We provide the solution
and put control in your hands.

All materials, natural or man-made, bear unique traces of their origin and history. The traces are typically minute, represented across thousands of variables. Materialytics Analysis (MA) detects these traces and uses them to characterize materials in well-classified groups e.g. components produced by XYZ Company, parts meeting quality standards, or minerals taken from Mine A. After a group has been statistically identified, materials with unknown histories can be analyzed and definitively matched or rejected as unknown.  The process is fast, minimally invasive and extremely effective, giving you the knowledge you need to reduce waste and increase confidence in the quality and value of your products. 

 

This process can be considered in three stages:

  1. Reference Collection
  2. Reference Database
  3. Material Classification

 

 

The first step in MA is to create a Reference Collection – collecting and carefully documenting samples with known provenance and history to use in characterizing the specified group. The Collection must be cataloged and meticulously maintained to assure the integrity of the results.

 

A Reference Collection is created by going into the factory or the field, and gathering and documenting samples (parts or materials) that represent the desired group e.g. good parts, legitimate components or minerals from Mine A . The resulting data from this group is what all other samples will be compared to. The size of a Reference Collection can vary, but the accuracy of Quantagenetics® classification increases with the size of the Reference Database it is accessing. This means that a small Reference Collection may be enough to answer specific questions. However, if you want to know not just whether the sample did or did not come from site A, but if it didn’t, did it come from sites B, C, D or E – you’ll need a broader Reference Collection that includes materials from those sites.

 

Reference Database

 

Physical analysis of all samples is performed by the Materialytics Sequencing Station (M2S®), which produces the digital data used in the final Classification step. Analysis is usually completed in minutes by a trained technician. It involves simple prepping of the sample, usually no more than washing it with isopropyl alcohol, placing it in the analysis chamber of our customized spectrometer, running a series of test shots, removing it, and preparing the resulting information for our classification technique.

 

This process is simple; however, it produces an immense amount of useful data for characterizing the sample. The Materialytics Sequencing Station (M2S) uses a spectroscopic technique to produce approximately 2.5 million data points per sample. While the basic spectroscopic technique is well known, the specifics of the equipment and procedures Materialytics has developed are distinctly different. Physical analysis with M2S produces data that is full of noise, interference, and artifacts of the testing process. No attempt is made to clean up the data, or take averages to reduce processing.  All of the data are preserved and processed in our final step, Classification.

 

Material Classification

 

In Step 3 the data produced during analysis is processed using Quantagenetics to classify the sample. Quantagenetics allows us to use all (approximately 2.5 million points) of the information generated by spectral analysis to reveal each material’s unique chemical and structural signature. By combining all of these signatures, we are able to characterize groups. Using Quantagenetics in cluster analysis allows samples of material to be classified with high accuracy, e.g.: good or bad, legitimate or counterfeit, from Mine A or Mine B.

 

The classification process begins with examining samples from a Reference Collection containing well-documented materials. Quantagenetics places the digital signatures it creates from each sample into a Reference Database. A signature is then created for the group as a whole. These processed signatures represent large amounts of data; we routinely handle terabyte files. The materials represented in each of these databases is designated as one “class” of material. Once the database has been developed, samples with unknown histories can then be analyzed. A digital signature is generated from an unclassified sample and then compared to the Reference Database. It then reports how close of a match the sample is.

 

This is sharply different from the usual practice of struggling to reduce the amount of data that must be processed by disposing of what appears to be “noise,” and removing “unimportant” data. These actions typically reduce the amount of data by more than 99%, so that available computing resources are able to process what remains. Human judgment determines what is “unimportant.” Previously, experts have always picked the data to pieces, seeking significant features that they understand, and can manipulate handily.  Unfortunately, there has never been a way of knowing what’s really worth keeping and what can sensibly be ignored. There has never been way to handle “big data.”


Quantagenetics® allows us to manage big data. In Materialytics Analysis the analysis algorithms use all of the data. Nothing is disregarded, because the critical information is contained in thousands of variables, whose individual significance has never been determined. Quantagenetics processing is able to deal with the vast amount of data with readily available computing resources. This allows it to compare real data from reference samples directly to real data from unknown samples, instead of comparing expert interpretations of more limited data. Quantagenetics classification on average uses over 1300 times more data than traditional chemical analysis techniques.

 

This approach allows Materialytics Analysis to classify materials with extraordinary accuracy. The process reveals information about some materials that has not previously been detected.