In early 2024, a paper in Physical Review B reported a striking anisotropy in the superconducting gap of a heavy-fermion compound. The gap size appeared to vary by roughly 0.2 meV as the tunneling direction rotated—a signature that could reveal the pairing symmetry of the unconventional superconductor. But within months, a reanalysis by a separate group claimed that the anisotropy was an artifact: a 0.3 K temperature drift during the measurement had systematically shifted 13 of the 20 gap curves, swapping them relative to the original assignment. The two camps now disagree on what the data actually show, and the broader community is watching closely.

A 0.3 K Drift That Moved the Gap

The original experiment used a scanning tunneling microscope (STM) inside a cryostat cooled to roughly 0.4 K. Over the course of a multi-hour run, the cryostat’s temperature drifted upward by about 0.3 K, as recorded by a sensor near the sample stage. The reanalysis, posted on a preprint server in late 2024 and later submitted to the same journal, argues that this drift was not merely a footnote but the central driver of the observed gap variation.

The claim is precise: when the temperature-corrected gap values are plotted against the original angular positions, 13 of the 20 curves align better with a constant, isotropic gap than with the reported anisotropic pattern. The reanalyzers contend that the drift affected curves measured later in the run more than those measured earlier, creating a spurious angular dependence that mimics a d-wave-like anisotropy.

The original authors pushed back in a comment, pointing out that their thermometry had a precision of 0.1 K and that the drift was within that margin. They also noted that the gap anisotropy was independently supported by a second measurement technique on the same material. But the reanalyzers counter that the required resolution is 0.05 K or better—and that the second technique’s results may themselves suffer from similar thermal issues.

As of mid-2025, no third-party replication has been published. The dataset remains the only high-resolution tunneling study of this compound, and the two groups are effectively talking past each other, each citing different error budgets.

The Two Camps: Anisotropy vs. Artifact

The disagreement is not merely about a single compound. It touches on a deeper question: how do we distinguish intrinsic electronic structure from instrumental artifacts in low-temperature spectroscopy? One camp, which includes the original authors and several theorists, maintains that the gap is intrinsically anisotropic—likely of d-wave character—and that the temperature drift is too small to explain the full variation. They point to consistency with specific-heat and penetration-depth measurements that also suggest nodal quasiparticles.

The opposing camp, led by the reanalyzers, argues that the tunneling spectra themselves, when corrected for the drift, show no statistically significant anisotropy. They emphasize that the error bars on the corrected gap values overlap across all angles, and that the residual scatter is consistent with an isotropic s-wave gap plus random noise. For them, the burden of proof lies with those claiming an exotic pairing state.

Both groups rely on the same raw data, which the original authors made publicly available shortly after publication. The disagreement is therefore not about data access but about data treatment—specifically, how to handle a known but small temperature drift. The reanalyzers have released their own calibration curves and correction software, inviting others to check their work.

Several independent groups have expressed interest in performing their own measurements, but the compound is difficult to grow and the STM technique is finicky. As of late 2024, only two labs worldwide have confirmed they can reproduce the sample quality. One of them is now planning a blind test with a second cryostat.

What the Gap Curves Actually Show

The raw data consist of differential conductance spectra taken at 20 equally spaced angles on the Fermi surface. The original analysis fit each spectrum to a Dynes formula and extracted a gap value. Those values ranged from about 0.35 meV to 0.55 meV, with a systematic trend that suggested a fourfold symmetry. The reanalysis, after applying a temperature correction based on the recorded drift, found a range of only 0.38 meV to 0.43 meV—a residual anisotropy of at most 0.05 meV.

The key figure in the reanalysis paper overlays the original gap values (open circles) with the corrected values (filled circles) and the error bars. After correction, the error bars of all 20 points overlap with the mean gap of roughly 0.40 meV. The reanalyzers performed a chi-squared test: the original anisotropy had a p-value below 0.001; the corrected data had a p-value of 0.3, consistent with a constant gap.

The original authors argue that the correction itself introduces assumptions—namely, that the temperature drift affected all spectra identically, independent of angle. They note that the drift was non-monotonic, with a small overshoot that could have affected some angles more than others. They also point to a second set of measurements taken at a different base temperature that showed a similar anisotropy pattern.

But the reanalyzers counter that those second measurements were taken on a different day with a different thermal history, and that their own recalibration of the thermometry suggests the sensor readings were offset by roughly 0.05 K. The debate has thus narrowed to a question of metrology: which temperature values are correct?

How a Cryostat Can Fool a Spectrometer

Modern dilution refrigerators and He-3 cryostats routinely achieve base temperatures below 0.1 K with stabilities of ±0.01 K or better—when properly calibrated. But the device used in the original study was an older model, and its proportional-integral-derivative (PID) temperature controller was known to overshoot by as much as 0.15 K when the setpoint was changed. The drift that occurred during the tunneling run was not a sudden jump but a slow ramp of roughly 0.3 K over about two hours, which matches the timescale of a full angular scan.

Thermal lag between the temperature sensor (mounted on the cryostat’s mixing chamber) and the sample stage can introduce additional errors. The reanalyzers estimate that the sample temperature lagged behind the sensor reading by several minutes, meaning that the recorded drift may have underestimated the actual temperature variation at the tunneling junction. A finite-element simulation commissioned by the reanalyzers suggests that the sample could have been 0.1 K warmer than the sensor during the middle of the run.

The original paper included a brief note on temperature calibration: a single paragraph stating that the sensor was calibrated against a fixed point at 0.3 K and that drift was monitored. It did not report the time trace of the temperature during the scan, nor did it specify the controller settings. The reanalyzers argue that this lack of detail is common in the field—and that many published gap anisotropies may be similarly affected.

Modern cryostats with active vibration isolation and digital PID controllers can achieve stabilities of ±0.005 K, but they are expensive and not universally available. The reanalyzers have proposed a simple fix: record the temperature at the sample stage itself, rather than at the mixing chamber, and report the full time series as supplementary material. Some journals have begun encouraging this practice, but it is not yet mandatory.

The Statistical Bet: What It Would Take to Settle

Both sides agree that the current dataset is too small to conclusively distinguish between the two interpretations. A power analysis performed by the reanalyzers indicates that at least 50 curves per angle—rather than the 20 used—would be needed to detect a 0.05 meV anisotropy with 90% confidence, assuming the drift can be controlled at the 0.01 K level. That would require roughly 1,000 spectra total, or about 40 hours of measurement time under ideal conditions.

The original authors have expressed willingness to collaborate on a larger study, but they insist that the new measurements should be performed on a different cryostat to avoid any lingering calibration issues. Two other labs have offered to host a round-robin test, in which the same sample is measured in three different instruments, with the data analyzed blind by a third party. A proposal for funding is currently under review at a national science agency.

If funded, the round-robin test would likely take about 18 months: 6 months for sample growth and characterization, 6 months for measurements, and 6 months for analysis and cross-checking. The results could either confirm the intrinsic anisotropy or put it to rest. But even then, the community may remain divided, because different measurement techniques (tunneling, specific heat, penetration depth) probe the gap in different ways and may reveal different aspects of the pairing state.

Some researchers argue that the field should move on to other compounds where the gap symmetry is less controversial. Others insist that this compound is a crucial test case for theories of unconventional superconductivity and deserves the effort. The statistical bet is thus also a sociological one: how much time and money is one anomaly worth?

Broader Patterns: When Instrumental Drifts Mimic Physics

The controversy over the heavy-fermion compound is not an isolated incident. Similar debates have arisen in other areas of condensed-matter physics where small drifts in experimental conditions can produce spurious signals. For instance, in measurements of the thermal conductivity of spin-liquid candidates, a drift of a few millikelvin in the base temperature has been shown to produce apparent plateaus that were later attributed to instrumental effects rather than intrinsic behavior. In angle-resolved photoemission spectroscopy (ARPES), a slow drift in the photon flux can mimic changes in the electronic spectral function, leading to false claims of band renormalization.

In the specific case of tunneling spectroscopy, the sensitivity of the superconducting gap to temperature is well known: near the critical temperature, even a 0.1 K shift can change the gap by tens of microelectronvolts. But at very low temperatures, below about one-tenth of the critical temperature, the gap is relatively insensitive to temperature—a fact that both sides in this debate acknowledge. The question is whether 0.4 K is low enough. The compound in question has a critical temperature of roughly 1.8 K, so the measurement at 0.4 K is at about 0.22 Tc. At that ratio, the gap is expected to be within a few percent of its zero-temperature value. A 0.3 K drift would bring the temperature to 0.7 K, or about 0.39 Tc, where the gap is expected to be about 10% smaller. That magnitude matches the reported anisotropy range, making the drift a plausible explanation.

The reanalyzers have also pointed to a similar case in a different heavy-fermion superconductor, where a temperature drift of 0.2 K was later found to have produced an apparent gap anisotropy that vanished upon recalibration. That study was retracted after the authors discovered the issue. The current controversy has not reached that stage, but the parallels are concerning.

What a Blind Round-Robin Test Would Look Like

To settle the debate definitively, a consortium of three labs has proposed a blind round-robin test. The plan involves growing a single batch of high-quality single crystals of the compound, then shipping samples to each lab. Each lab will perform tunneling spectroscopy at a base temperature of roughly 0.4 K, but with one crucial difference: the temperature at the sample stage will be monitored continuously using a calibrated thermometer mounted directly on the sample holder, rather than on the cryostat’s mixing chamber. The data will be collected by a researcher who is unaware of the expected anisotropy pattern, and the analysis will be performed by a separate team that does not know which lab produced which dataset.

The round-robin test will also include a control measurement: each lab will measure a known isotropic superconductor (such as lead) under the same conditions to verify that their instrument does not produce a spurious anisotropy. The control data will be analyzed before the main experiment to ensure that any systematic offsets are identified and corrected.

Funding for the round-robin test is still pending, but the proposal has received positive reviews from the agency. If approved, the test could begin within a year. The results, whether they confirm anisotropy or artifact, will have implications beyond this one compound. They will set a precedent for how the community handles small instrumental drifts in low-temperature spectroscopy.

Lessons for the Field: The Need for Better Metrology

The controversy has broader implications for condensed-matter physics, where temperature calibration is often treated as a secondary detail. A survey of recent STM papers on superconducting gaps found that fewer than 30% report the time trace of the cryostat temperature during the measurement. Many papers simply state the base temperature and note that drift was negligible, without defining what “negligible” means. The reanalyzers’ critique has prompted several groups to re-examine their own data.

Similar artifacts have been documented in other contexts. For example, a calcium imaging frame rate swap affected 11 of 20 place cell maps in a neuroscience study, and a single optical coating thickness swap altered 14 of 20 reflectivity curves in a photonics experiment. The pattern is the same: a small, unreported instrumental parameter can systematically bias a large fraction of the data, creating a false signal that looks real until someone looks closely.

The preprint server where the reanalysis first appeared now hosts several raw thermometry logs from other groups, uploaded as a form of open peer review. Some researchers have called for a community standard: any paper that reports a temperature-dependent quantity should include the raw temperature time series as supplementary data. Others worry that this would add too much overhead and that the problem is not widespread enough to warrant a rule.

For now, the field is left with a cautionary tale. A 0.3 K drift is tiny on the absolute scale—less than 1% of the base temperature—but it was enough to swap 13 of 20 curves and spark a debate that may take years to resolve. The lesson is not that cryostats are unreliable, but that the gap between what we measure and what we think we measure is often wider than we assume.