We design biomaterials backwards — from the function we want to the molecule.
The loop
Most materials research starts from a molecule and looks for something to do with it. We go the other way. We fix the target property first, then work backwards to the composition that produces it: a model proposes candidates, we synthesize them at the bench, measure, and feed the result back as the next round of training data. The loop only turns because computation and synthesis sit in the same lab. Short handle: molecular design with the bench attached.
Three layers
Layer | What | The question it answers |
Source | Nature — a catalogue of problems already solved | where the target function comes from |
Method | Reverse design — from function to molecule | what we do |
Output | Biomaterials | the field we work in |
Evolution fixes a function and then finds a molecule for it. That is reverse design, run for a very long time. Our first paper on grasshopper mandibles (Chem. Mater., 2015) started there, and the question has not changed since — only the direction we now run it in.
Data is the bottleneck
What inverse materials design actually lacks is not algorithms but data to learn from. Most of the effort so far has gone into organic synthesis and catalysis; polymer interfaces and polyphenol chemistry are close to empty. And for nature-derived materials the measurements barely exist — nobody has assembled them.
So we make them. Rather than digging a single organism and a single mechanism deeply — the conventional biomimetics move, and a crowded one — we mine nature as a data source, systematically, and feed what we measure back into the design loop.
What we build
•
Regenerative medicine — bone-regenerative scaffolds, artificial periosteum, bone adhesives, soft-tissue fillers, polymeric drug delivery, microneedles. Skin regeneration extends into cosmetic formulation.
•
Scent and VOC — quantifying how volatile molecules meet polymer surfaces; capture and deodorization materials.
•
Adhesion, coatings, flame retardancy, environmental sorbents — solvent-free adhesives, second-scale curing, nature-derived flame retardancy, heavy-metal capture.
These are not three fields. They are the same method with a different material loaded.
What we don't do
•
Pure informatics on public datasets only — if we don't make data, it isn't our seat
•
One-off demonstrations that leave no data behind — a model cannot learn from them
•
Contract measurement for experiments designed elsewhere — we don't take new ones
•
Head-on competition on catechol adhesion mechanism — we use it as a tool
•
Conventional biomimetics that digs one organism and one mechanism