β-ga 2 O 3 Material Properties of Beta-Gallium Oxide

Transcription

1 β-ga 2 O 3 Material Properties of Beta-Gallium Oxide James S. Speck 1 and Steven A. Ringel 2 1 University of California Santa Barbara 2 The Ohio State University

2 Work Plan Year 1 UCSB Task 1 UCSB Task 2 OSU Task 1 MBE Growth of β-ga 2 O 3 Unintentionally-doped (UID) Films Sn and Si Doping Studies Baseline Schottky Diodes and Initial Characterization OSU Task 2 Baseline Trap Spectrum for UID β-ga 2 O 3 Year 2 UCSB Task 3 UCSB Task 4 OSU Task 3 OSU Task 4 Year 3 UCSB Task 5 UCSB Task 6 OSU Task 5 Sn and Si Doping Studies (completion) Mg Doping Studies Bandgap States in n-type β-ga 2 O 3 / Completion of UID Studies Transport Measurements Mg Doping Studies (completion) Development of (Al,Ga) 2 O 3 growth Transport Measurements OSU Task 6 Bandgap States in Mg-doped β-ga 2 O 3 OSU Task 7 Evaluation of Traps in β-(al,ga) 2 O 3 and Heterostructures

3 Outline 1. New electronic material β-ga 2 O 3 2. Growth studies Homoepitaxy Heterostructures 3. Process and transport 4. Defect spectroscopy 5. Conclusion and ongoing work

4 Wide Bandgap Semiconducting Oxides Material Crystal structure Gap (ev) Features Substrate Issues/ Opportunities Growth technique Applications ZnO Wurtzite 3.40 polar Sapphire, GaN, ZnO dominant donor, p-doping,surface accumulation MOCVD TCO, sensors, light emitters In 2 O 3 Bixbyite 2.70 conductivity YSZ CeO 2 dominant donor, mobility, surface accumulation MBE TCO, sensors SnO 2 Rutile 3.65 conductivity TiO 2 dominant donor, p-doping,surface accumulation MBE TCO, sensors, light emitters Ga 2 O 3 Monoclinic Rhombohedr conductivity Sapphire bulk crystals; dominant donor; UV transparent MBE Power TiO 2 Rutile Anatase 3.10 high-k, conductivity Sapphire, TiO 2 dielectric; n-type conductivity Hybrid MBE CMOS; tunable capacitors; photocatalysis SrTiO 3 Perovskite 3.25 high-k, piezoelectric SrTiO 3 dielectric; n-type conductivity, piezoelectric Hybrid MBE CMOS; tunable capacitors

5 β-ga 2 O 3 Crystal structure -gallia structure a = 12.2 Å b = 3.0 Å c = 5.8 Å = 90.0 o =103.7 o = 90.0 o (010) (001) (100) Ga 2 O 3 bulk E.G. Villora et al, JCG 270, 420 (2004) (Edge-defined film-fed growth (EFG) method) - Pulling speed: 10 mm/h - Size: 70 mm x 50 mm x 3 mm H. Aida et al, JJAP 47, 8506 (2008)

6 β-ga 2 O 3 Physical Properties Direct, dipole disallowed bandgap 4.9 ev gap Naturally n-type Cleavage on (100) and (001) Poor growth on cleavage planes Excellent growth on noncleavage planes ARPES studies on β-ga 2 O 3 ~4.9 ev gap Valence bands n-type doping Sn, Si, Ge (Sb in development) Deep acceptor doping Mg Alloys and heterostructures possible Group III site β-(al,in,ga) 2 O 3 Mohamed et al., APL 97, (2010)

7 Growth Studies

8 Materials Growth and Characterization UCSB Substrates - Homoepitaxy (010) β-ga 2 O 3 from Tamura/Koha Dedicated Oxide MBE at UCSB Plasma-assisted MBE Growth Activated oxygen from O 2 plasma (1-2% atomic O) Sn, Ga, In, Sb, Mg effusion cells Base pressure ~10-10 Torr Oxygen plasma cleaning prior to growth Used extensively for SnO 2 and In 2 O 3 Epi 620 Materials Characterization Extensive suite of characterization tools AFM, HRXRD, SIMS, XPS, TEM, atom probe T-dependence Hall, I-V, CV,

9 Substrate Growth 620 (RF Plasma MBE) Oxygen plasma: - O BEP: ~1 x 10-5 Torr - Plasma power: 200 W - Active O BEP: ~1 x 10-7 Torr Experimental Details - -Ga 2 O 3 (010) (Tamura Corp.) - 5 x 5 mm 2 - XRD (020) -rocking scan : 83 arcsec deg miscut - Sn-doped (conductive) and Fe-doped (insulating) RMS 0.1 nm Sample cleaning: - Acetone and Isopropanol - In situ O plasma cleaning (30 min, 850 C) Virgin -Ga 2 O 3 (010) substrate h rms : 0.16 nm ω (deg) 200 nm Thickness fringes from rigid body displacement

10 Growth Parameter Refinement: Ga Flux Homoepitaxial Growth ~100 nm UID β-ga 2 O 3 T sub : 700 C Ф Ga - Ф O* = -0.2 nm/min Ф Ga - Ф O* = +0.2 nm/min h rms : 1.16 nm h rms : 1.52 nm O-rich growth results in favorable surface morphology Excess Ga: rms roughness, step-bunching, clusters at step edges (Clusters not removed by HCl not Ga droplets)

11 Growth Parameter Refinement: Substrate Temp and Growth Rate Homoepitaxial Growth ~100 nm UID β-ga 2 O 3 Ф Ga - Ф O* = -0.2 nm/min T sub : 600 C h rms : nm T sub : 650 C h rms :0.529 nm Increasing Substrate Temperature Enhanced Step-bunching / Surface Roughness Optimal Temp Range: C T sub : 700 C h rms :1.468 nm Peak of 3.2 nm/min (~200 nm/hr) at 60 Torr O 2 foreline 45% improvement over previous PAMBE growth

12 β-(al x Ga 1-x ) 2 O 3 Alloys: Solubility Equilibrium Al 2 O 3 -Ga 2 O 3 Phase Diagram Al 2 O 3 β-ga 2 O 3 Corundum β-gallia Solubility of Al 2 O 3 in β-ga 2 O 3 drastically reduced below 800 C by formation of AlGaO 3 intermediate compound In the typical β-ga 2 O 3 homoepitaxy temperature range solubility 20-30% Al 2 O 3 V. G. Hill et al, J Am Cer Soc 35, 135 (1952)

13 β-(al x Ga 1-x ) 2 O 3 (010) Grown by PAMBE at 600 C O-rich growth - ~60 nm layers - ~3 nm/min x = ~0.18 Grown beyond the phase stability limit at 600 C -Islanded surface -Weak layer peak -No fringes Phase stability limit of ~15% Al 2 O 3 at 600 C (Composition confirmed by EDS)

14 β-(al x Ga 1-x ) 2 O 3 (010) O-rich growth - ~60 nm layers - ~3 nm/min Distinct layer peak with x = ~0.18 at 650 C Increased phase stability limit at 650 C since layer peak with x = ~0.18 at 600 C was indistinguishable Peak roughness at phase stability limit Increasing phase stability limit with increasing temp Phase stability limit of ~18% Al 2 O 3 at 650 C Phase stability limit of ~20% Al 2 O 3 at 700 C

15 Process and Transport Studies

16 ICP- Superior etching technique BCl 3 20 SCCM ~15 mt chamber pressure 200 nm etch depth (all BCl 3 ) ICP yields much higher etch rates and smoother surfaces than RIE Circular TLM measurements on etched Sn-doped β-ga 2 O 3 substrates in progress to correlate roughness to contact resistance

17 Mobility (cm 2 /Vs) Preliminary Sn-doping Study O-rich growth - ~200 nm layers - ~3 nm/min 1 x 1 μm 2 AFM h rms : 0.40 nm Sn-doping inhibits step-bunching SnO 2 surface segregation for h >250 nm Electron concentrations spanning cm -3 Mobility of ~120 cm 2 /Vs (current record) ~1.5x GaN BFOM, similar HFOM

18 Contact and Defect Spectroscopy Studies

19 Current Density (A/cm -2 ) Capacitance (pf) Capacitance(pF) Initial Ni/β-Ga 2 O 3 (010) Schottky Diode Screening 200 nm 8 nm Ni Ti/Au UID β-ga 2 O 3 (010) Very high quality devices K Ideal C-V Diode size: 300 x 300 µm K n ~ 1.04 Leakage below detection limit Voltage (V) Voltage (V) 300 K No observable capacitance dispersion 0 1k 10k 100k 1M Frequency (Hz)

20 n-type background doping: UID β-ga 2 O 3 (010) Although unintentionally doped, n-type conductivity via C-V and Hall - Hall measured electron density of cm -3, with a mobility of 20 cm 2 /V-s - C-V revealed n-type doping density of cm -3 (assuming ε Ga2O3 =10 * ) n-type n detected by Hall 300 K N D = cm K 1 MHz K. Irmscher, et al., J. Appl. Phys. 110, (2011)

21 Photoyield 1/2 (a.u.) Photocurrent (pa) Barrier Height (ev) Ni/β-Ga 2 O 3 (010) Schottky barrier height SBH determined by internal photoemission (IPE) and confirmed by C-V Ni/Ga 2 O 3 SBH is 1.55 ev at 300 K with small T dependence K Bias = 0 V Energy (ev) 1.55 ev Energy (ev) Theoretical prediction of ideal SBH: Temperature (K) Work function for Ni ~ 5.1 ev Electron affinity for Ga 2 O 3 ~ 3.5 ev [1] Calculated ideal SBH ~ 1.55 ev Evidence that Ni/(010)Ga 2 O 3 SBH may be unpinned! - Plan to test different metals and orientations - Note Ni/(100) Ga 2 O 3 SBH = 1.1 ev [2] [1]. I. Lopez et al, Phys. Status Solidi A 209, No. 1, (2012) [2]. K. Irmscher, et al., J. Appl. Phys. 110, (2011)

22 dc/c dc/c ln( T 2 ) DLTS results to explore traps within ~ 1 ev of E c RW 10 s -1 V R = -0.5 V V F = 0 V DLTS spectrum E C ev E C ev Arrhenius plot E C ev Temperature (K) E C ev Temperature (K) E C ev N T (cm -3 ) This work Literature* E C ev low ~ mid E C ev mid E C ev low ~ mid /kT (ev -1 ) E C ev Red = (010) Ga 2 O 3 (this work) Black = (100) Ga 2 O 3 * Same DLTS traps appear in both crystal orientations for different sources Trap concentration for E C ev trap is similar; possible growth sensitivity for E C ev and E C ev * K. Irmscher, et al., J. Appl. Phys. 110, (2011)

23 log ( ) dc/c Optical Cross Section Spectrum from Photocapacitance transient analysis provides energy levels o n 1 E c ev DLOS exploring traps in 4.8 ev bandgap: energy level determination dc dt h t 0 E c ev E G = 4.8 ev Steady State Photocapacitance provides concentrations RT V R = -0.5 V V F = 0 V E c ev E G ~ 4.8 ev E c ev Energy (ev) Fitting with Lucovsky model [1] E ( h E ) ( h ) 1/2 3/2 o i i n ( h ) 3 [1] G. Lucovsky, Solid State Commun. 3, p299 (1965) Energy (ev) Energy level N T (cm -3 ) E C ev* E C ev *broadness indicates significant lattice relaxation, under study now 2 [2] A. Armstrong et al, J. Appl. Phys. 98, (2005).

24 Summary Research highlights: Optimized (010) β-ga 2 O 3 homoepitaxial growth Developed β-(al x Ga 1-x ) 2 O 3 /Ga 2 O 3 heterostructures Establish Al solubility limit Developed RIE and ICP etch technology of (010) β-ga 2 O 3 High electron mobility in β-ga 2 O 3 High quality Ni/Ga 2 O 3 Schottky diodes SBH appears to be unpinned at 1.5~1.6 ev over wide range of temperatures DLTS and DLOS trap studies Next steps: characterize MBE grown epilayers as function of doping (Si, Sn, etc), growth conditions Explore AlGaO/GaO heterostructures Measure SBH with different metal/orientation Variable temperature Hall, C-T to determine activation energy for background donor E C E C ev E C ev E C ev E C ev E C ev E V N T ~ 1x10 16 cm -3

25 Background Slides

26 Advanced Defect Characterization OSU Deep level transient/optical spectroscopy (DLTS/DLOS) Probe entire E G to ~ 5.5 ev <0.02 ev energy resolution Sensitive to N T >~ 10-5 N D E C DLTS E C to E C - 1eV DLTS/DLOS:C,G,Y,I.. DLOS E C 1 ev to E V E V Semi-transparent Schottky Nano-DLTS/DLOS/SPM n-gan, uid(n)-ingan, or n-algan n+gan:si epi GaN template Ohmic

27 DLTS Signal (pf) DLTS and DLOS Trap modulation: Quiescent Condition V 0 E C E T E V Electrical pulse to fill traps V 0 E C E T E V h E g Trap emission (Thermal/Optical) V 0 Capacitance transient E C E T E V DLTS: thermal emission Arrhenius plot & DLTS spectra Full coverage of E G DLOS: optical emission Steady State Photocapacitance ln(t 2 /e n ) /kT (ev -1 ) DLTS E F ~1 ev ~2.4 ev E c DLOS 2N D C/C (cm -3 ) 8x x x x10 16 N T2 N T1 E G Temperature (K) slope, peak height E T, N T E v 0 onset, step height Incident Energy (ev) E T, N T

28 Growth Rate of -Ga 2 O 3 by MBE -Ga 2 O 3 (100) O plasma: 200 W Growth T: 800 o C O 2 : 1 sccm Ga: 2x10-7 Torr (100) & (001) : Cleavage plane Small growth rate K. Sasaki et al, J. Cryst. Growth 378, 591 (2013) Ga 2 O 3 + 4Ga 3Ga 2 O Narrow growth window M. Tsai et al, J. Vac. Sci. Tech. A28, 354 (2010) Goal: Growth diagram of -Ga 2 O 3 (010)

29 Why DLOS DLTS can t see mid-gap states! 300K time constant E F ~1 ev DLTS E c E C E T1 = E C -0.6 ev ~30 ms ~1.4 ev?? E T2 = E C -1.3 ev ~63 yrs ~1 ev DLTS E v >3.4 ev E V Wide bandgap reveals limitations for trap detection approaches based on thermal carrier emission (e.g. DLTS) e n n v n N C Et E exp kbt C Exponentially worsens with increasing Eg even worse for AlGaN!

30 Energy (ev) Defining Energy States in III-Nitrides InN In x Ga 1-x N 0.60 ev 0.65 ev 2.12 ev 0.39 ev 0.91 ev 1.45 ev 1.76 ev 2.6 ev 2.5 ev GaN Vacuum level Al x Ga 1-x N In % N T ~ cm -3 Al % N N T >~ cm -3 T >~ cm -3 AlN E g (In x Ga 1-x N)= 3.42(1-x)+0.77x-1.43x(1-x) N-Vacancy E vac -E v = x complex Point defect E g (Al x Ga 1-x N)= x-x(1-x) 0.25 ev 0.87eV cluster E TDD related vac -E v = x(1-x) 0.9 ev 1.5 ev point defect 1.28 ev Carbon References: 2.4 ev Fang et al., APL 82, 1562 (2003). interstitial 1.63 ev 2.5 ev Hierro et al., APL 80, 805 (2002). Grp III vacancy 2.6 ev Hierro et al., PSSB 228, 309 (2001). Armstrong et al., APL 84, 374 (2004). 3.11eV Mg acceptor Hierro et al., APL 76, 3064 (2000) ev Hierro et al., APL 77, 1499 (2000) ev 3.93 ev Armstrong et al., APL 84, 374 (2004). Carbon K. B. Nam et al., APL 86, (2005). acceptor K band-gap variation of InGaN and AlGaN alloys InGaN bandgaps: J.Wu et al. APL80, 4741 (2004). S. X. Li et al., PRB 71, (2005). L. R. Bailey et al. JAP 104, (2008). AlGaN bandgaps: Yun et al., JAP 92, 4837 (2002). C. I. Wu, APL 74, 546 (1999). C. I. Wu, JAP4249 (1998)

31 Additional Slides

32 β-(al ~0.15 Ga ~0.85 ) 2 O 3 /Ga 2 O 3 (010) Superlattice at 650 C Cross-sectional TEM taken in [201] zone axis projection Roughly normal to bunched steps Onset of step-bunching Homogeneous alloy distribution and abrupt interfaces

33 β-(al ~0.15 Ga ~0.85 ) 2 O 3 (010) Temperature Optimization O-rich growth - ~60 nm layers - ~3 nm/min T sub : 600 C T sub : 625 C T sub : 650 C T sub : 675 C β-(al ~0.15 Ga ~0.85 ) 2 O 3 (010) β-ga 2 O 3 (010) h rms : 0.60 nm h rms : 0.72 nm h rms : 0.90 nm h rms : 1.12 nm Increasing Temperature Enhanced Step-bunching h rms : 1.12 nm h rms : 0.58 nm h rms : 0.50 nm h rms : 0.64 nm Inclusion of Al 2 O 3 Suppresses Step-bunching Smoothest β-(al ~0.15 Ga ~0.85 ) 2 O 3 (010) grown at 650 C

34 RIE of β-ga 2 O 3 (010), (-201), (100) Average of 3 timed etches BCl 3 /Cl 2 10/10 SCCM 10 mt chamber pressure 200 W RF 10 mt chamber pressure Reduced etch rate on (100) Suitable etch rates 200 W Pure BCl 3 etch most effective Reduced discrepancy in etch rates between planes with pure BCl 3 BCl 3 does not etch SPR 220 or NLOF 2020 PR (Cl 2 etches PR >100 nm/min) Active components and products of etch unknown presumably GaCl 3