Utility of Squaraine Dyes for Dye-Sensitized Photocatalysis on Water or Carbon Dioxide Reduction.

Red light-sensitized squaraine (SQ) dyes were developed and incorporated into dye-sensitized catalysts (DSCs) with the formula of SQ/TiO2/Cat, and their efficacies were evaluated in terms of performance on either water or carbon dioxide reduction. Pt nanoparticles or fac-[Re(4,4'-bis-(diethoxyphosphorylmethyl)-2,2'-bipyridine)(CO)3Cl] were used as each catalytic center within the DSC frame of SQ/TiO2/Pt (Type I) or SQ/TiO2/Re(I) (Type II). In order to convey the potential utility of SQ in low energy sensitization, the following catalytic reductions were carried out under selective lower energy irradiation (>500 nm). Type I and II showed different catalytic performances, primarily due to the choice of solvent for each catalytic condition: hydrogenation was carried out in H2O, but CO2 reduction in dimethylformamide (DMF), and SQ was more stable in aqueous acid conditions for hydrogen generation than CO2 reduction in DMF. A suspension of Type I in 3 mL water containing 0.1 M ascorbic acid (pH = 2.66) resulted in efficient photocatalytic hydrogen evolution, producing 37 μmol of H2 for 4 h. However, in photocatalysis of Type II (SQ/TiO2/Re(I)) in 3 mL DMF containing 0.1 M 1,3-dimethyl-2-phenyl-1,3-dihydrobenzimidazole, the TiO2-bound SQ dyes were not capable of working as a low energy sensitizer because SQ was susceptible to dye decomposition in nucleophilic DMF conditions, resulting in DSC deactivation for the CO2 reduction. Even with the limitation of solvent, the DSC conditions for the utility of SQ have been established: the anchoring group effect of SQ with either phosphonic acid or carboxylic acid onto the TiO2 surface; energy alignment of SQ with the flat band potentials (E fb) of TiO2 semiconductors and the reduction power of electron donors; and the wavelength range of the light source used, particularly when >500 nm.

Introduction

Converting solar energy into chemical energy represents a promising strategy for replacing fossil fuels with renewable and sustainable energy.[1−5] The buildup of a photocatalytic conversion system for such solar fuel production requires the optimum organization of a visible light-harvesting photosensitizer and reduction catalyst that can achieve an efficient light-driven flow of electrons from an electron donor to the catalyst for solar-to-chemical conversion, typically hydrogen evolution[2,6−10] or CO2 reduction.[11−17] Among several strategies, multicomponent dye-sensitized photocatalysis (DSP) for light-driven hydrogen production[18−22] and CO2 reduction[23−31] offer several advantages over conventional homogeneous strategies: facile preparation and tunability of the catalytic system with a rational design for optimizing activity, efficient photoelectron collection and charge separation of photosensitizers by semiconductor immobilization, and on-demand multi-electron transportation toward the reduction catalyst through the accumulation of long-lived electrons in the conduction band (CB) of the semiconductor.[32,33]

Despite such advantages, DSP systems can still experience performance improvement with a photosensitizer and its rational interfacial engineering with an inorganic semiconductor and component substitution at the semiconductor surface with the catalyst. Regarding research into the photosensitizer, one of the primary issues has been the visible-light absorbing dye that can efficiently harvest far-red and even near-IR solar radiance to maximize overall photocatalytic activities.[34−38] The use of a long-wavelength absorber typically known for use in porphyrin derivatives with reduction catalysts has been studied for light-driven catalysis,[15,39−41] and recent efforts in improved dye-sensitized catalyst materials have continued to emerge,[42,43] but only a limited number of red-light photosensitizers have been reported. In this regard, the squaraine derivative can be considered a potential alternative red and near-IR sensitizer[44−46] because of its high extinction coefficients in the region of 600–800 nm and its easy synthetic tunability. However, until recently, squaraine dyes have been relatively rarely used in photocatalysis, although the applications of squaraine dyes are as extensive as other visible light active antenna in the field of dye-sensitized solar cells[36−38,47−51] and organic solar cells.[52−54] This limited utilization is thought to be derived from weaknesses such as the low structural stability of the central cyclobutene ring to nucleophilic attack, lack of directionality in the excited state with a fluorescent lifetime (∼1 ns), and formation of nonfluorescent aggregates, which limit the use of squaraine dyes in photocatalysis.

The above DSP platform can be considered to be one of the most prominent strategies with positive effects of squaraine immobilization on semiconductors. The rapid consumption of photogenerated anionic and/or oxidized species on the TiO2 surface is expected to effectively alleviate the photodecomposition of anchored squaraine dyes during photolysis, consequently improving its durability with an enhanced charge separation/collection yield.[27] On the other hand, the introduction of squaraine dyes, long-wavelength sensitizers, in the DSP system could also be a viable way to avoid such considerable loss of light harvesting arising from extensive light scattering by dispersions of the hybrid particles in solution, because the penetration of the red and near-IR light is generally superior to that of the blue and green light with a relatively high frequency wave.

Scheme describes the essential components of the present photocatalytic hybrid systems using squaraine dyes (SQ; X = COOH or PO3H2) and catalysts (Cat = Pt nanoparticles for hydrogen generation and Re(I) for CO2 to CO conversion). Herein, we present the details of two representative hybrid models, Type I (SQ/TiO2/Pt) for hydrogen generation and Type II (SQ/TiO2/Re(I)) for CO2 reduction, which were prepared through the covalent anchoring of SQ on two functional TiO2/Pt and TiO2/Re(I) catalytic particles. The photocatalytic properties of the resulting SQ-sensitized TiO2 hybrid were found to differ substantially from the anchoring ability of SQ, reaction solvent media, and the energy levels of the components relative to the TiO2 CB edge according to the systems and reaction conditions. These investigations offer a useful guideline for optimizing the activity and durability of squaraine-based photocatalytic systems.

Scheme 1

Conceptual Representation of the Reaction System of Type I and II and the Components (Electron Donor, Pt or Re(I) Catalyst, and Squaraine Dye) Used in This Work

Results and Discussion

Materials: Preparation of Squaraine Derivatives and Integration of SQ/TiO2/Cat (Cat = Pt or Re(I))

For successful low energy sensitization, the electron-withdrawing 1-cyanovinyl-phosphonic or -carboxylic acid anchoring group is attached to the core indolenine part in order to achieve a directional electron flow from squaraine to the CB of TiO2 and also to improve the light harvesting ability by extending the optical absorption wavelength toward the red region. Two different anchoring groups (1-cyanoacrylic acid and (1-cyanovinyl)phosphonic acid) were used in order to investigate any possible anchoring effects on the catalytic activities of the systems. The linear dodecyl group was introduced to the squaraine moiety to avoid π–π aggregation between the squaraine dyes on the TiO2 surface.[48]

The two donor–acceptor type squaraine derivatives (SQ and SQ) were prepared through the modification of a previously described synthetic method (Scheme ). The asymmetric bromo-bis(indoline) squaraine (SQ) was prepared under the reported reaction conditions.[48] A Pd-catalyzed C–C coupling reaction of SQ with 5-formylthiophen-2-yl boronic acid gave the aldehyde precursor (SQ) with high yield (∼70%), which was further treated with cyanoacetic acid to yield SQ and with diethyl cyanomethylphosphonate to yield SQ. Subsequent hydrolysis of the phosphonic ester with bromotrimethylsilane (TMSBr) produced the desired SQ in moderate yield (∼40%). All products were identified through 1H NMR spectroscopy and ESI-mass spectrometry. The detailed procedures and characterization of all squaraine derivatives are provided in the Experimental Section.

Scheme 2

Synthetic Route of Squaraine Derivatives (SQ, SQ, and SQ)

Conditions: (i) 5-formylthiophene-2-yl boronic acid, Pd(dppf)Cl2, K2CO3, and toluene/MeOH; (ii) piperidine, cyanoacetic acid, and CHCl3; (iii) piperidine, diethylcyanomethyl phosphonate, and CHCl3; (iv) TMSBr, CHCl3, and MeOH.

The Re(I) catalyst, fac-[Re(4,4′-bis-(dihydroxyphosphorylmethyl)-2,2′-bipyridine)(CO)3Cl], was prepared according to a previously published method.[26] The platinized TiO2 particles (TiO2/Pt) were prepared according to the previously published procedures using Degussa P25.[22] The successful anchoring of SQ and Re(I) on TiO2 nanoparticles was confirmed through diffuse-reflectance spectra (DRS) showing the absorption maximum of squaraine dyes at ∼680 nm (inset in Figure ) and via the IR absorption peaks indicating three characteristic CO peaks of Re(I) (2027, 1914, and 1906 cm–1) (see Figure S1 of the Supporting Information). The supernatant separated after centrifugation of the SQ- and/or Re(I)-treated suspensions show negligible absorption of SQ and Re(I), confirming the quantitative fixation of the components. In this study, ascorbic acid (AscH)[55] and 1,3-dimethyl-2-phenyl-1,3-dihydrobenzimidazole (BIH)[56] were used as the electron donors for hydrogen evolution in water and for CO2 reduction in dimethylformamide (DMF), respectively.

Figure 1

Visible absorption spectra (solid line) and photoluminescence spectra (dotted line) of 2 μM solution of SQ (black) and SQ (red) in DMF. The inset shows DRS of the SQ/TiO2 powders.

Photophysical Properties

Figure and Table both show the UV/vis/NIR absorption spectra of SQ and SQ in DMF, which exhibit a typical narrow absorption band centered at ∼670 nm with a high molar absorption coefficient of (1.1–1.8) × 105 M–1 cm–1. As has been reported by Shi et al.,[48] the absorption bands of SQ and SQ are red shifted as compared to the previously reported pristine squaraine dye (SQ1), because of the effective intramolecular charge transfer (ICT) and π-elongation effects of the electron-withdrawing cyanovinyl moiety. The absorption spectra of SQ immobilized on the TiO2 powder (SQ/TiO2) are slightly red-shifted and broadened. This behavior can be understood in terms of the formation of J-aggregates of the adsorbed SQ on the TiO2 surface and/or the interaction between the sensitizer and TiO2, with optical effects such as scattering and reflection of TiO2.[57] The zero–zero transition energies (E0–0) were calculated based on the crossing point of the normalized absorption and emission spectra (Figure ), which are listed in Table . Overall, the two asymmetrical squaraine derivatives exhibit broader and additional absorption in the longer wavelength region, which should be favorable for light harvesting. Upon the excitation of SQ at 664 nm, red emissions centered at 680 nm (±2 nm) appear to show a mirror image against the absorption spectra with small Stokes shifts (Δv = 329–375 cm–1), indicating the delocalization and small reorganization energy of the excited state (Table ).[44−46]

Table 1

Photophysical Properties of SQ in DMF

dye	λmax [ε × 104]a	λmax,DRSb	λemc	E0–0 [eV]d	Stokes shift [cm–1]
SQCA	667(18)	701	682	1.84	329
SQPA	664(11)	687	681	1.85	375
SQ1e	636(15)	651f	659	1.91	548

Extinction coefficient: ε [cm–1 M–1].

DRS of SQ/TiO2 particles.

Excitation at 600 nm.

The zero–zero excitation energies (E0–0) were determined from the crossing point of the normalized absorption and emission spectra: E0–0 (eV) = 1240/λ crosspoint.

Taken from ref (36).

UV–vis absorption of the SQ1/TiO2 porous film on the FTO substrate.

Electrochemical Properties

The primary event of dye photosensitization is the electron injection from SQ into TiO2, for which the electrochemical redox potentials of SQ were determined using cyclic voltammetry. Table presents the half-wave oxidation and reduction potentials of SQ and SQ (Eox/red), which are 0.36/–1.30 and 0.33/–1.18 V versus the saturated calomel electrode (SCE), respectively. It should be noted that the redox potentials of SQ, the ethyl ester of SQ, are used in place of those of SQ, because the latter revealed no reliable electrochemical behavior, presumably due to the chemisorption of SQ on the Pt working electrode. The two squaraine dyes show relatively lower reduction potentials than the pristine squaraine dye (SQ1),[36,48] which is attributed to the enhanced ICT and extended π-conjugation caused by incorporating the electron-accepting ability of the cyanoacrylic- or (1-cyanovinyl)phosphonic acid moiety. A similar trend can also be observed in the density functional theory (DFT) calculations for the squaraine dyes: SQ and SQ (−(2.61 to 2.81) eV) have relatively lower LUMOs than pristine SQ1 (−2.49 eV) (see Figure ). The highest occupied molecular orbital (HOMO) energy levels of the push–pull SQ are mainly inherited from the indolenine unit, whereas their lowest unoccupied molecular orbital (LUMO) energy levels are mostly populated on the electron-withdrawing anchoring moieties, reflecting their better directionality in the excited state relative to SQ1. Figure summarizes the oxidation and reduction potentials of components {SQ, SED (sacrificial electron donor), and catalytic unit [Pt and Re(I)]} in their ground/excited states with the CB of TiO2 varying according to the solvent environment. The regeneration of SQ by the reduction of SQ•+ with SED (AscH or BIH) is an essential process that should determine the TON and long-term stability of SQ. The free-energy changes of this process (ΔGreg) can be given by the difference between the oxidation potentials (Eox) of SQ and SED (0.46 V for AscH and 0.28 V for BIH vs SCE, which are sufficiently negative (ΔGreg = −(0.08 to −0.21) eV for AscH and −(0.26 to 0.39) eV for BIH) for electron transfer from SED to SQ•+ to regenerate SQ. The excited state oxidation potential (Eox*) of SQ is relevant to electron injection from excited-state SQ into TiO2, and is estimated to be from the excitation energy (E0–0) and Eox of SQ (E(SQ)ox* = −1.48 V) and E(SQ)ox* = −1.51 V). The Eox* values were compared with the apparent flat-band potential (Vfb) of the TiO2 nanoparticle film determined via Mott–Schottky analysis, which is −(0.5 to 1.1) V versus SCE in water with variation of pH (0–12)[58−62] or −(1.5 to 2.6) V versus SCE in DMF depending on water contents (0–20 vol %).[26,28,29] In aqueous media, the Eox* values are much substantially negative than Vfb of TiO2 (typically ∼−0.5 V at pH = ∼2), so the electron injection from 1SQ* into TiO2 must be sufficiently exergonic to efficiently proceed. In DMF, on the other hand, Eox* of SQ is slightly more positive by 0.26 or 0.23 eV than Vfb of TiO2 [∼−1.74 V vs SCE in the presence of 3% (v/v) H2O].[26] According to the estimation, the electron injection in DMF should be slow. However, in reality, the fluorescence of SQ anchored on a TiO2 mesoporous film was quenched within a time domain of 25 ps, which was much shorter than the fluorescence lifetimes in DMF (140 ps for SQ and 1.72 ns for SQ) (Figure S2 and Table S1 of the Supporting Information).

Table 2

Electrochemical Property of Squaraine Derivatives (SQ)a

dye	Eox [V]	Ered [V]	Eox* [V]b	Ered* [V]c	ΔGreg,AscH [eV]d	ΔGreg,BIH [eV]d
SQCA	0.36	–1.30	–1.48	0.54	–0.08	–0.26
SQPE	0.33	–1.18	–1.51	0.67	–0.21	–0.39

In DMF.

Oxidation potential in the excited singlet state; Eox* = Eox – E0–0.

Reduction potential in the excited singlet state; Ered* = Ered + E0–0.

The recovery efficiencies by the reduction of SQ•+ with AscH or BIH are quantitatively compared with the relative driving force for dye regeneration (ΔGreg), which is calculated from the excited-state reduction potentials of 1SQ* (Ered*) and the oxidation potential of AscH (Eox = 0.46 V vs SCE in water) or BIH (Eox = 0.28 V vs SCE in DMF): ΔGreg,AscH = Eox(AscH) – Ered*(SQ); ΔGreg,BIH = Eox(BIH) – Ered*(SQ).

Figure 3

HOMO–LUMO levels of squaraine dyes (SQ) obtained from DFT calculation and spatial distributions for the frontier orbitals of SQ.

Figure 2

Relative energy levels (V vs SCE) of the excited-state oxidation/reduction potentials (Eox*/Ered*) and oxidation/reduction potentials (Eox/Ered) of SQ, the reduction potential {Ered[Re(I)]} of the Re(I) catalyst, the oxidation potentials [Eox(ED)] of electron donors, and flat-band potentials (Vfb) of TiO2 in DMF or H2O under various conditions.

Photocatalytic Hydrogen Evolution of the SQ/TiO2/Pt Catalyst (Type I)

The hydrogen-evolution experiment was undertaken under red-light (>500 nm) irradiation of 10 mg Type I hybrid particles dispersed in 3 mL water in the presence of AscH, ethylenediaminetetraacetic acid (EDTA), or triethanolamine (TEOA) as the sacrificial electron donor. In all the cases of photolysis, a concentration of 3.3 g TiO2/L (10 mg TiO2 nanoparticles in 3 mL reaction solvent) was used as the optimum condition of light absorption of the irradiated heterogeneous system.[63]Table summarizes the hydrogen product observed under various conditions, including the net formation amounts in μmol, turnover number (TONNP) in mol H2 (g TiO2)−1, and turnover frequency (TOFNP) in μmol H2 h–1 (g TiO2)−1. First, the effects of SQ-anchoring amounts on TiO2/Pt nanoparticles were investigated in order to identify the optimal conditions. In Type I hybrid, the photocatalytic H2 evolution increased with increasing loading amounts of SQ from 0.2 to 4.8 μmol to reach a maximum at 1.6 μmol, while no further increase of H2 was observed at higher SQ loading amounts (see entries 1–10 of Table and Figure b). The change in the anchoring group from −COOH to −PO3H2 significantly enhanced the photocatalytic activity (see TONNP and TOFNP in Table and Figure ).

Table 3

Photocatalytic Hydrogen Production with Type I Hybrid under Different Conditionsa

entry	system	tirr./h	H2/μmol	TONSQb	TONNPc	TOFNPc
1	SQCA(0.2)/TiO2/Pt	2	8.20	41.0	820	410
2	SQCA(0.4)/TiO2/Pt	2	16.0	40.0	1600	800
3	SQCA(1.6)/TiO2/Pt	2	29.3	18.3	2930	1465
4	SQCA(3.2)/TiO2/Pt	2	26.6	8.3	2660	1330
5	SQCA(4.8)/TiO2/Pt	2	29.8	6.2	2980	1490
6	SQPA(0.2)/TiO2/Pt	2	20.9	104.5	2090	1045
7	SQPA(0.4)/TiO2/Pt	2	28.3	70.8	2830	1415
8	SQPA(1.6)/TiO2/Pt	2	41.6	26.0	4160	2080
9	SQPA(3.2)/TiO2/Pt	2	41.0	12.8	4100	2050
10	SQPA(4.8)/TiO2/Pt	2	42.3	8.8	4230	2115
11	TiO2/Pt	3	d
12	SQCA(1.6)/TiO2	3	d
13	SQCA(1.6)/TiO2/Pte	3, w/o light	d
14	SQCA(1.6)/TiO2/Ptf	3, w/o BIH	d
15	SQCA(1.6)/s-TiO2/Pt	2	3.12	1.9	312	156
16	SQCA(1.6)/d-TiO2/Pt	2	8.37	5.2	837	419
17	SQPA(1.6)/s-TiO2/Pt	2	5.86	3.7	586	293
18	SQPA(1.6)/d-TiO2/Pt	2	5.04	3.2	504	252

The following standard conditions were employed: 10 mg of TiO2/Pt was commonly used as the sacrificial electron donor and immobilizer of the functional components (SQ and SQ); the hybrids were modified with various amounts of SQ (0.2–4.8 μmol). The total volume of the solution or suspension was 3 mL. All of the photocatalytic reactions were performed in distilled water containing 0.1 M AscH at greater than 500 nm emitted from an LED lamp (60 W, Cree Inc.).

TON indicates the turnover number in micromols H2 per micromol of SQ.

TONNP and TOFNP indicate the turnover number in micromol H2 (g TiO2)−1 and turnover frequency in micromol H2 h–1 (g TiO2)−1, respectively.

Not detected.

Under dark conditions.

Without electron donor (AscH).

Figure 4

Dependences of photocatalyzed H2 formation on SED with SQ(1.6 μmol)/TiO2/Pt (a) and on loading amounts of SQ with Type I hybrid after 4 h irradiation (b) in N2-saturated water containing 0.1 M AscH at ≥500 nm.

In order to verify the effects of the different TiO2 particles, Degussa P-25 (d-TiO2, ca. 25% rutile and 75% anatase) and homemade TiO2 nanosheets (s-TiO2, [001] facet-exposed TiO2)[64] were applied to this hybrid system. The results indicated that the d- and s-TiO2-based hybrids showed significantly poorer photocatalytic activities than Type I hybrid based on h-TiO2 (see entries 15–18 in Table ).

The results of the control experiments indicate that the co-loading of both SQ and Pt on TiO2 is essential for the efficient production of hydrogen, since SQ/TiO2 or TiO2/Pt without the immobilization of either SQ or Pt reveals almost no hydrogen evolution activities (see entries 11 and 12 in Table ). In the dark or in the absence of AscH, H2 evolution was either not detected at all or only occurred in a trace amount. In addition, very poor H2 production was observed when EDTA and TEOA were used as SED in place of AscH (Figures a and 5). These results can be understood in terms of the different reducing powers of the SEDs used [E(SED)ox = 0.57–0.82 V for TEOA, 0.57–0.92 V for EDTA, and 0.46 V for AscH vs SCE].

Figure 5

Effects of SEDs on photocatalytic H2 production under 2 h irradiation at >500 nm; 10 mg of SQ (1.6 μmol)/TiO2/Pt and 0.1 M AscH in 3 mL N2-saturated water.

The apparent quantum yield (AQY) of photocatalyzed H2 formation at 436 nm for the Type I hybrid catalyst (SQ/h-TiO2/Pt) in water containing 0.1 M AscH was determined to be (1.15 ± 0.07)%. However, note that the AQY determined is not the real quantum yield with scientific definition but reveals a relative measure for the estimation of hydrogen production efficiency, because the number of photons absorbed by the TiO2-bound SQ dye cannot be exactly estimated in the present heterogeneous system without optical effects (light scattering) of mesoporous TiO2 particles (of the TiO2 particle dispersion system). The poor AQY can partly be interpreted in terms of optical effects (extensive light scattering) of mesoporous TiO2 particles.

Photocatalytic Behavior of SQ/TiO2/Re(I) (Type II)

Previously, we have reported that the reduction of CO2 to CO can be effectively photosensitized by a blue-green active dye anchored on TiO2/Re(I) particles in DMF; Re(I) is CO2 to CO conversion which is fixed on TiO2 through PO3H2 anchoring. The successful red-light photosensitization of H2 evolution by SQ as described above led us to investigate the applicability of SQ photosensitization to CO2 to CO conversion based on TiO2/Re(I) particles in DMF. First, SQ/TiO2/Re(I) catalysts (Type II) were prepared by anchoring SQ on TiO2/Re(I) particles. Next, dispersions of 10 mg of Type II hybrid in 3 mL of CO2-saturated DMF in the presence of BIH as SED were irradiated with a white-light light-emitting diode (LED) through an optical filter cutting off the light shorter than 400 nm or than 500 nm. Figure a shows temporal plots of CO formation under >400 nm irradiation. The CO2 to CO reduction activity [TONCO = moles of CO formed per mol of Re(I)] of Type II using SQ was found to be kept for an extended period of time up to 70 h, reaching ∼165 with only slight leveling-off tendency, while the sensitization with SQ showed higher efficiencies by 23 h but also a levelling-off tendency under longer-time irradiation. Note that an induction period was commonly observed around ∼2 h in both cases. Figure b exhibits plots of CO formation in the initial irradiation time of ≤10 h for Type II with SQ under irradiation at >400 nm. The induction period was observed within 2 h under >400 nm irradiation and by a much longer period of time (∼7 h) under >500 nm irradiation (see Figure ). The gaseous products formed in the headspace of the reactor were then analyzed by GC. High-performance liquid chromatography analysis of the liquid-phase contents was performed to detect such liquid products as formic acid and oxalic acid. The results confirm that CO is the exclusive product with a trace amount of H2, while neither formate nor oxalic acid was detected in any amount. An isotope-tracing experiment for SQ/TiO2/Re(I) dispersions (in 13CO2-saturated DMF-d7) was performed by GC-mass. The 13C isotope abundance analysis of CO (>95%) indicates that CO2 is the source of CO produced (see Figure S3 of the Supporting Information).

Figure 6

Plots of CO formation versus time for SQ/TiO2/Re(I) (-■-) and SQ/TiO2/Re(I) (-●-) during (a) prolonged irradiation time (<70 h) and (b) shorter irradiation time (<10 h) in the presence of 2.5 vol % water; irradiation at >400 nm.

Figure 7

Plots of CO formation versus time for SQ/TiO2/Re(I) under irradiation at λ >400 nm (-■-) and at λ >500 nm (-●-).

However, in contrast to Type I in aqueous media, the initial intense blue SQ/TiO2/Re(I) hybrid particles (Type II) have become pale yellow after photolysis (see the inset of Figure ). As shown in Figure , the intensity of the long-wavelength absorption band at ∼680 nm is rapidly diminished with increasing photolysis time (0–10 h). Further, the observation of almost no catalytic activity of Type II hybrid under low energy irradiation (λ >500 nm) reflects such full loss of long-wavelength absorption. In turn, this result confirms that the SQ dyes are not capable of functioning as a low energy sensitizer during photolysis of Type II hybrid for CO2 reduction (see Figure and entry 5 of Table ). Note that the initial absorption intensities of SQ around 400 nm are maintained under continuous LED irradiation irrespective of the full loss of low energy absorption (mainly around 500 and 680 nm) (Figure ). This implies that undefined organic species (formed with photobleaching of SQ) can continuously play a role as visible-light harvesters due to the survived absorption capability at the ∼400 nm region. This result also has important implications for the steady CO production observed in the above SQ-based photocatalysis under >400 nm (Figure ). However, at this point, the exact structure of photomodified species is unclear, as its characterization is still underway.

Figure 8

Relative UV–vis absorbance and photographic images (inset) of the reaction solution filtered after photoreaction (0 to 10 h). These results indicate the magnitude of photobleaching of SQ. Photograph courtesy of M. J. Copyright 2018.

Table 4

Results of Photocatalytic CO2 Reduction with SQ/TiO2/Re(I) (Type II) under Different Conditionsa

					CO
entry	systems	irradiation time/h	water/vol %	wavelength/nm	TON	μmol
1	SQCA/TiO2/Re(I)	70	2.5	>400	112 ± 15	11.2 ± 1.5
2	SQPA/TiO2/Re(I)	70	2.5	>400	165 ± 18	16.5 ± 1.8
3	SQCA/ZrO2/Re(I)	7.5	2.5	>400	4.0	0.4
4	SQCA/TiO2/Re(I)	7.5	2.5	>400	27	2.7
5	SQCA/TiO2/Re(I)	10	2.5	>500	1.7	0.17
6	SQCA/TiO2/Re(I)	10	2.5	>400	31.2	3.12

The following standard conditions were employed: 0.1 M BIH and 10 mg of TiO2 are commonly used as the sacrificial electron donor and immobilizer of 0.1 μmol Re(I) catalyst and 0.8 μmol SQ, respectively. The total volume of the suspension was 3 mL. All photocatalytic reactions are performed in CO2-saturated DMF irradiated at >400 or 500 nm using an LED lamp (60 W, Cree Inc.).

Discussion

According to the experimental data and interpretation presented above, the overall photocatalytic cycle of Type I hybrid would follow the proposed reaction sequences (Scheme ). Upon low energy sensitization, ultrafast electron injection from 1SQ* to the TiO2 CB via oxidative quenching commonly occurs with very fast trapping of the injected electrons to give trapped electrons [TiO2(e–)] (process 1). The injected electrons in the bulk phase are generally long-lived, with most of them staying in the CB of TiO2.[32,33] Subsequently, TiO2 shuttles the electrons to the anchored reduction catalytic sites (Pt) simultaneously (process 2). At the Pt sites in aqueous media, the reduction of proton (H+) occurs, resulting in net hydrogen production (process 4). The overall photocatalytic sequence of hydrogen evolution is completed with the successive regeneration of SQ by the reduction of SQ•+ with AscH (process 3).

Scheme 3

Proposed Mechanism for Hydrogen Production in Heterogeneous Photocatalysis by SQ/TiO2/Pt (Type I)

In the above water-splitting hydrogen production of Type I hybrid, the observed anchoring group dependency (Table ) can be interpreted in terms of the higher binding affinity of phosphonic acid (compared to the carboxylic acid) which leads to a suppressed photobleaching phenomenon during photolysis, as observed in our previous work.[27] This hypothesis is supported by DRS comparison of Type I hybrid measured prior to and after photolysis: the DRS absorption spectra of SQ fixed on TiO2 maintains the original absorption feature at the ∼680 nm region after photolysis for 5 h, whereas a discernible reduction was observed for SQ, indicating some probability of photobleaching behavior of SQ detached under continuous irradiation (Figure S4 of the Supporting Information). Such anchoring group dependency is consistent with that observed in the photocatalytic CO2 reduction of Type II hybrid. The contrasting photoactivities with variation of TiO2 semiconductor are consistent with our previous explanation that each TiO2 source has different distributions of surface states/trap sites and grain boundaries associated with electron injection efficiencies from SQ* to the TiO2 CB, charge recombination between SQ•+ and trapped TiO2 CB electrons [TiO2(e–)], and the catalytic properties.[27] Regarding the electron donor dependency of catalytic activity observed above, apart from different reducing power of electron donors used, the significant change of its pH value with variation of SED should also be considered as another key factor affecting the overall activity because an acidic condition (H+) in aqueous media can help accelerate the H2 production in photocatalytic reaction. Measurements of the pH values for three SEDs were made as following; each 0.1 M AscH, EDTA, and TEOA in aqueous media resulted in a pH value of 2.66, 2.91, and 10.86, respectively. However, the observed SED dependency should be understood as a result of the different reducing powers of the SEDs rather than their pH difference because two similar acidic conditions using each AscH (pH = 2.66) and EDTA (pH = 2.91) as SED have dissimilar photocatalytic activities.

In an effort to elucidate the cause of low energy sensitization failure of SQ in Type II hybrid, time-resolved photoluminescence measurements for SQ or SQ/TiO2 (dyes anchored on TiO2, injecting case) were carried out to confirm whether the photoexcited electrons are injected to TiO2 particles, where SQ and SQ were fixed on TiO2-coated sapphire glass windows. In this study, homogeneous SQ in DMF was used as reference for the noninjection case to TiO2. Figure S2 shows the TRF profiles of the samples taken by 640 nm excitation. Interestingly, for the TiO2-free solutions, no fast decay component shorter than 100 ps was observed in their TRFs, whereas for the films, the lifetimes were dramatically decreased to about 20 and 26 ps for SQ/TiO2 and SQ/TiO2 films, respectively (see Table S1 of the Supporting Information), indicating that the fast electron injection process indeed occurs at the interface between SQ and TiO2 nanoparticles. This result is the exact opposite of what is observed on experiment: the uphill process (positive free Gibbs energy) (ΔG = −(0.02 to 0.22) V, from SQ* to TiO2) and the almost no photocatalytic activity under >500 nm irradiation. However, an electron injection can occur with not only the strong electronic coupling between the SQ LUMO orbital and TiO2 CB, but also the existence of various shallow surface trapping sites/bands that are 0.3–1.0 eV below the TiO2 CB, as has been reported in several dye/TiO2 systems.[65−69] At this point, it is not clear what factor drives the low photocatalytic activity in the DMF-mediated photolysis for CO2 reduction. We now think that there might exist other possible quenching pathways of photoexcited electrons via an undefined side reaction including photodecomposition of TiO2-bound SQ.

In addition, a probable formation of SQ/BIH+ (oxidized SED) ion-pair complexes in aprotic solvent media can be considered to be another reason of fast photobleaching, given that the SQ/cationic salt combination tends to undergo a photobleaching of SQ with the generation of active radical species.[70] This result differs from our previous ternary systems based on other photosensitizers (i.e., typical push–pull organic dye[26,28−30] and porphyrin[27]). In addition to the solvent dependency on photostability of SQ discussed above, we now think such low photostability of SQ may be partially derived from the intrinsic fragility of the squaraine structure. By contrast, the relatively less photobleaching of SQ in water media can be considered the result of the protection of the electron deficient cyclobutene ring of SQ by hydrogen bond formation between H2O molecules and the oxygen atoms of the cyclobutene ring. During photoreaction of Type I hybrid, the cyclobutene ring of SQ can be readily attacked by nucleophilic species such as ascorbate (deprotonated AscH) or undefined nucleophiles.[71]

Regarding the reduction process via anchored molecular Re(I) complexes in Type II hybrid, regardless of the kind of structure the real photosensitizing unit has during photolysis, the observed photocatalytic CO2 to CO conversion should follow the monomeric mechanism (suggested by our previous Re(I)-immobilized TiO2 system[26]) rather than the formation of the CO2-bridged dimer complex (LReCOOReL)[72−74] because the Re(I) complex was almost individually immobilized on the TiO2 surface (Scheme ).

Scheme 4

Possible Chemical Processes for Two-Electron Reduction of CO2 to CO Catalyzed by the TiO2-Bound Re(I) Complex in Photocatalysis of Type II Hybrid

Overall, based on the above photolysis results and the results of the photobleaching experiment of SQ, the photocatalytic efficiency of the SQ-based photocatalytic system seems to be associated more with the degree of photostability of SQ in different reaction solvents than the energy level-related thermodynamic electron injection issue.

Conclusions

In summary, we have synthesized two asymmetric squaraine dyes marked by linear dodecyl spacers and anchoring groups with π-conjugated electron-withdrawing cyano (−CN) moieties. The prepared push–pull squaraine dyes (SQ and SQ) have been successfully engaged via chemical anchoring to the TiO2/catalyst hybrid platform for photocatalytic hydrogen generation and CO2 reduction. Under lower energy irradiation (>500 nm), the SQ/TiO2/Pt catalyst (Type I) in AscH buffered (0.10 M) water showed an efficient photocatalytic hydrogen production with secured optimal electron flow from SQ to the Pt site coupled with the alleviated photobleaching of SQ, providing ∼42 μmol H2 for 2 h. However, in DMF-mediated photocatalysis of SQs/TiO2/Re(I) (Type II) for CO2 reduction, the low energy sensitization by SQ was not successful for the following reasons; (1) the significant photodegradation of SQ and (2) its energy-level mismatch (uphill process from SQ to TiO2). On the other hand, under high energy irradiation (>400 nm) conditions, the photodegraded SQ was found to function as a real photosensitizer after the induction period, giving a TONCO of ∼165 for 70 h. Although the final structure of the photodegraded SQ species is yet to be disclosed, we have gained some valuable information on how to design the optimum dye and adjust the reaction conditions within a photocatalytic system using a photodegradable organic antenna. We also found that several conditions should be considered and tuned for the long-wavelength photosensitization of the squaraine antenna including solvent-sensitive Vfb of TiO2, the chemical binding ability of the organic sensitizer on the TiO2 surface, environmentally sensitive photostability of organic photosensitizers, and reducing power of electron donors. The optimization of this hybrid system with further structural modification of the squaraine dye combining new alternate electron donor/acceptor building units is currently under investigation. We believe that the development of a highly active photocatalytic reduction system with excellent durability is possible through the use of this molecular engineering.

Experimental Section

Synthesis and Characterization

Two squaraine dyes (SQ and SQ) were synthesized following the synthetic procedure (see Schemes and S1 of the Supporting Information).[27,48]fac-[Re(4,4′-bis-(diethoxyphosphorylmethyl)-2,2′-bipyridine)(CO)3Cl] (Re(I))[26] and the sacrificial electron donor (BIH)[56] were prepared according to the method reported in the previous research.

SQPA

SQ (0.390 g, 0.554 mmol) and diethylcyanomethyl phosphonate (0.198 mL, 1.11 mmol) were added to a dry round-bottomed flask, which was then evacuated and filled with nitrogen three times. Next, anhydrous chloroform (30 mL) was added to dissolve the reactants. After the addition of piperidine (60 μL), the reaction mixture was then heated up to reflux for 8 h. Finally, the solvent was removed, and the residue was passed through silica gel chromatography [dichloromethane/methanol (10:1 v/v)] in order to obtain SQ as a blue solid (90 mg, 19%). The purified SQ (90 mg, 0.104 mmol) was added to a dry round-bottomed flask, which was then evacuated and filled with nitrogen three times. Dry CHCl3 (10 mL) was added to dissolve the reactants. Bromotrimethylsilane (TMSBr) (1.68 mL, 12.8 mmol) was added dropwise to the reactor containing SQ. The reaction mixture was refluxed for 12 h. To this solution methanol (10 mL) was added and stirred for 3 h. The solvent was then removed, and the residue was passed through silica gel chromatography (dichloromethane/methanol (5:1 v/v)) in order to obtain SQ as a blue solid (33.5 mg, 40%). 1H NMR (400 MHz, CDCl3): δ 8.04 (s, 1H; =CH), 7.90 (m, 1H; Ar–H), 7.68 (m, 1H; Ar–H), 7.63 (m, 1H; Ar–H), 7.56 (m, 1H; Ar–H), 7.52 (m, 1H; Ar–H), 7.36 (m, 3H; Ar–H), 7.19 (m, 1H; 3Ar–H), 5.85 (s, 1H; =CH), 5.82 (s, 1H; =CH), 4.77 (m, 2H; N–CH2), 4.13 (m, 4H; N–CH2), 1.95 (m, 12H; 4CH3), 1.75–1.14 (m, 23H; CH3 + 10CH2), 0.82 (s, 3H; CH3) (PO3H2 resonance not observed). ESI-MS (m/z): calcd for C47H56N3O5PS [M], 805.3678; found [M – H]−, 804.4196.

Preparation of Type I/II Hybrid Catalyst

Two commercially available TiO2 particles [Degussa P25 (d-TiO2) and Hombikat UV-100 (h-TiO2)] and one homemade TiO2 nanosheet (s-TiO2) were thoroughly washed with distilled water, ultrasonically treated in water, separated by centrifugation, and then dried in an oven under N2. Preparation of TiO2/Pt particles: surface-platinized TiO2 particles (TiO2/Pt, 0.3–1 wt % Pt loading) were prepared by a photodeposition method as described elsewhere.[75] The particles were collected by filtration or by centrifugation and then dried in an oven under N2. A water suspension of the TiO2 particles (0.5 g L–1) containing ∼1 M methanol and 0.03–0.1 mM chloroplatinic acid was irradiated with a 200 W mercury lamp for 30 min. After irradiation, the Pt-loaded TiO2 powders were separated by centrifugation, washed with distilled water, dried, and stored under N2 in the dark. The 0.1 g TiO2 or 0.1 g platinized TiO2 (TiO2/Pt) particles dispersed in an MC/MeOH solution of SQ or SQ (2–48 μmol) were allowed to stand overnight under stirring and then subjected to centrifugation. The collected particles were washed with the solvent and then dried in an oven under N2. For Type II hybrid, Re(I) complex was further chemiadsorbed to the SQ-deposited TiO2 powders for the preparation of SQ/TiO2/Re(I) (0.1 μmol Re(I) complex loading per 10 mg SQ/TiO2 particles).

Photocatalytic Hydrogen Production

Suspensions of SQ/TiO2/Pt particles (10 mg with 0.2–4.8 μmol dye) in 3 mL water containing 0.1 M AscH (pH = 2.66) were placed in a Pyrex cell (1 cm pass length; 6.0 mL total volume), bubbled with CO2 for 30 min, and sealed with a septum. A series of samples were placed on a homemade merry-go-round apparatus and then irradiated under magnetic stirring with an LED lamp (λ >500 nm, 60 W, model Fc-6051, Cree Inc.). The amounts of H2 evolved in the overhead space of the cell were determined through gas chromatography (HP6890A GC equipped with a TCD detector) using a SUPELCO Carboxen 1010 PLOT Fused Silica Capillary column.

Photocatalytic CO2 to CO Conversion

Suspensions of SQ/TiO2/Re(I) particles (10 mg with 0.8 μmol dye and 0.1 μmol Re(I) complex) in 3 mL DMF/water mixture solvent (water 2.5 vol %) containing 0.1 M BIH were placed in a Pyrex cell (1 cm pass length; 6.0 mL total volume), bubbled with CO2 for 30 min, then sealed with a septum. A series of samples were placed on a homemade merry-go-round apparatus and then irradiated under magnetic stirring with an LED lamp (λ >400 or 500 nm, 60 W, model Fc-6051, Cree Inc.). A water solution filter containing 30 wt % K2CrO4 (1 cm path length of light) was placed in front of an LED lamp for selective red light irradiation (>500 nm), 500 nm long-pass. The amounts of gas product evolved in the overhead space of the cell were determined using gas chromatography.

AQY Measurement

The AQY Φ(H2) for photocatalytic hydrogen production was determined for SQ/TiO2/Pt suspensions; a band-pass filter (420–450 nm) was used to isolate the 436 nm light from the emission light of a high-pressure mercury lamp (1000 W, model 6171, Newport Corporation), and the incident light flux was determined using a 0.2 M ferrioxalate actinometer solution.[76] The AQY of H2 formation for the hybrid system in the presence of 0.1 M AscH was determined in a linear time-conversion region.

Picosecond Time-Resolved Fluorescence Measurements

Picosecond time-resolved fluorescence measurements were made using a commercial picosecond fluorescence lifetime measurement system (Hamamatsu streak camera, C11200). The light source was a commercial optical parametric amplifier (TOPAS-prime, Light-Conversion) seeded with a commercial regenerative amplifier system (Spitfire-Ace, Spectra-Physics) operating at 1 kHz. The center wavelength and pulse energy were adjusted to 400 nm and about 1 μJ, respectively. The output was spectrally filtered by using a pair of prisms. A singlet lens was used to focus the excitation beam to the sample, and the photoluminescence was collected in a back-scattering geometry using a parabolic mirror. The emission was sent to a monochromator and detected with the streak camera. Magic angle detection was used to avoid the effect of polarization. Widths (fwhm) of the instrumental response function were about 180 ps in the 10 ns time window. All the data were acquired in a single photon counting mode using Hamamatsu U8167 software.