Temperature-Programmable Resistively Heated Micromachined

Article pubs.acs.org/ac

Temperature-Programmable Resistively Heated Micromachined Gas Chromatography and Differential Mobility Spectrometry Detection for the Determination of Non-Sulfur Odorants in Natural Gas J. Luong,†,‡ R. Gras,‡ H. J. Cortes,†,§ and R. A. Shellie*,† †

Australian Centre for Research on Separation Science (ACROSS), University of Tasmania, Private Bag 75, Hobart, Tasmania 7001, Australia ‡ Dow Chemical Canada ULC, Highway 15, Fort Saskatchewan, Alberta T8L 2P4, Canada § HJ Cortes Consulting LLC, Midland, Michigan 48642, United States ABSTRACT: A portable, fast gas chromatographic method for the direct measurement of the parts per billion level of sulfur-free odorants in commercially available natural gas is introduced. The approach incorporates a resistively heated, temperature-programmable silicon micromachined gas chromatograph that employs a standard capillary column for the fast separation of methyl and ethyl acrylate from the natural gas matrix. The separation approach is coupled to a micromachined differential mobility detector to enhance analyte detectability, and the overall selectivity obtained against the matrix is described. A complete analysis can be conducted in less than 70 s. Furthermore, these two compounds can be measured accurately in the presence of other common volatile sulfur-based odorants such as alkyl mercaptans and alkyl sulfides. Repeatability of less than 3% RSD (n = 20) over a range from 0.5 to 5 ppm was obtained with a limit of detection for the target compounds at 50 ppb (v/v) and a linear range from 0.5 to 50 ppm with a correlation coefficient of at least 0.997.

N

ment. Methyl acrylate and ethyl acrylate both have good vapor pressure and therefore are suitable to be analyzed by gas chromatography (GC). However, without a column with good selectivity toward alkyl acrylates, substantial interferences are observed if single-dimensional GC is used. Other analytical techniques such as olfactometry, membrane separation with ion mobility spectrometry, and the direct reading Drager chemical measurement (CMS) chip have been reported with various degrees of success.8−12 Common constraints encountered with these techniques include a high cost of ownership, the requirement of highly skilled analysts, unreliable results, and chromatographic matrix interferences. A portable, fast gas chromatographic approach based on a resistively heated, temperature-programmable silicon micromachined gas chromatograph that employs a standard capillary column and differential mobility spectrometry (DMS) detection for the measurement of sulfur-free odorants is introduced here. Despite some similarities to ion mobility

atural gas is an important fossil fuel with many uses in power generation in residential, commercial, industrial, and transportation spaces. Major constituents of commercially available natural gas are methane (90−95% or higher), ethane, propane, isobutane, n-butane, isopentane, and n-pentane, with other trace components such as hexanes, heptanes, carbon dioxide, oxygen, nitrogen, and water.1−3 Natural gas is colorless, odorless, and highly flammable, so safe use requires odorization with approved odorants to allow early detection in the event of a loss of containment. Sulfur-containing compounds such as alkyl mercaptans, alkyl sulfides, and cyclic sulfides are typically employed for this purpose.4−7 There are operational and environmental issues concerning use of sulfur-based odorants, including undesirable corrosion of pipelines and pumping station equipment, overdosing resulting in false leak alarms, and the generation of additional pollution to the environment in the form of acid gas (sulfur dioxide) when the odorized natural gas is consumed by end-users.8 Following the lead of European countries, Canadian natural gas distributors are contemplating the replacement of sulfur odorants with non-sulfur-based odorants such as methyl acrylate (CAS 96-33-3) and ethyl acrylate (CAS 140-88-5) in an effort to reduce harmful sulfur emissions to the environ© 2013 American Chemical Society

Received: January 8, 2013 Accepted: February 21, 2013 Published: February 21, 2013 3369

dx.doi.org/10.1021/ac400067j | Anal. Chem. 2013, 85, 3369−3373

Analytical Chemistry

Article

Figure 1. Topographic chart illustrating the impact of the detector temperature on the reactant ion product over a range of temperature from 100 to 150 °C. RIP = reactant ion product, PI = product ion, High = high sensitivity, and Low = low sensitivity.

spectrometry (IMS), DMS detection offers some significant advantages such as the incorporation of nonlinear mobility dependence in high radio frequency electric fields for ion filtering. In addition, DMS does not require ion pulses for operation; hence, the resolution is not dictated by the pulse width. Also, in DMS, the ions are introduced continuously into the ion filter, and all the tuned ions are passed through the filter. This offers improved sensitivity over the already sensitive IMS. A detailed description of DMS technology may be found elsewhere.13−17 In the present investigation, DMS is coupled to temperatureprogrammed micromachined GC that employs a standard capillary column for chromatographic separation. Lack of temperature-programming capability has been one of the major gaps encountered with commercially available silicon micromachined GC.18,19 Temperature-programming capability is critically needed to extend the range of applicability for matrixes having a wide boiling point range, such as natural gas. Combining resistively heated temperature-programming capability with micromachined GC is a logical approach since the technology offers low power consumption. Low power consumption is an important criterion for portable or transportable equipment, especially for successful field operation in below freezing point ambient temperature environments. In this application, temperature programming consumes less than 25 W. Resistively heated chromatographic systems also afford a respectable cooling time. For instance, cooling from 150 to 50 °C with the column assembly used requires a mere 25 s. This can substantially improve sample throughput and shorten the wait time for results to become

available in the event of a gas leak in either industrial or residential environments. Successful development and implementation to provide reliable and direct measurement of a range of non-sulfur odorants at the low part s per billion level without sample enrichment in less than 70 s is discussed.

■

EXPERIMENTAL SECTION

An Agilent CP-4900 micromachined gas chromatograph (Agilent Technologies, Middelburg, The Netherlands) was used as the analytical platform. The original isothermal column module was removed and replaced with a custom-fabricated Valco VICI resistively heated, temperature-programmable column module, equipped with a 10 m × 0.25 mm i.d. fused silica column coated with 1 μm of 100% poly(dimethylsiloxane) stationary phase. The column was interfaced with a micromachined silicon injector (Agilent Technologies) via the existing inlet interface manifold using Viton O-rings and compression fittings similar to those used for the original micromachined isothermal temperature module. The inlet temperature was 120 °C, and an injection time of 200 s was used throughout. Helium carrier gas was supplied at an inlet pressure of 10 psig. The outlet of the analytical column was connected to a custom, locally fabricated Siltek deactivated metal manifold interfaced to a micromachined DMS detector (Sionex, Bedford, MA) via heat-traced 25 cm × 0.25 mm i.d. deactivated, uncoated fused silica tubing with Viton O-rings and compression fittings similar to those used to interface the column inlet to the injector. Details of the development and configuration of the silicon micromachined gas chromatograph 3370

dx.doi.org/10.1021/ac400067j | Anal. Chem. 2013, 85, 3369−3373

Analytical Chemistry

Article

offers better sensitivity, a higher asymmetric field can help suppress interfering ion cluster formation. This, in turn, can provide enhanced selectivity for the target compounds and in some cases a better overall signal-to-noise ratio. An rf value of 1100 V with Vc values of −6 V for methyl acrylate and −3.2 V for ethyl acrylate in the positive channel were selected for the measurement of these two compounds of interest in the matrix mentioned. For the detection of methyl acrylate and ethyl acrylate, an rf voltage of 1100 V in the positive channel was applied. The following equations describe a simplified scheme for product ion formation for positive ion products for methyl acrylate and ethyl acrylate.

are described elsewhere.20 Temperature programming of the column unit was conducted using an external programming controller (Valco VICI, Houston, TX). The temperature was programmed from 50 °C (20 s) to 150 °C at 3.5 °C/s and maintained at 150 °C for 15 s. The on-board pump sampling duration was 30 s, and the transport gas was filtered, CO2-free air with a flow rate of 500 mL/min. Detector control and data handling capability were handled by Sionex Expert software version 2.15 (Sionex, Bedford, MA). The differential mobility detector temperature was set at 120 °C. Three-dimensional data for method optimization and twodimensional data for quantitative analysis were handled by Expert version 2.15, integrated with Varian Maitre Elite chromatographic data system version 3.2 (Varian, Middelburg, The Netherlands). For the detection of methyl acrylate and ethyl acrylate, a rf voltage of 1100 V in the positive channel was applied. Carrier and transport gases such as helium and air used for system performance studies were acquired from Air Liquide (Edmonton, Canada), while methyl acrylate and ethyl acrylate, both of ACS grade (99.5%), were obtained from Aldrich Canada (Oakville, Canada). Certified 100 ppm (v/v) methyl acrylate and ethyl acrylate standards in methane were acquired from Air Liquide. Secondary standards with methane as the diluent over the range from 100 ppb to 50 ppm (v/v) were prepared from the primary standards by serial dilution for calibration purposes with an Environics 2014 computerized gas dilution and blending system (Environics,Tolland, CT). A typical approach for gas sampling is to obtain a representative sample with a Siltek-treated 500 mL stainless steel sampling cylinder. The sample pressure is reduced to less than 10 psig with a Porter single-stage pressure regulator prior to sampling to prevent the potential of overpressurizing the micromachined injector. An alternative to the use of a passivated sampling cylinder in sampling involves the use of a new Tedlar sampling bag. For the application described, if a Tedlar bag is employed, the bag is not recycled. Once used, the bag is disposed of to prevent cross-contamination or absorption.

methyl acrylate: C4 H6O2 + H+(H 2O) → C4 H 7O2+ + H 2O

ethyl acrylate: C5H8O2 + H+(H 2O) → C5H 9O2+ + H 2O

Dimerization can also occur. methyl acrylate: C4 H 7O2+ + C4 H6O2 → C8H13O2+

ethyl acetate: C5H 9O2+ + C5H8O2 → C10H17O2+

Figure 2 shows a topographic chart of the two analytes of interest in a nonodorized natural gas matrix with methyl

■

RESULTS AND DISCUSSION Optimization of detector performance involved first selecting an appropriate detector temperature. With the type of polymeric material used as detector gaskets, the maximum operating temperature for the detector is restricted to 150 °C. Figure 1 shows a topographic chart illustrating the impact of the detector temperature on the reactant ion product (RIP) over a temperature range of 100−150 °C. The sensitivity of the differential mobility spectrometer is typically inversely proportional to the temperature applied. For the application described, a temperature of 120 °C was selected to prevent condensation of solutes with the high boiling point in the matrix, in this case, hexanes, heptanes, and water. It is advantageous to operate the detector at high flow rates when high sensitivity is required. A transport gas flow rate of 500 mL/min was chosen for the present application. The detector was further optimized for an appropriate rf voltage and its corresponding compensation voltage (Vc) to deliver the best selectivity and sensitivity for methyl acrylate and ethyl acrylate. A detailed discussion on the strategy in detector optimization and the dispersion plot has been reported previously.17,20,21 In general, the rf voltage is inversely proportional to the RIP intensity. While a lower rf voltage

Figure 2. Topographic chart of methyl acrylate and ethyl acrylate in a nonodorized natural gas matrix.

acrylate and ethyl acrylate having retention times of 37 and 46 s. Under the conditions used, the product ions formed were well separated from the RIP and free from chromatographic and spectral interferences. A nonpolar wall-coated open tubular capillary column was utilized to help improve the overall system selectivity by providing a coarse separation for methyl acrylate and ethyl acrylate from other potential interfering compounds that might exist in the natural gas matrix. The rather short temperatureprogramming cycle of 50−150 °C affords the removal of any potential higher boiling point compounds in the matrix, such as the C5−7 hydrocarbons that might have accumulated on the column in an expedient manner. This approach substantially speeds up the analysis when compared to conventional 3371

dx.doi.org/10.1021/ac400067j | Anal. Chem. 2013, 85, 3369−3373

Analytical Chemistry

Article

Figure 3. Chromatogram of a commercially available, nonodorized natural gas sample with the technique described.

Figure 4. Chromatogram of 100 ppb (v/v) each of methyl and ethyl acrylate in commercial natural gas.

in new, unused Tedlar bags and up to at least two weeks in Siltek deactivated sampling cylinders. With this technique, methyl acrylate and ethyl acrylate can be selectively and accurately measured without preconcentration in less than 70 s. A repeatability study was conducted using standards of methyl and ethyl acrylate at two different concentrations, 500 ppb (v/v) and 5 ppm (v/v), in commercially available nonodorized natural gas. Relative standard deviations of 2.9% and 2.1% at the 95% confidence level were determined for the 500 ppb (v/v) and 5 ppm (v/v) levels, respectively (n = 20). GC−DMS was shown to have a detection limit of 50 ppb (v/v) under the conditions used and was found to be linear over a range of up to 50 ppm (v/v)with the linearization software feature engagedwith a correlation coefficient of greater than

isothermal silicon micromachined GC. Figure 3 shows a chromatogram of a commercially available, nonodorized natural gas sample that contains 96% methane, percentage levels of byproducts such as ethane, propane, isobutane, butane, isopentane, pentane, hexanes, and heptanes, and ppm levels of oxygen, nitrogen, carbon monoxide, carbon dioxide, and water. The high degree of selectivity of the detector toward the target compounds under the conditions used was highlighted by the absence of a response of the detector to any of the above-mentioned interfering components. Figure 4 shows a chromatogram of 100 ppb (v/v) each of methyl and ethyl acrylate in natural gas. A storage study conducted showed that, at a concentration of 1 ppm (v/v) with a recovery of greater than 95%, methyl and ethyl acrylate are stable for up to 5 days 3372

dx.doi.org/10.1021/ac400067j | Anal. Chem. 2013, 85, 3369−3373

Analytical Chemistry

■

0.997. The attained linear range and detection limit meet the analytical method requirements as the dosage range for methyl acrylate and ethyl acrylate in natural gas is typically from 1 to 25 ppm (v/v).11,12 In the event of an overdosage, where the concentration of the analytes exceeds 50 ppm (v/v), sample dilution is recommended. Although not validated for the measurement of sulfur odorants, GC−DMS also has the potential for use in measuring common sulfur odorants in commercially available natural gas such as mercaptans, sulfides, and disulfides (Figure 5). The possibility for false positives

Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +61-3-6226-7656. Fax: +61-3-6226-2858. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Dr. HuaMin Cai, of Valco VICI, Houston, TX, and Dr. Jos Curvers, of Bruker Chemical Analysis BV, Middelburg, The Netherlands, are acknowledged for their invaluable assistance in prototyping the analytical platform. Dr. Ranaan Miller, of MIT, Cambridge, MA, and Professor Dr. Erkinjon Nazarov, of Draper Laboratory, Tampa, FL, are acknowledged for their technical guidance in differential mobility detection. Special thanks are due to Dave Walter, Dan Martin, Vicki Carter, and Andy Szigety, of the Analytical Technology Center, The Dow Chemical Co., for their encouragement and support. Robert Shellie is the recipient of an Australian Research Council Australian Research Fellowship (Project DP110104923).

■

REFERENCES

(1) Tenkrat, D.; Hlincik, T.; Prokes, O. In Natural Gas; Potocnik, P., Ed.; Sciyo/InTech: Rijeka, Croatia, 2010; pp 25−89. (2) Usher, M. Odor FadePossible Causes and Remedies; Elf Atochem North America, Inc.: Philadelphia, PA, 1999. (3) Fazzalari, F. A., Ed. Compilation of Odor and Taste Threshold Data; ASTM Data Series DS 48A; ASTM International: West Conshohocken, PA, 1978; pp 1−508. (4) Wehnert, P. Proc. Int. Sch. Hydrocarbon Meas. 2002, 747−750. (5) Litvinov, A.; Unchenko, P.; Nikolaev, I. Meas. Tech. 2007, 50, 548−550. (6) Firor, R.; Quimby, B. Hydrocarbon Process. 2003, 82, 79−81. (7) Kim, K. Sensors 2012, 11, 1405−1417. (8) Ruzsanyi, V.; Sielemann, S.; Baumbach, J. I. J. Environ. Monit. 2007, 9, 61−65. (9) Bernhart, M.; Reimert, R. gwf-Gas/Erdgas 1999, 140, 92−96. (10) Schunk, C.; Bernhart, M.; Reimert, R. gwf-Gas/Erdgas 1999, 140, 716−721. (11) Kröger, K.; Bernhart, M.; Reimert, R. gwf-Gas/Erdgas 2001, 142, 779−784. (12) Schmeer, F.; Reimert, R. gwf-Gas/Erdgas 2003, 144, 52−58. (13) Miller, R.; Eiceman, G.; Nazarov, E.; King, T. A MEMS Radio Frequency on Mobility Spectrometer for Chemical Agent Detection. Solid State Sensor and Actuator Workshop, Hilton Head Island, SC; The Foundation: Cleveland Heights, OH, 2000. (14) Luong, J.; Gras, R.; Van Meulebroeck, R.; Sutherland, F.; Cortes, H. J. Chromatogr. Sci. 2006, 44, 276. (15) Lambertus, G.; Fix, C.; Reidy, S.; Wheeler, M.; Miller, R.; Narazov, E.; Sacks, R. Anal. Chem. 2005, 77, 7563. (16) Eiceman, G.; Karpas, A. Ion Mobility Spectrometry; CRC Press: Boca Raton, FL, 1994. (17) Eiceman, G. Trends Anal. Chem 2002, 21, 259. (18) Luong, J.; Cai, H.; Gras, R.; Curvers, J. J. Chromatogr. Sci. 2012, 50, 245. (19) Terry, S. a Gas Chromatographic Air Analyzer Fabricated on Silicon Wafer Using Integrated Circuit Technology. Ph.D. Dissertation, Standford University, Stanford, CA, 1975. (20) Luong, J.; Nazarov, E.; Gras, R.; Shellie, R. A.; Cortes, H. Int. J. Ion Mobility Spectrom. 2012, 15, 179−187. (21) Curvers, J.; Van Schaik, H. Am. Lab. 2004, 36, 18.

Figure 5. Topographic chart of methyl acrylate, ethyl acrylate, and other sulfur-based compounds used as odorants in commercially available natural gas: 1, ethyl mercaptan; 2, propyl mercaptan; 3, methyl acrylate; 4, ethyl acrylate; 5, dimethyl sulfide; 6, methyl ethyl sulfide; 7, tetrahydrothiophene.

caused by chromatographic interferences is substantially minimized when compared to single-dimensional gas chromatography with common detectors such as micromachined thermal conductivity detection, flame ionization, photoionization, or electron capture owing to the synergy derived from temporal separation rendered by the chromatographic process and spatial separation with tunable detectability provided by spectrometric detection.

■

CONCLUSIONS A fast GC method based on temperature-programmable silicon micromachined technology that employs standard capillary column technology for separation in combination with differential mobility detection for the direct measurement of parts per billion levels of nonsulfur odorants such as methyl acrylate and ethyl acrylate in natural gas has been successfully developed and implemented. In the work described the test samples taken at the field were brought back to the laboratory for analysis; however, the system can be forward deployed for field analysis if required. A complete analysis can be conducted in less than 70 s with a relative precision of less than 5% over a range from 500 ppb to 50 ppm (v/v) with a linear range from 0.5 to 50 ppm (v/v) and a correlation coefficient of greater than 0.997. GC−DMS was found to be reliable with a well-purified transport gas, can be field deployable, and is suitable for use in monitoring said compounds in blending facilities, for fugitive emissions, or for residential (end-users) leak monitoring purposes. As well, the technique can also be used for the monitoring of common sulfur odorants in natural gas such as volatile mercaptans, sulfides, and disulfides. 3373

dx.doi.org/10.1021/ac400067j | Anal. Chem. 2013, 85, 3369−3373