In 2019, a catalysis group at a mid-sized European university published a striking result in the journal Organometallics: a palladium catalyst that had reliably delivered high turnover numbers in toluene suddenly performed far worse when the lab switched to tetrahydrofuran (THF). The same metal complex, the same substrate, the same temperature and pressure, yet the turnover number—the average number of times each active site converts a molecule of substrate—dropped by a factor of more than two. The finding appeared as a routine observation, the kind of result that often goes unnoticed. Yet for chemists who study how catalysts actually work, it underscored how sensitive these systems are to conditions that often go unreported.

A Single Solvent Swap Quietly Changed a Catalyst's Numbers

The catalyst in question was a palladium complex used in a carbon-carbon bond-forming reaction, a type of transformation that underpins much of pharmaceutical and materials chemistry. The lab had been running the reaction in toluene, a non-polar aromatic solvent, for months. Their yields were good, and the turnover numbers, typically in the range of several thousand, were consistent across batches. Then a graduate student, following a protocol from a new collaborator, tried the same reaction in THF. The reaction still proceeded, but the turnover number fell to just over a thousand. The student repeated the run, thinking it was an error. It was not.

The group's first instinct was to blame the catalyst batch. They had seen batch-to-batch variability before, so they re-synthesized the complex and tested it again in both solvents. The result held: in toluene, the catalyst turned over roughly three times as many substrate molecules per active site as it did in THF. The difference was not marginal. It was the kind of effect that could change a process from economically viable to not, or shift the outcome of a mechanistic study.

When the team wrote up the results, they included the solvent comparison as a cautionary note, a warning to readers who might want to replicate the reaction. But the more they thought about it, the more they realized that solvent effects of this magnitude are often underappreciated. A solvent is not just a medium; it is a participant in the catalytic cycle. It can coordinate to the metal, stabilize intermediates, or alter the rate of the rate-limiting step. The 2019 observation was not an anomaly. It was a window into how catalysis actually works.

The decision to publish the comparison, rather than bury it in a supplementary section, was a deliberate choice. It acknowledged that the conditions under which a catalyst is tested matter as much as the catalyst itself. And it raised a question that echoes across the discipline: how many published turnover numbers are, in effect, artifacts of a particular solvent, and how many would change if the solvent were swapped?

Why Turnover Number Is the Metric That Matters

Turnover number, often abbreviated TON, is a measure of catalyst efficiency. It tells you how many times, on average, a single active site converts substrate into product before it deactivates. Unlike yield, which can be high even with a poorly performing catalyst if you use enough of it, TON is a per-site measure. It is the number that process chemists care about when they are deciding whether a catalyst is worth scaling up.

A high TON means a small amount of catalyst can produce a large amount of product. That is economically and environmentally attractive. A low TON means you need more catalyst, which is expensive and generates more waste. So when a lab reports a TON, they are making a claim about practical utility. But TON is not a fixed property of a catalyst. It depends on the entire reaction environment, and solvent is one of the most important components of that environment.

Solvent can affect a catalyst in at least three ways. First, it can coordinate to the metal center, competing with substrates for binding sites. Second, it can change the solubility of reactants, products, and the catalyst itself, altering the effective concentrations. Third, it can stabilize or destabilize transition states and intermediates, shifting the energy landscape of the catalytic cycle. Any of these can change the rate of the reaction, the lifetime of the catalyst, or both.

Turnover number is a composite of these effects. It captures not just how fast the catalyst works, but how long it survives. A solvent that accelerates the reaction but also promotes catalyst decomposition might give a similar TON to one that is slower but gentler. Disentangling these contributions requires careful kinetic studies, which is exactly what the 2019 lab did.

The Solvent's Role in Catalytic Cycles

Why would THF cause such a dramatic drop in TON compared to toluene? The answer lies in the electronic and steric properties of the two solvents. Toluene is a non-coordinating solvent; it does not readily bind to metal centers. THF, on the other hand, is an ether with a lone pair of electrons on oxygen, and it can act as a ligand. It can coordinate to a palladium center, occupying a coordination site that might otherwise be available to the substrate. If THF binds more tightly than toluene, it can inhibit the binding of the substrate, slowing the catalytic cycle. More importantly, it can stabilize the resting state of the catalyst, making it harder for the substrate to displace the solvent. This would increase the activation energy for the turnover-limiting step, reducing the overall rate. A slower rate alone would not necessarily lower the TON, if the catalyst remained active long enough. But if the solvent also promotes catalyst decomposition, the cumulative effect is a lower TON.

There is another possibility. THF can act as a hydrogen-bond acceptor, and it can stabilize charged intermediates differently than toluene. In a catalytic cycle that involves a polar transition state, a more polar solvent like THF might stabilize that state, lowering the barrier and increasing the rate. That would be a positive solvent effect. The fact that the lab saw a negative effect suggests that coordination or decomposition dominates in this particular system.

The 2019 study did not resolve which mechanism was responsible. The authors proposed that THF coordination was the most likely culprit, based on NMR evidence of a THF-bound intermediate. But they acknowledged that other factors, such as trace water or peroxides in the THF, could also play a role. Solvents are not perfectly pure, and THF is known to form peroxides over time. A rigorous replication would need to control for solvent purity.

How the Lab Documented the Shift

To understand the solvent effect, the lab did not just measure final yields. They monitored the reaction in real time using in-situ NMR spectroscopy. This allowed them to track the concentration of product over time, and to observe the induction period, the time before the catalyst reaches its steady-state activity. In toluene, the induction period was short, on the order of a few minutes. In THF, it was noticeably longer, sometimes stretching to nearly an hour.

They also ran the reaction with several batches of catalyst, synthesized independently, to rule out batch-to-batch variation. Each batch showed the same pattern: higher TON in toluene, lower in THF. They controlled temperature and pressure carefully, keeping them constant across runs. They even tried a mixed solvent system, varying the ratio of toluene to THF, and found a smooth decrease in TON as the THF fraction increased. That monotonic trend was strong evidence that the solvent was the cause, not some hidden variable.

The team also collected kinetic data at different substrate concentrations. In toluene, the reaction showed classic saturation kinetics, consistent with a Michaelis-Menten-like model. In THF, the kinetics were more complex, with an apparent inhibition at high substrate concentrations. That suggested that THF was competing with the substrate for the metal center, a classic sign of competitive inhibition.

They published the raw kinetic data in the supplementary information, a practice that is still not universal in catalysis. That data allowed other researchers to reanalyze the results and test alternative models. A few groups did, and their reanalyses supported the coordination hypothesis, though they could not rule out other explanations. The transparency of the original report made those independent checks possible.

Comparisons Across Other Catalytic Systems

The 2019 observation is not isolated. Solvent effects of this magnitude have been documented in many catalytic systems. In palladium-catalyzed cross-coupling reactions, for example, the choice between toluene, dioxane, and dimethylformamide can change both the rate and the selectivity of the reaction. In some cases, a switch from a non-coordinating to a coordinating solvent has been shown to reduce the turnover number by an order of magnitude.

Nickel and cobalt complexes, which are increasingly used as cheaper alternatives to palladium, show similar sensitivities. A 2016 study on nickel-catalyzed amination found that the turnover number varied by more than a factor of five depending on whether the solvent was tetrahydrofuran or 2-methyltetrahydrofuran. The two solvents differ only by a methyl group, yet the effect was dramatic. Such findings underscore how subtle changes in solvent structure can have outsized consequences.

Meta-analyses of published catalytic data, while still rare, have begun to quantify these effects. One analysis of over 200 published reactions found that the reported turnover number varied by an average of 30% when the solvent was changed, even when the metal, ligand, and substrate were identical. The variation was larger for reactions in coordinating solvents, suggesting that coordination is a common source of solvent sensitivity.

These patterns have practical implications. If a process chemist is trying to reproduce a published procedure, the solvent purity and grade matter as much as the catalyst batch. A solvent that is nominally the same but contains different impurities can shift the outcome. This is why many labs now report the solvent supplier and grade, and why some journals have begun to require such details in their methods sections.

Practical Takeaways for Reproducible Catalysis

What can a working chemist take away from this story? First, always report the solvent, its purity, and its source. A statement like "reactions were run in THF" is not enough. The THF from one supplier may contain stabilizers or water that the THF from another does not. Those impurities can change the outcome, as can the age of the solvent bottle.

Second, check the catalyst's stability in the chosen solvent before running a full reaction. A simple UV-Vis or NMR experiment can reveal whether the catalyst degrades or forms a stable solvent adduct. If it does, the solvent may not be appropriate, or the conditions may need adjustment.

Third, run control experiments with alternative solvents. If a new catalyst is being developed, testing it in two or three solvents can reveal whether its performance is robust or solvent-dependent. This information is valuable, both for understanding the mechanism and for guiding scale-up.

Fourth, share raw kinetic data. A turnover number is a single point, but the full reaction profile contains much more information. By publishing time-resolved data, researchers allow others to see how the reaction proceeds, to identify induction periods, and to detect unusual kinetic behavior. This kind of transparency is essential for replication.

Finally, consider solvent effects in catalyst design. If a catalyst is intended for a specific application, the solvent should be part of the design criteria. A catalyst that works well in toluene but not in THF may be less useful than one that works well in both. By testing a range of solvents early in the development process, chemists can choose catalysts that are more robust.

The Broader Lesson: Small Variables, Big Consequences

The story of the 2019 turnover number is a reminder that catalysis is a condition-sensitive science. A single variable, the solvent, can change a catalyst's performance by a factor of two or more. This is not a failure of the science; it is a feature of it. Catalysts are molecular machines that respond to their environment, and the environment includes the solvent.

But it does mean that reproducibility requires precise reporting of conditions. A published turnover number is meaningless without the full context: the solvent, the temperature, the pressure, the concentrations, the batch of catalyst. Journals and funding agencies have begun to recognize this, and some now require more detailed methods sections. But the burden is also on individual researchers to be thorough in their descriptions.

There are reasonable disagreements about how much detail is enough. Some argue that a good scientist should be able to reproduce a result from a brief description, relying on common practice. Others counter that common practice varies from lab to lab, and that even small deviations can matter. The 2019 case lends weight to the latter view.

As one commentator put it, the solvent is not a spectator. It is a participant. The challenge for the field is to treat it as such, to report it carefully, and to understand its role. That is not a glamorous task, but it is essential. The next time a catalyst underperforms, the answer may not lie in the metal or the ligand, but in the liquid that surrounds them. The lab that swapped its solvent learned that lesson the hard way, and their published comparison is a gift to everyone who follows.