Effect of Support on the Activity of Ag-based Catalysts for Formaldehyde Oxidation

Ag-based catalysts with different supports (TiO2, Al2O3 and CeO2) were prepared by impregnation method and subsequently tested for the catalytic oxidation of formaldehyde (HCHO) at low temperature. The Ag/TiO2 catalyst showed the distinctive catalytic performance, achieving the complete HCHO conversion at around 95 °C. In contrast, the Ag/Al2O3 and Ag/CeO2 catalysts displayed much lower activity and the 100% conversion was reached at 110 °C and higher than 125 °C, respectively. The Ag-based catalysts were next characterized by several methods. The characterization results revealed that supports have the dramatic influence on the Ag particle sizes and dispersion. Kinetic tests showed that the Ag based catalyst on the TiO2, Al2O3 or CeO2 supports have the similar apparent activation energy of 65 kJ mol−1, indicating that the catalytic mechanism keep immutability over these three catalysts. Therefore, Ag particle size and dispersion was confirmed to be the main factor affecting the catalytic performance for HCHO oxidation. The Ag/TiO2 catalyst has the highest Ag dispersion and the smallest Ag particle size, accordingly presenting the best catalytic performance for HCHO oxidation.

, Na-promoted Pd/TiO 2 9 and Au/CeO 2 10 catalysts could convert HCHO to harmless CO 2 and H 2 O at ambient temperature even at high space velocity. However, the high price of Pt, Au, Pd blocks their wide application and promotes the flourishing studies of supported Ag-based catalysts [11][12][13][14][15][16][17] which has much lower price but shows the considerable efficiency for HCHO oxidation in low temperature. Qu et al. 14 prepared Ag/SBA-15 by post-grafting method and it displayed 100 °C as the complete conversion temperature. A recent study reported that 1.7% K-Ag/Co 3 O 4 could catalyze the oxidation of HCHO with 100% conversion at 70 °C 18 .
The activities of supported Ag catalysts are dramatically influenced by the supports, which have been attributed to the different states of sliver species on the supports 13 , diverse contents of the subsurface oxygen 19 , various oxygen storage and redox capacities 20 , etc. Ma et al. 12 prepared Ag/CeO 2 catalysts which exhibited distinct activities when the supports CeO 2 were of diversity in morphology led by different synthesized methods. Chen et al. 13 observed the dramatic variation in distribution of Ag nanoparticle diameters, redox capacities and HCHO desorption properties when Ag was loaded on the supports such as TiO 2 , MCM-41, SBA-15, etc. Shi et al. 17  on Ag catalysts and found that Ag/MnO x -CeO 2 showed much higher activity since Mn-Ce oxides could accelerate the partial oxidation of HCHO into HCOO − .
Due to the enormous distinction in activity displayed in supported Ag catalysts, it is of great significance to reveal the internal discipline of support effect. TiO 2 , γ -Al 2 O 3 and CeO 2 have been proved to be stable support for several catalytic reactions such as HCHO oxidation 12,[21][22][23] , NO x selective degeneration [24][25][26][27] , formic acid decomposition 28 , hydrogen production 29 , etc. Therefore, in this work, we prepared the Ag-based catalysts with supports of TiO 2 , γ -Al 2 O 3 and CeO 2 by impregnation method and then compared their performance for the catalytic oxidation of HCHO at low temperature. The dramatic difference about the catalytic activity was clearly observed on the three catalysts. Ag/TiO 2 exhibited the best activity, archiving 100% conversion of 110 ppm HCHO at around 95 °C with a gas hourly space velocity of 100 000 mL (g cat ·h) −1 . The Brunauer-Emmett-Teller (BET), X-ray diffraction (XRD), High-resolution transmission electron microscope (HRTEM), UV-vis diffuse reflectance spectroscopy (DRS), X-ray photoelectron spectroscopy (XPS), Temperature-programmed reduction (TPR) were next measured to characterize the catalysts. Based on the results, the internal rules influencing the activity of Ag based catalysts were discussed and elucidated.

Results and Discussion
Performance Test of HCHO Catalytic Oxidation. Figure 1 displays the HCHO conversions as a function of temperature over the Ag/TiO 2 , Ag/Al 2 O 3 and Ag/CeO 2 catalysts at a GHSV of 100 000 mL (g cat ·h) −1 with the inlet HCHO concentration of 110 ppm. The temperature dependence of conversion rates were dramatically affected by the supports, and the catalytic activities followed the order of Ag/ TiO 2 > Ag/Al 2 O 3 > Ag/CeO 2 . The Ag/TiO 2 catalyst showed the highest HCHO conversion rate at each testing temperature point and reached the 100% conversion at around 95 °C. The Ag/Al 2 O 3 sample exhibited much lower activity with complete conversion at 110 °C, and the HCHO over Ag/CeO 2 could not be entirely decomposed even when the temperature was ramped to 125 °C.
We also prepared the Ag/TiO 2 with different ratio as 4, 6, 10, 12 wt% and their activities are shown in Figure S1 and S2 (supporting information). The activity as a function of content displayed a typical volcano curve and the best activity emerges when the content was 8 and 10 wt%, which is the reason for choosing 8 wt% to implement this research. The pure supports (TiO 2 , Al 2 O 3 and CeO 2 ) showed no activity for HCHO oxidation in the testing temperature range (35-125 °C) (data not shown). Therefore, the loaded Ag should be mainly responsible for the huge distinction in catalytic performance. Similar support effect was also found among the Pt-based catalysts with support as TiO 2 , SiO 2 , Ce & Zr oxides for HCHO oxidation mainly due to the Pt dispersion variation on different supports 30 . A theoretical study on Au-based catalysts showed that different supports could affect the charge transfer from Au to support materials, therefore showing the different activity for glucose oxidation 31 . Due to the numerous explanations by many studies, investigating the factors that lead to such diversity of catalytic performances from the same Ag active species would be in considerable significance.

Structure Analysis.
To check the possibility of influence from surface physical property, the specific surface areas (BET), average pore diameters, and total pore volumes of the three samples were first measured, subsequently the specific reaction rate (R s ) at 80 °C were calculated. As shown in Table 1, the Ag/TiO 2 catalyst had the lowest surface area but presented the highest catalytic activity, leading to its largest R s of 2.18 nmol s −1 m −2 at 80 °C among the three samples. The Ag/Al 2 O 3 displayed the higher surface area, apparent activity and also R s than the Ag/CeO 2 catalyst. These results indicate that the physical properties have the limited influence on catalytic performance.
XRD measurement was next carried out to investigate the crystallographic structures. As shown in Fig. 2, the intensive peaks in the characteristic locations of TiO 2 with anatase structure (JCPDS 71-1167) and CeO 2 (JCPDS 43-1002) indicated the dominating content of the respective supports. Unlike the former two supports, γ -Al 2 O 3 (JCPDS 74-2206) was formed with amorphous structure, leading to the poor intensity in the pattern. In addition, the peaks of metallic silver could be also detected, which could be indexed to face-centered cubic crystal structures, while no diffraction due to silver oxide was observed. Two peaks (shown in the magnified inset graph) located at 38.1° and 44.3° could be assigned to diffraction line of the (111) and (200) planes of the metallic silver crystal, respectively (JCPDS 87-0597). The Ag (111) peak at 38.1° in Ag/TiO 2 pattern was overlapped with the TiO 2 (anatase) peaks of (103), (004), (112) facets. It was still observed that the peak intensities corresponding to silver were much different over three samples. Noting that the intensity of XRD pattern reflects the crystallinity and the crystal size, we therefore speculate that the Ag particle sizes over each sample are much different among samples. The largest Ag average particle size might emerge on Ag/CeO 2 which displays the most intensive peak for Ag crystal and the smallest Ag average particle size should exist on Ag/TiO 2 sample.
We next implemented the HR-TEM measurement to investigate the states and particles size distribution of Ag over three catalysts. The images shown in Fig. 3 are representative for the entire surface of the respective samples. As shown in the inset histogram of statistic results from 100 particles and the summary in Table 2, the Ag particles on Ag/TiO 2 displayed homogeneous distribution with a mean diameter of 3.4 nm, being close to the previous result 13 . In contrast, the Ag particles on Ag/Al 2 O 3 (Fig. 3b) grew larger with the average size of 11.3 nm and the distribution was also not as uniform as that on Ag/TiO 2 . Being different with the former two samples, the Ag particle over Ag/CeO 2 ( Fig. 3c) was undistinguishable with that of support CeO 2 on HR-TEM image. Therefore, mapping was next carried out to find out the morphology and location of Ag species on CeO 2 . It could be seen from the mapping (insert in Fig. 3c1) that the supported Ag was poorly dispersed on CeO 2 with large particle size (>30 nm), being consistent with the previous study 32 which reported that a few large Ag particles on CeO 2 could even reach to 400 nm. Taking into account of the incomparability between bright field and mapping images, we did not make statistic for Ag/CeO 2 . However, the HR-TEM images still approved the deduction from   the XRD results that the Ag on CeO 2 was rather poorly dispersed. It is also confirmed that the Ag particle size on TiO 2 was the smallest among the three kinds of catalysts.
Chemical Characterization. UV-Vis measurement was carried out to discover the details of the state of supported Ag. Figure 4 depicts the UV-vis spectra of both the catalysts and the supports. According to the previous literatures 13, 33 , the absorbance below λ = 238 nm should be assigned to the 4d 10 to 4d 9 5s 1 transition of highly dispersed Ag + ions; the band at 238-320 nm is attributed to small Ag n δ+ clusters, and the absorbance above 320 nm is due to the existence of the metallic Ag 0 . From the comparison between the spectra of the Ag loaded catalysts and pure supports, it was revealed that the absorbance in the whole scale generally were more intensive after Ag was loaded. The bands could be deemed as a description of distribution of different Ag species 33 . It could be speculated that the Ag over the catalysts mainly presented the metallic crystal status since the band area in Ag 0 region was larger than others. Another interesting observation was that the peak intensities in 220-350 nm were enhanced over Ag/TiO 2 and Ag/Al 2 O 3 but diminished over Ag/CeO 2 after Ag loading. One speculation was that a few large Ag particles on CeO 2 , which presented the Ag 0 state, blocked the absorbing of UV but enhanced the absorbance of visible light.
To further check the electronic state of surface Ag, the XPS spectra were measured. Figure 5 shows the Ag3d spectra of the three catalysts. The Ag/TiO 2 showed two peaks at binding energies of 367.9 eV (Ag3d 5/2 ) and 373.9 eV (Ag3d 3/2 ), which were close to those expected for metallic silver (368.0 eV and 374.0 eV) 34 , indicating that the Ag on TiO 2 was mainly in the metallic state, being consistent with the XRD and UV-vis results. This finding could be corroborated by silver oxide's thermal decomposition study 35 , showing that the Ag 2 O will decompose to metallic silver with the thermal treatment at 400 °C, which is lower than our calcination temperature of 450 °C. Since different support would lead to different catalyst peak area of Ag3d 5/2  1 The area is a relative one when that of Ag/CeO 2 is set as 1.00. 2 Statistic mean particle size by HRTEM, 3 Calculated particle size from dispersion results by H 2 -TPR. and complex electronic environment, the Ag3d peaks showed a slight shift over each catalysts, which could be also observed by many documents 12,18,36 .
It was most noted that the intensities of Ag peaks for three catalysts displayed dramatic distinction. As shown in Table 2, the peak areas of Ag3d 5/2 for Ag/TiO 2 and Ag/Al 2 O 3 were 3.29 and 1.34 times of that for Ag/CeO 2 sample. The peak area is the function of the atom numbers of element when the XPS spectra are measured in the same condition for the same Ag element 37 . Thus, the peak area was exclusively related to the number of Ag atoms in the scanning volume. When the Ag was highly dispersed over support, there would be much Ag atoms exposed to the surface and emitted photoelectrons, consequently led to the high intensity of Ag3d spectra lines. Therefore, the XPS results showed that Ag species on TiO 2 was well dispersed and the sequence of dispersion degree should be Ag/TiO 2 > Ag/Al 2 O 3 > Ag/ CeO 2 , which is well consistent with the former characterization results.
One preceding work 38 had reported the size effect over gold catalysts, which showed markedly enhanced catalytic activity for CO oxidation as the gold particles are smaller than 10 nm. Another study 39 reported the remarkable size effect of ruthenium particles also for the catalytic oxidation of CO. Our study is also prone to the size effect as the indispensible factor that lead to the dramatic distinction in the catalytic activities for HCHO oxidation. The particle size on of Ag on TiO 2 is the smallest among the three supports, consequently contribute to the considerable enhancement of the activity. H 2 -TPR experiments were next performed to investigate the reducibility of three samples. Figure 6 presented the H 2 -TPR profiles of both the Ag catalysts and the corresponding supports. The pattern of Ag/TiO 2 sample showed two reduction peaks at around 80 °C and 435 °C. Since the reduction peak of Ag 2 O powder normally centered at above 100 °C 40 , the peak located at 80 °C should be mainly ascribed to reduction of oxygen species absorbed on the dispersed Ag surface. The peak at 435 °C should be due to the reduction of surface capping oxygen of TiO 2 . In contrast, the pure TiO 2 presented a single reduction peak at 549 °C, indicating that the Ag presence facilitated the reduction of TiO 2 surface oxygen species.
Different from pure TiO 2 sample, no reduction peak was observed on pure γ -Al 2 O 3 in the examined temperature range (− 50-600 °C). Consequently, the Ag/γ -Al 2 O 3 only showed one peak at 74 °C, which should be exclusively related to the supported Ag species. As for the Ag/CeO 2 , a low temperature reduction at 87 °C also appeared, similar to other two samples. The TPR profile of pure CeO 2 is characterized by wide peak in the region of 300 °C to 600 °C, which is typical for the reduction of surface capping oxygen species on CeO 2 41 . The Ag loading also induced the shift of the CeO 2 reduction to low temperature at around 200 °C, which is consistent with the previous study 12 .
Three Ag supported catalysts all displayed the similar reduction peaks at around 80 °C, indicating the similar oxygen mobility over these catalysts. The difference lied on the H 2 consumption amount, which were calculated from the profiles and presented in Table 2. The Ag/TiO 2 catalyst had the largest H 2 consumption amount with 166.9 μ mol g −1 . The Ag/Al 2 O 3 and Ag/CeO 2 presented the amounts of 97.1 μ mol g −1 and 69.9 μ mol g −1 , respectively. This sequence was well aligned with the activity order, indicating the catalytic performance should be closely related to the amount of active oxygen species. The different H 2 consumption amounts might be originated from the disparity of Ag particles over these three catalysts. Ghosh et al. 42 reported that the supported metallic Ag nanoparticles could activate the molecular oxygen and then catalyze the oxidation reaction. The Ag/TiO 2 catalyst possessed the best Ag dispersion, which is beneficial to oxygen activation and then led to the most sufficiency of active oxygen species, consequently exhibited the best catalytic activity.  However, this method needs the 170 °C to perform the measurement of Ag dispersion, therefore not suitable to the Ag/CeO 2 catalyst which displays the mixed reduction of both oxygen species absorbed on Ag surface and CeO 2 support at this temperature. In contrast, the intensity of reduction peak at around 80 °C in H 2 -TPR profiles was proportional to the amount of oxygen species absorbed on Ag surface, therefore it may accurately respond to the Ag dispersion on each support. We then calculated the Ag particle sizes based on the H 2 consumption amounts and the results are given in Table 2. The diameters of Ag particles are 5.5, 9.5 and 13.1 nm on TiO 2 , Al 2 O 3 and CeO 2 , respectively. Although there is slight deviation between the statistics from HR-TEM and calculation by H 2 consumption 44 , the sequences of Ag particle sizes from both methods are the same for three catalysts.
The high dispersion of Ag species on support enhanced the amount of exposed active centers and consequently contributed to high activity. However, there is another possibility for the size effect that the reaction mechanism might be changed when the Ag particle size dropped 45 . Therefore, we next implemented the kinetic measurement to investigate the change of reaction mechanism. Figure 7 displays the Arrhenius plots of the three catalysts, which were tested at high GHSV condition to control the HCHO conversion below 15%. We then calculated the apparent activation energies and the results are presented   Table 2. All the three samples display almost the same activation energy with 67 kJ mol −1 , 65 kJ mol −1 , 63 kJ mol −1 for Ag/TiO 2 , Ag/Al 2 O 3 and Ag/CeO 2 catalysts, respectively. The turnover frequency at 95 °C is also calculated based on the amount of Ag sites calculated from the H 2 -TPR results. The data in Table 2 revealed that Ag/TiO 2 , Ag/Al 2 O 3 and Ag/CeO 2 had the similar TOFs of 0.0050 s −1 , 0.0060 s −1 , 0.0068 s −1 , respectively. These results indicate that the differences of both Ag particle sizes and support types did not clearly influence the reaction mechanism of HCHO oxidation, but mainly changed the pre-exponential factor of the rate law 46 . Thus, we further confirmed that the dispersion and particle size of supported Ag species was the main factor affecting the catalytic activity.
In summary, the Ag supported on TiO 2 , Al 2 O 3 and CeO 2 catalysts were prepared by impregnation method. The Ag/TiO 2 catalyst showed the highest catalytic performance among three catalysts, achieving the complete HCHO conversion at around 95 °C. The Ag particle size was found to be the main factor influencing the catalytic performance of Ag supported catalysts for HCHO oxidation. Different supports could drastically affect the Ag particle size, consequently influence the catalytic activity. TiO 2 was found to be the most suitable support for Ag to catalyze the oxidation of HCHO since it could induce the best dispersion and the smallest particle size of Ag species on surface. Despite of the considerable distinction in activities among three catalysts, the reaction mechanism of HCHO oxidation remained unchanged. It was also demonstrated that Ag/TiO 2 catalyst may potentially be utilized to industrial application for HCHO oxidation.

Methods
Catalysts Preparation. The three kinds of samples (8 wt% Ag/support) were prepared by solution impregnation of each support at room temperature, using an aqueous solution of AgNO 3 . After stirring for 1 h, excess water was removed in a rotary evaporator at 30-50 °C under vacuum until dryness. Then, the samples were dried at 100 °C overnight and calcined at 450 °C with ramping rate of 5 °C min −1 in static air for 3 h.
The TiO 2 with anatase structure and γ -Al 2 O 3 powder were commercialized samples. The CeO 2 support was prepared with the method similar to the previous documents 47,48 . In detail, 9 mmol cerium nitrate and 0.375 mol NaOH were dissolved in 30 ml and 50 ml of deionized water, respectively. After blending the two kinds of solution in a beaker, the mixture was stirred for 60 min and subsequently transferred into a Teflon-lined stainless autoclave at a temperature of 100 °C and held for 12 h. The fresh precipitates were thoroughly washed with deionized water and anhydrous ethanol. The solid obtained was dried at 60 °C in air for 24 h and calcined at 550 °C for 4 h in air.
Catalyst Characterization. The structure parameter, pore size and specific surface area of the samples were obtained by the BET plot using a Quantachrome Quadrasorb SI-MP at − 196 °C over the whole range of relative pressures. The pore size distribution was calculated by the desorption branch of the N 2 adsorption isotherm using the BJH method. Before the N 2 physisorption, the catalysts were degassed at 300 °C for 5 h. XRD patterns were measured on an X'Pert PRO MPD X-ray powder diffractionmeter with a Cu Kα (λ = 0.154056 nm) radiation operated at 40 kV and 40 mA. The 2θ angle ranged from 10° to 80° with a scan step of 0.02°.
HRTEM was performed on a FEI Tecnai G 2 F20 electron microscope operating at 200 kV with a supplied software for automated electron tomography. Typically, a drop of the nanoparticle solution was dispensed onto a 3-mm carbon-coated copper grid. Excess solution was removed by an absorbent paper, and the sample was dried at room temperature. Average particle size and particle size distribution were sampled by ~100 Ag particles from randomly chosen areas of the TEM image. The mean particle diameter (d M ) was calculated as the previous report 30 by the formula: where d i is the particle size and n i is the number of particles in size range between d i + Δ and d i .
The UV-vis DRS of the catalysts and supports were recorded with a UV-2600 spectrophotometer. The spectra were recorded in the 220-800 nm wavelength region with a collection speed of 80 nm min −1 at ambient temperature. XPS profile was obtained by an AXIS Ultra system, equipped with an Al Kα radiation (hv = 1486.6 eV) with anode operated at 225 W and 15 kV. The binding energy values were calibrated by C 1 s peak (284.8 eV).
H 2 -TPR were carried out on Chemisorption Analyzer (AutoChem 2920) equipped with a TCD detector. After sweeping by air at 300 °C and declining down to 0 °C in the flow of Ar, 10% H 2 /Ar and air successfully flowed through the samples. Then the temperature ramped to − 50 °C at the atmosphere of Ar. When the reference gas of 10% H 2 /Ar at a rate of 50 cm 3 min −1 stabilized the TCD signal, the temperature started to ramp from − 50 to 600 °C at a rate of 10 °C min −1 . The H 2 consumption was monitored by TCD after produced H 2 O removal. The dispersion (D) data of catalysts was determined by the H 2 consumption amount. The Ag/O 2 /H 2 stoichiometry was assumed as 2/1/2 and the particle size was calculated by the formula d (nm) = 1.34/D as the previous document 33,49 . Measurement of Catalytic Activity. The activity tests for the catalytic oxidation of HCHO over the catalysts (60 mg) were performed in a fixed-bed quartz flow reactor (i.d. = 4 mm) in an incubator. The catalysts were pressed in the same pressure and sieved to collect the portion of 40-60 mesh. Gaseous HCHO was generated by flowing nitrogen through the paraformaldehyde container in a water bath kept at 35 °C. The feed gas composition was 110 ppm HCHO, 20% O 2 balanced by N 2 . The total flow rate was 100 mL min −1 , corresponding to a gas hourly space velocity (GHSV) of 100000 mL (g cat ·h) −1 .
Kinetics measurement was implemented as the HCHO conversion being kept below 15%. Kinetics data were tested under the condition of HCHO concentration with 1400 ppm and GHSV with 476000 mL (g cat ·h) −1 for Ag/TiO 2 , 341000 mL (g cat ·h) −1 for Ag/Al 2 O 3 and 302000 mL (g cat ·h) −1 for Ag/CeO 2 catalysts. The total flows all contained 20% O 2 and N 2 was the balance gas.
As the same with our previous activity evaluating instruments and methods 6 , the inlet and outlet gases were monitored by FTIR (Nicolet iS50) equipped with 2 m gas cell and a DTGS detector; resolution: 0.5 cm −1 ; OPD velocity: 0.4747 cm s −1 . The collect region was 4000-600 cm −1 and the number of scans per spectrum was 16 times. HCHO and CO 2 was measured by the peaks located at 2897 (C-H vibration) and 2350 cm −1 (O-C-O vibration), respectively. The HCHO and CO 2 concentrations were quantified and calculated based on the peak area of CO 2 at 2350 cm −1 . Since no other carbon containing compounds except for CO 2 were detected in the effluents for all tested catalysts, the conversion was calculated from a carbon balance that 1 mol of HCHO forms 1 mol of CO 2 .