Ultrahigh-precision noble gas isotope analyses reveal pervasive subsurface fractionation in hydrothermal systems

Abstract

Mantle-derived noble gases in volcanic gases are powerful tracers of terrestrial volatile evolution, as they contain mixtures of both primordial (from Earth's accretion) and secondary (e.g., radiogenic) isotope signals that characterize the composition of deep Earth. However, volcanic gases emitted through subaerial hydrothermal systems also contain contributions from shallow reservoirs (groundwater, crust, atmosphere). Deconvolving deep and shallow source signals is critical for robust interpretations of mantle-derived signals. Here, we use a novel dynamic mass spectrometry technique to measure argon, krypton, and xenon isotopes in volcanic gas with ultrahigh precision. Data from Iceland, Germany, United States (Yellowstone, Salton Sea), Costa Rica, and Chile show that subsurface isotope fractionation within hydrothermal systems is a globally pervasive and previously unrecognized process causing substantial nonradiogenic Ar-Kr-Xe isotope variations. Quantitatively accounting for this process is vital for accurately interpreting mantle-derived volatile (e.g., noble gas and nitrogen) signals, with profound implications for our understanding of terrestrial volatile evolution.

Fig. 1.. Nonradiogenic Ar-Kr-Xe isotope variations across all volcanic noble gas samples analyzed in this study, reported using delta notation relative to the atmosphere.

(A and B) Predictions for steady-state isotope fractionation via diffusive transport fractionation (DTF) through CH₄, H₂O, N₂, and CO₂ are given as vectors. Samples with ∆⁴⁰Ar = δ⁴⁰Ar/³⁶Ar − 2*δ³⁸Ar/³⁶Ar > 250‰ (i.e., ⁴⁰Ar/³⁶Ar ≥ 373, if δ³⁸Ar/³⁶Ar = 0‰) are noted with asterisks.

Fig. 2.. δ¹³²Xe/³⁶Ar versus δ⁸⁴Kr/³⁶Ar composition of all volcanic noble gas samples analyzed in this study, reported using the delta notation relative to the atmospheric isotope composition.

(A) The heavy noble gas composition of AEW (56) is given for comparison, as a function of temperature (from 0° to 100°C). Symbols with and without black outlines have been analyzed by DMS on MAT253⁺ and MAT253 (both instruments in Seltzer Lab, WHOI), respectively. (A) and (B) Samples with ∆⁴⁰Ar > 250‰ (i.e., ⁴⁰Ar/³⁶Ar ≥ 373, if δ³⁸Ar/³⁶Ar = 0‰) are noted with asterisks.

Fig. 3.. Schematic representation of typical volcanic gas collection and illustration of DTF.

(A to C) Cross-section of the subsurface, where volcanic gas (mainly CO₂) ascends through shallow groundwater until it reaches the surface. Part of the gas is captured by an upside-down funnel submerged in surficial water. The captured gas then flows through the tubing system and copper tube, until it reaches the preevacuated Giggenbach bottle containing a 5 N NaOH solution (17, 51). Because of the pressure gradient across the NaOH solution, a fraction of the volcanic gas is sucked into the Giggenbach bottle, where CO₂ dissolves in solution and inert gases accumulate in the headspace. CO₂ bubble stripping of groundwater-derived noble gases is a ubiquitous process that accounts for the occurrence of air-derived components with AEW elemental abundances in volcanic gas systems (Fig. 2) (45). Groundwater-derived noble gases then experience diffusive transport against CO₂ [also referred to as “binary” or “mutual” diffusion; (20, 26)]. The difference between the diffusion coefficients of heavy (slow) and light (fast) isotopes (here, only ³⁶Ar and ³⁸Ar are shown for illustrative purpose) produces a depletion of the heavy isotopes (in this case, ³⁸Ar), whose magnitude depends on the atomic mass of the gas in medium (20).

Fig. 4.. Empirical relationship between δ¹²⁸Xe/¹³⁰Xe and δ⁸⁶Kr/⁸⁴Kr, used to correct Xe isotopes for DTF.

The slope of the regression (−0.5331 ± 0.0179; 95% CI, Mean Squared Weighted Deviation (MSWD) = 1.78) is obtained by using the error weighted least squares algorithm of (59), forcing the regression through the origin. This correction is required to yield unbiased access to deep Xe isotope signals originating from mantle and crustal sources. Note that the same approach can be used by using δ⁸²Kr/⁸⁴Kr instead of δ⁸⁶Kr/⁸⁴Kr (fig. S5).

Fig. 5.. Overview of Xe isotope systematics across Yellowstone National Park.

(A to F) Xe isotopic spectra of volcanic gas samples collected along the north-south transect from the edge of the Yellowstone caldera (Washburn spring) to its center (Mud Volcano). For each sample, the raw data and data corrected for DTF (Fig. 4 and fig. S6) are both displayed for comparison. Helium isotope compositions are given for (33). The schematic map of Yellowstone National Park shows the location of the four sampling sites analyzed in this study. Thermal areas across the Yellowstone National Park are shown as red circles. The evolution of the F_CM parameter [= (δ¹³⁶Xe* − δ¹²⁹Xe*)/δ¹³⁶Xe*] across the four samples analyzed in this study, as a function of the distance along the north-south transect from Washburn spring to Mud Volcano, demonstrates a clear evolution from a primarily crustal (F ~ 1) to a primarily mantle-sourced (F ~ 0) volcanic setting. Helium isotope data measured in the Barry Lab are reported for each sample (Supplementary Materials).

Fig. 6.. Xe isotopic spectra of volcanic gas samples collected in Eifel (Germany).

(A and B) The two samples collected at E-VQ and E-SQ show similar extents of DTF (e.g., comparable δ¹²⁸Xe/¹³⁰Xe) but distinct radiogenic and fissiogenic isotope excesses. Whether variations in mantle gas contributions from one site to the other, and from one sampling campaign to the other, primarily reflect changes in volcanic activity or climatic parameters remains unclear. After correction for DTF, the Xe isotope composition of E-VQ (Eifel) and MV1 (Yellowstone) is compared to one another by normalizing the MV1 spectrum (Fig. 5D) to the ¹²⁹Xe excess of E-VQ. The brown shaded area shows the uncertainty envelope of the ¹²⁹Xe-normalized MV1 spectrum. Slightly higher excesses in fissiogenic Xe isotopes in MV1 compared to E-VQ could potentially arise from the greater ¹²⁹Xe excess in the upper mantle (source of Eifel gas) compared to deep mantle (source of the Yellowstone plume) (8, 17).

Fig. 7.

¹³²Xe^*/¹³⁶Xe^* versus ¹³⁴Xe^*/¹³⁶Xe^* systematics of Yellowstone and Eifel samples with the highest mantle contributions reported in this study (i.e., MV1 and E-VQ, respectively), before and after DTF correction. (A and B) The pure uranium (U-Xe) and plutonium (Pu-Xe) fissiogenic end-members (17) are shown as black squares. Once corrected for DTF, MV1 and E-VQ samples both appear most consistent with fissiogenic Xe isotopes deriving from uranium.

Fig. 8.. δ⁸⁴Kr/³⁶Ar versus δ¹⁵N for Iceland and Yellowstone samples with air-like Δ₃₀ values of 16‰ and higher.

(37) The observed trend implies that N loss causing δ¹⁵N variations occurs together with preferential Kr and Xe losses relative to Ar. This is in contrast with predictions based on solubilities obtained in ideal conditions, where both Kr and Xe are expected to be more soluble than Ar. Previous studies therefore suggested that this may represent degassing of air-saturated water under extreme temperature and pressure conditions (37), where gas solubilities deviate considerably from a behavior governed by Henry’s law (60). Here, we show that the observed trend is compatible with DTF of a gas phase that was stripped off a deep groundwater component with the elemental composition of AEW at ~50°C. This suggests that δ¹⁵N of N₂, like noble gases, is affected by the same DTF processes that appear to be ubiquitous in volcanic gas systems worldwide (43).

Fig. 9.. δ⁸⁴Kr/³⁶Ar versus δ³⁸Ar/³⁶Ar, δ⁸⁶Kr/⁸⁴Kr, and δ¹²⁸Xe/¹³⁰Xe isotope systematics and their comparison with predictions from the idealized model of subsurface fractionation.

(A to C) This model simulates the evolution of deep groundwater-derived noble gases due to progressive interaction with shallow groundwater (see MATLAB code; Supplementary Materials). Each separate line represents a different trajectory of equilibration between shallow groundwater and an initial gas phase representing deep groundwater-derived noble gases that have been variably fractionated because of DTF, depending on the advection/diffusion ratios [from pure advection by CO₂ without fractionation (green line, noted A) to extensive fractionation due to diffusion through CO₂ (orange line, noted B); fig. S8]. While this conceptual model readily explains elemental ratios varying between AEW (here shown for 20°C water) and air, without requiring mixing with unfractionated air during sampling, samples with δ⁸⁴Kr/³⁶Ar greater than AEW (e.g., Chile) likely require addition of extra Kr and Xe from the contribution of subduction-derived components, as suggested by carbon isotope systematics (fig. S2).

Fig. 10.. Schematic representation of our idealized model of subsurface fractionation, from deep to shallow groundwater levels.

The deep groundwater reservoir is considered to be extensively degassed as a result of protracted noble gas stripping by CO₂ bubbles. Depending on the advection/diffusion ratio [referred to as the Péclet number (Pe)], groundwater-derived noble gases in the deep unsaturated zone may then undergo variable extents of (i) DTF against CO₂ and (ii) advective transport toward the shallow groundwater level. Last, as the CO₂ plume accumulates shallow groundwater-derived noble gases, the gas phase gets closer to being in equilibrium with AEW. If it eventually reaches equilibrium, then its elemental and isotopic composition becomes indistinguishable from air. Intermediate stages of DTF and equilibration with shallow groundwater are required to explain the range of isotopic and elemental compositions observed in natural volcanic gas samples worldwide (Fig. 9).