Nonadiabatic polaron dynamics from machine-learned hopping trajectories

In the world of materials science, some of the most important processes happen not with a bang, but with a tiny, localized stumble. Imagine a material as a vast, flexible grid of atoms. When an extra electrical charge, like an electron or a "hole" (a missing electron), enters this grid, it doesn't just glide smoothly through. Instead, it pulls the surrounding atoms toward itself, creating a small, self-made distortion in the lattice. This combination of the charge and the distortion it carries is called a polaron. These polarons are everywhere in the energy materials that power our world, from the solar cells on rooftops to the batteries in our phones. How they move and how they interact with light determines whether a device works efficiently or fails. For decades, scientists have struggled to watch these tiny travelers in action. The problem is that polarons move in fits and starts, hopping from one atom to another only when the thermal energy of the material is just right. These hops are rare events that happen over timescales far too long for traditional computer simulations to capture, and they often involve a complex exchange of energy between the moving charge and the vibrating atoms that standard physics models cannot fully describe.

A team of researchers has now built a new way to watch these events unfold, combining the speed of machine learning with the precision of quantum physics to simulate how polarons move in two very different materials: titanium dioxide and magnesium oxide. By creating a framework that can follow the path of a polaron for nanoseconds—a duration long enough to see many hops occur—they were able to observe exactly how these charges jump and how those jumps trigger changes in the material's electronic state. Their work reveals that not all hops are created equal. In titanium dioxide, the most common type of hop is a smooth, continuous slide, but a rarer type of hop is a chaotic, quantum leap that can actually flip the energy state of the electron, either dropping it to a lower energy level or kicking it up to a higher one. In magnesium oxide, the hops are almost always the smooth kind, meaning the charge stays put in its energy state for much longer. This distinction turns out to be critical for understanding how these materials behave in real-world applications like photocatalysis, where the ability of a charge to reach a surface and react with other molecules is the key to the process.

The researchers focused on three specific scenarios to test their new method. First, they looked at bulk titanium dioxide, a material widely used to split water into hydrogen and oxygen using sunlight. Here, they found that the electron polarons move by hopping between titanium atoms. Most of the time, these hops happen along a specific direction where the electron moves smoothly, staying in a low-energy state. However, when the electron hops in other directions, the situation changes dramatically. As the electron moves, the energy gap between its trapped state and the free-flowing energy levels of the material narrows significantly. This narrowing allows the electron to interact strongly with the vibrating atoms, creating a non-adiabatic event. In simple terms, the electron and the atoms stop moving in perfect sync, and the electron can suddenly jump between energy levels. The simulations showed that these rare, chaotic hops are responsible for the electron losing its energy quickly, dropping from an excited state back to the ground state in less than a picosecond at higher temperatures.

The team also discovered something surprising about the reverse process. In these same chaotic hops in titanium dioxide, the energy from the vibrating atoms can be transferred to the electron, pushing it from a calm, ground state up into an excited state. This is a thermally activated excitation, driven entirely by the mechanical motion of the atoms during the hop. It is a two-way street: the hop can cool the electron down, or it can heat it up, depending on the specific moment of the jump. This behavior was not seen in magnesium oxide. In that material, the holes (the positive charge carriers) move in a very orderly fashion, always taking the same path between oxygen atoms. Because the path is so smooth and the energy levels stay far apart, the holes never experience these strong interactions. They remain in their excited states for much longer, sometimes for tens of picoseconds, because there is no chaotic hopping to knock them down or push them up.

The study took this analysis a step further by looking at the surface of titanium dioxide, which is where the actual chemical reactions happen in a photocatalyst. In the bulk material, electrons move relatively freely, but at the surface, they get stuck in a layer just below the topmost atoms. To get to the surface and do any work, they must make a difficult jump from this subsurface layer up to the reactive surface layer. The simulations revealed that this specific journey is almost entirely non-adiabatic. It is a rare, high-energy event where the electron and the lattice vibrations are strongly coupled. When this jump happens, it is accompanied by a massive spike in the interaction between the electron and the atoms, and it frequently triggers the electron to jump to an excited state. This suggests that the very act of a polaron reaching the surface to do chemistry is a quantum mechanical event, not just a simple slide. The researchers found that at lower temperatures, these surface hops are so rare that the electron can stay trapped below the surface for hundreds of picoseconds, effectively waiting for the right thermal kick to make the leap.

By combining machine learning to generate long trajectories with advanced quantum simulations to analyze the electronic transitions, the researchers provided the first direct dynamical proof of a classification scheme that had previously been based only on static calculations. They confirmed that the distinction between smooth, adiabatic hops and chaotic, non-adiabatic hops is real and measurable. The smooth hops in titanium dioxide and magnesium oxide allow charges to move without changing their energy state, while the chaotic hops act as switches that can rapidly change the energy of the charge carrier. This finding changes how we understand the lifetime of excited charges in these materials. In titanium dioxide, the frequent chaotic hops mean that excited charges die out quickly, which can be a problem for efficiency but is essential for understanding the material's limits. In magnesium oxide, the lack of these chaotic hops means charges live much longer, which could be useful for different types of applications.

The implications for designing better energy materials are significant. If a scientist wants to create a photocatalyst that works efficiently, they need to ensure that the charges can reach the surface. The study shows that for titanium dioxide, getting a charge from the subsurface to the surface is a bottleneck that relies on these rare, non-adiabatic events. This means that simply lowering the energy barrier for the jump isn't enough; engineers might need to design surfaces that encourage these specific, strong interactions between the electron and the atoms. The research also highlights the importance of temperature. In titanium dioxide, raising the temperature from 400 to 600 Kelvin makes the chaotic hops happen much more often, causing the excited charges to lose their energy nearly ten times faster. In magnesium oxide, temperature has a much smaller effect because the hops remain smooth and orderly regardless of the heat.

Ultimately, this work bridges a gap between the slow, statistical world of how charges move and the fast, quantum world of how they change energy. It shows that the movement of a charge is not just a physical journey through space, but a dynamic process that constantly reshapes the energy landscape of the material. The researchers have shown that by watching these journeys in detail, we can see exactly when and why a charge loses its energy or gains it, providing a clear roadmap for how to manipulate these processes in future technologies. The ability to simulate these events over such long timescales, capturing the rare but critical moments where the rules of physics shift, offers a powerful new tool for understanding the materials that will power our future.