1. Introduction
Various photocatalytic systems, mainly consisting of photosensitizers and catalysts, have recently been developed for CO
2 reduction [
1,
2,
3,
4]. In these systems, the photosensitizer efficiently absorbs visible light and generates electrons that can transfer to the catalyst as an initial step of the photocatalytic process. The catalyst then converts CO
2 into energy-rich compounds, utilizing the transferred electrons. Intra- or intermolecular electron transfer (ET) is a prerequisite process in photocatalytic CO
2 reduction, and its dynamics can affect the overall performance of the photocatalytic system. Therefore, it is crucial to elucidate the detailed dynamics and mechanisms of ET in the photocatalytic reactions for developing efficient photocatalytic systems.
However, tracking the overall ET dynamics of photocatalytic systems during the CO
2 reduction process is challenging. The complete cycle of CO
2 reduction requires sacrificial electron donors (SEDs) and other additives that can aid CO
2 binding to the catalyst [
5,
6,
7,
8]. Furthermore, advanced photocatalytic systems often use metal-organic frameworks (MOFs) or semiconductors like TiO
2 as mediators that can collect and transport multi-electrons to enhance ET efficiency and reduction performance [
4,
9,
10,
11]. Hence, the overall reactions of photocatalytic systems commonly involve multiple ET processes whose time scales span a wide range from picoseconds even to minutes, complicating the reaction dynamics and mechanisms. Deciphering the complete ET processes in photocatalytic reactions demands thorough time-resolved studies on a wide time scale under delicately controlled experimental conditions.
In this study, we investigated the ET dynamics in photosensitization reactions of a series of zinc porphyrin derivatives (
ZnPs), which have different peripheral substituents shown in
Figure 1a, using ultrafast spectroscopies, i.e., picosecond time-resolved fluorescence spectroscopy and femtosecond transient absorption (TA) measurements.
Figure 1b shows the overall ET pathways tracked in this study. Our previous studies have shown that the
ZnPs can serve as a photosensitizer in the binary hybrid system with a heterogeneous TiO
2/Re(I) (Re(I) catalyst anchored to TiO
2 particle) for photocatalytic CO
2 reduction. However, the collisional ET from
ZnPs to TiO
2 was measured to be substantially slow (
k = ~10
−3 s
−1) with high activation energy (58 kJ/mol) determined by temperature-dependent and second-to-minute-resolved kinetics of UV/vis spectrum [
9]. Herein, equipped with a high time resolution, we revisited the ET reactions of
ZnPs with a SED and TiO
2 particle to unveil the fast dynamics possibly occurring in the photosensitization reactions. Stern-Volmer analysis of the fluorescence lifetimes yielded quenching rates that report on the ET dynamics between photoexcited
ZnPs (
ZnPs*) and SED. Additionally, global analysis of TA spectra revealed the dynamics of further ET reaction between one electron reduced species (OERS) of
ZnPs (
ZnPs•–) and TiO
2 particle, enabling us to estimate reaction rates that are significantly faster than those previously measured with low time resolution and even faster than the diffusion rates in solution. This suggests the ET dynamics of
ZnPs•– occurring very near the TiO
2 surface.
2. Results and Discussion
Figure 2a shows the steady-state UV/vis absorption spectra of
ZnPs. They have typical spectral features of metalloporphyrin with an intense Soret (B) band in the range of 400 to 470 nm and a Q-band in a region longer than 580 nm. The steady-state emission spectra appear as a mirror image of the Q-band (inset of
Figure 2a). As the substituent changes from silanyethynyl to ethynyl and ethyl, absorption and emission spectra blue-shifts due to weakened electron-donating ability for
ZnP→
ZnPAcet and broken π-conjugation for
ZnPAcet→
ZnPEt. The fluorescence lifetimes of
ZnPAcet and
ZnPEt are almost identical, having a time constant of 2.9 and 3.0 ns, respectively (
Table 1). However,
ZnP has a slightly shorter fluorescence lifetime with a time constant of 2.4 ns (
Table 1): its bulky substituent, silanyethynyl, can supply more nonradiative decay channels, presumably via vibrational relaxation, enhancing the fluorescence decay rate. Here, we did not observe any hint of aggregation under the sample concentration (10 μM) in the UV/VIS absorption spectra, and thus, the measured fluorescence lifetimes were not affected by aggregates in
ZnPs.
In the photosensitization reactions of
ZnPs, the first step is the ET reaction of photoexcited
ZnPs with BIH to form an OERS,
ZnPs•–. BIH plays a role as a SED in this reaction. Then, an excess electron of
ZnPs•– is transferred to TiO
2, which serves as an electron reservoir for ET to the Re(I) catalyst in the binary hybrid system, when the OERS encounters TiO
2 particles. To understand the overall dynamics of this reductive quenching mechanism, we first measured fluorescence quenching kinetics of
ZnPs with BIH, i.e., the dynamics of
ZnPs•– formation, as displayed in
Figure 2b. The fluorescence lifetimes of
ZnPs decrease with the addition of BIH because collisional ET with BIH quenches the fluorescence. The higher concentrations of BIH, the higher collisional frequency, and thus, the shorter lifetimes of
ZnPs. The relationship between the fluorescence lifetimes and the concentration of a quencher, [Q], is described by the Stern-Volmer equation: [
12]
where
τ0 and
τ are the fluorescence lifetimes in the absence and presence of quencher, respectively.
KSV is the Stern-Volmer quenching constant and is given by
KSV =
kqτ0 where
kq is the bimolecular quenching constant. According to Equation (1), the linear fits to the Stern-Volmer plot of the fluorescence lifetimes resulted in
KSV values, and corresponding
kq values were calculated from
τ0 (
Figure 2c and
Table 1). The calculated
kq values of
ZnP and
ZnPAcet are 6 to 8-fold larger than that of
ZnPEt, which is consistent with the driving force (ΔG) of ET from BIH to photoexcited
ZnPs: based on the oxidation and reduction potentials in the cyclic voltammetry measurements the estimated ΔG is −0.34/−0.36 eV for
ZnP/
ZnPAcet and −0.10 eV for
ZnPEt [
9]. Unlike conjugated substituents in
ZnP and
ZnPAcet, non-conjugated substituent, ethyl, in
ZnPEt does not lower the molecular orbital (MO) energy levels, keeping the reduction potential of photoexcited
ZnPEt close to the oxidation potential of BIH and yielding less negative ΔG. More negative ΔG generally coincides with a lower reaction barrier and, thus, a faster reaction rate as we observed larger
kq values for
ZnP and
ZnPAcet.
Interestingly, the
kq value of
ZnP obtained by the fluorescence lifetime measurements in this study is much smaller than the previously acquired value by quenching the fluorescence intensity (
kq [10
9 M
−1 s
−1] = 1.9 vs. 9.3). Different origins of quenching mechanisms can explain the discrepancy in the quenching rates between two methods. The quenching rate determined by the fluorescence lifetime measurements only reports on the collisional (or dynamic) quenching. In contrast, the method with the fluorescence intensity can include the quenching through the formation of a nonfluorescent ground-state complex between fluorophore and quencher (static quenching) as well [
12]. If both dynamic and static quenching happens, the fluorescence intensity measurements will result in larger
kq values than those from the fluorescence lifetime measurements even for the same quenching reaction. Therefore, the discrepancy in
ZnP’s
kq observed by two methods suggests the possibility of complex formation between
ZnP and BIH in the ground state. In many cases, transition metals form complexes with imidazole and benzimidazole derivatives in porphyrins and phthalocyanines [
13,
14,
15].
On the other hand, the difference in
kq values between the two methods is not significant for
ZnPAcet. (
kq determined by the fluorescence intensity quenching is 3.0 × 10
9 M
−1 s
−1) [
9]. The electron-donating from the substituent is less in
ZnPAcet than in
ZnP, which can reduce the complex formation between
ZnPAcet and BIH. For
ZnPEt, the fluorescence quenching by BIH was not significant in both intensity and lifetime measurements, and it seems unreliable to compare the values and discuss the complex formation.
To investigate the ET dynamics of the further photosensitization reaction beyond the formation of
ZnPs•–, we carried out femtosecond TA measurements initially with
ZnPs and then in the presence of BIH or/and TiO
2 as depicted in
Figure 3 (see also
Figure S2 in Supporting Information). In the absence of BIH and TiO
2, the TA spectra of
ZnPs do not show dramatic changes within the time delays of 7.4 ns (
Figure 3a). For
ZnPs and
ZnPAcet, the ground-state bleach signals of the Soret band near 430~450 nm and the Q-band near 630~640 nm slightly recover, concurring with modest growth of induced absorption signals near 475 nm. The TA spectra of
ZnPEt also exhibit similar behaviors except for minor decay of induced absorption signal near 460 nm. In the presence of BIH, the TA spectra of
ZnP and
ZnPAcet substantially evolve with time (
Figure 3b): both the ground-state bleach and induced absorption signals significantly decrease in magnitude with an increase of time delay, reflecting the ET reaction with BIH. (Note that the recovery of the ground-state bleach signal represents a mixed evolution of the ground-state recovery of
ZnPs* and the absorption of newly formed
ZnPs•–.) However, we could not observe any spectral feature attributable to
ZnPs•– presumably due to the overlaps of TA signals and the limited range of our probe wavelength. For the TA spectra of
ZnPEt, the presence of BIH does not seem to primarily affect their evolution in line with the minute change of fluorescence lifetimes observed in the quenching measurements. Finally, adding TiO
2 into the sample solution in the presence of BIH enhances the decrease of bleach and absorption signals in the TA spectra of
ZnP and
ZnPAcet, which implies the ET dynamics from
ZnPs•– to TiO
2 particle. In contrast, the TA spectra of
ZnPEt measured with the addition of BIH and TiO
2 reveal the same behaviors as those with only BIH added.
To analyze the evolution of TA spectra in detail, we globally fitted the TA data with the first-order kinetic model. The kinetic model for the best fit to all the TA data required three species, A→B→C→ground state (GS).
Figure 4 displays evolution-associated spectra (EAS) obtained from the fits and representative decay profiles. The kinetic parameters associated with each EAS are tabulated in
Table 2 as an inverse form of the time constant. Without any additives, the TA spectra of
ZnPs can be fitted with the associated time constants (
τ) of 0.7~2.2 ps for A→B, 2.3~3.4 ns for B→C, and more than 100 ns for C→GS. The first evolution, A→B, in a few picoseconds is ascribed to the internal conversion (IC) process from the higher singlet excited-state (S
n) to the lowest singlet excited-state (S
1) since the pump wavelength was tuned to the B-band excitation at 435 nm in the TA measurements [
16,
17,
18,
19]. Then, the B state of EAS, which corresponds to the S
1 state, evolves into the C state with
τB→C that matches the singlet state lifetime (
τ0) measured with the time-resolved fluorescence, enabling us to assign the C state of EAS as the triplet excited state (T
1) of
ZnPs. The C state lives longer than the upper limit of our apparatus’s time delay (~8 ns) and has a long time constant (~100 ns) from the global fit, confirming its triplet character.
Even with the addition of BIH or/and TiO
2, the EAS of A, B, and C states do not essentially change from those observed in
ZnPs, still representing the S
n, S
1, and T
1 states, respectively: their
τA→B values of 1.1~2.8 ps are the same as the IC process (S
n→S
1) in
ZnPs, and
τC→GS values of longer than 100 ns are agreeable to the triplet state decay (
Figure 4 and
Table 2). However, the presence of additives in the
ZnPs solution mainly affects the S
1 state lifetime,
τB→C. Like in the fluorescence quenching experiments where the ET reaction with BIH shortens the S
1 state lifetime of
ZnPs,
τB→C decreases in the TA measurements with the addition of 0.3 M BIH: the fitted values of
τB→C are 822 and 737 ps for
ZnP and
ZnPAcet and 2.0 ns for
ZnPEt (
Table 2). These
τB→C values are comparable to the S
1 state lifetimes for
ZnPs + 0.3 M BIH (
τZnPs+BIH), which are estimated to be 1.0, 0.9, and 2.4 ns for
ZnP,
ZnPAcet, and
ZnPEt, respectively, by
(where
τ0 and
kq are from
Table 1 and 0.3 is from the concentration of BIH). This demonstrates that the TA experiments reveal the ET dynamics of
ZnPs with BIH, which is consistent with the results from the Stern-Volmer analysis.
In
ZnP and
ZnPAcet, the
τB→C value further decreases with the addition of both BIH and TiO
2 (
Table 2), meaning that the decay of the S
1 state is additionally enhanced by the presence of TiO
2 in the solution. According to the reductive quenching mechanism, the electron transfer to TiO
2 should occur not directly from the photoexcited
ZnPs in the S
1 state (
ZnPs*) but from the OERS (
ZnPs•–) that is formed after the ET reaction with BIH. Indeed, the TA experiments with only TiO
2 added to the
ZnPs solution showed no change in the kinetics of TA spectra compared to the sample solution without additives (
Figure S2 and
Table S1), confirming that the direct reaction between
ZnPs* and TiO
2 is unlikely. However, the TA spectra of
ZnPs•– can not be resolved in our experiments, and tracking its explicit dynamics is impossible. Instead, the disappearance of
ZnPs•– by the ET reaction with TiO
2 will be incorporated in the decay of
ZnPs* because quenching of
ZnPs•– by TiO
2 generates the ground state species (
ZnPs) that has no TA signal, conceivably contributing to the decay of
ZnPs*, i.e., the additional enhancement of decay rate in the B→C evolution. Assuming that quenching by TiO
2 is the only factor for the
τB→C decrease, we can estimate the quenching rate of
ZnPs•– in the ET reaction with TiO
2 (
kq,OERS) as
where
τB→C,BIH and
τB→C,BIH+TiO2 are the time constants for the B→C evolution measured in the presence of BIH and BIH + TiO
2, respectively (
Table 2). The calculated
kq,OERS values of
ZnP and
ZnPAcet are
and
s
−1, respectively. In
ZnPEt, the formation of OERS is so slow that its quenching by TiO
2 is not observed within the experimental time window of TA measurements, i.e., no change in
τB→C with the addition of TiO
2. Compared to the collisional ET rates of
ZnP•– with TiO
2 previously determined by the UV/vis spectrum change (
s
−1), [
9] the
kq,OERS values of
ZnP and
ZnPAcet differ by many orders of magnitude. Given that the concentration of BIH was lower only by three times in the UV/vis spectrum measurements with the same concentration of TiO
2, the fast ET dynamics of the OERS observed in this study should reflect a different mechanism from the high activation energy process formerly revealed with a slow time resolution, suggesting another ET route.
In fact, the ET dynamics of the OERS happening on a few nanosecond time scales are even faster than the diffusion rate in solution. The diffusion rate constant of a dye in most organic solvents is typically on the order of 10
9~10
10 M
−1 S
−1 at room temperature [
20]. Under our experimental condition with 10 μM of
ZnPs, the concentration of the OERS formed after the ET between photoexcited
ZnPs and BIH will be much less than 10 μM. Then, the diffusion rate constant of the OERS will become far less than 10
4 s
−1 which is significantly slower than the
kq,OERS values obtained here. (Note that the particle size of TiO
2 used in this study spans from 0.2~3 μm, [
21], i.e., the particles are very large, and we can assume that their diffusion is almost negligible compared to
ZnPs). In this regard, the
kq,OERS values may report the ET dynamics of the OERS formed at the very vicinity of the TiO
2 surface. The interfacial electron transfer on the semiconductor surface generally occurs on picoseconds time scale or even faster times [
22,
23,
24,
25,
26,
27,
28] KC et al. reported that when the
ZnP derivative is covalently attached to the TiO
2 surface, the electron injection from
ZnP•– to TiO
2 occurs within 30 ps [
29]. Therefore, a few nanoseconds dynamics observed in this study can correspond to the ET dynamics of intermediate regime where the dye molecules in solution are very near the semiconductor surface without a direct connection like a covalent bond and electronic interaction between the dye and surface is negligible. Still, no diffusion of dye is required for the ET in this regime. This implies that the overall intermolecular ET process in photosensitization reaction can happen on multi-time scales, and thus, the spectroscopies with the time resolution of multi-time scales are required to uncover the complete intermolecular ET dynamics.
3. Materials and Methods
All reagents were purchased from Sigma Aldrich (St. Louis, MO, USA) and used without further purification. TiO
2 (Hombikat UV-100) was purchased from Huntsman (The Woodlands, TX, USA).
N,
N-Dimethylformamide (DMF) was distilled from calcium hydride and stored over molecular sieves. [5,15-bis[(triprop-2-yl silanyl)ethynyl]-10,20-bis[2,6-di(dodecyloxy)phenyl]porphinato]zinc (
ZnP), [5,15-diethynyl-10,20-bis[2,6-di(dodecyloxy)phenyl]porphinato]zinc (
ZnPAcet), and [5,15-diethyl-10,20-bis[2,6-di(dodecyloxy)phenyl]porphinato]zinc (
ZnPEt) samples and 1,3-dimethyl-2-phenyl-1,3-dihydrobenzimidazole (BIH) were synthesized according to the previously reported procedures [
9,
10].
Steady-state absorption and emission spectra were collected using a Cary 5000 UV-Vis-NIR (Agilent Technologies, Inc., Santa Clara, CA, USA) and Cary Eclipse (Varian, Palo Alto, CA, USA), respectively. For picosecond time-resolved fluorescence spectroscopy, a home-built cavity-dumped Kerr-lens mode-locked Ti:sapphire oscillator was used. The 800 nm output was doubled in frequency using a 100 μm thick BBO (β-barium borate) crystal to generate the excitation pulses at 400 nm. A parabolic mirror was employed to focus the excitation pulse onto the sample and to collect fluorescence with a confocal geometry. The collected light was sent to a monochromator (SP-2155, Teledyne Princeton Instruments, Thousand Oaks, CA, USA) and detected with a single-photon counting module (id 100-50, id Quantique). A commercial TCSPC board (SPC-130-EMN, Becker & Hickl Inc., Berlin, Germany) was used to record time-resolved fluorescence with a time resolution of about 50 ps. All the instruments were controlled in unison by using a home-built LabVIEW software (LabVIEW 2016, 16.0, 32 bit, National Instruments, Austin, TX, USA).
Femtosecond TA measurements were previously described in detail [
30]. Briefly, a Ti:sapphire regenerative amplifier system at 1 kHz (Spitfire Ace, Spectra Physics, Inc., Milpitas, CA, USA), which was seeded by a Ti:sapphire oscillator (MaiTai SP, Spectra Physics, Inc., Milpitas, CA, USA) and pumped by a diode-pumped Q-switched laser (Empower, Spectra Physics, Inc., Milpitas, CA, USA), was used for time-resolved measurements. An optical parametric amplifier (TOPAS prime, Spectra Physics, Inc., Milpitas, CA, USA) converted the 800 nm fundamental output into 435 nm for excitation. For TA measurements, a small residual of 800 nm fundamental light was focused on a water-filled cuvette to generate white light probe pulses directed to the computer-controlled translational delay stage. The mechanical chopper alternatively blocked pump pulses at 500 Hz to calculate TA spectra from the probe intensity detected by the CCD detector when the pump is on and off. Both pump and probe pulses were focused on a 1 mm-sandwich sample cell containing a magnetic stirring bar. The TA spectra were recorded by a commercial pump-probe spectrometer (Helios, Ultrafast Systems, LLC, Sarasota, FL, USA). Before kinetic analysis, the TA data were background-subtracted and chirp-corrected by using Surface Xplorer 4 (Ultrafast Systems, LLC, Sarasota, FL, USA). Kinetic data from multiple different wavelengths fit a first-order kinetic model using the global analysis programs written in MATLAB.