6 positions open at MCPL

Our phase-field finite element solver is built for free surfaces that deform and break — jets, drops, coating beads. It already holds material properties inside their physical bounds, follows a bubble rising through a viscoelastic liquid, and resolves ternary triple junctions. Two things are missing. This project is both.

• Out of axisymmetry. Fully three-dimensional interfaces coupled to viscoelastic constitutive models — where a jet breaks asymmetrically and a coating bead stops being a 2-D cross-section.
• Onto the GPU. The solver is CPU-bound, and that is the wall in front of every 3-D case. Porting it — assembly, solver, linear algebra — is the core of this project, not an afterthought.
• Against experiment. DoS-CaBER filament thinning and coating-bead visualization are measured down the hall, so the computations have something to be wrong against.

A separate project from the phase-field LBM work on the next screen — different solver, different physics, different person. We are recruiting for both.

A concentrated slurry remembers how it was sheared. Rheology tells you it changed; it does not tell you what changed. This position works both ends — the flow response, and the structure underneath it.

• Measure the nonlinear response. Step-up and step-down transients and large-amplitude oscillatory shear separate viscoelastic from thixotropic behavior, and expose the frictional contact network that flips a slurry between liquid-like and solid-like at rest.
• See the structure directly. Low-field NMR relaxometry reads solvent mobility at particle surfaces; impedance spectroscopy reads the conductive network. Both run on the same samples as the rheology, so a flow signature and a structural one sit side by side.
• Check it against simulation. Discrete-element simulation resolves the contacts that shear reorganizes; lattice Boltzmann resolves transport through the pore structure the probes report. Dispersant and binder chemistry enter both — the knobs a formulator can actually turn.
• Put it on the robot. The shear protocols are already moving onto our automatic rheometer. Part of the work is judging which probes can follow, and what that buys in a closed-loop formulation search.

Open at either level — one measurement studied deeply, or several combined into a single picture of the microstructure.

Materials-discovery labs ask which candidate is promising. Processability is the next question: can anything actually be made from it? A promising coin cell answers the first and says nothing about the second — that answer comes from whether the operating window of a real unit process opens for the material. Our self-driving platform settles it by running the process, and this position makes it do so for concentrated suspensions.

• Judge the slurry, not the test cell. Processability is decided on the intermediate that enters the process — solids loading, viscosity, yield stress, thixotropy — and on whether a slot-die window then opens without streaks, agglomeration, or drying cracks.
• Measure without stopping the run. Vision-based property measurement during mixing and an automatic rheometer on a robot, so the verdict forms while the experiment is still going.
• Send failures back. A window that will not open is information. Bayesian optimization over physics-informed models picks the next run, and the failure conditions return to candidate design — the loop that makes a laboratory autonomous rather than merely automated.
• Know where processability ends. A window that opens at 100 mm and 1 m/min is processability; whether it survives 800 mm and tens of m/min is scale-up, and a different failure. Statistical processability maps say which side a formulation sits on.

Measurement and decision are one loop, so the position is shaped around whichever end you are stronger at — open to doctoral and postdoctoral applicants alike.