IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 1, JANUARY 2013 429

Life Cycle Water Usage in CdTe PhotovoltaicsParikhit Sinha, Amy Meader, and Mariska de Wild-Scholten

Abstract—Life cycle water withdrawal for cadmium tel-luride photovoltaics (CdTe PV) ranges from approximately 382–425 L/MWh, with only ∼12% from direct on-site usage. The re-mainder is related to indirect water withdrawal from the use of gridelectricity and raw materials throughout the product life cycle. Ap-proximately half of life cycle water withdrawal is associated withmodule manufacturing, one-third from balance of systems (BOS),and the remainder from takeback and recycling. Primary contrib-utors to life cycle water withdrawal are the use of grid electricity,glass, and on-site water during manufacturing; steel, copper, in-verters, and on-site water in the BOS; and electricity, chemicaluse, and transport during takeback and recycling. During man-ufacturing, water consumption is approximately one quarter ofwithdrawal and is due to cooling tower water evaporation and siteirrigation. When deployed in the U.S. Southwest, a CdTe PV arraycan provide net displacement of life cycle water withdrawal of over1700–5600 L/MWh relative to grid electricity.

Index Terms—Environmental management, thin-film photo-voltaic (PV) modules and manufacturing, water resources.

I. INTRODUCTION

WHILE energy security and climate change have been im-portant market drivers for renewable energy adoption,

water security provides an additional driver. For example, inthe U.S., thermoelectric power plants have recently accountedfor over 40% of total freshwater withdrawals, even more thanfor agriculture (see Table I). This energy–water nexus associ-ated with traditional energy sources can be a potential concern,particularly in water-stressed regions.

Because solar photovoltaics (PV) use little to no water dur-ing operation (0–19 L/MWh) [2], [3], they provide a potentialpath forward to addressing the energy–water nexus. However,water is used on-site during PV module manufacturing, projectconstruction, and end-of-life recycling. There is also indirectwater usage from the use of grid electricity and raw materialsthroughout the product life cycle [4].

The purpose of this evaluation is to examine water usage fora specific PV technology, i.e., CdTe PV, by life cycle stage,and in relation to other electricity generation sources. For themodule manufacturing life cycle stage, a water balance is pre-sented to identify how water is consumed before it is treated

Manuscript received May 10, 2012; revised July 9, 2012; accepted August11, 2012. Date of publication September 27, 2012; date of current version De-cember 19, 2012.

P. Sinha is with First Solar, Tempe, AZ 85281 USA (e-mail:parikhit.sinha@firstsolar.com).

A. Meader is with First Solar, Perrysburg, OH 43551 USA (e-mail:amy.meader@gmail.com).

M. de Wild-Scholten is with SmartGreenScans, Groet 1873GH, TheNetherlands (e-mail: mariska@smartgreenscans.nl).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JPHOTOV.2012.2214375

TABLE IU.S. WATER WITHDRAWAL BY CATEGORY [1]

and discharged. The net quantity of water avoided by deploy-ing a utility-scale CdTe PV array in the U.S. Southwest is alsoconsidered.

II. METHODS

Water usage can be more specifically described with the termswithdrawal and consumption. Water withdrawal is the amountof water removed from all sources (surface, ground, rain, tap wa-ter, etc.) Water consumption is the amount of water evaporated,incorporated into products, or otherwise removed from the im-mediate water environment. Water consumption is derived fromwater withdrawal minus water discharge.

The 2010 life cycle inventory (LCI) data for module manu-facturing were obtained from First Solar’s plants in Perrysburg,OH, Frankfurt-Oder, Germany, and Kulim, Malaysia. Global av-erage data were obtained by weighting data across plants basedon 2010 module production. Note that the module manufactur-ing inventory data have also been submitted for inclusion inthe Ecoinvent (V.3) database in 2012. LCI data for balance ofsystems (BOS) were obtained from First Solar’s planned 550MWac Topaz Solar Farm in San Luis Obispo County, CA [5].LCI data for end-of-life module takeback and recycling werebased on First Solar’s U.S. facility in Perrysburg, OH (Held,2009 [6] adapted to U.S. grid electricity). The U.S. facility waschosen for recycling as it would be the recipient of end-of-lifemodules from a U.S. project. However, recycling operations areconsistent across the First Solar plants. Key LCI parameters aresummarized in Table II.

Life cycle assessment (LCA) of this data has been conductedwith Simapro (V. 7.3.2) software and Ecoinvent (V. 2.2) unit pro-cesses. Water withdrawal estimates are based on all Simapro rawmaterial water categories (groundwater, lake, river, ocean, salt,cooling, unspecified, etc.), excluding water used in running hy-droelectric turbines (turbine use). The latter is excluded becausewater withdrawals typically represent off-stream use (water re-moved from the natural body), whereas turbine use representsin-stream use (use of water in the natural water body) [7].

LCA water impacts per megawatthour energy output followInternational Energy Agency Task 12 LCA methodology guide-lines [8] and are based on Q4 2011 average module conversionefficiency of 12.2%. Project-specific data on performance ra-tio (0.812), plane of array irradiance (2199 kWh/m2 /year), and

2156-3381/$31.00 © 2012 IEEE

430 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 1, JANUARY 2013

TABLE IIKEY LCI PARAMETERS CORRESPONDING TO FIG. 1

TABLE IIIDIRECT AND TOTAL LIFE CYCLE WATER WITHDRAWAL (L/MWH)

FOR CDTE PV

module degradation rate (0.70%/year) have been obtained in thefourth quarter of 2011 from First Solar’s performance engineerutilizing PVsyst V. 5.52 software [9]. For modules, a 30-yearlifetime is evaluated, and for BOS mounting structures, a rangeof 30–60-year lifetime is evaluated.

Water withdrawal displaced by deploying a CdTe PV arrayin the U.S. Southwest is based on the mix of electricity gen-eration sources in the California grid (USEPA CAMX eGridsubregion) [10]. These generation sources in combination withLCA water withdrawal estimates by generation source [4] areused to estimate net water withdrawal avoided (L/MWh) by a

Fig. 1. Major contributors to life cycle water withdrawal in CdTe PV.

CdTe PV array

ND =

(∑GS

fGS × WWGS

)− WWCdTePV (1)

SINHA et al.: LIFE CYCLE WATER USAGE IN CDTE PHOTOVOLTAICS 431

TABLE IVNET DISPLACEMENT OF WATER WITHDRAWAL (L/MWH) THROUGH DEPLOYMENT OF A CDTE PV ARRAY IN CALIFORNIA

Fig. 2. Water balance (m3 per m2 module) for CdTe PV manufacturing in Frankurt-Oder, Germany, in 2010. Specifically metered data are indicated by thesymbol “†.” Other data are estimated and, therefore. have a higher degree of uncertainty.

whereND net displacement of water withdrawal

(L/MWh);GS generation source;fGS fractional contribution of generation source to

California grid;WWGS life cycle water withdrawal per generation

source (L/MWh);WWCdTePV life cycle water withdrawal by CdTe PV

(L/MWh).

III. RESULTS

Table III shows LCA water withdrawal across the life cy-cle stages of CdTe PV. Approximately half of life cycle water

withdrawal is associated with module manufacturing, one thirdfrom BOS, and the remainder from takeback and recycling. Thedirect (on-site) withdrawal of water is a small (∼12%) percent-age of total LCA water withdrawal. In the module manufacturinglife cycle stage, the major contributors to water withdrawal aregrid electricity, glass, and on-site water (primarily for modulemanufacturing, cooling tower, and site irrigation). Steel, cop-per, inverters, and on-site water (primarily for site preparationand dust suppression during construction) are important to theBOS stage. During takeback and recycling, electricity, chemi-cal use (hydrogen peroxide), and transport are also importantcontributors to life cycle water withdrawal (see Fig. 1).

Direct water withdrawal for module cleaning during PV op-eration is not shown in Fig. 1 because, in general, First So-lar does not recommend module cleaning in its project sites.An exception is in the Middle East region, where a brush

432 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 1, JANUARY 2013

cleaning method has been developed that uses no water. ForPV in general, module cleaning during operation can result inup to 19 L/MWh of water withdrawal [3].

It should be noted that direct water use in the BOS stagevaries by location. For example, First Solar’s Desert Sunlightproject is also a 550-MWac project. Located in the MojaveDesert region of California, direct water use during constructionfor dust suppression and site preparation is approximately twicethe amount assumed here for the Topaz Solar Farm, which islocated in the less arid Carissa Plains of California. Accountingfor this variability would increase direct water withdrawal from∼12% to ∼15% of total LCA water withdrawal.

The net water withdrawal displaced by deploying a CdTe PVarray in the U.S. Southwest (California) is estimated with (1).There is a wide range of life cycle water withdrawal by genera-tion source depending on whether or not cooling water is recir-culated [4]. On the low end, the net water withdrawal displacedwith a CdTe PV array in California exceeds 1700 L/MWh (seeTable IV). An alternative estimate for displacement is obtainedwith Ecoinvent (V. 2.2) unit process data on life cycle waterwithdrawal for California grid electricity generation sources.Based on this approach, the net water withdrawal displacedwith a CdTe PV array in California exceeds 5600 L/MWh (seeTable IV).

Although life cycle water withdrawal by CdTe PV is lowcompared with most other electricity generation sources (seeTable IV), there are still opportunities to reduce waterwithdrawal during manufacturing. These efforts begin by un-derstanding the manufacturing water balance. Fig. 2 showsthe balance of withdrawal, consumption, and discharge duringCdTe PV module manufacturing in First Solar’s Frankfurt-Oder,Germany, plant in 2010.

During manufacturing, water is primarily withdrawn for pro-duction of ultrapure water (UPW) for use in module production,as well as for cooling tower water and site irrigation. Water con-sumption is approximately one-quarter of withdrawal and isassociated with cooling tower water evaporation and site irriga-tion, which account for 57% and 43% of consumption, respec-tively. Following on-site treatment, wastewater that has beentreated for metal constituents (metal wastewater) is dischargedto municipal sewer and nonmetal wastewater is discharged toriver. To reduce water withdrawal, nonmetal treated wastewa-ter could potentially be reused for site irrigation or as sanitary(gray) water.

Water balance data have been shown for First Solar’s plant inGermany based on the availability of metering. Manufacturingprocess water use is consistent across First Solar plants, butthere are climate-related differences in water use for irrigationand cooling tower operation. In addition, while on-site treatedwastewater is discharged to rivers and sewers in Germany, itis discharged only to sewers in the U.S. and only to rivers inMalaysia. However, the on-site wastewater treatment process isconsistent across plants.

While data for on-site water consumption have been pro-vided, estimates for indirect, upstream water consumption arenot available. Estimates could potentially be obtained using theEcoinvent database by subtracting water discharge from water

withdrawal. However, because there are no categories for dis-charge to nature in Ecoinvent (V. 2.2) unit processes, a strictwater balance cannot currently be achieved and is an area forfurther research.

IV. CONCLUSION

Life cycle water withdrawal by CdTe PV (382–425 L/MWh)is low compared with most other electricity generation sources.As a result, when deployed, CdTe PV arrays can displace waterwithdrawal from grid electricity by over 1700–5600 L/MWhin the U.S. Southwest. Primary contributors to life cycle waterwithdrawal are grid electricity, glass, and on-site water dur-ing manufacturing; steel, copper, inverters, and on-site waterin the BOS; and electricity, chemical use, and transport duringtakeback and recycling. Developing a manufacturing water bal-ance provides a baseline for reducing water withdrawal duringmanufacturing.

REFERENCES

[1] J. F. Kenny, N. L. Barber, S. S. Hutson, K. S. Linsey, J. K. Lovelace, andM. A. Maupin, “Estimated use of water in the United States in 2005,” U.S.Geological Survey, Reston, VA, Circular 1344, 2009.

[2] U.S. Department of Energy, “Energy demands on water resources: Reportto congress on the interdependency of energy and water,” Office Environ.Mgmt., U.S. Dept. Energy, Washington, DC, Tech. Rep. 597330, 2006.

[3] National Renewable Energy Laboratory, “Fuel from the sky: Solar power’spotential for western energy supply,” National Renewable Energy Lab.,Golden, CO, Tech, Rep. NREL/SR-550-32160, 2002.

[4] V. Fthenakis and H. C. Kim, “Life-cycle uses of water in U.S. electricitygeneration,” Renewable Sustainable Energy Rev., vol. 14, pp. 2039–2048,2010.

[5] County of San Luis Obispo, “Final environmental impact report for theTopaz Solar Farm project,” County of San Luis Obispo, Dept. Planningand Building, San Luis Obispo, CA, vol. 1, App. 8A, 2011.

[6] M. Held, “Life cycle assessment of CdTe module recycling,” in Proc. 24thEur. Photovoltaic Solar Energy Conf. Exhib., 2009, pp. 2370–2375.

[7] S. Pfister, A. Koehler, and S. Hellweg, “Assessing the environmentalimpacts of freshwater consumption in LCA,” Environ. Sci. Technol.,vol. 43, pp. 4098–4104, 2009.

[8] V. Fthenakis, R. Frischknecht, M. Raugei, H. C. Kim, E. Alsema, M. Held,and M. de Wild-Scholten, “Methodology guidelines on life cycle assess-ment of photovoltaic electricity,” (Int. Energy Agency, Paris, France, Rep.IEA-PVPS T12-03:2011, 2nd ed., 2011). Photovoltaic Power SystemsProgramme.

[9] C. Schwartz, personal communication.[10] U.S. Environ. Protection Agency. (2011). eGrid2010 Version 1.1,

Year 2007 Summary Tab. [Online]. Available at: http://www.epa.gov/cleanenergy/energy-resources/egrid/index.html

[11] R. Frischknecht, N. Jungbluth, J. H. Althaus, G. Doka, T. Heck, S. Hell-weg, R. Hischier, T. Nemecek, G. Rebitzer, M. Spielmann, and G. Wernet,“Overview and methodology,” Swiss Centre for Life Cycle Inventories,Duebendorf, Switzerland, Ecoinvent Rep. No. 1, Tab. 7.3, 2007.

[12] R. Dones, C. Bauer, R. Bollinger, B. Burger, M. F. Emmenegger,R. Frischknecht, T. Heck, N. Jungbluth, A. Roder, and M. Tuchschmid,“Life cycle inventories of energy systems,” Swiss Centre for Life CycleInventories, Duebendorf, Switzerland, Ecoinvent Rep. No. 5, 2007.

Authors’ photographs and biographies not available at the time of publication.