Zhiyu Shi

Zhiyu Shi

he/him

Postdoctoral Scholar in UC Berkeley

Projects

Where does the energy of an electrical discharge go, and which reactions determine what it accomplishes? I study how a discharge can sustain a flame, change its transition to detonation, or transform a metal catalyst. Machine-learned molecular dynamics lets me examine the individual collisions and surface reactions behind these larger-scale processes.

From an electrical discharge to a sustained flame

How can a nanosecond electrical pulse start a flame that survives long after the pulse ends? For fuels such as ammonia, slow ignition can limit their use in propulsion. I develop simulations that follow the energy deposited by a discharge from electron collisions with molecules to the growth of a flame in a moving gas.

In ammonia–hydrogen mixtures, I found that a stronger field can divert energy away from the molecular processes that favor ignition. In hydrogen–air, sustaining excitation between pulses lowered the energy needed to ignite the flows I studied. These results show why ignition design depends on which reactions a discharge excites and how long their effects persist.

Curves show how the fractions of electron energy going into excitation, dissociation, and ionization vary with the reduced electric field in an ammonia–hydrogen–air mixture.
In ammonia–hydrogen–air, the electric field determines which molecular processes receive the discharge energy. Full size
Two rows of simulated hydrogen–air temperature maps follow the ignition region at 158 nanoseconds, 5 microseconds, and 0.2 milliseconds.
At equal deposited energy, the hybrid discharge (bottom) sustains a larger ignition region in fuel-rich hydrogen–air than nanosecond pulses alone (top). Full size

Related publications

When ignition triggers a detonation

Once a flame takes hold, the location and timing of the next ignition event can determine whether a shock and a reaction front join to form a detonation. That transition is desirable in detonation engines and dangerous in fuel-handling equipment. My hydrogen–oxygen simulations showed why a plasma can steer it in either direction. A discharge can hasten local ignition yet consume enough fuel ahead of the flame to delay detonation. Predicting the outcome requires following how chemical energy reinforces the shock.

In companion experiments on dimethyl ether led by Madeline Vorenkamp, our team used laser diagnostics and high-speed imaging to probe the role of low-temperature chemistry. In related work with Andy Thawko, we showed how ozone accelerates the coupling between ignition and a shock. Together, these studies show how reactions ahead of a flame can strengthen or weaken its coupling to a shock.

Flame-front velocity versus time for several discharge pulse counts, beside a diagram linking pulse count to ignition, shock formation, and detonation onset.
The velocity histories show when the flame accelerates. The diagram traces how changing the pulse count can advance or delay its transition to detonation. Full size

Related publications

When a plasma changes the catalyst itself

A discharge can reshape chemistry at a solid surface as well as in a flowing gas. In ammonia synthesis, that raises a fundamental design question. What if the plasma changes the identity of the working catalyst and opens a reaction pathway that the original metal could not sustain?

Experiments found that adding hydrogen to a nitrogen plasma could increase nitrogen uptake by nickel, even as less atomic nitrogen was available in the gas. Working with Yiteng Zheng and colleagues, I used plasma kinetics and density functional theory to explain this reversal. Hydrogen adsorbed on the surface helps nitrogen enter the metal. The resulting nitride could itself supply nitrogen for ammonia formation through a Mars–van Krevelen cycle. This suggests a way to tune the working catalyst during operation through the composition of the discharge.

Atomistic diagrams compare surface reaction pathways with a proposed cycle in which plasma-generated radicals help nitrogen enter nickel and lattice nitrogen can form ammonia.
Surface reactions and a proposed alternative route to ammonia in which nitrogen enters the nickel catalyst and can subsequently react with hydrogen. Full size

Related publications

Learning reaction dynamics from molecular collisions

Models of flames and catalysts depend on rate constants. Each one represents the outcome of countless molecular encounters, only a fraction of which form a new bond. I train interatomic potentials on quantum calculations so that enough of these encounters can be simulated to predict rates directly. For a benchmark reaction in hydrogen combustion, combining two levels of quantum chemistry reproduced the reference rates while cutting the cost of generating training data by more than 80%.

With Argonne collaborators, I have extended this approach to collisions that redistribute energy as well as those that form new products. With Aditya Lele and colleagues, I helped develop a force field to follow ammonia synthesis on iron under temperature modulation. These studies connect the motion of individual atoms to the reaction data that larger-scale models need.

A transfer-learned interatomic potential drives molecular collision trajectories. Two plots compare predicted and reference rate constants for hydrogen and oxygen atoms reacting with HO₂.
From quantum energies to collision statistics. The two panels at the bottom test the learned potential against reference reaction rates. Full size

Related publications