- Email: [email protected]
Multidimensional gas chromatography for the characterization of permanent gases and light hydrocarbons in catalytic cracking process
Journal of Chromatography A, 1271 (2013) 185–191
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Multidimensional gas chromatography for the characterization of permanent gases and light hydrocarbons in catalytic cracking process J. Luong a,b , R. Gras b , H.J. Cortes a,c , R.A. Shellie a,∗ a b c
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, MI 48642, USA
a r t i c l e
i n f o
Article history: Received 3 September 2012 Received in revised form 12 November 2012 Accepted 12 November 2012 Available online 19 November 2012 Keywords: Gas chromatography MDGC Planar microfluidic device Permanent gases Light hydrocarbons
a b s t r a c t An integrated gas chromatographic system has been successfully developed and implemented for the measurement of oxygen, nitrogen, carbon monoxide, carbon dioxide and light hydrocarbons in one single analysis. These analytes are frequently encountered in critical industrial petrochemical and chemical processes like catalytic cracking of naphtha or diesel fuel to lighter components used in gasoline. The system employs a practical, effective configuration consisting of two three-port planar microfluidic devices in series with each other, having built-in fluidic gates, and a mid-point pressure source. The use of planar microfluidic devices offers intangible advantages like in-oven switching with no mechanical moving parts, an inert sample flow path, and a leak-free operation even with multiple thermal cycles. In this way, necessary features such as selectivity enhancement, column isolation, column back-flushing, and improved system cleanliness were realized. Porous layer open tubular capillary columns were employed for the separation of hydrocarbons followed by flame ionization detection. After separation has occurred, carbon monoxide and carbon dioxide were converted to methane with the use of a nickel-based methanizer for detection with flame ionization. Flow modulated thermal conductivity detection was employed to measure oxygen and nitrogen. Separation of all the target analytes was achieved in one single analysis of less than 12 min. Reproducibility of retention times for all compounds were found to be less than 0.1% (n = 20). Reproducibility of area counts at two levels, namely 100 ppmv and 1000 ppmv over a period of two days were found to be less than 5.5% (n = 20). Oxygen and nitrogen were found to be linear over a range from 20 ppmv to 10,000 ppmv with correlation coefficients of at least 0.998 and detection limits of less than 10 ppmv . Hydrocarbons of interest were found to be linear over a range from 200 ppbv to 1000 ppmv with correlation coefficients of greater than 0.999 and detection limits of less than 100 ppbv . © 2012 Elsevier B.V. All rights reserved.
1. Introduction The characterization of oxygen, nitrogen, carbon monoxide, carbon dioxide (permanent gases) and light hydrocarbons is an important analysis for petrochemical and chemical industries [1–5]. These compounds are often encountered in various critical chemical processes like catalytic cracking for example [6–8]. Gas chromatography is a technique of choice due to the high degree of volatility of the compounds involved. Unfortunately, without the employment of cryogenic chromatography, no single chromatographic stationary phase is capable of adequately separating all of these analytes [9]. For instance, on a porous polymer based
∗ Corresponding author. Tel.: +61 3 6226 7656; fax: +61 3 6226 2858. E-mail address: [email protected] (R.A. Shellie). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.11.025
stationary phase like divinyl benzene, oxygen, nitrogen, and carbon monoxide are perfectly co-eluted. Hydrogen, oxygen, nitrogen, methane, and carbon monoxide can be well separated using a molecular sieve stationary phase; however carbon dioxide and heavier hydrocarbons are irreversibly adsorbed. The presence of water in the sample as vapor can also deactivate the stationary phase. Carbon molecular sieve or carbon black has been used to separate these analytes, but inadequate separation was encountered for oxygen and nitrogen especially when the concentrations of the two compounds are at the percent level [10–12]. Further, there is evidence of elevated reactivity of the adsorbent toward double and triple bond hydrocarbons like propylene, methyl acetylene, and 1,3-propadiene. To compensate for the lack of an appropriate stationary phase for the characterization of permanent gases and light hydrocarbons, a common chromatographic practice involves the use of a series of multi-port switching valves to selectively transfer the
186
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
Fig. 1. Analytical system configuration with two SilFlow three-port planar microfluidic devices for permanent gases and light hydrocarbons analysis.
effluent of one column to another, or bypass certain columns at various stages in the analysis to prevent the columns from being contaminated by the analytes in the matrix as in the case of carbon dioxide or water on a zeolite stationary phase like molecular sieve. Under such a scheme, it is common for a system to have three to five multi-port rotary or slider valves and a number of packed or micropacked columns or porous layer open tubular columns involved to achieve the separation required [13–15]. While this approach performs adequately, there are some important constraints, like the requirement of a dedicated gas chromatograph as a custom-built analyzer. The use of multi-port switching valves can lead to degraded chromatography due to excessive void volume encountered from connections and tubing, the lack of inertness along the sample flow path, and port-to-port cross leaks from valve rotor wear. The requirement for an additional external oven to house the valve assemblies, and the need for a highly accurate pneumatic system to minimize valve-timing shifts over the course of normal use are further complicating factors. These constraints can substantially add to the overall cost of ownership and have a negative impact on reliability and availability of the instrument for analysis. In the present article, we introduce a practical and effective integrated gas chromatographic system for the measurement of permanent gases and light hydrocarbons in one single analysis with the capability of addressing the above mentioned shortcomings. The system employs an arrangement involving the use of two threeport planar microfluidic devices in series of each other with built-in microfluidic gates and a mid-point pressure source to perform critical in-oven chromatographic tasks like column isolation and backflushing.
2. Experimental An Agilent 6890N gas chromatograph (Agilent Technologies, Wilmington, Delaware, USA) was used for all analyses. The chromatograph was equipped with one split/splitless inlet (operated at 150 ◦ C), one flame ionization detector (operated at 250 ◦ C), and
one flow modulated thermal conductivity detector (operated at 150 ◦ C), a nickel based methanizer (operated at 375 ◦ C) to convert carbon monoxide and carbon dioxide to methane, and a threechannel auxiliary pressure module. Two SilFlow three-port planar microfluidic devices, part number 123722 (SGE Analytical Science, Ringwood, Australia) were incorporated into this gas chromatograph as illustrated in Fig. 1 for the column isolation, flow diversion, and back-flushing in the analysis of permanent gases and light hydrocarbons. An Agilent universal, low pressure drop Ultra Inert liner part number 5190-3165 was used throughout. Injection volume was 1.0 mL and a split ratio of 2:1 was employed. Helium carrier gas was supplied to the column inlet using programmed pressure mode (30 psig (hold 8 min) – 2 psig (hold 4.5 min) @ 99 psig/min). The column midpoint pressure was also provided in programmed pressure mode (7.5 psig (hold 3.5 min) – 10 psig (hold 4.5 min) @ 99 psig/min then – 20 psig (hold 4 min) @ 99 psig/min). Detector gases were FID: hydrogen 45 mL/min, air 450 mL/min, nitrogen 25 mL/min; TCD reference flow 10 mL/min, make-up flow 3.5 mL/min. Columns used involved a 50 m × 0.32 mm id × 5 m CPPoraBOND Q and a 15 m × 0.32 mm id × 25 m Molecular Sieve 5A. A 5 m × 0.25 mm id, deactivated, uncoated fused silica tubing was used to connect the first three-port fluidic device to the methanizer while a 60 cm × 0.25 mm id deactivated, and uncoated fused silica tubing was used to connect the two three-port fluidic devices as described in Fig. 1. A Hewlett-Packard desktop computer equipped with a Pentium Core-Duo 2.5 MHz processor, 3 gigabytes of RAM, and 500 Gb hard drive with Windows XP Professional SVP-3 as operating system was used to host the software and process the data obtained. Data was collected with ChemStation software B.04.03SP1. Carrier and utility gases such as helium, nitrogen, hydrogen, and air used for system performance studies were acquired from Air Liquide (Edmonton, Canada). Primary gas standards were purchased from BOC (Edmonton, Canada). Secondary standards over a range from 100 ppbv to 1000 ppmv for hydrocarbons, and from 5 ppmv to 1% for oxygen and nitrogen were prepared from the primary standards by serial dilution. Samples for analysis were collected
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
187
Fig. 2. Separation of permanent gases and light hydrocarbons. (A) 1000 ppmv each of oxygen and nitrogen in helium on the TCD. (B) 1000 ppmv each of carbon monoxide, carbon dioxide, methane, ethane, ethylene, acetylene, propane and propylene in helium on the FID.
in new Tedlar bags. Tedlar bags were disposed of after use to prevent potential for contamination. The introduction of samples and standards was accomplished with a locally built Large Volume Gas Injection System (LVGIS) as described in details in Ref. [16]. 3. Results and discussion 3.1. Chromatographic system design To achieve the separation of permanent gases and light hydrocarbons without the use of multi-port valves, two three-port planar microfluidic devices were integrated together with a midpoint pressure source as depicted in Fig. 1. Each three-port device with nuts and ferrules installed weighs a mere 8 g. Embedded 50 m × 500 m fluid logic gates are incorporated with the flow architecture of the planar microfluidic devices [17–19]. A fluid logic gate is an orifice in a flow path, implemented to achieve a high flow velocity on the spot to prevent mixing or back diffusion of a gas against a flow stream, yet inducing minimal pressure drop. In the present arrangement, the first three-port functions as a splitter where the effluent of the first column is split to detector 1 via a fused silica restrictor, and to the second three-port device. In the second three-port device, a mid-point pressure is introduced. A second column and detector with different selectivities when compared to the first column and detector, were also connected to the second three-port device. With this configuration, depending on the difference between the inlet pressure (P1) and the mid-point pressure (P2), a number of critical chromatography applications can be performed such as second column isolation or analyte diversion, or back-flushing of the first column. For instance, If P1 P2, the effluent of Column 1 will flow to Detector 1 as well as to Column 2, and subsequently to Detector 2. At a certain value of P2, typically less than P1, but higher than the outlet pressure of Column 1, Column 2 can be isolated from the analytical system, yet receiving continuous carrier flow across the column with the effluent of Column 1 now flows solely
to Detector 1. If P2 is elevated substantially higher than P1, backflushing of Column 1 occurs. In the configuration proposed for the separation of permanent gases and hydrocarbons, a 50 m × 0.32-mm id × 5 m CP-PoraBOND Q was used as the first column while a 15 m × 0.32mm id × 25 m CP-MolSieve 5A was used as the second column. Injection of the sample is made onto the CP-PoraBOND Q column where the hydrocarbons and water, if present, are retained and separated while the non-retained solutes such as permanent gases pass through the column and onto the molecular sieve column. Once this is completed, the mid-point pressure is raised so that the molecular sieve column is now isolated and the effluent from the CP-PoraBOND Q flows into the restrictor, where all the hydrocarbons are detected by the first detector. The separations of hydrocarbons are carried out uneventfully on a divinyl benzene column. With the employment of a nickel-based methanizer operating at 375 ◦ C in a hydrogen rich reducing atmosphere, carbon monoxide and carbon dioxide are converted to methane, and these two components can be detected using the FID. All of these manipulations were conducted without the need of multi-port rotary valves. 3.2. Chromatographic system optimization and performance Chromatograms obtained from a 100 ppmv each of oxygen and nitrogen in helium on the TCD was shown in Fig. 2A and from a 1000 ppmv each of carbon monoxide, carbon dioxide, methane, ethane, ethylene, acetylene, propane and propylene in helium on the FID was illustrated in Fig. 2B, with an inlet pressure (P1) at 30 psig. With the restriction induced by the CP-MolSieve 5A (Column 2) and the restrictor from the second three-port planar microfluidic device, the outlet pressure of the CP-PoraBOND Q (Column 1) was measured by the on-board pressure sensor to be at 6.5 psig without applying any mid-point pressure (P2). Under this condition, the effluent of Column 1 flows into both the methanizer/Detector 1 and the inlet of Column 2/Detector 2. To determine a mid-point pressure whereby Column 2 can be isolated from the
188
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
Fig. 3. Isolation of molecular sieve column with an increase of mid-point pressure (P2) at 7.5, 9.0, and 10 psig respectively.
analytical system, as shown in Fig. 3, P2 was increased to higher pressure settings, namely 7.5, 9.0 and 10.0 psig respectively. The two analytes of interest, oxygen and nitrogen were detected by the TCD with a mid-point pressure setting of up to 9.0 psig. At 10 psig and beyond, these two compounds were not detected, indicating that under this condition, Column 2 is completely isolated from the effluent of Column 1. As a result, to protect the integrity of the molecular sieve column from the heavier hydrocarbons, carbon
dioxide, and water in the sample matrix, once the non-retained analytes in Column 1 are transferred or when the isolation of Column 2 is required, a P2 value of 10 psig was selected. Back-flushing of Column 1 can simply be achieved by lowering the pressure of P1 from 30 psig to 2 psig while raising the pressure of P2 to 30 psig. Figs. 4 and 5 show chromatograms of a synthetic reformed gas mixture containing oxygen, methane, carbon monoxide and carbon dioxide in a helium matrix, analyzed with
Fig. 4. Chromatogram of a synthetic reformed gas with oxygen, carbon monoxide, methane, and carbon dioxide–FID channel.
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
189
Fig. 5. Chromatogram of a synthetic reformed gas with oxygen, carbon monoxide, methane, and carbon dioxide (diverted)–TCD channel.
the apparatus described. In Fig. 4, as predicted, oxygen and carbon monoxide cannot be separated with the divinyl benzene column, whereas these two analytes are well resolved on the molecular sieve column with Rs > 20 (Fig. 5). After methane was eluted from the divinyl benzene column, the mid-point pressure was raised at 3.4 min to 10 psig to isolate but maintain carrier gas flow across the molecular sieve column. With the molecular sieve column isolated, carbon dioxide, which would have been irreversibly adsorbed on the molecular sieve column, eluted from the divinyl benzene column, converted by the methanizer to methane, and subsequently detected by the FID. The chromatographic performance of the system was found to be reliable under the conditions established. Reproducibility of retention times for all compounds were found to be less than 0.1% (n = 20). Reproducibility of area counts at two levels, namely 100 ppmv and 1000 ppmv were found to be less than 5.5% (n = 20) over a period of two days. Oxygen and nitrogen were found to be linear over a range from 20 ppmv to 10,000 ppmv with correlation coefficients of at least 0.998 and detection limits at 10 ppmv . Hydrocarbons of interest were found to be linear over a range from 200 ppbv to 1000 ppmv with correlation coefficients of greater than 0.999 and detection limits at 100 ppbv . Unlike the switching valve system, where permanent plumbing is required and reconfiguration of the analytical system can be time consuming and impractical, using this configuration a large permutation of column choices, dimensions, and selectivities can be exploited to meet the analytical needs. For example, the restrictor used can be replaced by yet another capillary column to offer extra selectivity, or a stop-flow arrangement can be implemented to achieve tune-able selectivity between the first and the second column. To demonstrate the high degree of flexibility this configuration has to offer, Fig. 6 shows the separation of hydrocarbons and primary alcohols, with the first column being a 30 m × 0.32 mm id × 1.2 m VF-Waxms and the second column being a 15 m × 0.32 mm id × 10 m Rt-Aluminum Oxide MAPD column. Two flame ionization detectors were employed. On the VF-Waxms column, the light hydrocarbons are not retained and eluted as one discrete peak as shown in Fig. 6A. The mid-point
pressure was then raised at 1.5 min to isolate yet at the same time maintain carrier gas flow on the Rt-Aluminum Oxide MAPD column where the light hydrocarbons were separated and detected by the second FID as illustrated in Fig. 6B. With the Rt-Aluminum Oxide MAPD column completely isolated from the analytical system, the alcohols, having been separated by the VF-Waxms column, were diverted toward and detected by the first FID as shown in Fig. 6A. A complete analysis for all the analytes of interest can be conducted in less than 10 min with a single instrument. If this analysis were to be performed using single column analysis, all the light hydrocarbons would have co-eluted on the VF-Waxms column as suggested in Fig. 6A while all the alcohols would have been irreversibly adsorbed on the Rt-Aluminum Oxide MAPD column which in turn will cause degradation to the column performance. While many of the capabilities described with this configuration can also be achieved with a classical pressure-based Deans’ switch, the configuration is economical and has proven to be straight forward to implement, and sufficiently rugged for production laboratory or field implementation. 3.3. Constraints The shedding of the particles used as a stationary phase, particularly from Column 1 located on the upstream of the first fluidic device, can cause flow restrictions to the device. A technique that was found useful in substantially minimizing (if not completely eliminating) this possibility involved purging the column at a high flow rate of 30 mL/min for 10 min to remove any particles loosened during column shipment followed by taking the column through a few thermal cycles before connecting the column to the device. Also, in comparison to single column analysis, due to the fact that the effluent of Column 1 is split into two channels, reduced sensitivity for oxygen and nitrogen was observed. However, if required, the detection limits for all analytes can be further reduced by a factor of two by increasing the size of the sample injected from 1 to 2 mL. It is worth noting that the molecular sieve column employed does not separate oxygen from argon. However, the separation of
190
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
Fig. 6. Overlay of chromatograms for the separation of oxygenated compounds and hydrocarbons in one single analysis with a configuration of two three-port SilFlow planar microfluidic devices. (A) Separation of light hydrocarbons and oxygenated with the VF-Waxms column. (B) Separation of light hydrocarbons with the Rt-Aluminum Oxide MAPD column.
these two analytes can be attained with above ambient temperature chromatography using commercially available thicker layer molecular sieve column technology.
measurement needs as demonstrated in the separation of hydrocarbons and oxygenated compounds. Acknowledgements
4. Conclusions Despite its importance and popularity, the actual hardware used to perform connections and the implementation of techniques involved with packed or micropacked columns are not fully compatible for use with capillary gas chromatography. Large thermal mass fittings can cause cold spots, leaks from graphitized/Vespel ferrules can negatively impact system reliability, and attempts to adapt the techniques to capillary column chromatography with commercially available products such as removable fused silica adaptors and ferrules, or glass press-fit connectors were met with limited success. The use of multi-port rotary or slider valves commonly employed in this application for switching purpose further exacerbate the problems. Migration from tube-based flow systems to planar micro-channel systems allows the delivery of flexible and innovative solutions with key features such as being leak-free over multiple thermal cycles, low thermal mass, in-column switching, and inertness. By employing a configuration consisting of two three-port planar microfluidic devices with built-in fluidic gates and a mid-point pressure source to perform chromatographic tasks like column isolation and back-flushing, critical analyses in petrochemical and chemical industries such as the measurements of permanent gases and hydrocarbons can be conducted in one single analysis with respectable performance. The described configuration has proven to be easy to implement and maintain, suitable for field or production lab implementation. This approach eliminated many chromatographic issues encountered with column connectivity especially in capillary gas chromatography, and the use of multi-port valves employed for switching purposes. With this concept, a large permutation of column choices, dimensions, and selectivities can be exploited to meet other chromatographic
Kaye Burns, Dan Martin, Mark Marinan, Jeff Mason, Dave Walter and Andy Szigety of Dow Analytical Technology Center were acknowledged for their support. Michelle Baker and Vicki Carter were recognized for their assistance. Robert Shellie is the recipient of an Australian Research Council Australian Research Fellowship (project number DP110104923). SilFlow and SilTite are trademarks of SGE Analytical Science, Ringwood, Australia. CP-PoraBOND Q, CP-Molsieve 5A, VF-Waxms are trademarks of Agilent Technologies, Middelburg, the Netherlands. Rt-Alumina BOND MAPD is a trademark of Restek Corporation, PA, USA. Vespel is a trademark of DuPont Company, DE, USA. References [1] A. Drew, Manual on Hydrocarbons Analysis, ASTM, West Conshohocken, PA, USA, 1998. [2] M.K. Sarkar, G.G. Haselden, J. Chromatogr. 104 (1975) 425–428. [3] F. Poy, L. Cobbelli, J. Chromatogr. 349 (1985) 17–22. [4] M. Janse van Rensburg, A. Botha, E. Rohwer, J. Chromatogr. A 1167 (2007) 102–108. [5] N. Snow, J. Chromatogr. 47 (2006) 443–483. [6] A. Corma, A.V. Orchillés, Microporous Mesoporous Mater. 35 (2000) 21–30. [7] N. Rahimi, R. Karimzadeh, Appl. Catal. A 398 (2001) 1–17. [8] N. Fonseca, L. dos Santos, H. Cerqueira, F. Lemos, F. Ramôa-Ribeiro, Y. Lam, M. de Almeida, Fuel 95 (2012) 183–189. [9] V.G. Berezkin, J. de Zeeuw, Capillary Gas Adsorption Chromatography, Huthig Publisher, Heidelberg, 1996, pp. 268–269. [10] Sigma-Aldrich, Carboxen GC PLOT Capillary Columns, Supelco Product Literature T-403146, 2003. [11] I. Wilson, C. Poole, Handbook of Methods and Instrumentation in Separation Science, Elsevier, London, United Kingdom, 2009, p. 230. [12] J. de Zeeuw, J. Luong, Trends Anal. Chem. 21 (2002) 594–607. [13] X. Li, Z. Guan, Analysis of permanent gases and light hydrocarbons using Agilent 7820A GC with three-valve system, Agilent Technologies Application note 5990-4667EN, 2009.
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191 [14] X. Wang, Parallel GC for complete RGA analysis, Agilent Technologies Application note 5989-7438EN, 2008. [15] Y. Zhou, X. Wang, R. Firor, Analysis of permanent gases and methane, Agilent Technologies Application note 5988-5960EN, 2003. [16] J. Luong, R. Gras, L. Sieben, R. Firor, J. Chromatogr. A 1216 (2009) 2740–2748. [17] J. Luong, R. Gras, H. Cortes, R. Shellie, in: L. Mondello (Ed.), Proceedings of the 36th International Symposium on Capillary Chromatography, Plenary Lecture L.E.12, Riva Del Garda, 2012, L. Mondello Publisher, Messina, 2012.
191
[18] P. Dawes, B. Barnett, R. Hibbert, in: P. Sandra (Ed.), Proceedings of the 34th International Symposium on Capillary Chromatography, Poster, Riva Del Garda, 2010, Pat Sandra Publisher, Kortrijk, 2010. [19] R. Freeman, P. Dawes, B. Barnett, R. Hillbert, Proceedings of the 35th International Symposium on Capillary Chromatography, Poster, San Diego, 2011, CASSS Publisher, San Diego, 2011.
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Multidimensional gas chromatography for the characterization of permanent gases and light hydrocarbons in catalytic cracking process J. Luong a,b , R. Gras b , H.J. Cortes a,c , R.A. Shellie a,∗ a b c
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, MI 48642, USA
a r t i c l e
i n f o
Article history: Received 3 September 2012 Received in revised form 12 November 2012 Accepted 12 November 2012 Available online 19 November 2012 Keywords: Gas chromatography MDGC Planar microfluidic device Permanent gases Light hydrocarbons
a b s t r a c t An integrated gas chromatographic system has been successfully developed and implemented for the measurement of oxygen, nitrogen, carbon monoxide, carbon dioxide and light hydrocarbons in one single analysis. These analytes are frequently encountered in critical industrial petrochemical and chemical processes like catalytic cracking of naphtha or diesel fuel to lighter components used in gasoline. The system employs a practical, effective configuration consisting of two three-port planar microfluidic devices in series with each other, having built-in fluidic gates, and a mid-point pressure source. The use of planar microfluidic devices offers intangible advantages like in-oven switching with no mechanical moving parts, an inert sample flow path, and a leak-free operation even with multiple thermal cycles. In this way, necessary features such as selectivity enhancement, column isolation, column back-flushing, and improved system cleanliness were realized. Porous layer open tubular capillary columns were employed for the separation of hydrocarbons followed by flame ionization detection. After separation has occurred, carbon monoxide and carbon dioxide were converted to methane with the use of a nickel-based methanizer for detection with flame ionization. Flow modulated thermal conductivity detection was employed to measure oxygen and nitrogen. Separation of all the target analytes was achieved in one single analysis of less than 12 min. Reproducibility of retention times for all compounds were found to be less than 0.1% (n = 20). Reproducibility of area counts at two levels, namely 100 ppmv and 1000 ppmv over a period of two days were found to be less than 5.5% (n = 20). Oxygen and nitrogen were found to be linear over a range from 20 ppmv to 10,000 ppmv with correlation coefficients of at least 0.998 and detection limits of less than 10 ppmv . Hydrocarbons of interest were found to be linear over a range from 200 ppbv to 1000 ppmv with correlation coefficients of greater than 0.999 and detection limits of less than 100 ppbv . © 2012 Elsevier B.V. All rights reserved.
1. Introduction The characterization of oxygen, nitrogen, carbon monoxide, carbon dioxide (permanent gases) and light hydrocarbons is an important analysis for petrochemical and chemical industries [1–5]. These compounds are often encountered in various critical chemical processes like catalytic cracking for example [6–8]. Gas chromatography is a technique of choice due to the high degree of volatility of the compounds involved. Unfortunately, without the employment of cryogenic chromatography, no single chromatographic stationary phase is capable of adequately separating all of these analytes [9]. For instance, on a porous polymer based
∗ Corresponding author. Tel.: +61 3 6226 7656; fax: +61 3 6226 2858. E-mail address: [email protected] (R.A. Shellie). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.11.025
stationary phase like divinyl benzene, oxygen, nitrogen, and carbon monoxide are perfectly co-eluted. Hydrogen, oxygen, nitrogen, methane, and carbon monoxide can be well separated using a molecular sieve stationary phase; however carbon dioxide and heavier hydrocarbons are irreversibly adsorbed. The presence of water in the sample as vapor can also deactivate the stationary phase. Carbon molecular sieve or carbon black has been used to separate these analytes, but inadequate separation was encountered for oxygen and nitrogen especially when the concentrations of the two compounds are at the percent level [10–12]. Further, there is evidence of elevated reactivity of the adsorbent toward double and triple bond hydrocarbons like propylene, methyl acetylene, and 1,3-propadiene. To compensate for the lack of an appropriate stationary phase for the characterization of permanent gases and light hydrocarbons, a common chromatographic practice involves the use of a series of multi-port switching valves to selectively transfer the
186
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
Fig. 1. Analytical system configuration with two SilFlow three-port planar microfluidic devices for permanent gases and light hydrocarbons analysis.
effluent of one column to another, or bypass certain columns at various stages in the analysis to prevent the columns from being contaminated by the analytes in the matrix as in the case of carbon dioxide or water on a zeolite stationary phase like molecular sieve. Under such a scheme, it is common for a system to have three to five multi-port rotary or slider valves and a number of packed or micropacked columns or porous layer open tubular columns involved to achieve the separation required [13–15]. While this approach performs adequately, there are some important constraints, like the requirement of a dedicated gas chromatograph as a custom-built analyzer. The use of multi-port switching valves can lead to degraded chromatography due to excessive void volume encountered from connections and tubing, the lack of inertness along the sample flow path, and port-to-port cross leaks from valve rotor wear. The requirement for an additional external oven to house the valve assemblies, and the need for a highly accurate pneumatic system to minimize valve-timing shifts over the course of normal use are further complicating factors. These constraints can substantially add to the overall cost of ownership and have a negative impact on reliability and availability of the instrument for analysis. In the present article, we introduce a practical and effective integrated gas chromatographic system for the measurement of permanent gases and light hydrocarbons in one single analysis with the capability of addressing the above mentioned shortcomings. The system employs an arrangement involving the use of two threeport planar microfluidic devices in series of each other with built-in microfluidic gates and a mid-point pressure source to perform critical in-oven chromatographic tasks like column isolation and backflushing.
2. Experimental An Agilent 6890N gas chromatograph (Agilent Technologies, Wilmington, Delaware, USA) was used for all analyses. The chromatograph was equipped with one split/splitless inlet (operated at 150 ◦ C), one flame ionization detector (operated at 250 ◦ C), and
one flow modulated thermal conductivity detector (operated at 150 ◦ C), a nickel based methanizer (operated at 375 ◦ C) to convert carbon monoxide and carbon dioxide to methane, and a threechannel auxiliary pressure module. Two SilFlow three-port planar microfluidic devices, part number 123722 (SGE Analytical Science, Ringwood, Australia) were incorporated into this gas chromatograph as illustrated in Fig. 1 for the column isolation, flow diversion, and back-flushing in the analysis of permanent gases and light hydrocarbons. An Agilent universal, low pressure drop Ultra Inert liner part number 5190-3165 was used throughout. Injection volume was 1.0 mL and a split ratio of 2:1 was employed. Helium carrier gas was supplied to the column inlet using programmed pressure mode (30 psig (hold 8 min) – 2 psig (hold 4.5 min) @ 99 psig/min). The column midpoint pressure was also provided in programmed pressure mode (7.5 psig (hold 3.5 min) – 10 psig (hold 4.5 min) @ 99 psig/min then – 20 psig (hold 4 min) @ 99 psig/min). Detector gases were FID: hydrogen 45 mL/min, air 450 mL/min, nitrogen 25 mL/min; TCD reference flow 10 mL/min, make-up flow 3.5 mL/min. Columns used involved a 50 m × 0.32 mm id × 5 m CPPoraBOND Q and a 15 m × 0.32 mm id × 25 m Molecular Sieve 5A. A 5 m × 0.25 mm id, deactivated, uncoated fused silica tubing was used to connect the first three-port fluidic device to the methanizer while a 60 cm × 0.25 mm id deactivated, and uncoated fused silica tubing was used to connect the two three-port fluidic devices as described in Fig. 1. A Hewlett-Packard desktop computer equipped with a Pentium Core-Duo 2.5 MHz processor, 3 gigabytes of RAM, and 500 Gb hard drive with Windows XP Professional SVP-3 as operating system was used to host the software and process the data obtained. Data was collected with ChemStation software B.04.03SP1. Carrier and utility gases such as helium, nitrogen, hydrogen, and air used for system performance studies were acquired from Air Liquide (Edmonton, Canada). Primary gas standards were purchased from BOC (Edmonton, Canada). Secondary standards over a range from 100 ppbv to 1000 ppmv for hydrocarbons, and from 5 ppmv to 1% for oxygen and nitrogen were prepared from the primary standards by serial dilution. Samples for analysis were collected
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
187
Fig. 2. Separation of permanent gases and light hydrocarbons. (A) 1000 ppmv each of oxygen and nitrogen in helium on the TCD. (B) 1000 ppmv each of carbon monoxide, carbon dioxide, methane, ethane, ethylene, acetylene, propane and propylene in helium on the FID.
in new Tedlar bags. Tedlar bags were disposed of after use to prevent potential for contamination. The introduction of samples and standards was accomplished with a locally built Large Volume Gas Injection System (LVGIS) as described in details in Ref. [16]. 3. Results and discussion 3.1. Chromatographic system design To achieve the separation of permanent gases and light hydrocarbons without the use of multi-port valves, two three-port planar microfluidic devices were integrated together with a midpoint pressure source as depicted in Fig. 1. Each three-port device with nuts and ferrules installed weighs a mere 8 g. Embedded 50 m × 500 m fluid logic gates are incorporated with the flow architecture of the planar microfluidic devices [17–19]. A fluid logic gate is an orifice in a flow path, implemented to achieve a high flow velocity on the spot to prevent mixing or back diffusion of a gas against a flow stream, yet inducing minimal pressure drop. In the present arrangement, the first three-port functions as a splitter where the effluent of the first column is split to detector 1 via a fused silica restrictor, and to the second three-port device. In the second three-port device, a mid-point pressure is introduced. A second column and detector with different selectivities when compared to the first column and detector, were also connected to the second three-port device. With this configuration, depending on the difference between the inlet pressure (P1) and the mid-point pressure (P2), a number of critical chromatography applications can be performed such as second column isolation or analyte diversion, or back-flushing of the first column. For instance, If P1 P2, the effluent of Column 1 will flow to Detector 1 as well as to Column 2, and subsequently to Detector 2. At a certain value of P2, typically less than P1, but higher than the outlet pressure of Column 1, Column 2 can be isolated from the analytical system, yet receiving continuous carrier flow across the column with the effluent of Column 1 now flows solely
to Detector 1. If P2 is elevated substantially higher than P1, backflushing of Column 1 occurs. In the configuration proposed for the separation of permanent gases and hydrocarbons, a 50 m × 0.32-mm id × 5 m CP-PoraBOND Q was used as the first column while a 15 m × 0.32mm id × 25 m CP-MolSieve 5A was used as the second column. Injection of the sample is made onto the CP-PoraBOND Q column where the hydrocarbons and water, if present, are retained and separated while the non-retained solutes such as permanent gases pass through the column and onto the molecular sieve column. Once this is completed, the mid-point pressure is raised so that the molecular sieve column is now isolated and the effluent from the CP-PoraBOND Q flows into the restrictor, where all the hydrocarbons are detected by the first detector. The separations of hydrocarbons are carried out uneventfully on a divinyl benzene column. With the employment of a nickel-based methanizer operating at 375 ◦ C in a hydrogen rich reducing atmosphere, carbon monoxide and carbon dioxide are converted to methane, and these two components can be detected using the FID. All of these manipulations were conducted without the need of multi-port rotary valves. 3.2. Chromatographic system optimization and performance Chromatograms obtained from a 100 ppmv each of oxygen and nitrogen in helium on the TCD was shown in Fig. 2A and from a 1000 ppmv each of carbon monoxide, carbon dioxide, methane, ethane, ethylene, acetylene, propane and propylene in helium on the FID was illustrated in Fig. 2B, with an inlet pressure (P1) at 30 psig. With the restriction induced by the CP-MolSieve 5A (Column 2) and the restrictor from the second three-port planar microfluidic device, the outlet pressure of the CP-PoraBOND Q (Column 1) was measured by the on-board pressure sensor to be at 6.5 psig without applying any mid-point pressure (P2). Under this condition, the effluent of Column 1 flows into both the methanizer/Detector 1 and the inlet of Column 2/Detector 2. To determine a mid-point pressure whereby Column 2 can be isolated from the
188
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
Fig. 3. Isolation of molecular sieve column with an increase of mid-point pressure (P2) at 7.5, 9.0, and 10 psig respectively.
analytical system, as shown in Fig. 3, P2 was increased to higher pressure settings, namely 7.5, 9.0 and 10.0 psig respectively. The two analytes of interest, oxygen and nitrogen were detected by the TCD with a mid-point pressure setting of up to 9.0 psig. At 10 psig and beyond, these two compounds were not detected, indicating that under this condition, Column 2 is completely isolated from the effluent of Column 1. As a result, to protect the integrity of the molecular sieve column from the heavier hydrocarbons, carbon
dioxide, and water in the sample matrix, once the non-retained analytes in Column 1 are transferred or when the isolation of Column 2 is required, a P2 value of 10 psig was selected. Back-flushing of Column 1 can simply be achieved by lowering the pressure of P1 from 30 psig to 2 psig while raising the pressure of P2 to 30 psig. Figs. 4 and 5 show chromatograms of a synthetic reformed gas mixture containing oxygen, methane, carbon monoxide and carbon dioxide in a helium matrix, analyzed with
Fig. 4. Chromatogram of a synthetic reformed gas with oxygen, carbon monoxide, methane, and carbon dioxide–FID channel.
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
189
Fig. 5. Chromatogram of a synthetic reformed gas with oxygen, carbon monoxide, methane, and carbon dioxide (diverted)–TCD channel.
the apparatus described. In Fig. 4, as predicted, oxygen and carbon monoxide cannot be separated with the divinyl benzene column, whereas these two analytes are well resolved on the molecular sieve column with Rs > 20 (Fig. 5). After methane was eluted from the divinyl benzene column, the mid-point pressure was raised at 3.4 min to 10 psig to isolate but maintain carrier gas flow across the molecular sieve column. With the molecular sieve column isolated, carbon dioxide, which would have been irreversibly adsorbed on the molecular sieve column, eluted from the divinyl benzene column, converted by the methanizer to methane, and subsequently detected by the FID. The chromatographic performance of the system was found to be reliable under the conditions established. Reproducibility of retention times for all compounds were found to be less than 0.1% (n = 20). Reproducibility of area counts at two levels, namely 100 ppmv and 1000 ppmv were found to be less than 5.5% (n = 20) over a period of two days. Oxygen and nitrogen were found to be linear over a range from 20 ppmv to 10,000 ppmv with correlation coefficients of at least 0.998 and detection limits at 10 ppmv . Hydrocarbons of interest were found to be linear over a range from 200 ppbv to 1000 ppmv with correlation coefficients of greater than 0.999 and detection limits at 100 ppbv . Unlike the switching valve system, where permanent plumbing is required and reconfiguration of the analytical system can be time consuming and impractical, using this configuration a large permutation of column choices, dimensions, and selectivities can be exploited to meet the analytical needs. For example, the restrictor used can be replaced by yet another capillary column to offer extra selectivity, or a stop-flow arrangement can be implemented to achieve tune-able selectivity between the first and the second column. To demonstrate the high degree of flexibility this configuration has to offer, Fig. 6 shows the separation of hydrocarbons and primary alcohols, with the first column being a 30 m × 0.32 mm id × 1.2 m VF-Waxms and the second column being a 15 m × 0.32 mm id × 10 m Rt-Aluminum Oxide MAPD column. Two flame ionization detectors were employed. On the VF-Waxms column, the light hydrocarbons are not retained and eluted as one discrete peak as shown in Fig. 6A. The mid-point
pressure was then raised at 1.5 min to isolate yet at the same time maintain carrier gas flow on the Rt-Aluminum Oxide MAPD column where the light hydrocarbons were separated and detected by the second FID as illustrated in Fig. 6B. With the Rt-Aluminum Oxide MAPD column completely isolated from the analytical system, the alcohols, having been separated by the VF-Waxms column, were diverted toward and detected by the first FID as shown in Fig. 6A. A complete analysis for all the analytes of interest can be conducted in less than 10 min with a single instrument. If this analysis were to be performed using single column analysis, all the light hydrocarbons would have co-eluted on the VF-Waxms column as suggested in Fig. 6A while all the alcohols would have been irreversibly adsorbed on the Rt-Aluminum Oxide MAPD column which in turn will cause degradation to the column performance. While many of the capabilities described with this configuration can also be achieved with a classical pressure-based Deans’ switch, the configuration is economical and has proven to be straight forward to implement, and sufficiently rugged for production laboratory or field implementation. 3.3. Constraints The shedding of the particles used as a stationary phase, particularly from Column 1 located on the upstream of the first fluidic device, can cause flow restrictions to the device. A technique that was found useful in substantially minimizing (if not completely eliminating) this possibility involved purging the column at a high flow rate of 30 mL/min for 10 min to remove any particles loosened during column shipment followed by taking the column through a few thermal cycles before connecting the column to the device. Also, in comparison to single column analysis, due to the fact that the effluent of Column 1 is split into two channels, reduced sensitivity for oxygen and nitrogen was observed. However, if required, the detection limits for all analytes can be further reduced by a factor of two by increasing the size of the sample injected from 1 to 2 mL. It is worth noting that the molecular sieve column employed does not separate oxygen from argon. However, the separation of
190
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191
Fig. 6. Overlay of chromatograms for the separation of oxygenated compounds and hydrocarbons in one single analysis with a configuration of two three-port SilFlow planar microfluidic devices. (A) Separation of light hydrocarbons and oxygenated with the VF-Waxms column. (B) Separation of light hydrocarbons with the Rt-Aluminum Oxide MAPD column.
these two analytes can be attained with above ambient temperature chromatography using commercially available thicker layer molecular sieve column technology.
measurement needs as demonstrated in the separation of hydrocarbons and oxygenated compounds. Acknowledgements
4. Conclusions Despite its importance and popularity, the actual hardware used to perform connections and the implementation of techniques involved with packed or micropacked columns are not fully compatible for use with capillary gas chromatography. Large thermal mass fittings can cause cold spots, leaks from graphitized/Vespel ferrules can negatively impact system reliability, and attempts to adapt the techniques to capillary column chromatography with commercially available products such as removable fused silica adaptors and ferrules, or glass press-fit connectors were met with limited success. The use of multi-port rotary or slider valves commonly employed in this application for switching purpose further exacerbate the problems. Migration from tube-based flow systems to planar micro-channel systems allows the delivery of flexible and innovative solutions with key features such as being leak-free over multiple thermal cycles, low thermal mass, in-column switching, and inertness. By employing a configuration consisting of two three-port planar microfluidic devices with built-in fluidic gates and a mid-point pressure source to perform chromatographic tasks like column isolation and back-flushing, critical analyses in petrochemical and chemical industries such as the measurements of permanent gases and hydrocarbons can be conducted in one single analysis with respectable performance. The described configuration has proven to be easy to implement and maintain, suitable for field or production lab implementation. This approach eliminated many chromatographic issues encountered with column connectivity especially in capillary gas chromatography, and the use of multi-port valves employed for switching purposes. With this concept, a large permutation of column choices, dimensions, and selectivities can be exploited to meet other chromatographic
Kaye Burns, Dan Martin, Mark Marinan, Jeff Mason, Dave Walter and Andy Szigety of Dow Analytical Technology Center were acknowledged for their support. Michelle Baker and Vicki Carter were recognized for their assistance. Robert Shellie is the recipient of an Australian Research Council Australian Research Fellowship (project number DP110104923). SilFlow and SilTite are trademarks of SGE Analytical Science, Ringwood, Australia. CP-PoraBOND Q, CP-Molsieve 5A, VF-Waxms are trademarks of Agilent Technologies, Middelburg, the Netherlands. Rt-Alumina BOND MAPD is a trademark of Restek Corporation, PA, USA. Vespel is a trademark of DuPont Company, DE, USA. References [1] A. Drew, Manual on Hydrocarbons Analysis, ASTM, West Conshohocken, PA, USA, 1998. [2] M.K. Sarkar, G.G. Haselden, J. Chromatogr. 104 (1975) 425–428. [3] F. Poy, L. Cobbelli, J. Chromatogr. 349 (1985) 17–22. [4] M. Janse van Rensburg, A. Botha, E. Rohwer, J. Chromatogr. A 1167 (2007) 102–108. [5] N. Snow, J. Chromatogr. 47 (2006) 443–483. [6] A. Corma, A.V. Orchillés, Microporous Mesoporous Mater. 35 (2000) 21–30. [7] N. Rahimi, R. Karimzadeh, Appl. Catal. A 398 (2001) 1–17. [8] N. Fonseca, L. dos Santos, H. Cerqueira, F. Lemos, F. Ramôa-Ribeiro, Y. Lam, M. de Almeida, Fuel 95 (2012) 183–189. [9] V.G. Berezkin, J. de Zeeuw, Capillary Gas Adsorption Chromatography, Huthig Publisher, Heidelberg, 1996, pp. 268–269. [10] Sigma-Aldrich, Carboxen GC PLOT Capillary Columns, Supelco Product Literature T-403146, 2003. [11] I. Wilson, C. Poole, Handbook of Methods and Instrumentation in Separation Science, Elsevier, London, United Kingdom, 2009, p. 230. [12] J. de Zeeuw, J. Luong, Trends Anal. Chem. 21 (2002) 594–607. [13] X. Li, Z. Guan, Analysis of permanent gases and light hydrocarbons using Agilent 7820A GC with three-valve system, Agilent Technologies Application note 5990-4667EN, 2009.
J. Luong et al. / J. Chromatogr. A 1271 (2013) 185–191 [14] X. Wang, Parallel GC for complete RGA analysis, Agilent Technologies Application note 5989-7438EN, 2008. [15] Y. Zhou, X. Wang, R. Firor, Analysis of permanent gases and methane, Agilent Technologies Application note 5988-5960EN, 2003. [16] J. Luong, R. Gras, L. Sieben, R. Firor, J. Chromatogr. A 1216 (2009) 2740–2748. [17] J. Luong, R. Gras, H. Cortes, R. Shellie, in: L. Mondello (Ed.), Proceedings of the 36th International Symposium on Capillary Chromatography, Plenary Lecture L.E.12, Riva Del Garda, 2012, L. Mondello Publisher, Messina, 2012.
191
[18] P. Dawes, B. Barnett, R. Hibbert, in: P. Sandra (Ed.), Proceedings of the 34th International Symposium on Capillary Chromatography, Poster, Riva Del Garda, 2010, Pat Sandra Publisher, Kortrijk, 2010. [19] R. Freeman, P. Dawes, B. Barnett, R. Hillbert, Proceedings of the 35th International Symposium on Capillary Chromatography, Poster, San Diego, 2011, CASSS Publisher, San Diego, 2011.