Photocatalytic CC bond cleavage in ethylene glycol on TiO2: A molecular level picture and the effect of metal nanoparticles

Journal of Catalysis 354 (2017) 37–45

Contents lists available at ScienceDirect

Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat

Photocatalytic CAC bond cleavage in ethylene glycol on TiO2: A molecular level picture and the effect of metal nanoparticles Xianchi Jin a,b,1, Chao Li c,b,d,1, Chenbiao Xu a,1, Dawei Guan a, Ajin Cheruvathur d, Yi Wang d, Jian Xu d, Dong Wei a, Hongwei Xiang c,d, J.W. (Hans) Niemantsverdriet d,e, Yongwang Li c,d, Qing Guo a,⇑, Zhibo Ma a,⇑, Ren Su d,⇑, Xueming Yang a,⇑ a

State Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian, Liaoning 116023, China University of Chinese Academy of Sciences, Beijing 100049, China State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, CAS, Taiyuan, China d SynCat@Beijing, Synfuels China Technology Co. Ltd., Leyuan South Street II, No.1, Yanqi Economic Development Zone C#, Huairou District, Beijing 101407, China e SynCat@DIFFER, Syngaschem BV, 6336 HH Eindhoven, The Netherlands b c

a r t i c l e

i n f o

Article history: Received 14 July 2017 Revised 2 August 2017 Accepted 3 August 2017

Keywords: Photocatalysis Polyol conversion CAC bond cleavage Rate determining step Ethylene glycol Cocatalyst Metal nanoparticles

a b s t r a c t Polyol conversion to value-added products is of great interest for the bio-diesel industry. Photocatalytic oxidation processes may offer a green approach for polyol conversion; however the lack of comprehensive mechanistic understanding from an interdisciplinary perspective limits or even misleads the design of highly selective and efficient photocatalysts for such process. Here we have studied the photocatalytic polyol conversion on pristine TiO2 and metal (Au, Pd, and Pt) nanoparticles (NPs) decorated TiO2 using ethylene glycol (EG) as the model compound. We have developed a mechanistic picture at molecular level by coupling in-situ surface science study on rutile (110) surface with in-situ vibrational-mass spectrometry study on TiO2 nanopowders. The CAC bond cleavage was found to be the only pathway in EG photo-conversion under deaerated conditions, leading to the formation of formaldehyde and hydrogen. We rationalized that the desorption of the surface adsorbed H (Hads) to be the rate determining step (RDS), making pristine TiO2 a poor photocatalyst that only catalyze the EG conversion at very low surface coverages. The addition of metal NPs on TiO2 surface promotes the desorption of Hads significantly, thus leading to an enhanced CAC bond cleavage performance at higher surface coverages that is more applicable. Ó 2017 Elsevier Inc. All rights reserved.

1. Introduction Polyols, ranging from diols, triols, sugars, cyclitols to celluloses, are a branch of alcohols that contain multiple hydroxyl groups. Polyols exist in large quantities in the world, either produced by natural processes or synthesized as by-products in chemical industry, making the value of most polyols low in general (e.g., $ 0.01– 0.08 USD/Ib of glycerol) [1]. Therefore the conversion of polyols to high-value added chemicals (i.e., polyurethanes, polyesters, polycarbonates) is of potential importance and interest in the chemical industry [2]. Conventional catalytic hydrogenolysis process has been used to crack the polyols to smaller molecules with high values by metal ⇑ Corresponding authors. E-mail addresses: [email protected] (Q. Guo), [email protected] (Z. Ma), [email protected] (R. Su), [email protected] (X. Yang). 1 These authors contributed equally to the manuscript. http://dx.doi.org/10.1016/j.jcat.2017.08.004 0021-9517/Ó 2017 Elsevier Inc. All rights reserved.

based catalysts under high pressure H2 environment [3–5]. Glycerol can be selectively converted into various products (i.e., 1,2-propanediol, EG, 1-propanol, ethanol) by carefully controlling the surface properties of the catalyst. Selective oxidation has also been applied to convert polyols to their corresponding aldehydes or acids under relatively mild conditions [5,6]. The polyols can also undergo reforming process to produce synthesis gas (CO + nH2) for Fischer-Tropsch or methanol synthesis [7]. However, all the aforementioned processes require significant energy input for heating, as well as nano-sized noble metals catalysts (i.e., Ru, Pt, Au). Photocatalytic polyol conversion provides an alternative process that can be performed using solar energy under ambient conditions [8–11]. The photo-generated electron-hole (e
—h+) pairs offer the possibility of either performing selective reduction or oxidation by controlling the reaction conditions [12,13]. The redox potential of the e
—h+ pairs can also be tuned by manipulating the electronic structure of the photocatalyst to realize selective conversion of polyols [14,15]. However, the quantum efficiency

38

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

and selectivity of photocatalytic polyol conversion still need to be enhanced for applications. Besides, the effect of metal NPs that are frequently applied as promoters remains unclear. Unfortunately, fundamental understanding of such a process is limited and still trapped at the level of global kinetic schemes. This is far behind the development in comparison to other photocatalytic reactions (i.e., H2 and O2 evolution, CO2 reduction), where reasonable theories have been developed by various surface science techniques or in-situ spectroscopy methods to guide the design of efficient photocatalysts [16–18]. However, cautions should be paid upon these unilateral conclusions that obtained by monotonous analytic methods, as they may limit or even mislead the material design. The heterogeneous catalytic conversion of polyols has been extensively investigated and molecular level mechanistic understandings have been obtained by surface science approaches [19–24], however, the photocatalytic process is completely different compared to the thermal dissociation of polyols that normally operated above 400 K. Thus it is urgently required to establish a clear picture of the photocatalytic polyol conversion process at molecular level. Here we have used the simplest polyol, ethylene glycol (EG), to understand the photocatalytic polyol conversion. We aim at establishing a clear mechanistic understanding of bond cleavage and the rate determining step (RDS) in the EG photo-conversion process by employing interdisciplinary analytic methods. The effect of EG surface coverage and noble metal NPs as promoters on EG photoconversion have been studied by coupling in-situ surface science analysis on TiO2 single crystal and in-situ vibrational spectroscopy combined with mass spectrometry on TiO2 nanopowders.

2. Experimental section 2.1. Surface science analysis procedures Scanning tunneling microscopic (STM) analyses were performed using a low temperature STM (Matrix, Omicron) at an ultra-high vacuum condition (UHV, base pressure <4e
11 mbar). The rutile TiO2 (1 1 0) (Princeton Scientific, 10  5  1 mm3) sample was cleaned by repeated cycles of Ar+ sputtering (1 keV, 1.5 mA, 10 min) and UHV annealing (850 K, 20 min). The clean, reduced TiO2 (1 1 0)-(1  1) was obtained and verified by STM. An electrochemically etched tungsten tip was used for all scanning during the experiment. All STM images were scanned in constant current mode (100 pA) at a bias voltage of +1.25 V at 80 K. EG (Sigma-Aldrich, 98%) was cleaned via several freeze-pumpthaw cycles. The experiments were performed at 80 K, and the gas lines were heated to increase the vaper pressure of EG. The tip was retracted back 20 mm during the EG exposure [25]. The temperature programmed desorption (TPD) experiments were performed using an UHV apparatus (base pressure <3e
11 mbar) [17]. A QMS (Extrel) was used as the detector and the UHV condition (<2e
12 mbar) could be maintained in the electron-impact ionisation during experiments. The rutile TiO2(1 1 0) crystal was cleaned by standard cycles of Ar+ sputtering and UHV annealing, and the cleanness was confirmed by Auger electron spectroscopy (AES) and a sharp (1  1) low energy electron diffraction (LEED) pattern. The population of the surface bridging oxygen vacancy (BBOv) measured by H2O TPD was 4– 5%. The purified EG was then dosed to the sample at 110 K through a home-built calibrated molecular beam doser. TPD spectra were collected with a heating rate of 2 K/s after irradiation with the sample facing to the QMS detector. A 355 nm nanosecond laser (HIPPO, Spectra-Physics) was used as the light source for all STM and TPD experiments with power intensity varied from 25 to 75 mW in different experiments. The laser was operated with a pulse time of 12 ns and a frequency of

50 kHz. The spot size of the beam was 6 mm in diameter, and a grazing angle of 30° was used for the irradiation. During the laser irradiation, the STM tip was also retracted about 20 mm to avoid the shadow effect. We have also estimated the local transient temperature raise [DT(t)] caused by the laser irradiation using the equation provided by E. Weitz and co-authors [26,27]:

DTðtÞ ¼ ðF=KÞðjt=pÞ1=2

ð1Þ
2

where F is the absorbed energy density (0.003 mJ cm ) calculated from the average laser fluence (83 mW cm
2/50 kHz = 0.00165 mJ cm
2) and the reflectivity of TiO2 (12% at 355 nm irradiation), and K and j are the thermal conductivity (0.2 W cm
1 K
1 at 100 K) and the thermal diffusivity of TiO2 (0.2 cm2 s
1 at 100 K), respectively [27]. The maximum surface temperature, 80 K+ DT(t), at the cessation of a 12 ns pulse is then calculated to be 80.5 K. Therefore the heating effect of the laser is negligible in our experiments. 2.2. Synthesis and characterization of photocatalysts Degussa AeroxideÒ P25 (P25) powders were used for spectroscopy analysis. Metal (Au, Pt, and Pd) NPs were loaded on P25 via a photo-deposition method [28]. 200 mg of TiO2 powders were added into 4 ml of deionized (DI) water, sonicated for 1 h, and then transferred to a beaker that contains 96 ml of 1:1 (v/v) waterethanol solution. Then the aqueous metal salt solution (HAuCl4 4H2O, H2PtCl6 6H2O, or PdCl2) that contains the equivalent amount of 2 mg metal (1 wt% metal loading) was added into the beaker. The mixed suspension was continuously stirred and deaerated by N2 to fully remove dissolved O2. The UV irradiation was then commenced for 0.5 h after the suspension was purged for 0.5 h under dark to deposit the metal on the TiO2. After irradiation the powders were collected by centrifugation and washed by DI water for three times, and eventually dried at 80 °C for 12 h. X-ray photoelectron spectroscopy (XPS) was performed to analyze the chemical composition and oxidation states of all samples. Transmission electron microscopy (TEM) was taken to characterize the metal NPs. 2.3. Photocatalytic process analysis The EG photo-dissociation on pristine and metal NPs loaded TiO2 powders was studied using an in-situ FTIR (Vertex 70, Bruker) combined with an in-situ QMS (HPR-20, Hiden). The setup allows us to precisely control the reaction condition and to probe the evolution of the reactant, the intermediates, and the products both in the surface and gas phase. The FTIR and QMS were connected via a multi reflection attenuated total reflection (ATR) flow cell (Harrick). Prior to each experiment, the photocatalyst materials (10 mg) were deposited on the Ge window of the ATR cell. The cell was leak tight and all experiments were performed under an O2 partial pressure of <1e
12 mbar (C(O2) < 10 ppm). The bypass pump valve (VD) was closed and the cell was kept in dark for 2 h to check the leakage. The IR spectra were recorded repeatedly every 5 min for all experiments. The schematic drawing of the apparatus, detailed experimental procedures, and data processing protocols (i.e., integration method of the vibrational peaks) can be found in the Supporting Information [25]. 3. Results and discussion 3.1. Pristine TiO2 We first studied the photo-oxidation process of EG on rutile TiO2 single crystal by in-situ surface science techniques. Fig. 1(a) shows a scanning tunneling microscope (STM) image of a typical

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

clean, reduced rutile (1 1 0) surface that was taken at 80 K with a positive voltage. The well-defined rutile (1 1 0) surface was chosen as it is the most stable and reproducible surface. It is also the dominant surface in rutile TiO2 according to the Wulff construction and crystallographic analysis [29,30]. The bright and dark rows are inplane 5-fold coordinated Ti (Ti5c) atoms and 2-fold coordinated bridging oxygen (Obr) atoms respectively. The bright spots between Ti5c rows are Obr vacancies (VObr) and are marked by dashed lines for positioning [31]. After dosing 0.01 monolayer (ML, 1 ML = 5.2  1014 molecules cm
2) of EG, the identical area shown in Fig. 1(a) was imaged in Fig. 1(b). We observed that the EG molecule (marked by a dashed circle) preferably adsorbed on the Ti5c rows. Whilst the photo-dissociation of simple monohydric alcohols (i.e., ethanol) on rutile (1 1 0) resulted in OAH and CAH bond

39

cleavage [16], the photocatalytic conversion of EG exhibited a completely different phenomenon (Fig. 1c–e). The spherical EG molecule was converted into two dumbbell-shaped molecules after irradiation for 20 min (Fig. 1c). The reaction is a photocatalytic process rather than a thermal process, as the local transient temperature raise caused by the laser irradiation is negligible (see estimation of the temperature raise in the experimental section) [25]. Further irradiation resulted in gradual desorption of these two dumbbell-shaped molecules, leading to the re-exposure of the original VObr and the appearing of two H atoms on the Obr (HObr, Fig. 1d and e) [25]. The HObr marked in Fig. 1d showed a reconstructed structure that is resulted from the interaction between the HObr and the VObr. In contrast to the formation of symmetrical HObr from the OAH and CAH bond cleavage that was observed in the dissociation of monohydric alcohol [16], these two H atoms

Fig. 1. In-situ STM images (2.7 nm  2.7 nm) of (a) pristine rutile (1 1 0); (b) after EG (200 pm in height) adsorption; (c)-(e) after UV irradiation for 20, 40, and 60 min. The bridging oxygen vacancies (VObr) are marked by dashed lines for positioning. EG, formaldehyde, and H atom adsorbed on the Obr (HObr) are marked by circle, hexagon, and square, respectively. The top-right bright spot in (b)-(e) is a water molecule (70 pm in height) introduced during dosing EG; (f) formaldehyde on rutile (1 1 0) surface. All images are recorded at 80 K.

40

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

Fig. 2. (a) TPD spectra (m/z = 30) recorded after 0.1 ML EG adsorbed on rutile (1 1 0) surface (black curve) and after UV irradiation for different times at 80 K. The light intensity was 50 mW cm
2. (b) Integrated TPD signal of formaldehyde as a function of the irradiation time. The line is exponential fitting to the data points.

formed in EG dissociation did not adsorb on the Obr symmetrically (see Fig. S1 in SI). Since EG is adsorbed on TiO2 via the two O atoms bound to two neighboring Ti5c, we consider that these two H atoms are most likely originated from the cleavage of two OAH bonds. Note that the HObr atoms are firmly bound to the TiO2 surface even above room temperature (RT) [16]. We assigned the dumbbellshaped molecules to be the CAC bond cleavage product, formaldehyde, which showed an identical structure when adsorbed on rutile (1 1 0) (Fig. 1f). The CAC bond cleavage process is independent of the VObr (see Fig. S1 in SI). Previous studies on photodissociation of methanol also observed the formation of formaldehyde with similar shape [32,33]. This hypothesis has been further confirmed by temperature programmed desorption (TPD) analysis on photocatalytic dissociation of 0.1 ML EG on rutile (1 1 0) surface, as shown in Fig. 2(a). Prior to UV irradiation, only a tiny desorption peak located at 440 K was observed, which can be indexed to the trace amount of the surface adsorbed EG (black curve) [21,34]. A desorption peak centered at 260 K was gradually evolved after UV irradiation has been commenced, which can be indexed to the surface adsorbed formaldehyde molecules [35–37]. Meanwhile a tiny desorption peak at 420 K appeared after UV irradiation for 40 min, which can be assigned to the formation of methanol during TPD analysis via the hydrogenation process of formaldehyde with the Hads [38]. No formation of other species (i.e., H2, H2O) can be detected (see Fig. S2 in the SI) [25]. We have further plotted the integrated peak areas of the surface adsorbed formaldehyde as a function of the irradiation time, as shown in Fig. 2(b). An exponential dependence between the formaldehyde desorption peak area and the irradiation time suggests that the photocatalytic EG dissociation process on TiO2 surface follows pseudo-first order kinetics with a rate constant of 0.04 min
1. Based on the STM and TPD results, we propose that EG underwent photo-oxidation on the rutile (1 1 0) surface via CAC bond cleavage solely, as shown in Eq. (2): hv

HOH2 C
CH2 OH ! 2HCHO þ 2HObr

ð2Þ

Note that this photo-oxidation of EG on rutile (1 1 0) surface at low temperature (80 K) showed a completely different mechanism compared to that of the conventional thermal dissociation of EG on rutile (1 1 0) surface reported previously [20,21,24]. The thermal dissociation of EG only takes place at elevated temperatures (>400 K) via both dehydration and dehydrogenation channels, which leading to the formation of ethylene and water, acetaldehyde and H2, respectively. It should be also noted that the surface science studies were performed on rutile TiO2 single crystal with a very low EG coverage (0.01–0.1 ML) under ultra-high vacuum at 80 K, whereas realistic photocatalytic reactions are normally performed using photocata-

lyst NPs or thin films at very high surface coverages (in liquid) under ambient conditions at RT. Therefore we aimed at extending the surface science conclusions to more applied photocatalysis by investigating the EG photo-dissociation on TiO2 nanopowders from low to high surface coverages at ambient conditions using in-situ spectrometry approaches. We have first investigated the photo-conversion of EG using pristine P25 TiO2 NPs at three different dosing volumes (0.4, 2, and 10 lL) by in-situ Fourier transform infrared-quadruple mass spectrometry (FTIR-QMS, Fig. S2 in SI) under deaerated conditions [25]. These dosing volumes corresponded to the surface coverages of  3, 17, and 82 ML of EG, which we assigned them to ML, thin layers (XL), and thick layers (XXL) of surface adsorbed EG, respectively (see estimation and Fig. S3 in the SI) [25]. As depicted in Fig. 3(a), we observed that EG molecules (m[OH] = 3333 cm
1, mas[CH3] = 2940, and ms[CH3] + ms[CH2] = 2877 cm
1)[39] can be photocatalytically converted into formaldehyde (m[CO] = 1717 cm
1)[40] via CAC bond cleavage under deaerated conditions at a very low surface coverage (ML). Tiny but detectable formaldehyde peaks were observed after the TiO2 powders were irradiated for 50 min. Further increasing the irradiation time resulted solely in the accumulation of formaldehyde. This phenomenon became less obvious when the EG surface coverage increased to a thin layer, where detectable formaldehyde peaks can be observed only after the sample had been irradiated for three hours (Fig. 3b). The FT-IR study on pristine TiO2 nanopowders also indicates that only a small portion of EG can be slowly photodissociated into formaldehyde, which agreed well with the TPD observations on TiO2 single crystal surface. It should be clarified that the thickness of the EG layer does not influence the light transmission, as the EG molecule exhibited a negligible absorption in the UV range (Fig. S4 in SI) [25]. We have further revealed the effect of EG surface coverage on the evolution of formaldehyde by analyzing the integrated m[CO] peaks (Fig.3c). The formation rate of formaldehyde decreased significantly upon increasing the EG surface coverage. Noticeably, the photo-conversion of EG to formaldehyde was completely inactivated even after 10 h irradiation when a thick layer of EG covered the TiO2 (Fig. S5 in the SI) [25]. This result is consistent with previous observations using pristine TiO2 for photocatalytic H2 evolution with alcohols as hole scavengers, as the charge recombination kinetics was significantly faster than that of redox reactions [18,41]. Whilst the formation of formaldehyde followed first order kinetics, the H2 evolution rates showed zero order kinetics, as evidenced by in-situ QMS (Fig. 3d) [25]. We also noticed that the H2 evolution rates dropped and eventually stopped following the increase of the EG surface coverage from ML to thick layer. We have rationalized the photocatalytic EG dissociation process by combining the in-situ surface science and spectroscopy analyses, as demonstrated in Scheme 1(a)–(c). Upon UV irradiation, the surface adsorbed EG molecules on the Ti5c rows (Scheme 1a) will undergo CAC bond cleavage, forming two parts of formaldehyde on the Ti5c and two surface adsorbed H atoms (Hads) on the Obr, which completely covered the TiO2 surface. The zero order H2 evolution rates indicate that the removal of Hads in the form of molecular H2 from Obr is the RDS of EG photo-conversion on pristine TiO2. While most EG molecules will adsorb on the Ti5c with neighboring HObr and cannot be dissociated (Scheme 1b), only those EG molecules that adsorbed on the Ti5c with two unoccupied neighboring Obr can be dissociated (Scheme 1c). Considering the absence of formic acid (m[CO2]  1600–1400 cm
1) [42], we suggest that the EG molecules were solely photo-oxidized to formaldehyde coverages under deaerated conditions following Eq. (3) and (4):

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

41

Fig. 3. (a) and (b) In-situ FTIR spectra recorded during photocatalytic EG conversion under deaerated conditions [C(O2) < 10 ppm] using pristine TiO2 powders with different EG coverages. The EG and formaldehyde peaks are labelled as I and II, respectively. (c) Evolution of the integrated formaldehyde peaks during the photocatalytic process at different EG coverages. The lines are exponential fitting to the data points. (d) Evolution of molecular H2 during the In-situ FTIR experiment determined by QMS.

Scheme 1. (a) Scheme of EG adsorption on TiO2 surface under dark conditions. The surface adsorbed EG will be photo-dissociated into formaldehyde and HObr. (b) Newcoming EG on Ti5c having neighboring Obr fully occupied with Hads cannot be dissociated; (c) New-coming EG on Ti5c having neighboring Obr without Hads can be further dissociated.

hv

HOH2 C
CH2 OH ! 2HCHO þ 2Hads slow

2Hads ! H2

ð3Þ ð4Þ

It should be noted that the rutile (1 1 0) surface used for surface science studies differs from the P25 TiO2 (78% of anatase, 20% of rutile and 2% of amorphous) used for spectroscopy studies. However, we consider that the major differences between these two systems (the specific surface area, the surface density of Ti5c, and the band position) do not influence the reaction mechanisms but only influence the photocatalytic activity. While the larger surface area of the TiO2 powders and a higher density of the Ti5c in anatase surface provide more active sites for the reaction [43], the band alignment of anatase and rutile in P25 improves the charge transfer [44]. Therefore we consider that all the major differences only influence the reaction rate but are unlikely to change the reaction mechanism of the photocatalytic EG dissociation. 3.2. Metal NPs supported on TiO2 It is obvious that the poor performance of pristine TiO2 hindered the photocatalytic polyol conversion for application. Since noble

metal NPs showed lower adsorption energy of Hads [45], we expected that the photo-conversion efficiency of EG can be improved by surface modification of TiO2 with metal NPs. We have synthesized Au, Pt, and Pd NPs decorated TiO2 powders with similar metal loadings (1 wt%) and particle sizes (5 nm), as demonstrated in Figs. S6 and S7 in the SI [25]. As shown in Fig. 4(a), (b) and Fig. S8 in SI [25], the presence of Au, Pt, and Pd NPs significantly improved the photo-conversion of EG to formaldehyde at the thin layer coverage under deaerated conditions. Both the AOH and ACH2 vibration peaks dropped coherently without peak shifting, indicating no intermediates with both AOH and ACH2 groups evolved. This result differs from the photocatalytic EG dissociation on Rh/TiO2 in water, where glycolaldehyde and acetaldehyde were observed as the main intermediates together with formaldehyde [10]. Whilst EG molecules oxidized exclusively into formaldehyde on Au/TiO2, the evolution of tiny amounts of CO (m [CO] = 2050 cm
1 and 1900 cm
1 for Pd/TiO2) was observed for Pt/TiO2 and Pd/TiO2 (see Fig. S8 in the SI) [25,46]. However, the carbonyl formation showed a negligible effect on EG photoconversion (Fig. 3c and Fig. S8d in the SI) [20]. The formation rates of formaldehyde remained to be first order for all samples, and the reactivity followed the order of Pt > Au  Pd  pristine TiO2. More

42

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

Fig. 4. (a) and (b) Representative in-situ FTIR spectra recorded during photocatalytic EG conversion at thin layer coverage (XL, 2 lL) under deaerated conditions [C(O2) < 10 ppm] using Au/TiO2 and Pt/TiO2 powders. (c) Evolution of the integrated formaldehyde peaks during the photocatalytic process using pristine and metal/TiO2. (d) Evolution of H2 in the gas-phase during the In-situ FTIR experiment determined by QMS.

Fig. 5. (a)–(c) Representative in-situ FTIR spectra recorded during photocatalytic EG conversion at thick layer coverage (XXL, 10 lL) under deaerated conditions [C(O2) < 10 ppm] using Au/TiO2, Pt/TiO2, and Pd/TiO2 powders, respectively. (d) Evolution of the integrated formaldehyde peaks during the photocatalytic process using pristine and metal/TiO2.

importantly, significant enhancement in H2 evolution rates was observed when metal NPs were used as promoters, and followed the same order of Pt > Au  Pd  pristine TiO2 (Fig. 3d). It is also worth noting that the H2 evolution rates changed to first order kinetics for all metal/TiO2 samples, implying that the metals indeed promote the Hads desorption thus resulting an enhanced photodissociation rate of EG into formaldehyde. The ratios between the formaldehyde formation rate and H2 evolution rate of all metal/TiO2 samples are 2:1, again indicating the photocatalytic EG dissociation follows Eqs. (3) and (4) that forming formaldehyde and H2. Noticeably, the concentration of formalde-

hyde reached a plateau after UV irradiation of 3 h when Pt/TiO2 was used as the photocatalyst, indicating formaldehyde is the final product in photocatalytic EG conversion under deaerated conditions. This has been also confirmed by our in-situ QMS analysis, where only trace amount of CO2 was detected. In contrast, significant amount of CO2 was observed when the EG photodissociation was performed with the presence of water [10]. Furthermore, the metal-decorated TiO2 photocatalysts also worked even at thick layer coverage (82 ML, XXL), which is equivalent to realistic photocatalysis that are normally performed in liquid phase, as shown in Fig. 5(a)–(c) and Fig. S9 in the SI [25].

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

43

Fig. 6. STM images (15 nm  15 nm) of (a) and (b), Au NPs on rutile (1 1 0); (c) and (d) 0.05 ML EG adsorbed on Au/rutile (1 1 0); (e) and (f) Au/rutile (1 1 0) after UV irradiation for 80 min, respectively. The reacted EG molecules are labelled by dashed shapes.

The use of Pt NPs still outperformed than that of using Au and Pd NPs as promoter in converting EG to formaldehyde, however, it should be also noted that the identity of metal NPs showed huge impact on the photocatalytic performance of EG dissociation at higher surface coverages (Fig. 5d). While the use of Pt NPs as promoter can further boost the photodissociation of EG at thick layer coverage compared to that of the reaction performed at thin layer coverage, the use of Au NPs only showed very weak enhancement in the photocatalytic performance. In contrast, the use of Pd NPs as promoter failed to further improve the performance at thick layer coverage of EG, and the reaction seems slowed down after UV irradiation for 5 h (Fig. 5d). Meanwhile, we do observe that significant amount of CO was formed after UV irradiation for 100 min on Pd/ TiO2 (Fig. 5c), which may poison the Pd/TiO2 photocatalyst. There-

fore the use of Pt NPs as promoter is more appropriate for scaling up applications of EG photoconversion. To examine the effect of metal NPs in the photo-conversion of EG on the molecular level, we deposited Au NPs on rutile (1 1 0) surface and investigated the reaction by in-situ STM. Fig. 6 (a) and (b) showed the representative metal rich and metal deficient areas of the as-deposited Au on rutile (1 1 0) surface, respectively. Statistical analysis revealed that the diameter of Au NPs varied from 2 to 8 nm with an averaged particle diameter of 4.8 nm (Fig. S10 in SI) [25]. This was comparable to that of the metal NPs deposited on the polycrystalline TiO2. After dosing 0.05 ML of EG, we observed that the EG molecules were mainly adsorbed on the rutile (1 1 0) surface (Fig. 6c and d) with a few EG on the Au-TiO2 interface (Fig. 6c). Note that EG on Au NPs

44

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45

Scheme 2. (a) Schematic demonstration of EG photoconversion on Au/rutile (1 1 0) surface. The photogenerated HObr can rapidly diffuse onto Au surfaces and desorb as H2, resulting in the depletion of Obr. The EG adsorbed on Ti5c have more neighbouring Obr without Hads thus can be continuously dissociated. (b) Overall photocatalytic EG dissociation cycle on metal/TiO2.

was difficult to distinguish due to the high electron density of Au. Since alcohol molecules are often used as electron donors in photocatalytic H2 evolution [16,47], we consider that only EG on the TiO2 surface can react with the photo-generated hole species located on the TiO2 sites. After irradiation for 80 min at 80 K, most EG molecules underwent CAC bond cleavage and formed exclusively formaldehyde (Fig. 6e and f). The unchanged bright dots after irradiation may be attributed to water and unreacted EG. Combining the in-situ spectroscopy and the STM results, we conclude that the presence of metal NPs does not alter the adsorption of EG. The Hads formed during the photocatalytic EG CAC bond cleavage rapidly transferred from Obr to the metal surface and eventually desorbed as H2 gas, as demonstrated in Scheme 2 (a) and (b). Such a promotion effect of metal NPs on Hads desorption was also observed in the water-gas shift reaction on Pt/TiO2 surface [48]. This rapid depletion of Hads provides extra Obr sites for the dissociation of the OAH bond of EG, thus promotes the CAC bond cleavage of EG and the formation of formaldehyde.

4. Conclusions Here we proposed a molecular level picture of the photocatalytic conversion of ethylene glycol on pristine TiO2 and metal NPs decorated TiO2 by coupling surface science techniques with in-situ spectroscopies. While in-situ STM and TPD revealed that the EG molecules were preferably adsorbed on the Ti5c of the TiO2 surfaces and solely converted into formaldehyde via CAC bond cleavage process, in-situ IR-MS spectroscopy suggested that such reaction was EG surface coverage dependent and the desorption of Hads was the rate determining step. The presence of metal NPs (Au, Pd, and Pt) did not change the EG adsorption sites and the reaction mechanisms, however it promoted the desorption of Hads to H2 significantly, thus improved the formaldehyde formation even at higher surface coverages of EG. We anticipate that our interdisciplinary investigations in revealing the reaction mechanisms from molecular-to-kinetic level will promote the understanding of the photo-conversion of more complicated polyols and help the design of novel photocatalyst materials.

Acknowledgment We thank financial support from the NSFC (projects number: 21503257, 21673236, 21673235, 21403224, 21503223), the CAS, Strategic pilot science and technology project of the CAS (XDB17010200), the Chinese Ministry of Science and Technology (2013CB834605), the Youth Innovation Promotion Association CAS, and the Key Research Program of the CAS.

Author Contributions RS and ZM conceived the study. XJ, CX, DG, DW, QG, and ZM performed the surface science studies; RS designed the in-situ FTIRQMS; XJ, CL, and AC carried out the in-situ FTIR-QMS experiments; YW performed the TEM characterizations. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. yThese authors contributed equally. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcat.2017.08.004. References [1] P.S. Kong, M.K. Aroua, W.M.A.W. Daud, Conversion of crude and pure glycerol into derivatives: a feasibility evaluation, Renew. Sust. Energy Rev. 63 (2016) 533–555. [2] BASF: www.intermediates.basf.com, and Shell Global: www.shell.com/. [3] T. Miyazawa, Y. Kusunoki, K. Kunimori, K. Tomishige, Glycerol conversion in the aqueous solution under hydrogen over Ru/C+ an ion-exchange resin and its reaction mechanism, J. Catal. 240 (2006) 213–221. [4] E.P. Maris, W.C. Ketchie, M. Murayama, R.J. Davis, Glycerol hydrogenolysis on carbon-supported PtRu and AuRu bimetallic catalysts, J. Catal. 251 (2007) 281– 294. [5] C.-H. Zhou, J.N. Beltramini, Y.-X. Fan, G.Q. Lu, Chemoselective catalytic conversion of glycerol as a biorenewable source to valuable commodity chemicals, Chem. Soc. Rev. 37 (2008) 527–549. [6] D.I. Enache, J.K. Edwards, P. Landon, B. Solsona-Espriu, A.F. Carley, A.A. Herzing, M. Watanabe, C.J. Kiely, D.W. Knight, G.J. Hutchings, Solvent-free oxidation of primary alcohols to aldehydes using Au-Pd/TiO2 catalysts, Science 311 (2006) 362–365. [7] M. Pagliaro, R. Ciriminna, H. Kimura, M. Rossi, C. Della Pina, From glycerol to value-added products, Angew. Chem. Int. Ed. 46 (2007) 4434–4440. [8] J.C. Colmenares, R. Luque, Heterogeneous photocatalytic nanomaterials: prospects and challenges in selective transformations of biomass-derived compounds, Chem. Soc. Rev. 43 (2014) 765–778. [9] X. Lang, X. Chen, J. Zhao, Heterogeneous visible light photocatalysis for selective organic transformations, Chem. Soc. Rev. 43 (2014) 473–486. [10] T.F. Berto, K.E. Sanwald, W. Eisenreich, O.Y. Gutiérrez, J.A. Lercher, Photoreforming of ethylene glycol over Rh/TiO2 and Rh/GaN:ZnO, J. Catal. 338 (2016) 68–81. [11] K.E. Sanwald, T.F. Berto, W. Eisenreich, O.Y. Gutiérrez, J.A. Lercher, Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. 344 (2016) 806–816. [12] R. Su, L. Kesavan, M.M. Jensen, R. Tiruvalam, Q. He, N. Dimitratos, S. Wendt, M. Glasius, C.J. Kiely, G.J. Hutchings, F. Besenbacher, Selective photocatalytic oxidation of benzene for the synthesis of phenol using engineered Au-Pd alloy nanoparticles supported on titanium dioxide, Chem. Commun. 50 (2014) 12612–12614. [13] X. Guo, C. Hao, G. Jin, H.-Y. Zhu, X.-Y. Guo, Copper nanoparticles on graphene support: an efficient photocatalyst for coupling of nitroaromatics in visible light, Angew. Chem. Int. Ed. 53 (2014) 1973–1977. [14] Y. Zhang, N. Zhang, Z.-R. Tang, Y.-J. Xu, Identification of Bi2WO6 as a highly selective visible-light photocatalyst toward oxidation of glycerol to dihydroxyacetone in water, Chem. Sci. 4 (2013) 1820–1824.

X. Jin et al. / Journal of Catalysis 354 (2017) 37–45 [15] R. Chong, J. Li, X. Zhou, Y. Ma, J. Yang, L. Huang, H. Han, F. Zhang, C. Li, Selective photocatalytic conversion of glycerol to hydroxyacetaldehyde in aqueous solution on facet tuned TiO2-based catalysts, Chem. Commun. 50 (2014) 165– 167. [16] J.Ø. Hansen, R. Bebensee, U. Martinez, S. Porsgaard, E. Lira, Y.Y. Wei, L. Lammich, Z.S. Li, H. Idriss, F. Besenbacher, B. Hammer, S. Wendt, Unravelling site-specific photo-reactions of ethanol on rutile TiO2(110), Sci. Rep. 6 (2016). [17] Q. Guo, C. Xu, Z. Ren, W. Yang, Z. Ma, D. Dai, H. Fan, T.K. Minton, X. Yang, stepwise photocatalytic dissociation of methanol and water on TiO2(110), J. Am. Chem. Soc. 134 (2012) 13366–13373. [18] R. Su, R. Tiruvalam, A.J. Logsdail, Q. He, C.A. Downing, M.T. Jensen, N. Dimitratos, L. Kesavan, P.P. Wells, R. Bechstein, H.H. Jensen, S. Wendt, C.R.A. Catlow, C.J. Kiely, G.J. Hutchings, F. Besenbacher, Designer titania-supported Au-Pd nanoparticles for efficient photocatalytic hydrogen production, ACS Nano 8 (2014) 3490–3497. [19] Z. Zhang, Y. Yoon, X. Lin, D. Acharya, B.D. Kay, R. Rousseau, Z. Dohnálek, OH group dynamics of 1,3-propanediol on TiO2(110), J. Phys. Chem. Lett. 3 (2012) 3257–3263. [20] D.P. Acharya, Y. Yoon, Z. Li, Z. Zhang, X. Lin, R. Mu, L. Chen, B.D. Kay, R. Rousseau, Z. Dohnálek, Site-specific imaging of elemental steps in dehydration of diols on TiO2(110), ACS Nano 7 (2013) 10414–10423. [21] Z. Li, B.D. Kay, Z. Dohnalek, Dehydration and dehydrogenation of ethylene glycol on rutile TiO2(110), Phys. Chem. Chem. Phys. 15 (2013) 12180–12186. [22] L. Chen, Z. Li, R.S. Smith, B.D. Kay, Z. Dohnálek, Conversion of 1,2-propylene glycol on rutile TiO2(110), J. Phys. Chem. C 118 (2014) 15339–15347. [23] L. Chen, Z. Li, R.S. Smith, B.D. Kay, Z. Dohnálek, Conversion of 1,3-propylene glycol on rutile TiO2(110), J. Phys. Chem. C 118 (2014) 23181–23188. [24] L. Chen, Z. Li, R.S. Smith, B.D. Kay, Z. Dohnálek, molecular hydrogen formation from proximal glycol pairs on TiO2(110), J. Am. Chem. Soc. 136 (2014) 5559– 5562. [25] Details for sample synthesis, characterisations, the protocol for in-situ FTIRQMS, more FTIR-QMS data, UV-vis spectra of EG, and STM iamges are available in the Supporting Information. [26] D. Burgess Jr., P.C. Stair, E. Weitz, Calculations of the surface temperature rise and desorption temperature in laser-induced thermal desorption, J. Vac. Sci. Technol. 4 (1986) 1362–1366. [27] S.J. Garrett, V.P. Holbert, P.C. Stair, E. Weitz, Wavelength dependence of the photodissociation and photodesorption of CD3I adsorbed on the TiO2(110) surface, J. Chem. Phys. 100 (1994) 4626–4636. [28] G.R. Bamwenda, S. Tsubota, T. Nakamura, M. Haruta, Photoassisted hydrogenproduction from a water-ethanol solution - a comparison of activities of AuTiO2 and Pt-TiO2, J. Photochem. Photobiol. A 89 (1995) 177–189. [29] M. Ramamoorthy, D. Vanderbilt, R.D. Kingsmith, 1st-principles calculations of the energetics of stoichiometric TiO2 surfaces, Phys. Rev. B 49 (1994) 16721– 16727. [30] R. Su, M. Christensen, Y. Shen, J. Kibsgaard, B. Elgh, R.T. Vang, R. Bechstein, S. Wendt, A. Palmqvist, B.B. Iversen, F. Besenbacher, Rapid synthesis of porous, mixed phase titania films with tailored orientation of rutile for enhanced photocatalytic performance, J. Phys. Chem. C 117 (2013) 27039–27046. [31] Y. Zhao, Z. Wang, X. Cui, T. Huang, B. Wang, Y. Luo, J. Yang, J. Hou, What are the adsorption sites for CO on the reduced TiO2(110)-1  1 Surface?, J Am. Chem. Soc. 131 (2009) 7958–7959. [32] Y. He, D. Langsdorf, L. Li, H. Over, Versatile model system for studying processes ranging from heterogeneous to photocatalysis: epitaxial RuO2(110) on TiO2(110), J. Phys. Chem. C 119 (2015) 2692–2702.

45

[33] D. Wei, X. Jin, C. Huang, D. Dai, Z. Ma, W.-X. Li, X. Yang, Direct imaging single methanol molecule photocatalysis on titania, J. Phys. Chem. C 119 (2015) 17748–17754. [34] X. Yu, Z. Zhang, C. Yang, F. Bebensee, S. Heissler, A. Nefedov, M. Tang, Q. Ge, L. Chen, B.D. Kay, Z. Dohnálek, Y. Wang, C. Wöll, Interaction of formaldehyde with the rutile TiO2(110) surface: a combined experimental and theoretical study, J. Phys. Chem. C 120 (2016) 12626–12636. [35] H. Qiu, H. Idriss, Y. Wang, C. Wöll, Carbon
carbon bond formation on model titanium oxide surfaces: identification of surface reaction intermediates by high-resolution electron energy loss spectroscopy, J. Phys. Chem. C 112 (2008) 9828–9834. [36] J. Haubrich, E. Kaxiras, C.M. Friend, The role of surface and subsurface point defects for chemical model studies on TiO2: a first-principles theoretical study of formaldehyde bonding on rutile TiO2(110), Chem. Eur. J. 17 (2011) 4496– 4506. [37] C. Xu, W. Yang, Q. Guo, D. Dai, T.K. Minton, X. Yang, Photoinduced decomposition of formaldehyde on a TiO2(110) surface, assisted by bridgebonded oxygen atoms, J. Phys. Chem. Lett. 4 (2013) 2668–2673. [38] H. Feng, S. Tan, H. Tang, Q. Zheng, Y. Shi, X. Cui, X. Shao, A. Zhao, J. Zhao, B. Wang, Temperature- and coverage-dependent kinetics of photocatalytic reaction of methanol on TiO2 (110)-(1  1) surface, J. Phys. Chem. C 120 (2016) 5503–5514. [39] T.T. Nguyen, M. Raupach, L.J. Janik, Fourier-transform infrared study of ethylene glycol monoethyl ether adsorbed on montmorillonite; implications for surface area measurements of clays, Clays Clay Miner. 35 (1987) 60–67. [40] G. Busca, J. Lamotte, J.C. Lavalley, V. Lorenzelli, FT-IR study of the adsorption and transformation of formaldehyde on oxide surfaces, J. Am. Chem. Soc. 109 (1987) 5197–5202. [41] M. Murdoch, G.I.N. Waterhouse, M.A. Nadeem, J.B. Metson, M.A. Keane, R.F. Howe, J. Llorca, H. Idriss, The effect of gold loading and particle size on photocatalytic hydrogen production from ethanol over Au/TiO2 nanoparticles, Nat. Chem. 3 (2011) 489–492. [42] G. Busca, A.S. Elmi, P. Forzatti, Mechanism of selective methanol oxidation over vanadium oxide-titanium oxide catalysts: a FT-IR and flow reactor study, J. Phys. Chem. 91 (1987) 5263–5269. [43] U. Diebold, The surface science of titanium dioxide, Surf. Sci. Rep. 48 (2003) 53–229. [44] D.O. Scanlon, C.W. Dunnill, J. Buckeridge, S.A. Shevlin, A.J. Logsdail, S.M. Woodley, C.R.A. Catlow, M.J. Powell, R.G. Palgrave, I.P. Parkin, G.W. Watson, T. W. Keal, P. Sherwood, A. Walsh, A.A. Sokol, Band alignment of rutile and anatase TiO2, Nat. Mater. 12 (2013) 798–801. [45] J. Greeley, T.F. Jaramillo, J. Bonde, I. Chorkendorff, J.K. Norskov, Computational high-throughput screening of electrocatalytic materials for hydrogen evolution, Nat. Mater. 5 (2006) 909–913. [46] E. Ivanova, M. Mihaylov, F. Thibault-Starzyk, M. Daturi, K. Hadjiivanov, FTIR spectroscopy study of CO and NO adsorption and co-adsorption on Pt/TiO2, J. Mol. Catal. A 274 (2007) 179–184. [47] W.-T. Chen, A. Chan, Z.H.N. Al-Azri, A.G. Dosado, M.A. Nadeem, D. SunWaterhouse, H. Idriss, G.I.N. Waterhouse, Effect of TiO2 polymorph and alcohol sacrificial agent on the activity of Au/TiO2 photocatalysts for H2 production in alcohol–water mixtures, J. Catal. 329 (2015) 499–513. [48] S.C. Ammal, A. Heyden, Water–gas shift catalysis at corner atoms of pt clusters in contact with a TiO2 (110) support surface, ACS Catal. 4 (2014) 3654–3662.