Physicists Build a Table-Top Way to Image an Electron’s 3D Shape

An electron inside a molecule has no single location. It has a shape, a smeared-out cloud of probability called a wavefunction, and that shape decides almost everything the molecule can do: how it grabs light, how it bonds, how it reacts. The trouble is that you cannot photograph it. The wavefunction is not a thing sitting still under a lens, waiting to be caught.

So a team at the University of Göttingen did the next best thing. They measured the part of it that instruments can reach, and let a computer work out the rest.

The technique is called photoemission orbital tomography, and it has been around since 2009, when a group led by Peter Puschnig first showed in Science that molecular orbitals could be reconstructed from photoemission data. Shoot light of the right energy at a well-ordered layer of molecules, and electrons come flying off. Catch those electrons, measure the direction and speed of each one, and you can map the momentum side of the orbital: essentially, a Fourier shadow of the real-space shape you actually want. Recording that shadow across a spread of light energies builds up the full three-dimensional picture. That, at least, is the principle.

Getting the third dimension has always been the expensive part. It meant booking time at a synchrotron, a facility the size of a building, and grinding through measurements at energy after energy after energy.

What the Göttingen group has done is shrink the whole thing onto a laboratory bench. Their light source is a high-harmonic-generation setup, driven by a fibre laser, that produces short bursts of extreme ultraviolet light tunable from roughly 13 to 71 electronvolts. No synchrotron required. They pointed it at a monolayer of PTCDA (perylene-tetracarboxylic dianhydride, an organic semiconductor) with a satisfying brick-wall pattern, sitting on a silver surface, and imaged its two frontier orbitals in full 3D.

“The wavefunction is a fundamental quantity in quantum mechanics, yet it cannot be directly observed or measured,” says Stefan Mathias, one of the physicists behind the work. That’s the whole problem in a sentence. Photoemission gets you the momentum part without disturbing the state; the sign, the phase, the piece no detector records, has to be reconstructed afterwards.

Teaching the Algorithm to Fill the Gaps

Which is where the second innovation comes in, and it might be the more important one. Reconstructing a phase you never measured is an old headache in imaging, and the usual fix is brute force: measure at so many closely spaced energies that you can just interpolate between them. That is exactly the tedium the synchrotron era was built around. Momentum microscopy paired with bench-top high-harmonic light sources had already begun chipping away at that dependence over the past few years, but the third dimension stayed stubbornly out of lab reach. The Göttingen team, working with mathematicians in the same building, rebuilt the reconstruction algorithm from scratch instead. Their method, called cyclic projections, hunts for an orbital that satisfies the sparse data and a handful of physical constraints at once, the molecule’s known symmetry, its rough size, the fact that most of space is empty, and recovers amplitude and phase together rather than interpolating intensity and hoping the sign sorts itself out near the tricky zero crossings. And because it leans on the physics rather than on sheer quantity of data, it needs far less of it.

How much less turned out to be the surprise. The researchers recorded ten datasets, one at each of ten different light energies, to have a thorough benchmark to test against. But when they fed the algorithm fewer and fewer of those measurements, the reconstruction held up remarkably well. Seven energies gave a comfortable balance of speed and reliability. And a full 3D image, they found, could be pulled from as few as four.

Four is not free, mind you. With only four energies the algorithm has more room to go wrong, settling into one of several plausible-but-wrong shapes unless the constraints are tightened, and the team is candid that reliability trades off against how little data you are willing to collect. Seven is the sweeter spot for a static measurement. Still, four worked, and the whole four-energy dataset took about eight hours to record, on a bench, in a normal lab.

To check the result, they compared their reconstructed orbitals against density-functional-theory calculations of the isolated molecule, and the two lined up closely, down to features finer than the spacing between the molecule’s own carbon atoms, around 0.75 angstroms.

From a Frozen Shape to a Moving One

The reason a bench-top version matters is not really the bench. It’s the laser. A synchrotron gives you a steady glow; this source gives you flashes, each one lasting tens of femtoseconds, a femtosecond being a millionth of a billionth of a second. Flashes let you do stop-motion.

“This technique might mean that stroboscopic videography becomes a reality,” says Wiebke Bennecke, the study’s first author, “allowing us to observe not just the shape of wavefunctions, but also to see how it changes.” The dream is an atomic-scale movie: hit a molecule, then watch its orbitals rearrange as it absorbs the light or starts to react, frame by femtosecond frame. Pump-probe photoemission has been filming electrons in solids this way for years, with groups such as Sie and colleagues at the Massachusetts Institute of Technology using tunable ultraviolet pulses to record what they called an ultrafast movie of a material’s electronic structure; doing the same for a full 3D orbital is the piece still missing. The static portraits in this paper are the still photographs that prove the camera works.

That movie has not been shot yet, and the paper is careful on the point; time-resolved 3D imaging is described as a route now open, not a result now in hand. A full dynamical run, by the team’s own reckoning, might take some eighty hours of measurement. But the sparse-data algorithm is exactly what makes such a thing thinkable, because every energy you can drop is hours you get back. Learn the molecule’s static shape first, from seven snapshots, and you can chase its moving one with four. For a field that spent fifteen years imaging orbitals that hold still, the prospect of watching one flinch is a fair distance to have travelled.

Frequently Asked Questions