Real-time determination of volatile organic compounds (VOCs) by ion molecule reaction – mass spectrometry (IMR-MS)

Abstract Comprehensive analytical validation studies of a developed ion molecule reaction – mass spectrometer (IMR-MS) were undertaken for the real-time determination of volatile organic compounds (VOCs) in air. The instrument was developed with a focus on promoting chemical ionization (CI) in the reaction chamber by direct sample loading and enhancing maintenance efficiency and reliability of the results. Instrument stability was assessed through a system check and pre-performance check process, and consequently, the instrumental and analytical conditions including the plasma generation, pressure, temperature, and flow rate were successfully optimized. Relevant performance characteristics, such as mass resolution, mass detection range, accuracy, and precision were also investigated by VOC standards composed of benzene, toluene, perfluorotoluene, propylbenzene, and octane. To evaluate whether the performance of the technology is comparable to already accepted techniques, the quantitative results of the IMR-MS were compared with those of a commercial mass spectrometer. This evaluation was successful and suggests the applicability of the technology for spillage accidents of hazardous chemicals and identification of odor-causing substances as well as for real-time gas analysis.


Introduction
Volatile organic compounds (VOCs) mostly arise from human-made chemicals and are produced in the manufacture of pharmaceuticals, plastics, and many other industrial substances. [1] VOCs comprise hydrocarbons that are often used in everyday life including liquid fuels with low boiling points, paraffin, olefin, and aromatic compounds. [1,2] From the viewpoint of atmospheric chemistry, they generate photochemical oxidizers, such as ozone through photochemical reactions with nitrogen oxides (NOx) in the atmosphere, which causes photochemical smog [3] and secondary pollutants. [4] Most VOCs, such as benzene, toluene, ethylbenzene, and aliphatic hydrocarbons, are known for their potential toxic, mutagenic, and carcinogenic effects on the human body. [5][6][7] Therefore, the determination of atmospheric VOCs is an essential part of monitoring the critical influence of environmental pollutants on human health and the global environment. Individual countries have set exposure limits for VOCs (such as 0.5-3 ppmv for benzene, 10-50 ppmv for toluene, 5-100 ppmv for ethylbenzene, 75-300 ppmv for octane) while continuously monitoring them and introducing related regulations. [8][9][10][11] Gas chromatographymass spectrometry (GC-MS) is widely used for VOC determination as a potent tool for obtaining precise qualitative and quantitative results. [12,13] However, GC-MS needs sample preparation and requires lengthy analysis, and there are limits on obtaining information about pollutants in real time. [14,15] In order to overcome these challenges, selected ion flow tube MS (SIFT-MS) was developed based on proton transfer reaction MS (PTR-MS) technology. [16] SIFT-MS has been widely used for the monitoring of VOCs from human breath, air, and archaeological artifacts. [17][18][19] Recently, ion molecule reactionmass spectrometry (IMR-MS) has been developed and introduced for online real-time monitoring of gases. IMR-MS applies a direct sample loading technology, which promotes the chemical ionization (CI) reaction of the sample gas and selected reagent ions, such as H 3 O þ , NO þ , or O þ 2 generated in an ion source. In this technique, the sample gas is inserted at an oblique angle directly into the reaction chamber where it mixes with the carrier gas or reagent ions flowing through the reaction chamber. Through this process, the sample gas is prevented from colliding with the wall of the chamber and prevents its neutralization. [20] A detailed drawing of a direct sample loading tube is provided in Figure  S1. A mechanical system was also developed in order to improve the reliability and longevity of the instrument. The instrument is an all-in-one system that can be mounted on a moving vehicle for real-time gas analysis, whereas general mass spectrometers, such as GC-MS and liquid chromatography-MS (LC-MS), have a structure in which the main chamber and the rotary pump are separated. In order to enhance stability, the mass spectrometer was designed to minimize the transmission of vibrations from the dry vacuum pump to the upper part of the system, that is to the mass spectrometer. This structure is expected to improve the maintenance and reliability of results and to be helpful when the instrument is installed in a vehicle.
While the benefits of cost-effective, online monitoring of VOCs may be anticipated, it should be verified that the analytical performance is comparable to that of an already accepted technique. To this end, comprehensive analytical validation studies are required before broad applications can be considered, and these are the focus of this work. Herein, we report (1) the evaluation of system performance in order to increase the practical use of the developed IMR-MS for real-time analysis of samples in the atmosphere and (2) the investigation of its applicability through a comparison of its quantitative results with those of a commercially available instrument.

Instrumentation
The developed IMR-MS, ACE-1100 (Younginace, Gyeonggi, South Korea), was equipped with two-stage quadrupole mass spectrometers: first mass filter and second mass filter. The first mass filter selectively passes reagent ions such as H 3 O þ , NO þ , and O þ 2 , and the second separates the specific product ions generated through reaction with an analyte. The instrument configuration of the IMR-MS is shown in Figure 1 and the dimensions of individual regions are provided in Table S1. The analyte gas and reagent ions are injected into the reaction chamber and subsequently, collisions occur. The reagent ions ionize analyte molecules by soft CI reactions including proton transfer, electrophilic addition, and hydride abstraction of charge exchange reactions and produce product ions.
The first and the second mass filter each have an attached turbo pump. The second mass filter is divided into two chambers, one is located close to the reaction chamber and the other is near the electron multiplier (EM). The pressure of the first mass filter reaches less than 7.0 Â 10 À4 torr, while the two chambers of the second mass filter reach less than 1.0 Â 10 À4 and 5.1 Â 10 À5 torr, respectively, when the turbo pump is operating at 90,000 rpm. The reaction chamber has a relatively low degree of vacuum as the sample, reagent ion, and flow gas are introduced into the chamber. By separating the chamber and the EM, the influence of the vacuum degree is minimized so that the EM may be stably operated in a high vacuum environment. A direct-acting 2-way standard solenoid control (DAS) valve controls the optimum air/moisture ratio in which plasma is generated, while at the same time maintaining the constant pressure of the ion source.
Since each analyte has selectivity by the reagent ion, an individual standard gas was used to confirm the product ions. The marker ion of the VOC standard gas was selected among the product ions in accordance with the 1.0 > 1,000 a Current of the reagent ion was detected by nano ammeter at the 1st mass filter. b Current of the reagent ion was detected by electromagnetic detector at the 2nd mass filter. reagent ion, and the unique marker ion was used for the quantitative determination of the analyte. The optimal reagent ion for the determination of each component is highlighted in bold in Table S2. The reaction rate between each reagent ion and component was the basic consideration for determining the optimal reagent ion, but the ion selectivity in the mass spectrum and high sensitivity were also considered.
The quantification method using IMR-MS is different from using GC-MS. In GC-MS, a standard material is used to obtain a calibration curve, and the curve is applied to unknown samples for the quantification of the target analyte. In IMR-MS, in consideration of the abundance of the reagent ion and the product ion participating in the reaction and the reaction rate, an absolute quantitation method is applied, without using standard material by Equation (1): [21] A where [A] is the concentration of a target component, y is a correlation factor of the instrument, p þ is the product ion, R þ is the reagent ion, and k is the reaction rate constant.

Materials
Six individual standard gases (ethylbenzene, propylbenzene, 1,2,4-trimethylbenzene, 3-methylbenzene, octane, decane; 5 ppmv) and standard gases (benzene, toluene; 1000 ppmv) were purchased from Korea Nano Gas (Gyeonggi, South Korea). The standard mixture gas in nitrogen in 10 L aluminum cylinders was also purchased from Korea Nano Gas. The standard mixture gas was composed of 7 compounds: benzene (78 g/mol), ethylene (28 g/mol), isobutane (58 g/mol), tetrafluorobenzene (150 g/mol), toluene (92 g/mol), perfluorobenzene (180 g/mol), and perfluorotoluene (236 g/mol). The standard gases were then diluted to serial concentrations in a 15 L Silonite canister (#29-11522VG, valve guard, Silonite coated true seal valve, 30 gauge) using a 4700 precision diluter with both the canister and the diluter purchased from Entech Instruments (Simi Valley, CA). The desired concentrations of the standard gases were prepared and transferred to Tedlar bags. Quantitative results for cross-validation using a commercially available instrument were provided by the analysis support service from the Center for Research Facilities at Sunchon National University.

Verification of system
Instrument stability is essential for obtaining reliable results that may be verified by a system check and pre-performance check. In the system check process, the vacuum system, including the dry and turbo pumps, was optimized to operate in a stable condition. Next, the DAS valve and plasma state were adjusted.
After enhancing the basic conditions for the system, specific values of pressure, temperature, and voltage were optimized for different states of a closed or open valve in the first and second mass filters and the reaction chamber through a pre-performance check process. Screenshots of the validation process for the IMR-MS are shown in Figure S2. Through this program, all maintenance factors were controlled and monitored at the same time. Appropriately developed conditions resulted in abundant reagent ion generation from the ion source and sufficient detection through all pathways to the detector.
The optimized conditions of the instrument after the two check processes are shown in Table 1.
The results include the most appropriate conditions of pressure in the ion source, reaction chamber, and mass filter; the temperature of the reaction chamber; and lens voltages in the first and second mass filter. The first mass filter has 8 lenses and the second has 9 lenses, each optimized for settings from À100 to 73 V for the former and from À30 to 12 V for the latter. In addition, when optimizing a condition or confirming the state of the instrument, an inspection was sequentially performed in accordance with the flowchart shown in Figure S3, which demonstrates the system check and the pre-performance check. The results suggest that the inspection criteria were obtained in accordance with the flowchart, resulting in satisfactory optimization.
An adequate quantity of reagent ions is important for increasing the sensitivity of the analysis, and sufficient reagent ions may be generated under the appropriate plasma state, for which an appropriate pressure and voltage are required. In addition, the pressure is related to the first mass filter, DAS valve, and needle valve. The appropriate plasma state is confirmed by observing the change in plasma color in Figure 2.
The color changed from (I) ignitor on to (IV), the most appropriate plasma state. In the optimal plasma state, the generation of reagent ions was at the maximum ( Figure S4). The color in Figure 2 (IV) was obtained under the optimized conditions for pressure, DAS valve, and needle valve (Table 1).

Performance evaluation
The performance of the instrument with regard to mass resolution, mass detection range, accuracy, and precision was investigated using standard gases. The evaluation criteria and the acceptable reference values were established based upon the quality control and quality assurance (QC/QA) for mass spectrometric analysis. [22][23][24][25] For the evaluation of mass spectral resolution, the low, middle, and high mass regions were measured and determined by full-width half-maximum (FWHM) [26] from the corresponding marker ion of the analyte. The measured values for the mass resolutions of three reagent ions and VOC gases by reaction with the reagent ion H 3 O þ are presented in Table 2. The mass spectrum for the representative mass resolution measurements of low, middle, and high mass are shown in Figure 3.
The results demonstrate that the mass resolution is stably maintained from m/z 19 to 240, and satisfies the evaluation criteria (< 0.7).
The mass detection range was evaluated by confirming the "scannable" range using reagent ions and standard mixture gas. The mass spectra of ethylene (m/z 28), isobutane (m/z 57), benzene (m/z 78), toluene (m/z 92), tetrafluorobenzene (m/z 150), perfluorobenzene (m/z 180), and perfluorotoluene (m/z 236), which are selective for reagent ion O þ 2 , indicate that the product ion may be detected up to m/z 420 in accordance with the change in the RF output value as shown in Figure 4.
Accordingly, it was determined that the mass detection range for the instrument was sufficient to determine the VOCs of interest in the atmosphere.
The minimum accuracy was 72.8% obtained from 20 ppmv toluene, and the maximum accuracy was 116.0% obtained from 5 ppmv ethylbenzene. The precision varied from 1.6 to 6.6% for all analyzed samples, and the averages of the accuracy were 90.0, 82.9, 116.0, 111.8, 95.5, 91.2, 76.3, and 74.6% for benzene, toluene, ethylbenzene, propylbenzene, 1,2,4-trimethylbenzene, 3-methylpentane, octane, and decane, respectively. The intra-and inter-day precision and stability for 2 hr of operation with benzene and toluene obtained by IMR-MS are also provided in Table S3 and Figure S5, respectively. The results satisfy the evaluation criteria required for the analysis of the environmental samples: 70 À 120% for accuracy and < 7% for precision. [16,21,27,28] Continuous measurement of benzene and toluene showed that their concentrations were constant for 2 hr, which is sufficient to verify the stability of the instrument.

Comparison of benzene measurements with those by a commercially available instrument
To verify the analytical performance of the IMR-MS, accuracy, precision, and linearity were compared with those of another real-time mass spectrometer (SIFT-MS) using benzene. The accuracy and precision were evaluated by measuring benzene at each concentration in the combination with IMR-MS. The average accuracy for each component by SIFT-MS was 81.2%, and the precision, represented by the relative standard deviation (RSD %), was 2.3%, indicating the similarity between the results obtained by IMR-MS and SIFT-MS.
The linear range from 0.3 to 20 ppmv of benzene was assessed for the two instruments. The results showed similarity with the two lines having a similar slope, and the results for the y-intercept and correlation coefficients (r 2 ) were y ¼ 0.6854x þ 0.1091 and r 2 ¼ 0.9979 for IMR-MS and y ¼ 0.7025x þ 0.1931 and r 2 ¼ 0.9951 for SIFT-MS. The results demonstrate that the IMR-MS and SIFT-MS showed comparable performance in accuracy, precision, and linearity. The concentration range was established based on the lowest exposure limit value being included in the linearity Figure 5. Accuracy and precision measurements of benzene, toluene, ethylbenzene, propylbenzene, 1,2,4-trimethylbenzene, 3-methylpentane, octane, and decane at three concentration levels.
range by referring to chemical exposure standards, enforcement regulations of Air Conservation Acts, Malodor Prevention Acts, material safety data sheets, and exposure limit concentrations of numerous countries with such standards. [8][9][10][11] The determination of benzene was compared with an F-test and T-test, a statistical method for comparing and assessing the outcome from two different factors. [29] For benzene (0.6 ppmv), the F-ratio statistic value was 0.381 and the F-rejection value one-sided test value was 0.053 (Table 3a); hence, the t-test was performed under the assumption of heteroscedasticity between the two instruments. The t-test shows that because t statistic value ｜À1.471｜ is smaller than the t-rejection value two-sided test value 3.182 (Table 3b), there is no difference in the average analysis results obtained by the instruments.
In addition, a statistical comparison of the quantitative results for each concentration of benzene (0.6, 3, 10, and 20 ppmv) shows that there are also no statistically significant differences (Table 3c). Therefore, the performance of ACE À1100 has been shown to be comparable to the commercial instrument.

Conclusion
IMR-MS was developed for online real-time determination of VOCs in the air using direct sample loading and an improved mechanical system to enhance instrumental stability. To verify the performance of this approach before considering broad applications, a comprehensive validation study was performed. The instrumental stability was confirmed by a system check and pre-performance check process, and the instrumental and analytical conditions including plasma generation, pressure, temperature, and flow rate were successfully optimized. In particular, the plasma state was improved by optimizing the pressure of the first mass filter, DAS valve value, and needle valve value, and the optimal condition was observed by the color change of the plasma.
The system performance was also investigated by VOC standard gases. Favorable results for mass spectral resolution ( 0.7), mass detection from m/z 10 to 400, accuracy (70 À 120%), and precision below 7% RSD were obtained. These results demonstrate that the performance of the instrument satisfies the essential requirements of a mass spectrometer for the determination of atmospheric VOCs.
In addition, to evaluate whether the performance is comparable to that of an accepted technique, quantitative measurements by IMR-MS were compared with those by SIFT-MS. The results for accuracy, precision, and linearity of benzene were comparable between the two instruments. Statistical comparison by F-and t-tests confirmed that the results were not statistically different. This successful evaluation suggests that the developed IMR-MS may be utilized in real-time measurements for atmospheric samples, spillage accidents of hazardous chemicals, identification of odor-causing substances, and other applications.