Membrane-Extraction Ion Mobility Spectrometry for in Situ Detection of

Mar 24, 2010 - Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831 and Dalian Institute of Chemical Physics, Chinese Academy of ...
0 downloads 0 Views 924KB Size
Anal. Chem. 2010, 82, 4089–4096

Membrane-Extraction Ion Mobility Spectrometry for in Situ Detection of Chlorinated Hydrocarbons in Water Yongzhai Du,† Wei Zhang,† William Whitten,† Haiyang Li,‡ David B. Watson,† and Jun Xu*,† Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831 and Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023 Membrane-extraction ion mobility spectrometry (ME-IMS) has been developed for in situ sampling and analysis of trace chlorinated hydrocarbons in water in a single procedure. The sampling is configured so that aqueous contaminants permeate through a spiral hollow poly(dimethylsiloxane) (PDMS) membrane and are carried away by a vapor flow through the membrane tube. The extracted analyte flows into an atmospheric-pressure chemicalionization (APCI) chamber and is analyzed in a specially made IMS analyzer. The PDMS membrane was found to effectively extract chlorinated hydrocarbon solvents from the liquid phase to vapor. The specialized IMS analyzer has measured resolutions of R ) 33 and 41, respectively, for negative- and positive-modes and is capable of detecting aqueous tetrachloroethylene (PCE) and trichloroethylene (TCE) as low as 80 and 74 µg/L in the negative ion mode, respectively. The time-dependent characteristics of sampling and detection of TCE are both experimentally and theoretically studied for various concentrations, membrane lengths, and flow rates. These characteristics demonstrate that membrane-extraction IMS is feasible for the continuous monitoring of chlorinated hydrocarbons in water. Contaminated groundwater in U.S. legacy sites is a major concern due to the potential hazard of certain pollutants such as dense nonaqueous phase liquids (DNAPLs). Chlorinated hydrocarbons (ClHCs) are particularly important because they constitute the major portion of DNAPLs. In order to assess the status of the pollutants and to assist remediation efforts, technologies for monitoring chlorinated solvents are needed. Current monitoring methods usually involve complex procedures of installation and maintenance of groundwater wells, sampling, shipping, and laboratory analysis using gas chromatography/mass spectrometry (GC/MS; EPA SW-846 Method 8260B). These procedures are time-consuming, labor-intensive, and expensive. It is desirable to have a field-deployable in situ sensor that combines sampling and detection of the chlorinated solvents from groundwater as a single procedure. * Corresponding author. Phone: (865)574-8955. Fax: (865)576-5235. E-mail: [email protected]. † Oak Ridge National Laboratory. ‡ Chinese Academy of Sciences. 10.1021/ac100162d  2010 American Chemical Society Published on Web 03/24/2010

Ion mobility spectrometry (IMS) technology as a field-deployable vapor analyzer has been used for detecting chemical warfare agents, toxic industrial compounds, and trace explosives. It is of great interest to use IMS as a field-deployable sensor for monitoring trace chlorinated solvents. Previously, GC-IMS has been used for ex situ analysis of contaminated water.1-3 GCdifferential ion mobility spectrometry (DMS) has been studied for detecting halocarbons in the vapor phase.4 In 2006-2007, Kanu et al.5 and Sevier et al.6 reported a subsurface IMS for detecting soil-gas chlorinated hydrocarbons. Different from mass spectrometry (MS), IMS is generally intrinsically operated in ambient atmospheric pressure, without a vacuum. As a result, IMS has proportionally higher sensitivity because more analyte molecules are present within the analyzer. Such a feature also allows combining IMS with membranes for direct sampling of water analyte. In such combination, the transport gas flow in a tubular membrane is compatible with the IMS drift flow. To sample chlorinated hydrocarbons from water, a membrane inlet system has been considered for converting chlorinated solvents in water to a stream of vapor that can be analyzed subsequently by IMS. Membranes, such as polydimethylsiloxane (PDMS), have been used as a pervaporation technology for organic solvent recovery.7,8 Such membrane separation technology is feasible for detection purposes such as sampling aqueous volatile organic compounds (VOC).9,10 In fact, membrane inlets for sampling water analytes have been used for (1) Sielemann, S.; Baumbach, J. I.; Pilzecker, P.; Walendzik, G. Int. J. Ion Mobility Spectrom. 1999, 2, 15–21. (2) Stach, J.; Arthen-Engeland, T.; Flachowsky, J.; Borsdorf, H. Int. J. Ion Mobility Spectrom. 2002, 5, 82–86. (3) Walendzik, G.; Baumbach, J. I.; Klockow, D. Anal. Bioanal. Chem. 2005, 382, 1842–1847. (4) Eiceman, G. A.; Krylov, E. V.; Tadjikov, B.; Ewing, R. G.; Nazarov, E. G.; Millerc, R. A. Analyst 2004, 129, 297–304. (5) Kanu, A. B.; Hill, H. H., Jr.; Gribb, M. M.; Walters, R. N. J. Environ. Monit. 2007, 9, 51–60. (6) Sevier, D.; Gribb, M.; Walters, R.; Imonigie, J.; Ryan, K.; Kanu, A.; Hill, H. H.; Hong, F.; Baker, J.; Loo, S. M. Geotech. Spec. Publ. 2006, 1 (147), 25–234. (7) Vankelecom, I. F. J.; Dotremont, C.; Morobe, M.; Uytterhoeven, J.-B.; Vandecasteele, C. J. Phys. Chem. 1997, B10, 2154–2159. Vankelecom, I. F. J.; Dotremont, C.; Morobe, M.; Uytterhoeven, J.-B.; Vandecasteele, C. J. Phys. Chem. B 1997, 101, 2160–2163. (8) Dutta, B. K.; Sikdar, S. K. Environ. Sci. Technol. 1999, 33, 1709–1716. (9) Smitha, B.; Suhanya, D.; Sridhar, S.; Ramakrishna, M. J. Membr. Sci. 2004, 241, 1–21. (10) Panek, D.; Konieczny, K. Sep. Purif. Technol. 2007, 57, 507–512.

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

4089

Figure 1. Schematic experimental setup for membrane-extraction IMS.

mass spectrometry.11-13 Membrane-extraction IMS is expected to avoid many difficulties associated with transport gas. For either IMS or MS, a hollow membrane layer is used with one side contacting the aqueous sample and the other side interfacing with a vapor flow (called “flow-over”). This “flow-over” is needed to remove permeants from the membrane surface, maintaining the permeation gradient. However, such flow-over may create difficulties for mass spectrometry due to conflicting effects of the transport gas. If a large flow of transport gas is used inside the membrane tube, it can dilute the sample concentration due to the low pressure required for MS.12 On the other hand, if no transport gas is used, the slow diffusion of permeant along the tube reduces the gradient of permeant concentrations across the membrane layer and consequently reduces permeant flow.13 These conflicting effects can be avoided in membrane-extraction IMS because this technique operates with a large drift flow at near atmospheric pressure. The transport gas flow is not expected to interrupt the drift flow, while delivering permeants to the ionization chamber of the IMS. In this work, poly(dimethylsiloxane) (PDMS) membrane-extraction IMS has been developed to extract trace levels of chlorinated hydrocarbons from water and to qualitatively and quantitatively detect these pollutants. EXPERIMENTAL SECTION The schematic experimental setup for membrane-extraction IMS experiments is shown in Figure 1. Laboratory air was dried by a molecular sieve and activated charcoal and then split into two streams via computerized flow controllers (Apex Schoonover, Inc., mass and volumetric flow controllers, 16 series). One stream, about 300 standard cubic centimeters per minute (SCCM), was used as the IMS drift gas in the direction counter to the ion drift velocity. The other stream (∼30 SCCM) was sent into a spiral-shaped PDMS membrane tube as the transport flow. The outer surface of the membrane was in contact with water spiked with a contaminant. The analyte permeated (11) Camilli, R.; Hemond, H. F. Trends Anal. Chem. 2004, 23, 307. (12) Short, R. T.; Toler, S. K.; Kibelka, G. P. G.; Rueda-Roa, D. T.; Bell, R. J.; Byrne, R. H. Trends Anal. Chem. 2006, 25, 637–646. (13) LaPack, M. A.; Tou, J. C.; Enke, C. G. Anal. Chem. 1990, 62, 1265–1271.

4090

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

through the membrane layer and was carried by the transport flow into the IMS ionization chamber. Downstream from the membrane extraction, a dew point sensor (COSA Instrument XR 71110 dew point meter, Transcat, Inc.) was used to monitor the moisture level in the vapor. A specially made IMS analyzer was used to qualitatively and quantitatively analyze permeants that originated from the water. Chlorinated solvents studied in this work included trichloroethylene (TCE, 99.5%, spectrophotometric grade, Sigma Aldrich) and tetrachloroethylene (PCE, > 99.9%, Chromasolv for HPLC, Sigma Aldrich). A chlorinated solvent was mixed with deionized water in a volumetric flask with a known concentration varying between 0.2 and 100 ppm by volume. We will refer to concentration by volume unless stated otherwise. The water (18 MΩ cm) was purified by Millipore Ultrapure Water Systems (Direct-Q 5). The flask of the contaminant-water solution was sealed and shaken in an ultrasonic bath for 10 min. The mixed solution was introduced into a thermostatic bath housing a spiral membrane tube, as shown in Figure 1. The solution was sealed and stirred continuously with a magnetic bar during the time when membrane extraction occurred. PDMS membrane tubes were obtained commercially (VWR Labshop) and used for extracting chlorinated solvents from water to vapor. PDMS tubes were chosen because they are highly permeable toward organic vapor and show good chemical stability. The tube i.d. was 1.47 mm and the wall thickness was 0.23 mm. The length of the membrane that was in direct contact with water varied from 8 to 150 cm. Since the as-received membrane was typically contaminated with solvents and impurities, these membranes were cleaned initially in a methanol bath followed by a methanol removal procedure. Then the membrane was further cleaned by flushing the inside of the membrane tube with dry air and annealing in hot water. This procedure took about 20 h. The entire cleaning procedure was monitored using an online miniature ion-mobility spectrometer. The permeation temperature was controlled in the range of 0-70 °C using a magnetic stirrer heater (CAT, MCS67). A specialized IMS analyzer was made which consisted of an atmospheric pressure chemical ionization (APCI) chamber, a

geneous everywhere at any given time. Hence, the TCE concentration in water can be determined using the following mass conservation equation.

Vw

Figure 2. IMS spectrum for dry air flowing into a 50 cm PDMS membrane tube. Drift bias is -2000 V, drift tube temperature is 50 °C.

Bradbury-Nielson (B-N) gate, a drift tube, a Faraday detection plate, and a preamplifier. The analyte was first sent into the APCI chamber using the transport flow. The ionization source used was 5-mCi 63Ni. These ions were then injected by the B-N gate into the IMS drift channel. The homemade B-N gate consisted of 13 interdigitized grids located in the same plane and supported by a circular printed circuit board material with an inner diameter (i.d.) of 10 mm. The IMS electrodes were made of stainless steel rings separated by alumina rings. The distance between the adjacent electrodes was 3.8 mm, and the inner diameter of the rings was 10.8 mm. A total of 18 resistors of 1 MΩ each with 1% variation were placed between adjacent ring electrodes in the drift cell. The current passing through these resistors was measured as 0.100 mA. Electric potentials were distributed to these electrodes to generate a uniform drift field, as well as to the source electrode to direct ions close to the B-N gate. The effective length of the drift tube is 6.2 cm. The drift gas had a flow rate of approximately 300 SCCM opposite to the ion travel. The working pressure inside the drift cell was atmospheric pressure, 760 Torr. The entire drift channel was temperature-controlled with a tape-heater. A homemade current-sensitive preamplifier was used with 10 kHz bandwidth and a gain of 2 × 109 V/A. The output voltage was sent to a digital oscilloscope (Tektronix TDS2024B, 200 MHz) and a subsequent computer for recording. Figure 2 shows a typical negative-mode IMS spectrum of dry air at 50 °C where the major peak is O2-(H2O)n.14 This IMS spectrometer is capable of producing a resolution of R ) 33 in negative ion mode and R ) 41 in positive ion mode at 50 °C. THEORETICAL CONSIDERATIONS A transient model was developed to calculate the distribution of a contaminant in water and carrier gas as a function of time. TCE is used as the contaminant. The model considers three interlaced transport phenomena of TCE: (1) mixing in the water, (2) permeation through the membrane layer, and (3) forced diffusion in the membrane tube. The mixing of TCE in water is driven by magnetic stirring, as described above. It is reasonable to assume that the induced flow is so vigorous that the concentration of TCE in water is homo(14) Eiceman, G. A.; Karpas, Z. J. Am. Soc. Mass Spectrom. 1994, 5, 177.

∂Cw ) -SbJ(t) ∂t

(1)

where cw is the molar concentration of TCE in water (moles per cubic meter), t is the time, Vw is the total volume of water, Sb is the surface area of the membrane, and J(t) is the permeation flux (moles per meters squared second) at time t. Details of the permeation flux, J(t), are discussed below. In eq 1, the left-hand side is the rate of TCE concentration decrease in water, while the right-hand side represents the rate of TCE removal from water via permeation. The permeation stage consists of absorption of TCE into the outer surface of the PDMS tube, diffusion through the porous membrane layer, and dissolution from the membrane inner surface to vapor. Detailed determination of the permeation flux of TCE through the PDMS layer is complex and requires input data such as the diffusion coefficient and sorption coefficient,15 which are not readily available. For simplicity, it is assumed that the steadystate permeation flux (Jss) can be calculated using the apparent permeability, as shown in the following equation. Jss ) P(cw - ca)

(2)

where P is the apparent permeability (meter per second) and cw and ca are the molar concentrations of TCE in water and vapor, respectively. It is noted that P is related not only to diffusion in the membrane layer but also to the adsorption on and dissolution from the membrane surfaces. Since the membrane layer has a certain thickness, there is a finite time required for TCE to diffuse through the layer. To describe timedependent behavior, the permeation flux ramps up from zero to the steady-state value using the equation below:13 N

J(t) ) Jss[1 + 2

∑ (-1)

n

e-n At] 2

(3)

n)1

where J(t) is the transient permeation flux at time t and A is a constant (inverse seconds) that depends on the membrane properties such as thickness and diffusion coefficient. It is found that the fifth order solution (i.e., N ) 5) is sufficiently accurate. To derive the governing equations for forced diffusion of TCE in the membrane tube by the transport gas flow, the following two assumptions are used. First, the flow is assumed to be laminar and incompressible. This assumption is reasonable since the Reynolds number is approximately 80 for an air flow velocity of 0.75 m s-1 and a membrane i.d. of 1.47 mm, much less than the typical threshold value of 2100 for the onset of turbulent flow in a round pipe.16 Second, the presence of TCE is assumed to not interfere with the air flow. This assumption is plausible since the concentration of TCE is typically very low in air. (15) Baker, R. W. Membrane Technology and Applications, 2nd ed.; John Wiley & Sons: New York, 2004. (16) Munson, B. R.; Young, D. F. Okiishi, T. H. Fundamentals of Fluid Mechanics; John Wiley & Sons: New York, 1990.

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

4091

Table 1. Input Data Used in the Coupled Fluid Flow and Mass Transport Calculation input -3

density (kg m ) viscosity (kg m-1 s-1) diffusion coefficient of TCE (m2 s-1)

water

air

998.2 0.001

1.225 1.79 × 10-5 6.1 × 10-619

Considering these two assumptions, the air flow velocity and TCE concentration can be described by the continuity equation, moment conservation equation, and TCE mass transport equation. Details of the first two equations are presented in many textbooks17 and are not repeated here. The TCE mass transport equation is the following.

[

( )]

∂ca ∂(uca) ∂(vca) ∂2ca 1 ∂ ∂ca + + ) Da + r 2 ∂t ∂x ∂r r ∂r ∂r ∂x

(4)

where ca is the molar concentrations of TCE in air, u and v are the axial and radial velocities of the air, respectively, x and r are the tube axial and radial distance, respectively, and Da is the diffusion coefficient of TCE in air. The model is numerically solved using ANSYS FLUENT,18 which is based on the finite volume method. Table 1 summarizes the input data used in the coupled fluid flow and mass transport model. In addition to those data, the values of the apparent permeability (P) and the membrane constant (A) are required. In the present study, their values are determined by fitting the experimental data. RESULTS AND DISCUSSION Membrane Cleaning. The as-received membrane tubing typically has many residual solvent impurities inside the tube wall and contaminants on both inside and outside surfaces. Membraneextraction IMS allows first detection of impurities released from the PDMS membrane during the cleaning procedure. Figure 3 shows IMS spectra (a) before and (b) after cleaning. The IMS instrument was operated in positive ion mode, at room temperature, and at ambient pressure, with drift-field strength of ∼290 V/cm. The membrane tube was pretreated by dipping the outside of the membrane into a methanol-water solution and by passing

Figure 3. Positive ion mode IMS spectra of chemicals released from PDMS membrane (a) before and (b) after cleaning. 4092

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

Figure 4. Negative ion mode IMS spectra for TCE, PCE, and CH3I spiked in water with concentrations of 10 ppm, respectively. The numbers listed are drift times with a -290 V/cm drift field.

nitrogen gas through the inside for 30 min. Then the membrane tube was immersed in a flask containing clean water (200 mL). Dry air having around 80 ppm of gaseous H2O flowed through the inside of the membrane tube. The entire membrane-water assembly was sealed and placed in a temperature-controlled housing. Initially, as shown in Figure 3a, the IMS spectrum showed a dominant peak at 8.7 ms drift time and impurity peaks with drift times in the 10-22 ms range. The 8.7 ms peak is attributed to H3O+(H2O)n (n ) 2∼3) resulting from APCI20 and is referred to as the reactant ion peak (RIP). This peak gives a reduced mobility of Ko ) 2.08 cm2 V-1 s-1, in good agreement with that reported in ref 5. After about 20 h of the treatment of flowing dry air at 100 SCCM and heating at 50 °C (water temperature), the impurity peaks disappeared and the RIP seems to be the only peak, as shown in Figure 3b. At this stage, the membrane is believed to be reasonably clean. Identification and Quantification of Chlorinated Hydrocarbons (ClHCs). The clean membrane was immersed in a flask that contained 200 mL of water and a selected contaminant, either TCE, PCE, or CH3I with concentrations of 10 (14.6 mg/L), 10 (16 mg/L), and 10 ppm (22.8 mg/L), respectively. Figure 4 shows the IMS spectra in negative ion mode for the three contaminants. All RIP positions are at 8.14 ms, corresponding to a reduced mobility (Ko) of 2.23 cm-2 V-1 s-1 and agreeing with that reported in ref 5. Interestingly, the drift times of the signature peaks for both TCE and PCE are the same, 7.14 ms (Ko ) 2.53 cm-2 V-1 s-1), while the drift time for CH3I is at 7.44 ms (Ko ) 2.43 cm-2 V-1s-1), characteristically different. The 7.14 ms (Ko ) 2.53 cm-2 V-1 s-1) negative ion peak is attributed to Cl-(H2O)n (n ) 1-3) because the reduced mobility is in excellent agreement with the Cl-(H2O)n mobility listed on page 86 of ref 20 and with the TCE sample reported (17) Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena, 1st ed.; John Wiley & Sons: New York, 1960. (18) ANSYS. http://www.ansys.com/products/fluid-dynamics/fluent/. (19) Bartelt-Hunt, S. L.; Smith, J. A. J. Contam. Hydrol. 2002, 56, 193–208. (20) Eiceman, G. A.; Karpas, Z. In Ion Mobility Spectrometry; CRC, Taylor & Francis: Boca Raton, FL, 2005; pp 84-86.

Figure 7. Reaction schemes for positive ion formation of PCE.

Figure 5. Negative ion mode IMS spectrum of 0.32 mg/L PCE in water.

Figure 6. Positive-mode IMS spectra: (a) RIP with membrane in deionized water, (b) membrane in PCE headspace air, (c) membrane in PCE-water solution, and (d) membrane in TCE-water solution.

in ref 5. It is well-known21 that the electron affinities for halogen atoms are high, leading to formation of negative halogen ions. Then, the halogen ions are expected to capture H2O to form halogen-water ion complexes. The difference of the drift times between Cl-(H2O)n or I-(H2O)n is consistent with the difference of Cl and I masses. Hereafter, we refer to Cl-(H2O)n as the chlorine ion. The origin of other peaks at larger drift time 9 ms is not completely known. Two possiblities are (1) product ions resulting from halogen ions interacting with the membrane and/or (2) halogen ion clusters. To determine the limits of detection (LOD) for chlorinated solvents, an IMS spectrum of 0.32 mg/L PCE, equivalent to 200 ppb in water, was measured as shown in Figure 5. The Cl- peak intensity is 4 times higher than the 3σ variation, suggesting that the LOD for PCE is 80 µg/L (49 ppb). A similar test was made for TCE in water. The LOD for TCE in water is 51 ppb, 74 µg/L. Such low limits suggest that membrane-IMS is feasible for sensitive monitoring of ClHCs in groundwater. Positive-mode IMS spectra are shown in Figure 6 for (a) membrane in clean water (carrier gas moisture level is ∼540 ppm), (b) membrane in PCE-containing air (PCE headspace (21) Spangler, G. E.; Collins, C. I. Anal. Chem. 1975, 47, 393–402.

concentration at room temperature, PCE vapor pressure is 2.47 kPa22), and (c) membrane placed in PCE-water solution with the saturated concentration (0.15 g/L). The values listed in the figure are the reduced mobilities. For the clean membrane in the deionized water, the IMS spectrum only shows the RIP, as expected. For the membrane in the PCE-water solution, the additional peaks are shown in Figure 6c. These peaks are attributed to product ions that have different numbers of chlorine clusters. To confirm the origin of these products from PCE, Figure 6b shows the measured IMS spectrum for membrane sampling from the PCE-containing air. Some of the peaks observed from the PCE-water solution (Figure 6c) are consistent with those observed in the PCE vapor, while some are not. Figure 6d shows positive-mode IMS spectrum after the membrane was placed in TCE-water solution (0.47 g/L). It is clear that the ion peaks for TCE are different from those for PCE. Some of these peaks are similar to a recent work reported by Kanu et al.5 and Sevier et al.6 and some are different. Although these positive ion spectra of ClHCs are complicated, they are capable of identifying individual ClHCs because the patterns of their product ions are different. We have found that it was difficult to observe positive ions from PCE unless the moisture level is below ∼400 ppm. Because of this reason, the above experiments were done at a high flow rate of the transport gas (100-300 SCCM) to reduce the moisture level. The mechanisms responsible for formation of positive ions of ClHCs are complicated and not well understood. The conventional mechanism of proton transfer from the RIP, H3O+(H2O)n, to a ClHC seems not to occur due to the unfavorable energetics.23,24 It is believed that formation of positive ions in APCI involves charge exchange with primary ions, such as H2O+, O2+, and N2+, as identified by APCI-mass spectrometry.23,24 Therefore, the moisture H2O (g), which is the origin of H3O+(H2O)n, competes with the ClHC to get the charge from the primary ions. Considering a high concentration of water, as is the case here, the mechanisms responsible for the positive ion formation of PCE are proposed as two steps, as shown in Figure 7. Step 1 is the charge exchange with a primary ion and step 2 is the ion-water interaction, resulting in product ions. C2HCl3O+ ion can also yield other smaller product ions by releasing HCl and CO and larger ions by clustering with H2O (g). Unlike the negative ion mode, positive-mode IMS can be used to identify a ClHC because its product ions have a signature of (22) Calculated using an equation from http://en.wikipedia.org/wiki/ Tetrachloroethylene_(data_page)#Vapor_pressure_of_liquid). (23) Dono, A.; Paradisi, C.; Scorrano, G. Rapid Commun. Mass Spectrom. 1997, 11, 1687–1694. (24) Nicoletti, A.; Paradisi, C.; Scorrano, G. Rapid Commun. Mass Spectrom. 2001, 15, 1904–1911.

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

4093

Figure 8. Time dependence of IMS signal intensity for the Cl-(H2O)n ion peak at 7.14 ms. The insert shows two consecutive runs of Cl-(H2O)n peak time-profiles over a large time window. The second run was recorded only for 250 min, which was sufficient to show the reproducibility of the permeation curves. Table 2. Experimental Conditions for TCE Detection TCE concentration: 146 mg/L

drift field: ∼290 V/ cm (on average)

membrane length: 50 cm

drift gas: 300 SCCM dry-air at 1 atm drift tube temperature: 50 °C no. of IMS scans: 16

water temperature: room temperature transport gas flow: 30 SCCM dry-air

the ClHC. Because of the high moisture level under current conditions, the sensitivities of positive-mode membrane-IMS to ClHCs are low. Time Dependence. To understand time-response for the membrane-extraction IMS sensor, the intensity of the negative chlorine ion peak was measured as a function of time after the membrane was immersed in TCE-containing water (200 mL), as shown in Figure 8. The zero-time is when the membrane was placed initially into the TCE-spiked water. Experimental conditions are given in Table 2. The time profile generally shows an increase of the TCE peak intensity as a function of time in early times (15 min), as shown in the figure. In order to investigate reusability of the membrane, we consecutively measured two time-profiles of the chlorine ion peak for two flasks containing the same concentration (80 ppm) of TCE in water, as shown as the insert of figure 8. In the first run, the chlorine peak intensity went to a very low value after 700 min permeation. Then the same membrane, without cleaning, was immediately put into the second flask. As shown in the insert, the second-run response was identical to the first profile within experimental errors. The reproducibility of the permeation curves shows that the loss of the Cl peak intensity is unlikely due to accumulation of TCE with membrane that leads to poor transport and instead is due to reduced concentration of TCE in water. In addition, the membrane was found to remain reusable for a few months. Various concentrations of TCE in water were measured using the membrane-IMS system. Their time dependent profiles are shown in Figure 9a. The increasing and decreasing phases of the profiles are approximately the same for all concentrations, while the chlorine IMS signal intensity (at 7.14 ms) increases with TCE concentration. The increase seems linear at low concentrations but tends to be saturated for high concentrations, as shown in Figure 9b. The similar time profiles at various concentrations suggest that linear processes are involved in the membrane-IMS system under current conditions. In other words, TCE molecules do not affect each other when partitioning on the membrane surfaces, diffusing inside the membrane layer, and transporting in the vapor. This is understandable because the TCE concentrations are low. The time profiles for 10 and 50 cm lengths of PDMS membrane tubes with the same i.d. and o.d. were measured, as shown in

Table 3. Calculated and Fitted Parameters for Two Flow Rates

Figure 10. Time profiles of the Cl ion peak for 10 and 50 cm lengths of PDMS membrane tubes.

Figure 11. Comparison between the experimental data and the calculated results for an air flow rate of 30 and 80 SCCM.

Figure 10. It is shown that the times to reach 90% of the highest intensity are 2.3 and 4.0 min for 10 and 50 cm lengths, respectively. For the 50 cm tube, the maximum intensity is approximately 70% higher than that for 10 cm tube. The response time for the 10 cm tube is not only short but also shows a long flat region. Both features are beneficial for developing field-based monitors because they require a short duty cycle and stable monitoring concentrations that are not altered by the permeation processes. The increase of IMS response intensity for a longer membrane is different from that observed for membrane-mass spectrometry.12,13 For a short length range (1-7 cm), the membrane-MS intensity can increase as the length increases.13 However, a longer membrane can be a pitfall in membrane-MS, because the slow diffusion process of membrane-MS without use of a transport gas can result in a reduced concentration difference across the membrane layer. If a transport gas is used, the analyte concentration can be diluted in MS.12 Therefore, in membrane-extraction IMS, the use of a flow rate in the range of 30-100 SCCM is feasible for IMS operation. Flow Rate Dependence. Two time profiles of the Cl peak were measured, as shown in Figure 11, after the 50 cm membrane was placed in a sealed 80 ppm TCE-water solution for two flow rates of the transport gas, (a) 30 and (b) 80 SCCM. Both the experimental data and the calculated results indicate that the TCE concentration in air increases initially, reaches a peak value, and then decreases. For the higher flow rate (80 SCCM), the Cl peak intensity at later times decays faster and the maximum intensity is lower than those for 30 SCCM. The faster decay is attributed to a larger permeation rate under a higher flow rate. This

Air flow rate

30 SCCM

80 SCCM

average air velocity (m/s) P (m/s) A (s-1)

0.28 1.3 × 10-5 7.5 × 10-3

0.75 4.7 × 10-5 7.5 × 10-3

correlation suggests that transport flow leads to a large difference of the concentrations across the membrane layer and increased apparent permeability. In Figure 11, the theoretical modeling of the time profiles for the two flow rates are plotted as solid lines. In these calculations, the parameters “P” and “A” were first fitted to the experimental data for 30 SCCM. Then “A” was maintained the same for both flow rates because it mainly depends on the permeating chemical and membrane, which are the same for both flow rate measurements. A new value of “P” was obtained by fitting to the 80 SCCM experimental data. As shown in Figure 11, modeling with the fixed “A” is reasonably consistent with experimental data, validating the coupled fluid flow and mass transport model. The fitting summarized in Table 3 shows that the apparent permeability increases with the increased air flow rate. Since this simplified model does not consider the details of TCE permeation, it is not able to explain the dependence of permeation flux on the flow rate. However, this trend is consistent with that reported in the literature.25 In that study, benzene was removed from a gas mixture through a membrane into an aqueous solution and the benzene mass transfer flux increased with the liquid flow rate. Detailed study on TCE permeation through the membrane layer is planned in the future to relax several assumptions used here and to gain a better understanding of the same. CONCLUSIONS Unlike conventional detection methods involving two separate procedures, i.e., sampling and subsequent analysis of TCE and PCE in water, a sensor that combines both a membrane tube and ion mobility spectrometry was developed as a single procedure for in situ monitoring of chlorinated hydrocarbons in water. Applying a transport gas though the membrane tube can increase permeation of the contaminants crossing the membrane layer while quickly sending analyte into the IMS analyzer without affecting analyzer detection. The results presented here provide a basis for developing field-deployable membrane-extraction IMS sensors for in situ monitoring of chlorinated solvents in groundwater. The time-dependent characteristics of sampling and detection of TCE have been both experimentally and theoretically studied for various concentrations, membrane lengths, and flow rates. These characteristics provide important clues in understanding permeation and sampling processes. The response times were obtained for various chemicals, membrane lengths, and flow rate of the transport gas. These response times are important to determine the duty cycles in developing fielddeployable sensors for monitoring ClHCs in water. Negative-mode IMS has very high sensitivity for monitoring TCE and PCE in water: 80 and 74 µg/L, respectively. However, (25) Xu, J.; Li, R.; Wang, L.; Li, J.; Sun, X. Sep. Purif. Technol. 2009, 68, 75–82.

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

4095

this polarity cannot resolve the type of chlorinated hydrocarbons. To identify individual ClHC molecules, positive ion mode IMS and/or hybrid IMS are needed. The positive ions for PCE were observed in this work. The framework developed here can be readily extended to pursue measurement of positive ions for a mixture of chlorinated hydrocarbons, such as PCE, TCE, DCE, and VC in the future.

portion of research was sponsored by the Laboratory Directed Research and Development Program. Oak Ridge National Laboratory is managed by UT-Battelle, LLC for the U.S. Department of Energy under Contract DE-AC05-00OR22725.

ACKNOWLEDGMENT Research was sponsored by the Strategic Environmental Research and Development Programs (SERDP). The modeling

Received for review January 19, 2010. Accepted March 8, 2010.

4096

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

AC100162D