Jing
Huang
a,
Mélina Gilbert
Gatty
a,
Bo
Xu
a,
Palas Baran
Pati
a,
Ahmed S.
Etman
b,
Lei
Tian
a,
Junliang
Sun
b,
Leif
Hammarström
a and
Haining
Tian
*a
aDepartment of Chemistry-Ångström Laboratory, Uppsala University, Box 523, SE 751 20, Uppsala, Sweden. E-mail: haining.tian@kemi.uu.se
bDepartment of Materials and Environmental Chemistry (MMK), Stockholm University, SE 106 91 Stockholm, Sweden
First published on 10th July 2018
Covalently linking photosensitizers and catalysts in an inorganic–organic hybrid photocatalytic system is beneficial for efficient electron transfer between these components. However, general and straightforward methods to covalently attach molecular catalysts on the surface of inorganic semiconductors are rare. In this work, a classic rhenium bipyridine complex (Re catalyst) has been successfully covalently linked to the low toxicity CuInS2 quantum dots (QDs) by click reaction for photocatalytic CO2 reduction. Covalent bonding between the CuInS2 QDs and the Re catalyst in the QD–Re hybrid system is confirmed by UV-visible absorption spectroscopy, Fourier-transform infrared spectroscopy and energy-dispersive X-ray measurements. Time-correlated single photon counting and ultrafast time-resolved infrared spectroscopy provide evidence for rapid photo-induced electron transfer from the QDs to the Re catalyst. Upon photo-excitation of the QDs, the singly reduced Re catalyst is formed within 300 fs. Notably, the amount of reduced Re in the linked hybrid system is more than that in a sample where the QDs and the Re catalyst are simply mixed, suggesting that the covalent linkage between the CuInS2 QDs and the Re catalyst indeed facilitates electron transfer from the QDs to the Re catalyst. Such an ultrafast electron transfer in the covalently linked CuInS2 QD–Re hybrid system leads to enhanced photocatalytic activity for CO2 reduction, as compared to the conventional mixture of the QDs and the Re catalyst.
Colloidal quantum dots (QDs), a kind of semiconductor nanoparticle with size smaller than their exciton Bohr radius, can be a suitable alternative to molecular photosensitizers for building an inorganic–organic hybrid system for photocatalysis. Firstly, because of the quantum confinement effect, their bandgap can be easily tuned by their size, to optimize their absorption of solar light.33,34 Also, their band energy levels can be adjusted in a facile manner by varying the QD size,33,34 composition,35,36 and the capping ligand,37,38 to increase the driving force for electron transfer from QDs to catalysts. Secondly, due to the large surface-to-volume ratio of QDs, molecular catalysts could adsorb on the QD surface, and the charge transfer reaction becomes much faster than in a case with freely diffusing catalysts, in which the charge transfer is primarily mediated by bimolecular collision.39,40 Weiss et al. have recently employed CuInS2 QDs as photosensitizers for FeTPP41 and FeTMA catalysts,42 and achieved impressive photocatalytic activity for these hybrid systems, due to ultrafast electron transfer from QDs to catalysts. Thirdly, in addition to being photosensitizers, semiconductor nanoparticles can also work as electron reservoirs for the catalyst, which can slow down charge recombination in the system and prolong the lifetime of a catalyst intermediate.43 Last but not least, semiconductor nanoparticles can function as scaffolds for the catalysts, hindering the formation of catalyst dimers.44
To achieve efficient photo-induced catalysis, strong electronic coupling between QDs and catalysts is necessary to enable fast electron transfer from QDs to catalysts, and in that way, the catalyst molecules could be fed with more electrons to accomplish CO2 reduction. Numerous studies have shown that, compared to unbound photosensitizers and catalysts, hybrid systems consisting of covalently-bound sensitizers and catalysts always show stronger electronic coupling and more efficient electron transfer, resulting in higher activity and selectivity for CO2 reduction.45–47 Moreover, when molecular catalysts are linked with QDs to form a hybrid superstructure, the immobilization process on substrates is also simplified. On the other hand, without covalent bonds, the original ligand on the QDs can act as a barrier that inhibits the adsorption of catalysts on the QD surface, hence, hindering electron transfer between them.45 To date, only a few methods of covalent attachment of catalysts onto the QD surface have been reported, and most of them are based on molecular catalysts with an anchoring group, such as thiol groups45 or phosphonate groups,47,48 which can bind the catalyst molecules as ligands to the QD surface. However, the synthesis and purification processes of such catalysts with thiol or phosphate groups are often arduous. Hence, a synthetically simpler and more tunable approach for immobilizing catalysts on the QD surface is highly desirable.
The “click” reaction with azide–alkyne cycloaddition is an ideal reaction for covalent linkage of two molecules owing to its high selectivity, high yield and fast reaction kinetics under mild reaction conditions.49–52 It is also suitable for covalently linking organic molecules to conductive surfaces.53,54 In this work, we covalently linked the catalyst Re(4-azidomethyl-4′-methyl-2,2′-bipyridine)(CO)3Cl (Re catalyst) with CuInS2 QDs by a copper-free click reaction, to synthesize the QD–Re hybrid system for CO2 photoreduction. This is the first time that a molecular catalyst was attached on QDs with covalent bonds through a copper-free click reaction for photocatalytic reactions. Each reaction component was easy to prepare, which can be a demonstration for a straightforward and versatile strategy to link molecular catalysts with inorganic semiconductor nanoparticles. We also provide evidence that the photocatalytic activity of these QD–Re hybrid systems was enhanced by efficient electron transfer between the QDs and the Re catalyst due to the covalent linkage.
Then the original ligands on QDs were exchanged with 3-MPA through a ligand exchange reaction. Specifically, a MPA ligand solution, prepared by mixing 0.3 mL of MPA with 0.7 mL of methanol, with the pH of the solution adjusted to 10 using 38% NaOH aqueous solution, was dropped into 5 mL of CuInS2 QD stock solution under strong stirring. After stirring for 40 minutes at room temperature, MPA capped QDs, which were in the lower part, were collected and purified with water and acetone, and finally dispersed in 10 mL of DMSO.
The Re catalyst was synthesized by following a reported method.31
Fig. 1 Synthetic routes of the covalently-linked CuInS2 QD–Re hybrid system. Insert: Photograph of CuInS2 QDs and the CuInS2 QD–Re hybrid system dispersed in DMSO. |
Fig. 2 shows the UV-Vis ground-state absorption spectra of the CuInS2 QDs, the CuInS2 QD–Re hybrid system, and the Re catalysts. The as-synthesized CuInS2 QDs exhibited a broad excitonic peak centered at around 540 nm with a long tail extended to 800 nm (detailed in Fig. S3†), which illustrates the fact that the CuInS2 QDs can absorb the entire range of visible light up to near infrared light. The absorption of the Re catalyst began from 500 nm, and featured a maximum absorption at around 375 nm. Upon covalent binding of the Re catalyst onto the surface of CuInS2 QDs, the CuInS2 QD–Re hybrid system displayed absorption features both from the CuInS2 QDs and the Re catalyst. The difference spectrum (the absorption spectrum of CuInS2 QD–Re subtracted from that of CuInS2 QDs, green dotted line in Fig. 2) resembled the absorption spectrum of the pure linker-Re catalyst, and the latter was prepared similarly to the CuInS2 QD–Re hybrid system, but without CuInS2 QDs in the reaction system (see in Fig. S4†). Both of them showed a slight bathochromic shift from that of the pure Re catalyst, probably due to the weaker electron withdrawing ability of the ligand. The absorption spectra confirmed the presence of the Re catalyst attached onto the CuInS2 QD surface in the CuInS2 QD–Re hybrid system, and the loading efficiency of catalyst molecules was about 41% (mole of the Re catalyst on per mole of CuInS2 QDs) as estimated from the increase in absorbance at 375 nm compared to that of the neat QD sample.
Fig. 2 UV-Vis absorption spectra of CuInS2 QDs in aqueous solution, the CuInS2 QD–Re hybrid system and the Re catalyst in DMSO. |
To further confirm that the Re catalyst was attached on the QDs, Fourier-transform infrared spectroscopy (FTIR) measurements were conducted and the results are presented in Fig. 3A. The FTIR spectrum of the Re catalyst displayed three characteristic absorption peaks due to carbonyl stretches at νco = 1898, 1912 and 2021 cm−1, which was in agreement with previous reports on similar Re(bpy)(CO)3L complexes.44,56 Additionally, introduction of the azide group in the Re catalyst structure (Fig. 1) produced another absorption peak νN3 at 2110 cm−1. For the CuInS2 QD–Re hybrid system, the νco stretches were preserved, while the azide group stretch vanished due to the click reaction between the linker molecule and the catalyst molecule. In comparison, there were only weak νco stretch bands for the control sample prepared without a linker molecule (Fig. S5,† sample labeled as QDsRe), implying that the Re catalysts can adsorb on the surface of QDs only to a slight extent in the absence of linker molecules. Furthermore, Re signals were detected from the CuInS2 QD–Re hybrid system by energy-dispersive X-ray (EDX) measurements, as shown in Fig. 3B. Therefore, by combining the above results, we conclude that our CuInS2 QD–Re hybrid system was formed through the linker assisted click reaction.
Time-correlated single photon counting (TCSPC) measurements of the CuInS2 QD–Re hybrid system confirmed the accelerated decay of the fluorescence of the QDs in the presence of the Re catalyst. From the data presented in Fig. 4 and the results from biexponential fits in Table 1, we can see that the PL lifetime of QDs in the mixed solution partially decreased from that of the as-synthesized CuInS2 QDs, and the decrease was mainly in the fast decay component (τ1 = 0.587 ns (77.3%) for the mixed sample, and τ1 = 2.39 ns (62.8%) for the as-synthesized CuInS2 QDs), which was consistent with the results from steady-state fluorescence measurements. In contrast, the PL lifetime of the CuInS2 QD–Re hybrid system dramatically decreased, with τ1 = 0.243 ns (98.8%), τ2 = 19.7 ns (1.2%), indicating that almost all the fluorescence decayed within 1 ns. The short PL lifetime of the CuInS2 QD–Re hybrid system confirmed that the fluorescence of the CuInS2 QDs was quenched by the covalently bound Re catalyst, and the high PL quenching yield suggested that the ET process in the system was efficient. With the average PL lifetime defined as 〈τ〉 = A1τ1 + A2τ2, where A1 and A2 are the amplitude fractions, the average lifetime for ET can be calculated according to eqn (1). The efficiency of ET is then given by eqn (2), where the latter term is equal to the ratio of the areas under the respective PL decay curve.
(1) |
(2) |
Fig. 4 Decays of the photoluminescence of CuInS2 QDs, CuInS2 QD–Re hybrid system, and the mixture solution of CuInS2 QDs and the Re catalyst, monitored at the emission wavelengths above 640 nm, after excitation at 470 nm in DMSO. The lifetimes obtained from the fitting procedure are listed in Table 1. |
Sample | QDs | QD–Re | QD–Re mixture |
---|---|---|---|
a Some deviations from a perfect fit (see χ2/Nd values, and residual plots in Fig. S9) indicated that the interfacial kinetics was somewhat more heterogeneous than that described by the fit. For the purpose of the present analysis, however, a biexponential fit was still sufficient to demonstrate the large differences between the samples. | |||
τ 1 (ns) | 2.39 (62.8%) | 0.243 (98.8%) | 0.587 (77.3%) |
τ 2 (ns) | 36.7 (37.2%) | 19.7 (1.2%) | 29.9 (22.7%) |
Weighted average PL lifetime <τ> (ns) | 15 | 0.48 | 7.2 |
χ 2/Nd | 1.54 | 1.70 | 1.52 |
According to eqn (1) and (2), and assuming that all PL quenching arose from electron transfer, the average ET lifetime (〈τET〉) and efficiency (ηET) of 0.49 ns and 97% were obtained for the QD–Re hybrid system. In other words, the majority of the electron transfer processes from CuInS2 QDs to the Re catalyst occurred within 1 ns. This efficient ET process is presumably because of a strengthened electronic coupling between the CuInS2 QDs and the Re catalyst, through forming the QD–Re covalent bond.39 Note that any process faster than ca. 30 ps cannot be resolved by the TCSPC experiment.
In order to further investigate the origin of the PL quenching, transient absorption spectroscopy measurements in the mid-IR region were performed. The reduced rhenium should give a clear signature in the infrared region due to strong νco stretches according to the literature.28,57–59 Fig. S10† and Fig. 5A show the time-resolved IR (TRIR) spectra of the CuInS2 QD–Re hybrid system under 520 nm excitation after subtraction of the background absorption of the conduction band electrons in the QDs. The spectra show bleaches of the ground-state CO bands of the Re catalyst at 2030, 1924, and 1897 cm−1, which were slightly blue-shifted as compared to the FTIR spectrum of the QD–Re hybrid system. This blue-shift may be due to the solvent effect since the TRIR spectra were recorded in solution, while the FTIR spectra were recorded in solid samples.44 Simultaneously, the appearance of a positive band at 2012 cm−1 could be observed, which is red-shifted by 18 cm−1 from the ground state of the CO stretching band at 2030 cm−1. Numerous studies have shown that the νco stretch bands of the one-electron-reduced species [Re(bpy)(CO)3Cl]− normally shift to a lower energy by around 20 cm−1 from that of the ground state.28,32,40,57–62 Therefore, the positive band at 2012 cm−1 could be assigned to the reduced form of the Re catalyst. Similar TRIR features were also observed in CdSe QD–Re(CO)3Cl(dcbpy) complexes,40 and these results suggested that the photoinduced electrons in conduction band of QDs could effectively transfer to the Re catalyst, generating the Re catalyst anion:
QDs* + Re → QDs+ + Re−. |
Fig. 5 (A) fs-TRIR spectra for the CuInS2 QD–Re hybrid system under 520 nm excitation and (B) comparison of the fs-TRIR spectra at 1 ps for the CuInS2 QD–Re hybrid system and the mixed sample of CuInS2 QDs and Re catalysts. Inset: UV-Vis absorption spectra of solutions of the CuInS2 QD–Re hybrid system and the mixed sample of CuInS2 QDs and Re catalysts used in the fs-TRIR measurements. For the fs-TRIR spectra, the background absorption of the conduction band electrons of the QDs has been subtracted to emphasize the molecular signature of the Re catalyst. The raw data (i.e. before background subtraction) can be found in the ESI in Fig. S13.† The region below 1950 cm−1 is not shown because of the overlapping TRIR signals of the samples with absorption of the solvent (DMSO). |
From Fig. 5A, we can find that the feature of the Re catalyst anion can be observed already at 300 fs after excitation, which means the electron transfer from the QDs to the Re catalyst must occur within less than 300 fs. The transient IR spectra of the Re catalyst alone recorded under the same conditions show negligible features of the Re excited state and singly reduced Re, which excludes direct excitation of the catalyst at 520 nm (Fig. S11†). Unsurprisingly, the reduced Re catalyst was also detected from the mixed sample of CuInS2 QDs and Re catalyst (Fig. S12†). However, the amount of catalyst anions was less than the ones in the CuInS2 QD–Re hybrid system. As shown by Fig. 5B, the signal of the reduced Re catalyst was larger in the CuInS2 QD–Re hybrid system though the amount of the Re catalyst was slightly higher in the mixed sample (see the inserted comparison spectra in Fig. 5B). These results suggested that the ET process was more efficient in the covalently linked hybrid system that in the simple mixture. The signals from the reduced Re catalyst decayed with an average lifetime of ≥1.1 ns, with a significant component >5 ns (see Fig. S14 and Table S1†). Thus, the slower ET components indicated by TCSPC will not lead to an obvious rise of the reduced Re catalyst IR signal, as formation occurs concomitant with (or is slower than) its subsequent decay.
Because a LED white light (≥420 nm) was used as the light source, the Re catalyst alone exhibited moderate photocatalytic activity toward CO2 reduction, with a TON of 8 after 6 h. No H2 was generated from the photocatalytic system (see GC traces in Fig. S15†), which is similar to other reports.44 When the Re catalyst was attached on the surface of the QDs in the CuInS2 QD–Re hybrid system, the TON of CO by the catalyst was significantly enhanced to 16 after 6 h of photocatalysis. In contrast, in the mixed solution of QDs and Re catalyst, the QDs inhibited the activity of the catalyst, presumably by parasitic light absorption, resulting in a lower TON of 5. These results demonstrated that the linkage between QDs and the catalyst was important for the improvement of the photocatalytic activity of the catalyst, which could contribute to efficient ET from the QDs to the catalyst and then benefit the electron accumulation of the catalyst for photocatalytic CO2 reduction. The lower TON for the mixed system compared to the Re catalyst alone can be explained by the strong visible light absorption of CuInS2 QDs, which competes with the light absorption of the Re catalyst.
Based on the previous studies by Ishitani and other groups,14,63,64 we proposed a photocatalytic mechanism of the CuInS2 QD–Re hybrid system for CO2 reduction. Without QDs, a bare Re catalyst is a classic catalyst for CO2 reduction with high selectivity, and the photocatalytic mechanism is well studied.14,44,65,66 As shown in Scheme 1, after absorbing light with a wavelength shorter than 500 nm, the Re catalyst can be excited, and then the excited catalyst Re* will quickly acquire an electron from the electron donor TEOA, generating one-electron reduced (OER) species of the Re complex. Notably, unlike the electro-catalysis of CO2 reduction by the Re catalyst (Fig. S8b†),56 the photo-induced process is conducted by the OER species of the Re complexes. The OER species play two significant roles in photocatalytic CO2 reduction: firstly, they can bind with CO2 in solution after ligand elimination, forming a [CO2 adduct]; secondly, a neighboring OER species can also feed electrons to the [CO2 adduct], converting CO2 to CO.14,63 When CuInS2 QDs were bound to the Re catalyst, forming the CuInS2 QD–Re hybrid system, the CuInS2 QDs can utilize the solar light with longer wavelengths to generate electron–hole pairs. Then the photoinduced electrons in the QDs can transfer to the Re catalyst, contributing to more OER species, as proved by the TRIR results, and the QDs can subsequently be regenerated by TEOA. In the QD–Re hybrid system, the average lifetime of OER, i.e. the reduced Re catalyst, was ≥1.1 ns (τ1 = 20.4 ps (54%), τ2 = 994 ps (31%), τ3 > 5 ns (15%), Fig. S12 and Table S1†), which is long enough to feed electrons to the [CO2 adduct] and reduce CO2. Compared to an unsensitized Re catalyst, there is an additional channel for the formation of OER species in the CuInS2 QD–Re hybrid system; therefore, the photocatalytic activity of the CuInS2 QD–Re hybrid system was improved.
After full optimization of the hybrid system, we aim to immobilize the QD–Re system on NiO electrodes to fabricate a photocathode for CO2 reduction.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8dt01631c |
This journal is © The Royal Society of Chemistry 2018 |