Analysis Method of Volatile Sulfur Compounds Utilizing Separation Column and Metal Oxide Semiconductor Gas Sensor

2.1 Configuration of GC/MOS system

The GC/MOS system consisted of an activated carbon filter (AC filter), sample inlet (S. valve), column (odor cell), detector (MOS gas sensor), and vacuum pump (micropump), as reported by Kim et al. [9]. A small vacuum pump mounted at the back of the system sucked air from the atmosphere, which passed through an activated carbon filter to create odorless air and was fed at a constant rate to the carrier gas. Because the vacuum pump sucks air from the atmosphere, there is no need for a separate carrier gas supply, which has the advantage of miniaturizing the system. This system can also supply air containing oxygen as a carrier gas rather than an inert gas, which has a significant impact on the reaction stability and recovery characteristics of MOS gas sensors. This differs from the work of Oh, Staples, and Viswanathan, who utilized surface acoustic wave (SAW) sensors as detectors, which used the inert gas helium (He) as the carrier gas, and configured a different carrier gas supply system [10,11].

An internal schematic of the GC/MOS system is shown in Fig. 1. The column portion separating the VSC components was filled with celite in a Teflon tube, as reported by Hanada et al. [7,9]. The length and inner diameter of the celite-filled column were 30 cm and 3 mm, respectively. The celite-filled Teflon tube and temperature sensor were inserted into a copper tube with an inner diameter of 12 mm and of the same length as the Teflon tube to create a double tube. In addition, the Teflon tubes to be filled with celite were cleaned thoroughly, washed evenly with 10 N H₃PO₄ aqueous solution or acetone mixture (0.05 N H₃PO₄ + acetone), and dried thoroughly with flowing nitrogen gas to improve the reproducibility of the VSC analysis and reduce the tail drag of the VSC component peaks. The celite in the Teflon tube was filled using a vacuum pump, and the quartz glass fibers sealing the holes on both sides of the tube were treated with an acid wash [12].

Fig. 1.

Internal components of the developed GC/MOS system

The detector mounted at the bottom of the column utilized an MOS gas sensor, which is similar to the human olfactory mechanism and responds to a variety of odorous substances. To drive the MOS gas sensor (TGS2602, Figaro Engineering Ltd., Japan) and the voltage divider circuit, 5 V DC was applied to the sensor's electrodes and heater, respectively. The voltage applied to the heater ensured that the surface temperature of the MOS gas sensor was always maintained between 400 and 450^oC. In addition, the MOS gas sensor was integrated into the onboard PCB (Fig. 2), and only the gas contact part of the MOS gas sensor was inserted into the acetal manifold to fix it; it was mounted as close as possible to the end of the column so that the VSC part separated by the column part could react quickly with the MOS gas sensor.

Fig. 2.

(a) Onboard PCB module. (b) Integrated MOS gas sensor.

The MOS gas sensor, heater temperature, sample injection unit, and real-time measurement output were controlled by a Windows-based program. Fig. 3 shows the internal components of the GC/MOS system and PC, which were built as a single unit and made of plastic for easy transportation, as no additional accessories such as a carrier gas supply were required.

Fig. 3.

Prototype of the GC/MOS analysis system.

2.2 Experimental Methods

To perform quantitative and qualitative analyses of the individual odor components of the VSCs, such as H₂S, CH₃SH, DMS, and DMDS, a GC/MOS system with an MOS gas sensor as a detector was used, and approximately 100 ppm of each standard gas (RiGAS, Korea) in nitrogen was used. The VSC standard gas prepared in an odorless bag or a Tedlar bag was connected to the odor sample injection holder shown in Fig. 1, and approximately 4 mL of the standard gas sample was automatically injected into the column using the measurement start algorithm of the GC/MOS system. The injected standard gas sample was separated by RT, and digital signals were obtained using a chromatogram. After the standard gas sample was injected into the column, the valve at the sample inlet was automatically changed to continuously supply odorless air that passed through an activated carbon filter. The flow rate of the air-carrier gas was set to approximately 6 mL/min. In addition, the temperature of the column section formed in the space between the Teflon tube and the copper tube was set to an isothermal condition, maintained at 30^oC by running a space heater installed outside the copper tube. Since components such as H₂S and CH₃SH are not well separated under high temperature conditions, as in the study by Kim and Lee [8], the optimal temperature condition for the column section was a relatively low 30^oC isothermal condition.

Additional standard gas samples were prepared at concentrations of 0.1, 0.5, 1.0, 5.0, and 10.0 ppm by diluting 100 ppm of VSC standard gas with odorless air in a 3 L odorless bag. H₂S, CH₃SH, DMS, and DMDS standards prepared at 0.1, 0.5, 1.0, 5.0, and 10.0 ppm, respectively, were measured in five replicates, and the peak maxima of the components separated by the GC/MOS system were used as the concentration output (peak height). The time was used as the RT of each component. From the chromatogram results of the GC/MOS system, the maximum output values of the component peaks allowed quantitative analysis of the VSC components, and the column RT, the time at which the maximum output value was identified, allowed qualitative analysis. The detailed analysis conditions are listed in Table 1.

Table 1.

Analysis conditions of the GC/MOS system

3.1 Separation Characteristics of VSC Components

The chromatogram results of the GC/MOS system equipped with an MOS sensor as a detector were used to confirm that each VSC component could be separated at different RTs, and the reliability of repeated measurements was investigated by varying the concentration of the VSC standard gas. Fig. 4 shows the chromatograms obtained using H₂S, CH₃SH, DMS, and DMDS standard gases at a concentration of 1.0 ppm each. At a carrier gas flow rate of 6 mL/min and a column temperature of 30^oC, we found that the VSC components were well separated and detected. The chromatogram results shown in Fig. 4 show that H₂S, which has the lowest boiling point (bp) of -60^oC among the standard gases, separated in the fastest time and was detected by the MOS gas sensor, followed by CH₃SH (bp = 6^oC), DMS (bp = 37^oC), and DMDS (bp = 110^oC). The separation characteristics were similar to the results of Yang et al. using GC/FPD [12].

Fig. 4.

Chromatograms of individual substances in the VSCs.

Therefore, it was indirectly confirmed that the GC/MOS system using MOS gas sensors could be used for the quantitative and qualitative analyses of H₂S, CH₃SH, DMS, and DMDS, which are representative VSC components.

In addition, when the concentration of VSC standard gas was analyzed at 0.1, 0.5, 1.0, 5.0, and 10.0 ppm, H₂S was separated at approximately 1.22 min, CH₃SH at approximately 1.87 min, DMS at approximately 2.87 min, and DMDS at approximately 18.80 min, respectively. The RT results are summarized in Table 2. Based on the variation in RT with concentration and component in Table 2, the relative standard deviation (RSD) for each component is approximately 3%, which can be assessed as a relatively stable separation characteristic. In particular, the significant %RSD deviations of the components with shorter RTs appeared to be due to arithmetic limitations imposed by the relatively short RTs, which resulted in large deviations for small differences. Even when the concentration of VSC components increased, the detection times of each component, i.e., RTs of approximately 1.2 min, 1.8 min, 2.8 min, and 18.8 min, were clearly distinguishable. The separation characteristics of the GC/MOS system were found to be excellent. We have demonstrated the applicability of a new system that can distinguish gas components while utilizing a single MOS gas sensor as a detector.

Table 2.

Comparison of the separation characteristics (RT) of VSC substances

3.2 Separation Characteristics as a Function of VSC Component Concentration

Fig. 5 illustrates the separation characteristics of the GC/MOS system as a function of the concentration of the individual VSC components, and Fig. 5 (a) shows the chromatograms obtained by sequentially increasing the concentration of the H₂S standard gas to 0.1, 0.5, 1.0, 5.0, and 10.0 ppm. Similarly, Fig. 5 (b) shows the chromatogram results for the CH₃SH standard gas, and Fig. 5 (c) shows the chromatogram results for the DMS and DMDS standard gases.

Fig. 5.

Chromatogram of the GC/MOS system with varying VSC concentrations. The (a), (b) and (c) are output peak value about H2S, CH3SH, DMS and DMDS standard gases, respectively.

The x-axis of the chromatogram represents the RT of the detected VSC components traveling through the column, and the y-axis represents the output (voltage) value of the MOS gas sensor. The maximum output value of the MOS gas sensor increased linearly with the concentration of H₂S, and the maximum output value was detected at approximately 1.2 min. Moreover, as the concentration of CH₃SH increased, the maximum output value of the MOS gas sensor increased proportionally, and the RT of the maximum output value was approximately 1.8 min.

Chromatograms obtained by analyzing a mixture of DMS and DMDS at 0.1, 0.5, 1.0, 5.0, and 10.0 ppm also show that the maximum output value of the MOS gas sensor increases with the concentration of the standard gas, with the maximum output value for DMS being approximately 2.8 min and for DMDS approximately 18.8 min, depending on the concentration of the standard gas. Consequently, the RT remained constant for each component.

Table 3 summarizes the results of five repetitions of the chromatogram analysis with different concentrations of H₂S, CH₃SH, DMS, and DMDS standard gases. The maximum output of the MOS gas sensor measured five times under the same concentration conditions varies by concentration, and the %RSD deviation of the components across the five concentration classes and five repetitions is 1.05 to 2.99 for H₂S, 0.52 to 6.14 for CH₃SH, 0.28 to 1.44 for DMS, and 1.15 to 5.28 for DMDS, respectively.

Table 3.

Comparison of detection characteristics with VSC concentrations

The GC/MOS system used in this study was configured to be automatically controlled by a PC operating program such that the diluted standard gas sample bag was automatically aspirated five times when connected to the sample inlet. It is believed that 4 mL of the standard gas sample was aspirated and left connected to the sample inlet without shaking the sample bag to mix for a 30 min analysis time. The repeatability of the GC/MOS system is expected to improve if more attention is paid to sample injection. As shown in Table 3, the maximum output of the MOS gas sensor increased proportionally with the concentration of the individual VSC components.

The correlations among these parameters are shown in Fig. 6. Fig. 6 (a) shows the concentration correlation of the H₂S components separated using the GC/MOS system. The H₂S concentration correlation formula for the maximum output value of the MOS sensor was calculated as y = 0.162 e^2.23x, where x represents the maximum output value of the MOS gas sensor and y is the concentration of the VSC component, with a correlation coefficient (R²) of 0.991, showing a very high correlation. The correlation between the maximum output of the MOS gas sensor and the concentration of the CH₃SH component in Fig. 6 (b) was y = 0.114 e^2.66x, with a high correlation coefficient of 0.997. The correlation equation for DMS in Fig. 6 (c) is y = 0.091 e^1.96x with a correlation coefficient of 0.994, and that for DMDS in Fig. 6 (d) is y = 0.104 e^4.67x with a correlation coefficient of 0.998. Overall, the concentration correlation between the VSC component and the MOS sensor output of the GC/MOS system was excellent. Therefore, it is easy to calculate the maximum output value of the MOS sensor extracted from the results of the chromatograms analyzed using the GC/MOS system by converting the concentration of each VSC component.

Fig. 6.

Correlation between the VSC concentration and the output value of the GC/MOS system (the peak height (V) is the output of the load resistor in the voltage divider circuit). The (a), (b), (c) and (d) are the concentration correlation curve about H2S, CH3SH, DMS and DMDS standard gases, respectively.

The chromatogram results of the VSC components analyzed using the above GC/MOS system revealed that individual VSC components with different RTs could be easily separated, and quantitative/qualitative analysis of each component was possible. In particular, the results are expected to be extended to the new concept of an electronic nose system that combines the advantages of MOS gas sensors that respond to different gas components using GC separation technology.

3.3 Separation Characteristics of VSCs for Mixed Components

The separation properties of the individual VSC components were analyzed in a mixture of components to investigate whether the GC/MOS system could effectively separate the VSC components, and the concentration of each component was quantified in a standard gas containing a mixture of four VSC components. The mixed standard gases were prepared by diluting the H₂S, CH₃SH, DMS, and DMDS components to a concentration of 1.0 ppm each in 3 L odorless bags in the same way as the diluted individual VSC components. The source components for all VSC standard gases were supplied by the manufacturer at approximately 100 ppm and diluted 100-fold based on the source component concentration to reach an arithmetic theoretical concentration of approximately 1.0 ppm.

Fig. 7 shows the chromatogram of a standard gas containing a mixture of H₂S, CH₃SH, DMS, and DMDS analyzed using the GC/MOS system, and the separation characteristics of the VSC components were compared by displaying the results in Fig. 4, which analyzed the individual components as a background (colored dotted lines). The results shown in the background are also from the same 1.0 ppm diluted standard gas; only the difference in maximum output for the H₂S component can be seen, while the RTs of the other components are almost identical.

Fig. 7.

Chromatogram of the GC/MOS system for the VSC gas mixture.

We also compared the arithmetic theoretical concentrations with the concentrations converted from the maximum output values for each VSC component of the chromatogram shown in Fig. 7. Table 4 summarizes the results of the chromatograms measured in triplicate with the mixed standard gas. The concentrations of each component were calculated by applying the maximum output value of each component to the concentration correlation equation in Fig. 6. The theoretical concentration of all VSC components is 1.0 ppm, and the concentrations of VSC components analyzed by the GC/MOS system were calculated in the range of 0.84 to 1.06 ppm. For H₂S, the calculated concentrations were approximately 0.84 ppm, CH₃SH was approximately 1.06 ppm, DMS was approximately 0.99 ppm, and DMDS was approximately 0.91 ppm. However, the deviation between the theoretical and actual analyzed concentrations of the H₂S component was relatively large among the mixed standard gases. The maximum output value of the MOS gas sensor corresponding to 1.0 ppm in the H₂S concentration correlation equation shown in Fig. 6 (a) is also far from the calculated value. Therefore, a concentration correlation equation that can reduce the error generated when the concentration correlation equation of the MOS gas sensor is applied to 1.0 ppm of H₂S must be derived.

Table 4.

Concentration results for the VSC gas mixture analyzed using the GC/MOS system

In addition, by comparing the separation characteristics of the gas components of the GC/MOS system as a detector and the GC/SAW system proposed by Yoo et al., it can be concluded that the GC/MOS system separates the peaks of the chromatogram more clearly. Although it is difficult to directly compare the separation characteristics of the results for acetone, ethanol, and methanol in Yoo et al.'s study, the selectivity of individual components separated by the column is determined to be low, and the GC/SAW system has a clear limitation as it cannot distinguish between ethanol and methanol when mixed [13].

Therefore, it was confirmed that the GC/MOS system was sufficiently capable of quantitatively/qualitatively analyzing VSC components, and it is expected that the separation characteristics of various odorous components other than VSC components can be studied to provide basic data for the application of a new electronic nose system.