Citation
193nm Excimer Laser Processing Wide Band-Gap Semiconductor Materials

Material Information

Title:
193nm Excimer Laser Processing Wide Band-Gap Semiconductor Materials
Creator:
Wang, Xiaotie
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (109 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Materials Science and Engineering
Committee Chair:
GILA,BRENT P
Committee Co-Chair:
APPLETON,BILLY RAY
Committee Members:
PEARTON,STEPHEN J
NORTON,DAVID P
REN,FAN
Graduation Date:
12/18/2015

Subjects

Subjects / Keywords:
Annealing ( jstor )
Atoms ( jstor )
Carbon ( jstor )
Fluence ( jstor )
Graphene ( jstor )
Ions ( jstor )
Laser annealing ( jstor )
Lasers ( jstor )
Pulsed lasers ( jstor )
Raman scattering ( jstor )
Materials Science and Engineering -- Dissertations, Academic -- UF
annealing -- gan -- graphene -- laser -- lift-off -- sic
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Materials Science and Engineering thesis, Ph.D.

Notes

Abstract:
A 193 nm wavelength ArF excimer laser was used to develop Laser Lift-Off (LLO) technology on AlGaN/GaN high electron mobility transistors and GaN-based LEDs. Currently, the chip yield and device reliability made using dicing or 254 nm laser lift-off technology for chip separation suffered due to cracks formed along the edge of the chips during the dicing or laser lift-off. To prevent the crack formation, 30 um wide grooves were drilled from the front side of the wafer along the device edges prior to the laser lift-off. To achieve large area laser lift-off, we studied using multiple laser pulses at low fluence instead of one pulse at high fluence, on which both simulation and experiment were conducted. Large area laser lift-off GaN based thin films with the multiple-pulse method has been attempted. On the other hand, we used the same laser system to Pulse Laser Annealing (PLA) of ion implanted SiC to synthesize graphene. Different ion implanted SiC substrates were annealed, and we found evidence of graphitization on ion implanted SiC surfaces at lower threshold fluences than pristine SiC substrate. Raman spectroscopy confirmed the existence of graphene G, 2D and D peaks. Amorphous and ion implanted SiC samples have been studied, both amorphous and catalytic effect of different ion species were revealed. Ge-implanted sample graphitization has been studied. Possible mechanism of graphitization was proposed. Pulse laser annealing pristine SiC at various conditions demonstrated the potential of optimizing graphene growth quality by changing the annealing condition, including ambient content, pressure, laser fluence, laser pulses and laser frequency and etc. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2015.
Local:
Adviser: GILA,BRENT P.
Local:
Co-adviser: APPLETON,BILLY RAY.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2016-12-31
Statement of Responsibility:
by Xiaotie Wang.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
12/31/2016
Classification:
LD1780 2015 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

193nm EXCIMER LASER PROCESSING WIDE BAND GAP SEMICONDUCTOR MATERIALS By XIAOTIE WANG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2015

PAGE 2

© 2015 Xiaotie Wang

PAGE 3

To my family, with all my love

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my sincere gratitude to the College of Engineering for letting me be a student and a gator here. It would be impossible to complete a work such as this without teamwork and collaborations. First, I would like to th ank Dr. Brent Gila for being my advisor and for providing support and guidance in every aspect of my research. This work could not have been accomplished without his great mentorship. I would also like to thank Dr. Fan Ren for his support during my master study and, later, in the research. My sincere appreciation is extended to Dr. Bill Appleton for his tremendous help and guidance, and to Dr. Steve Pearton and Dr. David Norton for their great advice and suggestions. I would like to thank my collaborator Di nesh K. Venkatachalam and Robert G. Elliman at the Australian National University (ANU) for providing broad area implanted sample s for pulse laser annealing. I would also like to thank former student Dr. Chien Fong Lo. We have had valuable conversations on every area of life and research. I appreciate the suggestions that he made for my research, some of which worked really ith him during the last few years. I would like to thank my colleague, coauthor, and friend, Tsung during my graduate studies. I would also like to thank Dr. Kara Berke, my coauthor and friend. Her detailed and organized work always keeps things running smoothly. I want to thank Camilo Velez for the opportunity to work with him on the laser fabrication of micro fluid devices, for the information th at he provided during my research, and for being a

PAGE 5

5 interpret Raman Spectra when I was a rookie. I have to thank the amazing engineering staff members in the Nano Research Facil ity (NRF). Al Ogden trained and helped me a lot when I first became a NRF user, and later in my research and teaching experience. David H ays gave me important training on the use of sputtering and the optical profilometer. Bill Lewis shared his knowledg e on photolithography. Andres Trucco and Thomas Sanders helped me set up the vacuum chamber for the laser annealing. Paula Mathis is the kindest person in the NRF, and I am so glad to work with her. To my family, I am truly and deeply indebted. Since my c hildhood, they have been constantly supportive. Their hard workings, dedication to the family and unconditional love have helped me to grow from a boy to a man, and from a man to a father. I truly wish they remain healthy and happy. Finally, thanks to my w ife Yan. She is the best and most special woman that I have ever met. She is the muse of my life, the mother of our beloved son, and a friend of my soul.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURE S ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 1.1 Motivation ................................ ................................ ................................ ......... 14 1.2 Dissertation Outline ................................ ................................ ........................... 15 2 BACKGROUND AND LITERATURE REVIEW ................................ ....................... 16 2.1 Laser Processing Materials ................................ ................................ ............... 16 2.1.1 Laser Types ................................ ................................ ............................. 16 2.1.2 Inte raction between Lasers and Materials ................................ ............... 16 2.1.3 Thermal Process ................................ ................................ ..................... 18 2.1.4 Photochemical Process ................................ ................................ ........... 18 2.1.5 Photophysical Process ................................ ................................ ............ 18 2.1.6 Differences between Conventional Heating Source and Lasers .............. 18 2.2 Laser Lift Off GaN Thin Film ................................ ................................ ............. 19 2.2.1 GaN on Substrates ................................ ................................ .................. 19 2.2.2 Laser Lift Off GaN (LLO) Thin Film ................................ .......................... 20 2.2.3 Challenges and Issues ................................ ................................ ............ 21 2.3 Graphene, Graphene Synthesis, and Characterization ................................ ..... 21 2.3.1 Graphene and Its Properties ................................ ................................ ... 21 2.3.2 Graphene Synthesis Techniques ................................ ............................ 22 2.3.2.1 Exfoliation ................................ ................................ ...................... 22 2.3.2.2 Melton Metal ................................ ................................ .................. 22 2.3.2.3 CVD Method ................................ ................................ .................. 23 2.3.2.4 Thermal Annealing Silicon Carbide (SiC) ................................ ....... 23 2.3.2.5 Others ................................ ................................ ............................ 24 3 193 NM EXCIMER LASER LIFT OFF FOR ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTORS ................................ ................................ ..................... 26 3.1 Single Pulse Laser Lift Off ................................ ................................ ................ 26 3.1.1Abstract ................................ ................................ ................................ .... 26 3.1.2 Introduc tion ................................ ................................ .............................. 26 3.1.3 Experiments ................................ ................................ ............................ 27

PAGE 7

7 3.1.4 Results and Discussion ................................ ................................ ........... 29 3.1.5 Conclusion ................................ ................................ ............................... 33 3.2 Laser Lift Off GaN with Multiple Pulses. ................................ ........................... 34 4 GRAPHENE SYNTHESIS BY PULSED LASER AND THERMAL ANNEALING ION IMPLANTED SIC ................................ ................................ ............................. 49 4.1 Abstract ................................ ................................ ................................ ............. 49 4.2 Introduction ................................ ................................ ................................ ....... 49 4.3 Experime nts ................................ ................................ ................................ ...... 51 4.3.1 Ion Implantation. ................................ ................................ ...................... 51 4.3.2 Pulsed Laser Annealing. ................................ ................................ .......... 51 4.3.3 Characterization. ................................ ................................ ..................... 52 4.4 Results and Discussion. ................................ ................................ .................... 52 4.5 Conc lusion ................................ ................................ ................................ ........ 55 5 SELECTIVE GROWTH OF GRAPHENE AND CARBON NANO STRUCTURES ON AMORPHOUS SIC VIA PULSED LASER ANNEALING ................................ ... 60 5.1 Introduction ................................ ................................ ................................ ....... 60 5.2 Experiments and Thermal Simulations ................................ ............................. 61 5.3 Results and Discussion ................................ ................................ ..................... 62 5.3.1 Thermal Simulation ................................ ................................ .................. 62 5.3.2 Amorphous Effect ................................ ................................ .................... 63 5.3.3 Catalytic Effect of Au + Implanted Samples ................................ .............. 67 5.3.4 Ge Implanted SiC ................................ ................................ .................... 68 5.4 Conclusions ................................ ................................ ................................ ...... 71 6 GRAPHENE SYNTHESIS BY PULSED LASER ANNEALING CRYSTALLINE SIC AND OPTIMIZATION ................................ ................................ ....................... 81 6.1 Introduction ................................ ................................ ................................ ....... 81 6.2 Thermal Simulation and Experiments. ................................ .............................. 8 1 6.3 Results and Discussion. ................................ ................................ .................... 82 6.3.1 Thermal Simulation. ................................ ................................ ................. 82 6.3.2 Fluence. ................................ ................................ ................................ ... 83 6.3.3 Frequency Effect. ................................ ................................ .................... 84 6.3.4 Pressure Effect. ................................ ................................ ....................... 86 6.3.5 SiC Solid Transformation From 4H To 3C. ................................ .............. 87 6.4 Conclusion ................................ ................................ ................................ ........ 89 7 CONCLUSION ................................ ................................ ................................ ...... 100 LIST OF REFERENCES ................................ ................................ ............................. 102 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 109

PAGE 8

8 LIST OF TABLES Table page 2 1 Wavelength for selective lasers 1 ................................ ................................ ......... 25 3 1 on a Si wafer and the estimated strain relaxatio n of the LLO HEMT. ................. 38 4 1 Comparison of the a SiC surface layer thicknesses produced by II and the PLA fluences at which graphitization begins for each. ................................ ........ 56 5 1 Tabulation of the implantation parameters used to form stoichiometric a SiC surface layers of various thicknesses on c SiC. ................................ ................. 72 5 2 Parameters of amorphous and crystalline SiC used in thermal simulation calculations. ................................ ................................ ................................ ........ 72

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Schematic of different types of electronic excitations in insulator/semiconductor. ................................ ................................ ..................... 25 3 1 Photographs of HEMT. ................................ ................................ ....................... 39 3 2 Photograph after LLO . ................................ ................................ ........................ 40 3 3 Device performance before and after LLO (I) . ................................ .................... 41 3 4 Device performance before and after LLO (II) . ................................ ................... 42 3 5 . ... 43 3 6 Raman spectrum E2 peak from a HEMT sample before and after the LLO process. ................................ ................................ ................................ .............. 44 3 7 Schematic map for x and y direction scanning, and Image after laser lift off. ..... 45 3 8 Drawings s how crack formation ................................ ................................ .......... 45 3 9 TEM images of GaN on sapphire as grown and GaN thin film after laser lift off. ................................ ................................ ................................ ...................... 45 3 10 Mu ltiple pulse LLO experiments before and after LLO ................................ ....... 46 3 11 Cross sectional schematic drawing of multiple pulse laser lift off. ...................... 47 3 12 Illustration of multiple pulses LLO ................................ ................................ ....... 48 4 1 SRIM calculations of the damage depth profiles produced by ion implantation of Au, Cu, and Ge in SiC. ................................ ................................ ................... 56 4 2 Schematic diagram of the193 nm ArF excimer laser used in our work (See text for details). ................................ ................................ ................................ ... 57 4 3 Optical images of ion implanted samples of 4H SiC after PLA in air at different pulse number and fluence.. ................................ ................................ .. 57 4 4 Raman spectra of ion implanted SiC before and after PLA at different fluences, and relative G Peak intensities versus fluence for Au, Cu, Ge implanted SiC. ................................ ................................ ................................ .... 58 4 5 Growth of FLG with nanoscale features by Au MIBL and PLA. .......................... 59

PAGE 10

10 5 1 Thermal simulation results estimating laser fluences required to melt certain thicknesses of amorphous SiC on c SiC. Melting temperature of amorphous SiC, T m is used as 2445K. ................................ ................................ .................. 73 5 2 Raman spectrum for pristine SiC, PLA pristine SiC, sample #1, sample #2, and sample #3. ................................ ................................ ................................ ... 74 5 3 HR XTEM images of sample #3 after pulse laser annealing with 100 pulses at 800mJ/cm 2 .. ................................ ................................ ................................ .... 75 5 4 HR XTEM image of pulse laser annealing of amorphous SiC without Au + implant, with 2000 laser pulses at 800mJ/cm 2 , and HR XTEM image of pulse laser annealing of amorphous SiC implanted with Au + ions, with 2000 laser pulses at 800mJ/cm 2 . ................................ ................................ ......................... 76 5 5 Raman spectrum for Pristine SiC (blue), pulse laser annealing with 2000 pulses at 800mJ/cm 2 on amorphous SiC without Au + implant (green) and pulse laser annealing with 2000 pulses at 800mJ/cm 2 amorphous SiC with Au + im plant (red). ................................ ................................ ................................ 77 5 6 Raman spectra of Ge implanted SiC before PLA (black), Ge implanted SiC after PLA with 50 pulses at 100mJ/cm 2 ( red), Ge implanted SiC after PLA with 50 pulses at 200mJ/cm 2 (green). ................................ ................................ 78 5 7 3D image of optical profilometer, 50 pulses, 100mJ/c m 2 , and 3D image of optical profilometer, 50 pulses, 200mJ/cm 2 . ................................ ....................... 79 5 8 HR XTEM images of Ge implanted sample.. ................................ ...................... 80 6 1 Thermal simulation of peak temperature at the end of 25 ns laser pulse at various fluence. Solid black line region is from thermal simulation, and red dash line is from estimation due to lack of constant. ................................ .......... 91 6 2 Raman spectra comparison before and after pulsed laser annealing with 1 100 pulses at 1J / cm 2 . ................................ ................................ ......................... 92 6 3 TEM images comparison among pulsed laser annealing SiC with 1 100 pulses at 1J / cm 2 in Ar.. ................................ ................................ ....................... 93 6 4 Raman Spectrum for frequency effect. ................................ ............................... 94 6 5 TEM images of PLA 375 pulses at 0.8J/cm 2 , 50 Hz , and PLA 375 pulses at 0.8J/cm 2 , 1 Hz. ................................ ................................ ................................ ... 95 6 6 Raman spectra of PLA c SiC in Ar ambient and vac uum condition. ................... 96 6 7 Illustration of 3C and 4H SiC at different zone axis. ................................ ........... 97

PAGE 11

11 6 8 XTEM images of PLA samples at different zone axis. ................................ ........ 98 6 9 TEM images of PLA c SiC @ 0.8J/cm 2 , showing 3c SiC layer evolution with incremental pulse number. ................................ ................................ ................. 99

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy 193nm EXCIMER LASER PROCESSING WIDE BAND GAP SEMICONDUCTOR MATERIALS By Xiaotie Wang December 2015 Chair: Brent Gila Major: Material s Science and Engineering A 193 nm wavelength ArF excimer laser was used to develop laser lift off (LLO) technology on AlGaN/GaN high elect ron mobility transistors and GaN based LEDs. Currently, the chip yield and device reliability cause the use of dicing or 254 nm laser lift off technology for chip separation to be unreliable, due to cracks formed along the edge of the chips during dicing o r laser lift off. In order to prevent this, 30 µm wide grooves were drilled from the front side of the wafer along the device edges, prior to the laser lift off. To achieve a large area for the laser lift off, we studied the use of multiple laser pulses at low fluence instead of one pulse at high fluence; to test this, both a simulation and an experiment were conducted. Using the multiple pulse method, we attempted to create a large area for laser lift off with GaN based thin films. Meanwhile, we used the same laser system for pulse laser annealing (PLA) of ion implanted SiC in order to synthesize graphene in selective areas. Various ion implanted SiC substrates were annealed, and we found evidence of graphitization on ion implanted SiC surfaces at lower th reshold fluences than pristine SiC substrates. Raman spectroscopy confirmed the existence of G, 2D and D peaks in the graphene.

PAGE 13

13 We studied a morphous and ion implanted SiC samples, revealing both amorphous and catalytic effects of different ion species dur ing pulsed laser annealing. Few layers graphene and carbon nano structure were synthesized. We examined the Ge graphitization. Pulse laser annealing of pristine SiC under various cond itions demonstrated the potential for optimizing graphene growth quality by changing the annealing condition, including ambient content, pressure, laser fluence, laser pulses, laser frequency, and so on.

PAGE 14

14 CHAPTER 1 INTR ODUCTION 1.1 Motivation The laser is one of the greatest inventions of the 20 th century. The theoretical foundation for the laser was established by Albert Einstein in the paper On the Quantum Theory of Radiation in 1917; however, it was not until 1960 tha t Theodor Maiman built the first laser at Hughes Laboratories. From 1962 to 1968, most of the important types of lasers were developed, including CO2 gas lasers, Nd:YAG lasers, semiconductor lasers, dye lasers, and other gas lasers. Following this, increas ingly reliable and durable lasers were built. In the late 1970s, lasers began to be used in industrial applications for purposes such as marking, cutting, welding, and drilling. After the 1980s, scientists used lasers more for surface related applications such as annealing, glazing, cladding, and thin film deposition. Nowadays, laser wavelengths available in the market range from far infrared to visible spectrum, and from UV spectrum to, surprisingly, soft X be as low as 1 mW and as high as 3000 W, depending on its type and size. Lasers can also be divided into pulsed lasers and CW (continuous wave) lasers. All this flexibility, combined with high speeds and accurate industrial processing, gives lasers an important role in modern ind ustry. The global market of laser processing was worth $9.33 billion in 2014, and is expected to grow to $20.05 billion by the end of 2020, based on Laser Processing Market Analysis and Segment Forecasts to 2020 by Grand View Research. Semiconductor processing with lasers forms around 30% of the current laser processing market share, and this area is still growing. Its uses range from wafer

PAGE 15

15 fabrication to defect analysis. This growth is due to the fact that surface processing with local laser treatment has the benefit of affecting fewer bulk materials, and because the pulsed laser interacts effectively within the surface layer. 1.2 Dissertation Outline This dissertation covers the two topics. One is laser lift off GaN based divecs and G aN free standing thin film. The other is laser annealing SiC to synthesis graphene. Chapter 1 reviews the laser processing industry history. Chapter 2 reviews the physics of laser matter interaction, laser lift off GaN, and graphene history and current man ufacture methods. Chapter 3 covers the laser lift off high electron mobility transistor (HEMT), and device performance before after the process and possible reason. It also proposes potential new method to minimize the crack formation during laser lift off by using multiple low fluence pulses. Chapter 4 studies the graphene synthesis by pulsed laser annealing (PLA) ion implanted silicon carbide. Chapter 5 studies the mechani sm 6 further studies more contributing effects during laser annealing, including fluence, pulse number, repetition rate and annealing ambient, and 4H to 3 C SiC solid phase transition is also reported.

PAGE 16

16 CHAPTER 2 BACKGROUND AND LITERATURE REVIEW 2.1 Laser Processing M aterials 2.1.1 Laser T ypes Laser technology has reached a stage where it is highly developed, and wavelengths can range from far infrared to X Ray. Based on lasing media, lasers can be classified as one of the following types. Table 2.1 lists these types and the typical wavelength they emit. 1 1. Gas lasers (e.g., CO2, excimer) 2. Liquid lasers (organic liquid dye) 3. Solid state lasers (e.g., Nd YAG) 4. Fib er lasers 5. Semiconductor lasers (e.g., diode laser) 6. Free electron lasers (FEL) Depending on their operation mode, lasers can radiate in continuous wave (cw) or in pulsed mode. Continuous wave lasers have a constant power output over time, while a pulsed la ser is any laser not classified as a continuous laser. Duration and repetition rate are important parameters for pulsed lasers. The pulse duration can be in the range of picoseconds (10 12 s) or even femtoseconds (10 15 s), and the repetition rate can be u p to 1 kHz. 1, 2, 3 2.1.2 Interaction b etween Lasers and M aterials macroscale, when laser light shine s on the surface of materials , reflection, absorption, and transmission occur. For opaque materials, Reflectivity = 1 absorptivity. For transparent materials, Reflectivity = 1 (transmissivity + absorptivity)

PAGE 17

17 Laws of reflection obey the Fresnel equations. Transmission obeys the Beer Lambert law, where I = I 0 e . I 0 is the initial light intensity, I is the light intensity through materials, z is the material thickness, and is the absorption coefficient, which is In the microscale, the lase r photon passes its energy to atoms or molecules in materials by following the path of interaction between photon electron, electron phonon, and phonon lattice; this process is referred to as electronic excitation and relaxation. Inter band excitation occu rs primarily for semiconductors, and intra band excitation for metals. F ig. 2.1 shows the different types of electronic excitations in a solid. 4 Only the highest valence band (VB) and lowest conduction band (CB) are shown here. The straight lines denote th e absorption or emission of photons, while the oscillating lines denote a non radioactive process. When photon energy hv is larger than the band gap Eg , electron pairs are formed. The relaxation process can occur through either the release of a photon hv and the return of an electron to the valence band, or the release of a phonon. The center of the figure portrays defects and traps that can be excited with a photon energy that is less than Eg . The multi photon process is also shown, which consists of two coherent photon absorptions that are either simultaneous or sequential. This is essential for some applications where a higher energy photon is not available for the single photon process; this provides the flexibility of using a lower energy laser to proc ess materials in the experiment. In metals, photons are primarily absorbed by conduction band electrons.

PAGE 18

18 2.1.3 Thermal P rocess During the thermal process, the thermalization of the excitation energy is described as the relaxation time T . In a thermal process, T << R , where R is the controlled relaxation time during which processes take place such as desorption of the means that the time for photon energy to transfer to thermal energy is m uch less than the chemical reaction rate, which means that before the chemical reaction is activated by thermal energy, the laser energy could be simplified as a heating source. 2.1.4 Photochemical P rocess During the photochemical process, thermalization of the excitation energy is slow and in this case, T R . Therefore, the photon energy is primarily applied in the chemical reaction, which means that the temperature remains static during laser irradiation. 2.1.5 Photophysical P rocess When both thermal a nd non thermal processes contribute to the overall processing rate, we call this a photophysical process, which is usually closer to the reality of laser matter interactions. 2.1.6 Differences between C onventional H eating S ource and L asers Although in a t hermal process laser can be regarded as a heating source, its outcomes may be quite different compared to a conventional heating source such as a tube furnace. In an extremely short period of time, in the range of nanoseconds or femtoseconds, lasers can in duce a 10 4 K rise in temperature in a localized volume of materials. In the case of a high density pulsed laser, the temperature ramping up rate

PAGE 19

19 could be as high as 10 15 /s. Meanwhile, the chemical reaction rate may appear slow and may be far from achieving equilibrium before the laser pulse ends, and the surface temperature drop significantly and quickly after the laser pulse ends, which could freeze the surface. Therefore, some of the meta stable states or phases may be achieved by this non equilibrium pro cess and some applications may become feasible such as laser annealing, recrystallization, glazing, surface cleaning, surface polishing, shock hardening, and so on. 2.2 Laser Lift Off GaN Thin F ilm 2.2.1 GaN on S ubstrates In the last decade, GaN based dev ices have achieved tremendous success due to the high quality of their epitaxy growth. It is essential to choose proper substrates to grow GaN epi layers. Silicon, silicon carbide, and sap phire have been widely used, and the high crystalline quality of these substrates is generally accepted by the market. Silicon is the material that has dominated the semiconductor industry for more than half a century. The advantages of using silicon to g row GaN are that it is well studied, high quality single crystal silicon wafers are available, and there is a proliferation of existing techniques that can be directly applied to device fabrication and packaging. It has a decent thermal conductivity which is almost the same as crystalline GaN, with 1.5 W/cm k at 25 °C. 5 However, the lattice misfit of 17% and the thermal expansion coefficient mismatch between Si and GaN could introduce unwanted stress Silic on carbide has very good thermal stability and conductivity of 3.0 3.8 W/cm k (varies between poly types), which offers a great advantage over silicon and sapphire. GaN based devices such as HEMT or LEDs are intended to have a high power density

PAGE 20

20 which chal a better candidate for high power applications. In addition, a 3.5% lattice mismatch can greatly reduce stress during epi growth. However, the cost of silicon carbide is much h igher than that of sapphire and silicon, which limits its application. Sapphire (Al 2 O 3 ) was one of the earliest substrates used for GaN growth; it has a good thermal stability. However, its poor thermal conductivity of 0.4 W/cm k 5 limits its reliability a nd use in high power applications. Moreover, the wafer cost of sapphire is lower than that of silicon carbide, but still much higher than that of silicon. Therefore, much effort has been put into finding low cost and high quality substrates with good therm al conductivity. 2.2.2 Laser Lift Off GaN (LLO) Thin F ilm L aser lift off GaN (LLO) on sapphire was first demonstrated in 1998 by W.S. Wong et al. 6 thick GaN layer grown on double sided polished sapphire substrate by epo xy. A single pulse of 600 mJ/cm 2 , 38 ns KrF (248 nm) excimer laser was applied through the transparent sapphire substrate; a 50 °C thermal annealing then completed the lift off process by melting the Ga (melting point 29.7 °C) residue at the GaN/sapphire interface. Later, the InGaN multiple quantum well (MQW) LEDs were also laser lifted off by KrF excimer laser. 7 The free standing LEDs were compared with the results before LLO to show a 23% reduction in output power. Chu et al. 8 also reported that transfe rring laser lift off of GaN LED to copper substrate, which has a much better thermal conductivity than sapphire, improved thermal dissipation. A higher operation current and power output were also achieved.

PAGE 21

21 In summary, the laser lift off technique constit utes an efficient method of fabricating free standing GaN film or transferring the GaN epi layer from the sapphire substrate to other substrates. This technique has a great potential for combining a low cost substrate and a high quality epitaxy. Additional ly, access to the backside of the GaN epi layer also enables increased device design and processing choices. 2.2.3 Challenges and I ssues Although various lasers have successfully de bonded GaN thin film from sapphire substrates, several issues remain unre solved, of which thin film quality after laser lift off is the most important. Cracks 9, 10 seem inevitable due to laser shock during the GaN decomposition and de bonding process. The reasons for crack formation, and methods for preventing it, are still und er investigation. Another issue is inconsistent device performance, as in the two LED samples in the previous paragraph. A detailed and comprehensive study is needed in order to reveal the secret behind this. 2. 3 Graphene, Graphene Synthesis, and C haracter ization 2. 3 .1 Graphene and Its P roperties G raphene is one of the atomic layers of the hexagonal lattice of sp2 bonded carbon atoms. As early as 1859, researchers were aware of lamellar structures in graphite oxide. 11 Due to the limited microscopic techniques at that time, it was not until 1948 that G. Ruess and F. Vogt captured the first TEM image of layers of graphite. 12 Attempts were made to mechanically exfoliate the graphene layers in the 1990s, and Andre Geim and Kostya Novoselov of the University of Manchester were the first group to successfully exfoliate the single graphene layer in 2004 13 ; they were awarded the Noble Prize for this work in 2010.

PAGE 22

22 Graphene has drawn huge attention since 2004 13 . It is also called electron mobility of 2.5 x 10 5 cm 2 V 1 S 1 14 15 ; excellent thermal conductivity above 3,000 WmK 1 16 ; and an optical absorption coeffi cient of 2.3% 17 . It can also be chemically functionalized. 18 20 . D materials such as silicone, 21 germanene, 22 BN, 23 MoS 2, 24 26 and so on. The fundamental study and fabrication techniques of graphene h ave been widely applied to these materials. This range of new discoveries provides endless possibility and flexibility, not only for the semiconductor industry but for the entire scientific community. 2. 3 .2 Graphene Synthesis T echniques 2. 3 .2.1 Exfoliation A dhesive 13 to achieve a single layer of graphene. Exfoliation was conducted multiple times with the layers reducing in number each time, until a single layer remained. In terms of quality, exfoliated graphene has the least amount of defects and the highest electron mobility. 27 However, its flake shape, size, and thickness is not completely controllable, which limits its application and drives researchers to look for alternative synthesis methods. 2. 3 .2.2 Melton M etal Amini et al. 28 reported using melting transition metal to synthesize single layer graphene (SLG) and few layer graphene (FLG). Nickel was placed in a graphite crucible The graphite crucible was the carbon source; only a very small amount was able to dissolve into the melting nickel. After 16 hours at 1500 °C, the sample was cooled down

PAGE 23

23 at a rate of 10 °C/min, and eventually returned to room temperature. Due to the decreasing solubility of carbon atoms in nickel at lowe r temperatures, carbon atoms precipitated on the nickel surface and sp2 bonded with each other, forming graphene. The Raman spectrum demonstrated the existence of single layer and few layer graphene. Besides nickel, copper, ruthenium, iridium, and other tr ansition metals also demonstrate the ability to assist graphene formation on a surface. 29 31 2. 3 .2.3 CVD M ethod Xuesong Li et al. 32 reported the chemical vapor deposition (CVD) method for ixture of methane and hydrogen, carbon atoms dissolved into copper film at 1000 °C, and precipitated to grow graphene while cooling back to room temperature. The researchers also found that this CVD process is self limited. By running the growth for 10 min utes, they could achieve the same quality as when running for 60 minutes. This self limiting process is due to the limited solubility of carbon in copper. Besides the copper graphene growth, they also demonstrated the ability to grow graphene on other subs trates such as silicon oxide and glass. 2. 3 .2.4 Thermal Annealing Silicon C arbide (SiC) Silicon carbide (SiC) was heated up to 1000 °C in a high vacuum, and the silicon sublimated from the surface and left the carbon to bond and form graphene on a surfac e. 32 33 Different faces on silicon carbide were also studied: silicon face SiC tended to generate single layer graphene (SLG), and C face SiC tended to form multiple layer graphene (MLG). However, the grain size was small, in the range of 30 200nm. Later, the SiC was thermally annealed at an elevated temperature of 1250 ~ 1600 °C in an 34 Due

PAGE 24

24 to the existing SiC processing techniques and application, the epitaxy growth obtained wa s held to be promising for potential application. 2. 3 .2.5 Others A few other techniques have recently been developed for synthesizing graphene. Roll to roll processing 35 was based on CVD graphene on a large, thin copper substrate. The graphene was coated w ith a layer of polymer, and the roll to roll process followed in order to separate the graphene and copper substrates, which were then ready to be transferred to other substrates. Unzipping of carbon nanotubes was demonstrated by Dmitry et al. 36 and Jiao e t al. 37

PAGE 25

25 Ta ble 2 1. Wavelength for selective lasers 1 L asing medium L aser type Wavelength Solid state R uby Nd:YAG Ti :Sapphire 649 nm 1,064 nm 650 1,100 nm Gas CO2 Excimer 10 193 nm (Ar F) 248 nm (K F) 308 nm (Xe Cl) 351 nm (Xe F) Semiconductor GaN Quantum cascade 0.4 mid far IR Free electron Far IR to vacuum UV F igure 2 1. Schematic of different types of electronic excitations in insulator/semiconductor.

PAGE 26

26 CHAPTER 3 193 NM EXCIMER LASER LIFT OFF FOR ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTORS 3.1 Single P ulse Laser Lift O ff 3.1.1Abstract AlGaN/GaN HEMTs grown on both side polished sapphire substrates were successfully lifted off with a 193 nm UV excimer laser system. The photon energy of the 193 nm laser is larger than the band gap of AlN and thus it can be used to lift off AlGaN HEMT structures with AlN or AlGaN interfacial layers grown on sapphire substrates prior to growth of the GaN buffer layers. The lifted off HEMT chip was warped and showed 25 42% reduction of the saturation drain current. There was no degradation observed either in the forward and reverse gate current voltage (I V) characteristics or on the drain punch through voltage. Based on comparisons of cross sectional electron micrographs, no additional dislocations were created in the HEMT structures after the laser lift shifts were used to estimate the strain relaxation of the laser lifted off samples. 3.1.2 Introduction AlGaN/GaN High Electron Mobility Transi stors (HEMTs) with high electron mobility and saturation drift velocities are promising for high power and high frequency applications 38 46 . Sapphire substrates are widely used to grow HEMT structures, but their poor thermal conductivity is a primary limi tation for GaN based power devices. The heat generated during device operation cannot be dissipated effectively, which also limits the device performance. SiC is an alternative substrate with high thermal conductivity, but the high cost and relatively poor quality of SiC substrates limits their application. Recently, laser lift off (LLO) technology 46 48 has been developed and widely

PAGE 27

27 used for GaN based light emitting diodes (LEDs) to improve thermal dissipation, operating current, and light extraction ratio 46, 49 51 . Typically, a 248nm laser is used to decompose a thin layer of GaN at the GaN/sapphire interface and to separate the GaN film from the sapphire substrate. This process enables the possibility of transferring the GaN film onto substrates with hig her thermal conductivity and lower cost. Although AlGaN/GaN HEMTs have been lifted off, reports of film quality and device performance of these HEMTs have been inconsistent 52 58 . It was demonstrated that film quality was limited by the size of the laser beam spot, with the central region of the lifted off sample having better quality than at the periphery 58 . This difference was attributed to the intensity gradient at the edge of the laser beam spot causing crack generation and crystal structure degradat ion 58 . A 10 15% reduction of the drain current was reported for LLO HEMTs 53 , but higher saturation drain current was also reported for LLO HEMTs 54 . However, no explanation was provided for either the device degradation or enhancement in these reports 53 , 54 . In this work, we demonstrate the use of a 193 nm laser to lift off HEMT structures. Transmission electron microscopy (TEM) was employed to examine the material quality of the LLO HEMTs. Drain and gate current voltage (I V) characteristics were used to ev aluate the effects of the LLO process on device performance. The were used to estimate the strain relaxation of the LLO HEMTs. 3.1.3 Experiments AlGaN/GaN HEMT structures were grown on c plane sapphire s ubstrates by metal organic chemical vapor deposition (MOCVD). The epilayers consisted of 2 µm thick carbon doped GaN buffer layers followed by a 55 nm thick undoped GaN layer,

PAGE 28

28 20nm of Al 0.25 Ga 0.75 N, and a 2.5 nm GaN capping layer. On wafer Hall measureme nts showed sheet carrier concentration, sheet resistance, and mobility of 8.5 × 10 12 cm 2 , 2 implantation was used to electrically isolate adjacent devices. Ti/Al/Ni/Au Ohmic metall ization was achieved by annealing at 850°C for 30 sec for source/drain contacts, Ni/ Au based 1 m gate length and gate width of 200 µm, source to gate and gate to drain distances of 1 and 1.5 µm, respectively. The devices were passivated with 200 nm of SiNx deposited by a Plasma Therm 790 PECVD system. The laser lift off process was carried out using a JPSATM IX 260 ArF excimer pler objective lens with a meniscus corrector was used to correct spherical aberration, and the focal length of the tripler was 10 cm. A metal mask was installed in the light path and the openings on the mask were imaged onto the target surface. The laser pulse duration was fixed at 25 ns and the maximum repetition rate was 100 Hz. The sample stage was designed to accommodate 6 inch wafers. Two He Ne lasers were used to guide the high accuracy air bearing x y linear motor sample stage, and the minimum st age movement was 0.1µm/step. The stage had an accuracy of +/ 3 m over a full range of motion with a stage velocity of 6 8 inches per second; the stage position could be programmed with a resolution of +/ 1 um and stage movement repeatability of +/ 1 um. The fluences used in this study ranged from 650 to 800 mJ/cm 2 for the LLO process and 1.3 mJ/cm 2 for groove drilling. To prevent micro crack formation along the edges of the LLO samples during the LLO process, 30 µm wide grooves were drilled

PAGE 29

29 through the entire AlGaN/GaN HEMT structure from the front side of wafers. Photoresist was first spin coated on the wafer to protect the device from debris generated during laser drilling. The HEMT samples were mounted on AlN or Si wafers with the epitaxial side of the samples contacted to the mounting substrate using adhesive or wax. Figure 3 1 shows photographs of the HEMT sample before and after groove drilling. After the LLO process, 5 minute etching in a 1:1 HCl:H 2 O solution was used to remove gallium droplets from the lifted off GaN surface. Acetone was then used to rem ove the mounting wax to obtain the freestanding LLO HEMTs. The dc characteristics of the HEMTs were measured with an HP 4156 semiconductor parameter analyzer. Cross sections of the HEMT structures prior to and after LLO process were prepared for TEM ex amination using a Nova 200 focused ion beam system. 3.1.4 Results and Discussion Figure 3 2 A shows an LLO HEMT placed on a flat glass slice. Interference patterns are clearly visible in the middle of the sample, which indicates that the LLO HEMT samples ar e not flat. The bending of the freestanding LLO HEMT sample could be due to relaxation of the built in strain from deposition of the SiNx film and metal contacts during the device fabrication, as well as the strain resulting from the lattice mismatch betw een AlGaN and GaN during material growth. Figure 3 2 B shows an image of the laser drilling defined edge for an LLO HEMT; the edge is very sharp and no micro cracks are evident. Without drilling grooves along the HEMT structure prior to the LLO process, t he edges of the laser lift off area would have cracks due to the impact of the shockwave created during the high fluence laser exposure, which would separate

PAGE 30

30 the laser exposed area and the unexposed region but with the presence of irregular edges and micro cracks. Figure 3 3A shows the drain I V curves of a typical HEMT before and after LLO process. Both curves exhibit good pinch off characteristics, but the saturation drain currents of the LLO HEMTs are decreased significantly relative to the control va lues. About a dozen of the LLO HEMTs were examined and the saturation drain current of these HEMTs after LLO process ranged from 25 42% lower than the HEMTs before LLO process. However, the threshold voltage of the HEMT was on l y slightly shifted, as show n in Figure 3 3B. The threshold voltage was obtained from the intercept with the x axis of the line fitted to the linear region of the square root of the drain current Since there were no micro cracks exhibited on the LLO HEMT samples, the saturation cur rent degradation could not be due to micro crack formation present on typical LLO samples. Schottky gate and drain punch through characteristics were also examined to determine whether the AlGaN Schottky contact layer or GaN buffer were damaged during the LLO process. As shown in Figure 3 4 A , gate I V characteristics of a typical HEMT prior to and after the LLO process were very similar, and the gate I V curves for some of the LLO HEMTs actually revealed slightly lower reverse leakage current and higher Schottky barrier height. These improvements might be due to the piezoelectric effect from the Schottky contact to AlGaN layer on the states of surface traps, however, this need to be confirmed with modeling. Based on the Schottky characteristics, there w as no evidence that the AlGaN layer was damaged during the LLO process. Figure 3 4 B illustrates the drain punch through voltages of the HEMT before and after the LLO process. The gate dimensions of the HEMT was 1µm × 200

PAGE 31

31 µm and the distance between the g ate electrode and drain contact was 38 µm. The HEMT showed a 650 V drain punch through voltage and there was no difference between the HEMT before and after the LLO process. Thus, the saturation drain current reduction of the LLO HEMTs must be related to the strain relaxation. In order to separate the effects of the strain induced by the lattice constant mismatch between AlGaN and GaN from the built in strains of SiNx film and metal contacts deposited on the HEMT structure, circular samples of AlGaN/GaN HE MT structures without SiNx or metal contacts were also lifted off to evaluate the partial relaxation of the lattice constant difference induced strain between AlGaN and GaN. Prior to the LLO process, a 20 µm wide circular groove was drilled from the fro nt side of the sample defining the radius of a circular sample, which was chosen to be 100 µm. When the LLO circular samples were placed on a Si wafer, circular diffraction patterns were observed, as illustrated in Figure 3 5 A . These circular diffraction patterns are 60 . The LLO circular sample was warped due to the partial relaxation of the strain between the AlGaN and GaN layers. The radius of the warped circular LLO HEMT sample, R, as shown in Figure 3 5 B , was estimated based o n the R = r k 2 / k , ( 3 1) Where R is the radius of the warped sample, k is the number of dark rings and k=1, 2, k is the radius of kth dark ring and is the wavelength of light. The partial relaxed strain induced deflection, d, is calculated using , if ( 3 2)

PAGE 32

32 , of the c ircular LLO HEMT was estimated by treating the LLO sample as a single beam. The tensile strain near the top surface of the beam is given by, = td/ L 2 ( 3 3 ) Where t is the sample thickness, d is the deflection and L is the length of the beam. Table I s ummarizes the calculated radius of the warped circular LLO HEMT sample, as well as the partial relaxed strain induced deflection and the partial relaxed strain for the circular LLO HEMT sample. To verify the partial relaxed strain estimated using Newton Raman scattering measurements were also performed on the same sample before and after the LLO process. For GaN, the E2 peak has been shown to be very sensitive to stress, and the E2 peak shift is proportional to the biaxial stress change ( 3 4 ) 1 (GPa) 61 . As illustrated in Figure 3 6, the wave number of the HEMT sample before and after LLO process are 569.01 cm 1 and 567.31 cm 1 , respectively. Thus 1.697 cm 1 0.2737 GPa. The relaxed strain resulting from this stress relaxation is obtained from 62 GaN / (1 ( 3 5 ) Where E GaN (0.183) 63, 64 . The partial relaxed strain estimated from the Raman spectrum E2 peak shift method is 7.87 × 10 4 of 2.4 × 10 4 estimated by the

PAGE 33

33 For the A l 0.25 Ga 0.75 N/GaN HEMT structures, the theoretical strain due to lattice mismatch between Al 0.25 Ga 0.75 N and GaN is 6.04 × 10 3 , which is derived from a linear interpolation between the lattice constants of GaN and AlN 47 : a(x) = ( 0.077x + 3.189) × 10 10 m ( 3 6 ) x Ga 1 x N and x is the atomic fraction of the Al in spectroscopy methods only accounted for 5 10% of the original strain 65 . As s hown in Figure 3 5 A , the interference patterns were observed in the middle of the sample, but not the device active area. Hence, the bending of the sample around the device active area was too large to create interference patterns owing to additional strai n relaxation for the strain built in by the SiN x film and metal contact. Multiple probes were also used to flatten the HEMT sample while measuring the drain current and higher drain current was obtained. Once three probes were used for the measurement, th e drain current reduced to the original level. Thus, the drain current reduction of the LLO HEMT was mainly due to strain relaxation and no apparent damage was created during the LLO process. 3.1.5 Conclusion AlGaN/GaN HEMT devices were lifted off using a 193 nm laser. Laser drilled grooves on the front side of the sample were employed to prevent the formation of micro cracks during the LLO process. Besides a 25 43% reduction of the saturation drain current, no degradation of other dc characteristics was observed. The LLO HEMT was bent due partial relaxation of the as grown strain as well as strain arising from were used to estim ate the partial relaxation of the strain owing to the lattice mismatch

PAGE 34

34 between Al 0.25 Ga 0.75 N and GaN. The quantity of partial relaxation for lattice mismatch only accounted for 5 10% of the drain current reduction. The majority of the drain current reduc tion was attributed to the relaxation of the strain caused by dielectric and metal depositions. 3.2 Laser Lift Off GaN with Multiple P ulses. In the last section of this chapter, we demonstrated the capability of doing laser lift off by one pulse at 800 mJ/cm 2 . To prevent the crack formation, grooves around devices were used. And this method also limits the quality and size of the laser lift off GaN thin film. To achieve a larger area laser lift off, a line scanning method was used in this work. Figure 3 7 A is the schematic map of doing laser lift off large area GaN thin film. 100x100 m laser beam was used; the step width of line scanning from x direction is 100um. Therefore the beam overlap between two connected laser pulses is 0um. After laser finished the first line, the next line below was processed and the step width of y direction is also 100um. The laser lift off process finished when all the areas are scanned. Figure 3 7 B is the image of GaN thin film after large area laser lift off based on the s trategy described in Figure 3 7 A . Clearly we can see cracks around laser beam edge, and stylus profiler indicates that some of these cracks can propagate through the 2 m thick GaN to the front side. Different overlap condition has been test from 1 to 20 m, but a similar crack formation is inevitable on the beam edge. The crack formation is related to the GaN decomposition during laser lift off. When GaN at the GaN/Sapphire interface absorbs the laser energy and transfers it into heat, the interface temper ature

PAGE 35

35 could rise up to 1600 K, 66 and GaN decomposes to Gallium and Nitrogen gas, and forms plasma. It can be classified as a photo thermal process. The more GaN decomposes, the more Gallium and Nitrogen are generated, and the easier to form a crack during the laser lift off separation process, as demonstrated in Figure 3 8 the cross sectional drawing of laser lift off at 800 mJ/cm 2 . These cracks limit the size of the laser lift off GaN thin film and limit its application eventually. We could use a larger la ser beam to generate less cracks and larger laser lift off thin film, however, an infinite big laser beam is not realistic. A better alternative should be developed with the same laser beam size, and minimizing the amount of the Gallium and Nitrogen is the key to prevent cracks. Besides the obvious crack formation, a further investigation of material integrity was also performed by cross section TEM. Figure 3 9 is the TEM images before and after laser lift off. The amount of the dislocation through the 2 um epi layer increased after laser lift off. There are two major reasons for the dislocation propagation. One is temperature and the other is mechanical deformation. Once temperature is above the brittle to ductile (BTB) temperature, 67 stress can relax by d islocation flow, and the BTB temperature for GaN is 1400 K. Based on the thermal simulation the peak temperature could reach 1600 K at 800 mJ/cm 2 , 66 and it is well above the BTB temperature. Mechanical deformation happens when the GaN thin film separates from the sapphire substrate. Due to the high pressure and temperature, the Gallium and Nitrogen plasma not only causes the crack on the laser beam edge mentioned previously but also deforms the lift off film as demonstrated in Figure 3 8. And this deformat ion could boost the dislocation propagation.

PAGE 36

36 Attempts with various fluences were conducted, from 400 to 1200 mJ/cm 2 , and the threshold fluence to lift off the GaN is 800 mJ/cm 2 . Above this fluence, laser can generate more plasma and make the cracks worse; below this fluence, laser cannot lift off the GaN successfully. Therefore, 800 mJ/cm 2 is the best condition could be used for single pulse laser lift off, but still creates cracks and generates the dislocations. To generate less gallium and nitrogen durin g the laser lift off, multiple laser pulses with lower fluence were used to lift off GaN thin film. Figure 3 10 A 66 is the image of sample before laser lift off. There are grooves drilled around 100x100um squares. Figure 3 10 B is the image of the one pulse at 800 mJ/cm 2 . A silver shining color indicates the existing of Ga metal, and the success of laser lift off. Figure 3 10 C and D are the images of 1 pulse and 5 pulses at 420 mJ/cm 2 , respectively, and there seems no changes before and after laser processing. But with 50 pulses, figure 3 10 E shows a start of laser lift off. And with 160 pulses, Figure 3 10 F shows a complete laser lift off. By doing multiple pulses at lower fluence, the interface peak temperature is expected to be lowered from 1600k to 1050k based on the simulation 66 During this process, the photon energy was used to break bonds between gallium and nitrogen, and also to generate heat, therefore, it is a photo chemical process. Figure 3 11 is the cross sectional drawing for multiple p ulse process. During 420 mJ/cm 2 multiple pulse laser lift off, the amount of gallium and nitrogen were limited to minimal amount, which lowers the possibility of the cracks and dislocations formation. Figure 3 12 A is the schematic map for the new laser lif t off strategy with multiple pulses at lower fluence. The numbers indicate the laser exposure sequence. And there is a 10 m overlap between two connected pulses. Figure 3 12 B is the optical

PAGE 37

37 image after laser lift off based on Figure 3 12 A . Although the s urface looks dirty, there is no cracks at the beam edge, and the surface roughness is 2~3nm measured by AFM. In summary, the cracks were generated and the dislocation propagated during the single pulse laser lift off. And the mount of gallium and nitrogen plasma was the key to control the quality of laser lift off GaN film. Multiple pulse laser lift off was developed and used to fabricate crack free laser lift off GaN thin film with good quality.

PAGE 38

38 Table 3 y placing circular LLO HEMT on a Si wafer and the estimated strain relaxation of the LLO HEMT. # of the Newton ring, k Radius of the Newton ring, r k (µm) Radius of the curvature for LLO HEMT, R (µm) Bended distance of the LLO HEMT, d, (µm) strain 1 48.5 4163 1.2 2.4 × 10 4 2 68.5 4152 1.2 2.4 × 10 4

PAGE 39

39 Figure 3 1. Photographs of HEMT A ) before and B ) after drilling grooves along the edges of the device.

PAGE 40

40 Figure 3 2. Photograph after LLO process. A ) Photograph of a LLO HEMT placed on a flat glass substrate. B ) Photograph of the laser drilled edge.

PAGE 41

41 Figure 3 3. Device performance before and after LLO (I) . A ) Drain I V characteristics of AlGaN/GaN HEMT before and after the LLO process. B ) Square root of the drain current as a function of the gate voltage of the AlGaN/GaN HEMT before and after the LLO process.

PAGE 42

42 Figure 3 4. Device performance before and after LLO (II). A ) Forward and reverse gate I V before and after laser lift off. B ) Drain punch through voltage measured at a Vg of 7V for the AlGaN/GaN HEMT before and after laser lift off.

PAGE 43

43 Figure 3 5. Photograph of Newton curved LLO sample . A ) Photograph of a LLO circular AlGaN/GaN HEMT sample placed on a Si B ) Schematic of the LLO circular AlGaN/GaN HEMT sample.

PAGE 44

44 Figu re 3 6. Raman spectrum E2 peak from a HEMT sample before and after the LLO process.

PAGE 45

45 Figure 3 7. Illustration of LLO process A ) Schematic map for x and y direction scanning, B ) Image after laser lift off. Figure 3 8. Drawings show crack formation Figure 3 9. TEM images A ) GaN on sapphire as grown B ) GaN thin film after laser lift off.

PAGE 46

46 Figure 3 10. Multiple pulse LLO experiments before and after LLO A ) Photograph of the HEMT samples prior to the LLO process and photographs exposed with different LLO conditions; B ) Exposed with a single pulse of laser at the fluence of 800 mJ/cm 2 . C ) Exposed with 1 laser pulse at the fluence of 420 mJ/cm 2 . D ) 5 laser p ulses at the fluence of 420 mJ/cm 2 . E ) 50 laser pulses at the fluence of 420 mJ/cm 2 . F ) 160 pulses at the fluence of 420 mJ/cm 2 . 66

PAGE 47

47 Figure 3 11. Cross sectional schematic drawing of multiple pulse laser lift off.

PAGE 48

48 Figure 3 12. Illustration of multiple pulses LLO A ) Schematic drawing for the large area laser lift off by using multiple pulse method. B ) Image after multiple pulse laser lift off, with no cracks at the beam edge.

PAGE 49

49 CHAPTER 4 GRAPHENE SYNTHESIS BY PULSED LASER AND THERMAL ANNEALING ION IMPLANTED SIC 4.1 Abstract In previous studies 68 72 techniques were presented whereby graphene (G) and few layer graphene (FLG) could be synthesized on SiC by ion implantation (II) and pulsed laser annealing (PLA) or thermal anneal ing. These processing techniques selectively graphitize regions of SiC only where ions have been implanted and without elevating the temperature of the surrounding substrate. Nanoscale features can be patterned over large areas by multi ion beam lithogra phy and subsequently converted to G or FLG via PLA in air. Samples were characterized using Raman spectroscopy, ion scattering/channeling, SEM, AES, and AFM, from which the degree of graphitization was determined to vary with implantation species, damage a nd dose, laser fluence and pulsing. This thesis is an extension of these studies that addresses the contrasting growth regimes and graphitization mechanisms that occur during II and PLA. 4.2 Introduction The design and synthesis of two dimensional (2D) m aterials has recently inspired its own branch of materials science. Since its discovery, 13, 73 76 graphene has been at the forefront of this latest discipline, and is emerging as a promising material system. 5 Graphene has demonstrated exceptional electrical, 77 optical, 78,17 chemical, 18 , 20 and mechanical properties, 79 , 80 including carrier mobility greater than 100,000 cm 2 / V s, 13 and unlike the other well studied carbon allotropes nanotubes and fu llerenes graphene is compatible with planar processing technologies developed for silicon. Some formidable obstacles remain before graphene devices will compete with their silicon counterparts. So far the highest quality devices have been fabricated on small

PAGE 50

50 flakes exfoliated from graphite, 13 , 73 but this app roach is not suitable for wafer scale device integration. Large area graphene has been grown on various metals using chemical vapor deposition (CVD); however, this technique requires the transfer of the graphene onto insulating substrates, chemical proces sing, etc.. 81 83 A promising technique for forming large area graphene directly on insulating substrates is to anneal SiC single crystals at high temperatures (~ 1400 C) in ultra high vacuum (UHV). 34 , 84 This results in Si sublimatio n, leaving behind a C rich interface leading to the growth of graphene suitable for the fabrication of electronic devices. 34 , 85 This process is, however, a costly and time consuming method for producing electronically isolated graph ene. Since graphene is a zero gap semimetal, graphene devices rely on quantum confinement, 86 , 87 strain, 88 doping, 85 87 or perpendicular electric field modulation 82 to achieve desired performances, but these require additional and often complicated processing steps. Currently, conventional processing technologies such as photolithography, e beam lithography, and dry etching (O 2 ) are used to fabr icate devices. This exposes the graphene sheets to various polymers and harsh chemical/mechanical treatment, thus leading to reduced mobility and unintentional surface and in terface properties is essential for maximizing device performance. The techniques of II and PLA offer opportunities to synthesize nanoscale graphene features on SiC without the need for some of these conventional processing and annealing techniques. The II and PLA techniques themselves each have multiple processing features and these are explored in this thesis. There have been promising reports of graphene synthesized on c SiC by PLA. For example, Perrone et al 93 used a

PAGE 51

51 q switched Nd:YVO 4 to anneal the C face of 4H SiC in Ar and reported evidence for graphene formation, and Lee et al 94 have reported that graphene can be grown on SiC single crystals in vacuum when irradiated with 500 pulses from a KrF excimer laser at fluences of 1 1.2 J/cm 2 . 4.3 Experiments 4 .3.1 Ion I mplantation. Commercially available 4H SiC was implanted over broad areas using the accelerator facilities at the Australian National University (ANU) by Dinesh K. Venkatachalam and Robert G. Elliman. The implant conditions were 60 keV Au at 3.6 x 10 16 Au/cm 2 , 40 keV Cu at 8.0 x 10 16 Cu/cm 2 , and 40 keV Ge at 3.5 x 10 16 Ge/cm 2 . Figure 4 1 shows the SRIM simulation results of these implant conditions. These implantations created amorphous surface layers on the 4H SiC that were measured by ion scattering/channeling measurements at ANU and are listed in Table 4 1. Ion implantation nano ribbons were implanted using the Raith multi ion beam lithography (MIBL), nanofabrication, and engineering (MionLiNE) syste m at the University of Florida 68 72 , with Au at 35 keV 5x10 16 Au/cm 2 . 4.3.2 Pulsed Laser A nnealing. The laser system that we used is JPSA IX 260 ArF excimer laser with 193nm wavelength, 25 ns pulse duration, 1 100 Hz repetition rates and 0.1 to 5 J/cm 2 fluences. Figure 4 2 is the schematic drawing of the laser system. The laser beam size could be adjusted from 20µm x 20µm to 400µm to 800 µm by changing masks. The laser beam deliver system is equipped with double 6 x 6 Fly Eye array beam homogenizer to gu arantee the beam uniformity, which is important when we are doing PLA. The motorized stage with 0.1µm accuracy can hold a 6 inch wafer, and can be programmed

PAGE 52

52 to run automatically during laser processing. In our experiments, both implanted and un implanted regions were annealed with the 193 nm pulse laser. Eight 45 µm x 45 µm square areas were sequentially annealed with 1, 2, 5, 10, 50, 100, 300 and 500 pulses, in each region, at various fluences ranging from 0.2 0.8 J/cm 2 as illustrated in Figure 4 3 4 .3 .3 Characterization . The PLA regions were analyzed using micro Raman with a 532 nm green laser, scanning electron microscopy (SEM), atomic force microscopy (AFM), and optical microcopy. Graphene has signature Raman peaks of G peak (1580 cm 1 ), a primary in plane vibrational mode; 2D peak (2690 cm 1 ), a second order overtone of a different in plane vibration; and D peak (1350 cm 1), defect peak. 95 And it makes Raman spectra ideal for graphitization indication. The size of the micro Raman beam was one micron and this system, with the precisely controlled motorized stage, is capable of mapping the entire PLA spot area. 4.4 Results and Discussion. Figure 4 3, shows optical images of 3 implanted areas after PLA at different fluences and pulses in air. The visibl e squares are the 45 um x 45 um PLA spots, and the surrounding areas are implanted but not annealed. The fluence and pulse number are listed along the top and left hand side. Comparing the color contrast of the images, there is more change after PLA with m ore pulses and higher fluences. I. Jung et al. reported the colors of graphene and graphene oxide multilayers on various substrates. 96 The author concluded that the color change was related to the graphene or graphene oxide layer thickness and dielectric films thickness or type. And it could be a simple and

PAGE 53

53 helpful way to indicate the amount of graphene or graphene oxide layers. In our case, the contrast change is possibly related to the formation of graphitic layers. Raman analysis of these laser annealed regions shows an increasing G signal of FLG with more pulses and higher fluence, and the areas PLA below 50 pulses do not show any G growth. Figure 4 4 summarizes Raman spectra of PLA spots with 50 pulses at dif ferent fluence, and Figure 4 4Aa, Bb and C c list the Au, Cu and Ge implanted sample re sults respectively. Figure 4 4a, b and c display relative G peak intensity versus fluences. We define the onsite graphitization fluence G as the fluence where the G band starts to rise at the certain pulse number, and it is 50 pulses in our case. Therefore the onsite graphitization fluence for Au, Cu and Ge implanted SiC is 0.4, 0.2 and 0.4 J/cm 2 with 50 pulses, respectively. In the same ma nner, we also find that the onsite graphitization fluence G for pristine SiC is 0.8 0.9 J/cm 2 (Not shown). In a previous paper , 68 S. Tongay et al. reported that when II SiC wafers were thermally annealed in vacuum at temperatures ~ 100 200 °C below the temperatures where graphitization of crystalline SiC occurs, that G and FLG were formed only where ions were implanted and the surrounding areas remained crystalline SiC. Therefore it is not surprising to see a lower onset graphitization fluence for the P LA ion implanted SiC regions compared with pristine SiC. Table 4 1 summarizes the onset graphitization fluence for the amorphous layer thickness created by II. There seems to be no clear trend to indicate that amorphous layer thickness results in differ ent onset fluences. Early experiments by Baeri showed that when amorphous layers produced on SiC crystals by II were annealed with a pulsed Ruby laser, that the fluence where the amorphous layer began to melt was

PAGE 54

54 strongly dependent on the thicknesses of th e amorphous layers . 97 His results showed that the melting laser fluence of surface a SiC layer was lower with increasing thickness of the a SiC layer thicknesses of ~ 30 nm began melting at ~ 1.3 J/cm 2 , but our G is much lower than this value. His experiments were with a different type laser and implanted ion species than ours but as results and thermal simulations presented later in this thesis show, melting is unlikely in our PLA results. We noted that Cu and G e implanted samples have similar amorphous layer thickness, but a 2 times difference of G , which implies that there could be a catalytic effect or doping effect that has been introduced by different ion species. Similarly, Cu has been used as a substrate and catalyst in CVD growth of graphene, which also implies that Cu could be a good catalyst during the PLA process. A further study will be presented in the next chapter. In this chapter, results on broad area implantation samples were presented, and disc ussed. In the industry, nano area implantation and laser annealing are commercially available and be integrated to existing microfabrication process. To demonstrate the previous paper , M. Lemaitre, S. Tongay, X. Wang et al. 69 , we PLA implanted nano ribbons and selectively graphitized only this area. In Figure 4 5, 4H SiC was patterned with Au ions into 2 m x 10 m nano ribbons, and annealed with PLA at 0.8 J/cm 2 , 100 pulses. Figure 4 5 A is the SEM image of this area Figure 4 5 B is the Raman map of the 2D band intensity, showing that only the implanted area was graphitized by PL A process . Micron size lines were patterned to show up more clearly with SEM but nanometer precision is achievable with the MIBL system.

PAGE 55

55 4.5 C onclusion We have demonstrated that II combining with PLA could synthesize few layer graphene selectively. We observed tha t the II region has much lower onset graphitization fluence than crystalline SiC, which contributes the selective graphene growth. These results also raise questions regarding the graphitization mechanism during PLA of the II regions. Both amorphous surfac e layer effects and catalytic or doping effects seem to contribute to the G growth, however a detailed and systematic study is required to address these questions. Moreover the laser processing parameters need to be optimized to gain better control on both the quantity and quality of the graphene layers.

PAGE 56

56 Table 4 1. Comparison of the a SiC surface layer thicknesses produced by II and the PLA fluences at which graphitization begins for each. Ion implantation condition s Am orphous SiC thickness (nm) Onset gra phitizatio n fluence (J/cm 2 ) Au, 60 keV 30.5 0. 4 Cu , 40 keV 40.5 0.2 Ge, 40 keV 41.5 0.4 Figure 4 1 . SRIM calculations of the damage depth profiles produced by ion implantation of Au, Cu, and Ge in SiC.

PAGE 57

57 Figure 4 2 . Schematic diagram of the193 nm ArF excimer laser used in our work (See text for details). Figure 4 3 . Optical images of ion implanted samples of 4H SiC after PLA in air at different pulse number and fluence. A ) Au implant sample. B ) Cu implant sample. C ) Ge implant sampl e.

PAGE 58

58 Figure 4 4. Raman spectra of ion implanted SiC before and after PLA at different fluences, and relative G Peak intensities versus fluence for Au, Cu, Ge implanted SiC.

PAGE 59

59 Figure 4 5. Growth of FLG with nanoscale features by Au MIBL and PLA A ) Reproduction of a plasmonic terahertz metamaterial consisting of a micro array of 2 µm wide lines. B ) Raman 2D band map of the metamaterial array. Scale bars are 1, 2, and 2µm, respectively. 69 .

PAGE 60

60 CHAPTER 5 SELECTIVE GROWTH OF GRAPHENE AND CARBON NANO STRUCTURES ON AMORPHOUS SIC VIA PULSED LASER ANNEALING 5.1 Introduction

PAGE 61

61 In this paper, experiments and analyses are presented to differentiate the process SiC on c SiC, and the same a SiC layers doped with a variety of dopants, and that occur as a result of pulsed laser annealing. We also show that under certain conditions additional graphitic na no structures are produced. 5.2 Experiments and Thermal S imulations Commercially available 6H Silicon carbide (SiC) samples obtained from II VI were used for ion implantation. Five c SiC samples were implanted with Si + and C + ions at various energies and doses to create stoichiometric surface layers of a SiC of desired thicknesses by Dinesh K. Venkatachalam and Robert G. Elliman at Australian National University. The amorphous surface layer thicknesses were simulated using SRIM evaluations of and verified by Rutherford Backscattering (RBS) at the Australian National University; detailed implantation conditions are listed in Table 5 1. Both implanted and un implanted pristine SiC were annealed using a JPSA IX 260 ArF laser at University of Florida (193 nm wa velength, 25 ns pulse width, 50 Hz repetition rate). In order to optimize the PLA parameter, the laser fluence was varied from 0.1 1.2J/cm 2 , and the number of pulses was altered, up to 2000 pulses. Raman spectra were obtained using the Horiba MicroRaman Sp ectrometer with a 532 nm green laser. Cross sectional transmission electron microscopy (X TEM) images were obtained by JEOL 2010F TEM. 3D surface images were taken using a Bruker Optical Profilometer. A finite element analysis was used to estimate the surf ace temperature during the PLA process. We used an unsteady state energy balance equation with rectangular coordinates (x , y , and z axes) 52 to approximate the surface temperatures of ion implanted samples after exposure to one laser pulse (pulse time = 2 5 ns), as shown by:

PAGE 62

62 (5 1) where T is temperature, is density, C p is heat capacity, is absorption coefficient, k is thermal conductivity (in units of W/cm K), z is distance from the sample surface, t is time, and I is absorbed power during laser exposure. This can be represented as a function of z and t by using (5 2) where I 0 is the incident laser intensity at the sample surface (z=0) and R is the reflectivity of the specimen. 5.3 Results and Discussion 5 .3.1 Thermal S imulation In our study, a nanosecond pulsed laser was used to heat up the amorphous SiC (a SiC) on the c SiC substrates. It is believed that silicon atoms are sublimated during this process, and the surface is enriched with carbon atoms to form gr aphene. However, the temperature during this 25 ns laser pulse is almost impossible to determine using instruments. Fortunately, thermal simulation by finite element analysis can be used to calculate the surface temperature. To study the mechanism of pulse laser annealing (PLA) of a SiC, we had to ascertain whether the laser fluence was high enough to melt the surface, and what kind of temperature range would result from our optimal laser annealing conditions. Figure 5 1 shows the laser fluence required to melt a certain thickness of a SiC on SiC, based on thermal simulation results. During the simulation we used 2445K as the a SiC melting temperature; 97,98 the rest of the thermal properties of a SiC and SiC are listed in Table 5 2. The trend of decreasing f luence with increasing

PAGE 63

63 thickness is due to the poor thermal conductivity of a SiC compared to c SiC. When both a SiC and c SiC were irradiated with 193 nm laser, part of the laser energy was reflected, and part was absorbed by SiC and transferred to heat. The heat can be dissipated through the bulk material underneath the absorbing surface, and a SiC with a poor thermal conductivity could conduct heat more slowly, and surface temperature increase faster. Therefore, surface temperature at the a SiC surface could rise up faster than c SiC at the same laser fluence. When we compare the a SiC samples with different thickness, although they have the same poor thermal conductivity, their different heat resistance is different, which is proportional to the materia l thickness. As a result of high heat resistance for thick amorphous layer, the heat could not be conducted very well, and the surface peak temperature could be higher even at the same laser fluence. Although thermal simulation provides us with a good meth od of determining the surface temperature during laser annealing, it can only be applied to the initial laser pulse on the material. This is because after the first laser pulse, part of the amorphous SiC may be crystallized; this phase change alters the th ermal property used in the simulation. Additionally, it renders the SiC unable to acquire the new thermal property of the amorphous and poly crystalline mixture. Therefore, the simulations for the second, third, or subsequent pulses are not going to provid e as accurate results as the first pulse. However, this still constitutes a useful method of roughly estimating the temperature range during laser annealing. Fortunately HR XTEM often provides confirmation of whether the sample surface melts or not as esti mated. 5 .3.2 Amorphous E ffect amorphous layer. Silicon and carbon ions were implanted into the SiC in identical doses

PAGE 64

64 of around 1.5×10 15 ions/cm 2 , but with various energies from 4 0 to 160 keV, in order to form different thicknesses of amorphous layers. Table 1 lists the implant conditions and the amorphous layer thicknesses. Samples 1, 2, 3, as well as pristine c SiC were used for pulsed laser annealing; the condition was 800 mJ/cm 2 with 100 pulses at 50 Hz. Raman spectra are widely used in research to indicate graphitic signals. Raman peaks at 1580 cm 1 , 2700 cm 1 , and 1350 cm 1 are named as G, 2D, and D peaks from graphene, respectively. 95 The G and 2D peaks are a result of the Stokes phonon energy shift by laser excitation (in our case, the 532 nm green laser); these two peaks indicate the existence of sp 2 carbon, and 2D is the defect peak. In Figure 5 2, the pink data line represents the pristine c SiC after laser annealing, and does not show any difference from the c SiC before annealing (blue line). This indicates that this fluence (800mJ/cm 2 ) and pulse number (100) are not enough for a pristine sample to be graphitized. However, under t he same conditions, the graphitic signal started to grow on sample 1, which had a 70 nm thick amorphous SiC on top; the G and D peaks rose slightly compared with the pristine SiC. We also observed that with the increase in the thickness of the amorphous la yer in sample 2 (99 nm) and sample 3 (124 nm), the G, 2D, and D peaks rose significantly, which meant that more and more sp 2 carbon bonds were formed during the laser annealing process. However, a closer look at the G and D peak intensity, as shown by I G a nd I D , reveals that they had a similar growth rate over different samples, which indicates a similar quality of graphitic carbon. Under the same laser annealing condition, a SiC had a higher surface temperature than c SiC; graphene grew selectively on the amorphous region but not on c SiC. We call this the 2 , we also attempted other

PAGE 65

65 fluences, ranging from 0.1 to 1.2 J/cm 2 ; the Raman spectra showed a similar trend, with creasing along with the amorphous layer thickness. In order to study the mechanism of PLA in a SiC/c SiC, what kind of structural changes the laser induces, and how amorphous SiC assists sp 2 carbon bond formation, we used HR XTEM to examine the near surfa ce region of sample 3 after PLA at 800mJ/cm 2 , 100 pulses. As indicated in Table 5 1 and Figure 5 3, sample 3 had an amorphous layer of 124 nm before PLA, with the fluence being enough to melt the surface. In Figure 5 3, the sample surface was melted by PLA during 25 ns laser duration, cooled down quickly after 25 ns, and froze until the next pulse. Figure 5 3 A shows an overview of the sample surface; the area above the white line represents the Cr protective layer that was deposited for TEM sample preparati on. The white area around 50 nm thick, below the line, represents amorphous SiC as labled. The region below the amorphous SiC was melted and re crystallized from amorphous SiC to poly crystalline SiC. The bottom region, with pristine SiC, was not affected during laser annealing. After taking a closer look, we also observed that most graphitic layers were concentrated at the interfaces of different phases. Figure 5 3 B shows a magnified image of the Cr/a SiC interface; there is few layer graphene (FLG) on top of the amorphous layer. It seems that the a SiC can form FLG at 0.8J/cm 2 , 100 pulses, which estingly, we also found some onion like nanostructures in the deeper amorphous region and in the white region in the polycrystalline layers. Figure 5 3 C shows a carbon onion of around 10 nm in size. Carbon onions were first observed by Iijima in 1980 1 01 , a fter heating

PAGE 66

66 carbon black rods to extremely high temperatures (>4000K). Other synthesis methods have since been recognized, including high energy electron irradiation of carbon SOOT, 1 02 arc discharge in water from a carbon target, 1 03 and carbon ion implant ation in Cu substrates. 1 04 All of these methods involve non equilibrium conditions such as rapid heating carbon containing precursors to high temperatures or ion implantation where ion induced damage cascades quench rapidly. With this in mind, the presence of carbon onions in sample 3 most likely indicates that a non equilibrium process, such as laser melting and rapid solidification of SiC, has occurred, which is consistent with the thermal simulation results that show the surface temperature leaps from ro om temperature to melting temperature and beyond during the 25 ns pulse duration, and quickly drops back to room temperature after 25 ns. In our case, the carbon onion formed from carbon atoms in the amorphous SiC. In the amorphous region, the possibility of carbon atoms bonding with each other in 3 dimensional space with a spherical and multi shell arrangement is much higher than that of 2 dimensional planar bonding. Carbon onions are likely to form in the amorphous region. However at the surface or interf aces, on the contrary, carbon atoms favor 2 dimensional bonding along the interfaces to form FLG, and these interfaces could be regarded as a template. With more laser pulses, there are more silicon atoms driven away near these FLG, more carbon atoms avail able for bonding around the FLG, and this additional bonding could further fold the FLG and form onion shaped nanostructures. All these TEM images indicate that the formation of FLG mainly stabilizes at interfaces such as the sample surface or amorphous/cr ystalline interface, and that carbon onions are more likely to grow in thick carbon rich amorphous regions away from interfaces. Therefore, a thinner

PAGE 67

67 amorphous layer enhances FLG formation and limits the possibility of carbon onions growth. 5 .3.3 Catalytic E ffect of Au + I mplanted S amples prepared by implanting equal amounts of Si + (1.5×10 15 ions/cm 2 , 10 keV) and C + (1.5×10 15 ions/cm 2 , 4 keV) into 6H SiC to form 20 nm thick amorp hous SiC surface layers. The first sample was used as a control sample and the second sample was implanted into the amorphous region with Au + (5×10 15 ions/cm 2 , 30 keV). The thickness of the amorphous layers was measured by ion backscattering at ANU and ver ified as 20 nm by HR XTEM images at the University of Florida. These two samples were then laser annealed with 2000 pulses at 800mJ/cm 2 in Argon ambient. Figure 5 4 A shows a TEM image of amorphous SiC without an Au + implant; it shows that two or three laye rs of graphene have formed. Figure 5 4 B shows a TEM image of amorphous SiC with Au + implants, which has nine or ten layers of graphene formed with uniform coverage. This result is consistent with Raman spectra as seen in Figure 5 5. Figure 5 5 compares the Raman spectra of the two samples after the PLA process with that of pristine SiC. Raman spectra were normalized at the SiC peak (SiC TO ) located at 1520 cm 1 . The G, D, and 2D peaks rose for both PLA samples, indicating that both were graphitized. A strong er relative intensity of all three peaks (shown by the red curve) for the a SiC with an Au + implant demonstrates that there was more sp 2 carbon on the surface. In addition, we observed that the 2D peak of the Au + implanted sample was located at 2697 cm 1 a nd that the 2D peak of the control sample was located at 2676 cm 1 . This red shift implies that there were more layers of graphene in the PLA Au + implanted sample, 1 3 which agrees with the evidence of the TEM images. The mechanism of the

PAGE 68

68 Au + used as substrates and catalysts for graphene growth in the CVD method. 32 , 106 The low solubility of carbon in these transition metals leads to precipitation of carbon atoms, which form graphene when they cool down; this is a self limiting process. Conversely, silicon and gold can form Au Si eutectic alloys at a much lower melting temperature; 1 07 such an alloy can dissolve certain amount of silicon atoms in a Si C system. Therefore, the m echanism we propose for graphene growth during PLA Au implanted SiC is , when the laser heats the surface to a high temperature, silicon atoms start to de bond with carbon atoms by either sublimation or eutectic alloying with gold, leaving the surface enric hed with carbon atoms that bond with each other to form graphene. However, it is still not clear that if the Au/Si bonding could increase the sublimation rate. Both interactions between Au/Si and Au/C encourage graphene growth. In our study, as a result of Au + implantation, more layers of graphene formed after laser annealing and we that we can better control the laser annealing process and selectively graphitize the Si C surface by ion implantation and pulse laser annealing. Since these two techniques are highly compatible with current device manufacturing methods, they have significant advantages for growing and patterning graphene on SiC. 5 .3.4 Ge I mplanted SiC Variou s implant species have been tested in our experiments including Au, Ga, Cu, and Ge; among these, the unusual color change and Raman spectrum of the Ge implanted sample stands out. Sample 4 was a 4H SiC wafer implanted with equal amounts of Si + (1.8×10 15 io ns/cm 2 , 160 keV) and C + (1.8×10 15 ions/cm 2 , 75 keV) to form a 217 nm thick a SiC surface layer, followed by implantation with Ge + 5×10 16 ions/cm 2 ,

PAGE 69

69 56 keV. We laser annealed this sample from 0.1 J/cm 2 to 1.2 J/cm 2 , with 50 pulses in air, and the color of the spot at a fluence of 200mJ/cm 2 turned black after PLA. A higher or lower fluence did not result in the same color change. The Raman spectrum of this unique spot is plotted in Figure 5 6 as the green curve, which can be compared with results before and after PLA of 50 pulses at 100 mJ/cm 2 . This green curve shows giant G and D peaks and a small 2D peak, which means that the sample surface was strongly graphitized although it is also defected; we did not see similar results under other PLA conditions. In order to study the morphology of this sample, a 3D optical profilometer was used to examine the sample surface. Figure 5 7 A shows a 3D optical image of the spot that was annealed with 50 pulses at 100 mJ/cm 2 ; it has a shallow flat pit around 5 nm deep, wit h an area size of 100× 100 , which is the same as the laser spot size. Figure 5 7 B shows a 3D optical image of the spot that was annealed with 50 pulses at 200 mJ/cm 2 , which reveals a height expansion of ~20 nm over the area that was annealed; the surface on this mesa was rough, which is unlike the flat surface in the other spot annealed at 100 mJ/cm 2 . To study how deeply the surface was affected by this rough texture, HR XTEM was used to explore both samples. Figure 5 8 A shows a TEM image of the 100 mJ/cm 2 spot. There is a 216 nm thick amorphous layer (d 1 ) on top of the c SiC and the layer above that is the Cr film that was deposited prior to sectioning the sample for HR XTEM. This sample surface did not melt, and there is no sign of any recrystallization or graphitization occurring, which agrees with the Raman spectra in Figure 5 6. Figure 5 8 B shows a TEM image of the 200 mJ/cm 2 spot. There are two layers in between the

PAGE 70

70 crystalline SiC and the protective Cr: one is the 170 nm thick amorphous layer (d 2 ) on top of the c SiC, and the other is a ~70 nm thick layer that is a combination of poly crystalline SiC and porous structures as interpreted from HR XTEM. The total thickness is ~240 nm. Normally, the density increases during the amorphous to crystallin e transition; however, on the contrary, the total thickness of 240 nm, compared with 216 nm, implies a thickness increase and a density drop. This decrease in density may be due to the porous structure created during PLA, which also explains the ~20 nm sur face expansion shown in Figure 5 7 B . Combined with the Raman spectra in Figure 5 6, this suggests that the 70 nm thickness of the poly crystalline SiC and porous structure is mixed with many graphene layers inside. This phenomenon of pores and a strong gra phitic signal was not seen in amorphous samples with other implants like Au, Cu, or Ga in our experiments under the same or different laser annealing conditions. Therefore, we believe that porous formation and graphitization are related to the Ge implants and specific annealing conditions. Similar highly porous surfaces have been reported as the result of ion bombardment of Ge with a variety of ions and under a variety of conditions. 1 08 In these cases craters formed only in the amorphous surface layers of G e crystals created by the ion bombardment, and it was observed that once the porous surfaces were exposed to air large amounts of oxygen and carbon were present, likely because of the increased surface area of the highly porous surface. Although we only ha ve a small amount of Ge implanted in the amorphous SiC, it appears that the Ge may assist in graphene formation as Au does, and the increased surface areas of the porous surface may provide increased nucleation sites for both sp2 bonded carbon and oxygen.

PAGE 71

71 Further study of this curious phenomenon is certainly needed to better determine the key processing parameters and mechanisms. 5.4 Conclusions We studied the surface temperature of pulse laser annealed amorphous SiC by pulse laser annealing of pure amorphous silicon carbide and Au + implanted silicon carbide separately; we also proposed possible mechanisms. Graphene nano structures were synthesized, characterized by Ra man spectra, and imaged using HR XTEM, including carbon onions, few layer graphene, and graphene.

PAGE 72

72 Table 5 1. Tabulation of the implantation parameters used to form stoichiometric a SiC surface layers of various thicknesses on c SiC. Sample Si (10 15 ion s/cm 2 ) Energy (keV) C (10 15 ions/cm 2 ) Energy (keV) d amorphous (nm) 1 1.5 40 1.5 18 69.7 2 1.5 60 1.5 27 98.9 3 1.5 80 1.5 37 123.8 4 1.8 160 1.8 75 216.7 Table 5 2. Parameters of amorphous and crystalline SiC used in thermal simulation calculations. Parameter Amorphous SiC Crystalline SiC (6H) Density, (g/cm 3 ) 2.66 3.21 C p (J/g K) 1.3 1.3 Thermal conductivity, (W/cm K) 0.011 4.9 (106 cm 1 ) 1.1 1.5 R (%) 34.6 40 Melting point, T m (K) 2445 3100

PAGE 73

73 Figure 5 1. Thermal simulation results estimating laser fluences required to melt certain thicknesses of amorphous SiC on c SiC. Melting temperature of amorphous SiC, T m is used as 2445K.

PAGE 74

74 Figure 5 2. Raman spectrum for pristine SiC, PLA pristine SiC, sample #1, sample #2, and sample #3.

PAGE 75

75 Figure 5 3. HR XTEM images of sample #3 after pulse laser annealing with 100 pulses at 800mJ/cm 2 . A) Overall view B ), C ), D ) and E ) are magnified images in the red box regions.

PAGE 76

76 Figure 5 4. HR XTEM images A ) HR XTEM image of pulse laser annealing of amorphous SiC without Au + implant, with 2000 laser pulses at 800mJ/cm 2 .B ) HR XTEM image of pulse laser annealing of amorphous SiC implanted with Au + ions, with 2000 laser pulses at 800mJ/cm 2 . 100 nm thick Cr was sputtered onto sample before FIB and x TEM imaging for protection purpose during the FIB sample preparation process.

PAGE 77

77 Figure 5 5. Raman spectrum for Pristine SiC (blue), pulse laser annealing with 2000 pulses at 800mJ/cm 2 on amorphous SiC without Au + implant (green) and pulse laser annealing with 2000 pulses at 800mJ/cm 2 amorphous SiC with Au + implant (red).

PAGE 78

78 Figure 5 6. Raman spectra of Ge implanted SiC before PLA (black), Ge implanted SiC after PLA with 50 pulses at 100mJ/cm 2 (red), Ge implanted SiC after PLA with 50 pulses at 200mJ/cm 2 (green).

PAGE 79

79 Figure 5 7. 3D image of optical profilometer. A ) 50 pulses, 100mJ/cm 2 , B ) 3D image of optical profilometer, 50 pulses, 200mJ/cm 2 .

PAGE 80

80 Figure 5 8. HR XTEM images of Ge implanted sample. A ) HR XTEM image after PLA with 50 pulses at 100mJ/cm 2 . B ) HR XTEM image after PLA with 50 pulses at 200mJ/cm 2 . Cr layer was sputtered onto samples after the PLA as protection layer during FIB and Pt layer was put onto the sample during FIB for extra protection.

PAGE 81

81 CHAPTER 6 GRAPHENE SYNTHESIS BY PULSED LASER ANNEALING CRYSTALLINE SIC AND OPTIMIZATION 6.1 Introduction In chapter 4, we studied selective graphene (G) growth by pulse laser annealing (PLA) ion implanted SiC (II), and raised questions regarding the graphene growth u implanted SiC with a thin amorphous layer on top. And we also found some graphene nanostructure and explored how to avoid them during PLA. In these chapters, PLA parameters were mentioned but never systematically presented. In this chapter, a detailed pa rameter optimization study and proposed graphitization mechanism will be demonstrated by PLA c SiC at various fluences, repetition rates and environments. SiC solid phase transformation from 4H to 3C was also found during PLA. 6.2 Thermal Simulation and Ex periments. Thermal simulation calculations were done using the same material properties as table 5 1 and were conducted in the same manner mentioned in chapter 6. SiC melting temperature was assumed as 3100K. Commercially available 4H SiC samples from II V I were used for pulsed laser annealing experiments. 100 m x100 m laser spot was used, which allowed multiple laser annealing spots within one sample. In order to optimize the PLA parameters, the laser fluence was varied from 0.1 1.0J/cm2; the number of p ulses was altered, up to 2000 pulses; different repetition rates were used, 50 Hz and 1 Hz for comparison; and the PLA was conducted in varous ambient environments including Argon, Hydrogen and vacuum. Raman spectrum were obtained using the Horiba MicroRam an Spectrometer with a 532 nm green laser, and its laser

PAGE 82

82 spot was 1 micron, capable of measuring within the 100×100 micron laser spots. High Resolution cross sectional Transmission Electron Microscopy (HR XTEM) images were obtained using a JEOL 2010F TEM a fter Raman spectrum, and a Cr sputter coating was used for surface Protection during cross sectioning . 6.3 Results and Discussion. 6.3.1 Thermal S imulation. To study how c SiC response to PLA, we simulated the surface peak temperature of SiC with a single pulse at different laser fluences ranging from 0.1 to 1.0 J/cm 2 . When the pulsed laser irradiates the SiC surface, the photon energy is absorbed, heat the surface, and the surface temperature starts to rise to the peak temperature until the end of lase pul se. Figure 6 1 is the thermal simulation results, showing the peak surface temperature at each fluence. We define the fluence that can just melt the top surface as the melting threshold fluence m , and the m for SiC is 0.905 J/cm 2 based on the simulatio n. Below the m , peak temperature follows the linear relationship with fluence, and the SiC remains in the solid phase although temperatures rise. Above the m , the surface reaches the melting temperature, and c SiC at the top surface changes from solid to liquid. Meanwhile the laser light reflection increases and absorption decreases. Therefore we expect a decrease as indicated by red line in the figure. However, the practical reflectivity while SiC melting and temperature increasing, is not available expe rimentally and was estimated as shown in the red dashed line in Figure 6 1.

PAGE 83

83 6 .3.2 Fluence. The fluences in our PLA experiments ranged from 0.1J/cm 2 to 1.0J/cm 2 to study how the melting process affects SiC structure and graphene growth. Raman spectra and HR XTEM were used to evaluate PLA effects at fluences both above and below m . Figure 6 2 plots the Raman spectra before and after 1J/cm 2 laser annealing with d ifferent number of pulses. Figure 6 2 A is for 1 laser pulse. It shows only a tiny increase of the G, D and 2D peaks, implying an existence of sp 2 carbon bonding. The SiC peak at 1520 cm 1 remaining the same indicates that most of the SiC remains intact aft er laser annealing. When the number of pulses is increased to 10 as in Figure 6 2 B, the SiC peak at 1520 cm 1 is greatly reduced, and the graphene G, D and 2D peaks rise, which means more graphitic layers were formed. 100 pulses, as shown in Figure 6 2 C , t he peak corresponding to SiC is almost unnoticeable compared with the strong G signal. It is reasonable to assume that more laser pulses could anneal the sample for a longer time and yield more graphene layers. However, Raman spectra are not the most defin itive method to indicate the graphene quality and quantity, because of the XTEM images were taken on these 1 to 100 pulse samples and are compared I Figure 6 3. For clarity Figu res 6 3 A , C and E are intentionally aligned so the surface is at the same level. In Figure 7 3 A , 1 laser pulse at 1J/cm 2 melted the c SiC at the near surface, and recrystallization yields a ~20 nm thick polycrystalline 3C SiC on the 4H c SiC substrate, an d a thin amorphous layer and few layers of graphene on top. The beauty of pulse laser annealing is that the sample temperature increases rapidly and returns to room temperature in microseconds after the laser pulse duration. This freezes the

PAGE 84

84 surface in no n equilibrium states and allows these surfaces to be examined. This is extremely helpful when we are studying the graphitization evolution with different PLA conditions. Figure 6 3 B magnifies the surface area in Figure 6 3 A , and clearly 2~3 layers of graph ene has formed on top of the surface, but the quality is not the best and these FLG seems discontinuous, which can explain why the D peak is obvious but the G and 2D is barely noticeable in Figure 6 2 A . Figure 6 3 C is the TEM picture after 10 laser anneali ng pulses at 1J/cm 2 . Apparently, the amorphous layer increases to around 60 nm, and the polycrystalline 3C SiC seems still around 20 nm but much more non uniform. The magnified TEM image in Figure 6 3 D shows that many more layers of graphene formed at the interface of the polycrystalline SiC and amorphous layer. Figure 6 3 E is the TEM image after 100 pulses. The total thickness affected by the laser is ~75 nm, which increased slightly compared with Figure 6 3 C . Interestingly, more and more SiC recrystallize s from the amorphous phase to polycrystalline phase, and it seems the amorphization and the recrystallization are competing with each other. During this process, some of the inclusions shows carbon nano structures like carbon onions in Figure 6 3 F . The tre nd of 1 to 100 pulses laser annealing at a fluence of 1J/cm 2 , which is high enough to melt the surface, tells that there are increasing layers of graphene when applying more pulses, but not in a desirable fashion such as single or multi layer, epitaxial g rowth. Therefore, high fluence rendering surface melting is not the optimum condition for the epitaxial graphene growth. A lower fluence below the melting threshold was also conducted and presented with other effects in the next section. 6.3.3 Frequency E f fect. To create better quality graphene, the lower fluence conditions (<1.0J/cm 2 ) were examined, and it was found that the fluence below 0.8J/cm 2 does not produce graphene

PAGE 85

85 in any way. Interestingly, an unusual effect between the laser repetition rate and 6 3 compares the Raman spectra of SiC after pulsed laser annealing at 0.8J/cm 2 in vacuum, the peak temperature is ~2200 K and is much lower than the c SiC melting tempera ture of 3100K according to the thermal simulation. Figure 6 4 A is the Raman spectrum of PLA at 0.8J/cm 2 , with multiple pulses at 50Hz in vacuum. With the increasing pulses, G and 2D peaks do not change significantly and only the D peak surfaces slightly, b ut it is still considerably smaller compared with pristine SiC. Figure 6 4 B is the Raman spectrum of PLA SiC at 0.8J/cm 2 , with multiple pulses at 1 Hz in vacuum. The areas that were annealed with 125 pulses or more demonstrate prominent graphitic signal. S more graphene growth. To confirm the graphene growth and quality HR XTEM images were taken from these samples. Figure 6 5 compares the TEM images of PLA at 50 Hz and 1Hz with 375 puls es at 0.8J/cm 2 . Figure 6 5 A shows little and discontinuous graphene growth, which is consistent with Raman spectra results for the same conditions. On the contrary, Figure 6 5 B shows 4 layers of graphene epitaxial growth on the C SiC surface, and there is also a thin amorphous layer on top containing unconnected graphene features. The epitaxial graphene on the c SiC substrate has very good quality based on these images, and we believe that the defect peak, D peak in Figure 6 4 B is primarily contributed from the loose graphene features in the amorphous region. growth mechanism in PLA SiC process. The temperature change during PLA increases

PAGE 86

86 rapidly during the 25 ns laser pulse, and then quickly drops back to room temperature in a few s, based on the thermal simulation 114 The only difference between these two conditions is the time interval between two subsequent laser pulses, which is 20 ms for 50 Hz and 1000 ms (1s) for 1Hz. E ven 20 ms is long enough for the surface to be cooled down to the room temperature comparing with a few s, but it seems that a longer wait time at room temperature is helping the graphene growth process. In the physical process of PLA c SiC, silicon atoms are given enough kinetic energy due to the temperature to de bond with carbon atoms, and then can desorb from the SiC surface. After silicon atoms leave the surface, the carbon atoms could bond with each other, and graphene formation happens. If the free silicon atoms do not leave or do not leave quickly enough, silicon atoms could inhibit the graphene growth by re bonding with carbon atoms. In terms of time scales, the de bonding and bonding process in the chemical reaction is in the range of a few nanose conds to microseconds while the surface is still hot, but the mass transportation is in the millisecond or second range when the surface is cool. It seems that the graphene growth during PLA c SiC is not reaction controlled but a diffusion controlled proce ss. Therefore, we believe this surface. A proper control of the silicon sublimation rate and time could be a key to control the quality of graphene synthesis. 6.3.4 Pressure E ffect. PLA in 1 atm of Ar, H 2 or N 2 ambient can significantly lower the graphene yield compared with PLA in vacuum in the 1 to 20 millitorr range . Figure 6 6 compares

PAGE 87

87 Raman spectra for PLA of c SiC in Ar ambient and vacuum. Figure 6 6 A shows results for PLA with100 pulses at 0.8J/cm 2 . The red (Ar) and green (vacuum) spectra have similar G peaks, but slightly different D and 2D peak. The green spectrum (vacuum) has a narrower D peak and higher 2D peak, implying slightly better graphene quality. Annealing with more laser pulses, such as 500 pulses, helps differentiate the two ambient conditions as shown in Figure 6 6 B where the G, D and 2D peaks rise significantly for the sampl e under vacuum conditions, and PLA with 500 pulses in Ar ambient is still very much the same as 100 pulses, and there is no sign of further graphitization. In section 6.3.3 Frequency effect, we discussed that the silicon sublimation rate could affect graph ene growth. The existence of Ar atoms in the environment creates a pressure that could slow or suppress the free silicon atoms from leaving the surface, and therefore lowers the graphene yield. The situation is similar to the growth of graphene by thermal annealing SiC at high temperatures. 33,9 4 an Ar ambient is more productive than annealing in vacuum. 6.3.5 Si C Solid Transformation From 4H T o 3C. During our studies HR XTEM images of pulsed laser annealing of 4H SiC single crystal at 0.8J/cm 2 , at 1Hz wi th different pulses, revealed that a layer of 3C SiC can form on top of 4H SiC, depending on the annealing conditions. Based on the thermal simulations, PLA at 0.8 J/cm 2 can heat the surface temperature up to ~ 2200K, which is much lower than the SiC melti ng temperature ~ 3100K, and the HR XTEM shows no characteristics of melting. Therefore, it is believed that this 4H to 3C SiC transition is solid state transformation instead of melting and recrystallization. Because of the nature of our PLA, the 4H SiC sa mple could be heated and cooled rapidly by a single pulse, repeatedly processed as desired by varying the number and frequency of

PAGE 88

88 pulses, and this 4H to 3C transformation could be preserved and examined at any point by HR XTEM. Figure 6 7 illustrates impo rtant details of the 4H and 3C SiC lattice structure when viewed from different zone axis. Looking from [11 20] for 4H SiC and [1 10] for 3C SiC, there is a distinct difference of lattice structure in the x view, showing the e 4H 3C SiC. However, if we look from the [1 100] for 4H SiC and [ 1 12] for 3C SiC, the x view of the two lattice structures appears to be the same. And this distinct difference was also captured by HR XTEM, show n in Figure 6 8. Two images (a) and (b) from the same 4H c SiC wafer were annealed in vacuum with 300 pulses , 25ns, 1Hz from an ArF laser, at a fluence of 0.8 J/cm 2 well below the PLA fluence at which c or decomposes. A thin crystalline 3C p hase forms on top of the 4H substrate as a result of PLA, and can be seen when viewed along [11 20], but not [1 100]. We also studied how the 3C layer thickness changed with different pulse number by taking HR XTEM images from spots PLA with different numb ers of pulses. Figure 6 9 is the TEM images of spots that were laser annealed with 0, 50, 100, 150, 200, 250, 300, 350 and 375 pulses. We observed that the thickness of 3C SiC formed varies with increasing pulse number. Once 3C SiC formed, the thickness de creases with increasing pulse number. And the thickness increase at 250 pulses and decreases again with increasing pulse number. There seems to be a competition between the thermal l y activated 4H to 3C and 3C to 4H transformation with more laser pulses. In the literature, there are a few possible reasons for the SiC poly type transformation. Phase changes may occur in SiC under applied thermo mechnanical stress. 109 Pirouz et al. 110 have shown polytypic

PAGE 89

89 transformation to occur in single crystal SiC when sub jected to external mechanical loading. Vlaskina and Shin 111 induced 6H SiC to 3C SiC polytypic transformation under vacuum conditions at temperatures between 1800 °C and 2000 °C. Activation energy is 2.34+/ 0.02eV. Some proposed transformation mechanisms for 3C to 6H include layer displacements (layer rotations) of individual close packed double layers of Si and C, that requires the migration of atoms inside the crystal when vacancies are created by the high temperature annealing (Woo et.al.) 112 ; and non random micro twinning (Sebastiak et.al.) . 113 Most of these cases suggest that the poly type change for SiC happens under non equilibrium conditions. Pulsed laser annealing can cause rapid/repeated heating and cooling and our observations show that th is can cause 3C and 4H transformations in what we are confident is a solid state transformation.. A further study is underway to understand the complete mechanism of this phenomenon. 6.4 Conclusion This chapter presented the study of multiple contributing effects during pulsed laser annealing. The graphene yield is better at the lower fluences (0.8J/cm 2 ) with multiple pulses, where the non melting conditions could result in epitaxial graphene related to the mass indicated graphene growth is enhanced under vacuum conditions, which was in concert with the thermal annealing SiC. We also observed a solid pha se transformation of 4H c SiC to 3C c SiC transformation as a result pf PLA with multiple pulses at 0.8J/cm2 in vacuum, which is also epitaxial. However, further study is necessary in the future to explain this process. In summary, pulse laser annealing is a powerful tool to control the

PAGE 90

90 graphene synthesis process by manipulating the processing parameters, and the non equilibrium condition could freeze the surface after the laser pulse, which offers us a great way to study the graphene growth mechanism.

PAGE 91

91 Figure 6 1. Thermal simulation of peak temperature at the end of 25 ns laser pulse at various fluence. Solid black line region is from thermal simulation, and red dash line is from estimation due to lack of constant.

PAGE 92

92 Figure 6 2. Raman spectra comparison before and after pulsed laser annealing with 1 100 pulses at 1J / cm 2 . A ) 1pulse. B ) 10 pulses. C ) 100 pulses.

PAGE 93

93 Figure 6 3 . TEM images comparison among pulsed laser annealing SiC with 1 100 pulses at 1J / cm 2 in Ar. A ) 1pulse. B ) Zoom in image of A ). C ) 10 pulses. D ) Zoom in image of C ) E ) 100 pulses. F ) Zoom in image of E ).

PAGE 94

94 Figure 6 4 . Raman Spectrum for frequency effect. A ) Pulsed laser annealing c SiC in vacuum, at 0.8J/cm 2 , at 50 Hz with 1, 5, 25, 125, 375, 625 and 1000 pulses. B ) Pulsed laser annealing c SiC in vacuum, at 0.8J/cm 2 , at 1 Hz with 1, 5, 25, 125, 375, 625 and 1000 pulse s.

PAGE 95

95 Figure 6 5. TEM images A ) PLA 375 pulses at 0.8J/cm 2 , 50 Hz. B ) PLA 375 pulses at 0.8J/cm 2 , 1 Hz.

PAGE 96

96 Figure 6 6. Raman spectra of PLA c SiC in Ar ambient and vacuum condition.

PAGE 97

97 Figure 6 7 . Illustration of 3C and 4H SiC at different zone axis.

PAGE 98

98 Figure 6 8 . X TEM images of PLA samples at different zone axis.

PAGE 99

99 Figure 6 9. TEM images of PLA c SiC @ 0.8J/cm 2 , showing 3c SiC layer evolution with incremental pulse number.

PAGE 100

100 CHAPTER 7 CONCLUSION In Chapter 3, AlGaN/GaN HEMT devices were lifted off using a 193 nm laser. Laser drilled grooves on the front side of the sample were employed to prevent the formation of micro cracks during the LLO process. Besides a 25 43% reductio n of the saturation drain current, no degradation of other dc characteristics was observed. The LLO HEMT was bent due partial relaxation of the as grown strain as well as strain troscopy methods were used to estimate the partial relaxation of the strain owing to the lattice mismatch between Al 0.25 Ga 0.75 N and GaN. The quantity of partial relaxation for lattice mismatch only accounted for 5 10% of the drain current reduction. The majority of the drain current reduction was attributed to the relaxation of the strain caused by dielectric and metal depositions. In Chapter 4, it has been demonstrated that II combining with PLA could synthesize few layer graphene selectively. We observe d that the II region has much lower onset graphitization fluence than crystalline SiC, which contributes the selective graphene growth. These results also raise questions regarding the graphitization mechanism during PLA of the II regions. Both amorphous s urface layer effects and catalytic or doping effects seem to contribute to the G growth, however a detailed and systematic study is required to address these questions. Moreover the laser processing parameters need to be optimized to gain better control on both the quantity and quality of the graphene layers. In Chapter 5, it has been studied that the surface temperature of pulse laser

PAGE 101

101 aser annealing of pure amorphous silicon carbide and Au + implanted silicon carbide separately; we also proposed possible mechanisms. Graphene nano structures were synthesized, characterized by Raman spectra, and imaged using HR XTEM, including carbon onion s, few layer graphene, and graphene. In Chapter 6, I presented the study of multiple contributing effects during pulsed laser annealing. The graphene yield is better at the lower fluences (0.8J/cm 2 ) with multiple pulses, where the non melting conditions co uld result in epitaxial graphene indicated graphene growth is enhanced under vac uum conditions, which was in concert with the thermal annealing SiC. We also observed a solid phase transformation of 4H c SiC to 3C c SiC transformation as a result pf PLA with multiple pulses at 0.8J/cm2 in vacuum, which is also epitaxial. However, furth er study is necessary in the future to explain this process. In summary, pulse laser annealing is a powerful tool to control the graphene synthesis process by manipulating the processing parameters, and the non equilibrium condition could freeze the surfac e after the laser pulse, which offers us a great way to study the graphene growth mechanism.

PAGE 102

102 LIST OF REFERENCES 1. U. Sowada, H.J. Kahler, D. Basting, World Lasers Almanac 1, 50 (1988) 2. F.K. Kneubühl, M.W. Sigrist, Laser (Teubner Taschenbücher Physik, Stuttgart, 1995) 3. K.J. Schmatjko, G. Endres, World Lasers Almanac 1, 46 (1988) 4. BäuerleW.Dieter. Laser Processing and Chemistry. (2011). 5. Stephen Pearton. GaN and Zn O based Mateirals and Devices. (2011) 6. W.S. Wong, T. Sands and N.W. Cheung, "Damage free Separ ation of GaN Thin Films from Sapphire Substrates," Appl. Phys. Lett. 72 (1998) pp. 599 601. 7. Study of GaN light emitting diodes fabricated by laser lift off technique 8. Chu, Chen Fu and Lai, Fang I and Chu, Jung Tang and Yu, Chang Chin and Lin, Chia Feng and Kuo, Hao Chung and Wang, S. C., Journal of Applied Physics, 95, 3916 3922 (2004) 9. XJ Su, KXu, YXu, GQRen, J C Zhang, JFWang and H Yang, J. Phys. D: Appl. Phys.46(2013) 205103 (6pp) 10. Xu et al. Materials Letters 56 (2002) 43 46 11. Brodie, B. C. (1859). "On the At omic Weight of Graphite". Philosophical Transactions of the Royal Society of London 149: 249 259. Bibcode: 1859RSPT..149..249B 12. Ruess, G.; Vogt, F. (1948). "Höchstlamellarer Kohlenstoff aus Graphitoxyhydroxyd". Monatshefte für Chemie (in German) 78 (3 4): 2 22 242. 13. Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV, Firsov AA. 2004 Electric field effect in atomically thin carbon films. Science 306, 666 669. 14. Mayorov, A. S.et al.Micrometer scale ballistic transport in encapsulated gr aphene at room temperature.Nano Lett.11,2396 2399 (2011). 15. Lee, C., Wei, X. D., Kysar, J. W. & Hone, J. Measurement of the elastic properties and intrinsic strength of monolayer graphene.Science321,385 388 (2008). 16. Balandin, A. A. Thermal properties of graph ene and nanostructured carbon materials.Nature Mater.10,569 581 (2011). 17. Nair, R. R.et al.Fine structure constant defines visual transparency of graphene. Science320,1308 (2008).

PAGE 103

103 18. : evidence for graphane.Science323,610 613 (2009). 19. Loh, K. P., Bao, Q. L., Ang, P. K. & Yang, J. X. The chemistry of graphene.J. Mater. Chem.20,2277 2289 (2010). 20. Nair, R. R.et al.Fluorographene: a two dimensional counterpart of Teflon.Small6, 2877 2884 (2010) 21. Aufray, Bernard and Kara, Abdelkader and Vizzini, Sébastien and Oughaddou, Hamid and Léandri, Christel and Ealet, Benedicte and Le Lay, Guy, Applied Physics Letters, 96, 183102 (2010) 22. P Bampoulis, L Zhang, A Safaei, R van Gastel, B Poelsema and H J W Zandvliet, J. Phys.: Condens. Matter 26 (2014) 442001 (5pp) 23. Pacile, D.; Meyer, J. C.; Girit, C. O.; Zettl, A. The two dimensional phase of boron nitride: few atomic layer sheets and suspended membranes. Appl. Phys. Lett. 2008, 92, 133107 1. 24. Radisavljevic , B., Radenovic, A., Brivio, J., Giacometti, V. & Kis, A. Single layer MoS2 transistors. Nature Nanotech. 6, 147 150 (2011). 25. Schwierz, F. Nanoelectronics: Flat transistors get off the ground. Nature Nanotech. 6, 135 136 (2011). 26. Radisavljevic, B., Radenovic , A., Brivio, J., Giacometti, V. & Kis, A. Single layer MoS2 transistors. Nature Nanotech. 6, 147 150 (2011). 27. Kusmartsev, F. V.; Wu, W. M.; Pierpoint, M. P.; Yung, K. C. (2014). "Application of Graphene within Optoelectronic Devices and Transistors". arXiv :1406.0809 28. Amini, Shaahin; Garay, Javier; Liu, Guanxiong; Balandin, Alexander A.; Abbaschian, Reza (2010). "Growth of Large Area Graphene Films from Metal Carbon Melts". Journal of Applied Physics 108 (9): 094321 29. Lenski, Daniel R.; Fuhrer, Michael S. (2011 ). "Raman and optical characterization of multilayer turbostratic graphene grown via chemical vapor deposition". Journal of Applied Physics 110: 013720. 30. Peter W. Sutter, Jan Ingo Flege & Eli A. Sutter, Epitaxial graphene on ruthenium, Nature Materials 7, 4 06 411 (2008) 31. F. J. Derbyshire, A. E. B. Presland and D. L. Trimm,Carbon, 1975, 13, 111. 32. Berger, C. et al. Ultrathin epitaxial graphite: 2D electron gas properties and a route toward graphene based nanoelectronics. J. Phys. Chem. B 108, 1991219916 (2004) .

PAGE 104

104 33. Berger, C. et al. Electronic confinement and coherence in patterned epitaxial graphene. Science 312, 11911196 (2006). 34. Emtsev et al, Towards wafer size graphene layers by atmospheric pressure graphitization of silicon carbide, Nature Materials 8, 203 20 7 (2009) Published online: 8 February 2009 35. Sukang Bae et al, Roll to roll production of 30 inch graphene films for transparent electrodes, Nature Nanotechnology 5, 574 578 (2010) doi:10.1038/nnano.2010.132 36. Dmitry et al. Longitudinal unzipping of carbon nan otubes to form graphene nanoribbons, Nature 458, 872 876 (16 April 2009) | doi:10.1038/nature07872 37. Jiao et al. Narrow graphene nanoribbons from carbon nanotubes, Nature 458, 877 880 (16 April 2009) | doi:10.1038/nature07919 38. A. Krost and A. Dadgar, Mater. S ci. Eng. B 93, 77 (2002). 39. M. Umeno, T. Egawa, and H. Ishikawa, Mater. Sci. Semicond. Process. 4, 459 (2001). 40. J. C. Carrano, T. Li, P. A. Grudowski, C. J. Eiting, R. D. Dupuis, and J. C. Campbell, J. Appl. Phys. 83, 6148 (2000). 41. T. D. Moustakas, E. Iliopoulos, A. V. Sampath, H. M. Ng, D. Doppalapudi, M. Misra, D. Korakakis, and R. Singh, J. Cryst. Growth 227, 13 (2001). 42. The Semiconductor Industry Association (SIA), International Technology Roadmap for Semiconductors, 1999 edition. 43. Y. F. Wu, D. Kapoln ek, J. P. Ibbetson, P. Parikh, B. P. Keller, and U. K. Mishra, IEEE T. Electron Dev. 48, 586 (2001). 44. Y. F. Wu, P. B. Keller, S. Keller, D. Kapolnek, S. P. DenBaars, and U. K. Mishra, IEEE Electron Dev. L. 19, 455 (1996). 45. P. Konminous, T. Kehagias, J. Kiose oglou, G. P. Dimitrakopulos, A. Sampath, T. D. Moustakas, G. Nouet, and T. Karakostas, J. Phys.: Condens. Matter 14, 13277 (2002). 46. W. S. Wong, T. Sands, N. W. Cheung, M. Kneissl, D. P. Bour, P. Mei, L. T. Romano, and N. M. Johnson, Appl. Phys. Lett. 75, 13 60 (1999). 47. E. A. Stach, M. Kelsch, E. C. Nelson, W. S. Wong, T. Sands, and N. W. Cheung, Appl. Phys. Lett. 77, 1819 (2000). 48. J. Xu, R. Zhang, Y. P. Wang, X. Q. Xiu, B. Shen, S. L. Gu, Y. Shi, Z. G. Liu and Y. D. Zheng, Materials Letters, 56, 43 (2002).

PAGE 105

105 49. J. Christenson, Y. C. Shen, C. Lowery, P. S. Martin, S. Subramanya, W. Gotz, N. F. Gardner, R. S. Kern, and S. A. Stockman, Appl. Phys. Lett. 78, 3379 (2001). 50. T. Fujii, Y. Gao, R. Sh arma, E. L. Hu, S. P. DenBaars, and S. Nakamura, Appl. Phys. Lett. 84, 855 (2004). 51. K. Bao, X. N. Kang, B. Zhang, T. Dai, Y. J. Sun, Q. Fu, G. J. Lian, G. C. Xiong, G. Y. Zhang, and Y. Chen, Appl. Phys. Lett. 92, 141104 (2008). 52. T. S. Kang, C. F. Lo, X.T. Wa ng, L. Liu, R. Finch, E. Douglas, S. J. Pearton, S. T. Hung, C. J. Chang, and F. Ren, J. Vac. Sci. Technol. B 29, 041202 1(2011). 53. J. Das, W. Ruythooren, R. Vandersmissen, J. Derluyn, M. Germain, and G. Borghs Phys. Stat. Sol. (c) 2, 2655 (2005). 54. H. Ji, J. Das, M. Germain and M. Kuball. Solid State Electronics 53, 526 (2009). 55. C. F. Chu, F. I. Lai, J. T. Chu, C. C. Yu, C. F. Lin, H. C. Kuo, and S. C.Wang: J. Appl. Phys. 95, 3916 (2004). 56. T. Wang, Y. Fang, X. Guo, G. Shen, Z. Cui, Thin Solid Films, 515, 3854 (2 007). 57. J. E. Northrup and J. Neugebauer, Phys. Rev. 53, R10477 (1996). 58. H. W. Choi, M. A. Rana, S. J. Chua, T. Osipowicz and J. S. Pan, Semicond. Sci. Technol. 17, 1223 (2002). 59. K. T. Lee, Y. C. Lee, S. H. Tu, C. L. Lin, P. H. Chen, C. Y. Liu, and J. Y. Chang , Jpn J. Appl. Phys. 47, 930 (2008). 60. Isaac Newton, Opticks, Dover, New York, 1070, p.1704. 61. T. Kozawa , T. Kachi , H. Kano , H. Nagase , N. Koide , and K. Manabe , J. Appl. Phys. 77, 4389 (1995). 62. S. N. G. Chu, F. Ren, S. J. Pearton, B. S. Kang, S. Kim, B. P. Gila, C. R. Abernathy, J. I. Chyi, W. J. Johnson, and J. Lin, Mater. Sci. Eng. A, 409, 340 (2005). 63. C. H. Tsai, S. R. Jian, J. Y. Juang, Appl. Surf. Sci. 254, 1997 (2008). 64. M. A. Moram, Z. H. Barber and C. J. Humphreys, J. Appl. Phys, 102, 023505 1 (2007). 65. B.S.Kang, S.Kim, F.Ren, K.Baik, S,J,Pearton, B.P.Gila, C.R.Abernathy, C. C.Pan, G.T.Chen, J. I.Chyi, V.Chandrasekaran, M.Sheplak, G. T.Chen, and S.N.G.Chu, Appl.Phys.Lett., 83(23), 4845(2003). 66. Kang T S, Wang X T, Lo C F 2012 J. Vac. Sci. Technol. B 30 011203

PAGE 106

106 67. GaN based materials and devices : growth, fabricatio n, characterization and performance, Michael Shur; Robert F Davis, World Scientific, 2004, page 73 68. S. Tongay, M. Lemaitre, J. Fridmann, A. F. Hebard, B. P. Gila, and B. R. Appleton, Appl. Phys. Lett. 100, 073501 (2012). 69. M. G. Lemaitre, S. Tongay, X. Wang, D. K. Venkatachalam, J. Fridmann, B. P. Gila, A. F. Hebard, F. Ren, R. G. Elliman, and B. R. Appleton, Appl. Phys. Lett.100, 193105 (2012). 70. B. Appleton and B. Gila, U.S. Patent No. pct/us2010/024723 (n.d.). 71. B. Appleton, B. Gila, M. Lemaitre, and S. Tongay, U.S. Patent No. 61/526,510 (n.d.). 72. B.R. Appleton, S. Tongay, M. Lemaitre, B. Gila, J. Fridmann, P. Mazarov, J.E. Sanabia, S. Bauerdick, L. Bruchhaus, R. Mimura, and R. Jede, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 272, 153 (2012). 73. K.S. Novoselov, D. Jiang, F. Schedin, T.J. Booth, V.V. Khotkevich, S.V. Morozov, and A.K. Geim, Pnas 102, 10451 (2005). 74. Y. Zhang, Y. W. Tan, H.L. Stormer, and P. Kim, Nature 438, 201 (2005). 75. K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, M.I. Katsnelson, I.V. Grigorieva, S.V. Dubonos, and A.A. Firsov, Nature 438, 197 (2005). 76. A.K. Geim, Science 324, 1530 (2009). 77. A.H. Castro Neto, F. Guinea, N.M.R. Peres, K.S. Novoselov, and A.K. Geim, Rev. Mod. Phys. 81, 109 (2009). 78. L.A. Falkovsky, Journal of Physics: Conference Series 129, 012004 (2008). 79. J.C. Meyer, A.K. Geim, M.I. Katsnelson, K.S. Novoselov, T.J. Booth, and S. Roth, Nature 446, 60 (2007). 80. J.S. Bunch, A.M. Van Der Zande, S.S. Verbridge, I.W. Frank, D.M. Tanen baum, J.M. Parpia, H.G. Craighead, and P.L. McEuen, Science 315, 490 (2007). 81. A. Reina, X. Jia, J. Ho, D. Nezich, H. Son, V. Bulovic, M.S. Dresselhaus, and J. Kong, Nano Lett. 9, 30 (2008). 82. X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner, A. Velamak anni, I. Jung, E. Tutuc, S.K. Banerjee, L. Colombo, and R.S. Ruoff, Science 324, 1312 (2009). 83. S. Lee, K. Lee, and Z. Zhong, Nano Lett. 10, 4702 (2010).

PAGE 107

107 84. Y. M. Lin, C. Dimitrakopoulos, K.A. Jenkins, D.B. Farmer, H. Y. Chiu, A. Grill, and P. Avouris, Science 327, 662 (2010). 85. J. Robinson, X. Weng, K. Trumbull, R. Cavalero, M. Wetherington, E. Frantz, M. LaBella, Z. Hughes, M. Fanton, and D. Snyder, ACS Nano 4, 153 (2009). 86. M.Y. Han, B. Özyilmaz, Y. Zhang, and P. Kim, Phys. Rev. Lett. 98, 206805 (2007). 87. B. Özyilm az, P. Jarillo Herrero, D. Efetov, and P. Kim, Applied Physics Letters 91, 192107 (2007). 88. Z.H. Ni, T. Yu, Y.H. Lu, Y.Y. Wang, Y.P. Feng, and Z.X. Shen, ACS Nano 2, 2301 (2008). 89. T.O. Wehling, K.S. Novoselov, S.V. Morozov, E.E. Vdovin, M.I. Katsnelson, A.K. Geim, and A.I. Lichtenstein, Nano Lett. 8, 173 (2007). 90. X. Wang, X. Li, L. Zhang, Y. Yoon, P.K. Weber, H. Wang, J. Guo, and H. Dai, Science 324, 768 (2009). 91. S. Tongay, K. Berke, M. Lemaitre, Z. Nasrollahi, D.B. Tanner, A.F. Hebard, and B.R. Appleton, Nanote chnology 22, 425701 (2011). 92. Y. Zhang, T. T. Tang, C. Girit, Z. Hao, M.C. Martin, A. Zettl, M.F. Crommie, Y.R. Shen, and F. Wang, Nature 459, 820 (2009). 93. D. Perrone, G. Maccioni, A. Chiolerio, C. Martinez de Marigorta, M. Naretto, P. Pandolfi, P. Martino, C . Ricciardi, A. Chiodoni, E. Celasco, L. Scaltrito, and S. Ferrero, Proceedings of the Fifth International WLT Conference on Laser Manufacturing (2009). 94. S. Lee, M.F. Toney, W. Ko, J.C. Randel, H.J. Jung, K. Munakata, J. Lu, T.H. Geballe, M.R. Beasley, R. S inclair, H.C. Manoharan, and A. Salleo, ACS Nano 4, 7524 (2010). 95. Saito, R.; Hofmann, M.; Dresselhaus, G.; Jorio, A.; Dresselhaus, M. S. Adv. Phys. 2011, 30, 413 550. 96. Inhwa Jung, Jong Soo Rhyee, Jong Yeog Son, Rodney S Ruoff and Kyong Yop Rhee, Nanotechnolo gy 23 (2012) 025 708 (8pp) 97. P. Baeri, C. Spinella, and R. Reitano, International Journal of Thermophysics 20, 1211 (1999). 98. A. Hedler, S. Urban, F. Falk, H. hobert, W. Wesch, Applied Surface Science 205 (2003) 240 248. 99. C. Dutto, E. Fogarassy, and D. Mathiot, Appl. Surf. Sci. 184, 362 (2001).

PAGE 108

108 100. S. G. Muller, R. Eckstein, J. Fricke, D. Hofmann, R. Hofmann, R. Horn, H. Mehling, and O. Nilsson, Mater. Sci. Forum 264, 623 (1998). 101. Sumio Iijima Journal of Crystal Growth 50 (1980) 675 683. 102. Ugarte, D. Chemical Physics L etters ( 1993) , 207, 473 479. 103. N. Sano , H. Wang, I. Alexandrou, M. Chhowalla, K. Teo, G. Amaratunga, K. Iimura, Journal of Applied Physics 2002, 92, 2783 2788. 104. 4792. 105. Otakar Frank et al. ACS Nano, 2011, 5 (3), pp 2231 2239 106. Tuba Oznuluer, Ercag Pince, Emre O. Polat, Osman Balci, Omer Salihoglu and Coskun Kocabas, Appl. Phys. Lett. 98, 183101 (2011). 107. A. L. Pinardi, S. J. Leake, Physical Review B 79, 045416 (2009) 108. B. R. Appleton, O.W. Holland, D.B. Poker, J. Naryan, and D. Fathy, Nuclear Instruments and Methods in Physics Research B7/8 (1985) 639 644 109. J.W. Yang and P. Pirouz, J. Mat. Res., 8 pp. 2902 (1993). 110. P.Pirouz and J.W. Yang, Ultramicroscopy 51, pp. 189 (1993). 111. Svitlana I. Vla skina and Dong Hyuk Shin, Jpn. J. Appl. Phys. 38, Part. 2, No. 1A/B, L27 L29 (1999). 112. Woo SikYoo et.al. Japanese Journal of Applied Physics, Vol.30, No. 3, 545 593, 1991 113. M.T. Sebastiak Phy. Status Solidi , Vol.84, Issue 2 (1984) 114. Z. Tian, N.R. Quick, A. Kar, Acta Mater. 54 (2006) 4273 4283.

PAGE 109

109 BIOGRAPHICAL SKETCH Xiaotie Wang was born in Baotou, China, in 1986. At the age of six , he moved to Tianjin , where he lived until 2005 , when he graduated from Xin Hua High School. He he Department of Materials Science and Engineering at the South China University of Technology , Guangzhou, in 2009. He attended the University of Florida, pursuing master degree in the Department of Materials Science and Engineering, and graduated in 201 1. enrollment in the Department of Material Science and Engineering at the University of Florida under the guidance of Dr. Brent Gila. He studied laser processing wide band gap s emiconductor materials, including laser lift off GaN thin film and GaN high electron mobility transistor (HMET), and the graphene synthesis and optimization by pulse laser annealing a SiC and c SiC.