Cadmium telluride solar cell

Articleissues
advert = October 2008
COI = October 2008
wikify = October 2008A cadmium telluride solar cell is a solar cell based on a cadmium telluride (CdTe) thin film, a semiconductor layer to absorb and convert sunlight into electricity [http://www.nrel.gov/pv/thin_film/pn_techbased_cadmium_telluride.html ] .

Background

Since inception, the dominant solar cell technology in the marketplace has been based on wafers of crystalline silicon. During the same period, the idea of developing alternative, lower cost PV technologies led to the consideration of so-called thin films and concentrators. Thin films are based on using thinner semiconductor layers to absorb and convert sunlight; concentrators, on the idea of replacing expensive semiconductors with lenses or mirrors. Both reduce cost, in theory, by reducing the use of semiconductor material. However, both faced critical challenges.

The first thin film technology to be extensively developed and manufactured was amorphous silicon. However, this technology suffers from low efficiencies and slow deposition rates (leading to high capital costs) and has not become a market leader. Instead, the PV market has grown to almost 4 gigawatts (GW) with wafer-based crystalline silicon with almost 90% of sales. [Various estimates of world module production in 2007 are about 4 GW (e.g., http://www.worldwatch.org/node/5449#notes).] Installation trails production by a slight time lag, and the same source estimates about 3 GW were installed in 2007.

During this period, two other thin films continued in development (cadmium telluride, and copper indium diselenide or CIS-alloys). The latter is beginning to be produced in start-up volumes of 1-30 MW per year by individual companies and remains an unproven, but promising market competitor due to very high, small-area cell efficiencies approaching 20%. [19.9% CIGS cell made at NREL: http://www.nrel.gov/news/press/2008/574.html]

Issues

Cell efficiency

Best cell efficiency has plateaued at 16.5% since 2001. [Since NREL’s X. Wu produced a 16.5% cell using advanced front TCO material that allowed more light while being more conductive than prior cells.] The opportunity to increase current has been almost fully exploited, but more difficult challenges associated with junction quality, CdTe properties, and contacting have not been as successful. However, until recently the number of active scientists in CdTe PV was small. [Most US R&D activities, which indeed means most world activities in CdTe, were organized and partially funded through NREL’s Department of Energy funded Thin Film PV Partnership (http://www.nrel.gov/pv/thin_film/ and http://www.nrel.gov/pv/thin_film/about.html), which included a CdTe national R&D team of about 50 members. The Partnership was ended with the advent of the recent DOE Solar America Initiative which de-emphasized technology-specific research.] Improved doping of CdTe and increased understanding of key processing steps (e.g., cadmium chloride recrystallization and contacting) are key to progress. Since CdTe has the optimal band gap for single-junction devices, it may be expected that efficiencies close to exceeding 20% (such as already shown in CIS alloys) should be achievable in practical CdTe cells. Modules of 15% would then be possible.

Process optimization

Process optimization allows higher throughput at lower cost. Typical improvements are larger substrates (since capital costs scale sublinearly, and installation costs can be reduced), thinner layers (to save material, electricity, and throughput time), and better material utilization (to save material and cleaning costs). Making components rather than buying them is also a traditional way for large manufacturers to shave costs. Today’s CdTe module costs are about $110/m² (normalized to a square meter) [This number is calculated by multiplying efficiency (10.7%) by 1000 to get output watts per square meter (107 W/m²), and then multiplying this number of watts by the stated cost of $1.04/W to get $111/m² , or $80 per module.] . It is not expected that this can be cut in half, but costs in the $75/m² seem achievable.

Thus a practical, long-term (10-20 year) goal for CdTe modules resulting from combining cost and efficiency goals would be $75/150 W, or about $0.5/W. [This number is calculated by dividing the cost per unit (e.g, $75/m²) by output per the same unit (15% produces 150 W per square meter): $75/150 W = $0.5/W.] With commodity-like margins and combined with balance-of-system (BOS) costs, installed systems near $1.5/W seem achievable. With Southern California sunlight, this would be in the 6 to 8 ¢/kWh range (e.g., based on economic and other assumptions used in algorithms such as in the DOE/NREL Solar Advisory Model). [Like any solar price model, the Solar Advisory Model (http://www.nrel.gov/csp/troughnet/models_tools.html#solaradvisormodel) is quite sensitive to assumptions. Different sunlight, tax rates, interest rates, discount rates, loan durations, temperature coefficients, annual degradation rates, initial de-rating versus standard conditions, inverter efficiencies and O&M, and others can each have as much as a 10% impact on ¢/kWh.]

Tellurium supply

Perhaps the most subtle and least understood problem with CdTe PV is the supply of tellurium. Tellurium (Te) is an element not currently used for many applications. Only a small amount, estimated to be about 800 metric tons (MT) per year, is available. Most of it comes as a by-product of copper, with smaller byproduct amounts from lead and gold. One gigawatt (GW) of CdTe PV modules would require about 90 MT (at current efficiencies and thicknesses), [There is about 3 gm/cc Te in CdTe. One micron over a square meter area is 1 cc, or 3 gm of Te. Typical CdTe layer thicknesses are about 3 microns, so there are 9 g/m². At 10% efficiency, one GW is 107 m², or 90 million g, which is 90 MT. However, it takes only about 0.5 micron thickness to absorb 90% of the sunlight in a CdTe layer. If cells were this thick, six times less Te would be needed per GW, or 15 MT/GW. Note that whatever Te is unused in processing will be recycled and used.] so this seems like a limiting factor. However, because tellurium has had so few uses, it has not been the focus of geologic exploration. In the last decade, new supplies of tellurium-rich ores have been located, e.g., in Xinju, China. [Publications of the Sichuan Xinju Mineral Resource Development Co., China] Since CdTe is now regarded as an important technology in terms of PV’s future impact on global energy and environment, the issue of tellurium availability is significant. Recently, researchers have added an unusual twist – astrophysicists identify tellurium as the most abundant element in the universe with an atomic number over 40. [From Cohen (1984) and Hein et al (2004), where Hein writes, “It has been suggested that Te is unique in the universe in that its cosmic abundance is as great or greater than any of other element with an atomic number higher than 40 (http://www.webelements.com), yet it is one of the least abundant elements in the Earth’s crust and in ocean water.”] This surpasses, e.g., heavier materials like tin, bismuth, and lead, which are common. Researchers have shown that well-known undersea ridges (which are now being evaluated for their economic recoverability) are rich in tellurium and by themselves could supply more tellurium than we could ever use for all of our global energy. [See Hein (2004) and Hein et al. (2003) for a complete discussion. The ridges occur at 400-4000 m depths “where currents have kept the rocks swept clean of sediments for millions of years. Crusts…forming pavements up to 250 mm thick.”] It's not yet known whether this undersea tellurium is recoverable, nor whether there is much more tellurium elsewhere that can be recovered.

Other issues

Another issue frequently mentioned, but already resolved, is the use and recycling of the heavy metal cadmium. First Solar has a self-imposed recycling regimen that provides a deposited amount (under a nickel a watt) that covers the costs of transport and recycling of the module at the end of its useful life. [First Solar describes its recycling program: http://www.firstsolar.com/recycle_modules.php; and NREL’s summary of thin film environmental issues: http://www.nrel.gov/pv/thin_film/pn_techbased_esh.html] Recycling has been fully demonstrated on scrap modules. In a validating test, Vasilis Fthenakis of the Brookhaven National Laboratory showed that the glass plates surrounding CdTe material sandwiched between them (as they are in all commercial modules) seal during a fire and do not allow any cadmium release. [Fthenakis et al. 2004] All other uses and exposures related to cadmium are minor and similar in kind and magnitude to exposures from other materials in the broader PV value chain, e.g., to toxic gases, lead solder, or solvents (most of which are unused in CdTe manufacturing). [e.g., Fthenakis and Kim 2006 for environmental issues; and Rose 1999 for manufacturing approaches]

A subtle issue with CdTe and with all thin films in relation to higher efficiency PV module technologies is the potential impact of commodity inflation. Lower efficiency modules incur a greater balance of system commodity cost per unit output. Thus such inflation can have a larger percentage impact on system cost. This is another reason that continued efficiency improvements are important.

olar tracking

Almost all thin film systems to-date have been non-solar tracking, because the output of modules has been too low to offset tracker capital and operating costs. But relatively inexpensive single-axis tracking systems can add 25% output per installed watt. [See the Solar Advisory Model (http://www.nrel.gov/csp/troughnet/models_tools.html#solaradvisormodel) for algorithms that switch back and forth between non-tracking and tracking to establish this percentage. It is also climate dependent.] Tracking also produces a smoother output plateau around midday, allowing afternoon peaks to be met.

History

Research in CdTe dates back to the 1950s, [Early publications by Goldstein, Vodakov, Cusano, R. Bube and D. Bonnet; Patents including R. Colman, July 28, 1964, US 3142586] because it was quickly identified as having a band gap (about 1.5 eV) almost perfectly matched to the distribution of photons in the solar spectrum in terms of optimal conversion to electricity. A simple heterojunction design evolved in which p-type CdTe was matched with n-type cadmium sulfide (CdS). The cell was completed by adding top and bottom contacts. Early leaders in CdS/CdTe cell efficiencies were GE in the 1960s, [D. A. Cusano led a group at GE in the 1960s.] and then Kodak, Monosolar, Matsushita, and Ametek.

By 1981, Kodak used close spaced sublimation (CSS) and made the first 10% cells and first multi-cell devices (12 cells, 8% efficiency, 30 cm²). [Tyan especially published both patents and papers of significance at Kodak and helped to establish CdTe as an important thin film option. ] Monosolar [B. Basol patented numerous aspects of electrodeposition and CdTe contacting for Monosolar. Monosolar was subsequently bought by SOHIO, which was then absorbed by British Petroleum. Electrodeposition continued at BP Solar until about 2002 when it was canceled along with all thin film work at BP.] and Ametek [ Peter Meyers, originally of Ametek, provides a thread stretching from Ametek through Solar Cells Inc. to First Solar. He is on Ametek patents US Patent 4,260,427, 1981; US Patent 4,710,589, 1987; and SCI/First Solar patents; see note 15] used electrodeposition, a popular early method. Matsushita started with screen printing but shifted in the 1990s to CSS. Cells of about 10% sunlight-to-electricity efficiency were being made by the early 1980s at Kodak, Matsushita, Monosolar, and Ametek. K. Zweibel has published a number of useful review articles on thin films and especially CdTe. This is from Zweibel (1995). A more up to date one is Noufi and Zweibel (2006).]

An important step forward occurred when cells were being scaled-up in size to make larger area products called modules. These products require higher currents than small cells and it was found that an additional layer, called a transparent conductive oxide (TCO), could facilitate the movement of current across the top of the cell (instead of a metal grid). One such TCO, tin oxide, was already being applied to glass for other uses (thermally reflective windows). Made more conductive for PV, tin oxide became and remains the norm in CdTe PV modules.

Professor Ting L. Chu of Southern Methodist University and subsequently of University of South Florida, Tampa, made significant contributions to moving the efficiency of CdTe cells to above 15% in 1992, a critical level of success in terms of potential commercial competitiveness. This was done when he added an intervening or buffer layer to the TCO/CdS/CdTe stack and then thinned the CdS to allow more light through. Chu used resistive tin oxide as the buffer layer and then thinned the CdS from several microns to under half a micron in thickness. Thick CdS, as it was used in prior devices, blocked about 5 mA/cm² of light, or about 20% of the light usable by a CdTe device. By removing this loss while maintaining the other properties of the device, Chu reached 15% efficiency in 1991, the first thin film to do so, as verified at the National Renewable Energy Lab (NREL). Chu used CSS for depositing the CdTe. For his achievements in taking CdTe from its status as “also-ran” to a primary candidate for commercialization, some think of Ting L. Chu as the key technologist in the history of CdTe development.

In the early 1990s, another set of entrants were active in CdTe commercial development, but with mixed results. A short-lived company, Golden Photon replaced Photon Energy, when it was bought by the Coors Company in 1992. Golden Photon, led by Scot Albright and John Jordan, actually held the record for a short period for the best CdTe module measured at NREL at 7.7% using a spray deposition technique. Meanwhile Matsushita, BP Solar, and Solar Cells Inc. were active. Matsushita claimed an 11% module efficiency using CSS and then dropped out of the technology, perhaps due to internal corporate pressures over cadmium. A similar efficiency and fate eventually occurred at BP Solar. BP used electrodeposition inherited from Monosolar by a circuitous route when it purchased SOHIO. SOHIO had previously bought Monosolar. BP Solar however never made a complete commitment to their CdTe technology despite its achievements and dropped it in the early 2000s. Another ineffective corporate evolution occurred at a European entrant, [http://www.antec-solar.de Antec] . Founded by CdTe pioneer Dieter Bonnet (who made cells in the 1960s), Antec was able to make about 7%-efficient modules, but went bankrupt when it started producing commercially during a short, sharp downturn in the market in 2002. Purchased from bankruptcy, it never regained the technical traction needed to make further progress. However, as of 2008 Antec does make and sell CdTe PV modules, the only other company besides First Solar to do so.

CI and First Solar

The single major success to emerge from the turmoil of the 1990s was Solar Cells Incorporated (SCI). Founded in 1990 as an outgrowth of a prior company, Glasstech Solar (founded 1984), led by inventor/entrepreneur Harold McMaster, [Harold McMaster envisioned the opportunity for low cost thin films made on a large scale. After trying amorphous silicon, he shifted to CdTe at the urging of Jim Nolan and founded Solar Cells inc., the precursor of First Solar; http://www.toledoblade.com/apps/pbcs.dll/article?AID=/20080429/COLUMNIST02/804290323] it switched from amorphous silicon to CdTe as a better solution to the high cost crystaline silicon PV. McMaster championed CdTe for its high-rate, high throughput processing. Technical leadership came from a team that included Jim Nolan, Rick Powell, and Peter Meyers, with consulting help from Ting Chu and Al Compaan (U. Toledo). SCI started with an adaptation of the CSS method then shifted to a vapor transport approach, inspired by Powell. [SCI CSS patent: Foote et al. Process for making photovoltaic devices and resultant product, United States Patent 5248349; and their vapor transport patent, featuring the movement of vaporized cadmium and tellurium atoms through a closed, silicon carbide distributor: Apparatus and method for depositing a semiconductor material, United States Patent 6037241. Both are now owned by First Solar.] In February 1999, McMaster sold the company to True North Partners, an investment arm of the Walton family, owners of Wal-Mart. [D. H. Rose, Oct. 1999, p. Viii (preface)] John T. Walton joined the Board of the new company, and Mike Ahearn of True North became the CEO of the newly minted First Solar.

Like all other PV companies pioneering a new technology, First Solar suffered setbacks. Initial module efficiencies were modest, about 7%. But progress was steady, and commercial product became available in 2002. But production did not reach 25 MW until 2005fact|date=October 2008. The company built an additional line in Perrysburg, OH, then four lines in Germany, supported by the then substantial German production incentives (about 50% of capital costs)fact|date=October 2008. In 2006 First Solar reached 75 MW of annual productionfact|date=October 2008 and announced a further 16 lines in Malaysia, the more recently announced lines have been operational ahead of schedulefact|date=October 2008. As of 2008, First Solar is producing at nearly half a gigawatt annual ratefact|date=October 2008 and is among the largest PV module manufacturers in the worldfact|date=October 2008.

ee also

* Cadmium telluride
* High efficiency solar cells
* Low cost solar cell
* Renewable energy
* Solar energy
* Solar cell
* Solar panel

Notes

References

*B. Basol, E. Tseng, R.L. Rod, 1981, Thin film heterojunction photovoltaic cells and methods of making the same, Monosolar, US patent 4388483.
*R. H. Bube, Dec. 1955, Photoconductivity of the Sulfide, Selenide, and Telluride of Zinc or Cadmium, RCA Laboratories, Princeton, N.J., Proceedings of the IREVolume: 43, Issue: 12, page(s) 1836-1850, ISSN: 0096-8390, Digital Object Identifier: 10.1109/JRPROC.1955.278046
*B. L. Cohen, 1984, “Anamolous behavior of tellurium abundances, Geochim. Cosmochim. Acta 38, 279-300
*D. A. Cusano, 1963, “CdTe Solar Cells and PV Heterojunctions in II-VI Compounds,” Solid State Electronics, 6, 217
*V. Fthenakis, H. C. Kim, 2006, “CdTe Photovoltaics: Life Cycle Environmental Profile and Comparisons,” European Material Research Society Meeting, Symposium O, Nice, France, May 29-June 2, 2006
*V. Fthenakis, H. C. Kim, 2007, “CdTe photovoltaics: Life cycle environmental profile and comparisons,” Thin Solid Films, Volume 515, Issue 15, 31 May 2007, Pages 5961-5963, doi:10.1016/j.tsf.2006.12.138; article at http://www.clca.columbia.edu/papers/CdTe_Photovoltaics_Life_Cycle_Environmental_Profile.pdf
*V. Fthenakis, M. Fuhrmann, J. Heiser, W. Wang, 2004, “Experimental Investigation of Emissions and Redistribution of Elements in CdTe PV Modules during Fires,” 19th European PV Solar Energy Conference, Paris, France, June 7-11, 2004, 5BV.1.32, http://www.nrel.gov/pv/thin_film/docs/fthenakis_2004_cdte_fires_paris_preprint.pdf
*B. Goldstein, Jan. 1958, “Properties of PV Films of CdTe,” Phys. Rev., v. 109, p. 601
*J. Hein, April 2004, “Cobalt-Rich Ferromanganese Crusts: Global Distribution, Composition, Origin and Research Activities,” Chapter 5 from Workshop on Minerals other than Polymetallic Nodules of the International Seabed Area, prepared by the Office of Resource and Environmental Monitoring, International Seabed Authority, Kingston, Jamaica, ISBN 976-610-647-9
*J. Hein, A. Koschinsky, and A. Halliday, 2003, “Global Occurrence of tellurium-rich ferromanganese crusts and a model for enrichment of tellurium,” Geochimica et Cosmochimica Acta, Vol. 67, No. 6, 1117-1127, doi:10.1016/S0016-7037(00)01279-6
*A. H. Hill, “Progress in Photovoltaic Energy Conversion,” NASA, Washington, DC, http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19660006117_1966006117.pdf
*D. A. Jenny and R. H. Bube, 1954, “Semiconducting CdTe,” Phys. Rev. 96, 1190, DOI: 10.1103/PhysRev.96.1190.
*R. Noufi and K. Zweibel, 2006, “High-Efficiency CdTe and CIGS Thin-Film Solar Cells: Highlights and Challenges,” National Renewable Energy Laboratory, Golden, CO 80401, USA, http://www.nrel.gov/pv/thin_film/docs/wc4papernoufi__.doc
*D. H. Rose et al., Oct. 1999, “Technology Support of High-Throughput Processing of Thin Film CdTe Panels,” NREL SR-520-27149, http://www.nrel.gov/docs/fy00osti/27149.pdf
*Richard Stevenson, August 2008, “First Solar: Quest for the $1 Watt,” IEEE Spectrum Online, http://spectrum.ieee.org/aug08/6464
*Y. S. Tyan, 1978, Polycrystalline thin film CdS/CdTe photovoltaic cell, Kodak, patent US4207119 (EP0006025)
*Y. S. Tyan and E. A. Perez-Albuerne, 1982, Integrated array of photovoltaic cells having minimized shorting losses, Kodak, US patent 4315096.
*Y. A. Vodakov, G. A. Lomakina, G. P. Naumov, Y. P. Maslakovets, 1960, “A P-N Junction photocell made of CdTe,” Soviet Physics, Solid State, v. 2, n. 1, p. 1
*X. Wu et al., Oct. 2001, “High Efficiency CTO/ZTO/CdS/CdTe Polycrystalline Thin Film Solar Cells,” NREL/CP-520-31025, http://www.nrel.gov/docs/fy02osti/31025.pdf
*K. Zweibel, 1995, “Thin Films Past Present and Future,” NREL/TP-413-7486, 18 p., http://www.osti.gov/bridge/servlets/purl/61140-cYAVRN/webviewable/61140.PDF
*K. Zweibel, J. Mason, V. Fthenakis, January 2008, “A Solar Grand Plan,” Scientific American.