(242h) Blind Source Separation in Raman and ATR-FTIR Spectroscopy: A Processing Case Study

Kocevska, S., Georgia Institute of Technology
Rousseau, R. W., Georgia Institute of Technology
Grover, M., Georgia Tech
Blind Source Separation in Raman and ATR-FTIR Spectroscopy:
A Processing Case Study

Maggioni G.M., Kocevska S., Rousseau R. W., Grover M. A.

The low-level radioactive waste in the USA needs treating through vitrification to achieve long-term, safe, and environmentally sustainable storage [1].

To avoid the deterioration of the equipment, the process operating conditions must remain within a relatively narrow range of conditions. Ultimately, the key variable to monitor and control to achieve process safety and efficiency are the temperature, the species present in the feed, and their (relative) concentrations.

Spectroscopic techniques, such as Infrared (IR) (in the form of ATR-FTIR) and Raman have been long used to analyse and monitor the composition of solutions and slurries. The standard approach for the application of these techniques relies on calibration, usually a time-consuming procedure that needs to carefully design a set of experiments with known species and concentrations to estimate the model parameters.

However, each waste-containing vessel has not only been produced at different time and at different conditions, but it has also been long stored separately from the other vessels, thus experiencing different conditions and history. Additionally, if the species present in the mixture change, a new calibration typically becomes necessary, requiring to halt the process, lest one risks to obtain wrong results.

In the present work, we have developed a protocol to identify the spectra of the pure species and to compute a reliable estimate of their relative concentrations in a manner as calibration-free as possible, based on Blind Source Separation (BSS) techniques [2]. The protocol builds upon two different well-established techniques, namely Independent Component Analysis (ICA) and Multivariate Curve Resolution - Alternate Least Square (MCR-ALS).

In particular, we have investigated the application of our protocol on Raman and ATR-FTIR spectroscopy, using a simulant of the actual low-radioactive waste. We investigate the ability of BSS methods to correctly determine the spectra of unknown species in solution and discuss the reliability of one-point calibration methods to estimate the system composition.

[1] Civil Nuclear Waste Disposal, Congressional Research Service, 2018

[2] Hyvarinen, A. (2012). Independent component analysis: recent advances. Philosophical Transactions of the Royal Society A 371(1984), 20110534–20110534.