Stepwise photoassisted decomposition of carbohydrates to H2

Abstract

•Stepwise carbohydrate decomposition realizes H2 production and storage•Photocatalysis over Ta-CeO2 splits carbohydrates into C1 LHCs•Heating promotes selective C–C bond breaking powered by visible light•Concentrated solar light splits glucose to C1 LHCs in a flow apparatus Hydrogen (H2), as an indispensable clean energy vector, has been well demonstrated to be produced via biomass photoreforming powered by solar light. For future biomass refining, biomass photoreforming deserves a high decomposition extent of biomass to maximize H2 production, as the greenhouse gas emissions from biomass acquisition and pretreatment will minimize per mass of H2. The main obstacle to high H2 yield is the far insufficient C–C bond breaking to convert biomass carbons into CO2 with maximization of H2 production. Here, we emphasize C–C bond breaking instead of direct H2 production. Such a “C–C bond first” strategy realizes conversion of carbohydrates into C1 liquid hydrogen carriers (LHCs, consisting of HCOOH and HCHO) over Ta-CeO2 photocatalyst and is demonstrated in a flow apparatus powered solely by solar energy. The LHCs can release H2 on-site where needed by either photocatalysis or thermocatalysis. This work provides a new perspective for H2 production by photocatalysis. Biomass reforming by harvesting solar energy can provide green hydrogen. Current biomass photoreforming provides H2 erratically and in limited yield although efficiently, owing to intermittent features of solar light and incomplete degradation of biomass C–C bonds. Here, we detour the flaws by prioritizing conversion of carbohydrates to liquid hydrogen carriers (LHCs, consisting of HCOOH and HCHO), appropriate for transportation. Subsequently, the LHCs are fully decomposed, releasing only H2 and CO2. This stepwise process enables complete scission of carbohydrate C–C bonds, affording 44 g of H2 per kg of glucose thereof. Intermittent solar light provides the photoenergy and heat to split glucose carbons to produce LHCs (2.5 mmol h−1) in a flow apparatus. This work demonstrates hydrogen production and storage by emphasizing the complete scission of biomass C–C bonds. Biomass reforming by harvesting solar energy can provide green hydrogen. Current biomass photoreforming provides H2 erratically and in limited yield although efficiently, owing to intermittent features of solar light and incomplete degradation of biomass C–C bonds. Here, we detour the flaws by prioritizing conversion of carbohydrates to liquid hydrogen carriers (LHCs, consisting of HCOOH and HCHO), appropriate for transportation. Subsequently, the LHCs are fully decomposed, releasing only H2 and CO2. This stepwise process enables complete scission of carbohydrate C–C bonds, affording 44 g of H2 per kg of glucose thereof. Intermittent solar light provides the photoenergy and heat to split glucose carbons to produce LHCs (2.5 mmol h−1) in a flow apparatus. This work demonstrates hydrogen production and storage by emphasizing the complete scission of biomass C–C bonds. Hydrogen (H2) has emerged as a sustainable and indispensable energy vector that can be produced from renewable energy resources.1Turner J.A. Sustainable hydrogen production.Science. 2004; 305: 972-974https://doi.org/10.1126/science.1103197Crossref PubMed Scopus (4261) Google Scholar,2Liu W. Cui Y. Du X. Zhang Z. Chao Z.S. Deng Y.L. High efficiency hydrogen evolution from native biomass electrolysis.Energy Environ. Sci. 2016; 9: 467-472https://doi.org/10.1039/C5EE03019FCrossref Google Scholar Storage and transportation of H2 remain challenging in the aspect of its worldwide applications.3Chen Z. Li P. Anderson R. Wang X. Zhang X. Robison L. Redfern L.R. Moribe S. Islamoglu T. Gómez-Gualdrón D.A. et al.Balancing volumetric and gravimetric uptake in highly porous materials for clean energy.Science. 2020; 368: 297-303https://doi.org/10.1126/science.aaz8881Crossref PubMed Scopus (268) Google Scholar In this context, formic acid (HCOOH) has been identified as a promising liquid hydrogen carrier (LHC), thermally stable up to 200°C,4Fein D.E. Wachs I.E. Quantitative determination of the catalytic activity of bulk metal oxides for formic acid oxidation.J. Catal. 2002; 210: 241-254https://doi.org/10.1006/jcat.2002.3683Crossref Scopus (32) Google Scholar,5Zhou J. Mullins D.R. Adsorption and reaction of formaldehyde on thin-film cerium oxide.Surf. Sci. 2006; 600: 1540-1546https://doi.org/10.1016/j.susc.2006.02.009Crossref Scopus (80) Google Scholar easy to transport, and be back decomposed with H2 release by thermo-, photo-, or electrocatalysis.6Kumar A. Daw P. Milstein D. Homogeneous catalysis for sustainable energy: hydrogen and methanol economies, fuels from biomass, and related topics.Chem. Rev. 2022; 122: 385-441https://doi.org/10.1021/acs.chemrev.1c00412Crossref PubMed Scopus (63) Google Scholar,7Boddien A. Mellmann D. Gärtner F. Jackstell R. Junge H. Dyson P.J. Laurenczy G. Ludwig R. Beller M. Efficient dehydrogenation of formic acid using an iron catalyst.Science. 2011; 333: 1733-1736https://doi.org/10.1126/science.1206613Crossref PubMed Scopus (641) Google Scholar,8Wei D. Sang R. Sponholz P. Junge H. Beller M. Reversible hydrogenation of carbon dioxide to formic acid using a Mn-pincer complex in the presence of lysine.Nat. Energy. 2022; 7: 438-447https://doi.org/10.1038/s41560-022-01019-4Crossref Scopus (21) Google Scholar Therefore, a very promising and challenging way of exploiting the hydrogen economy is based on the use of HCOOH, which must be produced from renewable resources.9Zhang P. Guo Y.-J. Chen J. Zhao Y.-R. Chang J. Junge H. Beller M. Li Y. Streamlined hydrogen production from biomass.Nat. Catal. 2018; 1: 332-338https://doi.org/10.1038/s41929-018-0062-0Crossref Scopus (73) Google Scholar,10Han X. Sheng H. Yu C. Walker T.W. Huber G.W. Qiu J. Jin S. Electrocatalytic oxidation of glycerol to formic acid by CuCo2O4 spinel oxide nanostructure catalysts.ACS Catal. 2020; 10: 6741-6752https://doi.org/10.1021/acscatal.0c01498Crossref Scopus (112) Google Scholar Broadly available next-generation biomasses containing both carbon and hydrogen are such kinds of renewable resources.11Huber G.W. Shabaker J.W. Dumesic J.A. Raney Ni-Sn catalyst for H2 production from biomass-derived hydrocarbons.Science. 2003; 300: 2075-2077https://doi.org/10.1126/science.1085597Crossref PubMed Scopus (847) Google Scholar,12Liao Y. Koelewijn S.F. Van den Bossche G. Van Aelst J. Van den Bosch S. Renders T. Navare K. Nicolaï T. Van Aelst K. Maesen M. et al.A sustainable wood biorefinery for low-carbon footprint chemicals production.Science. 2020; 367: 1385-1390https://doi.org/10.1126/science.aau1567Crossref PubMed Scopus (404) Google Scholar Both hydrogen production and storage can be achieved if biomass is converted to HCOOH. Cellulose, hemicelluloses, and their hydrolyzates can generate HCOOH by thermocatalysis,9Zhang P. Guo Y.-J. Chen J. Zhao Y.-R. Chang J. Junge H. Beller M. Li Y. Streamlined hydrogen production from biomass.Nat. Catal. 2018; 1: 332-338https://doi.org/10.1038/s41929-018-0062-0Crossref Scopus (73) Google Scholar processes that require stoichiometric oxidants with strong oxidation potential to break the C–C bonds. However, these carbohydrates suffer over-oxidation to CO2 on catalysts.13Zhang Z. Huber G.W. Catalytic oxidation of carbohydrates into organic acids and furan chemicals.Chem. Soc. Rev. 2018; 47: 1351-1390https://doi.org/10.1039/C7CS00213KCrossref PubMed Google Scholar,14Hou Y. Niu M. Wu W. Catalytic oxidation of biomass to formic acid using O2 as an oxidant.Ind. Eng. Chem. Res. 2020; 59: 16899-16910https://doi.org/10.1021/acs.iecr.0c01088Crossref Scopus (20) Google Scholar Producing strong oxidants to cleave C–C bonds represents the advantage of photocatalysis.15Nosaka Y. Nosaka A.Y. Generation and detection of reactive oxygen species in photocatalysis.Chem. Rev. 2017; 117: 11302-11336https://doi.org/10.1021/acs.chemrev.7b00161Crossref PubMed Scopus (1840) Google Scholar,16Uekert T. Pichler C.M. Schubert T. Reisner E. Solar-driven reforming of solid waste for a sustainable future.Nat. Sustain. 2020; 4: 383-391https://doi.org/10.1038/s41893-020-00650-xCrossref Scopus (70) Google Scholar Carbohydrates are easily over-oxidized with less HCOOH generation because of the relatively long carbon chains; nevertheless, more H2 is produced from carbohydrates concomitantly due to the charge balance of photocatalysts,17Wakerley D.W. Kuehnel M.F. Orchard K.L. Ly K.H. Rosser T.E. Reisner E. Solar-driven reforming of lignocellulose to H2 with a CdS/CdOx photocatalyst.Nat. Energy. 2017; 2: 17021https://doi.org/10.1038/nenergy.2017.21Crossref Scopus (302) Google Scholar which is the current consensus of exploiting photogenerated oxidants for H2 generation from biomass (Figure S1 path A).18Kuehnel M.F. Reisner E. Solar hydrogen generation from lignocellulose.Angew. Chem. Int. Ed. Engl. 2018; 57: 3290-3296https://doi.org/10.1002/anie.201710133Crossref PubMed Scopus (137) Google Scholar The principle is to generate holes or ⋅OH, which breaks the O–H bonds of biomass, mainly the carbohydrate components, generating oxygen-centered radicals that experience β-scission to cleave the C–C bonds.19Zhang K. Chang L. An Q. Wang X. Zuo Z. Dehydroxymethylation of alcohols enabled by cerium photocatalysis.J. Am. Chem. Soc. 2019; 141: 10556-10564https://doi.org/10.1021/jacs.9b05932Crossref PubMed Scopus (85) Google Scholar,20Nguyen S.T. Murray P.R.D. Knowles R.R. Light-driven depolymerization of native lignin enabled by proton-coupled electron transfer.ACS Catal. 2020; 10: 800-805https://doi.org/10.1021/acscatal.9b04813Crossref Scopus (51) Google Scholar However, the co-generated carbon-centered radicals and those derived from direct oxidation of the C–H bond by photogenerated oxidants tend to form C–C bonds when external or concurrently generated oxidants are absent (Figure S1 path B).21Kisch H. Semiconductor photocatalysis for chemoselective radical coupling reactions.Acc. Chem. Res. 2017; 50: 1002-1010https://doi.org/10.1021/acs.accounts.7b00023Crossref PubMed Scopus (123) Google Scholar,22Luo N. Nie W. Mu J. Liu S. Li M. Zhang J. Gao Z. Fan F. Wang F. Low-work function metals boost selective and fast scission of methanol C–H bonds.ACS Catal. 2022; 12: 6375-6384https://doi.org/10.1021/acscatal.1c06005Crossref Scopus (5) Google Scholar This mechanism of photochemical recondensation is adverse to C–C bond breaking, as biomass is converted to recalcitrant architecture like humin and wasted.23Huang Z. Luo N. Zhang C. Wang F. Radical generation and fate control for photocatalytic biomass conversion.Nat. Rev. Chem. 2022; 6: 197-214https://doi.org/10.1038/s41570-022-00359-9Crossref Scopus (18) Google Scholar Neglecting the principle for C–C bond scission limits the amount of H2 per mass of biomass.18Kuehnel M.F. Reisner E. Solar hydrogen generation from lignocellulose.Angew. Chem. Int. Ed. Engl. 2018; 57: 3290-3296https://doi.org/10.1002/anie.201710133Crossref PubMed Scopus (137) Google Scholar Besides, the intermittent feature of solar light provides H2 erratically. Tackling these challenges requires prioritized biomass conversion to HCOOH or equivalent C1 LHCs, appropriate for on-site hydrogen production independent of solar light irradiation.24Cao S. Chen Y. Wang H. Chen J. Shi X. Li H. Cheng P. Liu X. Liu M. Piao L. Ultrasmall CoP nanoparticles as efficient cocatalysts for photocatalytic formic acid dehydrogenation.Joule. 2018; 2: 549-557https://doi.org/10.1016/j.joule.2018.01.007Abstract Full Text Full Text PDF Scopus (96) Google Scholar,25Kar S. Rauch M. Leitus G. Ben-David Y. Milstein D. Highly efficient additive-free dehydrogenation of neat formic acid.Nat. Catal. 2021; 4: 193-201https://doi.org/10.1038/s41929-021-00575-4Crossref Scopus (49) Google Scholar However, fully breaking the C–C bonds to produce these active C1 LHCs is challenging, as the C1 LHCs are readily decomposed into side products either on photocatalysts or with the catalysis of acids generated concurrently.26Wang M. Liu M. Lu J. Wang F. Photo splitting of bio-polyols and sugars to methanol and syngas.Nat. Commun. 2020; 11: 1083https://doi.org/10.1038/s41467-020-14915-8Crossref PubMed Scopus (46) Google Scholar In an approach orthogonal to direct H2 production from biomass, here we devise the strategy of “C–C bond first” that carbohydrates from hydrolysis of lignocellulose are first converted to LHCs with nearly 100% yield of C1 products (Figure 1). Subsequent degradation of the LHCs provides H2 and CO2 exclusively. The photocatalytic oxidation of carbohydrates in the first step targets splitting carbohydrate carbons into C1 LHCs, distinct from those reported for oxidative functionalization of carbohydrates. This oxidation step was conducted over strongly distorted Ta-doped CeO2 (Ta-CeO2), and the kinetics of C–C bond scission, C–C bond coupling, and conversion of recalcitrant carbonyl intermediates were controlled to realize a high LHCs selectivity. The obtained aqueous solution of LHCs releases gaseous H2 where needed by either photocatalysis or thermocatalysis. Solar-light-driven catalytic oxidation of glucose in a flow apparatus demonstrates the allure of the biomass conversion to LHCs. The Ta content of Ta-CeO2 was determined to be 0.78 mol % by inductively coupled plasma atomic emission spectroscopy (ICP-AES). X-ray diffraction (XRD) patterns (Figure S2A) of CeO2, the contrast catalyst, and Ta-CeO2 show fluorite structure. The diffraction peaks of Ta-CeO2 negligibly shift (Figure S2B), in agreement with the low amount of Ta dopant. Representative high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of the Ta-CeO2 shows the morphology of a rectangular structure (Figure 2A) with an average size of about 7.7 nm (Figure S3). The lattice fringes with spacings of 0.27 and 0.29 nm (Figure 2A) are attributed to the (200) and (020) interplanar distances of fluorite CeO2, respectively. The HAADF-STEM image and the elemental mappings by energy-dispersive X-ray spectra (EDS) show the uniform dispersion of Ta and Ce elements (Figure 2B), implying that Ta is homogeneously incorporated into the matrix of CeO2 (Figure 2C). The inclusion of Ta in the CeO2 matrix is confirmed by X-ray absorption fine structure (XAFS) analysis of the Ta-CeO2 sample at the Ta L3 edge. The X-ray absorption near edge structure (XANES) spectrum of the Ta-CeO2 (Figure 2D) shows that Ta is in the +5 oxidation state as in Ta2O5. The fitting of the k1-weighted extended X-ray absorption fine structure (EXAFS) signal (Figure 2E) reveals that the average local structure around Ta atoms can be reconstructed, including ∼8 oxygen atoms at two different Ta–O distances and ∼2 Ce atoms in the second shell (results are summarized in Table S2). The possibility of observing a Ta–Ce distance in the second shell is a clear indication of the substitution of Ta5+ for Ce4+ in the CeO2 matrix. Despite this, the local environment around Ta ions is strongly distorted with respect to the cubic symmetry of the site hosting the cations in the fluorite structure of CeO2, in agreement with HAADF-STEM results. Moreover, the Ta–O and Ta–Ce distances are significantly shorter than those expected for the fluorite structure of CeO2 (Ce–O distance of 0.234 nm and Ce–Ce distance of 0.382 nm).27Shahin A.M. Grandjean F. Long G.J. Schuman T.P. Cerium LIII-edge XAS investigation of the structure of crystalline and amorphous cerium oxides.Chem. Mater. 2005; 17: 315-321https://doi.org/10.1021/cm0492437Crossref Scopus (75) Google Scholar The strong lattice distortion of the local structure around Ta can be easily justified considering the ionic radii of Ta5+ and Ce4+ in the cubic coordination: 0.074 and 0.097 nm, respectively.28Shannon R.D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides.Acta Cryst. A. 1976; 32: 751-767https://doi.org/10.1107/S0567739476001551Crossref Scopus (53685) Google Scholar Considering the smaller ionic radius of Ta5+, the local structure around Ta ions must be contracted with the oxygen anions moving toward the dopant, with consecutive relaxation of the successive shell of Ce ions. This distortion can be partially compensated by the formation of one Ce3+ expected in the proximity of each Ta5+ dopant as a result of charge compensation and that Ce3+ is larger than Ce4+ (ionic radii of 0.114 versus 0.097 nm),28Shannon R.D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides.Acta Cryst. A. 1976; 32: 751-767https://doi.org/10.1107/S0567739476001551Crossref Scopus (53685) Google Scholar in agreement with the results of Raman spectra (Figure S4). Compared with the as-prepared CeO2, which has an indirect band gap of 2.60 eV (see Note S1), the band gap of Ta-CeO2 is narrower (2.37 eV) owing to the lattice distortion and increased Ce3+ concentration (Figure S5). In direct photoreforming of biomass, the biomass is partially oxidized, H2 is thus produced in a limited yield, and biomass degrades to by-products that require tedious separations or disposal.18Kuehnel M.F. Reisner E. Solar hydrogen generation from lignocellulose.Angew. Chem. Int. Ed. Engl. 2018; 57: 3290-3296https://doi.org/10.1002/anie.201710133Crossref PubMed Scopus (137) Google Scholar The underutilization of biomass originates from incomplete scission of biomass C–C bond and radical coupling in an environment lacking oxidative oxygen radicals, such as O2⋅− and ⋅OH. Glycerol was used as a model molecule at the beginning of the research, although we realize that oxidative splitting of carbohydrate carbons is much more arduous than splitting glycerol carbons because of the longer carbon chains (Table S1). However, the simple and representative polyol structure is beneficial for intermediate detection and mechanism study. Photocatalytic oxidation of glycerol by O2 (5 bar) was evaluated at near room temperature (36°C) over Ta-CeO2 using a homemade photoreactor (Figures 3A and S6). The conversion of glycerol was 26%, and the yield of LHCs was 6% after photoirradiation for 10 h by blue light-emitting diodes (LEDs, 452 ± 10 nm, 200 mW cm−2), indicating that C–C bonds were broken incompletely as widely reported.29Montini T. Gombac V. Sordelli L. Delgado J.J. Chen X. Adami G. Fornasiero P. Nanostructured Cu/TiO2 photocatalysts for H2 production from ethanol and glycerol aqueous solutions.ChemCatChem. 2011; 3: 574-577https://doi.org/10.1002/cctc.201000289Crossref Scopus (158) Google Scholar The yield of LHCs was improved to 74% at 120°C, with glycerol conversion of 93% (43% apparent quantum efficiency). The photocatalytic glycerol oxidation can be conducted in air (25 bar), affording LHCs in 58% yield with glycerol conversion of 75% in a similar O2 partial pressure (5 bar, Figure S8). The selectivity of LHCs (65%, Table S3, entry 1) reduces at a higher reaction temperature (150°C), owing to the prevalent over-oxidation of HCOOH to CO2. This degradation of glycerol to LHCs also relies on photoirradiation, considering glycerol is scarcely converted in the dark even at 120°C (Figure S9A). Glycerol is converted to LHCs with high selectivity (80%) even irradiated with weak light intensity (18 mW cm−2), indicating the synergy of photo and thermal is essential for the splitting of glycerol carbons. When irradiated with a light intensity larger than 130 mW cm−2, the conversion of glycerol and selectivity of LHCs remained unchanged, illustrating that either the light absorption was saturated or the elementary steps involving photogenerated charges were no longer rate-limiting. Under the varied range of light intensities, the carbon balance which is defined as Equation 9 is within 95%–103% (Figure S9B). Glycerol was converted rapidly in the first 2 h under the optimized reaction conditions and was entirely converted after reacting for 20 h, according to the time profiles; the final yield of LHCs reached 86% (Figure S10A). During photocatalytic glycerol oxidation, the carbon balance increased from 72% at 2 h to nearly 100% when glycerol was fully converted (Figure S10B). The Ta-CeO2 photocatalyst shows minor deactivation during the first run and slightly higher deactivation (about 12% according to glycerol conversion) for the subsequent runs (Figure S11). The observed slight deactivation is not amazing and can be explained by the poisoning of Ta-CeO2 by HCOOH or formaldehyde (HCHO) during photocatalysis. LHCs consisting of HCHO and HCOOH could be produced over pristine CeO2 in 58% yield (Table S3, entry 4) after 10 h of reaction at 120°C; the conversion of glycerol was 73%, lower than that (93%) using Ta-CeO2. Of note, Ta-doped other metal oxides, such as TiO2 and Nb2O5, are not selective to C1 LHCs. The promotional role of Ta that is doped into CeO2 could be attributed to the abundantly present Ce3+, deriving from the substitution of Ta5+ for Ce4+ in the CeO2 matrix (see Note S1). The higher concentration of Ce3+ in Ta-CeO2 is expected to adsorb more reactant intermediates,30Campbell C.T. Peden C.H. Chemistry. Oxygen vacancies and catalysis on ceria surfaces.Science. 2005; 309: 713-714https://doi.org/10.1126/science.1113955Crossref PubMed Scopus (978) Google Scholar,31Yang C. Yu X. Pleßow P.N. Heißler S. Weidler P.G. Nefedov A. Studt F. Wang Y. Wöll C. Rendering photoreactivity to ceria: the role of defects.Angew. Chem. Int. Ed. Engl. 2017; 56: 14301-14305https://doi.org/10.1002/anie.201707965Crossref PubMed Scopus (30) Google Scholar thereby breaking the C–C bond scission to produce LHCs in high selectivity. The high concentration of Ce3+ in Ta-CeO2 also promotes electron transition from O 2p to Ce 4f orbits (Figure S12),32Xie S. Wang Z. Cheng F. Zhang P. Mai W. Tong Y. Ceria and ceria-based nanostructured materials for photoenergy applications.Nano Energy. 2017; 34: 313-337https://doi.org/10.1016/j.nanoen.2017.02.029Crossref Scopus (102) Google Scholar generating more photogenerated carriers for glycerol conversion. Doping Ta into CeO2 can weaken the adsorption of HCHO and HCOOH on CeO2 according to the results of temperature-programmed desorption (TPD) experiments (Figure S13; Note S6).33Bu Y.B. Chen Y.F. Jiang G.M. Hou X.M. Li S. Zhang Z.T. Understanding of Au-CeO2 interface and its role in catalytic oxidation of formaldehyde.Appl. Catal. B. 2020; 260: 118138https://doi.org/10.1016/j.apcatb.2019.118138Crossref Scopus (61) Google Scholar We studied the reaction routes of the photocatalytic glycerol oxidation. The radical intermediates were captured by benzyl acrylate (BA) and 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO). Quadrupole-time of flight mass spectrometer (Q-TOF MS) was used to analyze the products derived from the reaction of radical intermediates and trapping reagents. The adducts derived from the addition reaction of BA and glycerol were detected with the addition of 1 equivalent of BA (Figure S14A), implying that radicals derived from C–H bond scission were the intermediates. C2 intermediates in the form of glycolaldehyde and glycolic acid were detected according to the m/z of 214 and 75 (Figure S14B), attributed to TEMPO-glycolaldehyde adduct and glycolic acid, respectively.34Luo N. Montini T. Zhang J. Fornasiero P. Fonda E. Hou T. Nie W. Lu J. Liu J. Heggen M. et al.Visible-light-driven coproduction of diesel precursors and hydrogen from lignocellulose-derived methylfurans.Nat. Energy. 2019; 4: 575-584https://doi.org/10.1038/s41560-019-0403-5Crossref Scopus (156) Google Scholar Photocatalytic scission of the C–C bonds in these C2 intermediates also requires heat (Figure S15). Other intermediates were probed by control experiments. Photocatalytic conversion of hypothesized C3 intermediates (Table S3, entries 6, 7, and 8), including glycerol acid, glyceraldehyde, and 1,3-dihydroxyacetone, produces LHCs with yields of 8%, 55%, and 47%, respectively, suggesting that glyceraldehyde and 1,3-dihydroxyacetone are partially involved in the photocatalytic glycerol oxidation. We also tried to reveal possible C2 intermediates. Photocatalytic glycolaldehyde oxidation (Table S3, entry 9) was very fast, affording LHCs and CO2 with 80% and 15% yields in 3 h, respectively. Photocatalytic oxidation of glycolic acid with prolonged reaction time (6 h) produced LHCs (25%) and CO2 (27%) in low yields (Table S3, entry 10). The above results indicate that glycolaldehyde is the major intermediate and derives from the C–C bond scission of glycerol via the initial scission of the terminal C–H bond. Tentative reaction routes were proposed for the photocatalytic glycerol oxidation (Figure S16), combing with the results that photogenerated holes, electrons, and O2⋅− are essential to the photocatalytic glycerol oxidation (Figure S17). Irradiation of Ta-CeO2 generates holes and electrons. Photogenerated holes oxidize the C–H bond of glycerol, affording carbon-centered radicals mainly at the terminal carbon of glycerol (path I). The carbon-centered radicals react with O2⋅− derived from the reduction of O2 by photogenerated electrons, bringing about C–C bond scission with the generation of glycolaldehyde and HCOOH. The glycolaldehyde then experiences C–C bond scission, producing HCOOH and HCHO in equal amounts. Parallel to this reaction path (path I), the C–C bond scission of carbon-centered radicals at the middle carbon produces HCOOH, HCHO, and CO2 (path II). We then studied the catalytic mechanism of heat in promoting photocatalytic oxidation of polyols by monitoring the selectivity of C2+ products, of which the selectivity could be calculated by subtracting C1 products from all the products. A high proportion of C–C bonds is retained at low reaction temperatures according to the moderate selectivity (76%) of C2+ products for photocatalytic glycerol oxidation at 36°C (Figure 3B). The selectivity of C2+ products decreases when improving the reaction temperature, with a selectivity of 5% at 120°C. The undetected carbons were analyzed by Q-TOF MS (Figure S18). In the high-resolution MS, the peaks with m/z of 89.0245, 191.0737, 211.0830, and 255.2341 are ascribed to C3H6O3 and C6H10O4, C8H6O4, and C10H10O5, respectively. The intensities of MS peaks were stronger at 36°C than those at 120°C, in agreement with the results depicted in Figure 3B. Particularly, C3H6O3, ascribed to glyceraldehyde or 1,3-dihydroxyacetone, is produced from the dehydrogenation of glycerol, indicating that the elimination of H2 molecule from glycerol is more overwhelming than breaking the C–C bonds of glycerol at low temperature (36°C). The formation of C6H10O4, C8H6O4, and C10H10O5 during photocatalytic glycerol oxidation indicates a significant contribution of C–C bond coupling. Generally, the C–C bonds cannot be directly activated,35Liao F. Lo T.W.B. Sexton D. Qu J. Wu C.-T. Tsang S.C.E. PdFe nanoparticles as selective catalysts for C–C cleavage in hydrogenolysis of vicinal diol units in biomass-derived chemicals.Catal. Sci. Technol. 2015; 5: 887-896https://doi.org/10.1039/C4CY01159GCrossref Google Scholar,36Li Z. Yan Y. Xu S.M. Zhou H. Xu M. Ma L. Shao M. Kong X. Wang B. Zheng L. et al.Alcohols electrooxidation coupled with H2 production at high current densities promoted by a cooperative catalyst.Nat. Commun. 2022; 13: 147https://doi.org/10.1038/s41467-021-27806-3Crossref PubMed Scopus (32) Google Scholar and their scission is generally less prone than the dehydrogenation of hydroxyl moieties and C–C bond coupling between two radicals.37Liu M. Wang Y. Kong X. Rashid R.T. Chu S. Li C.-C. Hearne Z. Guo H. Mi Z. Li C.-J. Direct catalytic methanol-to-ethanol photo-conversion via methyl carbene.Chem. 2019; 5: 858-867https://doi.org/10.1016/j.chempr.2019.01.005Abstract Full Text Full Text PDF Scopus (40) Google Scholar Considering the above results, the low selectivity of C2+ products at high temperatures implies that heating may assist C–C bond scission. Therefore, both the products from dehydrogenation and C–C bond coupling are converted, preferentially generating HCOOH and HCHO. To prove the hypothesis, we measured the apparent activation energy (Ea) and the kinetic isotope effect (KIE) in the photocatalytic glycerol oxidation. The Ea measured at low light intensity (10 mW cm−2) represents that of dark conditions, considering glycerol is scarcely converted in the dark. The reaction rate constants, fitted from first-order kinetics (a linear relationship between the logarithm of glycerol concentration [Cglycerol] and the reaction time; Figure S19),38Park H. Yun Y.S. Kim T.Y. Lee K.R. Baek J. Yi J. Kinetics of the dehydration of glycerol over acid catalysts with an investigation of deactivation mechanism by coke.Appl. Catal. B. 2015; 176–177: 1-10https://doi.org/10.1016/j.apcatb.2015.03.046Crossref Scopus (52) Google Scholar were obtained at various temperatures under irradiation with light intensities of 10 and 200 mW cm−2, respectively (Figure 3C). The linear fit between the logarithm of rate constants and the reciprocal of the reaction temperature affords the Ea, which is determined to be 24.5 and 24.9 kJ mol−1 for the reactions with light intensities of 10 and 200 mW cm−2, respectively. This result suggests that the rate-limiting step of photocatalytic glycerol oxidation is independent of light intensity. Because the trajectory of the photocatalytic glycerol oxidation involves C–H and C–C bond scission, the KIE experiments were designed to evaluate which step is of kinetic significance. The conversion rates of glycerol and deuterated glycerol were determined at 200 mW cm−2. The KIE values for scission of the C–H bond at the middle and terminal carbons were measured to be 1.20 ± 0.02 and 1.19 ± 0.04, respectively (Table S4, entries 2 and 3). The secondary KIE values excluded the photocatalytic C–H bond scission as the rate-limiting step. Consequently, the C–C bond scission should be rate-limiting. Co