FyriumLabs
Uppsala, Sweden

Surface chemistry, interpreted by AI

Trustworthy surface chemistry, faster.

Fyrium Labs turns X-ray photoelectron spectroscopy — one of the most information-rich measurements in materials science — into a clear, defensible answer, without waiting on a scarce human expert.

Uppsala, Sweden — home of Ångström surface science Founder: 10+ years in data science, ML & NLP In dialogue with leading XPS instrument makers

The problem

The bottleneck isn’t the machine — it’s the interpretation.

Reading a spectrum by hand is slow, subjective, and expensive. Two skilled experts can draw the peaks differently — and skilled XPS analysts are scarce.

Existing software still expects the human to make every key decision, so results are hard to reproduce and hard to audit. That’s a real problem when the answer feeds a product, a published paper, or a safety decision.

The same spectrum fit three different ways by three analysts.
The same spectrum, fit three defensible ways by three analysts. That ambiguity is exactly what Fyrium is built to remove.
XPS schematic: X-rays strike a sample surface, electrons fly into an analyser, producing a labeled spectrum.
X-rays in, electrons out — and a spectrum that names what’s on the surface, how much, and in what chemical form.

What is XPS

A chemical fingerprint scanner for surfaces.

Shine X-rays at the surface of a material and measure the electrons that come flying off. Every element responds at its own signature energy — a barcode of which elements are present, how much of each, and what chemical form they’re in.

This is the measurement behind AI-XPS, the analytical engine at the core of Fyrium Labs.

BatteriesSemiconductorsSolar cellsCoatingsMedical implantsCorrosion research

Used everywhere real surfaces matter.

The solution

A careful, automatic analyst that shows its work.

Raw instrument files in, a clear report out — a fixed, transparent sequence, with no manual fiddling in between.

01Check the dataIs this measurement even usable?
02Identify what’s thereWhich elements the spectrum contains.
03Fit the peaksSeparate the overlapping bumps and measure each one — the hard part.
04Interpret the chemistryRusted iron or pure carbon — and flag anything uncertain.
05ReportA color-coded summary anyone can read.

Why it’s different

Three principles set it apart.

01

Blind, not biased

It reads what the data actually shows instead of assuming the textbook answer, then labels afterward.

02

Every step is auditable

You can trace exactly how it reached a conclusion — and it says clearly when it isn’t sure.

03

It grades itself

The system re-checks its own fits and can bring in outside AI second opinions on the tricky cases.

Who it’s for

Everywhere real surfaces matter.

Industrial R&D & quality control

Battery, semiconductor, coatings and materials teams who need consistent answers without waiting on a specialist.

Academic & national labs

Researchers who want reproducible, defensible results and higher throughput.

Instrument makers & core facilities

A way to give every user expert-level interpretation, not just the experts.

Non-experts on the team

People who need to understand and act on XPS results without becoming spectroscopists.

It takes a measurement that used to depend on a scarce human expert and makes the interpretation fast, consistent, and checkable.

Fyrium Labs

Surface chemistry you can act on — from raw spectrum to trusted answer.

Kevin Jones, Ph.D. The Imitation Group Fyrium Labs — AI-XPS
Fyrium Labs — AI-XPS Uppsala, Sweden · 2026