Sensitivity Assessment of Boron Isotope as Indicator of Contaminated Groundwater for Hydraulic Fracturing Flowback Fluids Produced from the Dameigou Shale of the Northern Qaidam Basin

Abstract

Hydrogeochemical processes occurring in contaminated groundwater and aquifer systems may reduce the sensitivity of boron isotopes as an indicator of hydraulic fracturing flowback fluids (HFFF) in groundwater. In this paper, based on the Chaiye-1 well (the first continental shale gas well in the northern Qaidam Basin), the hydrogeochemical processes affecting boron isotopes were analyzed in HFFF-contaminated Neogene (NG1 and NG2) and Quaternary (QG1) groundwater around the shale gas field. Then, a model for boron isotopes in HFFF-contaminated groundwater was constructed to assess the sensitivity of boron isotopes as an HFFF indicator. The results show that, limited by the range of pH values and saturation indices (SI) in HFFF-contaminated groundwater, the dissolution of alkali feldspar and precipitation of carbonate have little effect on the boron isotopes in shallow groundwater. For the NG2 aquifer system containing clay minerals, the δ11B of simulated contaminated groundwater (40.0–55.6‰) is always higher than that of the corresponding groundwater mixed conservatively (−6.4–55.6‰) due to preferential adsorption of boron isotopes onto clay minerals, indicating preferential adsorption would reduce the sensitivity of boron isotopes as an indicator of groundwater contamination from HFFF. For the scenario of HFFF contamination, when the mixing ratio of HFFF in contaminated groundwater increases by 5%, boron isotopes in Neogene (NG1 and NG2) and Quaternary (QG1) groundwater have detectable responses to HFFF contamination, suggesting δ11B is a sensitive indicator of HFFF contamination in shallow groundwater from the Dameigou Shale in the northern Qaidam Basin.

1. Introduction

Shale gas has attracted global attention due to its high energy efficiency and cleanliness as a result of the successful application of high-volume hydraulic fracturing (HVHF) [1,2]. Hydraulic fracturing flowback fluids (HFFF), which are produced by ubiquitous HVHF, contain human-made additives and naturally occurring chemicals at toxic concentrations [3,4,5]. The toxic substances in HFFF may contaminate the shallow groundwater under the following contingencies: (1) leakage from a well pipe through a faulty well casing such as a corroded and poorly joined casing during flowback [6]; and (2) accidental release during storage, transport, and disposal of HFFF after flowback [7]. Consequently, potential environmental impacts on groundwater by these toxic substances have raised public concerns [8,9,10,11,12], and forensic identification of HFFF contamination in groundwater has become a research hotspot [13,14,15,16,17,18,19,20].

Multiple studies have focused on the isotopic fingerprint characteristics of HFFF and proposed that δ²H-H₂O, δ¹⁸O-H₂O, δ³⁴S-SO₄, δ¹⁸O-SO₄, ⁸⁷Sr/⁸⁶Sr, and ²²⁸Ra/²²⁶Ra can be tracers for HFFF [18,21,22]. Among them, based on the mechanisms of hydrogeochemical formation of HFFF, the boron isotope ratio (δ¹¹B) was considered to provide a novel diagnostic signature for characterizing HFFF and distinguishing HFFF from natural water [7,18,22]. Warner et al. [23] stated that δ¹¹B values for HFFF from the Marcellus and Fayetteville Formations (marine shales) fall within a narrow range (25 to 31‰) with high B/Cl ratios (>0.1 × 10⁻¹), and were distinct from major river systems (δ¹¹B = 3 to 14‰, B/Cl = 0.2 × 10⁻⁴–0.1 × 10⁻¹) and shallow groundwater (δ¹¹B = 34 to 54‰, B/Cl < 0.3 × 10⁻⁴) in shale gas fields. Ni et al. [24] reported that the HFFF from the Permian Longmaxi Formation in the Sichuan Basin (δ¹¹B = 22.5 to 31.6‰) had similar δ¹¹B values to those of HFFF from the Marcellus Formation and were different from shallow groundwater (δ¹¹B = −7.8 to 6.4‰) [25]. Additionally, Cui et al. [16] proposed that extensive dissolution of feldspars occurring during hydraulic fracturing can impart a unique δ¹¹B fingerprint from continental shale (−30.1 to 10.2‰) to HFFF. Thus, δ¹¹B can be reliably used in North America and Asia to identify the release of HFFF to rapidly circulating shallow groundwater that is mainly recharged by rainwater (δ¹¹B: −1.5 to 34.7‰). These studies have demonstrated that δ¹¹B would be a useful indicator of HFFF contamination in shallow groundwater; however, few studies have focused on hydrogeochemical processes that may cause boron isotope fractionation and the mixing of exogenous boron when HFFF enter the aquifer. The disruption of the original water chemical balance in aquifer systems resulting from HFFF contamination can lead to a series of water–rock interactions [26,27]. The adsorption of boron onto clay [28,29] and precipitation of boron during neoformation of secondary phases [30] can affect the equilibrium exchange of B(OH)₃ and B(OH)₄⁻ in aquifers and change the relative content of boron in different structural phases, resulting in fractionation of boron isotopes [31,32]. Dissolution can also release boron from the mineral lattice and change the δ¹¹B in contaminated groundwater [27,33]. Thus, these processes may reduce the sensitivity of δ¹¹B as an HFFF indicator in groundwater and may even negate δ¹¹B as an HFFF indicator entirely.

The aim of this study is to assess the sensitivity of δ¹¹B as an HFFF indicator through a better understanding of the effects of hydrogeochemical processes on boron isotope evolution in HFFF-contaminated shallow groundwater. Our previous study reported that δ¹¹B of HFFF produced from the Dameigou Shale in the northern Qaidam Basin (−10.2 to −6.4‰) was lower than and distinct from that of the shallow groundwater (24.4 to 55.7‰) [16,34]. In this study, we focus on variations in boron concentrations and isotopes affected by hydrogeochemical processes in HFFF-contaminated shallow aquifers using a quantitative hydrogeochemical model. The results of this study provide a methodology for using boron isotopes to identify groundwater contaminated by HFFF produced from the Dameigou Shale in the northern Qaidam Basin.

2. Materials and Methods

2.1. Study Area

The study area is located within the Yuka Sag in front of the Qilian Mountains in the North Qaidam Margin block fault zone (Figure 1). The study area has a typical continental plateau climate characterized by drought. The average annual precipitation is only 89.4 mm, while the average annual evaporation is as high as 2167.1 mm, more than 24 times the precipitation. The Yuka Depression, which is surrounded by mountains on three sides and faces the river on one side, is relatively closed. The main landform types in the area are structurally denuded low mountains, river terraces, riverbeds, and other fluvial landforms. The shallow aquifer in the area is mainly composed of Quaternary Upper Pleistocene gravel and Neogene siltstone. Groundwater flow directions in the study area are controlled by topography and are generally consistent with the flow directions of rivers. The Chaiye-1 well (CY1), the first continental shale gas well in northwest China, is located in the river terrace of the Naoer River. Furthermore, the Dameigou Shale is the target stratum of the CY1 well [33]. The CY1 well, with a depth of 2250 m, was fractured in 2014. In addition, the Dameigou Shale (depth of 2000 m to 2045 m) is the target stratum of the fracturing [35]. The main components of fracturing fluid are slick water, gel liquid, and hydrochloric acid. The volume of fracturing fluid utilized for the CY1 vertical well was 890 m³ for the purpose of shale gas reservoir parameter evaluation, and the cumulative volume of flowback fluid was 765 m³ after this fracturing.

2.2. Sample Collection

Two types of water were collected in the study area: (1) five samples of HFFF produced from CY1 (CY1-1d to CY1-7d) were collected as a time series from the first to the last day of the artesian flowback process, and (2) two samples of shallow groundwater (NG1 and NG2) and one sample of spring water (QG1) from the Neogene and Quaternary aquifers, respectively, were collected before hydraulic fracturing in order to characterize the background hydrochemical compositions of groundwater. The HFFF sampling days during the flowback process are shown in

Table 1. Hydrochemical and isotopic results of the CY1 HFFF and shallow groundwater before fracturing.

Sample ID	Description	Na	K	Ca	Mg	Cl	HCO3	SO4	TDS	B	δ11B (‰)	pH	T (°C)
		mg/L
CY1-1d	HFFF, flowback day 1	1683	68	375	239	3486	415	412	6413	1.26	−9.6	6.33	29.1
CY1-2d	HFFF, flowback day 2	2815	77	410	170	4783	1368	90	8973	1.04	−10.2	7.00	38.5
CY1-3d	HFFF, flowback day 3	2897	80	359	162	4662	1602	89	8990	1.01	−7.1	7.02	40.1
CY1-5d	HFFF, flowback day 5	3263	83	316	129	5042	1688	41	9658	0.92	−8.3	7.73	43.6
CY1-7d	HFFF, flowback day 7	3467	89	294	113	5228	1764	46	10,056	0.86	−6.4	7.89	46.1
NG1	Groundwater, Neogene aquifer 1	3209	13	733	361	6014	33	1328	11,675	2.31	26.1	7.23	14.3
NG2	Groundwater, Neogene aquifer 2	4044	34	723	568	6555	100	3113	15,088	2.65	55.7	7.29	14.8
QG1	Groundwater, Quaternary aquifer	1796	66	224	329	2017	370	2052	6791	4.41	24.4	8.08	14.6

. NG1 and NG2 represent the groundwater from the two major shallow Neogene aquifers in the study area. QG1 represents the groundwater from the major shallow Quaternary aquifer around the CY1 shale gas field. The depths of the NG1, NG2, and QG1 aquifers are 145.0 to 150.5 m, 96.2 to 105.0 m, and 0 to 14.5 m, respectively.

In the field, water samples were filtered through a 0.45 μm nylon filter and then collected into sterilized HDPE bottles. Samples for trace element analyses were acidified to pH 2 with ultrapure nitric acid (HNO₃). The collected water samples were stored on ice while in the field and refrigerated in the laboratory at 4 °C until the analyses were completed.

2.3. Sample Analysis

Major anions were measured using ion chromatography (IC) via a Thermo Scientific Dionex ICS-4000 (precision ± 1%), except for HCO₃⁻. The HCO₃⁻ concentration was determined by phenolphthalein titration. Major cations and minor elements were measured using inductively coupled plasma optical emission spectroscopy (ICP-OES) via a PerkinElmer Model Optima 8300 (precision ± 1%). Boron isotopes were analyzed using inductively coupled plasma mass spectrometry (ICP-MS) via a PerkinElmer ELAN DCR-e. The average ¹¹B/¹⁰B ratio of NIST SRM 951 during this study was 4.0436 ± 0.0016 (n = 25). The major ions and minor elements were measured at the National Research Center for Geoanalysis, Chinese Academy of Geological Sciences. Boron isotopes were analyzed at the Center of Analysis, Beijing Research Institute of Uranium Geology.

3. Results and Discussion

3.1. Characteristics of Hydrochemical Compositions and Boron Isotope in HFFF and Shallow Groundwater

The hydrochemical and isotopic results of the HFFF and shallow groundwater analyses are shown in Table 1. HFFF are characterized by high total dissolved solids (TDS) (6.4 to 10.1 g/L), are weakly alkaline (6.3 to 7.89), and are dominant in Cl^- and Na⁺. Similarly, the shallow groundwater samples are also characterized by high TDS (6.8 to 15.1 g/L), are weakly alkaline (7.23 to 8.08), and are dominant in Cl⁻ (or Cl⁻ and SO₄²⁻) and Na⁺. Because the TDS of HFFF produced from Dameigou continental shale are significantly lower than those of HFFF produced from marine shales such as the Marcellus Formation in the United States (14.8 to 211.4 g/L) [36], the Bowland Formation in the United Kingdom (62.6 to 99.5 g/L) [37], the Longmaxi Formation in China (13.1 to 53.5 g/L) [38], and Polish shale formations (103.2 g/L) [39], and because the dominating ions of shallow groundwater are similar to those of HFFF in this study, the conventional tracers of HFFF such as TDS, Cl, and Na/Cl are not applicable to identifying groundwater contamination by HFFF in the study area. The SO₄²⁻ concentrations of the Dameigou Shale HFFF, except for the first flowback day (41 to 90 mg/L), are slightly higher than those reported for the Marcellus Formation (1.3 to 11.7 mg/L) [40] and Longmaxi Formation (23 to 89 mg/L) [41], but are lower than those reported for the Bowland Formation (101.8 to 120.1 mg/L) [37]. The SO₄²⁻ concentrations of shallow groundwater in the study area (1328 to 3113 mg/L) were much higher than those of the Dameigou Shale HFFF (41 to 412 mg/L), while the HCO₃⁻ concentrations of shallow groundwater (33 to 370 mg/L) were much lower than those of the Dameigou Shale HFFF (415 to 1764 mg/L). These results reflect that the saturation/unsaturation states of the main mineral components in HFFF and shallow groundwater are different from each other. Using pH values, temperature, and concentrations of major ions, the saturation index (SI) values of HFFF and shallow groundwater were calculated (

Table 2. The saturation index (SI) values of carbonates, sulfates, and chloride-type salts in the CY1 HFFF and shallow groundwater.

Sample ID	Calcite (CaCO3)	Aragonite (CaCO3)	Gypsum (CaSO4·2H2O)	Anhydrite (CaSO4)	Halite (NaCl)	Sylvite (KCl)
CY1−1d	−0.17	−0.32	−1.01	−1.26	−3.97	−4.95
CY1−2d	1.15	1.01	−1.72	−1.87	−3.64	−4.83
CY1−3d	0.98	0.85	−2.75	−2.89	−3.63	−4.83
CY1−5d	1.74	1.61	−2.17	−2.28	−3.56	−4.80
CY1−7d	1.90	1.77	−2.06	−2.14	−3.52	−4.77
NG1	−0.57	−0.72	−0.31	−0.76	−3.47	−5.39
NG2	−0.12	−0.27	−0.06	−0.50	−3.35	−4.96
QG1	0.84	0.69	−0.50	−0.95	−4.16	−5.11

). The SI values of gypsum, anhydrite, halite, and sylvite for the HFFF (−2.75 to −1.01, −2.89 to −1.26, −3.97 to −3.52, and −4.95 to −4.77, respectively) and shallow groundwater (−0.06 to −0.50, −0.95 to −0.50, −4.16 to −3.35, and −5.39 to −4.96, respectively) were less than 0, indicating that these minerals were in the unsaturated state in both the HFFF and shallow groundwater. Considering the pH and temperature changes in the contaminated groundwater, the listed minerals are always in the unsaturated state in contaminated groundwater, even at different degrees of contamination.

The boron concentrations of the HFFF (0.86 to 1.26 mg/L) decreased with the flowback day and were slightly lower than those of shallow groundwater (2.31 to 4.41 mg/L). Meanwhile, the δ¹¹B of the HFFF (−10.2 to −6.4‰) increased with flowback day and are significantly lower than those of shallow groundwater (24.4 to 55.7‰) and those of HFFF produced from marine shales such as the Marcellus Formation (25 to 31‰) [23] and Longmaxi Formation (22.5 to 31.6‰) [24]. These indicate that the δ¹¹B of groundwater is expected to have a significant response to continental HFFF contamination in the study area. The boron isotope characteristics of the HFFF were mainly evolved by the mixing of a shale formation water in which dissolution of non-marine borates (−30.1 to 10.2‰ with mean value of 4‰) has occurred during the formation of the Dameigou Shale, and the water–rock interactions were dominated by the dissolution of boron-bearing alkali feldspars during hydraulic fracturing and subsequent flowback periods [42]. Conversely, the boron isotope characteristics of the shallow groundwater were mainly imparted by the local rainwater (δ¹¹B = 16.7‰), which is the major source of groundwater recharge, and groundwater in specific aquifers (NG2) was further altered by clay adsorption [43,44]. Based on the different formation mechanisms and distinctive values of δ¹¹B in the HFFF and shallow groundwater, boron isotopes can be used as tracers for HFFF contamination in shallow groundwater.

3.2. Hydrogeochemical Processes Affecting Boron Isotopes in HFFF-Contaminated Aquifers

Hydrogeochemical processes, which would occur in HFFF-contaminated aquifer systems induced by disruption of the original water chemical balance, may be the main factors significantly affecting boron isotopes in contaminated groundwater, in addition to conservative mixing of groundwater and HFFF. These hydrogeochemical processes can affect the equilibrium and cause dynamic fractionation of boron isotopes in aquifer systems, and also influence boron isotopes largely through the dissolution of boron-containing aquifer media. Because boron has one main oxidation state (+3, i.e., B³⁺) and does not exist in natural water in the gaseous phase, redox reactions, microbial metabolism, and physical actions such as evaporation and volatilization do not result in boron isotope fractionation [45]. Therefore, the following hydrogeochemical processes, other than the conservative mixing of HFFF and groundwater, that may further modify δ¹¹B of contaminated groundwater were considered in this study:

3.2.1. Dissolution of Alkali Feldspars

During the mineralization process, B³⁺ can be trapped in the alkali feldspars (albite and K-feldspar) by substitution with Si⁴⁺ or Al³⁺, thus forming boron silicate compounds [46]. Because albite and K-feldspar are present in both the Neogene and Quaternary aquifers in the study area, it is necessary to consider the boron from incongruent dissolution of alkali feldspars in contaminated groundwater [33]. The conservative mixing model for groundwater and HFFF hydrogeochemical compositions was established using the Mixing Module of PHREEQC (Version 3.4.0, USGS, Reston, VA, USA) in order to assess the stability of alkali feldspars in the contaminated aquifer system. The CY1-7d HFFF sample, which has smaller differences in δ¹¹B from the background groundwater, was selected as the endmember of the contamination source.

Based on thermodynamic reactions, the contaminated groundwater is in the stability field of K-feldspar only when the HFFF flux reaches more than approximately 30% of groundwater (NG1, NG2, and QG1), as shown in the mineral stability diagram for K-feldspar (Figure 2a–c). Also, the contaminated groundwater is always in the instability field of albite, regardless of the flux ratio of HFFF in groundwater (Figure 2d–f). These data suggest that alkali feldspar dissolution would occur in both the Neogene and Quaternary aquifers in the study area. On the basis of chemical kinetics, Chou and Wollast [47] reported that the steady-state dissolution rates of albite and K-feldspar in hydrochloric acid at 25 °C and pH 6 to 8 are about 0.1 × 10⁻¹² mol/m²/s and 0.5 × 10⁻¹⁴ mol/m²/s, respectively. Moreover, Lasaga [48] stated that the dissolution half-life for a 1 mm crystal of K-feldspar is 5.2 × 10⁵ y at 25 °C and pH 5, and the half-life would be longer in the pH range of 6 to 8. These studies demonstrate that alkali feldspars in both the Neogene and Quaternary aquifers contaminated by HFFF dissolves at very low rates because the pH variations in the contaminated groundwater samples NG1, NG2, and QG1 are in the range of 7.23 to 7.89, 7.28 to 7.89, and 7.89 to 8.08, respectively. Therefore, under the premise of dynamic groundwater monitoring with a frequency of three months, which is common in shale gas fields, too little boron is released from the dissolution of alkali feldspars in the aquifers to affect the boron isotopes in HFFF-contaminated groundwater.

3.2.2. Carbonate Precipitation

Borate ions (¹⁰B(OH)₄⁻) can be incorporated into calcium carbonate (CaCO₃) lattices through carbonate precipitation, resulting in the kinetic fractionation of boron isotopes and an increase in δ¹¹B in aqueous solutions [32,49]. The results of conservative mixing show that the SI values of calcite and aragonite are greater than 0 in the contaminated groundwater samples NG1, NG2, and QG1 when the HFFF flux reaches more than 1% of the groundwater, indicating that carbonates in both the contaminated Neogene and Quaternary aquifers are in the saturation state. However, previous studies reported that carbonates precipitate in aqueous solution only when the saturation index is greater than 2 or 2.5 [50]. Consequently, boron isotope fractionation caused by carbonate precipitation in contaminated aquifer systems does not need to be considered in this study.

3.2.3. Boron Adsorption of Clay Minerals

In aqueous solutions, boron is mainly present in the form of boric acid (B(OH)₃). When the pH of the aqueous solution is higher than 6.7, B(OH)₄⁻ is generated through the hydrolysis of B(OH)₃, as shown in Equation (1), and the reaction process is mainly controlled by the pH [51]. The hydrolysis of B(OH)₃ is accompanied by equilibrium fractionation of boron isotopes [52]. Because the fractionation factor of B(OH)₃ is always higher than that of B(OH)₄⁻ under different temperature conditions, B(OH)₃ is always preferentially enriched in ¹¹B, and B(OH)₄⁻ is enriched in ¹⁰B (Equation (2)) [31,53]. Thus, the δ¹¹B of contaminated groundwater can be affected through selectively adsorbing ¹⁰B(OH)₄⁻ in the liquid phase onto clay minerals in the aquifers.

The aquifers represented by NG1 and QG1 consists of quartz, feldspar, and biotite, while clay minerals are absent, which plays a key role in boron isotope equilibrium fractionation. Hence, little boron isotope fractionation would be caused by adsorption between the solid and aqueous phases after HFFF leakage into the aquifers at NG1 and QG1. Therefore, the δ¹¹B of contaminated groundwater in the NG1 and QG1 aquifers can be generalized as conservative mixing. However, the aquifer represented by NG2 includes approximately 10% clay minerals. These minerals could cause equilibrium fractionation of boron isotopes between the aqueous and solid phases with preferential adsorption of ¹⁰B(OH)₄⁻, which was influenced by the variation of pH values (from 7.29 to 7.89,

Table 3. Summary of parameters and process data used and calculated for δ11B in contaminated NG2 groundwater, considering equilibrium fractionation caused by preferential adsorption.

Mixing Ratio of CY1−7d in Contaminated Groundwater		0.0%	1.0%	5.0%	10.0%	30.0%
NG1	[B]CMW (mM)	0.214	0.213	0.207	0.201	0.174
	δ11BCMW (‰)	26.1	26.0	25.5	24.8	21.7
	∆[B]contamination (mM)	/	−0.001	−0.007	−0.013	−0.04
	∆δ11B contamination (‰)	/	−0.1	−0.6	−1.3	−4.5
	SICalcite	−0.57	−0.22	0.38	0.69	1.17
	SIAragonite	−0.72	−0.37	0.23	0.54	1.01
QG1	[B]CMW (mM)	0.408	0.405	0.392	0.375	0.310
	δ11BCMW (‰)	24.4	24.3	24.0	23.7	22.0
	∆[B]contamination	/	−0.003	−0.016	−0.033	−0.098
	∆δ11B contamination	/	−0.1	−0.3	−0.7	−2.4
	SICalcite	0.84	0.85	0.90	0.95	1.12
	SIAragonite	0.69	0.70	0.74	0.79	0.96
Mixing ratio of CY1−7d in contaminated groundwater		50.0%	70.0%	90.0%	95.0%	100.0%
NG1	[B]CMW (mM)	0.147	0.120	0.094	0.086	0.080
	δ11BCMW (‰)	17.3	11.0	1.0	−2.4	−6.4
	∆[B]contamination (mM)	−0.067	−0.094	−0.12	−0.128	−0.134
	∆δ11B contamination (‰)	−8.8	−15.1	−25.1	−28.6	−32.6
	SICalcite	1.36	1.46	1.89	1.89	1.90
	SIAragonite	1.20	1.30	1.76	1.76	1.77
QG1	[B]CMW (mM)	0.244	0.178	0.113	0.096	0.080
	δ11BCMW (‰)	19.3	14.7	4.7	0.1	−6.4
	∆[B]contamination	−0.164	−0.230	−0.295	−0.312	−0.328
	∆δ11B contamination	−5.0	−9.6	−19.6	−24.2	−30.8
	SICalcite	1.25	1.37	1.47	1.49	1.90
	SIAragonite	1.10	1.22	1.32	1.34	1.77

Note: ∆δ¹¹B _{contamination} = δ¹¹B_{NG1 or QG1} − δ¹¹B_CMW, ∆[B]_{contamination} = [B]_{NG1 or QG1} − [B]_CMW; “/” indicates the value does not need to be acquired from NG2 groundwater without mixing with HFFF.

) resulting from increasing HFFF in contaminated groundwater. Thus, the effect of adsorption must be considered in the NG2 aquifer system.

B(OH)₃ + H₂O→B(OH)₄⁻ + H⁺

(1)

¹⁰B(OH)₃ + ¹¹B(OH)₄⁻→¹¹B(OH)₃ + ¹⁰B(OH)₄⁻

(2)

3.3. Modeling of δ¹¹B in Contaminated Groundwater Considering Boron Isotope Equilibrium Fractionation Caused by Adsorption

HFFF contamination can be divided into three steps to model the δ¹¹B in contaminated groundwater considering equilibrium fractionation.

(1) Equilibrium fractionation of boron isotopes between aquifer media and groundwater occurs at a background pH of 7.29 and temperature of 14.8 °C before HFFF contamination.

The first step is expressed as:

$$K{d}_B=\frac{{\left[B\right]}_{BS}}{{\left[B\right]}_{BW}}$$

(3)

$${\alpha }_B=\frac{{\delta }^{11}{B}_{BS}+1000}{{\delta }^{11}{B}_{BW}+1000}$$

(4)

where Kd_B is the background distribution coefficient of the aquifer system before contamination; [B]_BS and [B]_BW are the background boron concentrations of the aquifer media and groundwater, respectively; α_B is the background fractionation factor between the adsorbed and dissolved species of boron; and δ¹¹B_BS and δ¹¹B_BW are the background δ¹¹B values of the aquifer media and groundwater, respectively.

(2) Conservative mixing of background groundwater and HFFF causes a change in δ¹¹B in the aqueous phase.

The second step is expressed as:

$${\left[B\right]}_{CMW}={\left[B\right]}_{BW}(1-x_{HW})+{\left[B\right]}_{HFFF}{x}_{HW}$$

(5)

$${\delta }^{11}{B}_{CMW}=\frac{{\left[B\right]}_{BW}{\delta }^{11}{B}_{BW}(1-x_{HW})+{\left[B\right]}_{HFFF}{\delta }^{11}{B}_{HFFF}{x}_{HW}}{{\left[B\right]}_{CMW}}$$

(6)

where [B]_CMW and δ¹¹B_CMW are the boron concentration and δ¹¹B of conservative mixing between background groundwater and HFFF before adsorption occurred in the contaminated aquifer system, respectively; [B]_HFFF and δ¹¹B_HFFF are the boron concentration and δ¹¹B of HFFF, respectively; and x_Hw is the volume percentage of HFFF in the HFFF-contaminated groundwater. [B]_BW and δ¹¹B_BW are the same variables as described in Equations (3) and (4).

(3) The changed pH causes hydrolysis of B(OH)₃ (Equation (1)), which is associated with the boron isotope exchange reaction (Equation (2)) in the aqueous phase. Then, a new equilibrium fractionation of boron isotopes between aquifer media and conservative mixing water occurs after preferential adsorption/desorption of ¹⁰B(OH)₄ in the aqueous phase on clay minerals under the changed pH of contaminated groundwater.

The final step is expressed as:

$$K{d}_C=\frac{{\left[B\right]}_{CS}}{{\left[B\right]}_{CW}}=\frac{{\left[B\right]}_{BS}+{\left[B\right]}_{adsorbS}}{{\left[B\right]}_{CMW}-{\left[B\right]}_{adsorbedW}}$$

(7)

$${\left[B\right]}_{adsorbS}=u{\left[B\right]}_{adsorbedW}$$

(8)

$${\alpha }_C=\frac{{\delta }^{11}{B}_{CS}+1000}{{\delta }^{11}{B}_{CW}+1000}=\frac{\frac{{\delta }^{11}{B}_{BS}\times {\left[B\right]}_{BS}+{\delta }^{11}{B}_{adsorbed}\times {\left[B\right]}_{adsorbS}}{{\left[B\right]}_{BS}+{\left[B\right]}_{adsorbS}}+1000}{\frac{{\delta }^{11}{B}_{CMW}\times {\left[B\right]}_{CMW}-{\delta }^{11}{B}_{adsorbed}\times {\left[B\right]}_{adsorbedW}}{{\left[B\right]}_{CMW}-{\left[B\right]}_{adsorbedW}}+1000}$$

(9)

where Kd_c is the distribution coefficient of the contaminated groundwater with changed pH and temperature; [B]_CS and [B]_CW are the boron concentrations of the contaminated aquifer media and groundwater, respectively; and [B]_adsorbS and [B]_adsorbedW are the elevated boron concentrations of the aquifer media absorbed from conservative mixing water and the decreased boron concentration of mixing water caused by adsorption, respectively. In addition, u is the ratio of the mass of water to the mass of oven-dried material; α_C is the fractionation factor between the adsorbed and dissolved species of boron of the contaminated aquifer system; δ¹¹B_CS and δ¹¹B_CW are the δ¹¹B values of the contaminated aquifer media and groundwater after conservative mixing and adsorption/desorption, respectively; and δ¹¹B_absorbed is the δ¹¹B of absorbed boron from conservative mixing water to the aquifer media. [B]_BS, [B]_CMW, δ¹¹B_BS, and δ¹¹B_CMW are the same variables as described in Equations (3)–(6).

3.4. Sensitivity Assessment of Boron Isotopes as an Indicator of HFFF-Contaminated Groundwater

To assess the sensitivity of boron isotopes as an indicator of HFFF-contaminated groundwater, the seventh-day HFFF and background groundwater (NG1, NG2, and QG1) were selected as the endmembers of the contamination source and contamination receptors, respectively. The NG1 and QG1 samples contaminated by HFFF can be generalized as examples of conservative mixing (Table 3). However, the NG2 contamination was simulated using the model established in Section 3.3. Experimental data for K_d and α in the NG2 aquifer system containing 10% clay minerals are lacking. Instead, it was assumed that the NG2 aquifer medium is entirely composed of clay minerals, which have experimental data for K_d and α under a variety of pH and temperature conditions. If there is little boron isotope fractionation that plausibly occurs in the environment in the range of HFFF inputs for the above assumption, δ¹¹B could be a cogent tool for identifying HFFF from NG2 groundwater under real hydrogeological conditions.

The pH values for contaminated groundwater at different HFFF mixing ratios were obtained by the Mixing Module of PHREEQC (Version 3, USGS, Reston, VA, USA). The Kd and α values between the adsorbed and dissolved species of boron for a variety of pH values and a temperature of 15 °C can be checked from previous clay mineral experimental studies (

Table 4. Summary of parameters and process/result data used and calculated for δ¹¹B in contaminated NG2 groundwater considering equilibrium fractionation caused by preferential adsorption at different HFFF mixing ratios.

Mixing Ratio of CY1-7d in Contaminated NG2 Groundwater	0.0%	1.0%	5.0%	10.0%	30.0%
pH value	7.29	7.35	7.50	7.59	7.73
Kdmix at 15 °C [31]	1.868	1.932	2.123	2.256	2.503
αmix at 15 °C [31]	0.9 714	0.9 719	0.9 729	0.9 734	0.9 742
[B]CMW (mM)	0.245	0.244	0.237	0.229	0.196
δ11BCMW (‰)	55.7	55.5	54.6	53.5	48.1
[B]absorbedW (mM)	/	0.006 3	0.020 3	0.024 6	0.012 1
δ11Babsorbed (‰)	/	52.8	49.3	45.0	−8.1
[B]CW (mM)	/	0.238	0.217	0.204	0.184
δ11BCW (‰)	/	55.5	55.1	54.5	51.8
∆δ11Bpreferential adsorption	/	0.1	0.5	1.0	3.7
∆[B]contamination (mM)	/	−0.007	−0.028	−0.041	−0.061
∆δ11B contamination	/	−0.2	−0.6	−1.2	−3.9
SICalcite	−0.12	−0.2	−0.6	−1.2	−3.9
SIAragonite	−0.27	0.01	0.36	0.61	1.08
Mixing ratio of CY1-7d in contaminated NG2 groundwater	50.0%	70.0%	90.0%	95.0%	100.0%
pH value	7.79	7.83	7.87	7.88	7.89
Kdmix at 15 °C [31]	2.617	2.713	2.800	2.805	2.822
αmix at 15 °C [31]	0.9 745	0.9 747	0.9 749	0.9 749	0.9 749
[B]CMW (mM)	0.163	0.129	0.096	0.088	0.080
δ11BCMW (‰)	40.4	28.9	9.4	2.2	−6.4
[B]absorbedW (mM)	−0.012 1	−0.038 1	−0.065 1	−0.072 8	−0.079 9
δ11Babsorbed (‰)	158.6	100.3	88.3	86.0	84.5
[B]CW (mM)	0.175	0.167	0.161	0.161	0.160
δ11BCW (‰)	48.6	45.2	41.2	40.2	39.1
∆δ11Bpreferential adsorption	8.2	16.3	31.8	38.0	45.5
∆[B]contamination (mM)	−0.070	−0.078	−0.084	−0.084	−0.085
∆δ11B contamination	−7.1	−10.5	−14.5	−15.5	−16.6
SICalcite	1.29	1.41	1.49	1.50	1.90
SIAragonite	1.13	1.26	1.34	1.35	1.77

Note: Positive values of [B]_adsorbedW indicate the sediment adsorbed the B from mixing water, while negative values of [B]_adsorbedW indicate the sediment desorbed the B from exchangeable sites on the mineral surfaces to mixing water; ∆δ¹¹B_{preferential adsorption} = δ¹¹B_CW − δ¹¹B_CMW, ∆δ¹¹B _{contamination} = δ¹¹B_NG2 − δ¹¹B_CW, ∆[B]_{contamination}= [B]_NG2 − [B]_CW; “/” indicates the value does not need to be acquired from NG2 groundwater without mixing with HFFF.

). Then, using the collected data of Kd_B and α_B [29] and the measured data of [B]_BW and δ¹¹B_BW (Table 1) for the background equilibrium fractionation state before contamination, [B]_BS (0.458 mM/kg) and δ¹¹B_BS (25.5‰) are obtained in the first step (Equations (1) and (2)) of modeling boron isotopes. The [B]_CMW and δ¹¹B_CMW values can be obtained using the measured data of [B]_BW, [B]_HFFF, δ¹¹B_BW, and δ¹¹B_HFFF (Table 1) at different x_HW values in the second step (Table 4). Finally, δ¹¹B_CW can be calculated at the new equilibrium fractionation state after mixing and adsorption/desorption (Table 4).

The simulation results of groundwater contamination by HFFF show that δ¹¹B of the NG2 contaminated groundwater (δ¹¹B_CW: 40.0 to 55.6‰, Table 4) in a variety of HFFF mixing ratios (1 to 100%), considering the mixing of background groundwater and HFFF and the selective adsorption by clay minerals after mixing, is always higher than those of the corresponding contaminated groundwaters mixed conservatively (δ¹¹B_CMW:−6.4 to 55.6‰, Table 4), suggesting that preferential adsorption of clay minerals would reduce the sensitivity of boron isotopes as an indicator of HFFF contamination in groundwater. Although the ∆δ¹¹B_{preferential adsorption} (∆δ¹¹B_{preferential adsorption} = δ¹¹B_CW − δ¹¹B_CMW) gradually increases with an increasing degree of contamination (or HFFF mixing ratio), a slight increase of ∆δ¹¹B_{preferential adsorption} (<3.6‰) would be caused in the range of HFFF inputs (<30% HFFF) that might plausibly occur in the environment. This indicates that selective adsorption has little impact on the usability of δ¹¹B as an indicator of HFFF in NG2 groundwater in the study area.

Response curves of the contamination degree were also generated for the NG1, NG2, and QG1 groundwater interacting with the seventh-day HFFF (Figure 3). For the scenario of HFFF contamination, a flux equivalent to 5% of NG1, NG2, and QG1 groundwater can result in detectable shifts in the δ¹¹B (∆δ¹¹B_{contamination} = −0.6‰, −0.6‰, and −0.3‰, respectively) and boron concentration (∆[B]_{contamination} = −0.07 mg/L, −0.30 mg/L, and −0.18 mg/L, respectively) for the shallow groundwater (Table 3 and Table 4). When the HFFF flux reaches 30% of the NG1, NG2, and QG1 groundwater, the δ¹¹B (∆δ¹¹B_{contamination} = −4.5‰, −3.9‰, and −2.4‰, respectively) and boron concentration (∆[B]_{contamination} = 0.44 mg/L, 0.66 mg/L, and 1.07 mg/L, respectively) for all shallow groundwater are significantly affected. These results illustrate that boron isotopes can be an effective indicator for identifying groundwater contaminated by HFFF produced from the Dameigou Shale in the study area.

4. Conclusions

In the Neogene (NG1 and NG2) and Quaternary (QG1) aquifers of the study area, the potential hydrogeochemical processes that may affect boron isotopes in HFFF-contaminated groundwater (in addition to conservative mixing of HFFF and background groundwater) are incomplete dissolution of alkali feldspar minerals and carbonate precipitation. Although albite and K-feldspar dissolution occurs in HFFF-contaminated Neogene and Quaternary aquifer systems, the very low dissolution rates of these minerals, mainly controlled by pH, result in insignificant boron being produced from dissolution, and thus this process does not affect the sensitivity of boron isotopes as an HFFF indicator in the shallow groundwater of the study area. Because carbonate is in the non-precipitating state in a range of HFFF mixing ratios, boron isotope fractionation caused by carbonate precipitation in contaminated aquifer systems does not need to be considered for this study area.

The simulation of δ¹¹B for HFFF-contaminated groundwater shows that preferential adsorption onto clay minerals in aquifer media affects boron isotopes, reducing the sensitivity of δ¹¹B as an indicator of HFFF. However, for this study area, the usability of δ¹¹B as an HFFF indicator is likely to be unaffected for the range of HFFF inputs that might plausibly occur in the environment (<30% HFFF).

For the scenario of HFFF contamination, when the mixing ratio of HFFF in contaminated groundwater increased by 5%, boron isotopes in Neogene and Quaternary groundwater had a detectable response to HFFF contamination, suggesting δ¹¹B is a sensitive indicator of HFFF contamination in the shallow groundwater from the Dameigou Shale in the northern Qaidam Basin.