定制高端激光薄膜

光学薄膜论坛

 找回密码
 注册
搜索
本站邀请注册说明!
查看: 1031|回复: 0

[原创] Materials Challenges for Terrestrial Thin-Film Photovoltaics

[复制链接]
发表于 2008-2-27 09:23:35 | 显示全部楼层 |阅读模式
Materials Challenges for Terrestrial Thin-Film Photovoltaics
& v1 V& [3 s: \3 E8 kAlvin D Compaan. JOM. New York: Dec 2007. Vol. 59, Iss. 12; pg. 31, 6 pgs - b7 d. C& u! z% K8 z

' q9 E$ y: `5 x( E( B/ gAbstract (Summary)
$ H. N  l% `6 [3 u0 A6 P! W1 qPhotovoltaic (PV) materials are often classified into first, second, and third generations.. C% y- C' @9 }+ e+ D5 [7 t

/ \! m7 x8 `' |/ p »  Jump to indexing (document details)
& n3 q' ]% e0 N: y7 uFull Text (5087  words)
) y" R; L# H' G, g' LCopyright Minerals, Metals & Materials Society Dec 2007. }7 o( _1 ~3 ~5 c7 o
- G8 a2 G" U2 t( h3 M
[Headnote]
7 l6 H$ D; }2 b: D' d4 `8 uThe terrestrial photovoltaics market has been dominated from the beginning by crystalline silicon and by cast multicrystalline silicon. Continuing improvements in materials quality and innovative designs are responsible for keeping silicon solar cells at a market share of about 85%. However, in the past two years, thin-film solar modules based on amorphous silicon, cadmium telluride, and copper indium gallium diselenide have gained a strong foothold in the market, particularly in the United States and Europe. This article will briefly review the status of silicon solar cells and then discuss developments, opportunities, and materials challenges in thin-film solar cells.
) g) Z' a% u% B, k/ Q; f9 V0 y6 ]* X3 v: e  D3 H$ n* Q/ Y& Z1 Z

! {8 l) u4 x5 \2 K9 L" m  fINTRODUCTION
; U: D7 ~1 t% H4 ?/ e( V% T" z- r
The production of terrestrial solar modules has been growing at the stunning pace of 30% to 50% per year for the past eight years. Worldwide in 2005. the production of crystalline silicon and multicrystalline silicon wafer-based modules accounted for 84% of photovoltaics (PV) production. In addition, ribbon silicon modules, which also must be assembled from smaller wafers, accounted for another 3%. Thus, wafer silicon accounted for about 87% of PV production.1 However, 2006 represented a breakthrough year for thin-film PV module manufacturing. In the United States, dedicated thin-film (TF) manufacturers of cadmium telluride and amorphous silicon modules were number one and three in production. As both wafer silicon and TF solar modules mature, we must ask whether there are significant materials challenges remaining or whether the principal challenges are engineering, marketing, and publicpolicy. The purpose of this overview is to review the state-of-the-art in TF PV and to identify where important materials challenges arise.
- h8 A  P* R+ K7 B# u/ N% E: v- c- T
As reviewed in this issue by John Merrill and Donna Senft. high-efficiency, high-value cells for space have provided an important technology driver for PV. Historically this was the case for silicon and then for multifunction III-V cells. Now, as the efficiencies of solar cells are pushed well above the 31% blackbody Shockley-Queisser single-junction limit,2 reaching more than 40% for triplejunction cells under high concentration,3 these high-efficiency cells become more and more attractive for some terrestrial concentrator applications.
. Q; m5 Z" ~) l9 ?4 [; _
7 B) h/ w2 @9 qThin-film solar cells have developed mostly without this high-value market pull. This partly explains the slow development of amorphous silicon, cadmium telluride. and copper-indium-gallium-diselenide (C1GS) thin-film (TF) devices. The extremely rapid growth of TF modules recently is partly due to the strong market demand coupled with the silicon feedstock shortage but also an indication of technological advances that are the result of long and diligent work on the science and technology.
9 h. V8 @8 K" E7 k
& S8 e; P" J# W* a% ?Although large-scale manufacturing of TF modules is now successful (most notably by Kaneka, First Solar,4 and United Solar Ovonics5), a wide range of materials challenges remain in the TF PV field. Whereas a strong knowledge and experience base from the electronics and opto-electronics industry has supported silicon and I1I-V PV technology, that is not the case for the thin-film materials, except for a-Si:H. Amorphous silicon PV has some synergy with the flat panel display industry that uses a-Si:H for thin-film transistors (TFTs) for active matrix liquid crystal display technologies. However, a major distinction is that TFTs are majority-carrier devices (n-type field-effect transistors), whereas solar cells are bipolar devices. This significantly increases the demands on materials and processing science and technology.
" e7 I3 _: M# `9 T# S! F: Z8 Q3 d: U. |0 _
This overview will briefly review best efficiencies for all PV materials, revisit the physics of the silicon homojunction device, and provide an introduction to materials issues in PV fora-Si:H, CdTe. and CIGS TFs. It will not discuss dye-sensitized or polymer solar cells partly because there is very little commercial production yet.
0 i; y3 ~/ [' h0 s+ S2 H/ k  y6 r
1 e: h6 r7 X* V* \See the sidebar for an overview of PV materials and efficiencies.* L: M8 {6 a' C: ~3 k" i
1 x, [2 U- H( K- q4 A
WAFER-BASED MODULES-SILICON STILL DOMINATES THE MARKET8 |2 P. y: c" I

! _; |1 W2 K4 ?The typical wafer and band structure of a silicon cell is sketched in Figure 1, showing an extended neutral region in the bulk of the absorber and a depleted region near the junction where the built-in electric field is strong. (The transparent, conducting window/front contact is not shown nor is the metal back contact.12) Recall that the width of the depletion region for a one-sided step junction is inversely proportional to the square root of the doping concentration. (Silicon cells are almost always n-on-p structures since the electron diffusion length in p-Si is much greater than holes in similarly doped n-Si.) The major challenge for crystalline or multicrystalline silicon is that the absorption coefficient is relatively low between the indirect gap at 1.1 eV and the direct gap at 3.2 eV. Thus, most visible and near-infrared wavelengths penetrate tens of micrometers through the depletion region and into the neutral region where many of the carriers are generated. This puts atight constraint on the purity of the silicon material to achieve long minority carrier (electron) diffusion lengths. This is particularly difficult for the multicrystalline (cast) silicon wafers which are decorated with grain boundaries that have high densities of recombination centers. Also, the mc-Si will usually have higher intrinsic impurity levels. Recent progress has resulted from advanced gettering techniques to reduce the bulk impurity level.13,14 Wire saw techniques are just beginning to yield wafers less than 200 µm15 so the neutral region is long compared with the absorption length, and electrons spend most of their lifetime diffusing in the neutral region of the p-type absorber rather than drifting under the influence of the field in the depletion region. This will lead to poor carrier collection if the materials quality is not high.$ P! V9 n+ B' ?* |, \6 _
5 D1 n. o- A3 O" T3 H
As seen in Figure A, single-crystal silicon cells have reached efficiencies of 24.7% compared with predicted achievable efficiencies for AM 1.5 of 27%.16 This efficiency has been achieved with several innovative optical techniques and grid structures that maximize the active area. Note that the heavily doped region at the back contact produces band bending that will serve as an electron reflector and reduce unwanted electron collection while enhancing hole collection at the back contact.0 @. `6 F1 ?) E7 T6 J

) q* w# C. o6 b$ D/ l( h; YThe assembly of large modules from discrete solar cell wafers has the disadvantage of requiring multiple steps of interconnection and bonding, etc., to produce a large module from, for instance, 20 cm cells. However, a major advantage is that individual cells can be tested and low-performing cells rejected before assembly into modules. Such individualized testing before assembly cannot generally be done for thin-film modules.
" f, z5 x3 J# W# j2 c) X3 N' [' F2 B0 }/ q3 j: |
Because of the potential for large-area deposition and monolithic integration, thin-film modules are generally regarded as having much greater potential for cost reduction since the cell assembly process is avoided. But correspondingly there are much greater challenges in scaleup. The typical manufactured module efficiency for the TF materials is usually only 50% or less of the laboratory record cell, although again individual champion modules will exceed this. There are a variety of causes for this difference, including interconnect losses, optical transmission losses, and encapsulation losses. However, other components of the difference arise from behavior that may be intrinsic to the materials. Prominent among these losses is the issue of spatial non-uniformity of the photoelectronic response.
7 P/ Y: S* z, I3 |5 b. x6 H! D/ N, U5 s- _
( X( Z- O1 L$ A- GThree materials classes are now beginning to reach large-scale manufacturing: amorphous silicon, CIGS, and CdTe.: L* r  X9 E" q8 f3 a; C9 c5 w! [! _0 Y
" X+ s) g$ Y7 G0 r; K
THIN-FILM CELLS AND MODULES
0 X0 e/ ~) q; B6 c8 ~4 @
& w3 V" d7 I* [( `  gAmorphous and Nanocrystalline Silicon8 [5 f5 v0 C' @; T
. R$ y9 B: R$ P. h2 a: H8 k
Hydrogenated amorphous silicon was the first of the three inorganic thin-film materials to be identified as a good PV candidate. An excellent review of the materials physics and the cell fabrication technology is given by X. Deng and E. A. Schiff.17) d% |9 k: R6 v& E. R. G
3 Q" s$ n0 Q2 e$ j, _& K( a5 Q
There are two very important advantages of the a-Si:H material. First, it can be deposited on a variety of non-crystalline substrates (e.g., glass and stainless steel) from vapor phase molecular precursors, and second, the process is easily scalable to large areas. An additional feature is that the rf glow discharge deposition (plasma-enhanced chemical vapor deposition, or PECVD) can be done at temperatures between room temperature and 150°C so that the process can be used on polymer substrates. The low-temperature deposition also facilitates the fabrication of double and triple-junction devices since the second and third junctions and interconnect layers can be deposited at temperatures that do not damage the materials quality of the earlier-deposited layers. High-efficiency cells have been fabricated in single-junction, dual-junction, and triple-junction configurations.
# j5 B1 V' @# z% p- ^  Q. c
, V2 b5 ^, l  t& f/ Z( NThe record triple-junction cell has 13% stabilized efficiency18 with a structure of a-Si:H / a-SiGe:H / a-SiGeH with the bottom cell having more germanium than the middle cell and thicknesses adjusted to produce current match. The typical individual component cells have the n-i-p structure in which the intrinsic layer has a strong electric field that quickly sweeps photogenerated electrons and holes, respectively to the n-type and p-type layers, where they become majority carriers. The main absorber layer is high-quality intrinsic a-Si:H that has much better minority carrier lifetimes than either the n- or p-doped layers. Single-junction a-Si:H cells require a relatively thick a-Si:H intrinsic layer to absorb most of the light and consequently these single-junction cells suffer up to 30% degradation in performance during the first 1.000 hours of light soak. The triple-junction, spectrum-splitting devices are fabricated with much thinner intrinsic layers so that there is a much stronger electric field in each i-layer that sweeps the carriers into the electron and hole collector layers (n-type a-Si:H and p-type a-Si:H).
2 Y! n( F3 O/ R. u
: q2 L( W3 t, X$ Q* ^. y. MMaterials Issues
7 @8 ^9 ~/ O4 C) i7 w
. `) H, z! I5 ~The highest efficiency structures also employ a back reflector to improve red light collection in thin structures. These 3-J structures can have degradation as low as 15% after 1,000 hours of light soak. Nevertheless, the degradation due to the Stabler-Wronski or dangling bond instability is still a major materials issue. This light-induced degradation generally increases with increasing germanium content and creates difficulties in obtaining good red and near-infrared response. Recently it has been demonstrated that performing the PECVD deposition under temperature and hydrogen dilution conditions that lie at the boundary of the amorphous-to-nanocrystalline transition will yield improved stability and red response.19 United Solar has announced an initial efficiency of 15.1 % and stabilized efficiency of 13.3% using nc-Si for the bottom cell.5 Typically the best solar cell material is nc-Si grains in an amorphous matrix. This "protocrystalline" or "micromorph" material is the focus of considerable recent research activity.20
4 l7 V" M, @% H" `
7 J  g2 o4 \* H0 O8 XTriple-junction a-Si devices require two interconnect layers that must allow the cell current to flow with little voltage drop. In the epitaxial, multijunction III-V solar cells, this interconnect is a tunnel junction formed with heavily doped layers. In the amorphous (and also the polycrystalline) TF materials, degenerately doped material is generally not achievable. However, these p/n a-Si: H interconnect junctions appear to operate quite well as recombination junctions with very little voltage drop across them and their formation does not seem to be a significant materials issue.3 l$ T, e& M% D' f9 ?& H0 `
/ \) I: g! Y+ V+ w6 d0 C0 V0 f# i+ v& s
One of the limitations of the amorphous silicon technology is the relatively slow deposition rate of the rf PECVD deposition. Rates are generally less than 1 nm/s. Recently some groups have been successful in the use of VHF frequencies (-70-100 MHz) which can provide higher deposition rates.21 Others have successfully used microwave frequencies for higher deposition rates. The higher frequencies require considerable redesign of the electrodes to be able to maintain uniform deposition over large areas, but some success has been reported.
  a$ q0 L$ E2 i& n: d+ a6 M
: \6 H" c! q4 SCadmium Telluride
- j, s4 L- w. w/ `( f8 j
4 \8 q9 W& r# a6 Q# r5 w  @' ~( cThe first of the two common polycrystalline materials for thin-film PV is the class of II-VI semiconductors. B.E. McCandless and J.R. Sites have provided an excellent review of the CdS/CdTe solar cell status.22 CdTe always works best with CdS as the heterojunction partner, like CIGS. The current record efficiency is 16.5% achieved on borosilicate glass23 rather than soda-lime glass, which is greatly preferred for manufacturing because of the much lower cost. Typical best performance on soda-lime glass is about 14.5%. It should be noted that the champion cell used a special TCO consisting of cadmium stannate (CdSn^sub 2^O^sub 4^) and zinc stannate (ZnSn^sub 2^O^sub 4^), which is not yet available as a commercially prepared TCO nor has it been demonstrated in a large-area production-line environment.
1 V0 c+ Q7 y. c" ?0 b. F% z9 T/ N% e2 D
CdTe is unusual in the wide range of deposition methods that have produced high efficiencies. These included close-spaced sublimation and vapor transport deposition with substrate temperatures of 550-650°C. and also physical vapor deposition and magnetron sputtering24 with deposition temperatures of 250-300°C. However, all methods involve a post-deposition treatment of the CdS/CdTe structure in vapors of CdCl^sub 2^ at ≥390°C. This is the highest temperature in the cell fabrication and makes the process barely compatible with flexible substrates such as polymides. The best cell performance on a flexible substrate -11% using a laboratory-prepared polyimide spun on glass and then the cell structure peeled off after deposition.25
% c6 a' s! \2 [/ U6 }, B5 ]7 C6 I( i+ m( P6 S6 ^2 X1 a  w
Materials Issues6 z* v# C8 w8 k  B; G( J& Q

/ w2 ~) K, i  {3 AOne of the complicating issues with CdTe-based cells is the need to fabricate the devices so they operate in a superstrate configuration (see Figure 3). Thus, in operation the light enters the cell through the glass "superstrate;" then sequentially through the TCO, the CdS, and finally the CdTe. This allows the fabrication of the back contact last which permits the use of chemical etches to prepare the CdTe surface for the back contact, which in the best cells is an HgTe:Cu-doped mixture of graphite paste or "dag." However, other groups use metal combinations such as a bilayer of copper (3 nm) and gold (10 om).# q# p5 T) D1 ^) d% }/ f

9 g- Y0 J& H" ]# v& q- tIn addition, the highest efficiencies have been obtained with laboratory high resistivity transparent (HRT) coatings such as Zn^sub 2^SnO^sub 4^ and undoped SnO^sub 2^ with resistivities of about 1,000 ohm-cm to 5.000 ohm-cm. These serve an important function in reducing the effects of shunting and weak diodes.6 L: J! f. f1 ^; `! x( ~) Z- F: Z
8 k. z% [! m( X+ C: S
That CdTe deposition works so well with such a variety of methods is attributable to the tendency of the material to grow stoichiometrically as CdTe. However, there is a complicating factor and that is the need for a post-deposition annealing treatment in vapors of chlorine (and usually CdCl^sub 2^). This post-deposition treatment is often called "activation" because with some deposition methods the cell performance without this activation step will show only 1-3% efficiency but after treatment will show 10-14% efficiency. This appears to be related to grain boundary passivation and to a change in the density of native defects.
8 v$ C* ]5 q) _/ I5 v
2 Y) v5 X% {3 Z& Z( _$ NStubborn materials challenges in the CdS/CdTe cell include the poorly understood role of the chlorine-activation process, and the difficulty of obtaining low-resistance back contacts to the high work function p-CdTe. which effectively forces one to do the back contact last and use the superstrate configuration even though it is generally preferred to fabricate the junction and anti-reflection window layers last.26 The band diagram in Figure 2 sketches what is believed to be a typical hole barrier at the CdTe/contact interface, almost independent of the type of back contact.27 This back-contact barrier will enhance (unwanted) electron collection and impede hole collection.28 There is poor control of the heterojunction interface partly because it is formed at the beginning of the absorber layer deposition and partly because there is usually a considerable S-Te interdiffusion across the CdS/CdTe interface during subsequent processing steps. The CdS heterojunction partner absorbs above 2.4 eV and these photogenerated carriers are never collected, probably due to low carrier lifetimes in CdS. This problem cal Is for innovative approaches, possibly including carbon nanotubes such as attempted for window materials in CIGS.29 In common with CIGS and the other polycrystalline TFs. grain boundary passivation iscritically important but is poorly understood.. d8 y2 I& F6 q3 b& B

9 n8 \2 J  [" l' q1 fCopper Indium Gallium Diselenide# ]9 v+ d# M+ F: ?
; J3 N) {) O6 M1 M
Of the polycrystalline TFs, easily the highest efficiency TF single-junction device is CIGS with the gallium to (In+Ga) ratio of about 30% with device efficiency of 19.5%. An excellent review is provided by W.N. Shafarman and L. Stolt.30 The CIGS structure is a substrate structure built usually on soda-lime glass with a coating of molybdenum., }# S, Q, R- |

# X" A* ]9 A$ A: y! nDepositing a molybdenum layer that does not delaminate during the remainder of the cell fabrication has been achieved by some laboratories with a double layer, first using conditions that optimize adhesion and then a layer that is in lateral compressive strain. The champion CIGS cell is deposited by coevaporation in a three-step process that involves careful control of the copper, indium, and gallium fluxes as well as temperature.31 Other groups have used sputtering or other deposition of the elemental metal layers, either sequentially or simultaneously, followed by a selenization process.
1 c% _6 {" e: _6 t$ i
+ x1 h+ e- ]0 ]Materials Issues
6 ^' b" s/ [; I" d! D5 n2 A7 z$ E+ b! j  D8 k* z
In spite of the widespread achievement of high-efficiency CIGS devices by many laboratories, commercial success with large-area modules has to date been elusive. Recently, however, Wurth Solar has been successful with a pilot plant and is completing a -20 MW production facility. There are many start-up companies commercializing CIGS using a wide range of deposition methods from evaporation to sputtering to atmospheric pressure spray techniques with nanoparticle CIGS precursors.
5 N" y& Z& [: k8 T" k/ I& s# z+ F; t
The structure of a typical CIGS solar cell is shown in Figure 3. What is not apparent from this image and sketch is that for the highest-efficiency cells (with -30% gallium), the absorber layer is deposited in a compositionally graded structure. In the highest-efficiency cells, the gallium content increases from 5-8% near the front junction to about 10-15% at the back contact, while the indium content correspondingly decreases toward the back. This composition gradient yields an increasing conduction band energy that produces a driving force to aid collection of electrons (minority carriers in the p-CIGS) by the n-CdS. These composition gradients act similarly to the back surface field in the silicon structure. Variation in the Cu/In ratio very near the junction can be used to control the doping concentration. Like CdTe, CIGS uses a heterojunction structure with CdS as the n-type partner. The 2.4 eV band gap is transparent to most of the photons absorbed in CIGS, but above 2.4 eV the photons absorbed in CdS are not collected due to poor materials quality control.
9 K( l' k+ c5 l+ d/ I, @" B
7 q% C# \8 t1 e" C9 j3 }, J. KCOMMON CHALLENGES IN POLYCRYSTALLINE THIN FILMS/ x8 @( a# s. o& a' j0 _3 y0 B
6 R' p1 C# f! r; c5 n( Q
The front heterojunction partner should have a good lattice match to the absorber, good carrier lifetimes (not true for CdS). and especially an appropriate conduction band lineup. A. Kanevce et al. have modeled the situation forCIGS and compared with current-voltage measurements.32 They point out that the preferred conduction band alignment has a spike in the conduction bands.33 This will occur if the electron affinity of the CdS is less than the CIGS. For spikes no greater than 0.48 V and cell operation at or above room temperature, thermionic emission of electrons over this spike will be strong enough to sustain the photogenerated reverse current. If the conduction band offset is less than 0.1 eV, there will be excessive interface recombination that will reduce the cell performance.: \/ e5 h, P% Y

0 C" _) X8 f; q! M+ DOne of the remarkable facts about both CIGS and CdTe solar cells is that the polycrystalline TFs produce better solar cells than their crystalline counterparts. Some of this is explained by the difficulty (or perhaps lack of effort) of obtaining thin (epitaxial) layers of CdTe or CIGS. Thus, several reports exist of cells constructed with CdS deposited on bulk CdTe or bulk CI(G)S, but this is not equivalent since cell operation with a bulk crystal requires the carriers to traverse typically mill i meters of absorber material. These experimental difficulties aside, there appear to be fundamental reasons that these particular polycrystalline TFs yield high-efficiency solar cells and some materials such as the III-Vs do not. In CdTe the grain boundary passivation seems to be related to the presence of chlorine and oxygen in the activation ambient and possible formation of oxides of copper at the grain boundaries.34 In the chalcopyrites, suggestions remain controversial that compositional or structural changes at the grain boundaries lead to enhanced carrier collection.35,36 Passivation due to charged grain boundaries is more commonly accepted. For example, sodium is well known to diffuse from the soda-lime glass substrates and improve cell performance. One model holds that Na+ accumulation at grain boundaries establishes electron potential wells that repel holes and therefore inhibit electronhole recombination.375 w2 R3 K6 l- x% K6 P/ V

0 k: n! c* ?' o" e# GGrain boundaries represent nanoscale non-uniformities peculiar to the polycrystalline TFs. But other non-uniformities are ubiquitous in thin-film devices whether amorphous, nanocrystalline, or polycrystalline. These non-uniformities can be as simple and dramatic as pinholes caused by dust in the deposition process or as subtle as a structural non-uniformity that introduces a diode region with a slightly lower turn-on voltage that becomes forward biased under illumination when the surrounding material is biased at the maximum power point. V.G. Karpov et al.38 have modeled nonuniformities by a random diode network model with a few weak diodes. However, the effects of these nonuniformites can be largely removed through the use of appropriate high resistivity transparent layers and postdeposition treatments.39
" v- b5 _4 ?/ A( f& w1 N3 S" w, \0 Y2 [7 @
To minimize resistive losses in a thin-film module, one uses series interconnection of many cell strips to achieve high voltage. Although metal styli can be used for the scribing of many of these rather soft or brittle PV materials, pulsed laser ablation scribing has many advantages, but it also requires a careful choice of laser wavelength and pulse duration matched to the material. A comprehensive study was presented in Reference 40.$ c* E& b4 g! s

% \( J7 _/ r9 y3 z- jIn most applications for PV, a premium is placed on reliability and long life. Twenty-five to 30-year lifetimes are required to achieve the desired levelized cost of energy. Space applications often do not require such long durations since the spacecraft itself often does not remain in orbit that long or the mission finishes in much shorter times. Many of these durability issues will not be intrinsic to the active PV materials but will depend on the encapsulation materials (glass cover slips, glass superstrates or laminations, ethyl vinyl acetate, etc.). It is beyond the scope of this report to examine encapsulation and lamination issues.9 }2 ]8 `9 Z- V9 w0 }/ J) K! ~
' X7 g! S- t6 k  C* D9 y
Intrinsic stability of crystalline silicon and epitaxial III-V semiconductor devices is not considered to be a problem; however, there is generally a small decay in performance even of these crystalline materials at rates generally below the 1% (relative) decrease per year. Amorphous silicon is well known for the Stabler-Wronski instability in which a standard, single-junction a-Si:H cell typically will drop in performance by 20% to 30% in the first 1,000 hours of one-sun illumination, and triple junction cells, with their thinner i-layers, about 15%. Beyond 1,000 hours, however, a-Si: H cells are similar to c-Si in stability.
& w6 p6 i" P, Z
8 o, }. E! U2 S2 Y+ X8 @1 \Thin-film CIGS material appears to have cell stability that is as good as silicon. In fact, CIGS seems to have some capability of self-healing of radiation damage. Because it is a direct bandgap material, this provides some advantage over silicon. It is not yet clear whether it is superior to the III-V materials regarding radiation damage but since the initial efficiency is lower and CIGS is not yet available in a multijunction configuration, the multijunction III-Vs remain the choice for most space applications.
# \. U! S- @& p9 p+ Y' t1 s- ~3 @- D- ?  h
CdTe cells can show substantial degradation under sunlight, particularly when stored at open-circuit conditions. However, it is now understood that this is process-dependent and not an inevitable characteristic of the CdS/CdTe cell. Some of this instability appears to be due to the movement of copper from the back contact toward the CdS/CdTe junction, and other instability appears related to oxygen and water vapor exposure that can be corrected with good encapsulation.
* ]. a! }" P) o# B9 H5 b$ o: c# m8 f, R8 o: n
CONCLUSIONS
( _5 l6 u7 |/ U& B' g1 k# V  A' H7 J0 K: t$ c
The commercial success of thin-film solar cells has come very slowly, in part because the technology of wafer-based silicon PV continues to improve. Also, however, it required the focused efforts of many scientists and engineers to address the formidable materials challenges that arise in moving from modules assembled from discrete wafers to monolithically integrated large-area, thin-film modules. The thin-film modules require rapid deposition of multiple layers of metals, insulators, and semiconductors with demanding optical and electronic properties. These micrometer and submicrometer thick layers then need to have proven stability over 20 years or more in extreme conditions.6 W, g8 z- y7 X3 ?4 U9 o0 ?4 L
, Q+ \/ G. e2 n" t# n
Hopefully this overview has illustrated that much improvement in these thin-film devices can be realized if we can solve the many challenging materials issues that remain. We now know that thin-film PV is here to stay and it is possible, even likely, that TF modules will eventually exceed silicon PV world production as they started to do in the U.S. market in 2006. Rest assured there are plenty of materials challenges that must still be resolved in order for TFs to dominate and to provide cost-competitive grid electricity for the world. But it is clear that such a transformation is on the horizon.
5 ^# {, V# g3 w5 i% A$ s4 k" [# E
# k1 E- k! C  o- _% B# N2 ?[Sidebar] 7 u( e: n' T; n) K+ C* c( k
OVERVIEW OF PV MATERIALS AND EFFICIENCIES
: T& I! L* A* {- C$ o& G6 x  I  RPhotovoltaic (PV) materials are often classified into first, second, and third generations. First generation typically refers to cells and modules made from silicon wafers that usually are sliced with diamond wire saws from crystalline silicon boules (c-Si) pulled slowly from the melt, or from multicrystalline material solidified in a crucible (mc-Si). second-generation devices are typically thin-film inorganic coatings on inexpensive large-area substrates such as glass, stainless steel, or polymer. second-generation materials include primarily amorphous silicon (a-Si), cadmium telluride, and copper indium diselenide (CIS, CuInSe^sub 2^), or the CIS alloy with gallium (CIGS . Cu(InGa)Sej. with aluminum (CIAS, Cu(InAl)Se^sub 2^. or with sulfur (CIGSS, Cu(InGa)(SeS)^sub 2^). Second generation also usually includes film silicon, which is similar to a-Si but is recrystallized to form a polycrystalline layer.6 (In photovoltaics, polycrystalline (pc) generally refers to films with grain sizes of a couple of micrometers or less, while multicrystalline (mc) generally is taken to mean grains of millimeter to centimeter scale.) Generation-three devices include the Graetzel cell using inorganic nanoparlicles (TiO^sub 2^) and a liquid or gel dye sensitizer,7 and devices made from photovoltaic polymers/ (Note that the highest efficiency cells, made from triple-junction III-V materials, are crystalline and epitaxial and often placed into the first generation category but have efficiencies exceeding the Shockley-Queisser limit and therefore also are often described as third generation.)
8 N- z$ H( G3 R6 R/ X6 kThe theoretical maximum efficiency of a single-junction solar cell for terrestrial sunlight (AM 1.5) is about 28% and it occurs for materials with a bandgap of 1.5 eV. Large-area modules should theoretically be only a couple of percent lower if there is negligible dead area. (A "cell" is typically 1 cm^sup 2^ or less whereas a module has about 0.1-1 m^sup 2^ aperture area.) second-generation or thin-film PV modules typically have efficiencies below 10%. This compares with silicon wafer champion efficiency of 24% and typical production module efficiencies of 15-18%. Both a-Si TF and III-V epitaxial cells are regularly produced in double or triple junction, spectrum-splitting devices. (Triple junction cells theoretically can attain 45% efficiency.) Triple-junction III-V cells hold the efficiency record of 34% at one-sun. A major project by the U.S. Defense Advanced Research Projects Agency is underway to obtain 50% efficiency devices using clever spectrumsplitting optics and materials combinations. The team has recently announced a 42.8% efficiency device.9
6 S" m% E$ `* k  ^0 D: zSolar-cell efficiency is just the maximum electrical power divided by the total solar power incident. This power efficiency should be distinguished from the quantum efficiency at a specified wavelength which will be 90-100% over a wide spectrum in a good solar cell. The current state of the technology for small cells and the comparison with achievable limits are shown in Figure A.10 Note that the photon energy for maximum efficiency for a single junction device lies at -1.4 eV for one-sun illumination but the variation is slow between 1.1 eV and 1.7 eV so that all the materials including silicon. CuInSe^sub 2^. GaAs, and CdTe are very attractive PV materials. (The bumps and shoulders in the efficiency curves are directly a consequence of the CO2 and H2O absorption lines in the terrestrial spectrum.11) The efficiencies of the single-crystal materials, silicon, GaAs. and InP, lie very close to the achievable maximums. Most of the polycrystalline TF materials have considerable room for improvement, although the quaternary alloy, eopper-indium-gallium diselenide (CTGSe) with about 30% gallium and η[approximate]20% is the closest to predicted limits at AM1.5 for thin-film cells. 2 H  `. y& Q% n1 t/ f5 X
, d8 V- B3 l. l4 O! N
[Reference] 2 B- ?' ^7 h/ K+ @9 N" \; L7 D
References
6 i/ a3 F! d# G( S0 O8 N8 b7 g) ~1. PV News (Cambridge, MA:The Prometheus Institute for Sustainable Development, March 2006). 6 D1 V7 {1 l/ r  B
2. W. Shockley and H.J. Queisser, J. Appl. Phys, 32 (1961), p. 510. ! T2 k) K, z9 p9 y3 _" ^3 B' w
3. R.R. King, World Conference on Photovoltaic Energy Conversion-2006 (Piscataway, NJ: IEEE, 2006), p. 1757. , V2 N, p, Z2 X+ w4 @& B' ^- X
4. P.V. Meyers, in Ref. 3, p. 2024. 4 g" j0 f( [3 E/ e2 @
5. B. Yan et al., in Ref. 3, p. 1477. 3 S# l$ B- h. H3 W2 c0 ^
6. P.W. Basore, in Ref. 3, p. 2089. ; T! r. q8 F8 }' y& k
7. M. Graetzel, Nature, 414 (2001), p. 338. 1 [7 |7 V4 p4 [% y5 }
8. S. Forest, Proc. 2003 SPIE Meeting (Bellingham, WA: SPIE, 2003). , K! v% C3 s- }
9. Elizabeth Corcoran, "A Trick of the Light" (3 September 2007), http://members.forbes.com/global/2007/0903/088, html. 0 S* \% L0 r; Z: G% A9 S1 ~& H
10. L. Kazmerski, J. Electron Spectroscopy, 150 (2006), pp. 105-135. . W; g, A9 S' B- G! x
11. "Reference Solar Spectral Irradiance Air Mass 1.5," httpj/rredc.nrel.gov/solar/spectra/am1.5/
: z" U; X1 N8 b, ~, M4 h9 V$ @1 e12. Alvin Compaan, APS News Online (April 2005), www.aps.org/apsnews/0405/040514.cfm. 9 i* A8 {' s! t- \) |. ~3 C
13. B. Agostinelli et al., in Ref. 3, p. 1004. * ^. L, V8 `9 [6 p
14. M. McCann et al., in Ref. 3, p. 894. " j% @/ Z6 D) q% a, T
15. J. Wohlgemuth et al., in Ref. 3, p. 1099.
- o& O( G7 M/ N& I16. M.A. Green, Progress in Photovoltaics, 14 (2006), pp. 45-51. 1 a: \% Z/ j/ E+ T* g) g) \
17. X. Deng and E.A. Schiff, Handbook of Photovoltaic Science and Engineering, ed. by A. Luque and S. Hegedus (New York: John Wiley & Sons Ltd., 2003), pp. 505-566.
% B/ l( ]( S6 T! V0 D18. J. Yang, A. Banerjee, and S. Guha, Appl. Phys. Lett., 70 (1997), p. 2975. ' E  e3 p; T! q( q
19. X. Deng et al., in Ref. 3. p. 1461.
- n3 Q( p2 r( Z6 u5 c20. K. Yamamoto et at, Proc. 28th IEEEPV Specialists Conference-2000 (Piscataway, NJ: IEEE, 2000), p. 1428. ; e- s& w' T6 [  z7 ~  a' T
21. A.H.M. Smets, T Matsui, and M. Kondo, in Ref. 3, p. 1592.
4 {8 U2 Z& ], k3 T( ?8 x22. B.E. McCandless and J.R. Sites, in Ref. 17, pp. 617-662.
2 A& ?. i) `# ?! r3 [7 v' I23. X. Wu et al., DOE Solar Energy Technologies Program, FY2005 Annual Report (Program Review Meeting, Denver, CO, November 7-10, 2005). * Z$ d0 B1 B6 k) _) x4 g( Y
24. A. Gupta and A.D. Compaan, Appl. Phys. Ltrs., 84 (2004), p. 684. , v0 h- E6 J7 \, O4 y) U
25. A.N. Tiwari (Presentation at the Mat. Res. Soc. Proc. Symposium Y, April 9-13, 2007). / X0 I. N- l+ s9 _" W
26. R. Noufi and D. Zweibel, Conference Paper NREL/CP-520-39894 (2006), www.nrel.gov/docs/fy06osti/39894.pdf.
; \) Y, ~9 x1 S1 Q  ]4 c27. A.L. Fahrenbruch, in Ref. 3, p. 376. - y$ M# n' ?' r: J4 O4 t
28. J. Pan, M. Gloeckler, and J.R. Sites, J. Appl. Phys., 100(2006), p. 124505.
. z% |7 q* J- F29. M. Contreras et al., in Ref. 3, p. 428. % g) }/ C9 M: ]
30. W.N. Shafarman and L. Stolt, in Ref. 17, pp. 567-616.
0 Z* W( c: c; f6 f% E31. M.A. Contreras et al., Prog. Photovoltaics, 13 (2005), p. 209.
# O3 B0 R3 ~: W32. A. Kanevce et al., Mater. Res. Soc. Symp. Proc. Vol. 865. (Warrendale, PA: Materials Research Society, 2005). F5.32.1. 0 W( [. n! W( l+ J! ?$ W) C
33. L. Wienhardt et al., in Ref. 3, p. 412. / g1 M5 d1 j# u! r, U4 I
34. X. Liu, A.D. Compaan, and J.Terry, Proc. 31st IEEE Photovoltaic Specialists Conference-2005(Piscataway, NJ: IEEE, 2005), pp. 267-270. & {& ]1 Y) j. O
35. C. Persson and A. Zunger, Phys. Rev Ltrs., 91 (2003). p. 266401.
: X" F: m, d. \/ f36. Y. Yan, R. Noufi, an M.M Al-Jassim, Phys. Rev. Ltrs., 96 (2006), p. 205501. * {/ b3 h$ @: k: j: }
37. W.K. Metzger and M. Gloeckler, J. Appl. Phys., 98 (2005), p. 063701. / V  ~: W: o7 p; o( |- h/ P) [1 r
38. V.G. Karpov, A.D. Compaan, and Diana Shvydka, Appl. Phys. Lett.. 80 (2002), p. 4256. ( U5 o6 D3 j% f5 P3 b6 M4 }" U
39. Y. Roussillon et al., Appl. Phys. Ltrs., 84 (2004), p. 616. 2 D* `* o7 Q" w- t5 Q9 Z
40. A.D. Compaan, I. Matulionis, and S. Nakade, Optics and Lasers in Engineering, 34 (2000), pp. 15-45.
您需要登录后才可以回帖 登录 | 注册

本版积分规则

本站邀请注册说明!

小黑屋|手机版|Archiver|光学薄膜信息网  

GMT+8, 2024-6-18 19:35 , Processed in 0.040722 second(s), 19 queries .

Powered by Discuz! X3.4 Licensed

Copyright © 2001-2021, Tencent Cloud.

快速回复 返回顶部 返回列表