Portable Mass Spectrometry for In-Field Real/Time Water Pollution Monitoring: Validation and Pilot Study in Danube-Tisa-Danube Irrigation System

1. Introduction

The contamination of surface and groundwater resources by organic micropollutants - primarily volatile organic compounds such as aromatic hydrocarbons (BTEX) and chlorinated solvents-alongside agrochemicals and various other pollutants, represents a critical global environmental concern. These hazardous substances enter aquatic systems, including irrigation canals, drainage streams, and groundwater matrices, through industrial activities and accidental spillages [1,2]. Moreover, catastrophic natural events, such as major floods, can further exacerbate the situation by flushing industrial sites and permanently depositing pollutants into river overbank sediments and floodplains [3]. Within surface water bodies utilized for agricultural irrigation, such as the complex Hydro-system Danube-Tisa-Danube (DTD), the introduction of these toxic compounds poses a dual threat. On one hand, they can enter the agri-food chain through soil-to-crop transfer, potentially endangering food safety from "farm-to-fork". On the other hand, chronic exposure to VOCs through contaminated agricultural and environmental matrices is associated with adverse effects on the human immune and nervous systems, as well as vital internal organs. Due to the high volatility, toxicity, and ecological risks associated with these substances, developing highly sensitive laboratory analytical methods, real-time field monitoring networks, and advanced computational modeling tools is of paramount importance for comprehensive environmental risk assessments [2,4]. The environmental significance of VOC monitoring is further reinforced by increasingly stringent international regulatory frameworks governing the quality of surface waters. Within the European Union, the Water Framework Directive (2000/60/EC) [5] and the Environmental Quality Standards Directive (2008/105/EC, amended by 2013/39/EU) [6] establish Environmental Quality Standards (EQS), defining both annual average (AA-EQS) and maximum allowable concentrations (MAC-EQS) for hazardous organic contaminants. For example, the annual average environmental quality standard for benzene in surface waters is set at 10 μg L⁻¹, while chlorinated solvents such as 1,2-dichloroethane and chloroform are subject to similarly strict concentration limits. Comparable regulatory requirements are enforced by the United States Environmental Protection Agency (US EPA) through the National Recommended Water Quality Criteria under the Clean Water Act [7], where VOCs, including benzene, trichloroethylene (TCE), and tetrachloroethylene (PCE), are regulated at low μg L⁻¹ levels because of their ecological and human health risks. In Serbia, the Regulation on Limit Values of Pollutants in Surface and Ground Waters and Sediments [8], harmonized with the European legislative framework, prescribes equivalent threshold values for priority pollutants in surface waters intended for agricultural and environmental use. Compliance with these regulations requires reliable, rapid, and cost-effective analytical methods capable of supporting continuous environmental surveillance and early detection of contamination events.

Gas Chromatography-Mass Spectrometry (GC-MS) is the primary analytical technique for identifying and quantifying pollutants (e.g., organic pollutants, including volatile chlorinated hydrocarbons, BTEX, persistent organic pollutants – POPs, pesticide residues, etc.) in water networks [1,2,4,9]. Utilizing full mass spectra scans and contrasting data with extensive mass spectral libraries enables the precise, unambiguous identification of individual target compounds as opposed to relying on non-specific group parameters [2,4].

Given their high volatility and occurrence at low concentrations, the primary analytical challenge in VOCs detection lies in sample preparation. Water samples generally require extraction techniques such as Solid Phase Extraction (SPE) or liquid-liquid extraction prior to analysis [9]. However, for volatile chlorinated hydrocarbons and BTEX in aqueous matrices, the purge-and-trap (P&T) technique has proven superior to classical extraction methods by minimizing analyte loss due to volatilization during sample handling [1,4]. As a highly efficient and environmentally friendly alternative that eliminates the use of hazardous organic solvents, headspace solid-phase microextraction (HS-SPME) has also been widely adopted to isolate BTEX and petroleum residues from complex solid and liquid matrices [10]. Furthermore, the reliability and precision of SPME analysis for contaminated groundwater have been significantly enhanced by incorporating deuterated internal standards into the GC-MS protocol, which successfully compensates for complex matrix effects and variations in extraction efficiency [11].

Named advances in sampling and sample preparation during conventional laboratory workflow are welcomed and they surely improve sensitivity and accuracy of the analytical process. However, transport and storage still result in substantial volatile analyte losses. Additionally, conventional GC-MS analysis provides insight in specific time point, due to due to discrete sampling. Thus, continuous monitoring of the VOC concentration changes in emergent situations (e.g., oil spillage) is impossible.

Currently, to achieve rapid detection and continuous observation of water pollution, environmental monitoring frameworks are increasingly shifting toward automated sensor networks and field-deployable systems. This approach requires analytical instruments that can support rapid, sensitivity, and real-time quantification of VOCs directly in the field [12]. Although conventional laboratory-based techniques offer superior accuracy and sensitivity, they cannot provide fast, real-time, on-site analysis. For immediate field screening of volatilized compounds, portable multi-gas portable sensors are routinely deployed; however, these field instruments can be highly susceptible to cross-sensitivities, necessitating fixed laboratory confirmation to rule out false positives [11]. On the other hand, mass spectrometry-based techniques are well known to be very sensitive, selective, accurate, and precise. Among these, Membrane Inlet Mass Spectrometry (MIMS) is an old and simple technique which has gained renewed interest as a powerful technique for direct, real-time, and on-site environmental analysis [13,14]]. The fundamental mechanism of MIMS relies on pervaporation through a selective membrane (usually silicone-based), which allows target VOCs to pass directly into the mass spectrometer's ion source, entirely bypassing sample preparation [15,16]. Additionally, it provides both discrete analysis at specific time points, and continuous monitoring, with low power consumption. While this system has already been validated in the laboratory for BTX analysis [17], this study expands the target list to a wider array of volatile water pollutants, validates the system for the mixture of the selected compounds in the laboratory and in field – at 36 locations across the Danube-Tisa-Danube (DTD) irrigation canal in Serbia. In parallel, the similar MIMS system developed by the same research group has been successfully deployed for the on-site detection of volatile pollutants in the open sea [18]. Obtained results from both studies demonstrate MIMS instrument potential as a very promising screening instrument for environmental monitoring.

2. Results

2.1. Laboratory Validation of Portable Mass Spectrometer

The MIMS system capability testing included examination of several parameters that are usually examined during method validation process: selectivity, sensitivity (limits of detection and quantification), linearity, precision and accuracy. To validate the performance of the analytical method for target volatile organic compounds, specific criteria must be met across five key parameters. First, selective permeation requires the presence of characteristic mass fragments in scanned mass spectra for target VOC standards, as well as unit resolution for selected m/z fragments used in monitoring and quantification. Sufficient sensitivity is established by achieving limits of detection relevant to legislative standards in the low parts-per-billion (ppb) range, maintaining a signal-to-noise ratio (S/N) greater than 3. Additionally, a linear calibration range must yield a coefficient of determination (R²) greater than 0.95 across the optimal concentration range. Finally, method validation involves determining both the precision and accuracy of the newly developed analytical approach.

Listed parameters were assessed according to the following plan:

• Selectivity examination included scanning the gas phase of pure chemicals for all target VOCs to obtain their mass fragmentation patterns using MIMS system with an electron impact (EI) ion source and compared to the NIST database. Additionally, the mixture of all selected VOCs in the water matrix was scanned in order to confirm unit resolution.
• Sensitivity was examined by consecutive analyses of 12 water samples spiked at low ppb concentration levels to determine limits of detection (LOD) and quantification (LOQ) for each VOC.
• Linearity of the method was assessed by analyzing water samples spiked at 6 concentration levels and determining the coefficient of determination (R2).
• Precision was examined by analyzing 12 water samples spiked at concentration level 50 ppb (µg/L) – 6 series with 2 samples per each. From the results obtained several parameters were calculated:
• Sw (%) – intra-serial standard deviation – expressing the repeatability of the method,
• RSD (%) – relative standard deviation – expressing the intra-laboratory repeatability of the method.
• Accuracy was examined by calculating the recovery (%) values for the 12 water samples spiked at concentration level 50 ppb (µg/L).

In addition to the evaluation of the listed validation parameters using the newly developed MIMS sensor and method, a parallel analysis using conventional GC-MS method was performed for benchmarking purposes. Six water samples were spiked at a concentration level of 50 ppb for each target VOC and were analyzed to compare precision and accuracy between two techniques.

2.1.1. Selectivity Examination

Ethylbenzene and xylenes share the characteristic fragments m/z 91 and 106, and literature indicates that xylenes are typically more abundant in water matrices, ethylbenzene was excluded from the target analyte list. As a result, only xylenes will be monitored in further experiments, with the caveat that the reported xylene signals may include a minor contribution from ethylbenzene.

The first step in selectivity confirmation was the examination of characteristic mass fragments in the scanned spectra using the MIMS sensor. Figure 1 shows scanned spectra for benzene, toluene, xylenes, chlorobenzene, 1,2-dichlorethane, trichloroethylene and tetrachloroethylene. Acquired mass spectra were compared to the NIST database, and all evaluated compounds produced the expected mass fragments and corresponding fragment ratios.

Considering the obtained mass spectra, a single m/z fragment was selected for the following quantification for each of the target compounds. For this purpose, the most abundant fragment that was unique and did not overlap with fragments from any other VOC (Table 1) on the list was chosen.

The second step in selectivity evaluation was examination of the resolution for each of the selected mass fragments. Figure 2 shows scanned spiked water sample, with selected peaks marked. The unit resolution was confirmed for all examined target VOCs, except for chloroform. Namely, during the experiments, it was noticed that the ion current signal for chloroform was very unstable and drifted. It is possible that its high volatility causes it to evaporate very fast and the homogeneity in the water cannot be achieved. This kind of behavior was not noticed for any other examined VOCs. Thus, it was decided to exclude the chloroform from the target list. Obtained results indicate that selective analysis for target compounds can be achieved using new MIMS sensor.

2.1.2. Sensitivity Examination

Sensitivity of the new MIMS sensor and the corresponding analytical method were described via limits of detection (LOD) and limits of quantification (LOQ) for each VOC. For LOD and LOQ determination, 12 consecutive measurements of water samples spiked at low concentrations for each VOC were conducted. The level was defined after preliminary scans of the tested concentration levels for each VOCs and comparing the ion current intensities for the analytes and for the instruments baseline noise. The request was that for LOD signal to noise is ≥ 3. Following these preliminary experiments, the lowest calibration level for each VOC were used: 5ppb for benzene, toluene, chlorobenzene, 1,2-dichloroethane, trichloroethylene and tetrachloroethylene while 15ppb is for xylenes.

Table 2. Signal intensities of 12 consecutive measurements for LOD and LOQ determination.

Signal intensities (A)

VOC

1

2

3

4

5

6

7

8

9

10

11

12

Benzene

5.52E-11

4.74E-11

2.28E-11

2.27E-11

2.28E-11

2.18E-11

2.36E-11

2.31E-11

2.65E-11

2.68E-11

2.36E-11

2.29E-11

Toluene

1.51E-11

1.25E-11

4.80E-12

5.65E-12

5.84E-12

5.48E-12

6.00E-12

5.99E-12

6.94E-12

7.11E-12

6.24E-12

5.98E-12

Xylenes

3.21E-12

1.41E-12

1.41E-12

1.58E-12

1.48E-12

1.64E-12

1.70E-12

2.01E-12

2.03E-12

1.83E-12

1.75E-12

1.87E-12

CB

8.22E-12

6.77E-12

3.03E-12

3.00E-12

3.38E-12

3.08E-12

3.64E-12

3.58E-12

4.21E-12

4.32E-12

3.90E-12

3.63E-12

1,2-DCE

3.61E-11

1.74E-11

1.67E-11

1.63E-11

1.58E-11

1.62E-11

1.59E-11

1.71E-11

1.73E-11

1.56E-11

1.54E-11

1.59E-11

TCE

1.30E-11

1.17E-11

5.97E-12

5.76E-12

5.85E-12

5.41E-12

5.94E-12

5.81E-12

6.54E-12

6.62E-12

5.90E-12

5.81E-12

PCE

1.98E-12

1.61E-12

6.54E-13

5.94E-13

7.30E-13

5.20E-13

7.25E-13

7.21E-13

8.99E-13

9.34E-13

7.51E-13

7.36E-13

Obtained signals were used for LOD and LOQ calculations. The following formulas were used. For LOD we used formula LOD=3×stddev ÷ slope, for LOQ we used formula LOQ=10×stddev ÷ slope.

Table 2. Signal intensities of 12 consecutive measurements for LOD and LOQ determination.

Signal intensities (A)

VOC

1

2

3

4

5

6

7

8

9

10

11

12

Benzene

5.52E-11

4.74E-11

2.28E-11

2.27E-11

2.28E-11

2.18E-11

2.36E-11

2.31E-11

2.65E-11

2.68E-11

2.36E-11

2.29E-11

Toluene

1.51E-11

1.25E-11

4.80E-12

5.65E-12

5.84E-12

5.48E-12

6.00E-12

5.99E-12

6.94E-12

7.11E-12

6.24E-12

5.98E-12

Xylenes

3.21E-12

1.41E-12

1.41E-12

1.58E-12

1.48E-12

1.64E-12

1.70E-12

2.01E-12

2.03E-12

1.83E-12

1.75E-12

1.87E-12

CB

8.22E-12

6.77E-12

3.03E-12

3.00E-12

3.38E-12

3.08E-12

3.64E-12

3.58E-12

4.21E-12

4.32E-12

3.90E-12

3.63E-12

1,2-DCE

3.61E-11

1.74E-11

1.67E-11

1.63E-11

1.58E-11

1.62E-11

1.59E-11

1.71E-11

1.73E-11

1.56E-11

1.54E-11

1.59E-11

TCE

1.30E-11

1.17E-11

5.97E-12

5.76E-12

5.85E-12

5.41E-12

5.94E-12

5.81E-12

6.54E-12

6.62E-12

5.90E-12

5.81E-12

PCE

1.98E-12

1.61E-12

6.54E-13

5.94E-13

7.30E-13

5.20E-13

7.25E-13

7.21E-13

8.99E-13

9.34E-13

7.51E-13

7.36E-13

Obtained signals were used for LOD and LOQ calculations. The following formulas were used. For LOD we used formula LOD=3×stddev ÷ slope, for LOQ we used formula LOQ=10×stddev ÷ slope.

Table 3. Results obtained for LOQ and LOD using MIMS sensor.

VOC	SD (%)	Slope	LOD (µg/L)	LOQ (µg/L)
Benzene	1.099E-11	7.09E-12	5	16
Toluene	3.135E-12	2.16E-12	5	15
Xylenes (-o, -m, -p)	4.822E-13	1.91E-13	8	25
Chlorobenzene	1.614E-12	1.49E-12	4	11
1,2-Dichloroethane	5.741E-12	3.542E-12	5	16
Trichloroethylene	2.519E-12	1.27E-12	7	20
Tetrachloroethylene	4.396E-13	3.104E-13	5	14

*SD – standard deviation of 12 measurements.

Table 3. Results obtained for LOQ and LOD using MIMS sensor.

VOC	SD (%)	Slope	LOD (µg/L)	LOQ (µg/L)
Benzene	1.099E-11	7.09E-12	5	16
Toluene	3.135E-12	2.16E-12	5	15
Xylenes (-o, -m, -p)	4.822E-13	1.91E-13	8	25
Chlorobenzene	1.614E-12	1.49E-12	4	11
1,2-Dichloroethane	5.741E-12	3.542E-12	5	16
Trichloroethylene	2.519E-12	1.27E-12	7	20
Tetrachloroethylene	4.396E-13	3.104E-13	5	14

*SD – standard deviation of 12 measurements.

2.1.3. Linearity Examination

The linearity of the new analytical method using MIMS sensor was determined using in-house developed application for data processing, after construction of the calibration curves for each target VOC (Figure 3), in concentration ranges (Table 4).

2.1.4. Precision and Accuracy Examination

For the precision and accuracy of the new MIMS sensor and the method determination 12 water samples were spiked at concentration level of 50 µg/L. The samples were prepared in 6 series of two samples. In the same way, separate set of samples was prepared and analyzed by conventional HS-GC-MS technique, which is “gold standard” in VOCs analysis and serves for the benchmarking of the new technique. The analytes were quantified using corresponding calibration curves constructed separately on MIMS and GC-MS instruments as can be seen in Table 5.

The precision of the MIMS and HS-GC-MS methods was defined through repeatability and reproducibility of the method. The intra-serial standard deviation (Sw, %) was expressed as measure of repeatability, while relative standard deviation (RSD, %) within all analyzed samples was used for intra-laboratory reproducibility determination. Table 6 shows obtained results for precision determination for MIMS and HS-GC-MS techniques separately.

The accuracy of the MIMS and HS-GC-MS methods was expressed as recovery (%), which was calculated as:

$$\mathrm{R}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{y}\left(\mathrm{\%}\right)={\displaystyle \frac{\mathrm{q}\mathrm{u}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{i}\mathrm{f}\mathrm{i}\mathrm{e}\mathrm{d} \mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e} \mathrm{i}\mathrm{n} \mathrm{u}\mathrm{g}/\mathrm{L}}{\mathrm{s}\mathrm{p}\mathrm{i}\mathrm{k}\mathrm{e}\mathrm{d} \mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e} \mathrm{i}\mathrm{n} \mathrm{u}\mathrm{g}/\mathrm{L}}}\times 100.$$

In accordance with the set criteria according to AOAC [19], both MIMS and HS-GC-MS methods provided acceptable precision and accuracy.

Additionally, 6 water samples were spiked at the concentration level 50 µg/L and analyzed in parallel by MIMS and HS-GC-MS, and the results are presented in Table 7.

Parallel analysis of the spiked water samples at concentration level 50 µg/L (ppb) resulted in standard deviation between the two applied techniques for each sample lower than the sum of RSD % values determined experimentally for MIMS and HS-GC-MS, and within the required criteria set by AOAC for the relevant concentration level.

2.2. In-Field Tests

Validated MIMS system and corresponding analytical method were employed for direct and real-time analysis of target VOCs (Table 8) in real environment on the one part of Danube-Tisa-Danube irrigation system.

In total, 40 locations (Figure 4) were selected for this in-field validation of MIMS system. However, 4 locations were not accessible. Therefore, the total number of locations examined in this study is 36 analyzed by two analytical techniques – newly developed portable MIMS system and GC-MS, in laboratory conditions. MIMS system was used in real time on-site analysis, while GC-MS samples were collected in triplicate and analyzed on the same day in the laboratory. Fresh calibration standards were prepared prior to every analysis for each technique separately.

2.2.1. MIMS Results

In-house application was created for VOCs quantification. It enables processing of the scan and SIM (selected ion monitoring) raw files generated by MIMS. First step in current research data processing includes creation of calibration curves based on MIMS generated SIM (selected ion monitoring) files for each analyte, at different concentration levels. Once calibration curves are set, sample files are plotted against the calibration curves and quantified ppb values are obtained. VOCs concentrations are provided in .csv format.

During in-field validation of portable VOC sensor sampling was conducted at 36 locations along the DTD canal, and none of the examined VOCs were present in concentration above the LODs obtained during method validation. The results are presented in Table 9 with sample IDs for 36 locations only.

Even though there are no quantified values for examined VOCs throughout the irrigation system, at one location increased signals for benzene and toluene were detected by MIMS. This location is 200 meters from the gasoline station (Figure 5). Thus, obtained results are logical, as ambient concentrations of benzene and toluene primarily originate from anthropogenic sources, including motor vehicle exhaust, fossil fuel combustion (coal and oil), and fugitive evaporative emissions from gasoline service stations [21,22].

Figure 6 illustrates the active in situ deployment of the MIMS sensor. External battery for powering the MIMS can be seen as well as the battery for powering the pump for sample pumping.

2.2.2. GC-MS Results

Conventional GC-MS technique confirmed results obtained by MIMS sensor. No examined VOCs were above limits of quantification or limits of detection for MIMS sensor. However, some of VOCs were detected by GC-MS in some samples. Qualitative presence was confirmed by NIST database incorporated in GC-MS results processing software (Agilent® MassHunter Quantitative Analysis) and comparing mass spectra of VOCs detected in samples with mass spectra of calibration standards for corresponding VOC. Results are presented as detected (marked as “o” in Table 10) and non-detected (marked as “x”) and sample location numbered L7, L20, L23 and L37 were inaccessible at the moment of sampling. VOCs levels are not quantified, as detected values were out of the performed calibration range. These results serve as an indication of presence of examined VOCs.

As can be seen, toluene and tetrachloroethylene were detected at all examined locations. These are followed by chlorobenzene, which was detected at 26 locations, xylenes which were present at 9 locations, trichloroethylene which was present at 6 locations and benzene which was present at 2 locations. 1,2 – dichloroethane was not detected at any location.

The ubiquity of toluene and tetrachloroethylene (PCE) across all monitored sites is a direct consequence of their widespread anthropogenic application and environmental persistence. Toluene typically enters the canal network via urban stormwater runoff and industrial effluents, whereas PCE – primarily utilized as a solvent in dry cleaning and metal degreasing – exhibits high resistance to biodegradation in surface aquatic environments. The high detection frequency of chlorobenzene (at 26 locations) aligns with the regional profile of the Danube-Tisza-Danube canal basin, as this compound serves as a critical intermediate in pesticide and dye synthesis, directly reflecting the intensive agro-industrial activities characterizing the Vojvodina region. Conversely, xylenes (9 locations) and trichloroethylene (TCE) (6 locations) were detected less frequently. Although xylenes share similar pyrogenic and industrial sources with toluene, they undergo substantially faster photolytic and microbial degradation in open lotic systems, while TCE occurrences are typically confined to localized, point-source discharges from metal processing facilities. The sparse detection of benzene at only two locations is justified by its physicochemical properties. Despite its prevalence in petroleum derivatives, its exceptionally high vapor pressure drives rapid volatilization from surface waters into the atmosphere, compounded by strict regulatory restrictions implemented due to its high toxicity1. Finally, the complete absence of 1,2-dichloroethane is expected, as its historical use as an anti-knock additive in leaded gasoline has been entirely phased out, and its contemporary application in PVC synthesis is confined to strictly closed-loop industrial processes, preventing its adventitious release into open hydro systems.

2.2.3. MIMS vs HS-GC-MS Results

Given that we detected only benzene and toluene at one location using MIMS sensor, we can make comparisons of the results only for these results among the two techniques. At this specific location, quantified values obtained by MIMS were 2.87 ppb for benzene and 3.04 ppb for toluene, respectively. For the same location, quantitative values by GC-MS were 0.21 ppb for benzene and 0.55 ppb for toluene. It must be noted that these values are just indicative, as were out of the calibration range. It can be observed MIMS results were significantly greater than results obtained by delayed analysis by GC-MS. As we stated – this was the location near the gasoline station, and giving that these VOCs evaporate very fast it is logical that on-site technique could detect the greater amount, comparing to the delayed analysis by GC-MS. Even though it is only one location, this is very promising outcome and proof that on-site monitoring solutions are highly needed. On the other hand, at other locations, when examined pollutants were present, but not in sufficient amounts (below MIMS sensor limits of detection), these were still detectable by the conventional technique. Therefore, these results are the real confirmation that both approaches are needed for holistic environmental monitoring.

3. Materials and Methods

3.1. Chemicals and Reagents

All the VOCs used in this study were acquired from Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany). Membranes used in the membrane inlet were acquired from Technical Products Inc. of Georgia (Buford, GA, USA). To generate calibration curves, aqueous standards were freshly prepared at the deployment site by spiking water with serially diluted VOC stock solutions in HPLC-grade ethanol. All preparations were handled in sealed glass vials to prevent volatilization losses prior to immediate, on-site mass spectrometer calibration.

3.2. Instrumentation – Membrane Inlet Mass Spectrometer

Mass spectra (m/z 1–300, SIM mode) were acquired on a PrismaPro QMG 250 M3 quadrupole mass spectrometer (Pfeiffer Vacuum GmbH, Asslar, Germany) using 70 eV electron ionization via an yttria-coated tungsten filament (2000 µA current) and a combined Faraday/electron multiplier detector. The unit was housed in a custom vacuum chamber (UltraHighVacuum, UK) maintained at 4 × 10⁻⁶ Torr using a Pfeiffer HiPace 80 turbomolecular pump, an MVP 030-3DC backing pump, and an MPT 200 cathode pressure gauge (all supplied by Pfeiffer Vacuum GmbH, Asslar, Germany). Sample delivery during field analysis was facilitated by a 12 V DC submersible bilge pump. The system was controlled using PV MassSpec software (V.23.06)

The membrane inlet, sampling interface, and internal component layout of the portable mass spectrometer were described in our previous study [18]. MIMS system was packed in the protective case provided by Pelican (Torrance, California, USA). Dimensions of the case are 60 × 50 × 23 cm (length × width × height) with a mass of entire assembly being 29 kg.

3.3. Instrumentation – GC-MS

Headspace sampling was executed using an automated PAL3 autosampler coupled to an Agilent 8890 Gas Chromatograph (GC) with Agilent 5977B Mass Selective Detector (MSD) (Agilent Technologies, Santa Clara, CA, USA). Samples were equilibrated at 60 °C for 10 min under intermittent agitation (250 rpm; 10 s on, 20 s off). Headspace vapor (0.2 mL) was injected via a heated syringe (60 °C) into the GC inlet operated in split mode at 250 °C. Chromatographic separation was performed on an Agilent HP-5MS capillary column (30 m × 0.25 mm i.d., 0.25 µm film thickness) using helium as the carrier gas at a constant flow rate of 1.0 mL/min. The oven temperature was initially held at 31 °C for 2.5 min, ramped at 7 °C/min to 150 °C, and maintained for 1.0 min, resulting in a total chromatographic run time of 20.5 min. Data acquisition and spectral processing were performed using Agilent MassHunter Workstation software.

4. Conclusions

This study successfully validated and deployed a portable Membrane Inlet Mass Spectrometry (MIMS) system for direct, real-time on-site monitoring of target volatile organic compounds (VOCs) - including benzene, toluene, xylenes, chlorobenzene, 1,2-dichloroethane, trichloroethylene, and tetrachloroethylene - across 36 locations in the Danube-Tisa-Danube (DTD) irrigation canal network. Laboratory validation confirmed that the MIMS method fulfills AOAC criteria for selectivity, sensitivity (limits of detection between 4–8 µg/L), linearity (R² > 0.98), precision, and accuracy, showing strong analytical agreement with conventional headspace GC-MS. While general baseline canal concentrations remained below MIMS detection limits across the hydro-system, the sensor proved its field effectiveness near an active gasoline station, where on-site MIMS captured markedly higher concentrations of volatile benzene (2.87 µg/L) and toluene (3.04 µg/L) than delayed laboratory GC-MS analysis. This highlights key capacity of the MIMS to prevent volatile analyte losses caused by sample collection, handling, and storage. Ultimately, portable MIMS represents a robust, rapid, and solvent-free screening technology for continuous aquatic surveillance, demonstrating real-time field measurements. Together with fixed laboratory GC-MS, they effectively complement each other to achieve holistic environmental risk assessment and early pollution detection.