Traceable Reference Gas Mixtures for Sulfur-Free Natural Gas

Jun 2, 2014 - Andrew S. Brown*†, Adriaan M. H. van der Veen‡, Karine Arrhenius§, Michael L. Downey†, Daniel Kühnemuth§, Jianrong Li‡, Hugo ...
0 downloads 0 Views 766KB Size
Article pubs.acs.org/ac

Traceable Reference Gas Mixtures for Sulfur-Free Natural Gas Odorants Andrew S. Brown,*,† Adriaan M. H. van der Veen,‡ Karine Arrhenius,§ Michael L. Downey,† Daniel Kühnemuth,§ Jianrong Li,‡ Hugo Ent,‡ and Lucy P. Culleton† †

Analytical Science Division, National Physical Laboratory, Teddington, Middlesex TW11 0LW, United Kingdom Van Swinden Laboratorium, Thijsseweg 11, 2629 JA Delft, The Netherlands § SP Technical Research Institute of Sweden, Box 857, SE-501 15 Borås, Sweden ‡

ABSTRACT: The first reference gas mixtures of sulfur-free natural gas odorants that are traceable to the International System of Units (SI) have been produced and their compositions validated. These mixtures, which contain methyl acrylate and ethyl acrylate at amount fractions between 1.1 and 2.1 μmol mol−1, can be used to underpin measurements of sulfur-free odorants, which are increasingly being used to odorize natural gas in transmission networks as they have less harmful properties than traditional sulfur-containing odorants. The reference gas mixtures produced have been shown to be stable in passivated aluminum cylinders for at least 8 months and have been validated (to within 6% or less) by interlaboratory measurements at three National Measurement Institutes. The stability of methyl acrylate and ethyl acrylate in gas sampling bags has been investigated, and the challenges of analyzing 2-ethyl-3-methylpyrazine, which is used as a stabilizer in sulfur-free odorants, are also briefly discussed.

N

Sulfur-containing compounds have been used for odorizing natural gas for many years because their characteristics4 include that they are stable in natural gas, usable at low temperatures, do not leave residues or solid deposits, and have an odor that is strong at very low concentrations, not confusable with other frequently occurring odors, the same at different dilutions of natural gas with air. However, in addition to the risks to health and processing equipment outlined above, the combustion of natural gas odorized with sulfur-containing compounds also generates sulfur dioxide, which is a harmful environmental pollutant. In order to eliminate these potentially damaging effects of sulfur-containing compounds, sulfur-free odorants are now being used in a number of natural gas networks, mainly in Germany, but also in parts of Austria and China. Indeed, in 2013 the international standard for the odorization of natural gas, ISO 13734,4 was updated to incorporate sulfur-free odorants for the first time. Sulfur-free odorants also have potential applications as odorants for fuel cells powered by natural gas or LPG5 or hydrogen,6 where even the presence of trace levels of sulfur-containing compounds can cause permanent degradation to fuel cell performance. The most-common sulfur-free odorant in use in gas transmission networks is Gasodor S-Free,7 which contains methyl acrylate (MA), ethyl acrylate (EA), and 2-ethyl-3-

atural gas is used throughout the world to provide heating and power to homes and industry, as feedstock in industrial processes, and to generate electricity. Other major uses of natural gas include as a vehicle fuel and as a starting material for the production of hydrogen. The key role played by natural gas as an energy vector is outlined by the huge volume of natural gas consumed: in 2012, 3.31 Tm3 (trillion cubic meters) of natural gas was used worldwide, 1.08 Tm3 of which was used in Europe and 0.87 Tm3 in North America. This worldwide consumption represents a 31% increase in the last 10 years and a 163% increase in the last 30 years.1 As natural gas is transported via extensive national and international pipeline transmission networks, it is essential that any potential catastrophic leaks of gas can be detected by the general public. Although natural gas from some sources does contain naturally occurring odorous sulfur-containing compounds (such as hydrogen sulfide), the gas is typically scrubbed of these compounds and then artificially odorized with sulfurcontaining compounds (usually thiols or tetrahydrothiophene) before being injected into gas networks. This ensures that the gas contains defined levels of odorants so that it conforms to specifications for gas quality. As an example, natural gas in the U.K. national transmission system must contain no more than 5 mg m−3 of hydrogen sulfide and a total of no more than 50 mg m−3 of sulfur-containing compounds.2 Similar limits exist in other European countries with the maximum total permitted mass concentration of sulfur-containing compounds in the member states of the European Union ranging from 10 mg m−3 (Sweden) to 150 mg m−3 (Italy and Belgium).3 © 2014 American Chemical Society

Received: April 25, 2014 Accepted: June 2, 2014 Published: June 2, 2014 6695

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

Figure 1. Preparation strategy used to prepare mixtures NPL B1−B4.

mixtures are prepared by weighing gaseous or liquid components into cylinders) being the most accurate and most widely used method to ensure traceability for the measurement of gases.14 The preparation of gas mixtures by gravimetry is sufficiently mature that is has been the subject of an international standard, ISO 6142, for more than a decade.15 (Reference gas mixtures are also often prepared by dynamic volumetric methods, as described in the ISO 6145 series of standards.16) This paper summarizes the work performed at three National Metrology Institutes: NPL (National Physical Laboratory, U.K.), VSL (Van Swinden Laboratorium, The Netherlands), and SP (Technical Research Institute of Sweden) to develop and validate the first fully traceable reference gas mixtures of sulfur-free odorants in cylinders. The composition of these mixtures, which have been produced using existing methods in accordance with ISO 6142, are traceable to the SI through standards of mass and a purity assessment of the materials used for gas mixture preparation.15 NPL and VSL prepared the gas mixtures and performed initial validation tests using gas chromatography with mass spectrometer detection (GC/MS) and flame ionization detection (GC-FID). The mixtures were then distributed to the other laboratories for further validation and comparison, with analysis at SP taking place by thermal desorption (TD) gas chromatography (specifically TD-GC/MS and TD-GC-FID) calibrated using sorbent tubes spiked with known amounts of the sulfur-free odorants, either by direct injection of the liquid components onto the sorbent tubes or by transferring gas mixtures of the odorants from sample bags onto

methylpyrazine (EMP, added as a stabilizer) at mass fractions of 37.4% MA, 60.1% EA, and 2.5% EMP. Full details of the research and testing activities performed before the product was introduced into gas transmission networks are available in the literature.8 The odorization of natural gas in the German transmission network is regulated by the DVGW standard G280-1, 9 which sets a minimum total odorant mass concentration of Gasodor S-Free of 8.8 mg m−3 (at ambient temperature and pressure). In practice, a mass concentration between 10 and 12 mg m−3 is used. Commercial instruments are available which can analyze sulfur-free odorants in natural gas using gas chromatography (GC)10 or ion mobility spectrometry11. Details of a new, rapid GC method using a micromachined GC with differential mobility spectrometry that can measure MA and EA at volume fractions of 0.5−50 μL L−1 in natural gas with a precision of better than 5% relative have recently been published.12 However, in order to ensure that measurements of sulfur-free odorants are reliable and robust, it is crucial that the instruments used to perform the analysis are calibrated using reference gas mixtures with compositions which are both accurate and traceable to the International System of Units (SI). Although traceable reference gas mixtures of sulfurcontaining odorants have been available for some time and have been validated by comparison between National Measurement Institutes,13 no such capability has previously been established for standards of sulfur-free odorants. Reference gas mixtures can be produced by a variety of methods, with the gravimetric preparation method (where 6696

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

Figure 2. Example GC-FID chromatogram obtained from the analysis of mixture NPL A4.

Liquide). Volumes of 100−200 mL of these mixtures were then immediately transferred onto Tenax TA tubes (Supelco), which were used for calibrating the TD-GC system. In addition to the reference gas mixtures prepared at VSL from the individual components (MA, EA, and EMP), an additional gas mixture was prepared directly from the commercial Gasodor S-Free liquid, which is a liquid mixture of these three components. The method used to prepare this gas mixture is as described above, but the composition and purity of the Gasodor S-Free liquid were not analyzed prior to the preparation of mixture VSL 3, meaning that the calculated gravimetric amount fractions of the components in this mixture do not have metrological traceability (unlike those mixtures prepared from the individual components). The composition of the liquid mixtures as specified by the producer of Gasodor SFree was used to determine the amount fractions of MA, EA, and EMP in the gas mixture. GC Analysis. Analysis at NPL was performed using an Agilent 7890A GC equipped with 5975C MS and FID detectors. Separation of the components was achieved using a DB-624 column (75 m length, 0.53 mm internal diameter, 3 μm film thickness) (Agilent). The temperature of the column was held at 60 °C for 4.5 min, increased to 160 °C at 40 °C min−1, then held at 160 °C for 3 min. The MS detector was operated in selected ion monitoring (SIM) mode using ions of m/z = 55 and 85 for MA, 55 and 99 for EA, and 121 and 122 for EMP. Samples were injected directly from the gas mixtures into the GC at a flow rate of approximately 30 mL min−1. An example chromatogram is shown in Figure 2; the limits of detection of this method are approximately 8 nmol mol−1 MA for EA and 5 nmol mol−1 for EMP. When performing the analysis of gas mixtures obtained from other laboratories, the GCs were calibrated using in-house reference gas mixtures prepared in the manner described above. Analysis at VSL was performed using two different GCs: (1) An Agilent 6890N GC equipped with an FID detector. Separation of the components was achieved using a WCot fused-silica CP Sil 5CB column (60 m length, 0.32 mm internal diameter, 0.25 μm film thickness) (Varian). The temperature of the column was held at 40 °C for 4 min, increased to 195 °C at 10 °C min−1, then held at 195 °C for 5 min. Samples were injected directly from the gas mixtures into the GC at a flow rate of approximately 60 mL min−1. (2) An Agilent 6890N GC

the sorbent tubes. The preparation, initial validation, and stability of the reference gas mixtures are discussed in this paper, focusing on the work performed at NPL. The results of the interlaboratory analyses are then presented.



EXPERIMENTAL SECTION Reference Gas Mixtures. All reference gas mixtures were prepared gravimetrically in accordance with ISO 6142.15 The reference gas mixtures prepared at NPL were produced from methyl acrylate (MA) (Fluka), ethyl acrylate (EA) (Fluka), 2ethyl-3-methylpyrazine (EMP) (SAFC), and methane (CK Gases). The purity of each was determined by GC before use. Gas mixtures were prepared in 10 L aluminum cylinders (Luxfer) with the internal surfaces treated using the “Spectraseal” passivation technique (BOC) in order to minimize the adsorption of compounds to the cylinder walls. Prior to filling, the cylinders were evacuated for 24 h to a pressure of approximately 1.5 × 10 −7 mbar using a turbomolecular pump (Leybold Vacuum). An example preparation strategy is shown in Figure 1, in which the components are added by one of two methods: direct transfer of each individual component to the cylinder being filled (for component masses of more than approximately 12 g), or transfer via an intermediate stainless steel vessel (for smaller masses). The reference gas mixtures prepared at VSL were produced by a similar method to those at NPL, but were prepared in 5 L aluminum cylinders (Luxfer) with the internal surfaces treated using the “Aculife IV” passivation technique (Scott Gases). The individual components (MA, EA, and EMP) were introduced into the cylinder using a syringe;17 the methane was added by direct transfer of the gas into the cylinder. Calibration Solutions and Gas Mixtures. SP prepared five solutions of MA (Fluka), EA (Fluka), and EMP (SAFC) in diethyl ether (Fisher) containing a range of total odorant mass concentrations between 200 and 800 mg mL−1. Volumes of 2 μL of each of the solutions were injected on Tenax TA sorbent tubes (Supelco), which were used to calibrate the TD-GC system. Calibration was also performed by transferring 2.0−2.9 μg of a 5.4 μg μL−1 solution of the odorants into FlexFilm sample bags (SKC) and adding 2.4 L of synthetic biogas consisting of approximately 390 mmol mol−1 carbon dioxide, 600 mmol mol−1 methane, and 10 mmol mol−1 nitrogen (Air 6697

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

Table 1. Eight Reference Gas Mixtures of Sulfur-Free Odorants Produced at NPLa amount fraction/(μmol mol−1) mixture NPL A1

mixture NPL A2

mixture NPL A3

mixture NPL A4

MA EA EMP

104.12 ± 0.17 − − mixture NPL B1

− 99.08 ± 0.14 − mixture NPL B2

− − 14.90 ± 0.12 mixture NPL B3

1.251 ± 0.004 2.094 ± 0.004 0.0544 ± 0.0006 mixture NPL B4

MA EA EMP

100.09 ± 0.17 − −

− 99.88 ± 0.14 −

− − 15.00 ± 0.13

1.110 ± 0.004 1.639 ± 0.004 0.0519 ± 0.0006

a

The amount fractions have been determined by gravimetry. All uncertainties are expanded (k = 2) uncertainties and are the gravimetric uncertainties only (so do not include any contribution for uncertainty due to stability of the mixture). The values stated for EMP are indicative values only due to the difficulties with the preparation and analysis mixtures of this compound discussed in the text.

prepared at NPL following the preparation scheme outlined in Figure 1 (which shows the amount fractions of the components in the mixtures NPL B1−B4). Mixtures NPL B1−B4 (referred to hereafter as the NPL “B” mixtures) were prepared approximately 8 months after mixtures NPL A1−A4 (the NPL “A” mixtures). Performing a robust and traceable validation of the NPL “A” mixtures before the NPL “B” mixtures were produced was problematic as no other traceable mixtures containing these components existed for them to be compared against. The potential use of FID carbon response factors (i.e., the GC response divided by the gravimetric amount fraction divided by the number of carbon atoms) to validate the three components against each other was ruled out as a possibility as the expected response factors for each of the three components were not known and were very unlikely to be similar. Also, even if FID carbon response factors were able to be used, this approach is not fully traceable to the SI. The validation of all of the NPL “A” mixtures was therefore performed after the preparation of the NPL “B” mixtures as part of the stability tests now discussed. Stability of the Reference Gas Mixtures. Assessment of the stability of reference gas mixtures in cylinders is essential to quantify any changes in the composition of a gas mixture with time. Reference materials are supplied with a “shelf life” during which the certified amount fraction of each component in the mixture must be maintained (within the stated uncertainty).20 The stability of the reference gas mixtures prepared at NPL were tested by comparing the FID response factors (i.e., the GC response divided by the gravimetric amount fraction) of each component in the NPL “A” mixtures and the NPL “B” mixtures. Repeated alternative injections of the related NPL “A” and NPL “B” mixtures were performed in order to minimize the effects of detector drift. The results from these stability tests are shown in Figure 3, where the ratios of the FID response factors of the NPL “B” mixtures relative to the NPL “A” mixtures are plotted. The small size of the error bars for MA and EA is an indication of the excellent repeatability of the GC method, which is less than 0.2% for MA and EA, even in the multicomponent mixtures, where the amount fractions of these compounds are only between 1.1 and 2.1 μmol mol−1. Figure 3 shows the measured FID response factors for both MA and EA in the binary mixtures NPL A1, A2, B1, and B2 (which contain these components at amount fractions of approximately 100 μmol mol−1) and the multicomponent mixtures NPL A4 and B4 (between 1.1 and 2.1 μmol mol−1) agree within the experimental uncertainties. This indicates that

equipped with an atomic emission detector (AED) using the same column as the GC-FID. The temperature of the column was held at 40 °C for 4 min, increased to 80 °C at 10 °C min−1, held at 80 °C for 2 min, increased to 200 °C at 30 °C min−1, then held at 200 °C for 5 min. The sample flow rate was also 60 mL min−1. Analysis at SP was performed using Agilent 6890 GC with 5975C MS and FID detectors and a PerkinElmer Turbomatrix 650 thermal desorption unit. Separation of the components was achieved using an BPX5 column (50 m length, 0.32 mm internal diameter, 1 μm film thickness) (SGE). The temperature of the column increased from 30 to 80 °C at 2 °C min−1, then to 300 °C at 15 °C min−1 before holding at 300 °C for 10 min. The MS detector was operated in selected ion monitoring mode (SIM) using ions of m/z = 55, 55, and 121 for MA, EA, and EMP, respectively. Calculation of Uncertainties. The uncertainty in the amount fraction of each component in the gas mixtures was calculated in accordance with ISO 6142.15 The components of uncertainty from the weighing process, purity analyses and the molecular masses of the compounds in the mixture were combined in accordance with the Guide to the expression of uncertainty in measurement (GUM).18 The uncertainties in the results from the GC analyses were determined (also in accordance with the GUM) by combining the uncertainty contributions from the calibration standards and the repeatability of the GC analyses. A more detailed discussion of uncertainties in chromatographic analyses can be found in ref 19.



RESULTS AND DISCUSSION This study focused on the preparation of multicomponent reference gas mixtures with a total odorant mass concentration of 11 mg m−3, as this is the middle of the range of mass concentrations typically found in natural gas networks in order to ensure that the odorization requirements of ref 9 are met. Using a pressure of 1 bar and a temperature of 20 °C, this total mass concentration of 11 mg m−3 equates to amount fractions (mass concentrations in brackets) of each component of 1.15 μmol mol−1 (4.11 mg m−3) MA, 1.59 μmol mol−1 (6.61 mg m−3) EA, and 54 nmol mol−1 (0.28 mg m−3) EMP. The results presented in this section will focus on EA and MA as these are the compounds added to Gasodor S-Free specifically to act as odorants. The analysis of EMP, which acts as a stabilizer in the Gasodor S-Free mixture, is significantly more challenging and is also discussed briefly. Preparation and Validation of the Reference Gas Mixtures. Eight gas mixtures (shown in Table 1) were 6698

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

Figure 3. Results of the stability tests of the binary mixtures NPL A1− A3 and NPL B1−B3 (a) and the multicomponent mixtures NPL A4 and B4 (b). MA is represented by blue circles, EA by green squares, and EMP by red triangles. Note the different y-axis scales. The error bars indicate the expanded (k = 2) uncertainties of the measurements.

Figure 4. Results of stability tests of mixtures of sulfur-free odorants in synthetic biogas with total mass concentrations of MA and EA of 4.4 mg m−3 (light purple filled squares), 5.4 mg m−3 (purple hollow triangles), and 6.3 mg m−3 (dark purple filled circles). The data were obtained using GC/MS (in SIM mode) and are plotted as the measured mass concentration as a percentage of the “known” mass concentration of the original mixture calculated from the preparation methods. The expanded (k = 2) uncertainty in each point is approximately 8%. The dashed line indicates a recovery of 100%.

these two components are stable for a period of more than 8 months even at very low micromole per mole levels. The results for EMP show much larger differences between the NPL “A” and “B” mixtures (2.2% for the binary mixtures and 17% for the multicomponent mixtures). Although these differences are nearly accounted for by the measurement uncertainties, they are likely to be statistically significant and result either from problems experienced when preparing and analyzing mixtures of EMP or losses to the internal walls of cylinders and transfer lines. These issues are discussed in more detail below. The results also provide a strong evidence of the validation of the composition of the mixtures (i.e., the composition is that determined by the gravimetric preparation process). The possibility of the two sets of mixtures experiencing similar losses of the components immediately after preparation (e.g., by adsorption onto the cylinder walls) does still remain after these tests, but this can be examined by comparing the results from mixtures prepared in different cylinders. This is discussed further later in this section with the results from the interlaboratory analyses. Stability Tests of MA and EA in Sample Bags. The stability of MA and EA in FlexFilm sample bags was investigated at SP. Although the work discussed above shows that both of these compounds are stable for more than 8 months in passivated aluminum cylinders, studies of their stability in sample bags are of interest. Sample bags may be used to prepare gaseous mixtures of known concentrations for calibrating laboratory instruments, so an indication of the shortterm stability of MA and EA in these bags is required (it is unlikely that such gases will be stored for more than a few days). Sample bags may also be used to take samples of gases in the field, although for natural gas applications, this is more likely to be carried out using sampling cylinders. Figure 4 shows results from the GC/MS (SIM mode) analysis of three mixtures of sulfur free-odorants in synthetic biogas (see the Experimental Section) in FlexFilm bags. The total mass concentrations of MA and EA in the mixtures were 4.4, 5.4, and 6.3 mg m−3. The measurements of the gas mixtures performed on the same day that they were produced (t = 0 days) show recoveries of nearly 100%, implying that any immediate losses of these compounds in sampling bags is

negligible. Significant losses are, however, shown for each of the three subsequent measurements, with the relative magnitude of these losses typically being independent of the initial concentration of the compound. The losses, which are likely to be due to the compounds adsorbing onto the internal surfaces of the bags, exceed 50% after storage in the bags for 2 weeks. These results show that sample bags should only be used when the gas is to be used immediately after being introduced into the bag, which may be the case when bags are used to prepare gas mixtures of known concentrations for calibrating laboratory instruments. However, as MA and EA show significant losses when stored in these bags for a period of only a few days, gases should not be stored in the bags, and the bags are not suitable for taking samples of gases in the field this time period is much shorter than the usual time period between field sampling and laboratory analysis (including transportation). Instead, the compounds should be sampled either into passivated sampling cylinders or directly onto sorbent tubes if TD-GC analysis is to be used−additional tests at SP have shown that these compounds are stable for a period of up to 2 weeks when sampled onto Tenax TA sorbent tubes. (Recoveries were reported for MA of 94% after 2 days and 99% after 2 weeksboth of these results are consistent with no decay of MA being experienced within the expanded uncertainty of the results, which is approximately 8% relative.) These results also imply that the best strategy for calibrating a TD-GC system for analysis is to inject solutions of known concentrations of these compounds directly onto sorbent tubes. However, gas mixtures in bags may be used to dose tubes if the dosing takes place immediately after the gas mixture has been made. An alternative approach would be to “spike” a known amount of a tracer compound (such as an isotopic analogue of the odorants) to the sample immediately after it is taken. As it is expected that the decay rate of a isotopic analogue used as a tracer compound will be the same as the analyte, measurement 6699

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

Table 2. Four Gas Mixtures of Sulfur-Free Odorants Analyzed at Other Laboratoriesa analyzed by prepared from MA amount fraction/(μmol mol−1) EA amount fraction/(μmol mol−1) EMP amount fraction/(μmol mol−1)

mixture NPL B4

mixture VSL 1

mixture VSL 2

mixture VSL 3

SP only individual components 1.110 ± 0.004 1.639 ± 0.004 0.0519 ± 0.0006

NPL, then SP individual components 1.1748 ± 0.0038 1.6066 ± 0.0051 0.0552 ± 0.0004

SP only individual components 1.5161 ± 0.0049 2.0733 ± 0.0066 0.0713 ± 0.0005

NPL, then SP Gasodor S-Free liquid 1.163 ± 0.020 1.607 ± 0.018 0.0548 ± 0.0013

a

All uncertainties are expanded (k = 2) uncertainties. The values stated for EMP are indicative values only due to the difficulties with the preparation and analysis mixtures of this compound discussed in the text.

The results from the NPL analysis of MA in the VSL reference gas mixture (VSL 1) show similar absolute differences from the gravimetric amount fraction as the results from SP, but the smaller analytical uncertainties mean that the difference observed from the gravimetric amount fraction for MA (approximately 6%) is significant. The difference for EA is, however, again within the measurement uncertainty. Further work would be needed to investigate this discrepancy for MA in this mixture. It is, however, very unlikely that the discrepancy is due to a bias in the preparation of the mixtures at either lab as it has been shown that mixtures produced by VSL and NPL give very similar results when measured at SP. The stability of the mixtures prepared at VSL can also be ruled out as the source of the discrepancy, as the mixtures were analyzed at NPL before being analyzed at SP and it is the NPL results which report measured amount fractions lower than the gravimetric values. It also notable that the results from the NPL analysis of both MA and EA in the mixture prepared at VSL from the Gasodor S-Free liquid (VSL 3) show excellent agreement with the gravimetric amount fractions. These results give confidence that the analytical method used is robust, even if the gravimetric amount fractions in this mixture are not metrologically traceable as the composition or purity of the Gasodor S-Free liquid were not analyzed prior to the preparation of mixture VSL 3. Aside from the measurements of MA in one of the mixtures, the results of these interlaboratory analyses show generally good agreement between the three laboratories. The measurement uncertainties are large, but not excessively so for novel mixtures of challenging components at low amount fractions (1.1−2.1 μmol mol−1), and these uncertainties could be improved with further development work. These results also give further evidence to support the validation of the reference gas mixtures produced at NPL and VSL. It has already been shown that the mixtures are stable for many months, so as discussed above, the only remaining possible effect that may prevent the gravimetric amount fractions being reproduced by analysis are losses immediately after the components had been added to the cylinder. These relative losses may not be of the same magnitude in the cylinders used by NPL and VSL as the volumes (10 and 5 L, respectively) and the internal surface passivations of the cylinders used by the two laboratories are different. Measurement of 2-Ethyl-3-methylpyrazine. The results obtained in this work for EMP show much larger uncertainties and much less comparability than those for the two acrylate compounds. The measurement of this compound in sulfur-free odorants is, however, not essential as it is added to the mixture only as a stabilizer, but the challenges of measuring EMP is worthy of brief discussion. The chemical nature of EMP makes it very prone to adsorption onto the internal surfaces of the pipework used to

of the concentration of this tracer compound allows a correction factor to be calculated and applied. Interlaboratory Analysis of the Reference Gas Mixtures. The three reference gas mixtures prepared at NPL and VSL that were subsequently sent for measurement at other laboratories (SP and NPL) are shown in Table 2 (as mixtures NPL B4, VSL 1, and VSL 2), along with the one gas mixture prepared from the Gasodor S-Free liquid (VSL 3). The results from these measurements of MA and EA in these mixtures are shown in Figure 5.

Figure 5. Results from the interlaboratory analysis of MA (blue circles) and EA (green squares) in three multicomponent reference gas mixtures. The filled circles and squares indicate mixtures prepared from the individual components, the hollow circles and squares indicate mixtures prepared from the Gasodor S-Free liquid. Analysis was performed using GC/MS (in SIM mode) at the laboratories shown in square brackets. The error bars indicate the expanded (k = 2) uncertainty in the measurement, which is a combination of the repeatability of the GC analysis and the uncertainties in amount fractions of the components in the reference gas mixtures and calibration gases used by the laboratories.

The results in general show a good level of comparability, especially for the measurements performed at SP, which all agree with the gravimetric amount fractions of the mixtures produced at NPL and VSL within the measurement uncertainty. (The relatively large measurement uncertainty on these results is mainly due to the uncertainty in the procedure used to calibrate the TD-GC system.) The results for EA reproduce the gravimetric values slightly better than those for MA, but the difference between the results for the two components is negligible considering the measurement uncertainties. It is also interesting to note that no significant differences are observed between the results of the analyses of the mixtures produced at NPL and those produced at VSL. 6700

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

concentrations since these will have been determined using SI-traceable calibration gas mixtures.

prepare the gas mixtures and to sample the mixtures into the GC, even if passivated pipework is used. Absorption of the compound onto internal surfaces of the GC system (including the transfer valves) is also a major issue, especially at the very low amounts fractions (approximately 55 nmol mol−1) in the multicomponent mixtures prepared for this work. In one experiment performed as part of this work, traces of EMP from a previously sampled gas mixture were still being observed by GC-FID even after 15 measurements of pure nitrogen. Further work, perhaps including the development of new passivation technologies, would be needed to improve the analytical capabilities for this highly challenging compound.



AUTHOR INFORMATION

Corresponding Author

*Phone: +44 20 8943 6831. Fax: +44 20 8614 0448. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors acknowledge the funding of this work by the EURAMET (The European Association of National Metrology Institutes) and the European Commission as part of the European Metrology Research Programme. The NPL authors also acknowledge the funding received from the U.K. Government’s National Measurement Office from their Chemistry and Biology Knowledge Base Programme. The VSL authors acknowledge the funding received through the Measurement Standards contract with the Dutch Ministry of Economic Affairs. The Gasodor S-Free liquid used in the work was kindly provided by Symrise.



CONCLUSIONS Novel SI-traceable reference gas mixtures containing the three compounds in the most widely used commercial sulfur-free odorant mixture (methyl acrylate, ethyl acrylate, and 2-ethyl-3methylpyrazine) have successfully been produced and validated. The amount fractions of MA and EA in these mixtures were validated (to within 6% or less) by comparing a range of mixtures produced at different times, and by interlaboratory measurements performed at three National Measurement Institutes. Despite MA and EA being present in these mixtures at low amount fractions (between 1.1 and 2.1 μmol mol−1), good comparability between the mixtures has been demonstrated by the use of GC methods which can measure these compounds with excellent repeatability (typically below 0.2% for GC-FID measurements). Mixtures containing MA and EA at these low amount fractions have been shown to be stable in passivated aluminum cylinders for at least 8 months, meaning that these reference gas mixtures are suitable for long-term use as calibration gases for instruments performing the measurements of sulfur-free odorants in the field. Although longer-term stability tests have not yet been performed, it is likely that these mixtures will remain stable for many years, as the majority of reactive components that experience losses in gas mixtures begin to exhibit these losses immediately after the mixture has been prepared. MA and EA have, however, been found to be unstable in gas sampling bags, which is likely to be a result of the compounds adsorbing onto the internal surfaces of the bags. The use of bags to sample gases containing these components is therefore not recommended unless the gases are immediately sampled onto sorbent tubes, or the sample is “spiked” with a tracer compound. Performing accurate measurement of the third component in the mixtures, 2-ethyl-3-methylpyrazine, was found to be highly challenging, although this should not come as a surprise considering the highly adsorptive nature of the compound and the very low amount fraction (approximately 55 nmol mol−1) in the multicomponent mixtures produced for this work. It is proposed that more reliable measurements of EMP will be able to be made if improved passivation chemistries are developed for the internal surfaces of gas cylinders and transfer lines. The new traceable reference gas mixtures developed in this work will allow the natural gas industry to perform measurement of the concentration of sulfur-free odorants with increased accuracy and confidence, thus providing a tangible benefit to the transporters and end-users of natural gas, the operators of gas transmission networks. Members of the public will also benefit through reassurance that any leaks of natural gas containing sulfur-free odorants will be able to be detected as MA and EA will be present in the gas at the correct



REFERENCES

(1) BP Statistical Review of World Energy June 2013; BP PLC: London, U.K., 2013; http://www.bp.com/statisticalreview (accessed May 2014). (2) Gas Ten Year Statement: U.K. Gas Transmission; National Grid: Warwick, U.K., 2012. (3) Gas Quality Harmonisation: Cost Benefit Analysis; GL Noble Denton and Pöyry Management Consulting, 2012. (4) ISO 13734:1998 (including technical corrigendum 1:1998): Gas analysisorganic sulphur compounds used as odorantsrequirements and test methods. (5) Oka, J.; Yoshida, T.; Kondo, T.; Ito, S.; Katoh, K. J. Therm. Anal. Calorim. 2010, 99, 9−14. (6) Imamura, D.; Akai, M.; Watanabe, S. J. Power Sources 2005, 152, 226−232. (7) Symrise Gasodor Home Page. http://www.gasodor-s-free.de/en/ home.html (accessed May 2014). (8) Graf, F.; Kröger, K.; Reimert, R. Energy Fuels 2007, 21, 3322− 3333. (9) DVGW Arbeitsblatt G 280-1, Gasodorierung, 2007. (10) Varian CP-4900 DMD Sulphur-Free Odorants in Natural Gas Analyzer; Data Sheet SI-0223; Varian Inc.: Palo Alto, CA, 2004. (11) Ruzsanyi, V.; Sielemann, S.; Baumbach, J. I. J. Environ. Monit. 2007, 9, 61−65. (12) Luong, J.; Gras, R.; Cortes, H. J.; Shellie, R. A. Anal. Chem. 2013, 85, 3369−3373. (13) Konopelko, L. A.; Kustikov, Y. A.; Vishnyakov, I. M.; Pavlov, M. V.; Efremova, O. V.; Woo, J.-C.; Kim, Y.-D.; Wessel, R. M.; Ziel, P. R.; Milton, M. J. T.; Vargha, G. M.; Brown, A. S.; Uprichard, I. J. Metrologia 2010, 47, 08004. (14) Milton, M. J. T.; Vargha, G. M.; Brown, A. S. Metrologia 2011, 48, R1−R9. (15) ISO 6142:2001: Gas analysispreparation of calibration gas mixturesgravimetric method. (16) For example, ISO 6145:2001: Gas analysispreparation of calibration gas mixtures using dynamic volumetric methodspart 1: methods of calibration. The ISO 6145 series of standard consists of ten numbered parts, full details of which can be found via http://www.iso. org (accessed May 2014). (17) van der Veen, A. M. H.; Zalewska, E. T. Metrologia 2012, 49, 446−454. (18) JCGM 100:2008: Evaluation of measurement dataguide to the expression of uncertainty in measurement. 6701

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702

Analytical Chemistry

Article

(19) Konieczka, K.; Namieśnik, J. J. Chromatogr., A 2010, 1217, 882− 891. (20) ISO Guide 34:2009: General requirements for the competence of reference material producers.

6702

dx.doi.org/10.1021/ac501525d | Anal. Chem. 2014, 86, 6695−6702