Content uploaded by Florian Osselin
Author content
All content in this area was uploaded by Florian Osselin on Feb 03, 2023
Content may be subject to copyright.
Water-rock interactions and self-remediation: lessons from a
hydraulic fracturing operation in the Vaca Muerta formation,
Argentina
F. Osselin1,*, E.C. Gaucher2,+, P. Baldony-Andrey2, W. Kloppmann4, and B. Mayer3
1Institut des Sciences de la Terre d’Orléans, Université d’Orléans,CNRS,BRGM
UMR7327, 1A Rue de la Ferollerie, 45100 Orléans, France
2TotalEnergies, France
3Applied Geochemistry Group, Department of Geoscience, University of Calgary, 2500
University Dr. NW, Calgary, Alberta, T2N 1N4, Canada
4French Geological Survey (BRGM), Orléans, France
*Corresponding author
+Present address: RWI Group, Institute of Geological Science, University of Bern,
Switzerland
December 21, 2022
1 Abstract1
In order to analyze the effect of a new gelling agent for hydraulic fracturing, fluid samples from different2
stages of the operation (hydraulic fracturing fluid, coil tubing, flowback and produced waters) were col-3
lected from a well in the the Vaca Muerta formation in Argentina. Collected samples were analyzed for4
major and trace elements, first within a few days after sampling, then reanalyzed 6 months later and again5
2 years after sampling. Results show that the salinity of samples increased quickly with time, from 20006
mg/L up to 43,000 mg/l a month later, due to the mixing of hydraulic fracturing fluids with formation7
1
water. No evidence of water-rock reactions was observed. Results from the later analyses showed that the8
composition of the samples evolved with time with a sensible decrease of concentration for most trace ele-9
ments over the course of these two years (e.g. Ba from 137 mg/L to 55 mg/L, Mn from 8mg/L to 5mg/L)10
and heavy metals (e.g. As 100µg/L to 1µf/L, Co 160 µg/L to 1.4 µg/L, Cr from 160 µg/L to 26µg/L). Inter-11
pretation of the results shows that delayed, post-sampling, precipitation of barite in the preserved samples12
is the reason for such a decrease. This opens a very interesting option for mitigation and remediation of13
wastewaters from hydraulic fracturing as natural or even triggered precipitation of barite could involve14
most of the dissolved heavy metals and decrease strongly their concentrations.15
2 Introduction16
The exploitation of hydrocarbons trapped in tight formations is continuing its worldwide expansion with17
countries like USA or Canada producing a large percentage of their hydrocarbon resources through shale18
gas plays like the Marcellus, Eagle Ford, Permian Basin formations in the United-States (, EIA), the Mont-19
ney, Duvernay and Horn River formations in Canada (Board, 2017; of Canada, 2018; of Canadian Academies,20
2014). Numerous other countries are joining the movement, with recent hydraulic fracturing operations21
in Argentina where the giant reservoir of Vaca Muerta was discovered in 2010 (Administration, 2013) is22
exploited.23
Hydraulic fracturing and the exploitation of shale gas can appear as a controversial technology and its24
global expansion is met with criticism from ecological organizations (Brittingham et al., 2014). Indeed,25
concerns remain about the potential contamination of freshwater resources by fugitive methane emana-26
tions (Osborn et al., 2011; Humez et al., 2019) or by hydraulic fracturing fluids migration (Darrah et al.,27
2014; Warner et al., 2012; Bondu et al., 2021). Other environmental risks are linked to the very high water28
consumption (Gallegos and Varela, 2015; Gregory et al., 2011a; Vengosh et al., 2014; Kondash et al., 2018),29
as well as the treatment and recycling of thousands of cubic meters of potentially toxic wastewaters pro-30
duced per well (Thiel et al., 2015; Gregory et al., 2011b). Indeed, after hydraulic fracturing of the reservoir,31
when the well is opened to production, a large quantity of water (called flowback water) is produced (Kon-32
dash et al., 2017). These fluids are a mixture of hydraulic fracturing fluids with formation brine as well33
as products of water-rock interactions, the latter being frequently enhanced by the additives used for the34
optimization of the hydraulic fracturing operation itself (Osselin et al., 2018, 2019; Birkle, 2016; Birkle and35
Makechnie, 2022; Li et al., 2017; Lu et al., 2017; Phan et al., 2020). These flowback fluids are characterized36
2
by a very high TDS (total dissolved solids), usually several times the salinity of seawater and contain usu-37
ally non-negligible quantities of trace elements and heavy metals (e.g. As, Ni, Co, Cr) with a potentially38
high toxicity (Haluszczak et al., 2013; Blauch et al., 2009; Abualfaraj et al., 2014; Phan et al., 2015; Johnson39
and Graney, 2015; Ni et al., 2018; Bern et al., 2021). The disposal and remediation of these high salinity40
fluids is complicated and operators usually choose to reinject these fluids underground in nearby porous41
formations, usually into stratigraphically deep reservoirs, well below aquifers utilized for water supplies.42
However, considering the very high consumption of freshwater by hydraulic fracturing operations (several43
thousand cubic meters per stage), and especially in areas where the freshwater resource is already under44
strain, there is a strong incentive to recycle and reuse these flowback fluids for the next hydraulic fracturing45
operation (Liu et al., 2020). One way to simplify the processing of flowback fluids for reuse in hydraulic46
fracturing operations is to control the downhole water-rock interactions and the release of heavy metal47
and traces by the formation (Osselin et al., 2019; Lerat et al., 2018). An extensive understanding of the48
downhole geochemical behavior during hydraulic fracturing and subsequent flowback is then required in49
order to optimize and tailor the geochemistry of flowback fluids. In particular, it was shown that the use50
of oxidative breakers was not optimal because of its very aggressive behavior towards minerals like pyrite51
and organic matter releasing heavy metals in the flowback fluids (Renock et al., 2016; Yang et al., 2021).52
This study presents the analysis of flowback samples from a well operated in Argentina. This well53
presents the particularity of having been treated with a novel fracturing fluid mixture, not involving the54
classic couple guar gum and oxidant breaker. Instead, the gelling agent (i.e. the additive modifying vis-55
cosity for a better proppant transport into the fractures) was composed of fragile polymers. Upon opening56
the well to production, when a lower viscosity is desirable for smooth flowback (i.e., once the proppant57
has been transported deep into the hydraulic fractures), the decrease in viscosity is not achieved by an ag-58
gressive oxidative attack on the polymer but by a simple mechanical shearing and breaking of the polymer59
due to the rough and intersected nature of the hydraulic fractures network. If the polymer is forced into a60
steep angle because of the geometry of the hydraulic fractures, it will break into smaller pieces. Thus the61
downhole geochemistry is strongly different from other classic hydraulically fractured wells such as from62
the Marcellus or the Montney shales, because of the absence of oxidative action on the sulfide minerals and63
on organic matter.64
Several water samples were collected during the different stages of the procedure, the hydraulic fractur-65
ing itself, the coil tubing operation which opens the different stages to the main wellbore, and finally the66
flowback once the well is opened to production. All samples were analyzed for major elements and traces67
3
as well as several stable isotopes in order to elucidate the downhole behavior and see if the absence of ox-68
idative breaker was impacting beneficially the overall composition of returned waters in view of recycling69
and reuse of such waters for further hydraulic fracturing operations or simply for disposal.70
3 Material and Methods71
Study site and well completion The considered well is located in the province of Neuquen, North-West72
Patagonia, Argentina (Figure 1). The target formation is the Vaca Muerta formation, a carbonate marl with73
black shale and lime mudstone. The targeted interval (between 2600 mbs (meters below surface) and 280074
mbs) presents an average composition of 15% albite feldspar(NaAlSi3O8), 30% quartz (SiO2), 25% calcite75
(CaCO3), 22% illite (clay mineral), with a bit (3%) of pyrite (FeS2) and traces of dolomite and apatite.76
In this zone, several pads were constructed, with each pad being the starting point of three horizon-77
tal wells in a two pronged fork shape. The horizontal portion of the studied well was divided into 1578
stages. After the drilling of the well and the perforations, but before any hydraulic fracturing, the well was79
spearheaded with HCl to cleanse debris and prepare the formation for the hydraulic fracturing operation.80
Hydraulic fracturing was proceeded with a slickwater mixture with freshwater obtained from the overlying81
sandstone aquifer. The water used came from two different wells labeled WW3 and WW4.82
The 15 stages were proceeded from toe (the far end of the horizontal wellbore) to heel, and stages were83
separated from each other by small aluminum plugs. After all stages were fractured, the well was subjected84
to a coil-tubing operation to open the stages to production. This operation consists of inserting a flexible85
drill into the horizontal wellbore and drilling the plugs which opens the stages. During this operation,86
no water is produced and the whole system supposedly works as a closed loop. However, due to some87
overheating of the water and the equipment, it was necessary to add four trucks of freshwater during the88
operation with each truck containing 25m3of water. Before the addition, the same amount was bled from89
the closed loop and replaced with the freshwater from the trucks. One truck was added between the 7th90
and the 8th stage, another between the 13th and 14th stage and 2 trucks between 14th and the last plug for91
a total of 100m3of freshwater (Figure 2).92
Finally, the well was left to rest one week between the end of the coil-tubing operation and the beginning93
of flowback (i.e. opening of the well to production).94
Analytical Methods Samples were obtained on site during the coil-tubing procedure and the flowback95
(Oct-Nov 2015). Samples from the water wells used for the hydraulic fracturing fluids (WW3 and WW4)96
4
Figure 1: Location of the Neuquen reqion
5
were also obtained as well as one produced water sample from another hydraulically fractured well on97
another pad, which was put to production four months before the operation on the considered well. This98
sample, named PW, was considered to be representative of the formation brine. In total, three samples99
were taken during coil-tubing (CT1, CT2 and CT3), one after the first plug was drilled, one between plugs100
7 and 8, and one after the last plug was drilled. Unfortunately, the first CT sample was not kept. A total of101
6 flowback (FB1 to FB6) samples were obtained during the first 2 days of flowback at regular intervals.102
After filtration (0.1µm), the pH of the samples was measured before being preserved on site (acidifica-103
tion pH < 2 for cation analysis) and shipped a few days later to INDUSER (Induser Group SRL, Laboratory104
of Chemical and Microbiological Analysis, Buenos Aires, Argentina) for analysis. Samples were analyzed105
for all major and trace elements by the laboratory as well as numerous organic species. Cations were106
quantified by ICP-OES (inductively coupled plasma optical emission spectrophotometer) and anion con-107
centrations were measured by ion chromatography. Bicarbonate alkalinity was measured by titration. Total108
dissolved solids (TDS) content was calculated by adding all measured concentrations of dissolved species109
and the consistency of the results (QA/QC) was verified by checking that the ionic balance was below 5%110
for all samples.111
Stable isotope analysis were made at BRGM (French Geological Survey, Orléans, France). Water isotopes112
ratios (δ2H and δ18O ) were determined by CF-IRMS (continuous flow isotope ratio mass spectrometry) and113
results are reported in the δnotation δ= (Rspl/ Rstd −1)×1000, with Rspl the isotope ratio in the sample and114
Rstd the isotope ratio of a standard reference. For water isotopes, the standard reference is V-SMOW.115
The exploitation of the results from INDUSER showed that the detection limits from this laboratory116
were too high for the samples and very little could be read from these analyses (see Supplementary In-117
formation Tables 1-4). Additional analyses were then requested for the trace elements with some done118
at TOTALEnergies (Laboratory of Inorganic Chemistry, CSTJF, Pau, France) and some at BRGM (French119
Geological Survey). The TOTALEnergies analyses where proceeded in July 2016 (i.e. 8 to 9 months after120
sampling due to delays on the exportation of the samples). Concentrations of trace elements were measured121
at BRGM on Thermo Scientific XSERIES 2 ICP-MS with a precision generally better than 5%. Analyses were122
performed on April 2018 i.e. more than 2 years after sampling. Finally, radium quantification was made123
by ALGADE by gamma-spectrometry, using Pb214 and Bi214 for Ra226 and Ac228 for Ra228 according to124
the norm NF EN ISO 550 1070. Analyses were proceeded on 3 samples (PW, FB1 and FB6) in August 2016125
i.e. 9 months after sampling.126
6
QA/QC The quality of trace element analyses is highly dependent on the quality of sample preparation on127
site. Samples were routinely filtered with a 0.1 µm pore size syringe filter, following the recommendation128
of Claret et al. (2011) to avoid conventional filters with larger pore sizes 0.2 µm or 0.45 µm. Filtration was129
performed immediately after fluid sampling on the borehole fluid circulation loop.130
A QC check of the analyses was performed by checking with PHREEQC (Appelo and Postma, 2005;131
Parkhurst and Appelo, 2013), the charge balances and the equilibrium of the fluids with calcite. As this132
equilibrium is usually quickly reached in sedimentary systems and as calcite belongs to the mineralogy of133
the considered formation, it demonstrates that pH, alkalinity and calcium concentration are correctly an-134
alyzed and by extension major elements, based on the charge balance. For metals, redundancy of analyses135
in different laboratories was planned as a means of quality control, but the results of this cross-check are136
part of the discussion of the article.137
4 Results138
The behavior of the well can be divided into 2 phases: (i) the hydraulic fracturing and coil-tubing until139
October 8th 2015 and (ii) flowback and production from Oct. 8th 2015 as represented on Figure 2. During140
the first phase, the TDS increases by an order of magnitude, going from ≈1500 mg/L in the hydraulic141
fracturing fluid (average between WW3 and WW4), to 13,978 mg/L for sample CT1 in the midst of coil-142
tubing operations, and nearly doubling to 23,824 mg/L for CT2 at the end of coil-tubing. After the well is143
opened to flowback, the first sample presents a TDS value almost identical to the end of CT (22,999 mg/L).144
TDS increases quickly up to 41,996 mg/L after a few hours of flowback maintain stable for the remaining145
flowback period (i.e. 48 hours).146
Evolution of other major elements is represented on Figure 3 as a function of Cl, as chloride is usually147
considered conservative during the whole duration of the process (Li et al., 2017; Engelder et al., 2014). The148
goal of such a plot is to allow the identification of conservative and non-conservative species. Conservative149
species will correlate linearly to Cl concentrations, while the absence of correlation indicates the presence150
of mineral precipitation/dissolution and cation exchange.151
All major elements are evolving similarly to the TDS content with an increase between hydraulic frac-152
turing fluids and CT1 (e.g. Na from 650 mg/L to 5030 mg/L), a minor increase during CT (Na from 5030153
mg/L to 7910 mg/L), followed by a small drop between CT and FB1 (from 7910 to 7820 mg/L). A fast154
increase upon flowback (from 7820 to 13,550 mg/L) is followed by a period of almost no variation until the155
7
Figure 2: Evolution of TDS as a function of time during coil-tubing and flowback (INDUSER). The right
figure shows a zoom of the flowback period
end of the sampling period. Interestingly, Na, Ca and Mg concentrations plot linearly with Cl for HWW,156
coil tubing and flowback samples, except sample PW which plots either higher for Ca, Mg, and lower for157
Na and K. Finally, the drop between CT2 and FB1 is proportionally more pronounced for K (from 87 mg/L158
down to 56.2 mg/L) than for the other major species respectively to Cl.159
Alkalinity also shows an increase from 450 mg/L for the hydraulic fracturing fluid to 895 mg/L for CT1160
and up to 1480 mg/L for FB3. The PW value for the alkalinity is actually smaller than for the hydraulic161
fracturing fluid with a value of 136 mg/L.162
Barium presents an increase in concentration similar to the major elements but does not seem to plot163
linearly with Cl (Fig. 4). The first coil-tubing sample shows a value almost at zero, when CT2 plots at164
23.9 mg/L and the flowback concentrations are around 135 mg/L. On the opposite, sulfate concentration is165
high for hydraulic fracturing fluids (300 mg/L) and drops quickly during coil-tubing (CT1 at 200 mg/L and166
CT2 at 106 mg/L) before reaching around 35 mg/L during flowback (Fig. 4). There is no value for sulfate167
concentration for FB1. Radium is measured at 1.58 Bq/L in the initial flowback sample and increased to168
13.65 Bq/L in the last flowback sample. The produced water sample presents a Ra activity of 137.3 Bq/L.169
Water isotopes present a behavior similar to the major dissolved species, with a steady evolution from170
-88.8 and -95.3‰ for δ2H values of respectively WW3 and WW4 to -66.5‰ at the end of coil-tubing, then171
-65.0‰ for FB1 and a plateau ratio around -53.5‰ for flowback samples after the first (Fig. 5). Oxygen172
isotope values increase from -11.0 and -12.3‰ in WW3 and WW4 samples up to -5.9‰ for CT2 and -6.0‰173
at the beginning of flowback. The flowback plateau is around -4‰. Finally the produced water sample PW174
is measured at δ2H = -33.0‰ and δ18O = -1.5‰. A cross-plot δ18O versus δ2H shows that WW3 and WW4175
8
Figure 3: Plots of major species against Cl during coil-tubing and flowback (INDUSER)
Figure 4: Plot of Ba and sulfate against Cl during coil-tubing and flowback (INDUSER)
9
- 1 4 - 1 2 - 1 0 - 8 - 6 - 4 - 2 0
- 1 0 0
- 8 0
- 6 0
- 4 0
- 2 0
0 W W
C T
P W
F B
L M W L
δ2HH 2 O (
‰
)
δ18OH 2 O (
‰
)
Figure 5: Cross plot of δ2H vs δ18O and the LMWL (BRGM)
plot close to the Local Meteoric Water Line (LMWL) (Hoke et al., 2013) while subsequent samples depart176
from this line.177
Few samples with heavy metals concentrations beyond the INDUSER detection limit (Nov. 2015)178
present values between 100 and 200 µg/L. The highest values measured for heavy metals are 210 µg/L179
for As in PW, 190 µg/L for Co in FB1, 240 µg/L for Cu in FB1, 200 µg/L for Cr in CT2. Mn presents val-180
ues between 5 and 15 mg/L, B up to 85 mg/L for FB4 and Fe up to 200 mg/L for FB5. However, values181
measured from BRGM (April 2018) differ strongly. For example 0.67 µg/L of As in PW, 1.4 µg/L of Co in182
FB2, 1.7 µg/L of Cu in FB2 and 22.9 µg/L of Cr in CT2. For boron, FB samples drop from around 80 mg/L183
to 50 mg/L, while CT2 increases from 31.1 mg/L to 65.2 mg/L between INDUSER and BRGM results. Mn184
decreases from 7.5-8 mg/L in flowback to 5.5 mg/L and from 5.1 mg/L to 1.668 mg/L in sample CT2 (Fig-185
ure 6). Only three samples were quantified for Fe in BRGM, with sample CT2 presenting the largest drop186
between 97.1 mg/L for INDUSER down to 12 mg/L for BRGM, while PW concentrations decrease from 173187
mg/L to 155 mg/L. The concentrations of Fe in FB3 does not change with 122 and 125 mg/L in INDUSER188
and BRGM analysis respectively.189
Similarly Ba shows a decrease in concentration from INDUSER to TOTAL (July 2016) to BRGM for190
each sample, with drops between 15% of the initial value (PW) and around 60% for FB values (Figure191
6). Strontium presents a similar behavior, to the exception of PW with an increase from 1750 mg/L for192
INDUSER results up to 3002 mg/L in TOTAL analysis towards 2863.2 mg/L.193
10
PW CT2 FB2 FB3
0
220
440
660
880
Concentration (mg/L)
Sample
Induser
Total
BRGM
Barium
PW CT2 FB2 FB3
0
500
1000
1500
2000
2500
3000
Concentration (mg/L)
Sample
Strontium
PW CT2 FB2 FB3
0
5
10
15
Concentration (mg/L)
Sample
Manganese
PW CT2 FB2 FB3
0
20
40
60
80
Concentration (mg/L)
Sample
Boron
+ 9 months
+ 29 months
Figure 6: Evolution of Ba, Sr, Mn and B with time (i.e. according to the different laboratories). Error bars
are significantly smaller than the variation (e.g. 1mg/L Induser and 0.5µg/L BRGM for Ba)
11
5 Discussion194
TDS and water-rock interactions The evolution of salinity in flowback water has been extensively de-195
scribed in numerous publications and is the result of mixing of the hydraulic fracturing fluids with forma-196
tion brine as well as caused by chemical interactions between hydraulic fracturing fluid/formation water197
and reservoir rock (Osselin et al., 2018; Rowan et al., 2015; Hakala et al., 2021; Huang et al., 2020). These198
reactions usually involve salt dissolution if present in the mineralogy (halite, gypsum, anhydrite, calcite,199
dolomite) and potential oxidation of sulfide minerals (pyrite FeS2) and organic matter from fracturing wa-200
ter containing strong oxidants (O2, oxidant breaker) (Osselin et al., 2019; Xu et al., 2018).201
In the studied case, it is necessary to differentiate the behavior during coil-tubing from flowback. After202
hydraulic fracturing, the injected fluid stays in the fractures during the whole coil-tubing operation as there203
is no fluid production. The only exception is the addition of freshwater to cool down the fluid, which was204
preceded by bleeding of the same volume of water from the well. As a result, since the bleeding decreased205
the quantity of water in the fractures, a small proportion of formation water was allowed into the hydraulic206
fractures and mixed with the hydraulic fracturing fluid. This explains the increase of TDS for samples CT1207
and CT2 compared to WW3 and WW4. At the end of CT, the system was left to rest for a week with little208
change in water composition. As no fluid was produced, no formation water was allowed in the system and209
no change of TDS is detectable. This is another confirmation that the source of high TDS in flowback water210
is not only the dissolution of autochthonous salts but mostly the mixing with high TDS formation brine.211
The only remarkable feature between CT2 and FB1 is the small drop of K which is likely due to some cation212
exchange with the clay minerals of the reservoir, e.g. replacement of Ca by K in the exchangeable sites of213
the illite fraction (Essington, 2005).214
Then, as the well is opened and flowback begins, a very fast increase of TDS is observed as the forma-215
tion brine starts mixing with the hydraulic fracturing fluid. This increase is followed by a plateau where216
flowback water salinity stays roughly the same. One of the reasons of this plateau is that almost all the217
easily mobilized formation brine (i.e. mobile formation water close to the hydraulic fractures, probably in218
natural fractures) already mixed with the fracturing fluids between FB1 and FB2. After that point, the mix-219
ing slows down as the flow rate of formation water into the hydraulic fractures is smaller since the system220
is tapping into tighter permeability rocks and less mobile sources of formation brine (Osselin et al., 2018).221
Interestingly, pore water sample PW does not seem to correspond to the end-member for conservative222
mixing. Indeed, water isotopes are considered conservative in this context (Rowan et al., 2015) and PW223
12
does not fall (e.g. excess of +10‰ of δ2H ) on a line joining the hydraulic fracturing fluids, coil tubing224
fluids and flowback (Fig. 5). Since PW is a sample from another well fractured four months before, it225
is possible that it does not correspond to the formation water of the considered well, either because four226
months is potentially not long enough to reach the final composition of the formation water; because of227
some interactions between wells shifting the global conservative mixing; or simply because the formation228
water changes composition with the location of the well. Disregarding PW, it appears that Na, Ca, Mg and229
K are likely to be conservative (to the exception of some cation exchange with the clay minerals), while230
Ba and sulfate are probably not. Water isotopes further support this interpretation as they show a mostly231
conservative mixing behavior during the whole coil-tubing and flowback duration (Fig. 5). The very similar232
values in the water isotope ratios between CT2 and FB1 confirms the absence of mixing and water exchange233
during the whole week between the end of coil-tubing and the beginning of flowback.234
Saturation indexes of minerals have been modeled with PHREEQC v3 with the Pitzer database (Appelo,235
2015) and results are represented in Figure 7. Calcite is at equilibrium for PW and or CT, but shows a slight236
oversaturation for FB samples. This is perfectly consistent with the presence of calcite in the mineralogy of237
the formation. A slight precipitation of calcite may then be expected but the simulation results confirm that238
Na, Ca, Mg and K should be mostly conservative. The non-conservative behavior of Ba and sulfate can, on239
the other hand, be explained by the oversaturation of barite during the whole duration of the operation. PW240
– which can be considered as more or less representative of the formation brine despite not being the actual241
end-member of the mixing in the considered well – is quite Ba-rich (781 mg/L), while the freshwater used242
for the hydraulic fracturing operation is rather sulfate-rich (300 mg/L). Precipitation of barite is then likely,243
especially with the addition of the four 25m3trucks of sulfate-rich freshwater. The behavior of radium,244
even in the absence of the first end-member can be linked to barium behavior: in the first flowback sample,245
Ra and Ba are both very low due to barite precipitation incorporating Ra in the crystal structure (Scheiber246
et al., 2014). Then the Ra concentration increases due to mixing with formation water.247
Once the samples are preserved and acidified, calcite becomes strongly undersaturated and is not ex-248
pected to precipitate. On the other hand, barite (BaSO4) is not impacted by acidification and stays super-249
saturated in all samples.250
Evolution of trace concentrations with time in preserved samples As described in the result section, the251
concentrations of heavy metals and trace elements vary widely between the three laboratories. Rejecting the252
hypothesis that the analyses were erroneous, the reason for this discrepancy is that between the sampling253
13
Figure 7: Evolution of barite and calcite saturation indices in the samples and the effect of acidification
(Calculated from INDUSER data)
and the analyses, the samples were not really quenched by the acidification and evolved. Acidification is254
mainly used for preventing the precipitation of oxi-hydroxide iron complexes (Fe(OH)3) which are known255
to co-precipitate with other metallic cations (Appelo and Postma, 2005). However, as can be seen from the256
Ba evolution (Fig. 6 and from the SI from Figure 7, barite is very likely to have precipitated during the257
delayed analyses as its saturation index is not impacted by acidification. This precipitation plays a similar258
role as the iron oxi-hydroxides as barite co-precipitates with several traces such as Ra, As, Cu, Zn, Ag, Ni,259
Hg, Co, V, Pb, Mn and Cr and other rare-earth elements (Gupta, 1991; Crecelius et al., 2007). Most of the260
elements precipitate as a substitution to the Ba in the barite structure while some (Mn as permanganate261
MnO –
4or Cr as chromate CrO 2 –
4) substitute to sulfate (Tokunaga et al., 2016). This is consistent with262
the analytical results showing an overall decrease in all the mentioned cations as well as decrease of Ba263
with time. Strontium usually forms a solid solution with barite upon precipitation and presents indeed a264
decrease with time for all FB samples. The large increase between INDUSER and BRGM value for PW and265
CT2 is on the other hand not clear. As radium was only measured 9 months after sampling (similar time as266
TOTALEnergies analyses), it is likely that the actual values are higher than the ones measured.267
Boron shows a similar increase with a large increase between INDUSER and BRGM results for CT2.268
One possible explanation is the slow decomposition of some colloidal complexes with time releasing Sr269
and B into solution (Appelo and Postma, 2005) during the 2 years between INDUSER and BRGM analyses.270
However the filtration on 0.10µm filters should not let some colloidal particles through. A damaged fil-271
ter, a mishandling of the samples or simply the difficulty associated with Boron analyses may explain the272
14
discrepancy273
The behavior or trace elements over these two years leads to two important conclusions:274
• Acidification of samples is not enough to stabilize a sample with high sulfate and barium concentra-275
tions. Analyses have to be made as soon as possible on flowback waters (or any high salinity brine276
containing these two elements) in order to avoid any experimental error showing lower trace elements277
concentration than in reality. Similarly, if the behavior of B and Sr is linked to the decomposition of278
colloidal particles, then the filtration has also to be carefully controlled. The current case is quite279
extreme since most of the analyses have been made more than two years after sampling, but not280
completely uncommon.281
• Barium sulfate precipitation is a potential mitigation and remediation process to lower the concen-282
tration of heavy metals and potentially toxic species. Indeed, the concentrations of most trace ele-283
ments have decreased by several orders of magnitude during the two years period, reaching even-284
tually concentrations below the detection limit. This can provide an mitigation approach to recycle285
hydraulic fracturing waters by adding either Ba ions for sulfate rich wastewaters or sulfate in barium286
rich wastewaters (e.g. Marcellus). The precipitation of barium sulfate (barite) could scavenge a large287
amount of trace elements and heavy metals from the returned waters decreasing their concentrations288
below detection limits. Since hydraulic fracturing is using thousands of m3of freshwater and pro-289
ducing large amounts of highly saline wastewater, the recycling of such returned waters is a necessity290
to lower the pressure on water resources and a quick way to reduce the trace elements and heavy met-291
als concentration could help alleviate this issue. This however means that the potential solids would292
have to be treated with proper care as they would be incorporating radioactive species (Ra) and heavy293
metals that were scavenged during precipitation of barite.294
It is important that values for the heavy metals before barite precipitation in our case were still at295
low levels. The highest concentrations were between 100 and 200 µg/L. These values, while being higher296
than the drinking water guidelines (Organisation, 2017), are still orders of magnitude lower than other297
industrial pollution such as mine tailing (for example As average concentrations of 22.53 mg/L in the298
polluted groundwater linked to Chéni mine in France Bodénan et al. (2004)). Finally, the combination299
of barite precipitation with the absence of oxidative breaker and the low initial trace and heavy metals300
concentrations in formation water leads to high TDS, Na-Cl type, returned waters presenting low to very301
low concentrations for all minor species and thus low to very low toxicity except for the hypersalinity.302
15
6 Conclusion303
The presented study was initially designed to analyze the effect of a new gelling agent for hydraulic frac-304
turing operations. This new gelling agent does not require an strong oxidative breaker (such as persulfate),305
which limits the potential water/rock interaction during the hydraulic fracturing and the flowback pe-306
riod. Water samples were taken during the different periods of the process (coil tubing and flowback) and307
were analyzed for major and trace elements. Results showed that, as expected, little water/rock interaction308
occurred except the precipitation of barite.309
However, due to the high detection limits of the first Argentinian laboratory which analyzed the results,310
it was decided to duplicate the analyses both at TotalEnergies and at BRGM. Because of custom delays, these311
additional analyses were performed only 9 months later for some (TotalEnergies) and more than 2 years312
after sampling (BRGM). These extra results show a systematic decrease in the heavy metal content from the313
initial analyses. This decrease is interpreted in terms of post-sampling barite precipitation which scavenges314
heavy metals. This opens an potential mitigation technique for flowback and produced waters to improve315
their quality and go towards their recycling.316
7 Acknowledgment317
G. Lebas, M. Burgos Egido, K. Cailleaud, J. Lerat (TotalEnergies, France), M. Lopez, D. Rosenman and D.318
Gonzales (TotalEnergies Argentina) are warmly thanked for their help and for the concept of the research319
and the sampling of the flowback waters. Two anonymous reviewers are also thanked for their help in320
improving the manuscript. The interpretation of the data was done by F. Osselin and co-authors within321
the framework of the G-baseline project, co-funded by a strategic project grant of the Natural Sciences and322
Engineering Research Council of Canada (NSERC grant №463605), the French Research Agency (ANR-14-323
CE05-0050 grant) and TotalEnergies R& D. This research was undertaken thanks in part to funding from324
the Canada First Research Excellence Fund.325
8 Supplementary Information326
Analytical results for the different samples described in this study. When applicable, results from the 3327
laboratories (INDUSER, TOTAL and BRGM) are mentionned.328
16
Sample Date Time pH TDS Cl Alkalinity Na Ca Mg
DD/MM/YYYY HH:mm - mg/L mg/L mg/L mg/L mg/L mg/L
WW3 09/10/2014 - 9.1 1998 444 656 730 4 0.4
WW4 11/10/2015 - 9.1 1264 125 326 415 6.8 0.11
PW 13/10/2015 15:00 6 115795 71000 136 29657 9980 1360
CT1 19/10/2015 10:00 6.3 13978 7410 895 5030 264 30.7
CT2 31/10/2015 07:30 6.7 23824 12500 - 7910 630 66.4
FB1 08/11/2015 11:40 6.5 22999 12482 - 7820 633 62.9
FB2 08/11/2015 16:00 6.9 41996 24740 1180 13550 1350 183
FB3 09/11/2015 09:30 6.4 41858 25000 1480 13210 1130 134
FB4 09/11/2015 16:00 6.4 41790 24760 1340 13590 1060 127
FB5 10/11/2015 09:30 6.3 43982 25150 1460 14910 1290 153
FB6 10/11/2015 16:00 7 42716 25390 1460 13620 1160 143
Table 1: Selected results from the flowback water samples (Induser) - Major elements (<LQ = below quan-
tification limit)
Sample Date Time K SO4δ2Hδ18O
DD/MM/YYYY HH:mm mg/L mg/L ‰ ‰
WW3 09/10/2014 - 0.9 253 -88.8 -11.0
WW4 11/10/2015 - 1.3 404 -95.3 -12.3
PW 13/10/2015 15:00 330 <LQ -33.0 -1.5
CT1 19/10/2015 10:00 41 200 - -
CT2 31/10/2015 07:30 87 106 -66.5 -5.9
FB1 08/11/2015 11:40 56.2 - -65.0 -6.0
FB2 08/11/2015 16:00 170 32.8 -53.4 -3.6
FB3 09/11/2015 09:30 139 34.9 -53.7 -4.0
FB4 09/11/2015 16:00 132 35.1 -54.9 -4.1
FB5 10/11/2015 09:30 157 32.5 - -
FB6 10/11/2015 16:00 147 34.5 -52.8 -3.9
Table 2: Selected results from the flowback water samples (Induser) - Major elements (<LQ = below quan-
tification limit) - Continued
17
Sample Date Time Ba Sr Mn B Fe
Induser Total BRGM Induser Total BRGM Induser BRGM Induser Total BRGM Induser BRGM
DD/MM/YYYY HH:mm mg/L mg/L mg/L mg/L mg/L
WW3 09/10/2014 - <LQ - 0.1 <LQ - 0.174 < LQ 4.42 (µg/L) 0.7 - 0.646 0.64 -
WW4 11/10/2015 - <LQ 0.18 0.075 <LQ 0.27 0.213 < LQ 9.29 (µg/L) 0.4 0.2 0.189 0.53 -
PW 13/10/2015 15:00 781 734 660 1750 3002 2863.41 13.1 12.489 51.5 43 50.434 173 155
CT1 19/10/2015 10:00 2.9 - - 63.5 - - 4.21 - 23.3 - - 24.2 -
CT2 31/10/2015 07:30 23.9 5.5 4.043 111 145 129.021 5.1 1.668 31.1 30 65.229 97.1 12
FB1 08/11/2015 11:40 26.4 - - 219 - - 10.3 - <LQ - - 307 -
FB2 08/11/2015 16:00 131 - 51.804 421 - 306.326 10.6 0.761 79.7 - 51.467 198 -
FB3 09/11/2015 09:30 137 81 55.237 394 335 287.217 8.12 5.313 83 48 52.038 122 125
FB4 09/11/2015 16:00 130 - 59.898 389 - 305.807 7.51 5.490 84.7 - 55.176 121 -
FB5 10/11/2015 09:30 135 - 69.464 382 - 306.514 7.61 5.439 81.2 - 53.577 199 -
FB6 10/11/2015 16:00 145 - 70.472 366 - 316.825 8.08 5.539 - - 55.589 130 -
Quantification limit 1 0.0005 0.0005 1 0.001 0.001 0.05 0.001 0.2 0.005 0.005 0.2 0.001
Table 3: Selected results from the flowback water samples and evolution between INDUSER and BRGM -
Traces elements (<LQ = below quantification limit; - not measured)
Sample Date Time As Co Cu Cr Ra
Induser BRGM Induser BRGM Induser BRGM Induser BRGM Algade
DD/MM/YYYY HH:mm µg/L µg/L µg/L µg/L Bq/L
WW3 09/10/2014 - <LQ 1.61 <LQ <LQ <LQ 1.64 <LQ 1.64 -
WW4 11/10/2015 - 120 12.28 <LQ <LQ <LQ - <LQ <LQ -
PW 13/10/2015 15:00 210 0.67 <LQ 0.51 <LQ 1.81 <LQ 30.3 137.3
CT1 19/10/2015 10:00 <LQ - <LQ - <LQ - <LQ - -
CT2 31/10/2015 07:30 <LQ 0.83 <LQ 1.13 180 <LQ 200 22.9 -
FB1 08/11/2015 11:40 <LQ - 190 - 240 - 190 - 1.58
FB2 08/11/2015 16:00 <LQ <LQ 160 1.4 <LQ 1.7 160 25.7 -
FB3 09/11/2015 09:30 100 1.03 <LQ 0.6 <LQ 2.37 <LQ 73.3 -
FB4 09/11/2015 16:00 <LQ 0.96 <LQ 0.51 <LQ 2.49 <LQ 75.5 -
FB5 10/11/2015 09:30 <LQ 4.43 <LQ <LQ <LQ 2.59 <LQ 85.8 -
FB6 10/11/2015 16:00 <LQ 1.02 <LQ <LQ <LQ 2.57 <LQ 86.0 13.65
Quantification limit 0.10 0.10 0.10 0.10
Table 4: Selected results from the flowback water samples and evolution between INDUSER and BRGM -
Traces elements (<LQ = below quantification limit; - not measured) - Continued
18
References329
Abualfaraj, N., Gurian, P. L., and Olson, M. S. (2014). Characterization of marcellus shale flowback water.330
Environmental Engineering Science, 31:140716083132007.331
Administration, U. E. (2013). Technically recoverable shale oil and gas resources: An assessment of 137332
shale formations in 41 countries outside the united state.333
Appelo, C. and Postma, D. (2005). Geochemistry, Groundwater And Pollution, Second Edition. Taylor &334
Francis.335
Appelo, C. A. (2015). Principles, caveats and improvements in databases for calculating hydrogeochemical336
reactions in saline waters from 0 to 200°c and 1 to 1000atm. Applied Geochemistry, 55:62–71.337
Bern, C. R., Birdwell, J. E., and Jubb, A. M. (2021). Water–rock interaction and the concentrations of338
major, trace, and rare earth elements in hydrocarbon-associated produced waters of the united states.339
Environmental Science: Processes & Impacts, 23:1198–1219.340
Birkle, P. (2016). Recovery rates of fracturing fluids and provenance of produced water from hydraulic341
fracturing of silurian qusaiba hot shale, northern saudi arabia, with implications on fracture network.342
AAPG Bulletin, 100:917–941.343
Birkle, P. and Makechnie, G. K. (2022). Geochemical cycle of hydraulic fracturing fluids: Implications344
for fracture functionality and frac job efficiency in tight sandstone. Journal of Petroleum Science and345
Engineering, 208.346
Blauch, M. E., Myers, R. R., Moore, T. R., and Lipinski, B. A. (2009). Marcellus shale post-frac flowback347
waters – where is all the salt coming from and what are the implications? SPE 125740, SPE Regional348
Eastern Meeting, pages 1–20.349
Board, N. E. (2017). Canada’s energy future 2017 supplement: Natural gas production.350
Bodénan, F., Baranger, P., Piantone, P., Lassin, A., Azaroual, M., Gaucher, E., and Braibant, G. (2004).351
Arsenic behaviour in gold-ore mill tailings, massif central, france: hydrogeochemical study and investi-352
gation of in situ redox signatures. Applied Geochemistry, 19:1785–1800.353
Bondu, R., Kloppmann, W., Naumenko-Dèzes, M. O., Humez, P., and Mayer, B. (2021). Potential impacts354
19
of shale gas development on inorganic groundwater chemistry: Implications for environmental baseline355
assessment in shallow aquifers. Environmental Science and Technology, 55:9657–9671.356
Brittingham, M. C., Maloney, K. O., Farag, A. M., Harper, D. D., and Bowen, Z. H. (2014). Ecological risks357
of shale oil and gas development to wildlife, aquatic resources and their habitats. Environmental Science358
& Technology, 48:11034–11047.359
Claret, F., Tournassat, C., Crouzet, C., Gaucher, E. C., Schäfer, T., Braibant, G., and Guyonnet, D. (2011).360
Metal speciation in landfill leachates with a focus on the influence of organic matter. Waste Management,361
31:2036–2045.362
Crecelius, E., Trefry, J., McKinley, J., Lasorsa, B., and Trocine, R. (2007). Study of barite solubility and the363
release of trace components to the marine environment.364
Darrah, T. H., Vengosh, A., Jackson, R. B., Warner, N. R., and Poreda, R. J. (2014). Noble gases identify the365
mechanisms of fugitive gas contamination in drinking-water wells overlying the marcellus and barnett366
shales. Proc Natl Acad Sci U S A, 111:14076–14081.367
(EIA), E. I. A. (2019). Annual energy outlook 2019 with projections to 2050.368
Engelder, T., Cathles, L. M., and Bryndzia, L. T. (2014). The fate of residual treatment water in gas shale.369
Journal of Unconventional Oil and Gas Resources, 7:33–48.370
Essington, M. E. (2005). Soil and Water chemistry: an integrative approach. CRC Press.371
Gallegos, T. J. and Varela, B. A. (2015). Trends in hydraulic fracturing distributions and treatment fluids,372
additives, proppants, and water volumes applied to wells drilled in the united states from 1947 through373
2010—data analysis and comparison to the literature.374
Gregory, K. B., Vidic, R. D., and Dzombak, D. A. (2011a). Water management challenges associated with375
the production of shale gas by hydraulic fracturing. Elements, 7:181–186.376
Gregory, K. B., Vidic, R. D., and Dzombak, D. A. (2011b). Water management challenges associated with377
the production of shale gas by hydraulic fracturing. Elements, 7:181–186.378
Gupta, J. (1991). Determination of barium, strontium and nine minor and trace elements in impure barite379
and strontianite by inductively-coupled plasma atomic-emission spectrometry after dissolution in dis-380
odium ethylenediaminetetraacetate. Talanta, 38:1083–1087.381
20
Hakala, J. A., Vankeuren, A. N. P., Scheuermann, P. P., Lopano, C., and Guthrie, G. D. (2021). Predicting the382
potential for mineral scale precipitation in unconventional reservoirs due to fluid-rock and fluid mixing383
geochemical reactions. Fuel, 284:118883.384
Haluszczak, L. O., Rose, A. W., and Kump, L. R. (2013). Geochemical evaluation of flowback brine from385
marcellus gas wells in pennsylvania, usa. Applied Geochemistry, 28:55–61.386
Hoke, G. D., Aranibar, J. N., Viale, M., Araneo, D. C., and Llano, C. (2013). Seasonal moisture sources387
and the isotopic composition of precipitation, rivers, and carbonates across the andes at 32.5-35.5°s.388
Geochemistry, Geophysics, Geosystems, 14:962–978.389
Huang, R., Sun, W., Ding, X., Zhao, Y., and Song, M. (2020). Effect of pressure on the kinetics of peridotite390
serpentinization. Physics and Chemistry of Minerals, 47.391
Humez, P., Osselin, F., Kloppmann, W., and Mayer, B. (2019). A geochemical and multi-isotope model-392
ing approach to determine sources and fate of methane in shallow groundwater above unconventional393
hydrocarbon reservoirs. Journal of Contaminant Hydrology, 226:103525.394
Johnson, J. D. and Graney, J. R. (2015). Fingerprinting marcellus shale waste products from pb isotope and395
trace metal perspectives. Applied Geochemistry, 60:104–115.396
Kondash, A., Albright, E., and Vengosh, A. (2017). Quantity of flowback and produced waters from uncon-397
ventional oil and gas exploration. Science of the Total Environment, 574:314–321.398
Kondash, A. J., Lauer, N. E., and Vengosh, A. (2018). The intensification of the water footprint of hydraulic399
fracturing. Science Advances, 4.400
Lerat, J. G., Sterpenich, J., Mosser-Ruck, R., Lorgeoux, C., Bihannic, I., Fialips, C. I., Schovsbo, N. H.,401
Pironon, J., and Éric C. Gaucher (2018). Metals and radionuclides (mar) in the alum shale of denmark:402
Identification of mar-bearing phases for the better management of hydraulic fracturing waters. Journal403
of Natural Gas Science and Engineering, 53:139–152.404
Li, Y., Huang, T., Pang, Z., and Jin, C. (2017). Geochemical processes during hydraulic fracturing: a water-405
rock interaction experiment and field test study. Geosciences Journal, 21:753–763.406
Liu, D., Li, J., Zou, C., Cui, H., Ni, Y., Liu, J., Wu, W., Zhang, L., Coyte, R., Kondash, A., and Vengosh, A.407
(2020). Recycling flowback water for hydraulic fracturing in sichuan basin, china: Implications for gas408
production, water footprint, and water quality of regenerated flowback water. Fuel, 272:117621.409
21
Lu, J., Mickler, P. J., Nicot, J.-P., Choi, W., Esch, W. L., and Darvari, R. (2017). Geochemical interactions of410
shale and brine in autoclave experiments—understanding mineral reactions during hydraulic fracturing411
of marcellus and eagle ford shales. AAPG Bulletin, 101:1567–1597.412
Ni, Y., Zou, C., Cui, H., Li, J., Lauer, N. E., Harkness, J. S., Kondash, A. J., Coyte, R. M., Dwyer, G. S., Liu, D.,413
Dong, D., Liao, F., and Vengosh, A. (2018). Origin of flowback and produced waters from sichuan basin,414
china. Environmental Science and Technology, 52:14519–14527.415
of Canada, N. E. B. (2018). Neb - canada’s energy future 2018: Energy supply and demand projections to416
2040 - publication information and downloads.417
of Canadian Academies, C. (2014). Environmental Impacts of Shale Gas Extraction in Canada.418
Organisation, W. H. (2017). Guidelines for drinking-water quality: fourth edition incorporating the first419
addendum.420
Osborn, S. G., Vengosh, A., Warner, N. R., and Jackson, R. B. (2011). Methane contamination of drinking421
water accompanying gas-well drilling and hydraulic fracturing. Proceedings of the National Academy of422
Sciences of the United States of America, 108:8172–6.423
Osselin, F., Nightingale, M., Hearn, G., Kloppmann, W., Gaucher, E., Clarkson, C. R., and Mayer, B. (2018).424
Quantifying the extent of flowback of hydraulic fracturing fluids using chemical and isotopic tracer425
approaches. Applied Geochemistry, 93:20–29.426
Osselin, F., Saad, S., Nightingale, M., Hearn, G., Desaulty, A.-M., Gaucher, E., Clarkson, C., Kloppmann,427
W., and Mayer, B. (2019). Geochemical and sulfate isotopic evolution of flowback and produced waters428
reveals water-rock interactions following hydraulic fracturing of a tight hydrocarbon reservoir. Science429
of The Total Environment, 687:1389–1400.430
Parkhurst, D. L. and Appelo, C. (2013). Description of input and examples for phreeqc version 3 — a431
computer program for speciation , batch-reaction , one-dimensional transport , and inverse geochemical432
calculations. u.s. geological survey techniques and methods, book 6, chapter a43, 497 p. U.S. Geological433
Survey Techniques and Methods, book 6, chapter A43, pages 6–43A.434
Phan, T. T., Capo, R. C., Stewart, B. W., Graney, J. R., Johnson, J. D., Sharma, S., and Toro, J. (2015). Trace435
metal distribution and mobility in drill cuttings and produced waters from marcellus shale gas extrac-436
tion: Uranium, arsenic, barium. Applied Geochemistry, 60:89–103.437
22
Phan, T. T., Hakala, J. A., and Sharma, S. (2020). Application of isotopic and geochemical signals in un-438
conventional oil and gas reservoir produced waters toward characterizing in situ geochemical fluid-shale439
reactions. Science of the Total Environment, 714:136867.440
Renock, D., Landis, J. D., and Sharma, M. (2016). Reductive weathering of black shale and release of barium441
during hydraulic fracturing. Applied Geochemistry, 65:73–86.442
Rowan, E. L., Engle, M. A., Kraemer, T. F., Schroeder, K. T., Hammack, R. W., and Doughten, M. W. (2015).443
Geochemical and isotopic evolution of water produced from middle devonian marcellus shale gas wells,444
appalachian basin, pennsylvania. AAPG Bulletin, 99:181–206.445
Scheiber, J., Seibt, A., Birner, J., Cuenot, N., Genter, A., and Moeckes, W. (2014). Barite scale control at the446
soultz-sous-forêts ( france ) egs site.447
Thiel, G. P., Tow, E. W., Banchik, L. D., Chung, H. W., and Lienhard, J. H. (2015). Energy consumption in448
desalinating produced water from shale oil and gas extraction. Desalination, 366:94–112.449
Tokunaga, K., Uruga, T., Nitta, K., Terada, Y., Sekizawa, O., Kawagucci, S., and Takahashi, Y. (2016). Ap-450
plication of arsenic in barite as a redox indicator for suboxic/anoxic redox condition. Chemical Geology,451
447:59–69.452
Vengosh, A., Jackson, R. B., Warner, N., Darrah, T. H., and Kondash, A. (2014). A critical review of the risks453
to water resources from unconventional shale gas development and hydraulic fracturing in the united454
states. Environmental Science & Technology, 48:8334–8348.455
Warner, N. R., Jackson, R. B., Darrah, T. H., Osborn, S. G., a. Down, Zhao, K., a. White, and a. Vengosh456
(2012). From the cover: Geochemical evidence for possible natural migration of marcellus formation457
brine to shallow aquifers in pennsylvania. Proceedings of the National Academy of Sciences, 109:11961–458
11966.459
Xu, M., Binazadeh, M., Zolfaghari, A., and Dehghanpour, H. (2018). Effects of dissolved oxygen on water460
imbibition in gas shales. Energy and Fuels, 32:4695–4704.461
Yang, S., Liu, D., Yang, Z., Wang, C., Chen, X., Li, H., Li, Q., Yang, B., and Li, Y. (2021). Occurrence462
and mobility of trace elements during oxidant stimulation of shales in yichang, hubei province of china.463
Applied Geochemistry, 127:104913.464
23