Bioinspired study of energy and electron transfer in photovoltaic system

ABSTRACT This study focuses on understanding the fundamentals of energy transfer and electron transport in photovoltaic devices with uniquely designed nanostructures by analysing energy transfer in purple photosynthetic bacteria using dye-sensitised solar cell systems. Förster resonance energy transfer between the xanthene dye (donor of energy) and a new polymethine dye (acceptor of energy) was studied in dye-sensitised solar cells, which leads to a doubling of energy conversion efficiency in comparison to the cell with only the polymethine dye. The electron transport in the two different nanostructures of zinc oxide (nanorods and nanosheets) was investigated by spectroscopic methods (UV-vis spectrometer, time-resolved photoluminescence spectroscopy) and electrochemical potentiostat methods. The nanosheet structure of zinc oxide showed high short circuit current and long diffusion length. This fundamental study will lead to efficient artificial photosystem designs.


Introduction
Commercialisation of solar cells requires efficiency, stability, cost effectiveness and material availability. At present, a solar cell which could fulfil all these criteria still requires improvement of the existing technology and the development of new pathways by alternative mechanisms for solar energy conversion [1][2][3][4].
One promising area is the dye-sensitised solar cell (DSSC), which utilises a Ru-dye similar to plant chlorophyll, and applies a self-assembly principle, which reduces process costs [5][6][7][8]. Although DSSC is a special type of the solar cells and demonstrates power conversion efficiency up to 13%, the availability of Ru-dyes and the stability of liquid-based electrolytes are still the main concerns when developing this type of solar cell [9][10][11][12][13][14].
Natural photosynthetic systems are the result of perfectly assembled and efficient systems of natural dyes coupled with protein molecules that harvest sunlight and use its energy to drive metabolic reactions [15]. The solar energy conversion process includes critical steps of solar energy harnessing, charge separations and catalytic reaction. One of the strongest examples of this process is Purple photosynthetic bacteria [16] (Figure 1 (B)), which contain two types of light-harvesting complexeslight-harvesting complex I (LH-I) and light-harvesting complex II (LH-II)and surround the reaction centre (RC) in the bacteria's photosynthetic membranes. This assembly system retains the distinctive structure of a photosynthetic complex with a new type of light-harvesting protein and transport channels that allow efficient energy and electron transfer. The excitation transfer from LH-II -! LH-I -! RC occurs within 100 picoseconds at about 95% efficiency [17,18]. B800 BChls, bacteriochlorophylls (BChls) which act as antennas with the carotenoids, absorb the sunlight at 800 nm and funnel its energy to the ring B850 BChls (absorbing light at 850 nm) through the F€ orster resonance energy transfer (FRET) mechanism [19,20]. By understanding the structure of light-harvesting complexes, we can advance our understanding of the primary excitation transfer process just as the structure of the RC revolutionised the study of the primary electron transfer processes. The arrangement of the main pigment-protein complexes, among LH-I and the photosynthetic RC explained by recent electron microscopy data, suggest that the hierarchy arrangement of the electron transfer system in RC is mostly shown as cylinders (or rods) and coplanar structures (sheets) [16]. Moreover, due to the elegant structure of the biological organism, genetically engineered viruses [21][22][23] were applied to the solar energy conversion system in order to study solar cell and water splitting. Nanotechnology provides us with ways of building nanostructures as they exist in a natural photosynthetic system [17,24]. This study focuses on energy transfer and electron transport in a photovoltaic device coupled with uniquely designed nanostructures inspired by photosynthetic bacteria. A DSSC system provides a base for investigating fundamental problems including energy transfer and electron transport. In the study of electron transport properties, uniquely assembled nanostructures of ZnO were fabricated using electrochemical and chemical vapour deposition (CVD) approaches. The assembled nanostructures, including one-dimensional nanorods, and nanosheets, were investigated for electron transfer using characterisation methods (SEM, TEM and XRD), spectroscopic methods such as UV-vis spectrometer and time-resolved photoluminescence spectroscopy, and electrochemical potentiostat methods. In the study of energy transfer, FRET between the xanthene dye (donor of energy) and a new polymethine dye (acceptor of energy) is studied on the surface of TiO 2 films by time-resolved photoluminescence spectroscopy, and this fundamental study will lead us to better designed artificial photosystems.

Preparation of ZnO nanorods
Preparation of ZnO seed layers. To prepare ZnO seed layers, first 0.5 M of zinc acetate dehydrate [(Zn(CH 3 COO) 2 ¢2H 2 O, Sigma Aldrich)] in the mixed solution of monoethanolamine (C 2 H 7 NO, Sigma-Aldrich) and isopropyl alcohol was spin-coated onto ITO substrates at 3000 rpm five times [25]. Then, the ITO substrates were thermally heated at 400 C in air for 30 minutes to transform zinc acetate to ZnO.
Hydrothermal deposition. To fabricate ZnO nanorods, the hydrothermal approach was applied to grow ZnO nanorods vertically via heating the ZnO-seeded ITO substrates in solutions of 25 mM Zn(NO 3 ) 2 (Sigma-Aldrich) and 25 mM hexamethylenetetramine (C 6 H 12 N 4 , Sigma-Aldrich) at 90 C for 10 hours. Finally, the resulting substrates were cleaned with water/ethanol and annealed at 400 C for 30 minutes to eliminate any residual organics.

Preparation of ZnO nanosheets
In brief, in the synthesis of the ZnO nanosheets, an a three-electrode electrochemical configuration set up (consisting of an ITO substrate (working electrode), a graphite electrode (counter electrode) and a saturated Ag/AgCl (reference electrode), respectively) was applied to deposit the arrays of hexagonal Zn 5 (OH) 8 Cl 2 nanosheets first at ¡1.1 V and 50 C conditions for 30 minutes in the aqueous electrolyte of 0.05M Zn(NO 3 ) 2 and 0.1M KCl . Subsequently, a pyrolytic transformation of Zn 5 (OH) 8 Cl 2 into ZnO nanosheets was performed at 400 C in air.

Assembly of DSSC
Solar cells were prepared and assembled according to a previously described procedure [26]. Briefly, the prefabricated ZnO-based photoanodes, each with an active surface area of 0.2 cm 2 , were immersed in a dry ethanol solution containing 0.5 mM of N719 (Solaronix) at 50 o E for 1 h to load the dye at first. A Pt coated conductive glass substrate was applied as the counter electrode. The Pt layer was prepared on the ITO glass through electrodeposition approach as described elsewhere [27,28]. The DSSC was assembled by placing the Pt-coated counter electrode over the dyed ZnO-based photoanode separated by a 25 mm thick thermoplastic sealing spacer (Meltonix 1170-25, Solaronix). The cell's internal space was filled with a highly stable liquid electrolyte containing Iodolyte Z-150 (Solaronix; Redox couple: iodide/tri-iodide; redox concentration: 150 mM; additives: ionic liquid, alkylbenzimidazole, thiocyanate; solvent: 3-methoxypropionitrile).

Results and discussion
The two types of artificial mimicked photosystem were built and explored to understand FRET-energy transfer and electron transport properties based on the model of DSSC, respectively.
Designation of FRET-energy system and its study: To mimic the FRET system in purple bacteria, a donor/acceptor system consisting of a xanthene dye (rhodamine as a donor) and a polymethine dye (squarylium dye (SQ) as acceptor) was selected for study. This system is located on the surface of TiO 2 films in the DSSC because of their higher molar extinction coefficients (100,000-300,000 M ¡1 cm ¡1 ) than ruthenium-based complexes (low molar extinction coefficients (5000-20,000 M ¡1 cm ¡1 )) [29]. In order to prepare samples with proper concentrations of rhodamine and squarylium dye, the F€ orster critical distance and critical concentration (Supporting Table 1) were calculated to be 2.9 nm and 10 ¡2 M based on the spectral measurement data [30,31]. The critical distance between these two dyes was determined to be 2.9 nm, which is in the range of the values of purple bacteria [16]. As shown in Figure 1(B), in the FRET energy system of a purple bacteria the absorption energy of donor dye (absorption of Bacteriochlorophylls (BChls), 'B800 BChls', at 800 nm) must be higher than that of the absorption energy of acceptor (absorption of B850 BChls at 850 nm) [16], where light-harvesting pigments transfer their energy to the RC. Figure 2(A) shows that the fluorescence spectrum of the donor and the absorption spectrum of the acceptor overlaps, a prerequisite for effective electronic excitation energy transfer. Absorbance of rhodamine 6G and squarylium dye (SQ) are 525 and 625 nm, respectively. Therefore, Rhodamine should efficiently transfer the energy by the F€ orster process to the anchored SQ. Time-resolved fluorescence spectroscopy was applied to measure the fluorescence decay kinetics in the absorption band of the donor using radiation from a laser with a wavelength (λ gen ) of 488 nm. Peaks of donor and acceptor fluorescence were found to be 555 and 660 nm, respectively. The concentrations of the donor and acceptor in these measurements were chosen under the critical concentration, the requirement for transferring the greatest amount of the excitation energy to acceptor molecules. . Furthermore, FRET-energy transfer between rhodamine 6G and squarylium dye (SQ) ( Table 1) was studied in TiO 2 films in comparison with a sole donor and a sole acceptor at various concentrations of components [32]. The efficiency of energy transfer, E EG , was calculated according to the reference [33].
As shown in Table 1, the excited-state lifetime of the donor in the presence of acceptor molecules decreases with increasing concentration of the acceptor molecules. The lifetime of acceptor molecules has a value close to the lifetime of the donor, which confirms the implementation of the process of energy transfer between the selected dyes and reaches the highest energy transfer.
Likewise, the fluorescence intensity of the acceptor (in the presence of, and without donor molecules, under illumination by a xenon lamp) was also measured to evaluate their effects on the performance of the cellco-sensitisation of TiO 2 semiconductor film by the donor and acceptor molecules. Figure 2(B) indicated the luminescence intensity of  the acceptor is 2.5 times high as that of sole acceptor. Addition of donor molecules or acceptor molecules with respect to one another leads to the quenching of fluorescence of rhodamine or polymethine dyes in respective measurements: this explains efficient FRET energy transfer between donor and acceptor molecules in the presence of titania assembled films [34]. Figure 2(C) shows the strong photoluminescence at the low concentration range for both donor and acceptor. The quenching of the fluorescence intensity of dyes at the growth of donor concentration is the result of molecular aggregation of rhodamine dye. The strongest quenching for both donor and acceptor were equal to about 25% and registered at donor concentration of 10 ¡3 mol/L. After identifying the correct ratio between donor and acceptor molecules in the porous titania films, the conversion of light energy into electrical energy in the DSSC solar cells was investigated further. The energy conversion efficiency of the semiconductor films sensitised by the donor-acceptor compound was evaluated in parallel study with either the sole squarylium dye (SQ) or rhodamine 6G measured by solar cell tester and potentiostat. Figure 2(D) shows a current-voltage curve for solar cells sensitised by SQ dye molecules, (curve 1), sole donor molecules (curve 2) and donor-acceptor compound (curve 3). I sc of solar cells with only SQ, and sole rhodamine was equal to 0.62 and 0.12 mA/cm 2 . In the case of co-sensitisation of the solar cell with donor and acceptor at the above ratio, the I sc value was significantly increased to 1.7 mA/cm 2 . The power conversion efficiency of TiO 2 solar cells consisting of donor-acceptor is more than 2.5 times over a pristine cell. The optimum concentration of donor molecules for the co-sensitisation of solar cell is 10 ¡5 mol/L.
The curves of spectral photosensitivity of solar cell indicated (Supporting Figure 2) that the photosensitivity for the donor-acceptor system had a higher photosensitivity than that of the sole acceptor which supported the I-V curve results.
Study of nature-inspired electron transport system: In designing electron transport pathways for mimicking photosynthetic purple bacteria (Figure 1(B,C)), ZnO nanostructures with rods and sheets were selected, since ZnO with different nanostructures is easily fabricated, has a better electron mobility, and similar band gap structure with titania [35]. These nanostructures were main aggregate forms in the electron transport pathways in the RC and LH-I of purple bacteria. ZnO's nanorods and nanosheets were synthesised by both hydrothermal and electrochemical deposition methods, respectively, and are confirmed by scanning electron microscopy (SEM) (Figure 3(a,b)). Figure 3(c,d) indicated that nanorods and nanosheets possess wurtzite crystal structure. A unique crystal face (0001) in the nanosheet structure of zinc oxide is considered to be a polar plane. The influence of the polar plane, defect density and morphology of these nanostructures on photovoltaic and electron transport properties was evaluated by the following approaches: photoluminescence study, electrochemical impedance spectroscopy and analysis of dye-loading amount in the photoelectrodes in parallel with solar cell performance measurement. Photovoltaic performance of the DSSC cells constructed from nanorod and nanosheet arrays was listed in presence of the highly stable electrolytes, including the high boiling organic solvent and a mixture of ionic liquids with relatively low viscosity, using Ru-dye (N719) in Table 2 and Figure 4(A). Figure 4(A) indicated that the short-circuit current density (J sc ) of the nanosheet array (NSh cell) cell is almost five to six times more than that of the nanorod array (NR cell)-based cell, whereas, the open circuit voltage (V OC ) of the NSh cell is the opposite. In addition, the fill factor (FF) of the NSh cell is significantly lower than that of the NR cell. Since the same dye (N719) was utilised in the solar cell test, the big difference in the J sc of the two cells is due to different surface areas and different light scattering properties. One possible explanation is that the nanosheet structure has a feature size close to the light wavelength and may generate stronger light scattering, thus increasing the current density, a point that needs further confirmation. Another possible explanation is that the photoexcited electrons inject more efficiently from the dye molecules into ZnO through the polar face of ZnO since ZnO brings a large surface to the nanosheet array.
To prove the idea, the amount of the dye loaded in both structures was evaluated by UV-vis absorption measurement after desorbing the dye molecules from ZnO film in the base solution. The dye absorption spectrum of nanosheet array is significantly larger than that of nanorod array (Supporting Figure 3), which indirectly suggests that surface area of a nanosheet array is larger than that of a nanorod array.
The defect density of the zinc oxide is also one of the factors that could significantly affect the cell photovoltage. The photovoltage is determined by the increase of the quasi- Figure 3. SEM images of arrays of (a) ZnO nanosheet (on the left) and (b) nanorod (on the right). X-ray diffraction spectra of (c) ZnO nanorods and (d) nanosheets. Fermi level of the semiconductor, E Fn , under a constant illumination with respect to the dark value, E F0 , which equals the electrolyte redox energy [36]. Another way of understanding this is that defects can trap the excited electrons and diminish the concentration of the injected electron in the conduction band of the semiconductor. As shown in Equation (1), the quasi-Fermi level of the semiconductor is proportional to the electron concentration in the conduction band, n [36]: Therefore, the low value of V OC of NSh in comparison with V OC of the NR cell may be related to the effect of defects, which was confirmed by the photoluminescence (PL) properties of nanosheets and nanorods (Figure 4 (B)). The PL spectra of nanosheets (blue line) and nanorods (green line) (Figure 4 (B)) showed a sharp emission peak at 380 nm and a broad band with two peaks centred at 546 and 576 nm. The UV emission at 380 nm is ascribed to recombination of the free excitations or near band edge emission of the wide band-gap ZnO [36,37] and the visible light emission is related to defects, due to oxygen vacancies or zinc interstitials [38,39]. Furthermore, the black line exhibits a very strong UV emission and a negligible visible emission, while the red line shows a relatively strong visible emission. Their ratios of UV/visible emissions are 7.7 and 0.17 under our current measurement conditions, respectively. This result indicates that the nanosheets have much more defect density, or it is due to the different disorder or defects located at different crystal planes [40,41]. Figure 4(C ) shows the kinetics of the edge luminescence band (ELB) decay for the nanorods and nanosheets. The ELB of ZnO decays in the subnanosecond range [42]. As can be seen from Figure 4(C), the decay kinetics of ELB for both the nanorods and nanosheets is a nonexponential function. The lifetimes calculated from these curves are »1.1 and 2.33 ns for the nanorods and 0.71 and 1.7 ns for the nanosheets. As can be seen, the ELB of nanosheets decays faster than that of the nanorods.
In the DSSCs based on TiO 2 films, the majority of the photoinjected electrons are located in trap states in the nanostructured TiO 2 film [43,44], leading to a trapped electron density that greatly exceeds the conduction electron density. These electrons must be released thermally to the conduction band before they can reach the interface and transfer to I 3¡ ions. These trapping and detrapping transport models of photoinjected electrons have been confirmed by the data of electrical impedance spectroscopy (EIS), intensitymodulated photovoltage spectroscopy (IMVS) and photovoltage decay measurements [43][44][45].
Electron transport properties in the DSSC cells were further investigated by considering the parameters: two different morphologies and a defect density of the ZnO nanostructures by electrochemical impedance spectroscopy (EIS). The results were further analysed by the general transmission line model of ZnO nanostructured DSSCs based on the diffusion-recombination model proposed by Bisquert [46]. The Nyquist plots for impedance data of the NSh and NR cells and the fitting results are shown in Figure 4(D) and the related electron transport parameters were obtained from the Nyquist plots according to the reference [47]. The results show that electron densities at the steady state (n s ) in the conduction band of NSh-ZnO cell are approximately three to four times larger than these of NR-ZnO cell and indicated that larger number of electrons are injected in NSh, which are consistent with high J SC . Furthermore, the effective electron lifetime (t eff ) of NSh ZnO cell is higher than that of NR ZnO cell. As discussed in references [48,49], the electron diffusion length (L n = (D Ã t eff ) 1 = 2 is proportional to (t eff ) 1 = 2 . The collection diffusion length, L c , determines the short circuit current of a solar cell which is illuminated with light of near-bandgap wavelength [50,51]. Therefore, the high electron density and longer electron life time in NSh-ZnO cell are to some degree explain the high short circuit current of the cell.

Conclusions
In this paper, we investigated the FRET-energy transfer between molecules of rhodamine and polymethine dye in TiO 2 films and electron transport in different nanostructures of zinc oxides. The fluorescence decay kinetics of the energy donor revealed a reduction of the excited state lifetime of the donor in the presence of acceptor molecules and confirmed that the efficiency of energy transfer was 22% at an acceptor molecule concentration of 10 ¡4 mol/ L. Under white light illumination, the intensity of sensitised fluorescence of polymethine was almost doubled and its photovoltaic efficiency of solar cells was increased 2.5 times in the presence of rhodamine dye molecules. The main reason for the increase of the cell's efficiency is connected to the energy transfer process. These results are very important to understanding and completing the assembly of high efficient solar cells sensitised with organic dyes.
In addition, we studied the effect of ZnO nanorod and nanosheet on the solar cell efficiency of DSSEs. The nanosheets have a larger defect structure than the nanorods, which was confirmed by a relatively strong visible/defect emission. The short current of DSSC made of ZnO nanosheet was five to six times larger than that of the DSSC with nanorods. This observation is supported by two explanations: first, that nanosheet structures have more dye loading in contrast to nanorod structures, which was confirmed by the UV-vis absorption spectra of desorbed dye molecules; the second is that the nanosheet structure of zinc oxide possesses a relatively high electron density and an effective long electron life time in the NSh-ZnO cell in contrast to the NR-ZnO cell. However, we also consider the possibility that the electron photoinjection from the dye molecules into ZnO more effectively occurs through the ZnO polar plane (0001), making the dominant contribution to the surface area of the nanosheets, unlike the nanorods. The low photovoltage of DSSC with the nanosheet array was due to the low electron trap density, which can reduce the electron quasi-Fermi level. .