Air-Cooled Cold Trap Channel Integrated in a ... - ACS Publications

In Search of the Ultimate Benzene Sensor: The EtQxBox Solution. ACS Sensors 2017, 2 (4) , 590-598. DOI: 10.1021/acssensors.7b00110. Yuko Ueno,, Tsutom...
0 downloads 0 Views 124KB Size
Download PDF
Anal. Chem. 2002, 74, 1712-1717

Air-Cooled Cold Trap Channel Integrated in a Microfluidic Device for Monitoring Airborne BTEX with an Improved Detection Limit Yuko Ueno,* Tsutomu Horiuchi, and Osamu Niwa

NTT Lifestyle and Environmental Technology Laboratories, 3-1, Morinosato Wakamiya, Atsugi, Kanagawa 243-0198, Japan

We integrated an air-cooled cold trap (CT) channel in a microfluidic device for monitoring airborne benzene, toluene, ethylbenzene, and xylene (BTEX) gases and demonstrated its effect on improving the detection limit of the microfluidic device. The device consists of concentration and detection cells formed of 3 × 1 cm Pyrex plates. We first introduced a sample gas into the concentration cell, and the gas was adsorbed onto an adsorbent in the channel. We then raised the temperature using a thin-film heater and introduced the desorbed gas into the detection cell. To prevent dilution of the gas before detection, we propose an improvement to the concentration cell structure that involves the integration of the CT channel. We examined the CT effect by comparing three types of concentration cell with different channel structures. We found that we could detect a gas concentration about 2 orders of magnitude lower than in our previous work by optimizing the channel structure and integrating a CT channel. As an example of BTEX detection, we obtained a 0.05 ppm detection limit for toluene gas with a sampling time of 30 min. Airborne benzene, toluene, ethylbenzene, and xylenes (BTEX) are volatile organic compounds (VOCs) of great social and environmental significance, because they are hazardous to the human nervous system even at parts-per-billion concentrations. Benzene is classified as a human carcinogen and is a risk factor for leukemia and lymphomas.1 The regulated standard concentration of benzene is 1, 3, and 5 µg/m3 (0.33, 1.0, and 1.6 ppb) in the United States,2 Japan,3 and the European Union,4 respectively. The guidelines for the upper indoor concentration limits of toluene, ethylbenzene and xylenes in Japan are 260, 3800, and 870 µg/m3 (0.07, 0.88, and 0.20 ppm), respectively. Because they are found (1) See, for example: U.S. Environmental Protection Agency. Carcinogenic Effects of Benzene: An Update, 1998; EPA/60/P-97/001F; U.S. Government Printing Office: Washington, DC, 1998. (2) U.S. Environmental Protection Agency. National Air Toxics Program: The Integrated Urban Strategy, Report to Congress, 2000; EPA-453/R-99-007;U.S. Government Printing Office: Washington, DC, 2000. (3) Environmental Quality Standards in Japan sAir Qualitys, Environmental Quality Standards of Benzene, Trichloroethylene and Tetrachloroethylene; http://www.env.go.jp/en/lar/regulation/aq.html; Ministry of the Environment, Government of Japan, 1997. (4) Are we moving in the right direction? TERM 2000. Environmental issues series No. 12, 2000; http://reports.eea.eu.int/ENVISSUENo12/en/page008.html; European Environment Agency.

1712 Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

mainly in automobile exhaust gases, their concentration in the air differs with time and location. Therefore, it is important to determine their concentration in the air by onsite field monitoring. The onsite field monitoring of BTEX requires a portable detector that has parts-per-billion sensitivity and gas selectivity. The most widely used conventional method for VOC detection is gas chromatography/mass spectrometry (GS/MS).5 The TO-14 method developed by the United States Environmental Protection Agency (EPA) also employs the GC/MS method for VOC detection,6 because this method has several advantages, namely a parts-per-trillion detection limit, high selectivity, and high accuracy. However, in terms of field monitoring, it has a crucial disadvantage in that the size and weight of a GC/MS instrument can be reduced only to a certain degree without degrading its selectivity and sensitivity. In addition, purified gas should be used for such measurements, and this is inconvenient for onsite monitoring. Thus, a portable system is required for onsite monitoring that can detect each BTEX species. Recently, we developed a microfluidic device designed to detect and identify BTEX in the air.7 Our device is composed of concentration and detection cells with peripheral devices. Each cell is 3 × 1 cm. To measure BTEX gases, we first introduced polluted air into the concentration cell by using a palm size pump, and the gas was adsorbed onto an adsorbent in the channel. After sampling the gas for a specified time, we raised the temperature using a thin-film heater in order to desorb the concentrated gas from the adsorbent. Then, we introduced the desorbed gas into the detection cell, which was aligned between a light source and a spectrometer, and measured the absorption spectra of the gas. The detection limit of our previously reported device is 4 ppm. However, this is insufficient for BTEX in the air. Therefore, to achieve a lower detection limit, it is important to reduce the diffusion of the sampled gas to narrow the band before introducing the gas into the detection cell. We thought that the integration of a cold trap (CT) channel in the detection cell, downstream of the adsorbent between the (5) See, for example: Bruner, F. Gas Chromatographic Environmental Analysis: Principles, Techniques, Instrumentation; John Wiley & Sons: New York, 1993. (6) U.S. Environmental Protection Agency. Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air, 2nd ed.; EPA/ 625/R-96/010b; Method TO-14A.; U.S. Government Printing Office: Washington, DC, 1999. (7) Ueno, Y.; Horiuchi, T.; Morimoto, T.; Niwa, O. Anal. Chem. 2001, 73 (19), 4688. 10.1021/ac0110810 CCC: $22.00

© 2002 American Chemical Society Published on Web 03/05/2002

Figure 1. Diagram of the proposed method of integrating an air-cooled CT channel in a concentration cell.

adsorbent channel and the detection cell, might prevent dilution and, thus, maintain the desorbed gas band at a high concentration before its introduction into the detection cell. Figure 1 illustrates the idea of the air-cooled CT channel. A method that combines thermal desorption and a cold trap is usually employed in the GC method.8 To provide higher resolution, it is necessary to employ a liquid nitrogen supply to freeze the sample and a reheating system to collect the frozen sample quickly. However, in the terms of field monitoring, it is more important to realize the CT function without using complicated systems than to realize the highest possible resolution. Thus, we simply removed the thin-film heater from the channel downstream of the adsorbent to form the CT channel. The temperature of a channel without a heater will not increase because the air cools this area. In this paper, we report an optimized gas concentration cell consisting of preconcentration and air-cooled trap channels with a view to improving the detection limit of the device. We found that we could detect a gas concentration about 2 orders of magnitude lower than in our previous work by optimizing the channel structure and integrating an air-cooled CT channel in the concentration cell. As an example of BTEX detection, we obtained a 0.05 ppm detection limit for toluene gas with a sampling time of 30 min. EXPERIMENTAL SECTION Improvement of the Concentration Cell. Figure 2 shows photographs of the three types of concentration cell we examined, namely, a reference cell with a channel of the same structure as that used in our previous report (R), a cell with a wider channel than R (W), and a cell with the same shaped channel as W and integrated with a CT channel (WCT). We fabricated the three

Figure 2. Photographs of the three types of concentration cell, namely a reference cell with a channel of the same structure as that described in our previous report (R), a cell with a wider channel than R (W), and a cell with the same shaped channel as W and integrated with a CT channel (WCT).

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

1713

Figure 3. Top and side views of the trench shapes of the concentration cells, R (A) and W/WCT (B).

types of concentration cell and a detection cell using almost the same processes that we have described elsewhere.7 Figure 3 shows top and side views of the trench shapes of the concentration cells, R (A) and W/WCT (B). For W and WCT, we formed an upstream section that we packed with adsorbent to a width three times greater in R to provide sufficient capacity for gas adsorption. We formed a shallow section in the middle of the channel as a stopper to prevent any movement of the packed adsorbent. In our previous study, we obtained a maximum flow rate of ∼1.2 × 10-2 cm3/s. We thought that we might be able to obtain a sharper signal response by using a higher flow rate, because the desorbed gas may be introduced into the detection cell more quickly. Moreover, we could collect a larger amount of gas in the same sampling time. We found that the depth of the adsorbent stopper had a dominant effect with regard to determining the gas flow resistance. Therefore, we positioned the stopper 20 µm deeper (8) Mattinen, M. J.; Sandell, E.; Saarela, K. Toxicological and Environmental Chemistry 1993, 40, 133-142. Kostiainen, R. Chromatographia 1994, 38 (11/12), 709. Suzuki, S. Anal. Sci. 1995, 11, 953-960. Etela¨talo, E. K.; Kostiainen, O.; Kokko, M. J. Chromatogr. A 1997, 787, 205-214. Pankow, J. F.; Luo, W.; Isabelle, L. M.; Bender, D. A.; Baker, R. J. Anal. Chem. 1998, 70, 5213-5221.

1714

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

for W and WCT than for R. We used a dicing saw to form the trench and so were unable to make the wide section downstream of the stopper any narrower. The section furthest downstream was narrow to prevent excess diffusion of the desorbed gas. To form a different type of concentration cell, we also changed the pattern of the thin-film heater, as shown in Figure 2. We formed the W and WCT heating areas with a zigzag shape so that they would heat the whole section where the adsorbent was packed. We ensured that all of the heaters had the same resistance by varying the heater thickness. We achieved this by controlling the sputtering conditions during their deposition. The heating areas were 14.0 × 1.0, 26.0 × 3.0, and 15.5 × 3.0 mm for R, W, and WCT, respectively. Measurement. To provide an artificial environment of BTEXpolluted air, we prepared toluene mixture gases by diluting commercial-grade toluene (500 and 50 ppm), benzene (50 ppm) and o-xylene (50 ppm) gases with nitrogen in a gas blender consisting of mass-flow meters (Estec, SEC-400). We introduced this toluene mixture gas into the concentration cell using a Bimor pump (Kyokko Co., Ltd., BPF-465P) 72 × 72 × 32 mm in size. Each part of the pump that came into contact

with the flowing gas was made of fluoroplastic. We were able to vary the gas flow rate by changing the output frequency of the pump. Here, we set the output frequency at 100 Hz, which gave us a flow rate of about 1.2 × 10-2 and 1.1 × 10-1 cm3/s for R and W/WCT, respectively, at the point furthest downstream from the device. The toluene gases were first introduced into the concentration cell and then adsorbed by the adsorbent in the channel. The adsorbent we used was commercially obtained amorphous silicon dioxide powder (Kanto Chemical Co. Inc.). We packed ∼0.3 and ∼1.5 mg of adsorbent, respectively, into the R and W/WCT channels. We sampled the toluene gas for a certain time, and then we raised the temperature using the thin-film heater so that the concentrated toluene gas was desorbed from the adsorbent. We set the applied voltage at 12, 17.5, and 14.5 V for R, W, and WCT, respectively. Although we designed the thin-film heaters of the CT cells to provide the same temperature at the same applied voltage, we had to adjust them to increase the temperature at the adsorbent to ∼200 °C because of the different thermal radiation rates of the three types of cell. We measured the temperature at various cell positions by using an infrared thermometer (Keyence, IT2-02 controlled by IT2-50). The measurement spot size was φ ) 1.2 mm. We then introduced desorbed toluene gas into the detection cell along with a carrier gas, namely the original toluene mixture gas we used as a sample. In terms of field monitoring, it is important to remove as much of any excess component as possible, in this case, the purge gas. Interference from the original gas was almost negligible, because its concentration was >2 orders of magnitude lower, which was within the margin of accuracy of this device, as we will describe later. In the detection cell, we measured the absorption spectra of the toluene gas by using optical fibers fixed at the edges of the channel aligned between a light source and a spectrometer, respectively. Because the absorption spectrum of the toluene spreads over the UV region, we used a 30-W deuterium (D2) lamp (Soma Optics) as the light source. We measured the absorption spectra using a UV spectrometer (Fastevert S-2400, Soma Optics) that was specially modified for this device. RESULTS AND DISCUSSION Figure 4 shows the thermal characteristics of W (a) and WCT (b). We set the time at which we started supplying a voltage to the heater at 0 s. The temperatures at the adsorbent were the same with W and WCT and reached ∼200 °C. The temperatures downstream of the stopper reached ∼175 and ∼150 °C for W and WCT, respectively. The temperatures were lower than that at the adsorbent, even when these areas were also covered with heaters. It is possible that the air in this area is replaced by cool outside air during the measurement, thus preventing any temperature increase. However, these areas do not have cold trap effects, because the temperatures are even higher than the boiling point of toluene (110.6 °C). In WCT, the temperature in the CT channel was ∼50 °C. This shows that we were able to obtain this large temperature difference on the same chip without using any form of cooling system. This is due to the small heat capacity of the thin-film heater. Figure 5 compares the device response curves obtained at the detection cell when using the three different concentration cells for the detection of 30 ppm of toluene. The toluene absorption spectrum obtained with our microfabricated detection cell was

Figure 4. Thermal characteristics of W (a) and WCT (b). The temperature was measured from the opposite side of the cell on which the thin-film heater was sputtered. The time was set at 0 s (t ) 0) when the heater switch was turned on.

Figure 5. Changes in the absorption intensity of the toluene gas at 267.5 nm for AWCT (9), AW (b), and AR (2) against time. The time was set at 0 s (t ) 0) when the switch of the heater was turned on.

indicative of the specific peaks of gas-phase toluene9 at 267.5, 263.5, 261.0, 259.0, 255.0, and 247.0 nm. The peak at 267.5 nm is the sharpest. Therefore, we used the absorbance of this peak (A) to evaluate the performance of the device. We measured the changes in the absorbance of the flowing toluene gas at 267.5 nm against time for the three types of cell. Here, we denote the absorbance for R, W, and WCT as AR, AW, and AWCT, respectively. We set the time at zero seconds (t ) 0) when we turned the heater switch on. The toluene gas was sampled for 5 min for each concentration cell. The gas concentration of the original gas (C0) was 30 ppm. We measured the background spectrum prior to each measurement, and the detector exposure time was set to UV light at 500 ms, which is the maximum value before the detector sensitivity saturates. We accumulated 10 sets of 500-ms exposure data. Thus, the total time required to measure each spectrum was 5 s. We therefore define A as the average absorption intensity in 5 s, from t - 2.5 to t + 2.5. The absorption change was measured at 5-s intervals. To study the changes in shorter time steps, the (9) Ginsburg, N.; Robertson, W. W.; Matsen, F. A. J. Chem. Phys. 1946, 14, 511-517.

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

1715

Figure 6. C0 dependence of AWCT/AW measured at C0 levels from 0.2 to 10.0 ppm. The background level was subtracted from each set of data. The original gas was concentrated for 30 min.

measurement was repeated, this time delaying the measurement’s starting time by 2.5 s, and the two results were overlaid. Thus, we calculated A every 2.5 s from t ) 0 to t ) 50. AW and AWCT have their maximum values at t ) 10, but AR has its maximum at t ) 15. AR provided a much broader response curve, and the intensity at the maximum value was the weakest of the three. This shows that a much sharper response can be obtained by changing the cell structure in addition to changing the amount of adsorbent and the gas flow rate. AW provided a slightly broader response with a weaker maximum intensity than AWCT. This shows that employing a CT channel with the concentration cell had the effect of raising the signal intensity. Without a CT channel, the gas in the channel downstream of the adsorbent expands with heat during thermal desorption. The flow rate during desorption was reduced to ∼0.3 × 10-1 and ∼0.6 × 10-1 cm3/s for W and WCT, respectively. This reduction of the flow rate with W may accelerate the dilution of the desorbed gas. Therefore, it is important to build the CT channel into the cell adjacent to the adsorbent section. We denote the width of the response for W and WCT as WW, and WWCT, respectively. Because the amount of sampled toluene gas was the same in both cases, we assume the areas of the peaks (A × W) to be the same. With W, a dilution of the desorbed gas broadens the response (WW > WWCT) and, thus, reduces the signal intensity (AW < AWCT). If the difference between WW and WWCT should become larger with the lower C0, the difference between AW and AWCT should become larger as C0 decreases. To test these hypotheses, we analyzed the C0 dependence of the broadening of the response (WW/WWCT). AW and AWCT were measured for C0 in the 0.05-10 ppm region; toluene gas was sampled for 30 min for each measurement. Figure 6 shows the C0 dependence of AWCT/ AW. The results indicate that AWCT/AW increased significantly with lower C0 and is, thus, consistent with our hypotheses. Therefore, we can conclude that the broadening of the response without a CT channel may be more serious in the lower C0 region. A lower detection limit can be expected using the concentration cell with WCT. Next, we measured the detection limit using the three types of CT cell. Figure 7 plots the integrated spectral area in the 230275 nm region for R, W, and WCT (AR′, AW′, and AWCT′) against C0 in the 0.05-10 ppm region. The sampling times for the measurement were 60 and 30 min for AR′ and AW′/AWCT′, respectively. We did not plot the data at C0 values lower than 2 1716 Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

Figure 7. AWCT′ (9), AW′ (b), and AR′ (2) from 0.05 ppm (background) to 10.0 ppm. The background level was subtracted from each set of data. The original gas was concentrated for 60 and 30 min for AR′ and AW′/AWCT′, respectively.

Figure 8. Absorption spectra of detected 10 ppm of (a) benzene, (b) toluene, and (c) o-xylene by using WCT.

and 0.5 ppm for AR′ and AW′, respectively, because they were the same as the background level (ABG′ ) 0.0200), as described later. AWCT′ provided the strongest intensity over the whole range, followed by AW′ and AR′. The difference between these intensities increased significantly in the lower C0 region for the reason we gave earlier. When the signal-to-noise ratio was 3, the detection limits were ∼0.05, ∼1, and ∼4 ppm, and A′ at that point was 0.1267, 0.0780, and 0.0626, for WCT, W, and R respectively. We estimated the accuracy of this method to be three times the standard deviation of A′, ∼20%. The amount of accumulated sampled gas, calculated by multiplying the flow rate by the sampling time, was ∼4.5 times larger with W and WCT than that with R; however, the W/WCT detection limit was >4.5 times lower than that of R. This shows that changing the cell structure improves the detection yield after thermal desorption. We concluded that by optimizing the channel structure, we can achieve a 4-fold reduction in the detection limit, but by

integrating a CT channel in the concentration cell, we can achieve a 20-fold reduction. As a result, we successfully realized an 80fold reduction in the detection limit, as compared with our previous device. Figure 8 shows the absorption spectra of other gases detected by using WCT. We measured 10 ppm of (a) benzene and (c) o-xylene gas and compared those results with that obtained for 10 ppm of toluene (b) gas. The measurement conditions were the same in all three cases, namely, a sampling time of 30 min at a flow rate of 1.1 × 10-1 cm3/s. We observed the specific peaks of each compound. The results indicate that it is possible to detect and identify other BTEX gases in the same way as toluene by using our microfluidic device, although we do not show the results for m- and p-xylene and ethylbenzene. To avoid any reduction in adsorbent capacity as a result of the adsorption of other gases, for example, water, aliphatic compounds, and polycyclic aromatic hydrocarbons, during sampling we can employ a multibed adsorption method10 that combines various adsorbents in the same sampling channel, which is normally used to adjust the adsorption and desorption capacity to the analytes of interest. Since we measured only the 230-275-nm region, other gases did not interfere with the detection of BTEX. (10) Helmig, D.; Greenberg, J. P. J. Chromatogr. A 1994, 677, 123-132. Oliver, K. D.; Adams, J. R., Jr.; Daughtrey, E. H.; McClenny, W. A.; Toong, M. J.; Pardee, M. A.; Almasi, E. B.; Kirshen, N. A. Environ. Sci. Technol. 1996, 30, 1939-1945.

CONCLUSION We proposed integrating an air-cooled CT channel in the concentration cell to improve the detection limit of our microfluidic device for airborne BTEX detection. We compared three types of concentration cells and concluded that when wider channel structures with W and WCT were used, the increased gas flow rates were ∼9 times greater than with the narrower structure of R, and this sharpened the response of the device and increased the signal intensity. We also found that the response broadening without the CT became more serious when measuring low gas concentrations. By optimizing the channel structure, a 4-fold reduction in the detection limit could be achieved, but by integrating the CT channel in the concentration cell, we a 20-fold reduction could be achieved. We found that the WCT type provides the strongest signal intensity. As a result, we succeeded in achieving an 80-fold reduction in the detection limit compared with our previous device and obtained a 0.05 ppm detection limit for toluene gas with a sampling time of 30 min at a flow rate of 1.1 × 10-1 cm3/s.

Received for review October 9, 2001. Accepted January 15, 2002. AC0110810

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

1717