Status Of Amorphous And Crystalline Thin Film Silicon .

an aperture area of 0.75 m2 and a power rating of 40 Wp.This module may be the lowest-cost commerciallyavailable today, being offered at 2.25/Wp. EPV’sstrategy is to supply business partners withmanufacturing equipment and technology, at present fora-Si double-junction modules. EPV anticipates only entsofa-Si:H/a-Si:Hdouble-junctionmodules. Under the Thin-Film Partnership, EPV isdeveloping technology to manufacture a-Si:H/nc-Si:H(“micromorph”) cells and later modules to enhance theperformance of the a-Si product.Iowa Thin Film (ITF) is manufacturing a-Si cellsusing roll-to-roll deposition on polymer foils.Itestablished itself as a successful niche player makinglightweight flexible PV generators for a variety ofconsumer applications.AstroPower is a successful crystalline Si PVmanufacturer, using either reclaimed Si wafers from theelectronic Si industry or in-house deposited APEX multicrystalline Si wafers. As next-generation PVtechnology, the development of high-temperatureprocessed Si films on ceramic substrates is undertaken.FF change upon light-soaking or annealing showdifferences, and there is a lack of quantitative correlationbetween the Do generation and pc and FF losses, but bettercorrelation between pc and FF instead [6]. Such findingssuggest that perhaps too much effort went into developingmaterials with lower stabilized (Do) defect densities.Materials with lower stabilized Do densities weresynthesized, but they have not led to improved stabilizedsolar cell performance.It is also found that degradation and annealing behavior inmost instances can typically be separated into twomechanisms: a “fast” (low-temperature annealable) andslow (below 100oC non-annealable) mechanism. Anycomprehensive Staebler-Wronski model must account forboth mechanisms, which appear linked. The effects of thetwo mechanisms on degradation and annealing are largeenough to affect the power output of a-Si:H-based modulesby 10% [7]. Although previous work focussed onunderstanding the creation/annealing of localized metastableDo defects by breaking weak bonds, work on a-SiGe:Halloys suggests both a localized and global mechanismdominating the degradation and annealing behavior [8].Interest in mitigating Staebler-Wronski degradation remainshigh, because if it were possible to maintain today’s solarcell performance in the annealed state, 16%-efficientmultijunction cells and 10% commercial modules wouldlikely be a reality. However, to date, no stable a-Si:H orrelated alloys have been produced that result in desirablestabilized solar cell performance. It is now widely acceptedthat the degradation is “intrinsic” to a-Si:H and relatedalloys. The instability appears not to depend systematicallyon hydrogen content or on how tightly bonded the hydrogenis. In fact, some of the more stable (“deuterated and highhydrogen-dilution, lower substrate temperature”) cellrecipes were achieved with materials that contain largeramounts of loosely bonded hydrogen [9]. It appears that alack of scientific understanding and guidance frominadequate models relating material properties and solar cellperformance remains an obstacle toward progress.The prospects for significant increases in a-Si:H-basedstabilized PV module efficiencies (to reach 10% commercialmodules) appear somewhat diminished, unless someunexpected breakthrough occurs. This situation has the a-SiPV industry focusing on market applications where lowerefficiencies are less critical. Elimination of the a-SiGe:Halloy from the cells has also been considered (because of Gecost and Ge-alloying being an impediment for higherdeposition rates), but are not implemented by entities usingthem because even the modest efficiency loss associatedwith such change is deemed unacceptable. Outside theUnited States, we are not aware of significant new a-Simodule manufacturing capacity under development.Entities in Japan and Europe are rather focussing ondeveloping commercial micromorph devices with presumedhigher efficiency, holding off on adding manufacturingcapacity until such developments reach greater maturity.3. Amorphous Silicon StatusIn recent years, the wish list of the a-Si PVmanufacturers has been:(a) reduce light-induced degradation(b) increase the deposition rate by a factor of 2 – 10(c) find processes to fabricate a better narrow-bandgapbottom cell for a double- or triple-junction device.The ubiquitous way of achieving greater a-Si:H devicestability and greater cell voltages has been to deposit thelayers using high hydrogen dilution. The industry usesdc or rf plasma-enhanced chemical vapor deposition(PECVD), at low substrate temperatures ( 200oC).However, hydrogen dilution reduces the deposition rates.The a-Si national team has reached agreement thatadjusting the operating parameters (typically,temperature, power, pressure, and gas flow) on thepresent dc and rf PECVD deposition systems is notlikely to result in acceptable stabilized deviceperformance at higher deposition rates. In particular, thenarrow-bandgap a-SiGe:H bottom cell, already thelimiting component cell in a multijunction stack, hasstubbornly resisted all attempts to increase its depositionrate without significant loss in stabilized performance.Therefore, alternative deposition methods wereinvestigated to overcome this limitation, with limitedsuccess to-date. The most promising method appears tobe very high frequency (VHF) PECVD deposition [5].With respect to Staebler-Wronski degradation, it hasfinally become clear that the degradation of theelectronic properties, such as solar cell fill factor (FF) orthe photoconductivity (pc) of the intrinsic layer (i-layer),is not controlled by a simple, single, straightforwardmechanism. In the past, the generation of neutraldangling-bond defects (Do) was assumed to be adominating mechanism. It is now confirmed that thekinetics with which observables such as Do, pc, or cell4. Thin-Crystalline Si Film Solar CellsEarly in 2002, the Thin-Film Partnership assembled asubteam to the present a-Si team, addressing various issues2

cell results are obtained when the i-layer deposition rate isslow ( 0.5 nm/s). However, some deposition methods(such as “hot-wire” CVD) have shown that different filmstructures and modest cell efficiencies (about 5%) can beobtained at higher deposition rates (1 – 2 nm/s).Thesecond observation is that some nanocrystalline Si films are“porous” and therefore subject to post-deposition oxidation.After oxidation, solar cell performance degrades severely.Such films have to be avoided, because it has been reportedthat oxidation occurs even when such devices are “capped,”e.g., by a p-layer in a solar cell.An area of concern for nc-Si is that in recent years cellefficiency records have been somewhat stagnant. Given thecurrent performance, it appears that there is opportunity forincrementally improving a-Si:H-based multijunctions byusing the nc-Si cell as a bottom cell. Given the large varietyof nc-Si films that can be produced, it may well be possiblethat cell recipes, at greater deposition rates, can be foundthat result in even higher cell efficiency.Next, we will briefly discuss the higher-temperature,large-grain crystalline Si films.Several entities areworking on growing larger-grain-size thin Si films on glass,e.g., CalTech and the University of Delaware (IEC), relyingon “induced” crystallization schemes. It is presently notclear what the grain-size requirements may be for highefficiency solar cells. Two approaches to date have resultedin larger grain sizes. AstroPower uses an approach to thinfilm solar cells that relies on melt-recrystallizing a CVDdeposited ( 100-µm-thick “growth” layer) on a ceramicsubstrate. Then, active solar cell layers are depositedepitaxially by chemical vapor deposition. The active Silayers used are a 5-µm–thick back-surface field and contactlayer (boron-doped 0.1 Ωcm) and a very lightly borondoped 30 35-µm-thick absorber layer. The cell emitter isformed by a P-diffusion process. For years, AstroPowerworked on the development of a ceramic substrate thatwould allow deposition or recrystallization of Si near itsmelting temperature. The parameters for the best cellobtained are as follows: η 9.2%, VOC 0.543 V, FF 0.732, JSC 23.11 mA/cm2. Because the substrate isinsulating, monolithic interconnection of the cells can beaccomplished and has been demonstrated. Interestinglyenough, the current generated and collected in these 30 35-µm-thick absorbers is rather similar to those currentdensities measured in the only 1.5 – 2-µm-thick nc-Siabsorbers. There are indications that, contrary to popularbelief, current generation in thin, “optically enhanced” Sifilms is quite efficient; however, the collection of carriers,especially those generated in thicker films some distanceaway from the junction, remains a challenge.Wang et al. at NREL developed a process for growinglarge-grain films in the 750 – 950 oC substrate temperaturerange using an iodine vapor transport reaction [13].Heterojunction cells using the so-called “HIT” structuredeveloped by Sanyo [14] were fabricated at NREL.It appears quite beneficial to have team collaborationamong researchers pursuing both nanocystalline lowtemperature and large-grain high-temperature Si filmapproaches. Previously, thin crystalline Si solar cells havebeen attempted by groups having either a bulk (wafer orof crystalline thin-film solar cells. About a year into theprogram, the following results were obtained:(a) Several entities in the United States achieved “nearstate-of-the-art” nanocrystalline (nc-Si) cells (about7% efficient) using low-temperature depositionapproaches (the term “microcrystalline” has alsobeen used for such films; however, grain sizes arealways 1 µm).(b) Similar to findings by the group in Jülich [10], thebest cell efficiencies are found for small-grainmixed-phase films, not larger-grain films.(c) United Solar, using a reproducible butinhomogeneous deposition method, produced solarcells gradually varying from amorphous to nc-Si ilayers.(d) Because many researchers believe that larger grainsizes should lead to greater device performance,approaches are pursued to achieve crystalline Si thinfilms with larger grain sizes, involving hightemperature deposition or recrystallization.First, we will discuss the low-temperature approaches.Companies involved are United Solar, EPV, and MVSystems, as well as several universities and the group atNREL. Most work is geared toward replacing thebottom cell in current a-Si cells. The efficienciesachieved to date appear too low to justify developmentof single-junction nanocrystalline PV technology. Tofunction as a bottom cell, the nc-Si cell has to generate 24 mA/cm2 and have a sufficiently high red response togenerate 8 mA/cm2 in a triple-junction stack cell or 12mA/cm2 for a double-junction device. Using a state-ofthe-art backreflector, Yang et al. at United Solar wereable to demonstrate such performance with nc-Siabsorbers that are 2 µm thick, using a bottom celldeposited (at low deposition rates) by modified VHFPECVD [11].It appears from the optimization of such nc-Si cellsthat experimentally a regime is encountered wherehigher short-circuit current densities (JSC) are obtainedby “sacrificing” open-circuit voltage (VOC) or vice versa.Such a trade-off between voltage and current is notuncommon both for thin-film and wafer-basedcrystalline silicon [12]. It remains to be seen what thehighest voltages for the nc-Si cell – in conjunction withthe required high JSC values – will be.Betteroptimization and better understanding of the mechanismleading to such a trade-off are both required.Researchers at United Solar have prepared cellsspanning the amorphous to nanocrystalline transition. Itis found that cells with a large fraction of amorphousphase in their i-layer show large voltages, but also lightinduced degradation. Light-soaking actually enhancesVOC, but decreases FF. With such unstable i-layers, thehigh JSC values required for multijunctions could not bedemonstrated. It is important to point out that even thestable cell recipe with the largest grain size and bestordered i-layers did not result in the highest efficiency,similar to the findings of Luysberg et al. [10].Two additional universal findings about nc-Si cellsshould be mentioned. As in a-Si alloys, to date, the best3

[4] P. Nath, C. Vogeli, K. Jones, A. Singh, I. Garcia, and S.Guha, “Field Installed Peel and Stick PV Laminates forMetal Roofs,” Proc. of the 28th IEEE PhotovoltaicSpecialists Conf. (2000) 607.ribbon) crystalline Si background or an a-Si background.For wafers, combining such expertise at Sanyo [14] hasresulted in a fast-growing commercial product with a topefficiency rating in the entire PV industry.Ourexperienced a-Si cell makers have quickly adapted theirknow-how to make reasonable nc-Si cells.[5] S.J. Jones, T. Liu, X. Deng, and M. Izu, “a-Si:H-BasedTriple-Junction Cells Prepared at i-Layer Deposition Ratesof 10 Å/Sec using a 70 MHz PECVD Technique,” Proc. ofthe 28th IEEE Photovoltaic Specialists Conf. (2000) 845.5. SummaryIt is argued that the efficiency potential for a-Si solarcells evident today may be somewhat limiting theircompetitiveness for applications where PV product issold “by the watt,” as crystalline Si technologies appearto maintain an almost 2:1 efficiency lead. Lowerefficiencies may be less of a factor in building-integrated(BIPV) applications. Dedicated BIPV products maycurrently be considered a “niche” application, but willultimately have huge market volumes by themselves. Inthese applications, other a-Si performance advantagescould play a role, e.g., better temperature coefficients.“Micromorph” a-Si/nc-Si multijunctions offer thepotential to incrementally improve a-Si:H multijunctiontechnology. Based on the limited present knowledge, nodramatic increase in the performance over a-Si/a-SiGemultijunctions is to be expected. The ultimate potentialof the nanocrystalline cell is yet unknown. Althoughconventional wisdom equates larger grain size withhigher cell performance, near the a-Si to nc-Si transition,a regime has been found where increased grain sizes andbetter grain order lead to lower cell performance.Significant further improvement for the efficiencies oflarge-grain high-temperature deposited/recrystallizedcells is also required to make them suitable candidatesfor single-junction modules. As in the case of nc-Si, thepotential of such technology is presently unknown,requiring further consistent research to determinesuccess.[6] J.M. Pearce, R.J. Koval, R.W. Collins, C.R. Wronski,M.M. Al-Jassim, and K.M. Jones, “Correlation of LightInduced Changes in a-Si:H Films with Characteristics ofCorresponding Solar Cells,” Proc. of the 29th IEEEPhotovoltaic Specialists Conf. (2002) 1098.[7] B. von Roedern and J.A. del Cueto, “Model for StaeblerWronski Degradation Deduced from Long-Term, ControlledLight-Soaking Experiments,” Materials Research SocietySymposia Proceedings (2000) Vol. 609, A10.4[8] J.D. Cohen, J. Heath, K.C. Palinginis, J.C. Yang, and S.Guha, “Light-Induced Annealing of Deep Defects in LowGe Fraction a-SiGe:H Alloys: Further Insights into theFundamentals of Light-induced Degradation,” MaterialsResearch Society Symposia Proceedings (2001) Vol. 664,A12.5[9] J. Yang and S. Guha, “Amorphous Silicon AlloyMaterials and Solar Cells Near the Threshold ofMicrocrystallinity,” Materials Research Society SymposiaProceedings (1999) Vol. 557, 239.[10] M. Luysberg, C. Scholten, L. Houben, R. Carius, F.Finger, and O. Vetterl, “Structural Properties ofMicrocrystalline Si Solar Cells,” Materials Research SocietySymposia Proceedings (2001) Vol. 664, A15.2.6. AcknowledgementThis paper was made possible by the insight that theauthor, a program manager, gained with those doing theoriginal research. Sharing of results and continueddebate over the mechanisms controlling and limiting cellperformance are much acknowledged.[11] J. Yang et al., this conference.[12] M.Y. Ghannam, S. Sivoththaman, H.E. Engamel, J.Nijs, M. Rodot, and D. Sarti, “636 mV Open-CircuitVoltage Multicrystalline Silicon Solar Cells on PolixMaterial: Trade-Off between Short-Circuit Current andOpen-Circuit Voltage,” Proc. of the 23rd IEEE PhotovoltaicSpecialists Conf. (1993) 106.References[1] B. von Roedern, K. Zweibel, E. Schiff, J.D. Cohen,S. Wagner, S.S. Hegedus, and T. Peterson, “”ProgressReport on the Amorphous Silicon Teaming Activities,”CP394, NREL/SNL Photovoltaics Program Review,November 1996, (ed. C.E. Witt and J.M.Gee), AmericanInstitute of Conference Proceedings 394, 1997), 3.[13] T.H. Wang, T.F. Ciszek, M.R. Page, R.E. Bauer, M.D.Landry, Q. Wang, and Y.F. Yan “Atmospheric PressureIodine Vapor Transport for Thin-Silicon Growth,” Proc.2001 NCPV Program Review Meeting, NREL/EL-52031065.[2] S. Guha, J. Yang, A. Banerjee, K. Hoffman, S.Sugiyama, J. Call, S.J. Jones, X. Deng, J. Doehler, M.Izu, and H.C. Ovshinsky, “Triple-Junction AmorphousSilicon alloy Manufacturing Plant of 5 MW AnnualCapacity,” Proc. of the 26th IEEE PhotovoltaicSpecialists Conf. (1997) 607.[14] Y. Kuwano, S. Nakano, T. Takahama, T. Masuyama,M. Isomura, N. Nakamura, H. Haku, M. Nishikuni, H.Nishiwaki, and S. Tsuda, “A More than 18% Efficiency HITStructure a-Si//c-Si Solar Cellusing ArtificiallyConstructed Junction (ACJ),” Materials Research SocietySymposia Proceedings (2001) Vol. 258, 857.[3] S. Guha, 2003, private communication.4

Form ApprovedOMB NO. 0704-0188REPORT DOCUMENTATION PAGEPublic reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of thiscollection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 JeffersonDavis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.1. AGENCY USE ONLY (Leave blank)2. REPORT DATEMay 20033. REPORT TYPE AND DATES COVEREDConference Paper4. TITLE AND SUBTITLEStatus of Amorphous and Crystalline Thin Film Silicon Solar Cell Activities5. FUNDING NUMBERSPVP3.50016. AUTHOR(S)B. von Roedern7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)8. PERFORMING ORGANIZATIONREPORT NUMBERNational Renewable Energy Laboratory1617 Cole Blvd.Golden, CO 80401-3393NREL/CP-520-335689. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)10. SPONSORING/MONITORINGAGENCY REPORT NUMBER11. SUPPLEMENTARY NOTES12a.DISTRIBUTION/AVAILABILITY STATEMENT12b.DISTRIBUTION CODENational Technical Information ServiceU.S. Department of Commerce5285 Port Royal RoadSpringfield, VA 2216113. ABSTRACT (Maximum 200 words)This paper reviews the recent activities and accomplishments of the national Amorphous Silicon Team and a (crystalline) thinfilm-Si subteam that was implemented in 2002 to research solar cell devices based on thin crystalline Si based layers. Thispaper reports the evolution of team organization, the technical highlights from the recent team meetings, and an outlook oncommercialization potential.15. NUMBER OF PAGES14. SUBJECT TERMSamorphous silicon; crystalline silicon; Thin Film Partnership; national team; potential17. SECURITY CLASSIFICATIONOF REPORTUnclassifiedNSN 7540-01-280-550018. SECURITY CLASSIFICATIONOF THIS PAGEUnclassified19. SECURITY CLASSIFICATIONOF ABSTRACTUnclassified16. PRICE CODE20. LIMITATION OF ABSTRACTULStandard Form 298 (Rev. 2-89)Prescribed by ANSI Std. Z39-18298-102

In November 2002, BP Solar made the business decision to terminate all commercialization of thin-film PV technology. The plant in Toano was shut down and put up for sale. United Solar Systems manufactures modules on stainless steel ing usan Si/aa--SiGe/a-SiGe iptrle-junction cell structure. Until now, a line with 5 MWp annual capacity was used [2].