They Ran the Double-Slit Experiment Inside a Silicon Crystal

A team has performed Young's double-slit experiment using two neighboring atomic columns in a silicon crystal as the slits. The work, Atomic-scale double-slit interferometry with a focused electron probe, was published in Nature and announced by the University of Tokyo on August 20.

The two Si [110] columns are separated by 1.36 ångströms. A focused electron probe in a scanning transmission electron microscope was delocalized across both, and the resulting interference fringes were measured directly.

It is Young's 1801 experiment, run about ten million times smaller, inside a solid.

Why nobody had done it

Electron interference is not new. Electron biprism experiments demonstrated the wave nature of electrons decades ago, and double-slit versions using fabricated apertures have been done repeatedly, including at the single-electron level.

What had not been done was using the crystal itself as the apparatus. The obstacles are dimensional and practical. The slits have to be a fixed, known distance apart at ångström scale — which a crystal lattice provides for free, and which nanofabrication does not. The probe has to be coherent across both slits simultaneously, meaning the electron wavefunction must be delocalized over 1.36 Å while still being a focused probe you can position. And the detector has to resolve the fringe pattern against everything else a crystal does to an electron beam, which is a great deal.

Aberration-corrected STEM with fast pixelated detectors is what makes this tractable now. The instrument class that enables it has only existed at the required performance for a few years.

The result that matters is the vibration

The interference pattern is the demonstration. The physics is in what the pattern encodes.

At any temperature above absolute zero, the atoms in those columns are vibrating — thermally displaced from their equilibrium positions, continuously. In a double-slit geometry, a slit that moves is a slit whose contribution to the interference pattern changes. The atoms' motion is imprinted directly on the fringes.

Which means the fringes are readable in reverse. Measure the interference pattern, and you can extract how the two neighboring atoms are vibrating — including, critically, whether they vibrate together.

Correlated atomic motion is the entire subject of phonon physics, and phonons are how heat moves through a solid. Thermal conductivity in a semiconductor is not an average property of the bulk; it emerges from how vibrational modes propagate, scatter, and couple at the atomic scale. Until now, that structure has been probed by inelastic neutron scattering, X-ray scattering, and spectroscopy — all of which average over large volumes, and none of which tell you about these two atoms, here, next to this defect.

This technique is local. That is the difference.

What it is good for

The University of Tokyo announcement points directly at thermal design of semiconductors, and the connection is not a stretch.

Heat is the binding constraint on modern chips. Power density at advanced nodes has outrun the ability to remove the resulting thermal load, which is why clock speeds stopped climbing, why datacenter design is now largely a cooling problem, and why 3D-stacked packaging — the direction the entire industry is heading — is fundamentally limited by getting heat out from between the layers.

Thermal transport in real devices is dominated by interfaces, defects, dopants, and boundaries. Those are exactly the places where bulk-averaged phonon measurements fail, because the interesting behavior is local and the measurement is not. A method that reads correlated atomic vibration column by column, at specific sites, is a tool for understanding heat flow where it actually bottlenecks.

That does not translate into a cooler chip next year. It translates into better physical models of interfacial thermal resistance, which is one of the least well-characterized quantities in semiconductor engineering and one of the most consequential.

The other reason it is interesting

There is a foundational-physics dimension that will get less attention than it deserves.

The double-slit experiment is the canonical demonstration of quantum superposition, and its canonical lesson is that which-path information destroys the interference pattern. Here, the "slits" are not passive apertures — they are physical objects that couple to the electron and to a thermal bath, and that are themselves in motion.

The degree to which that motion carries which-path information, and the degree to which it merely modulates the fringes, is a question about decoherence in a system where you can position the probe, know the geometry exactly, and vary the temperature. That is an unusually clean experimental handle on a topic usually studied in far more contrived setups.

Whether the authors pursue that or leave it to others, the platform is now demonstrated.

The framing

There is a particular pleasure in this result that has nothing to do with its applications. Young's experiment with sunlight and a card in 1801 settled a two-century argument about the nature of light. The same geometry, scaled down by seven orders of magnitude and pointed at a silicon lattice, is now a measurement instrument for how atoms shake.

The experiment did not get simpler. The apparatus got small enough that the crystal could be the apparatus.