Earth-abundant absorber based solar cells onto low weight stainlesssteel substrate

Simón López-Marino a, Markus Neuschitzer a, Yudania Sánchez a, Andrew Fairbrother a,Moisés Espindola-Rodriguez a, Juan López-García a, Marcel Placidi a,Lorenzo Calvo-Barrio b,c, Alejandro Pérez-Rodríguez a,c, Edgardo Saucedo a,n

a Catalonian Institute for Energy Research (IREC), Jardin de les Dones de Negre 1, 08930 Sant Adrià del Besòs, Spainb Centres Científics i Tecnològics de la Universitat de Barcelona (CCiTUB), LLuís Solé i Sabarís 1-3, 08028 Barcelona, Spainc Departament de Electrònica (IN2UB), Universitat de Barcelona, Martí i Franquès 1, 08028 Barcelona, Spain

a r t i c l e i n f o

Article history:Received 20 December 2013Received in revised form16 July 2014Accepted 17 July 2014

Keywords:Cu2ZnSnSe4Stainless steelSemiconductorsThin filmsSolar cells

a b s t r a c t

In this paper we demonstrate the potentiality to produce solar cells based in earth abundant materialsdeposited onto low weight and flexible stainless steel substrates. Due to the increasing interest ofkesterite absorbers like those formed by Cu2ZnSn(S,Se)4, we deposited Cu2ZnSnSe4 layers by a two stagemethod onto stainless steel substrates coated with a Chromium diffusion barrier, a Mo back contact anda ZnO intermediate layer as substrate configuration, obtaining photovoltaic grade absorbers. Firstpreliminary results, using thermal treatments optimized for glass substrates, lead to a 3.5% efficiencysolar cell. This is a promising result, showing the potentiality of this low cost technology for thedevelopment of devices onto flexible and low weight substrates, compatible with roll-to-roll industrialimplementation.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

Thin films photovoltaic devices deposited onto low weight andflexible substrates, is a very promising way to reach the grid paritybetween the electricity produced by solar modules and classicalnon-renewable energies [1–3]. It brings the advantages to provide“green energy” at low prices, with the possibility of being easilyintegrated in a wide type of buildings and structures. Also, flexiblesubstrates can be used in roll-to-roll processes, which are extre-mely attractive from the fabrication point of view, allowing for acontinuous and cheap production of solar modules [1]. Recently,very encouraging results have been obtained in the developmentof CuIn1�xGaxSe2 (CIGS) based solar cells onto low weight sub-strates, in particular deposited onto polymers and metal foils, withefficiencies in the same order of the best results obtained for thetechnology developed onto classical soda lime glass substrate(SLG) [2–5]. Considering that the production throughput is usuallyhigher for flexible substrates, their advantages are clear.

Among the possible flexible substrates, CIGS solar cells have beenprepared onto polymers [2,3,6–8], stainless steel [5,9–11], aluminum[12], zirconia sheets [13], etc. Within these options stainless steel is

clearly one of the most promising because it fulfils all the require-ments necessaries to be a good substrate attractive for the PVindustry: good chemical, mechanical and thermal stability, widelyused in the industry, relatively cheap material, easily available in thinfoils (50 to 300 μm) and with very low surface roughness. As waspreviously highlighted, efficiencies obtained with these substrates(principally stainless steel and polymers) are in the order of thoseobtained on classical SLG substrates. Two particular issues have to beaddressed in order to allow for high efficiency values: the incorpora-tion of Na and the development of chemical barriers to avoid thediffusion of impurities from the substrate. The importance of theincorporation of Na in this type of technology has been extensivelyanalyzed and in general, a NaF layer deposited by e-beam at the backor front region of the absorber is used to control the concentration ofthis important element [3,6,7,14]. Additionally, different types ofbarriers have been employed with success, like: Al2O3 [15], ZnO [16],SiO2 [17] and Cr [18]. The effectiveness of all these materials has beendemonstrated, and advantages from one to other are not clear, but Crhas the advantage of matching better the thermal expansioncoefficient of Mo (for example: CTEMo 4.8�10�6 K�1, CTECr4.9�10�6 K�1, CTEAl2O3

6–8�10�6 K�1, CTESiOx 1–9�10�6 K�1)[19,20].

It is important to note that CIGS based technology has as amajor drawback the In and Ga scarcity, which can hinder thefuture mass deployment of photovoltaic modules based on this

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/solmat

Solar Energy Materials & Solar Cells

http://dx.doi.org/10.1016/j.solmat.2014.07.0300927-0248/& 2014 Elsevier B.V. All rights reserved.

n Corresponding author. Tel.:þ34 933562615.E-mail address: esaucedo@irec.cat (E. Saucedo).

Solar Energy Materials & Solar Cells 130 (2014) 347–353

material [21]. In this sense, many efforts are focused on thedevelopment of new earth abundant and low toxic materials,which can meet the increasing demand of electricity producedwith solar energy. Among the absorbers under study, a family ofcompounds generally called kesterites (Cu2ZnSn(S,Se)4 –CZTSSe–,Cu2ZnSnSe4–CZTSe–and Cu2ZnSnS4–CZTS–), are attracting a con-siderable interest, because they fulfill all the properties required tobe a good photovoltaic absorber: direct band gap with valuesbetween 1.0 eV and 1.50 eV, high light absorption coefficient andp-type conductivity [20–25]. Also, these materials are formed onlyby earth abundant and/or low toxic elements. Photovoltaic devicesproduced with these materials have shown so far efficienciesexceeding 10% onto classical Glass/Mo substrates, becoming a realalternative to the most mature CIGS and CdTe technologies [22].Typically, CZTSSe has to be prepared under Zn-rich and Cu-poorcomposition to obtain high efficiencies as one of the mostremarkable peculiarities of this family of materials [22–27].

So far, kesterite solar cells have been prepared onto classicalsoda lime glass substrates, covered by Mo as back contact, andusing the same technology developed for its close cousin material:CIGS [22–27]. But recently, a detrimental reaction between kester-ite and Mo has been demonstrated questioning the viability of thistype of substrate [27]. Nevertheless, a facile solution has beenproposed based in the use of a very thin intermediate layerbetween the Mo and CZTS, which prevents the contact betweenboth materials and in consequence the decomposition reaction[28]. As intermediate layer, a 10 nm in thickness ZnO film seems tobe enough to prevent the decomposition. In this work we presentan extension of the CZTSSe based technology to flexible and lowweight substrates, demonstrating its higher potentiality for theproduction of low cost, high throughput and eco-friendly devices.We implement a special back contact configuration including Cr asdiffusion layer for the substrate, Mo as back contact and ZnO asintermediate layer, in order to avoid the decomposition of kester-ite when is in direct contact with Mo as was previously commen-ted [28]. By using a two stage process consisting in the depositionof metallic stacks by DC-magnetron sputtering, followed by areactive thermal annealing under Se or S atmosphere, we obtaina photovoltaic grade material with efficiency exceeding 3% in afirst optimization of the complete device.

2. Material and methods

Austenitic stainless steel (SS) foils (0.05 mm in thickness) wereused as substrate. The substrates were mechanically polished toobtain a very smooth surface (average roughness lower than20 nm), and then carefully cleaned with a basic soap and rinsedwith deionized water (18 MΩ). Immediately after, they weresubmitted to an ultrasonic bath cleaning process with the follow-ing organic solvent sequence: acetone, methanol and isopropylalcohol. The time of the ultrasonic treatment for each solvent was10 min at a temperature of 55 1C. Finally, the substrates wererinsed with deionized water (18 MΩ) and dried at 60 1C with anitrogen flux. Previous to the chemical barrier and back contactdeposition, they were submitted to an additional surface treat-ment using radiofrequency (RF) plasma (100 W, 2�10�3 mbar Arpressure, room temperature, 5 min). This last process does haveany impact neither on the surface roughness, nor in the surfacecomposition of the substrate.

As substrate, a three layers configuration was used: chemicalbarrier, back electrical contact and intermediate layer. The chemi-cal barrier was a thin Cr layer (220 nm in thickness calibrated ontoglass substrate by X-ray fluorescence – XRF Fisherscope XVD-),deposited by direct current magnetron sputtering (DC-Sp) (120 W,2�10�3 mbar of Ar, room temperature, 15 min of time

deposition). An 800 nm in thickness Mo layer is used as backelectrical contact and 10 nm in thickness ZnO as intermediatelayer. The details of the deposition of Mo and ZnO intermediatelayer were published elsewhere [28]. The substrate size is2.5�2.5 cm2 in area in all cases.

To synthesize the CZTSe absorbers, we employed a two stageprocess consisting in the deposition of metallic stacks followed bya reactive annealing process. The structure of the metallic stackprecursor was the following: Cu(3 nm)/Sn(333 nm)/Cu(140 nm)/Zn(220 nm); all the metallic layers were deposited by DC-Sp (seeRefs. [25,28] for detailed description of the deposition process).The thicknesses were selected in order to have the following finalcomposition, further confirmed by XRF using calibration samplesgrown onto glass: [Cu]/([Zn]þ[Sn])¼0.75 and [Zn]/[Sn]¼1.25. Theannealing process is carried out in a three zones tubular furnace,using a graphite box (23.5 cm3 in volume). Two different thermaltreatments have been investigated: (i) a one step thermal processat 450 1C during 45 min (heating ramp 20 1C/min, total Ar pressureof 1 mbar, 50 mg of Se, 5 mg of Sn, natural cooling down); (ii) atwo step thermal process consisting in a first treatment at 400 1Cduring 30 min (heating ramp 20 1C/min, total Ar pressure of1 mbar) and a subsequent second treatment at 550 1C during15 min (heating ramp 20 1C/min, total Ar pressure of 1 bar). Bothtreatments were sequentially carried out, using 50 mg of Se and5 mg of Sn, with a natural cooling down to room temperature. It isimportant to note that the lowest temperature, 450 1C, will lead tolower crystalline quality when compared with higher tempera-tures. Nevertheless, this temperature regime will certainly mini-mize diffusion of contaminants from the absorber and will alsoreduce the induced thermal stresses during the cooling process. Incontrast, the temperature of 550 1C will certainly lead to biggergrain size and higher crystalline quality, usually tight to highdevices efficiency, but as it was stated above, higher thermalstresses could affect the final quality of the absorber. Therefore,the investigation of these temperatures, which values (or similarto) are widely reported in the literature for kesterite synthesis, arean interesting subject of study for this work [25,28–31].

Se concentration is close to 50% for both annealing tempera-tures, as is already expected for this material.

The samples were characterized using a FEI NovaTM NanoSEM230 microscope, an atomic force microscopy AFM in tapping andcontact modes (Park Systems XE-100), X-ray diffraction using aBruker D8 Advance equipment and Auger spectroscopy (Phi 670scanning Auger nanoprobe).

The as-annealed layers are submitted to a chemical etching inacidic KMnO4 solution followed by a Na2S etching (see Ref. [30] fordetailed description of the etching process). Immediately after, andwith the aim to complete the solar cells, a CdS layer is deposited bychemical bath deposition (60 nm in thickness), i-ZnO (50 nm) andZnO:Al (450 nm, 25Ω□) both by pulsed DC-Sp. 3�3 mm2 cells arescribed using a micro-diamond scriber (MR200 OEG) and then theIV-dark and illuminated curves (AM1.5 illumination conditions)are obtained using a pre-calibrated Sun 3000 Class AAA solarsimulator from Abet Technologies. The spectral response is mea-sured in a pre-calibrated Bentham PVE300 system, allowing us toobtain the external quantum efficiency (EQE) of the cells.

3. Results and discussion

The deposition of photovoltaic grade absorbers onto nonclassical substrates (i.e. onto substrates different than SGL), pre-sent many challenges as was already investigated and demon-strated for other thin films technologies like CIGS and CdTe [1–17,32,33]. Among them, the surface status (roughness, composi-tion, cleaning), the thermal properties of the material, the possible

S. López-Marino et al. / Solar Energy Materials & Solar Cells 130 (2014) 347–353348

diffusion of the constituent elements of the substrate and theabsence of Na, are the main factors to be considered. Fig. 1 shows3D AFM images for comparison between the as-cleaned substrate(Fig. 1a) and after the deposition of the Cr/Mo layers (Fig. 1b). Theaverage surface roughness of the substrate is 10 nm, whereas afterthe deposition of the Cr/Mo bi-layer this parameter is slightlyreduced towards 9 nm. The morphology of the surface seems to bealso slightly changed observing a more polygonal shape of the Mograins in comparison with the substrate grains. In principle, themorphological properties of the surface of the austenitic stainlesssteel (SS) fit the requirements to make it a potential substrate forphotovoltaic applications [5,9–11]. At this level, it is expected thatthe substrate has a negligible impact on the morphology of theprecursors (the Cu/Sn/Cu/Zn metallic stack), which looks verysimilar than those reported onto classical glass substrate.

We have tested two different thermal processes (low �450 1Cand high �550 1C) for the formation of CZTSe, as it is described inthe experimental section. Fig. 2 shows the top and cross sectionalviews of the as grown layers. The top view of the absorberannealed at 450 1C shows a rough morphology with a typical“cellular” configuration of the grains, which exceed 1 μm in size.Conversely, the absorber annealed at 550 1C exhibit some“faceted” crystals suggesting a high crystallization degree andcorrelating with the higher annealing temperature. Nevertheless,

the observed morphology in both cases is remarkably differentthan those typical for films prepared onto SLG as substrate, wherepolygonal grains are observed with sizes exceeding 1–2 μm[28,29]. The different morphology can be in principle due to twomain different effects. First, the higher differences between thethermal expansions coefficients between CZTSe and the sub-strates. The values of the thermal coefficients for the substratesare 8.6�10�6 K�1 for SLG and 18.4�10�6 K�1 for austeniticstainless steel. The higher thermal expansion coefficient of stain-less steel with respect to SLG could introduce additional stressesduring thermal annealing, explaining the surface grain morphol-ogy differences. Additionally, it is important to point out that inthis work we are not using any Na source which is crucial in thecase of CIGS for a correct crystallization [3,6,7,14]. The absence ofNa in the system, could also explain at least in part the apparentlyless crystallization degree on the surface.

Differences between both samples are highlighted when layersare observed in cross sectional view (Fig. 2c and d). The layerannealed at 450 1C exhibits smaller grains with values rangingfrom 200 nm to 500 nm as is already expected because of thelower temperature, and 1.7 mm in thickness approximately. Forsamples annealed at 550 1C grains exceeding 1 mm in size areclearly observed with similar thicknesses, in agreement with thehigher temperature of the thermal process. Nevertheless, a

Fig. 1. Three dimensional AFM images of SS (a) and SS/Cr/Mo substrates.

Fig. 2. Top and cross sectional view of absorbers annealed at 450 1C ((a) and (c), respectively) and at 550 1C ((b) and (d), respectively). SEM parameters: (a) 10 kV, WD 7 mm,12 K X, (b) 5 kV, WD 6.6 mm, 15 K X, (c) 5 kV, WD 5.2 mm, 15 K X, (d) 10 kV, WD 7.7 mm, 12 K X).

S. López-Marino et al. / Solar Energy Materials & Solar Cells 130 (2014) 347–353 349

capping layer formed by smaller grains is observed on the surface,decorating the larger grains in agreement with the surfacemorphology analysis. The observed grain sizes and morphologiescorrespond well with those reported in the literature for layersgrown at similar temperatures suggesting that the substrate doesnot affect in large extent the grain growth process [22–25,27,29].Nevertheless, the capping layer seems to be characteristic ofsamples grown onto stainless steel, which origin is not clear.Although compositional and structural characterization suggeststhat is formed by Cu2ZnSnSe4, maybe a contribution of few Sn–Sequantities could also be present as we will show later. It isimportant to remark that the absorbers presented in the Figureswere submitted to a chemical etching for the removal of ZnSesecondary phases present on the surface as was published else-where [30], discarding the possibility that the capping layer isformed by this phase.

The surface morphology of the absorbers was also investigatedby AFM as is shown in Fig. 3. Apparently, the absorber annealed at550 1C seems to be more uniform. This was confirmed analyzingthe AFM Figures., obtaining the following average roughness forthe different absorbers: 114 nm (450 1C onto SS), 95 nm (550 1Conto SS) and 91 nm (550 1C onto SLG, with a precursor with thesame composition that produces solar cells with 6% efficiency,image not shown). The morphology and roughness of the surfaceis very similar than those observed for absorbers grown onto SLGat equivalent conditions, confirming the potentiality of theseprecursors for the preparation of solar cells.

In Fig. 4 the PXRD patterns of CZTSe absorbers annealed at450 1C and 550 1C, as well as of the flexible stainless steel substratewith and without Cr barrier layer and Mo back contact are shown.The peaks of the substrate are well preserved after the selenizationprocess, indicating that it is stable in the conditions used for the

thermal processes. Also, the peaks corresponding to the Cr layerare detected before and after etching (even if the Cr layer is verythin), confirming that the chemical barrier is stable. The peakscorresponding to the (1 1 0), (2 0 0) and (2 1 1) directions of theMo back contact are observed for the substrate as well as for theas-annealed samples indicating that the Mo is only slightlyaffected by the selenization. Nevertheless and as is expected, theintensity of the Mo peaks diminish after selenization, suggestingthat some MoSe2 was formed (in fact the diminution of theintensity of these peaks is more marked in the absorber selenizedat 550 1C). This was further confirmed since the broad diffractionfeatures at 2θ¼31.821 and 2θ¼56.101, corresponding to MoSe2(PDF 01-072-1420) are observed after annealing and increases inintensity with the annealing temperature.

The presence of CZTSe absorber is confirmed by XRD showingpolycrystalline growth, because all peaks corresponding to thiscompound could be identified using the ICDD PDF 01-070-8930reference diffractogram. Regardless the annealing temperature,the (1 1 2), (2 0 4) and (3 1 2)/((1 1 6) are the main peaks [34],suggesting a low impact of this parameter on the polycrystallinenature and texture of these layers. Furthermore, very similar XRDpatterns have been reported for absorbers grown onto SLG [35,36],suggesting a low impact of the substrate on the grain orientation.This is expectable if we take into account that regardless thesubstrate, CZTS grows in contact with Mo in all cases. Furthermore,the appearance of diffraction peaks at 2θ¼30.621, 2θ¼33.701suggests the formation of SnSe2, already observed in the surfaceof our samples due to the condensation of the SnSe2 in excess usedduring the thermal process to avoid the Sn evaporation [37]. Theorigin of the diffraction peak in the sample at 2θ¼48.321 is not yetclear. In summary, the structural properties of the CZTSe baseddevices seems to be very similar than those reported for classicaldevices deposited onto SLG substrates.

To confirm that the used substrate configuration is effective toavoid the diffusion of detrimental elements from the substratetowards the absorber layer, and that the Mo layer withstands thehard Se conditions, Auger spectroscopy analysis were carried outin both types of samples. This is important because the higherthermal conduction, thermal expansion coefficient and the thinnerthicknesses of the stainless steel substrate with respect to SLG, canimpact dramatically on the behaviour of the back contact relativeto the thermal processes. Fig. 5 shows the Fe, Cr, Mo, Se, Cu, Zn andSn Auger compositional profiles of the devices for absorbers

Fig. 3. AFM images in tapping mode of absorbers prepared onto SS at 450 1C(a) and 550 1C (b).

Fig. 4. PXRD pattern of CZTSe samples annealed at 450 1C and 550 1C, and flexiblestainless steel substrate with and without Cr barrier layer and Mo back contact. TheCZTSe peaks were indexed using ICCD PDF 01-070-8930. After annealing at 550 1Cdiffraction peaks at 2θ¼30.621, 2θ¼31.821, 2θ¼33.701, as well as at 2θ¼48.321 and2θ¼56.101 increase in intensity indicating a possible stronger selenization of theMo back contact and Cr barrier layer and/or formation of secondary phases.

S. López-Marino et al. / Solar Energy Materials & Solar Cells 130 (2014) 347–353350

prepared at 450 1C (Fig. 5a) and 550 1C (Fig. 5b). Starting from thesurface, and although we have employed the same precursor inboth cases different Cu, Zn, Sn and Se profiles were observed. Atlower temperatures a Zn-rich and Cu-poor surface is observed aswas already reported for the selenide compound at the sametemperature onto glass substrate [31]. Nevertheless, the cationdistribution is affected by the higher annealing temperature as isclear from Fig. 5b. Whereas for samples annealed at 550 1C thesurface tends to be Sn-rich and Zn-poor, the Zn-excess introducedin the precursor is clearly accumulated towards the back region,matching the increase of the Zn and Se concentration at thisregion.

Similar trends were observed for the S case, where theaccumulation of ZnS was observed and reported at temperatureshigher than 500 1C [38].

Comparing the Fe profile for both cases, it is evident that Cr actsas very effective barrier in the temperature range of 450–550 1Cfor Fe from the austenitic stainless steel, avoiding the possiblecontamination of the absorber with this element. As it is expect-able, the higher the temperature, the higher the Fe penetration inthe Cr layer. Nevertheless, this detrimental element is not detectednor in the Mo layer, neither in the CZTSe one. Furthermore, theoverlapping between the Cr and Mo profiles is minimum, con-firming the stability and usefulness of this barrier layer for thetechnology proposed in this paper. Concerning to the Se diffusion,it is evident the formation of a MoSe2 layer as was alreadyobserved by XRD, and the thickness of this layer increases withthe annealing temperature. Thus, the complex SS/Cr/Mo/ZnOstructure used in this work exhibit all the desirable characteristicsto be a good substrate from the point of view of morphology andstability at the required annealing conditions.

To confirm the potentiality of these layers as earth abundantbased solar cells onto low weight substrates, devices where preparedusing the classical configuration [25]. Fig. 6a and b presents theilluminated and dark I–V curves at AM1.5 conditions and spectralresponse, respectively, of typical solar cells prepared with absorbersannealed at 450 1C and 550 1C. First devices prepared with thistechnology give efficiencies between 2% and 3.5% and in Fig. 6a thecells with maximum efficiency corresponding to absorbers preparedat 4501 (Eff.MAX¼3.5%) and 550 1C (Eff.MAX¼3.2%) are presented.Additionally, the optoelectronic parameters extracted from Fig. 6aaccording to the Sites and Mauk method (see ref 39) are summarizedin Table 1. Interestingly, both absorbers lead to very similar devices,with almost the same Jsc, FF, Rs and Rsh, although the majordifferences observed from the morphological, cationic in depthdistribution and secondary phase presence. The main differencebetween them is the Voc, which is �10% lower for the absorberprepared at 550 1C, which could be related to the ZnSe accumulationat the back and/or the SnSe accumulation at the surface. Examiningthe parameters extracted from the I–V curves, it seems clear that adiode quality factor under illumination of 1.6–1.8 is related torecombination at the space charge region [40,41]. Furthermore, avery high value of the saturation current factor (J0), certainlyindicates that these devices are dominated by recombination issues[42]. Another important matter is the crossover effect between theilluminated and dark I–V curves observed in Fig. 6a. This is a clearindication of a deviation from the ideal diode behaviour, beingresponsible for a decrease in the FF and the high Rs of both devices.This non ideal effect could explain at least to some extent the

Fig. 5. Auger profiles of Fe, Cr, Mo and Se in samples prepared at 450 1C (a) and550 1C (b).

Fig. 6. I–V illuminated curves with the corresponding optoelectronic parameters (a) and E.Q.E. (b) of solar cells with absorbers annealed at 450 1C (red) and 550 1C (black).(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

S. López-Marino et al. / Solar Energy Materials & Solar Cells 130 (2014) 347–353 351

difference between the optoelectronic parameters obtained underdark and illuminated conditions.

Finally, the E.Q.E. plots of both devices are shown in Fig. 6b. Inboth cases, very similar spectra are obtained with a maximum E.Q.E. of 70% for 600 nm wavelength. This is consistent with the verysimilar values obtained for the JSC. For higher wavelengths, the E.Q.E. decreases and this could be related to a non-optimized junction,correlating with the observed capping layer formed by smallgrains (see Fig. 2). The steep decrease of the EQE plot from600 nm implies as well important losses in the minority carrierscollection for the spectral region corresponding to the bulk [41].Strong recombination at the grain boundaries due to the absenceof a beneficial passivation effect of the Na, could explain this EQEshape. Na is tied to a reduction of deep recombination centers andmakes the acceptors shallower, thereby improving with its pre-sence recombination issues at the bulk [43]. This recombinationphenomenon located at the bulk is in agreement with the previousanalysis of the optoelectronic parameters extracted from the IVcurves. The band-gap of the material is estimated in 1.0 eV for bothabsorbers in agreement with the expected value for CZTSe.

We believe that despite the morphological similarities betweenthe samples produced on SS substrates and glass, as stated before,the absence of Na in the SS based absorbers is clearly reducing thesolar cell efficiency when compared to glass based devices. Na hasa major role as dopant in grain boundary passivation, enhancingVoc and FF, and it is generally accepted that increases theefficiency in CIGS and CZTS technologies [43–49]. Furtherimprovements on the composition, thermal processes, junctionformation and Na doping of the absorber are under progress, withthe aim to improve the conversion efficiencies presented in thispaper, which are encouraging.

4. Conclusions

We demonstrate the potentiality of earth abundant absorbersonto low weight and flexible substrates based in a two stageprocess, which opens a new perspective in the development ofcost effective solar cells. The use of a Cr layer as chemical diffusionbarrier seems to be very effective to avoid the diffusion ofimpurities, mainly Fe from the SS substrate. Two different tem-peratures were tested to selenize the metallic stacks precursors,showing similar capabilities for the production of high qualityabsorbers. At low annealing temperatures (450 1C) we obtainlayers with smaller dense grain structure with rougher surfacethan those produced at higher temperatures (550 1C). Conversely,at lower annealing temperatures the structural properties of thesubstrate are less affected as it can be expected. Nevertheless, wedemonstrate that the substrate configuration used in this workallows working at temperatures as high as 550 1C. In a first roughoptimization of the whole process parameters, we obtain devices

with 3.0–3.5% maximum efficiencies independently on the anneal-ing temperature. Future works are focused on the substrate andthermal processes optimization, as well as in the analysis of theimpact of Na doping. We believe that the results presented here,are encouraging in terms of the development of this low cost andhigh throughput compatible technology.

Acknowledgements

This research was supported by the Framework 7 ProgramUnder the Project KESTCELLS (FP7-PEOPLE-2012-ITN-316488) andby European Regional Development Founds (ERDF, FEDER Pro-grama Competitivitat de Catalunya 2007–2013). Authors from IRECand the University of Barcelona belong to the M-2E (ElectronicMaterials for Energy) Consolidated Research Group and theXaRMAE Network of Excellence on Materials for Energy of the“Generalitat de Catalunya”. E.S. thanks the MINECO (Spain) for the“Ramon y Cajal” fellowship (RYC-2011-09212), V.I. for the “Juan dela Cierva” Fellowship (JCI-2011-10782), A.F. for the FPU-MINECO(FPU12/05508) and M.E-R. for the FPI-MINECO (BES-2011-045774).

References

[1] M. Pagliaro, R. Ciriminna, G. Palmisano, Flexible solar cells, ChemSusChem 1(2008) 880–891.

[2] A. Chirila, S. Buecheler, F. Pianezzi, P. Bloesch, C. Gretener, A.R. Uhl, C. Fella,L. Kranz, J. Perrenoud, S. Seyrling, R. Verma, S. Nishiwaki, Y.E. Romanyuk,G. Bilger, A.N. Tiwari, Highly efficient Cu(In,Ga)Se2 solar cells grown on flexiblepolymer films, Nat. Mater. 10 (2011) 857–861.

[3] R. Caballero, C.A. Kaufmann, T. Eisenbarth, T. Unold, R. Klenk, H.-W. Schock,High efficiency low temperature grown Cu(In,Ga)Se2 thin film solar cells onflexible substrates using NaF precursor layers, Prog. Photovoltaics Res. Appl. 19(2011) 547–551.

[4] A. Chirila, P. Reinhard, F. Pianezzi, P. Bloesch, A.R. Uhl, C. Fella, L. Kranz,D. Keller, C. Gretener, H. Hagendorfer, D. Jaeger, R. Erni, S. Nishiwaki,S. Buecheler, A.N. Tiwari, Potassium-induced surface modification of Cu(In,Ga)Se2 thin films for high-efficiency solar cells, Nat. Mater. 12 (2014)1107–1111.

[5] F. Pianezzi, A. Chirilă, P. Blösch, S. Seyrling, S. Buecheler, L. Kranz, C. FellaA.N. Tiwari, Electronic properties of Cu(In,Ga)Se2 solar cells on stainless steelfoils without diffusion barrier, Prog. Photovoltaics Res. Appl. 20 (2012)253–259.

[6] R. Caballero, C.A. Kaufmann, T. Eisenbarth, M. Cancela, R. Hesse, T. Unold,A. Eicke, R. Klenk, H.W. Schock, The influence of Na on low temperaturegrowth of CIGS thin film solar cells on polyimide substrates, Thin Solid Films517 (2009) 2187–2190.

[7] V. Izquierdo-Roca, R. Caballero, X. Fontané, C.A. Kaufmann, J. Álvarez-García,L. Calvo-Barrio, E. Saucedo, A. Pérez-Rodríguez, J.R. Morante, H.W. Schock,Raman scattering analysis of Cu-poor Cu(In,Ga)Se2 cells fabricated on poly-imide substrates: effect of Na content on microstructure and phase structure,Thin Solid Films 519 (2011) 7300–7303.

[8] M.G. Faraj, K. Ibrahim, A. Salhin, Fabrication and characterization of thin-filmCu(In,Ga)Se2 solar cells on a PET plastic substrate using screen printing, Mater.Sci. Semicond. Process. 15 (2012) 165–173.

[9] R. Wuerz, A. Eicke, M. Frankenfeld, F. Kessler, M. Powalla, P. Rogin, O. Yazdani-Assl, CIGS thin-film solar cells on steel substrates, Thin Solid Films 517 (2009)2415–2418.

[10] K. Moriwaki, M. Nomoto, S. Yuuya, N. Murakami, T. Ohgoh, K. Yamane,S. Ishizuka, S. Niki, Cu(In,Ga)Se2 solar cells and submodules using newlydeveloped structure metal foil substrate with a dielectric layer, Sol. EnergyMater. Sol. Cells 112 (2013) 106–111.

[11] R. Wuerz, A. Eicke, F. Kessler, S. Paetel, S. Efimenko, C. Schlegel, CIGS thin-filmsolar cells and modules on enamelled steel substrates, Sol. Energy Mater. Sol.Cells 100 (2012) 132–137.

[12] D. Brémaud, D. Rudmann, M. Kaelin, K. Ernits, G. Bilger, M. Döbeli, H. ZoggA.N. Tiwari, Flexible Cu(In,Ga)Se2 on Al foils and the effects of Al duringchemical bath deposition, Thin Solid Films 515 (2007) 5857–5861.

[13] S. Ishizuka, A. Yamada, K. Matsubara, P. Fons, K. Sakurai, S. Niki, Developmentof high-efficiency flexible Cu(In,Ga)Se2 solar cells: a study of alkali dopingeffects on CIS, CIGS, and CGS using alkali-silicate glass thin layers, Curr. ApplPhys. 10 (2010) S154–S156.

[14] P. Blösch, S. Nishiwaki, A. Chirila, L. Kranz, C. Fella, F. Pianezzi, C. Adelhelm,E. Franzke, S. Buecheler, A.N. Tiwari, Sodium-doped molybdenum backcontacts for flexible Cu(In,Ga)Se2 solar cells, Thin Solid Films 535 (2013)214–219.

Table 1Optoelectronic parameters from illuminated and dark I–V curves of 450 and 550 1Cabsorbers.

Tanneal¼450 1C Tanneal¼550 1C

Light Dark Light Dark

η (%) 3.5 3.2JSC (mA/cm2) 24.7 25.2VOC (mV) 302 270FF (%) 47.1 46.7Rs (Ω/cm2) 1.3 6.7 1.4 2.1Rsh (Ω/cm2) 72.3 293.3 57.4 226.2A 1.8 2.1 1.6 1.7J0 (mA/cm2) 3.3�10�2 6.4�10�2 2.1�10�2 3.5�10�2

S. López-Marino et al. / Solar Energy Materials & Solar Cells 130 (2014) 347–353352

[15] S. Gledhill, A. Zykov, N. Allsop, T. Rissom, J. Schniebs, C.A. Kaufmann, M. Lux-Steiner, C.-H. Fischer, Spray pyrolysis of barrier layers for flexible thin filmsolar cells on steel, Sol. Energy Mater. Sol. Cells 95 (2011) 504–509.

[16] C.Y. Shi, Y. Sun, Q. He, F.Y. Li, J.C. Zhao, Cu(In,Ga)Se2 solar cells on stainless-steelsubstrates covered with ZnO diffusion barriers, Sol. Energy Mater. Sol. Cells 93(2009) 654–656.

[17] T. Satoh, Y. Hashimoto, S.-i. Shimakawa, S. Hayashi, T. Negami, Cu(In,Ga)Se2solar cells on stainless steel substrates covered with insulating layers, Sol.Energy Mater. Sol. Cells 75 (2003) 65–71.

[18] C. Platzer-Björkman, S. Jani, J. Westlinder, M.K. Linnarsson, J. Scragg, M. Edoff,Diffusion of Fe and Na in co-evaporated Cu(In,Ga)Se2 devices on steelsubstrates, Thin Solid Films 535 (2013) 188–192.

[19] F. Kessler, Dominik Rudmann, Technological aspects of flexible CIGS solar cellsand modules, Sol. Energy 77 (2004) 685–695.

[20] ASM International Materials Properties Database Committee, F. Cverna,Thermal expansion in: ASM Ready Reference: Thermal Properties of Metals,ASM International, Ohio, 2002, pp. 09-220.

[21] C. Candelise, M. Winskel, R. Gross, Implications for CdTe and CIGS technologiesproduction costs of indium and tellurium scarcity, Prog. Photovoltaics Res.Appl. 20 (2012) 816–831.

[22] T.K. Todorov, J. Tang, S. Bag, O. Gunawan, T. Gokmen, Y. Zhu, D.B. Mitzi, Beyond11% efficiency: characteristics of state-of-the-art Cu2ZnSn(S,Se)4 solar cells,Adv. Energy Mater. 3 (2013) 34–38.

[23] B. Shin, O. Gunawan, Y. Zhu, N.A. Bojarczuk, S.J. Chey, S. Guha, Thin film solarcell with 8.4% power conversion efficiency using an earth-abundantCu2ZnSnS4 absorber, Prog. Photovoltaics Res. Appl. 21 (2013) 72–76.

[24] I. Repins, C. Beall, N. Vora, C. DeHart, D. Kuciauskas, P. Dippo, B. To, J. Mann,W.-C. Hsu, A. Goodrich, R. Noufi, Co-evaporated Cu2ZnSnSe4 films and devices,Sol. Energy Mater. Sol. Cells 101 (2012) 154–159.

[25] A. Fairbrother, E. García-Hemme, V. Izquierdo-Roca, X. Fontané, F.A. Pulgarín-Agudelo, O. Vigil-Galán, A. Pérez-Rodríguez, E. Saucedo, Development of aselective chemical etch to improve the conversion efficiency of Zn-richCu2ZnSnS4 solar cells, J. Am. Chem. Soc. 134 (2012) 8018–8021.

[26] S. Siebentritt, S. Schorr, Kesterites—a challenging material for solar cells, Prog.Photovoltaics Res. Appl. 20 (2012) 512–519.

[27] J.J. Scragg, J.T. Wätjen, M. Edoff, T. Ericson, T. Kubart, C. Platzer-Björkman, Adetrimental reaction at the molybdenum back contact in Cu2ZnSn(S,Se)4 thin-film solar cells, J. Am. Chem. Soc. 134 (2012) 19330–19333.

[28] S. López-Marino, M. Placidi, A. Perez-Tomas, J. Llobet, V. Izquierdo-Roca,X. Fontane, A. Fairbrother, M. Espindola-Rodriguez, D. Sylla, A. Perez-Rodriguez, E. Saucedo, Inhibiting the absorber/Mo-back contact decomposi-tion reaction in Cu2ZnSnSe4 solar cells: the role of a ZnO intermediatenanolayer, J. Mater. Chem. A 1 (2013) 8338–8343.

[29] A. Fairbrother, X. Fontané, V. Izquierdo-Roca, M. Espindola-Rodriguez,S. López-Marino, M. Placidi, J. López-García, A. Pérez-Rodríguez, E. Saucedo,Single-step sulfo-selenization method to synthesize Cu2ZnSn(SySe1�y)4 absor-bers from metallic stack precursors, ChemPhysChem 14 (2013) 1836–1843.

[30] S. López-Marino, Y. Sanchez, M. Placidi, A. Fairbrother, M. Espindola-Rodriguez, X. Fontane, V. Izquierdo-Roca, J. López-García, L. Calvo-Barrio,A. Perez-Rodriguez, ZnSe etching of Zn-rich Cu2ZnSnSe4: an oxidation routefor improved solar-cell efficiency, Chem. Eur. J. 19 (2013) 14814–14822.

[31] A. Fairbrother, X. Fontané, V. Izquierdo-Roca, M. Placidi, D. Sylla, M. Espindola-Rodriguez, S. López-Marino, F.A. Pulgarín, O. Vigil-Galán, A. Pérez-Rodríguez,E. Saucedo, Secondary phase formation in Zn-rich Cu2ZnSnSe4-based solar

cells annealed in low pressure and temperature conditions, Prog. Photovol-taics Res. Appl. 22 (2013) 479–487.

[32] A. Romeo, M. Terheggen, D. Abou-Ras, D.L. Bätzner, F.J. Haug, M. Kälin,D. Rudmann, A.N. Tiwari, Development of thin-film Cu(In,Ga)Se2 and CdTesolar cells, Prog. Photovoltaics Res. Appl. 12 (2004) 93–111.

[33] J. Perrenoud, B. Schaffner, S. Buecheler, A.N. Tiwari, Fabrication of flexible CdTesolar modules with monolithic cell interconnection, Sol. Energy Mater. Sol.Cells 95 (2011) S8–S12.

[34] S. Schorr, The crystal structure of kesterite type compounds: a neutron and X-ray diffraction study, Sol. Energy Mater. Sol. Cells 95 (2011) 1482–1488.

[35] P.M.P. Salomé, P.A. Fernandes, A.F. da Cunha, J.P. Leitao, J. Malaquias, A. Weber,J.C. González, M.I.N. da Silva, Growth pressure dependence of Cu2ZnSnSe4properties, Sol. Energy Mater. Sol. Cells 94 (2010) 2176–2180.

[36] P.M.P. Salomé, P.A. Fernandes, A.F. da Cunha, Morphological and structuralcharacterization of Cu2ZnSnSe4 thin films grown by selenization of elementalprecursor layers, Thin Solid Films 517 (2009) 2531–2534.

[37] A. Redinger, D.M. Berg, P.J. Dale, S. Siebentritt, The consequences of kesteriteequilibria for efficient solar cells, J. Am. Chem. Soc. 133 (2011) 3320–3323.

[38] A. Fairbrother, X. Fontané, V. Izquierdo-Roca, M. Espíndola-Rodríguez,S. López-Marino, M. Placidi, L. Calvo-Barrio, A. Pérez-Rodríguez, E. Saucedo,On the formation mechanisms of Zn-rich Cu2ZnSnS4 films prepared bysulfurization of metallic stacks, Sol. Energy Mater. Sol. Cells 112 (2013) 97–105.

[39] J.R. Sites, P.H. Mauk, Diode quality factor determination for thin film solarcells, Sol. Cells 27 (1989) 411–417.

[40] A. Polizzotti, I. Repins, R. Noufi, S.-H. Wei, D.B. Mitzi, The state and futureprospects of kesterite photovoltaics, Energy Environ. Sci. 6 (2013) 3171–3182.

[41] I. Repins, C. Beall, N. Vora, C. DeHart, D. Kuciauskas, P. Dippo, B. To, J. MannW.-C. Hsu, A. Goodrich, R. Noufi, Co-evaporated Cu2ZnSnSe4 films and devices,Sol. Energy Mater. Sol. Cells 101 (2012) 154–159.

[42] S. Siebentritt, Why are kesterite solar cells not 20% efficient? Thin Solid films535 (2013) 1–4.

[43] J.V. Li, D. Kuciauskas, M.R. Young, I.L. Repins, Effects of sodium incorporation inco-evaporated Cu2ZnSnSe4 thin-film solar cells, Appl. Phys. Lett. 102 (2013)163905.

[44] S. Siebentritt, M. Igalson, C. Persson, S. Lany, The electronic structure ofchalcopyrites—bands, point defects and grain boundaries, Prog. PhotovoltaicsRes. Appl. 18 (2010) 390–410.

[45] R. Caballero, C.A. Kaufmann, T. Eisenbarth, A. Grimm, I. Lauermann, T. Unold,R. Klenk, H.W. Schock, Influence of Na on Cu(In,Ga)Se2 solar cells grown onpolyimide substrates at low temperature: impact on the Cu(In,Ga)Se2/Mointerface, Appl. Phys. Lett. 96 (2010) 092104.

[46] P.M.P. Salomé, A. Hulqtvist, V. Fjallstrom, M. Edoff, B. Aitken, K. Vaidyanathan,K. Zhang, K. Fuller, C. Kosik Williams, Cu(In,Ga)Se2 solar cells with varying Nacontent prepared on nominally alkali-free glass substrates, IEEE J. Photovol-taics 3 (2013) 852–858.

[47] M. Ruckh, D. Schmid, M. Kaiser, R. Schäffler, T. Walter, H.W. Schock, Influenceof substrates on the electrical properties of Cu(In,Ga)Se2 thin films, Sol. EnergyMater. Sol. Cells 41–42 (1996) 335–343.

[48] C.M. Sutter-Fella, J.A. Stückelberger, H. Hagendorfer, F. La Mattina, L. Kranz,S. Nishiwaki, A.R. Uhl, Y.E. Romanyuk, A.N. Tiwari, Sodium assisted sintering ofchalcogenides and its application to solution processed Cu2ZnSn(S,Se)4, thinfilm solar cells, Chem. Mater. 26 (2014) 1420–1425.

[49] G. Altamura, C. Roger, L. Grenet, J. Bleuse, H. Fournier, S. Perraud, H. Mariette,Influence of sodium-containing substrates on kesterite CZTSSe thin filmsbased solar cells, MRS Online Proc. Lib. 1538 (2013) 103–106.

S. López-Marino et al. / Solar Energy Materials & Solar Cells 130 (2014) 347–353 353