Make sense of your materials and formulations
— in the physics, and in your data

Materia Dynamics is a computational materials science consultancy for R&D teams across the chemical, energy, electronic, semiconductor, biomedical, and advanced manufacturing industries. We predict the properties of your material or formulation — thermal, mechanical, chemical, electronic, spectroscopic, biological, among others — directly from first-principles physics — before your team ever explores it, tests it in the lab, or scales it to production. At the same time, every material and formulation you work with generates a growing volume of information — and the next breakthrough is often already buried in your data. That's why we add artificial intelligence to surface the hidden patterns and turn them into clear direction. The result: a ranked shortlist of optimal formulations that shows you exactly where to take your R&D next. Your team tests only what works.

Materia Dynamics — Simulation Workspace
Target Properties
Thermal cond. Optimized
Viscosity Optimized
Adhesion Optimized
Surface energy Reviewing
Density Constrained
Method DFT + ML
Candidates 512
Candidate Ranking
Top 3 of 512 screened — sorted by composite score
TIM-A7  Top pick 94.2%
EP-C3  Strong 87.6%
SIL-D9  Review 71.8%
TIM-A7 — Property Detail
Thermal Conductivity 4.2 W/mK
Above target threshold (3.5 W/mK)
Viscosity @ 25 °C 12.4 Pa·s
Within process window
Adhesion (Al substrate) 3.1 J/m²
Exceeds minimum spec
Surface Energy 38 mN/m
Close to threshold — validate

Trial and error is costing you time, budget, and market position

Every lab iteration burns weeks, expensive reagents, and engineering hours. And there's no guarantee the result will be optimal.

4–12 mo.

Extended Development Cycles

Traditional formulation discovery requires multiple lab iterations, each consuming weeks. The cumulative timeline delays your product roadmap by quarters.

~90%

Prototypes Discarded

Most lab-tested candidates fail. The reagents, equipment time, and engineering hours spent on them are unrecoverable sunk costs.

Growing gap

Competitive Pressure

Companies investing in computational screening reach optimal formulations faster. The advantage compounds with every development cycle.

From your challenge to ranked candidates in three steps

You bring the engineering problem. We deliver a shortlist of optimal materials — ranked, with data, ready for validation.

01

Discover

We understand your engineering problem, the properties you need to optimize, and your process constraints. Together we define what success looks like.

02

Simulate

We model your materials from first-principles physics — DFT and molecular dynamics — and apply machine learning and graph neural networks to mine your data for hidden patterns. Hundreds of candidates screened in days, not months.

03

Deliver

You receive ranked optimal candidates with predicted property data, confidence scores, and direct recommendations for your production process.

Predict any material property from molecular structure alone

From thermal conductivity to biocompatibility — if it involves a material, a molecule, or a formulation, we can compute it.

You define which properties matter. We screen hundreds of candidates and deliver ranked results with confidence scores.

And when you bring existing data, we apply machine learning and graph neural networks to mine it for hidden patterns — turning information you already have into clearer direction.

Discuss your challenge
Analysis & Discovery 8 Active
Computational capabilities available for your project
Thermal & Electrical Conductivity Active
Physicochemical Properties Active
Surface Compatibility Active
Biocompatibility Screening Active
Formulation Optimization Active
Surface Energies Active
Data Mining & Pattern Discovery Active
Machine Learning & Graph Neural Networks Active

Replace months of lab trial-and-error with weeks of simulation

Traditional R&D tests 10–50 compounds over months. We screen 500+ computationally in weeks — with near-zero reagent waste and physics-backed confidence in every result.

~90% less waste Weeks, not months Physics-backed
Side-by-side comparison
Traditional R&D
Materia Dynamics
Timeline
4 – 12 months
2 – 6 weeks
Compounds
10 – 50
500+
Reagent waste
High
Near zero
Confidence
Best guess
Physics-backed
Deliverable
Lab report
Ranked candidates

Your next step forward is hidden in your data.

A clear, actionable deliverable for your R&D team

Every project ends with a structured output your engineers can act on immediately.

Ranked Candidate List

Top-performing materials sorted by composite score across your target properties. Each candidate includes confidence intervals.

Property Data Tables

Predicted values for every property you specified — thermal, mechanical, chemical, surface — with units and methodology notes.

Production Recommendations

Direct recommendations for your manufacturing process: which candidates to validate first, and what to expect in lab confirmation.

J. Ignacio Borge, Ph.D — Founder of Materia Dynamics

J. Ignacio Borge, Ph.D

Founder & Chief Scientist

Ph.D. in computational chemistry with research experience spanning three countries — from the University of Costa Rica's CELEQ research center to Bar-Ilan University in Israel, to the University of Alabama at Birmingham, where he worked on materials development for transistor applications.

Pharmaceutical consulting in the UK on drug desorption kinetics
R&D in the medical implant sector
Materials development for transistor applications
Trimpot® electronic components (Bourns)
Material separation and recovery consulting for textile recycling (Proquinal)

That combination of fundamental research and applied industry experience now serves a clear purpose: helping R&D teams predict how materials will perform before committing to production. His experience in predicting and optimizing material properties translates directly into ranked candidate lists, property predictions, and formulation recommendations your team can act on.

Materia Dynamics exists because the computational tools to predict material behavior already exist — they just need someone who has spent a career mastering them and knows how to put them to work for engineering teams.

Ph.D. Computational Chemistry M.Sc. Materials Science B.Sc. Chemistry Costa Rica LinkedIn

What R&D teams ask before getting started

Accuracy depends on the property and methodology. For well-characterized systems, DFT and molecular dynamics predictions typically fall within 5–15% of experimental values. We always report confidence intervals so your team knows exactly how much to trust each prediction before committing to lab validation.
A large part of the value lives in the data. Alongside physics-based simulation, we apply data mining, machine learning, and graph neural networks (GNN) to explore large volumes of materials data, find patterns across thousands of compositions, and extract knowledge that would otherwise stay hidden. GNNs are especially powerful for materials: molecules and crystals are naturally represented as graphs — atoms as nodes, bonds as edges — which lets the models learn structure–property relationships directly and predict behavior for candidates that have never been synthesized. This data-driven layer is what lets us screen hundreds of candidates quickly and rank them with confidence.
At minimum, we need your target properties (what you're optimizing for), any process constraints (temperature ranges, compatible chemistries), and your success criteria. The more context you share about the engineering problem, the better we can tailor the simulation. We typically cover all of this in a 30-minute discovery call.
Most projects deliver results in 2–6 weeks, depending on the complexity of the system and the number of candidates to screen. A single-property screening of 200 candidates can be done in under two weeks. Multi-property optimization with hundreds of variables takes closer to six.
Computational screening is designed to narrow the field, not replace lab validation entirely. We deliver ranked candidates with confidence scores — your team validates only the top performers. If a prediction is off, we recalibrate the model with the experimental data and refine the next iteration. The goal is to make your lab time dramatically more efficient.
No. We can work with target property specifications alone — you define what the material needs to do, and we search the candidate space computationally. If sharing formulation details would improve results, we work under NDA. Your intellectual property stays yours.
The founder's experience spans pharmaceuticals (drug desorption kinetics in the UK), medical implants, electronic components (Bourns), semiconductor materials (transistor development at the University of Alabama at Birmingham), and textile recycling (material separation consulting). The computational methods we use are industry-agnostic — whether the challenge lies in energy materials (batteries, fuel cells, photovoltaics, catalysts), chemicals, or advanced manufacturing, if the problem involves materials, molecules, or formulations, we can model it.

Ready to accelerate your R&D?

Schedule a 30-minute technical discussion. We'll assess your challenge and show you exactly what computational simulation can deliver.