Letter pubs.acs.org/ac
Highly Selective Zeolite Membranes as Explosive Preconcentrators Jie Zhao,†,‡ Ting Luo,† Xiangwen Zhang,‡ Yu Lei,§ Ke Gong,⊥ and Yushan Yan*,†,⊥ †
Department of Chemical and Environmental Engineering, University of California, Riverside, California 92521, United States Key Laboratory for Green Chemical Technology of Ministry of Education, School of Chemical Engineering & Technology, Tianjin University, Tianjin 300072, China § Department of Chemical, Materials and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut 06269, United States ⊥ Department of Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware 19716, United States ‡
S Supporting Information *
ABSTRACT: Highly selective thin zeolite MFI membranes are synthesized on porous stainless steel and α-alumina supports using a seeded growth method. An ultraviolet (UV) light treatment is employed as a low temperature alternative to remove the organic structure-directing agent (SDA) to avoid membrane cracking. The feasibility of the use of the MFI membranes as an explosive preconcentrator is examined by measuring the permeation of nitrogen (N2, an air surrogate) and 1,3,5-trimethylbenzene (TMB) (a 2,4,6-trinitrotoluene (TNT) surrogate) in a mixture of N2 and TMB. High N2/TMB selectivity (>10 000) and reasonable N2 flux (13.5 mmol/m2·s) are observed. On the basis of the flux, a hollow fiber array based preconcentrator is proposed and estimated to provide 1000× concentration within about 1 min using a hollow fiber with a 50 μm internal radius. This high performance explosive preconcentrator may open a new venue for the detection of subppb or lower level of explosives simply in conjunction with conventional explosives detectors.
C
Zeolite membranes have been developed for use in molecular sieving gas separation applications.7−11 Zeolite MFI is considered as a promising membrane material for concentrating explosives because its pore size (0.51−0.56 nm) is smaller than explosives molecules (e.g., TNT has a kinetic diameter of 0.85 nm12) but much larger than air components, N2 (0.38 nm), O2 (0.35 nm), and CO2 (0.39 nm).13 An ultrathin molecular sieving zeolite membrane may allow a large flux of air while denying the passage of explosives, leading to an efficient concentration of the explosives in a short time and consequently a (near) real time detection of explosives at ultralow concentrations (subppb or lower). The aim of this work is to synthesize continuous, crack-free, ultrathin molecular sieving zeolite MFI membranes on porous supports and test their use as explosive preconcentrators. The thin zeolite films are designed to offer the high molecular sieving selectivity while the inorganic supports provide the required mechanical integrity. The recently available ultraviolet (UV) treatment at ambient temperature14,15 was used for the structuredirecting agent (SDA) removal16−21 to avoid cracks that usually accompanied the conventional high temperature calcination.22,23 The membranes are characterized in terms of their flux and selectivity for explosive molecules and air. For safety reasons,
onventional explosives such as 2,4,6-trinitrotoluene (TNT) are still the most commonly used in terrorism attacks, and they have killed and injured a lot more people than biological, chemical, or radioactive substances that are ironically often on the forefront of the public’s mind.1 Conventional explosives are also widely used in antipersonnel mines that are estimated to be numbered at 110 million in over 70 countries. About 15 000 to 20 000 people, mostly civilians, and often children, are killed or maimed by landmines each year. Clearly, there is a pressing need for rapid and sensitive detection of these conventional explosives. The critical problem for detection of these conventional explosives by the current sensor devices is their low to ultralow concentrations under ambient conditions. For instance, at 25 °C, the concentration of 2,4,6-trinitrotoluene (TNT) vapor at saturation is 10 ppb, and the saturation concentration of cyclotrimethylenetrinitramine (RDX) is less than 1 ppb. Furthermore, in a large open air environment such as airports, train stations, and minefields, the concentrations can be diluted from ppb level to ppt level or even lower,2−6 and as a result, even the best-engineered sensors still have a challenging time detecting these explosives. Instead of exhaustively improving the intrinsic sensitivity of a sensor, a possibly simpler and more effective solution is to design a highly efficient explosive preconcentrator that can extend the detection limit of a sensor down by three or more orders of magnitude. © 2012 American Chemical Society
Received: May 18, 2012 Accepted: July 18, 2012 Published: July 18, 2012 6303
dx.doi.org/10.1021/ac301359j | Anal. Chem. 2012, 84, 6303−6307
Analytical Chemistry
Letter
Figure 1. Top and cross-sectional view SEM micrographs of zeolite MFI membrane on porous stainless steel (SS) and α-alumina (Al) supports. (a) SS support, (b) Al support, (c) as-synthesized membrane on SS, (d) as-synthesized membrane on Al, (e) calcined membrane on SS, (f) calcined membrane on Al, (g, i) UV treated membrane on SS, and (h, j) UV treated membrane on Al.
Both the stainless steel (SS) (Figure 1a) and the α-alumina (Al) (Figure 1b) supports showed packed particle morphology. Uniform and continuous MFI zeolite membranes were formed on both the SS (Figure 1c) and Al (Figure 1d) supports with no
trimethylbenzene (TMB) was chosen to represent TNT. The N2/TMB selectivity obtained in this study is expected to be the lower bound for N2/TNT selectivity because TNT molecules (0.85 nm12) are much larger than TMB (0.76 nm24). 6304
dx.doi.org/10.1021/ac301359j | Anal. Chem. 2012, 84, 6303−6307
Analytical Chemistry
Letter
Table 1. Gas Permeance and Selectivitya single gas permeation no.
samples
a b c d e
SS Al MFI-SS MSM MFI-Al MSM MFI-SS-C MSM MFI−Al-C MSM MFI-SS-UV MS MFI-SS-UV MSM MFI−Al-UV MS MFI−Al-UV MSM
f g h i j
mixture permeation
N2 flux, mmol/(m2·s)
N2 permeance, mol/(m2·s·Pa)
N2 flux, mmol/(m2·s)
N2 permeance, mol/(m2·s·Pa)
N2/TMB selectivity
N2 flux decrease from single to mixture permeation, %
91.6 144.0 0 0 11.20
4.1 × 10−6 9.9 × 10−6 0 0 1.8 × 10−7
82.7 165.0 0 0 6.2
3.4 × 10−6 1.3 × 10−5 0 0 7.6 × 10−8
0.8 0.6 N/A N/A 0.5
9.7 −14.6 N/A N/A 44.6
24.0
4.5 × 10−7
8.1
1.1 × 10−7
0.4
66.3
3.6
4.1 × 10−8
3.3
3.7 × 10−8
3330
8.6
8.5
1.1 × 10−7
6.6
8.3 × 10−8
6160
21.7
20.0
3.4 × 10−7
11.6
1.7 × 10−7
9510
41.7
20.6
3.6 × 10−7
13.5
2.0 × 10−7
10600
34.4
Notes: SS: porous stainless steel; Al: α-alumina; MFI: MFI film on support; C: calcination; UV: UV treatment; MS: membrane-support configuration with one zeolite membrane on the feed side; MSM: membrane-support-membrane with zeolite membranes on both sides of the support. a
on both sides of support (MSM configuration) have the best separation performance with a N2 flux up to 13.5 mmol/m2·s and a N2/TMB selectivity greater than 10 000. It is interesting to note that the MSM configuration offers higher flux than the MS (zeolite membrane only on the feed side of the membrane) one even though the permeating gases have to pass through two layers of zeolite membranes in the former case and only one layer in the latter case. This observation is consistent with the modeling results by Gardner et al.26 Briefly, the addition of the zeolite film on the permeate side prevents the sweeping gas (i.e., He) from entering the support pores, and this allows the permeating gas (i.e., N2) to avoid the slow diffusion through a thick (i.e., 1 mm) He film. Upon switching the feed from pure nitrogen to binary mixture, the N2 flux decreases by about 10− 20% for porous SS supported samples and about 40% for Al supported samples. Decreases in flux are due to a multicomponent feed, possible preferential adsorption of TMB, and smaller partial pressure differences. The separation differences between the two supports can be attributed to differences in support characteristics, such as the pore diameter, pore interconnectivity, and the porosity. Although the N2 flux drop on Al supported sample is much larger than that of a SS supported sample, the N2 flux and the N2/TMB selectivity of the Al membranes are still much higher than that of the SS supported samples. The key to the design of a membrane based preconcentrator is to maximize the ratio of the membrane area to concentrator volume, so that the time needed to reach a targeted concentration is minimized. It is also preferable to use a concentrator geometry that is commercially available. With these factors considered, a hollow fiber based preconcentrator (Figure 2) is proposed. Estimates of the time needed to reach a targeted level of concentration of TNT are also provided for different hollow fiber inside radii (Figure 3). It is feasible to reach 1000× concentration within a minute using a 50 μm radius hollow fiber. Although the estimate is carried out for a single fiber, the actual device could consist of multiple fibers with the same sensor or with different sensors to drastically improve the sensor’s ability to recognize a complex mixture. With the fast concentration speed, explosives at ultralow concentrations may be facilely detected even with conventional
noticeable differences in crystal morphology. Conventional high temperature calcination led to cracked membranes (Figure 1e,f), while UV treatment produced crack-free membranes (Figure 1g,h). Cross-sectional views (Figure 1i,j) show continuous coriented polycrystalline films with a thickness of about 3.5 and 4 μm for the SS and Al support, respectively. For the synthesis of zeolite MFI membranes, organic SDA molecules (i.e., tetrapropylammonium hydroxide) are used and these molecules are trapped inside the zeolite pores, blocking any transport through the zeolite pores. To remove these SDA molecules, high temperature calcination is commonly used, but this method often leads to cracks of the zeolite membranes (Figure 1e,f) due to differences in the thermal expansion coefficient between the support and zeolite film.14,25 In contrast, there is no crack formation on the UV-treated membranes (Figure 1g,h) since the UV treatment of the MFI membranes was performed at ambient temperature. Different from the previously reported results19 where a 4 h UV irradiation was sufficient for a 440 nm thick MFI zeolite membrane, a much longer UV exposure was used in this study due to thicker membranes. For example, the single-sided membrane on either SS or Al support required a 72 h UV treatment to remove the organic SDA from the 4 μm thick zeolite membranes. The SS and Al supports (samples a and b in Table 1) offer high N2 flux but no N2/TMB selectivity because of their large pore size (1.8 μm for Al and 0.2 μm for SS). Once zeolite MFI films are formed on the surface, the composite membranes become nonpermeable because of the occluded SDA molecules (samples c and d). After high temperature calcination (samples e and f), the N2 fluxes of the membranes are about 10−20% of the bare supports, but the N2/TMB selectivity is nonexistent. The significant reduction of flux is consistent with the expectation that it is much harder to permeate through the zeolite film than the support because zeolites have much smaller pores. The poor N2/TMB selectivity is also expected because of the formation of cracks during the calcination process. Compared with the high temperature calcination method, the UV treatment reduces the flux by 10−20% but improves the selectivity (e.g., >10 000), simply because no cracks were formed during the template removal process. The Al-supported membranes with zeolite films 6305
dx.doi.org/10.1021/ac301359j | Anal. Chem. 2012, 84, 6303−6307
Analytical Chemistry
Letter
loop and related equations.This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The support for this work was provided by the US National Science Foundation (CMMI 0730826) and the China Scholarship Council. The authors would like to thank Shuang Gu, Qianrong Fang, Rui Cai, Laj Xiong, and Kurt Jensen for their help.
■
Figure 2. Schematic of a hollow fiber based preconcentrator.
Figure 3. Estimate of the concentration factor ( f) versus time (t) for different hollow fiber internal radii (e.g., 75, 50, and 25 μm).
explosives sensors, opening up venues in trace explosives detection for homeland security. Furthermore, many improvements are available to increase the N2 flux by reducing the membrane thickness or the separation layer thickness.27 Further work is also ongoing to test the effects of moisture on the N2 flux and selectivity of the zeolite membranes.
■
REFERENCES
(1) Sanchez, C.; Carlsson, H.; Colmsjö, A.; Crescenzi, C.; Batlle, R. Anal. Chem. 2003, 75, 4639−4645. (2) Davis, M. E. Science 2003, 300, 438−439. (3) Lai, Z.; Tsapatsis, M. Ind. Eng. Chem. Res. 2004, 43, 3000−3007. (4) Lai, Z.; Bonilla, G.; Diaz, I.; Nery, J. G.; Sujaoti, K.; Amat, M. A.; Kokkoli, E.; Terasaki, O.; Thompson, R. W.; Tsapatsis, M.; Vlachos, D. G. Science 2003, 18, 456−460. (5) Xomeritakis, G.; Lai, Z.; Tsapatsis, M. Ind. Eng. Chem. Res. 2001, 40, 544−552. (6) Tavolaro, A.; Drioli, E. Adv. Mater. 1999, 11, 975−996. (7) Xiao, W.; Yang, J.; Lu, J.; Wang, J. Sep. Purif. Technol. 2009, 67, 58− 63. (8) Zhao, Q.; Wang, J.; Chu, N.; Yin, X.; Yang, J.; Kong, C.; Wang, A.; Lu, J. J. Membr. Sci. 2008, 320, 303−309. (9) Hasegawa, Y.; Kimura, K.; Nemoto, Y.; Nagase, T.; Kiyozumi, Y.; Nishide, T.; Mizukami, F. Sep. Purif. Technol. 2008, 58, 386−392. (10) Kanezashi, M.; O’Brien, J.; Lin, Y. S. J. Membr. Sci. 2006, 286, 213−222. (11) Hasegawa, Y.; Ikeda, T.; Nagase, T.; Kiyozumi, Y.; Hanaoka, T.; Mizukami, F. J. Membr. Sci. 2006, 280, 397−405. (12) Greathouse, J. A.; Ockwig, N. W.; Criscenti, L. J.; Guilinger, T.; Pohl, P.; Allendorf, M. D. Phys. Chem. Chem. Phys. 2010, 12, 12621− 12629. (13) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The properties of Gases and Liquids; McGraw-Hill: New York, 1987; p 741. (14) Heng, S.; Lau, P. P. S.; Yeung, K. L.; Djafer, M.; Schrotter, J. C. J. Membr. Sci. 2004, 243, 69−78. (15) Motuzas, J.; Heng, S.; Ze Lau, P. P. S.; Yeung, K. L.; Beresnevicius, Z. J.; Julbe, A. Microporous Mesoporous Mater. 2007, 99, 197−205. (16) Kiricsi, I.; Fudala, Á .; Kónya, Z.; Hernádi, K.; Lentz, P.; Nagy, J. B. Appl. Catal., A: Gen. 2000, 203, L1−L4. (17) Buchel, G.; Denoyel, R.; Llewellyn, P. L.; Rouquerol, J. J. Mater. Chem. 2001, 11, 589−593. (18) Parikh, A. N.; Navrotsky, A.; Li, Q.; Yee, C. K.; Amweg, M. L.; Corma, A. Microporous Mesoporous Mater. 2004, 76, 17−22. (19) Li, Q.; Amweg, M. L.; Yee, C. K.; Navrotsky, A.; Parikh, A. N. Microporous Mesoporous Mater. 2005, 87, 45−51. (20) Xiao, L.; Li, J.; Jin, H.; Xu, R. Microporous Mesoporous Mater. 2006, 96, 413−418. (21) Zhang, Y.; Tokay, B.; Funke, H. H.; Falconer, J. L.; Noble, R. D. J. Membr. Sci. 2010, 363, 29−35. (22) Pachtova, O.; Kocirik, M.; Zikanova, A.; Bernauer, B.; Miachon, S.; Dalmon, J. Microporous Mesoporous Mater. 2002, 55, 285−296. (23) Dong, J.; Lin, Y. S.; Hu, M. Z. C.; Peascoe, R. A.; Payzant, E. A. Microporous Mesoporous Mater. 2000, 34, 241−253. (24) Chauvin, B.; Fajula, F.; Figueras, F.; Gueguen, C.; Bousquet, J. J. Catal. 1988, 111, 94−105. (25) Geus, E.; Van Bekkum, H. Zeolites 1995, 15, 333−341. (26) Gardner, T.; Martinek, J.; Falconer, J.; Noble, R. J. Membr. Sci. 2007, 304, 112−117.
ASSOCIATED CONTENT
S Supporting Information *
Experimental method; XRD patterns of MFI zeolite membrane; schematic of the apparatus for permeation measurements; calibration curves of N2 and TMP partial volumes in the sample 6306
dx.doi.org/10.1021/ac301359j | Anal. Chem. 2012, 84, 6303−6307
Analytical Chemistry
Letter
(27) Yan, Y. S.; Davis, M. E.; Gavalas, G. R. J. Membr. Sci. 1997, 126, 53−65.
6307
dx.doi.org/10.1021/ac301359j | Anal. Chem. 2012, 84, 6303−6307
