Our Story
About D10 Analytics
The idea behind D10 Analytics was sparked at Oak Ridge National Laboratory to address an endemic industry-wide problem: labs generate terabytes of characterization data but only analyze a fraction of it. Onyx was built to close that gap.
D10 Analytics exists because of a contamination event that exposed the limits of manual analysis.. A safety-critical additively manufactured part was found to be contaminated — our team faced the task of analyzing numerous mixed powder lots to identify the contamination source, quantify its severity, and determine with statistical confidence whether other parts from the same lots would meet specifications. Manual review couldn’t analyze the quantity of data at hand or with the necessary speed. The pace of the laboratory demanded a model that would work on any dataset, from any alloy, on any microscope — without weeks of labeling or retraining for every new material system.
D10 Analytics and the Onyx Platform
We realized the solution wasn’t just to gather more data; it was to build a smarter way to analyze electron microscope data that could be used universally. We understood the need for a solution that could find the needle in the haystack — whether that needle is a single precipitate, an unusual secondary phase, or a critical corrosion product.
Why “D10”? The name D10 comes from the ten-sided die. Roll one enough times and the distribution converges — that’s the Law of Large Numbers, and it’s the principle Onyx is built on. If you’re making quality decisions from 10 images out of 10,000, you’re not doing statistics.
The Bigger Picture
Onyx is engineered for ultimate flexibility. It is hardware agnostic, designed to run on any dataset from any microscope, on any alloy, and in any industry. This independence frees your workflow from vendor limitations and allows you to integrate data seamlessly across your enterprise.
Onyx is a desktop application that processes all microscopy data locally — your raw data and analysis results never leave your machine. The only network communication is license verification and optional, user-controlled telemetry for error reporting. No data is uploaded to the cloud. This means your sensitive intellectual property stays exactly where it belongs: on your hardware. All output files are saved in the open-source HDF5 format. Only the proprietary machine learning model itself is encrypted, ensuring your data remains accessible in perpetuity.”
While Onyx’s current focus is on quantifying microstructure via Energy Dispersive Spectroscopy, our vision extends far beyond. We aim to create an end-to-end data analysis ecosystem that links empirical microstructure data with processing parameters and material properties. The ultimate goal is predictive insight: if you want to improve a material’s strength by 10%, Onyx will identify the precise microstructural and processing parameters needed to achieve that result.
Joseph Simpson
Founder
B.S. & M.S. Mechanical Engineering (Tennessee Tech), M.B.A. (University of Kansas). Former data scientist at Oak Ridge National Laboratory, focused on sparse sampling, QC of additively manufactured parts for nuclear reactors, and machine vision for electron microscopes.