Peter Griffiths on NIR, Mid-IR, and Raman

Spectroscopy recently interviewed Dr. Peter Griffiths, Professor Emeritus at the University of Idaho on his work in NIR, Mid-IR, and Raman.

In 2004, you received the first Gerald S. Birth Award for your work in developing the procedure to calibrate a wavelength standard across the total NIR region more accurately than previously possible. Since retiring from the University of Idaho, you said that your research has focused on the development of specifications for wavelength standards. Can you tell us more about what you’re doing in this area and how you think it will affect the application of NIR?

Griffiths: Well first of all, let me just say that since the two papers that were honored by the Gerald S. Birth award, I haven’t published any papers in near infrared and we really haven’t done any research in near infrared. All my work is in the areas of mid-infrared and Raman spectroscopy albeit using many of the data processing techniques that are used in near infrared. I was a member for five years of the U.S. Pharmacopeia’s General Chapters Experts Committee and during that time I was responsible, or in part responsible, for putting together the chapters on near infrared, mid-infrared, and Raman spectroscopy. During that time, it became clear to me that the specification that the USP had put onto wave inaccuracy for their instruments was something which was essentially impossible to meet in practice because the wavelengths of the standards vary with the type of instruments and the resolution that is used for the measurements. I volunteered to look into this in some detail and I have been working with a couple of other people on trying to set down more realistic wavelength standards. This very much follows on with the papers for which we were rewarded with the Birth award.

How has the application of NIR evolved throughout your career? Have there been instrumental advances in the technique?

Griffiths: There have been enormous instrumental advances in the technique. The first time that I came across the real strengths of NIR spectroscopy was in the late 1970s. We published a couple of papers on diffuse reflection mid-infrared spectroscopy. One of the companies that was involved with NIR at that time was Technicon, and they held a small symposium for a reasonably exclusive group of people that was working both in mid-infrared and near-infrared, for a couple of days at their facility in Tarrytown, New York. This was a very, very exciting session because there were all sorts of new ideas thrown around there. This was where I first really heard about what was, at that stage, the very new technique of near-infrared reflection analysis.

One of the people attending that conference for example was Tomas Hirschfeld, who was one of the missionaries of the technique and also people like Karl Norris, who is the absolute godfather of the early developments of NIR. The early instruments that were being used at that time were largely filter instruments, although Norris did most of his work on a scanning monochromator. Since then, there have been enormous advances in both the hardware and the software.

If you look at the number of different instruments that are being used for NIR spectroscopy, it’s really amazing. You’ve got filter instruments that include tunable filters, these could be liquid crystal tunable filters or tilting filters; there are monochromaters with a scanning grating; there are polychromators with a nonscanning grating and a detector array; there are Fourier-transform NIR spectrometers, there are NIR emitting diode spectrometers. There is just a whole bunch of them that have been developed and each has its own particular strengths. So, the instrumentation for NIR, and incidentally portable instruments have been developed, has undergone enormous development.

At the same time, whereas the early NIR measurements were done by relatively simple, short chemometric programs, the development of techniques like partial least squares and support vector machines began to change things further and a lot of this was because people could not interpret NIR spectra and had to use this multivariate statistical techniques. The whole field of chemometrics has been driven largely, in my view, by the field of NIR and driven because most people can’t interpret NIR spectra. They get a bunch of data and they use a computer to get concentration information.

Click here to listen the full podcast interview.

This interview has been edited for length and clarity