In a mid-sized catalysis lab at a university in northern Europe, a routine screening of twenty novel catalysts began producing yields that climbed steadily over several weeks. Some reactions reached 85–90%, far above the 55–65% baseline the team had come to expect from similar systems. The lead investigator, a postdoc with six years of experience, initially attributed the jump to a new ligand design. But an undergraduate researcher, tasked with repeating a subset of runs, noticed something odd: the same catalyst gave wildly different yields depending on the day.

The culprit turned out to be an uncalibrated syringe pump—a workhorse instrument used to deliver precise volumes of reagent into each reaction vial. Over the course of the study, the pump's flow rate had drifted, causing 13 of the 20 reactions to receive more catalyst than intended. The inflated yields were not random; they were systematically biased upward by as much as 40%. When the pump was recalibrated and the reactions re-run, only three catalysts exceeded 70% yield. The rest clustered in the 55–65% range, consistent with the lab's historical data.

This incident, described in a preprint posted to ChemRxiv in late 2024, has quietly circulated among catalysis researchers as a cautionary tale. It is not the first time small equipment failures have skewed large datasets—a reproducibility audit traced 19 failures to uncalibrated pH probes in a separate field—but the scale of the effect here is striking. The pump's drift was gradual, making it invisible to routine checks that only compared daily averages against a fixed target. The team's internal standard, a control reaction run every morning, did not flag the problem because the standard itself was affected by the same drift.

The pump's drift and its effect on yields are best understood by examining the timeline. Over the first two weeks, yields drifted upward. The postdoc leading the project was initially pleased: the new ligands seemed to be outperforming earlier generations. But the undergraduate researcher, a third-year student named Elena Vasquez, grew suspicious when she noticed that the yield for catalyst 7—a compound that had previously given 58%—jumped to 82% overnight. She asked to re-run the catalyst 7 reaction from scratch, using a freshly prepared stock solution. This time, the yield was 61%.

Vasquez then compared the pump's logged delivery volumes against the expected 0.1 mL. She found that on days with high yields, the pump had delivered roughly 0.14 mL—a 40% excess. The pump's calibration had drifted over several weeks, likely due to a worn seal in the syringe barrel. The effect was systematic: every reaction on those days received more catalyst, inflating yields across the board. Of the twenty catalysts tested, thirteen showed statistically significant increases, with effect sizes ranging from 15% to 40%.

The lab manager later described the incident as a “quiet failure”—one that produced plausible-looking data rather than obvious errors. Unlike a pump that stops working altogether, a drifting pump can generate results that appear to confirm a hypothesis. In this case, the false positives would have led the team to pursue three catalysts that were, in truth, no better than the median.

How the Malfunction Slipped Past Standard Checks

The pump's drift went undetected for several weeks because the lab's calibration protocol was incomplete. Daily logs recorded the pump's set flow rate and the total volume dispensed, but not the actual volume delivered per stroke. The team relied on an internal standard—a control reaction run each morning—to detect systematic errors. However, the control reaction used the same pump and the same catalyst stock, so it was biased in the same direction as the experimental runs.

Blind re-runs, in which a second researcher prepared fresh stock solutions and used a different pump, were the only way to uncover the problem. When those re-runs were conducted after Vasquez's alert, the corrected yields painted a very different picture. The lab manager acknowledged that the team should have cross-checked the pump's performance by weighing the delivered solvent weekly—a simple sanity check that would have revealed the drift early on.

The incident echoes a broader pattern in laboratory instrumentation. A 2023 survey of 400 chemistry labs found that only about a third performed regular calibration checks on syringe pumps, and fewer than half documented those checks in a way that could be audited. Many labs assume that automated liquid handlers are reliable out of the box, but seals wear, software glitches occur, and drift can accumulate slowly enough to escape notice.

In this case, the pump had been in service for three years without a formal recalibration. The manufacturer recommended annual calibration, but the lab's budget had not allocated funds for it. The team had also skipped daily volume checks because they were considered “too time-consuming” for a high-throughput workflow. The result was a dataset that looked clean but was fundamentally flawed.

The True Yield Distribution After Correction

After the pump was recalibrated and all twenty reactions were re-run in triplicate, the yield distribution narrowed dramatically. True yields for the twenty catalysts ranged from 51% to 74%, with a mean of 61%. Only three catalysts exceeded 70%, and none reached 80%. The original dataset had shown a mean of 73%, with seven catalysts above 80%.

The error was systematic, not random. Because the pump's drift affected all reactions on a given day equally, the relative ordering of catalysts was partially preserved—the best catalysts in the corrected dataset were also among the best in the original. But the magnitude of the differences was compressed. In the original data, the gap between the top and bottom catalyst was 45 percentage points; in the corrected data, it was 23 points. That compression could have misled the team about which catalysts were worth pursuing further.

The corrected results have now been replicated in two independent labs, each using freshly calibrated pumps. The replication studies show a narrow spread, consistent with the post-correction distribution. The team has submitted a correction to the journal that published the initial findings, and the preprint includes both the original and corrected datasets for transparency.

For the broader catalysis community, the case underscores an uncomfortable truth: even well-designed experiments can be derailed by small instrumentation faults. The effect size here—up to 40%—is large enough to change the interpretation of a screening campaign. If the three false-positive catalysts had been scaled up for process development, the waste of time and resources would have been significant.

Implications for High-Throughput Screening Protocols

High-throughput screening (HTS) has become a cornerstone of modern catalysis research. Automated liquid handlers can test hundreds or thousands of conditions in a single day, generating datasets that are too large for manual verification. But the syringe pump incident reveals a vulnerability: HTS protocols often prioritize speed over calibration, and the instruments themselves are rarely double-checked during a run.

A 2024 white paper from the American Chemical Society's Division of Catalysis Science and Technology called for standardized calibration checklists for automated liquid handlers. The proposed checklist includes daily volume verification by weight, weekly cross-checks against a second pump, and monthly recalibration by the manufacturer. Several labs have already adopted these measures, and early reports suggest they catch about one in twenty pumps that are drifting beyond acceptable tolerances.

Some researchers argue that low-cost flow sensors could be integrated into liquid handlers to detect drift in real time. A prototype sensor, developed by a group at the University of Toronto, costs roughly $50 per unit and can detect flow rate deviations of 2% or more. The sensor has been tested on four different pump models and flagged drift in two of them during a month-long trial. However, the sensor is not yet commercially available, and integrating it into existing workflows would require software changes that many labs are reluctant to make.

Others advocate for a simpler fix: building redundancy into the screening process. Instead of relying on a single pump, labs could run each reaction in duplicate on two different instruments, then compare the results. This approach doubles the workload but can catch systematic errors that affect only one instrument. The lab that experienced the syringe pump drift now runs all critical screens in duplicate on separate pumps, and the lab manager estimates that this practice adds about 15% to the total time per screen.

Lessons for Reproducibility in Catalysis Research

The syringe pump incident is part of a larger conversation about reproducibility in chemistry. A 2022 survey of 1,500 chemists published in Nature found that more than 70% had failed to reproduce at least one published experiment from their own lab. Instrumentation errors were cited as a contributing factor in about 20% of cases, but they are rarely reported in method sections.

Journal reviewers seldom ask about pump calibration, and even when they do, authors may not have the records to answer. The lab in this case did not keep detailed logs of calibration dates or results, making it difficult to pinpoint exactly when the drift began. After the incident, the lab implemented a digital logbook that timestamps every calibration check and sends alerts when a pump is due for recalibration.

Some researchers have proposed pre-registering equipment checks as part of study protocols, similar to how clinical trials pre-register analysis plans. A pilot program at the Center for Open Science allows researchers to submit a calibration checklist before starting data collection, and the checklist is then made public alongside the final paper. So far, fewer than 50 chemistry studies have used this option, but early adopters report that it forces them to think more carefully about instrument maintenance.

A community-driven repository of instrumentation failure modes could also help. The lab that experienced the syringe pump drift has contributed its story to a wiki maintained by the Reproducibility Network in Chemistry. The wiki currently contains descriptions of 34 equipment-related failures, including pH probe drift, balance miscalibration, and gas flow controller errors. Each entry includes the effect size observed and the fix that resolved it. The hope is that other labs can learn from these incidents without having to experience them firsthand.

Practical Fixes for Any Synthesis Lab

For labs that cannot afford duplicate instruments or real-time sensors, there are low-cost measures that can catch drift early. The simplest is to weigh the delivered solvent weekly. A syringe pump set to deliver 1 mL of water should produce a mass of roughly 1 gram (assuming room temperature). If the mass deviates by more than 2% from the expected value, the pump needs recalibration. This check takes about five minutes and requires only a balance that most labs already have.

Using two independent flow measurement methods is another safeguard. For example, a lab could verify the pump's performance using both gravimetric analysis (weighing the delivered liquid) and volumetric analysis (using a calibrated syringe to collect the output). If the two methods disagree, there is likely a problem with the pump or the measurement technique. The lab in this case now uses both methods weekly and records the results in a shared spreadsheet.

Sharing negative results—including near-misses like this one—can help the community as a whole. The undergraduate researcher who spotted the anomaly presented her findings at a lab meeting, and the lab later published a brief note in a methods journal describing the drift and its impact. That note has been cited by several other groups who experienced similar issues. The lab manager now encourages all team members to report any suspicious data, even if the cause is not immediately clear.

Looking forward, the catalysis community is beginning to treat calibration as an integral part of experimental design rather than an afterthought. Several funding agencies now require grant applicants to include a line item for instrument maintenance, and the National Science Foundation's Chemistry Division has begun asking for calibration plans in its review criteria. As one program officer put it, the goal is to make calibration checks as routine as recording the ambient temperature. The hope is that by embedding these checks into daily workflows, the field can reduce the frequency of quiet failures like the one that derailed this catalysis study.