ◦C.

Implantation-induced-damages resulted in marked changes in the structure of amor-phous Si3N4. It can be suggested that crystallization accompanying a partial decom-position of amorphous materials, and the annealing mechanism, favour the recou-pling of remaining Si-N bonds in a crystalline form of Si3N4 after thermal annealingat 1100 ◦C.

§A.4 Summary 107

Also, after implantation, dissociation of Si-N bonds may change the stoichiome-try of the matrix. Si-dangling bonds can either rebond with N or make new bondswith metals or Ge, corresponding to the bonding energy listed in Table A.3. Silicide

Table A.3: Bond energy of metal-silicide and GeSi bonds plus N bonds [7, 8]System Si-Si Si-N Co-Si Cu-Si Ni-Si Ni-N Ge-Si Ge-GeKJ/mol 310 437 274 221 318 539 297 257

bonds are the second preference bond after Si-N bond. It can be concluded that thephase transformation of Si3N4 is mediated by silicide precipitates.

The effect of loss of N2 was not considered in this study. It is probable that, af-ter implantation and formation of silicide bonds, N atoms would be free to leave thesystem in the form of N2 gas (944.84±0.10 is the N-N dissociation bond energy). Thisis subject of future work.

It is clear that a combination of mechanisms resulted in crystallization of thematrix. Implantation induced-damages, silicide bond formation and change of thestochiometry of the matrix could be the main reasons for phase transition.

Appendix B

H implantation effect

This appendix contains the experimental procedure of ion implantation of H followedby Ge in LPCVD Si3N4 layers. This experiment was designed to investigate the effectof H implantation on the structural properties of embedded NPs. Also, to comparethe results with embedded NPs in PECVD layers.

109

110 H implantation effect

B.1 Introduction

In this section, the structural properties of NPs as a function of formation conditionsare characterized to better understand the relative influence of H implantation. Asmentioned in Subsection 1.6.3, the major difference of LPCVD and PECVD Si3N4 lay-ers is the H content. The aim of this section is to study the role of H content on thestructural properties of NPs formed in LPCVD Si3N4 layers by multiple implanta-tions of H and Ge in LPCVD layers. Also, to investigate if addition of H could causea change in the bonding preference of NPs and the matrix. EXAFS was used to probethe structural properties of NPs. The obtained results were then compared with thestructural properties of NPs embedded in PECVD Si3N4 layers. For that reason thefluences of implanted Ge were kept similar to those used for PECVD Si3N4 layers.

B.2 Experimental procedure

0 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 00

1 0 0 0 0

2 0 0 0 0

3 0 0 0 0

4 0 0 0 0

5 0 0 0 0

6 0 0 0 0

0 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 00

1 0 0 0 0

2 0 0 0 0

3 0 0 0 0

4 0 0 0 0

5 0 0 0 0

6 0 0 0 0 G e Ha )

D e p t h ( Å )

b ) G e

D e p t h ( Å )

(Atom

s/cm3 )/(A

toms/c

m2 )

Figure B.1: H+Ge depth profiles. Sample was implanted with H first and then implantedwith Ge and the other sample just implanted with Ge of the same concentration.

LPCVD layers were implanted with Ge or a co-implantation of H and Ge with Geto investigate the role of H content on the structural properties of obtained NPs. TheGe and H distribution profiles calculated by SRIM are shown in figure B.1. Table B.1summarizes the Ge and H implantation parameters. Samples were annealed at 700and 900 ◦C under N2 gas for 1 hr to promote NP precipitation and growth.

Table B.1: Summary of Ge and H implantation parameters.Energy (keV) Concentration (at.%) Implantation Temperature (◦C) Fluence(atoms/cm2)

H 80 and 100 10 400 2.46 and 2.5× 1017

Ge 1000 12 400 4× 1017

§B.3 Result and discussion 111

B.3 Result and discussion

Here, we investigate the role of H implantation on the bonding environment andstructural properties of embedded NPs. Figure B.2 shows the XANES spectra forLPCVD layers of Ge and H+Ge implantation samples, annealed at 900 ◦C at theGe K-edge. The spectra of samples implanted with H is almost identical to sam-

1 1 0 8 0 1 1 1 0 0 1 1 1 2 0 1 1 1 4 0 1 1 1 6 0 1 1 1 8 0 1 1 2 0 00 . 0

0 . 2

0 . 4

0 . 6

0 . 8

1 . 0

1 . 2

1 . 4

H + G e G e

Absor

ption

(arb. u

nits)

E n e r g y ( e V )

A f t e r a n n e a l i n g a t 9 0 0 ° C

Figure B.2: XANES spectra at the Ge K-edge for Ge and H+Ge implanted LPCVD layers.

ples without H. For both cases the bonding environments are very similar to thec-Si33Ge66 standard. No obvious correlation between H implantation and NP bond-ing environment can be seen. Quantification of the H bonding from XANES analysiswould necessitate an appropriate measurement of Si-H and Ge-H standards, lackingin our system. Hence, more investigation is needed for a better understanding of thesystem.

Figure B.3 (a) shows k3-weighted EXAFS spectra, and figure B.3 (b) correspond-ing non-phase-corrected FTs for samples implanted with Ge and H+Ge annealed at900 ◦C. Refined fitting parameters for the first NN shell are listed in Table B.2. A

Table B.2: Rifined EXAFS fitting parameters.System CNGe RGe (Å) CNSi RSi (Å) σ2 (Å2)Ge+H 1.5±0.2 2.457±0.005 1.7±0.1 2.402±0.006 0.0033±0.0004Ge 1.5±0.2 2.45±0.005 1.9±0.1 2.402±0.005 0.0036±0.0005

single scattering path for SiGe random structure was used similar to that for fittingthe SiGe NP EXAFS data. The values of amplitude and energy shift were obtainedfor crystalline Ge standard and kept fixed for examined samples. Details of the fit-ting procedure can be found in Section 3.2. It can be seen that the two spectra are

112 H implantation effect

0 2 4 6 8 1 0 1 2- 8- 6- 4- 202468

0 2 402468

1 01 21 4

G e + H w i n d o w G e

a ) 9 0 0 ° Ck3 * χ

( k )( Å

- 3 )

W a v e n u m b e r K ( Å - 1 )

F T [ k3 *χ

( k )(Å

- 3 )]

b ) 9 0 0 ° C

G e + H G e


Figure B.3: (a) k3-weighted EXAFS spectra and (b) corresponding non-phase-corrected FTspectra of Ge and H+Ge samples.

identical. Also, it can be concluded H implantation did not affect the NP structuralproperties or bonding preference. In summary, no clear changes can be observedfrom the results.

Appendix C

XANES spectra of crystallinestandards

XANES spectra of NPs were compared to XANES spectra of different crystalline andamorphous standards. As mentioned in the text, no indication of Ge-O or Ge-Nwas apparent. Figure C.1 shows XANES spectra of c-Ge, GeO2 and Ge66Si34 spectra.The XANES spectra of Ge3N4 is also presented for further comparison(adopted fromRef.[[140]]).

Figure C.1: XANES spectra of crystalline Ge, GeO2 and Ge66Si34 standards. XANES spectraof crystalline Ge3N4 standard adopted from Ref.[[140].

For a quantitative analysis the bond lengths value of the first nearest neighbourof Ge with Si, O and N - achieved from the FT spectra of examined standards- arealso summarized in Table C.1. For the presented study, no indication of Ge-O orGe-N bonds was apparent for Ge implanted samples in PECVD nitride matrices.However, Si-Ge bonds were analysed for LPCVD Si3N4 matrix. We expected forma-tion of Ge NPs in LPCVD matrix due to its higher purity and less contaminationand stoichiometric structure. Hence, in order to identify the cause of the formationof GeSi NP further systematic study is required. The Ge only composition of theNPs in PECVD nitride matrices is consistent with the bond energy values as given

113

114 XANES spectra of crystalline standards

Table C.1: Bond length of first nearest neighbour derived from EXAFS data for crys-talline standards.

R(Å) Ref.Ge-O 1.73 [141]Ge-N 1.87 [140]Ge-Si 2.01 [99]Ge-Ge 2.44 [45]

in Table C.2. While Si and Ge are completely miscible and readily alloy, clearly Sipreferentially bonds to O and/or N.

Table C.2: Bond energies for Si- and Ge- bonds[7]System Si-Si Si-N Si-O Ge-N Ge-O Ge-Ge Ge-SiKJ/mol 310 437 799 395 658 264 297

Appendix D

Quantitative analyse of Ramanmeasurements

Peak position of Ge-Ge Raman-active mode for Ge NPs in PECVD Si3N4 and SiO1.67N0.14

matrices were analysed quantitatively and the results are summarized in Table D.1.The Ge-Ge peak position was fitted using the Voigt profile in Multipeak Fit package1.The Ge-Ge signal from amorphous and crystalline Ge standards are associated at 297and 300 cm−1, respectively[143]. Raman measurements in addition to XANES datawere used to predict the dominant phase of Ge NPs. Noticeably, as the concentrationand annealing temperature increases the measured values of peak position for NPsbecome closer to crystalline component.

Table D.1: Peak frequency for the main Raman-active mode (Ge-Ge) for Ge embed-ded NPs in PECVD Si3N4 and SiO1.67N0.14 matrices. The frequencies are in unitsof cm−1. Samples are referred first by their atomic concentration (at.%) followed by

their annealing temperature (◦C)Matrix Sample Ge-GeSi3N4 6-700

6-9009-7009-90012-70012-900

266.5±1.2277.8±0.3285.3±0.1285.7±0.4290.8±0.4292±0.1

SiO1.67N0.14 9-7009-9009-110012-70012-90012-1100

280.2±0.4285.6±0.8286.7±0.5287.3±0.7294±0.5295±0.5

1More details can be found in Ref.[[142]]

115

116 Quantitative analyse of Raman measurements

Bibliography

1. Das, S., Aluguri, R., Manna, S., Singha, R., Dhar, A., Pavesi, L. & Ray, S. K. Opti-cal and electrical properties of undoped and doped Ge nanocrystals. NanoscaleRes. Lett. 7, 143 (2012). (cited on pages xix and 4)

2. Ray, S. K., Maikap, S., Banerjee, W. & Das, S. Nanocrystals for silicon-basedlight-emitting and memory devices. J. Phys. D: Appl. Phys. 46, 153001 (2013).(cited on pages xix, 2, 3, 4, and 5)

3. Bregolin, F., Behar, M., Sias, U. & Moreira, E. Structural and photoluminescenceproperties of Si nanoclusters obtained by ion implantation into Si3N4 films. J.Lumin. 131, 2377 (2011). (cited on pages xix, 10, and 11)

4. Mirabella, S., Cosentino, S., Gentile, A., Nicotra, G., Piluso, N., Mercaldo, L. V.,Simone, F., Spinella, C. & Terrasi, A. Matrix role in Ge nanoclusters embeddedin Si3N4 or SiO2. Appl. Phys. Lett. 101, 011911 (2012). (cited on pages xix, 12, 76,92, and 94)

5. Heinig, K., Muller, T., Schmidt, B., Strobel, M. & Moller, W. Interfaces underion irradiation: growth and taming of nanostructures. Appl. Phys. A-Mater. Sci.Process. 77, 17 (2003). (cited on pages xix and 19)

6. Luo, Q., Dragomir-Cernatescu, I., Snyder, R., Rees, W. & Hess, D. Comparisonof nitrided HfO2 films deposited in O2 and N2 O by direct liquid injection CVD.J. Electrochem. Soc. 153, F1 (2006). (cited on pages xix and 23)

7. Luo, Y.-R. Chemical Bond Energies, chap. 6 (CRC Press, 2007). (cited on pagesxxv, xxvi, 107, and 114)

8. Vattuone, L., Yeo, Y. & King, D. Adatom bond energies and lateral interactionenergies from calorimetry: NO, O2, and N2 adsorption on Ni (100). J. Chem.Phys. 104, 8096 (1996). (cited on pages xxv and 107)

9. Bragg, W. H. Faradays Diary. Rev. Mod. Phys. 3, 449 (1931). (cited on page 2)

10. Suyver, J. F. Synthesis, Spectroscopy and Simulation of Doped Nanocrystals. Ph.D.thesis, Universiteit Utrecht, Netherlands (2003). (cited on page 2)

11. Ray, S. & Das, K. Luminescence characteristics of Ge nanocrystals embedded inSiO2 matrix. Opt. Mater. 27, 948 (2005). (cited on page 3)

117

118 BIBLIOGRAPHY

12. Park, C. J., Cho, K. H., Yang, W.-C., Cho, H. Y., Choi, S.-H., Elliman, R. G.,Han, J. H. & Kim, C. Large capacitance-voltage hysteresis loops in SiO2 filmscontaining Ge nanocrystals produced by ion implantation and annealing. Appl.Phys. Lett. 88, 071916 (2006). (cited on page 3)

13. De Blauwe, J. Nanocrystal nonvolatile memory devices. IEEE Transactions OnNanotechnology 1, 72 (2002). (cited on page 3)

14. Tiwari, S., Rana, F., Chan, K., Hanafi, H., Chan, W. & Buchanan, D. Volatile andnon-volatile memories in silicon with nano-crystal storage. In International Elec-tron Devices Meeting, 1995 - IEDM Technical Digest, 521 (IEEE, Electron DevicesSoc, 1995). 1995 International Electron Devices Meeting, Washington, DC, Dec.10-13, 1995. (cited on page 3)

15. Hanafi, H., Tiwari, S. & Khan, I. Fast and long retention-time nano-crystalmemory. IEEE Transactions On Electron Devices 43, 1553 (1996). (cited on page3)

16. King, Y.-C., King, T.-J. & Hu, C. MOS memory using germanium nanocrystalsformed by thermal oxidation of Si1−xGex. In Electron Devices Meeting, 1998.IEDM’98. Technical Digest., International, 115–118 (IEEE, 1998). (cited on page 3)

17. Kim, D., Kim, T. & Banerjee, S. Memory characterization of SiGe quantum dotflash memories with HfO2 and SiO2 tunneling dielectrics. IEEE Transaction onElectron Devices 50, 1823 (2003). (cited on page 3)

18. Lee, J. J., Wang, X., Bai, W., Lu, N. & Kwong, D.-L. Theoretical and experimen-tal investigation of Si nanocrystal memory device with HfO2 high-k tunnelingdielectric. IEEE Transactions on Electron Devices 50, 2067 (2003). (cited on page3)

19. Lee, P. F., Lu, X. B., Dai, J. Y., Chan, H. L. W., Jelenkovic, E. & Tong, K. Y. Memoryeffect and retention property of Ge nanocrystal embedded Hf-aluminate high-kgate dielectric. Nanotechnology 17, 1202 (2006). (cited on page 3)

20. Zacharias, M. & Fauchet, P. M. Blue luminescence in films containing Ge andGeO2 nanocrystals: The role of defects. Appl. Phys. Lett. 71, 380 (1997). (citedon pages 3 and 4)

21. Min, K. S., Shcheglov, K. V., Yang, C. M., Atwater, H. A., Brongersma, M. L.& Polman, A. The role of quantum-confined excitons vs defects in the visibleluminescence of SiO2 films containing ge nanocrystals. Appl. Phys. Lett. 68, 2511(1996). (cited on pages 3 and 4)

22. Takeoka, S., Fujii, M., Hayashi, S., & Yamamoto, K. Size-dependent near-infrared photoluminescence from Ge nanocrystals embedded in SiO2 matrices.Phys. Rev. 58, 7921 (1998). (cited on page 3)

BIBLIOGRAPHY 119

23. Das, S., Das, K., Singha, R. K., Dhar, A. & Ray, S. K. Improved charge injectioncharacteristics of Ge nanocrystals embedded in hafnium oxide for floating gatedevices. Appl. Phys. Lett. 91, 3118 (2007). (cited on page 3)

24. Maeda, Y. Visible photoluminescence from nanocrystallite Ge embedded in aglassy SiO2 matrix: Evidence in support of the quantum-confinement mecha-nism. Phys. Rev. 51, 1658 (1995). (cited on page 3)

25. Daniel, M. C., Tsvetkova, I. B., Quinkert, Z. T., Murali, A., M. De, V. M. R., Kao,C. C. & Dragnea, B. Role of Surface Charge Density in Nanoparticle-TemplatedAssembly of Bromovirus Protein Cages. ACS Nano 4, 3853 (2010). (cited onpage 3)

26. Takagahara, T. & Takeda, K. Theory of the quantum confinement effect onexcitons in quantum dots of indirect-gap materials. Appl. Phys. Lett. 61, 2187(1992). (cited on page 3)

27. Das, K., NandaGoswami, M., Mahapatra, R., Kar, G. S., Dhar, A., Acharya,H. N., Maikap, S., Lee, J.-H. & Ray, S. K. Charge storage and photolumines-cence characteristics of silicon oxide embedded Ge nanocrystal trilayer struc-tures. Appl. Phys. Lett. 84, 4379 (2004). (cited on page 3)

28. Kanemitsu, Y., Uto, H., Masumoto, Y. & Maeda, Y. On the origin of visiblephotoluminescence in nanometerâARsize Ge. Appl. Phys. Lett. 61, 2187 (1992).(cited on page 3)

29. Sharp, I. D., Xu, Q., Yi, D. O., Yuan, C. W., Beeman, J. W., Yu, K. M., Ager, J. W.,Chrzan, D. C. & Haller, E. E. Structural properties of Ge nanocrystals embeddedin sapphire. J. App. Phys. 100, 114317 (2006). (cited on page 3)

30. Das, S., Singha, R. K., Gangopadhyay, S., Dhar, A. & Ray, S. K. Microstructuralcharacteristics and phonon structures in luminescence from surface oxidized Genanocrystals embedded in HfO2 matrix. J. Appl. Phys. 108, 053510 (2010). (citedon page 4)

31. Zhang, J.-Y., Ye, Y.-H., Tan, X.-L. & Bao, X.-M. Effect of density of Ge nanocrys-tals on violet-blue photoluminescence of Ge+-implanted SiO2 film. J. App. Phys.86, 6139 (1999). (cited on page 4)

32. Shen, J., Wu, X., Tan, C., Yuan, R. & Bao, X. Correlation of electroluminescencewith Ge nanocrystal sizes in Ge-SiO2 co-sputtered films. Phys. Lett. A 300, 307(2002). (cited on page 4)

33. Zhang, J.-Y., Ye, Y.-H., Tan, X.-L. & Bao, X.-M. Voltage-controlled electrolu-minescence from SiO2 films containing Ge nanocrystals and its mechanism. J.Appl.Phys. A 71, 299 (2000). (cited on page 4)

120 BIBLIOGRAPHY

34. Chang, S.-T. & Liao, S.-H. Light emission and photoluminescence from high-kdielectrics containing Ge nanocrystals. J. Vac. Sci. Technol. B 27, 535 (2009). (citedon page 4)

35. Ortiz, M., Sangrador, J., Rodríguez, A., Rodríguez, T., Kling, A., Franco, N.,Barradas, N. & Ballesteros, C. Growth by LPCVD, crystallization and charac-terization of sige nanoparticles for nanoelectronic devices. phys. Status Solidi (A)203, 1284 (2006). (cited on page 5)

36. Zhu, J. G., White, C., Budai, J. D., Withrow, S. P. & Chen, Y. Growth of Ge, Si,and SiGe nanocrystals in SiO2 matrices. J. App. Phys. 78, 4386 (1995). (cited onpages 5 and 6)

37. Mogaddam, N. A. P., Alagoz, A. S., Yerci, S., Turan, R., Foss, S. & Finstad, T. G.Phase separation in SiGe nanocrystals embedded in SiO2 matrix during hightemperature annealing. J. Appl. Phys. 104, 124309 (2008). (cited on page 6)

38. Sørensen, A., Johnson, E., Bourdelle, K., Johansen, A., Andersen, H. & Sarholt-Kristensen, L. Sizes, structures and phase transformations of nano-sized thal-lium inclusions in aluminium. Philos. Mag. A 75, 1533 (1997). (cited on page6)

39. Markwitz, A., Rebohle, L., Hofmeister, H. & Skorupa, W. Homogeneously sizedistributed ge nanoclusters embedded in SiO2 layers produced by ion beamsynthesis. Nuc. Instrum. Methods Phys. Res. Sec. 147, 361 (1999). (cited on page6)

40. Williamson, A. J., Bostedt, C., Buuren, T. V., Willey, T. M., Terminello, L. J.,Galli, G. & Pizzagalli, L. Probing the Electronic Density of States of GermaniumNanoparticles. Nano Lett. 4, 1041 (2004). (cited on page 6)

41. Cantelli, V., von Borany, J., MÃijcklich, A. & Schell, N. Investigation of theformation and phase transition of Ge and Co nanoparticles in a SiO2 matrix.Nuc. Instrum. Methods Phys. Res. Sec. 238, 268 (2005). (cited on page 6)

42. Jensen, J. S., Pedersen, T. L., Pereira, R., Chevallier, J., Hansen, J. L., Nielsen,B. B. & Larsen, A. N. Ge nanocrystals in magnetron sputtered SiO2. Appl. Phys.A 83, 41 (2006). (cited on page 6)

43. Wellner, A., Paillard, V., Bonafos, C., Coffin, H., Claverie, A., Schmidt, B. &Heinig, K. H. Stress measurements of germanium nanocrystals embedded insilicon oxide. J. App. Phys. 94, 5639 (2003). (cited on page 6)

44. Araujo, L. L., Giulian, R., Sprouster, D. J., Schnohr, C. S., Llewellyn, D. J., Kluth,P., Cookson, D. J., Foran, G. J. & Ridgway, M. C. Size-dependent characterizationof embedded Ge nanocrystals: Structural and thermal properties. Phys. Rev. B78, 094112 (2008). (cited on pages 6, 33, 69, 74, 79, 92, and 94)

BIBLIOGRAPHY 121

45. Araujo, L. L., Foran, G. J. & Ridgway, M. C. Multiple scattering effects on theEXAFS of Ge nanocrystals. J. Phys.: Condens. Matter 20, 165210 (2008). (cited onpages 6, 94, and 114)

46. Giacomazzi, L. & Umari, P. First-principles investigation of electronic, struc-tural, and vibrational properties of a-Si3N4. Phys. Rev. B 80, 144201 (2009). (citedon page 6)

47. Ma, Y., Yasuda, T. & Lucovsky, G. Ultrathin device quality oxide-nitride-oxideheterostructure formed by remote plasma enhanced chemical vapor deposition.Appl. Phys. Lett. 64, 2226 (1994). (cited on page 7)

48. Bachhofer, H., Reisinger, H., Bertagnolli, E. & von Philipsborn, H. Transient con-duction in multidielectric silicon-oxide-nitride-oxide semiconductor structures.J. App. Phys. 89, 2791 (2001). (cited on page 7)

49. Aozasa, H., Fujiwara, I., AkihiroNakamura & Komatsu, Y. Analysis of car-rier traps in Si3N4in oxide/nitride/oxide for metal/oxide/nitride/oxide/sili-con nonvolatile memory. Japanese Journal of Applied Physics 38, 1441 (1999). (citedon page 7)

50. URL http://www.siliconfareast.com/sio2si3n4.htm. (cited on page 7)

51. URL www.azom.com/article.aspx?ArticleID=1114. (cited on page 7)

52. McCann, M., Weber, K. & Blakers, A. SHORT COMMUNICATION: Surfacepassivation by rehydrogenation of silicon-nitride-coated silicon wafers. Progressin Photovoltaics: Research and Applications 13, 195 (2005). (cited on page 8)

53. Bazin, B. Samnos technology. Solid-State Electron. 15, 649 (1972). (cited on page8)

54. Kim, B. Y., Luan, H. F. & Kwong, D. L. Nitride/oxide gate dielectric. IEEEStanford University 97, 463 (1997). (cited on page 8)

55. URL http://iopscience.iop.org/0960-1317/6/1/001. (cited on page 8)

56. Silicon Nitride for MEMS Applications: LPCVD and PECVD Process Compari-son (2014). URL www.plasmatherm.com. (cited on page 8)

57. Kuo, Y. PECVD silicon-nitride as a gate dielectric for amorphous-silicon thin-film-transistor - process and device performance. Journal of the ElectrochemicalSociety 142, 186 (1995). (cited on page 8)

58. Lin, Y.-T., Chen, C., Shieh, J.-M., Lee, Y.-J., Pan, C.-L., Cheng, C.-W., Peng, J.-T.& Chao, C.-W. Trap-state density in continuous-wave laser-crystallized single-grainlike silicon transistors. Appl. Phys. Lett. 88, 233511 (2006). (cited on page8)

http://www.siliconfareast.com/sio2si3n4.htm

www.azom.com/article.aspx?ArticleID=1114

http://iopscience.iop.org/0960-1317/6/1/001

www.plasmatherm.com

122 BIBLIOGRAPHY

59. Winderbaum, S., Romijn, I., Sterk, D. & van Straaten, B. PECVD nitride asa Gate Dielectric for Amorphous Silicon Thin Film Transistor. 21st EuropeanPVSEC, Dresden 186 (2005). (cited on page 8)

60. Jin, H. & Weber, K. J. The effect of LPCVD silicon nitride deposition on theSi-SiO2 interface of oxidized silicon wafers. J. Electrochem. Soc. 154, H5 (2007).(cited on page 9)

61. Jin, H., Weber, K., Deenapanray, P. & Blakers, A. Hydrogen reintroduction byforming gas annealing to LPCVD silicon nitride coated structures. J. Electrochem.Soc. 153, G750 (2006). (cited on page 9)

62. Bååk, T. Silicon oxynitride; a material for GRIN optics. Appl. Opt. 21, 1069(1982). (cited on page 9)

63. amrani, A. E., Menous, I., Mahiou, L., Tadjine, R., Touati, A. & Lefgoum, A.Silicon nitride film for solar cells. Renewable Energy 33, 2289 (2008). (cited onpage 9)

64. Huang, R., Lin, Z., Guo, Y., Song, C., X.Wang, Lin, H., Xu, L., Song, J. & Li, H.Bright red, orange-yellow and white switching photoluminescence from siliconoxynitride films with fast decay dynamics. Opt. Mater. Express 4, 205 (2014).(cited on page 9)

65. Sharma, S. K., Barthwal, S., Singh, V., Kumar, A., Dwivedi, P. K., Prasad, B. &Kumar, D. PECVD based silicon oxynitride thin films for nano photonic on chipinterconnects applications. Micron 44, 339 (2013). (cited on page 10)

66. Cen, Z. H., Chen, T. P., Ding, L., Ye, J. D., Liu, Y., Yang, M., Wong, J. I., Liu, Z. &Fung, S. Optical Transmission and Photoluminescence of Silicon Nitride ThinFilms Implanted with Si Ions. Eletrochem. Solid-State Lett. 12, H38 (2009). (citedon page 10)

67. Seol, K., Futami, T., Watanabe, T., Ohki, Y. & Takiyama, M. Effects of ion implan-tation and thermal annealing on the photoluminescence in amorphous siliconnitride. J. App. Phys. 85, 6746 (1999). (cited on page 10)

68. Keita, A.-S., Naciri, A. E., Delachat, F., Carrada, M., Ferblantier, G. & Slaoui,A. Dielectric functions of Si nanoparticles within a silicon nitride matrix. Phys.Status Solidi C 7, 418 (2010). (cited on page 11)

69. Wang, Y. Q., Wang, Y. G., Cao, L. & Cao, Z. X. High-efficiency visible photolu-minescence from amorphous silicon nanoparticles embedded in silicon nitride.Appl. Phys. Lett. 83, 3474 (2003). (cited on page 11)

70. Lee, S., Huang, S., Conibeer, G. & Green, M. Structural and optical study of Genanocrystals embedded in Si3N4 matrix. Energy Procedia 10, 20 (2011). (cited onpage 11)

BIBLIOGRAPHY 123

71. Sahu, B. S., Gloux, F., Slaoui, A., Carrada, M., Muller, D., Groenen, J., Bonafos, C.& Lhostis, S. Effect of ion implantation energy for the synthesis of Ge nanocrys-tals in SiN films with HfO2/SiO2 stack tunnel dielectrics for memory applica-tion. Nanoscale Res. Lett. 6 (2011). (cited on page 12)

72. Ehrhardt, F., Ferblantier, G., Muller, D., Ulhaq-Bouillet, C., Rinnert, H. & Slaoui,A. Control of silicon nanoparticle size embedded in silicon oxynitride dielectricmatrix. J. App. Phys. 114, 033528 (2013). (cited on page 12)

73. Perret-Tran-Van, S., Makasheva, K., Despax, B., Bonafos, C., Coulon, P. E. &Paillard, V. Controlled fabrication of Si nanocrystals embedded in thin SiONlayers by PPECVD followed by oxidizing annealing. Nanotechnology 21 (2010).(cited on page 12)

74. Meldrum, A., Haglund, R., Boatner, L. & White, C. Nanocomposite materialsformed by ion implantation. Adv. Mater. 13, 1431 (2001). (cited on pages 16and 19)

75. Ziegler, J. F., Littmark, U. & Biersack, J. P. The stopping and range of ions in solids /J.F. Ziegler, J.P. Biersack, U. Littmark (Pergamon New York, 1985). (cited on pages17, 37, 38, 59, and 74)

76. Bording, J. & Tafto, J. Molecular-dynamics simulation of growth of nanocrystalsin an amorphous matrix. Physical Review B 62, 8098 (2000). (cited on page 20)

77. URL http://www.genplot.com/doc/rump.htm. (cited on pages 21, 36, 58, and 72)

78. Mayer, J., Nicolet, M. & Chu, W. Backscattering Spectrometry. J. Vac. Sci. Tech.12, 356 (1975). (cited on pages 21 and 59)

79. Wellner, A., Paillard, V., Bonafos, C., Coffin, H., Claverie, A., Schmidt, B. &Heinig, K. H. Stress measurements of germanium nanocrystals embedded insilicon oxide. J. App. Phys. 94, 5639 (2003). (cited on page 23)

80. Kolobov, A. V. Raman scattering from Ge nanostructures grown on Si substrates:Power and limitations. J. App. Phys. 87, 2926 (2000). (cited on page 23)

81. Sharp, I. D., Xu, Q., Liao, C. Y., Yi, D. O., Beeman, J. W., Liliental-Weber, Z.,Yu, K. M., Zakharov, D. N., Ager, J. W., Chrzan, D. C. & Haller, E. E. Stable,freestanding Ge nanocrystals. J. App. Phys. 97, 124316 (2005). (cited on page 23)

82. Bianconi, A. Surface x-ray absorption spectroscopy: Surface EXAFS and surfaceXANES. Appl.Surf. Sci. 6, 392 (1980). (cited on page 25)

83. X-Ray Absorption Spectroscopy of Semiconductors, vol. 190 of 0342-4111 (Springer-Verlag Berlin Heidelberg, 2015). (cited on page 29)

http://www.genplot.com/doc/rump.htm

124 BIBLIOGRAPHY

84. Glover, C., McKinlay, J., Clift, M., Barg, B., Broadbent, A., Boldeman, J., Ridgway,M., Foran, G., Garrett, R. & Lay, P. The x-ray absorption spectroscopy beamlineat the Australian synchrotron. In Synchrotron Radiation Instrumentation Conference(2006). (cited on page 30)

85. URL http://www.synchrotron.org.au/aussyncbeamlines/

x-ray-absorption-spectroscopy/technical-information. (cited on page 31)

86. Decoster, S., Glover, C. J., Johannessen, B., Giulian, R., Sprouster, D. J., Kluth,P., Araujo, L. L., Hussain, Z. S., Schnohr, C., Salama, H., Kremer, F., Temst,K., Vantomme, A. & Ridgway, M. C. Lift-off protocols for thin films for use inEXAFS experiments. J. Synchrotron Radiat. 20, 426 (2013). (cited on page 31)

87. Ridgway, M. C., Azevedo, G. d. M., Elliman, R. G., Glover, C. J., Llewellyn,D. J., Miller, R., Wesch, W., Foran, G. J., Hansen, J. & Nylandsted-Larsen, A.Ion-irradiation-induced preferential amorphization of Ge nanocrystals in silica.Phys. Rev. B 71, 094107 (2005). (cited on page 31)

88. Bunker, G. Introduction to XAFS: a practical guide to X-ray absorption fine structurespectroscopy (Cambridge University Press, 2010). (cited on page 31)

89. Newville, M. Ifeffit: interactive XAFS analysis and FEFF fitting. Journal of syn-chrotron radiation 8, 322 (2001). (cited on page 33)

90. Newville, M., Lıvinš, P., Yacoby, Y., Rehr, J. J. & Stern, E. A. Near-edge x-ray-absorption fine structure of pb: A comparison of theory and experiment. Phys.Rev. B 47, 14126 (1993). (cited on page 33)

91. Newville, M., Ravel, B., Haskel, D., Rehr, J., Stern, E. & Yacoby, Y. Analysis ofmultiple-scattering XAFS data using theoretical standards. Physica B: CondensedMatter 208, 154 (1995). (cited on page 33)

92. URL http://www.fe�project.org/fe�project-fe�-download.html. (cited on page33)

93. Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysisfor X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat. 12, 537(2005). (cited on page 33)

94. URL http://www.synchrotron.org.au/aussyncbeamlines/

x-ray-absorption-spectroscopy/data-analysis. (cited on page 33)

95. Kelly, S. D. & Ravel, B. EXAFS energy shift and structural parameters. AIPConference Proceedings 882, 132 (2007). (cited on page 33)

96. Araujo, L. L., Foran, G. J. & Ridgway, M. C. Multiple scattering effects on theEXAFS of Ge nanocrystals. J. Phys.: Condens. Matter 20, 165210 (2008). (cited onpages 33 and 94)

http://www.synchrotron.org.au/aussyncbeamlines/x-ray-absorption-spectroscopy/technical-information

http://www.synchrotron.org.au/aussyncbeamlines/x-ray-absorption-spectroscopy/technical-information

http://www.feffproject.org/feffproject-feff-download.html

http://www.synchrotron.org.au/aussyncbeamlines/x-ray-absorption-spectroscopy/data-analysis

http://www.synchrotron.org.au/aussyncbeamlines/x-ray-absorption-spectroscopy/data-analysis

BIBLIOGRAPHY 125

97. Araujo, L., Kluth, P., de M. Azevedo, G. & Ridgway, M. Short-range thermaland structural properties of Ge nanocrystals. Nuc. Instrum. .Methods .Phys. Res.Sec. B 257, 56 (2007). (cited on pages 33 and 94)

98. Araujo, L., L., Kluth, P., de M. Azevedo, G. & Ridgway, M. C. Vibrationalproperties of Ge nanocrystals determined by EXAFS. Phys. Rev. B 74, 184102(2006). (cited on pages 33 and 94)

99. Ridgway, M., Yu, K., Glover, C., Foran, G., Clerc, C., Hansen, J. & Larsen, A.Composition-dependent bond lengths in crystalline and amorphized GexSi1−xalloys. Physical Review B 60, 10831 (1999). (cited on pages 37 and 114)

100. Deal, B. & Helms, C. The Physics and Chemistry of SiO2 and the Si-SiO2 Interface 2(Springer US, 1993). (cited on pages 46 and 76)

101. Schmidt, H., Geckle, U. & Bruns, M. Simultaneous diffusion of Si and N insilicon nitride. Phys. Rev. B 74, 045203 (2006). (cited on pages 46 and 92)

102. Chang, J. E., Liao, P. H., Chien, C. Y., Hsu, J. C., Hung, M. T., Chang, H. T.,Lee, S. W., Chen, W. Y., Hsu, T. M., George, T. & Li, P. W. Matrix and quantumconfinement effects on optical and thermal properties of Ge quantum dots. J.Phys. D: Appl. Phys. 45, 105303 (2012). (cited on page 47)

103. Cardona, M. & Thewalt, M. Isotope effects on the optical spectra of semicon-ductors. Rev. Mod. Phys. 77, 1173 (2005). (cited on pages 65, 76, and 80)

104. Gonzalezhernandez, J., Azarbayejani, G., Tsu, R. & Pollak, F. Raman, transmis-sion electron-microscopy, and conductivity measurements in molecular-beamdeposited microcrystalline Si and Ge - a comparative-study. Appl. Phys. Lett.47, 1350 (1985). (cited on page 65)

105. Richter, H., Wang, Z. & Ley, L. The one phonon Raman-spectrum in microcrys-talline silicon. Solid. State. Commun. 39, 625 (1981). (cited on page 65)

106. Fujii, M., Hayashi, S. & Yamamoto, K. Raman-scattering from quantum dotsof Ge embedded in SiO2 thin-films. Appl. Phys. Lett. 57, 2692 (1990). (cited onpages 65, 76, and 80)

107. Kluth, P., Johannessen, B., Giraud, V., Cheung, A., Glover, C. J., Azevedo, G.d. M., Foran, G. J. & Ridgway, M. C. Bond length contraction in Au nanocrystalsformed by ion implantation into thin SiO2. Appl. Phys. Lett. 85, 3561 (2004).(cited on page 69)

108. Qi, W., Wang, M. & Su, Y. Size effect on the lattice parameters of nanoparticles.J. Mater. Sci. Lett. 21, 877–878 (2002). (cited on page 69)

109. Frenkel, A., Hills, C. & Nuzzo, R. A view from the inside: Complexity in theatomic scale ordering of supported metal nanoparticles. J. Phys. Chem. B 105,12689 (2001). (cited on page 69)

126 BIBLIOGRAPHY

110. Soulairol, R. & Cleri, F. Interface structure of silicon nanocrystals embedded inan amorphous silica matrix. Solid State Sci. 12, 163 (2010). (cited on page 69)

111. Karouta, F., Vora, K., Tian, J. & Jagadish, C. Structural, compositional and opticalproperties of pecvd silicon nitride layers. J. Phys. D: Appl. Phys. 45, 445301 (2012).(cited on page 72)

112. Mirzaei, S., Kremer, F., Sprouster, D., Araujo, L., Feng, R., Glover, C. & Ridgway,M. C. Formation of Ge nanoparticles in SiOxNy by ion implantation and thermalannealing. Journal of Applied Physics 118, 154309 (2015). (cited on page 73)

113. Koops, G., Pattyn, H., Vantomme, A., Nauwelaerts, S. & Venegas, R. Extremelowering of the Debye temperature of Sn nanoclusters embedded in thermallygrown SiO2 by low-lying vibrational surface modes. Physical Review B 70 (2004).(cited on page 74)

114. Sato, S., Nozaki, S. & Morisaki, H. Density of states of the tetragonal-phasegermanium nanocrystals using x-ray photoelectron spectroscopy. Appl. Phys.Lett. 72, 2460 (1998). (cited on page 79)

115. Glover, C., Ridgway, M., Llewellyn, D., Kluth, P. & Johanessen, B. Formation andelectronic structure of germanium nanocrystals formed by ion beam synthesis.Nucl. Instrum. Methods Phys. Res. Sect. B-Beam Interact. Mater. Atoms 238, 306(2005). (cited on page 79)

116. Backman, M., Djurabekova, F., Pakarinen, O. H., Nordlund, K., Araujo, L. L. &Ridgway, M. C. Amorphization of Ge and Si nanocrystals embedded in amor-phous SiO2 by ion irradiation. Phys. Rev. B 80, 144109 (2009). (cited on page79)

117. Cheung, A., de M. Azevedo, G., Glover, C. J., Llewellyn, D. J., Elliman, R. G.,Foran, G. J. & Ridgway, M. C. Structural perturbations within Ge nanocrystalsin silica. Appl. Phys. Lett. 84, 278 (2004). (cited on pages 92 and 94)

118. Chew, H. G., Choi, W. K., Foo, Y. L., Zheng, F., Chim, W. K., Voon, Z. J., Seow,K. C., Fitzgerald, E. A. & Lai, D. M. Y. Effect of germanium concentrationand oxide diffusion barrier on the formation and distribution of germaniumnanocrystals in silicon oxide matrix. Nanotechnology 17, 1964 (2006). (cited onpage 92)

119. Pavesi, L. & Turan, R. Silicon nanocrystals: fundamentals, synthesis and applications(John Wiley & Sons, 2010). (cited on page 93)

120. Riley, F. Silicon nitride and related materials. J. Am. Ceram. Soc. 83, 245 (2000).(cited on page 98)

BIBLIOGRAPHY 127

121. Kim, H.-D., Lin, H.-T. & Hoffmann, M. J. Advanced Si-based Ceramics and Com-posites: Proceedings of the International Symposium on New Frontier of Advanced Si-Based Ceramics and Composites (ISASC 2004) Held in Gyeongju, Korea, June 20-23,2004, vol. 287 (Trans Tech Publications, 2005). (cited on page 98)

122. Okada, A. Automotive and industrial applications of structural ceramics inJapan. J. Eur. Ceram. Soc. 28, 1097 (2008). (cited on page 98)

123. Katz, R. Applications of silicon nitride based ceramics. Ind. Ceram. 17, 158(1997). (cited on page 98)

124. Allaire, F. & Langlois, R. Factors influencing the crystallization of ultrafineplasma-synthesized silicon nitride as a single powder and in composite SiC-Si3N4 powder. J. Mater. Sci. 27, 1265 (1992). (cited on page 98)

125. Szépvölgyi, J. & Mohai, I. Crystallization of an amorphous silicon nitride pow-der produced in a radiofrequency thermal plasma. Ceram. Int. 25, 711 (1999).(cited on page 98)

126. Hojo, J., Maeda, H. & Kato, A. Crystallization behavior of amorphous-siliconnitride powder. J. Ceram. Soc. Jpn. 96, 842 (1988). (cited on page 98)

127. Seher, M., Bill, J., Aldinger, F. & Riedel, R. Crystallization kinetics ofpolysilazane-derived amorphous-silicon nitride. J. Cryst. Growth 137, 452 (1994).(cited on page 98)

128. Li, Y., Wang, L., Yin, S., Yang, F. & Wu, P. Rapid crystallization process ofamorphous silicon nitride. J. Am. Ceram. Soc. 94, 4169 (2011). (cited on page 98)

129. Sharma, N., Hooda, M. & Sharma, S. Synthesis and characterization of LPCVDpolysilicon and silicon nitride thin films for mems applications. J. Mater. 2014(2014). (cited on page 98)

130. Temple-Boyer, P., Rossi, C., Saint-Etienne, E. & Scheid, E. Residual stress inlow pressure chemical vapor deposition SiNx films deposited from silane andammonia. J. Vac. Sci. .Technol. A 16, 2003 (1998). (cited on page 98)

131. Rousset, B., Furgal, L., Fadel, P., Fulop, A., Pujos, D. & Temple-Boyer, P. De-velopment of SiNx LPCVD processes for microtechnological applications. LeJournal de Physique IV 11, Pr3 (2001). (cited on page 98)

132. URL http://rru�.geo.arizona.edu/AMS/result.php. (cited on page 99)

133. Jehanathan, N., Saunders, M., Liu, Y. & Dell, J. Crystallization of silicon nitridethin films synthesized by plasma-enhanced chemical vapour deposition. Scriptamaterialia 57, 739 (2007). (cited on page 102)

http://rruff.geo.arizona.edu/AMS/result.php

128 BIBLIOGRAPHY

134. Jehanathan, N., Saunders, M., Liu, Y. & Dell, J. M. Crystallization and composi-tional changes in amorphous PECVD SiNx thin films. In International Conferenceon Smart Materials and Nanotechnology in Engineering, 64230C (International So-ciety for Optics and Photonics, 2007). (cited on page 102)

135. Nguyen, T., Ho, H., Kotecki, D. & Nguyen, T. Reaction study of cobalt andsilicon-nitride. J. Mater. Res. 8, 2354 (1993). (cited on pages 102 and 105)

136. Levinshtein, M. E., Rumyantsev, S. L. & Shur, M. S. Properties of Advanced Semi-conductor Materials: GaN, AIN, InN, BN, SiC, SiGe (John Wiley & Sons, 2001).(cited on page 103)

137. URL http://www.matweb.com/search/datasheet.aspx?matguid=

f8a93e43a88b4eb3acea7d7d25a9b34c&ckck=1. (cited on page 103)

138. Chen, L. J. Silicide Technology for Integrated Circuits (Processing), vol. 5 (Iet, 2004).(cited on page 103)

139. Xu, Y. & Ching, W. Electronic-structure and optical-properties of α and β phasesof silicon-nitride, silicon oxynitride, and with comparison to silicon dioxide.Phys. Rev. B 51, 17379 (1995). (cited on page 104)

140. Bull, C. L., McMillan, P. F., Itié, J.-P. & Polian, A. X-ray absorption and EX-AFS studies of Ge coordination and bonding in high-pressure nitrides: β-Ge3N4

(phenacite) and γ-Ge3N4 (spinel). Physica status solidi (a) 201, 909 (2004). (citedon pages 113 and 114)

141. Micoulaut, M., Cormier, L. & Henderson, G. The structure of amorphous, crys-talline and liquid GeO2. Journal of Physics: Condensed Matter 18, R753 (2006).(cited on page 114)

142. Bradley, M. Curve fitting in raman and ir spectroscopy: basic theory of lineshapes and applications. Thermo Fisher Scientific, Madison, USA, Application Note50733 (2007). (cited on page 115)

143. Schlecht, S., Yosef, M. & Fröba, M. Synthesis and raman spectroscopy ofnanoparticles of crystalline and x-ray amorphous germanium within meso-porous SiO2. Zeitschrift für anorganische und allgemeine Chemie 630, 864 (2004).(cited on page 115)

http://www.matweb.com/search/datasheet.aspx?matguid=f8a93e43a88b4eb3acea7d7d25a9b34c&ckck=1

http://www.matweb.com/search/datasheet.aspx?matguid=f8a93e43a88b4eb3acea7d7d25a9b34c&ckck=1

study the structural properties of ge nanoparticles …...study the structural properties of ge...

Documents