Sep. Purif. Technol.: Mediator-Free NADH Regeneration and Photoenzymatic CO₂ Reduction

First authors: Yihao Lin and Mei An
Corresponding author: Assoc. Prof. Chao Peng
DOI: 10.1016/j.seppur.2026.137403

Research highlights:

1. Two COF/MXene photocatalyst families, TDC-Nb_a and TDC-Ti_b, were constructed through aldehyde–amine condensation, enabling efficient 1,4-NADH regeneration without a soluble electron mediator. Optimized TDC-Ti_0.12 achieved a 1,4-NADH turnover frequency of 0.71 h⁻¹, exceeding TDC-Nb_0.18 (0.63 h⁻¹), pristine TDC (0.15 h⁻¹) and mediator-containing TDC (0.49 h⁻¹).
2. The TDC/Ti₃C₂T_x interface combined a larger Fermi-level difference, lower exciton-binding energy and a stronger preference for terminal NAD⁺ adsorption. The 2.17 eV adsorption-energy difference favored formation of 1,4-NADH.
3. Coupling the photocatalyst with formate dehydrogenase (FDH) created a photoenzymatic CO₂-reduction system. TDC-Ti_0.12 generated formate at 1485.1 μmol g⁻¹ h⁻¹ without [Cp*Rh], 2.3 times the rate of pristine TDC and higher than mediator-containing TDC.

Introduction

In February 2026, Separation and Purification Technology published work from Assoc. Prof. Chao Peng’s group at Wuyi University on mediator-free photocatalytic regeneration of the coenzyme NADH. The team bonded a vinylene-linked covalent organic framework (TDC) to aminated Nb₂CT_x or Ti₃C₂T_x MXene, then coupled the optimized photocatalyst with FDH for photoenzymatic reduction of CO₂ to formate.

Research Background

Enzymatic CO₂ conversion is attractive because FDH can reduce CO₂ to formate under mild conditions with high substrate specificity. The reaction, however, depends on nicotinamide adenine dinucleotide (NADH), whose cost limits scale-up. Many NADH-regeneration systems rely on the soluble rhodium mediator [Cp*Rh(bpy)H₂O]²⁺ (abbreviated [Cp*Rh]). In a photoenzyme cascade, the mediator adds cost and may interact with amino acids in a way that deactivates FDH.

Developing a photocatalyst that regenerates the biologically active 1,4-NADH isomer efficiently and selectively without a soluble mediator is therefore important for sustainable enzymatic CO₂ reduction. The present study combines systematic characterization, photoelectrochemical measurements and DFT calculations to explain why a COF/Ti₃C₂T_x interface improves both activity and regioselectivity.

Instrumentation Used in the Study

The research used a predecessor of the Perfectlight μGAS1001 Trace Gas Reaction Evaluation System. Designed for catalytic reactions that generate trace gases—including overall water splitting and photocatalytic CO₂ reduction—the system integrates control, gas circulation, automatic sampling/injection and reaction modules.

1. Catalyst Synthesis and Interface Construction

Figure 1. Solid-State NMR, PXRD and FT-IR

The successful formation of TDC was verified by ¹³C solid-state CP/MAS NMR, PXRD and FT-IR. NMR signals at 125 and 136 ppm were assigned to aromatic and overlapping vinylene/aromatic carbon, while the 170 ppm signal corresponded to the triazine ring. The new vinylene carbon appeared at 142 ppm, and the absence of a 190 ppm carbonyl signal indicated complete conversion of the DFB monomer.

PXRD peaks at 4.9° and 25.0° were assigned to the (100) and (001) planes. In the FT-IR spectrum, disappearance of the TMTY −CH₃ vibration at 1380 cm⁻¹ and DFB C=O vibration at 1690 cm⁻¹, together with emergence of a C=C band at 1631 cm⁻¹, confirmed formation of the vinylene-linked framework.

Figure 2. XPS Evidence for COF/MXene Bonding and Charge Redistribution

C 1s, N 1s and O 1s spectra contained the characteristic functionalities of both TDC and the aminated MXene sheets. The lower fraction of surface −NH₂ groups in TDC-Nb_0.18 and TDC-Ti_0.12 relative to the parent aminated MXenes showed that aldehyde–amine condensation consumed the surface amino groups.

Nb 3d and Ti 2p signals shifted by 0.6 and 1.0 eV, respectively, toward higher binding energy after composite formation, indicating interfacial charge redistribution. The larger shift for TDC-Ti_0.12 is consistent with stronger charge transfer at the TDC/Ti₃C₂T_x interface.

Figure 3. Morphology and Atomic-Scale Interface

TDC appeared as 200–500 nm fibers, while the aminated MXenes formed layered nanosheets with smooth surfaces and defined edges. After composite formation, TDC fibers were distributed over the nanosheets and linked through C=N bonds. TDC-Ti_0.12 showed a seaweed-like morphology in which the fibers grew directly on the nanosheets, creating close interfacial contact.

HRTEM resolved 1.03 nm fringes assigned to the Ti₃C₂T_x (002) plane and 1.14 nm fringes assigned to the TDC (001) plane. HAADF-EDX maps showed uniform C, N, O and Ti distributions, confirming a crystalline, atomically integrated COF/MXene interface.

2. Photocatalytic NADH Regeneration

Figure 4. Yield, Kinetics, 1,4-NADH Selectivity and TOF

Without [Cp*Rh], pristine TDC regenerated only 20.1% NADH after 24 minutes. Increasing the Nb₂CT_x content from 6% to 18% raised the yield from 30.7% to 39.8%, consistent with a larger number of charge-transfer interfaces. At 24% loading, the yield declined slightly, likely because excess MXene screened incident light. TDC-Nb_0.18 reached a first-order rate constant of 0.018 min⁻¹, higher than pristine TDC containing the homogeneous mediator (0.014 min⁻¹).

The Ti₃C₂T_x series performed even better. TDC-Ti_0.12 regenerated 44.2% NADH without [Cp*Rh] and 88.8% with the mediator, with corresponding rate constants of 0.022 and 0.089 min⁻¹. Its mediator-free yield exceeded the 41.3% obtained from pristine TDC with [Cp*Rh].

Alcohol dehydrogenase assays showed that adding MXene also increased selectivity toward the biologically active 1,4-NADH isomer. Without [Cp*Rh], TDC-Ti_0.12 reached 48.4% 1,4-NADH selectivity, compared with 45.3% for TDC-Nb_0.18 and 29.6% for TDC. Its mediator-free 1,4-NADH TOF of 0.71 h⁻¹ exceeded TDC-Nb_0.18 (0.63 h⁻¹) and mediator-containing TDC (0.49 h⁻¹).

3. Photoelectrochemical Properties

Figure 5. Light Absorption, Band Structure and Carrier Dynamics

The absorption edge shifted from approximately 496 nm for TDC to 504 nm for TDC-Nb_0.18 and 516 nm for TDC-Ti_0.12. The corresponding band gaps were 2.50, 2.46 and 2.40 eV. Mott–Schottky analysis placed the conduction-band potentials at −1.49, −1.58 and −1.64 V versus Ag/AgCl (pH 6.5), sufficiently negative to reduce NAD⁺ and [Cp*Rh].

Photocurrent densities increased from 0.06 μA cm⁻² for TDC to 0.09 μA cm⁻² for TDC-Nb_0.18 and 0.13 μA cm⁻² for TDC-Ti_0.12. Charge-transfer resistance decreased from 476 to 358 and 100 kΩ, respectively. The Ti-containing composite also showed the weakest steady-state photoluminescence, the longest fluorescence lifetime (5.18 ns) and the lowest exciton-binding energy (135.1 meV), all consistent with more efficient carrier separation.

4. Interfacial Charge-Transfer Mechanism

Figure 6. Fermi-Level Alignment and Electron Transfer

DFT-calculated work functions for TDC, NH₂-Nb₂CT_x and NH₂-Ti₃C₂T_x were 3.70, 4.80 and 5.38 eV; UPS measurements yielded 4.83, 5.48 and 5.99 eV. Although the absolute values differed with method, both datasets gave the same order: NH₂-Ti₃C₂T_x > NH₂-Nb₂CT_x > TDC in work function.

Photoexcitation moves electrons within TDC from the DFB unit toward the triazine-rich TMTY unit. Because the Fermi level of TDC is higher than that of either MXene, electrons transfer spontaneously from TDC to the MXene. The larger energy-level difference at the TDC/Ti₃C₂T_x interface supplies a stronger driving force, explaining the longer carrier lifetime, lower exciton-binding energy and higher NADH-regeneration rate of TDC-Ti_0.12.

5. Origin of 1,4-NADH Selectivity

Figure 7. Terminal vs Planar NAD⁺ Adsorption

DFT compared two NAD⁺ configurations: planar adsorption, which favors 1,6-NADH, and terminal adsorption at the C4 position, which favors 1,4-NADH. On pristine TDC, planar adsorption (−2.65 eV) was stronger than terminal adsorption (−2.01 eV), consistent with only 29.6% 1,4-NADH selectivity.

After MXene incorporation, terminal adsorption became strongly preferred. Terminal adsorption energies reached −5.24 eV on TDC-Nb_a and −5.75 eV on TDC-Ti_b, while planar adsorption energies were −3.63 and −3.58 eV. The terminal–planar difference was therefore 1.61 eV for the Nb composite and 2.17 eV for the Ti composite, explaining the stronger C4-site preference and higher 1,4-NADH selectivity of TDC-Ti_0.12.

6. Photoenzymatic CO₂ Reduction to Formate

Figure 8. Formate Production and Cascade Mechanism

Coupling the photocatalytic NADH-regeneration system with FDH enabled enzymatic reduction of CO₂ to formate. Without [Cp*Rh], formate-generation rates were 651.7 μmol g⁻¹ h⁻¹ for TDC, 1214.2 μmol g⁻¹ h⁻¹ for TDC-Nb_0.18 and 1485.1 μmol g⁻¹ h⁻¹ for TDC-Ti_0.12. The optimized Ti composite was therefore 2.3 times faster than pristine TDC and also exceeded mediator-containing TDC (1373.1 μmol g⁻¹ h⁻¹).

¹²CO₂/¹³CO₂ isotope tracing confirmed that CO₂ was the primary carbon source of the formate. Mechanistically, visible-light excitation separates charge in TDC; TEOA consumes the holes; electrons move into the MXene; the MXene surface reduces NAD⁺ to 1,4-NADH; and FDH uses the regenerated coenzyme to convert CO₂ into formate.

Summary

The work demonstrates that an atomically bonded COF/MXene interface can replace a soluble rhodium mediator while improving NADH-regeneration efficiency, 1,4-NADH regioselectivity and photoenzymatic CO₂ conversion. TDC-Ti_0.12 combined efficient carrier separation with preferential terminal NAD⁺ adsorption, delivering 44.2% NADH regeneration in 24 minutes, 48.4% 1,4-NADH selectivity and a 1,4-NADH TOF of 0.71 h⁻¹ without [Cp*Rh].

By linking interface energetics, adsorption configuration and enzyme-cascade performance, the study provides a mechanistic basis for designing greener mediator-free systems for coenzyme regeneration and CO₂ valorization.

About the Corresponding Author

Chao Peng is an associate professor and master’s supervisor in the School of Environmental and Chemical Engineering at Wuyi University. His research interests include multi-field coupled catalysis for green and low-carbon processes and the detection and catalytic treatment of emerging environmental contaminants.

Publication Information

Yihao Lin, Mei An, et al. “COF/MXene-Catalyzed 1,4-NADH Regeneration and Photoenzymatic CO₂ Reduction without Electron Mediator: Efficiency and Mechanistic Insights.” Separation and Purification Technology 394 (2026), 137403. https://doi.org/10.1016/j.seppur.2026.137403.

Related resources: Photocatalytic CO₂ Reduction Technical Resources · Sources of Error in Photocatalytic CO₂ Evaluation

Sales-Email: network@perfectlight.cn