Determination of Sulfur-Based Odorants in Commercially Available Natural Gas with Flow-Modulated Comprehensive Two-Dimensional Gas Chromatography

March 1, 2013
T. Hayward

,
J. Luong

,
R. Gras

LCGC North America

LCGC North America, LCGC North America-03-01-2013, Volume 31, Issue 3
Page Number: 224–231

A method able to quantitate odorants at levels as low as 100 ppb (v/v)

Alkyl mercaptans, alkyl sulfides, and cyclic sulfides are added in various blends to commercial natural gas as a safety precaution to identify its presence by smell. Although several different analytical methods exist to measure these compounds, few are all-encompassing techniques that can measure all the sulfur compounds simultaneously at appropriate levels. Pulsed-flow-modulated comprehensive two-dimensional gas chromatography with flame ionization detection was used to separate and identify all common sulfur compounds found in natural gas. The column set used in this method consisted of a polydimethylsiloxane-like stationary-phase selectivity in the first dimension with a low-bleed wax stationary phase in the second dimension. This set was optimized to separate key sulfur compounds found in natural gas from this hydrocarbon matrix. Detection limits of this method were determined to be 0.1 ppm (v/v), and a linear calibration range spanning from 0.1 to 30 ppm (v/v) was used to perform quantitative analysis on a residential natural gas supply. Reproducibility of peak volume of seven successive injections on these test analytes averaged 3.7%.

Natural gas is highly flammable, colorless, and odorless; therefore malodorous compounds are added as a method to identify the presence of this gas. Most commonly, these compounds consist of a blend of alkyl mercaptans, alkyl sulfides, or cyclic sulfides (1,2). These structures are prominent because of their desirable physical and chemical characteristics. For example, the blends need to be odorous, volatile, flammable liquids that have a low odor threshold and will not oxidize metal pipelines (1,2). The most common sulfur compounds used as odorants are tert-butyl mercaptan, tetrahydrothiophene, methylethyl sulfide, dimethyl sulfide, isopropyl mercaptan, n-propyl mercaptan, and sec-butyl mercaptan. Two or more of these compounds are added to natural gas before it is delivered to the consumer (1,2), and depending on environmental or economical factors, they will be blended in various combinations and percentages.

The ability to monitor these odorant blends in all their variant forms is beneficial to natural gas providers to certify their presence at detectable levels by smell and ensure consumer safety. North American regulations require an odorant to be detectable by smell when natural gas concentrations reach one-fifth of the lower explosive limit, which represents a level of about 1.25% in air (Federal Regulation, 49 CFR, 192.625). Most sulfur odorants have an odor threshold of approximately 1 ppb (3), and therefore are minimally required to be at concentrations of 100 ppb in natural gas. Generally, commercially available residential natural gas contains sulfur odorants with 0.5–10 ppm levels. At these levels, natural gas can be readily detected by smell at concentrations well below the lower explosive limit. It is also necessary to ensure that these odorants are not overdosed, which can lead to further expenses to the natural gas supplier. Excessive additions in natural gas can cost companies millions of dollars in unnecessary chemical cost as well as investigations because of false positive alarms reported by consumers. Higher levels of sulfur species within the distribution pipelines may also corrode these pipes and create potentially hazardous leaks.

Currently, odorants in natural gas are monitored by various sensors such as olfactory detection (ASTM D6273-08) and lead acetate strip tests (4–6). Gas chromatography (GC) has also been demonstrated to detect these compounds, which usually involves a selective detection method such as flame photometric detection (FPD) (7–10), sulfur chemiluminescence detection (SCD) (11–13), mass spectrometry (MS) (14), or ion mobility spectrometry (15). The use of FPD requires the sulfur compounds of interest to be well separated from the hydrocarbons in the matrix to reduce the effects of quenching (16–18), which involves multiple switching valves. SCD affords great selectivity of sulfur response over carbon with low detection limits, but has a higher cost of ownership. Furthermore, this necessitates equipment specifically dedicated to sulfur detection. MS in selected ion monitoring mode can also be used; however, if there is insufficient separation between the analyte of interest and the interference there is a possibility of a false positive. For example, in selected ion monitoring mode an abundance of ions can cause peak broadening because of electrostatic repulsions within a cloud of similarly charged ions (19). Ultimately, this could offset the centroid of the sulfur species of interest or these coeluted hydrocarbons may overlap and create a false positive sulfur response (19).

Comprehensive two-dimensional gas chromatography (GC×GC) is an emerging technique that uses the separation power of two columns with different selectivity (20–22). All analytes in this method undergo separation by both columns, and the process is facilitated by a rapid re-injection from a modulator. Flow modulation was first described by Seeley (23) and Amirav (24). Basic operation of modulation includes effluent from the first column filling a collection channel within the plate, and then that effluent is rapidly injected into the second column by switching flow through the channel creating a pulsed flow. Not only does the separation power increase in GC×GC, but signals are enhanced as modulation creates sharp narrow peaks that elute off the second column.

GC×GC has been coupled to selective detection systems such as FPD and SCD to analyze sulfur species in complex matrices (25–27). Here, GC×GC is used to measure prominent sulfur odorants found in natural gas in one method. Utilization of this technique can be beneficial to separate sulfur compounds of interest from the hydrocarbon matrix as well as to identify the species added to a natural gas sample. Furthermore, signal enhancement from modulation allows detection at low levels such that these sulfur species can be monitored with flame ionization detection (FID) over an appropriate range.

Experimental

Chromatographic Conditions

A model 7890 gas chromatograph (Agilent Technologies) was used for development of this method and was equipped with two split–splitless inlets, two flame ionization detectors, and a CFT GC×GC flow modulator. Flow was delivered to the modulator by a pneumatically controlled module with the system running in constant flow mode. Raw chromatograms were collected using Chemstation software version B.04.03 (Agilent) and were then deconvoluted to produce the 3D color plots using GC Image software version 2.2 (Zoex Corporation).

A 1-mL gas sample was injected manually into the split–splitless inlet at 275 °C. The system was operated in split mode with a split ratio of 2:1. The inlet liner used was an ultra-inert liner (Agilent), in efforts to maintain minimum activity of the sulfur analyte with the liner surface. Hydrogen (Praxair) was used as the carrier gas at a rate of 1 mL/min through the first column, controlled by electronic pressure control, and 22 mL/min through the second column, delivered from the pneumatically controlled module. An oven temperature program was used that began at an initial temperature of 40 °C for 2 min, then programmed to 250 °C at a rate of 5 °C/min. The FID system was maintained at 250 °C with an additional 20 mL/min of hydrogen to the carrier flow rate, 350 mL/min of air (Praxair), and 30 mL/min of nitrogen makeup gas (Praxair). The modulation period was optimized to be 4.0 s, with a fill time of 3.70 s and an inject time of 0.30 s.

The first-dimension (1D) column had dimensions of 30 m × 0.250 mm with a film thickness of approximately 3.0 μm. The stationary phase was a base-deactivated 100% polydimethylsiloxane (PDMS) (Restek). The second-dimension (2D) column was a 5 m × 0.250 mm, 1.0-μm df Agilent VF-WAXms column.

Chemicals

The sulfur test analytes used in this study were dimethyl sulfide, isopropyl mercaptan, n-propyl mercaptan, tert-butyl mercaptan, sec-butyl mercaptan, methylethyl sulfide, and tetrahydrothiophene, which were obtained from Sigma-Aldrich. Natural gas samples were supplied by Direct Energy.

Results

GC×GC Conditions and Optimization

Column Set Selection

Many methods used to analyze sulfur compounds in natural gas suppress matrix interferences by means of selective detection. With GC×GC, separation is maximized by the use of two columns with orthogonal selectivity. Column selection is a vital component in optimizing the separation capabilities, which in this method is used to separate the sulfur compounds of interest in the first dimension and then separate these compounds from the matrix by the second-dimension column.

To separate the sulfur analytes in the first-dimension column, a nonpolar, low beta ratio (β) stationary phase was chosen because it offers advantages of inertness, good separation of low retention factor (k) solutes, and high diffusivity. Recently, in collaboration with Restek, a low β value, inert stationary phase with selectivity similar to PDMS was developed. In using a flow modulator it is advised to use a maximum column internal diameter of 0.250 mm to avoid excessive pressure pulsing. Thus, the first-dimension column was made to a β equal to 21 with a film thickness of approximately 3.0 μ;m and an internal diameter of 0.250 mm.

The first-dimension column is utilized to separate the sulfur compounds of interest. The analytes spend the most time interacting with this phase, therefore it will have the most influence in the separation. Single-dimension separations of these sulfur compounds with this column produce chromatograms with Gaussian peaks that are well resolved; among a hydrocarbon matrix such as natural gas, however, coelutions occur that make detection and identification of these sulfur compounds difficult. Figure 1 shows single-dimension chromatograms of natural gas overlaid with a gas mixture of the sulfur compounds ranging at approximately 5 ppm (v/v). These concentrations represent upper limits of these species that would be added to natural gas. While compounds like isopropyl mercaptan and tetrahydrothiophene are separated from hydrocarbons in natural gas, the remaining sulfur compounds show either full or partial coelution. To analyze these compounds in a conventional GC method, selective detection is required. However, sulfur-selective detectors only provide information on the sulfur species present, and as previously mentioned, the coelutions present could pose a problem for quantitation with MS in selective ion monitoring mode. However, in GC×GC, the second dimension is utilized to separate the sulfur analytes from the matrix and detection is provided by FID; therefore all species present can be separated, monitored, and quantified.

Figure 1: Single-dimensional chromatogram of a 1-mL natural gas injection (blue trace) overlaid with sulfur standard mix (red trace). The analytical conditions are described in the text. Modulation was deactivated. Standard peaks: 1 = dimethyl sulfide (7.22 ppm [v/v]), 2 = isopropyl mercaptan (5.67 ppm [v/v]), 3 = tert-butyl mercaptan (4.71 ppm [v/v]), 4 = n-propyl mercaptan (5.86 ppm [v/v]), 5 = methylethyl sulfide (5.87 ppm [v/v]), 6 = sec-butyl mercaptan (4.88 ppm [v/v]), 7 = tetrahydrothiophene (6.02 ppm [v/v]).

The second-dimension stationary phase was determined to be a wax phase with a film thickness of 1.0 μm (β = 63). Previous work demonstrated selectivity that retains methyl mercaptan, which is eluted near heptane (28). Thus, the selectivity of the wax stationary phase will have more retention of the sulfur analytes over the hydrocarbon matrix, effectively providing the necessary separation in the second dimension.

Modulation Optimization

The column flow rates through the first and second dimension were 1 mL/min and 22 mL/min, respectively. These values are typical using a modulator to achieve optimally modulated peaks of the effluent coming off of the first dimension (23,29). Furthermore, these rates achieve a signal enhancement as the peaks eluted out of the second-dimension column are compressed with the higher carrier velocities.

The modulation period was developed and optimized to be 4.0 s, where the modulation number was at least three, meaning that the peak eluted from the first-dimension column was modulated at least three times. This results in accurate representation of the peak, while effectively maximizing the signal intensity (30,31). A sample of the raw chromatographic sulfur peaks are displayed in Figure 2. Flush and fill times of the modulation chamber were adjusted to obtain Gaussian peaks eluted from the second-dimension column, which are essential for accurate volume determination in the color plots that are used for quantitation.

Figure 2: Raw chromatogram of modulated sulfur standard gas mix. The inset shows a magnified view of the tert-butyl mercaptan peak. The chromatographic conditions are described in the text. Standard peaks: 1 = dimethyl sulfide (0.90 ppm [v/v]), 2 = isopropyl mercaptan (0.71 ppm [v/v]), 3 = tert-butyl mercaptan (0.59 ppm [v/v]), 4 = n-propyl mercaptan (0.73 ppm [v/v]), 5 = methylethyl sulfide (0.73 ppm [v/v]), 6 = sec-butyl mercaptan (0.61 ppm [v/v]), 7 = tetrahydrothiophene (0.75 ppm [v/v]).

Figures of Merit

Calibration was performed using test analytes of each of the sulfur compounds. Linearity is observed over a range of 0.1–30 ppm (v/v), which is representative for sulfur compounds typically found in natural gas (1,2). Correlation coefficients of each test analyte are all above 0.99. Recoveries of these sulfur compounds spiked into a natural gas matrix at approximately 0.7 ppm (v/v) were all above 95%.

Figure 3: Color plot of sulfur standard gas mixture approaching detection limits. The concentrations are as described in Table I, and the chromatographic conditions are as described in the text.

Detection limits were obtained by measuring the signal provided by the tallest peak in the raw chromatographic data and then ratioed to the peak-to-peak (p-p) noise. At the 100 ppb (v/v) level, the signal-to-noise ratio (p-p) was calculated to be approximately 10–27 for each sulfur compound. Table I displays the minimum detectable concentration for each of the sulfur compounds tested. A color plot of this mixture is shown in Figure 3. The well formed peaks seen in this color plot were achieved by deconvoluting the raw chromatograms (Figure 2), after which volumes can be determined for quantitation. Also shown in Table I are the minimum detectable concentrations and signal-to-noise ratios obtained without modulation demonstrating minimum detectable concentrations approximately six times larger at lower signal-to-noise values. These results demonstrate the improved detectability when using modulation, which leads to the possibility of detecting sulfur species at the lower limits required by federal regulations.

Table I: Limit of detection for the prominent sulfur compounds in natural gas with and without modulation

Reproducibility of seven successive injections at concentrations ranging from 475 to 730 ppb (v/v) is displayed in Table II. As shown, the percent relative standard deviation (%RSD) spans from 1.7% to 5.4%, with an overall average of 3.7%. Table II shows %RSD values determined on retention times in the first and second dimension, which demonstrate excellent reproducibility with %RSD values less than 1% in all analytes tested over a period of four weeks. Overall, the chromatographic conditions are reliable and robust and retention time marking can be used to identify these sulfur compounds when present in natural gas.

Table II: Reproducibility of peak volume, first- and second-dimension retention time on seven successive injections of a standard mixture of the prominent sulfur compounds in natural gas

Figure 4 shows these sulfur compounds spiked into a natural gas matrix at concentrations of about 0.6 ppm (v/v). Target analytes are well separated from the hydrocarbon matrix where coelutions occurred in single-dimension analysis (Figure 1). Natural gas contains hydrocarbons beyond hexane, and the first sulfur species of interest are eluted near pentane. This method may then be extended to analyze odorants found in propane that is distributed and used in residential households, as propane contains similar odorants used for identification of its presence.

Figure 4: Sulfur standard gas mixture spiked into a natural gas matrix. Concentrations of each sulfur component: dimethyl sulfide (0.90 ppm [v/v]), isopropyl mercaptan (0.71 ppm [v/v]), tert-butyl mercaptan (0.59 ppm [v/v]), n-propyl mercaptan (0.73 ppm [v/v]), methylethyl sulfide (0.73 ppm [v/v]), sec-butyl mercaptan (0.61 ppm [v/v]), and tetrahydrothiophene (0.75 ppm [v/v]). The chromatographic conditions are described in the text.

Sulfur Compounds in Natural Gas

A residential sample of natural gas was obtained by sampling a natural gas line into a Tedlar bag and injecting it into the GC×GC system. The resulting chromatogram is shown in Figure 5. Retention time identifies a blend of tert-butyl mercaptan and methylethyl sulfide in natural gas. The concentrations of the species (based on their peak volumes) were found to be 2 and 0.7 ppm (v/v), respectively by external calibration. This represents a blend of approximately 75:25 tert-butyl mercaptan to methylethyl sulfide, which is a common combination (1,2).

Figure 5: Natural gas sample taken from a residential source with identified sulfur components labeled.

The ease of determining the sulfur compounds in this matrix should be further noted. The sample was taken right from the natural gas line and directly injected into the GC×GC system, and the use of retention time templates allowed for the rapid identification of the sulfur compounds present. This method can be used to analyze sulfur compounds in natural gas from any provider as it simultaneously detects all prominent sulfur species. Furthermore, not only does this technique allow for detection of the sulfur compounds present, but also all of the hydrocarbons in the same analysis. This method may be very beneficial as a complementary technique to using a selective detection method such as SCD or MS.

Conclusion

A method has been developed to analyze key odorants found in commercially available natural gas by GC×GC with pneumatically controlled modulation. This technique uses the separation power of GC×GC to separate sulfur compounds from each other as well as from a hydrocarbon matrix. Furthermore, because this technique uses a modulator, increased signal detectability is achieved that allows for quantitation at levels as low as 100 ppb (v/v) of each sulfur compound. This method allows for direct injection of a natural gas sample for identification and quantitation of odorants added.

Acknowledgments

The authors would like to thank Jaap de Zeeuw (Restek) for his contribution of the low beta ratio, first-dimension column used in this method and the members of the Analytical Technology Center, particularly Vicki Carter, for their help and support. Funding for this project was made possible from contributions by Alberta Innovates Technology Futures and the Dow Analytical Technology Center Technology Renewal and Development 2012 Program.

References

(1) M.J. Usher, in Odor Fade – Possible Causes and Remedies (Elf Atochem North America, Inc., Philadelphia, Pennsylvania, 1999), pp. 1–12.

(2) D. Tenkrat, T. Hlincik, and O. Prokes, in Natural Gas, P. Potocnik, Ed. (Sciyo/InTech, Rijeka, Croatia, 2010), pp. 87–103.

(3) F.A. Fazzalari, Ed. Compilation of Odor and Taste Threshold Data (ASTM Data Series DS 48A, West Conshohocken, Pennsylvania, 1978), pp. 1–508.

(4) P.D. Wehnert, in Determination of Proper Odorization of Natural Gas (Proceedings of the International School of Hydrocarbon Measurement, 2002), pp. 747–750.

(5) A.V. Litvinov, P.O. Unchenko, and I.N. Nikolaev, Meas. Tech.50, 548–550 (2007).

(6) K.-H. Kim, Sensors11, 1405–1417 (2011).

(7) R.L. Firor and B.D. Quimby, Hydrocarbon Process.82, 79–81 (2003).

(8) L. Huber and H. Obbens, J. Chromatogr. 349, 465–468 (1985).

(9) J. Macak, J. Kubat, V. Dobal, and J. Mizera, J. Chromatogr. 286, 69–78 (1984).

(10) D.C. Pearson, J. Chromatogr. Sci. 14, 1521–1528 (1976).

(11) R. Shearer, D. O'Neal, R. Rois, and M. Baker, J. Chromatogr. Sci.28, 24–28 (1990).

(12) H.P. Tuan, H.-G. Janssen, and C.A. Cramers, J. High Resol. Chromatogr.18, 333–342 (1995).

(13) H.P. Tuan, H.-G. Janssen, E.M. Kuiper-van Loo, and H. Vlap, J. High Resol. Chromatogr. 18, 525–534 (1995).

(14) S.-W. Myung, S. Huh, J. Kim, Y. Kim, M. Kim, Y. Kim, W. Kim, and B. Kim, J. Chromatogr. A791, 367–370, (1997).

(15) J. Luong, R. Gras, R. Van Meulebroeck, F. Sutherland, and H. Cortes, J. Chromatogr. Sci.44, 276–282 (2006).

(16) L. Kalontarov, H. Jing, A. Amirav, and S. Cheskis, J. Chromatogr. A696, 245–256 (1995).

(17) W.A. Aue and X.-Y. Sun, J. Chromatogr. A 641, 291–299 (1993).

(18) M. Dressler, J. Chromatogr. A270, 145–150 (1983).

(19) R.E. Pedder, Application Note RA_2011A, Extrel CMS L.P., Pittsburgh, Pennsylvania (2002).

(20) P.J. Marriott, S.-T. Chin, B. Maikhunthod, H.G. Schmarr, and S. Bieri, TrAC34, 1–21 (2012).

(21) M. Adahchour, J. Beens, R.J.J. Vreuls, and U.A.Th. Brinkman, TrAC25, 438–454 (2006).

(22) M. Adahchour, J. Beens, R.J.J. Vreuls, and U.A.Th. Brinkman, TrAC25, 726–741 (2006).

(23) J.V. Seeley, N.J. Micyuc, J.D. McCurry, and S.K. Seeley, American Laboratory38, 24–26 (2006).

(24) M. Poliak, M, Kochman, and A. Amirav, J. Chromatogr. A1186, 189–195 (2008).

(25) S.-T. Chin, Z.-Y. Wu, P.D. Morrison, and P.J. Marriott, Anal. Methods2, 243–253 (2010).

(26) J. Blomberg, T. Riemersma, M. van Zuijlen, and H. Chaasbani, J. Chromatogr. A1050, 77–84 (2004).

(27) R. Hua, J. Wang, H. Kong, J. Liu, X. Lu, and G. Xu, J. Sep. Sci.27, 691–698 (2004).

(28) R. Gras, J. Luong, V. Carter, L. Sieben, and H. Cortes, J. Chromatogr. A1216, 2776–2782 (2009).

(29) P.McA. Harvey and R.A. Shellie, J. Chromatogr. A1218, 3153–3158 (2011).

(30) W. Khummueng, J. Harynuk, and P.J. Marriott, Anal. Chem. 78, 4578–4587 (2006).

(31) P.J. Marriott and W. Khummueng, LCGC Europe22, 1–12 (2009).

T. Hayward, R. Gras, and J. Luong are with Dow Chemical Canada, in the Analytical Technology Center in Fort Saskatchewan, Alberta, Canada. Direct correspondence to: thayward@dow.com