CsPbBr3–CdS heterostructure: stabilizing perovskite nanocrystals for photocatalysis

The instability of cesium lead bromide (CsPbBr3) nanocrystals (NCs) in polar solvents has hampered their use in photocatalysis. We have now succeeded in synthesizing CsPbBr3–CdS heterostructures with improved stability and photocatalytic performance. While the CdS deposition provides solvent stability, the parent CsPbBr3 in the heterostructure harvests photons to generate charge carriers. This heterostructure exhibits longer emission lifetime (τave = 47 ns) than pristine CsPbBr3 (τave = 7 ns), indicating passivation of surface defects. We employed ethyl viologen (EV2+) as a probe molecule to elucidate excited state interactions and interfacial electron transfer of CsPbBr3–CdS NCs in toluene/ethanol mixed solvent. The electron transfer rate constant as obtained from transient absorption spectroscopy was 9.5 × 1010 s−1 and the quantum efficiency of ethyl viologen reduction (ΦEV+˙) was found to be 8.4% under visible light excitation. The Fermi level equilibration between CsPbBr3–CdS and EV2+/EV+˙ redox couple has allowed us to estimate the apparent conduction band energy of the heterostructure as −0.365 V vs. NHE. The insights into effective utilization of perovskite nanocrystals built around a quasi-type II heterostructures pave the way towards effective utilization in photocatalytic reduction and oxidation processes.


Introduction
Semiconductor quantum dots are excellent building blocks for designing light-harvesting assemblies. 1,2 The ability to chemically modify the surface with a functionalized ligand or to couple with another semiconductor particle offers a variety of ways to harvest visible photons. [3][4][5] Since the 1990s, metal chalcogenide quantum dots (QDs), CdSe in particular, have served as the prototypical compound to elucidate excited state and charge transfer properties. [6][7][8] In recent years, another quantum dot system, viz., perovskite nanocrystals (CsPbX 3 , X ¼ Cl, Br, I), has emerged as a model semiconductor QD system to probe light induced optoelectronic and photocatalytic properties. [9][10][11][12][13] We have recently elucidated the photocatalytic aspects of CsPbBr 3 QDs by probing the interfacial electron transfer to methyl viologen 14,15 and ferrocenium cation. 16 To date, the use of these perovskite quantum dots in photocatalysis has been limited only to a few nonpolar solvents. [17][18][19][20][21] The weakly-binding organic ligand shell around perovskite nanocrystals does not provide sufficient stability in polar solvents. In order to expand the scope of the perovskite nanocrystals to a wide range of photocatalytic applications (e.g., solar hydrogen production or CO 2 reduction), it is important to provide protection against chemical transformation in the presence of a redox couple or in a polar medium.
One simple approach to achieve stability in polar solvents is to cap the semiconductor nanocrystals with a thin inorganic shell. Design of such heterostructures has been successfully employed for binary and ternary semiconductors like CdSe/ ZnS, 22,23 InP/ZnS, 24 and AgInS 2 /ZnS. 25 The heterostructure with type I or type II band alignment offers strategies to enhance emission of the core QD or improve charge separation within the heterostructure. Although a few reports exist to-date of capping CsPbBr 3 QDs with SiO 2 , 26 CdS, 27 or ZnS 28,29 shells, none of these heterostructures have shown a major leap in achieving improved performance with a long-term stability in polar solvents. Ambiguity still exists whether the added material forms a continuous shell around the perovskite core or forms smaller discontinuous islands on the surface. 26 Given the difficulty in imaging the thin inorganic shell around perovskite nanocrystals, because of the image contrast, one employs stability tests in a polar medium or its resistance to halide exchange to conrm surface modication. [30][31][32][33] Designing perovskite heterostructures with metal chalcogenide shells can have several distinct advantages: (i) providing stability towards increased polarity of the solvent, (ii) remediating surface defects by directly interacting with the vacancies, and (iii) allowing for type I or quasi-type II band alignment to promote increased charge recombination in the core (increased emission yield) or improved charge separation. [34][35][36] In this context, a CsPbBr 3 -CdS heterostructure offers an attractive means to tune the band energies, as their conduction bands are nearly isoenergetic (E CB of CsPbBr 3 and quantized CdS z À0.8 V versus NHE) [37][38][39] and facilitate charge separation. In addition, cubic CsPbBr 3 nanocrystals have a lattice constant (a) of 5.85Å (ref. 40) while that of CdS (zinc blende structure) is 5.83Å. 41 The similarity of the two values signies the possibility of having a less strained interface with reduced defect states. 42 We have now successfully prepared CsPbBr 3 -CdS heterostructures in a two-step method, and the optical properties of these structures are discussed.

Results and discussion
CsPbBr 3 -CdS heterostructure CsPbBr 3 quantum dots (QDs) dispersed in octadecene (ODE) were prepared using a previously reported procedure. 43 These QDs were then treated with cadmium diethyldithiocarbamate (Cd(DDTC) 2 ) at 110 C to obtain CdS capped CsPbBr 3 QDs. Experimental details on the synthesis of CsPbBr 3 and CsPbBr 3 -CdS QDs are presented in the ESI (Fig. S1 †). The transmission electron microscopy (TEM) images of the two nanocrystals are shown in Fig. 1A and B. These cubic particles are similar in size showing particles of lengths 8-9 nm (see Fig. S2 † for size distribution analysis). This shows that CdS capping in the heterostructure is relatively thin compared to the CsPbBr 3 core and it is difficult to identify with the image contrast in TEM images. This observation is consistent with earlier work which reports difficulty in characterizing the shell in a CsPbBr 3 heterostructure using TEM analysis. 44,45 These studies have attributed the imaging difficulty to low electron density contrast of the shell. However, other techniques such as elemental analysis can be useful to overcome these limitations.
We succeeded in establishing the presence of CdS in the CsPbBr 3 -CdS heterostructures through elemental analysis with TEM energy dispersive X-ray (EDX) spectroscopy. Fig. 1C-G which present elemental mapping, conrm the presence of Cd and S along with Cs, Pb, and Br. The elemental ratio (Fig. 1H) also suggests a relatively low concentration of CdS as compared to Cs, Pb and Br in the heterostructure. We can conclude that any CdS in the heterostructure is of the order of a monolayer. However, there is also the possibility of forming small clusters in and around the CsPbBr 3 QDs.
Evidence of surface modication with CdS was also seen through the change in the surface charge. Zeta potential measurements indicated that CsPbBr 3 -CdS QDs suspended in toluene carry more negative surface charge (À37.4 mV) than pristine CsPbBr 3 nanocrystals in toluene (À15.8 mV). This increased surface charge of CsPbBr 3 -CdS QDs enabled us to carry out electrophoretic deposition of a lm under the inuence of a DC eld (see ESI † for details; Fig. S3 †). The increased surface negativity and ability to be deposited as a lm under applied bias indicates a modied surface around CsPbBr 3 . Similar electrophoretic deposition was also possible when CsPbBr 3 nanocrystals were coated with a PbSO 4 -oleate shell. 46 It should be noted that pristine CsPbBr 3 QDs suspended in toluene cannot be deposited as lm using electrophoresis as it does not carry sufficient surface charge.
In addition to CsPbBr 3 and CsPbBr 3 -CdS heterostructures, we also synthesized CdS QDs with the same ligands following a similar experimental procedure (i.e. without CsPbBr 3 ). Fig. 2A shows the absorption spectra of CsPbBr 3 , CdS and CsPbBr 3 -CdS QDs in toluene. The CdS and CsPbBr 3 QDs exhibit characteristic excitonic peaks at 434 and 518 nm, respectively. Although the absorption spectrum of the CsPbBr 3 -CdS heterostructure shows two peaks that overlap with the absorption of individual QDs, the peak around 434 nm may also arise from the deposition of small size CdS particles on the surface of CsPbBr 3 nanocrystals. Such a decoration of CdS particles, if any, would give rise to a CdS excitonic peak in the heterostructures. The excitonic peak at 518 nm (CsPbBr 3 ) remains unaffected aer heterostructure formation, thus ruling out any interference of exchange of metal ions. Similarly, the tail absorption at longer wavelengths arises from the scattering effects, similar to what is seen in PbSO 4oleate capped CsPbBr 3 . 46 The emission spectra of these three nanostructures are shown in Fig. 2B. Whereas CdS QDs remain the least emissive, both CsPbBr 3 and CsPbBr 3 -CdS QDs are highly emissive. Additionally, the emission features of CsPbBr 3 QDs remain unchanged following the deposition of CdS: the emission maximum (521 nm) and the full width at half maximum (18 nm) of CsPbBr 3 are unaffected aer CdS modication. These results conrm that there is no substitution of cations during the heterostructures synthesis, and thus the emission characteristics of the parent CsPbBr 3 QDs are retained in the heterostructure. If there was any substitution of Pb 2+ with Cd 2+ we would expect a blue shi in the absorption and emission maxima of the perovskite QD. 47 The excitation spectra recorded at different emission wavelengths conrm the origin of the emission to arise from the CsPbBr 3 ( Fig. S4 in the ESI †).
Another interesting aspect of the capping with CdS is the enhancement in emission yield. The emission quantum yields as determined using an integrating sphere were 39% and 60% for CsPbBr 3 and CsPbBr 3 -CdS QDs, respectively (see ESI † for details). Passivation of surface defects by CdS is expected to suppress nonradiative processes and thus lead to increased emission yield. For example, capping of CdSe with CdS has resulted in the signicant enhancement of emission yield. [48][49][50][51] The ow of charge carriers from the CdS shell to CdSe core in these studies was established through emission and excitation spectral measurements. Control experiments were carried out to check whether introduction of Cd 2+ ions alone can induce similar changes in the emission properties. Fig. S5 † shows a decrease in emission yield and lifetime when CsPbBr 3 was treated with cadmium acetylacetonate (Cd(acac) 2 ) instead of Cd(DDTC) 2 . Additionally, treatment with Cd 2+ did not change the absorption of the QDs. This further conrms that the observed optical properties are due to the presence of CdS in the heterostructure.
We employed time-resolved emission measurements to monitor the excited state behavior of CsPbBr 3 before and aer CdS deposition. The emission decay at 520 nm was monitored for CsPbBr 3 and CsPbBr 3 -CdS samples (Fig. 2C). Each trace was analyzed using a biexponential decay t and the tting parameters are presented in Table S1. † Of interest is the increase in emission lifetime of CsPbBr 3 upon capping with CdS. A nearly seven-fold increase in average lifetime of CsPbBr 3 -CdS QDs (s ave ¼ 46.9 ns) was observed over that of pristine CsPbBr 3 QDs (s ave ¼ 7.0 ns). This shows that CdS deposition facilitates long-lived charge separation in CsPbBr 3 -CdS. In addition to surface passivation, we can also expect the formation of a quasi-type II heterojunction as shown in the scheme (Fig. 2D). Whereas direct charge carrier recombination is dominant in pristine CsPbBr 3 , the nearly isoenergetic conduction bands of CsPbBr 3 and CdS can facilitate delocalization of electrons across the two semiconductors, thus improving charge separation. Since the CdS layer is relatively thin, its contribution to the emission is expected to be small. The excitation spectra (Fig. S4 †) rule out the contribution from CdS to overall emission. The observed increase in lifetime parallels the emission yield enhancement seen in the CsPbBr 3 -CdS heterostructure.

Stability in polar environment
Attaining long term stability of CsPbBr 3 nanocrystals in polar solvents remains a challenge. CsPbBr 3 nanocrystals undergo rapid degradation in polar solvents, which has hampered their applications in photocatalysis. Even the addition of a small amount of polar solvents such as ethanol or water can induce chemical transformation/precipitation of CsPbBr 3 QDs and thus a loss of photoactivity. 9,52,53 Recently, it was reported that ZnS-capped CsPbBr 3 QDs were stable in a toluene : water biphasic mixture. The contact with water was made by periodic shaking since the two solvents are immiscible. 28 We also conducted a similar stability test of CsPbBr 3 and CsPbBr 3 -CdS nanocrystals using a biphasic mixture of toluene and water with periodic shaking. The emission spectra of CsPbBr 3 and  The biphasic solvent mixture approach does not represent an increase in the overall polarity of the medium. Ideally, an inorganic shell should prevent direct contact of CsPbBr 3 with a polar environment and maintain its photostability. We checked the stability of CsPbBr 3 and CsPbBr 3 -CdS by introducing a miscible polar solvent (ethanol) to a toluene suspension of the QDs and monitoring the absorption and emission spectra over time. Fig. 3A and B show the absorption and emission spectra recorded following addition of ethanol (15% v/ v) to toluene solution over a period of 60 minutes. The absorption of pristine CsPbBr 3 QDs shows enhanced absorbance with time due to scattering effects caused by turbidity as the ligands from QD surface become detached in the polar medium. 54 A $85% decrease in the CsPbBr 3 emission yield is seen immediately aer the addition of ethanol to the toluene solution. In addition, upon exposure to ethanol we also see a change in the absorption of CsPbBr 3 due to particle aggregation. These results conrm the susceptibility of CsPbBr 3 QDs to polar environment (Fig. 3C). On the other hand, CsPbBr 3 -CdS QDs exhibit only a small decrease ($15%) in emission with a relatively small change in the absorption during 60 min of exposure in toluene/ethanol mixed solvent. The CdS deposition provides the necessary protection for CsPbBr 3 , and thus decreases its susceptibility to ethanol-induced degradation.

Excited state interactions with an electron acceptor
Since CsPbBr 3 -CdS QDs were stable in toluene/ethanol mixed solvent, we were able to probe the excited state interactions and interfacial electron transfer with a cationic electron acceptor, ethyl viologen, EV 2+ . Emission spectra recorded at different concentrations of EV 2+ are shown in Fig. 4A. The quenching of photoluminescence conrmed the excited state interaction between CsPbBr 3 -CdS QDs and EV 2+ . As a control, we tested the solubility of EV 2+ in the toluene/ethanol mixed solvent separately by recording absorption spectra and conrming the probe molecules are fully soluble at the concentrations employed in this study (Fig. S7 †). Earlier studies have shown direct complexation between CsPbBr 3 and methyl viologen and elucidated the role of surface bound ligands in dictating the complexation constant. 14,15 Here, we were able to quench the emission at micromolar concentrations of EV 2+ , thus indicating a complex formation in the ground state between CsPbBr 3 -CdS and EV 2+ . 51 The equilibrium of the bound and unbound EV 2+ molecules (reaction (1)) can be expressed in terms of the apparent association constant, K app and emission yields (expression (2)). 55 The observed quantum yield (f f (obs)) takes into account the emission arising from EV 2+ -bound (f f 0 ) and pristine (f f 0 ) CsPbBr 3 QDs. With increasing concentration of EV 2+ , more CsPbBr 3 -CdS QDs bind to viologen and thus exhibit a decrease in the emission yield.
The photoluminescence quenching data was analyzed using expression (2). The emission intensity of the QDs at the emission maximum (which is proportional to quantum yield, (f f (obs))) was monitored at different concentration of EV 2+ . 55,56 The linear dependence of the double reciprocal plot Fig. 4B conrms the validity of the association between CsPbBr 3 -CdS and EV 2+ . The apparent association constant K app determined from the slope and intercept of the plot in Fig. 4B was 7.0 Â 10 4 M À1 . This complexation constant is 1-2 orders of magnitude smaller than the one observed for uncapped CsPbBr 3 and viologen (0.8-7.0 Â 10 6 M À1 ). 15 The decrease in K app further indicates that the presence of CdS reduces the surface interactions with the viologen. The K app value we obtain in this study is in line with literature values of CdS interacting with viologens. 57 To further establish the excited state interactions, we monitored the photoluminescence lifetime of the CsPbBr 3 -CdS QDs at different EV 2+ concentrations. Time-resolved luminescence decay traces were recorded in toluene : ethanol (85 : 15% v/v) using an excitation source at 370 nm. The lifetimes were tted to a biexponential kinetic expression (expression 3), and the tting parameters are given in Table S2. † The average lifetime (s ave ) decreased with increasing concentration of EV 2+ in accordance with the photoluminescence quenching seen in Fig. 4A. The decrease in average lifetime from 42.4 ns to 18.1 ns upon addition of 12 mM EV 2+ is indicative of a competing excited state deactivation pathway involving electron transfer from the CsPbBr 3 -CdS QDs to EV 2+ . Since this electron transfer is likely to occur within the time resolution ($1 ns) of our photoluminescence lifetime set up, we employed femtosecond transient absorption spectroscopy to resolve the electron transfer process.
Transient absorption spectra were recorded following laser pulse excitation at 400 nm (16 mJ cm À2 ). The transient absorption spectra of a representative CsPbBr 3 -CdS QD sample containing 0 and 4 mM EV 2+ are shown in Fig. 5A and B respectively. The negative absorption (exciton bleach) feature centered at $527 nm corresponds to the charge separated state within the NCs. [58][59][60] The charge separation which occurs within the laser pulse is seen in spectrum 'a' recorded with a probe delay of 1 ps. As electrons and holes recombine, a recovery in the bleached absorption is seen. The bleach recovery at 527 nm for the samples containing different amounts of EV 2+ are presented in Fig. 5C (see Fig. S8 † for longer time scale kinetics). With increasing EV 2+ concentration we observe a quick recovery of the bleach feature, thus conrming the presence of an additional deactivation pathway for the photogenerated electrons, viz., electron transfer to EV 2+ (reaction (4)).
The bleach recovery was analyzed using a biexponential kinetic t 35 and the tting parameters (a 1 , s 1 ) and (a 2 , s 2 ) corresponding to fast and slow components are presented in Table  S3. † While the fast component varied in the range 18.6-6.6 ps the long component varied in the range of 129.6-71.7 ps.
If we assign the decrease in the fast component to the electron transfer pathway from CsPbBr 3 -CdS to EV 2+ , we can obtain the rate constant for electron transfer through the expression (5).
If we substitute the fast time components (s 1 ) of CsPbBr 3 -CdS and that of CsPbBr 3 -CdS with 32 mM EV 2+ in expression 5 (Table S3 †), we obtain rate constant (k et ) of the electron transfer in the range of 3.8-9.8 Â 10 10 s À1 for three different EV 2+ concentrations (4-32 mM) or an average rate constant of 6.5 Â 10 10 s À1 . This rate constant for electron transfer is lower than the one obtained for electron transfer between oleic acid/ oleylamine capped CsPbBr 3 QDs and viologen (k et ¼ 3.6 Â 10 11 s À1 ). 15 As discussed in the emission quenching experiments, the surface interactions play a role in dictating the kinetics of interfacial electron transfer.
It is evident that modication of CsPbBr 3 with CdS slows down the electron transfer rate. Schemes 1A and B illustrate the two scenarios for achieving electron transfer, viz., without and with CdS modication. As shown earlier, when methyl viologen is directly bound to CsPbBr 3 , the charge separation is extended, with electrons residing in the viologen moiety and holes residing within the CsPbBr 3 QDs. 14 This extended charge separation as a bound pair was manifested as a long-lived transient bleach component. In contrast, the electron transfer with CsPbBr 3 -CdS is mediated through the CdS layer as expected by the quasi-type II band alignment (Scheme 1C). The reduced ethyl viologen (EV + c) is no longer directly bound to the CsPbBr 3 , but instead is now linked to the CdS layer. Similar CdSmediated electron transfer has been observed in CdSe-CdS heterostructures. 51 The bleaching recovery accelerates with increasing viologen concentrations as the electrons are depleted from the CsPbBr 3 core, mediated through CdS. The distinct difference between the two bleach recovery kinetics observed with pristine CsPbBr 3 and CsPbBr 3 -CdS QDs further highlights

Steady state photolysis and Fermi level equilibration
Although there have been several studies that demonstrate the photocatalytic properties through product identication, 17,18,36,61 direct spectroscopic identication of intermediates or electron transfer products is rather limited. If indeed the CsPbBr 3 -CdS heterostructure is responsible for photocatalytic reduction, we should be able to observe the buildup of stable viologen radical under continuous photoirradiation. 62,63 Deaerated CsPbBr 3 -CdS nanocrystal suspensions in toluene : ethanol (85 : 15 v/v%) mixed solvent containing 100 mM of EV 2+ were subjected to steady-state visible light illumination (>400 nm; 200 mW cm À2 ). Absorbance spectra were recorded periodically during the steady state photolysis experiment. Fig. 6A shows a representative difference absorbance spectrum, which shows distinct peaks at 405 nm and 608 nm, corresponding to the absorbance of EV + c. 64 Fig. 6B shows the growth of the 608 nm absorption over time, which attains a plateau aer about 45 minutes. The steady state concentration of EV + c increased with increasing concentration of EV 2+ . From the extinction coefficient of EV + c at 608 nm (3 ¼ 1.4 Â 10 4 M À1 cm À1 ) 63,64 we can determine the steady state concentration of the electron transfer product. The quantum yield (QY) of EV + c formation (F EV + c ) was determined using potassium ferrioxalate actinometry. 51,65 Details of the actinometry experiments are presented in the ESI. † In the present case we obtain a maximum quantum efficiency (F EV + c ) of 8.4% for the electron transfer. Although the initial electron transfer yield as monitored from the transient absorption could be as high as 73%, 15 the back-electron transfer during the steady state irradiation makes the net electron transfer yield lower than the value obtained immediately aer laser pulse excitation.
The steady state concentration of the reduction product EV + c is dictated by the forward and back electron transfer processes. Because of the use of ethanol as a hole scavenger, the back electron transfer rate constant (k bet ) is signicantly lower in the present experiments. The redox couple in contact with a semiconductor surface undergoes Fermi level equilibration that is dependent on the position of the conduction band and the potential of the redox couple. The equilibrium concentration of the reduced and oxidized species can be used to determine the at band potential of the semiconductor. The steady state concentration of [EV+c] ss is indicative of charge equilibration between the CsPbBr 3 -CdS nanocrystals and EV 2+ /EV + c couple. We employed the Nernst equation (expression (6)) to obtain the at band potential of the CsPbBr 3 -CdS heterostructure.
By substituting the redox potential of the EV 2+ /EV + c couple, E ¼ À0.449 V vs. NHE, 64  . It should be noted that the at band potential of the CsPbBr 3 -CdS heterostructure obtained in this study is based on the charge equilibration between the semiconductor QD and redox couple and may differ from the values obtained from theoretical estimates or bulk material. The estimate of at band potential provides an estimate of the energetics of the CsPbBr 3 -CdS heterostructure suspended in the solvent medium to execute photocatalytic processes.

Conclusions
The design of CsPbBr 3 -CdS heterostructure offers stabilization of perovskite nanocrystals for photocatalytic applications in polar medium. The salient feature of the CsPbBr 3 -CdS heterostructure is realized through its stability in mixed solvents, remediation of surface defects, and increased emission yield. The stability of the CsPbBr 3 structure in mixed solvents with increased polarity has allowed us to accumulate electron transfer product (reduced viologen) under steady state irradiation conditions with a quantum efficiency of 8.4%. The relatively high electron transfer efficiency observed in the present study shows how a heterostructure design of perovskite nanocrystals plays a crucial role in dictating the photocatalytic properties of stable perovskite nanocrystals.

Data availability
The data is available within the main text and ESI. †

Conflicts of interest
There are no conicts to declare.