Role of Surface States in Photocatalysis: Study of ... - ACS Publications

Article pubs.acs.org/cm

Role of Surface States in Photocatalysis: Study of Chlorine-Passivated CdSe Nanocrystals for Photocatalytic Hydrogen Generation Whi Dong Kim,† Ji-Hee Kim,‡ Sooho Lee,† Seokwon Lee,† Ju Young Woo,†,∥ Kangha Lee,† Weon-Sik Chae,Λ Sohee Jeong,∥ Wan Ki Bae,¶ John A. McGuire,⊥ Jun Hyuk Moon,∇ Mun Seok Jeong,‡,§ and Doh C. Lee*,† †

Department of Chemical and Biomolecular Engineering (BK21+ Program), KAIST Institute for the Nanocentury, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Korea ‡ Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon 24233, Korea ∥ Nanomechanical Systems Research Division, Korea Institute of Machinery and Materials (KIMM), Daejeon 34103, Korea Λ Analysis Research Division, Daegu Center, Korea Basic Science Institute, Daegu 41566, Korea ¶ Photo-Electronic Hybrids Research Center, Korea Institute of Science and Technology (KIST), Seoul 02792, Korea ⊥ Department of Physics and Astronomy, Michigan State University, East Lansing, Michigan 48824, United States ∇ Department of Chemical and Biomolecular Engineering, Sogang University, Seoul 04107, Korea § Department of Energy Science, Sungkyunkwan University, Suwon 24233, Korea S Supporting Information *

ABSTRACT: We examine the effects of chlorine-passivation of Cd surface atoms on photocatalytic H2O reduction by CdSe NCs. Transient absorption spectroscopy reveals that Cl passivation removes electron trap states in CdSe NCs, which is also reflected in an increase of photoluminescence quantum yield, e.g., from 9 to 22% after the Cl treatment. Size-tunable energy states in CdSe NCs enable the systematic investigation of surface defects and their effect on the photocatalytic hydrogen generation rate. It turns out that, depending on band-edge energy levels, the surface trap states may enhance or inhibit photocatalysis. Cl-treated CdSe NCs larger than 2.7 nm show a higher hydrogen evolution rate than untreated CdSe NCs of the same size as Cl treatment removes trap states with energy below the H2O reduction potential. In contrast, the same Cl treatment does not increase the photocatalytic rate of CdSe NCs smaller than 2.7 nm because both the conduction band edge and trap states are above the water reduction potential. The size-dependence of the effect of Cl treatment suggests that electron trap states in CdSe may promote photocatalytic activity by enhancing charge separation.

■

INTRODUCTION

Recent efforts to address the surface passivation of semiconductor NCs have highlighted the potential of halide atoms.10,11 For example, we have reported the formation of metal-halide complexes (e.g., PbX2) on PbSe NCs via simple halide treatments.12,13 The resulting PbSe NCs showed little to no shift in absorption peak position for weeks, while untreated NCs underwent an uncontrollable blue shift due to surface oxidation. Page et al. reported that Cl passivation on CdTe NCs leads to near-complete suppression of surface trapping of photogenerated charge carriers, resulting in the increase of PL QY.9 In contrast, in the case of CdSe NCs, it has been reported that the PL QY decreases upon Cl treatment, because Cl atoms act as “hole traps”, as evidenced in surface photovoltage spectroscopy and transient absorption (TA) spectroscopy.14,15

The size-tunable optical properties of colloidal semiconductor nanocrystals (NCs) have prompted a wave of developments in their use as sensitizers in photocatalysis and solar cells.1−4 Advances in the arrested precipitation approach have enabled the wet-chemical growth of colloidal semiconductor NCs with intricate control over size and shape.5,6 However, one of the major roadblocks to the commercial deployment of NCs has been the lack of control over surface trap state, which alter excited-state dynamics and quench photoluminescence (PL).7 The surface trap states result from insufficient passivation of surface atoms, which become sites for surface-related nonradiative recombination.8 Organic capping ligands used in colloidal synthesis of semiconductor NCs confer solubility and air stability on the NCs, but the passivation of surface atoms by organic ligands is incomplete, as indicated by the relatively low PL quantum yield (QY) ( 97%); sodium sulfate (>98%); sodium sulfide (90%); D2O (99.98%); and acetonitrile (>99.5%). Phenyl ether (99%) was purchased from TCI and D-chloroform (99.8%) from CIL, and these chemicals were used as-received. Synthesis of CdSe Nanocrystals (NCs). CdSe NCs of varying size were synthesized by a hot-injection method using a previously reported recipe by Lim et al.17 Briefly, in a three-neck flask, 70 mL of ODE solution containing 6.24 mL of OA and 0.51 g of CdO was heated to 230 °C under Ar to form a clear solution. 0.6 mL of 2 M TOP-Se solution was quickly injected into the flask and the injection was repeated every 10 min. Aliquots were taken at different times during the progress of the reaction: 40 s, 20, 30, 40, and 50 min. These aliquots contained CdSe NCs of different size: 2.3, 2.7, 2.9, 3.5, and 4.5 nm, respectively. The reaction mixture was cooled to room temperature by injection of cold toluene to terminate the reaction. The resulting products were collected by adding ethanol followed by centrifugation at 10 000 rpm. The precipitated CdSe NCs were redispersed and stored in toluene. Cl Treatment. One mL ammonium chloride solution dissolved in methanol was added to 10 mL CdSe solution (10 mg/mL) before and after purification to prepare in situ and ex situ Cl:CdSe NCs, respectively. The mixture was then stirred for 10 min at 60 °C. Both in situ and ex situ Cl:CdSe NCs were collected using centrifugation described above. To prepare modified in situ Cl:CdSe NCs, MUAcapped in situ Cl:CdSe NCs were dispersed in methanol (10 mg/mL), and 1 mL of methanol solution of ammonium chloride (0.025 M) was added. The mixture was stirred for 10 min, and then toluene was added to the mixture. Centrifugation at 10 000 rpm for 10 min yields precipitates, which were dissolved and stored in water. Synthesis of Pt-CdSe NCs. We synthesized Pt-CdSe NCs via a modified version of the procedures described by Bang et al.18 OA (0.20 mL, 0.63 mmol), oleylamine (0.20 mL, 0.61 mmol), 1,2hexadecanediol (43 mg, 0.17 mmol), and phenyl ether (10 mL, 63 mmol) were added in a flask (flask 1) and purged with Ar and heated to 200 °C after degassing for 30 min at 80 °C. In a separate flask (flask 2), platinum(II) acetylacetonate (25 mg) and CdSe NCs (100 mg) were mixed with 1,2-dichlorobenzene (10 mL) and heated at 65 °C for 10 min. The solution in flask 2 was injected into flask 1 at 200 °C.

■

RESULTS AND DISCUSSION Effect of Cl Treatment Methods. Cl-treated CdSe NCs (Cl:CdSe NCs) were prepared by injecting a methanol solution of ammonium chloride either into the NC growth solution immediately after the growth was terminated (in situ Cl:CdSe NCs) or into a toluene solution of washed CdSe NCs (ex situ Cl:CdSe NCs). We performed XPS analysis to examine the existence of Cl atoms on the surface of CdSe NCs. As shown in Figure S1a, b, a peak appears at 198.7 eV from both samples, indicating that Cl atoms are on the surface,15 while no Cl peak appears in an XPS spectrum of untreated CdSe NCs (Figure S1). Figure 1 shows shift of 1S exciton peak in relative to untreated CdSe NCs and PL QY after treatment with varying NH4Cl concentrations. The position and width of the 1S 963

DOI: 10.1021/acs.chemmater.5b04790 Chem. Mater. 2016, 28, 962−968

Article

Chemistry of Materials Scheme 1. Summarized Illustration of Cl Treatment Methods Presented in This Study

the ligand exchange, we added an extra Cl treatment step after the exchange. Cl:CdSe NCs used in the photocatalysis described hereafter were prepared via this modified in situ procedure. In addition, it has been reported that holes rapidly migrate (∼1 ps) to the thiol group of MUA molecules,23,24 and so hydrogen generation by MUA-capped CdSe NCs is due exclusively to electron dynamics. Holes trapped at NC surface atoms usually induce anodic corrosion resulting in formation of SeO2.25 Rapid hole transfer to MUA molecules is corroborated by our XPS results that Se does not undergo oxidation to SeO2 in the cases of both untreated and Cl-treated MUA-capped CdSe NCs under illumination, unlike OA-capped CdSe NCs (Figure S4). Therefore, surface atoms of CdSe NCs typically believed to be responsible for hole trapping do not contribute to photocatalytic activity change between untreated and Cltreated samples in the case of MUA-capped CdSe NCs. Size-Dependent Photocatalytic Properties. Figures 2a and b show a photograph of CdSe NCs of various sizes in aqueous solution and their absorption spectra, respectively. The size of the studied CdSe NCs ranges from 2.3 to 4.4 nm. To measure the photocatalytic activity of CdSe NCs, NC suspensions were irradiated under AM1.5 illumination in the presence of a mixture of sodium sulfide and sodium sulfite (0.25 M) as sacrificial agents. Figure 2c summarizes the hydrogen generation rate using CdSe NCs of varying size prepared with or without the Cl treatment. Interestingly, the relative hydrogen generation rate in Cl:CdSe NC samples increases at larger NC size. For a more quantitative perspective, we plotted the hydrogen generation normalized by the absorption cross-section as a function of energy difference between conduction band edge and water reduction potential (Figure 3).26 The rate constant kRed for electron transfer depends on the difference between the conduction band edge and proton reduction potential; kRed ∝ exp(ECB−ERed)/kT, where ECB is the conduction band level and ERed is the redox potential for hydrogen generation.27 In the case of untreated CdSe NCs, the high conduction band level in small NCs leads to the highest hydrogen generation rate as a consequence of a high

Figure 1. (a) 1S peak shift and (b) PL QY of CdSe NCs as a function of NH4Cl concentration after in situ and ex-situ treatment processes. (Δλ = λ (1S peak after Cl treatment) − λ (original 1S peak)).

exciton peak do not significantly change after introducing Cl in the case of in situ CdSe NCs, while both a shift and broadening of the 1S peak are observed in ex situ CdSe NCs as shown in Figure 1 and Figure S2. The blue-shift and broadening of the 1S peak are attributed to etching by Cl ions in the ex situ Cltreatment process.19 Shin et al. reported that chlorine ions etch the surface atoms of CdSe NCs, which was confirmed by decreased NC size and detection of cadmium trichloride (CdCl3−) in the NC solution.20 At a high NH4Cl concentration (e.g., 1.0 μM/mg), CdSe NCs started flocculating in solution, likely resulting from significant removal of capping ligands from the NC surface. Aggregated ex situ Cl:CdSe NCs dissolve in Nmethylformamide (NMF), because halide ions that are bound to Cd surface atoms provide electrostatic stabilization in NMF solution (Figure S3).21 Dispersibility in NMF clearly supports that ex situ Cl:CdSe NCs are passivated by Cl. In the case of in situ Cl:CdSe NCs, the Cl atoms react with the surface Cd atoms in the presence of excess capping ligands, which inhibit the etching of surface atoms (Figure S2).22 As shown in Figure 1b, the in situ treatment results in a notable increase of PL QY. To the best of our knowledge, this is the first observation that an increase of PL QY can be achieved by halide passivation of CdSe NCs. The increase in PL QY in the case of in situ Cl:CdSe NCs suggests that Cl atoms passivate unsaturated surface Cd atoms. To be used in photocatalytic water splitting, CdSe NCs must be in an aqueous suspension. To make the CdSe NCs soluble in water, we replaced oleic acid (OA) on CdSe NCs with MUA. NMR spectroscopy reveals that the ligand density decreases from 3.7/nm2 to 1.1/nm2 after ligand exchange into MUA on Cl:CdSe NCs (see the Supporting Information for details). As illustrated in Scheme 1, the ligand exchange results in an increased number of dangling bonds on the surface of Cl:CdSe NCs, which is detrimental to their optical and colloidal stability. (Figure S1d) To keep surface atoms passivated with Cl after 964

DOI: 10.1021/acs.chemmater.5b04790 Chem. Mater. 2016, 28, 962−968

Article

Chemistry of Materials

To unravel the dynamics of electrons in Cl-treated and untreated NCs, we carried out transient absorption (TA) measurements using a pulsed laser with 200 fs resolution, so as to probe electron dynamics in the picosecond time scale. Bleaching of the 1S peak in the absorption spectrum indicates the presence of electrons in the 1Se band-edge state. TA and TRPL results shown in Figure S4b indicate that TA indeed reflects the dynamics of electrons: The PL signal decays by about 90% within 1 ns in TRPL spectra due to migration of holes, whereas only 30−40% decay is observed in TA spectra, which are sensitive primarily to the population of 1Se electrons.26 Figure 4 shows the normalized degree of bleaching

Figure 2. (a) Photograph of MUA-capped CdSe NCs with varying size dispersed in deionized water after ligand exchange. (b) UV−vis absorption spectra of CdSe NCs with varying size. Absorbance was normalized with respect to weight of samples. (c) Experimental H2 generation rate from CdSe NCs with various size before and after Cl treatment.

Figure 3. Relative H2 generation rate normalized by the number of absorbed photons under AM 1.5 illumination plotted against energy difference between NC conduction band edge and water reduction potential before and after Cl treatment. Normalized H2 generation rate is calculated by dividing hydrogen generation rate by overlap integral between the AM 1.5 spectrum and absorbance of the respective NCs. Figure 4. Dynamics of bleaching at the 1S transition obtained from transient absorption spectra for (a) small and (b) large NC samples before and after Cl treatment at 298 K.

electron transfer rate. This result implies that the altered gap between the conduction band and redox potential is responsible for the different photoactivity and is in agreement with previously reported results.2,27,28 On the other hand, the photocatalytic activity of Cl-treated NCs shows a different size dependence: most notably, 2.3 nm Cl:CdSe NCs show lower activity than 2.7 nm NCs. Elucidation of Electron Transfer in Cl-Treated NCs. To explain the contrasting size dependence of photocatalytic activity between Cl-treated and untreated CdSe NCs, we examined the recombination dynamics using time-resolved photoluminescence (TRPL) spectroscopy. Figure S5a shows that about 50% of initial exciton population decays within 50 ps regardless of Cl treatment in the MUA-capped CdSe NCs, while OA-capped CdSe NCs exhibit relatively slow decay. The rapid PL decay in MUA-capped CdSe NCs is attributed to thiol groups on NC surface accepting holes on a time scale of 20−30 ps.29

of the 1S peak in TA spectra of Cl-treated and untreated CdSe NCs of two different sizes at room temperature. The initial fast decay (within 30 ps) is attributed to electron transfer from the 1S conduction edge state to surface traps.30 In Cl:CdSe NCs, the number of electron trap states is expected to be smaller as a result of Cl treatment, which explains the slower bleaching of the 1S absorption peak than in the case of untreated CdSe NCs. For instance, the electron population decays in 30 ps by ca. 33% and 52% of small and large bare CdSe NCs and by ca. 22% and 42% of small and large Cl:CdSe NCs, respectively, suggesting the diminution of electron trap states after the Cl treatment. Similar conclusions are drawn from cyclic voltammograms, in which the onset voltage for reduction is slightly shifted after the Cl treatment (See Figure S6). On the basis of the TA spectroscopy and CV analysis, it becomes clear that electron trap states significantly influence the difference in 965

DOI: 10.1021/acs.chemmater.5b04790 Chem. Mater. 2016, 28, 962−968

Article

Chemistry of Materials

trapping time of ca. 30 ps for the case of CdSe NCs.24 Because of this fast electron transfer, one would expect that Pt-CdSe NCs would show similar hydrogen-generation rate changes regardless of NC size and conduction band minimum. Indeed, as shown in Table 1, 2.3 and 4.5 nm CdSe NCs result in

size-dependence of photocatalytic activity between untreated and Cl-treated CdSe NCs. Effect of Cl Treatment on Photocatalysis. For 4.5 nm CdSe NCs, the photocatalytic rate of hydrogen evolution is 9 times higher in the case of Cl:CdSe NCs. Thibert et al. reported that the hydrogen-generation rate of CdSe NCs decreases significantly when the energy level of trap states lies below the reduction potential necessary for hydrogen generation, whereas elimination of surface trap states helps increase the photocatalytic rate after growth of a thin inorganic shell (e.g., CdS).4 Similarly, our Cl treatment is believed to remove surface trap sites whose energy levels are located at lower values than the water reduction potential. The removal of surface traps facilitates electron transfer from the conduction band edge of CdSe NCs to a proton. In contrast, in the case of CdSe NCs smaller than 3 nm (e.g., 2.3 and 2.7 nm), the photocatalysis becomes slower after the Cl treatment. As the conduction band edge shifts upward at small NC size, so do the surface trap states.31 Because the trap states are located above the H2 redox potential in these small NCs, a fraction of electrons trapped in the sites still can migrate to a proton. Moreover, the trapping results in enhanced charge separation, which can increase photocatalytic activity even further,32 since recombination of trapped carriers is much slower (approximately microsecond) than radiative recombination of excitons (∼20−50 ns).33 Therefore, untreated NCs with higher density of surface trap states exhibit superior photocatalytic activity in smaller NCs, as summarized in Scheme 2. Pt-CdSe Heterostructure NCs. Pt nanoparticles anchored on semiconductor photocatalysts are known to increase photocatalytic activity in water splitting.34 In particular, CdSe NCs with Pt nanoparticles on their surface (denoted hereafter as Pt-CdSe NCs) have shown electron transfer from CdSe to Pt that is fast (