Raman is a qualitative spectroscopic system specifically developed for molecular identification. However, with the advent of portable, high resolution Raman spectrometers and advanced chemometric modeling software, the use of Raman for highly accurate quantitative measurements of reactions and mixtures has been made possible.
In this article, B&W Tek's chemometric modeling software BWIQ is used to ascertain unknown concentrations of methanol in a tertiary mixture of water, acetonitrile, and methanol.
BWIQ, a multivariate analysis software package, is capable of analyzing spectral data and identifying internal relationships between spectra and sample classes or spectra and response data.
The software integrates conventional chemometric techniques, such as Principal Component Analysis (PCA) and Partial Least Squares Regression (PLSR), with new techniques such as Support Vector Machine (SVM) algorithms for non-linear datasets and adaptive iteratively reweighted Penalized Least Squares (airPLS) algorithms for automatic baseline correction.
Experiment and Results
For this experiment, B&W Tek’s portable Raman spectrometer i-Raman was selected, as it is the only system on the market with a high resolution of ~3cm-1. The spectrometer together with its TE cooled CCD and fiber optic probe was combined to the BCR100A double-pass cuvette holder.
Then, to develop a calibration curve, known acetonitrile, water, and methanol standards were prepared using 10mm disposable cuvettes as shown in Figure 1.
Figure 1. Chemometric analysis workflow
Subsequently, all measurements were obtained by setting an integration time of 7.5 seconds and 532nm excitation wavelength to ~30mW power output. In this experiment, averaging was not used and all spectra signify a single acquisition.
Then, approximately 94 spectra were obtained with varying concentrations of the three parts. The data collected thereafter was fed into the BWIQ software. Among the total of 94 spectra, 70 were arbitrarily selected for generating the calibration curve and the remaining 24 were utilized for validation.
The acquired data was preprocessed by performing an airPLS baseline correction. Then, smoothing was done using the Sovitzky-Golay technique followed by taking a Sovitzky-Golay 1st differential. Then through PLS1 regression, the calibration curve was produced.
Once the data was preprocessed and regressed, the measured and estimated concentrations were plotted to establish the model’s accuracy. Then the R2 value was determined to be greater than 0.999 for the estimated as well as the measured concentrations.
Likewise, the RMSE was determined to be less than 0.008 for both estimated and measured concentrations.
Finally, samples with unknown methanol concentrations were prepared in a mixture of water, methanol, and acetonitrile. B&W Tek’s i-Raman was used to measure these samples, which were entered into the BWIQ software. The software effectively determined the concentration utilizing the prior developed model.
From this study, it is evident that the i-Raman portable high resolution Raman spectrometer can be utilized with BWIQ software to effectively develop a quantitative model in order to ascertain the methanol concentration in tertiary mixtures of water, methanol, and acetonitrile.
This information has been sourced, reviewed and adapted from materials provided by B&W Tek.
For more information on this source, please visit B&W Tek.