When Electrons Learn to Dance: Steering Attosecond Currents with Light
16 May 2026, Yanjiang
Scientists steer attosecond electron currents in a scanning tunneling microscope using two-color laser pulses, achieving directional control and sub-femtosecond timing.
The scanning tunnelling microscope has been with us for four decades, and in all that time it has done one thing exquisitely well: it feels atoms. A needle-sharp tip hovers above a surface, a voltage is applied, and electrons tunnel through the vacuum gap — a current so sensitive that individual atoms register as bumps in the landscape. Move the tip laterally, and you trace the topography of matter itself. But there is a second dimension to this story, one that has stubbornly resisted human control: time. Each electron that tunnels does so in a flash so brief — attoseconds, billionths of a billionth of a second — that for years, no one could say precisely when it jumped, or in which direction, or whether that direction could be dictated at all. A preprint (arXiv:2507.10252) from a team led by Michael Krüger at the Technion—Israel Institute of Technology, working with collaborators at Université Paris-Saclay, now answers all three questions. They have clocked the attosecond tunnelling burst, steered its direction at will, and done so under ordinary laboratory conditions — no cryogenics, no vacuum chamber more elaborate than the STM itself.
What makes this possible is a trick with colour. The team fires two laser pulses at the tip-sample junction: an infrared fundamental and its second harmonic, precisely the shade of blue that oscillates twice as fast. When these two fields overlap in the nanoscale gap, their superposition produces an asymmetric waveform — a lopsided electric field whose peaks point preferentially in one direction. It is the optical equivalent of sculpting a wave at the beach so that every seventh breaker crashes harder to the left. The asymmetry gives electrons their marching orders. By turning a tiny knob that shifts the relative delay between the two colours, the researchers can flip the current’s direction — electrons flow from tip to sample, then, with a twist of the delay stage, from sample back to tip.
Two-colour laser pulses create asymmetric waveforms that steer ultrafast currents across a nanoscale junction. This control enables attosecond-precision measurements of electron motion in scanning tunnelling microscopes. (Source: arXiv:2507.10252)
But directional control is only half the achievement. The deeper challenge — the one that has kept ultrafast STM at the frontier of impossibility — is disentangling the genuine attosecond signal from the thermal noise that floods any experiment where laser light heats a metal junction. When you illuminate a nanoscale gap with intense pulses, the tip and sample expand, the gap narrows, and a spurious current flows that has nothing to do with sub-cycle tunnelling. Earlier experiments, including important work by Maier and colleagues (arXiv:2507.10206) that demonstrated attosecond charge transfer in STM, relied on the carrier-envelope phase of single-colour pulses to achieve control. But as Krüger’s team notes, the two-colour approach provides a cleaner separation: the directional current appears as an oscillation at a specific modulation frequency, while thermal effects are effectively filtered out by the lock-in detection scheme. A direct measurement with the tip frozen in place and zero applied bias removes any remaining ambiguity. What you see — an orange curve tracing the raw current versus two-colour delay — is the genuine attosecond burst, unclouded by thermal artifacts.
Now we arrive at the number that lingers in the mind: the burst lasts just 860 attoseconds. That is less than a femtosecond. To grasp it, consider that light itself, sprinting at three hundred thousand kilometres per second, covers barely a quarter of a micrometre in that interval — less than the wavelength of ultraviolet light. Inside the junction, the electron’s journey is even more compressed. The team’s theoretical modelling, which spans both single-active-electron Schrödinger calculations and full time-dependent density functional theory, reveals a three-step choreography. First, the laser field thins the tunnelling barrier, lowering it just enough that an electron near the tip can burrow through. This is not the gentle tunnelling of a steady voltage; it is non-adiabatic, meaning the electron gains energy while still beneath the barrier. Second, having emerged on the far side, it is further accelerated by whatever remains of the laser half-cycle. Third, it slams into the sample, depositing its kinetic energy as a sharp spike of current. The width of that spike, measured at half its maximum amplitude, is the 860 attoseconds the team reports.
One might wonder: is that number hard, invariant, a fundamental constant of the junction? It is not. The team’s simulations show that the burst duration depends on the peak field strength, the intensity ratio between the two colours, and the width of the vacuum gap. Vary any of these, and the duration shifts. Yet the remarkable finding — the one that gives confidence that we are seeing real physics rather than a modelling artifact — is the agreement between two entirely different theoretical descriptions. The single-active-electron model, which treats only one electron as mobile, and the full many-body density functional calculation, which tracks the correlated motion of the entire electron sea, converge on burst durations that differ by less than a tenth of the nominal value. When two theories with such different assumptions converge, the measurement they agree on earns a credibility that neither could command alone.
The spatial resolution is no less striking. Despite running the experiment in ambient conditions — no cryogenic vacuum, no vibration-dampened bunker — the team achieved a lateral resolution of about 2 nm, roughly ten silicon atoms laid side by side. Simultaneously, the topographic sensitivity remained sub-angström. This is the microscope doing two things at once: its traditional job of feeling the atomic landscape, and its new job of timing the electrons that cross it. The two signals, acquired together, reveal no crosstalk — the attosecond current does not degrade the spatial image, and the topography does not leak into the ultrafast measurement. For anyone who has wrestled with the tyranny of signal-to-noise in precision measurements, this is quietly extraordinary.
The road from a laboratory demonstration to a tool that other researchers can use is long, and the authors are candid about what remains to be done. The error budget for the burst duration — the systematic uncertainties that could shift the central value — has not yet been fully dissected. Thermal effects, while dramatically suppressed by the modulation scheme, are not completely absent; a quantitative accounting of the residual heating would strengthen the case that what is measured is purely electronic. But these are questions of refinement, not of existence. The central fact — that attosecond currents in an STM can be directionally controlled and temporally characterized — is established.
Here we should pause and place this work in a longer arc. Since the 1980s, ultrafast microscopy has advanced in pulses: picosecond resolution came first, then femtosecond, and now, with this experiment, the attosecond frontier has been reached in a scanning probe geometry. Each step required not just faster lasers but cleverer ways of isolating the signal of interest from everything else that happens when intense light strikes matter. The Krüger team’s two-colour lock-in approach is the latest in this lineage, and it may prove to be the most general: the same scheme could, in principle, be applied to other tip-sample combinations, other materials, other laser wavelengths. The attosecond burst becomes not an exotic phenomenon confined to one laboratory but a characterizable feature of tunnelling junctions in general.
What does this open, practically? The paper invokes the phrase “lightwave electronics” — a vision in which the oscillating electric field of light, rather than the steady voltage of a transistor, drives electronic circuits. If currents can be switched on attosecond timescales, information could be processed a million times faster than in today’s fastest silicon. The experiment reported here is not an engineering prototype; it is a physics experiment. But it demonstrates the core capability — directional control of a sub-cycle current — that any such future technology would require. The authors also point toward ultrafast microscopy of charge dynamics at surfaces. Imagine watching a single molecule undergo a chemical reaction, not frame-by-frame in femtoseconds, but in the attosecond freeze-frame where electrons actually rearrange. This experiment provides the trigger and the detector for that camera.
Perhaps the deepest implication, however, is one the paper does not dwell on: the marriage of attosecond science with atomic-scale spatial resolution forces us to confront what “tunnelling” actually means when the barrier itself is changing on the same timescale as the tunnelling event. The traditional picture — an electron popping from one side to the other with a fixed probability per unit time — derives from a world where voltages are steady, barriers are static, and time is a parameter rather than a participant. In Krüger’s junction, time is an active variable. The electron’s own clock, the phase of the optical field at the moment it enters the barrier, determines everything that follows: how much energy it gains, how far it travels, how sharply it arrives. This is not just faster tunnelling; it is tunnelling with a memory, tunnelling whose outcome depends on when it began. That subtlety, once fully understood, may reshape how we think about quantum transport in driven systems.
For now, what we have is a measurement. An orange curve, recorded at room temperature in a laboratory in Haifa, that traces the rise and fall of an electron current lasting less than a thousandth of a trillionth of a second. The curve is clean, repeatable, and flips sign exactly when the theory says it should. In an age where science news often arrives hedged with caveats and maybes, there is something bracing about a result that speaks for itself. The team built the experiment, the experiment worked, and the data agree with the calculation. The questions that remain are the right kind of questions: not whether we can control attosecond currents, but how precisely, in which materials, and what will we see when we point this new instrument at the world.
Laser pulses control the direction of electric current flowing across a microscopic gap with attosecond precision. This ultrafast switching opens the door to electronics that operate at speeds a million times faster than today’s chips. (Source: arXiv:2507.10252)
Editorial Note: The critical analysis in this article was informed by a structured review of related work cited by the paper’s authors. No independent experts were interviewed.
Yanjiang is an online editor of LoomSci.com
References
- Daniel Davidovich et al., Clocking and Controlling Attosecond Currents in a Scanning Tunnelling Microscope, arXiv:2507.10252
- Maier et al., Attosecond Charge Transfer in Atomic-Resolution Scanning Tunnelling Microscopy, arXiv:2507.10206