# Fileset

[d4cp02012j.pdf](https://mdr.nims.go.jp/filesets/f7363c52-5bb5-4aad-a98c-cb6e42a87750/download)

## Creator

[Goichiro Seo](https://orcid.org/0000-0003-2912-2848), [Ryoma Hayakawa](https://orcid.org/0000-0002-1442-8230), [Yutaka Wakayama](https://orcid.org/0000-0002-0801-8884), Ryosuke Ohnuki, Shinya Yoshioka, [Kaname Kanai](https://orcid.org/0000-0002-3952-5491)

## Rights



## Other metadata

[Mechanism of charge accumulation in potassium poly(heptazine imide)](https://mdr.nims.go.jp/datasets/d4e96456-0496-4b53-8ebc-c18be91a54ea)

## Fulltext

Mechanism of charge accumulation in potassium poly(heptazine imide)This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  20585Cite this: Phys. Chem. Chem. Phys.,2024, 26, 20585Mechanism of charge accumulation in potassiumpoly(heptazine imide)†Goichiro Seo, *a Ryoma Hayakawa, b Yutaka Wakayama, b Ryosuke Ohnuki,aShinya Yoshiokaa and Kaname Kanai aDark photocatalysis is the ability of a photocatalyst to accumulate photocarriers during light irradiationand consume them for redox reactions under dark conditions. This phenomenon of photocatalystsstoring photocarriers is known as charge accumulation. Dark photocatalysts can mitigate fluctuations inphotocatalytic reaction efficiency in response to fluctuating solar irradiation. Potassium poly(heptazineimide) (K-PHI) has attracted considerable attention due to its high photocatalytic efficiency and ability toundergo dark photocatalysis. However, the detailed mechanism of charge accumulation in K-PHIremains unclear because photochromism, potassium ion desorption, and charge accumulation occursimultaneously triggered by light irradiation, limiting the comprehensive understanding of thismechanism. To elucidate the charge accumulation mechanism in K-PHI, highly oriented K-PHI thin filmswere prepared. Then, their fundamental physical properties and optical response of their electricalproperties were investigated. We succeeded in separately observing photochromism, potassium iondesorption, and charge accumulation induced by light irradiation on K-PHI and proposed acomprehensive model to explain these phenomena. This study not only provides insights into theunique physical phenomena exhibited by K-PHI but also contributes to the development of solarenergy–storage materials in the future.IntroductionThe realization of a low-carbon society has fueled the study onvarious materials, such as photocatalysts, in recent years toefficiently use renewable energy. Photocatalytic applicationshave been widely explored, and efforts are being taken towardtheir sustainability, particularly in photocatalytic water split-ting for generating hydrogen or reducing carbon dioxide toproduce useful substances.1–4 However, these applicationsrequire continuous light irradiation, and the photocatalyticactivity may be affected due to fluctuations in solar radiationdue to changes in weather conditions.5 Thus, photocatalyticactivity is restricted to only daytime, limiting the application ofrenewable energy resources such as photovoltaics.6 Suchunstable photocatalytic activities can hinder the applicationof futuristic facilities, wherein atmospheric carbon dioxide willbe converted into fuel or water into hydrogen.7,8 This issuecan be addressed using dark photocatalysis or via persistentphotocatalysis, which can be performed in the dark.9,10 Therapid decrease in photocatalytic conversion efficiencies due tofluctuations in solar radiation can be moderated using darkphotocatalysts.Dark photocatalysis is a phenomenon in which a photoca-talyst accumulates charges under light irradiation and con-sumes the accumulated charge under dark conditions to oxidizeor reduce targets. Transition metal oxide composites10–15 havethus far been studied for dark photocatalysis; however,poly(heptazine imide) (PHI), a carbon nitride polymeric com-pound, was recently discovered to exhibit this property.16–18PHI has two advantages of being inexpensive to synthesizebecause it does not contain transition metals and very easy touse because it exhibits dark photocatalytic activity on its ownwithout the need to combine it with other materials.19,20Typical dark photocatalysts are composed of a photocatalystand pseudo-capacitor.9 Under light irradiation, photocarriersgenerated in the photocatalyst by light absorption are trans-ferred to the pseudo-capacitor bonded to it, causing chargeaccumulation. These accumulated charges are then consumed,leading to photocatalytic activity even under dark conditions.In this sequence, the phenomenon of charge accumulationplays an essential role in dark photocatalytic activity, andbesides accumulation in a pseudo-capacitor, in some cases it alsoa Department of Physics and Astronomy, Faculty of Science and Technology,Tokyo University of Science, 2641 Yamazaki, Noda, Chiba 278-8510, Japan.E-mail: 6222701@ed.tus.ac.jpb Research Center for Materials Nano architectonics (MANA), National Institute forMaterials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cp02012jReceived 15th May 2024,Accepted 8th July 2024DOI: 10.1039/d4cp02012jrsc.li/pccpPCCPPAPEROpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article OnlineView Journal  | View Issuehttps://orcid.org/0000-0003-2912-2848https://orcid.org/0000-0002-1442-8230https://orcid.org/0000-0002-0801-8884https://orcid.org/0000-0002-3952-5491http://crossmark.crossref.org/dialog/?doi=10.1039/d4cp02012j&domain=pdf&date_stamp=2024-07-22https://doi.org/10.1039/d4cp02012jhttps://doi.org/10.1039/d4cp02012jhttps://rsc.li/pccphttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jhttps://pubs.rsc.org/en/journals/journal/CPhttps://pubs.rsc.org/en/journals/journal/CP?issueid=CP02603020586 |  Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 This journal is © the Owner Societies 2024involves reduction of transition metals or charge accumulation indefect levels. PHI shows photocatalytic activity in the dark anddoes not require other materials to function as a pseudo-capacitoror contain transition metals in its structure. PHI also containssome amount of defects, similar to other carbon nitride materials;however, retaining the charge trapped in defects for a long time isdifficult.21–25 Therefore, the mechanism known so far for darkphotocatalysts cannot be simply applied to the mechanism of PHIdark photocatalytic activity. By elucidating the dark photocatalyticactivity of PHI and prolongation of charge accumulation time,transition metal-free dark photocatalysts can be developed in thefuture.Several models have been proposed for understanding thedark photocatalytic activity of PHI;26,27 however, all of them arebased on the conventional mechanism of dark photocatalyticactivity, and the role of PHI-specific phenomena such asphotochromism and desorption of potassium ions with lightirradiation has not been clarified. To understand the specificdark photocatalytic activity of PHI, its physical propertiessuch as electronic structure and electrical properties must bemicroscopically investigated as photochromism, potassiumion desorption and charge accumulation are closely related.However, the experimental techniques applicable to conven-tional PHI powder samples are limited, making microscopicinvestigation difficult. Herein, a method for fabricating K-PHIthin films (Fig. 1) was proposed, and their physical propertiessuch as crystal, chemical, and electronic structures and elec-trical properties were investigated. X-ray photoelectron spectro-scopy (XPS) and ultraviolet photoelectron spectroscopy andinverse photoemission spectroscopy (UPS/IPES) results haverevealed the fundamental properties of K-PHI, including thenumber of water molecules in the channel of the K-PHI crystal,electronic structure near the energy gap, ionization energy, andelectron affinity. The results of the electrical properties of K-PHIrevealed that it exhibits photochromism and K+ desorbs fromits structure upon light irradiation. As the electrical conductionof K-PHI is attributed to both ionic conduction by K+ andcarrier conduction by electrons and holes, selectively observingthe charge accumulation phenomenon is difficult. However, wesuccessfully distinguished between the two properties using aK-PHI thin film and observed charge accumulation in theexcited state. This result is consistent with that predicted fromour previously proposed charge accumulation model, i.e., theshallow trap charge accumulation (STCA) model.28 We discussthe STCA model that explains charge accumulation in K-PHI bycombining the results of the optical response of the basicphysical and electrical properties of K-PHI.Experimental and theoreticalMaterialsA single crystal of melem hydrate was obtained after melaminecalcination (5 g, purity: 99.0%, Wako Pure Chem., Ind., Ltd,139-00945) in a tube furnace (KTF035N1/Koyo Thermo SystemsCo.) by increasing the temperature to 310 1C at 1 1C min�1,followed by holding for 5 hours and then naturally cooling. Theresulting product was washed with N,N-dimethylformamide(DMF) and acetone and dried at 150 1C for 24 h. The driedsample was dissolved in dimethylsulfoxide (DMSO) to obtainmelem hydrate single crystals via vapor diffusion. Melon wassynthesized in the same way as melem in a tube furnace.Melamine (3 g) was placed at the bottom of a quartz test tubeand calcined by increasing the temperature to 550 1C at1 1C min�1 and holding for 5 h, followed by decreasing thetemperature at 2 1C min�1. K-PHI was obtained by mixingmelon (0.3 g) and potassium thiocyanate (KSCN, 0.15 g, purity:98.0%, Wako Pure Chem., Ind., Ltd, 164-04555) and calciningthem in a tube furnace by increasing the temperature at30 1C min�1 and holding for 1 h, followed by further increasingthe temperature at 30 1C min�1 and holding for 30 min; then,the temperature was decreased at 2 1C min�1.Thin film fabricationThe as-synthesized melem hydrate single crystals were placedin an aluminum crucible and evaporated onto an Au (300 nm)/Si(100) substrate in an ultra-high vacuum (10�6 Pa) to fabricatea 30 nm thin film. K-PHI was dispersed in N-methyl-2-pyrrolidone (NMP) and sonicated for 60 h for chemical clea-vage. The resulting dispersion was separated in a centrifugeovernight, and the supernatant liquid was considered as theK-PHI nanosheet dispersion (Fig. S1(b), ESI†). To this, chloro-benzene was mixed at a ratio of 60 vol%, and the nanosheet wasprecipitated via centrifugation overnight. The resulting concen-trated nanosheet liquid was dropped onto the substrate toobtain the K-PHI thin film. Subsequently, UV-O3 cleaning wasperformed (UV253E; Filgen) for 30 min, and the cleaningchamber was replaced with oxygen before cleaning. 3, 4, 9,10-Perylenetetracarboxylic diimide (PTCDI, 495% TokyoChemical Industry Co., Ltd P0984) film was fabricated by30 nm deposition in an ultra-high vacuum in the same wayas melem film on the Au (300 nm)/Si(100) substrate.Fig. 1 Molecular structure of K-PHI.Paper PCCPOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jThis journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  20587Measurement methodsX-ray diffraction (XRD) analysis was performed (Smart lab,Rigaku), with Cu Ka (0.15496 nm) as the X-ray source. Trans-mission electron microscopy (TEM) was performed using aJEM-2100F (JEOL), with an acceleration voltage of 80 kV andspherical aberration correction Cs = 0.5 mm, using Cu300 mesh(EM Japan Co., Ltd U1015-F10) as the grid. XPS was performedusing JPS-9030 (JEOL), with monochromatized Al Ka (hn =1486.6 eV) as the X-ray source. UPS equipment was created inthe laboratory using SES200 (SCIENTA) as the analyzer, and HeIa (hn = 21.22 eV) as the light source. IPES was similarlyperformed on the same sample on PSP vacuum technologyequipment. The band-pass detector comprised a NaCl-coatedCanneltron behind a SrF2 window and detected photons at hn =9.3 eV. Pulse IV measurements were performed using sourcemeasurement units (Keysight Technologies, B2912B). Electricalimpedance spectroscopy (EIS) was performed using an LCRmeter (Keysight Technologies, E4980A). All the measurementswere performed using a 4-probe system at room temperatureunder atmospheric conditions. The pyZwx software was used toobtain the impedance spectrum and fitted under appropriateequivalent circuit assumptions.29 These electrical measure-ments were performed on an Au (20 nm)/K-PHI/Au/Si devicestructure where the top Au electrode had a diameter andthickness of 0.1 mm and 20 nm, respectively.Density functional theory calculationsThe density of states (DOS) and partial DOS (PDOS) wereall obtained via density functional theory (DFT) simulations(Materials Studio) performed using Dmol3 as the module, DNPas the Dmol3 basis function, and SCAN, an m-GGA, as theelectron correlation approximation method.30–32 The cutoffradius of the orbitals was set to 10 Å for all calculations, andthe number of k-point meshes was set to only G points for the‘Orbital’ calculation and 9 � 9 � 16 or 1296 points for the othercases. The cutoff radius was determined from the changein total energy, ionization energy, and electron affinity withincreasing cutoff radius from 2 Å to 13 Å.Results and discussionStructural and chemical state analysis of K-PHIIn this paper, before discussing the charge accumulationphenomenon of K-PHI, the K-PHI thin films fabrication andtheir basic properties are discussed in detail. The mechanismof the charge accumulation of K-PHI is then discussed below,based on the results of experiments using K-PHI thin films assamples. K-PHI thin film was first synthesized, followed by itsnanosheets. Specifically, the synthesized K-PHI powder wassonicated in NMP for 60 h and centrifuged. The obtainedsupernatant liquid was dispersed with K-PHI nanosheets.Fig. 2(a)–(c) show the high-resolution transmission electronmicroscopy (HRTEM image) of the K-PHI nanosheet, magnifiedimage, and FFT image of the HRTEM image. The K-PHI nano-sheet has a size of B50 nm and with its c axis perpendicular tothe grid. Fig. 2(b) shows that white dots are periodicallyarranged in the image, with their periodicity corresponding tothe yellow circled points labeled 100 at 0.89 nm�1 in the FFTimage shown in Fig. 2(c). This indicates that the lattice pointsare arranged three-fold symmetrically in the K-PHI nanosheetwith a spacing of 1.1 nm, which is equal to the distancesbetween the channels in the K-PHI structure and the symmetryof their arrangement. Furthermore, the red circles labeled110 at 1.56 nm�1 show a three-fold symmetry, which are tiltedby 301 with respect to the yellow circles. This result is consistentwith a lattice constant of 0.64 nm in the PHI layer of K-PHI.Fig. 2(d) and (e) show the simulated TEM and FFT images of K-PHI, respectively. Fig. 2(d) shows a magnified view of thesimulated TEM image and the molecular structure of K-PHI.The lattice of white dots corresponds to the arrangement ofchannels present in the PHI layer of K-PHI. The simulated FFTimage (Fig. 2(e)) shows hexagonally aligned points, as inFig. 2(c), and their distances from the center are also consistentwith those in Fig. 2(e). These results indicate that the K-PHInanosheet retains periodically aligned channels and its struc-ture is equivalent to that of the PHI layer of K-PHI. K-PHI thinfilms were fabricated herein by depositing K-PHI nanosheetson the Au substrate. The photographs and UV-vis spectra of thefabricated K-PHI thin films are shown in Fig. S3 (ESI†). Fig. 2(f)shows the XRD results of the as-prepared K-PHI thin film.Fig. 2 (a) and (b) High-resolution transmission electron microscopy(HRTEM) images of K-PHI nanosheets and its magnified view. (c) FastFourier transform (FFT) image of the HRTEM image. Yellow and red circlesare located 0.89 and 1.56 nm�1, respectively, from the center. (d) and (e)TEM simulation based on the model structure of K-PHI and its FFT image.The yellow and red circles correspond to those in (c). (f) Thin-film XRDresults of K-PHI thin films fabricated by depositing K-PHI nanosheets onthe Au substrate. The out-of-plane and in-plane results and the simulateddiffraction pattern are shown from top to bottom. The inset shows that theK-PHI nanosheet is oriented perpendicular to the substrate.PCCP PaperOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012j20588 |  Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 This journal is © the Owner Societies 2024The out-of-plane and in-plane measurements and diffractionpattern of K-PHI are shown from top to bottom in the samefigure. The in-plane measurements show peaks at 2y = 7.981,13.81, 21.21, and 27.91. In particular, the d-spaces of the twolow-angle peaks are 1.1 and 0.64 nm, which are consistent withthe periodicity observed in Fig. 2(b) and (c). The XRD patternsof the K-PHI thin film do not match those of the K-PHI powdersample (Fig. S3(b) and (c), ESI†).17 The XRD profile of the K-PHIthin film is well explained by the crystal structure model witha = b = 901, g = 1201, a = b = 12.78 Å and c = 3.20 Å. Comparedwith the simulated diffraction pattern, the in-plane measure-ments show (100), (110), (210), and (001) peaks at low 2y.In particular, the (001) peak appears strongly in the out-of-plane results, indicating in the K-PHI nanosheet being orientedperpendicular to the substrate. Although K-PHI nanosheetfabrication has been studied previously, its isolation and thinfilm fabrication have not been investigated yet.33,34 Filmsfabricated from nanosheets of titanium oxides or niobiumoxides are oriented perpendicular to the substrate;35,36 however,the fabrication of oriented films from nanosheets of 2D-COF havenot been reported yet. Thus, to the best of our knowledge, thisstudy is the first to report such a method.The upper panel of Fig. 3(a) shows the N 1s spectrum of theK-PHI thin film and the lower panel shows the N 1s spectrum ofthe melem film. The fitting analysis results and the observedspectra are also shown; the assignments of the peaks used inthe analysis is based on Akaike et al.’s study.37 Specifically,three types of N 1s peaks can be observed: a nitrogen peak witha C–NQC bond (peak Ni), a nitrogen peak linking an aminogroup or a heptazine ring (peak Nii), and a nitrogen peakattached to three carbons at the center of the heptazine ring(peak Niii). As K-PHI and melem have similar chemical struc-tures, they have similar N peaks. The binding energies of Ni, Nii,and Niii peaks are almost equal, but their intensity ratios aredifferent. The Ni : Nii : Niii ratios for melem and K-PHI werefitted to be 6 : 3 : 1 and 12 : 3 : 2, respectively, which agree wellwith the observed data. These ratios are ideal values for eachsample, as shown by the molecular structures in Fig. 3(a),indicating that melem and K-PHI thin films have fewer impu-rities such as intermediate products. A more detailed examina-tion of the binding energies of the peaks shows that the Ni andNiii peaks have shifted by B0.1 eV between K-PHI and melem,whereas the Nii peak has shifted by B0.4 eV. This chemicalshift was caused by a change in the chemical state of Nii froman amino group to an imide bond: the hydrogen bonded to thenitrogen was replaced by carbon with higher electronegativityand the nitrogen was slightly positively charged, resulting in achemical shift to the higher binding energy side. The Nii peakintensity of K-PHI was smaller than that of melem becauseduring K-PHI synthesis, the amino groups in melem underwenta polymerization reaction that caused the desorption of ammo-nia and a decrease in the number of Nii peaks. In other words,these changes in the Nii peak indicate that K-PHI was producedvia the polymerization of melem. Fig. 3(b) and (c) show the O/Nand C/N values obtained from the analysis of the N 1s, C 1s,and O 1s XPS spectra. The average matrix relative sensitivityFig. 3 (a) The upper panel shows the XPS N 1s spectrum of the K-PHI thin film, and the lower panel shows the XPS N 1s spectrum of melem. The opencircles are measured data, and the solid lines and filled peaks are the results of fitting analysis. The attribution of each peak resolved by fitting analysis isshown in correspondence with the molecular structure, i.e., the color of each peak corresponds to the color of the circle surrounding the nitrogen atomin the molecular structure. The ratio of each peak area obtained from the analysis is also shown. Details of the background subtraction and correction forbinding energy are described in Note 2 (ESI†). (b) and (c) Oxygen to nitrogen ratios and carbon to nitrogen ratios of K-PHI and melem obtained from thequantitative analysis of XPS using the AMRSF method. Details of the quantitative analysis of XPS using the AMRSF method are given in Note 2 of the ESI.†Error bars in K-PHI are expanded uncertainties calculated with a coverage factor of k = 2 and have a confidence level of about 95%. (d) Model structure ofK-PHI deduced from the analysis results in (b) and (c) by considering the estimated number of atoms of each element per unit cell of K-PHI and thecalculated element ratios. All atoms in the molecular structure in (d) are shown according to the space-filling model with van der Waals radius.Paper PCCPOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jThis journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  20589coefficients were used for elemental sensitivity correction basedon the international standards.38 The PTCDI-deposited film,which contained carbon, nitrogen, and oxygen, was used as thereference sample. The error bars are expanded uncertaintiesbased on standard uncertainties obtained from the measure-ments of the five K-PHI thin films, with k = 2 as the inclusionfactor and a confidence level of B95%. Quantitative analysisand the related method are described in Note 2 in the ESI.† Asshown in Fig. 3(b), the C/N ratio of K-PHI is larger than that ofmelem because the amount of nitrogen in the sample decreasesas the amino group is desorbed from melem during poly-merization from melem to K-PHI. The C/N values of K-PHIare consistent with the ideal values of K-PHI within the experi-mental error. The XPS intensity ratios within the same andbetween different elements show that the K-PHI thin film isindeed a polymer with a heptazine ring as the building unit anda theoretically predicted K-PHI structure. Lotsch et al. usednuclear magnetic resonance (NMR) to identify the chemicalstructure of K-PHI and this work firstly reports the identifi-cation of chemical structure of K-PHI by quantitative analysisof XPS results.17 As shown in Fig. 3(c), K-PHI has higher O/Nvalues than melem. The oxygen detected in melem is possiblyderived from water adhering to the sample surface; thus, K-PHIcan be assumed to contain water molecules in its channels thatare unique to the crystal structure with the heptazine-basedframework. In other words, the large O/N ratio of K-PHI ispresumed to be due to the water-derived oxygen in the channelsof K-PHI. This is because of the humidity dependence of theionic conductivity and the requirement of water molecules inthe channel for a reasonable interpretation of the XRD resultsof K-PHI.17,39 These water molecules were not directly observedin the K-PHI. The C/N and O/N results are summarized inFig. 3(d). The number of carbon and oxygen atoms werecounted with respect to the number of nitrogen atoms, whichwas calculated by multiplying the C/N and O/N values by thenumber of nitrogen atoms. The K-PHI structure based on thecalculated number of atoms contained in the K-PHI structure isshown on the left side of Fig. 3(d) using a space-filling modelwith van der Waals (vdW) radii. Note that the channel containsapproximately five water molecules. The structure of K-PHI inthe left figure shows that the channel is almost entirelyoccupied by these five water molecules and potassium ions.Therefore, the average number of potassium ions in the chan-nel of K-PHI is estimated to be at most one. Thus, the proposedstructural model from our previous research may have to becorrected.28,40 Our previous report used a model where theaverage number of potassium ions in the channel was three,which is inconsistent with the results in Fig. 3 of this paper.However, DFT simulations show that the energy gap was alsoreduced when one potassium ion bound to the heptazine-basedframework in the K-PHI channel was desorbed due to lightirradiation (Fig. S7(c), ESI†). This indicates that K-PHI exhibitsphotochromism associated with potassium ion desorption evenif each channel contains only one potassium ion on average.To summarize the discussion in this section, the K-PHInanosheet is a polymer with a heptazine-based frameworkand each channel of the PHI layer contains, on average, fivewater molecules and one potassium ion. This result is the firstclear indication of the number of water molecules and potas-sium ions in the K-PHI channels, which remains unclear forpowdery K-PHI.Electronic structure of K-PHIFig. 4(a) shows the UPS/IPES spectra of the K-PHI and melemfilms. The K-PHI thin film was cleaned with UV-O3 for 30 minto remove any surface contamination. The effect of cleaningis discussed in Fig. S2 and S6 (ESI†); however, the K-PHInanosheet was not damaged during the process. As shown inFig. 4(a), the blue data denote the observed UPS/IPES spectra,the black solid lines denote the smoothed raw data, and thehorizontal axis shows the binding energy measured from thevacuum level. The solid red lines show the spectra with amagnified view near the energy gap, where Ith, Ath and Eg arethe ionization energy, electron affinity threshold and energygap, respectively. Here, Ith and Ath are determined by the crosspoints between a baseline and a tangent at the inflection pointsof the first peaks in the UPS and IPES spectra, respectively.Eg can be calculated based on the energy difference between Ithand Ath. The gray hatch-filled spectra denote the DOS obtainedfrom DFT simulations, and the solid and dotted black lines inthe hatch represent the N 2p and C 2p PDOS. The structuralmodel determined from the results in Fig. 1 and 2 was used forDFT calculations (Note 3, ESI†). A notable difference in the sizeof the energy gap is observed in the spectra of K-PHI andmelem: the energy gap of K-PHI is B1.5 eV smaller than that ofmelem because of the polymerized structure of K-PHI withmelem as the building unit. This observation is consistent withthat of Fig. 2. Moreover, the Ith values of K-PHI and melem werenot considerably different, but the Ath of K-PHI was consider-ably smaller than that of melem. To illustrate the changes inionization energy and electron affinity, Fig. 4(b) shows theenergy diagrams for melem, melon, and K-PHI. The left axisrepresents the binding energy measured from the vacuum level,whereas the numbers in the figure represent Ith and Ath. The Ithand Ath of melon were cited from a previous study.37 The reasonfor comparing melem, melon and K-PHI here is that thesecompounds have in common that they are based on theheptazine ring, which makes it easy to compare the differencesin electronic structure due to the degree of polymerization.Fig. 4(b) shows that the energy gap decreases in the order ofmelem, melon, and K-PHI. For melon and K-PHI, the redoxlevels of water are located within the energy gap. Moreover,melon and K-PHI exhibit photocatalytic activity in visible light,which corresponds with the results reported herein.16,17,41The Ith and Ath values of melem, melon, and K-PHI are shownin Fig. 4(b). The difference in the Ith and Ath values of thesematerials are at most B0.2 and B1.4 eV, respectively. ThePDOS in Fig. 4(a) shows that in both K-PHI and melem, aconduction band is formed by almost equal contributions fromnitrogen and carbon, whereas a valence band is formed almostexclusively by nitrogen. This suggests that the wave function ofthe conduction band is delocalized over the molecule or thePCCP PaperOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012j20590 |  Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 This journal is © the Owner Societies 2024polymeric framework, whereas that of the valence band islocalized to nitrogen. Fig. 4(c) shows the wave functions ofthe highest occupied molecular orbital (HOMO), HOMO�1, thelowest unoccupied molecular orbital (LUMO), and LUMO+1 ofmelem and the wave functions of the highest energy band(valence band maximum (VBM)), the second highest energyband (VBM�1) in the valence band, the lowest energy band(conduction band maximum (CBM)), and the second lowestenergy band (CBM+1) in the conduction band of K-PHI. TheLUMO and LUMO�1 of melem are delocalized over the hepta-zine ring via a p-bond between nitrogen and carbon. In con-trast, the wave functions of HOMO�1 and HOMO are localizedto nitrogen, indicating that these orbitals are formed by thelone pairs of nitrogen. These features can also be observed in K-PHI with the same heptazine structure as a building unit. Thus,when melem polymerizes to form K-PHI, the wave function ofthe valence band of K-PHI remains localized to the lone pair ofnitrogen. On the other hand, the LUMO of melem becomesdelocalized and a conduction band of K-PHI is formed. Thus,the ionization energy of K-PHI is almost similar to that ofmelem and only its electron affinity increases.The preceding sections cover the fundamental physicalproperties of K-PHI thin film have been discussed from theviewpoints of crystallography, chemistry, and electronic struc-ture. The subsequent sections will report on the electricalproperties of K-PHI thin films.Electrical properties of K-PHIFig. 5(a) shows the results of I–V measurements for the K-PHIthin films. The electric field swept in the order of A, B, C, D, and E.Fig. 4 (a) UPS and IPES (UPS/IPES) spectra of K-PHI and melem films. The K-PHI thin film was cleaned with UV-O3. The horizontal axis plots the bindingenergy measured from the vacuum level. The blue lines show the raw data, the solid black lines show the smoothed spectra of the raw data, and the solidred lines are magnified spectra near the energy gap. Ith, Ath, and Eg represent the ionization energy, electron affinity, and energy gap, respectively. Thegray filled spectra show the density of states (DOS) obtained via DFT calculations. The solid black and dotted black lines in the gray region represent theN 2p and C 2p derived partial DOS, respectively. (b) Energy diagrams with Ith and Ath for melem, melon, and K-PHI. The left axis shows the binding energymeasured from the vacuum level. The dotted lines indicate the redox levels of water. (c) Wave functions of HOMO�1, HOMO, LUMO, and LUMO+1for melem and the wave functions of the highest energy band VBM, the second highest energy band VBM�1 in the valence band, the lowest energyband CBM and the second lowest energy band CBM+1 in the conduction band of K-PHI, where the red and green lobes represent the positive andnegative signs of the wave functions. VBM and CBM are the valence band maximum and conduction band minimum, respectively. The nitrogen, carbon,hydrogen, and potassium ions are represented by blue, gray, white, and purple spheres, respectively.Paper PCCPOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jThis journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  20591The variation in current in the segment BC is particularlynoteworthy. In segment AB, the current increases proportionallyto voltage, whereas in segment BC, the current becomes negativeas the voltage decreases. This indicates that the current flows inthe opposite direction to the applied voltage, which cannot beexplained by the carrier conduction of electrons and holes orsimple ionic conduction. A similar phenomenon also occurs insegment DEB, indicating that segment BCDE has a characteris-tic hysteresis curve. These I–V properties can be explained byconsidering the ionic polarization in K-PHI. As shown in Fig. 2,the K-PHI thin film has a structure of K-PHI nanosheets stackedon the substrate, so the potassium ions can move vertically inthe channel with respect to the substrate and electrode surface.Therefore, when the voltage is applied to the top and bottomsurfaces of the film, potassium ions move toward the cathodebut only slightly shift from the position where they remaincharge neutral. The K-PHI thin film shows a large polarizationat point B because all the potassium ions in the film move,although the distance traveled by a potassium ion is not large. Insegment BC, the polarization built up in the thin film is releasedas the voltage decreases, thereby generating a current flow in thedirection opposite to the applied voltage. In other words, anegative current flow in the external circuit when potassiumions, which have been shifted from their charge-neutral positionby the applied voltage, are attracted by the PHI anion and returnto their original position. These scenarios are shown in detail inFig. S10 (ESI†). These findings indicate that the current observedin Fig. 5(a) are polarization currents generated due to the ionicpolarization of potassium ions. However, the carrier conductionof electrons and holes in K-PHI cannot be selectively observed innormal I–V measurements because of the non-negligible effect ofthe hysteresis in the polarization currents. To directly observecarrier conduction in K-PHI, ionic conduction must be sup-pressed. Herein, the pulse I–V measurement method was usedto suppress the hysteresis caused by the polarization currentduring I–V measurements. The corresponding results are shownin Fig. 5(b). The detailed pulse I–V measurement conditions,such as the interval between pulse voltages, are shown inFig. S11 (ESI†). Contrary to normal I–V measurements, aninterval exists between pulse voltages in pulse I–V measurementsduring which no voltage is applied. During this interval, thepotassium ions in the K-PHI thin film return to their originalposition before the voltage application, allowing the sample toreturn from the polarized state to the neutral state. This methodof voltage application ensures that it is always applied to theneutral state of the K-PHI during pulse I–V measurements, suchthat hysteresis associated with ion polarization does not occur.Fig. 5(b) shows that the current flowing in the forward andbackward directions of the voltage sweep are almost identical,without significant hysteresis. In the pulse I–V curve, the slopechanges around the threshold voltage (Vth).I = aVm (1)Eqn (1) is a general form that represents the voltage depen-dence of current, where I is the current, V is the voltage, and a isan arbitrary constant. Therefore, Ohmic polarization occurs inthe voltage region lower than Vth for the current with m = 1 ineqn (1). Contrarily, carrier conduction occurs in the voltageregion higher than Vth, which is expressed as m 4 1. When thecarrier injection from the electrode is not rate-limiting, carrierconduction in organic semiconductor can be described gener-ally with m C 2 according to the child rule. However, in thevoltage region where trap filling is rate-limiting (trap-fillinglimited), m 4 2.42,43 In perovskite materials, where both ionicand carrier conduction is observed, as in K-PHI, m varies basedon the balance between ionic and trap densities in thesample.44,45 Carrier conduction occurs in the K-PHI thin filmin the high voltage region where m 4 1. Moreover, K-PHIexhibits ionic conduction, but we found that carriers can beinjected into K-PHI and that it exhibits carrier and ionicconduction.39 The pulse I–V measurement results for the K-PHI thin film in Fig. 5(b) show a change in the current flow dueto light irradiation. Two changes were observed in the pulse I–Vcurves before and after light irradiation. (1) In the low-voltageregion, the intercept of the pulse I–V curve after light irradia-tion increases. The intercept of the I–V curve in double loga-rithmic graphs in Fig. 5(b) corresponds to a. This indicates thatthe ionic conductivity of the K-PHI thin film increases withlight irradiation. This result is consistent with the desorption ofpotassium ions in K-PHI due to light irradiation reportedFig. 5 (a) Results of I–V measurements of the K-PHI thin film. I is plottedagainst V. The electric field is swept in the order A, B, C, D, and E. (b) Thedouble logarithmic graph of pulse I–V measurement results for the K-PHIthin film. The yellow and blue circles show the pulse I–V curve before andafter light irradiation, respectively.PCCP PaperOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012j20592 |  Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 This journal is © the Owner Societies 2024previously.28,46 Here, it is beyond the scope of this study todiscuss the pH values inside the channel of K-PHI, but H3O+resulting from the dissociation of water molecules and partiallypresent H+ in the channel would be expected to show highconductivity. The effect of variations in local pH values in thechannel on ionic conduction of K-PHI will need to be investi-gated experimentally in the future, but it is not reasonable toassume here that these conductivities will increase upon lightirradiation. (2) Vth before light irradiation is 5.0 V, whereas thatafter light irradiation considerably reduces to 2.4 V. Thisdecrease in Vth indicates that light irradiation has reducedthe carrier injection barrier from the Au electrode to the K-PHI thin film. In general, the carrier injection barrier isdetermined by the energy difference between the Fermi energyof the electrode metal and the ionization energy or electronaffinity of the sample. As shown in Fig. S12 (ESI†), the energygap of K-PHI is narrowed by photochromism. Thus, a smallerdifference in the Fermi energies of the top of the valence bandand the bottom of the conduction band and Au electrode isobserved. Consequently, Vth is shifted to lower voltages due to asmaller carrier injection barrier caused by the narrowing of theenergy gap by light irradiation. Thus, photochromism occurs inK-PHI thin films due to the stimulation by light irradiation,which increases ionic conductivity and decreases the carrierinjection barrier. These changes caused by light irradiationconsiderably alter the optical and electrical properties of K-PHIthin films.The increase in ionic conductivity upon light irradiation inthe K-PHI thin film was further investigated by performingelectrical impedance spectroscopy (EIS). Fig. 6(a) shows theresponse of the impedance magnitude (|Z|) and phase differ-ence (y) to light irradiation. Here, |Z| is the magnitude of theimpedance in the complex plane, and y is the angle withrespect to the real axis of the complex plane. The yellowhatched area indicates the area where the sample was irra-diated with light. As shown in Fig. 6(a), the phase difference is901 before light irradiation; therefore, the impedance hasalmost only an imaginary component and the K-PHI respondssimilar to a capacitor. After light irradiation, the phase differ-ence is 151 and the imaginary component in the impedance isconsiderably reduced to almost a real component, making theK-PHI respond in a similar way to a resistor. Typically, in theabsence of an externally applied voltage, many materialsrespond in a similar way to a capacitor because of the absenceof charge carriers within them that can respond to an appliedAC voltage. However, if carriers are present in materialsthat can respond to external fields, such as mobile ions, theyrespond in a similar way to a resistor. K-PHI responds in asimilar way to a capacitor before light irradiation, as is the casewith many materials; however, it responds as a resistor uponlight irradiation because the potassium ions detach from theheptazine-based framework and become mobile. Fig. 6(b)shows the EIS results at different light intensities. The EISmeasurements were performed at equilibrium with saturatedimpedance at 1000 Hz (Fig. S13 (ESI†)). Without light irradiation,the EIS shows a typical response of a capacitor. In contrast, the EISspectra obtained at light intensities of 0.24 W m�2 and above showa semicircle that can be fitted by a Randles-type equivalent circuit,with the diameter of the semicircle decreasing with increasinglight intensity. For the Randles-type equivalent circuit, the dia-meter of the semicircle represents the resistance (Rp) of thesample. Fig. 6(b) shows that the resistance of the K-PHI thin filmdecreases with increasing light intensity: the ionic conductivity ofthe K-PHI thin film increases with the light intensity. The con-ductivity is plotted against light intensity in Fig. 6(c), which wascalculated from Rp obtained from the results of the fitting analysisof each EIS spectrum. The results show that the ionic conductivityincreases linearly with light intensity. It has been reported that theionic conductivity of K-PHI can be controlled by humidity, andthe results here show that the ionic conductivity of K-PHI can alsobe controlled by the light intensity.39Fig. 6(a) shows the variation in |Z| upon light irradiation,wherein the time constant (t) of the impedance change isconsiderably different during and after light irradiation. Duringlight irradiation, t = 66 s, whereas after light irradiation, t =826 s, i.e., approximately 12 times longer. The change in |Z| iscaused by the desorption and recombination of potassiumions. However, it is difficult to regard this asymmetry in t asbeing caused solely by the thermal motion of the ions. To clarifythis finding, other factors that inhibit the recombination ofFig. 6 (a) Response of impedance magnitude (|Z|) and phase difference (y) at 1000 Hz to light irradiation. The yellow hatched areas indicate time framesduring which light was irradiated and not irradiated. |Z| and y in the complex plane are shown in the figure. (b) Electrical impedance spectrum at each lightintensity. The circuit in the diagram is the equivalent circuit used for fitting. (c) Light intensity dependence of conductivity obtained from the fittinganalysis of the data in (b). The blue dotted line shows the result of linear fitting.Paper PCCPOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jThis journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  20593potassium ions with the heptazine-based framework are consid-ered. The desorption of potassium ions from the heptazine-basedframework is caused by the transition of K-PHI to an excited stateupon light irradiation. Conversely, as long as K-PHI is in an excitedstate, potassium ions will not recombine into the heptazine-basedframework. On the other hand, as mentioned in the introduction,K-PHI is a material that exhibits charge accumulation upon lightirradiation. Since the charge accumulation phenomenon is causedby maintaining the excited state generated by light absorption for along time, the recombination between potassium ions and theheptazine-based framework does not proceed quickly in K-PHIafter light irradiation and the impedance gradually changes.Mechanism of charge accumulation in K-PHIIn this section, we proposed the model to explain the charge-accumulated state (CA) of K-PHI. The ionic conductivity inK-PHI increases upon light irradiation due to photochromismand potassium ion desorption. In addition, CA appears as astable intermediate state. These phenomena in K-PHI areclosely related to each other, which have been explained bydeveloping a model that can comprehensively explain each ofthese phenomena. Fig. 7(a) shows an energy diagram. Thehorizontal axis represents the bonding distance between thenitrogen and potassium ions in K-PHI. The experimental factwe obtained in this study is attempted to be explained usingthis energy diagram. Based on this diagram, we discuss (A) theincrease in ionic conductivity and (B) photochromism of K-PHIupon light irradiation. As shown in Fig. 4(c), the wave functionof the VBM of K-PHI is localized in the nitrogen bound to thepotassium ion, whereas the wavefunction of the excited state ofthe CBM is spread over the entire heptazine ring. When theelectrons of nitrogen in the bridge site, which attracted potas-sium ions, are excited from the VBM to the CBM, the electrondensity distribution changes considerably and become deloca-lized over the heptazine-based framework. Thus, the bondbetween the potassium ion and nitrogen is cleaved and thepotassium ion is desorbed from the heptazine-based frame-work, causing the system to transition from Y* to B*, followedby the deactivation of B* to B. Here, B* and B are the blue statesresulting from the narrowing of the energy gap of K-PHI due topotassium ion desorption, indicative of photochromism.In many photochromic materials, these changes in molecularstructure caused by electron density redistribution are respon-sible for photochromism.47,48 In B, where the N–K bond iscleaved, the energy gap is narrowed and potassium ionsreleased from the heptazine-based framework become moremobile (Fig. S7(c) (ESI†)). These findings explain the phenom-ena (A) and (B) well. The light intensity dependence of (C)impedance and (D) asymmetry of t in the dark and light statesof the K-PHI is subsequently discussed. Photochromism andionic conductivity in K-PHI are correlated and photochromismshows a light intensity dependence, as observed in T-typephotochromic materials.49–51 This phenomenon generallyoccurs because the transition to the colored state by lightirradiation and the transition to the colorless state by thermaldeactivation reach an equilibrium. This equilibrium positionis determined by the balance between the excitation rate dueto light irradiation and thermal deactivation rate, i.e., theFig. 7 (a) Energy diagram showing the energies of the light-irradiated excited state and the subsequent charge accumulated state of K-PHI. Thehorizontal axis represents the bonding distance between nitrogen and potassium ions in K-PHI. Y, Y*, B, and B* in the diagram show the color of thesample due to photochromism, representing the ground state (yellow state), the excited state immediately after light absorption (yellow state),the ground state of the K+-desorbed structure (blue state), and the excited state of the K+-desorbed structure (blue state) respectively. CA represents thestate in which charges are accumulated on the K-PHI. k1, k2, k3, and k4 represent the rate constants for each reaction, and DE represents the activationenergy required to relax from CA to B. (b) The molecular structure of K-PHI in yellow and blue states. The blue color in the molecular structureconceptually depicts the electron density in the ground (Y) and excited (B*) states. (c) The left diagram is a schematic of K-PHI in CA. The energy diagramof K-PHI in CA is shown on the right.PCCP PaperOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012j20594 |  Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 This journal is © the Owner Societies 2024equilibrium position is determined by the light intensity undera constant temperature. A similar equilibrium state also existsbetween states Y and B in K-PHI, suggesting that the impe-dance of K-PHI shows light intensity dependence undera constant temperature. The higher the light intensity, thesmaller the impedance, and the decrease in impedance satu-rates with the irradiation time and reaches an equilibrium(Fig. S13(a) (ESI†)).We discuss how K-PHI reaches a stable intermediate stateCA upon light irradiation. The transition from B to the groundstate Y occurs with thermal deactivation at about room tem-perature. This transition cannot explain the phenomenonshown in Fig. 6(a), where the blue state is maintained fora long time without K+ returning to the heptazine-based frame-work. Furthermore, after deexcitation from B*, the systempresumably transfers directly to the B and then deactivates toY; only the rate constant k2 is involved in the relaxation. Thisindicates that the relaxation process of the impedance in K-PHIin the dark state can be described by a simple single exponen-tial function with respect to time. However, as shown inFig. S13(b) (ESI†), the relaxation process of K-PHI impedancecan only be reproduced by the superposition of two exponentialfunctions, indicating that the excited state of K-PHI undergoestwo relaxation processes. In addition, the dependence of timeconstant of the impedance relaxation on light intensity andirradiation time cannot be explained by considering k2 alone(Fig. S14 (ESI†)). This is because the transition from B to Y iscaused by thermal deactivation and the rate must be indepen-dent of light intensity. These findings indicate that K-PHI doesnot directly transition from B* to B during the relaxationprocess in the dark state but via another state. Therefore, ascenario is proposed in which the relaxation from the excitedstate B* of K-PHI involves a stable intermediate state CA(Fig. 7(c)). CA is a state where photocarriers are accumulatedin the blue K-PHI due to the charge separation of excitons inB*. The stabilization of the system with similar charge trans-fer has also been reported for charge-transfer photochromicmaterials.48,52 K-PHI deexcited by charge separation from B*has a blue state with a smaller energy gap than the yellow state.Therefore, the blue region of K-PHI works as a shallow trap forelectrons and holes as long as it is surrounded by a yellowregion with a larger energy gap (left side of Fig. 7(c)). An energydiagram representing this scenario is shown on the right sideof Fig. 7(c). To avoid the recombination of charge generated bycharge separation and form excitons, the electrons and holesare spatially separated and accumulated in separate blue K-PHIregions (Fig. 7(c)). The charge separation can assume to befacilitated by the electric field gradient created at the boundarybetween the blue and yellow regions. The accumulated chargeis then consumed in the dark state, resulting in the darkphotocatalytic activity of K-PHI. Although models of chargeaccumulation in traps have been developed for other darkphotocatalysts, K-PHI is unique as charge is accumulated in ashallow trap formed by its own photochromism.9,13,14,53We refer to this model as the STCA model.27 Finally, theCA relaxes to B by consuming charge carriers and excitonregeneration and recombination. The charge accumulated inthe blue K-PHI region must overcome the energy barrier DEcorresponding to the energy difference between B and Y to bereleased from the trap. This is governed by the rate constant k4,which explains another relaxation mechanism observed in theimpedance relaxation process upon light irradiation: K-PHIstabilizes in CA during deexcitation, resulting in a large impe-dance relaxation time constant in the dark state, as shown inthe abovementioned experimental fact (D). (D) and the fact thatK-PHI exhibits dark photocatalysis suggests that the transitionfrom CA to B takes a long time and k4 is very small. After theaccumulated charges are released, excitons regenerate andimmediately recombine. Accordingly, K-PHI falls into B, butK+ is reaccommodated in the heptazine-based framework andreturns to the ground state Y as the electron density distri-bution of PHI returns to the same as the ground state Y.Considering its use in dark photocatalysis, if only the holesaccumulated in K-PHI are consumed using a hole scavengersuch as 4-methyl-benzyl-alcohol (4-MBA), the recombinationprobability between electrons and holes is even smaller and thestate CA is maintained for a longer time.16–18ConclusionsIn this study, K-PHI nanosheets were successfully isolated fromK-PHI powder, and K-PHI thin films were subsequently pre-pared from these nanosheets. XRD and XPS results showed thatthe chemical and crystal structures of K-PHI thin film andK-PHI were similar. In addition, UPS/IPES measurements wereperformed to obtain an electronic structure that closely agreedwith DFT simulations, and the energy gap, ionization energy,and electron affinity were precisely measured as the fundamen-tal properties of K-PHI. Its electrical properties were alsodetermined. Pulse I–V measurements indicated that K-PHIwas an ionic and a carrier conductor and the threshold voltagefor carrier conduction shifted due to photochromism. EISmeasurements revealed that the ionic conductivity of K-PHIincreased proportionally to light intensity and an asymmetryexisted in the impedance relaxation process with and withoutlight irradiation, suggesting that K-PHI formed a charge-accumulated state in the dark state. To summarize theseresults, three phenomena induced by light irradiation in K-PHI,i.e., photochromism, potassium ion desorption, and charge accu-mulation, were observed with shifts in threshold voltage in thecarrier conduction region, a proportional increase in ionic con-ductivity with light intensity, and long-lived excited states withmoderated relaxation. Finally, a model was developed to compre-hensively explain these phenomena. The energy diagram suggeststhe bond between the nitrogen and potassium ions is cleavedduring the transition from Y* to B* and the existence of a CAbetween B* and B of K-PHI. The CA is constituted by charges thataccumulate in shallow traps created by photochromism in K-PHI.The discussion of CA is an important insight for understandingand developing dark photocatalyst activity in K-PHI. Moreover, thecharge accumulation phenomenon in K-PHI is unique as it doesPaper PCCPOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jThis journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  20595not require bonding with heterogeneous materials and does notinvolve transition metals. The reported findings thus providea new opening for the future development of energy-storingphotocatalysts.Author contributionsThis manuscript was written through contributions of allauthors.Data availabilityAll data included in this study are available upon request bycontacting the corresponding author.Conflicts of interestThere are no conflicts to declare.AcknowledgementsThe authors gratefully acknowledge Prof. Y. Idemoto and T.Ichihashi for the TEM measurements. This work was supportedby JST SPRING, Grant Number JPMJSP2151 and JSPS KAKENHIGrant Number 23KJ1967.Notes and references1 L. Wang and H. Xu, Two-dimensional conjugated polymerframeworks for solar fuel generation from water, Prog.Polym. Sci., 2023, 145, 101734.2 J. Xu, F. A. Roghabadi, Y. Luo, V. Ahmadi, Q. Wang andZ. Wang, et al., Recent advances in heterogeneous catalysisof solar-driven carbon dioxide conversion, J. Environ. Sci.,2024, 140, 165–182.3 J. D. Xiao, R. Li and H. L. Jiang, Metal–Organic Framework-Based Photocatalysis for Solar Fuel Production, Small Meth-ods, 2023, 7(1), 1–36.4 Q. Xu, Z. Xia, J. Zhang, Z. Wei, Q. Guo and H. Jin, et al.,Recent advances in solar-driven CO2 reduction over g-C3N4-based photocatalysts, Carbon Energy, 2023, 5, 2.5 D. Matuszko, Influence of the extent and genera of cloudcover on solar radiation intensity, Int. J. Climatol., 2012,32(15), 2403–2414.6 D. Schmidt, M. D. Hager and U. S. Schubert, Photo-Rechargeable Electric Energy Storage Systems, Adv. EnergyMater., 2016, 6(1), 1–11.7 N. S. Lewis, An integrated, systems approach to the devel-opment of solar fuel generators, Electrochem. Soc. Interface,2013, 22(2), 43–49.8 I. Ganesh, Conversion of carbon dioxide into methanol -A potential liquid fuel: Fundamental challenges and oppor-tunities (a review), Renewable Sustainable Energy Rev., 2014,31, 221–257.9 J. Y. Y. Loh, N. P. Kherani and G. A. Ozin, Persistent CO2photocatalysis for solar fuels in the dark, Nat. Sustainability,2021, 4(6), 466–473.10 M. Sakar, C. C. Nguyen, M. H. Vu and T. O. Do, Materialsand Mechanisms of Photo-Assisted Chemical Reactionsunder Light and Dark Conditions: Can Day–Night Photo-catalysis Be Achieved?, ChemSusChem, 2018, 11(5), 809–820.11 T. Tatsuma, S. Saitoh, Y. Ohko and A. Fujishima, TiO2-WO3photoelectrochemical anticorrosion system with an energystorage ability, Chem. Mater., 2001, 13(9), 2838–2842.12 P. M. Stanley, F. Sixt and J. Warnan, Decoupled Solar EnergyStorage and Dark Photocatalysis in a 3D Metal–OrganicFramework, Adv. Mater., 2023, 35, 1.13 P. Feng, X. Tang, J. Zhang, Y. Mei and H. Li, Persistentphotocatalysis effect of black peony-like BiOCl and itspotential full-time photocatalytic applications, RSC Adv.,2017, 7(53), 33241–33247.14 J. Li, Y. Liu, Z. Zhu, G. Zhang, T. Zou and Z. Zou, et al., A full-sunlight-driven photocatalyst with super long-persistentenergy storage ability, Sci. Rep., 2013, 3, 2–7.15 X. Fu, Y. Kong, M. Wang, T. Cai and Q. Zeng, MXene derivedTi3C2/TiO2/Ag persistent photocatalyst with enhanced elec-tron storage capacity for round-the-clock degradation oforganic pollutant, J. Colloid Interface Sci., 2024, 656,233–240.16 J. Kröger, A. Jiménez-Solano, G. Savasci, P. Rovó,I. Moudrakovski and K. Küster, et al., Interfacial Engineer-ing for Improved Photocatalysis in a Charge Storing 2DCarbon Nitride: Melamine Functionalized Poly(heptazineimide), Adv. Energy Mater., 2021, 11, 6.17 H. Schlomberg, J. Kröger, G. Savasci, M. W. Terban, S. Betteand I. Moudrakovski, et al., Structural Insights intoPoly(Heptazine Imides): A Light-Storing Carbon NitrideMaterial for Dark Photocatalysis, Chem. Mater., 2019,31(18), 7478–7486.18 V. W. Lau, D. Klose, H. Kasap, F. Podjaski, M. C. Pignié andE. Reisner, et al., Dark Photocatalysis: Storage of SolarEnergy in Carbon Nitride for Time-Delayed Hydrogen Gen-eration, Angew. Chem., Int. Ed., 2017, 56(2), 510–514.19 A. Gouder, A. Jiménez-Solano, N. M. Vargas-Barbosa,F. Podjaski and B. V. Lotsch, Photomemristive sensing viacharge storage in 2D carbon nitrides, Mater. Horiz., 2022,9(7), 1866–1877.20 V. Sridhar, F. Podjaski, J. Kröger, A. Jiménez-Solano,B. W. Park and B. V. Lotsch, et al., Carbon nitride-basedlight-driven microswimmers with intrinsic photochargingability, Proc. Natl. Acad. Sci. U. S. A., 2020, 117(40),24748–24756.21 Z. Liu, G. Wang, H. S. Chen and P. Yang, An amorphous/crystalline g-C3N4 homojunction for visible light photocata-lysis reactions with superior activity, Chem. Commun., 2018,54(37), 4720–4723.22 H. Kim, S. Gim, T. H. Jeon, H. Kim and W. Choi, DistortedCarbon Nitride Structure with Substituted Benzene Moietiesfor Enhanced Visible Light Photocatalytic Activities, ACSAppl. Mater. Interfaces, 2017, 9(46), 40360–40368.PCCP PaperOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012j20596 |  Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 This journal is © the Owner Societies 202423 I. Papailias, T. Giannakopoulou, N. Todorova, D. Demotikali,T. Vaimakis and C. Trapalis, Effect of processing temperatureon structure and photocatalytic properties of g-C3N4, Appl.Surf. Sci., 2015, 358, 278–286.24 Q. Tay, P. Kanhere, C. F. Ng, S. Chen, S. Chakraborty andA. C. H. Huan, et al., Defect Engineered g-C3N4 for EfficientVisible Light Photocatalytic Hydrogen Production, Chem.Mater., 2015, 27(14), 4930–4933.25 P. Niu, M. Qiao, Y. Li, L. Huang and T. Zhai, Distinctivedefects engineering in graphitic carbon nitride for greatlyextended visible light photocatalytic hydrogen evolution,Nano Energy, 2018, 44, 73–81.26 A. Rogolino, I. F. Silva, N. V. Tarakina, M. A. R. da Silva,G. F. S. R. Rocha and M. Antonietti, et al., ModifiedPoly(Heptazine Imides): Minimizing H2O2 Decompositionto Maximize Oxygen Reduction, ACS Appl. Mater. Interfaces,2022, 14(44), 49820–49829.27 Y. Markushyna, P. Lamagni, C. Teutloff, J. Catalano, N. Lockand G. Zhang, et al., Green radicals of potassium poly-(heptazine imide) using light and benzylamine, J. Mater.Chem. A, 2019, 7(43), 24771–24775.28 G. Seo, Y. Saito, M. Nakamichi, K. Nakano, K. Tajima andK. Kanai, Mechanism of charge accumulation ofpoly(heptazine imide) gel, Sci. Rep., 2021, 11(1), 1–12.29 K. Kobayashi and T. S. Suzuki, Free analysis and visualiza-tion programs for electrochemical impedance spectroscopycoded in python, Electrochemistry, 2021, 89(2), 218–222.30 J. Sun, A. Ruzsinszky and J. Perdew, Strongly Constrainedand Appropriately Normed Semilocal Density Functional,Phys. Rev. Lett., 2015, 115(3), 1–6.31 B. Delley, From molecules to solids with the DMol3approach, J. Chem. Phys., 2000, 113(18), 7756–7764.32 B. Delley, An all-electron numerical method for solving thelocal density functional for polyatomic molecules, J. Chem.Phys., 1990, 92(1), 508–517.33 C. Zhu, X. Luo, C. Liu, Y. Wang, X. Chen and Y. Wang, et al.,Defect-rich ultrathin poly-heptazine-imide-frameworknanosheets with alkali-ion doping for photocatalytic solarhydrogen and selective benzylamine oxidation, Nano Res.,2022, 15(10), 8760–8770.34 W. Wang, X. Fan, Z. Shu, J. Zhou and D. Meng, Sustainableand mild exfoliation of bulk crystalline carbon nitride intoultrathin nanosheets via ion-exchange in pure-water, Car-bon, 2023, 205, 76–85.35 Y. Shi, M. Osada, Y. Ebina and T. Sasaki, Single dropletassembly for two-dimensional nanosheet tiling, ACS Nano,2020, 14(11), 15216–15226.36 M. Osada, K. Akatsuka, Y. Ebina, H. Funakubo, K. Ono andK. Takada, et al., Robust high-k response in molecularly thinperovskite nanosheets, ACS Nano, 2010, 4(9), 5225–5232.37 K. Akaike, K. Aoyama, S. Dekubo, A. Onishi and K. Kanai,Characterizing Electronic Structure near the Energy Gap ofGraphitic Carbon Nitride Based on Rational Interpretationof Chemical Analysis, Chem. Mater., 2018, 30(7), 2341–2352.38 ISO 18118:2015 Surface chemical analysis—Auger electronspectroscopy and X-ray photoelectron spectroscopy—Guideto the use of experimentally determined relative sensitivityfactors for the quantitative analysis of homogeneousmaterials.39 J. Kröger, F. Podjaski, G. Savasci, I. Moudrakovski,A. Jiménez-Solano and M. W. Terban, et al., ConductivityMechanism in Ionic 2D Carbon Nitrides: From HydratedIon Motion to Enhanced Photocatalysis, Adv. Mater., 2022,34, 7.40 M. Hattori, M. Nakamichi, A. Yamaguchi, C. Miyazaki,G. Seo and R. Ohnuki, et al., Influence of Ion Exchange onPhotoresponsive Properties of Potassium Poly(heptazineimide), Chem. Mater., 2023, 35(3), 1283–1294.41 V. W. H. Lau, I. Moudrakovski, T. Botari, S. Weinberger,M. B. Mesch and V. Duppel, et al., Rational design of carbonnitride photocatalysts by identification of cyanamidedefects as catalytically relevant sites, Nat. Commun., 2016, 7.42 X. Tang, M. Chen, L. Jiang, M. Li, G. Tang and H. Liu,Improvements in Efficiency and Stability of Perovskite SolarCells Using a Cesium Chloride Additive, ACS Appl. Mater.Interfaces, 2022, 14(23), 26866–26872.43 Z. Fang, W. Chen, Y. Shi, J. Zhao, S. Chu and J. Zhang, et al.,Dual Passivation of Perovskite Defects for Light-EmittingDiodes with External Quantum Efficiency Exceeding 20%,Adv. Funct. Mater., 2020, 30(12), 1–9.44 V. M. Le Corre, E. A. Duijnstee, O. El Tambouli, J. M. Ball,H. J. Snaith and J. Lim, et al., Revealing Charge CarrierMobility and Defect Densities in Metal Halide Perovskitesvia Space-Charge-Limited Current Measurements, ACSEnergy Lett., 2021, 6(3), 1087–1094.45 E. A. Duijnstee, J. M. Ball, V. M. Le Corre, L. J. A. Koster,H. J. Snaith and J. Lim, Toward Understanding Space-Charge Limited Current Measurements on Metal HalidePerovskites, ACS Energy Lett., 2020, 5(2), 376–384.46 M. Nakamichi, M. Hattori, G. Seo, R. Ohnuki, S. Yoshiokaand K. Kanai, Formable Optoelectronic Functional Compo-sites with Potassium Poly(heptazine imide), ACS Appl. Elec-tron. Mater., 2022, 4(12), 5747–5751.47 A. Khodko, V. Khomenko, Y. Shynkarenko, O. Mamuta,O. Kapitanchuk and D. Sysoiev, et al., Ultrafast ring-closing reaction dynamics of a photochromic furan-baseddifurylethene, Chem. Phys. Lett., 2017, 669, 156–160.48 Y. Kobayashi and J. Abe, Recent advances in low-power-threshold nonlinear photochromic materials, Chem. Soc.Rev., 2022, 51(7), 2397–2415.49 K. Mutoh, Y. Nakagawa, A. Sakamoto, Y. Kobayashi andJ. Abe, Stepwise two-photon-gated photochemical reactionin photochromic [2.2]paracyclophane-bridged bis(imidazoledimer), J. Am. Chem. Soc., 2015, 137(17), 5674–5677.50 I. Yonekawa, K. Mutoh and J. Abe, Visible light intensitydependent negative photochromism of a binaphthyl-bridged phenoxyl-imidazolyl radical complex, Chem. Com-mun., 2019, 55(9), 1221–1224.51 I. Yonekawa, K. Mutoh, Y. Kobayashi and J. Abe, Intensity-Dependent Photoresponse of Biphotochromic MoleculeComposed of a Negative and a Positive Photochromic Unit,J. Am. Chem. Soc., 2018, 140(3), 1091–1097.Paper PCCPOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012jThis journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 20585–20597 |  2059752 Y. Kobayashi, T. Katayama, T. Yamane, K. Setoura, S. Ito andH. Miyasaka, et al., Stepwise two-photon-induced fastphotoswitching via electron transfer in higher excited statesof photochromic imidazole dimer, J. Am. Chem. Soc., 2016,138(18), 5930–5938.53 Z. Zou, X. Tang, C. Wu, D. Wang, J. Zhang and Z. Ci, et al.,How to tune trap properties of persistent phosphor: Photo-stimulated persistent luminescence of NaLuGeO4:Bi3+,Cr3+tailored by trap engineering, Mater. Res. Bull., 2018, 97,251–259.PCCP PaperOpen Access Article. Published on 09 July 2024. Downloaded on 11/9/2024 11:16:23 AM.  This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by/3.0/http://creativecommons.org/licenses/by/3.0/https://doi.org/10.1039/d4cp02012j