Changes in the atmospheric helium isotope ratio over the past 40 years

Authors


Abstract

[1] We have compared the helium isotope ratio in five samples of Pacific marine air spanning the 40 year period between 1973 and 2013 against a secondary gas standard. In a separate experiment we directly compared the 3He/4He ratio in air samples collected in 1973 and 2013 at the same location in La Jolla, California, eliminating any geographical bias. Both experiments are consistent with zero time rate of change for atmospheric 3He/4He. Our best estimate for the rate of change of the 3He/4He ratio in Pacific marine air is −0.0014 ± 0.0045%/yr (2σ), indicating that air helium is still a valid standard for terrestrial helium isotope measurements.

1 Introduction

[2] Atmospheric helium has been used as an isotopic standard for terrestrial helium isotope measurements for several decades. Because the residence time of helium in the Earth's atmosphere (~106 years) is much longer than the mixing time (~10 years), it has been assumed that the 3He/4He ratio in air should be uniform with respect to geographic location. About two decades ago, Sano et al. [1989] raised a different concern by reporting a 1–2% decrease in the 3He/4He ratio in air samples collected over a 10 year period at various sites in Japan. They attributed this decrease to the release of radiogenic helium with low 3He/4He as a byproduct of the burning of natural gas. If the atmospheric helium isotope ratio has decreased by several percent over time, this is an important matter since several modern laboratories are capable of measuring helium isotope variations as small as 0.2% in the 3He/4He ratio. The uncertainty in the Sano et al. [1989] measurements was fairly high, and Lupton and Graham [1991] questioned the assertion that the atmospheric helium isotope ratio is decreasing over time, reporting instead that the rate of change in the atmospheric 3He/4He ratio is −0.0074 ± 0.0372%/yr (2σ). Later, Lupton and Evans [2004] reported additional measurements of Pacific marine air over the 30 year period from 1973 to 2003 and found a rate of change of −0.0050 ± 0.0069%/yr (2σ), consistent with zero rate of change. Other studies have tried to extend the time span for assessing atmospheric helium variations by measuring air trapped in old porcelain [Matsuda et al., 2010] and in historical metallurgical slags [Pierson-Wickmann et al., 2001; Sano et al., 2010]. However, these studies were not able to place tighter constraints on the rate of change compared to the Lupton and Evans [2004] result.

[3] Recently, Brennwald et al. [2013] reported measurements of helium and other noble gases in archived air samples collected at Cape Grim, Tasmania, between 1978 and 2011 and found a larger rate of decrease of 3He/4He of −0.023 to −0.030 ± 0.016%/yr (2σ). Brennwald et al. [2013] made the case that their measurements are unique in that their samples were all collected at the same location. In this paper we report new measurements of air helium collected in 1973, 1993, 2006, and 2013 added to our time series of atmospheric helium measurements. We have also directly compared 2013 Pacific marine air against an archived tank of 1973 Pacific marine air with both samples collected at the same location in La Jolla, California, thereby eliminating any geographical bias. Our results, which are in disagreement with those of Brennwald et al. [2013], place strong constraints on the rate of change of 3He/4He in Pacific marine air over the 40 year period from 1973 to 2013.

2 Methods

[4] All of the measurements reported here were performed on a single mass spectrometer, a 21 cm radius, sector-type, dual-collector instrument specially designed for measuring terrestrial helium isotope ratios. Prior to 1991, the spectrometer was located at University of California Santa Barbara, and then in 1991, it was moved to NOAA/Pacific Marine Environmental Laboratory in Newport, Oregon. The instrument is operated in static mode, with 4He signal measured on a conventional Faraday cup and the weak 3He beam measured with an electron multiplier detector fitted with a pulse-amplifier counting system. After cleanup on hot titanium, samples are exposed to a charcoal finger held at 77°K, which removes the argon fraction as well as any residual N2, O2, etc. In the final stage of the preparation, the sample is exposed to a charcoal finger held at 38°K, which condenses 99.9% of the neon while leaving the helium in the gas phase. Thus, only the two isotopes of helium are present in the mass spectrometer during analysis, thereby eliminating the possibility of helium-neon interference effects. We have corrected all of our results for the effect of lost pulses in the 3He collector due to the finite dead time in the pulse-counting system. The instrument originally had an Amptek pulse-amplifier discriminator (PAD) with a rather large 500 ns dead time, requiring a significant correction to the 3He/4He ratio for some samples. In 2011 we replaced the Amptek PAD with an ARI Corp. F-100T PAD with 75 ns dead time, resulting in much smaller dead-time corrections. We have also corrected all our measurements for the combined processing line and mass spectrometer blank, which in every case was ≤0.1% of the sample size. The precision of a single mass spectrometer measurement averages ~0.2% (1σ) in the 3He/4He ratio for a single air sample analysis based on the reproducibility of repeat analysis of air standards. This is very close to the √N uncertainty that would be predicted based on Poisson statistics associated with the weak 3He signal, indicating that the precision is limited mainly by counting statistics rather than systematic variations.

[5] One approach to the problem of determining changes in the atmospheric helium isotope ratio is to measure the absolute 3He/4He ratio in air. However, a determination of the absolute atmospheric 3He/4He ratio is difficult since it requires the synthesis of a precisely known secondary standard with a 3He/4He ratio of ~10−5 or lower. In fact, the atmospheric 3He/4He ratio is only known to a precision of about 1% and is generally assumed to be (1.39 ± 0.01) × 10−6, which is the average of the results published by Mamyrin et al. [1970] and Clarke et al. [1976]. Rather than attempt to make an absolute determination of the air 3He/4He ratio, we have instead quantified relative changes in the ratio by comparing samples of Pacific marine air against a secondary isotopic standard, the MM geothermal gas (Yellowstone Park sample YP-123 collected from Murdering Mudpots fumarole in June 1979 [see Welhan, 1981]). The MM gas standard, which has a 3He/4He ratio of ~16.5 Ra (where R = 3He/4He and Ra = Rair), has been used as an isotopic standard in our laboratory for over 30 years.

[6] All of the air samples and the MM standard discussed here were housed in stainless steel tanks with an internal volume of 1 to 4 L, depending on the tank. All of the tanks were fitted with an aliquoting system consisting of two Nupro™ 4B valve drivers attached to a common valve body with an internal volume of ~0.13 cm3. In this way, small aliquots of air or MM gas were introduced directly to the mass spectrometer inlet system by simple volume expansion. No pumps were used to transport gases either in the filling of the tanks or in the analytical procedure. The entire system, including the mass spectrometer, the inlet system, and the tanks, was all metal without any o-rings. The air tanks were all filled with marine air on the U.S. Pacific coast by first evacuating the tanks to ultrahigh vacuum and then expanding the air into the tank through a Drierite column to remove water vapor. Prior to this expansion, the Drierite column was purged by using a small hand pump to draw ambient air through the column. Thus, all of the air tanks were at atmospheric pressure or slightly below, depending on how many gas aliquots had been withdrawn.

[7] As described previously in Lupton and Evans [2004], our 1973 air sample came from the air standard tank originally fitted to the Scripps Institution of Oceanography (SIO) mass spectrometer. That tank was originally filled with Pacific marine air at the end of the SIO research pier in 1973. The results reported here depend critically on the integrity of that 1973 La Jolla air standard, which is still housed in its original 4 L tank with the 0.13 cm3 aliquot volume attached. About 10,000 air aliquots had been drawn from this tank when we acquired it in 2003. Since it was originally filled to 1 atm, it should now have an internal pressure of ~0.8 atm. In 2003 a single aliquot from this tank yielded 4.102 × 10−7 cm3 STP helium as measured in the mass spectrometer and standardized against a precision capacitance manometer. This agrees quite well with what would be expected based on the number of aliquots withdrawn. In 2013, after ~100 additional aliquots had been withdrawn, a single aliquot yielded 4.096 × 10−7 cm3 STP helium, essentially identical to the 2003 value. This indicates negligible leakage into the tank over the 10 year period from 2003 to 2013. Furthermore, after the tank had been stored for 10 years with both aliquot valves closed, the aliquot volume was still at high vacuum, indicating that neither of the two aliquot valves has a cross-seat leak. For these reasons we are confident that the 1973 La Jolla air tank has retained its integrity during the intervening 40 years.

[8] As discussed above, our MM standard is a geothermal gas collected in 1979 at Yellowstone Park, Wyoming [Welhan, 1981]. In 1983 a split of this gas was expanded into a 4 L stainless steel tank fitted with an aliquot system similar to the air sample tanks. Because of the high helium content of the MM gas, the pressure in the tank was adjusted to ~1.3 mm Hg so that the helium released in one aliquot expansion would be manageable. Thus, the MM tank is at ~1/600 of atmospheric pressure, and any problems would be manifested as air leakage into the tank. We recently measured the following composition for the MM gas: He/Ne/(He/Ne)air = 1100, He/Ar/(He/Ar)air = 623, and 40Ar/36Ar = 380. The original MM gas as analyzed by Welhan [1981] had He/Ar/(He/Ar)air = 1260 ± 660 (2σ). Thus, the gas in our MM tank appears to have a composition nearly identical to the original collection. Even if one assumes that all of the neon in the MM gas represents air addition to the tank, then the 3He/4He ratio has only decreased by a negligible 0.09% due to this effect. We are thus confident that our MM standard tank has not changed since it was created in 1983.

[9] We also addressed the effect of pressure on linearity in the mass spectrometer. Our mass spectrometer is fitted with a Baur-Signer ion source which has very low mass discrimination and excellent linearity with respect to pressure effects. For example, the 3He/4He ratio changes by only 0.20% for a factor of 2 pressure change in the range where these measurements were made. To further eliminate any possible pressure effects, we bracketed the MM samples with air samples with slightly higher and lower 4He contents. A similar approach was used in our direct comparison between air samples.

3 Results

[10] In this paper we report new results for Pacific marine air collected in 1973, 1993, 2006, and 2013 measured versus the MM isotope standard. In a sense, these new results represent an extension of the time series measurements previously reported in Lupton and Evans [2004]. However, as discussed above, the new measurements were made with a faster 75 ns pulse-amplifier system as compared with the slower ~500 ns PAD used for the Lupton and Evans [2004] results. For 1973, 1993, and 2006 we have measurements with both pulse-counting arrangements, and this comparison indicates that the new faster PAD gives Rair/RMM values that are about 0.3% lower on the average (see Table 1). This indicates some small unresolved difference in the pulse-counting dead-time corrections [see Evans, 1955]. For this reason we have chosen to focus on the new measurements and treat them as a data set separate from the Lupton and Evans [2004] results in order to eliminate any possible bias.

Table 1. 3He/4He in Pacific Marine Air Relative to the MM Helium Isotope Standard
DateLocationRair/RMMb2σ2σ (%)
  1. a

    This sample consisted of aliquots taken from the air standard tank borrowed from the helium isotope spectrometer “GAD” at the Scripps Institution of Oceanography, originally collected at the end of the SIO pier in 1973.

  2. b

    Here R = 3He/4He. “MM” refers to the special helium isotope standard collected in 1979 at Murdering Mudpots fumarole in Yellowstone Park, Wyoming.

  3. c

    This sample was collected in 2013 at the end of the SIO pier in La Jolla, California, in the same location as the 1973 La Jolla air sample.

  4. d

    Published previously in Lupton and Evans [2004].

Measurements With Slower 500 ns Pulse-Amplifier System
1973La Jolla, CAa0.06063d0.000100.16
May 1985Santa Barbara, CA0.06060d0.000100.16
Feb 1988Santa Barbara, CA0.06051d0.000200.34
Jan 1990Santa Barbara, CA0.06068d0.000060.10
Feb 1990Santa Barbara, CA0.06061d0.000120.20
Jul 1993Newport, OR0.06065d0.000050.09
Jul 1993Newport, OR0.06058d0.000110.18
Sep 1997Newport, OR0.06059d0.000100.16
Sep 1997Newport, OR0.06052d0.000150.25
Feb 2003Newport, OR0.06052d0.000100.16
Jul 2006Newport, OR0.060560.000120.20
New Measurements With Faster 75 ns Pulse-Amplifier System
1973La Jolla, CAa0.0604030.0000790.13
Jul 1993Newport, OR0.0604120.0001160.19
Jul 2006Newport, OR0.0603860.0001380.22
Mar 2013Newport, OR0.0604490.0000440.07
May 2013La Jolla, CAc0.0604400.0000480.08

[11] As described above, the 1973 air sample came from the air standard tank originally fitted to the SIO mass spectrometer. That tank was originally filled with Pacific marine air at the end of the SIO research pier in La Jolla in 1973. As discussed above, we are confident that this air standard tank has retained its integrity over the 40 intervening years. The 1993 and 2006 samples are both air standards that we use for the routine calibration of our mass spectrometer. A 2013 air sample was collected recently in Newport, Oregon, purposely for this study. In order to remove any geographic bias in our sampling, in May of 2013 we arranged to have an additional air sample collected in La Jolla at the end of the SIO pier in the same exact location as the original 1973 sample. This sample was also compared against the MM standard and is included in this study.

[12] Our results are shown in Figure 1, which shows Rair/RMM plotted versus year (where R = 3He/4He) for the five determinations spanning 1973 to 2013 made with the new faster pulse-counting system. The error bars shown are all 2σ, corresponding to a 95% confidence interval. The complete suite of air versus MM measurements is listed in Table 1, including values determined with the older PAD published in Lupton and Evans [2004]. As shown in Table 1, the 1973 air was collected in La Jolla, California, and there is one additional La Jolla air sample that was collected in 2013. The other samples were collected in Santa Barbara, California, or in Newport, Oregon.

Figure 1.

Comparison of the 3He/4He in Pacific marine air relative to the MM geothermal standard versus time for the 40 year period from 1973 to 2013 for five measurements made with our faster pulse-counting system. Error bars are ±2σ. Solid line is a weighted linear regression fit to the data. The shaded region represents the 95% or 2σ confidence envelope for our results, using the method of Wonnacott and Wonnacott [1979]. Dashed line represents the slope from our direct comparison of 1973 versus 2013 La Jolla air normalized to the mean of the Rair/RMM results. Data are listed in Table 1.

[13] A weighted linear regression fit to these five measurements gave 0.0017 ± 0.0032%/yr (2σ) for the rate of change of the 3He/4He ratio for Pacific marine air. This corresponds to a slight increase of 0.071% in the 3He/4He ratio over the 40 year period from 1973 to 2013. The shaded region indicates the 95% confidence band based on the method of Wonnacott and Wonnacott [1979]. This new result is in agreement with the older estimate of −0.0050 ± 0.0069%/yr (2σ) reported by Lupton and Evans [2004] for the 30 year period from 1973 to 2003. It is important to note that both results are also consistent with zero rate of change in the 3He/4He ratio at the 95% confidence level.

[14] In addition to comparing marine air against the MM standard, we also conducted two separate experiments in which we directly compared the 1973 La Jolla air tank against our two different collections of 2013 air. In each case, this was done by connecting the 1973 and 2013 air tanks and aliquot systems directly to the mass spectrometer processing line and alternating analyses between the two air samples. This approach eliminates any bias with the dead-time corrections or possible problems with the MM standard. A total of 15 comparisons between 1973 La Jolla air and the 2013 Newport air gave R1973/R2013 = 1.00104 ± 0.0019 (2σ), indicating a small decrease of 0.10% over the 40 year period. A similar set of comparisons between 1973 La Jolla air and 2013 La Jolla air gave R1973/R2013 = 1.00056 ± 0.0018 (2σ), indicating a small decrease of 0.06% over the 40 year period. These results are consistent with our Rair/RMM time series results within the errors and are both also consistent with zero rate of change at the 95% confidence interval.

4 Discussion

[15] As mentioned in section 1, Brennwald et al. [2013] measured helium isotope ratios in archived air samples collected at Cape Grim, Tasmania, between 1978 and 2011 and found a rate of change of −0.023 to −0.030 ± 0.016%/yr (2σ). Figure 2 compares the Brennwald et al. [2013] results with the 95% confidence envelope for our air/MM time series measurements (from Figure 1). In order to make this comparison, we have normalized our results to the average of their results at the median point. The dashed line in Figure 2 represents the slope of our direct comparison of 1973 versus 2013 La Jolla air, which represents our best estimate of the rate of change. The plot makes it clear that the Brennwald et al. [2013] best fit slope implies a rate of decrease in atmospheric 3He/4He that is ~15 times higher than our best estimate. However, their measurements are about 3 times less precise than ours, with the result that the 95% confidence envelope for our air/MM measurements is roughly consistent with their 2σ error bars. However, the converse is not true, i.e., their best fit slope does not fit within our 95% confidence envelope.

Figure 2.

Comparison of our results with those of Brennwald et al. [2013]. Error bars are all ±2σ. Solid line is the linear regression fit to the Brennwald et al. [2013] results. The shaded region is the 95% confidence envelope for our results from Figure 1 which has been normalized to the average of their results for purposes of comparison. Dashed line shows the slope from our direct comparison of 1973 versus 2013 La Jolla air.

[16] In their paper, Brennwald et al. [2013] also make the case that their measurements are unique in that their samples were all collected at the same location. They discuss the possibility that latitudinal or interhemispherical variations in atmospheric helium isotope ratios may exist. On this basis, they dismiss most of the previously published measurements including our previous work [Lupton and Evans, 2004] in which samples were collected at various locations in California and Oregon. In fact, Sano et al. [2010] reported evidence that the atmospheric helium isotope ratio is about 0.33% higher at latitude 50°N compared to 50°S. On the other hand, Jean-Baptiste and Fourré [2012] found that the mean air He isotope ratios are identical for the Northern and Southern Hemispheres. In this paper we have directly addressed concerns about latitudinal variations by sampling La Jolla air at the end of the SIO pier in 2013 at exactly the same location as the earlier 1973 La Jolla air sample. As shown in Figure 1 and Table 1, our measurements of 2013 Newport, Oregon, air and 2013 La Jolla, California, air are identical within 0.03%, indicating little if any latitudinal effect between 32.8°N and 44.6°N. In terms of any interhemispherical variations, this could be addressed by comparing a Southern Hemisphere air sample such as one of the Cape Grimm archive samples against our MM standard or against one of our Northern Hemisphere air samples.

[17] Recently, Mabry et al. [2013] have undertaken measurements of Cape Grimm Air Archive (CGAA) samples in order to verify the Brennwald et al. [2013] results. Mabry et al. [2013] found very little decrease in 3He/4He for CGAA samples collected between 1978 and 2011, in disagreement with the Brennwald et al. [2013] published results. Thus, the higher-precision measurements of Mabry et al. [2013] indicate that the Southern Hemisphere CGAA samples show only a small rate of change in agreement with our Northern Hemisphere results. The question remains as to why the Brennwald et al. [2013] results exhibit an apparent linear decrease in 3He/4He over time.

[18] Assuming that the main effect on atmospheric 3He/4He is due to the burning of natural gas, it is of interest to compare our results with the global rate of natural gas consumption. Our comparison of 1973 and 2013 La Jolla air gave a rate of change for air 3He/4He of −0.0014 ± 0.0045%/yr (2σ), corresponding to a release of 1.3 ± 4.2 × 1010 mol 4He/yr. The current global consumption of natural gas is estimated at 1.3 × 1014 mol/yr, but the average over the last 40 years is somewhat lower, about 1.0 × 1014 mol/yr [IEA, 2013]. If the burning of natural gas is the sole source of the change in atmospheric 3He/4He, then a natural gas He concentration of 0.013 ± 0.042% would be required to produce the change we observe. This is an order of magnitude lower than the 0.1–0.25% estimated for the mean global natural gas helium content [Oliver et al., 1984]. If one uses the maximum rate of decrease of air 3He/4He consistent with our results at the 2σ level, this yields a natural gas He content of 0.055%. This still implies that 75% of the helium emitted by natural gas burning is missing. We have examined the U.S. Bureau of Mines data base for the analysis of 17,278 natural gas samples in the continental U.S. [U.S. Bureau of Mines (USBM), 1992]. An unweighted average of these data yields an average He content of 0.41% [USBM, 1992]. However, when these results are weighted according to the open flow rate of the wells, the average decreases to 0.23%. This indicates that the He content is inversely correlated with the size of the reservoir, suggesting that the He content of the natural gas that is actually being consumed may be much lower than 0.2%, perhaps in agreement with our 3He/4He measurements.

5 Summary and Conclusions

[19] We have made direct measurements of helium isotope ratios in samples of Pacific marine air collected over the 40 year period from 1973 to 2013. The simplest and possibly the most instructive result was our direct comparison of 1973 La Jolla air against 2013 La Jolla air, which gave an estimated rate of change of −0.0014 ± 0.0045%/yr (2σ), equivalent to a change of only −0.056 ± 0.0018% over the 40 year period. Our direct comparison of 1973 La Jolla air against 2013 Newport, Oregon air gave a similar result. The best fit to our air versus MM comparison for five air samples collected in 1973, 1993, 2006, and 2013 gave a slightly positive change of 0.0017 ± 0.0032%/yr (2σ). Although all three experiments that we conducted are consistent with zero rate of change at the 95% confidence level (see Table 2), the average of our results suggests a small decrease in atmospheric 3He/4He over time. Our measurements place strong constraints on any possible decrease, indicating that the atmospheric helium isotope ratio has decreased less than 0.2% between 1973 and 2013. This is a very small change that is not analytically detectable in most laboratories, indicating that air helium remains a viable standard for terrestrial helium isotope measurements.

Table 2. Comparisons of 3He/4He in Pacific Marine Air 1973 Versus 2013
DescriptionR1973/R2013a2σRate of Change (%/yr ± 2σ)
  1. a

    Here R = 3He/4He.

  2. b

    The La Jolla air samples for 1973 and 2013 were both collected at the same location at the end of the SIO pier in La Jolla, California.

  3. c

    These measurements were made with the faster 75 ns pulse-counting system installed.

Direct comparison 1973 La Jolla, California air versus 2013 Newport, Oregon airb1.00104c0.0019−0.0026 ± 0.0048
Direct comparison 1973 La Jolla, California air versus 2013 La Jolla, California airb1.00056c0.0018−0.0014 ± 0.0045
Linear regression fit to marine air versus MM (from Table 1, faster PAD)0.99929c0.0013+0.0017 ± 0.0032
Brennwald et al. [2013] estimate from CGAA samples____−0.023 to −0.030 ± 0.016

Acknowledgments

[20] We thank D. Graham and R. Poreda for useful discussions, H. Kueker for collecting the 2013 La Jolla air sample, and A. Lau for assistance with the statistical analysis. This work was supported by the NOAA Earth Ocean Investigations Program (formerly the NOAA Vents Program). This is PMEL contribution 4047.

[21] The Editor thanks Matthias Brennwald and an anonymous reviewer for their assistance in evaluating this paper.

Ancillary

Advertisement