Computational Materials Science, as a Service
We calculate what you'd otherwise have to guess at — before you touch a lab bench.
Materials Quest Research Group runs DFT and machine-learning studies on electrode materials, catalysts, and 2D materials for academic groups and industry R&D teams who need results, not a new hire.
What we do
Four ways to get computation off your plate
Each engagement is scoped as a fixed deliverable — a report, a dataset, a ranked candidate list — not an open-ended retainer.
DFT electronic structure & properties
Band structure, density of states, formation energy, and defect calculations for electrode, catalyst, and 2D-material candidates.
ML-accelerated screening
Machine-learned interatomic potentials and surrogate models to screen thousands of candidate structures or compositions fast.
Custom workflow builds
Automated pipelines connecting structure generation, DFT/ML calculation, and analysis — handed off as code you keep.
Training & collaboration
Workshops and embedded support for groups building in-house DFT/ML capability, plus co-authored academic collaborations.
How an engagement runs
Three stages, one deliverable
Scope the system
We pin down the material system, the property you need answered, and what "done" looks like — usually within one call.
Run the calculations
DFT and/or ML calculations run on our compute allocation, with interim checkpoints if the engagement runs longer than two weeks.
Hand off results
A written report with figures, raw output files, and — for method-heavy work — the code, so it's reproducible on your end.
Research themes
Work we've actually done, not stock imagery
Every figure below comes from our own calculations. This is the depth of analysis an engagement delivers.
Computational design of energy materials
We map how defects, dopants, and ion transport govern performance, then screen candidate chemistries at scale before anyone touches a lab bench. Two examples from that work: the defect physics behind cathode rate capability, and a high-throughput screen of a 2D anode family — 26 transition metals across 4 crystal phases against 4 carrier ions, over a hundred candidates ranked by stability, voltage, and capacity.
Catalysts for thermal and electrocatalysis
From adsorption energetics on a bare surface to full reaction networks under applied potential, we model both thermally- and electrochemically-driven catalytic systems and connect DFT energies to measurable yields and selectivity. The examples below span three reaction classes: CO₂ hydrogenation over a phase-separated bimetallic catalyst, the full electrochemical CO₂ reduction network, and biomass oxidation validated directly against experimental yield data.
Machine learning potentials for larger and longer scale simulation
DFT alone can't reach the system sizes or timescales many real problems need. We train machine-learned interatomic potentials (MLIP) on DFT data to drive molecular dynamics and Monte Carlo simulations orders of magnitude larger and longer — tracking phase transformations, diffusion, and surface processes across thousands of atoms and nanoseconds rather than tens of atoms and picoseconds. The examples below apply this to hydrogen storage: spillover into a metal host, a Monte Carlo study of a hydride phase transformation, and molecular dynamics of hydrogen uptake into a nanoparticle.
Interpretable ML for materials discovery
We train surrogate models on DFT data to screen candidates orders of magnitude faster than direct calculation — and we don't stop at accuracy. SHAP analysis identifies which physical descriptors actually drive predictions, so the model tells you why a material works, not just that it does.
About the company
Run by researchers who do this work themselves
Materials Quest is a computational research group operating under the academic services of Suranaree University of Technology (SUT). We are open for collaboration.
Operated under the academic services ofSuranaree University of Technology
Fixed scope, fixed price
You get a quote before we start, tied to a defined deliverable — not billed hours on an open task.
Reproducibility by default
Inputs, parameters, and convergence criteria are documented so results hold up under review.
Confidential by default
Industry engagements are covered by NDA on request; nothing is published without your sign-off.
Academic-rate option
Reduced pricing for university groups, with co-authorship available in place of a fee.
Get in touch
Tell us what you're trying to answer
A short description of the material system and the question you need answered is enough for us to scope a quote — usually within two business days.