Raman spectroscopic evaluation of precision of oxygen isotope ratio of carbon dioxide

Abstract

We measured the Raman spectra of CO₂ fluid at 30 MPa and room temperature (19.1–21.3°C). The spectra showed peaks attributed to the CO₂ combining various isotopes such as ¹²C¹⁶O₂, ¹³C¹⁶O₂, and ¹²C¹⁶O¹⁸O. The relative ratio of ¹²C¹⁶O₂ to ¹²C¹⁶O¹⁸O peaks represents the oxygen isotope ratio of CO₂. To evaluate the precision of the peak ratios, we measured the CO₂ Raman spectra at different exposure times. We examined the standard deviation (1 σ) of the peak intensity ratios and area ones for 20 measurements at each exposure time. The standard deviations of the intensity ratios and area ones were 2.7 and 3.1%, respectively, at a maximum exposure time of 494 s, conversion to peak intensity of ¹²C¹⁶O₂ yields about 2,000,000 counts. The uncertainties are 2.5–3 times greater than expected from the noise of a CCD camera and photon statistics. The oxygen isotope ratios (δ¹⁸O) of natural samples have a range of variation of about 16.6%. Compared to that value, the precision we obtained from this study is very small. Raman spectroscopy can be combined with microscopy to analyze areas as small as approx. 1 μm in diameter. Therefore, oxygen isotope measurement using Raman spectroscopy has potential for application to natural samples as a new method for small CO₂ fluids such as fluid inclusions.

Introduction

Fluid inclusions have important information for elucidating fluid circulation system in the Earth. Fluid density has been used to estimate the depth at which a host mineral including the fluid had existed (e.g., Roedder and Bodnar, 1980; Yamamoto et al., 2002). Combined information about the depths and isotopic compositions of fluids can inform whereabout and the origin of fluid circulating us in the Earth’s interior. The isotopic ratios of noble gases (e.g., Burnard et al., 1998), nitrogen (e.g., Yamamoto et al., 2020), and carbon (e.g., Yamamoto et al., 2001) extracted by ablating and crushing deep-derived minerals in vacuum have been used to examine characteristics of the fluids in the fluid inclusions, but they are not desirable as a means of exploring the origin of the fluids because the extracted gases might be mixtures derived from fluid inclusions of several kinds, each with a different origin. Therefore, the isotopic ratios of individual fluid inclusions must be analyzed for the support of accurate estimation of the fluid origin. SIMS (Secondary Ion Mass Spectrometry) and spectroscopy are available for in-situ analyses of the isotopic compositions of fluid inclusions even with diameters of a few micrometers. Unfortunately, SIMS analysis requires solidification of the fluid (Yurimoto et al., 2014), which might lead to rupture and isotopic zoning within the inclusions. Micro-FT-IR (Fourier Transform Infrared) Spectroscopy technique offers some potential for use in isotopic analyses of fluid components in fluid inclusions (e.g., Richter et al., 2002; Sakai et al., 2022). However, a cylindrical analytical region of absorbance analysis is not suitable for the analysis of individual fluid inclusions. Micro-Raman spectroscopy is the only technique for in-situ analysis of the isotopic composition of individual fluid inclusions without changing the fluid state.

As for the isotope analysis using micro-Raman spectroscopy, five reports of the relevant literature have described carbon isotope analysis of CO₂ (Arakawa et al., 2007; Li et al., 2018; Yokokura et al., 2020; Cui et al., 2021; Wang et al., 2022). One report describes nitrogen isotope analysis of N₂ (Yamamoto and Hagiwara, 2022). CO₂-dominant fluids are commonly observed as fluid inclusions in mantle-derived rocks (Andersen and Neumann, 2001). Therefore, the carbon isotope analysis applicable to a fluid inclusion is expected to be a useful tool for investigating the origins of fluids. However, considering the possible existence of fluids originating from a mixture of three or more endmembers, proving the origin of fluids based solely on δ¹³C is difficult. If we were able to measure not only the carbon isotope ratio of CO₂ but also the oxygen isotope ratio simultaneously, then we would be able to distinguish fluid origins more clearly (e.g., Luciani et al., 2022).

In the Raman spectrum of CO₂, a peak of ¹²C¹⁶O¹⁸O is evident in addition to those of ¹²C¹⁶O₂ and ¹³C¹⁶O₂, which are used for carbon isotope analysis (Fig. 1). The ratios of the intensity and area of the ¹²C¹⁶O¹⁸O and ¹²C¹⁶O₂ peaks would indicate the oxygen isotope ratio (δ¹⁸O). The range of variation in δ¹⁸O presented in geological samples is approximately 16.6% (Holden et al., 2018). Therefore, we investigate whether the oxygen isotope ratio of CO₂ can be determined by Raman spectrometry with precision that falls within the range of ±8.3%.

Fig. 1.

(a) Raman spectrum of CO₂ at 30 MPa with accumulation time of 494 s. (b) Enlarged spectrum of (a).

Experimental

CO₂ fluid

To detect subtle Raman peaks attributable to trace amounts of ¹⁸O in CO₂, we used CO₂ compressed to a high-pressure cell with sapphire windows (PC-400 MS; Syn Corp.) at room temperature. The windows are made of high-purity sapphire glasses. Fluorescence from the windows did not affect the data of this study. First, the cell was filled with liquid-phase artificial CO₂ (99.5% purity, Fukuoka Oxygen Co., Ltd). The pressure in the cell was then increased by reducing the volume with an apparatus using a pressurizer. The CO₂ pressure was adjusted to 30 MPa because high-pressure CO₂ is often found as fluid inclusions in mantle xenoliths, with density of approx. 0.95 g/cm³ (Andersen and Neumann, 2001), which corresponds to about 30 MPa at room temperature. The pressure had been monitored by a PG-2TH digital pressure transducer with a manometer (WGA-650A; Kyowa Electronic Instruments Co., Ltd.). The accuracy of the pressure transducer is ±0.2%. The pressure and temperature during the measurements were 30 ± 0.1 MPa and 19.1–21.3°C respectively.

Micro-Raman spectroscopy

Raman spectra were measured using diode-pumped solid-state laser (532 nm; Changchun New Industries Optoelectronics Technology Co., Ltd.) excitation, a 75 cm focal length spectrometer (Acton SP-2750; Princeton Instruments, Inc.), and CCD camera (1650 × 200 pixel, 16 μm width, iVac; Andor Technology). The laser was focused through a 10× objective with super long working distance (T Plan SLWD; Nikon Corp.; N.A. = 0.2). The laser power was 1 W at the source and 0.4 W at the surface of the sapphire window of the high-pressure cell; the pixel resolution of this Raman spectrometer when using a grating of 1800 lines/mm is 0.29 cm^–1/pixel at about 1334 cm^–1. The spectrum was measured in a single window between 1004.48 and 1483.57 cm^–1. This range covers two peaks: the ¹²C¹⁶O¹⁸O peak at about 1362 cm^–1 and the ¹²C¹⁶O₂ peak at about 1368 cm^–1. Neon atomic emission lines were used for wavenumber calibration.

The obtained Raman peaks of ¹²C¹⁶O₂ and ¹²C¹⁶O¹⁸O were fitted to a Voigt curve using Fityk software (Wojdyr, 2010).

Results and Discussion

Signal count dependence of Raman spectra

From measuring Raman spectra of CO₂ at 30 MPa at 10 different exposure times, the obtained intensity of ¹²C¹⁶O₂ was found to be approx. 600 to approx. 2,000,000 counts. Figure 1 shows the Raman spectrum of CO₂ at a maximum intensity of ¹²C¹⁶O₂ as approx. 20,000,000 counts. Two prominent peaks are visible at about 1280 cm^–1 (ν^–) and 1386 cm^–1 (ν⁺) (Fig. 1a) which are split by Fermi resonance of symmetric stretching and bending vibrations. In Fig. 1b, the small peaks at the wavenumber position lower than that of ν⁺ (1386 cm^–1) are ¹²C¹⁶O¹⁸O (1369 cm^–1) and ¹³C¹⁶O₂ (1363 cm^–1) peaks, which are also split by Fermi resonance.

The relative intensity of the peaks to the noise is crucially important for obtaining the proper shape of the peaks (Fig. 2). When the ¹²C¹⁶O₂ peak intensity is approx). 600 counts, the ¹²C¹⁶O¹⁸O peak was indistinguishable from noise (Fig. 2j). Under the analytical conditions used for this study, the noise effect, especially on the shape of the ¹²C¹⁶O¹⁸O peak, is significant when the ¹²C¹⁶O₂ peak intensity is less than 7500 counts (Fig. 2h, i). The noise effect becomes less marked in Fig. 2b when the intensity of ¹²C¹⁶O₂ peaks exceeds 20,000 counts. Higher intensity of ¹²C¹⁶O₂ peak is associated with weaker noise effects. However, the long exposure time necessary to obtain high intensities leads to another difficulty for Raman spectrum analysis. Yamamoto and Hagiwara (2022) reported that when they measured the nitrogen isotope ratio of N₂ fluid using the same instrument used for this study, the effects of possible deformation of the spectrometer caused by changes in room temperature appeared on the observation of Raman peaks at exposure times longer than a half hour. Therefore, we acquired no data beyond 2,000,000 counts, or approx. 15 min.

Fig. 2.

CO₂ Raman spectra obtained at various accumulation times. An arrow indicates the ¹²C¹⁶O¹⁸O peak (ν⁺₁₈). The range of the vertical axis is 6% of the height of the ¹²C¹⁶O¹⁸O peak.

¹²C¹⁶O¹⁸O/¹²C¹⁶O₂ peak intensity ratio and peak area ratio

Table 1 presents the mean and standard deviation of the ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂ peak intensity ratio (I₄₆/I₄₄) and its area ratio (A₄₆/A₄₄), which were measured 20 times for each intensity range of ¹²C¹⁶O₂. The relation between the ¹²C¹⁶O₂ intensity and I₄₆/I₄₄ or A₄₆/A₄₄ is shown in Fig. 3.

Table 1. Peak intensity ratios and peak area ratios of ¹²C¹⁶O¹⁸O and ¹²C¹⁶O₂ Raman peaks in nine intensity regions

Time (s)	Intensity	I46/I44	1σ	1σ (%)	A46/A46	1σ	1σ (%)
1.0	3657	0.002453	0.000760	31.0	0.002158	0.000756	35.1
2.0	7522	0.001610	0.000533	33.1	0.001816	0.000448	24.6
6.0	21651	0.002062	0.000336	16.3	0.001605	0.000267	16.6
10.0	37336	0.001790	0.000162	9.0	0.001454	0.000195	13.4
20.0	80763	0.001552	0.000131	8.4	0.001387	0.000134	9.6
52.0	202339	0.001532	0.000105	6.8	0.001352	0.000121	8.9
77.0	293612	0.001487	0.000113	7.6	0.001305	0.000095	7.3
107.1	473329	0.001552	0.000092	5.9	0.001197	0.000082	6.9
494.0	1928076	0.001547	0.000041	2.7	0.001172	0.000037	3.1

Note: Data were obtained from 20 measurements of CO₂ fluid at 30 MPa at room temperature. The intensities of the 20 data in each intensity region have variation of less than 5%.

Fig. 3.

Counts of ¹²C¹⁶O₂ Raman peaks of 30 MPa CO₂ fluid versus peak intensity ratios (I₄₆/I₄₄) (a) and peak area ratios (A₄₆/A₄₄) (b) of ¹²C¹⁶O¹⁸O and ¹²C¹⁶ O₂ Raman peaks. Data with error bars in (a) and (b) are means and standard deviations of 20 peak intensity ratios, and peak area ones for each of the nine intensity regions.

The mean values of I₄₆/I₄₄ and A₄₆/A₄₄ decrease with increasing intensity of ¹²C¹⁶O₂ (Fig. 3a). The I₄₆/I₄₄ and A₄₆/A₄₄ means with high ¹²C¹⁶O₂ intensity sometimes do not overlap with the I₄₆/I₄₄ and A₄₆/A₄₄ means with low ¹²C¹⁶O₂ intensity, even considering uncertainties. This tendency is also recognized from the relation between the nitrogen isotope ratio of N₂ (¹⁴N¹⁵N/¹⁴N₂) and ¹⁴N₂ intensity obtained by the same instrument used in this study (Yamamoto and Hagiwara, 2022). That earlier report suggests that it results from noise effects on the shapes of low-intensity peaks. The high intensity ratio and area ratio of ¹⁴N¹⁵N/¹⁴N₂ in the lower count regions would be attributable to the discrimination threshold between the tail of the smaller ¹⁴N¹⁵N peak and noise, which were intermingled with each other in the curve-fitting process. On the other hand, in the higher count regions (>100,000 counts for a main peak (¹⁴N₂) or >700 counts for a small ¹⁴N¹⁵N peak), the effect of noise on the shape of ¹⁴N¹⁵N peak becomes negligible. The same can be said for the sent study. When the intensity of ¹²C¹⁶O₂ is obtained at 2,000,000 counts, the intensity of ¹²C¹⁶O¹⁸O is much higher than 700 counts. Therefore, we were able to obtain reliable I₄₆/I₄₄ and A₄₆/A₄₄ at least in the 2,000,000 counts region. However, the A₄₆/A₄₄ at an intensity of ¹²C¹⁶O₂ of 2,000,000 counts is 0.0011720 (37), which is quite different from the I₄₆/I₄₄ of 0.001545 (42). This difference might reflect the difference in the shape of the ¹²C¹⁶O¹⁸O and ¹²C¹⁶O₂ peaks. When discussing the absolute value of ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂, it is necessary to correct the measured value using a standard CO₂ with a known ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂.

Figure 4 shows the relation between ¹²C¹⁶O₂ intensity and precision (1 σ) of I₄₆/I₄₄ or A₄₆/A₄₄. The precision improves as the ¹²C¹⁶O₂ intensity increases, and reached 3.13% and 2.68% for I₄₆/I₄₄ and A₄₆/A₄₄, respectively, at around 2,000,000 counts. These values are less than half the target precision of 8.3%. Therefore, the oxygen isotope analysis technique using a micro-Raman spectrometer would be applicable to geological samples. However, the variation range of δ¹⁸O in mantle is about 0.5% (Sharp et al., 2018). Further reduction of precision is expected to be necessary for application of the present method to mantle fluid.

Fig. 4.

Counts of ¹²C¹⁶O₂ Raman peaks versus standard deviations of 20 measurements in various intensity ranges for I₄₆/I₄₄ and A₄₆/A₄₄. The dashed line is the sum of squares of the shot noise (σ_S²) calculated for each count. The solid line is the sum of the squares of the noise (shot noise (σ_S²), dark noise (σ_D²) and readout noise (σ_R²)) of the CCD camera for each count.

Effects of room temperature change

Although the target value of uncertainty was achieved, the reduction rate of uncertainty with increasing intensity was found to be worse than expected.

The uncertainty in ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂ derived from the noise of the CCD camera (i.e., dark noise σ_D and readout noise σ_R) used for this study and that resulting from photon statistics (σ_S) can be calculated according to the calculation method presented by Yamamoto and Hagiwara (2022). The noise calculation results are presented in Table 2. Similar to the precision of I₄₆/I₄₄ and A₄₆/A₄₄ in the measurements, the calculated noise values decrease with increasing ¹²C¹⁶O₂ intensity. However, the rate of decrease in the noise uncertainty is markedly larger than that in the uncertainty of the measurements. The precision of the measurements normalized to the noise uncertainty is presented in Fig. 5. For reference, we included the normalized precisions of ²⁹N₂/²⁸N₂ obtained by the same instrument (Yamamoto and Hagiwara, 2022). The overall trend is similar; it does not differ by molecule. The findings suggest that the increase in uncertainty dissociation associated with this increase in intensity might depend on the equipment characteristics.

Table 2. Calculated uncertainties (%) of ¹²C¹⁶O¹⁸O /¹²C¹⁶O₂ caused by CCD noise and photon statistics

Time (s)	Intensity	Squared shot noise (σS2)	Squared dark noise (σD2)	Squared read out noise (σR2)	All noise
1.0	3657	23.73	0.31	52.15	57.30
2.0	7522	16.55	0.21	10.65	19.68
6.0	21651	9.75	0.13	2.20	10.00
10.0	37336	7.43	0.10	1.62	7.60
20.0	80763	5.05	0.06	1.06	5.16
52.0	202339	3.19	0.04	0.68	3.26
77.0	293612	2.65	0.03	0.57	2.71
107.1	473329	2.09	0.02	0.42	2.13
494.0	1928076	1.04	0.01	0.22	1.06

Note: All noise is the square root of the sum of the squared shot noise (σ_S²), squared dark noise (σ_D²), and squared readout noise (σ_R²) of a CCD device. σ_S represents the square root of the signal counts of a peak. σ_D denotes the dark current arising from thermal energy in the silicon structure of a CCD. σ_R is defined as the noise in the amplifier on the CCD chip, which converts the stored charge of each pixel into an analog voltage. Values of σ_D and the σ_R were obtained from blank measurements of the spectrometer at various exposure times and numbers of accumulation.

Fig. 5.

Counts of ¹²C¹⁶O₂ Raman peaks versus the standard deviations of 20 measurements in various intensity ranges for I₄₆/I₄₄ and A₄₆/A₄₄ normalized by the calculated all noises of the CCD. For reference, data are shown when measuring ¹⁴N¹⁵N/¹⁴N₂ of N₂ (Yamamoto and Hagiwara, 2022), which were obtained using the same apparatus used for the present study.

Although, Yamamoto and Hagiwara (2022) proposed the possibility that the discrepancy was attributable to room temperature fluctuations, they were unable to prove it because the room temperature had not been monitored during measurements.

We monitored both the Raman spectra of CO₂ and room temperature at 10 s intervals. Figure 6 shows the result representing the relation of room temperature fluctuation with positions, intensities, areas, intensity ratios, and area ratios of two main peaks of CO₂ at approx. 1286 cm^–1 and approx. 1388 cm^–1. ¹²C¹⁶O¹⁸O is not visible in short measurements. Therefore, approx. 1282 cm^–1 and approx. 1387 cm^–1 of ¹²C¹⁶O₂, which have stronger intensity, were used. The room temperature fluctuates with a cycle of approx. 25 min. This room temperature fluctuation seems to be linked to the positions, intensities, areas, area ratios, and intensity ratios. During measurement, the intensity ratio and the area ratio varied as much as 16%. If more than several minutes are necessary to take 20 measurements for the precision evaluation of ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂, then the influence of room temperature fluctuations is unavoidable. Obtaining 2,000,000 counts data for ¹²C¹⁶O₂ takes about 3 hours for 20 measurements. Therefore, the main reason for the much higher uncertainty than that expected from the noise calculation, especially at high intensity ranges, is probably room temperature fluctuation. Resulting wavenumber fluctuations change the relative positions of the peaks to the CCD, leading to changes in the shapes of the detected peaks, and thereby changes in both the peak intensity ratios and the area ratios (Arakawa et al., 2007; Yamamoto and Hagiwara, 2022). To reduce the uncertainty, it would be useful to minimize the range of room temperature fluctuation and to complete the measurements quickly.

Fig. 6.

Fluctuations of (a) room temperature, (b) peak position of ¹²C¹⁶O₂ (ν^–), and ¹²C¹⁶O₂ (ν⁺) (c) intensity of ¹²C¹⁶O₂ (ν^–) and ¹²C¹⁶O₂(ν⁺), (d) area of ¹²C¹⁶O₂ (ν^–) and ¹²C¹⁶O₂(ν⁺), and (e) intensity ratios and areas of ¹²C¹⁶O₂ (ν^–) to ¹²C¹⁶O₂ (ν⁺).

Implication for application of Raman mass spectrometry for natural samples

Finally, we summarize some points to apply the Raman mass spectrometry for natural CO₂ fluid.

The effects of fluid inclusion size

The focal region of the incident laser beam elongated in the direction of the incident beam. Assuming a spherical fluid inclusion with several micrometer diameter, some of the light will leave the inclusion and excite a host mineral in the direction of the incident beam, even if the beam is aimed at the center of the inclusion. On the other hand, the thickness of the fluid-filled area in the high-pressure cell used in this study is more than 1000 μm (Hagiwara et al., 2021a). Therefore, in order to obtain CO₂ Raman peaks with high intensities from natural CO₂ fluid inclusions, it requires a longer measurement time than in the case of the present study. Yamamoto and Kagi (2008) measured Raman spectra of CO₂ in fluid inclusions in mantle xenoliths and obtained ¹²C¹⁶O₂ peak at approximately 30–100 counts/sec, resulting in that approximately 10 hours are necessary to obtain intensity of 2,000,000 counts. Measuring inclusions with large size and high CO₂ densities would be effective to shorten the measurement time.

The effects of host mineral properties

A host mineral surrounding inclusions affects the Raman spectrum of the inclusions. Following conditions should be kept in mind when applying the Raman mass spectrometry to natural CO₂ inclusions.

1) Fluorescence emitted from a host mineral may mask a Raman spectrum of an inclusion. This effect may be reduced by changing the wavenumber of incident laser.

2) Raman peaks of a host mineral may overlap with those of an inclusion. This could be resolved by subtracting the Raman spectrum of the host mineral, which is obtained separately.

3) The Raman spectrum of an inclusion may depend on the crystallographic orientation of a host mineral, possibly due to the birefringence of the host mineral (Caumon et al., 2020). When measuring an inclusion within a mineral with birefringence, it is important to examine whether it also affects ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂.

4) The Raman spectrum of an inclusion is affected by the absorption of the incident laser beam by a host mineral. One of the effects is the attenuation of incident laser power. It can be reduced by measuring inclusions close to sample surface. The other is that the increase in internal pressure of an inclusion induced by laser heating may cause apparent fluctuation in ¹²C¹⁶O¹⁸O/¹²C¹⁶O₂. Pressure change in CO₂ causes fluctuation in the wavenumber of CO₂ Raman peaks. As with the effect of changes in room temperature, the wavenumber fluctuation caused by the pressure change would change the relative position of the peaks to the CCD, leading to change of the shape of detected Raman peaks. The apparent pressure dependence may be corrected by estimating the CO₂ pressure of the inclusion using Raman barometry for CO₂ fluid proposed by Hagiwara et al. (2020, 2021b).

Conclusions

We measured the Raman spectra of the CO₂ fluid at 30 MPa. The precisions of I₄₆/I₄₄ and A₄₆/A₄₄ of the ¹²C¹⁶O₂ and ¹²C¹⁶O¹⁸O peaks are, respectively, 0.00117 (37) (2.6%) and 0.001547 (42) (3.1%) in the intensity region of approx. 2,000,000 counts for ¹²C¹⁶O₂ peak. These values are much smaller than the variation of δ¹⁸O found in nature (approx. 16.6%). The uncertainties of I₄₆/I₄₄ and A₄₆/A₄₄ caused by the CCD noise and photon statistics were calculated as about 1% at 2,000,000 counts for ¹²C¹⁶O₂ peak. Therefore, the precision in the measurements has main causes other than the uncertainty derived from the photon detection device. The I₄₆/I₄₄, A₄₆/A₄₄, and wavenumber positions of the relevant peaks were apparently related to room temperature fluctuation. The fluctuation of wavenumber positions would change the relative positions of the peaks to the CCD, leading to variation in I₄₆/I₄₄ and A₄₆/A₄₄. Minimizing temperature fluctuations around a spectrometer and rapid measurements can be expected to improve the precision of I₄₆/I₄₄ and A₄₆/A₄₄.

Acknowledgments

We thank the two anonymous reviewers who reviewed the manuscript and provided various useful comments.

This study was supported by Grants-in-Aid for Scientific Research (Nos. 20H02000 and 20K20931) and for JSPS Fellows (JP19J21537 and JP22J00081) from the Japan Society for the Promotion of Science.

References

•  Andersen,  T. and  Neumann,  E.-R. (2001) Fluid inclusions in mantle xenoliths. Lithos 55, 301–320. https://doi.org/10.1016/S0024-4937(00)00049-9
•  Arakawa,  M.,  Yamamoto,  J. and  Kagi,  H. (2007) Developing micro-Raman mass spectrometry for measuring carbon isotopic composition of carbon dioxide. Appl. Spectrosc. 61, 701–705. https://doi.org/10.1366/000370207781393244
•  Burnard,  P. G.,  Farley,  K. A. and  Turner,  G. (1998) Multiple fluid pulses in a Samoan harzburgite. Chem. Geol. 147, 99–114. https://doi.org/10.1016/S0009-2541(97)00175-7
•  Caumon,  M.-C.,  Tarantola,  A. and  Wang,  W. (2020) Raman spectra of gas mixtures in fluid inclusions: Effect of quartz birefringence on composition measurement. J. Raman Spectrosc. 51, 1868–1873. https://doi.org/10.1002/jrs.5605
•  Cui,  S.,  Wang,  W.,  Cheng,  C.,  Yao,  Y.,  Qin,  C. and  Sun,  Q. (2021) Raman quantitative measurement for carbon isotopic composition of CO₂: Theory and method. Chem. Geol. 585, 120570. https://doi.org/10.1016/j.chemgeo.2021.120570
•  Hagiwara,  Y.,  Torimoto,  J. and  Yamamoto,  J. (2020) Pressure measurement and detection of small H₂O amounts in high‐pressure H₂O–CO₂ fluid up to 141 MPa using Fermi diad splits and bandwidths of CO₂. J. Raman Spectrosc. 51, 1003–1018. https://doi.org/10.1002/jrs.5865
•  Hagiwara,  Y.,  Kawano,  T.,  Takahata,  K.,  Torimoto,  J. and  Yamamoto,  J. (2021a) Temperature dependence of a Raman CO₂ densimeter from 23°C to 200°C and 7.2 to 248.7 MPa: Evaluation of density underestimation by laser heating. J. Raman Spectrosc. 52, 1744–1757. https://doi.org/10.1002/jrs.6188
•  Hagiwara,  Y.,  Yoshida,  K.,  Yoneda,  A.,  Torimoto,  J. and  Yamamoto,  J. (2021b) Experimental variable effects on laser heating of inclusions during Raman spectroscopic analysis. Chem. Geol. 559, 119928. https://doi.org/10.1016/j.chemgeo.2020.119928
•  Holden,  N. E.,  Coplen,  T. B.,  Böhlke,  J. K.,  Tarbox,  L. V.,  Benefield,  J.,  de Laeter,  J. R., Mahaffy P. G.,  O’Connor,  G.,  Roth,  E.,  Tepper,  D. H.,  Walczyk,  T.,  Wieser,  M. E. and  Yoneda,  S. (2018) IUPAC periodic table of the elements and isotopes (IPTEI) for the education community (IUPAC Technical Report). Pure Appl. Chem. 90, 1833–2092. https://doi.org/10.1515/pac-2015-0703
•  Li,  J.,  Li,  R.,  Zhao,  B.,  Guo,  H.,  Zhang,  S.,  Cheng,  J. and  Wu,  X. (2018) Quantitative measurement of carbon isotopic composition in CO₂ gas reservoir by Micro-Laser Raman spectroscopy. Spectrochim. Acta A: Mol. Biomol. Spectrosc. 195, 191–198. https://doi.org/10.1016/j.saa.2018.01.082
•  Luciani,  N.,  van der Lubbe,  J. H. L.,  Verdegaal-Warmerdam,  S. J. A.,  Postma,  O.,  Nikogosian,  I. K.,  Davies,  G. R. and  Koornneef,  J. M. (2022) Carbon and oxygen isotope analysis of CO₂ trapped in silicate minerals. Chem. Geol. 602, 120872. https://doi.org/10.1016/j.chemgeo.2022.120872
•  Richter,  D.,  Erdelyi,  M.,  Curl,  R. F.,  Tittel,  F. K.,  Oppenheimer,  C.,  Duffell,  H. J. and  Burton,  M. (2002) Field measurements of volcanic gases using tunable diode laser based mid-infrared and Fourier transform infrared spectrometers. Opt. Lasers Eng. 37, 171–186. https://doi.org/10.1016/S0143-8166(01)00094-X
•  Roedder,  E. and  Bodnar,  R. J. (1980) Geologic pressure determinations from fluid inclusion studies. Annu. Rev. Earth Planet. Sci. 8, 263–301. https://doi.org/10.1146/annurev.ea.08.050180.001403
•  Sakai,  S.,  Otsuka,  T.,  Matsuda,  S.,  Sakairi,  Y.,  Uchida,  R.,  Sugahara,  K.,  Kano,  A. and  Yang,  D. (2022) Subnanomolar sensitive stable isotopic determination in CO₂ by tunable infrared laser absorption spectroscopy. Anal. Chem. 94, 6446–6450. https://doi.org/10.1021/acs.analchem.1c05458
•  Sharp,  Z. D.,  Wostbrock,  J. A. G. and  Pack,  A. (2018) Mass-dependent triple oxygen isotope variations in terrestrial materials. Geochem. Perspect. Lett. 7, 27–31. https://doi.org/10.7185/geochemlet.1815
•  Yamamoto,  J. and  Hagiwara,  Y. (2022) Precision evaluation of nitrogen isotope ratios by Raman spectrometry. Anal. Sci. Adv. 3, 269–277. https://doi.org/10.1002/ansa.202200020
•  Yamamoto,  J. and  Kagi,  H. (2008) Application of densimetry using micro-Raman spectroscopy for CO₂ fluid inclusions: A probe for elastic strengths of mantle minerals. Euro. J. Mineral. 20, 529–535. https://doi.org/10.1127/0935-1221/2008/0020-1825
•  Yamamoto,  J.,  Watanabe,  M.,  Nozaki,  Y. and  Sano,  Y. (2001) Helium and carbon isotopes in fluorites: Implications for mantle carbon contribution in an ancient subduction zone. J. Volcanol. Geotherm. Res. 107, 19–26. https://doi.org/10.1016/S0377-0273(00)00315-2
•  Yamamoto,  J.,  Kagi,  H.,  Kaneoka,  I.,  Lai,  Y.,  Prikhod’ko,  V. S. and  Arai,  S. (2002) Fossil pressures of fluid inclusions in mantle xenoliths exhibiting rheology of mantle minerals: Implications for the geobarometry of mantle minerals using micro-Raman spectroscopy. Earth Planet. Sci. Lett. 198, 511–519. https://doi.org/10.1016/S0012-821X(02)00528-9
•  Yamamoto,  J.,  Takahata,  N.,  Sano,  Y.,  Yanagita,  M.,  Arai,  S. and  Prikhod’ko,  V. S. (2020) Nitrogen and noble gas isotopic compositions of mantle xenoliths from Far Eastern Russia: Implications for nitrogen isotopic characteristics of mantle wedge fluid. Earth Planet. Sci. Lett. 534, 116109. https://doi.org/10.1016/j.epsl.2020.116109
• Yokokura, L.  Hagiwara,  Y. and  Yamamoto,  J. (2020) Pressure dependence of micro-Raman mass spectrometry for carbon isotopic composition of carbon dioxide fluid. J. Raman Spectros. 51, 997–1002. https://doi.org/10.1002/jrs.5864
•  Yurimoto,  H.,  Itoh,  S.,  Zolensky,  M.,  Kusakabe,  M.,  Karen,  A. and  Bodnar,  R. (2014) Isotopic compositions of asteroidal liquid water trapped in fluid inclusions of chondrites. Geochem. J. 48, 549–560. https://doi.org/10.2343/geochemj.2.0335
•  Wang,  W. Q.,  Cui,  S. and  Sun,  Q. (2022) Raman spectroscopic measurement guidelines for carbon isotopic composition of CO₂. Vib. Spectrosc. 123, 103449. https://doi.org/10.1016/j.vibspec.2022.103449
•  Wojdyr,  M. (2010) Fityk: A general-purpose peak fitting program. J. Appl. Crystallogr. 43, 1126–1128. https://doi.org/10.1107/S0021889810030499