The Science That’s Been Under Our Feet All Along

A hundred years in the making

The foundations of X-ray diffraction were laid in the early 1900s. In 1912, Max von Laue discovered that X-rays passing through a crystal produce a unique scattering pattern, a fingerprint of its atomic structure^[1]. Nobel Prize, 1914. The Braggs built the mathematical tools to decode those patterns^[2]. Another Nobel, 1915. Debye and Scherrer extended the method to powdered samples, making it practical for almost any material. That was 1916^[3].

For most of the century that followed, the technique relied on laboratory X-ray tubes, capable, but limited in brightness and resolution. That changed in the early 1980s, when Hastings, Cox and colleagues published a landmark synchrotron X-ray powder diffraction study in the Journal of Applied Crystallography (1984)^[4]. The improvement in data quality was substantial. Three years later, Cox and Hastings showed that combining synchrotron data with Rietveld refinement, the method for calculating precise mineral quantities from diffraction patterns, produced accuracy that no laboratory instrument could match^[5].

That became the gold standard. The decades since have only reinforced it.

What a synchrotron is, briefly

A synchrotron is a large circular particle accelerator. Electrons travel around a ring at near light-speed and emit extremely intense X-ray beams in the process. The light produced is billions of times brighter than a laboratory X-ray tube, and far more precisely controllable. Mineralytics works with two of the world’s leading facilities through their industry partner networks, the Swiss Light Source at the Paul Scherrer Institute in Switzerland, and the European Synchrotron Radiation Facility in Grenoble, France.

When that beam hits a mineral sample, the resulting diffraction pattern has a clarity that conventional instruments cannot produce. Trace minerals become visible. Poorly crystalline phases, the ones that behave unpredictably in processing, can be identified and quantified. Overlapping mineral signatures that blur together in standard XRD resolve into distinct peaks.

The other thing worth mentioning: a measurement takes seven to fifteen seconds per sample. That means hundreds of samples in a single session. That is a different scale of information entirely.

It was already being used in mining, just not by miners

Academic researchers were applying synchrotron mineralogy to mining-relevant problems long before the industry noticed. One example: a study of the Giant Gold Mine in Yellowknife, Canada, used synchrotron micro-XRD to characterise arsenic-bearing phases in the tailings grain by grain^[6]. Not just detecting arsenic, but identifying exactly which minerals hosted it and in what form. The authors noted that bulk analysis alone would have left ambiguities the synchrotron approach resolved directly. At a site where arsenic speciation had real environmental and regulatory consequences, that distinction was not a footnote.

By 2022, a comprehensive review in Minerals could describe synchrotron techniques as essential for characterising amorphous and poorly crystalline phases, the materials conventional XRD is not built to handle^[7]. The method had earned its place across geoscience, environmental science and materials research over decades of serious use.

This is not new technology being proposed for mining. It is established science that mining has been slow to adopt.

Why it matters for decisions

Mining is full of expensive decisions made on incomplete information. What’s in the ore body, really? Which phases will cause problems in the circuit? Where are the trace minerals that affect recovery?

SR-XRD addresses this from two directions. Resolution, it sees what conventional analysis misses. And throughput, it generates enough data to map variability at the deposit level rather than inferring it from too few samples. Together, that changes the risk profile of every decision that depends on mineralogical data.

I come from a background where information quality and decision quality are understood to be directly connected. That principle doesn’t become less true underground.

One more reason to care: AI needs data it can trust

The research is already pointing in this direction. In 2021, researchers at Argonne National Laboratory trained a machine learning model on synchrotron X-ray data and achieved mineral classification accuracy of 97% or higher^[8]. The model worked because the training data was reliable enough to learn from.

This matters because the industry is moving toward automated mineralogy and ML-assisted decisions. Those systems will be as good as the data they are built on, not better. Investing in measurement quality now is not just about the analysis you get back today. It is about the foundation you are building for everything that comes next.

Standing on the shoulders of giants

The science behind SR-XRD was built by Nobel laureates and refined by generations of researchers. The facilities that run it represent decades of scientific investment. The method has been tested and validated for over a century.

It is ready. It has been ready. The gap is not in the technology, it is in how widely the industry has been willing to use it.

I find that both a little frustrating and deeply exciting. Mostly the latter.