At 136 picometres, a pair of neighbouring silicon atomic columns has served as the two slits in what Naoya Shibata's group at the University of Tokyo calls the first atomic-scale reproduction of Thomas Young's double-slit experiment.

The team channelled a focused electron probe from a 4D scanning transmission electron microscope (4D-STEM) along two adjacent columns of aligned silicon atoms. The columns sit 136 pm apart in a pure silicon crystal roughly 10 nm thick, with the pixellated detector placed about 10 cm downstream. The resulting electron interference fringes span some seven orders of magnitude less than those in Young's 1801 light demonstration with ~1 mm slits and a ~1 m propagation distance.

The work appears in Nature.

What did the Tokyo team actually build?

Conventional crystal interferometry collapses because every neighbouring atom pair behaves as a double slit, and the repeated unit cells smear the signal from any single pair. Shibata's group solved that by exploiting the probe precision and pixellated diffraction capture of 4D-STEM: the electron wave overlapped only the two target columns, isolating their contribution.

"Our key idea was to use these two atomic columns as the two slits," team member Takehito Seki said. "By channelling the probe along both columns at once, we turned them into two coherent scattering sources — effectively making the crystal itself into an atomic-scale double-slit interferometer."

The measured pattern behaves like a textbook Young experiment: clear maxima and minima, with fringe spacing set by the 136 pm column separation and the 10 cm propagation distance.

Why do the fringes survive at high temperature?

Thermal motion is the obvious enemy of any interferometer this small. Each silicon atom vibrates, and in a classical picture that jitter should randomise the phase between the two electron pathways and erase the fringes.

The team found the opposite. Distinct fringes remained visible from 300 K to 900 K. The mechanism, Shibata explains, is correlation rather than amplitude: "In a which-path picture, relative motion of the two atomic columns makes the two electron pathways more distinguishable and reduces the interference, whereas correlated in-phase motion preserves coherence." Phonons in adjacent columns couple, so the two slits move together rather than independently.

That detail matters because it turns the atomic-scale double slit into a phonon sensor as well as an electron interferometer.

What could this deliver for chip R&D?

Shibata frames the result as a metrology milestone first and an application story second. Direct measurement of phonon correlations between single pairs of atoms is currently out of reach for inelastic neutron or Raman scattering, both of which average over many unit cells.

"It also opens a route to probing local lattice dynamics at the scale of individual atomic bonds, including at interfaces and defects, where local lattice dynamics can strongly influence thermal transport," Shibata says. "In the long run, such atomic-scale insight could help inform better thermal management strategies for future chip technologies."

For R&D managers, the practical read-across is narrow but concrete: a 4D-STEM workflow that already exists in advanced characterisation labs can now resolve bond-scale vibrational coupling in silicon and, in principle, other crystals. Whether the technique becomes routine depends on acquisition time — 4D-STEM datasets remain minutes per pixel rather than milliseconds — and on whether the correlated-phonon signature survives in materials more disordered than pure silicon.

The forward question is whether the Shibata–Seki geometry transfers to the stacks where thermal bottlenecks now sit: gate oxides, the interfacial layers in 3D NAND, and the disordered silicides used in advanced interconnects. Success there would put atomic-column phonon data on a thermal-budget spreadsheet.