Geochemical Modelling of Eperimental O2 SO2 CO2 Reactions of Reservoir Cap Rock and Overlying Cores - Pearce - 2019

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Applied Geochemistry 109 (2019) 104400

Contents lists available at ScienceDirect

Applied Geochemistry
journal homepage: www.elsevier.com/locate/apgeochem

Geochemical modelling of experimental O2–SO2–CO2 reactions of reservoir, T


cap-rock, and overlying cores
Julie K. Pearcea, , Dirk M. Kirsteb, Grant K.W. Dawsona, Victor Rudolphc, Suzanne D. Goldinga

a
School of Earth and Environmental Sciences, University of Queensland, QLD, 4072, Australia
b
Department of Earth Sciences, Simon Fraser University, Vancouver, Canada
c
School of Chemical Engineering, University of Queensland, QLD, 4072, Australia

ARTICLE INFO ABSTRACT

Editorial handling by Dr D. Wolff-Boenisch Carbon dioxide streams stored geologically from industrial sources e.g. coal oxy-fuel firing sources, steel or
Keywords: cement processing, may contain gas impurities including O2, NOx and SOX which may have higher reactivity to
CO2 geosequestration rock than pure CO2. Supercritical CO2 with O2 and SO2 impurities (0.16 and 2% respectively) was reacted with
Oxygen and sulphur dioxide impurity core from the Precipice Sandstone, Evergreen Formation, and Hutton Sandstone of the Surat Basin, Australia.
Precipice sandstone The Precipice Sandstone is a low salinity target reservoir for CO2 storage, where the Evergreen Formation would
Evergreen formation be the overlying cap-rock, and the Hutton Sandstone is an overlying aquifer. High and low water regimes were
Hutton sandstone investigated with core sections either submerged in bulk gas saturated water, or suspended in the water satu-
rated supercritical gas head at reservoir in situ conditions of 60 °C and 12 MPa. During O2–SO2–CO2 water rock
reactions in bulk solution, a low solution pH (~1–2) was measured owing to the relatively high gas impurity
concentrations used. Dissolution of reactive minerals from the core including Fe-chlorite, ankerite, siderite, and
plagioclase was directly observed. The Evergreen Formation cap-rock and Hutton Sandstone cores reacted here
contained ~ 5–10 times higher total concentrations of elements such as Fe, Al, Ti, Ba than the Precipice
Sandstone core. Several cations were mobilised to solution including Mn, Co, Zn, Sr, Si, and Mg. Increasing
dissolved concentrations of Mn, and Co were strongly correlated with Fe (R2 = 0.99) during reaction of the
Precipice Sandstone core. Dissolved Co and Sr were strongly correlated with Fe during reaction of the Evergreen
Formation core. Dissolved Mn was somewhat correlated with Fe (R2 = 0.99) from reaction of the Hutton
Sandstone core, from both carbonate and chlorite dissolution. During gas-water reaction with the Hutton
Sandstone core, dissolved element concentrations subsequently decreased, by incorporation into newly formed
Fe oxide precipitates observed with Cr and Ni signatures. Geochemical models were built and validated with
good agreement attained with experiments, predicting Fe-chlorite and minor carbonate minerals were the main
minerals dissolving. Reactive surface areas initially estimated by a modified geometric method needed sub-
stantial increases for example in the case of chlorite (up to 2000x), and decreases for carbonates (2–10x) to
match experimental data. A 30 year simulation of reservoir rock reactivity showed that the lowered pH was
buffered by mineral dissolution. Compared to previously published pure CO2-water-rock reactions, the con-
centrations of ions such as Si generally were initially higher through reaction of more silicate minerals at the
lower generated pH, and the oxidation of sulphides. Compared to SO2–CO2-water reactions, the concentrations
of cations were similar, but with O2 present dissolved Fe subsequently decreased as Fe-oxide, sulphate and clay
minerals precipitated. Cap-rock showed reactivity to wet supercritical O2–SO2–CO2 fluid in the supercritical gas
headspace. Corrosion of minerals including ankerite and Fe-chlorite were observed, along with precipitation of
oxide, sulphate and silicate minerals on rock surfaces. These results have implications for impacts in CO2 storage
sites where gas impurities may be co-injected with CO2 and accumulate in the near wellbore region.

1. Introduction or cement processing. Separation and purification of CO2 from high


concentrations of inert gases such as N2, Ar can introduce high costs.
Carbon capture and storage (CCS) involves the capture of CO2 Coal oxy-fuel firing can produce a flue gas with a high proportion of
streams from point sources such as coal fired power stations, steel, gas, CO2, reducing the need for extensive purification, and is an option for


Corresponding author. UQ Centre for Coal Seam Gas and School of Earth and Environmental Sciences, University of Queensland, QLD, 4072, Australia.
E-mail address: j.pearce2@uq.edu.au (J.K. Pearce).

https://doi.org/10.1016/j.apgeochem.2019.104400
Received 16 May 2019; Received in revised form 9 August 2019; Accepted 13 August 2019
Available online 13 August 2019
0883-2927/ © 2019 Elsevier Ltd. All rights reserved.
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

CO2 capture and storage. Several oxy-fuel firing plants have been as- increase or decrease element mobility with implications for water
sessed or trialled in Australia (e.g. The Calide Oxy-fuel Project)(Spiro, chemistry and mineral trapping of CO2 (Little and Jackson, 2010; Lu
2014) and internationally (e.g. Vattenfall Oxy-fuel Pilot Program, et al., 2010). The subsequent precipitation of oxides such as hematite or
Germany, and the White Rose Project, UK) (Ting et al., 2014). CO2 Fe oxyhydroxides can, however, beneficially incorporate dissolved
streams from coal oxy-fuel firing or cement and lime processing may trace metals.
contain up to a few percent of impurities including O2, SOx and NOx Storage in high salinity aquifers has generally been considered
which have reactivity to minerals such as carbonates, pyrite, plagio- worldwide, however, CO2 storage in low salinity aquifers is proposed in
clase and clays when in contact with formation water (Wang et al., Queensland, Australia (Hodgkinson and Grigorescu, 2012). A storage
2012). Oxygen content in injected CO2 streams generally from different site in the Gippsland Basin, Victoria, Australia also has a fresh water
capture sources is expected to range from ~30 ppm to ~5 vol%, with an intrusion in the proposed injection site (Michael, 2014). Although the
upper limit of 4% suggested by European regulations (de Visser et al., majority of CCS studies focus on deep saline aquifers, several authors
2008; IEAGHG, 2011; Jia et al., 2009; Jia et al., 2007). Injection and have performed low salinity experimental studies (at pressure and
storage of impure CO2 streams geologically has been suggested to temperature relevant to shallow aquifers) mainly to determine likely
provide potential economic benefits through lower capture and pur- changes to water quality in the case of a pure CO2 leak to overlying
ification costs, however, relatively less is understood of the geochemical freshwater aquifers. Reduced pH from carbonic acid formation, and
impacts compared to pure CO2 (Gaus, 2010; Glezakou et al., 2012; subsequent dissolution of reactive minerals (mainly carbonates) were
Thaysen et al., 2017). observed to increase the concentration of dissolved elements including
Storing CO2 subsurface has implications for the redox potential and Ca, Mn (Little and Jackson, 2010). The concentration in solution of
water chemistry of storage reservoirs and overlying aquifers, an im- major and trace elements including Fe, Al, U, As, Cr was also reported
portant consideration for monitoring efforts (Humez et al., 2014). CO2- to be affected by redox changes causing desorption from rock core,
water-rock reactions may also change the porosity and permeability of although vessel corrosion may be a source (Lu et al., 2010). The in-
reservoir and cap-rock through dissolution and precipitation processes creasing dissolved concentrations of several elements (Sr, Co, Mn, Ni,
with implications for CO2 stream migration and containment (Dávila Tl, Zn) in CO2-water-rock experiments with carbonate cores have been
et al., 2016; Farquhar et al., 2015; Kummerow and Spangenberg, 2011; attributed to carbonate rather than sulphide dissolution (Humez et al.,
Pearce et al., 2019). Experimental studies allow the collection of data to 2014; Wunsch et al., 2013, 2014). Dissolved cations are however site
validate and improve geochemical models for longer term predictions and mineralogy specific, and the element sources from sandstones are
of impacts to water chemistry and rock properties. There is however not well understood. A recent experimental study on Surat Basin core
generally relatively little data on the geochemical effects of co-injection reactivity to pure CO2 reported mobilised concentrations of dissolved
of O2 in stored CO2 streams (de Visser et al., 2008; Kather, 2009). The Co, Zn, and Mn with Co and Zn showing a correlation with increasing
overwhelming majority of water-rock interaction studies at CO2 storage CO2 pressure up to 3.8 MPa at 45 °C though no mineral sources were
conditions have been performed with pure CO2, with an increasing discussed (Horner et al., 2014).
number of studies including co-injected SO2 +/− traces of O2 from air Two complementary experimental and modelling studies on the
e.g. (Dawson et al., 2015; Kirste et al., 2017; Palandri et al., 2005; reactivity of pure CO2 or CO2 with SO2 impurities at low salinity in situ
Pearce et al., 2015, 2016a; Wilke et al., 2012). In addition recently a conditions (12 MPa and 60 °C) with similar sections of Surat Basin core
portion of SO2 (~60% at equilibrium) has been shown to remain in the from Queensland, Australia, as those reacted here also indicated po-
supercritical gas head in SO2–CO2- water or brine dissolution experi- tential changes to near well bore water chemistry (Farquhar et al.,
ments (Amshoff et al., 2018). 2015; Pearce et al., 2015). Dissolved elements including Fe and Ca,
Co-injected oxygen has received even less attention, even though were mobilised to solution at generally higher concentrations from seal
most injected streams will contain traces of O2 from air. Pyrite oxida- core in comparison to reservoir core. In the presence of SO2 a lower pH
tion caused water acidification and element mobilisation from Gothic was generated and dissolved Fe concentrations from reactive silicates
shale with gypsum and Fe oxide formation when reacted in the pre- was elevated up to 20 times.
sence of 1–8% O2 in CO2 (Jung et al., 2013). A gas mixture containing Supercritical (sc)CO2 has historically been thought to have little
4% SO2 and 4% O2 in CO2 was also reacted with Mano Dolostone in reactivity with rock except when dissolved in formation water to form
high salinity brine (Renard et al., 2011). The dissolution of calcite and carbonic acid (Gaus, 2010). However, formation water can also dissolve
clays, and precipitation of Ca sulphate and barite were attributed to into the scCO2 phase (Spycher et al., 2003). Buoyant scCO2 may ac-
inclusion of the co-injected gases. Both authors observed oxygen to be a cumulate under cap-rock (structural trapping) and become water sa-
reactive component, as have others investigating the reactivity of well turated (Shao et al., 2011). The reactivity of water saturated (wet)
materials such as steel (Choi et al., 2010). Wet basalt cores reacted with scCO2 (pure CO2 or with H2S or SO2 impurities) with wetted or non-
1% SO2 and 1% O2 in CO2 also resulted in formation of the sulphate wetted rock and well materials has been recently observed, however
minerals jarosite, alunite and gypsum (Schaef et al., 2014). Calcite relatively little experimental work has been performed compared to
cemented cap-rock was recently reacted with O2 and SO2 in CO2 with experiments with rock submerged in brine or formation water. With
calcite dissolution increasing overall porosity but precipitation of work on the wet supercritical phase or gas head mostly focussed on
gypsum observed which may decrease porosity or permeability over pure mineral studies mainly with pure CO2 (Choi et al., 2010; Felmy
longer time scales (Pearce and Dawson, 2018; Pearce et al., 2016b). et al., 2010; Glezakou et al., 2012; Loring et al., 2011). Further in-
Clay rich cap-rock was also reacted, with precipitation of Fe-oxides vestigation of this phase was suggested owing to its potential reactivity
observed to close nanoscale pores (Pearce et al., 2016a, 2018). with cap-rock (Gaus, 2010). Indeed recent studies of reservoirs with
SO2 acidifies formation water and may induce reactive rock dis- naturally high concentrations of CO2 have revealed high rock reactivity
solution with subsequent potential changes to water quality and rock in the gas-leg with precipitation of carbonates and chlorite conversion
porosity/permeability. The additional presence of O2 can drive the to kaolinite (Higgs et al., 2015). As oxygen has a low relative solubility,
formation of sulphuric acid (Equ. 1) (Palandri et al., 2005). accumulation in the near well bore region in the buoyant supercritical
gas phase has been predicted (Jung et al., 2013). In certain storage
SO2 (g) + H2O + ½O2 ↔ HSO4− + H+ 1
sites, contact of an impure wet scCO2 phase with the sealing cap-rock
As the mobility of some elements (e.g. Fe) released from rock into may take place; hence the potential for reactivity of this phase with cap-
formation water is affected by redox processes, the addition of co-in- rock should also be assessed (Black et al., 2014).
jected gases such as O2 to a CO2 stream could be expected to either In this work, CO2 with SO2 and O2 impurity gases has been ex-
perimentally reacted with reservoir and cap-rock core from a potential

2
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

CO2 storage site. Both core submerged in water with the dissolved gas siltstone-fine sandstone, generally contained quartz, kaolin, k-feldspar,
mixture, or suspended in the wet scCO2 phase are investigated experi- ankerite, chlorite, Fe-oxide, and siderite. H 279 contained quartz with
mentally through changes to water chemistry and rock. Geochemical feldspars, chlorite, siderite (Table 1), with organic matter present, and
modelling of experiments with core reacted submerged in water was Fe-oxides.
also performed to identify dissolving or potentially precipitating mi-
nerals on the time scale of experiments. Similar core depth sections and 2.2. Experimental procedure
conditions were used to previous studies to allow comparison to pure
CO2 or CO2–SO2 water rock reactivity (Farquhar et al., 2015; Pearce Experiments were conducted in an apparatus based on unstirred
et al., 2015). Parr® reactors with custom built thermoplastic liners as described in
detail previously (Pearce et al., 2015). Reactions were performed at
60 °C and 12 MPa to simulate conditions in the Surat Basin, Australia
2. Method (Raza et al., 2009). The Precipice Sandstone in the Surat Basin has fresh
to brackish formation water, although a reliable composition at the
2.1. Materials study wellbore has not been determined (Hodgkinson and Grigorescu,
2012). Formation water measurements from nearby bores indicate
CO2 storage has been proposed in the Precipice sandstone of the measured pH of 6.7–6.9, conductivity in the range 200–5000 μS/cm
Surat Basin, Queensland, Australia, with the low permeability and alkalinity 90–1100 mg/L as CaCO3, with Cl concentrations of
Evergreen formation to act as the sealing cap-rock (Hodgkinson et al., 8–1170 mg/kg (Feitz et al., 2014). Milli Q water was used in the ex-
2010). The overlying Hutton sandstone could act as a secondary seal, or periments with an initial water-rock equilibration in N2 to allow com-
has been suggested to provide a region for subsurface monitoring of parison to previous SO2–CO2-water-rock and CO2-water-rock reactions
CO2 leakage. Four core sections from the three formations were tested. with similar core sections, and to simplify interpretation and prevent
The Precipice Sandstone hosts fresh to brackish water, unlike the ma- masking of dissolved analytes (Pearce et al., 2015).
jority of sites investigated for CO2 storage internationally (Hodgkinson Three sets of experiments were performed with the rock cores;
and Grigorescu, 2012; Hodgkinson et al., 2010).
Sub-sample blocks were used from the GSQ Chinchilla-4 well core of 1) Deoxygenated MQ water (70 mL) was added to the reactor, with the
the Precipice Sandstone, referred to here after as P 386 sampled at a solid core blocks suspended in a thermoplastic basket immersed in
depth of 1193 m, the Evergreen Formation E 368 sampled at 1138 m the water. Heated vessels were purged of residual air with a low
and E 289 sampled at 898 m, and the Hutton Sandstone H 279 sampled pressure N2 flush. A Teledyne ISCO syringe pump (500HP) was used
at 868 m. These general core sections have been previously miner- to pressurise vessels to 12 ± 0.5 MPa, initially with N2 gas.
alogically characterised in detail (Farquhar et al., 2013, 2015). These Following the initial baseline water – rock reaction period to ob-
sub-sample sections were similar but not adjacent to those reacted in serve any initial release from water-rock interaction, as in previous
SO2–CO2 –water and pure CO2 – water at similar conditions by our studies, water samples were obtained. N2 gas was subsequently re-
group as described in detail previously (Pearce et al., 2015). The GSQ placed with a gas mixture of 0.16% SO2, and 2% O2 in a balance of
Chinchilla-4 well is approximately 10 km south-south east of Miles in CO2 (Coregas) using the syringe pump. The concentration of SO2
Queensland, Australia, at a latitude of 26.72722 and longitude of and O2 gas was chosen to simulate a simplified potential average
150.2014. Note that the E 368 sample from the Evergreen Formation at composition of industrially captured CO2 from oxy-fuel fired power
1138 m may also be classified as part of the upper Precipice Sandstone plants (Kather, 2009; Notz et al., 2011). The potential average
by other authors. Briefly, P 386 contained generally low reactivity composition used does not necessarily reflect likely concentrations
minerals, major quartz with kaolinite, smectite, and illite, and addi- expected to be injected into the Surat Basin. Excess supercritical
tional trace minerals (including calcite) based on previous QEMSCAN fluid was maintained during experiments, with a supercritical fluid
and XRD adjusted by SEM observations of the blocks reacted here gas head in the reactors above the water line. Fluid was sampled
(Table 1). More reactive minerals were present in the reacted core from periodically over ~16 days reaction for analysis, before depressur-
the Evergreen Formation and Hutton Sandstone; these cores were more isation and collection of rock material.
heterogeneous in nature, with some minerals in trace quantities only 2) Additional experiments were separately performed with wet solid
detectable by SEM-EDS of the sub-samples reacted here. E 368 con- core blocks suspended in the supercritical fluid at the top of reactors
tained quartz, k-feldspar, kaolinite and plagioclase feldspar, as well as in a thermoplastic basket with core not in direct contact with the
Fe-chlorite, calcite and illite and biotite. E 289, an interlaminated 40 mL of MQ water initially added to the bottom of the reactor. The
rectors were heated, and then the same gas mixture (Coregas;
Table 1 O2–SO2–CO2) was used to pressurise the reactors to 12 MPa at 60 °C.
Initial model input mineral volume % based on data from Pearce et al. (2015) A portion of water (~0.58%) would subsequently be expected to
and Farquhar et al. (2015), modified for core sub-sample observations (e.g.
dissolve into the supercritical phase on heating and pressurisation
trace calcite observed in P 386 sample reacted here). The core depths are given
(Duan and Sun, 2003). Bulk water was periodically sampled for
in brackets.
analysis over ~28 days via a dip tube into a small sampler volume
P 386 E 368 E 289 H 279 (described in Pearce et al., 2015) to avoid pressure loss. Reaction
(1193 m) (1138 m) (898 m) (868 m)
conditions were otherwise generally the same as the first set of ex-
Quartz 94.0 45.50 43.00 50.90 periments.
k-Feldspar 0.10 27.00 10.00 6.00 3) Blank tests were also conducted at the conditions of both experiment
Plagioclase Feldspar 0.10 8.00 4.00 7.00 sets 1 and 2, but without rock material present to identify any po-
Kaolin 2.55 12.4 18.00 7.00
tential cation contamination from corrosion of connections in the
Illite/Muscovite 2.70 20.00 3.00 9.00
Smectite 0.20 0.40 0.50 reaction vessel.
Biotite 0.50 2.00 8.00
Fe–Mg Chlorite 0.20 3.10 4.00 7.00 2.3. Analytical methods
Calcite 0.05 1.10
Ankerite 10.00 4.00
Adjacent core sections to the sub-samples used in reactions were
Siderite 0.05 4.00 0.10
Fe Oxide 0.05 2.00 0.50 subject to whole rock analysis. Whole rock fusions and elemental ana-
lysis by inductively coupled plasma optical emission spectroscopy (ICP-

3
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

OES) was performed to determine major elemental compositions of the CO2 fugacity was calculated based on a solubility model at the ex-
core sub-samples reported as percentage oxides with an error of less perimental PT conditions (Duan and Sun, 2003). Initial values of mi-
than 5% (PerkinElmer Optima 3300 DV ICP-OES instrument with a 3σ neral reactive surface areas are given in Table 2, calculated as described
detection limit of 0.001 mg/L) by methods described previously (Pearce previously in (Pearce et al., 2015) based on the grain size determined
et al., 2015). Loss on ignition was also determined, along with trace through SEM and modified for spheres or clay plate morphology by the
element analysis by acid digestion (HCl–HNO3) and inductively coupled method of (White, 1995). Mineral reactive surface areas were subse-
plasma mass spectrometry (ICP-MS) on a Thermo Fisher Series Xi quently adjusted to reach a fluid compositional match with experi-
quadropole ICP-MS (RSD error less than 10%, for most elements less mental data. These values varied by rock experiment and the final va-
than 5%). Two portions of both rock samples E 289 and H 279 were lues used are given in the results section.
analysed owing to the rock sample heterogeneity. Rock elemental Subsequently, 30 year geochemical models were built using the
analyses were also performed on the rock samples post O2–SO2–CO2 mineral content of the rocks but with an equilibrated formation water
-water-reaction. Pore size distributions of separate subsamples of the chemistry from (Feitz et al., 2014). The volume of water used was based
cores were measured by mercury intrusion with the method described on the rock porosity, and the O2 fugacity was not fixed. The experi-
previously (Farquhar et al., 2015). mentally fitted reactive surface area final values were generally sub-
Scanning Electron Microscopy with Energy Dispersive Spectrometry sequently upscaled for the reservoir by a factor of 0.1, and for carbo-
(SEM-EDS) was performed on rock sample surfaces pre and post reac- nates were 0.05 cm2/g (Kirste et al., 2017).
tion (low-vacuum JEOL6460LA environmental SEM with an EDS de-
tector) to detect corrosion effects and formation of new precipitates, by 3. Results and discussion
methods described previously (Dawson et al., 2015). In most cases the
same position could be imaged before and after reaction, avoiding 3.1. Rock elemental compositions
ambiguity between previous diagenetic alteration, and dissolution/
precipitation occurring during gas-water-rock reactions. Prior to, and Whole rock major elemental analysis as oxides and loss on ignition
post reaction, rock sample skeletal density was also determined via (LOI) are shown in Table 3 for specific sub-sections of core. These were
helium pycnometry (Micromeritics AccuPyc 1330) on selected samples generally in good agreement with those measured previously for non-
to detect physical changes. Core sample E 368 was crushed and ana- adjacent sections of core (Farquhar et al., 2015). The H 279 rock sample
lysed by XRD (UQ CMM) post reaction to attempt to identify newly reacted in this study was observed to contain visibly more organic
formed precipitates. Other samples were mainly used up in digestions matter and Fe bearing phases (mainly Fe oxides, biotite, and siderite),
post-reaction to identify changes to elemental composition (or changes as reflected in the higher LOI and Fe oxide content of rock sample H
deemed likely too small to detect by XRD). 279b (Table 3). In general the P346 sample has the highest Si content
Solution pH and conductivity were immediately measured after and lowest Ca, consistent with the mineral content in Table 1 showing
sampling from batch reactors with ex situ probes (TPS WP81, the highest quartz content and lowest calcite or ankerite content. E368
error ± 0.01), the pH represents maximum values as CO2 may have has the highest K content consistent with the highest K-feldspar content.
degassed from solution. A portion of solution was filtered (0.45 μm), Whole rock trace element analysis of the core sections by digestion
diluted 10 times, and acidified to 2% HNO3 which preserved the sam- indicated approximately 5–10 times higher concentrations of most trace
ples prior to element analysis by ICP-OES. Further portions of solution elements in the Evergreen Formation and Hutton Sandstone cores when
(prior to O2–SO2–CO2 gas injection and at the termination of experi- compared to the Precipice Sandstone core, as shown in Table 4. Pre-
ments) were also filtered, diluted and acidified with ultrapure HNO3 vious microprobe analysis of these cores indicated chlorite is Fe-rich,
prior to trace element analysis by ICP-MS. Selected unacidified samples and also contains Mg and less Mn. Calcite contains Mg and Mn, with Sr
were analysed for sulphate, chloride and alkalinity by ALS in both the calcite and clays. Ti was mainly present in ilmenite, and also
Environmental by methods described in detail previously (Pearce et al., chlorite and muscovite.
2015). Organic matter, oxides and carbonate cements were generally ob-
served to be sporadic and spatially variable in the cores and may re-
2.4. Geochemical modelling present sources of trace elements, especially in Evergreen Formation
and Hutton Sandstone core sections, as shown in Table 4. Trace
Experiments with rock core submerged in water with dissolved amounts of phosphate, sulphide and fine-grained Cu containing mi-
O2–SO2–CO2 gas were modelled with the React module of the nerals were also observed in cores occasionally.
Geochemist's Work Bench (GWB), with methods described in detail
previously (Pearce et al., 2015). Initial mineral input volume % are 3.2. Porosity and pore throats
given in Table 1, which were based on previous core characterisation
adjusted slightly according to SEM observations of minerals on the sub- Mercury intrusion and micro CT porosities of the cores and sub-
samples reacted here. A lower Evergreen Formation calcite-cemented plugs were reported previously (Farquhar et al., 2015). The Precipice
sample E 368 (Table 1) was additionally modelled to investigate likely Sandstone sample P 386 had the highest porosities of 15.2% and 18.4%
pH buffering above the reservoir at the contact with the Evergreen by mercury intrusion and micro CT (with a resolution of 2.2 μm) re-
Formation. Mineral kinetic parameters are given in Table 2, the ther- spectively. Evergreen Formation core containing calcite cement, E 368,
modynamic database for mineral and aqueous species is based on the had mercury and resolvable micro CT porosities of 8.8% and 0.2% re-
EQ3/6 database with modifications as described previously (Delany spectively. The fine-grained E 289 had mercury and micro CT porosities
and Lundeen, 1989; Pearce et al., 2015). Mineral data is from the of 6.1%, and 2% including organic content. H279 had porosities of
compilation of (Palandri and Kharaka, 2004), except for chlorite from 7.9% and 1.9% respectively, and the calcite cemented H256 10.1% and
(Lowson et al., 2005, 2007), siderite and ankerite from (Steefel, 2001), 1.1%. The μCT characterisation was performed on sub-plugs, and the
and illite from (Köhler et al., 2003). Input water chemistry was based mercury intrusion on separate core offcut material. This may have in-
on experimentally measured values after the water-rock soak at 60 °C troduced some heterogeneity. Since the μCT characterisation is a cal-
and 120 bar N2 prior to O2–SO2–CO2 gas injection. A mass of SO2 gas culated porosity and had a resolution of 2.2 μm in this case it was
was added to replicate experimentally observed concentrations, as de- reasonable for the reservoir sandstone P346, but less useful for fine-
scribed previously, and a fixed fugacity of O2 was used (tested values grained clay-rich rocks containing a lot of small pores below the re-
increased from −40 to use a value of −10 to match experimental pH solution. For the cap-rocks and overlying aquifer cores the mercury
values and precipitation profiles of experimental ion concentrations). intrusion porosities were likely more reliable.

4
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Table 2
Kinetic parameters for modelling as described in detail in (Pearce et al., 2015) As are initial reactive surface areas used to model experiments, and subsequently
adjusted with the final values used defined for each experiment in the results section. Γ is the pre-exponential factor for nucleation.
Mineral k25(acid) Ea(acid) kJ/mol n k25(neut) mol/cm2/s Ea(neut) kJ/mol Asa cm2/g K(precip) mol/cm2/s Γ
mol/cm2/s

Chalcedony 1.70E-17 68.70 10 K(diss) 2E+10


K-feldspar 8.71E-15 51.7 0.500 3.89E-17 38.00 10 K(diss) 2E+10
Albite 6.92E-15 65.0 0.457 2.75E-17 69.80 10 K(diss) 2E+10
Kaolinite 4.90E-16 65.9 0.777 6.61E-18 22.20 70 K(diss)/10 2E+10
Illite/Muscovite 1.91E-16 46.0 0.600 8.91E-20 14.00 70 K(diss) 2E+10
Smectite 1.05E-15 23.6 0.340 1.66E-17 35.00 150 K(diss) 2E+10
Biotite 1.45E-14 22.0 0.525 2.82E-17 22.00 70 K(diss) 2E+10
Fe–Mg-Chlorite 1.62E-14 25.1 0.490 1.00E-17 94.30 70 K(diss) 2E+10
Calcite 5.01E-05 14.4 1.000 1.55E-10 23.50 10 K(diss) 1E+10
Ankerite 1.59E-08 45.0 0.900 1.26E-13 62.76 10 K(diss)/1e5 3E+10
Siderite 1.59E-08 45.0 0.900 1.26E-13 62.76 10 K(diss) 2E+10
Fe-oxide 4.07E-14 66.2 1.000 2.51E-19 66.20 70 K(diss) 1E+10

a
Initial.

Pore throat size distributions for the four core sub-sections de- Fig. 2A) were corroded post-reaction (Fig. 2B) with the quartz grain
scribed here and additionally a calcite cemented Hutton Sandstone surfaces generally unchanged. Kaolin and Ti–Fe oxides observed on the
sampled from 799 m depth of the same well core (H256) are shown in sample surface pre-reaction, showed only minor corrosion post reaction
Fig. 1. The Precipice Sandstone (reservoir) sample P386 has the highest (Fig. 2C–F). Some minor movement of kaolin booklets appeared to have
population of large pores with the majority in the macro pore range, in occurred. Ca signatures observed pre-reaction were absent post-reac-
agreement with the observation of open porosity in SEM. The Evergreen tion as were NaCl, and Ca-phosphate suggesting they dissolved during
Formation (seal or cap-rock) samples have the highest occurrence of the reaction. In general significant mineral surface changes were
smaller pores in the mesopore range. The Evergreen E298 fine-grained however not observed for this sample.
shale has the majority of its pore throats below 1000 Å, likely asso- Corrosion of kaolin from P 386 was not observed in previous reac-
ciated with clays and organic matter. The Hutton Sandstone (overlying tions at similar conditions with either pure CO2 or SO2–CO2 gas streams
aquifer) samples have intermediate pore throat distributions variable (Farquhar et al., 2015; Pearce et al., 2015).
over a wide size range. The high proportion of small pore throats as-
sociated with clays in the cap-rocks and overlying aquifer core is con-
sistent with the lower porosities reported by micro CT where the ma- 3.3.1.2. Evergreen. The E289 core sample had a relatively fine grain
jority of pores are below the technique resolution in that case. size, with organic matter present (Fig. 3A). Some siderite and ankerite
present on this E 289 sub-sample surface tended to have no Mg content
unlike sub-samples reacted previously (Farquhar et al., 2015; Pearce
3.3. Dissolved gas-water-rock reactions et al., 2015). The rock sample mass used was 1.4823 g.
Fine grained siderite and ankerite were observed, with EDS in-
3.3.1. Rock changes dicating ankerite contained Ca, Fe and also some Mn ± Mg (supple-
3.3.1.1. Precipice. SEM-EDS surveys of the mineral surfaces were mentary material). Siderite also sometimes contained Mg ± trace sig-
performed. Where possible the same area probed pre-reaction was natures of Mn. These carbonates were often associated with chlorite or
returned to post-reaction to observe mineral surfaces and EDS occasional biotite and chloritized biotite both Fe-rich ± Mg signatures.
signatures. This allowed more confident assignment of changes owing Occasional zircon grains were observed and a monazite (supplementary
to reaction rather than previous diagenetic changes of the rock. material). Also trace amounts of Ti–Fe-oxides sphalerite, pyrite, and
The Precipice P 386 sample had variable grain size, with trace plagioclase solid series. K-feldspar and plagioclase grains were not fresh
amounts of Fe bearing mineral phases. These included Fe-oxide cements surfaces pre-reaction, and had relatively high surface areas through
and grain coatings (also identified previously in petrography) Ti–Fe- previous diagenetic changes (Fig. 3C and inset, and E).
oxide previously identified as ilmenite and siderite (Fig. 2A) (Farquhar Siderite and ankerite on the surface of E 289 was observed to be
et al., 2013, 2015). Ca signatures associated with trace calcite on completely corroded post-reaction (Fig. 3A and B). Post-reaction, minor
kaolinite were observed in SEM-EDS. EDS signatures also indicated a pitting of K-feldspar surfaces (Fig. 3D), and more significant pitting of
trace amount of fine grained Ca-phosphate and an occasional NaCl albite and Ca–Na-plagioclase surfaces (Fig. 3F) were observed. Chlorite
signature. Trace amounts of albite were also identified. Occasional fine was also corroded, although some residual chlorite remained. Occa-
grained pyrite, sphalerite, and Cu–Sn oxide/carbonate were also pre- sional chlorite and chloritized biotite booklets appeared to have totally
sent in the reacted sub-section (Fig. 2G and H, and supplementary dissolved or dislodged from the surface during reaction (Fig. 3G)
material). especially where they were associated with carbonates that dissolved.
The minor amounts of siderite (e.g. bright spot in the centre of Organic matter appeared to still be present after reaction, and the

Table 3
Whole rock elemental composition of adjacent core sections reported as element (%) oxides from fusions and ICP-OES, and loss on ignition (%). H 279b indicates the
sub-sample with greater visible organic matter (coal) content.
Al2O3 CaO Fe2O3 K2O MgO MnO Na2O P2O5 SiO2 TiO2 Total LOI

P 386 3.24 0.00 0.12 0.31 0.01 0.00 0.68 0.02 98.48 0.08 102.95 0.73
E 368 15.30 0.77 2.65 4.48 0.78 0.02 1.32 0.11 72.03 0.64 98.12 5.56
E 289 15.96 0.22 3.40 2.76 0.70 0.04 1.46 0.06 68.52 0.88 94.00 7.71
H 279 10.15 0.02 3.79 1.60 0.37 0.05 1.19 0.03 79.33 0.48 97.02 3.14
H 279b 11.47 0.04 4.30 1.81 0.46 0.04 1.23 0.04 80.18 0.63 100.21 3.99

5
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Table 4
Whole rock trace elemental composition (mg/kg), two sections of E 289 and H 279 were analysed owing to the rocks heterogeneous nature, H 279b sub-sample had
visibly greater organic matter (coal) content.
Li Be B Sc Ti V Cr Co Ni Cu Zn

P 386 2.5 0.1 12.9 0.6 455.5 3.9 6.5 2.3 3.8 41.0 15.4
E 368 15.4 1.2 20.4 4.4 3065.4 41.7 15.9 7.5 7.8 20.3 52.7
E 289 30.0 1.9 32.5 10.8 4830.6 70.1 53.1 9.7 19.0 16.6 75.2
E 289b 12.0 0.6 148.0 5.4 333.8 40.9 104.5 2.5 5.7 14.3 25.5
H 279 21.7 1.1 40.9 6.6 2824.8 48.2 31.1 7.4 14.1 9.2 56.2
H 279b 24.3 1.3 28.5 8.1 3668.5 61.4 43.1 9.4 16.3 50.6 88.7

As Se Rb Sr Zr Nb Mo Cd Ba Pb U

P 386 0.8 0.1 2.5 10.4 27.3 184.0 0.2 0.0 24.0 2.9 0.4
E 368 5.5 0.3 98.6 108.6 112.7 760.3 0.6 0.1 714.1 14.6 1.3
E 289 3.5 0.8 108.4 141.7 249.2 1552.6 0.6 0.3 524.2 17.8 2.8
E 289b 25.9 1.5 23.2 310.6 27.0 359.4 1.1 13.1 93.9 7.7 1.9
H 279 3.4 0.4 53.0 56.7 115.1 999.3 0.5 0.1 288.6 10.7 1.2
H 279b 4.2 0.5 67.7 73.0 143.6 1329.3 0.8 0.2 330.3 13.1 1.5

Fig. 1. Pore throat size distributions (mercury intrusion) of the four cores described here, and also a calcite cemented Hutton Sandstone.

surface of the rock appeared rougher. (Farquhar et al., 2015).


Corrosion of ankerite and chlorite appeared to be more extensive in Post reaction, chlorite and siderite were corroded (Fig. 4A, B, C, D).
the current study than in previous studies of similar rock sample re- Post-reaction, some areas of the rock sample surface had an orange-
activity with pure CO2 or SO2–CO2 and water (Pearce et al., 2015). brown colouration, and in these areas fine-grained Fe ± Mn bearing
Elemental S signatures were not observed post O2–SO2–CO2 reaction on phases with EDS signatures suggesting oxides were observed by SEM to
the E 289 sample likely owing to oxidation by co-injected oxygen. In have been precipitated. Sulphate Fe, O and S signatures were also ob-
previous reaction with SO2–CO2-water in the absence of oxygen, ele- served (Fig. 4E). Kaolin did not show obvious corrosion post reaction
mental S formation was observed. (e.g. Fig. 4D, F (6)). Minor pitting on k-feldspar, albite and Na–Ca-
plagioclase surfaces was observed (Fig. 4F).
The surface corrosion on H 279 observed in this study appeared
3.3.1.3. Hutton. The Hutton 279 sub-sample reacted here appeared to visibly more extensive than in previously published reactions with pure
contain a relatively higher organic matter, Fe-oxide, and siderite CO2 or SO2–CO2 (Pearce et al., 2015).
content (Fig. 4A, and supplementary material) when compared with
sub-samples previously reacted from this depth section of core. Siderite
was observed in SEM-EDS to be variable with Fe and also Fe–Mg 3.3.2. Water chemistry and geochemical modelling
signatures ± Mn and occasional associated trace signatures of Ca. 3.3.2.1. Precipice. Changes in measured ex-situ pH, conductivity, and
Biotite was occasionally observed often chloritized. The Ca content dissolved element concentrations during P 386 O2–SO2–CO2 water rock
(Table 3) was lower for this core sub-section than for similar sub- reaction are detailed in Table 5a, b, and Supplementary Information.
sections reacted previously (Farquhar et al., 2015). Trace signatures of After O2–SO2–CO2 gas injection, solution pH sharply decreased to 1.23.
Ca-phosphate, Cu–Fe-sulphide, apatite, barite cements, Cu–Sn oxide/ Conductivity increased coincident with an increase in the concentration
carbonate, zircon and plagioclase solid series were observed. Generally of total S in solution (Table 5 b). The dissolved concentration of S
more Ti–Fe oxide cement (Fig. 4B, inset), identified as illmenite in continued to increase gradually until the termination of the experiment,
previous microprobe analysis, was observed in this sub-sample and pH generally continued to decrease to ~1.13. On depressurisation

6
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 2. P 386 SEM pre and post gas-water-rock reaction. A) Surface view of Fig. 3. E 289 SEM images pre and post gas-water-rock reaction. A) Surface view
quartz grains, note central bright siderite spec, upper left bright rutile spot, and pre-reaction, with bright fine grained ankerite and chlorite (1), next to dark
B) surface view post-reaction. C) Kaolin pre-reaction and D) kaolin post-reac- organic matter, inset view (500x) of ankerite and chlorite. B) Surface view post-
tion with minor corrosion (1). E) Pre-reaction Ti–Fe oxide surrounded by reaction with ankerite corroded. C) K-feldspar pre-reaction with inset view
quartz, and F) post-reaction showing only very minor corrosion. G) Pre-reac- (2000x) of a sanidine surface showing the high surface area. D) K-feldspar post-
tion, pyrite (2) on kaolin. H) Pre-reaction, Cu–Sn signatures (3) on kaolin. reaction with minor pitting on surface. E) Ca–Na-plagioclase surface pre-reac-
tion, and F) post-reaction with the Ca–Na-plagioclase surface corrosion. G)
Fine-grained ankerite/siderite and chlorite pre-reaction, and H) post-reaction
the pH of the quench fluid was higher at 1.26, with residual fluids near
with ankerite dissolved and chlorite dissolved or detached from the surface.
the core sample having a pH of ~1.5. Solution conductivity increased to
20.49 ms/cm3 by experiment termination. This indicates dissolution
and conversion of SO2 gas to form sulphuric acid. The measured adjacent sub-section, indicting either some contribution from the re-
dissolved S is assumed to exist mainly as measurable dissolved actor or from oxidative dissolution of trace amounts of sulphide mi-
sulphate through oxidation including by co-injected oxygen, and nerals such as sphalerite, or carbonates, both present in higher con-
additionally by oxide minerals when rock was present. centrations in the reacted sub-sample. The blank experiment with no
The concentration of elements in solution from silicate and trace rock or minerals present indicated some contribution of elements in
carbonate dissolution generally increased over 384 h. However, Fe solution from reactor corrosion including up to 17 mg/kg of Fe,
concentration increased to 120 mg/kg then subsequently decreased 2.66 mg/kg Ni and 2.19 mg/kg Cr (Table 5). The pH in the blank ex-
after ~120 h. This decrease was likely due to precipitation of Fe bearing periment was slightly higher than the P 386 experiment, with dissolved
phases such as hematite, goethite or Fe-oxyhydroxide. Dissolved Fe S concentration lower. Dissolved S in the blank experiment originated
measured during reaction was correlated with Mn (R2 = 0.99), and from SO2 dissolution and oxidation by co-injected O2. A gas head ex-
with Co with an R2 of 0.99. Concentrations of elements including As, isted above the aqueous phase containing excess O2–SO2–CO2 gas. Fe
Co, Rb and V were higher at the termination of experiments as shown in oxide minerals present in the P 386 rock sample were available to
Table 5 and Supplementary Information An initial alkalinity of further oxidise SO2 generating more sulphuric acid (lower pH) and a
~149 mg/kg as CaCO3 before mixed gas injection was measured and higher measurable sulphate concentration in solution. Oxidation of SO2
used for modelling input, with subsequent measured alkalinity below likely fixed S in solution as sulphate which was measurable ex-situ,
detection, the dissolved Cl concentration was below 1 mg/kg. unlike dissolved gases such as H2S which could be desorbed on sam-
Element concentrations in solution during the P 386 (and other rock pling and depressurisation (measured dissolved total S was therefore
samples) experiments were generally higher indicating the main con- assumed to be sulphate).
tribution from rock reaction. The concentration of dissolved Zn and Co Previous reactions of P 386 with SO2–CO2 gas indicated lower levels
were slightly higher than might be expected from the rock content of an of sulphate in solution likely owing to the generation of both sulphate

7
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

this core likely contributing to the higher reactive surface areas needed
(Farquhar et al., 2015). For siderite, surface areas were decreased from
10 to 5, for ankerite from 10 to 3, and hematite from 70 to 0.01 cm2/g.
This is consistent with the cementing nature of these minerals.
Simulated in situ pH was in good agreement with experimentally
measured ex situ values (Fig. 5C). Trace amounts of siderite, chlorite,
calcite, and chalcedony (note amorphous silica has been generally ob-
served in this depth section of Precipice core), were the most reactive
phases, with additionally some dissolution of kaolinite, illite, albite,
and k-feldspar predicted (Fig. 5D). Precipitation of 0.01 g of hematite
matched the decrease in Fe observed experimentally (Fig. 5B, D).

3.3.2.2. Evergreen. During reaction of the E 289 rock sample, solution


pH sharply decreased after O2–SO2–CO2 gas injection, however, some
mineral buffering is evident though the subsequent slight increase from
pH 1.57 to 1.62 (Table 5). A higher pH of 1.79 was measured after
depressurisation on quench fluid, and residual fluid near the core
sample had a pH of ~2.3. Solution conductivity was lower than during
P 386 reaction, increasing up to 7.37 ms/cm3. The concentration of
most elements in solution continued to increase during the experiment,
with dissolved S, Fe, Al and Si the most abundant (Table 5). The
concentration of sulphate was measured for selected samples, and
increased from initially 9 mg/kg ( ± 10%), to 2446 mg/kg at 216 h and
4175 mg/kg at 408 h indicating the majority of measured dissolved S
existed as sulphate. Fe, Al and Mg likely partly originated from the Fe
and Mg rich chlorite observed in SEM to be leached of Fe. Ca and some
component of the Fe and Mg were also likely from ankerite and siderite
dissolution. Mn content in the unreacted rock was observed in EDS
analysis to be mainly associated with carbonates ankerite and siderite,
and also chlorite. Dissolved Co and Sr were correlated with Fe (with R2
of 0.997 and 0.994, respectively). The initial concentration of Cl was
below 1 mg/kg, with alkalinity before mixed gas injection ~229 mg/kg
CaCO3 and subsequently below detection. Trace elements in solution
including Pb, As, V, Rb generally had increased after 408 h reaction,
except for Sn which showed a decrease in concentration
(Supplementary Information Table S5). The majority of trace
Fig. 4. H 279 SEM pre and post gas-water-rock reaction. A) Surface view pre- elements in solution are expected to originate from mineral
reaction with bright areas including siderite (1), Fe-chlorite, and Fe-oxides, and
dissolution or surface desorption as the concentrations were above
darker kaolin and organic matter. B) Surface view post-reaction with some
blank reaction levels (Table S5).
bright chlorite and siderite corroded. Inset bright Ti–Fe-oxide, and dark organic
matter still present post-reaction. C) Fine grained siderite on chlorite (2), and
The concentration of total S in solution was lower than for the P 386
kaolinite (3) pre-reaction. D) Post-reaction view from C), siderite and chlorite reaction (Figs. 5a and 7a), and similar to the blank experiment level
were corroded. E) Fine grained Fe-oxides (4) observed to be formed post-re- without rock. The lower mass of the E 289 rock sample and lower
action. F) Post-reaction, minor pitting on albite surfaces (5), bright Ti-oxide, amount of Fe oxides available (Fe-oxides were not observed in SEM) for
and kaolinite mainly appears unaltered (6). G) Pre-reaction Cu-sulphide (7) on further dissolved SO2 oxidation (Equ. 1) to measurable sulphate sub-
chlorite. H) Cu–Sn signatures (8) on clay. sequent to initial oxidation by O2 gas (Equ 2) could be one reason.
Previous work has indicated surface mediated reactions (hence greater
and H2S under more reducing conditions (Pearce et al., 2015). Higher with higher mass/surface area of rock) acting as a proton sink could be
Fe-oxide content (oxidising capacity) was also observed in SEM on the P the reason for increased CO2 gas dissolution and higher than expected
386 rock subsample surface reacted in experiments here, and is ex- dissolved inorganic carbon content in CO2 water rock experiments
pected to have contributed to oxidation of SO2. (Humez et al., 2014). Although SO2 dissolution mechanisms are not
Experimentally measured dissolved S and Si concentrations were well understood, the generation of bicarbonate in mineral buffering
subsequently converted assuming they were SO42− and SiO2 respec- reactions (e.g. Equ. 3) could remove protons and shift the solution
tively for comparison to model outputs. Increases in sulphate con- equilibrium of Equ. 2 to conversion of more SO2 producing more sul-
centration through gradual SO2 conversion in experiments could not be phate/sulphuric acid. This process would be controlled by the amount
modelled as further addition of SO2 was not possible during the model. of available minerals (carbonates and possibly oxides) and hence mi-
Geochemical modelling of the P 386 experiment is shown in Fig. 5 neral mass in the system. This was also reported separately for a Marl
(lines). A reasonable match with experimental dissolved element con- cap-rock reaction, where an initial presence of sulphate yielded HSO4
centrations (icons in Fig. 5 A, B) was obtained by increasing reactive deprotonation (~1% of the total sulphate) (Dávila et al., 2017).
surface areas of several silicates and decreasing those of carbonates. SO2 + Fe2O3 + 4H+ ↔ H2SO4 + 2Fe2+ + H2O (2)
The reactive surface area of chalcedony was increased from 10 to
1000 cm2/g, chlorite from 70 to 10000, kaolinite from 70 to 700, and SO2 + H2O +½O2 ↔ HSO4- + H+ ↔ SO42− + 2H +
(3)
anorthite from 10 to 20 cm2/g. This is consistent with the fine grain
CaCO3 +H+ → HCO3− + Ca 2+
(4)
sizes of these clays, and initially altered/weathered nature of the sur-
faces of plagioclase and quartz resulting in higher initial surface areas. The O2–SO2–CO2 water reaction of the lower Evergreen (E 368)
In addition colloidal silica/chalcedony has been reported in sections of core, although not reacted experimentally here (owing to limited

8
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Table 5
a, b and c: Water chemistry during O2–SO2–CO2 water rock reaction; pH, conductivity (mS/cm3), major and minor elements in solution by ICP-OES (mg/kg),
DL = detection limit. * Ni and Cr concentrations are higher than expected, as described in the text.
DL Time/h pH Cond. mS/cm3 Al Ba Ca Fe K

5.40E-03 1.12E-04 2.74E-03 1.49E-02 5.99E-04

Blank 384 1.35 9.68 0.20 0.00 1.18 17.27 0.31

P 386 0 4.55 0.03 0.04 0.01 16.55 1.41 0.11


48 1.23 15.23 4.15 0.04 21.96 20.97 0.38
120 1.19 17.66 4.87 0.07 51.46 120.05 0.47
216 1.23 18.43 5.33 0.05 46.77 97.44 0.53
288 1.13 19.41 5.71 0.05 41.42 89.39 ND
384 1.13 20.49 6.38 0.05 36.89 40.07 0.59

E 289 0 5.60 0.08 0.05 0.00 13.27 1.97 0.31


48 1.57 4.86 29.18 0.11 58.42 185.41 6.57
120 1.48 4.60 58.57 0.11 65.91 326.75 4.62
216 1.56 4.61 88.55 0.09 54.74 421.86 6.14
288 1.61 6.99 114.77 0.08 41.61 440.02 ND
384 1.62 7.37 98.51 0.07 42.09 447.74 7.15

H 279 0 6.01 0.16 0.36 0.02 62.69 2.32 0.49


72 1.61 10.24 183.66 0.13 73.45 555.09 2.20
168 1.70 8.31 229.12 0.08 47.30 700.01 8.07
240 1.82 6.98 267.20 0.03 44.20 793.68 8.57
336 1.86 7.46 351.58 0.01 54.26 950.99 4.49
408 1.84 7.31 343.28 0.01 65.27 817.43 0.20

DL Time/h Mg Mn Na P S Si Sr

4.10E-04 1.96E-04 1.80E-02 5.12E-03 1.09E-01 1.32E-02 2.16E-04

Blank 384 0.10 0.38 1.78 0.11 1233.85 0.70 0.00

P 386 0 0.28 0.08 1.29 0.16 3.70 0.70 0.02


48 1.51 0.36 2.90 0.42 1398.95 8.17 0.08
120 1.94 2.74 5.48 0.13 1448.88 11.09 0.10
216 1.87 2.39 2.51 0.20 1604.24 11.83 0.10
288 1.98 2.21 2.45 0.29 1707.70 12.78 0.10
384 1.95 1.04 6.40 0.66 1697.15 13.33 0.11

E 289 0 0.15 0.08 22.28 0.10 3.54 2.03 0.01


48 10.79 4.45 43.01 ND 894.42 28.41 1.03
120 20.90 7.57 47.60 0.64 1033.65 53.41 1.61
216 30.52 8.69 51.96 1.03 1101.80 80.36 1.99
288 35.56 8.61 52.02 ND ND 88.16 2.09
384 35.03 8.48 53.21 1.15 1139.61 97.05 2.09

H 279 0 0.67 0.00 25.22 0.40 11.88 3.54 0.05


72 37.57 0.01 25.86 ND 1412.19 124.34 0.61
168 47.25 0.01 25.37 1.52 1424.18 153.75 0.72
240 55.91 0.01 26.04 0.71 1492.66 172.41 0.85
336 74.99 0.01 32.00 0.85 1760.70 205.35 1.13
408 75.16 0.01 30.64 0.84 1595.15 196.26 1.17

DL Time/h Ni* Cr* Zn Co Cu

8.21E-04 3.35E-04 6.08E-04 5.24E-04 5.58E-04

Blank 384 2.66 2.19 0.08 0.05 0.23

P386 0 0.23 0.15 0.1 0.01 0.05


48 0.65 1.16 0.75 0.06 0.31
120 17.03 23.86 0.52 0.42 0.41
216 14.04 18.65 0.54 0.34 0.42
288 12.10 16.82 0.57 0.32 0.34
384 3.80 5.86 0.57 0.12 0.76

E289 0 0.39 0.29 0.05 0.02 0.01


48 1.90 5.12 1.24 0.76 0.66
120 2.51 5.88 2.08 1.36 0.80
216 1.42 3.53 2.63 1.65 0.68
288 1.69 2.98 – – 0.71
384 1.09 1.31 1.90 1.76 0.56

H279 0 0.30 0.16 0.14 < DL 0.07


(continued on next page)

9
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Table 5 (continued)

DL Time/h Ni* Cr* Zn Co Cu

8.21E-04 3.35E-04 6.08E-04 5.24E-04 5.58E-04

Blank 384 2.66 2.19 0.08 0.05 0.23

72 3.30 4.83 2.75 0.36 0.73


168 3.65 3.75 2.77 0.42 1.01
240 4.60 4.54 3.18 0.45 1.05
336 5.23 4.90 4.04 0.59 1.16
408 4.22 3.31 3.91 0.59 0.86

available core), was modelled to investigate likely pH buffering at the Fig. 7) was obtained in geochemical modelling of E 289 reaction (lines
interface or reservoir rock and seal. Baffling of acidified formation Fig. 7). The surface area of chalcedony was increased from 10 to
water could be expected to occur in the reactive lower Evergreen core 10,000, K-feldspar from 10 to 7000, albite from 10 to 9000, and
where horizontal permeability is higher than vertical permeability chlorite from 70 to 200,000 cm2/g to match the experimental ion re-
(Farquhar et al., 2015). A previously determined mineralogy was used lease. This is consistent with the initially altered/weathered appearance
as input with an initial formation water chemistry from previous water- of these grains and mineral surfaces, resulting in initially higher surface
rock equilibration of E 368 samples (unpublished data)(Farquhar et al., areas. The surface area of illite was however decreased from 70 to
2015). Rock sample total mass and surface areas from the H 279 model 0.1 cm2/g (note on the rock surface illite was generally not easily ob-
were generally used. Model outputs are shown in Fig. 6, with a high served and may have been covered by other minerals). The surface
concentration of dissolved sulphate (Fig. 6A), Fe and Ca (Fig. 6B) areas of siderite were decreased from 10 to 5, and ankerite from 10 to
predicted. The solution pH was predicted to be buffered to ~2 after 0.1 cm2/g, consistent with their cementing morphology. Solution pH
408 h reaction. Calcite and chlorite were the most reactive minerals was in reasonable agreement, with a slightly higher predicted pH ~2
present, with 90% of chlorite and 100% calcite predicted to react likely reflecting the inability to model additional SO2 dissolution during
(Fig. 7D). Minor illite (0.003 g) and chalcedony precipitation was pre- the reaction which maintained a slightly lower experimental pH
dicted, with chalcedony precipitation likely indicating the incongruent (Fig. 7C). Chlorite and ankerite were the main minerals dissolving and
dissolution of chlorite as described below. Smectite (nontronite), he- contributing to dissolved elements with also K-feldspar, albite, and illite
matite, jarosite, and goethite were saturated in the model; gypsum and dissolution (Fig. 7D). In the model of E 289 reaction, 87% of the
anhydrite were not saturated. chlorite present in the rock sample reacted. Chalcedony precipitation
A good match with experimental element concentrations (icons was included in the model to match experimental observations of SiO2

Fig. 5. Model outputs from P 386 O2–SO2–CO2 water reaction (lines) with experimental data as icons: A) and B) Concentration (mg/kg) of fluid components in
solution over 384 h's reaction, note the different scales in the y axis. C) Solution pH, D) minerals (g) dissolved and precipitated (chalcedony not shown).

10
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 6. Model outputs from E 368 O2–SO2–CO2 water reaction (lines): A) and B) Concentration (mg/kg) of fluid components in solution over 384 h's reaction, note the
different Y axis scales. C) Solution pH, D) minerals (g) dissolved (chalcedony not shown).

Fig. 7. Model outputs from E 289 O2–SO2–CO2 water reaction (lines) with experimental data as icons: A) and B) Concentration (mg/kg) of fluid components in
solution over 384 h's reaction, note the different Y axis scales. C) Solution pH, D) minerals (g) dissolved (chalcedony not shown).

11
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 8. Correlation of dissolved concentrations (mg/kg) in solution during O2–SO2–CO2 water reaction. A) Co and Sr during reaction of E289 core. B) Fe and Sr during
reaction of E289 core. C) Fe and Al during reaction of E289 core. D) Sr and Mg during reaction of H279.

concentration. However, this likely reflects the incongruent dissolution proxy for the silica-rich chlorite structure remaining after leaching of
of Fe rich chlorite. Under acidic conditions the dissolution of chlorite the Fe and Al ions to correct the aqueous silica concentrations. Smectite
has been shown elsewhere to be incongruent with preferential leaching (nontronite), hematite, jarosite, and goethite were saturated in the
of Fe and Al to leave a silica rich layer (Brandt et al., 2003). The pre- model, however allowing their precipitation could not be reconciled
cipitation of secondary silica phases was not observed in experiments with matching the experimentally observed water chemistries, also
here, and has not been observed during chlorite dissolution experi- observable quantities of these minerals were not observed to be formed
ments performed elsewhere (Black and Haese, 2014; Smith et al., on mineral surfaces in SEM.
2013). Hence, the precipitation of chalcedony in models was used as a

Fig. 9. Model outputs from H 279 O2–SO2–CO2 water reaction (lines) with experimental data as icons: A) and B) Concentration (mg/kg) of fluid components over
408 h's reaction, note the different Y axis scales. C) Solution pH, D) minerals (g) dissolved (chalcedony not shown).

12
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 10. E 368 pre- and post- O2–SO2–CO2 wet supercritical fluid reaction. A) Surface view pre-reaction indicating the courser grain size compared to the E 289
Evergreen core. B) (Ca–Mg) Fe-aluminosilicate (amphibole) surrounded by degraded k-feldspar present pre-reaction. C) Fe–Mg-chlorite (1) and k-feldspar (2) grains
pre-reaction. D) Colloidal silica (3), Ti oxide (4) and k-feldspar (5) pre-reaction. E) Post reaction, a disaggregated silicate grain with Fe oxide coating, including Mg
and Ba elemental signatures, surrounded by Ca-sulphate flat bladed crystals (6). F) EDS spectrum of Ca-sulphate shown in E formed during reaction.

3.3.2.3. Hutton. Experimental pH decreased sharply after mixed gas subsequently fell below detection after mixed gas injection. Trace
injection during H 279 reaction; subsequently it was buffered to pH element concentrations in solution were generally higher compared to
1.84 by reaction termination (Table 5a). Residual fluid near the core the Precipice Sandstone and Evergreen Formation core experiments
sample had a higher pH on depressurisation of ~2.2. Solution prior to gas mixture injection. However many dissolved trace element
conductivity increased to 10.24 ms/cm3 after mixed gas injection and concentrations including Ti, V, Cu, As, Pb appeared to have decreased
subsequently decreased slightly to 7.31 ms/cm3. Fe concentration by the experiment termination (Supplementary Information Table S5).
reached 951 mg/kg then decreased slightly to 817 mg/kg by This indicates their co-precipitation with the observed precipitated Fe-
experiment termination. Dissolved Fe likely originated from chlorite oxides, or adsorption onto precipitated mineral surfaces. Adsorption on
and siderite, with subsequent precipitation of Fe oxides (observed in organic matter surfaces is also possible as this sample contained coal.
SEM) contributing to the slight reduction in dissolved Fe concentration The concentration of S measured in solution was generally high,
after 336 h. Element concentrations including Mg, were generally reaching 1760 mg/kg at 336 h with a subsequent slight decrease by the
higher in solution compared to the E 289 and P 386 experiment, end of the experiment (Table 5b). Again mineral buffering removing
likely partly owing to the larger mass of this core sample, however, Mn acidity likely caused the conversion of further SO2 to sulphate/sul-
and Na concentrations were lower (Table 5b). Although Sr only phuric acid in solution. The relatively high Fe-oxide content of this
increased to ~1 mg/kg, it had a strong correlation with Mg with an rock, as for the P 386 core, likely contributed to further SO2 oxidation
R2 of 0.998 (Fig. 8). The correlation of Mg with Sr indicates potentially (with oxidation by co-injected O2) to reach a similar sulphate con-
a similar source for these elements. Previous microprobe analysis of centration in solution as observed during P 386 reaction. The sub-
similar H 279 sections indicated Mg content in illmenite (Ti–Fe oxide) sequent slight decrease in sulphate concentration could indicate pre-
and muscovite as well as chlorite (Farquhar et al., 2015). Dissolved Mn cipitation of S bearing phases (Fe sulphates) in agreement with Fe and S
was somewhat correlated with Fe (R2 = 0.99). Fe was also correlated signatures observed in post-reaction SEM on the rock surface.
with Si (R2 = 0.99). On reaction, bicarbonate alkalinity was initially Modelled Fe, Al, Na, Ca and K concentration in solution (lines;
~510 mg/kg as CaCO3 in good agreement with previous experimental Fig. 9A and B) were in good agreement with experimental observations
work on other sections of the Hutton core (Horner et al., 2014), and (icons Fig. 9A and B). The surface area of chalcedony was increased

13
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 11. E 289 pre- and post- O2–SO2–CO2 wet supercritical fluid reaction. A) Chlorite (1), and muscovite/biotite (2) pre-reaction. B) Ankerite pre-reaction. C) K-
feldspar (3) and sphalerite (4) pre-reaction. D) Albite grain (5), muscovite/biotite (6), ankerite (7), and quartz (8) pre-reaction. E) Post reaction surface covered by
fine-grained precipitates with Fe S and O signatures in EDS. F) Magnified view of fine grained euhedral Fe–Na–K–Al-sulfate and Fe-sulphate precipitates.

from 10 to 1000 cm2/g, K-feldspar from 10 to 50, and chlorite from 70 formation core sample, E 368, was reacted in the wet O2–SO2–CO2 sc
to 200,000 cm2/g (consistent with the fine-grained nature of the fluid. The E 368 sample was previously characterised to contain
chlorite). Predicted Mg was high and likely reflects a lower Mg content reactive phases including calcite and chlorite as detailed in Table 1.
present in chlorite in the rock than that used in the model. Ankerite Smectite was previously identified as saponite (Farquhar et al., 2013).
surface area was decreased from 10 to 0.5 cm2/g, with siderite at This lower Evergreen core section has a courser grain size than E 289,
10 cm2/g. Very little ankerite was observed on the rock surface, an SEM image of the surface is shown in Fig. 10A. K-feldspar and Fe-
therefore the lower needed surface area may represent ankerite being chlorite were observed pre-reaction to have high surface areas owing to
covered by other minerals and not accessible to fluids. Predicted solu- previous diagenetic changes (Fig. 10 B, C). Both Fe-chlorite, and
tion pH was ~1.7 at 55 h in good agreement with experimental values chlorite containing mainly Fe with also some Mg were identified.
(Fig. 9C), but was subsequently up to 1 pH unit higher than measured Traces of biotite and a Fe–Ti phyllosilicate (possibly oxy phlogopite)
values. This likely reflects the inability of adding further SO2 to the were present. Amorphous silica and Ti oxides (Fig. 10 D) were also
model during reaction. Chlorite and siderite were the most reactive observed pre-reaction, in addition to the major minerals listed in
minerals, with K-feldspar and albite also dissolving. Loss of 75% of the Table 1. Sphalerite, FeS/pyrite, phosphate, garnet, and Cu–Sn
chlorite was predicted after 408 h. The precipitation of chalcedony and minerals were additionally observed.
very minor illite (0.0003 g) were predicted, however chalcedony pre- After reaction, rock disaggregation had occurred (owing to the
cipitation likely reflects the incongruent dissolution of chlorite as in the presence of swelling clays) and grains had a greenish colouration. Flat
E 289 model. Saturated phases in the model were nontronite (smectite), bladed crystals of precipitated Ca sulphate with a gypsum morphology
glauconite, hematite, jarosite and alunite. were observed around grains replacing calcite cement, with Fe oxide
coatings on grain surfaces and barite traces (Fig. 10 E, F). XRD analysis
of the rock sample post-reaction (Supplementary Information Table S6)
3.4. Wet supercritical fluid-rock reactions identified abundant quartz, major orthoclase, minor kaolinite, albite,
illite, Na-anorthite, and trace chlorite/montmorillonite. Ca-sulphate
3.4.1. Rock changes minerals, however, were not detected and were certainly below the
3.4.1.1. Evergreen wet supercritical fluid reaction. A lower Evergreen

14
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 12. H 279 pre- and post-wet O2–SO2–CO2 su-


percritical fluid reaction. A) Pre-reaction mixed
Fe–Mg-chlorite (1) with fine grained siderite, k-
feldspar (2), siderite traces (3). B) Post-reaction
view in A) showing Fe-chlorite and carbonates cor-
roded, with inset magnified view (700x). C) Pre-
reaction Ti-oxide, and D) post-reaction view of C
with precipitates covering pre-existing Ti-oxide
with Fe, O, K and S signatures (4) and fine grained
Fe oxides (5). E) Magnified view of D (4) showing
fine grained booklets of silicate and/or sulphate
precipitates (6) and Fe oxides (7). F) Post-reaction
surface view of long radiating crystals of Al–Fe –K-
sulphate precipitates (8).

XRD detection limit (~5%). (Fig. 12A).


The E 289 wet supercritical fluid (WS) reacted core sample, referred Fe-chlorite and siderite were corroded post-reaction (Fig. 12B) ap-
to as E 289WS, was similar before reaction to the E 289 sub-sample pearing to increase porosity and open up fine cracks around grains in
described in section 3.2.1.2 and Fig. 2, containing reactive silicates some areas. Fine grained precipitates with an Fe-(Na–K-)Al-sulfate
including Fe-chlorite (Fig. 11A), and carbonates including pore-filling signature covered the majority of the sample surface, which was visibly
and fine-grained ankerite (Fig. 11B). Ankerite and siderite contained orange/brown in areas. Fine-grained Fe-sulphate signatures had also
Mg and Mn signatures in EDS. K-feldspar and albite surfaces were not precipitated on biotite with occasional Cr, Ni, Cu and Mn signatures in
pristine pre-reaction (Fig. 11 C, D). Occasional grains of Na–Ca-plagi- EDS. Some precipitated Fe oxides also had Cr and Ni signatures. Ti-
oclase, rutile and Ti–Fe phlogopite were observed along with organic oxides present pre-reaction (Fig. 12C), had thin mats of fine-grained
matter. Trace sphalerite was also observed, along with occasional S and precipitates covering their surfaces post reaction (Fig. 12D). Some
Ni signatures or Ca and S signatures on chlorite (Fig. 11C). precipitates had the appearance of micron-sized clay booklets. Long
After reaction, the E 289WS rock sample surface was covered in radiating crystals were also observed to have been formed during re-
fine-grained precipitates (Fig. 11 E) making observations of underlying action, with an EDS signature suggesting an Al–Fe–K sulphate such as
mineral corrosion impossible. Precipitates were generally ~2 μm size alunite (Fig. 12F). The rock sample mass used was 0.8295 g. Alunite
euhedral crystals with Fe–Na–K(Al) sulphate signatures in EDS, poten- precipitation has been predicted elsewhere in geochemical model re-
tially jarosite (Fig. 11F). Some precipitates had Fe, S and O signatures of actions of CO2 and SO2 with saline brine and feldspar rich reservoir
a Fe-sulphate. rock (Xu et al., 2007).

3.4.1.2. Hutton Sandstone wet supercritical fluid reaction. The H 279 WS 3.5. 30 y geochemical model
sub-sample contained similar amounts of organic matter and Fe-oxides
as the H 279 rock sub-sample described in section 3.2.3.1 and Fig. 4. The P346 sample mineralogy was used as input for a 30 year
Albite and K feldspar grains were again not pristine with rough CO2–SO2–O2 model. The reactive surface areas determined from the
surfaces. Occasional Fe–Ca-silicate and rutile grains were present. Fe model of the gas-water-rock experiment were decreased for reservoir
(Mn) oxides existed as cements and grain coatings. Trace amounts of scale (as described in the methods) and used for input. The predicted
Zn–Fe-sulphide and Ca-phosphate were also observed. Some clay and change in minerals and the pH are shown in Fig. 13. The trace amount
oxide surfaces had Ca and S signatures in EDS. Fe-chlorite and siderite of calcite was included in the mineral input, and this dissolved along
containing Mn and trace Ca signatures were also present pre-reaction with chlorite mainly. Precipitation of goethite was an initial transitory

15
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Fig. 13. Predicted change in minerals and pH over 30 year CO2–SO2–O2 reaction using the reservoir P346 mineral input including the trace amount of calcite.

phase consuming O2, and subsequently anhydrite, ankerite, kaolinite with the observation of increased silicate mineral dissolution at the
and traces of siderite, and pyrite precipitated. Overall for this low re- lower pH generated in O2–SO2–CO2 experiments. The concentrations of
activity rock, rock volume decreased very slightly with no impact on dissolved Zn, Co, Ni and Cr are also higher in the O2–SO2–CO2 ex-
porosity. The lowered pH was buffered after 30 years to 4.77 by mineral periments compared to pure CO2 experiments by ~2–20 times. This is
dissolution and the groundwater. likely owing to oxidation of small amounts of sulphides including
sphalerite and pyrite and increased dissolution of clays, although the
3.6. Comparison to pure CO2 and other reactions core heterogeneity may have also played a role. Mobilised concentra-
tions of As, Cr, Co, Ni and Zn were also reported in experimental re-
Compared to previous experiments with pure CO2 where pH was actions of different cores from the Surat Basin with up to 38 bar pure
~4.5–6, the pH generated is much lower with SO2 and O2 gases present CO2 at 45 °C although the potential source minerals were not discussed
and small amounts of carbonates in the reacted rock (Farquhar et al., (Horner et al., 2014). Compared to previous SO2–CO2 experiments the
2015). The generated pH with O2–SO2–CO2 gas is slightly lower than dissolved concentration of Al is ~0–5 times higher, Si is similar or
with SO2–CO2 gas (~2–3), likely due to additional oxidation of the SO2 slightly higher in the O2–SO2–CO2 experiments, Sr, Mg, and Mn con-
gas (and potentially oxidation of sulphides present in the rock) (Pearce centrations are also similar except for H279 with a much lower Mn
et al., 2015). The Surat Basin is a very large area, with variation in the (Pearce et al., 2015). Here, concentrations of dissolved Fe appeared
formation properties across the basin, and some areas with little data. however to subsequently decrease with time in O2–SO2–CO2 experi-
In the area studied here in the North region, the Precipice Sandstone ments, especially for Hutton H279 as Fe was incorporated into Fe-oxide
aquifer has a reported pH range at depth of 6.7–8.2, with the Evergreen and sulphate minerals. Note changes to water chemistry are expected to
Formation reported pH range 7.7–7.9, and Hutton Sandstone 7.5–8.1 be higher in experiments than in aquifers where higher alkalinities,
(Feitz et al., 2014). All are within the recommended pH 6.5–8.5 for dilution and transport may minimise changes (Kharaka et al., 2010).
groundwater resources (NHMRC and NRMMC, 2011). Bicarbonate al- Dissolved sulphate concentration was high in all O2–SO2–CO2 ex-
kalinity in the three formations is variable, reported from ~50 to periments. Compared to the equivalent SO2–CO2 experiments, total S
1260 mg/kg in the region and generally higher in the Hutton Sand- was 3–5 times higher in O2–SO2–CO2 experiments from the co-injected
stone. Lowered pH through gas injection could be expected to be buf- O2 oxidising SO2 to sulphate which is not desorbed on sampling. The
fered by reservoir alkalinity over longer timescales, especially in car- concentration of SO2 present in the injection gas would partly control
bonate cemented zones, as predicted in the longer simulations the concentration of dissolved sulphate. Lower concentrations of SO2
performed here. The experiments performed here were at initial low than used here could be expected to be injected subsurface owing to
alkalinities in the range of the reservoir, and with low carbonate con- pipeline regulations. Although accumulation of impurity gases in the
tent cores. Pure CO2 experiments (Farquhar et al., 2015), and SO2–CO2 near well bore have the potential to increase concentrations initially
reactions of the same core sections have been reported elsewhere during CO2 injection, the precipitation of sulphates and S bearing mi-
(Pearce et al., 2015). Table 6 compares the observations from the nerals over longer time periods may subsequently decrease dissolved
current study with those other gas stream experiments. sulphate concentration.
Mineral precipitation was, generally, not observed in pure CO2 re- A previous experimental study reacted 0–8% O2 in CO2 with gothic
actions of the same or similar cores from the same well (Farquhar et al., shale and observed the initial increase and subsequent decrease of
2015; Horner et al., 2014). Compared to pure CO2 experiments, the dissolved Fe (and Ni) concentration as Fe oxides were precipitated.
dissolved concentration of Fe was ~10–30 times, Si ~5–15 times, Al Lower dissolved U concentrations with higher injected O2 were also
~10–100 times, Mg 2–4 times, and Mn 2–15 times higher in attributed to co-precipitation with Fe oxides (Jung et al., 2013). The
O2–SO2–CO2 experiments (Farquhar et al., 2015). This is consistent precipitation of Fe oxides and Ca sulphate were observed in the current

Table 6
Comparison of experiment observations from this study and the published pure CO2 or SO2–CO2 reactions of the same core. Observations differed with core type but
are generalised here.
Gas stream pH Dissolution Precipitation

CO2 4.5–6 Traces of carbonates, chlorite


SO2–CO2 2–3 Carbonates, chlorite, albite, K-feldspar, kaolinite S, gypsum
O2–SO2–CO2 1.1–1.9 Carbonates, chlorite, albite, K-feldspar, kaolinite, ilmenite. Generally stronger corrosion and Fe oxide + Cr, Fe-sulphate
clay collapse
Wet gas head Carbonates, note precipitates obscured observations of corrosion Gypsum, Fe oxide + Cr,Ni, Fe-sulphate, alunite,
O2–SO2–CO2 jarosite

16
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

study. Fe oxide and gypsum formation have been reported in a previous 4. Conclusions
study on O2–CO2 brine reaction of Gothic shale cap-rock containing
pyrite, with gypsum also formed in O2–SO2–CO2 reaction of basalt and With the presence of both 0.016% SO2 and 2% O2 impurities in CO2-
in SO2–CO2 water reaction with calcite cemented Hutton Sandstone water-rock reactions, sulphuric acid generation resulted in a low solu-
(Jung et al., 2013; Pearce et al., 2015). tion pH (~1–2) in experiments. Relatively low reactivity quartzose
If all the 1% calcite in the lower Evergreen sample E 368WS reacted reservoir rock from the Precipice Sandstone had minimal pH buffering
here were converted during wet sc fluid reaction to gypsum, this would capacity, although Fe–Ti oxides (ilmenite) and Fe-oxide grain coatings
result in a 1% volume increase of the rock (by mass balance), or 0.3% likely contributed to oxidation of co-injected SO2. Dissolution of re-
volume increase if the calcite were converted to anhydrite. If 50% of the active minerals from Hutton Sandstone and Evergreen Formation core
calcite present were converted, with the other ~50% transferred to was directly observed including mainly Fe-chlorite, ankerite, siderite,
solution and transported away (as suggested by experiments), the vo- and also albite and K-feldspar. The initially low solution pH was buf-
lume increase would be 0.3% with formation of gypsum, or 0.1% vo- fered to a greater extent in these experiments. Increases in concentra-
lume increase with anhydrite formation potentially reducing CO2 mi- tions of several cations were observed. Dissolved Mn, Co, +/− Sr, Mg,
gration. Measurements of changes to core porosity and permeability and Si were correlated with Fe. Precipitation of minerals including Ca-
through experiments and natural analogue studies are needed in future or Fe-sulphates and Fe oxides appeared to subsequently reduce some
to validate predictions and models. There is longer-term potential for dissolved element concentrations, with Cr and Ni signatures occasion-
self-sealing of cap-rock, however in the time scales of the experiments ally observed trapped in new mineral precipitates. Compared to pre-
and models performed here rock mass decreased after impure CO2- vious pure CO2 experiments, stronger clay, plagioclase and K-feldspar
water or wet supercritical fluid reaction indicating initial corrosion. The mineral dissolution was observed through mineral surface corrosion
Precipice Sandstone sample reacted with O2–SO2–CO2 gas dissolved in features in SEM, and higher concentrations of ions such as Al, Fe, and Si
water lost 0.5% mass, with the Evergreen and Hutton samples losing up at the lower pH. Higher concentrations of dissolved cations associated
to 2.4% mass. The H 279 WS and E 289 WS samples reacted in wet with oxidative dissolution of sulphides, including Zn, were also mea-
supercritical fluid lost 1.3 and 1.7% of their original masses respec- sured. The stronger clay and plagioclase dissolution can provide more
tively. cations for precipitation of carbonate, sulphate, clay and sulphide mi-
Park et al., reacted sandstone containing chlorite with CO2–SO2 at nerals especially in the more reactive cap-rocks at the gas water in-
0–1% impurity concentrations, and subsequently modelled the terface. This may help self-seal cap-rocks over longer timescales. The
CO2–SO2 reactions over 25 days (Park et al., 2018). They also found an potential for oxidative dissolution of sulphides however should be
overall decrease in rock mass owing to a generated low pH and dis- further investigated as they could release heavy metals.
solution of chlorite mainly in agreement with this study. The low Longer term experiments with lower gas impurity concentrations
content of calcite in their case resulted in no precipitation of gypsum or including NOx to replicate expected PCC or oxyfuel capture gases are
anhydrite. In the 30 year simulation of reaction of the Precipice recommended in future work to investigate likely reactions resulting
Sandstone core performed here, the rock volume also decreased slightly from injection of CO2 from coal combustion sources. Investigating the
overall with essentially no effect on porosity. This was owing to the low impacts to rock porosity and permeability which impact CO2 migration
amount of reactive minerals present. The small amount of calcite was are also suggested.
converted to anhydrite, however the volume change was too small to Geochemical models were in good agreement with experiments,
affect the porosity. Bacon et al., built a reactive transport model for CO2 indicating Fe-chlorite and minor carbonates were the most reactive
and CO2 with 1% SO2 reacting with the saline Rose Run Sandstone or minerals, with prediction of Fe oxide precipitation. Initial predictions of
Copper Ridge dolomite for 10 years (Bacon et al., 2009). They predicted solution pH were in good agreement, although experimentally further
dissolution of calcite and to some extent dolomite, and precipitation of SO2 likely dissolved from a supercritical gas head during the experi-
anhydrite close to the injector, however the volumes were not sig- ments further reducing pH which could not be modelled using the
nificant enough to change the reservoir properties. In contrast Xu et al., current method. Incorporating the correct (or best possible) mineral
predicted increases to porosity in the near well bore acidified zone with stoichiometry's, and calculating the initial reactive surface areas mod-
CO2 and SO2 reaction of a plagioclase rich sandstone over 10–10,000 ified based on observed grain size and mineral habit improves model
years (Xu et al., 2007). Precipitation of alunite and anhydrite decreased predictions. Parameterising with experimental data further improves
porosity further from the injector, with ankerite, silica, Na-smectite, predictive capacity. Surface areas had to be generally increased for si-
dawsonite, siderite and magnesite precipitation also predicted. Com- licates and fine grain size clays, and decreased for cementing carbo-
paring model outcomes with alteration observed in natural analogue nates to match experimental data in this study. Using the parameterised
data from basins with natural occurrences of CO2 is vital for validation upscaled reactive surface areas in a 30 year simulation predicted that
of the predictions. Kaszuba et al., studied the Madison Limestone as a porosity and pH would not be significantly impacted in a low reactivity
natural analogue for carbon and sulphur co-sequestration (Kaszuba sandstone with co-injection of O2 and SO2. Further work should include
et al., 2011). With sources of Ca and Fe from calcite and dolomite, the building of a reactive transport model for co-injection including the
anhydrite trapped sulphate, and subsequently pyrite, and carbonates re- overlying formation to determine if continued co-injection can strip
precipitated. In our case calcite and chlorite are the main mineral impurity gases in close proximity to the injector and further decrease
sources of Ca and Fe but the predicted minerals precipitated are con- pH.
sistent with the findings of Kaszuba. The alteration of chlorite to side- Observable reactivity of siltstone and sandstone (cap-rock) core to
rite or ankerite and kaolinite has been observed in other natural ana- wet supercritical O2–SO2–CO2 fluid indicated further work is needed to
logue situations (Higgs et al., 2015; Watson et al., 2004). Reactivity of determine likely changes to storage cap-rock integrity in contact with
silicates including Fe-chlorite and plagioclase were observed in current buoyant supercritical CO2. Corrosion of reactive minerals including
experiments to release Fe and Ca cations, a source of ions for mineral ankerite and chlorite, and precipitation of new (Fe) oxide and sulphate
trapping of CO2 as siderite, ankerite or calcite. Sulphate has ad- phases were observed on rock surfaces. Future work investigating
ditionally been shown elsewhere to increase the dissolution rates of changes to cap-rock porosity and permeability through reaction of
silicates such as plagioclase, hence SO2 co-injection could potentially impure wet supercritical CO2 (low water availability regime) is sug-
enhance mineral trapping of CO2 (Min et al., 2015). gested to determine if reactions will result in self-sealing.

17
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Acknowledgements with carbonate minerals under co-sequestration conditions: a combined experimental


and theoretical study. Geochem. Cosmochim. Acta 92 (0), 265–274.
Higgs, K.E., Haese, R.R., Golding, S.D., Schacht, U., Watson, M., 2015. The Pretty Hill
The authors acknowledge the funding provided by the Formation as a natural analogue for CO2 storage; an investigation of mineralogical
Commonwealth of Australia and industry sponsors through the and isotopic changes associated with sandstones exposed to low, intermediate and
CO2CRC Program "Reactive Rocks" (Australia) for part of this work, and high CO2 concentrations over geological time. Chem. Geol. 399, 36–64 Measuring
and predicting the geochemical impacts of CO2 storage on reservoir rocks.
Ralf Haese for management and technical feedback in the Reactive Hodgkinson, J., Grigorescu, M., 2012. Background research for selection of potential
Rocks program. The University of Queensland, Australia, is acknowl- geostorage targets—case studies from the Surat Basin, Queensland. Aust. J. Earth Sci.
edged for financial support for part of this project through a new staff 60 (1), 71–89.
Hodgkinson, J., et al., 2010. The Potential Impact of Carbon Dioxide Injection on
research start-up fund. Susan Farquhar is thanked for access to rock Freshwater Aquifers: the Surat and Eromanga Basins in Queensland 9781921489570.
core samples. Dept. of Employment, Economic Development and Innovation, Geological Survey of
Special thanks to Dean Biddle for technical assistance with the high Queensland, Brisbane.
Horner, K.N., Schacht, U., Haese, R.R., 2014. Characterizing long term CO2-water-rock
pressure and temperature experiments. Marietjie Mostert of the
reaction pathways to identify tracers of CO2 migration during geological storage in a
Environmental Geochemistry Laboratory, CGMS, University of low-salinity siliciclastic reservoir system. Chem. Geol. https://doi.org/10.1016/j.
Queensland, is acknowledged for her assistance with chemical analyses, chemgeo.2014.09.021.
as is Alison Law. We acknowledge the facilities, and the scientific and Humez, P., Lions, J., Négrel, P., Lagneau, V., 2014. CO2 intrusion in freshwater aquifers:
review of geochemical tracers and monitoring tools, classical uses and innovative
technical assistance, of the Australian Microscopy and Microanalysis approaches. Appl. Geochem. 46 (0), 95–108.
Research Facility at the Centre for Microscopy and Microanalysis, The IEAGHG, 2011. Effects of Impurities on Geological Storage of CO2 2011/04 June, 2011.
University of Queensland. Three reviewers are thanked for their con- Jia, L., Tan, Y., Anthony, E.J., 2009. Emissions of SO2 and NOx during Oxy−Fuel CFB
combustion tests in a mini-circulating fluidized bed combustion reactor. Energy Fuels
structive comments that improved this manuscript. 24 (2), 910–915.
Jia, L., Tan, Y., Wang, C., Anthony, E.J., 2007. Experimental study of oxy-fuel combustion
Appendix A. Supplementary data and sulfur capture in a mini-CFBC. Energy Fuels 21 (6), 3160–3164.
Jung, H.B., Um, W., Cantrell, K.J., 2013. Effect of oxygen co-injected with carbon dioxide
on Gothic shale caprock–CO2–brine interaction during geologic carbon sequestration.
Supplementary data to this article can be found online at https:// Chem. Geol. 354 (0), 1–14.
doi.org/10.1016/j.apgeochem.2019.104400. Kaszuba, J.P., Navarre-Sitchler, A., Thyne, G., Chopping, C., Meuzelaar, T., 2011.
Supercritical carbon dioxide and sulfur in the Madison Limestone: a natural analog in
southwest Wyoming for geologic carbon-sulfur co-sequestration. Earth Planet. Sci.
References Lett. 309 (1–2), 131–140.
Kather, A., 2009. CO2 quality and other Relevant issues. In: 2nd Working Group Meeting
on CO2 Quality. Cottbus, Germany.
Amshoff, P., Weger, T., Ostertag-Henning, C., 2018. Dissolution kinetics of CO2 and CO2-
Kharaka, Y.K., et al., 2010. Changes in the chemistry of shallow groundwater related to
SO2 mixtures in water and brine at geological storage conditions of 16 MPa and
the 2008 injection of CO2 at the ZERT field site, Bozeman, Montana. Environ. Earth
333 K. Int. J. Greenh. Gas Cont. 79, 173–180.
Sci. 60 (2), 273–284.
Bacon, D.H., Sass, B.M., Bhargava, M., Sminchak, J., Gupta, N., 2009. Reactive transport
Kirste, D., Pearce, J., Golding, S., 2017. Parameterizing geochemical models: do kinetics
modeling of CO2 and SO2 injection into deep saline formations and their effect on the
of calcite matter? Procedia Earth Planet. Sci. 17, 606–609.
hydraulic properties of host rocks. Energy Procedia 1 (1), 3283–3290.
Köhler, S.J., Dufaud, F., Oelkers, E.H., 2003. An experimental study of illite dissolution
Black, J.R., Carroll, S.A., Haese, R.R., 2014. Rates of mineral dissolution under CO2
kinetics as a function of ph from 1.4 to 12.4 and temperature from 5 to 50°C.
storage conditions. Chem. Geol. https://doi.org/10.1016/j.chemgeo.2014.09.020(0.
Geochem. Cosmochim. Acta 67 (19), 3583–3594.
Black, J.R., Haese, R.R., 2014. Chlorite dissolution rates under CO2 saturated conditions
Kummerow, J., Spangenberg, E., 2011. Experimental evaluation of the impact of the
from 50 to 120°C and 120 to 200 bar CO2. Geochem. Cosmochim. Acta 125 (0),
ineraction of CO2-SO2, brine, and reservoir rock on petrophysical properties: a case
225–240.
study from the Ketzin test site, Germany. Geochem. Geophys. Geosyst. 12 (5), 1–10.
Brandt, F., Bosbach, D., Krawczyk-Bärsch, E., Arnold, T., Bernhard, G., 2003. Chlorite
Little, M.G., Jackson, R.B., 2010. Potential impacts of leakage from deep CO2 geose-
dissolution in the acid ph-range: a combined microscopic and macroscopic approach.
questration on overlying freshwater aquifers. Environ. Sci. Technol. 44 (23),
Geochem. Cosmochim. Acta 67 (8), 1451–1461.
9225–9232.
Choi, Y.-S., Nesic, S., Young, D., 2010. Effect of impurities on the corrosion behavior of
Loring, J.S., et al., 2011. In situ infrared spectroscopic study of forsterite carbonation in
CO2 transmission pipeline steel in supercritical CO2−Water environments. Environ.
wet supercritical CO2. Environ. Sci. Technol. 45 (14), 6204–6210.
Sci. Technol. 44 (23), 9233–9238.
Lowson, R.T., Brown, P.L., Comarmond, M.C.J., Rajaratnam, G., 2007. The kinetics of
Dávila, G., Cama, J., Luquot, L., Soler, J.M., Ayora, C., 2017. Experimental and modeling
chlorite dissolution. Geochem. Cosmochim. Acta 71 (6), 1431–1447.
study of the interaction between a crushed marl caprock and CO2-rich solutions
Lowson, R.T., Comarmond, M.C.J., Rajaratnam, G., Brown, P.L., 2005. The kinetics of the
under different pressure and temperature conditions. Chem. Geol. 448, 26–42.
dissolution of chlorite as a function of pH and at 25°C. Geochem. Cosmochim. Acta 69
Dávila, G., Luquot, L., Soler, J.M., Cama, J., 2016. Interaction between a fractured marl
(7), 1687–1699.
caprock and CO2-rich sulfate solution under supercritical CO2 conditions. Int. J.
Lu, J.M., Partin, J.W., Hovorka, S.D., Wong, C., 2010. Potential risks to freshwater re-
Greenh. Gas Cont. 48, 105–119.
sources as a result of leakage from CO2 geological storage: a batch-reaction experi-
Dawson, G.K.W., Pearce, J.K., Biddle, D., Golding, S.D., 2015. Experimental mineral
ment. Environ. Earth Sci. 60 (2), 335–348.
dissolution in Berea Sandstone reacted with CO2 or SO2-CO2 in NaCl brine under
Michael, K., 2014. Nearshore Quifer Modelling, ANLEC R&D Science Day. Enabling CCS
CO2 sequestration conditions. Chem. Geol. 399, 87–97 Measuring and predicting the
Deployment, Sydney, Australia.
geochemical impacts of CO2 storage on reservoir rocks.
Min, Y., Kubicki, J.D., Jun, Y.-S., 2015. Plagioclase dissolution during CO2–SO2 cose-
de Visser, E., et al., 2008. Dynamis CO2 quality recommendations. Int. J. Greenh. Gas
questration: effects of sulfate. Environ. Sci. Technol. 49 (3), 1946–1954.
Cont. 2 (4), 478–484.
NHMRC, NRMMC, 2011. Australian Drinking Water Guidelines Paper 6 National Water
Delany, J.M., Lundeen, S.R., 1989. The LLNL Thermodynamic Database. Lawrence
Quality Management Strategy. National Health and Medical Research Council,
Livermore National Laboratory Report UCRL, pp. 21658.
National Resource Management Ministerial Council, Commonwealth of Australia,
Duan, Z., Sun, R., 2003. An improved model calculating CO2 solubility in pure water and
Canberra.
aqueous NaCl solutions from 273 to 533 K and from 0 to 2000 bar. Chem. Geol. 193
Notz, R.J., Tönnies, I., McCann, N., Scheffknecht, G., Hasse, H., 2011. CO2 capture for
(3–4), 257–271.
fossil fuel-fired power plants. Chem. Eng. Technol. 34 (2), 163–172.
Farquhar, S.M., Dawson, G.K.W., Esterle, J.S., Golding, S.D., 2013. Mineralogical char-
Palandri, J.L., Kharaka, Y.K., 2004. A Compilation of Rate Parameters of Water-Mineral
acterisation of a potential reservoir system for CO2 sequestration in the Surat Basin.
Interaction Kinetics for Application to Geochemical Modeling. USGS Open File Report
Aust. J. Earth Sci. 60 (1), 91–110.
2004-1068. USGS Open File Report 2004-1068.
Farquhar, S.M., et al., 2015. A fresh approach to investigating CO2 storage: experimental
Palandri, J.L., Rosenbauer, R.J., Kharaka, Y.K., 2005. Ferric iron in sediments as a novel
CO2-water-rock interactions in a low-salinity reservoir system. Chem. Geol. 399,
CO2 mineral trap: CO2-SO2 reaction with hematite. Appl. Geochem. 20 (11),
98–122.
2038–2048.
Feitz, A.J., Ransley, T.R., Dunsmore, R., Kuske, T.J., Hodgkinson, J., Preda, M., Spulak,
Park, J., Choi, B.-Y., Jeong, J.O., Shinn, Y.J., 2018. Impact of SO2 on alteration of re-
R., Dixon, O., Draper, J., 2014. Geoscience Australia and Geological Survey of
servoir rock with Ca-deficient conditions and poor buffering capacity under a CO2
Queensland Surat and Bowen Basins Groundwater Surveys Hydrochemistry Dataset
geologic storage condition. Geofluids 2018, 13.
(2009-2011). Geoscience Australia, Canberra Australia. https://doi.org/10.4225/
Pearce, J., Dawson, G., 2018. Experimental determination of impure CO2 alteration of
25/5452D04771CCF.
calcite cemented cap-rock, and long term predictions of cap-rock reactivity.
Felmy, A.R., et al., 2010. The role of intermediates during metal carbonation of forsterite
Geosciences 8 (7), 241.
in wet supercritical CO2. Geochem. Cosmochim. Acta 74 (12) A284-A284.
Pearce, J.K., et al., 2018. Impure CO2 reaction of feldspar, clay, and organic matter rich
Gaus, I., 2010. Role and impact of CO2-rock interactions during CO2 storage in sedi-
cap-rocks: decreases in the fraction of accessible mesopores measured by SANS. Int. J.
mentary rocks. Int. J. Greenh. Gas Cont. 4 (1), 73–89.
Coal Geol. 185, 79–90.
Glezakou, V.-A., Peter McGrail, B., Todd Schaef, H., 2012. Molecular interactions of SO2
Pearce, J.K., et al., 2019. A combined geochemical and μCT study on the CO2 reactivity of

18
J.K. Pearce, et al. Applied Geochemistry 109 (2019) 104400

Surat Basin reservoir and cap-rock cores: porosity changes, mineral dissolution and 100°C and up to 600 bar. Geochem. Cosmochim. Acta 67 (16), 3015–3031.
fines migration. Int. J. Greenh. Gas Cont. 80, 10–24. Steefel, C.I., 2001. GIMRT, Version 1.2: Software for Modeling Multicomponent,
Pearce, J.K., Dawson, G.K.W., Law, A.C.K., Biddle, D., Golding, S.D., 2016a. Reactivity of Multidimensional Reactive Transport. User's Guide, UCRL-MA-143182. Lawrence
micas and cap-rock in wet supercritical CO2 with SO2 and O2 at CO2 storage con- Livermore National Laboratory, Livermore, California.
ditions. Appl. Geochem. 72, 59–76. Thaysen, E.M., Soler, J.M., Boone, M., Cnudde, V., Cama, J., 2017. Effect of dissolved
Pearce, J.K., et al., 2016b. Mineralogical controls on porosity and water chemistry during H2SO4 on the interaction between CO2-rich brine solutions and limestone, sandstone
O2-SO2-CO2 reaction of CO2 storage reservoir and cap-rock core. Appl. Geochem. and marl. Chem. Geol. 450, 31–43.
75, 152–168. Ting, T., Stanger, R., Wall, T., 2014. Oxyfuel CO2 compression: the gas phase reaction of
Pearce, J.K., et al., 2015. SO2 impurity impacts on experimental and simulated CO2- elemental mercury and NOx at high pressure and absorption into nitric acid. Int. J.
water-reservoir rock reactions at carbon storage conditions. Chem. Geol. 399, 65–86 Greenh. Gas Cont. 29 (0), 125–134.
Measuring and predicting the geochemical impacts of CO2 storage on reservoir rocks. Wang, J., et al., 2012. The effect of impurities in oxyfuel flue gas on CO2 storage capacity.
Raza, A., Hill, K.C., Korsch, R.J., 2009. Mid-Cretaceous uplift and denudation of the Int. J. Greenh. Gas Cont. 11 (0), 158–162.
Bowen and Surat Basins, eastern Australia: relationship to Tasman Sea rifting from Watson, M.N., Zwingmann, N., Lemon, N.M., 2004. The Ladbroke Grove-Katnook carbon
apatite fission-track and vitrinite-reflectance data. Aust. J. Earth Sci. 56 (3), 501–531. dioxide natural laboratory: a recent CO2 accumulation in a lithic sandstone reservoir.
Renard, S., et al., 2011. Geochemical study of the reactivity of a carbonate rock in a Energy 29 (9–10), 1457–1466.
geological storage of CO2 : implications of co-injected gases. Energy Procedia 4, White, A.F., 1995. Chemical weathering rates of silicate minerals in soils. Rev. Mineral.
5364–5369. Geochem. 31 (1), 407–461.
Schaef T., H., et al., 2014. Mineralization of basalts in the CO2-H2O-SO2-O2 system. Wilke, F.D.H., Vásquez, M., Wiersberg, T., Naumann, R., Erzinger, J., 2012. On the in-
Environ. Sci. Technol. 48, 5298–5305. teraction of pure and impure supercritical CO2 with rock forming minerals in saline
Shao, H., Ray, J.R., Jun, Y.-S., 2011. Effects of salinity and the extent of water on su- aquifers: an experimental geochemical approach. Appl. Geochem. 27 (8), 1615–1622.
percritical CO2-induced phlogopite dissolution and secondary mineral formation. Wunsch, A., Navarre-Sitchler, A.K., Moore, J., McCray, J.E., 2014. Metal release from
Environ. Sci. Technol. 45 (4), 1737–1743. limestones at high partial-pressures of CO2. Chem. Geol. 363 (0), 40–55.
Smith, M.M., Wolery, T.J., Carroll, S.A., 2013. Kinetics of chlorite dissolution at elevated Wunsch, A., Navarre-Sitchler, A.K., Moore, J., Ricko, A., McCray, J.E., 2013. Metal release
temperatures and CO2 conditions. Chem. Geol. 347 (0), 1–8. from dolomites at high partial-pressures of CO2. Appl. Geochem. 38 (0), 33–47.
Spiro, C., 2014. Callide Oxyfuel Project: Lessons Learned. Global CCS Institute. Xu, T.F., Apps, J.A., Pruess, K., Yamamoto, H., 2007. Numerical modeling of injection and
Spycher, N., Pruess, K., Ennis-King, J., 2003. CO2-H2O mixtures in the geological se- mineral trapping of CO2 with H2S and SO2 in a sandstone formation. Chem. Geol.
questration of CO2. I. Assessment and calculation of mutual solubilities from 12 to 242 (3–4), 319–346.

19

You might also like