Defect-Mediated Gold Substitution Doping in ZnO Mesocrystals and

Nov 18, 2015 - Ming-Han LiuYun-Wen ChenTien-Sung LinChung-Yuan Mou. ACS Catalysis 2018 8 (8), 6862-6869. Abstract | Full Text HTML | PDF | PDF w/ ...
0 downloads 0 Views 4MB Size
Download PDF
Research Article pubs.acs.org/acscatalysis

Defect-Mediated Gold Substitution Doping in ZnO Mesocrystals and Catalysis in CO Oxidation Ming-Han Liu,† Yun-Wen Chen,‡ Xiaoyan Liu,†,∥ Jer-Lai Kuo,‡ Ming-Wen Chu,*,§ and Chung-Yuan Mou*,† †

Department of Chemistry and Center of Condensed Matter Science, National Taiwan University, Taipei 10617, Taiwan Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 10617, Taiwan ∥ Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China § Center of Condensed Matter Science, National Taiwan University, Taipei 10617, Taiwan ‡

S Supporting Information *

ABSTRACT: Highly dispersed supported gold with strong metal−support interaction is a desirable material for heterogeneous catalysis. Unlike current dispersion strategies of depositing gold from solution to support, we report herein a new method of producing highly dispersed gold clusters on ZnO mesocrystal. The gold clusters appeared on defect-rich twin-brush ZnO mesocrystals (TB-ZnO) via an in-and-out process: (i) a mixed Au/Zn oxide was formed first after deposition-precipitation of Au with AuCl4−, (ii) fine Au clusters grow from the underneath to the surface of the oxide after heating. The TB-ZnO behaved like a sponge allowing gold atoms to heavily disperse into the wurtzite structure of ZnO with Zn-substitution by gold. After mild thermal treatment, the embedded gold emerged from underneath of the ZnO support to form highly dispersed Au nanoparticles of ∼2 nm on the ZnO surface. DFT calculation shows energetically favored Au-doping in TB-ZnO and facile defect-mediated migration of Au in it. The material (Au/TB-ZnO) gave outstanding activities for the catalysis of CO oxidation. The use of mesocrystals of metal oxide as supports, with rich vacancy defects, provides a new route for preparing highly dispersed and active supported metal catalysts. KEYWORDS: mesocrystal, gold, ZnO, defect, substitution, CO oxidation, catalysis

■

INTRODUCTION Understanding and exploiting metal−support interactions are long-standing goals of researchers in heterogeneous catalysis. In the last two decades, metal oxide-supported gold nanocatalyst for oxidation has been a paradigmatic problem in catalysis.1 The roles of supports in gold-based catalytic oxidation have been examined in size/shape control of gold particles,2 electron transfer between gold and oxide,3 and oxygen activation.4 However, the role of support defects in metal−support interaction is not well-understood. “Defect Engineering” by doping the oxide support5 for tuning metal−support interaction is emerging as a key strategy for preparing an outstanding supported metal catalyst. However, most works were focused on the modulation of electron flow from or to the metal nanoparticles which gives pining of nanoparticles and affects the adsorption of substrates.5,6 A more important doping effect is in the participation of activation of reactant (such as dioxygen) in heavily doped mixed oxides. Mixed oxides of two metals in a common oxide matrix can give novel electronic and catalytic properties.7,8 Catalysis by doped oxides is an emerging field of wide interest. In the problem of supported Au-based oxidation catalyst, doped oxide could be an effective oxidant. For example, undoped and gold-doped CuMnOx catalysts were © 2015 American Chemical Society

reported to follow the Mars van Krevelen (MvK) mechanism in CO oxidation, with gold-doped Au/CuMnOx promoting more oxygen activation.9 Recently, Sutter et al. reported a synthesis of Au-rich Au−In mixed oxide, by oxidation of AuIn alloy, which is capable of adsorbing both CO and O2 and, further, gave CO2 at room temperature.10 It could be the result of loose binding of the O atom in Au−O−In linkage. In CO oxidation, Au nanoparticle (adsorbing CO) next to a nearly “weakly bound” oxygenlinkage would be a highly active center. Thus, it will be a highly interesting situation where gold is supported on a gold-doped metal oxides, e.g., Au is in both Au nanoparticle and forming mixed oxide as support (denoted as Au/MOx:Au). This may create many Au−O bonds next to Au nanoparticles on the support which may help in oxygen activation. In this paper, we report a new method of making Au supported on ZnO that would give a catalyst Au/ZnO:Au where gold were both supported on and doped in the ZnO. The ZnO support is in a special morphology of mesocrystal Received: June 15, 2015 Revised: November 11, 2015 Published: November 18, 2015 115

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis

exploited with a corresponding probe size of ∼1 Å (∼120 pA). The EDX spectrometer is a Bruker XFlash-5030, liquidnitrogen-free silicon drift detector with an optimized collection solid angle of ∼0.13 sr. Ultrathin slices of Au/TB-ZnO catalyst were prepared by ultramicrotome techniques. Cathodoluminescence spectra were measured using JEOL JSM-6500F scanning electron microscope equipped with Gatan MonoCL3 operated at 15 kV. Electron paramagnetic resonance (EPR) spectra were collected at 300 K with a Bruker EMX 10/12 spectrometer working under the X-band in CW-mode (9.46 GHz, 10 mW). Prior to EPR analyses, 10−15 mg of sample was placed inside a 4 mm O.D. quartz tube with greaseless stopcock for evacuation until the residual pressure lower than 10−3 Torr. X-ray absorption spectra (XAS) of Au/TB-ZnO catalysts were recorded at beamline 17C1 of National Synchrotron Radiation Research Center in Hsinchu, Taiwan. A double Si(111)-crystal monochromator was employed for energy selection whose resolution (ΔE/E) is 2 × 10−4. The catalysts were treated at 200 °C under 10 vol % O2 in He and then spread on Scotch tape. All spectra were recorded at room temperature in a transmission mode. Au foil as standard was measured simultaneously by using the third ionization chamber so that energy calibration could be performed. The X-ray absorption data were processed by Athena software package.17 Simulation Methods. In our DFT calculations, all electron wave functions are expanded in plane-wave basis sets with 400 eV energy cutoff in conjunction with pseudopotentials constructed in projector augmented-wave (PAW) method.18 The Au, Zn, and O atoms’ valence electrons included are 6s5d, 4s3d, and 2s2p orbitals, respectively. Perdew−Burke−Ernzerhof (PBE) generalized gradient approximation (GGA)19 is applied to calculate the electron exchange-correlation energies. A 3 × 3 × 3 Γ-centered Monkhorst−Pack k-point mesh20 is applied for taking care of long-range electron−electron interactions and making sure the convergence of energy estimation. The diffusion energy barriers are estimated via climbing nudged elastic band (cNEB) method which is implemented in VASP (Vienna Ab initio Simulation Package).21 All of the simulation models were built by constructing ZnO bulk with Zn-vacancies w/and w/o Au-substitutions. The details of how to construct the simulation models and more simulation parameters are described in the Supporting Information and Table S1.

form. Mesocrystals are formed by oriented attachment of crystallographically aligned nanocrystals,11,12 which give rise to the formation of porous single-crystal-like assembly. The assembled mesocrystal containing rich defects and voids have been demonstrated to possess potentially tunable electronic13 and photocatalytic14 properties. Normally, Au atoms would be very difficult to be doped into a well-annealed metal oxide; here we however take the advantage of a large number of vacancy defects existing in the mesocrystals of ZnO15 to allow heavy doping of Au in ZnO. To achieve this goal, we explore defective twin-brush-like ZnO mesocrystals (TB-ZnO),16 consisting of orientatedattached nanocrystals, to support gold nanoparticles via alkaline precipitation of AuCl4−. Mesocrystals of ZnO were synthesized hydrothermally by using gum Arabic as a structure-directing agent.16 After deposition of gold by deposition−precipitation, we find spontaneous heavy doping of gold into the lattice of the ZnO support. After heating, mobile gold atoms moved from under the sponge-like zinc oxide mesocrystal to form supported Au nanoparticles which give rise to extraordinarily catalytic activities in the low temperature oxidation of CO.

■

EXPERIMENTAL METHODS Synthesis of Au/TB-ZnO Catalyst. Twin-brush ZnO mesocrystals were prepared hydrothermally by using gum Arabic as a structure-directing agent as previously reported.16 The TB-ZnO mesocrystals were calcined at 300 °C for 3 h. Gold clusters were deposited on the calcined TB-ZnO mesocrystals by the deposition−precipitation (DP) method.1 Briefly, 10 mL of 0.5 wt % HAuCl4 solution was neutralized to pH value of 7. Afterward, it was dropped slowly into a solution containing 1 g of TB-ZnO, followed by adjusting pH value to 9.0. The resulting solution was stirred for 2 days vigorously at room temperature. The collected sample was washed a couple of times and centrifuged to be separated from AuCl4− solution. Then, the sample was dried in oven at 60 °C. Inductively coupled plasma mass spectrometry (ICP-MC) analysis of gold loading on the support was 4.1 wt % close to the nominal value (5 wt %). The catalyst was treated at 200 °C for 1 h under 10 vol % O2 in He, followed by doing CO oxidation and material characterization (i.e., ICP-MS, EPR, EXAFS analyses, and STEM observation). Catalytic Performance in CO Oxidation Reaction. Prior to the evaluation of catalytic activity, the catalyst was pretreated for 1 h under 10 vol % O2 in He (flow rate: 30 mL/min) at 200 °C. CO oxidation reaction was carried out at atmospheric pressure in a nearly isothermal packed-bed flow reactor (thickness ∼1 mm). Total feed flow rate into reactor was 90 mL/min with a volume ratio of CO/O2/He of 1:6:93. The CO conversions were determined by an online gas chromatograph equipped a 60/80 Carboxen 1000 packed column (1-2390-U, Supelco; 6820 GC system, Agilent Technologies) and a thermal conductivity detector (TCD). Characterizations. The crystal structure and phase composition of Au/TB-ZnO catalyst were investigated by using a Philips X’Pert diffractometer with Cu Kα radiation (λ = 1.5418 Å). Morphology and size of Au and TB-ZnO were examined by transmission electron microscope operated at 200 kV (TEM, JEOL JEM-2010) and field emission scanning electron microscope (FESEM, JEOL JSM-7600F). STEM images were conducted on an aberration-corrected electron microscope (JEOL-2100F). The HAADF collection angles are 72−192 mrad. A probe convergent angle of ∼20 mrad was

■

RESULTS AND DISCUSSIONS After calcination of the synthesized ZnO, we obtained porous single crystals of twin-brush form (TB-ZnO) with high surface area and the exposed surfaces mostly of the nonpolar {101̅0} planes (Figure 1a). A uniform twin-brush-like morphology (TB-ZnO, see also Figure S1a) was obtained with mean diameter at 1.0 ± 0.2 μm. ZnO is a polar crystal with intrinsic dipole along c-axis. The symmetrical shape of twin-brush structure is the result of dipolar cancelation to avoid divergent electrostatic energy. Nitrogen adsorption isotherm gives a larger surface area of 27.67 m2/g, which is due to textural porosity between the nanocrystals of ZnO. Subsequently, Au/ TB-ZnO catalyst with 4.1 wt % gold-loading was prepared by using TB-ZnO as support with deposition-precipitation (DP) method, followed by a pretreatment under O2. The TEM image of a sliced particle shows brushes-featured nanorods and the electron diffraction along c-axis indicates a single crystal oriented attachment of nanorods with exposed 116

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis

found on the as-prepared Au/TB-ZnO. TEM image of asprepared Au/TB-ZnO reveals a homogeneous contrast without particles decoration (Figure 1b). With O2-pretreatment at a higher temperature of 200 °C, few discernible Au nanoparticles (NP) were found on the external surface of ZnO under SEM observation (Figure 1c). XRD pattern reveals, only the presence of wurtzite ZnO without the reflections of gold (Figure 1c). Taking into account the nominal gold-loading of ∼4.1 wt % and laboratory XRD sensitivity of ∼2 wt %, this absence implies a large party of loaded Au may not appear as individual Au NPs in the material system. We thus conducted further TEM investigations of the microtomed Au/TB-ZnO. TEM observation of sliced Au/TB-ZnO sample with pretreatment at 200 °C reveals high density of dispersed Au nanoparticles with size of 1.9 ± 0.3 nm (Figure 1d). Apparently, very small Au NPs emerge after thermal treatment at 200 °C. In addition, the same catalyst after having been subject to a round of low-temperature CO reaction ramping from −20 to 90 °C shows basically the same size characteristic, 2.0 ± 0.3 nm (Figure S2). This indicates a strong metal− support interaction in the Au/TB-ZnO system. The mesocrystal of TB-ZnO support is nanoporous, as indicated by the dark and hole-like contrasts (arrows, Figure 2a) in the mass sensitive high-angle annular dark-field (HAADF) imaging in scanning transmission electron microscope (STEM). The characteristic atomic-number (Z) sensitivity of HAADF leads to a notable contrast between Au (Z, 79) and Zn (Z, 30). A close inspection of several tens of STEM-HAADF images at atomic resolution of comparable quality indicates two intriguing features: (i) patches of continuous (arrow 1 in Figure 2b) and dot-like (arrows 2 in Figure 2b and 2c) bright contrasts regularly appearing in ZnO and (ii) Au nanoparticles, ca. 2−3 nm, supported on ZnO displaying an Au-ZnO structural

Figure 1. (a) SEM image of calcined TB-ZnO mesocrsytal. (b) TEM image of the sliced as-prepared Au/TB-ZnO. (c) Backscattered electron image and XRD pattern (in log scale) of Au/TB-ZnO pretreated under oxygen at 200 °C. The two red bars in the XRD indicate the peak positions of gold if it existed. (d) TEM image of the sliced corresponding sample c (the particle size in d is 1.9 ± 0.3 nm).

surfaces mainly of {101̅0} nonpolar facets (see Figure S1b). We deposited gold onto TB-ZnO from a solution of AuCl4−. Interestingly, there was no obvious gold particles could be

Figure 2. (a) HAADF STEM image of a crushed piece of Au/TB-ZnO catalyst at low-magnification, showing some representative mesopores (darker contrast with arrows). (b) STEM image taken from the edge of Au/TB-ZnO piece. Arrows 1 and 2 are regarded as patch- and cluster-like Au in ZnO, respectively. (inset) Fourier transform of the corresponding atomic-resolution image, showing the wurtzite structure of ZnO. (c) STEM image showing an enlarged image of clustered Au-doping in ZnO and a hemispherical gold NP attached on the ZnO surface (arrow 3). (d) EDX analysis of the cluster-like Au in ZnO. (e) EDX analysis of the patch-like Au in ZnO, nicely revealing the corresponding Au-doping nature into ZnO. (f) Enlarged STEM image of part c, exhibiting a clustered Au-doped ZnO region notably following the lattice of the wurtzite. (g) Schematic representation of the Au-doped TB-ZnO structure. 117

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis interaction as revealed by the contrast modulation at the junction (arrow 3, the inset of Figure 2c). We note here that the patch of continuous bright contrast is not due to focussensitive artificial contrast changes, since the specimen thickness (40−50 nm) allows the unveiling of atomic-scale structural detail only at a very narrow focus range. Moreover, the atomically resolved bright contrasts in the individual ZnO nanoparticles (arrows 1 and 2, Figure 2b), reflecting the presence of Au, are nicely arrayed with the Zn-matrix contrast (oxygen otherwise invisible due to the noticeable Z difference with Au and Zn) free from any symmetry-forbidden reflection in the corresponding fast Fourier-transform pattern (inset, top right panel; [100] projection). This remarkable structural coherence between the nominal Zn columns and the potentially Au-implanted ones appears to suggest the substitutional Au in ZnO. We then carefully examined this possibility by using spatially resolved chemical imaging through STEM in conjunction with energy-dispersive X-ray spectroscopy (EDX) (Figure 2d and 2e; Au, Lα1; Zn, Kα1,2).22 To obtain a distinguishable gold chemical mapping image from Au-doped ZnO matrix, we empirically acquired the atomically sensitive STEM-EDX maps (Figures 2d and e) under practical collection condition of pixel dwell time ∼20 ms. However, with an increase in pixel dwell time, no spatial-resolved image at atomic resolution can be achieved due to the non-negligible weighting factor of the stray-electron contribution. Notably, the bright STEM-HAADF contrasts in Figure 2d are consistent with the Au-chemical contrasts, firmly establishing that the onset of the bright STEM-HAADF contrasts does arise from the Au therein. With these structural (Figure 2b and c) and chemical (Figure 2d and e) characterizations, it is now unambiguous that the substitution of Zn by Au indeed occurs in ZnO mesocrystals. Moreover, these patch- (e.g., arrow 1 in Figure 2b) and clusterlike (e.g., arrow 2 in Figure 2b and c) Au atoms in the ZnO are most likely located in the close vicinity of the ZnO free surface, where a substitution of Zn by Au would be most feasible.23 The enlarged image of Figure 2c shown in 2f clearly shows Ausubstituted area nicely follows the same wurtzite lattice of ZnO, with the corresponding substitution being schematized in Figure 2g. To characterize the catalyst ensemble, X-ray absorption spectroscopy was conducted to examine the valence state and local structures of substitutional gold. The Au/TB-ZnO sample (Figure 3a), curve iii, gave a lower white line intensity versus asprepared Au/TB-ZnO, curve iv, indicating more Au(0) after 200 °C heating. This is consistent with the STEM observation of the emergence of Au nanoparticles after 200 °C pretreatment. Figure 3b shows the structural fitting of the k3-weighted extended X-ray absorption fine structure (EXAFS) spectrum with the corresponding structural information given in Table 1. Indeed, a satisfactory fitting in Figure 3b could not be achieved without the incorporation of Au−O and Au−O−Zn bonds in addition to the metallic Au−Au bond of ∼2.86 Å in the Au metal (visibly contributed by the Au NPs). Remarkably, the refined Au−O and Au−Zn distances of ∼1.97 and ∼3.21 Å (Table 1), respectively, are quite close to the typical Zn−O (1.974 Å) and Zn−Zn (in Zn−O−Zn, 3.208 Å) distances in wurtzite ZnO. This is more evidence that Au atoms have doped into the lattice of ZnO, with substitution at Zn-site. We also note that the coordination number of Au−O (2.28) is less than 4 which would be the case if all Au−O are in ZnO lattice substituting Zn.24 This indicates many of the Au−O are from

Figure 3. (a) Normalized XANES spectra Au LIII-edge of (i) Au foil, (ii) (CH3)2SAuCl compounds, (iii) Au/TB-ZnO, (iv) as-prepared Au/ TB-ZnO, and (v) HAuCl4. (b) Fitted result of k3-weighted EXAFS spectrum in R-space at Au LIII-edge (without phase correction) taken from Au/TB-ZnO catalyst. (XANES: X-ray absorption near edge structure. EXAFS: extended X-ray absorption fine structure.)

surface. The Au−O−Au coordination number (3.96) is quite close to 4 which corroborates with the clustering phenomenon of doped Au within ZnO as we observed in Figure 2. Also, contributions from metallic gold are observed due to the presence of Au nanoparticles on the surface. It can be concluded that the ZnO mesocrystals are, in effect, decorated by Au NPs on the surface and heavily doped by Au ions in ZnO lattice of Au/TB-ZnO. Apparently, the gold nanoparticles emerged from under the mixed oxide of ZnO:Au after a heating at 200 °C. The Au/TB-ZnO catalysts (at 4.1 wt % Au loading), asprepared and pretreated under oxygen at 200 °C, were tested in CO oxidation in the temperature range from −50 to 90 °C with a very small amount of only 3.5 mg Au/TB-ZnO at a high flow velocity (GHSV: 1.54 × 106 mL/gcat·h). The as-prepared catalyst, where no Au NP was found on surface, shows nearly zero activity (Figure 4a). However, the activity for the Au/TBZnO sample pretreated at 200 °C gave an excellent activity. The absolute catalytic rate for the catalyst at 20 °C is remarkably high at 8.21 molCO/gAu·h (or 9.35 × 10−5 molCO/ gcat·s) (Table S2). The CO oxidation activity is outstanding and stable, much higher than our recently reported Au/ZnO nanorod catalyst systems.25 In fact, its activity is the highest one compared to the previously known records for Au/TiO2 and Au/FeTiO2 system (Table S2). We have done additional measurement of reaction rates at a GHSV of 1.8 × 105 mL/gcat· h, i.e., at a 8.6 times lesser space velocity. The conversions are 100% except for the lowest temperature of 25 °C (Figure S3). Also, we treated the pretreated Au/TB-ZnO sample with cyanide (KCN) to wash away the Au nanoparticles. The resulting catalyst also shows no activity (Figure 4a) and no remaining Au NP was observed on the surface of ZnO (Figure 4b). A subsequent heat treatment on the cyanide-treated Au/ TB-ZnO results some gold nanoparticles re-emerging on the surface of TB-ZnO (Figure 4c). The excellent catalytic activity of the Au/TB-ZnO is associated with a very small gold nanoparticle size (∼2 nm) due to strong metal−support interaction (SMSI). A confinement effect due to the smaller domain of ZnO nanocrystals probably give a help on restricting and preventing the sintering of Au NPs upon thermal pretreatments. To demonstrate the capability of TB-ZnO for finely dispersing Au, a heavily loaded gold catalyst was prepared at 18 wt % of Au-loading (ICP-MS determined).The size of Au NPs still gave an impressively small value of ∼3.8 nm (Figure 118

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis Table 1. Refined Results of the EXAFS Analysis at Au LIII-Edgea sample

shell

CN

Au foil Au/TB-ZnO

Au−Au Au−O Au−Au Au−Au Au−Zn

12 2.28 3.88 3.96 2.06

R (Å)

σ2 × 103 (Å2)

ΔE0 (eV)

r (%)

± ± ± ± ±

6.87 3.00 6.51 6.51 6.51

2.9 8.7 8.7 8.7 8.7

0.62 0.67

2.86 1.97 2.86 3.11 3.21

0.02 0.01 0.02 0.03 0.03

CN: coordination number for the absorber−backscatter pair. R: average absorber−backscatter distance. σ2: Debye−Waller factor. ΔE0: inner potential correction. The data ranges used for data fitting in k-space and R-space are Δk 3.1−11.8 1/Å and ΔR 1.3−3.7 Å, respectively.

a

Figure 5. (a) Cathodoluminescence spectra for TB-ZnO (brown) and Au/TB-ZnO (dark cyan) samples. (b) EPR spectra of commercial ZnO powder (Comm. ZnO, black), TB-ZnO particles (TB-ZnO, brown), and Au/TB-ZnO catalyst (dark cyan). Spin counts were obtained.

By comparison, the decreasing of green emission at 510 nm after Au loading can be attributed to some Zn vacancies at nonpolar surfaces have been occupied by gold, once we accept the assignment of 510 nm to Zn-vacancies.26 Electron paramagnetic resonance (EPR) measurement was further conducted to assess the concentration of oxygen vacancy (Vo). The result indicates TB-ZnO has an exceptionally strong signal at g = 1.960 (Figure 5b), which is assigned to singly ionized oxygen vacancy (Vo·).25,29 Spin count measurement of the defects gives 2.7 × 1013/mg for TB-ZnO, whereas the spin count for the commercial ZnO powder is only 2.2 × 1011/mg, less than 1% of our TB-ZnO sample. This indicates that the ZnO mesocrystals obviously possess much more oxygen vacancies than crystalline ZnO. A much reduced EPR signal of Vo· with 2.8 × 1012/mg spin count was taken from gold-deposited TB-ZnO. This decreasing is from the Vo· turning into EPR silent as an electron transfer from gold nanoparticle upon oxygen pretreatment.25 Recently, Cui et al. demonstrated Mo-doped CaO(nonreducible) can help activate adsorbed oxygen.30 Here, Au-doped ZnO may also help the oxygen activation in catalysis. A series of density functional theory (DFT) calculations were further performed to understand the strong doping of Au in ZnO and its easy migration due to Zn-vacancy defects in TBZnO. The doping (Edop) and formation (Eform) energies for various Au-doped ZnO models were calculated according to eqs 1 and 2:

Figure 4. (a) CO conversions of Au/TB-ZnO, cyanide-treated and asprepared catalysts are plotted in the colors of dark cyan, brown, and black, respectively. (b) TEM image of cyanide-treated Au/TB-ZnO. (c) TEM image of the cyanide-treated sample (part b) which subsequently underwent one more heat treatment, showing that some gold nanoparticles re-emerged on the surface of TB-ZnO.

S4). Simple estimates from sizes indicate that the Au NPs on surface could not be accounted for the large 18% loading. There seems to be a substantial fraction of Au still residing within the ZnO lattice. To elucidate the unique structural of TB-ZnO, TB-ZnO particles were examined by cathodeluminescence (CL) spectroscopy, which presents a sharp emission at 385 nm due to near band edge (NBE) transition of ZnO, and a green broad emission at 510 nm from Zn-vacancies near the {101̅0} nonpolar surfaces (Figure 5a).26 We should note here that the assignment of green fluorescence of ZnO was a debated issue. There were reports assigning the green emission to oxygen vacancy.27 However, more recent spectroscopic and computational studies have given very strong evidence for associating the green fluorescence (510 nm) of ZnO as transition involving Zn vacancy acceptor defect levels.26,28 The Zn-vacancy interpretation have been proved convincingly by positron annihilation studies.26c,d On the other hand, the green-yellow emission near 560 nm could be assigned due to O-vacancy. Moreover, we have concluded from EPR that there are 2 orders of magnitude higher concentration of oxygen vacancy than normal ZnO. By charge balance, there should also be large amount of Zn-vacancy.

Edop = Etotal(Audop) − Etotal(VZn) − nE(Au fcc)

(1)

Eform = Etotal(Audop) − Etotal(ZnOwurt ) − nE(Au fcc) + nE(Znhcp) 119

(2) DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis

The highest m/n ratio in our simulation models is 1.5 in which 4 Au atoms doped in a tetrahedron conformation belong to a 6/4 class. According to Figure 6a, the class having the lowest doping energy is 3/3, in which the 3 doping Au atoms form a triangle, whereas higher m/n ratios (m/n > 1) tend to decrease energy gain from Au doping. Hence, Au doping atoms prefer to cluster together and form dimers, trimmers, and small clusters, but the tendency to make bigger Au clusters with m/n ratios higher than 1 is getting lower. The corresponding formation energies (Figure S5) also indicate a similar conclusion with the lowest value at m/n = 5/4. In Figure 6b, we show the hopping energy barriers of one Au or Zn atom hoping to an adjacent Zn-vacancy site. The barrier height of Zn-hopping is more than twice of Au-hopping. Hence, we conclude that doped Au atoms migrate much faster than Zn atoms in TB-ZnO. Combining these results, we propose a plausible Au doping process. Because the Au-doping energies are negative for defect-rich ZnO and Au atoms can migrate much faster than Zn atoms, Au atoms tend to fill the Zn vacancies before the defect-rich ZnO could anneal itself to a perfect ZnO crystal. However, the positive formation energies(Figure S5) for nondefective ZnO imply that when there are enough thermal energies for Au and Zn atoms to rearrange, Au atoms would be expelled out of a perfect ZnO lattice to form nanoparticles on its surface. For further understanding the extensive dopings of Au in the lattice of as-synthesized Au/TB-ZnO, we employed X-ray photoelectron spectroscopy (XPS) and Raman spectroscopy to study the catalysts. Figure 7a shows the XPS spectra of asprepared sample with 7.3 wt % gold-loading, i.e. before any high temperature treatment. Because of the overlapping of Zn 3p and Au 4f peaks, the deconvolution of the XPS spectra was carefully done with contributions from Zn 3p and Au species (Au(I), Au(III), and Au(0)). First, in the as-prepared Au/TBZnO (Figure 7a), one can see little Au(0) signal from reduced Au nanoparticles. This agrees with our TEM observation that no Au nanoparticles could be observed on the surface of asprepared catalyst. The Au species as detected by XPS, existed mostly as Au(III) (70.4%) and Au(I) (27.4%) species (Table S3). Given the high Au loading of 7.3%, much of the Au species should still exist inside the lattice as we proposed. Also, it is striking that the Zn 3p (Figure 7b) profile seems to be composed of two components, partially from wine and blue deconvolution spectra in Figure 7a. Previously, Kundu et al. reported a ZnO-Au nanohybrids and in their XPS spectra a Zn−Au contact interaction was assigned for the high binding energy component of Zn 3p.31 We also see this high binding energy component in Figure 7a, as indicated by the dashed arrow. Compared to TB-ZnO (without Audoping, as shown in Figure S6), the Zn 3p high binding energy side cannot be fitted well without a contribution of Zn−Au component (wine color). Upon thermal treatment at 200 °C under oxygen, Au nanoparticles re-emerge from TB-ZnO to give clear Au(0) peaks (73.6%) in XPS (Figure 7b). The very strong Au(0) peaks cannot solely be the result of transformation from surface Au(III) and Au(I) species. The emerging of Au from lattice is a better explanation. After sample pretreatments at 200 °C−O2 the catalysts still show a little amount of Au(III) (8.7%) and Au(I) (17.7%) (Table S3). The support, i.e., ZnO:Au therefore became a reducible support and further may play a role in metal−support interaction during the catalysis. Raman spectroscopy was measured to estimate the heavy doping of Au in TB-ZnO (Figure 7c). A sharp and strong

where Etotal(Audop), Etotal(ZnOwurt), and Etotal(VZn) are the model total energies of Au-doped ZnO, pristine wurtzite ZnO, and defective ZnO with Zn vacancies. E(Aufcc) and E(Znhcp) are one atom self-energies of respective face-centered cubic (fcc) and hexagonal close-packed (hcp) crystalline metal, and n stands for the number of Au atoms for doping or the number of Zn atoms for removal. The calculation parameters used are detailed in Table S1. In the calculation of doping energy, the defects were constructed inside the ZnO structure (not on the surface) and the energy optimization is local, i.e. no diffusion allowed. It is found that all the doping energies are negative, but the formation energies are positive (more than 4 eV on average, as shown in Figure S5). Negative doping energies mean Au atoms prefer to occupy existing Zn vacancies. However, the positive formation energies (defined with respect to perfect ZnO cystal) indicate ZnO sample and Au atoms would eventually become phase separated if there are kinetic route for some partial self-healing. In Figure 6a, the doping energies are presented in terms of the

Figure 6. Theoretical calculation of Au-doped ZnO by DFT. (a) Correlation plots of doping energy versus the ratio of no. of Au−Au bonds (m) to no. of doping Au atoms (n). The symbols with yellow balls and gray sticks indicate the corresponding geometry structure of Au doping. (b) Energy barriers for Au and Zn hopping to the adjacent Zn-vacancy site. (b inset) Cartoon scheme for Au or Zn hopping.

m/n class, which is defined as the ratio of total Au−Au bonds number to total number of doping Au atoms. One Au−Au bond is counted if two Au atoms are doped at adjacent Znvacancy sites; other situations are regarded as no Au−Au bond. Hence, a m/n class means there are m Au−Au bonds in n-Auatom doped ZnO model. A higher m/n ratio indicates the Au atoms forming a cluster instead of dispersing into ZnO bulk. 120

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis

catalytic activity upon supporting gold.1 In our Au/TB-ZnO however, there are still substantial amount of Au(III) doped inside the support when Au nanoparticles emerged on surface as catalyst. In a way, the doped support TB-ZnO:Au could function as a reducible metal oxide because of the presence of Au−O−Zn within support. A gold-assisted redox mechanism may thus be responsible for the high activity in CO oxidation.35 A detailed study of catalytic mechanism of this catalytic system is undergoing in our laboratory. Finally let us summarize the experimental evidence for the heavy-doping of Au in ZnO we provided in this paper.

Figure 7. X-ray photoelectron spectroscopy analysis of (a) as-prepared 7.3 wt % Au/TB-ZnO catalyst and (b) oxygen-pretreated (at 200 °C) 4.1 wt % Au/TB-ZnO catalyst. The blue curve stands for background. Wine, navy, dark cyan, magenta, and dark yellow curves represent Zn− Au bonded Zn 3p, Zn 3p for ZnO2, zerovalent Au 4f, single-valent Au 4f, trivalent Au 4f peaks, respectively. (c) Raman spectra of (i) ZnO nanorods, (ii) commercial-ZnO powder, (iii) calcined TB-ZnO support, 4.1 wt % Au/TB-ZnO, (iv) as-prepared, and (v) with 200 °C O2-pretreatment. The red column represents a position at 542 cm−1 for the indication of the broaden peak from Au-doped TB-ZnO.

1. STEM-HAADF observation showed contrast change over a large patch which can only be explained by Audoping of ZnO lattice. 2. Large contributions due to Au−O−Zn coordination are observed from fitting of EXAFS data. 3. Re-emergence of Au nanoparticles is observed upon heating of catalyst after KCN washing of surface Au species. 4. DFT computational study of Au-doping of defect-rich ZnO shows extensive doping is energetically favorable. Migration of Au in Au-doped ZnO is also feasible. 5. XPS study shows Au−Zn linkage in Au/TB-ZnO catalyst with the absence of Au(0) before heating but emergence of Au(0) after heating in oxygen. 6. Raman study shows that the pattern for Au-doped ZnO is very different from pure ZnO. 7. A very heavy doping level at ∼18% Au is obtained without observing many Au nanoparticles on the surface. The total combined evidence for rich Au-doping in mesocrystal of ZnO is overwhelming. This method is a very unique for Au nanoparticle deposition from below the surface of mixed Au, Zn oxides. Finally, we obtain a Au nanocatalyst that is highly active in the catalytic oxidation of CO.

peak at ca. 438 cm−1 (E2(high) mode) and the peak at 333 cm−1 (E2H−2L) would be due to a good crystallinity of commercial and nanorods of ZnO.32 For the Au/TB-ZnO catalyst, the E 2(high) and E 2H−2L modes were silenced because the incorporation of Au atoms into the ZnO crystal destroyed the crystalline symmetry of ZnO. Furthermore, a broad peak at ca. 542 cm−1 in Au/TB-ZnO only (Figure 7c iv for as-prepared, v for 200 °C-treated sample) was assigned to quasi-LO mode which has been attributed to intrinsic host-lattice defects related to doping.33 The fact that as-prepared sample show the same Raman pattern as 200 °C-treated sample indicates heavy doping of Au into ZnO mesocrystal occurred in the synthesis stage already. Because of the high lattice energy of most of metal oxides, gold would not be able to disperse into a perfect metal oxide crystal. Indeed, in practically all of the supported Au catalysts reported so far, the deposited gold was only found on top of the support. As we have shown in our calculation (Figure S5), this is also true for normal ZnO crystal with no defects. This is why normally one only obtained Au on top of the usual ZnO supports, not within ZnO.25,31,34 We discover however heavy doping of Au within ZnO mesocrystal lattice because of the large amount pre-existing vacancies. It is known most of the nonreducible metal oxides, such as ZnO, do not give good

In summary, we have successfully employed defect-rich mesocrystals of ZnO (TB-ZnO) as a novel support for a heavy doping of gold. The large amount of Zn- and O-vacancy defects in TB-ZnO give a strong driving force for introducing gold into the lattice of ZnO. STEM, EXAFS, XPS, and Raman spectroscopy gave strong evidence for the heavy doping of Au in the lattice of ZnO. The ZnO mesocrystal thus behaved in a sponge-like fashion for accommodating large amount of gold atoms. These Au atoms eventually move out of ZnO lattice to form highly dispersed Au nanoparticles around 2 nm after moderate thermal treatment. A test of CO oxidation activity gave a record of catalytic activity as compared with other Aubased catalysts. The Au/TB-ZnO catalyst may be also highly active for other Au-based catalytic reactions. In this work, the use of ZnO mesocrystal of high concentration of defects give new meaning of the strong metal−support interaction between gold and ZnO support,25 e.g., the interaction is so strong that mixed oxide of Au/ZnO was formed initially and gold nanoparticles could then emerged from within the lattice. The use of mesocrystal of metal oxides as support for catalyst may provide a new approach for designing and developing new catalysts by controlling strong metal−support interactions.

■

121

CONCLUSIONS

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122

Research Article

ACS Catalysis

■

(17) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537− 541. (18) Blöchl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 17953−17979. (19) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (20) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188−5192. (21) Henkelman, G.; Jónsson, H. J. Chem. Phys. 2000, 113, 9978− 9985. (22) (a) Chu, M. W.; Liou, S. C.; Chang, C. P.; Choa, F. S.; Chen, C. H. Phys. Rev. Lett. 2010, 104, 196101. (b) Chang, C. P.; Lin, J. G.; Jeng, H. T.; Cheng, S. L.; Pong, W. F.; Shao, Y. C.; Chin, Y. Y.; Lin, H. J.; Chen, C. W.; Yang, J. R.; Chen, C. H.; Chu, M. W. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 075129. (23) (a) Yan, Y.; Al-Jassim, M. M.; Wei, S.-H. Appl. Phys. Lett. 2006, 89, 181912. (b) Zhang, X. D.; Wu, P.; Shen, Y. Y.; Zhang, L. H.; Xue, Y. H.; Zhu, F.; Zhang, D. C.; Liu, C. L. Appl. Surf. Sci. 2011, 258, 151− 157. (24) Abrahams, S. C.; Bernstein, J. L. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1969, B25, 1233−1236. (25) Liu, X. Y.; Liu, M.-H.; Luo, Y.-C.; Mou, C.-Y.; Lin, S. D.; Cheng, H.; Chen, J.-M.; Lee, J.-F.; Lin, T.-S. J. Am. Chem. Soc. 2012, 134, 10251−10258. (26) (a) Fabbri, F.; Villani, M.; Catellani, A.; Calzolari, A.; Cicero, G.; Calestani, D.; Calestani, G.; Zappettini, A.; Dierre, B.; Sekiguchi, T.; Salviati, G. Sci. Rep. 2014, 4, 5158. (b) Leung, Y. H.; Chen, X. Y.; Ng, A. M. C.; Guo, M. Y.; Liu, F. Z.; Djurišić, A. B.; Chan, W. K.; Shi, X. Q.; Van Hove, M. A. Appl. Surf. Sci. 2013, 271, 202−209. (c) Singh, V. P.; Das, D.; Rath, C. Mater. Res. Bull. 2013, 48, 682−686. (d) Č ížek, J.; Valenta, J.; Hruška, P.; Melikhova, O.; Procházka, L.; Novotný, M.; Bulíř, J. Appl. Phys. Lett. 2015, 106, 251902. (27) Qian, K.; Huang, W. X.; Fang, J.; Lu, S. S.; He, B.; Jiang, Z. Q.; Wei, S. Q. J. Catal. 2008, 255, 269−278. (28) Niu, W.; Zhu, H.; Wang, X. F.; Ye, J. D.; Song, F. Q.; Zhou, J. F.; Gu, S. L.; Shi, Y.; Xu, Y. B.; Zhang, R. Appl. Phys. Lett. 2015, 107, 021902. (29) Drouilly, C.; Krafft, J.-M.; Averseng, F.; Casale, S.; Bazer-Bachi, D.; Chizallet, C.; Lecocq, V.; Vezin, H.; Lauron-Pernot, H.; Costentin, G. J. Phys. Chem. C 2012, 116, 21297−21307. (30) Cui, Y.; Shao, X.; Baldofski, M.; Sauer, J.; Nilius, N.; Freund, H.J. Angew. Chem., Int. Ed. 2013, 52, 11385−11387. (31) Kundu, P.; Singhania, N.; Madras, G.; Ravishankar, N. Dalton Trans. 2012, 41, 8762−8766. (32) Wang, X.; Xu, J.; Yu, X.; Xue, K.; Yu, J.; Zhao, X. Appl. Phys. Lett. 2007, 91, 031908. (33) Phan, T.-L.; Vincent, R.; Cherns, D.; Dan, N. H.; Yu, S.-C. Appl. Phys. Lett. 2008, 93, 082110. (34) Noei, H.; Birkner, A.; Merz, K.; Muhler, M.; Wang, Y. M. J. Phys. Chem. C 2012, 116, 11181−11188. (35) Li, L.; Wang, A. Q.; Qiao, B. T.; Lin, J.; Huang, Y. Q.; Wang, X. D.; Zhang, T. J. Catal. 2013, 299, 90−100.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.5b02093. Additional SEM/TEM images, XPS spectrum for bare TB-ZnO, computed formation energy for Au substituted TB-ZnO, catalytic conversion at slower flow rate, calculation parameters, XPS analysis results for Au/TBZnO catalysts, and comparison table for catalytic activities (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (M.-W.C.). *E-mail: [email protected] (C.-Y.M.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was funded by MOST (Ministry of Science and Technology), Taiwan. We thank C.-Y. Tang, S.-J. Ji, and C.-Y. Chien of the Instrument Center of MOST for assistance in SEM/EDX and TEM observations. X. Liu thanks the National Science Foundation of China (21303194, 21476227) and the Hundred Talents Program of DICP for support. X-ray absorption experiments were performed at National Synchrotron Radiation Research Center (NSRRC).

■

REFERENCES

(1) Liu, X. Y.; Wang, A.; Zhang, T.; Mou, C.-Y. Nano Today 2013, 8, 403−416. (2) Comotti, M.; Li, W.-C.; Spliethoff, B.; Schüth, F. J. Am. Chem. Soc. 2006, 128, 917−924. (3) (a) Yoon, B.; Häkkinen, H.; Landman, U.; Wörz, A. S.; Antonietti, J.-M.; Abbet, S.; Judai, K.; Heiz, U. Science 2005, 307, 403− 407. (b) Wang, Y.-G.; Yoon, Y.; Glezakou, V.-A.; Li, J.; Rousseau, R. J. Am. Chem. Soc. 2013, 135, 10673−10683. (4) Widmann, D.; Behm, R. J. Angew. Chem., Int. Ed. 2011, 50, 10241−10245. (5) Chua, Y. P. G.; Gunasooriya, G. T. K. K.; Saeys, M.; Seebauer, E. G. J. Catal. 2014, 311, 306−313. (6) Schumann, J.; Eichelbaum, M.; Lunkenbein, T.; Thomas, N.; Á lvarez Galván, M. C.; Schlögl, R.; Behrens, M. ACS Catal. 2015, 5, 3260−3270. (7) McFarland, E. W.; Metiu, H. Chem. Rev. 2013, 113, 4391−4427. (8) Pala, R. G. S.; Metiu, H. J. Catal. 2008, 254, 325−331. (9) Morgan, K.; Cole, K. J.; Goguet, A.; Hardacre, C.; Hutchings, G. J.; Maguire, N.; Shekhtman, S. O.; Taylor, S. H. J. Catal. 2010, 276, 38−48. (10) Sutter, E. A.; Tong, X.; Jungjohann, K.; Sutter, P. W. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 10519−10524. (11) Liu, Y.; Zhang, Y.; Wang, J. CrystEngComm 2014, 16, 5948− 5967. (12) Cö lfen, H.; Antonietti, M. Mesocrystals and Nonclassical Crystallization; John Wiley & Sons, Ltd: Chichester, UK, 2008. (13) Bian, Z.; Tachikawa, T.; Zhang, P.; Fujitsuka, M.; Majima, T. Nat. Commun. 2014, 5, 3038. (14) (a) Peng, Y.; Wang, Y.; Chen, Q.-G.; Zhu, Q.; Xu, A. W. CrystEngComm 2014, 16, 7906−7913. (b) Wang, J.; Hou, S.; Zhang, L.; Chen, J.; Xiang, L. CrystEngComm 2014, 16, 7115−7123. (15) Wang, M.; Zhang, Y.; Zhou, Y.; Yang, F.; Kim, E. J.; Hahn, S. H.; Seong, S. G. CrystEngComm 2013, 15, 754−763. (16) Liu, M.-H.; Tseng, Y.-H.; Greer, H. F.; Zhou, W.; Mou, C.-Y. Chem. - Eur. J. 2012, 18, 16104−16113. 122

DOI: 10.1021/acscatal.5b02093 ACS Catal. 2016, 6, 115−122