Nanoscale Perovskite‐Sensitized Solar Cell Revisited: Dye‐Cell or Perovskite‐Cell?

Abstract A general and straightforward way of preparing few‐nanometer‐sized well‐separated MAPbIxBr3−x (MA=methylammonium) perovskite photosensitizers on the surface of an approximately 1 μm thick mesoporous TiO2 photoanode was suggested through a two‐step sequential deposition of low‐concentrated lead halides (0.10–0.30 m PbI2 or PbBr2) and methylammonium iodide/bromide (MAI/MABr). When those nanoscale MAPbIxBr3−x perovskites were incorporated as a photosensitizer in typical solid‐state dye‐sensitized solar cells (ss‐DSSCs), it could be verified clearly by the capacitance analysis that nano‐particulate MAPbI3 perovskites play the same role as that of a typical dye sensitizer (MK‐2 molecule) although their size, composition, and structure are different.

Over the last three decades, third-generation solarc ells have evolvedg reatly into different advanced structures by combining variousn ano-or molecular materials and using easys olution-based processing. [1][2][3][4][5][6] In this third generation,t here are mainly two different types of cells depending on the operational mechanism:o ne is as ensitized type, and the other is a film type. Chargeg eneration after light absorption and then charge transportation to each electrode occur in different ma-terials for the former (sensitized), [1,2] but in the same material for the latter (film). [3][4][5][6] The characteristic features of the sensitized cell come mainly from decoupling light harvesting and charget ransporting, thus leadingt o1 )film fabrication at low cost;2 )versatile combinationso fm ain components;a nd 3) effectivea bsorption of diffuse light, making it suitable for running small devices indoors. [1,2,7] The most successful example of as ensitized solar cell came from as eminal report in 1991, in which molecular dyes anchored onto the surface of mesoscopic TiO 2 film as ap hotosensitizer produced significant current by injecting chargesi nto each electron-and hole-transporting materiala fter absorbing incident light. [1,8] This dye-sensitized solar cell (DSSC) also evolved first as al iquid type and then a solid type using solid-state hole-transporting material, among which the most successful one wasSpiro-OMeTAD {2,2',7,7'-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9'-spirobifluorene}. [1,9,10] Later,s emiconducting quantum dots (QDs) also attracted much attention as ar eplacement for typical molecular dyes toward an ew class of inorganic photosensitizer in the same DSSC structure and have shown promising results recently in a liquid-type cell after many trials. [2,11] Even in the history of recent perovskite-based solar cell research, the first few were reported as as ensitized type cell, not ab ulk film type. [12] However,w ithout more detailed investigations on the structures and workingm echanismso fn anoscale perovskite-sensitized cells, almosta ll relatedr esearch efforts have been directed toward bulk perovskite films owing to the extremelyh igh efficiency obtained in the film-type perovskite solar cells. [5,6,13] Meanwhile, molecular dye-or inorganic QD-sensitizeds olidstate (ss) solar cellsb ased on mesoporous metal-oxide films have experiencedavery slow rise in overall powerc onversion efficiencies (PCEs) although their importance was clearly recognized in terms of using stable solid hole-conductors relative to volatile liquid electrolytes. [9, 10, 11b, 14] PCEs of typical ss-DSSCs with Spiro-OMeTAD as the best hole-transporting material (HTM)h aves teady risen up to approximately 7.5 %o ver 1.0-2.0 mmt hick TiO 2 film during the last two decades since the first report of 1.8 %. [9,15] Recently,o ne exceptional case of a "zombie cell" was reported because it was assembled first as a liquid-type DSSC with aCu(I/II) redox couple and was observed to be working as as olid-typec ell after leakageo fe lectrolyte solventa nd drying only with the redox couple left inside mesoporous TiO 2 film. This extraordinary type of cell has displayed approximately1 1% PCE after optimizations over ar elatively thick TiO 2 film composed of approximately 4 mmt ransparent and 4 mms cattering layers. [16] As for QD-sensitized solid cells, the best PCE is approximately 8% with polymer HTMs owing to ar elatively low open-circuit voltage (V oc )f rom many defectinducedr ecombinationsa tt he QD itself or the interfaces with electron-transporting material (ETM) and HTM. [17] Overall, typical dye-or QD-sensitized solid solar cells using at hin mesometal-oxide film less than approximately 2.0 mmt hick have reached6 -8 %P CE so far,a lthough many efforts have been made with ap owerful molecular dye or QDs as ap hotosensitizer situated between am eso-ETMa nd solid HTM layer.T herefore, more ideal photosensitizersw ith stronger absorptivity and better defect tolerance are still neededf or the enhanced PCEs of ss-DSSCs. While pursuing this goal and being inspired by current perovskite bulk film-based solar cell research, we believed that nanoscale perovskites could be agood candidate of sensitizer for ss-DSSCs based on at hin meso-film of metal oxides. Even when considering the very early resultso fp erovskite solar cells and recent reports on nanoscale perovskites, [12,18] this study on nano-perovskite photosensitizers is still in avery early stage and needs amore general preparation methodf or well-defined and predictable nanoscale perovskite photosensitizers and ag eneral workingm echanismb ased on systematic investigations for furtherp rogress.I nt his study, we show as imple but effective route for preparing nanoscale MAPbI x Br 3Àx (MA = methylammonium) perovskite photosensitizers on relatively thin mesoporous TiO 2 films by using at wostep deposition of low-concentrated precursors. When capacitance of these samples as af unctiono ff requency is analyzed and compared with MK-2 sensitizer in standards s-DSSCs,s imilar capacitances are observed, pointing to the chemical capacitance of the mesoporousT iO 2 layer.T hisf inding indicates clearly that nano-perovskites are working as ap hotosensitizer based on the same principle as the typicalm olecular sensitizers in DSSCs.
In the few very early resultso fp erovskite solar cell research that were reported as MAPbI 3 -sensitized type cells, [12] the amount of precursor used was almost the same as that forpreparing currentb ulk perovskite films, usually with av alue of approximately 1.0 m.S oon, it was established that am ore highly concentrated precursor solution ( % 1.3 m)s hould be spincoated over av ery thin mesoporous metal-oxide film (< 300 nm thick) for current record-efficiency optoelectronic films. [5,6,19] However, in our recent reports,i tw as clearly found that relativelyl ow-concentration precursors (Pb 2 + /MAI) are enough to prepare nanoscaleM APbI 3 perovskites over approximately 0.65 mmt hickm eso-TiO 2 film. [18a] Here, as shown in Scheme 1, av ery general two-step methodw as suggested for preparing af ew-nanometer-sized MAPbI x Br 3Àx as ap hotosensitizer by depositing precursorsi nsitu in the cell structure of ss-DSSCs. First, lead halide (PbI 2 or PbBr 2 )w as scattered and adsorbedo nt he bare surface of approximately 1 mmt hick meso-porousT iO 2 film by spin coating as mall aliquot of low-concentrated ( 0.3 m)P bI 2 (or PbBr 2 )s olution (Scheme1b). Then, a small amounto fM AI/MABrs olution was dropped over the spinningP bI 2 -o rP bBr 2 -adsorbed TiO 2 electrode to induce the formation of MAPbI x Br 3Àx on the surfaceo fT iO 2 (Scheme 1c). However,w hen these two low-concentrated precursors were dissolved together in one chemical bath ands pin-coated by one-step deposition, less homogeneous nano-MAPbI 3 perovskites were obtained with al ower PCE compared with the results from the current two-step deposition when tested in the configuration of ss-DSSC( Figure S1 in the Supporting Information). In the one-step procedure, some aggregates could be formed more easily because the solvent( DMF) used also dissolves the deposited materialb etter than tBuOH/chlorobenzene in the two-step process, thus inducing the dissolved material to be aggregated and gathered at the upper part of the film duringt he spin-coating process. Thus, the two-step deposition is more effective in preparing well-defined nanoscale perovskites homogeneously over meso-metal-oxide films in the case of using low-concentrated precursors olutions.
To check the morphological statuso fd eposited MAPbI 3 on meso-TiO 2 filma fter following Scheme 1, electron microscopy analysisu sing SEM and TEM was appliedt oo btain images of TiO 2 /MAPbI 3 at different magnifications. Figure 1a,b shows a top surface and ac ross-sectional area, respectively,a fter spincoating ana liquot of 0.30 m PbI 2 and 0.06 m MAI solution successivelyo ver approximately 1 mmt hick meso-TiO 2 film. They look like bare TiO 2 film without any deposits because the sizes of MAPbI 3 deposit from relatively low concentrations of precursors ( 0.3 m)a re too small to be detected in the typical magnification range of SEM.T hus, the TiO 2 film appears to have no deposits in Figure 1a,b.H owever,w hen the film was scanned with energy-dispersive X-ray spectroscopy (EDX) over the cross-section area, the main elements (Pb and I) of MAPbI 3 perovskitesw ere detectedt ob ed istributed homogeneously along with an overlap of Ti from TiO 2 ,which indicates the presence of nanoscale MAPbI 3 on the surface of TiO 2 even though it was not seen in the SEM images ( Figure S2 in the Supporting Information). More directly,w hen the MAPbI 3 -deposited TiO 2 film was cut into the smallest piece possible and put over a TEM grid for imaging, ac lear distribution of few-nanometer-sized small dots of MAPbI 3 perovskites around TiO 2 particles was seen in high-resolution (HR)TEM ( Figure 1c). The deposition pattern and size of MAPbI 3 look almostt he same as those of typical SILAR (successive ionic layer adsorption and reaction)-deposited QD sensitizers such as CdS, [20] PbS, [21] CdSe, [11b, 22] and others [23] on the surface of meso-metal-oxide films because their formationm echanism is very similart ot he adsorption of the first precursor and subsequent reaction with as econd one for nanoscale growth on the surface. By increasing the concentration from 0.1 to 0.3 m in the first step, the amount of nanoscale MAPbI 3 deposited could be increased gradually as confirmed by enhanced opticald ensity in absorbance ( Figure S3 in the Supporting Information). Morphological and structural analyses by XRD ( Figure S4 in the Supporting Information)clearly confirmed that nanoscale MAPbI 3 perovskites are distributed well on meso-TiO 2 film. When checking both steady and time-resolved photoluminescence, nano-perovskite MAPbI 3 dots on ZrO 2 showedablue shifto fa pproximately 25 nm in the maximum emissionp eak and af aster decay of electron lifetimec omparedt ot hose of at ypical bulk perovskite film (FigureS5i nt he Supporting Information). These could be further evidence of nanoscale perovskite formation by showing as tronger effect of confinement in as maller nanoscale volume. [24] These few-nanometer-sized MAPbI 3 were expectedt op lay the role of ap hotosensitizer comparable to molecular dye or inorganic QDs in the mesoporous film-based sensitized solar cells reported thus far.W hen these nanoscale MAPbI 3 produced by Scheme1 were incorporated into at ypical ss-DSSC with approximately 1 mmt hick TiO 2 film, they showeda pproximately 4-7.2 %P CEs depending on the startingP bI 2 concentrations used (Figure 2a nd Figure S6 in the Supporting Information). With more MAPbI 3 deposits in the nanoscale regime, much incident light is absorbed, and thus more photocurrents are induced as summarized in Ta ble 1. The main reason for PCE enhancement was the increase of short-circuit current (J sc )i n more concentratedp recursors used. The best result with 0.3 m PbI 2 was 7.2 %u nder the standard 1sun condition, comparable to the best ones from dye-or QD-sensitized solid cells with a similar cell structure and materials reported so far. [15b, 17b] When at ypical organic dye (codeda sM K-2) was tested in the same cell-structure for comparison of the sensitizer performance, only approximately 2.6 %P CE was obtained. Moreover,t he current nanoscale perovskite-sensitized cells have not been fully optimized, and there is much room for furtherd evelopment toward moree fficient sensitized cells. This enhancement would be possible by checking main components and their interfaces:1 )the mesoporous electron-transporting layer of the  TiO 2 filmc ould be adjusted to host more nano-perovskite sensitizers effectively by controlling its thickness, pore size, and surfacestates;2)adifferent selection of precursors and deposition routes could be used for more adsorption and charge generation;3 )a more facile infiltration of Spiro-OMeTAD or other HTMa nd its intimatec ontact with nano-perovskites to reduce resistance could be achieved;a nd 4) other optimizations at the interfaces of mesoporous films could be explored to reduce recombination.
To understand the workingp rinciples, impedances pectroscopy characterization was performedf or three low-concentrated precursor-based perovskite cells along with one typical organic dye (MK-2) cell. The same trend was observed for all analyzed samples presenting two demi-arcs. The Nyquist plots of the impedance spectra of all the samples measured at V oc by differentl ight illumination conditions are depicted in Figure S7 in the Supporting Information, and the same pattern is observed. Figure3a, bs hows the Bode plots of the measured capacitance under 0.1 sun illumination at an appliedb ias of 0.2 and 0.5 V, respectively.T hese resultsc learly show that all four kinds of analyzed samples presentt he same capacitance in a range of low and intermediate frequencies (< 10 4 Hz). It is well known that this capacitance in the case of DSSCs,s uch as the MK-2-sensitized cell, corresponds to the chemical capacitance of am esoporousT iO 2 layer. [25,26] This coincidence of capacitan-ces clearly indicates that all samples work as as ensitized type, in which photo-excited electrons in the sensitizer (dye or nano-perovskites) are injected into the mesoporous TiO 2 layer, raising the electron Fermi level in the TiO 2 . [27] Moreover,p hotoexcited holes are injected into the Spiro-OMeTAD acting as a hole transporting layer.T his behaviori sd ifferent from that observed in bulk film-based perovskite solar cells, for which the capacitance in the intermediate-frequency range (10 2 -10 4 Hz) decreases in comparison with the sensitizedt ype by the decrease in the chemical capacitance, whereas the capacitance in the low-frequency region (1-10 2 Hz) increases. [28] The percolation of carriers along the perovskite layer with low-density band-gap states is the origin of the decrease in chemical capacitance observedinbulk filmofperovskite solar cells. In contrast, in Figure 3a,b the same capacity is observed regardless of the amounto rt he type of sensitizer,i ndicating that this is the chemical capacitance of the common part in all samples, that is, the mesoporous TiO 2 layer.
After understanding the workingp rinciples of the fabricated devices,i tw as possible to furtherc haracterizea nd compare the performance of the different typeso fd evices with an analysis of the impedance spectra. [29] The impedance spectra have been fitted using the equivalent circuits for all solid DSSCs. [25,28] The recombination resistance, R rec ,f or the different samples is plotteda td ifferent open circuit potentials (by varying the illumination) in Figure 3c.S imilar trends were obtained from all samples regardless of the type of sensitizer or its amount.B y fitting the slope of Figure 3c,i ti sa lso possible to obtain the ideality factor of the device, n,w hich also provides important information about the recombination mechanisms. [30] In our case, all samples present similar n values of approximately 1.2-1.3, pointingt oasurface-mediated Shockley-Read-Hall recombination. [31] The currentt wo-step deposition of PbI 2 adsorption and its reactionw ith MAI for ar epresentative perovskite MAPbI 3 could be expanded to am ore general formula MAPbI x Br 3Àx for modulating the absorption range over the entire visible spectrum. By adsorbing PbI 2 or PbBr 2 first and then applying am ixture of MAI and MABr in different ratios, different range-sensitizing MAPbI x Br 3Àx were prepared depending on the ratio of MAI and  MABr in the second step. When PbI 2 was adsorbed first and a mixture of MABr/MAI was applied with ag raduali ncrease in the ratio, the current onset point shifted from 800 to 600 nm (Figure 4a). In the case of initial PbBr 2 adsorption, the onset point movedf rom 700 to 550 nm (Figure 4b). Thus, by controlling the first adsorbed metal halide and the appliedr atio of one or two alkyl halides, it is possible to prepare various nanoscale MAPbI x Br 3Àx photosensitizers on meso-TiO 2 film with controlled absorption ranges. From the current results, it seems possible to prepare more versatile nanoscale perovskites, including lead-free ones, over mesoporousf ilms for various applications such as photosensitizers, photocatalysts, light-emitters, and others with at ailor-made sensitizing range by controlling the ratio of halides.
In summary,w ell-defined and controllable nano-MAPbI x Br 3Àx perovskite photosensitizers could be prepared on the surface of approximately 1 mmt hick mesoscopic TiO 2 photoanode throughasimple two-step deposition of low-concentrated precursors ( 0.3 m)i nc ontrast to highly concentrated ones (> 1.0 m)u sed in typical bulk films. UV/Vis absorption, XRD, photoluminescence, and SEM/TEMm easurements all confirm that MAPbI x Br 3Àx perovskite photosensitizers could be grown gradually on TiO 2 film in the nanoscale regime. In aw ay of pursuingm ore ideal and powerful photosensitizers for ss-DSSCs based on relativelyt hin meso-TiO 2 films of 1.0-2.0 mm thickness, these nano-perovskites were incorporated and tested in ss-DSSSc as as ensitizeri nsteado ft ypical molecular dyes and showed ap romising initial PCE of over 7.0 %w ith more optimization points left for further improvements. From impedance spectroscopica nalysis, it wasr evealed clearly that both nano-perovskites and molecular dyes are workinga sa separatep hotosensitizer under the same principle, in which they are only ac harge generatorf or transfer after light absorption, not ac hargea ccumulator for transport in ss-DSSCs. Therefore, these well-separated nano-perovskites on TiO 2 can be considered as an ew dye-like sensitizer with ac omposition and structure of metal halide perovskites. In addition, owing to their appropriate sizes, thesen ano-perovskites could be combined with molecular dye [32] or QDs on mesoscopic TiO 2 film for making more efficient hybrid sensitizers to further improve the performance of ss-DSSCs.