Dynamics of Electron Transfers in Photosensitization Reactions of Zinc Porphyrin Derivatives

Abstract

Photocatalytic systems for CO₂ reduction operate via complicated multi-electron transfer (ET) processes. A complete understanding of these ET dynamics can be challenging but is key to improving the efficiency of CO₂ conversion. Here, we report the ET dynamics of a series of zinc porphyrin derivatives (ZnPs) in the photosensitization reactions where sequential ET reactions of ZnPs occur with a sacrificial electron donor (SED) and then with TiO₂. We employed picosecond time-resolved fluorescence spectroscopy and femtosecond transient absorption (TA) measurement to investigate the fast ET dynamics concealed in the steady-state or slow time-resolved measurements. As a result, Stern-Volmer analysis of fluorescence lifetimes evidenced that the reaction of photoexcited ZnPs with SED involves static and dynamic quenching. The global fits to the TA spectra identified much faster ET dynamics on a few nanosecond-time scales in the reactions of one-electron reduced species (ZnPs^•–) with TiO₂ compared to previously measured minute-scale quenching dynamics and even diffusion rates. We propose that these dynamics report the ET dynamics of ZnPs^•– formed at adjacent TiO₂ without involving diffusion. This study highlights the importance of ultrafast time-resolved spectroscopy for elucidating the detailed ET dynamics in photosensitization reactions.

1. Introduction

Various photocatalytic systems, mainly consisting of photosensitizers and catalysts, have recently been developed for CO₂ 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 CO₂ into energy-rich compounds, utilizing the transferred electrons. Intra- or intermolecular electron transfer (ET) is a prerequisite process in photocatalytic CO₂ 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 CO₂ reduction process is challenging. The complete cycle of CO₂ reduction requires sacrificial electron donors (SEDs) and other additives that can aid CO₂ binding to the catalyst [5,6,7,8]. Furthermore, advanced photocatalytic systems often use metal-organic frameworks (MOFs) or semiconductors like TiO₂ 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 TiO₂/Re(I) (Re(I) catalyst anchored to TiO₂ particle) for photocatalytic CO₂ reduction. However, the collisional ET from ZnPs to TiO₂ 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 TiO₂ 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 TiO₂ 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 TiO₂ 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→ZnP_Acet and broken π-conjugation for ZnP_Acet→ZnP_Et. The fluorescence lifetimes of ZnP_Acet and ZnP_Et are almost identical, having a time constant of 2.9 and 3.0 ns, respectively (

Table 1. Quenching Kinetic Parameters of ZnPs by BIH.

	τ0 (ns) 1	KSV (M−1) 2	kq (109 M−1 s−1) 3
ZnP	2.4	4.44	1.9
ZnPAcet	2.9	6.87	2.4
ZnPEt	3.0	0.99	0.3

¹ Fluorescence lifetimes measured in the absence of BIH (λ_ex = 400 nm; λ_det = 645, 635, and 620 nm for ZnP, ZnP_Acet, ZnP_Et, respectively). ² Stern-Volmer quenching constant. ³ bimolecular quenching constant.

). 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 TiO₂, which serves as an electron reservoir for ET to the Re(I) catalyst in the binary hybrid system, when the OERS encounters TiO₂ 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]

$$\frac{{\tau }_0}{\tau }=1+K_{\mathrm{SV}}\left[Q\right]$$

(1)

where τ₀ and τ are the fluorescence lifetimes in the absence and presence of quencher, respectively. K_SV is the Stern-Volmer quenching constant and is given by K_SV = k_qτ₀ where k_q is the bimolecular quenching constant. According to Equation (1), the linear fits to the Stern-Volmer plot of the fluorescence lifetimes resulted in K_SV values, and corresponding k_q values were calculated from τ₀ (Figure 2c and Table 1). The calculated k_q values of ZnP and ZnP_Acet are 6 to 8-fold larger than that of ZnP_Et, 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/ZnP_Acet and −0.10 eV for ZnP_Et [9]. Unlike conjugated substituents in ZnP and ZnP_Acet, non-conjugated substituent, ethyl, in ZnP_Et does not lower the molecular orbital (MO) energy levels, keeping the reduction potential of photoexcited ZnP_Et 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 k_q values for ZnP and ZnP_Acet.

Interestingly, the k_q 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 (k_q [10⁹ 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 k_q values than those from the fluorescence lifetime measurements even for the same quenching reaction. Therefore, the discrepancy in ZnP’s k_q 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 k_q values between the two methods is not significant for ZnP_Acet. (k_q determined by the fluorescence intensity quenching is 3.0 × 10⁹ M⁻¹ s⁻¹) [9]. The electron-donating from the substituent is less in ZnP_Acet than in ZnP, which can reduce the complex formation between ZnP_Acet and BIH. For ZnP_Et, 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 TiO₂ as depicted in Figure 3 (see also Figure S2 in Supporting Information). In the absence of BIH and TiO₂, the TA spectra of ZnPs do not show dramatic changes within the time delays of 7.4 ns (Figure 3a). For ZnPs and ZnP_Acet, 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 ZnP_Et 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 ZnP_Acet 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 ZnP_Et, 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 TiO₂ into the sample solution in the presence of BIH enhances the decrease of bleach and absorption signals in the TA spectra of ZnP and ZnP_Acet, which implies the ET dynamics from ZnPs^•– to TiO₂ particle. In contrast, the TA spectra of ZnP_Et measured with the addition of BIH and TiO₂ 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. Kinetic Parameters from the Global Fits of the TA Spectra of ZnPs.

	Additives	kA→B	kB→C	kC→GS
ZnP		(2.2 ± 0.1 ps)−1	(2.3 ± 0.1 ns)−1	<<(100 ns)−1
	+BIH	(2.2 ± 0.2 ps)−1	(822 ± 21 ps)−1
	+BIH + TiO2	(2.8 ± 0.3 ps)−1	(614 ± 15 ps)−1
ZnPAcet		(0.7 ± 0.1 ps)−1	(3.4 ± 0.2 ns)−1
	+BIH	(1.1 ± 0.1 ps)−1	(737 ± 8 ps)−1
	+BIH + TiO2	(2.0 ± 0.1 ps)−1	(535 ± 5 ps)−1
ZnPEt		(1.6 ± 0.1 ps)−1	(2.9 ± 0.3 ns)−1
	+BIH	(1.1 ± 0.1 ps)−1	(2.0 ± 0.1 ns)−1
	+BIH + TiO2	(1.2 ± 0.1 ps)−1	(2.0 ± 0.1 ns)−1

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 (S₁) 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 S₁ state, evolves into the C state with τ_B→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 (T₁) 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 TiO₂, the EAS of A, B, and C states do not essentially change from those observed in ZnPs, still representing the S_n, S₁, and T₁ states, respectively: their τ_A→B values of 1.1~2.8 ps are the same as the IC process (S_n→S₁) in ZnPs, and τ_C→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 S₁ state lifetime, τ_B→C. Like in the fluorescence quenching experiments where the ET reaction with BIH shortens the S₁ state lifetime of ZnPs, τ_B→C decreases in the TA measurements with the addition of 0.3 M BIH: the fitted values of τ_B→C are 822 and 737 ps for ZnP and ZnP_Acet and 2.0 ns for ZnP_Et (Table 2). These τ_B→C values are comparable to the S₁ state lifetimes for ZnPs + 0.3 M BIH (τ_ZnPs+BIH), which are estimated to be 1.0, 0.9, and 2.4 ns for ZnP, ZnP_Acet, and ZnP_Et, respectively, by

$$\frac{1}{{\tau }_{\mathrm{ZnPs}+\mathrm{BIH}}}=\frac{1}{{\tau }_0}+0.3k_{\mathrm{q}}$$

(2)

(where τ₀ and k_q 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 ZnP_Acet, the τ_B→C value further decreases with the addition of both BIH and TiO₂ (Table 2), meaning that the decay of the S₁ state is additionally enhanced by the presence of TiO₂ in the solution. According to the reductive quenching mechanism, the electron transfer to TiO₂ should occur not directly from the photoexcited ZnPs in the S₁ state (ZnPs*) but from the OERS (ZnPs^•–) that is formed after the ET reaction with BIH. Indeed, the TA experiments with only TiO₂ 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 TiO₂ 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 TiO₂ will be incorporated in the decay of ZnPs* because quenching of ZnPs^•– by TiO₂ 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 TiO₂ is the only factor for the τ_B→C decrease, we can estimate the quenching rate of ZnPs^•– in the ET reaction with TiO₂ (k_q,OERS) as

$$k_{\mathrm{q},\mathrm{OERS}}=\frac{1}{{\tau }_{\mathrm{B}\to \mathrm{C},\mathrm{BIH}+{\mathrm{TiO}}_2}}-\frac{1}{{\tau }_{\mathrm{B}\to \mathrm{C},\mathrm{BIH}}}$$

(3)

where τ_B→C,BIH and τ_{B→C,BIH+TiO2} are the time constants for the B→C evolution measured in the presence of BIH and BIH + TiO₂, respectively (Table 2). The calculated k_q,OERS values of ZnP and ZnP_Acet are $4.12\times {10}^8$ and $5.12\times {10}^8$ s⁻¹, respectively. In ZnP_Et, the formation of OERS is so slow that its quenching by TiO₂ is not observed within the experimental time window of TA measurements, i.e., no change in τ_B→C with the addition of TiO₂. Compared to the collisional ET rates of ZnP^•– with TiO₂ previously determined by the UV/vis spectrum change ($5.12\times {10}^{-3}$ s⁻¹), [9] the k_q,OERS values of ZnP and ZnP_Acet 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 TiO₂, 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 10⁹~10¹⁰ 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 10⁴ s⁻¹ which is significantly slower than the k_q,OERS values obtained here. (Note that the particle size of TiO₂ 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 k_q,OERS values may report the ET dynamics of the OERS formed at the very vicinity of the TiO₂ 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 TiO₂ surface, the electron injection from ZnP^•– to TiO₂ 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. TiO₂ (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 (ZnP_Acet), and [5,15-diethyl-10,20-bis[2,6-di(dodecyloxy)phenyl]porphinato]zinc (ZnP_Et) 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.

4. Conclusions

In conclusion, we investigated the ET dynamics of ZnPs during the photosensitization reaction where ZnPs* react with BIH, forming ZnPs^•– that can transfer an excess electron to TiO₂. Here, the time-resolved emission and absorption spectroscopies uncovered more detailed dynamics and mechanisms in the ET reactions compared to the previously reported steady-state measurements. The Stern-Volmer analysis of the fluorescence lifetime quenching experiments revealed a large discrepancy in k_q of ZnP between the steady-state and time-resolved measurements, suggesting that both static and dynamic quenching processes exist in the reaction of ZnP and BIH. Most of all, the TA measurements unveiled the fast ET dynamics between the OERS and TiO₂, which occurs on a time scale of a few nanoseconds. This fast time scale is even faster than the diffusion rate in solution, suggesting that diffusion is not involved in this ET reaction. We propose that the ET from the OERS formed at the very vicinity of the TiO₂ surface is responsible for the fast time scale ET dynamics.

Designing more efficient photosensitizers for photocatalytic CO₂ reduction requires a thorough understanding of the ET mechanisms in the photosensitization reaction. This study demonstrates that the ET dynamics of photosensitizers can span multi-time scales and highlights the importance of time-resolved spectroscopies with multi-scale time resolution for elucidating the reaction mechanisms of photosensitization.