Open Access
Add to BibSonomy
Abstract
We propose a high-resolution, broad-spectral-range spatial heterodyne Raman spectrometer (SHRS) having separate filters and multi-gratings (SFMG). A prototype of the SFMG-SHRS is built using multi-gratings with four sub-gratings having groove densities of 320, 298, 276, and 254 gr/mm and separate filters with filter bands corresponding to the sub-gratings. We use the SFMG-SHRS to measure the Raman spectra of inorganic and organic compounds with various integration times, laser power, and transparent containers, compare measurements of microplastics with and without the separate filters, and measure mixtures of inorganic powders and organic solutions. The designed SFMG-SHRS makes high-resolution, broad-spectral-range Raman measurements with improved signal-to-noise ratios and visibility of weak Raman peaks even in the presence of fluorescence.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Since the concept of the inelastic scattering of light was proposed [1] and discovered by the Indian scientist Chandrasekhara Venkata Raman in 1928 [2], the technology of the Raman spectroscopy has been rapidly developed and widely used in many fields, including industry [3], geology [4,5], biology [6,7], and physics [8,9]. As an improved version of Fourier interferometer with no moving parts, the spatial heterodyne spectrometer (SHS) is based on the Michelson interferometer, in which the mirrors in the two interference arms has been replaced by two dispersion grating [10,11]. The high stability, compact structure, large throughput, and broad spectral range or high spectral resolution makes the Raman spectrometer based on SHS (SHRS) emerge as the times require [12] and has been developed rapidly. Experiment results of many literatures prove that SHRS are suited to material analysis and is widely used in the identification of minerals [13,14], measurement of chemical compounds [15,16], planetary exploration [17], and remote sensing of the atmospehre [18].
Although the SHRS performs excellently in measuring the Raman spectrum, owing to the restriction of the Nyquist sampling theorem [19], it is not possible to achieve both high spectral resolution and a broad spectral range simultaneously in one measurement process of the conventional SHS. For example, an SHS with a spectral resolution of 7.11 cm−1 and 1360 pixels has a spectral range of ∼4500 cm−1 [20], which satisfies the band measurement requirements of the vast majority of samples, but its low spectral resolution is not suited to the identification of mixed targets [21] or feature peaks having small wavenumber intervals. In contrast, an SHS with a high spectral resolution of 2.47 cm−1 and 1024 pixels has a spectral range of 1262 cm−1 [14] and misses strong Raman peaks such as that of the C–H stretching of organic compounds near 3000 cm−1 [22]. The concept of a multi-grating [23] has been introduced to address the problems of spectral resolution and measurement band constraints of the conventional SHS.
A multi-grating is a collection of sub-gratings with different groove densities, allowing for the simultaneous measurement of multiple measurement bands in a single measurement process, ultimately achieving a high-spectral-resolution and broad-spectral-range measurement. The measurement ranges of the different sub-gratings are relatively independent of one another and the spectral range in which each sub-grating can measure accounts for a part of the overall spectral range. However, there is a problem with this arrangement. When measuring with a normal grating in the SHS, we can improve the signal-to-noise ratio (SNR) by adding a filter in the collection light path to filter out light outside the measurement range. In contrast, when measuring with a multi-grating in the SHS, owing to the characteristics mentioned above, even if a filter is added in the collection light path, there remains much light for each sub-grating that does not fall within the measurement range, and the light outside the measurement range of the sub-grating becomes background or stray light that interferes with the measurement.
To solve the above issue, in the present paper, we propose a new type of SHRS based on separate filters and multi-gratings (SFMG-SHRS) and report an experiment performed on the SHRS. The filter band of each separate filter was adapted to the detectable spectral range of the corresponding sub-grating for filtering in the detectable spectral range of each sub-grating during a single measurement process. In comparison with the conventional SHRS, the SFMG-SHRS not only makes high-resolution and broadband measurements but also increases the SNR of the spectra and weakens the effect of fluorescence in the measurements. The paper describes the design principles of the SFMG-SHRS in detail. In this study, each multi-grating had four sub-gratings with groove densities of 320, 298, 276, and 254 gr/mm, and corresponding separate filters were used. After building a prototype, we measured inorganic compounds and organic solutions for various integration times, laser power, and transparent containers. A comparative experiment was carried out for the measurement of microplastics with and without the separate filters, and relevant discussions are presented in the paper. In addition, Raman spectra of mixtures of inorganic compounds and organic solutions were measured.
2. Principle
2.1 Basic theory
Figure 1 presents the measurement optical path of the SFMG-SHRS, which comprises two parts connected by a fiber. In the excitation and collection optical path, the combination of a linear variable filter, mirror, and collimation lens L1 is convenient for adjusting the focusing position and laser power on the sample. In addition, the collimation lens L1 collects the excited Raman signal light. After being filtered by the filters, the Raman signal light is converged by collimation lens L2 and enters the input port of the fiber. Beyond the output port of the fiber is the principle prototype of the SFMG-SHRS, which basically has the structure of a Michelson interferometer. The output signal light is first collimated by collimation lens L3 set at the focal distance from the output port of the fiber and is then split into two beams by a beam splitter. The two beams pass through field-widening prisms Pi (i = 1, 2) and separate filters and then diffracted by multi-gratings MGi before returning. As shown in Fig. 1, each multi-grating MGi has four sub-gratings SGij (j = 1, 2, 3, 4) with different groove densities and is set in the Littrow configuration, which is expressed as
where αL is the Littrow angle of multi-grating MGi and σLj and 1/dj are respectively the Littrow wavenumber and groove density of the corresponding sub-grating SGij. The Littrow wavenumber σLj = 1/λLj, where λLj is the Littrow wavelength. Equation (1) is the diffraction equation of the signal light with the Littrow wavenumber. For signal light with an off-Littrow wavenumber σ, the diffraction equation isVertically tilting multi-grating MG1 by an angle ɛ introduces a spatial frequency fy and generates a two-dimensional interference fringe, which avoids the ambiguity associated with “ghost” and “true” spectra of the one-dimensional interference fringe. The spatial frequency fy is expressed as
Fig. 1. Configuration of the measurement optical path of the spatial heterodyne Raman spectrometer including separate filters and multi-gratings.
On the basis of the above analysis, the two-dimensional interference fringe is written as
The spectral resolution of a spectrometer is a performance indicator of its ability to qualitatively distinguish two spectral peaks that are very close to each other. This resolution is limited by the beam’s divergence angle and the maximum optical path difference (OPD) [24]. Adopting the definition of the resolution limit, the spectral resolution δσ and resolution power R of the SFMG-SHRS are given by
Here, WMG is the effective width of the multi-grating used for imaging, which is calculated as
where WIA is the width of the image area on the detector. Owing to the two-dimensional interference fringes generated by the vertical tilt of multi-grating MG1, the spectral range Δσ of each of the two sub-gratings SG1j and SG2j is determined asThe throughput of a spectrometer is an important consideration. It is measured as the number of signal light rays received by the spectrometer and strongly relates to the sensitivity of the designed system. The SFMG-SHRS is an improved version of the conventional SHS and has the same collection solid angle as the conventional SHS [25], which is expressed as
where ΩSHS is the collection solid angle of the conventional SHS. The optical throughput of the SFMG-SHRS is then written as where TSFMG is the throughput of the SFMG-SHRS, AMG is the effective area of the entrance aperture, and WMG and HMG are respectively the effective width and height of the multi-grating used for imaging. The field of view and thereby collection solid angle are greatly increased by introducing a pair of field-widening prisms [11,26]. The rotation angle and apex angle of field-widening prism Pi in the optical path are calculated according to where n is the refractive index of the field-widening prism, γp is the angle of rotation between the normal of the prism’s incident surface and the optical axis, and αp is the apex angle of the field-widening prism. By introducing the field-widening prisms with parameters based on the calculations of Eqs. (1), (12), and (13) in the spectrometer design, the field of view is increased by two orders of magnitude [26].2.2 Calibration theory
To estimate the performance of the designed SFMG-SHRS on an experimental breadboard, a calibration procedure was performed for the spectrometer. The calibration procedure establishes the spectral response of the designed spectrometer for a standard calibration light source (e.g., a neon or mercury lamp). In the calibration process, after two known emission lines are obtained from the Fourier transform of the interference fringes generated by sub-gratings SG1j and SG2j, the Littrow wavelength and corresponding Littrow wavenumber for these two sub-gratings are calculated according to
On the basis of Eq. (1), when a groove density of 1/dj is assumed, the Littrow angles of the sub-gratings and the overall multi-grating are obtained according to
Combining Eqs. (6) and (15), the width of the image area on the CCD is calculated as
After determining the Littrow angle of the multi-grating and image width on the detector, the effective width of the multi-grating is calculated using Eq. (8). According to the above analysis, when a pair of sub-gratings SG1j and SG2j are calibrated and the groove densities of all sub-gratings are known parameters, the Littrow wavenumbers of the other sub-gratings can also be calculated using Eq. (1).
3. Experiment
3.1 Design
When designing the parameters of the multi-gratings, we need to combining the spectral performance required by SFMG-SHRS with the parameter of the optical elements used to build the experiment breadboard.
The first step is to determine the size parameter of the multi-grating based on the detector used for the experiment and the ruling technique of the multi-grating. According to the Eq. (6)and Eq. (9), we can clearly see that the width of the imaging and the pixel number of the detector are the key parameters. The CCD we planned to use in the experiment is iKon-L 936 of Andor, which has a detection area of 27.6 × 27.6 mm2 and Np × Np = 2048 × 2048 pixels. Paired with an imaging lens group, it can easily achieve precise imaging of regions ranging from 30 to 50 mm2 at the image plane. In terms of the ruling multi-grating, the height of each sub-grating is ∼1/4 times the total height of the multi-grating. Considering that the ruling distance (i.e., the height) of each sub-grating should not be too short, as it would be difficult to engraving grating grooves, and there should be a margin left for the clamping area when the multi-grating is fixed. So, the ruling area of the multi-grating is determined as 50 × 50 mm2, and the effective width of the multi-grating used for imaging WMG is determined as ∼45 mm.
After determining the size parameter, the suitable groove densities and the Littrow angle of the multi-grating can be selected by choosing multiple groups of parameters to calculate the appropriate spectral performance. Due to the introduction of the multi-grating with four sub-gratings, at the beginning of the experiment, we designed a broad overall spectral range of ∼4500 cm−1 and a high spectral resolution of ∼0.7 cm−1. Based on the analysis in section 2.1, to ensure the integrity of the obtained Raman spectrum, the spectral range Δσ of each sub-grating is about 1300 cm−1. Based on Eq. (9), when the Δσ, Np, and the WMG are determined, the spectral resolution δσ and the Littrow angle αL can be calculated as
The laser power we planned to use to excite the Raman light is 532 nm, so the wavenumber of 532 nm can be seen as the zero wavenumber in the spectral range. Each sub-grating's detectable spectral range expands symmetrically around its Littrow wavenumber σLj, covering from σLj - Δσ/2 to σLj + Δσ/2. If we set 532 nm as the Littrow wavelength λL1 of the sub-grating SGi1, a significant portion of the detection spectral range will be wasted in the negative wavenumber region. Therefore, based on previous experimental experience, we chose the Littrow wavelength λL1 = 546 nm. By substituting the σL1 and the αL in Eq. (1), the groove density 1/d1 can be calculated as
Since the groove density was supposed to be an integer, we choose 1/d1 as 320 mm−1, and the Littrow angle and the spectral resolution in this condition are shown in Table 1. To guarantee the integrity of the of the obtained spectrum, the wavenumber difference between the Littrow wavenumber of the SGij and SGij + 1 should less than 1300 cm−1, ensuring a substantial spectral overlap range. We set the difference between two adjacent σLj and σLj + 1 as ∼1260 cm−1, ensuring a spectral overlap area of 40 cm−1. Considering that groove densities must be integers, we conducted numerous calculations and precise adjustments of the value of σLj and 1/dj using Eq. (1). As a result, we determined the groove densities and their corresponding Littrow wavenumbers, which are presented in Table 1. The Littrow wavelengths can also be derived accordingly.
Table 1. Designed parameters of the multi-gratings
3.2 Breadboard
Figure 2 presents the optical layout of the experimental breadboard. The important parameters of the optical components set on the breadboard are listed in Table 2. In the excitation optical path, the power supplied to the solid-state 532-nm laser (Changchun New Industries Optoelectronics Tech. Co., Ltd) was linearly adjusted from 0 to 400 mW with a linear variable filter and collimated onto the sample by collimation lens L1 having a 15-mm focal length. In the collection optical path, we placed two 532-nm long-pass edge filters (532-LAB-80AC-2.5, CNI) and one 700-nm short-pass filter (84-714, Edmund) after the collimation lens L1 to reject the laser light and fluorescent light, respectively. The input port of a 1.5-mm-diameter silica fiber was placed at the focal plane of collimation lens L2 with a 50-mm focal length to receive and transfer the filtered signal light.
Fig. 2. Experiment breadboard of the spatial heterodyne Raman spectrometer based on separate filters and multi-gratings.
Table 2. Essential parameters of the optical elements used in the experimental breadboard system
The collimation lens L3 with a 150-mm focal length was placed at the focal length after the output port of the silica fiber to collimate the emitted light and transfer it into the prototype of the SFMG-SHRS. The SFMG-SHRS comprised a cubic beam splitter with dimensions of 50.8 × 50.8 × 50.8 mm3 (model no. 20BC17MB.1, Newport), four pairs of filters with different pass bands, two multi-gratings set in the Littrow configuration, and one multi-grating vertically tilted by a small angle ɛ to generate two-dimensional interference fringes. Figure 3 presents the separate filters and multi-gratings manufactured by the Changchun Institute of Optics, Fine Mechanics and Physics. In addition, two fixing frames (Changchun UP Optotech (Holding) Co., Ltd) were used to fix the filters for the different sub-gratings. The CCD was cooled to −60 °C to reduce thermal noise, and a camera lens (AF-S Micro Nikkor 105mm 1:2.8G ED, Nikon) was used to image the two-dimensional interference fringes.
The two-dimensional fast Fourier transform was adopted to transfer the two-dimensional interference fringes to the two-dimensional spatial frequency domain. To accurately convert the frequencies into the corresponding spectrum, a phase correction method was adopted in the restoration of Fourier-transform spectroscopy [27], and the wavelet transform was then used to correct the baseline of the signal and remove the background [28].
3.3 Calibration
Figure 4 presents the interference fringes and corresponding calibration results of the neon lamp for the SFMG-SHRS. Figure 4(a) shows that the two-dimensional interference fringes were evenly divided into four parts corresponding to the four pairs of sub-gratings. The interferogram of the sub-gratings SGi2 used for two-dimensional Fourier transform has an area size of 2048 × 512, and was transferred to the spatial frequency domain by fast Fourier transform as shown in Fig. 4(b). Since the interference fringes are tilted, that is, two-dimensional distribution, the positive and negative of the corresponding spatial frequency fx distribution represent the two directions of the fringe tilt. By taking the spatial frequency fx at fy = -10 and plot it in one dimension to determine its correspondence with the Raman peaks in the standard neon spectrum for restoration, thereby obtaining the corresponding Raman spectral distribution. In a comparison with the standard spectrum of the neon lamp, we see that the spatial frequencies of 350 and 626 correspond to the wavelengths of 588.189 and 594.483 nm. On the basis of Eqs. (14)–(16) and the groove density of SGi2, we calculate the Littrow wavelength λL2 = 580.397 nm, spectral resolution δσ = 0.6522 cm−1, and Littrow angle αL = 4.9611°. The width of the image area on the CCD can be calculated as WIA = 44.16 mm, and the width of the imaged multi-grating is then calculated as WMG = 44.33 mm. Using Eq. (1), the Littrow wavelengths of other sub-gratings are obtained as λL1 = 540.494 nm, λL3 = 626.660 nm, and λL4 = 680.938 nm. Considering the error in instrument alignment and adjustment, the instrument calibration results are generally consistent with the designed parameters in Table. 1.
Fig. 4. (a) Two-dimensional interference fringes of the neon lamp produced by the SFMG-SHRS. (b) Two-dimensional spatial frequencies obtained through the fast Fourier transform of the interference fringes on SGi2 and one-dimensional spatial frequencies distribution at fy = −10. (c) Measured spectrum on SGi3 for a mercury lamp after calibration.
The fringes in Fig. 4(a) and final spectral lines obtained in Fig. 4(c) show that the interference fringes of the different sub-gratings were well restored to the corresponding spectrum, the spectral ranges of the different sub-gratings partially overlapped, and the overall detectable spectral range was continuous. Setting the wavenumber of 532 nm as the zero wavenumber, the overall spectral range was from −372.41 to 4779.18 cm−1, and the overlap in the spectral ranges of each pair of adjacent sub-gratings was 63.662 cm−1. The full width at half maximum at 640.225 nm was approximately 1.281 cm−1, which was in almost exact agreement with the theoretical spectral resolution.
The designed SFMG-SHRS has resolution power R = 28,821 (obtained using Eq. (7)) and a solid angle ΩSHS = 2.18 × 10−4 sr (obtained using Eq. (10)). According to the height parameter HMG = WIA = 44.16 mm and the width parameter WMG = WIA/cosαL = 44.33 mm, the throughput of the SFMG-SHRS is TSHS = 4.268 × 10−3 cm2 sr. The Kaiser Holospec f/1.8 has one of the highest throughputs among commercially available dispersive spectrometers [25] and makes for a useful comparison with the SFMG-SHRS. For a slit-based dispersive spectrometer, the spectral resolution δσDS and collection solid angle ΩDS are expressed as [29]
where RDS and Ws are respectively the reciprocal linear dispersion and the width of the entrance slit of the dispersive spectrometer. The Kaiser Holospec f/1.8 has an F-number of f/1.8 and thus a solid angle ΩDS = 0.2424 sr. The grating commonly used in the Holospec has RDS = 2.4 cm−1pixel−1, and a detector with a pixel pitch a = 13.5 µm gives RDS = 0.1778 cm−1µm−1. A slit no wider than 1.834 µm is thus required to achieve a spectral resolution of 0.6522 cm−1 as for the SFMG-SHRS. The length of the slit of the Holospec is approximately 0.8 cm. Applying Eq. (11), the throughput of the Holospec is obtained as TKH = 3.556 × 10−5 cm2 sr. The throughput of the multi-grating in the SFMG-SHRS is at least two orders of magnitude higher than the throughput of the Holospec. In addition, we also introduced field-widening prisms, which can further expand the throughput of the SFMG-SHRS.4. Results and discussion
4.1 Raman spectral and SNR analyses of sulfur and ethanol for various integration times
Figure 5(a) shows the measured Raman spectra of solid sulfur for different integration times and the same laser power of 30 mW. Considering the inflammability of the solid sulfur, we used a relatively low laser power to study the characteristics of the vibrational Raman spectrum of sulfur for different integration times. The three peaks at 153, 218, and 472 cm−1 are clearly visible and well detected. The strongest Raman peak at 218 cm−1 is due to symmetric bond-bending of the S8 molecule whereas the weaker Raman peaks at 153 and 472 cm−1 are due to the antisymmetric bond-bending mode and symmetric bond-stretching mode of the S8 molecule, respectively [30].
Fig. 5. (a) Measured Raman spectra of solid sulfur for different integration times and the same laser power of 30 mW. (b) Signal-to-noise ratio (SNR) plot of the sulfur Raman spectra for different integration times and the same laser power of 30 mW.
To quantify the signal performance of the Raman spectrum measured by the designed SFMG-SHRS, the SNR of the spectrum is given by
where Ip is the amplitude of the Raman characteristic peak signal and RMSnoise is the root-mean-square (RMS) value of the noise in the recovered Raman spectrum. By taking the amplitude of the maximum Raman characteristic peak at 218 cm−1 and calculating the RMS of the Raman spectrum of sulfur for different integration times, we plot the SNR of sulfur versus integration time as shown in Fig. 5(b). In Fig. 5(b), as the integration time increases from 1 to 32 s, the plot of the SNR of the Raman spectrum is initially steep and then gradually flattens out. The nonlinear relationship curve arises because, while extending the integration time diminishes the influence of certain noise sources, it concurrently introduces additional sources of noise [31]. The rate of the signal increase is not as fast as the rate of noise increase, resulting in the inability to further improve SNR.Furthermore, we compare the Raman spectrum measurement performances of the SFMG-SHRS and a commercial spectrometer [32]. The commercial spectrometer i-Raman Pro has a wavenumber range from 65 to 3400 cm−1 and a spectral resolution finer than 3.5 cm−1@614 nm. As shown in Fig. 6, under the same measurement conditions, the amplitude of the Raman signal at 218 cm−1 measured by the SFMG-SHRS is at least 200 times that of the signal measured by the i-Raman Pro.
Fig. 6. Comparison of the sulfur Raman spectra measured by the designed SFMG-SHRS and the i-Raman Pro instrument of B&W Tek for an integration time of 0.5 s and laser power of 8 mW.
Figure 7(a) shows measured Raman spectra of ethanol (ACS, 99.5%) for different integration times and the same laser power of 160 mW. Figure 7(b) plots the Raman signal in 1600 ∼3200 cm−1 with logarithmic scale for the y-axis to clearly show the noise and peak intensity. The stretching modes of–CC and –CO have peaks at 882 and 1054 cm−1, respectively. The Raman peak at 1097 cm−1 is assigned to the librations and rock vibrations of –CH3. The Raman peaks at 1282 and 1453 cm−1 are assigned to the twisting of –CH2 and the bending vibrations of–CH3 and –CH2, respectively. The Raman peak at 2878 cm−1 is attributed to the symmetric stretching modes of –CH2 and –CH3 groups, and the symmetric and antisymmetric stretching of –CH3 groups has peaks at 2930 and 2976 cm−1, respectively [33]. Figure 7(a) clearly shows that the Raman intensity of the C–H stretching band (2800∼3000 cm−1) is stronger than that of the C–H bending band (1200∼1500 cm−1) and that of the C–O and C–C stretching bands (800∼1100 cm−1).
Fig. 7. (a) Measured Raman spectra of ethanol for different integration times and the same laser power of 160 mW. (b) The Raman signal in 1600 ∼3200 cm−1 plotted with y-axis in log scale.
4.2 Raman spectral analyses of titanium dioxide and fumaric acid for various laser power
Figure 8(a) shows the measured Raman spectra of titanium dioxide (TiO2, 99.8%, metals basis) for different laser power and the same integration time of 50 s. The strongest Raman peak at 144 cm−1 is assigned to the O–Ti–O bending vibration mode v6(Eg). Two other O–Ti–O bending vibration modes v5(Eg) and v4(B1g) have peaks at 197 and 397 cm−1, respectively. The two close Raman peaks centered at 513 and 519 cm−1 are assigned to the Ti–O bond stretching vibration modes v3(A1g) and v2(B1g), respectively. In addition, the Raman peak at 639 cm−1 is assigned to the Ti–O bond stretching vibration mode v1(Eg) [30,34]. As in Fig. 7(a) for ethanol, the Raman peak at 197 cm−1 is difficult to identify when the laser power is lower than 40 mW, but the character of the titanium dioxide Raman spectrum can still be identified when the laser power is 20 mW because other Raman peaks are still clearly visible, even though the amplitudes of the peaks are lower than those at laser power of 40 mW.
Fig. 8. (a) Measured Raman spectra of titanium dioxide for different laser power and the same integration time of 50 s. (b) Measured Raman spectra of fumaric acid for different laser power and the same integration time of 12 s.
Figure 8(b) shows the measured Raman spectra of fumaric acid (C4H4O4, AR, 99.5%) for different laser powers and the same integration time of 12 s. The assignment of the Raman peaks of the fumaric acid is given in Table 3 [35,36]. Figure 8(b) and Table 3 show that the vibrational bands of the out-plane bending appear at 650∼1000 cm−1, the vibrational bands of the in-plane bending and rocking appear at 1300∼1500 cm−1, and the stretching bands mainly appear at 1600∼3100 cm−1. The Raman characteristic peaks are well measured as the laser power increases from 20 to 200 mW for the same integration time of 12 s. The Raman spectral analyses of titanium dioxide and fumaric acid thus show that the SFMG-SHRS is capable of detecting inorganic compounds and organic solutions.
Table 3. Assignment of the fumaric acid Raman characteristic peaksa
4.3 Raman spectrum detection and analysis of microplastics with and without separate filters
There are various types of microplastic with complex characteristic Raman peaks, and many of them exhibit strong fluorescence, making it difficult to conduct Raman measurements. We measured four types of microplastic, namely polyoxymethylene (POM), polyphenylene oxide (PPO), polycarbonate (PC), and polyamide (PA). In the measurement of each type of microplastic, the sample was fixed at the same position for the same integration time and laser power in two situations, namely with the separate filters set in front of the multi-grating as shown in Fig. 1 and without the use of the separate filters). By taking the magnitude of the highest Raman peak in each spectrum as a value of 1, we normalize the Raman spectra of the microplastics as shown in Fig. 9, the Raman spectrum of the microplastics obtained in the two situations have not undergone baseline correction processing. There are clear differences in the noise level and peak values between the measurements with and without separate filters. For each Raman spectra shown in Fig. 9, the SNR is calculated by substituting the amplitude of the maximum peak and the RMS of the noise in Eq. (22). The assignment of the Raman peaks of the microplastics is given in Table 4.
Fig. 9. Normalized Raman spectra of microplastics obtained with (red line) and without separate filters (blue line) under the same laser power of 100 mW: (a) polyoxymethylene with an integration time of 30 s, (b) polyphenylene oxide with an integration time of 15 s, (c) polycarbonate with an integration time of 80 s, (d) polyamide with an integration time of 90s.
Table 4. Assignment of the microplastics Raman characteristic peaksa
Figure 9 and Table 4 show that the Raman peaks assigned to the C–H stretching vibration mode of the microplastics in the high wavenumber range (2800∼3000 cm−1) are clearly higher than the Raman peaks assigned to other vibration modes in the low wavenumber range (500∼1800 cm−1). The assignment of the Raman peaks in the low wavenumber range is more complex in that besides the stretching mode, there are also deformation, bending, rocking, and twisting modes. In the low wavenumber range, the Raman peaks assigned to bending and stretching modes in the 1300∼1700-cm−1 band are much higher than the other Raman peaks.
As shown in Fig. 9(a) and 10(c), the amplitudes of the Raman peaks of POM and PC measured with and without sperate filters are not greatly different, but there is an obvious difference in the noise level between the two situations. In Fig. 9(a), the SNRs of the POM Raman spectrum measured with and without the separate filters are 89 and 44, respectively. In Fig. 9(c), the SNRs of the PC Raman spectrum measured with and without separate filters are 121 and 69, respectively. In Fig. 9(a) and 9(c), the SNR of the spectrum measured with the separate filters is twice that of the spectrum measured without the separate filters.
Fig. 10. Measured Raman spectra of sodium sulfate in various types of container for an integration time of 50 s and laser power of 120 mW.
In Fig. 9(b) and 9(d), the differences in the noise level between the two situations are similar, but the amplitudes of the PPO and PA Raman peaks in the low wavenumber range measured with the separate filters are clearly higher than those measured without the separate filters. In Fig. 9(b), the SNRs of the PPO Raman spectrum measured with and without separate filters are 139 and 110, respectively. The PPO Raman peaks at 835 and 962 cm−1 in the spectrum measured with the separate filters are easily identified whereas those in the spectrum measured without the separate filters cannot be identified from the background noise. In Fig. 9(d), the SNRs of the PA Raman spectrum measured with and without separate filters are 157 and 137, respectively. The PA Raman peak at 930 cm−1 in the spectrum measured with the separate filters is easily identified whereas that in the spectrum measured without separate filters cannot be identified from the background noise.
The analysis above reveals that the SFMG-SHRS weakens the effect of fluorescence in the measurement of the microplastics and thus increases the SNR and improves the visibility of the weak Raman signal. The SFMG-SHRS is designed to effectively detect a sample such as a microplastic sample with fluorescent properties, which is of importance to industrial development, material testing, and environmental protection.
4.4 Raman spectrum detection and analysis of sodium sulfate powder and methyl salicylate in different containers
Figures 10 and 11 respectively present the Raman spectra of sodium sulfate (Na2SO4) and methyl salicylate in different transparent containers, namely a glass bottle, plastic bottle, and plastic bag. The bottles had thickness of 1 mm and the plastic bag had thickness of 0.02 mm.
Fig. 11. Measured Raman spectra of methyl salicylate in various types of container for an integration time of 8 s and laser power of 200 mW.
Figure 10 presents the Raman spectra of sodium sulfate (AR, 99%) measured for an integration time of 50 s and laser power of 120 mW. There are obvious vibration bands corresponding to different vibration and stretching modes. The main Raman peak at 992 cm−1 is due to the symmetric stretching mode v1(A1). In the lower wavenumber region (400∼700 cm−1), the bending vibration mode v2(E) has peaks at 449 and 466 cm−1 and the bending vibration mode v4(F2) has peaks at 620, 632, and 647 cm−1 [46]. In the higher wavenumber region (1000∼1200 cm−1), the anti-symmetric stretching mode v3(F2) has peaks at 1101, 1131, and 1152 cm−1. The measured SNRs of the main Raman peaks at 992 cm−1 for the glass bottle, plastic bottle, and plastic bag are 367, 185, and 437, respectively.
Figure 11 presents the Raman spectra of methyl salicylate (AR, 99%) measured for an integration time of 8 s and laser power of 200 mW. There are obvious main vibration bands. The assignment of the Raman peaks of the methyl salicylate is given in Table 5 [47]. Figure 11 and Table 5 show that the stretching mode mainly appears in the high wavenumber region (1200∼3100 cm−1) and has a strong Raman peak. The Raman peaks of deformation, rocking, and twisting modes mainly appear in the low wavenumber region (100∼1200 cm−1), with most are much weaker than the Raman peaks assigned to the stretching mode. The measured SNRs of the Raman peak at 2964 cm−1 are 138, 85, and 103 for the glass bottle, plastic bottle, and plastic bag, respectively.
Table 5. Assignment of the methyl salicylate Raman characteristic peaksa
In Fig. 10, Raman peaks assigned to the weak vibration and stretching modes of sodium sulfate are clearly visible for the three different containers. In Fig. 11, Raman peaks assigned to medium and strong deformation modes of methyl salicylate are clearly visible for the three different containers, but the Raman peaks assigned to weak deformation modes at 858, 965, and 1039 cm−1 are not effectively identified for the plastic bottle and plastic bag. In addition, the measured SNR and intensity of the Raman peaks are clearly lower for the sodium sulfate and methyl salicylate stored in the plastic bottle than for the sodium sulfate and methyl salicylate stored in the glass bottle of the same thickness or the plastic bag owing to the higher reflection of the laser and lower transmission of the Raman signal. The designed SFMG-SHRS performs well in measuring inorganic and organic compounds stored in a glass bottle, plastic bottle, or plastic bag and has good potential in the detection of toxic chemicals and hazardous substances stored in transparent containers.
4.5 Raman spectrum detection and analysis of the mixture of inorganic powders and organic solutions
Figure 12 presents the measured Raman spectra of a mixture of sodium sulfate (Na2SO4, AR, 99%), sodium nitrite (NaNO2, AR, 99%), and calcium carbonate (CaCO3, AR, 99%) and the individual inorganic powders for an integration time of 45 s and laser power of 130 mW. These powders were mixed in a 1:1:1 volume ratio. The measured inorganic powders are within the detectable spectral range of sub-gratings SGi1 and SGi2, and all the Raman peaks are clearly visible.
Fig. 12. Measured Raman spectra of a mixture of inorganic powders and the individual inorganic powders for an integration time of 45 s and laser power of 130 mW.
In Fig. 12, in the Raman spectra of the mixture of the inorganic powders, the peaks at 449, 466, 620, 632, 647, 992, 1101, 1131, and 1152 cm−1 correspond to the Raman spectrum of sodium sulfate [46]. The weak vibration modes v2(E), v3(F2), and v4(F2) of the sodium sulfate are still clearly visible in the spectrum of the mixture. In particular, the visibility of the weak Raman peak at 1101 cm−1 is unaffected by the strong Raman peak at 1085 cm−1 beside it. The peaks at 120, 154, 828, and 1328 cm−1 correspond to the Raman spectrum of sodium nitrite [48]. The peaks at 155, 281, 781, and 1085 cm−1 correspond to the Raman spectrum of calcium carbonate [49,50]. The two adjacent Raman peaks at 154 and 155 cm−1 overlap each other in the spectrum of the inorganic mixture because the wavenumber difference between these two peaks is smaller than the full widths at half maximum of the peaks. In addition, in the spectra of sodium nitrite, the amplitude of the Raman peak at 154 cm−1 is only half that at 120 cm−1 whereas in the spectrum of the inorganic powder mixture, the amplitude of the overlapping Raman peaks at 154 and 155 cm−1 is almost equivalent to that of the single Raman peak at 120 cm−1. Obviously, the amplitude of the mixed Raman peak is the superposition of the intensity of the Raman peak of sodium nitrite at 154 cm−1 and that of the Raman peak of calcium carbonate at 155 cm−1. However, this does not prevent us from identifying the composition of the inorganic powders according to the well-distinguished Raman characteristic peaks at 992, 1085, and 1328 cm−1 and characteristic vibration bands in the spectrum of the mixture.
Figure 13 presents the measured Raman spectra of a mixture of cyclohexane (AR, 99.7%), acetone (GC, 99.8%), and carbon tetrachloride (CCl4, GC, 99.6%) and the individual organic solutions for an integration time of 30 s and laser power of 200 mW. These solutions were mixed in a 1:1:1 volume ratio. The measured organic solutions are distributed in the detectable spectral range of sub-gratings SGi1, SGi2, and SGi3, and all the Raman peaks are clearly visible.
Fig. 13. Measured Raman spectra of a mixture of organic solutions and the individual organic solutions for an integration time of 30 s and laser power of 200 mW.
In Fig. 13, in the Raman spectrum of the organic mixture, the peaks at 803, 1030, 1267, 1446, 2853, 2923, and 2938 cm−1 correspond to the spectrum of cyclohexane [51,52], the peaks at 530, 786, 1066, 1220, 1428, 1706, 2848, 2924, 2963, and 3006 cm−1 correspond to the spectrum of acetone [53], and the peaks at 218, 314, 452, 762, and 789 cm−1 correspond to the spectrum of carbon tetrachloride [30,54]. Many of the Raman peaks of the three organic solutions are close to one another. The Raman peak of cyclohexane at 803 cm−1 along with the Raman peak of acetone at 786 cm−1 and the Raman peak of carbon tetrachloride at 789 cm−1 overlap each other in the Raman spectrum of the mixture of the organic solutions. The Raman peak of carbon tetrachloride at 762 cm−1 can still be identified separately. The strongest Raman peak of acetone at 2924 cm−1 and the Raman peak of cyclohexane at 2923 cm−1 overlap in the Raman spectrum of the mixture. As in the analysis of the Raman peaks of the mixture of the organic powders, the amplitudes of the mixed Raman peaks in the spectrum of the mixture of inorganic solutions are higher than the amplitudes of the single Raman peaks in the spectra of the individual organic solutions. The high-intensity characteristic Raman peaks of cyclohexane and acetone are both in the detectable spectral range of sub-grating SGi3, and the high-intensity characteristic Raman peaks of carbon tetrachloride are in the detectable spectral range of sub-grating SGi1. The analysis of the mixed and individual inorganic powders and organic solutions thus shows that the designed SFMG-SHRS is capable of identifying the composition of mixtures of inorganic powders and organic solutions.
5. Summary and conclusions
In this paper, a high-resolution, broad-spectral-range SHRS that combines separate filters and multi-gratings was proposed and experimentally tested. The unique concept of the designed SFMG-SHRS is the combination of the separate filters with corresponding sub-gratings for filtering in different detectable spectral ranges in a single measurement process. In comparison with conventional SHRS, the SFMG-SHRS not only enables high-resolution and broadband measurement but also decreases the effect of fluorescence in the measurement.
In a comparison of the sulfur Raman spectra measured by the SFMG-SHRS with that measured by a commercial spectrometer under the same measurement conditions, the SFMG-SHRS showed good potential in the measurement of Raman spectra. Spectral analysis of the experimental results for sulfur, ethanol, titanium dioxide, fumaric acid, sodium sulfate, and methyl salicylate showed that the SFMG-SHRS is able to measure inorganic compounds and organic solutions for various integration times, laser power, and transparent containers. A comparative experiment was carried out for the measurement of microplastics with and without the separate filters. The complex characteristic Raman peaks of the microplastics were well detected using the separate filters. The results of the comparative experiment show that the combination of the separate filters and multi-gratings reduces the effect of fluorescence in the measurement of the microplastics, and increases the SNR and visibility of weak Raman peaks. The SFMG-SHRS also enables analysis of the components of a mixture through the measurement of the Raman spectrum of the mixed target with high spectral resolution in a broad spectral range.
The designed SFMG-SHRS thus has good potential in making high-spectral-resolution and broad-spectral-range Raman measurements including those involving weak peaks and strong background noise.
Funding
Jilin Province Research Projects in China (20220204091YY); National Natural Science Foundation of China (52227810, U2006209, 62075261, 62205333, 61975255).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request
Reference
1. A. Smekal, “Zur quantentheorie der dispersion,” Naturwissenschaften 11(43), 873–875 (1923). [CrossRef]
2. C. V. Raman and K. S. Krishnan, “A New Type of Secondary Radiation,” Nature 121(3048), 501–502 (1928). [CrossRef]
3. J. Pan, M. Lv, H. Bai, et al., “Effects of metamorphism and deformation on the coal macromolecular structure by laser Raman spectroscopy,” Energy Fuels 31(2), 1136–1146 (2017). [CrossRef]
4. -. Z. Meizhen, -. S. Weijun, and -. Z. Zhirong, “Laser Raman microcsope and its application in geology - petroleum geology & experiment,” 30(3), 302–307 (2008).
5. H. G. M. Edwards, J. Jehlička, and A. Culka, “Portable Raman Spectroscopy in Field Geology and Astrobiology Applications,” in Portable Spectroscopy and Spectrometry (2021), pp. 377–400.
6. K. Dodo, K. Fujita, and M. Sodeoka, “Raman Spectroscopy for Chemical Biology Research,” J. Am. Chem. Soc. 144(43), 19651–19667 (2022). [CrossRef]
7. G. Pezzotti, “Raman spectroscopy in cell biology and microbiology,” J. Raman Spectrosc. 52(12), 2348–2443 (2021). [CrossRef]
8. G. Pezzotti, “Raman spectroscopy of piezoelectrics,” J. Appl. Phys. 113(21), 211301 (2013). [CrossRef]
9. A. M. Zheltikov, “The Raman effect in femto- and attosecond physics,” Phys.-Usp. 54(1), 29–51 (2011). [CrossRef]
10. J. M. Harlander, F. L. Roesler, and S. Chakrabarti, “Spatial heterodyne spectroscopy: a novel interferometric technique for the FUV,” in (SPIE, 1990), 120–131.
11. J. M. Harlander, “Spatial heterodyne spectroscopy: interferometric performance at any wavelength without scanning,” (The University of Wisconsin-Madison, 1991).
12. N. R. Gomer, C. M. Gordon, P. Lucey, et al., “Raman Spectroscopy Using a Spatial Heterodyne Spectrometer: Proof of Concept,” Appl. Spectmsc. 65(8), 849–857 (2011). [CrossRef]
13. M. J. Egan, S. M. Angel, and S. K. Sharma, “Optimizing data reduction procedures in spatial heterodyne Raman spectroscopy with applications to planetary surface analogs,” Appl. Spectrosc. 72(6), 933–942 (2018). [CrossRef]
14. M. J. Egan, S. M. Angel, and S. K. Sharma, “Standoff spatial heterodyne Raman spectrometer for mineralogical analysis,” J. Raman Spectrosc. 48(11), 1613–1617 (2017). [CrossRef]
15. G. Hu, W. Xiong, H. Luo, et al., “Raman Spectroscopic Detection for Simulants of Chemical Warfare Agents Using a Spatial Heterodyne Spectrometer,” Appl. Spectrosc. 72(1), 151–158 (2018). [CrossRef]
16. K. A. Strange, K. C. Paul, and S. M. Angel, “Transmission Raman measurements using a spatial heterodyne Raman Spectrometer (SHRS),” Appl. Spectrosc. 71(2), 250–257 (2017). [CrossRef]
17. E. M. Kelly, M. J. Egan, A. Colón, et al., “Remote Raman Sensing Using a Single-Grating Monolithic Spatial Heterodyne Raman Spectrometer: A Potential Tool for Planetary Exploration,” Appl. Spectrosc. 77(5), 534–549 (2023). [CrossRef]
18. J. M. Harlander, F. L. Roesler, J. G. Cardon, et al., “Shimmer: a spatial heterodyne spectrometer for remote sensing of Earth’middle atmosphere,” Appl. Opt. 41(7), 1343–1352 (2002). [CrossRef]
19. W.-L. Zhang, Z.-Y. Liu, H. Wang, et al., “Research status of spatial Heterodyne spectroscopy–A review,” Microchem. J. 166, 106228 (2021). [CrossRef]
20. T. Nathaniel and C. Underwood, “A compact spatial heterodyne remote raman spectrometer for Mars exploration,” 62nd International Astronautical Congress 2011, IAC 2011 2, 1261–1271 (2011).
21. Q. Chu, Y. Sun, C. Sun, et al., “Broadband, high-resolution spatial heterodyne Raman spectroscopy measurement based on a multi-Littrow-angle multi-grating,” Opt. Express 31(19), 31284–31299 (2023). [CrossRef]
22. L. Wang, T. Ishiyama, and A. Morita, “Theoretical investigation of C–H vibrational spectroscopy. 1. Modeling of methyl and methylene groups of ethanol with different conformers,” J. Phys. Chem. A 121(36), 6687–6700 (2017). [CrossRef]
23. J. Qiu, X. Qi, X. Li, et al., “Broadband transmission Raman measurements using a field-widened spatial heterodyne Raman spectrometer with mosaic grating structure,” Opt. Express 26(20), 26106–26119 (2018). [CrossRef]
24. V. Saptari, Fourier Transform Spectroscopy Instrumentation Engineering, SPIE Press monograph (SPIE Optical Engineering Press, 2004), Chap 3.
25. P. D. Barnett, “The development of a miniature spatial heterodyne Raman spectrometer for applications in planetary exploration and other extreme environments,” (University of South Carolina, 2016), Chap 2, Chap 3.
26. J. M. Harlander and F. L. Roesler, “Spatial heterodyne spectroscopy: a novel interferometric technique for ground-based and space astronomy,” in (SPIE, 1990), 622–633.
27. C. R. Englert, J. M. Harlander, J. G. Cardon, et al., “Correction of phase distortion in spatial heterodyne spectroscopy,” Appl. Opt. 43(36), 6680–6687 (2004). [CrossRef]
28. Y. Song, G. Yongying, X. Wei, et al., “Baseline correction of spatial heterodyne spectrometer using wavelet transform,” Infrared & Laser Eng. 45(11), 1117009 (2016). [CrossRef]
29. J. D. Ingle Jr. and S. R. Crouch, Spectrochemical analysis (Prentice Hall College Book Division, United States, 1988), Chap 3.
30. J. Lin and Y.-Q. Li, “Ultralow frequency Stokes and anti-Stokes Raman spectroscopy of single living cells and microparticles using a hot rubidium vapor filter,” Opt. Lett. 39(1), 108–110 (2014). [CrossRef]
31. M. Zhang, G. Li, S. Wang, et al., “The influence of different integration time on stoichiometric analysis in near infrared grating spectrometers,” Infrared Phys. Technol. 86, 130–134 (2017). [CrossRef]
32. J. Qiu, X. Qi, X. Li, et al., “Raman measurements using a field-widened spatial heterodyne Raman spectrometer,” J. Raman Spectrosc. 50(10), 1602–1613 (2019). [CrossRef]
33. S. Burikov, T. Dolenko, S. Patsaeva, et al., “Raman and IR spectroscopy research on hydrogen bonding in water–ethanol systems,” Mol. Phys. 108(18), 2427–2436 (2010). [CrossRef]
34. T. Ohsaka, F. Izumi, and Y. Fujiki, “Raman spectrum of anatase, TiO2,” J. Raman Spectrosc. 7(6), 321–324 (1978). [CrossRef]
35. M. C. Bayer, C. Jessen, and A. J. Kornath, “Preparation and Characterization of Protonated Fumaric Acid,” Z. Anorg. Allg. Chem. 646(7), 333–339 (2020). [CrossRef]
36. Y. Du, H. X. Fang, Q. Zhang, et al., “Spectroscopic investigation on cocrystal formation between adenine and fumaric acid based on infrared and Raman techniques,” Spectrochim. Acta, Part A 153, 580–585 (2016). [CrossRef]
37. G. Zerbi and P. J. Hendra, “Laser-excited Raman spectra of polymers: Hexagonal and orthorhombic polyoxymethylene,” J. Mol. Spectrosc. 27(1-4), 17–26 (1968). [CrossRef]
38. H. Tadokoro, M. Kobayashi, Y. Kawaguchi, et al., “Normal Vibrations of the Polymer Molecules of Helical Configuration. III. Polyoxymethylene and Polyoxymethylene-d2,” J. Chem. Phys. 38(3), 703–721 (1963). [CrossRef]
39. K. C. Khulbe, C. Y. Feng, J. M. A. Tan, et al., eds. (Springer US, Boston, MA, 2001), pp. 231–303.
40. J. J. Lindberg, G. Lundstrom, S. L. Korppi-Tommola, et al., “Dichroism and depolarization of vibration spectra of polyphenylene oxide,” Acta Chem. Scand. A 32(4), 1 (1978).
41. C. Zimmerer, I. Matulaitiene, G. Niaura, et al., “Nondestructive characterization of the polycarbonate - octadecylamine interface by surface enhanced Raman spectroscopy,” Polym. Test. 73, 152–158 (2019). [CrossRef]
42. B. H. Stuart, “Temperature studies of polycarbonate using Fourier transform Raman spectroscopy,” Polym. Bull. 36(3), 341–346 (1996). [CrossRef]
43. D. Bertoldo Menezes, A. Reyer, and M. Musso, “Investigation of the Brill transition in nylon 6,6 by Raman, THz-Raman, and two-dimensional correlation spectroscopy,” Spectrochim. Acta, Part A 190, 433–441 (2018). [CrossRef]
44. A. Marcellan, P. Colomban, and A. Bunsell, “(Nano)structure, skin/core and tension behaviour of polyamide fibres,” J. Raman Spectrosc. 35(4), 308–315 (2004). [CrossRef]
45. X. J. Gu, “Raman spectroscopy and the effects of ultraviolet irradiation on polyimide film,” Appl. Phys. Lett. 62(13), 1568–1570 (1993). [CrossRef]
46. K. Ben Mabrouk, T. H. Kauffmann, H. Aroui, et al., “Raman study of cation effect on sulfate vibration modes in solid state and in aqueous solutions,” J. Raman Spectrosc. 44(11), 1603–1608 (2013). [CrossRef]
47. H. T. Varghese, C. Yohannan Panicker, D. Philip, et al., “IR, Raman and SERS studies of methyl salicylate,” Spectrochim. Acta, Part A 66(4-5), 959–963 (2007). [CrossRef]
48. C. K. Asawa and M. K. Barnoski, “Raman Spectra and Mode Frequency Shifts of Ferroelectric Sodium Nitrite at 77 and 294°K,” Phys. Rev. B 2(1), 205–213 (1970). [CrossRef]
49. H. N. Rutt and J. H. Nicola, “Raman spectra of carbonates of calcite structure,” J. Phys. C: Solid State Phys. 7(24), 4522–4528 (1974). [CrossRef]
50. G. Behrens, L. T. Kuhn, R. Ubic, et al., “Raman spectra of vateritic calcium carbonate,” Spectrosc. Lett. 28(6), 983–995 (1995). [CrossRef]
51. M. J. Pelletier, “Effects of Temperature on Cyclohexane Raman Bands,” Appl. Spectrosc. 53(9), 1087–1096 (1999). [CrossRef]
52. K. B. Wiberg and A. Shrake, “A vibrational study of cyclohexane and some of its isotopic derivatives—I: Raman and infrared spectra and assignments of cyclohexaae and cyclohexane-d12,” Spectrochim. Acta. A. 27(7), 1139–1151 (1971). [CrossRef]
53. V. Calvino-Casilda, K. Stawicka, M. Trejda, et al., “Real-time Raman monitoring and control of the catalytic acetalization of glycerol with acetone over modified mesoporous cellular foams,” J. Phys. Chem. C 118(20), 10780–10791 (2014). [CrossRef]
54. D. Tuschel, “Practical group theory and Raman spectroscopy, part II: Application of polarization,” Spectroscopy 1, 1 (2014).
