Single atom–cluster synergy in Ag catalysts enables chiral glyceric acid from biomass

Abstract

Accessing optically pure chemicals directly from biomass via chemical catalysis remains challenging because of the need for high selectivity across complex, multistep transformations. Here, we report a robust silver-based catalyst featuring synergistic single atoms and nanoclusters that function as complementary active sites in the anaerobic oxidative carbon-carbon cleavage of bio-derived feedstocks. This catalyst enables the production of optically pure glyceric acid from a broad range of biomass sources including raw biomass with unprecedented yield (59 to 96%) and enantiomeric excess (97 to >99%). Structural and mechanistic studies reveal that silver clusters, stabilized by surface oxygen vacancies, activate dioxygen through π-backbonding, while electron-deficient silver single atoms preferentially bind the chiral substrate via ionic interactions, increasing the energy barrier for mutarotation and preserving its stereochemical identity. Oxygen spillover between the two sites facilitates precise oxidative Cα–Cβ bond cleavage. Together, these features overcome the longstanding dual challenges of low productivity and poor enantioselectivity in chemocatalytic biomass upgrading.

Ag₁-Ag_n/CeO₂-250 catalyst with Ag single atoms and clusters jointly contribute to biomass valorization to d-glyceric acid.

INTRODUCTION

Not only does biomass represent the most abundant renewable source of organic carbon on Earth, but it also serves as the largest natural reservoir of chiral centers. Many naturally occurring sugars, polyols, and polysaccharides in biomass exist in optically pure forms; however, this valuable stereochemical richness remains largely underused (1–3). In current biorefinery schemes, biomass is primarily processed into fuels such as bioethanol or into achiral platform chemicals such as 5-hydroxymethylfurfural, levulinic acid, and γ-valerolactone (4–9). Despite decades of effort, chemocatalytic strategies to convert biomass into high-value chiral molecules remain scarce (10–14). Unlike traditional asymmetric synthesis, which requires laborious construction of chiral centers using costly catalysts and ligands, the use of biomass feedstock merely needs the chiral centers to be preserved and transferred to target products, offering a more direct approach. To date, the intrinsic chirality of biomass, which is a specific and one of the most valuable features of biofeedstocks, is largely wasted.

One promising pathway to access chiral chemicals from biomass is the oxidative transformation of polyols and monosaccharides into optically pure carboxylic acids. Among the targeted products, glyceric acid stands out as a valuable chiral building block with broad applications in pharmaceuticals, biodegradable polymers, and agrochemicals (15–19). However, achieving both high yield and high enantiomeric excess (ee) in these oxidative conversions remains challenging. This difficulty arises from the inherent complexity of biomass-derived substrates, which undergo competing cascade and side reactions such as mutarotation, leading to low enantio- and chemoselectivities (20–23). Developing efficient catalytic systems that can navigate these challenges while maintaining stereochemical integrity is therefore a key objective in the valorization of biomass into chiral fine chemicals. Stereoretention in biocascade conversions critically depends on suppressing planar intermediates such as enols and carbocations. For example, Ag nanoclusters supported on Al₂O₃ have been used to promote carbonyl oxidation, enabling the formation of chiral carboxylic acids from xylose with high enantiopurity (24). Notably, this catalyst has long been used in industry for the oxidative conversion of ethylene, where Ag nanoparticles or clusters serve as the primary sites for O₂ activation (25–27). However, while effective at initiating oxidation, the system is not optimized for sugar activation. It performs poorly with substrates bearing trans-configured hydroxyl groups, which is a common motif in native biomass such as glucose, arabinose, and fructose, (28, 29) and delivers limited activity (~30% yield). This narrow substrate scope suggests that additional functionality is needed to selectively transform complex polyols while preserving stereochemistry.

Designing an effective catalyst for sugar oxidation requires distinct active sites. It requires sites for O₂ activation to provide oxidative potential and sites for selective sugar/polyol coordination and transformation. Metal nanoclusters such as Ag are well suited for O₂ activation (30–34) but lack the site specificity needed for stereocontrolled interactions with hydroxyl-rich substrates. In contrast, atomically dispersed metal species can offer well-defined coordination environments (35–40) that favor selective C─C bond rearrangement and stereoretention. We therefore envisioned a dual-site catalyst integrating Ag nanoclusters and Ag single atoms spatially separated on an oxygen conductive support, where clusters drive oxidation, single atoms provide precise binding and activation of sugars, and the support transports oxygen. This single atom–cluster support synergy may overcome the activity and selectivity limitations in the multistep sugar/polyol conversion to chiral organic acids.

Following this strategy, we report an Ag-based catalyst that leverages the cooperative interplay between single atoms and nanoclusters to achieve high-efficiency, stereoretentive sugar oxidation. In this system, Ag clusters promote O₂ activation, while atomically dispersed Ag species provide well-defined sites for selective adsorption and transformation of hydroxyl-rich substrates. The catalyst is supported on ceria engineered with surface oxygen vacancies, which enhance the generation and migration of reactive oxygen species and enable strong interactions between electron-deficient Ag single atoms and metallic Ag clusters. These structural and electronic features together enable precise control over α carbon–β carbon (Cα–Cβ) bond cleavage and enable the production of optically pure glyceric acid with exceptional yield (up to 96%) and ee (>99%) from a broad range of biomass feedstocks, including raw lignocellulosic materials.

RESULTS

Structural characterization

Ag/CeO₂ catalyst was synthesized through an incipient wet impregnation method. The obtained precursor was subsequently calcined in N₂ atmosphere at 450°C, which enabled the decomposition of Ag salt to metallic Ag, and particularly promoted metallic Ag dispersion by oxygen vacancies on CeO₂ (the details were shown in fig. S1) (41–44). The coexistence of Ag single atoms and clusters was achieved by turning Ag loading, with optimal loading identified as ~2.5 wt % (table S1). For comparison, catalysts containing exclusively single atoms or nanoclusters were also synthesized, with Ag loadings of ~0.5 and ~5%, respectively. The three catalysts with distinct Ag speciation were denoted as Ag₁/CeO₂ (single atoms), Ag_n/CeO₂ (nanoclusters), and Ag₁-Ag_n/CeO₂ (coexistence of both species). In addition, H₂ pretreatment of the CeO₂ support before Ag loading was conducted to introduce additional oxygen vacancies (O_v) for improved redox capacity. On the basis of H₂-TPR (temperature-programmed reduction) results (fig. S2), three reduction temperatures (180°, 250°, and 350°C) were selected, and the reduced CeO₂ was denoted as CeO₂-T (T was the reduction temperature).

The dispersion state of Ag species on CeO₂ is assessed by transmission electron microscopy (Fig. 1, A to G). Aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) images of Ag₁-Ag_n/CeO₂ at different magnifications (Fig. 1, B and F) validate the coexistence of Ag single atoms (in red circle) and clusters (in green circle with an average size of 1.0 ± 0.2 nm about 10 to 14 Ag atoms). In contrast, the AC-HAADF-STEM images of Ag₁/CeO₂ (Fig. 1A and fig. S3) show numerous isolated bright spots, attributed to Ag single atoms uniformly anchored on the CeO₂ (111) planes, while that of Ag_n/CeO₂ (Fig. 1D) shows considerable Ag nanoparticles with larger average size (1.7 ± 0.2 nm) and little isolated Ag atoms. The AC-HAADF-STEM image of Ag₁-Ag_n/CeO₂-250 (Fig. 1, C and G) demonstrates that the coexistence of atomically dispersed Ag and Ag clusters (Ag_n, approximately 6 to 10 atoms) remains on H₂-reduced CeO₂, with a minor decline of the particle size distribution to 0.7 ± 0.2 nm. The intensity profile along the line in Fig. 1G (Fig. 1H) demonstrates the obvious difference of Ag sites, where the isolated Ag atoms show high intensity while that with low intensity is attributed to Ag clusters. The three-dimensional (3D) fitting maps (Fig. 1I) highlight again the presence of single Ag atoms and neighbored Ag clusters, and the energy-dispersion x-ray spectroscopy (EDS) mapping (Fig. 1J) confirms the well dispersion of Ag. The H₂ pretreatment of the CeO₂ support increases its O_v concentration (more detailed discussion can be found in the Supplementary Materials), which facilitates the stabilization and dispersion of Ag species (Fig. 1K; Ag exposed percentage peaks at a pretreatment temperature of 250°C).

Fig. 1. The characterization of Ag/CeO₂ catalysts.

AC-HAADF-STEM image of (A) Ag₁/CeO₂, (B and F) Ag₁-Ag_n/CeO₂, (C and G) Ag₁-Ag_n/CeO₂-250, and (D) Ag_n/CeO₂. (E) High-resolution transmission electron microscope image of Ag₁-Ag_n/CeO₂-250. (H) Brightness intensity profiles of lines marked in the AC-HAADF-STEM images of (G). a.u., arbitrary units. (I) The 3D fitting maps of Ag₁-Ag_n/CeO₂-250 at high magnification. (J) The EDS elemental mapping of Ag₁-Ag_n/CeO₂-250 in (E). Ag exposed percentage on different Ag catalysts (K).

X-ray absorption near-edge spectroscopy (XANES) and extended x-ray absorption fine structure (EXAFS) are carried out to assess the electronic structure and local chemical environment of Ag in Ag₁-Ag_n/CeO₂-250 catalyst. The absorption edge of Ag k-edge XANES spectra (Fig. 2A) is a little higher compared to Ag foil and slightly shifts toward Ag₂O. This demonstrates that the valence state of Ag species in Ag₁-Ag_n/CeO₂-250 is situated between 0 and +1, and the average chemical valence is estimated to be +0.33. The k³-weighted EXAFS (Fig. 2B) for Ag₁-Ag_n/CeO₂-250 display an obvious Ag─Ag path (2.86 Å) and a secondary Ag─O path. The coordination number (CN) of Ag─Ag is evaluated to be ~4.5, while two CNs of Ag─O give 1.2 and 1.7, respectively (table S2); that is, Ag in clusters is coordinated with around four Ag atoms, while single Ag atom is bridged with one or two oxygen atoms. The obviously higher R-space of Ag─O and lower CN than that of Ag₂O reference suggest the formation of isolated Ag atom, which probably bridges with O atom on CeO₂ surface via Ag─O─Ce. The lower CN of Ag─Ag than that (CN = 12.0) in the Ag foil reference verifies the formation of Ag clusters without obvious agglomeration. X-ray photoelectron spectroscopy (XPS) of Ce 3d (Ce 3d_5/2 and Ce 3d_3/2) on Ag₁-Ag_n/CeO₂-250 showed that Ag loading leads to obvious increase in Ce³⁺/Ce⁴⁺ ratio possibly due to the interaction between Ag and CeO₂ support resulting in electron transfer from Ag to CeO₂ (figs. S4 to S6 and table S3), while Ag 3d (Ag 3d_5/2 and Ag 3d_3/2) XPS spectra (Fig. 2C) confirm that Ag exists in both Ag⁺ and Ag⁰ forms. The wavelet transform EXAFS (WT-EXAFS) analysis (Fig. 2, D to F) for Ag k-edge has been further conducted to visualize the coordination configuration, which allows information to be displayed in both R-space and k-space. The strong WT signal focused at 9.7 Å⁻¹ is derived from Ag─Ag contribution. Besides, a WT signal with moderate intensity at 4.8 Å⁻¹ was attributed to Ag─O contribution.

Fig. 2. The structural identifications of Ag₁-Ag_n/CeO₂-250 catalysts.

(A) XANES. (B) EXAFS. (C) Ag 3d XPS. (D to F) WT-EXAFS contour plot of Ag foil, Ag₂O, and Ag₁-Ag_n/CeO₂-250. FT, Fourier transform.

Selective chiral glyceric acid production

The activity of the as-prepared catalysts was first evaluated using d-arabitol as a represented substrate derived from C₅ sugar because of the same stereostructure at C2 and C4, while NaOH was added as a promoter in O₂ atmosphere. The results (Fig. 3A) indicated that Ag₁-Ag_n/CeO₂ promoted the conversion of d-arabitol, selectively affording d-glyceric acid with a yield as high as 84.2%, which was notably higher than that achieved by either Ag₁/CeO₂ and Ag_n/CeO₂ (more information could be found in tables S4 to S14). The results also revealed that the production rate, normalized by the molar amount of Ag (in mole_product per mole_Ag per hour; Fig. 3B), from Ag₁-Ag_n/CeO₂ remained higher than that of either Ag₁/CeO₂ or Ag_n/CeO₂. In the absence of Ag catalysts, NaOH promoted d-arabitol conversion but caused complete racemization of glyceric acid (table S8), whereas Ag catalysts enabled d-glyceric acid production with excellent ee (>99%). We performed a comparative study using Au or Cu supported on CeO₂ (table S10), which afforded only 41.2 and 16.6% yields of d-glyceric acid, respectively. Even worse, Au supported on CeO₂ gave low ee (54.6%), while a racemic mixture was obtained in the case of Cu-anchored on CeO₂, highlighting the crucial role of Ag in achieving stereoselective transformation. When O_v content in the CeO₂ support was increased by H₂ reduction before Ag loading, d-glyceric acid yield initially increased with rising H₂ reduction temperature and achieved an unprecedented maximum (95.6% yield and >99% ee) over Ag₁-Ag_n/CeO₂-250 catalyst (Fig. 3A), offering the highest d-glyceric acid production rate (Fig. 3B). However, further increasing the reduction temperature to 350°C led to a decline in d-glyceric acid yield, which correlated with the reduced Ag dispersion observed on the catalyst surface.

Fig. 3. The catalytic activity of Ag/CeO₂ catalysts for chiral glyceric acid production.

(A) d-glyceric acid yield and ee from different Ag-based catalysts. (B) d-glyceric acid production rate from different Ag-based catalysts. (C) Substrate extension. Reaction conditions: (A and B) 0.1 g of d-arabitol, 0.1 g of catalyst, 373 K, two equivalent NaOH, 0.6 MPa of O₂, 30 min (the details are in the Supplementary Materials); (C) 0.04 g of monosaccharide, 0.04 g of Ag₁-Ag/CeO₂-250, 373 K, two equivalent Na₂CO₃, 0.6 MPa O₂, 60 min. Two-step reaction was used when derivatives from raw corncob and rice straw were used as feedstock and more details are illustrated in the Supplementary Materials.

We subsequently evaluated the catalytic performance across a series of substrates bearing diverse hydroxyl group configurations and carbon chain structures. The Ag₁-Ag_n/CeO₂-250 exhibited consistently high performance across a broad range of biomass-derived substrates, including polyols, aldoses, and ketoses, affording glyceric acid in impressive yields of 76.5 to 95.6% with ee exceeding 99% (Fig. 3C). These results represent the highest levels of yield and enantioselectivity reported so far for glyceric acid synthesis from biomass feedstocks (table S15). Carbon balance and total organic carbon analysis further support the high efficiency and minimal by-product formation under optimized conditions (table S16). d- and l-glyceric acid were obtained respectively from their corresponding enantiomeric precursors, hinting at the preservation and transfer of the biomass-derived chiral centers. Encouraged by the broad substrate scope, we further tested the system with unrefined raw biomass. Even derivatives from raw biomass (corncob and rice straw) as crude materials afforded ~60% yield of d-glyceric acid with ~97% ee (table S16), which, to the best of our knowledge, is the first example of obtaining an optically pure molecule through C─C bond cleavage from raw biomass via chemical catalysis. In contrast, when using Ag₁/CeO₂-250 and Ag_n/CeO₂-250 while keeping total Ag mass constant, the catalytic performance dropped. For various biomass-derived model substrates, Ag_n/CeO₂-250 afforded glyceric acid in yields ranging from 54.6 to 74.0%, while Ag₁/CeO₂-250 delivered only 44.6 to 68.1%. With raw biomass, the gap widened further, where Ag₁/CeO₂-250 and Ag_n/CeO₂-250 yielded merely 31.5 to 33.2% and 35.4 to 36.7%, respectively (tables S17 and S18). The durability test of catalyst showed that there was just slight decrement in d-glyceric acid yield after 5 cycles and the ee was always remained at >99% (tables S19 and S20).

Single atom–cluster synergy

Reaction pathway

To reveal the specific roles of single atoms and clusters, we first elucidated the reaction pathway by in situ nuclear magnetic resonance (NMR) spectroscopy using isotopic labeling ¹³C2–d-arabitol due to the highest d-glyceric acid yield (also labeled as ¹³C4–d-arabitol due to the symmetrical structure). As shown in Fig. 4 (A and B), the peak at δ = 70.4 parts per million (ppm), corresponding to C2(4) in d-arabitol, gradually diminished with prolonging reaction time. Concurrently, a new peak at δ = 71.6 ppm, assigned to C2 in glyceraldehyde appeared with its intensity increased first and then decreased, while a further peak at δ = 73.6 ppm, attributing to C2 in glyceric acid, progressively intensified. While it is well accepted that the terminal ─OH first transforms to aldehyde group during the process of polyol oxidation (45, 46), little ¹³C2- or ¹³C4-labeled d-arabinose was detected with the consumption of d-arabitol, possibly due to its high reactivity. Nonetheless, signals corresponding to glycolic and formic acids increased over time, likely resulting from further oxidation or decomposition of the C2 intermediate (glycolaldehyde) generated via Cα─Cβ cleavage.

Fig. 4. The reaction pathway on d-arabitol transformation to d-glyceric acid.

(A) The diagram of ¹³C flow in ¹³C2(4)–d-arabitol. (B) In situ ¹³C NMR spectra for ¹³C2(4)–d-arabitol conversion over Ag₁-Ag/CeO₂-250 catalyst (the measurement conditions can be found in the Supplementary Materials). (C) The diagram for the chirality transference from d-arabitol to d-glyceric acid. (D) The residual d-arabitol percentage and d-glyceric acid yield during d-arabitol conversion. (E) The corresponding specific rotation in (D) and simulated mixture. (F) The residual d-glyceraldehyde percentage and d-glyceric acid yield during d-glyceraldehyde conversion. (G) The corresponding specific rotation in (E) and simulated mixture. Reaction conditions: (B) 0.1 g of ¹³C2(4)–d-arabitol, 0.1 g of Ag₁-Ag_n/CeO₂-250, 373 K, two equivalent NaOH, 0.6 MPa of O₂; (D to G) 0.5 g of d-arabitol or d-glyceraldehyde, 0.1 g of Ag₁-Ag_n/CeO₂-250, 373 K, two equivalent NaOH, 0.6 MPa O₂.

To verify chirality transfer path, we monitored the time-dependent specific rotation ([α]) of the reaction solution (Fig. 4, C to G). The [α] value steadily increased with time and matched that of a standard mixture containing equivalent d-arabitol and d-glyceric acid. Similar experimental phenomena were observed when d-glyceraldehyde was used as reactant. This strongly validated that the chiral center at C2(4) in d-arabitol was successfully transferred to d-glyceric acid during C─C cleavage, ruling out configuration changes via epimerization reaction or other side reactions (47, 48). On the basis of these results, the reaction pathway on d-arabitol transformation to d-glyceric acid was proposed as follows: (i) the oxidation of terminal ─OH in d-arabitol to d-arabinose, (ii) Cα─Cβ cleavage in d-arabinose to produce equivalent d-glyceraldehyde and glycolaldehyde intermediates, and (iii) the subsequent d-glyceraldehyde oxidation to d-glyceric acid along with glycolaldehyde oxidation to glycolic acid. Following such a reaction pathway, the chiral center at C2(4) in d-arabitol was delivered to C2 of d-glyceraldehyde.

Role of Ag₁ and Ag_n

On the basis of the proposed reaction pathway, we further investigated the specific active sites involved in the multistep transformation of d-arabitol. Molecular dynamics (MD) studies were first used to probe the competitive adsorption of d-arabitol and O₂ on Ag₁ or Ag_n. Catalyst model for Ag₁-Ag_n/CeO₂ in calculation was established on the basis of AC-HAADF-STEM and EXAFS results, and the optimized structure was well consistent with experimental results (Fig. 5, A and B). We first conducted Bader charge analysis for Ag in optimized model, and the result confirmed that Ag combining with O of CeO₂ induced electron transfer from Ag to O, where Ag₁ provided more electron shift than Ag_n (Fig. 5C). MD results (Fig. 5, D and E) verified that O₂ is more inclined to be adsorbed on Ag_n cluster than Ag₁ in terms of Ag─O interaction, while d-arabitol was preferentially adsorbed on Ag₁ by combining with O at terminal ─OH also via an Ag─O interaction. The interaction energy (table S21) of Ag_n-O₂, Ag_n–d-arabitol, Ag₁-O₂, and Ag₁–d-arabitol were calculated to be −42.3, −18.9, −24.6, and − 37.7 kcal/mol, confirming the preferential adsorption of O₂ on Ag_n and d-arabitol on Ag₁. Bonding analyses revealed that O₂ preferentially adsorbs on Ag_n due to stronger π backbonding, while d-arabitol favors Ag₁ owing to their higher electrostatic (ionic) interaction (detailed discussion can be found in the Supplementary Materials, including tables S22 to S24).

Fig. 5. The investigation on the role of Ag₁, Ag_n, and CeO₂ on d-arabitol conversion to d-glyceric acid.

(A) The calculation model of Ag₁-Ag_n/CeO₂-250. (B) Differential charge density analysis for Ag₁-Ag_n/CeO₂-250. (C) Typical Bader charge of Ag atom on model in (A). (D) MD investigation on the competitive adsorption of d-arabitol and O₂ on Ag₁ or Ag_n. (E) Radial distribution function analysis. (F to H) The CO-DRIFT spectra on Ag₁-Ag_n/CeO₂-250. (I) The Raman spectra on Ag₁-Ag_n/CeO₂-250. (J) AIMD (ab initio MD) investigation on oxygen spillover.

Diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy analysis (Fig. 5, F to H) was further conducted to elucidate the interactions between carbonyl intermediates (d-arabinose and glyceraldehyde) and the dual active sites with CO as a probe molecule. The spectroscopy consisted of three absorbance bands with peaks at around 2050, 2005, and 1930 cm⁻¹. The signal at 2050 cm⁻¹ was assigned to the linear-adsorption of CO on Ag₁ (Ag₁-CO), while the peaks at 2005 and 1930 cm⁻¹ were attributed to the linear adsorption (Ag_n-CO) and bridge adsorption [Ag_n-(CO)_m] of CO on Ag_n, respectively (49, 50). When CO was changed to a mixture of CO and O₂, the fraction of Ag₁-CO obviously enhanced, while Ag_n-CO content sharply decreased. The ratio of peak area (Ag₁-CO versus Ag_n-CO) increased gradually with O₂ partial pressure increasing. For instance, the ratio of Ag₁-CO to Ag_n-CO increased from 1.23 to 1.49 when the percentage of O₂ was raised to 10%. Upon O₂ partial pressure increasing to 40%, the ratio continuously increased to 2.26. This indicates a preferential interaction between carbonyl intermediates and Ag₁ under O₂-rich conditions. The absence of either Ag₁ or Ag_n sites might impair the activation of substrate/intermediate or O₂, respectively, leading to a marked decline in glyceric acid selectivity. Isotope tracing experiment using ¹³C2(4)–d-arabitol as substrate evidenced the generation of more ¹³C-glyoxylic acid and ¹³C-oxalic acid resulting from the deep oxidation using Ag_n/CeO₂ (fig. S7). In the case of Ag₁/CeO₂, far less Ag_n to activate O₂ caused the generation of more lactic acid derived from the dehydration and isomerization of glyceraldehyde intermediate, in addition to the formation of dihydroxyacetone as the incomplete oxidation product (table S25).

Role of CeO₂ and O_v

XPS (figs. S8 to S10) and Raman (Fig. 5I) spectroscopy revealed abundant O_v on the Ag₁-Ag_n/CeO₂-250, where the surface vacancies (O_v-surf at 945 cm⁻¹) increasing with H₂ reduction temperature and correlating positively with catalytic activity, suggesting that O_v-surf rather than bulk vacancies (O_v-bulk at 535 cm⁻¹) likely plays a key role in promoting d-glyceric acid formation. AIMD (ab initio MD) simulation (Fig. 5J) illustrated that the existence of O_v on catalyst surface induced O₂ (adsorbed by Ag_n) to be located between Ag_n and O_v. It was observed that with initially extending simulation time, one of O atom in O₂ approached O_v gradually, while the other moved to Ag_n. At 800 fs, the distance between O atom in O₂ and Ce (adjacent to O_v) decreased from ~4.2 to 2.9 Å, while that between the other O atom and Ag_n slightly reduced from ~2.5 to 2.2 Å. On the contrary, O─O distance in O₂ sharply increased from ~1.1 to 2.8 Å. As a consequence, O₂ was dissociated by the cooperation of Ag_n and O_v, which was in agreement with the experimental results. Thereafter, the dissociated O atom (binding to O_v) moved toward Ag₁ at 800 to 1800 fs. In that case, the distance of O─Ag₁ decreased from ~3.9 to 2.4 Å, while the distance from Ce increased gradually and lastly stayed at ~6.2 Å. These results gave an indication of oxygen spillover, which allowed the oxidation of d-arabitol (adsorbed on Ag₁). O₂-TPD (temperature programmed desorption) results (fig. S11) showed that the amount of physically and chemically adsorbed O₂ was higher than exposed Ag on catalyst surface, further supporting the occurrence of oxygen spillover.

Density functional theory study

Density functional theory (DFT) study was subsequently conducted to comprehensively elucidate the cooperative contributions of Ag₁, Ag_n, and O_v (mainly O_v-surf) in d-arabitol transformation to d-glyceric acid (Fig. 6A). Figure 6B illustrated that the energy barrier for O₂ activation to produce two O* by the cooperation of Ag_n and O_v (9.6 kcal/mol) was much lower than that by sole Ag_n (18.5 kcal/mol) or O_v alone (26.9 kcal/mol). This confirmed the predominant role of Ag_n-O_v in O₂ activation, in well agreement with Raman and MD results. With respect to d-arabitol activation, upon the adsorption of d-arabitol on Ag₁, the H of terminal ─OH in d-arabitol was removed by NaOH (Fig. 6C and fig. S12). Thereafter, d-arabinose was formed after C1-H removal by O* with energy barrier (17.1 kcal/mol). We also comparatively investigated the case that d-arabitol was adsorbed on Ag_n, but higher energy barrier (20.3 kcal/mol) was obtained. Similarly, d-arabitol adsorption by O_v alone gave as high as energy barrier (28.7 kcal/mol). These results suggested d-arabitol adsorption on Ag₁ before Ag_n and O_v, in agreement with CO-DRIFT and MD results. Notably, d-arabitol adsorbed by the cooperation of Ag₁ and O_v (fig. S13) also showed a comparable energy barrier (16.6 kcal/mol) than that adsorbed on Ag₁ alone (17.1 kcal/mol).

Fig. 6. DFT study.

(A) Schematic illustration for O₂ activation and d-arabitol transformation. (B) Energy profile for O₂ activation to O*. (C) Energy profile for d-arabitol transformation to d-glyceric acid (glycerate). (D) Reaction mechanism on C2(4) mutarotation. TS, transition state. IM, intermediate.

The subsequent Cα─Cβ bond cleavage may proceed via two possible pathways: a retro-aldol reaction to form glyceraldehyde or an oxidative C─C cleavage. The results indicate that the oxidative C─C cleavage has a lower energy barrier of 17.0 kcal/mol, compared with direct C─C cleavage on Ag₁ (23.3 kcal/mol) or the conventional base-catalyzed retro-aldol pathway by NaOH (42.1 kcal/mol; fig. S14). This suggests that C─C bond cleavage and aldehyde oxidation likely proceed in a synergistic manner during d-glyceric acid formation. This avoids the isomerization of hydroxyl aldehyde intermediates, preserving chirality during the reaction. Last, the d-glyceric acid formation via H removal at the terminal C only requires a low energy barrier of 9.2 kcal/mol. Glyceraldehyde detected in in situ ¹H NMR suggests that the retro-aldol pathway was not completely suppressed. Nevertheless, the interaction between glyceraldehyde and Ag₁ increases the energy barrier for base-catalyzed C2(4)-R mutarotation to C2(4)-S from 14.5 to 38.0 kcal/mol, thereby ensuring the retention of chirality from glyceraldehyde to glyceric acid. The hybridization of sutured C was sp^2.3 without catalyst, while it changed to sp^2.7 when interacting with Ag₁ (fig. S15). With the further participation of O* on Cα─Cβ cleavage, Ag₁ site made hybridization continuously change to sp³, thereby stabilizing C₃ intermediate (Fig. 6D).

DISCUSSION

To sum up, we fabricated a robust Ag₁-Ag_n/CeO₂ catalyst with the coexistence of Ag single atoms and clusters, which exhibited high efficiency for the valorization of biomass to access chiral glyceric acid with excellent optical purity. d-glyceric acid yield up to 95.6% with >99% ee was achieved using d-arabitol as feedstock, which has not been reported before. Experimental results combined with multiscale theoretical calculations unveiled the pronounced synergistic effect between Ag single atoms and clusters, in addition to O_v on catalyst surface, which contributed to O₂ activation and effective d-arabitol transformation in tandem. This allowed the precise C─C cleavage and impeded the mutarotation at C2(4) in d-arabitol, thus enabling the reservation and successful deliver of chirality in feedstock to d-glyceric acid. This strategy maximums atom utilization efficiency of precious Ag and optimizes the catalytic activity, which may open a previously unexplored window to expand atomic size heterogeneous catalyst in biomass upgrading for synthesizing optically pure chiral chemicals. Compared with traditional asymmetric catalysis, which requires the use of chiral ligands or chiral environments to induce stereoselectivity, this approach presents a fundamentally different and potentially more sustainable pathway to access enantiomerically pure products.

MATERIALS AND METHODS

Reagents

AgNO₃ [analytical reagent (AR)], Cu(NO₃)₂ (AR), HAuCl₄ (AR), NaOH (AR), and polyvinylpyrrolidone K30 (PVP; AR) were purchased from KESHI (Chengdu, China). CeO₂ (no. XFI35) was obtained from XFNANO (Jiangsu, China). Glycolic acid (99%), d-glyceric acid (99%), d-glyceraldehyde (95%), glycolaldehyde (40 wt % in solution), dihydroxyacetone, and tartaric acid (98%) were purchased from Adamas-beta (Shanghai, China). Formic acid (99%) and oxalic acid (99%) were purchased from Sigma-Aldrich (Shanghai, China). Racemic glyceric acid (20 wt % in solution) and lactic acid (20 wt % in solution) were purchased from TCI (Shanghai, China). d-/l-arabitol, d-/l-xylose (99%), d-/l-arabinose (99%), d-/l-sorbose (99%), and d-fructose (99%) were obtained from J&K Scientific (Beijing, China). ¹³C2(4)–d-arabitol (99.9%) was purchased from SHANGHAI ZZBIO CO. LTD. (Shanghai, China). Rice straw was obtained from Fushun in Liaoning Province of China, which was smashed by miniature plant sample mill (1400 rpm, 180 W, 2 g/min), and 60 to 100 mesh of rice straw powder was selected as feedstock. Corncob powder (60 to 100 meshes) was provided by Shandong Futaste Investment Co. Ltd. (Shandong Province, China). All the chemicals were used without further purification. Ultrapure water of 18.25 megohms/cm (298 K) was used.

Catalyst preparation

Ag/CeO₂ catalyst was prepared via a deposition-precipitation method. A desired amount of AgNO₃ was first dissolved in 4 ml of deionized water and then ultrasonicated for 10 min. A total of 1000 mg of CeO₂ and 100 mg of PVP were added into 40 ml of deionized water at room temperature with a magnetic stirring to obtain a suspension. AgNO₃ aqueous solution was then added dropwise into the suspension. Thereafter, the mixture was kept at stirring for 1 hour at room temperature. The resultant suspension was ultrasonicated in ice-water bath for 1 hour. Subsequently, 4 ml of Na₂CO₃ (twofold moles of AgNO₃) aqueous solution was added dropwise into the suspension under magnetic stirring and kept for 3 hours. The suspension was then stood and aged for 2 hours at room temperature. Last, the precipitate was filtered and washed with deionized water and ethanol. The resultant precipitate was dried at 80°C under magnetic calcined in N₂ at 450°C for 2 hours with a heating rate of 5°C/min. Other Ag catalysts with different supports or CeO₂-supported other metal catalysts (Au and Cu) were prepared by the similar method.

Ag₁-Ag_n/CeO₂-T catalyst was prepared by the following methods. CeO₂ was reduced at 10% H₂/Ar (50 ml/min) for 1 hour at different temperature, which was recorded as CeO₂-T (T is the reduced temperature for CeO₂). In addition, 1000 mg of CeO₂-T (usually CeO₂-250) and 100 mg of PVP were added into 40 ml of deionized water at room temperature with a magnetic stirring to obtain a suspension. AgNO₃ aqueous solution (0.039 g of AgNO₃ dissolved in 4 ml of deionized water and then ultrasonicated for 10 min) was then added dropwise into the above suspension. Thereafter, the mixture was kept at stirring for 1 hour at room temperature. The resultant suspension was ultrasonicated in ice-water bath for 1 hour. Subsequently, 4 ml of Na₂CO₃ (twofold moles of AgNO₃) aqueous solution was added dropwise into the suspension under magnetic stirring and kept for 3 hours. The suspension was then stood and aged for 2 hours at room temperature. Last, the precipitate was filtered and washed with deionized water and ethanol. The precipitate obtained was dried at 80°C for 8 hours and then calcined in N₂ at 450°C for 2 hours with a heating rate of 5°C/min.

Activity test

The conversion of d-arabitol or other saccharides was conducted in a 100-ml Parr microreactor equipped with a mechanical stirring. Upon mixing substrates and catalysts in aqueous solution with initial O₂ pressure of 0.6 MPa, the Parr reactor was heated from room temperature to the target temperature (100°C) via programmed temperature control within 15 min. Reaction timing commenced (t = 0) only after reaching the target temperature, followed by isothermal maintenance for 30 or 45 min to ensure complete substrate conversion. The duration of this isothermal period constitutes the reported reaction time. When the reaction was finished, the reactor was cooled to room temperature in air. After separating the solid by filtration, the liquid mixture was analyzed.

Two-step method was used for the conversion of corncob or rice straw: 0.50 g of raw material was pretreated in 50 ml of aqueous solution containing 0.60 g of maleic acid at 140°C for 3 hours. After reaction, the pH of solution was adjusted to about 9.5 by NaOH to ensure that the residual maleic acid was transformed to maleate. Then, maleate was separated by extraction in tetrahydrofuran (THF)/NaCl/water two-phase system (detailed procedure could be found in the Supplementary Materials). In that case, sugars were enriched in THF phase. THF was then removed by distillation to obtain sugar solution, which was used as the feedstock for the next reaction over Ag₁-Ag_n/CeO₂-250 catalyst. As for corncob/rice straw as raw material, almost all hemicellulose was dissolved, and the dissolution of cellulose was less than 5%.

Product analysis

To quantify the products in the liquid mixture, the reaction solution was analyzed by high-performance liquid chromatography (HPLC) (Waters, e2695) equipped with an Aminex column (HPX-87 column, Bio-Rad). The temperatures of detector and column were set as 35° and 50°C, respectively. H₂SO₄ aqueous solution (5 mM) was used as the mobile phase with a flow rate of 0.6 ml/min. Before analysis by HPLC, the sample was acidized by H₂SO₄ aqueous solution (5 mM). The procedure involved the dilution of the sample by fourfold H₂SO₄ (5 mM) aqueous solution. At that time, the pH of sample was 2.0, where ~97% glycerate has been transformed to glyceric acid. The amount of liquid products was obtained by comparing with the standard calibration curves of each sample obtained from commercial sources.

To detect the optical purity of glyceric acid produced, the reaction solution was further analyzed by HPLC (Agilent Technologies 1200 Series) equipped with a MCI GEL packed column (MCI GEL CRS10W column). The temperatures of detector and column were both set at 40°C. CuSO₄ aqueous solution (1 mM) was used as the mobile phase with a flow rate of 0.6 ml/min. The samples were first acidized by H₂SO₄ aqueous solution (5 mM) and then were extracted by THF in which most of the liquid products could be extracted to organic phase. After separating the organic phase, the THF in organic phase was removed by distillation, and the remained liquid products were analyzed by HPLC. The amount of glyceric acid enantiomer was calculated by comparing to the standard calibration curves of optically pure d- or l-glyceric acid from commercial source.

The conversion (Con. %) of polyols and monosaccharides was calculated as the following equation

$$\begin{array}{c}\begin{array}{c}\text{Con}.(\%)\hfill \\ =\frac{\text{Moles of initial feedstock}-\text{moles of remained feedstock}}{\text{Moles of initial feedstock}}\hfill \\ \times 100\%\hfill \end{array}\hfill \end{array}$$ (1)

The conversion (Con. %) of raw biomass was calculated as the following equation

$$\begin{array}{c}\text{Con}.(\%)\hfill \\ =\frac{\text{Mass of initial hemicellulose}-\text{mass of remained hemicellulose in feedstock})}{\text{Mass of initial hemicellulose in feedstock}}\hfill \\ \times 100\%\hfill \end{array}$$ (2)

The yield of d-glyceric acid from polyols or monosaccharides was calculated on the basis of the assumption that 1 M polyol or monosaccharide can afford only 1 M d-glyceric acid as the following equation

$$\begin{array}{c}\text{Yield} (\%)=\frac{\text{Moles of} \mathrm{ᴅ}\text{-glyceric acid}}{\text{Moles of initial feedstock}}\times 100\%\hfill \end{array}$$ (3)

The yield of d-glyceric acid from raw biomass (corncob and rice straw) was calculated on the basis of the moles of xylose in the hemicellulose of raw biomass, where it was assumed that all the hemicellulose consisted of xylose with an average molecular weight as 132 g/mol per unit. The dissolution of cellulose was less than 5%

$$\begin{array}{c}\text{Yield} (\%)=\frac{\text{Moles of} \mathrm{ᴅ}‐\text{glyceric acid}}{\text{Mass of hemicellulose in feedstock}/132}\times 100\%\hfill \end{array}$$ (4)

The ee of d-/l-glyceric acid was calculated by the following equations

$$\begin{array}{c}\text{ee} (\%) \text{to} \mathrm{ᴅ}‐\text{glyceric acid}\hfill \\ =\frac{\text{Moles of} \mathrm{ᴅ}‐\text{glyceric acid}-\text{moles of} \mathrm{ʟ}‐\text{glyceric acid} }{\text{Moles of} \mathrm{ᴅ}‐\text{glyceric acid}+\text{moles of} \mathrm{ʟ}‐\text{glyceric acid} }\times 100\%\hfill \end{array}$$ (5)

$$\begin{array}{c}\text{ee} (\%) \text{to} \mathrm{ʟ}‐\text{glyceric acid}\hfill \\ =\frac{\text{Moles of} \mathrm{ʟ}‐\text{glyceric acid}-\text{moles of} \mathrm{ᴅ}‐\text{glyceric acid} }{\text{Moles of} \mathrm{ʟ}‐\text{glyceric acid}+\text{moles of} \mathrm{ᴅ}‐\text{glyceric acid} }\times 100\%\hfill \end{array}$$ (6)

d-glyceric acid production rate is the number of d-glyceric acid per unit catalyst and time at the end of reaction (~100% of d-arabitol conversion in our work), which was calculated by the following equation

$$\begin{array}{c}\text{Production rate} (\text{molproduct}/\text{molAg} \text{per} \text{hour})\hfill \\ =\frac{\text{Moles of} \mathrm{ᴅ}‐\text{glyceric acid} }{\text{Moles of} \text{Ag} \text{on} \text{catalyst} \times \text{reaction time} \left(\text{hour}\right)}\hfill \end{array}$$ (7)

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S18

Tables S1 to S30

References