For decades, the critical temperature of a superconductor has been treated as a bulk property: a single number measured across a crystal containing billions of atoms. But that number is an average, and averages hide the atomic-scale disorder that actually determines performance. Dopant atoms, strategically placed to tweak the electronic structure, do not all contribute equally. Some sit in lattice positions that strengthen electron pairing; others disrupt it. Until recently, no instrument could tell which was which. New single-atom imaging techniques, built on scanning tunneling microscopy, now resolve individual dopant sites and correlate their positions with local changes in the superconducting gap—the direct fingerprint of Tc enhancement.

The Dopant Site Mystery That Limits Superconductors

Superconductivity emerges when electrons, normally repulsive, form Cooper pairs that move without resistance. The critical temperature, Tc, marks the point where this pairing becomes energetically favorable. In most practical superconductors, Tc is tuned by chemical substitution—replacing a fraction of atoms with different elements that alter the density of states or the pairing interaction. For example, in yttrium barium copper oxide (YBCO), substituting yttrium with other rare-earth elements or partially replacing copper with zinc or nickel can dramatically affect Tc. Similarly, in magnesium diboride (MgB2), doping with carbon or aluminum modifies the two-gap superconductivity. The problem is that dopants are not distributed uniformly at the atomic scale. They occupy different crystallographic sites, create local strain fields, and sometimes cluster. Each configuration shifts the local electronic environment in a distinct way.

Bulk measurements, whether transport, magnetization, or specific heat, average over millions of such sites. The result is a single Tc value that represents a weighted mean of many microscopic contributions. This averaging has stymied efforts to understand which dopant arrangements are beneficial. A crystal might contain a mixture of sites that locally raise Tc and others that suppress it, yet the bulk measurement reports only the net effect. Researchers have long suspected that optimizing dopant placement could push Tc higher, but without site-resolved data, the guesswork remained.

Early attempts to correlate dopant concentration with Tc produced inconsistent results across labs. Some groups reported sharp increases with small substitutions; others saw plateaus or even decreases. The discrepancies were often chalked up to sample quality or measurement differences, but the real culprit was likely the uncontrolled distribution of dopant sites. A crystal with the same nominal composition could have very different local arrangements depending on growth conditions, cooling rates, or post-annealing. The field needed a way to see the atoms themselves.

That need has now been met. By combining scanning tunneling microscopy with spectroscopic mapping, researchers can identify individual dopant atoms, determine their lattice occupancy, and measure the superconducting gap directly above each site. The technique does not average; it interrogates one atom at a time. The result is a new kind of structure-property relationship, one that connects atomic coordinates to electronic behavior with precision that was unattainable even a few years ago.

Scanning Tunneling Microscopy at Single-Atom Resolution

Scanning tunneling microscopy (STM) works by bringing a sharp metal tip within a nanometer of a conducting surface and measuring the tunneling current that flows when a bias voltage is applied. The current is exquisitely sensitive to the distance between tip and sample, and to the local density of electronic states. By rastering the tip across the surface, STM produces topographic images with sub-nanometer resolution, sufficient to resolve individual atoms on many surfaces. But resolving atoms is not the same as resolving their electronic influence.

The key advance came from spectroscopy: by varying the bias voltage and recording the differential conductance, STM maps out the local density of states as a function of energy. In a superconductor, this spectrum shows a gap—a region of suppressed conductance around the Fermi level. The width of that gap is directly proportional to the pairing strength, and therefore to the local Tc. By acquiring a grid of spectra across a region of the surface, researchers can build a spatial map of the gap, revealing how it varies from site to site.

Groups at IBM Research and TU Delft were among the pioneers of this approach. In the late 2010s and early 2020s, they demonstrated that STM could not only image dopant atoms but also distinguish between different lattice positions based on subtle topographic signatures. For instance, a 2021 paper in Nature Physics from the group at TU Delft reported atomic-scale imaging of zinc impurities in a cuprate superconductor, revealing site-dependent suppression of the superconducting gap. The technique requires cryogenic operation—typically below 4 Kelvin—to preserve the superconducting ground state and to minimize thermal drift that would blur atomic-scale features. Even at these temperatures, the measurements are painstaking: a single high-resolution map can take days to acquire.

Recent improvements in tip preparation and electronics have pushed the resolution even further. Researchers can now achieve atomic-column sensitivity, meaning they can resolve not just the top layer of atoms but also the positions of subsurface dopants that perturb the surface electronic structure. This is crucial because many dopants in high-Tc materials sit in layers below the surface. The ability to see them indirectly through their electronic signature has opened a window into the bulk that was previously closed.

Correlating Atomic Positions with Local Tc Enhancement

With the technique in hand, the first systematic studies began to appear. A 2022 paper from a collaboration between TU Delft and the University of Tokyo reported on a cuprate superconductor doped with zinc and nickel impurities. Zinc is known to suppress superconductivity, while nickel has a milder effect. The STM maps revealed that the suppression was not uniform: the gap decreased sharply at the zinc site itself, but remained relatively intact a few nanometers away. Nickel, by contrast, produced a shallower depression that extended further.

The effect sizes are striking. In some measurements, the local gap above a favorably positioned dopant was roughly 15% larger than the average gap of the undoped crystal. That is a substantial enhancement, and it suggests that some sites are doing more than their fair share of the work. The challenge is to identify which sites those are. The STM data show that the gap enhancement correlates with the local strain field: dopants that create a compressive strain in the nearby lattice tend to increase the pairing strength, while those that create tensile strain tend to suppress it.

Sample sizes in these studies are modest but meaningful. Researchers typically image a region of about 50 by 50 nanometers, containing on the order of 200 dopant atoms. That is enough to build statistics on the distribution of gap values across different site types, but not enough to claim universal rules. The field is still in the descriptive phase, cataloging which configurations produce which effects. The hope is that these catalogs will eventually become predictive.

A crucial observation is that the local gap is not the same as the bulk Tc. The bulk Tc is set by the most connected path of strong pairing, not by the average. A few exceptionally good sites might raise the overall Tc if they are arranged in a percolating network, whereas isolated good sites do little. This insight shifts the question from “which dopants are good?” to “which arrangements of dopants are good?”—a much richer problem that single-atom imaging is uniquely suited to address.

Disentangling Chemical Substitution from Strain Effects

One of the most vexing issues in dopant physics is separating the chemical effect—the change in electronic structure due to the different valence or size of the dopant atom—from the strain effect—the distortion of the lattice around the impurity. Both alter the local density of states, and both can influence pairing. STM data, combined with density functional theory (DFT) calculations, have begun to separate these contributions.

DFT simulations that place a dopant atom in a specific site and allow the lattice to relax around it reproduce the observed gap variations remarkably well, provided the strain field is included. When the same calculations are run with the dopant’s chemistry but no lattice relaxation, the agreement degrades. This suggests that strain is the dominant factor, at least for the systems studied so far. The chemical identity of the dopant matters mostly insofar as it determines the size and sign of the strain it induces.

Machine learning has entered the picture as a tool to classify site types from STM topographs. Convolutional neural networks trained on images of known dopant arrangements can identify the lattice position of a new atom with high accuracy, even when the topographic signature is subtle. This speeds up the analysis considerably, allowing researchers to map hundreds of sites in a single session. The models also pick up correlations that humans might miss, such as the tendency of certain dopants to cluster near defects.

But the separation is not clean. In some materials, the electronic effect of a dopant extends beyond its nearest neighbors, and the strain field is long-ranged, so the local gap at a given site depends on the entire surrounding configuration. This makes the mapping from atomic position to Tc enhancement a many-body problem. The field is still developing the language to describe these effects, but the single-atom data provide the raw material for building more accurate models.

From Model Crystals to Real Device Interfaces

Most single-atom imaging studies have been performed on high-quality single crystals grown in the lab. These are ideal for fundamental physics, but practical superconductors are thin films, wires, or tapes, often with grain boundaries, defects, and interfaces. The question is whether the site-dependent behavior observed in bulk crystals carries over to these more complex geometries.

Early evidence suggests that it does. STM studies on thin films of iron-based superconductors, such as cobalt-doped BaFe2As2, have shown that dopant atoms near grain boundaries behave differently from those in the bulk. At a boundary, the lattice is already strained, and additional strain from a dopant can either reinforce or counteract the local distortion. In some cases, dopants near boundaries actually enhance the local gap, because they relieve the tensile strain that suppresses pairing. This is a counterintuitive result that would be invisible in bulk measurements.

The implication for device fabrication is clear: if we can control where dopants sit, we can engineer more favorable strain landscapes. That is a tall order, since atomic-scale placement during growth is extremely difficult. But the new imaging techniques offer a way to verify the results of growth attempts. A film can be grown, imaged atom by atom, and the correlation between dopant positions and local gap can be used to refine the growth recipe. This feedback loop is already being used in a few labs.

Practical superconductors also contain many dopant species, not just one. The interactions between different dopants—whether they attract or repel each other—add another layer of complexity. STM is well-suited to study these interactions because it can image multiple species simultaneously and measure how their proximity affects the gap. The hope is that, by understanding these interactions, researchers can design co-doping strategies that maximize the number of beneficial sites.

What This Means for Room-Temperature Superconductivity

The ultimate goal of much of this research is to raise Tc to room temperature, which would transform energy transmission, computing, and transportation. The single-atom imaging technique does not directly bring that day closer, but it provides a tool for optimizing the materials we already have. If we can identify the atomic configurations that maximize local pairing, we might be able to raise Tc of existing compounds by a few tens of percent, which is not nothing.

However, the technique has limits. The most promising room-temperature superconductors, such as the hydrides that superconduct under extreme pressure, cannot be studied with STM because they are not stable at ambient conditions and are typically synthesized in diamond anvil cells. The STM tip cannot reach them. Other probes, such as nuclear magnetic resonance or muon spin rotation, provide bulk information but not atomic-scale spatial resolution. The transfer of single-atom imaging to these materials will require new experimental platforms.

Even within the materials that can be studied, the gains are incremental. A 15% increase in local gap does not translate directly into a 15% increase in Tc, because the bulk Tc depends on the global connectivity of strong pairing regions. The field is cautiously optimistic that optimizing dopant placement could push Tc higher, but the improvements are more likely to be measured in a few degrees Kelvin than in a leap to room temperature.

There is also a healthy skepticism about whether the rules learned from one family of superconductors apply to others. Cuprates, iron-based, and nickel-based superconductors have different pairing mechanisms, and what works for one may not work for another. The single-atom imaging approach is powerful precisely because it is agnostic to the mechanism; it measures the local gap directly. But the interpretation of those measurements still requires a theory of how local pairing gives rise to global Tc, which is far from settled.

The Next Instrument: Toward Operando Atomic Imaging

The current state of the art is static: images are taken at cryogenic temperatures, after the material has been grown and cooled. But many questions require watching dopants move or change their environment in real time. For example, do dopants diffuse during annealing? Do they rearrange under applied electric fields? To answer these, researchers are developing operando STM, where the imaging is performed while the sample is being modified.

One promising direction is combining STM with RF reflectometry, which allows faster acquisition of spectra without sacrificing spatial resolution. This could enable time-resolved measurements of dopant dynamics, such as the hopping of an atom from one site to another. The challenge is that such hopping events are rare and fast, so the instrument must be both sensitive and fast. Prototypes are in development, but they are not yet routine.

Another open challenge is beam damage. Some STM modes use high bias voltages that can displace atoms or alter the sample. Even low-energy tunneling can occasionally induce atomic motion, especially in soft materials. Researchers must be careful to distinguish intrinsic behavior from tip-induced artifacts. This is a particular concern for organic superconductors, which are more fragile than inorganic ones.

Sample drift is a constant enemy. Even at cryogenic temperatures, thermal expansion and mechanical vibrations can cause the tip to shift by nanometers over the course of a measurement. Active feedback systems that correct for drift in real time are improving, but they add complexity. The community goal is to reach a state where atomic-scale imaging is as routine as optical microscopy, but that day is still years away.

The promise of operando imaging is not just to observe but to control. If we can watch a dopant move into a beneficial position, we might be able to nudge it there deliberately. That would be a true revolution in materials synthesis. Until then, the single-atom imaging results offer a detailed map of the landscape we are trying to navigate—a map that was blank just a decade ago.

For a field that has long been dominated by bulk measurements, the shift to atomic-scale resolution is profound. It is reminiscent of how the invention of the scanning tunneling microscope in the 1980s transformed surface science, or how single-molecule fluorescence changed biochemistry. The difference is that here we are not just looking at atoms; we are looking at the electronic consequences of their arrangement. That is a new kind of sight, and it is already changing how we think about superconductivity.