Photo splitting of bio-polyols and sugars to methanol and syngas

Methanol is a clean liquid energy carrier of sunshine and a key platform chemical for the synthesis of olefins and aromatics. Herein, we report the conversion of biomass-derived polyols and sugars into methanol and syngas (CO+H2) via UV light irradiation under room temperature, and the bio-syngas can be further used for the synthesis of methanol. The cellulose and even raw wood sawdust could be converted into methanol or syngas after hydrogenolysis or hydrolysis pretreatment. We find Cu dispersed on titanium oxide nanorod (TNR) rich in defects is effective for the selective C−C bond cleavage to methanol. Methanol is obtained from glycerol with a co-production of H2. A syngas with CO selectivity up to 90% in the gas phase is obtained via controlling the energy band structure of Cu/TNR.

tube and reduced at 300 o C for 2 h in pipe furnace with a heating rate of 2 o C min -1 and 25 mL min -1 flow of H2.

Preparation M/P25 (M=Fe，Co，Ni)
The P25 supported catalyst with 1 wt % weight of metal was prepared by the impregnation method. Typically, 0.5 g of P25 was dispersed in 20 mL of deionized water, into which a certain volume of M(NO3)2 aqueous solutions (0.1 mol· L -1 ) was added. The mixture was stirred for 24 h at room temperature and then heated at 100 o C overnight to remove water. The acquired powders were put in a quartz tube and calcined at 450 o C for 2 h in pipe furnace with a heating rate of 2 o C min -1 and 25 mL min -1 flow of air.

Characterizations
Fourier transform infrared (FT-IR) spectra were collected on a Bruker Tensor 27 FT -IR spectrometer. About 30 mg of Cu/TNR samples was pressed into a plate. After collecting the background of Cu/TNR, 10 μl of formic solution was dropped into the plate which is dried in vacuum at 60 o C for 10 min. Then, the adsorption spectra was collected via subtraction of the Cu/TNR background. The X-ray powder diffraction (XRD) patterns were obtained using a Rigaku D/Max 2500/PC powder diffractometer with Cu Kα radiation (λ = 0.15418 nm). High resolution transmission electron microscopy (HRTEM) was performed using JEOL JEM-2100 electron microscope operated at 220 kV. The X-ray photoelectron spectroscopy (XPS) measurements were performed on a Thermo ESCALAB 250Xi spectrometer equipped with a monochromated AlKα X-ray source (hυ = 1486.6 eV, 15 kV, 10.8 mA). The samples were dried in vacuum at 120 °C for 12 h. The charge neutralizer system was used for all of the analyses. The base pressure was 1 × 10 -8 Pa. High resolution spectra were recorded with 20 eV pass energy. The pass energies correspond to the Ag3d5/2 FWHM of 0.65 eV. The data was acquired with 0.05 eV steps. The binding energy (BE) was calibrated to the C1s signal (284.6 eV) as a reference. The curve fitting procedure was performed using an approximation based on a combination of the Gaussian and Lorentzian functions with the subtraction of a Shirley-type background. The X-ray absorption fine structure (XAFS) experiment was performed at the bending magnet beamline BL12B of SPring-8 (8 GeV, 100 mA) belong to National Synchotron Radiation Research Center, in which the X-ray beam was monochromatized with water-cooled Si (111) double-crystal monochromator and focused with two Rh coated focusing mirrors with the beam size of 2.0 mm in the horizontal direction and 0.5 mm in the vertical direction around sample position. The samples were filled in a phi10 aluminium tube sealed with graphite tape in Ar gas protected glovebox for quasi in situ XAFS measurements. All of the samples were measured by both transmission and fluorescence modes at Cu K-edge. The spectra were analyzed and fitted using an analysis program Demeter(Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for Xray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat 12, 537-541 (2005).). The electron paramagnetic resonance (EPR) were performed on Bruker spectrometer at X-band under room temperature, with a field modulation of 100 kHz. The microwave frequency was kept at 9.401 GHz. For in situ ESR measurements, 2Cu/TNR were dispersed in a mixed solution of 95 vol% MeCN and 0.5 vol% H2O containing DMPO (0.1 M), which was used as a spin-trapping agent, by ultrasonic treatment. Then, the suspension was injected into a glass capillary and the glass capillary was placed in a sealed glass tube under Ar atmosphere. The sealed glass tube was placed in the microwave cavity of EPR spectrometer and was irradiated with Hg lamp during EPR measurements at room temperature.

DFT calculations
All of the first-principles electronic structure calculations were carried out using the Vienna ab initio simulation package (VASP), 6 one density functional theory implementation. The exchange correlation potential was described by the Perdew-Burke-Ernzerhof (PBE) 7 formulation of the generalized gradient approximation (GGA). The ion-electron interactions were represented by the projector augmented wave (PAW) 8 method, while the valence electrons (2s 2 2p 4 of O, 3s 2 3p 6 3d 2 4s 2 of Ti, and 3d 10 4s of Cu) were expanded by a plane wave basis set with an energy cutoff of 400 eV. The k-point sampling was performed using the Monkhorst-Pack scheme. 9 The electronic self-consistent minimization was converged to 10 -5 eV, and the geometry optimization was converged to 10 -4 eV. The self-interaction error (SIE) was mitigated using the DFT+U method by Dudarev and his colleagues. 10 A typical U value of 4.5 eV was used for Ti.
The lattice constants of anatase TiO2 were optimized to be a = 3.855 Å and c = 9.661 Å, in good agreement with the experimental constants, a = 3.782 Å and c = 9.502 Å. 11 We used them to build a p(2 × 4) TiO2(101) slab with 12 atomic layers and a vacuum of 15 Å. Atoms in the bottom 6 atomic layers were fixed to their bulk positions, while the rest were allowed to fully relax. A 4 × 3 × 1 k-point mesh was used.

Reaction procedure and product analysis
The reaction was carried out in home-made LED photoreactors. Typically, 10 mg of substrate and 10 mg of catalyst were added into 1 mL of solvent in a 6.5 mL of quartz tube reactor, then the system was completely replaced with Ar before sealed with a cap. This quartz tube reactor could stand up 0.5 MPa pressure. The quartz tube was then irradiated with 365 nm LED light (18 W, 55 Mw cm -2 ) via side irradiation. The reaction temperature was kept between 25-35 o C. After the reaction, gas-phase products were analyzed by mass spectroscopy (MS) and gas chromatography (GC) equipped with a TCD detector and TDX-01 column. The gas was injected into the mass spectroscopy and GC via an injector. The liquid phased was analyzed by the GC equipped with FID detector and GDX-02 column, and high performance liquid chromatography (HPLC) equipped with H column. Propanol was added into the reaction liquid as the internal standard. The catalyst was filtered and the supernatant was used for the GC and HPLC analysis.
The H2 and CO2 were quantified by mass spectroscopy using an internal standard method. Ar was used as the internal standard. The molecule molecular ion peak area was used for calculation. The differential response was calibrated using the response factor. The detailed calculation was as follows: VH₂ = AH₂/AAr×VAr×KH₂ VCO₂= ACO₂/AAr×VAr×KCO₂ RH₂= VH₂/mcatalsyt In these equations: VH₂ : the volume of H2 in the quartz tube reactor; VCO₂: the volume of CO2 in the quartz tube reactor; RH₂: hydrogen generation volume per mg of catalyst; VAr: the volume of Ar in the quartz tube reactor (5.5 mL); AH₂ : the peak area of H2 (m/z=2); AAr: the peak area of Ar (m/z=40); ACO₂ : the peak area of H2 (m/z=44); KH₂ : the H2 response factor (0.1) related to Ar; KCO₂ : the CO2 response factor (2.65) related to Ar.
The CO and CH4 were quantified by GC equipped with FID detector and TDX-01 column. He was used as the carrier gas. CO2, quantified by MS, was used as an internal standard to calculate the amount of CO and CH4. The detailed calculation was as follows: VCH₄ = ACH₄/ACO₂×VCO₂×KCH₄ VCO= ACO/ACO₂×VCO₂×KCO In these equations: VCH₄ : the volume of CH4 in the quartz tube reactor; VCO: the volume of CO in the quartz tube reactor; ACH₄ : the peak area of CH4; ACO: the peak area of CO; ACO₂ : the peak area of H2; KCH₄ : the CH4 response factor (1.79) related to CO2; KCO : the CO response factor (0.8) related to CO2.
The yields are calculated based on sum of total carbon in the products. Yp=(Np☓Cp)/(Nsubstrate☓Csubstrate)☓100, where Np is the molar of product, Cp is the number of carbon in the product, Nsubstrate is the molar of substrate, and Csubstrate is the number of carbon in the substrate.
The molar of gas products is calculated based on the gas equation. N = PV/RT=101.3☓V/(8.314☓T)，where N is the molar of gas, V is the volume of the gas product, T is the temperature.
The volume of gas is depended on the temperature. We calculated the molar of gas product at room temperature (298 K).

Supplementary Discussion Reaction conditions optimization
The solvent and copper loading amount of Cu/TNR showed a great effect on the photoreforming of glycerol to methanol. The water-organic solvent mixture solvent was used as the solvent. MeCN, dioxane, DMSO and DMF were used as the organic solvent. MeCN shows the highest yield of methanol among these solvents. The ratio of organic solvent to water affects the reactions. The methanol yield first increases and then decreases with the increase of MeCN concentration. MeCN with 80% volume concentration achieves the highest yield of methanol. Cu/TNR with 0.25-5 wt % copper loading were investigated in the photo-reforming of glycerol. A 1-2 wt % copper loading shows the best performance.

The reaction route study
The liquid phase products were detected by HPLC. HPLC analysis was performed on an Agilent system equipped with a RID-6A refractive index detector. A hydrogen column (Hi-Piex H, 300 ☓7.7 mm) was used with pure water as eluent (flow rate of 0.6 ml min -1 ). The temperature of the column was kept at 65 º C. Ten microlitres of the sample were injected. Oxalic acid, glycolic acid, formaldehyde, formic acid, ethylene glycol, hydroxypropanone, 1, 3-propanediol and methanol were detected ( Figure S4). The observed products in the reaction of glycerol were separately employed as reactants ( Figure 1). Hydroxypropanone and 1, 3-propanediol were probably formed via dehydration of primary and secondary hydroxyl groups, respectively, which could not be converted into methanol under the standard reaction conditions, indicating they were not the intermediates to methanol. EG was a possible intermediate on rout to methanol as it is detected in the reaction and could be converted to methanol in high yield. Glycolic acid and oxalic acid were the overoxidaiton product of EG and mainly converted into CO2, indicating they are contributed to the formation of CO2. Formaldehyde and formic acid were the overoxidation product of methanol and were decomposed to CO, CO2 and H2 under the reaction conditions. Based on these results, the reaction route was proposed. Glycerol first undergoes C-C bond cleavage to form EG and methanol. EG can further undergo C-C bond cleavage to form methanol. The dehydration of glycerol generates hydroxypropanone and 1, 3-propanediol byproducts. The overoxidation of EG and methanol leads to the formation of overoxidation products, such as glycolic acid, oxalic acid, formaldehyde and formic acid, which were finally converted to CO, CO2 and H2.
The C-C bond cleavage of glycerol may produce EG and hydroxymethyl radicals. We then try to capture the radicals. The carbon-based radical can add to the C=C bond. Styrene was added into the reaction to capture the radicals. After the reaction, the reaction liquid was qualified by GC-MS. 3-Phenylpropanol and 4-phenylbutane-1,2-diol were detected, confirming the formation of EG and methoxyl carbon radicals via C-C bond cleavage.

Conversion of cellulose
It is a challenge to direct convert cellulose due to the poor solubility. We tried to convert cellulose via a two-step strategy. Firstly, the hydrogenolysis of cellulose to a mixture of polys and then photo-reforming of the reaction liquid to methanol or syngas. The reaction liquid was provided from Tao Zhang group (Dalian Institute of Chemical Physics, Chinese Academy of Sciences). The catalytic conversion of cellulose (Merck, microcrystalline) was carried out in a stainless-steel autoclave (Parr Instrument Company, 100 mL) at an H2 pressure of 4 MPa (measured at room temperature) and at 518 K for 30 min. For each reaction, 0.5 g of cellulose, 0.05g of Raney Ni, 0.05 g of tungstic acid and 50 mL of water were put into the reactor and stirred at a rate of 1000 r min -1 . After the reaction, the catalyst was filtered. The water reaction solution was treated with Ba(OH)2 to remove the tungstic acid, and the pH value of the resulting solution is 9-10. Cellulose was completely converted. The yields of polyols were as follows: 6.9% of sorbitol, 3.3% of erythritol, 2.2% of glycerol, 63.6% of EG and 7.7% of 1, 2-propanol.
The procedure for the photo-reforming of the cellulose reaction solution is as follows. Procedure A: 10 mg of 2Cu/TNR, 0.2 mL (equal to 2 mg of the cellulose) of the reaction solution, 0.8 mL of MeCN, were added into the quartz tube reactor, then the system was completely replaced with Ar before sealed with a cap. The quartz tube was then irradiated with 365 nm LED light (9 W) for 12 h. The reaction temperature was kept between 25-35 o C.
Procedure B: 10 mg of 0.1Cu/TNR, 0.05 mL (equal to 2 mg of the cellulose) of the concentrated reaction solution, 9.5 mL of MeCN, were added into the quartz tube reactor, then the system was completely replaced with Ar before sealed with a cap. The quartz tube was then irradiated with 365 nm LED light (9 W) for 72 h. The reaction temperature was kept between 25-35 o C.
Under the procedure A, 16% yield of methanol was obtained based on the cellulose. Under the procedure B, 50% and 5% yield of CO and CO2 were obtained from cellulose, respectively. The ratio of H2/CO is 6.7.

Conversion of native biomass
The beech sawdust was treated with an acid solution to convert the hemicellulose and cellulose into soluble sugars. The resulting sugar solution was further used as the substrate for the photoreforming reaction. Typically, 0.5 g of each biomass sample was loaded into 50-mL beakers with the addition of 7.5 mL of a 72 wt % H2SO4 solution. The mixture was left at room temperature for 2 h under stirring. Afterward, the slurry was transferred into a round-bottom-flask and 90 mL of water was added to reach a H2SO4 concentration of 3 wt %. The solution was heated under reflux (100 o C) under magnetically stirring for 6 h. The resultant solution was filtered and the filtrate was analyzed by HPLC. The as-obtained solution was treated with Ba(OH)2 to removed H2SO4. The formed BaSO4 solid was filtered, and the filtrate was concentrated to 10 mL. The pH value of the final solution was 9-10. Then, the as-obtained solution was subjected to further photo reaction. Typically, 10 mg of 0.1Cu/TNR and 50 μl of the solution were added to 0.95 mL of MeCN in a 6.5 mL of quartz tube reactor, then the system was completely replaced with Ar before sealed with a cap. This quartz tube reactor could stand up 0.5 MPa pressure. The quartz tube was then irradiated with 365 nm LED light (18 W). The reaction temperature was kept between 25-35 o C. The results were shown in table S5.
After acid treatment, based on the mass of the beech sawdust, 21 wt % glucose, 14 wt % xylose, 6 wt % fructose and 4 wt % formic acid were left in the solution, which accounts for 45 wt % of the beech sawdust. The soluble products in the solution come from the (hemi)cellulose part of beech sawdust. Based on the (hemi)cellulose, 88 wt % of (hemi)cellulose was released to the solution as small molecules. The carbon content of beech sawdust was determined to 49 wt % by element analysis. Based on the carbon analysis, about 33% carbon of beech sawdust was left in the solution. After photoreaction for 6 h, based on the carbon in the solution, 11% of methanol, 44% of CO and 11% of CO2 were formed, which accounts for 66% and 22% of carbon in the solution and beech sawdust, respectively. Further prolonging the reaction time to 24 h generates 50% of CO, 23% of CO2 and 23% of CH4, which accounts 96% and 32% of carbon in the solution and beech sawdust, respectively.

Carbon mass balance analysis
The liquid phase products were analyzed by HPLC. After photoreaction, the liquid phase products are mostly the same for different substrates, which is probably due to the similar polyols structure. The identified compounds are as follows: oxalic acid, glycolic acid, formaldehyde, formic acid, ethylene glycol, hydroxypropanone, 1, 2-propanediol, 1, 3-propanediol and methanol. For some substances, there are some unidentified compounds, which accounts for the relatively lower carbon balance. The results were shown in Table S1-5. The carbon balance is calculated based on the following equation.
Carbon balance =ΣMi×ni/M×n×100 In these equations: Mi: the molar of the products; M: the molar of the substance; ni: the carbon number of the products; n: the carbon number of the substance; Beech sawdust is derived from the group of Fang Lu (Dalian Institute of Chemical Physics, Chinese Academy of Sciences). The composition has been analyzed, 12 which comprises 47 wt % of cellulose, 21 wt % of hemicellulose and 21 wt % of lignin. The total carbon in beech sawdust and the solid after acid treatment of beech sawdust were 45 wt % and 64 wt %, respectively. The total carbon content of cellulose and hemicellulose was calculated to be 60%. The carbon content of lignin is calculated to be 40%. The carbon balance for the beech sawdust is calculated based on cellulose and hemicellulose.
The apparent quantum yield (AQY) was measured over 2Cu/TNR with UV LEDs light (input power of 50 W, 365 nm) by top irradiation. We measured the apparent quantum yield was calculated based on the following equation.

Apparent quantum yield (AQY) measurements
The AQY was measured over 2Cu/TNR in MeCN-H2O (8:2) solution with UV LEDs light (input power of 50 W, 365 nm) by top irradiation. We measured the apparent quantum yield was calculated based on the following equation where nmethanol, nmethanol, and nhydrogen are the molar amount of methanol, EG and hydrogen, respectively, and NA, I, t and S represent Avogadro's constant, light intensity, reaction time and irradiation area, respectively.

Catalyst characterization
To understand the form and structure of the copper species in the Cu/TNR catalysts prepared under different conditions, Cu K-edge XAFS was used for local structure analysis of the Cu species as shown in Figure 5b, along with those of Cu, Cu2O and CuO. Cu2O exhibit a low energy peak in the region between 8982 eV, which has been assigned as a Cu ls→4p transition. 13 The X-ray absorption pre-edge feature of CuO shows a weaker peak at about 8987 eV. In addition, after the absorption maximum, the absorption intensities of Cu and Cu2O show a little decrease, while that of CuO decreases more sharply. At low copper loading, 0.1Cu/TNR shows a peak at 8987 eV, while with increasing the copper loading, the peak in the pre-edge is not obvious, but it can be estimated that the copper species in them exist mainly as Cu 2+ because their sharply decreased absorption features are similar to that of CuO. Figure 5b shows the radial structure functions (RSF) of Cu, Cu2O, CuO, and Cu/TNR samples obtained by k3-weighted Fourier transformation. According to the literature, 14,15 metallic Cu shows a strong peak at about 2.10Å corresponding to the Cu-Cu distance. Cu2O shows two strong peaks at about 1.48 and 2.72Å corresponding to the Cu-O and Cu-Cu distances, respectively. CuO shows three distinct peaks at about 1.53Å (Cu-O), 2.46Å (Cu-Cu) and 2.90Å (Cu-Cu). At low copper loading, the presence of Cu-O peak and absence of Cu-Cu peak for 0.1Cu/TNR indicates the Cu 2+ single sites are highly dispersed in the TiO2. As the copper loading increase, weak Cu-Cu peaks gradually appeared, indicating the CuO clusters or nanoparticles formed. These results suggest that the copper species gradually transformed from single copper dopant to CuO particles.
The above characterization results demonstrated that there are two kinds of copper species: the doped Cu 2+ and CuOx nanoparticles. At low copper loading amount, highly dispersed Cu 2+ dopants are dominant, and with increasing the copper loading amount, CuOx nanoparticles gradually became the major copper species.

Supplementary Figures
Supplementary Figure 1 The effect of the solvent. Reaction conditions: 10 mg of glycerol, 10 mg of 1Cu/TNR, 0.8 mL of organic solvent, 0.2 mL of water, 365 nm LED (18 W, 55 mW cm -2 ) irradiation for 12 h.

Supplementary
[a] The carbon balance is calculated based on cellulose and hemicellulose.