Largely heralded as a "wonder material" since its discovery in 2004, graphene has now entered the era of industrial manufacturing. Although several techniques for large-scale graphene manufacturing have been developed, one important question remains: how to easily and rapidly characterize the quality of graphene and other carbon nanomaterials to effectively monitor and control the production process, where simple and robust analytical tools are required. In this study, three types of materials (graphene powders, graphene powder coated sheets, and carbon nanofibers) were analyzed using portable Raman spectroscopy. This efficient analysis is ideal for at-line or online material analysis for manufacturers of graphene and other carbon nanomaterials to do materials characterization, product quality control, and process monitoring.

	Carbon nanomaterials constitute a variety of carbon allotropes including graphene, graphene oxide, carbon nanotubes, and carbon nanofibers. Each of these materials exhibits unique properties in electrical conductivity, thermal conductivity, and mechanical strength thanks to the distinct structures of each allotrope. For instance, graphene is a two-dimensional (2D) material formed from a hexagonal lattice of carbon atoms, whereas graphite is made up of stacked individual layers of graphene. Graphene's strength, superconductivity, and excellent heat conductivity makes it a very attractive material as a conductor in memory chips and batteries for electronics (1), and its flexibility has the potential to invigorate infrastructure, aeronautics, and biotechnology. Graphene oxide, an alternative form of graphene made from oxidized graphite that has been exfoliated to form a few layers or a single layer, is used to desalinate water and remove radioactive isotopes (2). Carbon nanotubes, which are tubes with walls formed from graphene sheets, have attracted interest from industry to potentially generate high-surface-area catalysts in fuel cells with the addition of different functional groups (3). Carbon nanotubes can either be single-walled nanotubes (SWNTs) or multiwalled nanotubes (MWNTs). Recently, MWNTs have found practical use in lithium-ion batteries to increase the lifespan and capacity of portable electronics (4). SWNTs have been slower to find commercial application, but have potential for use as transistors in microelectronics because of their bandgap and as biosensors because of their biomolecular compatibility (4). A carbon nanofiber consists of graphene layers stacked as cones or plates; recently, carbon nanofiber has found applications in construction materials thanks to its flexibility and durability (5).

	Because of the intense interest surrounding their unique properties and use in many practical applications, graphene and other graphene-based nanomaterials have now entered the era of large-scale manufacturing. Graphene can be made through various processes including

		mechanical exfoliation, a low-cost technique that involves isolating graphene from bulk graphite (6),

		chemical vapor deposition (CVD), which is performed on a relatively larger scale but at a high cost and with expensive equipment (7,8), and

		exfoliation and reduction of graphene oxide, which is synthesized through the oxidation of graphite powder on a larger scale, but can result in extensive defects (9).

	Although recent market forecasts predict the value of the current global graphene market to be only in the low tens of millions of dollars (10), the value of the global graphene market is estimated to reach half a billion dollars by 2025 (11). It may be decades before the potential of these carbon nanomaterials are fully realized, but research and development spending on graphene is likely to reach hundreds of millions of dollars in the coming years.

	To reach these market estimates, one important question for large-scale graphene manufacturing remains: how to easily and rapidly characterize the quality of graphene and other carbon nanomaterials to effectively monitor and control the production process? Raman spectroscopy has been used extensively by carbon nanomaterials research communities in recent years because of its ability to characterize materials from their molecular vibrations (12,13). The Raman spectra of carbon nanomaterials are typically characterized by only three major bands: the G band, the D band, and the 2D band (also known as the G? band). Though simple, the spectra of these nanomaterials are rich in information about their quality and their microstructures as revealed by the peak positions, peak shapes, and peak intensities. The G band appears around 1582 cm

 vibrational mode (13,14); it is an indication of the crystallinity of the material. The dispersion of the G band can be observed in disordered graphene materials, and the dispersion is proportional to the degree of disorder (15). The D band at around 1350 cm

 is attributed to the structural disorder near the edge of the microcrystalline structure that decreases the symmetry of the structure (14). The Raman peak intensity ratio of these two bands, 

, can be used to characterize the degree of disorder of the materials (15). The 2D band appears around 2700 cm

 depending on the laser excitation wavelength and is related to the number of graphene layers (3).

	The Raman spectra of monolayer graphene (red), carbon nanotubes (black), single crystal graphite (green), and carbon black powder (blue) are shown in Figure 1. High-quality graphene is characterized by the sharp, symmetric single peak of the 2D band. The graphite spectrum displays a high amount of order and thus crystallinity characterized by its prominent G band and the absence of a D band. Since graphite is composed of many layers of monolayer graphene, the 2D band observed in the graphite spectrum is much broader and asymmetric than the graphene 2D band, indicating multiple components from several phonon modes. Carbon nanotubes display unique features of the G band. Because of the confinement and curvature of graphene layers in forming nanotubes, the G band becomes asymmetric for MWNT and more likely splits into two bands, the G+ and the G band for SWNT (15). The spectrum of carbon black, which has the lowest amount of crystallinity, displays a strong D band, a broad G band, and a high 

 ratio, indicating a disordered structure.


	Figure 1: Raman spectra of graphene (red), carbon nanotubes (black), graphite (green), and carbon black (blue). The spectra have been manually offset.

	Raman spectroscopy provides rich information about the characteristics of carbon black, graphite, graphene, and other carbon nanomaterials. Although confocal Raman microscopy can provide high-resolution characterization of carbon nanomaterials, the high cost of this instrumentation coupled with the complexity in operation and data interpretation is inadequate for monitoring in a large-scale manufacturing setting, where simple and rapid analytical tools are required. In this study, a high-throughput portable Raman spectrometer is used to characterize three types of materials: graphene powders, carbon nanofibers, and carbon black. This quick analysis can be applied as an at-line or online technique for graphene and other carbon nanomaterials manufacturers to perform materials characterization, product quality control, and process monitoring.

	An i-Raman Pro HT system (B&W Tek) with a laser excitation of 532 nm via a fiber-optic sampling probe (spot size ~100 µm) was used for measurements of all samples. The system uses a high-throughput Raman spectrometer with a back-thinned charged-coupled device (CCD) thermoelectrically cooled to -25 °C. For materials in powdered forms, no microscope was used. A probe holder with an adjustable 

 stage was used to support the fiber-optic probe over an aluminum pan containing a given carbon sample. The 

-focus was used to optimize the working distance from the probe to the sample. BWSpec acquisition software was used for data collection, spectral processing, and programming the automatic calculation of peak intensities and ratios for online measurements or off-line batch processing.

	Three types of carbon nanomaterials were analyzed: graphene powders, carbon nanofibers, and carbon black powders. For graphene powders, a laser power of 40 mW with an integration time of 60 s was used to acquire Raman spectra. For carbon nanofiber and carbon black powder samples, an integration time of 90 s was used with a laser power of ~21 mW. Three spectra were collected for each sample. No spectral averaging was used.


	Figure 2: Raman spectrum before (red trace) and after (blue trace) airPLS baseline correction.

 sample stage of the probe holder, 15 measurements from different locations of one graphene powder were collected to assess the uniformity of the sample. Two types of preprocessing were applied to the spectra prior to the peak ratio calculation. First, an adaptive iteratively reweighted penalized least squares (airPLS) background correction was used to remove any fluorescent background present in the spectra. Figure 2 shows an example of before (red trace) and after (blue trace) baseline correction. A Savitzky-Golay smoothing algorithm with a window size of 7 was also applied to all spectra. Figure 3 shows five manually offset representative spectra from different areas of the graphene powder sample. The D band (1346 cm

) can be observed. A prominent band at 2328 cm

 is the Q-branch Raman peak for atmospheric nitrogen (N

) because of the nitrogen–nitrogen bond (16). The N

Raman peak is observed because of the high-throughput nature of the Raman spectrometer.


	Figure 3: Raman spectra of five locations on a graphene sample. The spectra have been manually offset by 1000 intensity units.

 is an important parameter in assessing the level of disorder in the crystal structure of a sample. Table I presents the calculated peak intensities after background removal and intensity ratios for the D and G bands of the five representative spectra. The 

ratios of the 15 measurements ranged from 0.1246 to 0.1462. The average 

 of all 15 scans was 0.1386, with a percent relative standard deviation (%RSD) of 4.07%. This result indicates a high level of uniformity within the sample.

	Two carbon nanofiber samples and four carbon black samples were also tested. Three spectra were collected at different locations for each sample. For all collected spectra, an airPLS background correction was used to remove any fluorescent background present in the spectra. A Savitzky-Golay smoothing algorithm with a window size of 5 was also applied to all spectra. Figure 4 shows the manually offset representative spectra for each sample. The four carbon black samples display typical carbon black Raman signatures that contain D and G bands but no 2D band. The two carbon nanofiber samples show prominent D bands, indicating a high level of disorder. The G bands in the carbon nanofiber spectra also exhibit some asymmetry. The asymmetry is likely because of the curvature of the nanofiber, with the slight splitting of the G band induced by curvature of graphene layers when forming the nanofiber (15).


	Figure 4: Raman spectra of two carbon nanofiber samples and four carbon black samples. The spectra have been manually offset by 5000 intensity units for clarification.

	BWSpec software was used to automatically calculate the 

 values as spectra were collected. Table II presents the calculated 

 peak intensity ratios measured for the four carbon black samples and two carbon nanofiber samples. The carbon nanofiber sample 2 has the highest level of order of all samples with an average 

 of 0.4706, whereas the carbon nanofiber sample 1 has an especially high level of disorder with an average 

	Raman spectroscopy can also be very useful for identifying contaminants or starting materials in carbon nanomaterial final products. For example, the carbon black 3 and carbon black 4 samples contain up to 10% hematite (Fe

) because of their manufacturing processes. Figure 5 shows the Raman spectra of carbon black 3 and 4 compared to the Raman spectrum of hematite. Raman peaks at 213 cm

 are in good agreement with the Raman spectrum of hematite.


	Figure 5: Raman spectra of carbon black 3 (purple) and carbon black 4 (orange). Raman peaks at 213 cm

 are consistent with a hematite Raman spectrum (black). The spectra have been manually offset by 1000 cm

	Having robust and simple quality control tools for carbon nanomaterials such as portable Raman spectroscopy will help facilitate the future growth of the global graphene market. Although simple, a Raman spectrum of a carbon nanomaterial can provide a vast amount of information to characterize materials including carbon black, graphite, carbon nanotubes, and graphene. Portable Raman spectroscopy is capable of quickly characterizing carbon nanomaterials, revealing critical information regarding material qualities such as sample crystallinity and level of disorder using various parameters such as D- and G-band intensity ratios. Graphene manufacturers can easily use the analysis to obtain at-line or online measurements for material characterization, product quality control, and process monitoring.

	(1) C. Ahn, S.W. Fong, Y. Kim, S. Lee, A. Sood, C.M. Neumann, M. Asheghi, K.E. Goodson, E. Pop, and H.S.P. Wong, 

	(3) C. Luo, H. Xie, Q. Wang, G. Luo, and C. Liu, 

	(4) M.F.L. DeVolder, S.H.Tawfick, R.H. Baughman, andd J. Hart, 

	(7) C. Mattevi, H. Kim, and M. Chhowalla, 

	(9) S. Stankovich, D.A. Dikin, R.D. Piner, K.A. Kohlhaas, A. Kleinhammes, Y. Jia, Y. Wu, S.T. Nguyen, and R.S. Ruoff, 

https://www.grandviewresearch.com/industry-analysis/graphene-industry

	(12) M.S. Dresselhaus, A. Jorio, M. Hofmann, G. Dresselhaus, and R. Saito, 

	(14) A.C. Ferrari, J.C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K.S. Novoselov, S. Roth, and A.K. Geim, 

 are with B&W Tek in Newark, Delaware. Direct correspondence to: 

Articles

Raman spectroscopy is ideal for at-line or online material analysis for manufacturers of graphene and other carbon nanomaterials to do materials
characterization, product quality control, and process monitoring.



Portable Raman Spectroscopy for At-Line Characterization of Carbon Nanomaterials

Gas analysis systems used for mud logging in the oil and gas industry provide information that is critical to optimize the drilling process and for ensuring safety on-site. As drilling speeds increase, measurement of hydrocarbons and nonhydrocarbons in real time becomes more challenging for traditional methods like gas chromatography (GC). Raman spectroscopy offers a promising alternative for multicomponent analysis of complex gas mixtures, particularly when high-sensitivity compact Raman instrumentation is combined with advanced data analysis. Here we share a new data-driven Raman spectroscopy (DDRS) method capable of simultaneously measuring 12 hydrocarbon and nonhydrocarbon gases in the presence of matrix interferences. Validation of its performance using both standard gas mixtures and two real mud-logging data sets compared favorably against the GC method, demonstrating the feasibility of this technique for online, high-throughput quantitative analysis of gases in oil and gas exploration and recovery, as well as many other industries.

 details the composition and characteristics of the rock cuttings, mud, and gases brought to the surface by borehole drilling during oil and natural gas exploration. It provides valuable information about the quality and status of a drill site, logging the position and mix of hydrocarbons to facilitate efficient extraction and provide advance warning of dangerous gas levels (1). As rocks are crushed in the drilling process, a variety of gases indicative of the reservoir are released, including hydrocarbons such as methane, ethane, propane, isobutane, normal butane, isopentane, and normal pentane as well as a variety of nonhydrocarbons: CO

. Rapid, continuous analysis of the composition of these complex mixtures is essential to help workers respond quickly to changes during drilling.

	Gas Chromatography Versus Raman Spectroscopy

	Although gas chromatography (GC) is the most widely used technique for the evaluation of multicomponent gases (2), it is limited both in speed (3) and its ability to distinguish between hydrocarbons and nonhydrocarbons simultaneously. GC also carries ongoing costs in the form of consumable columns, associated maintenance, and the need for trained operators.

	Raman spectroscopy, in contrast, can be performed in just seconds, requires no sample preparation, and does not consume the sample. It has very little sensitivity to water vapor, and can probe all the relevant hydrocarbon and nonhydrocarbon gases concurrently. The challenge in making this technique widely deployable lies in achieving the sensitivity and specificity required to quantify each component within the complex, widely varying gas mixtures extracted from boreholes.

	Since the first application of Raman spectroscopy to natural gas detection in 1980 (4), many variations have been proposed to enhance the signal, with successful deployment in some process analysis and onsite applications. Although specialized techniques can go so far as to achieve sub-parts per million detection limits, if Raman is to replace GC in the industry, the instrumentation must also be compact and portable enough for onsite deployment while still retaining sufficient sensitivity for meaningful quantitative analysis. Improvements in detection efficiency are part of the equation, but systems must also be able to quantify both hydrocarbon and nonhydrocarbon gases within the same matrix in a timely manner.

	Analysis Challenges and Proposed Solution

	The Raman spectra of complex gas mixtures of the type seen in mud logging are composed of many overlapping bands, and are also subject to considerable matrix interference. Those challenges demand the development of new chemometric analysis methods that are capable of deconvoluting superimposed spectra and neglecting background and matrix effects. The present body of work introduces a new approach called

) (5). DDRS is an application of partial least squares (PLS) analysis that combines higher-density discrete wavelet transform (HDWT) with variable selection influenced by the data itself, with the goal of isolating the most useful spectral features for target gas quantification in complex multicomponent mixtures. Instead of analyzing Raman spectra based on matching individual peaks at discrete wavelengths, DDRS uses an HDWT representation of the Raman spectrum to extract the most relevant spectral bands for the analysis of each given component.

 breaks down data into its frequency band components, allowing sharp changes in signal to be easily discriminated from noise. This concept has been applied to good effect in JPEG compression of images, which uses two-dimensional (2D) discrete wavelet transform (DWT) to find the local changes in brightness that truly define an image, filtering out noise and allowing the image to be compressed without loss of crucial details. Similarly, DWT has proven useful for background subtraction in Raman spectra (6).

	DWT samples data in both the frequency and time domains. In the case of Raman spectra, the frequency domain correlates to peak width, while the time domain correlates to peak position. Raman spectra contain a rich amount of information in the form of both peak position and width, facilitating analysis in the frequency domain (peak width resolution) to mitigate the effects of background and noise, as well as in the time domain (peak wavelength position) to isolate analyte-specific spectral bands from the presence of other species and matrix interference.

	In developing the DDRS methodology, two key elements of a typical DWT approach were modified to overcome known weaknesses and improve its applicability to the analysis of multicomponent gases. The first improvement related to the transform itself. DWT down-samples data which, in the case of spectra, has the effect of degrading spatial resolution in the time (peak position) domain. To offset this effect, an oversampling technique called 

higher-density discrete wavelet transform

) was used. HDWT was developed to improve both the time and frequency resolution of DWT (7), and to mitigate a known vulnerability of traditional DWT to the alignment of the signal in time (or peak position, in the case of Raman). When applied to overlapping spectral bands in mixtures, the improved spectral resolution of HDWT allowed interfering components to be more easily distinguished. It also has the practical benefit of making the transform less sensitive to wavelength shifts, which is extremely important for Raman instrumentation operating in harsh environments and over a wide range of temperatures.

	The second improvement defining DDRS lies in selection of the variables for PLS analysis-the unique spectral features that will allow each component to be identified and quantified with minimal overlap and greatest accuracy. A random frog algorithm is typically used for this purpose, but because this approach often results in the selection of unimportant variables, a template-oriented frog algorithm (TOFA) was proposed, in which variables are weighted by an estimate of their importance. For a detailed description of the HDWT and TOFA methodology and application, readers are referred to the literature (5).

	Together, these two improvements to traditional DWT define a DDRS analysis method intended to isolate the optimal combination of spectral features for each component gas with minimum overlap, and then construct high-quality calibration models for gas analysis. To test its potential for extracting quantitative information, the method was put through its paces with a series of known multicomponent gas mixtures. It was also compared to GC analysis in continuous monitoring studies at two actual gas logging sites.

	To be viable for use in oil and gas exploration, a Raman spectroscopy system must be compact, robust, and capable of delivering repeatable data over a wide range of conditions, often harsh. The system built to test the DDRS method (Figure 1) was designed with onsite measurement in mind, integrating a 532-nm, 300-mW diode-pumped solid-state laser (WSLS-532-300 m, Wavespectrum Laser Group Limited) with a high numerical aperture Raman spectrometer for optimum throughput (Wasatch Photonics 532 Raman spectrometer, f/1.3). The spectrometer utilized a 1624-lines/mm volume phase holographic (VPH) grating in an aberration-corrected transmissive design with a thermoelectrically cooled Hamamatsu charge coupled device (CCD) detector (1024 × 64 pixels). Free-space coupling to the spectrometer's 25-µm slit was chosen to maximize use of the f/1.3 aperture for maximum sensitivity, thus enabling short integration times while maintaining spectral resolution of ~9 cm


	Figure 1: Design of the Raman gas analysis system. A 532-nm laser was coupled into the gas cell for excitation of Raman scattering, with longpass filtering of signal before detection by a high-sensitivity Raman spectrometer.

	Laser light was delivered to the gas cell via a longpass dichroic beamsplitter designed to reflect the 532-nm laser and pass the Raman scattered light while excluding Rayleigh scattering (Figure 1). A 20-mm focal length lens focused laser light into the gas cell through a sapphire window, also collimating emitted Raman scattering for transmission through the longpass dichroic. The telescope in the laser path expanded the beam to enable a tighter focus of laser light into the gas cell. Rayleigh scattering in the signal path was further suppressed by a longpass filter, after which the Raman scattered light was focused onto the 25-µm entrance slit of the spectrometer by a 30-mm focal length achromatic lens. Throughout the measurements, the gas cell temperature and pressure were monitored, and pressure–flow was precisely controlled using regulator valves.

	Using this gas analysis system, static measurements of 164 standard gas mixtures were performed, in addition to continuous measurements at two mud-logging sites. The standard gas mixtures were divided into three groups, all provided by Messer China Ltd. The first group of 74 samples was composed of hydrocarbons, and included seven alkanes: methane, ethane, and propane at 0.01–20%, isobutane and normal butane at 0.01–6%, and isopentane and normal pentane at 0.01–1.6%. The second group of 52 samples was composed of nonhydrocarbons and included CO

 in concentrations of 0.01–10.5%. In both cases, the distribution of concentrations in the samples were assigned following the principle of uniform design to simulate practical mug-logging gas samples. An additional 38 samples of O

 gas were tested separately because of the risk of explosion. Of the total 164 sample mixtures, 19 were chosen randomly for use as an independent validation set, and the remainder were used for training the models.

	Individual spectra collected for each of the seven hydrocarbon gases exhibit unique signatures (Figure 2), but because of their similar bond structure, many of their Raman bands overlap closely. This overlap is problematic for detecting low concentrations of heavier hydrocarbons like butane and pentane, which are easily swamped by the more abundant, lighter gases. As a result, the spectra of the representative standard gas mixtures show Raman activity in similar bands, but with distinct spectral differences indicative of their differing compositions. The nonhydrocarbon gases, in contrast, were easily discriminated from one another in mixture, because of their well-separated peaks.


	Figure 2: Raman spectra for the seven hydrocarbon gases under study, illustrating the high degree of overlap of the spectral bands key for identification and discrimination of each component. Adapted with permission from reference 5, copyright 2017 Elsevier.

PLS Versus DDRS: Analysis of Standard Mixtures

	To properly assess the success of the newly developed DDRS analysis method, it was compared against plain PLS modeling for the standard gas mixtures. For each of the 12 target gases, PLS modeling was performed and evaluated. The number of PLS factors varied from 2 to 10, and yielded poor prediction precision in general, with root mean square error of prediction (RMSEP) values of up to 0.1113 and R values in the range 0.708–0.996 (Figure 3). Given the high number of PLS factors required and the overall poor quality of fit, PLS alone did not deliver the precision or robustness needed for accurate, reliable detection of multiple gas components in mud logging.

	The same spectra were then reanalyzed using multivariate calibration models under the newly developed DDRS methodology. First, concentrations of the gases in each mixture type (hydrocarbon versus nonhydrocarbon) were assigned according to the principles of uniform design. Then HDWT was performed on each Raman spectrum, setting the decomposition scale to 6. After the sampling probability of each HDWT coefficient for the TOFA was calculated, the parameters of the TOFA were set. After 20 iterations, the TOFA results were collected in a matrix, and the frequency of each variable was calculated. This approach allowed a series of PLS models to be constructed with an increasing number of most frequent variable, from which the PLS model with minimum RMSEP was selected as the final model.

	Looking at the results of modeling each target gas component (Table I), it can be seen that DDRS delivers high prediction precision with a relatively low number of required variables and only two PLS factors. RMSEP values are significantly lower across the board for DDRS versus plain PLS modeling (Figure 3), while 

 values were consistently >0.969, indicating a much better quality of fit. Additionally, an F-test to compare the two approaches confirmed that DDRS is a significantly better calibration model for component analysis in the gas mixtures.


	Figure 3: Comparison of root mean square error of prediction (RMSEP) values for each of the 12 component gases under study as calculated for plain PLS modeling versus DDRS modeling.

	In terms of practical applicability, it is important that DDRS address the particular challenges of applying Raman spectroscopy to mud-logging gas analysis. Firstly, it must be able to successfully isolate a single gas target from within a highly overlapped and unpredictable matrix. This requirement is well evidenced by the results for methane, for which prediction results were improved tremendously by DDRS. Secondly, the model must be capable of quantifying low concentrations of heavier alkanes such as butane and pentane in mixtures dominated by lighter components. DDRS also performed extremely well in this respect, as can be seen by the excellent agreement between predicted and measured values for isobutane, normal butane, isopentane, and normal pentane (Figure 4).


	Figure 4: Measured versus predicted values for (a) isobutane, (b) normal butane, (c) isopentane, and (d) normal pentane as calculated by the DDRS model. Adapted with permission from reference 5, copyright 2017 Elsevier.

	With confidence in DDRS established, the limit of detection (LOD) for each target gas component was calculated according to the International Union of Pure and Applied Chemistry (IUPAC) (8), using pure helium in the system as a measure of the baseline noise. This approach yielded LOD values for the target gases of 0.012–0.086% (Table I), which are sufficiently sensitive to be relevant in the mud-logging industry.

	DDRS Versus GC: A Comparison in the Field

	Having established the capability of DDRS to quantify the target gas components within simulated gas mixtures, the method was benchmarked against the industry standard method of GC at two actual mud-logging sites. For this purpose, the system was connected serially between the extractor and GC system (CPS-KQ-VI, Shanghai China Petroleum Instrument Co., Ltd.) at each site, using a small Raman gas cell volume to minimize gas sample delay between the measurements. The extracted gas was measured for methane, first by the DDRS system (every 6 s), then by the GC system (every 2 min). Measurements were taken for 13 h at the first site and for 6 h at the second (Figure 5). In both cases, DDRS measurements tracked very closely with GC results, providing much faster results and with better time resolution. This speed of analysis is particularly important for the detection of thin or hidden hydrocarbon reservoirs, and represents a significant tangible benefit for use in mud logging.


	Figure 5: Comparison of GC (black curve) and Raman gas analysis system (red curve) results for methane, as measured at two mud-logging sites, (a) for 13 h, and (b) for 6 h. Adapted with permission from reference 5, copyright 2017 Elsevier.

	The use of a high-sensitivity Raman system in combination with a newly developed analysis methodology termed DDRS shows multiple benefits for multicomponent gas analysis in the mud-logging industry. DDRS makes use of higher-density wavelet transform (HDWT) together with template oriented frog algorithm (TOFA) to find an optimal combination of spectral features with minimum overlap, and then constructs high-quality calibration models for gas analysis. Its ability to isolate both hydrocarbon and nonhydrocarbon target gas components within an uncontrolled and variable matrix with high prediction accuracy makes it a promising analysis method at mud-logging sites. When tested under the rigors of two real-world mud-logging sites, DDRS compared favorably with GC. In delivering much faster, better resolved measurements, the system shows excellent potential for high-throughput analysis of multiple gas components in real time, which could be extended to online multicomponent analysis in many other gas and fuel systems.

	(3) S.A. Barclay, R.H. Worden, J. Parnell, D.L. Hall, and S.M. Sterner, 

	(5) X. Han, Z. Huang, X. Chen, Q. Li, K. Xu, and D. Chen, 

	(6) Y. Hu, T. Jiang, A. Shen, W. Li, X. Wang, and J. Hu, 

	(8) M. Safaeian, D. Solomon, and P.E. Castle, 

is a professor of Biomedical Engineering at the College of Precision Instrument & Opto-Electronics Engineering at Tianjin University in China. 

is the vice president of marketing at Wasatch Photonics in Durham, North Carolina. Direct correspondence to: 

This new data-driven Raman spectroscopy (DDRS) method is capable of simultaneously measuring 12 hydrocarbon and nonhydrocarbon gases in the presence of matrix interferences.



Data-Driven Raman Spectroscopy in Oil and Gas: Rapid Online Analysis of Complex Gas Mixtures

It’s said that “seeing is believing.” Unfortunately no single imaging technique can reveal every feature of a sample because each relies on a different interaction with the material being investigated. A more comprehensive understanding can be achieved by using several techniques and correlating the results. This advantage is particularly pronounced when using methods that, on the one hand, deliver images of the morphology of a sample and, on the other hand, give information suitable for determining its molecular composition. Confocal Raman imaging and scanning electron microscopy are techniques that are ideally suited to correlating structural and chemical information as shown here on a variety of samples.

	As has been demonstrated in the past, correlating morphological, molecular, and chemical properties from the same region of interest (ROI) of a complex specimen yield a much better understanding than confining oneself to studying only one of these characteristics. On the micrometer or even nanometer scale, the use of two complementary microscope technologies is the method of choice for such a thorough analysis. Over the past decade, high-performance analytical techniques relevant to correlative microscopy have come to include optical microscopy, scanning electron microscopy (SEM), atomic force microscopy (AFM), tomography, energy dispersive X-ray spectroscopy (EDS), focused ion beam (FIB), and secondary ion mass spectrometry (SIMS), among others. For example, the combination of a confocal light microscopy and electron microscopy (CLEM) has been used more, especially in cell biology, to characterize not only fixed but also living cells in great detail (1).

	The combination of SEM-EDS and Raman imaging is an ideal correlative technique for many applications. It includes SEM's ability to image structural surface features of a specimen at extremely high resolution, down to 1 nm, with a remarkable depth-of-field, resulting in images of great three-dimensional (3D) structural detail. EDS, often incorporated in an SEM system, can identify the chemical elements of the sample. Additionally, Raman spectroscopy can identify the sample's molecules and their characteristics. By acquiring a Raman spectrum at every pixel of the ROI, a Raman image can be generated, visualizing the local distribution of the constituent molecules, degree of crystallinity, crystal symmetry and orientation, and stress and strain states in the sample. The resolution of this technology is diffraction limited to about 150 nm. The three technologies-SEM, EDS, and Raman imaging-are well established and robust (2,3). The development of approaches for correlative Raman imaging and SEM has been described elsewhere (4).

	Since the first Raman-SEM instrument was assembled from home-built parts nearly 20 years ago, the development of Raman-SEM systems has significantly advanced to meet the requirements of scientists. Retrieving the ROI, which had long been a tedious and difficult procedure, has been made easy by locating the Raman unit within the SEM vacuum chamber and guiding it with integrated software. By automatically transferring the sample between the electron beam and the Raman microscope within the vacuum chamber, confocal Raman imaging can be performed without compromising SEM performance (5). Here, correlative Raman imaging-SEM (named RISE microscopy) was performed with several systems. In all of them, the objective and the scanning stage of a confocal Raman microscope (apyron, WITec) was located inside the vacuum chamber of the SEM. The location of the objective is offset from the SEM column, thus freeing space for other detectors that require the electron beam within the small chamber. After EDS and SEM imaging, the sample is simply transferred to the Raman measuring position using the SEM stage (Figure 1). Samples are automatically transferred from one measuring position to the other and stay within the vacuum chamber of the SEM system for the entirety of the procedure, thus streamlining the workflow and drastically improving the instrument's ease of use.


	Figure 1: The principle of correlative Raman-in-SEM imaging. For easy correlation of Raman and SEM images, a setup was developed that allows Raman imaging within the vacuum chamber of a stand-alone SEM system. The samples are automatically transferred from one measuring position to the other, streamlining the workflow and drastically improving the instrument's ease of use.

	Two different SEM systems (Mira, Tescan, and Sigma 300, ZEISS) were used for the analysis of cassiterite and pearl, and hematite, respectively.

	The Raman systems were equipped with a spectrometer (UHTS 300, WITec), a 100x vacuum objective (N.A. 0.75) and a 532-nm excitation laser. Raman spectra were analyzed with multivariate spectral analysis methods (6).

) is the main source of tin (Sn) and has been extensively studied. In this ore, Sn is often associated with other metals such as titanium (Ti), niobium (Nb), tantalum (Ta), indium (In), and tungsten (W). These elements can be identified and distinguished from SnO

 has characteristic Raman bands at 638 and 741 relative wavenumbers (Figure 2). The inclusion of other elements as well as variations in the crystal lattice structure influences this typical spectrum. For example, the Raman band at 840 relative wavenumbers (red spectrum in Figure 2) is produced by a vibrational mode of Nb. SEM and cathodoluminescence (CL) imaging were used to determine the precise ROI, EDS identified trace elements, and the following Raman analysis showed the distribution and crystal orientations of the molecules in the SnO

 crystal (7). This example also convincingly illustrates that confocal Raman imaging can be a quick procedure: The confocal Raman image comprising 40,000 pixels was acquired in less than 1 h and displays both very high spectral sensitivity and local resolution simultaneously.


	Figure 2: Raman-SEM image of cassiterite. The confocal Raman image that was generated based on the spectrum of each pixel of the ROI was superimposed onto a CL image of the same area. The Raman bands at 638 and 741 relative wavenumbers identify SnO

 while the other spectra with additional Raman bands result from additional elements and varying lattice compositions in the crystal. Raman image parameters: 200 × 200 µm

; 200 × 200 spectra, integration time of 80 ms per spectrum, 1 mW laser power.

	Iron ore contains many minerals, including variations of iron oxide. Thus, before processing iron ore, an extensive mineralogical characterization is required. Today, this characterization is done by optical microscopy and SEM (8). Differentiation of iron ore minerals that have similar oxygen contents with SEM is difficult. However, Raman spectroscopy can easily distinguish iron oxides from one another. Here, a piece of hematite (Fe

) was analyzed first with SEM, then with the integrated confocal Raman microscope as shown in Figure 3. SEM admittedly depicts some structural characteristics, but cannot differentiate between the oxides. However, the Raman spectra indicate the occurrence of several crystal forms of hematite and goethite (FeO(OH)). The distribution of these minerals is shown in the Raman image that has been overlaid onto the SEM image.


	Figure 3: Raman-SEM image of hematite. The analyzed sample consists of several crystal forms of hematite (red, blue, green, orange, pink) and of goethite (light blue, cyan). The Raman image was derived from the Raman spectra acquired with the Raman microscope of the combined Raman-SEM system and overlaid onto the SEM image. Raman image parameters: 50 x 50 µm

; 150 × 150 spectra, integration time of 150 ms per spectrum, 10 mW laser power.

	The shiny opalescence of pearls results from the mineral aragonite, an orthorhombic polymorph of calcium carbonate that is produced by a biomineralization process. This structure can be visualized with SEM. Some pearls show biomineralization defects characterized by reduced iridescence (also called "milky pearl") (9). These defects are related to the change in the mineralization form of calcium carbonate, from orthorhombic aragonite to hexagonal vaterite. Such a defect was studied with correlative Raman imaging and SEM. The crystallographic structures of the two polymorphs are indicated in the SEM image (Figure 4). The polymorphs have different Raman spectra, so they can be easily identified, imaged, and correlated with the morphology determined from the SEM data. Furthermore, small changes in relative Raman peak intensities highlight anisotropies in the aragonite phase.


	Figure 4: Raman-SEM imaging of a pearl's surface. Raman spectroscopy can differentiate not only aragonite (blue, red) and vaterite (green), two polymorphs of calcium carbonate, but also different phases of aragonite in the pearl. The image is an overlay of the SEM image of the pearl's surface with the confocal Raman image that was created from the Raman spectra at each pixel of the ROI. Raman image parameters: 40 × 40 µm

; 150 × 150 spectra, integration time of 100 ms per spectrum, 5 mW laser power.

	Here, the feasibility and usefulness of correlative Raman imaging and scanning electron microscopy was explored on several samples, namely cassiterite, hematite, and CaCO

polymorphic components of a pearl. All three example measurements reveal the power of Raman imaging as a complementary method to EDS in the determination of the molecular composition of minerals. The locating of trace elements in the cassiterite sample by combining confocal Raman imaging with SEM and EDS imaging is demonstrated. The sensitivity of Raman imaging to polymorphism and small changes to crystal lattice vibrations is highlighted in the measurements on iron ore and pearl. Because Raman imaging reveals the kind of molecules in a sample and their locations in detail, these chemical data can be precisely correlated with the morphological features revealed by high-resolution electron microscopy and the elemental distribution obtained by EDS.

Scanning Electron Microscopy and X-Ray Microanalysis

, 3rd Ed. (Springer, Heidelberg, Dordrecht, London, and New York, 2003).

, J. Toporski, T. Dieing, and O. Hollricher, Eds. (Springer Int. Publishing, Cham [CH], 2018).

	(5) J. Jiruse, M. Hanicinec, M. Havelka, O. Hollricher, W. Ibach, and P. Spizig, 

Software Requirements and Data Analysis in Confocal Raman Microscopy

, 2nd Edition, J. Toporski, J. Dieing, and O. Hollricher, Eds., (Springer Int. Publishing, Cham [CH], 2018).

	(7) G. Wille, C. Lerouge, and U. Schmidt, 

	(8) E. Donskoi, J. Manuel, P. Austin, A. Poliakov, M. Peterson, and S. Hapugoda, 

 is with BRGM in Orleans, France. Direct correspondence to: 

Confocal Raman imaging and scanning electron microscopy are techniques that are ideally suited to correlating structural and chemical information as shown here on a variety of samples.

Exploring the Unknown with Correlative Raman Imaging and Scanning Electron Microscopy

Rheology coupled with in situ Raman spectroscopy (rheoRaman) was used to examine the isothermal crystallization of cocoa butter. Two unique isolation protocols were used to form cocoa butter polymorphs form III and form IV. The hyphenation of these two independent analytical techniques, rheology and Raman spectroscopy, enabled a more holistic depiction of the crystallization process and provided unique insights into the formation of cocoa butter polymorphs.

	Cocoa butter is an edible vegetable fat extracted from the cocoa bean. Cocoa butter is commonly used in home and personal care products (such as ointments and lotions) and it is a vital ingredient in chocolate. Cocoa butter forms the continuous phase within chocolate confections and is responsible for the chocolate's texture, snap, gloss, melting behavior, and resistance to fat bloom. These physical characteristics are a direct result of cocoa butter's triacylglycerol (TAG) composition and overall crystalline structure.

	In general, TAG molecules assume a tuning fork configuration and the TAG "forks" assemble to form crystal lattice structures. During crystallization, the TAG molecules slow down as the cocoa butter oil cools and the TAGs come to rest in contact with one another, forming subcrystalline cells (1). After the subcells are formed, they are thermodynamically driven to aggregate into larger and more-stable crystalline structures (2). The self-assembly of subcell structures and their further aggregation is governed by a balance of intra- and intermolecular interactions. Depending on the molecular level packing and orientation of the TAGs, cocoa butter can form different types of crystal lattice structures, commonly referred to as 

	Researchers have identified six unique cocoa butter polymorphs (3)- form I, II, III, IV, V, and VI-each with a distinct molecular signature. Form V is the polymorph most commonly found in commercially available chocolates, and it is believed that all other polymorphs lead to inadequate properties or a shortened shelf life. For example, form IV is often associated with poorly or under-tempered chocolate; whereas form VI is considered to cause fat bloom, the greyish surface coating sometimes observed on old or temperature-abused chocolate. As a result, the chocolate tempering process is critical because it ensures that the cocoa butter has assembled into the desired crystalline polymorph.

	Overall, cocoa butter crystallization is a highly complex, multistage process. Understanding the isothermal crystallization behavior of cocoa butter is vital for improving chocolate manufacturing processes and maintaining product quality. Cocoa butter crystallization has been widely studied using a variety of analytical techniques, including differential scanning calorimetry (DSC), X-ray diffraction (XRD), X-ray scattering, rheology, and Raman spectroscopy (2,4–7). Although cocoa butter crystallization has been thoroughly examined from the molecular, microstructural, and macroscopic levels, a lack of consensus about the existence of some cocoa butter polymorphs, the overall crystallization kinetics (especially in the presence of additives), and the underlying mechanisms of the crystallization process still persist.

	In this study, rheology coupled with in situ Raman spectroscopy was used to examine the isothermal crystallization of cocoa butter. Raman spectroscopy is a highly sensitive, relatively fast, and nondestructive technique that can probe the molecular structure and conformation in both liquid and solid TAG assemblies, as well as intra- and inter-TAG interactions. With simultaneous Raman and rheological measurements, molecular-level interactions and conformational shifts during the isothermal crystallization of cocoa butter were directly correlated with the changes in bulk viscoelastic properties, providing unique insight into the multifaceted crystallization behavior of cocoa butter.

	Commercially available, organic cocoa butter (

	Rheological measurements were performed using a Thermo Scientific HAAKE MARS 60 rheometer equipped with a serrated 35-mm-diameter plate rotor at a gap height of 1 mm. The serrated plate was used to prevent slip at the sample-rotor interface. All measurements were conducted using oscillatory shear, with a frequency of 1 Hz and a constant strain of 0.1%; data were collected every 10 s. Cocoa butter samples were loaded onto the rheometer at 60 °C and allowed to equilibrate for 10 min to erase any crystal structures or shear history from sample loading. After the equilibrium step, samples were exposed to one of the following temperature ramp and isothermal hold procedures to generate cocoa butter polymorph form III or form IV (2):

		Temperature decreased from 60 °C to 14 °C at a rate of 2 °C/min

		Temperature held constant at 14 °C for 60 min

		Temperature decreased from 60 °C to 22 °C at a rate of 10 °C/min

		Temperature held constant at 22 °C for 120 min

	The melt-to-solid phase transition of cocoa butter was probed rheologically using small amplitude oscillatory shear measurements, where the storage modulus G' and loss modulus G" were measured as a function of time. The overall magnitudes of G' and G", as well as the ratio of G"/G' = tan(δ), determine the general viscoelasticity and overall resistance to deformation for a given material. The term "tan(δ)" is also referred to as the loss or damping factor where δ is the phase angle defined as the shift or lag between the input strain and resultant stress sine waves (or vice versa) during an oscillatory shear measurement. Values of tan(δ) less than unity indicate elastically dominant (solid-like) behavior, whereas values greater than unity indicate viscously dominant (liquid-like) behavior. Unlike the individual moduli, tan(δ) can be used to quantify overall brittleness of a material and is commonly used to assess glass transition behavior. In general, as tan(δ) becomes smaller, the more G' deviates from G", and the more brittle (or glass-like) the material becomes.

	Raman spectroscopy measurements were performed using a Thermo Scientific iXR Raman spectrometer. The iXR system used a 532-nm laser operated at 10-mW power at the sample. Data acquisition and processing were carried out using the Thermo Scientific OMNIC for Dispersive Raman software. Sequential Raman spectra (in parallel with the rheological measurements) were collected over a predetermined time window using the time series collection function of the SERIES software within the OMNIC for Dispersive Raman package. Spectra were collected as an average of four 2-s sample exposures.

	The Raman spectrometer (Thermo Scientific iXR Raman spectrometer) and rheometer (Thermo Scientific HAAKE MARS 60) were coupled together using the Thermo Scientific HAAKE rheoRaman module. The iXR Raman spectrometer was free-space coupled to the rheometer with an optical train that used a series of mirrors to direct the incident laser into the rheoRaman module (Figure 1).


	Figure 1: Schematic diagram of the rheoRaman system (MARSXR) showing side and top views of the rheometer sample stage. The iXR Raman spectrometer is free-space coupled to the MARS rheometer using lens tubes and mirrors that direct light into a 20x objective. The objective focuses the incoming laser (green dashed line) and collects the back-scattered Raman light (yellow) coming out of the sample sitting atop the rheometer stage.

	The sample was positioned between a sandblasted glass bottom plate and the serrated 35-mm plate rotor to avoid slippage at the sample–plate interfaces. An electrical heating element within the rheoRaman module provided temperature control from below the sample, while an active electrical hood was used to provide temperature control from above (eliminating the potential for a temperature gradient within the sample). Cooling of the sample was supplied from a temperature-controlled water bath circulator.

) Raman spectra for the liquid-phase cocoa butter and the two investigated solid phase cocoa butter polymorphs (forms III and IV) are shown in Figure 2. Prominent Raman features were observed in both the C–H stretching region (2700–3050 cm

) and the fingerprint region (1000–1800 cm

). More specifically, the lower Raman shift features include the carbonyl (C=O) stretching region (1700–1800 cm

), and the C–C stretching region (1000–1200 cm


	Figure 2: The full Raman spectra for the melt phase and two polymorphs (forms III and IV) of cocoa butter.

	The C-H stretching regions for the melted and solidified cocoa butter specimens are highlighted in Figure 3. Two strong peaks were observed at ~2850 cm

, which are attributed to symmetric and asymmetric CH

 stretching, respectively (2). The symmetric vibrational modes at 2850 cm

 were dominant in the liquid (melt) phase, indicating reduced lateral interactions between CH

 groups and a low degree of intermolecular interactions. Conversely, the asymmetric vibrations at 2882 cm

 were dominant in the solid phase, which suggests increased lateral interactions resulting from closely packed crystalline structures. Thus, the 2850 cm

 bands are strong indicators of amorphous and crystalline content, respectively (8). Subsequently, the 

 peak intensity ratio was used to dynamically track crystal formation during the in situ rheoRaman measurements.


	Figure 3: Raman spectra of the C-H stretching region (2700–3050 cm

) for the melt phase and two polymorphs (forms III and IV) of cocoa butter. 

	Although they are less intense than the C-H stretching region, approximately eight unique spectral features were identified in the fingerprint region (1000–1800 cm

; Figure 4). When comparing the cocoa butter melt state to the crystalline cocoa butter polymorphs, the most significant changes were observed in the C–C stretching region (1000–1200 cm

). Two well-defined features emerged at 1130 cm

 during the solidification process, which originate from the symmetric and asymmetric C-C stretching, respectively (9,10). In the melt phase, all C-C stretching bands were relatively weak and broad because of the disordering effects of methyl gauche conformations. However, as the cocoa butter solidified, the backbone methyl groups were ordered into the trans conformation, signified by the emergence of the peak at 1130 cm

 spectral marker was also used to track the crystalline-phase transition within cocoa butter via in situ rheoRaman measurements.

 Raman spectral range for the melt phase and two polymorphs (forms III and IV) of cocoa butter.

	To form cocoa butter polymorph form III, the temperature was ramped down from 60 °C to 14 °C at a rate of 2 °C/min. During cooling and before reaching the isothermal temperature of 14 °C, both the elastic and viscous moduli (G' and G", respectively) began to increase exponentially at ~20 min and ~20 °C (Figure 5a). From 20 to 25 min, as the temperature was further decreased and then held at 14 °C, G' and G" increased by approximately six and five orders of magnitude, respectively. The rapid increase in the moduli indicates the start of the solidification process, where the cocoa butter specimen transformed from a semisolid material into a firm solid. After the temperature reached the constant value of 14 °C, the elastic modulus began to plateau at 30 min, whereas the viscous modulus reached a maximum at ~25 min and then gradually decreased from ~25 to 55 min. From 55 min and beyond, both G' and G" remained constant. After 55 min, G' was about two orders of magnitude greater than G", suggesting that the cocoa butter was now a rigid, glass-like solid. From a purely rheological standpoint, the cocoa butter appeared to have fully crystallized.


	Figure 5: (a) Rheology: G' and G" (filled and open circles, respectively; plotted on the left y-axis) and the temperature trace (solid line; plotted on the right y-axis) and (b) Raman: the I

 (right y-axis) intensity ratios for cocoa butter during the melt-crystalline transition to polymorph form III from 60 °C to 14 °C. The vertical dashed line at 20 min indicates the increase in both G' and G" and the Raman intensity ratios.

	The rheological measurements for cocoa butter form III were further corroborated using in situ Raman spectroscopy (Figure 5b). Similarly to G' and G", both the 

 peak intensity ratios remained relatively constant during the first ~20 min of the measurement. Then the 

 spectral markers began to abruptly increase at ~20 min, indicating the formation of crystalline structures within the cocoa butter. As the cocoa butter further crystallized, both spectral ratios continued to increase from 20 to 70 min. After ~70 min, the Raman peak intensity ratios began to stabilize as the cocoa butter TAGs arranged into their final crystal structure. Significant scatter in the 

 ratio, especially at the earlier melt state and at the latter part of the crystallization process. This scatter was most likely because of the low intensity of the 1130 cm

 feature in comparison to the strong Raman bands in the C-H stretching region (2850 cm

	Notably, the initial increase in G' and G" directly aligned with the rise of the 1130 and 2882 cm

 spectral ratios. The concurrent increases in both the rheological response and Raman feature intensity suggests that the bulk solidification during the formation of cocoa butter polymorph form III was directly associated with the formation of crystalline structures. Interestingly, as the change in rheological behavior began to subside at 55 min, the Raman spectral features were still increasing and they did not become steady until ~70 min and beyond. It is postulated that although the overall bulk hardening of the cocoa butter had stopped, the crystalline domains were still rearranging at the molecular level and that the cocoa butter had not reached its final crystalline lattice structure until the end of the measurement.

	To form cocoa butter polymorph form IV, the temperature was rapidly decreased from 60 °C to 22 °C at a rate of 10 °C/min. Note that since the cooling step was so brief, only the data from the isothermal step is shown in Figure 6. During the initial portion of the isothermal hold from 0 to 5 min (immediately following the rapid decrease in temperature from 60 °C to 22 °C), both G' and G" increased as the cocoa butter transformed from a melted liquid to a soft semisolid (Figure 6a). This initial increase in modulus is most likely because of a delay between the set temperature and the internal temperature of the loaded sample. After the sample had reached thermal equilibrium and was at the isothermal set point of 22 °C, the moduli were relatively stable from 10 to 25 min. From 25 to 50 min, however, both G' and G" began to gradually increase and then from 50 to 80 min, the moduli rapidly increased, where G' and G" increased by approximately five and four orders of magnitude, respectively. The exponential increase in the moduli indicates a solidification process, where the cocoa butter transformed from a pliable semisolid to a more robust, hardened solid. At 80 min and beyond, growth in the elastic modulus slowed and eventually plateaued, showing no further significant change past 100 min. The viscous modulus, however, reached a slight plateau from 80 to 100 min but continued to decrease from 100 min and beyond.


	Figure 6: (a) Rheology: G' and G" (filled and open circles, respectively; plotted on the left y-axis) and tan(δ) (plotted on the right y-axis) and (b) Raman: the I

 (right y-axis) peak intensity ratios for cocoa butter during isothermal crystallization at 22 °C. The vertical dashed line at 45 min indicates the increase of G' and G", while the dashed line at 65 min indicates the decrease in tan(δ) and increase in the Raman intensity ratios.

	During the increase in G' and G", a rapid decrease in the loss factor tan(δ) was observed from ~65 min and beyond (Figure 6a, right 

-axis). The decrease in the loss factor indicates a deviation in overall magnitude between G' and G". As the cocoa butter hardened, the increase in G' exceeded the increase in G", triggering the decrease in tan(δ). At the end of the 120 min isothermal study, G' was more than a full order of magnitude greater than G" and the loss factor was approaching 0.01, indicating the cocoa butter had transitioned into a brittle glass-like solid.

	The observed rheological behavior was further confirmed using simultaneous Raman spectroscopy (Figure 6b). Initially, both the 

 peak intensity ratios remained unchanged during the first ~65 min of the isothermal study. Then a sharp increase of the 

 ratios began at ~65 min, indicating the formation of crystal structures within the cocoa butter. As the cocoa butter further crystallized, both spectral markers continued to increase from 65 to 100 min. Beyond 100 min, the growth in both Raman features had subsided and the peak intensity ratios began to stabilize.

	Overall, the rate of increase in the 1130 and 2882 cm

 peak intensity ratios were similar to the rate of change for both G' and G" during the formation of cocoa butter polymorph form IV (Figures 6a and 6b). However, there was a noticeable 15–20 min lag between the observed increase in G' and G" and the rise of the Raman intensity ratios. The sharp upturn in G' and G" indicates an increased resistance to deformation (that is, a bulk hardening of the cocoa butter), signaling the start of the solidification process. The Raman spectral markers, on the other hand, are indicators of crystal formation. Thus, the time delay between the rheology and Raman profiles suggests that cocoa butter first hardens into an amorphous solid, followed by a transformation from an amorphous to a crystalline solid. This morphological transformation was signified by the subsequent increase in the Raman band intensities associated with crystal cocoa butter structures (the 1130 and 2882 cm

peaks). The temporal separation of the rheological and Raman spectral profiles indicates a clear distinction between bulk hardening of the cocoa butter and the formation of crystalline domains during the formation of polymorph form IV.

	Interestingly, the increase in the Raman spectral features (

) directly correlated with the observed reduction in tan(δ) (Figures 6a and 6b). The loss factor is an indication of material brittleness and crystalline structures are commonly known to be brittle. Thus, it is reasonable to infer that the formation of crystal domains at the molecular level (as indicated by Raman) coincides with the overall brittleness of the cocoa butter. As a result, the loss factor may be a more revealing indicator of bulk cocoa butter crystallization than G' and G" alone.

	Simultaneous rheology and Raman spectroscopy measurements were used to examine the isothermal crystallization of cocoa butter. This multimodal analytical technique allowed the bulk mechanical properties of cocoa butter (G', G", and tan[δ]) to be directly correlated with conformational changes at the molecular level (

) in real-time. With a slow cooling (2 °C /min) and subsequent isothermal hold at 14 °C, the concurrent increases in both the rheological response and Raman feature intensity suggests that the bulk solidification during the formation of cocoa butter polymorph form III coincided with the formation of crystalline structures. With a rapid cooling (10 °C/min) followed by an isothermal hold at 22 °C, however, there was a noticeable time lag between the rheological response (G' and G") and the Raman spectral profiles during the formation of cocoa butter polymorph form IV. The observed time delay indicates that cocoa butter crystallized by first hardening into an amorphous solid, manifested by a sharp increase in G' and G" while the Raman features remained unchanged. The amorphous solid then underwent a morphological transition to form a crystalline solid, signified by the increase in Raman features associated with crystal cocoa butter structures (1130 and 2882 cm

). Without coupling these two separate analytical techniques, the observed amorphous-solid to crystalline-solid transformation would have been left undetected. Alone, each technique suggests a single stage process, however, only when the two techniques are coupled is the multiphase crystallization process revealed, further exemplifying the unique analytical capability unleashed by hyphenating rheology with in situ Raman spectroscopy.

Crystallization of Lipids: Fundamentals and Applications in Food, Cosmetics, and Pharmaceuticals

 (John Wiley & Sons, Hoboken, New Jersey, 2018).

	(2) S. Bresson, D. Rousseau, S. Ghosh, M. El Marssi, and V. Faivre, 

	(4) K. Dewettinck, I. Foubert, M. Basiura, and B. Goderis, 

The hyphenation of rheology and Raman spectroscopy enabled a more holistic depiction of the crystallization process and provided unique 
insights into the formation of cocoa butter polymorph form IV.


Special Issues-06-01-2018

Portable Raman Spectroscopy for At-Line Characterization of Carbon Nanomaterials

Data-Driven Raman Spectroscopy in Oil and Gas: Rapid Online Analysis of Complex Gas Mixtures

Exploring the Unknown with Correlative Raman Imaging and Scanning Electron Microscopy

Gaining Insight into Cocoa Butter Polymorph Formation Through In Situ Rheology and Raman Spectroscopy

Spectroscopy June 2018 Raman Technology issue PDF