July 1999

Solid-state lasers, holographic optics, and cooled CCDs have enabled the development of a new generation of rugged Raman process analyzers.

Many industries critically depend on the ability to make in-process chemical composition measurements of both end products and intermediates. Typical examples of this range from determining the crystallinity of polymer sheets to measuring gasoline octane. Such quantitative monitoring not only allows manufacturers to optimize production processes, but can provide direct feedback to ensure key parameters remain at specified values.

Unfortunately, most of these measurements have traditionally entailed off-line analysis of captured samples in a plant's analytical laboratory. The most widely used techniques involve GC (gas chromatography) and HPLC (high-pressure liquid chromatography). These may require overall measurement times of up to one hour, which can have a huge cost impact on a process. Indeed, the holy grail of process monitoring has always been real-time, in-situ measurements. With recent advances in photonic technology, Raman spectroscopy is uniquely able to finally deliver this. As a result, surprisingly, industrial Raman instrumentation is now one of the fastest-growing areas of analytical instrumentation, already in use at such giants as Dow Corning, ICI, DuPont, Monsanto, Akzo Nobel, Exxon, and British Petroleum.

Figure 1. Like infrared absorption, Raman spectroscopy is usually used to probe signature vibration resonances of molecules. In the case of Raman, though, only part of the original photon energy is absorbed, leaving a lower-energy photon.

Raman Spectroscopy
Raman scattering was first discovered in 1928 but for many years enjoyed only limited popularity as a laboratory tool for physical chemists. In Raman scattering, a high-energy photon causes vibrational excitation of a molecule; the photon is essentially re-emitted with its energy reduced by this vibrational energy quantum (see Figure 1). If Raman-scattered light is dispersed in a spectrometer, it is seen to consist of spectral peaks and lines that are shifted from the original wavelength by characteristic amounts.

In order to ensure sharp spectra, it is important that the exciting light be monochromatic. Given that Raman is a low-probability effect, the excitation source must also be very bright, so that there is sufficient returned signal. Together, these requirements make the laser an ideal Raman excitation source. In theory, this laser can be at any arbitrary fixed wavelength, but in practice certain spectral regions are preferred. (Shorter visible wavelengths give more intense signals, because Raman intensity scales with (photon energy)⁴. Near-infrared excitation is used with samples where visible Raman could be obscured by sample fluorescence.)

Raman offers a number of advantages for on-line process control over other spectroscopic techniques, such as FTIR (Fourier transform infrared absorption) and NIR (near-infrared absorption). For example, Raman provides clean, easily resolved spectral features that can be readily related to composition. Raman can be performed using visible light, which can be transmitted through low-cost glass fibers, allowing for remote measurements in a plant. Also, by relying on backscattered Raman, the sampling setup can be very simple. Using a simple imaging lens or microscope objective, noncontact measurements can be made through glass windows and bottles, for example. Packaged pharmaceuticals can even be analyzed in plastic blister packs. Furthermore, even dilute solutes in aqueous solutions can be detected because water itself produces only a weak Raman response.

Turnkey Rugged Spectrometers

Given these advantages, why did it take over 60 years for Raman to finally begin gaining acceptance for direct process control? The answer is very simple. The evolution of photonic technology has recently enabled a revolution in the way practical Raman is implemented, in terms of both performance and ease of use.

In the past, Raman required an expensive and bulky water-cooled laser. A large triple monochromator was used to scan the spectrum and separate the weak Raman signal from scattered laser light. The entire setup would fill a large optical table, consume lots of electrical power and water, and require constant skilled realignment.

The past ten years, however, have seen the maturation of three key technologies: compact, solid-state visible and near-infrared lasers for excitation, highly efficient holographic notch filters to remove scattered laser light, and cooled, low-noise CCD detectors to enable simultaneous detection of the entire Raman spectrum. Together, these enabled Raman spectrometers to be packaged as turnkey benchtop instruments that are small, rugged, and simple to operate. Just as important, the combination of efficient elimination of laser scatter and low-noise detection enables these instruments to detect materials down to the parts-per-million level. In addition, because these spectrometers are all-solid-state, with no moving parts, they can also be packaged in industrial NEMA configurations as on-line process analyzers. The photo at the head of this article shows a typical example: the Kaiser HoloProbe. One of the most important features of these fiber-coupled analyzers is the incorporation of a rugged multichannel fiber sequencer, which allows the software to sample up to eight different points in a plant on demand.

Applications: Improving PET Products at ICI

Raman is now being used both to improve processes as well as to monitor and control in real time. As with any new process monitoring/control technique, the first step is to use Raman in the lab to understand precisely how the data relates to product composition and quality factors. For many years ICI has conducted this type of research at its Wilton Research Centre (Middlesborough, UK), for example on PET poly(ethylene terephthalate). This is one of the most important polyestermer materials, used to make everything from textiles to soft-drink bottles and touch keyboards.

The mechanical properties of PET products are strongly dependent on the morphology of the PET material. This includes the extent of crystallinity and the degree to which the polymer chains are aligned in a particular axis or plane. These are ultimately determined by production parameters such as melt temperature, draw speed, cooling rate, etc. In blow-molded products, for example, local PET morphology can vary greatly throughout the product, an article by Neil Everall, a business research associate at ICI, explains. It has been known for some time that polymer crystallinity and orientation could be measured from the intensity of key spectral peaks, and that orientation could be deduced by measuring spectral intensity as a function of laser and signal polarization.

Monitoring Chlorosilane Production at Dow Corning

An application that has already proved its value in the on-line manufacturing environment is the use of Raman to monitor chlorosilane production at Dow Corning. Methylchlorosilanes (MexSiCly) are important intermediates in the industrial production of silicones. The problem is that these are generated by the catalytic reaction of silicon and chloromethane, which produces a mixture of chlorosilanes and methylchlorosilanes. These must then be separated and purified by sequential distillation. Since separation is costly in terms of both time and energy, it should be monitored in real time to maximize plant efficiency.

Until recently, the intermediate and final process streams in this separation were monitored by extracting hot chlorosilane samples from the process streams and using GC analysis in the laboratory. This extraction was less than simple because the hot chlorosilanes react with moisture in the air, producing corrosive HCl. Furthermore, overall sampling and analysis times were as long as 50 minutes.

Figure 3. The Raman data at Dow Corning revealed a system oscillation that GC had been too slow to discern. Data courtesy of Dow Corning.

Then a group led by Ronda Grosse and Elmer D. Lipp at Dow Corning's Process Research Department showed that a single Raman analyzer can replace several GC instruments. More importantly, it can make true on-line measurements with acquisition times as short as 20 seconds. Also, the measurement can be made through a window in the process stream, thereby avoiding air contact with the sample. The power of Raman's ability to make real-time measurements is demonstrated in Figure 3. This data from one of Dow Corning's pilot plant-separation columns shows a fast cycling, which the slower sampling of GC had never detected. This revelation was a significant factor in convincing the production engineers to adopt Raman in the plant. Another advantage noted by the Dow group is overall economics. Explains Grosse, "While GC instruments can be relatively inexpensive, the fact is that Raman doesn't require the expensive consumables associated with GC [pure carrier gases, replacement columns and septa]. In addition, by using a multichannel Raman analyzer (Kaiser HoloProbe), we replace four GC instruments with a single Raman instrument."

Conclusion

For many years, Raman spectroscopy remained a relatively minor laboratory technique. Now the convergence of several important photonic technologies has enabled Raman instrumentation to be packaged as a compact, cost-effective turnkey system. As more engineers come to realize the value and power of Raman, it will continue to gain prominence in such diverse fields as pharmaceuticals, petrochemicals, polymers, biotechnology, and environmental monitoring.

For more information, contact the authors of this article, Michael J. Pelletier, Senior Scientist, and Will K. Kowalchyk, Applications Scientist, Kaiser Optical Systems, Spectroscopy Products, Ann Arbor, MI. E-mail: will@kosi.com.

 

Top