Band Structure Engineering and Defect Passivation of Cu x Ag1-x InS2/ZnS Quantum Dots to Enhance Photoelectrochemical Hydrogen Evolution.

The AgInS2 colloidal quantum dot (CQD) is a promising photoanode material with a relatively wide band gap for photoelectrochemical (PEC) solar-driven hydrogen (H2) evolution. However, the unsuitable energy band structure still forms undesired energy barriers and leads to serious charge carrier recombination with low solar to hydrogen conversion efficiency. Here, we propose to use the ZnS shell for defect passivation and Cu ion doping for band structure engineering to design and synthesize a series of Cu x Ag1-x InS2/ZnS CQDs. ZnS shell-assisted defect passivation suppresses charge carrier recombination because of the formation of the core/shell heterojunction interface, enhancing the performance of PEC devices with better charge separation and stability. More importantly, the tunable Cu doping concentration in AgInS2 CQDs leads to the shift of the quantum dot band alignment, which greatly promotes the interfacial charge separation and transfer. As a result, Cu x Ag1-x InS2/ZnS CQD photoanodes for PEC cells exhibit an enhanced photocurrent of 5.8 mA cm-2 at 0.8 V versus the RHE, showing excellent photoelectrocatalytic activity for H2 production with greater chemical-/photostability.

Introduction

Colloidal quantum dot (CQD) photoelectrochemical (PEC) cells have improved significantly in performance for hydrogen production over the past decade,[1,2] due to their excellent optoelectronic properties. The size-, structure-, and chemical-composition-dependent bandgap of CQDs provides an opportunity to harvest a wide solar spectrum up to the near-infrared (NIR) range.[3] This property makes CQDs attractive photosensitizer candidates for photoelectrode materials and allows for application in the visible and NIR-bandgap photovoltaic and optoelectronic devices,[4,5] such as solar cells,[6,7] photodetectors,[8,9] and light-emitting diodes (LEDs).[10−13] Moreover, multiple exciton generation in CQDs has the potential to greatly increase the utilization of the solar fluence and boost the solar-to-hydrogen (STH) conversion efficiency of CQD PEC cells.[14] Rapid progress in CQD PEC cells has been achieved by improvements in the PEC cell efficiency and stability. Energy band engineering, surface passivation, and device architecture optimization are three major requirements for improving light absorption and increasing charge collection, which further increases their promise.[15,16]

Generally, a typical glass|FTO|TiO2|CQD-based photoanode (Figure a) for PEC cells displays a heterojunction with a suitable interfacial energy band alignment in a PEC cell configuration (Figure b). When irradiated by light, photogenerated exciton dissociation occurs at the CQD/TiO2 heterogeneous interface, and the photogenerated electrons transfer into TiO2 with a less negative conduction band (CB) and move to the metal counter electrode to conduct a H2O reduction reaction for H2 generation.[17,18] The photogenerated hole transfers to a less positive valence band (VB) via the contact interface and is further consumed by the S2–/SO32– scavenger in the electrolyte.[19] In previous reports, it is proved that the aforementioned efforts to improve the H2 production performance are essential.[2,20] However, further advances still depend on the suitable electronic band structure, efficient charge transfer, and high photochemical stability of the core/shell CQD (Figure c) itself.

Figure 1

Manipulating the catalytic activity of photoanodes by modulating the structure of CQDs: (a) Schematic diagram for the CQD-based photoanode with the structure of glass|FTO|TiO||CQDs. (b) Figure and predictable band alignment of CQD-based photoanode photoelectrochemical cells. (c) Band structure of Type I CQDs. (d) Schematic illustration for band structure engineering and defect passivation of Cu-doped AgInS2/ZnS CQDs. (e) The luminescence photograph of AgInS2 and AgInS2/ZnS CQDs synthesized at different reaction temperatures under UV lamp excitation at 325 nm. (f) Photoluminescence decay curves of AgInS2 and AgInS2/ZnS CQDs synthesized at 180 °C.

Band structure engineering, such as size/shape controlling, doping strategy, and chemical composition management have demonstrated enhanced electron-extraction efficiency in CQD PEC cells.[21] In CQDs, intentional electronic doping results in exchange charges with the CB and VB via radiative and nonradiative transitions, affecting the electric/optical properties of host CQDs.[15,22] Significant improvements in electronic band alignment, luminescence excited-state lifetime, and quantum efficiency for CQDs have been achieved by the incorporation of aliovalent dopants such as Cu, Mn, and Ni.[23−25] Among those, Cu-doped CQDs have been well synthesized, as the various host CQDs progressed from heavy metal (Pb and Cd) to heavy metal free (I–III–VI: Cu, Ag, Zn, In, S).[26] The Cu doping in host CQDs is investigated extensively and derived from the Cu oxidation state, Cu-related emission, and its stability.[27] In an acceptable mechanism, Cu T2 energy states between the CB and VB of host CQDs induce the elimination of the intrinsic band-edge emission of host CQDs, resulting in a broad and intense emission in a wide wavelength range with a much longer excited-state lifetime.[28,29]

However, the Cu-doping-related emission still suffers from the photostability problem, due to photooxidation.[30] The Cu-state-related emission is ascribed to the CB energy of host nanocrystals, arising from charge-transfer recombination of the e–CB → Cu2+, existing between Cu-localized holes and delocalized CB-like electrons.[31] Additionally, the large surface-to-volume ratio of the as-synthesized CQDs leads to the appearance of surface traps/defects, resulting in added opportunities for electronic defect formation. Shell-assisted defect passivation for Cu-doped CQDs is a proven effective approach to combat these losses. The overgrowth of a shell leads to surface passivation, heterojunction formation (e.g., type I heterojunction, Ohmic junction, and Schottkey junction), and photostability.[32,33] In core–shell CQD systems, the inorganic shell can reduce surface-related traps/defects by passivating the CQD surface. Ideally, the defect passivation has been carried out by a combination of doped CQDs and core/shell CQDs. The band structure engineering and defect passivation strategy will build the effective transmission path of the photoinduced carrier, which could have the potential to improve the CQD characteristics.

Here, we report a grown-doping strategy (Figure d) to design and synthesize a less toxic I–III–VI2 group AgInS2 (AIS) colloidal quantum dot with ZnS shell-assisted defect passivation. The insertion of a Cu dopant with high diffusivity in AIS CQD lattices regulates the energy band structure and photoluminescence spectra of CuAg1–InS2/ZnS CQDs with type|core/shell heterojunction structure. The best substitution ratio is x = 0.5. The Cu0.5Ag0.5InS2/ZnS CQDs show the optimized band alignment, tunable spectral region, and increased charge carrier lifetime. Moreover, the charge separation in the CQD/TiO2 interface is efficiently enhanced by forming the heterojunction interface hindering the charge recombination. As expected, the PEC cells based on Cu0.5Ag0.5InS2/ZnS CQDs exhibit a saturated photocurrent density of 5.8 mA cm–2 with good stability toward solar-driven H2 production. This work opens a new avenue for heavy metal-free CuAg1–InS2/ZnS CQDs as efficient photoelectrocatalysts to hydrogen generation.

Results and Discussion

The AgInS2 and AgInS2/ZnS CQDs (Figure e) were synthesized at different reaction temperatures, showing the tunable photoluminescence emission colors (Figure S1b and Figure S1c). After ZnS shell growth, AgInS2/ZnS CQDs exhibit longer photoluminescence lifetimes compared with that of AgInS2 CQDs (Figure f and Table S1). This indicates that ZnS shell-assisted surface passivation can reduce the defect-state concentration and eliminate some surface-related nonradiative centers, which are beneficial for photocatalytic applications.[34] Among all AgInS2-based CQDs, AgInS2/ZnS CQDs synthesized at 180 °C show the best photoactivity (Figure S1d), and hence AgInS2/ZnS CQDs (180 °C) were chosen as the model catalyst for study in this work. More details of the CQD synthesis can be seen in the Experimental Section of the Supporting Information.

The core/shell CuAg1–InS2/ZnS CQDs with different Cu/Ag molar ratios (x = 0, 0.25, 0.5, 0.75, and 1) were synthesized at 180 °C by a hot-injection method. The Cu-to-Ag atomic ratio was supported by an inductively coupled plasma–optical emission measurement (Table S2). For clarity, the AgInS2/ZnS, Cu0.25Ag0.75InS2/ZnS, Cu0.5Ag0.5InS2/ZnS, Cu0.75Ag0.25InS2/ZnS, and CuInS2/ZnS are denoted as Cu0Ag1, Cu0.25Ag0.75, Cu0.5Ag0.5, Cu0.75Ag0.25, and Cu1Ag0 hereafter, respectively. The tunable dopant Cu(I) ions are doped into the AgInS2 structure during the growth of host core CQDs with Cu atoms instead of Ag atoms (Figure a). As shown in Figure b, all CuAg1–InS2/ZnS samples exhibit the diffraction peaks of X-ray diffraction (XRD), consistent with a tetragonal structure of AgInS2 (AIS, see JCPDS database file 25-1330)[35] and CuInS2 (CIS, see JCPDS database file 47-1372),[36] implying the crystal structure was not changed after doping Cu into the CuAg1–InS2/ZnS CQDs (Figure S1a). Moreover, carefully calibrated XRD spectra (right image of Figure b) suggest the increasing Cu concentration makes a red-shift of the diffraction peaks to larger angles. This is because the ionic radius of Cu (74 pm) is smaller than that of Ag (114 pm), leading to a lattice contraction after the Cu doping. Furthermore, the chemical compositions of CuAg1–InS2/ZnS-alloyed CQDs were investigated by X-ray photoelectron spectroscopy (XPS). Figure c displays high-resolution scans of Cu 2p, showing two major fitting peaks at 930.6 and 951.3 eV corresponding to Cu 2p3/2 and Cu 2p1/2 for Cu+, respectively. In Figure d, the binding energies at 367.6 and 373.7 eV are assigned to Ag 3d5/2 and Ag 3d3/2 for Ag+ ions, respectively. In addition, the typical In 3d, Zn 2p, and S 2p level spectra are shown in Figure S2, indicating that the normal valence states for CuAg1–InS2/ZnS are 3d5/2 3d3/2 for In, 2p3/2 2p1/2 for S, and 2p3/2 2p1/2 for Zn.[37]

Figure 2

Structural characterization of CuAg1–InS2/ZnS CQDs: (a) Schematic presentation of AgInS2 chalcopyrite structure. (b) XRD patterns of CuAg1–InS2/ZnS CQDs (x = 0, 0.25, 0.5, 0.75, and 1). The zoomed-in image on the right side shows the AgInS2 (112) and CuInS2 (112) diffraction peaks. Binding energy of (c) Cu 2p and (d) Ag 2d in CuAg1–InS2/ZnS CQDs (x = 0, 0.5, and 1). TEM images of (e) Cu0Ag1 CQDs and (f) Cu0.5Ag0.5 CQDs. (g) STEM image, (h) HRTEM image, and (k) selected-area electron diffraction pattern of Cu0.5Ag0.5 CQDs. (i) Calculated FFTs of HRTEM images from the yellow boxed region in (h). (j) Lattice averaged image from the yellow area in (h) reveals the atomic arrangement along the [000] direction. The atomic arrangement fits well with the chalcopyrite AgInS2 crystal structure. Here and in (j), colors represent the following: red, copper; gray, silver; blue, indium; yellow, sulfur.

The CuAg1–InS2/ZnS-alloyed CQDs were further characterized by transmission electron microscopy (TEM). In Figure e and 2f, the average dimeter of the as-prepared Cu0Ag1 and Cu0.5Ag0.5 CQDs was about 8.6 and 8.8 nm, respectively (Figure S3), suggesting the nanoparticle size remained almost constant with the variation of the Cu doping ratio. Furthermore, scanning transmission electron microscopy (STEM) analysis confirms the uniform size distribution of Cu0.5Ag0.5 CQDs (Figure g). The high-resolution transmission electron microscopy (HR-TEM) images further show that the Cu0.5Ag0.5 CQDs were largely single crystalline in nature, as shown in Figure h. The interplanar distance of 3.34 Å corresponds to the (112) plane for typical chalcopyrite AgInS2. The Cu0.5Ag0.5 CQDs were imaged with one crystallographic direction, viz., [000], indicated by the yellow area. When fast Fourier transformation (FFT) is calculated from the area, the characteristic lattice of chalcopyrite AgInS2 is directly evident in Figure i. The clear atomic arrangement fits well with the AgInS2 crystal structure in the direction, as can be seen in the models depicted in Figure j, confirming the chalcopyrite crystal structure of the Cu0.5Ag0.5 CQDs. In addition, electron diffraction circles from the (112) and (204) crystallographic planes were observed in Figure k, showing the intensity profile of the diffraction rings corresponding to the XRD data.

To investigate the potential of these Cu-doped CQDs for PEC H2 production, the CuAg1–InS2/ZnS CQD-sensitized TiO2 photoanodes (Figure a) were prepared by depositing CQDs into the mesoporous TiO2 (m-TiO2) film substrates by the EPD method (detailed information in the Supporting Information).[38] The schematic diagram of glass|FTO|TiO2|CuAg1–InS2/ZnS CQD-sensitized photoanodes is displayed in Figure b. The corresponding Cu0.5Ag0.5 CQD-sensitized TiO2 photoanode has a 24.1 μm thickness of the m-TiO2 layer (Figure c). The PEC performance of the CuAg1–InS2/ZnS CQD-sensitized photoanodes was measured in the dark and under simulated solar illumination (AM 1.5 G, 100 mW cm–2). The measurements were performed using the architecture of a standard three-electrode configuration with a working electrode (as-fabricated CQD-sensitized photoanode), Pt counter electrode, and Ag/AgCl reference electrode (Figure d and 3f). As shown in Figure g and Figure S4, the photocurrent density (J) of the Cu0Ag1 and Cu1Ag0 CQD photoanodes is about 3.6 and 3.4 mA cm–2 at 0.9 VRHE (reversible hydrogen electrode; detailed information in the Supporting Information), respectively, while the bare TiO2 photoanode shows a very low saturated J = 0.38 mA cm–2 (Figure S5). The results confirm that high hydrogen production activity with the enhanced photocurrent density is mainly ascribed to the CQD photoelectrocatalysts deposited on the TiO2 film. For comparison, the saturated photocurrent density of the Cu0.5Ag0.5 CQD photoanode-based PEC is improved to 5.8 mA cm–2, showing the 1.7 times enhancement compared to that of pristine CQDs because of the Cu doping. In addition, the corresponding calculated H2 generation rate (see Figure S6 in the Supporting Information) is up to 67.4 μmol cm–2 h–1 for Cu0.5Ag0.5 CQDs from 36.4 μmol cm–2 h–1 for Cu0Ag1 CQDs, which is obtained by using the measured photocurrent as a reference.[39] These results demonstrate that Cu doping in AIS CQDs can effectively improve the H2 generation performance of the CQD-sensitized system.

Figure 3

PEC properties of CuAg1–InS2/ZnS CQD-sensitized TiO2 photoanodes: (a) photograph, (b) illustrative schematic, and (c) cross-sectional SEM image of the glass|FTO|TiO2|CQD-sensitized photoanode. (d) Photograph and (f) scheme of CQD-sensitized PEC cells. (e) Photograph of the Pt counter electrode. (g) Photocurrent measurement with linear sweep voltammetry for CQD-sensitized PEC cells in the dark and under AM 1.5 G illumination at 100 mW cm–2. (h) Photocurrent density–bias potential dependence and (i) photocurrent density–time dependence (at 0.5 V) of CQD-sensitized PEC cells in the chopped AM 1.5G irradiation. (j) Normalized steady-state photocurrent density as a function of time of CQD-sensitized PEC cells at 0.5 V (vs RHE) under AM 1.5 G illumination at 100 mW cm–2.

Meanwhile, Figure h shows the photocurrent density versus the applied bias (vs RHE) for the Cu0Ag1 and Cu0.5Ag0.5 CQD PEC cells in the S2–/SO32–electrolyte during the on–off illumination cycles. Significantly, a spike in the photocurrent density of Cu0.5Ag0.5 CQDs is more obvious tha that of pristine CQDs in Figure h and Figure S4. It is mainly attributed to good charge separation and fast charge transport due to the introduction of Cu doping. Furthermore, the photocurrent density versus reaction time of CQDs based on PEC cells at 0.5 VRHE upon standard on/off irradiation is displayed in Figure i. It can be seen that the obvious enhancement in photocurrent density of the Cu0.5Ag0.5 CQD PEC cell is obtained compared to that of the Cu0Ag1 CQDs. Higher photocurrent density indicates that more photogenerated electrons can be transferred to the counter electrode derived from the CQD photoanode. The stability measurements of the PEC cell based on CuAg1–InS2/ZnS CQD photoanodes were investigated at 0.5 V versus RHE under AM 1.5 G solar illumination (100 mW cm–2). In Figure j, the photocurrent density of Cu0Ag1 CQD-based PEC cells rapidly decreases during illumination, showing a large degradation (≈52%) of its initial value after 4000 s. However, 60% of the initial value for the Cu0.5Ag0.5 CQD photoanodes can be maintained after 4000 s solar illumination, and a drop equals to 50% after 10 000 s, exhibiting the improvement in chemical- and photostability in this CuAg1–InS2/ZnS CQD-sensitized photoanode for PEC cells due to Cu doping.

To investigate the photophysical origins of the enhanced photoelectrocatalytic performance with different Cu doping concentrations, the optical properties of the CuAg1–InS2/ZnS CQD contents were studied. In Figure a, PL peak position first red-shifted and subsequently blue-shifted with a peak at 750 nm as the Cu doping content increases. It is in accord with the trend of an absorption edge shift with or without a ZnS shell (Figure S7). The corresponding spectral data are displayed in the Supporting Information Table S3. The optical bandgaps (Eg) of Cu0Ag1, Cu0.25Ag0.75, Cu0.5Ag0.5, Cu0.75Ag0.25, and Cu1Ag0 are calculated to be 2.46, 2.22, 1.91, 2.08, and 2.17 eV, respectively, according to the classical Tauc method (Figure b). It finds that the Eg first augments and subsequently decreases with a minimum value of 1.91 eV for Cu0.5Ag0.5 CQDs (Figure c). The observation reveals that the Cu ions are successfully doped into the AgInS2 CQDs to produce light absorption and PL emission and change the lattice energy band. This result provides evidence and support to previous reports of the extent of the Cu-to-Ag ratio in the optical property alteration by the lattice structure compatibility theory.[40] It also confirms the potential of Cu doping engineering to enhance the optical property and optimize the band structure of the AgInS2 CQDs.

Figure 4

Optical properties and band structure of CuAg1–InS2/ZnS CQDs: (a) UV–vis absorption spectra and photoluminescence (PL) spectra and (b) Tauc plots of UV–vis absorption spectra for CuAg1–InS2/ZnS CQDs (x = 0, 0.25, 0.5, 0.75, and 1). The black lines represent the extrapolation of the linear portion of the absorption edges, α, absorption coefficient; h, Planck’s constant; and ν, photon frequency. (c) PL peak position and Eg versus Cu doping content. (d) Valence band (VB) XPS spectra and (e) the corresponding energy band diagram for CuAg1–InS2/ZnS CQDs (x = 0, 0.5, and 1).

To further characterize the electronic states and band structures of CuAg1–InS2/ZnS CQDs with different Cu doping concentrations, the valence band positions of Cu0Ag1, Cu0.5Ag0.5, and Cu1Ag0 were measured by the secondary electron cutoff and valence band spectra from He I excitation energy (Figure d), which are calculated to be 1.43, 1.33, and 1.41 eV, respectively. Based on the measured band energies, the band alignment diagram of Cu0Ag1, Cu0.5Ag0.5, and Cu1Ag0 CQDs is shown in Figure e, demonstrating that the conduction band (ECB) potentials vs NHE are calculated as −1.03, −0.58, and −0.76 eV, respectively. Compared to the pristine AIS or CIS CQDs, both the conduction band (CB) and valence band (VB) of the Cu-doped CQD are found to shift negatively. This finding further confirms that the band structure can been designed by tailoring the Cu dopant concentration, narrowing the bandgap and leading to an obvious enhancement of visible-light absorption. This also demonstrates that the tunable band structure is expected to increase the charge transfer efficiency by drift of the photogenerated electrons to the CQDs/TiO2 interface and of the photogenerated holes to the CQDs/electrolyte interface.[41]

Thus, we further investigated the effect of Cu doping in the AgInS2 atomic electronic structure by density functional theory (DFT) calculations. Figure a shows the pristine atomic structure of Cu0Ag1 with space group I-42d (122) with the chalcopyrite lattice. The unit cell contains 4 Ag, 4 In, and 8 S atoms, containing the AgS4 and InS4 tetrahedra. In Figure b, the doped structure of Cu0Ag1 has a direct band gap energy of 1.99 eV. After Cu doping, the calculated band gap energy of Cu0.5Ag0.5 is decreased to 1.77 eV, which is close to the experimental value. Moreover, the calculated total density of states (TDOS) spectra show that the valence band maximum (VBM) of Cu0Ag1 consists of the Ag and S atom hybrid orbitals, and the conduction band minimum (CBM) is mainly attributed to the In and S atom hybrid orbitals. It is obvious that Cu doping shifts the VBM of Cu0.5Ag0.5 upward and narrows the band gap because of the participation of the Cu atom orbitals (Figure c). It could be the reason that the d orbital of the Cu atoms possesses slightly higher energy than that of the Ag d orbital.[42] Therefore, the Cu doping can tailor both the band gap energy and band edge position of AIS CQDs. These results suggest that the Cu doping effect is able to create the efficient transportation and separation of the photogenerated charge for CQDs/TiO2 heterojunction photoanodes for photocatalytic hydrogen generation in the PEC cell.[43]

Figure 5

DFT calculations: (a) Atomic structures of Cu0Ag1 and Cu0.5Ag0.5 used in DFT calculations. The energy band structure and density of states (DOS) of (b) Cu0Ag1 and (c) Cu0.5Ag0.5 calculated with the PBE0 level, where the energy level was aligned with respect to the vacuum level (Evac).

To gain insight into the separation efficiency of photogenerated electrons and holes, time-resolved photoluminescence (PL) spectra of Cu0Ag1, Cu0.5Ag0.5, and Cu1Ag0 CQDs were displayed in Figure b under 350 nm excitation, and the corresponding data were fitted by a biexponential model[44] (see eqs S1–S2 in the Supporting Information). Compared to pure CQDs, the PL decay time of the Cu0.5Ag0.5 CQDs increases upon doping with Cu content of 50% (Table S1), now having a prolonged lifetime of 397.9 ns. This extension of carrier lifetimes demonstrates the effective spatial separation of charge carriers, which are beneficial for photoelectrocatalytic application in the CQD system.[45] In fact, the longer lifetime is attributed to the exciton relaxation derived from the host CQD systems via the additional Cu T2 energy states.[27,29]

Figure 6

Photoluminescence properties of CuAg1–InS2/ZnS CQDs: (a) Schematic diagram of the energy gap structure of ZrO2/TiO2/CQD/Elec, showing the charge-transfer process. (b) Photoluminescence decay curves of CuAg1–InS2/ZnS CQDs (x = 0, 0.5, and 1). Photoluminescence decay curves of (c) Cu0.5Ag0.5 and (d) Cu0Ag1 CQDs on TiO2, ZrO2, and ZrO2 wetted with the electrolyte (ZrO2 + Elec).

To further investigate the electron transfer rate (Ket) and hole transfer rate (Kht), the transient PL spectra of CuAg1–InS2/ZnS CQDs deposited on TiO2 or ZrO2 films with or without dipping the S2–/SO32– electrolyte are shown in Figure c,d. The Ket or Kht is calculated by using the following eq (43,46,47)where τQDs/TiO2 and τQDs/ZrO2 are the PL lifetime of CuAg1–InS2/ZnS CQDs deposited on TiO2 and ZrO2 substrates, respectively, indicating the efficiency of the photogenerated carrier separation and transport,[48] and τQDs/ZrO2/Elec represents the PL lifetime of CQDs deposited on ZrO2 with the presence of the electrolyte (Figure a). The carrier transfer rates are reported in Table S4. Generally, the electrons or holes transfer from CQDs to TiO2 or ZrO2 with a hole scavenger with a suitable energy level. The Ket value is 1.9 times larger than Kht, suggesting a more efficient electron transfer from Cu0.5Ag0.5 CQDs toward that of TiO2, compared to that of Cu0Ag1 CQDs (1.2 times). These results indicate that an efficient photoexcited electron and hole transfer can be achieved by controlling the Cu doping, confirming the enhanced performance of Cu-doping CQDs in the PEC solar H2 generation.

Conclusions

In summary, heavy metal-free CuAg1–InS2/ZnS CQDs with different Cu doping profiles were prepared by in situ growth with a hot-injection method. The effects of Cu doping on light absorption and band energetics of AgInS2 CQDs were studied with a ZnS shell-assisted defect passivation strategy. It is found that the Cu doping obviously shifts the band edge positions of AgInS2 CQDs and optimizes the optical properties. The Cu doping improves the efficiency of the photogenerated charge separation by tailoring a suitable energy band structure. Based on these findings, the Cu0.5Ag0.5InS2/ZnS CQD-sensitized TiO2 photoanode for the PEC cell shows an excellent photocurrent density of 5.8 mA cm–2. In addition, the doping effects and shell defect passivation also lead to an improvement in stability with 50% surplus of the photocurrent density for 10 000 s. These results demonstrate that the use of this designed Cu doping strategy with defect passivation can be used as a guideline for improving the photoelectrocatalytic activities of heavy metal-free I–III–VI CQD light absorbers for PEC hydrogen generation.

Experimental Section

Materials

Copper iodide (CuI, 99.999%), silver nitrate (AgNO3, 99.99%), indium acetate (In(Ac)3, 99.99%), sulfur (S, 99%), zinc acetate dehydrate (Zn(Ac)2·2H2O), 1-octadecence (ODE, 99.6%), 1-dodecanethiol (DDT, 96%), trioctylphospine (TOP, 90%), oleylamine (OLA, 70%), sodium sulfite (Na2SO3), sodium sulfide nonahydrate (Na2S·9H2O), acetone (>99%), methanol (99.93%), and ZrO2 nanoparticle powder (≈100 nm sized) were bought from Sigma. Ti-nanoxide was purchased from Solaronix. Titania (TiO2) paste (18 NR-AO, 20–450 nm sized) was obtained from Dyesol (Queanbeyan, Australia). The FTO/glass substrates were bought from XOP Glass Co., Ltd. All chemicals were used without further purification.

Synthesis of CuAg1–InS2/ZnS CQDs

The synthesis of CuAg1–InS2/ZnS CQDs with tunable molar ratio of Ag/Cu followed the reference with slight modifications.[34] The CuAg1–InS2/ZnS CQD was synthesized via a hot-injected method using different Ag and Cu precursors. Typically, 0.051 g of AgNO3 and 0.09 g of In(Ac)3 were mixed with 24 mL of ODE in a flask and then purged by N2 gas for 30 min. Subsequently, the temperature was raised to 100 °C. Then 3 mL of DDT was added in the flask and heated to different reaction temperature (130, 180, and 210 °C, respectively). Then, 0.02 g of S and 1.3 mL of OLA mixed solution were quickly injected into the flask with vigorous stirring for 30 min. For ZnS shell in situ growth, 0.762 g of Zn(Ac)2, 0.0384 g of S, and 6 mL of TOP mixed clear solution were quickly injected into the above reaction solution and then kept at the corresponding temperature for 2 h. After the reaction finished, the anhydrous ethanol was added into the reaction solution, and the obtained suspension was centrifuged. Finally, the as-synthesized CuAg1–InS2/ZnS CQDs were dispersed in toluene. The additive amount of CuI is the 0, 25, 50, 75, and 100% the molar weight of AgNO3, respectively. The same process was used to synthesis a series of the CuAg1–InS2/ZnS CQDs.

Preparation of TiO2 and ZrO2 Films

The compact TiO2 layers were deposited on clear FTO/glass substrates by spinning TiO2 precursor solution (Ti-Nanoxide) at 5000 r.p.m. for 30 s and then sintered at 500 °C for 30 min. Subsequently, the mesoporous TiO2 layers were deposited on the above-mentioned substrates by blade-coating 18 NR-AO TiO2 paste. Finally, the films were sintered at 500 °C for 30 min. For the mesoporous ZrO2 films, the same blade-coating process and thermal treatment procedure were performed by using the commercial ZrO2 paste.

Fabrication of CQD-Based Photoelectrochemical Cells

The CQD-based photoelectrochemical (PEC) cells were constructed by a typical three-electrode configuration, consisting of a CQD-TiO2 electrode, a Pt counter electrode, a saturated Ag/AgCl reference electrode, and the electrolyte. The CuAg1–InS2 and CuAg1–InS2/ZnS CQDs were deposited on a TiO2/FTO/glass substrate with a current bias of 100 V for 2 h in CQD toluene solution according to the electrophoretic deposition method.[49] Subsequently, the surface of the as-prepared CQDs-TiO2 electrodes was protected by depositing a ZnS capping layer via the SILAR deposition cycle[47] to complete PEC cell fabrication. The active area of the photoanode ranges is 0.15 ± 0.05 cm2.

Characterization

X-ray diffraction of the CQD samples was carried out with a Philips X’Pert PRO X-ray diffractometer. X-ray photoelectron spectroscopy measurement was performed on a VG Escalab 220i-XL electron spectrometer by using a Twin Anode X-ray Source. Inductively coupled plasma-optical emission measurements were carried out with an Agilent 5100 ICP-OES system. Transmission electron microscopy images of the CQD samples were recorded using a JEOL ARM200CF TEM/STEM at an accelerating voltage of 200 kV. UV–vis spectra were performed on a Shimadzu UV-2600 spectrophotometer. Steady-state photoluminescence measurements were performed by using the Horiba Jobin Yvon. Time-resolved photoluminescence measurements were taken by a Pico-Quant FmbH fluorescence lifetime spectrometer. The morphology images of cross-sectional CQD-TiO2 electrodes were obtained with a JSM-7401F scanning electron microscope. Current–potential measurements of the CQD-based photoelectrochemical cells were performed in a potential window of −0.6 V to +0.9 V versus the reversible hydrogen electrode (RHE) (VRHE = VAg/AgCl + 0.1976 + 0.059 × pH) by using simulated one-sun illumination (AM 1.5G, 100 mW cm–2) with a SLB-300A solar simulator with a CHI-760D electrochemical workstation.

Theoretical Method

Structure optimization and electronic structure calculations were recorded by using density functional theory (DFT) with the projector-augmented wave (PAW) method,[50] as implemented in the Vienna Ab initio Simulation Package (VASP).[51,52] A cutoff energy of 500 eV and an 8 × 8 × 8 Monkhorst–Pack k-point mesh were used in all calculations. All structures are fully optimized with the maximum atomic force below 0.01 eV Å–1, and the convergence criterion of electronic energy was set to 10–5 eV. The Perdew–Burke–Ernzerhof (PBE)[53] and hybrid PBE0[54] exchange-correlation functional were used for structure optimization and electronic structure calculation, respectively. Band structures within the hybrid functional scheme were obtained for efficiency by using the smoothed Fourier interpolation which was implemented in the BoltzTraP2 code.[55] For band alignment, the energy levels, including the valence band maximum (VBM), conduction band minimum (CBM), and Fermi level, are referenced to the vacuum level. To obtain the vacuum level of the bulk system, the slab models with a vacuum layer of 20 Å were constructed using the 1 × 1 × 5 supercell of conventional cells. Then the value of the energy level with respect to the vacuum level can be defined uniformly as E = (Ebulk – Erefbulk) – (Evacslab – Erefslab,far), where Ebulk is the energy level of the bulk system; Evacslab is the vacuum level of the slab system; and Erefbulk and Erefslab,far are the macroscopically averaged electrostatic potential in the bulk system and the region far away from the surface of the slab system, respectively.