Anal. Chem. 1995, 67,1992-1997
High-Precision Determination of 2H/1Hin Ha and H20 by Continuous-Flow Dsotope Ratio Mass Spectrometry Simon J. Prosser**tand Charles M. Scrimgeoufl Europa Scientific Ltd., Electra Way, Crewe CW7 1Z4, U.K. and Chemistry Department, Scottish Crop Research Institute, Invergowrie, Dundee DO2 5DA, U.K.
The major obstacle to accurate and precise measurement of 2H/1H in H2 by continuous-flow, isotope ratio mass spectrometry has been the interference of the 4Hef carrier peak with the neighboring 2H1H+peak. This has been overcome using a novel mass spectrometer with high dispersion, giving an abundance sensitivity at mlz 3 of < 1 ppm. Samples of H2 with natural abundance 2H/1H can be analyzed routinely in 1 2 min with a precision of &1.5%0. In combination with the equilibration of H2O with H2, this technique provides a fast, accurate method of analysis of water, with a precision of f3.0%0at natural abundance. "here is no measurable memory between samples of either water or H2 gas. The technique is accurate over a wide range of enrichment and is useful for tracer experiments as well as natural abundance studies. Continuous-flow isotope ratio mass spectrometry (CF-IRMS) has simplified the measurement of many of the so-called 'light element' stable isotopes (H, C, N, 0, and S) for a large number of applications. Analysis of 15N,I3C, and lSO to a precision of 0.10.2% is now routine for CF-IRMS.'z2 For many applications this challenges the precision available by traditional, dual-inlet isotope ratio mass spectrometry @I-IRMS), once the variability due to off-line sample preparation is taken into account. Measurement of 34Sby CF-IRMS to -0.5%0 has also been r e p ~ r t e d . The ~ , ~ only light element hitherto unmeasured by CF-IRMS is H. There are a number of reasons for this. Low-resolution mass spectrometers designed for DI-IRMS of Hz often had an extra, shorter spur added to an existing IRMS and were unable satisfactorily to resolve the zHIH+ion beam from the 104-105 times larger *He+ ion beam. Neither was there a suitably fast and simple method of preparing Hzfrom HzO that would benefit from the speed and simplicity of CF-IRMS. The interference of 3Het on the zHIH+ion beam is not important-it is a constant (and very small) proportion of the He carrier and is easily corrected for by a background subtraction. A simple mathematical proof demonstrates that there is no effect Europa Scientific, Ltd. Scottish Crop Research Institute. (1) Prosser, S. J. Int. J. Mass Spectrom. Ion Processes 1993,125,241-266. (2) Barrie, A; Debney, S.;Workman, C. T.; Pullan, C. In International Symposium on Nuclear and Related Techniques in Soil/Plant Studies on Sustainable Agriculture and Environmental Preservation; IAEA Vienna, 1994; IAEA-SM334 / 37. (3) Haystead, A. In Stable Isotopes in Plant Nutrition, Soil Fertilib and Environmental Studies; IAEA Vienna, 1991; pp 3-25. (4) Geisemann, A; Jager H.-J.; Norman, A L.; Krouse, H. R; Brand, W. A Anal. Chem. 1994,66,2816-2819. +
1992 Analytical Chemistry, Vol. 67,No. 73,July 7, 7995
on 6 values caused by the suppression of the 3Hef background during the introduction of a pulse of Hz. Traditionally, 2H/1H has been measured in HzO by reduction of the HzO to Hz over a heated metal, either in a batch process using zinc5f or by on-line reduction in a uranium f u m a ~ e .The ~ on-line technique is subject to memory effects and can require up to 10 replicate samples to be analyzed to achieve reliable results. The results achieved using the batch preparation with Zn are variable across different batches of Zn reagent, although much progress has been made in making the Zn reduction technique more reliable.8-10 A further complication of reduction techniques is that they require pure water; any dissolved material can contaminate the metal surface and prevent complete reaction. Samples must either be pre-prepared by distillation" or kept physically separate until evaporated during the heating process.12 The Zn batch technique is also less satisfactory for enriched samples due to the presence of occluded natural abundance waters. A fundamentally different approach is to equilibrate HzO with Hz. This equilibration needs a catalyst in order to proceed quickly enough to be viable. Platinum is known to catalyze the equilibration; however, it operates on the vapor/gas interface and must be kept dry to work. This has been accomplished using Pt supported on a porous hydrophobic polymer (Hokko beads, Shoko Corp., Tokyo, J a ~ a n ) . ' ~ .This ' ~ achieves equilibration within 1 h and has been used as the basis of an automated batch preparation method.'j The fractionation factor between HzO and H? is highly temperature-dependent, the d2H of Hz gas in equilibration with HzO changes by -6% per degree,16 and so temperature control to i 0.01 "C is necessary. The equilibration technique allows a wide variety of aqueous fluids to be analyzed without the need (5) Coleman, M. L.; Shepard. T. J.; Durham, J. J.; Rouse, J. E.; Moore, G. R. Anal. Chem. 1982,54, 993-995. (6) Kendall, C.; Coplen, T. B. Anal. Chem. 1985,57, 1437-1440. (7)Banie. A; Coward, W. A Biomed. Mass Spectrom. 1985,12,535-541. (8) Tanweer. A; Hut, G.; Burgman, J. 0. Chem. Geol. 1988,73,199-203. (9) Tanweer, A. Anal. Chem. 1990,62,2158-2160. (10) Hayes, J. M.; Johnson, M. W. Indiana University, Bloomington, IN, 1988. Unpublished work. (11) Penman, A D.; Wright, I. A Biomed. Environ. Mass Spectrom. 1987,14, 339-342. (12) Wong, W. W.; Cochran, W. J.; Kiish, W. J.; Smith, E. 0.; Lee, L. S.; Klein, P. D. Am. J, Clin. Nutr. 1987,905-913. (13) Horita, J. Chem. Geol. 1988,72,89-94. (14) Horita, J. Chem. Geol. 1989.79,107-112. (15) Coplen, T. B.; Wildman, J. D.; Chen, J. Anal. Chem. 1991,63, 910-912. (16) Rolston, J. H.; Hartog, J. den; Butler, J. P.J. Phys. Chem. 1976,80, 10641067. 0003-2700/95/0367-1992$9.00/0 0 1995 American Chemical Society
V A-
11
Gas
Figure I. Schematic of the modified 20-20 mass spectrometer showing the extra spur and large separation between lH1H+and 2H1H+. Also shown is a simplified schematic of the gas inlet system.
for pre-distillation. An alternative method,17described in greater detail below, uses a readily available, Pt-on-alumina, catalyst that slows the equilibration time. The effect of temperature fluctuations is greatly reduced by the long equilibration time as they are averaged out, and temperature control of the equilibration process has been found not to be necessary. The slow equilibration also allows samples with bound water, e.g., plant tissue and soil, to be analyzed without pre-extraction of the water, at least for tracer studies.** Despite some advantages of equilibration over the reduction method, it does have some shortcomings. Compared with reduction methods, equilibration requires more sample; 0.1-1 mL versus 1-10 pL. Hydrogen in isotopic equilibrium with H20 at room temperature has a 2H/1H ratio some four times less than that of the H ~ 0 . lThe ~ natural abundance ratio is already very low (2HPH = -0.00015, m/z 3/2 = -0.0003), so any further depletion of 2H makes precise measurement even more problematic. All the methods of determining d2H have relied on DI-IRMS to analyze the prepared gas. The internal precision (-0.15%, 2010) of such IRMS is overkill for the reproducibility of the preparation methods, -1-Z0h, and requires -20 min to analyze each sample. A simpler, faster measurement technique employing CF-IRMS is described below. EXPERIMENTAL SECTION Instrumentation. A 20-20 isotope ratio mass spectrometer'
(Europa Scientific, Crewe, U.K.) was modified to allow simultaneous collection of m/z 2 (lHIH+)and 3 (lH2H+)ions. The large relative mass difference between the two ions of interest results in a large difference in the radius of their paths through the analyzer. The flight tube of the mass spectrometer was 'flared' to accommodate the larger outer radius of the m/z 3 ion beam, and an extra collector spur was added (Figure 1). A magnetic (17) Scrimgeour, C. M.; Rollo, M. M.; Mudambo, S. M. K T.; Handley, L. L.; Prosser, S. J. Bid. Mass Spectrom. 1 9 9 3 , 22, 383-387. (18) Scrimgeour, C. M. J. Hydro/., in press.
field strength of 0.12 T with accelerating potential of -4 x lo3V allowed m/z 2 to be collected in the major collector of the standard triple collector assembly, radius 10.9 cm, while the additional single collector assembly, radius 13.3 cm, measured m/z 3. The standard collector is placed with a sector angle of 120" and a focal length of 9.6 cm; the additional collector is placed with a sector angle of 98.8" and a focal length of 31.7 cm. The distance between the focal points of the two ion beams of m/z 2 and m/z 3 is 24.0 cm; it is this large dispersion that ensures that there is no interference at m/z 3 from m/z 2 or from the He carrier at m/z 4. Gas samples were introduced to the mass spectrometer via the continuous-flow gas injection interface described by Prosser et al.,19 a simplified schematic is shown in Figure 1. In brief, sample gas was injected into a flow of He carrier gas (-60 mL min-l); the sample swept through a drying tube [Mg(C10&, 0.2 m long, 5 mm i.d.) and packed-column GC (Carboseive-G, 0.5 m long, 5 mm i.d.). A small amount of the eluent (-0.05 mL min-l) flowed via a crimp into the ion source of the mass spectrometer. The ratio of m/z 3 to m/z 2 of injected pulses of H2 entering the mass spectrometer were measured by simultaneous integration of the ion beams over time. Procedures. Hydrogen gas was taken from a high-pressure cylinder of 99.999%H2 (Electrochem Ltd., Stoke-on-Trent, U.K.) and transferred into a 0.5 L foil gas bag (Becton Dickinson Ltd., Cowley, Oxford, U.K.) fitted with a three-way valve. Aliquots were drawn from the foil bag, using a 2 mL gas-tight syringe fitted with an isolation valve, and injected into the CF interface. Water samples were equilibrated with H2 using the method described by Scrimgeour et aI.;I70.4 mL of water was equilibrated with -13 mL of the tank H2 gas in 13 mL, septum-topped vials in the presence of a Pt-on-alumina catalyst for 3 days at room temperature. After equilibration, gas was withdrawn from the vials using a 2 mL gas-tight syringe and injected into the CF interface. Hydrogen in equilibrium with water is very depleted in 2H; at 20 "C the fractionation factor is 3.9688; H2 in equilibrium with (19) Prosser, S. J.; Brookes, S. T.; Linton, A; Preston, T. Bid. Muss Spectrom. 1 9 9 1 , 2 0 , 724-730.
Analytical Chemistry, Vol. 67,No. 73,July 7, 7995
1993
V-SMOW at this temperature would have d2H= -748.00?. If such gas were measured against a natural abundance standard, the accuracy would suffer because of the large isotopic d ~ e r e n c e between the gases. Because the permil scale of the Hz is apparently compressed by the large fractionation during equilibration and must be re-expanded to obtain the isotopic composition of the original water, the precision also suffers as the analytical uncertainty is expanded by the same factor as the permil scale. To overcome these problems, samples were measured against Hz equilibrated with a reference water. The fractionation factor, a, is given by
2b.)
0.0- -
38
4.1
Acceleracmg Voliagu (kV1
where R, is the 2H/1H of the water after equilibration and Rg is the 2H/1H of the Hz gas in equilibrium with it. The isotopic composition of a water sample is given by
Figure 2. (a) 4He+ and aHIH+ peaks as they are scanned across the outer spur collector using the acceleration voltage. (b) Expanded scale for the m/z3 peak from which the upper limits of the abundance sensitivity were determined.
ppm). The dilution series extended to -920 ppm excess over the reference water (defined as 0 ppm excess) in six steps. Samples of the international standards V-SMOW and V-SLAPwere also prepared using this method.
combining this with eq 1
RESULTS AND DISCUSSION
Equation 3 demonstrates that if reference and sample waters are equilibrated at the same temperature so that a is constant, then a cancels from the top and bottom of the equation and the d2HW of the sample versus reference water is equivalent to the d2H, between the HZ in equilibrium with the waters. This removes the errors associated with re-expandmg the permil scale and, most importantly, makes the measurement entirely independent of the fractionation factor and thus temperature. It is not necessary to know the value of the fractionation factor or temperature so long as it is constant across a batch of samples and references. The effect of any temporal temperature variation is minimized by the fast analysis time and slow equilibration process. The effects of spatial variations are minimized by keeping the vials close together during the equilibration process, at the place where they will be sampled. Equation 3 gives the isotopic composition of the water at the end of the equilibration, d2HW,which is governed by the original isotopic composition of the water (@How)and Hz (dZHog)and the molar ratio of HZto HzO (e): d2Ho, = d2H, - e(d2H0, - d2Hg)
(4)
If the molar ratio of HZ to HzO is low enough, the isotopic composition of the water does not change significantly during the equilibration, and no correction for the original isotopic composition of the H2 is necessary. Due to an oversight, the procedure described above provides a molar ratio above 0.02, which results in a 2.5%scaling error. We now recommend only 1mL of Hz for 0.4 mL of H20, which brings the scaling error within the overall precision of the technique. A balance of He is used to raise the vial slightly above atmospheric pressure to help guard against ingress of air. The equilibration method was used to prepare a dilution series from -99% 2H20 and reference water at d2Hv.s~ow= 32.0?? (160.7 1994 Analytical Chemistry, Vol. 67, No. 13, July 1, 1995
Mass Spectrometer Performance Characteristics. The mass spectrum between m/z 4 and 3, scanned across the single collector spur, is shown in Figure 2a. The trace was obtained under normal continuous-flow conditions with a mix of a p proximately 201 of He to Hz bleeding into the mass spectrometer. The indicated pressure in the source housing was 4 x mbar, and ion beam intensities were m/z 2, 2.27 x A; m/z 3, 8.42 x A; and m/z 4,1.39 x A The dispersion was measured at m/z 3 as 9.1 cm, equivalent to an effective dispersion radius of 27.3 cm. The resolution was measured as 40 using the ‘5%peak height’ definition, which is given by
R = m/Am,
(5)
where R is resolution, m is the mass of the ion in amu, and Amy is the width of the peak in amu at 5%of the peak height. This is technically equivalent to the ‘10%valley’ definition that is given by the highest mass of two adjacent ion beam peaks of equal intensity, differing in mass by 1 amu, which shows a valley between the peaks of 10%of the peak height. In practice, the ‘5% peak height’ method is easier to use. The mass spectrum shown in Figure 2a is magnified in Figure 2b to show greater detail around m/z 3. The large physical separation between the ion beams has resulted in negligible interference between them. The upper limit for the abundance sensitivity of the tailing from m/z 2 under the m/z 3 beam is 1 ppm; the upper limit for m/z 4 tailing under m/z 3 is 0.5 ppm. Figure 3 shows the peak shapes for m/z 2 and 3 as they are simultaneously scanned across their respective collectors. Also shown is the ratio between these two beams. The flat tops and steep sides of the peaks indicate well-focused ion beams; the flat top of the ratio trace shows excellent stability over a range of 25
v.
Gas Samples. Measurement of isotopes using Hz gas is complicated by ion/molecule reactions forming H3+ ions in the ion source. These are collected with the 2H1H+ions and result
R, = kl,
major 2.40E-8 A nunor749E-lZA rabo 3.28E-4
+ R,
(7)
This constant was then used to extrapolate to major ion beam intensity I,,, = 0 for each point to find the offset ratio, Ro. A 6 value corrected for the H3+ effect was then calculated:
400
4.05 Accelerating Voltage (kV)
Figure 3. Peak scan of m/z 2 and 3 and the 3/2 ratio showing the quality of focusing and coincidence of the ion beams and the 25 V range of acceleration potential for which the ratio remains stable. Table 1. Analytical Preclslon for Injection of Natural Abundance H z Gas
sample size (mL) 1.0 1.0 1.0 1.0 1.o 0.5 0.75 1.0 1.25 1.50
major
beam d2H, area (%reo ( O h ) 100.07 100.95 101.52 102.14 102.63 51.89 78.08 93.32 128.48 150.70
mean
SD
-1.76 2.39 3.70 4.46 5.66 3.0 2.9 -123.91 -57.64 -16.18 73.34 132.60 1.64 102.3
62HcolT (%o)
mean SD
-1.40 0.47 0.30 -0.55 -0.62 -0.4 0.8 1.38 -0.26 1.68 0.03 1.67 0.9 0.9
in an overestimation of the measured 2H/'H ratio. The amount of H3+ formed is proportional to the square of the amount of Hz, and so there is a linear relationship between the measured ZH/ 'H ratio and the pressure of HZ(intensity of the m/z 2 ion beam) in the mass spectrometer. Theoretically, if this relationship is extrapolated back to zero ion beam intensity, the ratio is then correct. Using traditional DI-IRMS, the problems of H3+ are overcome by using a linear correction for major ion beam (m/z 2) intensity in conjunction with accurate matching of the major ion beam intensities of reference and sample gases. Using CFIRMS, it is not always possible to match the sizes of samples and references, and so mathematical correction must be relied upon. Table 1 shows the extent of the 'H3+ effect' for a particular source tuning condition. The major beam area and d2HraW are the original experimental data.
d2H,, = (Rms/Rmr- 1) x 1000 ("A)
(6)
where R,,,, is the measured ratio for the sample and A,, is the measured ratio for the reference. Without correction, even variations of a few percent of major beam area greatly change the d2HraW and render the technique useless. The ratio values were then corrected by linear regression through the data series of measured ratios (Rm)versus major ion beam intensity (Z,J to find the proportionality constant (k) from
where Ro, is the offset ratio for the sample and Rk is the offset ratio for the reference. These corrected values (62Hcorr>show how predictable the effect is. There is no significant difference in analytical precision for sample sizes varying by f50%, and sample sizes matched as closely as the injection technique would allow (Table 1). CF-IRMS is capable of measuring the 2H/1H of natural abundance Hz with a precision of fl%. In order to measure the proportionality constant (k) using DIIRMS, it is necessary to vary the pressure in the inlet and note the change in ratio. Using CF-IRMS, each peak provides intensity versus ratio data as the peak rises and falls. The possibility of using this information to perform an individual H3+ correction on each peak was investigated. However, it was found that differences in transport times for light and heavy molecules through the injection interface and mass spectrometer crimp resulted in a time dependence of the ratio, which was superimposed on the pressure dependence; the two effects proving difficult to deconvolute. In practice, with a measurement cycle of
