# Fileset

[Bombsch_Adv_Funct_Mater2024.pdf](https://mdr.nims.go.jp/filesets/1d9bdc5f-bb3e-47eb-84e8-8e2efa1140f4/download)

## Creator

[Jakob Bombsch](https://orcid.org/0000-0002-0820-162X), [Tim Kodalle](https://orcid.org/0000-0002-8792-9669), [Raul Garcia‐Diez](https://orcid.org/0009-0000-9374-1083), Claudia Hartmann, Roberto Félix, [Shigenori Ueda](https://orcid.org/0000-0001-9425-0614), [Regan G. Wilks](https://orcid.org/0000-0001-5822-8399), [Christian A. Kaufmann](https://orcid.org/0000-0001-9168-2032), [Marcus Bär](https://orcid.org/0000-0001-8581-0691)

## Rights

Share — copy and redistribute the material in any medium or format for any purpose, even commercially.
Adapt — remix, transform, and build upon the material for any purpose, even commercially.[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Chemical Interface Structures in CdS/RbInSe<sub>2</sub>/Cu(In,Ga)Se<sub>2</sub> Thin‐Film Solar Cell Stacks](https://mdr.nims.go.jp/datasets/3c1bd446-17ca-43f9-b5bd-7e3bec0e6f1a)

## Fulltext

Chemical Interface Structures in CdS/RbInSe2/Cu(In,Ga)Se2 Thin‐Film Solar Cell StacksRESEARCH ARTICLEwww.afm-journal.deChemical Interface Structures in CdS/RbInSe2/Cu(In,Ga)Se2Thin-Film Solar Cell StacksJakob Bombsch, Tim Kodalle, Raul Garcia-Diez, Claudia Hartmann, Roberto Félix,Shigenori Ueda, Regan G. Wilks, Christian A. Kaufmann, and Marcus Bär*Performance-enhancing heavy alkali-based post-deposition treatments (PDT)of Cu(In,Ga)Se2 (CIGSe) thin-film solar cells absorbers often induce theformation of a Rb- In-Se phase on the CIGSe absorber. Co-evaporation of aninterfacial RbInSe2 (RISe) layer between buffer and absorber can also benefitcell performance. A detailed analysis of the chemical interface structures inCdS/RISe/CIGSe layer stacks is performed using hard X-ray photoelectronspectroscopy (HAXPES). For comparison, stacks without RISe and based onRbF PDT CIGSe absorbers are also studied. When aiming for the directco-evaporation of a RISe layer on the CIGSe absorber, the formation of anadditional In-Se phase is found. For the RbF PDT CIGSe absorbers, the studyonly finds small amounts of Rb and no indication for a RISe layer formation.Examining layer stacks prepared via additional chemical bath deposition(CBD) of CdS reveals a clear impact of the presence of Rb (or of Rb-containingspecies) on the CIGSe surface. In these cases, an increase of theinduction/coalescence period is found at the beginning of the CBD bufferlayer growth process and the formation of Cd─Se bonds; thereafter, a morecompact CdS layer growth is observed.1. IntroductionThe implementation of post deposition treatments (PDTs) basedon heavy alkali metal fluorides into the deposition process ofJ. Bombsch, R. Garcia-Diez, C. Hartmann, R. Félix, R. G. Wilks, M. BärDepartment Interface DesignHelmholtz-Zentrum Berlin für Materialien und Energie GmbH (HZB)12489 Berlin, GermanyE-mail: marcus.baer@helmholtz-berlin.deT. Kodalle, C. A. KaufmannPVcomBHZB12489 Berlin, GermanyS. UedaSynchrotron X-ray Station at SPring-8National Institute for Materials Science (NIMS)1-1-1 Kouto, Sayo, Hyogo 679-5148, JapanThe ORCID identification number(s) for the author(s) of this articlecan be found under https://doi.org/10.1002/adfm.202403685© 2024 The Authors. Advanced Functional Materials published byWiley-VCH GmbH. This is an open access article under the terms of theCreative Commons Attribution License, which permits use, distributionand reproduction in any medium, provided the original work is properlycited.DOI: 10.1002/adfm.202403685Cu(In,Ga)Se2 (CIGSe) absorber layers hasled to significant increases in device effi-ciency of resulting solar cells over the re-cent years.[1–3] One of the reported effectsof such PDTs is the formation of an alkali-In-Se phase, which was predicted for the al-kali metals K, Rb, and Cs[4] and has been ob-served in the cases of KF and RbF PDTs.[5–8]While the overall effect of optimized PDTsis positive, a too pronounced alkali-In-Sephase is frequently associated with an en-ergetic barrier having a detrimental impacton the fill factor (FF) in the case of RbFPDT.[9,10] To examine the distinct impactof the Rb-In-Se (RISe) phase on the solarcell performance, Kodalle et al. replaced theRbF PDT by a direct RISe co-evaporationof varying duration.[9] Despite a detrimen-tal impact on the FF for longer depositiontimes, an overall increase in device per-formance could be detected due to signif-icant gains in open circuit voltage (VOC).To understand this efficiency in-crease, we used hard X-ray photoelectronspectroscopy (HAXPES) to perform a detailed analysis of thechemical structure of RISe/CIGSe heterostructures manufac-tured in the same way as in ref [9]. For comparison, an untreatedS. UedaResearch Center for Electronic and Optical MaterialsNIMS, 1-1 Namiki, Tsukuba, Ibaraki 305-0044, JapanR. G. Wilks, M. BärEnergy Materials In-Situ Laboratory Berlin (EMIL)HZB12489 Berlin, GermanyM. BärDepartment of X-ray Spectroscopy at Interfaces of Thin FilmsHelmholtz-Institute Erlangen-Nürnberg for Renewable Energy (HI ERN)12489 Berlin, GermanyM. BärDepartment of Chemistry and PharmacyFriedrich-Alexander-Universität Erlangen-Nürnberg (FAU)91058 Erlangen, GermanyAdv. Funct. Mater. 2024, 2403685 2403685 (1 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbHhttp://www.afm-journal.demailto:marcus.baer@helmholtz-berlin.dehttps://doi.org/10.1002/adfm.202403685http://creativecommons.org/licenses/by/4.0/http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadfm.202403685&domain=pdf&date_stamp=2024-05-06www.advancedsciencenews.com www.afm-journal.deand a RbF PDT CIGSe absorber were also studied. In addition,we deposited a CdS thickness series on some of those samples toinvestigate the impact of the respectively altered surfaces on thebuffer layer growth.2. Results and DiscussionSurvey spectra of the differently treated CIGSe absorbers and theRISe reference[11] are shown in Figure S1 (Supporting Informa-tion). They indicate the presence of all CIGSe related elements onall samples except for the RISe reference, where no signs of Cuand Ga are visible. In addition, oxygen- and carbon-related sig-nals are observed, which might be attributed to surface contami-nants, which were either not removed by the NH4OH etching oradsorbed to the surface in the time between rinsing and measure-ment (e.g., when the samples were handled in ambient conditiondirectly prior to HAXPES measurements). However, the inclu-sion of some O and C into the bulk of the absorber (during depo-sition) is also a possible scenario. In addition, the alkali elementsNa and Rb are indicated to be present at all sample surfaces withtwo exceptions: The (control) untreated CIGSe absorber showsno indications for the presence of Rb, and the RbF PDT absorbersurface does not exhibit any sign for the presence of an apprecia-ble amount of Na, in agreement with previous findings of RbFPDT leading to Na free CIGSe surfaces.[12,13] For a thorough in-vestigation of the chemical surface structure, detail-spectra of therelated shallow core levels were obtained. The Rb 3d, Ga 3p, Cu3p, and Na 2s spectra can all be fitted with one component forall samples (see Figure S2, Supporting Information), indicatingthat the presence of Rb on the RbF PDT CIGSe and RISe/CIGSesamples does not result in the formation of additional chemicalenvironments for Ga, Cu, or Na.However, the Se 3d and In 4d related peaks undergo signifi-cant changes induced by the RISe evaporation which needs to beaccounted for by the introduction of additional contributions inthe fit model. It is possible to model the changes in line shapeof these spectra by curve fit analysis using only one additionalspectral contribution; however, such a model leads to significantinconsistencies as discussed extensively in the Supporting Infor-mation in conjunction with Figures S3 and S4 (Supporting Infor-mation). To reach meaningful data interpretations fits employingtwo additional spectral contributions are needed, as presented inFigure 1, where the Se 3d and Ga 3d/In 4d spectra of the differ-ent samples are displayed alongside with that of the RISe refer-ence. The “a” fit component of the Se 3d and In 4d spectra isattributed to CIGSe throughout, with observable shifts betweensamples attributable to changes in electronic structure as a re-sult of differing surface treatments. The “a” component and the“d” component of the In 4d spectra, attributed to oxidized In (asregularly observed on CIGSe[6,14]) are the only spectral contribu-tions for the untreated and RbF PDT absorber data. Additional“b” and “c” contributions are required to properly fit the spectraof the RISe/CIGSe sample set. The position of the “b” contribu-tion, which is shifted to lower BE, agrees with an assignment to aRISe phase for both Se and In.[6] This assignment is corroboratedby the main species of the RISe reference appearing at similar po-sitions and are labelled Seb and Inb. The secondary “c” contribu-tion appears at similar BE in the RISE reference and RISe/CIGSesample measurements; appearing in both, Se 3d and In 4d, itFigure 1. HAXPES (6 keV) spectra of RISe/CIGSe samples prepared usingdifferent RISe deposition times (resulting in different thicknesses) as wellas of a RbF PDT absorber, a bare (untreated) CIGSe, and a RISe ref. [11]for comparison. a) Se 3d and b) Ga 3d/In 4d spectra are displayed. Dataare shown with a linear background subtracted. Respective fits using pairsof Voigt profiles to represent the respective doublets are displayed alongthe data as well as the respective residuals. The vertical dashed lines markthe position of the “a” species as a visual guide; the separation betweenspecies was held constant in the fitting procedure.might be related to an additional In-Se bond environment. [NB:the widths of the fit functions in the RISE reference spectra havea ≈0.1 eV higher Gaussian width than in the other spectra, andthe separation between the “b” and “c” components differs. Thesechanges may indicate that the origin of the peaks in the RISE ref-erence differs in some way despite the close agreement.] This as-sumption is further corroborated by a previous Raman study onidentically produced RISe reference samples, indicating the pres-ence of one or more In–Se-type species besides the main RISephase.[15] However, it is difficult to make any statements on theprecise nature of the In–Se-type compound(s) and whether theyare identical on the RISe reference and the RISe/CIGSe samples.The absolute BE positions of all different core levels and speciesare given in Table S1 (Supporting Information).Adv. Funct. Mater. 2024, 2403685 2403685 (2 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deIn contrast to the clear changes of the chemical environment ofSe and In of the RISe/CIGSe samples, no indications for the for-mation of additional species can be found for the RbF PDT CIGSeabsorber, which is in contrast to previous studies, where the for-mation of a Rb-In-Se phase could be observed as a result of RbFPDT.[6,8] However, the relatively high nominal [Cu]/([In]+[Ga]) ra-tio (CGI) of 0.95 of the samples used in this study (comparedto, e.g., 0.9 in ref. [8]) might reduce the amount of RISe phaseformed to below the detection limit of our HAXPES measure-ments, which are also less surface sensitive than the conventionalXPS measurements used in ref. [13].The nominal thickness of the cover layer was calculated fromthe attenuation of the CIGSe absorber core level spectral signa-tures (Cu 3p, Ga 3p, Sea 3d, and Ina 4d; Figure 1; Figure S2, Sup-porting Information) using the Lambert–Beer law and assuminghomogeneous coverage and an IMFP of 8.6 nm (as would be ex-pected for a RISe layer) as derived by QUASES IMFP TPP2Mcode.[16,17] The results, displayed in Figure 2a, indicate increasingcover layer thicknesses with increasing RISe deposition times, asexpected. The RbF PDT (leftmost in Figure 2a) causes a smallintensity decrease of the core level spectral intensity comparedto the untreated absorber, consistent with an average cover layerthickness of (0.2 ± 0.2) nm, which is thinner than a monolayerof all suspected phases; therefore a non-closed layer is assumed,in agreement with the previously observed “nanopatterning” onsimilar absorbers.[13] The cover layer thicknesses estimated fromthe different photoemission lines are all in agreement, indicat-ing the absence of significant diffusion of any absorber-relatedelement toward the sample’s surface as a result of RISe deposi-tion. To corroborate this, the [Ga]/([Ga]+[In]) (GGI) and CGI ra-tios were calculated based on the total intensities of the given el-ements and considering only the CIGSe-related species (Cu, Ga,and Ina) and are displayed as red triangles in Figure 2b,c, respec-tively. The GGI and CGI considering only the CIGSe-related con-tributions remain constant within the error bars, showing thatthe absorber composition itself is unaffected by RISe deposition.When the calculated using total intensities (i.e., all In photoe-mission signal), the GGI and CGI decrease with increasing RISedeposition times, in agreement with the deposition of a Ga- andCu-free cover layer.The overall [Na]/([Cu]+[In]+[Ga]+[Se]) and[Rb]/([Cu]+[In]+[Ga]+[Se]) ratios were calculated as mea-sures for the respective alkali content in the samples and aredisplayed in Figure 2d. While Na is completely removed fromthe surface region by the RbF PDT, in agreement with previousobservations,[12,13] no such Na depletion is found after the RISedeposition. While the Na content slightly changes for the differ-ent RISe thicknesses/RISe deposition time, no clear trend can befound, indicating a different interaction mechanism between Rband Na in case of the RISe deposition compared to the RbF PDT,which might positively impact the device efficiency as discussedin ref. [9].While the RbF PDT leads to some Rb deposition on the sam-ple surface, its amount is significantly less, even compared to theshortest RISe deposition time. The deposited Rb amount is alsosignificantly smaller compared to that resulting from a combinedNaF/RbF PDT on low temperature (LT) absorbers,[6] the value ofwhich is presented as an open triangle in Figure 2d for compari-son. This corroborates the comparable small quantitative impactFigure 2. a) RISe layer thickness calculated on the basis of core level at-tenuation. Calculated thicknesses are given for each core level individu-ally. b) Total [Ga]/([Ga]+[In]) ratio (GGI) and GGI calculated using onlyGa and Ina species and therefore only referring to the GGI in the CIGSephase. c) Total [Cu]/([Ga]+[In]) ratio (CGI) and CGI calculated using onlyCu, Ga, and Ina species and therefore only referring to the CGI in the CIGSephase. d) [Rb]/([Cu]+[In]+[Ga]+[Se]) and [Na]/([Cu]+[In]+[Ga]+[Se]) ra-tios as a measure of respective alkali content. For comparison, the[Rb]/([Cu]+[In]+[Ga]+[Se]) ratio of low-temperature CIGSe after RbF PDTfrom ref. [6] is given. e) [Inb]/[Rb] and [Seb]/[Rb] ratios as a measure ofstoichiometry in the additionally formed “b phase”. The gray dashed linesindicate the respective positions for a stoichiometric RbInSe2 compound.f) [Inc]/[Sec] ratio as an indication of the stoichiometry in the additional“c compound”. The dashed lines indicate the expected composition forIn2Se3, CuIn5Se8, and CuIn3Se5.of the RbF PDT on the chemical structure of the CIGSe absorber(with nominal CGI = 0.95 and in particular on the chemical en-vironment of Se and In) in this study, as already discussed above.The Rb content increases for increasing RISe deposition times,reaching a [Rb]/([Cu]+[In]+[Ga]+[Se]) ratio of ≈0.04 (±0.01) for6 min. However, this is significantly less than expected for a stoi-chiometric RISe layer for the following reason: From the effectivecover layer thicknesses (Figure 2a) and the decrease in total GGIand CGI (black circles, Figure 2b,c) we can conclude that, in thecase of the 6 min RISe sample, about half of the signal originatesfrom the absorber and half from the cover layer. However, theamount of Rb on the 6 min sample is less than 20% of what isAdv. Funct. Mater. 2024, 2403685 2403685 (3 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deobserved on the pure RIS sample, which itself seems to possessa Rb-depleted surface region ([Rb]/([Cu]+[In]+[Ga]+[Se]) ≈0.23,while the value for stoichiometric RbInSe2 would be 0.33). There-fore, the amount of Rb is less than 40% of what would be expectedin the stated scenario, indicating that the deposited cover layer issignificantly Rb depleted compared to the RISe reference or stoi-chiometric RbInSe2, even though the deposition conditions werenominally identical. As significant diffusion of CIGSe related el-ements into the cover layer can be excluded, as discussed above,this indicates that a significant fraction of the Rb diffuses out ofthe deposited layer during or after deposition. This assumptionis also corroborated by the fact that glow-discharge optical emis-sion spectrometry (GDOES) measurements on identically pro-duced samples show significant amounts of Rb inside the bulkof the CIGSe absorber after a RISe evaporation also indicating asignificant diffusion of Rb into the CIGSe.[9,18]The observed Rb deficiency in RISe layers would be in agree-ment with different scenarios: A very Rb-deficient Rb–In–Se-type layer could be formed, similar to the Cu deficient vacancycompound frequently suggested to form on the CIGSe frontsurface.[6–8] However, it could also be, that an additional separatephase forms parallel to RbInSe2, which would be in agreementwith the additional Sec and Inc species discussed above.To investigate the stoichiometry of the RISe phase, the Inb/Rband Seb/Rb ratios were calculated and are plotted in Figure 2e.Both ratios agree within the error with a Rb:In:Se stoichiometryof 1:1:2 (i.e., with a proposed RbInSe24). To get a quantitative per-spective on the In-Se related second species, the Inc/Sec ratio wasderived and is depicted in Figure 2f. While the determined ratioswould be, e.g. in agreement with the formation of In2Se3, or aCu:In:Se = 1:5:8 or 1:3:5 ordered vacancy compound, an unam-biguous assignment would require further investigation. Note,also a mixture of different In-Se containing phases would be areasonable explanation.To elaborate how the different surfaces impact the CdS bufferlayer growth, CdS layers of different thicknesses were depositedon the untreated and RbF PDT absorbers as well as on the 2and 4 min RISe/CIGSe samples. Survey scans of the samples(Figure S5, Supporting Information) display the presence of Cdand S and attenuation of CIGSe related peaks after CdS deposi-tion. Analysis of the Se 3s/S 2s, Cd 4s/Rb 3d/Ga 3p, Cu 3p, Se3d, and Ga 3d/ In 4d detail spectra (Figures S6–S8, SupportingInformation) indicates the presence of one S species, which is as-signed to CdS, and two Cd species, of which the dominant speciesis also assigned to CdS, while a minor contribution at higher BEis attributed to oxidized Cd, probably related to the air exposureof the samples prior to measurement. No additional species arerequired to fit the chalcopyrite related core levels after the CdSdeposition, however, in the case of the Se 3d peak, changes in thepeak shape lead to a relative intensity change of the different Secontributions, further discussed below.To get a better impression of the buffer layer growth dynam-ics, the attenuation of the Ga 3p, Cu 3p, Se 3d, and In 4d peaks(Figures S6 and S7, Supporting Information) was used to cal-culate the buffer layer thickness using the Lambert-Beer lawand an IMFP of 9.8 nm assuming homogeneous CdS cover-age (Figure S9, Supporting Information). The average values ob-tained from the different core levels are displayed in Figure 3a.When comparing the CdS layer thicknesses for the differentFigure 3. a) CdS layer thickness calculated as averages from indi-vidual core level attenuation (Figure S9, Supporting Information). b)[Rb]/([Cu]+[Ga]+[In]+[Se]) ratios as a measure of Rb content. c) [Inb]/[Rb]and [Seb]/[Rb] ratios as a measure of stoichiometry in the additionalformed “b phase” related to RISe. The grey dashed lines indicate the re-spective positions for a stoichiometric RbInSe2 compound. d) [Inc]/[Sec]ratio as an indication of stoichiometry in the additional “c compound”,consisting of In and Se. The dashed lines indicate the expected composi-tion for In2Se3, CuIn5Se8, and CuIn3Se5.(RISe/)CIGSe samples, substantial differences are found: Whilea 2 nm layer is formed on the untreated CIGSe layer alreadyafter 30 s, the derived layer thickness is close to zero after thesame time on the RbF PDT CIGSe sample and the absorberswith RISe, indicating a delayed/slower growth of the buffer onthe Rb-containing samples in the initial growth regime. Suchan “induction/coalescence period” at the start of the CdS CBDhas been observed before.[19–21] As the duration of this stage isstrongly dependent on the chemical and morphological proper-ties of the substrate,[22] it is most likely impacted by the pres-ence of Rb at the treated absorber surface, e.g., by Cd-Rb ionexchange,[19] explaining the observed differences. After the ini-tial “induction/coalescence period;” beginning with the 3 minAdv. Funct. Mater. 2024, 2403685 2403685 (4 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.desamples, the CdS leads more strongly attenuates the absorber-related signals of the RbF PDT sample than of the untreated ab-sorber, consistent with Raman/SEM results[19] which indicate su-perior coverage.[1,23] A combined Raman/SEM study on similarsamples indicates the latter.[19] After 15 min CBD no CIGSe re-lated peak is visible on the RbF PDT sample (Figures S6 and S7,Supporting Information), which makes the determination of adistinct CdS layer thickness impossible, and also indicates a fullyclosed, compact CdS layer with thickness exceeding 3–4 × IMFP.The CdS thickness on the 2 min RISe/CIGSe sample, in contrast,is still smaller than the one on the untreated CIGSe after 3 min,indicating a longer “induction/coalescence period”, which mightindeed scale with the amount of Rb present at the sample surface.To monitor the behavior of the Rb upon CdS deposition, the[Rb]/([Cu]+[In]+[Ga]+[Se]) ratio is calculated and displayed inFigure 3b, indicating no significant Rb increase or decrease withCdS deposition time (note that, due to the very low Rb content onthe RbF PDT CIGSe sample, the Rb peak cannot be detected any-more in samples submitted to CBD treatments longer than 30 s).This lack of a trend in Rb content indicates that the Rb does notseem to diffuse into the buffer layer as Na is known to do.[24]To investigate the impact of the CdS deposition on the ob-served RISe and the In-Se species formed upon RISe deposition,the [Inb]/[Rb] and [Seb]/[Rb] as well as the [Inc]/[Sec] ratios are dis-played in Figure 3c,d. While the [Inc]/[Sec] and [Inb]/[Rb] ratiosshow no significant changes when the CdS layer is deposited,indicating the chemical RISe/CIGSe structure to be unaffectedby the CBD process of CdS, this is different in the case of the[Seb]/[Rb], which shows a pronounced increase with CBD time.This could indicate a pronounced Se enrichment in the RISephase during CdS layer growth, which is rather improbable, asno Rb-In-Se phase with a 1:1:4 stoichiometry is known. It couldalso indicate that the used fit model is insufficient and that an ad-ditional Se species might be formed upon CdS deposition. A pos-sible candidate would be Cd-Se bonds, which have been shownto form at CdS/CIGSe interfaces.[25]To investigate this, the Se 3d peak was fitted with a new model,allowing for an additional species (Sed) and fixing the Seb species,which accounts for the RISe phase, to the expected [Seb]/[Rb] =2 ratio based on the amount of Rb on the respective sample; theresult is displayed in Figure 4a. The additional species, Sed, com-pensates for the CdS CBD-induced shape changes of the peak.The position of that peak at ≈53.8 eV BE is very close to the RISerelated Seb species. However, the position is also in the energeticrange where the Se 3d line of Cd─Se bonds is expected.[26–28]To further investigate the hypothesis of Cd-Se bond formation,the Cd 4d peak of the 3 min CBD CdS/ 2 min RISe/CIGSesample was investigated, as on this sample the largest amountof Cd─Se bonds is expected according to the Se 3d fit. As theCd 4d peak is at very low BE (≈11 eV) and therefore might bestrongly impacted by hybridization, no fitting procedure is ap-plied, but it is instead compared to the respective peak of the3 min CdS/CIGSe sample on which no additional Se species wasobserved. Both peaks were normalized and plotted overlayingeach other as displayed in Figure 4b. To further visualize the dif-ference between the two spectra, a difference spectrum is addedto the figure with a magnification factor of 10. This differencespectrum indicates an additional species is present for the 3 minCBD CdS/2 min RISe/CIGSe sample at higher BE (Cd 4d5/2 atFigure 4. a) Se 3d spectra of the 30 s and 3 min CBD CdS on 2 min and4 min RISe/CIGSe samples. Data are shown with a linear background sub-tracted. Respective fits are displayed along the data as are respective resid-uals. In the displayed fit model, the Seb peak was restricted to a 2:1 ratioto the amount of Rb on the respective samples obtained from the spec-tra displayed in Figure S6 (Supporting Information). b) Cd 4d spectra of3 min CBD CdS layers on CIGSe and 2 min RISe/CIGSe samples. Peaksare normalized and the difference between them is displayed with an off-set and magnified by a factor of 10 as a gray line. c) Cation/anion ratio as[Cda]/([S]+[Sed]) ratio, with and without the inclusion of [Sed] on the anionside. The dashed line marks a 1:1 cation/anion ratio as expected for CdSand CdSe.11.4 ± 0.2 eV) compared to the main peak (Cd 4d5/2 at 11.0 ±0.1 eV); a position that agrees well with previous observationson CdSe contributions,[29] corroborating the suggestion of Cd-Sebond formation on the RISe treated samples. We cannot excludethat Cd─Se bond formation happens – to a lesser extent than onthe RISe/CIGSe sample – for the CdS/CIGSe and the CdS/RbFAdv. Funct. Mater. 2024, 2403685 2403685 (5 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.dePDT CIGSe samples, as it has been reported for similar heteroin-terfaces in literature.[29,30] The Cd-Se formation and the durationof the “induction/coalescence period” of the CBD process (char-acterized primarily by Cd(OH)2 adsorption on the CIGSe[21]) bothappear to be enhanced by the presence of Rb.To investigate how the buffer stoichiometry develops with de-position time and how it relates to the discussed formation ofthe Cd-Se bonds, the [Cda]/[S] and [Cda]/([S]+[Sed]) ratios werecalculated based on the displayed fits of the Cd 4s, S 2s, and Se3d peaks (Figures S6–S8, Supporting Information) and are dis-played in Figure 4c. In the early stages of the CBD process thereseems to be a significant S deficiency in all cases when consid-ering the [Cda]/[S] ratio (Cda is the main species; the secondaryspecies Cdb seen at higher BE is attributed to oxidized Cd[31]) TheS-deficiency is, however, more pronounced in samples contain-ing more Rb. If the Sed species, attributed above to Cd─Se bond-ing, is included along with S (i.e., calculating [Cda]/([S]+[Sed])),the anion deficiency vanishes and the cation to anion ratio in thebuffer layer is constant throughout the deposition process on thesamples displaying a Sed species. This again is a clear indicationfor a pronounced Cd─Se bond formation in the early stages of theCBD. The S deficiency also occurs for the CdS layer CBD grownon untreated CIGSe and on RbF PDT CIGSe absorbers, possiblyindicating that Cd─Se bonds may also be formed to a lesser ex-tent on these absorbers, with the related additional Se speciesbeing below our detection limit. Making a more precise state-ment on the Cd–Se-type species, however, is difficult. ThomasLepetit observed the formation of CdSe clusters in the early stagesof CdS CBD on KF PDT CIGSe absorbers[30] but not on a KFfree reference sample. Assuming the alkali element as the driv-ing force for this formation and similar effects of K and Rb (orNa), the Cd─Se bonds could therefore relate to CdSe. However,also the presence of Cd-In-(Se,S,O,OH) compounds formed atthe CdS/CIGSe interface, especially on alkali treated samples, hasbeen discussed.[19,25,30] Although our investigation shows no in-dication of In being involved in an additional phase, parts of thedetected Cd-Se bonds might still relate to such a phase with theIn contribution being below our detection limit.3. ConclusionThe chemical interface structures of (CdS/)RISe/CIGSe thin-film solar cell stacks have been studied and compared to that of(CdS/)CIGSe and (CdS/) RbF PDT CIGSe samples. While only asmall amount of Rb and no indication for RISe formation can befound for the RbF PDT CIGSe absorber (presumably attributedto the high CGI of 0.95 and the relatively high IMFP of HAX-PES compared to conventional XPS), there are clear indicationsof RISe formation for the RISe/CIGSe samples. However, the de-posited cover layer is Rb-deficient compared to a stoichiometricRbInSe2 phase, presumably due to Rb diffusion into the absorber.This Rb-deficient cover layer is described as a combination of sto-ichiometric RbInSe2 and an additional In-Se phase.When depositing a CBD CdS layer on the studied samples, aclear impact of the presence of (a) Rb (containing species) be-comes apparent: in the early stages of the deposition no sig-nificant coverage of the absorber can be observed on the Rb(species) containing samples, with the duration of this “induc-tion/coalescence period” of the CBD process seemingly scalingwith the amount of Rb (species) on the sample. In the early stagesof the CBD, formation of Cd─Se bonds is observed in particularon the RISe/CIGSe samples.Linking our findings to the corresponding solar cell param-eters published in ref. [9], the increasing VOC with increasingRISe deposition time might be related to defect passivation bythe RISe, the additional In-Se phase, or the Rb-diffusion into thebulk of the CIGSe, while the decreasing FF for solar cells with3 min or longer RISe deposition, might be related to an increas-ing series resistance and/or the formation of a current extractionbarrier due to the thicker cover layer. These competing effectslead to optimal device performance when RISe deposition is inthe 2–4 min range.These findings may contribute to understand the (beneficial)effect of the presence of Rb (species) at the buffer/absorber inter-face on the performance of respective solar cell devices.4. Experimental SectionSample Preparation and Handling: The samples were prepared at PV-comB using a high temperature (max. 530 °C) multistage process on Mo-coated soda-lime glass. The nominal [Cu]/([Ga]+[In]) (CGI) element ratiowas 0.95. While one CIGSe absorber was kept free of Rb for comparison,another absorber underwent an RbF PDT consisting of the co-evaporationof RbF and Se for 10 min at a substrate temperature of 280 °C and a sub-sequent annealing step at the same temperature for 5 min without addi-tional evaporation of RbF or Se. On the other absorbers, RISe layers withnominal thicknesses between 5 nm (1 min RISe evaporation) and 30 nm(6 min RISe evaporation) were deposited at a substrate temperature of530 °C using the same procedure as described in ref. [11], leading to anominal deposition rate of 5 nm min−1. Note that the RbF PDT as wellas RISe deposition were performed directly after finishing the CIGS prepa-ration, without breaking the vacuum. A more detailed description of theabsorber synthesis can be found in refs [9,18]. In addition, RISe was co-evaporated (thickness significantly larger than the detection depth) on Mo-coated soda-lime glass as reference. Details of the production and prop-erties of this sample can be found in ref. [11]. To investigate the impactof the RbF PDT and the RISe deposition on the CdS buffer layer prop-erties and buffer/absorber interface structure, CdS layers were depositedvia chemical bath deposition (CBD) on the RISe/CIGSe samples as wellas on the RbF PDT and the (control) untreated absorber. The time of theCBD was varied (30 s, 3 min, 15 min, with the latter being the standarddeposition time for high-performing solar cells) to produce a buffer thick-ness series. Details of the buffer layer deposition can be found in ref. [18]15 min CdS/RISe/CIGSe samples with 2 and 4 min RISe layers lead to thebest-performing devices.[9] After fabrication, the samples were briefly ex-posed to air and stored in N2 atmosphere, before being etched in aqueousNH4OH solution for 3 min at room temperature to remove surface con-taminants, subsequently rinsed in H2O, double bagged in inert nitrogenatmosphere, and shipped to the SPring-8 synchrotron in Japan for HAX-PES measurements (6 keV). For these measurements it was inevitable toexpose the samples (again) to air for ≈2 h during mounting and introduc-tion into the measurement system.Synchrotron-Based Photoelectron Spectroscopy: HAXPES experimentswere conducted at beamline BL15XU[32] of the SPring-8 storage ring. Thebase pressure of the HAXPES setup was <10−8 mbar. It was equipped witha Scienta R4000 electron energy analyzer in a near-normal emission anglegeometry. X-rays are horizontally polarized, with the direction of polariza-tion normal to the analyzer entrance slit. Spectra were recorded using cal-ibrated photon energies of 5.95 keV (referred to as 6 keV in this work)monochromatized with a Si(111) double-crystal monochromator and ahigh-resolution Si(333) channel-cut monochromator. This results in an in-formation depth (defined as three times inelastic mean free path – IMFP– accounting for ≈95% of the measured signal) of ≈25 nm which wasAdv. Funct. Mater. 2024, 2403685 2403685 (6 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.desufficient to monitor most of the interfaces of the samples and layers de-scribed above. A pass energy of 200 eV was used for all measurements,resulting in a combined analyzer plus beamline resolution of ≈0.25 eV.The binding energy (BE) was calibrated by referencing the Au 4f7/2 peak ofa grounded clean Au foil to a binding energy of 84.00 eV.Curve Fit Analysis: Core levels were fitted using linear backgrounds andVoigt profiles, keeping – if more than one species was present – the in-terspecies energy separation of one core level constant for all excitationenergies and keeping the shape of the Voigt identical for identical core lev-els. For core-levels with a splitting due to spin orbit coupling (i.e., all corelevels with l > 0), two Voigt profiles with a fixed splitting and a fixed ratioaccording to1+2(l+ 12 )1+2(l− 12 )were used.[33]Stoichiometry Calculation: For calculating elemental ratios, shallowcore levels (Cd 4s, Rb 3d, Cu 3p, Na 2s, Se 3d, Ga 3d, In 4 d) were onlycorrected by photoionization cross-section,[34,35] as the kinetic energy ofthe respective photoelectrons was very similar and thus any impact of dif-ferent IMFP and a different analyzer transmission was negligible. The S2s core level was additionally corrected by analyzer transmission functionand IMFP,[16] as its binding energy differs by ≈200 eV from that of theshallow core levels.Cover Layer Thickness Calculation: Buffer layer thicknesses “d” werecalculated using the Lambert-Beer-law: d = IMFP ∗ ln( I0Id) with I0Idbe-ing the ratio of absorber-related photoemission intensities prior and afterbuffer layer deposition. For the two calculated buffer layers, IMFP valuesof 8.6 nm (RISe) and 9.8 nm (CdS) were used.[16]Supporting InformationSupporting Information is available from the Wiley Online Library or fromthe author.AcknowledgementsJ.B. acknowledges support from the Graduate School Materials for So-lar Energy Conversion (MatSEC) as part of Dahlem Research School. TheHAXPES measurements at SPring-8 were performed under an approval ofNIMS Synchrotron X-ray Station (Proposal No. 2019A4910) and was sup-ported by NIMS microstructural characterization platform as a programof “Nanotechnology Platform” (project No. 12024046) of the Ministry ofEducation, Culture, Sports, Science and Technology (MEXT), Japan.Open access funding enabled and organized by Projekt DEAL.Conflict of InterestThe authors declare no conflict of interest.Data Availability StatementThe data that support the findings of this study are available from the cor-responding author upon reasonable request.Keywordschalcopyrite thin-film solar cells, HAXPES, RbF-PDT, RbInSe2Received: February 29, 2024Revised: March 31, 2024Published online:[1] A. Chirilă, 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, Nat. Mater. 2013, 12, 1107.[2] P. Jackson, D. Hariskos, R. Wuerz, O. Kiowski, A. Bauer, T. M.Friedlmeier, M. Powalla, Phys. Status Solidi RRL – Rapid Res. Lett.2015, 9, 28.[3] M. Nakamura, K. Yamaguchi, Y. Kimoto, Y. Yasaki, T. Kato, H.Sugimoto, IEEE J. Photovolt. 2019, 9, 1863.[4] M. Malitckaya, H.-P. Komsa, V. Havu, M. J. Puska, J. Phys. Chem. C2017, 121, 15516.[5] E. Handick, P. Reinhard, R. G. Wilks, F. Pianezzi, T. Kunze, D.Kreikemeyer-Lorenzo, L. Weinhardt, M. Blum, W. Yang, M. Gorgoi,E. Ikenaga, D. Gerlach, S. Ueda, Y. Yamashita, T. Chikyow, C. Heske,S. Buecheler, A. N. Tiwari, M. Bär, ACS Appl. Mater. Interfaces 2017,9, 3581.[6] J. Bombsch, E. Avancini, R. Carron, E. Handick, R. Garcia-Diez, C.Hartmann, R. Félix, S. Ueda, R. G. Wilks, M. Bär, ACS Appl. Mater.Interfaces 2020, 12, 34941.[7] N. Taguchi, S. Tanaka, S. Ishizuka, Appl. Phys. Lett. 2018, 113, 113903.[8] N. Maticiuc, T. Kodalle, J. Lauche, R. Wenisch, T. Bertram, C. A.Kaufmann, I. Lauermann, Thin Solid Films 2018, 665, 143.[9] T. Kodalle, T. Bertram, R. Schlatmann, C. A. Kaufmann, IEEE J. Photo-volt. 2019, 9, 1839.[10] T. P. Weiss, S. Nishiwaki, B. Bissig, R. Carron, E. Avancini, J. Löckinger,S. Buecheler, A. N. Tiwari, Adv. Mater. Interfaces 2017, 5, 1701007.[11] T. Kodalle, R. K. M. Raghupathy, T. Bertram, N. Maticiuc, H. A.Yetkin, R. Gunder, R. Schlatmann, T. D. Kühne, C. A. Kaufmann,H. Mirhosseini, Phys. Status Solidi RRL – Rapid Res. Lett. 2019, 13,1800564.[12] E. Avancini, R. Carron, T. P. Weiss, C. Andres, M. Bürki, C. Schreiner,R. Figi, Y. E. Romanyuk, S. Buecheler, A. N. Tiwari, Chem. Mater. 2017,29, 9695.[13] T. Kodalle, M. D. Heinemann, D. Greiner, H. A. Yetkin, M. Klupsch,C. Li, P. A. Aken, I. Lauermann, R. Schlatmann, C. A. Kaufmann, Sol.RRL 2018, 2, 1800156.[14] J. Bombsch, E. Avancini, R. Carron, E. Handick, R. Garcia-Diez, C.Hartmann, R. Félix, D. Abou-Ras, S. Ueda, R. G. Wilks, M. Bär, Adv.Energy Sustain. Res. 2021, 2, 2100101.[15] M. Guc, T. Kodalle, R. Kormath Madam Raghupathy, H. Mirhosseini,T. D. Kühne, I. Becerril-Romero, A. Pérez-Rodríguez, C. A. Kaufmann,V. Izquierdo-Roca, J. Phys. Chem. C 2020, 124, 1285.[16] S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interface Anal. 1993, 20, 77.[17] S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interface Anal. 1994, 21, 165.[18] T. Kodalle, Unraveling the Structural and Optoelectronic Effects of Rbon Chalcopyrite Solar Cells. PhD Thesis 2020.[19] T. Kodalle, L. Choubrac, L. Arzel, R. Schlatmann, N. Barreau, C. A.Kaufmann, Sol. Energy Mater. Sol. Cells 2019, 200, 109997.[20] L. Choubrac, G. Brammertz, N. Barreau, L. Arzel, S. Harel, M. Meuris,B. Vermang, Phys. Status Solidi A 2018, 215, 1800043.[21] P. O’Brien, J. McAleese, J. Mater. Chem. 1998, 8, 2309.[22] D. Lincot, J. Vedel, In Tenth E.C. Photovoltaic Solar Energy Confer-ence, (Eds: A. Luque, G. Sala, W. Palz, G. Santos, P. Helm), SpringerNetherlands, Dordrecht 1991, pp 931.[23] T. M. Friedlmeier, P. Jackson, D. Kreikemeyer-Lorenzo, D. Hauschild,O. Kiowski, D. Hariskos, L. Weinhardt, C. Heske, M. Powalla, In 2016IEEE 43rd Photovoltaic Specialists Conf. (PVSC), Portland, OR, USA,June 2016, pp 0457.[24] H. A. Yetkin, T. Kodalle, T. Bertram, A. Villanueva-Tovar, R. Klenk, B.Szyszka, R. Schlatmann, C. A. Kaufmann, In 2020 47th IEEE Photo-voltaic Specialists Conf. (PVSC), Calgary, AB, Canada 2020, pp 1192.[25] P. Yang, R. G. Wilks, W. Yang, M. Bär, ACS Appl. Mater. Interfaces 2020,12, 6688.[26] R. Islam, D. R. Rao, J. Electron Spectrosc. Relat. Phenom. 1996, 81, 69.[27] H. Rupp, U. Weser, Bioinorg. Chem. 1975, 5, 21.Adv. Funct. Mater. 2024, 2403685 2403685 (7 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.de[28] M. Polak, J. Electron Spectrosc. Relat. Phenom. 1982, 28, 171.[29] L. Ley, R. A. Pollak, F. R. McFeely, S. P. Kowalczyk, D. A. Shirley, Phys.Rev. B 1974, 9, 600.[30] T. Lepetit, Influence of KF Post Deposition Treatment on the Poly-crystalline Cu(In,Ga)Se2/CdS Hetero-Junction Formation for Photo-voltaic Application, Doctoral Thesis, Université de Nantes, Nantes,2015.[31] D. Kreikemeyer-Lorenzo, D. Hauschild, P. Jackson, T. M. Friedlmeier,D. Hariskos, M. Blum, W. Yang, F. Reinert, M. Powalla, C. Heske, L.Weinhardt, ACS Appl. Mater. Interface 2018, 10, 37602.[32] S. Ueda, Y. Katsuya, M. Tanaka, H. Yoshikawa, Y. Yamashita, S.Ishimaru, Y. Matsushita, K. Kobayashi, AIP Conf. Proc. 2010, 1234,403.[33] F. de Groot, A. Kotani, Core Level Spectroscopy of Solids, 1st ed., CRCPress, Boca Raton, FL 2008.[34] M. B. Trzhaskovskaya, V. I. Nefedov, V. G. Yarzhemsky, At. Data Nucl.Data Tables 2001, 77, 97.[35] M. B. Trzhaskovskaya, V. K. Nikulin, V. I. Nefedov, V.G. Yarzhemsky, At. Data Nucl. Data Tables 2006, 92,245.Adv. Funct. Mater. 2024, 2403685 2403685 (8 of 8) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403685 by National Institute For, Wiley Online Library on [31/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.de