, '

Electricity from Photovoltaic Solar Cells

Flat-Plate Solar Array Project

10 Years of Progress

October 1985

•• ,United StatesDepartment of Energy .JPL

Jet Propulsion LaboratoryCalifornia Institute of Technology

NI\SI\National Aeronautics andSpace Administration

Photovoltaic Module Progress

1975 1985Flat or nonconcentrating module prices have droppedas module efficiencies have increased. Prices are in1985 dollars for large quantities of commercial products.

~

«LUa...f­f­«~<f}

LUUa:a...

75

5

MODULE PRICE

;01<10°.

>­uZLU

u5tt

LU

(f)

a::«LU

>-

1975 1985Typical module lifetimes were less than 1 year butare now estimated to be greater than 10 years.(Ten-year warranties are now available.)

Technology advancement in crystalline silicon solar cellsand modules (non-concentrating).

Union Carbide Corporation (UCC) funded the nowoperational silicon refinement production plant with1200 MTlyear capacity. DOEIFSA-sponsored effortswere prominent in the UCC process researchand development.

The automated machine interconnects solar cellsand places them for module assembly. The second­generation machine made by Kulicke and Soffa wascost shared by Westinghouse Corporationand DOEIFSA.

Module evolution is represented by typical modules fromthe series purchased by DOEIFSA between 1975 and 1985.The modules were designed and manufactured by industryto FSA specifications and were evaluated by FSA.

More technology advancements are shown oninside of back cover. Use of modules in photovoltaicpower systems are shown on outside of back cover.

FRONT COVER PHOTO:

The Block I module (1975) beingheld in front of four Block V modulesrepresents the progress ofa la-year cooperativeindustryluniversitylDOEIFSAeffort.

Electricity from Photovoltaic Solar Cells

Flat-Plate Solar Array Projectof the

U.S. Department of Energy'sNational Photovoltaics Program

10 Years of Progress

October 1985

Elmer Christensen

. United Statesi Department of Energy"

.lPLJet Propulsion LaboratoryCalifornia Institute of Technology

NI\S/\National Aeronautics andSpace Administration

The Role of Photovoltaics:An Historical PerspectiveThe world's demand for more energy continues,

and will continue in the foreseeable future. Theeconomic growth of developing nations depends uponincreasing energy supplies. In technically advancedcountries, long-range economic growth depends onthe availability of inexpensive energy. Historically, inthe United States, the main source of energy haschanged from wood to coal to petroleum, and it isinevitable that change will continue as fossil fuels aredepleted. Within a lifetime, it is expected that mostU.S. energy will come from diverse sources, includingrenewables, instead of from a single type of fuel.There are many indications that the percentage of thetotal consumed energy supplied by electrical powerwill continue to increase.

The use of photovoltaics (PV) is a promising way togenerate electricity that could provide significantamounts of power in the years to come. It is a form ofsolar energy that uses solar cells to generateelectricity cleanly and reliably, directly from sunlightwithout moving parts. PV power systems are simple,flexible, modular, and adaptable to many differentapplications in an almost infinite number of sizes andin many environments. Although photovoltaics is aproven technology that is cost-effective for hundredsof small applications, it is not yet cost-effective forlarge-scale utility use in the United States. Foreconomical widespread use, the cost of generatingPV power must continue to be decreased by reducingthe initial system cost, by reducing land or arearequirements, and by increasing the operationallifetime of its systems.

Early needs of the space program for electricalpower were satisfied by crystalline silicon solar cells.During the 1960s and 1970s, it was demonstratedthat PV was a dependable electrical power source forspacecraft. In this time interval, solar-cell quality andperformance improved and the costs decreased.However, the costs were still much too high forwidespread use on Earth. There was a need to .drastically lower the manufacturing costs of solar cellsif they were to be a practical, widely used terrestrialpower source.

In 1975, the U.S. Government initiated a terrestrialPV research and development (R&D) project with the

ii

objective of reducing the cost of manufacturing solarcells and modules. This effort, which was assigned tothe Jet Propulsion Laboratory (JPL) in Pasadena,California, evolved from more than a decade and ahalf of spacecraft PV power-system experience at JPLand from recommendations of a conference held in1973 at Cherry Hill, New Jersey, on SolarPhotovoltaic Energy.

After about 2 years of technical progress andeconomic analyses by the Flat-Plate Solar Array(FSA) Project, a growing community of peoplebecame convinced that the Project's goals wererealistic and that PV power systems had the potentialof being competitive with those based on fossil fuels.Additional federally supported efforts, sponsored bythe U.S. Department of Energy (DOE), covering allaspects of PV power-system R&D, were initiated atthe Solar Energy Research Institute (SERI), SandiaNational Laboratories, and other organizations. Thesebroadened efforts included concentrator modules, PVsystems studies, application demonstrations, andresearch on cell materials other than silicon anddevices such as thin-film cells. Most of the DOE fundshave sponsored R&D in private organizations anduniversities. The result has been an effectiveGovernment-university-industry team that hascooperated to advance PV technology rapidly.

In parallel, a small but fast-growing terrestrial PVindustry evolved, with products that wereeconomically competitive for small stand-alone PVpower systems. A few megawatt-size utility-connectedPV installations, made possible by Governmentsponsorship and tax incentives, have demonstratedthe technical feasibility and excellent reliability of largemulti-megawatt PV power-generation plants.

Although fuel prices have leveled off, worldwideinterest in photovoltaics continues, especially incountries where electricity is more expensive than inthe United States. It is anticipated that the field ofphotovoltaics, with its tremendous potential, willregain its place in the sun as fossile fuel energysources deplete and new energy technologiescontinue to evolve.

Flat-Plate Solar ArrayProject Summary

The Flat-Plate Solar Array (FSA) Project, aGovernment-sponsored photovoltaics project, wasinitiated in January 1975 (previously named the Low­Cost Silicon Solar Array Project) to stimulate thedevelopment of PV systems for widespread use. Itsgoal then was to develop PV modules with 10%efficiency, a 20-year lifetime, and a selling price of$0.50 per peak watt of generating capacity (1975dollars). It was recognized that cost reduction of PVsolar-cell and module manufacturing was the keyachievement needed if PV power systems were to beeconomically competitive for large-scale terrestrialuse.

The project was initiated at JPL to meet thesegoals through R&D of all phases of flat-plate moduletechnology, from solar-cell silicon material refinementthrough verification of module reliability andperformance. The Project sponsored paralleltechnology efforts with periodic progress reviews andcontinuing sponsorship of only the most promisingoptions. A module manufacturing cost-analysiscapability was developed that permitted cost goalallocations to be made for each module technology,based upon potential for achievement. Economicanalyses, done as technical progress was achieved,permitted assessments to be made of each technicaloption's potential for meeting the goal and of theProject's overall progress toward the national goal.

Excellent technical progress across the entireproject was accomplished over the years, withgrowing interest and participation by the privatesector. More recently, effective energy conservationpractices, a leveling of energy prices, a perceptionthat energy prices are not increasing endlessly, and achange in Government emphasis has altered thepicture for photovoltaics. DOE's NationalPhotovoltaics Program was redirected to longer-rangeefforts that the private sector avoids because ofhigher risk and longer payoff time. Consistent withthese new directions, FSA has been concentrating itsefforts on overcoming specific critical technologicalbarriers that inhibit large-scale PV use. Theseactivities require an in-depth, long-range, integratedeffort that industry cannot reasonably be expected toundertake alone. This work is being performed by ateam that includes universities, industry (which sharesthe cost), and Government laboratories. This teamhas worked together successfully for 10 years.

An estimate that PV-generated power should costabout $0.15/kWh in the 1990s to be competitive inutility central-station generation plants is the basis fora DOE Five-Year Research Plan. The high cost of PVgenerating equipment, especially modules, remains

iii

the major limitation. Continuing solar-cell and modulecost reductions are still the key activities. However, itis now known that, for a utility plant, area-relatedcosts are significant enough that flat-plate moduleefficiencies must be raised to between 13 and 17 %(more energy generated per fixed-area dollar), andmodule life should be extended to 30 years. It isbelieved by FSA reseachers that both of theserequirements and the corresponding cost reductionsare possible, but are achievable only with a dedicatedeffort. Present FSA activities emphasize highefficiency, long life, reliability, and low cost for solarcells, modules, and arrays.

Major ProjectAccomplishments

«I Established the basic technologies for allaspects of manufacturing and evaluated non­concentrating crystalline silicon PV modules andarrays for terrestrial use.

• Established a new low-cost, high-purity silicon­feedstock-material refinement process.

111 Made significant advances in quality and costreduction of:

• Silicon sheet for solar cells.

• Higher efficiency solar cells.

• Manufacturing processes.

111 Established PV module/array engineering/designand evaluation knowledge and capabilities.

111 Established PV module encapsulation systems.

111 Devised manufacturing and life-cycle costeconomic analysis capabilities.

«I Transferred the above technologies to theprivate sector by dynamic interactive activitiesincluding hundreds of contracts, comprehensivemodule development and evaluation efforts, 25Project integration meetings, research forums,presentations at hundreds of technical meetings,and advisory efforts to industry on specifictechnical problems.

III Stimulated the establishment of a commercialPV industry in the United States.

PhotovoltaicsCommunityInteractions

Following the OPEC oil embargo of 1973, aconference of photovoltaics experts was held atCherry Hill, New Jersey, to assess the viability ofphotovoltaics as a terrestrial electrical energysource. In response to a 1974 request from theNational Science Foundation, JPL proposed a PVmodule R&D effort based upon the Cherry Hillconference recommendations. This was accepted inJanuary 1975 and a Project, now called Flat-PlateSolar Array, was initiated.

The JPL Project quickly established relationshipswith university research teams, non-profitorganizations, and commercial firms interested in thedevelopment of low-cost photovoltaics. Thefundamental strategy adopted by JPL was to involveindustry and university teams in a highly interactiveeffort along parallel paths of technologydevelopment, and to select the best options. Theeffort has been highly goal oriented using a schemeof cost allocations to guide the progress of theactivities.

By 1978, the MIT-Lincoln Lab; MIT-Energy Lab;Brookhaven National Lab, the NASA-Lewis ResearchCenter; Sandia National Lab; a newly formed SERI;the U.S. Army MERADCOM; and others had becomemembers of the Federal program under the newlyestablished U.S. Department of Energy (DOE), whichreplaced the Energy Research and DevelopmentAdministration (ERDA). The JPLlFSA became verymuch involved through many technical interfaceswith all of the DOE Laboratories in order to carry outthe national mandate.

Perhaps the most important strategy in ensuringtechnology transfer has been to directly involve therecipient photovoltaics community. The single, mostprominent effort in the technology transfer strategy

has been the conduct of 25 Project IntegrationMeetings (PIMs) over the 10-year life of the Project.The PI Ms have been conducted to enable the rapidexchange of data between all FSA participants,assess recent progress, gain perspective on trendsand new developments, and to guide the FSAProject's planning and priority adjustment. Thecommunication has been open between the Federalprogram and the PV community at all levels of detailregarding problem-solving and choice-making. Ateach PIM, contractors presented their achievementsand problems encountered since the previous PIM.These presentations to their peers and the ensuingdiscussions stimulated extensive technicalexchanges and progress. DOE managementparticipated in every PIM, lending the guidanceneeded for credible technology transfer.

Changes in the overall political environmentresulted in the FSA reducing its scope of activitiesstarting in 1981. The DOE thrust became lessoriented toward the direct development of the U.S.photovoltaics industry infrastructure and more towardthe research into long-term, high-risk, high-payofftechnical options that industry would not likelyundertake themselves.

Continuing under the new charter, FSA developedthe idea of Research Forums on specific technicalproblems of special difficulty wherein a workinggroup of experts both within and outside the PVcommunity would convene. Of the many such forumsheld to date, everyone has resulted in new ideasand contractual relationships to help solve theproblem.

The legacy of the FSA Project consists ofsignificant measurable technical and economicachievements. An equally enduring part of the legacymay well turn out to be the way in which the Projectwas able to work both effectively and efficiently withthe PV community to achieve the progress to date.

The FSA organization has been changedthroughout the years to meet the programmaticneeds and has evolved to its present form as shownbelow in the Project Organiza.tion Chart.

FLAT·PLATE SOLAR ARRAYPROJECT

W. T. Callaghan, ManagerD. G. Tustin, Admin. Manager

II I I I

PROCESS MATERIALS RELIABILITY AND PROJECT ANALYSISENGINEERING

DEVELOPMENT AND DEVICESSCIENCES

AND INTEGRATION

D. 8. Bickler, M. H. Leipold,A. G. Ross,

P. A. McGuire.Manager Manager Manager

Manager

II I I I

SILICON SILICON SHEET DEVICE RESEARCH WEB TEAMMATERIALS

A. O. Morrison, A. D. Morrison, A. H. Kachare, R. R. McDonald.Manager Manager Manager Manager

Iv

ContentsProject Overview . . . . . .

Project Technical Thrusts.

FSA Project Technologies

Project Analyses and Integration

Silicon Material

Silicon Sheet .

High-Efficiency Solar Cells

Process Research and Development

Modules and Arrays

Module and Array Evolution

Engineering Sciences and Design

Encapsulation

Reliability Physics

Module Performance and Failure Analysis.

Appendix A: Glossary .

Appendix B: FSA Project Contractors

...........1

. 2

................... 5

.7

.12

.23

.38

.42

.53

.58

.66

.70

.77

.82

.88

.91

Acknowledgment

The progress of the past 10 years in crystallinesilicon PV technology (non-eoncentrating) has beenaccomplished only because there has been a strongmotivation to work on a very challenging andworthwhile activity in which many people believe.Without the sustaining support of the FederalGovernment and cost sharing by private industry,the excellent achievements across the completebroad scope of activities would not have progressedso rapidly. The thousands of people in many diverselarge and small organizations deserve credit for theexcellent work performed individually and as a team.This report, which summarizes 10 years of FSAactivities, was possible only with the help of manyother members of the FSA Project.

Prepared by the Jet Propulsion Laboratory, California Institute of Technology,for the U.S. Department of Energy through an agreement with the NationalAeronautics and Space Administration.

The JPL Flat-Plate Solar Array Project is sponsored by the U.S. Department ofEnergy and is part of the Photovoltaic Energy Systems Program to initiate amajor effort toward the development of cost-competitive solar arrays.

This report was prepared as an account of work sponsored by an agency of theUnited States Government. Neither the United States Government nor anyagency thereof, nor any of their employees, makes any warranty, express orimplied, or assumes any legal liability or responsibility for the accuracy I com­p~eteness, or usefulness of any information, apparatus, product, or processdisclosed, or represents "that its use would not infringe privately owned rights.

Reference herein to any specific commercial product, process, or service by tradename, trademark, manufacturer, or otherwise, does not necessarily constitute orimply its endorsement, recommendation, or favoring by the United StatesGovernment or any agency thereof. The views and opinions of authorsexpressed herein do not necessarily state or reflect those of the United StatesGovernment or any agency thereof.

This publication reports on work done under NASA Task RE-152, Amendment66, DOE / NASA IAA No. DE-AIOI-76ET20356.

Project Overview

Objectives

• To develop the flat-plate PV array technologiesrequired for large-scale terrestrial use late in the1980s and in the 1990s:

• To advance crystalline silicon PV technologies.

• To develop the technologies required to convertthin-film PV research results into viable moduleand array technology.

• To stimulate transfer of knowledge of advanced PVmaterials, solar cells, modules, and arrays to thePV community.

• Develop module and array specifications,develop performance and lifetime criteria, andassess contemporary module and arraytechnologies.

• Continue to fund and cost-share efforts withindustry, universities, and other institutions.

• Continue economic analyses to evaluatetechnical options and overall progress.

The definition of a solar cell, module and array isshown in Figure 1.

Plans DIRECT SUNLIGHT ON A SlllCONSOLAR CELL {10 em OR 4 in, DiAl

SOLAR CELL

TRANSPARENTSEAL

METALLIC CONTACTSTOP AND BOTTOM

-"-li~iiiiiiii-i~"'--' PRODUCES ABOUT1 WATT OF DIRECTCURRENT ELECTRICALPOWER 1::::112 volt,2amporosl

• Perform advanced solar cell, module, and arraytechnology R&D that the private sector can useto establish economical central station androoftop applications, competitive with large­system power-generation costs of $0.15/kWh inthe 1990s.

• To sponsor or cost-share efforts that industryreasonably would not fund because of high risksand/or long-time requirements:

• Emphasize efforts to reduce technical barriersto the development of high-performance,economical, and long-life solar cells, modules,and arrays.

• Establish the technologies for a low-cost,15 % efficient, 30-year-life, crystalline-siliconflat-plate module.

• In cooperation with SERI, establish thetechnology for a low-cost, 12% efficient, long­life, thin-film module.

MODULE - MANY SOLAR CElLSIN A WEATHERPROOF UNIT

ARRAY - MANY MODULES ELECTRICALLYAND PHYSICALLY CONNECTED TOGETHER

Figure 1. Definitions

Project Technical ThrustsThe three flat-plate-array technical thrusts now

required to meet the DOE Program goal of PV powersystem generations costs of $O.15/kWh are:

• High efficiency.

• Low cost.

• Reliability.

To achieve the overall goal, individual goals have

been set for each thrust and for specific contributingefforts; the contribution and potential contribution ofindividual efforts are periodically assessed, as is thecombined contribution toward each thrust. The Projectas a whole contributes to each thrust; the synergistic,integrated value of that contribution is important, andis likewise assessed. The goals of each thrust are theresult of years of integrated assessments of userneeds to meet competition and the potential of eachtechnology to meet the requirements. These effortsare all interrelated.

Ob"jectlves:EI Lower Cost Encapsulation Module/ArrayEI Higher Efficiency Materials & DesignEI Higher Reliability Processes Technology

J ~ tSilicon Silicon Sheet Process Advanced Prototype

Refinement f--Formation Technology ModulesEI Ribbon grown to - EI Sheet to solar

Processes EI JPL specificationsdesired thickness cells - EI Designs by Industry

EI Mutual design review~ AND

Material & Cell EI Cells assembled EI Fabrication by Industry

Characterization l- into modules& Enhancement t

t t t t t t t

Project Analysis and Integration TestEI Price allocation guidelines EI PerformanceEI Economic analyses -- EI Durability/ReliabilityEI Probability of success Failure AnalysisEI Industry assessment EI JPL tests

EI Applications

--- Information Flow, Both Directions Results to Industry

2

High Efficiency: Crystalline Silicon

Ell Factors contributing to more efficient solar cells,modules, and arrays:

ED High-purity silicon material.

ED High-quality silicon sheet as grown ordeposited.

ED Advanced solar cell designs.

ED Advanced module and array designs.

ED Advanced cell processes.

ED Advanced module processes.

Ell R&D is required to:

ED Understand the influence on efficiency of:

• Silicon growth-related defects including:

• Impurities.

• Structural defects.

• Advanced solar cell designs.

• Advanced cell processing.

ED Develop advanced silicon sheet growthcapabilities for high quality sheet that is alsoinexpensive.

ED Develop advanced material and solar cellproperty measuring techniques andequipment.

ED Develop advanced solar cell designs andfabrication processes.

ED Develop advanced module and array designs.

High Efficiency: Thin-Film Solar Cells

Ell Thin-film solar cells consist of thin layers ofsemiconductor materials deposited on asubstrate so that the cell can absorb andconvert usable light to electricity in a smallthickness. The potential for depositing cells andcell interconnections in a continuous productionline holds promise for low-cost thin-film modulesthat require small amounts of semiconductormaterials.

Ell Achievement of higher-efficiency cells andmodules is the key to widespread use of thin­film technology for PV power generation.

• Basic research on materials and devices leadingto higher performance is being sponsored and/orperformed by SERI for DOE, and by the privatesector.

3

• Translation of thin-film research and the blendingof crystalline silicon PV experience intocommercial manufacturing technologies for thin­film PV power units is now being accelerated byindustry and by FSA in support of SERI. FSA isinvolved in:

ED Developing thin-film cell and module processappropriate for large-area modules.

ED Developing and verifying module and arraytechnologies for reliable units.

Low Cost: Crystalline Silicon

• Factors contributing to low-cost solar cells,modules, and arrays are:

ED Low-cost, high-purity silicon semiconductormaterial.

ED Rapid growth or deposition of silicon in sheetform of a quality consistent with cost-efficientsolar cells.

ED Efficient solar-cell designs that areinexpensive to fabricate.

ED High-volume cell processes using inexpensivematerials that have high yields of cost­effective solar cells.

ED Efficient module designs using inexpensivematerials that are easily mass-produced withhigh yields of long-life, reliable modules.

• R&D is required to achieve:

ED Completion of silicon refinement research.

ED Cost-effective silicon sheet formation resultingin high-quality, inexpensive sheet.

ED High-efficiency solar cell designs for cost­effective processing and performance.

ED Mass-production processes with high yields ofhigh-efficiency, inexpensive solar cells andmodules.

ED Verification of long-life effectiveness ofmodule and encapsulation materials.

Low Cost: Thin-Film Solar Cells

Ell Comprehensive manufacturing technologyefforts have not been accomplished as yetbecause material, device, and basic cell andmodule designs are still evolving.

ED A thorough assessment and subsequent effortsare required when the research results aredefinitive.

Reliability: Crystalline Silicon

Module and array reliability means consistent highperformance for 30 years.

e Factors contributing to high reliability are:

e Stable, long-life materials.

e Compatible solar cell and module materials.

e Appropriate cell and module designs.

e Suitable cell and module processes.

• High quality of manufacturing workmanship.

e Capability of integration into PV powersystems and installation.

e Types of reliability or performance degradationare:

e Infant mortality.

e Wear-out or aging.

e Random failures.

e Environment-induced soiling (weather­variable).

e Causes of degradation are:

• Cell cracking.

e Corrosion.

• Cell-interconnect fatigue.

• Module material aging.

e Module material fatigue.

e Insulation failure.

4

e Soiling or dirt accumulation on face ofmodule.

• Structural failure.

e Design and fabrication flaws.

e Methods of reliability verification are:

e Development of precise performance­measurement capabilities.

e Accelerated testing of materials and modules.

• Module qualification testing.

e Field testing of modules and arrays.

e R&D is required for:

e Advanced module and array specificationdefinitions.

• Material aging and stability research.

e Advanced mOdule and array designtechnology research.

e Solar-cell, module, array, and PV systemcompatibilities.

e Advanced module design verification.

• Correlation of accelerated testing with fieldtesting and with operational results.

Reliability: Thin-Film Solar Cells

e Unknown, evaluation just beginning.

e Most of the crystalline-silicon technology, notedabove, will apply to thin-film hardware, butprobably with unique variations.

FS Project Technologies

SILICON SHEET• SILICON IS GROWN WITH SPECIFIC ELECTRICAL AND

CRYSTALLOGRAPHIC CHARACTERISTICS

• INGOTS ARE SLICED INTO WAFERS

• RIBBONS ARE GROWN TO DESIRED THICKNESS

ACCOMPLISHMENTS

• NEW SEMICONTINUOUS, LARGER CZ INGOTS GROWN• NEW SHAPED·INGOT TECHNOLOGIES EVOLVED• FASTER, MORE EFFICIENT SAWING DEVELOPED• GROWTH OF WIDE, FLAT·RIBBON ACHIEVED

RESEARCH REQUIRED• INVESTIGATE CRYSTALLIZATION RATE LIMITATIONS

AND INFLUENCE OF PRIMARY GROWTH PARAMETERS• CONTINUE CHARACTERIZATION OF SILICON SHEET

MATERIAL VERSUS GROWTH RATES• CONTINUE BASIC INVESTIGATION OF HIGH·QUALlTY,

RAPID CRYSTAL GROWTH; ESPECIALLY RIBBON GROWTH

RIBBONS

INGOT

WAFER

NEW SILICON REFINEMENT PROCESSESACCOMPLISHMENTS

• LARGE·SCALE SILANE PRODUCTION PLANT BY UCC IN OPERATION• PRACTICAL CONVERSION OF SILANE TO SILICON IN FBR DEMONSTRATED• OTHER PROCESS AND PROCESS IMPROVEMENTS ADOPTED BY INDUSTRY

RESEARCH REQUIRED

• ESTABLISH PURITY OF SILICON FROM UCC PROCESS• VERIFY ECONOMICS OF PROCESS

SILICON MATERIAL• QUARTZ (SI02lIS REFINED BY REMOVING 02 AND IMPURITIES• THE PRODUCT IS VERY HIGH PURITY SILICON

HIGH·EFFICIENCY SOLAR CELLSSOLAR CELL EFFICIENCY IS A FUNCTION OF:• SILICON SHEET QUALITY• CELL DESIGN• CELL FABRICATION PROCESSES

UNIFORMITY AND REPEATABILITY OF ABOVE

25 PROBABLE MAXIMUM ACHIEVABLE_

CELL AND MODULE PROCESS RESEARCH AND DEVELOPMENT• SOLAR CELLS ARE FABRICATED FROM SILICON SHEET

CELLS ARE ELECTRICALLY INTERCONNECTED AND ENCAPSULATEDIN A SUPPORTING STRUCTURE

o L19:-:7::5---------,-:Y=-EA,.-R::---------,-19=-=a,.,J5

eft 20>-()

Z

~ 15

u::u.UJ...J...JUJ()

LABORATORY CELLS_-------..._~'---------l

ACCOMPLISHMENTS

• KEY BARRIERS TO HIGHER·EFFICIENCYSOLAR CELLS IDENTIFIED

• HIGH·EFFICIENCY SOLAR CELLS MODELEDAND ANALYZED

• NEW SOLAR CELL CHARACTERIZATIONTECHNIQUES DEVELOPED

RESEARCH REQUIRED

• CONTINUE INVESTIGATION OF LOSS MECHANISMSWITHIN SOLAR CELLS

• CONTINUE DEVELOPMENT OF SOLAR CELL CHARACTERIZATIONANALYSIS AND MEASUREMENT TECHNIQUESCONTINUE RESEARCH CELL DEVELOPMENT AND TESTING

ACCOMPLISHMENTS

• IMPROVED CELL SURFACE TREATMENTS, JUNCTION FORMATION, AND METALLIZATIONHAVE RESULTED IN LESS EXPENSIVE YET MORE EFFICIENT SOLAR CELLS

• SIMPLIFIED FABRICATION SEQUENCES SUITABLE FOR AUTOMATION DEVELOPED• PROTOTYPE MASS·PRODUCTION EQUIPMENT DEVELOPED

RESEARCH REQUIRED

• INVESTIGATE HIGH·EFFICIENCY SOLAR CELL PROCESSING• INVESTIGATE ELECTRICALLY CONDUCTIVE SILICIDES, CORROSIVE

EFFECTS ON METALLIZATION, AND THE EFFECTS ON JUNCTION FORMATION• CONTINUE INVESTIGATING CRITICAL PROBLEMS OF ION IMPLANTATION,

METALLIZATION, INTERCONNECTIONS, AND MODULE ASSEMBLY

5

MODULE/ARRAY ENGINEERING SCIENCES AND DESIGNDESIGN AND TEST METHODS HAVE BEEN DEVELOPED AND USED TO IMPROVETHE PERFORANCE, ENVIRONMENTAL DURABILITY, AND SAFETY OFPHOTOVOLTAIC MODULES AND ARRAYS

FIRE TESTOF MODULES

ARRAYSTRUCTURETEST

ACCOMPLISHMENTS AND ONGOING ACTIVITIES

• MODULE AND ARRAY DESIGN REQUIREMENTS GENERATED FOR FUTURELARGE·SCALE APPLICATIONS

• IMPROVED DESIGN CONCEPTS, ANALYSIS, AND TEST METHODSAVAILABLE AND UNDER DEVELOPMENT (CIRCUIT DESIGN, SAFETY,STRUCTURES, RELIABILITY, ETC.)

• PERFORMANCE EVALUATION METHODS AND DESIGN AND TESTSPECIFICATIONS DEVELOPED THROUGH INTERACTION WITH INDUSTRY

RESEARCH REQUIRED

• PERFORM ABOVE FUNCTIONS FOR MODULESMADE FROM THIN·FILM SOLAR CELLS

RELIABILITY PHYSICSPOWER GENERATION WITHOUT DEGRADATION REQUIRES:

• STABLE, LONG·L1FE MATERIALS• GOOD, COMPATIBLE CELL, MODULE, ARRAY AND PV SYSTEM MATERIALS

AND DESIGN• GOOD FABRICATION AND INSTALLATION WORKMANSHIP

ENCAPSULATION

SOLAR CELLS AND INTERCONNECTS ARE PROTECTED FROMTHE ENVIRONMENTS

TRANSPARENT COVER--

SOLAR CELLS ANDELECTRICALINTERCONNECTIONS

BACK COVER----~••••••••II,..ACCOMPLISHMENTS

• EXPERIENCE INDICATES 30·YEAR MODULE LIFE IS ACHIEVABLE• SELECTED MATERIALS AND PROCESSES EXPECTED TO MEET GOALS

AND AUTOMATED PROCESSING NEEDS• MATERIALS INDUSTRY RESPONDING TO NEEDS AS DEFINED BY FSA

RESEARCH REQUIRED

• CONTINUE INVESTIGATION OF ENCAPSULATION MATERIALS TOACHIEVE LONG·TERM PHOTOTHERMAL STABILITY

• CONTINUE DEVELOPMENT AND USE OF ACCELERATED AND LONG·LIFE,DURABILlTY·TESTING TECHNIQUES

• CONTINUE INVESTIGATION OF ENCAPSULANT, METALLIZATION, ANDOTHER INTERFACIAL BONDING STABILITY AND CORROSION CRITERIAAND MECHANISMS

MODULE PERFORMANCE AND FAILURE ANALYSES• PHOTOTYPE ADVANCED MODULES HAVE BEEN DESIGNED AND MANUFACTURED

BY INDUSTRY TO JPL SPECIFICATIONS• MODULE ELECTRICAL PERFORMANCE AND ENVIRONMENTAL DURABILITY

ASSESSMENT BY JPL

(J)

ex:«UJ>-

1975 1985

ACCOMPLISHMENTS

• DEVELOPED ANALYSIS TECHNIQUES FOR EVALUATING PARAMETRICINFLUENCES ON RELIABILITY

• DEVELOPED MANY TEST TECHNIQUES AND EQUIPMENT FOREVALUATING MODULE RELIABILITY

• MODULE LIFE·TIMES HAVE INCREASED DRAMATICALLY

RESEARCH REQUIRED

• CORRELATE FUTURE FIELD OPERATIONAL RESULTS WITH EXISTINGDATA

• EVALUATE AND IMPROVE THIN·FILM MODULE RELIABILITY

PROGRESS IN MODULE PERFORMANCE

• MODULE PERFORMANCE AND DURABILITY IMPROVEMENT WITH EACHNEW PROCUREMENT

• LATEST MODULES (5th PROCUREMENT) INCORPORATE INNOVATIVEDESIGN FEATURES

• CONTINUING CHARACTERIZATION AND DIAGNOSTIC ANALYSES OFMODULE DEFICIENCIES/FAILURES ARE TRANSFERRED TO INDUSTRY

PROJECT ANALYSIS AND INTEGRATION• DEVELOPED PV ANALYSIS METHODS, ESPECIALLY FOR

MANUFACTURING COSTS• ESTABLISHED GOALS INTERNAL TO FSA AND MEASURED TECHNICAL

PROGRESS BY USE OF ANALYSIS TECHNIQUES

6

Project Analysisand Integration

Those responsible for initiating and organizing theFSA Project recognized the need for coordination

. within the Project and with private industry toachieve the objectives established for the Project. Tothat end, the Project Analysis and Integration (PA&I)effort has served as a channel of communication forProject activities, private industry, and the U.S.Department of Energy.

PA&I performs planning and integration activitiesto coordinate the various R&D tasks. Goals areestablished for Project R&D activities that reflect theobjectives of the National Program. Analyticalsupport is given to the Project Office, providing achannel for the integration of individual task work intothe overall Project. Integration activities have alsosupported the coordination of research activitiesbetween the Project and other research laboratories.

PA&I performs analytical studies to measure theprogress toward accomplishing Project goals and todetermine which R&D tasks are the most likely toresult in significant performance and costimprovements. Guidelines for module productioncosts are established by PA&I which are consistentwith the National Program goals. Performance andcost criteria are set for separate module elements,creating objectives for the individual R&D tasks.Sensitivity studies are performed to see howtechnical advances in the Project and changes in theeconomic environment affect Project objectives.

Background

FSA research and development has encompassedall phases of flat-plate module technology, from thebasic feedstock materials for the photovoltaicmanufacturing processes, through vehfication ofmodule reliability and performance in the field. Whenthe Project was organized in 1975, it was recognizedthat this broad spectrum of activities would requirethe Project to be somewhat unique at JPL becauseof the number of industrial organizations, Governmentlaboratories, academic institutions, and other entities,such as utilities, with which JPL would have tointeract. It was apparent at the outset thatcommunication between JPL and the otherparticipating organizations would be an importantfactor in the success of the effort. In addition, theProject was one of the first, and certainly the largestat JPL, for which technical performance was not theultimate goal but only a contributing factor to theprincipal measure of success,economicperformance. Thus, it was also apparent thatanalytical methods would be needed to translatetechnical performance into future economicperformance. The PA&I Task was formed as part of

7

the original Project structure to address the problemof communication and to help guide the Projecttechnical developments toward the National PVProgram's economic goals.

Project Activities

In 1975, analytical capabilities were needed toevaluate competing technologies and establishrealistic guidelines for their development. Initially,cost goals were allocated to silicon material, siliconsheet, encapsulation materials, cell fabricationprocesses, and module assembly. As economicanalysis capabilities were developed, therebyallowing Project-wide technology assessments to bemade, these goal allocations were revised. Thetechnical objectives that were established were usedto determine the division of resources throughout theProject. As time passed, the completeness andsophistication of the analysis capabilities matured.

The assessing of technology status and thetracking of progress has been one of the principalPA&I activities throughout the duration of the FSAProject. It has established a set of standards bywhich progress can be measured. The standards arepart of an overall methodology developed by PA&I sothat research findings could be reported in acommon format. The methodologies are being usedto track and assess progress within the Project so asto derive the maximum technical and economicbenefits from research and development work.Technical reports and research activity assessmentsare an integral part of the overall Project effort.

The FSA Project participation in the early efforts ofDOE to establish a National PV Program Plan wascoordinated and implemented by PAM It authoredseveral major sections of the first DOE ProgramPlan. The planning and organization of thePhotovoltaic Technology Development andApplications Lead Center at JPL in 1978 was a majorPA&I activity at that time.

Today's Status

PA&I provides ongoing support to the Project inthe form of up-to-date technology assessments.ManUfacturing technology descriptions are revised toreflect new technical developments,and newmanufacturing cost estimates are made using theSAMICS .(Solar Array Manufacturing Industry CostingStandards) methodology. Examples of this analysiscapability are the 1985 production cost estimateupdates for Czochralski (Cz) and dendritic web siliconsolar cell modules.

Module cost estimates were made for state-of-the­art ,Cz silicon solar cell technology, assuming theconstruction of a 25-MW-per-year production facilityand current material costs. Assuming 3 shifts perday, full-time operation and module efficiencies of13.5%, production prices were estimated at $1.47 per

peak watt (1985) dollars. This represents significantprogress over early assessments for this technology.

Module cost estimates were made for state-of-the­art, dendritic web silicon solar cell technology,assuming three shifts per day, full-time operation at a25-MW-per-year production rate and current materialcosts. Assuming module efficiencies of 13.7 %, pro­duction prices were estimated at $1 .48/Wp or more.This is based upon a web growth rate of 10 cm2/minor less (today's best growth rate).

The dendritic web solar cell technology, while stillunder active development, promises to be a mostattractive photovoltaic option. The principalimpediment to economic feasibility of dendritic webhas been the low crystal growth rates. If thetechnology development presently in progress issuccessful in increasing the dendritic web growthrates to 30 cm2/min, then the production price canbe reduced to $0.86/Wp (1985 dollars). If modules atthis price are placed in a PV electrical· generationplant, electricity could be produced at $0.17/kWh,very close to the DOE Program goal of $0.15/kWh.

Accomplishments

PA&I supports a systems analysis capability forevaluating developments within the Project andassessing progress. The present status of thiscapability and its accomplishments are bestsummarized in terms of the methodologies oranalytical tools maintained for Project analysis andintegration efforts. These analytical tools haveprovided a consistent and reliable framework formeasuring developments throughout the Project.

SAMICS

One of the most important analytical toolsdeveloped and supported through PA&I is the SolarArray Manufacturing Industry Costing Standards(SAMICS). SAMICS has proven itself as a crediblecosting methodology for an emerging technology.The SAMICS methodology provides a standardizedway for comparing cost estimates of manycompeting technical options.

SAMICS has been used repeatedly to assess thecost of photovoltaic module production at outputlevels representative of commercial ventures usingstate-of-the-art technology. SAMICS has also beenused for projections of what might reasonably beexpected at some future date for the technologygiven successful completion of research activities.These snapshots of technology development inprogress have provided FSA valuable insights intothe likelihood of achieving desired reductions in thecost of photovoltaic module production.

8

The SAMICS methodology is a complete procedurefor evaluating the cost of manufacturing PV solararrays. The SAMICS methodology includes:

• Computer programs designed for userimplementation on personal computers (SAMISPC, SAMPEG, IPEG-4).

• Formats for the preparation of completemanufacturing process descriptions.

• A detailed cost account catalog which caneasily be modified to meet user requirements.

• A standard set of financial assumptions makingdirect technology comparisons possible.

• A report format that provides detailed valueadded estimates for all process steps.

• The option to vary any of the technical,financial, or overhead parameters.

A series of competitive procurements, designatedBlock I through Block V, was initiated by FSA toencourage the adoption of new technology andstimulate reduction in the cost of photovoltaicmodule production. SAMICS has made it possible toassess the technical progress made. Detailedprocess descriptions reflecting the latest technologydevelopments were converted into manufacturingcost estimates under commercial productionconditions. These were then compared with previousblock purchases and the cost goals of the NationalProgram.

SIMRAND

Analytical tools. have been developed which weredesigned specifically for support of programmanagement activities. The FSA Project mustcontinually make decisions regarding support levelsfot various projects. These decisions are necessarilybased on information that is quite uncertain innature. The SIMRAND (Simulation of Research andDevelopment Projects) model was developed as atool that allows this uncertainty to be directlyincorporated into the information used by decision­makers.

The SIMRAND model assesses the probability thata combination of R&D activities will meet a given setof price goals. The descriptions developed bySIMRAND for each technology reflect the uncertainnature of the database used. Technologies can thenbe compared to see which has the highest probabilityof reaching Project goals. The decision process ismodeled by means of alternative networks thatrepresent the feasible technologies being considered.SIMRAND performs repeated simulations to

0.0175 MAINT. MECH.

O.026kWhlmln

400-cufllmlnFUMES

1.191E-4IblmInSOLDER

5.24cull/mlnCOMPRESSED AIR

24 CONNECTORS/min

REQUIRED PRODUCT:

TESTe

99.9% VIELD

ISTR1NG/9CELlS

JPL-+FORMAT A

CELLSTANG

PROOUCT:

STRINGS/min

ECONOMIC ANALYSESSAMICS - THE SOLAR ARRAY

MANUFACTURING INDUSTRY COSTING STANDARDS

SUM OFDIRECTWORI<

STATION REQTS

INTERNAL DATA BASE

9

SAMICS OUTPUTSDESCRIPTION OF

SIMULATED FACTORY(INCLUDING FINANCES)

A FACTORY lOR SEQUENCEOF FACTORIESllS BEINGBUILT, STAFFED, ANDOPERATED IN THE COM­PUTER; THE SIMULAliON

CONCENTRATES ON THEINDIRECT REQUIREMENTS(OVERHEADl AND THE

OTHER ECONOMIC ASPECTSOF THE COMPANY (OR

COMPANIES

determine the likely distribution of outcomes for eachtechnology. The network reflecting the varioustechnology options is continually reevaluated duringthe simulation to find the optimum subset of researchtasks.

SIMRAND is a management decision methodologythat incorporates information uncertainty. TheSIMRAND model (Figure 2) includes:

• The capability of evaluating relatively complexnetworks of technical options representing thefeasible choices.

• The ability to replace process parameters withprobability distributions reflecting informationuncertainty.

The SAMICS and SIMRAND models and theAllocation Guidelines have been applied individuallyto specific technical problems and as an integratedset when longer range management decisions havebeen required. Technical options have beenevaluated by applying the manufacturing costsimulations to processing sequences and comparingthe results with the cost-goal allocations. Manufac­turing cost point estimates generated by SAMICShave been combined by using SIMRAND with theuncertainties associated with those estimates to formprobability distributions of technical performanceand/or cost. The inclusion of uncertainty has allowedProject management insight into the risks associatedwith the technical options being considered.

LCP

• Computer program available for implementationon a personal computer.

• The option to include decision-maker'spreferences regarding risk into the decisionprocess.

Figure 2. Module Production Network

Assessment of a PV systems capability tocompete with conventional energy resources requiressimulation of its performance over its lifetime andrecognition of regional variations in solar energyresources. The Lifetime Cost and Performance (LCP)model was developed and is now used to simulatethe energy output of a PV power plant over its usefullife. LCP relates the effect of hourly weather condi­tions, system design and component efficiencies,electrical design, long run effects of exposure, andalternative operation and maintenance strategies tosystem cost, performance, and value (Figure 3). Fixedarray, one-axis tracking and two-axis tracking systemscan be evaluated at different locations throughout thecountry.

LCP simulates the performance of PV systemsusing actual solar resource measurements asrecorded in weather data tapes. The LCP modelincludes:

• The ability to use insolation data, temperaturemeasurements, and precipitation and dirtaccumulation measurements to model locationspecific performance.

STEP 5

MOOUlECEll

STEP 4STEP 3

SAW1N~CRYSTAlUZATlON

I STEPZ ISTEP 1

SILICONPURIFICATION

A joint JPLlSERI silicon sheet material studysuccessfully used SIMRAND to incorporateuncertainty about potential technical progress andmaterial costs into the outlook for each silicon sheettechnology. The product was a set of distributionsindicating the possible range of outcomes for eachresearch activity. The study results were instrumentalin redirecting the technology mix pursued by theNational PV Program.

• The ability to evaluate the effects of balance ofsystem component failures and electricalmismatches within modules.

• A computer model that can be implemented ona personal computer.

LCP, which was used to assist Acurex Corp. inmaking economic assessments of alternative designs

10

for the first phase of the Sacramento Municipal UtilityDistricts proposed 100-MW PV installation. The perfor­mance of each design alternative was simulated overits lifetime using weather data characteristic of thelocation and the parameters of the PV array system.

from a variety of causes have been assessed.PVARRAY can model the effect of specific solar cellfailure mechanisms on system performance and cor­rect for various failed module replacement strategies.

PVARRAY simulates the power output of a PVarray over its lifetime. The PVARRAY model includes:

• Computer program that can be implemented ona personal computer.

• The capacity to consider alternative seriesparallel wiring schemes.

• The ability to evaluate alternative replacementstrategies for different cell failure rates anddiode placements.

The ability to compare performance of differentmodule designs.

•

The combination of PVARRAY, SAMICS, and LCPhas been used at JPL to estimate the net presentvalue of energy from a photovoltaic system over itsuseful lifetime. Manufacturing costs for the moduleproduction sequence have been estimated usingSAMICS, and the life-cycle cost and performancehave been simulated using the LCP and PVARRAYmodels. The three programs have allowed PA&I tocalculate the cost of different module designs anddetermine the time-d~pendent economic impact ofdesign changes by simulating array performance andlife-cycle cost (Figure 4).

JULY 1

Figure 3. Time-of-Day Output and Customer LoadProfile

PVARRAY

The PVARRAY model simulates array performanceover time. Using PVARRAY, the power profiles ofmodules representing various state-of-the-art andadvanced designs have been obtained and theeffects of the current-voltage (I-V) changes resulting

0 . I;, I I I I , I I I I I I I '--'.... ,,,.. .... "" ,.,. "u, "" ..... .... 100._ tMi - n" '0" '01" 'OIl n .., [ • • 20161986 1996 2006

Figure 4. Cell Degradation and Lifetime Performance for Three Different Module Designs

11

Silicon MaterialSemiconductor-grade silicon is the material used to

fabricate crystalline-silicon solar cells, most solid-statedevices, and most integrated circuits. It is very highlypurified, with tight specifications limiting theconcentrations of specific impurities. Themanufacturing yields during the production of solid­state devices and the performance of the devices aredirectly related to the quality of the silicon.Semiconductor-grade silicon, as produced in theSiemens reactor for the semiconductor industry, hasbeen and continues to be used by the PV industry forthe production of the great majority of the solar cells.The cost of the silicon is a minor part of the cost ofproducing semiconductor devices, but it has been andstill is a significant fraction of the cost of manufacturingsolar cells. Integrated circuits are very small and manycan be fabricated on one silicon wafer, but only onesolar cell is fabricated from a wafer. The small marketprovided by solar-cell users has not justified industry'sdevelopment of an inexpensive high-volume siliconproduction technique that would result in low-costsilicon material. Considerable FSA funding has beendevoted toward advancing silicon purificationtechnology and processes.

The objective of the silicon refinement effort is toovercome the technical Qarriers to low-cost siliconpurification in order to achieve a process capable ofproducing high-quality, low-cost silicon (appropriate forphotovoltaic use), at a price consistent with FSAobjectives.

The goals are to establish silicon material processesappropriate to:

• Purity adequate for efficient crystalline-siliconsolar cells that meet the requirements of the FSAProject.

• Market price of $19/kg (1985 dollars).

• High-volume production plants «1000 MT/year).

Background

Silicon, the second most abundant element in theEarth's crust, occurs naturally primarily as silicondioxide and as silicates: sand, quartz, etc. Largequantities of quartzite, a crystalline form of silicondioxide, are reduced in arc furnaces to metallurgical­grade silicon (mg-Si) for use by man'1 industries. It sellsfor $1 to $2/kg and is about 98 to 99% silicon. Amuch purer silicon is required to fabricatesemiconductor devices and solar cells. A small portion(a few thousand metric tons) of mg-Si is used annuallyin processes to produce semiconductor-grade silicon.

Most solar cells have been and still are fabricatedfrom silicon with a total impurity concentration less

12

than one part per million, which possessessemiconductor properties. In 1975, the availablematerial of this quality was produced by the Siemensprocess and was referred to as semiconductor-gradesilicon. The price then was greater than $50/kgandwas very sensitive to the supply-and-demandfluctuations. It was known that solar cells could bemade from a less-pure silicon. How much less pure,with which impurities, and how much of each impuritycould be tolerated, was unknown, nor was such amaterial commercially available.

Project Activities:1975T01981

The status of the Siemens process for theproduction of semiconductor grade silicon was:

Advantages

High purity (more than adequate at that time)

Acceptable form

Proven process

Potential for improvements:

Decreased energy use

Increased production rate

Disadvantages

High-energy consumption process

Low conversion rate and yield

Large amount of waste by-product

High labor costs

High product cost

A plan was formulated within FSA to develop asilicon refinement process to meet PV sheet and solarcell requirements. The plan was to develop low-costsemiconductor-grade silicon processes; to develop asolar-grade silicon refinement process (a less-pure, butunspecified material that should be even lessexpensive); and to determine the effects of variousimpurities and their concentrations on siliconrefinement processes and on solar cell performance.The plan consisted of three process developmentphases for both the semiconductor-grade and the solar­grade process (Table 1). The intent was to establishthe practicality of the processes by production, energyconsumption,and economic criteria. To classifycandidate processes and define the processspecifications, information was required about theeffects of specific impurities on silicon and on solar cell

Phase

Laboratory-ScaleExperiments

Process DevelopmentUnits (PDUs)

Experimental ProcessSystem DevelopmentUnits (EPSDUs)

Table 1. Planned Process Research Phases

Purpose

Demonstrate process feasibility

Determine reaction kinetics and yields

Identify reactor-material problems

Obtain silicon deposition rates of 0.01 to 1 kglh

Obtain product purity information

Demonstrate practicality of reactors and critical process steps

Obtain reactor scale-up information

Characterize critical process components

Establish suitability of reactor materials

Determine process capability by chemical engineering analyses using acquireddata and extended process design

Perform preliminary economic analyses

Establish technical readiness of integrated processes

Design and build pilot plant EPSDUs with silicon deposition rates of up to100 MT/year

Obtain operating data to characterize complete process including steady-stateoperation

Obtain data for optimization of design and operation of a full-scale productionplant

Confirm product purity at steady-state conditions

Confirm economic analyses

performance. These purity requirements directlyaffected the process designs and, thus, the economicsof the candidate processes. The impurity studiesconsisted of two phases: the measurements of theeffects on the final product of impurities that are foundin mg-Si or that might be introduced during a processstep, and the determination of the interrelationships ofimpurities and the processing steps and their combinedeffects on solar cell performance.

The feasibility of 11 potentially low-eost processeswere investigated. Their technical feasibilities wereassessed by initially studying basic chemical reactionsand general chemical engineering designs, which led toan analysis of the chemistry and chemical engineering

13

parameters based upon data from laboratory-scaleexperiments. After an investigation lasting 1 to 3 years,each of the promising specific processes was thenevaluated for its practicality.

Scaled-up reactors were used for chemicalengineering studies to derive data for mass and energybalances, process flows, kinetics, mass transfer,temperature and pressure effects, and operatingcontrols. This information was used in analyses andextrapolations that resulted in improved processconcepts and designs. The processes being examinedin June 1977 and a number of supporting stUdies areshown in Table 2. By this time some processdevelopments had progressed rapidly.

Table 2. Silicon Material Refinement Contractors in 1977

Contractor

Semiconductor-Grade Production Processes

Technology

AeroChem Research Laboratories, Princeton,New Jersey

Battelle Memorial Institute, Columbus,Ohio

Union Carbide, Sistersville; West Virginia

Motorola, Phoenix, Arizona

Silicon halide-alkali metal flames

SiC 14 reduction by Zn using FBR technology

Pyrolysis of SiH4 derived from redistribution ofSiHC 13 (With a feed of metallurgical grade silicon)

Chemical vapor transport and pyrolysis of SiF2generated from SiF4 reaction with metallurgical gradesilicon

Northrop Research, Hawthorne, California

C. T. Sah Associates, Urbana, Illinois

Spectrolab, Inc., Sylmar, California

Solar-Cell-Grade Specifications

Lifetime and diffusion length measurements

Model and studies of effects of impurities on solarcell properties

Solar cell fabrication and analysis

Westinghouse Electric, Pittsburgh, Pennsylvania

Monsanto Research Corp., St. Louis, Missouri

Investigation of effects of impurities on solar cellperformance

Investigation of effects of impurities on solar cellperformance

Solar-Cell-Grade Production Processes

AeroChem Research Laboratories, Princeton,New Jersey

Dow Corning, Hemlock, Michigan

Stanford Research Institute, Menlo Park, California

Schumacher, Oceanside, California

Texas Instruments, Dallas, Texas

Westinghouse Electric, Pittsburgh, Pennsylvania

Use of a nonequilibrium plasma jet

Arc furnace process using purer source materials

Na reduction of SiF4

Reduction of SiHBr3 in an FBR

Carbothermic reduction of Si02 in a plasma

Plasma-are-heater reduction of SiC 14 with Na

Lamar University, Beaumont, Texas

Commercial Potential of Processes

Evaluations of relative commercial potentials ofsilicon-production processes developed under theSilicon Material Task

14

After analytical and experimental studies indicatedthat a process was feasible and potentially economical,and the practicality of its reactors and critical stepswere demonstrated, a comprehensive economicanalysis was made. Professor Carl Yaws, of LamarUniversity, and JPL personnel made these economicanalyses which not only calculated the costs toproduce silicon, but also made comparativeassessments of the progress of each process. Themost technically and economically promisingprocesses were then carefully reviewed before theywere deemed ready for design and development of apilot plant. The verification of a process in a pilot plantwas necessary to validate silicon refinementeconomics and product purity, because of limitedcapabilities in scaling-up chemical processes byanalytical means alone.

Purification processes involving deposition of thesilicon from silane and dichlorosilane wereemphasized. Because these two substances can bepurified relatively easily and, because of their highreactivity, they can be more easily and completelydecomposed or reduced to form silicon than cantrichlorosilane, which is used in the conventionalSiemens process. Research on other processes thatoffered promise for making a less-pure solar-cell-gradesilicon by refinement of metallurgical-grade silicon werecontinued because of the potential for even lower-costsilicon.

A comprehensive study of the effects of impuritiesand their concentrations on semiconductor-gradesilicon and crystalline-solar-cell performance wascarried out by a number of organizations, with themajor focus of this effort at Westinghouse ElectricCorp. A functional empirical model was formulatedbased upon the relationship that impurity atomsprimarily decrease minority carrier recombinationlifetime, reduce the short-circuit current of a cell, andessentially act independently. Experimental dataconfirmed this and provided quantitative relationshipsof the effects of specific elements on cell efficiency.

Project Activities: 1981 to Present

In 1981, the funds available were reducedconsiderably, which resulted in many activities beingdropped and support being given only to the mostpromising, advanced efforts. At about the same time,the then newest PV system cost assessments indicatedthat higher-efficiency solar cells and modules werenecessary to offset the area costs of an array. Analysisindicated that the power generation cost reductionspossible by the use of higher-efficiency solar cells weregreater than cost reductions possible by use of solar­cell-grade silicon. Cell efficiency is directly related tosilicon purity. Therefore, solar-cell-grade silicon seemedto be impractical economically and efforts werereduced. The Union Carbide Corporation silaneprocess and the Hemlock Semiconductor Corp.dichlorosilane process developments were continued.

15

In support of these efforts, a basic investigation ofthe technology of a high-performance silane fluidized­bed reactor (FBR) has been continued at JPL; researchto characterize the aerosol phenomena involved in thegrowth of particles formed by the thermaldecomposition of silane is being performed at theCalifornia Institute of Technology, and an engineeringdevelopment of the silane-to-silicon FBR is beingperformed by Union Carbide Corporation. Continuingstudies within FSA of the influence of impurities insilicon on solar cell performance are described hereunder the heading, "High Efficiency Solar Cells."

Today's Status

• A silane process has been developed by theUnion Carbide Corp. that seems to be capable ofmeeting FSA goals, including price.

• A pilot plant (1 OO-MT/year capacity) hasdemonstrated the following (see Figure 5):

• The silane portion of the Union Carbide Corp.silane process has met FSA goals.

• The silane produced in the first section of theUnion Carbide Corp. process is very highpurity and will be a low-cost material suitablefor amorphous-silicon solar cells and/or otherproducts when and if they become large­quantity products.

• The R&D results of the silane-to-silicondeposition FBRs of the process are good.Research at JPL, which established thefundamental technology for FBR silane-to­silicon pyrolysis, will end in 1985. The FBRdevelopment at Union Carbide Corp., whichhas been sponsored by DOE, will continuewith Union Carbide Corp. funds.

• Silane process production plants (completelyfunded by Union Carbide Corp.) include:

• A 1200-MT/year plant which started operationearly in 1985 converting mg-Si to silane, basedupon DOE-sponsored technology. Komatsuchemical-vapor-deposition (CVD) reactorsconvert silane into high purity silicon at a pricethat is expected to be less than that ofSiemens-produced silicon. It is almost at fullcapacity now (see Figure 6).

• A second 1200-MT/year plant is underconstruction with completion scheduled for1987.

• A third plant of 3000-MT/year capacity is indesign, with completion scheduled late in1988. Use of a FBR is under investigation.

• The plants listed above will significantly increaseworld production of silicon. World capacity toproduce semiconductor-grade silicon was about6000 MT/year at the end of 1984.

Figure 5. A 1OO-MT Silane Process Pilot Plant, Washougal, Washington

Figure 6. A 1200-MT Silane Process Production Plant, Moses Lake, Washington

16

Significant Accomplishments

The silane process developed at Union CarbideCorp. under an FSA contract seems to have thecapability to meet the FSA goal of semiconductor­grade silicon at $19/kg (1985 dollars). This processcomprises two parts: the preparation of pure silane(SiH4) from metallurgical-grade silicon, and thedeposition of high-purity silicon from the silane using aFBR. Operation of a 100-MT/year pilot plant hasverified the technical and economical practicality of thesilane portion of the process. The rate of productionand the purity of the silane from the Union CarbideCorp. pilot plant are consistently high. A silane sampletested by Brookhaven National Laboratory was thepurest silane it had ever tested. Good research anddevelopment results for low-eost pyrolysis of silane tosilicon are coming from the collaborative fluidized-bed­reactor efforts at Union Carbide Corp. and JPL. TheUnion Carbide Corp. effort is determining the steady­state operational conditions for the design of aproduction prototype reactor for silane feedconcentrations of 20 to 25 %. For more cost-effectivepyrolysis, JPL research is being performed at silaneconcentrations up to 100 % with emphasis on 40 to60% concentrations. The problems of reactor clogging,bed slugging, and excessive production of silicon dusthave been solved along with raising the silane feedlimits from 2 to 100% concentration. The Union Car­bide Corp. FBR has successfully run for 66 h at 20%silane. Unresolved problems are suitable silicon seedparticle material, prevention of contamination, productpurity confirmation, and verification of productioneconomics.

The progress of 10 years of effort has resulted inUnion Carbide Corp. building a 1200-MT/year siliconproduction plant (now in production) in which the silaneis deposited as silicon in Komatsu chemical vapordeposition reactors. (These reactors were alsooperated in conjunction with the pilot plant.) The siliconproduced is of very high purity. It is expected that thecost will be less than Siemens-process-producedsilicon, but not as much less as with a FBR.

The dichlorosilane chemical vapor deposition

process of Hemlock Semiconductor Corp. wassuccessfully carried through the demonstration of thepracticality of the reactor and the critical process steps.A PDU was designed, constructed, and tested. ThisPDU was used to investigate the making ofdichlorosilane from trichlorosilane. A pilot plant was notsponsored by DOE/JPL because process economicsindicated a product price less than for the conventionalSiemens process, but greater than that of the silaneprocess. This technology is available for use in aproduction plant and Hemlock may have assimilated itinto their ongoing production of silicon, althoughHemlock has never divulged its commercializationplans to JPL.

Numerous FSA studies also contributed toadvancements in trichlorosilane and modified Siemens­reactor technologies. A thorough examination of themany silicon processes has resulted in significantimprovement in the understanding of many othersilicon refinement technologies. The solar-eell-gradesilicon efforts were stopped because of (1) largebudget reductions, (2) the advanced state of the UnionCarbide Corp. silane process, (3) improvedunderstanding of the need for higher-purity silicon toobtain high cell efficiency, and (4) the economic valueof higher-efficiency solar cells.

A good fundamental understanding of the effects ofimpurities in silicon upon the properties of siliconmaterial and upon the performance of solar cells hasbeen determined by extensive studies, principally bythe Westinghouse R&D Center, and supported byothers.

Silane Process(Union Carbide Corporation)

Preparation of Silane from Metallurgical-GradeSilicon and Conversion of Silane intoSemiconductor-Grade Silicon

The process flow diagram is shown in Figure 7.

SiCI4

SiHCI3

H2IMPURITIES

FLUIDIZEDBEDREACTOR

DISTILLATION

~~XDE L..- S_iC...,:14=-- ---I

Si

Figure 7. Silane Process Flow Diagram

DISTILLATION

SiHCI3

SiH2CI2

SiHCI3

SiHCI3SiCI4SiH2CI2

REDISTRI·BUTIONREACTOR

17

SiH2CI2SiHCI3SiH4

REDISTRI·BUTIONREACTOR

FLUIDIZEDBEDREACTOR

DISTILLATION

SILICONGRANULES

Past Accomplishments

• Process feasibility demonstrated in laboratory­scale experiments for silane production and forsilane conversion to silicon by thermaldecomposition in free-space reactors and FBRs.

• Practicality of reactors and process steps forcontinuous, steady-state silane productiondemonstrated in scaled-up process developmentunits.

• Thermodynamic values and reaction kineticsmeasured in hydrochlorination and redistributionreactors.

• Vapor-liquid equilibria calculated for chlorosilanes.

• Detailed process design for 100-MT/year EPSDUcompleted (silane portion).

• Engineering design and economic analysis for1000-MT/year plant completed (silane portion).

• Hydrochlorination reaction characterized inextended studies (Solarelectronics, Inc.) consistsof:

41 Equilibrium constants and reaction kineticsmeasured over wide temperature and pressureranges.

41 Reaction rate equation and reaction modelformulated.

41 Suitable materials of construction for reactordetermined.

• Conventional free-space reactor for conversion ofsilane to silicon discarded because of unsuitablefine powder product.

• FBR R&D plan formulated.

More Recent Accomplishments

• Low-cost silane process technical readinessdemonstrated by operation of pilot plant (EPSDU)at Washougal, Washington:

41 High-purity silane produced under optimized­continuous steady-state conditions.

41 High-purity silicon produced using Komatsuchemical-vapor-deposition reactors.

41 Operating data obtained for design andoperation of production plant.

18

• Engineering model FBR successfully operated atWashougal, Washington:

41 Process operating parameters assessed with a6-in.-diameter FBR.

e Long-duration runs with modified reactorextended range to continuous operation for66 h using 20% silane.

41 Conditions for continuous, steady operationand for producing high-purity silicon remain tobe defined.

• FBR research at JPL:

41 Characterization for high-silane-eoncentrationoperation extended to 100% silane using6-in.-diameter reactor and cooled distributorplate (see Figure 8).

41 Operation at various high silane concentrationsdemonstrated, e.g., 80% silane for 3 h atdeposition rate of 3.5 kg/h.

Figure 8. JPL Research Fluidized Bed Reactor

• Feasibility of periodic removal of productdemonstrated.

• Scavenging mechanism describing high beddeposition and very low fines formationdemonstrated.

• Fine particle growth in silane free-space reactorat California Institute of Technology:

• Practicality of redistribution and depositionreactors as well as integrated unit demonstratedin scale-up process development unit (seeFigure 10).

• Redistribution reactor with Dowex catalystcharacterized for kinetics and conversion yield.

• Dichlorosilane characteristics determined toassess hazard:

• Aerosol theory modified to describe formationand growth of fine particles in silane pyrolysissystem.

• Reactor developed and experimentalconditions investigated with the objective ofgrowing silicon particles to size suitable forseed in an FBR.

• Purity of product from both facilities:

• Autoignition, explosion severity, hydrolysis,and explosive output measured;dichlorosilane found to be considerablymore hazardous than hydrogen ortrichlorosilane.

• Process redesigned, including elimination ofdichlorosilane storage, to minimizeproblems.

• Cleansing procedures established to ensurepurity of initial seed particles.

• Intermediate and advanced design depositionreactors studied:

• Quartz liners installed in FBR to preventcontamination from reactor wall.

• Final purity needs to be determined.

• Objectives developed from economic goal.A 2 gh-1 cm-1 deposition rate, 40 mol %conversion yield, and 60 kWh/kg energyuse.

Dichlorosilane CVD Process(Hemlock Semiconductor Corporation)

Preparation of Dichlorosilane fromMetallurgical-Grade Silicon and Deposition ofSemiconductor-Grade Silicon fromDichlorosilane

• Results from intermediate reactor studiesindicated reactor modifications required:

• Deposition rate, conversion yield, andenergy usage objectives achievedseparately.

• HCI etching step introduced to removesilicon wall deposits.

The process flow diagram is shown in Figure 9.

• Process feasibility for conversion of metallurgical­grade silicon to dichlorosilane and for chemical­vapor deposition of dichlorosilane demonstrated.

• Results from advanced reactor with cooledwall showed all objectives achievable byoptimizing conditions.

• Semiconductor-grade purity demonstratedfor product silicon:

Si

SiHCI3SiCI4

SiH2CI2HCI

DISTILLATION H2

REDISTRIBUTIONREACTOR

SiHCL3SiCI4SiH2CI2

"-- ---'SiHCI3

BORONREMOVAL

SiHCI3SiH2CI2BORON IMPURITIES

SiHCI3 SiHCI3SiCI4 SiH2CI2SiH2CI2

DISTILLATION

Si CI4SiHCI3SiH2CI2

SiCI4

FLUIDIZEDBEDREACTOR

MET·GRADE

Si

Figure 9. Dichlorosilane CVD Process Flow Diagram

19

Figure 10. Process Development Unit

• Boron, phosphorus, carbon, and metalconcentrations commensurate with, or lessthan, those of commercial semiconductor­grade silicon.

• Energy conversion efficiency of photovoltaiccells made from this material shown to beequivalent to that of baseline cells fabricatedfrom commercial semiconductor-gradesilicon.

• EPSDU sized to 100-MT/year outlined.

• Economic analysis for a 1000-MT/year plantindicated $34.80/kg (1985 dollars) for productcost (without profit).

Effects of Impurities in Silicon

The objective was to relate silicon purity to theeconomics of silicon refinement processes by the

20

determination of impurity effects on material and solar­cell properties. The approach was to characterizematerials and solar cells prepared from deliberatelydoped ingots.

The studies included the following parameters:

• Deep-level impurity concentrations in singly andmultiply doped ingots.

• Methods and growth rates for single crystals .

• Melt-replenishment methods.

• Base dopant type and concentration (resistivity).

• Anisotropic distributions of deep-level impurities.

• Grain boundary structure (impurity interactions inpolycrystalline cells).

CD Thermal and gettering treatments during cellprocessing.

• High-temperature aging of cells.

• Effects on high-efficiency PV solar cells (minoractivity).

Tolerable concentrations of impurities were found tobe determined by the effects of: (1) specific metallicimpurities on the performance of cells, (2) boron andphosphorus on resistivity and single-crystal yield iningot growth, and (3) total impurity concentration in thesilicon on single-crystal growth yield. The effects ofimpurities on cell efficiency were related empiricallythrough the base diffusion length and were shown tobe species-dependent. For example, the concentrationscausing a 10% decrease in efficiency ranged from0.1 ppba for tantalum to 10 ppma for copper (seeFigure 11). The tolerable concentrations in the poly­crystalline silicon are dependent on the methods usedfor crystal growth. Crystalline structure breakdownoccurs when the total impurity concentration exceeds acritical value for a particular set of conditions of ingotor sheet growth. For example, the critical concentrationis in the range of 200 to 500 ppma when a 1O-cm­diameter Cz crystal is pulled at a rate of about 8 cm/h.(The available database for these trade-off studies has

been incorporated into a slide rule.) The impurity limitscan be relaxed considerably for a near-equilibrium, uni­directional solidification in which liquid-solid impuritysegregation is the concentration controlling factor.Unfortunately, these studies have not been carriedexperimentally into various forms of silicon ribbon andsolar cells made from silicon ribbons.

In a fundamental approach to the impurity effectsproblem, a transmission-line-equivalent circuit modelwas developed by C.T. Sah to compute the exactsteady-state characteristics of one-dimensional siliconsolar cells. The model was used for calculations oflimiting concentrations of specific impurities for definedcell structures and efficiency. A model was alsodeveloped to study the effects of defects across theback-surface-field junction on the performance' of high­efficiency cells. A new theory to distinguish anacceptor-like deep level from a donor-like deep levelwas developed using measured values of the thermalemission and capture cross sections. This theory alsoprovides information concerning the lattice distortionaround an impurity atom before and after the captureor emission of an electron or hole at the impuritycenter.

1012 1013 1014 1015 1016 1017 1018

METAL IMPURITY CONCENTRATION (atoms/cm 3)

N - TYPE SILICON

Fe

wziii 1.0 ~-----~=---_::::--"""==:---"""==:----~en<tal"#:. 0.8

>=~ 0.6wC3u::tt 0.4ClwN

~ 0.22a:~ 0.0 L-.L--l-L...I.l--L-L-Ul_L-J.-Lll-....L.....L.Lll-----l-----l-Ll.l-...l-...LJ...Ll-----l-----l...LLJ

1011

METAL IMPURITY CONCENTRATION (ppms)

10-5 10-4 10-3 10-2 10-1 1.0

P - TYPE SILICON

wz

1.0:::;wen<tal

~0_8

>=u 0.6zwC3u::u.. 0.4wClWN:::;

0.2<t2a:0Z 0.0

1011 1012 1013 1014 1015 1016 1017 1018

METAL IMPURITY CONCENTRATION (atoms/cm3)

METAL IMPURITY CONCENTRATION (ppms)

10-5 10-4 10.3 10-2 10-1 1.0

Figure 11. Impurity Concentration Effects on Solar Cell Efficiencies, n-Type and p-Type Silicon

21

Conclusions

• Decrease of cell conversion efficiency is uniquelydependent on specific impurities and theirconcentrations.

• General empirical model correlates measuredimpurity effects on cell performance.

• Significantly less performance reduction causedby most deleterious elements in n-base cells.

• Reduction of electrical activity of impurities byprecipitation in vicinity of grain boundaries inpolycrystalline material is dependent on impuritydiffusion coefficients.

• Removal of impurity elements by thermal andgettering procedures is dependent on diffusion.

Ell Composition of acceptable polysilicon feedstockdepends on factors of single-erystal growthtechnique, melt-replenishment strategy, and cell­processing sequence.

Ell Long-term, ambient-temperature cellperformance is not significantly changed furtherby presence of deep-level impurities.

22

• Only about 10% variation in cell performance isattributable to nonuniform impurityconcentration.

Ell Model and data indicate that higher-efficiencycells are increasingly sensitive to impurities.

Research and Development Needs

• Complete understanding of silicon particle growthmechanisms and particle size control in afluidized bed reactor.

• Establish product purity and other characteristicsof FBR-produced silicon.

Ell Complete characterization of JPL FBR foroperation using high silane concentrations andproducing semiconductor-grade silicon.

• Establish engineering design and operation ofFBR for long-duration steady-state production ofsemiconductor-grade silicon.

• Verify low-cost production of polysilicon in UnionCarbide Corp. silane/FBR EPSDU.

• Define effects of silicon produced by thesilane/FBR process on solar-eell efficiency.

Si Iicon SheetIn 1975, all of the silicon sheet used for

manufacturing semiconductor devices, includingsolar cells, consisted of wafers sliced fromCzochralski (Cz) crystals by diamond-embeddedinside-diameter (ID) saws. This semiconductor­industries technology had evolved to a level ofsophistication that was adequate for solar cells foruse on spacecraft. However, the cost of producingthese wafers was prohibitive for large-scaleterrestrial PV uses. The cost of the silicon materialand the processes required to produce the thin­sheet constituted the majority of the manufacturingcosts of the early terrestrial PV modules. It wasknown that solar cells could be made from siliconless perfect than single-crystal Cz wafers, but theeffects on solar cell performance were notunderstood. Potentially less expensive silicon sheethad been experimentally grown in a laboratory;however, the quality of solar cells that might bemade from these new forms of silicon wasunknown. Therefore, an extensive investigation ofsilicon materials, its processing into thin sheets, andits characteristics for PV use was initiated by theFSA Project.

The objective of the silicon sheet activity is todevelop a sheet growth technology that will meet thecost and quality requirements for large-scale PVapplications. Initially, the objective had been todefine, develop, and demonstrate processtechnologies capable of producing large-area siliconsheet material for production of low-cost solar arraysat a price of $0.50 per peak watt (1975 dollars) by1986.

The goals were to:

ED Develop silicon sheet growth technologies tomeet specific cost and quality criteria (includingribbon growth, ingot growth, and ingot slicing).

ED Develop growth and slicing equipment to meetthese criteria.

ED Verify equipment performance.

The goals now are to establish crystalline silicongrowth technology appropriate for:

ED Silicon sheet of quality suitable for fabricationinto solar cells with efficiencies up to 20 %.

ED Sheet at a cost compatible with FSA goals.

The key technical issues are to understand thetechnologies so that sheet growth can be controlledto meet:

ED High-area throughput rates that meet qualityand economic needs.

23

ED A sheet quality with acceptable crystallographicand impurity characteristics, i.e., type, quantity,and uniformity of distribution.

ED Material stability with little stress/strain.

Background

Based upon the silicon sheet R&D performedduring the early 1960s, considerable potential existedin 1975 for lower-cost silicon sheet that wassurmised to be of adequate quality for terrestrialsolar cells, i.e., growth of ribbons, casting of sili.con,and advanced (less expensive) Cz and float-zone(FZ) crystal-growth techniques. Three new ribbongrowth techniques, by which silicon is grown directlyto the thickness of a solar cell, had been conceivedand their feasibility had been demonstratedexperimentally: the dendritic-web, Stepanov, andedge-defined film-fed growth (EFG) methods.

Direct growth of silicon ribbons that did notrequire cutting to obtain the appropriate thickness forsolar cells obviously was a way to eliminate costlywafering and silicon-material waste. The growing orcasting of silicon ingots in rectangular shapes so thatthey could be cut into square or rectangular wafers,for more efficient solar cell packing into PV modules,also seemed to be economically plausible. However,if cast-ingot technology was to be cost-effective, theslicing of the large ingots into wafers had to beaccomplished more economically than withcontemporary single-blade ID saws. A commoncharacteristic of these new silicon sheet forms wasthat they were not ideal, with imperfections andimpurities of greater densities and more unevendistributions than those of Cz crystals. The electronicproperties of these less-than-perfect forms of sheetwere not well known in the 1970s, so many questionshad to be answered. What would be the effect onsolar cell performance? How would performance beaffected by the material purity, the size andcharacteristics of the grains, and the quality of thecrystallographic structure? What quality of siliconsheet could be obtained consistently from the varioussheet-growth techniques? How would the cost toproduce sheet using each specific technique varywith sheet quality? To answer these questions,considerable research was needed andmanufacturing cost-analysis capabilities had to bedeveloped that could be used to periodically assessthe progress of the competing technologies.

Project Activities: 1975 to 1981

In 1975, under the stimulus created by the needfor large-area low-cost silicon sheet and theavailability of Government funds, several moregrowth techniques than mentioned above weredevised. The emphasis was on ribbon technologiesbecause they seemed to offer the greatest potentialfor inexpensive PV power. However, ingot

technologies were also funded because there wasgreater assurance that these efforts could succeed,An initial plan was devised that included a four-phaseeffort: research and development on sheet-growthmethods (1975-1977); advanced development ofselected growth methods (1977-1980); prototypeproduction development (1981-1982); development,fabrication, and operation of growth production plants(1983-1986),

Concurrently, with the early phases, a cost-to­manufacture economic analysis capability wasdeveloped and refined to maturity. Periodicassessments of the economic value of each sheettechnology were performed, based upon the latestprogress of sheet and other PV technologies. Thesheet technologies showing the greatest promise foreconomic viability were selected for advanceddevelopment, and the less promising were dropped.The silicon sheet technologies funded by FSA duringthe late 1970s are shown in Tables 3 and 4. Thelists of active contracts in June 1977 and September1978 show that in this period, the less promisingribbon techniques were dropped, and that newadvanced Cz growth efforts (and die and containermaterial studies) were initiated. Early in 1981, amajor review of the remaining sheet techniques wasperformed. The dendritic web, EFG, and silicon-on­ceramic (SOC)' efforts continued to be supported.Also, the advanced Cz (by Kayex Corp.), heat­exchange method (HEM), a Semix, Inc., castingprocess, and multiwire ingot slicing receivedcontinued support.

The Cz growth technique is shown inFigure 12. Semiconductor silicon chunks, the"charge," are melted in a quartz crucible within asealed grower furnace filled with argon. A silicon"seed" of the appropriate crystallographic orientationis lowered until it is in contact with the molten siliconsurface, and is then withdrawn slowly. The single­crystal silicon that grows on the seed has the samecrystallographic structure as the seed, as long as thefurnace conditions and seed withdrawal are properlycontrolled. The dimensions of the Cz crystal or ingotare determined by the quantity of the silicon chargeand the rate of seed withdrawal. The dopant andimpurity levels in the crystal are determined by thetrace elements in the charge, except that additionaloxygen and carbon are introduced during the growthprocess from the quartz (Si02) crucible and thegraphite furnace components. Control of impurities,structural defects, and their uniformities have beenimproved considerably by careful analysis,experimentation, and improved grower design andoperational control. Goals and status are shown inTable 5.

Cast-silicon ingots (Figures 13 and 14) have theadvantage of being cast in such a form that square orrectangular wafers can be sliced directly from theingot. Some cast ingots are large enough that theycan be slabbed into four or nine square ingots beforebeing sliced into wafers. Cast ingots are lessexoensive than Cz ingots because their solidification

24

rates'are'faster. However, they are polycrystalline.The quality, size, and uniformity of grains in the ingot,plus the non-uniform distribution of impurities, areparameters that influence the electronic quality ofspecific cast ingots. The efficiencies of solar cellsmade from these materials are generally less thanthose of Cz cells, and are highly dependent uponthe ingot quality. Goals and status are shown inTables 6 and 7.

Three ingot slicing techniques on which FSA hassponsored R&D are ID, multiblade, and multiwire(shown in Figures 15, 16 and 17). The ID and multi­wire saws have cutting materials embedded in asofter matrix and are subject to wearout as the cuttingparticles are dulled or eroded away. The multibladesaw and a multiwire saw (also funded by DOE) use aslurry containing abrasive particles. In all cases, cool­ing of the cutting area with a fluid is required. Thesawing occurs by the tearing away of silicon, causingsurface damage to the sawed wafer. The surfacedamage is different for each sawing technique; ineach case it must be removed, usually by etching thewafer in acid. Goals and status are shown in Table 8.

The most advanced and still the most promisingribbon technologies are the EFG methods and thedendritic web methods (shown in Figures 18 and 19),

In the EFG method, the molten silicon movesupward between parallel walls of a graphite die bycapillary action. A silicon seed is lowered to makecontact with the molten silicon in the top of the die,and is then pulled upward. The shape of the resultingribbon grown on the seed is defined by the shape ofthe top of the die. Relatively large quantities of ribboncan be grown quickly, especially by pulling severalribbons simultaneously from the same melt. MobileSolar Energy Corp. devised, with its own funds, andis producing nine ribbons from one melt by its"nonagon" concept (a tubular shape with nine flatsides). The quality of the EFG ribbon is its majorlimitation: it produces solar cells with lower efficien­cies than are possible with Cz or FZwafers. Themolten silicon reacts with the graphite dies and, asthe silicon cools, silicon carbide precipitates are incor­porated into the ribbon. The result is a ribbon that hasa disrupted crystalline structure with uneven impuritydistributions. Goals and status are shown in Table 9.

Ribbon grown by the dendritic,web method is ofhigher quality than EFG ribbon and other ribbons. Itselectronic quality can be almost as good as that ofCz wafers, except that it has a "twinning plane" inthe plane of the ribbon throughout the ribbon length.By precise control of temperatures and temperaturegradients, during and after seeding and growthinitiation, parallel silicon filaments, called"dendrites," can be grown in the silicon melt. Undercarefully controlled withdrawal, the dendritescontinue to grow and a film of molten silicon formsand solidifies between them. The only solid materialthat the molten silicon contacts is the quartz cruciblewhich minimizes contamination and results in a

Contractor

Mobil-Tyco Solar EnergyWaltham, Massachusetts(JPL Contract 954335)

IBMHopewell Junction, New York(JPL Contract 954144)

RCAPrinceton, New Jersey(JPL Contract 954465)

Univ. of So. Carolina,Columbia, South Carolina(JPL Contract 954344)

MotorolaPhoenix, Arizona(JPL Contract 954376)

Westinghouse ResearchPittsburgh, Pennsilvania(JPL Contract 954654)

HoneywellBloomington, Minnesota(JPL Contract 954356)

RockwellAnaheim, California(JPL Contract 954372)

General ElectricSchenectady, New York(JPL Contract 954350)

Univ. of Pennsylvania,Philadelphia, Pennsylvania(JPL Contract 954506)

Crystal SystemsSalem, Massachusetts(JPL Contract 954373)

Crystal SystemsSalem, Massachusetts(JPL Contract 954373)

VarianLexington, Massachusetts(JPL Contract 954374)

Table 3. Silicon Sheet Contractors: June 19 77

Technology Area

Ribbon Growth Processes

Edge-defined, film-fed growth

Capillary action shaping technique

Inverted Stepanov growth

Web-dendritic growth

Laser zone ribbon growth

Dendritic web process

Sheet Growth Processes

Dip coating of low-cost substrates

Chemical vapor deposition on low-cost substrates

Chemical vapor deposition on floating siliconsubstrate

Hot-forming of silicon sheet

Ingot Growth Process

Heat-exchanger ingot castinga

Ingot Cutting

Multiple wire sawinga

Breadknife sawing

aSingle contract provides for both ingot casting and mUltiple wire sawing.

25

Contractor

Mobil-Tyco Solar EnergyWaltham, Massachusetts(JPL Contract 954355)

Motorola, Inc.Phoenix, Arizona(JPL Contract 954376)

Westinghuuse ResearchPittsburgh, Pennsylvania(JPL Contract 954654)

Honeywell Corp.Bloomington, Minnesota(JPL Contract 954356)

RCA LabsPrinceton, New Jersey(JPL Contract 954817)

Crystal Systems, Inc.Salem, Massachusetts(JPL Contract 954373)

Kayex Corp.Rochester, New York(JPL Contract 954888)

Siltec Corp.Menlo Park, California(JPL Contract 954886)

Texas InstrumentsDallas, Texas(JPL Contract 954887)

Varian Vacuum DivisionLexington, Massachusetts(JPL Contract 954374)

Varian Vacuum DivisionLexington, Massachusetts(JPL Contract 954884)

Battelle LabsColumbus, Ohio(JPL Contract 954876)

Coors PorcelainGolden, Colorado(JPL Contract 954878)

Eagle PicherMiami, Oklahoma(JPL Contract 954877)

RCA LabsPrinceton, New Jersey(JPL Contract 954901)

TylanTorrance, California(JPL Contract 954896)

Table 4. Silicon Sheet Contractors: September 1978

Technology Area

Shaped Ribbon Technology

Edge-defined film-fed growth

Ribbon growth, laser zone regrowth

Dendritic web process

Supported Film Technology

Silicon on ceramic substrates

Epitaxial film growth on low-eost silicon substrates

Ingot Technology

Heat exchanger method (HEM), cast ingot, andmultiwire fixed abrasive slicing

Advanced Cz growth

Advanced Cz growth

Advanced Cz growth

Multiblade slurry sawing

Advanced Cz growth

Die and Container Materials Studies

Silicon nitride for dies

Mullite for container and substrates

CVD silicon nitride and carbide

CVD silicon nitride

Vitreous carbon

26

ADVANCED CZOCHRALSKICRYSTAL GROWTH SYSTEM

ISOLATIONVALVE

MULTIPLE INGOT CAPABILITY

CRYSTAL REMOVED

HOPPER WITH SILICON INSERTED

LOWERED TO CRUCIBLE - UNLOADED

SECOND CRYSTAL GROWN

CYCLE REPEATED

--~~

GROWINGCRYSTAL

FURNACECHAMBER

SEEDCRYSTAL -----hi'kil1

MOLTENSILICON

INGOT PULL ANDROTATION -----<'"1MECHANISM

DOOR

CZOCHRALSKI INGOT

Figure 12. Advanced Czochra/ski Crystal Growth by Hamco Division (Kayex Corp.)

HEAT EXCHANGER METHOD

VIEW PORT

INSULATION

HELIUM GAS

1:--, 'L,u"'l.Jj:--HEATING ELEMENT

POWER LEADS~;I.';..-;e'v.- CRUCIBLE

• TO VACUUM PUMP

PYROMETER

CHAMBER

SILICON MELT

SEED CRYSTAL

Figure 13. Ingot Casting by Heat Exchanger Method by Crystal Systems, Inc.

Figure 14. Ingot Casting by Semicrystalline Casting Process by Semix, Inc.

27

Table 5. Advanced Czochralski Crystal Growth by Hamco

GOALS: STATUS R&D NEEDS

" 150 kg OF INGOTS/ " 150 kg OF INGOTS/ "INGOT QUALITY:CRUCIBLE CRUCIBLE STRUCTURAL PERFECTION

" 15-cm-DIA INGOT " 15-cm-DIA INGOT " GROWTH PARAMETEROPTIMIZATION

" 2.5 kg/hr THROUGHPUT " 2.2 kg/hr THROUGHPUT1986 INGOT ADD-ON PRICE'(1980 $/Wp)

" AUTOMATION " UNDER DEVELOPMENTAT END OF CONTRACT(LATER ACHIEVED BY HAMCO)

" 15% ENCAPSULATED SOLAR " 15% SOLAR CELL "FSA PRICECELL EFFICIENCY EFFICIENCY' , ALLOCATION 0.112

"90% YIELD "86% YIELD " SAMICSPROJECTION 0.162

'ASSUMES 0.73 m2 OF WAFERS/kg OF INGOT

"ENCAPSULATED CELL EFFICIENCY N.A.

Table 6. Ingot Casting by Heat Exchanger Method by Crystal Systems, Inc.

GOALS: STATUS: R&D NEEDS

" 30 em x 30 em x 15 em " 34 em x 34 em x 17 em " IMPROVED INGOT QUALITYSHAPED INGOT (35 kg) SHAPED INGOT (35 kg) " INCREASED INGOT YIELD

" >90% SINGLE CRYSTAL " > 95% SINGLE CRYSTAL 1986 INGOT ADD-ON PRICE"" 2.0 kg/hr GROWTH RATE " 1.9 kg/hr GROWTH RATE (1980 $/Wpl

" 15% ENC. SOLAR CELL " 15% SOLAR CELL " FSA PRICEEFFICIENCY EFFICIENCY ALLOCATION 0.194

">95% YIELD ">95% YIELD " SAMICSPROJECTION 0.17

" ASSUMES 1 m 2 OF WAFERS/kg OF INGOT

Table 7. Semicrystalline Casting Process by Semix

GOALS: STATUS: R&D NEEDS

" 30 em x 30 em x 19 em " 20 x 20 x 15 " LARGE QUANTITY OFSHAPED INGOT (40 kg) (14 kg) LARGE AREA CELLS AT

15%

" SEM ICRYSTALLINE " SEMICRYSTALLINE " DEMONSTRATION OFHIGH THROUGHPUTPROCESSING

" 15% SOLAR CELL " 15% SOLAR CELLEFFICIENCY EFFICIENCY 1986 ADD-ON PRICE"

(1980$!Wp)

" 98% INGOT YIELD " 83% INGOT YIELD "FSA PRICEALLOCATIONS 0.256

" 2.3 kg PER HOUR " 2.3 kg PER HOUR" SEMIX PROJECTION

CRYSTALLIZATION 0.115

RATES

"USING 0.81 m2 OF WAFERS/kg OF INGOTS

28

SILICON TECHNOLOGY CORP

1.0.* SAW

IMBEDDEDDIAMONDPARTICLES

'INNER DIAMETER

Figure 15. Inside Diameter Ingot Saw by Silicon Technology Corp.

HOFFMAN

MULTIBLAOE SAW

SLURRY WITHCUTIING L-_-\-.....COMPOUND

Figure 16. Multiblade Saw by P. R. Hoffman Co.

CRYSTAL SYSTEMS, INC.

MULTI-WIRE SAW

/ROCKINGINGOTHOLDER

Or)

• Figure 17. Multiwire Saw by Crystal Systems, Inc.

29

Table 8. Advanced Wafering Technologies by Crystal Systems Inc., P.R. Hoffman, andSilicon Technology Corp.

GOALS STATUS ADDITIONAL R&D(BEST SINGLE DEMONSTRATION) REQUIRED

• DEVELOP FUNDAMENTAL UNDERSTANDING15 em DIA 10 x 10 em square 15 em DIA 10 x 10 em OF WAFERING PROCESS, FOR

20 WAFERS/em 25 WAFERS/em OF 17 WAFERS/em 25 WAFERS/emINCREASING THROUGHPUT

OF INGOT INGOT • INCREASED WAFER THROUGHPUT

• REDUCED EXPENDABLE COSTS0.5 WAFER/min 1 WAFER/min 0.42 WAFER/min 1.25 WAFER/min • SENSITIVITY TO EXTENDED OPERATION

0.82 m2/kg' 1.0 m2/kg' 0.69 m2/kg" 1.0 m2/kg 1986 WAFER ADD-ON PRICE(1980 $/Wp)

95% YIELD 95% YIELD 95% 95% • FSA PRICE ALLOCATION0.081 (15 em DIA)

• SAMICS 0.110 (15 em DIA)

'NO LONGER UNDER DEVELOPMENT

"SQUARE METER OF WAFERS/KILOGRAM OF INGOT

clean, smooth ribbon. The dendrites are easilyremoved after the ribbon has cooled. The majordisadvantage of this method is the low productionrate of quality ribbon. Significant difficulties havebeen encountered as faster growth rates andsustained operation for many hours have beenattempted. The problems of growing low-stressribbons at economical growth rates continue toplague the web efforts. Goals and status are shownin Table 10.

The silicon-on-ceramic (SOC) technique was thethird promising non-ingot growth method. It wasunique in that it had high growth rates; but theefficiency of solar cells made from it was low.Consequently, when the need for higher efficiencybecame widely accepted and the Project funds werecut, SOC efforts were dropped (see Figure 20 andTable 11).

In 1981, the direction of the National PhotovoltaicsProgram was altered to emphasize long-term, high­risk research efforts, and funding was significantlyreduced. Concurrently, there was a growingawareness that PV module efficiencies had to beraised considerably if PV was to be cost effective forlarge utility applications. Higher-efficiency modulesand solar cells require higher-quality silicon sheet.Consequently, the production-development-orientedthird an,d fourth phases were discontinued, and ingotgrowth, ingot slicing, and ribbon technologies otherthan dendritic web and EFG were no longer funded.

The status of the sheet technologies that weredropped because of programmatic changes areshown in Table 12, as they were at time of contractcompletion. An exception, low-angle silicon sheet(LASS) growth method, was funded for about 2 years.

30

Project Activities: 1981 to Present

With a change in Project directions, R&D onspecific growth technologies was then replaced bygeneric growth R&D, More recently, a concertedeffort to increase the area growth rates ot hlgh­quality dendritic web are now supported by aconsortium of utilities, DOE and Westinghouse.The status of web and EFG in 1985 is shown inTable 13.

As time passed, ribbon-buckling problems andreduced ribbon quality that were encountered whenhigher ribbon growth rates were attempted led to theconclusion that a more fundamental understanding ofthe impediments to high-speed silicon ribbon growthand quality was required. This resulted in theestablishment of an interactive university-industry­Government effort that continues to actively pursuetheoretical and experimental ways to achieve high­speed, high-quality silicon crystal growth. Computermodels have been developed and are being used tostudy the stresslstrain relationship in the siliconsolidification zone and the nearby areas in theribbon. The analytical results are then experimentallyevaluated on ribbon-growth equipment atWestinghouse and Mobil Solar. In support, high­temperature silicon material properties are beingdetermined and sophisticated instrumentation andmeasurement techniques continue to be developed.Progress and problems of the various efforts arediscussed at frequent meetings and the resultsiterated into the appropriate activities. A low-angle(horizontal) silicon sheet growth effort was alsoinitiated to explore the potential value of this veryhigh-speed-ribbon-growth technique (Figure 21 andTable 14). This technique has the potential to evolveinto a practical growth technology, but is not nowfunded by DOE because of a shortage of funds. Itwas funded in an attt3mpt to gain generic informationabout high-speed silicon growth.

EFG* RIBBON

DIE

CAPILLARYACTION

MOLTENSILICON

'EDGE-OEFINED FILM-FED GROWTH

3-to 1O-em WIDE RIBBON GROWERS

f

Figure 18. Edge-Defined Film-Fed Ribbon Growth by Mobil Solar Energy Corp.

WEB DENDRITICSHEET GROWTH

Figure 19. Dendritic-Web Ribbon Growth by Westinghouse

t:~t MOVABLE DISPLACER

SILICON__ _ MELT

GRAPHITE ROLLERS

Figure 20. Silicon-on-Ceramic by Honeywell

31

Table 9. Edge-Defined Film-Fed Ribbon Growth by Mobil Solar Energy Corp.

GOALS:

• SIMULTANEOUS GROWTH OF'4 RIBBONS, 10 em WIDE EACH,AT 4 em/min FOR 8 hrWITH AUTOMATIC CONTROLS

• MULTIPLE RIBBON GROWTH*STATION

• 0.20 mm RIBBON THICKNESS

• 12% ENCAPSULATED SOLARCELL EFFICIENCY

• 80% YIELD

STATUS:

• 40 em2/min (10 em WIDE'x 4.0 em/min) (4 hr DEMOOF ONE RIBBON)

• DEMONSTRATED'(5 RIBBONS AT 5 em WIDE)(3 RIBBONS AT 10 em WIDE)

Ell 0.15-mm RIBBON THICKNESS

Ell >12% SOLAR CELL"EFFICIENCY

R&D NEEDS:

• IMPROVED SHEET QUALITY

Ell INCREASED THROUGHPUTGROWTH SPEED

1986 SHEET ADD-ON PRICE(1980 $/Wp)

• FSA PRICEALLOCATION 0.205

• SAMICS'PROJECTION 0.194

'MOBIL DEVELOPING COMPETITIVE NONAGON EFG TECHNOLOGY"ENCAPSULATED CELL EFFICIENCY N.A.

Table 10. Dendritic-Web Ribbon Growth by Westinghouse

GOALS (LATE 1986): STATUS:

e 30-cm2/min AREA e 13-cm2/min AREA GROWTH RATEGROWTH RATE FOR 50 METER·LONG FOR SHORT RIBBON LENGTHS ANDCRYSTAL AND CONSTANT MELT LEVEL GROWTH RATE

e 17.5% SOLAR CELL EFFICIENCY e 16.9% CELL EFFICIENCY

e PROCESS AUTOMATION e AUTOMATION BEING INCORPORATED

e 8-hr DEMONSTRATION RIBBON GROWTHFOR 8-3/4 hr AT CONSTANTMELT LEVEL

Table 11. Silicon-on-Ceramic by Honeywell

R&D NEEDS:

e INCREASED THROUGHPUT

e GROWTH AND STABILITY

1986 SHEET ADD-ON PRICE(1980 $/Wp)

e FSA PRICEALLOCATION 0.292

e SAMICSPROJECTION 0.160

GOALS:

• <$6/m2 SUBSTRATE COST

• 350-em2/min AREA GROWTHRATE (2 SHEETS)

• 0.10-mm FILM THICKNESS

• 11% ENCAPSULATED SOLARCELL EFFICIENCY

• 95% YIELD

STATUS:

• MULLITE SUBSTRATE IDENTIFIED<$6/m2 PROJECTED

• 60-em2/min AREAGROWTH RATE (1 SHEET)

• <0.10-mm FILM THICKNESS

• 10.5% SOLAR CELL"EFFICIENCY ON DIP-COATED SOC

ADDITIONAL R&D REQUIRED:

• IMPROVED SHEET QUALITY

e INCREASED THROUGHPUT

• OPTIMIZED CELL DESIGN

1986 SHEET ADD-ON PRICE(1980 $/Wp)

e FSA PRICEALLOCATION 0.190

e SAMICSPROJECTION N.A.

'NO LONGER UNDER DEVELOPMENT"ENCAPSULATED CELL EFFICIENCY N.A.

32

Table 12. Technology Status at Completion of Contract

MATERIAL SOLAR CELL EFFICIENCY ADD-ONSHEET DIMENSIONS THROUGHPUT UTllI~ATION

%AMl PRICETECHNOLOGY (em) (kglhr)tor (m2Ihr) m Ikg ENCAPSULATED BARE CELL SIWp(1980S)

ESTIMATED' TRINGOT TR GOAL STATUS TR GOAL STATUS TR GOAL STATUS TR GOAL STATUS ALLOCATION

Adv. CZ 15 DIA 15 DIA 2.5t 2.2t n,a. n.a. 15 13.5 0.112

HEM 30x30x15 34x34x17 1.0t 1.0t n.a. n.a. 15 11.5 0.194

Semix Casting30x30x19 20x20x15 2.3.... 2.3.... n.a. n.a. 15 11.0 0.245'"Process

15 DIA & 15 DIA & 0.53 0.42 0.73 0.65 n.8. n.a. 0.081Adv. Wafering"10xl0 SO 10xl0 SO 0.60 0.75 1 1 n.a. n.a. 0.062

SHAPED SHEET

LASS 1 @ 15 3.6 3.6 12

SOC 2 @ 12.5 1 @ 102.1 0.3 4.3 4.3 11 9.5 0.190x 0.010 x 0.010

'HIGHER EFFICIENCY INDIVIDUAL CELLS HAVE BEEN FABRICATED"DATA GIVEN FOR ID WAFERING. OTHER WAFERING APPROACHES WILL ALTER THE FIGURES

'''ADDED VALUE FOR CASTING AND WAFERING....CRySTALLIZATION RATE kglhr

Table 13. Technology Status of Active Contracts

MATERIAL SOLAR CELL EFFICIENCY ADD-ONSHEET DIMENSIONS THROUGHPUT UTllI~ATION %AMl PRICETECHNOLOGY (em) (m2/hr) m /kg ENCAPSULATED BARE CELL S!Wp(1980S)

C~RRENT CURRENT CURRENT ESTIMATED' TRTR GOAL STATUS TR GOAL STATUS TR GOAL STATUS TR GOAL STATUS ALLOCATION

SHAPED SHEET

EFG" 4@ 10WIDE 3@ 10WIDE0.96 0.60 2.15 2.15 12 14 0.205x 0.020 THICK X 0.020 THICK

WEB - - 0.18 0.05 <;;;3.2 <;;;3.2 15 16.9 0.292

'HIGHER EFFICIENCY INDIVIDUAL CELLS HAVE BEEN FABRICATED"NONAGON RIBBON CONFIGURATION FUNDED BY MOBIL SOLAR HAS HIGHER VALUES.

STATUS (BEST INDIVIDUAL ACHIEVEMENT)

Table 14. LOW-Angle Silicon Sheet Growth byEnergy Materials Corp.

demonstrated individually. But until a few monthsago, progress had been very slow in the efforts toachieve all growth requirements simultaneously andto grow web continuously for hours while meeting therequirements. To obtain a fresh look at this growthtechnique, a special JPL Web Team was formed andhas recently made interesting advancements(Table 15). During the last few months, Westinghousehas also made good progress (Table 16). The majorquestion still remains: How economically can high­quality web be grown under production conditions?

25 mils

450 em2/min

15 em

1 RIBBON AT 85 em/min

12.9%

RIBBON THICKNESS:

PULL SPEED:

THROUGHPUT:

EFFICIENCY:

RIBBON WIDTH:RIBBON

SCRAPER

MELT

HEAT FLOW MODIFIER

CRUCIBLE

Figure 21. Low-Angle Silicon Sheet Ribbon Growthby Energy Materials Corp.

Of all of the sheet technologies, dendritic web isviewed today as having the greatest potential forhigh-quality sheet with the best economical value.However, close temperature control of the siliconliquid-solid interface regions in the grower arerequired. By 1981, a number of web-growthrequirements to meet Project goals had been

33

Table 15. JPL Web Team

Objective

The objective of the JPL Web Team is to conduct an independent R&D activity that will increase thelikelihood that dendritic web technology will achieve the DOE goals.

Approach

Team effort consists of a combination of analytical and experimental work.

Operate a modified Westinghouse web growth system.

Measure high temperature physical properties of silicon.

Use improved stress web model to determine a web temperature profile resulting in higher areagrowth rates of web with satisfactory quality.

Use thermal model of web to determine a growth system thermal configuration that will yield thedesired web temperature profile.

Use thermal analysis of web-growth interface and the susceptor-crucible melt to determine agrowth system thermal configuration that will improve stability of growth process.

Design, fabricate, and test the new thermal configurations and feed results back into the models.

Accomplishments

High temperature properties of silicon have been measured.

A stress model of growing web has been developed and used.

A thermal model of growth system configuration to control web temperature has been developed.

A thermal model of susceptor, crucible, and melt has been developed.

Westinghouse growth system has been installed with an improved data acquisition system.

Numerous web ribbons have been grown.

Table 16. Dendritic- Web Growth Progress by Westinghouse (Westinghouse data)

Mid

Development 1977 1978 1979 1980 1981 1982 1983 1984 1985

Area Growth Rate, cm2/min

1) Transient (Lengths Of Several Centimeters) 2.3 8 23 27 42

2) Ouasi-Steady-State (Lengths Of 30 To 100 cm) 7 13

3) Steady-State (Meters Of Length, Hours Of Growth) 4 8

Maximum Undeformed Width, Centimeters 2.4 3.5 4.0 4.4 5.5 5.8 6.7

W150-Undeformed Width At 150 IJ.m Thickness2.0 2.7 3.2 4.9(An Inverse Measure Of Buckling Stress)

Maximum Area Throughput For Single Furnace in9,000 27,000

5-Day Week (cm2/week)

Maximum Demonstrated Solar Cell Efficiency,14 15 15.5 15.9 16.2 16.9

AM1 13

34

All three ingot technologies (shown in Figures 12,13, and 14) are in commercial use today forterrestrial solar cells. The majority of the wafers forterrestrial solar cells are still sliced from Cz single­crystal ingots. In 1975, the solar cells fabricatedspecifically for terrestrial use were made from thecircular wafers sliced from Cz cylindrical crystals.Today, Cz crystals generally are slabbed lengthwise,converting cylinders into polygons that are then slicedinto polygon wafers, thereby enabling the solar cellsto be more efficiently "packed" into modules. Now,much larger Cz crystals are grown and multiplecrystals can be grown from the same crucible byusing melt-replenishment. Formerly, a new cruciblewas required for each crystal grown.

Today, ID sawing remains the predominantcommercial, and most economical, slicingtechnology. More economical sawing remains themajor impediment to continued long-term use of ingottechnologies for PV. Studies of the basicmechanisms involved in the removal of silicon,including the influence on cutting rates and surfacedamage of various fluids, are promising. Theseresults, as yet, have not been applied in an attemptto develop a new saw or to improve existing saws.

The best-quality silicon sheet today is obtained byFZ crystal growth; the next best is by Cz crystalgrowth. Very high-efficiency solar cells have beenmade in the laboratory from FZ wafers. However, FZtechnology is not used commercially for PV becauseof its high cost. FZ technology development hasnever been part of the FSA Project. Excellentprogress in reducing the cost of growing Cz crystalswas made during the late 1970s, primarily by FSA­sponsored efforts. During the early 1980s, continuingprogress in improved Cz crystal quality has beenmade for and by the integrated'circuit industry tomeet its needs for more compact and sophisticateddevices. The achievement of even higher-quality Czcrystals by MCz (Cz crystals grown in a magneticfield), improved understanding of growth phenomena,and better equipment is continuing for more stringentfuture integrated-circuit industry needs. Theseactivities fit in well with the PV need for higher­efficiency cells. However, the major obstacle tofuture use of Cz or FZ wafers for PV is the cost ofwafering. The added cost of the wafering step isincidental to the cost of manufacturing integrated­circuit devices, but it looms as the single most costlyitem if Cz or FZ technology is to be used for futuresolar cells (one cell/wafer) and modules. A very lowkey, but promising effort on understanding the basicmechanism of silicon fracturing during sawing hasbeen continuing. With the uncertainties in theeconomical value of ribbon growth, it would beprudent to increase efforts leading to moreeconomical slicing and an evaluation of the technicaland economical potential of advanced Cz andadvanced FZ crystal growth for higher-efficiency,low-cost PV cells and modules.

35

Progress Since 1975

Ingot Technology

Ell Cz Ingots

'" Ingot diameters increased from 3 to 6 in.

'" Throughput rates were increased threefold.

'" Ingot quality was improved.

'" Sheet costs were decreased to one-seventh.

Ell Cast Ingots

'" Two concepts have proven feasible: HEMand Semix (semicrystalline).

'" The same two techniques have advanced tocommercial products.

Ell Wafering

.. Developed and demonstrated increased ingotslicing capabilities.

'" Reduced production costs.

Ribbon Technology

Ell Thirteen ribbon and non-ingot technologies wereinvestigated.

Ell Two ribbon technologies continue to besupported:

'" Dendritic web, by Westinghouse ElectricCorp.

'" EFG, by Mobil Solar Corp. (now funded onlyas generic ribbon growth).

'" One ribbon technology (EFG) is nowcommercially available. One ribbontechnology (web) is scheduled to bedemonstrated by late 1986.

'" Significant improvements in understandinghigh-speed crystalline silicon growthphenomena are being obtained.

o Low-angle silicon sheet, by Energy MaterialsCorp. (contract ended June 1985).

Recent Achievements

Ell Thin, flat, essentially stress-free dendritic-webribbon width has been increased to 6.7 cm(Westinghouse).

• Using a new grower with dynamic controls,dendritic-web linear growth rates have beenincrea~ed to 3.0 em/min for 150-fLm-thick ribbonsunder transient growth conditions (Westinghouse).

• Dendritic-web transient area growth rate for shortribbon lengths was increased from 27 to 42cm2/min (Westinghouse: see web growthprogress, Table 16).

• Continuous web ribbon growth with meltreplenishment was increased to 7-m length at7 cm2/min using technology developed for FSA(Westinghouse).

• Integration of the theoretical studies of heat flowand interface stability by Massachusetts Instituteof Technology (M IT), and plastic flow near theinterface region by Harvard University, were inte­grated with residual stress data by Mobil Solar.

• Microhardness tests of silicon in various fluidenvironments has revealed that the surfacehardness of silicon varies apparently as a functionof the fluid's dielectric constant. Tests wereconducted with toluene, acetone, ethanol,methanol, glycerol, and deionized water(University of Illinois at Chicago).

• A research forum and other meetings on high­speed growth and characterization of crystals forsolar cells have resulted in a definition of theareas requiring research.

• A comprehensive plan for increasing ribbonquality and growth capability has been devisedand is being partially implemented.

Research and Development Needs

Silicon sheet material R&D is needed for ribbon oringot in the cost-to-produce range of $90/m2, suitablefor production-type processing into high-performance17 to 20% efficiency solar cells.

• Requirements

e Diffusion length equal to or greater than300 fLm.

e A 60- to 100-cm/s bulk recombinationvelocity.

• Research Areas

Basics of physics and chemistry of siliconmaterials at room temperature and near themelting point:

36

e Stress/strain effects on ribbon and bulksilicon during growth.

e Effects of impurities on stress/strain in silicongrowth.

e Fracture mechanics and surface chemistryeffects on:

G Device efficiency.

G Cell and module performance lifetime.

G Fabrication and processing rates andyields.

Ribbon Technology

• A general thermal stress model of ribbongrowth integrating creep behavior, growthspeed, and ribbon width should be completedand applied to all ribbon growth processes.

• Correlation of observed buckling phenomenaand stress model.

• Data on basic silicon material properties (i.e.,creep behavior and stress moduli) attemperature range of interest (800 to 1400 0c).

• Correlation of structure (i.e., dislocations andgrain boundaries) effects with generated stress.

• Impurity effects on stress generation.

• Verification of stress model with experimentalgrowth systems in industry.

It Construction of or modification of a growthsystem that integrates all information generatedto produce optimum silicon ribbons at highspeed.

Ingot Technology

It Improved understanding of the effects ofimpurities on the growth of uniformly high­quality single crystal ingots.

• Improved understanding of defect formationduring single-crystal ingot growth.

It Verification of the feasibility of growthtechniques that improve quality of ingots.

It Improved understanding of slicing phenomenathat will lead to higher quality, more economicalwafers.

Current Research and Plans

• Stress and strain relationships in wide, thinribbons grown at high speeds:

€I Develop methods of measuring thermal fieldsin growing ribbons.

€I Develop methods of measuring residual stressin ribbons.

€I Develop methods of observing ribbondeformation in real time.

€I Measure the high-temperature mechanicalproperties of silicon.

€I Develop models for plastic deformation andelastic buckling in growing ribbons.

€I Verify models by relating the observed thermalfields to observed ribbon stress and strain.

€I Use verified models to define ribbon thermalfields and growth dynamics that will yield low­stress growth environments.

37

€I Modify present ribbon growth processes to usethe data obtained on the above.

€I Design, fabricate, and test low-stress ribbongrowth systems.

• Impurity effects and interface stability duringribbon growth:

€I Model heat flow and interface stability criteriafor ribbon growth.

€I Test and verify model.

• Sheet quality effects on solar cell performance:

€I Characterize ribbon material grown undervarious conditions.

€I Fabricate baseline cells on ribbons grownunder various growth-process conditions.

€I Relate cell performance to ribbon growthprocess variables and specific defects.

€I Optimize growth processes that compensatefor specific ribbon defects.

High-EfficiencySolar Cells

During the past 10 years, as FSA effortscontinued, it has been perceived that the originalProject goal of 10% module efficiency (from 6% in1975) would be met as a result of the planned FSAefforts. It was anticipated that this would occur assolar cell efficiencies improved and as modulepacking factors increased, i.e., more active solar cellarea per unit of module area. Logic indicated thatcell efficiency should increase as low-cost celldesign and fabrication technology matured, and asmass production and quality assurance were intro­duced. Module packing factors of 0.90 were thoughtto be achievable from the then 0.54 value. (1.0 equalstotal coverage.) These presumptions have proven tobe accurate because, today, production modules fab­ricated using single-crystalline silicon cells haveefficiencies as high as about 11 %.

Economic analyses now indicate that moduleefficiency must be about 15 % for photovoltaics to beeconomically competitive for central-stationapplications, which is now expected to be the firstwidespread use of photovoltaics in the United States.The achievement of 15 % efficiency modules requiresthe repeatable fabrication of large-area laboratorysolar cells with 20 % or greater efficiency,demonstrating that high-efficiency solar cellprinciples are understood. This knowledge then mustbe converted into low-cost production cell technologythat will yield uniform quality production cells of 17 to18 % efficiency and, subsequently, moduleefficiencies of 15 % .

The objective is to perform research that identifiesand resolves key generic limitations to increasedefficiency of crystalline silicon solar cells as requiredfor large power generation applications.

The goal is to establish device technology requiredfor repeatable fabrication of large-area laboratorycrystalline silicon solar cells of greater than 20%efficiency.

Background

Years ago, the need for high-power output fromspacecraft solar cells motivated a sustained researchfor higher-efficiency cells. With the realization thatphotovoltaics had the potential to become a practicalterrestrial power source, the research broadened.Consequently, the performance of silicon solar cellshas been gradually increasing as a result of thesesystematic efforts to gain an understanding of howsolar cells function, how to reduce cell losses bybetter cell designs, and by improved processing. Alarge number of relatively small advances, over manyyears, have added to our knowledge of solar celltechnology and to increased cell efficiency. Early

38

progress in cell designs and process techniqueswere largely empirical, which has been strengthenedby better understanding of the underlying principlesof operation. These activities have led to a generalagreement regarding the factors that limit cellperformance and to those that can be changed tofurther increase the performance of crystalline siliconsolar cells.

These efforts by many researchers andtechnologists over many years have graduallyincreased efficiencies from about 6 % when siliconsolar cells were devised in 1954 to the followingefficiencies at present:

• Production cells made from Cz ingots forterrestrial use have efficiencies up to 14 %.

• Research cells made from Cz ingots havereached 17 to 18%.

• Research cells made from float-zone siliconhave reached 19 % efficiency.

• Theoretical limit of crystalline silicon cells isabout 30%.

• Probably maximum achievable limit is about25%.

Project Activities: 1982 to Present

The general approach being used to increase solarcell efficiency to greater than 20% has beenestablished. The required changes in certain cellparameters to achieve this high efficiency arebelieved to have been identified. However, the exact'cell parametric changes and the cell designmodifications required so that large-area 20%laboratory cells can consistently be fabricated havenot been devised. To assist in better understandinghow a cell functions and how various parts of a cellinfluence its performance, the following figures anddescription are included.

The main features of a crystalline silicon solar cellare shown in Figure 22. The cell is made from asemiconductor material doped (minute amounts of aspecific element have been introduced undercarefully controlled conditions) in a manner so as tomake the semiconductor sheet photosensitive. Thetypical silicon cell made from p-type silicon (dopedwith boron) sheet material is modified by the diffusionor implantation of a phosphorus dopant to anextremely thin layer (now n type) on the surface thatwill become the top of the cell (emitter). Thesemiconductors must be doped correctly, so that then-type material has large electron densities and thep-type material has large hole densities. (A hole is avacancy in a silicon atom's outer electron ring afteran electron has been freed.) After the nip junction isformed, a basic requirement for PV energyconversion has been met by the creation of an

I

4 SQ. CM100 MW/SQ. CM25 C

; 640.9 mvolts; 143.5 mamps; 539.8 mvolts; 134.2 mamps; 72.4 mwatts; 0.787; 18.10

CELL AREAINTENSITYTEMP

VOCISCVm1mPmFF% EFF

/ISC

160 .-----------------------,MAX POWERPOINT

120 f-

40 I-

«E,.:ffi 80 I-a:a:::>o

I I I I i Io0L--L.--I-~_::'_0.2~-l--..L-.L--:0:'-:.4-.L.-....L-L.-...,.OL.6--l.L.-....JVOLTS

Figure 23. Current-Voltage Curve of a High-EfficiencySolar Cell

electrical field, which separates electrons and holeswithin the cell. Near the junction, most of thephotogenerated holes from the n side transfer to thep side; similarly, most of the photogenerated electronstransfer to the n side. The nIp junction is permanentand the electrical field is stable. When light shines ona cell, the light that is not reflected is absorbed bythe semiconductors. When a photon enters the celland is of sufficient energy (appropriate wavelength)and strikes one of the four outer (valence) electrons ofa silicon atom, that electron is dislodged. This leavesa silicon bond missing an electron, called a hole, andfrees an electron to move about in the cell. Electri-cal current flow in a solar cell, when it does occur, isdue to both the motion of electrons and to effectivemotion of the holes. When sunlight is absorbed inthe cell and frees electrons and holes from theirbonds, the electrical field separates the charges andestablishes an electrical potential, or voltage. If thecell top and bottom contacts are connected to anexternal circuit, an electrical current (amperage) willflow as long as light continues to be absorbed bythe cell.

Sunlight

~ ~ ~ ! MATERIAL, CHEMICAL AND ELECTRI·CAL PROPERTIES(THICKNESS, DEFECTS. GB, IMPURI·TIES, RESISTIVITY, CARRIER LIFE­TIME, ETC)

GROWTH METHODS (CVDI, EVAPORA·TION, OXIDATION, REFLECTANCE,REFRACTIVE INDEX, LATTICE STRAIN,SURFACE PASSIVATION (SiOx• Ta20SSINx)

DOPANT, DOPING METHODS, DOPINGCONCENTRATION, DOPING PROFILE(GRADIENTS VS UNIFORMI, LOCALIZEDEFFECTS AT GBS, DEFECTS

~~BUSSBAR

AR COATING

N + TOP EMITTERJUNCTION

nip Junction

FrontContact

Figure 22. Crystalline Silicon Solar Cell

/~:::::;:=========:::::f"Anti·ReflectionCoating

Figure 24. Key Factors that Influence Cell Efficiency

t::~====~:2======BACKSURFACE FIELDc: BACK METAL CONTACT

AREA COVERAGE, ME:TALS (TI·Pd·Ag·, Ai, Ni, Ag, Cu) EVAPORATION,SCREEN PRINTING SINTERING

TOP METAL CONTACT, METALS (HPd·Ag; Ni, Cu, Ai, Ag), METHODS(EVAPORATED, SCREEN PRINTED),CONTACT AND GRID RESISTANCE,STABILITY WITH TEMPERATURE

THICKNESS AND QUALITY OF BASE,MINORITY CARRIER RECOMBINA·TION, SURFACE PREPARATION,SHARP VS GRADUAL (AI, B)CONCENTRATION DEPTH

BASEP-Si SHEET

The primary thrust of the FSA high-efficiencyefforts has been and continues to be directedtoward reducing solar cell bulk and surface losses,with specific focus on reducing the minority carrierrecombination rates throughout the cell. Threemethods being used are: increased minority carrierlifetime, reduced volume of silicon for bulkrecombination, and reduced surface recombination.By obtai~ing silicon sheet with properties so that the

For over a decade, there have been significantimprovements in crystalline silicon solar cellperformance, mainly as a result of increased cellshort-circuit current output. (Figure 23 is an I-V curveshowing the current and voltage of a typical high­efficiency solar cell fabricated at JPL.) Theimprovements have resulted by use of shallowjunctions, back surface fields, back surfacereflectors, textured surfaces, and multilayerantireflection coatings. Shallow junctions help toreduce the effects of cell front surface losses. Back­surface fields reduce the effects of back-surfacelosses. Back-surface reflectors increase photonabsorption. Textured surfaces and multilayerantireflection coatings reduce light reflection lossesat the front surface. Figure 24 shows a solar cell withkey factors that influence cell efficiency. The potentialfor further improvements in a cell's short-circuit cur­rent are limited. However, single-junction crystallinesilicon cell performance can be increased by increas­ing the cell open circuit voltage. This is the areaof endeavor now being emphasized by FSA high­efficiency silicon solar cell research.

39

recombination rates in the bulk material do notdominate cell performance, the emphasis can thenbe on reducing recombination in the thin emitterregion and at the cell surface. Future cells will bemultilayer structures with each layer thicknessdesigned according to the electrical properties ofthat layer and the other layers.

Within FSA, the efforts being implemented aredirected toward gaining a better understanding of thephysics of carrier losses and their influence on celldesign by the use of comprehensive modeling, by thedevelopment and use of new and improvedmeasurement techniques, and by new and improvedcell fabrication techniques. The surfacerecombination velocity of a solar cell (a measure ofsurface losses) is generally greater than 105 cmlsAn effort is under way to reduce the surfacerecombination velocity, to less than 103 cmls by·addressing critical problems of controllingrecombination losses at surfaces and interfaces (forexample, the investigation of transparent conductingmaterials). To better understand and control surface­interface phenomena, surface passivants (such asSiNx, polycrystalline silicon, and ultra-thin oxidelayers) are under development. Also, reliablemeasurement techniques to characterize passivants,surfaces, and interfaces are evolving. As part ofsurface loss investigations, work on transparentconducting materials is emphasizing study of theinteraction of a polymer with a silicon surface andthe effects on optical and electrical characteristics.Studies to understand and control bulk lossmechanisms by improvements in silicon sheetstructural, electrical, and chemical properties areleading to improved minority carrier diffusion lengths.The complicated and varied nature of sheet is afunction of its growth conditions and its processing,including solar cell fabrication.

Characterization of bulk material and deviceperformance evaluation are using or will usea varietyof measurement techniques for chemical, structural,and electrical characterization. These include opticalelectron and ion beam analyses, X-ray characteriza­tion, light and dark current-voltage relationships,spreading resistance, diffusion lengths by surfacephotovoltage, scanning electron microscopy, deep­level transient spectroscopy, secondary ion micro­probe spectroscopy, and Zeeman spectroscopy.

Measurement techniques are being developed thatare critical to gaining a basic understanding of lossmechanisms, to correlating model parameters withmeasured ones, and to optimizing cell performance.For example, the short-circuit current-decay (SCCD)method (Figure 25) has been developed at the Uni­versity of Florida. It simultaneously measures thecarrier lifetime in the bulk of the cell and the recom­bination velocity at the back surface interface. Withthis method, a forward bias is applied to the solarcell to set up a steady-state condition. Then, byrapidly applying a zero bias across an extremely small

40

resistance, a short circuit is obtained. This causesthe pin junction region and the base and emitterregions to discharge. If the discharge times relatedto the junction and emitter regions are much smallerthan the time related to the base region, the lifetimeand recombination velocity can be determined.

PULSER3 GENERATOR

R1

LOGAMPLIFIER

Figure 25. SCCD Cell Loss Mechanism MeasurementMethod

One of the main advantages of this technique isthat the discharge time of the pin junction (on theorder of 10-11 s) is much less than the dischargetimes associated with either the emitter or the base.This makes it very easy to interpret the observedtransient current. The SCCD method is simple, fast,and less expensive than methods used.to measureother key solar cell parameters. It can be routinelyused during cell fabrication steps. It is beingevaluated as a possible method for determiningcarrier lifetimes and recombination velocities in thethin emitter region.

A highly sensitive microwave reflectancemeasuring system with computer data acquisitionhas been designed and built at JPL. The experimentalresults illustrate that the m,icrowave technique canprobe accurately the decay of photogeneratedminority carriers in silicon sheet materials thatdepend on the minority carrier lifetime and surfacerecombination velocity. Under controlled surfaceconditions, one can measure bulk minority carrierlifetimes in silicon sheets, such as silicon wafersgrown by Czochralski and floating-zone techniquesand dendritic web ribbons. With known lifetime in thebulk material, the technique has been demonstratedto be an effective tool for monitoring the surfacerecombination velocity. Preliminary results reveal thatthe silicon surface polished by Syton has a smallsurface recombination velocity when it is freshlyetched in a diluted hydrofluoric solution, but therecombination velocity increases drastically when thesurface is exposed to the air. The effect becomessaturated after several hours. Currently, JPLscientists are working on an adequate analysismethod to extract quantitative results fromexperimental data. The microwave technique isreliable, nondestructive, effective, and easy to use. It

can be used on as-grown and processed sampleswithout any contact.

Electron beam-induced current technique tomeasure minority carrier lifetime in n + and surfacerecombination velocity at n + front surface is beinginvestigated.

More complete analytical solar cell modeling andanalysis capabilities development has led to theestablishment of a program [Solar-Cell EfficiencyEstimation Methodology and Analysis (SEEMA)] thatuses computer models to evaluate and designadvanced silicon cells. After an extensive search,FSA located two models to use in SEEMA. One ofthese, Solar-Cell Analysis Program in One Dimension(SCAP1 D), was designed by Purdue University. It is ageneral purpose program used to analyze and designsilicon solar cells. The other SUPREM-II, wasdesigned by Stanford University. It is used to calcu­late information about the emitter doping profile whichis then used as input to the SCAP1 D model.

Researchers in FSA validated the utility of SCAP1 Dby evaluating the metal-insulator-nIp junction (MINP)structure of the high-efficiency silicon cell reportedby Martin Green of Australia. Using experimentaldata provided in the literature as input, SCAP1 D'sevaluation closely agreed with Green's experimentalresults. Researchers also used SCAP1 D to perform asensitivity analysis on the structure by changingvarious design parameters one at a time, whileholding the other parameters constant. Their analysesindicated that a cell with the MINP structure couldobtain 20% efficiency by increasing the bulk minoritycarrier lifetime from its present 20 ~m to 250 fLm.This would make the diffusion length much greaterthan the cell thickness of 280 mm (Figure 26). Underthese conditions, using a back surface field wouldfurther improve the cell's efficiency.

For future use, JPL is establishing improvedversions of the two models. Purdue's upgradedmodel, SCAP2D, performs two-dimensional analysis.This will enable researchers to analyze complicatedstructural designs that are difficult to reduce toone-dimensional equivalents. Stanford's newSUPREM-1I1 also has additional features forcalculating electrical properties after each processstep.

Recent Accomplishments

• New techniques to simultaneously measure bulkminority carrier lifetime and back surfacerecombination velocity developed, including SCCDand microwave techniques.

• Key barriers to obtain high-efficiency cellsidentified by use of comprehensive numericalanalysis techniques.

• Use of new passivants for front and back cellsurfaces showing promising results.

• High-efficiency solar cells modeled.

• Theoretical analysis of a new cell design, called"Floating Emitter Solar Cell Cell Transistor,"indicates that the cell has the potential to achievegreater than 20 % efficiency.

Research Needed

• Establish relationships between residue defectsand recombination mechanisms in highest qualityavailable silicon sheet.

• Investigate recombination mechanisms in low-costsilicon sheet in bulk, in all layers, and at allsurfaces.

• Develop technique for measurement of frontsurface recombination velocity.

• Establish correlations between material defects,device fabrication steps, and device performance.

• Develop new higher-efficiency device concepts.

• Improve modeling capabilities.

Lifetime (J.1 s)

20

19

20

16 • Continue characterization of silicon sheets anddevices as their quality improves.

20 50 75 100 200

Cell Thickness ( J.1 m)

ED Continue device modeling and experimentalverification of high-efficiency device structures.

Figure 26. Sensitivity of Efficiency to Solar CellThickness for Various Minority-Carrier Lifetime Values

• Investigate cell efficiency improvements for newadvanced cell processes.

41

Process Research and Development

In 1975, the processes for fabrication of solarcells, for use on Earth, were either those used formaking space solar-cells or their derivatives. Thesetechniques, borrowed from diode productiontechnology, were very labor intensive and materialintensive with respect to the total product price. Acidetching and cleaning steps were extensively usedalong with photolithographic grid-line definition.

The performance of individual solar cells was notconsistent. Fabricated in batches, even theperformances of those cells for use in space variedsignificantly. The best of these small rectangular"space cells" were assembled on panels, as closelypacked as possible, to produce the maximum powerper unit area. The lower performance cells were notused in spacecraft power systems because of high­performance requirements. Instead, these rejectspace cells were used for terrestrial applications. Bythat time, solar cells made specifically for terrestrialuse had become competitive with reject space cells.These circular cells were made from one completeCz wafer. The cell processing was similar, butsimpler and less expensive. For example, semi­conductor industry-reject wafers were used. Theresult was that terrestrial solar-cell performance wasless than space-cell performance. The challenge in1975 was to reduce the cost of terrestrial solar cells,but still retain good quality and performance.

The objectives are to develop cell fabrication andmodule assembly process technologies required tomeet cost and performance requirements forterrestrial PV power production.

The goals were to:

• Identify and develop low-cost cell and moduleproduction processes.

• Design facilities and equipment to perform theprocesses.

• Demonstrate fabrication of low-cost PV cellsand modules in a pilot line environment.

• Implement the transfer of mass productiontechnology to industry.

The goals now are to develop processes to:

• Fabricate high-efficiency cells from availablesilicon sheet substrates.

• Assemble cells into higher-efficiency longer-lifemodules.

42

The key technical issues are to develop processesso that:

• Silicon sheet substrate quality is not degraded.

• A variety of high-efficiency cell designs can befabricated.

Background

Crystalline silicon solar cells were invented at BellLaboratories in the mid-1950s. These solar cells orsolar batteries, as they were called, are still the mostcommonly used solar cells. A solar ,cell is fabricatedby a series of processes, from a silicon sheet dopedwith either an n or p type impurity. The othertype of impurity (p or n) is then added to anextremely thin (about 0.3 fLm) layer on the surfacethat will be the top of the cell. This creates the nipjunction (or pin) where the photovoltaic effect canoccur in the solar cell. The addition of metallicelectric current collectors, on the top and on thebottom of the wafer, permits electricity that isgenerated within the cell to flow from the cell. Thetop current collector consists of uniformly spacedfine lines that minimize the amount of sunlight that isblocked from entering and from being absorbed bythe solar cell. Cell fabrication, although it seemssimple, is sophisticated with improvements inprocess and equipment continuing even after30 years of progress. Completed solar cells areelectrically interconnected and then encased to forma module that supports and protects the cells.

Project Activities: 1975 to 1981Within FSA, the research and development

required for solar cell fabrication and moduleassembly were conducted by a group familiar withthe processes and equipment used to produce solarcells and semiconductor devices. A plan consistingof five phases, which was compatible with the otherFSA activities, was devised and implemented. Theplan consisted of:

I. Technology Assessment: Determine availabilityand applicability of cell and module processes.

II. Process Development: Develop applicable oldand required new processes.

III. Facility and Equipment Design: Developautomated equipment and facility designs.

IV. Experimental Plant Construction: Incorporateprevious efforts into a pilot line to define yieldsand remaining key cost factors.

V. Conversion to Mass Production: Transfertechnology to industry.

The Technology Assessment contracts awarded toMotorola, RCA and Texas Instruments in early 1976assessed the state of the art of PV processtechnologies in terms of cost large volumeproduction potential, ability to be automated,reliability, and process synergistics. Identification ofthose processes with the most potential for cost­effective implementation and recommendations forfurther study of those processes were alsoperformed. More than 50 processes covering allareas of cell fabrication and module assembly werestudied. Processes evolved that clearly distinguishedterrestrial cells from space cells. As an example,copper metallization was identified as being morecost effective for top and bottom electricalconductors than either solder or silver. Metallizationmethods were expanded from space-orientedvacuum evaporation to include electroless plating,electrolytic plating, thick film metal (printing), andreflow solder. Wafer surface treatments wereexpanded to include brushing, plasma etching,texture etching, and new cleaning solutions. Junctionformation, in addition to the standard space cellprocess (gaseous diffusion), was expanded to includespin-on and spray-on of liquid dopants, solid dopantdiffusion, doped oxides, ion implantation, andadvanced ion implantation. Other good candidate cellprocesses that were added included silicon nitride(CVD), oxide growths, mechanical edge grinding, andlaser scribing. Assembly methods included the useof: glass substrates, glass superstrates, andconductive adhesives; new encapSUlation materialssuch as silicones, PVB, EVA; and new cellinterconnection methods. Selected processes werestudied regarding their "sensitivity to variables" suchas purity or condition of input materials and suppliesand processing parameters, primarily time andtemperature. Efforts at JPL were concentrated onprocess verification and synergisms along with cellmetallization analysis programs.

The problem facing the Project rather early in itsinception was how to compare the potentialproduction costs of competing processes andprocess sequences being investigated by a numberof different researchers. This was clearly brought outduring the Technology Assessment phase in whichcost comparisons between the different contractorson the same processes were significantly different.This pinpointed a need for a standard methodologythat allowed relative comparisons of the potentialproduction costs attributable to competingprocesses, and an estimate of the actual cost tomass produce using a specific process.

Three key costing analyses methods developed byFSA PA&I personnel are:

• Interim Price Estimation Guidelines (IPEG), acosting system simple enough to be run on ahand calculator. This method uses standardcoefficients for costing inputs such asequipment, direct labor, material, floor space,and utilities.

43

" Solar Array Manufacturing Industry Simulation(SAM IS), a factory simulation which runs on amain frame computer. This document requiresthe filling out of detailed costing sheets calledFormat As which are then fed into a computerfor accurate and consistent process costs.

• Solar Array Manufacturing Industry CostStandards (SAMICS), a catalog of materialcosts, labor costs, and other input costs whichassisted the contractors in inputting standardcosts.

These three techniques were verified andupgraded continually throughout the life of theProject and used extensively.

At the conclusion of the assessment effort, it wasdetermined that cost effective processes should:

" Consume a minimum of material and supplies.

" Use low-cost, recyclable supplies.

• Have high production yields.

" Result in high-efficiency modules.

Before the assessment effort, automation andcapital equipment were thought to be major costfactors. It was shown that when labor is a significantcost factor in a process, the process should beautomated. Most semiconductor processes were, toa significant extent, already automated and,generally, capital costs for equipment could bedistributed over large production runs resulting in lowcosts per unit.

Requests for Process Development proposalswere distributed to the entire PV industry and werebroadly scoped to include new and novel processesand those processes identified in the TechnologyAssessment Phase as requiring further development.Also, unsolicited proposals from PV contractors wereencouraged, reviewed, and acted upon. Twenty PVindustry contractors and two universities participatedin these studies, which reflected the thinking of thePV industry at that time. A partial review of some ofthe contractors and their diversified activities duringthis phase are:

" Applied Solar Energy Corp.: Low-cost copper­plated contacts; ion-planted cell evaluation.

" Lockheed Missiles and Space Division:Sprayed-on fluorocarbon encapsulation.

" Mobil Solar Energy: Module assembly.

• Motorola: Antireflective (AR) coatings;palladium/nickel/soldermetallization.

" Sensor Technology (later called PhotowattInternational): Nickel/copper metallization.

Ell RCA: Expitaxial cell processing.

Ell Solarex: Wafer thickness studies; solar breederstudies; high efficiency module; nickel/soldermetallization.

Ell Sollos, Inc.: Nickel metallization.

Ell Spectrolab: Thick-film metallization.

Ell Spire Corp.: Ion implantation; pulsed electronbeam annealing.

Ell University of Pennsylvania: Process ar:lalysisand evaluation.

Ell Westinghouse: Cell processing using dendriticweb.

All contracts during this phase required processcost evaluations to be made using the costingmethodology and the writing of processspecifications. The costing documentation includedall of the direct inputs for equipment cost andperformance; required labor, floor space, utilities,maintenance, supplies and input materials for eachprocess. The process specifications included detaileddescriptions of equipment, chemicals and materialsused, and process parameters including time,temperature, pressure, etc.

Selection of an individual process depended uponwhat other processes were used before and afterthat process. Accordingly, an additional effort, thedevelopment of process sequences, was initiated.Discrete process steps were combined to formprocess sequences that were synergistically soundand cost effective, such as shown in Figure 27.Contracts were awarded to develop and demonstrateviable, cost-effective sequences. It became clear thatthere was no one single process sequence that wasmost cost effective, but rather a number of processsequences could be demonstrated to be viable andcost effective. Photowatt (then called SensorTechnology) demonstrated a process sequence with

spray-on dopant junction formation and printed metalmask followed by plating; Spectrolab demonstrated aspin-on dopant junction formation and a thick-filmmetallization system; Motorola demonstrated ionimplantation junction forming, a CVD silicon nitrideAR coating and plasma etching; Solarexdemonstrated cost-effective processing ofpolycrystalline material; Westinghouse indicated thatconventional semiconductor processes such asgaseous diffusion and vacuum metallization could bemade cost effective with large volume productionequipment.

Collectively, the process sequences showed theresults of the cross-fertilization of technology thatFSA emphasized. Copper metallization was used intwo of the sequences. RCA cleaning technology wasused extensively, as was laser scribing and thick filmsystems. The sequences illustrated the large numberof possible low-cost process combinations that couldbe used to make a cell. The process sequencesdemonstrated that there was a new breed of solarcells, terrestrial solar cells, fabricated distinctlydifferent than the space cell from which it evolved.

Most of the process R&D efforts have beenconducted by private industry and universitiesthrough subcontracts with FSA which, to date, haveincluded over 60 contracts. A parallel effort was thecreation of a Process Research Laboratory at JPL inwhich the contractor process work could be verifiedand, in many cases, directly supported. Independentresearch has also been conducted in the laboratoryby JPL personnel.

A major tool used to transfer the evolving processtechnology to the entire PV industry and to obtaincross fertilization has been the distribution of processspecifications for verification. Near the end of theProcess Development Phase, more than 140 processspecifications covering cell and module fabricationhad been documented and distributed to interestedPV industry members. A feedback request resulted inindustry verification of more than 40 of theseprocesses. The results of these efforts are shown in

TYPICAL SEQUENCE 1977 TYPICAL $2.80/WATT SEQUENCE* TYPICAL $O.70/WATT SEQUENCE-

METALLURGICALSILICON

GROW INGOT

PHOTORESIST

GOLD PLATE

SOLDER COAT

CLEAN FLUX --------SILICON PREP SHEET FAB -------­MOOULE FAB

'--------..---- "-----.--' '---.----' "---..------'SILICON PREP SHEET fAB CELL fAa MODULE FAD

'----...------' ~ '----.---' '-------"..-'SILICON PREP SHEET FAa CElL fAR MODULE FAB

Figure 27. Modu/e Process Sequences

44

Figure 28. The JPL Process Research Laboratory wasalso very much involved in the process verification

effort by providing a place for supplementaryexperimentation and verification.

An important function of FSA is to promote theadvancement of the photovoltaics industry by the transferof technology

~I~

Former Technology TransferProcedures circa 1981

Processing TechnologyTransfer Status

1981

PIM

r-_S~~IREFf-rI I Iii

Available

Surface Preparation 19Junction Formation 18Metallization 13Module Assemblv 16

Totals 66

• VERIFY SELECTEDPROCESSES

• ASSESS SENSITIVITYTO PROCESS VARIABLES

• VERIFY ECONOMICDATA fROMCONTRACTOR

UnderEvaluation Confirmed

7 54 56 35 0

22 13

Technology transfer continues but in a lessformal manner

RECENT TYPICAL TECHNOLOGY TRANSFER

• MEL T REPLENISHMENT EQUIPMENTWestinghouse Electric Corp. is using equipmentdeveloped by Kayex Corp. and modified at Westinghouseto replenish silicon in a dendritic·web growth process.

• CulnSe2 MODULE LAMINA TlONBoeing Eng'g & Construction used the JPL ProcessResearch Task laboratory's equipment and processesto laminate CulnSe2 cells.

• THERMALL Y INSULATED TOOLING FOR LAMINATORSSunelco Corp., Tideland Signal Corp., Solarex Inc. andBoeing Eng'g & Construction have shown great interest inthermal insulation techniques used in the JPL ProcessResearch Task's laminator (capable of producing 4 x 4·ftmodules) and are adopting them.

• TERRESTRIAL MODULE PRODUCTION LINETideland has set up a production operation in Australiabased on processes transferred from Spectrolab, Inc.,developed under contract with JPL.

• UNSUPPORTED ("CREDIT·CARD") MODULE LAMINATIONThe first unsupported (no superstrate or substrate)module was fabricated by Spectrolab using the JPLProcess Research Task's 4 x 4-ft laminator.

• CELCAL PROGRAM AVAILABLE FROM COSMICAn engineer-interactive computer program for solar·cellmetallization design has been transferred to and isavailable from Computer Software Management andInformation Center (COSMIC).

• AUTOMATED ROLLING-SPOT BONDER CELL STRINGINGKulicke and Solla, Inc., has built an automated machinefor Westinghouse based on a concept developed undercontract with JPL.

• THICK·FILM INK AVAILABILITYSilver and aluminum thick·film inks are available fromElectrink, Inc., based on formulations developed underJPL contracts.

Figure 28. Technology Transfer to Industry

45

The important assembly equipment that evolvedfrom the near·term effort were the automated cell

In late 1979, a bill sponsored by Senator Tsongasof Massachusetts provided additional funds for near·term reduction of solar energy production costs bysoliciting ideas from industry. The following contractswere implemented:

Organization

ARCO Solar

Kulicke & Soffa

Motorola

Photowatt

Sollos

RCA

Activities

Automated cell interconnection

Automated cell interconnection

Wax patterning.Thin cell processing

Polysilicon cell processing;surface texturizing

Thick·film metallization

Megasonic cleaning

interconnect machines by ARCO Solar and Kulicke &Soffa and a vacuum laminating system by ARCOSolar. First·generation machines are shown inFigure 29.

The Facilities and Equipment Design Phase wasnever heavily stressed because most of thistechnology was already being developed for non·PVsemiconductor devices. In addition to the specialnear·term funding, the development of otherprocess equipment was funded. Important processequipment developments were ion implantationequipment for large area junction formation and apulsed electron beam wafer annealing unit by SpireCorp., and a robotic cell interconnection andmodule assembly system by Tracor MB Associates.The robotic equipment was especially meaningfulbecause it demonstrated that cells could be handledrobotically in large·volume quantities withoutbreakage. These machines demonstrated thepotential for high·volume production.

AUTOMATED SOLAR CELL ELECTRICALINTERCONNECT (SERIES) ASSEMBLYBY KULICKE AND SQFFA

1ST INTERCONNECT SOLDER

2ND INTERCONNECT SOLDER

EDGE SEAL AND FRAME ASSEMBLY(TO BE DEVELOPED)

Figure 29. First-Generation Module Assembly

46

In late 1980, two Module Experimental ProcessSystem Development Unit (MEPSDU) contractswere awarded to provide the first controlled pilot­line data on cell and module processing. They wereto reflect the culmination of the technology gainedfrom the earlier phases. Prior to this time, thecontractors produced cells and modules in alaboratory environment and then extrapolated thedata for input into the SAMIS factory simulationprogram. In the MEPSDU contracts, each of the twocontractors was required to construct and operatea pilot line. The contractors were required to recordthe production rate, yield, and process parameters.Three technical demonstration runs were required,two of which allowed no major adjustments,modifications, or repairs during and between runs.Solarex, Inc. and Westinghouse were competitivelyselected for the MEPSDU contracts. The twocontractors chose diametrically oppositeapproaches as shown in Figure 30. The Solarexprocess sequence would produce lower efficiencycells using low-cost processes and polycrystallinesilicon substrates. The Westinghouse processsequence would result in higher-efficiency cellsusing more expensive processes. The Solarexprocess sequence used semicrystalline siliconwafers, whereas the Westinghouse processsequence used dendritic web silicon. Ultrasonicbonding of cell interconnections to the solar cellswas to be used by Westinghouse and solderedconnections by Solarex. Vacuum lamination was the

OBJECTIVE: VERIFY PROCESS AND EQUIPMENTCONCEPTS AND TRANSFER TECHNOLOGY TO INDUSTRY

proposed module assembly technique for bothprocesses.

However, within 6 months the MEPSDU contractswere reduced in scope because of Project budgetreductions. Within a year and a half, the MEPSDUefforts were cancelled entirely and the contractorswere redirected toward research-oriented efforts inaccordance with new Project guidelines.

Project Activities: 1981 to Present

The first three phases were completed, or almostcompleted, as planned with significant reductions incell and module fabrication costs. Projectredirection starting in 1981 resulted in cancellationof the MEPSDU and Mass Production Phases.

Good progress had been made toward meetingthe individual parts of the original Project goals.However, an end-to-end integrated sequence wasnever accomplished. Efforts were then directedtoward basic process research and, later, towardhigher-efficiency cell process research. With newemphasis on higher efficiency and longer life,development of more advanced and sophisticatedtechnology was initiated. Also, an examination ofthe needs of the PV industry, at that time, and theresults of the earlier efforts led to the conclusionthat metallization and overall process yields weretwo key items in which efforts should be continued.

APPROACH: DESIGN AND DEVELOP EXPERIMENTALPROCESSING EQUIPMENT WHICH WILL HANDLE BOTHFIRST AND SECOND GENERATION MATERIAL.

r==r I\ \ \ \ \ \~N c; F~

\\\\\\(I P I P

SI'R.AY ONP+

DOPANT DIFFUSION HEAT ALUMINUM PRINT SINTER

ANTI·REFLECTION (A·A)

t P L~: t~ E~RGASEOUS P GASEOUS

P- P-

DIFFUSION DIFFUSION ETCH GLASS DIP COAT

~US~Cl~~~ ISO~n~~NA.R S~LDERM f~~ MP0:l~:MASK MASK ANTI·REFLECTION IA·RI

~.N [r:tS:lf ~: r! NII P E=:+P+ Ni NiLASER SCRIBE Ni CLEAN & SOLDER DIP NICKLE PLATE ETCH PRINT MASK SPRAY COAT

JU~RCu A-R A-R t::::ti:

PR rHOTORESiST (PfrRRm'ft EL E~ r E3

:1~:P- II ~_'-

p.PR-METAL

LASER SCRIBE COPPER PLATE REJECTION METALLIZEIVACUUM) EXPOSE-DEVElOp·ETCH DIP COAT

CELL STRINGING IULTRA-SONIC BOND)

Figure 30. Module Experimental Process System Development Unit

47

The Westinghouse MEPSDU effort was redirectedto investigate liquid-dopant processing for junctionformation, which ultimately replaced the gaseousdiffusion process on the Westinghouse pilot line. TheSolarex MEPSDU effort was redirected towardinvestigating process mechanisms affecting lifetimeand performance of cells made from polycrystallinematerial. A second generation of automated cellinterconnection equipment evolved from theWestinghouse MEPSDU effort. The initial effort wascost-shared with JPL and, after MEPSDUcancellation, was continued to completion byWestinghouse. Kulicke and Soffa, under contract,developed, designed, and fabricated the ultrasonicbonding machine, which is capable of .interconnecting one cell every 3.5 seconds or a 180cell module every 10.5 minutes. The machine isshown in Figures 31 and 32. Figure 33 shows inter­connected dendritic web cells.

Figure 31. Second-Generation Cell Stringing Machine

"\MOLECULAR BOND.\

Figure 32. Rolling Spot Ultrasonic Bonding

48

Figure 33. Ultrasonic Interconnect Bonding ofDendritic Web Solar Cells

Metallization was determined to be the majorremaining higher-cost factor for cell processing.Figure 34 shows the results of cost analysis effortson metallization systems. A number of innovativemetallization ideas were generated as a result of theCell Metallization Research Forum held in 1983. Theobjectives of the Forum were to clarify and definethe state of the art of current solar cell metallizationsystems to: report on performance experience withthese systems, including advancements made andproblems encountered; describe advancedprocessing techniques under development; and todiscuss expectations for future improvements.

Process control became a major area of concernas industry yields were not as high nor as predictableas had been expected. Industry proprietary concernsand lack of large-scale engineering pilot lines limitedthe availability of data. To overcome the lack ofinformation and to explore processing for higher­efficiency cells, process and device modelingcapabilities were investigated. For example, SUPREM(developed for the semiconductor industry byStanford University) provides a comprehensivecomputer process simulation program that allows theprediction of device structures from any fabricationtechnology sequence. An example of a device modelis SPCOLAY, by the University of Pennsylvania, aone-dimensional PV cell model which has beenparticularly useful in defining the interactionsbetween cell substrate material parameters andpreferred cell processes. The soft boundary shown inFigure 35 defines the "Domain of Practical Signifi­cance" for those characteristics which are prevalentin present state-of-the-art materials.

Candidate Processes/Systems 24.--,.---,---,---,--,,.--,----.,.-----,

400

"DOMAIN OFPRACTICALSIGNIFICANCE"

BSF THICKNESS, f..Lm

10NO BSF

~10::.:::::::;;;;;;;;--~ 5

"ff~&iJ#oP"{ 10.1

NO BSF

1010.10.01

NO BSF

BULK RESISTIVITY, D.·em

22

>-"~ 20LU

Uu::u..LU

ZClc;:;a:LU

~ 18Clu

Figure 35. Low- and High-Bulk Resistivity CellBehavior; Effects of Bulk Resistivity, Diffusion;Length, and Back Surface Field

2.80/W, Block IV"

JPl BPU"2.80/W, Block IVJPl BPU

JPl BPU, Sol/los

OATA SOURCE

ASECWestinghouse

Solarex, Motorola

Motorola

Spectrolah"

Spectrolab, Sol/los

Illinois Tool Works

$90$14

$9.4

$8

COSTlm2

$8 to $35$7 to $34

$2.4$1.4

$2.0

$5 to $16$2

$6

PROCESS/SYSTEM

Current Results

Evaporation (EV)

• SOA (TiIPd/Ag)• Advanced lTilNi .-to Cu plating')

Screen print (PRINT)

• SOA (Ag paste)• 1g90 (Ag paste')

• SOA (AI paste)• 1g90 (AI paste')

• 1990 (Mo/Sn')

Electroless plating

• SOA (Print resist, Ni·plate, Sinter,Wave solder)

• Advanced (PR, Ni plate, Sinter, Cuplating')

Midlilm' (Ag) (MID)

Midfilm' (MoISn)

Ion plating' (Ti/Ni/Cu) (ION)

.. Advancement of Photovoltaic state-of-the·art"Ag price from $10 to $5010'

PRINT RESISTELECTROLESSNilSOLOER 0

FRONT METALLIZATION COST. $lm2

PURPOSE OF THIS ANAL YS/S:

• Compare costs and effectiveness of state'of-the'art (SOA) metallization and projectedmetallization approaches

• Estimate the potential impact of R&D in thisarea

APPROACH:

• Use data from FSA contractors as the majorsource for calculating cell metallizationcosts with IPEG2

• Use JPL's grid optimization model toestablish electrical performance ratiosbased on geometrical and electrical data(ratio to a cell without resistive or shadowlosses)

• Develop metallization cost vs performancetrade'off slopes for several PV systems; thecell metallization system farthest above atrade·off slope has the greatest value

• Study as yet has not included processcompatibility, reliability, or unconventionalmetallization systems

Figure 34. Metallization Cost Comparison

The term "directed energy research" was coinedto describe the increased emphasis on high­efficiency processes, which covers the use ofmicrowave, laser, and thermal pulse energy sources.All of these new energy sources are of technicalinterest because they allow many cell processes tobe performed without heating the bulk of thesubstrate. PV cell fabrication by "cold" processing isused to avoid degradation of bulk substrateproperties, particularly minority carrier lifetime.

Results from directed energy research have beenvery gratifying. Westinghouse demonstrated thatthermal pulse (heat lamps) can be used tosimultaneously form nand p + junctions using liquiddopants. Previous processing required morehandling, cleaning, and the use of expensive oxidemasks. Excimer lasers can be scaled up and havebeen used by ARGO Solar and Spire for annealingion-implanted junctions. Superwave Technologydeveloped a microwave-powered plasma systemused for thin-filmed silicon nitride deposition as wellas a passivating silicon dioxide layer. Argon laserswere used by Westinghouse in a successful newmetallization method in which lasers scan waferscovered with metal, bonding the metal to the solarcell where the "laser writing" has occurred,eliminating the costly photolithography process(Figure 36).

49

SCANNING MIRRORS J :4==== Ar+LASER BEAM

"""=::::::1t=:::::=;;l> LENS

DEPOSITEDMETAL FILM

SILICON______ SUBSTRATE

SAMPLE BASE TEMPERATURE 75°CFOCUSED LASER SPOT DECOMPOSES SPUN-ON FILMMETALLIZATION PATTERNS ARE FORMED BY DIRECT WRITING

Figure 36. Laser Pyrolysis of Spun-On Metallo­Organic Film

One of the new metallization research effortsfunded was the development of metallo-organicdecomposition (MOD) films by Purdue University. TheMOD process uses liquid compounds containingmetal atoms bonded to organic compounds such asneodecanoic acid. Figures 37 and 38 illustrate thewide variety of possible compounds that can besynthesized for research or for use in production.Decomposition of these compounds by theapplication of heat between 250 °C-300 °C results inthe formation of fugitive gases such as carbondioxide and deposition of a thin layer of metal. Manydifferent metals have been successfully tried such astitanium, chromium, silver, bismuth, and palladium.Combinations of metals have also been successfulwith especially good results obtained from a bismuth­silver formulation. Choice of organic acid and variousviscosity adjusters allows application of MOD inks byspinning, dipping, spraying, screen printing, and inkjet printing. Multiple layers have also been success­fully deposited, and lines as narrow as 3 fLm havebeen achieved.

HETERO-ATOM(0, S, N, P, As)

+I COMMON SOLVENT

+

I RHEOLOGY ADJUSTORS III

~+

PRINT, SPRAY, DIP, ETC

I RAW FILM I+

I HEAT III

ICONDUCTOR I

Figure 37. Metallo-Organic Deposition Processing

50

SUBSTITUENTS I % ABUNDANCE

R1 = CH3

R2 = CH3 31

I-- R3 = C6H13

R1 = CH3

R2 > CH3 67

R3 < CH6H13

R1 > CH3

R2 > CH3 2

R3 < C5Hl1

Figure 38. Alternative Formations of Metallo-OrganicDeposition Precursor Chemicals

Today's Status

Low-cost processes have been developed for celland module fabrication steps such as:

• Surface Preparation

.. Hot alkaline damage etch.

.. Spray, dip, roller, spin, and meniscus coatingof anti reflective films.

• Junction Formation

.. Gaseous diffusion of 6-in.-diameter cells.

.. Liquid-dopant junctions.

.. Liquid-dopant back surface fields.

.. Aluminum-paste back surface fields.

• Metallization

.. Silver thick-film ink systems.

.. Copper cell metallization systems.

.. Pioneering of molybdenum-tin inks.

• Module Assembly

.. Ultrasonic cell interconnection.

.. Automated cell stringing.

.. EVA encapsulation processes.

.. Double-chamber vacuum lamination.

The above processes seem to have the potentialto meet the original Project goals.

A complete processing sequence using all thesteps, the individual yields, and the process para­metric sensitivities was not completed.

The need for higher-efficiency and longer-lifemodules, but manufactured at essentially the samecosts has imposed the need for new, more advancedtechnology and processes.

Accomplishments

The following is a list of accomplishments withthe associated contractor(s) credited.

Processes

ED Surface Preparation

• A cost-effective etching processing for sawdamage removal prior to junction formation(Sensor Technology, Lockheed).

• Anistropic etching process to produce largearea textured surfaces for enhanced photoncollection (Motorola, Sensor Technology,Solar Technology).

• Developed a number of antireflectioncoatings with concurrent processes,including mUlti-layer evaporation, spraying,dipping and spinning, and the requiredmaterial development (RCA, Westinghouse).

• Polycrystalline silicon processing (Solarex).

• Microwave enhanced CVD (Superwave).

• Surface passivation of front emitter areasusing silicon dioxide, silicon nitride, andsilicon oxynitride (Motorola, Westinghouse,Spire).

e Megasonic cleaning (RCA).

e Plasma pattern etching (Motorola).

ED Junction Formation

e Large area, large volume, gaseous diffusionprocessing using POCls, and PHs and BCls(Sensor Technology, Westinghouse).

e Simultaneous front and back junctionformation using liquid dopants and infrared(IR) rapid heating pulse techniques(Westinghouse).

• Ion implantation of both front and backjunction at a rate of 1200 wafers/hour (Spire).

e Non-mass analyzed ion implantation of bothback and front junctions (Motorola, Spire).

51

• Laser annealing, electron pulse beamannealing, and rapid thermal pulse annealingof ion implanted junctions as well asconventional thermal annealing to maintainlifetime of the initial silicon material (Spire,ARCO Solar, Lockheed).

• Spin-on, spray-on, and meniscus coating ofliquid dopants (Spectrolab, SensorTechnology, Westinghouse).

• Laser scribe (Sensor Technology).

ED Metallization

e Thick film screenable cost-effectivemetallization processes using Ag, AI, AgAI,Cu, and MOD AgBi (Spectrolab, Bernd RossAssoc., RCA, Purdue, Electrink, Sollos, Solec).

• Metallo-Organic decomposition films (MODfilms) using low-temperature processes« 300°C) (Purdue, Electrink, Westinghouse).

• Reliable plating systems using Pd and Nifollowed by solder build-up by immersion(Motorola, Solarex).

• Pyrolytic decomposition of MOD films using alaser as the thermal source (Westinghouse).

e Development of a generic fabricationprocess for producing MOD film (Purdue,Electrink).

• Diffusion barriers (Caltech).

ED Module Assembly

• Ultrasonic bonding of interconnects(Westinghouse, Kulicke & Soffa).

• Automated soldering of interconnects (ARCO,Kulicke & Soffa, Foothill Engineering).

• Laminated encapsulation in a partial vacuum.

e PVB and attendant processes for 4 x 4 ftmodule (ARCO Solar, RCA).

e EVA and attendant processes (RCA).

e PV shingle assembly (General Electric).

e Silane primers, antioxidants, and UV screen­ing agents (Springborn Laboratories)

• Electrostatic/ultrasonic edge sealing (Spire)(see Figure 39).

Figure 39. Hermetic Edge Sealing

ALUMINUM TOALUMINUM SEALBY ULTRASONiCWELD

\

/ALUMINUM TOGLASS SEALBY ES8

ANTICIPATED BENEFITS

III REDUCED H20 VAPORTRANSMISSION

III REDUCED 02 TRANSMISSION

III REDUCED AIRBORNEPOLLUTANT TRANSMISSION

GI UV AND TEMPERATURESTABLE SEAL

ENCAPSULANT

SOLARCELL

GLASSSUPERSTRATE

APPLICATIONS

III MODULES IN SEVERE ENVIRONMENTS

- COASTAL AND MARINE SITES- HOT AND HUMID REGIONS

GI REMOTE LOCATIONS

- REDUCED MAINTENANCEAND REPLACEMENT COSTS

GI MODULES WITH SENSITIVECOMPONENTS

GI EXTENDED LIFETIME FORSTANDARD MODULES

• Automated cell stringing machine (ARCOSolar, Kulicke & Soffa).

• Ultras.onic automated cell stringing fordendritic web (Westinghouse, Kulicke &Soffa).

• A laminator capable of producing large areamodules (JPL, ARCO Solar, Spire, ASEC).

" Robotic module assembly (TracorMB Associates).

Analyses

Ell SPCOLAY (University of Pennsylvania). PV cellmodelling program for investigation of processvariable sensitivities.

Equipment

Ell Surface Preparation

• Megasonic cleaning equipment (RCA).

" Microwave enhanced CVD System(Superwave).

Ell Junction Formation

" Ion implanter with 4-mA capability forphosphorous and throughput of 400 wafers/h(Spire).

" Non-mass analyzed ion implant equipment(JPL, Motorola, Spire).

" Pulsed electron beam annealing system (Spire).

• Laser annealing system (Spire, ARCO Solar,Lockheed).

• Meniscus coating machine for dopantadditions to dendritic web material(Westinghouse).

Ell Metallization

• Ink jet printer-computer controlled (Purdue).

• Argon ion laser decomposition system(Westinghouse).

" Dry film/mid film (Spectrolab, Applied SolarEnergy Corp.).

Ell Module Assembly

" Anti-reflectance treatment of glass(Motorola).

52

Ell Zero depth concentrator effect (JPL, ScienceApplications, J. Mark, General Electric).Mathematical model of the effect of lighttrapping by non-specular reflection in thickfilms.

Ell Grid line optimization (JPL, Texas Instruments,Motorola, RCA, Spectrolab). Mathematicalmodels and computer optimization programs toallow design of minimum power loss collectorgrids.

Ell $2.00/watt (JPL). "Strawman" process analysesfor low cost cell and module fabrication.

Ell $0.50/watt (JPL). "Strawman" process analysesusing projected, automated processes.

Ell Diffusion barriers (Caltech). Development of thescientific basis for formation of diffusionbarriers, especially those of amorphous films.

Ell Solar breeder (Solarex). An analysis of aphotovoltaics production facility powered byphotovoltaics.

Ell Wafer thickness (Motorola, Solarex). An analysisof wafer thickness cost tradeoffs.

Ell Polycrystalline cell processing '(Solarex, SensorTechnology). Development of techniques forprocessing polycrystalline silicon material.

Ell Solar Area Production System Handbook(Theodore Barry Associates). A compilationof costing variables related to large volumeproduction of solar cells.

Ell Tandem Junction cell (Texas Instruments). Ananalysis of a novel photovoltaic device structureto produce electrical power.

Modules and ArraysIn 1975, crystalline-silicon solar cells for use on

spacecraft had already been manufactured for about15 years. There was significant progress madeduring these years in understanding how solar cellsfunctioned and how to improve their performance.However, there was little information about how todesign, fabricate, and use PV modules and arrays forterrestrial applications. There was an immediateneed for module and array knowledge andexperience. A complete technology had to beestablished including environmental and designspecifications, engineering concepts and designknowledge,encapsulation materials and processing,module assembly techniques, quality assurancetechniques, module performance and durabilityevaluation techniques and facilities, and failureanalysis techniques and capabilities. An intensiveeffort to meet this need was initiated with the start ofthe FSA Project in 1975.

The objectives of the module and array activitiesfor both crystalline silicon and thin-film technologiesare to:

• Develop design requirements and specificationscompatible with various operational and safetyenvironments for a variety of applications.

• Develop the engineering sciences technologyrequired to integrate low-eost, efficient celltechnologies into cost effective and safemodules and arrays that meet the operationalrequirements of future large-scale applications.

• Develop the technologies required to achieveand to verify reliable 30-year life flat-plate PVmodules.

53

• Verify adaptability of technology to meetrequired operational conditions.

• Transfer the above technology to industry.

The goals, as established in 1983, are to developflat-plate (non-concentrating) module and arraytechnology as follows:

• Modules with crystalline silicon solar cells.

• High efficiency, 15%.

• Low cost, $90/m2 (1982 dollars).

• Thirty-year reliable operation.

• Modules with thin-film solar cells.

• Efficiency, 12%.

• Low cost, $70/m2 (1982 dollars).

• Thirty-year reliable operation.

In 1975, the goals established were to develop flat­plate module and array technology that industrycould use as early as 1986 to achieve:

• Module prices less than $0.50 per peak wattFOB (1975 dollars), marketing and distributioncosts not included.

• Production rates sufficient to attain economiesof scale.

• Operating lifetimes in excess of 20 years.

• Module efficiency greater than 10%.

Background

Early in the 1970s, small modules were madeusing reject space cells sealed into heavy cast-glassmodules that were durable, but expensive (Figure 40).However, typical mid-1970s terrestrial modules,designed for more economical manufacturing, had alifetime ranging from 6 months to 2 or 3 years (Fig­ure 41). The solar cells for these modules were madeusing circular wafers directly sliced from a Cz ingotand processed by essentially space-cell technology.The electrically interconnected cells were usuallyplaced on a layer of silicone rubber covering a fiber­glass panel, covered with a second layer of pouredsilicon,8' rubber and cured. The' entirel process!was!verylabor intensive and subject to product variations. Themodule designs and processing were often by trial

and error. The products had high failure rates, causedby both immature designs and poor processing.

During the past 10 years, the FSA-industrypartnership has significantly advanced the basictechnology for terrestrial module and array designand fabrication, using crystalline-silicon solar cells.This was accomplished by concurrent experimentaland analytical research efforts supported by iterativefabrication and testing of developmental andcommercial modules, as described in Module andArray Evolution, Engineering Sciences and Design,Encapsulation, Reliability Physics, and ModulePerformance and Failure Analysis. The evaluation ofmodules using thin-film solar cells is just beginning,as is the adoption of existing knowledge and thedevelopment of unique new techniques for thin-filmmodules.

Figure 40. Durable Modules Made from Space SolarCells in Early 1970s.

54

Figure 41. Mid-1970s Terrestrial Photovoltaic Modules

Project Activities: 1975 to 1981Initially, in 1975, contemporary modules were

purchased and evaluated; concurrently,developmental work was initiated on modulespecifications, designs, materials, and processing.These studies evaluated both the technical andeconomical feasibilities of various technicalconcepts. During the next few years, simultaneousassessment of the needs, capabilities, deficiencies,and potential solutions across the entire spectrum ofmodule technologies was accomplished by anexcellent cooperative effort by industry, universities,

55

and Government. This effort resulted in a continuingiterative evolution of new knowledge and capabilities.Practical, realistic environmental protectionrequirements and approaches were developed alongwith techniques for evaluating module capabilities formeeting these specifications. Practical moduledesign knowledge grew as encapsulation materialsand processing methods were developed andevaluated, and the compatibility of the modules withPV power systems was established. As thetechnology evolved, it was incorporated by industryinto durable hardware. This was accomplished by theProject sponsoring cell and module technology

developments, the purchase of advanced modules,the evaluation of these modules, and the analysis ofmodule problems and failures (Figure 42). A flow ofinformation from operational PV systems, as shown inFigure 43, helps the technology development activi­ties shown in Figure 42. Five sequential purchases("block buys" of modules designed and fabricated byindustry to JPL specifications) have resulted in aneffective cooperative effort by the FSA Project andindustry. The large quantities of modules purchased

RELATED- TECHNOLOGIES

RESEARCH

t

(large at the time) for the first three block buys wereintended to increase industry production capacity aswell as provide modules for evaluation and field test­ing. Block IV and V modules were purchased inlimited quantities in the 1980s, for test and evaluationonly. These Project efforts were instrumental indeveloping practical flat-plate module design andmanufacturing technology and in stimulating industryto incorporate much of it into its commercial products.

t

MODULE EVALUATIONPROTOTYPE MODULEQUALIFICATION AND

ACCELERATED TESTING

MODULE ANDARRAY DESIGN~REQUIREMENTS

-

DESIGN SYNTHESISANALYSES,

EXPERIMENTS, -AND COST

EFFECTIVE SOLUTIONS

DURABILITY MODELINGLIFE PREDICTION ~

AND ACCELERATIONTEST CAPABILITIES

FAILURE ANALYSISQUANTIFICATION OF 1..0.-

DEGRADATION r­MECHANISMPARAMETERS

IN-FiElD EVALUATIONPRODUCTION MODULES

AND ARRAYSLONG·TERM

OPERATIONALCONDITIONS

30-YEARLIFE

TECHNOLOGY

The design requirements were periodically updated based upon new knowledge of application needs, environmenta-Iconditions, realism of evaluation and testing techniques and hardware performance.

Figure 42. Module/Array Technology Development

INFORMATION FLOW

FIELD SITES ANDAPPLICATIONS

FSAENGINEERING

INDUSTRYFSA PROJECT

SYSTEMS ENGINEERINGTESTS AND APPLICATIONS

STANDARDS ACTIVITIES

• SYSTEM NEEDS/SENSITIVITIES• ARRAY FIELD EXPERIENCE• TECHNOLOGY CAPABILITY• RAW WEATHER DATA

• ENGINEERING ANALYSIS• ENVIRONMENTAL ASSESSMENT• EXPLORATORY TESTING• DESIGN OPTIMIZATION

• ENVIRONMENTAL STRESS LEVELS• PERFORMANCE CRITERIA• PERFORMANCE DATA• DESIGN GUIDELINES• LOW-COST DESIGNS• ANALYSIS TOOLS• TEST METHODS

Figure 43. Information and Flow

56

Project Activities: 1981 to Present

Since 1981, the evolution of module technologyand the adoption by industry has continued withemphasis now on high-efficiency, cost-effectivemodules that meet safety and reliability requirementsof future large-scale applications such as for utilities.Efforts on rooftop and residential applications arefewer now that it is anticipated that utilityapplications will be the first large-scale use ofphotovoltaics in the United States.

A second parallel effort, just starting, isestablishing the practicality of modules made usingthin-film solar cells. This requires going through manyof the above sequences using as much of theexisting knowledge and capabilities as possible;however, many unique new techniques and newknowledge will have to be developed and evaluated.As an example, new accurate, repeatable electricalperformance measurement techniques for thin-filmcells and modules are being developed and verifiedso that degradation can be accurately quantified andperformance measurements standardized. With theexpanding availability of first-generation thin-filmmodules, performance evaluation and assessment ofreliability characteristics are under way.

Today's Status

Crystalline Silicon Modules

• Estimated life of contemporary productionmodules is more than 10 years.

• Module efficiency is as high as +11% instandard production modules.

• Encapsulation materials have been life-tested to20 years.

• Comprehensive terrestrial module design tech­nology and knowledge exist, which have beengained from 10 years of analytical, design,fabrication, and test activities in activecooperation with industry.

57

• Modules of 0.9 MW sold for just under $5 perpeak watt (1983 SMUD purchase).

• Estimated module price would be about $1.47per peak watt if a multi-tens of megawattsproduction plant were built with the latestexisting technology (see PA&I section of report).

Thin-Film Modules

Prototype modules (other than consumer products)are only beginning to become available forevaluation:

• Reliability is unknown.

• Efficiency is about 3 to 5%.

• Prices are about the same as for crystalline­silicon modules.

Thin-film module differences requiring new orexpanded research:

• New cell environmental durability(temperature/humidity/ultraviolet) failure modes.

• Altered hot-spot heating failure mechanisms.

• Short-circuit cell failure modes and effect onsize and series/parallel redundancy.

• New cell electrical-interconnect failure modes.

• Altered glass breaking strength.

• Flexible substrate technology demands.

• High-cell stresses caused by glass bending.

• Non-linear electrical response and effect onmeasurement.

• Cell-to-cell electrical variability and effect ofelectrical mismatch on circuit design.

odule andArray Evolution

The PV terrestrial modules have evolved during thepast 10 years so that they are now practical, reliable,and cost effective for many applications. There havebeen major improvements in design, fabrication,performance, and reliability. These improvements haveresulted in large reductions in module prices and life­cycle PV power generation costs. An important factorcontributing to this evolution has been the series of fivemodule procurements known as "block buys." Themodule design and development were performed byindustry to FSA specifications; the module performance

evaluations were performed by FSA. These activitiesserved three functions: (1) they encouraged industry toincorporate the latest technology developed in the PVprogram, (2) the various modules were evaluated byuniform repeatable tests that enabled identification ofdesign and/or fabrication deficiencies, and (3) theyenabled a cooperative manufacturer/evaluator quickresponse to module problems that occurred.

A module from each of the five block buys isshown in Figure 44. Table 17 lists the representativecharacteristics of each block of modules. The photo­graph, the list of characteristics, and the five trendcharts (Figures 45 through 49) portray the progress offlat-plate PV modules with crystalline silicon solarcells during the past 10 years.

IV58 Walls'

GENERALmCTRlC

SOLARPOWER

21

SOlAREX

!5 Watts'

Figure 44. Examples of Block I Through V Modules

58

Table 17. Representative Characteristics of Block Modules

I II III IV V

AREA Im21 0.1 0.4 0.3 0.6 1.1WEIGHT Ikgl 2 5 5 9 17SUPERSTRATE OR TOP COVER SILICONE RUBBER SILICONE RUBBER SILICONE RUBBER GLASS GLASSSUBSTRATE OR BOTTOM COVER RIGIO PAN RIGIO PAN RIGIO PAN FLEXIBLE SHEET FLEXIBLE LAMINATEFRAME NO YES YES YES NOCONNECTIONS TERMINALS J·BOX TERMINALS PIGTAILS PLUG·INENCAPSULATION SYSTEM CAST CAST CAST LAMINATED LAMINATEOENCAPSULATION MATERIAL SILICONE RUBBER SILICONE RUBBER SILICONE RUBBER PVB EVACELLS

QUANTITY 21 42 43 75 117SIZE (mm) DlA: 76 DlA: 76 DlA: 76 95 x 95 100 x 100CONFIGURATION ROUNO ROUNO ROUND SHAPED SHAPEDMATERIAL CZ CZ CZ CZ CZJUNCTION NIP NIP NIP NIP p+ NIP

FAULT TOLERANCEPARALLEL CELL STRINGS NONE NONE NONE 3 6INTERCONNECT REDUNDANCY NONE MINOR MINOR MUCH MUCHBY·PASS DIDOES NO NO NO YES YES

PACKING FACTOR 0.54 0.60 0.65 0.7B 0.89NDCT" 43 44 48 4B 48PERFORMANCE AT 2B·C CELL TEMP.b

POWER, MAX. IWI B 24 26 54

I112

MODULE EFFICIENCY 1%1 5.8 6.7 7.4 9.1 10.6ENCAPSULATED CELL EFFICIENCY (%1 10.6 11.2 11.5 11.8 12.3

"NOMINAL OPERATING CELL TEMPERATURE: CELL TEMPERATURE IN OPEN·CIRCUITED MODULE EXPOSED TO 80 mW/cm2 INSOLATION IN AMBIENT OF 20·C, 1 m/! WIND VELOCITY.

bAT 100 mW/cm2, AM 1.5 INSOLATION.

II I I I I I I I II- -

ICELL TEMP: 25·C

I-INSOLATION: 100 mW/cm2, AM1.5 -

II

l- I IVIII T -

I II ESTIMATED

l- I V-I

II I I I I I I I I

40

30

10

75 76 77 78 79 80YEAR

81 82 83 84

12.5IV V

"'-

I I>-'10.0u

i:Ec:; III

~ 7.5 II I~ I I::::> Ic CELL TEMP: 25'Cc:;; 5.0

INSOLATION: 100 mWlcm2, AM1.5

75 82 84

Figure 45. Module Cost Trend

200

VI100 IV;:

Ia:'~

II IIIa:>~

I I~

20::::>cc

CELL TEMP: 25·C:;;10

INSOLATION: 100 mWlcm2, AM1.5

-

-

IJI

"'- 15.0,.::u IV Vi:E III

I [c:;

12.5 II

I~ I

I~ I~

~

10.0u

53I-

:5 CELL TEMP: 25'C::::> 7.5'" INSOLATION: 100 mWlcm2, AM1.5<l-e<ui:E

5.075 84

Figure 48. Cell Efficiency Trend

Figure 46. Module Efficiency Trend

84

I I IIV

I I

75 76 77 78 79 80 81

YEAR

Figure 47. Module Power Trend

1.0 - I I

cr - IIc I II-U

I;:;:

'" 0.5 i-2:;;:ue<<I- -

o'----_'--1_1'----_'--1_'--1_'--1_-'--1_-'--I-'-I-:-:-~I---::-:-'75 76 77 78 79 80 81 82 83 84

YEAR

Figure 49. Packing Factor Trend

59

The following sequence of events typifies theactivities of a recent block buy:

(II FSA prepares design and test specification basedupon prior experience.

(II FSA conducts competitive procurementculminating in award of parallel contracts foralternate design concepts.

(II Contractor performs module design.

(II FSA conducts design review.

(II Contractor fabricates 10 prototype modules.

(II FSA performs module qualification tests (andfailure analysis, as applicable).

(II Contractor modifies design and/or processingprocedures to correct problems revealed byqualification tests.

(II FSA conducts design review.

(II Contractor fabricates 10 modules.

(II FSA performs module qualification test (andfailure analysis, as applicable).

(II Contractor modifies design and/or processing asnecessary and supplies modules for retest.

(II FSA completes final testing.

(II FSA prepares and issues User Handbookdescribing construction details and performanceof successful module design by each contractor.

This series of module procurements has producedthe following results in PV module technology:

(II A systematic upgrading of module quality andperformance by the incorporation of newtechnology into improved designs that are moreamenable to mass production.

Ell Establishment and continuing upgrading of designand performance specifications based upon the

60

needs of the users and the desires of themanufacturers.

(II Development and continuing improvement inmodule evaluation capabilities and equipmentincluding electrical performance, physicalqualification tests, and correlation of both withfield experience.

(II Systematic development of module failureanalysis techniques, equipment andsupplementary analysis, and experimentationleading to the solution of module problems.

(II Adoption by the international PV community ofthe FSA-developed specifications and designfeatures of the Block V module.

(II Variations of three of the Block V modulesbeing purchased for the second phase of the1-MW PV central station power plant being

The success of these activities can be attributed, atleast in part, to the following:

(II The competitive nature of the moduleprocurements which included updated designand test specifications and a work statementindicating design parameters requiringimprovement.

(II Extensive advances in cell, module, andprocessing technology performed primarily byDOE/JPL-sponsored R&D.

(II Participation by FSA personnel having access tothe entire spectrum of PV technology and PVfield operational results including items such asquality assurance.

(II The excellent cooperation established andcontinued at the working level by the PV industryand FSA personnel throughout all these technicalactivities.

Figures 50 through 54 show the five block buysand Tables 18 through 20 show their characteristics.

• ENVIRONMENTAL TESTS LIMITED TO:TEMPERATURE CYCLEHUMIDITY SOAK

• MANY DESIGN IMPROVEMENTSDURING PRODUCTION

• ELECTRICAL PERFORMANCE PERMANUFACTURERS RATINGS

Figure 50. Block I: 1975-1976, Off-the-Shelf Design,54 kW

• FIRST LAMINATED MODULE

• CELL INTERCONNECT AND TERMINAL REDUNDANCY

• QA SPECIFICATION INTRODUCED

• ELECTRICAL PERFORMANCE CRITERIA115.8 VOLTS - 60°C CELL TEMPERATURE)

• STANDARD ARRAY SIZE AND MOUNTING

• INTRODUCTION OF GROUNDING SAFETY PROVISIONS

• EXPANDED ENVIRONMENTAL QUALIFICATION TESTING

• THERMAL CYCLE• HUMIDITY CYCLE• STRUCTURAL LOADING

Figure 51. Block 1/: 1976-1977, Designed to FSASpecification, 127 kW

• DESIGN AND TEST SPECIFICATIONS ESSENTIALLY SAME AS BLOCK II

• IMPROVEMENTS IN DESIGN AND PRODUCTION PROCESSES RESULTINGFROM BLOCK II EXPERIENCE

• MORE UNIFORM QA STANDARDS

Figure 52. Block 11/: 1978-1979, SimilarSpecifications, 259 kW

61

• TYPICAL DESIGN FEATURES• LAMINATED MODULE CONSTRUCTION• FAULT TOLERANT CELL AND CIRCUIT DESIGNS• LARGER POWER OUTPUT• CELLS WITH BACK SURFACE FIELDS• GLASS FRONT FACE

• INNOVATIVE DESIGN FEATURES

• SHAPED CELLS• ION IMPLANTED CELLS• SEMICRYSTALLINE CELLS• ETHYLENE VINYL ACETATE ENCAPSULANT• BATTEN - SEAM ROOFING SUBSTRATE• FRAMELESS MODULE• INTEGRAL BYPASS DIODES

Figure 53. Block IV: 1980-1981: Industry Designs Reviewed by FSA, 26 kW of Prototype Modules

62

• TYPICAL DESIGN FEATURES

• LARGER POWER OUTPUT

• MODULE EFFICIENCY> 10% (EXCEPTRIBBON CELL MODULE)

• GLASS TOP COVER

• ETHYLENE VINYL ACETATEENCAPSULANT

• LAMINATED COMPOSITE FILM BACKCOVER

• LAMINATED MODULE CONSTRUCTION

• FRAMELESS MODULE

• SHAPED CELLS (HIGHERPACKING FACTOR)

• PARALLEL CELL STRINGSFAULT TOLERANT CELL AND CIRCUITDESIGNS

BYPASS DIODES

INNOVATIVE DESIGN FEATURES

• MAJOR INCREASE IN AREA ANDPOWER OUTPUT

• MET MORE STRINGENT QUALIFICATIONTESTS

• VIRTUAL ELIMINATION OF THEFOLLOWING CATASTROPHIC FAILUREMODES

• UNACCEPTABLE CELL CRACKS

• INTERCONNECT FAILURES

• HOT·SPOT FAILURES

• HAIL DAMAGE• MODULE WITH CELLS MADE FROM

SILICON RIBBON (EFG) GROWN TOTHE CORRECT THICKNESS

Figure 54. Block V: 1981-1985: Industry Designs Reviewed by FSA, Small Module Quantitiesfor Evaluation Only

63

Table 18. Module Cell and Circuit Characteristics

CELL CIRCUIT

SIZE BASE SERIES PARAllEl CROSS BY·PASSMANUFACTURER MOOEl NO. QNTY (mm) SHAPE MATl JUNCTION CELLS CELLS TIES OIOOESSENSOR TECH. V·13·AT 25 500lA ROUNO CZ NIP 25 - - -

I SOLAREX 7B5 18 7601A NIP 18 - - -SOLAR POWER E-l0·229·1.5 22 8701A PIN 22 - - -SPECTROLAB 060513·8 20 500lA NIP 20 - - -SENSOR TECH. 20·10·1452·J 44 5601A t 44 - - -

II SOLAREX A·0221·D 42 7601A 42 - - -SOLAR PDWER E·1D008·C 40 102 OIA PIN 40 - - -SPECTROLAB 022962-G 120 500lA NIP 40 3 - -ARCO SOLAR 10699·C 41. 7601A

!41 - - -

MOTOROLA P·0170·770-J 48 7601A 12 4 11 -III SENSOR TECH. 20·10·1646 44 5601A 44 - - -

SOLAREX A·0221·G 42 7601A ROUND W/1 flAT 42 - - -SOLAR POWER E·l0008·F 40 102 OIA ROUND PIN 40 - - -ARCO SOLAR 012110·E 35 103 DIA ROUND W/2 flATS NIP 35 - - 1ASEC 60-3062·F 136 7601A ROUND t 34 4 5 1G.E.8 47J254977Gl·C 19 100 OIA ROUND Wll FLAT 19 - - -

IV'MOTOROLA MSP43D40·G 33 100 x 100 QUASI·SQUARE NIP p+ 33 - - -PHOTOWATT ML·1961·D 72 7601A ROUND

!12 6 - -

SOLAREX 560-BRC 72 95 x 95 SQUARE SEM~XTl 36 2 35 36SOLAREX8 580·BT·R·C 72 95 x 95 SQUARE SEM~XTl 12 6 11 12SPIRE 058·0007-A 108 64 x 64 QUASI-SQUARE CZ 36 3 11 2ARCO SOLAR 004·014168·2 72 97 x 97 QUASI·SQUARE CZ NIP 12 6 3 1G.E.

847E258449G2·A 72 100 x 100 QUASI·SQUARE CZ NIP 36 2 34 3

MSEC8R8·180·12-D 432 95 x 48 RECTANGULAR EFG NIP 36 12 2 1

V SOLAREX C·120·10A 117 101 x 101 RECTANGULAR SEMI·XTl NIP 13 9 1-SPIREs 058·0006·B 72 91 x 91 QUASI-SQUARE CZ N/P·P+ 36 2 2 3

NOTE: 8RESIDENTIAL MODULE

Table 19. Module Performance Characteristics

SAMPLE PERFORMANCE

AT 100 mWlem2• AM 1.5. 28°C CEll TEMP. AT 100 mWiem • AM 1.5. NOCTb

Pma, VPmax IPma, Ise FILL MOOULE CElL IPmax Ise Fill MOOULE CEll NOCTb

MANUFACTURER MOOEL NO. (WI (VI (AI IAI FACTOR EFF. (%1 EFF.el%1 (AI (AI FACTOR EFF. (%1 EFF.c(%1 (OCI

SENSOR TECH. V·13·AT 4.7 9.8 0.48 4.8 9.4 39SOLAREX 785 8.7 7.0 1.24 6.5 10.6 48SOLAR POWER E·l0·229·1.5 13.2 9.6 1.38 5.8 10.2 49SPECTROLAB 060513·8 4.7 9.4 0.50 5.9 12.0 35SENSOR TECH. 20·10·1452·J 11.4 20.7 0.55 24.8 0.60 0.77 6.8 10.6 10.4 18.7 0.56 23.4 0.59 0.75 6.3 9.6 43SOLAREX A·0221·0 20.5 18.0 1.14 24.3 1.43 0.59 6.0 10.7 18.7 16.3 1.15 22.4 1.44 0.58 5.5 9.8 47SOLAR POWER E·l0008·C 33.8 18.0 1.88 23.5 1.98 0.73 7.4 10.7 31.3 16.6 1.89 22.0 1.98 0.72 6.9 9.7 46SPECTROLAB 022962·6 30.0 lB.2 1.65 23.0 1.B6 0.70 6.6 12.7 2B.5 17.3 1.65 21.9 1.88 0.69 6.3 11.7 41ARCO SOLAR 10699·C 22.8 18.2 1.25 23.3 1.38 0.71 8.4 12.2 20.6 16.5 1.25 22.0 1.40 0.67 7.6 11.0 50MOTOROLA P·0170·770·J 26.2 5.9 4.45 7.1 4.82 0.76 7.7 11.8 23.6 5.3 4.45 6.6 4.88 0.73 7.0 10.8 53

III SENSOR TECH. 20·10·1646 11.3 20.2 0.56 24.6 0.62 0.74 6.8 10.5 10.2 18.6 0.55 23.0 0.62 0.72 6.1 9.4 43SOLAREX A·0221·G 21.7 17.8 1.22 23.7 1.40 0.65 6.5 11.6 19.7 16.4 1.20 22.1 1.41 0.63 5.8 10.4 46SOLAR POWER E·l0008·F 34.8 18.3 1.90 23.6 1.97 0.75 7.7 11.2 32.2 17.2 1.87 22.0 1.98 0.74 7.1 10.3 46ARCO SOLAR 012110·E 35.7 16.6 .2.15 21.0 2.42 0.70 9.6 12.6 32.4 15.0 2.16 19.6 2.42 0.68 8.7 11.4 46ASEC 60·3062·F 84.6 16.5 5.11 20.2 5.40 0.78 10.1 13.6 77.4 15.0 5.16 19.2 5.45 0.74 9.3 12.6 47G.E" 47J254977GI·C 18.8 8.5 2.21 11.0 2.53 0.68 9.6 12.6 15.3 7.1 2.16 9.6 2.53 0.63 7.8 10.3 58

IV MOTOROLA MSP43040·G 37.3 16.2 2.30 19.5 2.50 0.76 8.8 11.6 34.3 15.1 2.27 18.4 2.52 0.74 8.1 10.6 49PHOTOWATT ML·1961·0 38.6 5.68 6.79 6.98 7.58 0.73 7.2 11.6 34.9 5.10 6.84 6.5 7.62 0.70 6.6 10.6 47SOLAREX 580·BHC 62.6 16.1 3.90 19.6 4.50 0.71 8.2 9.6 57.3 14.4 3.98 18.1 4.58 0.69 7.5 8.8 49SOLAREXa 580·BT·R·C 60.8 5.31 11.4 6.60 13.2 0.69 8.1 9.3 54.5 4.70 11.6 6.2 13.3 0.66 7.3 8.4 56SPIRE 058·0007·A 57.0 16.2 3.52 20.3 3.64 0.77 11.4 13.6 50.6 14.2 3.58 18.6 3.67 0.74 10.1 11.9 49

AT 100 mWlem2• AM 1.5. '25°C CElL TEMP. AT 100 mWlcm AM 1.5. NOCT

ARCO SOLAR 004·014168·2 84.1 5.82 14.5 7.16 15.9 0.74 11.3 12.6 75.0 5.20 14.4 6.56 16.1 0.71 10.1 11.2 49G.E.· 47E258449G2·A 81.7 17.0 4.61 20.9 5.65 0.69 10.5 11.7 65.4 13.3 4.92 17.7 5.69 0.85 6.4 9.3 65MSECa R.·180·12·0 185. 15.3 12.1 18.9 13.3 0.74 8.4 9.4 165. 13.2 12.5 17.9 13.7 0.67 7.5 8.4 48SOLAREX C·120·10A 139. 5.84 23.8 7.47 26.7 0.70 10.3 11.7 123. 5.18 23.7 6.79 27.2 0.67 9.1 10.3 46SPIREa 058·0008·B 70.7 16.1 4.39 20.7 4.79 0.71 10.1 13.3 62.7 14.5 4.32 18.9 4.84 0.89 9.0 11.8 47d

NOTES: ·RESIOENTIAL MOOULEbNOMINAL OPERATING CEll TEMPERATURE: CEll TEMPERATURE IN OPEN·CIRCUITEO MOOULE EXPOSEO TO 80 mW/lcm2 INSOLATION IN AMBIENT OF 20°C. 1 mI. WINO VELOCITYcENCAPSULATEO CElldRACK·MOUNTEO

64

Table 20. Module Mechanical Characteristics

AR£Ab LENGTH WIDTH MASS SUPERSTRATE SUBSTRATE ENCAPSULANT ElECTRICAL PACKING

MANUFACTURER MODel NO. Iml Imlt (m!t (kg! OR TOP COVER OR BOTTOM COVER ENCAPSULANT METHOD FRAME CONNECTIONS FACTOR

SENSOR TECH. V·13·AT 0.097 0.57 0.17 1.3 RTY-615 ALUMINUM RTV·615 CASTING NONE TERMINALS 0.51

i SOLAREX 7B5 0.133 0.51 0.26 1.1 SYlGARO 184 NEMA·G 10 BOARD SYLGARO 184 PIGTAILS 0.61SOLAR POWER E·l0·229·1.5 0.229 0.61 0.37 2.6 D.C. R4·3117 NEMA·G 10 BOARD SYlGARO 184 J·BOX/CABLE 0.57SPECTROLAB 060513·8 0.090 0.66 0.12 1.6 GLASS ALUMINUM RTY-615 TERMINALS 0.49SENSOR TECH. 20·10·1452·J 0.168 0.582 0.289 1.5 RTV-S15 ALUMINUM RTV·615 TERMINALS 0.64

if SOLAREX A·0221·0 0.335 0.579 0.579 4.1 SYLGARO 184 NEMA·G 10 80ARO SYLGARO 184 ALUM. J.BOX 0.56SOLAR POWER E·l0008·C 0.454 1.168 0.389 7.6 D.C. XL·2577 GfR POLYESTER BOARD SYLGARO 184 NONE J·BOX 0.69SPECTROLAB 022962·G 0.453 1.168 0.388 6.1 GLASS MYlAR PV8 LAMINATION ALUM. PlUG·IN 0.52ARCO SOLAR 10699·C 0.270 1.168 0.231 3.7

Rrt615

TEOLAR PVB LAMINATiON AlUM. TERMINALS 0.69MOTOROLA P·0170·770·J 0.340 0.583 0.583 6.6 STAINLESS STEEL 0.C.03·6527A

CAsrG ST. STEEl t 0.65

Ii SENSOR TECH. 20·10·1646 0.166 0.582 0.286 3.7 ALUMINUM RTV·615 NONE 0.65SOlAREX A·0221·G 0.335 0.579 0.579 4.4 SYLGARO 184 NEMA·G 10 BOARD SYLGARO 184 ALUM. J·BOX 0.56SOLAR POWER E·l0008·F 0.454 1.168 0.389 7.4 D.C. R4·3117 GFR POLYESTER BOARD SYlGARO 184 NONE I 0.69ARCO SOLAR 012110.£ 0.372 1.219 0.305 5.2 GLASS TED/SIlTED PV8 LAMINATION AlUM.

PllTAILS0.76

ASEC SO·308H 0.834 1.199 0.696 13.5 TEOLAR PV8 ALUM.. 0.74G..E." 47J254977G~C 0.196 0.618 0.669 4.0 MEAD PANHOARO G.E. SCS2402 NONE FlAT·CABLE 0.76

IV MOTOROLA MSP43040-o 0.426 1.198 0.356 5.8 TEOIALITEO PV6 ST. STEEL J·60X 0.76PHOTOWAIT ML·1961·0 0.532 1.199 0.444 7.4 TEO/AlITEO PV8 ALUM. PlUG·IN 0.62SolAREX 58D-BH·C 0.782 1.200 0.635 13.9 TEOLAR EVA ALUM. PIGTAILS 0.85SOlAREX" 580·BT·RoC 0.749 1.193 0.628 11.2 TEOLAR NONE PIGTAILS 0.87SPIRE 058·0007·A 0.504 1.200 0.417 7.8 MYLAR·AL·COAT ST. STEEl PlUG·IN 0.85ARCO SOLAR lr04=014168·2 0.745 TIff 0.610 12.0 TEO/PEIITEIi" ALUM. J·BOX 0.90G.E." 47E256449G2·A 0.778 1.228 0.633 13.6 TEOIPET/ALITEOd•• NONE FLAT CABLE 0.90MSEC" R..180·12·0 2.154 1.791 1.203 29.5 PET/ALITEO·

!J·BOX 0.89

V SOLAREX C·120·10A 1.331 1.391 0.957 23.6 PET/MYLAR/TEO· PLUG·IN 0.88SPIRE" 058.flO08·6 0.675 1.134 0.595 7.3 TEOLAR PLUG·IN 0.76

"RESIDENTIAL MODULEbexPOSEO AREACOVERALL DIMENSION

"PLus SHINGLE MATERIAL"PET.POlYESTER FILM. POlYTHYLENE TEREPHTHALATE

65

Engineering Sciencesand Design

At the start of the FSA Project, there was a needto establish a complete new PV module and arraytechnology for terrestrial applications. The use ofsolar cells in space was a very different applicationthan any found on Earth. Thus, there was a need todefine potential terrestrial applications, and todevelop the operational and environmentalrequirements and technology needed for themanufacture and use of PV products in theseapplications.

The objective is to develop reliable, low-costmodule and array engineering and designtechnologies for integrating cells and modules intocost-effective and safe arrays for large-scaleapplications.

• Identify and develop functional requirementsand specifications to ensure modulecompatibilty with the various operational andsafety environments of future PV applications.

tl Develop engineering design methods and dataand advanced safe, reliable design conceptsfocused at the needs of future large-scale PVapplications.

• Verify the adaptability of the technology to meetrequired operational conditions.

Background

The use of crystalline silicon solar cells in spacehad demonstrated that PV power systems were apractical and reliable method of generating electricalpower. The solar cells and the "solar panels," asdesigned, were light-weight, resistant to radiation,capable of withstanding rapid heating and cooling,and were cost effective as used in space. The ideaof using photovoltaics as a major power source inthe terrestrial environment meant that the newfunctional requirements of these applications had tobe identified and the necessary technologiesdeveloped. Not only were the environmentalrequirements significantly different with the additionof wind and moisture-related weathering, but theinstitutional requirements were vastly changed withadded considerations of building and safety codes,user interaction, product liability concerns, andinteraction with the community of installation tradesand operation and maintenance personnel.

Project Activities: 1975 to 1981

Deficiencies and failures were prevalent in the first(pre-1975) terrestrial modules. In addition to the earlyreliability problems, the modules were found to be illsuited for use in the first kilowatt-scale, high-voltage

66

systems. Their size was too small, the mechanicalinterfaces (bolt patterns) varied from module tomodule (even within the same manufacturers model,they were unsafe electrically for use at highvoltages), and they lacked means of electricallyinterconnecting them cost effectively" To define theneeded changes, the FSA Project worked closelywith the systems and applications researchers atNASA Lewis Research Center and Sandia NationalLaboratory to understand the problems beinguncovered and the future applications beingidentified. Next, FSA researchers contracted withleading industrial teams, knowledgable of the detailedtechnical characteristics of future applications, toidentify and develop concepts and requirements forflat-plate PV modules specifically suited to theseuses. Large architect and engineering (A&E) firmswith extensive experience building large central utilitypower plants defined optimum modules and arraysfor utility-scale applications; simultaneously,residential and commercial architectural firmsdeveloped the detailed requirements and conceptsappropriate for these sectors, including considerationof applicable building and safety codes and laborpractices.

As the early deficiencies were gradually resolved,the above activities coalesced into a coherent arraysubsystem design effort (Figure 55) addressed tooptimizing the overall module and array designs forthe intended applications so as to reduce costs andincrease efficiency and lifetime. The environmentalconditions that PV modules "and arrays are subjectedto during field operation were assessed, measured,and defined. This covered mean operating conditionsin addition to extremes for conditions such as UV/solar weathering, temperature, temperature cycling,hail impact, and combinations of the above (Fig-ure 56). Means of quantifying and reducing theoperating temperatures of modules were developed;detailed wind tunnel analyses were conducted todefine expected wind loading levels, the effectivenessof wind barriers, and dynamic flutter magnitudes.Based on the structural loading studies, new modulestructural designs were developed. These designsincluded frameless modules and low-cost ground­mounted array structures (Figure 57). A newmethodology for determining the breaking strengthof glass was also developed.

To improve the reliability of the arrays, complexcircuit analyses were used to define optimum series­parallel electrical circuit redundancy approaches atthe array and module level, and optimum mainte­nance and replacement trade-offs (Figure 58); otherdesign studies explored low-eost electrical inter­connection techniques.

One of the key institutional barriers uncovered inthe electrical circuit studies was the unique electricalsafety attributes of PV which are inconsistent withmany standard safety practices. Unlike conventionalelectrical sources, PV modules cannot be turned offand they cannot generate the overcurrent needed to

1975 TO 1984

YearTechnology

75 76 77 70 79 00 01 02 OJ 04

Thermal Design (Groundl c: "::::J1 c ::3. c==:>llThermal Design (Rooll C

~f= :iii.

PV-Thermal Modules C c::: t:::::lGnd-Mtd Array Structures 1=::::;&.Residential Arrays c:: =Series-Parallel Analysis c f------ P c j:::;a:Bypass Diode Integration

Electrical Connectors c ~ c = faModule Safety DesignArray Safety Systems

Flammability 0\. = :a. tl=*:PC Interface

.., = Major ContributionIA = Significant Contribution

Figure 55. Engineering Sciences Developments

OBJECTIVES

• DEVELDPTESTREOUIREMENTLEVELSAND DURATlDNS,TEST PROCOURESAND APPARATUS FOR DETECTING

~~~~f~~':,.,~ATION MODES CR'

• DETERMINE RELATIVE TOLERANCE OFMODULE DESIGNS TO SPECIFIEO ENVIRONMENTAL STRESSES AND IDENTIFV~~~~~~~t:AMETERS INFLUENCING

PHILOSOPHY

• MAXIMUM TEST REPEATABILITY TOALLOW CORRELATION AND COMPARISON

• MINIMUM CHANGE IN TESTS TO PERMITLONG TERM CORRELATIONS

• MINIMUM TEST COMPLEXITY AND DURATlON TO REDUCE COST

• EASILV IMPLEMENTED flV MANUFACTUR~:~J~~~ COMMERCiAl TESTING ORGANI

EXPERIMENTAL HAIL RESISTANCETEST APPARATUS

HAIL IMPACT RESEARCH

OBJECTIVES

• ~~~~~~~:~J~st!o~tG';:~~:E~~gHg~u~~g~~F::llc~i:~~~CHANISMS

• DEVELOP STATISTICAL DATA DefiNINGPROBABIlITV OF IMPACT BV VARIOUSHAIL SIZES IN GEOGRAPHIC REGIONS OFUNITEOSTATES

• CONDUCT MODULE HAIL RESISTANCECDSTI8ENEfITASSESSMENT

SUSCEPTIBLE PARTS

• CELLS IESPEClALlV EDGES NEAR ELECTRlCALCONTACTSI

• ENCAPSULANT SYSTEM ICORNERS ANOEOGES, POINTS OFSUPERSTRATE

~~~~~RF~6~I~J~E~~~R~~I~SUU~~~1TI HAIL IMPACT DAMAGEON DEVELOPMENTAL MODULE

!3/4-in.ICE8ALLI

HAil IMPACT RESISTANCE

~~

CLEAR SILICONE POTTING FAILURE RANGE

o 10_in ACRVlIC SHEET NO DAMAGE

009_in.ANNEALEOGLASS~.lALUMSUBSTRATEI

012·;n ANNEALED GLASS

0.12-ln. TEMPERED GLASS

01B_in. TEMPERED GLASS

o O.S 1.0 1.5 2.0

IfAILSTONEOIAMETER lin I

Figure 56. Environmental Requirements Example

OBJECTIVEDEFINITION OF ARRAY DESIGNTRADE-OFFS AND GENERATIONOF DESIGN GUIDELINES, RE­QUIREMENTS, AND CONCEPTS

APPROACH• DEVELOP ADVANCED

DESIGN CRITERIA• DEVELOP ANALYSIS TOOLS

• TESTS• EQUATIONS• COMPUTER PROGRAMS

• GENERATE REPRESENTATIVEDESIGNS AND DESIGNTRADE-OFF DATA

BURIED TRUSS STRUCTURE

4' X 4' MODULE

TRUSS STRUCTURE IS BURIED IN1 1/2' WIDE x 31/2' DEEP x 9'LONGTRENCHES

REFILL AND TAMP TRENCHES USINGREMOVED EARTH

PROTOTYPE STRUCTURE

Figure 57. Advanced Array Structures

VIBRATION TEST

blow fuses or circuit breakers in the event of a shortcircuit or short to ground. To develop the neededsafety technology, JPL contracted with UnderwritersLaboratorieswho developed detailed module safetyrequirements and conceptual approaches to theentire electrical safety system for a complete PV

67

power system (Figure 59). On the basis of this work,requirements for UL listing of modules have beendeveloped, and a new Article 690, covering requiredelectrical safety features in high-voltage PV powersystems, has been included in the 1984 revision ofthe National Electrical Code.

OBJECTIVE• DEVELOP RELIABILITY OATA, DESIGN GUIOElINES. AND REQUIREMENTS FOR

flAT ARRAYS AND MODULES

APPROACH• DEVELOP SERIES PARALLEL ElECTRICAL RELIABILITY MODEL• DETERMINE ARRAY SENSITIVITY TO KNOWN COMPONENT FAILURE MOOES• DEVELOP ARRAY/MODULE CIRCUIT CONFIGURATION GUIDELINES TO MINIMIZE

ARRAY DEGRADATION DUE TO COMPONENT FAILURES• GENERATE TEST TECHNIQUES AND ACQUIRE RELIABILITY DATA FOR MOOULE

COMPONENTS. (CEllS SOLDER JOINTS. CONNECTORS, .llcJ• OEFINE COST _ BENEFITS ASSOCIAUO WITH COMPONENT RELIABILITY

IMPIlOVEMENTS• DEFINE MAINTENANCE PRACTICES FOR MINIMUM LIFE

CYCLE COSTS

RELIABILITY/DURABILITY

GENERALIZED REtlABltlT'1'/DURAllILlTY GOAtliFE-CYCLE ECONOMIC PERFORMANCESHALLDE EQUIVALENT TO $O.101W100EFFlCIENCY. NO OEGRADATION FOil 20 VUfO

BASELINE

NORMAU2EO[ [LaUIVALENTPOWEROUTPUT

,;--~",-~-j20 30

SERIES/PARALLEL NOMENCLATUREr->.

MOOU" I3 PARALLEL STfllNGS2 SERIES BlElCKS2 CEllS PER SUBSTRING2 DIODES PER MO.DULE---~-

BRANCH CIRCUIT3 PARALLEl STRINGS6 SERIES BtOCKS2 CELLS PER SUBSTRING _1 DIODE PER SERIES llLOCK

ARRAY POWER DEGRADATION vsCIRCUIT REDUNDANCY

CEll FAILURE RATE ~ 00001 PER YEAR8 PARALLEL X 2400 SERIES X N SERIES BLOCKS

"~"o".,,"s.,oc.."

0.5 .

,o 5 10 15 20

T1ME!YEARSI

L1FE·CYCLE ENERGY COST vs CIRCUIT CONFIGURATION

FRAME/--=­STRUCTURE

GROUNO

CIRCUIT GROUNO

Figure 58. Electrical Circuit Reliability

applications. Detailed studies were conductedcharacterizing the operating temperatures of roof­mounted modules and defining residential arraydesign guidelines including array wiring practices andbypass-diode packaging concepts. Testing of the newenvironmentally durable encapsulants indicatedsubstantial progress in the area of reliability, but astep backward relative to flammability and suitabilityfor module applications with a high fire risk.Research was initiated to characterize the flameresistance of new module designs and to developflame-resistant design approaches and encapsulantmaterials (Figure 60).

Figure 59. Isolation/Grounding Safety Concepts

The most important technique for getting module­manufacturing industry participation in themodulelarray design process and for transferring newknowledge into practice, was to make periodicprocurements of modules. These purchases ofadvanced-design modules to successively moredefinitive and more appropriate requirements, andthe subjection of these modules to qualification lifetests and field tests, resulted in a very dynamic andfruitful cooperative industry/Government effort.

There were a number of design specifications thatevolved over the years, each building on the previousone together with the technology developments andthe application and testing experience. The latestBlock V Module Design and Test Specification wasissued in two versions in 1981; one specification isfor Residential Applications, and the second is forIntermediate Load Applications, i.e., general purposeuses.

Project Activities: 1981 to Present

Since 1981, the evolution of module and arrayrequirements and technology has continued withemphasis on completing the technology base forresidential applications and supporting the designand fielding of the first megawatt-scale utility

68

FLAMMABILITY RESEARCH

Figure 60. Flammability Research on

Rooftop Module

The need for improved integration between thearray and the power conditioner led to extensivemodeling of array output versus time for various sitesin the United States. Characterizations included

statistical data on maximum array voltage, currentand power levels and the performance efficiency ofvarious maximum-power tracking algorithms.

With the gradual completion of the engineeringtechnologies for crystalline-silicon modules,Engineering Sciences research has gradually phaseddown in recent years and turned to the needs of theemerging new thin-film modules. These needs seemto be well served by the technology base resultingfrom the 10 years of research on crystalline-siliconmodules and arrays.

Significant Accomplishments

Specific design data and recommendations thathave evolved include:

It Module design specifications includingenvironmental endurance test requirements

@ Module electrical safety design requirementsand practices (U L 1703)

It Safety system design concepts andrecommendations (National Electrical CodeArticle 690)

@ Detailed assessments of residential andcommercial bUilding codes and the implicationson the use of photovoltaics

It Module product-liability guidelines

@ Array wire selection and safety designguidelines

It Array series/parallel electrical circuit designguidelines including grounding and bypassdiode design guidelines

69

It Module electrical circuit design guildelinesincluding interconnect redundancy and cellcontacting techniques

It Module electrical terminal needs and designs

@ Module electrical insulation system designguidelines and testing techniques

It Module and array thermal design guidelines forcooler operation, resulting in increased poweroutput and longer life

@ Standardized module thermal testing methods

@ Fire resistant module designs and encapsulantmaterials

@ Residential array design approaches

It Low-cost ground-mounted array designapproaches including frameless modules

ED Guidelines for optimum maintenance/replacement of failed modules in the field

ED Data on array cleaning costs and automatedwashing techniques

It Wind pressure loads on flat modules and arrays

ED Structural capabilities of glass as used inmodules, and glass sizing algorithms

ED Design guidelines for optimally interfacingarrays with power conditioners

EncapsulationIn 1975, the encapsulation requirements of PV cells

for terrestrial use had not been comprehensivelyassessed, needs had not been defined, nor potentiallow-cost encapsulation materials investigated.Therefore, the FSA Project set out to assess theneeds in order to identify, and/or develop newmaterials and new material technologies that wereinexpensive and could protect solar cells for years.The assessment led to the establishment ofencapsulation needs, goals, and detail requirementsand a plan on how to meet the goals.

The objectives of the encapsulation activities wereto define, develop, demonstrate, and qualifyencapsulation systems, and materials and processesthat meet the FSA Project module life, cost, andperformance goals; and to develop and validateencapsulation system life prediction: methodologybased on modeling life-limiting failure modes and onconducting and analyzing aging tests.

The goals are to develop encapsulation systemtechnology adequate for:

• Module life of 30 years (increased from 20years).

• Encapsulation materials, including substrateand/or superstrate, costs not exceeding $14/m2

(1980 dollars).

• Initial optical transmission of 90% and a loss ofless than 5% after 20 years of use.

• Capability to withstand an electrical breakdownvoltage of 3000 V dc.

• Structural integrity and durability to withstandhandling and weather.

• Cost-effective processing in an automatedfactory with high yields.

Background

Typical terrestrial PV module fabrication prior to1975 was by hand assembly. The solar cells wereinterconnected by hand soldering. The strings of cellswere encased by hand-poured silicone rubber, whichusually resulted in non-uniform products. Thedurability of a module was unpredictable, often withdelamination and/or cracking of the encapsulant. Themodule materials were expensive and theencapsulation concepts, designs, and processeswere not amenable to mass production. Consistentmodule quality was a problem because of a lack ofgood processing knowledge, capabilities, and quality

70

assurance, for example, requirements for and controlof encapsulant curing. The product quantities wereso small that manufacturers could not afford to do athorough job of studying materials and processes. Acomprehensive long-range encapsulation technologyplan was needed and was formulated by the FSAProject.

Project Activities: 1975 to 1981

In the implementation of the plan to meet theobjectives and goals, five activities were followed: (1)the generation of specifications and functionalrequirements for encapsulation materials; (2) theidentification or development of low-cost materialssatisfying the specifications and functionalrequirements; (3) engineering material usage; (4)recognition of life and/or weathering deficiencies inthe low-cost materials; (5) generation of necessarydesign approaches or material modifications toenhance life or weathering stability; and (6) life­prediction methodologies for encapsulation systems.

Two basic encapsulation systems for terrestrialmodules evolved: substrate and superstrate(Figure 61). This nomenclature refers to the methodof structurally supporting the encapsulated cells andtheir electrical interconnections, i.e., sub or bottomside and super or top sunnyside. Today, the super­strate design is universally used because temperedglass is a practical top cover for use with crystallinesilicon solar cells. Thin-film cells have the potentialfor use in flexible modules; however, a practicalflexible top cover is unavailable. Consequently, thepreponderance of the effort, to date, has centeredon the technology for laminating the active compo­nents in a PV module and their function are shownin Figure 62.

The key component is the pottant, which is atransparent, elastomeric material that encases thecells and their electrical circuitry. In the fabricatedmodule, the pottant provides:

• Maximum optical transmission in the siliconsolar cell operating wavelength range of 0.4 to1.1 f-Lm.

Ii Retention of a required level of electricalinsulation to protect against electricalbreakdown, arcing, etc., with the associateddangers and hazards of electrical fires, andhuman safety.

Ii The mechanical properties to maintain spatialcontainment of the solar cells andinterconnects, and to resist mechanical creep.The level of mechanical properties also mustnot exceed values that would impose unduemechanical stresses on the solar cells.

DES!GN

GLASS (STRUCTURAL)

SPACERiSCRIM

PODANT

",:~:ifj~Ci~SOLAR CELLS

SMeERISCRIM

PODANT

BACK COVER fiLM

SUBTOTAL

COST EST

(~J",~L

$5.00-10.00

0.100.10

0.70-2.00

0.10-0.20

COST EST1_2 ><1.2 m SIZE$1.10-2JlOlm2

Figure 61. Encapsulation System Designs

Also, the pottant must be commercially availableand be readily processible for automated fabrication.Because there is a significant difference between thethermal-expansion coefficients of polymeric materialsand the silicon cells and metallic interconnects,stresses developed from the thousands of dailythermal cycles can result in fractured cells, brokeninterconnects, or cracks and separations in thepottant material. To avoid these problems, the pottantmaterial must not overstress the cell andinterconnects, and must itself be resistant tofracture. From the results of a theoretical analysis,experimental efforts, and studies of availablematerials, the pottant must be a low-moduluselastomeric material.

The top cover is in direct contact with all of theweathering elements: ultraviolet, humidity, water,oxygen, sunlight, etc., and must protect underlyingmaterials from degrading effects. The outer surface

71

must also be resistant to atmospheric soiling, and beabrasion resistant, anti-reflective, and easilycleanable.

The back cover protects the cells, interconnects,pottant, and the substrate, if there is one, fromweather and from electrical shorting. The covershould be weatherable, hard, mechanically durableand tough, and have a white, high-emissiVity surfacefor cool module operation.

The edge of the laminate needs to be protected sothat delamination does not begin and destroy themodule. A gasket is used if the module is to have aframe or it may be used for the installation offrameless modules.

During module fabrication, primers are required toobtain good adhesion of the polymeric pottant toglass, solar cells, and polymeric films that are usedfor back covers.

MODULE ENCAPSULATIONFUNCTIONAL DESIGN ELEMENTS

PRIMERS/ADHESIVES(AS REQUIRED)

- DIELEC:rRIC-----( • ELECTRICAL ISOLATION

• ENCASES SOLAR CELLSAND INTERCONNECTIONS

• HIGH LIGHT TRANSMISSION

MODULE ELEMENTS

TOP SURFACE(AS IS. MODIFIED.OR COATED)

REQUIRED FUNCTIONS/CHARACATERISTICS

• LOW SOILING• EASY CLEANABILITY• ABRASION RESISTANT• ANTIREFLECTIVE

{

• MAXIMUM LIGHT TRANSMISSION• UV SCREEN

---- • STRUCTURAL SUPERSTRATE(IF REQUIRED)

- SPACER, POROUS---f . AIR RELEASEl . MECHANICAL SEPARATION

- SUBSTRATE -----{

_TOP COVER

- SPACER, POROUS

- SOLAR CELLS

-POTTANT-----~

-PODANT

• STRUCTURAL SUPPORT(IF REQUIRED)

- BACK COVER----{ • BACKSIDE PHYSICAL PROTECTIONl . BACKSIDE WEATHERING BARRIER

{

• GOOD BONDINGCOMPLETED MODULE - • GOOD HEAT TRANSFER

• GOOD PROCESSABILITY

Figure 62. Module Encapsulation Functional Design Elements

Many of the low-cost polymeric materials thatsatisfy module engineering and encapsulationprocessing requirements are not intrinsically weatherstable. Therefore, emphasis has been applied tospecific efforts to overcome life andlor weatheringdeficiencies, to enhance weather stability, and todevelop accelerated aging techniques. Specific itemsinclude the development of chemically attachablestabilization additives (ultraviolet screening agents,anti-oxidants, etc.), computer-assisted kineticmodeling of outdoor weathering reactions, and theuse of new and novel outdoor-heating racks andcontrolled-environment reactors as accelerated agingtechniques.

Transparent polymeric pottants are subject tothree weathering actions: ultraviolet reactions,thermal oxidation, and hydrolysis. During theassessment of various pottants, each polymer wasstudied for its ability to withstand the aboveweathering actions and the cost of the product. Itwas determined that all the lowest cost polymerswere subject to all three weathering actions, anumber of expensive polymers were subject to noweathering actions, and a few intermediate costpolymers were resistant to weathering (at

temperatures up to 80°C), except for ultravioletweathering. Of the four intermediate cost polymers,two were casting liquids and two were dry films thatcould be used for fabricating laminated modules.The dry films are practical for PV module usesbecause of the ease and simplicity of handling andprocessing. Initially, ethylene vinyl acetate (EVA) wasthe only low-cost dry film available; therefore,emphasis was placed upon efforts to adopt it for PVuse. The search for a second dry film polymercontinued in case EVA did not fulfill the encapsulationrequirements. Ethylene methyl acrylate (EMA) wasdiscovered 2 or 3 years later, but by then it seemedthat EVA would satisfy the needs. A lack of sufficientfunds inhibited a thorough investigation of EMAalthough, to date, there are no known EMAdeficiencies for use as a PV module pottant. Theapproach used was to develop an EVA that could bethoroughly evaluated while work progressed on anadvanced EVA.

Other encapsulation topics addressed related tolife and durability included: (1) soiling, (2) electricalinsulation, (3) primers and adhesives, (4) encapsulantsystem design analysis, and (5) materialcharacterization (see Figures 63 and 64).

72

ENCAPSULANT SYSTEM DESIGN ANALYSISRESPONSE OF MODULE COMPONENTS TO ENVIRONMENT

O~~~~

WIND: 100 mphTEMPERATURE: _40° + 90°CHUMIDITY: 0° - 100%SOLAR: 100 mW/cm2

WILL PROVIDE CRITERIAFOR MATERIAL SELECTIONAND MODULE DESIGN

-SOLAR

CELLSTRESS

POTTANT THICKNESS REQUIRED FOR SPECIFIC MODULE DESIGN AND ENVIRONMENT

Figure 63. Encapsulant System Design Analysis

MATERIAL CHARACTERIZATIONCOVER SCREEN ABSORPTION OF HARMFUL UV RADIATION:REQUIRED FOR MATERIALS SELECTION AND LIFE ASSESSMENT

450oL.. )•._~._~_~.~-4a. -J__ _..J

300 350 400

WAVELENGTH, A(nm)

20

40

IKORADj

UV SCREENGUT-OFFCURVES

60 (VINYl TINUVIN IN MMAj

80

SOLARABSORPTION

(%)

1001---_

Figure 64. Material Characterization

Project Activities: 1981 to Present

The PV industry has been using a first-generationEVA in commercial modules for more than 5 years.This material has proven to be a surprisingly stablepolymer when correct module design and processinghave been used. The life-limiting characteristics ofthis EVA have been found from experiments,qualification testing, field operation, and acceleratedtesting. With this knowledge, an advanced EVA thatcorrects minor deficiencies of the earlier EVA hasbeen devised. On the basis of the results of 1 year ofevaluating the advanced EVA, it seems to havecorrected the previous materials deficiencies. At thistime, there are no known reasons why this new EVAshould not satisfy the requirements for a 30-yearmodule life. However, testing needs to be done to

verify the life of the pottant and of a completeencapsulation system using this pottant.

In experimental studies of EVA aging, it seemsthat EVA does not undergo thermal oxidation attemperatures up to 50 DC. However, it does undergophoto-oxidation. EVA, which yellows and degradeswithin 1000 h when subjected to normal weatherconditions at 50 DC, did not yellow or degrade in20,000 h when protected by an ultraviolet screeningfilm. Therefore, the life of an EVA pottant in outdoorservice at 50 DC (or less) is directly related to thepermanence of the module's protection fromultraviolet.

The advanced EVA material (formulated bySpringborn Laboratories) uses a curing agent (TBEC)

73

ADVANCED ADDITIVES DEVELOPMENT

MATERIALS IMPROVEMENT USING ADDITIVES

DEVELOPMENT OF CHEMICALLY BOUNDUV SCREENING ADDITIVES

CONCEPT DEVELOPMENTS INADVANCED STABILIZATIONADDITIVES PROVIDED DIRECTIONSFOR A MORE WEATHER STABLE,EASIER TO PROCESS, EVA FORMULATION

FUNCTION

• BASIC RESIN

• PEROXIDE CURING AGENTREPLACING LUPERSOL 101

• FASTER CURE AT LOWERTEMPERATURES

• IMPROVED LONG·TERMSTORAGE STABILITY

• UV SCREENING ­CHEMICALLY ATTACHABLEFOR BETTER WEATHERINGPERMANENCE

• UV ANTI·OXIDANT, A HINDEREDAMINE LIGHT STABILIZER(HALS) -

HIGH MOLECULARWEIGHT ADDITIVE FORWEATHERINGPERMANENCE

IMPROVED ADVANCED EVA FORMULATION

• UV·3346 (AM. CYANAMID)

• UV·2098 (AM. CYANAMID)

COMPONENT

• EVA - DUPONT ELVAX 150

• LUPERSOL TBEC (PENN WALT)

LED TO :>

HCH3H H H/

/~,?/~,~/~ ~~

COaCH, @:(:N ....~OH

CHEMICALLY ATTACHED UV SCREENIN ACRYLIC (MMA) CHAIN FORUSE IN TRANSPARENT COVER FILMOR POTlANT MATERIALS

WITH METHYLMETHACRYLATE (MMA)

COPOLYM...:.ER_IZE__

VINYL TINUVIN(SYNTHESIZED ATUNIV. OF MASS.)

Figure 65. Advanced Encapsulation Materials Research

which cures faster at a lower temperature in additionto having a longer storage or shelf life; it also has achemically attachable, nondepleting ultraviolet screencomponent and has a nondepleting polymeric weatherstabilizer (see Figure 65). The test results to datehave been excellent.

A definition of the intrinsic dielectric strength ofinsulating materials which can be considered as afundamental material property has been proposed.This is being used to quantify the aging characteris­tics of pottants in regard to their electrical insulationproperties.

The components of an encapsulation system needto be bonded to each other permanently. Significantadvances have been made in this chemical bondingtechnology during the past 10 years and especiallyduring the last 2 years when the number of adhesivesrecommended for use in the PV industry has beenreduced from 23 to 3 and the adhesives themselvesare of improved quality (see Figure 66).

Predictive aging models that connect change ofencapsulation material properties and rates ofchanges to module performance and life have beendeveloped. A correlation of accelerated outdoor agingto normal outdoor aging by elevating the temperatureof encapsulants has been determined (see Figure 67).Equipment for accelerated testing has also beendeveloped.

PROBLEM

DELAMINATION OFMODULE INTERFACESDUE TO OUTDOORWEATHERING

SOLUTION

GENERATE ACTUAL CHEMICAL BONDSBETWEEN MATERIAL INTERFACES;SUCH AS, EVA AND GLASS, BY EMPLOYINGSILANE CHEMISTRY, WHICH RESULTS INHIGH BOND STRENGTH, AND LONG·TERMOUTDOOR WEATHER STABILITY

Power loss of modules caused by 3 years of out­door soiling has been reduced significantly by the useof fluorocarbon coatings on the module top surface.On glass surfaces, this amounted to a time-averagedloss reduction of from 2.65 to 1.55%.

o SILANE

CHEMICALLY BONDED • EVA

INTERFACE INTERPHASE, ,I

I0- SI-

O-SI-I O-SI-

BULK

IO-SI- EVA

O-SI-

O-SI-• O-SI-

MOLECULAR·LEVEL MODELOF A SILANE (CHEMICAL)BONDED INTERFACE -GLASS AND EVA

Figure 66. Chemical Bonding Research

The use of flexible top cover materials is receivingincreased interest by thin-film module manufacturers .However, there is no proven top layer material otherthan glass. It is the judgment of Project personnel thatneither Tedlar nor Acrylar materials, as envisioned,will become an acceptable top material. Present ultra­violet additives to Tedlar are depleted too quickly andthe mechanical properties of Acrylar are deficient forhandling, processing, and in-field operation. Yearsago, the study of substrate encapsulation systemswas downgraded so that, with the available funds,superstrate encapsulation could be emphasized. Ifflexible module designs are to become a reality, anew film material must be developed for use as a toplayer, combining properties such as long-term life,low-cost, weatherability, ultraviolet permanence, goodtransparency in the visible spectrum, and goodcharacteristics during film processing, handling, andencapsulation processing.

74

INCREASING THE SUNS INTENSITY(8 SUNS), AT DSET-ARIZONA

THIS INITIAL FSA LEARNINGEXPERIENCE EVOLVED TO A2ND GENERATION OUTDOORAGING CONCEPT

INCREASING TEST TEMPERATURE AT 1 SUN,STRINGBORN LABS, ENFIELD, CONNECTICUT

NATURAL OUTDOOR AGINGOF POLYPROPYLENE

• UNITS ARE CALLEDOUTDOOR PHOTOTHERMALAGING REACTORS (OPTAR)

• UNITS OPERATE AT70,90, AND 105°C

• IN A TRIAL RUN, UNITSSUCCESSFULLY YIELDEDACCELERATING AGINGRATES OF POLYPROPYLENEAT 70, 90, AND 105°C,WHICH EXTRAPOLATEDTO THE ACTUAL AGINGTIME OF POLYPROPYLENEAT ROOM TEMPERATURES

ACTUAL

ACCELERATEDTESTS

"l!r\..

}.•'\.

I"'" ......

"'<!>o 0- I--

(J) 110(Jw° W.a:woa:w~ 90:::lew!CC(J)~a:<Ca: 70wzoD.;: -::Eo !Zw::c wI- (J) (Jo . 40z~

-~o~<CI-:::. 20

200 500 2000

AGING TIME, HOURS

Figure 67. Outdoor Accelerated Aging Testing of Module Designs and Low-Cost Encapsulation Materials

Table 21. Encapsulation Materials-1985

PROJECTED COST INPRODUCTION QUANTITIES

MODULE ELEMENT MATERIAL STATUS NOTES $/m2 (1980 DOLLARS) THICKNESS (mm)*

SUPEASTRATE GLASS - LOW·IRON TEMPERED (SUNADEX) STRONGEST, BUT DEFECT SENSITIVE 5.50· 8.50 3.0·5.0(STRUCTURAL) - LOW.IAQN ANNEALED THICKNESS DETERMINED BY HAIL OR WIND 5.00· 6.00 3.0·5.0

PLASTIC FILMS FLUOROCARBON - (TEDLAA) .. DuPONT UV SCREENING FILM 1.00 - 3.00 0.025· 0.10(UV SCREEN) ACRYLIC - (ACRYLAR) .. 3M UV SCREENING FILM 0.50· 0.70 0.05 ·0.08

PLASTIC FILMS FEP " EXCELLENT WEATHER RESISTANCE 109 1(NON·UV SCREEN) FLUOREX ... EXPERIMENTAL FLUOROCARBON FILM ,170 1

PAlMERS A-11861 .. FOR BONDING EVA TO GLASS - -

SURFACE TREATMENTS ANTI·REFLECTION TREATMENT FOR GLASS ... BEING DEVELOPED -ANTI·SOILING TREATMENTS .. 0 - -

POTTANT SILICONE RUBBER (RTV 615. SYLGARD) HIGH COST CASTING lI0UID 30.00 ·60.00 1.5 ·30POL YVINYL BUTERAL (PVB SAFLEX) THERMOPLASTIC IN SAFETY GLASS LAMINATION 3.60 0.7ETHYLENE VINYL ACETATE (EVA) MELTS AND CURES DURING LAMINATION 1,00 - 2.00 0.5· 1.0ETHYLENE METHYL ACRYLATE (EMA) .. SOFTENS AND CURES DURING LAMINATION 1.00 - 2.00 0.5 . 1.0ALIPHATIC POL YETHER URETHANE ® 1.50· 3.00 0.5 . 1.0

SPACER NON-WOVEN GLASS MAT (CRANEGLAS) FOR AIR RELEASE AND ELECTRICAL ISOLATION SPACING 0.075fm2 0.125

SUBSTRATE GLASS. TEMPERED OR ANNEALED DIELECTRIC AND THERMAL MATCH 5.50 - 8.50 3.0 ·5.0(STRUCTURAL) STEEL .. HIGH STRENGTH 5.40 0.91

ALUMINUM (SHEET AND EXTRUSIONS) COST AND THERMAL EXPANSION DISADVANTAGES 10.00 1.5

BACK COVER POLYESTER FILM (MYLAR) DIELECTRIC FILM 1,00/m2 0.13

FLUOROCARBON, PIGMENTED (TEDLAR) DIELECTRIC FILM 1.00 0.05

ALUMINUM FOIUPOLYMER LAMINATES .. PROVIDES HERMETICITY AND DIELECTRIC 1.00 . 2.00 0.05· 0.13

STEEL FOIUPOLYMER LAMINATES .. BETTER THERMAL EXPANSION MATCH THAN Al 1.00 - 2.00 0.05 - 0.13FIBER GLASS COMPOSITE .. FLAME RESISTANT BACK COVER - -

USED IN CURRENT PV MODULES

•• COMMERCIAL MATERIAL BEING EVALUATED FOR PV MODULES

•• ·MATERIAL UNDER DEVELOPMENT AND EVALUATION FOR PV

* 0001 In = 0,025 mm; 0.040 in = 10 mm

75

loday's Status

It Superstrate encapsulation designs, materials,and material processing are compatible withmass production requirements for modulesusing crystalline silicon solar cells and shouldmeet cost allocations.

It Encapsulation materials have been life-tested to20 years.

It Advanced encapsulation materials are expectedto have a 30-year life expectancy.

It More testing and field operational results areneeded to determine actual module lifecapabilities.

Significant Accomplishments

• Assessed module encapsulation requirementsand potential materials for use under terrestrialoperating conditions.

It Developed a polymeric pottant material (EVA)and processing techniques suitable for massproduction of PV modules.

It Developed back-cover concepts and materials.

It Developed primers and adhesives for glasssuperstrate modules using the above materials.

76

It Superstrate and substrate module designconcepts devised and verified by passingmodule qualification tests.

It Devised photodegradation resistant additivesfor pottants and devised and carried outdegradation tests that verified their value.

It Devised life prediction modeling and tests, andperformed studies of module encapsulationlifetimes.

It Encapsulation materials life-tested to 20 years.

It Materials industry responded very well to JPL­defined material development needs and haveestablished commercial products.

Encapsulation System R&D Needs

It Thin-film module encapsulation systems.

It Flexible top cover films.

It Durable bonding technology for thin-filmmodules.

It Life verification of above and superstratemodule encapsulation systems for crystallinesilicon solar cell modules.

Reliability PhysicsReliability has always been an inherent part of FSA

activities. In fact, efforts in all parts of the Project havebeen directed toward a low-eost, long-life product. Asterrestrial PV technology evolved, the more easy-to­solve failure mechanisms were addressed first; inparallel, excellent progress was made on long-termstudies of encapsulation materials and designs for long­life modules. As each aspect of module technologyevolved, as an entity, the need to understand all thedegrading mechanisms and their interactions within amodule became apparent. These efforts have beenconsolidated and are called Reliability Physics.

The objective is to develop the reliability anddurability knowledge required to produce long-life PVmodules and arrays:

III Develop an understanding of module long-termdegradation and failure mechanisms andtechniques for reducing degradation.

III Develop a reliability prediction capability.

III Develop accelerated test capabilities.

III Verify material, device, and module reliabilities.

Background

In the early 1980s, when the DOE PV Programemphasis was shifted to electric utility applications,module and array lifetime goals were increased to30 years from 20 years. This was considered to be apractical goal based upon technology advancements,typical utility planning for 30-year operational times,and the economic advantages of longer operation ofpower systems.

The longer-life requirement necessitated that a morebasic understanding be determined of all thephenomena involved in module/array degradation. Along-range integrated approach of analyses andexperiments covering all important parameters that canshorten the life and/or reduce the power generatingcapabilities of PV arrays was established. Within FSA,these efforts are now grouped under Reliability Physicsand include the durability of cells, cell metallization, cellinterconnects (within a module), encapsulationmaterials and their use, effects on module life offabrication processing; and the influence of operationaland environmental conditions.

Project Activities

During the course of the project, a steady stream ofmodule failure mechanisms (Figure 68) was observedand identified through module testing, applicationexperiments, and failure analyses. To resolve the

77

reliability problems a systematic research effort wasundertaken with parallel efforts focused on the mosttroublesome failure mechanisms. Initial emphasis,placed on the development of test methods useful forquantifying the reliability weaknesses during themodule design phase, led to the continuing evolutionand development of module qualification tests. Thesetests and their severity were selected to fail oldermodule designs with known deficiencies and to passmodules with good in-the-field performance. As moduledesigns and fabrication techniques improved, emphasiswas gradually shifted toward achieving a physicalunderstanding of the longer-term, less-well-understoodfailure mechanisms and devising qualification tests anddesign solutions for them. Research activities wereinitiated studying interconnect fatigue (Figure 69),sheet glass strength (Figure 70), module soiling (Fig­ure 71), hot-spot heating (Figure 72), breakdown ofelectrical insulation materials (Figure 73) and variousothers. All of these activities have led to comprehen­sive amounts of directly usable data and knowledgeas shown in Figure 74. This information has beenquickly transferred to industry and continues to beincorporated into production module designs andfabrication techniques.

Over the years, the reliability/durability research inFSA has evolved into a general methodology with sixmajor elements:

III Identification of key degradation mechanisms.

• Establishment of mechanism-specific reliabilitygoals.

III Quantification of mechanism parameterdependencies.

III Development of degradation prediction methods.

III Identification of cost-effective solutions.

• Testing and failure analysis of trial solutions.

Although a substantial degree of synergism existsamong these elements, it is useful to address eachseparately.

Identification of the failure mechanism is a criticalstep. During the early years of the Project, modulequalification testing was effective in detecting modulefailure. With the better quality of contemporarymodules, the best evidence of a generic problem iswell documented field failures. This requires carefulmonitoring of field applications with statisticallysignificant numbers of modules, an active problem­failure reporting system, and a comprehensive failureanalysis capability. The voltages and currents in anoperating system also play an important role in hot­spot heating failures, shorts to ground and in-circuitarcs. Failures that show up only after prolonged fieldexposure must be identified by a variety of acceleratedtests, unless waiting for years is acceptable.

UNiFORMLY LOADED, SIMPLY SUPPORTED,

RECTANGULAR GLASS SHEET

MODULE FAILURE TYPES

LIFE PREDICTIONOBJECTIVE: OEVELOP AND VERIFY DEGRADATION RATE

MODELS FOR EACH WEAR-OUT TYPE FAILURE

10001000

CENTER STRESSES ARE CRITICALBELOW BREAK IN CURVES

\

10 ~"lL.J...L.L-'-':':'::--!-.l-LL.LllU_.l-L-'-'-LUJ.l_L..L-'-LJ.wJ10 100 1000

100

LOAD INTENSITY FACTOR

a:o~()

~ 1000>­~u;zw~

~

'"'"wa:~

'"

- RANDOM

I OVERSTRESS

- RANDOMFLAWS

WEAR-OUT IAGINGI

- CORROSION- MATERIAL DEGRADATION- FATIGUE ICYCLINGI- WEAR (DIMENSIONALI- DAMAGE ACCUMULATION

(INFANT MORTALITYI

MANUFACTURINGPROCESS QUALITY

- MATERIALS- CONTAMINATION- DIMENSIONS- ASSEMBLY- HANDLING

FAILURERATE

TIME IN OPERATION

Figure 68. Module Failure-Rate Classification Figure 70. Glass Breakage and Sizing Research

INTERCONNECTSTRAIN-CYCLE FAILURE

TEST DATA

TABLE MY - LOS ANGelES. CATORRANce, CA

~30

z

~

g '"zu

~ 10400 X

PHOTOGRAPHS OF PARTICLES

DEPOSITED ON MODULES

(Scanning Electron Microscope)

400 X

GENEU,UDES'ClNMETHDDOlOC'ES,TR...lIE-a"o...r .... GUlOu'..E5....NOADVANCED D(s'GNS'Q/I KEYMOOUlf COMPONENTS lNVOlVINGMOOUlE.o.SSEMDlYlEVElTIIAOEOff$

APPROACH

... IOENlIfYlMPOIIT.o.NTCOMPONENTDUIC)N$IIEOUIIII'HI.o.SSEMBlYlEVElOI'l'IMIZATlO><

.OfVnOPNECE5SAnTlIUI<)NMETHOllOlOGIUANOTfSTnC'''IIOUU

... CONOUCrANAlVSUTOOEfmEDE$Il)N TnAOE.Qff5. IOfNTIFTDESl(lNGUIOElINE$AND OEVElOPAOVA"CEO COMPON[NT DUKlNS

Figure 69. Electrical Interconnect Fatigue Research Figure 71. Module Soiling Research

78

EXPERIMENTAL APPARATUS OBSERVED MODULERESPONSE vs CELL TEMPERATURE

Figure 72. Hot-Spot Endurance Test Development

HIGH VOLTAGE THIN FILMINSULATION BREAKDOWN

RESEARCH APPARTUS

-MYLAR lEUA ·04801,1B 09211111C 1420111o 30001,1

-- TEDLARE lOOSG 30 TA, 1 milF 100BG30TR,lm,1G 200 SG 40 TA, 201;1

0.99

THIN FILMTEST BREAKDOWN DATA

Technology Vear75 76 77 78 79 80 01 82 83 84

Bond Delamination Cl

Interconnect Fatigue Cl C ~Galvanic Corrosion el

Electrochemical Corrosion e:1=Photothermal Degradation

Structural Failure 1=="Hail Impact != 'fOI

Glass Breakage =11"Cell Fracture Mechanics =y e:13Voltage Breakdown ~Hot-spot Heating =1'.Soiling

.., = Major ContributionA = Significant Contribution

Figure 73. Insulation Breakdown Research

79

Figure 74. Reliability and Durability Developments,1975 to 1984

Thirteen principal failure mechanisms of flatcrystalline-silicon modules, grouped into four types ofdegradation are shown in Table 22. The key failuremechanisms identified to date, and target degradationlevels for each that are consistent with a 30-yearlifetime, are shown together with their economicsignificances and target allocation levels. The units ofdegradation in the third column provide a convenientmeans of quantifying failure levels of individualmechanisms according to their approximate timedependence. Columns 4 and 5 indicate the level ofdegradation for each mechanism that will result in a10% increase in the cost of delivered energy from alarge PV system. Because the mechanisms willgenerally occur concurrently, the total cost impact isthe sum of the 13 cost contributions. Column 6 of thetable carries the analysis a step further and suggests astrawman allocation of allowable degradation amongthe 13 mechanisms to achieve a total reliabilityperformance consistent with expectations of a 30-yearlifetime. The total effect of the allowable levels is a20 % increase in the cost of energy over that from aperfect, failure-free system. The distribution among themechanisms reflects the judgment of FSA personnel atthis time. It is important to update these allocations asfield operation experience becomes available in thenext few years.

A specific degradation or failure mechanism isdependent upon a large number of field-stressparameters, system operating parameters, component­design parameters, and manufacturing-processingparameters. In order to understand the test data, aquantitative understanding of the importance of eachprincipal parameter is required. This is best derivedfrom understanding the chemical and physicalprocesses in the degradation, plus a qualitative insightinto the mechanism physics as can best be determinedby specific tests and analyses.

A specific degradation prediction can be made bydetermining the time history of an applied stress of theappropriate parameter. A variety of theseenvironmental stress characterizations have beendeveloped at JPL including hail impact probability,wind loading pressures, and array voltage and currentdurations. Solar radiation surface meteorologicalobservations (SOLMET) weather data have been usedto model ultraviolet, temperature, and humidity levelsto which modules are exposed. These models havebeen combined with other parametric variables inmaking module useful-life predictions.

Cost effectiveness of modules and arrays requirestrade-offs of degradation rates, failure rates, andlifetimes against initial-manufacturing costs, fieldmaintenance costs, and lost energy revenues. Life­cycle costing analysis integrates these variables andpermits trade-offs to be made. In the identification ofcost-effective solutions, it is important to minimizesensitivities to processing variations and design andmaterial uncertainties. These analyses permit rationaljudgments and cost-effective levels of reliability to beselected.

The first and last step in a reliability program isextensive testing of the complete product. Such testingcuts across the entire reliability development effortfrom beginning to end. It plays a key role in theidentification of failure mechanisms, in the illuminationof parameter dependencies, in the selection of cost­effective solutions, and in the verification of finaldesigns. By the end of the reliability effort, tests withquantitative correlation to field applications should beavailable, based on the parameter dependenciesdetermined and the prediction algorithms developed. Akey element of complete-product testing is inclusion ofthe synergistic reactions between all of thecomponents, including likely interfacing components on

Table 22. Life-Cycle Cost Impacts and Allowable Degradation Levels

level for 10% Allocation

Type ofUnits Energy Cost for Economic

Degradation Failure Mechanism of Increase* 30·Year· PenaltyDegrad. life

k = 0 k = 10 Module

ComponentOpen-circuit cracked cells %/yr 0.08 0.13 0.005 Energy

failures Short-circuit cells %/yr 0.24 0.40 0.050 EnergyInterconnect open circuits %/yr2 0.05 0.25 0.001 Energy

PowerCell gradual power loss %/yr 0.67 1.15 0.20 Energy

degradationModule optical degradation %/yr 0.67 1.15 0.20 EnergyFront surface soiling % 10 10 3 EnergyModule glass breakage %/yr 0.33 1.18 0.1 O&MModule open circuits %/yr 0.33 1.18 0.1 O&M

Module Module hot-spot failures %/yr 0.33 1.18 0.1 O&Mfailures Bypass diode failures %/yr 0.70 2.40 0.05 O&M

Module shorts to ground %/yr2 0.022 0.122 0.01 O&MModule delamination %/yr2 0.022 0.122 0.01 O&M

life-limiting Encapsulant failure due Years27 20 35

End ofwearout to loss of stabilizers of life life

*k = DIscount rate

80

the user's side of the interface. This latter considerationis one of the reasons field-application data are moreconvincing than laboratory test data because theyinclude interaction with the user.

Significant Accomplishments

E!I Developed long-term soiling data for variousmodule surface materials and synthesis ofantisoiling surface treatments.

E!I Developed analysis tools, test methods, anddesign data for the control of solar-eell hot-spotheating.

E!I Developed data on probability of impact ofvarious-sized hailstones, and means of survivinghail impact.

E!I Developed analysis tools and design data forpredicting the breaking load of glass sheets.

E!I Developed design data and test methods on thephotothermal stability of module encapsulants.

81

E!I Developed test methods and performance dataon the reliability of solar-eell metallizationsystems.

E!I Characterized the electrical breakdown strengthof polymeric dielectric films.

E!I Characterized the parameters involved inelectrochemical corrosion of modules.

E!I Developed analysis tools and design data forpredicting interconnect fatigue and designinglong-life, reliable cell interconnections.

E!I Developed design data and qualification testingtechnique to ensure the reliability of bypassdiodes.

E!I Developed design data and testing techniques onthe fracture strength of crystalline-silicon wafersand cells.

E!I Developed a comprehensive set of modulequalification tests, testing facilities, and failureanalysis facilities.

odule Performanceand Failure Analysis

In 1975, at the start of the FSA Project, litte or noquantitative data were available on the electricalefficiency of the early PV modules or on the exactcause of the reliability problems common to modulesin use at that time. Building on JPL's history ofmaking precise measurements and performing failureanalyses as part of its space photovoltaic activities,the Project initiated the measurements, testing, andfailure analysis of terrestrial modules in 1975. Thiswas an important part of the Block I moduleprocurement activities and it has continuedthroughout the duration of the Project.

The objectives of the module performancemeasurement and failure analysis activity are tosupport the module development effort:

.. By providing accurate electrical measurementsof state-of-the-art and advanced developmentmodules, including developing and providinginternational recognized electrical performancemeasurements and facilities.

.. By conducting various characterization andenvironmental qualification tests including thedevelopment of unique testing methods andfacilities.

.. By establishing a problem-failure reportingprocedure, conducting detailed analysis ofmodule failures, and developing advancedfailure analysis techniques and facilities.

Background

At the start of the JPL Project, it was recognizedthat to succeed with the development of advancedmodule technologies there must be accuratemeasurements data to continually determine thestatus of the technology, and to guarantee accuratecommunication among the large number ofresearchers involved. The need to accurately definecosts and cost standards was the responsibility ofthe Project Analysis and Integration portion of theProject; the need to accurately measure efficiencyand reliability became this activity's responsibility.This evaluation function served to determine whetherdesign and fabrication innovations were successful intheir objectives, and to identify deficiencies thatshould be the object of design and fabricationmodifications or the subject of further research.

Project Activities

Electrical Performance Measurement

The electrical performance of a new moduledesign is the necessary criterion for determining

82

whether module innovations have resulted inincreased efficiency. Therefore, progress would havebeen impossible to assess without the developmentof precise and accurate methods of performing thismeasurement. Because the measurement of modulesunder natural sunlight is prohibitively time-consuming,use is made of solar simulators. Early in the Project,a standard irradiance (magnitUde and spectraldistribution) was defined and various solar simulatorswere put into commercial use by the manycompanies and laboratories involved in the program.Because none of the simulators could duplicate thestandard spectrum, methods were developed forcorrecting the simulator measurement to the value ofpower output that would have been obtained if themodule had been exposed to a standard irradiance.

In 1975, electrical performance measurement ofsolar cells and arrays had already undergone con­siderable development in connection with the use ofthese devices on spacecraft over the prior 15 years.Measurement for that application was simpler thanmeasurement for the terrestrial application becausethe solar irradiance spectrum above the Earth'satmosphere is more stable and more easily measuredand described. It was also feasible to devoteextensive resources to the measurement of a singlespacecraft module (or array) because of its high costand need as part of space mission success.

In the terrestrial application, the irradiancespectrum is far more difficult to measure and definebecause it is subject to spectral absorption andscattering in the atmosphere and because it consistsof both a direct and diffuse component. Because thecommercial cost of an individual module must below, the resources allotted to its measurement mustalso be low.

The first method consisted of performing a primary(natural sunlight) calibration of a solar cell that was asample of the cells used in the modules; it could bepresumed to have the same spectral response. Usingthis cell as a reference cell for adjusting themagnitude of irradiance from any simulator, theresultant measurement was then independent of thesimulator spectrum. Subsequently, a new methodknown as mismatch factor correction was developed.With this method, which required knowledge of thesimulator spectrum and of the spectral response ofboth the reference cell and the module, it waspossible to perform a computer correction of themeasured power value even though the referencecell spectral response did not match that of themodule.

The disadvantage of the above two methods isthat either they assume that the spectral response ofthe module is the same as that of the cell, or theyrequire the difficult or impossible operation ofmeasuring the spectral response of the module. Asolution to these problems consisted of designing asimulator/filter combination that produced the

---- ASTM E891-82 DIRECT IRRAD-- FILTERED LAPSS DIRECT IRRAD

NORMALIZED TO lOOOW/m 2

600 700 800 900 1000 1100 1200 1300WAVELENGTH (nanometers)

Figure 76. Spectrallrradiance (AM 1.5 Direct LAPSSVersus ASTM AM 1.5 Direct)

180

160

EI40N

TE

~120

g100 ,'E I,w 80 Iu ,z~

,0 60 I« I

'" I,

~ 40 II

II

20 I,I

o '300 400 500

The simulator used in this system is known as theLarge Area Pulsed Solar Simulator; it produces thespectrum shown in Figure 75 when it is not filtered.The filtered spectra for the direct normal case andthe global case are shown in Figures 76 and 77,respectively, along with the desired standard spectra.These spectral matches are close enough in bothcases so that the error due to the mismatch is notgreater than 1%.

standard spectrum. Such a system has now beenimplemented for both of the current irradiancesdefined in the U.S. standards documents. These arethe airmass 1.5 direct normal irradiance (ASTM E891-82) and the airmass 1.5 global spectrum(combining direct and diffuse components, ASTM E892-82). With this system, the reference cell needonly be of the same generic material as the module'scells, and very large modules can be measuredbecause their spectral response need not be known.

r---...--.I.. ........

\ ,-\ ; III"VIIIV

I

---- ASTM E892-82 GLOBAL IRRAD- FILTERED LAPSS GLOBAL IRRAD

NORMALIZED TO I000W/m2

500 600 700 800 900 1000 1100 1200 1300

WAVELENGTH (nanometersl

I

60 "/

40 :II

20 I

180

300 400o I

160

J 140E

~ 120jb 100'Et5 80z~o«'"'"---

wUz 80~0«'" 60~w>

40;::::;j'" 20

0300 400 500 600 700 BOO 900 1000 1100 1200 1300

WAVELENGTH (na nometersl

Figure 75. Spectrallrradiance (JPL UnfilteredLAPSS)

Figure 77. Spectrallrradiance (AM 1.5 Global LAPSSVersus ASTM AM 1.5 Global)

83

Qualification Testing

In addition to determining whether a new moduledesign provides a desired increase in efficiency, theelectrical performance measurement of a modulealso serves as the principal criterion for determiningthe success of a new module in terms of its ability towithstand the environmental and electrical stressesexpected in the field. An accelerated program ofthese stresses has been applied to all new modulesdeveloped under the FSA Project as well as to othercommercial modules available from U.S. and foreignsources. The test program consists of a set offormally defined qualification tests whose descriptionand sequence are illustrated and described in detail in

Figure 78. As the figure also shows (under theheading "Qualification Test Specifications"), thespecifications evolved during a series of moduledevelopment programs over a 10-year period. Theevolution has resulted from the information providedby the tests themselves, from module field testexperience, and from module research.

Qualification tests have been performed onmodules of about 150 different designs. With fewexceptions, the tests have revealed defectsnecessitating module design or manufacturingprocess changes. This is proof of the critical role ofthis type of testing in identifying modulemodifications needed, and in identifying requiredresearch in materials, processes, andlor design.

ELECTRICAL PERFORMANCE MEASURED AND PHYSICAL DURABILITY ASSESSED

INITIAL EVALUATION

- elECTRICAL PERfORMANCE

- VISUAL AND MeCHANICALINSPECTION ANDCHARACTERIZATION

- INSULATION TEST

- NOMINAL OPERATING CELLTEMPERATURE INOCTIDETERMINATION

- VOLTAGE AND CURRENTTEMPERATURE COEFFICIENTDETERMINATION

L... THERMAL CYCLEf'll"" TEST

-+- EV f+. HUMIOI'iEsrEEZING ~ EV 1+ M'CHAN'CAC I ... -,. I... ~ I ...

LOADING i"'" v 1-pIi' EV HAil ~

"SO _ a....__--I a...._'_M._m_...I

.....

FINALEVALUATION

ELeCTRICALPERFORMANCE

VISUAllNSPECTtON

INSULATION TEST

EVALUATION OFCHANGES

HAIL TEST APPARATUS

HUMIDITY-FREEZING CYCLE

ElECTnlCAl PERFORMANCEVISUAL INSPECTION

W ........".... '.." ..'MUM~ESISTA..C[TOG~OU..Ofonu~OSEOCO..OIlCmnS

i~'!~Z~\',~::'~~'t.:;,::~~:,

<O"<;YO'!IO'C,CYCUSAS,"O'CATlO

'0 ~CVCUSAY""~H,n-c~o..,.cOil\DCYClESMlIS.. m'.,oIO'CTO_lIS'C

~~~~~i~!>O 'O-'f,2l CYCLES AS

,~~:~:,'::~~~,~ lA8 ~ur ..o "7

~t'.:'G~~""Ell"'TEII2<""'m."'OF

2O ........ 0'A... ETEnHAIL·21m1<

:IOOll !>O_AMAX CURIIE"TATVOLTAGE

:,ICONTINUED FOR5(I~1~~ __

:'\ ,/j '- ...J

I. , ,TIME(hr)

QUALIFICATION TEST SPECIFICATIONS

TYPICAL THERMAL CYCLING TEST

I------ MAXIMUM CYCLE TIME - ..I

l/lO"CIh,MAXIMUM

MICHA,,'CAllOAO'''GCYClE

• liES _ ~ESJOE..T!AL,,- '''1E~MEOIATE lOAO

LARGE-AREASOLAR SIMULATOR

TESTING

CYCLIC PRESSURE LOADING TEST APPARATUS

Figure 78. Module Qualification

84

Qualification tests are performed on a smallsample of modules of any given design. The proof ofsatisfactory performance, including reliability,requires field testing the modules in the variousenvironments in which PV power stations would beinstalled. Such a program was instituted early duringthe Project history using the sites listed in Figure 79.The performance observed in the tests at these sites,and at other array installations that were part of theDOE program, resulted in important modifications tothe earlier qualification test specifications. Theseprovided for better accelerated testing to reveal thepotentiality of failures because of such phenomenaas hail impact, thermal stress, and cell back-biasing.The early history of failure statistics is shown inFigure 80. Because of reductions in support of fieldtesting, statistics are not available for the Block IV orBlock V modules.

Failure Analysis

When problems occur during qualification tests, orfield tests, or at array installations, it is necessary to

MODULE FIELDTESTING-16 SITES

PASADENA. CA IJPLIPRIME SITE FOR FIELDTESTING

HOUGHTON. MITYPICAL REMOTE SITE

perform an in-depth failure analysis to find the exactcause of the problem. A Problem Failure Reportingsystem was established by the FSA Project in 1975to provide formal reporting of all failures, regardlessof the site of occurrence. The reports and the failedmodules were delivered to failure analysis personnelwho then applied a variety of sophisticated tech­niques (some derived from the NASA space explora­tion program and some developed explicitly for theterrestrial PV program) to isolate the specific cause ofthe failure (Figure 81). A highly detailed report wasprepared describing each such analysis andpresenting the results, with recommendations forcorrecting the module deficiencies. ThesePerformance and Failure Analysis Reports weresupplied to the manufacturer of the module and toJPL personnel responsible for module developmentresearch. During the various module developmentprograms, such analyses were instrumental incorrecting design and processing problems to theextent that new modules incorporating therecommended changes were then successful inpassing the qualification tests.

FIELD TEST SITES

EXTREME WEATHER

FORT GREELY. ALASKA (ARCTIC)FORT CLAYTON. CANAL ZONE (TROPIC)

MARINE

KEY WEST. FLORIDASAN NICOLAS ISLAND. CALIFORNIAPOINT VICENTE. CALIFORNIA

HIGH DESERT

ALBUQUERQUE. NEW MEXICODUGWAY. UTAH

GOLDSTONE. CALIFORNIA

MOUNTAIN

MINES PEAK. COLORADOTABLE MOUNTAIN. CALIFORNIA

URBAN COASTAL

NEW ORLEANS. LOUISIANANEW LONDON. CONNECTICUT

MIDWEST

CRANE. INDIANA

UPPER GREAT LAKES

HOUGHTON. MICHIGAN

NORTHWEST

SEATILE. WASHINGTON

URBAN SOUTHWEST

PASADENA. CALIFORNIA

Figure 79. Module Field Testing (16 Sites)

85

NEEDS FOR LONG-LIFE, DURABLE MODULES/ARRAYS

• MODULE AND ARRAY FIELD TESTING

• REQUIRED BECAUSE MODULE QUALIFICATION TESTING HAS NOT BEEN ENTIRELYSUCCESSFUL IN PREDICTING AND/OR DUPLICATING MODULE FIELD FAILURES

• KNOWLEDGE OF MODULE WEAR-OUT MECHANISMS WHICH, OTHER THAN CELLINTERCONNECT FAILURES, HAVE NOT BEEN DETECTED AS YET; POTENTIALLONG-TERM OPERATION DEFICIENCIES/FAILURES ARE EVIDENT (CRACKED CELLSAND COVER GLASS)

• REDUCTION OF SINGLE MODULE ENDURANCE FIELD TESTING IN FAVOR OFMODULE TESTING IN ARRAYS

FAILURE" RATES FOR MODULES AT FSA FIELD TEST SITES

326 MODULES DEPLOYED12 FAILURES

• FIELD TESTING IN REAL-USE POWERSYSTEM CONFIGURATIONS

• DETECTION OF DEFICIENCIES/FAILURESWHICH OFTEN HAVE BEEN A SURPRISE

• PERFORMANCE VERIFICATION, UNDEROPERATIONAL CONDITIONS, OF NEWADVANCED MODULES, WITH INNOVATIVEDESIGN FEATURES

• KNOWLEDGE OF MODULE/ARRAYPERFORMANCE AND SAFETY PROBLEMSWITH THE REMAINDER OF THEPOWER SYSTEM AND WITHUTILITY INTERACTIONS

• CENTRAL STATION: ESTABLISHMENTOF CREDIBILITY OF MODULE/ARRAYSAS POWER GENERATORS FORCENTRAL STATION USE; i.e.,COST EFFECTIVE WITHLONG LIFE OPERATION

484236

163 MODULES DEPLOYEDo FAILURES

3024

BLOCK I

261 MODULES DEPLOYED48 FAILURES

021

UJ..J

< 18u..InUJ 15..J:::>00 12:Eu..0 9UJC)

< 6I-ZUJ(.)

3a::UJa..

TIME IN FIELD, MONTHS·FAILURE - OUTPOWER DECREASE

GREATER THAN 25%

Figure 80. Needs for Long-Life, Durable Modules/Arrays

MODULE PROBLEM/FAILURE ANALYSISX-RAY OF OVERHEATEDCELLS SHOWING MELTED SOLDER

LASER SCAN OF MODULEOUTPUT SHOWING CRACKED CELL

PHOTOMICROGRAPH OF HAIL DAMAGE

HIGH VOLTAGEBREAKDOWNOF INSULATOR

• PROBLEMS/FAILURES AT TEST/APPLICATIONSITES REPORTED

• JPL AND MANUFACTURER EVALUATEP/F AND DETERMINE CORRECTIVE ACTIONS

• MANUFACTURER CHANGES MODULEDESIGN OR WORKMANSHIP AS NEEDED

CELL INTERCONNECT FAILURE

Figure 81. Module Problem/Failure Analysis

86

Significant Accomplishments

• Environmentally tested more than 150 differentmodule designs to date over the 10-year period.

• Processed 1200 reports of failures. involving460 major failure analyses.

• Developed unique environmental testingfacilities including hail guns, mechanical cyclic­loading apparatus, hot-spot test equipment. andnominal operating cell temperature (NOCT) testracks and instrumentation.

• Established outdoor field test-rack facilitiesthroughout the United States in variousgeographical climates.

87

• Developed module inspection techniques andguidelines.

• Developed a unique Large Area Pulsed SolarSimulator with closely matched airmass 1.5direct and airmass 1.5 global spectra.

• Calibrated numerous reference cells that haveserved as the primary industry standards overthe past 10 years.

• Developed a simple. accurate method ofsecondary calibration of reference cells; havecalibrated cells for many U.S. and foreign PVmanufacturers.

• Developed a unique laser scanner instrumentfor the failure analysis of large-area moduLes.

Appendix A: GlossaryA

A&E

AM

AR

a-Si

ASME

ASTM

BSF

BSR

C

CAST

COSMIC

CVD

CVT

Cz

DCS

DOE

EFG

EMA

EPSDU

ERDA

EVA

FAST

FBR

FSA

FZ

h

HEM

Isc

I-V

ID

Ampere(s); Angstrom(s)

Architects and engineering

Air mass (e.g., AM1 = unit air mass)

Antireflective

Amorphous silicon

American Society of Mechanical Engineers

American Society for Testing and Materials

Back-surface field

Back-surface reflection

Celsius (temperature scale)

Capillary Action Shaping Technique

Computer Software Management Information Center

Chemical vapor deposition

Chemical vapor transport

Czochralski (classical silicon crystal growth method)

Dichlorosilane

U.S. Department of Energy

Edge-defined film-fed growth (silicon-ribbon growth method)

Ethylene methyl acrylate

Experimental process system development unit

Energy Research and Development Administration

Ethylene vinyl acetate

Fixed Abrasive Slicing Technique

Fluidized-bed reactor

Flat-Plate Solar Array Project

Float-zone (silicon crystal growth method)

Heat transfer coefficient; hour(s)

Heat-exchange method (silicon-crystal ingot-growth method)

Short-circuit current

Current-voltage

Inside diameter

88

IPEG

IR

J sc

JPL

Kg

LAPSS

LASS

MEPSDU

mg-Si

MINP

MOD

MSEC

MT

NASA

NEC

NOCT

O&M

Pmax

PA&I

PDU

PIM

ppba

ppm

PU

PV

PVB

R&D

RH

s

SAMICS

SAMIS

SCAP1D

Improved Price Estimation Guidelines

Infrared

Short-circuit current density

Jet Propulsion Laboratory

Kilogram

Large-area pulsed solar simulator

Low-angle silicon sheet growth method

Module experimental process system development unit

Metallurgical-grade silicon

Metal insulator, n-p

Metallo-Organic deposition

Mobil Solar Energy Corp.

Metric ton(s)

National Aeronautics and Space Administration

National Electrical Code

Nominal operating cell temperature

Operation and maintenance

Maximum power

Project Analysis and Integration Area (of FSA)

Process development unit

Project Integration Meeting

Parts per billion, atomic

Parts per million

Polyurethane

Photovoltaic(s)

Polyvinyl butyral

Research and development

Relative humidity

Second(s)

Solar Array Manufacturing Industry Costing Standards

Standard Assembly-Line Manufacturing Industry Simulation

Solar-Cell Analysis Program in One Dimension

89

SCCD

SEEMA

SERI

Si

SIMRAND

SMUD

SOA

SOC

SOLMET

STC

T

TCM

TCS

UL

Voc

Wp

Short-circuit current-decay

Solar-Cell Efficiency Estimation Methodology and Analysis

Solar Energy Research Institute

Silicon

SIMulation of Research And Development Projects

Sacramento (California) Municipal Utility District

State of the art

Silicon-on-ceramic

Solar radiation surface meteorological observations

Silicon tetrachloride

Temperature

Transparent conducting material

Trichlorosilane

Underwriters Laboratories, Inc.

Open-circuit voltage

Peak watt(s)

90

Appendix B: FSA Project Contractors

co

COMPANY COMTRACT TITLE OP~ AREA

AeroChel'l Research Laboratories, Inc. Si Halide-Al"'.ali. I1etal FlaPles as a Source of Solar Grade Si SIMATAeroChel'l Research Laboratories, Inc. Developl'lent Processes for the Production of Solar Grade Si frm Si Halides [; Alkali Metals SINATAeroChePl Research Laboratories, Inc. Developl'lent of tlodel and COl'lPuter Code for Si Reactions SINATAerochel'l Research Laboratories, Inc. Synttlesis of Silane and Si in a Non-E~ilibrim Plasna .Jet SIMATAlA Research Corp. Integrated Residential PV Array Developl'l€nt .. RESApplied Solar Energy Corp. Developl'lent of Hi!1J-Efficiency Solar cells .. DRApplied Solar Energy COrp. Evaluation of Io~n-I'~lanted Solar Cells POApplied Solar Energy Corp. Developl'lent of Hi!1J-Efficiency (14%) Solar cell Array Modules POApplied Solar Energy Corp. Assessl'lent of Present State-of-the-Art SaloJing Tech. of Large Dial'leter Ingots for Solar Sheet l1aterial POApplied Solar Energy Corp. High-Efficiency, Lor~-Life Terrestrial Solar ParEl POApplied Solar Energy COrp. Design, fi.tlrication, Test and Price ANllysis of Third Generation Desi!)'l Solar Cell l'tldUle MPfAApplied Solar Energy COrp. Developl'lent of Low-Cost CCfltacts to 5i Solar cells POApplied Solar Energy COrp. Laboratory services .. POApplied Solar Energy Corp. Microcrystalline Si Growth for Heterojunction for Solar Cells ORApplied Solar Energy COrp. Interl'ledlate Load Nodules for Test and Evaluation RESApplied Solar Energy Corp. Solar Cell Process DeveloJ)'lent .. POARGO Solar, Inc. Solar Cell Panel Develo~ent Effort POARGO Solar, Inc. VacUUPl Die cast of Si Sheet for PV Applications SINATARGO Solar, Inc. Design, fi.tlricatiCfl, Test. lNalification and Price Pnalysis of Third Generation Design Solar cell Modules MPFAARGO Solar. Inc. Autmated Solar Panel Assel'1bly Line POARGO Solar, Inc. Design of Block V Solar Cell Modules - 19B1 MPFAARGO Solar, Inc. Block V Docmentation and Solar cell Modules .. RESARGO Solar. Inc. Adapt Pulsed Excil'ler Laser Processing Tecmiques to fabricate Cost-Effective Solar cells .. POAriZONl State University Conduct an Evaluation and Calibration of a Czochralski (Cz) Crystal Growth Systel'l .. SINATAstrosystel'lS, Inc. Si filA Solar Cell Process .. SHEETBattelle I1el'lOrial Institute Part 1. - Evaluation of Si Production SUiti.tlle for Solar Cells SINATBattelle I1el'lOrial Institute Nodule Encapsulation Task of Low-Cost Si Solar Array Project SINATBattelle I1el'lOrial Institute Study ProgIaPl to Develop 1; Evaluate Die am Container Materials SHEETBattelle I1el'lOrial Institute Study ProgIaPl to Develop 1; Evaluate Die am Container Materials SHEETBechtel National, Inc. Terrestrial Central Power Utility Array Life-Cycle Pnalysis PA6IBechtel NationaL Inc. Developl'lent of ReQirel'lents for Terrestrial Solar Arrays RESBernd Ross Associates Developl'l€nt of an All Metal Thick fill'l Cost Effectille Metallization S,lstel'l for Solar Cells POBernd Ross Associates Econol'lical Il'Iproved Thick fill'l Solar Cell Contact POBurt Hill Kosar RittelAann Associates Operation and Maintenance Cost Data for Residential Photovoltaic Nodules/Panels RESBurt Hill Kosar RittelAann Associates COAl'lerciaVIndustrial PV l'tldules Requirerlents Study RESBurt Hill Kosar RittelAann Associates Residential PV Modules ReQjireAents Study RESC. T. Sah Associates A St.udy of Effect of Il'lpuri ties in Si Material SINATC. T. sah Associates A Study of Relationships of Material Properties and Hi-Eff Solar cell Perfomance on Material COl'IPOsition .. ORCalifornia Institute of Tecmology InvestioatiCfl of free Space Reactor Si Production .. SINATcalifornia Institute of Tecmology Diffusion Barrier Studies POCarnegie Mellon UrQllersity Exploratory Study of Product Safety and Product Liability Considerations for Photovoltaic Module/Array Devices RESCase IJestern Reserve i.X1iversi ty Encapsulation Syst€l'l Studies for Low Cost Si Solar Array DRCttronar Corp. Al'lorphous Si Module Research .. RESClel'lson University Investigation of Reliability Attributes am Accelerated Stress factors on Terrestrial Solaz Cells .. RESCoors Porcelain Co. Study PrograPl to Develop and Evaluate Substrate and CCfltaiJler Materials SIMATCornell University Investigation of Physical Structure and the Chenical Nature of Defects in Si $t,eet Material .. POCornell University Cheracterization of StructuraL Electrical E; Chel'lical Properties of 5i Sheet t1aterial SINATCrystal SystePlS. Inc. Nul ti-Wire Wafering Technology DeveloPRent by a flx€':l Abrasive Slicing Tedmique (f~lST) SINAT

enI\)

CDI1PANY CONTRACT TITLE 0lPEJI AREA

Crystal Systens, Inc. Heat-Excha1ger- Ingot casting/Slicing Process, Si Sheet Growth SHEET000 Corning Corp. Task I - Si Naterial Part 3 SINAT000 Corning Corp. Develop Si Encapsulation $yS'tens for Terrestrial Silicon Solar Arrays POEagle-ficher Ind.. Inc. Study Progrart to Develop and Evaluate Die aa1d Container Materials SHEETElectriN<., Inc. Laboratory services for the Devel. & Charac. of Screenable Mats. Utilized in the Fabrication of Solar Cells .. POEndurex Corp. Solar Cell Nodule lCil-flating Process and Testing POEnergy I1aterials Corp. Electrochenical Production of Si .. SINATEnergy Materials Corp. Low-Angle Si Sheet Growth SHEETEnergy Materials Corp. Gaseous Nel t Replenishlfflt Systen SINATEnergy Materials Corp. Technology Develo~t for LOIol-Angle Si Sheet (LASS) Growth Technique for PI) Applications .. SHEETGeneral Electric Co. Developnent & Testing of Shingle-Type Solar Cell Modules POGeneral Electric Co. Integrated Residential Phatovoltaic Array Developnent RESGffleral Electric Co. Design, fabricatiCil, Test, Qualification and Price Analysis of Third Generation Design Solar ceu Nodules NPfAGeneral Electric Co. Design of Block V Solar Cell Nodule - 1981 NPFAGeneral Electric Co. Floating SUbstrate Sheet Growth Process LASS Task for LSSA Project SHEETGeneral Electric Co. Bypass Diode Encapsulation Study RESGnostic Concepts. Inc. Industrialization Stuctj PA&IHenlock Seniconductor Corp. Developnent of a Polysilicon Process Based on Chem.cal lJa~r Depositicn .. SINATHoneywell. Inc. Dip coating Process. Si Sheet GrOlYth DeveloPf'lent for the Large Area SUiCCfl Sheet Task SHEETIBN Corp. Si Rit:tJCil Grolorth Using a capillary Shaping Technique SHEETlIT Research Institute TechniCal St.pport in Reliability Engineering of Photovoltaic Modules RESIllinois Tool loklrl<s. Inc. DeRonstration of Gapabili t)I t.o Metallize Solar cells by ICfl-flating POJ. C. SCtIJl'Iacher CO. High velocity ContiNJoos-Flow Reactor for Production of Solar Grade Si SINATKa'jex Corp. Low-Cost CZochralski Crystal Growing Techmlogy SINATKa'jex Corp. Developnent of Advanced Nethods for CCiltiNJOOS C-zochIalski Grwth SINATKa'jex Corp. AdIIa1ced CZ Grolorth Process to Produce LC4I-Cost 150 Kg Si Ingots frOl'l a Single Crucible for Techmlogy Readiness SINAlKinetic Coatings. Inc. Phase 2 of the Array AutoPlated Asseroly Task POKulicl<e & Soffa Industries Autonated Solar NOOule AssePlbly Line POLanar Universit.y Process feasibility Stuliy in SUpport of Si Material SINAlLocl<heed t1issiles & Space Co. Spraylon Fluorocarbon Encapsulation for Si Solar cell Arrays RESLocl<heed t1issiles & Space Co. Transparent Solar Cell Mowle Developnent Effort POLocl<heed tti.ssiles & Space Co. Phase 2 of the Array Autol'lated Asseroly Task POLocl<heed t1issiles & Space Co. Evaluation of Laser Amealing of Solar cells PO117 International. Inc. Five (5) Kilowatt Solar Cell Nodules NPfAN7 International. Inc. Three (3) KilC41atts of Solar cell Nodules NPfSNas~chus~tsI~tituteofkmroM~ Hi Speed Growth of Si Sheets in Inclined Meniscus Configurations .. SHEETNas~chus~ts Institute of Temrolo~ Investigation of the I+jdrcgenatiCil of SiCl4 SINATNaterials Research. Inc. Analysis of Defect Structure in Si .. SINATNaterials Research. Inc. Quantitative Analysis of Defects in Si SINATNaterials Research. Inc. Analysis of Defect Structure in Si ORtti.tre Corp. tti.tre Solar Energy Test Systel'l Evaluation POMOOil Solar Energy Corp. Solar Modules POMOOU Solar Energy Corp. Design of Block V Solar Cell Modules - 19B1 NPfAMOOil Solar Energy Corp. Edge-Defined filn fed Growth for Si Growth Devel~fflt SHEETMOOU Solar Energy Corp. Grys:ta1 Growth by Edg~Defined filn-fed Growth TechniQJe (EfG) SHEETMOOil Solar Energy Corp. Block V Doctnentation and Solar cell Hodules RESNCil~nto Research Corp. Si Material Task - Part 2 SINATNotorola. Inc. PV Module Electric.al Tel1'lination 0esif11 Requirenents Study RES

<00J

COMPANY CONTRACT TITLE OPEII AREA

Motorola, loc. Design. Fcbricate. Test, QJalification Eo Price Analysis of 3rd Generation Design Solar cell Nodules NPFAMotomla, loc. Si Material Task. Part 1 SiNATMotorola, loc. Processing ExperiPlents on the non-Cz Si Sheet POMotorola, loc. EstalllishPlent of a Production-Re.aay Nfg. Process Utilizing Thin-Si Substrates for Solar Cells POMotorola, loc. Phase 2 of the Array Autonated Assel'lbly Task PO

Motorola. loc. Metallization of Large Si Yafers PO

Motorola, loc. Solar Cell J1Ol1Jles with Parallel-oriented lntercomections PO

Motorola, loc. Studies ard Testing of Anti-Reflective Coatings for Soda-LiPle Glass PO

Motorola, loc. Devel~Plent of a I1ethod of Producing Etch Aesistant Yax Patterns on Solar Cells PO

Motorola, Inc. Laser-Zone Growth in a Ritbon-to-fUbbon Process (ATA), Si Sheet Growth De'lelopAent SHEETMotorola, loc. Anti-Reflective Coatings on Large Area lAlass Sheets SHEETMotorola, loc. Anti-Reflective Coatings ~plied by Acid-Leaching Process SHEETMotorola, loc. Phase I of the Autonated Array Assel'lbly Task mNat'l Research for Geosciences Labs, Inc. Evaluation of LSA Test and Perfornaoce I1ethods l'PFANorth Carolina AT&T Fourdation, loc. Study of Effects of Row-to-Roll Shading on PV Arrays

.. RESNorth Carolina State Lhiversity Rapid Themal Processing of Ion IFllla-t:ed Si

.. ORNorthrop Corp. Lifetifle l1easurel'lent Analysis SINATOregon State Lhiversi ty Devel~Plentof Fluidized-Bed 5i Tec~JIlology SINATP. R. I-«lffl'lan (tt>rlin Inl1Js.) Slicing of Single Crystal {'; Polycrystalline 5i Ingots usifl;l Nulti-fllade Saws SHEETP. R. I-«lffPlan (tt>rlin Inl1Js.) Nultiple-Blade sawing <t1lS) of Si Ingots SHEETPennsylvania State Lhiversity Use of Low Energy Hydrogen Ion lAPlants in Hi-Efficieocy CI'jStalline Si Solar Cells .. mPhotowatt Intl,Inc. Phase 2 of the Array Autonated i\ssel'lbly Task POPhotowatt Intl,Inc. Devel~Plent of Technique for All Coatings Eo Nickel ('; C~per Metallization of Solar Cells .. POPhotowatt Intl.loc. InterPlediate Load Nodule for Test & Evaluation RESPhotowatt Intl.Ioc. In-Depth Study of Si LJafer SUrface Te;<turing POPhotowatt Intl. Inc. Devel~Plent of Low-Cost Pol'jSilicon Solar Cells POPhotowatt Intl.lnc. 4) Kilowatts Solar cell Nodules RESPhotowatt Intl.lnc. Eight Kilowatts of Solar cell Nodules RESPhotowatt.Intl.Inc. Devel~Plent of low-Cost. Hi-Energy/lili t-Area Solar cell Modules POPhotowatt. Intl. Inc. Design. Fabricate. Test. QJalification Eo Price Analysis of 3rd Generatioo Design Solar cell Nodules l'PFAPolytectYlic Institute of New Yorl< - Albany Devel~Plent & DeI'lonstratioo of S)lnthetic Procedures for Polyl'leric UV Stabilizers & Absorbers .. RESPurl1Je Research Fourdation IrK Jet Printing of Silver Metallization for PV .. POPurl1Je Research Foundation MOO Silver I1etallization for PV .. PORCA Corp. Epitaxial Si Growtll for Solar Cells SHEETRCA Corp. Ptlase I of the Automted Array AsseI'lbly Task PORCA Corp. Phase 2 of the Array Autonated Assel'lbly Task PORCA Corp. Inverted Stepanov Tech (1ST) for Si Sheet Growth DevelopAent SHEETRCA Corp. Devel~Plent of I1egasonic Cleaning for Si IoJafers PORCA Corp. Evaluation & Verification of Epitaxial Process sequence for Si SOlar cell PromJction PORCA Corp. Design of Bloc!< II Solar Cell l1Ol1J1es - '9B1 l'PFARCA Corp. Develop. & Evaluation Die Nats. for Prol1Jcing Si Ribbons by the Inverted Ribbon Growth Process SINATResearch Triangle lnsti11Jte CMprehenisve Si Solar cell COA~er tlodeling .. mRocl<\4ell International Corp. Devel~ & DeI'Ionstrate l'1etrodS for Accelerated & Abbreviated Testing of Encapsulated Materials PORocl<\4ell International Corp. ChePlical Vapor Deposition Growth Si Sheet Growth Develo~ent SHEETRocl<\4ell International Corp. Study Progran for Encapsulation Materials Interface for Low-Cost Si Solar Array PO~mingOectronAna~~s~oouw~ Material Itlalysis of P\I Cells and Nodule Test SaAj:lles .. RESSCience Applications. Ioc. Analysis of Cost-Effective PV Panel Design Coocepts Usifl;l Light Trapping POSilicon Tectnology Corp. Advanced IO Yafering Techrology SHEET

<0-l'o

COMPANY ~TRACT TITLE OPEM AREA

Silicoo Tecmology Corp. Develc~:lI'Ient of Methods of Prowcing LASS tJy the Slicing of Silicon Ingots Using 10 saws SHEETSiltee Corp. Develqll'lent of Advanced Methods for CootirlJous Cz Growth .. SHEETSiltee Corp. Enhanced ID Slicing Technology for Silicon Ingots SHEETSolar Energy Research Institute Purification of Metallurgical Grade Si tJy CheFlical Vapor Transport .. SINATSolar Energy Research Institute Solidll1elt Interface Studies of High-Speed S:i Sheet Growth SHEETSolar Energy Research Institute Optinization of Si Crystals for Hi-Efficiency Si Solar cells .. SHEETSolar Pouer Corp. Design. FcDricate. Test. Q.lalification I; Price Analysis of 3rd Generatioo Design Solar cell Nodules NPFASolar TecmolO'y International Solar Cell Panel Developllent Effort POSolarelectronics. Inc. Hydrochlorination of SiCl4 I; Metallurgical Grade Si SIMATSolarex Corp. Process Research of Polycrystalline Si Haterial POSolarex Corp. Block V DocU'lentatioo and Solar cell !'Mules .. RESSolarex Corp. Design of Block V Solar Cell I10Wles - 19B1 NPFASolarex Corp. Process Research of Polycrystalline Si Haterial .. POSolarex Corp. Phase 2 of the Array Autol'lated Assevlbly Task POSolare>: Corp. Analysis of the Effects of IApurities on Si SIMATSolarex Corp. Si Solar cell Noduless with a Total PQI,Ier Capability of 30 Kilowatts RESSolarex Corp. Tecmical feasibility & Effective Cost of Various Wafer Thicknesses for Manufacturing POSolarex Corp. Higr,..Efficiency, High-Density Terrestrial Solar Panels POSolare>: Corp. Solar Breeder feasibility Investigatioo POSolarex Corp. Design. fabricate. Test. Q.lalification I; Price Analysis of 3rd Generatioo Design Solar cell Nodules NPfASolarex Corp. Nitre Solar Energy Test SystePl E'Yaluatioo NPfASolavolt International InterPlediate Load Nodules for Test and Evaluation .. RESSolec International. Inc. Aotireflecti'Ye Coatings to Single Crystal or Polycrystalline Solar cells ~ Use of Roller Coating Nethod .. POSolenergy Corp. InterPlediate Load Nodules for Test and E'Yaluatioo .. RESSollos. Inc. A Nell Nethod of Netallizaticn for Si Solar cells POSollos. Inc. Investigation of Nickel Silicon Hetallization Process POSpeetrolab. Inc. High Resolution, Low Cost Solar cell COntact DeveloDAent POSpectrolab. Inc. Phase 2 of the Array Autol'lated Assef'lbly Task POSpectrolab, Inc. Design, Oe....elopPlent & Prototype Production of Si Solar cell Nodules NPFASpectrolab, Inc. High Resolution. Low-Cost Solar cell Contact DeveloDAent POSpeetrolab, Inc. Analysis of Effects of IJIlpurities Intent-ionally Incorporated into 5i SIMATSpectrolab. Inc. Si Solar cell Process OevelopPlen'l Fabrication and Analysis SINAT$peetrolab. Inc. llevelopPlent of Metallization Process .. POSpectrolab. Inc. Design, Analysis, Ii Test verification of Advanced EncapsulatiCil System .. RESSpire Corp. DevelopPlent & fabrication of a SOlar cell JlTIction Process SysteA POSpire Corp. HerPletic Edge Sealing of PIJ Nodules POSpire Corp. Ion I~lantation of Non-Cz Si .. POSpire Corp. verification & IApleAentation of ElectrCaStatic Bonding Technologies RESSpire Corp. Design of Bloev. V SOlar Cell I10Wles - 19B1 NPfASpire Corp. Block V OocUAentation and Solar Gell ItldUles " RESSpire Corp. Adapt Pulse E:<ciner Laser Processing Technology to Fabricate Cost-Effective Solar Cells .. POSpire Corp. High-Efficiency Solar Cell I10dules .. RESSpire Corp. DevelopPlent of Pulsed Processes for the Manufacture of Solar cells POSpire Corp. Design. fabricate, Test. Q.lalification & Price Analysis of 3rd Generation Design Solar cell Nodules NPfASpire Corp. DevelopandTestmca~wation~terials POSpire Corp. Solar Cell Panel Developllent POSpire Corp. Assess Governflent ().ned JPL Advanced CZO crystal GrOUtfl SystePl l10del SHEETSpire Corp. (SlAulated Pl)lsics. Inc. ) Hi-Rate Lou-Energy Expenditure fabrication of Solar Cells PO

<0(Jl

COI1PI\H't' COtITRACT TI TLE OPOI IIREA

Springllorn Labs, Inc. Nodule Encapsulation Task .. RESSpringllorn Labs, Inc. Fabricaticn of Solar cell Nodules RESSRI International Novel O[.plex ValXlr-ElectrochePlical Method for Si Solar cells SINATStanford University Study of ParaYleters of Heavily Doped Si .. ORStanford University Interfacial Barriers in Hi-Efficiency Crystalline Si Solar cells .. SHEETState Universi ty of New Yorl<-ll1.barYi OXygen & carbcn-Related Defects on Hi-Efficierq Si Solar Cells .. SHEET$l.Perwave TeclYlOlogy, Ire. Nicrowa'/e Heating in a Non-Reactive FBR .. SINAT$l.Perwave TeclYlOlogy, Ire. Depositing 5emconductor Layers Using Microwa'ie Tec~f1iques

.. POTexas Instr~ents, Inc. Si Material Task 1. Part 3 SINATTexas Instnl'lents, Inc. Closed Cycle Process for Low-Cost Solar Si Using a P.otary Chaffier Reactor SINAITexas InstrU1ents, Inc. Si Crystal Gr0\4th & Waferino Process Il'Iprovef4ents SHEETTexas InstrU1ents. Inc. Phase 2 of the Array AYtol'lated Asseroly Task POTexas Instr~ents, Inc. Adv~ced Methods for Continuous Cz GrOYth, Si Sheet Growth Developl'lent SHEETTexas InstrU1ents. Inc. Phase I of the Autol'lated Array Assel'lbly Task POTexas Research & EnlJineering Institute Si Production Process Evaluation SINAIThe Aerospace Corp. Coopositicn NeaSlJreYlents try Analyt...i.cal catalysis SINATThe Boeing Co. Feasibility StudY of Solar Dolle Encapsulation of P\I Arrays ESflPThe Nitre Corp. Mitre Solar Energy Test SystePi Evaluation POTheodore Barry & Associates SLwort of the LifetiYle Cost Performnce (LCP) Modeling Effort PMITheodore Barry & Associates StudY to SUpport OeveloDl'lent of Solar Array Nufacturing Industry Costing Standards PMITheodore Barry & Associates SOlar Array Prolilction SysteA (SAPS) Production Progra'l Hi3"ldbook & Mathenatics Model POTracor 113 Associates Phase 2 of the Array AYtol'lated Asset'lbly Task POTracor 113 Associates AutOAation EQuij:flent Develoj:flent and Modification POTracor 113 Associates Phase 2 of the Array AYtol'lated Assel'lbly Task POTracor 113 Associates Use of Glass Reinforced Concrete (GRC) as a Substrate for PV Molilles POTyl~ Corp. Vitregraph Coaings on Mullite SHEETlk'iderwriters Labs. Inc. PV Mol1Jle and Array Materials .. RESlk'iderwriters Labs. Inc. SOlar ArraylNodule safety ReQuirefllents .. RESLhioo Carbide Corp. SOlar Cells on ttllticrystalline Slbstrates of Si Refined try OS/PtIS Process SINATLhioo Carbide Corp. Sil~e to Si Process .. SINATLhiversity of california at Los Angeles OptiAization MethodS & Si Solar cell Nuyerical Models .. ORLhiversity of california at Los Angeles Si Solar cell FabriC<ltioo Tecmology SHEETLhiversity of california at Los Angeles Delayed fracture of Si SHEETLhiversity of california at Los Angeles Devel~YK4nr of tlethods of Characterizing Si S~.eet Naterial SHEETLhiversity of california at Los Angeles Si Sheet wittl Molecular BeaPi [pitaxy for Hi. -Efficieocy SOlar cells .. ORLhiversity of california at SlJ'lta Barbara Visible-Irradiaree Optical Pr~erties of O:Jnducting PolyRers .. ORLhiversity of california at SlJ'lta Cruz Tine Resolved Spectrosc~ic HeaSlJreYlents .. RESLhiversity of Florida SUrface and Allied Studies in Si SOlar cells .. ORuniversity of Illinois StudY of the Abrasive Wear Rate of Silicon .. SHEETLhiversity of Kentucl<.y Stres~StrainAM~sisofSi~~

.. SHEETLhiversity of Massachusetts S~thetic Procelilres for PolYlleric IN Stabilizers and Absorbers POLhiversity of I'lissouri Determne Effects of PresSlJre of Reactant Gases Where Si is in Contact wi ttl Die & Cootainer & Cootainer Metals SHEETLhiversity of Pennsylvania Analysis & EValuation of Processes ~d EQI.dj:flent POLhiversity of Pennsylvania Hot formng of Si. Si Sheet Growth Oeveloj:flent SHEETLhiversity of Pennsylvania Analysis & Evaluation of I'blille Experil'lental Process S)lsteA OevelopYlent Unit Processes POLhiversity of Pennsylvania Chairmn of Advisory Comittee on LOY Cost Solar cells .. SHEETLhiversity of Pennsylvania Si Solar cells of Near 20\ Efficiency .. ORLhiversity of SOuth Carolina l4eb-Dendritic Ribbon Growth SHEET

CD(j)

COMPANY COIrTRACT TITLE OPal AREAUniversity of Southern california Neasurel'lent of the Absorption Coefficients of Si SHEETUniversity of Southern california Accelerated Aging and Analysis of PolYl'lers POUniversity of Southern california Study of Deep Level Il'lPurities and Defects in Si .. DRUniversity of Toronto Photodegradation in Solar Cell Modules of Slbstrate & Superstrate Design .. RESUniversity of Washirv;lton Si Inversion Layer Solar Gells & Carrier Lifetine Measurel'lents .. DRVarian Associates Developl'lent of Advanced Methods for CZO Growth SHEETVarian Associates Slicing of Si into Sheet t1aterial SHEETWashington Universi tJ' at St. Louis Cz Crystal Modeling Study .. SII1I\TWashington Universi tJ' at St. Louis FluidiZed Bed Reactors for Production of Si fran Silane .. SII1I\Twestirv;lhouse Electric Corp. Dendritic Web-Type Solar Gell Nini-Modules .. RESwestinghouse Electric Corp. Developl'lent of Si Sheet by Dendritic IJeb Ribbon Process .. RESWestinghouse Electric Corp. Methods of Production of Large Areas of Si Sheet b'y 'the Web Dendrite Process SHEETWestinghouse Electric Corp. Si Dendritic LJeb Material Process DevelopFlent POwestinghouse Electric Corp. Phase 2 of the Array Autonated Assel'lbly Task POwestinghouse Electric Corp. Process for High Gapacity Arc Heater Prod of Si for Solar Arrays POWestinghouse Electric Corp. Part 2-Develop & Define Requirenents of Purity for Solar Gell Grade Si POWestirv;lhouse Electric Corp. Process Research of non-Cz Si Naterial POwestinghouse Electric Corp. Single Crystal 5i Dendritic Web Rbbon & Hi-Efficiency Solar Cells .. SINI\Twestinghouse Electric Corp. Dendritic Web Process DevelopFlent .. SHEETwestinghouse Electric Corp. Laser Assisted Solar cell Metallization Processing .. POWestirv;lhouse Electric Corp. High-Efficiency Solar Cells on 5i Web .. DRWestinghouse Electric Corp. Process Researctl of Non-Cz Si Material .. POWilkes College ElectroCheAical Research on Solar Cells aJ)j PV natetials .. RES\.lyle Laboratories Durability/Reliability PerforPlance Criteri<l r; Test I1ethods for Array SUbsystel'l EleAents .. RESXerox Corp. Center Punched Solar cell Nodule Developrqent Effort POLEGEND: .,Cz - CZochralskiDR - Device Rese-3rchNPFA - I10dule Perfornalce r; Failure AnalysisOPEN - ~en or Continuing ContractsPA&I - Project Analysis l; IntegrationPO - Process DeveloprqentPV - Photovoltaic(s)RES - Reliability £; Engineering SCienceSHEET- Advancoo Silicon SheetSi - SiliconSINAI- Silicon I1aterials

More Technology Advancements

Dendritic web silicon ribbons are grown to so/ar-{;ellthickness. Progress is shown by experimental ribbonsgrown in 1976 and 1978 and a ribbon grown in aWestinghouse Electric Corporation pilot plant.

The edge-defined film-fed growth silicon ribbons aregrown to so/ar-{;ell thickness. A DOE/FSA-sponsoredresearch ribbon grown in 1976 is shown next to anine-sided ribbon grown in a Mobil Solar EnergyCorporation funded configuration.

Czochralski silicon crystals as grown aresawed into thin circular wafers. (Support forthis effort was completed in 1981.)

FRAME

SEAL

GASKET

Typical superstrate module design is shown with theelectrically interconnected solar cells embedded in alaminate that is structurally supported by glass.Materials and processes suitable for mass productionhave been developed using this laminated design.

GLASS (STRUCTURAL}

Prototype modules have passed UL 790 Class Aburning brand tests which are more severe thanthis spread of flame test. This 13.7% efficiency prototype module was made

by Spire Corporation. If solar cells more recentlyfabricated from float-zone silicon wafers were madeinto a module, the module efficiency should beabout 15%.

Photovoltaic Applications1975

u.s. Coast Guard buoywith photovoltaic-powerednavigational light.

Later...

House in Carlisle, Massachusetts, with a 7.3-kWphoto voltaic roof-top array. Excess photo voltaic­generated power is sold to the utility. Power isautomatically supplied by the utility as needed.

1985

Photo voltaic-powered corrosion protectionof underground pipes and wells.

A 28-kW array of solar cells for crop irrigationduring summer, and crop drying during winter(a DOE/University of Nebraska cooperative project).

1.2 MW of photovoltaic peaking-power generationcapacity for the Sacramento Municipal Utility District.(The 8 x 16 ft panels are mounted on a north-southaxis for tracking the sun.)

JPL 400-279 10/85(5101-279)