Inorganic Thin Films: Future Perspectives Global Climate Energy Project Solar Energy Workshop: Thin-Film Photovoltaics October 19, 2004 John P. Benner Division Manager Electronic Materials and Devices National Center for Photovoltaics
Future Perspectives from 1975 2004 CdTe 25.9 0.845 0.755 16.5
20 Best Research-Cell Efficiencies Thin-Film Inorganics CuInSe CdTe 2 NREL NREL 16 Amorphous silicon Univ. of So. Florida NREL Efficiency (%) 12 (stabilized) Kodak Boeing ARCO BP Solar Photon Energy Boeing Univ. of So. FL EuroCIS Boeing AMETEK 8 Matsushita Monosolar Kodak Boeing United Solar Boeing ECD Univ. of Maine 4 0 RCA 1975 1980 1985 1990 1995 2000 2005 026587181
World PV Cell/Module Production (MW) 800 700 ~50 MW Thin-Film in 2003 744.1 600 561.8 500 400 300 Rest of world Europe Japan U.S. 287.7 390.5 200 100 0 201.3 125.8 154.9 33.6 40.2 46.5 55.4 57.9 60.1 69.4 77.6 88.6 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Source: PV News, March 2004
U.S. Thin-Film Manufacturing 25 a-si CIS CdTe 20 Projected 15 10 5 0 1996 1999 2002 2005
Think Big -- Very Big 2-5 GW Factories GW Annual Module Production 40 35 30 25 20 15 10 5 0 1990 1995 2000 2005 2010 2015 2020
Ground Rules for GW Scale Factories Dedicated float glass line 5x reduction in glass cost ~$4/m 2 finished Redundant cluster production tools 8x reduction in capital cost ~100 20MW deposition clusters @ $2M each Recycled Effluents 75% utilization Advanced Packaging Factory Aluminum Extruding and Fabrication M. Keshner, et al, Hewlett Packard Final Rpt NREL#ADJ-3-33631-01
Solar Factory Module Cost Comparisons for Completed Solar Panels Cost Summary 20 MW Plant 2 GW Plant (all numbers are per sq. meter) Complete solar panel ready for simple attachment onto a roof Coated Glass $ 23.62 Net Gain $ 4.62 5 x Operating Expenses $ 4.00 $ 1.50 2.5x Materials and depreciation a - Si $ 2.33 + $ 13.35 $ 0.31 + $ 2.67 5x CdTe $ 3.46 + $ 10.00 $ 2.31 + $ 2.00 7.5x CuInGa Se 2 $ 13.96 + $ 13.35 $ 9.31 + $ 2.67 7.5x Assembly, Packaging $ 41.71 & Interconnect $ 10.50 4x Overall process yield 60 % 93 % 1.55x Total manufacturing cost per w att a - Si ( 7%) $ 2.02 $ 0.30 CdTe (11%) $ 1.25 $ 0.21 CuInGa Se 2 (12%) $ 1.34 $ 0.26 + 225 volts _ Notes: If CdTe and CuInGaSe2 could use effective light trapping and be reduced in thickness to 0.4 um like a-si, then their cost per Wp would be $.19 and $.19, respectively. If a-si could use a second junction of a-sige or uc Si, its efficiency would be circa 10% and its cost per Wp would be $.21. September 20, 2004 hp confidential page 10
Solar Factory Module Cost Comparisons for Completed Solar Panels Cost Summary 20 MW Plant 2 GW Plant (all numbers are per sq. meter) Complete solar panel ready for simple attachment onto a roof Coated Glass $ 23.62 Net Gain $ 4.62 5 x Operating Expenses $ 4.00 $ 1.50 2.5x Materials and depreciation a - Si $ 2.33 + $ 13.35 $ 0.31 + $ 2.67 5x CdTe $ 3.46 + $ 10.00 $ 2.31 + $ 2.00 7.5x CuInGa Se 2 $ 13.96 + $ 13.35 $ 9.31 + $ 2.67 7.5x Assembly, Packaging $ 41.71 & Interconnect $ 10.50 4x Overall process yield 60 % 93 % 1.55x Total manufacturing cost per w att a - Si ( 7%) $ 2.02 $ 0.30 CdTe (11%) $ 1.25 $ 0.21 CuInGa Se 2 (12%) $ 1.34 $ 0.26 + 225 volts _ Notes: If CdTe and CuInGaSe2 could use effective light trapping and be reduced in thickness to 0.4 um like a-si, then their cost per Wp would be $.19 and $.19, respectively. If a-si could use a second junction of a-sige or uc Si, its efficiency would be circa 10% and its cost per Wp would be $.21. September 20, 2004 hp confidential page 10
Percentage of Capacity 120% 100% Time to Production: Processes must be better characterized 80% 60% 40% 20% 0% α-si:h - BP 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Yield CIGS - Shell Dec-01 1-Jul Dec-00 Sep-00 Mar-00 Dec-99 Mar-99 Jan-99 Sep-98 Nov-97 Feb-97 Month/ Year CdTe First Solar
16.5% Efficient CdTe Solar Cells Back-contact (C:HgTe:CuTe) CdTe (~10 µm) CdS (0.07-0.1 µm) Zn 2 SnO 4 (0.1-0.2 µm) Cu diffusion Front-contact (In) Anneal in CdCl 2 CdS x Te 1-x O content in CdS CdS ZTO interdiffuse Cd 2 SnO 4 (0.15-0.3 µm) Glass substrate
Thin Film Cells are.. Thin Advantages Low material consumption High throughput potential Module patterning Improved carrier generation profile Drift collection Flexible Semi-transparent Challenges Unique Materials Interdiffusion Grain size Low-lifetime Drift collection Characterization
Effect of Back Contacts Deposition Temperature on Thin-Film CdTe Solar Cell Performance 25 35 45 55 65 75 200 240 280 320 360 400 Fill Factor (%) Contact Deposition Temperature ( C) 450 550 650 750 850 200 240 280 320 360 400 Open Circuit Voltage (mv) Contact Deposition Temperature ( C)
Cu Diffusion from CdCl 2 and Contact Processes High-Resolution SIMS of Cu Concentration Un-quantified 10 21 Quantified Cu Profiles As grown CdCl2 treated W5 USF W11 NREL W7 NREL ZnTe Secondary ion counts 10 6 10 5 10 4 10 3 10 2 ZnTe:Cu no contact no CdCl 2 no contact wet CdCl 2 207 C 285 C 335 C 390 C CdTe CdS SnO 2 Concentration (at/cm 3 ) 10 20 10 19 10 18 ZnTe CdTe CdS SnO 2 10 1 10 0 0 1 2 3 Depth (µm) 4 5 10 17 10 16 0 1 2 3 Depth (µm) 4 5 6
Combined EBIC of ZnTe:Cu Contacted Devices
Think High High efficiency Multi junction Highly ordered, oriented films Single crystal High rate deposition
Polycrystalline Thin Film Tandem Solar Cell CdTe top cell Achieved 50% transmission, 12.7% efficiency CIS bottom cell Achieved 14.5% efficiency 7059 Cornning glass CTO ZTO S-CdS:O CdTe Cu x Te back-contact ITO c-zto / i-zno CBD-CdS CIS Mo Soda-lime glass In contact Ni/Al grids Ni/Al grids In contact FY06 milestone: 15% efficient 4-terminal device will be met one year early
Red QE equals USSC bottom cell 1.0 0.8 a-sige Q_L1067 Q_T2358 0.6 QE µc-si 0.4 0.2 0.0 300 400 500 600 700 Wavelength (nm) 800 900 1000
Film c-si on glass concept epitaxially thickened c-si c-si seed layer glass Many approaches to both seed and epitaxy under study See, review by Berg mann & Werner, Thin Sol Films 2002
Ni-seeded c-si template / HWCVD c-si glass Ni Solid-phase crystallized a-si glass 300 C HW poly c-si Si lifetime > ~10 µs Poly c-si (Ni) H in grain boundaries? glass Poly-Si growth rate 1 Å/s --> slow at ~3 hr per µm heavy H 2 dilution Richardson et al, MRS Spring A, 2004
Single Grain Si Films Induced by Hydrogen Plasma Seeding Single nucleus achieved for holes <0.6 µm Bo et al,jvst B. May 2002
Ta wire improves epitaxy ~ 3Å/s at 270 C a-si:h cone strained c-si (100) substrate Ta filament: about 350 nm epitaxy W filament: 50 to 100 nm epitaxy
Think Small Defects and nanostructure Thinner Devices
CIGS, Ga/(In+Ga)=28.5% AFM SKPM G2 G2 G1 G1 5 µm 5 µm Height (nm) 100 0 2 4 6 8 10 Distance (µm) 100 Potential (mv) Potential height: ~150 mv Depletion width: 150~400 nm.
Cu 0.9 (In 1-x Ga 0.30 ) 1.1 Se 2 Jo (ma/cm^2) 1.00E+00 1.00E-01 1.00E-02 1.00E-03 1.00E-04 1.00E-05 1.00E-06 Jo A 5 4 3 2 1 Ideality Factor A Height of potential peak (mv) Efficiency η (%) 200 150 100 50 0 32 28 24 20 16 12 (a) (b) Predicted from band gap M. A. Green, Solar Cells, P. 89 Measured efficiency 0 0.2 0.4 0.6 0.8 1 1.2 X = Ga/(In+Ga) 8 0 20 40 60 80 100 Ga Content Ga/(In+Ga) (%)
Difficulty for Ga>30% Difficult to dope n-type Difficult to form n-type Cu-poor layers S.-H. Wei et al APL 1998
Quantum Efficiency of CIGS Solar Cells
How is a Crystal 10% Cu Poor? Cu 2 Se CuInSe 2 CuIn 3 Se 5 Cu 2 In 4 Se 7 CuIn 5 Se 8 Neutral Defect Complex (2V Cu + In Cu 2+ ) Zhang et al Phys Rev B 1998 Phase Segregation Material immediately surrounding dislocations and grain boundaries in device-quality CI(G)S will have higher bandgap The α/β hole mirror disappears at [Ga]/[Ga+In] 35% Stanbery TBP
Low Cost Processes Large-Area Optical and Electronic Materials 10000 1000 FPD $/M 2 100 10 1 Fuel Cell Bipolar plate 1 PV Coated Glass Glass 10 100 1000 10000 Million M 2 per Year Solar Fuels Electrode Paint
Advances in PV System Design Achieve Cost Advantages Uni-Solar Amorphous Silicon Field Applied Roofing Products in units to 128W (18 x16, 17 lbs, 33V & 3.88 A) United Solar Shingles
Summary Inorganic Thin-Film PV is on the threshold of increasing market presence. Potential for further improvement 10x reduction in $/m2 2-3x increase in module efficiency Current fundamental understanding in all material systems contains large gaps Large entry investments will demand improved understanding and predictive capability. Shared production infrastructure simplifies start-up and growth
Presented with Great Appreciation for the original work, contributions, discussions and figures from my colleagues: Mowafak Al-Jassim Sally Asher Howard Branz Miguel Contreras Tim Coutts Tim Gessert Falah Hasoon Chun-Sheng Jiang Rommel Noufi K. Ramanathan Manuel Romero Su-Huai Wei Xuanzhi Wu Yanfa Yan Alex Zunger Ken Zweibel Marvin Keshner (HP) B. Stanbery (HelioVolt)