From the time of Thomas Edison to the present day, materials discovery has largely been “trial and error” – tedious, expensive and slow

  • Research today is still conducted by educated guess
  • Even the best researchers can be biased to familiar ideas or incremental optimization to a “local maxima”
  • It is not practical to simulate every compound
  • Only a fraction of known compounds have empirically measured property values

We live in a world of materials limited by human bias and what has been measured

Thomas Edison in his West Orange laboratory. AP Photo/J. Walter Thompson
    Material credited to STScI on this site was created, authored, and/or prepared for NASA under Contract NAS5-26555.

Imagine a world where, in minutes, the key properties of 50,000 materials known to exist can be predicted

  • Any property – whether mechanical, electrical, chemical, thermodynamic or other
  • Imagine the ability to quickly and rationally sort through thousands of materials in a single click to find the one that enables a critical application, that sparks an innovative design, that solves the problem that you have been wrestling with for years

Enter Xaedra. The world's most advanced AI platform that predicts properties based on atomic information

  • Xaedra utilizes neural-network machine learning (ML) methodology
  • Its database includes 53,000 known crystal structures
  • Every crystal structure is processed into an atomistic “fingerprint” enabling application of ML
  • The ML predicts – not simulates – properties of previously uncharacterized materials
  • Additional training loops improve correlation

Xaedra: Elegant, physics “first principles” based Machine Learning for Material Discovery

Xaedra is potentially a breakthrough approach to material discovery Rational Material Design

  • Xaedra both opens the horizon and shortens the time for material discovery
  • Predictions allow an enormous number of candidates to be narrowed down to a few highly likely candidates
  • For most challenges framing in terms of custom properties is key
  • Training data is still required but lab resources can be largely be redeployed to confirmation and validation

Xaedra has successfully predicted well over a million property datapoints from small training datasets

    The physics “First Principles” AI methodology of Xaedra simply works. Take a look at the platform. From roughly 47,000 experimental properties Xaedra has currently predicted over 1.2 million property datapoints. Most of these have well over 90% coefficient of determination. Excellent fidelity – WITHOUT an expensive, slow, limited scope, incremental “big data” AI approach.

Let us help you discover a breakthrough material today!

  • Xaedra staff includes experienced analytical chemists, physicists and ML experts who take a practical, industry focused approach to materials discovery
  • We work with you on a project to solve your particular materials discovery challenge
  • This typically requires identification of “custom” property(ies), creative ways to generate training datasets, and careful selection of material candidates (including “virtual” candidates)
  • All known compounds can be predicted as well as “virtual” compounds –compounds that have not yet been synthesized, but are predicted to be stable and may have utility for your specific application
  • In our project model, all IP for materials belongs to you
  • Contact us today

Xaedra - How does it work?

Cognitive systems build understanding of the world by making decisions based on past experiences. Scenarios which are very different from those encountered in the past may lead to errors but provide data to enrich the knowledge base of the system. Such systems should progressively improve as the variety of situations increases. Recent developments in artificial intelligence enabled to quantify these vague notions of “understanding” or “experience” and made them mathematically precise.

Xaedra employs methods of machine learning to enable selection of candidate materials based on a simple search for materials properties specific to the application at hand. Of course, existence of data is a prerequisite for any machine learning problem, it should be available, reliable, and curated, or effort has to be put in to create that data.

The Xaedra platform consists of a database containing known inorganic chemical compounds, their composition, and associated crystallographic information. However, that experimental identification and verification of compound crystal structure does not provide any information on the specific properties it might have. Some of the available materials property data can be found in various forms ranging from peer-review journals, to material property databases. However, the information is always very sparse. Xaedra gets around lack of empirical property data by using atomic level information and a neural network to quantitatively predict previously unknown material properties. Thus, making it possible to do a properties-based search for materials, even where no empirical property data exists.

At the heart of Xaedra is an atomic structure descriptor – a data structure uniquely identifying a material by associating it with an atomic level fingerprint. The descriptor enables application of convolutional neural network and ultimately construction of a regression model that can be trained to establish a mapping between a compounds atomic structures and intensive properties of corresponding materials. This surrogate model is used to predict property values for all known materials which are then stored in the database empowering the simple search. This, finally opens a path to rational material design – a design-driven approach to materials exploration and discovery.

What academia is saying about materials and AI



Meet the innovators leading the materials revolution.

Pawel Pisarski Ph.D.

Pawel Pisarski is Manager for R&D at XAEDRA, in which role he is responsible for all development of machine learning methodology for XAEDRA’s material prediction platforms. His expertise includes artificial intelligence and high-performance computing as well as physics and chemistry. He joined Lumiant in 2013 and has over 5 years of research into machine learning applied to material prediction, resulting in the XAEDRA 1.0 and XAEDRA 2.0 platforms. Dr. Pisarski did his postdoctoral research at Queen’s University in Kingston, Ontario, and he previously studied in Poland, where he obtained a Ph.D. in theoretical physics.

Zohrab Ahmadi Ph.D.

Zohrab Ahmadi is Manager for Material Discovery Operations at XAEDRA. He has extensive experience in catalysis, materials synthesis, nanotechnology, and fluid-surface chemistry, with over six years of industrial experience as a research scientist, R&D chemist, and product development lead roles. Dr. Ahmadi has published over 30 peer-reviewed and preceding conference papers, and invented four patented and patent-pending technologies. He received his Ph.D. in chemistry from University of Victoria in 2013 with a focus on catalysis, and Postdoctoral from University of Calgary in materials science.

Charlie Baker P.E.

Charlie Baker is General Manager for XAEDRA and responsible for Business Development. He is a Registered Professional Engineer and has 40 years of experience in top engineering and marketing executive positions at Honda, General Motors, Johnson Controls, and Harley-Davidson.

What people are saying about Xaedra

Case Study

Lumiant Corporation researchers had invented TitanMade L465, which is a composite of gamma titanium aluminide and aluminum oxide created from titanium oxide and aluminum using Self-propagating high-temperature Synthesis (SHS). The Lumiant researchers wanted to understand what other composites could be created by SHS. The actual search using Xaedra is described in the accompanying video.

From this comprehensive Xaedra search one follow-on composite identified was titanium silicide with titanium carbide (Ti5Si3-TiC), which was subsequently patented and trademarked as TitanMade S by Lumiant. Xaedra correctly predicted the ability to synthesize TitanMade S using SHS, which was not intuitively obvious to the experienced researchers. As described in the video the discovery and synthesis of TitanMade S in the lab took a total of 10 days. Confirmation testing of all material properties of TitanMade S is pending, but given the nature of its composition, TitanMade S would likely exhibit ceramic-like properties. Both the specific strength and specific modulus would likely be higher than available technical ceramics. Likewise, the fracture toughness of TitanMade S would likely exceed that of the best technical ceramics. The hardness of TitanMade S would be extremely high, and using TitanMade L465 as a representative predictor, the Vickers hardness of TitanMade S would be in the range of 2800-3100, which is very close to the hardness of boron carbide (B4C).

The key benefits of TitanMade S are:

  • Very high hardness
  • High specific strength, stiffness, and fracture toughness
  • Extremely high melt point (refractory)
  • Low energy requirement of manufacturing process

Potential uses of TitanMade S include high end technical armour, especially armour that requires ultra high temperature resistance, as well as jet engine components and automotive engine components. The high hardness of TitanMade S would also provide excellent wear resistance. Within the overall scope of high-performance refractory mechanical components in wear applications, TitanMade S would likely be very cost-competitive compared to fairly limited set of current alternatives.