CLEFS CEA - N° 44 - WINTER 2000-2001

25

SOLAR ENERGY

Thermodynamic solar electricity

Focusing the sun rays on a single pointor on a line can produce high temperatures.This principle of concentration, known sinceantiquity, uses parabolic collectors, cylin-drical-parabolic collectors, or “solar ther-mal-electric” power plants, for which a mul-titude of multi-directional heliostats focusthe sun’s energy on a single boiler locatedat the top of a tower. This enables the hea-ting of heat-transfer fluids, generally oil ormolten salts, in a temperature range run-ning from 250 to 800 °C, depending on thetechniques used. These heat-transfer fluidsthen heat the steam, which drives a turbo-alternator, as in standard thermal powerplants.

Prototype units from around a few dozenkilowatts (kW) up to a dozen megawatts(MW) have been built throughout the worldover the past two decades. In the Pyrenees,the Thémis power plant, with a power outputof 2 MW, was put into operation at the begin-ning of the 1980’s. The largest commercialdevelopment was nonetheless accomplishedby the Luz Corp. company, which, in thesame decade, built three power plants withcylindrical-parabolic collectors amountingto a total power rating of 354 MWe, sup-plying the Southern California Edison net-work – which feeds Los Angeles – with peakperiod electricity during summer afternoons.In spite of the constructor’s bankruptcy tenyears ago, these power plants have been ope-rating non-stop and their productivity hasimproved. They now confirm the relative

THE THREE WAYS TO EXPLOIT SOLAR ENERGY

The three ways to exploit solar energy “directly” are undergoing further developments in orderto improve their performances and, above all, their economic viability. The oldest way is the useof thermal collectors, for direct heating of premises or water circuits. It is an attractive solutionthat has been promoted by guarantees, subsidies and technological improvements. Anotherone, based on thermodynamic, was developed some years ago with installations in France suchas the Odeillo solar furnace or the Thémis 2 MW power plant, and is now used on a largerscale in California. The direct transformation of sunlight into electricity, i.e. photovoltaic power,constitutes the potentially most promising way forward. For some ten years, the CEA has beenworking (box, p. 26) on developing its advantages and minimizing constraints.

CEA/Coulon

Photovoltaic cells made ofpolycrystalline silicon tested atthe CEA by the Genec inCadarache (Bouches-du-Rhône).

maturity of this way of producing power,with the cost-price of electricity at around0.1 Euro per kWh.

A potential improvement of 20 to 30% ispossible, especially if the steam is directlyproduced in the collectors and the mirrorsare optimised. The United States, Israel and,in Europe, Germany and Spain, are conduc-ting joint research on these topics. In the fra-mework of financing facilities made avai-lable by the Global Environment Facility,new plants are to be built within the next twoto three years in several countries includingEgypt, India and Brazil.

Thermal solar energy

Thermal solar energy is used mainly intwo applications: heating of domestic hot

CLEFS CEA - N° 44 - WINTER 2000-2001

water and heating of premises. For theseapplications, the glazed collectors usedensure efficiency of around 50% at the requi-red temperatures. Four square meters are suf-ficient to satisfy the hot water needs of afamily of four, with an average investment of3 000 Euro, and ten to twenty square metersare enough to heat a single-family house.

An auxiliary heating system is necessaryfor cold weather periods and, on a yearlyaverage, a solar heating system covers around50 to 60% of a family’s needs, thus provi-ding the same amount of savings. Depen-ding on the types of substitute back-upenergy, return on investment is between 6and 12 years.

Every year such thermal collectors pro-duce from 200 to 800 kWh per m2, depen-ding on the needs and methods of use. Thelowest values correspond to intermittent usesof domestic hot water at high temperatures(more than 55 °C), with the highest valuesbeing obtained in the case of continuous low-temperature heating. For this type of appli-cation, often called “direct solar floor”, theheat-transfer fluid from the collectors is injec-ted directly into building floors at a tempe-rature of 25 to 30 °C. On the one hand, thisdesign produces fairly comfortable housing-units and, on the other hand, it offers one ofthe best ways to achieve technical and eco-nomic profitability.

The development of the European marketwas relatively slow, at around 250 000 m2

per year in the 1980’s, due to numerous coun-

ter-references caused by the lack of trainingavailable for installers. But the introductionof new concepts, such as solar performanceguarantees, along with the activity of theGerman, Austrian and Dutch markets duringthe 1990’s, brought about substantial growth,with an increase in two and four-fold factorincrease in surface-areas sold over the lastfive years compared with the previousdecade. In the short term, the Europeanannual market is estimated at several mil-lion square meters. In France, the recentnational program, Helios 2006, is intended tofurther increase distribution of these pro-ducts by providing users with financial incen-tives.

Current technological developments areintended to lower costs, by means of impro-ved integration and implementation in thestructural frame of buildings.

Photovoltaic solar electricity

Real advantages and genuineconstraints

Photovoltaic solar energy is an interes-ting way to reduce electricity distributioncosts in certain regions. Readily available inmost countries located between the equatorand the 45th parallels, it is a remarkablyreliable energy source that has a completelypositive energy balance and environmental

report card (box C, How does a photovol-taic solar cell work?).

As a resource, solar energy is relativelywell distributed geographically and is thusavailable in numerous places. Henceforth,solar cells or photovoltaic modules can beused to produce electricity on the spot, i.e.where the needs are actually located. This isthe essential advantage of solar electricity:it means none of the distribution costs inhe-rent in conventional solutions, whether thisinvolves the use of generating sets suppliedby fossil energies (diesel, petrol or gas), orthe extension of a main electricity power lineup to the point of use.

In fact, in the first case, it is necessary totake account of the availability and supplycost of the fuel up to the site in question,as well as routine maintenance. In thesecond case, the costs of extending or rein-forcing a line considerably increases theprice of a kWh, especially if energy needsare low. If they are high, the depreciationon each kWh will be reduced proportio-nally.

The absence of any mechanical movementor flow of fluids ensures photovoltaic elec-tricity with an exceptionally high level ofreliability: the most widely-sold models,based on silicon crystalline, are now cur-rently backed up by guarantees of aroundtwenty years and even longer service-lives.The energy balance is positive, since a pho-tovoltaic module returns the energy neces-sary for its production in two to four years of

SOLAR ENERGY

26

B. Charlon/GAMMA

Multi-directional mirrors on theThémis solar power plant site, anexample of the implementation ofthermodynamic solar electricity.

Genec at the age of ten

The Genec (Cadarache researchgroup for new energies), founded in1991 under the name “Groupementénergétique de Cadarache” (Cadaracheenergy group), is responsible forresearch and technological develop-ment in the field of the rational use ofphotovoltaic solar energy and thermalenergy. As a result of cooperation bet-ween the CEA and the Ademe (Agencyfor the environment and energycontrol), the Genec has acquired exten-sive experience in laboratory testingand work on real sunshine, as well asin technical assistance for photovol-taic electrification programs, especiallyin developing countries.

The Genec users association bringstogether industrialists, cooperatingorganizations and the Genec group forthe purpose of optimising work andinformation and creating partnerships.

27

exposure to the sun, depending on its manu-facturing technology.

The main factor limiting the feasibilityof a photovoltaic system is the amount ofenergy required, which must be in line withthe resource’s possibilities. To give a feworders of magnitude, lighting needs amoun-ting to a few hours a day will require a daily“output capacity” of several dozen Wh,which corresponds to a surface of photo-voltaic modules well under one per squaremeter. Domestic needs associated with amodern level of comfort, including televi-sion, hi-fi and household appliances, orpumping facilities for the distribution ofvillage water, will require several kWh, oreven, in some cases, around ten kWh perday. In all these scenarios, installations ofaround one square meter or ten squaremeters do the job, and they rarely causeintegration difficulties.

On the other hand, the use of solar elec-tricity is rarely cost-effective for higherneeds, when these are located in one placeonly: this energy source is not appropriatefor high-energy intensities.

Photovoltaic solar electricity can be usedin two ways: to supply electricity to isola-ted sites or to inject electricity into an elec-tricity power supply system.

Increase in the supply of isolated sites

The first application is the oldest and mostwidespread with regard to the millions ofsystems installed throughout the world. Thistype of application began in the 1960’s for

satellites, in which photovoltaic solarmodules clearly dominated most other solu-tions, mainly owing to questions of weightand reliability.

The first ground-based applications cameinto widespread use in the 1970’s, mainlyfor professional requirements (the power sup-ply of weather stations or telecommunica-tion relays). In the 1980’s, numerous mar-ket gaps appeared, one after the other:maritime and air beaconing and lighting sys-tems, the cathodic protection of pipelines orpylons, urban furniture and, above all, ruralelectrification, which mainly covers needssuch as interior lighting, audiovisual equip-ment and water pumping. In the near future,the supply of water to isolated sites and itssubsequent treatment in order to render itdrinkable, will represent a very substantialmarket, since considerable environmentaldifficulties are foreseeable in this field.

The most recent years have seen anincrease in the number of facilities built ineach of these sectors. In France, 9% of seabuoys are equipped this way. In developingcountries, all telecommunication stations ormicrowave stations use this energy source.Rural electrification programs are now car-ried out in blocks of several thousand sys-tems.

The main characteristic of this initial cate-gory for the application of solar electricityis that it requires the use of batteries whenelectricity needs are not in phase with pro-duction, that is, the resource, or, put other-wise, the sun (box F, Storage batteries, cellsand batteries: steadily improving perfor-mance).

Accelerated growth forelectricity, “as the sun shines”

The second, more recent application andthus, for the time being, one which is lesswell-developed, is experiencing an evenhigher growth rate. It is based on the ideaof transforming the direct current from thephotovoltaic modules directly into alterna-ting current, identical to the current which isused in low-voltage electricity power sup-ply systems. The electricity which is thusgenerated “as the sun shines” is either consu-med in situ, or injected into the network.The cost-related advantage and the currententhusiasm for this solution result from thefact that electricity generated in this waycan be sold to electricity distribution com-panies.

Such electricity can be “hooked up” tothe grid following a centralised layout withphotovoltaic power plants of a few mega-watts: power plants were built this way inthe United States in the mid-1980’s. Thecomplementary approach uses the distri-buted characteristic of the resource and islimited to “domestic” facilities, that is, faci-lities of around a few kilowatts. The pre-cursors in the field of “solar roofs” wereSwitzerland and Germany at the end of the1980’s. Several operations or demonstra-tion programs, whereby modules are instal-led on building façades or in roofs, were car-ried out during the 1990’s. At present largeprograms are under way, mainly in Japanand northern Europe, and are advancing at arate of a thousand roof-mounted systems permonth.

CLEFS CEA - N° 44 - WINTER 2000-2001

CEA/Coulon

Test on the CEA/Cadarache solarsimulator of double-wall roofmodel permitting natural aircurrents (created in this case byblades) to lower the temperaturein a housing-unit.

CLEFS CEA - N° 44 - WINTER 2000-2001

Cost-prices consistentlydropping

In 1999, the market amounted to a volumeof 200 MW of electric power modules, anda total turnover for the photovoltaic sectorof over 1.5 billion Euro. Growth is high, with18% per year on average over the past tenyears, and gathering speed: up 27% over thepast five years, and up 34% per year overthe past three years. New large-scale distri-bution programs of photovoltaic roofs inJapan, and subsequently in Germany, are atthe origin of this upswing.

Forecasts for the year 2010, based ongrowth estimates of 20 and 25% per year,show an annual market of 1 500 and 2 300MW, respectively. For the year 2020, accor-ding to the same estimates, the marketshould be between 9 and 21 GW. At pre-sent, the main producing countries are, inorder of importance, Japan, the UnitedStates and Europe, with respective marketshares of 40, 30 and 20%. The main partiesworking in this area are the major petro-leum industrial or electronic groups: BPAmoco, Siemens, Kyocera, Sharp andSanyo. Photowatt, a company of French ori-

28

SOLAR ENERGY

CEA/GENEC

Lamp posts with individualphotovoltaic cell power supplyin Antibes (Alpes-Maritimes).

gin with facilities in the Rhône-Alpes, isnumber one on a European scale, and num-ber seven worldwide.

The selling price of modules on the inter-national market is currently around 3.35Euro per watt (price, ex-works in mass-pro-duction runs). Prices and sales volumes overtwenty years show a steady progression, cor-responding to price cuts of fifty percentevery ten years. The forecast for the futureis a price of around 1.5 Euro per watt in theyear 2010. For a photovoltaic system of afew KW, hooked up to the grid, the cost-price for the user, including installation andpower inverter, is about twice the price ofthe module ex-works, i.e. a little more than6 Euro per watt. For a system equippedwith battery storage, prices vary between 9and 12 Euro per watt, depending on theapplications. ●

Philippe Malbranche and Olivier Dieudonné

GenecTechnological Research Division

CEA/Cadarache

CLEFS CEA - N° 44 - WINTER 2000-2001

29

From ingot to wafer

A photovoltaic cell has the ability todirectly convert solar rays into electricity(box C, How does a photovoltaic solar cellwork?). From an electronic standpoint, it isa diode. The most widely used semicon-ductor material for its manufacture is crys-talline silicon: the width of its bandgapallows a high conversion efficiency, its tech-nology is well mastered, and it is very abun-dant. Presently, over 80% of solar cells aremade with crystalline silicon. Amorphoussilicon represents only about 10% of themarket, since cells manufactured with thismaterial have conversion efficiencies thatare half that of crystalline silicon cells. Theportion of other materials (CdTe, CIS, etc.)remains very low.

Photovoltaic cells are manufactured fromvery fine silicon wafers (thickness from 200-350 µm, with a surface area ranging from10 × 10 cm2 to 15 × 15 cm2), cut from mono-crystalline or multicrystalline silicon ingots.The Czochralski or CZ process is used tomanufacture monocrystalline ingots. A seed,which itself is monocrystalline, is immersedat the surface of a bath of slightly overhea-ted silicon in a silicon crucible, and pulledat a constant speed. Solidification takes placeand reproduces the seed's crystalline pattern.If the pulling conditions are properly control-led, this operation produces a monocrystal-line silicon cylindrical ingot of with nodefects. This process allows the productionof cells with excellent conversion efficien-cies (16 to 17% in production), but remainsfairly expensive.

The process used to manufacture multi-crystalline ingots involves silicon melting insilicon crucibles. This method consists inslowly cooling liquid silicon in a given direc-tion in order to achieve a solidification struc-ture that is uniformly oriented in this direc-tion. It yields an ingot with fairly large grains(1 cm) that can be cut into cells having agood efficiency (14%). It is a slow but inex-pensive process.

The silicon obtained must be very pure,since the presence of impurities, even slight,

Photovoltaic solar modules: from crystalline silicon to thin film

Photovoltaic solar cells presently use silicon as a raw material, as do most microelectroniccomponents. After having used the manufacturing scraps from the microelectronics sector,today solar cell production is beginning to follow its own specifications and adopt new methods,such as thin film, or even the use of plastic materials. The CEA is working on several of these newtechnologies.

CEA/Coulon

From a silica stone to a siliconblock.

will reduce the cells' efficiency: the impu-rity concentrations tolerated range from 0.1to a few dozen ppm (parts per million) andare even of the order of a few dozen ppb(parts per billion) for some impurities. Oncesolidified, the crystalline silicon blocksobtained are cut into ingots and then wafers.Wafer cutting is performed with a wire saw,a long process that consumes a lot of mate-rial: the saw makes a cut approximately200 µm wide, and almost half the siliconis lost.

Cells and modules

Overall, silicon wafer manufacturerepresents 40% of the price of a module.This module fulfils several functions: theconnection of the cells to each other inorder to supply the desired voltage (usuallya series of 36 cells for a 12 V output), andprotection of the cells from environmentalstress (erosion, humidity, hail, salt, UVs,etc.). In order to reduce costs, savings mustbe sought at all stages: silicon purification,

CLEFS CEA - N° 44 - WINTER 2000-2001

ingot manufacture and wafer cutting. Oneattractive solution consists in producingribbons of silicon, which eliminates wafercutting. However, this technique, extensi-vely developed over the past twenty years,has not been able to establish itself, in par-ticular due to the inferior quality of siliconobtained.

For a cell to work several functions arerequired: maximum absorption of light overthe entire solar spectrum, efficient collec-tion of the carriers (electrons and holes)

generated by the photons, and the esta-blishment of an electrical connection withthe exterior circuit. The first step of itsmanufacture is therefore a chemical attackon the surface, to clean it and make it rou-gher, and therefore less reflective. The junc-tion is then formed by diffusion of dopants(the wafers used are generally p-doped butphosphorous is introduced by diffusion inorder to n-dope the silicon to a depth ofabout a micrometer), and an anti-reflectivecoating is deposited. The metal grids usedto collect the current, very narrow to avoidcreating any shadowing effect, are thenapplied by screen printing. An annealingoperation then allows the creation of theelectrical contact between the silicon andthe collecting grids. The conversion effi-ciencies of these cells range from 14 to 17%,depending on the type of material used(multi- or monocrystalline) and the cellmanufacturing process. An aluminium rib-bon designed to make the interconnectionsbetween cells is then welded onto the grids.Then the cells are individually tested, sor-ted according to conversion efficiency, andassembled into modules. The irradiated sur-face of the cells is bonded to a plate of tem-pered glass which ensures mechanicalprotection. The back of the cells is protectedwith a sheet of glass or plastic. Thesebonding operations are performed with apolymer which ensures protection fromhumidity.

30

SOLAR ENERGY

How does a photovoltaic cell work?

Silicon was chosen for the manufac-ture of photovoltaic solar cells becauseof its electronic properties, characteri-zed by the presence of four electrons onits peripheral layer (columnIV in Mendeleiev's table). Insolid silicon, each atom isbonded to four neighbors,and all the electrons in theperipheral layer help thebonding process. If a siliconatom is replaced by an atomfrom column V (phospho-rous for example), one of theelectrons no longer partici-pates in the bonds, and cantherefore move around inthe network. There isconduction by one electron,and the semiconductor isconsidered of the n-doped

type. If on the contrary a sili-con atom is replaced by anatom from column III (boron forexample), an electron is then missing tomake all the bonds, and an electron cancome to fill this hole. This is called hole

C

conduction, and the semiconductor isconsidered of the p-doped type. Atomssuch as boron or phosphorous are sili-con dopants.

zone that was initially n-doped becomespositively charged, and the zone that wasinitially p-doped becomes negativelycharged. An electrical field is therefore

created between the n andp zones, which tends topush the electrons into then zone, and a balance isestablished. A junction hasbeen created, and by addingmetal contacts onto the nand p zones, one thencreates a diode.

When this diode is irra-diated, the photons areabsorbed by the material,and each photon givesbirth to an electron and ahole (this is called an elec-

tron-hole pair). The junc-tion of the diode separatesthe electrons from theholes, creating a voltage

between the n and p contacts, and acurrent circulates if a resistor is pla-ced between the diode contacts(figure).

- + -+

I

contact on n zone

contact on p zone

n-doped zonephoton absorption

carrier generation

carrier collection

p-doped zone

When a type n semiconductor is pla-ced into contact with a type p semicon-ductor, the excess electrons in the nmaterial diffuse in the p material. The

Crystalline silicon cells assembledinto modules on the CEA/Cadaracheexperimental site.

J.-F. Mutzig

CLEFS CEA - N° 44 - WINTER 2000-2001

31

The technological obstacles

Up until now, the volumes of silicon usedin the photovoltaics industry were low (3000metric tons per year, i.e. about 10% of thesilicon consumed in the microelectronics sec-tor), and the silicon came from scraps (ingottops and tails, polycrystalline silicon) fromthe microelectronics sector, which supplieda silicon with a purity of a few ppb, higherthan that required for photovoltaic cells.However, the growth forecasted for the pho-tovoltaics sector now makes it necessary tofind an autonomous source that could sup-ply silicon at a low cost with a level of puritysuited to its needs. Among the possible alter-natives to the use of microelectronic scraps,the fluidized bed process using metallurgi-cal silicon as a basic material appears themost promising. It would allow the produc-tion of silicon of satisfactory quality at anaccessible cost. The German chemical groupBayer is presently studying the industriali-zation of this process.

Today's ingot manufacturing process isslow. The industry is moving towardsmethods involving continuous casting in acold crucible, whereby the melted silicon isintroduced, confined by an electromagneticfield, and solidified without any contact withthe walls. This makes it possible to conti-nuously obtain large-sized ingots (35 × 35 cm2

in section and 3 m long) with a crystallinestructure compatible with photovoltaic appli-cations. This type of process has severaladvantages in comparison with conventio-nal processes: the electromagnetic confine-ment prevents contamination from the cru-cible, productivity is multiplied by 10, andthe ingot's homogeneity allows productionof cells with a very targeted conversionefficiency, so that cell sorting is no longernecessary.

Concentration cells

In a concentration system, a small-sizedcell is placed in the focal spot of an opticalsystem (figure 1). The surface area of thesilicon cell is thus divided by the concen-tration factor, which can be as high as 300.The cell can be made of very good qualitysilicon with microelectronic technologies,which authorizes higher conversion effi-ciencies (above 20%) than those of cellsused without concentration (15%). Howe-ver, the system must be placed on an orien-table support in order to track the sun. Thecost of the system is then essentially deter-mined by the optical system, which can bemade of plastic and therefore remains fairlycheap, the assembly, and the tracking sys-tem. The low concentration systems (lessthan 50) use cylindrical optical systems thatrequire the sun be tracked according to one

axis, and the high concentration systems(over 100) use a system with two axes. Thisapproach is particularly well-suited to elec-tricity generation in mini power stations,and should allow to reduce the cost of thegenerated energy. In order to improve theseconcentration systems, researchers are tryingof course to increase the cells' efficiency,but also to develop optical systems withshorter focal distances in order to reducepanel thickness and therefore the spacerequired.

Future prospects: thin film modules

The scientific andtechnological obstacles

The low consumption of raw materialand the direct production of the materialusing conventional deposition techniqueson a low-cost support material such as glass,

PHOTOWATT INTERNATIONAL

100 mm multi-crystalline siliconcell.

optical systemof concentration

silicon cell Figure 1. Schematic diagram of a concentration cell.

CLEFS CEA - N° 44 - WINTER 2000-2001

steel or polymer, without requiring any othercostly shaping stages such as sawing, makethe thin film technology a particularly attrac-tive solution for photovoltaic cells (see box).Crystalline silicon, due to its electronicband structure, is not naturally the best-sui-ted semiconductor material for thin film pho-tovoltaic conversion. Indeed, the absorptiondepths for the portion of the solar spectrumlocated in the close infrared range are high,and can be measured in dozens of microns.Nevertheless thin film monocrystalline sili-con cells have their advantages, as we willsee below.

In order to achieve a thin film cell, it ispreferable to use semiconductors with adirect electronic bandgap (rather than indi-rect like silicon) and with a value suited tothe solar spectrum (of the order of 1.5 eV).With such materials, the entire solar spec-trum can be absorbed in a thickness of aboutone micrometer. Moreover, the shallowabsorption depth makes the cell fairly tole-rant of defects acting as recombination cen-ters for the charge carriers (electrons andholes).

Numerous research efforts have thereforebeen carried out on a fairly large numberof materials over the past forty years, inorder to achieve thin film photovoltaic cellswith good conversion efficiency at a lowcost. Historically, two materials have beenespecially studied and tested in industriali-zation: amorphous silicon and cadmium tel-luride (CdTe), both deposited on glass. Inspite of the research efforts, the maximumefficiency of this type of cell remains unfor-tunately limited for significant cell sizes.This is due to the difficulty in achievingthis type of material with a low defect den-sity and thus good electronic properties.Moreover, amorphous silicon suffers from

an aging effect linked to the instability ofthe hydrogen in its structure, and the pre-sence of cadmium, a heavy metal with atoxicity comparable to that of mercury,makes the CdTe fairly inappropriate forconsumer applications. Recently, remar-kable progress was made with another typeof material, chalcopyrites, such as the refe-rence materials copper indium di-selenide(CuInSe2) called CIS. This path of researchhas therefore rapidly become the referencein thin film research.

But other possibilities have not yet beenexplored, and efforts are being made in manyother directions: manufacture of silicon mate-rial with very small grains leading to a modi-fication of the electronic band structure (so-called "micro-crystalline" silicon), multi-junction cells based on amorphous silicon-germanium alloys, and cells based on orga-nic semiconductor materials.

One of the most important aspects of thinfilm research is also linked to the possibilityof applying these films directly to a flexiblematerial which could then be applied to aconstruction material. This poses anotherchallenge: to find support polymers that resisthumidity and thus prevent any corrosion ordeterioration of the semiconductor materialand electrical connections.

The field of research on these thin filmmaterials remains fairly wide open. Theentire range of possible materials has notbeen studied in detail. Numerous efforts willstill be necessary in order to achieve anapplication that combines minimum costwith maximum efficiency. Finally, the ele-mentary physical mechanisms that controlthe initial efficiencies and the deteriorationfactors are still poorly understood in all ofthe thin film materials studied, whatevertheir maturity.

32

SOLAR ENERGY

(0,3 )(0,005 )(2 )(1 )(3000 )

ZnO upper contactn-CdS/p-CIS junction

CIS layermolybdenum back contact

glass plate

Figure 2. Schematic diagram ofthe stacking of a CIS (copper

indium di-selenide) photovoltaiccell and of the electrical

interconnection between twocells in series. The thickness of

the different layers is indicated inparentheses.

CLEFS CEA - N° 44 - WINTER 2000-2001

33

Technologies currently beingdeveloped or industrialized

In the past few years, while the conversionefficiencies of solar cells based on thin filmsof CdTe have stagnated at around 16% in thelaboratory, constant progress has been achie-ved with CIS semiconductor material. Anefficiency value of 18.8%, a record for thinfilms, was achieved in the laboratory with acell of small size. For modules measuring30 × 30 cm2, performances of 12.8% havealready been achieved. Moreover, funda-mental research has shown that this materialhas the quality of remaining very stable underirradiation, as well as the ability to auto-rege-nerate. This auto-regeneration can be explai-ned by the chalcopyrite structure of this mate-rial, which is not completely ordered, andby the fact that numerous defects such as thelack of atoms, or foreign atoms (sodium,iron, gold, etc.) are present, and that the cop-per atoms it contains are relatively mobile.These results have stimulated renewed inter-est in this field of research, especially in Ger-many.

The stacking structure of a CIS cell (figure2) includes a glass support and a molybde-num electrode deposited by cathode sput-tering. The junction between the cadmiumsulfide (CdS) of the n type and the CIS ofthe p type is obtained by depositing a 50-nanometer film of CdS in a chemical bath(CBD-Chemical Bath Deposition) on theCIS. The upper contact consists of a trans-parent zinc oxide (ZnO) electrode doped withaluminium. Several deposition techniqueshave been used to obtain the CIS film(vacuum evaporation, selenization of themetallic precursors, cathode sputtering orspraying, electrodeposition). The best per-formances have been achieved with the tech-nique based on co-evaporation of the ele-ments. The cell's final surface is thenencapsulated in a layer of polymer and a pro-tective glass plate.

The main industrial actors in this field areSiemens Solar (Germany, USA), Wurth Solar(Germany) and Matsushita (Japan). InFrance, no industrialist has yet entered thisfield, whereas in the research sector, the elec-trochemistry laboratory of the ENSCP(National school of chemistry in Paris) hasbeen working on this material for severalyears, with international renown. EDF, Saint-Gobain Recherche and the ENSCP are par-ticipating in a project that would allow thedevelopment in France of an original typeof CIS thin film based on electrodeposition.

Thin film monocrystallinesilicon cells

Even if, as we have seen, silicon requiresthicknesses greater than 10 microns to

achieve sufficient conversion efficiency, avery attractive solution to manufacture pho-tovoltaic cells at a low cost consists in takinga film a few dozen microns thick from amonocrystalline silicon wafer and applyingit to a low cost support, such as glass or cera-mic.

This would make it possible to use onlythe amount of silicon that is strictly neces-sary for the cell to work, and avoid losseslinked to wire sawing, while still preserving

Photowatt-CEA:

intensified cooperation

The working partnership establi-shed between Photowatt Internatio-nal, the leading European verticallyintegrated manufacturer of photovol-taic wafers, cells and modules, theCEA/Cadarache and the Leti in Gre-noble should be increasing in thecoming years, in the areas of moduleintegration in the building sector andthe development of new photovoltaiccell manufacturing processes. Photo-watt is pursuing its strategy of growthin a rapidly expanding internationalmarket. Having progressed from anannual production capacity of 5 MWin 1997 to 18 MW today, PhotowattInternational also proposes an expan-ded product range, with two sizes ofwafers and cells, and a line of modulesgenerating from 10 to over 100 peak

watts. This has allowed the companyto broaden its geographical presence:Photowatt is now equally present inEurope (the "grid-connected" market,particularly in Germany), America(thanks in particular to its Americanparent company Matrix Solar Tech-

nologies, a subsidiary itself of theCanadian company Automation Too-ling Systems), and the Asian-Pacificarea (Japan is the largest market inthe world). The objective is to rankamong the top five international manu-facturers. Photowatt attaches parti-cular importance to the developmentof new technologies and new manu-facturing processes. With the partici-pation of Ademe, the company devotesapproximately 15% of its turnover totechnology research and development.Its main objectives in the short andmedium terms are: 1) a continuouscasting furnace for silicon in a coldcrucible, making ingot manufactureten times faster than the techniquepresently used, with a reduction by afactor of 2 of the added value, and 2)the development of an innovative cellmanufacturing process allowing asignificant increase in the conversionefficiency. Other projects for the lon-ger term involve silicon refining andthe development of ultra-thin cells(100-150 µm).

Manufacturingmulticrystalline siliconcells at the PhotowattInternational plant inBourgoin-Jallieu (Isère).

PHOTOWATT INTERNATIONAL

a material and technology that have provento be especially reliable. Several processesare currently being studied in laboratories inJapan, Germany, Australia and France. Theyall have in common the weakening of thesilicon in depth in order to be able to detachthe surface film located above this fragilezone and apply it to a large-sized, low-costsupport (figure 3). An additional advantage:the performance of the technological stageson the cells is done at the scale of a large

CLEFS CEA - N° 44 - WINTER 2000-2001

surface module and no longer for each sili-con wafer, which allows significant reduc-tion of the cost per unit of surface area. TheCEA/Leti (Electronics and information tech-nology laboratory), which has been a pio-neer in developing this type of approach formicroelectronic applications, proposes tostudy its application to photovoltaic cells inthe framework of a contract with Ademe,and in collaboration with the INSA engi-neering school in Lyon.

Organic cells: moving towardstotal polymer composition

Research and development on solar cellsbased on organic materials or polymers ismotivated by the advantages that these mate-rials present: low cost, unlimited raw mate-rial, simple implementation, low tempera-ture technologies, large surface areas, flexiblesystems, etc. Moreover, this solution wouldmake it possible to process the substrate(mechanical support), the active materialwhere the photovoltaic conversion takes

34

SOLAR ENERGY

JKU Linz

Flexible photovoltaic celldeveloped at the University of

Linz (Austria).

low cost support

monocrystalline silicon

thin film paving

cell technology

Figure 3. Schematic diagram ofthe transfer of a thin film and itsapplication to a large size, lowcost support, with performanceof the technological stages of cellmanufacture at the scale of themodule.

place, and the encapsulation, all with a singletechnology.

Today organic photovoltaic cells exist thathave a conversion efficiency higher than10%. They are based on the so-called Grät-zel technology which consists in a junctionbetween an organic polymer and a liquidelectrolyte.

The photovoltaic generation is located inthe polymer and the electrolyte ensures thecharge transfer and the voltage (electromo-tive force) via its junction with the polymer.This type of cell is being developed in Swit-zerland by Solaronix for low power appli-cations, and in Germany by the INAP (Ins-titut für Angewandte Photovoltaik) for highpower applications. The presence of liquidelectrolyte is one of the major drawbacks ofthis technology, due to its lack of stabilityover time (evaporation) and limited range ofoperating temperatures.

Researchers are therefore now focussingon a total polymer solution. In this field, thebest performance achieved to date is aconversion efficiency of 3.6%. The numberof laboratories working in the field, at a stagethat is still relatively upstream, remains fairlylimited, but several publications, patents andlectures illustrate the emergence of this sub-ject of research which could enter the deve-lopment phase around the year 2010. TheCEA/Leti has acquired significant experiencein this area by working on the first Euro-pean contract on the subject, and has pro-posed an original way to improve the col-lection performances of the charge carriersin this type of structure. A national projectbringing together the Leti, several univer-sity laboratories and the TotalFinaElf groupis presently being organized to developpolymer photovoltaic cells for energy gene-ration applications. In a first phase, thisproject will constitute an upstream researchactivity. ●

Claude Jaussaud, Jean-Pierre Joly,Alain Million and Jean-Michel Nunzi

Technological Research DivisionLeti electronics and information

technology laboratoryCEA/Grenoble

CLEFS CEA - N° 44 - WINTER 2000-2001

35

The interface betweenthe user and the resource

The photovoltaic system is the interfacebetween the user and the resource, it pro-cesses the energy captured by the photo-voltaic modules according to the types ofapplications. In addition to an associationof photovoltaic modules, an inverterconverts the direct current into alternatingcurrent for use on an electric grid. The usermay then either use the energy that he hasproduced or re-inject it into the electric gridif, for example, the operator's purchasingconditions are appealing to him. The inver-ter may set off a pump in the case of a sun-shine direct coupled pumping system day-time, water is then forced back into areservoir sized according to the require-ments of the village and released ondemand.

If it's necessary to store the electricalenergy produced, a storage bank is introdu-ced. This bank will be managed via a regu-lator: it charges when the sunshine allows itand supplies the user as required. This typeof storage allows, on one hand, to compen-sate for the day-night alternation as well asfor several consecutive days of bad weatherconditions, and on the other hand, to meetpower requirements that are substantiallyabove those that the photovoltaic generatormay supply instantly.

Hybrid solutions

This type of architecture becomes morecomplex for more important applications: inorder to avoid the setting-up of a sizeableand therefore costly storage bank, an auxi-liary generator such as a Diesel generatormay be selected. It will then be a photovol-taic system said to be "hybrid", meaning thatit associates a photovoltaic generator withan energy source, conventional or not. If theweather conditions are favorable, the asso-ciation of several renewable sources (pho-tovoltaic, wind power or micro-hydraulic)may even be considered.

The research and development directionsfor all of these systems aim mainly at loweringthe electricity production costs. This decreaseis obtained by playing on the cost/performanceratio of each component making up the sys-tem, but also on the more general factors ofsystem management and architecture.

In the order of importance of initial invest-ment costs for a hybrid photovoltaic system,for example, the photovoltaic generator repre-sents on average 35%, the battery bank 20%,the other components 20%, the additionalenergy sources (Diesel generator) 10%, thelogistics and the installation 15%. The inte-gration of O&M (Operation and mainte-nance), intervention and replacement ofequipment (the storage bank for example)disrupts the "life cycle" costs breakdown:the battery bank then represents 50%, thephotovoltaic generator 20%, the additionalenergy sources 15%, the other components10%, and the logistics and installation 5%.

Research on photovoltaic modules hasdirect effects on the cost of the systems interms of increasing the performance andconversion efficiency of the cells (morepower at a constant price), but also in termsof reducing the production costs by wor-

Photovoltaic systemsBeyond the module that holds the cells exposed to the sun, the exploitation of photovoltaicelectricity uses "systems" whose future development requires research efforts, on each com-ponent, to improve their economy and reliability. This is especially true for the essential linkthat is storage, as the main constraint weighing on the industry is the need to store electricitybetween periods of sunshine. The CEA is attempting, among other things, to model the ope-ration of photovoltaic systems by integrating the evolutions of the storage characteristics overtime.

Installation destined to meet theelectricity and waterrequirements of a holiday campin Alice Springs in the Australianoutback. It associatesphotovoltaic modules, a batterybank, a Diesel generator, a boringpump, a desalination unit and afresh water tank.

Ph. Malbranche

CLEFS CEA - N° 44 - WINTER 2000-2001

king on the manufacturing processes (pho-tovoltaic module less expensive at equalpower, see Photovoltaic solar modules: fromcrystalline silicon to thin film). It is reaso-nable today to estimate that a gain marginexists that would allow a 30% reduction inthe cost of a complete system. The two othermost important research directions shouldnot be neglected. In the storage bank, thepotential gain is estimated at 50% if, forexample, it is possible to double the life-time of the battery. For the other compo-nents, the goal is to minimize the conver-sion losses: the "solar" inverters are muchmore efficient than standard inverters. Theirefficiency reaches 95% at their rated powerwith a very low self-consumption, whilethe efficiency of Uninteruptible Power Sup-ply inverters is barely 80% with a far higherself-consumption. The price varies as aresult.

Increasing reliability,energy management andnew architecturesIn the case of complete systems, the main

technological levers that guide research

actions fall essentially into three categories.The first concerns the reliability of the sys-tems. In the short and long term, it's one ofthe most effective means of action: itconcerns, for example, reducing the fre-quency of maintenance in isolated sites orconferring services lives on all of the com-ponents that are as long as those of the pho-tovoltaic modules. This requires a betterknowledge of the reliability of the systemand the installation of monitoring and userassistance devices, but also a certain stan-dardization of the components and anincreasingly important quality assuranceand normative approach. One part of thestudies undertaken by the Cadarache Groupfor new energies (Genec) aims at the ela-boration of international standards for sys-tem testing.

The second lever is the management ofenergy flows: in the medium term this solu-tion seems the most promising. It consists inoptimizing the use of the energy flows fromall of the components, by attempting to havethe photovoltaic modules producing as nearas possible to their maximum power, byhaving better knowledge of battery mana-gement strategy, by optimizing the associa-tion of energy sources in order to favor theuse of renewable energies, all while corres-ponding to an economic optimum. Theimplementation of "predictive" managementmethods, meaning that they forecast thefuture state of the resources or the system,currently makes up one of the importantdirections for the studies undertaken atGenec.

The third lever acts on the overall archi-tecture of the system: it's certainly the mostinnovative point that will develop productswhose architecture will be very differentfrom the first systems. This trend is veryobvious for "grid-connection" type appli-cations. Increasingly diversified productsare being developed to allow a more simpleand modular integration into buildings: rooftiles, slate or "solar" roofing, semi-trans-parent façade or roof windows and evenmultifunctional components simultaneouslyensuring several functions, such as themechanical rigidity of the building, insula-tion, solar protection, capture of thermalenergy and generation of photovoltaic elec-tricity. But this is also the case for the otherapplications where the technical answerswill allow the final product to better meetthe requirements, or even to find new appli-cations.

If the improvements to the complete sys-tems are mainly linked to the technologicalinnovation process, the gains forecasted inthe field of electrochemical storage – likethe photovoltaic conversion examined ear-lier – are in the field of competence of up-stream research.

36

SOLAR ENERGY

A second home in the center ofFinland supplied with electricity

by a photovoltaic systemdelivering approximately 5.6 kWhper day in summer. The roof-top

modules also ensure thefunctions of roofing and

insulation.

Neste Oy NAPS

Batteries undergoingaccelerated testing at Genec,CEA/Cadarache.

CEA/GENEC

37

The storage of the electricityof photovoltaic origin

Stand-alone photovoltaic generatorsrequire electric storage to ensure a uninter-rupted energy supply regardless of the meteo-rology. According to the applications, thissupply will be ensured, for example, for twoto three days for certain small domestic sys-tems and for approximately two weeks forprofessional applications, such as lighthousesand maritime buoys.

Different types of batteries

The different types of applications requiredifferent types of batteries to guarantee theservice provided to the end-user. Fromamong these technologies, the lead acid bat-tery, despite having been around for over onehundred years, currently, and for many yearsto come, offers the best answer in terms ofprice and service lifetime. Certain sites wherethe operating or climatic constraints are par-ticularly severe may be equipped with nickel-cadmium batteries, but their prohibitive costprevents the generalization of their use. Thenew pairs (lithium-ion, lithium hydride metal)present interesting solutions for low capa-city portable applications, but they are alsotoo expensive (box F, Storage batteries, cellsand batteries: steadily improving perfor-mance).

Experiments are being performed in cer-tain northern countries (Germany, Finland) touse fuel cells as a generator associated withconventional storage. In this case, the sto-rage is ensured by hydrogen reserves pro-duced by the electrolyzers supplied from thephotovoltaic generators. This type of inter-seasonal storage does not self-discharge. Ifthe prices were sufficiently competitive tocompensate for the current low efficiency ofthis technology, it would resolve the pro-blems at our latitudes due to the variation inelectric production between summer andwinter.

The stand-alone systems will use startertype flat-plate lead acid batteries for the ins-tallations with peak installed capacity approa-ching one hundred watts. The most impor-tant installations will be equipped withtubular batteries more adapted to a dailycycle, but with a cost per kilowatt-hour 1.5to 2 times higher. This type of battery equipsinstallations from several hundred watts toseveral kilowatts peak and all the professio-nal applications for reliability and safety rea-sons (television and telecommunication relaystations and maritime lighthouses). The sea-led lead acid battery is essentially used inrestrictive environments that only allow veryinfrequent maintenance, such as for theequipment in maritime buoys and in confinedinstallations.

Storage in a photovoltaic system contri-butes a non-negligible amount to the totaloperating cost due to its successive repla-cements during the service lifetime of asystem. Indeed, according to the techno-logy and use of lead acid batteries, theirservice lifetime may vary from 2 to 15years. In addition, the total storage costdoes not follow the same decrease as theone obtained on the other components of aphotovoltaic system. One of the currentobjectives is to double the service lifetimeof low-priced batteries with a design closeto that of starter batteries and to prolong to20 years the service lifetime of positivetubular stationary industrial batteries. Inorder to do this, industry level research isaimed at the design of new products thatare more adapted to the constraints of pho-tovoltaic applications.

Better management ofbattery lifetime

A second work direction concerns theimprovement of battery management sys-tems: this concerns conserving them whilemaking them work in less constraining statesof charge. These improvements require a bet-ter knowledge of the degradations observedon-site. The work currently taking place atGenec, within the scope of the contracts withthe Ademe, EDF, the European Commissionand industries, allows the identification and

the study of the relevant parameters that areat the origin of the degradation mechanisms.

The degradations observed on batteriesreturning from the field are essentially hardsulfation, the loss of active matter cohesionand, to a lesser extent, the corrosion of thegrids. To successfully complete these works,the laboratory is equipped with charge-discharge cycling, electrochemical studyand optical and chemical characterizationmeans. The knowledge acquired will allowthe modeling of these mechanisms and theirintegration into adaptive management algo-rithms that will evolve according to theconstraints actually sustained by the bat-tery subject to its specific operating condi-tions. The battery will then be managedaccording to its own state, its own beha-viour and no longer according to parame-ters pre-established during its installation,which will strongly contribute to the impro-vement of the service provided to the end-user and, in the long run, of the storage ser-vice lifetime. ●

Pascal Boulanger and Daniel DesmettreGenec

Technological Research DivisionCEA/Cadarache

CLEFS CEA - N° 44 - WINTER 2000-2001

Tests on a prototype battery atGenec.

CEA/GENEC