Anal. Chem. 2000, 72, 2664-2666
On-Line D/H Analysis for Water, Natural Gas, and Organic Solvents by Manganese Reduction O. Shouakar-Stash,*,† R. Drimmie,† J. Morrison,‡ S. K. Frape,† A. R. Heemskerk,† and W. A. Mark†
Department of Earth Sciences, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada, and Micromass UK Limited, Floats Road, Wythenshawe, Manchester M23 9lZ, U.K.
A new technique for on-line sample preparation and D/H determination is described. The technique is suitable for the preparation of fresh and brine waters, as well as natural gases and organic solvents. A 5-µL sample of water or hydrogen equivalent is injected and reduced by means of hot manganese metal in a specially designed reaction tube surrounded by a tube furnace and attached directly to the mass spectrometer inlet without modification.The hydrogen gas flows directly into the MS to be analyzed by reference/sample comparison. The reproducibility varied between 0.7 and 1.8‰ for all liquid and gas samples. The accuracy of this technique is confirmed by analysis of IAEA standard waters V-SMOW, GISP, and SLAP, as well as NGS-3 (IAEA methane intercomparison material). Hydrogen isotope analysis of water and organic compounds by mass spectrometry is performed on the hydrogen gas obtained by the quantitative reduction of water and organic compounds. In the past zinc2 and uranium1 were commonly used in reduction systems. Other reagents used are chromium,3 carbon in a nickel tube,5 and platinized magnesium.4 Manganese is another reduction agent, which can be used to produce hydrogen gas. Manganese was chosen because it is thermodynamically more efficient than zinc in water reduction6 and lacks the radioactive byproducts created when uranium is used. Manganese is easily obtained, is less costly than most of the other reagents, and does not require any pretreatment. This technical note contains a description of the method used to determine deuterium isotope ratio by manganese reduction. Quantitative reduction of water proceeds as in the equation 900 °C
H2O + Mn 98 H2 + MnO
(1)
A promising method based on eq 1 has been developed for the * Corresponding author (phone) (519) 885-1211 ex. 6465; (fax) (519) 7460183; (e-mail)
[email protected]. † University of Waterloo. ‡ Micromass UK Limited. (1) Bigleisen, J.; Perlman, M. L.; Prosser, H. C. Anal. Chem. 1952, 24, 13561357. (2) Coleman, M. L.; Shepherd, T. J.; Durham, J. J.; Rouse, J. E.; Moore, G. R. Anal. Chem. 1982, 54, 993-995. (3) Gehre, M.; Hoefling, R.; Kowski, P.; Strauch, G. Anal. Chem. 1996, 68, 4414-4417. (4) Halas, S.; Jasinska, B. Isot. Environ. Health Stud. 1996, 32, 105-109. (5) Motz, J. E.; Edwards, T. W. D.; Buhay, W. M. Chem. Geol. 1997, 140, 145-149. (6) Tanweer, A.; Han, L.-F. Isot. Environ. Health Stud. 1996, 32, 97-103.
2664 Analytical Chemistry, Vol. 72, No. 11, June 1, 2000
Figure 1. Schematic of the on-line device.
simple and rapid conversion of a large variety of sample types [fresh and brine water (total dissolved solids >100 g L-1), natural gases, and organic compounds] to pure hydrogen for isotopic measurement on a mass spectrometer, using an on-line device. EXPERIMENTAL SECTION Apparatus. The system (Figure 1) is fitted directly to the sample inlet of a VG Micromass 602C mass spectrometer. The quartz reduction tube and the borosilicate glass cap were fabricated at the University of Waterloo. The furnace was constructed of commercially available ceramic elements, which are controlled by a variable transformer. The temperature is monitored with an Omega CN9000 type K thermocouple, inserted to the midpoint of the furnace. A manual valve was added between the reduction tube and the spare foreline vacuum port on the 602C mass spectrometer. This allows for the evacuation of the reduction tube, through the roughing pump, while analysis of the previous sample proceeds. 10.1021/ac991384i CCC: $19.00
© 2000 American Chemical Society Published on Web 04/25/2000
Table 1. Test of Accuracy of the On-Line Mn Reduction Technique for Water Samples Using IAEA International Standards and In-House EIL Standards IAEA standards δD (‰)
no. of runs mean value (‰) std dev
EIL standards δD (‰)
V-SMOW (0.00)
GISP (-189.50)
SLAP (-428.00)
EIL-14 (10.70)
EIL-10 (-206.00)
EIL-15 (-492.00)
13 -0.3 1.5
13 -189.5 1.0
13 -426.2 1.8
16 10.5 0.7
16 -205.8 1.3
16 -492.3 0.7
Preparation of the Reduction Tube. Pyrex glass wool is inserted in the narrow part of the tube (5 cm), and quartz wool is placed on top of the Pyrex wool (7 cm). Then 50 g of Mn (-50 mesh, 99+%, Aldrich Chemical Co. Inc., Catalog No. 26,615-9) is poured on top of the quartz wool as shown in Figure 1. A viton O-ring seal is used on the glass ball-and-socket injection port which is clamped together with a metal clip. The septum (Chromatographic Specialties Inc., Catalog No. C13413) is set on top, Teflon side out, and the cap is screwed down firmly. The reduction tube is mounted onto the sample inlet with the Mn centered in the furnace. The tube is left under roughing vacuum overnight with the furnace off and the MS sample inlet closed. To minimize temperature fluctuations, quartz wool is packed in both ends of the furnace around the reaction tube. The next morning, the variable transformer is turned up until a stable operating temperature is achieved (900 ( 5 °C). Water Samples. Using a syringe (Hamilton Co., Microliter 701), 5 µL of water sample is injected. The sample evaporates quickly and is reduced upon contact with the manganese. The produced hydrogen gas is allowed to flow from the furnace to the reservoir through the sample inlet for 20 s. Then the sample inlet is closed and the hydrogen is left for 200 s to cool, until both sample and reference gases have reached the same temperature. The reference and the sample sides are then balanced to a major ion beam of 6.0 × 10-9 A and analysis is started. At this point, the reduction tube is pumped out through the foreline roughing port in preparation for the next sample injection. Natural Gas Samples. There is no hydrocarbon separation available on this system at this time. This technique can only be used for pure gas samples or total combustible hydrocarbon. Preseparation on a GC can be performed to analyze a specific compound in a mixture. In this case, 3 cm3 of methane or ethane gas is injected. To allow the reaction to go to completion, the gas is left in the reaction tube for 4 min. The sample inlet is opened for 20 s to allow the hydrogen gas to expand into the sample reservoir, after which the inlet sample is closed and the hydrogen is left for 200 s to cool. The measurement starts after the pressure is equalized between the reference and the sample inlets at a major ion beam of 6.0 × 10-9 A. Organic Samples. In the case of liquid organic samples, a suitable amount of pure phase solvent (∼4-10 µL) is injected according to the quantity of hydrogen in the molecule; e.g., 4 µL is enough for decane. After injection, the sample remains in the reaction tube for 9 min to allow the reaction to go to completion. The sample inlet is then opened for 20 s to allow the hydrogen gas to expand into the reservoir. The inlet sample is closed and the hydrogen is left for 200 s to cool to the same temperature as the reference gas. The measurement starts after the pressure is
Table 2. Test of Accuracy of On-Line Mn Reduction Technique for Natural Gas Samples Using Both IAEA and In-House EIL Standards IAEA stds δD (‰)
no. of runs mean value (‰) std dev
EIL stds δD (‰)
NSG-3 (-175.95)
EIL-8 (-111.300)
EIL-7 (-157.871)
10 -176.9 0.9
10 -112.2 0.7
3 -158.8 0.7
equalized between the reference and the sample inlets at a major ion beam of 6.0 × 10-9 A. Using 5 µL of water/sample, 200 reductions to pure hydrogen can be carried out from 50 g of manganese. Natural gases are chemically more stable than water, thus requiring a longer reaction time. The complete conversion of methane to pure hydrogen requires 4 min at 900 °C. During the reaction time, the valve between the reduction tube and the mass spectrometer remains closed. Furthermore, organic and chlorinated solvents require an even longer time of 9 min at 900 °C. Calibration and Standardization. The deuterium contents are expressed in terms of δD, i.e., per mil (‰). The δD values are calculated as follows:
δD ) (Rsample - Rreference)/Rreference × 1000
(2)
where Rsample and Rreference are the sample and reference ratio of minor and major beams, respectively, as measured on the mass spectrometer. The calibration of δD values with respect to the international δD scale for water samples analyzed by this technique was performed by analyzing V-SMOW, SLAP, and GISP (Table 1). Calibration of δD for organic compounds was done by analyzing NGS-3 [IAEA natural gas, natural gas intercomparison material (98.8% CH4)] and EIL-8 (in-house methane standard) (Table 2). In the Environmental Isotope Laboratory (EIL), deuterium results are corrected to a regression line drawn from standards run daily. These standards are in-house laboratory waters whose values are calibrated to the IAEA International standards SMOW, SLAP, and GISP and are monitored biannually for change. Like the IAEA standards, the EIL waters cover a wide range; EIL-14 has a δ of +10.4, EIL-10 has a δ of -206, and EIL-15 has a δ of -492 (Table 1). RESULTS AND DISCUSSION: The high correlation coefficient (R2 ) 0.999 924) of the linear regression in Figure 2, based on the EIL and IAEA standard Analytical Chemistry, Vol. 72, No. 11, June 1, 2000
2665
Table 5. Comparison between EIL and Micromass Results for Chlorinated Solvents (TCE and TCA) from Different Sources and Years EIL
Figure 2. Regression line of water calibration. Table 3. Test of Memory Effect of On-Line Mn Reduction Technique Using Three Water Standards order of meas
V-SMOW δD (‰)
order of meas
GISP δD (‰)
order of meas
EIL-15 δD (‰)
1 2 3 4
-2.1 -1.9 -1.7 0.9
5 6 7 8
-190.0 -189.7 -188.3 -189.1
9 10 11 12
-489.6 -491.6 -493.2 -491.8
mean value (‰) std dev
-1.2
-189.3
-491.6
1.4
0.7
1.5
Table 4. Comparison between EIL and Micromass Results of Organic Compounds EIL
Micromass Factory
compound
n
mean δD (‰)
std dev
n
mean δD (‰)
std dev
decane dodecane undecane methyl decanoate methyl pentadecanoate methyl decanoate (capric acid) squaline
3 2 2 3 2 3
-84.6 -104.9 -225.4 -233.5 -57.5 -227.6
1.1 1.8 0.2 1.2 1.2 1.0
3 3 3 3 3 3
-84.5 -105 -221.6 -232.1 -58.3 -206.1
1.4 1 1.5 1.2 0.2 1.5
2
-167.7
0.9
3
-167.6
1.5
results (87 data results) presented in Table 1, indicates that the technique- and machine-dependent fractionations are predictable over an approximately 500‰ range of δD values and can be calibrated by using a simple linear relationship. The results of the isotope analyses of organic compounds are shown in Tables 4 and 5. To compare the accuracy and the reproducibility of the analyses, all samples were analyzed in the EIL using the above technique, as well as by an on-line pyrolysis technique consisting of a Euro Vector Euro EA elemental analyzer interfaced in continuous-flow mode to a Micromass IsoPrime IRMS at the Micromass Factory in the U.K. The measurements agree well for both laboratories. The reproducibility of the δD values using this technique varied between 0.7 and 1.8 for all liquid and gas samples except TCE PPG 93, which had a standard deviation of (4.3, which may be due to sample impurities.
2666 Analytical Chemistry, Vol. 72, No. 11, June 1, 2000
compound
n
mean δD (‰)
TCA ICI 93 TCA PPG 93 TCA PPG 95 TCA VULCAN 093 TCA VULCAN 593 TCE DOW 92 TCE DOW 95 TCE ICI 93 TCE PPG 93 TCE PPG 95
3 2 2 2 2 2 2 2 3 2
-6.1 -18.4 14.7 22.2 15.1 466.9 583.8 519.8 681.9 570.2
Micromass Factory std dev
n
mean δD (‰)
std dev
1.2 0.3 1.4 0.3 1.3 0.7 0.2 0.8 4.3 1.8
3 3 4 5 5 3 3 3 3
-10.7 17.7 16.9 17.3 458.9 578.4 529.9 669.4 572.2
0.3 0.2 0.5 1.2 2.4 3.3 1.8 0.9 0.8
To prove the accuracy of the δD values using this technique, the international standards V-SMOW, SLAP, GISP, and NSG-3 were determined. The measured δD values for these standards agree with defined values. A memory effect test was carried out on three standards covering a range of ∼500‰ (Table 3). Each standard was measured four times. The standard deviations of the mean values from all measurements are within the accuracy of the manganese technique presented in this paper. As was shown in Table 3, There is no detection of memory effect when standards are apart by ∼200‰, as in the case of V-SMOW and GISP. However, when the difference is more than 200‰, a shift of 2-3‰ was detected in the first injection, as in the case of GISP and EIL-15. To avoid any memory effect, one can flush 5 µL through the tube during the pumping period prior to a run. This simple procedure was proven to be effective, and it does not add any extra time to analysis time. The results of the isotope analyses of chlorinated solvents (trichloroethylene, TCE; and trichloroethane, TCA) from different sources and years are shown in Table 5. Injections of 20 and 9 µL for TCE and TCA, respectively, have been used in these cases. CONCLUSON Deuterium analysis for water, natural gases, and organic solvents using the on-line reduction by manganese is simple and easy to perform in any laboratory with a dual-inlet deuterium isotope ratio mass spectrometer. The reproducibility is comparable and possibly even better than previous techniques. The memory effect is minimal and easy to avoid. ACKNOWLEDGMENT We thank CRESTech and NSERC for financial support for S.K.F. and the EIL. Comments by anonymous reviewers are appreciated. Received for review November 30, 1999. Accepted February 24, 2000. AC991384I
