1. Introduction
Perovskite nanocrystals (NCs) exhibit significant promise as a photocatalytic material due to several scientific advantages, including high efficiency, tunable bandgap, and low-cost production [
1,
2,
3]. Specifically, perovskite NCs’ high light absorption coefficients and excellent charge carrier mobility have resulted in high photoconversion efficiencies, rendering them a highly attractive candidate for the development of next-generation solar energy conversion and environmental remediation technologies [
4]. Furthermore, the tunable bandgap of perovskite NCs allows for the facile modification of their electronic properties via compositional changes, enabling the development of perovskite-based catalysts with tailored properties for specific applications. Moreover, perovskite materials are relatively low-cost and amenable to large-scale production via solution-based methods, thereby making them highly accessible for practical applications. However, the primary challenge associated with perovskite pertains to its chemical stability [
5]. As a result of its ionic binding character, perovskite NCs are highly susceptible to ion migration under polar media, which leads to low chemical stability under water and deterioration of their outstanding optical properties. The incorporation of a shell structure into perovskite NCs can potentially mitigate these challenges, enhancing their chemical and operation stability [
6].
In recent times, core-shell perovskite NCs, including CsPbBr
3/organic shell and CsPbBr
3/inorganic shell, have been developed to enhance the chemical stability of the perovskite structure. Although CsPbBr
3/organic shells are relatively easy to fabricate, the core and shell composition is primarily bonded through physiosorption. However, this physiosorbed moiety can be readily separated due to surrounding pH, which may result in perovskite structure degradation or co-catalyst poisoning that could affect the low catalytic efficiency of perovskite-based catalysts [
7]. Alternatively, perovskite/inorganic shells, such as CsPbBr
3/SiO
2, CsPbBr
3/Al
2O
3, and CsPbBr
3/ZnS, have also been developed [
8,
9,
10]. Among these, CsPbBr
3/SiO
2 NCs have garnered considerable attention as photonic materials under water, particularly for bio-imaging, photocatalyst, and photoelectrocheical catalyst. Song et al. demonstrated that CsPbBr
3/SiO
2 core-shell NCs fabricated using 3-aminopropyl-triethoxysilane (APTES) exhibited improved water resistance compared to pristine perovskite NCs [
11]. This encapsulation strategy enhances the chemical stability of perovskite NCs while maintaining their consistent optical properties.
In catalyst applications, perovskite nanoparticles undergo repeated absorption and desorption events, leading to increased susceptibility to poisoning and degradation compared to their use in bio-imaging applications [
12]. The stability of the perovskite structure can be impacted by various factors, including exposure to polar molecules, such as oxygen and moisture from ambient air, phase transitions, thermal stress, and illumination. The presence of surface defects resulting from strong absorption and/or desorption during redox reactions can facilitate ion migration in the presence of polar molecules. Moreover, phase transitions, thermal stress, and illumination can exacerbate defect generation and lead to defect proliferation [
13]. Therefore, in prior studies, core-shell perovskite NCs were developed for use as photocatalysts or photoelectrochemical catalysts; and were primarily utilized in organic solvents such as acetonitrile, dichloromethane, alcohols, and halo acids due to their ability to stabilize the PbBr
6 octahedral structure of perovskite. Although some research has demonstrated the efficacy of these NCs in aqueous solutions, these studies were typically conducted under neutral pH conditions.
Interestingly, the exceptional reduction ability of perovskite, due to the position of its CB, which is typically negative enough for H
2 generation or CO
2 reduction, prompted further examination of their photocatalytic behavior in water-based reactions. For instance, the Co
2%@CsPbBr
3/Cs
4PbBr
6 were developed as a photocatalyst for CO
2 reduction in water [
14]. The reaction effectively converted carbon dioxide to carbon monoxide with a conversion rate of ~11.95 μmol/g/h under 300 W Xe lamp, and the reaction remained stable for 20 h. However, only a few studies have demonstrated the photoelectrochemical properties of perovskite. Li et al. developed a CsPbBr
3/a-TiO
2 core/shell heterostructure to enhance the photocurrent and stabilize the crystal structure of the inner CsPbBr
3 [
15]. The material was able to sustain 8 h in a 0.1 M Na
2SO
4 aqueous solution during the photoelectrochemical reduction. Compared to pristine CsPbBr
3 NCs, the core/shell CsPbBr
3/a-TiO
2 NCs exhibited a higher photocurrent, but still showed a low value of 10
−6 A under illumination, which is considerably lower than that of other photoelectrochemical catalysts. This can be attributed to the low conductivity of the shell material of core-shell perovskite NCs. While a thick shell component can effectively prevent the penetration of water, resulting in high chemical stability, the insulating behavior of the shell severely impede the transport of carriers to active sites [
16]. Thus, perovskite NCs with a controlled shell thickness, low-water-penetration rate, and high photoluminescence quantum yield (PLQY) are necessary to realize an efficient photoelectrochemical catalyst.
In this study, we developed two-time cross-linkable core-shell perovskite NCs using two types of silanes to achieve both excellent chemical stability and high conductivity. The vinyl-functional group is a widely used functional group that can polymerize with the help of a photo initiator under illumination, leading to C-C coupling under mild conditions. Along with silyl ether functional group as the primary cross-linking site, the vinyl-functional group was chosen as a secondary cross-linking site. The first cross-linking reaction was spontaneously initiated in the solution state during the purification process, forming an ultrathin Si-O-Si amorphous matrix on the surface of the perovskite NCs. The second cross-linking reaction was intentionally induced in the film state by exposing it to UV light. Due to the two-step cross-linking reaction (i.e., solution-state cross-linking and subsequent film-state cross-linking), CsPbBr3/SiO2 NCs with high packing density were generated, resulting in excellent chemical stability of core-shell perovskite NCs. The high packing of CsPbBr3/SiO2 NCs facilitates the carrier transport behavior between NCs, resulting in a high photocurrent of over 10−5 A observed for our cross-linked CsPbBr3/SiO2 NCs film. Finally, several types of cross-linked B-site doped CsPbBr3/SiO2 NCs films were fabricated, and Ag-doped CsPbBr3/SiO2 NCs demonstrated the lowest onset potential (−190 mV at 100 µA/cm2) for HER in a 0.5 M H2SO4 aqueous solution due to effective band modulation.
2. Results and Discussion
Our research group previously developed core-shell CsPbBr
3/SiO
2 NCs using a one-pot hot-injection method with APTES, which demonstrated exceptional chemical stability despite the relatively thin shell thickness of approximately 2–3 nm [
16,
17]. APTES was utilized as the precursor for the amorphous SiO
2 shell formation. Following the hot injection process, the perovskite NCs were subjected to methyl acetate during the purification step. The silyl ether functional groups in APTES underwent hydrolysis and condensation with the aid of methyl acetate, leading to the formation of a Si-O-Si cross-linked matrix on the surface of the perovskite NCs [
18]. As a result, the synthesized perovskite NCs exhibited a core-shell composition, consisting of a perovskite core and an amorphous SiO
2 shell covered with surface ligands such as residual silane, oleic acid, and oleylamine. The core-shell CsPbBr
3/SiO
2 NCs with a 2–3 nm shell thickness demonstrated relatively stable features in various polar solvents, as we have previously reported, although their stability was insufficient for survival in water. In order to enhance the resistance of the perovskite NCs to water, it was necessary to increase the thickness of the shell. To achieve this, we added twice amount of APTES during the synthesis process, which showed 5–20 nm shell thickness. However, this led to a change in the crystal structure of the perovskite. As shown in
Figure 1a, the XRD pattern of the core-shell CsPbBr
3/SiO
2 NCs synthesized with a low amount of APTES displayed dominant peaks at ~15.0°, ~21.4°, ~26.5°, and ~31.1°, which were assigned to the (100), (110), (111), and (200) crystal planes, respectively, for cubic Pm-3m symmetry [
19]. On the other hand, the core-shell CsPbBr
3/SiO
2 NCs synthesized with a high amount of APTES showed dominant peaks at ~15.0°, ~21.4°, ~26.5°, ~30.5°, and ~34.0°, assigned to the (101), (121), (022), (202), and (222) crystal planes, respectively, for orthorhombic Pnma symmetry. This thermodynamically favorable phase change from the cubic to orthorhombic phase was due to the strong bonding of the amine functional group in APTES on the surface of perovskite NCs. In the case of core-shell CsPbBr
3/SiO
2 NCs synthesized with a low amount of APTES, the majority of NCs formed a core-shell structure with each NC existing separately (
Figure 1b). However, core-shell CsPbBr
3/SiO
2 NCs synthesized with twice the amount of APTES exhibited an aggregated feature between NCs with an inhomogeneous thick shell (
Figure 1c,d). This aggregation could impede carrier transport during photoelectrochemical reactions.
To achieve both excellent chemical stability and high conductivity, we designed cross-linkable core-shell perovskite NCs using two different silane molecules (
Figure 2a). These two silane moieties played different roles: one for amorphous shell formation and the other as a cross-linking site in a film state. Unlike APTES, vinyltrimethoxysilane lacks a specific functional group that strongly interacts with the surface of perovskite NCs, allowing it to move more freely near the surface. As a result, the relatively small size of vinyltrimethoxysilane effectively filled in the voids of the Si-O-Si cross-linked matrix generated by APTES during the first cross-linking reaction. Thus, CsPbBr
3/SiO
2 NCs synthesized using APTES/vinyltrimethoxysilane exhibited a well-packed Si-O-Si matrix on the surface of perovskite NCs. The PLQY of core-shell CsPbBr
3/SiO
2 NCs was investigated as a function of the amount of vinyltrimethoxysilane used in the synthesis process, as the optical properties of perovskite NCs can be affected during functionalization reactions. The results showed that the highest PLQY was obtained with CsPbBr
3/SiO
2 NCs using 0.1 mL of vinyltrimethoxysilane, and the full width at half maximum (FWHM) of the emission peak also displayed a similar trend (
Figure 2b,c). We posit that a small amount of vinyl, which was insufficient to cover all the voids between the Si-O-Si matrix created by APTES, resulted in a low PLQY and large FWHM. Conversely, using a large amount of vinyl beyond 0.1 mL facilitated the polymerization reaction and led to the aggregation of core-shell perovskite NCs.
The chemical stability of CsPbBr
3/SiO
2 NCs was assessed using three distinct polar solvents. The PL spectra of CsPbBr
3/SiO
2 NCs in an aqueous system exhibited an enhancement of approximately 11% for up to 90 min, attributable to recrystallization facilitated by low water penetration (
Figure 3a). The chemical stability of CsPbBr
3/SiO
2 NCs was observed to have improved even before the second cross-linking reaction, as compared to the CsPbBr
3/SiO
2 NCs synthesized with solely APTES, which demonstrated PL loss after 60 min [
18]. This suggests that a well-packing shell was produced for the APTES/vinyltrimethoxysilane system. The PL spectra of CsPbBr
3/SiO
2 NCs in ethanol showed a similar trend to that observed in the aqueous system (
Figure 3b). The PL enhancement was sustained for up to 30 min, after which the PL proportionally decreased. The PL spectra of CsPbBr
3/SiO
2 NCs in IPA did not reveal a significant PL enhancement, but indicated no significant PL loss for up to 120 min. Most of the PL emission disappeared within 3 h in this system.
In this experimental setup, we deliberately induced a secondary cross-linking reaction at the film state by exposing it to UV light (1 W, 380–400 nm). The aim was to reduce the distance between the NCs and promote more efficient carrier transport. Since the activation of cross-linking reactions in vinyl-functional groups necessitates a high level of energy, we employed a photo initiator to expedite the reaction under mild conditions.
Figure 4a illustrates the molecular structure of the photo initiator employed. Under UV light, the photo initiator generates a radical which, in turn, breaks the double bond of the vinyl functional group. This results in C-C coupling with the double bond of oleic acid or oleylamine, culminating in the creation of a cross-linked network between the NCs [
20].
The second cross-linking reaction was evaluated through Fourier-transform infrared (FT-IR) analysis (
Figure 4b). The FT-IR spectrum of the as-coated film using CsPbBr
3/SiO
2 NCs without a photo initiator revealed prominent peaks at 1456 cm
−1, corresponding to aliphatic C-H bending or scissoring; and at ≈1022 cm
−1, corresponding to -C=C- stretching, arising from oleic acid, oleylamine, and vinyl functionalized CsPbBr
3/SiO
2 NCs. These peaks did not show significant changes before and after UV irradiation in the absence of a photo initiator. On the other hand, the as-coated film using CsPbBr
3/SiO
2 NCs with a photo initiator exhibited a new peak around 1718~1722 cm
−1, associated with the C=O stretching of the photo initiator for both before and after UV irradiation [
19]. After UV irradiation, the -C=C- stretching peak exhibited a decrease in intensity due to the facilitated UV cross-linking reaction by the photo initiator. We also analyzed the XRD pattern of the corresponding film used for FT-IR, which revealed peaks at ~15.0°, ~21.4°, ~26.5°, and ~31.1°, assigned to the (100), (110), (111), and (200) crystal planes, respectively, for cubic Pm-3m symmetry. This indicated that no phase transition occurred during the addition of the photo initiator and UV treatment, which was critical in enabling the realization of the secondary cross-linking reaction.
The TEM image further supported this result, showing that the core-shell CsPbBr
3/SiO
2 NCs had a high crystalline structure after both the 1st and 2nd cross-linking reactions. It is worth noting that the CsPbBr
3/SiO
2 NCs synthesized using the APTES/vinyltrimethoxysilane system still exhibited an ultrathin passivation layer after the 2nd cross-linking reaction, unlike the inhomogeneous aggregated feature of CsPbBr
3/SiO
2 NCs synthesized solely with APTES (
Figure 5a,b). As a result, the PL spectra of the film produced by CsPbBr
3/SiO
2 NCs showed PL enhancement until 20 min of UV exposure and only slightly decreased after 30 min of UV irradiation, but still exhibited higher PL intensity than that of the film before UV irradiation (
Figure 5c).
Finally, we investigated the photoelectrochemical properties of CsPbBr
3/SiO
2 NCs on the HER reaction using two different systems, namely sole APTES and APTES/vinyltrimethoxysilane. The sole APTES system exhibited low dark current around 10
−6 A and photocurrent around 10
−6 A (
Figure 6a), whereas the APTES/vinyltrimethoxysilane system demonstrated a significantly higher photocurrent (~4 × 10
−5 A) and onset potential effectively shifted lower (−219 mV at 10 µA/cm
2) due to the high packing of CsPbBr
3/SiO
2 NCs, facilitating carrier transport between NCs (
Figure 6b) and reducing the electron-hole recombination at the surface. The stability test results in
Figure 6c show that the current level of the cross-linked CsPbBr3/SiO
2 NCs film remained sustained for up to 2 h, while the bare NCs died within 5 min. To modulate the active site of perovskite materials based on reaction thermodynamics, we fabricated four types of cross-linked doped CsPbBr
3/SiO
2 NCs films, incorporating dopants during the synthesis process of the core CsPbBr
3. These dopants included Bi-doped CsPbBr
3/SiO
2 NCs, Cu-doped CsPbBr
3/SiO
2 NCs, Ni-doped CsPbBr
3/SiO
2 NCs, and Ag-doped CsPbBr
3/SiO
2 NCs. Among them, Ag-doped CsPbBr
3/SiO
2 NCs exhibited the lowest onset potential for HER (approximately −190 mV vs. the reversible hydrogen electrode (RHE) at 100 µA/cm
2), followed by −267 mV at 100 µA/cm
2 for Cu-doped CsPbBr
3/SiO
2 NCs, −312 mV at 100 µA/cm
2 for Ni-doped CsPbBr
3/SiO
2 NCs, and −503 mV at 50 µA/cm
2 for Bi-doped CsPbBr
3/SiO
2, respectively, in a 0.5 M H
2SO
4 aqueous solution (
Figure 6d). The effectiveness of Ag-doped CsPbBr
3/SiO
2 NCs in promoting the redox reaction was attributed to their provision of an effective active site.
To clarify the doping effect, we investigated the band state of each sample. The UV–Vis spectra of Ag-doped NCs displayed distinct phenomena compared with pristine NCs without UV treatment or other doped NCs (
Figure 7a). Pristine NCs and Cu-doped NCs exhibited nearly identical band edges, indicating a similar band gap. In contrast, Ni-doped NCs and Bi-doped NCs exhibited a blue-shifted band gap, while Ag-doped NCs exhibited a red-shifted band gap. Ultraviolet photoelectron spectroscopy (UPS) was conducted on all five samples, and the work function of each film was calculated using W
F = 21.2 eV (He I radiation) − E
Cut-off (
Figure 7b). The work function of pristine NCs was 3.6 eV, while Ni-doped NCs and Bi-doped NCs were 4.2 eV. Cu-doped NCs exhibited a work function of 4.3 eV, and Ag-doped NCs exhibited a work function of 4.4 eV. Ag-doped NCs demonstrated the closest work function towards the H
2O/H
2 reaction potential. Unlike other doped NCs, Ag-doped NCs displayed a relatively unclear band edge in the UV–Vis spectra and UPS analysis, suggesting that Ag doping induced mid-gap states, which facilitates carrier transport during the PEC performance. Consequently, the PL spectra of Ag-doped NCs exhibited a red-shifted emission wavelength with reduced PL intensity (
Figure 7c). In addition to the band modulation, the conductivity of Ag-doped NCs was also enhanced. The I-V curve of four different samples was measured (
Figure 7d), and all doped samples exhibited improved conductivity compared to pristine NCs without UV treatment. Notably, Ag-doped NCs displayed the highest current level among the four different doped NCs. Therefore, we believe that Ag-doped sample exhibited the lowest onset potential due to band modulation and its high conductivity.