Efficient photocatalytic hydrogen evolution with ligand engineered all-inorganic InP and InP/ZnS colloidal quantum dots

Photocatalytic hydrogen evolution is a promising technique for the direct conversion of solar energy into chemical fuels. Colloidal quantum dots with tunable band gap and versatile surface properties remain among the most prominent targets in photocatalysis despite their frequent toxicity, which is detrimental for environmentally friendly technological implementations. In the present work, all-inorganic sulfide-capped InP and InP/ZnS quantum dots are introduced as competitive and far less toxic alternatives for photocatalytic hydrogen evolution in aqueous solution, reaching turnover numbers up to 128,000 based on quantum dots with a maximum internal quantum yield of 31%. In addition to the favorable band gap of InP quantum dots, in-depth studies show that the high efficiency also arises from successful ligand engineering with sulfide ions. Due to their small size and outstanding hole capture properties, sulfide ions effectively extract holes from quantum dots for exciton separation and decrease the physical and electrical barriers for charge transfer.


Supplementary Tables
Supplementary Table 1. ICP-OES results for QDs before and after ligand exchange. InP QDs (525 nm) and InP/ZnS QDs (525 nm, 15 min) powders were treated with aqua regia for dissolution. As Zn precursors were introduced during the synthesis of InP QDs for size control (cf. Methods in main text), a small amount of Zn could be detected for InP QDs. It should be pointed out that when more zinc(II) iodide was added to obtain QDs with smaller size (480 nm and especially 440 nm), the size distribution of the QDs became broader. As a result, a certain amount of QDs with smaller size did not precipitate from the system, but remained in the supernatant during purification.

In
These QDs should be discarded as well. 5 In addition, we also found that with higher amounts of zinc(II) iodide added, the unreacted raw material maintained a similar color to the QDs itself. This renders the purification process more difficult, because the precipitation of QDs is often calibrated through the color changes of the supernatant.

Supplementary Note 2: Influence of pH and other parameters on QDs in photocatalytic media.
In addition to the description in the main text, we furthermore increased the acidity of the solution gradually (by decreasing the pH in steps of 0.5 units) and we noticed that precipitation set in only after the pH dropped to 2.5 ( Supplementary Fig. 19). This value is much lower than the pKa of H2S/HS -(6.9-7.0). This result agrees well with a previous study reporting on the precipitation of thiol capped Cd-based QDs by X.
Peng's group. 6 The pKa of -S/-SH for thiol was around 10, but ligand removal (and precipitation of QDs) occurred at pH 4-6 for different series of QDs. Both results show that once the ligands are bound on the surface of QDs, their original nature can be drastically changed.
When the pH was adjusted to 4.5 in zeta potential measurements (Supplementary Table 4 Introduction of H2A/NaHA decreased the zeta potential of the system as well, probably due to adsorption of a small amount of H2A/NaHA on the surface of the QDs. 8 Although H2A/NaHA probably does not interact strongly with QDs, the large amount of H2A/NaHA (0.2 M) might increase the interaction. Given the probable inaccuracy of DFT calculations, we also carried out further experiments, such as steady state and time resolved spectroscopy, to study the specific energy level of the defect states. However, such attempts were not successful. For example, we did not observe any trap-state emission of InP based QDs, neither in the visible light nor in the near infrared light range. 8 Considering the diversity and complexity of the surface states of QDs, 9 the defect states may have a wide energy level distribution, and to date there is still a lack of widely recognized methods for the determination of QD defect energy levels. Therefore, the energy level positions of either intrinsic defects or surface S 2ligands were indicated indistinctly here with several line symbols ( Supplementary Fig. 28), which is a common practice in the description of defect states of QDs.  Fig. 25b). As for the structure of InP with intrinsic defects, a (2×2) period InP (001) slab model was employed and the coverage of In vacancy was 1/4 ML (Supplementary Fig. 25c). A layer of ZnS (001) was combined with InP to mimic the structure of InP/ZnS (Supplementary Fig. 25d).

Supplementary Method 2: Calculation of InP and InP/ZnS QD concentrations.
As it is difficult to find a straightforward and appropriate empirical formula to directly calculate the QD concentration of InP QDs in contrast to Cd-based QDs, 15 we first derived the InP unit/molecule concentration from absorption spectra. The basic assumption is that the intrinsic absorption coefficient of colloidal nanocrystals is close to that of bulk materials in the short wavelength range. Knowing the intrinsic absorption coefficient of bulk InP, we can use these values to calculate the molecular and mass concentration of InP QD solutions. 16 Different absorption wavelengths of the InP QDs were used for the calculation, in particular 310 nm, 17,18 350 nm, 19 and 413 nm. 16 In our case, we have selected 413 nm for concentration estimations to minimize the influence of ZnS, which may display absorption in the shorter wavelength range. Calculations were performed as follows: 16 The intrinsic absorption coefficient is given by: for water at 413 nm were obtained from other reports. 20,21 The amount of InP unit/molecule cuvette is calculated as: where L is the cuvette length (m), Vm is the InP molar volume, and A is the measured absorbance of QDs at 413 nm (A should be at a suitable value range). Here, L is 0.01 m and Vm is 3.0 × 10 -5 m 3 mol -1 . 16 The InP unit /molecule concentration molecule is then obtained from the following equation: For hexane as solvent, .
For water as solvent, .
To further calculate the concentration of QDs, the number of molecules per QD, N, was estimated. Taking InP/ZnS QDs (525 nm) for instance, we here assume that the QDs are spherical. The corresponding diameter of QDs was averaged at 2.7 nm from the TEM statistics (see Supplementary Fig. 8). The number of InP molecules per QD was then calculated as follows: 22 Two methods were applied to synthesize CdSe QDs with comparable absorption peaks and sizes to InP QDs (525 nm). The first route via octadecene (ODE) was modified from literature 24,25 and is referred to here as S44 CdSe-A. Briefly, 0.4 mmol of CdO, 0.5 mL of oleic acid (OA), 1.5 mL OLA, and 8 mL of ODE were mixed in a 50 mL three-neck flask, and the mixture was heated to 240 °C. Then 1 mL of ODE solution containing 100 µL TOP-Se (2 M) was injected. The reaction was kept for 30 s and cooled down to room temperature quickly.
The second route (here: CdSe-B) via trioctylphosphine oxide (TOPO) was taken without changes from literature. 26 Purification and ligand exchange for CdSe QDs were conducted as described for InP QDs in the Methods in the main text.
The photocatalytic hydrogen evolution experiment was carried out at a concentration where InP QDs and CdSe QDs in the reaction system had the same absorbance at 525 nm (i.e. the wavelength of the LED light source). The remaining hydrogen evolution conditions were the same. We emphasize that these control experiments are just for reference purposes and do not permit absolute conclusions, because the optimal conditions and the most suitable catalysts for hydrogen evolution may differ for different QD types.

Supplementary Method 4: Calculation of the TON value and hydrogen evolution rate of InP/ZnS QDs.
The TON value based on QDs and hydrogen evolution rate was calculated from Supplementary Fig. 13a We also estimated the hydrogen evolution rate based on QD mass. The mass of InP/ZnS QDs was determined as 0.42 ± 0.03 mg in a typical photocatalytic experiment from four parallel gravimetric measurements (the amounts of QDs were proportionally increased for precise weighing). Accordingly, the hydrogen evolution rate is 45 mmol·g -1 ·h -1 .
Besides, we calculated the mass of QDs from the molar amount of InP QDs, which was derived from the optical absorbance of QDs. For a hydrogen evolution system with InP/ZnS QDs containing 2 μmol of InP For monochromatic LED light sources (λ = 465 nm), the accurate illumination power for a certain area of the reaction mixture (1 cm 2 ) was measured using a digital photodiode power meter (Newport, model 842-PE).
Photocatalytic hydrogen evolution was performed in a custom-made spectro-cell with a total volume of 7 mL .
Amount of incident photons per second: Accordingly, we also determined other AQY/IQY of the system at several other wavelengths, and the results are listed in Supplementary Table 3. In addition, good correlation of the AQY values with the absorption spectra of InP/ZnS QDs is shown in Supplementary Fig. 14 pentahydrate (TMAH), should be added first. 5 In a typical process, 1 mL of InP or InP/ZnS QDs (about 3-5 mg mL -1 ) in HEX was mixed with 1 mL of 0.1 M MPA/MUA and 0.2 M TMAH in NMF. The mixture was kept stirring at room temperature until a total phase transfer of QDs from HEX to NMF was achieved. The washing process for MPA and MUA was the same as applied for S 2capping, while higher acetone amounts were required to precipitate the QDs. The photocatalytic hydrogen evolution experiment was carried out with the same concentration of QDs (1.6 μM) with different ligands.

Supplementary Method 8: Ligand exchange of InP/ZnS QDs with Cland PO4 3-.
To further show the significance of S 2ligands, two other inorganic ligands, namely Cland PO4 3-, were chosen here for comparison because of their feasibility for ligand exchange and their stability in ambient environments. 29,30 It should be pointed out that the use of NaCl or KCl did not lead to a successful phase transfer of QDs into water, so that InCl3 and ZnCl2 were used as Clsources instead. This phenomenon is very similar to previous reports. 29 In a typical process, 1 mL of InP or InP/ZnS QDs (about 3-5 mg mL -1 ) in HEX was mixed with 1 mL of 0.025 M Na3PO4·12H2O in FA, or 0.05 M ZnCl2/InCl3 in NMF. The mixture was kept under stirring until a total phase transfer of QDs from HEX to FA/NMF was achieved, but here more time or higher temperature are needed for a total phase transfer of the QDs in contrast to S 2-. The washing process for Cland PO4 3was the same as that applied for S 2capping. The photocatalytic hydrogen evolution experiment was carried out with the same concentration of QDs (1.6 μM) with different ligands. For photocatalytic experiment comparison with QDs-Cl, we finally used QDs obtained from ligand exchange with InCl3, as the hydrogen evolution rate for InCl3 is better than that of QDs after ligand exchange with ZnCl2.