Research spotlight: Sushree Dash

The strongest light-twisting power in a bismuth iron garnet film sits in its outer two nanometers, a layer only a few atomic rows thick. There, at a green laser wavelength of 532 nanometers, the material's magneto-optic gyrotropy — its ability to rotate light's polarization under a magnetic field — runs about seven times higher than in the bulk. Sushree S. Dash, Gregory Odegard and Miguel Levy set out to explain why, and the answer they published in Optical Materials Express points to a reshaped electronic band structure and a narrowed bandgap at the surface.

The reason this matters now is practical. Bismuth-substituted iron garnets are the standard material inside optical isolators, the one-way valves of fiber-optic networks that stop reflected light from destabilizing lasers. Chipmakers want to shrink those isolators onto photonic integrated circuits — chips that steer light instead of electricity — but that requires thin films, and thin films were long assumed to be too weak magneto-optically. The new work shows the surface does the heavy lifting, which reframes how designers should approach the next generation of miniaturized nonreciprocal devices.

For anyone who has streamed video or sent a file across a long-haul fiber link, those isolators were in the path. Cheaper, smaller isolators mean denser and lower-power optical hardware in data centers and telecom networks. Before any of that can happen, designers need a predictive account of why the surface behaves differently — and that is what this paper supplies.

The twist, explained at a bar

Faraday rotation is easier to picture than its name suggests. Linearly polarized light — light whose electric field oscillates in a single plane — is really a blend of two circular polarizations, one spinning clockwise and one counterclockwise. Inside a magnetized material, those two components feel slightly different speeds, so one falls behind the other. When the two halves recombine on the far side, the plane of polarization has turned. Think of two dancers crossing a stage, one spinning clockwise and the other counterclockwise, at unequal speeds; the direction they face together at the far side depends on how much one has lagged. That accumulated turn is the Faraday rotation, and gyrotropy is the material property that sets how strongly it happens.

The effect has been known since Michael Faraday's experiments in 1845, but modern telecommunications leans on it. An optical isolator pairs a 45-degree Faraday rotator with polarizers so that light passes forward while back-reflections get rotated away and blocked. Garnet films, especially bismuth-substituted ones, offer strong rotation with low absorption in the near-infrared, which is why they dominate commercial isolators. A 2019 measurement by the same Michigan Technological University group first spotted the sevenfold surface amplification; the new paper supplies the electronic explanation that was missing.

Rebuilding the surface's energy map

The team attacked the mystery with computation and experiment. First, density functional theory (DFT) — a quantum-mechanical method that approximates how electrons arrange themselves in a solid — compared the electronic states of surface atoms with those in the bulk. DFT produces a band structure, a map of the energy levels electrons can occupy, drawn as bands separated by forbidden gaps.

On top of that map, Dash applied a transition dipole matrix (TDM) model, a calculation that takes the allowed electronic transitions from occupied to empty states and turns their strengths and energies into a predicted Faraday rotation. In plain terms, the TDM model is a translator: feed it a band structure, and it predicts how much a material will twist light at each wavelength. The team ran it on both bulk and surface states. The full study, Band structure reconfiguration and surface Faraday rotation in Bi-substituted iron garnets, is open access in Optical Materials Express.

What the band structure showed

The computations found that surface atoms are not a slightly blurred copy of the bulk. Their band structure is reconfigured, and the bandgap — the energy gap electrons must cross to move from filled to empty states — narrows near the surface. Run the TDM model on those surface states and the predicted Faraday rotation climbs, matching the observed surface enhancement. The same electronic reconfiguration that narrows the gap also amplifies the magneto-optic response.

“The surface band structure reconfiguration and the resulting bandgap reduction are what drive the enhanced Faraday rotation,” Dash explains in the discussion of the results. “Once the model captures that, the surface response stops looking like an anomaly and becomes predictable.”

The modeling did more than explain a single wavelength. By extending the surface-versus-bulk comparison across the full visible and near-infrared ranges, the team predicted that the amplification persists significantly across the telecom band, the infrared window near 1310 and 1550 nanometers where fiber-optic networks operate. A companion density functional theory study of the parent yttrium iron garnet compound had already shown a surface bandgap reduction and stronger spin-orbit coupling near the surface; the new paper extends that picture to the bismuth-substituted material used in devices.

Thinner films, narrower bandgap

On the experimental side, the team used X-ray photoelectron spectroscopy (XPS) to probe the chemical and electronic states at the surface and UV-Vis spectroscopy to measure how the films absorb light. Both pointed in the same direction: as films were thinned from 200 nanometers down to 40 nanometers, the bandgap shrank. A smaller gap lowers the energy required for optical transitions, which shifts and strengthens the absorption features tied to Faraday rotation.

Those measurements anchor the theory in something observable. The DFT and TDM calculations are predictive; the XPS and UV-Vis data confirm that real films follow the trend the model describes, connecting the sevenfold surface effect to a measurable electronic change rather than a numerical artifact.

Where the caution lies

Every computational study carries caveats. DFT is known to underestimate bandgaps in many materials, so the absolute gap values deserve care even when the surface-versus-bulk trend is robust. The TDM model's agreement with experiment is anchored at 532 nanometers, and the telecom-band predictions are extrapolations that still await direct measurement. The films were grown by liquid-phase epitaxy — a high-quality route in which crystalline layers form from a molten flux — on gadolinium gallium garnet substrates. Whether sputtered or otherwise deposited films of the kind more likely to appear on silicon photonic chips reproduce the same surface reconstruction remains open. Earlier work that documented the surface reconstruction in bismuth iron garnets established the structural basis for these electronic differences.

One photonics researcher who was not involved in the study offered a measured take: “The electronic picture is coherent and it explains a genuinely puzzling observation. The engineering risk is that a two-nanometer surface layer is fragile. Capping or patterning can disturb the very reconstruction that creates the enhancement, and even routine thermal processing could erase it, so the practical test is whether the effect survives a real device process.”

The work lists support from Photonica, Inc. and from Michigan Technological University's Applied Chemical and Morphological Analysis Laboratory and Electron Microscopy facility. Industry support of that kind is worth keeping in view when weighing the study's significance.

Why telecom engineers are watching

The practical target is the optical isolator. Today's isolators are discrete, millimeter-scale components built around bulk garnet crystals. In photonic integrated circuits, where light runs through waveguides etched onto chips, an isolator has to be tiny and compatible with wafer-scale fabrication. Thin garnet films are the leading candidate, but making them thinner has traditionally meant sacrificing magneto-optic strength. The surface-enhancement results invert that logic: the thinnest films can be the most efficient per unit thickness, provided the surface is preserved.

The implications ripple outward in a few concrete directions:

• Isolators could shrink because a few nanometers of surface deliver more rotation than bulk films of the same thickness.
• Telecom-band operation speaks directly to the 1310 and 1550 nanometer fiber windows used across long-haul networks.
• A predictive TDM model gives materials designers a screening tool before fabricating films, cutting the trial-and-error cost of new garnet compositions.
• On-chip lasers could gain protection from back-reflections without the bulky external isolators they need today.

Dash has since moved to Chalmers University of Technology in Gothenburg, Sweden, as a postdoctoral researcher, where her current work has turned toward palladium-copper-platinum alloy films for hydrogen sensing. The magneto-optics results represent the closing chapter of her PhD research at Michigan Technological University.

The next ten years

Within five to ten years, this insight could migrate from journal pages into device design rules. If surface reconstruction can be deliberately engineered — through growth conditions and capping layers, or by patterning the surface atom by atom — the two-nanometer amplifier becomes a designed feature rather than an accident of thinning. The TDM model gives other groups a way to screen candidate garnet compositions and surface treatments before spending weeks in a cleanroom.

More broadly, the study is a reminder that when materials are grown thin, the surface is a distinct material in its own right. The properties that matter for light and for magnetic order, along with electronic transport, can all be set in the outermost atomic layers. That is where the next generation of compact optical components will be won or lost.

Photo by Brecht Corbeel on Unsplash