For more than 30 years, the field of trapped-ion quantum computing has known it had a noise problem near chip surfaces — and had no precise way to measure it. Researchers at ETH Zurich published a technique in June 2026 that resolves both issues: a single trapped beryllium ion, suspended and repositioned above a quantum chip with micrometer precision, can now build a fully three-dimensional map of the electromagnetic fields that degrade quantum bits — and do so with sensitivity no previous instrument inside a chip trap has matched. For engineers building the next generation of quantum processors, the result means that a 30-year empirical guessing game about which chip materials produce the least noise can, for the first time, be replaced with direct measurement.

The paper, by Tobias Sägesser, Shreyans Jain, and colleagues at the ETH Zurich Institute for Quantum Electronics — published online in Science Advances on June 19, 2026 — establishes a sensitivity record of 10 nanovolts per meter for oscillating electric fields, measured in a single second of wait time. For scale: the electromagnetic field from a mobile phone, measured from several kilometers away, is still roughly 10,000 times stronger than the signals the ETH team can now resolve at micrometer distances from a chip surface.

Why Electric Field Noise Kills Quantum Bits

Trapped-ion quantum computers store information in the electronic states of individual charged atoms, suspended in carefully designed electromagnetic fields and manipulated with laser pulses. In the early decades of the field, those traps filled entire rooms. Miniaturization has since compressed them onto millimeter-scale chips, bringing ions within a hair's breadth of a solid surface — and directly into the electromagnetic environment that surface creates.

That proximity has a severe cost. Electric field noise originating from chip surfaces can jostle a trapped ion's motional state, driving it out of the quantum ground state from which precise gate operations are performed. Researchers have measured heating rates as high as 7,000 quanta per second in typical ion traps; for a fault-tolerant quantum gate to reach 99.9% fidelity, that rate must be below roughly 100 quanta per second. This gap is not a minor calibration problem — it is one of the primary physical limits on how many qubits can be reliably controlled and how fast quantum operations can run.

The noise scales with ion-electrode distance in a particularly unfavorable way: heating rates climb approximately in proportion to the inverse fourth power of that distance, meaning that as chip traps shrink toward the geometries required for large-scale quantum computing, the noise gets dramatically worse, not better. What makes the problem especially stubborn is that no one has established where the noise comes from. Surface adsorbates, fluctuating patch potentials, and thermally activated processes have all been proposed as mechanisms, but none of the competing models can fully explain all experimental observations. Without a precise spatial map of where the noise is concentrated and what it looks like in three dimensions, distinguishing between those models has been impossible. As Prof. Jonathan Home, who leads the ETH Zurich group, put it: "For more than thirty years, researchers have tried to find out where the electric field noise close to a chip comes from."

How the Penning Trap Enables a 3D Scan

The ETH Zurich technique turns the very object being disturbed — a single ion — into the instrument that measures the disturbance. The enabling technology is a chip-based Penning trap, which the same research group first demonstrated in a March 2024 paper in Nature: Penning micro-trap for quantum computing. That earlier paper established full quantum control of a single ion in the Penning trap and the ability to move it freely in the plane above the chip. The 2026 Science Advances paper is the direct fulfillment of the research agenda the group outlined there: systematic characterization of the electric field noise environment as a function of position.

A Penning trap differs from the radio-frequency Paul traps used in most commercial and research trapped-ion systems in a fundamental way: instead of oscillating electric fields in the radio-frequency range, it uses only static electric and magnetic fields. That distinction has two significant consequences for noise sensing, as Jain explained. First, it removes the geometric constraint that limits RF traps to linear ion positioning — the Penning trap can move the ion freely in three dimensions. Second, and more importantly for this measurement, the absence of any oscillating trapping fields means the instrument is not masking the very signals it is trying to detect. In an RF trap, the oscillating field used to confine the ion is electromagnetic noise itself; separating that background from the much weaker stray fields emanating from chip surfaces requires untested assumptions. In the Penning trap, that problem disappears.

The measurement sequence operates as follows. First, laser beams cool the beryllium ion until it reaches its quantum mechanical ground state of motion — it effectively comes to a standstill. Electrode voltages then shuttle the ion to a precise target location above the chip surface; the scan covers a 200-by-200-micrometer area at heights ranging from 50 to 450 micrometers. Once positioned, the researchers simply wait. The stray electromagnetic fields from the chip drive the ion to oscillate inside the trap, gradually building up its motional energy. After a set interval, additional laser pulses read out how much the ion's quantum state has changed, encoding the local field strength. For static electric fields, the team tracks how far the stray field deflects the ion from its equilibrium position, visible under a microscope. For magnetic fields, they measure shifts in the ion's internal energy levels.

The result is a three-dimensional point-by-point field map — the first of its kind with the spatial resolution required to compare directly against competing theoretical models. The Penning trap's additional advantage, Home noted, is that it can be temporarily disconnected from all external voltage sources, eliminating environmental interference that previously forced researchers to rely on unverified assumptions about the field environment.

What the Sensitivity Record Means in Practice

The technique's headline number is the sensitivity achieved for oscillating electric fields: 10 nanovolts per meter within a measurement time of one second. Lead author Sägesser et al., Science Advances 2026 confirmed this represents a new record for this class of measurement inside a chip trap.

That number is significant for two reasons. First, it means the technique can detect field variations that are comparable in magnitude to the weakest signals predicted by some of the microscopic theoretical models of anomalous heating — making it possible, for the first time, to confirm or rule out specific physical mechanisms rather than merely observing their aggregate effect. Second, and more immediately practical, it means the instrument is sensitive enough to detect differences between surface materials and fabrication processes at the level of detail that matters for chip engineering, not just order-of-magnitude comparisons.

Prior efforts to characterize electric field noise in chip traps used simpler scanning geometries — primarily linear translations available in RF Paul traps — which limited the spatial information that could be extracted and required assumptions about the trap's own contribution to the noise background. The ETH technique removes both constraints simultaneously.

What This Result Builds On

The Science Advances paper is not a standalone result — it is the second anchor of a deliberate multi-year research program. The 2024 Nature paper by the same group (Jain, Sägesser, Hrmo, Kienzler, Home, and colleagues) established the Penning micro-trap as a viable quantum computing platform and explicitly identified noise characterization as the next step: that paper noted that compatibility with CMOS manufacturing "demands an evaluation of how close to the surface ions could be operated for quantum computing and will require in-depth studies of heating." The 2026 Science Advances paper delivers exactly that evaluation.

The research arc matters because it positions the Penning trap as more than an academic curiosity. Commercial trapped-ion quantum computers today are built almost exclusively on RF Paul traps, which face scaling challenges from the high-voltage RF signals required, power dissipation constraints, and the geometric restrictions on ion placement. The ETH group's systematic demonstration that a chip-scale Penning trap can not only perform quantum operations but also characterize the noise environment at record sensitivity opens the question of whether the Penning platform could become the preferred architecture for large-scale trapped-ion systems — a question that now has a concrete experimental program behind it, rather than only a theoretical argument.

A Diagnostic Tool for the Quantum Industry

Home described the near-term application of the technique as a materials characterization tool. The ions can scan different regions of a chip built from different electrode materials, directly measuring which surface produces the lowest electric field noise. The manufacturing processes used to fabricate chip traps — deposition methods, surface treatments, cleaning procedures — could similarly be evaluated and ranked by their noise properties before any full quantum system is assembled.

That capability addresses a bottleneck that currently forces researchers to build and test entire quantum processors before noise characteristics can be evaluated. With the scanning probe, that evaluation could move to the materials-screening stage, potentially compressing years of empirical trial-and-error. Noise reduction efforts to date — including argon-ion bombardment cleaning, which the NIST group demonstrated could achieve 100-fold noise reduction in 2012 — have been applied without a clear picture of the spatial structure of the noise being treated. The ETH technique provides that picture for the first time.

Over a longer horizon, the group identified the possibility of adapting the approach to in-situ, real-time diagnostics — a single trapped ion acting as a permanent watchdog inside an operating quantum processor, monitoring the electromagnetic environment as computation proceeds.

The full author list on the paper is Tobias Sägesser, Shreyans Jain, Pavel Hrmo, Alexander Ferk, Matteo Simoni, Yingying Cui, Carmelo Mordini, Daniel Kienzler, and Jonathan Home. The work was conducted at the Institute for Quantum Electronics at ETH Zurich.

Frequently Asked Questions

Why has electric field noise in ion traps been so hard to solve for more than three decades?

The noise originates from microscopic surface processes on trap electrodes — likely fluctuating patch potentials from surface adsorbates or thermally activated surface chemistry — and scales with ion-electrode distance in a steep inverse-fourth-power relationship. That means miniaturized chips, which bring ions closer to surfaces for better control, also expose them to dramatically more noise. Multiple theoretical models have been proposed, but none fully explains all observations, because until now there was no instrument that could map the noise in three dimensions at sufficient resolution to distinguish between competing mechanisms. Researchers have been treating symptoms (surface cleaning, cryogenic cooling) without a diagnostic that could identify the source.

How does the Penning trap approach differ from conventional RF ion traps for this kind of measurement?

A conventional radio-frequency Paul trap confines ions using oscillating electric fields in the RF range. That oscillating field is itself electromagnetic noise — it masks the much weaker stray signals from the chip surface, and it restricts how freely the ion can be repositioned (primarily to linear translations). The Penning trap uses only static electric and magnetic fields for confinement, removing the masking problem entirely and allowing the ion to be moved anywhere in three dimensions above the chip. The result is a scanning probe that can detect oscillating fields at 10 nanovolts per meter — a sensitivity level the RF approach cannot reach with an ion in a chip trap.

What does the 10-nanovolts-per-meter sensitivity record actually mean for quantum computing?

At that sensitivity level, the scanner can detect field variations comparable to the weakest signals predicted by microscopic models of anomalous heating — the phenomenon where ions gain energy from chip surfaces at rates orders of magnitude above what ordinary thermal physics would predict. That means researchers can now, for the first time, compare measured field maps directly against specific theoretical predictions to confirm or rule out proposed mechanisms. Practically, it means chip engineers can screen candidate electrode materials and surface treatments for their noise properties before committing to a full quantum processor build, potentially replacing years of trial-and-error with systematic materials science.

Does this result suggest the Penning trap could replace conventional RF traps in quantum computers?

The two-paper arc from this ETH Zurich group — the 2024 Nature demonstration of quantum control in a chip-scale Penning trap, followed by the 2026 Science Advances noise characterization result — constitutes the first sustained experimental program to establish the Penning approach as a serious quantum computing architecture, not just a precision measurement tool. RF traps face known scaling challenges with high-voltage RF signals, power dissipation, and geometric ion-placement constraints that the Penning approach avoids. Whether those advantages translate into a competitive quantum computing platform depends on future work: extending from a single-ion probe to multi-qubit operations in the Penning geometry. That program is now explicitly underway at the ETH Zurich TIQI Penning trap research group.