Potassium hydroxide, 3,5-dichlorosalicylaldehyde were purchased from Wako Fujifilm. Manganese(II) chloride tetrahydrate, zinc(II) acetate dihydrate, aniline, sodium nitrite and sodium hydroxide were purchased from Kanto Chemical Co., Inc. Salicylaldehyde, p-anisidine and 4-aminobenzoic acid were obtained from Tokyo Chemical Industry. 4,4’-((1S,2S)-1,2-diammonioethane-1,2-diyl)dibenzoate and (2S, 3S)-3-aminoaspartic acid, (E)-2-hydroxy-5-((4-methoxyphenyl)diazenyl)benzaldehyde and (E)-2-hydroxy-5-((4-carboxyphenyl)diazenyl)benzaldehyde were prepared according to the literature procedures [1] [2] [3] [4] [5]. AgNPs (30 - 80 nm) were synthesized using a modified polyol method [6] [7].

Preparation schemes of the four complexes were shown in Figure 1.

Elemental analyses (C, H and N) were performed using a Perkin-Elmer 2400 II CHNS/O analyzer at the Tokyo University of Science and using an Elementer vario MICRO cube at TORAY Research Center. Infrared (IR) spectra were recorded as KBr pellets on a JASCO FT-IR 4200 plus spectrophotometer in the range 4000 - 400 cm⁻¹ at 298 K. Electronic (UV-vis) spectra were obtained on a JASCO V-650 UV-vis-NIR spectrophotometer in the range 800 - 200 nm at 298 K. Circular dichroism (CD) spectra were obtained on a JASCO J-820 spectropolarimeter in the range 800 - 250 nm at 298 K. ¹H-NMR spectra were recorded on a JEOL JMN-300 spectrometer (300 MHz). Electrochemical (cyclic voltammetry, CV) measurement were carried out on a BAS SEC2000-UV/VIS and ALS2323 system with Ag/AgCl electrodes range of −2.0 - 1.5 V vs. Ag/Ag⁺. Raman spectroscopy was recorded using JASCO NRS-3200. I - V measurement was obtained Yamashita Denso YSS-80 at 298 K. XPS measurement was carried out using Mg Kα source (10 kV, 25 mA) on a SHIMADZU ESCA3400. Powder X-ray diffraction patterns were collected at 298 K with Pohang Light Source II 2D Supramolecular Crystallography Beamline (PLSII-2D-SMC). Nanosecond time-resolved TA measurements were carried out with a Ti: sapphire laser (LibraHE, Coherent). The second harmonic (400 nm) of the laser was used for excitation. The intensity of the laser pulse (70 mW) was measured with an air-cooled thermopile sensor (PM2, Coherent). A xenon lamp (C4251, Hamamatsu) was used as the probe light source. The light transmitted through the sample was passed through monochromator (CM110, Spectral Products) and detected with a Si photodiode (DET10A/M, Thorlabs). The photocurrent from the detector was amplified with an amplifier (DHPVA-201, Femto), and signals were processed with a digital oscilloscope (WaveRunner8254, Teledyne LeCroy) and analyzed with a computer. The DC offset of the photocurrent from the detector was subtracted by using the AC-coupled mode of the amplifier. TCSPC measurements were carried out using FluoroCube (Horiba). For the excitation light source, a pulsed laser diode (Horiba, NanoLED; wavelength of 410 nm, pulse width of 200 ps, and repetition rate of 1 MHz) was used. For streak camera measurements, the second harmonic (400 nm) light of the Ti: sapphire laser (Solstice, Spectra Physics; pulse width of 100 fs, and repetition rate of 1 kHz, 3.3 mW) was irradiated to the samples and photoluminescence was corrected using C4334-01 (Hamamatsu Photonics).

Calculations of all complexes were performed using the Gaussian 09W software Revision D.02 (Gaussian, Inc.) [8]. Starting models (four-coordinated square planar) of structural optimization is based on preliminary results of Rietveld analysis and TG-DTA. The gas phase geometry optimizations were carried out by using TD-DFT with B3LYP functional. The vertical excitation energy was calculated with the Lanl2dz for each complexes, with the 6 - 31 + G (d) basis set for H, C, N and O method based on the singlet ground state geometry.

For example, UV-vis spectra for MZn and AgNP appeared at 391, 450 nm and 270, 450 nm, respectively, whose the strongest peaks around 380 nm could be assigned to be p-p* transition of trans-form as shown in Figures 2(a)-2(d). Substitution to Mn(II) ion and carboxyl groups resulted in short wavenumber shift of UV-vis spectra (MZn 391 > MMn 387 > CZn 371 > CMn 364 nm). The peaks shifted to long wavelength region on formation MZn + AgNP composite (Figure 3). CD spectra for MZn appeared at 300 and 410 nm, though induced CD could

not be observed clearly for MZn + AgNP composite for all complexes due to unsuitable arrangement of dipole moments of chiral complex and AgNP [4] [5]. Low-energy shift of XPS also suggested adsorption of metal complexes onto AgNPs: 3d_5/2 and 3d_3/2 are 368.328 and 374.312 eV for AgNPs, 368.159 and 374.145 eV for MMn + AgNP, 368.157, 374.138 eV for MZn + AgNP, 368.244, 374.39 eV for CMn + AgNP, and 368.206, 374.182 eV for CZn + AgNP, respectively. On adsorbing on AgNP, as shown in Figure 4, fluorescence spectra of carboxyl complexes (CMn and CZn) exhibited more clear quenching than that of methoxy complexes (MMn and MZn).

Time-correlated single photon counting (TCSPC) were carried out to investigate the photoexcited state and the reaction intermediate state when each complex and its composite system (8 samples in total) were illuminated [9]. The fluorescence decay plots (Figure 5) were analyzed to obtain the fitting parameters (Table 1) to the exponential functions (1) and (2) using an ORIGIN software. It was found that this task could be achieved by compositing silver nanoparticles. The relaxation constants are faster in the order τ₁ > τ₂ > τ₃. A₁, A₂, and A₃ are the component ratios of the respective relaxation components. The average fluorescence lifetime was calculated by the following equation.

I ( t ) = A 1 exp ( − t / τ 1 ) + A 2 exp ( − t / τ 2 ) + A 3 exp ( − t / τ 3 ) (1)

t aver = ( A 1 τ 1 + A 2 τ 2 + A 3 τ 3 ) / ( A 1 + A 2 + A 3 ) (2)

Only for MZn + AgNP composite, τ_avg was longer than sole complex (MZn), which indicated that the average fluorescence lifetime of other complexes is

shortened by the addition of AgNPs ascribed to effective energy transfer through carboxyl groups. Due to Mn(II) ion unsuitable for excitation, MMn + AgNP composite showed the shortest τ_avg of 1.55 ns. In this way, the results observed were caused by the energy coupling between the plasmon resonance and the emission of the fluorescent molecules, which shortens the electronic recombination time from the excited state to the ground state and accelerates the radiation relaxation. A previous study [9] reported that the presence of metal nanoparticles in the vicinity of fluorescent molecules shortens the fluorescence lifetime of the fluorescent molecules. This suggests that AgNP could not exist in the vicinity of the composite due to aggregation of AgNPs in MZn + AgNP. While CZn has many carboxyl groups to adsorb on AgNPs. Therefore, when AgNP were combined with the complex, the carboxyl groups of the complex were able to bind to AgNPs and maintain a close proximity.

Discussion in 3.2 suggests that MZn + AgNP may be suitable for sensitizer of DSSC among the four complexes, and cells [1] assembled as Figure 6, in which adsorption of MZn to AgNP was monitored with Raman peaks at 1417 and 1042 cm⁻¹. The DFT calculation shows that the positional relationship of the orbitals before and after excitation is in an advantageous position with respect to the adsorption substituents. Redox potentials from CV suggested that the four complex may be appropriate for sensitizer electrochemically (LUMOs are higher than −0.540 V for TiO₂, HOMOs are lower than 0.460 V for I − / I 3 − ). Photoelectric conversion efficiency could be evaluated to be 0.433%, 0.0377%, and 0.342% for AgNP, MZn, and MZn + AgNP, respectively based on I - V curves. In contrast, employing MZn shorten t_ave, the corresponding values are 0.433%, 0.0837%, and 0.655% for AgNP, CZn, and CZn + AgNP, respectively. The results indicated that other factors associated with cells may make it inefficient besides excited state of MZn + AgNP.