Photovoltaics Challenges and Opportunities in Materials, Machines and Manufacturing K.V. Ravi Applied Materials Inc. 1
What s Your Mindset? Tough Guy Tree Hugger The energy content of all the crude oil left in the ground = 15 hours of sunlight hitting the earth Who s the girly man now? From: Brent Nelson - NREL 2
Driving forces for Solar 1. Increasing costs of conventional energy - Oil > $11/barrel; top tier pricing for electricity in the PG&E territory today - > 3 cents/kwh 2. Climate change, environmental issues - CO2 emissions 3. Energy security - political instabilities in oil producing regions, terrorism 4. Increasing demand for energy from China, India and other emerging economies 5. Increasing levels of government support feed in tariffs, tax credits 3
What is needed to make solar happen? Government Policies Active promotion of this energy source Incentives, tax breaks, legislative initiatives- give solar the same advantages as have been given to carbon/nuclear/hydro based energy generation options Education, public awareness, sustained political will Avoid irrational response to fluctuating oil prices Sustained policies over decades Invest in education Science/Technology/Manufacturing Very large scale manufacture GW scale factories Highly automated, efficient manufacturing capability machines, processes, factory automation, --- Continuing and accelerating research in materials, processes, devices, machines 4
All Path s Lead to Solar Pure Economist s Path: Use what s cheapest, no environmental concern tar sands coal oil Solar s hal e gas All Other Renewables Low CO2 Path: Invest in the future now Invest in Solar before it s too Late: tar Intelligent Solar Investment Now sa Solar neducation: public & future techies coal ds oil gas All Other Renewables From: Brent Nelson - NREL 5
Global crystalline silicon solar cell production 5 15 to 25 GW 45 4279 4 35 21 212 MW 3 25 2 1815 2536 69% 15 1 5 22 1999 42% 287 2 4% 41 21 39% 56 22 34% 75 23 68% 1256 24 45% 25 4% 26 27 28 Year From: Photon International, 28 6
Billions of square inches of silicon 6 5 4 3 2 1 Area of silicon used per year 13 6.6 Photovoltaics ICs 18 7 23 7.6 32 8 45 8.6 58 9.2 25 26 27 28 29 21 Year Metric Tons (X 1) $/Kg 3 25 2 15 1 5 2 18 16 14 12 1 Projection by Photon International 8 6 4 2 3 5 2 292 25 27 21 212 Year Polysilicon Capacity Direct selling price Indirect selling price 32 32 45 1 6 175 7 24 25 26 27 Year 2 Polysilicon Pricing 7
Key market inflection point is coming $/kwh $1.2 $1. $.8 $.6 $.4 $.2 US Residential $/kwh OR/WA AZ/TX/NM Bay Area, CA $. 199 2 21 22 23 Source: Solarbuzz, NREL, RWE Schott, EIA Utility costs Until crossover, subsidy bridges gap between PV and Utility cost PV Costs/Watt falling Efficiency improvements Learning Curve Scale manufacturing Utility Rates Increasing Modeled at 25 yr CAGR of 2.4% Additional upward bias from external factors PG&E service territory Time of use rates as high as $.4/kWh Top pricing tier at $.32/kWh Average Price Parity Coming Crossover in US: ~3-1 yrs Crossover in Europe; 6-8 yrs Crossover in Japan: Now PV increasingly cost competitive with residential utility power 8
Today: Production Cost of Electricity 25 (in the U.S. in 22) 25-5 2 Cost, /kw-hr 15 1 5 1-4 2.3-5. 6-8 5-7 6-7 Cost Coal Gas Oil Wind Nuclear Solar 9
Scale manufacturing and cost reduction 1 Historical Module Cost ($/W) * 1 198 $21.83/W 1985 $11.2/W 199 $6.7/W 1995 $4.9/W 2 $3.89/W Projected Scale to Enable Learning Curve 25 $2.7/W 1 1 1 1 1, 1, 1E5 Cumulative Volume (MW) Production line size (Megawatts per Year):.5 (198) 5 (2) 5 (25) 1 (21) Lines Per Factory 2 3 4 1 * 22 Dollars Source: Navigant Consulting 1
Market share of different Technologies- 27 α-si -4% Ribbon silicon - 3% CdTe 6% CIGS/CIS <1% Single Crystal (CZ) 33% Crystalline Silicon represents ~9% of the PV market The ratio between single crystal and multicrystal may change as we go forward: Multicrystal - 53% In favor of single crystal as higher efficiency cell processing concepts can be more easily implemented with single crystal silicon and advances, such as silicon recharge between ingot growth runs, may further shrink the cost differential between multicrystal and single crystal silicon. From: Navigant Consulting, 28 In favor of multicrystalline ingots if quality improvements narrow efficiency gap with Cz and further cost reductions are achieved as cast ingot technology is easier to scale and easier to wafer ( no OD grinding-less waste and time) 11
Crystalline Silicon PV manufacturing Value Chain Summary Definitions Polysilicon manufacture Ingot manufacture Wafering Cell manufacture Laminate Module and BoS Installation Saw Damage Emitter Formation Passivation Metallization Edge Isolation Test and Sort Poly Silicon Ingots Wafering Saw Damage/ Texture etch Emitter Formation Cell Production / Manufacture Passivation/ AR coating Metallization Edge Isolation Inspect / Test Cell interconenct/ Laminate Module Installation Materials Production Materials Processing Packaging Deployment 12
Module cost breakdown - $/W based on multicrystalline silicon technology (Current) $.15 $.6 $.17 Materials Depreciation Labor OH $.15 $.19 $1.37 (65%) Interest SG&A Modules $.4/W 3 % Wafers $.73/W 53 % Total cost - $ 2.9/W Cells $.23/W 17 % Materials cost breakdown 13
Czochralski process intrinsically limited to ~25 mm ( for solar) and, perhaps, 3 mm for ICs Cast ingot approach not limited in size due to fundamentally different method of heat extraction during crystal growth Heat Extraction Czochralski Heat Extraction 3 mm ingot Wafers per year per tool 35 3 25 2 15 1 5 Czochralski 2 mm ingots Cast ingots 15 275 425 1 Ingot weight- Kg 2 6 95 22 Ingot dimensions- mm (dia. or square) Heat Extraction 6 mm square ingot ~2 mm square bricks Cast Ingots Heat exchanger method 14
Watts/Gram of Silicon 1.9.8.7.6.5.4.3.2.1 Efficiency 15% 17% 19% 21% 3 25 2 15 12 1 Wafer thickness - Microns Efficiency calculations of n + p c-si solar cell vs. wafer thickness at different rear surface recombination velocities. L n is diffusion length. Aberle, Prog. Photovolt: Res. Appl. 8, pg 473 (2) 15
Crystalline Silicon Junction/surface lifetime/ recombination dominated Bulk lifetime dominated Relative External Quantum Efficiency, % a-si:h/μc-si:h Cell Spectral Response a-si:h junction μc-si:h junction 1 8 6 4 2 AM 1.5 global spectrum.3.4.5.6.7.8.9 1. 1.1 1.2 Wavelength, microns 5 4 3 2 1 Number of Sunlight Photons (m -2 s -1 micron -1 ) E+19 16
Typical Solar Cell Structure Ag Grid SiN x AR Coating Efficiency loss mechanisms 1. ohm-cm 1 2 3 4 n + emitter p-si Al-BSF Ag Contact Energy 1 3 Thermalisation losses photo excited pairs lose energy in excess of band gap low energy photons pass through the device without doing useful work - limits efficiency to ~ 44% Junction and contact voltage losses Recombination of photo excited electron - hole pairs Bulk recombination Surface recombination Theoretical upper efficiency limit of a silicon solar cell ~ 33% 1 4 2 4 3 Usable energy (qv) 17
Materials and Cell efficiency where are we today? 4 35 3 25 2 15 1 5 Attributes of Silicon: Availability Efficiency Cost Reliability Non-toxicity Manufacturability Highest reported cell efficiency - % Lab cells not in production Production cells Multjunction Concentrator Cells Single crystal silicon Multicrystalline Silicon Single crystal Silicon, production Multicrystalline Silicon, production Cu(In,Ga)Se2 CdTe Amorphous silicon Organic solar cells 18
Efficiency (%) 4 36 32 28 24 2 16 12 8 4 Best c-si Cell efficiencies-impact of materials quality and device design (Data from NREL) No. Carolina State University Solarex Crystalline Si Cells Large area ( > 14 cm 2 ) (Small area) Single crystal Single crystal Multicrystalline Westing-ARChouse Rear junction/interdigitated device Multicrystalline Spire c-si (n-type) n p n p Inter-digitated contacts to inter-digitated junctions Stanford Varian UNSW UNSW Georgia Tech Sharp Contacts Grain boundaries α-si Emitter UNSW α-si BSF n-type, high lifetime, single crystal silicon UNSW UNSW Georgia Tech UNSW Homojunction device Hetrojunction device c-si (n-type) N+ P P+ Contacts Contacts Sunpower Sanyo (27) FhG-ISE Kyocera (26) p-type, multicrystalline silicon 1975 198 1985 199 1995 2 25 19
Efficiency! - Extrapolations, Exaggerations and Downright lies!! Best module efficiency-% 25 2 15 1 5 Single Crystal Multi Crystal Average module efficiencies: Single Crystal - 14.1% Multi crystal - 13.7 % a-si CGS First Solar -CdTe Sharp Suntec Motec Solar world Kyocera Device structures that take advantage of the higher quality of single crystal silicon JA Solar BP Schott Sanyo Sunpower Best reported cell efficiencies Reported conversion efficiency numbers are all over the map! Not everyone follows a standardized test method* Best cell efficiencies reported at technical and trade conferences do not translate into production module efficiencies Module efficiencies, based on volume manufacturing, are substantially lower then the best reported cell efficiencies Large gap between R&D and manufacturing * For example see - IEEE Spectrum April 28 p 37 Thin Films Crystalline Silicon Selected PV manufacturers From: Photon International, 28 2
PV industry today is characterized by: Machines/Manufacturing Small volume ( 5 to ~ 2 MW/year) factories Crystalline silicon (wafer based) technology has not changed substantially for over 3 years (with some notable exceptions) Non silicon thin film technologies (E.G. CdTe, CIGS, etc.) are characterized by proprietary processes with home built/one of a kind tools, practiced by a few, mostly single, companies Thin film silicon ( α-si, tandem structures) beginning to emerge with integrated, turn key factories AMAT Sunfab Thin Film Lines Very high efficiency concepts based on single crystal compound semiconductors and complex multijunction structures are even further away from volume manufacturing 21
Benchmarking? C-Si PV industry today No standards, no industry road map Low volume factories Minimal inter-tool automation Little factory automation (software) No SPC Low yields, large bin splits Large companies integrating individual tools purchased from different tool manufacturers Small and new companies getting state of the art turnkey lines from several manufacturers Contrast with the Semiconductor industry Industry standards (SEMI) Industry road maps (ITRS) Common tools differentiation in design, integration, applications, markets Very high yields Very complex materials systems and processes Sophisticated, automated mega fabs PV industry needs to intelligently copy from the semiconductor and display industries A big issue is differentiation if common tools and processes become prevalent Differentiation cannot be based on design, process and equipment but needs to be based on scale manufacturing, lower cost of manufacture ( higher yields, efficiencies), applications and market development 22
SunFab Thin Film Line 23 23
SunFab PECVD 5.7 Enable new 5.7m2 glass substrate standard (2.2m x 2.6m) 24 24
Where is silicon going? Significant opportunity for thin film silicon α-si and tandem structures Low efficiency offset by low cost Large integrated modules 5.7 meters square Application in large power systems, BIPV not constrained by area Two directions to date with c-si based on traditional thinking Low cost multicrystalline silicon wafers derived from directionally solidified silicon ingots lower efficiency Higher cost, higher efficiency single crystal silicon Future directions Efficiency enhancements with thin film silicon multiband gap structures, light trapping etc. Continuing cost reductions with multicrystalline ingots/wafers Larger ingots up to 1 Ton (1Kg), thinner wafers Efficiency advancements thru material and process innovation Cost reduction and efficiency enhancements with single crystal silicon n-type silicon, thinner wafers Advanced device structures Convergence between thin films and c-si ( hetrojunctions; multi band gap devices --) 25
PV trajectory of the future? Silicon today α-si integrated factories at 6 to 1% efficiency emerging Multicrystalline silicon dominates, Single crystal for high efficiency - Efficiency range for conventional homojunction devices 15 to 17% Silicon tomorrow Efficiency and price pressures from thin film technologies CdTe, α-si Increasing drive towards high efficiency devices with crystalline silicon 25 Efficiency-% 2 15 1 5 Rear junction /interdigitated device c-si (n-type) n p n p α-si Emitter α-si BSF Hetrojunction device c-si (n-type) a-si a-si/c-si CdTe c-si(m) c-si(cz) c-si (backcont) c-si (HIT) Inter-digitated contacts to inter-digitated junctions n-type, high lifetime, single crystal silicon Contacts 26
Extending Core Nanomanufacturing Competencies into Solar 1.E-1 1.E-2 $ per Transistor $/Transistor 1.E-3 1.E-4 1.E-5 1.E-6 196 197 198 199 2 21 Experience from Semiconductor Manufacturing Cost / Area (relative) 1 8 6 4 2.7m 2 1.4m 2 Substrate Area 2.8m 2 Cost/Area 4.4m 2 5.2m 2 1999 21 23 24 26 Experience from LCD Manufacturing German Advisory Council on Global Change (WBGU), 24 27
28
Energy Consumption (kwh/wp) Energy Paybeck Period (y) 7 6 5 4 3 2 1 Energy Consumption (kwh/wp) Energy payback period, Germany (y) Energy payback period, California (y) Cell efficiency % Wafer thickness - μm 1995 25 21 15 17 2 45 25 15 Hess et. al. 22 nd European Photovoltaic Solar Energy Conference, Sept 27, Milan, Italy 29