Publication date: 15th December 2025
Heterogeneous photocatalysts provide an ideal platform for sustainable chemical synthesis as they are often inexpensive to synthesize at scale and allow facile catalyst recovery and reuse.[1-3] Particulate photocatalysts are most commonly used as dispersions in the bulk of a liquid phase, but this poses intrinsic limitations to efficient catalyst use and scale-up as well as photon harvesting and utilization (Beer-Lambert law), as particles away from the outermost layers of the reaction vessel may not be exposed to light.[4]
Immobilization strategies of photocatalysts on solid supports are therefore being explored, with the aim of replacing dispersed powders with monolithic materials or thin films. Floatable photocatalysts[5,6] are assembled by immobilizing semiconductors on floatable (low-density) support matrices. These materials have recently been demonstrated in emerging technologies for water purification,[7,8] energy harvesting[9,10] solar fuel synthesis,[11] and plastic photoreforming.[6]
Unlike monolithic packed-bed reactors and photosheets, floatable photocatalysts offer a potential platform to compartmentalize products and charges in photoredox catalysis. While they have been shown at the gas-liquid interface, there are no reports on the segregation of liquid species. Efforts to achieve the compartmentalization of liquid phase redox reactions have been mostly inspired by biological systems,[12] using artificial synthetic and colloidal nanoreactors based on liposomal structures.[13-17]
We introduce liquid|solid|liquid (L|S|L) photocatalysis.[18] Here, macroscopic separation of stacked liquid aqueous phase, photocatalytic solid layer, and liquid organic phase allows paired redox synthesis of hydrophilic and hydrophobic chemical species with continuous operation of both liquid streams, which makes it an ideal framework for liquid-liquid flow photocatalysis.
In our first demonstration of this innovative concept, we use a carbon nitride/polypropylene (CNx/PP) composite immobilized between water and organic solvents to pair the synthesis of clean aqueous H2O2 and added-value aldehydes from biomass-derived chemicals, including 1-butanol from fermentation processes, or lignocellulose as a by-product of the pulp industry.
We demonstrate a facile and robust method to prepare metal-free, resource-benign floatable photocatalysts based on solvent-free thermal processing with readily scalable carbon nitride and low-density plastics. The protocol provides a general approach for the preparation of low-density photocatalysts from shredded plastic and any powdered material of interest for catalytic or other purposes, with limited constraints on their thermal or chemical stability. Our fabrication procedure also provides a route for plastic upcycling, in the interest of circular economy.
In applications for paired aqueous|organic photocatalysis, we report the synthesis of 2.7±0.5 mmol L–1 h–1 aqueous H2O2 and 1.5±0.4 mmol L–1 h–1 butanal from 1-butanol oxidation at room temperature under blue LED irradiation. We also demonstrate kraft lignin upcycling in ethyl acetate using concentrated solar irradiation for combined solar light and heat management in a continuous flow process, and we observe kraft lignin depolymerization with a drop in weight-average molecular weight (Mw) from 2746 Da in pristine kraft lignin to 1400 Da with of 5.55±1.90 µmoles of H2O2 extracted after 16 h photocatalysis. We emphasize that the possibility of seamless flow operation highlights the versatility and potential for sustainable chemical synthesis using liquid|solid|liquid photocatalysis.
Our integrated, compartmentalized L|S|L photocatalytic reactors target sustainable synthesis on multiple fronts, including plastic upcycling, biomass and industrial byproduct valorization, abatement of separation costs, and it opens promising unexplored avenues for continuous flow photocatalysis in immiscible liquid-liquid media.
We gratefully acknowledge support by the Cambridge Trust (PhD studentship to A.R.), UK Research and Innovation (EP/S022953/1 to A.R., ERC Advanced Grant EP/X030563/1 to S.B., C.P. and E.R.; PhD scholarship to P.K.K.), UK Department of Science, Innovation and Technology and the Royal Academy of Engineering Chair in Emerging Technologies programme (CIET-2324-83 to E. R.), and the Natural Sciences and Engineering Research Council of Canada (NSERC) Postdoctoral Fellowship programme (to S.L.). The authors would like to acknowledge the Henry Royce Institute (Cambridge Royce Facilities grant EP/P024947/1 and Sir Henry Royce Institute – recurrent grant EP/R00661X/1), Dr. Heather Greer, Dr. Nigel Howard, Dr. Glenn Quek, Dr. Dongseok Kim, Prof. Clare Grey, Natalie Potter, Flora Mammadova, and Prof. Pawel Dydio for assistance with instrumental analysis. The authors would also like to thank Dr. Lu Chen, Ariffin Bin Mohamad Annuar, and Dr. Zhipeng Huang for helpful discussion.
