Solar Energy Materials 23 (1991) 117-128 North-Holland
Solar Energy Materials
Recent advances in thin-film solar cells Jack L. Stone Solar Energy Research Institute, 1617 Cole Boulecard, Golden, CO 80401, USA "What goes around, comes around," as the saying goes. Once again, the focus on renewable energy has come about as a result of the crisis mentality. Higher fossil fuel prices, uncertain supplies from the Middle East, and growing concerns for the environment have put clean sources of energy high oa the public's agenda. Fortunately, over the last ten years research progress for the various photovoltaic options has been 0uietly, but steadily, going forward. Progress for all of the important thin-film options is reviewed in this paper, and future requirements in terms of expanded markets and larger production capacity are discussed.
1. Introduction
Since the 1970s, when serious interest in terrestrial photovoltaics started, the number of options that received worldwide research attention have continually been narrowed. Today, the number of serious candidates numbers perhaps a dozen for flat-plate and concentrator systems. The thin-film options center primarily around amorphous silicon and its alloys, polycrystalline copper indium diselenide (CIS) and its alloys, polycrystalline cadmium te!luride, polycrystalline silicon on low-cost substrates, and crystalline gallium arsenide with reusable substi"ates. Various single-junction and multi-junction approaches are being considered for higher efficiency. All of the options appea~- to be scaleable and amenable to mass production, have the potential for efficiencies greater than ten percent, and have demonstrated reliability in the short term, which gives confidence for twenty- to thirty-year lifetimes. Research groups around the world are addressing the remaining questions associated with the light degradation of amorphous silicon, encapsulation requirements for CdTe, a demonstrated production worthiness of CIS and crystalline GaAs, and demonstrated truly thin films for polycrystalline silicon.
2. Amorphous silicon thin films
Efficiency progress in laboratory-scale devices has not improved significantly over the last several years, as much of the worldwide effort has been focused on scaling to larger sizes and designing manufacturing capacity in the 1-I0 MW plant sizes. Although early market entries were made for consumer products, today's research appears to be directed toward small to medium power applications; larger, low-efficiency modules for remote applications; and new device designs and 0165-1633/91/$03.50 © 1991 - Elsevier Science Publishers B.V. All rights reserved
J.L. Stone / Recent adrances in thin-film solar cells
118
Table 1 Performance of single-junction amorphous silicon devices Voc (V)
J.~c FF ( m A / c m 2)
0.967 0.850 0.857 0.895
17.7 19.0 18.7 18.5
Eft. (%)
0.703 12.0 0.742 12.0 0.749 12.0 0.723 12.0
Area Organization (cm 2) 0.03 1.05 1.0 !.0
0.891 19.13 0.950 17.8
0.701 0.70
11.95 0.255 11.8 1.0
0.915
18.1
9.705
11.7
0.886 0.91
17.6 18.19
0.740 11.5 0.693 11.5
0.894 0.923 0.89
18.20 18.56 17.72
0.688 11.2 1.0 0.66 11.28 1.2 0.706 11.1 NG
1.0 1.0 0.09
GL = graded layer SL = superlattice SS = stainless steel Text. = textured NG = not given BL = buffer layer
Description
Osaka University' TKD-SEL Mitsui Toatsu Hitachi
SnO 2 / p ( a - S i C X E C R ) / i / n / A g / g l a s s ITO/Text. SnO 2 / p ~ a - S i C ) / i / n TCO/metal oxide/pta-SiC)/i/n Glass/Text. SnO 2 / t r i p l e p(a-SiC)/ i/n/Ag S o , rex TCO/SLp(a-SiC)/GL(SiC)/i/n ECD SS/n(/zc-Si)/i(a-Si : H: F ) / p(/.tc-Si/ITO) Sanyo Text. TCO/p(a-SiCXphoto-CVD)/ i/n/Ag,SC Fuji Electric TCO/p(a-SiC)/i/n Tokyo Inst. of Tech. T C O / p ( ~ 5 - d o p e d ) / i / n / A I (all photo-CVD) Kanegafuchi Text. SnO 2 / p ( a - S i C ) / i / n / A g Teijin Polymer/ AI / SS / p / i/ n / ITO SHARP G l a s s / T C O / p ( a - S i C ) BL(a-SiC)/ i/n/ITO/Ag MTG = milky tin oxide on glass TCO = transparent conducting oxide TFS = thin film silicon (a-Si)
SC = saperchamber
materials for the next-generation, utility-type power modules. The light-induced degradation for single-junction devices is typically 10%-20% for a well designed product, and degradation below 10% has been reported on selected samples.
2.1. Single-junction deHces State-of-the-art efficiencies of 12% have been routinely achieved by many laboratories worldwide. It appears that an improvement of one or two percentage points may be the best that can be achieved from current best-quality materials prepared by glow discharge. Improvement beyond this may well require a significant improvement in material quality, perhaps using an alternative deposition scheme. Very good quality material has been obtained by remote-plasma-enhanced chemical vapor deposition (CVD) and hot-wire deposition. Device results have not been reported, but they should become available shortly. Table 1 is an abbreviated compilation of the best published results for single-junction devices. It should be noted that it may be difficult to reach intercomparative conclusions because an international standard for measurements has not been used. Researchers at SERI are working with the world community to gain consensus as to the specifications for efficiency measurements. Throughout this paper, I have indicated those efficiency values that have been measured under rigorous SERI conditions. Achieving efficiencies necessary to reach values for cost-effective electricity will require multi-junction device structures that more effectively utilize the majority of
J.L. Stone / Recent advances in thht-fihn solar cells
119
Table 2 Performance of multi-junction amorphous silicon cells
Voc
,l~,:
(V)
( m A / c m 2)
0.867 0.545
FF
Eft. (%)
Area (cm 2)
Organization
13.4 23.2
0.610 0.766
7.1 9.7 16.8
0.033 0.16
Osaka University
0.871 0.432
16.4 17.9
0.72 0.68
10.3 5.3 15.6
4.0 4.0 4.0
ARCO Solar
TFS CIGS Total for the 4-Terminal Class/a-Si/CIS/Glass Hybrid
1.478 2.545 0.880 0.715
14.34 7.66 20.5 10.1
0.63 0.701 0.499 0.598
15.04 13.7 NG NG 13.3
0.0648 0.248 1.0 1.0
Osaka University ECD Osaka University
2.54 2.32 1.79 1.75 1.75 2.36 1.72 2.20
7.49 7.3 8.45 8.83 8.99 6.77 9.10 6.74
0.70 0.73 0.748 0.725 0.682 0.67 0.67 0.57
13.3 '') 12.4 11.3 11.2 10.7 10.7 10.5 ''~ 8.5
0.249 1.0 1.0 1.0 1.0 1.0 0.25 0.09
ECD Sumitomo Fuji electric Kanegafuchi SHARP SHARP Solarex Mitstabishi
SS = stainless steel ") SERf-verified.
NG = not given
Description
a-Si Polycrystalline Si Total for the 4-Terminal Tandem
a-Si/x-Si Two Terminal Tandem SS/a-SiGe/a-Si/a-Si/ITO/Grid a-Si CdS/CdTe Total for the Hybrid Combination SS/a-SiGe/a-Si/a-Si Glass/a-Si/a-Si/a-SiGe Glass/a-Si/a-Si Glass/a-Si/a-Si Glass/a-Si/a-Si SS/a-SiGe/a-Si/a-Si a-SiC/a-SiGe/Glass a-Si/a-Si/a-SiGe
TFS = thin film silicon (a-Si)
CIS = copper indium diselenide
the available solar spectrum. Such structures, in two- and three-junction configurations, not only display higher efficiencies but an improvement in stability because of the thinner layers. 2.2. Multi-junction det'ices A variety of configurations have been reported, including two-, three-, and four-terminal structures. All approaches utilize the same features, either a number of different materials or different thicknesses of the same material, to achieve a more complete utilization of the solar spectrum. Performance results for a representative group of these approaches are given in table 2. Very impressive results in the 13%-16% efficiency range have lent confidence to the multi~unction approach for obtaining high performance with these thin films. Research work is directed toward achieving 18% cell structures, which will be required to produce 15% modules (a value required to complement the inherent low cost for producing low-cost electricity).
2.3. Power modules The ultimate use of any photovoltaic rrodu!e is for power applications. Performance data are shown in table 3. Results are impressive; a number of approaches
0.85 24.0 9.0 64.5 47.8 25.38 36.7 16.0 15.9 12.7 48.7 22.8 23.58 23.87 20.75 25.12 19.26
11.96
0.81 39.0 42.16 12.1
1560 1630 270 100 15.6 14.4 2210 132 1802 114 410 1490 12.5 440 119 Jsc = 8.9 410 2066 1154 530 1352 1459
0.67 0.635 0.65 0.68 0.61 0.654 0.68 0.648 0.65 0.55 0.68 0.62 0.63 0.642 0.65 0.50 0.63 0.67 0.64 0.66 0.66 0.49
8.5 8.4 8.3 a~ 8.2 8.1 8.0 NG 7.7 7.5 ") 7.5 7.1 ~'~ 7.0 7.0 6.8 6.7 6.7 6.4 a~ 6.3 ''~ 6.2 a) 6.38 5.0 a~ 3.4 a) 7.8 a) 8.55 NG NG 7.9 '~ NG NG NG NG NG NG 6.7 `" 6.7 ''~ NG 5.8 ~'~ 3.7 a)
NG NG NG
8.8 NG 8.8 `') IUU
4800 600 660 147 3200 930 5235 2852 ! 200 4536 3990
11)06
4800 900 100 100 100 NG 100 11613 400
9o NG 844 NG NG NG 4700 90 NG NG 908 NG NG NG NG NG NG 4939 2625 NG 3882 3653
No 40.3 7.43 0.82 NG NG 36.7 0.77 74.3 NG 7.14 33.9 NG NG 0.893 NG 5.9 32.83 17.54 7.26 22.47 13.65
SS/a-Si
Glass/a-Si Glass/SnO2/p-i-n/AI Glass/a-Si AI/'n/i/p(a-SiC)/SnO 2/Glass Glass/a-Si Glass/a-Si Metal/n/i/p(a-SiC)/SnO 2/Glass Glass/a-Si cj G l a s s / T C O / p i n / m e t a l d) M e t a l / n / i / p ( a - S J C ) / S n O 2/G~ass Glass/a-Si Glass/a-Si Glass/a-Si Folymer/p/i/n Glass/a-Si
Metal / n / i/ p(a-SiC) / T C O /Glass
olass/a-al rmm orlo Glass/a-Si (4 × 1200 cm2 ) Glass/a-Si G l a s s / T C O / p i n / m e t a l I,) Metal/n/i/p(a-SiC)/TCO/Glass
Fuji Electric A R C O Solar Sanyo Sanyo Fuji Electric ARCO Solar Nankai Univ. Chronar Kanegafuchi Solarex Fuji Electric Sumitomo Kanegafuchi Sanyo Kanegafuchi Chronar ARCO Solar Chronar Teijin Solarex Sovonics
}
100 1200 1200 1204 4800 100 4800 147 NI3 NG 1007 4104 4110 1007 2886
NG NG
900 900
c) mass-produced;
NG NG NG 9.27 * NG NG NG NG 8.4 a} 7.2 ~} NG 6.6 a~ 5.8 a) NG 6.0 a)
7.6 2.9 10.5
9.1 3.2 12.3
h} i-layer deposited at 15 ,~,/s; SS = stainless steel.
0.633
10.1 10.05 9.7 7.24 * 9.06 8.9 8.52 8.5 NG NG 6.2 a~ 5.8 "~ 5.1 ~j~ 5.7 '~ 5.3 a~
0.686 0.696 0.687
0.675 0.661 0.667 0.69 0.66 0,64 0.64 0.67 0.637 0.61 0.59
NG NG
0.62 0.59
206.4
NG NG
0.68 0.61
1640 J.~c= 7.8 859 82 609 555 385 1558 19560 193 1119
611.6 441 317
1.03 0.81
262 228
~) SERl-verified; NG = not given:
2,41 39.3 53.7 66.7 39.3 19.1 71.4 22.2 17.0 17,01 25.32 22,76 1.68 48.78 22.95
47.2 20.8
43.5 19.2
Mu~i-juncthgn
NG 12.06 11.7 8.7 43.5 NG 40.9 1.25 6.84 6.03 6.26 23.91 21.0 5,76 15.14
30.2 11.2 41.4
7.69 2.66 10.35
Siemens Solar Industries
Mitsubishi Fuji Electric Fuji Electric Solarex Fuji Electric Kanegafuchi Fuji Electric Sanyo ECD Sovonics Solarex Sovonics ECD Solarex Chro, nar
Semitransparent TFS Filtered CIS Tandem a-Si/a-Si/a-SiGe
G~,:.~s/a-Si/a-Si
a-Si/a-Si/a-SiGe a-Si/a-Si (4 × 1200 cm 2) a-Si/a-Si a-Si/a-Si Glass/a-Si/a-Si SS/a-SiGe/a-Si/a-Si SS/a-Si/a-Si/a-SiGe Glass/a-SiC/a-SiGe SS/a-Si/a-Si SS/a-Si/a-Si (single cell) Glass/a-Si/a-Si
a-Si/a-Si a-Si/a-Si
ARCO Solar
Semitransparent TFS Filtered CIS Tandem
d) see-through (transmittance 15%).
NG NG NG 940 NG NG NG NG 818 838 NG 3646 NG NG 2547
3883
3970
843 843
e~ e~
.r',
122
J.L. Stone / Recent adt'ances in thin-film solar ceUs
deliver in excess of 10 W of power and, most impressively, a single power module recently delivered 74 W before degradation. This power module is the largest thin-film module, measuring 2.5 × 5 feet. A 10 MW facility is under consideration for producing these modules in large lmmbers, and over 400 kW will be deployed at the Photovoltaics for Utility Scale Applications (PVUSA) project in California.
3. Polycrystalline thin films There exist a large class of materials known to be polycrystalline that have the proper physical properties for efficient solar cells. Being neither crystalline nor amorphous, these materials have proven to be important candidates for the most efficient of the thin-film options. Research has been narrowed over the past years to a small group of materials, but exploratory research continues on a number of lesser-known combinations. 3. I. Copper indium diselenide In 1982, researchers were able to demonstrate a 10%-efficient device, an achievement necessary to demonstrate the performance and cost goals set by the Department of Energy's pbotovoltaic program. Early on, researchers recognized that CIS did not exhibit light-induced degradation and that several deposition approaches showed promise for production-viable processing. Currer~.'.y, the leading candidate is electron-beam (e-beam) evaporation or sputtering of sequential layers of copper and indium, followed by selenization. Table 4 shows the current best results for a number of CIS approaches. It is important to note that the small-scale laboratory technologies have been successfully scaled to larger, practical sizes, as shown in table 5. Although the 1.0 eV band gap of CIS is not ideal. achieving 14% gives credibility to the claim that CIS-based materials can reach even higher efficiencies. Much of today's research is centered on the use of alloys, either by substituting sulphur for selenium or gallium for indium. Such alloys allow the synthesis of a full range of band gaps. Researchers believe that 15% singlejunction modules can be reached. In addition to single-junction devices, several four-terminal cascade combinations show promise for even higher efficiencies. These inc!ude C!S/a-Si and CIS/GaAs. The report of greater than 15% for CIS/a-Si represents one of the highest efficiencies for a thin film. Although these results are very impressive, there is evidence that a two-terminal device using only the I - I I I - V I materials for higher efficiency is possible. CIS can be used as the red cell (low band gap) in combination with CuGaSe 2 as the blue cell (high band gap). Practical efficiencies of 23% are predicted to be possible. 3.2. Cadmium telluride It has long been recognized that CdTe has the necessary properties to be a high-efficiency solar cell. Many years ago, single-crystal CdTe cells c.,',16bited
NG Evap. NG Evap. e-beam evap./ selenization Evap. Evap. NG
Z n O / t h i n CdS/CIS (Ga) Ar/ZnO/thinCdZnS/Cu(Galn)Se 2 Z n O / t h i n CdS/CIS CdS/CIS Z n O / t h i n CdS/CIS(Ga) 35.2 37.4 36.0 40.1
I0.9 a) 10.6 ,o 10.1 10.0
463 442 419 391
508 555 446 446
41.0 35.3 36.7 38.9
14.1 12.9 a) 12.5 I 1.3 a)
(mV)
L¢
( m A / c m 2)
Eft. (c,f)
0.075 0.94 0.3 3.5
3.5 0.96 3.56 0.93
0.677 0.657 0.700 0.653 0.666 0.639 0.664 0.641
Area (cm 2)
FF
(a) (a) (a) (a)
(a) (a) (a) (a)
ISET IEC Stuttgart ARCO Solar
ARCO Solar Boeing ARCO Solar SERI
Group
Measured at 83 m W / c m 2 No CdS
G a / I n < 1()%
G a / l n < !(1% G a / I n = 0.27
Comments
All currents normalized to 100 mW/cm2; (a) active and (t) total area; AR = antireflection coating; NG = not given; all measurements global AM 1.5. ~') Based on SERI measurement.
ITO/CdZnS/CIS ZnCdS/CIS Z n O / t h i n ZnSe/CIS
Process
Type
Table 4 Performance of polycrystalline coppe~ indium diselenide cells
t,~
E"
r~
5"
5"
.v-
J.L. Stone / Recent advances in thin-]ihn solar cells
124
Table 5 Performance of polycrystalline CIS and CdTe submodules Type CIS ZnO/thin C d S / CIS(Ga)/Mo, 55 cells ZnO/thin C d S / CIS(Ga)/Mo, 53 cells CdZnS/CIS/Mo CdTe Glass/SnO 2 / C d S /CdTe Gtass/SnO2/CdS /CdTe Glass/CdS/CdTe
Process
Eft. (%)
NG
1!.2 ~'~ 0.641
l~c (A)
NG
9.7
2.51
Evap.
9.6 ~t~ 0.774
Spray Electrodeposition Screen print/ sintering
7.3 '"
0.52
6.8 '') 0.125 6.1
0.171
V,,c (mV)
FF
25.5
0.639
23.9
0.631
1.78
20.5 9.63 51.9
0.639
Area (cm 2)
Pewer (W)
Group/ Comments
938
10.5
Siemens Solar Industries
3905
37.8
Siemens Solar Industries
91.4
0.877
Boeing
0.57
838
6.1
Photon Energy
0.57
100
0.70
AMETEK
0.583
853
5.19
Matsushita
All currents normalized to I00 mW/cm2; all measurements global AM 1.5, and aperture area unless noted. ~ Based on a SERI measurement.
efficiencies in excess of 10%. Recently, researchers have shown essentially similar results for thin-film cells. Table 6 shows results that closely parallel the achievements previously shown for CIS. Several potentially low-cost deposition approaches have been demonstrated. These include reactive sputtering, screen printing, spraying, close-space vapor transport, and electrodeposition, Early problems concerning a stable contact to p-type CdTe have been alleviated by using an n-i-p structure featuring a zinc telluride contact layer. Outdoor tests confirm the stability of this structure. Both CdTe and CIS modules are scheduled for deployment in the PVUSA test bed during the coming year. The best CdTe module and submodule results are given ~n table 5.
4. Crystalline thin films As with their non-thin-film counterparts, thin films of crystalline or semi-crystalline silicon and gallium arsenide exhibit the highest conversion efficiency of the thin-film groups. The materials have a large infrastructure based on past investments from defense programs and the electronics industry. The key questions for these thin-film technologies center on ultimate cost and performance. Research is also driven by the need for low-cost or re-usable substrates and, in the case of gallium arsenide, a process that efficiently utilizes the expensive source materials. Most of the research to date has been on laboratory devices, but recently
Electrodeoosition Spray Electrodeposition Electrodeposition Electrodeposition CSVT CSVT CSVT MOCVD
.,
Process
Active area ~- total area on these superstrate cells. ~}SERl-measured.
Glass/ITO/CdS/CdTe SnO 2/CdS/CdTe Glass/SnO2/CdS/CdTe G!ass/SnO 2/CdS/CdTe/ZnTe G|ass/ITO/CdS/CdHgTe In 203/CdS/CdTe Glass/l'O/CdTe Glass/SnO 2/CdS/CdTe/HgTe Glass/SnO2/CdTe/ZnTe
Type
Table 6 Performance of CdTe devices
1.112 0.79
4
0.1
1.48
0.02 0.31 1.0 1.068
Area (cm 2)
10.6 10.5 ~'~ 10.5 10.6 ~o 9.7 a~
11.2 a}
13.1 12.3 a} 12.0
Eft. (%) 72O 783 783 767 62O 75O 663 745 73O
(mA/cm") 27.9 25.0 23.0 22.36 27.0 17 28.1 21.62 22.2
¢oc (mV)
Lc
0.65 62.7 0.67 0.6963 0.63 0.62 0.56 0.658 0.599
FF
Univ. of Queensland Photon Energy BP Solar AMETEK ISET/Monosolar Kodak (at 75 mW/cm 2) ARCO Solar SMU Georgia Tech
Group
J.L. Stone / Recent adt'ances in thin-film solar cells
126
Table 7 Performance of thin-film polycrystalline silicon and single-crystal gallium arsenide V,,c (V)
Jw ( m A / c m -~)
FF
Eft. (%)
Area (cm 2)
Organization
Description
78.1
Astro Power Astro Power
100/~ thin-film poly Si on ceramic 100 t~ thin-film poly Si on ceramic
4.00 16.00
Kopin Kopin
5/z CLEFT 5 ~ CLEFT
Polycrystalline silicon thh~ film 0.600 0.557
33.00 23.9
0.792 0.642
15.7 8.5
1.020
(10.9) a~
Single-co,stal gallium arsenide thin film 1,011 4.034
27.55 6.55
0.838 0.796
23.3 21.0
~'~ Private communicalion, data not available.
larger-sized submodules have begun to appear. The most current results for the crystalline thin films are shown table 7.
4.1. Silicon thin films Encouraging results have been recently reported for thin films of polycrystalline silicon on a low-cost ceramic substrate with light trapping. A series of design rules were developed that allow high efficiency to be demonstrated using only 5-50 micrometers of silicon. With this thickness of silicon, the device design rules require light trapping, planar films with single-crystal grains at least twice the thickness, a minority carrier diffusion length at least two times the thickness, benign grain boundaries, and a substrate that provides mechanical support and a back-plane conductor. With 50 micrometers of silicon (or thinner), a metallurgical barrier is required between the silicon layer and the ceramic substrate. The results detailed in table 7 are for a material grown on a proprietary ceramic substrate 400 micrometers thick. The entire process is amenable to low-cost production. The substrates can be coated in a continuous line. Modeling predicts that efficiencies in excess of 19% can be achieved. To take advantage of the monolithic approach to fabricating thin-film modules, researchers have proposed a scheme for using this technology to produce large-area modules. Conducting and nonconducting ceramics are integrated into the process. Tandem approaches have been proposed, whereby the thin-film silicon cell is placed in series with a-Si. Experimental results are still unavailable at the time of writing this paper.
4.2. Thin-film gallium arsenide GaAs, like CdTe, possesses a nearly ideal band gap for a high-efficiency conversion of solar energy. Shown in table 7, recent results for thin films of crystalline GaAs have yielded devices with efficiencies greater than 20%. Films of GaAs 5-10 micrometers thick are formed by the cleavage of lateral epitaxi~l films for transfer (CLEFT) process. The ability to cleave larger areas of the material
J.L. Stone / Recent adcances hi thh~-fihn solar cells
127
appears to offer the possibility of larger-area modules that should have greater than 20% efficiency. The research questions center on whether the process can be automated to provide low-cost cells for fabricating the modules.
5. Conclusions
The risks associated with thin-film technologies have been considerably reduced by the pursuit of multi-path options. All approaches have been demonstrated at efficiencies greater than 10% for laboratory-scale cells, and submodules approaching 10% have demonstrated the scalability of the technologies. Larger power modules are in their early stages of development, with field testing being carried out on relatively low-efficiency approaches. Most of the early applications of thin films have been directed toward consumer products and small power battery charging. The next product areas with significant market potential appear to be in automotive power (sun roofs, battery trickle charging, etc.) and in replacing higher-cost crystalline technologies for remote applications, where high performance is not considered to be a premium. Electric utilities are gaining familiarity with thin films through the PVUSA emerging technologies 20 kW segments. As the costs continue to decline and the performance continues to increase, new and larger app!ications will become available. The key to cost reduction will be the construction of new, larger, cost-effective production facilities and the introduction of the newer, higher-efficiency technologies being developed in research laboratories worldwide. Conti.-.ued market growth over the last few years, along with new industrial entries into the field, indicate a healthy situatio~ for the future of photovoltaics.
6. References
Because of the space limitations for this paper, it is impossible to give a detailed reference for all of the results given in the text and tables. Many of the performance results were obtained directly from researchers at the cited organizations. The reader interested in citations for the various entries is directed to the conference proceedings for the three major international photovoltaic conferences that are held each six months on a rotating basis. In particular, the following volumes would be of interest: the proceedings of the IEEE Photovoltaic Specialists Conferences (conferences 12-21 are appropriate for thin-film references representing the period 1977-1990), the proceedings of the International Photovoltaic Science and Engineering Conferences (conferences 1-4 represent the period 1984-1989), and the Commission of the European Communities Photovoltaic Solar Energy Conferences (conferences 1-9 represent the period 1977-1989). The final authority rests with the individual researchers at the cited institutions.
128
J.L. Stone / Recent fldt ances in thin.fi!m solar cells
Acknowledgements The author is grateful to the many researchers from around the world who took the time to share their latest results. Such cooperation is ideal for advancing our technologies. This work was sponsored by the United States Department of Energy under Contract No. DE-AC02-83CH10093. The Solar Energy Research Institute is operated for the DOE by Midwest Research Institute.