UNDERSTANDING REACTIVE TRANSPORT OF MARCELLUS SHALE …

The Pennsylvania State University

The Graduate School

Department of Energy and Mineral Engineering

UNDERSTANDING REACTIVE TRANSPORT OF MARCELLUS

SHALE WATERS IN AQUIFERS

A Dissertation in

Energy and Mineral Engineering

by

Zhang Cai

2018 Zhang Cai

Submitted in Partial Fulfillment

of the Requirements

for the Degree of

Doctor of Philosophy

August 2018

ii

The dissertation of Zhang Cai was reviewed and approved* by the following:

Li Li

Associate Professor of Civil & Environmental Engineering

Dissertation Advisor

Chair of Committee

Jeremy M. Gernand

Associate Professor of Mineral Processing and Geo-Environmental Engineering

Hamid Emami-Meybodi

Assistant Professor of Petroleum and Natural Gas Engineering

Nathaniel R Warner

Assistant Professor of Civil & Environmental Engineering

Luis F. Ayala H.

Professor of Petroleum and Natural Gas Engineering

Head of the Department of Department or Graduate Program

*Signatures are on file in the Graduate School

iii

ABSTRACT Flowback and produced waters from Marcellus Shale gas extraction (MSWs)

typically contain high levels of salinity and pollutants including trace metals, which raise

public concerns on drinking water quality. Extensive studies have focused on evidences

linking the potential water contamination to the shale gas development and the interactions

of MSWs with minerals and different types of waters in batch systems. However, the

natural attenuation and reactive transport of MSW chemicals in natural aquifers remains

elusive due to the facing challenges: (i) the different time scales and magnitude of MSW

release under various receiving water conditions, (ii) the complex aquifers composed of

multiple minerals with differing reactivity, and (iii) the ubiquitous occurrence of spatial

heterogeneity in natural subsurface.

Numerical experiments indicates that in clay-rich sandstone aquifers, ion exchange

plays a key role in determining the maximum concentration and the time scale of released

cations in receiving natural waters. In contrast, mineral dissolution/precipitation play a

minor role. The relative time scales of recovery rr, a dimensionless number defined as the

ratio of the time needed to return to background concentrations over the residence time of

natural waters, vary between 5-10 for Na, Ca, and Mg, and between 10-20 for Sr and Ba.

In rivers and sand and gravel aquifers with negligible clay content, rr values are close to 1

because cations are flushed out at ~ 1 residence time. These values can be used as first

order estimates of time scales of released MSWs in natural water systems.

Mineralogy regulates the types of reactions that occur and the extent of solute

immobilization from MSW release. In the clay-rich column, trace metals are retarded by

ion exchange but also are retained via mineral precipitation (~50-90%). In the calcite-rich

iv

column, trace metals are retained through precipitation and solid solution partitioning by

75-99%. In the quartz column, the trace metals are retained the least.

As to spatial heterogeneity, we set up two two-dimensional heterogeneous cells with

the same vermiculite-to-quartz mass ratio but different spatial patterns as compared to a

“Uniform” column: the “1/4-zone” and “1/2-zone” cells have rectangular vermiculite

clusters at a quarter and a half lengths of the cells, respectively, and the “Uniform” column

has uniformly distributed vermiculite and quartz. Spatial heterogeneity regulates not only

the extent, but also the dominant types of clay-MSW interactions. In comparison to

Uniform media, heterogeneous media minimizes the vermiculite-MSW interaction with

the decrease of trace metal (Mn, Cu, Zn, Pb, Cd) immobilization by 1-2 orders of

magnitude. This implies the higher risk on drinking water quality in natural heterogeneous

aquifers. Consequently, this study has significant implications on predicting natural

attenuation and reactive transport of complex contaminants from MSW release in the

natural subsurface.

v

TABLE OF CONTENTS

LIST OF FIGURES ....................................................................................................... VIII

LIST OF TABLES ........................................................................................................... XII

ACKNOWLEDGEMENTS ........................................................................................... XIII

CHAPTER 1 INTRODUCTION ........................................................................................ 1

1.1. BACKGROUND AND MOTIVATION ..................................................... 2

1.2. OBJECTIVES .............................................................................................. 7

1.3. DISSERTATION STRUCTURE ................................................................. 8

CHAPTER 2 HOW LONG DO NATURAL WATERS “REMEMBER” RELEASE

INCIDENTS OF MARCELLUS SHALE WATERS: A FIRST ORDER

APPROXIMATION USING REACTIVE TRANSPORT MODELING ............ 10

ABSTRACT ...................................................................................................... 11

2.1. INTRODUCTION ...................................................................................... 12

2.2. METHODS................................................................................................. 15

2.2.1. Problem setup ...................................................................................... 15

2.2.2. Properties of natural waters and MSWs .............................................. 16

2.2.3. Characteristics of Marcellus Shale water release incident .................. 19

2.2.4. Reactive transport modeling ................................................................ 21

2.2.5. Quantification of release impacts ........................................................ 26

2.3. RESULTS AND DISCUSSION ................................................................ 27

2.3.1. Controlling processes in the sandstone aquifer ................................... 28

2.3.2 Effect of release characteristics in the sandstone aquifer ..................... 37

2.3.3 Effect of receiving water bodies ........................................................... 40

2.3.4 Impacts of the release incidents ............................................................ 43

2.3.5 Discussion ............................................................................................. 44

2.4. CONCLUSIONS ........................................................................................ 48

CHAPTER 3 MINERALOGY CONTROL ON REACTIVE TRANSPORT OF

MARCELLUS SHALE WATERS ...................................................................... 50

ABSTRACT ...................................................................................................... 51

3.1. INTRODUCTION ...................................................................................... 52

vi

3.2. MATERIALS AND METHODS ............................................................... 55

3.2.1. Mineral preparation ............................................................................. 55

3.2.2. Mineralogical composition and column property measurement ......... 56

3.2.3. Water composition ............................................................................... 57

3.2.4. Flow-through experiments ................................................................... 60

3.2.5. Reactive Transport Modeling (RTM) .................................................. 60

3.2.6. Quantification of injection and outlet mass ......................................... 63


3.3.1. Difference in column physical properties ............................................ 64

3.3.2. Temporal evolution of pH ................................................................... 65

3.3.3. Reactive transport of trace metals in columns ..................................... 67

3.3.4. Reactive transport of Ba, Sr and SO4 .................................................. 71

3.3.5. Reactive transport of Na, Ca, Mg, and K in columns .......................... 73

3.3.6. Chemical retention in columns ............................................................ 74

3.3.7. Discussion ............................................................................................ 76

3.4. CONCLUSIONS ........................................................................................ 77

CHAPTER 4 CONTROLS OF MINERAL SPATIAL PATTERNS ON THE REACTIVE

TRANSPORT OF MARCELLUS SHALE WATERS ....................................... 81

ABSTRACT ...................................................................................................... 82

4.1. INTRODUCTION ...................................................................................... 82

4.2. MATERIALS AND METHODS ............................................................... 85

4.2.1. Mineral preparation ............................................................................. 85

4.2.2. Two-dimensional cell design ............................................................... 86

4.2.3. Spatial distribution patterns and cell property measurement............... 87

4.2.4. Water composition ............................................................................... 89

4.2.5. Flow-through experiments ................................................................... 90

4.2.6. Chemical analysis ................................................................................ 90

4.2.7. Quantification of inlet and outlet mass ................................................ 90


4.3.1. Physical property differences .............................................................. 92

vii

4.3.2. Temporal evolution of pH ................................................................... 92

4.3.3. Reactive transport of trace metals ....................................................... 93

4.3.4. Reactive transport of Ba, Sr and SO4 in three cases ............................ 96

4.3.5. Reactive transport of Na, Ca, Mg, and K ............................................ 97

4.3.6. Mass balance of chemicals in three cases ............................................ 98

4.4. CONCLUSIONS ...................................................................................... 100

CHAPTER 5 CONCLUSIONS AND FUTURE WORK ............................................... 104

5.1. TIME SCALES AND MAGNITUDE OF MSW RELEASE UNDER

VARIOUS NATURAL WATERS.............................................................................. 105

5.2. MINERALOGY ....................................................................................... 106

5.3. SPATIAL HETEROGENEITY ............................................................... 107

5.4. FUTURE WORK ..................................................................................... 108

REFERENCE .................................................................................................................. 114

APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 3 ........................... 126

APPENDIX B DATA ARTICLE FOR CHAPTER 3 .................................................... 134

APPENDIX C SUPPORTING INFORMATION FOR CHAPTER 4 ........................... 149

APPENDIX D PERMISSION TO INCLUDE PUBLISHED PAPER IN THE THESIS

........................................................................................................................... 168

viii

LIST OF FIGURES

Figure 2. 1 (A) The numbers of Marcellus Shale water release accidents in

Pennsylvania from 2005 to June 8, 2015, with 78% of spills occurred in

Northeastern PA. Red spot indicated the location of Bradford County. The yellow

numbers are the numbers of spills. (B) A schematic diagram of 1-Dimensional

modeling setup. We assume a release point where the Marcellus Shale waters are

introduced into the surface water (river) or groundwater (aquifers). The release

can occur through spills, discharge, leakage, seepage, among others. ............. 15

Figure 2. 2. Evolution at the release point for Br under four scenarios. All four color

lines overlap. The grey shaded zone represents the release period. Due to its non-

reactive nature, the inclusion of different processes does not affect their evolution.

........................................................................................................................... 28

Figure 2. 3. Evolution at the release point for (A) Ca (mg/L) in logarithmic scale, (B)

Ca on exchange sites (mol/g solid), (C) Mg (mg/L) in logarithmic scale, (D) Mg

on exchange sites (mol/g solid), (E)Na (mg/L) in logarithmic scale, (F) Na on

exchange sites (mol/g solid), (G) calcite reaction rate (mol/m2/s) (negative

indicates dissolution and positive values indicate precipitation), and (H) pH. Grey

line overlaps with the black line. ...................................................................... 30

Figure 2. 4. Evolution at the release point for (A) Ba in water (mg/L), (B) Ba on surface

(mol/g solid), (C) Sr (mg/L), (D) Sr on surface (mol/g solid). Ion exchange

controls concentrations of these species while mineral dissolution and

precipitation play a minor role. ......................................................................... 33

Figure 2. 5. Spatio-temporal evolution of Br concentration in the sandstone aquifer in

the MIX+DISS/PPT+IEX case on Days 11, 25, 27 and 160. Release starts on day

10 and ends on day 25. The other tracer Cl behaves the same as Br. ............... 34

Figure 2. 6. Spatio-temporal profiles of major species in the sandstone aquifer under

the MIX+DISS/PPT+IEX scenario on Days 11, 25, 27 and 160. Left Column is

for aqueous concentrations (mg/L); right column is for concentrations on solid

surface (mol/g solid). Rows from the top to bottom: Ca (A and B), Mg (C and D),

Na (E and F), Ba (G and H), and Sr (I and J). ................................................... 35

Figure 2. 7. Profiles of Br, Ca, Ca on solid surface, Na, Na on solid surface in the

sandstone aquifer during release (left column) and after release (right column)

under the three release cases. The High, Medium, and Low release rates are

1.11×10-7 m3/s for 15 days, 5.55×10-8 m3/s for 30 days, and 1.11×10-8 m3/s for

150 days, respectively. The “During Release” curves are on day 10 after the

release starts. The “after Release” curves are on day 5 after release stops. ...... 39

Figure 2. 8. Profiles of Br, Ca, Ca on solid surface, Na, Na on solid surface during

release (left column) and after release (right column) in the sandstone aquifer,

sand and gravel aquifer, and river, respectively. The release rate is 1.11×10-7 m3/s

for 15 days. The “During Release” is on day 10 after the release starts. The “after

Release” is on day 5after release stops. ............................................................ 41

Figure 2. 9. The memory index of natural waters: Cmax and (A) recovery and (B)rr of

major species in the river (filled squares), SG aquifer (filled triangles), and S

ix

aquifer with high release (filled circles), medium release (crossed circles), and

low release rates (open circles). Both are calculated from the modeling output of

spatio-temporal concentration evolution. The Cmax is determined as the maximum

aqueous concentration during release. The recovery is the time scale for each

species to return to within 5% difference from its background concentrations in

natural waters. The relative recovery time rr, calculated as the ratio of recovery

over r, is a measure of the time scale that natural waters remember the incident

relative to their residence time. Each species is represented by one color, with

dashed line of the same color being their drinking water standard. In S aquifer

with abundant clay, rr values depend on cation affinity to solid surface with rr

between 5-10 for Na, Ca, and Mg, and 15-20 for Sr and Ba. ........................... 43

Figure 3. 1. Bromide breakthrough curves (BTCs) for Qtz (blue), Cal (green), and Vrm

(red) from experiments (dots) and from simulations (lines). The BTC of the Vrm

column is much wider than the other two columns, indicating a more

heterogeneous column than the other two due to the large contrast in grain size

and property between quartz (350-420 um) and vermiculite (75-150 um). ...... 64

Figure 3. 2. Temporal evolution of inlet (dash lines) and outlet (dots) pH in (A) Qtz

(blue), Cal (green) and Vrm (red) columns before, during, and after a MSW

release for about 0.48 residence times; Although the inlet pH in groundwater was

~ 8.2, the outlet pH varied significantly due to different reactions in different

columns. The outlet pH decreases in the Qtz column while increases in the Cal

and Vrm columns. The Qtz and Vrm columns “recover” quickly from the MSW

perturbation compared to the Cal column. (B) modeling output (lines) for Vrm

column under three cases with different processes in the model: PPT for mineral

dissolution/precipitation only, IEX for ion exchange only, and IEX+PPT for ion

exchange with mineral dissolution/precipitation. The IEX+PPT line overlaps with

the IEX line, indicating the dominant role of IEX in determining pH in the Vrm

column. .............................................................................................................. 65

Figure 3. 3. Left: Breakthrough data (dots) and modeling output (lines) of metals in

Qtz (blue), Cal (green), Vrm (red) columns; right: comparison of modeling output

in the Vrm column under three scenarios (including mineral

dissolution/precipitation (PPT only), ion exchange without mineral

dissolution/precipitation (IEX only), and ion exchange with mineral

dissolution/precipitation (IEX+PPT)). The comparison indicates that both ion

exchange and mineral precipitation contribute to the decrease of metals and their

retention within the column. Only a fraction of metal ions return back to the

solution. ............................................................................................................. 67

Figure 3. 4. Left: Breakthrough data (dots) and model output (lines) of (A) SO4, (B)

Ba, and (C) Sr experimental data with the right: comparison of three cases with

different process scenarios in the Vrm column (D) SO4, (E) Ba, and (F) Sr. In the

Vrm column, sulfate remains the same as inlet, indicating that barite and celestite

precipitation do not occur and Ba and Sr are exchanged onto vermiculite, which

gradually release out later over a long period of time. In the Cal and Qtz columns,

x

sulfate concentration decreases sharply during MSW release, indicating the

precipitation of sulfate-containing minerals. .................................................... 71

Figure 3. 5. Breakthrough data (dots) and modeling output (lines) of (A) Na, (B) Ca,

(C) Mg, and (D) K in Qtz (blue), Cal (green), Vrm (red) columns. Presorbed Mg

and K are ion exchanged out from the clay so their concentrations increase. After

MSW release, sorbed Na is slowly released back to the aqueous leading to a long

tail...................................................................................................................... 73

Figure 3. 6. Injected and outlet mass of species among Qtz (blue), Cal (green) and Vrm

(red) columns on logarithmic scale (A) Trace metals (Mn, Cu, Zn and Pb); (B)

Anions and cations (Br, Cl, Na, Ca, Mg, K, Ba, Sr, and SO4). ......................... 74

Figure 4. 1. Conceptual figure of deformed MSW plume and preferential flow path

with different mineral reactions in natural heterogeneous aquifer. The

complexities of aquifer may affect the ultimate reactive transport of chemicals

upon the MSW release. ..................................................................................... 84

Figure 4. 2. (A) A schematic of 2D cell of 40.0 cm×12.0 cm×1.0 cm (1/2-zone), with

2 zones of clay (dark brown) embedded within quartz sand (light brown). Glass

beads and honeycomb were positioned at the bottom of the cell to generate

homogeneous flow at the entry point. The flow however did segregate within the

cell due to the uneven distribution of clay and quartz. (B) A picture of the flow-

through experiments. The background groundwater was injected to pre-

equilibrate with minerals for about 6.0 residence times before and after the

injection of MSW pulse. ................................................................................... 86

Figure 4. 3. Temporal evolution of inlet (dash lines) and outlet (dots with connected

lines) (A) Br and (B) pH in the Uniform (blue), 1/4-zone (green) and 1/2-zone

(red) cases before and after a MSW release between 0 and 0.50 residence times.

The C0 represents the inlet concentrations during the MSWs leakage. Br in the

Uniform column has the shortest breakthrough tail compared to the other two

heterogeneous cells. Although the inlet pH was managed to be around 8.13, outlet

pH and Br vary significantly due to different vermiculite spatial patterns and

different extent of mineral-water interactions. Values of outlet pH are higher than

inlet pH in the Uniform column and are lower than the inlet pH in the 1/4-zone

and 1/2-zone cells. In the Uniform column, pH returns to the pre-injection

condition faster than in the other two heterogeneous cells. .............................. 91

Figure 4. 4 Breakthrough curves of (A) Zn, (B) Pb, (C) Cu, and (D) Mn (dots with

connected lines) in the three cases. The three solid light lines are Br BTCs for

comparison. Cd was also measured but not shown here. Gray dash line represents

the inlet. Trace metals have the lowest peaks and are retained the most in the

Uniform column compared to the other two heterogeneous cells. ................... 93

Figure 4. 5. Breakthrough curves of (A) SO4, (B) Ba, and (C) Sr from different cases.

In the Uniform column, SO4 concentrations remain similar to the inlet, indicating

negligible precipitation of sulfate-containing minerals (barite and celestite). Ba

and Sr were exchanged on vermiculite early and released out later, as indicated

by the late time increase in the Uniform column. In the 1/4-zone and 1/2-zone

xi

cells, sulfate concentration decreased sharply during MSW release, indicating the

precipitation of sulfate-containing minerals. .................................................... 96

Figure 4. 6. Breakthrough data of (A) Na, (B) Ca, (C) Mg, and (D) K in the Uniform,

1/4-zone and 1/2-zone cases. The extent of Mg and K increase vary among the

three cases. In the Uniform column, pre-sorbed Mg and K are ion exchanged out

the most so their concentration peaks are the highest and their mass increase by 7

to 10 times compared to the 1/4-zone and 1/2-zonecells. Based on mass balance

calculation, almost all sorbed Na is released back to the water phase within 25

residence times in the Uniform column, while it is still retained in the other two

heterogeneous cells. .......................................................................................... 97

Figure 4. 7. Inlet and outlet mass of chemical species among the Uniform (blue), 1/4-

zone (green), and 1/2-zone (red) cases on logarithmic scale (A) Trace metals (Mn,

Cu, Zn, Pb and Cd); (B) Anions and cations (Br, Cl, Na, Ca, Mg, K, Ba, Sr and

SO4). The inlet and outlet mass in the Uniform column is proportionally scaled

down. The retention of trace metals is maximized in the Uniform column while

minimized in the 1/4-zone and 1/2-zone cells. The reaction extents are maximized

therefore leading to largest increase of Mg and K, and largest decrease of Ba, Sr,

Ca, and Na in the Uniform column.

........................................................................................................................... 10

0

xii

LIST OF TABLES

Table 2. 1 Mineral composition and flow velocity in the natural waters ................. 17 Table 2. 2 Composition of natural waters and Marcellus Shale water (mg/L) ........ 19 Table 2. 3 Reaction network, Reaction thermodynamics, and Kinetics for mineral-

water interactions .............................................................................................. 22 Table 2. 4 Simulation scenarios for Marcellus Shale water release ......................... 25 Table 2. 5 aCases with contamination detected during direct discharge of MSW into

rivers.................................................................................................................. 46

Table 3. 1 Physical and geochemical properties of the columns ............................. 57 Table 3. 2 Compositions of background groundwater and Marcellus Shale waters

(mg/L) ............................................................................................................... 59 Table 3. 3 Calculated saturation index during the MSW release ............................. 69

Table 4. 1 Physical and geochemical properties of the heterogeneous cells and 1D

Uniform column ................................................................................................ 88 Table 4. 2 Compositions of groundwater and Marcellus Shale waters (mg/La) ..... 89 Table 4. 3 Saturation index of minerals during the MSW release ........................... 95

xiii

ACKNOWLEDGEMENTS

First and foremost I would like to express my sincere gratitude to my advisor Dr.

Li Li, for her immense knowledge and time, patience, great guidance and continuously

great support during my whole Ph.D. study. She set an excellent example for us. I have

learned a lot from her rigorous scientific attitude and enthusiasm for science. Without my

advisor, this thesis will be another story. Besides, I am grateful to my dissertation

committee members, Dr. Jeremy M. Gernand, Dr. Hamid Emami-Meybodi, and Dr.

Nathaniel R Warner for their time and insightful comments to widen my research from

various perspectives.

I thank Dr. Sridhar Komarneni for his intellectual and knowledgeable help with my

research and for letting me use his laboratory machines. I also thank my fellows’ help.

Hang Wen is always helpful and patient when I had problems on numerical modeling. It is

also pleasure to work with or alongside the past and present fellows and friends: Li Wang,

Xin Gu, Huaibin Zhang, Sruthi Kakuturu, Vikranth Surasani, Travis Tasker, Yingchi

Cheng, Rebecca Fogarty, Jessie Chao, Wei Zhi, Dacheng Xiao, Chen Bao, Fatemeh

Saleihikhoo, Peyman Heidari, and Jaime Harter. All discussions with them help me

understand things better. A special thanks to my friends Zhenzihao Zhang, Guanjun Ding,

Qiumei Zhou, Travis Tasker, and Kirsten Stephens for enriching the life at Penn State.

Lastly, I would like to appreciate my loving, supportive, and patient wife Yefei

Wang and my parents during my whole Ph.D. study. Their infinite love and faithful support

make me strong to confront any challenge during my life. I am also grateful to the help

from my other family members.

1

Chapter 1

Introduction

2

1.1. Background and Motivation

The advanced hydraulic fracturing technology has significantly increased the shale

gas production in the Marcellus formation, one of the largest shale-gas plays in the United

States. There have been up to 11,000 hydraulically fractured wells distributed across

Pennsylvania (Brantley et al., 2018). In parallel to such rapid increase, releases of flowback

and produced waters from Marcellus shale gas extraction occasionally occur through the

pathways including the wellhead, pit/tank, drill rig, flowline, transport, well casing, among

others, (Myers, 2012; Rozell and Reaven, 2012; Vidic et al., 2013a), which pose a potential

risk on the natural water resources. As to the causal mechanism, 40% is due to the

equipment failure with the rest 60% occurring from the human error (Patterson et al., 2017) .

Images such as burning tap water further increase the public’s concerns on drinking water

quality. Here, the flowback water from the hydraulic fracturing and produced water from

the shale gas production are abbreviated as Marcellus Shale waters (MSWs) (Barbot et al.,

2013a; Olmstead et al., 2013; Osborn et al., 2011c), which are typically characterized by

high concentration of total dissolved solids (often > 200,000 mg/L), elevated

concentrations of anions (Br, Cl), cations (Na, Ca, Mg, K, Ba, Sr), toxic trace metals (such

as Cu, Mn, Zn, Pb, Cd), and potentially natural occurring radioactive materials (NORM)

(Acharya et al., 2011; Chapman et al., 2012; Olmstead et al., 2013).

Recent evidences indicate that the natural water contamination are potentially

associated to the hydraulic fracturing activity. A record of 229 spills have been reported in

Pennsylvania. The spill volumes have a median of 38 gallons and can be up to 232,604

gallons (Brantley et al., 2014a; PADEP, 2005-2015). The detected compounds used for

3

hydraulic fracturing and anomalies in major ion concentrations in the monitoring wells

provide evidence of impact from hydraulic fracturing on subsurface drinking water sources

in Pavillion, Wyoming (DiGiulio and Jackson, 2016; Digiulio et al., 2011). In 2015, the

identification of a typical compound (2-n-Butoxyethanol) in hydraulic fracturing fluid at

nanogram-per-liter concentration in the aquifer linked the drinking water contamination to

the shale gas development (Llewellyn et al., 2015). Direct discharge event has been also

reported. For example, Warner et al (2013b) found Cl and Br concentration were 6,000 and

21,000 times higher than their corresponding background levels. Ferrar et al. (2013)

reported that Ba and Sr in river surpassed the US Maximum Concentration Level (MCL)

of drinking water after a deliberate discharge. Two recent studies have also indicated the

produce water may pollute the groundwater by mobilizing the preexisting colloidal

contaminants and significantly enhance the metal mobilization in soils (Chen et al., 2017;

Sang et al., 2014). These findings and investigations raise the concerns of risks on natural

water resources regarding the impact from the accidental releases and highlight the

importance to investigate the reactive transport of multicomponent chemical species from

MSWs when release occurs.

To understand the impact, we first need to unravel the fundamental physical /

chemical processes that control the ultimate reactive transport and fate of chemical species

from MSWs in natural waters. Physical mixing of different waters occurs immediately

upon release, which suggests that the relative magnitude of MSW release rate and the

background flow rate in the receiving waters can play an important role in determining

their concentrations. For example, Trefry and Trocine (2011) observed the precipitation of

barite in the 1:9 mixtures of produced water with seawater, but not in 1:99 or in 1:199

4

mixtures. As to the chemical processes, for example, although Sr can co-precipitate with

Ba producing the barium-strontium sulfate, barite precipitation reaction dominates and

governs the overall reaction rate (Vidic, 2015). In solution with abundant SO4 (thousands

of mg/L), the SrSO4 precipitation could be easily observed after all barium was consumed

in the mixture of flowback water and acid mine drainage (Kondash et al., 2013). On the

other hand, Ca seldom precipitates because the CaSO4 solubility product is 2 to 5 orders of

magnitude higher compared to those of SrSO4 and BaSO4.

Existing studies primarily focused on the reaction between flowback water and acid

mine drainage or seawater in the batch experiment, however, few investigate the reactive

transport of chemical species from MSWs considering the subsurface water conditions.

Flow, transport and multiple water-rock interactions coexist in the subsurface waters which

typically consist of different minerals with differing reactivity. For example, sand and

gravel aquifers often contain unreactive minerals. Carbonate aquifers contain abundant

carbonates. Sandstone aquifers usually have rich clay minerals with high cation exchange

capacity. Reactive mineral such as carbonate dissolves at rates of magnitude higher than

that of quartz dissolution under far-from equilibrium conditions (Tester et al., 1994). Upon

the MSW release, the same chemical species may experience various water-rock

interactions, such as mineral dissolution / precipitation, ion exchange, surface

complexation, in different subsurface systems with different mineralogy. In sand and

gravel aquifers Ba can precipitate as barite but adsorb onto clay minerals in clay-rich

aquifers due to the ion exchange (Frye et al., 2012; Inglezakis et al., 2005; Potgieter et al.,

2006; Schulthess and Huang, 1990; Srivastava et al., 2005). In carbonate aquifers, calcite

dissolution can lead to pH increase and potential precipitation of other minerals (Soler et

5

al., 2008). For example, Mn can precipitate as MnCO3 followed the dissolution of calcite.

The mineral precipitation may impact the physical properties such as permeability and

porosity thereafter affecting the contaminant transport (Phillips et al., 2000). Trace metals

can be also incorporated into calcite thorough solid solution partitioning therefore further

reducing their aqueous concentration in carbonate aquifers (Rimstidt et al., 1998). While

in clay rich aquifer, competitive adsorption of cations on clay can occur (Inglezakis et al.,

2005; Potgieter et al., 2006; Schulthess and Huang, 1990; Srivastava et al., 2005). The trace

metals may compete with cations like Ba, Sr for the clay surface sites (Jacobs and Waite,

2004). As a result, the multiple water-rock interactions will impact the ultimate transport

and fate of chemicals in subsurface water systems of different mineralogy with differing

reactivity.

Natural subsurface systems are inherently heterogeneous in physical and chemical

properties. Physical heterogeneity refers to the variations in physical properties such as

permeability and porosity. For example, in natural subsurface, permeability often varies

orders of magnitude (Newell et al., 1990a). Physical heterogeneity can lead to uneven

distribution of water and affect the solute transport processes, which has been studied for

more than four decades (Adams and Gelhar, 1992; Dagan, 1984; Dagan, 1990; Dagan et

al., 2013; Freeze, 1975; Gelhar and Axness, 1983; Gelhar et al., 1992; Heidari and Li, 2014;

Johnson et al., 2003; Le Borgne et al., 2008). The solute is expected to transport through

the highly permeable zones and forms the preferential flow paths, appearing an early arrival

and long tail breakthrough curve. The phenomenon of strong macrodispersion can be

expected in the porous media with highly physical heterogeneous variability (Dagan, 1984;

Gelhar and Axness, 1983; Gelhar et al., 1992). Physical heterogeneity can be characterized

6

by geostatistical measures such as correlation length. The extent of physical heterogeneity

varies significantly in natural subsurface systems. Long correlation length in low

permeability zones increases longitudinal dispersivity significantly and tend to have earlier

start of breakthrough curves and longer tails (Espinoza and Valocchi, 1997a; Fiori et al.,

2010; Jin and Brantley, 2011; Mohamed et al., 2010).

Chemical heterogeneity, which refers to the spatial patterns of different mineral

types and reactivity, has received increasing attention. In natural subsurface, different

minerals coexist and distribute as uniform patterns in one extreme to layered or clustered

patters on the opposite side. The various distribution patterns are determined by the

composition of source rocks, depositional environments, and depositional and diagenetic

processs (Carozzi, 1993). For example, carbonate minerals are distributed as scattered

cementations and clays are present as layered lenses (Peters, 2009; Reinson and Foscolos,

1986; Salehikhoo and Li, 2015; Viking, 1982). Meanwhile, various minerals present

differing reactivity. For example, quartz reaction rate is orders of magnitude lower than

those of clay minerals (Kump et al., 2000). Recent studies regarding the effect of chemical

heterogeneity on the reactive transport of solutes have been focused on the context of

sorption/desorption (Deng et al., 2013; Espinoza and Valocchi, 1997b; Seeboonruang and

Ginn, 2006; Tompson, 1993; Wang and Li, 2015b), pollutant degradation (Zhang et al.,

2010), and mineral dissolution (Li et al., 2007; Li et al., 2014a; Molins et al., 2012;

Salehikhoo and Li, 2015; Salehikhoo et al., 2013a; Wen and Li, 2017). They showed that

the reaction extents were different at different spatial scales. Zhang et al. (2010 found the

faster rates of biomass growth and contaminant degradation in the homogeneous

micromodel consisting of uniform cylindrical posts relative to the heterogeneous

7

micromodel aggregated of large and small cylindrical posts separated by interstitial pore

spaces. Liu et al. (2014 reported much lower rate of U(VI) desorption in media with layer

structure paralleled to flow direction than that with the relatively homogeneous porous

media. Our previous column experiments indicated that the magnesite dissolution rate was

1.6-2 times lower in heterogeneous column with one flow-parallel magnesite zone

embedded in the quartz matrix than that of the well mixed column (Li et al., 2014a;

Salehikhoo and Li, 2015; Salehikhoo et al., 2013a). Wang and Li (2015b found the sorption

of Cr(VI) was 1.4 order of magnitude smaller in flow-parallel illite zone packed column

relative to the homogeneous column. These studies explored the importance of chemical

heterogeneity on reaction extents and on the reactive transport of single chemical species.

However, there is a significant lack of studies on systematically understanding the effect

of mineral spatial patterns on the multiple geochemical reaction types and extents that

control the reactive transport and behavior of complex chemical species like those from

MSWs. As such, accurate understanding of reactive transport, natural attenuation of

complex chemical species from MSW release in natural heterogeneous subsurface remains

elusive.

1.2. Objectives

In this work, we evaluate the reactive transport of complex chemical species from

MSW release in natural waters through numerical experiments, column experiments and

two dimensional cell experiments. The numerical experiments overcome the limitation of

experiments on porous medium complexity and release condition, which makes better

8

understanding of impact from MSW release on the drinking water quality. The work of

numerical experiments, column and 2D cell experiments will help answer the following

questions: 1) How do the key processes govern the reactive transport and fate of major

cations from MSWs, and how do the time scales and magnitude of MSW release affect the

water quality under various release rates and receiving water conditions; 2) How and how

much does mineralogy composition affect the dominant processes that control the ultimate

reactive transport and retention of multicomponent chemical species during the MSW

release; (3) How and how much do mineral spatial patterns determine the natural

attenuation and retention of chemical species from MSWs release into clay-rich porous

media.

1.3. Dissertation Structure

Chapter 1 provided an introduction for my research work. Chapter 2 utilizes the

reactive transport modeling to understand key processes that govern the reactive transport

and fate of major cations from Marcellus Shale waters (MSWs); and to quantify time scales

and magnitude of the release impacts on water quality under various release and receiving

water conditions. Chapter 3 uses both column experiments and reactive transport model to

mechanistically understand the role of mineralogical composition on the control of reactive

transport and retention of representative trace metals (Mn, Cu, Zn, and Pb), major cations

(Na, Ca, Mg, K, Ba and Sr), and anions (Br, Cl and SO4) from MSWs in natural

groundwater. Chapter 4 uses 2D cell experiment to systematically explore the role of

mineral spatial pattern in determining the natural attenuation and retention of complex

9

chemical species from MSW release into clay-rich porous media. Chapters 2-3 are already

published in academic journals Geochemical Transaction and Science of the Total

Environment. Chapter 4 has been submitted to Energy & Fuels. Chapters 2-4 have been

heavily edited by co-author Dr Li Li. Chapter 5 gives the summary of the research.

10

Chapter 2

How Long Do Natural Waters “Remember” Release Incidents of Marcellus

Shale Waters: a First Order Approximation Using Reactive Transport

Modeling

The work of this chapter was published in Geochemical Transactions, 2016, 17:6.

https://doi.org/10.1186/s12932-016-0038-4.

11

Abstract

Natural gas production from the Marcellus Shale formation has significantly

changed energy landscape in recent years. Accidental release, including spills, leakage,

and seepage of the Marcellus Shale flow back and produced waters can impose risks on

natural water resources. With many competing processes during the reactive transport of

chemical species, it is not clear what processes are dominant and govern the impacts of

accidental release of Marcellus Shale waters (MSW) into natural waters. Here we carry

out numerical experiments to explore this largely unexploited aspect using cations from

MSW as tracers with a focus on abiotic interactions between cations released from MSW

and natural water systems. Reactive transport models were set up using characteristics of

natural water systems (aquifers and rivers) in Bradford County, Pennsylvania. Results

show that in clay-rich sandstone aquifers, ion exchange plays a key role in determining

the maximum concentration and the time scale of released cations in receiving natural

waters. In contrast, mineral dissolution and precipitation play a relatively minor role. The

relative time scales of recovery τrr, a dimensionless number defined as the ratio of the

time needed to return to background concentrations over the residence time of natural

waters, vary between 5 and 10 for Na, Ca, and Mg, and between 10 and 20 for Sr and Ba.

In rivers and sand and gravel aquifers with negligible clay, τrr values are close to 1

because cations are flushed out at approximately one residence time. These values can

be used as first order estimates of time scales of released MSW in natural water systems.

This work emphasizes the importance of clay content and suggests that it is more likely

to detect contamination in clay-rich geological formations. This work highlights the use

12

of reactive transport modeling in understanding natural attenuation, guiding monitoring,

and predicting impacts of contamination for risk assessment.

2.1. Introduction

The development of unconventional natural gas in the Marcellus Shale formation

has grown rapidly in recent years. Significant concerns arise in parallel due to their possible

impacts on water resources. Here Marcellus Shale waters (MSW) are defined as waters

from gas wells including both flowback and produced waters. Marcellus Shale waters are

typically characterized by high total dissolved solids (TDS, usually >200,000.00 mg/L),

elevated concentrations of Br, Cl, major cations (Na, Ca, Mg, K), as well as Ba and Sr,

often accompanied by natural occurring radioactive materials (Acharya et al., 2011;

Chapman et al., 2012; Haluszczak et al., 2013; Olmstead et al., 2013). Accidental release

of MSWs has been reported to occur through impoundments, drilling site discharge, spills,

among others (Myers, 2012; Rozell and Reaven, 2012; Vidic et al., 2013b). Although these

major ions are of less environmental concern than toxic metals, their high concentrations

can still pose adverse effects on human health. For example, Br may produce bromate

through ozonation, a human carcinogen (Haag and Hoigne, 1983). High Ba concentration

can cause muscle weakness and affects blood pressure, nervous and circulatory system

(Brenniman et al., 1981; Judd and Levy, 1991). Their release can deteriorate water quality

and aquatic ecological systems (Myers, 2012). In 2013, four northeastern amphibian

species have been recorded to be adversely affected by 50-1000 mg/L chloride, suggesting

small accidental releases can impede breeding habitats (Kiviat, 2013). They are also

13

important indicators of fracturing fluid, flowback and produced water, and brine

contamination in aquifers or rivers (Brantley et al., 2014a; Mastrocicco et al., 2011; Vidic

et al., 2013b).

Recent evidence highlighted the risk of MSW leakage into natural waters. Direct

discharge of MSW into surface waters has been frequently reported (Ferrar et al., 2013;

Hagström and Jackanich, 2011; Warner et al., 2013a). In Pennsylvania, a total of 229 spills

occurred from 2005 to 2015 (PADEP, 2005-2015), as illustrated in Fig. 2.1A. High

concentrations of methane, saline brine (Osborn et al., 2011a; Warner et al., 2012b) and 2-

n-Butoxyethanol (often used in the fracturing fluids) (2015) were found in drinking

groundwater aquifers in Pennsylvania, indicating potential leakage associated with

Marcellus Shale gas development. The discharge of MSW has been found to increase

downstream Br and Cl concentrations by more than three orders of magnitude (Ferrar et

al., 2013; Warner et al., 2013a). Ferrar et al. (2013) found Ba and Sr surpassed the US

Maximum Concentration Level (MCL) after a deliberate MSW discharge. Sang et al. (2014)

reported 32-36% of heavy metals associated with colloids mobilized by flowback water

flush.

These studies raise questions regarding the impacts of release incidents. How long

and to what extent do natural waters (rivers and aquifers) “remember” the release of MSW?

In other words, how long do MSW stay in natural waters? The ultimate transport and fate

of released chemical species can be affected by many processes (Fig. 2.1B). Mixing of

different waters occur immediately upon release, which means that the relative magnitude

of water release rate and the background flow rate in the receiving waters can play a

significant role in determining their concentrations (Bearup et al., 2012). Cations in the

14

Marcellus Shale waters can participate in multiple water-rock interactions, including

mineral dissolution and precipitation, ion exchange, and surface complexation when clay

minerals are abundant. The geochemical conditions of receiving aquifers, therefore, can be

important in determining dominant reactions, natural attenuation potential, and impacts of

accidental releases (Bertsch and Seaman, 1999). There has been a significant lack of key

measures that quantify and predict reactive transport and fate of chemical species from

MSW.

The objective of this study is to 1) understand key processes that govern the reactive

transport and fate of major cations from MSW; and to 2) to quantify time scales and

magnitude of the release impacts on water quality under various release and receiving water

conditions. It is important to note that here we focus on abiotic interactions, instead of

microbe-mediated biodegradation of organic contaminants. Heavy metals are not included

in this study as they deserve a separate study. The insights learned here can facilitate

fundamental understanding of natural attenuation and assess environmental risks.

Simulations were done under conditions relevant to natural waters in Bradford County in

the Pennsylvania, where local residential concerns on water quality arise in parallel with

the large number of drilled wells (Howarth et al., 2011). We use the multicomponent

reactive transport model CrunchFlow (Steefel, 2009), which solves conservation equations

with respect to mass, momentum, and energy. It has been extensively used to understand

and predict reactive transport of contaminants, and water-rock interaction in porous media

(Li et al., 2016; Wen et al., 2016b). To the best of our knowledge, this work is among the

early studies that use reactive transport modeling tools to understand the impacts of

Marcellus Shale waters in natural water systems.

15

2.2. Methods

2.2.1. Problem setup

As shown in Fig. 2.1B, MSWs are introduced into homogeneous and isotropic

natural water systems including ground water in sandstone (S) aquifers and sand and gravel

(SG) aquifers and surface water. This represents a base case scenario with major focus on

the coupling of transport and geochemical reactions without considering spatial

heterogeneities. The interactions between chemical species in MSWs and sediment

(typically <2 vol.%) in rivers are assumed negligible.

Figure 2. 1 (A) The numbers of Marcellus Shale water release accidents in Pennsylvania

from 2005 to June 8, 2015, with 78% of spills occurred in Northeastern PA. Red spot

indicated the location of Bradford County. The yellow numbers are the numbers of spills.

(B) A schematic diagram of 1-Dimensional modeling setup. We assume a release point

where the Marcellus Shale waters are introduced into the surface water (river) or

groundwater (aquifers). The release can occur through spills, discharge, leakage, seepage,

among others.

16

The S aquifers and SG aquifers were chosen as representative aquifers because they

dominate in Northeastern Pennsylvania (Swistock, 2007). They differ in mineralogical

compositions, with the S aquifers containing much more clay. We chose a branch of the

Susquehanna River to represent the river. The release characteristics of MSWs, including

release rates, time duration, and therefore total volumes, can vary significantly. All these

factors can influence the impacts of accidental release on natural water compositions.

2.2.2. Properties of natural waters and MSWs

Natural water systems. We used the characteristics of a sandstone aquifer with

dominant clay mineral of 21.7% in the Catskill Formation in Bradford County, PA. The S

aquifer has a groundwater velocity of 0.20 m/day and is predominantly a low-rank

graywacke with major minerals being quartz, mica (represented by muscovite) and other

clays, and trace amount of carbonate (mostly calcite) (Xu et al., 2004). In contrast, the Sand

and Gravel aquifer has a groundwater velocity of 0.40 m/day and a lower clay amount than

that of the S aquifer (Denny et al., 1963; Rogers, 1989). For rivers we choose conditions

relevant to the Susquehanna River segment in Bradford County, PA (Fulton, 1878),

considering 2% (v/v) of suspended sediments (Fisher and Stueber, 1976). The major

difference between the surface and subsurface water systems are the orders of magnitude

higher flow rates and the negligible presence of solid phases compared to the aquifers.

17

Table 2. 1 Mineral composition and flow velocity in the natural waters

Mineral Mineral Formula Volume Fraction

aS

Aquifer

bSG

Aquifers

cRiver

dPrimary minerals

Quartz SiO2 4.1310-1 5.8010-1 6.7410-3

K-Feldspar KAlSi3O8 3.5010-1 1.8010-1 7.4010-4

Muscovite KAl2(Si3Al)O10(OH)2 1.0510-1 0.00 0.00

Sericite KAl2(Si3Al)O10(OH)2 4.2010-2 0.00 0.00

Clinochlore-14A Mg5Al2(Si3O10)(OH)8 2.8010-2 0.00 0.00

Daphnite-14A Fe5Al2(Si3O10)(OH)8 2.8010-2 0.00 0.00

Kaolinite Al2Si2O5(OH)4 1.4010-2 9.0010-4 0.00

Illite K0.6Mg0.25Al1.8(Al0.5Si3.5O10)(OH)2 0.00 0.00 1.0810-2

Calcite CaCO3 3.5010-2 6.0010-4 1.5010-3

Dolomite

Suspended sediments

CaMg(CO3)2

----

0.00

---

9.6010-3

---

2.6010-4

2.0010-2

Porosity 3.0010-1 3.9010-1 9.8010-1

Total - 1.00 1.00 1.00

Flow velocity (m/day)

Permeability (m2)

e2.0010-1

e5.0010-13

f4.0010-1

f5.0010-12

g 2.76104

-

a (Glass et al., 1956; Pirc, 1979; Rogers, 1989; Xu et al., 2004; Zhou et al., 2007). b (Denny et al., 1963; Domenico and Schwartz, 1998; Heath, 1983; Rogers, 1989; Trapp and

Horn, 1997; Williams et al., 1998). c (Fisher and Stueber, 1976; Oram, 2012; Reed and Stuckey, 2002; Schulze et al., 2005).

18

d Four secondary minerals, including gypsum, celestitie, barite, and gibbsite, are

initially assigned with a volume fraction of 10-10 for precipitation in simulated

natural water domain (Glass et al., 1956; Rogers, 1989). e Porosity and flow velocity are within the typical range for S aquifers in this area (Pirc, 1979;

Zhou et al., 2007). f Porosity and flow velocity are within the typical range for SG aquifers (Domenico and

Schwartz, 1998; Heath, 1983). g (Oram, 2012; Reed and Stuckey, 2002; Schulze et al., 2005).

Water composition. The three natural waters differ in their chemical composition

(Williams et al., 1998) (Table 2.2). The surface water has higher concentrations of sulfate

and cations including iron, potassium, and zinc, while the ground waters are richer in

calcium, magnesium, and sodium. The major difference between the surface and

subsurface water systems are the orders of magnitude higher flow rates and the negligible

presence of solid phases compared to aquifers. The MSW composition was chosen to be in

the low concentration level of produced and flowback waters.

19

Table 2. 2 Composition of natural waters and Marcellus Shale water (mg/L)

Species aS Aquifer bSG

cRiver dMarcellus Shale water Aquifer

pH 7.4 7.44 7.37 6.9

Br 2.0210-2 2. 00 10-2 1.2910-2 1.87102

Cl 7.99100 5.89100 8.20100 2.92104

SO4 9.9810-1 1.39101 1.54101 6.60100

Al - - - 2.0010-1

Ba 1.2010-1 1.6010-1 2.1410-1 1.01103

Cd - - - 4.9810-2

Ca 4.24101 3.62101 1.57101 1.59103

Cu - - - 2.5010-1

Fe 1.0010-1 5.0010-2 5.9910-2 3.44101

Pb 1.0010-2 1.0010-2 1.0010-2 3.0010-2

Mg 1.64100 6.98100 3.07100 1.50102

Mn 6.0010-3 6.0010-3 - 1.02100

K 2.80100 1.27100 9.0110-1 6.40102

Na 1.85101 1.09101 8.37100 1.32104

Sr 2.9010-1 2.8210-1 - 3.90102

Zn 4.0010-5 4.0010-5 1.78101 1.7010-1

Alkalinity e1.77102 e1.49102 9.88101

f2.45102

as HCO3-

Notes: Water chemistry data are among the range of reported literature. a (Watkins and Cornuet, 2012; Williams et al., 1998). b (Boyer et al., 2012; Warner et al., 2012b). c (America’s Natural Gas Alliance, 2013). d (Hayes, 2009).

e Alkalinity (as HCO3-) was calculated based on equilibrium with calcite using

CRUNCHFLOW and is in the range of reported value of 51-366 mg/L for sandstone

aquifer water and of 85 - 195 mg/L for sand and gravel aquifer water. f Alkalinity is directly from literature. Charges are balanced in all natural waters.

2.2.3. Characteristics of Marcellus Shale water release incident

A total of 9179 unconventional wells were installed in the Marcellus Shale

formation in Pennsylvania from 2005 to 2015 (PADEP, 2005-2015). A total of 229 spill

20

accidents have occurred, dictating a spill possibility of 2.40% per well in average. The spill

volumes varied from 0.003 to about 11.35 m3 with the median value being 0.144 m3

(Gradient, 2013). With the same spill volume, a release can occur at small rates for a long

duration or high rates for a short time frame. The MSWs reached groundwater by seeping

into groundwater aquifers, which is a relatively slow process. Here we assume a net water

volume of 0.144 m3 reaching natural waters; the actual spill water can be much larger as

the vadose zone tends to trap a large percent of spilled water (Gradient, 2013). Here we do

not explicit consider vadose zone processes. The spill rates are varied to examine the

importance of release characteristics.

We define the dilution factor (DF):

MSW NW

MSW

Q QDF

Q

(2.1)

Where QMSW and QNW are the volumetric flow rates (m3/s) of MSW and the

receiving natural waters, respectively. The QNW values are calculated as the product of flow

velocity (m/day) and cross-sectional area of 1 m2 in the model. As such, we focus on

understanding processes at the immediate vicinity of the leakage point and flow path. The

DF quantifies the extent of dilution upon release into natural waters. A high DF value

means that the released MSW is quickly diluted by the fast background natural waters. It

is important to note here that fluid injection into an aquifer typically only causes limited

mixing at the fringes. Here by assuming well-mixed intruding fluid and background water

at the injection point, we can use this as a rough estimation of the relatively magnitude of

the injection fluid rate versus the background fluid rate.

21

2.2.4. Reactive transport modeling

Upon accidental release into natural water systems, the chemical species in the

MSWs interact with natural waters and solid phases. Major processes include mixing,

transport, and various types of water-rock interactions.

2.2.4.1 Reactive transport equations

Reactive transport models (RTM) have been extensively used to understand

complex interactions among physical, chemical, and biological processes in porous media

(Amos and Mayer, 2006; Bao et al., 2014; Salehikhoo et al., 2013b; Zheng et al., 2009).

The governing mass conservation equation for a chemical component i that participates in

ion exchange reactions can be written as follows:

1

( ){ ( ) }

Nri

i i ir r

r

C SC C v R

t t

iD u (2.2)

Here is porosity, Ci is total concentration (mol/m3 pore volume) of i, t is time (s), Di

is diffusion/dispersion tensor (m2/s), u is flow velocity (m/s), Nr is total number of kinetic

reactions that involve species i, vir is stoichiometric coefficient of species i associated with

reaction r, Rr is the rate of chemical reactions in which the species i is involved (mol/m3/s).

The diffusion / dispersion coefficients and flow velocities are set constant with a

disperisivity of 1.0 cm (Gelhar et al., 1992). Here kinetic reactions include mineral

dissolution and precipitation. Ion exchange and aqueous complexation are considered as

fast and are equilibrium-controlled. This equation implies that mass change rate of species

i depends on diffusion/dispersion represented by the first term in the right hand side (rhs),

22

advection described by the second term in the rhs, and reaction described by the third term.

The term S

t

represents mass exchange with solid phase through ion exchange, with

being solid bulk density (g/ m3 pore volume), and S being solid phase concentration of i

(mol/g). This term is essentially a storage term taking into account mass accumulation of i

on the solid phase (Valocchi et al., 1981). The geochemical system here includes 18

chemical components (Table 2.2) and 14 kinetic mineral reactions (Table 2.3).

Table 2. 3 Reaction network, Reaction thermodynamics, and Kinetics for mineral-water

interactions

No. Minerals Reactions alog Keq dlogk

((mol/m2)/s)

eSSA

Kinetic reactions

1 Quartz SiO2(s) ⟺SiO2(aq) -4.00 -13.41 f0.017

2 K-Feldspar KAlSi3O8 + 4H+ ⟺ Al3+ + K+ + 2H2O +

3SiO2(aq)

-0.27 -12.41 g0.098

3 Clinochlore-14A Mg5Al2Si3O10(OH)8 + 8 H+ ⟺ 5Mg2+ +

2Al(OH)4- + 3SiO2(aq) + 4H2O

67.24 -12.52 h1.10

4 Daphnite-14A Fe5Al2Si3O10(OH)8 + 8 H+ ⟺ 5Fe2+ + 2

Al(OH)4-+ 3SiO2(aq) + 4H2O

52.28 -12.52 h 1.10

5 Muscovite KAl2(Si3Al)O10(OH)2 + 10 H+ ⟺ K+ + 3Al3+ +

3SiO2(aq) + 6H2O

13.58 -13.55 i14.28

6 Kaolinite Al2Si2O5(OH)4 + 6 H+ ⟺ 2Al3+ + 5H2O +

2SiO2

6.81 -13.18 j14.70

23

7 Illite K0.6Mg 0.25Al1.8Al0.5Si3.5O10(OH)2 + 8 H+ ⟺

0.25 Mg2++0.6 K++2.30 Al3+ + 3.50 SiO2(aq) +

5 H2O

9.02 -11.60 k65.00

8 Sericite KAl2(Si3Al)O10(OH)2 + 10 H+ ⟺ K+ + 3Al3+ +

3SiO2(aq) + 6H2O

13.58 -13.55 l57.00

9 Dolomite CaMg(CO3)2(s) ⟺ Ca2+ + Mg2+ + 2 CO32- -16.70 -7.53 m0.25

10 Calcite CaCO3 (s) ⟺ Ca2+ + CO32- -8.48 -5.81 n0.48

11 Gypsum CaSO4 (s) ⟺ Ca2+ + SO42- +2H2O -4.48 -2.79 o7.00

12 Celestite SrSO4 (s) ⟺ Sr2+ + SO42- -5.68 - p1.22

13 Barite BaSO4 (s) ⟺ Ba2+ + SO42- -9.97 -7.90 n1.47

14 Gibbsite Al(OH)3 (s) + 3H+ ⟺ Al3+ + 3H2O 8.11 -11.50 q6.50

Ion exchange bCation Exchange Capacity (CEC) clogK

(Vanselow) S aquifer SG aquifers

15 NaX⟺Na+ + X− 5.0 ×10-5 eq/g 3.0×10-5 eq/g 0.00

16 KX⟺K+ + X− -0.69

17 CaX2⟺Ca2+ + 2X− -0.39

18 MgX2⟺Mg2+ + 2X− -0.30

19 BaX2⟺Ba2+ + 2X− -0.45

20 SrX2⟺Sr2+ + 2X− -0.45

a (Wolery et al., 1990a); b (Meunier, 2005); c (Appelo and Postma, 1993; Li et al., 2010); d (Palandri

and Kharaka, 2004); e SSA values are from the laboratory studies in the literature which are

generally observed to be faster than those from the fields (Brantley et al., 2008; White and Brantley,

2003); f(Wollast and Chou, 1988); g(White and Brantley, 2003); h(Brandt et al., 2003); i(Chakraborty et al., 2007a); j(Carrollwebb and Walther, 1988); k(2007b); l(Perng et al., 2006);

m(Sherman and Barak, 2000); n(Tomson et al., 2003); o(Brandt and Bosbach, 2001); p(N.Setoudeh

et al., 2011); q(Russell et al., 2009).

24

The RTM was implemented within a 10 meter one-dimensional domain with 100

grid cells and a fixed spatial discretization of 0.1 meters. The spatial discretization was

chosen as the lowest one that results in the same modeling output as those from spatial

resolutions higher than 0.1 meters. The injection point is the first grid cell. As such, we are

simulating the first 10 meter immediately down gradient of an injection point. We choose

not to do numerical experiments in a large spatial domain of kilometers because the goal

here is to understand dominant geochemical processes that govern natural attenuation of

Marcellus Shale waters. A domain of 10 meter is sufficient for such purpose. As will be

discussed later, the dimensionless time derived from this work is not confined to the

physical length of simulated domain. Running simulation at large spatial scales however

presents additional challenges, largely because reaction parameters in literature are mostly

measured in relatively small scale laboratory systems at the spatial scale of 100 – 102

centimeters. Reaction parameters, in particular reaction kinetic constants and effective

surface areas, are often orders of magnitude lower in large scale heterogeneous systems

(Bao et al., 2014; Dagan et al., 2013; Li et al., 2014b; White and Brantley, 2003). Running

simulations at the scale of kilometers therefore requires overcoming upscaling of reaction

processes, which has been a long-standing and unresolved puzzle (Li et al., 2006; Navarre-

Sitchler and Brantley, 2007).

We examined 5 cases with different types of water systems and release

characteristics (Table 2.4). The three release rates were determined by using reported

dilution factors in literature (Gradient, 2013; NJDEP, 2013; USEPA, 1996) and Equation

(2.1).

25

Table 2. 4 Simulation scenarios for Marcellus Shale water release

Receiving water systems Release

rate (m3/s)

Release

duration (days)

aDilution

Factor (DF)

Residence

Time (days)

Sandstone Aquifer 1.11×10-8 1.50×102 2.09×102 1.50×101

5.55×10-8 3.00×101 4.27×101 1.50×101

1.11×10-7 1.50×101 2.18×101 1.50×101

Sand and Gravel Aquifers 1.11×10-7 1.50×101 4.27×101 9.75×100

River 1.11×10-7 1.50×101 2.87×106 3.55×10-4

Note: a Values are from literature (Gradient, 2013; NJDEP, 2013; USEPA, 1996).

2.2.4.2 Mineral dissolution and precipitation

Mineral reactions are listed in Table 2.3 with their equilibrium constants and

reaction kinetics. In the systems in this paper, most waters are at close to neutral conditions

so we only use rate laws based on neutral mechanisms and follow the classical transition-

state-theory-based (TST) rate law (Lasaga, 1998):

(1 )Ca

eq

IAPR kA

K (2.4)

Here RCa is the rate of calcite dissolution (mol s-1), A is the reactive surface area

(m2). The ion activity product (IAP) is defined as , and Keq is the equilibrium

constant. The IAP/Keq measures the distance from equilibrium. If IAP is lower than Keq,

the water is under saturated and calcite dissolves; if IAP is higher than Keq, the system is

over saturated and calcite precipitates. The equilibrium constants are from the standard

EQ3/6 geochemical database (Wolery et al., 1990b).

2 23Ca CO

a a

26

2.2.4.3 Ion exchange

Ion exchange is represented as follows (Appelo and Willemsen, 1987; Vanselow,

1932):

(2.5)

Here (aq) and (s) refer to aqueous and exchanged phases, respectively; X- denotes

negatively charged exchange sites occupied by cations Au+ and Bv+ of charge u and v for

A and B, respectively. Ion exchange reactions are commonly calculated through the

Vanselow convention using cation mole fractions on the exchange sites (Bethke, 2008).

The overall cation exchange capacity was calculated based on volume fraction and surface

area of clay minerals including muscovite, illite, kaolinite, clinochlore-14A and daphnite-

14A. The selectivity coefficients in Table 2.3 indicate cation affinity to solid surface. The

species Ba and Sr have higher affinity than Ca and Mg, which in turn have higher affinity

than Na and K. This means that under similar concentration conditions, Ba and Sr tend to

be exchanged onto clay surface first before Ca, Mg, and K. The very high Na concentration

in Marcellus Shale waters also induces the exchange of Na onto solid surface compared to

Ca and Mg.

2.2.5. Quantification of release impacts

We define several terms to quantify release impacts on natural water composition.

The maximum concentration in receiving waters during release, Cmax, quantifies the

magnitude of the impacts. The residence time r is calculated by the domain length divided

( ) ( ) ( ) ( )u v

v uuBX s vA aq vAX s uB aq

27

by the natural water flow velocity; it quantifies the time scale at which a non-reactive

species stays in the domain of interest. The recovery time, recovery, is the time scale for each

species to return to within 5% difference from its original concentration. Because different

species involve different types of water-rock interactions (e.g., mineral precipitation versus

ion exchange), this time scale can vary drastically among species. The relative recovery

time rr is defined as the ratio of recovery over r. The rr quantifies the time duration (relative

to residence times) that the released chemical species still remain in the simulation domain.

All these terms are calculated based on modeling observations from output of numerical

experiments. Note that rr is a dimensionless number and its value is not constrained to the

length or time scale of the calculation domain. The rr is similar to the concept of effective

retardation coefficient and is species specific (Valocchi et al., 1981). For instance, the

retardation factors of Ba and Sr are 111 and 60, respectively under neutral condition

(Zabochnicka-Świątek et al., 2010). The cations generally follow the retardation sequence

of Mg<Ca<Sr<Ba (Brady, 1996; Zhang et al., 2001). A higher affinity to solid surface leads

to a larger retardation and therefore a slower movement, longer residence time and

ultimately longer rr and memory.

2.3. Results and discussion

Section 2.3.1 focuses on understanding processes that control transport and fate of

major species in the S aquifer. Section 2.3.2 assesses the role of release rates. Section 2.3.3

compares reactive transport of major species under different release rates under different

natural water conditions.

28

2.3.1. Controlling processes in the sandstone aquifer

Here we examine the spatio-temporal evolution of major species after release into

the S aquifer under 4 scenarios of increasing process complexity: a case including only

mixing without any reactions (“MIX”), a case including mixing and mineral

dissolution/precipitation (“MIX+DISS/PPT”), a case with mixing and ion exchange

without mineral dissolution/precipitation (“MIX+IEX”), and a case including mixing,

mineral dissolution/precipitation, and ion exchange (“MIX+DISS/PPT+IEX”). The release

occurred from day 10 to day 25 at the rate of 1.11×10-7 m3/s in all cases. Before the release

accident, initial water-rock equilibria are established by continuously injecting natural

waters into the simulated domain until their compositions are stabilized.

2.3.1.1 Temporal evolution at the release point

Figure 2. 2. Evolution at the release point for Br under four scenarios. All four color lines

overlap. The grey shaded zone represents the release period. Due to its non-reactive nature,

the inclusion of different processes does not affect their evolution.

29

Br and Cl. The breakthrough of Br and Cl in the four scenarios are the same due

to their non-reactive nature (Fig. 2.2). The concentrations increase upon release and return

to background concentration when the release stops. Their concentrations in the MSW are

185.00 and 29,252.00 mg/L, respectively. With the dilution factor of 21.85, the calculated

Br and Cl concentrations during release are 8.82 and 1,404.00 mg/L, respectively,

approximating their MSW concentrations divided by the dilution factor plus the

background concentration.

30

Figure 2. 3. Evolution at the release point for (A) Ca (mg/L) in logarithmic scale, (B) Ca

on exchange sites (mol/g solid), (C) Mg (mg/L) in logarithmic scale, (D) Mg on exchange

sites (mol/g solid), (E)Na (mg/L) in logarithmic scale, (F) Na on exchange sites (mol/g

solid), (G) calcite reaction rate (mol/m2/s) (negative indicates dissolution and positive

values indicate precipitation), and (H) pH. Grey line overlaps with the black line.

Na, Ca, and Mg. The Na concentration ([Na]) is the highest (13,200.00 mg/L)

among the three species in the MSW. The Na exchanges with presorbed Ca and Mg at the

31

solid concentrations of 2.72×10-7 mol/g and 2.55×10-8 mol/g, respectively. The Ca

therefore depends on the mixing with the ground water, dissolution and precipitation of

calcite, and ion exchange. In the MIX case, Ca behaves similarly to Cl. The [Ca] in the

MIX+DISS/PPT case is lower than that in the MIX case because of calcite precipitation,

as indicated by the positive calcite rate in Fig. 2.3G. In the MIX+IEX case, the [Ca]

increases sharply upon release, which echoes the fast Ca decrease on the surface in Fig.

2.3B and Na increase on the solid phase in Fig. 2.3F. This indicates that the quick increase

is caused by the ion exchange between Na and the presorbed Ca. This desorbed Ca leads

to calcite precipitation with sharply increasing rates during release (Fig. 2.3G positive

calcite rates), which decreases aqueous Ca significantly and cause calcite dissolution

afterwards (Fig. 2.3G negative calcite rates). At the time when release stops, the

precipitation even draws Ca concentration to below background concentration. The system

eventually relaxes back to the original state. Despite the differences in MIX+IEX and

MIX+DISS/PPT+IEX cases, similar [Ca] in these two cases indicate the dominance of ion

exchange and relatively minor role of calcite dissolution/precipitation when both processes

coexist. Compared to the MIX+DISS/PPT case, the increase in [Ca] in the

MIX+DISS/PPT+IEX case also leads to much higher calcite precipitation rate during

release (Fig. 2.3A and 2.3G).

Similar to Ca, Mg also participates in mineral dissolution/precipitation

(Clinochlore-14A and dolomite) and ion exchange (Table 2.3). Compared to Ca, its

concentration is about an order of magnitude lower in both background and MSW. Its

evolution at the release point resembles that of Ca (Fig. 2.3C). Although not shown here,

dolomite is close to equilibrium while the dissolution rate of Clinochlore-14A is in the

32

order of 10-10 mol/s. Comparison between the 4 cases show that Mg behaves similarly to

Ca and is primarily controlled by ion exchange. The highly elevated Na in MSW leads to

massive exchange on the solid surface. After the release stops, Na slowly desorbs, resulting

in a long tail for over more than 150 days (Fig.2.3E and 2.3F). Conversely, Ca and Mg sorb

back to the solid (Fig. 2.3B and 2.3D), which results in lower aqueous Ca and Mg

concentration when compared to the background concentration and calcite dissolution, as

indicated by the negative calcite rates in the MIX+DISS/PPT+IEX case after the release.

They eventually return to background concentrations after continuous groundwater

flushing and reach equilibrium again.

The original pH values are 7.40 and 6.90 in the S aquifer and MSW, respectively.

Values of pH drop upon release in all cases (Fig. 2.3H). In the MIX and MIX+IEX cases,

pH drops slightly and returns immediately to its background when release stops. In the

other two cases that involve mineral dissolution and precipitation, pH drops much more

significantly during the release, primarily due to calcite precipitation. In the

MIX+DISS/PPT+IEX case, because the ion exchange kicks out sorbed Ca and increased

aqueous [Ca], the higher calcite precipitation rates lead to more significant pH decrease

(Fig. 2.3G). The calcite dissolution leads to pH increase for an extended period of time

until Ca dominates the solid surface again. In general, the pH curve mirrors the shape of

calcite rate. The pH values relax back to its background immediately after the release in all

cases except the MIX+DISS/PPT+IEX case where pH is mostly controlled by calcite

dissolution and precipitation reactions.

33

Figure 2. 4. Evolution at the release point for (A) Ba in water (mg/L), (B) Ba on surface

(mol/g solid), (C) Sr (mg/L), (D) Sr on surface (mol/g solid). Ion exchange controls

concentrations of these species while mineral dissolution and precipitation play a minor

role.

Ba and Sr. Barium and strontium exchange with presorbed cations Ca and Mg,

leading to decreased aqueous [Ba] and [Sr], and increased aqueous [Ca] and [Mg] in

MIX+DISS/PPT+IEX(Fig. 2.4). After the release stops, Ba and Sr slowly desorb over a

longer period of time. Although not shown here, barite and celestite precipitate in

negligible rates, indicating the dominant role of ion exchange.

2.3.1.2 Spatio-temporal evolution in the MIX+DISS/PPT+IEX case

Here we examine the spatio-temporal evolution of major species in the

MIX+DISS/PPT+IEX case where all processes are included. The release occurs between

day 10 and 25 at the rate 1.11×10-7 m3/s.

34

Figure 2. 5. Spatio-temporal evolution of Br concentration in the sandstone aquifer in the

MIX+DISS/PPT+IEX case on Days 11, 25, 27 and 160. Release starts on day 10 and ends

on day 25. The other tracer Cl behaves the same as Br.

Tracers. During release, the [Br] amd [Cl] in the down gradient rapidly increase

(Fig. 2.5). After the release, Cl returns to background concentration starting from the

release point. The high concentration zone gradually migrates out of the domain until the

system returns to its background.

35

Figure 2. 6. Spatio-temporal profiles of major species in the sandstone aquifer under the

MIX+DISS/PPT+IEX scenario on Days 11, 25, 27 and 160. Left Column is for aqueous

concentrations (mg/L); right column is for concentrations on solid surface (mol/g solid).

Rows from the top to bottom: Ca (A and B), Mg (C and D), Na (E and F), Ba (G and H),

and Sr (I and J).

36

Reactive species. The 5 major cations can be categorized into 2 groups (Fig. 2.6).

Group I includes Ca and Mg (top two rows), both of which are in the MSW and are

originally on exchange sites. They are mobilized through ion exchange with cations in the

MSW, primarily Na, Ba, and Sr. During release, their aqueous concentration peaks in some

zone while the corresponding solid concentration show “valley” of low concentrations. The

peaks expand over time during the release. At the end of the incident, their aqueous

concentrations are lower than the background concentrations due to their exchange back to

the surface. Correspondingly, their solid phase low concentration valleys migrate down

gradient slowly over a much longer time scale, long after the release stops on day 25. The

depletion zone also becomes wider and shallower due to dispersion as they migrate out of

the system.

The Group II species consist of Na, Ba, and Sr, which are abundant in the MSW

and exchange with solid surface upon release, displacing Ca and Mg. During release they

all show highest aqueous and solid concentrations at the release point, while quickly

decrease down gradient. Both peak aqueous and solid concentrations increase over time

during release. After the release stops, these cations on the exchange sites gradually

become remobilized back into the aqueous phase through ion exchange. Compared to

Group I species, Group II species show peaks in both aqueous and solid phases that migrate

at similar rates down gradient. The concentration peaks become wider and shallower over

time.

Quantification of memory indexes from spatio-temporal profiles. The

“maximum concentration” Cmax and the “recovery time” recovery can be calculated from the

spatial profiles discussed above (Fig. 2.5-2.6). These two measures differ significantly

37

from one species to another. The Cmax of tracers (Br and Cl) are controlled by the mixing

process. After release the system returns to their background after approximately one

residence time. For Group I species (Ca and Mg), Cmax values are higher than those

estimated by their dilution factor because they are mobilized from the solid surface during

release. For Group II species (Na, Ba, and Sr), their peak concentrations equal to or are

lower than those predicted by dilution factor because they exchange onto solid surface. The

memory or the time scales of the reactive species are dictated by their affinity to the surface.

On day 160, the peak for Na has disappeared, indicating its migration out of the system. In

contrast, the peaks of Ba and Sr are approximately at 8 meter at that time. As indicated in

the ion exchange coefficients in Table 2.3, the affinity to the surface is Ba/Sr > Ca/Mg >

Na. The Ba and Sr therefore migrate out of the system much slower. The rr values are 6.79,

9.25, 9.38, 20.09, 18.76 for Na, Ca, Mg, Ba, Sr, respectively. This means that it takes 6.79

residence times to flush out Na, 9.25/9.38 residence time for Ca/Mg, and 20.09/18.76 for

Ba/Sr, which are consistent with their affinity to the solid surface. This gradient of time

scales consistent with their gradient of the affinity to the solid surface is similar to the

chromatographic effects in literature (Valocchi et al., 1981).

2.3.2 Effect of release characteristics in the sandstone aquifer

Three cases were compared with the same release volume of 0.144 m3 however at

different release rates. The “High” release rate is 1.11×10-7 m3/s for 15 days, the same as

the case in the section 4.1. The “Medium” rate is 5.55×10-8 m3/s for 30 days. The “Low”

38

release rate is 1.11×10-8 m3/s for 150 days (Table 2.4). The corresponding dilution factors

are 21.85, 42.70, and 209.54, respectively.

39

Figure 2. 7. Profiles of Br, Ca, Ca on solid surface, Na, Na on solid surface in the sandstone

aquifer during release (left column) and after release (right column) under the three release

cases. The High, Medium, and Low release rates are 1.11×10-7 m3/s for 15 days, 5.55×10-

8 m3/s for 30 days, and 1.11×10-8 m3/s for 150 days, respectively. The “During Release”

curves are on day 10 after the release starts. The “after Release” curves are on day 5 after

release stops.

40

Figure. 2.7 shows the spatio-temporal evolution for Br (tracer), Ca (Group I), and

Na (Group II). In general, the higher release rate, the higher impact on the water chemistry.

For the tracers, Cmax are essentially the MSW concentrations divided by the corresponding

dilution factor in each case. For the reactive species, the low release rate leads to much

lower aqueous and / or solid concentrations than in the high rate case. In addition, it takes

shorter time to flush out Na in the low release rate case and therefore the system recovers

sooner.

2.3.3 Effect of receiving water bodies

The river has the highest flow velocity (2.76104 m/day) compared to the S aquifer

(0.20 m/day) and SG aquifers (0.40 m/day). The S aquifer has 21.7 vol% of clay content

compared to 0.9% in the SG aquifers and zero in the river. The release occurred at the same

high rate of 1.11×10-7 m3/s for 15 days. The dilution factors for the three receiving natural

waters are 21.85, 42.70, and 2.87×106, for S aquifer, SG aquifers, and river, respectively.

41

Figure 2. 8. Profiles of Br, Ca, Ca on solid surface, Na, Na on solid surface during release

(left column) and after release (right column) in the sandstone aquifer, sand and gravel

aquifer, and river, respectively. The release rate is 1.11×10-7 m3/s for 15 days. The “During

Release” is on day 10 after the release starts. The “after Release” is on day 5after release

stops.

42

Figure 2.8 shows the effects of receiving water characteristics on the reactive

transport of major species. With orders of magnitude higher flow velocity, the river has no

memory of MSW - all concentrations are at the background concentration. The MSW

however leaves their footprint on the ground water aquifers. The [Br] during the release is

lower in the SG aquifers than in the S aquifer due to the higher flow velocity in the SG

aquifers. Note that the background [Br] in the two aquifers are also different, with lower

[Br] in the SG aquifers. The reactive species behave similarly to the tracers in the SG

aquifers because of the low clay content and the lack of ion exchange. The higher dilution

factor in the SG aquifers lead to a concentration about half of the maximum [Na] in the S

aquifer at the release point, while in the down gradient [Na] is higher in the SG aquifers

because negligible ion exchange occurs compared to that in the S aquifer. The [Na] and

[Ca] return to the background concentration much faster in the SG aquifers than in the S

aquifer.

43

2.3.4 Impacts of the release incidents

Figure 2. 9. The memory index of natural waters: Cmax and (A) recovery and (B)rr of major

species in the river (filled squares), SG aquifer (filled triangles), and S aquifer with high

release (filled circles), medium release (crossed circles), and low release rates (open

circles). Both are calculated from the modeling output of spatio-temporal concentration

evolution. The Cmax is determined as the maximum aqueous concentration during release.

The recovery is the time scale for each species to return to within 5% difference from its

background concentrations in natural waters. The relative recovery time rr, calculated as

the ratio of recovery over r, is a measure of the time scale that natural waters remember the

incident relative to their residence time. Each species is represented by one color, with

dashed line of the same color being their drinking water standard. In S aquifer with

abundant clay, rr values depend on cation affinity to solid surface with rr between 5-10

for Na, Ca, and Mg, and 15-20 for Sr and Ba.

Values of Cmax and recovery quantify the impacts and time scales of release accidents.

The numerical experiments indicate that Cmax of Ca, Na and Cl are 2 orders of magnitude

higher than Ba, Sr and Mg (Fig. 2.9). The river has the lowest Cmax compared to the

groundwater aquifers with all Cmax values below the drinking water standard (Canada, 2014;

Edition, 2011; USEPA). In the SG aquifers, all species behave as if they are non-reactive

with their Cmax values proportional to their concentrations in the original MSW. Only the

Cmax of Cl and Na exceed the drinking water standard. In the S aquifer, however, almost

44

all species exceed drinking water standards in the High and Medium release rate cases. In

the Low release rate case, only Br and Ba exceed the drinking water standard. The recovery

values vary by orders of magnitude and depend on specific characteristics of natural waters,

release incidents, and individual species (Fig. 2.9). The S aquifer remembers the incident

longer compared to SG aquifers due to the lower flow velocity and higher clay content.

The Low release rate case to recover much fast back to the background concentration than

the Medium and High cases. Their corresponding relative time scales, rr, however, vary

only from 1.0 to a maximum of about 20 (Fig. 2.9B). In fact, under all conditions where

chemical concentrations are controlled by the mixing process, values of rr are close to 1.

This includes non-reactive species in all natural waters at all release rates, and reactive

species in natural waters with negligible clay content (rivers and SG aquifers). Only in S

aquifer with abundant clay, rr values depend on cation affinity to solid surface with rr

between 5-10 for Na, Ca, and Mg, and 15-20 for Sr and Ba.

2.3.5 Discussion

Environmental Implications. Despite the fact that multiple minerals are involved

in dissolution and precipitation, these reactions play a relatively small role compared to ion

exchange. This highlights the importance of clay content in determining the time scales

and impact of incidental release on natural waters.

The results have interesting implications in understanding reactive transport,

monitoring, and detection of contaminants from Marcellus Shale waters in natural water

systems. In a controversial example involving unconventional gas wells in a sandstone

45

formation near Pavillion, Wyoming, EPA detected contamination in shallow monitoring

wells from 2010 to 2011 (Digiulio et al., 2011). Synthetic organic compounds used in

hydraulic fracturing fluids were detected in monitoring wells; [Cl] and [K] were found

more than one order of magnitude higher in a monitoring well than the background

concentrations. In a second time sampling in the same wells in April and May 2012, some

previously detected compounds (e.g., xylenes, toluene) were not found and a number of

other compounds have lower concentrations than the previous analysis. As shown in the

spatio-temporal figures, there are only certain “time windows” that the signature of MSW

can be observed in a specific location, which indicates the ephemeral nature of

contamination events and the transient and elusive contamination plumes. This imposes

significant challenges to monitoring and detection of groundwater contamination (Brantley

et al., 2014a).

There have been several cases that discharged MSW were detected in rivers. For

example, at the discharge point, [Cl] and [Br] were 6000-fold and 12,000-fold higher than

that in the stream background, both exceeding the drinking water standard (Warner et al.,

2013a). This is a case where the MSW was discharged into river with large volume and

therefore the dilution factor of 739 is more than three orders of magnitude lower than that

in our model (2.87×106). During dry season, low flow rates in rivers lead to lower DF

(Rozell and Reaven, 2012), which also increase the possibility of contamination detection.

Table 2.5 shows a few cases where elevated chloride concentrations were reported when

MSWs were discharged into river. The DF values in these cases, estimated as the ratio

between the flow rates of the river and the discharge rate, vary between 510 and 1246.

46

These values are 3-4 orders of magnitude lower than the DF value in the incidental release

case in this work.

Table 2. 5 aCases with contamination detected during direct discharge of MSW into rivers

Receiving water

body

MSW

Release

rate (×10-3

m3/s)

River flow

rate

(m3/day)

DF

[Cl] in

discharge

outlet

Calculated

[Cl] in

river

(mg/L)

Monitored [Cl]

in river (mg/L)

(mg/L)

Blacklick Creek 6.7 432,000 739 80542 107.78 195.00±175.00

Monongahela River 111.1 4,893,000 510 28879 56.62 136.80±2.70

Ten Mile Creek 11.3 1,223,000 1246 44915 35.84 61.90±2.49

a The release rates and flow rates are from literature; DF is calculated as the ratio of the reported

river flow rate over the MSW release rate; [Cl] in the discharge outlet are measured values from

literature; Calculated [Cl] in river are estimated by dividing the measured [Cl] in the discharge

outlet with DF. Monitored [Cl] was directly from literature. References for this table include (Ferrar

et al., 2013; Kyshakevych and Prellwitz, 2001; Warner et al., 2013a) .

Limitations. This study is for the specific hydrological and geochemical conditions

in Northeastern Pennsylvania in homogeneous systems of one-dimensional 10 meters

immediately down gradient of the release point where the impacts on natural waters are

most significant. This is different from three dimensional natural water systems in reality

that have larger dispersion and spreading. As such, the calculation here likely overestimates

Cmax and recovery and therefore represents the worst case scenario. However, the quantitative

term defined here, especially the relative recovery time rr, is dimensionless and is not

restricted to the length scale of the simulation domain. For example, if the estimated rr for

a particular species is 5.0, it means that the time needed for recovery is five times of the

water residence time. This estimation can be used for systems of different lengths and of

flow velocities, because residence times scale with length and flow velocity. As such, rr

47

provides the approximation needed for estimating memory or time scales of release

incidents in natural waters. In addition, as long as geochemical conditions remain relatively

similar, the dominant water-rock interactions are similar.

Here we mainly focus on water-rock interactions of major cations in the MSW

without considering redox reactions and biodegradation of organic contaminants that can

be present in MSWs. If organic contaminants are present and used by microbe as carbon

source, biodegradation reactions will transform organic contaminants into dissolved

inorganic carbon, which can increase the concentrations of bicarbonate significantly.

Under such circumstances, carbonate precipitation may play a much more significant role,

as indicated in literature (Bao et al., 2014; Li et al., 2010; Li et al., 2009).

This study also considers homogeneous systems. Natural groundwater aquifers are

typically layered with heterogeneous distribution of hydrological and geochemical

properties (Haggerty and Gorelick, 1995; Heidari and Li, 2014). Such spatial

heterogeneities have long been reported to cause order-of-magnitude longer tail for non-

reactive tracers (Dentz et al., 2004; Luo and Cirpka, 2008; Sudicky et al., 2010) and lower

reaction rates (Salehikhoo and Li, 2015; Wang and Li, 2015a; Wen et al., 2016b). The

specific characteristics of different natural water systems, including spatial distribution of

clay lenses and layers, therefore, will play a significant role in determining the recovery

time of natural waters from incidental release.

In addition, we did not consider the vadose zone processes. Vadose zone processes

will affect how much spill volume and chemicals will get into aquifers. However, the major

reactive transport processes in natural waters will remain the same and the time scales that

the released chemicals remain in aquifer will still be determined by their affinity to the

48

solid surface – this aspect is not going to change whether we include vadose zone processes

or not.

2.4. Conclusions

Recent studies on MSWs have mostly focused on evidence linking altered water

composition to possible release of Marcellus Shale waters. Process-based understanding

and quantification on reactive transport of accidentally released chemicals, however, are

largely lacking. Here we use major cations as tracers of release events and reactive

transport numerical experiments to illustrate key processes that determine the impacts of

accidental release.

The magnitude of the impacts is quantified by Cmax, the maximum observed

concentration during release, while the time scale of the impact, recovery, the required time

duration to recover to within (1005%) of its back ground concentration. We also define a

dimensionless number rr that is the relative ratio of the recovery compared to the residence

time of natural waters r. Our results show that in rivers and SG aquifers with negligible

clay content, mixing process controls Cmax and recovery of all species. The dilution factor

determines Cmax while recovery approximates the residence time. In clay-rich natural water

systems, ion exchange plays a dominant role compared to mineral dissolution and

precipitation. The S aquifers with abundant clay selectively remember Sr and Ba for 10-20

residence times due to their higher affinity to clay surface compared to 5-10 residence times

for Na, Ca, and Mg. This highlights the importance of clay content in both monitoring and

49

natural attenuation of chemicals from Marcellus Shale waters. This suggests that under

otherwise similar conditions, it is more likely to detect contamination in clay-rich

geological formations because it takes longer for chemicals to return to its original state in

these formations.

This work highlights the usefulness of reactive transport modeling in process

understanding and in guiding sampling and monitoring in natural water systems. Findings

from this work facilitates prediction of contaminant transport and fate, quantifies impacts

of released MSWs in natural waters, and provides insights on risk assessment and strategies

for sustainable shale gas development.

Acknowledgments

This work is funded by DOE National Energy and Technology Laboratory (NETL). The

findings and conclusions here do not necessarily reflect the view of the funding agency.

Coauthor including Dr. Li Li is appreciated. We acknowledge Susan L. Brantley for

providing information on accidental Marcellus Shale flow back/produced water spill. We

thank Reviewers for their meticulous and insightful reviews that has helped improve the

manuscript.

50

Chapter 3

Mineralogy Control on Reactive Transport of Marcellus Shale Waters

The work of this chapter was published in Science of the Total Environment, 2018, 630:1573-1582. https://doi.org/10.1016/j.scitotenv.2018.02.223

https://doi.org/10.1016/j.scitotenv.2018.02.223

51

Abstract

Produced or flowback waters from Marcellus Shale gas extraction (MSWs)

typically are highly saline and contain chemicals including trace metals, which pose

significant concerns on water quality. The natural attenuation of MSW chemicals in

groundwater is poorly understood due to the complex interactions between aquifer minerals

and MSWs, limiting our capabilities to monitor and predict. Here we combine flow-

through experiments and process-based reactive transport modeling to understand

mechanisms and quantify the retention of MSW chemicals in a quartz (Qtz) column, a

calcite-rich (Cal) column, and a clay-rich (Vrm, vermiculite) column. These columns were

used to represent sand, carbonate, and clay-rich aquifers. Results show that the types and

extent of water-rock interactions differ significantly across columns. Although it is

generally known that clay-rich media retard chemicals and that quartz media minimize

water-rock interactions, results here have revealed insights that differ from previous

thoughts. We found that the reaction mechanisms are much more complex than merely

sorption and mineral precipitation. In clay rich media, trace metals participate in both ion

exchange and mineral precipitation. In fact, the majority of metals (~50–90%) is retained

in the solid via mineral precipitation, which is surprising because we typically expect the

dominance of sorption in clay-rich aquifers. In the Cal column, trace metals are retained

not only through precipitation but also solid solution partitioning, leading to a total of 75–

99% retention. Even in the Qtz column, trace metals are retained at unexpectedly high

percentages (~20–70%) due to precipitation. The reactive transport model developed here

quantitatively differentiates the relative importance of individual processes, and bridges a

52

limited number of experiments to a wide range of natural conditions. This is particularly

useful where relatively limited knowledge and data prevent the prediction of complex rock-

contaminant interactions and natural attenuation.

3.1. Introduction

Natural gas extraction in Marcellus Shale has raised significant concerns about its

impacts on natural water resources (Maloney et al., 2017; Sang et al., 2014). Flowback

water from hydraulic fracturing and Marcellus Shale produced water from shale gas

extraction, here abbreviated as Marcellus Shale waters (MSWs), are characterized by high

organic content, total dissolved solids (TDS) (usually >200,000 mg/L), and elevated

concentrations of base cations, anions, and trace metals (Acharya et al., 2011; Chapman et

al., 2012; Haluszczak et al., 2013; Olmstead et al., 2013; Patterson et al., 2017; Shih et al.,

2015). The release of MSWs, either accidental or intentional, can deteriorate natural water

quality and aquatic ecosystems. As of 2016, 229 MSW spills have been reported by the

Pennsylvania Department of Environmental Protection (PADEP) with spill volumes up to

232,604 gallons. Intentional release, including road spreading of produced MSWs, is also

a common practice for deicing and dust control, which can lead to the accumulation of

chemicals in soils (Myers, 2012; Skalak et al., 2014). It is important to understand the

reactive transport and fate of chemicals from MSWs in natural water systems such as

aquifers.

Various studies have investigated the reactions between MSWs with acid mine

drainage (AMD), seawater, and surface soils. Trefry and Trocine (2011) observed barite

53

precipitation when mixing MSWs and seawater. Strontium and barium have been observed

to co-precipitate as Ba-Sr sulfate (Vidic, 2015). Kondash et al. (2013) reported SrSO4

precipitation in mixed flowback and AMD where SO4 concentrations reach thousands of

mg/L. Calcium rarely precipitates due to the 2 to 5 orders of magnitude higher solubility

of CaSO4 than that of SrSO4 and BaSO4. High concentrations of Ca and other divalent

cations, however, can affect SrSO4 and BaSO4 precipitation through lattice positioning

(Hennessy and Graham, 2002; Jones et al., 2004; Jones et al., 2008). Column experiments

have shown that produced water can contaminate groundwater by remobilizing existing

colloidal pollutants (Sang et al., 2014) and significantly enhance metal transport in soils

(Chen et al., 2017).

In natural groundwater systems, flow, transport, and complex geochemical

reactions occur simultaneously and have profound impacts on the ultimate fate of

contaminants. In addition, natural aquifers often consist of multiple minerals of differing

reactivity. Sand and gravel aquifers occur ubiquitously and do not contain much reactive

minerals. Carbonate aquifers provide 22% of groundwater supply in the United States and

25% of drinking water to the global population (Maupin and Barber, 2005; Quinlan and

Ewers, 1989). Different minerals can interact with MSWs through multiple reactions

including mineral dissolution and precipitation, ion exchange, solid solution partitioning,

and surface complexation (Apodaca et al., 2002; Brantley et al., 2008; DeSimone et al.,

2009; Elango and Kannan, 2007). For example, Ba can precipitate mostly as BaSO4 in sand

and gravel aquifers but can adsorb onto clay in clay-rich aquifers. In carbonate aquifers,

the water chemistry is often dominated by carbonate dissolution and precipitation, and

potential incorporation of trace metals into carbonate solid phases. In clay rich aquifers,

54

surface complexation and ion exchange reactions are often prevalent (Frye et al., 2012;

Inglezakis et al., 2005; Potgieter et al., 2006; Schulthess and Huang, 1990; Srivastava et

al., 2005). Trace metals can compete with cations such as Ba, Sr, and Na for surface sites

(Abollino et al., 2008; Jacobs and Waite, 2004). As a result, the same chemicals can

experience very different rock-water interactions in aquifers of distinct mineralogy.

Given the complexity of the MSW water chemistry and aquifer mineralogy, it has

been challenging to identify dominant processes that influence the ultimate fate of

chemicals. Our work in Chapter 2 using reactive transport modeling (Cai and Li, 2016)

have indicated that breakthrough times of MSW chemicals are much longer in clay-rich

aquifers than in sand and gravel aquifers due to the retardation through the ion exchange

reactions between chemicals and clay, suggesting the predominant influence of ion

exchange. The goal of this work is to use column experiments and reactive transport

modeling to mechanistically understand processes that control the ultimate fate of

representative trace metals (Mn, Cu, Zn, and Pb), major cations (Na, Ca, Mg, K, Ba and

Sr), and anions (Br, Cl and SO4) from MSWs in natural groundwater. The process-based

reactive transport modeling can pinpoint controlling parameters that influence these

chemicals.

55

3.2. Materials and methods

3.2.1. Mineral preparation

Columns were packed with different minerals to represent sand and gravel aquifers,

carbonate aquifers, and clay-rich aquifers. Calcite was used to represent carbonate minerals

because of its prevalence (Lindsey et al., 2009). Vermiculite is a common silicate clay

mineral with layered structure (Jackson and Inch, 1989; Rogers, 1989) and high cation

exchange capacity (dos Anjos et al., 2014). In this work we choose vermiculite as the model

clay also because it does not swell as much as other clays, which help bypass complications

arising from the formation of lumps and cracks and clogging in the column. The

vermiculite column is used to represent the clay rich aquifers with relatively high CEC

value (Lani and Schoonen, 2010; Rogers, 1989; Spradlin, 2015).

Mineral grains of 350 ~ 420, 125 ~ 150, 75 ~ 125 µm were used for quartz, calcite,

and vermiculite, respectively, to represent the physical and geochemical characteristics of

different minerals. A relatively small grain size between 75 and 125 µm was chosen for

clay minerals because they typically occur as low permeability minerals (Koltermann and

Gorelick, 1996). Grain sizes lower than 75 µm was not possible because they are easily

flushed out of the column. Grain sizes between 350 and 420 µm were used for quartz sand

because they typically have the highest permeability among the three minerals (Koltermann

and Gorelick, 1996; Salehikhoo and Li, 2015).

X-ray Diffraction (XRD) analysis and Inductively Couple Plasma Atomic Emission

Spectroscopy (ICP-AES) of calcite identified trace amounts of impurities including 0.17%

of magnesite, 0.05% of sodium, and 0.05% of strontium. Vermiculite samples analyzed

56

using LI-COR CO2–H2O Analyzer (LI-7000) (Bazilevskaya et al., 2015) indicate 1.29%

(mass) of calcite. The chemical composition of vermiculite is detected by ICP-AES and is

listed in the supporting information. The XRD pattern of vermiculite in Figure S1 indicates

a blend of vermiculite, mica and amphibole. The Cation Exchange Capacity (CEC) of

vermiculite was measured to be 43.2 meq/100g within the range of typical clays (Lani and

Schoonen, 2010; Rogers, 1989; Spradlin, 2015).

3.2.2. Mineralogical composition and column property measurement

Three columns (5 cm in diameter by 50 cm in length) were packed with quartz (Qtz),

quartz-calcite (Cal), and quartz-vermiculite (Vrm) with detailed physical and chemical

properties outlined in Table 3.1. Measurement details are documented in the Supporting

Information. The Qtz column contained 100% quartz as solid materials while the other two

columns were packed with calcite (Cal) or vermiculite (Vrm) and quartz. Columns were

“wet packed” following previously reported procedure (Salehikhoo et al., 2013a). The end

of the columns was capped with a 25 μm polytetrafluoroethylene (PTFE) frit to hold

mineral grains inside the columns. The effective permeability of the columns differs with

different grain sizes.

57

Table 3. 1 Physical and geochemical properties of the columns

Columns Qtz Cal Vrm

Total volume (cm3) 1013.45 1013.45 1013.45

Quartz (Qtz, gram) 1799.64 1795.50 1589.58

Calcite (Cal, gram) a- 203.96 -

Vermiculite (Vrm, gram) - - 140.20

Volume percent (%) - 7.45 5.42

Grain size Quartz (μm) 350-420 350-420 350-420

Grain size Calcite (μm) - 125-150 -

Grain size Vermiculite (μm) - - 75-125

b ave（%） 32.14 31.77 31.22

ckeff ±STDEV (10-13 m2) 11.15±0.81 7.19±0.49 7.67±0.76

dαL (m) 0.006 0.019 0.055

dDL (10-8 m2/s) 1.74 5.46 15.80

Residence times (hours) 15.61 15.43 15.19

Cation Exchange Capacity

(meq/100g)

- - 43.20

aThe dash “−” indicates not applicable b ϕave: Average porosity c keff: Effective permeability

dL is local longitudinal dispersivity and DL is the longitudinal hydrodynamic dispersion

coefficient.

3.2.3. Water composition

We synthesized the Marcellus Shale waters based on the data from the literature

(Shih et al., 2015). Most of the chemical compositions are on the median concentration

range of flowback and produced waters. Groundwater composition was also based on

58

reported compositions in literature (Watkins and Cornuet, 2012) (Table 3.2). Most MSW

literature emphasized the high concentrations of Na, Ca, Ba, Sr (Chapman et al., 2012;

Olmstead et al., 2013; Patterson et al., 2017). Several studies have also documented trace

metals at concentration levels that are much higher than drinking water standards

(Abualfaraj et al., 2014; Haluszczak et al., 2013; Shih et al., 2015; Ziemkiewicz and He,

2015). In our study we chose a relatively high concentration level of trace metals to

represent the worst case scenario.

59

Table 3. 2 Compositions of background groundwater and Marcellus Shale waters (mg/L)

Species

Groundwater

Inlet

Groundwater Outlet MSW

Qtz Cal Vrm

pH 8.18±0.03 8.14±0.02 8.30±0.02 8.58±0.08 6.67

Br 0.06±0.02 0.04±0.02 0.07±0.03 0.05±0.02 888

Cl 32.80±1.13 32.62±0.58 32.96±0.50 33.81±0.77 37797

SO4 12.64±0.36 12.60±0.20 12.98±0.33 12.23±0.35 10.44

Na 20.91±0.91 19.66±0.36 22.33±0.31 20.26±0.46 13661.62

Ca 16.16±0.70 15.70±0.41 17.71±0.17 5.07±0.11 6220.32

Mg 2.30±0.17 2.65±0.10 2.26±0.02 8.78±0.16 579.05

K 2.69 ± 0.16 2.61±0.09 2.84±0.03 14.57±0.28 240.65

Ba 0.08 ± 0.01 0.09±0.01 0.07±0.01 0.03±0.01 1690.45

Sr 0.12 ± 0.02 0.12±0.03 0.13±0.01 0.05±0.01 1040.19

Mn 0.0170±0.0050 0.0160±0.0020 0.0020±0.0010 0.0020±0.0010 19.896

Zn 0.0056±0.0012 0.0058±0.0015 0.0015±0.0004 0.0017±0.0009 20.685

Cu 0.0014±0.0003 0.0016±0.0004 0.0017±0.0003 0.0004±0.0002 1.533

Pb 0.00005±0.00002 0.00004±0.00002 0.00005±0.00003 0.00002±0.00002 0.171

-The outlet pH, Ca, Mg, K, and trace metals are different among the three columns. - Anions (Br, Cl, SO4), cations (Na, Ca, Mg, K, Ba and Sr), and trace metals (Mn, Zn, Cu, and

Pb) were measured using different instruments and therefore have different significant numbers

because of different analysis approach.

60

3.2.4. Flow-through experiments

Background groundwater was injected continuously upward at the flow velocity of

0.247 m/day, which is within the typical range of groundwater flow velocity (Newell et al.,

1990b). The residence times approximate 15.50 hours and differ slightly for each column

due to the small difference in pore volumes. The groundwater was injected to displace the

initial pore fluid in columns and to pre-equilibrate with the solid minerals until the pH and

effluent chemical species became relatively stable. After that the MSW was injected at a

rate of 0.028 m/day (30.40 µl/min) for 7.70 hours in all three columns. (Maloney et al.,

2017). The flow rate of MSW release is chosen based on potential seepage rates through

fractures and faults into groundwater aquifers, which is smaller than the background

ground water rate. The seepage rate is typically slow due to the absence of driving force

(Freeze and Witherspoon, 1967; Milici and Swezey, 2006). A total of 14.00 ml of MSW

was injected into each column. Effluent water samples were taken every 30 minutes for a

total of ~310, ~310, and ~ 410 hours for the Qtz, Cal, and Vrm columns using an automatic

sampler. Values of pH were measured immediately after sample collection. The Vrm

column was monitored for longer time because it takes longer time to flush out

contaminants from clay-rich materials.

3.2.5. Reactive Transport Modeling (RTM)

The column experiments were simulated in one dimension using the extensively

used reactive transport code CrunchFlow (Steefel and Lasaga, 1994). An example

61

governing mass conservation equation for a primary species i that participates in both

mineral dissolution and precipitation and ion exchange reactions is as follows:

(3.1)

Here is porosity, Ci is the concentration (mol/m3 pore volume) of species i, t is

the time (s), Di is the diffusion/dispersion coefficient (m2/s), u is the flow velocity (m/s),

Nr is the total number of kinetic reactions that involve species i, vir is stoichiometric

coefficient of species i associated with reaction r, Rr is the rate of chemical reaction r in

which the species i is involved (mol/m3/s). Here kinetic reactions are mineral dissolution

and precipitation. Ion exchange is fast and is typically equilibrium-controlled. This

equation implies that mass change rate of species i depends on diffusion/dispersion

represented by the first term in the right hand side (rhs), advection described by the second

term in the rhs, and reactions described by the third term. The term represents change

of mass on the solid phase with being solid bulk density (g/ m3 pore volume) and

being solid phase concentration of species i (mol/g). This term acts as a storage term taking

into account mass accumulation of i on the solid phase (Valocchi et al., 1981). The aqueous

and solid concentrations are related through the mass laws of ion exchange, as will be

discussed later. The geochemical system here includes 16 primary species, 14 secondary

species, 15 kinetic reactions (Table A1, Appendix A) and 11 ion exchange reactions in the

form of half-reactions defined in CrunchFlow (Table A2, Appendix A). The full ion

exchange reaction is generated by combing two half reactions.

Solid solution partitioning. The solid solution partitioning can occur when trace

metals (Me) substitute for Ca in the lattice of calcite. The newly formed solid phase can be

1

( ){ ( ) }

rN

i ii i i ir r

r

C SD C uC v R

t t

iS

t

iS

62

represented as Ca1-xMexCO3(s), where x is the fraction of sites occupied by Me (Glynn and

Reardon, 1990; Tesoriero and Pankow, 1996). The incorporation of trace metals into the

calcite structure via solid solution partitioning can significantly decrease the trace metal

concentration in natural aquifers (Andersson et al., 2016; Davis et al., 1987; Rimstidt et al.,

1998). In solid solution partitioning, the distribution coefficient DMe quantifies the

partition of trace metals between carbonate mineral and aqueous solution,

(3.2)

(3.3)

where XMeCO3 and XCaCO3 are the mole fractions of MeCO3 and CaCO3 in the solid

solution, respectively. Larger DMe value means preferential partitioning into calcite. Pb,

Mn, Zn and Cu are typical trace metals that can incorporate into calcite. In this work, the

DMe values were calibrated using breakthrough curve data, with values of 19, 21, 20 and

80 for Pb, Mn, Zn and Cu, respectively. The detailed implementation of solid solution

partitioning in CrunchFlow is discussed in the supporting information.

Model Calibration. The model was calibrated for porosity and dispersivity using

the tracer data. The physical properties were then fixed and only reaction parameters were

changed to reproduce the chemistry data. In the Qtz column, the kinetic rate constants (k)

and specific surface area (SSA) were adjusted for mineral precipitation and dissolution

reaction, which are within the range reported in the literature (Table A2). These parameters

are critical in capturing the BTCs of reactive speices in the Qtz column. These parameters

were then used in the other two columns, although the volume fraction of minerals such as

2 2

3 3(s) (s) CaCO Me MeCO Ca

23( )

23( )

/ C

/ C

s

s

MeCO Me

Me

CaCO Ca

XD

X

63

calcite are different so the total surface area of calcite in different columns are actually

different. The calibration in the Cal and Vrm columns by varying k and SSA values over

orders of magnitude (data not shown here) indicates that including mineral dissolution /

precipitation is not sufficient to capture the breakthrough curves of Na, Ca, Mg, K, Ba, Sr

and trace metals. We therefore introduced the solid solution partitioning in the Cal column,

which reproduced the trace metal data by adjusting DMe. In the Vrm column, we use the

parameters for mineral dissolution and preciptiation from other columns and adjusted ion

exchange selectivity coefficients (Table A3) to capture the breakthrough curves.

3.2.6. Quantification of injection and outlet mass

For each column, the total injection and outlet mass (mg) were calculated as follows

to quantify the amount of solute retention in each column:

(3.4)

(3.5)

where Q is the flow rate (L/h); Ci is the concentration (mg/L) of species i in the

outlet; T is the total running time (h) of the column experiment; Ci,GW and Ci,MSW are the

concentrations of species i in groundwater and MSW, respectively; VGW and VMSW are the

total injected volume of groundwater and MSW, respectively.

, ,Minj i GW GW i MSW MSWC V C V

,0

MT

out i tQ C dt

64


3.3.1. Difference in column physical properties

Figure 3. 1. Bromide breakthrough curves (BTCs) for Qtz (blue), Cal (green), and Vrm

(red) from experiments (dots) and from simulations (lines). The BTC of the Vrm column

is much wider than the other two columns, indicating a more heterogeneous column than

the other two due to the large contrast in grain size and property between quartz (350-420

um) and vermiculite (75-150 um).

Breakthrough curves (BTCs) for bromide from experimental and modeling outputs

show good agreement (Figure 3.1). Although all columns were packed with uniform

distribution of minerals, the BTCs differ between columns. The Qtz column has a “narrow”

breakthrough whereas that of the Vrm is much wider with early breakthrough and long tail.

The dispersivity values that reproduced the experimental data vary from 0.006 m for the

Qtz column to 0.055 m for the Vrm column. This observed higher physical heterogeneity

in Vrm column aligns with conclusions from other studies that clay-rich porous media in

65

general are more tortuous than those composed of sand and calcite (Heidari and Li, 2014;

Latour et al., 1995; Shen and Chen, 2007; Wang and Li, 2015b).

3.3.2. Temporal evolution of pH

Figure 3. 2. Temporal evolution of inlet (dash lines) and outlet (dots) pH in (A) Qtz (blue),

Cal (green) and Vrm (red) columns before, during, and after a MSW release for about 0.48

residence times; Although the inlet pH in groundwater was ~ 8.2, the outlet pH varied

significantly due to different reactions in different columns. The outlet pH decreases in the

Qtz column while increases in the Cal and Vrm columns. The Qtz and Vrm columns

“recover” quickly from the MSW perturbation compared to the Cal column. (B) modeling

output (lines) for Vrm column under three cases with different processes in the model: PPT

for mineral dissolution/precipitation only, IEX for ion exchange only, and IEX+PPT for

ion exchange with mineral dissolution/precipitation. The IEX+PPT line overlaps with the

IEX line, indicating the dominant role of IEX in determining pH in the Vrm column.

The inlet pH values were managed to be around 8.2 (Figure 3.2). During the MSW

injection, the outlet pH in all columns decreases because of the lower MSW pH. The

effluent pH in Qtz column returns to the background pH quickly whereas it takes much

longer time for the Cal and Vrm columns to return to their background pH (Figure 3.2).

The outlet pH of the Qtz column after release remains almost the same as the inlet pH,

66

which is expected as quartz is largely non-reactive. The Cal column increases slightly from

8.20 to 8.30. The model reproduces the pH trend before, during, and after the MSW

injection. The simulation indicates that calcite was undersaturated and dissolving, which

explains the slight increase in Ca concentration and pH (Figure 3.5D and Table 3.2). As

shown in Figure 3.2B with three simulated scenarios with different processes, scenarios

with ion exchanges (IEX and IEX+PPT) reproduce the pH values well, whereas the PPT-

only case suggests much lower effluent pH that is the same as inlet pH. This indicates that

ion exchange plays a dominant role in controlling pH in the Vrm column.

67

3.3.3. Reactive transport of trace metals in columns

Figure 3. 3. Left: Breakthrough data (dots) and modeling output (lines) of metals in Qtz

(blue), Cal (green), Vrm (red) columns; right: comparison of modeling output in the Vrm

column under three scenarios (including mineral dissolution/precipitation (PPT only), ion

exchange without mineral dissolution/precipitation (IEX only), and ion exchange with

mineral dissolution/precipitation (IEX+PPT)). The comparison indicates that both ion

exchange and mineral precipitation contribute to the decrease of metals and their retention

within the column. Only a fraction of metal ions return back to the solution.

68

The trace metal BTCs are very different in the three columns (Figure 3.3). In

general, the peaks of the Cal and Qtz columns show similar timing but very different

magnitude, with much higher peaks and breakthrough in the Qtz column, indicating it

retains smaller amount of metals. The BTCs from the Vrm column are typically wider but

with much lower peaks. Reactive transport modeling indicates that both mineral

precipitation and ion exchange occur. In the Cu figure (Figure 3.3G), compared to scenario

with ion exchange reaction, the Cu profile without ion exchange reaction has much higher

concentration and fails to capture the tail, indicating the critical role of ion exchange

reaction on Cu reactive transport. Although not shown here, precipitate rates and

equilibrium constants have been changed by more than one order of magnitude but still

failed to capture the Cu tail. The model overestimated the peak concentration, possibly

because our one site ion exchange model does not represent the strong sorption of Cu on

clay edges that can further reduce its aqueous concentration (Malandrino et al., 2006). The

model can generally capture the BTCs in the Vrm column when both ion exchange and

mineral reactions (IEX+PPT) are included; precipitation alone cannot capture the BTCs of

trace metals (Figure 3.3). The relative significance of the two reactions varies depending

on specific species. The calculated saturation indices of all potentially-precipitating

minerals are listed in Table 3.3. In each column, there are minerals with positive saturation

indices, indicating mineral precipitation in all three columns. In the Qtz column, metals

precipitate mostly as hydroxides whereas mostly as carbonate in the Cal column.

69

Table 3. 3 Calculated saturation index during the MSW release

Minerals Vrm Cal Qtz

Trace metals: carbonates

MnCO3 1.377 1.337 0.938

ZnCO3 0.677 1.095 -1.856

CuCO3 -2.147 -1.195 -2.949

PbCO3 -0.818 0.761 -0.564

Trace metals: hydroxide

Mn(OH)2 -1.168 -0.521 -1.012

Zn(OH)2 -0.286 -0.31 0.233

Cu(OH)2 1.607 0.862 1.586

Pb(OH)2 0.456 0.466 1.114

Ba, Sr and Ca

BaSO4 0.882 1.359 1.907

SrSO4 -2.892 -3.254 -2.413

CaSO4 -3.108 -3.313 -2.501

BaCO3 2.508 2.214 1.87

SrCO3 1.402 1.109 0.765

CaCO3 2.062 1.792 1.448

70

In the Cal column, both mineral precipitation and solid solution partitioning need

to be included to reproduce the trace metal breakthrough. Among all trace metals, Cu has

the largest distribution coefficient into calcite. The model observed the precipitation of

MnCO3, PbCO3, CuCO3, and ZnCO3. We cannot capture the peaks of trace metal

breakthrough curves by using mineral precipitation only. This is consistent with

observations in literatures that the incorporation of trace metals into calcite lattice through

solid solution partitioning can form Ca1-xMexCO3 (Andersson et al., 2016; Rimstidt et al.,

1998; Tesoriero, 1994; Tesoriero and Pankow, 1996).

71

3.3.4. Reactive transport of Ba, Sr and SO4

Figure 3. 4. Left: Breakthrough data (dots) and model output (lines) of (A) SO4, (B) Ba,

and (C) Sr experimental data with the right: comparison of three cases with different

process scenarios in the Vrm column (D) SO4, (E) Ba, and (F) Sr. In the Vrm column,

sulfate remains the same as inlet, indicating that barite and celestite precipitation do not

occur and Ba and Sr are exchanged onto vermiculite, which gradually release out later over

a long period of time. In the Cal and Qtz columns, sulfate concentration decreases sharply

during MSW release, indicating the precipitation of sulfate-containing minerals.

As shown in Figure 3.4A and 3.4D, outlet SO4 concentrations in the Vrm column

remained similar to the inlet during the MSW release. The outlet Ba and Sr decrease

significantly at first, and slowly increase at later time (Figure 3.4B and 3.4C), indicating

72

that Ba and Sr are immobilized first via ion exchange reactions, leaving not much Ba and

Sr for SO4 for mineral precipitation. As such, SO4 is negligibly retained in the Vrm column.

The right panels in Figure 3.4 show no significant difference between IEX and IEX+PPT

cases, further confirming the predominant role of ion exchange.

In the Qtz and Cal columns, however, mineral precipitation contributes to the

lowering of Ba and Sr concentrations. Table 3.3 indicates that Ba precipitates as BaSO4

(SI=1.907) while Sr precipitates as SrCO3 (SI values are 1.109 and 0.765 in Cal and Qtz

columns, respectively) instead of SrSO4 (Its SI is -3.254 and -2.413 in Cal and Qtz columns,

respectively), which agrees with findings from the literature (Vidic, 2015). Ba reacts

rapidly with SO4 within 30 min while Sr takes days to reach equilibrium. The incorporation

of Ba and Sr into calcite lattice in the Cal column is not important. The breakthrough curves

of Ba and Sr, however, are still similar to a tracer, because sulfate concentration is about 2

orders of magnitude lower than Ba and Sr in MSW (Table 3.2) so that it cannot lower their

concentrations significantly. In this case, SO4 precipitates and is retained more in the Qtz

and Cal columns than that in the Vrm column.

73

3.3.5. Reactive transport of Na, Ca, Mg, and K in columns

Figure 3. 5. Breakthrough data (dots) and modeling output (lines) of (A) Na, (B) Ca, (C)

Mg, and (D) K in Qtz (blue), Cal (green), Vrm (red) columns. Presorbed Mg and K are ion

exchanged out from the clay so their concentrations increase. After MSW release, sorbed

Na is slowly released back to the aqueous leading to a long tail.

As shown in Figure 3.5, Mg and K concentrations increase significantly during the

MSW injection in the Vrm column but not in other columns. On the other hand, Na and Ca

decrease most in the Vrm column among the three columns. This is because Mg and K are

the exchangeable cations in the interlayers of vermiculite. During the MSW injection, high

concentrations of Na and Ca displace out Mg and K. After the release, the sorbed Na slowly

release back to the aqueous phase leading to an extended rising tail. Correspondingly, Ca,

Mg and K are exchanged onto vermiculite and their concentrations decrease to levels below

the background concentration. The system slowly relaxes back to the background condition.

74

This suggests that the dominance of ion exchange in the Vrm column in controlling the

BTCs of Na, Ca, Mg, and K. In contrast, in the Qtz and Cal columns, these cations behave

relatively similarly and do not have such dramatic changes in concentrations except that

species in the Cal column show long tails. In natural aquifers, clay minerals often have

presorbed cations. When MSW release occurs, the high concentration of Na from MSWs

can compete with presorbed cations and exchange them out, as has been reported by Sang

et al. (2014. These authors showed that 32-36% presorbed trace metals were exchanged

out and mobilized from colloids due to the intrusion of flowback water.

3.3.6. Chemical retention in columns

Figure 3. 6. Injected and outlet mass of species among Qtz (blue), Cal (green) and Vrm

(red) columns on logarithmic scale (A) Trace metals (Mn, Cu, Zn and Pb); (B) Anions and

cations (Br, Cl, Na, Ca, Mg, K, Ba, Sr, and SO4).

Figure 3.6 compares injected and outlet mass of each chemical in each column.

Species that have relatively similar injected and outlet mass fall on the 1:1 lines, meaning

75

almost all injected chemicals are flushed out after about 20 residence times. Species that

are partially retained in the columns fall below the 1:1 line. As expected, for non-reactive

tracers such as bromide and chloride, they are on the 1:1 line in all three columns. Reactive

species behave very differently across the three columns. In general, most species in the

Qtz column are on the 1:1 line as they are mostly flushed out. The ones that are not on the

1:1 line are Zn, Cu, Pb, and SO4 because they precipitate out as hydroxides and sulfate

minerals even in the Qtz column. In the Cal column, trace metals and Ba, Sr, and SO4

precipitate so they fall below the 1:1 line. Most trace metals are retained in the column at

about ~90%, as indicated by the 1:10 line. The Vrm column has the most dramatic

difference between the injected and flushed masses. The most notable ones are Sr and Ba

with outlet mass of about 1.5 orders of magnitude lower than the injected mass, indicating

only about 5% are flushed out. All major cations are mostly trapped, falling close to the

1:10 lines. Mg and K are about one order of magnitude higher than those from the injected

mass because of the displacement by Na and Ca. Sulfate is all flushed out because clay

does not retain much of SO4. We did not observe changes in permeability during the

experiment. Because of the relatively low concentrations, the precipitate mass calculated

based on the mass difference from inlet and outlet correspond to volumes of 2.19×10-3 cm3,

3.62×10-3 cm3, 1.43×10-4 cm3 compared with the corresponding pore volume of 326 cm3,

321 cm3, 316 cm3 in Qtz, Cal, and Vrm columns.

76

3.3.7. Discussion

In the experimental work presented here, we chose a representative set of MSW

data at the median range of reported values, a few end-member mineralogy to represent

aquifer rock compositions, and one inlet groundwater composition. In natural systems the

environmental conditions can vary significantly, including variations in composition of

MSW, groundwater, and mineralogy, among others. Water-rock interactions can be

affected by other factors such as redox states, organic carbon, and microbe-mediated

reactions. These variations can have significant impacts on the transport, reactions, and

retention of chemicals. For example, under low pH conditions, less precipitates will form,

which can reduce the chemical retention in the solid via mineral precipitation. In Cal

column, however, calcite can dissolve more under low pH and still drive to higher pH

conditions (Wen et al., 2016a). Its dissolution may offer more carbonate for trace metals to

precipitate out therefore retaining more trace metals in the Cal column. In the Vrm column,

abundant H+ at low pH can lead to less sorption of positively-charged metals on solid

phases (Malandrino et al., 2006). However, it is not our intention here to exhaustively study

all variables. Our goal here to understand dominant processes in mineralogically different

aquifer systems, identify key differences in the reactive transport of chemicals, and explore

how these process differences lead to differences in contaminant retention and retardation.

The environmental geochemical community in generally know that clay-rich media

lead to sorption (e.g., surface complexation and ion exchange) and therefore retardation,

and that quartz media result in minimum water-rock interactions. The experimental data

and modeling results here however revealed several important insights that differ from

77

previous thoughts. Our results show that the reaction mechanisms of MSW chemicals are

much more complex. For example, in clay rich media, we observed that trace metals

participate not only in the ion exchange but also in mineral precipitation. In fact, the

majority of metals is retained in the solid via mineral precipitation, which is surprising

because typically we expect clay-rich aquifers would retard metals. Even in the Qtz column,

the trace metals were retained by 20 ~ 70%, which is surprising. In the Cal columns, the

trace metals are retained not only through precipitation but also solid solution partitioning,

which lead to a total of 75 ~ 99% retention. All these results differ from previous thoughts.

Although the experiments focus on a few specific conditions, the reactive transport

model developed here however overcome such limitations. The model quantitatively

differentiates the relative importance of multiple processes, therefore helps understand the

reaction mechanisms and predicts the transport and fate of chemicals under a wide range

of different conditions. The model therefore bridges laboratory work and natural conditions,

especially where relatively limited knowledge and data limit insights and prediction of

complex mineral-rock and contaminant interactions.

3.4. Conclusions

Here we use column experiments and reactive transport modeling to understand the

role of aquifer mineralogy in determining the reactive transport and fate of chemical

species from MSW. Results show that mineralogy exerts a significant control on the types

of reactions that occur and the extent of solute immobilization. An interesting example is

that Ba and Sr form precipitates in the Qtz and Cal columns, whereas mostly participate in

78

ion exchange reactions in the Vrm column. Although most chemicals behave similarly to

a conservative tracer in the non-reactive Qtz column, they are at least partially immobilized

in the more reactive Cal and Vrm columns through the formation of carbonate, hydroxide,

and sulfate precipitates. In the Vrm column, many species also participate in ion exchange

reactions, leading to the slow flushing of low level chemicals over tens of residence times.

However, metals retained in the Vrm column through the mineral precipitation do not

release back to the aqueous after MSW stopped. Trace metal species partition into

carbonate through solid solution partitioning in the Cal column, which does not occur in

other columns.

These results have interesting environmental implications in understanding the

natural attenuation processes and environmental monitoring. Reactive aquifers such as

carbonate and clay-rich aquifers tend to retain and retard most trace metals and Ba and Sr

that are characteristics of MSWs. In clay-rich aquifers, however, many contaminants tend

to linger at some level as ion exchange reaction slowly release the chemicals, posing a

long-term risk for water resources (Akob et al., 2016; Cozzarelli et al., 2017). The

immobilization of chemicals in precipitates reduce aqueous concentration and temporarily

reduces water quality risks. These chemicals however may be subject to mobilization again

when perturbation occurs (Frye et al., 2012). In addition, natural systems often contain pre-

sorbed chemicals. This work indicates that the highly saline MSWs tend to displace out

pre-sorbed chemicals. In this work, these chemicals are Mg and K. If metals and other

contaminants pre-sorb on clays, their mobilization can have large impacts on water quality.

The results also indicate that once aquifers are exposed to MSWs, the majority of the trace

79

metals can remain in reactive aquifers for a long time as precipitates. Although this means

lower impacted water quality in the short term, it imposes long-term risks.

This work also applied the multi-component reactive transport modeling to quantify

the relative importance of different reaction processes and identify the dominant reaction

mechanism. Reactive transport modeling solves governing equations that couple flow,

transport, and multi-component reactions that are relevant to the environmental fate of

chemicals (Lichtner, 1985). It has been used for applications including leakage detection

(Zhang et al., 2014) and remediation (Bao et al., 2014; Li et al., 2017; Steefel et al., 2015)

in natural environments. Given the process-based understanding and the integration with

data, reactive transport modeling can be used to extrapolate water quality under conditions

that experiments are not carried out, therefore providing a powerful tool for environmental

risk assessment.

Acknowledgments

Coauthors including Dr. Li Li, Dr. Hang Wen, and Dr. Sridhar Komarneni are appreciated.

We acknowledge Matthew Gonzales and Laura Liermann from College of Earth and

Mineral Sciences in providing help for the analyses of cations including trace metals. Xin

Gu from Department of Geoscience and Huaibin Zhang from College of Agricultural

Sciences assisted with the vermiculite sample analysis. Sruthi Kakuturu helped with

sample collection. This work was supported by the U.S. Department of Energy (DOE)

Subsurface Biogeochemistry Research program DE-SC0007056. The findings and

conclusions here do not necessarily reflect the view of the funding agency. We

acknowledge the Co Editor-in-Chief Dr. Jay Gan for handling the manuscript and three

80

anonymous reviewers for their constructive comments that have significantly improved the

manuscript.

81

Chapter 4

Controls of Mineral Spatial Patterns on the Reactive Transport of Marcellus

Shale Waters

The work of this chapter has been submitted to Energy & Fuels, 2018.

82

Abstract

This work examines the largely unexplored role of mineral distribution patterns in

determining reactive transport of Marcellus Shale waters in heterogeneous aquifers. Two

two-dimensional heterogeneous cells (40 cm by 12 cm by 1 cm) were built with the same

amount of clay (vermiculite) embedded in quartz however with different spatial patters of

clay: the “1/4-zone” and “1/2-zone” cells have rectangular vermiculite clusters at a quarter

and a half lengths of the cells, respectively. The reactive transport processes in these cells

were compared to those of a “Uniform” column with uniformly distributed vermiculite and

quartz and the same vermiculite-to-quartz mass ratio. Effluent chemistry data show that the

heterogeneous patterns reduce the trace metal (Mn, Pb, Zn, Cu) retention (41-86% and 23-

69% in the 1/4-zone and 1/2-zone cells) through mineral precipitation, compared to 74-93%

in the uniform case. Pre-sorbed Mg and K are exchanged out by 7-10 times more in the

Uniform column than those heterogeneous cells. Spatial patterns also regulate the dominant

reactions: Ba mostly precipitates as barite in the heterogeneous cells whereas it mostly

exchanges onto vermiculite in the Uniform case. These findings underscore the importance

of spatial patterns in controlling rates and types of reactions and ultimately transport and

fate of chemicals in natural water systems. These findings have significant implications on

predicting natural attenuation and assessing risks in the natural subsurface.

4.1. Introduction

Marcellus Shale gas extraction leads to the production of Marcellus Shale Waters

(MSWs), here defined as including flowback waters from hydraulic fracturing and

83

produced waters during shale gas extraction. These waters are characterized by high

organic content, total dissolved solids (TDS) (usually >200,000 mg/L), elevated

concentrations of anions, major cations, and trace metals (Barbot et al., 2013b; Chapman

et al., 2012; Haluszczak et al., 2013; Olmstead et al., 2013; Shih et al., 2015). Accidental

releases of MSWs have been reported to occur and can pose high environmental risks on

natural water resources (Brantley et al., 2014b; Osborn et al., 2011b; Vengosh et al., 2014;

Warner et al., 2012a). It is important to understand how chemicals in MSWs react with

minerals in groundwater systems, which determine the retention, release, and attenuation

of chemicals. The interactions of MSWs with minerals and different types of waters (e.g.,

acid mine drainage water) have been extensively studied in well-mixed batch reactor

systems (Kondash et al., 2013; Liberati, 2015; Trefry and Trocine, 2011). In our previous

work, we have also examined the role of aquifer mineralogy in controlling MSW-mineral

interactions and the response of water chemistry to pulsed MSW release in homogeneous

columns where minerals are uniformly distributed (Cai et al., 2018a; Cai et al., 2018b).

Natural groundwater systems however generally exhibit spatial heterogeneity with

co-occurring minerals of varying reactivity and water-conducting capacity distributed in

different spatial patterns. For example, aquifers typically have low-permeability clay lenses

embedded in layers of permeable zones (Bertoldi et al., 1991; Zheng and Gorelick, 2003).

Spatial variations in permeability leads to the formation of preferential flow paths (Cortis

et al., 2004; Dagan, 1990; Gelhar and Axness, 1983; Gelhar et al., 1992; Neuman and

Tartakovsky, 2009) and therefore different contact times between water and reacting

minerals (Wen and Li, 2017). Mounting evidence has shown that mineral spatial patterns

play a critical role in determining the extent of reactions (Al-Khulaifi et al., 2017; Atchley

84

et al., 2013; Li et al., 2014a; Liu et al., 2014; Perujo et al., 2017; Rolle et al., 2009;

Salehikhoo et al., 2013a). Wang and Li (2015b) observed 1.4 order of magnitude lower

Cr(VI) adsorption in column packed with large illite clusters than those with evenly

distributed illite. Salehikhoo and Li (2015 found that magnesite dissolution rates can be 2

orders of magnitude lower in porous media with large clusters of low permeability

magnesite. These studies shed light on how and to what extent mineral spatial

heterogeneity governs geochemical reactions. These existing experimental studies have

primarily focused on reaction systems with relatively simple water chemistry. It is not clear

how and how much spatial heterogeneities influence reactions in complex water systems

such as those in MSWs with highly elevated concentrations of many chemicals (Figure

4.1.).

Figure 4. 1. Conceptual figure of deformed MSW plume and preferential flow path with

different mineral reactions in natural heterogeneous aquifer. The complexities of aquifer

may affect the ultimate reactive transport of chemicals upon the MSW release.

The objective of this work is to systematically examine the role of mineral spatial

patterns in determining the reactive transport of chemicals from MSWs in clay-rich

systems. We ask the question how and how much do mineral spatial patterns determine the

natural attenuation and retention of MSW chemicals? We use two 2D cells in this work

85

and compare them to a 1D Uniform column (Cai et al., 2018a). The “Uniform” column has

uniformly-distributed vermiculite within quartz sand; the “1/4-zone” cell has five

vermiculite zones at the length of 1/4 of the cell; The “1/2-zone” cell has vermiculite grains

distributed in two zones of 1/2 length of the cell. Both cells have the same vermiculite-to-

quartz mass ratios as the Uniform column except with different mineral spatial patterns.

Vermiculite is used as model clay here because it occurs ubiquitously in the natural

subsurface. The insights and principles gained here however should be applicable to other

types of clays as well.

4.2. Materials and methods

4.2.1. Mineral preparation

Mineral grains of 350 ~ 420 and 75 ~ 125 µm were used for quartz and vermiculite,

respectively, to represent the physical and geochemical characteristics of different minerals.

Vermiculite is a common clay mineral with layered structure (Jackson and Inch, 1989;

Rogers, 1989) and high Cation Exchange Capacity (CEC) (dos Anjos et al., 2014). We

choose vermiculite as the model clay because it does not swell as much as other clays and

does not cause complications arising from the formation of lumps and cracks and clogging

in the cells. Vermiculite samples were analyzed by LI-COR CO2–H2O Gas Analyzer (LI-

7000) in the Biogeochemistry Laboratory, Department of Crop and Soil Sciences, Penn

State (Bazilevskaya, 2015), which indicated 1.29% (mass) of calcite.

86

Figure 4. 2. (A) A schematic of 2D cell of 40.0 cm×12.0 cm×1.0 cm (1/2-zone), with 2

zones of clay (dark brown) embedded within quartz sand (light brown). Glass beads and

honeycomb were positioned at the bottom of the cell to generate homogeneous flow at the

entry point. The flow however did segregate within the cell due to the uneven distribution

of clay and quartz. (B) A picture of the flow-through experiments. The background

groundwater was injected to pre-equilibrate with minerals for about 6.0 residence times

before and after the injection of MSW pulse.

4.2.2. Two-dimensional cell design

The 2D cell (40.0 cm×12.0 cm×1.0 cm) consists of an inlet port, a chamber filled with

5 mm-diameter glass beads and a honeycomb with 188 hexagonal cells (2cm×2mm, cell

87

height/cell side length=10, space between cells are 0.5 mm), one transparent box, and one

outlet sampling port. The chamber of glass beads and honeycomb homogenized the flow

before entering the mineral packed cell. A cap with a 5 degree angle relative to the

horizontal direction was used to avoid dead end zones. A 30 micron

polytetrafluoroethylene (PTFE) frit was used to hold the mineral grains inside the cells. To

avoid water leakage, silicon gasket was installed along the whole cell wall.

4.2.3. Spatial distribution patterns and cell property measurement

The two 2D cells were wet packed. The packing procedure is detailed in the

Supporting Information. The 2D cells have the same vermiculite-to-quartz mass ratio

(0.087 ± 0.001) of as the 1D column. The cells differ from the Uniform column in spatial

distribution patterns characterized by relative correlation length (CL), defined as the

relative length of the clay zone versus the total length of the cell (L) in the main flow

direction. The uniform column (“Uniform”) packed by homogeneously mixing quartz and

vermiculite grains to the extent possible to maximize the clay-quartz contact was from our

previous work (Cai et al., 2018a). It can be considered as having an infinite number of

small vermiculite zones at the length of grain size. The 1/4-zone cell (“1/4-zone”) has five

rectangular zones of 10.00 cm × 2.24 cm with the length of the clay zone one-quarter of

the total cell length in the main flow direction. The clay-quartz geometric contact area is

122.40 cm2, much smaller compared to the calculated 3968.12 cm2 of geometric surface

area if well mixing the same vermiculite mass. The 1/2-zone cell (“1/2-zone”) has two

zones of 20.00 cm × 2.80 cm with the clay zone half of the cell length. This case has a clay-

88

quartz contact geometric area of 91.2 cm2. The calculated relative correlation lengths are

0.00028, 0.25, 0.50 for the Uniform, 1/4-zone, and 1/2-zone, respectively. The properties

of the heterogeneous cells differ due to different vermiculite spatial distribution pattern

(Table 4.1).

Table 4. 1 Physical and geochemical properties of the heterogeneous cells and 1D

Uniform column

Visualized Schematics

Cases Uniform 1/4-zone 1/2-zone

Quartz (gram) 1599 772 765

Vermiculite (gram) 140.20 66.13 65.64

Vermiculite mass percent (%) 8.10 7.89 7.90

Quartz grain size (μm) 350-420 350-420 350-420

Vermiculite grain size (μm) 75-125 75-125 75-125

Length of clay zone (cm)

Width of clay zone (cm)

a-

-

10.00

2.24

20.0

2.80

Relative CL (clay length/cell length) 0.00028 0.25 0.50

b ave (%) 31.22 34.72 34.50

ckeff (±STDEV) (10-13 m2) 7.67 (±0.76) 8.98 (±0.45) 10.09 (±0.37)

a. The dash “-” indicates not applicable b. ϕave: Average porosity

c. keff: Effective permeability; The permeability of the vermiculite zone and sand zone is estimated to be

0.13 × 10-12 m2 and 1.11 × 10-12 m2 based on our previous experiments.

89

4.2.4. Water composition

The Marcellus Shale water and groundwater were synthesized based on data from

literature (Shih et al., 2015; Watkins and Cornuet, 2012). Marcellus Shale waters have

much higher concentrations for all species compared to the groundwater (Table 4.2). We

chose a relatively high concentration level of trace metals to represent the worst case

scenario (Abualfaraj et al., 2014; Haluszczak et al., 2013; Shih et al., 2015; Ziemkiewicz

and He, 2015).

Table 4. 2 Compositions of groundwater and Marcellus Shale waters (mg/La)

Species Groundwater Inlet Groundwater Outlet MSW

Uniform 1/4-zone 1/2-zone

pH 8.13 (±0.03) 8.58 (±0.08) 8.00 (±0.09) 7.77 (±0.02) 6.67

Br 6.37 (±0.84) ×10-2 5.45 (±1.55) ×10-2 6.10 (±0.83) ×10-2 7.23 (±1.36) ×10-2 8.88×102

Cl 3.30 (±0.12) ×101 3.38 (±0.08) ×101 3.36 (±0.01) ×101 3.17 (±0.23) ×101 3.78×104

SO4 1.23 (±0.02) ×101 1.22 (±0.04) ×101 1.26 (±0.01) ×101 1.23 (±0.04) ×101 1.04×101

Na 2.09 (±0.10) ×101 2.03 (±0.05) ×101 1.98 (±0.03) ×101 2.05 (±0.02) ×101 1.37×104

Ca 1.74 (±0.05) ×101 0.51 (±0.01) ×101 1.36 (±0.01) ×101 1.37 (±0.01) ×101 6.22×103

Mg 2.19 (±0.04) ×100 8.78 (±0.16) ×100 5.43 (±0.03) ×100 5.15 (±0.02) ×100 5.79×102

K 0.28 (±0.01) ×101 1.46 (±0.03) ×101 1.18 (±0.02) ×101 0.69 (±0.03) ×101 2.41×102

Ba 8.49 (±0.14) ×10-2 3.45 (±0.11) ×10-2 7.83 (±1.35) ×10-2 7.48 (±0.49) ×10-2 1.69×103

Sr 1.32 (±0.01) ×10-1 4.83 (±0.15) ×10-2 1.27 (±0.01) ×10-1 1.31 (±0.01) ×10-1 1.04×103

Trace

Metals

Mn 1.70 (±0.50) ×10-2 0.23 (±0.06) ×10-3 1.15 (±0.02) ×10-2 2.40 (±0.50) ×10-2 1.98×101

Zn 5.61 (±1.21) ×10-3 1.70 (±0.90) ×10-3 1.83 (±0.12) ×10-3 4.49 (±0.26) ×10-3 2.06×101

Cu 1.90 (±0.50) ×10-3 0.40 (±0.20) ×10-3 1.75 (±0.12) ×10-3 1.85 (±0.34) ×10-3 1.53×100

Pb 5.40 (±1.80) ×10-5 2.00 (±2.00) ×10-5 4.00 (±1.25) ×10-5 4.20 (±1.07) ×10-5 1.71×10-1

Cd 5.02 (±2.00) ×10-5 UDLb 2.20 (±0.37) ×10-5 4.08 (±0.93) ×10-5 1.22×10-1

90

4.2.5. Flow-through experiments

Groundwater was injected upward into the heterogeneous cells at a flow velocity

of 0.197 m/day. The residence times are 15.99 ±0.68 hours in the 1D Uniform column and

2D heterogeneous cells. The groundwater was injected to pre-equilibrate with solid phase

for 6 residence times until the effluent pH and chemical species were stabilized. The MSW

was then injected at a rate of 0.017 m/day (18.40 µl/min) for 7.70 hours. The ratio of

column/cell size-to-MSW injection volume was kept to be 62±9 between the 2D cells and

1D column. Outlet water samples were collected every 45 minutes for a total of ~ 380 hours

(~16 days) using the automatic sampler. The pH of each water sample was immediately

measured after each collection.

4.2.6. Chemical analysis

All collected effluent samples were filtered through a 0.22 micron membrane. Anions

(Br, Cl, SO4), cations (Na, Ca, Mg, K, Ba and Sr), and trace metals (Mn, Zn, Cu, and Pb)

were measured using Dionex DX120 ion chromatograph, Inductively Coupled Plasma

Atomic Emission Spectrometer (ICP-AES), and Inductively Coupled Plasma Mass

Spectrometry (ICP-MS), respectively.

4.2.7. Quantification of inlet and outlet mass

For each 2D cell, the total injection and outlet mass (mg) were calculated after the

MSW release experiments as follows:

91

(4.1)

(4.2)

where Q is the flow rate (L/h); Ci,t is the outlet concentration (mg/L) of species i at time t;

T is the time needed for chemicals to return to its background condition in each 2D cell;

Ci,GW and Ci,MSW are the concentrations of species i in groundwater and MSW, respectively;

VGW and VMSW are the total injected volume of groundwater and MSW, respectively.


Figure 4. 3. Temporal evolution of inlet (dash lines) and outlet (dots with connected lines)

(A) Br and (B) pH in the Uniform (blue), 1/4-zone (green) and 1/2-zone (red) cases before

and after a MSW release between 0 and 0.50 residence times. The C0 represents the inlet

concentrations during the MSWs leakage. Br in the Uniform column has the shortest

breakthrough tail compared to the other two heterogeneous cells. Although the inlet pH

was managed to be around 8.13, outlet pH and Br vary significantly due to different

vermiculite spatial patterns and different extent of mineral-water interactions. Values of

outlet pH are higher than inlet pH in the Uniform column and are lower than the inlet pH

in the 1/4-zone and 1/2-zone cells. In the Uniform column, pH returns to the pre-injection

condition faster than in the other two heterogeneous cells.

, ,Minj i GW GW i MSW MSWC V C V

,0

MT

out i tQ C dt

92

4.3.1. Physical property differences

As indicated in Figure 4.3A, the breakthrough curves (BTCs) of Br are very different

and indicate different physical properties in three cases. In the Uniform column, the Br

BTC is narrow and returns to background condition within 4 residence times. It is however

unsymmetric and does not follow the symmetric shape of a standard advection-dispersion

model, indicating a slight extent of heterogeneities although it was meant to be completely

homogeneous. In the 1/4-zone and 1/2-zone cells, the Br BTCs are similar as the Uniform

column in early times however have much longer tails and take more than 10 residence

times to drop to the background level. In particular, the 1/2-zone exhibits two apparent

peaks, demonstrating essentially “bimodal” distribution of flow velocities. The large peak

occurs first corresponding to the large water volumes coming out from highly permeable

sand zones compared to the much smaller peak with lower flow velocity coming out of low

permeability vermiculite zone (Ramasomanana et al., 2013).

4.3.2. Temporal evolution of pH

The inlet pH values were managed to be similar around 8.13. During the MSW release,

the outlet pH in all cases decreases because of the low MSW pH. The pH in the Uniform

column recovers quickly compared to the 1/4-zone and 1/2-zonecells. In particular, the 1/2-

zone cell has a wide BTC and a second peak between 1 to 2 residence times. The outlet pH

in the Uniform column increases to 8.58 while the other two do not increase as much. As

the exchangeable cation in the interlayers of vermiculite, the outlet Mg increases (from

2.19 mg/L at inlet) to 8.78 mg/L (Table 4.2) in the Uniform column compared to 5.43 and

93

5.15 mg/L in the 1/4-zone and 1/2-zone cells, respectively. Similarly, outlet K increases to

14.57, 11.89, and 6.90 mg/L in the cases of Uniform, 1/4-zone and 1/2-zone, respectively.

Figure 4. 4 Breakthrough curves of (A) Zn, (B) Pb, (C) Cu, and (D) Mn (dots with

connected lines) in the three cases. The three solid light lines are Br BTCs for comparison.

Cd was also measured but not shown here. Gray dash line represents the inlet. Trace metals

have the lowest peaks and are retained the most in the Uniform column compared to the

other two heterogeneous cells.

4.3.3. Reactive transport of trace metals

The first peaks of BTCs occur at similar residence times however with different

magnitudes in three cases. The Uniform column has the lowest concentration levels of Cu

and Mn, indicating it retains these trace metals the most. The Zn and Pb however have

relatively similar concentration levels in all three cases. The 1/2-zone cell typically has the

highest concentration level except Pb and retains the least. The trace metals have a second

94

peak in the 1/2-zone cell that is consistent with the Br BTC. As indicated in Table 4.3, the

saturation index of trace metals calculated from the measured water chemistry are mostly

positive and highest in the Uniform column among all three cases, indicating mineral

precipitation. In the 1/2-zone and 1/4-zone cells, almost all saturation indexes of carbonates

and hydroxides are negative suggesting mineral dissolution instead of precipitation. This

may be caused by different dissolving extent of calcite (1.20 % mass in vermiculite).

Because much less water passes through the vermiculite zones that has calcite in the 1/4-

zone and 1/2-zone cells, less calcite dissolves (Table B1), leading to lower pH and

carbonate concentrations and therefore much smaller saturation index for potential

precipitates for MSWs leakage (Table 4.3). Instead, ion exchange reaction may play a

critical role in the trace metal decrease for the 1/4-zone and 1/2-zone cells, as indicated by

the slow release of low level metals over a long period of time, which is similar to the

observation in our previous work (Cai et al., 2018a). The higher peaks in 1/2-zone also

indicate lowest extent of ion exchange in the 1/2-zone cell.

95

Table 4. 3 Saturation index of minerals during the MSW release

Minerals Uniform 1/4-zone 1/2-zone

Trace metals: Carbonates

MnCO3 1.37 0.91 0.66

ZnCO3 0.67 -2.77 -3.06

PbCO3 -0.81 -0.86 -0.95

CuCO3 -2.14 -2.2 -2.4

Trace metals: Hydroxide

Mn(OH)2 -1.16 -3.89 -4.3

Zn(OH)2 -0.28 -0.33 -0.77

Pb(OH)2 0.45 -5.94 -6.36

Cu(OH)2 1.6 0.75 0.42

Ba, Sr and Ca

BaSO4 0.88 3.13 3.08

SrSO4 -2.89 -0.34 -0.4

CaSO4 -3.1 -3.1 -3.1

BaCO3 2.5 0.43 0.13

SrCO3 1.4 0.76 0.45

CaCO3 2.06 1.21 0.95

Note: Saturation index (SI= log(IAP/Ksp) with IAP meaning ion activity product, Ksp

representing the solubility product) is an index showing whether a mineral will tend to

precipitate or dissolve in the solution. If SI >0, the mineral may precipitate. When SI <0, the

mineral may dissolve. If SI=0, it means the solution and mineral are at chemical equilibrium.

96

Figure 4. 5. Breakthrough curves of (A) SO4, (B) Ba, and (C) Sr from different cases. In

the Uniform column, SO4 concentrations remain similar to the inlet, indicating negligible

precipitation of sulfate-containing minerals (barite and celestite). Ba and Sr were exchanged on

vermiculite early and released out later, as indicated by the late time increase in the Uniform column.

In the 1/4-zone and 1/2-zone cells, sulfate concentration decreased sharply during MSW release,

indicating the precipitation of sulfate-containing minerals.

4.3.4. Reactive transport of Ba, Sr and SO4 in three cases

The outlet SO4 concentrations from the Uniform column remain similar to the inlet,

indicating negligible precipitation of sulfate-containing minerals including BaSO4 and

SrSO4 (saturation index of SrSO4 < 0, Table 4.3) (Figure 4.5). The outlet Ba and Sr

significantly decrease early on and increase later, indicating ion exchange reaction (Cai et

al., 2018a). In the 1/2-zone and 1/4-zone cells, however, SO4 concentration decreases while

Ba and Sr do not increase later, indicating the occurrence of precipitation. As indicated in

Table 3, saturation indexes of the precipitates BaSO4, BaCO3, and SrSO4 are positive. This

is because evenly distributed vermiculite in the Uniform column offers more water-

accessible exchangeable sites for Ba and Sr, leading to much lower concentration to

prevent the occurrence of BaSO4 and SrSO4 precipitation. This also indicates that although

the three cases are packed with the same vermiculite-to-quartz mass ratios, different

vermiculite spatial patterns can lead to the dominance of different types of reactions, which

97

has not been observed before. Existing studies on the role of spatial heterogeneity have

observed differences in rates and extent of reactions but not types of reactions (Li et al.,

2014a; Liu et al., 2014; Salehikhoo and Li, 2015; Wang and Li, 2015b; Wen and Li, 2017).

Figure 4. 6. Breakthrough data of (A) Na, (B) Ca, (C) Mg, and (D) K in the Uniform, 1/4-

zone and 1/2-zone cases. The extent of Mg and K increase vary among the three cases. In

the Uniform column, pre-sorbed Mg and K are ion exchanged out the most so their

concentration peaks are the highest and their mass increase by 7 to 10 times compared to

the 1/4-zone and 1/2-zonecells. Based on mass balance calculation, almost all sorbed Na is

released back to the water phase within 25 residence times in the Uniform column, while

it is still retained in the other two heterogeneous cells.

4.3.5. Reactive transport of Na, Ca, Mg, and K

During the injection of MSWs, high concentrations of Na and Ca displace out Mg

and K in the interlayers of vermiculite. Therefore, Mg and K concentrations increase during

the MSW release but vary among the three cases (Figure 4.6). In the Uniform column, the

98

released outlet mass of Mg and K due to ion exchange are 7 - 10 times larger compared to

the 1/2-zone and 1/4-zone cells. Mirroring the increase of Mg and K, Na and Ca decrease

in all three cases with corresponding magnitude in each case. This indicates that the

vermiculite spatial pattern regulates the extent of ion exchange. After the MSW release,

sorbed Na slowly releases back to the solution. In the Uniform column, the sorbed Na

decreases by 27.8% after 4 residence times and almost releases back to the solution by 25

residence times. However, ~ 10% Na are still retained in the 1/4-zone and 1/2-zonecells

after 25 residence times. This is because inner low-permeability vermiculite zone with

sorbed Na is less accessible so that it takes longer for desorbed Na to transport out of the

low-permeability vermiculite zone. This is similar to observations that low permeability

zones lead to much lower rates of U(VI) desorption than those with relatively homogeneous

flow (Liu et al., 2014).

4.3.6. Mass balance of chemicals in three cases

Figure 4.7A and 4.7B compare the inlet and outlet mass by 25 residence times. The

non-reactive species such as Br and Cl fall on the 1:1 line, meaning these species are all

flushed out and not retained in all cases. Species below the 1:1 line have less outlet mass

than inlet mass, indicating at least partially retained in the cases. The closer to the 1:1 line,

the less retention. Species above the 1:1 line have more outlet than inlet mass, indicating

mass addition to the water phase in the case. Although all cases were packed using the

same vermiculite-to-quartz mass ratio, the extent of retention / addition differs. In general,

the Uniform column has the highest extent of water-reactive mineral interactions so that

99

almost all chemicals deviate from the 1:1 lines. The trace metals are retained and fall close

to 1:10 line; Ba and Sr are exchanged and fall close to 1:10 line; Mg and K are displaced

out of the clay and are close to 10:1 lines. On the contrary, in the 1/4-zone and 1/2-zone

cells, because most water flows through the non-reactive quartz zone, water-mineral

reactions are minimized, leading to much less retention / addition (closest to the 1:1 line)

for almost all chemicals. The mass outfluxes of Mn, Zn, Cu, Cd, and Pb are 2 to 50 times

more from the 1/2-zone cell than from the Uniform column. Similar observations have been

documented for other trace metals. Species such as Ca in the 1/2-zone cell are close to the

1:1 line, indicating mostly flushed out. The differences between the 1/4-zone and 1/2-zone

are relatively minor compared their differences to the Uniform column.

The SO4 (blue cross) in the Uniform falls on the 1:1 line indicating negligible

precipitation of sulfate minerals. The retention of Ba (95.5%) is close to the 1:10 line,

primarily because the mass calculated is for within 25 residence times. As the breakthrough

curves indicate (Figure 4.5), more mass are flushed out in later times. In heterogeneous

cells, however, SO4 (green and red cross) falls below the 1:1 line due to sulfate mineral

precipitation. Based on the mass balance calculation, BaSO4 precipitation contributes 10.0%

to 36.9% retention of Ba in the 1/4-zone cell, and 14.0% to 32.5% retention of Ba in the

1/2-zone cell. This suggests chemical species experience different reaction types due to

mineral spatial patterns.

100

Figure 4. 7. Inlet and outlet mass of chemical species among the Uniform (blue), 1/4-zone

(green), and 1/2-zone (red) cases on logarithmic scale (A) Trace metals (Mn, Cu, Zn, Pb

and Cd); (B) Anions and cations (Br, Cl, Na, Ca, Mg, K, Ba, Sr and SO4). The inlet and

outlet mass in the Uniform column is proportionally scaled down. The retention of trace

metals is maximized in the Uniform column while minimized in the 1/4-zone and 1/2-zone

cells. The reaction extents are maximized therefore leading to largest increase of Mg and

K, and largest decrease of Ba, Sr, Ca, and Na in the Uniform column.

4.4. Conclusions

In this work, we use 2D heterogeneous cells and 1D Uniform column to

systematically understand the role of vermiculite spatial patterns in determining the reactive

transport and fate of chemical species from MSWs. Our results show that the extent of

MSW-mineral interactions is much more significant in the Uniform column that maximizes

the water-reactive mineral interactions than in the 1/2-zone and 1/4-zone cells. Specifically,

after 25 residence times, 14 - 77% of trace metals were flushed out of the 1/2-zone and 1/4-

zone cells, compared to 7 - 26% from the Uniform column. The outlet mass of Mg and K

from the Uniform column are about 10 and 7 times larger than those from the 1/2-zone and

101

1/4-zone cells, respectively. This is in accordance with the previous studies that underscore

the role of heterogeneity in reducing rates and extent of geochemical reactions.

The difference in the extent of MSW-mineral interaction is primarily caused by

differences in accessible ion exchange sites (IES) of vermiculite, although the total

vermiculite content is the same in all three cases. The measurement of separately packed

quartz and vermiculite columns has a permeability contrast of 0.11, with the permeability

of the vermiculite zone being 0.13 × 10-12 m2 and sand zone being 1.11 × 10-12 m2. This

means the flow velocities in the low-permeability vermiculite zone is much lower so that

the transport is dominated by diffusion, with the diffusion length (DL) being half width of

the vermiculite zone. Specifically, to access all IES within the vermiculite zone in the 1/4-

zone cell (DL=1.12 cm), the time needed is ~ 92.0 hours (=DL2/D* with a diffusion

coefficient D* of 3.8 × 10−10 m2/s), about 6 times longer than the residence time through

the whole cell (15.2 hours). As such, approximately 17% of IES are accessible. In the 1/2-

zone cell, only about 15% of IES are accessible, which is the smallest among the three

cases. Similarly, the “reactivity” is also much smaller for carbonate. The vermiculite used

here contains trace amount of carbonate, which is sufficient to dissolve out to increase pH

and precipitate trace metals in the Uniform column however not as much in the

heterogeneous cells. As a result, the trace metals are retained much less in the

heterogeneous cells.

This work for the first time shows that vermiculite spatial patterns regulate the

reaction type. In the Uniform column, Ba has negligible BaSO4 precipitates and primarily

exchanges onto the vermiculite and slowly releases back to the aqueous phase. In the 1/4-

zone and 1/2-zone cells, however, Ba participates in both mineral precipitation and ion

102

exchange reactions. Therefore, some Ba is trapped as mineral precipitates and does not

return to the aqueous phase. This differs from previous findings that spatial heterogeneities

mostly reduce reaction rates and lower the extent of reactions but not the reaction type (Li

et al., 2014a; Salehikhoo and Li, 2015; Salehikhoo et al., 2013a; Wang and Li, 2015b).

Clay minerals are often present as layered lenses in the natural subsurface. Results

here show the lower retention of trace metals and longer tailing in the 1/2-zone and 1/4-

zone cells, implying that the trace metals can linger for longer. The finding that the high

salinity of MSWs can replace out presorbed cations (Mg and K) from the clay mineral

suggests that presorbed contaminants, such as uranium (Alam and Cheng, 2014; Fox et al.,

2012), can be displaced as a result of MSW release.

Reactive minerals are often in the low permeability zone in natural subsurface.

Fractured system represents another example of fast-flowing water in highly conductive

fractures while most reactive minerals are in low permeability matrix (Wen et al., 2016a).

As such, the retention of contaminants through reacting with reactive minerals can be

overestimated by orders of magnitude by assuming homogeneous aquifer systems. With

smaller reactivity, the heterogeneous systems tend to see contaminants in water over a

much larger domain without precipitation and ion exchange. When ion exchange does

occur, it will also take much longer period of time for contaminants to be flushed out of

low permeability zones. Results from this study shed light on the importance of mineral

spatial patterns in understanding the mechanisms of and predicting the natural attenuation

of MSWs.

103

Acknowledgements

Coauthors including Dr. Li Li and Dr. Hang Wen are appreciated. We acknowledge

Matthew Gonzales and Laura Liermann from College of Earth and Mineral Sciences in

providing help for the analyses of cations including trace metals. Sruthi Kakuturu helped

with sample collection. We would also like to thank Travis Tasker from College of Civil

and Environmental Engin0eering in the discussion of the MSW composition. This

research was supported by the National Science Foundation (EAR-1452007). The

findings and conclusions here do not necessarily reflect the view of the funding agency.

104

Chapter 5 Conclusions and Future Work

105

In this research, we investigated the effects of time scales and magnitude of MSW

release, mineralogy, and spatial heterogeneity on the ultimate reactive transport, natural

attenuation and retention of complex chemical species from MSW release using reactive

transport model (RTM)-CrunchFlow, column experiments, and two-dimensional cell

experiments.

5.1. Time Scales and Magnitude of MSW Release under Various Natural Waters

The time scales and magnitude of MSW release on natural waters are quantified by

τrecovery, which is defined as the time needed for chemical species to recover to within 100±5%

of its background concentration, and Cmax, which is defined as the maximum observed

concentration during the release, respectively. In rivers and sand and gravel aquifers with

negligible clay content, mixing process controls Cmax and τrecovery of all chemical species

and they behave similarly as non-reactive tracers. The dilution factor determines Cmax while

τrecovery approximates their corresponding residence time. In clay-rich natural water systems,

ion exchange dominates when compared to mineral dissolution and precipitation. In

sandstone aquifers with rich clay mineral, Sr and Ba have two times much larger τrecovery

due to their higher affinity to the solid phase than Na, Ca, and Mg. This indicates that it is

more likely to detect pollution in clay-rich water systems when MSW release occurs

because it takes longer time for chemical species to return to its background condition. This

finding highlights the importance of RTM in process understanding, pollutant prediction,

and environmental impact quantification when MSW release to natural receiving waters.

106

5.2. Mineralogy

Mineralogy plays an important role in controlling the reaction types and the extent

of chemical species retention during the reactive transport of MSWs. For example, in the

Qtz and Cal columns, Ba and Sr participate in mineral precipitation, whereas mostly

participate in ion exchange reaction in the Vrm column with abundant clay content.

Although most chemical species behave similarly to the conservative tracer in the non-

reactive Qtz column, they are at least partially immobilized in the more reactive Cal and

Vrm columns through forming the carbonate, hydroxide, and sulfate precipitates. In clay-

rich Vrm column, trace metals participate in both ion exchange and mineral precipitation

reactions with around 50-90% immobilized in the column through mineral precipitation.

In the Cal column, trace metals are retained by around 75-99% through mineral

precipitation and solid solution partitioning. While the unreactive Qtz column retains the

least of trace metals among the three columns through the mineral precipitation. These

findings suggested that when MSW release occurs carbonate and clay-rich aquifers tend to

immobilize and retard most trace metals, Ba, and Sr. Contaminants that retarded by ion

exchange reaction in clay-rich aquifers tend to take longer time to release back to the

aqueous phase and as such posing a long-term risk on drinking water quality, which implies

that in clay-rich aquifers, a long-term field monitoring should be carried out to evaluate the

environmental impact if the MSW release occurs. Contaminants that retained through

mineral precipitation reduce the aqueous concentration and therefore water quality risks.

However, they may be re-mobile when release re-occurs. Significant amount of pre-sorbed

Mg and K is exchanged out from the solid phase to aqueous phase by high salinity MSWs.

107

Some aquifers consist of pre-sorbed naturally occurring trace metals and radioactive

elements (Ayotte et al., 2011). If high salinity MSW release occurs in such aquifers, these

elements may be displaced out and significantly deteriorate the drinking water quality

(Sang et al., 2014). This work also highlights the critic role of multi-component reactive

transport model in quantifying the relative importance of individual reaction process, and

in providing an insight for field monitoring, contaminant transport prediction, and

environmental risk assessment.

5.3. Spatial Heterogeneity

Spatial heterogeneity regulates not only the extent but also the types of MSW-

mineral interactions. The extent is maximized and much more significant in the Uniform

column than in the 1/2-zone and 1/4-zone cells. After 25 residence times, 2-3 times less

trace metals are flushed out of the Uniform column as compared to those of the 1/2-zone

and 1/4-zone cells, which implies the maximized retention in the Uniform column. The

pre-sorbed Mg and K from the Uniform column are displaced out ~ 10 and 7 times larger

than those from the 1/2-zone and 1/4-zone cells, respectively. This is in accordance with

previous literature findings that heterogeneity can reduce rates and lower extent of

geochemical reactions. However, our work here for the first time indicates vermiculite

spatial patterns can regulate the reaction types as well. In the Uniform column, Ba primarily

sorbs onto the vermiculite through the ion exchange reaction and then slowly releases back

to the aqueous phase rather than precipitates as BaSO4 being retained in cell. However, in

the 1/4-zone and 1/2-zone cells, Ba participates in both mineral precipitation and ion

108

exchange reaction. As such, some Ba is retained as mineral precipitates and does not release

back to the aqueous phase. This is different from the long known findings that

heterogeneities mostly reduce the rates and extents of reactions but not the reaction types.

Reactive minerals such as clays often distribute as low permeability zones in natural

subsurface. As such, without considering the spatial heterogeneity the pollutants that

retained through interacting with reactive minerals can be overestimated by orders of

magnitude and the time scales that needed for the impacted aquifers to return to background

condition can be underestimated. Meanwhile, if ion exchange reaction does occur, it will

take much longer time for pollutants to be flushed out of low permeability zones compared

to the homogeneous system. This work underscores the importance of mineral spatial

heterogeneity in understanding the processes that affect the natural attenuation and

retention and reactive transport and fate of complex chemical species after MSW occurs.

5.4. Future Work

Based on the current findings and limitation of this work, I list some points for

future work:

(i) Hydrogeochemical factors. In our work, we chose a representative set of MSW

chemicals, mineralogy, and groundwater chemistry. The water chemistry of

Marcellus shale flowback and produced waters, however, has spatial and temporal

variability during the shale gas extraction (Barbot et al., 2013b; Chapman et al.,

2012). Natural waters can also exhibit blends of different minerals. The interactions

between water and rock can be also affected by many other factors such as redox

109

state, saturation of exchangeable cations on clay minerals, organic carbon, microbe-

mediated reactions, etc. For example, the redox-sensitive species such as Mn can

have higher mobilization in an anoxic condition especially in the aquifers with

organic matter depleting the oxygen (Brumsack, 2006; Deutsch and Siegel, 1997;

Gounot, 1994). Lower pH condition helps reduce the retention of contaminants

through mineral precipitation. In a carbonate-rich aquifer, however, dissolution of

carbonate minerals can lead to increased pH and carbonate ion (CO32-)

concentration therefore leading to more precipitation of trace metals (Wen et al.,

2016a). While in a clay-rich aquifer, low pH can lead to less sorption of positively-

charged metals on solid phase making them much more mobile (Malandrino et al.,

2006). These various hydrogeochemical conditions can be considered in the further

experiment study.

(ii) Organic contaminants in MSWs. Organic contaminants include the compounds

originally from shale formations (e.g. benzene, toluene, ethlbenzene, and xylene

(BTEX), polycyclic aromatic hydrocarbons (PAHs)), hydraulic fracturing fluids

(e.g. 2-butoxyethanol), and downhole transformations of organics (e.g. halogenated

organic compounds) (Butkovskyi et al., 2017). The presence of organics and

organic acids, such as acetate, butyrate, and formate, in MSWs may reduce the

precipitation of Ba (Hakala et al., 2017) and trace metals (Park et al., 2011),

therefore enhancing their mobility in the aquifers and posing further risk on

drinking water quality upon the MSW release. This factor can be incorporated for

future study.

(iii) Randomly heterogeneous porous media. In our current work, vermiculite and

110

quartz were packed with regular mineral spatial patterns. The subsurface,

however,may exhibit strong heterogeneity. For example, the Macrodispersion

Experiment (MADE) site has a large variance of hydraulic conductivity of 4.5

(Rehfeldt et al., 1992), which is much larger than the other sites such as the Twin

Lake site (0.031) (Killey and Moltyaner, 1988), the Cape Cod site (0.26) (LeBlanc

et al., 1991), and the Borden site (0.29) (Sudicky, 1986). To extrapolate the

laboratory regular shaped heterogeneous setting to conditions with randomly

heterogeneous patterns, two-dimensional reactive transport modeling can be

developed and Monte-Carlo simulation can be introduced to generate random

spatial patterns at multiple permeability variances and correlation lengths.

Therefore a better environmental risk assessment and a general understanding of

reactive transport of complex chemical species from MSW release in randomly

and strongly heterogeneous subsurface can be achieved.

(iv) Factors considered for the risk assessment. In order to investigate the risk

assessment due to MSW release on drinking water quality, factors such as (1)

the distance between the wells and spills, (2) indicator of MSW contamination

in drinking water, (3) the real effect on drinking water, etc. can be considered

in the future study. For example, the nearer distance it has between the wells

and spills, the easier the groundwater wells will be contaminated. In

Pennsylvania, the distance of spills to surface water varies with the average

value of 268 m (Maloney et al., 2017). Moreover, the common depth to

groundwater table in Pennsylvania is shallow and ranges from 20 ft (6 m) to

250 ft (76 m). This further poses a higher risk on drinking water contamination.

111

As to the indicator of MSW contamination, we can choose unreactive chemical

species (e.g. Br or Cl) as the early indicator as their concentration are high in

MSWs but low in the groundwater. For example, Cl has a median value of 10.0

mg/L in groundwater in Pennsylvania (McBroom, 2013). In our study, during

the MSW release, we measured the maximum concentration of Cl is larger than

1895 mg/L in all experimental studies, which is ~190 times compared to the

median Cl concentration in aquifer in Pennsylvania. However, additional

chemical analysis, such as the 87Sr/86Sr ratio, typical chemicals (e.g. 2-

butoxyethanol) in hydraulic fracturing fluids, should be further analyzed to

discern the source of contamination, as in Pennsylvania there are other

contamination sources such as acid mine drainage, Appalachian brines. As to

the real effect on the drinking water quality, the release of MSWs can lead to

the rising chemical concentration in the aquifer therefore affecting the drinking

water quality. For example, when Ba exceeds 2 mg/L in drinking water, it can

result in kidney problems and high blood pressure to adults and delay the

physical or mental development to children (USEPA). In our experimental

study, Ba exceeds 2 mg/L with the duration of 1.21 residence time (RT) in Qtz

column, 1.49 RT in Cal column, 0 RT in Vrm column, 2.14 RT in 1/4-zone cell,

and 1.86 RT in 1/-2 zone cell. This indicates that in natural aquifers where

heterogeneity ubiquitously exists the MSW release tends to have longer impact

on drinking water quality. All these factors can be further investigated in the

future study.

(v) Transport through the vadose zone. The vadose zone is the unsaturated zone

112

that lies above the groundwater table, which is a major factor controlling the

water movement from the land surface to the aquifer. The vadose zone is

important in studies related to pollutant transport and interactions between

surface water and groundwater (Ravi et al., 1998). Contaminants, such as trace

metals, organics, can adsorb onto the soil particles in the vadose zone upon the

MSW release and slowly desorbed and migrated to groundwater during the

precipitation events or occurrence of another MSW release, which can

potentially pose a long adverse effect on the groundwater quality. The transport

of pollutants from MSW release in the vadose zone depends on the release

volume, the depth to the groundwater table, the infiltration rate, and

stratigraphy which can be considered in the future study. For example, in

Pennsylvania, the common depth to groundwater table ranges from 20 ft (6 m)

to 250 ft (76 m) (Fleeger, 1999). Generally, the lower groundwater table depth,

the easier contamination it can bring to the aquifers. The parameter of

infiltration rate can be estimated by techniques such as Green-Ampt Models

(Green and Ampt, 1911), which derived the physically based equation

describing the water infiltration in the soil. With lower infiltration rate, the

release will take much longer time to move through the vadose zone and reach

to the aquifer therefore affecting the final concentration to the aquifer.

(vi) Implementation of findings to other states and Utica Shale. Although the current

research focused on flowback and produced waters in Marcellus Shale formation,

the reactive transport model developed here however overcome such limitations.

For example, the model quantitatively differentiates the relative importance of

113

multiple processes, therefore helps mechanistically understand the reaction

mechanisms and predicts the transport and fate of chemicals under a wide range of

conditions. Findings from this research can inform the environmental impact on

aquifers in other states such as Colorado, New Mexico and North Dakota where the

spills of flowback and produced waters occur in close proximity to natural water

resources (Maloney et al., 2017). The geochemistry characteristics of

flowback/produced waters is almost similar between Utica shale and Marcellus

shale. This means our result can also be implemented to the flowback/produced

waters from Utica shale. Generally, Sr concentration from Utica shale is much

higher than that from Marcellus shale, which means in clay-rich aquifer under the

same release scenario, we tend to detect the Sr in groundwater from the Utica shale

compared to that from the Marcellus shale.

114

Reference

Abollino O, Giacomino A, Malandrino M, Mentasti E. Interaction of metal ions with

montmorillonite and vermiculite. Applied Clay Science 2008; 38: 227-236.

Abualfaraj N, Gurian PL, Olson MS. Characterization of Marcellus shale flowback water.

Environmental Engineering Science 2014; 31: 514-524.

Acharya HR, Henderson C, Matis H, Kommepalli H, Moore B, Wang H. Cost Effective Recovery

of Low-TDS Frac Flowback Water for Re-use. GE Global Research, 2011.

Adams EE, Gelhar LW. Field study of dispersion in a heterogeneous aquifer: 2. Spatial moments

analysis. Water Resources Research 1992; 28: 3293-3307.

Akob DM, Mumford AC, Orem W, Engle MA, Klinges JG, Kent DB, et al. Wastewater disposal

from unconventional oil and gas development degrades stream quality at a West Virginia

injection facility. Environmental science & technology 2016; 50: 5517-5525.

Al-Khulaifi Y, Lin Q, Blunt MJ, Bijeljic B. Reaction rates in chemically heterogeneous rock:

Coupled impact of structure and flow properties studied by X-ray microtomography.

Environmental Science & Technology 2017; 51: 4108-4116.

Alam MS, Cheng T. Uranium release from sediment to groundwater: Influence of water chemistry

and insights into release mechanisms. Journal of contaminant hydrology 2014; 164: 72-87.

America’s Natural Gas Alliance A. Bradford and Susquehanna County, Pennsylvania

Retrospective Case Study Characterization Report. 2013.

Amos RT, Mayer KU. Investigating ebullition in a sand column using dissolved gas analysis and

reactive transport modeling. Environmental science & technology 2006; 40: 5361-5367.

Andersson MP, Dideriksen K, Sakuma H, Stipp SLS. Modelling how incorporation of divalent

cations affects calcite wettability–implications for biomineralisation and oil recovery.

Scientific reports 2016; 6: 28854.

Apodaca LE, Bails JB, Smith CM. Water quality in shallow alluvial aquifers, upper Colorado river

basin, Colorado, 1997. JAWRA Journal of the American Water Resources Association

2002; 38: 133-149.

Appelo CAJ, Postma D. Geochemistry, Groundwater and Pollution: A. A. BALKEMA

PUBLISHERS 1993.

Appelo CAJ, Willemsen A. Geochemical calculations and observations on salt water intrusions, I.

A combined geochemical/minxing cell model. Journal of Hydrology 1987; 94: 313-330.

Atchley AL, Maxwell RM, Navarre-Sitchler AK. Human health risk assessment of CO2 leakage

into overlying aquifers using a stochastic, geochemical reactive transport approach.

Environmental science & technology 2013; 47: 5954-5962.

Ayotte JD, Gronberg JM, Apodaca LE. Trace elements and radon in groundwater across the United

States, 1992-2003: US Department of the Interior, US Geological Survey, 2011.

Ball JW, Nordstrom DK. User's manual for WATEQ4F, with revised thermodynamic data base and

test cases for calculating speciation of major, trace, and redox elements in natural waters.

1991.

Bao C, Wu H, Li L, Newcomer D, Long PE, Williams KH. Uranium bioreduction rates across

scales: Biogeochemical hot moments and hot spots during a biostimulation experiment at

rifle, Colorado. Environmental science & technology 2014; 48: 10116-10127.

Barbot E, Vidic NS, Gregory KB, Vidic RD. Spatial and temporal correlation of water quality

parameters of produced waters from devonian-age shale following hydraulic fracturing.

Environ Sci Technol 2013a; 47: 2562-9.

115

Barbot E, Vidic NS, Gregory KB, Vidic RD. Spatial and temporal correlation of water quality

parameters of produced waters from Devonian-age shale following hydraulic fracturing.

Environmental science & technology 2013b; 47: 2562-2569.

Bazilevskaya E, Rother G, Mildner DF, Pavich M, Cole D, Bhatt MP, et al. How oxidation and

dissolution in diabase and granite control porosity during weathering. Soil Science Society

of America Journal 2015; 79: 55-73.

Bearup LA, Navarre-Sitchler AK, Maxwell RM, McCray JE. Kinetic metal release from competing

processes in aquifers. Environ Sci Technol 2012; 46: 6539-47.

Bertoldi GL, Johnston RH, Evenson KD. Ground water in the Central Valley, California; a

summary report, 1991.

Bertsch PM, Seaman JC. Characterization of complex mineral assemblages: Implications for

contaminant transport and environmental remediation. Proceedings of the National

Academy of Sciences of the United States of America 1999; 96: 3350-3357.

Bethke CM. Geochemical and giogeochemical reaction modeling: Cambridge University Press,

2008.

Boyer EW, Swistock BR, Clark J, Madden M, Rizzo DE. The impact of Marcellus gas drilling on

rural drinking water supplies, 2012.

Brady PV. The Physics and Chemistry of Mineral Surfaces: CRC Press, 1996.

Brandt F, Bosbach D. Bassanite (CaSO4 center dot 0.5H(2)O) dissolution and gypsum (CaSO4

center dot 2H(2)O) precipitation in the presence of cellulose ethers. Journal of Crystal

Growth 2001; 233: 837-845.

Brandt F, Bosbach D, Krawczyk-Barsch E, Arnold T, Bernhard G. Chlorite dissolution in the acid

pH-range: A combined microscopic and macroscopic approach. Geochimica Et

Cosmochimica Acta 2003; 67: 1451-1461.

Brantley S, Vidic R, Brasier K, Yoxtheimer D, Pollak J, Wilderman C, et al. Engaging over data

on fracking and water quality. Science 2018; 359: 395-397.

Brantley SL, Kubicki JD, White AF. Kinetics of water-rock interaction. Vol 15: Springer, 2008.

Brantley SL, Yoxtheimer D, Arjmand S, Grieve P, Vidic R, Pollak J, et al. Water resource impacts

during unconventional shale gas development: The Pennsylvania experience. International

Journal of Coal Geology 2014a; 126: 17.

Brantley SL, Yoxtheimer D, Arjmand S, Grieve P, Vidic R, Pollak J, et al. Water resource impacts

during unconventional shale gas development: The Pennsylvania experience. International

Journal of Coal Geology 2014b; 126: 140-156.

Brenniman G, Kojola W, Levy P, Carnow B, Namekata T. High barium levels in public drinking

water and its association with elevated blood pressure. Archives of Environmental Health:

An International Journal 1981; 36: 28-32.

Brumsack H-J. The trace metal content of recent organic carbon-rich sediments: implications for

Cretaceous black shale formation. Palaeogeography, Palaeoclimatology, Palaeoecology

2006; 232: 344-361.

Bucca M, Dietzel M, Tang J, Leis A, Köhler SJ. Nucleation and crystallization of otavite, witherite,

calcite, strontianite, hydrozincite, and hydrocerussite by CO 2 membrane diffusion

technique. Chemical geology 2009; 266: 143-156.

Cai Z, Li L. How long do natural waters “remember” release incidents of Marcellus Shale waters:

a first order approximation using reactive transport modeling. Geochem. Trans. 2016; 17:

82-97.

Cai Z, Wen H, Komarneni S, Li L. Mineralogy control on reactive transport of Marcellus Shale

waters. Sci. Total. Environ. 2018a; 630: 1573–1582.

Cai Z, Wen H, Komarneni S, Li L. Time series of effluent chemistry data on mineralogy controls

on the reactive transport of Marcellus Shale waters. Data In Brief 2018b.

116

Canada H. Guidelines for Canadian Drinking Water Quality - Summary Talbe. In: Branch HEaCS,

editor, Ottawa, Ontario., 2014.

Carozzi AV. Sedimentary petrography: PTR Prentice Hall, 1993.

Carrollwebb SA, Walther JV. A Surface Complex-Reaction Model for the Ph-Dependence of

Corundum and Kaolinite Dissolution Rates. Geochimica Et Cosmochimica Acta 1988; 52:

2609-2623.

Chakraborty S, Wolthers M, Chatterjee D, Charlet L. Adsorption of arsenite and arsenate onto

muscovite and biotite mica. J Colloid Interface Sci 2007a; 309: 392-401.

Chakraborty S, Wolthers M, Chatterjee D, Charlet L. Adsorption of arsenite and arsenate onto

muscovite and biotite mica. Journal of colloid and interface science 2007b; 309: 392-401.

Chapman EC, Capo RC, Stewart BW, Kirby CS, Hammack RW, Schroeder KT, et al. Geochemical

and Strontium Isotope Characterization of Produced Waters from Marcellus Shale Natural

Gas Extraction. Environmental Science & Technology 2012; 46: 3545-3553.

Chen SS, Sun Y, Tsang DC, Graham NJ, Ok YS, Feng Y, et al. Insights into the subsurface transport

of As (V) and Se (VI) in produced water from hydraulic fracturing using soil samples from

Qingshankou Formation, Songliao Basin, China. Environmental Pollution 2017; 223: 449-

456.

Cortis A, Chen Y, Scher H, Berkowitz B. Quantitative characterization of pore-scale disorder

effects on transport in “homogeneous” granular media. Physical Review E 2004; 70:

041108.

Cozzarelli IM, Skalak K, Kent D, Engle MA, Benthem A, Mumford A, et al. Environmental

signatures and effects of an oil and gas wastewater spill in the Williston Basin, North

Dakota. Science of the Total Environment 2017; 579: 1781-1793.

Dagan G. Solute transport in heterogeneous porous formations. Journal of fluid mechanics 1984;

145: 151-177.

Dagan G. Transport in heterogeneous porous formations: spatial moments, ergodicity, and effective

dispersion. Water Resources Research 1990; 26: 1281-1290.

Dagan G, Fiori A, Jankovic I. Upscaling of flow in heterogeneous porous formations: Critical

examination and issues of principle. Advances in water resources 2013; 51: 67-85.

Davies C. Ion Association. Butterworths, London, , 1962.

Davis JA, Fuller CC, Cook AD. A model for trace metal sorption processes at the calcite surface:

Adsorption of Cd2+ and subsequent solid solution formation. Geochimica et

Cosmochimica Acta 1987; 51: 1477-1490.

Deng H, Dai Z, Wolfsberg AV, Ye M, Stauffer PH, Lu Z, et al. Upscaling retardation factor in

hierarchical porous media with multimodal reactive mineral facies. Chemosphere 2013; 91:

248-257.

Denny CS, Lyford WH, Goodlett JC. Surficial geology and soils of the Elmira-Williamsport region,

New York and Pennsylvania, with a section on forest regions and great soil groups. USGS,

1963.

Dentz M, Cortis A, Scher H, Berkowitz B. Time behavior of solute transport in heterogeneous

media: transition from anomalous to normal transport. Advances in Water Resources 2004;

27: 155-173.

DeSimone LA, Hamilton PA, Gilliom RJ. Quality of water from domestic wells in principal

aquifers of the United States, 1991-2004: overview of major findings: US Department of

the Interior, US Geological Survey, 2009.

Deutsch WJ, Siegel R. Groundwater geochemistry: fundamentals and applications to contamination:

CRC press, 1997.

117

DiGiulio DC, Jackson RB. Impact to underground sources of drinking water and domestic wells

from production well stimulation and completion practices in the Pavillion, Wyoming, field.


Digiulio DC, Wilkin RT, Miller C, Oberley G. Investigation of Ground Water Contamination Near

Pavillion, Wyoming. United States Environmental Protection Agency, Office of Research

and Development, National Risk Management Research Laboratory, Ada, Oklahoma

74820, 2011.

Domenico PA, Schwartz FW. Physical and chemical hydrogeology. Vol 44: Wiley New York,

1998.

dos Anjos VE, Rohwedder JR, Cadore S, Abate G, Grassi MT. Montmorillonite and vermiculite as

solid phases for the preconcentration of trace elements in natural waters: Adsorption and

desorption studies of As, Ba, Cu, Cd, Co, Cr, Mn, Ni, Pb, Sr, V, and Zn. Applied Clay

Science 2014; 99: 289-296.

Edition F. Guidelines for Drinking-water Quality. WHO chronicle 2011; 38: 104-8.

Elango L, Kannan R. Rock–water interaction and its control on chemical composition of

groundwater. Developments in environmental science 2007; 5: 229-243.

Espinoza C, Valocchi AJ. Stochastic analysis of one-dimensional transport of kinetically adsorbing

solutes in chemically heterogeneous aquifers. Water Resources Research 1997a; 33: 2429-

2445.

Espinoza C, Valocchi AJ. Stochastic analysis of one‐ dimensional transport of kinetically

adsorbing solutes in chemically heterogeneous aquifers. Water resources research 1997b;

33: 2429-2445.

Ferrar KJ, Michanowicz DR, Christen CL, Mulcahy N, Malone SL, Sharma RK. Assessment of

Effluent Contaminants from Three Facilities Discharging Marcellus Shale Wastewater to

Surface Waters in Pennsylvania. Environmental Science & Technology 2013; 47: 3472-

3481.

Fiori A, Boso F, de Barros FP, De Bartolo S, Frampton A, Severino G, et al. An indirect assessment

on the impact of connectivity of conductivity classes upon longitudinal asymptotic

macrodispersivity. Water resources research 2010; 46.

Fisher RS, Stueber AM. Strontium Isotopes in Selected Streams within Susquehanna River Basin.

Water Resources Research 1976; 12: 1061-1068.

Fox PM, Davis JA, Hay MB, Conrad ME, Campbell KM, Williams KH, et al. Rate‐limited U (VI)

desorption during a small‐scale tracer test in a heterogeneous uranium‐contaminated

aquifer. Water Resources Research 2012; 48.

Freeze RA. A stochastic ‐ conceptual analysis of one ‐ dimensional groundwater flow in

nonuniform homogeneous media. Water Resources Research 1975; 11: 725-741.

Freeze RA, Witherspoon P. Theoretical analysis of regional groundwater flow: 2. Effect of water‐table configuration and subsurface permeability variation. Water Resources Research 1967;

3: 623-634.

Frye E, Bao C, Li L, Blumsack S. Environmental Controls of Cadmium Desorption during CO2

Leakage. Environmental Science & Technology 2012; 46: 4388-4395.

Fulton J. Report Of Progress In Bradford And Tioga Counties (1878): Harrisburg, 1878.

Gelhar LW, Axness CL. Three‐dimensional stochastic analysis of macrodispersion in aquifers.


Gelhar LW, Welty C, Rehfeldt KR. A critical review of data on field‐scale dispersion in aquifers.

Water resources research 1992; 28: 1955-1974.

Glass HD, Potter PE, Siever R. Clay Mineralogy of Some Basal Pennsylvanian Sandstones, Clays,

and Shales: GEOLOGICAL NOTES. AAPG Bulletin 1956; 40: 750-754.

118

Glynn PD, Reardon EJ. Solid-solution aqueous-solution equilibria; thermodynamic theory and

representation. American Journal of Science 1990; 290: 164-201.

Gounot A-M. Microbial oxidation and reduction of manganese: consequences in groundwater and

applications. FEMS Microbiology Reviews 1994; 14: 339-349.

Gradient. National Human Health Risk Evaluation for Hydraulic Fracturing Fluid Additives.

Halliburton Energy Services, Inc., Houston, TX, 2013.

Haag WR, Hoigne J. Ozonation of bromide-containing waters: kinetics of formation of

hypobromous acid and bromate. Environmental Science & Technology 1983; 17: 261-267.

Haggerty R, Gorelick SM. Multiple‐ rate mass transfer for modeling diffusion and surface

reactions in media with pore‐scale heterogeneity. Water Resources Research 1995; 31:

2383-2400.

Hagström E, Jackanich K. Hydraulic Fracturing Litigation Is on the Rise. Hydraulic Fracturing

Digest 2011.

Haluszczak LO, Rose AW, Kump LR. Geochemical evaluation of flowback brine from Marcellus

gas wells in Pennsylvania, USA. Applied Geochemistry 2013; 28: 55-61.

Hayes T. Sampling and Analysis of Water Streams Associated with the Development of Marcellus

Shale Gas (Final). Report to Marcellus Shale Coalition. [Gas Technology Institute], 2009,

pp. 249.

Heath R. Basic ground-water hydrology: US Geological Survey Water-Supply Paper 2220.

Washington, DC: US Geological Survey, 1983.

Heidari P, Li L. Solute transport in low‐heterogeneity sandboxes: The role of correlation length

and permeability variance. Water Resources Research 2014; 50: 8240-8264.

Hennessy A, Graham G. The effect of additives on the co-crystallisation of calcium with barium

sulphate. Journal of crystal growth 2002; 237: 2153-2159.

Howarth RW, Ingraffea A, Engelder T. Natural gas: Should fracking stop? Nature 2011; 477: 271-

275.

Inglezakis V, Zorpas A, Loizidou M, Grigoropoulou H. The effect of competitive cations and

anions on ion exchange of heavy metals. Separation and Purification Technology 2005; 46:

202-207.

Jackson R, Inch K. The in-situ adsorption of 90Sr in a sand aquifer at the Chalk River Nuclear

Laboratories. Journal of Contaminant Hydrology 1989; 4: 27-50.

Jacobs P, Waite T. The role of aqueous iron (II) and manganese (II) in sub-aqueous active barrier

systems containing natural clinoptilolite. Chemosphere 2004; 54: 313-324.

Jin LX, Brantley SL. Soil chemistry and shale weathering on a hillslope influenced by convergent

hydrologic flow regime at the Susquehanna/Shale Hills Critical Zone Observatory. Applied

Geochemistry 2011; 26: S51-S56.

Johnson G, Gupta K, Putz D, Hu Q, Brusseau M. The effect of local-scale physical heterogeneity

and nonlinear, rate-limited sorption/desorption on contaminant transport in porous media.

Journal of contaminant hydrology 2003; 64: 35-58.

Jones F, Oliviera A, Parkinson G, Rohl A, Stanley A, Upson T. The effect of calcium ions on the

precipitation of barium sulphate 1: calcium ions in the absence of organic additives. Journal

of crystal growth 2004; 262: 572-580.

Jones F, Piana S, Gale J. Understanding the kinetics of barium sulfate precipitation from water and

water–methanol solutions. Crystal Growth and Design 2008; 8: 817-822.

Judd RM, Levy BI. Effects of barium-induced cardiac contraction on large-and small-vessel

intramyocardial blood volume. Circulation research 1991; 68: 217-225.

Killey R, Moltyaner G. Twin Lake tracer tests: Setting, methodology, and hydraulic conductivity

distribution. Water Resources Research 1988; 24: 1585-1612.

119

Kiviat E. Risks to biodiversity from hydraulic fracturing for natural gas in the Marcellus and Utica

shales. Annals of the New York Academy of Sciences 2013; 1286: 1-14.

Koltermann CE, Gorelick SM. Heterogeneity in sedimentary deposits: A review of structure‐

imitating, process‐ imitating, and descriptive approaches. Water Resources Research

1996; 32: 2617-2658.

Kondash AJ, Warner NR, Lahav O, Vengosh A. Radium and barium removal through blending

hydraulic fracturing fluids with acid mine drainage. Environmental science & technology

2013; 48: 1334-1342.

Kump LR, Brantley SL, Arthur MA. Chemical weathering, atmospheric CO2, and climate. Annual

Review of Earth and Planetary Sciences 2000; 28: 611-667.

Kyshakevych RG, Prellwitz HS. MONONGAHELA AND YOUGHIOGHENY RIVERS, POOLS

2 AND 3 RIVERBANK GEOLOGY, CONDITIONS, AND ACCESS REPORTS. In:

Collins T, editor. Carnegie Mellon University, Pittsburgh, 2001, pp. 55.

Lani M, Schoonen MA. The Influence of Cation Exchange on the Chemistry of Long Island

Groundwater. State University of New York at Stony Brook, 2010.

Lasaga A. Kinetic Theory in the Earth SciencesPrinceton Univ. Press, Princeton 1998.

Latour L, Kleinberg RL, Mitra PP, Sotak CH. Pore-size distributions and tortuosity in

heterogeneous porous media. Journal of Magnetic Resonance, Series A 1995; 112: 83-91.

Le Borgne T, Dentz M, Carrera J. Lagrangian statistical model for transport in highly

heterogeneous velocity fields. Physical review letters 2008; 101: 090601.

LeBlanc DR, Garabedian SP, Hess KM, Gelhar LW, Quadri RD, Stollenwerk KG, et al. Large‐scale natural gradient tracer test in sand and gravel, Cape Cod, Massachusetts: 1.

Experimental design and observed tracer movement. Water Resources Research 1991; 27:

895-910.

Li L, Maher K, Navarre-Sitchler A, Druhan J, Meile C, Lawrence C, et al. Expanding the role of

reactive transport models in critical zone processes. Earth Science Reviews 2016.

Li L, Maher K, Navarre-Sitchler A, Druhan J, Meile C, Lawrence C, et al. Expanding the role of

reactive transport models in critical zone processes. Earth-Science Reviews 2017; 165:

280-301.

Li L, Peters C, Celia M. Effects of mineral spatial distribution on reaction rates in porous media.

Water resources research 2007; 43.

Li L, Peters CA, Celia MA. Upscaling geochemical reaction rates using pore-scale network

modeling. Advances in Water Resources 2006; 29: 1351-1370.

Li L, Salehikhoo F, Brantley SL, Heidari P. Spatial zonation limits magnesite dissolution in porous

media. Geochimica Et Cosmochimica Acta 2014a; 126: 555-573.

Li L, Salehikhoo F, Brantley SL, Heidari P. Spatial zonation limits magnesite dissolution in porous

media. Geochimica Et Cosmochimica Acta 2014b; 126: 555-573.

Li L, Steefel CI, Kowalsky MB, Englert A, Hubbard SS. Effects of physical and geochemical

heterogeneities on mineral transformation and biomass accumulation during biostimulation

experiments at Rifle, Colorado. Journal of Contaminant Hydrology 2010; 112: 45-63.

Li L, Steefel CI, Williams KH, Wilkins MJ, Hubbard SS. Mineral Transformation and Biomass

Accumulation Associated With Uranium Bioremediation at Rifle, Colorado.


Liberati BP. Removal of Naturally Occurring Radioactive Material From Flowback/Produced

Water From the Hydraulic Fracturing Process. Ohio University, 2015.

Lichtner PC. Continuum Model for Simultaneous Chemical-Reactions and Mass- Transport in

Hydrothermal Systems. Geochimica et Cosmochimica Acta 1985; 49: 779-800.

Lindsey BD, Berndt MP, Katz BG, Ardis AF, Skach KA. Factors affecting water quality in selected

carbonate aquifers in the United States, 1993-2005, 2009.

120

Liu C, Shang J, Shan H, Zachara JM. Effect of subgrid heterogeneity on scaling geochemical and

biogeochemical reactions: a case of U (VI) desorption. Environmental science &

technology 2014; 48: 1745-1752.

Liu H, Korshin GV, Ferguson JF. Investigation of the kinetics and mechanisms of the oxidation of

cerussite and hydrocerussite by chlorine. Environmental science & technology 2008; 42:

3241-3247.

Llewellyn GT, Dorman F, Westland J, Yoxtheimer D, Grieve P, Sowers T, et al. Evaluating a

groundwater supply contamination incident attributed to Marcellus Shale gas development.

Proceedings of the National Academy of Sciences 2015; 112: 6325-6330.

Luo J, Cirpka OA. Traveltime-based descriptions of transport and mixing in heterogeneous

domains. Water Resources Research 2008; 44.

Malandrino M, Abollino O, Giacomino A, Aceto M, Mentasti E. Adsorption of heavy metals on

vermiculite: influence of pH and organic ligands. Journal of Colloid and Interface Science

2006; 299: 537-546.

Maloney KO, Baruch-Mordo S, Patterson LA, Nicot J-P, Entrekin SA, Fargione JE, et al.

Unconventional oil and gas spills: Materials, volumes, and risks to surface waters in four

states of the US. Science of the Total Environment 2017; 581: 369-377.

Mastrocicco M, Prommer H, Pasti L, Palpacelli S, Colombani N. Evaluation of saline tracer

performance during electrical conductivity groundwater monitoring. Journal of

contaminant hydrology 2011; 123: 157-166.

Maupin MA, Barber NL. Estimated withdrawals from principal aquifers in the United States, 2000.

2005.

Merkel BJ, Planer-Friedrich B. Groundwater Geochemistry: Springer-Verlag Berlin Heidelberg,

2008.

Meunier A. Clays. New York, Berlin: Springer, 2005.

Milici RC, Swezey CS. Assessment of Appalachian Basin Oil and Gas Resources: Devonian Shale–

Middle and Upper Paleozoic Total Petroleum System U.S. Department of the Interior, 2006,

pp. 70.

Minyard ML, Burgos WD. Hydrologic flow controls on biologic iron (III) reduction in natural

sediments. Environmental science & technology 2007; 41: 1218-1224.

Mohamed M, Hatfield K, Hassan A, Klammler H. Stochastic evaluation of subsurface contaminant

discharges under physical, chemical, and biological heterogeneities. Advances in Water

Resources 2010; 33: 801-812.

Molins S, Trebotich D, Steefel CI, Shen C. An investigation of the effect of pore scale flow on

average geochemical reaction rates using direct numerical simulation. Water Resources

Research 2012; 48.

Myers T. Potential Contaminant Pathways from Hydraulically Fractured Shale to Aquifers. Ground

Water 2012.

N.Setoudeh, Zamani MAA, N.J.Welham. Carbothermic Reduction of Mechanically Activated

Mixtures of Celestite and Carbon. World Academy of Science, Engineering and

Technology 2011: 526-529.

Navarre-Sitchler A, Brantley S. Basalt weathering across scales. Earth and Planetary Science

Letters 2007; 261: 321-334.

Neuman SP, Tartakovsky DM. Perspective on theories of non-Fickian transport in heterogeneous

media. Advances in Water Resources 2009; 32: 670-680.

Newell CJ, Hopkins LP, Bedient PB. A Hydrogeologic Database for Ground‐Water Modeling.

Ground Water 1990a; 28: 703-714.

Newell CJ, Hopkins LP, Bedient PB. A Hydrogeologic Database for Ground‐Water Modeling.

Groundwater 1990b; 28: 703-714.

121

NJDEP. Development of a dilution-attenuation factor for the impact to ground wter pathway. In:

Protection NJDoE, editor, Trenton, NJ. , 2013.

Olmstead SM, Muehlenbachs LA, Shih JS, Chu ZY, Krupnick AJ. Shale gas development impacts

on surface water quality in Pennsylvania. Proceedings of the National Academy of

Sciences of the United States of America 2013; 110: 4962-4967.

Oram B. Pennsylvania Groundwater Quality: Your Private Well: What Do the Results Mean?: B.

F. Environmental Consultants Inc., 2012.

Osborn SG, Vengosh A, Warner NR, Jackson RB. Methane contamination of drinking water

accompanying gas-well drilling and hydraulic fracturing. Proceedings of the National

Academy of Sciences of the United States of America 2011a; 108: 8172-8176.


accompanying gas-well drilling and hydraulic fracturing. proceedings of the National

Academy of Sciences 2011b; 108: 8172-8176.


accompanying gas-well drilling and hydraulic fracturing. Proc Natl Acad Sci U S A 2011c;

108: 8172-6.

PADEP. DEP OFFICE OF OIL AND GAS MANAGEMENT COMPLIANCE REPORT,

http://www.depreportingservices.state.pa.us/ReportServer/Pages/ReportViewer.aspx?/Oil

_Gas/OG_Compliance, 2005-2015.

Palandri JL, Kharaka YK. A Compliation of rate parameters of water-mineral interaction kinetics

for application to geochemical modeling. U.S. Department of the Interior, 2004, pp. 71.

Patterson LA, Konschnik KE, Wiseman H, Fargione J, Maloney KO, Kiesecker J, et al.

Unconventional oil and gas spills: Risks, mitigation priorities, and state reporting

requirements. Environmental Science & Technology 2017; 51: 2563-2573.

Perng YS, Wang EIC, Lu CC, Kuo LS. Performance of sericite as a pigment for coating paper.

Appita Journal 2006; 59: 378-+.

Perujo N, Sanchez-Vila X, Proia L, Romaní A. Interaction between physical heterogeneity and

microbial processes in subsurface sediments: a laboratory-scale column experiment.


Peters CA. Accessibilities of reactive minerals in consolidated sedimentary rock: An imaging study

of three sandstones. Chemical Geology 2009; 265: 198-208.

Phillips DH, Gu B, Watson DB, Roh Y, Liang L, Lee S. Performance evaluation of a zerovalent

iron reactive barrier: mineralogical characteristics. Environmental Science & Technology

2000; 34: 4169-4176.

Pirc S. Uranium and other elements in the Catskill Formation of east-central Pennsylvania.

Pennsylvania State University, 1979.

Pokrovsky O, Schott J. Surface chemistry and dissolution kinetics of divalent metal carbonates.


Potgieter J, Potgieter-Vermaak S, Kalibantonga P. Heavy metals removal from solution by

palygorskite clay. Minerals engineering 2006; 19: 463-470.

Quinlan JF, Ewers RO. Subsurface drainage in the Mammoth Cave area. Karst Hydrology. Springer,

1989, pp. 65-103.

Ramasomanana F, Younes A, Ackerer P. Estimation of macrodispersion in 2 ‐ D highly

heterogeneous porous media using the Eulerian‐Lagrangian localized adjoint method.


Reed LA, Stuckey MH. Prediction of velocities for a range of streamflow conditions in

Pennsylvania: US Department of the Interior, US Geological Survey, 2002.

http://www.depreportingservices.state.pa.us/ReportServer/Pages/ReportViewer.aspx?/Oil_Gas/OG_Compliance

http://www.depreportingservices.state.pa.us/ReportServer/Pages/ReportViewer.aspx?/Oil_Gas/OG_Compliance

122

Rehfeldt KR, Boggs JM, Gelhar LW. Field study of dispersion in a heterogeneous aquifer: 3.

Geostatistical analysis of hydraulic conductivity. Water Resources Research 1992; 28:

3309-3324.

Reinson G, Foscolos A. Trends in sandstone diagenesis with depth of burial, Viking Formation,

southern Alberta. Bulletin of Canadian Petroleum Geology 1986; 34: 126-152.

Rimstidt JD, Balog A, Webb J. Distribution of trace elements between carbonate minerals and

aqueous solutions. Geochimica et Cosmochimica Acta 1998; 62: 1851-1863.

Rogers RJ. Geochemical Comparison of Ground Water in Areas of New England, New York, and

Pennsylvania. Ground Water 1989; 27: 690-712.

Rolle M, Eberhardt C, Chiogna G, Cirpka OA, Grathwohl P. Enhancement of dilution and

transverse reactive mixing in porous media: Experiments and model-based interpretation.

Journal of contaminant hydrology 2009; 110: 130-142.

Rozell DJ, Reaven SJ. Water pollution risk associated with natural gas extraction from the

Marcellus Shale. Risk Analysis 2012a; 32: 1382-1393.

Rozell DJ, Reaven SJ. Water pollution risk associated with natural gas extraction from the

Marcellus Shale. Risk Anal 2012b; 32: 1382-93.

Russell R, Rinehart D, Smith H, Peterson R. Development and Characterization of Gibbsite

Component Simulant. Pacific Northwest National Laboratory Richland, Washington, 2009.

Salehikhoo F, Li L. The role of magnesite spatial distribution patterns in determining dissolution

rates: When do they matter? Geochimica et Cosmochimica Acta 2015; 155: 107-121.

Salehikhoo F, Li L, Brantley SL. Magnesite dissolution rates at different spatial scales: The role of

mineral spatial distribution and flow velocity. Geochimica et Cosmochimica Acta 2013a;

108: 91-106.

Salehikhoo F, Li L, Brantley SL. Magnesite dissolution rates at different spatial scales: The role of

mineral spatial distribution and flow velocity. Geochimica Et Cosmochimica Acta 2013b;

108: 91-106.

Sang w, Stoof CR, Zhang W, Morales VL, Gao B, Kay RW, et al. Effect of Hydrofracking Fluid

on Colloid Transport in the Unsaturated Zone. Environmental Science & Technology 2014.

Schulthess C, Huang C. Adsorption of heavy metals by silicon and aluminum oxide surfaces on

clay minerals. Soil Science Society of America Journal 1990; 54: 679-688.

Schulze K, Hunger M, Döll P. Simulating river flow velocity on global scale. Advances in

Geosciences 2005; 5: 133-136.

Seeboonruang U, Ginn TR. Upscaling heterogeneity in aquifer reactivity via exposure-time concept:

Forward model. Journal of contaminant hydrology 2006; 84: 127-154.

Shen L, Chen Z. Critical review of the impact of tortuosity on diffusion. Chemical Engineering

Science 2007; 62: 3748-3755.

Sherman LA, Barak P. Solubility and Dissolution Kinetics of Dolomite in Ca–Mg–HCO3/CO3

Solutions at 25°C and 0.1 MPa Carbon Dioxide. Soil Science Society of America Journal

2000; 64: 1959-1968.

Shih J-S, Saiers JE, Anisfeld SC, Chu Z, Muehlenbachs LA, Olmstead SM. Characterization and

analysis of liquid waste from Marcellus Shale gas development. Environmental science &

technology 2015; 49: 9557-9565.

Skalak KJ, Engle MA, Rowan EL, Jolly GD, Conko KM, Benthem AJ, et al. Surface disposal of

produced waters in western and southwestern Pennsylvania: Potential for accumulation of

alkali-earth elements in sediments. International Journal of Coal Geology 2014; 126: 162-

170.

Soler JM, Boi M, Mogollón JL, Cama J, Ayora C, Nico PS, et al. The passivation of calcite by acid

mine water. Column experiments with ferric sulfate and ferric chloride solutions at pH 2.

Applied Geochemistry 2008; 23: 3579-3588.

123

Spradlin JD. Trace Metal Characterization and Ion Exchange Capacity of Devonian to

Pennsylvanian Age Bedrock in New York and Pennsylvania in Relation to Drinking Water

Quality. Syracuse University, 2015.

Srivastava P, Singh B, Angove M. Competitive adsorption behavior of heavy metals on kaolinite.

Journal of Colloid and Interface Science 2005; 290: 28-38.

Steefel CI. CrunchFlow. Software for Modeling Multicomponent Reactive Flow and Transport.

User’s Manual. Lawrence Berkeley National Laboratory, Berkeley, USA 2009.

Steefel CI, Appelo CAJ, Arora B, Jacques D, Kalbacher T, Kolditz O, et al. Reactive transport

codes for subsurface environmental simulation. Computational Geosciences 2015; 19: 445-

478.

Steefel CI, Lasaga AC. A coupled model for transport of multiple chemical species and kinetic

precipitation/dissolution reactions with application to reactive flow in single phase

hydrothermal systems. American Journal of science 1994; 294: 529-592.

Sudicky EA. A natural gradient experiment on solute transport in a sand aquifer: Spatial variability

of hydraulic conductivity and its role in the dispersion process. Water Resources Research

1986; 22: 2069-2082.

Sudicky EA, Illman WA, Goltz IK, Adams JJ, McLaren RG. Heterogeneity in hydraulic

conductivity and its role on the macroscale transport of a solute plume: From measurements

to a practical application of stochastic flow and transport theory. Water Resources Research

2010; 46.

Swistock B. A Quick Guide to Groundwater in Pennsylvania. College of Agriculture Science, The

Pennsylvania State University, 2007, pp. 11.

Tesoriero AJ. Solid solution/aqueous solution partitioning of divalent cations of calcite:

Implications for the mobility of metals in natural waters. Doctor of Philosophy. Oregon

Health & Science University, 1994.

Tesoriero AJ, Pankow JF. Solid solution partitioning of Sr 2+, Ba 2+, and Cd 2+ to calcite.

Geochimica et Cosmochimica Acta 1996; 60: 1053-1063.

Tester JW, Worley WG, Robinson BA, Grigsby CO, Feerer JL. Correlating quartz dissolution

kinetics in pure water from 25 to 625 C. Geochimica et Cosmochimica Acta 1994; 58:

2407-2420.

Ticknor KV, Saluja PP. Determination of surface areas of mineral powders by adsorption

calorimetry. Clays and Clay Minerals 1990; 38: 437-441.

Tompson AF. Numerical simulation of chemical migration in physically and chemically

heterogeneous porous media. Water Resources Research 1993; 29: 3709-3726.

Tomson MB, Fu G, Watson MA, Kan AT. Mechanisms of Mineral Scale Inhibition. Society of

Petroleum Engineers 2003; 18.

Trapp H, Horn MA. Ground Water Atlas of the United States: Delaware, Maryland, New Jersey,

North Carolina, Pennsylvania, Virginia, West Virginia. Segment 11: The Survey, 1997.

Trefry JH, Trocine RP. Chemical Forms and Reactions of Barium in Mixtures of Produced Water

with Seawater. Produced Water. Springer, 2011, pp. 127-145.

USEPA. Drinking Water Contaminants - Standards and Regulations. In: (EPA) USEPA, editor,

https://www.epa.gov/dwstandardsregulations.

USEPA. Soil screening guidance: user's guide. In: Agency USEP, editor. United States

Environmental Protection Agency, Office of solid waste and emergency response,

Washington, DC, 1996, pp. 89.

Valocchi AJ, Street RL, Roberts PV. Transport of ion‐exchanging solutes in groundwater:

Chromatographic theory and field simulation. Water Resources Research 1981; 17: 1517-

1527.

https://www.epa.gov/dwstandardsregulations

124

Vanselow AP. The utilization of the base exchange reaction for the determination of activity

coefficients in mixed electrolytes. Journal of the American Chemical Society 1932; 54: 5.

Vengosh A, Jackson RB, Warner N, Darrah TH, Kondash A. A Critical Review of the Risks to

Water Resources from Unconventional Shale Gas Development and Hydraulic Fracturing

in the United States. Environ Sci Technol 2014.

Vidic RD. Sustainable Management of Flowback Water during Hydraulic Fracturing of Marcellus

Shale for Natural Gas Production. University of Pittsburgh, Pittsburgh, PA 2015.

Vidic RD, Brantley SL, Vandenbossche JM, Yoxtheimer D, Abad JD. Impact of Shale Gas

Development on Regional Water Quality. Science 2013a; 340.

Vidic RD, Brantley SL, Vandenbossche JM, Yoxtheimer D, Abad JD. Impact of shale gas

development on regional water quality. Science 2013b; 340: 1235009.

Viking A. GONTROLS ON CLAY. MINERAL AUTHIGENESIS IN THE VIKING

SANDSTONE, CENTRAL ALBEBTA. I. SHALLOW DEPTHS. Canadian Mineralogist

1982; 20: 141-150.

Wang L, Li L. Illite spatial distribution patterns dictate Cr (VI) sorption macrocapacity and

macrokinetics. Environmental science & technology 2015a; 49: 1374-1383.

Warner NR, Christie CA, Jackson RB, Vengosh A. Impacts of shale gas wastewater disposal on

water quality in western Pennsylvania. Environ Sci Technol 2013a; 47: 11849-57.

Warner NR, Jackson RB, Darrah TH, Osborn SG, Down A, Zhao K, et al. Geochemical evidence

for possible natural migration of Marcellus Formation brine to shallow aquifers in

Pennsylvania. Proceedings of the National Academy of Sciences 2012a; 109: 11961-11966.

Warner NR, Jackson RB, Darrah TH, Osborn SG, Down A, Zhao KG, et al. Geochemical evidence

for possible natural migration of Marcellus Formation brine to shallow aquifers in

Pennsylvania. Proceedings of the National Academy of Sciences of the United States of

America 2012b; 109: 11961-11966.

Warner NR, Kresse TM, Hays PD, Down A, Karr JD, Jackson RB, et al. Geochemical and isotopic

variations in shallow groundwater in areas of the Fayetteville Shale development, north-

central Arkansas. Applied Geochemistry 2013b; 35: 207-220.

Watkins DM, Cornuet TS. Evaluation of Geology and Water Well Data Associated with the EPA

Hydraulic Fracturing Retrospective Case Study Bradford County, Pennsylvania. WESTON,

Oklahoma city, 2012, pp. 360.

Wen H, Li L. An upscaled rate law for magnesite dissolution in heterogeneous porous media.

Geochimica et Cosmochimica Acta 2017; 210: 289-305.

Wen H, Li L, Crandall D, Hakala A. Where lower calcite abundance creates more alteration:

enhanced rock matrix diffusivity induced by preferential dissolution. Energy & Fuels

2016a; 30: 4197-4208.

Wen H, Li L, Crandall D, Hakala JA. Where Lower Calcite Abundance Creates More Alteration:

Enhanced Rock Matrix Diffusivity Induced by Preferential Carbonate Dissolution. Energy

& Fuels 2016b.

White AF, Brantley SL. The effect of time on the weathering of silicate minerals: why do

weathering rates differ in the laboratory and field? Chemical Geology 2003; 202: 479-506.

Williams JH, Taylor LE, Low DJ. Hydrogeology and Groundwater Quality of the Glaciated Valleys

of Bradford, Tioga, and Potter Counties, Pennsylvania. Water Resource Report. 68. United

States Geological Survey, Pennsylvania Geologican Survey, 1998.

Wolery TJ, Daveler SA. EQ6, a computer program for reaction path modeling of aqueous

geochemical systems: theoretical manual, user's guide and related documentation (Version

7.0): Lawrence Livermore Laboratory, University of California Livermore, CA, 1992.

125

Wolery TJ, Jackson KJ, Bourcier WL, Bruton CJ, Viani BE, Knauss KG, et al. Current status of

the EQ3/6 software package for geochemical modeling. Acs Symposium Series 1990a; 416:

13.

Wolery TJ, Jackson KJ, Bourcier WL, Bruton CJ, Viani BE, Knauss KG, et al. Current status of

the EQ3/6 software package for geochemical modeling. ACS symposium series. 416.

Oxford University Press, 1990b, pp. 104-116.

Wollast R, Chou L. Rate Control of Weathering of Silicate Minerals at Room Temperature and

Pressure. Physical and Chemical Weathering in Geochemical Cycles. 251. Springer

Netherlands, 1988, pp. 11-32.

Xu TF, Apps JA, Pruess K. Numerical simulation of CO2 disposal by mineral trapping in deep

aquifers. Applied Geochemistry 2004; 19: 917-936.

Zabochnicka-Świątek M, L. S, Stanczyk-Mazanek E. Hazard evaluation of the inorganic

contaminants (B, Ba, Sr, Cu, Zn) expansion under various pH values in the Quaternary

Aquifer. Polish Journal of Environmental Studies 2010; 2: 255-260.

Zhang C, Kang Q, Wang X, Zilles JL, Muller RH, Werth CJ. Effects of pore-scale heterogeneity

and transverse mixing on bacterial growth in porous media. Environmental science &

technology 2010; 44: 3085-3092.

Zhang L, Anderson N, Dilmore R, Soeder DJ, Bromhal G. Leakage detection of Marcellus shale

natural gas at an upper Devonian gas monitoring well: A 3-D numerical modeling approach.


Zhang P-C, Brady PV, Arthur SE, Zhou W-Q, Sawyer D, Hesterberg DA. Adsorption of barium

(II) on montmorillonite: an EXAFS study. Colloids and Surfaces A: Physicochemical and

Engineering Aspects 2001; 190: 239-249.

Zheng C, Gorelick SM. Analysis of solute transport in flow fields influenced by preferential

flowpaths at the decimeter scale. Groundwater 2003; 41: 142-155.

Zheng LG, Apps JA, Zhang YQ, Xu TF, Birkholzer JT. On mobilization of lead and arsenic in

groundwater in response to CO2 leakage from deep geological storage. Chemical Geology

2009; 268: 281-297.

Zhou Q, Birkholzer J, Rutqvist J, Tsang C-F. Sensitivity Study of CO2 Storage Capacity in Brine

Aquifers with Closed Boundaries: Dependence on Hydrogeologic Properties. Sixth Annual

Conference on Carbon Capture and Sequestration, 2007.

Ziemkiewicz PF, He YT. Evolution of water chemistry during Marcellus Shale gas development:

A case study in West Virginia. Chemosphere 2015; 134: 224-231.

126

Appendix A Supporting information for Chapter 3

A1. Mineral characterization

Vermiculite is a common layered silicate clay mineral. Its mineral composition was

analyzed by XRD as indicated in Figure A1. The XRD pattern shows that the vermiculite

is a blend of vermiculite, mica and amphibole. Its chemical composition was analyzed by

ICP-AES and ICP-MS (Table A1). Mg and K are typically the principle exchangeable ion

present in the interlayers of vermiculite clay; other cations detected in this clay are

coordinating cations (e.g. Si, Al). The cation exchange capacity (CEC) determined by the

ammonium acetate method is 43.20 meq/100 g, which is similar to 40.08 meq/100 g of

vermiculite reported in literature (Malandrino et al., 2006). The Qtz and Cal column do not

have cation exchange capacity in our study. The surface areas of quartz, calcite and

vermiculite were measured using the Brunauer-Emmett-Teller (BET) method

(Micromeritics ASAP-2020 surface analyzer).

127

Figure A1. XRD pattern of vermiculite shows that it contains regularly interstratified mica-

vermiculite (11.96Ȧ peak), vermiculite (14.41Ȧ peak) and mica (10.12 Ȧ peak). This sample was

estimated to contain approximately 50% mica based on the peak intensities of regularly

interstratified mica-vermiculite (11.96Ȧ peak) and mica (10.12 Ȧ peak). All the other peaks are

related to the three main phases given above. However, the trace peak at 8.483 Ȧ may be due to

amphibole impurity.

Table A1. Chemical composition of vermiculite

Composition Weight percent (%) Si02 37.4

MgO 20.5

Al2O3 8.62

CaO 7.63

Fe2O3 8.12

K2O 5.04

BaO 0.07

MnO 0.06

Na2O 0.22

P2O5 3.77

SrO 0.06

TiO2 1.09

Cr2O3 0.09

ZnO 0.01

128

Figure A2. (A) The Schematic of flow-through column experiments. The groundwater was injected

to pre-equilibrate with minerals for 6.0 residence times before and after the injection of MSW pulse.

(B) A Picture of the column experiment setup.

A2. Determination of column porosity and permeability

The porosity of columns was calculated using the water used for column packing

divided by the total volume of the columns. To determine permeability, a Crystal

Engineering pressure gauge (XP2i-DP) was used to measure the pressure gradients along

each column at six steady state flow rates from 0.5, 1.0, 2.0, 3.0, 4.0 to 5.0 ml/min. At each

129

flow velocity, the pressure gradient was measured three times. The effective permeability

was calculated using Darcy’s law based on the measured flow rates and pressure gradients.

A3. Chemical species analysis

Effluent samples were collected every half an hour using an auto-sampler. Values

of pH were measured immediately after sample collection. Anion samples were filtered by

0.22 μm membrane, diluted and transferred into 0.5 mL vials for analysis on a Dionex

DX120 ion chromatograph. Cations samples were diluted and analyzed using a Perkin-

Elmer Optima 5300DV inductively coupled plasma-atomic emission spectrometer (ICP-

AES). Trace metals were measured using Inductively Coupled Plasma Mass Spectrometry

(ICP-MS).

A4. Reaction network, thermodynamics, and kinetics

Based on the model calibration, mineral reactions are listed in Table A2 with their

equilibrium constants and reaction kinetics. The reaction rates follow the transition-state-

theory-based (TST) rate law (Lasaga, 1998):

(A1)

Here Ri,tot is the total reaction rate of multiple reactions that the species i is involved

in (mol s-1), Ai,j is the reactive surface area per unit volume (m2/m3) of mineral j that

involves species i, and km,j is the rate constant ((mol/m2)/s). The ion activity product (IAPj)

is 2 2

3Ca COa a

, for example, for calcite dissolution, and Keq,j is the equilibrium constant of

mineral reaction j with values from the standard EQ3/6 geochemical database (Wolery et

al., 1990b). The value of IAPj/Keq,j quantifies the distance to equilibrium.

, , ,

1 ,

[1 ( )]nk

j

i tot m j i j

j eq j

IAPR k A

K

130

A representative ion exchange reaction combining the half reactions in Table A3 is

as follows (Ba2+ as an example) (Appelo and Willemsen, 1987; Vanselow, 1932):

(A2)

The selectivity coefficient can be expressed as

(A3)

(A4)

Here (aq) and (s) refer to the aqueous and solid species, respectively; X- denotes

negatively charged exchange sites on vermiculite; and are the fractions of species

Ba and Mg exchanged on vermiculite, respectively; S is the solid phase concentration on

vermiculite; C is the aqueous concentration; is the activity coefficient calculated using

the Davies approximation (Davies, 1962). The selectivity coefficients (Ksc) indicate cation

affinity to solid surface. In general, the affinity to the solid surfaces are in the order of trace

metals > Ba, Sr > Ca, Mg > Na and K (Merkel and Planer-Friedrich, 2008). This means

that under similar concentration conditions, trace metals tend to sorb onto clay surface first

before other cations. However, a low-affinity cation can still exchange onto clay surface

when its concentration is high compared to other species. The high concentration of Na in

MSW can lead to the exchange of Na onto solid surface compared to Ca and Mg.

2+ 2+

2 2Ba (aq)+MgX (s) BaX (s)+Mg (aq)

MgC( / Mg)

C

Ba

Mg

S Mg

sc

S Ba Ba

fK Ba

f

, Ba Mg

MgBaS S

total total

SSf f

S S

BaSf MgSf

131

Table A2. Reaction network, Reaction thermodynamics, and kinetics

Chemical reactions Minerals aLog Keq bLog k (mol/m2

/s)

cSSA

(m2/g)

Aqueous complexation (at equilibrium) H2O H+ + OH- -14.00 H2CO3

o H+ + HCO3- -6.35

HCO3- H+ + CO3

2- -10.33 MgHCO3

+ Mg2+ + HCO3- -1.04

CaHCO3+ Ca2+ + HCO3

- -1.11 SrHCO3

+ Sr2+ + HCO3- -1.18

BaHCO3+ Ba2+ + HCO3

- -0.98

MnHCO3+ Mn2+ + HCO3

- -1.95 CuHCO3

+ Cu2+ + HCO3- -2.70

ZnHCO3+ Zn2+ + HCO3

- -2.10 CdHCO3

+ Cd2+ + HCO3- -1.50

PbHCO3+ Pb2+ + HCO3

- -2.90

Mn(OH)2(aq) Mn2+ + 2OH- 22.2 Cu(OH)2(aq)+2H+ Cu 2+ + 2H2O 13.68

Zn(OH)2(aq)+2H+ Zn 2+ + 2H2O 16.90 Cd(OH)2(aq)+2H+ Cd 2+ + 2H2O 20.35 Pb(OH)2(aq)+2H+ Pb2+ + 2H2O 17.12

BaSO4(aq) Ba2+ + SO42- -2.70

SrSO4(aq) Sr2+ + SO42- -2.29

CaSO4(aq) Ca2+ + SO42- -2.30

Mineral reactions

SiO2(s) ⇔ SiO2(aq) Quartz -4.00 -13.41 0.01

CaSO4(s) Ca2+ + SO42- Gypsum -4.36 -2.79 1.44

CaCO3(s) Ca2+ + CO32- Calcite -8.48 -7.80 0.56

MgCO3(s) + 2H+ Mg2+ + HCO3- Magnesite 2.50 -4.21 1.87

MnCO3(s) Mn2+ + CO32- Rhodochrosite -11.13 -5.00 0.91

CuCO3(s) Cu2+ + CO32- CuCO3 -9.63 -5.90 5.00

CdCO3(s) Cd2+ + CO32- Otavite -12.10 -3.00 1.37

PbCO3(s) Pb2+ + CO32- Cerussite -13.13 -5.00 1.00

ZnCO3(s) Zn2+ + CO32- Smithsonite -10.00 -5.00 2.90

BaSO4(s) ⇔ Ba2+ + SO42- Barite -9.97 -8.0 1.00

BaCO3(s) ⇔ Ba2+ + CO32- Witherite -8.56 -8.57 2.75

Ba(OH)2(s) + 2H+ Ba2+ + 2H2O Ba(OH)2 24.49 -5.00 1.00

SrSO4(s) + H+⇔ Sr2+ + HSO4 Celestite -6.63 -5.66 1.22

SrCO3(s) ⇔ Sr2+ + CO32- Strontianite -9.27 -9.00 1.40

Sr(OH)2(s) + 2H+ Sr 2+ + 2H2O Sr(OH)2 27.52 -5.00 1.00

Note: aEquilibrium constant (Keq) values are from EQ3/6 (Ball and Nordstrom, 1991; Wolery and

Daveler, 1992). b,cKinetic rate constants and SSA are winthin the range in literature (Bucca et al., 2009;

Liu et al., 2008; Palandri and Kharaka, 2004; Pokrovsky and Schott, 2002; Ticknor and Saluja, 1990).

132

A4. Solid solution partitioning in CrunchFlow

In our simulation, we use reactions (RS1) and (RS2) to realize the reaction 6.

They are mathematically equivalent with the mechanism of solid solution partitioning as

described below.

(RS1)

(RS2)

The reaction (RS1) is kinetically-controlled and can be calibrated based on DMe

value from literature (Rimstidt et al., 1998). A large DMe value indicates a strong

preferential partitioning of Me into calcite. The reaction (RS2) is a thermodynamic-

controlled reaction and proceed from left to right therefore precipitating the MeCO3 as

solid phase.

Table A3. Half-reaction ion exchange selectivity coefficienta.

Ion exchange reactions LogK

NaX ⇔ Na+ + X- 0.00

HX ⇔ H+ + X- -1.20

KX ⇔ K+ + X- -0.90

CaX2 ⇔ Ca2+ + 2X- -0.22

MgX2 ⇔ Mg2+ + 2X- -0.86

BaX2 ⇔ Ba2+ + 2X- -0.42

SrX2 ⇔ Sr2+ + 2X- -0.42

MnX2 ⇔ Mn2+ + 2X- -0.82

CuX2 ⇔ Cu2+ + 2X- -3.62

ZnX2 ⇔ Zb2+ + 2X- -2.20

PbX2 ⇔ Pb2+ + 2X- -2.70 aSelectivity coefficients for ion exchange reaction are calculated based on our batch experiment

and calibrated to fit the observed experimental BTCs.

2 2

3 3(s) (aq) CaCO Me MeCO Ca

3 3(aq) (s)MeCO MeCO

133

References

Appelo, C.A.J. and Willemsen, A. (1987) Geochemical calculations and observations on salt water

intrusions, I. A combined geochemical/minxing cell model. Journal of Hydrology 94, 313-

330.

Ball, J.W. and Nordstrom, D.K. (1991) User's manual for WATEQ4F, with revised thermodynamic

data base and test cases for calculating speciation of major, trace, and redox elements in

natural waters.

Bucca, M., Dietzel, M., Tang, J., Leis, A. and Köhler, S.J. (2009) Nucleation and crystallization of

otavite, witherite, calcite, strontianite, hydrozincite, and hydrocerussite by CO 2 membrane

diffusion technique. Chemical geology 266, 143-156.

Davies, C. (1962) Ion Association, Butterworths, London, .

Lasaga, A. (1998) Kinetic Theory in the Earth SciencesPrinceton Univ. Press, Princeton.

Liu, H., Korshin, G.V. and Ferguson, J.F. (2008) Investigation of the kinetics and mechanisms of

the oxidation of cerussite and hydrocerussite by chlorine. Environ Sci Technol 42, 3241-

3247.

Malandrino, M., Abollino, O., Giacomino, A., Aceto, M. and Mentasti, E. (2006) Adsorption of

heavy metals on vermiculite: influence of pH and organic ligands. Journal of Colloid and

Interface Science 299, 537-546.

Merkel, B.J. and Planer-Friedrich, B. (2008) Groundwater Geochemistry. Springer-Verlag Berlin

Heidelberg.

Palandri, J.L. and Kharaka, Y.K. (2004) A Compliation of rate parameters of water-mineral

interaction kinetics for application to geochemical modeling. U.S. Department of the

Interior, p. 71.

Pokrovsky, O. and Schott, J. (2002) Surface chemistry and dissolution kinetics of divalent metal

carbonates. Environ Sci Technol 36, 426-432.

Rimstidt, J.D., Balog, A. and Webb, J. (1998) Distribution of trace elements between carbonate

minerals and aqueous solutions. Geochimica et Cosmochimica Acta 62, 1851-1863.

Ticknor, K.V. and Saluja, P.P. (1990) Determination of surface areas of mineral powders by

adsorption calorimetry. Clays and Clay Minerals 38, 437-441.

Vanselow, A.P. (1932) The utilization of the base exchange reaction for the determination of

activity coefficients in mixed electrolytes. Journal of the American Chemical Society 54, 5.

Wolery, T.J. and Daveler, S.A. (1992) EQ6, a computer program for reaction path modeling of

aqueous geochemical systems: theoretical manual, user's guide and related documentation

(Version 7.0). Lawrence Livermore Laboratory, University of California Livermore, CA.

Wolery, T.J., Jackson, K.J., Bourcier, W.L., Bruton, C.J., Viani, B.E., Knauss, K.G. and Delany,

J.M. (1990) Current status of the EQ3/6 software package for geochemical modeling, ACS

symposium series. Oxford University Press, pp. 104-116.

134

134

Appendix B Data Article for Chapter 3

Title: Time series of effluent chemistry data on mineralogy controls on the

reactive transport of Marcellus Shale waters

Abstract

Produced or flowback waters from Marcellus Shale gas extraction (MSWs)

typically contain high levels of salinity and contaminants including trace metals, which

pose significant concerns on water quality. This data article document column

experimental data that accompany the original article Cai et al.[1]. Effluent chemistry

data from three flow-through columns with different mineralogical compositions are

presented here: a quartz (Qtz) column, a calcite-rich (Cal) column, and a clay-rich (Vrm,

vermiculite). The same pulse of MSWs was injected in each of the three columns to

mimic the leakage in natural systems. The effluent chemistry records the response of

water chemistry to such perturbation for 25 residence times. The collected time series

data include major cations (Na, K, Mg, Ca, Ba, Sr), anions (Br, Cl, SO4), and metals

(Cd, Zn, Mn, Cu, Pb).

Keywords: Environmental pollution; Column experiment; Marcellus Shale waters;

Trace metals;

135

135

Table B1. Specifications about the data article.

Subject area Environmental Sciences

More specific subject area

Geochemistry

Type of data Table

How data was acquired

The following instruments are used, including Perkin-Elmer Optima 5300DV inductively coupled plasma-atomic emission spectrometer (ICP-AES), X Series II-SBM and X Series II-MFM Inductively Coupled Plasma - Mass Spectrometry (ICP-MS), Dionex™ Aquion™ Ion Chromatography (IC) System, SevenMult pH meter (METTLER TOLEDO).

Data format Analyzed

Experimental factors All aqueous solutes are filtered by 0.45 μm filter before ICP-AES and ICP-MS analysis. The mineral and chemical composition of vermiculite are analyzed.

Experimental features We conducted flow-through column experiments (50 cm by 5 cm) with different minerals and collected the effluent samples.

Data source location University Park, Pennsylvania, USA

Data accessibility Data is available in this article.

Related research article

Data is submitted as a companion data article for the published research article (Cai et al., 2018a).

Value of the Data

Data highlights the importance of mineralogy in controlling natural attenuation

processes of MSWs.

Data can be further used in the water quality risk assessment associated with

contamination from MSW release.

Data can be used to develop prediction models and design experiments under other

environmental conditions such as different MSWs, groundwater chemistry, and

mineralogy.

B1. Data

Shale gas extraction from Marcellus formation, one of the largest shale gas play

in the United States, has raised significant concerns about its impacts on natural water

resources associated with the rising reported spills (Brantley et al., 2018; Brantley et al.,

2014b; Cai and Li, 2016; Maloney et al., 2017; Patterson et al., 2017; Vidic et al.,

2013b). Produced or flowback waters from Marcellus formation (abbreviated as

136

136

Marcellus Shale waters, MSWs) typically have high concentrations of salinity and

contaminants (e.g., trace metals). These waters pose potential risks on water resources.

The column experiments explored the role of mineralogy in the reactive transport of

MSWs. To represent sand, carbonate, and clay-rich aquifers, three columns (5 cm in

diameter by 50 cm in length) were packed with quartz (Qtz), calcite-rich (Cal), and

clay-rich (Vrm, vermiculite). The same pulse of MSWs was injected in each of the three

columns to mimic the leakage in natural systems. The effluent data from the columns

document the response of water chemistry to such MSW perturbation to the

groundwater water and elucidate important processes that control the reactive transport

of MSWs.

B2. Experimental Design, Materials, and Methods

Please refer to the research article and SI for detailed set up of the columns (Cai

et al., 2018a). The Effluent samples were collected every half an hour using an auto-

sampler from the flow through column (5 cm in diameter by 50 cm in length). Anion

samples were filtered by 0.22 μm membrane, diluted and transferred into 0.5 mL vials

for analysis in a Dionex DX120 ion chromatograph. Cations were diluted and analyzed

using a Perkin-Elmer Optima 5300DV inductively coupled plasma-atomic emission

spectrometer (ICP-AES). Trace metals were measured using Inductively Coupled

Plasma Mass Spectrometry (ICP-MS). The temporal element geochemistry of effluent

samples in Qtz, Cal, and Vrm column are listed in Table B1, B2, and B3, respectively.

The pH in three columns were measured immediately by pH meter (METTLER

TOLEDO) after sample collection (Table 4).

Acknowledgments

137

137

We acknowledge Matthew Gonzales and Laura Liermann from College of Earth

and Mineral Sciences in providing help for the analyses of cations including heavy

metals. Xin Gu from Department of Geoscience and Huaibin Zhang from College of

Agricultural Sciences assisted with the vermiculite sample analysis. Sruthi Kakuturu

helped with sample collection. This work was supported by the U.S. Department of

Energy (DOE) Subsurface Biogeochemistry Research program DE-SC0007056.

138

138

Table B2. Time series of effluent geochemistry in the Qtz column (units: mg/L)

Time

(hr) Br(ppm) Cl(ppm) SO4(ppm) Na(ppm) Ca(ppm) Mg(ppm) K(ppm) Ba(ppm) Sr(ppm) Mn(ppm) Zn(ppm) Cu(ppm) Pb(ppm) Cd(ppm)

0.00 0.05 32 12.52 19.01 15.43 2.63 2.16 0.09 0.12 0.0160 0.0081 0.0016 0.00004 0.00003

4.50 0.06 32.96 12.53 18.53 15.46 2.62 2.15 0.09 0.13 0.0140 0.0078 0.0012 0.00005 0.00003

7.00 0.02 32.56 12.89 18.44 15.23 2.65 2.16 0.28 0.12 0.0180 0.0135 0.0011 0.00004 0.00003

12.00 0.1 32 12.45 19.18 15.74 2.68 2.18 0.35 0.12 0.0206 0.0091 0.0005 0.00003 0.00003

14.50 1.01 31 12.3 32.71 25.09 4.56 5.65 0.65 0.21 0.048 0.0118 0.0009 0.00004 0.00003

15.00 1.16 82.11 13.08 109.06 71.47 12.36 9.33 1.78 0.57 0.1318 0.0618 0.001 0.00084 0.00003

15.50 23.48 1154.01 12.9 386.23 196.71 31.05 15.19 5.35 4.68 0.3585 0.1992 0.0058 0.00084 0.0001

16.00 48.54 2243.09 12.86 754.47 360.78 39.93 22.12 16.76 38.37 0.482 0.2026 0.0154 0.00808 0.00012

16.50 61.03 2766.44 8.36 1002.94 411.9 41.36 23.03 62.01 72.63 0.5584 0.252 0.0276 0.0074 0.00016

18.00 66.2 2980.77 7.52 1058.58 437.93 33.31 21.25 113.06 79.44 0.7068 0.2553 0.0433 0.00606 0.00018

19.50 54.31 2459.86 5.5 867.11 355.74 23.87 17.68 90.03 64.32 0.9103 0.2274 0.0387 0.00538 0.00144

21.00 44.81 2076.1 3.91 - - - - - - - - - - -

22.00 - - - 628.46 260.72 20.03 12.82 61.52 47.45 0.7382 0.2196 0.0265 0.00493 0.00205

22.50 35.64 1688.72 2.12 - - - - - - - - - - -

24.00 29 1268 2.5 - - - - - - - - - - -

24.50 - - 5.4 400 207.95 12.21 12.3 48.16 38.69 0.6176 0.1186 0.0173 0.00252 0.00200

27.00 12.09 558.64 7 275 127.18 2.99 4.5 24.18 23.65 0.3779 0.1033 0.01 0.00042 0.00140

27.50 - - 8.1 - - - - - - - - - -

29.50 - - - 59.54 30.01 2.05 3.96 6.29 5.72 0.0871 0.0431 0.0037 0.00042 0.00010

30.00 2.72 131.93 - - - - - - - - - - - -

31.50 0.06 36.56 12.85 - - - - - - - - - - -

33.00 0.07 37.52 12.53 - - - - - - - - - - -

34.50 0.06 37.25 12.52 18.08 14.65 2.6 3.04 3.4 2.41 0.0455 0.0141 0.0016 0.00004 0.00005

39.50 0.05 36.75 12.53 18.65 14.84 2.59 2.28 0.76 0.57 0.015 0.0128 0.0015 0.00005 0.00005

139

139

44.50 0.05 32.62 12.89 18.73 15.33 2.61 2.07 0.36 0.2 0.019 0.008 0.0016 0.00004 0.00004

47.00 - - - - - - - - - - - - -

54.50 0.06 31.5 12.98 19.02 14.75 2.65 2.16 0.09 0.18 0.0170 0.0078 0.0011 0.00002 0.00003

59.50 0.04 32.61 12.78 18.8 14.78 2.57 2.14 0.08 0.15 - - - - -

64.50 0.05 31.42 12.65 18.9 14.57 2.62 2.16 0.09 0.14 - - - - -

69.50 0.05 32.53 12.79 18.86 14.82 2.6 2.18 0.08 0.14 0.0180 0.008 0.0012 0.00004 0.00003

79.50 0.04 32.1 12.51 18.79 15.13 2.54 2.02 0.07 0.14 0.0150 0.0081 0.0015 0.00003 0.00003

99.50 0.05 32.35 12.56 18.75 15.37 2.63 2.15 0.08 0.14 0.0170 0.0083 0.0015 0.00004 0.00003

120.00 0.06 31.8 12.68 18.63 15.41 2.58 2.11 0.09 0.12 0.0160 0.0082 0.0013 0.00004 0.00003

140.00 0.03 32.2 12.85 18.46 15.46 2.59 2.18 0.08 0.13 0.0180 0.0076 0.0016 0.00003 0.00003

160.00 0.04 32.32 12.71 18.53 15.32 2.63 2.16 0.08 0.12 0.0170 0.0077 0.0014 0.00004 0.00003

200.00 0.05 31.72 12.65 18.92 14.75 2.56 2.15 0.07 0.12 0.0160 0.0081 0.0015 0.00003 0.00003

220.00 0.05 32.47 12.94 18.5 14.87 2.64 2.13 0.08 0.13 0.0180 0.0079 0.0016 0.00004 0.00003

250.00 0.04 32.38 12.73 18.64 14.57 2.61 2.12 0.09 0.12 - - - - -

270.00 0.04 31.98 12.68 18.95 14.65 2.63 2.14 0.09 0.12 - - - - -

Note: All analytes measured by ICP-AES except Br, Cl and SO4 (ion chromatograph), Cu, Zn, Pb and Mn (ICP-MS).

Anions (Br, Cl, SO4), cations (Na, Ca, Mg, K, Ba and Sr), and trace metals (Mn, Zn, Cu, and Pb) were measured using different instruments and therefore

have different significant numbers because of different analysis approach. Hyphen (-) indicates we did not measure it.

140

140

Table B3. Time series of effluent geochemistry in the Cal column (units: mg/L)

Time

(hr) Br(ppm) Cl(ppm) SO4(ppm) Na(ppm) Ca(ppm) Mg(ppm) K(ppm) Ba(ppm) Sr(ppm) Mn(ppm) Cu(ppm) Zn(ppm) Pb(ppm) Cd(ppm)

0.25 0.07 32.09 12.85 22.47 17.39 2.29 2.86 0.09 0.14 0.0002 0.0017 0.0017 0.00004 UDL

2.25 0.1 33.08 12.96 22.64 17.18 2.27 2.82 0.08 0.13 0.0003 0.0017 0.002 0.00003 UDL

5.75 - - - 22.29 16.98 2.25 3.44 0.1 0.12 0.0009 - - - UDL

6.25 0.06 31.96 13.01 21.92 17.12 2.25 3.56 0.11 0.16 0.0008 0.0017 0.0016 0.00009 UDL

7.25 0.69 42.09 12.01 180.53 115.88 2.97 4.68 0.36 1.23 0.0049 0.0002 0.0018 0.00002 UDL

8.25 6.3 348 11.32 240.61 111.01 13.42 8.97 2.94 6.25 0.0042 0.002 0.0149 0.000125 UDL

8.75 10.39 631.49 9.51 - - - - 10.36 18.81 0.0078 - - - UDL

9.25 16.53 965.96 9.2 386.37 147.96 15.02 9.32 15.31 18.98 0.0057 0.0078 0.0357 0.0003 UDL

9.75 21 1251 9.94 500.38 308.53 30.65 16.16 41.37 38.72 0.0104 - - - UDL

11.25 35.7 1937 9.25 769.97 266.73 24.91 11.64 47.8 42.59 0.012 0.0116 0.0482 0.00072 UDL

11.75 40.75 2182.53 6.59 858 288.19 27.98 13.4 57.69 48.16 0.0138 0.0226 0.0978 0.00146 UDL

12.25 45.37 2401.78 8.07 953.97 318.06 29.95 15.31 66.8 54.24 0.0159 0.027 0.1468 - UDL

12.75 46.01 2436 8.18 993.63 325.92 30.87 14.12 67.66 56.15 0.0138 0.0257 0.117 0.00247 UDL

13.25 48.75 2565.5 8.98 1036.73 333.15 30.87 16.79 71.36 58.71 0.0148 0.0269 0.1025 - UDL

13.75 52.01 2703 6.74 1061.14 349.78 32.91 15.22 78.08 60.5 0.0169 - - - UDL

14.25 52.03 2714.55 7.35 1100.57 349.46 32.32 15.2 76.39 62.14 0.0164 0.0289 0.1096 0.00247 UDL

14.75 54.32 2832.96 5.16 1150.25 362.92 33.45 12.54 79.34 64.54 0.0172 0.0277 0.0992 - UDL

15.25 55.66 2887.66 5.79 1141.29 365.29 34.09 12.78 79.44 64.62 0.0185 0.0277 0.0988 0.00247 UDL

15.75 - - - 1156.66 361.25 32.98 15.64 79.86 64.17 0.019 0.0301 0.1127 - UDL

16.25 49.56 2591 5.65 1092.5 323.68 29.44 15.66 70.02 59.53 0.0159 0.0277 0.0686 - UDL

16.75 44.72 2362.7 7.21 968.44 295.25 27.04 13.48 63.3 52.06 0.0159 0.0227 0.0937 0.00247 UDL

17.25 37.95 2122.39 7.72 847.66 256.52 23.41 13.91 54.16 44.78 - 0.0218 0.0752 0.00247 UDL

18.25 - - - 712.52 216.33 19.39 10.24 43.46 37.08 0.0078 0.016 0.0805 0.00124 UDL

18.75 25.11 1471.47 7.2 603.69 181.5 16.76 10.23 35.05 31.37 0.0083 0.0156 0.0812 - UDL

20.25 18.09 1090.3 7.03 455.85 142.02 13.17 8.66 24.73 23.73 0.0074 0.0093 0.0447 0.00093 UDL

141

141

21.75 13.08 811.48 8.42 345.92 108.71 10.06 9.06 17 18.31 0.0045 0.0091 0.0452 0.00124 UDL

23.25 10.23 618.19 8.9 267.14 87.01 8.08 5.91 12.8 14.45 0.0039 0.0081 0.0284 - UDL

24.75 8.8 544.96 10.43 246.03 81.65 7.61 9.84 9.77 12.88 0.0032 0.0064 0.0242 0.00041 UDL

26.25 6.89 439.31 10.36 185.93 67.86 6.6 8.54 8.51 10.6 0.0028 0.0058 0.0352 - UDL

27.25 6.31 354.44 10.41 155.84 58.01 5.62 7.27 6.66 8.96 0.0021 - - - UDL

28.75 4.96 279.76 12.24 129.99 49.92 4.87 6.29 4.43 7.76 0.0026 0.0052 0.02 0.00082 UDL

30.25 3.58 207.13 12.44 100.37 41.53 4.25 6.17 2.93 5.96 0.0017 - - - UDL

31.25 2.85 166.36 12.6 83.02 36.26 3.82 4.99 2.45 4.93 0.0019 0.0034 0.0088 0.00041 UDL

32.75 2.18 129.42 12.77 68.77 32.29 3.44 4.71 1.94 3.92 0.0015 - - - UDL

34.25 1.69 99.71 12.73 56.61 28.16 2.99 7.31 1.51 3.39 0.0011 0.0027 0.0061 0.00021 UDL

35.75 1.14 68.02 12.99 45.72 24.96 2.77 4.89 0.95 2.64 0.0011 - - - UDL

36.75 - - - 42.11 23.33 2.62 4.87 0.53 2.4 0.0011 - - - UDL

39.75 - - - 30.65 19.81 2.35 2.62 0.26 1.51 0.0016 0.0021 0.0033 0.00019 UDL

42.25 - - - 26 18.44 2.22 5.74 0.13 0.73 0.0008 - - - UDL

44.75 - - - 24.6 17.93 2.18 6.53 0.10 0.52 0.0009 0.0025 0.0023 0.00003 UDL

46.75 - - - 23.36 17.54 2.22 4.04 0.08 0.35 0.0007 - - - UDL

54.75 0.08 39.01 12.98 21.32 17.1 2.18 5.68 0.09 0.18 0.0007 0.0023 0.0017 0.00003 UDL

59.75 0.06 31.92 13.02 21.24 17.02 2.17 4.65 0.08 0.18 0.0009 0.002 0.0016 0.00003 UDL

64.75 - - - 21.49 17.27 2.21 3.38 0.09 0.18 0.001 0.0017 0.0018 0.00004 UDL

74.75 0.07 32 12.98 22.33 17.25 2.23 3.55 0.08 0.18 0.0009 - - - UDL

79.75 - - - 22.26 17.31 2.23 3.54 0.10 0.16 0.001 - - - UDL

84.75 - - - 22.34 17.52 2.23 2.86 0.09 0.17 0.0009 0.0018 0.0017 0.00005 UDL

89.75 0.08 31.2 12.93 - - - - - - - - - - UDL

97.25 - - - 22.51 17.58 2.25 2.77 0.10 0.17 0.001 - - - UDL

102.25 0.08 32 12.92 22.17 17.86 2.25 2.73 0.11 0.16 0.0009 - - - UDL

107.25 - - - 22.2 17.64 2.23 2.68 0.09 0.16 0.0009 - - - UDL

117.25 - - - 21.62 17.52 2.19 3.8 0.07 0.15 0.0009 - - - UDL

129.75 - - - 23.13 18.28 2.28 2.82 0.09 0.15 0.0009 - - - UDL

142

142

138.75 - - - 22.62 18.23 2.27 3.09 0.10 0.13 0.0008 - - - UDL

158.25 - - - 22.75 17.78 2.24 2.77 0.09 0.14 0.0009 - - - UDL

169.75 - - - 23.16 18.44 2.29 2.76 0.10 0.14 0.0009 - - - UDL

179.75 - - - 23.39 18.66 2.3 2.86 0.09 0.15 0.0009 - - - UDL

209.75 - - - 22.82 18.7 2.25 2.81 0.08 0.14 0.0008 - - - UDL

214.75 - - - 22.98 18.32 2.2 2.78 0.10 0.15 0.0009 - - - UDL


UDL means under detection limit.



143

143

Table B4. Time series of effluent geochemistry in the Vrm column (units: mg/L)

Time

(hr) Br(ppm) Cl(ppm) SO4(ppm) Na(ppm) Ca(ppm) Mg(ppm) K(ppm) Ba(ppm) Sr(ppm) Mn(ppm) Cu(ppm) Zn(ppm) Pb(ppm) Cd(ppm)

0.25 0.08 33.81 12.89 19.09 5.23 8.34 11.53 0.03 0.05 0.0003 0.0005 0.0025 0.00001 UDL

2.25 0.07 33.45 12.61 18.59 5 7.97 11.58 0.04 0.05 0.0002 0.0003 0.0023 0.00002 UDL

4.75 0.05 32.31 12.53 18.96 5.05 8.16 12.1 0.04 0.05 0.0002 0.0004 0.0024 0.00001 UDL

7.75 0.18 32.31 12.56 19.35 4.99 8.2 11.97 0.04 0.05 0.0002 0.0003 0.0027 0.00003 UDL

8.75 2.27 132.03 12.61 31.96 16.75 25.66 18.68 0.13 0.17 0.0057 0.0003 0.0247 0.0001 UDL

9.75 10.06 503.38 12.98 55.8 67.56 98.54 35.71 0.54 0.71 0.0206 0.0005 0.0102 0.0001 UDL

10.75 18.89 933.86 12.95 86.98 120.6 178.29 47.03 0.99 1.25 0.0365 0.0015 0.0621 0.00056 UDL

11.75 25 1192.36 12.92 166.5 150.71 226.38 54.63 1.28 1.54 0.0492 0.0014 0.0633 0.00056 UDL

13.25 32.71 1534.43 12.94 276.34 165.68 253.84 61.04 1.41 1.61 0.0571 0.001 0.0594 0.00092 UDL

15.25 38.39 1797.98 12.99 401.17 167.44 263.07 65.08 1.53 1.63 0.0248 0.0008 0.0248 0.00076 UDL

16.25 40.55 1873.38 12.74 466.78 167.97 264.38 66.46 1.6 1.67 0.0264 0.0015 0.0452 0.00016 UDL

17.25 40.79 1895.99 12.31 505.61 163.37 255.47 66.78 1.62 1.69 0.0265 0.0008 0.033 0.00008 UDL

18.75 29.43 1389.85 12.35 421.47 111.71 175.02 55.42 1.05 1.12 0.0221 0.0028 0.0974 0.00076 UDL

20.25 24.34 1165.75 12.31 361.57 82.67 130.99 48.01 0.79 0.83 0.0162 0.0008 0.0532 0.0006 UDL

23.25 16.31 810.25 12.95 275.2 55.18 87.95 39.44 0.52 0.55 0.0263 0.0006 0.0407 0.00032 UDL

26.75 - - - 219.51 44.9 71.63 33.83 0.43 0.47 0.0233 0.0004 0.0307 0.00005 UDL

32.25 - - - 127.5 33.58 53.91 33.39 0.29 0.33 0.0178 0.001 0.0279 0.00005 UDL

34.75 5.39 305.08 12.2 98.39 26.31 42.68 28.17 0.24 0.27 0.016 0.0004 0.0249 0.00003 UDL

39.75 4.11 210 12.96 77.44 20.37 34.89 25.72 0.18 0.2 0.0143 0.0006 0.0327 0.00001 UDL

44.75 1.52 100.72 12.96 58.28 8.43 14.3 16.49 0.07 0.08 0.0059 0.0003 0.0051 0.00006 UDL

49.75 0.44 48.59 12.87 39.04 3.73 5.36 12.23 0.03 0.04 0.0019 0.0003 0.0071 0.00008 UDL

54.75 0.22 36.05 12.95 36.54 2.83 3.76 10.19 0.02 0.03 0.0013 0.0007 0.0081 0.0001 UDL

59.75 0.08 33.81 12.96 35.88 2.5 2.88 9.06 0.02 0.02 0.0013 0.0002 0.0025 0.0001 UDL

64.75 0.08 33.45 12.89 34.82 2.37 2.79 8.61 0.02 0.02 0.0011 0.0003 0.0049 0.00005 UDL

144

144

69.75 - - - 35.71 2.11 2.55 11.71 0.02 0.02 0.001 0.0003 0.0027 0.00003 UDL

79.25 0.08 33.79 12.57 42.12 2.5 3.1 8.86 0.02 0.02 - - - - UDL

79.75 - - - 40 2.44 3.03 8.19 0.02 0.02 0.0006 0.0003 0.0019 0.0001 UDL

84.75 0.08 33.1 12.62 42.92 2.53 3.17 9.13 0.01 0.02 - - - - UDL

89.75 - - - 40.2 2.37 2.92 7.97 0.02 0.02 0.0006 0.0002 0.0018 0.00005 UDL

94.75 0.08 33.65 12.24 44.12 2.38 2.93 8.71 0.02 0.02 - - - - UDL

99.75 - - - 42.8 2.3 2.77 19.4 0.02 0.02 0.0001 0.0003 0.0024 0.00003 UDL

104.75 0.08 33.79 12.36 43.73 2.3 2.77 8.87 0.01 0.02 - - - - UDL

109.75 0.08 33.1 12.46 41.81 2.29 2.76 9.82 0.02 0.02 - - - - UDL

114.75 - - - 39.4 2.51 2.64 12.1 0.02 0.02 0.0005 0.0002 0.0025 0.00007 UDL

119.75 0.08 33.65 12.89 38.49 2.71 2.95 14.66 0.02 0.02 - - - - UDL

129.75 - - - 35 3.15 3.2 13.8 0.03 0.03 0.0008 0.0001 0.0022 0.00004 UDL

134.75 0.08 33.63 12.7 35.95 3.94 3.75 14.18 0.02 0.04 - - - - UDL

139.75 0.08 33.53 12.31 34.11 4.48 4.04 13.93 0.03 0.04 - - - - UDL

144.75 - - - 34.3 4.58 4.14 12.3 0.03 0.04 - - - - UDL

149.75 0.08 32.94 12.49 33.64 4.96 4.43 13.21 0.04 0.05 - - - - UDL

159.75 0.08 33.48 12.5 32.34 5.22 4.66 11.75 0.04 0.05 0.0001 0.0007 0.0027 0.00008 UDL

174.75 - - - 30.88 5.5 4.85 11.31 0.04 0.06 0.0002 0.0015 0.003 0.00008 UDL

189.75 0.08 33.11 12.23 30.69 6.1 5.3 11.74 0.04 0.06 0.0001 0.001 0.0024 0.00009 UDL

204.75 0.08 33.17 12.42 29.4 6.29 5.41 11.51 0.05 0.06 0.0002 0.0008 0.0024 0.00006 UDL

219.75 0.08 33.31 12.26 28.36 6.48 5.55 11.58 0.05 0.07 0.0003 0.0007 0.0028 0.00006 UDL

249.75 0.08 33.7 12.51 29.07 7.64 6.43 13.08 0.05 0.08 0.0001 0.0006 0.0022 0.00008 UDL

262.25 0.08 33.61 12.43 25.68 7.17 6.03 11.69 0.06 0.08 - - - - UDL

274.75 0.08 33.32 12.47 26.08 7.78 6.52 12.47 0.1 0.09 0.0001 0.0007 0.0029 0.00006 UDL

287.25 0.08 33.14 12.68 24.95 7.64 6.33 12.17 0.14 0.1 - - - - UDL

299.75 0.08 33.84 12.6 24.58 7.64 6.31 12.11 0.18 0.1 0.0002 0.0006 0.0020 0.00008 UDL

312.25 0.08 33.1 12.57 23.81 7.62 6.23 12.08 0.24 0.11 0.0001 0.0007 0.0023 0.00009 UDL

324.75 0.08 33.8 12.49 22.56 7.92 6.4 11.92 0.3 0.13 0.0001 0.0009 0.0027 0.00006 UDL

145

145

337.25 0.08 33.6 12.53 22.8 8.04 6.44 12.06 0.37 0.15 - - - - UDL

349.75 0.08 33.29 12.55 22.33 8.17 6.48 12.1 0.41 0.16 - - - - UDL

362.25 0.08 33.33 12.45 21.97 8.44 6.57 12.24 0.55 0.22 - - - - UDL

374.75 0.08 33.34 12.65 21.68 8.49 6.51 11.99 0.62 0.25 - - - - UDL

387.25 0.08 33.27 12.44 21.32 8.85 6.65 12.3 0.73 0.31 - - - - UDL

399.75 0.08 33.5 12.41 21.28 8.92 6.66 11.99 0.78 0.35 - - - - UDL

412.25 0.08 33.42 12.47 22.12 9.16 6.79 11.37 0.85 0.4 - - - - UDL

424.75 0.08 33.18 12.45 20.6 8.92 6.53 10.65 0.85 0.39 - - - - UDL


UDL means under detection limit.



146

146

Table B5. Time series of effluent pH in the three columns

Time (hr) Qtz Cal Vrm

0.25 8.13 8.30 8.48

0.75 8.14 8.31 8.50

1.25 8.13 8.30 8.49

1.75 8.14 8.29 8.51

2.25 8.15 8.30 8.49

2.75 8.14 8.30 8.52

3.25 8.16 8.29 8.50

3.75 8.16 8.31 8.45

4.25 8.13 8.31 8.47

4.75 8.15 8.27 8.49

5.25 8.14 8.28 8.45

5.75 8.16 8.26 8.45

6.25 8.15 8.25 8.48

6.75 8.14 8.26 8.50

7.25 8.14 8.19 8.47

7.75 8.13 8.07 8.51

8.25 8.14 7.95 8.48

8.75 8.13 7.87 8.42

9.25 8.16 7.75 8.29

9.75 8.14 7.76 8.14

10.25 8.15 7.72 8.07

10.75 8.15 7.69 8.00

11.25 8.14 7.70 8.00

11.75 8.14 7.65 8.00

12.25 8.12 7.65 7.97

12.75 8.14 7.67 7.97

13.25 8.13 7.62 7.97

13.75 8.12 7.62 7.99

14.25 8.1 7.69 8.00

14.75 8.04 - 7.97

15.25 7.87 7.67 7.97

15.75 7.73 7.69 7.92

16.25 7.67 7.65 7.96

16.75 7.67 7.75 7.99

17.25 7.66 7.72 8.02

17.75 7.68 7.72 8.03

18.25 7.68 7.74 8.08

18.75 7.66 7.74 8.13

19.25 7.66 7.77 8.16

19.75 7.69 7.79 8.18

20.25 7.68 7.78 8.24

20.75 7.69 7.80 8.26

21.25 7.68 7.82 8.30

21.75 7.72 7.87 8.32

22.25 7.7 7.86 8.33

147

147

22.75 7.73 7.93 8.34

23.25 7.73 7.91 8.38

23.75 7.74 7.90 8.38

24.25 7.75 7.94 8.37

24.75 7.75 7.96 8.42

25.25 7.78 7.99 8.42

25.75 7.78 7.99 8.38

26.25 7.79 8.01 8.42

26.75 7.81 - 8.41

27.25 7.82 8.05 8.43

27.75 - - 8.45

28.25 - 8.05 8.46

28.75 7.92 8.05 8.45

29.25 7.95 8.09 8.46

29.75 8.03 8.07 8.45

30.25 8.09 - 8.45

30.75 - - 8.45

31.25 8.13 8.16 8.45

31.75 8.14 8.15 8.45

32.25 8.16 8.14 8.46

32.75 8.16 8.16 8.47

33.25 8.16 8.16 -

33.75 8.15 8.20 -

34.25 8.16 8.18 -

34.75 8.15 8.17 -

35.25 8.16 8.20 -

35.75 8.16 8.23 -

36.25 8.15 8.22 8.46

36.75 8.16 - -

37.25 8.15 - -

38.75 - 8.25 8.45

39.25 - 8.26 -

40.25 - 8.24 -

40.75 - 8.24 -

41.25 - 8.25 -

41.75 - 8.25 -

42.25 - 8.26 8.50

42.75 - 8.27 -

43.25 - 8.27 -

43.75 - 8.26 -

44.75 - 8.29 -

45.25 - 8.29 8.53

45.75 - - 8.52

46.25 - - 8.53

46.75 - 8.30 8.50

47.75 - - 8.48

50.75 - - 8.48

52.25 - 8.29 8.45

148

148

53.25 - - 8.54

54.75 - - 8.55

55.25 - - 8.52

58.75 - 8.27 -

59.75 - 8.28 -

60.25 - 8.28 -

63.75 - - 8.52

68.75 - - 8.62

69.25 - - 8.52

69.75 - 8.29 8.55

72.75 - - 8.55

73.25 - - 8.54

73.75 - - 8.55

74.25 - - 8.54

74.75 - - 8.54

76.75 - - 8.53

79.75 - 8.30 -

80.25 - - 8.54

89.75 - - 8.52

99.75 - 8.29 8.55

120.75 - - 8.54

124.75 - 8.30 -

126.25 - - 8.57

141.25 - - 8.57

145.25 - - 8.57

146.25 - - 8.52

146.75 - - 8.51

147.25 - - 8.52

149.75 - 8.29 -

166.75 - - 8.52

174.75 - 8.28 -

176.75 - - 8.55

198.75 - - 8.56

199.25 - - 8.52

199.75 - - 8.53

200.25 - - 8.54

210.25 - - 8.52

227.25 - - 8.50

243.75 - - 8.56

259.75 - - 8.54

276.25 - - 8.53

292.25 - - 8.52

309.25 - - 8.50

325.75 - - 8.52

149

Appendix C Supporting information for Chapter 4

2D Heterogeneous Cell Packing Procedure. A “wet packing method”

(Minyard and Burgos, 2007) was used such that 2 cm water height above the solids was

constant when incremental masses of solids were loaded to the cell. For the 1/2-zone

heterogeneous cell, a rectangular thin pipe (2.80 cm × 0.95 cm × 50.00 cm) was

vertically positioned at 2.13 cm away from the left edge of cell to align the vermiculite

in place. In each increment, one portion of vermiculite and quartz was added into and

outside of the pipe, respectively. After each step, the pipe was gently pulled up to avoid

mixing of vermiculite and quartz. Cell was tapped on the front and rear sides to avoid

the air bubbles and ensure the uniformity. After filling the first 20 cm section, we slowly

removed the pipe and switched to the diagonal position. The pipe was positioned at 2.13

cm away from the right edge of cell and then we completed the other half zonation with

the same packing procedure as for the first 20 cm section. For the 1/4-zone cell, two

rectangular thin pipes (2.24 cm × 0.95 cm × 50.00 cm) were vertically placed into the

cell with the same gap of 2.50 cm between the pipes and edges using the same packing

procedure for the 1/2-zone cell. After filling the first 10 cm, pipes were gently removed.

Small portions of quartz were then added incrementally on top of the first 10 cm matrix

until the section between 10 and 15 cm of cell was filled. Then one pipe was vertically

placed back in the center of the cell with the same packing procedure for the 1/2-zone

cell to pack the section between 15 and 25 cm. After completion, the pipe was removed

and the quartz was incrementally added to the section between 25 and 30 cm. The last

150

10 cm section between 30 and 40 cm was filled with the same packing procedure as the

first 10 cm with two pipes. After that, the cells were secured with the end-cap and then

connected to the groundwater reservoir.

Determination of Cell Porosity and Permeability. The porosity of each cell

was calculated using the water used for cell packing divided by the total volume of the

cell. To determine permeability, a Crystal Engineering pressure gauge (XP2i-DP) was

used to measure the pressure gradients along each column at six steady state flow rates

from 0.5, 1.0, 2.0 3.0, 4.0 to 5.0 ml/min. At each flow velocity, the pressure gradient

was measured three times. The effective permeability was calculated using Darcy’s law

based on the measured flow rates and pressure gradients.

Figure C1. Picture of the flow-through 2D heterogeneous cell experiments (A) 1/2-zone

cell; (B) 1/4-zone cell. The groundwater from blue tank was injected to the cells by white

peristaltic pump to pre-equilibrate with minerals inside the device for 6 residence times

before the MSW injection and was kept being injected after the stop of MSW injection.

The groundwater was continuously injected at 12.67 ml/hour. The cell has a dimension

151

of 40 cm×12 cm×1 cm installed with a 3D printed honeycomb consisting of 188

hexagonal cells to help the laminar flow.

Table C1. Effluent saturation index before and after MSW injection

Minerals Before MSW

Injection

After MSW Injection

1/2-zone 1/4-zone Uniform

Trace metals: carbonates

MnCO3 -2.02 -2.16 -2.06 -2.37

ZnCO3 -5.23 -5.62 -5.73 -5.17

PbCO3 -1.98 -2.37 -2.09 -1.80

CuCO3 -2.94 -3.11 -2.86 -2.90

Trace Metals: hydroxide

Mn(OH)2 -6.55 -7.12 -6.84 -6.58

Zn(OH)2 -2.50 -3.33 -3.26 -2.12

Pb(OH)2 -6.96 -7.78 -7.32 -6.46

Cu(OH)2 -0.70 -1.31 -0.88 -0.35

Ba, Sr, Ca

BaCO3 -2.81 -3.15 -2.87 -2.66

SrCO3 -2.10 -2.39 -2.18 -1.94

CaCO3 -0.22 -0.61 -0.34 -0.18

BaSO4 -0.16 -0.22 -0.21 -0.59

SrSO4 -3.25 -3.26 -3.32 -3.67

CaSO4 -2.95 -3.05 -3.04 -3.49

152

Table C2. Time series of effluent geochemistry in the Uniform column (units: mg/L)

Time (hr) pH Br Cl SO4 Na Ca Mg K Ba Sr Mn Cu Zn Pb Cd

0.25 8.48 0.08 33.81 12.89 19.09 5.23 8.34 11.53 0.03 0.05 0.0003 0.0005 0.0025 0.00001 UDL

0.75 8.50 - - - - - - - - - - - - - UDL

1.25 8.49 - - - - - - - - - - - - - UDL

1.75 8.51 - - - - - - - - - - - - - UDL

2.25 8.49 0.07 33.45 12.61 18.59 5 7.97 11.58 0.04 0.05 0.0002 0.0003 0.0023 0.00002 UDL

2.75 8.52 - - - - - - - - - - - - - UDL

3.25 8.50 - - - - - - - - - - - - - UDL

3.75 8.45 - - - - - - - - - - - - - UDL

4.25 8.47 UDL

4.75 8.49 0.05 32.31 12.53 18.96 5.05 8.16 12.1 0.04 0.05 0.0002 0.0004 0.0024 0.00001 UDL

5.25 8.45 - - - - - - - - - - - - - UDL

5.75 8.45 - - - - - - - - - - - - - UDL

6.25 8.48 - - - - - - - - - - - - - UDL

6.75 8.50 - - - - - - - - - - - - - UDL

7.25 8.47 - - - - - - - - - - - - - UDL

7.75 8.51 0.18 32.31 12.56 19.35 4.99 8.2 11.97 0.04 0.05 0.0002 0.0003 0.0027 0.00003 UDL

8.25 8.48 - - - - - - - - - - - - - UDL

8.75 8.42 2.27 132.03 12.61 31.96 16.75 25.66 18.68 0.13 0.17 0.0057 0.0003 0.0247 0.0001 UDL

9.25 8.29 - - - - - - - - - - - - - UDL

9.75 8.14 10.06 503.38 12.98 55.8 67.56 98.54 35.71 0.54 0.71 0.0206 0.0005 0.0102 0.0001 UDL

10.25 8.07 UDL

10.75 8.00 18.89 933.86 12.95 86.98 120.6 178.29 47.03 0.99 1.25 0.0365 0.0015 0.0621 0.00056 UDL

11.25 8.00 UDL

11.75 8.00 25 1192.36 12.92 166.5 150.71 226.38 54.63 1.28 1.54 0.0492 0.0014 0.0633 0.00056 UDL

12.25 7.97 - - - - - - - - - - - - - UDL

12.75 7.97 - - - - - - - - - - - - - UDL

153

13.25 7.97 32.71 1534.43 12.94 276.34 165.68 253.84 61.04 1.41 1.61 0.0571 0.001 0.0594 0.00092 UDL

13.75 7.99 - - - - - - - - - - - - - UDL

14.25 8.00 - - - - - - - - - - - - - UDL

14.75 7.97 - - - - - - - - - - - - - UDL

15.25 7.97 38.39 1797.98 12.99 401.17 167.44 263.07 65.08 1.53 1.63 0.0248 0.0008 0.0248 0.00076 UDL

15.75 7.92 - - - - - - - - - - - - - UDL

16.25 7.96 40.55 1873.38 12.74 466.78 167.97 264.38 66.46 1.6 1.67 0.0264 0.0015 0.0452 0.00016 UDL

16.75 7.99 UDL

17.25 8.02 40.79 1895.99 12.31 505.61 163.37 255.47 66.78 1.62 1.69 0.0265 0.0008 0.033 0.00008 UDL

17.75 8.03 - - - - - - - - - - - - - UDL

18.25 8.08 - - - - - - - - - - - - - UDL

18.75 8.13 29.43 1389.85 12.35 421.47 111.71 175.02 55.42 1.05 1.12 0.0221 0.0028 0.0974 0.00076 UDL

19.25 8.16 - - - - - - - - - - - - - UDL

19.75 8.18 - - - - - - - - - - - - - UDL

20.25 8.24 24.34 1165.75 12.31 361.57 82.67 130.99 48.01 0.79 0.83 0.0162 0.0008 0.0532 0.0006 UDL

20.75 8.26 - - - - - - - - - - - - - UDL

21.25 8.30 - - - - - - - - - - - - - UDL

21.75 8.32 - - - - - - - - - - - - - UDL

22.25 8.33 - - - - - - - - - - - - - UDL

22.75 8.34 - - - - - - - - - - - - - UDL

23.25 8.38 16.31 810.25 12.95 275.2 55.18 87.95 39.44 0.52 0.55 0.0263 0.0006 0.0407 0.00032 UDL

23.75 8.38 - - - - - - - - - - - - - UDL

24.25 8.37 - - - - - - - - - - - - - UDL

24.75 8.42 - - - - - - - - - - - - - UDL

25.25 8.42 - - - - - - - - - - - - - UDL

25.75 8.38 - - - - - - - - - - - - - UDL

26.25 8.42 - - - - - - - - - - - - - UDL

26.75 8.41 - - - 219.51 44.9 71.63 33.83 0.43 0.47 0.0233 0.0004 0.0307 0.00005 UDL

154

27.25 8.43 - - - - - - - - - - - - - UDL

27.75 8.45 - - - - - - - - - - - - - UDL

28.25 8.46 - - - - - - - - - - - - - UDL

28.75 8.45 - - - - - - - - - - - - - UDL

29.25 8.46 - - - - - - - - - - - - - UDL

29.75 8.45 - - - - - - - - - - - - - UDL

30.25 8.45 - - - - - - - - - - - - - UDL

30.75 8.45 - - - - - - - - - - - - - UDL

31.25 8.45 - - - - - - - - - - - - - UDL

31.75 8.45 UDL

32.25 8.46 - - - 127.5 33.58 53.91 33.39 0.29 0.33 0.0178 0.001 0.0279 0.00005 UDL

32.75 8.47 - - - - - - - - - - - - - UDL

33.25 - - - - - - - - - - - - - - UDL

33.75 - - - - - - - - - - - - - - UDL

34.25 - - - - - - - - - - - - - - UDL

34.75 - 5.39 305.08 12.2 98.39 26.31 42.68 28.17 0.24 0.27 0.016 0.0004 0.0249 0.00003 UDL

35.25 - - - - - - - - - - - - - - UDL

35.75 - - - - - - - - - - - - - - UDL

36.25 8.46 - - - - - - - - - - - - - UDL

36.75 - - - - - - - - - - - - - - UDL

37.25 - - - - - - - - - - - - - - UDL

38.75 8.45 - - - - - - - - - - - - - UDL

39.25 - - - - - - - - - - - - - - UDL

39.75 - 4.11 210 12.96 77.44 20.37 34.89 25.72 0.18 0.2 0.0143 0.0006 0.0327 0.00001 UDL

40.25 - - - - - - - - - - - - - - UDL

40.75 - - - - - - - - - - - - - - UDL

41.25 - - - - - - - - - - - - - - UDL

41.75 - - - - - - - - - - - - - - UDL

155

42.25 8.50 - - - - - - - - - - - - - UDL

42.75 - - - - - - - - - - - - - - UDL

43.25 - - - - - - - - - - - - - - UDL

43.75 - - - - - - - - - - - - - - UDL

44.75 - 1.52 100.72 12.96 58.28 8.43 14.3 16.49 0.07 0.08 0.0059 0.0003 0.0051 0.00006 UDL

45.25 8.53 - - - - - - - - - - - - - UDL

45.75 8.52 - - - - - - - - - - - - - UDL

46.25 8.53 - - - - - - - - - - - - - UDL

46.75 8.50 - - - - - - - - - - - - - UDL

47.75 8.48 - - - - - - - - - - - - - UDL

49.75 - 0.44 48.59 12.87 39.04 3.73 5.36 12.23 0.03 0.04 0.0019 0.0003 0.0071 0.00008 UDL

50.75 8.48 - - - - - - - - - - - - - UDL

52.25 8.45 - - - - - - - - - - - - - UDL

53.25 8.54 - - - - - - - - - - - - - UDL

54.75 8.55 0.22 36.05 12.95 36.54 2.83 3.76 10.19 0.02 0.03 0.0013 0.0007 0.0081 0.0001 UDL

55.25 8.52 - - - - - - - - - - - - - UDL

58.75 - - - - - - - - - - - - - - UDL

59.75 - 0.08 33.81 12.96 35.88 2.5 2.88 9.06 0.02 0.02 0.0013 0.0002 0.0025 0.0001 UDL

60.25 - 0.08 33.45 12.89 34.82 2.37 2.79 8.61 0.02 0.02 0.0011 0.0003 0.0049 0.00005 UDL

63.75 8.52 - - - - - - - - - - - - - UDL

68.75 8.62 - - - - - - - - - - - - - UDL

69.25 8.52 - - - - - - - - - - - - - UDL

69.75 8.55 - - - 35.71 2.11 2.55 11.71 0.02 0.02 0.001 0.0003 0.0027 0.00003 UDL

72.75 8.55 - - - - - - - - - - - - - UDL

73.25 8.54 - - - - - - - - - - - - - UDL

73.75 8.55 - - - - - - - - - - - - - UDL

74.25 8.54 - - - - - - - - - - - - - UDL

74.75 8.54 - - - - - - - - - - - - - UDL

156

76.75 8.53 - - - - - - - - - - - - - UDL

79.25 - 0.08 33.79 12.57 42.12 2.5 3.1 8.86 0.02 0.02 - - - - UDL

79.75 - - - - 40 2.44 3.03 8.19 0.02 0.02 0.0006 0.0003 0.0019 0.0001 UDL

80.25 8.54 - - - - - - - - - - - - - UDL

84.75 - 0.08 33.1 12.62 42.92 2.53 3.17 9.13 0.01 0.02 - - - - UDL

89.75 8.52 - - - 40.2 2.37 2.92 7.97 0.02 0.02 0.0006 0.0002 0.0018 0.00005 UDL

94.75 - 0.08 33.65 12.24 44.12 2.38 2.93 8.71 0.02 0.02 - - - - UDL

99.75 8.55 - - - 42.8 2.3 2.77 19.4 0.02 0.02 0.0001 0.0003 0.0024 0.00003 UDL

104.75 - 0.08 33.79 12.36 43.73 2.3 2.77 8.87 0.01 0.02 - - - - UDL

109.75 - 0.08 33.1 12.46 41.81 2.29 2.76 9.82 0.02 0.02 - - - - UDL

114.75 - - - - 39.4 2.51 2.64 12.1 0.02 0.02 0.0005 0.0002 0.0025 0.00007 UDL

119.75 - 0.08 33.65 12.89 38.49 2.71 2.95 14.66 0.02 0.02 - - - - UDL

120.75 8.54 - - - - - - - - - - - - - UDL

124.75 - - - - - - - - - - - - - - UDL

126.25 8.57 - - - - - - - - - - - - - UDL

129.75 - - - - 35 3.15 3.2 13.8 0.03 0.03 0.0008 0.0001 0.0022 0.00004 UDL

134.75 - 0.08 33.63 12.7 35.95 3.94 3.75 14.18 0.02 0.04 - - - - UDL

139.75 - 0.08 33.53 12.31 34.11 4.48 4.04 13.93 0.03 0.04 - - - - UDL

141.25 8.57 - - - - - - - - - - - - - UDL

144.75 - - - - 34.3 4.58 4.14 12.3 0.03 0.04 - - - - UDL

145.25 8.57 - - - - - - - - - - - - - UDL

146.25 8.52 - - - - - - - - - - - - - UDL

146.75 8.51 - - - - - - - - - - - - - UDL

147.25 8.52 - - - - - - - - - - - - - UDL

149.75 - 0.08 32.94 12.49 33.64 4.96 4.43 13.21 0.04 0.05 - - - - UDL

159.75 - 0.08 33.48 12.5 32.34 5.22 4.66 11.75 0.04 0.05 0.0001 0.0007 0.0027 0.00008 UDL

166.75 8.52 UDL

174.75 - - - - 30.88 5.5 4.85 11.31 0.04 0.06 0.0002 0.0015 0.003 0.00008 UDL

157

176.75 8.55 - - - - - - - - - - - - - UDL

189.75 - 0.08 33.11 12.23 30.69 6.1 5.3 11.74 0.04 0.06 0.0001 0.001 0.0024 0.00009 UDL

198.75 8.56 - - - - - - - - - - - - - UDL

199.25 8.52 - - - - - - - - - - - - - UDL

199.75 8.53 - - - - - - - - - - - - - UDL

200.25 8.54 - - - - - - - - - - - - - UDL

204.75 - 0.08 33.17 12.42 29.4 6.29 5.41 11.51 0.05 0.06 0.0002 0.0008 0.0024 0.00006 UDL

210.25 8.52 - - - - - - - - - - - - - UDL

219.75 - 0.08 33.31 12.26 28.36 6.48 5.55 11.58 0.05 0.07 0.0003 0.0007 0.0028 0.00006 UDL

227.25 8.50 - - - - - - - - - - - - - UDL

243.75 8.56 - - - - - - - - - - - - - UDL

249.75 - 0.08 33.7 12.51 29.07 7.64 6.43 13.08 0.05 0.08 0.0001 0.0006 0.0022 0.00008 UDL

259.75 8.54 - - - - - - - - - - - - - UDL

262.25 - 0.08 33.61 12.43 25.68 7.17 6.03 11.69 0.06 0.08 - - - - UDL

274.75 - 0.08 33.32 12.47 26.08 7.78 6.52 12.47 0.1 0.09 0.0001 0.0007 0.0029 0.00006 UDL

276.25 8.53 - - - - - - - - - - - - - UDL

287.25 - 0.08 33.14 12.68 24.95 7.64 6.33 12.17 0.14 0.1 - - - - UDL

292.25 8.52 - - - - - - - - - - - - - UDL

299.75 - 0.08 33.84 12.6 24.58 7.64 6.31 12.11 0.18 0.1 0.0002 0.0006 0.002 0.00008 UDL

309.25 8.50 - - - - - - - - - - - - - UDL

312.25 - 0.08 33.1 12.57 23.81 7.62 6.23 12.08 0.24 0.11 0.0001 0.0007 0.0023 0.00009 UDL

325.75 8.52 - - - - - - - - - - - - - UDL

324.75 - 0.08 33.8 12.49 22.56 7.92 6.4 11.92 0.3 0.13 0.0001 0.0009 0.0027 0.00006 UDL

337.25 - 0.08 33.6 12.53 22.8 8.04 6.44 12.06 0.37 0.15 - - - - UDL

349.75 - 0.08 33.29 12.55 22.33 8.17 6.48 12.1 0.41 0.16 - - - - UDL

362.25 - 0.08 33.33 12.45 21.97 8.44 6.57 12.24 0.55 0.22 - - - - UDL

374.75 - 0.08 33.34 12.65 21.68 8.49 6.51 11.99 0.62 0.25 - - - - UDL

387.25 - 0.08 33.27 12.44 21.32 8.85 6.65 12.3 0.73 0.31 - - - - UDL

158

399.75 - 0.08 33.5 12.41 21.28 8.92 6.66 11.99 0.78 0.35 - - - - UDL

412.25 - 0.08 33.42 12.47 22.12 9.16 6.79 11.37 0.85 0.4 - - - - UDL

424.75 - 0.08 33.18 12.45 20.6 8.92 6.53 10.65 0.85 0.39 - - - - UDL


Anions (Br, Cl, SO4), cations (Na, Ca, Mg, K, Ba and Sr), and trace metals (Mn, Zn, Cu, and Pb) were measured using different instruments and therefore have

different significant numbers because of different analysis approach. Hyphen (-) indicates we did not measure it. “UDL” means under detection limit.

159

Table C3. Time series of effluent geochemistry in the 1/4-zone heterogeneous cell (units: mg/L)

Time pH Br Cl SO4 Na Ca Mg K Ba Sr Mn Cu Zn Pb Cd

0.375 7.94 0.06 33.66 12.08 19.54 13.71 5.13 11.90 0.10 0.13 0.0106 0.0023 0.0045 0.00004 0.00002

1.125 7.91 - - - - - - - - - - - - - -

2.625 7.93 - - - 20.03 13.62 5.19 11.63 0.09 0.13 0.0109 0.0024 0.0048 0.00003 0.00003

4.875 7.91 - - - - - - - - - - - - - -

6.375 - 0.06 33.60 - 19.89 13.66 5.13 11.97 0.12 0.13 0.0109 0.0019 0.0050 0.00005 0.00003

7.125 7.91 - - - - - - - - - - - - - -

7.875 7.93 0.06 33.39 12.08 30.22 13.76 5.17 11.56 0.12 0.13 0.0109 0.0016 0.0047 0.00026 0.00002

8.625 7.91 - - - - - - - - - - - - - -

9.375 7.88 0.10 38.22 12.10 52.97 14.45 5.35 12.26 0.12 0.53 0.0110 0.0072 0.0240 0.00018 0.00008

10.125 7.88 - - - - - - - - - - - - - -

10.875 7.9 0.77 99.48 11.76 97.38 26.49 8.07 12.94 0.12 0.97 0.0113 0.0067 0.0238 0.00021 0.00016

11.625 7.85 4.30 422.58 11.59 123.17 84.28 25.25 19.83 0.57 1.16 0.0123 0.0042 0.0164 0.00036 0.00018

12.375 7.67 13.96 1305.95 10.95 426.57 229.60 58.82 36.18 2.47 7.94 0.1274 0.0083 0.0749 0.00265 0.00039

13.125 7.55 - - - - 348.15 63.90 45.38 30.56 - 0.1765 - - - -

13.875 7.58 27.45 2381.14 12.23 - 386.97 64.00 44.95 75.21 64.72 0.1841 0.0162 0.1291 0.00406 0.00125

14.625 7.58 - - - - 391.64 64.10 43.86 81.51 - 0.2469 - - - -

15.375 7.55 28.50 2430.57 12.11 922.13 392.82 65.66 45.53 90.15 69.37 0.4071 0.0371 0.1891 0.00094 0.00094

16.125 7.58 - - - - - - - - - - - - - -

16.875 7.57 29.00 2500.67 12.01 922.30 392.86 67.03 46.88 95.06 69.19 0.7142 0.0446 0.2075 0.00125 0.00187

18.375 7.57 23.99 2110.93 11.66 774.81 301.88 52.96 40.25 81.28 58.80 0.7131 0.0351 0.1906 0.00078 0.00276

19.125 7.61 - - - 759.50 290.74 51.23 41.07 76.78 53.88 0.6942 0.0365 0.2192 0.00125 0.00234

19.875 7.6 - - - 656.73 252.68 46.36 37.30 65.90 46.72 0.5894 0.0443 0.2244 0.00169 0.00247

20.625 7.58 12.49 1171.39 12.21 506.80 198.59 38.82 33.30 50.21 35.93 0.4659 0.0283 0.1200 0.00052 0.00208

21.375 7.63 - - 9.84 400.18 155.50 34.24 31.76 38.64 25.86 0.3743 0.0161 0.0812 0.00031 0.00109

22.125 7.66 - - 12.06 316.08 118.49 29.08 29.27 26.70 20.44 0.2901 0.0140 0.0756 0.00057 0.00073

22.875 7.7 7.16 683.97 8.30 260.14 97.29 26.35 26.74 21.60 16.50 0.2410 0.0146 0.0758 0.00026 0.00052

160

23.625 7.73 - - - - - - - - - - - - - -

24.375 7.74 - - - - - - - - - - - - - -

25.125 7.76 3.39 428.50 8.84 150.56 57.85 19.20 28.82 11.78 9.16 0.1569 0.0101 0.0483 0.00021 0.00021

25.875 7.77 - - - - - - - - - - - - -

27.375 - 1.78 294.95 9.18 108.85 40.33 16.34 24.06 7.69 5.89 0.1151 0.0051 0.0295 0.00016 0.00014

28.125 7.79 - - - - - - - - - - - - - -

28.875 - - - - - 36.17 15.73 15.52 6.00 - 0.1071 - - - -

29.625 - 1.19 230.80 9.22 - - - - - - - - - - -

30.375 7.82 - - - - 32.67 15.16 15.06 5.06 - 0.1128 - - - -

32.625 7.81 - - - - 29.76 14.62 18.65 4.47 3.66 0.0910 0.0040 0.0239 0.00010 0.00008

34.875 7.82 - - - - 27.70 14.00 14.89 4.08 - 0.0852 - - - -

35.625 7.82 0.84 164.34 9.34 - 26.02 13.62 14.82 3.77 - 0.0833 - - - -

37.875 7.83 53.13 24.09 12.68 14.11 3.69 2.82 0.0708 0.0028 0.0092 0.00015 0.00011

38.625 7.81 0.70 137.64 9.41 - - - - - - - - - - -

39.375 7.83 - - - - - - - - - - - - - -

40.125 7.84 - - - - - - - - - - - - - -

40.875 7.85 - - - 48.00 21.29 11.70 14.63 2.96 1.08 0.0667 0.0027 0.0091 0.00005 0.00006

41.625 7.85 - - - - - - - - - - - - - -

42.375 7.84 - - - - - - - - - - - - - -

43.125 7.82 0.62 118.68 9.65 45.50 20.76 12.03 14.77 2.75 0.93 0.0649 0.0015 0.0088 0.00003 0.00002

43.875 - - - 9.72 44.43 20.30 11.17 14.67 2.66 1.84 0.0635 0.0024 0.0066 0.00004 0.00007

44.625 7.84 - - - - - - - - - - - - - -

45.375 7.85 - - 9.27 42.82 19.72 11.45 14.13 2.54 1.69 0.0620 0.0024 0.0057 0.00004 0.00007

46.125 - - - 9.42 41.94 19.45 10.91 13.19 2.35 1.54 0.0611 0.0023 0.0061 0.00004 0.00005

47.625 - 0.50 108.61 12.03 41.95 19.36 10.80 13.24 2.33 1.46 0.0610 0.0031 0.0062 0.00005 0.00007

48.375 7.85 - - - - - - - - - - - - - -

52.125 7.88 - - - - - - - - - - - - - -

55.125 - 0.39 91.89 12.14 36.34 17.24 9.59 11.78 1.67 0.98 0.0530 0.0020 0.0055 0.00004 0.00005

161

58.875 7.89 0.34 86.26 12.19 - - - - - - - - - - -

59.625 7.89 - - - - - - - - - - - - - -

60.375 7.9 - - - - - - - - - - - - - -

62.625 - 0.29 79.00 12.18 33.13 16.44 8.81 11.14 1.21 0.69 0.0486 0.0021 0.0057 0.00005 0.00005

64.125 7.9 - - - - - - - - - - - - - -

66.375 - 0.27 74.29 12.23 - - - - -

67.125 7.89 - - - - - - - - - - - - - -

70.125 - - - - 33.95 15.58 7.80 11.32 0.95 0.54 0.0445 0.0020 0.0099 0.00004 0.00004

73.875 - 0.22 65.95 11.49 - - - - -

77.625 - 0.20 61.53 11.68 30.12 14.80 6.92 10.51 0.75 0.44 0.0390 0.0012 0.0084 0.00009 0.00008

85.125 - 0.17 55.88 11.67 28.93 14.33 6.03 11.49 0.58 0.37 0.0335 0.0019 0.0060 0.00004 0.00004

85.875 7.9 - - - - - - - - - - - - - -

92.625 - 0.15 49.93 11.67 27.25 13.76 5.63 9.20 0.51 0.33 0.0296 0.0020 0.0088 0.00011 0.00003

94.125 7.92 - - - - - - - - - - - - - -

97.125 7.91 - - - - - - - - - - - - - -

100.125 - 0.14 45.68 - 27.32 13.45 5.33 8.76 0.45 0.31 0.0257 0.0021 0.0053 0.00003 0.00004

107.625 - 0.12 42.25 - 25.41 13.30 4.98 8.32 0.38 0.27 0.0235 0.0015 0.0027 0.00002 0.00004

115.125 - 0.11 39.60 11.94 25.12 12.96 4.65 8.02 0.35 0.25 0.0191 0.0015 0.0041 0.00003 0.00004

122.625 - 0.09 37.72 - 25.82 13.46 4.73 8.15 0.32 0.24 0.0207 0.0016 0.0036 0.00004 0.00004

130.125 - 0.09 36.23 11.78 25.11 13.28 4.54 7.97 0.30 0.24 0.0186 0.0016 0.0031 0.00022 0.00003

137.625 - 0.08 35.20 12.35 24.65 13.15 4.44 7.92 0.26 0.23 0.0177 0.0014 0.0032 0.00003 0.00003

152.625 - 0.07 34.90 12.06 24.95 12.97 4.35 7.89 0.23 0.21 0.0159 0.0013 0.0027 0.00002 0.00004

160.125 - 0.07 34.76 - 22.13 12.92 4.53 8.17 0.22 - 0.0111 - - - -

167.625 - - - - 23.45 12.99 4.38 8.04 0.23 0.21 0.0138 0.0015 0.0027 0.00021 0.00004

175.125 - - - - 23.02 13.14 4.44 8.11 0.21 0.20 0.0101 0.0015 0.0028 0.00003 0.00003

190.125 - - - - 23.07 13.12 4.55 8.24 0.19 0.20 0.0122 0.0015 0.0023 0.00003 0.00005

201.375 - 0.06 33.39 12.23 23.02 13.68 4.73 8.18 0.18 0.18 0.0101 - - - -

205.125 - - - - - - - - - - - 0.0016 0.0029 0.00002 0.00003

162

216.375 - 0.06 33.37 12.25 23.22 13.53 4.71 8.21 0.17 0.18 0.0122 0.0024 0.0059 0.00001 0.00007

231.375 - 0.06 33.30 12.19 23.10 13.52 4.73 8.50 0.17 0.18 0.0101 0.0027 0.0065 0.00002 0.00003

246.375 - 0.06 33.43 12.19 22.60 13.38 4.86 8.58 0.16 0.18 0.0122 0.0025 0.0035 0.00006 0.00004

261.375 - 0.06 33.61 12.19 20.96 13.38 5.01 8.72 0.15 0.18 0.0106 0.0025 0.0026 0.00007 0.00003

276.375 - 0.06 33.60 12.22 21.40 12.73 4.89 8.48 0.13 0.17 0.0109 0.0025 0.0042 0.00003 0.00004

291.375 - 0.06 34.10 12.23 23.00 14.22 5.03 8.71 0.13 0.16 0.0109 0.0025 0.0034 0.00004 0.00001

306.375 - 0.06 34.16 12.09 23.19 14.07 5.07 8.51 0.13 0.16 0.0109 0.0046 0.0087 0.00001 0.00003

313.875 - 0.06 34.39 11.99 22.79 14.09 5.15 8.45 0.13 0.16 0.0101 0.0024 0.0038 0.00009 0.00002

321.375 - 0.06 34.70 12.09 22.07 13.69 5.13 8.59 0.12 0.16 0.0122 0.0017 0.0019 0.00003 0.00001

322.875 - - - 12.14 - - - - - - - - - - -

330.875 - - - - 21.80 14.06 5.07 8.58 - - 0.0106 0.0024 0.0036 0.00004 0.00003

333.875 - - - 12.23 - - - - - - - - - - -

346.125 - - - 12.25 - - - - - - - - - - -

347.375 - - - - 21.81 14.08 5.15 8.52 - - 0.0109 0.0025 0.0033 0.00003 0.00003

358.625 - - - 12.23 - - - - - - - - - - -

363.625 - - - - 21.79 14.06 4.97 8.46 - - 0.0109 0.0024 0.0036 0.00003 0.00002

370.875 - - - 12.25 - - - - - - - - - - -

380.125 - - - - 21.78 14.07 5.07 8.60 - - 0.0109 0.0024 0.0033 0.00004 0.00002

383.125 - - - 12.23 - - - - - - - - - - -

395.375 - - - 12.25 - - - - - - 0.0101 0.0025 0.0035 0.00004 0.00002

396.625 - - - - 21.80 14.05 5.15 8.50 - - - - - - -

407.875 - - - 12.23 - - - - - - - - - - -

413.125 - - - - 21.81 14.09 5.13 8.44 - - 0.0122 0.0024 0.0033 0.00004 0.00003

420.125 - - - 12.25 - - - - - - - - - - -




163

Table C4. Time series of effluent geochemistry in the 1/2-zone heterogeneous cell (units: mg/L)

Time pH Br Cl SO4 Na Ca Mg K Ba Sr Mn Cu Zn Pb Cd

0.375 7.76 0.05 31.65 12.80 20.37 12.47 5.46 6.90 0.12 0.11 0.0239 0.0025 0.0055 0.00003 0.00004

1.125 7.73 - - - - - - - - - - - - - -

1.875 7.74 - - - - - - - - - - - - - -

2.625 7.75 - - - - - - - - - - - - - -

3.375 7.71 - - - - - - - - - - - - - -

4.125 7.75 - - - - - - - - - - - - - -

4.875 7.74 - - - - - - - - - - - - - -

5.625 7.77 - - - - - - - - - - - - - -

6.375 7.73 0.05 31.40 12.00 20.42 12.48 5.43 6.23 0.11 0.11 0.0235 0.0025 0.0086 0.00002 0.00004

7.125 7.74 - - - - - - - - - - - - - -

7.875 7.75 0.16 40.62 12.12 20.80 12.54 5.43 6.48 0.10 0.11 0.0221 0.0023 0.0091 0.00001 0.00003

8.625 7.73 - - - - - - - - - - - - - -

9.375 7.63 0.24 47.66 11.53 24.90 17.89 6.50 11.08 0.10 0.15 0.0229 0.0029 0.0104 0.00002 0.00005

10.125 7.6 - - - - - - - - - - - - - -

10.875 7.57 2.64 250.27 12.29 88.64 51.39 14.17 21.42 0.05 0.44 0.0775 0.0074 0.0708 0.00010 0.00005

11.625 7.54 5.38 481.60 11.72 191.53 89.63 25.60 29.78 0.32 1.73 0.1264 0.0111 0.0641 0.00022 0.00017

12.375 7.51 12.43 1076.29 11.81 398.46 182.52 36.99 33.05 5.92 16.30 0.2247 0.0508 0.3228 0.00040 0.00040

13.125 7.42 - - - - - - - - - - - - - -

13.875 7.36 22.53 1928.64 9.81 729.60 298.41 47.18 37.44 61.97 50.06 0.3923 0.0311 0.1942 0.00150 0.00025

14.625 7.32 - - - - - - - - - - - - - -

15.375 7.29 27.28 2329.68 10.70 896.35 363.70 53.67 40.19 85.48 61.88 0.8442 0.0369 0.2034 0.00390 0.00150

16.125 7.28 - - - - - - - - - - - - - -

16.875 7.27 26.61 2273.00 11.08 888.34 351.34 49.79 36.24 84.25 60.36 0.9090 0.0462 0.2565 0.00200 0.00225

17.625 7.28 - - - - - - - - - - - - - -

18.375 7.31 20.74 1777.81 11.07 691.37 262.41 42.03 26.60 66.04 47.90 0.7519 0.0344 0.1869 0.00180 0.00200

19.125 7.35 - - - - - - - - - - - - - -

164

19.875 7.42 - - - - - - - - - - - - - -

20.625 7.42 13.57 1172.60 10.08 471.92 173.13 36.44 30.65 35.59 28.73 0.5458 0.0196 0.1213 0.00150 0.00128

21.375 7.43 - - - - - - - - - - - - - -

22.125 7.49 - - - - - - - - - - - - - -

22.875 7.48 10.46 910.38 9.63 351.92 138.00 28.89 27.92 26.60 22.80 0.4513 0.0155 0.0730 0.00137 0.00100

23.625 7.43 - - - - - - - - - - - - - -

24.375 7.45 - - - - - - - - - - - - - -

25.125 7.43 5.76 513.20 10.53 194.12 78.22 21.10 28.75 14.58 12.52 0.3181 0.0122 0.0693 0.00066 0.00048

25.875 7.44 - - - - - - - - - - - - - -

26.625 7.45 - - - - - - - - - - - - - -

27.375 7.47 3.05 284.85 10.58 103.18 43.37 18.52 24.42 8.59 6.23 0.1942 0.0072 0.0299 0.00035 0.00025

28.125 7.49 - - - - - - - - - - - - - -

28.875 7.49 - - - - - - - - - - - - - -

29.625 - 2.33 224.15 10.90 - - - - - - - - - - -

30.375 7.5 - - - - - - - - - - - - - -

31.125 7.5 - - - - - - - - - - - - - -

31.875 - 1.95 192.05 11.05 63.66 31.20 21.07 14.30 5.07 3.49 0.1620 0.0056 0.0245 0.00010 0.00012

33.375 7.56 -

34.125 7.53 1.93 190.62 11.37 - - - - - - - - - - -

36.375 7.54 2.06 200.90 12.75 56.05 30.71 28.48 17.93 4.00 2.46 0.1758 0.0060 0.0795 0.00010 0.00086

38.625 7.52 2.25 217.00 12.93 - - - - - - - - - - -

39.375 7.56 - - - - - - - - - - - - - -

40.125 7.58 - - - - - - - - - - - - - -

40.875 7.56 2.35 225.39 13.13 58.08 31.24 34.43 19.36 3.08 1.66 0.1746 0.0050 0.0836 0.00028 0.00102

43.125 7.51 2.29 220.97 13.27 61.07 29.72 34.41 19.26 2.65 1.37 0.1627 0.0050 0.0734 0.00005 0.00088

47.625 - 1.25 132.59 11.72 50.60 19.89 19.44 14.38 1.60 0.82 0.1014 0.0050 0.0549 0.00016 0.00078

51.375 7.55 1.16 125.59 12.90 - - - - - - - - - - -

52.125 7.55 - - - - - - - - - - - - - -

165

52.875 7.6 - - - - - - - - - - - - - -

55.125 - 0.86 99.70 12.94 44.45 16.90 13.43 11.56 1.15 0.52 0.0828 0.0050 0.0478 0.00015 0.00057

58.875 - 0.70 86.08 12.22 - - - - - - - - - - -

62.625 - 0.56 74.60 12.24 37.66 14.94 9.01 11.97 0.75 0.35 0.0632 0.0044 0.0133 0.00001 0.00006

64.125 7.63 0.56 74.32 12.22 - - - - - - - - - - -

66.375 7.68 0.45 65.50 12.32 - - - - - - - - - - -

70.125 - 0.37 58.73 12.26 34.04 13.19 6.72 10.27 0.54 0.26 0.0501 0.0035 0.0088 0.00003 0.00004

73.875 7.69 0.31 53.29 12.27 - - - - - - - - - - -

77.625 - 0.27 49.98 12.60 32.14 12.33 5.48 9.13 0.41 0.22 0.0390 0.0033 0.0103 0.00002 0.00003

84.875 7.7 - - - - - - - - - - - - - -

85.125 - 0.19 42.37 12.41 29.71 12.47 4.87 8.25 0.31 0.20 0.0321 0.0028 0.0092 0.00002 0.00003

91.625 7.73 - - - - - - - - - - - - - -

92.625 - 0.17 41.44 12.39 - - - - - - - - - - -

96.375 7.71 - - - - - - - - - - - - - -

100.125 - 0.15 39.69 12.24 28.80 12.08 4.37 7.78 0.25 0.18 0.0278 0.0029 0.0075 0.00002 0.00004

101.375 7.72 - - - - - - - - - - - - - -

107.625 - 0.12 37.39 12.95 - - - - - - - - - - -

115.125 - 0.10 35.87 12.90 27.90 12.06 4.08 7.42 0.21 0.16 0.0233 0.0044 0.0064 0.00003 0.00003

121.875 - - - 12.86 - - - - - - - - - - -

122.625 - - - - 27.34 11.95 4.00 7.49 0.21 0.16 0.0227 0.0022 0.0063 0.00002 0.00002

123.375 - 0.09 35.07 - - - - - - - - - - - -

137.625 - - - - 25.60 12.32 4.14 6.90 0.20 0.16 0.0231 0.0019 0.0062 0.00008 0.00003

138.375 - 0.07 34.08 - - - - - - - - - - - -

152.625 - - - - 24.81 13.12 4.08 6.23 0.21 0.17 0.0242 0.0017 0.0077 0.00001 0.00001

153.375 - 0.07 33.20 - - - - - - - - - - - -

160.125 - - - - 24.47 13.41 3.81 6.48 0.20 0.16 0.0245 0.0019 0.0095 0.00009 0.00002

160.875 - 0.07 32.90 - - - - - - - - - - - -

168.375 - 0.06 32.60 - - - - - - - - - - - -

166

183.375 - 0.05 32.00 - - - - - - - - - - - -

190.125 - 0.05 31.70 12.86 25.29 13.52 3.86 6.39 0.15 0.14 0.0236 0.0045 0.0069 0.00007 0.00003

198.375 - 0.05 31.60 - - - - - - - - - - - -

190.875 - 0.05 31.73 - - - - - - - - - - - -

205.125 - 0.05 31.20 12.81 25.85 13.22 3.88 6.94 0.14 0.14 0.0233 0.0056 0.0058 0.00005 0.00005

220.125 - 0.05 31.01 12.83 24.40 12.73 3.96 7.03 0.13 0.13 0.0227 0.0041 0.0070 0.00005 0.00004

235.125 - 0.05 31.99 12.72 25.59 12.95 4.40 7.29 0.13 0.13 - - - - -

236.625 - - - - - - - - - - 0.0242 0.0034 0.0086 0.00005 0.00003

250.125 - 0.05 31.65 12.84 25.13 12.74 4.62 7.25 0.12 0.13 0.0233 0.0034 0.0060 0.00005 0.00003

265.125 - 0.05 31.94 12.89 25.12 13.04 4.91 7.33 0.12 0.13 0.0227 0.0038 0.0069 0.00005 0.00002

280.125 - 0.05 31.48 12.82 24.19 13.10 4.59 7.33 0.11 0.14 0.0231 0.0034 0.0058 0.00005 0.00003

298.625 - - - 12.83 23.08 12.72 5.06 7.25 - - 0.0241 0.0033 0.0070 0.00005 0.00004

310.875 - - - 12.88 - - - - - - - - - - -

313.625 - - - - 22.46 12.94 5.30 7.33 - - 0.0227 0.0033 0.0069 0.00004 0.00004

330.125 - - - - 21.21 12.73 5.37 7.33 - - - - - - -

335.625 - - - 12.30 - - - - - - - - - - -

337.875 - - - - - - - - - - 0.0241 0.0037 0.0058 - -

346.875 - - - - 21.46 13.12 5.43 7.25 - - - - - - -

348.125 - - - 12.83 - - - - - - - - - - -

354.625 - - - - - - - - - - 0.0231 0.0035 0.0070 - 0.00005

360.625 - - - 12.87 - - - - - - - - - - -

362.875 - - - - 20.84 13.09 5.41 7.33 - - - - - - -

370.875 - - - - - - - - - - 0.0226 0.0033 0.0069 0.00005 0.00004

373.125 - - - 12.83 - - - - - - - - - - -

379.375 - - - - 20.71 12.73 5.45 7.33 - - - - - - -

385.375 - - - 12.85 - - - - - - - - - - -

387.375 - - - - - - - - - - 0.0243 0.0035 0.0058 - 0.00004

395.875 - - - - 20.59 12.95 5.41 7.25 - - - - - - -

167

397.875 - - - 12.88 - - - - - - - - - - -

403.875 - - - - - - - - - - 0.0233 0.0037 0.0058 0.00004 0.00004

410.375 - - - 12.83 - - - - - - - - - - -

412.375 - - - - 20.34 12.74 5.42 7.33 - - 0.0227 0.0033 0.0057 - -

422.625 - - - 12.90 - - - - - - - - - - -


Anions (Br, Cl, SO4), cations (Na, Ca, Mg, K, Ba and Sr), and trace metals (Mn, Zn, Cu, and Pb) were measured using different instruments and therefore have

different significant numbers because of different analysis approach. Hyphen (-) indicates we did not measure it.

168

Appendix D Permission to include published paper in the thesis

D1. Copyright information for chapter 2

Open Access

© The Author(s) 2016

This article is distributed and licensed under the terms of the Creative Commons

Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/),

which permits unrestricted use, distribution, and reproduction in any medium, provided

you give appropriate credit to the original author(s) and the source, provide a link to the

Creative Commons license, and indicate if changes were made. The Creative Commons

Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/)

applies to the data made available in this article, unless otherwise stated.

Title: How Long Do Natural Waters “Remember” Release Incidents of Marcellus Shale

Waters: a First Order Approximation Using Reactive Transport Modeling

Author: Zhang Cai; Li Li

Publication: Geochemical Transaction

Publisher: Springer International Publishing

I would like to… reuse in a thesis/dissertation

I would like to… use full article

My format is… both print and electronic

I am the author of this article… Yes

I will be translating… No

http://creativecommons.org/licenses/by/4.0/

http://creativecommons.org/publicdomain/zero/1.0/

169

D2. Copyright information for chapter 3

Title: Mineralogy controls on reactive transport of Marcellus Shale waters

Author: Zhang Cai; Hang Wen; Sridhar Komarneni; Li Li

Publication: Science of The Total Environment

Publisher: Elsevier

Date: 15 July 2018

© 2018 Elsevier B.V. All rights reserved.

I would like to… reuse in a thesis/dissertation

I would like to… use full article

My format is… both print and electronic

I am the author of this article… Yes

170

I will be translating… No

Curriculum Vitae

Zhang Cai

EDUCATION The Pennsylvania State University Petroleum and Natural Gas Engineering Ph.D. 2018

Nankai University (PRC) Environmental Science M.S. 2012

Nankai University (PRC) Environmental Engineering B.S. 2009

SELECTED PUBLICATIONS

1. Z Cai, H Wen, L Li. Controls of mineral spatial patterns on the reactive transport of Marcellus

Shale waters. Submitted to Energy & Fuels.

2. Z Cai, H Wen, S Komarneni, L Li. Mineralogy controls on reactive transport of Marcellus

Shale waters. Science of the Total Environment. 2018, 630, 1573-1582.

3. Z Cai, L Li. How long do natural waters “remember” release incidents of Marcellus Shale

waters: a first order approximation using reactive transport modeling. Geochemical

Transactions. 2016. 17 (1), 82-97.

4. Z Cai, Q Zhou, S Peng, K Li. Promoted biodegradation and microbiological effects of

petroleum hydrocarbons by Impatiens balsamina L. with strong endurance. Journal of

Hazardous Materials. 2010,183 (1-3), 731-737.

5. X Wang, Z Cai, Q Zhou, Z Zhang, C Chen. Bioelectrochemical stimulation of petroleum

hydrocarbon degradation in saline soil using U-tube microbial fuel cells. Biotechnology and

Bioengineering.2012, 109 (2), 426-433.

6. Q Zhou, Z Cai, etc. Ecological Remediation of Hydrocarbon Contaminated Soils with Weed

Plant. Journal of Resources and Ecology. 2011, 2 (2), 97-105.

7. S Peng, Q Zhou, Z Cai, etc. Phytoremediation of petroleum contaminated soils by Mirabilis

Jalapa L. in a greenhouse plot experiment. Journal of Hazardous Materials. 2009, 168 (2-3),

1490-1496.

8. Z Zhang, Q Zhou, S Peng, Z Cai. Remediation of petroleum contaminated soils by joint action

of Pharbitis nil L. and its microbial community. Science of the Total Environment. 2010. 408

(22), 5600-5605.

9. W Liu, Q Zhou, Z Zhang, T Hua, Z Cai. Evaluation of cadmium phytoremediation potential in

Chinese cabbage cultivars. Journal of agricultural and food chemistry. 2011, 59 (15), 8324-

8330.

10. C Chen, Q Zhou, Z Cai. Effect of soil HHCB on cadmium accumulation and phytotoxicity in

wheat seedlings. 2013. Ecotoxicology, 1-9.

11. C Chen, Z Cai. Physiological and Antioxidant Responses in Wheat (Triticum aestivum) to

HHCB in Soil. Bulletin of Environmental Contamination and Toxicology. 1-6.

12. X Peng, H Yu, X Wang, Q Zhou, S Zhang, L Geng, J Sun, Z Cai. Enhanced performance and

capacitance behavior of anode by rolling Fe3O4 into activated carbon in microbial fuel cells.

Bioresource Technology. 2012. 121, 450-453.

Date post:	18-Dec-2021
Category:	Documents
Upload:	others
View:	5 times
Download:	0 times

UNDERSTANDING REACTIVE TRANSPORT OF MARCELLUS SHALE …

Documents