A photofunctional bottom-up bis(dipyrrinato)zinc(II) complex nanosheet

Two-dimensional polymeric nanosheets have recently gained much attention, particularly top-down nanosheets such as graphene and metal chalcogenides originating from bulk-layered mother materials. Although molecule-based bottom-up nanosheets manufactured directly from molecular components can exhibit greater structural diversity than top-down nanosheets, the bottom-up nanosheets reported thus far lack useful functionalities. Here we show the design and synthesis of a bottom-up nanosheet featuring a photoactive bis(dipyrrinato)zinc(II) complex motif. A liquid/liquid interfacial synthesis between a three-way dipyrrin ligand and zinc(II) ions results in a multi-layer nanosheet, whereas an air/liquid interfacial reaction produces a single-layer or few-layer nanosheet with domain sizes of >10 μm on one side. The bis(dipyrrinato)zinc(II) metal complex nanosheet is easy to deposit on various substrates using the Langmuir–Schäfer process. The nanosheet deposited on a transparent SnO2 electrode functions as a photoanode in a photoelectric conversion system, and is thus the first photofunctional bottom-up nanosheet.

Another type of nanosheets, namely molecule-based bottomup nanosheets, is emerging. This series of nanosheets is fabricated directly from atomic, ionic and molecular components. The concept was proposed nearly a century ago, although it has only recently been realized 21 . For example, Schlüter and King created independently single-layer cycloaddition-induced anthracene [22][23][24][25] nanosheets that featured carbon-carbon covalent bonds. 2,2 0 :6 0 ,2 00 -Terpyridine/metal complex nanosheets based on a metal-ligand coordination linkage were reported by Schlüter 26,27 . Frauenrath synthesized a carbon-rich nanosheet using carbonization of an amphiphilic hexayne molecule 28 . Other examples include surface metal-organic frameworks 29,30 , surface covalent-organic frameworks [31][32][33][34] , metal-surface-mediated monolayer formation in the vacuum phase [35][36][37][38][39] , and single-layer or few-layer metal-organic frameworks and covalent-organic frameworks delaminated from the crystal phase [40][41][42][43][44][45][46] . A significant advantage of the bottom-up synthesis is that structures can be customized through the selection of components (for example, metal ions and organic ligand molecules). Therefore, the bottom-up approach may broaden the diversity and utility of nanosheets. Although previous reports on bottom-up nanosheets have concentrated on the fabrication and analysis of various two-dimensional structures, no functionality has yet been demonstrated.
Given this background, we have sought to create functional bottom-up nanosheets, including an electrically conductive nickel bis(dithiolene) nanosheet [47][48][49] . The present work describes a bis(dipyrrinato)zinc(II) complex nanosheet synthesized from a three-way dipyrrin ligand and zinc(II) acetate. The spontaneous and reversible coordination of dipyrrin ligands with metal ions 50,51 makes them suitable as building blocks for bottom-up nanosheets. The bis(dipyrrinato)zinc(II) complex motif acts not only as a connecting point but also as a photofunctional moiety: it has a strong absorption in the visible and near-infrared region (tunable by substituents) [52][53][54] . In addition to the synthesis and identification of the bis(dipyrrinato)zinc(II) complex nanosheet, we report its large domain size (sides of 410 mm), guest inclusion, stepwise layering and photoelectric conversion ability.

Results
Synthesis and morphology of multi-layer N1. Figure 1 depicts bis(dipyrrinato)zinc(II) complex nanosheet N1 synthesized from three-way dipyrrin ligand L1 and zinc(II) acetate. To verify the macroscopic formation of N1, the first process considered was a liquid/liquid interfacial synthesis 47 using an aqueous zinc(II) acetate solution (upper layer, 5.0 Â 10 À 2 mol l À 1 ) and a dichloromethane solution of L1 (lower layer, 1.0 Â 10 À 4 mol l À 1 ; Fig. 2a). A spontaneous reaction at room temperature led to the generation of multilayer N1 at the water/oil interface, which appeared as an orange film. Multi-layer N1 may be transferred from the interface onto various substrates. For example, on an indium tin oxide (ITO) substrate it appears as a transparent film (Fig. 2a). Multi-layer N1 is not soluble in either water or organic solvent, reflecting the polymeric structure proposed in Fig. 1. Optical and scanning electron microscopic images reveal a uniform, flat, film-like morphology (Fig. 2b,c). The presence of cracks and wrinkles also indicates the sheet structure. The typical thickness of multi-layer N1 on a silicon(111) substrate modified with 1,1,1,3,3,3hexamethyldisilazane (HMDS/Si(111)) was observed to be 700 nm by atomic force microscopy (AFM, Fig. 2d), which corresponds to B580 layers considering the thickness of singlelayer N1 (vide infra). The thickness of N1 may be controlled by the concentration of L1 in dichloromethane and may span 6-800 nm (5-670 layers) using L1 solutions with concentrations of 1.0 Â 10 À 6 to 5.0 Â 10 À 4 mol l À 1 (Fig. 2e). We note that the liquid/liquid interfacial synthesis is essential for the synthesis of multi-layer N1. A conventional single-phase synthesis, performed in either dichloromethane at room temperature or N,Ndimethylformamide at 105°C, resulted in a solid material far from a film texture (Fig. 2f,g). In fact, its disordered structure was verified by X-ray photoelectron spectroscopy (XPS, vide infra).
Synthesis of single-layer or few-layer N1. Next, we synthesized single-layer N1. Figure 3a shows the fabrication procedure, an air/ liquid interfacial synthesis [47][48][49] . A very tiny amount (5 ml) of a dilute dichloromethane solution of L1 (7.4 Â 10 À 5 mol l À 1 ) was gently dropped onto the surface of an aqueous phase containing zinc(II) acetate (5.0 Â 10 À 2 mol l À 1 ) at ambient temperature. After prompt evaporation of dichloromethane, spontaneous air/ water interfacial complexation occurred between L1 and zinc(II) ions, and single-layer N1 was produced at the interface. The resultant single-layer N1 was almost invisible to the naked eye but could be transferred onto various substrates via horizontal deposition known as the Langmuir-Schäfer method (Fig. 3).
We note that few-layer N1 was also synthesized by increasing the dichloromethane solution of L1 to 20 ml. The single-layer or fewlayer N1 on a flat substrate was then subjected to a series of analyses.
XPS of N1. XPS was conducted for single-layer N1 on HMDS/ Si(111) to assess its constituent elements (N and Zn) and their bonding properties (Fig. 4a,b). For reference, dipyrrin ligand L1, mononuclear bis(dipyrrinato)zinc(II) complex M1 (Fig. 4c) and multi-layer N1 deposited on highly ordered pyrolitic graphite (HOPG) were also subjected to XPS. Ligand L1 did not show a Zn 2p 3/2 peak, whereas the peak was present at similar binding    55 , whereas M1 shows a single N 1s peak at 398.8 eV, which originates from its coordination to the zinc(II) centre, which makes the two nitrogen atoms equivalent to one another. Both multi-layer N1 and singlelayer N1 displayed a single peak for N 1s at almost the same binding energy as M1 (399.0 eV). In addition, the abundance ratio calculated from the peak area corrected using the photoionization cross-section is consistent with the ideal value of N:Zn  Fig. 1). Also noteworthy is that the products of the single-phase reactions ( Fig. 2f,g) exhibited nitrogen abundancies in excess of the ideal stoichiometric ratio (N:Zn ¼ 5.4:1 and 5.8:1, Supplementary  Fig. 2). This result indicates that the single-phase reaction products leave uncoordinating dipyrrin moieties, thereby possessing disordered structures.
AFM of single-layer or few-layer N1. The AFM analysis revealed the flat sheet texture and morphology of single-layer N1. Figure 5a,b shows the height and phase images, respectively, of single-layer N1 on an HMDS/Si(111) substrate. The phase image clearly distinguishes single-layer N1 and the bare substrate: the nanosheet possesses a phase value that is lower by 2.4°. The height image shows a domain with one side 410-mm long, which is a noteworthy size for bottom-up nanosheets [29][30][31][32][33][34][35][36][37][38][39] . To confirm the single-layer nature of the sheet, part of N1 was scratched with the AFM tip at a force of 8.6 Â 10 2 nN. Figure 5c,d shows the height images before and after scratching, and the scratched region is highlighted with a blue square. This treatment resulted in the removal of N1, leaving an intact HMDS/Si(111) surface. A crosssection analysis traversing one of the scratched edges demonstrated that N1 was 1.2-nm thick (Fig. 5e), which is consistent with the size of the bis(dipyrrinato)zinc(II) complex motif. A higher-force scratch (1.2 Â 10 4 nN) resulted in a more drastic change in the AFM image ( Supplementary Fig. 3). In this case, both single-layer N1 and the HMDS/Si(111) surface were destroyed. Few-layer N1 with larger domain sizes was fabricated by increasing the amount of L1 ( Supplementary Fig. 4). The phase image is flat, whereas the topological image reveals steps and divided domains. This result indicates that the scanned area was completely covered with few-layer N1. The wrinkles in the layers may be evidence of the sheet structure and were removed by thermal annealing at 120°C ( Supplementary Fig. 5).
In-plane periodicity. Scanning tunnelling microscopy (STM) revealed that single-layer N1 on HOPG exhibited a moiré pattern composed of two lattices: one is the in-plane hexagonal periodicity of N1 and the other is derived from HOPG ( Supplementary Fig. 6). The in-plane periodicity was also confirmed by selected area electron diffractions (SAEDs) in transmission electron microscopy for multi-layer N1 ( Supplementary  Fig. 7). They represent two sets of hexagonal diffractions, which are consistent with in-plane diffraction patterns reproduced from crystal lattices comprising piles of single-layer N1, which were optimized using a molecular mechanics calculation (Supplementary Fig. 8; Supplementary Information). This series of analyses ensures the hexagonal in-plane periodicity of N1.
Optical properties and layering. Figure 6a presents the ultraviolet/visible spectra of L1 and M1 in toluene and of few-layer N1 on a quartz substrate. Ligand L1 displayed an intense absorption band at 446 nm, which is derived from the 1 p-p* transition of the dipyrrin p-system. Complexation with a zinc(II) ion is known to induce a redshift in the 1 p-p* band 53 , and in fact M1 displayed a 49-nm wavelength shift relative to L1. Few-layer N1 also showed an absorption band in the visible region, with the absorption maximum being closer to that of M1 than to that of L1, which also indicates that the complexation of L1 with zinc(II) ions was complete. Using the 1 p-p* band as a probe, single-layer N1 was accumulated stepwise on a quartz substrate. The single-layer nanosheet was fabricated on the air/water interface of a Langmuir-Blodgett trough, which was deposited repeatedly on a quartz substrate at a constant surface pressure using the Langmuir-Schäfer method. Figure 6b presents the ultraviolet/ visible spectra of the modified quartz substrate. The peak absorbance of the 1 p-p* band at 500 nm is proportional to the number of deposition processes (Fig. 6c), which indicates the quantitative, layer-by-layer accumulation of single-layer N1.
Guest inclusion. Uptake of a fluorescent dye, Rhodamine B, to N1 with large pores was demonstrated ( Supplementary Fig. 9). On immersing multi-layer N1 on an ITO or quartz substrate in a dichloromethane solution of Rhodamine B, N1 was stained the colour of the dye (Supplementary Fig. 9a). Guest-incorporated N1 displayed fluorescence from Rhodamine B ( Supplementary  Fig. 9b,c).
Photoelectric conversion. N1 was employed as the active layer of a photoanode to demonstrate its functionality. A transparent SnO 2 working electrode was decorated with 36-layer N1, and a three-electrode system was set up (Supplementary Fig. 10). Triethanolamine (TEOA) was added to an electrolyte solution as a sacrificial electron donor. An anodic current was observed only when the working electrode was irradiated with 500-nm light, corresponding to the absorption maximum of N1 (Fig. 7a). The action spectrum shown in Fig. 7b demonstrates that the photocurrent was maximized with 500-nm light, and no response was observed using light at lo420 nm or l4560 nm, a region of negligible absorption for N1. Control experiments lacking either N1 or TEOA did not show current responses at all (Fig. 7c,d). Therefore, the observed photocurrent is derived from the photocatalytic oxidation of TEOA sensitized by N1. Surprisingly, to the best of our knowledge, this report is the first on photoelectric conversion using a bis(dipyrrinato)zinc(II) complex sensitizer despite its excellent light absorption ability. We then studied the relationship between the quantum yield of the photoelectric conversion and thickness of N1 ( Fig. 7e; Supplementary Fig. 11). Single-layer N1 exhibited the highest value (0.86%), which decreases gradually with the growth of the thickness of N1, leading to negligible photoresponses at thicknesses of over B300 layers. For the photocurrent, the maximum is located at B100-150 layers ( Supplementary Fig. 12). To demonstrate the superiority of N1, we also prepared the two types of mononuclear bis(dipyrrinato)zinc(II) complex sensitizers: plain zinc(II) complex M2, which was dropcasted onto a SnO 2 electrode to form a physisorbed film (Supplementary Fig. 13a) and M3 with carboxy groups, which underwent chemisorption onto a SnO 2 surface to form a self-assembled monolayer ( Supplementary Fig. 14a). These two photoanodes resulted in much lower conversion efficiencies (0.030 and 0.069%, Supplementary Figs 13 and 14), thereby justifying the superiority of bottom-up nanosheet N1 over conventional molecular films. The nanosheet structure of N1 affords appropriate porosity and suppresses molecular aggregation; these features presumably make N1 a better sensitizer. In addition, the coexistence of insolubility (to avoid redissolution into media) and manipulability (to facilitate deposition and layering) of N1 is also advantageous for potential applications.

Discussion
A photofunctional bottom-up nanosheet containing the photoactive bis(dipyrrinato)zinc(II) complex motif was fabricated. A nanosheet with atomic thickness was prepared via an air/water interfacial synthesis, during which spontaneous complexation proceeded between a three-way dipyrrin ligand and zinc(II) ions at the interface. The nanosheet was identified using ultraviolet/ visible spectroscopy and XPS, which revealed the complete formation of the bis(dipyrrinato)zinc(II) complex motif. The single-layer nanosheet was confirmed by AFM via a scratch experiment. The domain size of the single-layer nanosheet reached 10 mm on one side, which is large for bottom-up nanosheet materials. Repeated deposition of the single-layer nanosheet on a flat substrate resulted in its quantitative layering. The nanosheet efficiently collectes visible light at around 500 nm,
Apparatus for the identification of molecular compounds. 1 H (500 or 400 MHz) and 13 C (125 or 100 MHz) nuclear magnetic resonance (NMR) spectra were recorded on a Bruker-DRX500, JEOL ECX-400 or JEOL AL-400 spectrometer. Fast atom bombardment mass spectrometry (FAB-MS) and electrospray ionization time-of-flight spectrometry were conducted using a JEOL JMS-700 MStation and Micromass LCT Premier XE mass spectrometer, respectively.
Synthesis of L1. 5 0 -(4-Formylphenyl)-[1,1 0 :3 0 ,1 00 -terphenyl]-4,4 00 -dicarbaldehyde (450 mg, 1.2 mmol) and 2-methylpyrrole (0.63 ml, 7.5 mmol) were dissolved in dichloromethane (100 ml) under a nitrogen atmosphere. One drop of trifluoroacetic acid was added, and the solution changed from light yellow to bright red and was stirred at room temperature for 3 h. When complete consumption of the aldehyde was confirmed by thin-layer chromatography, a solution of chloranil (848 mg, 3.4 mmol) in dichloromethane was added, and the resultant mixture was stirred for an additional 15 min. The reaction mixture was washed with water, dried over magnesium sulfate, filtered and evaporated. The crude product was purified by column chromatography on aluminum oxide (activity II-III) with dichloromethane as an eluent to yield a deep-yellow powder (231 mg, 25%). 1  Synthesis of multi-layer N1. A 50-ml cylindrical glass vial (3.2 cm in diameter) was used as the reaction container. L1 was dissolved in dichloromethane to a concentration of 1.0 Â 10 À 4 mol l À 1 , and 10 ml of the solution was poured into the vial. Then, pure water (10 ml) was layered gently onto the organic phase, which served as a buffer layer. After 2 h, aqueous zinc(II) acetate solution (0.1 mol l À 1 , 10 ml) was slowly added to the buffer layer. The reaction system was left undisturbed for 4 days to obtain an orange film at the liquid/liquid interface. The resulting multi-layer N1 could be deposited on various substrates.
Single-phase reactions between L1 and zinc(II) acetate. Two types of reaction were performed. Stoichiometric amounts of zinc(II) acetate dihydrate (4.9 mg, 0.022 mmol) and L1 (12.3 mg, 0.015 mmol) were dissolved in dichloromethane (5 ml) in a glass vial. The reaction mixture was allowed to stand at room temperature for 1 day. The resultant product was filtered and washed with dichloromethane, ethanol and water to yield a brown solid (Fig. 2f). In the other type of reaction, stoichiometric amounts of zinc(II) acetate dihydrate (5.1 mg, 0.023 mmol) and L1 (12.2 mg, 0.015 mmol) were dissolved in N,N-dimethylformamide (5 ml) in a glass vial. After sealing the vial tightly, the reaction mixture was heated at 105°C for 1 day in an oil bath. After being cooled down to room temperature, the resultant product was filtered and washed with dichloromethane, N,N-dimethylformamide and water to obtain a brown solid (Fig. 2g).
Layering of single-layer N1. A Langmuir-Blodgett trough (KSV 2000, KSV NIMA) was used as a reaction container. Before a zinc(II) acetate aqueous solution (0.05 mol l À 1 ) was poured into the trough as a subphase, the trough was washed with ethanol. The surface of the aqueous solution was cleaned by suctioning the surface of the subphase several times on compressing the trough barrier. Then, a dichloromethane solution of L1 (7.4 Â 10 À 5 mol l À 1 , 27.7 ml) was dropped gently onto the surface of the subphase. Under this condition, the ideal coverage of singlelayer N1 reached 90% of the trough area. After spontaneous evaporation of the organic solvent, the reaction system was left undisturbed for 4 h, such that singlelayer N1 was produced at the air/liquid interface. The trough was then compressed to yield a trough area of 69.8 cm 2 (90% of the initial trough area) at a surface pressure of 0.04 mN m À 1 . Under these conditions, single-layer N1 was transferred onto a quartz substrate via the Langmuir-Schäfer method. The trough was then shrunk by the same area as the substrate, and the next transfer was conducted. This process was repeated to layer single-layer N1.
Guest inclusion. Multi-layer N1 (thickness: 700 nm) was deposited on a quartz or ITO substrate. The modified substrate was immersed in a dichloromethane solution of Rhodamine B (5.0 Â 10 À 5 mol l À 1 ) for 15 h and was then rinsed with dichloromethane to yield Rhodamine B-encapsulated multi-layer N1.
Analyses for N1. XPS was conducted using a PHI 5,000 VersaProbe (Ulvac-Phi, Inc.). Al Ka (15 kV, 25 W) was used as the X-ray source, and the beam was focused on a 100-mm 2 area. The spectra were analysed with MultiPak Software and standardized using the C 1s peak at 284.6 eV. AFM and STM were performed using an Agilent Technologies 5,500 scanning probe microscope under an ambient condition. AFM was performed in high-amplitude mode (tapping mode) with a silicon cantilever PPP-NCL (Nano World). The probe for STM (Pt-Ir alloy, 4:1, 0.25 mm in diameter) was cut from a wire using a nipper to obtain a sharp edge. An optical microscope image was taken using a VHX-100 (Keyence Corporation). A fieldemission scanning electron microscopic image was collected using a JEOL JSM-7400FNT. Transmission electron microscopy images/SAED patterns were recorded at accelerating voltage of 75 kV using a Hitachi HF-2000 equipped with an AMT-CCD camera. The sample was prepared by depositing multi-layer N1 (thickness: 700 nm) on a carbon film supported by a copper grid (ELS-C10, stem Co., Ltd) directly from the liquid/liquid interface. To acquire electron diffractions, we focused on the edge of multi-layer N1. To reproduce the obtained SAED pattern, three-dimensional structures of multi-layer N1, which comprise piles of singlelayer N1, were considered (given as a Supplementary Information and shown in Supplementary Fig. 8). Here we treated AA-, AB-and ABC-stack models, which are often encountered in layered materials 59,60 . The three-dimensional lattice was optimized at the molecular mechanics level of theory with the UFF VALBOND 1.1 force field on an Accelrys Cerius2 ver3.1 program package. The unit cell was assumed to be trigonal such that the a, b, and g angles were constrained to be 90°, 90°, and 120°, respectively. First, we performed a calculation on the AA-stacking structure with its molecular geometries and cell lengths being fully optimized. Initial structures for the AB-and ABC-stack models were constructed from the optimized AA-stack one by giving parallel displacement to the B and/or C layers. The SAED patterns were simulated by implementing CrystalMaker 2.6.3 and SingleCrystal 2.3 (CrystalMaker Software Ltd) ( Supplementary Fig. 7b-d). Ultraviolet/visible absorption spectra were recorded on a JASCO V-570 spectrometer in transmission mode. A quartz or SnO 2 substrate modified with N1 was set vertical to the probe light. Fluorescence and excitation spectra were recorded on a HITACHI F-4500 fluorospectrometer.
Photoelectric conversion. N1 with thicknesses r155 layers was deposited on a transparent SnO 2 electrode using the repeated Langmuir-Schäfer procedure shown in Fig. 6b,c. On the other hand, N1 with thicknesses of 4155 layers was fabricated directly using the liquid/liquid interfacial synthesis (Fig. 2a) and deposited on the electrode. Before photoelectric conversion, each N1 physisorbed on a transparent SnO 2 electrode was always subjected to ultraviolet/visible spectroscopy, acquiring spectra from four different positions. This series of measurements allowed us to ensure the uniformity of N1 and to quantify the thickness of N1 using the average absorbance at 500 nm and that of single-layer N1 (0.00101 from the slope of Fig. 6c). The average absorbance at 500 nm was also used in calculating the quantum yield of photoelectric conversion (vide infra). The modified SnO 2 electrode was used as a working electrode (photoanode). Silver and platinum wires were employed as reference and counter electrodes, respectively. The resulting three-electrode system was immersed in a homemade photoelectrochemical cell ( Supplementary Fig. 10), which was filled with an acetonitrile solution of tetra-nbutylammonium perchlorate (0.1 mol l À 1 ) containing TEOA (0.05 mol l À 1 ) as a sacrificial donor reagent. The cell was sealed and deoxygenized by argon bubbling for 30 min. Monochromatic light for the action spectrum shown in Fig. 7b (400-600 nm in every 20 nm) was extracted from a xenon lamp (MAX-302, Asahi Spectra Co., Ltd), the photon flux of which was monochromated by a monochromator (CT-10, JASCO Corporation). For the other experiments, 500-nm light was used exclusively, which was provided by the xenon lamp equipped with a band-pass filter. The active area of the electrode was 0.264 cm 2 , which was determined by a fluorocarbon rubber o-ring ( Supplementary Fig. 10). The electrode potential and photocurrent acquisition of the photoelectric conversion system were controlled using an electrochemical analyser (ALS 750A, BAS Inc.).
The photoanode was fixed at 0.0 V versus the silver reference electrode for the data shown in Fig. 7a-d and 0.15 V for the other data, including those for M2 and M3. At both potentials, no anodic current induced by the direct oxidation of TEOA in the dark was observed. The quantum yield of photoelectric conversion, f, was calculated using equation (1): where n e is the mole of electrons that flows in the circuit per unit time (in mol s À 1 ) and n p is the mole of photons absorbed by the sensitizer per unit time (in mol s À 1 ). n e and n p were calculated using equations (2) and (3): where i is the current flow (in A), F is the Faraday constant (9.65 Â 10 4 C mol À 1 ), W is the photon flux of incident light (in J s À 1 ), l is the wavelength of the irradiated light (5.00 Â 10 À 7 m), A is the absorbance at the irradiated wavelength, N A is the Avogadro constant (6.02 Â 10 23 mol À 1 ), h is the Planck constant (6.63 Â 10 À 34 Js) and c is the velocity of light (3.00 Â 10 8 m s À 1 ). A representative data set for the determination of f for N1 is shown in Supplementary Fig. 11. i was calculated using equation (4): where i L is the average light current for the first cycle (10 s) and i D is the average dark current just before the illumination of light. A photon counter (8230E and 82311B, ADC Corporation) was employed for the quantification of W. For every sample, W was measured independently. A typical value for W was B0.20 mW. When i was below the measurable level (o1.0 nA), W was increased to 1.0-2.0 mW such that the photocurrent signal was amplified. For referential mononuclear complex M2, film formation was conducted as follows: two droplets from a disposable Pasteur pipette of an acetonitrile solution of M2 (1.86 Â 10 À 5 mol l À 1 ) were dripped onto a SnO 2 substrate (B0.8 cm 2 ). The substrate was then dried under vacuum for 10 min to allow the solvent to evaporate. The modified SnO 2 electrode was subjected to photoelectric conversion, following the procedure for N1 except that aqueous 0.1 mol l À 1 sodium sulfate was used as an electrolyte solution; this change prevented the M2 film from redissolution. Referential mononuclear complex M3 was immobilized on a SnO 2 electrode through the carboxy group using the self-assembled monolayer procedure 61 . A SnO 2 substrate was immersed in a dimethylsulfoxide solution of M3 (1.96 Â 10 À 3 mol l À 1 ) for 1 day at room temperature, and the decorated substrate was rinsed with dimethylsulfoxide and dried by an argon blow. The modified SnO 2 electrode was subjected to photoelectric conversion using the same procedure as N1.