Suitable Semiconductor Properties for Photoelectrochemistry Evaluated using Nitrides

Univ. Jaume-I Castellon de la Plana, Spain, 2012.09.18 1 Suitable Semiconductor Properties for Photoelectrochemistry Evaluated using Nitrides Katsushi Fujii Global Solar and Initiative : GS + I, the University of Tokyo

Outline Introduction (Semiconductor for Photoelectrochemistry (PEC)) Photoelectrochemical water splitting to generate hydrogen Recent progress of PEC and semiconductor properties Nitride semiconductors Semiconductor Impurity Concentration and PEC Donor concentration of n-type and conversion efficiency Effects of thin p-type layer near the semiconductor surface Effects of Defects for PEC Effects of dislocation (x-ray rocking curve) Effects of point-like defects (impurity concentration) Artificial Structures Near the Semiconductor Surface Thin GaN-layer on In x Ga 1-x N Multi-Quantum Well (MQW) structures of GaN and In x Ga 1-x N Summary 2

Ideal Photoelectrochemical Reaction Water splitting to produce H 2 and N 2 by n-type semiconductor Oxidation 4 2H O O g + H + 2 ( l) 2( ) 4 + 4 e O 2 H 2 Reduction + 4H + 4e 2H 2( g) Electrolyte Light Pt Counterelectrode n-type Semiconductor Working Electrode Ag/AgCl/NaCl Reference Electrode V A Potentiostat Precise Measurement of electrode potential (energy)

Recent Approach using Device Structural Ideas p-gainp+n/p-gaas [1] 3junction a-si SC [2] 5 η eff ( E = 0 rev V P CE 0 ) j 100 η eff : Energy efficiency [%] E rev0 : Voltage for water splitting = 1.23 V V CE : Applied Voltage [V] J p : Photocurrent density [V] P 0 : Power density of light [V] V CE = 0 V (no bias addition) η eff = 12.4% p Wireless structure η eff = 1.75 2.4% [1] O. Khaselev, J.A. Turner, Science 280 (1998) 425. [2] S.Y. Reece, D.G. Nocera et al., Science 334 (2011) 645. University of Tokyo

Photoelectrochmical Water Splitting is Structural Improvement for Water Splitting is Progressing Target is high energy conversion efficiency Introducing the idea of semiconductor device structures Semiconductor Properties? The properties are important for the device design, however Many of researches point out the properties, but There are not so much evidence because Semiconductors for Photoelectrochemistry is Many of them are powder-like Even bulk, polycrystalline or easy to dissolve into electrolyte Thus, it is difficult to evaluate 6 Investigate the Relationships using Nitride Semiconductors

Why Nitrides? Characteristics of Nitrides 8.0 7 Band-gap [ev] 6.0 4.0 2.0 AlN GaN InN 0.0 0.30 0.32 0.34 0.36 0.38 Lattice Parameter [nm] III : V = 1 : 1 AlN Eg = 6.2 ev 200 nm GaN Eg = 3.4 ev 365 nm InN Eg = 0.6 ev 2066 nm Visible Light Relatively high quality, large area, a few micron thickness single crystal is obtained on sapphire. 1. InN GaN AlN from UV to IR 2. Chemically stable Difficult to chemical etch 3. Band edge energy GaN can split water 4. Structurable p/n junction, hetero-junction, etc

Energy [ev] Band Edge Energies of GaN The semiconductor can split water when water reduction / oxidation energy in between the band edge energy Si Dissolved into electrolyte ZnTe CdSe CdS ZnSe ZnS GaN GaAs InP GaP O 2 /H 2 O TiO 2 PbO φ = 0 V H + /H 2 Bi 2 O 3 WO 3 SnO 2 Not so many semiconductors can split water. Fe 2 O 3 SrTiO 3 ph = 0 Weak reduction ability of water BaTiO 3 MnTiO 3 FeTiO 3 KTaO 3 8 φ [V vs NHE] University of Tokyo

Band Edge Energy of GaN (ph Dependence) ph dependence of GaN band edge energy 9 Electrode potential [V vs Ag/AgCl/NaCl] -2.0 Conduction band -3.0-1.0 H 2 /H + -4.0 0.0-5.0 1.0 H 2 O/O 2-6.0 2.0 Valence band -7.0 3.0-2 0 2 4 6 8 10 12 14 16 ph Band edge energies of p-type GaN are the same as that of n-type GaN [1,2]. [1] K. Fujii, K. Ohkawa, Jpn. J. Appl. Phys. 44 (2005) L909. [2] N. Kobayashi et al., Jpn. J. Appl. Phys. 44(2005) L784. Electron energy [ev] GaN can split water! p-type GaN n-type GaN Impedance Meas. Mott-Schottky Plot Cp Calc. Model Rs Cp Rp GaN : 3.42 ev Frequency Scan : Mott-Schottky Plot AC Amplitude : 20 mv University of Tokyo

What kind of electrolyte is suitable? With and without illumination 2.0 in the dark 1.0 0.0-1.0-2.0-1.0 0.0 1.0 2.0 Bias [V vs counterelectrode] Without illumination : With illumination : 2.0 1.0 0.0 under illumination NaOH H 2 SO 4 10 1.0 mol/l H 2 SO 4 : ph 0.5 Na 2 SO 4 : ph 6.0 NaOH : ph 13.5 Na 2 SO 4-1.0-2.0-1.0 0.0 1.0 2.0 Bias [V vs counterelectrode] Typical Schottky diode characteristics Photocurrent observed clearly Basic electrolyte is good due to photocurrent density at zero bias

Without Bias : Carrier Concentration of GaN Photoelectrochemical reaction is Electron-hole pairs, which are generated by light absorption, are separated by the electric field of depletion layer High carrier concentration e - Light h + n-type 12 Recombination φ CB φ VB, α 10-5 [cm -1 ] Low carrier concentration Does depletion layer thickness, which was controlled by semiconductor carrier concentration, affect photoelectrochemical reaction? e - e - Light h + h + n-type ZERO bias φ CB φ VB

Current density [ma/cm 2 ] Without Bias: Carrier Concentration Affects? Carrier concentration dependence 2.0 1.5 1.0 0.5 0.0-0.5 n-gan Carrier Conc. Carrier Concentration 2.6x10 18 cm -3 1.0x10 18 cm -3 3.1x10 17 cm -3 1.7x10 17 cm -3 1.2x10 17 cm -3 5.5x10 16 cm -3 2.1x10 16 cm -3-1.0 0.0 1.0 2.0 Bias [V vs counterelectrode] 13 in 1 mol/l HCl Xe 150 W n-type GaN (0002)ω FWHM < 300 arcsec Carrier concentration : High/low C.C. Low current densities at at high/low carrier conc. i.e., Carrier concentration affects photo electrochemical properties (not only the width of space charge region). [1] M. Ono,, K. Fujii, T. Ito, Y. Iwaki, A. Hirako, T. Yao, K. Ohkawa, J. Chem. Phys. 126 (20079 054708.

Without Bias : Effect of Carrier Concentration Photocurrent density dependent on carrier concentration (Bias = 0.0 V vs counterelectrode) 14 Current density [ma/cm 2 ] 0.8 0.6 0.4 0.2 0.0 10 16 10 17 10 18 10 19 Carrier concentration [cm -3 ] Maximum point is exist Two (or more) mechanisms LIGHT Gas Carrier concentration for maximum photo current density : 10 17 ~10 18 cm -3 [1] n-gan 1mol/L Electrolyte HCl A Pt Metal [1] M. Ono, K. Fujii, T. Ito, Y. Iwaki, A. Hirako, T. Yao, K. Ohkawa J. Chem. Phys. 124 (2007) 054708.

Carrier Concentration is key for Efficiency 15 Series resistance of the system Semiconductor resistance plus electrolyte Rs Light Electrolyte Current density [ma/cm 2 ] 0.8 0.6 0.4 0.2 0.0 10 16 10 17 10 18 10 19 Carrier concentration [cm -3 ] Space Charge Region E C E F E V Semiconductor Space charge region and penetration depth of light Light Electrolyte Semiconductor E C E F E V Penetration depth of light Independent of carrier concentration ~100 nm Space charge region (SCR) Depend on carrier concentration (SCR) 2 /(Carrier conc.)

Hydrogen Generation without Bias Carrier concentration optimized n-type GaN Optimized carrier concentration is; n = 1.6 10 17 cm -3 0.1 1.0 mol/l HCl, V CE = 0.0 V, 150 W Xe-lamp Energy conversion efficiency[1] 0.0 0.0 η eff = 0.5 % 0 60 120 180 240 300 (considering H 2 O splitting) Time [min] η eff = 0.6 % (considering Cl - oxidation E rev = 1.36 V) LIGHT Gas volume generated[ml] 0.5 0.4 0.3 0.2 Calculated Experimental 0.6 0.4 0.2 16 Current density [ma/cm 2 ] Gas 1mol/L Electrolyte HCl Hydrogen generation from water without bias. n-gan A Pt Metal [1] S.U.M. Khan, M. Al-Shahry, W.B. Ingler Jr., Science 297 (2002) 2243.

Samples for these experiments 17 GaN layers were grown on sapphire substrates by metal-organic vapor phase epitaxy (MOVPE). Electrical properties (from single layer grown by the same condition) n-type GaN Si-doped n = 1.510 18 cm -3 µ = 260 cm 2 /Vs p-type GaN p = 1.810 17 cm -3 Mg-doped µ = 13 cm 2 /Vs (after 750 o C 20 min activation anneal) GaN with p/n-junction Sample structure n-type GaN (for reference) Ti/Au (10/50 nm) Light GaN:Mg x nm Ti/Au (10/50 nm) Light GaN:Si LT-GaN Sapphire (0001) 1.6 µm x = 5, 10, 20, 50, 100 GaN:Si LT-GaN Sapphire (0001) 1.6 µm

Photocurrent Density at Zero Bias Dependence on surface p-layer thickness 18 Photocurrent density : 0.5 0.4 0.3 0.2 0.1 0.0 at ZERO bias Electrode p-type GaN n-type GaN (1.6 µm) Sapphire substrate 0 20 40 60 80 100 p-layer thickness [nm] < 10 nm Increases with p-type layer thickness from 10 nm to 20 nm Decreases drastically > 20 nm Decreases with p-type layer thickness The potential profile near the interface is changed by the surface p-type layer. Comparison : 10 nm / 0 nm p-type layer

Electron energy [ev] Conduction Band-Edge Energy at Zero-Bias Especially for 0, 5, 10, 20 nm p-type layer -3.5-4.0-4.5-5.0 p-layer thickness 0 nm 5 nm 10 nm 20 nm V add = 0.0 V Ec 0 10 20 30 Distance from interface [nm] 0.6 0.5 0.4 0.3 0.2 0.1 bias = 0.0 [V vs counterelectrode] 0.0 0 20 40 60 80 100 p-layer thickness [nm] Depletion layer thickness: p-type layer : 0 nm < 5 nm < 10 nm < 20 nm The shape changes at p = 20 nm (Existing maximum value) Photocurrent decreases rapidly 19

Time Dependence in1.0 mol/l NaOH Time dependence of photocurrent density with zero bias 21 Zero bias 1.0 0.8 0.6 0.4 0.2 0.0 26 min 1.0 mol/l NaOH zero bias 312 min 470 min 1336 min 0 500 1000 1500 Time [min] Increased after once decreased!? Photocurrent density is; Rapid decrease once, increasing slowly to maximum, decreasing after that What is the origin? Time dependence surface checking Improving trial without using co-catalyst! n-gan:si 1.0 10 18 cm -3, 4.2 µm

Stability: What Is the Key for Stable Photoelectrochemical Reactions? Proposed model 0.8 0.6 0.4 0.2 0.0 Defect disappeared Defect generation 0 500 1000 1500 Time [min] Etched face orientation n-gan:si 1.0 10 18 cm -3, 4.2 µm Remove the origins of the defect formation is BELIEVE to the key for the stability! sample sample sample ESTIMATED stable sample [1] K. Sato, K. Fujii, K. Koike, T. Goto, T. Yao, Phys. Stat. Sol. (c) 6, (2009) S635. Different conditioned surface from the bulk Surface layer disappeared at first sample sample Degradated sample 22 Zero bias Progress of surface etching Defect formation and recovering Formation of rough surface morphology

n-type GaN for Stable Reaction n-type GaN (0001)+c undoped Si-doped (1.5) Si-doped (1.5) Si-doped (1.0) Si-doped (0.5) Carrier Conc. [cm -3 ] Donor Conc. [cm -3 ] 4.6 10 16 not available X-ray 2θ-ω (0002) FWHM [arcsec] X-ray ω (0002) FWHM [arcsec] X-ray 2θ-ω (10-12) FWHM [arcsec] X-ray ω (10-12) FWHM [arcsec] 340 310 1230 690 5.0 10 17 1.9 10 18 240 230 1300 690 8.4 10 17 1.1 10 18 230 190 550 320 1.9 10 17 3.3 10 17 250 190 530 300 5.5 10 16 1.3 10 16 240 190 530 290 Focus to the defects of n-type GaN (1)Different XRC (Especially (10-12)) (2)Different Si doping (defined by donor concentration) The number under the Si-doped indicates the flow rate of SiH 4 100 ppm diluted by H 2 at the growth (sccm). grown by MOVPE 23 Zero bias

Experiments Potentiostat A V Light 24 Zero bias (1) Long time stability Electrolyte : NaOH 1.0 mol/l Illuminated light : 270 280 mw/cm 2 Xe-lamp Bias : Nothing (connect to counterelectrode) Time : 650 min (Evaluated at 600 min) (2) Electrochemical properties Electrolyte : H 2 SO 4 0.5 mol/l Bias : -1.4 +1.0 V vs Ag/AgCl/NaCl Impedance from C-V (Mott-Schottky plot Donor Conc.) n-type GaN Working Electrode Electrolyte Ag/AgCl/NaCl Reference Electrode Pt Counterelectrode

Dependence on X-ray Rocking Curve (ω-scan) Photocurrent density Max. photocurrent = 1 1.0 NaOH 1.0 mol/l NaOH 1.0 mol/l 1.0 270 mw/cm 2 Xe-lamp max i = 1.0 0.8 Si-GaN: (10-12)XRC 0.8 FWHM 320 arcsec 0.6 un-gan un-gan (x10) 0.6 Si-GaN (10-12)XRC FWHM = 320 arcsec 0.4 0.4 Si-GaN (10-12)XRC 0.2 0.2 FWHM = 690 arcsec Si-GaN: (10-12)XRC FWHM 690 arcsec 270 mw/cm 2 Xe-lamp 0.0 0.0 0 200 400 600 0 200 400 600 Time [min] Relative-photocurrent density [-] Time [min] 25 Zero bias The higher initial photocurrent density is obtained with the smaller XRC- FWHM for the Si-doped GaN. Relative changes of Si-doped GaN are similar. (undoped GaN is stable.) undoped GaN (10-12) XRC FWHM : 690 arcsec University of Tokyo

Dependence on the amount of Si doping Photocurrent density Max. photocurrent = 1 1.0 NaOH 1.0 mol/l 1.0 NaOH 1.0 mol/l 270-280 mw/cm 2 Xe-lamp max i = 1.0 0.8 Donor Conc. 0.8 1.1x10 18 cm -3 (1.5) 0.6 un-gan un-gan (x10) 0.6 1.3x10 16 cm -3 (0.5) 0.4 0.2 3.3x10 17 cm -3 (1.0) 1.3x10 16 cm -3 (0.5) 0.0 0 200 400 600 Time [min] Relative-photocurrent density [-] 0.4 0.2 Donor Conc. 1.1x10 18 cm -3 (1.5) 3.3x10 17 cm -3 (1.0) 270-280 mw/cm 2 Xe-lamp 0.0 0 200 400 600 Time [min] The lower Si doping shows the lower initial photocurrent. and shows the higher stability (similar to undoped GaN). (The smaller XRC FWHM samples of Si-doped GaN were used except for undoped one.) 26 Zero bias Si-doped GaN () shows SiH 4 100 ppm flow rate (sccm) University of Tokyo

Photocurrent and the Surface after 600 min After 600 min (Si-doped GaN) 0.8 0.6 0.4 0.2 600 min after light illumination 270-280 mw/cm 2 Xe-lamp Before reaction High XRC 0.0 1.0x10 16 1.0x10 17 1.0x10 18 Donor Concentration [cm -3 ] Surface after 650 min The lower Si doping shows the higher photocurrent density at 600 min. Stable photocurrent samples show flat surface after the reaction. ( Si-doped including the larger XRC FWHM sample.) 27 Zero bias (Optical microscope) The numbers in photos are the donor concentration. University of Tokyo

Samples for Experiments GaN cap layer with In x Ga 1-x N bulk layer structures (Single crystal; x In = 0.19) 29 undoped GaN cap (0, 5, 15, 50 nm) n-type GaN:Si undoped GaN (0001) Sapphire (substrate) undoped In x Ga 1-x N bulk (0.2 µm) Low temperature GaN buffer Grown by conventional Metal-Organic Vapor Phase Epitaxy (MOVPE) References is GaN bulk single crystal Carrier concentration of n-type GaN is 2 10 18 cm -3. The Univ. of Tokyo

Samples for Experiments In x Ga 1-x N / GaN multi-quantum well (MQW) structures (Single crystal; x In = 0.19) undoped GaN cap (5 nm) 30 5 pairs n-type GaN:Si undoped GaN (0001) Sapphire (substrate) undoped In x Ga 1-x N well (5 nm) undoped GaN barrier (2, 5, 10 nm) Low temperature GaN buffer Grown by conventional Metal-Organic Vapor Phase Epitaxy (MOVPE) References are In x Ga 1-x N (0.2 µm) and GaN bulk single crystal Carrier concentration of n-type GaN is 2 10 18 cm -3. The Univ. of Tokyo

Experiments 31 Potentiostat Light V A Electrolyte (1) Electrochemical properties Electrolyte : H 2 SO 4 0.5 mol/l Bias : -1.0 to +1.0 V vs Ag/AgCl/NaCl Impedance from C-V (Mott-Schottky plot Resistance) (2) Photoelectrochemical properties Electrolyte : H 2 SO 4 0.5 mol/l Illuminated light : 90 mw/cm 2 Xe-lamp Bias : -1.4 to +1.0 V vs reference electrode (RE) -1.0 to +1.0 V vs counterelectrode (CE) GaN barrier/mqw Working Electrode Ag/AgCl/NaCl Reference Electrode Pt Counterelectrode The Univ. of Tokyo

Impedance Results of GaN barrier on In x Ga 1-x N 1/C 2 [10 12 cm 4 /F 2 ] 50 40 30 20 10 0.5 mol/l H 2 SO 4 InGaN xin = 0.19 GaN cap 5 nm GaN cap 15 nm GaN cap 50 nm InGaN bulk x0.1 GaN bulk 0 1 GaN = 5 nm Bias 0.8 and 1.0 V : 1/C 2 drastically changes GaN = 15, 50 nm Some strange changes of 1/C 2 -V were observed Series resistance for GaN = 5 nm is the highest. GaN = 5 nm acts as electron trap? Series Resistance (Rs) [Ω] [1] K. Fujii et al., 2011MRS Proc. 1387 (2012) 11-1387-e06-04. 1000 100 10 0.5 mol/l H 2 SO 4 InGaN xin = 0.19 GaN cap 5 nm InGaN bulk GaN cap 50 nm GaN cap 15 nm GaN bulk 32 n-type GaN:Si undoped GaN (0001) Sapphire (substrate) The Univ. of Tokyo

Mott-Schottky Summary of GaN barrier on In x Ga 1-x N 33 Flatband potential [V vs Ag/AgCl/NaCl] 0.2 0.0-0.2-0.4-0.6-0.8 0.5 mol/l H 2 SO In 4 x Ga 1-x N GaN bulk 0 20 40 60 80 100 GaN cap thickness [nm] Flatband potential : GaN < 5 nm InGaN (x = 0.19) like GaN > 15 nm GaN like n-type GaN:Si undoped GaN (0001) Sapphire (substrate) The Univ. of Tokyo

Cyclic Voltammetry for GaN barrier on In x Ga 1-x N 0.5 mol/l H 2 SO 4, Bias vs Ag/AgCl/NaCl GaN cap 50 nm GaN cap 15 nm GaN cap 05 nm 2 under illumination 1 0-1 in the dark -2-3 -4-5 2 under illumination 1 0-1 in the dark -2-3 -4-5 2 under illumination 1 0-1 in the dark -2-3 -4-5 34 Si-doped GaN bulk 2 under illumination 1 0-1 in the dark -2-3 -4 InGaN bulk 2 1 0-1 -2-3 -4 under illumination in the dark Thicker GaN shows some GaN photocurrent InGaN turn-on bias +0.25 V vs Ag/AgCl/NaCl n-type GaN:Si -5-5 Scan speed : 20 mv/s undoped GaN (0001) Sapphire (substrate) The Univ. of Tokyo

Cyclic Voltammetry for GaN barrier on In x Ga 1-x N 0.5 mol/l H 2 SO 4, Bias vs Ag/AgCl/NaCl 10 2 60 mv/dec -0.64 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 GaN cap 50 nm GaN cap 15 nm GaN cap 05 nm under illumination in the dark 10 2 60 mv/dec -0.64 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 under illumination in the dark 10 2 60 mv/dec -0.44 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 under illumination in the dark 35 10 2 60 mv/dec -0.59 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 Si-doped GaN bulk under illumination in the dark InGaN bulk 10 2 60 mv/dec -0.69 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 under illumination in the dark Majority carrier shows some 60 mv/dec --- reaction limited reaction No specific flatband effect Scan speed : 20 mv/s n-type GaN:Si undoped GaN (0001) Sapphire (substrate) The Univ. of Tokyo

Impedance Results for In x Ga 1-x N/GaN 5QW Compared with In x Ga 1-x N and GaN bulk samples (measured from impedance measurements) 1/C 2 [10 12 cm 4 /F 2 ] 25 20 15 10 5 0 InGaN F.B. = -0.11 V GaN F.B. = -0.68 V 1.5 Series Resistance [Ω] 10 6 GaN barrier barrier 10 nm 10 5 10 4 10 3 10 2 0.5 mol/l H 2 SO 4 InGaN well 5 nm barrier 5 nm barrier 2 nm InGaN bulk GaN bulk 10 1-1.0-0.5 0.0 0.5 1.0 36 n-type GaN:Si undoped GaN (0001) Sapphire (substrate) GaN bulk sample is the lowest resistance, In x Ga 1-x N bulk sample is the next. Resistance of In x Ga 1-x N / GaN 5QW samples increase with GaN thickness. [1] K. Fujii et al., J. Phys. Chem. C 115 (2011) 25165. The Univ. of Tokyo

Mott-Schottky Summary of In x Ga 1-x N/GaN 5QW In x Ga 1-x N and GaN bulk samples Flatband potential [V vs Ag/AgCl/NaCl] 0.2 0.0-0.2-0.4-0.6 InGaN bulk 0.5 mol/l H 2 SO 4 GaN bulk -0.8-5 0 5 10 15 20 25 GaN barrier thickness [nm] 37 n-type GaN:Si undoped GaN (0001) Sapphire (substrate) In x Ga 1-x N / GaN 5QW samples do NOT show the Mott-Schottky relationships. This is because the wells act as carrier (electron) traps. The Univ. of Tokyo

38 Static Photocurrent of In x Ga 1-x N/GaN 5QW vs CE Compared with In x Ga 1-x N and GaN bulk samples 1.0 0.8 0.6 0.4 0.2 0.0-0.2-0.4 0.5 mol/l H 2 SO 4 InGaN well = 5 nm GaN barrier barrier 2 nm InGaN bulk GaN bulk Bias [V vs CE] barrier 10 nm barrier 5 nm 0.00-5 0 5 10 15 20 25 Photocurrent of In x Ga 1-x N / GaN MQW samples is; similar to In x Ga 1-x N bulk sample for thickness of GaN with 5 and 10 nm and similar to GaN bulk sample for thickness of GaN with 2 nm. Highest photocurrent at zero bias was obtained at GaN = 2 and 5 nm! 0.20 0.15 0.10 0.05 InGaN bulk 0.5 mol/l H 2 SO 4 at +0.0 V vs CE GaN bulk GaN barrier thickness [nm] n-type GaN:Si undoped GaN (0001) Sapphire (substrate) The Univ. of Tokyo

Cyclic Voltammetry for In x Ga 1-x N/GaN 5QW 0.5 mol/l H 2 SO 4, Bias vs Ag/AgCl/NaCl InGaN 5nm/GaN 10nm x5 InGaN 5nm/GaN 5nm x5 InGaN 5nm/GaN 2nm x5 2 1 0-1 -2-3 -4-5 under illumination in the dark 2 1 0-1 -2-3 -4-5 under illumination in the dark 2 1 0-1 -2-3 -4-5 under illumination in the dark 39 n-type GaN:Si undoped GaN (0001) Sapphire (substrate) InGaN bulk 2 under illumination 1 0-1 in the dark -2-3 -4 Si-doped GaN bulk 2 under illumination 1 0-1 in the dark -2-3 -4 Thick GaN barrier shows high resistivity Thinner GaN barrier shows bulk GaN-like -5-5 Scan speed : 20 mv/s The Univ. of Tokyo

Cyclic Voltammetry for In x Ga 1-x N/GaN 5QW 0.5 mol/l H 2 SO 4, Bias vs Ag/AgCl/NaCl InGaN 5nm/GaN 10nm x5 InGaN 5nm/GaN 5nm x5 InGaN 5nm/GaN 2nm x5 10 2 60 mv/dec -0.49 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 under illumination in the dark 10 1 10 0 10-1 10-2 10-3 10-4 10 2 60 mv/dec -1.14 at 1E0 under illumination in the dark 10 1 10 0 10-1 10-2 10-3 10-4 10 2 60 mv/dec -0.59 at 1E0 under illumination in the dark 40 n-type GaN:Si undoped GaN (0001) Sapphire (substrate) InGaN bulk 10 2 60 mv/dec -0.49 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 under illumination in the dark Si-doped GaN bulk 10 2 60 mv/dec -0.59 at 1E0 10 1 10 0 10-1 10-2 10-3 10-4 under illumination in the dark Thicker GaN does not shows 60 mv/dec. --- NOT reaction limited --- related electron transfer in QW structure? Thin GaN barrier shows GaN-like i-v Scan speed : 20 mv/s The Univ. of Tokyo

Summary 1. Investigated semiconductor properties for photoelectrochemistry. The properties are important for semiconductor devices, are also important for photoelectrochemical devices. 2. Not only the depletion layer thickness depended by impurity concentration but also the shape (monotonically change) and the conductivity of the out region are also important to improve the conversion efficiency. 3. Not only the dislocation reduction but also the point defect reduction is important to improve the reliability without usion co-catalyst. 4. Surface fine structure changes the (photo)electrochemical properties of bulk-side semiconductor, but the effect probably different in the case of the equilibrium condition (without carrier flow) and steady condition (with carrier flow under light illumination, for example). 42

Acknowledgment 1. Tokyo University of Science Prof. Kazuhiro Ohkawa, Masato Ono, Takashi Ito, Yasuhiro Iwaki 43 2. Tohoku University Prof. Takafumi Yao, Prof. Takenari Goto, Prof. Meoung Whan Cho Kayo Koike, Mika Atsumi Takashi Kato, Keiichi Sato 3. The University of Tokyo Prof. Yoshiaki Nakano, Prof. Masakazu Sugiyama Akihiro Nakamura, Futami Sano