Composition effect on the carrier dynamics and catalytic performance of CuInS₂/ZnS quantum dots for light driven hydrogen generation

The efficient conversion and storage of solar energy represent a promising approach to address the increased energy demand and environmental problems caused by the combustion of fossil fuels.^1–3 One attractive strategy for solar energy conversion is light-driven water splitting to generate H₂ based on semiconductor photocatalysts.^4,5 Since the first report of light-driven photocatalytic H₂ generation based on semiconductors,^6,7 various semiconductor nanocrystals have been reported as photocatalytic materials.^8–10 Among them, CuInS₂ (CIS) quantum dots (QDs), known as colloidal semiconductor nanocrystals with low toxicity, have become one of the most promising semiconductor photocatalytic materials for H₂ generation due to their large optical absorption coefficient, tunable bandgap, and high photostability.^11–14 However, majority of CIS based systems reported are decorated with cocatalysts, where CIS mainly serves as a photosensitizer.^15–21 In contrast to these reports, we recently found that CIS/ZnS core/shell QDs alone can efficiently catalyze the H₂ evolution reaction (HER) in an aqueous solution with an exceptionally long lifetime.²² While this finding demonstrated the promise of CIS/ZnS as a single-component photocatalyst for the HER, the solar-to-H₂ quantum yield is quite low (<1%). Given the promising properties of this single-component system for the HER, it is necessary to further explore this system to improve its quantum efficiency for H₂ generation.

Previous studies have shown that one critical limitation of CIS QDs to their application in photovoltaics and photocatalysis is associated with the surface related defect states, which results in defect-induced nonradiative transition and poor emission quantum yield.^23–26 A number of experimental and theoretical studies have demonstrated that the surface defect states in CIS QDs can originate from sulfur vacancy (V_S), copper site substituted by indium (In_Cu), copper vacancy (V_Cu), indium vacancy (V_In), and indium site substituted by copper (Cu_In), where V_S and In_Cu act as the donors and V_Cu, V_In, and Cu_In act as acceptors.^22,27–29 In general, in Cu rich CIS QDs, the main transitions are from the conduction band to V_In or Cu_In and from V_S to the valence band. In Cu deficiency CIS QDs, the main transitions should be from V_S to V_Cu and from In_Cu to V_Cu.^30–32 These literature data suggested that the defect states in CIS are largely dependent on their composition, and altering the composition of CIS may serve as a viable approach to tune their photophysical properties and photocatalytic activity for the HER.

In this work, we report the effect of cation composition on the carrier dynamics, charge separation, and photocatalytic performance of CIS/ZnS QDs for light driven H₂ generation in an aqueous solution, where the ratio of Cu to In cations was systematically varied. We show that decreasing Cu to In ratio (increasing Cu deficiency) leads to slightly faster electron-hole pair recombination but significantly enhanced photocatalytic performance for the HER. This can be explained by improved overall charge separation due to faster electron transfer (ET) from the sacrificial donor to CIS/ZnS QDs in the sample with higher Cu deficiency resulting from the more positive valence band edge.

Chemicals and Materials. Copper(i) iodide (CuI, 98%), indium(iii) acetate [In(OOCCH₃)₃, 99.99% metals basis], 1-dodecanethiol (DDT, 98%), and toluene (anhydrous, 99.8%) were purchased from Alfa Aesar. 1-octadecene (ODE) and oleylamine (OLA) were purchased from Sigma-Aldrich. The rest chemicals and solvents were purchased from VWR. All chemicals were used as received without further purification.

Synthesis of CIS/ZnS QDs. In a typical synthesis for CIS with a Cu:In ratio of 1:2,³³ CuI (10 mg, 0.05 mmol), In(OAc)₃ (29.2 mg, 0.1 mmol), and 1-dodecanethiol (DDT) (5 ml) were mixed in a three-necked flask. The reaction mixture was degassed with N₂ for 30 min at room temperature. After that, the mixture was heated to 150 °C until a clear yellow solution was formed. Then, the mixture was heated to 230 °C and kept at this temperature for 5 min. For further growth of the ZnS shell, the Zn stock solution (4 ml) (see below for details) was added dropwise into the reaction mixture, after which the mixture was heated to 240 °C and kept at this temperature for 1 h. The zinc stock solution was prepared in advance by mixing Zn(OAc)₂ (0.219 mg, 1 mmol), oleylamine (OLA) (1 ml), and ODE (3 ml) in a flask, which is followed by degassing with N₂ for 30 min and heating to 100 °C until a colorless solution was formed. The resulting CIS/ZnS QDs were precipitated out using acetone, followed by centrifuging and decanting the supernatant. Half of the obtained samples were dispersed in toluene for optical property measurement, while the other half were used to prepare water soluble CIS/ZnS QDs. To synthesize CIS/ZnS QDs with different Cu to In ratios (i.e., 1:5, 1:8, and 1:10), the amount of CuI was kept the same while the amount of the In source was varied.

The Synthesis of Water Soluble CIS/ZnS QDs. The water soluble CIS/ZnS QDs were prepared through ligand exchange procedure in the presence of mercaptopropionic acid (MPA).³⁴ In a typical synthesis, 2 ml of MPA, 3 ml of N, N-dimethylformamide (DMF), and 100 mg of CIS/ZnS QDs were added into a 25 ml flask, and degassed with N₂ for 30 min. The mixture was then heated to 130 °C and kept at this temperature for 30 min, which results in the formation of a clear solution. The CIS/ZnS was precipitated out by adding 40 ml of 2-propanol and centrifugated at 6000 rpm for 10 min. The obtained MPA capped CIS/ZnS QDs were dispersed in 15 ml H₂O.

Photocatalytic Hydrogen Generation. All photocatalytic reactions were performed in a 11 ml vial under the illumination of a 405 nm light-emitting diode (LED) (3 mW). Typically, 0.1 ml of the CIS/ZnS QD solution was mixed with ascorbic acid (AA) (1.00 ml, 0.2M in H₂O, pH = 5) and H₂O (2.9 ml) in a sample vessel. The vial was sealed, degassed with nitrogen, and transferred to photocatalysis apparatus. The amount of evolved hydrogen (taken from 200 µl of the headspace of the vial) was quantified by using an Agilent 490 micro gas chromatograph (5 Å molecular sieve column).

Standard Characterization. Steady-state UV–visible absorption spectra were taken using an HP Agilent 8453 spectrophotometer (UV-Visible) and a Cary 5000 spectrometer (UV-Visible-NIR). Steady-state emission spectra and time-correlated single photon counting (TCSPC) lifetime were measured using a Photon Technology International QuantaMaster 40 spectrofluorometer. To collect the optical properties, the prepared water soluble CIS/ZnS sample solution was placed in a 10 mm cuvette. The steady-state emission spectra and TCSPC were collected with 0.1 OD absorption at 400 nm. A Rigaku MiniFlex II diffractometer with Cu Kα radiation was used to collect X-ray diffraction (XRD) patterns. Energy dispersive X-ray (EDX) spectroscopy data were collected on a Hitachi S-4800 scanning electron microscope with an EDAX XM2–60S energy dispersive spectrometer.

Transient Absorption (TA) Spectroscopy Measurement. The TA setup is a bump and probe system which is based on a regenerative amplified Ti-Sapphire laser system (Solstice, 800 nm, <100 fs FWHM, 3.5 mJ/pulse 1 kHz repetition rate). The traveling-wave optical parametric amplifier of white-light continuum (TOPAS prime) will generate the tunable pump (235–1100 nm) with 75% of the output Ti-Sapphire laser. The sapphire window on a translation stage can generate the tunable UV-visible probe pulses (430–750 nm) with the other 25% of the output Ti-Sapphire laser. The femtosecond TA measurements were performed in a Helios ultrafast spectrometer (Ultrafast Systems LLC). The prepared water soluble CIS/ZnS samples were placed in 2 mm cuvettes and all samples were under stirring during measurements to avoid multiexciton generation.

Figure 1(a) shows the XRD patterns of CIS/ZnS with different Cu to In ratios. The Cu to In ratios of these samples were measured by EDX (Table S1) and were used to label these samples. All samples show three main peaks with 2θ values at 27.8°, 46.9°, and 54.9°, which agree well with the literature data and support the formation of CIS/ZnS.^35,36 Figure 1(b) shows the UV-visible absorption spectra of CIS/ZnS with four different ratios, which are featured by shoulder absorption bands at around 500 nm corresponding to the optical bandgap.³⁷ With decreasing Cu to In ratio (increasing Cu deficiency), these absorption bands shift to a shorter wavelength, suggesting increasing bandgap between conduction and valence bands. Via the least-square fit of the linear region of a (Ahν)² vs hν plot (A, absorbance; h, Planck’s constant and ν, frequency),³⁸ the bandgaps of these CIS/ZnS QDs were estimated to be 2.25 eV, 2.40 eV, 2.50 eV, and 2.65 eV for QDs with Cu_0.54InS₂, Cu_0.28InS₂, Cu_0.17InS₂, and Cu_0.12InS₂, respectively [Inset of Fig. 1(b)].

In addition, the conduction band edges of these QDs were estimated by Mott-Schottky plots (Fig. S1),³⁹ where a similar conduction band edge (∼0.75 eV vs Ag/AgCl) was obtained for these CIS/ZnS QDs, suggesting that altering the Cu to In ratio within the 0.54:1 to 0.12:1 range has negligible impact on the conduction band edge. This also suggests that the change of bandgap with the Cu to In ratio is likely due to the shift of the valence band position, i.e., the valence band edge moves to more positive potential with increasing Cu deficiency. These results are consistent with previously reported data,^40,41 suggesting the validity of the models we used to estimate the bandgap and conduction band edge.

The emission spectra of these samples following excitation at 400 nm are shown in Fig. 1(c). The emission peak position shifts from 600 nm to 554 nm with increasing Cu deficiency. This follows the same trend as their absorption spectra, further supporting that increasing Cu deficiency results in the increase of the bandgap of CIS/ZnS. The emission quantum yield (QY) is estimated to be 20%, 11%, 9%, and 8% for QDs with Cu_0.54InS₂, Cu_0.28InS₂, Cu_0.17InS₂, and Cu_0.12InS₂, respectively, which decreases with increasing Cu deficiency and suggests more defect states in the sample with higher Cu deficiency (Fig. S2). In order to gain insight into the average emission lifetime of CIS/ZnS QDs with different Cu to In ratios, the time resolved emission spectra of these samples were measured following 400 nm excitation. Figure 1(d) shows the time-resolved emission lifetime decay kinetics of CIS/ZnS QDs with different Cu to In ratios. It can be seen that Cu_0.12InS₂ has the fastest decay, indicating the shortest average emission lifetime. The decay curves of the samples were fitted with a tri-exponential decay function,

where A₁, A₂, and A₃ are fractional contributions of PL decay lifetimes of τ₁, τ₂, and τ₃, respectively.

The fitting results are listed in Table S2. The obtained average lifetimes of these samples are 133.3 ns for Cu_0.54InS₂, 123.4 ns for Cu_0.28InS₂, 117.5 ns for Cu_0.17InS₂, and 113.2 ns for Cu_0.12InS₂, which shows that the emission lifetimes of these samples are similar to the minor trend showing decreasing emission lifetime with increasing Cu deficiency. These results also suggest that different Cu to In ratios in the CIS/ZnS QDs have negligible impact on electron-hole radiative recombination time.

The carrier recombination dynamics of CIS/ZnS QDs with different Cu to In ratios was further investigated using femtosecond transient absorption (TA) spectroscopy to unravel the early time carrier relaxation dynamics. The TA spectra of these samples were all collected under 400 nm excitation with 20 µJ/cm² power. Under this excitation power, the average number of exciton per QD is estimated to be 0.19 for Cu_0.54InS₂ (Fig. S3).⁴² Shown in Fig. 2 are the TA spectra of CIS/ZnS QDs, where the insets show the early time spectra. The TA spectra of all four CIS/ZnS QD samples were featured by a negative band and a broad positive absorption, where the former can be assigned to the exciton bleach band and the latter can be attributed to the electron absorption.^22,43 It can be clearly seen that the exciton bleach band shifts to a shorter wavelength with increasing Cu deficiency, which is consistent with the UV-Vis absorption and emission results above. As shown in the insets of Fig. 2, the early time spectra (<1 ps) for all four samples show an ultrafast rising component for the electron absorption signal. This component can be attributed to the electron trapping and/or cooling process, which is based on both the literature and our own experimental results. Previous data have shown that this rising component can be attributed to the electron trapping process,^26,43–45 while our experiments show that the rising component disappears when 450 nm light was used as the excitation wavelength (Fig. S4), suggesting that the cooling process contributes to this rising component.

To further evaluate the carrier dynamics of these CIS/ZnS QDs, we compared the exciton bleach recovery and the kinetic traces of the electron absorption signals of these samples. The kinetic traces were normalized at a later time in order to have a better comparison for the early time kinetics. Due to the small signal of the exciton bleach band as well as the overlap between the exciton bleach and the electron absorption signal for Cu_0.12InS₂, we are not able to compare its early time spectral evolution for the exciton bleach signal with other samples. Nevertheless, the exciton bleach recovery of Cu_0.54InS₂, Cu_0.28InS₂, and Cu_0.17InS₂ becomes faster with increasing Cu deficiency, suggesting that the Cu deficiency can affect the electron-hole recombination process. This appears to conflict with emission lifetime measurement discussed above, where Cu deficiency has negligible impact on the electron-hole recombination time. This discrepancy can be explained by the different time windows those time resolved emission and absorption spectroscopies measure, where the former probes the recombination dynamics at > 10 ns window due to its limited time resolution (∼10 ns), while the later measures the early time recombination dynamics at < 5 ns due to the limited time window of our TA spectroscopy. On the other hand, as shown in Fig. 3(b), the kinetic trace of the electron absorption decays slightly faster in the sample with higher Cu deficiency, which is consistent with the exciton bleach recovery. These results can be more clearly seen in the fitting results. As shown in Fig. 3 and Table S3, the kinetic traces of all samples can be fit by a four-exponential decay/rise function. Due to the presence of multiple carrier trapping states and recombination channels in CIS QDs, it is challenging to correlate each time constant with the relaxation pathway. However, according to the time scales of carrier trapping and recombination reported by previous studies,^23,25,26,44 the ultrafast rising/decay component (τ₁) which is close to the instrument resolution limit (∼0.2 ps) can be attributed to the electron cooling and/or trapping to the shallow electron trap states; τ₂ is likely due to the additional carrier trapping by deep-level trap states, and τ₃ and τ₄ can be attributed to the recombination of the electrons from these trap states with the holes from the valence band. However, due to the presence of ≫ 5 ns time constant which is beyond the time window of our TA spectroscopy, we are not able to accurately determine the carrier recombination time. Nevertheless, since τ₂ and τ₃ decrease slightly with increasing Cu deficiency, it suggests that the recombination process is enhanced slightly in the sample with higher Cu deficiency.

To evaluate the correlation of the composition of CIS/ZnS QDs with their function for catalysis, we examined the catalytic performance of these samples for H₂ generation under 405 nm LED irradiation. The sample container is a glass vial with 4 ml reaction solution and 7 ml head space. The reaction solution includes 1 ml ascorbic acid (AA) (pH = 5) as the electron donor, 0.1 ml of CIS/ZnS aqueous solution (∼0.2 mg CIS/ZnS), and 2.9 ml H₂O. The glass vial was sealed with a rubber stopper and purged with N₂ for at least 10 min before irradiation. Figure 4(a) shows the amount of H₂ generated as a function of irradiation time. It can be clearly seen that the H₂ generation efficiency increases significantly with the increasing Cu deficiency, where the amount of H₂ generated for Cu_0.12InS₂ is almost 100 times higher than that of Cu_0.54InS₂. In order to have a valid comparison of the H₂ generation efficiency among different CIS/ZnS QDs to account for the difference of absorbed photons under LED illumination (405 nm), we evaluated the H₂ generation efficiency in terms of photon-to-H₂ quantum yield according to the method published previously.⁴⁶ The obtained photon-to-H₂ quantum yields are ≪ 1%, 0.7%, 3.8%, and 5.2% for the Cu_0.54InS₂, Cu_0.28InS₂, Cu_0.17InS₂, and Cu_0.12InS₂, respectively, which increases with increasing Cu deficiency. While higher photon-to-H₂ quantum yields (∼5%–20%) have been reported previously in colloidal photocatalytic systems,^46–50 all of these systems include a cocatalyst that plays a major role in boosting the photocatalytic efficiency. To the best of our knowledge, CIS based photocatalytic systems without cocatalysts show a much lower photon-to-H₂ quantum yield (≤5%).^22,51 Indeed, the photon-to-H₂ quantum yield in our system with Cu_0.12InS₂ (5.2%) is slightly higher than that of the previously best-performing system (5%).⁵¹ These results together suggest that the presence of Cu deficiency can significantly enhance the photocatalytic performance of CIS/ZnS QDs and altering the Cu to In ratio is a viable approach to further optimize the photocatalytic performance of CIS/ZnS QDs for the HER.

Note that CIS/ZnS QDs with higher Cu deficiency show slightly faster charge recombination but much better photocatalytic activity. This appears to conflict with the commonly accepted model, where the elongated electron-hole pair lifetime is beneficial for H₂ generation.^22,46,51,52 In order to gain insight into the origin of this conflict, we examined the electron transfer (ET) process from AA to CIS/ZnS QDs, which is another key pathway that determines the overall efficiency for H₂ generation. Electron transfer from AA to CIS/ZnS QDs was evaluated by measuring the steady state emission quenching of CIS/ZnS QDs as a function of AA concentration followed by Stern-Volmer analysis. As shown in Fig. 4(b) and Fig. S6, we found that the ET rate constants are 0.58 × 10⁸ M⁻¹ S⁻¹, 1.02 × 10⁸ M⁻¹ S⁻¹, 1.77 × 10⁸ M⁻¹ S⁻¹, and 2.74 × 10⁸ M⁻¹ S⁻¹ for QDs with Cu_0.54InS₂, Cu_0.28InS₂, Cu_0.17InS₂, and Cu_0.12InS₂, respectively, which increase considerably with increasing Cu deficiency, following the same trend as their photocatalytic performance. Combining the results from the carrier dynamics of CIS/ZnS QDs measured by TA spectroscopy, emission quenching experiment, and their photocatalytic performance for H₂ generation, we believe that the ET from AA to CIS/ZnS QDs plays an important role in overall charge separation and catalytic activity for H₂ generation, where the faster ET process in the sample with higher Cu deficiency results in more efficient overall charge separation and thus significantly enhanced photocatalytic performance. The faster ET in CIS/ZnS QDs with higher Cu deficiency may result from their lower valence band, which leads to a larger driving force for the ET to occur from AA to CIS/ZnS QDs.

In summary, we report the dependence of carrier and charge separation dynamics of CIS/ZnS QDs and their photocatalytic performance for light driven hydrogen generation on the cation composition (i.e., Cu to In ratio). With increasing Cu deficiency (decreasing Cu to In ratio), we show that the electron-hole pair recombination in CIS/ZnS QDs measured by TA spectroscopy is a little faster while the photocatalytic activities for H₂ generation are significantly improved. This can be explained by the notably enhanced ET from AA to CIS/ZnS QDs resulting from the lower valence band edge of QDs (larger driving force for ET) for samples with higher Cu deficiency, which facilitates overall charge separation. This work suggests that altering the cation composition in CIS/ZnS QDs is a viable approach to further improve their photocatalytic performance for solar to fuel conversion.

See the supplementary material for the Mott-Schottky plots, kinetics comparison for each QD at different probe wavelengths, tables that list the fitting parameters for time resolved emission and absorption kinetic traces, and Stern-Volmer plots for emission quenching experiments.

This work was supported by National Science Foundation (Grant No. DMR-1654140) and ACS-PRF (Grant No. 57503-DNI6). Sizhuo Yang and Wenhui Hu acknowledge the John J. Eisch fellowship during the 2018–2019 and 2019–2020 academic years, respectively.