Can aliphatic anchoring groups be utilised with dyes for p-type dye sensitized solar cells?

A series of novel laterally anchoring tetrahydroquinoline derivatives have been synthesized and investigated for their use in NiO-based p-type dye-sensitized solar cells. The kinetics of charge injection and recombination at the NiO-dye interface for these dyes have been thoroughly investigated using picosecond transient absorption and time-resolved infrared measurements. It was revealed that despite the anchoring unit being electronically decoupled from the dye structure, charge injection occurred on a sub picosecond timescale. However, rapid recombination was also observed due to the close proximity of the electron acceptor on the dyes to the NiO surface, ultimately limiting the performance of the p-DSCs.

from the dye structure, charge injection occurred on a sub picosecond timescale. However, rapid recombination was also observed due to the close proximity of the electron acceptor on the dyes to the NiO surface, ultimately limiting the performance of the p-DSCs.

Introduction
The dye-sensitized solar cell (DSC) is a low cost alternative to crystalline silicon photovoltaics that converts sunlight into electricity using a dye adsorbed on a transparent, nanostructured semiconductor electrode, surrounded by a redox electrolyte. Almost all the current research in photocatalysis and dye-sensitized solar cells is focused on n-type systems, typically based on TiO2. 1 Unlike the standard DSC, which has a passive counter electrode, tandem devices simultaneously use two photoelectrodes to harvest a greater proportion of the spectrum more efficiently. 2,3 Incorporation of a photocathode in tandem with a TiO2-based n-type photoanode in a single device should give rise to a substantial increase in voltage and efficiency. By choosing sensitizers which absorb high energy photons on one electrode and low energy photons on the other, more of the solar spectrum can be utilised. 4 Tandem DSCs have not yet beaten the best n-type DSCs because the poor performance of dye-sensitized photocathodes limits the overall efficiency. Therefore it is necessary to improve the light conversion efficiency of dye-sensitized photocathodes in the long wavelength region of the solar spectrum so that they can match the best dye-sensitized photoanodes which typically collect and convert light between 400-700 nm.
Several groups have developed panchromatic organic sensitizers for TiO2. [5][6][7] These generate high photocurrents (ca. 14 mA cm -2 for dyes TH304 and T4BTD-A in references 5 and 7), however, the efficiencies are rarely higher than those obtained using dyes with a narrower absorption band. This is because shifting the absorption spectrum to longer wavelengths requires the frontier orbital separation of the dye to be reduced, bringing the HOMO and LUMO closer in energy. Lowering the energy of the LUMO lessens the driving force for electron injection into the TiO2 conduction band. This can be offset by adding lithium ions to reduce the energy of the conduction band edge, however this also leads to a decrease in the photovoltage. The same effect is observed when the energy of the HOMO is increased, as this reduces the driving force for electron transfer to the oxidised dye from the redox mediator, thus increasing the rate of the competing recombination reactions between the electrons in the TiO2 and holes in the dye. By using a tandem cell configuration, where the differences in energy between the redox couple and the valence and conduction band edges of the respective photocathode and anode are smaller at one electrode than the other, the low energy photons can be used to generate current without sacrificing the overall device voltage. To develop highly efficient tandem solar cells, the efficiency of the photocathode must be increased to match the conventional dye-sensitized TiO2 solar cells. 8 Recently, significant improvements to NiO-based p-DSCs have been made by designing dyes which promote charge separation and limit the recombination between the sensitizer and/or electrolyte and the semiconductor. [9][10][11][12][13][14] Currents obtained in these devices have reached over 8 mA cm -2 at 1 sun illumination. 4,15 Earlier, we reported a high performance dye based on a triphenyl amine electron donor and a cationic indolium acceptor unit that absorbed in this region. 4 The tandem DSCs incorporating this dye generated a photocurrent that was larger than any previously reported tandem DSC. However, it was notably lower than that of the single junction p-DSC. Inspection of the IPCE spectra of the individual n-DSC and p-DSCs showed that there was still significant overlap between the two dyes used at opposite electrodes in the tandem DSC.
Here we describe a series of tetrahydroquinoline derivatives ( Figure 1) which absorb up to 850-900 nm when adsorbed to mesoporous NiO via a lateral propionic acid chain. In the majority of donor-π-acceptor organic sensitizers reported, the anchoring group, typically carboxylic acid, is positioned close to an electron donating group. The electron donor is typically a modified triaryl amine, which fixes the excited state reduction potential (ED*/D-) to 1-1.4 V vs.
NHE. 16,17 By separating the anchoring group from the electron donor, a wider choice of derivatives is possible and fine-tuning of the energy levels can be achieved. This could minimise energy wastage in the system and allow the shifting of the absorption towards the infrared. These new dyes are likely to inject charge (h + ) using a different mechanism to dyes that have been extensively studied in the past, dependent on their geometry on the semiconductor surface. shed new light on the importance of dye-dye, dye-semiconductor and dye-electrolyte interactions to photoelectrochemical systems, such that modifications to the dye design can be made to improve the device efficiency.

Experimental Section
Solvents were dried by standard procedures. All other chemicals were purchased from commercial sources and used without further purification.

Analytical Measurements.
1 H NMR spectra were recorded with a Varian INOVA 400 NMR instrument. MS data were obtained with GCT CA156 (UK) high-resolution mass spectrometer (HRMS) or HP1100 LC/MSD (USA) mass spectrometer. UV-visible absorption spectra of the dyes in solutions were recorded in a quartz cell with 1 cm path length on a HP 8453 spectrophotometer.
Electrochemical redox potentials were obtained by cyclic voltammetry and differential pulse voltammetry using a three-electrode cell and an IviumStat potentiostat. A solution of Bu4NPF6 (0.1 M) in CH3CN was used as supporting electrolyte. The working electrode was a glassy carbon electrode, the auxiliary electrode was a Pt mesh, and the reference electrode was Ag + /Ag in a solution of the supporting electrolyte. Ferrocene was used as an external calibrant. Samples for FTIR spectroscopy were prepared as either a KBr pressed disk (dye only) or by adsorbing the dye on a mesoporous NiO film deposited on a CaF2 window (Crystran). The NiO films were prepared by spraying a saturated solution of NiCl2 in acetylacetone onto the surface of the CaF2 window which was pre-heated to 450°C on a hotplate; this was then allowed to cool slowly to room temperature to give a compact film of NiO. The mesoporous layer was then deposited on top of the compact layer using an F108-templated precursor solution containing NiCl2 (1 g), Pluronic® co-polymer F108 (1 g), Milli-Q water (3 g) and ethanol (6 g) and the excess was removed with a glass rod. The film was sintered at 450 °C for 30 minutes and an additional layer of precursor solution was applied and sintered to increase the film thickness.
Infrared absorption spectra were recorded on a Thermo Nicolet 380 FTIR spectrometer for 16 scans at a resolution of 2 cm -1 .

Density Functional Theory Calculations.
Frequency calculations were performed for Dye 1 using the Q-Chem software package. 18 The neutral Dye 1 molecule and its corresponding mono-anion were evaluated using the restricted and unrestricted B3LYP methods, with the 6-311G(d,p) basis set. Initial geometries were optimised at the same level of theory, to minimum energy structures in order to allow for harmonic vibrational analysis, which was performed using analytical second derivatives of the energy with respect to nuclear displacement. The resulting frequencies have been scaled by 0.9682. 19

Ultrafast Transient Absorption and Time-Resolved Infrared Spectroscopy.
Ultrafast TRIR spectra in CH2Cl2 solution were collected using the Nottingham ultrafast TRIR apparatus described in detail previously. 22 Samples for TRIR and TA spectroscopic measurements of the dyes on NiO were prepared as described above. The ULTRA laser system used for TRIR and TA spectroscopies on these samples has been described in detail elsewhere. 24

Dye synthesis
The synthesis of

Optical and Electrochemical Characterization.
The UV-visible absorption spectra of Dyes 1-4 in CH2Cl2 solution and adsorbed on NiO films are shown in Figure 2 and the results are summarised in Table 1. Differential pulse voltammograms of Dyes 2 & 4 are given in Figure S3 in the ESI.   All the dyes exhibit one prominent band in the absorption spectra in solution, appearing at ca. 600 nm with a shoulder at ca. 450 nm, which are both attributed to π-π* transitions. These values are similar to that observed for the previously reported dye HY103, which has the same donor and acceptor moieties as the dyes in this study. 23,28 The strong electron withdrawing nature of the (3-cyano-4,5,5-trimethyl-2(5H)-furanylidene)maleonitrile acceptor unit appears to red-shift the absorption maximum for these dyes compared to dyes reported previously which incorporate a tetrahydroquinoline electron donor and a cyanoacrylic acid acceptor (e.g. However, in the dyes presented here, the anchoring group is electronically decoupled from the dye so the shift in the electronic spectra must be attributed to the association of the bulk dye with the metal oxide surface. To investigate the association of the dyes with the NiO surface we measured the FTIR absorption spectra of the dyes alone (in a KBr pressed disc) and adsorbed on NiO (deposited on CaF2 windows). Both spectra are shown for Dye 1 and Dye 1|NiO in Figure 3 and comparisons of the key peaks are given in Table S1 in the ESI. It is clear that the peak corresponding to the C=O stretch at 1751 cm -1 present in the KBr spectrum was absent in the spectrum of the dye adsorbed onto NiO. 37,38 This is strong evidence for the dye anchoring through the carboxylic acid. 39 The peak at 2227 cm -1 corresponding to the C≡N stretch does not shift when the dye is adsorbed onto NiO. 40 However, the shoulder at 2209 cm -1 observed in the KBr spectrum was absent when the dye was immobilised on NiO. It is possible that this the result of a secondary binding mode to NiO via maleonitrile group on the acceptor unit. Alternatively the shoulder could arise from hydrogen bonding between dye molecules in the solid state which is disrupted when the dye is immobilised on NiO. An expanded view of these peaks are given in Figure S2 in the ESI.

Solution TRIR Measurements
To examine the photophysical behaviour of Dye 1 in CH2Cl2 solution we have used TRIR in both the ν(CN), organic ν(CO) and fingerprint regions. ps-TRIR spectra recorded in all regions following 650 nm excitation are shown (at selected delays) in Figure 4 below. on the anchor is not affected by the electronic transition. All the transient bands decay to reform the parent, with a lifetime of ca. 70 (± 15) ps ( Figure S5 in the ESI).

Ultrafast Transient Absorption Measurements
Ultrafast transient absorption measurements were undertaken of the dyes adsorbed onto NiO in both the visible and near-IR regions (500-1100 nm). The spectra recorded are shown in Figure 5.  500 nm (Table 2) assigned as the anionic dyes. Additionally, a broad feature is observed in the near-IR region of the spectra similar to that reported previously attributed to the oxidation of the NiO. 16,[41][42][43] The decay of both species for each dye were determined by fitting biexponentials and the fitted lifetimes are shown in Table 2. The lifetimes for all four dyes on NiO were < 30 ps and as such it might be expected that the dye regeneration step in a p-DSC device will not be able to compete with recombination. This will limit the photocurrent generated by p-DSCs incorporating these devices vide infra. In order to investigate the pushpull mechanism further we have undertaken TRIR measurements which allows more structural insight.

Ultrafast TRIR measurements
The ground state FTIR spectrum of Dye 1|NiO provided in Figure 3 contains bands at 1353, 1447, 1477, 1544 and 1605 cm -1 . Figure 6 shows the TRIR spectra of Dye 1|NiO at different delay times after excitation at 650 nm. The TRIR spectrum of the film 1 ps following photolysis (650 nm) shows that the parent bands are bleached and new features are produced at 1320, 1414, 1475, 1508, and 1585 cm -1 . As illustrated in Figure 6b, these bands are not stable and decay to reform the parent with time-constants of 2.1 ps and 24 ps. From our experiments, we were unable to resolve the rate of initial charge separation to form the reduced dye radical anions, which appears to take place on a sub-picosecond timescale. 14 Ultrafast charge separation is well documented for metal-free sensitizers on NiO when carboxylic acid anchoring groups conjugated to the chromophore are used.
Scaled harmonic frequency calculations were carried out for Dye 1 at the B3LYP/6-311G(d,p) level on both the neutral and mono-anionic forms. The band positions resulting from these calculations provide a good match with the TRIR spectra in Figure 6(a), and offer additional support for the assignment of the experimental spectra as depletion of the neutral species, followed by formation of the anionic state. A complete list of the associated frequencies and intensities can be found in Table S4 in the ESI.
These experiments suggest that charge separation between the dye and NiO can take place despite the electronic decoupling of the anchor. This is consistent with a previous reports that demonstrate sub-picosecond charge-separation from NiO with simple ruthenium and iridium complexes with electronically decoupled methyl phosphonic acid anchors, and perylenes which bind through benzoic acid. 17,44,45 However, we were surprised to find such rapid electron transfer for flexible, aliphatic anchoring groups in this study. We also investigated Dye 2|NiO, Dye 3|NiO and Dye 4|NiO using TRIR ( Figure S4 in the ESI).

p-Type Dye-Sensitized Solar Cells.
The photovoltaic performance of p-DSCs based on these sensitized by Dyes 1-4 is summarized in Table 3. Photocurrent density-voltage curves, dark current curves and IPCE spectra are given in Figures S6-9 the ESI. Under standard AM 1.5G sunlight irradiation (100 mW cm -2 ), the overall conversion efficiency of the Dye 2-sensitized solar cell is the highest. The VOC obtained for these devices were lower than those typically obtained for p-DSCs (90-120 mV) and the trends in VOC followed the trends in the onset of the dark currents. All four dyes in this study generated relatively low photocurrents, similar to a number of previously reported dyes which absorb at longer wavelengths. 46,47 However, upon inspection of the IPCE spectra for these dyes ( Figure S8  The IPCE at 400 nm for the p-DSCs in this study were in the region of 20-25%. I3 -/Isolutions are strongly coloured with an absorption maximum at 360 nm. The photocurrent response seen at λ <450 nm could be result of charge injection by the diiodide radical following the photolysis of I3 -. 32,[47][48][49] However, the photocurrent generated by a dye-free NiO device (0.2 mA cm -2 ) was lower than those obtained with the dyed p-DSCs. The UV-visible absorption spectra of the dyes in Figure 2 contain a second, higher energy absorption band at ca. 450 nm. Zhu et al.
previously reported that energy transfer from the excited dye to I3can result in charge injection by the I2 •radical. 48 IPCE spectra of Dye 2 and undyed NiO taken from 330-550 nm are given in Figure S9 in the ESI, from this it is apparent that light absorption by the dye does not contribute to the photocurrent in these devices. In agreement with reference 48, there is a slight enhancement to the photocurrent at the wavelengths attributed to injection by I2 •radical when the dye is present. One possible explanation for this is that the dye interacts with the electrolyte and increases the local concentration of I3close to the NiO surface. [50][51][52] However we note that there can be significant variation in the photocurrent generated in the absence of a dye due to slight differences in the film thickness (±50 nm) and exposed NiO surface.

Conclusions
Photoinduced charge-separation between the series of laterally anchored dyes and NiO occurred on an ultrafast timescale (τinj < 1 ps). This is despite the anchoring group being electronically decoupled from the surface. The driving force for charge-separation is slightly lower than is usually found for organic sensitizers on NiO and, therefore, less energy is wasted in the injection step and more light is collected at longer wavelengths. This supports our efforts to shift the absorption maxima of the dyes to lower energy as required for tandem DSCs.
However, results from the IPCE measurements indicate that very little photocurrent is being generated at the absorption maxima of these dyes. The ultrafast TA and TRIR experiments revealed that rapid recombination between the photoreduced dye and the h + at the NiO surface competes with regeneration of the dye by I3in the electrolyte. Evidence from the FTIR and UV-visible absorption spectroscopies is suggestive of electronic interaction between the dye and the NiO, suggesting the dye lies in close proximity to the NiO surface. This close proximity is a likely cause of the rapid recombination. TRIR has been used to investigate charge separation between NiO and a sensitizer. For these experiments, the charge-transfer mechanism appears to be straight forward -we appear to form one photo-induced product from one ground state. However we envisage that this tool will be increasingly important for more complex systems with more additional intermediate states or in the presence of additional species such as in our work to investigate photocatalysis using dye-sensitized NiO.