Pauline Leary is a CBRNE SME at Noble, specializing in miniature and portable spectroscopic instruments.

Pauline E. Leary

Portable Raman spectrometers have become smaller over the last 20 years, while their performance has increased. This has been made possible by closer coupling of all the components, use of transmission gratings rather than reflection gratings, and general advances in electronics, displays, and battery technologies. An obvious question to ask is whether this trend can continue. This paper describes the technologies and evolution of these instruments, existing limitations, the current landscape of miniature Raman spectrometers, and the state of the art. Finally, the paper also looks at what emerging technologies could be applied in this area, and how those could lead to new applications.

The rationale for portable spectroscopic instrumentation is now very familiar—allowing the spectrometer to be taken to the sample, as opposed to bringing the sample to the spectrometer, thereby moving the laboratory to the point of need, the location of the sample (1). Portable instruments provide the ability to make informed decisions on the spot, and by rapidly delivering actionable results where and when needed, you change or transform the way the customer does business, whether that customer be a Hazardous Materials (HAZMAT) technician, a scrap metal dealer, someone in pharmaceutical manufacturing, or a member of the military. These capabilities are made possible by “point-and-shoot” sampling interfaces (2) and integrated data systems with appropriate databases, spectral libraries, calibrations, and algorithms (3–6). Continuing developmental drives include the desire to reduce the size, weight, and power consumption (SWaP) of these instruments, expand the available databases, improve identification algorithms (especially for mixtures), improve detection and identification limits, and, in the case of portable Raman instruments, mitigate the effects of fluorescence from real-world samples. A relevant question, then, is “How small can a Raman spectrometer become, and still be useful for field analyses?” To address that question, it’s useful to start with a brief description of the evolution of Raman spectrometers as a whole. In all the discussion, it is recognized that portable Raman instruments do not have the performance (spectral range, resolution, signal-to-noise ratio [S/N], and sampling versatility) of their laboratory counterparts, but they do provide extremely valuable screening of samples, so as to inform the next steps to be taken.

The Evolution of Laboratory Raman Instruments

Figure 1 highlights some of the significant milestones in the evolution of Raman spectrometers, starting in the early 1950s with the introduction of the Cary 81, the first fully-integrated, highly optimized, scanning (double monochromator) Raman instrument. The introduction of laser excitation in the 1960s (7,8), supplanting mercury arcs, made laboratory instruments considerably more friendly, but instruments of that period were still scanning (double and triple monochromators). It took until the 1990s for several technologies—diode lasers (9), array detectors (10), holographic optical elements and efficient laser rejection filters (11), and fiber optic probes (12)— to come together (13–16) to make what we would recognize today as a modern Raman spectrometer (17–20). Those technologies, along with developments in consumer electronics, enabled the next step—the miniaturization of Raman spectrometers, using single-stage spectrographs, and the possibility of making them portable.

FIGURE 1: The evolution of fully-integrated handheld Raman spectrometers and their key technologies. The 1990s saw the introduction of technologies that make portable Raman instruments possible, with the first fully-integrated handheld 785 nm excitation systems introduced in 2005. Since that time, the instruments have shrunk in size while improving their performance, and several handheld 1064 nm excitation instruments have been introduced.

Technologies for, and the Evolution of, Handheld Raman Instruments

The first stage in the development of a field-portable instrument is often the incremental step of shrinking a laboratory instrument down to the size of a briefcase (21,22). Some of these are described in early surveys of portable Raman instruments (23–25), weighing about 10 pounds, and occupying volumes around 60,000 cm

). A different approach is to start from scratch with miniature components, and build up an instrument. One example used a multi-mode solid-state diode laser, giving an eventual spectroscopic resolution of about 30 cm

 (26,27). Another product was based on as-small-an-enclosure design as possible, and a handle was placed on it, along with a wireless link to a laptop. The second stage, in 2005, was the introduction of the first fully-integrated, portable Raman spectrometer, engineered and optimized for its intended market. This was from Ahura Scientific, and was designed for use by first responders and hazardous materials personnel. It represented a considerable reduction in SWaP, being about 30 cm x 15 cm x 7.6 cm in size and weighing 1.8 kg. The resulting volume (~3420 cm

) is a factor of 20 less than the briefcase-sized instruments of just a few years earlier. A third thread has been development of instruments for planetary exploration (Mars rovers), where obviously SWaP are major considerations, and here the spatial heterodyne spectroscopy design has been implemented (28,29), but, to the author’s knowledge, this technique has not been applied to commercial field instruments. Finally, Fourier transform (FT) Raman instruments (30–32) have not made much impact in the portable sphere. A briefcase-sized instrument, weighing about 12.7 kg and using 1064 nm excitation, was introduced as early as 2006, and there was a recent report (33) of an unnamed handheld FT-Raman instrument used for street drug identification.

Today, the technologies for handheld and portable Raman instruments are well understood (34,35), instruments have been critically reviewed (36–38), and they are available from many vendors. As well as continuing efforts to shrink the SWaP, recent developments have focused on four main areas: expansion of spectral libraries; development and implementation of improved mixture search algorithms; fluorescence mitigation; and trace detection.

Expansion of spectral libraries has been most active in the field of narcotics, where novel “new psychoactive substances” (39,40) materials appear “on the street” with alarming frequency. The case of determining likely constituents from a single spectrum (that is, one obtained by a portable spectrometer) is a difficult challenge. Over the years, a number of different algorithms have been developed and described in the scientific literature (41–47), but individual manufacturers may not describe their particular algorithms or their implementation in detail.

Obscuration of the Raman signal by fluorescence in the case of real-world, colored, and mineral samples is obviously very well known, and in the limiting case, shot noise from the fluorescence can completely obscure the Raman signal. Many approaches have been used to mitigate fluorescence (48), including baseline correction and other mathematical techniques (49–53), using long-wavelength excitation (for example, 1064 nm), using deep-ultraviolet (UV) excitation (shorter than 250 nm), sequentially shifted excitation (54–57), time-gating (58,59), and, in some cases, combinations of these techniques (60). One portable instrument implements a version of the sequentially-shifted method (61), and dual laser packages are now commercially available (62,63). Time-gated approaches using single photon avalanche diode (SPAD) detectors are discussed in a following section; a laboratory instrument implementing this principle is commercially available (64), but not yet a portable or handheld instrument (65,66).

Most vendors of handheld Raman instruments also supply surface-enhanced Raman spectroscopy (SERS) substrates, which not only enhance some Raman signals, but can also act to quench fluorescence. SERS is a huge field in and of itself, and the reader is referred to the chapter by Hargreaves (35) for its use in portable instruments. One commercial product (67) incorporates the ability to “read” automated colorimetric swab tests (68), providing a trace detection capability.

Figure 2 shows the size and weight of typical portable Raman instruments available today, comparing them with the 2005 Ahura product. What is remarkable is that the size of these instruments has diminished, while their performance—S/N for the same collection time and resolution—has increased. The first-generation portable Raman instruments, typified by the Ahura Scientific TruDefender and TruScan, were based on a reflective Čzerny-Turner (single-stage) design with fiber-coupled components. Second-generation instruments eliminated fiber coupling, were significantly smaller, and were more tightly integrated, using free-space optical coupling. These instruments improved the S/N (with the same range and resolution) by about a factor of 5 over their predecessors. The current generation of instruments are a little larger than a pack of playing cards, with yet improved S/N, possibly as much as a factor of 10, made possible via transmission grating designs (69). Those instruments using 1064 nm excitation are significantly larger than their 785 nm counterparts, which is due to several factors, including the physically larger size of InGaAs detectors, compared to Si-based charge-coupled device (CCD) and complementary metal–oxide–semiconductor (CMOS) arrays.

FIGURE 2: Size and weight of handheld Raman instruments from 2005. The purple marker in the top right represents the first fully-integrated system. The two yellow markers on the bottom left represent spectrometers that do not have an embedded data system, but communicate to a smartphone for control, data display and work-up. The earlier briefcase-sized instruments are not included in this plot.

For a stand-alone, completely-integrated instrument with its own data system, there is a practical minimum size, defined by the need for a display and controls. Many of these instruments will be operated by personnel wearing protective gear, implying limited visibility, and wearing gloves (70). Therefore, touchscreen displays will not work; raised molded buttons are used for control. In addition, there is often a need for more versatile sampling (such as sample presentation to the spectrometer) than point-and-shoot. For an instrument in the size range of 20 x 15 x 10 cm, sampling accessories can be configured in a “snap-on” mode, for instance, for liquid vials, a right-angle (or downward-looking) adapter, a fiber-optic probe, or a modest-distance stand-off adapter. These all require a significant-sized port to which to attach (at least 1 x 1 cm).

To minimize the size of a portable Raman spectrometer further, the approach has been to use a smartphone as the data system. This enables instruments as small as 11.2 x 3.9 x 3.4 cm from one vendor, and 6.3 x 3.9 x 1.7 cm from a second vendor, for a volume of 42 cm

, weighing just 63 g. This represents a size reduction of more than 1000 over the first “portable” instruments. That latter instrument delivers 16–19 cm

 resolution over a range of 400–2300 cm

 Raman shift, and is positioned as a “SERS reader” (71). One proposed configuration resembles a smart car key, and could therefore be used as a “wearable,” which can be used for clandestine work.

A configuration where remote control of an instrument is useful, even if it is self-contained with its own data system, is when it is attached to a robot—for instance, the new generation with “doglike” motion capabilities. This configuration can be used to explore locations containing suspected explosives, chemical agents, or other toxic materials. These instruments need to be able to communicate back to a position outside the hot zone.

Emerging and Future Technologies for Miniature Raman Spectrometers

Excitation of Raman spectra in the deep-UV has potential advantages, including fluorescence avoidance, ability to use solar-blind detectors, and a large increase in Raman scattering efficiency due to its υ

 dependence. Balanced against that are the immaturity of compact solid-state sources operating at shorter than 250 m, the lifetime of components, and eye safety considerations. The first instrument of this type has just been announced (72).

The development of stand-off Raman instruments has been ongoing for at least 20 years, driven by both military and planetary exploration missions (73–75). As noted above, stand-off adapters are available for some portable Raman systems, and a dedicated instrument, operating at up to a 2 m distance, is commercially available (76).

One possibility for further size reduction, while maintaining an integrated design, is to use the camera in a cellphone, and an effort in that area has been described (77). The project was called “spectrometer on a phone” (SOAP), and was designed using commercial-off-the-shelf (COTS) components. The system measured 4.0 x 4.0 x 1.4 cm, and was coupled directly to the cell phone detector camera optics. However, it achieved roughly only ~40 cm

 resolution. Nonetheless, it indicates what is possible, bearing in mind that some applications may only require low resolution to perform a quick screening of a sample in the field.

Single Photon Counting Avalanche Diodes (SPADs)

Detector arrays are available using single photon avalanche photodiode (SPAD) technology (78). These detectors can produce a time-based record of arriving photons, on a time scale of a nanosecond or less. They are well-suited to low light level applications, like Raman spectroscopy, and a significant number of publications have investigated their development and use (79,80). A SPAD array has already been used in a commercial Raman spectrometer (62), and its use permits the synchronization of opening a gate on the detector to the pulse of a laser (81,82). Because Raman scattering is instantaneous, and fluorescence is slightly delayed (83,84), this provides a method for discriminating against fluorescence while using visible light excitation (for example, 532 nm), without any mathematical processing (85). Commercial SPAD arrays are available in, for instance, linear (320 x 1) and 2D (320 x 320) configurations (86).

In 2022, NASA awarded a research program to Physical Sciences, Inc., entitled Photonic Integrated Raman Spectrometer (PIRS), to develop an on-chip Raman spectrometer (87,88). The project abstract states that it “will use a silicon nitride photonics platform to develop a dual-stage spectrometer that enables simultaneous high-bandwidth and high-resolution spectroscopy with direct readout. The overall form factor of the spectrometer will be less than a square centimeter while retaining spectral resolution of better than 0.2 nm” and that “the technology developed within this program can find application for material characterization in landing vehicles, plume sampling craft, and satellite-based scientific sensors. Specific missions that would benefit from PIRS include Goddard’s Ocean Worlds Science Exploration and Analogs (OSEAN) and the Mars Exploration Rover.”

The question is “How is the current miniature size achieved, and can we go any smaller?” From an optical design point of view, it is a matter of throughput—AΩ, the product of the limiting area and the solid angle collected—spectral coverage and of spectroscopic resolution, combined with what is commercially available (for example, array detectors) at a reasonable cost. The reader is referred to specialized articles (89) and monographs (90) on these topics, but we can briefly outline some of the issues.

Raman spectrographs, suitable for integration into portable equipment, are available from several vendors, with a resolution of ~9 cm

 for just a spectrograph, to around 1700 cm

, including a laser and probe. But these are significantly larger than the smallest systems on the market today.

To make a smaller instrument, every item must be addressed, and compromises made with spectral resolution, Raman shift coverage, and optical throughput. In design terms, this means looking at the detector selection, the focal length and f/number of the optics, and the dispersion (lines per mm) of the grating. These issues were described in detail in the book by McCreery (18), in the context of laboratory instruments. Are there any other approaches or “tricks” that can help? In a portable, open-beam instrument, the laser intensity can’t be increased due to eye-safety regulations. “Photonic” or “plasmonic” approaches (91,92) may not help, as they rely on bright sources. Echelle spectrographs promise the combination of high resolution and broad spectra range, but are complex designs (93). Detector cooling can reduce dark current, but may not be practical in terms of power consumption and heat dissipation in a miniature device. Finally, a non-destructive readout might help for low-light applications (like astronomy [94] and Raman spectroscopy).

The critical commercial component is the detector, typically a line sensor, so it makes sense to start there and work backwards. A typical CMOS detector for spectroscopy today has 2048 pixels, with a size of 14 μm x 200 μm, with a total detector package width of 42 mm—quite large in our context—and a version with 1024 pixels (28 mm width) is the same overall package size. Both would probably provide more spectral resolution than is required for a miniature sample identification device. In very round numbers, a 512-element linear sensor would result in 4 cm

 Raman shift coverage, giving an effective spectral resolution of ~12 cm

 varies over the coverage region in a grating-based instrument). Utilizing a 512-element sensor, with 25 μm pixel pitch, brings the detector package width down to less than 2.5 cm. Using the standard equations relating dispersion, focal length, and grating groove density, if we chose 2.5 cm focal length, then we would require a grating with ~2000 lines/mm for the desired spectral coverage, well within the state of the art. Therefore, a spectrograph with 2.5 cm internal dimensions is quite achievable, and it then remains to select the appropriate input slit and to throughput-match the collection optics.

The ultimate limitation in Raman spectrometer size is getting sufficient scattered photons to the detector to collect a reasonable signal-to-noise spectrum in a reasonable time. But what is reasonable? For spectroscopic signal to noise, a factor of 100, measured as height of the strongest band to the baseline noise in a featureless region, is more than enough for qualitative (sample identification) analyses. For a point-and-shoot application, a few seconds measurement time is practical when the instrument is actually handheld, but longer measurement times, up to a minute, are acceptable where the instrument is placed on a surface. If the optical throughput of a miniature spectrograph drops so that its typical sample measurement time is unacceptable, then it would not be a marketable product. The use of the SERS effect can clearly reduce sample measurement times or enable lower detection limits.

Smaller instruments imply lower costs, and that can lead to wider deployments in existing applications, plus new applications. Just a few are mentioned here.

Stand-off Raman, and the integration of portable Raman instruments with “doglike” robots (for cases involving explosives or suspected chemical weapons) have already been mentioned. The major law enforcement applications concern drugs (95) and explosives, and we can expect growth there. A related field in the public health arena is “community drug checking” (96), providing information to health care providers about novel substances in an illicit drug market, and promoting improved collaboration between local health care providers and law enforcement. Detection of counterfeit drugs (97) is another public health issue, especially in low- and middle-income countries (LMICs), where the cost of instrumentation is a barrier to wide usage; lower cost portable Raman spectrometers would aid considerably there.

Portable Raman instruments can be used in the laboratory as safe, non-destructive, and non-invasive screening tools, both to determine the confirmatory tests to use (such as gas chromatography–mass spectrometry [GC–MS] and liquid chromatography–mass spectrometry [LC–MS]), and to eliminate the use of wet chemistry techniques. Bodily fluid identification and gunshot residue characterization are possible emerging fields (98). Bodily fluid identification may require extensive development to encompass all types of fluid, variable human populations, and interference from substrates. Finally, a possible application is the investigation of war crimes, where the scene may not be able to be secured, and evidence needs to be gathered rapidly under adverse circumstances.

Current generation, fully integrated, portable Raman spectrometers are currently available as small as around 10 cm x 10 cm x 5 cm, weighing less than 1 kg, and capable of supporting a variety of sampling attachments. It may be difficult to reduce the size still further, while maintaining the same spectral coverage, spectral resolution, and spectroscopic signal-to-noise. However, spectrometers relying on an external data system (such as a smartphone) can be smaller—significantly so if compromises are made in those three key parameters. The smallest instrument available today is positioned as a SERS reader, taking advantage of the much larger signal obtained using that technique. With smaller size and lower costs, we can expect a rapid expansion in the applications of these instruments (1).

(1) Crocombe, R. A.; Leary, P. E.; Kammrath, B. W. 

The Role of Applications in Portable Spectroscopy. In

Portable Spectroscopy and Spectrometry, Volume 2, Applications

. Crocombe, R. A.; Leary, P. E.; Kammrath, B, W., Eds;. John Wiley, 2021; pp. 1–17.

(2) Ellis, D. I.; Muhamadali, H.; Haughey, S. A.; Elliott, C. T.; Goodacre, R. Point-and-Shoot: Rapid Quantitative Detection Methods for On-Site Food Fraud Analysis – Moving Out of the Laboratory and into the Food Supply Chain. 

(3) Carron, K.; Cox, R. Qualitative Analysis and the Answer Box: A Perspective on Portable Raman Spectroscopy. 

Identification and Confirmation Algorithms for Handheld Spectrometers, in Field-Portable Instrumentation, Encyclopedia of Analytical Chemistry,

(5) Zhang, L.; Lee, L. M.; Schreyer, S. K. 

Identification and Confirmation Algorithms for Handheld Analyzers. In Portable Spectroscopy and Spectrometry, Volume 2, Applications. 

Crocombe, R. A.; Leary, P. E.; Kammrath, B, W., Eds; John Wiley, 2021; pp. 19–42.

Library and Method Development for Portable Instrumentation: A Case Study Approach. In Portable Spectroscopy and Spectrometry, Volume 2, Applications.

 Crocombe, R. A.; Leary, P. E.; Kammrath, B, W., Eds; John Wiley, 2021; pp. 43–63.

Advances in Raman Instrumentation and Sampling Techniques. In Raman Spectroscopy: Theory and Practice. 

Szymanski, H. A., Ed., Plenum Press, 1967; pp. 44–81.

(9) Angel, S. M.; Carrabba, M; Cooney, T. F. The Utilization of Diode Lasers for Raman Spectroscopy. 

(10) Denson, S. C.; Pommier, C. J. S.; Denton, M. B. The Impact of Array Detectors on Raman Spectroscopy. 

(11) Owen, H. The Impact of Volume Phase Holographic Filters and Gratings on the Development of Raman Instrumentation. 

(12) Carrabba, M. M.; Rauh, D. Apparatus for Measuring Raman Spectra over Optical Fibers. US 5 112 127, 1992.

(13) Williamson, J. L.; Bowling, R. J.; McCreery, R. L. Near-Infrared Raman Spectroscopy with a 783-nm Diode Laser and CCD Array Detector. 

(14) Wang, T.; McCreery R. L. Evaluation of a Diode Laser/Charge Coupled Device Spectrometer for Near-Infrared Raman Spectroscopy. 

(15) Adar, F.; Delhaye, M.; DaSilva, E. Evolution of Instrumentation for Detection of the Raman Effect as Driven by Available Technologies and by Developing Applications. 

Evolution and Revolution of Raman Instrumentation – Application of Available Techniques to Spectroscopy and Microscopy. In Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line. 

Lewis, I. R.; Edwards, H. G. M. Eds. Marcel Dekker, 2001, pp. 11–40.

Raman Instrumentation. In Analytical Applications of Raman Spectroscopy, 

Dispersive Raman Spectrometers. In Raman Spectroscopy for Chemical Analysis,

 Wiley-Interscience, 2000; pp. 149–220.

Nondispersive Raman Spectrometers, In Raman Spectroscopy for Chemical Analysis, 

Modern Raman Spectroscopy – A Practical Approach,

(21) Crocombe, R. A.; Leary, P. E.; Kammrath, B, W. 

Introduction to Portable Spectroscopy. In Portable Spectroscopy and Spectrometry, Volume 1, Technologies and Instrumentation. 

Crocombe, R. A.; Leary, P. E.; Kammrath, B, W., Eds;. John Wiley, 2021; pp. 1–13.

(22) Wright, C. W.; Harvey, S. D.; Wright, B. W. Detection of Hazardous Chemicals Using Field-Portable Raman Spectroscopy. Proc. SPIE 5048, Nondestructive Detection and Measurement for Homeland Security, (accessed 2003-07-06); DOI: 

(23) Eckenrode, B. A.; Bartick, E. G.; Harvey, A. D.; Vucelick, M. E.; Wright, B. W.; Huff, R. A.. Portable Raman Spectroscopy Systems for Field Analysis. 

https://www.ojp.gov/ncjrs/virtual-library/abstracts/portable-raman-spectroscopy-systems-field-analysis

(25) Hargreaves, M.; Green, R.; Jalenak, W.; Brown, C.; Gardner, C. 

Handheld Raman and FTIR Spectrometers. In Infrared and Raman Spectroscopy in Forensic Science,

 Chalmers, J. M.; Edwards, H. G. M.; Hargreaves, M. D. Editors; John Wiley & Sons, 2011; pp. 275–287.

(26) Clarke, R. H.; Londhe, S.; Premasiri, W. R.; Womble, M. E. Low-Resolution Raman Spectroscopy: Instrumentation and Applications in Chemical Analysis. 

10.1002/(SICI)1097-4555(199909)30:9%3C827::AID-JRS454%3E3.0.CO;2-S

(27) Clarke, R. H.; Womble, M. E. Sample Analysis Using Low Resolution Raman Spectroscopy. US 5 982 484, 1999.

(28) Angel, S. M.; Gomer, N. R; Sharma, S. K; McKay, C. Remote Raman Spectroscopy for Planetary Exploration: A Review. 

(29) Waldron, A; Allen, A; Colón, A; Carter, J. C.; Angel, S. M. A Monolithic Spatial Heterodyne Raman Spectrometer: Initial Tests. 

Fourier Transform Raman Spectroscopy: Instrumentation and Chemical Applications; 

Fourier Transform Raman Spectroscopy: From Concept to Experiment; 

(32) Chase, B. FT–Raman Spectroscopy: A Catalyst for the Raman Explosion? 

(33) Assi, S.; Abbas, I; Kielisczyzk, K.; Wade, O.; Arafat, B. Characterization of Street Drugs Using Handheld Fourier Transform Raman Spectroscopy. 

(34) Rathmell, C.; Bingermann, D.; Zeig, M; Creasey, D. 

Portable Raman Spectroscopy: Instrumentation and Technology. In Portable Spectroscopy and Spectrometry, Volume 1, Technologies and Instrumentation.

 Crocombe, R. A.; Leary, P. E.; Kammrath, B, W., Eds;. John Wiley, 2021; pp. 115-145.

SERS and SORS. In Portable Spectroscopy and Spectrometry, Volume 2, Applications. 

Crocombe, R. A.; Leary, P. E.; Kammrath, B, W., Eds;. John Wiley, 2021; pp. 347-276.

(36) US Department of Homeland Security, Office of Science and Technology and National Urban Security Technology Laboratory, Handheld Raman Spectrometers. Market Survey Report, March 2021. SAVER-T-MSR-28. 

https://www.dhs.gov/science-and-technology/saver/handheld-raman-spectrometers

(37) Landscape Study of Field Portable Devices for Chemical and Presumptive Drug Testing; Forensic Technology Center of Excellence (2018). U.S. Department of Justice, National Institute of Justice, Office of Investigative and Forensic Sciences. 

https://forensiccoe.org/landscape-study-of-field-portable-devices-for-presumptive-drug-testing/

(38) Jehlička, J.; Culka, A. Critical Evaluation of Portable Raman Spectrometers: From Rock Outcrops and Planetary Analogs to Cultural Heritage – A Review. 

(39) World Drug Report 2013; United Nations Office on Drugs and Crime. 

https://www.unodc.org/wdr2013/en/nps.html

(40) World Drug Report 2022, Booklet 4; United Nations Office on Drugs and Crime. 

https://www.unodc.org/res/wdr2022/MS/WDR22_Booklet_4.pdf

(41) Lo, S-L.; Brown C. W. Infrared Spectral Search for Mixtures in Medium-Sized Libraries. 

(42) Arnó, J.; Andersson, G.; Levy, D.; Tomczyk, Zou, C. P.; Zuidema, E. Advanced Algorithms for the Identification of Mixtures Using Condensed-Phase FT-IR Spectroscopy. Proc. SPIE 8032, Next-Generation Spectroscopic Technologies IV, 80320Z (12 May 2011); DOI: 

(43) Yaghoobi, M.; Wu, D.; Clewes, R. J; Davies, M. E. Fast Sparse Raman Spectral Unmixing for Chemical Fingerprinting and Quantification. Proc. SPIE 9995, Optics and Photonics for Counterterrorism, Crime Fighting, and Defence XII, 99950E (24 October 2016); DOI: 

(44) Chandler, L.L.; Huang, B.; Mu, T. A Smart Handheld Raman Spectrometer with Cloud and AI Deep Learning Algorithm for Mixture Analysis, Proc. SPIE 10983. Next-Generation Spectrosc. Technol. XII, 2019; p. 7. DOI: 

(45) Mozaffari, M. H.; Tay L-L. Raman Spectral Analysis of Mixtures with One-Dimensional Convolutional Neural Network, ArXiv Prepr. ArXiv2106.05316, 2021. DOI: 

(46) Cooman, T.; Trejos, T.; Romero, A. H.; Arroyo, L. E. Implementing Machine Learning for the Identification and Classification of Compound and Mixtures in Portable Raman Instruments. 

(47) Gozdzialski, L.; Wallace, B.; Stege, U.; Hore, D. Linear programming for spectral mixture analysis. 

https://analyticalscience.wiley.com/do/10.1002/was.00050199

(48) Wei, D.; Chen, S.; Liu, Q. Review of Fluorescence Suppression Techniques in Raman Spectroscopy. 

(49) Galloway, C.M.; Le Ru, E.C.; Etchegoin, P.G. An Iterative Algorithm for Background Removal in Spectroscopy by Wavelet Transforms. 

(50) Zhang, Z. M.; Chen, S.; Liang, Y. Z.; Liu, Z. X.; Zhang, Q. M.; Ding, L. X.; Ye, F.; Zhou, H. An Intelligent Background-Correction Algorithm for Highly Fluorescent Samples in Raman Spectroscopy. 

(51) Schulze, H .G.; Foist, R.B.; Okuda K.; Ivanov, A.; Turner, R.F.B. A Model-Free, Fully Automated Baseline-Removal Method for Raman Spectra. 

(52) Cadusch, P. J.; Hlaing, M. M.; Wade, S. A., McArthur, S. L., Stoddart, P. R.. Improved Methods for Fluorescence Background Subtraction from Raman Spectra. 

(53) Liu, H.; Zhang, Z.; Liu, S.; Yan, L.; Liu, T.; Zhang, T. Joint Baseline-Correction and Denoising for Raman Spectra. 

(54) Shreve, A. P.; Cherepy, N. J.; Mathies, R. A. Effective Rejection of Fluorescence Interference in Raman Spectroscopy Using a Shifted Excitation Difference Technique. 

(55) Zhao, J.; Carrabba, M. M.; Allen, F. S. Automated Fluorescence Rejection Using Shifted Excitation Raman Difference Spectroscopy. 

(56) McCain, S. T.; Willett, R. M.; Brady, D. J. Multi-Excitation Raman Spectroscopy Technique for Fluorescence Rejection. 

(57) Cooper, J. B.; Abdelkader, M.; Wise, K. L. Sequentially Shifted Excitation Raman Spectroscopy: Novel Algorithm and Instrumentation for Fluorescence-Free Raman Spectroscopy in Spectral Space. 

(58) Kögler, M.; Heilala, B. Time-Gated Raman Spectroscopy – A Review. 

(59) Chiuri, A.; Angelini, F. Fast Gating for Raman Spectroscopy. 

Articles

There is a growing desire among spectroscopists for having instruments
small enough to be taken to the sample, as opposed to bringing the sample
to the instrument. The result is that Raman spectrometers are becoming
more miniaturized. Because these instruments come at a lower cost and offer distinct advantages over traditional spectrometers, the expectation is that a rapid expansion of when these instruments are applied will come forthwith. We offer a preview of how future miniaturized Raman spectrometers might look.

Portable Raman Spectrometers: How Small Can They Get?

Morphologically directed Raman spectroscopy (MDRS) is a novel and reliable tool that would enable criminalists to obtain more discriminatory information from forensic samples than their current capabilities. MDRS combines automated particle imaging and Raman spectroscopy into one instrument. Particle imaging is performed to determine particle size and shape distributions of components in a blended sample. Particle size is an important physical property of particulate samples and can be used in conjunction with Raman spectroscopy in the analysis of a range of samples of forensic interest, including illicit and counterfeit drugs, soils, gunshot residue (GSR), and white powders. Although measurement of particle size distributions is routinely carried out across a wide range of industries and is often a critical parameter in the manufacture and analysis of many products and substances, it is not widely used in the forensic sciences. Raman spectroscopy is used in forensic science to determine the molecular chemistry of materials because it is rapid, reliable, allows for analysis without contacting the sample, is nondestructive, and enables detection at low concentrations. Combining these two analytical techniques into a single platform allows the individual components present within a blend or mixture to be independently characterized and compared.

	Morphologically directed Raman spectroscopy (MDRS) combines the power of automated particle imaging with Raman microspectroscopy into a single platform. Particle imaging is performed to determine particle size and shape distribution of components in a blended sample. These are important physical properties of particulate samples and may have a direct influence on a sample's performance. For example, the size and surface area of a particle can be related in a significant way to the physical, chemical, and pharmacologic properties of a drug. Particle size distributions (PSDs) are routinely measured across a wide range of industries because they are important, and sometimes critical, to the manufacture and performance of substances and products. In spite of this importance, they are not widely used as methods for classification, identification, or individualization in the forensic sciences, other than in forensic soil examinations. However, characterization of materials by their microscopic morphology is used to analyze a plethora of forensic science samples, including drugs, geological materials (that is, soil minerals, rocks, and so forth), glass, dust, gunshot residue (GSR), pollen and diatoms, hairs and fibers, cosmetics, other anthropogenic materials (such as various building materials), general unknowns, and white powders used in hoax powder attacks. Raman spectroscopy is useful for determining molecular and physical chemistry because it is fast, reliable, nondestructive, and a noncontact method. Methods based on Raman spectroscopy are also used in the forensic sciences (1) for the analysis of many types of physical evidence including illicit drugs (2–5), explosives (6–8), paint (9–14), fibers (15–17), ink (14,18), and general unknowns (7). There are also several research articles on the use of Raman spectroscopy for the analysis of other types of forensic evidence, with a significant focus on analysis of the organic components of GSR (17,19–25).

	Independently, both automated particle imaging and Raman microspectroscopy are valuable methods. Even when combined into a single instrument, analysis based upon each method's independent evaluation may be useful. However, the power of a combined analytical scheme is greater than the sum of the individual approaches. Together, the data from these two methods may provide insight about the sample including its manufacturing method, history, and quality. This type of information may be invaluable during analysis of evidence in forensic casework.

	When performing MDRS, the sample's morphological data is collected using a light microscope with an automated stage. This allows for the sorting of particles based on various physical parameters. Once this particle data is collected, these parameters are then used to automatically select particles for chemical analysis using Raman microspectroscopy. The ability to perform particle selection for chemical analysis using physical parameters removes subjectivity in the measurement. In addition, the automation of the stage and of the particle selection removes the need to expose the analyst to the sample for the prolonged periods of time that would be required if manual measurements were performed (26).

	The combination of morphological and PSD analysis with Raman spectroscopy has the potential to provide unexploited information about a plethora of samples of forensic interest. This research examined the use of MDRS for the forensic analysis of four evidence types: illicit and counterfeit drugs, soils, GSR, and white powders. It is shown that MDRS can be used for the comparison of questioned and known samples, material, and mixture identification or classification as well as potential provenance determination.

	The experiments were performed using a Morphologi G3-ID particle imaging and Raman spectroscopy system (Malvern Panalytical). Samples were automatically dispersed onto a quartz plate at a pressure of 4 bar using the integrated Sample Dispersion Unit (SDU) on the Morphologi G3-ID instrument. Automated imaging analysis was performed directly on the quartz sample dispersion plate to obtain particle morphology data. All image acquisition settings were defined in a standard operating procedure (SOP) and included brightfield illumination and the use of a 10-times objective.

	Particles with a circle equivalent diameter (CED) greater than 7.0 µm and solidity greater than 0.75 were tagged for chemical targeting by the image analysis software. Morphological data was collected for more than 150,000 individual particles for each sample. Of these, 3000 particles per sample were targeted for Raman spectroscopic analysis over the spectral range of the instrument (150–1850 cm

). Raman spectra were obtained with the coupled Kaiser optical systems RamanRxn1 Raman spectrometer using a 785-nm semiconductor laser with a power of <500 mW and an exposure time of 2 s. The sampling spot size of the instrument was 3 µm. Both the particle morphology data and the Raman spectra were analyzed using the Morphologi software. All experiments were performed in triplicate.

	Spectral analysis required the creation of a spectral reference library, which was made using standards purchased from Sigma-Aldrich. The individual particle spectra from each sample were then compared against the reference spectral library and identifications made based on a correlation search algorithm where a correlation value close to one indicates a close match to that reference, and a value close to zero indicates no match. Using these correlation values, the particles were chemically classified and PSDs of the individual component populations were generated.

	A five-component mixture of illicit drugs was prepared (equal parts by weight) using cocaine, phenobarbital, pentobarbital, D-methamphetamine, and amphetamine.

	Two suspected synthetic cathinones, marketed as "bath salts," were purchased over the internet: "Arctic Rush" and "Fast Forward."

	Three counterfeit Viagra samples were obtained via the internet, where two were from Singapore and one was from India. Sampling was performed by removing a small amount of white powder from the core of each tablet, and all analyses were performed in triplicate.

	Soil samples were collected from four sites along one road in Connecticut (27). The 60–120 µm mineral fraction was separated by sieving and washing. Then 7 mm

 of minerals was deposited onto the quartz plate using evaporative dispersion. The mineral data were compared to a library made by importing the Raman spectra of minerals from the RRUFF database (28), and mineral identification was made only when there was a correlation score greater than 0.85.

	GSR was collected from cotton targets onto a quartz plate from four test fires using the same ammunition. The muzzle-to-target distances varied for each test fire, and included distances of 3, 6, 12, and 24 in.

	Three commercially available artificial sweetener blends were purchased locally, all with dextrose as the bulking agent but containing different sweetening agents. Sweet'N Low (designated as sample A) has saccharin as its sweetening agent, Splenda (Sample B) contains sucralose, and Equal (sample C) contains aspartame. An additional generic sucralose-based sweetener (NutraTaste Gold) was also purchased locally for comparison with the Splenda. Samples of the pure sweetener components were purchased from Sigma-Aldrich for addition to the spectral reference library.

	All five of the drugs were identified using MDRS. Further, the different components had different particle morphologies (Figure 1) and PSDs, which could be used in the comparison of samples from different seizures to evaluate whether they come from a common source. The particle size and shape of a substance can be useful for determination of the manufacturing process and, therefore, for comparative source attribution. Different methods of preparation can result in particles with different crystal structures and habits. Usually, slow crystallization methods form larger crystals, and rapid crystallization methods form smaller crystals.


	Figure 1: Particle images showing the different morphologies of (a) amphetamine, (b) D-methamphetamine, and (c) cocaine.

	The two suspected synthetic cathinones were analyzed by MDRS. The first, "Arctic Rush," did not contain any synthetic cathinones, and instead was composed of L-DOPA, caffeine, and an unidentified chemical. The presence of caffeine is a common cutting agent because it is a stimulant. L-DOPA is commonly used for the treatment of patients with Parkinson's disease, but there have been cases of people abusing L-DOPA as a means of enhancing the dopamine rush, which explains why it would be found in a mixture that people take to induce euphoria. It has been shown to increase aggressive behavior when taken in conjunction with methamphetamine (29). Inositol and phenethylamine were detected in the second, "Fast Forward." Inositol is a sugar alcohol that is commonly used as a cutting agent for narcotics (30). Phenethylamine has a similar chemical structure to synthetic cathinones, but there is a difference in one of the functional groups (31). Phenethylamine has a similar effect to that of synthetic cathinone, often ending in a state of excited psychosis or death (32).

	MDRS was able to identify four components in the counterfeit Viagra pills, including the active pharmaceutical ingredient (API) sildenafil citrate, and three excipients: lactose, starch, and talc. In an authentic Viagra sample, one would expect to see microcrystalline cellulose, calcium phosphate dibasic, croscarmellose sodium, and magnesium stearate. These were not detected in the counterfeit samples. Of particular interest is that the sample from India only contained the API and starch, while the two from Singapore also contained the lactose and talc (Figure 2). This has the potential to serve as a useful feature for comparison that could be used for counterfeit source attribution.


	Figure 2: A graph showing the relative percentage count of each of the components (lactose, sildenafil citrate, starch, and talc) in the three counterfeit Viagra samples, with the one from India in blue and the two from Singapore in green and red.

	The mineral morphologies of for each of the four samples were compared, and it was seen that they were all very similar. This is expected since they were collected along the same road in Connecticut. Other research has shown that the morphology of quartz can be used to differentiate different mineral environments (33–38), which could be exploited with MDRS. The PSDs for each location also proved to be indistinguishable.


	Table I: The average percent particle counts for each mineral for the four soil sample locations. The four locations can be differentiated because of the presence of some minerals in some sites that are not seen in the others, such as diopside in location C and epidote in location A, or because of differences in mineral counts, such as the higher percentage of muscovite in location B than the other sites.

	However, the samples for each location were differentiated when comparing the percent particle counts for each mineral (Table I). Multivariate statistical analysis was performed using R software (The R Foundation), using code written by N.D.K. Petraco, PhD, and adapted for this research by B.W. Kammrath, PhD. Principal component analysis (PCA) was performed on the particle count data, and in three principal components, 87.7% of the variance of the data was captured (Figure 3). Further analysis of the three-dimensional (3D) scores plot shows that there is good separation between the four data sets, with good clustering among replicates, which indicates the mineral counts achieved by MDRS can be used for sample discrimination.


	Figure 3: Three-dimensional scores plot of the autoscaled PCA results from the particle counts of the four soil locations. The three replicates cluster well for each of the four locations. 

	The collected residue from each of the four targets contained particles consistent with GSR as identified by the presence of nitrocellulose. The morphologies and PSDs of the particles were also analyzed (Figure 4). The PSDs have distributions centering on approximately the same circular-equivalent (CE) diameters, although there does appear to be a relationship between the volume distribution and the muzzle-to-target distance. This visual relationship, where the greater muzzle-to-target distance had larger volume distributions, needs to be explored in more detail with additional test fires using various ammunitions before this conclusion can be validated.


	Figure 4: Raman spectra (a) and particle image (b) of one GSR particle showing a spherical morphology and the presence of nitrocellulose. (c) The PSD of nitrocellulose-identified particles for all four muzzle-to-target distances (3, 6, 12, and 24 in. shown in red, blue, green, and brown, respectively).

	The automated imaging data alone were not enough to definitively identify the components within each of the artificial sweetener samples by particle size or shape. However, with the assistance of Raman chemical identification, individual components can be classified. Furthermore, PSDs were generated and compared for each of the chemical classes identified within a sample. Figure 5 shows the overall PSDs, the PSD of the dextrose, and PSD of the sweetening agent for each of the analyzed blends. The PSD of the entire sample and their distribution of dextrose bulking agents are the same, which shows that the PSD of the entire sample is primarily caused by the bulking agent. In addition, the PSD of the active sweetener component is markedly different from the dextrose, but does not have a detectable effect on the PSD of the entire sample. In a traditional bulk Raman analysis, smaller particles present in a lower volume are masked by the larger volume of dextrose, and thus cannot be identified. However, the particle specific targeting of MDRS enables the individual sweetening agent to be identified, which would allow for classification based on sweetener presence, and its component PSD to be analyzed.


	Figure 5: The PSD of Sweet'N Low (blue), Splenda (green), and Equal (red) for (a) all particles in a sample of each of the sweetener blends, (b) the particles identified as dextrose in a sample of each of the sweetener blends, and (c) the sweetener component in a sample of each of the sweetener blends.

	When comparing the results of the Splenda with the generic sucralose-based sweetener, the PSDs of the dextrose were the same for both samples. However, the PSDs of the sweetening agent were different (Figure 6). Thus, it was shown that even though these two samples are made with the same chemicals, the individual components have different PSDs that enable their differentiation. Consequently, the PSDs of the sweetening agent could prove to be a class characteristic of the specific manufacturer, thus its analysis could be used in determining and differentiating the particular brand of artificial sweetener.


	Figure 6: The PSD of Splenda (red) and the generic sucralose-based sweetener (green) for (a) the particles identified as dextrose in a sample of each of the sweetener blends and (b) the sucralose in a sample of each of the sweetener blends.

	MDRS is a nondestructive, relatively fast, and automated way to collect chemical and particle size information about samples of forensic interest. MDRS can be used in the forensic analysis of illicit and counterfeit drug mixtures, soil minerals, and hoax white powders by comparing two or more samples, not only based on their concentrations and identities of components, but also on the morphologies and size distributions of the particles. For illicit and counterfeit drugs, this determination of an individual components' particle size or shape distributions can provide additional information for connecting individual drug seizures and suppliers, tracing drug distribution routes and networks, and potentially identifying their geographical origin. For forensic soil analysis, MDRS enables PSDs and morphological examinations of each mineral component, which provides valuable information that can be used for comparison and source determination. In the analysis of hoax powders, MDRS can identify and characterize particles of trace components within mixtures that may be "hidden" in bulk Raman analysis. In addition, two components in a mixture with similar PSDs can be differentiated with MDRS because of the ability to chemically target specific particles, and two mixtures with the same components can be differentiated with MDRS because of their different PSDs.

	MDRS is a new approach to GSR analysis that combines Raman identification with particle size and shape information, thus making it comparable to current methods, specifically analysis by scanning electron microscopy with energy dispersive X-ray spectroscopy.

	The use of MDRS for the analysis of illicit and counterfeit drugs, soil minerals, GSR, and white powders is a novel and reliable tool that would enable criminalists to obtain more information from forensic evidence than is currently used for investigations and adjudications.

	(1) J.M. Chalmers, H.G. Edwards, and M.D. Hargreaves, 

Infrared and Raman Spectroscopy in Forensic Science 

	(2) C. Eliasson, N.A. Macleod, and P. Matousek, 

	(3) C.A. Penido, M.T.T. Pacheco, and L. Silveira, 

	(4) C.A.F.O. Penido, M.T.T. Pacheco, R.A.. Zangaro, and L Silveira,

	(6) E.M.A. Ali, H.G.M. Edwards, M.D. Hargreaves, and I.J. Scowen,

	(7) B. Cletus, W. Olds, P.M. Fredericks, E. Jaatinen, and E.L. Izake, 

	(8) E.M.A. Ali, H.G.M. Edwards, and R. Cox, 

	(9) J.D. Gelder, P. Vandenabeele, F. Govaert, and L Moens, 

	(10) P. Buzzini, G. Massonnet, and F.M. Sermier, 

, L. Kobilinsky, Ed. (John Wiley & Sons, Inc., Hoboken, New Jersey, 2011), pp. 131–224.

	(12) S.P. Stewart, S.E.J. Bell, W.J. Armstrong, G. Kee, and S.J. Speers, 

	(13) A. Michalska, A. Martyna, J. Zieba-Palus, and G. Zadora, 

	(17) K.C. Doty, C.K. Muro, J. Bueno, L. Halámková, and I.K. Lednev, 

	(18) S.E.J. Bell, S.P. Stewart, Y.C. Ho, B.W. Craythorne, and S.J Speers, 

	(20) J.Bueno, V. Sikirzhytski, and I.K. Lednev, 

	(21) M. López-López, J.J. Delgado, and C. García-Ruiz, 

	(22) M.López-López, J.L. Ferrando, and C. García-Ruiz, 

	(25) J. Zeng, J. Qi, F. Bai, J.C.C. Yu, W.-C. Shih, 

	(27) B. Kammrath, A. Koutrakos, J. Castillo, C. Langley, and D. Huck-Jones, 

https://doi.org/10.1016/j.forsciint.2017.12.034

	(28) B. Lafuente, R.T. Downs, H. Yang, and N. Stone, in 

Highlights in Mineralogical Crystallography 

, T. Armbruster and R.M. Danisi, Eds. (W. De Gruyter, Berlin, Germany, 2015), pp. 1–30.

	(30) P. Lociciro, P. Hayoz, P. Esseiva, L. Dujourdy, F. Besacier, and P. Margot, 

	(32) B. Dean, S. Stellpflug, A. Burnett, and E. Engebretsen, 

(Cambridge University Press, Cambridge, UK, 1973).

	(37) S.J. Culver, P.A. Bull, S. Campbell, R.A. Shakesby, and W.B. Whalley,

 is with the Henry C. Lee College of Criminal Justice and Forensic Sciences, Department of Forensic Science, at the University of New Haven in West Haven, Connecticut. 

 is with the Henry C. Lee College of Criminal Justice and Forensic Sciences, Department of Forensic Science, at the University of New Haven and the University of Verona in Verona, Italy. 

 is with Smiths Detection in Edgewood, Maryland. 

 are with Malvern Panalytical in Westborough, Massachusetts. 

 is with the Malvern Panalytical in Worcestershire, UK. Direct correspondence to: 

Can morphologically directed Raman spectroscopy obtain more discriminatory information from forensic samples than current tools?

Pauline E. Leary

Articles

Portable Raman Spectrometers: How Small Can They Get?

Morphologically Directed Raman Spectroscopic Analysis of Forensic Samples