Klein - Tarnas - Bach - 2020 - Abiotic Sources of Molecular Hydrogen On Earth
Klein - Tarnas - Bach - 2020 - Abiotic Sources of Molecular Hydrogen On Earth
Klein - Tarnas - Bach - 2020 - Abiotic Sources of Molecular Hydrogen On Earth
net/publication/339687849
CITATIONS READS
47 350
3 authors, including:
Some of the authors of this publication are also working on these related projects:
HYDROTHERMADEC - Geochemical and ecological impacts of hydrothermal processes at intra-oceanic island arcs, using the example of the Kermadec Arc View project
All content following this page was uploaded by Wolfgang Bach on 18 May 2021.
1811-5209/20/0016-019$2.50 DOI: 10.2138/gselements.16.1.19
T
he capacity for molecular hydrogen (H2 ) to hydrogenate oxygen and as in Hawai‘i (USA), where H 2
carbon is critical to the origin of life and represents the basis for all was identified spectroscopically
and estimated to constitute up
known life-forms. Major sources of H2 that strictly involve nonbiological to 1.8 mol% of the volcanic
processes and inorganic reactants include (1) the reduction of water during gas (Cruikshank et al. 1973).
the oxidation of iron in minerals, (2) water splitting due to radioactive decay, Giggenbach (1987) analyzed
fumarolic gases from White
(3) degassing of magma at low pressures, and (4) the reaction of water with Island (New Zealand) and found
surface radicals during mechanical breaking of silicate rocks. None of these that redox conditions and H 2 /H2O
processes seem to significantly affect the current global atmospheric budget ratios are governed by the SO2 –H2S
equilibrium during degassing from
of H2 , yet there is substantial H2 cycling in a wide range of Earth’s subsurface
the underlying magma. Studies of
environments, with multifaceted implications for microbial ecosystems. volcanic gas plumes at Mt. Etna
Keywords : molecular hydrogen, volcanic degassing, hydrothermal alteration, (Italy) and at the phonolitic lava
radiolysis, rock comminution lake at Mt. Erebus (Antarctica)
show a strong correlation of SO2
ABIOTIC SOURCES OF H2 ON EARTH with H 2 concentrations (Fig. 2),
supporting Giggenbach’s findings for White Island gases
Major sources of H2 in the lowest layer of Earth’s atmosphere (Aiuppa et al. 2011; Moussallam et al. 2012).
(i.e., the troposphere) include the photochemical oxida-
tion of CH4 and other simple organic carbon compounds Pressure is a significant control on magmatic H 2 degassing,
(1.98 × 1013 mol y−1), anthropogenic emissions from burning because at low pressure sulfur partitions more strongly into
fossil fuel and biomass (1.54 × 1013 mol y−1), and biological the vapor phase and the equilibrium
N2 fixation (2.98 × 1012 mol y−1) (Ehhalt and Rohrer 2009). 2H O + H S = SO + 3H (1)
2 2 2 2
For comparison, Worman (2015) estimates a flux of up
13
to 1 × 10 moles of abiotic H2 from submarine geologic is shifted to the right (Gaillard et al. 2011) (Fig. 2A). At Mt.
sources to the ocean per year globally. However, much of Etna, the low pressures of degassing led to a high time-
the H2 generated in marine settings is consumed at depth averaged H2 flux of 3.2 × 10 8 mol y−1 (Aiuppa et al. 2011).
and never reaches the atmosphere, suggesting substantial In submarine settings, magmatic degassing occurs at higher
H2 cycling in the deep ocean and subseafloor. Abiotic H2 pressures and should release little H2 (Gaillard et al. 2011).
has been detected in a broad array of subaerial and subma- Indeed, highly vesicular mid-ocean ridge basalt (MORB)
rine environments, including oceanic spreading centers, “popping rocks” from the Mid-Atlantic Ridge (which can
transform faults, passive margins, convergent margins, and explode upon decompression during sample recovery) have
intraplate settings. The main geologic sources responsible low H2 contents in the CO2 -dominated gas phase.
for H2 formation and the estimated annual fluxes per source
are summarized in Figure 1 and discussed in the following Holloway and O’Day (2000) argued for H2 release to the
sections. Processes that do not lead to H2 emanations at ocean during MORB crystallization because the ferric iron
Earth’s surface are not discussed. content of crystalline MORB is significantly higher than
that of MORB glass. Crystallization of MORB magma may
Magmatic H2 generate as much as 6.3 × 1012 mol H 2 y−1 globally, which
would exceed the annual MORB CO2 production (Holloway
Emission of magmatic gases containing H2 can be observed
and O’Day 2000). High H2 concentrations of hydrothermal
during, or immediately after, a volcanic eruption, such
vent fluids in the aftermath of volcanic events may support
the idea of transiently high H2 fluxes in ridge crest environ-
1 Woods Hole Oceanographic Institution ments, but such excursions may be due to hydrothermal
Department of Marine Chemistry and Geochemistry fluid–rock interactions (see below) rather than magmatic
266 Woods Hole Rd, MS# 8
processes (Lilley et al. 2003). For this reason, H2 produc-
Woods Hole, MA 02543, USA
E-mail: fklein@whoi.edu tion from submarine volcanism remains speculative and
2 Brown University requires further study.
Department of Earth, Environmental and Planetary Sciences
Globally, the amount of H2 emitted from subaerial volca-
324 Brook St., Box 1846
Providence, RI 02912, USA noes is on the order of 4.96 × 10 9 mol y−1 (Moussallam et
E-mail: jesse_tarnas@brown.edu al. 2012). Although this seems trivial in comparison with
3 University of Bremen the global annual tropospheric H2 flux of ~3.82 × 1013 mol
MARUM and Geoscience Department
Klagenfurter Str. 2, 28359 Bremen, Germany
E-mail: wbach@marum.de
S2
Hydrothermal H2 1 33 100
H2
Depth (m)
Large-scale abiotic H2 formation is evident in hydrothermal
S
CO
faults, fracture zones, subduction zones, passive margins, H2
2
iron oxidation
2 FeO + 0.5 O2(aq) = Fe2O3 (2)
1
then the oxidation of iron by oxygen-depleted water in
the root zone of a hydrothermal system can generate H2
as follows: 0
2 FeO + H2O = Fe2O3 + H2(aq) (3) 20 40 60 80 100 120 140 160 180 200
SO2 [ppm]
In this generalized reaction, FeO and Fe 2 O3 represent
the ferrous and ferric iron components in reactant and (U pper) Calculated compositions of volatiles in degas-
F igure 2
products minerals, respectively. Large parts of the oceanic sing magma as a function of pressure. Degassing at
lithosphere are composed of rocks with abundant ferrous high pressure favors CO2, CO, H2O, H2S, and S2 (in order of
decreasing abundance), whereas degassing at lower pressures
iron–bearing minerals that are prone to oxidation by
favors H2O, CO2, H2, and SO2. Note the two distinct depth scales
seawater. This process is surprisingly complex because for lithostatic pressure (brown: density is ~3 g/cm3) and hydrostatic
iron distribution during hydrothermal alteration involves pressure (blue: density is ~ 1 g/cm3). Modified from Gaillard et al.
several Mg–Fe minerals (having solid solutions), and their (2011) with permission from Springer Nature. (Lower) Graph of H2
(ppm) at Mt. Etna (Italy) showing linear correlation with SO2
compositions vary with temperature, protolith rock compo-
(ppm). Modified from A iuppa et al. (2011) with permission from John
sition, and water-to-rock ratio. Wiley and Sons.
1.0
200 Skelton et al. (2005) suggest it took only 105 to 106 years to
0.95 serpentinize much of the ocean–continent transition zone
during the late stage of continental breakup, which would
150
yield a H2 production rate of 1 × 1010 to 2 × 1011 mol y−1
XMg Brucite
0.90
at the Iberian margin. Other magma-poor passive margins
100 0.85 in North America, Africa, and Australia probably produced
similar amounts of H2 over similarly short timescales. It
0.80 appears that, during the finite timespan of active serpen-
50 tinization at magma-poor passive margins, these systems
0.75 may rival mid-ocean ridges in their annual H2 output.
ydrite
until they are released by dissolution or fracturing of the Dante
ite anh
epidote
olivine host. Because such fluid inclusions can form in any Bastille
–1 Pyrite Hulk
olivine-bearing rock type, H2 generation by serpentiniza- Pyrrhotite
anorth
Puffer
tion is likely more widespread than is currently perceived.
MEF
log aH2S(aq)
Sea Cliff
H2 in Mafic-Hosted Hydrothermal Systems –2 Lucky Strike
In ridge segments with robust magma supply, hydro-
MARK-1/2
thermal systems are usually hosted in mafic (basaltic or Piccard
gabbroic) rocks. Most mafic-hosted hydrothermal systems TAG
vent fluids with H 2 concentrations of ~0.05 to ~1 mM, –3 Hematite Edmond
similar to H 2 concentrations measured during basalt–
Magnetite
seawater hydrothermal laboratory experiments. Seyfried
and Ding (1995) proposed a four-step model to explain the
formation and subsequent buffering of H2 in mafic-hosted –4
hydrothermal systems. The first stage, during which H2 is
produced, involves the oxidation of primary iron sulfide
(pyrrhotite) and the iron (fayalite) component of olivine in
gabbro under relatively low water-to-rock mass ratios. The –5
–6 –5 –4 –3 –2 –1 0
fluid then evolves toward less reducing conditions where log aH2(aq)
H2(aq) and H 2 S (aq) activities are controlled by the mineral
assemblages pyrite–pyrrhotite–magnetite (stage 2), pyrite– F igure 4 Activity–activity diagram for H2(aq) and H2S (aq) illus-
magnetite (stage 3), and anhydrite–anorthite–clinozo- trating the stability fields of the minerals hematite,
magnetite, pyrite, and pyrrhotite at 400 °C and 50 MPa. Also shown
isite–pyrite–magnetite (stage 4). Figure 4 depicts measured
are measured compositions of vent fluids emanating from mafic-
H2(aq) and H2 S (aq) concentrations (activity coefficients are hosted hydrothermal systems. Activity coefficients of H2(aq) and
assumed to be close to unity) in vent fluids emanating H2S (aq) were assumed to be close to unity. Fluid analyses from
from mafic-hosted seafloor hydrothermal systems in the hydrothermal systems that experienced magmatic events are not
included. Most fluid compositions fall on the metastable extension
system Fe–O–S, which includes the minerals pyrrhotite,
of the pyrite–pyrrhotite equilibrium (thick dashed black line)
pyrite, magnetite, and hematite. Also shown are the H2(aq) suggesting that this assemblage controls the concentration of H2(aq)
and H2S (aq) activities controlled by the mineral assemblages in mafic-hosted vent fluids. The predicted H2(aq) and H2S (aq) activi-
of anhydrite–anorthite–clinozoisite and pyrite–pyrrhotite, ties corresponding to the assemblage anorthite–anhydrite–clinozo-
isite (epidote) are shown for comparison (gray dashed line). Phase
through the following reactions:
separated fluids emanating from the deepest known mafic-hosted
3CaAl2 Si2O8 (anorthite) + CaSO4 (anhydrite) + 4 H2(aq) hydrothermal system (Piccard) cooled from higher temperatures to
the venting temperature of 397 °C and appear to have equilibrated
= 2Ca 2Al3Si3O12 (OH) (clinozoisite) + 2H2O + H2 S (aq) (4) with the assemblage magnetite–pyrrhotite. Locations of vents as
follows: the Sully, Dante, Bastille, Hulk, and Puffer vents are all in
FeS (pyrrhotite) + H2 S (aq) = FeS2 (pyrite) + H2(aq) (5) the Main Endeavour Field, Juan de Fuca Ridge, northeast Pacific
Ocean; MEF = Main Endeavour Field; Sea Cliff vent (Gorda Ridge,
There is a remarkable spread in the H 2(aq) and H 2 S (aq) off west coast of California and Oregon); Lucky Strike and
MARK-1/2 vents from the Mid-Atlantic Ridge; Piccard (Beebe
concentrations of vent fluids. Differences in temperature
Hydrothermal Vent Field in the Mid-Cayman Spreading Centre);
and pressure in the root zone of hydrothermal systems— TAG (Trans-Atlantic Geotraverse on the Mid Atlantic Ridge);
and other effects like phase separation, cooling, and Edmond vent (Central Indian Ridge). Fluid data from L illey et al.
mineral precipitation—influence the compositions of vent (2003), Seewald et al. (2003), and McD ermott et al. (2018).
fluids. Cooling of hydrothermal fluid can generate H2 at
the expense of H2 S (McDermott et al. 2018) through the
reaction
favored (McDermott et al. 2018). While this process can
Fe ++ + 2H2 S = FeS2 + 2H + + H2(aq) (6) lead to significant H 2 enrichments in the vapor phase,
the H 2(aq) and H2 S (aq) concentrations at Piccard (red stars
However, most fluid compositions fall close to the in Fig. 4) may be controlled by the equilibrium between
metastable extension of pyrite–pyrrhotite equilibrium. If magnetite and pyrrhotite during cooling. In contrast, some
the assemblage anhydrite–anorthite–clinozoisite–pyrite– vent fluids show strong, apparently unbuffered, excursions
magnetite regulates H2(aq) and H 2 S (aq) activities in mafic- in H2(aq) and H2S (aq) related to magmatic events, such as the
hosted hydrothermal systems, then the H2 S (aq) -rich fluids ‘A’ vent on the East Pacific Rise. Hydrogen was likely formed
should have lower H 2(aq) concentrations and vice versa. during fluid–rock interaction at high temperatures, but
This supports the idea that the reaction of pyrrhotite with fluids did not have time to equilibrate with either magne-
H2 S (aq) to form pyrite can generate and control H2(aq) in tite, pyrrhotite, or pyrite before venting at the seafloor.
mafic-hosted hydrothermal systems. Because the pyrite– Note that fluid compositions from ‘A’ vent were omitted
pyrrhotite equilibrium is sensitive to temperature, the from Figure 4 for the sake of clarity.
strong correlation between H2(aq) and H 2 S (aq) would also
The estimated annual H 2 production from high-temper-
suggest that these species last equilibrated with pyrite and
ature basalt alteration along the global mid-ocean ridge
pyrrhotite at ~400 °C.
system is 0.4–6 × 1010 mol y−1, with a preferred value of
In contrast to most of the fluid compositions that are consis- 3 × 1010 mol y−1 (Worman 2015). These annual production
tent with the (metastable) pyrite–pyrrhotite equilibrium, rates were estimated by applying a H 2 concentration of
the composition of fluids from the Piccard Hydrothermal ~2 mM that is controlled by the quartz–fayalite–magnetite
Vent Field (also known as the Beebe Hydrothermal Vent (QFM) equilibrium to an annual flux of high-temperature
Field) on the Mid-Cayman Rise fall close to the magne- hydrothermal fluids (1013 kg) through the global mid-ocean
tite–pyrrhotite equilibrium (Fig. 4). There is evidence ridge system. Although a concentration of 2 mM or less
for phase separation at Piccard at temperatures >>491 °C is reasonable, we emphasize that QFM only serves as a
where volatile partitioning into the vapor phase is strongly theoretical approximation. Much of the H 2 generated
REFERENCES Klein F, Grozeva NG, Seewald JS (2019) Cienega Winery, central California.
Abiotic methane synthesis and serpen- Pure and Applied Geophysics 122:
Aiuppa A and 5 coauthors (2011) tinization in olivine-hosted fluid inclu- 376-391
Hydrogen in the gas plume of an open- sions. Proceedings of the National
vent volcano, Mount Etna, Italy. Journal Seewald JS, Cruse A, Saccocia P (2003)
Academy of Sciences of the United Aqueous volatiles in hydrothermal
of Geophysical Research: Solid Earth States of America 116: 17666-17672
116, doi: 10.1029/2011JB008461 fluids from the Main Endeavour Field,
Kumagai Y and 5 coauthors (2013) northern Juan de Fuca Ridge: temporal
Bach W, Edwards KJ (2003) Iron and Hydrogen production in gamma radiol- variability following earthquake
sulfide oxidation within the basaltic ysis of the mixture of mordenite and activity. Earth and Planetary Science
ocean crust: implications for chemolith- seawater. Journal of Nuclear Science and Letters 216: 575-590
oautotrophic microbial biomass produc- Technology 50: 130-138
tion. Geochimica et Cosmochimica Seyfried WE Jr, Ding K (1995) Phase
Acta 67: 3871-3887 LaVerne JA, Tandon L (2005) H 2 and Cl 2 equilibria in subseafloor hydrothermal
production in the radiolysis of calcium systems: a review of the role of redox,
Cruikshank DP, Morrison D, Lennon and magnesium chlorides and hydrox- temperature, pH and dissolved Cl on
K (1973) Volcanic gases: hydrogen ides. Journal of Physical Chemistry A the chemistry of hot spring fluids
burning at Kilauea Volcano, Hawaii. 109: 2861-2865 on mid-ocean ridges. In: Humphris
Science 182: 277-279 SE, Zierenberg RA, Mullineaux
Lefticariu L, Pratt LA, LaVerne JA,
D’Hondt S and 11 coauthors (2009) Schimmelmann A (2010) Anoxic pyrite LS, Thomson RE (eds) Seafloor
Subseafloor sedimentary life in the oxidation by water radiolysis products Hydrothermal Systems: Physical,
South Pacific Gyre. Proceedings of — a potential source of biosustaining Chemical, Biological, and Geological
the National Academy of Sciences energy. Earth and Planetary Science Interactions. Geophysical Monograph
of the United States of America 106: Letters 292: 57-67 Series 91, American Geophysical Union,
11651-11656 Washington DC, pp 248-272
Lilley MD, Butterfield DA, Lupton JE,
Dzaugis ME, Spivack AJ, Dunlea Olson EJ (2003) Magmatic events can Sherwood Lollar B, Onstott TC,
AG, Murray RW, D’Hondt S (2016) produce rapid changes in hydrothermal Lacrampe-Couloume G, Ballentine
Radiolytic hydrogen production in the vent chemistry. Nature 422: 878-881 CJ (2014) The contribution of the
subseafloor basaltic aquifer. Frontiers Precambrian continental lithosphere
in Microbiology 7, doi: 10.3389/ Lin L-H, Slater GF, Sherwood Lollar to global H 2 production. Nature 516:
fmicb.2016.00076 B, Lacrampe-Couloume G, Onstott 379-382
TC (2005) The yield and isotopic
Ehhalt DH, Rohrer F (2009) The tropo- composition of radiolytic H 2, a Skelton A, Whitmarsh R, Arghe F, Crill
spheric cycle of H 2 : a critical review. potential energy source for the deep P, Koyi H (2005) Constraining the rate
Tellus B: Chemical and Physical subsurface biosphere. Geochimica et and extent of mantle serpentinization
Meteorology 61: 500-535 Cosmochimica Acta 69: 893-903 from seismic and petrological data:
Fryer P (2012) Serpentinite mud volca- implications for chemosynthesis and
McDermott JM, Sylva SP, Ono S, German tectonic processes. Geofluids 5: 153-164
nism: observations, processes, and CR, Seewald JS (2018) Geochemistry of
implications. Annual Review of Marine fluids from Earth’s deepest ridge-crest Telling J and 13 coauthors (2015) Rock
Science 4: 345-373 hot-springs: Piccard hydrothermal comminution as a source of hydrogen
Gaillard F, Scaillet B, Arndt NT (2011) field, Mid-Cayman Rise. Geochimica et for subglacial ecosystems. Nature
Atmospheric oxygenation caused by a Cosmochimica Acta 228: 95-118 Geoscience 8: 851-854
change in volcanic degassing pressure. Miller HM and 5 coauthors (2017) Türke A, Nakamura K, Bach W (2015)
Nature 478: 229-232 Low temperature hydrogen produc- Palagonitization of basalt glass in the
Giggenbach WF (1987) Redox processes tion during experimental hydration flanks of mid-ocean ridges: implica-
governing the chemistry of fumarolic of partially-serpentinized dunite. tions for the bioenergetics of oceanic
gas discharges from White Island, New Geochimica et Cosmochimica Acta 209: intracrustal ecosystems. Astrobiology
Zealand. Applied Geochemistry 2: 161-183 15: 793-803
143-161 Moussallam Y and 5 coauthors (2012) Worman SL (2015) Global Rates of
Holloway JR, O’Day PA (2000) Production Hydrogen emissions from Erebus Free Hydrogen (H 2 ) Production by
of CO2 and H 2 by diking-eruptive volcano, Antarctica. Bulletin of Serpentinization and other Abiogenic
events at mid-ocean ridges: implications Volcanology 74: 2109-2120 Processes within Young Ocean Crust.
for abiotic organic synthesis and global PhD Thesis, Duke University, https://
Ribeiro JM, Lee C-TA (2017) An imbal- dukespace.lib.duke.edu/dspace/
geochemical cycling. International ance in the deep water cycle at subduc-
Geology Review 42: 673-683 handle/10161/10490
tion zones: the potential importance of
Klein F, Bach W, McCollom TM (2013) the fore-arc mantle. Earth and Planetary Worman SL, Pratson LF, Karson JA, Klein
Compositional controls on hydrogen Science Letters 479: 298-309 EM (2016) Global rate and distribution
generation during serpentinization of of H 2 gas produced by serpentinization
Sato M, Sutton AJ, McGee KA (1984) within oceanic lithosphere. Geophysical
ultramafic rocks. Lithos 178: 55-69 Anomalous hydrogen emissions from Research Letters 43: 6435-6443
the San Andreas fault observed at the