Thinning of CIGS solar cells Part II Cell characterizations

Thinning of CIGS solar cells:Part II:Cell characterizationsZ.Jehl a ,F.Erfurth a ,N.Naghavi a ,⁎,L.Lombez a ,I.Gerard b ,M.Bouttemy b ,P.Tran-Van b ,A.Etcheberry b ,G.Voorwinden c ,B.Dimmler c ,W.Wischmann d ,M.Powalla d ,J.F.Guillemoles a ,D.Lincot aaInstitut de Recherche et Développement sur l'Energie Photovoltaïque (IRDEP-UMR 7174EDF-CNRS-ENSCP),6quai Watier,78401Chatou Cedex,France bILV-UMR 8180,Universitéde Versailles St Quentin,45Av.des Etats Unis,78035Versailles Cedex,France cWuerth Elektronik Research GmbH,Industriestr.4,70565Stuttgart,Germany dZentrum für Sonnenenergie-und Wasserstoff-Forschung (ZSW),Industriestr.6,70565Stuttgart,Germanya b s t r a c ta r t i c l e i n f o Available online 15January 2011Keywords:Chalcopyrite CIGSe Solar cellsUltra-thin films EtchingIn this paper,the in fluence of reducing the thickness of the CIGSe absorber layer by bromine etching from 2.5μm to 0.5μm on electrical and optical solar cell properties is addressed.We observe a decrease in ef ficiency which is mainly caused by a reduced short circuit current,whereas the fill factor and the open circuit voltage are stable.Even without deliberate light trapping or anti-re flection coating,an ef ficiency of 10.3%is obtained for a 0.5μm thick CIGSe absorber.A smoothing of the absorber surface is observed during the etching,its in fluence on the cell parameters will be discussed.©2011Elsevier B.V.All rights reserved.1.IntroductionDuring the last decade increased attention has been paid to the preparation of very thin absorber layers in solar cells based on Cu(In,Ga)(S,Se)2(CIGSSe)[1–4].The reduction of the absorber layer thickness has been identi fied as a key issue for a further improvement of the competitiveness of the CIGSSe technology,which will impact two aspects:firstly the reduction of the consumption of indium per Wp and thus the costs and availability for large production volumes;secondly the increase of the production speed for the CIGSSe deposition and thus the reduction of equipment costs.Representative results have already been obtained at Uppsala University [1]and at the University of Delaware and the National Renewable Energy laboratory (NREL)in USA [2–4].The approach presented in this paper is to start from the standard glass/Mo/CIGSe structure and to thin the absorber layer by etching from the top using a chemical bromine solution.From this it will be possible to engineer thinner CIGSe layers while keeping exactly the same back contact con figuration.The effect of thinner layers,down to 0.5μm obtained by this methodology on solar cell properties will be addressed,allowing to gain very important information on the thickness dependence of key device parameters (photovoltaic parameters,recombination processes,and optical losses)while keeping a standard front and back side process.The advantage of the thinning approach is to overcome the dif ficulties arising from roughness variations when dealing with the deposition of absorbers with different thicknesses.It also gives a guaranty that the back contact interface is the same in all cases which might not be the case when varying the deposition time.This method leads to flat and specular absorber surfaces with easily controlled layer thickness [5].This can permit a good quanti fication of optical losses in thin CIGSe solar cells.The effect of thinning the absorber layer down to 0.5μm on the electrical and optical solar cell properties is addressed.A good understanding of the thinning CIGSe properties will make it easier to design suitable front and back contacts for further cell improvement separately from the absorber formation.2.ExperimentalThe investigated Cu(In,Ga)Se 2absorbers (CIGSe)were deposited on Mo covered glass by coevaporation at Wurth Solar [6].The absorber layers with a standard thickness of 2.5μm were chemically etched using an aqueous solution of HBr/Br 2.This method is described in detail by Bouttemy et al.[7].By increasing the etching time stepwise from 4min to 15min we obtained layers with a nominal layer thickness of 2μm,1.5μm,1μm,700nm,and 500nm,respectively,estimated by atomic absorption spectroscopy during the etching process.The electrical properties of cells were characterized by current voltage measurements at 25°C under illumination (AM 1.5global spectrum).Individual cells of 0.1cm 2were delimited by mechanical scribing.Absolute spectral response measurements were made with a monochromator (Spectral Products CM 110)under chopped illumination and a lock-in technique.The surface morphology of the samples was investigated by scanning electron microscopy (SEM)using a Leo Supra 35field emission gun (FEG).The AFM images were obtained with a D3100microscope and Nanoscope IIIa controller,using contact mode with DNP-20tips.The optical re flectivity was measured in the range of 250to 2500nm using a Perkin-Elmer Lambda 900UV/VIS/NIR spectrometer.Photoluminescence measurements with an excitation energy of 514nm were applied using a homemade setup.Thin Solid Films 519(2011)7212–7215⁎Corresponding author.E-mail address:negar.naghavi@edf.fr (N.Naghavi).0040-6090/$–see front matter ©2011Elsevier B.V.All rights reserved.doi:10.1016/j.tsf.2010.12.224Contents lists available at ScienceDirectThin Solid Filmsj o u r n a l h o m e p a g e :w w w.e l s ev i e r.c o m /l o c a t e /ts f3.Results and discussionThe SEM pictures in Fig.1show the cross sections of the initial absorber surface and of two etched absorber layers with a nominal thickness of 2000nm and 500nm,respectively.In addition the cross sections of the same samples are shown after the deposition of the standard CdS/i-ZnO/ZnO:Al layer structure.In contrast to the rough surface of the as-deposited absorber,the surfaces of the etched samples are much smoother.A strong surface modi fication already occurs within the first minutes of etching,as can be seen for the 2000nm sample.With increasing etching time a further smoothing of the surface is observed.On top of all etched absorbers a very thin super ficial layer is observed whose origin is not clear.It seems to be caused by the etching process and might be related to the formation of Se 0-enrichment at the surface [7],but further experiments are needed to verify the nature of this thin layer.The smoothing effect of the HBr/Br 2etching on CIGSe absorbers is con firmed by AFM measurements.Fig.2shows the root mean scare (RMS)parameters of the surface roughness on a scanning area of 50×50μm 2applied on a supplemental sample series as a function of the absorber thickness.The RMS decreases from about 220nm for a non etched 2500nm thick CIGSe sample down to 75nm for a 500nm thick CIGSe absorber.This con firms that the roughness of the absorber surface is clearly reduced,especially within the first minutes of etching (down to 2μm thickness)as was also observed on SEM pictures (Fig.1).The strong decrease of the roughness at the beginning of the etching declines with increasing etching time.The impact of the surface smoothing on the re flectivity of the absorber samples is illustrated in Fig.3a,showing the total re flectivity of three absorber films with a thickness of 2500nm (non-etched),1500nm and 500nm,respectively.As the thickness is reduced interference fringes emerge in the IR range.In this wavelength range the light is not absorbed by the CIGSe material.The appearance of these fringes is caused by the smoothing of the absorber surface providing a more homogeneous thickness of the absorber film,which thus acts as interference layer.The as deposited reference CIGSe sample does not exhibit any interference fringes due to its high texturation.In the CIGSe absorption range (i.e.wavelength between 1100nm and 400nm),the total re flectivity increases with increasing etching time.The enhancement of the re flectivity for the 500nm sample represents a relative loss of light intensity penetrating the absorber of about 25%compared to the as-deposited non etched CIGSe sample.When cells are completed by deposition of the standard CdS/i-ZnO/ZnO:Al layer structure,the difference in re flectivity between standard and etched CIGSe is strongly reduced (Fig.3b).This pseudo anti-re flecting effect occurs due to the improved adaptation of the optical index at the interfaces air/ZnO/CIGSe compared to the interface air/CIGSe,as the optical index of ZnO is about 2and of CIGSe is about 2.9[8].The ZnO window layer causes additional interference fringes even in the CIGSe absorption range,which affect the effective light intensity penetrating the absorber bulk.In the non-absorption range these fringes superpose with the interference fringes originating from the absorber layer,giving a non-periodic fringe structure.To investigate the impact of the etching on the solar cell parameters,we performed J(V)measurement under illumination using standard conditions (AM 1.5,25°C).The obtained photovoltaic parameters are shown in Fig.4.The fill factor remains roughly constant for all samples,which indicates the good quality and homogeneity of the material throughout the thickness.The open circuit voltage slightly increases with increasing etching time,which can be attributed to the grading of the band-gap inside the absorber:The band-gap of the absorber surface is smaller than the bulk value.When the “original ”absorber surface is removed,the band-gap of the “new ”surface increases,approaching the bulk value with increasing etching time.This interpretation is con firmed by the results of photoluminescence (PL)measurements,which are shown in Fig.5.The peak position corresponds to the band gap energy within the investigated layer.For CIGSe the penetration depth of this method is about a few 100nm.Hence,the observed shift of the peak with increasing etching time to higher energies mainly indicates an increase of the band-gap within this surface region.The shift of a few 10meV between the non-etched absorber and the 500nm sample matches very well with the increase of V oc between both samples.The fact that V oc does not decrease with reduced thickness is a strong advantage of this etching technique.In comparison Ramanathan etal.Fig.1.SEM images (cross sections,65°tilt)of standard non etched (2500nm)and etched (2000nm and 500nm)CIGSe samples without and with CdS/ZnO/ZnO:Al windows layer.7213Z.Jehl et al./Thin Solid Films 519(2011)7212–7215showed that the absorber layer thickness has a strong impact on V oc for directly grown thin CIGSe films.A strong degradation in ef ficiency by about 50%for sub-micron CIGSe absorber films was reported [4].Thus even for cells with an absorber layer thickness of only 500nm we still achieve ef ficiencies of 10.3%(see Fig.4).The observed drop of the ef ficiency with decreasing layer thickness is mainly caused by the almost linear decrease of J sc from 28.1mA cm −2for the reference absorber (2500nm)to 20mA cm −2for the 500nm sample.This decrease can be attributed to both,the reduction of the CIGSe absorber thickness and the smoothing of the absorber surface.The latter enhances the surface re flectivity of the etched cells and thus diminishes the absorption of light.This correlation is nicely demonstrated in Fig.6,showing external quantum ef ficiency (EQE)measurements.The drop of the short circuit current with decreasing layer thickness is represented by the overall decrease of the quantum ef ficiency.First estimations indicate that this decrease even in the short wavelength range at about 550nm is mainly caused by reduced absorption due to the thinning of the absorber layer.In contrast to the results of Lundberg et al.[1]we do not consider possible recombination at the back contact to in fluence the spectralresponse of the solar cells discussed in this paper.Additionally,interference fringes can be observed for the etched CIGSe samples originating from the ZnO layer.The positions of the maxima correspond to the positions of the minima of the re flective curves (compare Fig.3b),and vice versa.Fig.6also shows EQE spectra which are corrected for the re flectivity losses (dotted lines)compared to the non-etched sample (see Fig.3b).Therefore the EQE data was multiplied by the adjusted ratio of the re flectivity data of the standard absorber R standard and the etched sample R etched (1−R standard )/(1−R etched ).These data show the direct impact of the absorber layer thinning on the EQE,the interference fringes of the ZnO are vanished.In the near infrared wavelength range (800nm –1000nm),the decrease in quantum ef ficiency between the non-etched and the 500nm sample is stronger than the decrease in the wavelength range of 500nm –700nm.This behavior cannot only be explained by the higher penetration depth of light for higher wavelengths.The loss in J sc only caused by the reduction of light absorption due to the absorber thinning between a 1500nm and 500nm thick absorber layer was estimated to be smaller than 10%,taking the (constant)absorption of the ZnO layer and a global AM 1.5solar spectrum into account.The observed decrease in J sc shown in Fig.4Fig.3.a)Total re flectivity for three different thick (2500nm,1500nm and 500nm)CIGSe absorbers.2500nm CIGSe absorber is non etched with natural surface texturation.1500nm and 500nm etched samples are smooth.b)Total re flectivity of complete solar cells for the same samples.(The kinky structure at about 900nm is an artefact caused by the re flectivity measurementsetup.)Fig.4.Photovoltaic parameters for prepared cells with different CIGSe thickness.The plotting direction is with decreasing absorber layerthickness.Fig. parison of normalized photoluminescence signal of CIGSe layers with thickness varying from 2500nm down to 500nm.Fig.2.Measured RMS versus CIGSe thickness after etching.Thinner absorber samples are smoother caused by the longer etching time.7214Z.Jehl et al./Thin Solid Films 519(2011)7212–7215is much larger,which cannot be explained by optical losses,since the re flectivity data of both samples is almost identical (see Fig.3b).Hence,the strong decrease of J sc indicates additional electronic losses,which could be caused by enhanced back contact recombinations.This interpretation is supported by the strong decrease of the EQE in the high wavelength range.Higher wavelengths create more electron hole pairs in the volume and near the back side of the absorber,which than can recombine at the back contact.The position of the onset of the graph at the low energy side indicates the band-gap of the absorber material.The shift to higher energies is thus caused by the already discussed increase of the surface band-gap with decreasing CIGSe film thickness.4.ConclusionWe succeeded in etching CIGSe absorber layers with a standard thickness of 2.5μm down to 500nm without altering the material quality.Ef ficiencies greater than 10%for an absorber layer thickness of 500nm have been achieved.The ef ficiency losses compared to the non-etched absorber can be assigned to the reduction of J sc ,which is caused by the thinning of the absorber material and by a slight enhancement of re flectivity.This enhancement is caused by the smoothing of the absorber surface during the etching,which was observed by SEM and AFM characterization.Furthermore we reported the increase of the surface band-gap induced by etching,which was observed in EQE as well as in photoluminescence measurements.The value of a few 10meV nicely corresponds to the enhancement of V oc for the etched absorber samples.AcknowledgementThis work is supported by the ANR's HABISOL program within the ULTRACIS project.References[1]O.Lundberg,M.Bodegard,J.Malmstrom,L.Stolt,Prog.Photovoltaics 11(2003)77.[2]W.N.Shafarman,R.X.S.Huang,S.H.Stephens,Conference Record of the 2006IEEE4th World Conference on Photovoltaic Energy Conversion,1–2(2006)420.[3]K.Ramanathan,J.C.Keane, B.To,R.G.Dhere,R.Nou fi,Proc.20th EuropeanPhotovoltaic Sol.Energy Conf..Barcelona.Spain.(2005)1695.[4]K.Ramanathan,R.Nou fi,B.To,D.L.Young,R.Bhattacharya,M.A.Contreras,R.G.Dhere,G.Teeter,4th World Conference on Photovoltaic Solar Energy Conversion,2006,p.380,Hawaii.[5] B.Canava,J.F.Guillemoles,J.Vigneron,D.Lincot,A.Etcheberry,J.Phys.Chem.Solids 64(2003)1791.[6]M.Powalla,M.Cemernjak,J.Eberhardt,F.Kessler,R.Kniese,H.D.Mohring,B.Dimmler,Sol.Energy Mater.Sol.Cells 18–19(2006)3158.[7]M.Bouttemy,P.Tran-Van ,I.Gerard,A.Causier,A.Etcheberry,J.L.Pelouard,Z.Jehl,N.Naghavi,G.Voorwinden,B.Dimmler,M.Powalla,J.F.Guillemoles,D.Lincot,Thin Solid Film (unpublished).[8] A.Campa,J.Krc,J.Malmström,M.Edoff,F.Smole,M.Topic,Thin Solid Films 515(2007)5968.parison of external quantum ef ficiency between non etched (2500nm thick)and etched (1500nm and 500nm thick)CIGSe based solar cells.The dashed lines show the EQE data after correction of the re flectivity losses caused by surface smoothing.(The kinky structure at about 900nm for the “1500nm ”sample is an artefact caused by the re flectivity measurement setup.)7215Z.Jehl et al./Thin Solid Films 519(2011)7212–7215。