Quantitative monitoring of dissolved gases in a ﬂ ooded borehole: calibration of the analytical tools

. Gas monitoring is a prerequisite to understanding the exchange, diffusion, and migration processes of natural gases within underground environments, which are involved in several applications such as geological sequestration of CO 2 . In this study, three different techniques (micro-GC, infrared, and Raman spectroscopies) were deployed on an experimental ﬂ ooded borehole for monitoring purposes after CO 2 injection. The aim was to develop a real-time chemical monitoring device to follow dissolved gas concentrations by measurements in water inside the borehole but also at the surface through a gas collection system in equilibrium with the bore-hole water. However, all three techniques must be calibrated to provide the most accurate quantitative data. For this, a ﬁ rst step of calibration in the laboratory was carried out. A new calibrations were required to determine partial pressure and/or concentrations of gases in water or in the gas collection system. For gas phase analysis, micro-GC, FTIR spectroscopy, and Raman spectroscopy were compared. New calibration of the micro-GC was done for CO 2 , CH 4 , and N 2 with uncertainty from ±100 ppm to 1.5 mol% depending on the bulk concentration and the type of gas. The FTIR and Raman spectrometers were previously calibrated for CO 2 , and CO 2 , N 2 , O 2 , CH 4 , and H 2 O, respectively with an accuracy of 1 – 6% depending on concentration scale, gas and spectrometer. Dissolved CO 2 in water was measured using a Raman spectrometer equipped with an immersion probe. The uncertainty on the predicted dissolved CO 2 concentration and partial pressure was ±0.003 mol (cid:1) kg (cid:3) 1 and ±0.05 bar, respectively.


Introduction
Since the last decades, research projects related to the monitoring of the underground gas content within a borehole are still a topical subject that is directly involved in several environmental preservation applications. For instance, in the face of global warming and climate change, numerous Carbon Capture, Utilization, and Storage projects (CCUS) are performed around the world [1,2] to reduce the anthropogenic greenhouse gas emission into the atmosphere such as the Century Plant CCUS project in the USA, the Sleipner CCS project in Norway [3], the Shenhua CCUS project in China [4], and the Lacq-Rousse CCS project in France [5,6], etc. Within such application, the monitoring strategy of the soil gas content variation must be deployed before CO 2 injection (to establish the baseline concentration variation and to define the alert limitation) and during and after CO 2 injection (to detect as soon as possible any CO 2 leakage and environmental impacts). The geological sequestration of nuclear waste also received huge attention from the scientific community with many studies in which the confining properties of the host rock of the geological storage site were investigated to assess its long-term safety [7][8][9][10]. For that, the evolution of the concentration of dissolved gases (CO 2 , CO, CH 4 , C 2 H 6 , C 3 H 8 , N 2 , H 2 , O 2 , H 2 S, etc.) in the pore water is the key information to understanding the fluid-rock interaction mechanisms and the transport properties of solutes within the host rock [8,10].
Intense efforts have been made on the metrological development including the optimization of the experimental protocol and the data processing or the enhancement of the accuracy of the measurement [8,[11][12][13][14]. Since the chemical composition of the (dissolved) gases within the underground medium is not fully known prior, the combination of different techniques is envisaged in order to detect and quantify in-situ and continuously as much gas as possible. Nowadays, the Fourier Transform InfraRed (FTIR) and the Raman vibrational spectroscopies are widely used for continuous measurement of natural gas analyses [14][15][16][17][18]. Quantitative measurement using vibrational spectroscopy is not a straightforward analysis but requires first precise calibration of the signal of each gas with respect to the properties of interest (e.g., molar proportion, partial pressure). Since the signal from vibrational spectroscopy is particularly sensitive to different instrumental parameters (excitation wavelength, response function of the device) as well as the operational environmental conditions (pressure, temperature), the calibration is therefore device-specific and should be regularly re-calibrated to ensure the highest accuracy of measurement. It is to note that FTIR spectroscopy presents some intrinsic limitations. For instance, the spectra of the combination band of CO 2 (centered at 3610 cm À1 ) are strongly affected by the H 2 O vapor contribution at some concentration ranges and required specific additional processing [16]. Also, the homonuclear diatomic gases (such as N 2 , O 2 ) cannot be detected by FTIR spectroscopy whereas they are an important indicator of the biological activity of the soil (e.g., soil respiration processes) and the oxygenation conditions [17,[19][20][21][22]. On the other hand, the Raman technique can detect the most common natural gases (including CO 2 , CH 4 as well as H 2 , O 2 , and N 2 . . .) but with a lower sensitivity compared to that of FTIR spectroscopy. Combining these two techniques can thus bring interesting and complementary information to each other. Besides, the micro gas chromatograph (micro-GC), which is a molecular separative technique, is a powerful technique with very high sensitivity and reproducibility for natural gas analyses. Using at least two modules with two different columns and carrier gas, all the pre-cited gases can be analyzed.
The in-situ and continuous measurement of the dissolved gas concentration within an aquifer or the aqueous phase of the borehole is still challenging due to the technical limitation of the instruments (e.g., low sensitivity of the Raman spectroscopy, the lack of accurate calibration data). Assuming that the partial pressures of the gases dissolved within the liquid and the gaseous phases above the liquid are in equilibrium [23], the dissolved gas concentration with an aquifer or a flooded borehole can be thus indirectly measured by analyzing its gaseous form. For this purpose, the specific polymeric membrane-based completion SysMoG designed by Solexperts was developed which permitted the collection of gaseous species in equilibrium with dissolved species but excluded water [5]. The collected dissolved gases can thereby be transported from the aquifer or the flooded depth borehole via a gas circulation system to different analytical techniques to be analyzed. Besides, the dissolved concentration or the partial pressure of CO 2 can also be derived from the pore water ex-situ analysis, e.g., calculated from the measured alkalinity or the total inorganic carbon content using PHREEQC software [8,24]. However, the pore water extraction requires a careful sampling procedure to avoid any sample chemical alteration (due to the contact with the atmosphere for example) which in turn may result in an error of the measured dissolved gas concentration. In a previous study [8], there was a noticeable discrepancy between the partial pressure of dissolved CO 2 measured in the gaseous phase and the one calculated from water composition in sampled water which could not be fully interpreted. Thus, more experiments are still needed to investigate the gas membrane transfers to better interpret the evolution of the dissolved gas concentration. For this purpose, an experimental borehole of 51-meter depth drilled by Solexperts is devoted to developing and testing the experimental protocol and new analytical instruments before deploying them in the field. The borehole is equipped with a specific membrane-based completion Sys-MoG and a closed gas circulation system which allows continuous gas analysis by different techniques. Several CO 2 injections into water within the borehole were performed to simulate CO 2 leakage. The micro-GC, Raman, and FTIR spectroscopies are connected all together to the gas circulation system to do the in-situ analysis.
The purpose of this study is to calibrate and compare the performance of the instruments in quantifying dissolved gases before using them online with the instrumented borehole. Here the calibration procedure and the new calibration data for the quantitative analysis of CO 2 , CH 4 , and N 2 concentrations are detailed. The application of these calibrations to the measurement in the borehole during CO 2 injection will be published in a dedicated paper [25].

Description of the experimental borehole
The experimental site consists of a borehole of 51-meter depth and 10 cm in diameter, drilled by Solexperts (Vandoeuvre-lès-Nancy, France). The borehole is entirely isolated from its surrounding medium, i.e., there is no contact between the borehole and the adjacent soil or the groundwater. The piezometer level within the borehole is freely adjusted by adding or removing water. The hydrostatic pressure at the bottom of the borehole is~3-5 bars and the temperature is~10°C. For the in-situ continuous measurement of the gas concentration, a SysMoG completion is installed into the borehole and connected to the analytical instruments placed at the ground via a gas circulation system.
The gases dissolved in the aqueous phase equilibrate across the hydrophobic membrane into a Gas Collection Chamber (GCC) (Fig. 1). A gas circulation line connects the GCC of the completion to the analytical module placed at the ground in a closed system. The analytical instruments are set up at the ground including an FTIR spectrometer, a micro gas chromatograph (micro-GC), and two Raman spectrometers for continuous monitoring of the concentration of the different gases (e.g., CO 2 , CH 4 , N 2 , O 2 , and H 2 O). One Raman spectrometer is equipped with an immersion probe to directly measure gas concentration dissolved in water inside the borehole. The other instruments, i.e. FTIR, micro-GC, and the second Raman spectrometer, are used to analyze the gases from the gas collection chamber.

Laboratory experimental setup to determine dissolved CO 2 concentration by Raman spectroscopy
The signal of the Raman RNX2 spectrometer (Kaiser Optical Systems) coupled with an immersion probe (Wet-Head™) is calibrated to determine the concentration of CO 2 dissolved in water. It is done by acquiring the Raman spectra of a water solution equilibrated with a known pressure of CO 2 at a known temperature. Indeed, according to Henry's law [23] the concentration of a gas dissolved in a solution at equilibrium is proportional to its partial pressure within the vapor phase, which is above the liquid phase and can be calculated by equation (1): where C CO 2 is the concentration of the dissolved CO 2 (molÁkg À1 ), p is the partial pressure of CO 2 in a gaseous phase, K H is Henry's law constant for the solubility of CO 2 at 298.15 K, (i.e., K H = 0.034 molÁkg À1 Ábar À1 ) [26]. The Henry's law constant at a given temperature T (K T H ) can be calculated by K H using equation (2): The Raman spectra of CO 2 dissolved in water at different equilibrium states are recorded by using the experimental setup shown in Figure 2. The calibration apparatus consists of a 1000-mL airtight stainless-steel cylinder (Swagelok, Fig. 2.1) in which ultrapure Milli-Q water is equilibrated with CO 2 at controlled pressures. One end of the cylinder is connected by a union cross (Swagelok) and three valves to a pump ( Fig. 2.2), a pressure transducer (Fig. 2.3), and a bottle of CO 2 (Fig. 2.4). The three valves are firstly closed and the cylinder is then inversed and fill with about 500 ml of water into the cylinder via its bottom end ( Fig. 2.5). The immersion probe is then inserted into the cylinder in a way to ensure that the head of the immersion probe is still under the water level when the cylinder is closed and inversed again (Fig. 2.6). The cylinder is closed by a specific for ensuring there is no gas or water leakage and then vertically fixed with a stand. The Raman probe is connected to a Raman RXN2 spectrometer (Fig. 2.7) by a 100-m long optical fiber. Before introducing CO 2 into the cylinder, the entire system is evacuated under a vacuum to remove the air inside the cylinder. Once the purge is done, CO 2 is introduced into the cylinder at the desired pressure via the gas inlet ( Fig. 2.4) by using a pressure regulator connected to the compressed CO 2 bottle (99.99% purity, Air Liquid). The internal pressure is monitored by the pressure transducer ( Fig. 2.3). Room temperature is also monitored. Thus CO 2 gradually dissolves into water at controlled conditions (P and T). CO 2 and H 2 O Raman spectra are simultaneously recorded over the spectral range 1350-1850 cm À1 with the Raman RXN2 spectrometer. The wavelength and power of the excitation radiation are 532 nm and~100 mW, respectively. The acquisition time is 30 s per accumulation and 3 accumulations for each measurement. An example of spectra of CO 2 dissolved in water is shown in Figure 3. The obtained spectra were then processed by LabSpec6 software (Horiba) to determine the peak area of CO 2 and H 2 O after baseline subtraction. As the area of the lower band of CO 2 at~1285 cm À1 is too small, only the area of the upper band of CO 2 at~1381 cm À1 was considered. Besides, the Raman H-O-H bending mode of water is observed over a spectral range from 1450 to 1800 cm À1 which superimposes the signal of the sapphire windows of the immersion probe at 1522 cm À1 (Fig. 3). Thus, the total peak area of the H 2 O band was calculated after subtracting the signal of sapphire.
The internal pressure of the tanker was firstly set at 2 bars. The time needed to reach the equilibrium is more than 3 days (Fig. 5). When pressure increases from 2 bars (at equilibrium) to 3 bars, the time needed to reach the new equilibrium state at 3 bars is less than one day. Afterward, the pressure is successively decreased to~2.5, 1.5, 1, 0.7, 0.5, 0.3, and 0.1 bar by adjusting the pressure regulator of the gas bottle and/or opening the valve (Fig. 2.4). The equilibrium state at these lower pressures is reached after few hours. Shaking of the entire tanker can help to speed up the equilibrium process.

Micro-GC calibration
The micro-GC-490 (Agilent) was calibrated for the quantitative measurement of CO 2 , CH 4 and N 2 . CH 4 and N 2 were quantified using a module with a Molsieve 5Å column (10 m) with argon as carrier gas and CO 2 using the module with a PoraplotU column (10 m) with helium as carrier gas. For the two modules, the injection time is 50 ms, the injector temperature is 90°C, the carrier gas pressure is 30 psi (~2 bar) and the detectors are micro catarometer (lTCD). The column temperature is 145°C for the Molsieve module and 80°C for the PoraplotU module. Peak integration is done with Soprane Software (SRA). Calibrations were established using reference gases of known concentration (commercial standard gases with certified concentration, the uncertainty is less than 1%) or gas mixtures prepared in the laboratory from pure gases by using a gas mixer  (3), and a bottle of pure compressed CO 2 (4). The Raman probe (6) is introduced via the other end of the cylinder which is then closed by a special tor to ensure there is no water leakage. The Raman probe is connected to a Raman RXN2 spectrometer (7) by a 100-m optical fiber. (GasMix™-AlyTech). The uncertainty on the concentration of the prepared gas mixtures is less than 3%. To ensure the highest accuracy of the calibration data at a low concentration range, the concentrations of CO 2 and CH 4 of some prepared gas mixtures, when below 4000 ppm, are re-checked by the G2201-i analyzer (Picarro, Inc.). Upon the calibration procedure, the entrance of the micro-GC is directly connected to the bottle of standard gas or the outlet of the AlyTech gas mixer (Fig. 4). To introduce the gases into the micro-GC, an external pump is used and turned on for about 20 s to remove any atmospheric air in the system. Once this is done, the reference gases are introduced at a pressure of less than 1.5 bar. Thereby, the whole volume of the external pump and connecting tubes is filled only by the reference gas. The reference gas is then injected into the analysis modules to be analyzed thanks to the internal pump (Fig. 4).

FTIR spectrometer calibration
The IR measurements are performed using a portable FTIR Bruker Alpha spectrometer (Bruker Optics GmbH, Germany) equipped with a 5 cm path-length stainless-steel gas cell. The latter is equipped with two ZnSe (Zinc Selenide) windows and is directly connected to the gas circuit for continuous measurement of gas concentration. The IR spectra are scanned 16 times over the spectral range 600-4000 cm À1 .
The calibration data of the FTIR apparatus used herein (including the spectrometer and the gas cell) for the quantification of CO 2 concentration over a range from 100 to 60,000 ppm was established in previous work (see Fig. 4 in [22]). The calibration curve links the CO 2 concentration (ppm) to the variation of the peak area of the fundamental band m 3 of CO 2 at~2350 cm À1 . According to the authors, the accuracy of the CO 2 concentration measurement is less than 6% at <2000 ppm, less than 4% up to 5000 ppm, less than 2% up to 30,000 ppm, and less than 1% up to 100,000 ppm.

Raman spectrometer calibration
The molar fraction (X i , mol%) of gaseous species within a mixture can be determined from the Raman scattering cross-section (r i ) of the vibrational mode of gases and the corresponding peak area (A i ) by using equation (3) [27]: where f i is the instrumental efficiency which depends on the wavelength (peak position). As the spectrometer intensity response was calibrated using a reference emission lamp (HCA, Kaiser Optical System), all f i was equal for all gases. Therefore, the r i values from the literature of N 2 , CO 2 , CH 4 , O 2 , and H 2 O for the excitation wavelength of 532 nm can be used [28][29][30][31] (Tab. 1)  The Raman measurements of the gases flowing within the gas circulation lines are performed using a Raman RXN1 spectrometer equipped with a 532 nm Nd-YAG excitation laser and coupled with a gas probe (AirHead ™ , Kaiser Optical Systems, Inc.) via a 5-meter-long optical fiber. The air probe is embedded in a dedicatedly designed stainless steel gas cell equipped with a sapphire window and an array of mirrors to amplify the Raman scattering intensity that is connected to the gas circulation systems. The Raman analyses can be thereby continuously performed without changing the internal pressure and the composition of the collected gases. The peak area of the studied gases is measured by integrating their Raman spectrum over a predefined spectral interval, i.e., between 1380 cm À1 and 1397 cm À1 for the upper band of CO 2 (at 1388 cm À1 ), between 2319 cm À1 and 2341 cm À1 for the N 2 band, between 1545 cm À1 and 1566 cm À1 for O 2 , and from 3637 cm À1 to 3665 cm À1 for H 2 O.

Calibration data for quantitative measurements 3.1 Quantifying dissolved CO 2 concentration by Raman spectroscopy
The evolution of the CO 2 -to-H 2 O Peak Area Ratio (PAR) as a function of time is presented in Figure 5. In general, the PAR increases gradually with time following the increase of the amount of dissolved CO 2 , then becomes unchanged once the equilibrium state is reached. The average value of PAR at each equilibrium state and its standard deviation are reported in Table 2 with the corresponding concentration of CO 2 . The calibration curves are fitted from all experimental data points, linking the variation of the PAR to the concentration of the dissolved CO 2 (Fig. 6a) and the partial pressure of the CO 2 (Fig. 6b). The uncertainty (1r) of the  The average peak area ratio is calculated from all experimental data points at equilibrium (Fig. 5). The calibration curve is fitted from all experimental data points, linking the variation of the PAR to the concentration of the dissolved CO 2 (Fig. 6a) or the partial pressure of the CO 2 (Fig. 6b). The uncertainty (1r) of the predicted CO 2 dissolved concentration and the partial pressure are ±0.003 molÁkg À1 and ±0.05 bar, respectively. predicted dissolved CO 2 concentration and the partial pressure is about ±0.003 molÁkg À1 and ±0.05 bar, respectively.

Gas concentration determined by Micro-GC
In general, the calibration curves of the micro-GC-490 are a linear correlation between the peak area and the concentration (ppm). The signal saturates when the CO 2 concentration is above 60 mol%. Consequently, in the present study, only two calibration curves of CO 2 are provided for low (from 500 to 9000 ppm) and intermediate concentration ranges (up to~60%) (Fig. 7). The uncertainty of the predicted CO 2 concentration (at 1r, estimated from the prediction intervals of the regression equation) is less than ±100 ppm for the concentration range under 9000 ppm, <±1000 ppm for the concentration range under 100,000 ppm, and <±0.5% for the concentration range under 60 mol%.
In the same way, calibration data was established over a concentration range from 0 to 5000 ppm for CH 4 , and from 1 to 100% for N 2 (Fig. 8). The uncertainty of the predicted concentration of CH 4 and N 2 (1r, estimated from the prediction intervals) is less than about ±100 ppm and 1.5 mol%, respectively.

Gaseous CO 2 concentration determined by FTIR
In the present study, the CO 2 concentration within the gaseous phase of the borehole is expected to range from about 500 ppm (near the atmospheric concentration) to 60 mol%. Since the intensity of the IR fundamental band of CO 2 at 2350 cm À1 is quickly saturated above 60,000 ppm, the peak area of the combination bands of CO 2 at 3610 cm À1 (m 1 + m 3 and 2m 2 + m 3 ) is therefore used. It is noted that there is an interference between the IR spectra of the combination bands of CO 2 and that of H 2 O vapor around 3610 cm À1 . The subtraction of the signal of H 2 O vapor from that of CO 2 is thus required to ensure the highest accuracy of the concentration measurement. More details about FTIR data processing can be found in Taquet et al. [14]. Moreover, it is also to note that the existing calibration data of the FTIR Alpha spectrometer was  The Author(s): Science and Technology for Energy Transition 78, 21 (2023) usually established at atmospheric pressure, whereas the intensity and the area of the IR spectra vary as a function of pressure. In this study, the total pressure within the gas circulation system is however expected to vary from the atmospheric pressure to~up to 2-3 bars due to a large amount of CO 2 dissolved in the aqueous phase, then penetrating across the membrane into the GCC. The calibration data for such a wide pressure and concentration range is not available yet. In the borehole, the FTIR spectra are simultaneously recorded with Raman and micro-GC whose intensity does not depend on the total pressure. Therefore, new calibration curves linking the IR peak area of the combination band of CO 2 to the concentration measured by other techniques (Raman and micro-GC) are thereby established for the pressure and concentration range encountered in the borehole by comparison with Raman spectroscopy and GC [25].

Gas concentration determined by Raman spectroscopy
Partial pressure and mol% in the gas mixture are determined by Raman spectroscopy using literature data (Tab. 1) and pressure measurement. Previous works using the same Raman spectrometer and the same probe reported uncertainties of about 0.5-1 mol% at atmospheric pressure [14,18]. The detection limit depends on the cross-section of each gas: the best detection limit is for CH 4 and the lowest one for N 2 (Tab. 1) at about 0.002 mbar and 0.01 mbar, respectively.

Conclusion
In preparation for on-site measurement of gas concentration in a flooded borehole, laboratory calibrations are carried out for micro-GC, FTIR spectroscopy, and Raman spectroscopy. Raman spectroscopy is the only of the three techniques able to quantitatively measure the concentration of CO 2 dissolved in water, with an accuracy of 0.003 molÁkg À1 or 0.05 bar for the corresponding partial pressure in equilibrium with the solution. For the gas phase, Raman spectroscopy can detect all the studied gases (CO 2 , CH 4 , N 2 , O 2 , H 2 , H 2 O) with an accuracy better than 1 mol % and a detection limit lower than 0.01 mbar. The Micro-GC is calibrated to quantify CO 2 , CH 4 , and N 2 with a quite much lower detection limit (<100 ppm) and higher accuracy (100 ppm -5% depending on the concentration range) (i.e. 0.2 mbar -0.1 bar at 2 bar of total pressure). Finally, FTIR can quantify CO 2 concentration in the gas phase over the full concentration range (500 ppm -100%) with accuracy depending on the concentration range (1-6%, i.e., 0.02-0.12 bar at 2 bar of total pressure). At concentrations higher than 60 mol% and pressure higher than atmospheric pressure, the infrared spectrometer is not yet calibrated. The next acquisition in the flooded borehole will lead to a new calibration in this high-level range by comparison with micro-GC and Raman spectroscopy data [25]. The performance and applicability of each technique and the SysMoG polymeric membrane responsivity and resistance over a wide concentration-pressure range will be evaluated and compared during the measurement run on the instrumented borehole.