Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 237 (2018) 261–274

Fractionation of Mg isotopes by clay formationand calcite precipitation in groundwater with long residence

times in a sandstone aquifer, Ordos Basin, China

Hong Zhang a, Xiao-Wei Jiang a,⇑, Li Wan a, Shan Ke b, Sheng-Ao Liu b,Guilin Han c, Huaming Guo a, Aiguo Dong c

aMOE Key Laboratory of Groundwater Circulation and Environmental Evolution, China University of Geosciences, Beijing 100083, ChinabState Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences, Beijing 100083, China

c Institute of Earth Sciences, China University of Geosciences, Beijing 100083, China

Received 16 June 2017; accepted in revised form 22 June 2018;

Abstract

Compared with the numerous studies on river and soil waters, studies on Mg isotopes of groundwater are limited. In thisstudy, a sandstone aquifer in the Ordos Basin, China with contrast contents of Mg in shallow and deep groundwater is selectedto examine the behavior of Mg isotopes during groundwater circulation. The d26Mg values of shallow groundwater are withinthe range of widely reported results of groundwater, while those of deep groundwater are found to be as light as �3.30‰ to�2.13‰. Assuming that shallow groundwater is an endmember, 87Sr/86Sr ratios show that calcite dissolution has contributionto low d26Mg of deep groundwater, but mixing alone cannot explain the coupled low d26Mg and lowMg contents. The removalof Mg in deep groundwater is found to be mainly caused by incorporating into neoformed clay minerals, which further lowersd26Mg. For the deep groundwater samples denoted as G1 and G3, the relationship between d26Mg and 1/Mg has beenquantitatively explained by the superposition of calcite dissolution and clay formation with a fractionation factor (aclay–water)of 1.0003. For samples denoted as G2, in addition to calcite dissolution and clay formation, high proportion of Mg in theresidual solution are further removed via precipitation of low-Mg calcite, which leads to increased d26Mg. There are increasinglystronger degrees of clay formation in G3, G1, and G2 due to the increasingly longer travel distances and travel times ofgroundwater from recharge to discharge areas. This study enhances understanding on the factors controlling Mg isotopesof groundwater, as well as the geochemical processes of subsurface water-rock interactions in sandstone aquifers.� 2018 Elsevier Ltd. All rights reserved.

Keywords: Groundwater geochemistry; Magnesium isotopes; Rayleigh distillation; Residence time; Groundwater circulation; Flowing well

1. INTRODUCTION

The water cycle is a fundamental process involvingmovement of water and transport of elements on the Earth.The water-rock interactions during the water cycle could

https://doi.org/10.1016/j.gca.2018.06.023

0016-7037/� 2018 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: 29 Xueyuan Road, Haidian Dist.,Beijing 100083, China.

E-mail address: jxw@cugb.edu.cn (X.-W. Jiang).

lead to different hydrochemical and isotopic compositionsof water (Ingebritsen et al., 2006; Edmunds and Shand,2008). As a major element in water, magnesium participatesin various geochemical processes, which could cause signif-icant fractionation of Mg isotopes (24Mg, 25Mg and 26Mg)(Tipper et al., 2010). Therefore, Mg isotopes, which belongto stable isotopes, have been widely applied to tracegeochemical processes in river water and soil water(Pogge von Strandmann et al., 2012; Mavromatis et al.,

Fig. 1. A comparison of published d26Mg values of different watersand rocks. Data of groundwater is from Jacobson et al. (2010),Immenhauser et al. (2010), Tipper et al. (2012a, 2012b) Geske et al.(2015), Ma et al. (2015); data of soil water is from Tipper et al.(2010, 2012b), Pogge von Strandmann et al. (2012), Ma et al.(2015); data of seawater is from Young and Galy (2004), Hippleret al. (2009), Teng et al. (2010), Tipper et al. (2010), Mavromatiset al. (2016); data of river water is from Tipper et al. (2006), Brenotet al. (2008), Pogge von Strandmann et al. (2008), Jacobson et al.(2010), Wimpenny et al. (2011), Tipper et al. (2012a, 2012b), Leeet al. (2014), Fan et al. (2016), Mavromatis et al. (2016); data ofrainwater is from Tipper et al. (2010, 2012b), Bolou-Bi et al. (2012);data of drip water is from Young and Galy. (2004), Immenhauseret al. (2010); data of carbonates is from Buhl et al. (2007), Brenotet al. (2008), Immenhauser et al. (2010); and data of silicates is fromHuang et al. (2009), Liu et al. (2010). The d26Mg values ofgroundwater measured in the current study are also shown in thetop of the figure.

262 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

2016). In previous studies, Mg isotopes of river and/or soilwater are found to be controlled by dissolution of minerals(Brenot et al., 2008; Tipper et al., 2010, 2012a), precipita-tion of carbonates (Lee et al., 2014; Fan et al., 2016), neo-formation of clay minerals (Tipper et al., 2010, 2012a,2012b; Wimpenny et al., 2014), adsorption onto clay miner-als (Pogge von Strandmann et al., 2012; Huang et al., 2012;Ma et al., 2015) and uptake by plants (Pogge vonStrandmann et al., 2008; Bolou-Bi et al., 2012). Due tothe simultaneous controls by several factors, the dominantcontrols on the fractionation of Mg isotopes in waterremain unclear, and the predictive capacity of Mg isotopegeochemistry is limited compared with other widely utilizedisotopes such as Sr isotopes (Jacobson et al., 2010; Tipperet al., 2012b).

Although groundwater circulation is an indispensablecomponent of the water cycle, there are limited studies onMg isotopes of groundwater. In order to reveal the behav-ior of Mg isotopes during groundwater circulation fromrecharge to discharge areas, intensive sampling in bothrecharge and discharge areas is a prerequisite. In a recentfield study in the sandstone aquifer of the Ordos Basin,Wang et al. (2015a) found that deep groundwater in dis-charge areas has much lower Mg contents than shallowgroundwater in recharge areas, which provides a uniqueopportunity to examine the behavior of Mg isotopesaccompanying with Mg removal. Therefore, the sandstoneaquifer of the Ordos Basin is selected as the current studyarea. The d26Mg values of shallow groundwater are foundto be within the normal range between �1.63‰ and�0.59‰ (Fig. 1) as reported in previous studies with a totalof 20 groundwater samples (Jacobson et al., 2010;Immenhauser et al., 2010; Tipper et al., 2012a, 2012b;Geske et al., 2015; Ma et al., 2015). However, the d26Mgvalues of deep groundwater in the discharge areas of thecurrent study are all found to be below the lower limit ofreported values in groundwater. This study aims to quanti-tatively examine the major processes controlling the lowd26Mg and low Mg contents in deep groundwater withthe aid of major ions and 87Sr/86Sr ratios of groundwater.

2. METHODS

2.1. Study area

The Ordos Basin located in northwestern China is thesecond largest sedimentary basin of China. Due to thesemi-arid climate, groundwater is the main water supplyin the basin (Hou et al., 2008; Jiang et al., 2018). In thenorth part of the Ordos Basin, the thick Cretaceous sand-stone with sporadic clay lenses is the main aquifer. The thinunconsolidated Quaternary sediments overlying the Creta-ceous sandstone also constitute an aquifer in topographiclows with shallow water table.

The study sites of the current study include the DositRiver Watershed and the Wudu Lake Catchment (Fig. 2).The Dosit River Watershed covering an area of around11,000 km2 can be considered as a sub-basin of the OrdosBasin, with the Dosit River being the lowest dischargepoints and the major surface water body. The Wudu Lake

Catchment adjacent to the Dosit River Watershed coversan area of around 200 km2 and is one of the numeroussmall catchments in the Ordos Basin, with Wudu Lakebeing the lowest discharge point and the only surface waterbody. In both study sites, there are numerous wells with dif-ferent depths drilled into the Cretaceous sandstone or theQuarternary sediments for irrigation and domestic uses,which provide an excellent opportunity for intensivegroundwater sampling in both recharge and dischargeareas. More detailed descriptions of the Dosit River Water-shed can be found in Jiang et al. (2014) and Wang et al.(2015a), and of the Wudu Lake Catchment can be foundin Wang et al. (2015b) and Jiang et al. (2017). Note thatboth groundwater and surface water are alkaline in thenorth part of the Ordos Basin, which is one of the fewregions with numerous soda lakes due to its arid climate.

2.2. Sampling

In the summers of 2015 and 2016, 20 groundwatersamples were collected in the study area. Among them, 16samples were from the Dosit River Watershed, and 4 werefrom the Wudu Lake Catchment. Before sampling, electri-cal conductivity (EC), temperature, and pH were measured

Fig. 2. The topography and distribution of groundwater and rock samples in the Dosit River Watershed and Wudu Lake Catchment.

H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274 263

in the field. All groundwater samples were stored in LDPEbottles after passing through 0.45 lm filters, and the sam-ples for dissolved cations, and Sr and Mg isotopes wereacidified to pH = 2 with concentrated ultrapure HNO3.To examine the control of lithology on the isotopes ofgroundwater, 2 rock samples (labeled as R1 and R2) werecollected from the sandstone outcrops near the samplingwells in the Dosit River Watershed.

In the Dosit River Watershed, the 16 groundwater sam-ples represent three different hydrogeological conditions. 5samples (S1 through S5) were collected from shallow wellsdrilled into the sandstone aquifer in topographic highs,which represent shallow groundwater from the sandstoneaquifer in recharge areas; 8 samples (D1 through D8) werefrom deep wells drilled into the sandstone aquifer intopographic lows, which represent deep groundwater fromthe sandstone aquifer in discharge areas; 3 samples(Q1 through Q3) were from shallow dug wells in theQuaternary deposits in topographic lows, which representshallow groundwater from the unconsolidated-depositaquifer in discharge areas. In the Wudu Lake Catchment,the 4 groundwater samples (D9 through D12) were all col-lected from deep wells drilled into the sandstone aquifer intopographic lows, representing deep groundwater from thesandstone aquifer in the discharge area of the catchment.

The 12 deep groundwater samples in discharge areaswere collected at the outlets of flowing wells drilled intothe unconfined sandstone aquifer. For flowing wells in

unconfined-aquifer basins, Zhang et al. (2018) found thatgroundwater flowing out of the well is mainly from theaquifer corresponding to the deep part of the flowing well,and groundwater in the shallow part of the aquifer wouldnot enter the well. During the upward movement fromthe deep to the outlet of a flowing well, the decreasing pres-sure would inevitably lead to degassing and removal of suchions as Ca, Mg, Sr and HCO3 involving the carbon cycle. Itis assumed that other hydrochemical components notinvolved in the carbon cycle would not be influenced bydegassing and precipitation of carbonates. Note that evenif the samples were pumped from the deep part of the aqui-fer, degassing could not be avoided.

2.3. Geochemical analyses

To obtain the hydrochemical components of groundwa-ter, HCO3 and CO3 were analyzed by titration in the field,Cl, SO4, and NO3 were analyzed by ion chromatography(ICS-900, Dionex), and Na, K, Mg, Ca, Al and Sr by induc-tively coupled plasma (ICP-OES-900, Thermo). The chargebalance errors are all within 5%. SiO2 was analyzed by usingthe molybdenum blue spectrophotometer method with a T6UV Spectrophotometer. The PHREEQC program(Parkhurst and Applo, 1999) implemented with the llnl.datthermodynamic database (Delany and Lundeen, 1990) wasused to obtain the partial pressure of CO2 (PCO2) and thesaturation indices of minerals in groundwater. To determine

264 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

the mineralogy of rock samples, crushed powders of rocksamples were tested by a D/MAX2500 X-ray diffractometerin Xi’an Center of Geological Survey, China. The details ofmineralogical measurements can be found in Liu et al.(2015).

Mg isotopes of groundwater and rock samples were ana-lyzed by a Neptune Plus MC-ICP-MS at the Isotope Geo-chemistry Laboratory of China University of Geosciences,Beijing. Rock samples were firstly dissolved in HF-HNO3

mixture in Teflon vessels, and the solutions were driedand re-dissolved in HCl-HNO3 mixture, and were driedagain and re-dissolved in concentrated HNO3. Groundwa-ter samples were dried in Teflon vessels and dissolved inconcentrated HNO3 directly. Then the solutions of dis-solved rock samples and of dissolved groundwater sampleswere dried and re-dissolved in 1 N HNO3 in preparation forchromatographic separation. Magnesium was separatedfrom the matrix using ion exchange resin AG50W-X8 fol-lowing the procedure described in Teng et al. (2007) andKe et al. (2016). The same column procedure was repeatedto achieve the required level of purification with the wholeprocedure blank less than 10 ng. The final solutions wereheated to dryness in a vented laminar-flow hood and thendissolved in 3% HNO3 for mass spectrometry. More detailsof measuring Mg isotopes can be found in Ke et al. (2016).Mg isotope ratios are reported in delta notation relative tothe Dead Sea metal Mg standard (DSM3), which can beexpressed as

dxMgðÞ¼ xMg= 24Mg� �

sample= xMg= 24Mg� �

DSM3�1

h i�1000;

where x = 25 or 26. The precision and accuracy of the mea-surements were validated by analyzing the solution stan-dard GSB and the rock standard BHVO-2. The averaged26Mg values of GSB and BHVO-2 are �2.03 ± 0.04(2SD, n = 8) and �0.24 ± 0.04 (2SD, n = 3), respectively,which are identical to the recommended values within theanalytical uncertainty (Teng et al., 2015; Ke et al., 2016).

Sr isotope ratios of groundwater and rock samples wereanalyzed by a Neptune Plus MC-ICP-MS at the Ministry ofEducation Key Laboratory of Surficial Geochemistry ofNanjing University. Rock samples were firstly dissolved inHF-HNO3 mixture, then the solutions were dried and dis-solved in concentrated HNO3 for twice, and were driedand dissolved in concentrated HCl for twice. Finally, thesolutions with dissolved rocks were dried and dissolved in3N HNO3 in preparation for purification. Groundwatersamples were dried and dissolved in 3N HNO3 directly inpreparation for purification. The purification of Sr was fin-ished using Sr-Spec resin following the procedure describedin Aciego et al. (2009) with the whole procedure blank <1ng, which is negligible compared with the total Sr amountin the samples. The results of 87Sr/86Sr ratios were correctedby normalizing the measured 86Sr/88Sr to 0.1194. Moredetails on the procedures of measuring Sr isotopes can befound in Zhang et al. (2016). The standard SRM-987 weremeasured periodically to check the precision and accuracyof isotopic analyses, while the rock standard BCR-2 andthe seawater standard IAPSO were used to verify the chem-ical procedures. The mean 87Sr/86Sr ratios of SRM-987,

BCR-2 and IAPSO are 0.710254 ± 0.000005 (2SE, n =15), 0.705034 ± 0.000017 (2SE, n = 2) and 0.709241 ±0.000036 (2SE, n = 2), respectively.

3. RESULTS

Major elemental concentrations, 87Sr/86Sr ratios andd26Mg values of the groundwater samples are listed inTable 1. The major minerals, 87Sr/86Sr ratios and d26Mgvalues of rock samples are listed in Table 2.

3.1. General chemistry of groundwater and sandstone

The general chemistry of groundwater, including pH,PCO2 and the concentrations of major ions, versus welldepth is shown in Fig. 3. Compared with shallow ground-water in recharge areas, deep groundwater in dischargeareas has higher pH, but lower PCO2, HCO3 + CO3, Ca,Mg and K. Except for the 3 groundwater samples collectedfrom shallow wells in Quaternary deposits, shallow anddeep groundwater in the sandstone aquifer have similarcontents of Cl, SO4 and Na.

According to measurements of mineralogy of sandstonesamples, quartz, plagioclase, K-feldspar are the main min-erals, while calcite, hornblende, montmorillonite, illite andchlorite also exist (Table 2). The proportions of calcite inthe two rock samples are 0.9% and 2.1%, respectively. Pre-vious studies revealed that the mineral compositions of Cre-taceous sandstones are generally the same in the study area(e.g., Su et al., 2011), implying that the different hydro-chemical compositions of shallow and deep groundwatercould not have been caused by differences in minerals.

3.2. Mg and Sr isotopes of groundwater and rock samples

For the two sandstone samples, the d26Mg values arefound to be �0.49‰ and �0.39‰, respectively, which areclose to those of silicate minerals (Fig. 1), and the 87Sr/86Srratios are found to be 0.7128 and 0.7147, respectively,which are within the range of silicate minerals and calcite(Clark, 2015).

Similar to the dependence of most major ions of ground-water on well location and well depth, the d26Mg values ofgroundwater are also found to be dependent on well loca-tion and well depth (Fig. 3k). Although the 8 shallowgroundwater samples (S1 through S5, and Q1 throughQ3) are located in different aquifers of a watershed, theyhave similar d26Mg ranging between �1.29‰ and�1.07‰. In contrast, the d26Mg values of the 12 deepgroundwater samples from the sandstone aquifer in differ-ent discharge areas are found to range from �3.30‰ to�2.13‰, which are much lower than those of shallowgroundwater. Because the flowpaths of deep groundwaterin discharge areas originate from recharge areas, this studyfocuses on the behavior of Mg isotopes of groundwaterduring the circulation from the shallow of recharge areasto the deep of discharge areas in the sandstone aquifer.

According to Rao et al. (2015), rain water in the studyarea has 87Sr/86Sr ratios ranging from 0.7105 to 0.7110.87Sr/86Sr ratios of shallow groundwater from the sandstone

Table 1Major elemental concentrations, Sr and Mg isotopic compositions of groundwater. Uncertainties are quoted as twice the standard err (2 SE) or standard deviation (2 SD).

Sample Latitude(N)

Longitude(E)

Depth(m)

pH T (�C)

SO4

(ppm)Cl(ppm)

HCO3

(ppm)CO3

(ppm)NO3

(ppm)K(ppm)

Na(ppm)

Ca(ppm)

Mg(ppm)

Sr(ppb)

Al(ppb)

iO2

ppm)

87Sr/86Sr 2SE d25Mg(‰)

2SD d26Mg(‰)

2SD

S1 38.9162 107.6706 7 8.07 17.2 140.18 106.41 259.31 0 10.98 3.12 165.60 32.44 29.20 788 30 5.60 0.7112 0.000014 �0.66 0.03 �1.29 0.07S2 38.9472 107.6920 31 7.98 18.3 116.32 106.99 202.94 0 3.79 3.32 115.35 39.64 26.86 1080 56 9.12 0.7114 0.000016 �0.67 0.04 �1.29 0.07S3 38.9541 107.7072 150 8.07 18.4 108.86 88.54 225.49 0 2.29 3.04 108.86 35.72 24.70 588 24 5.62 0.7113 0.000014 �0.68 0.04 �1.29 0.07S4 38.9377 107.7249 70 8.14 14.6 55.70 32.50 169.12 0 47.99 2.34 48.65 40.94 22.45 549 32 1.13 0.7115 0.000015 �0.67 0.04 �1.28 0.03S5 38.9382 107.7264 180 8.16 13.3 64.80 48.28 214.21 0 49.71 3.68 114.44 30.40 13.33 363 23 7.76 0.7114 0.000012 �0.58 0.01 �1.11 0.04Q1 38.9030 107.6668 10 7.82 14.2 542.63 305.25 428.43 0 33.58 4.55 451.79 66.03 60.61 703 36 2.31 0.7110 0.000017 �0.61 0.07 �1.16 0.06Q2 38.9010 107.6733 4 8.11 18.8 343.82 228.79 383.33 0 6.94 1.54 382.31 30.38 15.22 1130 27 5.44 0.7113 0.000016 �0.58 0.04 �1.14 0.03Q3 38.9501 107.8798 3.5 7.78 17.5 73.28 62.49 225.49 0 47.82 3.57 115.61 29.63 13.21 360 81 4.90 0.7115 0.000017 �0.55 0.03 �1.07 0.02D1 38.8756 107.6014 350 9.44 22.1 63.81 55.58 89.10 23.90 0 0.23 109.93 6.15 0.06 13 17 5.08 0.7100 0.000016 �1.41 0.04 �2.67 0.03D2 38.9068 107.6424 300 9.33 23.4 53.16 46.39 97.20 23.90 0 0.26 103.84 2.00 0.05 12 11 3.57 0.7102 0.000017 �1.24 0.06 �2.37 0.02D3 38.9104 107.6746 518 9.11 25.1 85.85 68.20 121.48 12.65 0 0.25 141.46 3.32 0.13 14 4 3.91 0.7102 0.000014 �1.31 0.02 �2.48 0.05D4 38.8967 107.6775 250 9.17 15.7 60.50 42.90 90.19 22.18 0 0.23 100.30 2.70 0.16 15 25 1.70 0.7101 0.000014 �1.34 0.01 �2.59 0.06D5 38.8890 107.7529 350 9.06 17.1 41.10 26.29 126.36 9.03 0 0.39 86.65 3.68 0.96 107 13 1.99 0.7102 0.000014 �1.72 0.03 �3.30 0.06D6 38.9230 107.8181 400 8.85 25.3 64.27 40.20 143.49 8.83 0 0.52 109.94 5.29 1.84 151 13 2.55 0.7104 0.000016 �1.51 0.02 �2.92 0.05D7 38.9390 107.8284 248 8.56 18.6 61.82 55.71 135.29 0 0 0.72 100.83 6.08 2.21 174 15 3.37 0.7101 0.000013 �1.47 0.01 �2.83 0.03D8 38.9629 107.9078 450 8.73 22.1 53.56 43.80 126.42 17.72 0 0.42 101.10 6.52 1.67 182 820 3.94 0.7102 0.000014 �1.73 0.05 �3.30 0.06D9 38.9051 108.2988 300 9.02 17.4 57.66 76.15 170.11 15.93 0 1.14 130.06 8.08 6.25 439 3 3.38 0.7102 0.000016 �1.22 0.06 �2.34 0.04D10 38.9109 108.3386 300 8.60 29.1 36.69 6.29 178.21 0 0 0.89 74.06 8.06 4.02 474 23 4.69 0.7102 0.000013 �1.48 0.02 �2.85 0.04D11 38.8901 108.3448 160 8.32 17.5 43.68 36.88 186.31 0 0 1.68 63.20 13.92 21.19 1060 9 2.13 0.7098 0.000015 �1.10 0.03 �2.13 0.02D12 38.8840 108.3528 800 8.94 21.8 41.54 33.78 153.91 15.93 0 0.69 103.93 5.10 4.22 346 8 3.47 0.7097 0.000014 �1.41 0.02 �2.71 0.02

H.Zhan

get

al./Geochim

icaet

Cosm

ochim

icaActa

237(2018)

261–274265

or

S(

11121111111111111111

Tab

le2

TheSran

dMgisotopic

compositions,an

dtheproportionsofmajormineralsin

thesandstonesamples.

Sam

ple

Latitude

(N)

Longitude

(E)

87Sr/86Sr

2SE

d26Mg

(‰)

2SD

d25Mg

(‰)

2SD

Quartz

%Plagioclase

%K-feldspar

%Calcite

%Hornblende

%Montm

orillonite

%Illite

%Chlorite

%

R1

38.9398

107.7234

0.71

470.00

0017

�0.39

0.03

�0.20

0.05

44.4

36.9

9.1

0.9

1.2

3.0

2.0

2.5

R2

38.9553

107.7042

0.71

280.00

0015

�0.49

0.04

�0.24

0.02

34.2

49.0

3.3

2.1

7.4

1.5

1.0

1.5

266 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

aquifer in the recharge area vary from 0.7112 and 0.7115,which are higher than those of rain water due to water-rock interactions with sandstones, which have higher87Sr/86Sr ratios. However, 87Sr/86Sr ratios of deep ground-water from the sandstone aquifer in discharge areas rangefrom 0.7097 to 0.7104, much lower than shallow groundwa-ter in the recharge area (Fig. 3l). Because carbonates usu-ally have a low 87Sr/86Sr ratio ranging between 0.707 and0.709 (Clark, 2015), the lower 87Sr/86Sr ratios in deepgroundwater indicate that the contribution of carbonatedissolution in discharge areas is more significant than thatin recharge areas.

In the sandstone aquifer, from recharge to dischargeareas, as 87Sr/86Sr ratios decrease, the d26Mg values alsodecrease (Fig. 4). Therefore, calcite dissolution could havecontributed to the lower d26Mg in deep groundwater. The-oretically, the concentrations of Ca, Mg and Sr of ground-water should increase along the flowpaths due to thedissolution of calcite, which contains abundant Ca as wellas trace Mg and Sr. In fact, compared with groundwaterin recharge areas, groundwater in discharge areas has muchlower K, Ca, Mg and Sr (Fig. 3g–j), indicating that thereare geochemical processes removing K, Ca, Mg and Sr dur-ing groundwater circulation.

4. DISCUSSION

Rainfalls are the main source of groundwater recharge,and the hydrochemical components of infiltrated watercan be significantly changed by water-rock interactions inthe subsurface. As long as the climate and lithology are sim-ilar in a large-scale basin, similar water-rock interactionscould lead to comparable hydrochemistry and isotopes ofyoung groundwater in different recharge areas of the basin.Therefore, it is assumed that in the sandstone aquifer, shal-low groundwater in recharge areas could be utilized as anendmember of deep groundwater in discharge areas. Thissection aims to identify the specific geochemical processesdeep groundwater had undergone during the circulationfrom recharge to discharge areas, which lead to significantreductions in d26Mg and Mg contents.

4.1. Identification of major processes

Based on previous studies on Mg isotopes of water,d26Mg could be increased by weathering of silicate minerals(Tipper et al., 2012a), precipitation of carbonates (Tipperet al., 2006; Lee et al., 2014), and/or ion-exchange(Jacobson et al., 2010), and could be decreased by dissolu-tion of calcite (Lee et al., 2014; Fan et al., 2016), formationof clay minerals (Tipper et al., 2008; Teng et al., 2010;Opfergelt et al., 2012; Wimpenny et al., 2014) and/oradsorption onto clay minerals (Pogge von Strandmannet al., 2012; Fan et al., 2016). Because silicate weatheringdepends on PCO2, the degree of silicate weathering couldbe significant in the shallow part with high PCO2 but woulddecrease with travel distance and residence time due to theconsumption of CO2 in a closed system. When PCO2 is low,the dissolution rates of carbonates are thousands of timesquicker than silicate minerals (Holland, 1978; Suchet and

Fig. 3. The depth-dependent hydrochemistry and isotopes of groundwater. (a) pH; (b) PCO2; (c) HCO3 + CO3; (d) SO4; (e) Cl; (f) Na; (g) K;(h) Ca; (i) Mg; (j) Sr; (k) d26Mg; (l) 87Sr/86Sr.

H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274 267

Probst, 1993; Berner and Berner, 1996; Roy et al., 1999;Williams et al., 2007). Therefore, for deep groundwater indischarge areas with low PCO2, calcite dissolution plays amore significant role in the hydrochemistry and isotopiccompositions of groundwater, which has been supportedby the lower 87Sr/86Sr ratios.

From recharge to discharge areas, the concentrations ofNa in groundwater change little (Fig. 3f), and those ofK decrease (Fig. 3g), indicating that ion-exchange couldbe excluded. K could not be precipitated in natural watersdue to its high solubility, and could hardly be adsorbed byexisting clay minerals (Freeze and Cherry, 1979; Drever,1997; Appelo and Postma, 2005; Clark, 2015). Instead, the

removal of K could be caused by neoformation of clay min-erals like illite. The removals of Mg, Ca and Sr could becaused by several processes, including incorporation intoneoformed clay minerals like montmorillonite, adsorptiononto existing clay minerals, and precipitation into carbon-ates. Although Mg, Sr and K could be incorporated intodifferent clay minerals, the good linear relationship betweenMg and K, as well as between Sr and K in deep groundwater(Fig. 5), indicates that Mg and Sr are mainly removed byneoformation of clay minerals, not by adsorption onto exist-ing clay minerals. Moreover, as pointed out in Section 2.2,degassing would inevitably lead to removal of Ca, Mg andSr by precipitation into carbonates.

Fig. 4. The relationship between d26Mg and 87Sr/86Sr ofgroundwater.

268 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

Therefore, it is inferred that calcite dissolution, clay for-mation, and carbonate precipitation might be the majorgeochemical processes controlling the variation of Mg iso-topic compositions from recharge to discharge areas. Inthe following sections, changes in the Mg isotopic composi-tions due to the three geochemical processes are quantita-tively analyzed.

4.2. Calcite dissolution

From recharge to discharge areas, due to the decreaseddissolved CO2 in a closed system, the degree of calcite dis-solution largely outweighs that of silicate weathering.Moreover, the increasing pressure from the shallow to thedeep of the aquifer leads to a higher solubility of calcite(Macdonald and North, 1974; Langmuir, 1997). Calcitedissolution can be expressed as follows:

CaCO3 +H2O+CO2 �Ca2þ +2HCO3� ð1Þ

Theoretically, calcite dissolution leads to increased pH,Ca and HCO3. In the current study, although pH increasesfrom recharge to discharge areas, there are decreasingtrends of Ca and HCO3 + CO3 due to their removal byother processes. In fact, the decreased 87Sr/86Sr ratios in

Fig. 5. Plots of Mg versus K (a) and Sr versus K

discharge areas reflect the contribution of calcite dissolu-tion during the circulation from recharge to dischargeareas.

To quantify the contribution of calcite dissolution to thehydrochemistry of deep groundwater in discharge areas,shallow groundwater in recharge areas and calcite are con-sidered to be two endmembers and it is assumed that thereis no fractionation during congruent dissolution of calcite.Because 87Sr/86Sr ratios of water would not be changedby carbonate precipitation and clay formation, it is reason-able to use the 87Sr/86Sr ratios of deep groundwater to iden-tify the contribution of calcite dissolution. The equationscan be written as follows:

87Sr= 86Sr� �

SG� f SGþ 87Sr= 86Sr

� �CAL

� f CAL ¼ 87Sr= 86Sr� �

DG

ð2Þf SG þ f CAL ¼ 1 ð3Þwhere (87Sr/86Sr)SG, (

87Sr/86Sr)CAL and (87Sr/86Sr)DG referto the 87Sr/86Sr ratios of shallow groundwater, calcite anddeep groundwater, respectively; fSG and fCAL representthe contribution of shallow groundwater and calcite,respectively. The 87Sr/86Sr ratio of shallow groundwater isdetermined by the mean of the 87Sr/86Sr ratios of the fivesamples. Due to the abundant silicate minerals and tracecalcite in the Cretaceous sandstone samples, the leachingexperiment failed to obtain the Mg content and 87Sr/86Srratio representative of calcite in the sandstones because ofthe inevitable dissolution of silicate minerals in HAc.According to Veizer et al. (1999), the 87Sr/86Sr ratios ofCretaceous carbonates range between 0.707 and 0.708,therefore, both 0.707 and 0.708 are used to represent(87Sr/86Sr)CAL. fCAL is found to range between 0.29 and0.48 when (87Sr/86Sr)CAL equals 0.708, and between 0.23and 0.37 when (87Sr/86Sr)CAL equals 0.707. Consideringthe uncertainty of (87Sr/86Sr)CAL, fCAL is assumed to rangebetween 0.20 and 0.50 in the following analysis.

The plot of d26Mg versus Ca/Mg is useful to identifymineralogical controls (calcite and silicate) on Mg isotopes(Galy et al, 2002; Fan et al., 2016), while the plot of d26Mgversus 1/Mg can be used to identify various geochemicalprocesses influencing Mg isotopes (Pogge von

(b) of deep groundwater in discharge areas.

Fig. 6. (a-b) The plots of d26Mg versus Ca/Mg (a) and versus 1/Mg (b) showing the effect of calcite dissolution; (c) The plot of d26Mg versus1/Mg showing the effect of clay formation on deep groundwater; (d) The plot of d26Mg versus 1/Mg showing the effect of calcite precipitationon G2 of deep groundwater. Data of CAL-S used in (a) and (b) is from Bolou-Bi et al. (2009).

H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274 269

Strandmann et al, 2008; Lee et al., 2014). Fig. 6a and bshows d26Mg versus Ca/Mg and 1/Mg of shallow ground-water, deep groundwater, and CRPG (Centre deRecherches Petrographiques et Geochimiques) referencelimestone, which is labeled as CAL-S (Yeghicheyan et al.,2003; Bolou-Bi et al., 2009). According to Bolou-Bi et al.(2009), the concentrations of Ca and Mg of CAL-S are400,000 and 2200 ppm, respectively, while the d26Mg valuesof CAL-S equals �4.38‰. The d26Mg values of deepgroundwater fall between the d26Mg values of shallowgroundwater and CAL-S. Based on the differences inCa/Mg as well as 1/Mg, and the location of groundwatersamples, the 12 deep groundwater samples can be dividedinto three groups. The first is D(G1) (including samplesD5 through D8) in the Dosit River Watershed, the secondis D(G2) (including samples D1through D4) in the DositRiver Watershed, and the third is D(G3) (including samplesD9 through D12) in the Wudu Lake Catchment.

If calcite dissolution is the only process influencing thed26Mg values and contents of Ca and Mg, deep groundwa-ter samples should lie between the mixing lines. Accordingto the fCAL determined by 87Sr/86Sr ratios, deep groundwa-ter samples should be bounded by the two mixing lines andthe two lines with fCAL equaling 20% and 50% (Fig. 6a andb). Unfortunately, all deep groundwater samples are foundto be outside the possible ranges of d26Mg versus Ca/Mg,and d26Mg versus 1/Mg. This indicates that in addition tocalcite dissolution, there are other geochemical processes

influencing the Mg isotopes, as well as Ca and Mg alongthe flowpaths from recharge to discharge areas.

4.3. Clay formation

In the sandstone aquifer, pH of groundwater in rechargeareas varies between 7.98 and 8.16, while that in dischargeareas ranges between 8.32 and 9.44. The alkaline environ-ment, especially in the deep part, is beneficial for clay for-mation (Drever, 1997; Langmuir, 1997). As mentioned inSection 4.1, the removal of K should be mainly caused byclay formation, and the good linear relationship betweenK and Mg indicates that they are probably removed by asimilar process. Moreover, there are trends of decreasingAl and SiO2 from recharge to discharge areas, and mont-morillonite and illite are supersaturated in both rechargeand discharge areas. Therefore, it is reasonable to infer thatclay formation contributes to Mg removal in the sandstoneaquifer. Many studies have shown that clay formationcould lead to lower d26Mg values in solutions (Tipperet al., 2006; Tipper et al., 2008; Teng et al., 2010;Opfergelt et al., 2012; Wimpenny et al., 2014; Ryu et al.,2016). For deep groundwater in discharge areas, thedecreasing d26Mg values in G1 and G3 are accompaniedby decreasing Mg (Fig. 7), which is in accordance withthe effect of clay formation on Mg isotopes.

To quantify the effect of clay formation on Mg isotopesof deep groundwater, shallow groundwater in recharge

Fig. 7. The d26Mg versus Mg in deep groundwater showing theeffect of clay formation.

270 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

areas subjected to dissolution of more calcite, which isdenoted as F0 on the plot of d26Mg versus 1/Mg, is consid-ered as the initial condition of deep groundwater beforeclay formation (Fig. 6c). The sample D11 is within therange of F0, indicating that Mg in this sample had seldombeen incorporated into clay minerals. Based on the Ray-leigh calculation, the d26Mg values of deep groundwaterdue to clay formation can be obtained based on the follow-ing equation:

d26Mg ¼ d26MgF0 þ 1000 aclay-water � 1� �

ln RMg ð4Þwhere d26MgF0 is the d26Mg value of the initial conditionF0, RMg is the proportion of residual Mg in groundwaterafter clay formation, and aclay-water is the fractionation fac-tor between the neoformed clay minerals and water.

Teng et al. (2010) pointed out that the fractionation fac-tor (aclay-water) controlled by the neoformation of clay min-erals generally ranges between 1.00005 and 1.0004. Here,aclay-water = 1.0003 is chosen to calculate the theoreticalrelationship of d26Mg versus 1/Mg in deep groundwater(Fig. 6c). The possible ranges of d26Mg versus 1/Mg whenRMg equals 30% and 4%, which are the upper and lowerlimits of RMg for samples in G1 and G3, are labeled asF1 and F2, respectively. By accounting for a series ofRMg varying between 4% and 30%, the possible ranges ofd26Mg versus 1/Mg in G1 and G3 can be obtained, whichare bounded by the dotted lines shown in Fig. 6c. All deepgroundwater samples of G1 and G3 except for sample D11are inside the possible ranges, indicating that Mg isotopesof deep groundwater in G1 and G3 are mainly controlledby the simultaneous control of calcite dissolution and clayformation. Note that other values of aclay-water in the rangebetween 1.0003 and 1.0004 could also be used to explain thecontrol of clay formation on Mg contents and isotopes inthe study area.

Based on the relationship between Mg and K (Fig. 5a),it is inferred that G2 has the highest degree of clay forma-tion due to the lowest concentrations of K. By assumingthat the low concentrations of Mg in G2 are controlledby clay formation only, Fig. 6c shows the possible ranges

of d26Mg versus 1/Mg when RMg equals 0.7% (the upperlimit) and 0.2% (the lower limit), which are labeled as F3and F4, respectively. Unfortunately, d26Mg of all samplesin G2 are higher than the expected d26Mg by consideringcalcite dissolution and clay formation. Therefore, thereshould be other process(es) controlling the contents and iso-topic compositions of Mg.

4.4. Calcite precipitation

In the discharge areas of a deep basin, the upward move-ment of groundwater in the aquifer as well as in the flowingwells leads to a decreasing pressure of groundwater, accom-panying with escape of dissolved CO2 and precipitation ofcarbonates (Freeze and Cherry, 1979), the process of whichis called degassing or decarbonation (Toth,1999). As shownin Fig. 3, from recharge to discharge areas, there are simul-taneous decreases in Mg, Ca and HCO3 + CO3. Thedecreasing HCO3 + CO3 could be caused by precipitationof carbonates. To form 1 mol CaCO3 or MgCO3 duringdegassing, 2 mol HCO3 would be consumed. Consideringthe transformation among HCO3, CO3 and dissolvedCO2, the plots of half of dissolved inorganic carbon,DIC/2, versus Ca, and DIC/2 versus Ca + Mg, are usedto identify the causes of reductions in Mg, Ca and HCO3

+ CO3 from recharge to discharge areas. As shown inFig. 8, the decreased Ca is generally balanced by thedecreased DIC/2, but the decreased Ca + Mg is much lar-ger than the decreased DIC/2. Moreover, calcite has beenfound to be supersaturated in all groundwater samples.Therefore, it is reasonable to acknowledge that calcite isthe main precipitate due to degassing. In this process,although magnesite has been found to be undersaturated,it is inevitable that minor Mg would be precipitated(Saulnier et al., 2012; Mavromatis et al., 2013).

Because precipitated carbonates have low d26Mg, it iswidely acknowledged that precipitation of Mg-bearing cal-cite could result in higher d26Mg in the residual solutioncompared with the initial solution before precipitation.To verify whether the higher d26Mg values in G2 than theexpected values are caused by precipitation of low-Mg cal-cite, it is assumed that the incorporation of Mg into calcitefollows the Rayleigh fractionation law. The fractionationfactor between precipitated calcite and water, acalcite–water,can be calculated as

acalcite�water ¼ d26Mg� d26MgP01000 ln Mg� ln MgP0ð Þ þ 1 ð5Þ

where MgP0 and d26MgP0 are the Mg content and thed26Mg value, respectively, in the initial solution before pre-cipitation, which is denoted as P0. According to a reviewpaper by Saenger and Wang (2014), acalcite-water changesbetween 0.99919 and 0.99968. Based on this range, the ini-tial condition P0 corresponds to solutions with RMg varyingbetween 2.1% and 2.8%, i.e., 97.2–97.9% Mg had beenremoved by clay formation.

When RMg equals 2.1%, the range of P0, with an averageMg content equaling 0.019 mmol/L and an average d26Mgequaling �3.47‰, is shown in Fig. 6d. At this state, Cahas seldom been removed by precipitation yet, therefore,

Fig. 8. The plots of DIC/2 versus Ca (a) and versus Mg + Ca (b) showing the effect of precipitation.

Fig. 9. A schematic plot showing the main geochemical processescontrolling the isotopes and contents of Mg during groundwatercirculation from recharge to discharge areas.

H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274 271

the ratio of Ca/Mg could be very high. From P0 to the finalstate of G2, the ratios of Mg removal by precipitation arefound to be at least 66%, and the magnitudes of increasesin d26Mg are found to be at least 0.8‰. The ratios of Caremoval to Mg removal by precipitation are found to bein the range from 258 to 349 mol/mol. Therefore, the pre-cipitate during degassing can be considered as low-Mgcalcite.

In previous studies on the effect of calcite precipitationon Mg isotopes (Immenhauser et al., 2010; Mavromatiset al., 2013), the ratios of Mg precipitated into calcite tothe initial Mg in solution are low, therefore, the increasesin d26Mg are unobservable. For example, in Mavromatiset al. (2013), the ratios of Mg precipitated into calcite tothe initial Mg were less than 1.3%, which did not inducea measurable change in the Mg isotopic compositions ofthe solution. In the current study, the ratio of Mg precipi-tated into low-Mg calcite to the total Mg in the initial solu-tion is at least 66%, and could be as high as 93%, which isprobably the main reason causing the significant increasesof d26Mg in G2.

4.5. The control of residence time on clay formation

The extremely low content of Mg before precipitation inG2, which is due to its highest degree of clay formation,provides a great opportunity to directly observe the effectof precipitation on the increased d26Mg of groundwater.Therefore, G3, G1 and G2 have increasingly strongerdegrees of clay formation. According to previous studies,the process of clay formation is usually accompanied withlong residence times of water (Steefel and Van Cappellen,1990; Maher et al., 2004; Fantle and DePaolo, 2007;Maher, 2010; Tipper et al., 2012b). It is interesting to exam-ine whether the degree of clay formation could be related tothe travel distances and residence times from recharge todischarge areas.

The residence time of a groundwater sample since itsrecharge can be determined by a numerical model ofgroundwater flow (de Dreuzy and Ginn, 2016). Based onWang et al.’s (2016) three-dimensional groundwater flow

model of the Dosit River Watershed, the mean residencetime of G1 (samples D5 through D8) is around 14,000years, while that of G2 (samples D1 through D4) is around18,000 years. This is in accordance with the fact that G2 haslonger distances away from the divide of the Dosit RiverWatershed (Fig. 2). Although residence times of G3 (sam-ples D9 through D12) in the Wudu Lake Catchment arenot available, the lowest degree of clay formation is inaccordance with the nearest distance away from the divide.

The long residence times of deep groundwater in the dis-charge area of the Dosit River Watershed are supported bygroundwater age measured by 14C. The ages of groundwa-ter sampled in the middle (297–773 m) and the deep (777–954 m) sections of borehole B2, which is located severalkilometers to the northwest of well D1 (Fig. 2), were mea-sured to be around 20,000 years (Jiang et al., 2012).

5. CONCLUSIONS

This study examines the major controls on the lowd26Mg and low Mg contents of deep groundwater in

272 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

discharge areas of a sandstone aquifer, which are summa-rized in Fig. 9. The decreasing d26Mg from recharge to dis-charge areas are found to be caused by calcite dissolution,superposed by clay formation during groundwater circula-tion, which significantly removes Mg. The fractionationof Mg isotopes controlled by clay formation is largelydetermined by the residence time of groundwater. The aug-mentation of fractionation due to the long residence timesof water in various reservoirs deserves furtherinvestigations.

This study demonstrates that when Mg has been signif-icantly removed by clay formation, groundwater in thedeep part of the sandstone aquifer could have high Caand low Mg contents. During the upward movement ofdeep groundwater, low-Mg calcite with Ca/Mg ratios of258 and 349 mol/mol were precipitated, which furtherremoves Mg with a removal rate of >66%, and leads tohigher d26Mg than that before precipitation. The ratio ofCa/Mg and removal rate of Mg by precipitation identifiedin this study would be helpful to future experiments onthe effect of precipitation on Mg isotopes.

This study shows that the lower limit of d26Mg ofgroundwater could be down to �3.30‰, instead of being�1.63‰ as reported in previous studies. Therefore, theheterogeneity of Mg isotopic compositions of groundwateris much more significant than previously observed. Due tothe ubiquity of groundwater discharge to rivers, lakes andoceans, the mass balance of Mg isotopes in these water bod-ies should be reexamined in the future.

ACKNOWLEDGMENTS

This study is supported by the National Natural Science Foun-dation of China (41522205), the National Program for Support ofTop-notch Young Professionals, the Foundation for the Author ofNational Excellent Doctoral Dissertation (201457), and the Funda-mental Research Funds for the Central Universities of China. Wethank Prof. Gaojun Li for his assistance on Sr isotope measure-ments. We also thank E. T. Tipper and two anonymous reviewers,and the Associate Editor, Andrew Jacobson, for their constructivesuggestions.

REFERENCES

Aciego S. M., Bourdon B., Lupker M. and Rickli J. (2009) A newprocedure for separating and measuring radiogenic isotopes (U,Th, Pa, Ra, Sr, Nd, Hf) in ice cores. Chem. Geol. 266, 194–204.

Appelo C. A. J. and Postma D. (2005) Geochemistry, Groundwater

and Pollution, second ed. Balkema Publisher, Leiden.Berner E. K. and Berner R. A. (1996) Global Environment: Water,

Air, and Geochemical Cycles. Prentice-Hall, Englewood Cliffs,New Jersey.

Bolou-Bi E. B., Vigier N., Brenot A. and Poszwa A. (2009)Magnesium isotope compositions of natural reference materi-als. Geostand. Geoanal. Res. 33, 95–109.

Bolou-Bi E. B., Vigier N., Poszwa A., Boudot J. P. and DambrineE. (2012) Effects of biogeochemical processes on magnesiumisotope variations in a forested catchment in the VosgesMountains (France). Geochim. Cosmochim. Acta 87, 341–355.

Brenot A., Cloquet C., Vigier N., Carignan J. and France-LanordC. (2008) Magnesium isotope systematics of the lithologically

varied Moselle river basin, France. Geochim. Cosmochim. Acta

72, 5070–5089.Buhl D., Immenhauser A., Smeulders G., Kabiri L. and Richter D.

K. (2007) Time series d26Mg analysis in speleothem calcite:Kinetic versus equilibrium fractionation, comparison withother proxies and implications for palaeoclimate research.Chem. Geol. 244, 715–729.

Clark I. (2015) Groundwater Geochemistry and Isotopes. CRCPress, Boca Raton, Florida.

de Dreuzy J. R. and Ginn T. R. (2016) Residence times insubsurface hydrological systems, introduction to the SpecialIssue. J. Hydrol. 543, 1–6.

Delany J. M. and Lundeen S. R. (1990) The LLNL thermochemicaldatabase. Lawrence Livermore National Laboratory Report(UCRL-21658).

Drever J. I. (1997) The Geochemistry of Natural Waters, third ed.Prentice-Hall, Upper Saddle River, NJ.

Edmunds W. M. and Shand P. (2008) Natural Groundwater

Quality. Blackwell Publishing, Oxford, UK.Fan B. L., Zhao Z. Q., Tao F., Li X., Tao Z., Gao S. and He M.

(2016) The geochemical behavior of Mg isotopes in theHuanghe basin, China. Chem. Geol. 426, 19–27.

Fantle M. S. and DePaolo D. J. (2007) Ca isotopes in carbonatesediment and pore fluid from ODP Site 807A: the Ca2+(aq)-calcite equilibrium fractionation factor and calcite recrystal-lization rates in Pleistocene sediments. Geochim. Cosmochim.

Acta 71, 2524–2546.Freeze R. A. and Cherry J. A. (1979) Groundwater. Prentice-Hall,

Upper Saddle River, NJ.Galy A., Bar-Matthews M., Halicz L. and O’Nions R. K. (2002)

Mg isotopic composition of carbonate: insight from speleothemformation. Earth Planet. Sci. Lett. 201, 105–115.

Geske A., Lokier S., Dietzel M., Richter D. K., Buhl D. andImmenhauser A. (2015) Magnesium isotope composition ofsabkha porewater and related (Sub-) recent stoichiometricdolomites, Abu Dhabi (UAE). Chem. Geol. 393–394, 112–124.

Hippler D., Buhl D., Witbaard R., Richter D. K. and Immen-hauser A. (2009) Towards a better understanding of magne-sium-isotope ratios from marine skeletal carbonates. Geochim.

Cosmochim. Acta 73, 6134–6146.Holland H. D. (1978) The Chemistry of the Atmosphere and Oceans.

Wiley, New York.Hou G. C., Liang Y., Su X., Zhao Z., Tao Z., Yin L., Yang Y. and

Wang X. (2008) Groundwater systems and resources in theOrdos Basin, China. Acta Geol. Sin. 82, 1061–1069.

Huang F., Glessner J., Ianno A., Lundstrom C. and Zhang Z.(2009) Magnesium isotopic composition of igneous rockstandards measured by MC-ICP-MS. Chem. Geol. 268, 15–23.

Huang K. J., Teng F. Z., Wei G. J., Ma J. L. and Bao Z. Y. (2012)Adsorption- and desorption-controlled magnesium isotopefractionation during extreme weathering of basalt in HainanIsland, China. Earth Planet. Sci. Lett. 359–360, 73–83.

Immenhauser A., Buhl D., Richter D., Niedermayr A., Riechel-mann D., Dietzel M. and Schulte U. (2010) Magnesium-isotopefractionation during low-Mg calcite precipitation in a limestonecave – field study and experiments. Geochim. Cosmochim. Acta

74, 4346–4364.Ingebritsen S., Sanford W. and Neuzil C. (2006) Groundwater in

Geologic Processes, second ed. Cambridge Univ. Press, Cam-bridge, UK.

Jacobson A. D., Zhang Z., Lundstrom C. and Huang F. (2010)Behavior ofMg isotopes during dedolomitization in theMadisonAquifer, South Dakota. Earth Planet. Sci. Lett. 297, 446–452.

Jiang X. W., Wan L., Ge S., Cao G. L., Hou G. C., Hu F. S., WangX. S., Li H. and Liang S. H. (2012) A quantitative study on

H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274 273

accumulation of age mass around stagnation points in nestedflow systems. Water Resour. Res. 48, W12502.

Jiang X. W., Wan L., Wang J. Z., Yin B. X., Fu W. X. and Lin C.H. (2014) Field identification of groundwater flow systems andhydraulic traps in drainage basins using a geophysical method.Geophys. Res. Lett. 41, 2812–2819.

Jiang X. W., Sun Z. C., Zhao K. Y., Shi F. S., Wan L., Wang X. S.and Shi Z. M. (2017) A method for simultaneous estimation ofgroundwater evapotranspiration and inflow rates in the dis-charge area using seasonal water table fluctuations. J. Hydrol.

548, 498–507.Jiang X. W., Wan L., Wang X. S., Wang D., Wang H., Wang J. Z.,

Zhang H., Zhang Z. Y. and Zhao K. Y. (2018) A multi-methodstudy of regional groundwater circulation in the Ordos Plateau,NW China. Hydrogeol. J. 2, 1–12.

Ke S., Teng F. Z., Li S., Gao T., Liu S. A., He Y. and Mo X. (2016)Mg, Sr, and O isotope geochemistry of syenites from northwestXinjiang, China: tracing carbonate recycling during Tethyanoceanic subduction. Chem. Geol. 437, 109–119.

Langmuir D. (1997) Aqueous Environmental Geochemistry. Pren-tice-Hall, Upper Saddle River, NJ.

Lee S. W., Ryu J. S. and Lee K. S. (2014) Magnesium isotopegeochemistry in the Han River, South Korea. Chem. Geol. 364,9–19.

Liu S. A., Teng F. Z., He Y., Ke S. and Li S. (2010) Investigationof magnesium isotope fractionation during granite differentia-tion: implication for Mg isotopic composition of the continen-tal crust. Earth Planet. Sci. Lett. 297, 646–654.

Liu Y., Zhu Z. M., Chen C., Zhang S. P., Sun X., Yang Z. S. andLiang W. (2015) Geochemical and mineralogical characteristicsof weathered ore in the Dalucao REE deposit, Mianning-Dechang REE Belt, western Sichuan Province, southwesternChina. Ore Geol. Rev. 71, 437–456.

Ma L., Teng F. Z., Jin L., Ke S., Yang W., Gu H. O. and BrantleyS. L. (2015) Magnesium isotope fractionation during shaleweathering in the Shale Hills Critical Zone Observatory:accumulation of light Mg isotopes in soils by clay mineraltransformation. Chem. Geol. 397, 37–50.

MacDonald R. W. and North N. A. (1974) The effect of pressureon the solubility of CaCO, CaF and SrSO in water. Can. J.Chem. 52, 3181–3186.

Maher K. (2010) The dependence of chemical weathering on fluidresidence time. Earth Planet. Sci. Lett. 294, 101–110.

Maher K., DePaolo D. J. and Lin J. C. L. (2004) Rates of silicatedissolution in deep-sea sediment: in situ measurement using234U/238U of pore fluids. Geochim. Cosmochim. Acta 68, 4629–4648.

Mavromatis V., Gautier Q., Bosc O. and Schott J. (2013) Kineticsof Mg partition and Mg stable isotope fractionation during itsincorporation in calcite. Geochim. Cosmochim. Acta 114, 188–203.

Mavromatis V., Rinder T., Prokushkin A. S., Pokrovsky O. S.,Korets M. A., Chmeleff J. and Oelkers E. H. (2016) The effectof permafrost, vegetation, and lithology on Mg and Si isotopecomposition of the Yenisey River and its tributaries at the endof the spring flood. Geochim. Cosmochim. Acta 191, 32–46.

Opfergelt S., Georg R. B., Delvaux B., Cabidoche Y. M., BurtonK. W. and Halliday A. N. (2012) Mechanisms of magnesiumisotope fractionation in volcanic soil weathering sequences,Guadeloupe. Earth Planet. Sci. Lett. 341–344, 176–185.

Parkhurst D. L. and Applo C.A.J. (1999) User’s guide toPHREEQC (version 2) - A computer program for speciation,batch-reaction, one-dimensional transport, and inverse geo-chemical calculations. U.S. G. S. Wat Res. Inv. Report. 99-4259.

Pogge von Strandmann P. A. E., Burton K. W., James R. H., vanCalsteren P., Gislason S. R. and Sigfusson B. (2008) Theinfluence of weathering processes on riverine magnesiumisotopes in a basaltic terrain. Earth Planet. Sci. Lett. 276,187–197.

Pogge von Strandmann P. A. E., Opfergelt S., Lai Y.-J., SigfussonB., Gislason S. R. and Burton K. W. (2012) Lithium,magnesium and silicon isotope behaviour accompanyingweathering in a basaltic soil and pore water profile in Iceland.Earth Planet. Sci. Lett. 339–340, 11–23.

Rao W., Han G., Tan H. and Jiang S. (2015) Chemical and Srisotopic compositions of rainwater on the Ordos DesertPlateau, Northwest China. Environ. Earth Sci. 74, 5759–5771.

Roy S., Gaillardet J. and Allegre C. J. (1999) Geochemistry ofdissolved and suspended loads of the Seine River, France:anthropogenic impact, carbonate and silicate weathering.Geochim. Cosmochim. Acta. 63, 1277–1292.

Ryu J. S., Vigier N., Decarreau A., Lee S. W., Lee K. S., Song H.and Petit S. (2016) Experimental investigation of Mg isotopefractionation during mineral dissolution and clay formation.Chem. Geol. 445, 135–145.

Saenger C. and Wang Z. (2014) Magnesium isotope fractionationin biogenic and abiogenic carbonates: implications for pale-oenvironmental proxies. Quaternary Sci. Rev. 90, 1–21.

Saulnier S., Rollion-Bard C., Vigier N. and Chaussidon M. (2012)Mg isotope fractionation during calcite precipitation: anexperimental study. Geochim. Cosmochim. Acta 91, 75–91.

Steefel C. I. and Van Cappellen P. (1990) A new kinetic approachto modeling water–rock interaction—the role of nucleation,precursors, and Ostwald ripening. Geochim. Cosmochim. Acta

54, 2657–2677.Su X., Wu C., Dong W. and Hou G. C. (2011) Strontium isotope

evolution mechanism of the Cretaceous groundwater in OrdosDesert Plateau. J. Chengdu Univ. Technol. 38, 348–358 (inChinese with English Abstract).

Suchet P. A. and Probst J. L. (1993) Modelling of atmospheric CO2

consumption by chemical weathering of rocks: application tothe Garonne, Congo and Amazon basins. Chem. Geol. 107,205–210.

Teng F. Z., Wadhwa M. and Helz R. T. (2007) Investigation ofmagnesium isotope fractionation during basalt differentiation:implications for a chondritic composition of the terrestrialmantle. Earth Planet. Sci. Lett. 261, 84–92.

Teng F. Z., Li W. Y., Rudnick R. L. and Gardner L. R. (2010)Contrasting lithium and magnesium isotope fractionationduring continental weathering. Earth Planet. Sci. Lett. 300,63–71.

Teng F. Z., Li W. Y., Ke S., Yang W., Liu S. A., Sedaghatpour F.,Wang S. J., Huang K. J., Hu Y., Ling M. X., Xiao Y., Liu X.M., Li X. W., Gu H. O., Sio C. K., Wallace D. A., Su B. X.,Zhao L., Chamberlin J., Harrington M. and Brewer A. (2015)Magnesium isotopic compositions of international geologicalreference materials. Geostand. Geoanal. Res. 39, 329–339.

Tipper E. T., Galy A., Gaillardet J., Bickle M., Elderfield H. andCarder E. (2006) The magnesium isotope budget of the modernocean: constraints from riverine magnesium isotope ratios.Earth Planet. Sci. Lett. 250, 241–253.

Tipper E. T., Galy A. and Bickle M. J. (2008) Calcium andmagnesium isotope systematics in rivers draining the Himalaya-Tibetan-Plateau region: Lithological or fractionation control?Geochim. Cosmochim. Acta 72, 1057–1075.

Tipper E. T., Gaillardet J., Louvat P., Capmas F. and White A. F.(2010) Mg isotope constraints on soil pore-fluid chemistry:evidence from Santa Cruz, California. Geochim. Cosmochim.

Acta 74, 3883–3896.

274 H. Zhang et al. /Geochimica et Cosmochimica Acta 237 (2018) 261–274

Tipper E. T., Calmels D., Gaillardet J., Louvat P., Capmas F. andDubacq B. (2012a) Positive correlation between Li and Mgisotope ratios in the river waters of the Mackenzie Basinchallenges the interpretation of apparent isotopic fractionationduring weathering. Earth Planet. Sci. Lett. 333–334, 35–45.

Tipper E. T., Lemarchand E., Hindshaw R. S., Reynolds B. C. andBourdon B. (2012b) Seasonal sensitivity of weathering pro-cesses: hints from magnesium isotopes in a glacial stream.Chem. Geol. 312–313, 80–92.

Toth J. (1999) Groundwater as a geologic agent: an overview of thecauses, processes, and manifestations. Hydrogeol. J. 7, 1–14.

Veizer J., Ala D., Azmy K., Bruckschen P., Buhl D., Bruhn F.,Carden G. A. F., Diener A., Ebneth S., Godderis Y., Jasper T.,Korte C., Pawellek F., Podlaha O. G. and Strauss H. (1999)87Sr/86Sr, d13C and d18O evolution of Phanerozoic seawater.Chem. Geol. 161, 59–88.

Wang H., Jiang X. W., Wan L., Han G. and Guo H. (2015a)Hydrogeochemical characterization of groundwater flow sys-tems in the discharge area of a river basin. J. Hydrol. 527, 433–441.

Wang J. Z., Jiang X. W., Wan L., Worman A., Wang H., Wang X.S. and Li H. (2015b) An analytical study on artesian flowconditions in unconfined-aquifer drainage basins. Water

Resour. Res. 51, 8658–8667.Wang J. Z., Worman A., Bresciani E., Wan L., Wang X. S. and

Jiang X. W. (2016) On the use of late-time peaks of residencetime distributions for the characterization of hierarchicallynested groundwater flow systems. J. Hydrol. 543, 47–58.

Williams E. L., Szramek K. J., Jin L., Ku T. C. W. and Walter L.M. (2007) The carbonate system geochemistry of shallow

groundwater-surface water systems in temperate glaciatedwatersheds (Michigan, USA): significance of open-systemdolomite weathering. Geol. Soc. Am. Bull. 119, 515–528.

Wimpenny J., Burton K. W., James R. H., Gannoun A., MokademF. and Gıslason S. R. (2011) The behaviour of magnesium andits isotopes during glacial weathering in an ancient shieldterrain in West Greenland. Earth Planet. Sci. Lett. 304, 260–269.

Wimpenny J., Colla C. A., Yin Q. Z., Rustad J. R. and Casey W.H. (2014) Investigating the behaviour of Mg isotopes during theformation of clay minerals. Geochim. Cosmochim. Acta 128,178–194.

Yeghicheyan D., Carignan J., Valladon M., Bouhnik Le Coz M.,Samuel J., BenBakkar M., Bruguier O., Keller F., Pin C.,Pourtales L., Henin O., Mace J., Morin N., Guilmette C. andMarin L. (2003) The new carbonate reference material Cal-S:preliminary results. Abs. Geoanal. 2003, 146.

Young E. D. and Galy A. (2004) The isotope geochemistry andcosmochemistry of magnesium. Rev. Mineral. Geochem. 55,197–230.

Zhang W., Chen J., Ji J. and Li G. (2016) Evolving flux of Asiandust in the North Pacific Ocean since the late Oligocene.Aeolian Res. 23, 11–20.

Zhang Z. Y., Jiang X. W., Wang X. S., Wan L. and Wang J. Z.(2018) A numerical study on the occurrence of flowing wells inthe discharge area of basins due to the upward hydraulicgradient induced wellbore flow. Hydrol. Processes. 32, 421–435.

Associate editor: Andrew D. Jacobson