See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323703351

Geochemical Effects of Millimolar Hydrogen Concentrations in Groundwater:

An Experimental Study in the Context of Subsurface Hydrogen Storage

Article in Environmental Science and Technology · March 2018

DOI: 10.1021/acs.est.7b05467

CITATIONS READS

19 542

5 authors, including:

Marton Berta Frank Dethlefsen

Christian-Albrechts-Universität zu Kiel Christian-Albrechts-Universität zu Kiel
14 PUBLICATIONS 41 CITATIONS 35 PUBLICATIONS 398 CITATIONS

SEE PROFILE SEE PROFILE

Markus Ebert
Christian-Albrechts-Universität zu Kiel
55 PUBLICATIONS 870 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Uncertainties of geochemical modelling by geochemical codes and thermodynamic databases View project

KORA TP3.1 Hannover Südstadt View project

All content following this page was uploaded by Marton Berta on 13 November 2018.

The user has requested enhancement of the downloaded file.

 Article

Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX pubs.acs.org/est

Geochemical Effects of Millimolar Hydrogen Concentrations in

Groundwater: An Experimental Study in the Context of Subsurface
Hydrogen Storage
Marton Berta,* Frank Dethlefsen,* Markus Ebert, Dirk Schaf̈ er, and Andreas Dahmke
Department of Applied Geology, Aquatic Geochemistry and Hydrogeology, Institute of Geoscience, Kiel University,
Ludewig-Meyn-Straße 10, 24118 Kiel, Germany
*
S Supporting Information

ABSTRACT: Hydrogen storage in geological formations is

one of the most promising technologies for balancing major
fluctuations between energy supply from renewable energy
plants and energy demand of customers. If hydrogen gas is
stored in a porous medium or if it leaks into a shallow aquifer,
redox reactions can oxidize hydrogen and reduce electron
acceptors such as nitrate, FeIII and MnIV (hydro)oxides, sulfate,
and carbonate. These reactions are of key significance, because
they can cause unintentional losses in hydrogen stored in
porous media and they also can cause unwanted changes in the
composition of protected potable groundwater. To represent
an aquifer environment enclosing a hydrogen plume,
laboratory experiments using sediment-filled columns were
constructed and percolated by groundwater in equilibrium with high (2−15 bar) hydrogen partial pressures. Here, we show that
hydrogen is consumed rapidly in these experiments via sulfate reduction (18 ± 5 μM h−1) and acetate production (0.030 ± 0.006
h−1), while no methanogenesis took place. The observed reaction rates were independent from the partial pressure of hydrogen
and hydrogen consumption only stopped in supplemental microcosm experiments where salinity was increased above 35 g L−1.
The outcomes presented here are implemented for planning the sustainable use of the subsurface space within the ANGUS+
project.

■ INTRODUCTION
The energy production by wind turbines and solar panels
fluctuates due to variable wind velocity and solar radiation,
while the energy demand of consumers also varies creating a
significant challenge for the stability of future energy supplies
based mainly on renewable sources. One concept to address
this issue is the “power to gas” approach, where the generation
of a gas with high energy content uses the excess electrical
energy at times where the public energy demand is smaller than
the produced energy. Large scale gas storage is a potentially
necessary element in this concept, allowing the gas to be
available for reuse when the demand for energy exceeds the
supply. The overall difference between the amount of energy
supplied by renewable energy sources and the amount of
energy demanded usually varies in time, but this gap may reach
10 GWh−10 TWh, calculated as an example on a time scale of
days to months on a German regional scale power network.1
Electrochemical generation, storage, and consumption of
molecular hydrogen (in the following just “hydrogen”) has
the potential to balance such fluctuating gaps.1,2 The volume of
hydrogen corresponding to the amount of energy to be
balanced in an energy network ranges between millions to
billions of normal cubic meters.2 The subsurface may offer a
comparatively cost-efficient and safe option for storing such
large quantities of gaseous hydrogen.2−9 Moreover, sustainable,
renewable-based energy systems will likely require a series of
subsurface energy storage technologies simultaneously. Such
technologies include chemical (hydrogen, synthetic methane),
mechanical (compressed air), or heat energy storage, facing the
considerable amount of energy that needs to be stored and the
lack of other available storage technologies in the short or
medium term fulfilling volume, cost, and response time
requirements.1
Power-to-gas energy storage including geological hydrogen
storage is expected to play a potentially deciding role in future
energy networks, considering the following reasons. First, large-
scale development of renewable energy sources became a
cornerstone of all national energy policies following the Paris
Agreement,10 even if those actions are still unsatisfying
compared to the volume of development required to reach
climate protection goals already agreed on internationally.11
Second, consequently, estimations on for instance German
energy supplies project up to 100% renewable power
Received: October 25, 2017
Revised: March 10, 2018
Accepted: March 12, 2018
Published: March 12, 2018

© XXXX American Chemical Society A DOI: 10.1021/acs.est.7b05467

Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

Table 1. Hydrogen Oxidizing Reactions with Associated Gibbs Free Energy Yields under Standard Conditions and Neutral pH
and the Concentrations of Hydrogen Typical for Environments with Characteristic Metabolic Processes

reactions
ΔG0 [kJ·mol(H2)−1] H2 [nmol·L−1] references
aerobic hydrogen oxidation 2H2 + O2 → 2H2O −237 n.d. 73

hydrogen oxidation coupled to denitrification 5H2 + 2H+ + 2NO3− → N2 + 6H2O −224 0.03 25, 46, 73
producing N2

hydrogen oxidation coupled to nitrate reduction 4H2 + 2H+ + NO3− → NH4+ + 3H2O −150 <0.05 40, 73
producing NH3

hydrogen oxidation coupled to MnIV reduction H2 + MnO2 → Mn(OH)2 −163 <0.05 25, 46, 74

hydrogen oxidation coupled to FeIII reduction H2 + 2Fe(OH)3 → 2Fe(OH)2 + 2H2O −114 0.2 25, 46, 73

hydrogenotrophic sulfate reduction 4H2 + H+ + SO42− → HS− + 4H2O −57 1−2 25, 40, 46, 73

hydrogenotrophic methanogenesis 4H2 + H+ + HCO3− → CH4 + 3H2O −34 5−10 25, 40, 46, 73, 75

hydrogenotrophic acetogenesis 4H2 + H+ + 2HCO3− → CH3COO− + 4H2O −26 100< 25, 40, 75

generation by 2050.12 Third, if more than ca. 80% of the energy
is produced by renewable sources in a power network, then the
inclusion of large-scale storage facilities, like subsurface
hydrogen storage, is suggested to be a prerequisite for stable
and profitable operation.1 Overview studies on technical aspects
of subsurface hydrogen storage were published in the past few
years13−16 as a part of an increasing research and industrial
interest, but field or experimental studies specifying potential
environmental impacts or operational issues linked to storing
large amounts of pure hydrogen at elevated pressures in
geological media are missing so far. However, studies on
hydrogen-containing gas mixtures stored in porous town gas
reservoirs hint that hydrogenotrophic redox reactions poten-
tially consume hydrogen at such conditions.14,17−21 These
reactions may cause losses from the stored gas, and they may
also change the composition of reservoir water. Furthermore,
accidental leaking of hydrogen into shallow aquifers22 may
result in conditions where an enlarged hydrogen partial
pressure (i.e., equal to the hydrostatic pressure in case of a
pure hydrogen gas phase) will cause a high dissolved hydrogen
concentration in the shallow groundwater, probably initiating
typical redox reactions associated with hydrogen oxidation.
Hydrogen-driven redox reactions are predominantly micro-
biologically catalyzed23 and well-known from aquatic geo-
systems, where hydrogen is the electron donor and O2, NO3−,
FeIII and MnIV(hydr)oxides, SO42−, or dissolved CO2 are
usually the terminal electron acceptors (Table 1). These redox
reactions may produce NO2−, N2O, N2, NH4+, MnII, FeII, H2S,
CH3COOH, or CH4, which might be released into the pore
water or precipitated into various mineral phases. Reaction
products such as NO2−, H2S, or CH4 may have a negative effect
on the composition of the groundwater in terms of its usability
for other applications, e.g., drinking water production.24
Moreover, groundwater itself is usually a protected natural
good. In pristine aquifers, fermentative degradation of peptides,
saccharides, and further organic substances produces hydro-
gen,25 while other microorganisms consume it, usually resulting
in nanomolar concentrations of dissolved hydrogen in a steady
state.25−41 A competition for hydrogen may then result in the
development of a typical redox sequence (i.e., decreasing redox
B DOI: 10.1021/acs.est.7b05467
potential with increasing time or flow path length) because the
electron accepting processes coupled to hydrogen oxidation
with the higher energy yield (e.g., sulfate reduction) out-
compete the others (e.g., acetogenesis) by keeping the
hydrogen concentrations below a threshold concentration
required for that process (Table 1).42,43 Among other
possibilities, such reactive environments may be numerically
simulated assuming a dual substrate limitation by hydrogen
concentration on the one hand and the concentration of sulfate
or total dissolved inorganic carbon species on the other.
Furthermore, not only the occurrence but also the kinetics of
the redox reactions is usually dependent on the concentration
of dissolved hydrogen in those hydrogen-limited reactive
environments, and such processes are often modeled using
Monod kinetics.44−48
In situ remediation of contaminated aquifers may lead to
elevated dissolved hydrogen concentration, e.g., when applying
zerovalent iron (ZVI)49−51 or hydrogen releasing com-
pounds,52 in hydrogenotrophic denitrification systems,53−57 or
during the sulfidogenic treatment of wastewaters containing
dissolved metal ions.58,59 Increased dissolved hydrogen
concentration is thereby either an aim of the technology
(e.g., denitrification or sulfidogenic treatment) or a side-effect
(e.g., ZVI application for reductive dehalogenation). Usually,
those systems also show the establishment of redox sequences
known from pristine aquifers, and the kinetics of the occurring
redox reactions also depend on the hydrogen concentrations
created by hydrogen-producing primary reactions.46 Further-
more, the hydrogen concentration found in such reactive
environments is mostly limited to a few hundred micromoles
per liter.
In the context of the storage of pure gaseous hydrogen at
pressures between 30 and 400 bar60−62 in the subsurface,
dissolution of hydrogen may result in concentrations of several
millimoles per liter groundwater due to the roughly linear
correlation between increasing depth and increasing hydrostatic
pressure as well as between increasing pressure and increasing
gas solubility following Henry’s law. Also, an unintended
leakage of pure hydrogen gas into a shallow aquifer (up to 140
m depth) will result in a much higher dissolved hydrogen
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

Table 2. Chemical Composition of the Medium Grained Sandy Sedimenta

Nab Kb Cab Mgb Mntotb Fetotb FeIIc FeIIIc Sredd Corge Cinorge
3.68 8.8 23.2 0.797 0.121 3.91 <0.02 0.39 0.49 0.27 4.9
a
Conc. in g/kg. bAfter total digestion of sediment sample in hydrochloric acid, nitric acid, hydrofluoric acid, and perchloric acid. cExtractable after 21
days in 5 M HCl according to the methods of Heron et al.79 dAnalyzed as chromium reducible sulfur (CRS) similarly to the methods of Canfield et
al.80 eAnalyzed with Ströhlein Coulomat Type 702.

Table 3. Composition of the Groundwater and the Inflow Solution of the Column Experimenta
groundwater inflow solution
parameter unit n n
elec. cond. [μS cm−1] 721 ± 9 16 630 ± 38 7
pH [−] 7.2 ± 0.2 16 8.5 ± 0.3 8
O2 [mg L−1] 0.011 ± 0.007 12 n.m.
EHSHE [mV] −122 ± 59 16 n.m.
T [°C] 9.8 ± 0.2 16 20 ± 2
Na+ [mmol L−1] 0.72 ± 0.03 13 0.33 ± 0.08 21
K+ [mmol L−1] 0.04 ± 0.001 13 0.02 ± 0.01 21
Ca2+ [mmol L−1] 3.49 ± 0.06 13 1.2 ± 0.5 21
Mg2+ [mmol L−1] 0.234 ± 0.005 13 0.11 ± 0.03 21
Sitot [mmol L−1] 0.468 ± 0.008 13 0.21 ± 0.03 21
Fetotb [μmol L−1] 32 ± 7 13 0.3 ± 1.2 20
Mntotb [μmol L−1] 3.77 ± 0.06 13 0.5 ± 1.0 20
Cl− [mmol L−1] 0.81 ± 0.03 15 0.315 ± 0.007 22
SO42− [mmol L−1] 0.95 ± 0.04 15 0.33 ± 0.04 22
NO3− [μmol L−1] 6 ± 37 14 0.9 ± 2.0 22
PO43− [μmol L−1] <1.05 14 <1.05 22
F− [μmol L−1] <5.3 14 <5.3 22
Br− [μmol L−1] <1.25 14 <1.25 22
CH3COO− [μmol L−1] <3.33 14 30 ± 38c 25
CH4 [μmol L−1] 3±1 12 <1 20
TIC [mmol L−1] 5.4 ± 0.2 15 1.5 ± 0.2 22
TOCd [mmol L−1] n.m. 0.26 ± 0.09 22
NPOCe [mmol L−1] 0.4 ± 0.1 14 n.m.
a
Mean and standard deviation from n analyses; n.m.: not measured. Note: a table containing the average outflowing concentrations in the same data
structure can be found in the Supporting Information, Supporting Table 1. bDissolved Fe and Mn species were apparently oxidized and precipitated
during the storage of the raw groundwater collected from the field. These precipitates were filtered out when preparing the inflow solution so that
total Fe and Mn concentrations were significantly lower in the inflow solution than in the groundwater. cAcetate has not been added to the inflow
solution or to any other material used for the experiments when preparing the material and the setup. Its appearance in minor quantities is most
probably explained by acetogenic microbial activity taking place in the high-pressure mixing cell. dTOC was calculated as a difference of TC and TIC
measurements. eNPOC: nonpurgeable organic carbon; measurement of TC after acidifying (HCl) and stripping (O2) the sample for 5 min.

concentration than the nanomolar quantities usually present in
aquifers (Table 1). Elevated hydrogen concentrations will
probably result in a surplus in hydrogen supplies for the
microbial community, which usually strives for hydrogen.
Therefore, the microorganisms may start consuming hydrogen
rapidly through a series of consequent redox reactions46,63 or
through simultaneous redox processes64 without the establish-
ment of typical redox sequences. On the other hand, the
groundwater above a gas storage reservoir may have an elevated
salt content and a saltwater or brine intrusion might accompany
a gas leakage into a shallow aquifer. Elevated salinity potentially
inhibits the microbiologically catalyzed redox reactions due to a
limited halotolerance of the microorganisms present.65−72
In order to estimate the potential effects the subsurface
storage of hydrogen could have on groundwater, including the
effects following an unintended leakage into a shallow aquifer,
an improved understanding of hydrogenotrophic redox
reactions at high dissolved hydrogen concentration is required,
but investigations on those systems are missing so far, especially
regarding reaction kinetics. Available models on the spatial or
temporal zonation of redox processes arise in part from
C DOI: 10.1021/acs.est.7b05467
electron donor limitation, while after a gaseous hydrogen
intrusion into an aquifer the availability of hydrogen as electron
donor is not limiting any more. The knowledge of establishing
the sequence of redox reactions, their reaction kinetics, and
reaction products is necessary for any risk assessment or spatial
planning attempts and for modeling efforts at reservoir or
regional scales.7,8 Furthermore, monitoring and operation
strategies for a gas storage site also require knowledge about
the fate of the stored gas in shallow aquifers completed by an
estimation of the impact the stored gas may have through
volatile products of redox reactions (e.g., H2S or CH4).
The study presented here shows results of a high-pressure
column experiment using natural aquifer sediment and
groundwater equilibrated with hydrogen partial pressures of
2−15 bar, representing a scenario of hydrogen gas leakage into
a shallow or medium deep aquifer. The results enabled the
development of a simplified reaction model scheme describing
simultaneous redox reactions at elevated dissolved hydrogen
concentrations at flow through conditions, including reaction
kinetics and rate coefficients. Furthermore, results of a simple
testing routine for estimating the halotolerance of a microbial
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
community are shown. The outcome of the study is a first step
in an improvement of the understanding of hydrogen oxidizing
redox reactions in the context of hydrogen gas storage. The
contributions of this work will also be integrated into the
Compendium on Concepts and Methods for Quantitatively
Deducting the Spatial Demand of Subsurface Energy Storage
prepared by the ANGUS+ research project facilitating future
improvements in planning the sustainable use of the subsurface
space.6−8
■
MATERIALS AND METHODS
Experimental Setup. Sediment from a pristine shallow
Quaternary aquifer (WGS84: N54.0773, E10.0750, Table 2)
obtained from drilling cores between 4 and 12 mbgs was used
in the experiments. The sediment was homogenized and stored
wet in an Ar atmosphere to preserve the redox conditions as
well as possible. A pre-experiment conditioned the partially
oxidized sediment for approximately two months before the
high-pressure column experiment started using a flow through
batch reactor at atmospherical pressure and enriched by
hydrogen gas at approximately 1 bar.64 The batch experiments
for testing halotolerance contained unconditioned sediment.
Groundwater from the same well the sediment cores originated
from was sampled periodically (Table 3) and stored in HDPE
canisters before being utilized in the column experiment or in
gastight Al-impregnated bags (Tesseraux) for batch experi-
ments. Vacuum filtration (cellulose nitrate membrane 0.45 μm,
Sartorius) and dilution using deionized water (1:2.5) of the
groundwater prevented the formation of precipitates potentially
clogging the pumps and capillary tubes supplying the column
experiment. Equilibrating this initial solution of diluted
groundwater to various p(H2) in a flow through mixing cell
directly upstream the column generated in-line the inflow
solution for the high-pressure column experiment (Table 3).
The basic setup of the high-pressure column experiment
itself is described in detail elsewhere.76−78 In short, the high-
pressure column used (Ø = 4 cm, L = 50 cm, electrochemically
passivated titanium grade 2, with 4 ports along the flow path,
from Werner Kluge Engineering GmbH, Kiel, Germany) was
equipped with stainless steel and hastelloy fittings and tubes
(Swagelok, OH, USA) allowing an investigation in a gastight
system at total pressures of up to 100 bar. One HPLC pump
(Knauer GmbH, Berlin, Germany) pumped the initial solution
into a mixing cell where hydrogen (grade 5.0) was added at a
constant pressure monitored in the headspace by a
precalibrated pressure gauge (Swagelok, OH, USA). A second
HPLC pump pumped the generated inflow solution through
the column from the bottom to the top. A variable backpressure
valve at the end of the flow line ensured a total pressure in the
system of up to 10 bar higher than the applied gas partial
pressures in order to prevent degassing in the experimental
apparatus. In the experiments presented here, hydrogen gas at
p(H2) between 2 and 15 bar was applied during different
experimental stages (Supporting Information, Supporting
Figure 1). A pumping rate of 0.5 mL min−1 created a flow
speed of approximately 2 m d−1, assuming an effective porosity
of 0.3 ± 5% based on equivalent experiments,76,77 which results
in a total residence time of approximately 6.3 h. The total run
time of the column experiment was 242 days. A technical issue
with the HPLC pumps interrupted the flow through the
column at day 105 for 40 days between the 2 and 3 bar runs;
but in this phase, the total pressure in the column experiment
D DOI: 10.1021/acs.est.7b05467
Article
was kept constant, and after that discontinuity, the groundwater
percolation in the experiment was resumed.
A pressure-resistant sampling cell76−78 was used to sample
the in- and outflowing solutions as well as solutions from the
sampling ports along the flow path without discharging
pressure in the sediment column. After releasing the pressure
from the sampling cell, the water in it was sampled, treated, and
analyzed for cation, anion, and total as well as inorganic carbon
concentrations. While the above-mentioned parameters could
be determined precisely, taking a sample from a solution
saturated by a p(H2) above the atmospheric pressure resulted in
rapid hydrogen outgassing during pretests. Thus, before
samples for hydrogen analyses were extracted from the
sampling cell, the water to be sampled was diluted under the
column’s in situ water pressure within the sampling cell by
factors between 1:5 and 1:25. The dilution was carried out by
pumping well-defined amounts of the hydrogen free initial
solution into the sampling cell, simultaneously to the percolated
sample. This dilution lowered the hydrogen concentrations in
the sampling cell to below the saturation concentration at 1 bar
p(H2): this method delivered reasonable hydrogen concen-
trations and negligible losses in pretests. Pressure resistant
sensors (Corr Instruments, San Antonio, TX, USA) connected
to the sampling cell enabled the measurement of pH and
electrical conductivity during each sampling event at in situ
pressure conditions.
For the halotolerance test, 115 mL glass serum vials were
used as microcosms and filled with 50 g of untreated sediment
and 75 g of groundwater. An enrichment of the groundwater
with CaSO4 and NaCl salt adjusted a TDS level between 0.35
and 350 g L−1 (Table 4). A thorough sparging with pure
Table 4. TDS Level and Added Salt Concentration (in mg
L−1) in the Halotolerance Testing Schedule
TDS level CaSO4 NaCl
350 0 0
1100 375 375
3500 1575 1575
11 000 2000 8650
35 000 2000 32 650
110 000 2000 107 650
350 000a 2000 347 650
a
A minor fraction of the added salt did not dissolve.
hydrogen gas for several minutes ensured an initial dissolved
hydrogen concentration nearly in equilibrium with p(H2) = 1
bar. Furthermore, by sealing the vials using aluminum crimps
containing a Teflon plated butyl rubber septum as fast as
possible after the sparging, a nearly pure hydrogen atmosphere
in the headspace was prepared. The microcosms were stored
still and upside down in dark boxes at room temperature (20 ±
2 °C), and samples for hydrogen analysis were taken after
distinct time intervals from the free water phase after shaking
and opening a serum vial. Three separate batches were
prepared for each sampling event, which took place after 0, 6,
15, 23, and 48 days.
Sample Preparation and Analytics. Gas chromatography
(7890B GC with a TCD detector coupled to a 7697A
Headspace Sampler, Agilent Technologies) was used for
hydrogen analytics after a three interval linear calibration
from 25 μM to 10.5 mM based on defined volumes of
hydrogen gas (99.999 vol %) added by Hamilton syringes into
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

closed 20 mL headspace vials containing 5 mL of deionized

water, sealed with a Teflon-coated butyl rubber septum crimp,
stored upside down. Freshly prepared control standards
measured over several months showed a standard deviation
of 7.5% (n = 546). Hydrogen concentrations were measured in
5 mL samples as described for the standards using up to three
parallel and unfiltered samples, typically within an hour after
sampling. All other samples were filtered (0.2 μm, regenerated
cellulose, Sartorius) and filled into vials for cation, anion
(HDPE), and inorganic and organic carbon (glass, sealed with
the same crimps used for hydrogen samples) measurements.
Cation samples were acidified (cc. HNO3 s.p.), and all samples
were stored at 4 °C in the dark until the measurement. An ICP-
AES (Vista AX, Varian) analyzed the total elemental abundance
of Na, K, Ca, Mg, Fe, Mn, and Si. An ion chromatograph
analyzed (Metrohm 881) the anion (formate, acetate, sulfate,
chloride, nitrate, phosphate) concentrations, and a TIC-TOC
analyzer (multi N/C 2000 from Analytik Jena) measured the
total inorganic carbon (IC) and total carbon (TC) as well as
the nonpurgeable organic carbon (NPOC) concentrations, all
at the day of sampling. Ammonium concentrations were
measured photometrically, following DIN 38406E5. The
saturation index for, e.g., calcite was calculated as the ten-
based logarithm of the ratio between the ion activity product of
its dissolved components (IAP) and the corresponding
equilibrium constant for solubility (K) using PHREEQC81,82
and phreeqc.dat completed by a redox decoupled species for
acetate. Data storage and evaluation, including the model fitting
process, was done using MS Excel.

■
RESULTS AND DISCUSSION
Experimental Results. The column experiment started
using an inflow solution in equilibrium with a p(H2) of 5 bar,
and after 56 days, the hydrogen partial pressure was increased
to 15 bar for 18 days. Then, after the distribution of
geochemical parameters in the column had reached steady
state, the partial pressure of hydrogen was decreased back to 5
bar for 17 days followed by experimental phases with even
lower p(H2) of 2 and 3 bar for 16 and 42 days, respectively.
Finally, a third 5-bar period of 22 days completed the
experiment. The results indicated that the flow interruption
of 40 days between the 2-bar and 3-bar periods did not affect
the system significantly and, therefore, the time interval of the
flow interruption was neglected; the results are shown on a
continuous time scale (Figure 1). These results show a
continuous change between the in- and outflow regarding
dissolved concentrations of sulfate, acetate, total inorganic
carbon species (IC), calcium, and hydrogen, starting without
any detectable lag time at the beginning of the experiment.
Figure 1 demonstrates decreasing concentrations of sulfate (by
160 ± 100 μM; n = 77), calcium (500 ± 400 μM, n = 79), IC
(800 ± 400 μM, n = 73), and dissolved hydrogen (1700 ± 700
μM, n = 47), while the acetate concentration increased (180 ±
50 μM, n = 61). Furthermore, a sulfidic smell was clearly
identifiable in the solution flowing out of the column, and dark
precipitates, most probably iron and manganese sulfides or
possibly their carbonates and/or hydroxides, were also
observed in a transparent tube which discharged the water
from the outflow valve. An electron balance can be read from
Figure 4, as the modeled hydrogen concentration reflects values
calculated from the stoichiometry of electron consuming sulfate
reduction and acetogenesis, which are compared to the
measured hydrogen values.
E DOI: 10.1021/acs.est.7b05467
Figure 1. Dissolved concentrations of sulfate, acetate, hydrogen,
calcium, and total inorganic carbon species (IC) as well as the pH
value in the column’s in- and outflow solutions versus the experimental
run time in days. Vertical gray lines show the consecutive p(H2)
changes in the inflowing solution. (A technical note: the “15-bar”
period would only add 3 data points to the “Hydrogen” chart and 2
points to the “pH” chart, but it would, in exchange, degrade the clarity
of the whole figure. These points were therefore allowed to go off-scale
on those charts instead of removing them from the whole figure.)
These changes in the composition of groundwater were in a
good agreement with the assumption of an established
hydrogenotrophic microbial community apparently oxidizing
hydrogen mainly by reducing sulfate and reducing carbon
dioxide through acetogenesis. Moreover, the decrease in
calcium concentrations linked to increasing pH values along
the flow path suggested the precipitation of calcium carbonate,
which also contributed to the decrease in IC (Figure 2). The
dissolved Fetot concentrations stayed below the detection limit
of 0.02 mg L−1 in all samples while the concentration of
dissolved Mntot exceeded the detection limit of 0.01 mg L−1
only in a few samples, suggesting that FeIII and MnIII/IV
reducing processes were of minor importance. However, solely
measuring dissolved Fetot concentrations serves only as an
indicator for identifying FeIII-reduction processes, because
precipitating mineral phases (i.e., FeS, FeCO3) may limit the
solubility of FeII. The same is valid for any MnIII/IV reduction
processes. As the experiments are ongoing, mineralogical
evaluation of the solid phase was beyond the subject of this
work and, therefore, the role of FeIII and MnIII/IV reduction is
not conclusively evaluated. However, the comparatively small
FeIII content of the sediment used (0.55 ± 0.39 g kg−1)
supported the assumptions on a minor importance.
The pH usually increased from 8.7 to 9.6 along the flow path
and up to 11.8 during the 15-bar period of the experiment. The
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
Figure 2. Dissolved hydrogen, sulfate, total inorganic carbon species,
acetate, and calcium concentrations, pH, and electrical conductivity as
well as the saturation index81,82 of calcite for the solutions percolating
the high-pressure column plotted versus the residence time in the
experiment. Different profile lines represent sample series taken during
consecutive experimental phases using diluted groundwater equili-
brated with different hydrogen partial pressures as inflowing solution
(e.g., “5 bars (2nd profile)”) represents the second experimental phase
with 5 bar hydrogen partial pressure; see Figure 1. H2 in 15 bar
experiment not depicted.
concentrations of other groundwater constituents (i.e., sodium,
potassium, magnesium, chloride) showed no mentionable
differences between the in- and outflows of the column
experiment, apart from an unsteady Sitot increase (50 ± 80 μM,
n = 62). The increase of silica concentrations can indicate that
weathering of silicate aquifer minerals took place, possibly
enabled by the increasing pH. An overall consequence of the
observed hydrogeochemical processes was a decrease in the
electrical conductivity of the groundwater by 150 μS cm−1.
The composition of the inflow solution varied slightly due to
a repeated renewal of the stock in freshly produced ground-
water that was used for its preparation, reflecting the
background fluctuation of a natural aquifer. The variable H2
inflow concentration caused by different applied p(H2) shows
no correlation with these fluctuations. Furthermore, the extent
of concentration changes in dissolved species between the in-
and outflow generally shows no correlation to the hydrogen
inflow concentration, and variations in these changes did not
show a clear trend either. An exception is the loss in sulfate
concentration, where a temporarily decreasing tendency was
observed between day 56 (ΔSO42− = 80 μM) and day 78
(ΔSO42− = 22 μM), followed by a reestablishment to initial
values and then a slight increasing tendency at the beginning of
F DOI: 10.1021/acs.est.7b05467
Article
the 3-bar period of the experiment. However, as no connection
between the variations in sulfate decrease and the dissolved
hydrogen concentration in the inflow can be seen (Figure 5),
these fluctuations in sulfate reduction rates are possibly caused
by variations in microbial activity not controlled by any of the
monitored parameters, including minor fluctuations in
inflowing sulfate concentrations representing natural changes
in the groundwater-based solution used to percolate the
columns.
The concentration profiles along the flow path gathered
during all six experimental phases (Figure 2) support the
assumption that redox reactions, most probably catalyzed by
microorganisms, were responsible for the observed changes in
dissolved concentrations of hydrogen, sulfate, and acetate.
Moreover, the concentration profiles acknowledged the
independency of the reaction rates from the initial dissolved
hydrogen concentration derived from monitoring the in- and
outflow concentrations (Figure 1). All concentration profiles
show an average decrease in dissolved concentration of sulfate
from 0.33 to 0.22 mM, of IC from 1.45 to 0.70 mM, of calcium
from 1.13 to 0.84 mM, and of hydrogen by 1.2 mM as well as
an average increase in acetate concentration from 0.05 to 0.2
mM slowing down toward the end of the flow path. The results
imply that all reactions started immediately and concurrently
along the whole flow path. The simultaneous reaction progress
established throughout the total length of the column indicated
that a classical redox sequence, i.e., subsequent and separate
sulfate reducing and acetogenic zones, was not established in
the column experiment exhibiting hydrogen surplus at any time.
Furthermore, the almost linear decrease in hydrogen, sulfate,
IC, and calcium concentration suggests that the reactions
responsible for the concentration changes proceeded more or
less evenly along the total length of the flow path. Only the
acetate formation rate decreased toward the end of the column.
The concentration profiles provided an estimation of an
average net hydrogen consumption rate of 0.25 mM h−1, which
was nearly balanced by the 0.2 mM h−1 that was calculated on
the basis of the stoichiometry of hydrogenotrophic sulfate
reduction and acetogenesis. Although their rates could not be
quantified, FeIII and MnIII/IV reduction most probably also
contributed to the hydrogen consumption and they would
therefore explain a part of the hydrogen surplus. The rest of the
gap in this redox balance was probably caused by a loss of
hydrogen during the sampling procedure or by a diffusive loss
through pumping heads, tubes, connectors, etc. of the system
apparatus, reflecting the serious challenge to work with the
highly fugitive hydrogen gas dissolved at partial pressures much
higher than the ambient pressure. Hydrogen consuming
processes other than the discussed ones cannot be totally
excluded, but the results do not suggest any other relevant
reactions.
Methane (detection limit: 1 μM) or formate (detection limit:
4 μM) production were not detected in any of the samples,
although they were expected.83 The increasing pH value from
8.7 to 9.6 along the flow path of the column experiment might
have inhibited methane formation because pH levels above 7−9
can prevent methanogenic activity84,85 although studies at town
gas storage sites and hydrogen-limited aquifers have shown
methane formation at conditions comparable to the column
experiment.14,17−19 Usually, methane production is known to
take place at rates of up to a third of the acetate production
rates32 and even the presence of acetate itself in the column
experiment made methane production expectable.86 Therefore,
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
the high pH value characterizing the experiment probably in
combination with a lack of specialized microbes could be the
reason for suppressed methane formation. However, the
continuation of the lab experiment at neutral pH values
allowed methanogenesis to start.87 Interestingly, earlier
predictions assumed a groundwater acidification as a con-
sequence of a surplus in molecular hydrogen,21 which is clearly
denied by the presented experimental results.
Overall, the column experiment represents a flow through
system characterized by the surplus of dissolved hydrogen and
dominated by spontaneously starting hydrogen oxidizing
reactions, mainly parallel sulfate reduction and acetogenesis
catalyzed by microorganisms. Dissolved hydrogen was removed
almost completely during the 2-bar period of the experiment
only, indicated by a nonlinear decrease in hydrogen
concentration along the flow path (Figure 2) and small
concentrations of hydrogen in the outflow (between 27 and
158 μM) (Figure 1). However, the temporary depletion of
dissolved hydrogen did not stop acetate production, pH
increase, or calcium precipitation. The lack of dependency of
the observed reactions on the initial p(H2) or actually dissolved
H2 concentration indicated that the system reached quasi
steady-state conditions. Thus, the established microbial
community was apparently not sensitive to the pressure
variations or changing dissolved H2 concentration applied in
the experiment at least as long as a surplus of dissolved
hydrogen was provided.
The results of complementary batch experiments applying a
range of different total dissolved solids (TDS) concentrations
showed that high concentration of TDS inhibited hydrogen-
consuming redox reactions (Figure 3). The microcosms
Figure 3. Development of dissolved hydrogen concentrations
normalized to the initial total moles of hydrogen at different total
dissolved solids (TDS, mg L−1) concentrations. The black line
represents a blank series with quartz sand and deionized water; all
standard deviation bars were calculated from triplicate samples. Note
that the blank series identified no major artifacts or experimental side
effects (e.g., diffusive losses, sorption, etc.).
containing groundwater with TDS higher than 35 g L−1,
which corresponds to the average salinity of seawater, showed
constant hydrogen concentrations during a time span of at least
48 days. These concentrations were equivalent to the blank
experiments containing quartz sand and hydrogen-saturated
deionized water. In all microcosms containing TDS of 11 g L−1
or less, a significant and similar decrease in hydrogen
concentration was observed after a lag time of less than 6
days. The relative concentration of dissolved hydrogen reached
values between 60% and 30% after 48 days in these batch series.
Thereby, all these microcosms exhibited comparable hydrogen
G DOI: 10.1021/acs.est.7b05467
Article
consumption rates in the different experimental series.
Hydrogen consumption being inhibited by increasing the
salinity acknowledged that microbial processes were responsible
for hydrogen oxidation. This further suggests that TDS may
control or limit hydrogen oxidation, if microbial activity takes
place. Such a TDS limit will probably depend on the site-
specific microbial community of an aquifer.
The microbially mediated redox reactions starting sponta-
neously in the preconditioned sediment used in the column
experiment and after a short lag time in the halotolerance test
suggest that the unspecified natural microbial communities in
the aquifer material were ready to use high concentrations of
hydrogen as electron donor as well as sulfate and inorganic
carbon as electron acceptors. Furthermore, the results show
that at p(H2) higher than 2 bar the increasing dissolved
hydrogen concentration had no effect on the type nor on the
velocity of the established redox reactions. The results of the
halotolerance test combined with the column experiment
supported the idea of a fast establishment of microbially
catalyzed redox reactions in case of a hydrogen gas leakage into
shallow or medium deep aquifers, even in a scenario including
salt water intrusion accompanying the gas leakage. Storage of
hydrogen gas in porous saline aquifers will probably also initiate
these redox reactions, if the salt content in the aquifer does not
exceed some tens of grams per liter.
Kinetics. Two process models using different approaches
were consecutively developed for describing the observed
reactions. First, dual substrate Monod kinetics was initially
parametrized on the basis of literature values and it was then
fitted to the experimental results by varying the maximum
reaction rates and the Monod half-saturation constants (Table
5, Figure 4) for describing microbial metabolic processes
potentially limited by substrate availability. However, Monod
kinetics appears as either zeroth or first order kinetics in the
case that steady-state microbial activity can be assumed and if
reactant concentrations are significantly higher or lower than
the Monod half-saturation constants, respectively.45,88,89 Thus,
a simplified combination of (pseudo) zeroth and (pseudo) first
order rate equations and stoichiometric balance equations
represented the other kinetic process modeling approach
discussed here.
The first modeling approach based on dual substrate Monod
kinetics81 can describe the experimental data as illustrated in
Figure 4 using eqs 1−3 with the parameter values fitted to the
experimental results (Table 5). Thereby, consumption of
dissolved hydrogen and IC were defined on the basis of
stoichiometric balance equations using eqs 4 and 5,
respectively. Thus, unspecific hydrogen loss in the experimental
setup was not overinterpreted and carbonate precipitation,
which could not be quantified due to the lack of mineral
analyses, was regarded.
d[Ca 2 +]
− = kPrec
dt (1)
d[SO4 2 −] [H 2] [SO4 2 −]
− = vmaxSR × ×
dt MK H2−SR + [H 2] MK SO4 2 − + [SO4 2 −]
(2)
d[Ac−] [H 2] [IC]
− = vmaxAcG × ×
dt MK H2−AcG + [H 2] MK IC + [IC]
(3)
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

Table 5. Literature and Experimental Parameters for the Calculation of the Monod Kinetics Regarding Sulfate Reduction (SR)
and Acetogenesis (AcG), Such as the Maximal Reaction Rate (vmax) and the Monod Half-Saturation Constants (MK) for the
Corresponding Substratesa
vmax SR (h−1) MK SO42− (mM) MK H2−SR (mM) vmax AcG (h−1) MK IC (mM) MK H2−AcG (mM)
Literature Values
Boudreau and Westrich44 1.6
Dale, Regnier, and Van Cappellen45 0.002988 1 0.00001 0.002052 0.0001
Lovley and Goodwin46 0.0014
Watson, Oswald, Mayer, Wu, and Banwart47 0.000324 0.16 0.001
Xu, Jaffé, and Mauzerall48 0.23 0.0029
selected starting value before fitting process 0.002988 1 0.00001 0.002052 3 0.0001

Fitted to Experimental Result

2 bar model 0.05 1 0.00001 0.55 17 0.0001
3 bar model 0.12 1 0.00001 0.6 17 0.0001
5 bar (1st model) 0.072 1 0.00001 0.5 17 0.0001
5 bar (2nd model) 0.075 1 0.00001 0.55 17 0.0001
5 bar (3rd model) 0.12 1 0.00001 0.6 17 0.0001
15 bar model 0.052 1 0.00001 0.27 17 0.0001
averaged results 0.07 ± 0.04 1 0.00001 0.5 ± 0.2 17 0.0001
a
Note that all MK values for hydrogen are 4 to 5 orders of magnitude lower than the dissolved hydrogen concentrations in the high-pressure
experimental columns (Figure 2), representing concentrations expected to take place in an aquifer after a gaseous hydrogen intrusion took place.

Figure 4. Results of the simplified models using first order kinetics for acetate production and zeroth order kinetics for sulfate reduction and Ca-
carbonate precipitation (colored lines) and results of the model based on dual substrate Monod kinetics (black dotted lines lying perfectly on the
colored lines) and measured dissolved concentrations of hydrogen, sulfate, IC, calcium, and acetate (symbols) along the flow path profile in the high-
pressure column experiment during different experimental phases. The residence time was calculated using mean flow and porosity; dispersion was
neglected.

d[H 2] d[Ac−] d[SO4 2 −] Following the second-simplifying-modeling approach, sulfate

− =4 +4 reduction was calculated by zeroth order kinetics (eq 6) and
dt dt dt (4)
acetogenesis was described by pseudo first order kinetics in
dependence of IC (eq 7). Hydrogen and IC decrease as well as
d[IC] d[Ac−] d[Ca 2 +] carbonate precipitation were calculated in the same way as it
− =2 +
dt dt dt (5) had been done in the Monod kinetic modeling approach (eqs 4,
H DOI: 10.1021/acs.est.7b05467
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

5, and 1, respectively). The kinetic rate constants for sulfate several orders of magnitude higher than the corresponding
reduction (kSR), acetogenesis (kAcG), and carbonate precip- Monod half-saturation constants. Moreover, no literature values
itation (kPrec) were fitted to concentration profiles of Ca2+, are known for the half-saturation coefficient “MK IC”. Overall,
SO42−, acetate ([Ac−]), IC, and dissolved hydrogen. This for further applications, we suggest the more robust assumption
simplified model was equally able to represent the concen- of (pseudo) first and zeroth order kinetics, as these were
tration profiles (Figure 4) using rate coefficients for sulfate directly supported by the experimental results. For including
reduction and acetogenesis between 9 and 26 μM h−1 (average further or different pathways of the reduction of inorganic
18 ± 5) and 0.015 and 0.033 h−1 (average 0.030 ± 0.006), carbon species, eqs 4 and 5 can be extended by terms
respectively (Table 6). The fitted calcite precipitation rate representing reactions such as methane or formate production.
coefficients varied between 39 and 88 μM h−1 (average 57 ± Figure 5 shows the (pseudo) first and zeroth order rate
17). coefficients for acetogenesis, sulfate reduction, and carbonate
precipitation, respectively, gathered by fitting the model and
d[SO4 2 −]
− = k SR plotted against the initial dissolved hydrogen concentration.
dt (6) For the rate coefficients, the model input data were the six
concentration profiles along the flow path as well as all
d[Ac−] concentration differences between the inflow and outflow
− = kAcG × [IC]
dt (7) solution. The averaged values considering the in- and outflow
concentrations from each experimental period are mostly in
Table 6. (Pseudo) Zeroth Order Rate Coefficients for good agreement with the results provided by the concentration
Hydrogenotrophic Sulfate Reduction (kSR) and calcium profiles. Overall, the rate coefficients may vary by a factor of 2
carbonate precipitation (kPrec) and Pseudo First Order Rate to 3, but they indicated no dependency on the initial hydrogen
Coefficient with Respect to IC for Acetogenesis (kAcG) concentration, suggesting a quasi-steady state of the microbial
kSR (μM·h−1) kAcG (h−1) kPrec (μM·h−1)
community. Moreover, no recognizable trend in the microbial
activity versus the experimental run time (Figure 1) suggests no
average 18 ± 5 0.030 ± 0.006 57 ± 17
overall growth in the active biomass. The mean specific sulfate
reduction rate of 42 nM·cmsediment−3·h−1 or the mean
2 bar model 9 0.030 55
3 bar model 24 0.033 72
acetogenesis of 85 nM·cmsediment−3·h−1, which were calculated
5 bar (1st model) 17 0.028 40
from the column experiment, are smaller than rates described
5 bar (2nd model) 17 0.030 47
for other reactive aqueous environments exhibiting high
5 bar (3rd model) 26 0.033 88 dissolved hydrogen concentrations. Experimental studies58,90,91
15 bar model 13 0.015 39 on bioreactors yielded sulfate reduction rates between 3060
and 79 500 nM·cmsediment−3·h−1. However, a study90 on
acetogenesis in a Cretaceous shale and sandstone environment
The model using the combination of (pseudo) first or zeroth showed rates reaching from 110 to 160 nM·cmsediment−3·h−1
order reaction kinetics exactly reproduced the model curves parallel to sulfate reduction. While the acetogenesis rate was at
calculated using Monod kinetics and, therefore, both modeling least in a similar range, the sulfate reduction was 2 or 3 orders
approaches could be equally suitable. However, the Monod of magnitude smaller in the column experiment. Several reasons
kinetics model is strongly under-parametrized due to the might be responsible for the comparatively low metabolic
comparatively small changes in the total dissolved inorganic activity, e.g., a lack of nutrients (Table 3), composition of the
carbon species and sulfate concentrations as well as due to the established microbial community, or the comparatively high pH
constant availability of hydrogen in concentrations that were value.

Figure 5. Rate coefficients describing sulfate reduction, acetate formation, and calcium carbonate precipitation in the column experiment versus the
initial dissolved concentration of hydrogen. Vertical and horizontal error bars represent standard deviations in the measured hydrogen
concentrations and in reaction rate constants fitted for every in- and outflow concentration (“_in−out”), respectively. The larger symbols (“_port”)
represent the six fits based on concentrations over the flow path (Table 5, Figure 4).

I DOI: 10.1021/acs.est.7b05467
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
The Monod model as well as the simplified zeroth and first
order model supported the assumption that the simultaneously
proceeding and microbially catalyzed hydrogenotrophic sulfate
reduction and acetogenesis were the main redox processes
established in the column experiment, accompanied by
carbonate precipitation induced by the increasing pH value.
Nevertheless, sulfate reduction by acetate oxidation is also a
well-known microbial metabolic process86 and a model
considering this process instead of direct hydrogenotrophic
sulfate reduction could also be fitted to the results of the
column experiment. However, such a model would be under-
parametrized using the available data set, because acetotrophic
sulfate reduction would consume acetate and produce carbon
dioxide, which is then reduced again to acetate by oxidizing
hydrogen. Consequently, the acetate oxidation rate would be
twice as high as the sulfate reduction rate for stoichiometric
reasons and the rate of acetogenesis would increase by the same
value, resulting in identical model results. The experimental
results support a parallel and not a sequential occurrence of
acetogenesis and sulfate reduction, as both processes started
immediately at the beginning of the flow path, thereby
providing no evidence for sulfate reducing processes other
than hydrogenotrophic sulfate reduction. However, an
appearance of acetotrophic sulfate reduction cannot be totally
ruled out.
Implications. The main purpose of this study is to support
the planning, operation, and monitoring of hydrogen storage
facilities as well as to contribute to planning the sustainable use
of the subsurface space. An important contribution to these
workflows is the investigation of a hydrogen-rich groundwater
environment in which simultaneously proceeding redox
reactions were identified, instead of a redox sequence consisting
of separate metabolic zones. This poses a sharp contrast
compared to the hydrogen-limited groundwaters in pristine or
contaminated aquifers known from the literature so far. The
presented findings can be relevant for large scale modeling
efforts, design of monitoring networks at hydrogen gas storage
sites, or laboratory investigations serving a site specific risk
assessment. Using pressure-independent elementary kinetics
instead of sophisticated Monod rate equations in large scale
transport reaction models will probably save computational
capacity and may enable easier scenario modeling. Zeroth or
simple first order kinetics might overestimate the metabolic rate
in the case of exhausting reaction partners, but this effect
should be small at conditions with high initial reactant
concentrations. Furthermore, the independence of the observed
redox reactions from the dissolved hydrogen concentration
implies in a converse argument that laboratory investigations on
hydrogen leakage scenarios can be performed at system
pressures lower than in situ conditions, which will drastically
reduce laboratory efforts. Thereby, the presented halotolerance
test provides an easy procedure to estimate the effect of a salt
water intrusion accompanying a hydrogen leakage into shallow
or medium deep aquifers, if the natural material from a
particular site is available. The results of the halotolerance test
should not be transferred directly to any other groundwater
environment, because deep saline aquifers might have a well-
adapted and highly halotolerant microbial community possibly
enabling the consumption of newly introduced hydrogen even
if the salinity of the groundwater in such an aquifer is higher
than 35 g L−1. However, in less saline aquifers, a rapid
hydrogenotrophic sulfate reduction as well as a CO2 reduction
has to be expected, and the reaction products will modify the
J DOI: 10.1021/acs.est.7b05467
Article
groundwater and the stored gas. Therefore, targets of a
monitoring network covering a hydrogen gas storage site in
order to detect a gas leakage should be focused on the effects
reactions can have on groundwater (e.g., appearance of acetate,
depletion of sulfate, decreasing electric conductivity) instead of
the stored gas initially dissolved in a shallow aquifer.
■
*
ASSOCIATED CONTENT
S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.est.7b05467.
Table of experimental parameters; average hydrogen
concentrations in different experimental runs (PDF)
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail: mb@gpi.uni-kiel.de (M.B.).
*E-mail: frank.dethlefsen@ifg.uni-kiel.de (F.D.).
ORCID
Marton Berta: 0000-0001-8467-2792
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
We would like to thank the funding provided by the German
Federal Ministry of Education and Research (BMBF) for the
ANGUS+ project (03EK3022), as well as the support of the
Project Management Jülich (PTJ). Furthermore, we would like
to thank the five anonymous reviewers for suggestions to
improve this paper.
■
REFERENCES
(1) Sterner, M.; Stadler, I. Energiespeicher − Bedarf, Technologien,
Integration; Springer-Verlag: Berlin Heidelberg, 2014; p 758.
(2) Pfeiffer, W. T.; Al Hagrey, S. A.; Köhn, D.; Rabbel, W.; Bauer, S.
Porous media hydrogen storage at a synthetic, heterogeneous field site
- Numerical simulation of storage operation and geophysical
monitoring. Environ. Earth Sci. 2016, 75, 1177.
(3) Taylor, J. B.; Alderson, J. E. A.; Kalyanam, K. M.; Lyle, A. B.;
Phillips, L. A. Technical and economic-assessment of methods for the
storage of large quantities of hydrogen. Int. J. Hydrogen Energy 1986,
11 (1), 5−22.
(4) Evans, D. J. An appraisal of underground gas storage technologies
and incidents, for the development of risk assessment methodology; RR
605; Crown: Norwich, 2008.
(5) Gao, D.; Jiang, D. F.; Liu, P.; Li, Z.; Hu, S. G.; Xu, H. An
integrated energy storage system based on hydrogen storage: Process
configuration and case studies with wind power. Energy 2014, 66,
332−341.
(6) Bauer, S.; Beyer, C.; Dethlefsen, F.; Dietrich, P.; Duttmann, R.;
Ebert, M.; Feeser, V.; Gorke, U.; Kober, R.; Kolditz, O.; Rabbel, W.;
Schanz, T.; Schafer, D.; Würdemann, H.; Dahmke, A. Impacts of the
use of the geological subsurface for energy storage: an investigation
concept. Environ. Earth Sci. 2013, 70 (8), 3935−3943.
(7) Bauer, S.; Pfeiffer, T.; Boockmeyer, A.; Dahmke, A.; Beyer, C.
Quantifying induced effects of subsurface renewable energy storage.
Energy Procedia 2015, 76, 633−641.
(8) Kabuth, A.; Dahmke, A.; Beyer, C.; Bilke, L.; Dethlefsen, F.;
Dietrich, P.; Duttmann, R.; Ebert, M.; Feeser, V.; Gorke, U. J.; Kober,
R.; Rabbel, W.; Schanz, T.; Schäfer, D.; Würdemann, H.; Bauer, S.
Energy storage in the geological subsurface: dimensioning, risk analysis
and spatial planning: the ANGUS plus project. Environ. Earth Sci.
2017, 76, 23.
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
(9) Dethlefsen, F.; Beyer, C.; Feeser, V.; Köber, R. Parameterizability
of processes in subsurface energy and mass storage. Environ. Earth Sci.
2016, 75 (10), 1−25.
(10) United Nations. Report of the Conference of the Parties on its
twenty-first session, held in Paris from 30 November to 13 December 2015;
FCCC/CP/2015/10; United Nations Framework Convention on
Climate Change: Paris−Le Bourget, France, 2016; p 36.
(11) International Renewable Energy Agency. Perspectives for the
energy transition: Investment needs for a low-carbon energy system;
International Renewable Energy Agency (IRENA): Abu Dhabi, 2017;
p 204.
(12) UBA. Energieziel 2050:100% Strom aus erneuerbaren Quellen;
Umweltbundesamt: Dessau-Roßlau, Germany, 2010; p 196.
(13) Audigane, P.; Ebrahimiyekta, A.; Pichavant, M. Evaluation of
hydrogen migration and geochemical reactivity into underground. In
6th International Conference on Fundamentals & Development of Fuel
Cells, Toulouse, France, 2015.
(14) Panfilov, M. Underground and pipeline hydrogen storage. In
Compendium of Hydrogen Energy; Gupta, R., Basile, A., Veziroğlu, T.
N., Eds.; Woodhead Publishing: Oxford, 2016; pp 91−115.
(15) Bünger, U.; Michalski, J.; Crotogino, F.; Kruck, O. 7-Large-scale
underground storage of hydrogen for the grid integration of renewable
energy and other applications. In Compendium of Hydrogen Energy;
Woodhead Publishing: Oxford, 2016; pp 133−163.
(16) Gniese, C.; Bombach, P.; Rakoczy, J.; Hoth, N.; Schlömann, M.;
Richnow, H.-H.; Krüger, M. Relevance of deep-subsurface micro-
biology for underground gas storage and geothermal energy
production. In Geobiotechnology II: Energy resources, subsurface
technologies, organic pollutants and mining legal principles; Schippers,
A., Glombitza, F., Sand, W., Eds.; Springer Berlin Heidelberg: Berlin,
Heidelberg, 2014; pp 95−121.
(17) Amigáň, P.; Greksák, M.; Kozánková, J.; Buzek, F.; Onderka, V.;
Wolf, I. Methanogenic bacteria as a key factor involved in changes of
town gas stored in an underground reservoir. FEMS Microbiol. Lett.
1990, 73 (3), 221−224.
(18) Panfilov, M. Underground storage of hydrogen: in situ self-
organisation and methane generation. Transp. Porous Media 2010, 85
(3), 841−865.
(19) Buzek, F.; Onderka, V.; Vancura, P.; Wolf, I. Carbon-isotope
study of methane production in a town gas-storage reservoir. Fuel
1994, 73 (5), 747−752.
(20) Morozova, D.; Zettlitzer, M.; Let, D.; Würdemann, H.
Monitoring of the microbial community composition in deep
subsurface saline aquifers during CO2 storage in Ketzin, Germany.
Energy Procedia 2011, 4, 4362−4370.
(21) Reitenbach, V.; Ganzer, L.; Albrecht, D.; Hagemann, B.
Influence of added hydrogen on underground gas storage: a review
of key issues. Environ. Earth Sci. 2015, 73 (11), 6927−6937.
(22) Evans, D. J. A review of underground fuel storage events and
putting risk into perspective with other areas of the energy supply
chain. In Underground gas storage: worldwide experiences and future
development in the UK and Europe; Evans, D. J., Chadwick, R. A., Eds.;
The Geological Society: London, 2009; Vol. 313, pp 173−216.
(23) Truche, L.; Jodin-Caumon, M. C.; Lerouge, C.; Berger, G.;
Mosser-Ruck, R.; Giffaut, E.; Michau, N. Sulphide mineral reactions in
clay-rich rock induced by high hydrogen pressure. Application to
disturbed or natural settings up to 250 degrees C and 30 bar. Chem.
Geol. 2013, 351, 217−228.
(24) German Drinking Water Act. Trinkwasserverordnung, Verordnung
über die Qualität von Wasser für den menschlichen Gebrauch In BGBl. I S.
2977 and 3154, 2001.
(25) Hoehler, T. M.; Alperin, M. J.; Albert, D. B.; Martens, C. S.
Thermodynamic control on hydrogen concentrations in anoxic
sediments. Geochim. Cosmochim. Acta 1998, 62 (10), 1745−1756.
(26) Zinder, S. H. Physiological ecology of methanogens. In
Methanogenesis: Ecology, Physiology, Biochemistry & Genetics; Ferry, J.
G., Ed.; Springer US: Boston, MA, 1993; pp 128−206.
(27) Chapelle, F. H.; Lovley, D. R. Competitive-exclusion of sulfate
reduction by Fe(III)-reducing bacteria - a mechanism for producing
K DOI: 10.1021/acs.est.7b05467
Article
discrete zones of high-Iron ground-water. Groundwater 1992, 30 (1),
29−36.
(28) Jakobsen, R.; Albrechtsen, H.-J.; Rasmussen, M.; Bay, H.; Bjerg,
P. L.; Christensen, T. H. H2 concentrations in a landfill leachate plume
(Grindsted, Denmark): in situ energetics of terminal electron acceptor
processes. Environ. Sci. Technol. 1998, 32 (14), 2142−2148.
(29) Kotelnikova, S. Microbial production and oxidation of methane
in deep subsurface. Earth-Sci. Rev. 2002, 58 (3−4), 367−395.
(30) Chapelle, F. H.; O’Neill, K.; Bradley, P. M.; Methe, B. A.; Ciufo,
S. A.; Knobel, L. L.; Lovley, D. R. A hydrogen-based subsurface
microbial community dominated by methanogens. Nature 2002, 415
(6869), 312−315.
(31) Conrad, R.; Goodwin, S.; Zeikus, J. G. Hydrogen metabolism in
a mildly acidic lake sediment (Knaack Lake). FEMS Microbiol. Lett.
1987, 45 (4), 243−249.
(32) Conrad, R. Contribution of hydrogen to methane production
and control of hydrogen concentrations in methanogenic soils and
sediments. FEMS Microbiol. Ecol. 1999, 28 (3), 193−202.
(33) Lovley, D. R.; Chapelle, F. H.; Woodward, J. C. Use of dissolved
H2 concentrations to determine distribution of microbially catalyzed
redox reactions in anoxic groundwater. Environ. Sci. Technol. 1994, 28
(7), 1205−1210.
(34) Chapelle, F. H.; Haack, S. K.; Adriaens, P.; Henry, M. A.;
Bradley, P. M. Comparison of eH and H2 measurements for delineating
redox processes in a contaminated aquifer. Environ. Sci. Technol. 1996,
30 (12), 3565−3569.
(35) Heimann, A. C.; Blodau, C.; Postma, D.; Larsen, F.; Viet, P. H.;
Nhan, P. Q.; Jessen, S.; Duc, M. T.; Hue, N. T. M.; Jakobsen, R.
Hydrogen thresholds and steady-state concentrations associated with
microbial arsenate respiration. Environ. Sci. Technol. 2007, 41 (7),
2311−2317.
(36) Jørgensen, B. B. Bacteria and Marine Biogeochemistry. In
Marine Geochemistry; Schulz, H. D., Zabel, M., Eds.; Springer Berlin
Heidelberg: Berlin, Heidelberg, 2000; pp 173−207.
(37) Smith, N. J. P.; Shepherd, T. J.; Styles, M. T.; Williams, G. M.
Hydrogen exploration: a review of global hydrogen accumulations and
implications for prospective areas in NW Europe. In Petroleum geology:
north-west Europe and global perspectives: proceedings of the 6th
Petroleum Geology Conference; Dore, A. G., Vining, B. A., Eds.;
Geological Society of London: London, 2005; pp 349−358.
(38) Telling, J.; Boyd, E. S.; Bone, N.; Jones, E. L.; Tranter, M.;
MacFarlane, J. W.; Martin, P. G.; Wadham, J. L.; Lamarche-Gagnon,
G.; Skidmore, M. L.; Hamilton, T. L.; Hill, E.; Jackson, M.; Hodgson,
D. A. Rock comminution as a source of hydrogen for subglacial
ecosystems. Nat. Geosci. 2015, 8 (11), 851−855.
(39) Lin, L.-H.; Slater, G. F.; Sherwood Lollar, B.; Lacrampe-
Couloume, G.; Onstott, T. C. The yield and isotopic composition of
radiolytic H2, a potential energy source for the deep subsurface
biosphere. Geochim. Cosmochim. Acta 2005, 69 (4), 893−903.
(40) Cordruwisch, R.; Seitz, H. J.; Conrad, R. The capacity of
hydrogenotrophic anaerobic-bacteria to compete for traces of
hydrogen depends on the redox potential of the terminal electron-
acceptor. Arch. Microbiol. 1988, 149 (4), 350−357.
(41) Hoehler, T. M.; Alperin, M. J.; Albert, D. B.; Martens, C. S.
Field and laboratory studies of methane oxidation in an anoxic marine
sediment: Evidence for a methanogen-sulfate reducer consortium.
Global Biogeochemical Cycles 1994, 8 (4), 451−463.
(42) Gujer, W.; Zehnder, A. J. B. Conversion processes in anaerobic
digestion. Water Sci. Technol. 1983, 15 (8−9), 127.
(43) Rotiroti, M.; Jakobsen, R.; Fumagalli, L.; Bonomi, T.
Considering a threshold energy in reactive transport modeling of
microbially mediated redox reactions in an arsenic-affected aquifer.
Water 2018, 10 (1), 90.
(44) Boudreau, B. P.; Westrich, J. T. The dependence of bacterial
sulfate reduction on sulfate concentration in marine sediments.
Geochim. Cosmochim. Acta 1984, 48 (12), 2503−2516.
(45) Dale, A. W.; Regnier, P.; Van Cappellen, P. Bioenergetic
controls on anaerobic oxidation of methane (AOM) in coastal marine
sediments: a theoretical analysis. Am. J. Sci. 2006, 306 (4), 246−294.
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
(46) Lovley, D. R.; Goodwin, S. Hydrogen concentrations as an
indicator of the predominant terminal electron-accepting reactions in
aquatic sediments. Geochim. Cosmochim. Acta 1988, 52 (12), 2993−
3003.
(47) Watson, I. A.; Oswald, S. E.; Mayer, K. U.; Wu, Y.; Banwart, S.
A. Modeling kinetic processes controlling hydrogen and acetate
concentrations in an aquifer-derived microcosm. Environ. Sci. Technol.
2003, 37 (17), 3910−3919.
(48) Xu, S.; Jaffé, P. R.; Mauzerall, D. L. A process-based model for
methane emission from flooded rice paddy systems. Ecol. Modell. 2007,
205 (3−4), 475−491.
(49) Ebert, M. Elementares Eisen in permeablen reaktiven Barrieren zu
in-situ Grundwassersanierung − Kenntnisstand nach zehn Jahren
Technologieentwicklung. Habilitation Thesis, Kiel University, Kiel,
Germany, 2004.
(50) Reardon, E. J. Zerovalent irons: styles of corrosion and inorganic
control on hydrogen pressure buildup. Environ. Sci. Technol. 2005, 39
(18), 7311−7317.
(51) Schütz, M. K.; Libert, M.; Schlegel, M. L.; Lartigue, J.-E.;
Bildstein, O. Dissimilatory iron reduction in the presence of hydrogen:
a case study of microbial activity and nuclear waste disposal. Procedia
Earth Planet. Sci. 2013, 7, 409−412.
(52) Sandefur, C. A.; Koenigsberg, S. S. The use of hydrogen release
compound for the accelerated bioremediation of anaerobically
degradable contaminants: the advent of time-release electron donors.
Remediation Journal 1999, 10 (1), 31−53.
(53) Ergas, S. J.; Shumway, L.; Fitch, M. W.; Neemann, J. J.
Membrane process for biological treatment of contaminated gas
streams. Biotechnol. Bioeng. 1999, 63 (4), 431−441.
(54) Ergas, S. J.; Reuss, A. F. Hydrogenotrophic denitrification of
drinking water using a hollow fibre membrane bioreactor. J. Water
Supply Res. Technol. 2001, 50 (3), 161−171.
(55) Smith, R. L.; Ceazan, M. L.; Brooks, M. H. Autotrophic,
hydrogen-oxidizing, denitrifying bacteria in groundwater, potential
agents for bioremediation of nitrate contamination. Appl. Environ.
Microbiol. 1994, 60 (6), 1949−1955.
(56) Liessens, J.; Vanbrabant, J.; Devos, P.; Kersters, K.; Verstraete,
W. Mixed culture hydrogenotrophic nitrate reduction in drinking-
water. Microb. Ecol. 1992, 24 (3), 271−290.
(57) Xia, S. Q.; Zhong, F. H.; Zhang, Y. H.; Li, H. X.; Yang, X. Bio-
reduction of nitrate from groundwater using a hydrogen-based
membrane biofilm reactor. J. Environ. Sci. 2010, 22 (2), 257−262.
(58) Nevatalo, L. M.; Bijmans, M. F. M.; Lens, P. N. L.; Kaksonen, A.
H.; Puhakka, J. A. Hydrogenotrophic sulfate reduction in a gas-lift
bioreactor operated at 9 degrees C. J. Microbiol. Biotechnol. 2010, 20
(3), 615−621.
(59) van Houten, B. H.; Meulepas, R. J.; van Doesburg, W.; Smidt,
H.; Muyzer, G.; Stams, A. J. Desulfovibrio paquesii sp. nov., a
hydrogenotrophic sulfate-reducing bacterium isolated from a syn-
thesis-gas-fed bioreactor treating zinc- and sulfate-rich wastewater. Int.
J. Syst. Evol. Microbiol. 2009, 59 (Pt 2), 229−233.
(60) Pfeiffer, W. T.; Beyer, C.; Bauer, S. Hydrogen storage in a
heterogeneous sandstone formation: dimensioning and induced
hydraulic effects. Pet. Geosci. 2017, 23, 315.
(61) Feldmann, F.; Hagemann, B.; Ganzer, L.; Panfilov, M.
Numerical simulation of hydrodynamic and gas mixing processes in
underground hydrogen storages. Environ. Earth Sci. 2016, 75 (16),
1165.
(62) Sainz-Garcia, A.; Abarca, E.; Rubi, V.; Grandia, F. Assessment of
feasible strategies for seasonal underground hydrogen storage in a
saline aquifer. Int. J. Hydrogen Energy 2017, 42 (26), 16657−16666.
(63) Lovley, D. R. Minimum threshold for hydrogen metabolism in
methanogenic Bacteria. Appl. Environ. Microbiol. 1985, 49 (6), 1530−
1531.
(64) Mascus, C. Laboratory experiments evaluating geochemical
reactions in aquifers due to hydrogen intrusion. Christian-Albrechts-
Universität zu Kiel, Kiel, 2015.
(65) Oren, A. The bioenergetic basis for the decrease in metabolic
diversity at increasing salt concentrations: implications for the
L DOI: 10.1021/acs.est.7b05467
Article
functioning of salt lake ecosystems. Hydrobiologia 2001, 466 (1−3),
61−72.
(66) Omar, S. A.; Abdelsater, M. A.; Khallil, A. M.; Abdalla, M. H.
Growth and enzyme-activities of Fungi and Bacteria in soil salinized
with sodium-chloride. Folia Microbiol. 1994, 39 (1), 23−28.
(67) Wichern, J.; Wichern, F.; Joergensen, R. G. Impact of salinity on
soil microbial communities and the decomposition of maize in acidic
soils. Geoderma 2006, 137 (1−2), 100−108.
(68) Dincer, A. R.; Kargi, F. Salt inhibition kinetics in nitrification of
synthetic saline wastewater. Enzyme Microb. Technol. 2001, 28 (7−8),
661−665.
(69) Uygur, A.; Kargi, F. Salt inhibition on biological nutrient
removal from saline wastewater in a sequencing batch reactor. Enzyme
Microb. Technol. 2004, 34 (3−4), 313−318.
(70) Yan, N.; Marschner, P.; Cao, W.; Zuo, C.; Qin, W. Influence of
salinity and water content on soil microorganisms. International Soil
and Water Conservation Research 2015, 3 (4), 316−323.
(71) Windey, K.; De Bo, I.; Verstraete, W. Oxygen-limited
autotrophic nitrification−denitrification (OLAND) in a rotating
biological contactor treating high-salinity wastewater. Water Res.
2005, 39 (18), 4512−4520.
(72) Oren, A. Microbial life at high salt concentrations: phylogenetic
and metabolic diversity. Saline Syst. 2008, 4 (1), 2.
(73) Thauer, R. K.; Jungermann, K.; Decker, K. Energy conservation
in chemotrophic anaerobic bacteria. Bacteriol Rev. 1977, 41 (1), 100−
80.
(74) Conrad, R.; Wetter, B. Influence of temperature on energetics of
hydrogen metabolism in homoacetogenic, methanogenic, and other
anaerobic-bacteria. Arch. Microbiol. 1990, 155 (1), 94−98.
(75) Stevens, T. O.; Mckinley, J. P. Lithoautotrophic microbial
ecosystems in deep basalt aquifers. Science 1995, 270 (5235), 450−
454.
(76) Haase, C.; Dahmke, A.; Ebert, M.; Schafer, D.; Dethlefsen, F.
Suitability of existing numerical model codes and thermodynamic
databases for the prognosis of calcite dissolution processes in near-
surface sediments due to a CO2 leakage investigated by column
experiments. Aquat. Geochem. 2014, 20 (6), 639−661.
(77) Berta, M.; Dethlefsen, F.; Ebert, M.; Gundske, K.; Dahmke, A.
Surface passivation model explains pyrite oxidation kinetics in column
experiments with up to 11 bar p(O2). Environ. Earth Sci. 2016, 75,
1175.
(78) Berger, S. Laborversuche zur Quantifizierung von Wasserstoff-
Oxidationsreaktionen in Aquifer-Sedimenten. Kiel University, Kiel, 2015.
(79) Heron, G.; Crouzet, C.; Bourg, A. C. M.; Christensen, T. H.
Speciation of Fe(II) and Fe(III) in contaminated aquifer sediments
using chemical extraction techniques. Environ. Sci. Technol. 1994, 28
(9), 1698−1705.
(80) Canfield, D.; Raiswell, R.; Westrich, J. T.; Reaves, C.; Berner, R.
A. The use of Chromium reduction in the analysis of reduced
inorganic sulfur in sediments and shales. Chem. Geol. 1986, 54, 149−
155.
(81) Appelo, C.; Postma, D. Geochemistry, groundwater and pollution,
Second ed.; Taylor & Francis: Leiden, 2005.
(82) Parkhurst, D. L.; Appelo, C. A. J. Description of input and
examples for PHREEQC version 3A computer program for
speciation, batch-reaction, one-dimensional transport, and inverse
geochemical calculations. In U.S. Geological Survey Techniques and
Methods, book 6, A43; USGS: Reston, VA, 2013; p. 497.
(83) Lever, M. A. Acetogenesis in the energy-starved deep biosphere
− a paradox? Front. Microbiol. 2012, 2, 284.
(84) Cai, M. L.; Liu, J. X.; Wei, Y. S. Enhanced biohydrogen
production from sewage sludge with alkaline pretreatment. Environ.
Sci. Technol. 2004, 38 (11), 3195−3202.
(85) Yuan, H. Y.; Chen, Y. G.; Zhang, H. X.; Jiang, S.; Zhou, Q.; Gu,
G. W. Improved bioproduction of short-chain fatty acids (SCFAs)
from excess sludge under alkaline conditions. Environ. Sci. Technol.
2006, 40 (6), 2025−2029.
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
 Environmental Science & Technology Article

(86) Jesußek, A.; Köber, R.; Grandel, S.; Dahmke, A. Aquifer heat
storage: sulphate reduction with acetate at increased temperatures.
Environ. Earth Sci. 2013, 69 (5), 1763−1771.
(87) Metzgen, A.; Berta, M.; Dethlefsen, F.; Ebert, M.; Dahmke, A.
Impact of pH on hydrogen oxidizing redox processes in aquifers due to
gas intrusions. In EGU General Assembly, Vienna, Austria, April 23−28,
2017.
(88) Schäfer, D.; Schäfer, W.; Kinzelbach, W. Simulation of reactive
processes related to biodegradation in aquifers: 1. Structure of the
three-dimensional reactive transport model. J. Contam. Hydrol. 1998,
31 (1−2), 167−186.
(89) Schäfer, D.; Hornbruch, G.; Schlenz, B.; Dahmke, A.
Contaminant spreading assuming different kinetic approaches to
simulate microbial degradation. Grundwasser 2007, 12 (1), 15−25.
(90) Krumholz, L. R.; Harris, S. H.; Tay, S. T.; Suflita, J. M.
Characterization of two subsurface H2-utilizing bacteria, Desulfomi-
crobium hypogeium sp. nov. and Acetobacterium psammolithicum sp. nov.,
and their ecological roles. Appl. Environ. Microbiol. 1999, 65 (6),
2300−2306, http://aem.asm.org/content/65/6/2300.long
(91) Vallero, M. V. G.; Lettinga, G.; Lens, P. N. L. High rate sulfate
reduction in a submerged anaerobic membrane bioreactor (SAMBaR)
at high salinity. J. Membr. Sci. 2005, 253 (1−2), 217−232.

M DOI: 10.1021/acs.est.7b05467
Environ. Sci. Technol. XXXX, XXX, XXX−XXX

View publication stats