Solar Reforming of Biomass with Homogeneous Carbon Dots

Abstract A sunlight‐powered process is reported that employs carbon dots (CDs) as light absorbers for the conversion of lignocellulose into sustainable H2 fuel and organics. This photocatalytic system operates in pure and untreated sea water at benign pH (2–8) and ambient temperature and pressure. The CDs can be produced in a scalable synthesis directly from biomass itself and their solubility allows for good interactions with the insoluble biomass substrates. They also display excellent photophysical properties with a high fraction of long‐lived charge carriers and the availability of a reductive and an oxidative quenching pathway. The presented CD‐based biomass photoconversion system opens new avenues for sustainable, practical, and renewable fuel production through biomass valorization.

Abstract: A sunlight-powered process is reported that employs carbon dots (CDs) as light absorbers for the conversion of lignocellulose into sustainable H 2 fuel and organics. This photocatalytic system operates in pure and untreated sea water at benign pH (2-8) and ambient temperature and pressure. The CDs can be produced in a scalable synthesis directly from biomass itself and their solubility allows for good interactions with the insoluble biomass substrates. They also display excellent photophysical properties with a high fraction of long-lived charge carriers and the availability of a reductive and an oxidative quenching pathway. The presented CD-based biomass photoconversion system opens new avenues for sustainable, practical, and renewable fuel production through biomass valorization.
Photocatalysis allows for the utilization of solar energy to produce renewable H 2 , but most reported systems still require precious-metal components, purified water or an expensive sacrificial electron donor (ED). [1] Photoreforming (PR) can use sunlight to convert biomass waste into H 2 and organic chemicals. [2] Instead of oxidizing water as in classical artificial photosynthesis, [3] PR employs preferentially abundant and inedible lignocellulose as an ED to quench holes (h + ) in a photoexcited photocatalyst, leaving behind low-potential electrons to drive proton reduction. [4] PR commonly relies on UV-absorbing TiO 2 colloids with noble metal cocatalysts (Pt, RuO 2 ), [5] and toxic CdS in organic solvents (CH 3 CN) [6] or alkaline conditions (pH > 14). [7] Carbon nitride (CN x ) has been shown for visible-light driven PR of biomass under benign aqueous pH, [8] but the heterogeneous nature of CN x restricts effective substrate/ photocatalyst interactions to occur. [2b, 6] Previous PR systems have also shown conversion yields 22 % (under strongly alkaline conditions) and required purified water, [5][6][7][8] which limit their utility, sustainability and economics.
Here, we introduce homogeneous carbon dots (CDs, Figure 1) produced from controlled, scalable calcination of cellulose (a-cel-CDs at 320 8C, Figure S1), [9] or commercial precursors such as citric acid (resulting in amorphous CDs, a-CDs at 180 8C, and graphitic CDs, g-CDs at 320 8C), [10] and aspartic acid (resulting in graphitic N-doped CDs at 320 8C, g-N-CDs; see SI) [10b, 11] for biomass PR. The non-toxic, biocompatible CDs are employed as light absorbers, together with a Ni bis(diphosphine) H 2 evolution cocatalyst (NiP, [12] Figure S2), to produce H 2 and organics in purified and untreated water under benign conditions (Figure 1 b). Transient absorption (TA) spectroscopy provides insight into the electron transfer dynamics of the PR systems.
The a-cel-CD/NiP system provides a benchmark activity of 13 450 mmol H 2 (g CDs ) À1 h À1 ( Figure S3, Table S5). [10,11,14] The a-cel-CDs display a maximum internal quantum efficiency (IQE at l = 360 nm, I = 4.05 mW cm À2 ) of 11.4 %, which compares favorably with g-N-CDs (5.3 %) and a-CDs (1.4 %). [10b] Future improvements in the development of the CDs should focus on high IQEs in the visible region. The photo-stability of the CD/NiP systems is currently limited by the fragile ligand framework of NiP, which degrades after a few hours of operation either due to formation of radicals from EDTA oxidation or ligand displacement from the Ni center. [13] 4-Methylbenzyl alcohol (30 mmol) instead of EDTA produced 3.7 AE 0.2 mmol H 2 after 6 h irradiation with a-cel-CD/NiP ( Figure S4, Table S6).
We then studied various insoluble biomass (a-cellulose, xylan and lignin; Figure 1 a) and soluble biomass model substrates and alcohols of industrial relevance (ethanol, glycerol; Figure S5). PR in aqueous phosphate solution (KP i ; pH 6 and 25 8C) with the CDs showed activity under benign conditions (Figures 2 b, S6, Tables S6-S9), with the acel-CDs showing again the best activity (Figures 2 b).
The highest H 2 yields after 24 h were observed with galactose (8.8 AE 0.2 mmol) and glycerol (8.5 AE 0.1 mmol), which correspond to turnover numbers of NiP (TON NiP ) of 177 AE 4 and 170 AE 2, respectively. Control experiments without ED, CDs or NiP showed negligible or no H 2 evolution ( Figure S7 and Table S7). The lowest H 2 yields were observed for lignin (0.03 mmol) due to its strong light absorption and robust cross-linked polyphenolic structure. [15] However, a much higher H 2 yield (7.8 AE 0.5 mmol, Table S6) was observed at lower lignin quantities (0.5 mg) due to improved light penetration through the CD solution (Figure 2 b, empty bar). PR of a-cellulose and xylan produced 5.0 AE 0.2 and 3.6 AE 0.3 mmol H 2 , respectively, similar to a heterogeneous CN x /NiP system. [8a] However, in contrast to heterogeneous systems that show substrate-dependent H 2 yields, homogeneous CDs photoreform soluble and insoluble biomass with a similar efficiency.
PR of a-cellulose with the a-cel-CD/NiP system was subsequently studied in KP i (pH 4.5, 6 and 8), H 2 SO 4 (pH 2) and 10 m KOH (% pH 15) ( Figure S8). The highest H 2 yields after 24 h were observed at pH 6 (5.0 AE 0.2 mmol H 2 ) and pH 8 (3.6 AE 0.2 mmol H 2 ). The efficiency was decreased approximately four times (1.2 AE 0.1 mmol H 2 ) in strong acid (pH 2), and PR did not proceed under extremely basic conditions (10 m KOH) due to the chemical instability of NiP ( Figure S8 and Table S10). [13] The biomass conversion yield (CY, %) was determined in KP i pH 6 with a-cel-CD/NiP at various a-cellulose loadings (0.8-1.65 mg, Figure S9, Table S11). A CY of 13.4 % was achieved at 0.8 mg a-cellulose (12 hrs), whereas re-additions of NiP (50 nmol) to repair the PR system in situ allowed a CY of 34.1 % (48 h, Figure S9). [13] This is higher than CYs reported for CdS/CdO x (9.7 %) [7] and CN x /Pt (22 %) [8a] under strongly alkaline conditions.
products. PR of galactose/glucose resulted in C 6 H 12 O 6 and C 6 H 10 O 5 isomers.
PR of a-cel-CDs (2.2 mg) with biomass substrates (100 mg) was then studied in untreated sea water (adjusted pH 6; Figures S18, S19, Tables S12-S14). The H 2 yields are comparable to purified water as reaction medium, suggesting that impurities/background organics do not hinder photocatalysis as observed for TiO 2 -based systems, but may rather act as EDs. [16] The highest H 2 yields were again achieved with galactose (8.4 AE 0.1 mmol, 24 h). The g-N-CDs showed 2-7 times lower H 2 yields in sea water compared to purified water ( 2.3 AE 0.1 mmol, 24 h), presumably due to surface N-doping that may provide adsorption sites for contaminants from the impurity-rich water. [16a] a-CDs in sea water show low H 2 yields ( 0.3 mmol), comparable to purified water. Thus, undoped CDs maintain good photocatalytic performances under realworld conditions. [16a] TA spectroscopy was employed to study the photophysics and charge transfer properties of a-cel-CDs, on fs-ns (fs-TA) and ms-s (ms-TA) timescales. fs-TA spectra (355 nm excitation, under Ar) resulted in a broad absorption feature in the visible region ( Figure S20), which decays % 2 fold faster upon adding EDTA, with the decay halftime changing from % 20 to 40 ps (Figure 3 a). This indicates that the absorption contains a partial contribution from photoinduced h + that are scavenged by EDTA ( % 0.1 ns), [8c, 16c] most likely by pre-adsorbed ED species.
On ms-s timescales, a blue-shifted, long-lived signal is observed in the absence of EDTA ( Figure S21), which is effectively quenched by O 2 and thus originates primarily from electrons. These are long-lived, trapped charge carriers with residual signals ( % 100 ms) even without EDTA, similar to previous reports for C 3 N 4 , [17] and metal oxide photocatalysts. [18] Addition of NiP as electron scavenger for a-cel-CDs resulted in (i) quenching of the electron signal ( % 0.5 ms) and (ii) appearance of a negative signal, assigned to the groundstate bleach of NiP due to its reduction by CDs, at 500 nm (Figures 3 b, S22). [10b, 12, 19] This suggests the direct electron transfer from CDs* to NiP, even without EDTA, therefore demonstrating an oxidative quenching mechanism. Titration of CDs with NiP ( Figure S23) revealed a linear relationship between the electron decay rates (at 500 nm) and NiP concentration, and an oxidative quenching rate of 1.09 AE 0.04 10 8 m À1 s À1 . This mechanism will have a low overall yield, as without EDTA most electrons recombine on faster timescales (! 100 ms), consistent with negligible H 2 production (Table S7). Nevertheless, the ability of long-lived trapped electrons to reduce NiP indicates that they retain reactivity, with trap energies above the NiP reduction potential.
Consistent with the fast hole scavenging process ( % 0.1 ns), addition of EDTA resulted in prolonged electron signals at 700 nm (Figure 3 c), indicative of reductive quenching. Signals at 500 nm were not prolonged with EDTA, suggesting multiple electronic states in a-cel-CDs. [20] Nevertheless, these results show both oxidative and reductive quenching for a-cel-CDs, which is different from that observed for g-N-CDs and a-CDs under similar conditions. In the latter cases, NiP À can only be formed with EDTA, [10b] most likely due to differences in energy of the trapped charges between these samples. For a-cel-CDs, the appearance of the NiP À signal at 500 nm at long times (Figures S22e) is indicative of reasonably efficient photoinduced NiP reduction ( Figure 4).
Previous studies on g-N-CDs showed a bimolecular recombination lifetime of t 50 % = 9 ps, with a residual 6 % of long-lived carriers (5 ns) to drive H 2 production. [10b] Herein, using similar excitation fluence/buffer conditions, the a-cel-CD bimolecular recombination lifetime is t 50 % = 45 AE 5 ps (i.e., 5 times slower), with the proportion of long-lived Figure 3. Normalized a) ( % 1 ps) fs-TA kinetics between 500 and 520 nm, b) ( % 50 ms) ms-TA kinetics (electrons) at 500 nm, c) ( % 50 ms) ms-TA kinetics (electrons) at 700 nm of a-cel-CDs with EDTA and/or NiP. d) Normalized ( % 50 ms) ms-TA kinetics (electrons) of a-cel-CDs at 500 nm with NiP and various biomass EDs (0.1 m). Inset shows the bleach region of DA which corresponds to NiP À . Conditions: KP i (pH 6.6) with NiP (50 nmol) upon excitation at 355 nm with an energy of 1 mJ cm À2 . (> 5 ns) carriers being about 15-20 % (Figure 3 a). We can thus propose two reasons for improved photocatalysis with acel-CDs: (i) existence of both oxidative and reductive quenching mechanisms and (ii) a-cel-CDs show slower bimolecular recombination processes and higher yields of long-lived carriers, which enable higher H 2 yields both under model (Figure 2 a) and real-world conditions ( Figure S18).
Finally, ms-TA spectra of a-cel-CDs with biomass were collected to analyze their capacity to quench the photogenerated h + . Biomass addition induced a similar oxidative quenching mechanism as with EDTA (Figures S24), but with a 50 % lower yield of NiP À (Figure 3 d). The slower h + extraction is assigned to the less accessible biomass compared to EDTA, which results in increased recombination and thus fewer long-lived electrons that can be extracted by NiP. This agrees with photocatalysis, where twice the H 2 yield was observed with EDTA compared to biomass (Figure 2). It is also possible that long-lived, trapped h + accumulate in CDs with biomass as ED due to the oxidative quenching pathway by NiP (Figure 4, white panel), facilitating oxidation of the challenging lignocellulosic substrates.
In summary, we report the development of a homogeneous PR system using CDs as light absorbers, which use the nexus of natural resources for coupled sustainable fuel production with biomass utilization and chemical synthesis. CDs prepared from biomass have well-suited photophysical characteristics such as the availability of an oxidative quenching pathway to convert challenging substrates and a high fraction of long-lived charge carriers. The cellulose-derived CDs allow for solar-driven fuel synthesis from lignocellulosic biomass under benign conditions with the prospect to simultaneously produce valuable chemicals in solution. The PR systems operate with a noble-metal-free cocatalyst and maintain their photocatalytic activity even in untreated sea water, which creates promising perspectives for the development of energy self-sufficient and low-carbon economies.