Effect of extending conjugation via thiophene-based oligomers on the excited state electron transfer rates to ZnO nanocrystals

Oligothiophene dyes with two to five thiophene units were anchored to oleate-capped, quantum-confined zinc oxide nanocrystals (ZnO NCs) through a cyanoacrylate functional group. While the fluorescence of the bithiophene derivative was too weak for meaningful quenching studies, the fluorescence of the dyes with three, four and five thiophene rings was quenched upon binding to the NCs. Ultrafast pump-probe spectroscopy was used to observe the singlet excited states of the free dyes dissolved in dichloromethane as well as attached to a ZnO NC dispersed in the same solvent. When the dyes were bound to ZnO NCs, ultrafast spectroscopic measurements revealed rapid disappearance of the singlet excited state and appearance of a new transient absorption at higher energy that was assigned to the oxidized dye based on the similar absorption observed upon oxidation of the dye using nitrosonium ion. The appearance lifetimes of the oxidized dyes were assigned to the excited state electron transfer and were 36 ± 2, 22.3 ± 3.9, 26.5 ± 1.5 and 19.4 ± 0.8 ps for bi-, ter-, quarter- and quinquethiophene dyes respectively. Two factors contributed to the similarity in the electron transfer lifetime. First the excited state energies of the dyes were similar, and second, the free energy for electron transfer reaction was sufficiently large to move the event into the energy-independent regime.


Introduction
Monodisperse zinc oxide nanocrystals (NCs) have been shown to serve as a useful platform for studying light-induced charge separation as found in dye-sensitized solar cells. Using a variety of dyes, including porphyrins, 1 ruthenium tris(bipyridine) complexes 2 and terthiophenes, [3][4][5] our past work has addressed the impact of dye structure, excited state energy of the dye, anchoring group and ZnO NC diameter on the excited state electron transfer lifetime. Thiophene oligomers offer the ability to tune the dye absorption through systematic extension of the oligomer chain, and associated conjugation, with the potential to cover the entire visible spectrum within a chemically homologous series. The ability to cover a broad range of the spectrum is advantageous in the context of light harvesting, but the question that immediately arises in the context of a dye sensitized photovoltaic is how the extension in conjugation influences the rate of electron transfer.
One factor is the donor-acceptor energetic alignment. However, changes in the absorption energy in the thiophene oligomer series are dominated by increases in the HOMO energy with length 6 and this suggests that energetic alignment between the excited donor and the ZnO accepter should remain largely unaffected. Another factor is the donor-acceptor distance. [7][8][9] In a series of thiophene oligomers, as the oligomer chain is extended one might expect a decrease in the donor acceptor coupling as the average distance increases, with an associated reduction in the electron transfer rate. A prior study on TiO2 adjusted the donor acceptor distance through extension of a conjugated bridge, however this had only a minor influence on the absorption spectrum and the electron injection rate was not resolved. 10 The hypotheses investigated in this report is that a series of thiophene oligomer dyes can be used to cover the visible spectrum while maintaining rapid, efficient electron transfer by mitigating the systematic increase in donor-acceptor distance with an electron withdrawing linker that localizes the excited state near the donor-acceptor interface.
Recent reports have established cyanoacrylates as effective dye terminating groups in dyesensitized solar cells. [11][12][13][14] In addition to the carboxylic acid functional group serving as the anchor to metal oxide surface, the entire cyanoacrylate moiety is a powerful electron acceptor, which reduces the HOMO-LUMO gap and localizes the excited state electron density in close proximity to the metal oxide surface. 6,14,15 In this study, we bound a series of oligothiophene dyes to ZnO NCs through a cyanoacrylate anchoring group (Figure 1) and measured their excited state electron transfer rates using ultrafast pump-probe spectroscopy. Compounds 3T and 4T are known to be effective interface modifiers in dye-sensitized solar cells, 6,16 and 5T itself has been demonstrated an effective dye for solar cells. 17,18 This series of compounds is of particular interest because, despite large differences in their absorption spectra, the energies of the LUMOs, which are primarily localized on the cyanoacrylate moiety and the adjacent thiophene ring, of 2T, 3T, 4T and 5T were reported to be similar. 6 As such, rate measurement of excited state electron transfer could reveal whether or not the rates are independent of the chromophore structure past the anchoring group and the first thiophene. Such behavior could conceivable allow one to tune and expand the light harvesting bandwidth of a solar cell using multiple dyes with minimal problems resulting from substantially different excited state electron transfer rates. performed under argon or air using 1 cm quartz cuvettes. The presence of air did not quench the fluorescence. Fluorescence spectra were compensated for changes in the instrument's sensitivity at longer wavelengths. The synthesis and characterization of 2T, 3T, 4T, and 5T have been published elsewhere. 6 Figure 2 shows the electronic adsorption and fluorescence spectra of the dyes in dichloromethane solution.

Experimental
ZnO Nanocrystal Synthesis. The synthesis of ZnO NCs was adapted from Gamelin and coworkers. 19  mmol) over 8 min and allowed to stir at 40°C for 25 min. The nanocrystals were purified by precipitation with ethyl acetate (~9 mL) and isolated after centrifugation. To cap the ZnO NCs with oleate groups, the NCs from the above preparation were combined with 5 mL of a 12.7 mM solution of oleic acid in CH2Cl2 and sonicated. The capped NCs were precipitated with 9 mL of absolute ethanol and the mixture centrifuged. After removal of the supernatant, CH2Cl2 was added (10 mL) and the mixture was sonicated.
The onset of absorption in the UV region was used to measure the NC diameter. The oleic acid-capped NCs had diameters of 4.3 nm. Nanocrystal concentrations were estimated assuming full conversion of starting material and were used within 24 hours of synthesis.
Quenching Experiments. For 3T and 4T, samples were prepared in CH2Cl2 using varying aliquots (37 µL to 5.5 mL of a 2.28 × 10 -5 M ZnO NC dispersion) and a sufficient volume of a stock dye solution to bring the final dye concentration to 1.80 x10 -5 M following dilution to the total sample volume of 7.0 mL. This generated a range of dye-to-ZnO NC ratios of ~150:1 to 1:1.
For 5T, samples were prepared in CH2Cl2 using varying aliquots (33 µL to 3.28 mL of a 1.92 × nm. Sample emission was detected using an avalanche photodiode at a right-angle geometry at the exit of a double monochromator (Jobin-Yvon TRIAX-320). Fluorescence quantum yields, ΦF, were calculated using the simple point method employing Rhodamine 6G as standard reference.
All emission spectra were collected at 90º relative to the excitation light. Optical densities were less than 0.2, and all solutions were de-aerated with argon for 5 minutes before the spectrum was recorded. previously. 5 Briefly, the output of a home-built, amplified, Ti:sapphire based laser system producing 100 fs, 0.5 mJ, 1.55 eV pulses at 1 kHz was split to create the pump and probe. The majority of the energy, 90%, was either frequency doubled in a BBO crystal or used to pump a home-built non-colinear OPA producing excitation pulses at 3.1 eV or in the range 1.8-2.6 eV, respectively. The remaining 10% of the energy at 1.55 eV was focused into a 2 mm thick sapphire window to create a continuum probe. The pump and probe were crossed in a 1 mm optical pathlength sample cell with suprasil quartz windows (Starna), and the samples were flowed through the sample cell at a rate of 35 ml/min. After the sample, the probe continuum was columnated, refocused into a monochromator (Princeton Instruments 2150), and dispersed onto a linear silicon diode array with 256 elements (Hamamatsu) providing an effective resolution of 2 nm. The probe spectrum was recoded for every laser pulse, while the pump was modulated at half the laser repetition rate using a mechanical chopper wheel. The difference in optical density, DOD, was determined for each pulse pair (pump on -pump off), and then averaged over 50,000 pulse pairs at each delay time. Pump pulse irradiance was typically 140 µJ/cm 2 . There was no change in the absorption spectrum before and after the pump-probe experiments, and thus no evidence of photodegradation. The instrument response was 80 fs (Gaussian, FWHM).

Results
In dichloromethane the electronic absorption spectra ( Figure Figure 3 for 3T, 4T, and 5T using ZnO NCs as the quencher (2T was not studied as the fluorescence was too weak). Each graph exhibited an initial linear region, which was consistent with static quenching, however, maximum quenching appeared as the dye:ZnO NC ratio approached ~20:1. Further quencher addition caused a decrease in emission until a plateau was reached at a dye:ZnO NC of approximately two. The general shape of the Stern Volmer graphs observed for these dyes paralleled those reported in earlier studies, where the suggested causes were concentration quenching 21 or competition between dye binding sites that were or were not quenched. 3,4 The current work focused on the ultrafast quenching kinetics at a dye to ZnO NC of 2:1 in order to minimize any complications attributed to dye-dye interactions. 2T. Figure 4A shows the observed spectra of 2T at select time points following excitation at 3T. Figure 4C displays the pump-probe spectral changes at select time points for a solution of 3T in CH2Cl2 following a 2.38 eV pump pulse. Immediately following excitation the difference spectrum exhibited a GB at 2.64 eV and a positive signal, due to the singlet excited state having a maximum intensity at 2.07 eV. As a function of time, a signal with maxima at 2.11 and 2.35 eV appeared and remained present to the longest time delay measured (900 ps) Photolysis of a dichloromethane solution of 3T in a 2:1 ratio with ZnO NCs ultimately led to a difference spectrum with a new maximum located at 2.03 eV. This was sufficiently close in energy to the oxidized dye ( Figure 4D) to assign the peak to the product of excited state electron transfer. In addition to the peak at 2.03 eV, the GB no longer measurably decayed on the time scale of the experiment, indicating another process (i. e. oxidation of the dye) occurred to prevent recovery of the ground state during the lifetime of these measurements.
4T. Figure 4E shows the spectral evolution in the visible region of 4T in CH2Cl2 upon 2.38 eV pump excitation. Two distinct temporal events were discernable from 0 -900 ps coupled with a change that occurred on a time scale beyond the limit of the current experiments (> 900 ps). 5T. Following a pump pulse at 2.38 eV, TA formed at 1.47 eV, and the feature seemed to stretch from the end of the detection range in the near IR (1.30 eV) to 1.66 eV ( Figure 4G). Due the interference of the residual light from the fundamental of the laser around 1.55 eV used to create the probe pulse, the absorption was interrupted, so the whole feature could not be studied.
The signal did not return to the baseline by 900 ps. The ground state hole feature peaked at 2.55 eV and there was no stimulated emission for 5T. There was also a broad feature centered around 2.07 eV.
In the dye and ZnO NC solution, a broad TA similar to the dye alone was initially seen. As this feature decayed, the emergence of another absorption appeared at 1.66 eV ( Figure 4H). The absorption was attributed to the oxidized dye and matched well with chemical oxidation results ( Figure 4H). When fitting the spectral changes of solutions containing a 2:1 ratio of dye to ZnO NCs at these same probe energies, there was the same initial instantaneous jump with the creation of the excited state, and then a subsequent single exponential rise that was assigned to the appearance of the cation. The time constants for creation of the cation from the fits shown in Figure 5  computational study of unsubstituted bithiophene and concluded that the S1 lifetime would be 1.8 ps, which agreed with the current study. 24 Three exponential terms, two decays and one rise, were required to model the change in spectral intensity at 2.35 eV ( Figure 6). The smallest time constant (1.2 ± 0.1 ps) corresponded to the early time feature in the dye-alone pump-probe spectrum. In the fit, the time constant of the second decay and the rise were required to be the same. The 36 ± 2 ps time constant was assigned to the excited state electron transfer resulting in the appearance of the oxidized dye.

Discussion
The conduction band minimum energy of 4.3 nm ZnO NCs was calculated using the values for the change in the band gap as a function of size reported by Sarma and coworkers 25 and adding it to the bulk ZnO band gap and the work function measured on a nanocrystalline ZnO film.
The electronic structures of 2T, 3T, 4T and 5T were reported previously along with their lowest excited state energies in CH2Cl2. 6 In each of these molecules the LUMO orbital density was shown to be delocalized over the cyanoacrylic group and the adjacent thiophene unit. As this functionality is identical in each compound, there was little change in the energy of the LUMO among the dyes, a fact found to be consistent between the computational and experimental electrochemical measurements. 6 Transient absorption of the excited states was examined using ultrafast pump-probe spectroscopy on solutions where the number of dyes per ZnO NC was kept low (~2:1) to minimize the impact of dye aggregation. At higher dye to NC ratios (~20:1), Stern Volmer experiments display behavior consistent with concentration quenching. Based on steady state spectroscopic measurements, the bands of the stimulated emission and ground state hole in the difference spectra were known. Figure 4 shows that all the compounds exhibit transient absorption bands in the visible-near IR region range corresponding to the excited state. The main excited state absorption band overlaps in all cases with the bleach in ground state absorption and the stimulated emission spectrum to some extent, which complicates the exact assignment of the absorption maximum.
The energies of the main transient absorption bands agree with previous studies on extended thiophene oligomers that are unsubstituted, including the clear trend when going from two to five thiophenes where the excited state absorption band appears at lower energies. 22 Measurement of the excited state decay of the free dyes was critical for comparison to the changes observed for the dye/ZnO NC dyads. Both the dye alone and the dye/NC systems exhibited multiphasic dynamics. The lifetimes observed for 3T, 4T and 5T were similar to related terthiophene-based dyes that did not have cyanoacrylate anchoring groups. 3,5 The smallest dye, 2T, exhibited much shorter lifetimes. When ZnO NCs were combined with the dyes in an approximate 2:1 dye to NC ratio, a distinct transient absorption was observed that corresponded closely to the absorption observed from chemical oxidation of the dye using nitrosonium ion. The appearance of this absorption was assigned to electron transfer from the excited state of the dye to the ZnO NC. The excited state electron transfer time constants (Table 1) Integration over E accounts for transfer into available conduction band states. Given the similarities in the LUMOs for all of the dyes, with the excited state charge density enhanced at the cyanoacrylate anchoring group, as well as the similarity in the link to the ZnO NCs, we anticipate that the electronic coupling, H(E) will be essentially the same for all of the dye/ZnO NC dyads.
Addition of subsequent thiophene rings systematically reduces the energy of the absorption band primarily due to an increase in the energy of the HOMO. 6 The significant changes in the absorption energy across this series of compounds does not result in a significant change in the relative energies of the excited state relative to the ZnO NC conduction band. As shown in Table  1, the free energy change for electron transfer from the excited state of the dye to the conduction band of the ZnO NC, ΔGo, for the dyes are within 0.09 eV of one another. In addition to the small differences in the driving force, the free energy changes (ΔGo ~ -1 eV) exceeds our expectation for the reorganization energy, and is thus sufficiently large to provide access to barrierless (l + E = -DG0) transfer in all cases, greatly reducing the dependence of the rate on the exponential term in eq. 2. Consistent with this, the electron transfer time constants decrease by less than a factor of two in going from 2T to 5T.          to ZnO. Each color corresponds to the averaged ΔOD in the visible range for a selection on time delays between the pump and probe in picoseconds. The blank space around 1.55 eV is due to residual light from the fundamental in the continuum probe. For the oxidized dye absorption, peaks are shown only for the oxidized dye and any ground state dye absorption has been omitted.