# Fileset

[Adv Funct Materials - 2026 - Kobayashi - Water Adsorption‐Induced Color Sensor  Insight Into the Sensing Mechanism and.pdf](https://mdr.nims.go.jp/filesets/3fda8ea9-0444-4940-acd3-5bfbe6f89e3a/download)

## Creator

Tai Kobayashi, Ryosuke Nishikubo, Yusuke Tomiyori, Fumitaka Ishiwari, Daisuke Asakura, Eiji Hosono, Miho Kitamura, [Hisao Kiuchi](https://orcid.org/0000-0001-9139-8218), Yoshihisa Harada, Akinori Saeki

## Rights

[Creative Commons BY-NC Attribution-NonCommercial 4.0 International](https://creativecommons.org/licenses/by-nc/4.0/)

## Other metadata

[Water Adsorption‐Induced Color Sensor: Insight Into the Sensing Mechanism and Interfacial Engineering for Improved Responsiveness](https://mdr.nims.go.jp/datasets/4b04ff93-379a-4178-b03f-3be299e9584c)

## Fulltext

Water Adsorption‐Induced Color Sensor: Insight Into the Sensing Mechanism and Interfacial Engineering for Improved ResponsivenessAdvanced Functional Materialswww.afm-journal.deRESEARCH ARTICLEWater Adsorption-Induced Color Sensor: Insight Into theSensing Mechanism and Interfacial Engineering forImproved ResponsivenessTai Kobayashi1 Ryosuke Nishikubo1,2 Yusuke Tomiyori3,4 Fumitaka Ishiwari1,2,5 Daisuke Asakura6Eiji Hosono6 Miho Kitamura7 Hisao Kiuchi3,8 Yoshihisa Harada3 Akinori Saeki1,21Department of Applied Chemistry, Graduate School of Engineering, The University of Osaka, Suita, Osaka, Japan 2Innovative Catalysis Science Division,Institute for Open and Transdisciplinary Research Initiatives (ICS-OTRI), The University of Osaka, Suita, Osaka, Japan 3Institute for Solid State Physics (ISSP),The University of Tokyo, Kashiwa, Chiba, Japan 4AGC Inc., Yokohama, Kanagawa, Japan 5Faculty & Graduate School of Urban Environmental Science,Tokyo Metropolitan University, Tokyo, Japan 6Research Institute for Energy Efficient Technologies, National Institute of Advanced Industrial Science andTechnology (AIST), Tsukuba, Ibaraki, Japan 7NanoTerasu Center, National Institutes for Quantum Science and Technology (QST), Sendai, Miyagi, Japan8Research Center for Energy and Environmental Materials (GREEN), National Institute for Materials Science (NIMS), Tsukuba, Ibaraki, JapanCorrespondence: Ryosuke Nishikubo (nishikubo@chem.eng.osaka-u.ac.jp) Hisao Kiuchi (kiuchi.hisao@nims.go.jp) Yoshihisa Harada(harada@issp.u-tokyo.ac.jp) Akinori Saeki (saeki@chem.eng.osaka-u.ac.jp)Received: 21 January 2026 Revised: 11 May 2026 Accepted: 17 June 2026Keywords: antimony | color-sensing | photodetector | photovoltaics | TiO2ABSTRACTColor-sensitive photodetectors are at the forefront of next-generation light- and image-sensing research. In this context, aphotovoltaic device employing SbSI:Sb2S3 and TiO2 as the visible light absorber and electron transport material, respectively,exhibits a unique wavelength-dependent photovoltage effect and its reversible output in response to changing humidity levels.However, the specific functions of the individual layers of the device and their interaction with water remain largely unknown. Inthis study, we perform in situ X-ray absorption spectroscopy (XAS) of thin films under humidity-controlled conditions to revealthe interaction between water and the photovoltaic layers. The oxygen K-edge XAS data reveal the adsorption of water moleculesonto TiO2. Ultraviolet (UV) irradiation of TiO2 can trigger photochemical reactions of the adsorbed water at the interface, whichis considered to contribute to changes in the photovoltage. Based on this finding, we introduce hydrophilic layers between TiO2and SbSI:Sb2S3 to promote the accumulation of water in the devices, which successfully enhance both the wavelength sensitivityand response speed. Finally, we demonstrate color recognition with various irradiation intensities, thereby broadening the scopeand applications of unique wavelength-responsive single-cell photovoltaics.1Poira(Tm©AhIntroductionhotodetectors are essential components of information technol-gy that enable diverse optical applications, such as light sensing,mage capture, and optical communications [1–3]. A criticalequirement for next-generation photodetectors is responsivenesscross multiple wavelengths. Photodetectors and photovoltaicPV) devices typically exhibit a linear increase in the short-his is an open access article under the terms of the Creative Commons Attribution-NonCedium, provided the original work is properly cited and is not used for commercial purp2026 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbHdvanced Functional Materials, 2026; 36:e76809ttps://doi.org/10.1002/adfm.76809circuit current density (JSC) with incident light intensity (Pin),whereas the open-circuit voltage (VOC) increases marginally withthe logarithmic value of Pin, as described by Shockley’s diodeequation [4, 5]. Besides, the PV output is generally independentof the irradiation wavelength if the photon energy is greater thanthe bandgap energy of the photoactive layer. Thus, thewavelengthof the irradiated light can be distinguished using color filters(e.g., red, green, blue (RGB)) or spatial separation through prisms.ommercial License, which permits use, distribution and reproduction in anyoses.1 of 10http://www.afm-journal.dehttps://doi.org/10.1002/adfm.76809https://orcid.org/0000-0002-0533-9180https://orcid.org/0000-0002-0200-4510https://orcid.org/0000-0001-9139-8218https://orcid.org/0000-0002-4590-9109https://orcid.org/0000-0001-7429-2200mailto:nishikubo@chem.eng.osaka-u.ac.jpmailto:kiuchi.hisao@nims.go.jpmailto:harada@issp.u-tokyo.ac.jpmailto:saeki@chem.eng.osaka-u.ac.jphttp://creativecommons.org/licenses/by-nc/4.0/https://doi.org/10.1002/adfm.76809http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadfm.76809&domain=pdf&date_stamp=2026-07-23IdapdpllwurmWRodfwwtfti5rdtctowiHsftvAiwwiewilfrmvbtpehotce2 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Crean 2022, we reported the unprecedented filter-less wavelengthependence of the photovoltage in a single cell composed ofn antimony sulfiodide and sulfide (SbSI:Sb2S3, named ASIS)hotoactive layer [6, 7]. The device structure was fluorine-oped tin oxide (FTO)/TiO2/ASIS/PCPDTBT/Au [PCPDTBT:oly(cyclopentadithiophene-benzothiadiazole)], which exhibitedow and high photovoltages under ultraviolet (UV) and visible-ight irradiation, respectively (Figure S1a,b). Interestingly, thisavelength-dependent photovoltaic effect (WDPE) is enhancednder high-humidity conditions. Although SbSI and Sb2S3 haveeceived significant attention in the areas of PV [8–11] ferroelectricaterials and sensors [12–14] and photocatalysts [15, 16], theirDPE behavior remains largely unexplored.ecently, remarkable progress has beenmade in the developmentf multicolor-sensitive photodetectors. Two types of devices exist:ual-bandphotodetectors and color-selective photodetectors. Theormer is constructed from two different photoactive materialsith different bandgaps that exhibit responses in two distinctavelength regions (450–500 and 650–800 nm) [17–19]. However,hey are unable to discern the irradiated wavelength without ailter. In contrast, color-selective photodetectors exhibit dynamicunability of color detection [20–23]. For instance, Lan et al.nvented an organic photodetector that selectively detects 400–00 or 520–620 nm light under positive and negative biases,espectively [21]. They also demonstrated a 2D image-capturingevice. Min et al. reported the wavelength-specific photocurrentransient of a perovskite photodetector [24],which showed chargearrier dynamics partly similar to those of our WDPE. Notably,heir single-junction self-powered photodetector successfully rec-gnized a wide range of colors (350–750 nm) in the photocurrentaveform. Color-selective detection is advantageous for enhanc-ng spatial resolution and reducing the size of imaging devices.owever, color detection based on the output voltage using aingle-junction cell has only been achieved within our WDPEramework. This characteristic allows for the simultaneous detec-ion of light intensity andwavelength from the output current andoltage, respectively.significant challenge associated with the WDPE phenomenons understanding its mechanism. Szaniawski et al. reportedavelength-dependent photovoltages in Cu(In,Ga)Se2 PVs inhich the breakdown voltage temporarily decreased upon UVrradiation [23]. These phenomena were attributed to the differ-nces in the penetration depths of UV and visible light. However,e observed that WDPE remained unaffected by the direction ofrradiation, whether from the bottom or top, suggesting that theight penetration depth was not the origin of the WDPE. There-ore, WDPE likely originates from more complex photochemicaleactions involving charge carriers, water (or polar solvent)olecules, and the materials that compose the device. Througharious evaluations, we elucidated three key properties: (i)WDPEehavior depends significantly on the selection of the electronransport material (ETM) and hole transport material (HTM). Inarticular, a device containing TiO2 (ETM) exhibited particularlynhanced WDPE. (ii) Strong WDPE behavior occurs under high-umidity conditions. (iii)WDPE is likely related to the generationf trap sites induced by UV irradiation. Property (i) suggests thathe interfaces between these materials and the active layer play arucial role. Both transient photovoltage (TPV) [25] and chargextraction by linearly increasing the voltage (CELIV) [26, 27]of 10exhibited accelerated recombination under UV irradiation. TPVmeasurement results indicate that the lifetime decreased from31.5 to 14.9 ms upon UV irradiation, and a comparable decay inlifetimewas also observed in CELIVmeasurement (Figure S1c,d).Electron spin resonance (ESR) measurements under visible-lightirradiation [28, 29] exhibited reduced hole densities in the HTMupon additional UV irradiation (Figure S1e,f). This contrastswith the behavior of conventional PV devices, providing directevidence of accelerated recombination and/or trapping ratherthan an increase in hole density upon UV irradiation.Despite these findings, the layers that play specific roles in theWDPE remain unclear. Various measurements of PV devices,including current density–voltage (J–V) scanning, TPV, CELIV,and impedance spectroscopy, inherently comprise contributionsfrom multiple layers, making it challenging to isolate the roleof each layer. Although ESR clearly detected a signal attributedto holes in the organic HTM, no signals from other inorganiclayers, such as TiO2 and ASIS, were observed. To assess thepresence of low-density water molecules at the interface andstudy photochemical reactions under light exposure, we mustutilize an advanced spectroscopic technique with intense X-raysfrom a synchrotron facility.In this study, we performed in situ X-ray absorption spectroscopy(XAS) to determine the location of trapped water molecules inthe FTO/TiO2/ASIS/PCPDTBT/Au device. A specially designedambient-pressure cell was used to enable humidity-controlledmeasurements of the film samples. The intensity of the oxygen(O) K-edge XAS spectra increased under high-humidity condi-tions. This suggests that water molecules can easily penetratethe device structures and adsorb onto TiO2 surfaces. The watermolecules adsorbed on the TiO2 layer were correlated with adramatic change in VOC under UV irradiation. Subsequently,photocatalytic reactions are induced on TiO2, whichmay produceionic species such as H+ andOH−, working as temporal trap sites.Based on these results, hydrophilic polymers were introducedinto the device to improve the sensitivity and response speedof the wavelength-dependent photovoltage. Polystyrene sulfonicacid significantly improved the response speed. Among repre-sentative self-powered color-sensitive photodetectors, our deviceexhibited the fastest response to photovoltage-based wavelengthrecognition (Table S1). These findings are expected to pavethe way for improving the multicolor sensing ability of WDPEdevices.2 Results and Discussion2.1 Mechanism of WDPE2.1.1 Analysis of Water Penetration Behavior via InSitu XASFour thin-film samples with different stacking structures onglass substrates were prepared for O K-edge XAS measurements:single layers of TiO2, ASIS, and PCPDTBT, and a triple layer ofTiO2/ASIS/PCPDTBT. Themeasurement setup of the in situ XASis shown in Figure S2. Water adsorption was evaluated basedon the change in the X-ray absorption intensity at the O K-edge(520–550 eV). The obtained O K-edge XAS spectra of the glassAdvanced Functional Materials, 2026tive Commons LicenseFIGURE 1 (a) O K-edge XAS spectra of a triple layer measured under controlled humidity. (b) Corrected difference XAS spectra ratios (δ’) of singlelayers and a triple layer. The correction procedure is shown in Figure S5. The peaks attributed to H2O gas and liquid-like H2O are marked with blue andred dots, respectively. The pre-edge peak attributed to H2O–TiO2 interaction is marked with red circles. The spectra of SbSI, HTM, and a tri-layer areshifted along the vertical axis for easier comparison. (c) Schematic of water molecule penetration and adsorption onto the layered samples. The SEMimage of ASIS on TiO2 is appended, which is taken from our previous report [6].stitw5toptdtRaccrncsFswA 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Creaubstrate, TiO2, ASIS, HTM single layers, and TiO2/ASIS/HTMriple layer are shown in Figure S3 and Figure 1a. The broad,ntense absorption at 535–550 eV in these samples was attributedo the oxygen in the TiO2 and glass (SiO2) substrates [30, 31] alongith trivial contributions from water. Sharp peaks appeared at31.0 and 533.5 eV in the TiO2 single layer and TiO2/ASIS/HTMriple layer (Figure S3c; Figure 1a), which are attributed to thexygen in TiO2 [32]. The measurement for each sample waserformed in the order of first: dry (S1), second: wet (S2), andhird: dry (S3) (where S denotes the XAS spectrum). A slight butistinct increase and then decrease in intensity was observed inhe 535–550 eV region in response to the humidity value (wet: 70%H and dry: 0% RH). This change is attributed to the adsorptionnd desorption of water. To compensate for the large backgroundoming from the glass substrate and layer components, wealculated the difference spectra ratios (δ). Owing to the X-ay-induced hydrophilization (Figure S4) [33], S1 and S3 wereot the same. Consequently, the hydrophilization effect wasompensated for as much as possible by linear interpolation, ashown in Figure S5.igure 1b shows the corrected difference spectra ratios (δ’) ofingle- and triple-layer samples. The difference spectra ratiosithout correction are shown in Figure S6 as a reference. Thedvanced Functional Materials, 2026peaks of H2O gas and liquid-like H2O are marked by blue andred dots, respectively, and the broad difference spectrum withsmall sharp peaks at 534, 536, and 537 eV is attributed to amixture of liquid and gaseous water [32, 34]. Water gas is derivedfrom humidified gas flowing through the space between theSiC membrane and film (a few microns). Our observations areconsistent with those of Nordlund et al., who reported that wateradsorbed onto a metal catalyst exhibits O K-edge XAS spectrasimilar to those of ice and liquid water [35]. Notably, the TiO2 andTiO2/ASIS/HTM films exhibited a pre-edge at 530.5 eV markedby red circles. A similar extra peak was also observed in H2O-adsorbed TiO2, as described in a previous report [36], suggestingthe adsorption of H2O onto TiO2 [37, 38]. The triple-layer filmalso exhibited an enhanced broad signal at 538–543 eV, which isspecific to solid H2O. Assuming a simple SiO2/TiO2/H2O triple-layer model and calculating the integral intensity ratio at 550 eVin the glass/TiO2 spectra (Note S1 and Tables S2 and S3), thethickness of the adsorbedwater was estimated to be 10.0–12.4 nm.Because the ASIS layer is considerably thin (around 40 nm) andhas a mesoporous morphology (Figure 1c), water molecules seemto penetrate it easily. A possible concernwhen comparing the bareTiO2 and the fully layered sample is the signal attenuation causedby the overlayers. However, the HTM (PCPDTBT) and ASISlayers do not contribute to the O Kα fluorescence intensity but3 of 10tive Commons Licensefo(tdrfAabGdbmplchtlpasspaeSi2TtdcstsrtUOatgcr(ai(aT(HRA4 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Creaunction purely as attenuation filters because they do not containxygen. Comparing a 3-layer (glass/TiO2/H2O) and a 5-layerglass/TiO2/H2O/ASIS/HTM) model (Tables S2 and S3) revealshat although the absolute spectral intensity and the intensityifference decrease by 18–38% considering the error margin, δemain unchanged. However, δ’ of the TiO2 and TiO2/ASIS/HTMilms exhibit distinct profiles. This difference suggests that theSIS layer may affect the wettability of TiO2, which playsn important role in the WDPE response time, as discussedelow.iven that WDPE is most pronounced in TiO2/ASIS/HTMevices and depends significantly on humidity, the interactionetween TiO2 and water is considered essential to the WDPEechanism. To further probe the interfacial reactions, we alsoerformed the Ti L3-edge XAS on the TiO2/ASIS/HTM tri-ayer film under controlled humidity (Figure S7a). Characteristichanges were observed in the Ti L3-edge XAS spectra afterumidification, such a spectral change in the pre-edge region,ogether with a ≈0.1 eV shift of the peak at 457.9 eV towardower photon energy. Notably, these changes are consistent with arevious report where TiO2 nanoparticles in aqueous dispersionnd under dry conditions were compared [39]. This result alsoupports the possibility of H2O penetration into the stackedtructure and its interaction with TiO2. Sb M-edge XAS was alsoerformed; although the fluorescence line of the Sb M-edge ispproximately 100 eV lower than that of the OK-edge, the limitednergy resolution (≈100 eV) of the SDD detector causes the weakb M-edge emission to be obscured by the overlapping tail of thentense O K-edge emission (Figure S7b)..1.2 Role of the TiO2 Layer onWDPEo validate the importance of TiO2 inWDPE, we studied the spec-ral responses of our devices. Figure 2a shows the wavelength-ependent JSC and VOC at 50% RH. The device structure andorresponding external quantum efficiency (EQE) spectra arehown in Figure S8. Importantly, the change inVOC (ΔV) betweenhe initial (0 s) and delayed (20 s) values was observed in thehort wavelength region (<400 nm) (Figure 2b). This region cor-esponds to the bandgap energy of TiO2 (Figure S9a), suggestinghat TiO2 is mainly responsible for the decrease in photovoltage.nder UV irradiation, the adsorbed H2O may split into H+ orH−, which are considered to be trapped at the ASIS/HTMnd/or TiO2/ASIS interface and serve asmetastable electron/holerap sites (Figure 2c). These ionic species are considered toradually recombine and disappear, resulting in reversible deviceharacteristics. We previously reported that the photovoltageesponse changed sublinearly with the irradiation wavelength340–530 nm) [6]. Light with energy lower than the TiO2 bandgaplso induces a photovoltage drop to some extent. Consider-ng the large Urbach energy of the TiO2 film (650–680 meV)Figure S9b), photoabsorption by the midgap states of TiO2 canlso induce a decrease in photovoltage.o examine the necessity of the HTM and photoactive layerASIS), configurations of devices with control, devices withoutTM, and devices without ASIS were also evaluated at 50%H. Figure 3a shows the J–V curves of the HTM-less andSIS-less devices. Despite the much lower power conversionof 10efficiency than that of the control device, the HTM-less and ASIS-less devices exhibited diode behaviors. Importantly, the HTM-and ASIS-less devices also exhibited WDPE in time-dependentphotovoltage (TDPV) (Figure 3b,c). In addition, we previouslyreported that WDPE was not observed in devices in which TiO2was replacedwith SnO2 or CdS as ETMs [6]. These results supportthe importance of the TiO2 layer in WDPE. Notably, devices withdifferent active layers, such as Pb perovskite andAgxBiyIx+3y, havenever exhibited WDPE [6]. This is probably due to the densefilm formation and/or the specific interfacial interaction withTiO2, which hinder the interfacial photocatalytic reaction. Incontrast, ASIS layer is advantageous for WDPE property becauseof its porous characteristics and the efficient carrier transportproperty. The porous morphology facilitates H2O penetration. Inaddition, ASIS has efficient carrier transport property with thehigh EQE value (≈60%) [40], which facilitates observing the effectof increased trap-site.We previously reported the influence of UV irradiation oncarrier dynamics using transient photovoltage (TPV) and charge-extraction by linearly increasing voltage (CELIV) [6], and demon-strated that it results in the accelerated TPV transient (FigureS1c). However, the change in trap density has not been quantified.Herein, space-charge limited current (SCLC) measurement wasperformed with and without UV irradiation (Figure S10). Beforeand after a temporal UV irradiation (30 s), SCLCwas immediatelymeasured. The result exhibited an increase in trap density from1.32 × 1017 to 1.53 × 1017 cm−3, suggesting the generation of defectstate upon UV irradiation.2.2 Improvement of Color-SensingCharacteristicsInspired by the results showing that water can penetrate a deviceand adhere to the ASIS and/or TiO2 layers, we introduced a thinhydrophilic polymer layer at the TiO2/ASIS or ASIS/HTM inter-face to promote water penetration and adsorption and enhancethe photochemical reaction of TiO2. PSS (polystyrene sulfonicacid) and poly(vinyl alcohol) (PVA) were used as the hydrophilicpolymers. Notably, the introduction of the polymers resulted in adramatic improvement in the wavelength-photovoltage response(Figure 4a–e). The corresponding photocurrent decays are dis-played in Figure S11. The reduced photocurrent is attributed tothe introduction of an insulating layer such as PVA and PSS. Theresponse timewas calculated using a double exponential fitting ofthe TDPV curve obtained under UV irradiation. In particular, theintroduction of the PSS layer dramatically improved the responsespeed to 150–390 ms (Figure 4f) compared to a few seconds(2–4 s) of the control (no interlayer) device. This rapid response isbeneficial in sensing applications.We further investigated the time-dependent photovoltageresponse under periodic UV pulse irradiation and continuousvisible light irradiation. In the control device (Figure 4g), theadditional rectangular UV irradiation (0.5 s) initially induced arapid increase in the photovoltage owing to the increase in totalphoton density. This effect was clearly observed in the periodicresponse without irradiation of the 515 nm continuous light(Figure S12). Subsequently, a rapid decrease in photovoltage(= WDPE) of ΔV = 35 ± 2.2 mV was observed. The recoveryAdvanced Functional Materials, 2026tive Commons LicenseFIGURE 2 (a) Spectrum-dependent photovoltage and responsivity. The retention time for each wavelength was set to 20 s. The blue and orangelines represent the photovoltage at the beginning and end of each step. (b) Spectrum-dependence of the change in photovoltage (ΔV) during the retentiontime. The bandgap energy of TiO2 calculated from its photoabsorption spectrum is 3.02 eV (Figure S9). (c) The suggested mechanism of WDPE basedon the data shown in Figures 1 and 2a,b.Advanced Functional Materials, 2026 5 of 10 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 LicenseFIGURE 3 (a) J–V curves of control, HTM-less, and ASIS-less devices measured under a 100mW cm−2 Xe lamp. The solid and dotted lines indicatethe forward and reverse scans, respectively. TDPV decays of (b) HTM-less device and (c) ASIS-less device. The corresponding photocurrent density (Jph0)is displayed. The photon densities of the irradiated 315 nm (blue line) and 515 nm (green line) continuous lights are tuned to yield comparative currentdensities (6.1 × 1016 and 7.1 × 1016 photons cm−2 s−1 for the 315 and 515 nm lights in b), 3.3 × 1016 and 6.7 × 1017 photons cm−2 s−1 for the 315 and 515 nmlights in (c).t1operroraprSW(id(tcbttTi3Wcottaacbta6 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Creaime (τRec) after switching off the UV light was calculated to be4.1 s using double exponential analysis. The good repeatabilityf the photovoltage response is also notable. Surprisingly, theeriodic response of the PVA- and PSS-incorporated devicesxhibited not only improved ΔV of 45 ± 2.0 and 63 ± 3.6 mV,espectively, but also shortened τRec values of 11.6 and 3.12 s,espectively (Figure 4h,i and Table 1). Thus, the introductionf hydrophilic polymers improves the UV sensitivity andesponse/recovery speed, which is highly preferred for sensingpplications. Furthermore, among representative self-poweredhotodetectors, our device exhibited a considerably fasteresponse and restoration time for wavelength recognition (Table1).e further demonstrated its color-sensing performanceFigure 5).Upon irradiationwith 375 and 530nm lights at differentntensities, the voltage change (ΔV/V0) showed a distinct trendepending on the light intensity ratio. At a high 375 nm intensityI375) and low 530 nm intensity (I530) corresponding to the leftop part in the figure, ΔV/V0 became significantly negative. Inontrast, low I375 and high I530 irradiation, corresponding to theottom right, resulted in a decreased change. Additionally, theotal light intensity (or photon density) can be determined fromhe short-circuit current, as confirmed in our previous study [6].hus, our single-junction photodetector detects both the lightntensity and color composition without a filter.Conclusione performed in situ XAS measurements under humidity-ontrolled conditions to reveal the water penetration behaviorf PV devices. Adsorption of H2O onto TiO2 was observed inhe O K-edge XAS spectra. By comparing the layered samples,he H2O molecules were found to penetrate the devices anddsorb onto the mesoscopic TiO2/ASIS composites. Accordingly,photochemical reaction between the adsorbedwater and chargearriers in TiO2 was considered to have occurred. The possi-ly generated reactive species such as H+ or/and OH− wererapped at the ASIS/HTM and/or TiO2/ASIS interfaces, servings metastable charge traps. Based on these results, a hydrophilicof 10polymer was incorporated into the device structure to enhancewater permeation and retention. The response speed of the devicewith a PSS layer (τUV = 150 ms) was improved by one orderof magnitude. Surprisingly, the PSS-incorporated device alsoexhibited an improved color sensitivity (ΔV) and recovery speed(τRec = 3.12 s). These rapid responses are advantageous for sensingapplications. Notably, our single-junction photodetector detectedboth the light intensity and color composition without a filter.Our work offers new insights into the mechanism of the WDPEand strategies for enhancing the wavelength responsiveness ofsingle-junction multicolor sensing devices.4 Experimental Section4.1 SbSI:Sb2S3 (ASIS) Film PreparationSbSI:Sb2S3 films were prepared according to our previous report[6, 40]. First, 0.22 m SbI3 and 0.18 m Sb(EtXa)3 were dissolvedin dimethyl sulfoxide (DMSO). The solution was spin-coated at2300 rpm for 30 s, followed by thermal annealing at 160◦C for5 min. The solution was spin-coated again, followed by thermalannealing at 240◦C for 10 min. The thermal decomposition ofmetal-xanthate salts provides metal sulfides, which is known asChugaev elimination [41].4.2 Photovoltaic Device FabricationAn FTO substrate was cleaned with detergent, deionized water,and isopropyl alcohol. For TiO2 deposition, a compact TiO2layer was deposited onto the FTO/glass via spray pyrolysisusing a solution of titanium diisopropoxide bis(acetylacetonate)(Sigma Aldrich) in ethanol (1:12 v/v) at 450◦C. A 200-nm-thickmesoporous-TiO2 layer was deposited onto the compact TiO2layer by spin coating (5000 rpm, 30 s) diluted TiO2 paste inethanol (paste/ethanol = 1:7 w/w), followed by sintering at500◦C for 20 min. The substrates were then transferred to aN2-filled glove box (O2, H2O < 0.1 ppm). An SbSI:Sb2S3 layerwas deposited using the aforementioned procedure. HTM solu-tion (10 mg mL−1 PCPDTBT in o-dichlorobenzene) was thenAdvanced Functional Materials, 2026tive Commons LicenseFIGURE 4 TDPV decays of (a) the control device without interlayer, (b) with PVA at the TiO2/ASIS interface, (c) with PVA at the ASIS/PCPDTBTinterface, (d) with PSS at the TiO2/ASIS interface, and (e) with PSS at the ASIS/PCPDTBT interface at 50%RH. The intensities of the 375 and 515 nm lightswere modulated to yield same photocurrent values, as shown in Figure S11. (f) Response time (τUV) of the devices with different structures calculated byusing double exponential fitting. Photovoltage response of (g) control, (h) PVA onASIS, and (i) PSS onASIS devices under periodic 375 nmUV irradiation(0.5 ± 0.1 s) with constant 515 nm irradiation. The UV irradiation was started after the photovoltage under 515 nm had stabilized. The insets show thefitting analysis of the recovery time after UV irradiation using double exponential equation.sF4IFic(wA 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Creapin-coated at 2000 rpm and annealed at 100◦C for 10 min.inally, a 70 nm thick Au electrode was deposited..3 Hydrophilic Polymer Interlayerncorporationor the devices with PVA or PSS, a thin polymer interlayer wasntroduced at the TiO2/ASIS and/or ASIS/HTM interface by spinoating (2000 rpm, 30 s) the corresponding polymer solutionPVA: 5 mg mL−1 dissolved in DMSO, PSS: 0.5 vol% dilutedith H2O) onto the underlying layer, followed by drying, beforedvanced Functional Materials, 2026depositing the subsequent layer. A PSS solution was prepared bydiluting an 18 wt% PSS aqueous solution.4.4 MeasurementsPhotoabsorption spectroscopy (PAS) was performed using aJASCO V-730 spectrometer. Current–voltage curves were mea-sured using a source meter unit (ADCMT Corp., 6241A) under a100 mW cm−2 Xe lamp (Asahi Spectra MAX-303). The EQE spec-tra were measured using a Bunko Keiki SM-250KD instrumentequippedwith aKeithley 2401 sourcemeter. Themonochromated7 of 10tive Commons LicenseTABLE 1 Summary of photovoltage and photocurrent response characteristics of TiO2/ASIS/HTM (control), TiO2/ASIS/PVA/HTM, andTiO2/ASIS/PSS/HTM devices.Control PVA-incorporated PSS-incorporatedR515 (mAW−1) a 125.0 ± 28.1 42.7 ± 0.4 79.5 ± 17.2τUV (s) b 2.56 ± 1.03 0.95 ± 0.25 0.25 ± 0.13ΔV (mV) c 35 ± 2.2 45 ± 2.0 63 ± 3.6τRec (s) d 14.1 11.6 3.12aResponsivity (R = I/P, where I and P represent the photocurrent and light power, respectively) under 515 nm light irradiation.bUV response time (Figure 4f).cUV-induced photovoltage change, anddRecovery time from Figure 4g–i.FIGURE 5 Dependence of voltage change ratio (ΔV/V0) of a PSS-incorporated device upon the simultaneous irradiation of lights ofdifferent colors and intensities. ΔV and V0 are the voltage change uponshort 375 nm irradiation (0.5± 0.1 s) and base photovoltage under constant530 nm irradiation, respectively, measured using the procedure describedin Figure 4g–i. The background color is a fitting by radial basis function(RBF) interpolation. The data of a control device are also shown in FigureS13.l(4ShKmOiTwsmwoah8 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Creaight power was calibrated using a silicon photovoltaic (PV) cellBunko Keiki model S1337-1010BQ)..5 In Situ XASMeasurementsoft X-ray absorption measurements were performed underumid conditions at the BL-2B beamline of the Photon Factory,EK (Tsukuba, Japan). The partial fluorescence yield (PFY)ode using a silicon drift detector (SDD) was employed for theK-edge, Ti L3-edge, and Sb M-edge XAS measurements byntegrating the fluorescence lines originating from each edge.he energy resolution of the SDD detector (Amptek, FAST SDD)as limited to approximately 100 eV. The film sample with theize of 5 mm × 5 mm was loaded onto a dedicated gas-floweasurement cell (see Figure S2). A 150 nm thick SiC membraneas placed on the film sample. The measurement cell was fixednto the measurement window. A humidity sensor, gas inlet,nd gas outlet were connected to the cell to precisely control theumidity [42–44]. Humid feed gas was allowed to flow throughof 10a small space between the film sample and the SiC membrane.After the humidity reached the target value, it was maintainedfor more than 10 min to promote H2O adsorption and desorptionbefore starting themeasurement. TheOK-edge XAS spectra werenormalized only for the incident X-ray intensity (I0) and baselineremoval. The energy axes of the XAS spectra were corrected usingthe peak from the TiO2 reference [36].4.6 TDPV and TDPCMeasurementsA PV device was fabricated for TDPV and TDPC. The cathodeand anode of the device were connected to an oscilloscope anda source meter, respectively. The output voltage of the sourcemeter was set to ground (0 V). The intensities and the spot sizesof the 375 and 515 nm continuous wave lasers were modulatedusing neutral density (ND) filters and a concave lens (f= 50 mm),respectively. A 2 mm × 2 mm mask was placed in front of thedevice. The termination of the oscilloscope was 1 MΩ for theTDPVmeasurements and 50 Ω for the TDPCmeasurements. Thecontrol and sample data were obtained at approximately the samehumidity (50% RH) with using a chamber to exclude the effect ofvapor.4.7 StatisticsThe average and standard deviation (SD) of the photodetectionperformances were calculated, and the values were described asmean ± SD in the corresponding figure caption or table. Thenumber of data points for each statistic is also indicated.AcknowledgementsThis work was supported by the Japan Society for the Promotionof Science (JSPS) with the KAKENHI Grant-in-Aid for ScientificResearch (A) (JP24H00484), KAKENHI Grant-in-Aid for Early-CareerScientists (JP23K13826), KAKENHI Grant-in-Aid for Scientific Research(B) (JP25K01857), KAKENHI Grant-in-aid for Scientific Research (A):Dynamic Exciton (JP20H05836), KAKENHI Grand-in-Aid for ScientificResearch on Innovative Area: Aquatic Functional Materials (JP19H05717and JP22H04541), KAKENHI Grant-in-Aid for Transformative ResearchAreas (A) “Supra-ceramics” (JP22H05145 and JP23H04626), Japan Scienceand Technology Agency (JST) CREST (JPMJCR23O2), JST SPRING(JPMJSP2138), a research grant from The Mazda Foundation, YashimaAdvanced Functional Materials, 2026tive Commons LicenseEmo2CTDTcRB(2cL3IA4nh5“R6dF27“oF8H19Hd1FC11MAa1LE11St1NLA 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. 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 Creanvironment Technology Foundation. Soft X-ray absorption measure-ents were performed at the Photon Factory BL-2B with the approvalf the Photon Factory Program Advisory Committee (Proposal No.023G662).onflicts of Interesthe authors declare no conflicts of interest.ata Availability Statementhe data that support the findings of this study are available from theorresponding author upon reasonable request.eferences1. P. Wang, S. S. Liu, W. J. Luo, et al., “Arrayed Van Der Waalsroadband Detectors for Dual-Band Detection,” Advanced Materials 292017): 1604439, https://doi.org/10.1002/adma.201604439.. S. Lefler, R. Vizel, E. Yeor, et al., “Multicolor Spectral-Specific Sili-on Nanodetectors Based on Molecularly Embedded Nanowires,” Nanoetters 18 (2018): 190–201, https://doi.org/10.1021/acs.nanolett.7b03873.. E. A. Plis, N. Gautam, M. N. Kutty, et al., “Performance of Long-Wavenfrared InAs/GaSb Strained Layer Superlattice Detectors for the Spacepplications,” Proceedings of the SPIE 8164 (2011): 81640S.. W. Shockley, “The Theory of p-n Junctions in Semiconductors and p-Junction Transistors,” Bell System Technical Journal 28 (1949): 435–489,ttps://doi.org/10.1002/j.1538-7305.1949.tb03645.x.. F. S. Goucher, G. L. Pearson, M. Sparks, G. K. Teal, and W. Shockley,Theory and Experiment for a Germanium p−n Junction,” Physicaleview 81 (1951): 637–638, https://doi.org/10.1103/PhysRev.81.637.2.. R. Nishikubo, S. Li, and A. Saeki, “Unprecedented Wavelength Depen-ence of an Antimony Chalcohalide Photovoltaic Device,” Advancedunctional Materials 32 (2022): 2201577, https://doi.org/10.1002/adfm.02201577.. T. Kobayashi, R. Nishikubo, Y. Chen, K. Marumoto, and A. Saeki,Wavelength-Recognizable SbSI:Sb2S3 Photovoltaic Devices: Elucidationf the Mechanism and Modulation of Their Characteristics,” Advancedunctional Materials 32 (2023): 2311794.. R. Nie, H. S. Yun,M. J. Paik, et al., “Efficient Solar Cells Based on Light-arvesting Antimony Sulfoiodide,” Advanced Energy Materials 8 (2018):701901, https://doi.org/10.1002/aenm.201701901.. R. Nie, J. Im, and S. I. Seok, “Efficient Solar Cells Employing Light-arvesting Sb0.67Bi0.33SI,”AdvancedMaterials 31 (2019): 1808344, https://oi.org/10.1002/adma.201808344.0. Y. C. Choi, E. Hwang, andD.H. Kim, “Controlled Growth of SbSI Thinilms from Amorphous Sb2S3 for Low-Temperature Solution Processedhalcohalide Solar Cells,” APLMaterials 6 (2018): 121108, https://doi.org/0.1063/1.5058166.1. K. Mistewicz, W. Matysiak, M. Jesionek, et al., “A Simple Route foranufacture of Photovoltaic Devices Based on Chalcohalide Nanowires,”pplied Surface Science 517 (2020): 146138, https://doi.org/10.1016/j.psusc.2020.146138.2. S. Inagaki, M. Nakamura, H. Hatada, et al., “Growth of Visible-ight-Responsive Ferroelectric SbSI Thin Films by Molecular Beampitaxy,” Applied Physics Letters 116 (2020): 072902, https://doi.org/10.063/1.5142642.3. K. T. Butler, J. M. Frost, and A. Walsh, “Ferroelectric Materials forolar EnergyConversion: PhotoferroicsRevisited,”Energy&Environmen-al Science 8 (2015): 838–848, https://doi.org/10.1039/C4EE03523B.4. K. Mistewicz, M. Nowak, R. Paszkiewicz, and A. Guiseppi-Elie, “SbSIanosensors: FromGel to Single Nanowire Devices,”Nanoscale Researchetters 12 (2017): 334, https://doi.org/10.1186/s11671-017-1854-x.dvanced Functional Materials, 202615. M. Tamilselvan and A. J. Bhattacharyya, “Antimony Sulphoiodide(SbSI), a Narrow Band-Gap Non-Oxide Ternary Semiconductor withEfficient Photocatalytic Activity,” RSC Advances 6 (2016): 105980–105987.16. C. Wang, M. Zhang, Y. Fang, et al., “SbSI Nanocrystals: An ExcellentVisible Light Photocatalyst with Efficient Generation of Singlet Oxygen,”ACS Sustainable Chemistry & Engineering 6 (2018): 12166–12175, https://doi.org/10.1021/acssuschemeng.8b02498.17. S. Dhar, P. Chakraborty, T. Majumder, and S. P. Mondal, “CdS-Decorated Al-Doped ZnO Nanorod/Polymer Schottky Junction UV–VisDual-Wavelength Photodetectors,” ACS Applied Nano Materials 1 (2018):3339–3345.18. Y. Zhang, W. Xu, X. Xu, J. Cai, W. Yang, and X. Fang, “Self-PoweredDual-Color UV–Green Photodetectors Based on SnO2 Millimeter Wireand Microwires/CsPbBr3 Particle Heterojunctions,” The Journal of Physi-cal Chemistry Letters 10 (2019): 836–841, https://doi.org/10.1021/acs.jpclett.9b00154.19. M. Dai, H. Chen, R. Feng, et al., “A Dual-Band Multilayer InSeSelf-Powered Photodetector with High Performance Induced by SurfacePlasmon Resonance and Asymmetric Schottky Junction,” ACS Nano 12(2018): 8739–8747, https://doi.org/10.1021/acsnano.8b04931.20. L. Li, H. Chen, Z. Fang, et al., “An Electrically Modulated Single-Color/Dual-Color Imaging Photodetector,”AdvancedMaterials 32 (2020):1907257, https://doi.org/10.1002/adma.201907257.21. Z. Lan and F. Zhu, “Electrically Switchable Color-Selective OrganicPhotodetectors for Full-Color Imaging,”ACSNano 15 (2021): 13674–13682.22. Y. Chang, L. Huang, Y. Zhou, J.Wang, andW. Zhai, “Controlled Local-ized Phase Transition of Selenium for Color-Selective Photodetectors,”ACS Applied Materials & Interfaces 14 (2022): 5624–5633, https://doi.org/10.1021/acsami.1c22909.23. P. Szaniawski, J. Lindahl, T. Törndahl, U. Zimmermann, andM. Edoff,“Light-Enhanced Reverse Breakdown in Cu(In,Ga)Se2 Solar Cells,” ThinSolid Films 535 (2013): 326–330, https://doi.org/10.1016/j.tsf.2012.09.022.24. L. Min, Y. Zhou, H. Sun, et al., “Carrier Dynamic IdentificationEnables Wavelength and Intensity Sensitivity in Perovskite Photodetec-tors,” Light: Science & Applications 13 (2024): 280, https://doi.org/10.1038/s41377-024-01636-6.25. A. Maurano, C. G. Shuttle, R. Hamilton, et al., “Transient Optoelec-tronic Analysis of Charge Carrier Losses in a Selenophene/FullereneBlend Solar Cell,” The Journal of Physical Chemistry C 115 (2011):5947–5957, https://doi.org/10.1021/jp109697w.26. G. Juska, K. Arlauskas, M. Viliu, K. Genevicius, R. Österbacka,and H. Stubb, “Charge Extraction by Linearly Increasing Voltage: ANew Method for Determining Carrier Mobility in Thin Films,”AmericanPhysical Society 62 (2000): 235–238.27. A. J. Mozer, G. Dennler, and A. Pivrikas, “Time-Dependent Mobilityand Recombination of the Photoinduced Charge Carriers in ConjugatedPolymer/Fullerene BulkHeterojunction Solar Cells,” Physical Review B 72(2005): 035217, https://doi.org/10.1103/PhysRevB.72.035217.28. T. Nagamori and K. Marumoto, “Direct Observation of Hole Accu-mulation in Polymer Solar Cells during Device Operation Using Light-Induced Electron Spin Resonance,” Advanced Materials 25 (2013): 2362–2367, https://doi.org/10.1002/adma.201204015.29. M. Namatame, M. Yabusaki, T. Watanabe, Y. Ogomi, S. Hayase,and K. Marumoto, “Direct Observation of Dramatically Enhanced HoleFormation in a Perovskite-Solar-Cell Material Spiro-OMeTAD by Li-TFSIDoping,” Applied Physics Letters 110 (2017): 123904, https://doi.org/10.1063/1.4977789.30. M. Suzuki, S. Murayama, N. Umesaki, and T. Tanaka, “Hydroxyl-Group Identification Using O K-Edge XAFS in Porous Glass FabricatedbyHydrothermal Reaction and Low-Temperature Foaming,”Molecules 24(2019): 3488, https://doi.org/10.3390/molecules24193488.31. S. Benkoula, O. Sublemontier, M. Patanen, et al., “Water Adsorptionon TiO2 Surfaces Probed by Soft X-ray Spectroscopies: Bulk Materials vs.9 of 10tive Commons Licensehttps://doi.org/10.1002/adma.201604439https://doi.org/10.1021/acs.nanolett.7b03873https://doi.org/10.1002/j.1538-7305.1949.tb03645.xhttps://doi.org/10.1103/PhysRev.81.637.2https://doi.org/10.1002/adfm.202201577https://doi.org/10.1002/aenm.201701901https://doi.org/10.1002/adma.201808344https://doi.org/10.1063/1.5058166https://doi.org/10.1016/j.apsusc.2020.146138https://doi.org/10.1063/1.5142642https://doi.org/10.1039/C4EE03523Bhttps://doi.org/10.1186/s11671-017-1854-xhttps://doi.org/10.1021/acssuschemeng.8b02498https://doi.org/10.1021/acs.jpclett.9b00154https://doi.org/10.1021/acsnano.8b04931https://doi.org/10.1002/adma.201907257https://doi.org/10.1021/acsami.1c22909https://doi.org/10.1016/j.tsf.2012.09.022https://doi.org/10.1038/s41377-024-01636-6https://doi.org/10.1021/jp109697whttps://doi.org/10.1103/PhysRevB.72.035217https://doi.org/10.1002/adma.201204015https://doi.org/10.1063/1.4977789https://doi.org/10.3390/molecules24193488i13HS3M(3tV(3XR3ST13P(i3Pao3oC14Efh4SX14tta4aLH24YO3SAIS1 16163028, 2026, 67, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.76809 by Hisao Kiuchi - National Institute For , Wiley Online Library on [20/08/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are gsolated Nanoparticles,” Scientific Reports 5 (2015): 15088, https://doi.org/0.1038/srep15088.2. A. R. Abid, M. Mailhiot, N. Boudjemia, et al., “The Effect of Relativeumidity on CaCl2 Nanoparticles Studied by Soft X-ray Absorptionpectroscopy,” RSC Advances 11 (2021): 2103–2111.3. Y. B. Kwon, B. M. Weon, K. H. Won, J. H. Je, Y. Hwu, and G.argaritondo, “X-ray-Induced Changes in Wettability,” Langmuir 252009): 1927–1929, https://doi.org/10.1021/la804081k.4. A. Nilsson, D. Nordlund, I. Waluyo, et al., “X-ray Absorption Spec-roscopy and X-ray Raman Scattering of Water and Ice; An Experimentaliew,” Journal of Electron Spectroscopy and Related Phenomena 1772010): 99–129, https://doi.org/10.1016/j.elspec.2010.02.005.5. D. Nordlund, H. Ogasawara, K. J. Andersson, et al., “Sensitivity of-ray Absorption Spectroscopy to Hydrogen Bond Topology,” Physicaleview B 80 (2009): 233404, https://doi.org/10.1103/PhysRevB.80.233404.6. M. A. Van Spronsen, X. Zhao, M. Jaugstetter, et al., “Interfaceensitivity in Electron/Ion Yield X-ray Absorption Spectroscopy: TheiO2–H2O Interface,” Journal of Physical Chemistry Letters 12 (2021):0212–10217.7. F. Fasulo, G. Piccini, A. B. Muñoz-García, M. Pavone, and M.arrinello, “Dynamics of Water Dissociative Adsorption on TiO2 Anatase101) at Monolayer Coverage and below,” The Journal of Physical Chem-stry C 126 (2022): 15752–15758, https://doi.org/10.1021/acs.jpcc.2c03077.8. K. R. Hahn, A. Tricoli, G. Santarossa, A. Vargas, and A. Baiker, “Firstrinciples Analysis of H2OAdsorption on the (110) Surfaces of SnO2, TiO2nd Their Solid Solutions,” Langmuir 28 (2012): 1646–1656, https://doi.rg/10.1021/la204124p.9. T. Petit, J. Ren, S. Choudhury, et al., “X-Ray Absorption Spectroscopyf TiO2 Nanoparticles in Water Using a Holey Membrane-Based Flowell,” Advanced Materials Interfaces 4 (2017): 1700755, https://doi.org/10.002/admi.201700755.0. R. Nishikubo, H. Kanda, I. Garcia-Benito, et al., “Optoelectronic andnergy Level Exploration of Bismuth and Antimony-Based Materialsor Lead-Free Solar Cells,” Chemistry of Materials 32 (2020): 6416–6424,ttps://doi.org/10.1021/acs.chemmater.0c01503.1. Y. T. Alharbi, F. Alam, A. Salhi, M. Missous, and D. J. Lewis, “Directynthesis of Nanostructured Silver Antimony Sulfide Powders fromMetalanthate Precursors,” Scientific Reports 11 (2021): 3053, https://doi.org/10.038/s41598-021-82446-3.2. K. Yamazoe, Y. Higaki, Y. Inutsuka, et al., “Enhancement ofhe Hydrogen-Bonding Network of Water Confined in a Polyelec-rolyte Brush,” Langmuir 33 (2017): 3954–3959, https://doi.org/10.1021/cs.langmuir.7b00243.3. R. Watanabe, T. Sakamoto, K. Yamazoe, J. Miyawaki, T. Kato,nd Y. Harada, “Ion Selectivity of Water Molecules in Subnanoporousiquid-Crystalline Water-Treatment Membranes: A Structural Study ofydrogen Bonding,”Angewandte Chemie International Edition 59 (2020):3461–23465.4. K. Yamazoe, Y. Higaki, Y. Inutsuka, J. Miyawaki, A. Takahara, and. Harada, “Critical In-Plane Density of Polyelectrolyte Brush for therdered Hydrogen-Bonded Structure of Incorporated Water,” Langmuir8 (2022): 3076–3081, https://doi.org/10.1021/acs.langmuir.1c02895.upporting Informationdditional supporting information can be found online in the Supportingnformation section.upportingFile: adfm76809-sup-0001-SuppMat.pdf.0 of 10 Advanced Functional Materials, 2026overned by the applicable Creative Commons Licensehttps://doi.org/10.1038/srep15088https://doi.org/10.1021/la804081khttps://doi.org/10.1016/j.elspec.2010.02.005https://doi.org/10.1103/PhysRevB.80.233404https://doi.org/10.1021/acs.jpcc.2c03077https://doi.org/10.1021/la204124phttps://doi.org/10.1002/admi.201700755https://doi.org/10.1021/acs.chemmater.0c01503https://doi.org/10.1038/s41598-021-82446-3https://doi.org/10.1021/acs.langmuir.7b00243https://doi.org/10.1021/acs.langmuir.1c02895 Water Adsorption-Induced Color Sensor: Insight Into the Sensing Mechanism and Interfacial Engineering for Improved Responsiveness 1 | Introduction 2 | Results and Discussion 2.1 | Mechanism of WDPE 2.1.1 | Analysis of Water Penetration Behavior via In Situ XAS 2.1.2 | Role of the TiO2 Layer on WDPE 2.2 | Improvement of Color-Sensing Characteristics 3 | Conclusion 4 | Experimental Section 4.1 | SbSI:Sb2S3 (ASIS) Film Preparation 4.2 | Photovoltaic Device Fabrication 4.3 | Hydrophilic Polymer Interlayer Incorporation 4.4 | Measurements 4.5 | In Situ XAS Measurements 4.6 | TDPV and TDPC Measurements 4.7 | Statistics Acknowledgements Conflicts of Interest Data Availability Statement References Supporting Information