Химическое сообщество объединяется для решения проблемы воспроизводимости экспериментов

Key Insights - Catalytic chemistry is vital to many economically valuable processes. - Problems with reproducibility arise when important factors influencing reaction outcomes aren’t sufficiently documented. - Chemists are coming together as a community to establish best practices for record keeping, data management, and publishing. In January, 50 chemists held a 3-day workshop in an airport hotel in Denver. The group was comprised of researchers from various career stages and sectors, including academia, industry, and national labs. All of them were connected by a mutual research theme—homogenous catalysis—and the aim of how to make their science more reproducible. “There’s a constant low-level noise,” especially in the past decade or so, about a reproducibility crisis in science, says Rory Waterman, a chemistry professor at the University of Vermont. The workshop discussion arose from an initiative he put together after a US National Science Foundation letter in 2022 called for proposals on the topic of advancing reproducibility. Other scientific subfields, including 2D materials and biomedical research, have initiated similar conversations. "There are hard dollar signs to making an initial discovery that has poor translatability." Fortunately, the consensus from the workshop was that “homogeneous catalysis, on the whole, is pretty stinking reliable,” Waterman says. There are perennial challenges, of course, but hardly approaching the level of a crisis. In the interest of keeping things that way, the attendees brainstormed a collection of best practices for promoting reproducibility at every level of the research ecosystem. After soliciting feedback from prominent researchers who hadn’t been at the workshop, they published their recommendations on a preprint server with an open invitation for more chemists to contribute thoughts on the matter (ChemRxiv 2026: DOI: 10.26434/chemrxiv.15004014/v1). “I think the most important thing that we came up with is that there isn’t a silver bullet,” says Jillian Dempsey, a professor at the University of North Carolina who served on the workshop’s administrative committee. Ensuring that catalysis research is rigorous and that one lab’s findings can be validated by other labs is a community effort, and everyone has a role to play—from bench researchers and their team leads to journal publishers and scientific societies. Technology transfer and trust in science Catalytic reactions are essential to modern life and the global economy, Waterman says. One estimate from 2023 states that catalytic processes account for over 35% of worldwide gross domestic product. “People make drugs, they make polymers, they make other materials, they make other consumer goods. They make agricultural compounds,” and the list goes on, Waterman says. "When something is not reproducible, then that is not necessarily treated like an isolated incident. It’s treated as a symptom of the [entire scientific] community" The importance of research in this area has been recognized repeatedly. Four Nobel Prizes in Chemistry in the last 25 years—2001, 2005, 2010, and 2021—have been awarded to researchers working on homogenous catalysis. That impact would not be possible if the discoveries had not been reproducible, Waterman says. And yet finding a procedure in the scientific literature and then struggling to get it to work in the lab as reported is a near-universal experience for bench chemists. “There’s nothing worse than reading a published paper and then trying to reproduce that yourself and finding some facet or variable that was important was not captured properly,” says David Laitar, a principal research scientist at Dow and a member of the workshop’s scientific advisory committee. Reproducibility struggles have a very literal cost, Laitar says. “There are hard dollar signs to making an initial discovery that has poor translatability” from the bench toward a commercial process. Having agreed-upon community standards and best practices for reporting such procedures would save time and money, he adds. “The dream is [my research] gets used someday for something useful, and so I definitely want things to be repeatable,” says Courtney Roberts, an associate professor of chemistry at the University of Minnesota Twin Cities. Being proactive about reproducibility also goes hand in hand with addressing issues about trust in science, according to Roberts. “When something is not reproducible, then that is not necessarily treated like an isolated incident. It’s treated as a symptom of the [entire scientific] community,” she says. “We’re implementing best practices so that we can potentially regain the trust of people who haven’t trusted scientists over the last few years” and stay accountable to funding agencies. Reproducible recordkeeping There are a staggering number of variables that can influence the fate of a catalytic reaction: temperature, solvent, exposure to oxygen, humidity, light, the type of reaction vessel, the order in which reagents are added—even the position of a flask on a stir plate can change experimental results. And sometimes key factors don’t reveal themselves until someone from another lab tries to replicate a procedure, only to find that the whole thing was dependent on a trace impurity in a reagent or a contaminated stir bar. “All the details matter until they don’t,” Waterman says. Which means the foundation of reproducible research is rigorous experimental design and documentation. In this field, he says, bad actors are much less common than well-meaning researchers simply not keeping a clean lab notebook. “It’s not necessarily that anyone did any malfeasance. . . . Science has so many variables, and it’s hard to really control every single one,” Laitar says. And the number of variables increases as a reaction is scaled, so it’s crucial to have a reliable and detailed experimental record for those who might want to reference or replicate your work in the future, he says. “If you make a discovery that does get commercialized, that plant will last longer than your career, potentially.” Ultimately, individuals are responsible for their own experimental design and record keeping. But principal investigators and team leaders are responsible for setting standards for their lab’s procedures and data management. They should also be having routine conversations with their teams about research ethics and best practices, Dempsey says. She personally reviews her students’ notebooks every year and conducts an annual data management review. Beyond the bench In a publish-or-perish culture there might be a temptation to rush to submit journal papers. But if the whole community gets together and agrees that it’s worth taking the time to make sure published reactions are thoroughly investigated and well documented, it will help everyone, Roberts says. And if there’s grassroots adoption of best practices in individual labs, trainees will carry them forward to their own labs and shift the culture. Roberts says less stigma should be attached to correcting work found to be flawed after its publication. Currently, corrections hurt researchers’ job prospects. She would like to see a system for addressing problems in papers that does not punish people for honesty. Different journals often have slightly different guidelines for how to structure reaction procedures and compound characterization, Waterman says. Inconsistent guidelines create confusion for chemists publishing new research as well as people trying to replicate it. Ideally, publishers would have best-practice checklists that are the same across multiple journals so that people know what is expected of them whether they’re submitting to Nature or Nature Catalysis. In the past, print journals limited the amount of supporting information that could accompany a paper. But online journal publishing now gives researchers the ability to include much more, Laitar says. “We can bolster our supporting information with a lot more details and primary source data than we could just a decade ago. There’s really no reason not to include more.” "All the details matter until they don’t" Another way that journal publishers could support reproducibility is by having supporting information checked as part of the review process. Roberts says the independent journal Organic Syntheses provides a good model because all of the procedures are checked for reproducibility before publishing are checked for reproducibility before publishing. Scientific societies have a part to play as well, Waterman says—for example, by setting curriculum standards and providing training resources for data management and writing papers. Ultimately, addressing reproducibility is an ongoing effort that is “going to take a lot of buy-in across a lot of people and places,” Dempsey says. “It’s not one person’s responsibility. It’s everybody’s responsibility.”