As applications for optical spectroscopy expand, instruments keep pace.

Applications of optical spectroscopy are growing, providing more signals, allowing users to detect more, see more and do more. Images: P&P OpticaInnovations in optical spectroscopy have helped the technology reach a point where performance previously seen only in laboratory settings can be obtained in the field with compact and easy-to-use systems. These improvements, made to detectors, software and overall design, have greatly affected instrument characteristics such as speed, miniaturization, price and reliability.

As a result, optical spectroscopy has become important for many new applications. For example, short-wavelength infrared (SWIR) systems have proven highly effective in sorting, grading and quality control for material identification and continuous monitoring of a production line. These systems provide the ability to use faster, wider conveyor belts, or inspect more complex and varied materials. Multichannel systems, such as the HyperChannel technology used in the 2012 R&D 100 Award-winning Hyperspectral Multichannel Spectrometer from P&P Optica, Waterloo, Ont., are ideally suited for applications where spectrally similar materials can be distinguished and sub-pixel-scale information can be extracted in the presence of strong background signal. The small form factor and flexibility of this type of spectroscopy system allows for bedside examination of patients repetitively or monitored continuously for a long period of time. Spectroscopy systems are providing more signal, allowing users to detect more, see more and do more.

Furthermore, access to improved components, such as collection optics including fibers, diffractive elements, detectors and data-processing algorithms, improve practical application of spectroscopy and reduce cost as compared to some other sensing and imaging technologies.

Market pressures
Recent trends in market forces are encouraging to the growth of optical spectroscopy and its applications. A noteworthy trend in the market is the general globalization of resources and the centralization of processing of raw materials. For example, just a few companies, such as TOMRA Systems, Key Technology or John Bean Technologies, are responsible for the processing of a significant amount of the food produced globally. Food-processing companies acquire food from multiple sources, including large multi-national farms and small operations. Thus, grading of raw materials, in this case food, is required to achieve efficiencies of production. To do so in a fast and reliable way is critical. Optical spectrometers offer the ability to quickly inspect and grade incoming food products by size, origin and consideration of the chemical composition of food such as starch, water or sugar content.

Another industry where centralization has occurred is oil refining. Only about 170 refineries were functioning in North America in 2008, with a total of 700 around the world. These refineries process oil from many sources, and although they tend to be specialized, the source of crude dictates refining processes. Most of the oil is transported by trains and pipelines, and tends to come in many varieties, chemical compositions and complexity. An increasing need exists to analyze oil prior to its arrival at the refinery. One method of analysis is to introduce optical spectroscopy, from Fourier transform infrared (FT-IR) to near infrared (NIR). This allows for minimal handling of samples and quick analysis, unlike the more complex laboratory instruments such as gas or liquid chromatography combined with mass spectroscopy.

Technology highlights
Optical spectroscopy, as a traditionally multidisciplinary science, has lagged behind other technologies in the development of sensitivity, size reduction and portability. However, in the last two decades, this trend has reversed. The pursuit of novel manufacturing methods of diffractive optical elements, including microelectromechanical systems (MEMS) and reduction in the cost of high-quality lasers, have led to the creation of new types of gratings, as well as to significant enhancement of existing grating manufacturing methods to reduce costs and increase efficiencies. Better detectors, as well as improved mass production of optical elements, have led to a steady decrease in the size of optical spectrometers, with many companies, such as Ocean Optics, Thermo Fisher Scientific or P&P Optica, producing smaller, portable systems. Even handheld units, which are still limited by the amount of light that can be collected by small optical elements, have been getting steadily better by taking advantage of lower noise, faster charge-coupled device (CCD) and complementary metal–oxide–semiconductor (CMOS) detectors.

Spectrometers are no longer confined to laboratories; the decrease in the size of optical spectrometers has resulted in producing smaller, robust, portable systems including some handheld units. For example, many traditional spectrometers were limited by poor optical elements and light scattering on reflective gratings and curved mirrors. The mass-production of better types of gratings, such as transmission-based gel gratings, has enabled the same size of spectrometers to be limited by detector noise. That means the same-sized optical spectrometer, previously limited by its optical construction to a signal-to-noise ratio (SNR) of 300:1, can now be expected to stretch a linear detector to reach a SNR of well over 3000:1.

Other detector technologies have also made a tremendous impact on optical spectrometers. Traditional technologies such as CCD and CMOS cannot extend beyond about 1,050 nm. Better manufacturing processes of indium gallium arsenide (InGaAs)-based detectors have led to significant cost reduction in NIR spectroscopy, as well as SWIR up to 2,200 nm. Mercury cadmium telluride (MCT) can further extend the spectral range to 2,500 nm, allowing identification of materials which were traditionally difficult to analyze using visible spectroscopy.

The introduction of 2-D area detector-based spectrometers for non-laboratory use has also greatly improved the sensitivity of portable and industrial systems. A 12-mm slit for example, oriented along the height of the detector by binning the columns of the pixels vertically, can be used in some systems to collect signal, improving the SNR by a factor of almost five as compared to the linear detector arrays, even if everything else is identical. This stems from the fact that SNR improves by a square root of available measurement elements (in this case 0.5-mm-high detector rows). A high-performance optical spectrometer can now reach an SNR of well above 100,000:1 in the same acquisition time as typical systems achieving an SNR of 1000:1. Such a spectrometer can therefore detect materials which have 100 times smaller concentrations, or take 1/100th of the time to acquire identical spectrum as its less capable cousin.

Large-slit 2-D area detector-based spectrometers can also be used as multichannel systems or hyperspectral imagers with high spectral resolution of hundreds of color bands instead of traditionally tens of bands available with other imaging technologies. It’s now possible to mount a slit-based system above a fast-moving conveyor belt, or even on an airplane to inspect multiple objects at the same time. Thus, in any application which requires chemical differentiation of substances, a single system can be used continuously to inspect the entire process. This is different from the traditional use of spectrometers as point-measuring devices only.

The final development which will continue to impact the growth of optical spectroscopy is the constant evolution of processing algorithms used for spectral analysis. Chemometrics allows the transformation of large amounts of spectral information obtained from spectrometers into information used for decision making, such as actual chemical composition or other properties of interest. Methods such as principal component analysis (PCA) and partial least square (PLS) regression are now supplemented with new methodologies, such as genetic algorithms, wavelet analysis and even sparse data methods for spectral pattern recognition.