Luke Wylie, Joshua P. Barham, Barbara Kirchner
4 min
In photoredox catalysis (PRC), researchers often assume that catalysts and substrates act as independent entities that collide randomly in solution. However, recent evidence suggests that pre-existing non-covalent assemblies—specifically pi-stacked complexes—may be essential for efficient electron transfer. This study investigates how solvent choice influences the formation, structure, and longevity of these catalyst-substrate assemblies to explain why certain solvents facilitate successful reactions while others result in failure.
The authors employed specifically parameterized polarizable molecular dynamics (MD) simulations to model the interactions between triarylamine radical cationic photocatalysts and arene substrates. By simulating these systems in different solvents—acetonitrile (MeCN), dimethylformamide (DMF), and dichloromethane (DCM)—the researchers calculated radial distribution functions (RDFs), angular distribution functions (ADFs), and assembly lifetimes. This computational approach allowed them to observe the probability of pi-stacking and the formation of larger aggregates that are difficult to probe with traditional static quantum chemical calculations.
The simulations reveal that solvent choice is a critical, non-spectator variable in PRC. MeCN was found to promote the formation of long, alternating oligomeric assemblies between the catalyst and substrate, which correlates with high experimental product yields. In contrast, DMF disrupts these interactions, leading to a higher proportion of monomers and poor reactivity. The study also confirms that these pi-stacked assemblies are remarkably stable, with lifetimes exceeding 60 ps, providing ample time for photoexcitation and subsequent electron transfer. Furthermore, the data suggest that these assemblies form even in the ground state, prior to electrochemical oxidation, meaning the catalyst and substrate are already organized before the reaction begins.
This work challenges the conventional view of solvents as mere reaction media in photoredox catalysis. By demonstrating that solvents actively tune the aggregation state of the catalyst-substrate pair, the study provides a mechanistic basis for optimizing reaction conditions. These insights suggest that synthetic chemists should treat solvent selection as a primary tool for controlling reactivity and selectivity, rather than just a solubility factor.
Assemblies of photoredox catalysts and their target substrates prior to photoexcitation is a phenomenon naïvely overlooked by the majority of synthetic chemists, but can have profound influences on reactivity and selectivity in photocatalytic reactions. In this study, we determine the aggregation states of triarylamine radical cationic photocatalysts with various target arene substrates in different solvents by specifically parameterized polarizable molecular dynamics simulations. A π-stacking interaction previously implicated by more expensive, less-representative quantum calculations is confirmed. Critically, this study presents new insights on: i) the ability of solvents (MeCN vs DMF) to make or break a photocatalytic reaction by promoting (MeCN) or demoting (DMF) its catalyst-substrate assemblies, which is a determining factor for reactivity, ii) the average "lifetimes" of assemblies in solution from a dynamic simulation. We find that both in the ground state and the photoexcited state, the cationic radical assemblies remain intact for periods often higher than 60 ps, rendering them ideally suitable to undergo intra-assembly electron transfer reactions upon photoexcitation. Such aspects have not addressed by previous studies on synthetic photocatalytic reactions involving non-covalent assemblies.
Alex: [processing] So the solvent is acting as a molecular chaperone — enforcing a geometry that makes the subsequent electron transfer efficient. What are the limits of this picture? [[RP_SECTION:methodological-constraints|Methodological constraints]]
Sam: [measured, direct] The main constraint is the simulation methodology itself. These are classical force fields — polarizable, but without explicit electronic structure. They can capture the aggregation geometry reliably, but they're likely to miss short-lived charge-transfer effects that occur during the actual electron transfer event. So the model is a strong predictor of pre-excitation organization, but it's still an approximation of the underlying physics at the moment of reactivity.
Alex: [connecting the dots] Which means it's a useful screening tool, but you'd want higher-level quantum chemistry before making strong mechanistic claims about the transfer step itself. [[RP_SECTION:implications-for-reaction-design|Implications for reaction design]]
Sam: [nodding] That's the honest read. But even as a screening tool, the implication is significant. If aggregation behavior can be predicted computationally before you run a single experiment, you could design the solvation environment to maximize oligomerization from the outset — shifting solvent selection from empirical trial-and-error toward something more deliberate.
Alex: [reflective] It reframes what the solvent is doing entirely. We usually treat it as a background variable. Here it's the primary driver.
Sam: [quiet conviction] And that reframing has real consequences for how you approach reaction design. When the success of a photoredox reaction is decided in the ground state — by the solvent geometry, before the light is even on — then optimizing the photochemistry without accounting for that pre-excitation assembly means you're solving the wrong problem. That's what makes this worth paying attention to.