English

• Detection of hazardous volatile organic compounds (VOCs) in environmental waters is crucial for assessing environmental safety and human health risks [1,2]. However, the concentrations of most VOCs in water are extremely low, which poses considerable challenges for their determination, especially in remote locations lacking sample preparation conditions and analytical instruments [3]. Water samples are usually collected in the field and transported to central laboratories for analysis, a process fraught with potential issues. Firstly, the volatility of VOCs can lead to loss of analytes during storage and transportation, thereby compromising the accuracy of the analysis. Secondly, sensitive VOCs analysis requires the collection of large-volume water samples and their transportation to a laboratory, a process that is both labor-intensive and costly.

  To address these challenges, low-cost on-site sampling and cryopreservation offer a viable solution. On-site sampling simplifies the sample storage and transportation by separating the analytes from the samples and extracting analytes into small volumes of solid or liquid phases [4–8]. This is especially true for the solid extraction phase as it is solvent-free and allows the analytes to be preserved in a dry form, which greatly facilitates sample packaging, storage and transport [9]. Additionally, some analytes are more stable in the solid extraction phase than in water samples [10]. Various techniques have been employed for on-site sampling of water VOCs, including solid phase extraction, stir bar sorptive extraction, solid-phase microextraction (SPME), needle-based microextraction techniques, and vacuum-assisted techniques [4,5,11–17]. Among them, SPME is the most popular technique due to its easy operation and portability for on-site use. However, this technique has limitations such as fragility, low sample capacity and high cost of commercial SPME. Furthermore, the commonly used direct immersion extraction method is significantly affected by high matrix interference. For VOCs, using headspace mode can minimize co-extractives since only volatiles are sampled. Purge and trap (P&T) is a sorbent-based headspace extraction method that enriches volatiles on an inert support by continuously stripping analytes with an inert gas stream through the matrix [18]. The P&T technology is a dynamic headspace sampling method that utilizes a larger volume of extraction phase, resulting in higher extraction efficiency and low matrix interference. This technology has the potential to be used for developing a portable sampling device, while the sorbent tubes used in the technology are suitable for dry preservation [19,20]. However, few studies have reported on miniaturized P&T instruments for field sampling of VOCs in water. Moreover, for highly volatile compounds, ambient temperature is not sufficient for their long-term preservation, and a cooled environment is typically necessary [20,21]. Refrigerators are often used to preserve extracted phases, but this can raise costs and require additional components. For on-site sample preparation, to enhance portability and reduce costs, it is desirable to integrate the cooling system with the extraction system to create a compact and unified operating device.

  In this work, a high-throughput miniaturized purge-and-trap (µP&T) device integrating semiconductor refrigeration storage was successfully developed for on-site extraction of VOCs from water samples and their preservation under low-temperature and dry conditions. Sorbent tubes were used for extraction and preservation of VOCs, facilitating their take-back or delivery to a central laboratory where the analytes trapped in the sorbent tubes could be thermally desorbed and accurately and sensitively analyzed using gas chromatography (GC). Cyclic volatile methylsiloxanes (cVMSs) are a class of VOCs characterized by high production volumes. Due to their persistent presence and long residence times, cVMSs have emerged as contaminants of concern in aquatic systems. In this study, four specific cVMSs—hexamethylcyclotrisiloxane (D3), octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane (D5), and dodecamethylcyclohexasiloxane (D6)—were selected as model VOCs to evaluate the performance of the proposed device (Table S1 in Supporting information).

  The µP&T device consists of two integrated modules: A main module and an accessory module (Fig. 1a). The two modules are assembled using two PLA shells (Figs. S1 and S2 in Supporting information) designed in SolidWorks and printed by a 3D printer and integrated into an aluminum alloy suitcase (315 mm × 208 mm × 260 mm). The main module mainly consists of a µP&T unit and a semiconductor refrigeration unit. The µP&T unit is composed of a miniature air pump (3 W), an air particulate filter, a gas purifier filled with 10 g of styrene divinylbenzene copolymer and activated carbon, an air resistor, 50 mL gas wash purge vessels, quartz sorbent tubes (110 mm length × 2 mm i.d. × 3 mm o.d.) and gas-water separators. Nine 50-mL gas wash purge vessels are positioned in a 9-position sample tray, capable of pretreating nine water samples simultaneously (Fig. 1c). The semiconductor refrigeration unit consists of a semiconductor chilling plate (TEC1-12706a, 21.7 W, 40 mm × 40 mm), an aluminum groove, an aluminum fin block, an aluminum bracket and two micro fans (Fig. 1b). The aluminum groove attached to the cooling side of the semiconductor chilling plate is utilized as a cooling block for the cold preservation of sorbent tubes. The aluminum fin block, affixed to the heat-dispersing side of the plate, facilitates heat dissipation in conjunction with the micro fans. Additionally, a vent was designed at the back of the suitcase, near the semiconductor refrigeration unit, and incorporates a fan to enhance airflow with the external environment, thereby improving heat dissipation. The temperature is controlled by a digital thermostat. A rechargeable 20000 mAh lithium battery (12 V) serves as the power source. The power level is monitored by an indicator and can be recharged via the charging port. Three switches are used to control the main power supply, the air pump, and the semiconductor refrigeration unit, respectively. The accessory module consists of NaCl bags, pH buffer solution tubes, internal standard vials, a glass rod and two micro syringes. The entire device weighs 3 kg and can be easily carried by hand or drone (DJI, M300RTK) (Fig. 1d).

  Figure 1

  

  Figure 1.  (a) Schematic of the µP&T device. (b) Schematic of the semiconductor refrigeration unit. (c) Top-down view and front cross-section view of the main module. (d) Photograph of the assembled µP&T device.

  DownLoad: Full-Size Img PowerPoint

  Before extraction, all sorbent tubes should be conditioned under argon at certain temperature (Table S2 in Supporting information) for 15 min. Water samples (50 mL) were transferred into purge vessels, and 25 ng of internal standard was spiked. The purge vessels were immediately sealed and then placed in the device for equilibration. After 1 min of equilibration, the purge vessels were promptly connected to the gas-water separators and sorbent tubes. Then, the power switch and pump switch were turned on and the purge-trap process began with air generated by the pump. The air was first purified to remove atmospheric particulate matter and potential contaminants (Fig. S3 in Supporting information). Consequently, the purified air was divided into nine channels and then utilized as purge gas. During this process, the analytes were transferred from the aqueous phase to the sorbent tubes. After extraction, the sorbent tubes were sealed and stored in the cooling block. The entire preparation process with the µP&T device took 20–30 min. After extraction, the sorbent tubes were transported to the laboratory and analyzed by thermal desorption-gas chromatography-silicon emission detector (TD-GC-SiED) (Figs. S4–S6 and Table S3 in Supporting information).

  An ideal sorbent for the extraction of VOCs was first investigated. Five commercially available sorbents including Tenax GR [13], Tenax TA [22,23], activated carbon [24], silica gel [24] and XAD resin [25] were studied (Fig. S7 in Supporting information). Among these five materials, Tenax GR and Tenax TA exhibited higher P&T efficiencies, suggesting that these two materials share similar properties and possess greater adsorption efficiencies for the four VOCs. In addition, these two sorbents are general-purpose phases suitable for VOCs in the range of C₆–C₃₀ [13]. Compared to Tenax TA, Tenax GR has lower water retention [26], making it more suitable for dry preservation and subsequent GC analyses. Therefore, Tenax GR was selected as the sorbent for the µP&T device.

  To achieve the optimal extraction performance for the VOCs, different parameters that affected the extraction efficiency of the µP&T device were optimized using standard solutions containing the tested VOCs (2.0 µg/L). A small amount of green inorganic salts was used to adjust the pH of water and reduce the solubility of VOCs. The maximum extraction efficiency was achieved at pH 7 (Fig. S8a in Supporting information) because VOCs are more stable in a neutral molecular state, which facilitates their purging and subsequent trapping using sorbent tubes. Consequently, the pH was adjusted to 7 by adding 5 mL of a phosphate buffer prepared by mixing appropriate amounts of 0.2 mol/L anhydrous NaH₂PO₄ and 0.2 mol/L anhydrous Na₂HPO₄. The addition of sodium chloride can increase the ionic strength, reducing the solubility of cVMSs, thereby enhancing their transfer to the headspace and adsorbent, and improving P&T efficiency. However, NaCl also increases the solution viscosity, slowing purging kinetics. A 10% salt solution was selected for the subsequent experiments (Fig. S8b in Supporting information). The flow rate of the purge gas and purge time are key parameters affecting P&T efficiency. These parameters are influenced by VOCs volatility (boiling point and vapor pressure), sample matrix interactions (e.g., solubility), and adsorption in the trapping tube (related to molecular weight). As the flow rate increased from 40 mL/min to 70 mL/min, responses for D3, D4, and D5 rose, peaking at 70 mL/min, while D6 showed an optimal response at 80 mL/min (Fig. S8c in Supporting information). To enhance the analytical sensitivity of D6, a flow rate of 80 mL/min was determined as optimal (Fig. S8f in Supporting information). Low-volatility siloxanes (D5 and D6) are more sensitive to changes in gas flow rate. Short purge times are insufficient for low-volatility analytes, while longer times risk losing more volatile compounds (D3 and D4), which may be blown off from the sorbent. Thus, an optimal purge time of 15 min was established (Fig. S8d in Supporting information). According to our previous work [27], temperature significantly affects the extraction efficiency of VOCs. However, heating requires high energy. Therefore, no heating was utilized in the µP&T device. To mitigate the effects of temperature variations at different sampling locations on the analytical results, 25 ng of M4Q was added as an internal standard to each water sample. The effect of ambient temperature on the peak area ratio of analytes to the internal standard was evaluated (Fig. S8e in Supporting information), with no significant changes observed across the tested temperature range. The activities of VOCs in solution can be affected by factors such as suspended solids, ionic strength and dissolved organic matter (DOM) in the water, which vary widely in water systems. The absolute signals of the VOCs were affected by suspended solids (Figs. S8g and i in Supporting information) and ionic strength (Fig. S8j in Supporting information). However, the internal standard method successfully corrected for these effects (Figs. S8h and k in Supporting information). The DOM (humic acid used in this study) had a negligible effect on the studied VOCs, likely due to their low hydrophobicity (Fig. S8l in Supporting information).

  Under optimal experimental conditions, the performances of the device were evaluated and are summarized in Fig. S9 (Supporting information) and Table S4 (Supporting information). The linear correlations for all measured VOCs are greater than 0.9806. The inter-run and inter-tube reproducibility (n = 9) was within the ranges of 1.1%–2.5% and 2.6%–5.1%, respectively. The limits of detection (LODs) in standard solutions defined as the 3σ criterion (σ representing the signal-to-noise ratio) were 3.2, 2.5, 5.3 and 6.1 ng/L for D3, D4, D5 and D6, respectively. The reusability and stability of the Tenax sorbent tubes were also evaluated (Fig. S10 in Supporting information). The sorbent tubes can be reused for at least 100 cycles without a degradation in their adsorption performance. Additionally, the device exhibits outstanding mechanical properties (Fig. S11 in Supporting information). The use of a high-capacity lithium battery enables the device to operate continuously for over 8 h. Table S5 (Supporting information) compares the performance of the proposed method with other similar methods. To the best of our knowledge, this is the first instance in which semiconductor refrigeration storage with precise temperature control has been integrated into a field sampling device. This innovation enables real-time refrigeration of the extraction phase in the field, reducing reliance on traditional transportation refrigeration equipment, enhancing sampling flexibility and adapting to a variety of environmental conditions. Moreover, precise temperature control minimizes sample variability caused by temperature fluctuations, thereby improving the accuracy and reliability of test results. Other advantages of the proposed device include high sample throughput, environmental sustainability, and convenience for field applications, as it eliminates the need for stirring or heating.

  To assess the importance of cold preservation, we compared the storage stability of VOCs in water samples directly with VOCs extracted using sorbent tubes and stored in a customized semiconductor refrigeration unit, which primarily consists of a semiconductor chilling plate (Fig. 2a). At ambient temperatures ranging from 11.6 ℃ to 40.0 ℃, the temperature of the cooling block drops rapidly after the power supply is activated, reaching its minimum within 30 s (Fig. 2b). Even at ambient temperature as high as 40.0 ℃, the temperature of the cooling block can still be lowered to 4.0 ℃. A real wastewater sample and a spiked standard solution (0.5 µg/L) stored directly at ambient temperature (21.5 ℃) were analyzed (Figs. 2c and d). The concentrations of VOCs in both samples decrease rapidly with the increase of storage time, especially for the highly volatile D3 and D4. These results confirm that traditional methods of direct storage and transportation of water samples lead to the loss of VOCs, resulting in errors during sample analyses. In contrast, the loss of VOCs adsorbed in sorbent tubes was significantly reduced when stored in the semiconductor refrigeration unit at room temperature (Fig. 2e). As the storage temperature decreased to 4 ℃, the loss of analytes significantly decreased, with less than 5% of the analytes lost after 11 days of storage. These results indicate that a cold environment is more favorable for the storage of VOCs. The storage stability of the device at 4 ℃ was further investigated (Fig. 2f). The analytes remained stable for at least 20 days and over a month except for D3, proving the device's suitability for field sampling of VOCs in remote areas.

  Figure 2

  

  Figure 2.  (a) Photograph of the semiconductor chilling plate. (b) The temperature profile of the cooling block. (c) A wastewater sample stored directly at room temperature. (d) A spiked standard solution stored directly at ambient temperature. (e) The effects of storage temperature and time on the recoveries of VOCs stored by the proposed method. (f) The stability of VOCs stored at 4.0 ℃ by the proposed method.

  DownLoad: Full-Size Img PowerPoint

  To verify the feasibility, portability and on-site sampling capability of the µP&T device, multiple VOCs sampling tests were conducted on real environmental water from ten wastewater treatment plants (WWTPs) and Sancha lake in Chengdu, Sichuan. The overall workflow is illustrated in Fig. 3a. Water samples were extracted on-site (Fig. 3b and Movie S1 in Supporting information). In hazardous or extreme environments where manual sampling is impractical, drones can directly collect samples and return them for processing. After extraction, the sorbent tubes were preserved and transported to the laboratory at 4 ℃ for further analysis. Notably, drones also assist in transporting the µP&T device to or from difficult-to-reach locations, overcoming terrain challenges and reducing safety risks (Fig. 3b and Movie S2 in Supporting information). In all wastewater samples (Fig. 3c and Table S6 in Supporting information), the concentration of the detected VOCs rang from below the LOD to 0.070 µg/L, 0.011–0.309 µg/L, 0.025–0.614 µg/L and below the LOD to 0.147 µg/L for D3, D4, D5, and D6, respectively. In contrast, four cVMSs were not detected in the five surface water samples. To further validate the accuracy and practicality of the proposed method, a spike recovery test was conducted on all samples. As shown in Table S6, the recoveries of all analytes range from 82.5% to 100.3%, indicating that the method provides acceptable accuracy for quantifying VOCs.

  Figure 3

  

  Figure 3.  (a) The overall workflow for detecting target VOCs. (b) Photographs of on-site application of the µP&T device (IS, internal standard). (c) Spatial distribution of VOCs detected in WWTP influents.

  DownLoad: Full-Size Img PowerPoint

  In conclusion, a battery-powered µP&T device integrated with a semiconductor cooler has been developed for the on-site extraction and cold preservation of VOCs in water samples. The device integrates sampling, extraction, preservation, and injection into one step, overcoming the challenges of preserving large-volume liquid samples and the inconvenience of transporting them. The semiconductor cooler greatly extends the sample storage time, effectively reducing analytes loss during collection, storage and transportation, thereby ensuring the reliability and accuracy of analytical results. The system’s high throughput significantly enhances the experimental efficiency. Additionally, the device is robust and easy to manufacture using inexpensive, commercially available materials and 3D printing technology. Its operation is simple and eco-friendly. Furthermore, the device can be customized to extract various VOCs by selecting the appropriate sorbent based on the properties of the analytes. The method provides a cost-effective solution for monitoring environmental water quality in small or remote areas with limited resources. However, some drawbacks remain. The device requires an accessory module to accommodate components for pH and ionic strength adjustments. Moreover, manual operation is required for processes such as sampling, conditioning, attaching and removing the trap tube, and placing it into the cooler. Future improvements could focus on developing a more compact version that eliminates the accessory module while enhancing device’s autonomy.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Yuan Yang: Writing – original draft, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. Yue Wang: Writing – review & editing. Xi Wang: Writing – review & editing, Resources. Hanshuang Li: Writing – review & editing. Xiaoli Wu: Writing – review & editing. Yurong Deng: Writing – review & editing, Supervision. Chengbin Zheng: Writing – review & editing, Supervision, Project administration, Conceptualization.

  Acknowledgments

  The authors gratefully acknowledge the National Natural Science Foundation of China (No. 22306146) and the PhD Scientific Research Startup Foundation of Xihua University (No. RX2200002003) for their financial support.

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111395.

  
  
• 1. [1]

     N.S. Chary, A.R. Fernandez-Alba, TrAC Trends Anal. Chem. 32 (2012) 60–75. doi: 10.1016/j.trac.2011.08.011

  2. [2]

     J. Zhang, W. Huang, R. Wu, et al., Environ. Sci. Technol. 57 (2023) 20864–20870. doi: 10.1021/acs.est.3c07193

  3. [3]

     X. Jin, Y. Wu, M. Santhamoorthy, et al., Chemosphere 308 (2022) 136182. doi: 10.1016/j.chemosphere.2022.136182

  4. [4]

     M. Sajid, Trends Environ. Anal. Chem. 36 (2022) e00175. doi: 10.1016/j.teac.2022.e00175

  5. [5]

     W. Hao, D.B. Cardin, Anal. Chem. 95 (2023) 3959–3967. doi: 10.1021/acs.analchem.2c03414

  6. [6]

     X. Liu, Z. Liu, Z. Zhu, et al., Anal. Chem. 89 (2017) 3739–3746. doi: 10.1021/acs.analchem.7b00126

  7. [7]

     J. Dong, J. Liu, P. Xing, et al., Anal. Chem. 95 (2023) 6271–6278. doi: 10.1021/acs.analchem.2c04789

  8. [8]

     H. Li, Y. Luo, Z. Li, L. Yang, Q. Wang, Anal. Chem. 84 (2012) 2974–2981. doi: 10.1021/ac3001995

  9. [9]

     E. Hanhauser, M.S. Bono, C. Vaishnav, A.J. Hart, R. Karnik, Environ. Sci. Technol. 54 (2020) 2646–2657. doi: 10.1021/acs.est.9b04695

  10. [10]

      F.A. Casado-Carmona, J.M. Jiménez-Soto, R. Lucena, S. Cárdenas, Anal. Chim. Acta 1189 (2022) 339186. doi: 10.1016/j.aca.2021.339186

  11. [11]

      S. Molaei, A. Saleh, V. Ghoulipour, S. Seidi, Chromatographia 79 (2016) 629–640. doi: 10.1007/s10337-016-3069-1

  12. [12]

      J.J. Grandy, V. Singh, M. Lashgari, M. Gauthier, J. Pawliszyn, Anal. Chem. 90 (2018) 14072–14080. doi: 10.1021/acs.analchem.8b04544

  13. [13]

      A.C. Paiva, J. Crucello, N. de Aguiar Porto, L.W. Hantao, TrAC Trends Anal. Chem. 139 (2021) 116252. doi: 10.1016/j.trac.2021.116252

  14. [14]

      X. Mao, B. Hu, M. He, W. Fan, J. Chromatogr. A 1260 (2012) 16–24. doi: 10.1016/j.chroma.2012.08.062

  15. [15]

      S. Huang, S. He, H. Xu, et al., Environ. Pollut. 200 (2015) 149–158. doi: 10.1016/j.envpol.2015.02.016

  16. [16]

      J.J. Grandy, V. Galpin, V. Singh, J. Pawliszyn, Anal. Chem. 92 (2020) 12917–12924. doi: 10.1021/acs.analchem.0c01490

  17. [17]

      Y. Liu, H.F. Li, J.H. Zhang, T. Maeda, J.M. Lin, Chin. Chem. Lett. 21 (2010) 730–733. doi: 10.1016/j.cclet.2009.12.024

  18. [18]

      Y. Pan, F.F. Deng, Z. Fang, et al., Chin. Chem. Lett. 32 (2021) 3440–3445. doi: 10.1016/j.cclet.2021.05.067

  19. [19]

      M. He, S. Lv, R. Yin, et al., Anal. Chem. 96 (2024) 2767–2773. doi: 10.3390/molecules29122767

  20. [20]

      Y. Chen, J. Li, X. Hou, et al., J. Chromatogr. A 1619 (2020) 460947. doi: 10.1016/j.chroma.2020.460947

  21. [21]

      M.A. Vargas-Muñoz, C. Palomino, G. Turnes, E. Palacio, J. Environ. Chem. Eng. 11 (2023) 110503. doi: 10.1016/j.jece.2023.110503

  22. [22]

      L. Zhang, X. Chen, G. Luo, et al., Sci. Total Environ. 912 (2024) 169106. doi: 10.1016/j.scitotenv.2023.169106

  23. [23]

      Q.B. Li, X.F. Wang, X. Wang, Y. Lan, J. Hu, Environ. Sci. Proc. Impacts 22 (2020) 973–980. doi: 10.1039/c9em00445a

  24. [24]

      J. Wang, L. Liao, L.A. Wang, L. Wang, Biomass Bioenergy 158 (2022) 106347. doi: 10.1016/j.biombioe.2022.106347

  25. [25]

      F. Okan, M. Odabasi, B. Yaman, Y. Dumanoglu, Environ. Sci. Technol. 55 (2021) 4522–4531. doi: 10.1021/acs.est.1c00227

  26. [26]

      T. Lomonaco, P. Salvo, S. Ghimenti, et al., J. Breath Res. 15 (2021) 037102. doi: 10.1088/1752-7163/abf0b4

  27. [27]

      Y. Yang, Y. Lin, J. Yang, et al., Anal. Chim. Acta 1191 (2022) 339288. doi: 10.1016/j.aca.2021.339288

• 

  Figure 1  (a) Schematic of the µP&T device. (b) Schematic of the semiconductor refrigeration unit. (c) Top-down view and front cross-section view of the main module. (d) Photograph of the assembled µP&T device.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) Photograph of the semiconductor chilling plate. (b) The temperature profile of the cooling block. (c) A wastewater sample stored directly at room temperature. (d) A spiked standard solution stored directly at ambient temperature. (e) The effects of storage temperature and time on the recoveries of VOCs stored by the proposed method. (f) The stability of VOCs stored at 4.0 ℃ by the proposed method.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) The overall workflow for detecting target VOCs. (b) Photographs of on-site application of the µP&T device (IS, internal standard). (c) Spatial distribution of VOCs detected in WWTP influents.

  下载: 全尺寸图片 幻灯片
•