Overview
In recent years there has been an emphasis on changing the way we view instrumentation and a desire to make instruments smaller and lower in cost [1-3]. This trend was motivated by the rapid growth in telecommunication technology in the late1990s. During the past two to three years, a new set of miniaturized (small scale) instrument platforms and products have been produced, including Raman, Infrared (IR and NIR), UV/Vis, and non-optical technologies like Mass Spectrometry and NMR. However, this article focuses on new technologies applied to the near infrared (NIR). In the interest of brevity, four example platforms representative of the newer enabling technologies that are being implemented will be highlighted as they have made miniaturization possible. Practically, these include “no-moving” parts systems, where “moving” relates to a macro movement, such as a mechanically scanned grating, a moving mirror, or a mechanical chopper. As a clarification, some of the systems considered provide movement at the microlevel, such as Micro Electro Mechanical Systems (MEMS) and chip-scale components, including a MEMS FTIR and DLP-based spectral engines. One of the platforms covered is completely solid state, being fabricated as a functional “spectrometer on a chip” concept. This paper will review these new technologies and will provide examples of spectra that reflect the current performance level for common applications. In addition to making systems smaller, the emphasis will be on systems that are truly low in cost, where low cost is from a few $100s up to about $5000.
Introduction: Moving Away from Traditional Thinking
For the purpose of this article, near infrared is defined as the wavelength range from 750nm to 2600nm; from the edge of the visible to the edge of the mid-IR. Traditional process and laboratory spectrometers tend to be large and expensive. There is an active movement toward smaller systems, and a trend to make the measurement at the physical location of the “sample,” whether it is in production, a manufacturing process, or within the environment. Also, the use of smaller instruments favors distributed measurement schemes (multiple locations in processes) and “remote” monitoring approaches for environmental and “field-based” measurements. To meet this need, the recent trend is toward portable and handheld measurement systems. (Note the terms used—these may be defined as “meters,” “sensors,” “scopes,” or “tools,” rather than instruments.) In the context of the pharmaceutical industry, this means that PAT schemes can be more completely embraced [4].
Current trends fit these recent enabling technologies that support miniaturization, such as MEMS, handheld computing, wireless, Bluetooth, Smart Phones, etc. These technologies can be classified as “game changers,” and, relative to the trends being discussed, we may consider game changers for measurement systems as spectral sensors (spectrally oriented detectors), and when fully integrated with electronics and optics, as “Spectral Engines.” The latter becomes the functional part of the final measurement platform.
Key Enabling Technologies for Miniaturization and Low Cost
Many enabling technologies can be cited, but for the construction of a replacement to the traditional spectrometer, the following have been the most relevant: thin film filter technologies, MEMS (as applied to spectroscopy, such as new detectors, spatial modulators and solid state thermal emitters), Lasers, LEDs and alternative light sources, fiber optic assemblies, and high performance detector arrays. Note that that the use of these newer technologies does not exclude existing technologies that are implemented within commercially available instruments.
Narrow band-pass filters, covering wavelengths from the deep-UV (220nm) to the mid-IR (15microns, 15,000nm), can be used individually or in specially combined assemblies to provide the required spectral range(s). Note that filters may be diced (cut and reduced in size) and mounted directly on or just above a detector to produce a spectrally selective detector. These may include multi-element patterned filters produced by microlithography techniques, and linear variable filters (LVF), or continuously variable filters on a single substrate. Of these, LVFs are the most versatile, and these can be configured to cover wavelength ranges from the long UV (350nm) to the mid-IR (15microns, 15,000nm).
MEMS fabrication is now used for infrared components, such as sources/emitters and detectors, and also for functional spectrometer components, including tunable Fabry-Perot (FP) filters and scanning Michelson interferometers. The latter example is discussed in more detail later as a fully functional micro FTIR spectrometer. MEMS sources and detectors are widely available for mid-infrared applications. The emitters may be pulsed to provide a mechanically free modulated IR output. Detectors, in the form of pyroelectric devices, are packaged as single, dual, and quad combinations, and also as linear arrays, nominally as 128, 256, and 512 packages. The big motivator behind the use of MEMS fabrication is that the devices are typically built from silicon, and in common with silicon computer chips, these devices are scaleable and the costs come down in volume. However, like computer chips, the upfront engineering costs are extremely high. Therefore, there needs to be a strong higher volume driver application to make such devices worthwhile. One MEMS-based application will be covered later. This application, which is driven by a high volume market, is the digital micro-mirror device (DMD), better known as the digital light projection (DLP) chip. In this case, the spectrometer is very much a secondary application, relative to volume usage.
Lasers and LEDs are a natural fit, size-wise, as efficient sources for microspectrometer systems. A wide range of tunable laser diodes (TDLs) are now available, primarily in the near-infrared spectral region, and broadly tunable mid-infrared lasers are available in the form of quantum cascade lasers (QCLs). In the broadly tunable format, these lasers essentially function as a spectrometer, where tuning ranges can be as broad as 2-3 microns wide. LEDs, while not tunable, may be obtained over a broad range of wavelengths covering from the UV (250nm) to the NIR (1450nm) and more recently out to the upper mid-IR (around 3 microns). The devices at the extremes (UV and the mid-IR) are more specialized and are expensive. However, LEDs with wavelengths from 360nm to 1450nm are relatively inexpensive and readily available, and can be packaged in multi-wavelength (multi-die) formats.
Two other important enabling technologies are fiber optics and linear photodiode arrays. Fiber optics have become a standard and widely used method for interfacing small spectrometer systems, with the first commercial implementations on small instruments being available in the early 1990s. In another format, as a fused array of fibers, devices known as fiber-faceplates are being used for optical interfacing, where better light control, such as collimation, and image reshaping/geometry conversions (such as from rectangular to circular) are required. High performance linear photodiode arrays are available for all spectrally important wavelength ranges, from the UV-visible to the mid-infrared. High resolution silicon arrays are the most widely used, and these can cover from 190nm to 1100nm, with the shorter wavelengths (190nm to 350nm) requiring special coatings. The traditional near infrared (NIR) are normally covered by InGaAs and lead salt, PbS and PbSe, arrays. The price of InGaAs arrays remains relatively high, but the quality and consistency is improved, and resolutions as high as 512 pixels are now available. However, for low cost systems128 and 256 arrays in the 900nm to 1700nm range are quite popular. Extended InGaAs arrays covering a range out to 2600nm are available, but these are high priced, and require cooling to be used for practical measurements due to a high intrinsic dark current noise. For very low cost systems it is necessary to use silicon-based arrays, where good performance, high resolution (1024 pixels or higher) arrays are available with costs as low as $10 or $20 in moderate volume.
New Technologies: Spectral Engines and Miniature Spectrometers
In combination, the enabling technologies defined above form a toolbox which is the basis of the latest generation of spectral engines and miniature spectrometers. Four example systems are now described in brief as practical implementations. Two are array-based systems, and the other two are implementations that use digital encoding (programmable DLP and the MEMS Michelson interferometer) that allow the use of a single detector, rather than an array. Note that a single detector offers a cost benefit as the high price of many commercial spectrometers is dictated by the inherent “high” cost of certain array detectors.
The selection of a detector array, whether a photo-detector or thermal detector, is based on a number of key parameters, of which spectral range, sensitivity/performance, size, and cost are among the most important. Two different arrays are considered here: the InGaAs photo-detector array (900nm to 2600nm) and the silicon photodiode array (PDA, 700nm to 1100nm). In the case of a “spectral engine,” introduced about two years ago, the choice of InGaAs was made because of the desire to make a compact and fully integrated midrange NIR spectral engine. In this context, the term “mid-range” refers to a spectral region starting around 900nm and finishing just before the “mid-IR” at 2600nm. As noted earlier, there are various options for InGaAs detectors, which include different performance related to sensitivity and associated noise characteristics. In the example discussed, the designer opted to use an un-cooled version of the detector because of the smaller footprint and reduced overhead (related to the power, size, and thermal requirements associated with a cooled detector).
In the example of the engine shown in Figure 1(A/B), the device is produced as an optimum combination of an LVF packaged in close contact to the InGaAs detector, Figure 1C. The goal in this design was to keep the package as small as possible. To that end, the detector and its associated electronics is packaged with a microprocessor to produce a true, calibrated spectral output via a USB interface, where the USB provides the power for the detector electronics and the dual light sources. A highly customized optical interface provides a lightguide conduit between the sample and the LVF-detector combination. This component serves a dual-function: it ensures that as much light as possible from the sample reaches the detector, and it provides light collimation, a requirement of the LVF to provide spectral purity for the dispersed radiation falling on the detector. This mechanical arrangement is intended to provide as close as possible 1:1 correlation between the projected image of the LVF (providing the wavelength separation) and the detector elements of the array.
Figure 1. The fully integrated micro NIR Spectrometer engine (A/B) featuring LvF technology combined with a inGaAs detector array (C) generating a spectral output (d).The spectral ranges currently covered by the latest version of this compact system are 950nm to 1650nm. An extended version covering out to ~2200nm has been discussed by the developer, and may be available in an evaluation prototype form. Example spectral data are shown in Figure 1D for a sample of powdered aspirin in a polyethylene bag, referenced against Spectralon (also in a polyethylene bag).
In the second example, Figure 2, featuring a linear detector array a new solid-state device based on a 1024 silicon diode array is discussed. The device is designed as an ultralow-cost discrete semiconductor component, based on a semiconductor-style assembly. The chipscale component, as shown, which includes the spectral separation component, imaging optics, and the detector, is currently protected by a small enclosure (Figure 2A), which is less than 20mm square. (The actual active spectrometer device is only 9mm in length.) Future manifestations of this device may include a standard electronics package, such as a 16-pin dip package. A complete system assembled for demonstration purposes, including electronics, is shown in Figure 2B. This assembly interfaces via a USB cable, which provides the power for the system, as well as the communications. Two versions of the device will be made available during summer 2014 covering the visible spectrum (400nm to 700nm) and the near infrared (sw-NIR), final range to be determined, though currently 750nm to 1050nm is being considered. Example spectral results are included in Figure 2 (C visible and D sw-NIR). While this device can be offered as a spectrometer, its real practical utility is as an embedded component in a process or a system, such as a handheld meter or instrument. One strategy could include a combination featuring two devices covering both the visible and the NIR, where color and composition may be measured simultaneously.
Figure 2. Concept prototypes (A/B) for demonstration of a solid state interferometric chip-based spectrometer with visible and/or NIR spectral output (C/d).A unique new spectrometer concept is currently available in evaluation unit format. The spectrometer is built on an existing MEMS technology that is used in digital projectors, known as a DLP “chip.” The DLP device illustrated in Figure 3A is a computer addressable array of micro-mirrors, where the mirrors effectively function as microscopic light switches when correctly illuminated. An example DLP light engine, used as the functional heart of a new generation of small portable projectors, can be adapted for this application. The micro-mirrors function as pixels and can be addressed in the same manner as pixels in a two-dimensional photodiode array. Because a spectrum is two-dimensional and can be represented as columns of pixels, it is possible to program the DLP to acquire (scan) a spectrum by illuminating the array by the spectral output of a light dispersion device (Figure 3C), such as a grating, a prism, or even an LVF. By incrementally moving the image across the DLP, it is possible to build up a spectrum from the output of a single element detector. The spectral output at the detector from the DLP has great versatility and can be controlled in terms of line width (number of pixels in a column), line height (which governs the light output for a given wavelength), and position (which enables output from individual wavelengths, like a filter function). In addition, this arrangement allows for digital encoding of a spectrum where the output can be uniquely modulated, as in the case of a Hadamard transform instrument.
Figure 3. The DLP spatial modulator – transitioning from projector engine (A/B) to spectrometer engine (C).Figure 4 shows examples of spectral output from a current prototype evaluation unit (4A), which can be easily configured for liquids (Figure 4C, spectra of seed oils) and solids (D spectra of powdered actives from OTC pharmacy products). As an example of the flexibility of the DLP, two sets of resolution are illustrated for low (~115 pixel equivalents) and high (~430 pixel equivalents) where a major benefit of the low resolution is spectral acquisition speed; complete spectra are obtained in 0.5 seconds, with a peak SNR (as high as 50,000:1). The current evaluation unit (4A) is moderately sized at approximately 4.5in x 3.5in x 1.5in. However, it is worth noting the size of the small projector module (Figure 3B) as this indicates the opportunity for size reduction; an integration of this light engine with a spectral detection module is currently being prototyped and is also configured for transmission and reflection measurements.
Figure 4. DLP prototype module (A) shown with variable data/spectral resolution (B) function and capability to switch from liquid (C) to solid (d) samples.The use of MEMS fabrication for optical measurement systems has been discussed for several years. Such systems are sometimes integrated into dedicated measurement systems—in particular those used for blood and other body fluid testing, and often include micro-fluidics. A new stand-alone system that features a MEMS-fabricated Michelson interferometer, the heart of an FTIR is illustrated in Figure 5A/B. This system was introduced internationally last December and domestically in March at Pittcon. The system is currently configured for operation in the NIR producing spectra with a range out to 1700nm, as indicated in Figure 5 (C/D). Systems will be available as evaluation units and the main optical interface will be a 200micron optical fiber, allowing the spectra of a wide range of sample types to be measured.
Figure 5. MEMS fabricated Michelson interferometer (A) integrated in a spectrometer module (B) – providing Nir spectral range (C/d).Summary
This article has reviewed new technologies that can be applied and integrated for the measurement of optical spectra, primarily in the near infrared spectral regions. The main theme here is applying technologies that enable instruments to be made smaller and in some cases, less expensive. In one case, the technology is built as an electronic chip that can be embedded as a true spectral detection measurement system that can be used without the need of a separate instrument. In this case the system electronics can become the spectrometer. A major benefit of smaller and possibly lower cost “spectrometers” is that a dynamic process may be monitored in situ, and at multiple points. Also, the systems reviewed can be used for liquids and solids. We can expect these and variants to become ubiquitous for a wide range of applications in the pharmaceutical industry and beyond. Ultimately, we may expect such systems to show up in the home as an integral component of home safety and healthcare.
- In keeping with the publisher’s wishes for a non-commercial, “unbiased” article, the names of the manufacturers/suppliers and their products that are referenced in the article have been omitted. Some of the systems involved have already been introduced into the market at some level, and so the reader may already be aware and recognize the technology, the commercial product, and/or the supplier. Because most of the systems are better described as “spectral engines” rather than final instruments, this anonymity approach works because the “System” or “Device” is most likely buried within the final measurement system. Bottom line: an understanding of the functionality of a spectral engine is more important than knowledge of the commercial vendor.
References
- Coates, J.P., A review of Current and New Technology Used in Instrumentation for Industrial Vibrational Spectroscopy, Spectroscopy, 14, (10), pp. 20-34, 1999.
- Coates, J.P., New Micro Spectrometers: Building on the Principle that Simple is Beautiful, Spectroscopy, 15, (12), 2000.
- Coates, J.P., Spectrometric and Photometric Detectors: Opening the Doors to Miniaturized Spectroscopy, Lab International, September 2008, pp 25-27.
- Ritchie, G. et al., American Pharmaceutical Review, Spectroscopic Techniques and PAT Roundtable, December 2013.
Author Biography
Dr. John Coates was educated and started his career in the UK. His first position was as an analytical chemist working for Castrol Oil Company. He graduated with the Royal Society of Chemistry (RSC) and obtained his Ph.D. in analytical chemistry at Brunel University with an initial focus on Raman Spectroscopy. Dr. Coates has 50 years of industrial experience. After moving to the USA as a Senior Staff Scientist at Perkin Elmer Corporation, Dr. Coates accepted positions at Spectra-Tech (Stamford, CT) and then at Nicolet Instruments (Madison, WI), and eventually returning to a position at PerkinElmer’s Real-Time Systems Division (Wilton, CT), a joint-venture with Dow Chemical Company (Midland, MI). In 1996 Dr. Coates opted to leave the corporate world and formed his own company, Coates Consulting LLC.
Since 1996 he has built Coates Consulting LLC, a network consultancy focused on applied and industrial instrumentation, optical spectroscopy, and analytical instruments and sensors for dedicated applications. His main focus for the company is on instrument miniaturization and spectral sensors. Dr. Coates has devised and developed more than 50 different instrument and sensor products for dedicated analyses and has worked with major corporations in major industry sectors; including pharma, environmental, industrial (chemical and consumer products), aerospace, computer technology, and medical. He was co-founder of two businesses for the development of products based on miniaturization: Sentelligence and microSpectral Sensors. In addition, Dr. Coates held Director positions at Global Technovations, Inc. (now On-Site Analysis, Inc.), a company that markets field-based oil analyzers, and MCEC (Measurement and Control Engineering Center) at the University of Tennessee.