PB95-191250
                                                          EPA No. 530-R-95-012
                                    University of Nevada

\5                                           Reno


1.

^AJ                          Geochemical Modeling of Mine Pit Water:
                        An Overview and Application of Computer Codes
                      A thesis submitted in partial  fulfillment  of  the


                    requirements for the degree of Master  of  Science  in

                                        Hydrogeology
                                             by




                                        David A.  Bird




                          Professor W. Berry Lyons,  Thesis  Advisor


                                        December 1993

-------
         1993
   David A. Bird
All Rights Reserved

-------
                                          ACKNOWLEDGEMENTS


                      This thesis is the second of a two part study funded by the United
                States Environmental Protection Agency.  Part one was completed by
                Margaret Saunders Macdonald,  who graduated from UNR in 1992 with a
                Master's Degree in Hydrogeology.  Much of the information in this thesis
                was taken from Meg's report,  including many of the citations pertaining
                to the physical and chemical  characteristics of pit lakes.
                      Thanks are in order for many people who contributed to the
                completion of this thesis.  If I have forgotten anybody,  I apologize, as
                the list is quite long:

                *  The members of my committee.  Professors W. Berry Lyons, Glenn C.
                   Miller, and Stephen H. Wheatcraft for their.invaluable assistance,
                   support,  and encouragement toward the completion of this thesis.
                *  My office mates,  Ann Carey,  Georgia Doyle, Kevin Johannesson,  Bill
                   Ludwick,  Phil Murphree, Bwire Ojiambo,  Eric Swanson, and Jim Thomas
                   for support and enlightening geochemical/hydrological  discussions.
                *  Kathy Sertic, Dave Jones,  and Doug Zimmerman at the Kevada Division
                   of Environmental Protection,  for their assistance,  and for allowing
                   access to files.
                *  Bill Upton of Placer Dome  for granting permission to visit the Cortez
                   Mine site and sample the pit water.   Mark List and Eric Vokt of
                   Cortez Gold Mines for their sincere efforts in providing data,  and
                   Eric Vokt for helping sample the Cortez Pit lake.
                *  Rab Bustos, Eric Seedorf,  and Ron Zuck of Magma Mining Co.  for
\^                 providing data and tours,  and allowing sampling of the Ruth District.
                *  Chuck Zimmerman of Newmont for allowing access to the  Universal Gas
                   Pit site,  and Pam Gilbert  for conducting the tour.
                *  Georgia Doyle for assistance in sample collection,  organization,  and
" )                 geocnemical insight.  Thanks also for playing office traffic cop
\)                 during rush hour.
                *  Anne Marie Harris for critical review of selected sections.
                *  Carl Palmer of Oregon Graduate Institute,  -and Andy Davis of PTI
                   Environmental .Services for Helpful discuss ions "regardincjpft "water _ ~
                   modeling.      .-'...              .   ...... .	    . .,,_... ._._
                *  Steve Wesnousky and the gang in Meotectonics for 'allowing .access  to
                   computers.
                *  Professor Gary Vinyard of  the UNR Biology Department,  for the  use of
                   lake sampling equipment.

                      I would like to thank my advisor,  Berry Lyons,  for  his motivation
                and leadership,  for encouraging me to tackle this project,  for his
                patience during my endless barrage of questions  regarding elementary
                aqueous geochemistry,  and for having more faith in  me  than I did.
                      A special note of thanks is reserved for Rhonda  for her support,
                guidance, and unbelievable patience through 2 years of graduate school.
                I probably could not have done it without her.
                      This thesis is dedicated to my mother,  whose  courage,  strength,
                and sense of humor in the face of personal tragedy  was a  source of
                inspiration,  and reminded me  of the importance of making  use of the
                short time we have.
V\
                                                  ii

-------
                               ABSTRACT
      The impacts of mining on water quality in the western United
States have become the focus of increased environmental concern and
regulatory effort in recent years.  An assessment of potential mining
impacts on local water quality is necessary because of possible adverse
effects including acid mine drainage, elevated trace metal
concentrations,  and high dissolved solids such as sulfate.  In Nevada
alone, at least 26 open pit mining operations now, or will in the
future, require dewatering to allow excavation below the water table.
Open pits now requiring dewatering will likely see the eventual
development of pit lakes after mine-closure.  The water quality that
evolves in pit lakes will be a function of many variables including, but
not limited to:   host lithology and buffering capacity; structure
(fractures, faults); type, mass, and morphology (stratiform, massive,
disseminated,  structurally controlled) of ore, alteration, and gangue;
groundwater temperature, flow rate and aquifer morphology (isotropy,
homogeneity);  pit geometry and size;  biological activity; and climate
(precipitation,  evaporation, wind velocity) .
      Although a wide variety of geochemical modeling software packages
are available, none are designed specifically for the purpose of
modeling pit lake geochemistry, and no regulatory framework or standard
exists for such modeling efforts.  This study was designed to evaluate
hydrogeochemical modeling software that might be applicable to modeling
post-mining, pit water geochemistry.   Data from the Cortez Mine, a
carbonate-hosted, open pit,  precious metal mine in Nevada, are used in
an inverse model to determine geochemical mass transfer that has
occurred between the mine wallrock and the pit lake.  These results
guide the development of a forward reaction path model that may be used
for future mine sites.
      For inverse modeling,  the geochemical mass transfer code BALANCE
(USGS) was used because of its ability to incorporate trace metal
phases.  The reaction path (forward)  model,  
-------
                          TABLE OF CONTENTS
SIGNATURE PAGE 	  i
ACKNOWLEDGEMENTS 	 ii
ABSTRACT 	-	 iii
TABLE OF CONTENTS 	 iv
LIST OF FIGURES 	 vii
LIST OF TABLES 	 viii

1.  INTRODUCTION	  1
       Purpose of Study	  5
       Scope of Study	  5
       Previous Work 	  7
       General Modeling Background  	  8
       Geochemical Modeling Applied to Mine Water Quality  	  9
       Data Disk	 12

2.  GEOCHEMICAL OVERVIEW 	 14
       Ionic Strength	 14
       Activity Coefficients 	 15
       Ionic Balance 	 25
       Mass Balance 	 27
          Mass Transfer 	 28
       Equilibrium Thermodynamics 	 29
          Saturation Index 	 30
          Reversible vs.  Irreversible Reactions	 32
          Incongruent Dissolution 	 32
          Solubility vs.  K,p  	 33
          Temperature/pressure Dependency 	 34
       Chemical Speciation 	 35
          Limitations 	 38
       Oxidation/Reduction (Redox)  	 39
       Geology	 40

3.  MODEL DEVELOPMENT AND APPLICATION	.	43
       Conceptualization'	.	 43
          Information Desired*	..'..".^ ...'.-.'.	 43
          Input Required 	 47
       The Numerical Model 	 50
          Development	 50
          Execution	 53
          Interpretation and Sensitivity Analyses	 54
          Calibration	 55
          Verification/Validation 	 56

4.  SOFTWARE	 58
       Basic Input 	 58
       Database Limitations	 58
       Speciation Modeling Codes	 59
          WATEQF and WATEQ4F 	 0
       Inverse Modeling Codes 	 62
          BALANCE	 63
          NBTPATH	r	 63
       Forward Modeling Codes 	 64
          MINTEQA2 	 64
             Adsorption Models  	 68
          PHREEQB	 69
             Limitations 	 73
          PHRQPITZ 	 74


                                   iv

-------
5.   PIT WATER MODELING CONSIDERATIONS  	  75
       Chemical Factors 	  75
          Classification of Deposit 	  75
          Wallrock Mineralogy 	  77
             Acid Mine Drainage 	  81
             Dissolved Solids 	  86
             Trace Elements 	  88
          Oxidation/Reduction (redox)   	  97
          Adsorption/coprecipitation 	  98
             MINTEQA2 Adsorption Models  	 104
                Non-Electrostatic Adsorption Models  	 104
                Electrostatic Adsorption Models	 105
          Groundwater/Aquifer Geochemistry  	 no
          Reaction Kinetics 	 Ill
          Equilibrium Thermodynamics 	 112
          Biological Activity 	 113
          Ion Exchange 	.	 116
       Physical Factors 	 117
          Evapoconcentration 	 117
          Limnology	 119
             Geothermal Input 	 121
          Atmospheric Gas Exchange 	 122
          Rock/Water Ratio	 123
          Number of Inputs/Outputs in System 	 125
          Time Scale 	 126
          Hydraulic Gradient 	 127
          Anthropogenic Disturbance 	 128
       Other Factors 	 128
          Database Limitations 	 128
          Downgradient Impacts 	.	 129

6.   PIT WATER MODELING APPROACHES 	 131
       Rate-independent Dissolution 	.	 131
       Rate-dependent Dissolution	 134
       Coupled	 135

7.   MODELING RESULTS 	137
       Speciation/Bquilibriura Models	 137
          Cortez Pit 	 138
          Universal Gas Pit 	 140
          Discussion	 143
       Inverse Model 	 147
          Input 	 147
             Hater Chemistry 	;	 147
             Phases 	 149
          Results 	 150
       Forward Models 	 153
          Mass Transfer 	 153
          Precipitation 	 156
          Calibration	 157
          Adsorption	160
          Second Iteration	 162
       Summary	 166
       Sensitivity Analyses 	'.	 167
          Pyrite Dissolution	 167
          Anoxia Progression 	 170

8.   CONCLUSIONS 	 172

9.   RECOMMENDATIONS 	 173

-------
10.   REFERENCES 	 176

APPENDIX A (Debye-Hvickel a and b parameters)  	 188
APPENDIX B (Cortez pit water mass transfer models
     calculated in BALANCE 	 189
                                    Vi

-------
                           LIST  OF FIGURES


Number                                                               Page

1-1.    Location map for Nevada open pit mines
           requiring dewatering  	  2
1-2.    Fence diagram of part of  the Atlantic Coastal Plain
           showing hydrochemical  facies 	  10
1-3.    Evolution of hydrogeochemical  codes 	  ll

2-1.    Activity coefficients vs. log  /I 	  20
2-2.    Variation of the activity coefficient for yCa** according
           to the three forms of  the Debye-Huckel equation	  21

3-1.    Scenario requiring inverse modeling methods  	  44
3-2.    Scenario requiring forward modeling methods  	  46
3 -3.    Actual vs. net reaction path 	  55

4-1.    A comparison of databases for  some geochemical
           speciation and mass transfer codes 	  59
4-2.    Flowchart diagramming the MINTEQA2 procedural loop  	66
4-3.    Comparison of PHREEQE and MINTEQA2 	  72

5-1.    Cross section of hypothetical  open pit 	  76
5-2.    General classification and nomenclature of common
           plutonic and volcan-ic rock  types 	 80
5-3.    Silica species activity vs.  pH 	 87
5-4 .    Aluminum solubility vs. pH 	 89
5-5.    Arsenic Eh/pH diagram 	 92
5-6.    Iron Eh/pH diagram	.	 94
5-7.    Contours of dissolved iron as  a function of pe and pH,
           assuming pCO,   10'a,  IS  10'a 	 95
5-8.    Adsorption behavior of cations 	-	 100
5-9.    Adsorption of lead.on alumina..--?>...., .;...,. .-...,...-.;.,..... .-101
5-10.   Adsorption behavior of an arsenic1 species v.-;;.; .1; ,.?;;*. v* .-102^
5-11.   Schematic representation, of  the', surface, charge/potential
           relationships: used in the constant- capacitance a^-;:fI
           diffuse-layer models	 100
5-12.   Schematic model of the triple  layer model 	 109
5-13.   Precipitation in Nevada 	 118
5rl4.   Evaporation in Nevada		 119
5-15.   Oxygen and Eh profiles at Berkeley Pit 	 120
5-16.   Seasonal changes in lake profiles 	 121

6-1.    Rate-independent dissolution model 	132
6-2.    Rate-dependent dissolution model	 136

7-1.    Location map for Cortez and  Carl in Mines 	138
7-2.    Evolution of pH as a function  of pyrite dissolved
           and host rock	 169
                                  Vll

-------
                            LIST OP  TABLES


Number                                                              Page

1-1.   Hater chemistry, Berkeley and  Liberty Pits  	 4
1-2.   Computer codes evaluated  	 6
1-3.   Water chemistry, Yerington pit 	 12
1-4.   Water chemistry, Cortez and Universal Gas pits  	 13

2-1.   Equations for activity coefficient  (y, )  	 22
2-2.   Comparison of activity coefficients
          modeled in PHREEQE vs. PHRQPITZ  	 23
2-3.   Comparison of the Debye-Huckel equation, Davies equation,
          and the MacXnnes Assumption	'.'	 24

-1.   Moles of element per kilogram  of  rock in Cortez pit
          wallrock, and concentrations (mmol/1) of dissolved
          solids in pit water  	 133

7-1.   Cortez Pit water chemistry  (original)  	 139
7-2.   Cortez Pit water chemistry  (new)  	 139
7-3.   Input concentrations for  Cortez Pit water
          chemical modeling simulations  	 140
7-4.   Water chemistry, Universal Gas pit  	 141
7-5.   file names and contents of speciation model output  files  ... 142
7-6.   Cortez pit water speciation, portion  of output file
          CZSP01W4.OUT showing saturation indices  	 142
7-7.   Universal Gas pit water speciation, portion of output
          file UGSP01W4.OUT showing saturation indices  	 144
7-8.   Comparison of portions of output  files for
          Cortez pit water speciation simulations  	 145
7-9..   Comparison of portions of output  files for
          Universal Gas pit water speciation simulations 	 146
7-10.  Chemical analyses for Well SC-5B, Cartin Trend, Nevada  ..;-.. 149
7-11.  Mineral and gas phases selected for Cortez  Pit water
          inverse model	150
7-12.  Mass transfer model calculated by BALANCE	 15- -
7-13.  Minerals used  in mass transfer reaction models  ...'.	 154
7-14.  Concentration of pit water after  PHREEQE
          mass transfer model  (CZRXOiPH.OOT)  	 155
7-15.  Results of Cortez pit water precipitation
          simulation  in MINTBQA2 (CZPR01MT.OUT)  	156
7-16.  Cortez pit water speciation, output file CZSP01W4.OUT	157
7-17.  Precipitation  calibrations  (MINTEQA2)  	 160
7-18.  Output file  showing equilibrium distribution  of Cortez  pit
          water after adsorption model (CZAD01MT.OUT)  	161
7-19.  Adsorption  Calibrations  (MINTEQA2)  		 162
7-20.  Results of  BALANCE  model  in second  iteration	 163
7-21.  Concentration of pit water  after  PHREEQB mass transfer
          model;  second iteration  (CZRX02PH.OUT)  	164
7-22.   Portion of  output file  showing equilibrium distribution
          of Cortez pit  water  after  adsorption model,
           second  iteration (CZAD02MT.OUT)  	 165
7-23.   Comparison of adsorption model (second iteration)
           and actual Cortez  pit water chemistry	166
7-24.   File names  and contents of  forward  model output  files	 167
 7-25.   Evolution of pR as  a function of
           pyrite  dissolved and host  rock	168
 7-26.   Simulated anoxia in Cortez  pit lake 	 170


                                  viii

-------
                          1.   INTRODUCTION

      In the past two decades, the number of open pit precious-metal
mines in the western United States has increased significantly.  The
economics and technology of today's mining industry allow the extraction
of ore from great depths by open pit methods, which has raised concerns
about potential impacts on local and regional groundwater systems.
Mines excavating below the level of the local water table require
removal of groundwater through dewatering in order to keep the mine area
dry.  Mine dewatering will affect the local hydrologic system by
creating a cone of depression in the piezometric surface, and steepening
the hydraulic gradient in the immediate area.  The deeper the mine, the
more dewatering required,  and the deeper and broader will be the cone of
depression.  On its proposed completion in 2001, the Gold Quarry pit in
Eureka County, Nevada is projected to be 460 meters deep, 270 meters
below the level of the regional water table (PTI,  1992).   The cone of
depression is projected to be 64 kilometers in diameter at its maximum
width (HCI, 1992).
      When a mine is decommissioned and dewatering ceases, the cone of
depression will start to recover, and the pit may begin to fill with
water.  Several decades may elapse before the regional groundwater table
returns to pre-mine conditions and the pit fills to its steady state
depth.  If allowed to fill at an undisturbed rate,  the Berkeley pit in
Butte, Montana, would require 27 years to reach maximum depth at the
level of the ambient water table (Davis and Ashenberg, 1989) .  The Gold
Quarry pit is predicted to reach 95% of the final level approximately 20
years after cessation of dewatering (HCI, 1992).
      In Nevada alone, at least 26 open pit mine sites are currently
water-filled, or have active and/or proposed dewatering operations
 (MacDonald, 1992).  Figure 1-1 shows the locations of these mines.
Although most pits that will ultimately contain standing water are still

-------
                                                    Big Springs
Steeper*      Twin Creeks  X Horizon Twamt
     GetchellXX        xJ"* GoUs(>ike
          Bullion MonarcnX*' Geneail
    Lone IreeX            XcoM Quarry

       Fortitude X    X Mule Canyon

  McCoj/CoveX       XGoId Acres
                      x
                                                            Bnckhorn
                                  Austin Gold Venture
                                                               Buth/iaerty
              Yeringtoa
                       FMC-Ietchup Flat
                                X
                  Aurora Partnership
                 k   X
              X Hound Mountain Gold


            X Cypress Tonopah

        XBoaslGae
Figure 1-1:   Location map for Nevada open pit mines requiring  dewatering
(from  Macdonald,  1992).

-------
being mined, seven pit lakes are known in Nevada (Macdonald, 1992):


       Name                   County                   Type of mine

      Boss                    Nye                     Precious Metal
      Cortez                  Lander                 , Precious Metal
      Liberty                 white Pine              Porphyry Copper
      Ruth                    White Pine              Porphyry Copper
      Tuscarora               Elko                    Precious Metal
      Universal Gas           Eureka                  Precious Metal
      Yerington               Lyon                    Porphyry Copper


      Large volumes of water are being pumped in dewatering processes,

with staggering projections.  The Gold Quarry mine is expected to

require a pumping rate of over 50 million gallons per day  (over 58,000

acre feet per year) in the year 2001  (HCI, 1992).

      In recent years, the mining industry has become the  target of

environmental concerns pertaining to water quality,  including the water

chemistry that will evolve in the pit lake as a  result of  rock-water

interaction between the inflowing groundwater and the pit  wall minerals.

As recently as 10 years ago, the only extent to  which mining companies

were required to address the issue of post-mining pit water quality was

to include  the following statement in the permit (Harris,  1992):


               Upon closure, the open pit will  infill
                  and become a permanent  lake.


      Until the enforcement of water  quality regulations  in the  last

four years, mining companies were only  required to monitor.the  effects

of their  operations on  local water quality after the inception  of

mining.   This has  resulted  in some serious environmental  consequences,

such as acid mine  drainage  and  contaminated pit lakes.

      The most serious  pit  water problem known  exists at  the  Berkeley

pit, a  current Superfund  site  (see Davis and Ashenberg,  1989; Baum  and

Knox,  1992).  Similar scenarios,^but  of smaller scale,  exist  at the Ruth

and  Liberty pits,  located in the Ruth District  in eastern Nevada.  The

water  chemistry  of the Berkeley and  Liberty pits is  shown in Table  1-1.

-------
                                                                       4
Sites such as these,  with very low pH and high trace metal
concentrations,  have stimulated reevaluation of the regulatory framework
surrounding the  permitting of new mining operations.
TABLE 1-1: Water chemistry, Berkeley and Liberty pits (ng/1) .

Alkalinity, bicarbonate
Chloride
Fluoride
TOS
Sulfate
Aluminum
Arsenic
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Lead
Silica
Zinc
PH
Source : *
*

Berkeley 
(100m depth)
0.0
20
NA
NA
7060
206
0.7
NA
1.9
506
NA
218
1040
NA
25
272
162
73
NA
NA
496
3.0
Davis and Ashenberg, 1989
UNR sampling (1993)
NDBP files
NA > Not available
Liberty 
(surface)
0.0
40.1
0.16
NA
3780
90
< 0.002
0.0
0.036 *
522
NA
51 
59
NA
s.oa
351
146 
53.3
0.0
48
67 
3.02


 rv- m *
      Regulatory agencies now expect greater detail in studies--'v*1*--  -
predicting the impacts of surface mining activities on surface and
groundwater resources.  Companies with proposed mining operations are
being asked to assess these impacts during the permitting process before
mining can proceed.  Such assessments now must also be provided during
permitting for expansions of existing operations.  Many assessments
incorporate detailed geochemical models to predict the long term
chemistry of pit lakes.
      Geochemists rely on a variety of modeling methods to predict the
impacts of mining on local and regional hydrologic systems.  Hydrogeo-
chemical equilibrium and reaction path models are used in conjunction
with limnological, and in some cases numerical flow modeling computer
codes, to predict the geochemistry that will result in the pit lakes.

-------
                                                                        5
Purpose of Study
      The primary objective of  this  work is to evaluate  the  suitability
of the hydrogeochemical computer modeling codes BALANCE, MINTEQA2,
PHREEQE, WATEQF, and WATEQ4P to the  task of modeling post-mining pit
water geochemistry.  The advantages  and  disadvantages of these codes are
discussed, and considered in regard  to their utility for pit water
modeling.  Detailed descriptions pertaining to the operation of each
software code are given in chapter 4.  These are directed towards
readers with limited experience using the codes,  and are intended to
summarize the important features of  each.   Chapter 2 contains a detailed
discussion of introductory aqueous geochemistry,  and how the concepts
are integrated into chemical models.  This chapter is intended for
readers with limited aqueous geochemistry background.
      The underlying questions  that  the  thesis addresses are:  Can post-
mine pit water be predicted using the hydrogeochemical codes listed
above, how many of the variables can be  integrated into the model,  and
what level of accuracy and validity  can  be expected in the results?  The
                              - -. : ^  ... _ -  .   :. ^K ^ =;=_-,   -  ._  .r-r^T-.-r =. z. r-^  - - r- :
study attempts to demonstrate that pit water chemical modeling,  with an
understanding of the variables, can  be accomplished with these software
packages.

Scop* of Study
      As the focus of the study was  on computer modeling, minimal field
work was performed.   All computer programs used in the study are
available from the respective author and/or  federal agency where they
were developed.  Most aqueous geochemical, lithochemical, and
mineralogical data were either provided by site personnel,  or obtained
from the literature or public files.  Samples  were collected from the
Cortez and Liberty pits for the purpose of providing more detail in
existing geochemical sample suites.

-------
      The initial purpose of the  study was to learn how to use the
computer codes, BALANCE, MZNTEQA2,  and PHREEQS,  and evaluate their
suitability to pit water chemical modeling.   WATEQP and WATEQ4F were
subsequently added to the evaluation.   In the process of evaluation,
occasional references and comparisons  are made to other codes, such as
EQ3/6  (Wolery, 1992), HYDROGEOCHKM  (Yeh,  1989),  PHREEQM (Nienhus et al,
1991), and PHRQPITZ  (Plummer et al,  1988).  The hydrochemical computer
codes considered in the study are listed in Table 1-2.  All of the codes
evaluated are DOS based and PC-compatible.
TABLE 1-2:
Software
Code
BALANCE
MINTEQA2
NETPATH
PHREEQE
WATEQF
WATEQ4F
Computer codas evaluated.
Author
Parfchurst, et al. (1980)
Allison, et al (1991)
Plummer. et al (1991)
ParWiurst. et al (1980)
Plummer, et al (1984)
Ball and Nordstrom (1991)
Source
uses
EPA
uses
DSGS
OSGS
DSGS
                  EPA:   United States Environmental Protection Agency
                  DSGS:  united State* Geological Survey
      Mo attempt was made to  examine any programs other than the
hydrogeochemical codes listed above.  Limnological software (e.g. CE-
QUAL-R1; U.S. Army, 1986) and numerical flow modeling programs (e.g.
MINEDW, HCI,  1992a; MODFLOW,  McDonald and Harbaugh, 1984)  have also been
used in pit water modeling, but were not evaluated in this study.  A
consideration of groundwater  flow modeling is beyond the scope of the
study, as  the emphasis is on  water quality.
      The  Cortez pit of  Lander County in east-central Nevada,  and the
Universal  Gas pit in Eureka County, were chosen as example sites for
modeling simulations.  Chemical analytical data from the Cortez pit was
used as  input in the computer code BALANCE (Parkhurst et al, 1982) to
determine  chemical  mass  transfer.  The results from BALANCE were used as

-------
                                                                        7
input for PHREEQE and MINTBQA2  to attempt  to duplicate the actual Cortez
pit water chemistry.
      Speciation simulation codes used were HATEQF,  HATEQ4F,  MINTBQA2,
and a version of PHREEQE expanded to include trace metals  believed by
the author to be important in mine water quality.  Speciation
simulations, utilizing one or mare of the  aforementioned codes, were
performed on the Cortez, Universal Gas, Liberty, and Berkeley pits to
gain an understanding of the chemical speciation and saturation states
of the pit lakes.  The simulations of the  Cortez pit water were used  to
guide interpretations of the pit water evolution predicted by subsequent
"inverse" models (i.e. BALANCE) and "forward"  models (i.e.  PHREEQE, and
MINTEQA2).
      The water in the Cortez and Universal Gas pits might  be
representative of many that will evolve in sediment-hosted disseminated
precious metal deposits.  However, they are not likely representative of
some of the high sulfide systems in Nevada,  such as  Rabbit Creek, which
has a 25 foot thick stratibound zone containing up_tc>; 75%  total sulfide_
(Bloomstein et al, 1991).   :    -__ .  . -. -?.  :.. .t-::.::  v-...:--..   :

Previous Work
      Little is known about pit water quality, because few open pits
exist that contain standing water.  Open pit mining  techniques have only
seen widespread application to-precious metal  deposits in  the last 10 to
20 years, and most pits excavated below the water  table are still being
actively mined  (Macdonald, 1992).
      Numerous studies have been done on water quality in mining
environments  (Caruccio et al, 1976; Chapman et al, 1983; Davis and
Ashenberg, 1989; Davis and Runnells, 1987;  Filipek et  al,  1987; Herlihy
et al, 1988; Huang and Tahija, 1990; Karlsson  et al, 1988;  Macdonald,
1992; Nordstrom and colleagues, 1977, 1979b, 1982, 1985 (2),  1990;

-------
                                                                        8
Potter and Nordstrom, 1977; Rampe and RunnelIs, 1989; Steffen Robertson
and Kirsten, 1989; Hicks et al, 1991; Wicks and Groves, 1993) .  Most of
these works studied acid mine drainage environments.
      Glynn et al (1992) define forward chemical modeling as the
application of an assumed reaction model to an initial condition to
predict chemical composition of water and rock as a function of reaction
progress, and inverse chemical modeling as the use of observed chemical,
isotopic, petrographic, and hydrologic information at initial and final
points to define reaction models that are consistent with the data.
Studies by Plummer et al (1983), Plummer (1984),  and Plummer et al
(1990) lay the groundwork for development of forward models through
application of inverse modeling results.  Chemical models were developed
in these studies and applied to the Madison Aquifer in the northern
U.S., and to the Florida Aquifer.  Helgeson and colleagues (1968, 1969)
were the first to apply computer techniques to mass transfer in
geochemistry (Nordstrom et al, 1979a).             -
      Pit water geochemical.model ing is a new discipline -in the mining 
industry, done primarily .by-hydrologic and geochemical :consultants.  .Few
modeling studies have been submitted for regulatory review (Gold Quarry
Mine, PTI Environmental Services and Hydrologic Consultants,  Inc.; Lone
Tree Mine, Hydro-Search, Inc.; Betze Mine,  BNSR Consulting and
Engineering and Dr. James I. Drever).
      Comparative studies have been performed for hydrogeochemical codes
(Nordstrom et al, 1979a; INTBRA, 1983),  but nothing has been performed
on the scale of this study specifically for pit lake geochemical
modeling.

General Modeling Background
      A model is a simplification of reality.   A hydrogeochemical model
is an attempt to represent, through mathematical equations describing

-------
                                                                       9
thermodynamic relationships and species/mineral  stabilities,  a system of
chemical components or reactions in an aqueous environment.   The model
may be constructed to represent a variety of size and time scales.  The
environment could be a lake, a stream, groundwater, or, in the cases
considered in this study, a pit lake and the adjacent groundwater.
      The origins of aqueous geochemical modeling can be traced to
Back's papers (1961, 1966) on hydrochemical facies  (Figure 1-2).  The
Garrels and Thompson (1962) seawater speciation  model is often cited as
the work that launched the quantitative aspect of chemical modeling, and
established the framework for many of the computer codes used today.
      Figure 1-3 shows the evolution of the more popular hydrogeo-
chemical computer programs in the last 30 years.  The Garrels and
Thompson seawater speciation model was the first milestone, and two
subsequent events, the 1979 and 1989 ACS Chemical Modeling Symposia,
inspired the outgrowth of new or revised codes.  Those codes  surviving
the last 14 years of evolution have seen significant revision.  The
trend in the late 1970's toward many different^cpderrgaye way^isythe
1980's to refinement and improvement b^^existing >d*svr,Mo~.single "code '
has been developed capable of treating the wide  range of environmental
problems to which equilibrium calculations have  been applied, nor would
such a code be practical (Plummer,  1984) .

Geocheinical Modeling Applied to Mine Water Quality
      An attempt to produce a. comprehensive model that can be applied to
many mining scenarios will probably meet with unsatisfactory results.
The variation in geologic, hydrolpgic, physical, and chemical parameters
                                               *
that determine pit water geochemistry can produce different water
qualities even among geologically similar deposits.  Examples are
illustrated by the pit water chemistry for the Yerington pit  (Table 1-3)
versus the Liberty and Berkeley pits  (Table 1-1), all of which are

-------
                                                                      10
                                              EXPLANATION
Figur* 1-21  Fane* diagram of part of the Atlantic Coastal Plain showing
hydrochaaical facias (from Back, 1961).

porphyry copper deposits.  The Liberty and Ruth pits have experienced
anthropogenic disturbance (treatment and addition of tailings) . but are
still similar to the Berkeley pit.
      The Yerington pit water quality is much better than either of the
other porphyry copper systems.  Clearly, a comprehensive model for
porphyry copper terrains must consider the variables that could

-------
                                                                       11
potentially control these differences.
        1*10
                                                                 MOO
       Modified from Basso* end Melcnotr (1990)
            rigur* 1-31   evolution of hydrogochKicl codes
            (roa Olynn  t 1, 1992).
      The Cortex pit and the Universal Gas pit waters show slight
variations in water chemistries (Table, 1-4), even though both are
derived from carbonate aquifers with a minor siliceous component.  The
variation in ore, gangue,  and alteration mineralogy, which help
determine the elements released to solution, is sufficient to introduce
notable differences in the respective water chemistries.
      As these two examples demonstrate, modelers of mine pit water are
presented with a wide variety of parameters, even among genetically

-------
                                                                      12
similar deposits,  that must be considered in describing the  inputs  to
the models.
TABLE 1-3: Kater chemistry, Yrington pit.
Sources HDBP files.

Alkalinity, bicarbonate
Chloride
Fluoride
TOS
Sulfate
Arsenic
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Lead
Silica
Zinc
pH

(DDB)
134.0
40.0
1.77
628
242
0.014
0.034
0.008
230
0.004
0.232
o.sei
< 0.001
6.9
22.3
0.076
74.0
0.012
HA
0.081
8.21
Data Disk
      All of the output files generated in  the computer modeling
exercises are included on two 3H". DOS formatted  (1.4 MB), floppy  _
diskettes, contained with the thesis. -I the dis'fcs are missing or
unreadable, printouts of the files can be examined in the main library
archives at the University of Nevada, Reno.

-------
                                                                  13
TABLB 1-4:  Water chemistry, Cortex and Universal Ga pits.
(value* in ppa).
                            Cortet 
Universal Gaa 
Alkalinity, bicarbonate
Chloride
Conductivity, in pmnos/cm.
Fluoride
Ammonia
Nitrate Nitrogen
Nitrate
Solids, Dissolved (TOS)
Sulfate
Aluminum
Arsenic
Barium
Cadmium
Calcium
Cobalt
Chromium
Copper
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Nickel
Lead
Selenium
Silica
Strontium
Thallium
Tungsten
Vanadium
Zinc
pM
282.3
24.4
NA
2.4
NA
0.207
NA
432.3
90.2
< 0.02
0.0383
0.0603
NA
45.4
NA
< 0.01
< 0.007
0.134
0.00046
11.7
18.1
0.0017
68.63
NA
0.0043
NA
34.43
NA
NA
NA
NA
0.002
8.067
Source:  Cortes Gold Mines (1992). or ONR/Cortes Gold
joint sampling (1993) . - - 
 Oeraghty s> Miller, Inc.
* Nestmont Gold.
MA  Not available


77. S
342
903
0.394
0.13 *
1.3 
5.7
691
30.7
0.174
< 0.180
0.12
< 0.007
145
0.02
< 0.01
< 0.007
0.134
< 0.5
3.74
38
0.071
50
< 0.015
< 0.05
< 0.13
19.49
0.514
< 0.15
0.051
< 0.007
< 0.005
8.67
*in.._. .... -.

 "- . -, * -.'


-------
                                                                      14
                      2.   OEOCHEMICAL OVERVIEW

      Most geochemical modeling codes  incorporate  the fundamental
mathematical relationships of aqueous  geochemistry, including ionic
strength, activity, equilibrium, speciation, and solubility.  In
solutions containing numerous ions and chemical species, the
calculations of ionic strength, activity coefficients, and chemical .
speciation become too cumbersome to be attempted manually, and are best
handled by computer.
      A general knowledge of the basic physical principles of aqueous
geochemistry is important in understanding how the programs solve
problems.  These principles are at the root of geochemical computer
modeling, since the necessity of rapid computational ability in solving
these problems was the driving force that inspired the development of
computer modeling software.

Zonie Strength
      Ionic strength considers the higher degree of electrostatic
effectiveness of polyvalent ions in solution, which would be neglected
in  simple consideration of total molal concentration  (Drever, 1988).
Ionic strength  (Equation 1) is a required parameter for calculation of
activity coefficients, using molal concentrations  from the input data:
                      I . * I mpS             (1)
The variable m, is the molal concentration of the ith ion,  and zt is the
charge  on the ith  ion.  Equation  (1) illustrates the greater weight
given polyvalent ions in the calculation of  ionic  strength, i.e. charge
is  raised to the power of  two.  If the component is an uncharged
species,  such as H4SiO4,  then  z^  0,  causing the  term for that species
to  fall out of  the equation.  Uncharged  species, therefore, do not
 contribute  to  the  calculation of  solution ionic strength.

-------
                                                                      15
    An example of the relative significance  of  ionic  strength is  seen by
comparing the Sierra Nevada ephemeral  spring water  of Garrels and
Mackenzie (1967), which had an ionic strength of  0.000485  molal,  or
10'1-1  (calculated by author in WATEQF), to aeawater, which has an ionic
strength of around  0.6799 molal, or 10-0-17  (Parkhurst  et  al,  1980),  more
than 3 orders of magnitude higher.

Activity Coefficients
      Nearly all geochemical computer  models are  based on  the ion
association theory, which describes the behavior  of ions in  solution  in
terms of activity.  The activity of an ion in solution can be defined as
its "effective concentration"  (Drever, 1988), and incorporates the
assumption that charged ions exert a different  influence over adjacent
ions depending on the ion's size, charge, and the solution ionic
strength.  The ratio of a species' activity  to  its molal concentration
is called its activity coefficient.  The equation that adjusts molal
concentration (mj  by the activity coefficient  (yt)  to obtain activity
(a,)  is:          "~ =
                          < - 7* ' %
      A consideration of activity is essential  because only  in ideal
solutions does the molal concentration of an ion  or species  equal its
activity (Drever, 1988), but the condition of ideality does  not exist in
natural waters.   Electrostatic interactions  between charged  species, and
between ions and solution, impart non-ideal  behavior to the  system.
Under such conditions, the concentration of  an  ion is  best described by
its activity.
      Morel and Bering (1993) define an ideal system as "one in which
the free energy of a species is independent  of  the nature and concentra-
tion of other species, and state that this  occurs in  either of the two
following cases:

-------
                                                                        16
 1.  The system is very dilute,  and all  individual solute molecules are far  apart
    and effectively 'ignorant*  of  each  other (i.e.,  they have no energetic
    interactions and their individual free energies  are unaffected by each
    other's presence).  This im the "infinite dilution* reference state.
 2.  The major solutes  (those accounting for the bulk of the dissolved species)
    are considered to be at a fixed concentration and whatever effects they have
    on the free energy of another  species are accounted for in the standard
    value  (MI )  of  the chemical  free energy of  that  species.   This  is  the 'fixed
    composition* reference state.
       Since neither of these  cases is  encountered in natural waters,
 activity coefficients are needed to  describe interactions among  ions  and
 species in natural waters.
       Several  theories have evolved  to explain the  activities of species
 in solution, and  calculate activity  coefficients.  The appropriate
 formulas for calculating activity coefficients differ depending  on the
 solution ionic strength.

       Dby-Huckal Equation:  In relatively dilute  solutions  (I  s 10']),
 deviations  from ideal behavior  are primarily caused by long-range
 electrostatic  interactions (Stumm and  Morgan,  1981) .   At these ionic
 strengths,  a simple, single-ion activity coefficient formula, the Debye-
 Huckel equation,  is assumed to  give  an adequate description of ion
 interactions for  the purpose  of calculating activity coefficients.  The
 Debye-Huckel equation assumes that ions are point charges, the
 interactions are  entirely electrostatic,  and arrangement of ions about
i one another conforms to a Boltzmann  distribution (Drever, 1988) .  The
 Debye-Huckel equation assumes that ions behave "as  charged particles  of
 finite sizes in an electrostatic field of uniform intensity"  (Hem,
 1985) .  Therefore, the Debye-Huckel  equation contains no term to account
 for size  or hydration effects of the ion.
       The simplest form of determining the  activity coefficients (7,)  is
 also  the  simplest form of  the Debye-Huckel  equation:

                          log  7,  - -Az^/I               (2)

-------
                                                                       17
where z is the charge on the ith ion, and  I  is  the  ionic  strength of the
solution.  Equation 2 is valid to ionic strengths of  about 10"' molal.
The constant A is expressed as  (Truesdell  and Jones,  1974) :
                  (1.82483 X 10*) (d*)
            A -  -  (moles ")  (10 g HaO)*
where d is the density of water, T is 'the absolute temperature,  and c  is
the dielectric constant of water.
      At higher ionic strengths, the Debye-Huckel equation  becomes
inaccurate, because the formula predicts impossibly high concentrations
of ions in close proximity to one another (Drever, 1988), and tends to
underestimate the degree of ion association  (Truesdell and  Jones,  1969) .

      Extended Debye-Huckel Equation:  For ionic strengths  up to 10*1
molal, the extended Debye-Huckel formula, which incorporates  two
additional constants to account for ionic interactions, provides a
better approximation: -.-:--         .   .     .  .    .-.;. .....    -.-.^..
                    log
                              1 + Ba /I
The constant a represents the hydrated radius of the particular ion,  and
B is expressed as (Truesdell and Jones, 1974) :
            (50.2916 x 10J) (d*)
      B   - (cm'*)  (moles )  (10' g H,O)*
The constant A is commonly referred to as the "Debye-Huckel d  constant,"
or simple DHA.

      Robinson-Stokes Dby-Huckel Equation:  A modified version of the
extended Debye-Huckel equation, for use at higher ionic strengths,

-------
                                                                      18
incorporates a second term with another adjustable  parameter, b
(Truesdell and Jones, 1974; Robinson and Stokes,  1955):

                         -Az,/I
              log ft -	  + bl          (4)
                        1 + Ba /I

The b parameter is constant for a given ion, and  accounts  for the
decrease in concentration of solvent that occurs  at  higher ionic
strengths.  The bl term causes an increase in the activity coefficient
with increased ionic strength (Drever, 1988) .  Ball, et al (1979)
considered this equation to be more reliable than either the extended
Debye-Huckel or the Davies equation, and thus incorporated it into the
WATEQ2 code.  Several USGS codes  refer to Equation  (4) as  the "WATEQ
Debye-Huckel equation."
      The A and B constants,  and the Debye-Huckel & and b  parameters
(shown in Appendix A) are tabulated in many aqueous geochemistry texts,
and are incorporated into computer speciation codes.  The  A and B
constants are calculated from the dielectric constant, density,  and
temperature (Earner, 1968).  For deviations from 2SC, they require   - .:
temperature and pressure correction before being applied to calculation
of activity coefficients,  a task which all computer codes  perform.

      Davics Equation:  The Davies equation  (Equation 5) incorporates
semi-empirical data to account for ion interactions.  It is generally
accurate at ionic strengths up to about 10' 3 (0.5 molal) .  Davies
eliminated the parameters a, and b, the constant B, and added the
empirically derived.linear term (cl) , where c lies between 0.2 and 0.3.
Davies' original derivation of the equation set c at 0.2,  which he later
changed to 0.3 believing it provided a better fit to experimental data
(Davies, 1962).  The Davies equation is used in many cases by computer
codes because the A parameter required for the Debye-Huckel equations

-------
                                                                      19

frequently cannot be estimated (Ball and Nordstrom, 1991) .
                       /I
log
                                       /I
                                                -  cl        (5)
      The principal advantage of the Davies equation is to provide, a

"quasi-constant value of the activity coefficients in the range I  0.3

to 0.7 M" (Morel and Bering, 1993).  The ionic strength of natural

waters rarely exceeds 0.7 M, and the inaccuracies shown by the Davies

equation in the 0.3 to 0.7 M range are usually less than errors

introduced from other sources (Morel and Bering, 1993).  Different

computer codes use different values for c in the Davies equation.  In

PHREEQE, WATEQF and WATEQ4F, c = 0.3, whereas in MINTEQA2, c - 0.24.
                                                     )
      A comparison of activity coefficients for different ions using

three of the aforementioned equations (Debye-Huckel,  extended Debye-

Huckel, and Davies)  is shown in Figure 2-1.  Pankow (1991) observed that

the term -0.2Z in the Davies equation causes a minimum near log /I  0
                                               -:.-*	   -. -  =_.*- ^ -
in the plot of yt vs.  log  /I.  As  the plot  shows,  YI does'not decrease

steadily as I increases, but rather increases for large ionic strength.

This occurs because the amount of solvent available for solvation of

ions decreases as the ionic strength increases  (Bockris and Reddy,

1970).

      Figure 2-1 also illustrates that activity coefficients approach

1.0 in very dilute solutions, causing activity to approach molal

concentration.  At higher ionic strengths,  activity coefficients

generally decrease,  with the noted exception of those calculated by the

Davies equation, which begin increasing again after log /I - 0.  Similar

behavior is shown in Figure 2-2, which plots tCa1* calculated from the

three Debye-Huckel equations, vs. ionic strength.

      Table 2-1 summarizes the four previously discussed formulas, and

-------
                                                                      20
their range of applicability.
                -25
Figure 2-It  Activity coefficient* vs.  log /I.   The Debye-Ruckel
equation deviates as ionic strength increases,  producing lower  activity
coefficients than the Davies or extended Debye-HucJcel  equations (from
Pankow, 1991).                                   .--,...
      Ion interaction modalst  Beyond  ionic strengths  of  0.5,  ion
interactions become so great that deviations from the  ideal  solution
behavior are attributed mostly to short-range interionic  forces  (Stumn
and Morgan, 1981) which are more appropriately described  by  ion
interaction models.
      The Brensted-Guggenheim model was  one of the early  models  that net
with  success  (Harvie and Heare, 1980), but the ion-interaction models of
Pitzer  (1973, 1979, 1980) are probably the most popular today.
      Drever shows a simplified form of  the Pitzer formula:
(D-H)
        RT
                            4J (I)nyn,
(6)

-------
                                                                       21
where G.x  is  the  excess Gibba free energy per kilogram of water, D-H
represents a Debye-Huckel  term,  XtJ represents binary interactions, and
uijk represents ternary interactions,  which are significant only at very
high ionic strengths  (Drever,  1988).   Harvie and Heare (1980) have
performed what many believe  to be the most successful application of
Pitzer models to brine solutions (Nordstrom and Ball, 1983).
PHRQPITZ  (Plummer et al, 1988),  and SOLMINEQ.88 (Kharaka et al, 1988
utilize ion-interaction theories,  and can be applied to modeling highly
concentrated solutions.  For purposes of modeling mine pit water,
              1.2
              10
              08
           >06
              0.4
              0.2
              0
                     0.001
0.0!      0.1
  Ionic Sutnfth
10
Figure 2-2  Variation of  the activity coefficient for -yCa** according
to the three forms of the  Deby-Huck*l equation (from Orever,  1981;  Sq.
(2-7)  basic Debya-Bflckel,  Kq.  (2-8)   Extended Debye-Huckel, Iq.  (2-
10)  Robinson-Stokes Deby-BQckel).
it is unlikely that ionic  strengths will exist high enough to warrant
use of the  ion interaction models.   Even the most concentrated pit water
known, the  Berkeley Pit, has an ionic strength of 0.3 molal (calculated
by author in WATEQ4F, see  datadisk file BPSP01W4.OUT), which does not
approach the levels seen in brines and seawater.

-------
                                                                       22
TABLE 2-1: Equations for  activity coefficient (y,),  adapted froa Pankow
(1991).  Applicable ionic strength range obtained from Stunua and Morgan
(1981).
                                                   Applicable
 Equation Name               Formula                 Ionic Strength Range
Debye-HOckel log y, * -Az,/l I < 10"' 
Ext
-Alj



1 * Bi /I
Robinson-Stoke* -Az,/I
1 + Bi /I
Davies 1 /I

= s


1 1 * /I

- Cl I < ID'"

      Ion Pairs:  Above  ionic strengths of l(Tl, departures from the
behavior predicted by  the  Debye-Huckel  theory are thought to be due to
short-range  interactions,  such as those responsible for the formation of
ion pairs  (Garrels and Thompson,  1962).
      The formation  of ion pairs has two effects (Drever, 1988).  First,
charged ions come together to form uncharged species, thus decreasing
ionic strength,  and  second,  the concentrations of free ions such as Ca**
and SO42' decrease as they  become associated in ion pairs.  This can
produce misleading results in calculations of ionic strength and
activity coefficients.  Codes that use  equation (1) to calculate ionic
strength, and contain  uncharged ion pairs in the database, may generate
suspiciously low ionic strength.  The Pitzer models, which incorporate
very  few uncharged ion pair species, give a more realistic depiction of
ionic strength.   Table 2-2 shows a comparison of values calculated in
both  WATEQF and PHRQPITZ for various water samples.
       Maclnnes Convention:  The Maclnnes convention  (Maclnnes, 1919), or
mean  salt  method, can be invoked in some codes at the user's request, to
estimate  activities  of free ions.  The  convention assumes that the

-------
                                                                       23
Table 2-2:  Comparison of activity coefficient! modeled in PHRBBQB vs.
PHRQPXTZ  (top line of each pair of simulations is from PHRBBQB, bottoa
line is from PHRQPITZj from Olynn et al, 1992).
1
r~
Jea* .. SO

Coul. NT
fttntn
trrlf C., TX
1(4
S
Sv*tvttr. VT
D..4
into
M4f
Se*
ra-lu
Co.. m
-4.
0.999
0.999
0.997
0.9(1
0.911
0.912
0.919
O.M2
0.7(3
0.900
O.t99
0.60*
O.M7
o.tot
O.Ml
O.M1
loale
Itrcaftk
0.00(1
0.00*4
0.102
0.114
O.*f
0.71
" 2.JJ
2.34
*.14
.17
3.41
.57
9.3(
9.40
10.9*
10.91
4.73
13.34
Uf
-1.67
-1.71
-1.14
-1.23
-1.49
-3.S
-i.il
-1.25
MM
-3.17
-3.7

Clett
6.12
0.1S
-0.41
0.33
0.12 .
O.IS
0.34
0.2
W*
1.**
1.04
-2. (3 -0.05
-1.03 -0.29
-0.73 -1.30
-0.01 -2.(2
-0.29 -l.2
-0.01 -1.17
lfellttl CTMM
6.61
0.21
-1.27
-0.97
2.53
2.73
0.33
0.49
__
3.99
2.72
1.12
0.71
-2.7t
o.a
-0.2*
-2.11
-2.7(
-6.41
-0.42
-0.39
-0.*4
-0.(7
-1.03
0.13
0.04
-4.93
-3.94
-O.JI
0.02
0.33
0.45
0.09
-0.22
, l*t1
-16.11
-10. M
- 4.25
- 4.23
- 2.T5-
-2.49
- i.ir
- ohl
-0.07
- 1.41
-o.M '
- 0.07
-O.(l
- 0.37
- 1.34
- 1.25

single-ion activity coefficients of  K* and  Cl' are equal to each other

and to the mean activity coefficient of  KCl at all ionic strengths.  By

definition:
            If
  IT.  


T./-KC1
            then
            and
                                           YJ./.HaCl

                                            T./.KC1
                                             Tr*./.KCl
            and
                                            T./-KC1

            and so forth.

This method gives reasonable estimates  for the activity coefficients of

-------
                                                                             24
free  ions since K* and Cl" salts  do not normally  form strong ion pairs

(Millero and Schreiber,  1982) .

       Truesdell and Jones'  (1969)  comparison of the  Debye-Huckel

equation, Davies equation,  and the Maclnnes Assumption are  shown in

Table 2-3.
TABLE  2-3:  Comparison of  the Debye-HQckel  equation,  Davies equation,
and  the KacXnnes Assumption (modified from  Trueadell  and Jones,  1969).
       Method
                 Advantage
                                                             Limits
(A)  Debye-Huckel Equation:
                 Bi /I
                  Justified from
                  theoretical studies.
                  Can be used at all
                  temperatures.
                                    i must be estimated from
                                    experimental data.  Ionic
                                    strength must be less
                                   than 0.1 for most mono-
                                   valent  ions, less than
                                    0.05 for most divalent
                                    ions.  If i is carefully
                                    chosen, equation may be
                                    accurate at greater
                                    concentrations.
(B)   Davies Equation:
    log
-As(
                                   CO,*' unless corrected
                                   for ion association.
       Heutral  specie*t  Although neutral species are excluded from

calculations of ionic strength, they are not  immune from the influence

of  activity coefficients.   Activity coefficients (>*)  of uncharged

species can be approximated by the  following  formula  (Helgeson,  1969) :


                          yt  -  10-"
where I is  the ionic strength of  the solution.   This approach  is  used in

-------
                                                                      25
the USGS codes and MINTEQA2 for all neutral species except H20.
      According to this formula, as ionic strength approaches zero
(dilute solutions),  the activity coefficients of uncharged species
approach one.  With increasing ionic strength, the activity coefficient
rises slightly above one.   The probable reason for this behavior is that
much of the water in concentrated solutions forms the hydration shells
of ions, making less water available to solvate uncharged species
(Drever, 1988).
      Limitations:  A model is only as good as the assumptions on which
it is based.  Nordstrom et al  (1979a)  express reservations about ion
association theories and the non-thermodynamic assumptions from which
they were derived.  The activity coefficients used to describe the non-
ideal behavior of ions represent semi-empirical equations with inherent
uncertainty.  The assumption of ion association may actually be a naive
representation of the true interactions of "ions" in aqueous solution
(Nordstrom et al, 1979a).
      The inconsistency of the equations and thermodynamic data used in -
different codes may produce discrepancies.  Nordstrom et al (1979a)
demonstrated this by running the same  input through 14 different codes,
and comparing calculated results for molality, activity coefficient, and
saturation index.  In some cases,  the  discrepancies between codes
exceeded several orders of magnitude.
      A significant  source of  uncertainty could be.the activity of
uncharged species.  Reasons for this are the lack of reliable
information on the activity of neutral ion pairs, and the fact that they
                                                           f
often comprise the dominant species in aqueous systems (Nordstrom et al,
1979a).

Ionic Balance
      Ionic balance  refers simply to the balance between cations and

-------
                                                                      26
anions in solution.  To determine ionic balance, concentrations of
individual anions and cations must be converted to equivalents.  Since
equivalents are generally too large for application to natural waters,
the convention of milliequivalents per liter  (meq/1) is commonly used..
To convert from concentration in mg/1 or ppm to meq/1, the following
formula is used (adapted from Mazor, 1991):

                            mg/1
                  meq/1 *  	  x charge
                            gfw

By summing the positive meq/1 values and comparing with the negative
meq/1 values, the accuracy of the ionic balance for the particular water
analysis is revealed.
      A comprehensive analysis for the major elements in a water sample
should reflect the ionic balance with minimal error, i.e. < 10% (Lyons,
personal communication).  Plummer (1984) recommends that analyses with
more than 5% charge imbalance should be checked carefully.  All natural
waters are charged balanced  (Plummer et al, 1983),  so a balance
discrepancy indicates an error or omission somewhere, either in sample
collection, analysis, transport, or data input.  Prior to conducting any
detailed study of a water chemical analysis, such as a modeling effort,
the chemist should verify the validity of the chemical analysis by
checking its ionic balance.
      For waters with high trace metal and H* concentrations  (as in some
mine-related waters), omission of ions normally regarded as "trace
metals" and H* in the analyses may result in a significant imbalance.
An example is the Berkeley pit, in which Al, Fe, Cu, Zn exist at higher
concentrations than  some of  the major ions  (see Table i-1).  Therefore,
sampling and analyses must be conducted in the context of the aqueous
and geologic environment studied, i.e. knowledge of minerals present and
that  may contribute  to solution chemistry.

-------
                                                                       27
      All computer programs have some provision  for  determining the
ionic balance,  and display a printout of the results.   Some codes shut
down if the  ionic balance of the input analysis  exceeds some error
tolerance, such as 30% difference between cations and  anions.   Others
may only provide an error message.  If the program runs,  but reports a
significant  charge imbalance,  the results should be  interpreted with
caution.
      Some computer codes, including WATEQ4F, report the  ionic  balance
in terms of  equivalents per million (EPM).  PHREEQE  reports the ionic
balance as the  difference between cations and anions in molality.

Mass Balance
      Mass balance is an ambiguous term in the literature that  may  refer
to one of the following two concepts:
1.  Change in mass of a particular element,  compound, or  chemical species during
    dissolution or precipitation along a  reaction path.   Mass balance can be
    considered as a "budget" of sources from which the dissolved constituents in
    a water originate.  A simplified equation can be written (Plunmer et  al,
    1983; Plummer, 1964):
     Initial solution composition + fieactant phases 	>
                     Final solution composition + Product  phases
      Plummer  (1984)  and Plummer et al (1983)  provide good discussions
of this aspect  of mass balance,  and how it affects the progression  of
calculations in subsequent geochemical processes.  This process is  also
referred to  as  "mass transfer*  if the mass balance reaction involves the
shifting of  chemical constituents from the solid to the aqueous phase
and vice versa  (dissolution/precipitation),  or from the aqueous to  the
gaseous phase  (degassing/ingassing).
      The concept of mass balance also applies to conservation of
electrons if the problem involves redox,  such as sulfate reduction,
pyrite oxidation,  or other transfer of electrons from one species to
another.  Hydrated electrons do not exist in effective concentrations in

-------
                                                                        2F
solution  (Thorstenson,  1964)  so that if electron transfer does  occur
through a redox reaction,  the electrons transferred are  conserved among
the dissolved species  (Plummer, 1984).  Mass balance equations  for
hydrogen and oxygen  are often not included in chemical models,  because
of the impracticality  of analytically determining the total masses of
these elements in  solution (Plummer et al, 1983).
      The most difficult aspect of mass balance modeling is the non-
unique nature of modeled results that usually occurs (Plummer,  1984) .
2.  Conservation of mass in  the calculation of chemical  speciation, sometimes
    referred to as mole  balance.  When partitioning the  total mass of a
    particular ion among its various species, the computed sum of the free and
    derived  (complexes)  species must be equal to the given total concentration
    (Nordstrom et al,  1979a), for example:
             Total[Ca]  - CaCOj  + CaOH  + CaHCO, + CaSO,
Garrels and Thompson (1962) provide an example of mass balance  on  total
sodium for some possible species:

          an*' TOTAL  nwa'uncooplexed +  JWaHCO,* + DWaCO,' + n*S
-------
                                                                       29
release 1 mmol/1 of Ca3'  ion  and 1  mmol/1 of HCO,'.  The mass transfer
that occurred is the  transfer of 1 mmol/kg of solid calcite into
solution as dissolved Ca'* and HCO,'.
      Incorporation of mass  transfer calculations into a model  is
usually a necessity for any  forward reaction path simulation.   Mass
transfer occurs during reversible equilibration  reactions between
minerals and solution, and the subsequent  speciation, as in the
dissolution of calcite described above.  Mass transfer also occurs in
irreversible reactions such as dissolution of pyrite from pit wall
rocks.  Mass transfer calculations require some  extent of user
manipulation regarding which minerals to include,  and in the case of
irreversible reactions, the quantities of  minerals  involved.

Equilibrium Thermodynamics         '
      As presented by Drever (1988),  for a hypothetical reaction  in
which a moles of A ion reacts with b moles  of  B  ion to form  c moles of C
and d moles of D:                                   -.-..
                     aA + JbB  cC + dD       (7)

at equilibrium the following equality will  hold  true:
                                            (8)
where a is the activity of the particular ion, and K^ is the
equilibrium constant for the reaction (also referred to as K^, for
solubility product constant, or K,) .   The standard means  of  expressing
the relationship depicted in equation (7) is to place the reactants on
the left side, and the products on the right, which would correspond to
dissolved ions on the left and solid mineral on the right.   As an
example, the equation may correspond to the reaction between ferric iron
and hydroxide to form ferric hydroxide and water:

-------
                                                                       30
                Fe1* + 30IT  -  FeO(OH)  + H,O     (9)

If the activities of water and pure  solids  are assumed to be unity,  as
is customarily done, then an equilibrium equation  can  be written as
follows:

                a[Fe"]  a[OH-]J    K^          (10)

      Equilibrium constants can also be derived from basic thermodynamic
data (such as the free energy of  formation  AG(01) as presented by Drever
(1988).

      Saturation Index:  The product a[Fe3*]    a [OH']1 is called the ion
activity product (IAP) , and at equilibrium,  IAP =  K^,.   The quantity
lAP/K^, is called the  saturation  index (SI) ,  and at equilibrium will  be
1.0  (or log SI = 0) .  The saturation index  is  most commonly expressed in
logarithmic form, since the values may span many orders of magnitude:

                                  - IAP        -            ...
                   log SI - logto  	           (11)
                               ...  K^         .       :..  :-.-  ...	

      If the system is not at equilibrium,  then the IAP will not equal
                                               1
the  K^,, and reaction  (7) will  tend to proceed in one direction or the
other.  If SI < 1  (log SI < 0), the  system  will be undersaturated with
respect to the particular mineral, and the  mineral will tend to dissolve
into the solution.  In equation. (7),  if the mineral and water are
represented by the  components  cC  and dD respectively,  then the  reaction
will tend to proceed  from right to left.  If SI >  1 (log SI > 0),  the
system will be supersaturated  (also  referred to as oversaturated)  with
respect to the mineral, the reaction will tend to  proceed from  left  to
right, and the mineral will tend  to  precipitate from solution.
      The thermodynamic data in a computer  code's  database may  come  from
several different  sources.  One potential discrepancy that may  be

-------
                                                                      31
encountered in tabled thermodynamic data is that results of one
experiment (values for AGe or K,p)  may not match the results for the same
species from another researcher's experiment.  The determination of
thermodynamic data has been a subject of active research among chemists,
and the data are continuously being revised and expanded.  New
thermodynamic data periodically find their way into the computer codes,
so the user should be aware of the sources of the data, and the
potential differences in modeling results that may occur.
      Errors in thermodynamic and analytical data will cause a range of
uncertainty for the SI that must be considered when interpreting the
output.  This uncertainty will vary according to the complexity of the
mineral stoichiometry and input data errors {Ball and Nordstrom, 1991).
Nordstrom et al (1979b)  chose an "equilibrium zone" around the
saturation index equal to the estimated uncertainty of the solubility
product constant.   Within these limits, the solution is considered to be
in equilibrium with respect to the mineral phase, and only outside the
limits is the mineral considered over or under saturated.   . s	 .
      Examination of saturation indices for natural waters will-often
reveal many mineral phases that are oversaturated by several orders of
magnitude.  These phases may not necessarily be precipitating in the
system, even though thermodynamics say they should.  Minerals must often
overcome a level of energy known as the "activation energy" before
precipitation can occur.  As reactants go to products, they must pass
through an intermediate stage of higher energy than the reactants that
                                           j
ultimately will form (Drever, 1988).  Precipitation may also be hindered
by a lack of available nucleation and growth sites (Davison and Rouse,
1988) .
      Modelers must also be aware of cases of "partial equilibrium"
{Plummer et al, 1983) .  Although a mineral may actually be dissolving or
precipitating in a groundwater system, the SI calculation may indicate

-------
                                                                      32
equilibrium with respect to the mineral.   Partial equilibrium  occurs
when one or more slow mineral-water reactions, or changes  in pressure or
temperature, drive a larger set of faster  reactions, the latter  of which
may continually shift to maintain equilibrium  (Helgeson, 1968) .
      Plummer  (1984) concluded that a mineral was in equilibrium along
flow path if it was both saturated and had zero mass transfer.   The
mineral was in apparent equilibrium if speciation calculations showed
saturation in the system, but had non-zero mass transfer along the flow
path.  If the mineral has non-zero mass transfer along the flow  path,
but is not saturated, it is reacting irreversibly.  Minerals that are
not saturated in the system and have zero  mass transfer are either not
present along the flow path or, for kinetic reasons may be considered,
inert on the time scale of the flow system (Plummer, 1984).

      Reversible vs irreversible reaction*:  Modelers use the term
reversible to describe a reaction involving a mineral that may reach
equilibrium in solution.  The mineral may  dissolve or precipitate along
reaction path as thermodynamics demand to  maintain a state of
equilibrium in the system.  An irreversible reaction is one in which the
mineral is not expected to reach equilibrium in the system, or is unable
to because of thertnodynamic conditions.  Irreversible reactions
generally involve slow dissolution of one  or more minerals that  do not
reach equilibrium (Plummer et al, 1983) .

      Xncongruent dissolution:  Many aluminosilicate minerals dissolve
incongruently. leaving a residual clay mineral, such as K-feldspar
weathering to kaolinite  (Drever, 1988):

         2KAlSi,O +  2H* + 9H,0     Al2Si,Os(OH)4 + 2K* + 4H4SiO4

This behavior presents problems in chemical modeling, since the  mineral

-------
                                                                      33
may not demonstrate  reversible,  equilibrium solubility behavior
(Nordstrom et al, 1990).  Although a modeling simulation may show a
solution to be supersaturated with respect  to a certain silicate mineral
phase, one should not expect to  see the mineral precipitating in the
field.  This may be  a common problem in chemical modeling,  since many
simulations will involve silicate  assemblages.   Therefore,  equilibrium
constants for many silicates should be  used with caution.   The problem
applies to feldspars, smectites, illites, chlorites,  amphiboles,  micas,
pyroxenes, and pyrophyllites  (Ball  and  Nordstrom,  1991) .

      Solubility v. K^i  Values for K,p do not necessarily correlate
with mineral solubility, and cannot be  used to  predict  relative
solubilities of minerals because of complications  introduced to the K^,
equation by polyvalent ions.  Sawyer and McCarty (1978)  provide a good
illustration of this point, using barium sulfate and  calcium fluoride as
examples.  At 20C,  the solubility  of these compounds is:

                 BaSO,    l.l x 1
-------
                                                                      34
      Temperature/pressure dependency:   Dissociation/equilibrium
constants are thermodynamic constants,  and are independent  of  the
solution ionic strength, but not  independent of temperature and pressure
(Garrels and Thompson, 1962) .  The derivation from first principles
shows the relationship between the equilibrium constant and the standard
free energy of reaction, which also  illustrates the dependency on
temperature and pressure (Drever, 1988) :
                                                 "
 ace   aD*               -AG
	    exp
                        RT
Values for AGe and Kq are experimentally derived,  generally at  25C and
1 atmosphere pressure (standard temperature  and pressure, STP) .
Deviations from STP will change the value of AG and K^,,  and hence
change the thermodynamic behavior of the particular mineral  or  aqueous
species.  Most minerals exhibit higher solubility  with higher
temperature, resulting in higher concentrations of dissolved species.
                                                .. C" ' "  " -i a. ' r. C = _ .
Calcite is the notable exception, which becomes less soluble with higher
temperature.  Gases also show higher solubility in colder solutions.
      Computer codes use one of two formulas to correct constants for
temperature deviations from 25C.  The preferred formula is  (Allison et
al, 1991; Ball and Nordstrom, 1991; Parkhurst et al, 1980; Plummer et
al, 1984):                                                             %

          log Kf - A + BT + C/T + DLog(T)  + BT* + P/T + GT*   (12)

      Unfortunately, the constants  (A through G) are only available for
a limited number of chemical species and minerals. Only 38  species in
WATEQ4F have the constants available  (Ball and  Nordstrom, 1991) , 34 in
PHREEQE  (Parkhurst et al, 1980), and only 25 of the more than 1000
species in the MINTEQA2 database have the constants (Allison et al,
1991) .  For species without the constants, the  Van't Hoff equation is

-------
                                                                        35
used:

           logK, - log K^  -  (AHr/2.303R) x  (1/T-t/T*)       (13)

For temperatures far  from  25C, the variation of  AH,0 with temperature
should be recalculated  from heat capacity data  {Drever,  1988)

Chemical Spcciation
      A species is defined as a chemical entity such as  an ion,
molecule, solid phase,  etc.,  that is present in solution (Drever,  1988).
Species are generally grouped on the basis of the major  cation.
Chemical analyses typically express the concentration of a particular
ion in terms of the "total"  Na, K,  Ca, Mg, etc., and the sum of  the
molal concentrations  of  the  ion in each species will equal the "total"
ionic concentration.
      An illustration of speciation is seen in the Garrels and Thompson
model (1962),  in which  they express their hypothesized speciation
distribution of the major  cations in seawater:

             Na* TOTAL  -  Ha* + NaHCO,0 * MaCO,. +  NaSO*.:     .       .  .  .
             K* TOTAL  -  K* + KSO"
             Ca2* TOTAL   -   Ca2* + CaHCO/ + CaCO,0 + CaSO4
             Mg" TOTAL      Mga* + MgHCO,* + MgCO,e + MgSO
      Using sodium, they show that a mass balance relation can be
written for each analyzed  constituent:
      ni*' TOTAL  JUia'uncoaplexed + uitoHCO,' + flWaCO,- + awS04-       (14)
where m*' TOTAL is the molal concentration of total sodium.
      Species can also be  grouped on the basis of a particular system.
For example,  in the system CaCO,-H20-COj, possible species may  include
Ca2*, CO,, C01U,,. HaCOJf HCO,',  CO,*',  H*,  OH", H,0aj,  HjO,,,,  CaCO,,.,,  plus
                                                 /
various complexes  (Drever,  1388) .  The MINTEQA2 code includes  the
following ten "soluble"  species in a CaCOj solution at equilibrium:

-------
                                                                       3f
      Ca", CaOIT, CaCO,', CaHCO,',  H,COlf  HCCV, CO,1', H',  OH', HjO

      The determination of aqueous species  distribution is accomplished
by means of a chemical model similar  to  that developed by Garrels and
Thompson (1962), which was the  first  application of  the method of
                                                                     
successive approximation.  Although their model  only considered 17
species, Garrels and Thompson stated, "the  manipulations involved in
solving these (equations) simultaneously are tedious,   which  is why such
calculations today are left to  computers.
      Geochemical speciation codes write solute  reactions as  association
(formation) reactions, whereas  the solid reactions are  written as
dissociation (dissolution) reactions  (Ball  et al,  1979) .   The
association and dissociation equations are  sets  of nonlinear  mass action
and mass balance equations that are based on the equilibrium
relationships discussed in the  previous  section.   Codes used  in this
study (MINTEQA2, PHREEQE, WATEQF, WATEQ4F)  determine  speciation by
solving these equations through the mathematical approach known as the
continued fraction method (Wigley, 1977) .
      An example of a speciation  calculation, as it  is  performed by the
WATEQ codes, is demonstrated by Truesdell and Jones  (1974) .   Anionic
weak acid species, such as silicic acid, are computed first.   By
combining mass action and mass  balance equations,  the speciation
distribution can be determined  from the  total analytical  concentration,
pH, the equilibrium constant, and the activity coefficient.   The WATEQ
example demonstrates silica speciation  (H,SiO,~ and H4SiO4)  beginning with
the mass action equations:
         \
                   H4SiO4  -  H* +
                                                       (15)
       and         H,Si04-  H* + H,SiO4-J
rearranging these equations gives:

-------
                                                                       37
                          "(H4Si04) (yH4Si04)
                                                               (16)
    and
                      '(H2Si04-2) (yH2Si04'2) (lO'"")
                          "(HjSiO/J (yH4Si04)

The mass balance  (or mole balance)  equation for total silica  (silicic
acid and silicate ions)  is:

           "sitoti  "H4Si04 + "HgSiO/  + "H2Si04-2     (17)
The mass action equations can be combined with the mass balance
equations to solve  for "H4SiO4/ as  shown by:
"H4SiO4
                                                                    (18)
                             K,10pH
                yH4SiO4
The quantity *H4Si04 is then substituted back into the mass action
equations  (15) to  solve  for "I^SiO/ and "H2SiO4"2.  A similar procedure is
used for speciation of other  components,  such as phosphate, borate,  and
sulfide.  Carbonate-bicarbonate distribution also includes pH and
alkalinity after correction for other weak acid radicals  (Truesdell  and
Jones, 1974) .
      Garrels and  Thompson  (1962)  derived a total of 17 species from the
major ions in seawater,  requiring 17  independent equations.  They admit
that their model is a first approximation,  through their assumption  that
interactions among the major  ions result only in the formation of ion
pairs.  Despite the shortcomings,  the eight major ions analyzed, and
their associated species, constitute  over 99 percent of the dissolved
solids of sea water.  Subsequent studies have expanded the number of
known species in seawater to  at least 60 (Parkhurst,  et al 1980) .

-------
                                                                       38
       Limitation*:   The results of speciation calculations, such as
 those performed by computer,  are only as good as the input data.  The
 use of unreliable equilibrium constants or activity coefficients can
 completely change the predominant form of a complexed species, and
 therefore the interpretation  of the water chemistry (Millero,  1975,
 1977; Nordstrom and Ball,  1983).   This can have impacts on subsequent
 modeling for points further along a reaction path,  and will propagate
 any error introduced.
       The number of species that can be modeled depends on the
 availability of data for the  chemical model.  For computer simulations,
 this will be a function of the size of the database.  Addition of
 desired species or minerals to the code's thermodynamic database may be
 necessary before modeling is  attempted.  For example,  it would be futile
 to attempt to model the chemistry of a silicate aquifer if the only
 silica-bearing minerals contained in the code's thermodynamic  database
 were quartz and amorphous silica.  Furthermore, a particular model that
. has a larger database of aqueous species for * particular element will
 predict lower concentration of free ion (Nordstrom-etal,.l979a).-  Tfc	.
 model predicting higher concentration of free ion may be invalid.
       Species that are rare or absent in natural waters may be important
 in anthropogenically influenced systems, but thermodynamic data may not
 yet exist.  An example was suggested by Nordstrom et al (1979b)  who
 noted that data for ion triplets such as Fe(SO),' are  not accurately
 known and possible complexes  such as FeHSO" and Fe(HSO,), have not been
 properly identified.
       A model is "saturation sufficient" if the code database  is
  X
 suitably comprehensive to define saturation indices for a given set of
 plausible phases in the system (Plummer et al, 1983).   An incomplete
 database may predict an erroneous saturation index for a mineral, which
 could be propagated through the modeling effort.  They cite the two most

-------
                                                                       39
common examples  of saturation  insufficient data as the absence of
analyses for dissolved iron and aluminum.
      Recent studies show that organic matter may be responsible for
complexation of  up to 100% of some metals  in natural waters (Morel and
Hering, 1993).   Since geochemical code databases are mostly limited to
inorganic species, abundant organic complexation may render a  computer
simulation highly inaccurate.  The user should be aware of  the amount of
TDC  (total dissolved carbon) in the system,  and potential inaccuracies
that may arise through the exclusion of organic complexation.
      To add species to a computer 'code's  thermodynamic database,  the
user generally must provide values for AHf and K,.  Prior sections
demonstrated how these data are used to calculate speciation and
saturation indices.  The AH," and Ktp data can be obtained from published
sources.

Oxidation/Reduction (Rdox)
      A redox species is defined as. a  species of-any element which: .can
exist in more than one oxidation state in natural?aqueous-environments-  -:
(Parkhurst et al, 1982).   Examples are ferrous  (Fe**)  vs. ferric iron
(PeJ*), arsenite  (AsO,*')  vs.  arsenate (AsO,J'), and nitrate (NO/) v.
ammonia (NH/) .   Other elements with redox  chemistry include Cu,  Eg,  Ma,
S, and Tl.   Redox reactions usually are kinetically controlled  and many
are microbially mediated (Ball and Nordstrom, 1991) .
      Applying a field Eh to rigorous  geochemical problems  involving
redox can be risky,  since redox couples do not tend to  reach equilibrium
                      \
with each other  in natural waters (Lindberg and Runnells, 1984).  Redox
potentials measured in natural waters usually represent mixed
potentials.  Most system are likely in internal disequilibrium and
determining which couple  is most responsible for  the measured value will
be difficult without separate analyses for each component (Lindberg and

-------
                                                                       40
Runnells, 1984).  Aa and Se, and probably all oxyanions,  do not give
reversible potentials at a platinum electrode {Runnells and Lindberg,
1990; Runnells and Skoda, 1990).  Only dissolved iron,  dissolved
sulfide, and possibly dissolved uranium and vanadium are  likely to give
reversible potential measurements for a platinum electrode,  and then
only when the concentrations are high enough (Ball  and  Nordstrom,  1991).
The HS*/S(V'. can  be discounted  as  redox  controls because sulfate is not
involved in reversible redox reactions at  low temperatures  (Lindberg and
Runnells, 1984) .

Gaology
      Under normal circumstances, the chemical  composition  of  a
terrestrial water is directly controlled by rock/water  interaction in
the watershed or aquifer.  Exceptions might include anthropogenic
contamination, such as spills or agricultural runoff.   The  "major  ions"
will generally indicate from which type of  lithology/mineralogy a  water
has evolved, such-as-carbonate or silicate.   The major  ions most
commonly seen in solution from .weathering -of carbonate, and .silicate
rocks are:  Ca2', MgJ*, Na*.  K*.  Cl-, HCO,\  and SOa*.   Concentrations of
these ions in natural waters typically range from 10** to  10"a molal.
      Understanding the local geology will  be vital in  successfully
interpreting any model, even simple speciation  models.  Forward
modeling, in which future water chemistry  is predicted  after
interactions with minerals, will obviously require  detailed knowledge of
the minerals.  The same is true for inverse modeling, which requires
detailed knowledge of mineral mass transfer as  a water  body evolves
along a flow path.
      The solubility and thermodynamic behavior of  common minerals  will
aid an understanding of water chemistry.   The modeler must recognize
implausible minerals or reactions in simulation results.  A familiarity

-------
                                                                              /
                                                                      41
with the behavior of minerals and aqueous species will allow more
accurate interpretations of modeling results.  Examples are  seen in the
thermodynamic behavior of carbonates, sulfates, silicates, and sulfides.
      Carbonates dissolve congruently and exhibit reversible
dissolution/precipitation behavior.   They are fairly soluble, and have
relatively rapid kinetics.  Carbonates are generally responsible for the
bulk of calcium, magnesium, and bicarbonate in solution.
      Sulfates,  such as gypsum and anhydrite, are also very  soluble,
reversible,  and can contribute high concentrations of calcium and
sulfate to solution.
      Silicates are more difficult to model, since they dissolve
incongruently,  have a wide range of solubilities,  and do not exhibit
reversible dissolution/precipitation behavior.  As an example, Garrels
and Mackenzie (1967) discovered in the Sierra Nevada spring  study that
plagioclase weathers disproportionately higher than other silicates and
contributes the bulk of ions to solution.  Quartz and K-feldspar remain
as solid residues, eventually removed by mechanical weathering.  The
results of this study suggest that dissolved SiO,  in the Sierra spring
water,  and likely most natural waters,  comes from silicate weathering,
not quartz dissolution.
      Sulfides are fairly soluble and weather rapidly,  but may form an
oxidation product such as iron hydroxide, or dissociate during oxidation
and form free ferric iron, sulfate, and hydrogen ion.  Unlike most
carbonates and silicates, however,  the stability of iron species is
redox dependent.
      There are probably no groundwater systems that are in overall
chemical equilibrium with their host mineralogy (Plummer,  1984).   This
likely .applies to rivers and lakes as well.   Although more soluble
minerals like carbonates and sulfates may reach equilibrium,  most
mineral phases probably will not.  Therefore, the Saturation Index of

-------
                                                                      42
most minerals will be some distance from 1.0, indicating a tendency for
either dissolution or precipitation.  Most waters are generally
undersaturated with respect to the minerals of the local lithologies,
favoring continued mineral dissolution.

-------
                                                                      43
              3.   MODEL DEVELOPMENT  AND APPLICATION

      The objectives of chemical modeling may include any or all of the
following: to determine 1) what chemical reactions have occurred, 2) the
extent to which reactions have proceeded 3) the conditions under Vhich
the reactions occurred (open vs. closed, equilibrium vs.  disequilibrium,
constant vs. variable temperature), and 4) how the water quality and
mineralogy will change in response to natural processes and
perturbations to the system (Plummer, et al, 1983).
      The process of developing and applying a chemical model is
accomplished in a series of steps, similar to the development of
groundwater flow numerical models.  Development of the conceptual model
is followed by development and testing of the numerical model,
calibration, and validation.  Once these steps have  been successfully
completed, the model is ready for application.

Conceptualization
   .   The information desired about a hydrogeochemical system will guide
the development of a conceptual model on which to base the numerical
model.  Conceptual model development starts by first determining the
information desired from the model, and the input required to run the
model.
      Information desired;  Chemical modeling simulations  generally fit
into one of three categories - speciation,  inverse,  and forward.   The
                                 ,                   /
complexity of the model will be proportional to the  quantity  and quality
of information desired.
      A speciation model is basically a "snapshot* of a water sample at
a point in time,  and is the simplest type model to learn and  apply.   A
speciation model will be sufficient if the only desired information is
the distribution of chemical species, ionic balance,  saturation  indices.

-------
                                                                      44
or determination of possible mineral phases in contact with the water.
No reactions or mass transfer are modeled, and no predictions are made
of future water chemistry.
      If information is desired regarding chemical mass balance of
minerals and dissolved species between two points along a hydrologic
flow path (i.e. mineral dissolution/precipitation between two wells),
then an invars* model is required.. (Plummer et al, 1983).  inverse
modeling demands more complete and precise input data, as well as more
geologic and geochemical insight.   An example of a scenario requiring
           s
inverse modeling techniques is shown in Figure 3-1.  The desired
information is mass transfer between the two wells, which is determined
from the water chemistry of each and a set of hypothetical,  user-
specified mineral phases.  The mass transfer between well #2 and the
lake in Figure 3-1 could also be considered an inverse modeling problem.
      Included under mass transfer are all calculations for which there
is a recomputation of the distribution of species in response to changes
        Welltfl
               Flow
Figure 3-l>  Scenario requiring inverse modeling method* (from Glynn t
al, 1992).
in the composition, temperature and/or pressure  of the fluid (Plummer,
1984).  This includes mineral solubility,  dissolution, precipitation,
irreversible reactions in partial equilibrium systems, adsorption,

-------
                                                                      45
mixing of water, etc.  The mass transfer  calculations predict the
amounts of minerals transferred among aqueous, gaseous and solid phases
as a function of irreversible reactions and/or thermodynamic
constraints.
      Inverse modeling may require little more real data input than the
speciation models, but the mental input required is substantial.  Since
inverse modeling involves the determination of mass balance/mass
transfer, the user needs both comprehensive chemical analyses of the
waters, and thorough knowledge of the mineralogy of the system.
However, the determination of these mineral phases requires a level of
common sense regarding geologic processes and mineral stabilities.  As
stated in the NETPATH manual: "The validity of the mass-balance models
depends significantly on the geochemical  insight of the modeler in
selecting appropriate phases in the model."
      Inverse modeling is not constrained by thermodynamics,  and may
imply reactions that are thermodynamically impossible (Parkhurst et al,
1982) .  It may therefore be necessary to verify the mass balance
calculations by speciation modeling.   Inverse model results can be used
as input for forward models,  provided thermodynamic constraints are not
violated.  Plummer and colleagues have applied inverse and forward
modeling techniques to the Florida and Madison Aquifers (1983,  1984,
1990) .
      Inverse models may also carry errors through the simulation
unchecked.  For example,  if the ionic balance input to inverse  models
                                                              D
carries a significant error,  it may be carried into the mass  transfer
calculations, resulting in faulty molal transfers and erroneous
interpretations regarding the chemical evolution of the water.
      The results of an inverse model will be valid only if the  two data
sets are from the same hydrologic flow path (i.e.  along a groundwater
stream line, or two points on a river).   Attempts to model changes in

-------
water chemistry between two unrelated waters will be meaningless
(Plummer, 1984).
      Forward modeling is required if the desired information includes
predictions about water chemistry that might arise through chemical
reactions, biological activity,  ion exchange, adsorption,  or other
process contributing to the chemical evolution of a water body (Plummer,
1984).  Forward models generally have less data available than inverse
models, hence forward methods may be required in situations where an
inverse model should be utilized,  if only one analysis is available
along flow path.   This scenario  is illustrated in figure 3-2.   The
setting is identical to the inverse model,  but no second well  exists.
The desired information might be the groundwater chemistry at  the
location of Well  #2 from the previous example.
      Well#l
             Row
Figure 3-2t   Scenario requiring forward modeling methods
             (from Glynn et el,  1992).
      The most complex application of  a  forward model  is  the  reaction
path simulation.   Reaction path modeling is  designed to determine  the
chemical composition of an aqueous solution,  and the masses of minerals
dissolved and precipitated based  on a  set of hypothetical reactions and
        V
thermodynamic constraints imposed by the user.   Reaction  path models

-------
                                                                      47
require a thermodynamic model, an initial water composition, and an
assumed set of irreversible reactions and/or mineral-water equilibrium
constraints.  Using these, the model predicts the evolution of water and
rock as a function of reaction progress  (Plummer et al, 1983).  Reaction
progress is measured in terms of a progress variable, such as pH,
temperature, or the moles of a reactant dissolved into solution.  It
kinetic data are available, time may be chosen as the progress variable
(Plummer et al, 1983) .  A valid reaction path model, therefore, requires
detailed knowledge of mineral suites in the system, pH and redox states,
biological processes, and gas exchange.
      Most modeling exercises, both forward and inverse, will usually
generate multiple results.  To eliminate implausible results and isolate
the best model, the researcher must then draw upon geological and
geochemical knowledge, or in some cases geologic "common sense" if no
data are available.  Even then, only an approximation of the actual
system may be obtained.  As stated by Plummer et al (1983) :  "Rarely, if
ever, will the unique reaction which corresponds with reality be
isolated."
      Input required;   A speciation model requires a chemical analysis
for the water, and physical parameters such as pH, Eh and temperature.
Thermodynamic constants are required for each dissolved species and
mineral of interest, but are built into the databases of all speciation
codes considered in this study.  However, the thermodynamic database of
each code is different, and the researcher must be aware of the
differences and be prepared to modify thermodynamic data if necessary
and if the code allows.  An example of a small discrepancy in
thermodynamic data between codes is seen in the solubility constant for
quartz in PHREEQE  (log K^ = -4.0477) vs. WATEQP  (log K,, = -4.075).
Although this difference is minor and probably will not introduce
significant error in calculations involving quartz, the example

-------
                                                                      48
illustrates the potential differences that may be encountered in
thermodynamic data.
      Program output will include chemical species distribution, ionic
balance, and saturation states of all plausible mineral phases contained
in the software database.  Speciation modeling is often an integral
component of subsequent interpretation of inverse or forward simulations
(Plummer et al, 1983) .   The inverse program NETPATH runs WATEQF
speciation models for each water sample before performing any mass
balance/mass transfer simulations.
      The inverse model requires two additional bits of information
beyond the speciation model:  Water chemistry from a second water sample
along the same evolutionary flow path as the first,  plus a set of
plausible mineral and gas phases in contact with both waters along flow
path.  The specified mineral and gas phases will interact with the first
water to produce the second while satisfying mass balance among all
components.
      Sulfur and carbon isotopic information can also be incorporated to
help define mass transfer along flow path in groundwater (Plummer,  1984;
Plummer et al, 1983, 1990) .  Although mass transfer calculations can be
accomplished without isotopic data,  they provide an additional
constraint that helps eliminate implausible results.   The inverse model
assumes the effects of hydrodynamic dispersion are negligible (Plummer
et al, 1983).
      The output from an inverse model is a set of scenarios indicating
the possible mass transfer that occurred among the selected mineral
phases to generate the second water from the first.   Thermodynamic
constraints are not an explicit part of the mass balance methodology, so
it is usually necessary to check each model for thermodynamic violations
 (Plummer et al, 1983).   This can be accomplished by speciation
calculations at the endpoints of the flow path.  For example, an inverse

-------
                                                                      49
model may predict  dissolution  of  a mineral along flow path,  but
speciation models  at  each  endpoint show the mineral as oversaturated.
Either the inverse model has generated invalid results,  or the mineral
is undersaturated  somewhere between  the endpoints.   Examples of  inverse
model application  are provided by Plummer (1984)  and Plummer et  al
(1983, 1990).
      The most difficult aspects  of  inverse modeling are selection of
hypothetical phases,  and calibration of the model through  elimination of
implausible modeling  results.  The selection of plausible  mineral phases
requires detailed  information  on  the mineralogy of  the system, combined
with geologic "common sense."  The value of the mass balance
calculations is directly proportional  to the amount of analytical data
available (Plummer et al,  1983) .  The  modeler may be unable  to identify
all mineral phases present and reacting in the system (due to inadequate
data), which will  adversely affect the mass balance calculations, and
hence, the validity of the model.
      The forward  model is the most  demanding and involves the highest
degree of uncertainty, because less  information is  usually available
than for the inverse  model.  Forward modeling requires educated guesses
regarding reversible  and irreversible  chemical reactions that determine
the chemical evolution of the  water.   The forward model also requires
definition of initial conditions!(i.e.  a starting water sample), but the
'final conditions can  vary depending  on the location along  flow path at
which the simulation  terminates.  The  model of the  forward problem is
complete when all  appropriate  equilibrium or apparent  equilibrium
mineral-water reactions are included (Plummer,  1984).
      Plummer (1984)  discusses the importance of  adjusting initial water
input concentrations  to attain electrical neutrality.  This  step can be
ignored if the modeler chooses to maintain the analytical  integrity of
the initial solution.  The modeler must decide when it is  better to

-------
                                                                      50
leave the analytical data unadjusted,  in which  case the charge  imbalance
will be distributed among the computed mass  transfer coefficients.   For
this study, analytical data was left  intact  and charge imbalance was
carried into subsequent simulations.
      The output of a forward model is a prediction of water chemistry,
and perhaps an estimate of mineral mass transfer and mass distribution
generated as reactions occurred along flow path.  Unlike inverse model
results, forward model results will not contain violations of
thermodynamic constraints.  The selection of plausible results  will  be
based on information such as final dissolved concentrations, mass
distribution or transfer, or parameters such as pH or Eh.
      Inverse modeling is the method  of choice  when the necessary
information is available.  Forward modeling  is  regarded as a method  of
last resort when information is unavailable. Inverse modeling can also
be thought of as determining what has happened,  whereas forward modeling
predicts what will happen.
      The  ideal scenario for development of  a forward model is  when
information is available to allow calibration with an inverse model, a
procedure  that has been applied in this study.   If a reaction is found
through forward modeling that satisfies the  net mass transfer
constraints defined by the inverse model, then  the calculated path is
thermodynamically valid  (Plummer et al, 1983).

Th Numerical Model
      Development;  The development of the numerical model involves
gathering  and compiling data, and incorporation of these and other
parameters into the conceptual model  to give the simulation unique
characteristics representative of the field  situation.  This information
represents input  to the  computer modeling codes.
      What constitutes input  "data" may not  be  explicitly definable, and

-------
                                                                      51
may be subject to considerable user discretion.  Examples include the
selection of plausible mineral phases, estimates of  irreversible
reaction stoichiometries, ratios, and/or masses, or  selection of ion
sources and sinks.  Input data therefore will be a combination of actual
field data, and a set of user-defined variables or parameters that will
act upon, or be acted upon by, the actual data.
      Field data can include water chemistry, aquifer or wallrock
mineralogy, mineral percentages, precipitation/evaporation rates, and
perhaps groundwater flow rates.  The user-defined parameters might
include mixing ratios of different water parcels, irreversible mineral
dissolution masses, kinetic data, time frames, or minerals involved in
reversible equilibrium reactions.
      Both inverse and forward hydrogeochemical models will require
initial and boundary conditions, not unlike numerical groundwater flow
models.  The boundaries, i.e. the beginning and end  of the flow path,
depend on the information desired and the availability of data (Plummer,
1964).  The boundaries of an inverse model are strictly defined as the
initial and final points along the modeled flow path.
      The boundaries of a forward model may be less  rigorously defined,.
hence subject to higher uncertainty.  The initial condition is generally 
the chemical analysis for the initial water.  The final condition
depends on the objectives of the study, and may be varied through
sensitivity analyses.  For example, the initial condition in pit water
modeling is the chemistry of groundwater immediately upgradient of the
pit.  The final condition may be the pit water after a specific time
period, at a specified water level, or after a series of reactions have
occurred, or it might be groundwater at some point hydraulically
downgradient from the pit.
      During model development, many specific questions must be answered
which will influence the selection of modeling parameters and software.

-------
                                                                      52

Plummer (1984) posed the following questions in defining the inverse

model of the Madison Aquifer in the northern U.S.  These questions will

likely be applicable to development of any inverse or forward model:


*  What minerals are present and what are their abundances?

*  How does mineral abundance, including trace mineralogy, vary
   spatially in the system?

*  What is the actual composition (elemental substitution, exchangeable
   ions, etc.) of each mineral and how does this vary spatially?

*  What is the isotopic composition of the minerals and how does this
   vary spatially?

*  Are there any regional trends in mineralogy or composition that can
   be related to direction of flow?

*  From thin section or SEM examination, which minerals appear to be
   secondary and which are being replaced?

*  Is there evidence of coatings or zoned crystals?  And,  if so, how
   does the composition of the coating vary in the crystal:

*  Is the mineralogy of more permeable rocks in the system different
   from that in less permeable zones?

                                                                ~s
      A crucial step in forward model development is the recognition of

potential irreversible reactions (Plummer, 1984), such as:        -


*  Oxidation of organic matter, as during sulfate reduction.

*  Dissolution of minerals that rarely reach equilibrium in the ground
   water environment, e.g. primary silicates like pyroxenes, feldspars
   or trace minerals.

*  Gain or loss of gases in the system, such as methane, oxygen or
   carbon dioxide.

                         /
      Recognition of reversible reactions will be a key component of

either model.  Identification of mineral phases with which the water is,

or could be,  in equilibrium will help constrain the model and isolate

implausible results.

      Additional questions that may need answering to help define the

system  during model development include:

*  Is the system open to gases such as CO2,  02, CH4, in a reversible
   reaction?

-------
                                                                      53
*  What is the scale, both temporal and spatial, of the  simulation  (i.e.
   is the pit lake being modeled at incremental depths or at ultimate
   depth; what distance from the pit do we want to model)?
*  Do we need to incorporate reaction kinetics?
*  Do we need to model trace elements?

      Researchers must be careful not to overconstrain the model.  This
can be done by violating the phase rule,  or by specifying too many fixed
components or too few variables.  An example is provided by Peterson et
al (1987) for carbonate equilibria.  The variables are alkalinity, pH,
and pCOj.   If all of  these  are  fixed,  then a  forward simulation that
predicts changes to the system is not possible.  The execution may
terminate and indicate that a phase rule violation has occurred, or may
simply remove one of the variables.
      When all questions have been answered regarding model 
conceptualization, the level of sophistication required  from the
      /
software can be determined.  Since each program is designed for a
different purpose, each modeler should use a program based on the
research objectives.   The software are discussed in the next chapter.
                                                               "" 
      Execution:  The execution of the actual numerical model proceeds
in a series of steps.  Model calibration may require an  iterative
process based on the results of sensitivity analyses,  as illustrated
below:
            r
1)  Run the model.
2)  Interpret the results.
3}  Sensitivity analyses.
4)  Calibrate.
      Running the model is self-explanatory in the context of this
study, since only "canned" computer software packages are considered.
The only constraints are the availability of computer memory and
mathematical processing capability, provided the data have been entered
correctly into the model.

-------
                                                                       54
       Interpretation and Sensitivity Analyses;   Interpretation of
modeling  results will be site-specific,  and is  not a skill  that can be
acquired  by reading a manual.  The  user  must draw on a wide spectrum of
geochemical,  geologic, and modeling expertise to separate the  plausible
results from the implausible or  invalid.   The reader is again  referred
to Plummer et al  (1983) for examples of  model calibration.
       The selection,of the most  plausible result might be best achieved
through sensitivity analyses that provide a range of possibilities.
Sensitivity analyses will provide the combined  benefit of bracketing the
results within the window of uncertainty, as well as seeing the effect
of varying input parameters on modeling  output  as results move from
implausible to plausible within  simulations.
       The problem of excess plausible phases can be .complicated by the
existence of mineral phases of variable  composition,  solid  solution or
impurities that can change the mineral stoichiometry from that assumed
by ideality.  This exemplifies the  utility of field examination or
petrographic data for the geologic  system being modeled.
       Plummer et al  (1983) discuss  the difference between possible
 reaction paths and the net reaction path, which is illustrated in Figure
 3-3.  The net reaction path is depicted  by path 2 in each figure,
 whereas possible reaction paths  are depicted in paths 1, la, and 3.  In
 each case, PC  (Polk City) is  the initial water  and H (Wauchula)  is the
 final water.  Relative rates  of  dissolution/precipitation may  cause
                                       %
 curved paths, but the net reaction  path generally is a straight line.
 The actual reaction path may  be  definable through small incremental
 steps in reaction simulation.  This type of reasoning may be required in
'many modeling exercise  to eliminate implausible results.
       There are several  ways  of  eliminating implausible reaction models
 from  further consideration.   First  the computed mass transfer  should be
 consistent with the saturation indices {Plummer 1984).  Petrographic or

-------
                                                                       55
SEM data, plus isotopic data can help eliminate implausible models.
'0.0   0.4   0.8   12   1.6
     SULf ATE. IN MMOL/L
                                        0.0   0.4   OA   12   1.8
                                            SULFATE, N MMOA
Figure 3-3:  Actual vs. net reaction path.   Predicted variation of
magnesium, calcium, pR, and 513C in ground water between Polk City  (PC)
and Wauchula (W)  (from Plummer  et  al,  1983).
      Conditions that may limit  the  success  of  chemical modeling are

(Plummer, 1984):


* Fracture flow, causing different residences times at different
  locations in  the aquifer.

* Vertical mixing from either leakage or  recharge,

* Chemically stratified flow systems (changes in chemistry with depth),

* Groundwater systems that have  been altered hydrochemically such as
  through injection.

                                                               f

      Calibration:  Calibration  is the process  of adjusting specific

input parameters, such as initial or boundary conditions, and rerunning

-------
                                                                      56
the model in an attempt to converge on the desired  result.  Calibration
may be an iterative process requiring several  loops through steps  1 to 4
to attain convergence.  Calibrating an inverse model can be more
difficult than a forward model, because in the inverse model, the  final
condition is known, and any model which produces different results is
obviously invalid.  If model calibration is not possible due to faulty
input data, then more data collection and compilation may be necessary.
      In the forward situation, the modeler generally has a less
definitive idea of what to expect in the final condition.  Calibration
may not be so much an attempt to obtain a final result, but to refine
the output to bracket more believable geochemical values.
      The success of calibration depends on the availability of real
world examples with which the model can be compared.  Not all forward
models will have the luxury of an example location,  in which case.the
best educated guess may be the only reference  with  which the model may
be compared.  Pit water modeling is a case in  which a-limited number of
real world examples exist for comparison.
      When a model is calibrated, the input of a certain combination of
parameters and boundary conditions will reproduce field measured data as
output  (Hang and Anderson, 1982) .  However, the results may not be
unique, and multiple models may remain after all tests have been
exhausted.  In this case, the problem is non-unique and will remain so
until appropriate new data are introduced  (Plummer,  1984) .
      Verification/Validation;  The final test of a model is to
determine whether  it successfully simulates field observations.  When a
numerical groundwater flow model meets these criteria, it is said to be
calibrated and verified  (Wang and Anderson, 1982) .   The goal of
numerical model verification is  to demonstrate that the model can
simulate some historical hydrologic event  for  which field data are
available.

-------
                                                                      57
      The process of verification and validation applies to chemical
models as well.  Peterson et al  (1987) refer to validation as "the
coherence, to some acceptable accuracy, of laboratory and field data,*
and to verification as meaning that "the coding and mathematical
algorithms in the computer code were certified to be correct.*  Models
may be verified by comparing the results obtained from the code with
results from other codes, or the computations could be checked by hand
(Peterson et al, 1987).
      Once the model has been calibrated and verified/validated', it is
ready for application.

-------
                                                                      58
                             4.   SOFTWARE

      This chapter discusses the software codes that were used in this
study for application of the pit water models  (BALANCE, MINTEQA2,
PHREEQE, WATEQF, and WATEQ4F).  Each code is designed  for a different
purpose, and therefore each has different input requirements,
capabilities, and limitations.  The output  files generated by each code
vary in complexity and length, both of which increase  with the number of
functions the program is asked to perform.
      Speciation modeling codes are discussed first, .followed by the
inverse, then the forward modeling codes.   The order presented also
parallels the difficulty of use,  and the variety of functions each code
can perform.  The user friendliness of a code is generally inversely
proportional to its capabilities.
      Basic input:  All simulations require as input the concentrations
of dissolved ions obtained from chemical analyses.  Most codes require
temperature and pH, although PHREEQE can calculate pH  depending on the
concentrations of other components (alkalinity, pCO,)-;   If no
temperature is provided,  codes will generally default to 25C.   If the
problem involves redox, either dissolved oxygen or Eh/pe must be
specified, or the concentrations of separate redox couples must be
provided from which the code can calculate the Eh.
      If analytical data for a particular ion are not available,  no
speciation or saturation calculations will be performed for any species
or minerals of which the ion is a component, and that data will be
missing from the output.  This is a critical concept.  Missing
analytical data for particular ions will introduce deficiencies in the
modeling of speciation and mineral saturation indices.
      Database Limitationsi  The size of each program's thermodynamic
database will limit its effectiveness in assigning a valid and complete
suite of aqueous species and minerals to the modeled system.   This may

-------
                                                                           59

be  the most significant limiting  factor among  the various codes.   Figure

4-1 shows a comparison of the different databases for several  codes.

WATEQF has a much smaller database  than WATEQ4F,   The EQ3/6 and MINTEQA2

databases exceed  all  others,  containing a large collection of  trace

metal thermodynamic data appropriate  for mine-related modeling

applications.
                  rxxorrrz
                         SOLM1MEQU
SOLVTEO
CKILLEX
                                         RACT
                                               WATEOF
                                                           WATSOX
                                                                E
-------
                                                                      60
are only designed to calculate activity coefficients,  ionic balance,
chemical speciation, and saturation states  (including  gas partial
pressures) of a water analysis.  The only input data required are  field
parameters (temperature, pH, and solution density) and concentrations of
dissolved aqueous components  (expressed as  total calcium, sodium,
sulfate, alkalinity, etc.).  Optional field parameters include pe/Eh,
dissolved oxygen, and conductivity.
      HATEQF and WATBQ4F:  The WATEQ codes  of the USGS,  for which  WATEQF
and WATEQ4F are the latest versions, are among the most widely used
geochemical modeling programs.  They have easily understood menu-driven
input packages, and provide clear, concise  output files.  They can be
applied to stand-alone speciation studies,  or incorporated into more
comprehensive studies to help interpret forward or inverse models.
      The original WATEQ was written by Truesdell and  Jones in 1973
 (published by USGS in 1974, Journal of Research) in PL-1 (Programming
Language/One).  WATEQ contained a thermodynamic database consisting of
                                           i
22 master species, 100 aqueous species, and 56 minerals.  Plummer  et al.
 (1976) translated the PL-1 version into FORTRAN IV  (WATEQF, USGS Water-
Resources Investigations 76-13), and made minor revisions including
addition of manganese species and minerals.  WATBQ2  (Ball et al, 1979)
 incorporated  10 additional trace elements and many additional complexes
and minerals.  USGS publication WRI 78-116  (Ball et al., 1980) intro-
duced some revisions and corrections to WATEQ2, and WATEQ3  (Ball et al,
 1981) added uranium species.  The latest version, WATEQ4F  (Ball and
Nordstrom, 1991), contains a  thermodynamic  database with 32 master
 species and over  600 aqueous  species and minerals.
      The current version of  WATEQF  (Plummer et al,  1984) is called
 program WATEQF.PATH, which  creates input  files  for  the NETPATH program.
 WATEQF  first  calculates activity  coefficients,  speciation, ionic
 balance,  and saturation  indices of each water  analysis. WATEQF include^

-------
                                                                      61
provisions for entering isotopic analysis to be used  in mass transfer
calculations.  The files are then available to be loaded into NETPATH
for mass balance calculations in the inverse model.   The greatest
shortcoming of WATEQF in the context of mine water modeling is the
absence of any trace metal data other than iron.  WATEQF is therefore
inadequate for application to speciation modeling of  mine waters
containing trace metals beyond iron.  WATEQ4F has a larger thermodynamic
database, with many trace elements applicable to mine water modeling,
including arsenic, cadmium, copper, lead, nickel, silver, uranium, and
zinc.  Neither WATEQF or WATEQ4F contain thermodynamic data for mercury
species.
      WATEQ4F has greater flexibility in handling redox problems than
WATEQF.  WATEQ4F has 9 separate areas in which redox  calculations are
applied, and 14 means with which to calculate or input an Eh value (Ball
and Nordstrom, 1991).  A useful application of WATEQ4F is to verify
field Eh measurements by calculating redox potential  from the
concentrations of each component in a redox couple (e.g. Fea*/Fe3*, or
NH//NCV) .  WATEQ4F  can calculate the system pe/Bh from any of several
redox couples, then can redistribute the remaining redox couples bases
on the calculated pe/Eh.
      The WATBQ codes can calculate speciation in water samples ranging
in temperature from 0 to 100 C.  However, thermodynamic solubility
                               %
constants are specified for 25C, and adjustments to  K^ values via the
Van't Koff equation for departures from 25C increase the uncertainty in
the modeling results (Ball and Nordstrom, 1991).
      WATEQF and WATEQ4F compute charge imbalance by  the following
formula:

          (Sum of Cation Species - Sum of Anion Species)
 &t x   	:	  * 100
          (Sum of Cation Species + Sum of Anion Species)/2

-------
                                                                      62
      WATEQF provides an error message if the ionic balance exceeds  30%,
and asks if you wish to proceed anyway.  As stated earlier, it  would be
unwise to proceed with an error of that magnitude, since it usually
indicates an error or omission somewhere along the line from  sample
collection to data input.  WATEQ4F terminates the run and provides an
error message if charge imbalance exceeds 30%.
      WATEQF uses the Robinson-Stokes Debye-Huckel equation to  calculate
activity coefficients for Ca", Mg2*, Na*. K',  cr, SO43', COj*', and HCO/.
The user has the option of selecting either the Debye-Huckel  equation,
or the Davies equation (c  0.3) for all other activity coefficients.
      WATEQ4F does not allow the user to choose between methods for
calculation of activity coefficients.  The extended Debye-Huckel
equation (equation 3) is used for polysulfide species, carbonates, H',
and Sr species.  The Robinson-Stokes Debye-Huckel equation  (equation 4)
is applied to those species for which the b parameter is available,  and
the Davies equation .(c  0.3) is used for calculation of all  other
activity coefficients.
      The thermodynamic database of both WATEQF and WATEQ4F are hard-
coded, meaning that they are part of the source code, and no  additions
or corrections can be made to the database without recompiling  the
source code.

Inverse Modeling Codec
      Inverse modeling is the calculation of net geochemical  mass
transfer reactions between an initial and final water along a hydrologic
flow path  (Plummer et al, 1991) .  BALANCE and NETPATH are the only codes
evaluated in the study with inverse modeling capability.  These codes
require chemical analyses from two different water samples along the
same evolutionary path, plus plausible phases with which the  water
reacts to generate mass transfer dissolution or precipitation products.

-------
                                                                      63

These plausible phases generally are mineral solids, but may also

include gases,  ion exchangers, or (if a mixing problem is being modeled)

other aqueous solutions (Parkhurst et al, 1982).  The output produced by

an inverse model will be the mass transfer that occurred, in terms of

molality of components added to or removed from solution, between the

initial water and the mineral phases to produce the final water

composition.

      BALANCB:   BALANCE was developed in 1982, and was designed to help

define and quantify chemical reactions between ground water and minerals

(Parkhurst et al,  1982).  The program calculates the amounts of phases

entering or leaving the aqueous phase (mass transfer)  to account for the

changes in chemical composition between two solutions along the same

hydrologic flow path.  The purpose of the program is to derive balanced

reactions of the form {Parkhurst et al,  1982) :
                       /

          Initial solution + Reactant phases -->
                                  Final solution + Product phases


      BALANCE is designed specifically for mineral-water interactions,
                                                       f
but can solve any set of linear equations formulated by the user

(Parkhurst et al,  1982).  This includes: 1) mass balance on elements, 2)

mixing end-members waters, 3) oxidation-reduction reactions, and 4)

simple isotope balance.  Examples of each are provided in the BALANCB

manual (Parkhurst et al, 1982).

      The primary advantage of BALANCE lies in the ability of the user

to manually add elements and minerals to the database.  BALANCB is well

suited for inverse modeling in mining environments, since the user can
                                             'i
add trace metals such as arsenic, mercury, zinc, or others.

      NZTPATHt  NETPATH is a revision of BALANCE, and offers

improvements in the construction and management of input and output

files.  Isotopic fractionations can be incorporated more easily in

NETPATH.  The ability is retained to compute mixing proportion of two

-------
                                                                      64
initial waters and net geochemical reactions that can account for  the
observed composition of a final water.  NETPATH also allows
incorporation of evapoconcentration in the determination of mass
transfer.
      NETPATH has no provision for manually entering elements, such as
trace metals, into the calculations.  Only new minerals for which
elements already exist in the database can be defined.  Unfortunately,
NETPATH contains no trace metal data beyond iron.  This shortcoming
renders NETPATH inadequate for mass balance calculations on mine waters
containing any trace metals other than iron.  NETPATH could become an
important tool in determining mass transfer in pit wall dissolution if
these capabilities were incorporated.

Forward Modaling Codes
      Forward modeling can range from a simple equilibrium simulation to
one in which chemical evolution is followed as a function of reactions
with a suite of minerals.  Forward modeling, may also involve predicting
the evolution of the water down a hypothetical flow path which  -  '"- -  '-.
encounters a number of different processes and environments.  This type
of modeling is known as reaction path modeling.
      The variables that may be encountered along the flow path can
become numerous and complex, making the modeling effort difficult and
subject to multiple interpretations.  Reaction kinetics, adsorption, gas
exchange, biologic activity, and many other processes may influence the
                         1         '--*.,     '   i
chemical evolution of a water body along flow path.
      MINTBQA2:  MINTEQA2 is an equilibrium speciation software package
with the largest thermodynamic database of all codes considered in this
study.  The program can calculate ion speciation, solubility,
adsorption, oxidation-reduction, gas phase equilibria, and
precipitation/dissolution of solid phases  (Peterson et al, 1987) .

-------
                                                                      65
MINTEQA2 can accept a finite mass for any solid considered  for
dissolution.  The code contains many trace metals of interest in mining,
including arsenic, cadmium, cesium, chromium, copper, mercury,  lead,
selenium, silver, thallium, and zinc.
      The original MINTEQ was developed at Batelle Pacific Northwest
Laboratory by Felmy et al  (1984).  MINTEQ combined the thermodynamic
database of WATEQ3 with the mathematical structure of MINEQL  (Schecher
and McAvoy, 1991).  The latest edition of MINTEQA2, version 3.11 was
published December 1991, and incorporates the input file generator
PRODEFA2 version 3.11.
      For the level of sophistication it provides, MINTEQA2 is  the most
user friendly of the forward codes.  The interactive file generator,
PRODEFA2, allows easy construction of input files for both forward and
speciation models.
      MINTEQA2 solves multi-component chemical equilibrium problems much
the same way as other codes, by simultaneous solution of the nonlinear
mass action expressions and linear mass balance relationships.  MZNTEQA2
uses the mass action expressions to modify the mass balance equations -_
into the form necessary for the calculations.  This procedure is
illustrated by Peterson et al (1987) .  The user must be aware that
MIKTEQA2 uses formation constants rather than dissolution constants.
      MINTEQA2 performs a computational loop of iterating to
equilibrium, checking for precipitation or dissolution, and shifting
mass between the aqueous and solid phases until equilibrium is achieved
and there are no oversaturated "possible" solids and no undersaturated
"existing" solids.  The reader is referred to the manual for definitions
applied to various types of variables  (e.g. solids) in the code.
MINTEQA2 uses the Newton-Raphson approximation method to refine
estimates within each iterative loop.  Figure 4-2 is a flowchart
diagramming the procedural loop MINTEQA2 follows in solving a chemical

-------
equilibrium  problem.

       MINTEQA2 provides two options for  calculating activity

coefficients.   If  the user selects the Robinson-Stokes  Debye-Huckel
          0t
       Mnipullion '
                                    Tmp.0fiy
                       Unit*     KMC  Dil*ctric ConM,
                     Moltlity
                               Kt
                                    0by-Huckl A ( 
                        (log K it nw tmp<*tur|
                 Initul Activity
                 GIMM
          TM*I Inorginic
            Crben
               (from <*ocilwn conilims)
                                       Cemput* Ntw
                                       Componnt Aelivitit* 5OIVE
                                         ..= -
             *> Swtngth
             l.0.8Inuf
       ACTVTY *tivity Co.r
       ACTVlr IDivwtorO-H)
Sohrt Mtrw
(e*uiti*n hminciioA  SIMQ
                          Equilibrium frobtem
                                         *ub*Mutien)

                                     CompuM J
-------
                                                                      67
equation as  the  "modified Debye-Huckel equation."   MINTEQA2  uses  a
slightly different version of the Davies  equation,  where  the last term
is 0.241  (Allison et al, 1991).  MINTEQA2 starts the iterative process
by estimating the activities if none is provided.   The concentration of
each component is divided by 10 to obtain an  initial activity guess.
      MINTEQA2 offers some flexibility in the data  input  requirements.
A measured value of pH or pe may be specified as fixed, or MINTEQA2 can
calculate equilibrium values.  As with HATEQ4F, MINTEQA2  can calculate
the system pe/Eh from a variety of redox  couples, then can redistribute
the remaining redox couples based on the  calculated pe/Eh.   Also,  a
mineral may  be specified as presumed present  at equilibrium,  but  subject
to dissolution if equilibrium conditions  warrant, or definitely present
at equilibrium and not subject to complete dissolution (Allison et al,
1991) .  The  ionic strength can also be fixed  or computed.  MINTEQA2
                                       /
offers useful options in the manipulation and variation of pH, pe, and
in controlling the influence of gases.
      MINTEQA2 designates solid phases as either possible, finite, or
infinite.  The user specifies infinite phases and the amount present -..
(moles) .  The solid may then dissolve if  equilibrium conditions warrant,
up to the total amount specified.  Finite solids are also user defined,
and are available for complete dissolution up to equilibrium, as
solution thermodynamics dictate.  Both solids are redesignated as
possible solids if they dissolve completely,  in which case they may
reprecipitate if they become over saturated.   If dissolution is desired
beyond the equilibrium concentration, the solid must be "hand dissolved*
(Peterson et al, 1987}, in which the components of  the solid are entered
                                           \
as Type 1 components in PRODEFA2.
      The modeler may allow mineral precipitation if oversaturation
occurs, or they may be excluded from precipitation.  MINTEQA2 contains a
sweep option, in which a range of values  or concentrations can be

-------
                                                                      68
entered to evaluate the effect on the system from the perturbation.
This option is useful in sensitivity analyses.
      Expansion or revision of the thermodynamic database is easier  in
MINTEQA2 than in any other code.  The program is not  hard-coded, and
addition of components, species, and minerals is achieved interactively
through PRODEFA2.  These additions may be added  to the permanent
database, or simply included for the current  problem  being executed.
      Adsorption Model*:  One of the most attractive  features of
MINTEQA2 is the incorporation of adsorption models, including a limited
amount of surface complexation thermodynamic  data.  Seven adsorption
models are available:
                    \
            1)  Activity K,,
            2)  Activity Langmuir
            3)  Activity Freundlich
            4)  Ion exchange
            5)  Constant capacitance
            6)  Triple-layer
           . 7)  Diffuse-layer

      Only one adsorption- model may be chosen per simulation, but within
that up to five different surfaces (i.e. adsorbent, mineral phases such
as ferric hydroxide, or manganese hydroxide)  may be defined  for a single
program execution, with up to two types of sites per  surface.  This
capability is consistent with experimental data  published  on adsorption
surfaces such as hydrous ferric oxide (Dzombak and Morel,  1990),  which
appears to possess two different sites with different  surface energies
.and different adsorptive capacities.  .  -.
      The user must provide information regarding site density,  specific
surface area, adsorbent concentration,  and surface potential.  The
definition of these variables distinguishes one  adsorption model  from
another.  There is no intrinsic difference within MINTEQA2 that
distinguishes one surface from another,  nor one  site on a  surface from
another  (Allison et al, 1991).

-------
                                                                      69
      With the exception of one auxiliary input file for the diffuse
layer model,  the authors of MINTEQA2 have chosen to omit thermodynamic
constants for adsorption reactions, and leave the selection of them to
the discretion and problem-specific knowledge of the user.  They chose
this route because natural adsorbent phases often occur as mixtures of
impure amorphous substances that vary widely in chemical behavior among
sites (Allison et al',  1991).
      The large database and adsorption capability make MINTEQA2 a very
useful tool for mine water quality modeling.  The primary shortcoming of
MINTEQA2 is its limited ability to model reaction path geochemical
processes.  The program cannot dissolve ions into the pit water to
concentrations beyond equilibrium.  Once equilibrium is reached with
respect to the most solubility mineral thermodynamically plausible, the
dissolution process stops.  If a higher concentration is desired, the
user must "hand dissolve" the minerals.
      PHRBBQE:  PHREEQE (PH-REdox-EQuilibrium Equations) is designed to
model geochemical reactions, and can calculate pHj. redpx potential, and.
mass transfer as a function of reaction progress (Parkhurst et al,  -  .._-,-
1980).  In addition to most capabilities of the speciation codes, the
program can also determine the composition of a solution in equilibrium
with multiple phases.
      PHREEQE can simulate addition of reactants to a solution, mixing
of two waters, and titration of one solution with another.  In each of
                                                         I
these cases,  PHREEQE can simultaneously maintain the reacting solution
at equilibrium with multiple phase boundaries (Parkhurst et al, 1980).
The program can perform a sequence of simulations in a single computer
run.
      PHREEQE allows the entire reaction path to be modeled in one input
data set.  The building of this data set can be a long process if the
user does not have a firm grasp of the desired reactions from the start.

-------
                                                                      70
Furthermore, a mistake in one of the early steps will be propagated and
magnified through each remaining step until the simulation runs  its
course, likely rendering the entire simulation useless.  During  reaction
simulations, the program calculates pH, pe, total dissolved
concentrations of ions and species, the mass transfer of phases  between
the aqueous, solid, and gaseous phases, and saturation indices of  all
plausible minerals.
      The main disadvantage of PHREEQE is that the database is
significantly smaller than the MINTEQA2 or either WATEQ database.
However, the advantage of PHREEQE is that the elements, species, and
mineral phase databases are external to the computer code  (not hard-
coded) , and are easily modified or expanded.  The aqueous model  is
completely user-definable with respect to elements and species,  and
components are easily added or revised.  For this study, 8 elements, 113
aqueous species, and 130 minerals were permanently added to the  PHREEQE
thermodynamic database.
       PHREBQB's database can theoretically be expanded well beyond the
size  specified in the manual, to a level exceeding other codes with
larger databases  (Parkhurst, personal communication).  This cannot be
done  with the assistance of the input file generator (PHRQINPT), and
requires editing of the databases and source code, and recompilation.
       PHREEQE draws from three equations to calculate activity
coefficients:  the extended Debye-Huckel equation  (Equation 3),  the
WATEQ Debye-Hflckel equation  (Equation 4), or the Davies equation
 (Equation 5).  The non-linear equations are solved using a combination
of two techniques:  (1) a continued fraction approach, as in Wigley
 (1977),  is  used for mass balance equations, and  (2) a modified Newton-
Raphson technique is used for all other equations.  The reader is
 referred to the PHREEQE manual for thorough discussions of these methods
 (Parkhurst  et  al,  1980) .
                                  \

-------
                                                                      71
      PHREEQE has Che ability to dissolve masses of solids into  solution
well beyond system equilibrium.  This represents an advantage over
MINTEQA2, which cannot perform this function automatically.  PHREEQE can
compute the amount of irreversible reaction required for the solution
composition to reach the intersection of an assigned phase boundary,
with or without the inclusion of other mineral-water apparent equilibria
(Plummer, 1984).  PHREEQE can model mixing of two waters at any
specified proportion, a potential advantage in pit water models
requiring simulation of mixed conditions.  A comparison of PHREEQE and
MINTEQA2 is shown in figure 4-3.
      The major distinction between the reaction path capability of
PHREEQE, and true reaction path codes such as EQ3/6, is that PHREEQE
solves for the solution composition and mass transfer only at requested
points in reaction progress (Plummer, 1984) .  The disadvantage of
increased user manipulation is countered by better computational
efficiency.  Plummer (1984) cites a comparison of PHREEQE and EQ3/6
performed by INTBRA  (1983) that gave identical results when using the
same aqueous model and thermodynamic data.    	                 -
      To solve solution chemistry problems, PHREEQE uses equations
representing the following:

*  Total masses of each element in the system.  The total concentrations
   of the elements must be known for PHREEQE to begin any calculation at
   a reaction increment.  The total concentration must satisfy mass
   balance.
*  Mass action equations for ion pairs.  These are usually represented
   by formation constants, also referred to as equilibrium constants.
*  Electrical neutrality.  PHREEQB can adjust the pH of the system to
   bring the solution to electrical neutrality.
*  Phase equilibria.  An additional equation is required for each
   mineral added to the system, and is provided by the solubility
   product constant for the mineral.
*  Conservation of electrons (for problems involving redox).  PHREBQB
   keeps track of those species whose valence can change over the range
   of pe-pH conditions covered by the chemical stability of water.

-------
                                                                      72

Capability to irreversibly add or suteiract a
oet stoichiometric reactioo, in specified or
equal-increment step*?
Capability to ma two waten or to titrate ooe
witn the other?
Capability to add a oet sioichiometric
reaction umfl a mineral phase boundary a
reached?
Capability to change equilibration
temperature in equal or specified increment
steps?
Automatic charge balancing with a specified
eatioo/anion or wiib pH?
Adsorption, Surface compilation and loo
exchange?
Only predpitate solids if supersaturated?
Fa activities of given species (1 per
component)?
Exclude given speaesAninerab?
(excluded specka wflj cause an inconsistent
thermo. database)
SoBd-aolution Aqueous-sotutfoe equilibria?
Density correction? .
Maximum number of components that can
be entered in a given simulation?
1/uhI.ivmhtefn MMbv9

Capability to use toe solution made in one
problem a Input tor the oca?
Input program?
Graphical output?
Spreadsheet output?
PHREEQE
Yet
(extensive)
Yea
Yet
Yei
Yet
Ho
(PHREEQM does ioo exchange)
No
(PHREEQM does it)
No
Yet
No
Yet
30
Y
Ye
Ya
No
No
KONTEQA2
No
(except dissolution to equilibrium of 
fixed quantity of solid)
No
No
No
(separate problems must be ran)
No
Ya
Yes
Ye
Yet
No
No
35
(database has more, but code must
be recompiled)
Ye
No
Yet
No
Yd
Figur* 4-3s  Comparison batwaen MINTBQA2 (varion 3.0)  and PKREBQI
(1990)} from Olynn t 1, 1992.

-------
                                                                      73
      Limitations:  PHREEQE cannot remove  minerals from solution via
precipitation and adjust the resulting  solution concentration,  unless
the mineral is specified as a reversible reaction.  This procedure could
become tedious if PHREEQE were used alone  to model such reactions, which
is why MINTEQA2 was incorporated in this study for modeling
precipitation reactions.
      PHREEQE includes equations for  charge balance and conservation of
electrons, but mass balance constraints are not imposed on O and H  (i.e.
the model assumes a constant mass of  water) .   This assumption  can lead
to errors in modeling reactions involving  hydration and dehydration of
minerals, and redox conditions near or  beyond the stability of water
 (Plummer, 1984) .  The only constraint on H2 and O, are equilibrium and
electron balance constraints, so there  are no limits on the amounts of
H, or O, that can be made or destroyed in the computations performed to
satisfy the reaction constraints of a simulation.  If the masses of H,
and 0, involved in chemical reactions become  significant relative to 1
kilogram of water, then the simulation  may start to deviate from reality
 (Parkhurst et al, 1980).  The assumption is valid as long as the mass of
water involved in heterogeneous and homogeneous reactions is small
relative to one kilogram of water  (-55.5 moles; Plummer et al,  1983).
The error introduced when modeling natural waters is usually negligible.
      A more significant problem occurs if PHREEQB is used to  model
 systems in which large amounts of water are involved in mineral
precipitation or dissolution, such as might be encountered in  brines.
 For example, precipitation of 1 mole  of natron (Na,CO,  10H3O)  from 1
 liter of  solution would remove  10 moles of H,0 from the  aqueous phase,
 causing an  increase  in concentration of all remaining constituents in
 the  solution.  The  increase  in  concentration would be about 20 percent,
 but  would not  be  taken  into  account  in PHREEQE's present computation
 system.

-------
                                                                      74
      PHRQPITZ:  PHRQPITZ is designed for modeling highly concentrated
solutions such as brines.  PHRQPITZ is basically identical to PHREEQB in
the general format for input and output, with two major exceptions:  1)
PHRQPITZ incorporates the Pitzer parameter ion-interaction formulas to
calculate ionic activities, and 2) the database is restricted to species
and minerals (e.g. trona, mirabilite, etc.) typically found in brines.
To revise and expand PHRQPITZ to allow mine water modeling would be more
painstaking than the effort required for PHREEQE, so the user would be
better advised to revert to PHREEQE for such a project.

-------
                                                                      75
              5.   PIT WATER MODELING  CONSIDERATIONS

      There are many factors that can influence the chemical evolution
of pit water, and which require varying degrees of attention when
modeling the system.  Some variables are obviously more important than
others, and many are interrelated.  A change in one variable can affect
several others.  These considerations guide the model conceptualization
and must be factored into the numerical model development.  The
recognition and/or definition of these variables basically define the
numerical model.
      Figure 5-1 shows a cross section of a hypothetical pit lake,  and
illustrates some of the many factors and processes that will influence
pit water chemical evolution.  The factors can be categorized generally
as either physical or chemical.

Chemical Factors

      Classification of Deposit:  The genetic classification of the
mineral deposit will allow broad generalizations on the type of water
quality expected at the mine site.  Deposits are generally classified by
type of ore, morphology of the ore,  and type of mineralogy or lithology
in which the ore was deposited (host rock or wall rock).   Examples
include carbonate hosted disseminated gold deposit,  porphyry copper
deposit, stratiform massive sulfide, and stratibound base metal deposit.
      If the ore host is carbonates, acid mine drainage is unlikely,  and
if the ore is disseminated, high quantities of metal sulfides may be
unlikely.  However, if the ore is porphyry copper or massive sulfide,
both high metals and acidic waters may result,  as was seen in Table 1-1.
      The most common type of mineral deposit in Nevada is sediment-
hosted, disseminated gold deposit (Bonham,  1991).   Deposits  hosted  in
other lithologies,  such as volcanic or metamorphic rocks,  comprise  a

-------
                               Adapted from A. Dwis (1992)
5-1t   Cromm Motion of  hypothetical open pit.
                                                                             -o
                                                                             o\

-------
                                                                       77
smaller subset of disseminated precious metal  deposits.
      Ore may also be categorized depending on the  extraction and
processing techniques required, such as mill-grade,  refractory,  or
leach-grade, which may refer to sulfidic, carbonaceous, oxide or
siliceous ore.
      Nallrock mineralogy:  No single factor will have greater bearing
on the bulk chemistry dissolved into the pit water  than the  composition
of the rocks in the pit wall.  Mineral deposits commonly consist of  four
major mineralogic/lithologic suites:

        (1)  Host rock or wall rock    (3) Ore minerals
        (2)  Gangue minerals           (4) Alteration minerals

      Host rocks for most of the precious metal mines in the Great Basin
are Paleozoic age (240-570 million years old) sedimentary rocks,  such  as
limestone, dolomite, siltstone, and shale.  Tertiary age (2-63 million
years old) volcanic rocks comprise the second most  abundant  host rock.
Of 107 bulk-mineable precious metal deposits in Nevada, 77 are hosted  in
sedimentary rocks, 27 in volcanics,  and 3 in plutonics (Bonhara,  1991).
       \                                                  *
Porphyry copper deposits, by definition,  appear in plutonic,
porphyritic, igneous rocks.  The Ruth and Yerington  districts  are
examples of porphyry copper deposits in Nevada.
      Gangue minerals are those that were deposited by ore-forming
processes cogenetically with ore minerals, but have no economic value.
The most common gangue minerals are silica,  silicates, and carbonates,
and to a lesser extent oxides,  fluorides,  and sulfates (Guilbert and
Park, 1986).  Gangue can also include sulf ide*B of accessory  trace metals
deposited with the ore,  such as pyrite (FeS,),  arsenopyrite   (FeAsS),
orpiment  (As,S,), realgar (AsS), stibnite  (Sb,S,), and others.  A  gangue
mineral at one mine may be an ore mineral at a different mine.
      Ore minerals are those which contain the elements of economic
interest,  such as gold,  silver,  copper,  lead,  or zinc.  Examples of ore

-------
                                                                       78
 minerals include electrum (Au-Ag mixture),  gold-tellurides  (Au + Te),
 argentite (Ag,S) ,  chalcocite  (Cu,S) , sphalerite  (ZnS), galena  (PbS),  and
 pyrite.
       Alteration minerals are those that have undergone changes in
 composition as a result of physical or chemical means, especially by
 hydrothermal fluids (Guilbert and Park, 1986).  Alteration minerals  in
 hydrothermal deposits are generally clays,  including kaolinite, illite,
 sericite, chlorite, and micas.
       Depending on the site-specific characteristics of the deposit,
 dissolution of host rocks and gangue will likely contribute the majority
 of the major ions to solution, while trace metals could be contributed
 by both ore and gangue.  Alteration minerals, being primarily insoluble
. aluminosilicate clays, will contribute a small mass of major ions, and
 possibly trace metals.  They could also modify the pit water chemistry
 through ion exchange.                              '
       An illustration of the variety of water chemistries that can
 evolve from different host rocks is seen in the analytical data of
 Tables 1-1, 1-3, and 1-4.  The pit waters derived :from porphyry copper
                                 ,           '                     \
 deposits, in general, show the highest concentrations of metals.  The
 Berkeley Pit marks the worst case scenario of known porphyry copper pit
 waters, whereas the Yerington Pit represents the best case.  Although
 Yerington is a porphyry copper deposit like Butte and Ruth, the water is
 less contaminated.  Only two metals, Mn and Fe, exceed Federal
 standards, and the pH of the water is near neutral.  The Yerington pit
 has better water quality for two reasons.  First, the porphyry has a
 chrysocolla oxidation cap (copper silicate) rather than a sulfide cap,
 and the  high pyrite zone was eroded off in the Tertiary, leaving little
 acid generating material in the mine area  (Macdonald, 1992) .  Yerington,
 therefore, has significantly less trace metals and acid-generating
 sulfides,  resulting in neutral pH and  lower concentrations of dissolved

-------
                                                                      79
metals in the pit water.
      Pit waters derived from carbonate host  rocks tend to have lower
TDS, neutral pH, and lower concentrations of  metals  (Tables 1-1, 1-3, 1-
4).  This is primarily due to the acid-neutralizing  capability of the
carbonate host rocks, which buffers the pH of the water to the neutral
range.  Dissolution of carbonate minerals produces bicarbonate and
consumes hydrogen ions, making less acid available to dissolve metals
from the host rock.  The neutral pH, in turn,  favors the formation and
stability of hydroxides of Fe, Mn, and Al, which could remove up to 100%
of many trace metals from solution via adsorption (Balistrieri and
Murray, 1982).
      Water derived from volcanic-hosted precious metal deposits could,
potentially, be similar to waters in porphyry copper pits, since the
host rocks are of similar chemical composition.  Unfortunately, very few
pit lakes exist in volcanic-hosted mines to confirm this hypothesis.
Figure 5-2 shows that quartz monzonite consists of approximately 5-20%
quartz, 35-65% plagioclase feldspar, and 35-65% alkali feldspar.
Therefore, volcanic host rocks such as quartz latites will be closely
related chemically to porphyry copper mines in quartz monzonite.  The
buffering capacity should be relatively low as in porphyry copper
terrains, and the quantities of trace metals  may be high.   Alkaline or
quartz rich volcanic rocks will have equally  poor acid-neutralizing
capacity.  However, quantities of sulfides may be considerably less in
volcanic hosted precious metal deposits.
      An important factor that will complicate pit water chemical
modeling is the possibility that several mineralogical and lithological
suites might be present in. one pit.  As the pit fills, the water may
encounter different mineralogic or lithologic suites, causing different
rock/water interactions.  A typical example exists at the Gold Quarry
Pit  (PTI, 1992), in which the rising groundwater will encounter first

-------
limestone, then siltstone.   The likely result is that the initial water

levels will display neutral  pH levels, due to buffering by the
               fc
        MUhkWtpar
                       I**"*  I  moruomW   moruodontt
Figure 5-2  General clascifieation and nomenclature of common plutonie
and volcanic rock type*.  Classification bases on relative percentages
of quarts/ alkali feldspar,  and plagioclase,  measured in volume percent
(From Rurlbut and Klein,  19th ed.,  1977}
carbonate.  Ka the water encounters siltstone,  the pH may start to

decline as the effect of the carbonate rock is  offset by the absence

-------
                                                                      81
buffering capability in the siltstone.  However, this scenario might
only occur if the pit lake fails to turn over regularly.  The fact that
the pit water will always be in contact with a thick carbonate rock
suite may allow sufficient buffering that the acid generating potential
of the siltstone becomes negligible.
      Typical precious metal deposits exhibit a zone of oxidized
minerals overlying an unoxidized zone consisting of sulfidic and/or
carbonaceous rocks.  Alteration or gangue mineral suites are generally
present, such as pervasive silicification, decalcification,  or argillic
(clay) alteration, which disturb or obscure the primary mineralogy.
This type of alteration assemblage is common in sediment-hosted
disseminated gold deposits (Percival et al,  1988) .   The host rock may
grade from unaltered to altered zones, or from silicified to calcareous
or carbonaceous, making masses of specific minerals difficult to
quantify and model.  As the incoming grpundwater encounters the
different mineralogical regimes, different reactions may occur before,
during, and after the water enters the pit.  .Far:.from.the pit, the
system may approximate .a closed system, becoming .-more open to  .".   tr~
atmospheric gases as the water approaches the pit wall.   An oxidation
rind may,extend into the pit wall (PTI, 1992),  in which case the
approaching groundwater may encounter progressively decreasing
quantities of unoxidized minerals (e.g. sulfides),  and more oxides (e.g.
goethite).

      Acid Mine Drainage;  Acid mine drainage (AMD)  is  probably the most
widely studied aspect of mine-derived environmental contamination.  One
recent study says that AMD is the greatest problem caused by mining
(U.C. Berkeley Mining Waste Study Team, 1988) .   The three key
ingredients needed to produce acid mine drainage are a  sulfide mineral,
an oxidizing agent  (e.g. atmospheric oxygen  or ferric iron), and water

-------
                                                                       82
 (Nordstrom, 1985).  The  most  common sulfide mineral is pyrite (FeS,),
although sulfides of other minerals (Cu,  Zn,  As)  oxidize rapidly pnd
will also produce AMD.   Acid  mine waters  most commonly form by the
oxidation of pyrite under moist,  oxygenated conditions typical of many
active or inactive coal  and sulfide ore deposits  (Nordstrom,  1985).
      The ability of a rock sample to generate net acidity is a function
of the relative content  of acid generating and acid consuming minerals
 (SRK. 1989) .  As mentioned above,  typical sediment-hosted precious metal
                                              
deposits in Nevada will  contain both acid generating and acid consuming
minerals.. The balance between  the two will determine the extent to
which rock/water interaction  produces acidic  water.
      Recorded pH values from AMD are as  low as less than -1.0 (Iron
Mountain, California; Nordstrom and Alpers,  1990).   Dissolved metal
concentrations have been recorded as high as  46,000  ppm Cu (Butte, MT;
Nordstrom, 1985), 50,000 ppm  Zn (Baldwin,  Burma; Nordstrom,  1985), 43
mg/1 Cd (Iron Mountain,  CA; Nordstrom and Alpers,  1990),  56 mg/1 As
 (Iron Mountain, CA; Nordstrom and Alpers,  1990), 55,600 ppm Fe (Iron
Mountain, CA; Nordstrom  and Alpers,  1990).  10,000 ppmJU. (Comatock,  MV;
Nordstrom, 1985), and 420,000 mg/1 SO4J" (Iron  Mountain, CA; Nordstrom
and Alpers, 1990).
      The reaction that  oxidizes  pyrite and generates hydrogen ion and
sulfate is  (SRK, 1989):
           FeS, + 7/20, + H,0     Fe2* 4- 2SO,*- + 2H*       (1)

Under sufficiently oxidizing  conditions (dependent on both Eh and pH),
 ferrous iron, Fe (II), will oxidize to ferric  ion, Fe(III) :

              Fe" +  1/40, +IT  -  Fe1*  * 1/2H,O          (2)

 This reaction consumes hydrogen ion,  acting as a buffer at around pR
 2.0, and may explain why acid mine waters rarely attain pH levels below

-------
                                                                       83
about 2,0 (Nordstrom,  1985).  The  kinetics of this reaction are
relatively slow, about 10'* millimoles/hour (Singer and Stumm, 1970).
At pH values above 2.3 to 3.5,  the solution will be in the iron
hydroxide stability field, and  Fe(III)  may precipitate as Fe(OH)j,  again
generating hydrogen ions  (SRK,  1989):

                Fe1'   + 3H,0 =   Fe(pH)j(., + 3H*             (3)

This series of reactions generates 5 hydrogen ions and consumes 1,  for a
net of 4 hydrogen ions generated per mole of  pyrite oxidized.   The
overall process can be represented as:

     FeSj +  15/4O2 + 7/2HjO  =   Fe{OH)J(., +  2SO,2' + 4H*        (4)

At very low pH, pyrite can be oxidized  by ferric iron in the absence of
oxygen, via (Nordstrom, 1977):

        FeSj -t-  14Fe3' +  8H,0     15FeJ* + 2S04S'  + 16H*       (5)

      The kinetics of  this reaction are rapid, on the order of 0.002
milli-moles FeS, per hour (Garrels and Thompson,  1960} .  If reaction S
were the only means of oxidizing pyrite,  reaction 2 could not  generate
enough ferric iron to  sustain reaction  5 at the  rate of 0.002  mmols/hr.,
and acid generation would be limited by the rate of reaction 2 (Singer
and Stumm, 1970).  Unfortunately,  reaction 1  is  faster than reaction 2,
and a shortage of oxygen or water  in most AMD situations is unlikely.
      Pyrite oxidation is a self-maintaining  mechanism,  since  the rate
increases with lower pH, which  oxidizes more  pyrite and generates more
H* (and another oxidizing agent, Fe3*),  further lowering  the pH, and
continuing the cycle  (Nordstrom et al,  1979b) .   AMD from mine  pit walls,
mine shafts, or waste  dumps can theoretically continue until  all  the
sulfide has been oxidized, a process that may take centuries or
millennia to run to completion.

-------
                                                                      84
      According to SRK  (1989) , the primary  chemical  factors  which
determine the rate of acid generation are:

   pH
   Temperature
   Oxygen content of the gas phase, if saturation  is < 100%
   Oxygen concentration in the water phase
   Degree of saturation with water (water content)
*  Chemical activity of Fe3*
*  Surface area of exposed metal sulfide
*  Chemical activation energy required to initiate acid generation.

      Experiments show that the bacteria Thiobacillus ferrooxidans can
enhance the rate of pyrite oxidation (by reaction 2) six orders of
magnitude (Lacey and Lawson,  1970).  This would produce more than enough
ferric iron to sustain reaction 4, and AMD  could proceed in the absence
of oxygen.  The limiting factor in this case will be the growth rate of
T. ferrooxidans (Nordstrom, 1985).
      Small, "framboidal" pyrite crystals (<10'* m diameter)  are the
most reactive form of pyrite  (Caruccio et al, 1970).  The reactivity of
this form stems from the fact that framboidal pyrite exhibits the
largest surface area per mass of pyrite than any other form of pyrite.
However, surface area per mass could be equally as large for a highly
fractured sulfide ore body (Nordstrom,  1985) .
      In unmined ore deposits, insufficient oxygen is available to react
with sulfide minerals for AMD to occur at a rate that causes discernible
impacts on waters.  Reaction  (1) shows that 7/2 moles of O2 generate  2
moles of H*,  but dissolved oxygen in  soil waters is generally less than
                                   i                     I
0.6 mmol 02 per liter (Drever, 1988).   Once  mining exposes  sulfide
minerals to the atmosphere, and if sufficient meteoric or ground water
are available, pyrite oxidation and acid generation will ensue.
      Acid can be neutralized by calcium carbonate, via the reaction;
 (SRK, 1989):
               CaCOj + H*   --> Ca*'   +  HCO/

-------
                                                                      85
      Other carbonate minerals provide acid neutralizing capability, but
will generally be too scarce in most mine environments to have a
significant effect.  The exception is dolomite  (CaMg(CO,),) ,  which can
form thick depositional sequences similar to  limestone.  Sodium
carbonate minerals are shown to have greater  buffering capability than
calcium carbonates (Davison and House, 1988} .  Neutralization by  the
sodium salt leads to a final alkalinity greater than that obtained using
the calcium salt.  Since sodium carbonates are more soluble and have
faster dissolution kinetics, a smaller volume of rock should be required
to neutralize a given volume of acid water.   However, with the exception
of evaporitic terrains or in alluvium overlying mineral deposits,  sodium
                            f
carbonates will be greatly subordinate to calcium and magnesium
carbonates in the systems of interest in Nevada.
      Silicates, and some hydrous iron and aluminum oxides, also  consume
hydrogen during weathering, but generally have limited buffering
capability (SRK, 1989).  For example, A1(OH,)  can neutralize  acidic
solutions by the reaction:
              A1(OH,)  +   3H*   >  A1J*  +3H,O

      Although dissolution of carbonates, and other acid neutralizing
minerals, will suppress the production of hydrogen ions, the reaction
will still increase the levels of TDS in the solution.
      Precipitation of amorphous ferric hydroxide can armor buffering
minerals (Davis and Runnel Is, 1987), and "hide* them from acid
solutions.  This is discussed more thoroughly in other sections.
      AMD can be modeled after some difficult variables are defined,
such as the mass of sulfide available for oxidation, the kinetics of the
oxidation reaction, rate of introduction of water into the system, and
the volume of solution into which the dissolution occurs.  A model of
pyrite oxidation has been developed  (Davis and Ritchie, 1986) and

-------
                                                                       86
applied in pit water chemical modeling  (PTI,  1992).
      Prediction of potential AMD has been predicted by laboratory
experiments involving sulfide bearing rock from the  mine (PTI,  1992;
SRK, 1989).  A balance of acid generating rock  vs. acid neutralizing
rock should reveal the potential for AMD to be  sustained in the long
term.  In the absence of testing, chemical modeling  can provide a  guess,
by integrating an estimate of the mass of sulfide that  will be  dissolved
and the mass of buffering minerals as well (PTI, 1992) .

      Dissolved solids:  High concentrations of  the  major  ions  will
present less threat toxicologically than trace metals,  but will
contribute to overall water quality degradation  by increasing total
dissolved solids (TDS).  High TDS in mine-derived waters are caused by
two major processes (Nordstrom and Ball, 1985)r  1)  the oxidation  of
metallic sulfides,  such as pyrite, sphalerite, chalcopyrite, galena, and
                                                             s
arsenopyrite to produce high concentrations of trace metals  and sulfate,
and 2) acid dissolution of silicate bedrock (feldspar,  micas, clays,
etc.) that produce high concentrations of aluminum,  silica,  calcium,
magnesium, sodium,  and potassium.
      The dissolution of carbonate host rocks will introduce Ca2*,  Mg**,
and HCOj*  into solution.  Dissolution of silicate host rocks will add
dissolved species of Si, Al (depending on pH), Na, K, Ca, Mg, Fe, and
HCOj".  If evaporites are present, concentrations of  SO,, Cl, Na, and K
could be increased.  At near-neutral pH expected in  carbonate-hosted pit
waters, the concentrations of Ca, Mg, and HCO,"  should be controlled by
carbonate equilibria.  With increasing pH, carbonate control will yield
to silicate control.
      Attenuation of dissolved metals in a drainage  basin with  distance,
or in a lake with time, will occur due to oxidation,  precipitation,
adsorption, and dilution (Nordstrom and Ball,  1985) .
      Silica:  Dissolved silica generally does not reach extreme

-------
                                                                      87
concentrations, due to solubility controls  by silicate minerals such aa
kaolinite or other clays  (Drever, 1988).  Only above about pH 9 does
silica solubility increase beyond approximately 10"* molal (activity of
dissolved species) as shown by  figure  5-3.   In AMD situations,  silica
concentration can also be influenced by adsorption (Chapman et  al,
1983) .
                                                      14
Figure 5-3t  Activities of dissolved silica species ia equilibrium with
at 25*C, as a function of pH (from Drever, 1988).
      Sulfate:  Sulfate has an EPA standard, so prediction of sulfate
concentration and migration is important.  Sulfate normally behaves
conservatively in mine-derived solutions (Davis and Runnells, 1987),
although it may compete for adsorption sites at low pH (Chapman et al,
1983) , and may precipitate if subjected to Eh/pH changes (Filipek et al,
1987).  Sulfate concentrations will generally increase in solution as
sulfides are oxidized,  and may not be affected by neutralization.
      Nordstrom and Ball (1985),  and Davis and Runnells (1987) suggest
that sulfate is probably the best conservative tracer during downstream

-------
                                                                      88
dilution of acid mine waters, because it usually exists at high
concentration at the effluent source, and should be relatively
unaffected by precipitation or adsorption processes.  This behavior may
make sulfate the best indicator among major ions of acid mine drainage
in many systems, after trace metals are removed by adsorption and pH is
neutralized by dilution.

      Trace Elements;  Trace elements are defined as those elements that
generally appear in waters at concentration of less than 1 mg/1  (Drever,
1988).   Trace elements are of concern and must be monitored because of
their potential toxicity to aquatic and terrestrial'life (see review in
Macdonald, 1992) .   Trace metals can reach extremely high concentrations
4-n mine-related waters, as was discussed previously under acid mine
drainage.
      The behavior of trace metals in pit water will be difficult to
model for several reasons.  First, the quantity of metals dissolved from
the pit wallrock into the water will be difficult to predict, due to the
masses of metals contained in the wallrock, the morphology and
mineralogy of the metals  (disseminated vs. massive, sulfide vs. oxide),
the availability of metal to fluids  (associated with fractures,
armoring, etc.), the nature of the solution (saturation state, pR, Eh),
and variable rates of reaction due to all of the above.  Second, the
mass that remains as dissolved metal in the pit water will depend on
several factors, most notably the presence of Fe hydroxides and other
solids which can remove trace metals from solution by adsorption, and
the pH/Eh of the fluid, which will control the stability of species in
solution.  In general, trace metal concentrations will be higher at
lower pH, decreasing rapidly as the pH approaches neutrality.  Third,
the  thermodynamic data  (K^,  AH,0)  for some trace metal species are
questionable  (Nordstrom,  1992, personal communication).  If the

-------
                                                                       89
thermodynamic data is inaccurate,  the  model will be inaccurate.
      The trace metal analytical data  from pit waters  should be  used
with caution, because of potential problems in sample  collection,
storage, and analyses (Lyons, personal communication).  Analytical  data
for As, Ag, and Cu could be acceptable,  but data for Cd, Hg,  Pb, and Zn
could be cause for concern.
      The following paragraphs discuss the trace metals that  may be
important in pit water chemical modeling.
      Aluminum;  Aluminum concentrations are generally very  low  in
natural waters, usually less than  1 (tg/1.  but increase rapidly at low  or
high pH (Drever, 1988).   The minerals  believed to control aluminum
concentrations, gibbsite and kaolinite,  are very insoluble at neutral pH
(Nordstrom, 1982) as shown by figure 5-4.   Studies  indicate that
gibbsite controls aluminum solubility  for  slightly  acid to neutral pH
(above pH S.7, Davis and Runnel Is, 1987; at the Leviathan mine, above pH
              "8 -e
                     AMOH);
Al (OH|;
                                       At (OH)
                            /\\\
                                  6
                                 pH
      10
Figure 5-4 s  Activities of dissolved aluminum species in equilibrium
vith gibbsite  [Al(OB),]  at  25 *C,    function of pH (from Drever, 1988)

-------
                                                                      90
4.5, Nordstrom and Ball, 1985) .  Davis and Runnel Is  (1987) found  that
A1OHSO, is the  control  at  lower pH values.   Nordstrom and Ball (1985)
suggest that the solubility of aluminum at Leviathan below pH 4.5  is
controlled by the kinetics of the leaching rate of aluminum from bedrock
and soils.
      Aluminum can reach high concentrations in acid mine waters.  The
concentration of dissolved aluminum at a depth of 100m in the Berkeley
Pit is 206 mg/1 (Davis and Ashenberg, 1989) .  With rising pH, such as
through dilution or neutralization, aluminum minerals should precipi-
tate.  Filipek et al (1987) observed that gibbsite and kaolinite became
supersaturated at pH 5.25 and above, and suggested that precipitation of
a hydrolyzed aluminum mineral will occur above pH about 4.9.       In
carbonate hosted pit waters, with neutral pH, aluminum concentrations
should be too low to warrant concern.  However, Al will be a greater
concern in acid drainage situations that have potential to pollute
freshwater systems.  At low pH, the extreme sensitivity of aluminum
solubility to pR changes (Figure 5-4) can cause release of significant
amounts of Al.  Davis et al (1991) showed that a pH drop from 5.2  to 5.1
could release  .3 mg/1 Al into solution, increasing Al concentration over
the chronic toxicity threshold for trout embryo.  Poor plant growth in
soils located near acid mine drainage sites is also attributed mainly to
toxic concentrations of dissolved Al (van Breemen, 1973) .
      Several possible mechanisms can account for attenuation of
dissolved aluminum concentrations.  At neutral pH, Al will precipitate
as  residual weathering products such as clays.  Busenberg and Clemency
                     ^
 (1976)  found that mica and montmorillonite are rapidly precipitated from
the ions  released by the weathering of feldspars.  Aluminum may
precipitate as illite clay, aluminum sulfate, or amorphous Al(OH),
 (Davis  and Runnells, 1987) .  Aluminum can also be removed from solution
                                             N.
via adsorption onto solids  such as hydroxides  (Chapman et al, 1983) .  In

-------
                                                                       91
 AMD waters,  gibbsite, alunite,  basaluminute, and jurbanite are the most
 common precipitates  (Florence and Batley, 1980; Nordstrom, 1982; Chapman
 et al,  1983; Rampe and Runnells, 1989) .  Al mobility has also been shown
 to be governed by fluoride activity (Plankey et al, 1986).
       Arsenic:  Arsenic has multiple valence states and can form over
 245 mineral  compounds (Lynch,  1988).   The kinetics of oxidation and
 reduction of arsenic species are believed to be slow (Seyler and Martin,
 1989),  which causes both As(III) and As(V) species to be present in some
 solutions.   Biological activity may be the reason for this behavior
 (Masscheleyn et al,  1991).
       Sources of arsenic in precious  metal mines include orpiment
 (ASjSj), realgar (AsS), arsenopyrite (FeAsS), arsenic-bearing oxides,
      
 and iron sulfides.   Arsenic concentrations are dependent.on many
 factors,  including redox,  pH, adsorption, biological activity,  and
 kinetics (Macdonald,  1992).  Arsenic  species have a high affinity for
 adsorption onto hydroxides (Pierce and Moore,  1982).  Since most arsenic
 species are  anionic,  adsorption increases with decreasing pH,  the
 opposite behavior shown by cations (Balistrieri and Murray,  1983;
 Dzombak and  Morel,  1990;  Davis  and Leckie,  1980).   As (III)  compounds are
 generally more toxic than As(V)  compounds,  and inorganic As compounds
 are more toxic than organic As  compounds  (Bitton and Gerba,  1984).
 Removal of arsenic from solution can occur through precipitation of
, scorodite (FeAs04- 2H20)  and by adsorption onto iron hydroxides
 (Nordstrom and Ball,  1985;  Pierce and  Moore,  1982).   Much information is
 available on the role of  biota  in controlling arsenic chemistry  (e.g,
 see review in Macdonald,  1992),  but a  discussion is  beyond  the scope  of
 this study.
      Under  the oxidizing and neutral  pH  conditions  likely  found in most
 sediment-hosted pit  waters,  As(V)  species are  more  stable (Figure 5-5;
 see also speciation  files,  this  report),  and generally outnumber As(III)

-------
                                                                       92
species.  With decreasing pH  and/or Eh, As(III) species will become more
stable, as demonstrated in the  Berkeley pit, in which total As(V)
exceeded total As (III) by less  than one order of magnitude (Davis and
Ashenberg, 1989) .
                             SYSTEM   - - As-O-H
                                       25-C, 1 bar
                -0.8
Figure 5-Si  Kh-pH diagram for part of the system As-S-O-H.   The assumed
activities of dissolved species  are As  10*', 8  10"' (from Brookins,
1988)
      Cadmium:  Cadmium has been  observed to behave conservatively in
AMD systems  (Davis et al 1991; Chapman et al,  1983) .   Cadmium has  high
adsorption affinity for amorphous hydroxides at neutral  pH (Davis  et  al,
1987; McBride, 1980), but is strongly inhibited by competition  for
adsorption sites from other ions  (Balistrieri and Murray,  1982).
Cadmium has an EPA standard, so a prediction of mining related  impacts
is necessary.  Under circum-neutral  and oxidizing conditions, the
dominant  (unadsorbed) species are free CdJ* and CdCO,  (Brookins, 1988) .

-------
                                                                      93
                                *
In AMD situations, the dominant species  should be  free CdJ*.
      Copper:  Copper has a redox chemistry (Brookins, 1988),  so will be
sensitive to Eh-pH changes imposed on mine  water.  Copper has  been seen
to behave conservatively in AMD conditions  (Chapman et al, 1983),  even
when subject to dilution (Filipek et al,  1987).  However, Davis  et al
(1991) observed non-conservative behavior of  copper in the Clear Creek
AMD system, where copper concentrations  associated with particulate
fractions were an order of magnitude higher than in the dissolved
                                                   /
fraction.  Davis et al (1991) explain that  this may be due to  the pHso
(the pH at which 50% of the metal remains in  solution) of Cu vs.  Zn (Cu
=4.5, Zn = 5.5).  Their experimental data  says that  >90% of Cu  is
expected to be adsorbed at pH 6.0.  Their MINTEQA2 simulations indicated
that Cu2* and CuSO, are the dominant species at low pH,  whereas Cu(OH)3e
dominates at neutral pH.   All Cu minerals were undersaturated,
indicating Cu removal via adsorption.  In AMD  situations, copper
concentrations can be high enough to be  controlled by mineral phase
solubility.  At their Daylight Creek (NSW,  Australia) study area,
Chapman et al (1983) observed precipitation of a copper mineral  they
suggested was Cu2(OH),COj.
      Iron:  Iron has multiple valence states and  is  strongly controlled
by redox.  In most natural (oxygenated)  surface waters, iron should
generally be in the ferric (Fe1*) state,  which is very insoluble.
Therefore, iron mobility will be controlled by the precipitation of
ferric hydroxides.  Under more reducing  conditions, such as in
groundwater, ferrous iron  (FeJ*) may dominate  other iron species  and FeJ*
precipitation will not occur.  As the Eh-pH diagram for iron species
                              
shows  (Figure 5-6), many natural waters,  at near neutral pH and  slightly
oxidizing conditions, may lie near the phase boundary between Fe2* and
Fe (OH),.
      Iron concentrations are very sensitive  to pH as well as redox.

-------
                                                                       94
Davis, et al  (1991) discovered  that  dissolved iron concentrations in a
Colorado stream were determined by the solubility of different amorphous
Figur* 5-fit  Sh-pH diagram for part of  th* ay tea F-O-R assuming
F(OH), as stabl* F(XXI)  phas*.   Aasumad activity of dissolved
F  10-*,  (from  BrooJcins, 1981)
ferric hydroxide forms present.  As pH  increased due to dilution by more
alkaline tributaries, iron waa removed  from solution by the reaction:
            Pe (OH) ,*   -f   OIT
Fe(OH),(s).
      This was verified by calculating the saturation index for
ferrihydrite as a function of distance from the source of the AMD  (Davis
et al, 1991).  The extreme sensitivity of iron to pH and pe is seen  in
Figure 5-7, which shows that a shift in pe or pH can result in a change

-------
                                                                      95
in iron concentrations by several orders  of magnitude.
                    I    i    ill    ill
                234   56789   10  11
Figure 5-7:  Con tour  of dissolved iron as a function of pa and pH,
assuming pCO,  NT1,  CS  10'' .
      Jarosite will commonly precipitate in AMD waters if the pH is
between 1.5-2.5, and sufficient dissolved iron is present (Nordstrom et
al, 1979b) .  In some extremely acidic AMD streams, Jarosite
precipitation might not occur close to the source, but will appear
downstream as the pH rises to the appropriate range.  Jarosite
precipitation occurs as predicted by chemical modeling at the Berkeley
Pit (Davis and Ashenberg,  1989).
      Under most conditions, free ferric iron is never more than about
8% of the total iron.   Nordstrom et al (1979b) found in four AMD streams
that FeSO,  can constitute up to  50%  of the  total  dissolved  ferrous  ion.
Organic complexing of iron may be important in some waters,  but in AMD
terrains the organic matter is probably fully protonated and has only a

-------
                                                                      96
minor effect on complexing  (Nordstrom et al,  1979b).
      The solubility data for ferric hydroxide  and  ferrous  sulfide
minerals show much variability depending on crystallinity (Ball  et  al,
1980).  For this reason researchers should consider a  range of phases in
chemical models involving iron (Plummer, et al,  1983).   Further  compli-
cations are introduced by the definition of dissolved  vs. particulate
iron, the distinction between the two often being defined operationally
as the size fraction that passed through the  filter used in sampling.
      Lead:  Lead is very particle reactive,  with high affinity  for
adsorption onto hydroxide precipitates.  With sufficient ECO,, the
dominant lead species in circum-neutral, oxidizing  waters should be
PbCOj (Brookins,  1988).   Lead generally appears in very  low amounts in
most precious-metal deposits  (Percival et  al, 1988).   Low concentra-
tions, combined with high particle reactivity,  may  preclude lead from
being a problem in pit waters.
      Manganese:  Manganese has been observed to behave  conservatively
in low pH, oxidizing environments  (Rampe and  Runnella,. 1989;  Davis.et.
al,  1991) and even in diluted AMD systems  (Filipek. et  al, 1987)..--
Manganese is expected to precipitate in more  oxidizing,  alkaline
conditions  (Davis et al, 1991).  Manganese hydroxides  are considerably
better scavengers of trace  metals from solution than  iron hydroxides
 (review by Chao and Theobald,  1976) , and therefore  could be an important
control of trace metals  in  pit water.
      Mercury:  Mercury  is  insoluble and concentrations  are generally
very low  in natural waters  (Fitzgerald, 1979).   However, mercury is very
common  in precious metal deposits,  and may be present in resultant  pit
water (see Table  1-4).   Mercury  exhibits a redox chemistry in natural
waters  (Brookins,  1988), and has an EPA primary standard.
       Zinc:   Zinc has  no redox chemistry,  and has shown  to be both
 conservative  (Filipek,  et  al,  1987; Chapman et  al,  1983) and non-

-------
                                                                      97
conservative  (Bencala et al, 1987) in AMD systems.  Zinc has a
moderately high adsorption affinity for amorphous ferric hydroxide
(Tessier et al, 1985; Karlsson et al, 1988).  The dominant Zn species in
circum-neutral, oxidizing waters should be either free Zn2* or ZnCO,
(Brookins, 1988), depending on EC03.

      Oxidation/Reduction (Redox)i  Redox potential will be a critical
mechanism controlling dissolved metal concentrations, speciation, and
mineral phase stability in pit water.  Oxidizing conditions  (positive
Eh) favor stability of Fe-hydroxides, hence enhancing adsorption of
trace metals onto the Fe-hydroxides and removal from solution.  Reducing
conditions (negative Eh) favor destabilization of Fe-hydroxides and
possible dissolution, which will cause desorption of trace metals to
                    i
solution.  Metals released to solution could potentially be transported
out of the pit into a downgradient aquifer and contaminate water
supplies.
      Both natural and pit lakes can become chemically stratified,
causing vertical gradients of both, dissolved oxygen (0,) and  redox
potential, as in the Berkeley Pit (Davis and Ashenberg, 1989) .  The
likelihood of vertical stratification in pit lakes will depend on the
factors that determine the extent of mixing in the pit lake  (thermal
input, wind velocity, etc.).  The ability of redox potential to
influence lake chemistry through stratification can be modeled by
dividing the lake into vertical cells, each with its own redox
potential, dissolved oxygen, or relative activities of specified redox
couples.
      Verification of the existence of a redox change can be shown by
the relative concentrations among redox couple speciation between two
locations along a hydrologic flow path, such as between groundwater and
a pit.  As Plummer  (1983) demonstrated with sulfide species,  if

-------
                                                                      98
concentrations go from nondetectable to detectable,  then the problem
involves redox.
      For the most accurate chemical model application, one should
analyze for concentrations of each redox couple of  interest, such as
ferrous and ferric iron, that might be important  in the water  (Nordstrom
et al, 1979b).   Nordstrom et al  (1979b) outlined  some fairly rigorous
criteria that must be met before a measured Eh can  be related to a
specific redox couple.  These criteria are generally not met in natural
waters, but AMD waters may prove the exception.   In 60 samples of AMD
waters, Nordstrom et al (1979b) found that measured Eh correlated well
with Eh calculated from the ferrous/ferric couple with the Nernst
equation.  They also discovered that the Eh calculated using the O2/H,O
couple was higher than the measured value.  The redox state of the water
is thus determined by the ferrous-ferric ratio, and O, is  not  in
equilibrium with the ferrous-ferric couple.

      Adiorption/eoprccipitationi  Adsorption onto  mineral surfaces is
generally believed to be the dominant controlling mechanism for trace
element concentrations in natural waters  (Drever, 1988), and could also
control concentrations of dissolved metals in pit water.  Metals such as
Al, As, Cd, Cu, Fe, Hg, Mn, Hi, Pb and Zn can readily adsorb onto
particulate matter (Karlsson et al, 1988; Tessier et al, 1985; reviews
by Turekian, 1977, and Murray and Brewer, 1977) .  Important inorganic
sorbents include hydroxides of Al, Fe, Mn, and Si.   Despite the
complexity of natural waters and the many parameters that can control
adsorption,  the agreement of field data with laboratory experiments and
with  theory  is relatively good  {Tessier et al, 1985).
       The stability of iron hydroxide solid is strongly dependent on
both  pH and  Eh.  The Eh-pH stability diagram for  iron  (Figure 5-6) shows
that  Fe(OH), is more stable in higher pH,  oxidizing waters.  A decrease

-------
                                                                      99
in Eh (anoxia) or pH will cause the solution to move out of  the  Fe(OH),
stability field, into the Fe**  field, which could  potentially dissolve
the Fe(OH),  and release  any sorbed metals into solution.  This is a
concern in pit water chemical evolution, if carbonate buffering  is
absent or becomes inhibited through processes such as armoring.  Anoxia
may also develop if turnover fails to occur.
      Fe and Mn concentrations have been observed to decrease along a
flow path that experiences a pH increase (Hicks and Groves,  1993) .  The
pH dependent solubilities of both hydroxides are  seen in the reactions:
                 + 02 + 4H2O   =   2Fe(OH), + 2H*
            2Mnz* + Oj + 6H,O   =   2Mn(OH)3 -t- 4H*

Wicks and Groves observed that this decrease is accompanied by
precipitates of Fe and Mn hydroxides in the stream.
      High sensitivity to pH is further seen in a laboratory study by
Davis et al (1991) , which demonstrated that approximately 90% of Zn was
adsorbed at pH of 6.8, and 50% at pH 5.5.  They suggest that in the
event of blowout (rapid discharge of AMD to surface water. body) ,
sediments could potentially desorb significant non-point loads of Zn.
      Adsorption reactions can be described in a simplified manner by
formation or surface complexation constants similar to solubility
product constants.   The adsorption of a metal ion, M, onto an oxide
surface can be represented by (Benjamin and Leckie, 1981; Balistrieri
and Murray, 1983) :     .

                              V.
                   SOH,  +  M '    SOM  *  XH*
where SOH, represents the free surface  sites,  SOM represents the surface
complex, and x is the average number of H* ions released per M adsorbed.
If 'KA is considered  as an "average" equilibrium constant, and
expressing the above variables in terms of concentrations, then the

-------
                                                                     100
surface complexation constant is:
                           [SOM] [H'J
                           tSOH,] [M]

      As the above equation demonstrates, 'KA is pH dependent.   The
adsorption behavior of cations generally conforms to the curves shown  in
Figure 5-8.  Adsorption of-cations is insignificant at low pH, because
hydrogen ions outcompete other cations for adsorption sites.  Increasing
pH increases adsorption of cations in simple systems and thus should
also increase KA according to the  above  equation.  (Tessier et al,  1985).
Cation adsorption is analogous to hydrolysis in that both increase and
release protons with higher pH (James and Healy, 1972).
             100
              80
              60
              40
              20
                                                10
 Figure 5-81   Experimental data  and computed curves  for  adsorption of
 metals on SiO,  (James and Healy, 1972), from Drever (1988).
       Adsorption of  trace metals  onto  iron oxyhydroxides typically
 increases from near  0%  to near 100%  as the pH increases through a narrow

-------
                                                                     101
critical range of -2 pH units, referred to  as  the "adsorption edge1
(Tessier, et al, 1985).  Without exception, Tessier,  et al's experiments
for adsorption of Cd, Cu, Pb, Ni, Zn onto oxic lake sediments showed
sloping curves with higher adsorption with  increasing pH.   Lead
adsorption with varying pH is shown in figure  5-9A.  In these examples,
adsorption tails off to zero at about pH 4.0,  but approaches 100% at
about pH 7.0.  Experiments by Davis et al  (1991)  showed that the
affinity of metal ions for pure amorphous ferric  oxyhydroxide is Cd  < Mn
< Zn < Cu.
      100
      so
s       t
    H
                                          H>CO
                                           .4
                                                 XT
                                                             -t
Figure 5-9:  Adsorption of lead on alumina (y-Al,O,) .   A)  as a function
of pH for different surface site concentration*7  B) as  function of
surface site concentration at different pH (from Morel and Bering,
1993).
      Anion adsorption is a mirror image of cation adsorption, stronger
at lower pH, and weaker at high pH (Balistrieri and Murray, 1982).  An
example is arsenic, which forms primarily anionic species in solution.
Figure 5-10 shows the adsorption behavior of an arsenic species.
      The extent of adsorption of trace metals onto iron hydroxides is
believed to depend strongly upon certain characteristics of the
adsorbent surface, such as porosity and specific surface area
(Kinniburgh and Jackson,  1981).

-------
                                                                      10'
      Adsorption density  (moles of metal adsorbed per mole of
adsorbent), r, can be written for adsorption onto iron oxyhydroxides:
                              [SOM]
where FeT is the total concentration  of  iron present  as oxyhydroxides
(Tessier, et al, 1985).  Adsorption increases with increasing.site
density, as shown by Figure  5-9B.
       to
       
-------
                                                                     103
showed that ionic strength does not significantly influence  the
adsorption of Cu and Pb.  The reason may be that  adsorption  reactions
result in no net change in surface charge, and  therefore are not
susceptible to changes in surface charge caused by ionic strength
variations.
      Major ions can enhance or inhibit the adsorption of trace metals.
Balistrieri and Murray (1982) showed that Mg2* suppresses trace metal
adsorption by decreasing the number of available  sites, whereas S04*" can
enhance the adsorption of cations by changing the electrostatic
conditions.  Benjamin and Leckie (1980) showed  that there is competition
between metals for sites on surfaces of YFeOOH  and -yAljOj even though
the available surface sites are far in excess of  adsorbing species.
Including the adsorption behavior of ions such  as Mg8* and SO,1" in the
pit water models would consider the potentially important competitive
effects of these ions.
      The presence of organic matter can change the adsorption model
significantly.  Organic matter complexes certain  trace metals strongly/
particularly Cu (Mantoura,  et al 1978). and can adsorb-on suspended    -~-
surfaces (Balistrieri and Murray,  1982},  The latter can change the
surface characteristics of suspended solids so  that they acquire the
chemical behavior of organic functional groups  such as (-COOH)
(Balistrieri and Murray,  1982) .
      Smith and Jenne (1991) discovered that aging effect the ability of
amorphous ferric hydroxides to sorb metals.  Older solids exhibited a
higher degree of crystallinity making less sites  available for
adsorption.
      Quantifying and modeling adsorption are very difficult, and
several theories are currently in use.  Although  surface complexation
constants see wide use,  Morel and Hering warn "... no universal
equilibrium constant can be simply defined and  various adsorption models

-------
                                                                     104
differ principally by the manner in which the electrostatic  interaction
term is calculated," and Dzombak and Morel caution that  "equilibrium
constants (for adsorption) are not, in fact, constant."  Despite  the
problems, factoring adsorption into pit water chemical models  will
provide a means of removing metals from the pit water.
      The number of input parameters for an adsorption model will depend
on the complexity of the model selected.  At the minimum, the  required
parameters are:
*  An adsorption constant (K^)
*  Number of adsorption sites available, usually a function of:
         concentration of adsorbent.
         type of adsorbent,  and its charge in solution.
         Surface area of adsorbent available to solution.
*  Concentration of adsorbate in solution.
                                                           N
MINTEQA2 Adsorption Models
      Seven adsorption models are available in MINTEQA2.  Surface
reactions in MINTEQA2 are written in terms of the neutral surface site
SOH, and the equations are written as formation constants. -_ .   	  .
      Non-Electrostatic Adsorption Models.-  The simplest adsorption
models are the activity K,,,  Langmuir,  and Freundlich models.   The
activity K* and Freundlich models make the oversimplifying assumption
that an unlimited supply of surface sites is available.  This  assumption
renders competition between different adsorbing species meaningless, and
the adsorbing surface cannot become saturated no matter how  large the
supply of adsorbing ions.  The activity K^ model is adequate if the
concentrations of the adsorbing metals are low, and the pH and ionic
strength are relatively constant  (Peterson et al, 1987).
      The Langmuir adsorption model requires that the number of
available surface sites be specified.  This marks an improvement  over
 the  activity K and Freundlich models, since it eliminates the problem

-------
                                                                     105
of unlimited surface sites.
      The ion exchange model assumes that the surface  site is initially
occupied by an exchangeable ion that is released  into  solution during
the exchange process (Allison et al, 1991).  The  user  must supply
reaction stoichiometries, exchange constants, and the  ions participating
in the exchange process.
      Electrostatic Adsorption Models:  The constant capacitance,
diffuse layer, and triple-layer adsorption models include  the effects of
surface charge and potential on the adsorptive behavior  of ions and
adsorbents in a system.  This influence is incorporated  into  the mass
action equations by including terms that modify the activities of
sorbate ions as they approach charged adsorbent surfaces.   The
activities are modified by accounting for the electrical work necessary
to penetrate the zone of electrostatic potential  extending away from the
surface (Allison et al, 1991).  These three models treat trace metal
surface reactions as complexation reactions analogous  to the  formation
of complexes in solution.  The surface complexation models in MINTEQA2
were developed to describe surface reactions of amorphous  metal oxide in
aquatic systems, having been successfully applied to them  in  prior
experiments.
      In all 3 models, a charge (o) on the surface is  assumed to be
balanced by a charge (
-------
                                                                     106
balanced by the charge in the diffuse  layer  such that o + <7d - o
(Allison etal, 1991).
      The electrostatic potentials associated with the surface charge
can influence ion activities in solution.  The result is that the
activities of background and electrolyte  ions near the solid surface are
different from the concentration of the same ions in the bulk
electrolyte.  The activity difference  is  caused by the electrical work
required to move ions across the potential gradient between the charged
surface and the bulk solution.  The activity change between these
regions is a function of the ion charge (z)  and the electrical potential
(*) near the surface.  The relation can be expressed by the Boltzmann
equation (Allison et al, 1991) :

                   {X.*}  -  (X'} [e-w']
where:
         !X/}    activity of an ion X  of  charge z near the surface.
         X1}    corresponding  activity of  X in bulk solution outside the
                influence of the charged  surface.
        e'*""*  -  Boltzmann  factor
        z  m  charge of ion        .......    .--  . r.  -"-"".   ..-.  -.   	:
        P  *  Faraday constant     	                       -  -
        R    ideal gas constant
        T    absolute temperature

      Surface complexation reactions take a  variety of forms, depending
on the adsorbing ion and whether the surface is represented as neutral,
protonated, or deprotonated.  An example  of  how a reaction is expressed
in MINTEQA2 can be illustrated by starting with a basic protonation
reaction:      ' .
                    SOH + H.*  <	>   SOH,*

for which the mass action expression is:

                           {SOH/}
                     K - 	
                         {SOH}

-------
                                                                     107
The activity of the surface hydronium ions must  be  corrected for the
work performed in moving them to the charged  surface  where the reaction
occurs.  The corrected activity is expressed  as:
                   {H/}   -   {H*} (
                                   /
The corrected term is then substituted into the mass  action expression
to give:
                               {SOH,*}
                     K  -
                         {SOH} {H*}  [*/].

A similar result is seen for the deprotonation reaction  (see Allison et
al, 1991).
    - To define a diffuse layer adsorption model in MINTEQA2,  the user
must provide the following information:

      *  TSOH * Surface site density (moles of sites/1) .
      *  S  "  Specific surface area  of the solid (m2/g) 
      *  C,    Concentration of solid in the suspension (g/1) .
      *  Surface reactions in terms of MINTEQA2 components,  including
         the surface complexation (formation) constants,
Values for T^,  SA, and K^, are available from the literature,  and C, can
be determined from the concentration of the solid in the system.
      The constant capacitance and diffuse -layer models are very
similar, differing only in the function relating total surface charge o.
to surface potential *. .   The constant capacitance  model ia a special
case of the diffuse- layer model for solutions of high ionic strength and
surfaces of low potential (Allison et al,  1991) .
      The constant capacitance and diffuse- layer model specify only one
layer in which specifically adsorbed ions define the surface charge a.
That plane is referred to as the o-plane,  and its surface charge and
potential are designated a.  and *. .  Hydronium and hydroxyl  ions form

-------
                                                                      10P

the bulk of the surface charge  (s.) on the  o-plane.  Figure  5-11  shows a

schematic model of the diffuse  layer  model associated  with the surface

of a solid such as amorphous  ferric oxide.
                                      CONSTANT CAMCITANCf HODO.


                                         Utrvtl UYEft MOOCL
Figure  5-11i  Schematic representation of the surface  charge/potential
relationships used in the constant capacitance and diffuse-layer aodela
(from Allison et al, 1991).
       The  triple-layer model  is  slightly more complex  than the diffuse

layer  model, and considers  the adsorption layer to be  composed of two
                             y
constant capacitance  layers bounded by a diffuse layer of  electrolytes

 (Peterson  et al, 1987).  A  schematic is shown in Figure  5-12.   The

triple-layer model  in MINTEQA2 allows only protonation and deprotonation

reactions  in the o-plane.   Adsorption of other cations and ions occurs

-------
                                                                     109
in the -plane  (or inner Helmholtz  layer),  and non specifically adsorbed
ions reside in the diffuse  layer or 'd' plane  (outer Helmholtz  layer).
Figure 5-12t  Schematic representation of surface species and
charge/potential relationship* in the triple-layer model* (from Allison
et al, 1991).
Input parameters for the triple-layer model include two capacitance
terms and three electrostatic components.  Davis and RunnelIs (1987)
list values for surface site density and capacitance values for
adsorption in the triple-layer model.
      In the Gold Quarry pit geochemical model,  PTI (1992)  restricted
their adsorption models to the diffuse layer and triple layer models.
By fitting the model results to their experimental  data for each

-------
                                                                     110
specific metal ion of interest (arsenic, cadmium, copper, and lead),
they determined which model best described the adsorptive behavior of
each ion.

      Oroundwater/Aquiftr Geochemistry:  Before applying any
hydrogeochemical model to an aquatic system, the modeler needs a high
quality, comprehensive chemical analysis for each water to be modeled.
Without accurate knowledge of groundwater chemistry, the model will
probably be invalid, and may carry erroneous information into water
quality prediction, monitoring, and remediation.
      In most pit water scenarios, it is reasonable to assume that the
aquifer mineralogy near the pit will be the same as or similar to the
pit wallrock mineralogy, with the possible exception of ore-related
minerals such as -trace metals.  Water well chemistry in the vicinity of
the Universal Gas pit (Geraghty & Miller, 1991) shows major ion
concentrations consistent with an aquifer comprised of interbedded
carbonate rocks and clastic sedimentary units  (i.e., calcium, magnesium,
                                      *
alkalinity, silica).  This chemistry.may be representative of other.   - /
upgradient aquifers associated with precious-metal pit waters.  The
upgradient water may be expected to be in equilibrium with the dominant
mineral assemblage.
      The equilibrium state of the groundwater  (mineral saturation
indices) may influence the mineral mass transfer once the water
approaches the pit and comes in contact with wallrock minerals.
Groundwater in equilibrium with the aquifer may reach equilibrium faster
with the pit wallrock, or be less undersaturated with respect to
wallrock mineralogy.  The result might be less pit wall mineral
dissolution, and less mass transfer from the wallrock to the pit water.
                                                      
      Some ore deposits have unusual geologic/geochemical settings that
might cause interesting scenarios in local groundwater geochemistry.  Ar

-------
                                                                     Ill
example is the Ruth district in Nevada.  Ruth is a porphyry copper
deposit, with quartz monzonite porphyry the dominant host rock,  typical
of such systems.  However, this intrusive is set within a regional
                                                            u
stratigraphic sequence consisting of carbonate rocks.  When the  Ruth pit
water, with low pH and high metals, flows out of the pit and mixes with
the carbonate groundwater, the resulting water will likely be
significantly different from either of the precursors.  The carbonates
will probably buffer the groundwater toward neutral pH, causing  metals
to precipitate from solution.

      Reaction Kinetics:  Consideration of kinetics in pit water models
would greatly improve the validity of the model, but will require  more
                             A
detailed techniques and experiments.  Applying experimentally derived
kinetic data to complex field situations is difficult.
      Many factors influence rates of reaction, the most important
perhaps being the surface area of reactant available.  This will be
constantly changing .in field situations during dissolution/precipitation
reactions.  Without a knowledge of the reactive surface :a_rea in contact
with a unit volume of water flowing through the pore spaces, kinetic
data will remain somewhat empirical and of questionable value from one
system to another (Plummer, 1984} .  Apparent rates of reaction can be
derived from inverse modeling if the mass transfer results are combined
with estimates of residence times.  This procedure has been applied to
groundwater systems by Plummer (1984), using isotopic data.
      Quantitative interpretation of reaction kinetics in AMD situations
is difficult due to the complicated chemical system of buffering (Wicks
and Groves, 1993).  Rates of CaCO, dissolution measured in the field are
several orders of magnitude slower than rates predicted from the Plummer
et al  (1978) dissolution rate law  (Wicks and Groves, 1993) .  The
experiments of Wicks and Groves revealed that dissolution of Icelandic

-------
                                                                     112
Spar CaCOj in Camp's Gulf Branch was inhibited.  They also noted
precipitation of hydroxide floe on  the calcite crystals, and  suggest
that the floe inhibited the dissolution by armoring  the calcite.
Armoring may be a serious problem in pit waters, by  preventing
dissolution of potential solution buffering minerals.  However, armoring
may also affect acid generating minerals in a  similar manner.
      Rates of reaction of pit: wallrock minerals might decrease over
time, as the saturation state increases and mineral  availability
decreases.  More soluble minerals,  or minerals associated with
structures or zones of high permeability, might dissolve rapidly during
early stages of pit submergence.  As these minerals  are removed or
become armored, dissolution rates may slow.
      The.kinetics of pyrite oxidation have received great attention.
Davis and Ritchie (1987) developed  a model for oxidation of pyrite that
has been applied in pit water geochemical models (PTI, 1992) .  The
kinetic models generally agree that oxygen availability is the limiting
factor in pyrite oxidation. . The rate, of oxygen .diffusion into the pit
wall will be a function of the permeability of the wallrock.
      None of the codes evaluated in this study explicitly consider
kinetics.  The rate of reaction may be controlled in PHREEQE and
MINTEQA2 by specifying incremental  quantities  of components
participating in reactions for dissolution, precipitation, or other
processes.  EQ3/6 (Wolery, 1992) incorporates  a limited ability to model
reaction kinetics.

      Equilibrium Thermodynamicst   Speciation  models of existing pit
waters generally show several mineral phases thermodynamically over-
saturated.  Precipitation of minerals from solution  can produce some
important changes in pit water chemistry, such as reduction of dissolved
solids, removal of metals, and armoring of reactive  minerals.

-------
                                                                     113
      Aluminosilicate thermodynamics  (feldspars, pyroxenes,  amphiboles,
micas, olivines) introduces complications into modeling.  Most
aluminosilicates weather incongruently, leaving a solid residue  behind,
such as kaolinite or other clay minerals  (Drever, 1988).  Many
weathering products can be generated, such as mixtures of clays,  which
could armor other reactive minerals in the system.  Modeling results
involving aluminosilicate thermodynamics should be interpreted with  '
caution.  As stated by Ball and Nordstrom (1991), the use of solubility
product constants (K^,) for many silicate minerals  (smectites, illites,
chlorites, micas, feldspars,  amphiboles, pyroxenes, and pyrophyllites)
is not recommended because these phases have not demonstrated reversi-
ble, equilibrium solubility.

      Biological Activity:  Biological activity may be important  in
controlling several aspects of pit water geochemistry, such  as redox,
reaction kinetics,  and chemical speciation.   Biological effects were not
incorporated into the modeling simulations of this-study,  but some of
the more important aspects deserve mention.          ~   	.. ;     ,. :
      Bacterially mediated oxidation of ferrous iron can enhance pyrite
oxidation by up to six orders of magnitude (Lacy and Lawson, 1970).
Thiobacillus ferrooxidans is  the most common bacteria that oxidize
sulfur and iron, but at least 18 others are known (SRK, 1989).  Bacteria
can also accelerate the oxidation of other sulfides common in mineral
deposits  (Lundgren and Silver, 1980).  SRK (1989)  lists the following
factors which will determine  the bacterial activity, and the associated
rate of acid generation:
      *  Biological activation energy
      *  Population density of bacteria
      *  Rate of population growth
      *  Nitrate concentration
      *  Ammonia concentration
      *  Phosphorus concentration
      *  Carbon dioxide content
      *  Concentrations of any bacterial inhibitors

-------
                                                                     114
      Oxidation of organic matter may  influence  the level  of dissolved
oxygen in the lake, perhaps accelerating anoxia,  which will  affect the
O2 available for chemical oxidation.  Depletion of 0, with depth in
lakes will commonly be from oxidation  of organic matter, but may also  be
from chemical oxidation  (Hutchinson, 1957).
      Microbially mediated sulfate  reduction, followed by  reduced sulfur
mineral precipitation, can attenuate the acidity and high  concentrations
of iron and sulfate in acid mine drainage.  Sulfate  reduction can
partially counteract acidic inputs  in  freshwater lakes  by  generating
bicarbonate alkalinity via the reaction (Wicks et al,  1991):

      SO,''  +  2CorgMie  +  2H20   -->   H,S   +  2HCO/     (1)

This reaction is microbially mediated; the  sulfate reduction step
requires the presence of organic matter in  the sediments to  provide a
carbon source for the bacteria (Wicks  et al, 1991).   The sulfide  formed
in this way must be stored in a reduced form in  the  sediments to  sustain
the alkalinity generated (Hicks et  al., 1991).   At low  pH, the sulfide
may be released to the atmosphere as hydrogen sulfide gas  (Doyle, .1976).
                       x                         '
If the pH is above 7, HS~ will form rather than  H,S,  which  will  form
solid sulfides if any reactive iron is present (Drever, 1988).
      Wicks et al  (1991)  examined sediments of 15 coal  strip mine lakes
for evidence of sulfate reduction.  The end products  of reaction  (1)
were found in the sediments of all  15  lakes, indicating that sulfate
reduction occurs followed by authigenic sulfide  mineral formation.  Both
sulfide minerals and organic sulfur were observed in  the lake sediments.
Organic sulfur phases generally dominate the distribution  of end
products of sulfate reduction in slightly acidic, sulfate-rich, iron-
poor systems  (Wicks et al, 1991) .   Organic  sulfur phases also may
dominate  in AMD lakes, but in iron-rich sediments, sulfate reduction end
products are primarily inorganic  (Herlihy et al,  1980).  If  Pe

-------
                                                                     115
availability is limiting, an increase in sulfate or organic matter
content will not lead to an increase in the FeS or pyrite content  of  the
sediments.  In Nevada pit lakes, inorganic reduced sulfur minerals
should dominate the authigenic sulfur, since they would be associated)
with Fe-rich and S04-rich sediment.
      A variation of reaction  (1)  is provided by Bell et al  (1987) :

    2Fe(OH)J(.,  + SO,1' + CHjCOO'  +  H,  -->
                    FeS,., + FeJ*  +  2HCCV + 3H,0 + 3OH'       (2)

which shows that authigenic pyrite formation is controlled by several    '
factors.  The extent to which reaction (2) proceeds may be limited by
the concentration of any one of  the reactants, but organic matter  is
believed to be the limiting factor in this process (Berner, 1971).
Evidence of this in the pit lakes  included high correlation between
organic matter content and the sulfide minerals, and the low correlation
of porewater Fe and porewater sulfate concentrations with any of the S
mineral pools  (Hicks et al, 1991).
      Hicks et al (1991)  observed a correlation between age of strip
mine lakes and health (water quality).  The age of the lakes correlated
with the amount of reduced S mineral in the lake sediments.  The oldest
lake was only 23 years old, yet  showed signs of self-remediation through
authigenic pyrite formation.  The lithologies included limestone,
sandstone, and shale, and the pH ranged from 2.6 to 8.0, conditions
potentially similar to future Nevada pit lakes.  The effects of lake age
are:  (1) a lessened acidic, sulfate- and Fe-rich load delivered to the
sediments as the pyrite weathers from the spoil pile; and (2)
amelioration by the alkalinity generating reactions that occur in the
sediments.  The time schedule  (time required for lake to self-remediate:
acid neutralization, decreased amount of pyrite oxidation) of any
specific lake is determined by the amount of pyrite present in the spoil

-------
                                                                     116
that will eventually weather and produce  acidity,  the thickness  of  the
vadose zone, the precipitation rate in the  area, and ground water flow
direction (Wicks et al, 1991).
      Photosynthesis can influence the pH of aquatic systems  on  a cyclic
basis (Wetzel, 1983; Drever, 1988).  The  photosynthesis  reaction
(Drever, 1988) :

106CO, + 16NO,- + HP04a' + 122H,0 + 18IT + (trace elements, energy)
                                      P! + 1380,
utilizes carbon dioxide and hydrogen, thus increasing the pH.
Respiration, the reaction in reverse, produces hydrogen which decreases
the pH.  The variation imposed by photosynthesis can also influence the
sorptive behavior of trace metals.  Fuller and Davis  (1989) observed
significant daily fluctuations in trace metal dissolved and sorbed
concentrations in their study of Whitewood Creek, South Dakota, which
they attributed to uptake/release of trace metals by adsorbents as pH
changed due to photosynthesis .
      Chemical modeling may reveal biological activity not immediately
verifiable or evident in field observations.  Nordstrom et al  (I979b)
noted that jarosite was oversaturated in chemical speciation
calculations for some streams, but no obvious yellow iron precipitate
was seen in the reaches.  Field observations indicated that jarosite
precipitation occurred in the micro -environment of bacterial colonies.
Nordstrom et al  (1979b) conclude that this suggested the existence of a
kinetic barrier which hinders jarosite precipitation, but does not
hinder ferric hydroxide precipitation, and that this barrier is overcome
by the surfaces of bacterial colonies.

      Ion Exchange:  Ion exchange can be a significant control on the
chemistry of water in contact with sediments  (Drever, 1988) .  Ion

-------
                                                                     117
exchange may therefore influence cation  ratios  in pit  water,  perhaps
causing anomalous Na'/Ca2' or  Mg2*/Caa* ratios.   Possible ion exchange
reactions are:
                  MgX + Caa*  -  CaX + Mga*
                  Na2X +  Caa*    CaX  + 2Na*

where X denotes a clay mineral.  In the  first reaction, Mgr* in a clay
mineral is replaced by a Caa* ion,  and in the second, two  Na*  atoms  in a
clay are replaced by a CaJ* ion.  The first reaction removes one Ca2* ion
from solution and adds one Mg2*  ion to solution,  while  the second
reaction removes one Ca2* from solution and adds two Na' ions  to
solution.
      In experiments involving acidic tailings  fluid and  calcite-bearing
drill core, Davis and Runnells  (1987)  concluded that ion  exchange on
montmorillonite clay was responsible  for release of Ca2* to solution.
Ion exchange can be verified by chemical analysis of ion  exchangers
(WBL, personal communication).                             ......

Physical Factor*
      Bvapoconcentrations  Surface water bodies  in  the  arid climate of
the Great Basin are subject to high evaporation  rates which may exceed
the precipitation rates  (Figures 5-13 and 5-14;  see also  Herczeg and
Imboden, 1988).  For example, in 1984, Big Soda  Lake, near Fallen,
Nevada experienced 9 cm of precipitation,  but 120 cm of evaporation
(Kharaka et al, 1984).  Evapoconcentration of the pit lake should
increase the concentrations of dissolved ions in solution.
      Evaporation rates can be estimated by a wide variety of methods
(Gray, 1970).  Pan rates may overestimate the actual evaporation rate
due to higher heat loss than natural  water bodies,  requiring correction
by a conversion factor (Fetter, 1988).   Pan rates may also underestimate

-------
                                                                      118
            Note: Depths over 16
            inches not shown
V
                                                M-I
                                                 1       i
                                              \
Figure 5-13i  Average annual precipitation  in Nevada  (inches/year);  fro*
Nevada Division of Hater Planning  (1992).
actual rates due to wind speed across the lake and other  factors  (Gray,
1970).
     . A geochemical model conducted for Barrick Goldstrike's proposed
Betze Pit predicts that evapoconcentration will increase  the levels of
conservative elements by 2.25 times when steady state is  attained, some
200 years after mine closure  (BUI, 1991).

-------
                                                                     119
Figure 5-14  Average annual lake surface evaporation in Nevada
(inches/year), from Navada Division of Water Planning (1992).
      Xiianologyt  Many physical limnological factors will influence the
chemistry of pit lakes, including vertical mixing, temperature, and
dissolved gas profiles.  The extent to which vertical mixing occurs will
determine the likelihood of turnover  (Hetzel, 1983).  Turnover and
mixing will influence several aspects of pit lake chemistry, such as
redox, mineral phase and aqueous species stability, and biological
activity.  Important driving forces determining the occurrence of

-------
                                                                      120
turnover include wind energy,  chemical density  differences, solar
radiation, and lake geometry (Wetzel, 1983).  Pit  lakes could
potentially become anoxic  at depth, as the O, and Eh profiles  at  the
Berkeley Pit illustrate  (Figure  5-15).  As discussed earlier,  anoxic,
reducing conditions at the pit bottom, at low to neutral pH, could cause
elevated dissolved iron  concentrations.  Regular turnover and mixing
will favor a decrease in iron concentrations and precipitation of ferric
hydroxides.  A possible  drawback of iron hydroxide  precipitation is
increased turbidity in the water column, which  can  potentially reduce
the heat budget in a pit lake and cause stratification (Parsons,  1977).
 -* I
 -
 ~M
-MO- 
-IJO-
            -14*
              oo""V  'e   i
                                        0

                                       -Ifr

                                       .
                                      -'
                                      -140


                                            h
                                  tip  ijp
                                   (mV)
ii*
Figure 5-15i  Dissolved oxygen and Bh profiles  at  the Berkeley Pit (f
Davis and Ashenberg,  1989)          '
      The temperature of a pit lake will  influence its chemistry.
Colder water can hold more dissolved gases such as O, and CO,, which
will  influence many other factors.  Seasonal  changes in temperature and
gas profile of typical lakes are well known  (Figure 5-16) .
      The surface  area to depth ratio in  pit  lakes will be less than
most  natural lakes,  with the closest morphological analog being perhaps
Crater Lake, Oregon (Macdonald, 1992; Lyons et al,  in press).  Such

-------
                                                                     121
geometry may decrease the likelihood of  wind induced fall and spring
turnover, and increase the possibility of perennial anoxia in part of
the water column.
Spnnj lurnoMr Sumnxr itrMitaticA Fn tufnevw WinUr Mrttificttioft
0,V
IT o\
i 1
i i
ii i i i 1 1
-

,_





T 0,




1, . i ,














A ' ' T
IT o,
1
1
1
i
ii i i i
01020300 1020300 1020300 10 JO
S*













30
TtmpCCJ C , C C
Figure 5-15:   Idealized profiles of temperature and dissolved oxygen in
oligotrophic  (non-productive) lakes (after Hetzel, 1975).  The increase
in dissolved  O2 with depth is due to the greater solubility of  0, at
lower temperatures (from Drever, 1988) .
      The location and orientation of some mine site locations may
actually enhance the 'wind speed across the lake.  Cortex experiences an
apparent funnelling effect,  perhaps caused by the orientation of the
elongate pit roughly parallel to prevailing wind directions (DAB,
personal observation).   High freeboard (height of pit wall or shore
above water level)  of  a pit might inhibit the wind velocity across pit
lakes.
      Limnological models may help development and calibration of  pit
lake geochemical models.   The model CE-QUAL-R1 has been used to model
the limnology of the Gold Quarry Pit, using Pyramid Lake as an example
(Davis, 1993).  Pyramid Lake was deemed the "closest analog" even  though
the geometries of the  two lakes are very different (Lyons et al, in
press).
      Geothennal input:  Due to the anomalously high regional,  crust al
heat flow in the Great Basin, much of the groundwater is warmer than

-------
                                                                      122
 average.   Much of  the  geothermal water rises along the young, "Basin and
 Range,"  semi-vertical  faults that dissect Nevada (NGDC, 1983).  Since
 the ore  mineralization at  many Cenozoic age mines in Nevada is
 associated with faults,  it follows that some mine sites will be situated
 in areas of geothermal activity (Macdonald, 1992).   Therefore, the
 potential exists in many post-mine situations for the inflowing
 groundwater to be  significantly warmer than average.
       If warm groundwater  enters the base of a colder pit lake,  density
 driven vertical mixing may occur,  causing turnover on a frequent basis
 as the wanner,  less dense  water ascends to the surface.  This occurs at
 Crater Lake,  Oregon, a normally very cold mountain lake that experiences
 deep,  geothermal inflow and regular turnover (Williams and Von Her2en,
 1983) .  However, if the warmer water enters high in a cold pit lake,
*       i                    .
 thermal  stratification may result, causing a stable situation in which
 mixing and turnover are unlikely.

       Atmospheric  Gas  Exchange:  The ability of. the pit. lake to take up
 atmospheric CO, and O,  will influence the redox potential, pH, and the
 solubility and stability of mineral phases.  The pCO,  will affect pR and
 carbonate solubility,  which in turn influence concentrations of Ca, Mg,
 and alkalinity. The indirect impacts on pH-dependent metal solubilities
 could be significant.   Plummer et al (1983) found that the calculated
 mass transfer of pyrite was enhanced 10-fold in their system when closed
 to CO, versus being open to GO, input.
       The exchange of  0, will influence redox conditions  and the
 solubilities and concentrations of redox-dependent minerals.  Aa
 discussed earlier, the impact of Eh dependent metal solubilities could
 be equally significant, and may make the difference between trace or
 toxic quantities of metals released into the environment.
       The diffusion of Ot  into the pit wallrock could be  an  important

-------
                                                                     123
factor controlling dissolution of trace metals  and sulfate into the  pit
water.  Current models of pyrite oxidation assume  that the rate limiting
step is the rate of O, diffusion into the rock  (Davis and Ritchie,
1986) .  The Gold Quarry model applied sensitivity  analyses incorporating
three different porosities that affected O, diffusion into the pit, and
hence the mass transfer  (PTI, 1992).  The three models gave
significantly different mass transfer results.
      Many lakes with near neutral pH are slightly supersaturated with
CO, relative to atmospheric pCO,  (Wetzel, 1983) . Soil zone pCO,  is
generally much higher than atmospheric, whereas deep groundwater pCO,
may be significantly less than atmospheric  (Drever,  1988) .   The pCO, of
phreatic groundwater in limestone is usually above atmospheric  as well
(Drever, 1988).
      By manipulating the partial pressures of  CO2 and 0,,  the model  can
simulate oxic vs. anoxic, or open vs. closed systems.   Fixing the pC02
at atmospheric assumes the system is open to the atmosphere.  The lake
chemical model can be divided into a set of stratified cells, each with
a different pCO, or pO2.           "   --.-.   -                  '--.-

      Rock/water ratio:  The amount of rock exposed to solution (i.e.
rock/water ratio between pit wallrock and groundwater + pit  water) could
exert important control over the mineral mass transfer.  The rock/water
ratio will be a function of wallrock porosity,  permeability,  and the
water flux into the system.  The rate of water  flowing into  the pit will
directly influence the rock/water ratio, which  will have an  effect on
dissolution/precipitation rates.  Rapid inflow  rate may cause a higher
degree of undersaturation in the water, thus enhancing the dissolution
of wallrock minerals.
      Fractures can greatly increase the permeability,  and hence,  the
groundwater flow rate, through an aquifer.  This could have  a variety of

-------
                                                                     124
effects in pit water evolution.  First, if ore minerals such as sulfides
are associated with fractures and other structures, the sulfides will be
exposed to greater volumes of water than other minerals, causing higher
masses of metals dissolved into the water.  However, if ore minerals are
not associated with structures, or if secondary fracture permeability
has been introduced by blasting, the water flowing through fractures
could interact with relatively unaltered rock, in which case the
dissolution of metals may be suppressed.
      More than one aquifer may intersect the pit, such as an alluvial
(phreatic) aquifer, underlain by a series of confined or semi-confined
aquifers.  There may be a much deeper, fracture-controlled, aquifer,
perhaps in carbonates, similar to that being encountered in the Carlin
trend mines of Eureka, Co., NV  (HCI, 1992).
      Slope stability in submerged pits could be a complicating 'factor
introducing variation into mine water quality modeling.  Wall failure in
submerged pits may introduce large errors into the rock surface area
predictions of chemical models.  Slope stability of the pit wall has the
potential to  increase the rock/water ratio rapidly and dramatically'.
Sloughing has decreased the depth of the Berkeley Pit by 38 meters
 (Davis and Ashenberg, 1989), and the Ruth pit by 37 meters (Woodward-
Clyde Consultants, 1992).
      The most detailed model of pit water chemical evolution will
involve an estimate of the amount of water passing over minerals in the
pore spaces of the wall rock, perhaps in concert with kinetic data for
dissolution rates of  each mineral.  These data could be combined with an
estimate of pit  lake  volume and the masses of total metal to be
dissolved  into  it.  The pit inflow rates could be calculated from
groundwater  flow rate as determined from well data.  The rock/water
 ratio  would  require an estimate of pit wall porosity, which would be
 complicated by the complexity  of structures commonly seen in

-------
                                                                     125
hydrothermal ore deposits.  A consideration of rock water ratio should
incorporate a finely discretized grid, due to the probable variation in
the porosity and mineral percentages from site to site in the deposit.
An estimate of pit lake volume could be obtained from mine plans, as
could the volume or surface area of pit wall available for fluid
interaction.  By combining these data, one can estimate a flux (and a
rock/water ratio) through a given parcel of rock.  Ideally, this
information can be integrated into chemical models to determine the
volume of water that reacts with a specific volume of rock over a
specific time interval.
      Modeling the rate of inflowing groundwater substantially elevates
the level of sophistication of the overall pit water modeling exercise,
and was beyond the scope of this study.

      Number of Inputs/Output* in System:  The simplest modeling
scenario will be one in which the water balance can be expressed as:

                            (PJ  - E  (P8) + B
where:
      Pt m sources  of water to  the pit  (groundwater + precipitation)
      P0  sinks  of water  from  the pit  (pit outflow to groundwater)
      E * evaporation
In other words, the pit has filled to ultimate depth,  the system has
reached steady state,  and no significant inputs or outputs other than
those stated above exist.
      The contribution to the pit lake from sources other than
groundwater, such as surface runoff and precipitation,  will complicate
the model.  The largest contribution of metals into the Berkeley Pit
comes not from groundwater,  but from surface water draining mine
tailings  (Huang and Tahija,  1990) .  The mass of metals from this  source

-------
                                                                      126
may  be  impossible  to quantify and model,  and would require a detailed
water/solids budget  to the pit.
      The  rate  of  local precipitation may be a positive factor in
helping to dilute  the pit water in some regions.   In the arid west,
however, the site  of most of North America's precious metal pits,
precipitation will likely be far outweighed by evaporation (Figures 5-
13,  5-14).  The rate of precipitation in the Ely,  Nevada area (location
of Ruth mines,  and general location of the Cortez  Mine)  is 14 inches per
year, whereas the  evaporation rate from the Ruth pit lake is 45
inches/year (NDEP  files).  Net evaporation will probably be ubiquitous
across  Nevada and  the entire Great Basin,  and evapoconcentration will
contribute to higher dissolved solid concentrations in pit waters.
Evapoconcentration can be modeled by manual subtraction in MZNTEQA2,  but
is more easily  accomplished in PHREEQE.
      In most situations, the pit water will likely flow out of the pit
and  into the adjacent aquifer, under the influence of the local
hydraulic  gradient.:.   However, if -the pit lies on a 'steep slope,  the ~ -" -
water could potentially flow over the rim on the downs lope side of  the -
pit. This could introduce contaminants into a surface watershed much
more rapidly, and  over a larger area,  than would transport via
groundwater.                                        -

      Tla* Scales  The 
-------
                                                                     127
in contact with different lithologic/mineralogic  suites at different
stages as the pit fills.  A common sedimentary sequence appearing in
precious metal deposits is interlayered carbonates  and siltstones or
shales, as exists at Gold Quarry  (PTI, 1992) .   Two  potentially
significant scenarios could arise.  The first  is  of greatest concern,
that is if a non-buffering lithology is overlain  by a buffering
lithology which may not be in contact with  the rising pit water until
well into the future.  A more innocuous scenario  would be one in which
the shale overlies the carbonate, so that the  rising pit water is always
in contact with carbonates and some buffering  capacity is always
available as the pit fills.
      A pit water will probably reach equilibrium with some of the host
mineralogical suites in the wallrock (results  shown in later chapter),
                                            '         /
but the attainment of equilibrium may be many  years in the future.
      Modeling the pit only at ultimate depth  can be risky.  Ignoring
all the intervening years may neglect decades  of  contamination, with
dangerous consequences to local water quality.  An  example is the
Berkeley Pit, which is projected to top out (without human intervention)
in the year 2009, 27 years after dewatering ceased  (Davis and Ashenberg,
1989) .  Conditions at the Berkeley pit today are  such that immediate
attention is necessary (Duaime, 1992), and  local  officials are rapidly
trying to solve the problem before the pit  water  reaches the rim (Baum
and Knox, 1992).

      Hydraulic Gradient]  The direction and gradient of groundwater
flow may be an important influence on pit wall dissolution, as
illustrated by Hicks et al  (1991).  If sulfide minerals appear on the
hydraulically upgradient side of the pit, chances are that mineral
dissolution will be enhanced more so than the  situation in which the
metals exist on the downgradient side.  The downgradient side might be

-------
                                                                     128
more prone to armoring by mineral precipitation, and perhaps subject to
less dissolution than the upgradient  side.
      High groundwater flow rates may lead  to development of a seepage
face on the upgradient side of  the pit,  which may act to destabilize
that section of the pit wall and initiate pit wall failure.

      Anthropogenic Disturbance!  Post-closure anthropogenic disturb-
ances can have positive or negative impacts on water quality, but the
complications introduced by such disturbances might make modeling
efforts impossible.
      Pumping water from the pit lake may induce destabilizing
conditions, such as mixing or aerating and  moving anoxic-stable species
into oxic portions of the pit lake.
      At Ruth, Nevada, alkaline gold  mill tailings were discharged into
the Ruth Pit for several years  until  August 1987.  This resulted in a pH
rise to near neutral levels, and a rise in  cyanide levels to 5-7 mg/1 by
fall of 1989  (Dames & Moore, 1990).          - .         -
      Additionally H,SO  was used to leach gold from nearby  tailings,  _
some of which ran into one of the pits. Chemical analyses  for the Ruth
Pit from 1984-86 show pH values in the range of 3-5, whereas 1987
samples show pH up  to 6.7  (NDEP files).  In 1990, pH of both the Ruth
and Kimbley pits were in the neutral  range, whereas the pH  of the
Liberty pit was  in  the 2-3 range.
       Because of  the  significant anthropogenic disturbances in the Ruth
District and  Berkeley pits, they were not included in the modeling
exercises.

Other Factors
       Database limitationss   The largest source of discrepancy in
 chemical models is in the thermodynamic database  used by each model

-------
                                                                     129
(Nordstrom et al, 1979a).   The major species in dilute solutions are
relatively unaffected, but the discrepancies increase with higher ionic
strength and/or decreasing constituent concentration.  The problem
becomes more pronounced for trace metal speciation, which can show large
variation from small changes in equilibrium constants, pH, redox
potential, or temperature.

      Downgradient impacts:  The water in pit lakes should flow under
the driving force of a regional groundwater hydraulic gradient.  Unless
the water flux out of the  pit (evaporation + flow) exceeds the inflow
(which should result in a  dry pit),  the water should flow out of the pit
and enter an aquifer or watershed downgradient,  taking dissolved solids
with it.  Contaminants, if contained in pit lakes, probably pose little
threat to ecosystems or human water sources.  However, if allowed to
migrate into an aquifer or watershed that might be a source of water for
human consumption, the contaminants in pit water at some mining sites
could become a serious concern.  Downgradient impacts could be the most
important aspect of mine water quality, in the context of environmental
contamination, perhaps deserving equal attention in the overall mine
water quality modeling exercise.
      Once the pit water enters the aquifer and becomes groundwater,  the
>
system may become closed to atmospheric gas exchange.  Reducing
conditions may ensue, which will effect mineral solubilities and aqueous
speciation.  Sulfides of iron and other metals might precipitate.
However, if Eh/pH conditions are similar between the pit and the
aquifer, the metals might  remain mobile, be transported sufficient
distances, and threaten contamination of municipal water supplies.   The
                                                                /
presence of buffering minerals, such as carbonates, will maintain the pH
in the neutral range.  The Ruth district is surrounded by carbonate
rocks, which could have dramatic effects on the acidic and metal-laden

-------
                                                                      130
water that flows from the pits.
      The pit lake could conceivably  act  as  a net  discharge area
depending on its location in the regional hydraulic flow regime, and the
combined fluxes of evaporation, pit outflow,  and groundwater inflow.  In
such cases, contaminants will be contained in the  pit,  with no migration
downgradient.
      Although most groundwater flow  rates are typically slow,  on the
order of a few meters per year, large pits have the potential to
transport tremendous volumes of water and dissolved solids.   As an
example, the flow rate into the Berkeley  Pit  is estimated to be 7.6
million gallons per day (Davis and Ashenberg,  1989}, while  evaporation
is estimated at 0.08 millions gallons per day (Camp Dresser and McKee,
1988).  Under eventual steady conditions,  the balance is 7.52 million
gallons per day that could potentially flow out of the  pit  into the
downgradient aquifer.  Using iron as  an example, if the dissolved
concentration at 100 meters depth  (1040 mg/1)  is considered an average,
and assuming that all dissolved iron  is carried out, then over 29 metric
tons of iron could be transported out of  the  pit each day into the
aquifer downslope:

                  (1040 tag)      (1 kg)      (1 ton)     (3.7854 1)
   Ton* F per day  	 x  	  x 	  x  	  x 7.52 MOD
                  (1 liter)     (10* mg)    (1000 kg)      (1 gal)
                       29.6 metric tons  of Fe  per day.

Similar calculations reveal potential daily transport of 14.1 metric
tons of Zn, 6.2 metric tons of Cu, 5.9 metric tons  of Al, 0.05  metric
tons of Cd, and 201 metric tons of SOJ~.

-------
                                                                     131
                6.   PIT WATER MODELING  APPROACHES

      The development of the conceptual and numerical models for this
study proceeded in the following stages.  First,  the Cortez pit was
selected as an example location for the simulations.  An inverse model
was developed and used to determine mass transfer between the pit wall
and an upgradient,  initial  water to generate the  final pit water.  The
inverse model results were  used as input to the forward model in an
attempt to duplicate the real world condition.   Some of the questions
posed during conceptualization of both the inverse and forward models
include:
*  Is the quality of the available chemical analyses for the sites
   acceptable?  Are enough  data available for a valid model?
*  What minerals might reach equilibrium in the system,  such as the
   major host lithology (e.g. calcite, dolomite,  gypsum, silicates)?
   These will be defined as reversible reactions  in the numerical model.
*  What minerals are participating in irreversible (dissolution)
   reactions, and what are  their mass transfers?
*  What are the relative rates of dissolution of  pit wall minerals, and
   how can they be incorporated into the model?
*  What is the extent of gas exchange in the system,  and how does it
   influence pit water chemistry.
*  How does adsorption vs.  mineral precipitation  influence mineral
   partitioning between the aqueous and solid phase?
      This study addresses  two possible approaches for modeling pit
water chemical evolution.   The first is referred  to as the rate-
independent dissolution approach, and the second  as the rate-dependent
dissolution approach.

      Rate-independent dissolution:  The rate-independent dissolution
approach assumes uniform dissolution of pit wall  minerals independent of
reaction kinetics or mineral solubilities.  The model assumes that a
fixed thickness of pit wall (e.g. 1 foot) dissolves around the interior

-------
submerged surface of the pit, with resultant mass  transfer of all
constituent elements into solution.  A schematic representation is shown
in figure 6-1.  The method ignores the differential weathering that
could result from different mineral solubilities and/or different
dissolution kinetic rates.
                       Fixed thickness (i.e. 1 foot)
Figure 6-1s  Schematic cross-section of rat*-independent pit wall
dissolution model.
      The rate-independent approach greatly simplifies the system such
that it can be easily modeled.  Xn the absence of definitive or reliable
data on mineral proportions or dissolution rates, this may be  the only
approach for dissolving elements into the pit water.  The information
required to perform this simulation includes:

      1)  Inflowing groundwater chemistry.
      2)  Local precipitation and evaporation rates.
      3)  Masses of all minerals in pit wall rock.
      4)  Volume of pit lake  (ultimate or incremental depth) .
      5)  Aqueous speciation and equilibrium model,
         and adsorption model if desired.

-------
                                                                     133
      The approach simply dissolves the minerals into the pit lake,
giving a bulk concentration of solid per volume of solution in the pit
lake, from which the concentration can be easily calculated.  An
equilibrium model such as MINTEQA2 can be applied to simulate changes in
aqueous concentration after mineral precipitation, and an adsorption
model can be used to simulate further removal of selected dissolved
constituents.
      The risks of the rate-independent dissolution approach are
obvious.  Pit wall dissolution is likely not congruent,  since more
soluble minerals such as sulfides and carbonates should dissolve faster
and in greater quantities than silicates, clays, or oxides/hydroxides.
Also, minerals associated with structures and less competent rock such
as alluvium will probably dissolve in greater mass.
      The improbability of the rate-independent dissolution model is
demonstrated by table 6-1, which shows the mineral masses for the Cortez
pit wall rock and the Cortez pit water, plus mole ratios  for both. Table
6-1 reveals that iron is the least abundant element (on a mole-basis)  of
those tabled for the pit wall rock.  With the exception of aluminum, ..iron
Table 6-1 : Moles of element
wallrock, and concentrations
per kilogram of rock in Cortes pit
(mmol/1) of dissolved solids in pit water.
mole ratio
wallrock * (vs. Fe)
C
Ca
Mg
Si
K
S
Al
Fe
Source :
10.3 1030
5.7
4.7
4.6
0.2
0.02
. 0.5
0.01
* Wells and Mullens
** See Table 1-4.
570
467
465
16
2
49
1.
(1973) ;
pit water mole ratio
(mmol/1) ** (vs. Fe)
.4.63
1.13
0.74
0.57
0.30
0.94
0.0
0.0024
Wells et al
1928
472
310
70
125
239
0
1
(1969) .
is also the least abundant in the water.  However,  sulfur goes from a

-------
                                                                     134
mole ratio (S:Fe) of 2:1 in the wallrock  to a  ratio of 239:1 in the
water.  The likely source of sulfate  to the pit water is dissolution of
sulfides (shown by models in next chapter), which should introduce iron
and sulfate into the pit water in a 1:2 ratio.  Since sulfate can be
assumed to act conservatively in the  water  (not precipitating or
adsorbing), the S:Fe ratio of 239:1 indicates  that a significant mass of
iron is removed from solution, either by precipitation, adsorption, or
other mechanism.  In actuality, the iron  is removed via precipitation of
ferric-hydroxide, as demonstrated by  models in next chapter.
      The inescapable conclusion is that a rate-independent dissolution
model that dumps iron and sulfur into the pit  water in the proportions
that they exist in the pit wallrock will  ignore the processes that
control the partitioning of iron between the aqueous and solid phases.
This method will grossly underestimate the amount of iron that dissolves
into the pit water,  by approximately  2 orders  of magnitude.  A "bulk"
chemistry  (concentration before precipitation  and adsorption processes
are modeled)  two orders of magnitude higher than the ultimate    -----
concentration in solution is precisely the amount predicted by the ... ,
forward models in the next chapter. vx                
      Additional discrepancies between the wallrock and the pit water
are seen in aluminum and silica.  Dissolution of aluninosilicates is
expected to introduce silica to the pit water, but at neutral pH
aluminum will remain behind as clay or other weathering residue.   This
makes the behavior of silica difficult to predict,  but assuming
congruent dissolution of silicate minerals will likely overestimate the
amount of silica in solution, and may overestimate the buffering that
occurs from dissolution of silicates.

      Rate-dependent dissolution:  The rate-dependent dissolution model
acknowledges the probability that different minerals dissolve at

-------
                                                                     135
different rates, and incorporates this  into the mass transfer
calculations.  The rate dependent dissolution approach is a case of a
reaction path modeling exercise.
      In addition to the five parameters required for the rate-
independent dissolution model, the rate-dependent model requires
information on relative rates of reaction of pit wall minerals.  This
information may be available from experimental data in the literature,
but is best acquired from site-specific laboratory experiments such as
batch or column tests,  to determine relative rates of reaction of pit
wall minerals.
      A schematic of the rate-dependent dissolution model is shown in
figure 6-2, in which minerals associated with structures, ore zones, or
less competent rock units experience greater dissolution.
      This study has tried to duplicate a rate-dependent approach by
determining the mass transfer that has occurred in the system and
applying that information to forward models.  Although no information
has been incorporated regarding pit wall mineral masses or mineral
dissolution rates,  the mass transfer obtained from the inverse model has
been successfully applied in the forward model to duplicate the actual
field situation.  This exercise is analogous to the approach referenced
in Plummer (1984) in which kinetic information (apparent rates of
reaction)  is obtained indirectly through inverse modeling.

      Coupled:  The coupled, or reaction transport,  approach attempts to
combine aqueous geochemical reactions with the equations of hydrologic
advective transport.  Very few coupled codes are available (Bngesgaard
and Kipp,  1992; Nienhuis, 1991; Yeh and Tripathi,  1989),  as this is a
new sub-discipline of geochemical modeling and few studies applying
these codes have been performed.  Although consideration of the coupled
approach is beyond the scope of this thesis,  coupled models have the

-------
                                                                    136
potential to become important tools in future modeling studies.
Figure 6-2:  Schematic cross-section of rate-dependent pit wall
dissolution modal.

-------
                                                                     137
                        7.   MODELING  RESULTS

      The Cortez pit in Lander County, Nevada was selected as the
example site for the exercise that matched the forward and inverse
chemical models.  The Cortez Mine was an open pit precious metal mine
that was active up until the mid 1970's (Eric Vokt, Cortez Gold Mines,
personal communication).
      Speciation models were also performed for the Universal Gas pit in
Eureka County,  Nevada, which has been inactive since about 1983 (Denver
Knight Piesold, 1991).  These mines are classified as sediment-hosted,
disseminated,  precious-metal deposits.  The primary host lithology for
mineralization at both sites is the Silurian Roberts Mountain Formation
(Srm).  The Srm is primarily carbonate rocks (limestone and dolomite)
with minor siliceous interbeds (siltstone).  The Universal Gas pit is
located one kilometer northwest of the Carlin Gold Mine.  The locations
of the Cortez  and Carlin mine sites are shown in figure 7-1.
      Mining activity stopped at the Cortez and Universal Gas sites
several years  ago, and the water in the pits has apparently reached
static conditions.  These pits were chosen for several reasons:'  1) they
may be representative of lakes that will fill many pits left behind by
precious metal operations;  2) they are mostly undisturbed by
anthropogenic  inputs;  3) they have relatively simple water balance
situations, i.e. inflow is primarily from groundwater, plus periodic
storm event surface runoff;  4) the necessary hydrogeochemical and
\                                  ,
lithochemical  data are available, and of apparently acceptable quality
(determined by visual inspection and calculation of ionic balance).
                         \

Spcciation/Bquilibriua Models
      Speciation/equilibrium simulations were performed using pit water
data from the  Cortez pit and the Universal Gas pit.  For comparison

-------
                                                                  138
        NEVADA"
                            CARLIN  GOLD
      O&ENO
     I
                                              ELKO
                                       CORTEZ GOLD
                                          DEPOSIT
                            Sink
O
i
 EUREKA
Figure 7-1:   Location sap for Cortex and Carlin Mine* (from Nells et al,
1969).
purposes,  pit water analyses were run through WATEQF,  WATEQ4F, MINTEQA2,
and PHREEQE.  The simulations served three purposes:  1)  to gain an
understanding of the general state of the water, such as  chemical
speciation and saturation indices of mineral phases; 2) to compare the
capabilities of the different programs;  3) to check the  validity of the
data for application to the inverse simulations (BALANCE), and for
subsequent reaction path simulations (PKRBEQB, MINTEQA2).
      Cortex pit i  The original analytical data obtained  from Cortex
Gold Mines (Table 7-1) exhibited an ionic balance error of approximately
18% (data disk file CZSP01H4.0OT, speciation model of the average of the
three analyses).  The analyses also lacked data for dissolved silica.
The 1993 sampling effort was designed to fill the gaps in the existing
data set, and to improve upon the ionic balance.

-------
                                                                       139
TABLK 7-1:   Cortex pit water chemistry.
(sampled  June 1992},  values in ppm,  NA
 Sourcet   Cortez Gold Mines
 not available.
             Ca
             K
             Na
             HC03
             SS4

             Ii'
             Fe
             Mn
             As
             Pb
             pH
 end*       Middle

 44.2
 11.3
 72.8
 18.0
225
 86.5
 24.8
  1.78
 NA
  0.061
  0.145
  0.005
  0.038
 <0.005
 <0.0005
 <0.005

  8.02        8.07        8.13

  + Used as model input.
West
end
43.1
11.1
71.4
17.7
225
81.9
26.9
1.76
NA
0.060
<0.050
<0.003
0.040
0.007
0.00138
0.006

Average
43.5
11.3
72.2
17.8
226
64.67
26.53
1.77
NA
0.0603 +
0.134 +
0.0017 +
8.0383 +
.0043 +
0.00046 +
0.002 +
             8.073
      The  new data (Table 7-2; CZSP02W4.OUT) show a  much better ionic


balance  (-4%),  and also reveal that the original  ionic imbalance was


likely caused by an erroneous value for HC03".   The other major ions
TABLE 7-2:   Cortez pit water chemistry.  Source:  UNR/Cortez Gold Mines
joint sampling effort (1993).  Values in mg/1.


Ca
K
Na
H&>3
IT
N03
F
As
S102

pH
Eh (nV)
EC \fj9f\o\
Temp (C)

Surface
(0 ft.)
44.8
11.7
68.3
17.9
283
87.3
24.5
0.13
2.4
0.034
34.9

7.97
136
680
21.0
* Measured
+ Used as n
Mid
(20 ft.)
45.5
11.8
69.3
18.2
282
96.9
24.5
0.40
2.4
0.030
34.4
\
8.14
149
20.2
in the field.
lodel input.
Bottoa
(40 ft.)
45.9
11.6
-68.3
18.2
282
86.4
0 09
2.*4
0.024
34.0

8.09

20.2


Average
45.4 _^-+ 	 -""
11.7 +
68.63 +
18.1 +
282.3 +
90.2 +
24.4 +
0.2067 +
2.4 +
0.0293 +
34.43 +

8.07 + *
120 + *
682 *
20.5 + *

show reasonably good agreement between data sets.  The  1992 data set


included a  complete trace metal suite, so the 1993 sampling effort did


not include trace metals.  For model input, the major element suite from


the April 1993  collection was used due to the better major element ionic

-------
                                                                       140
balance.   These data were used in conjunction with the trace element

data collected by Cortez Gold Mines  in  1992.   The mixing of these  data

sets is not  expected to introduce any significant error to the model.

The combined data sets used as model input  are shown in Table 7-3.
TABLE 7-3:   Input concentrations for Cortex  pit water chemical modeling
simulations.   Source:   * ONR/Cortez Gold Mines joint effort  (1993);
** Cortex Gold Mines sample (1992).
             Alkalinity, bicarbonate
             Chloride
             Fluoride
             Nitrate Nitrogen
             Solids, Dissolved (TDS)
             Sulfate
             Arsenic
             Barium
             Calcium
             Iron
             Mercury
             Potassium
             Magnesium
             Manganese
             Sodium
             Lead
             Silica
             Zinc

             pH
                                         (ppml
282.3
 24.4
  2.4
  0.0467
432.3
 90.2
  0.0383
  0.0603
 45.4
  0.134
  0.0004
 11.7
 18.1
  0.0017
 8.63
  0.0043
 34.43
  0.002

  8.067
      Universal Gas Pit:   Table 7-4 shows  the  chemical analytical data

from the Universal Gas pit which were used as  input for the models.  The

1991 analyses were used for all components with the exception of nitrate

and ammonia.   The activity ratios for the  nitrogen species were used to

calculate  the pe/Eh of the water.

      The  chemical analyses for the Universal  Gas pit water shows some

similarities  to the Cortez pit water, but  also some key differences,

such as chloride, aluminum, sulfate, alkalinity,  calcium,  and a few

trace metals.  The input and output for all  speciation simulations for

the Cortez and Universal Gas pits, and for well SC-5B, are on datadisk

1.  The names and contents of the files are  listed in Table 7-5.

      The  speciation simulations for the Cortex pit water show some

predictable results.  A portion of the actual  output file (CZSP01W4.0UT)

is shown  in Table 7-6.  In all output sets,  both pit waters and the well

-------
                                                                            141
TABLE 7-4:   Hater chemistry,  Universal Gas pit.  Source:
Miller (values in ppm) .
                                                                 Oeraghty &
      Alkalinity, bicarbonate
      Alkalinity, total
      Chloride, titrimetric
      Chloride
      Conductivity, in mhos/cm.
      Cyanide, total
      Cyanide, weak acid diss.
      Fluoride
      Ammonia
      Nitrate Nitrogen
      Nitrate
      Solids, Dissolved  (TDS)
      Sulfate
      Silver
      Aluminum
      Arsenic
      Boron
      Barium
      Cadmium
      Calcium
      Cobalt
      Chromium
      Copper
      Iron
      Mercury
      Potassium
      Maor.es ium
      Sodium
      Nickel
      Lead
      Selenium
      Silicon
      -Tin
      Strontium
      Tellurium
      Titanium
      Thallium
      Tungsten       -->-,=.
      Vanadium              .
      Ziac
                                               Sample date
                                         4/23/91       3/12/90
       PH
  77.5
  85.5
 342
  NA
  NA
  NA
  NA
  0.394
  NA
  0.111
  NA
 691
  30.7
 < 0.02
  0.174
 < 0.180
  0.185
  0.12
 < 0.007
 145
  0.02
 < 0.01
 < 0.007
  0.134
 < 0.5
  3.74
  38
  50
 < 0.015
 < 0.05
 c 0.13
  9.11
 < 1.3
  0.514
 < 0.075
 < 0.001
 < 0.1S
: -0.051.
  0.00.7.
 < 0.005-
                                                            .2
                                                            .13
111.3
 97
 NA
198
903
  0.06
< 0.02
  0.
  0.
  1.3
  5.7
550
 23
< 0.01
< 0.2
< 0.01
  0.12
  0.388
< 4.005
 94
< 0.05
< 0.01
< 0.025
< 0.1
< 0.0002
  8
 21.1
 38
< 0.04
< 0.003
< 0.005
  53.7
  NA
  NA
  NA
  NA
  NA
                                                          < 0.02

                                                           7.74
                    Source:  *   Westmont Gold (NDEP  files)
                              Geraghty * Miller, Inc.
are near equilibrium with respect to the  carbonate  minerals  aragonite,

calcite, dolomite,  magnesite, and siderite.  The pit waters  are slightly

oversaturated, whereas the well  is slightly undersaturated.   This is

expected, since  both pits are situated  in carbonate host rocks, and well

SC-SB is emplaced in the carbonate Srm  formation.

       The pit waters are also near equilibrium with respect  to several

silica and iron  oxide phases.  The silica phases are no surprise, since

siltstones are interbedded with  the carbonates in all three  settings.

       The speciation of  the Universal Gas pit water shows a  slightly

more  saline composition  (ionic strength  0.017, or 10'1-77) than the

-------
                                                                                    142
Table 7-5:   File  names and contents of  speciation model output files,
                         Cortex pit  water Bpeciatlon simulations
CZSP01WF.lt)
CZSPOIWF.OOT
CZSP01W4.DAT
CZSP01W4.0UT
CZSP02W4.DAT
WATBQF input file
WATBQF output file
WATEQ4F input file
WATEQF output file
WATEQ4F input file  *
  CZSP01MT.IN
  CZSP01MT.OUT
  CZSP01PH.IM
  CZSP01PH.OOT
  CZSP02W4.0UT
     MINTEQA2 input file
     MINTEQA2 output file
     PHREEQE input file
     PHREEQE output file
     WATEQ4F output file 
OGSP01WF.IN
DGSP01WF.OUT
UGSP01W4.DAT
OGSP01W4.0UT
                      Dniversal Gas pit water speciation simulations
WATBQF input file
WATBQF output file
WATEQ4F input file
WATEQ4F output file
  OGSP01MT.IH
  OGSP01MT.OOT
  OGSP01PH.IN
  OSSP01PH.OUT
     MINTEQA2 input file
     MIHTEQA2 output file
     PHREEQE input file
     PHREEQE output file
SC5B01W4.DAT
SCSB01W4.OUT
SCSB01MT.IN
SC5B01MT.OUT
                            Well SC-5B speciation  simulations
WATEQF input file
WATEQF output file
MINTEQA2 input file 
MINTEQA2 output file 
  SCSB02MT.IN
  SC5B02MT.OUT
  SC5B01PH.IK
  SCSB01PH.OOT
     MINTEQA2 input file
     MIKTEQA2 output file
     PHREEQE input file
     PHREEQE output file
 *   Second Cortez pit water speciation,  using 92-93 composite.

 **  This simulation executed to determine Eh using specified  redox couple for the actual
     speciation in next simulation.
 Table  7-6:   Cortez  pit water  speciation, portion of output file
 CZSP02N4.00T showing saturation  indices.
                    Phase

                Aragonite
                Barite
                Calcite
                Cerrusite
                Chalcedony
                Chryaotile
                Cristobalite
                Diopside
                Dolomite  (d)
                Dolomite  (e)
                Fe3(OH)8
                FeOH)2.7C1.3
                Ferrihydrite
                Fluorite
                Goethite
                Greenalite
                Gypsun
                Hematite
                Maghemite
                Magnesite
                Magnetite
                Quartz
                Sepiolite(c)
                Siderite  (d)
                Siderite  (c)
                Silica gel
                Si02  (a)
                Talc
                Tremolite
                ZnSi03
                   lAP/KT

                    .449
                    .1C8
                    .S9C
                  -1.77B
                    .357
                   -.878
                    .401
                   -.983
                    .516
                   1.084
                   3.266
                   6.487
                   1.939
                   -.423
                   7.830
                   2.182
                  -1.846
                  17.321
                   7.273
                   -.080
                  19.185
                    .800
                   -.034
                   -.555
                   -.143
                   -.180
                   -.498
                   3.473
                   6.279
                   1.286
 Log  ZAP

 -7.860
 -9.876
 -7.860
-14.962
 -3.248
 31.888
 -3.248
 19.274
-15.900
-15.900
 23.488
  3.447
  6.830
-11.077
  6.830
 22.992
 -6.427
 13.659
 13.659
 -8.040
 23.489
 -3.248
 15.846
-11.005
-11.005
 -3.248
 -3.248
 25.393
 63.941
  4.421
 Log KT

 -.309
-10.044
 -8.456
-13.185
 ->.604
 32.766
 -3.649
 20.257
-16.415
-16.984
 20.222
 -3.040
  4.891
-10.654
 -1.000
 20.810
 -4.581
 -3.662
  6.386
 -7.960
  4.304
 -4.048
 15.880
-10.450
-10.862
 -3.068
 -2.749
 21.920
 57.662
  3.135

-------
                                                                     143
      Cortez pit water  (ionic strength  0.00845 or  10'J 7) .  A portion
of the output file from speciation model UGSP01W4.0UT  is shown in  Table
7-7.  The pH in the pit (8.67) is higher than the groundwater  (7.06),
suggesting that the buffering effect of carbonates is  enhanced by  the
system being open to CO,.
      Discussions  The advantages of each code become  apparent when the
output files are compared for the speciation models of  the Cortez  pit
water (Table 7-8) and the Universal Gas pit water (Table 7-9).
      WATEQF:  As the tables show, WATEQF contains no database for
several trace elements present in the Cortez pit water, and that may be
important in other pit waters derived from precious metal hydrothermal
deposits (As, Hg, Pb, Zn) .   This deficiency limits the  effectiveness of
WATEQF in pit water modeling applications,  and its use  is not
recommended.
      WATEQ4F:  WATEQ4F can manage all of the aforementioned elements
except mercury.  In addition, HATEQ4F can handle other potentially
important mine-derived trace metals, including Ag,  Cd,  Cs,  Cu, Ni, Rb,
Se, and U.  WATEQ4F also maintains a very flexible and comprehensive
approach to redox conditions.  WATEQ4F should serve adequately in
speciation/equilibrium of all pit waters,  except those for which mercury
speciation modeling is desired, and should be included within the
overall pit water modeling exercise.
      MINTEQA2:  The size of the databases  in MINTEQA2  and WATEQ4F are
similar.  The number of minerals in the MINTEQA2 model  is slightly
higher for each sample than in the WATEQ4F model,  whereas the number of
aqueous species is slightly higher in WATEQ4F.   Both of these codes are
relatively easy to use, and generate the same basic information in the
speciation model output files.  MINTEQA2 offers more flexibility than
WATEQ4F in some aspects (e.g. calculation of activity coefficients,
adjusting partial pressures of gases),  but  the advantages

-------
                                                                    144
Table 7-7: Universal Gas pit water speciation,
UGSP01W4.0UT showing saturation indices.

Phase
Adularia
Albite
Allophane(a)
Allophane(P)
Annite
Anorthite
Aragonite
Ba3lAs04)2
Barite
Beidellite
Boehmite
Calcite
Chalcedony
Chlorite 14A
Chlorite 7A
Chrysotile
Clinoenstite
Cristobalite
Diaspore
Diopside
Dolomite (d)
Dolomite (c)
Fe3 (OH) 8
FeOH)2.7C1.3
Ferrihydrite
Fluorite
Gibbsite (c)
Goethite
Hematite
Huntite
Illite
Kaolinite
Kmica
Laumontite
Leonhardite
Maghemite
Magneaite
Magnetite
Manganite
Montmoril BP
Montmoril AB
Montmoril Ca
Phillipsite
Prehnite
Pyrophyllite
Quartz
Rhodochrs (d)
Rhodochrs (c)
Sepiolite (d)
Sepiolite(c)
Silica gel
Si02 (a)
Strontianite
Talc
Tremolite
Wairakite
Log
IAP/KT
.691
-.522
-.971
.116
34.496
-.495
.946
11.779
-.193
2.299
.164
1.069
.032
10.082
6.710
3.676
-.556
.067
1.869
2.036
1.400
1.950
.224
6.581
1.873
-1.622
.637
7.764
17.537
-.670
2.272
3.021
- - - 7 . 574
3.712
. 15.259
7.143
.280
16.709
-1.549
5.244
4.970
2.099
.671
2.630
6.402
.461
-.740
-.000
-.609
2.291
-.502
-.808
-.904
7.438
16.122
. -.540

Log IAP
-19.882
-18.524
7.895
7.895
-51.149
-20.209
-7.391
-38.331
-10.163
-42.973
8.748
-7.391
-3.520
78.462
78.462
35.876
10.786
-3.520
, 8.748
21.930
-15.140
-15.140
20.446
3.541
6.764
-12.222
8.747
6.764
13.529
-30.638
-37.995
10.456
20.27?
-27:248
-54.497
13.529
-7.749
20.446
23.791
-29.669
-24.718
-42.928
-19.203
-9. 065
-41.912
-3.S20
-11.130
-11.130
18.051
18.051
 -3.520
-3.520
-10.175
28.837
72.696
-27.248
portion o output file

Log XT
-20.573
-18.002
8.866
7.778
-85.645
-19.714
-8.336
-50.110
-9.970
-45.272
8.584
-8.480
-3.551
68.380
71.752
32.200
, 11.342
-3.587
6.879
19.894
-16.540
-17.090
20.222
-3.040
4.891
-10.600
8.110
-1.000
-4.008
-29.968
-40.267
7.435
.-. . -12.J03 	
-30.960
_ ___-. 756
" 6;38C - -
-8.029
3.737
25.340
-34.913
-29.688
-45.027
-19.874
-11.695
-48.314
-3.980
-10.390
-11.130
18.660
15.760
-3.018
-2.712
-9.271
21.399
56.574
-26.708
for simple speciation models are relatively insignificant.   The biggest
advantages of MINTEQA2 are seen in the implementation of forward models,
which are performed later.

-------
                                                                     145
Table 7-8:
peciation











Comparison of
initiation* .

Ionic strength
Total aqueous
gpecies
Total minerals
Number of
iron species
Number of
arsenic species
Number of
mercury species
Number of
lead species
Number of
zinc species
log pCO,
Oversaturated
minerals
portion*
HATEQF
.0093
59
59
14
0
0
0
0
-2.67
6
of output
KATBQ4P j
.0093
12t
130
26
9
0
20
IS
-2.67
20
files
1INTEQA2
.0116
127
143
20
9
13
20

16
-2.37
24
for Cortex pit water
PHRKEpg
.0093
57
IOC
B
3
3
3
- 7
-2.67
16
      PHREEQE:  The unrevised PHREEQE  (as obtained from the USGS)
contains no thermodynamic data for the trace elements As, Hg, Pb, and
Zn.  PHREEQE was customized in this study, specifically for the Cortez
pit water simulations,  by adding the following eight elements to the  .
permanent database:

                  Arsenic            Mercury
                  Cadmium            Silver
                  Copper             Thai1ium
                  Lead        .       Zinc
Additionally, 113 aqueous species and 130 minerals composed of these
elements were permanently added to the" PHREEQE thermodynamic database.
This expanded version of PHREEQE can now model a limited number of
                                                                i
minerals and aqueous species for all ions of interest in the Cortez pit
water, including Hg (Table 7-8) .  The ability to modify or expand the
database represents the biggest advantage of PHREEQB over HATEQF or
WATEQ4F.

-------
                                                                      146
Table 7-9:  Comparison of portion*  of  output !! for Universal Gas pit

water speciation simulation*.




                             MATEQF   WATEQ4F   MINTEQA2   PHREEQE

            Ionic strength      .0170    .0170     .0175     .0170
Total aqueous
species
Total minerals
Number of
iron species
Number of
arsenic species
log pCO,
Over saturated
minerals
81
89

14

0
-3.91

30
119
109

26

9
-3.91

40
96
118

20

9
-3.S9

45
45
77

S

2
-3.92

27
      The principal disadvantage  of  PHREEQE is that it is the most


difficult of these codes to learn.   This  drawback,  plus the smaller off-


the-shelf database, make PHREEQE  less  desirable for simple speciation


models than either WATEQ4F or  MINTEQA2.   Learning to apply PHREEQE to


even simple tasks, such as ordinary  speciation simulations, is a chore


which researchers may  find impractical.   MINTEQA2 or WATEQ4F can be


learned  in less  time,  and have a  larger trace metal database than even
                                          *v

the expanded. PHREEQE.


      The time necessary to expand the PHREEQE database with the


necessary elements, species and minerals  can be immense (several weeks


for this study), but after its done  once,  the code can be used for a


variety  of simulations.  The ability to move easily between speciation


and reaction path models provided significant flexibility in the pit


water models, and represents the  biggest  advantage of PHREEQE.    The


output files  for the Cortez and Universal Gas pits illustrate the


relative sizes  of the'thermodynamic  databases of, each code, and the


usefulness of each  to  the application  of  speciation modeling.  Clearly,


the  databases of WATEQ4F and MINTEQA2  exceed the others, and provide the


most  comprehensive  speciation  models.  The only advantage WATEQ4F has

-------
                                                                     147
over MINTEQA2 for speciation modeling is in speciating a redox problem
for which no pe/Eh is available, but for which data are available for a
specific redox couple.  WATEQ4F can speciate in only one iteration,
whereas MINTEQA2 requires two.

Inverse Model
      The inverse model attempts to determine the chemical mass transfer
that has occurred along a hydrologic flow path.  The USGS computer code
BALANCE (Parkhurst et al, 1980) was used to determine the mass transfer
that occurred between the upgradient groundwater and the Cortez pit
water.  In the case of pit water modeling, the mass transfer occurs
during rock/water interaction between the upgradient groundwater and pit
wall minerals  atmospheric gases.
      Input:  The input required for an inverse model includes chemical
analyses for two waters  (initial and final) along a hydrologic flow
path, and a set of mineral phases (including gases) believed to be
responsible to produce the second water from the first through
dissolution and precipitation of the mineral phases.
      Water chemistry:  The final water (i.e. the "final well") for the
pit water inverse simulation is obvious, namely the pit water chemical
analysis.  As the analytical data in Table 7-2 illustrate,  there is no
significant chemical stratification evident in the Cortez pit water, so
an average of the three depths was deemed satisfactory.  Furthermore,
the water has possibly experienced mixing anyway due to active pumping
from the lake.
      Selection of an initial water for the Cortez pit water model
presents a small problem.  According to mine personnel, there is no well
located upgradient of the Cortez pit that can be used as representative
input water to the pit, so an exact initial water chemistry is not
available.  Monitoring wells located downgradient or lateral to the pit

-------
                                                                     148
show anomalously high levels of some trace metals  (NDEP,  1992) ,  and have
probably been influenced by ore mineralogy, or contaminated by anthropo-
genic activity.   Whatever the source of the metals,  these waters would
not be representative of groundwater immediately upgradient of the
Cortez Pit.  Consequently, chemical analytical data  for these  wells were
not used in the simulations.
      However, the host rock formation at both the Cortez and  the
Universal Gas sites is the Silurian Roberts Mountain Limestone (Srm).
The water chemistry from a well (Well SC-SB) approximately 200 meters
northeast of the Universal Gas pit and hydraulically upgradient (Denver
Knight Piesold,  1991) may reasonably represent the groundwater upgradi-
ent of the Cortez pit.  Monitoring well SC-5B was, therefore,  selected
as the initial water for the Cortez pit water models.
      Well SC-5B is approximately 115 feet deep, and taps a deep
alluvial aquifer at the alluvium/bedrock contact.  The water chemistry
in Well SC-5B (Table 7-10) is consistent, with that of the Universal Gas
pit, indicating that this aquifer is the likely source of the  Universal
Gas pit water (Knight Piesold, 1992).
      Unfortunately, the use of an initial water not along the hydro-
logic flow path makes the inverse model invalid.  However,  the framework
of the model remains valid, and only the actual input numbers  are
different.  Once the correct data become available,  they  can be easily
incorporated into the model to generate valid, site-specific results.
The  scope of this study only permitted the use of available data and
resources.  Funding was not available to drill new monitoring  wells.
      Some general predictions of possible mass transfer  are possible
 through brief examination of the two water chemistry samples.   For  ions
 which show higher concentrations in  the pit water than in the  well  (e.g.
 alkalinity, sulfate,  magnesium, sodium) we may predict that one or  more
 mineral phases  containing these components must be  dissolving  along flow

-------
                                                                     149
path.  For ions which show lower concentrations  in the pit water  (e.g.,
calcium, iron, zinc) some type of removal mechanism is at work, such as
mineral precipitation or adsorption.
TABLE 7-10:  Chemical analyses for Well SC-5B, Universal Gas site,
Nevada (sampled 3/12/90, source: Geraghty 6 Miller).
                                             pom
Alkalinity, bicarbonate
carbon
Chloride, titrimetric
Fluoride
Solids, Dissolved (TDS)
Ammonia
Nitrate Nitrogen
Nitrate
Boron
Sulfate
Sulfur
Barium
Calcium
Iron
Lithium
Potassium
Magnesium
Manganese
Sodium
Silica !Si03)
Silicon . ,
Aluminum
Arsenic v
Copper
LeaS
Mercury
Zinc
PH
112.2
22.1
(6
O.S
322
0.07
0.7
3.1
0.11
37
12
< 0.2
51
0.34
NA
S
13.8
0.079
31
83.22
38.90
< 0.2
0.01
< 0.02S
< 0.003
0.0002
0.022
7.0C
      Phases:  An inverse model requires a set of potential mineral
phases, including gases, that may react with the initial water to result
in the final water chemistry.  Selection of potential mineral phases was
based on publications and reports for the geology of the Cortez mine
site, the Roberts Mountain formation, and the aquifer geochemistry
(Denver Knight Piesold, 1991; Radtke et al, 1987; Wells and Mullens,
1973; Wells et al, 1969).
      The primary host lithology at both the Cortez and Universal Gas
sites is the Silurian Roberts Mountain Formation  (Srm), which is
believed to be up to 470 meters thick (Denver Knight Piesold, 1991).
The Srm consists of dolomite and limestone with siliceous  (chert),
silty, and argillaceous interbeds.  A minor host at the Universal Gas

-------
                                                                     150
mine is the Ordovician Vinini Formation  (Ov), consisting of shales,
siltstones, and chert with minor guartzite and limestone.  The phases
selected for the inverse model are shown in  Table 7-11.
Table 7-11: Mineral and gas phases selected for Cortex Pit water
inverse model.
Calcite
Ca/Na EX
Illite
Dolomite
Goethite
Gypsum
Ca-Montmo
Na-Montmo

K-Mica
Galena
CO, gas
Fluorite
Pyrite
SiO,
K-Montmo
=SOH : Zn

Magnesite
Sphalerite
Rhodochrosite
Arsenopyrite
Cinnabar
Sphalerite
Plagioclase
"CH2O" (organic

NaCl
MnO,
Gibbsite
Barite
K- Feldspar
Kaolinite
Mg/Na EX
matter)
      Results:  BALANCE found 1716 possible combinations of phases that
could account for the water evolution, but only 2 that satisfied the
constraints specified.  The calculated mass transfer of the selected
model is shown in Table 7-12.  A positive number indicates addition to
Table 7-12: Mass transfer model
(concentrations in juaol/kg.):







Plagioclase + F
Fluorite + F
-SOH : Zn F
Galena + F
Cinnabar + F
Arsenopy + F
Nad + F
Barite + F
Calcite F
calculated by BALANCE, first iteration
106.9250
50.0130
- .3059
.0208
.0013
.3772
1583.3785
.4391
2787.7720
MnOj F r 1.4091
Gypsum -3031.0075
Pyrite + 1238.1746
K-Montmo -767.3370


Illite
SiO2
Goethite
707.6120
-740.1979
1242.1746
 solution via mineral dissolution or in-gassing, and a negative number
 indicates removal  from solution by mineral precipitation or out-gassing.

-------
                                                                     151
The phase =SOH:Zn was arbitrarily., chosen as a  removal  mechanism for
zinc, and the negative number indicates removal  from solution via
adsorption.  The earlier predictions of dissolution  of mineral phases of
carbonate  (calcite) , sodium (Nad), magnesium  (illite), and  sulfate
(pyrite) have held true, as have predictions of  precipitation of mineral
phases of calcium  (gypsum),  and iron (goethite).  To help  interpret
whether these are believable,  we can refer to  the output from the
WATEQ4F calculations.
      The WATEQF output for the Cortez pit water data  shows  several
minerals near equilibrium or oversaturated, including  carbonates, iron
oxide minerals,  and silica phases.  These minerals may actually be
precipitating at some point along the hydrologic flow  path (e.g. in the
pit water) , so the models that predict precipitation of these phases may
be plausible.
      An important contrast in the chemical analyses for the  initial and
final waters pertains to dissolved zinc.  Most trace element
concentrations are higher in the final  (pit) water than in the initial.  .
(well), with the exception of zinc.  This prompted the definition of a
generic sink for zinc in the final water.  The phase *SOH:Zn,  possible
adsorption onto a solid phase, was arbitrarily chosen  to represent a
removal mechanism for zinc.
      A few comments are warranted regarding the BALANCE model results.
This model may only remotely resemble the actual mass  transfer that has
occurred along the flow path from Well SC-5B to  the  Cortez pit water.
Potential effects of evapoconcentration or other processes (i,e,
increases in concentration of "conservative" ions) were not considered
in the model.  Hence, BALANCE had to devise sources  for some  ions
(chosen from the phases provided by the modeler), such as  halite (NaCl)
and pyrite  (FeS,),  which showed an increase between the initial and
final waters.  The mass transfers of these ions  determined by the model

-------
                                                                     152
may significantly exceed that which  actually occurred  in  the  system,  if
the mass transfer occurred at all.
      To account for the desired final  concentration of S,  the  model  had
      ^
to increase pyrite dissolution, which also  increased Pe in  the
proportion equal to the stoichiometric  ratio of pyrite  (1:2).   However,
the final Fe concentration (Cortez pit  water analysis) is less  than the
initial  (Well SC-5B), so BALANCE removed the excess Fe by precipitating
goethite  (FeOOH).
      Calcite dissolution accounted  for the mass transfer of  bicarbonate
alkalinity from the well to the pit  water,  but resulted in  an excess  of
calcium, which was removed via precipitation of gypsum  (CaSO,-2H20) .
      Chloride in the initial water  is  higher 186 ppm) than in  the  final
(24.4 ppm), and perhaps reflects artifacts  of anthropogenic influence.
Constraining the, model by including  chloride would have forced  BALANCE
to devise a means of removing chloride  from the final water via
precipitation.  The only chloride phases in the BALANCE database are
evaporites  (e.g. halite, NaCl and sylvite,  KCl), but the Cortez pit
water ia  far too dilute to favor precipitation of evaporite minerals. 
For this  reason, chloride was not constrained in the BALANCE  models.
      Sodium was accounted for through  dissolution of plagioclase
 (selected stoichiometric ratio Al^Caj.jNao.jSi,.,), and halite (MaCl) .
The existence of halite in the Cortez pit system is hypothetical, but
other potential sodium sources are also difficult to determine.  The use
of plagioclase to account for all of the sodiua would have  increased  the
mass  transfer of Al, Ca, and Si proportionally and may have caused  the
model to fail.
       The effect of  interacting variables is once again demonstrated  in
 the modeling effort, as well as the  "balancing- act that the  code must
 perform to  obtain a  fit to the mass  transfer model.  Some of  the model
 results shown  in Table 7-12 can be  rightly  questioned, such as  the

-------
                                                                     153
precipitation of gypsum and dissolution of halite.  The modeler must use
good geochemical "common sense* to interpret the validity of the
possible models generated by the code.
      The limitations of the model, and the inability to account for all
of the processes at work in the chemical evolution of the Cortez pit
water, prevent the formulation of a model that matches the real world
exactly.  However inaccurate the results may be, they represent the best
mass transfer model obtainable for the Cortez pit water at this point.
Hence, with the mass transfer calculated, the inverse results can now be
applied to the forward model.

Forward Models
  i
      The forward model determines the chemistry of the final water, in
this case pitwater, which results from reactions between the upgradient
groundwater and the pit wall rock, and subsequent precipitation and
adsorption reactions in the pit water.  The results from the BALANCE
model  (Table 7-12) were run through PHREEQE to model the addition .of
dissolved species to the pit water via mass transfer from the pit wall.
The minerals that were chosen to introduce specific major elements and
trace metal into the pit in the required concentrations are shown in
Table 7-X3.
      Mass Transfert  An assumption that may be incorporated into pit
water chemical models is that the water will eventually come to
equilibrium with some of the primary host mineralogies.  Speciation
simulations for existing pit waters  (Cortez, Table 7-6, CZSP02W4.OUT;
Universal Gas, Table 7-7,  UGSP01W4.OOT) show equilibrium with respect
to several carbonate and silica phases  (e.g. calcite,  dolomite,
amorphous silica, quartz) which may be primary minerals in the carbonate
and siltstone lithologic units.  Another example is the Liberty pit
water,  for which the speciation model  (LISP01PH.OOT)  shows equilibrium

-------
                                                                      154

with several silica minerals  (amorphous silica,  chalcedony, quartz) that

might comprise the quartz monzonite  host rock.   However,  aluminosilicate

phases that may also comprise the host  (albite,  microcline, muscovite,

kaolinite, and gibbsite) are significantly undersaturated.   As later

models demonstrate (sensitivity analyses),  a solution in equilibrium

with these aluminosilicate phases will  likely be buffered in the range

of pH 5-6.  Since this pH is significantly greater than that seen in the
                    i
majority of igneous hosted pit waters,  the assumption of equilibrium

with aluminosilicate minerals is probably not valid.   Data  for the

Berkeley pit are lacking in silica concentrations (saturation

insufficient),  so no comparison is available.
      Table 7-13:  Minerals used in mass  transfer reaction model*.
            Ion

            Calcium

            Carbon (alkalinity)

            Chloride
            Magnesium
            Potassium

            Sodium

            Silica
            Sulfate
            Arsenic
            Barium
            Fluoride
            Iron
            Lead
            Manganese
            Mercury
            Zinc
 Source Mineral _

Calcite  (CaCOj)
Dolomite  (CaMg(C03),
Calcite  (CaCO,)
Dolomite  (CaMg(CO,),
Halite (NaCl)
Dolomite  (CaMg(C03),
K-Feldspar  (KAlSi,0.)
K-Mica (KAljSijO,0(OH),
Halite (NaCl)
Plagioclase  (Ca0 sNa0 SA11
K-Feldspar  (KAlSi3O,j
K-Mica (KAljSi,010(OH),
Plagioclase  (Ca0 sNa0 ,Alt
Sulfides listed below.
Arsenopyrite  (FeAaS)
Barite (BaSO4)
Fluorite  (CaF,)
Pyrite (FeS,)
Galena (PbS)
Rhodochrosite (MnCO,)
Cinnabar  (HgS)
Sphalerite  (ZnS)
SO()
                                                                ,Ot)
      Since calcite and dolomite constitute  a large  fraction of  the  host

rocks at the Cortez deposit, they were  specified as  reversible reactions

in PHREEQE  (allowed to attain equilibrium) .   The pH, Eh,  and partial

-------
                                                                     155
pressures of O, and CO, were taken from the earlier pit water speciation
models of the Cortez and Universal Gas  pit waters,  which indicated that
the pCO, is generally in the  range of 10'2", slighter higher than
atmospheric.
      A portion of the PHREEQE output  file CZRX01PH.OUT showing the
final concentration of the pit water is shown  in Table  7-14.
Table 7-14:  Concentration of pit water after  PHRBBQH ma*
model (CZRX01PH.OOT).
        transfer
         ELEMENT
                      MOLALITY
                                    LOG MOLALITTf
MG/L
Ca
Mg
Na
K
Fe
Mn
M
Ba
Si
Cl
bicarb.
S
N
8
F
AS
Zn
Pb
Hg





1.9170600-03
1.6316180-03
2.985860D-03
1.2793700-04
1. 2446680-03
1.438S94D-06
1.6039SOD-04
4.3910000-07
1.6S2966D-03
4.010174D-03
alk.
2. 8629550-03
5.001722D-05
1. 0180070-05
1.2635520-04
S. 1076940-07
3.3663680-07
2.0BOOOOD-08
1.12748UD-08
PR
DC
IONIC STRENGTH
TEMPERATURE
BLECTRICAL BALANCE
-2.7174
-2.7874
-2.S249
-3.8930
-2.9049
-S.8421
-3.7948
-6.3574
-2.7817
-2.3968

-2.S432
-4.3009
-4.9922
-3.8984
-6.2918
-S.4728
-7. 819
-7.9479
7.6893
2.474S
.0168
20.5000
1.10640-04
76.8
39.7
68.6
S.O
69.5
0.079
4.3
0.0603
99.3
142.2
132.6
275. 8
0.7
0.11
2.4
0.038
0.022
0.0043
0.0023
-
~



      Examination of this output file reveals some discrepancies with
the actual Cortez pit water, and even some unrealistic numbers.  Most
notable, the aluminum concentration is much higher than the actual
                        i
Cortez pit water, and even much higher than is possible at neutral pH.
The calcium, magnesium, iron, and sulfate concentrations are also higher
than actual.
      The reason for the discrepancies is that PHREEQE cannot remove
ions from solution via mineral precipitation unless the minerals are
specified in reversible reactions.  The BALANCE model found several
phases that would be expected to precipitate in the Cortez pit water

-------
                                                                     156
mass transfer model, but PHREEQE cannot  simulate the  reactions.
      Precipitation:  To account for precipitation reactions,  the next
step is to run the PHREEQE output through MINTEQA2, and model  the
changes that occur to the solution after minerals are removed  by
precipitation.  Table 7-15 is a portion  of the output file for the
precipitation model  (CZPR01MT.OUT) .
Table 7-15:  Results of Cortex pit water precipitation  simulation in
MXNTBQA2  (CZPR01KT.OUT).
NAME
               DISSOLVED
                                SORBED
                                               PRECIPITATED
MOL/KG PERCENT
N02-1
C03-2
Hg2+2
Pb+2
H3AS04
Na+l
K+l
Zn+2
Mn+2
F-l
S04-2
H3B03
Cl-1
H3AS03
Al+3
Ba+2
Fe+2
N03-1
NH4+1
Kg (OH) 2
Mn+3
Fe+3
H4S104
Mg+2
Ca+2
1.
2.
5.
2.
2.
2.
1.
3.
1.
1.
2.
1.
4.
5.
7.
2.
7.
2.
1.
1.
3.
1.
1.
6.
6.
9SOE-1B
232E-03.
760E-09
077E-08
151E-07
987E-03
280E-04
369E-07
439E-C6
265E-04
674E-03
019E-OS
015E-03
039E-16
304E-09
65SE-10
9S5E-16
OSSE-26
130B-05
6538-16
338B-29
1538-15
2528-06
665E-04
978E-04
Sun of CATIONS -
PERCENT DIFFERENCE
100.0
51.4
100.0
100.0
42.3
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
0.0
0.1
100.0
100.0
100.0
100.0
100.0
0.0
0.1
40.8
36.4
S.344E-03
. 36.91
NON- CARBONATE ALKALINITY
EQUILIBRIUM
EQUILIBRIUM
EQUILIBRIUM
IONIC STRENGTH (m)
pH
P


MOL/KG PERCENT
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
Sua of ANIONS
(ANIONS -
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
MOL/KG PERCENT
0
2
0
0
2
0
0
0
0
0
0
0
0
0
1
4
0
0
0
0
0
1
1
9
1
.OOOE-01
.113B-03
.OOOB-01
.OOOE-01
.928E-07
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.S95E-04
.3928-07
.0008-01
.OOOB-01
.0008-01
.0008-01
.0008-01
.2468-03
.6508-03
.6808-04
.220B-03
0
48
0
0
57
0
0
0
0
0
0
0
0
0
100
99
0
0
0
0
0
100
99
59
63
.0
.6
.0
.0
.7
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.9
.0
.0
.0
.0
.0
.0
.9
.2
.6
MG/L
<<
134
.0023
.0043
.0161
68.7
S.O
.022
.079
2.4
276
.11
142.3

1.97E-04
3.6SE-05
4.448-14

.2
3.88B-11
1.83E-24
6.448-11 _
.0752-
16.2
27.97
1.160B-02
CATIONS) / (ANIONS +
CATIONS)
9. 6218-07
.01222
 (.067
2.474


or


Eh


"


144.17


DV






      Table 7-15 depicts a more realistic  scenario, and  is  starting  to
approach  the measured Cortez pit water chemistry.  The dissolved
concentrations of several species have been reduced by the  precipitation
of various mineral phases.
      The dissolved aluminum concentration is back down  to  an acceptable
level for neutral pH waters  (-10'* mol/kg.) , due to the precipitation of
aluminum  phases.  Calcium and magnesium  concentrations are  approaching

-------
                                                                    157
the measured concentrations as well, having been slightly reduced
through precipitation of carbonate phases.
      However, the MINTEQA2 model has precipitated other minerals that
have reduced the levels of several important components well below their
measured concentrations in the pit water.  The ions most affected are
iron, silica, barium, and arsenic.  Iron has been almost entirely
removed from solution by the precipitation of hematite.  Barium and
arsenic have been reduced through precipitation of Ba(AsO4)2, and
dissolved silica has been reduced through precipitation of Ca-
nontronite.  Furthermore, sulfate is about three times higher than the
actual concentration, and chloride  is almost an order of magnitude
higher.
      The model clearly needs refining.  Before proceeding with the
adsorption model, the possible solubility controls for these components
need to be evaluated.  This marks the beginning of the calibration loop
for the modeling exercise.
       Calibration:   Referring back to Table 7-6,  the  Cortez  pit water
 speciation model  (CZSP02W4.OUT)  showed oversaturation with several
 mineral phases,  including carbonates,  silicates,  and  iron oxides (shown
 again in Table 7-16).   If the water analysis is run through MINTEQA2,
Table 7-16: Cortez pit water speciation, output file CZSP01W4.OUT
Phase
Aragonite
Calcite
Cristobalite
Dolomite (c)
FeOH)2.7C1.3
Goethite
Hematite
Magnetite
Talc
ZnSi03
Log tAP/KT
.449
.596
.401
1.084
6.487
7.830
17.321
19.185
3.473
1.286
Phase
Barite
Chalcedony
Dolomite (d)
Fe3 (OH) 8
Ferrihvdrite
Greenalite
Maghemite
Quartz
Tremolite

Log IAP/KT
.168
.357
.516
3.266
1.939
2.182
7.273
.800
6.279

 allowing precipitation to remove the components of oversaturated

-------
                                                                     158
minerals, the concentrations of some ions in the simulated final
solution drop several orders of magnitude below actual concentrations.
Due to kinetics and other factors  (incongruent dissolution/precipitation
behavior, ambiguous Kgp data), the  possibility of all of these
mineralsactually precipitating in  the pit water is remote.  This leads
to the conclusion that the most oversaturated phases are not likely
controlling the solubilities and concentrations of the ions.  For trace
metals,  the most likely control is adsorption.
      At this point in model interpretation, the modeler must once again
draw on  geochemical common sense.  An understanding of geological and
geochemical processes under  earth's  surface conditions helps eliminate
implausible results and isolate the  more probable scenario(s).   Some of
the phases shown in Table 7-16 can be eliminated from the model  because
they do  not exhibit reversible dissolution/precipitation behavior under
low temperature conditions.  Examples are  cristobalite, hematite,
magnetite, talc, quartz, and tremolite.
       To account  for  the solubility  and adsorption  controls  throughout
the  inverse  and forward model  exercise,  the model must be  calibrated by
working backwards  from the  final  condition.   By systematically excluding
 the  oversaturated phases  in MINTEQA2 precipitation  models,  the minerals
most  likely controlling the concentrations of dissolved  iron,  silica,
 arsenic, and barium in solution can be determined.   This  exercise will
 also determine the iron partitioning that occurs between the solid and
 aqueous phases for later adsorption models.  Once the precipitation
 calibration loops are complete,  then the partitioning of trace metals
 believed controlled by adsorption can be calibrated (As,  Hg, Pg, Zn).
       File CZPR10MT.OUT is the output file for the calibration run used
 to determine the iron partitioning.  The file shows the mass of iron
 precipitated and the resulting iron in solution.  The mass of amorphous
 ferric  hydroxide precipitated (as FeOOH), to result in a dissolved  iron

-------
                                                                           161
in Table 7-18.   The first adsorption iteration has  removed significant
percentages of As,  Hg, Pb, and Zn, as well as minor amounts of  Ba,  Ca,
and sulfate.  As expected, the final concentrations of the trace metals
have  been reduced below their actual concentrations in the pit  water,
necessitating a  calibration  loop similar to that used for precipitation.
Table  7-18:
IDX
Output file showing equilibrium distribution of Cortz pit
water after adsorption nodal  (CZADOUfT.OUT).
       NAME
       DISSOLVED
     MOW KG   PERCENT
   SORBED
MOL/KG   PERCENT
 PRECIPITATED
MOL/KG   PERCENT
732
SOO
950
180
270
280
60
360
410
00
470
150
2
61
330
281
1
471
361
460
770
100
140
S04-2
Na+1
Zn+2
Cl-1
F-l
Fe+2
H3A903
Hg2+2
K+l
Pb+2
Mn+2
Ca+2
H2O
H3As04
H+l
Fe+3
E-l
Mnt-3
Hg(OH)2
Mg+2
H4S104
Ba+2
CO3-2
2.862E-03
2.986E-03
3.544E-08
4.010E-03
1.264E-04
5.863E-07
6.648E-18
5.635E-09
1.279E-04
3.304E-11
1.439E-06
7.23SE-04
3.9S2E-06
3.362E-09
2.194E-03
9.277E-07
1.127E-08
3.947E-29
1.83SE-16
1.1738-03
8.-912E-OS
8.686B-08
2.189B-03
100.0
100.0
10. S
100.0
100.0
0.0
16.4
100.0
100.0
0.2
100.0
37.7
113.6
0.7
100.1
100.0
100.0
100.0
60.0
71.9
5.4
19.*
68.5
2.870E-07
O.OOOE-01
3.012E-07
O.OOOE-01
O.OOOE-01
O.OOOE-01
3.400B-17
O.OOOE-01
O.OOOE-01
2.077E-08
O.OOOE-01
1.738E-06
-4.725E-07
S.074E-07
-3.166E-06
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.22SS-16
O.OOOB-01
O.OOOB-01
2. 5428-10
O.OOOB-Ol
0.0
0.0
89.5
0.0
0.0
0.0
83. 6
0.0
0.0
99.8
0.0
0.1
13.6
99.3
-0.1
0.0
0.0
0.0
40.0
0.0
0.0
0.1
0.0
3.S20E-07
O.OOOE-01
O.OOOE-01
O.OOOB-01
O.OOOE-01
1.243E-03
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.192E-03
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
4.S93E-04
1. 5648-03
3.S20B-07
1.0081-03
0.0
0.0
0.0
0.0
0.0
100.0
0.0
0.0
0.0
0.0
0.0
62.2
0.0
0.0
o.o.
0.0
0.0
0.0
0.0
28.1
94.6
80.2
31. S
     Charge Balance:  SPBCIATBD     -                             ;:..'
      Sum Of CATIONS .  6.253K-03 Sum of ANIONS   1.138B-02
      PERCENT DIFFERENCE   2.909E+01  (ANIONS - CATIONS)/(AHIONS  CATIONS)
      EQUILIBRIUM IONIC STRENGTH (m)    1.294E-02
      BQOILIBRIOM pH              "   8.067
      EQUILIBRIUM p                 2.520   or Bh  -  146.67 rv
                     * DIFFOSB LAYER ADSORPTICM MOOD, 
                     ** Parameter* For Adsorbent Number 1 
         Blectroatatie Variables:  paiO . 0.003137
                               paib  0.000000
                               paid  o.oooooo
         Adsorbent Concentration (g/1)-.  0.074
         Specific Surface Area (aq. aeters/g):  COO.00
                                  sigO  0.000831
                                  aigb - 0.000000
                                  aigd - o.oooooo
       Calibration of trace metal adsorption is slightly less time
 consuming than precipitation calibration.  To achieve the correct trace
 metal partitioning, the bulk concentrations are simply increased until
 the final dissolved concentration is obtained.  The results of the
 adsorption  calibrations are shown in file CZAD02MT.OUT,  and are

-------
                                                                     162
summarized in Table 7-19.  Table 7-19 reveals that the final
concentrations of each trace metal ion is reduced by approximately  1-2
Table 7-19t Adsorption Calibrations (MINTBQA2)


AJ
Hg
Hg{OH)2
Hg2+2
Pb
Zn
Bulk
loo mol
-5.22
-7.77


-5.79
-6.47
Final
loo mol
-5.81
-8.07
-15.58
-8.07
-7.74
-7.55

mq/1
0.116
0.0017 '
S.27E-11
0.0009
0.0038
0.0018
orders of magnitude from the initial bulk concentration.  It must also
be noted in the results of CZAD02MT.OUT that mercury cannot be
calibrated by this procedure,  because the only Hg species for which the
MINTEQA2 diffuse layer adsorption model has complexation constants is
Hg(OH)z.   The most common mercury species  in the pit water,  according to
the MINTEQA2 speciation model (CZSP01MT.OUT) , is Hg,,,, by several orders
of magnitude.  Therefore, the most abundant mercury species will not    .
even be considered by the adsorption model.                     - ;
      All the information needed to run a complete forward model for pit
water chemical evolution is now available.

      Second iteration:  The first model iteration demonstrates that,
for modeling purposes, the existing Cortez pit water chemistry is not
necessarily the "final water.*  Thermodynamic constraints and
partitioning from solubility and adsorptive controls necessitate an
iterative calibration process to fine tune the mass transfer results.
To continue the calibration, results from CZAD02MT.OUT  (Table 7-19)  must
be entered again into BALANCE to determine the mass transfer in the
context of the newly determined solubility and adsorption controls.
      In the second iteration, BALANCE found 1820 possible combinations

-------
                                                                     159
concentration in solution of 0.134 mg/1,  is  5.201E-04 mol/kg Fe,  or
0.0735 g/1 FeOOH.  This mass of precipitated iron  was used as input to
the adsorption models to define the amount of sorbent available in
solution.
      Accordingly, file C2PR18MT.OUT  is the  output file  for the
calibration of silica partitioning.   The  barium partitioning was
determined simply by excluding the solid  Ba(AsO4}3, and allowing
solubility control by barite.  Consequently,  arsenic control was  also
removed by the exclusion of Ba(AsO4},.
      Table 7-17 shows the results of the precipitation  calibrations.
To end up with 0.134 mg/1 {10"s-*s mols/kg) dissolved Fe in solution,  a
mass of 10'J-2* mols/kg Fe must be dissolved from the pit  wall into
solution, combined with solubility control as specified  by the exclusion
option.  For a final concentration of 34.43  mg/1 SiO, (10 24 mol/kg.),  a
mass of 10"1'" mol/kg must dissolve.   For  a final Ba concentration of
0.0603 mg/1 (10"* M mol/kg), a mass transfer  of 10**" mol/kg is required.
      Calibration of the Cortex pit water model indicates that.the bulk
concentration of dissolved iron is approximately 2 orders of magnitude ...
higher  (10"1-2* mols/kg) than the final concentration of 10"*-" mols/kg.
Eleven iron minerals were excluded in the MINTEQA2 model to determine
this aspect of iron solubility control  (CZPR10MT.OOT).   Exclusion of six
silica phases revealed that the silica bulk  concentration is
approximately one order of magnitude  higher  than the final concentration
of io-J-M mols/kg.                     -
      Calibration of the only phase controlling the modeled zinc
concentration  (ZnSiO,)  resulted in a bulk concentration of over 200 mg/1
to result in 0.002 mg/1 in final pit  water,  over 5 orders of magnitude
difference.  This result suggests that zinc  concentration is less likely
controlled by mineral solubility than adsorption onto solids.
      As a time saving measure, most  of the  calibration  runs were

-------
                                                                            160


performed independently of  the other components.   In  some cases,  the




            Tabl* 7-17:  Precipitation Calibrations  (NINTSQX2)


                   Bulk           Final
                   log mol      log  mol        mg/1           Excluded

       Fe         -3.28         -5.62          0.135         Hematite
                                                            FeOH)2.7C1.3
                                                            Magnetite
                                                            Goethite
                                                            Lepidocrocite
                                                            Mag-ferrite
                                                            Maghemite
                                                            Ferrihydrite
                                                            Fe3(OH)8
                                                            Jarosite K
                                                            Siderite

       SiO,        -2.32         -3.23        34.78           Quartz
       H4Si04      -2.32         -3.23        55.63           Cristobalite
                                                            Chalcedony
                                                            Si02  (a,gl)
                                                            Si02  (a,pt) .
                                                            ZnSiO3

       Ba         -4.47         -6.36          0.0603         Ba(As04)2
final run gave slightly different results when included with all other dependent

components.  For example the iron precipitation calibration, which determined

that 5.201E-04 mol/kg Fe mass transfer resulted in 0.0735-g/l FeOOH precipitated

and 0.134 mg/1 dissolved Fe in the  final solution, gave slightly different

results when all precipitation and  adsorption models were included (see

CZAD01MT.OUT).  To  completely calibrate the model, an iterative loop would  be

necessary similar to those used for each individual component.

       Adsorption:  The next  step is to determine  the trace  metal partitioning

as a result of adsorption onto mineral solids.  The only solid considered as  an

adsorbent was amorphous ferric hydroxide, for which an abundance of adsorption

data and constants  are available.   The input parameters for amorphous ferric

hydroxide  in  the  diffuse layer adsorption model are (Dzombak and Morel,  1990):


       Concentration  of adsorbent in solid (g/1):    0.074 g/1
       Specific surface area (m3/g) :                 (00
       Site concentration (a/1 or m/g) :              I.953B-05 (site 1)
                                                3.S81S-03 (site 2)
       Surface potential (volts):                   Defined by KIMTEQA2 diffuse layer
                                                adsorption aodel



The first adsorption model,  before any calibration adjustments,  is  shown

-------
                                                                     163
of phases that could account for the water evolution, and 18 that
satisfied the constraints specified.  The model selected is marked by an
{*) in Appendix B,  and is shown in Table 7-20.
Table 7-20 Results of
Plagioclase
Fluorite
SOH:Zn
Galena
Cinnabar
Arsenopy
CO, gas
NaCl
Pyrite
Barite
K-Feldsp
Calcite
Gypsum
Kaolinite
Rhodochr
Dolomite
BALA;
+ F
+ F
F'
+ F
+ F
+ F
F
+ F
+ F
+ F
+ F
F
.

*
HCK model in second iteration.
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3609.4638
108.0000
512.0656
33.6800
171.3500
-1593.7647
-511.5883
-2378.9350
-1.4091
386.7400
      These results were again run through PHREEQB in the second
iteration, to produce the bulk water chemistry before precipitation and
adsorption controls were modeled.  These results are shown in Table 7-21
(file CZRX02PH.OUT).
      Mote once again some minor discrepancies involving ions predicted
to precipitate as mineral phases, such as calcium, magnesium, silica,
and sulfate.  These can be calibrated further in PHREEQB and MINTEQA2 to
more closely approximate the bulk concentrations arrived at in the prior
MINTEQA2 iteration.
      The results of the second PHREEQE model are run through M1NTEQA2,
resulting in the output file (CZAD02MT.OUT), a portion of which is shown
in Table 7-22.  A comparison of the model results and the actual Cortez
pit water chemistry is shown in Table 7-23.  Examination of these
numbers shows that the measured Cortez pit water chemistry still has not
been achieved.  Of special interest is arsenic, for which the model

-------
                                                                       164

predicts to exceed the actual concentration by a  factor of about three

 (and in violation  of federal primary drinking water standards).   A final

calibration iteration was performed to bring arsenic in line with the

actual concentration in the Cortez pit water  (CZAD03MT.OUT).  Reduction

of the bulk arsenic concentration (from 6.09E-06  molal)  to 3.3E-06 molal
Table 7-21i  Concentration of pit water after PHRBBQB
model; second iteration (CZRX02PH.OOT).
           ELEMENT

           Ca
           Mg
           Na
           K
           Fe
           Mn
           Al
           Ba
           Si
           Cl
           C
  MOLALITY

2.170879D-03
1.836022D-03
2.9858450-03
1.279370D-04
5.241170D-04
1.4385940-06
4.-5865500-03
3.368000D-05
9.0298910-03
2.5347740-03
1.9187980-03
LOG MOLALITY
                               mac* transfer
MG/L
           bicarbonate alkalinity
           S              1.4567130-03
           N              5.0017220-05
           B              1.0180070-05
           F.             1.2635520-04
           AB             6.089929O-06
           Zn             3.3663680-07
           Pb
1.634000D-06
1.6974810-08
                   PH -   7.6458             "
                   PB -   2.5180
                   IONIC STRENGTH     .0164
                   TEMPERATURE  -  20.5000
                   ELECTRICAL BALANCE -   1.1064D-04
-2.6634
-2.7361
-2.5249
-3.8930
-3.2806
-5.8421
-2.3385
-4.4726
-2.0443
-2.5961
-2.7170

-2.8366
-4.3009
-4.9922
-3.8984
-5.2154
-6.4728
-5.7867 .
-T.7702
87.0
45.3
68.6
5.0
29.27
.0079
123.8
4.6
542.6
89.9

117.1
139.8
3.1
.11
2.4
,4563
.022
>>- .3386
...I-.. .0034
 brought the dissolved arsenic concentration in the final pit water down

 to 0.0154 mg/1.   The effect on other ions in solution was minimal, and

 only noticeable for those metal cations which  compete with arsenic for

 adsorption sites.
                   i
       Tables 7-23 and 7-24 show that most other ions are within their

 actual concentrations by a factor of two or three.  This may be

 considered acceptable margin of error for major ions.  These

 Concentrations could be fine-tuned through continued calibration, a

 process that may continue for several iterations until the correct mass

-------
                                                                        165

transfer values, solubility controls, and adsorption controls are

identified.   The third iteration is as far as  this study was taken.   The

output  files for all of the forward simulations that are on datadisk are

listed  in Table 7-24.
Table  7-22:   Portion of output  file showing equilibrium distribution  of
Cortex pit water after adsorption model, second iteration
(CZAD02KT.OOT).
  NAME
  S04-2
  Zn+2
  Mn-t-2
  Cl-1
  F-l
  Fe+2
  H3As03
  Hg2+2
  K+l
  Pb+2
  Ca+2
  Na+1
  H3AsO4
  Fe+3
  Hg(OH)2
  Mn-t-3
  Ba+2
  Mg+2
  H4S104
  CO3-2
   DISSOLVED
MOL/KG   PERCENT
    SORBED
MOL/KG   PERCENT
  PRECIPITATED
MOL/KG   PERCEHT
1.423E-03
2.798E-08
1.439E-06
2.535E-03
1.264E-04
7.601E-08
3.377E-15
8.485E-09
1.279E-04
1.800E-08
6.238E-04
2.384E-03
1.S50E-06
1..364E-07
2.661E-16
3.842E-29
1.3S5E-07
1.063B-OS
1.547E-03
2.118B-03
97.7
8.3
100.0
100.0
100.0
0.0
23.1
100.0
100.0
1.1
28.7
79.8
25.5
100.0
68.3
100.0
0.4
0.6
17.1
._ 72.^,3.,
1.131E-08
3.086E-07
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.126E-14
O.OOOE-01
O.OOOE-01
1.616E-06
2.689E-06
O.OOOE-01
4.540E-06
O.OOOE-01
3.524E-17
O.OOOE-01
1.659E-10
O.OOOE-01
O.OOOB-01
O.OOOB-01
0.0
91.7
0.0
0.0
0.0
0.0
76.9
0.0
0.0
98.9
0.1
0.0
74.5
o.o
11.7
0.0
0.0
0.0
0.0
0.0
3.354E-05
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
S.239E-04
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.543E-03
6.018E-04
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
3.354E-05
1.82SE-03-
7.482E-03
8.133E-04
2.3
0.0
0.0
0.0
0.0
100.0
0.0
0.0
0.0
0.0
71.1
20.2
0.0
0.0
0.0
0.0
99.6
99.4
82.9
27.7
     Charge Balance: SPBCIATBD

       Sum of CATIONS -  3.622E-03 Sum of ANIONS   7.473E-03
       PERCENT DIFFERENCE >   3.470E+01   (ANIONS - CATIONS)/(ANIONS 4 CATIONS)
       EQUILIBRIUM IONIC STRENGTH (m)  -   7.442B-03
       EQUILIBRIUM pH                 *   8.067
       EQUILIBRIUM pe                 -   2.520   or Eh  -   146.67 v


               *******  DIFFUSE LAYER ADSORPTION MODEL **

                 *** Parameters For Adsorbent Number 1 ***
        Electrostatic Variables:  psiO  -.027403      sigO   -.005778
                                 psib  0.000000      sigb  - 0.000000
                                 paid  0.000000      sigd  - 0.000000
        Adsorbent Concentration  (g/1):   O.074
        Specific Surface Are* (sq. meters/g):   600.00
      This  exercise has hopefully demonstrated that the  calibration

procedure can help the model converge to the desired result,  but may

take some time.

-------
                                                                     166
Table 7-23:  Comparison of adsorption modal  (sacond itaration)  and
actual Cortex pit vatar chemistry.
            SO4-2
            Zn+2
            Total Mn
            Cl-1
            F-l
            Total Fe
            Total As
            Total Hg
            K+l
            Pb+2
            Ca+2
            Na+1
            Ba-i-2
            Mg+2
            S102
            bicarb, alk.

            pH
Model

136.7
 ,0.0018
  0.079
 89.9
  2.4
  0.012
  0.116
  0.0017
  5.0
  0.0037
 25.0
 54.8
  0.019
  0.26
 92.9
129.2

  8.07
 Actual

 90.2
  0.002
  0.0017
 24.4
  2.4
  0.134
  0.0383
  0.0004
 11.7
  0.0043
 45.4
 68.63
  0.0603
 18.1
 34.43
282.3

  8.07
Summary

      The overall pit water modeling exercise  as performed in this study

can be outlined as follows:                                     .


1)  Compile data for pit and groundwater geochemistry.

2)  First iteration; determine mass transfer with BALANCE.

3)  Apply BALANCE mass  transfer results in PHREEQB  to generate a bulk
    chemistry of pit water.

4)  Apply PHREBQB results to MINTEQA2  to determine  changes in solution
    chemistry from mineral precipitation.

5)  Evaluate  discrepancies.        '

6)  Calibrate the mass  transfer model  with MINTEQA2 to determine
    solubility control  of oversaturated phases, and the mass of
    amorphous ferric hydroxide precipitated for adsorption model.

7)  Calibrate mass transfer to determine adsorption partitioning and
    control.

6)  Second  iteration; apply results after  precipitation/adsorption
    models  in BALANCE to determine mass transfer.   New  "final water" is
    the  bulk  chemistry  of first  iteration.
 7)   Apply BALANCE results in PHREEQB.

-------
                                                                        167
8)   Apply PHREEQE results  in MINTEQA2, for final  determination of
     precipitation/adsorption model.
Table 7-24:  File names  and contents of forward model output files.
CZPR01PH.IN
C2PR01PH.OUT
CZPR01PH.IN
CZPR01PH.OUT
CZAD01MT.IN
CZAD01MT.OUT
CZAD02MT.IN
CZAD02MT.OUT
                     Cortez pit water precipitation simulations
PHREEQE input file
PHREEQE output file
PHREEQE input file
PHREEQE output file
CZPR18PH.IN
CZPR18PH.OUT
                      Cortez pit water adsorption simulations
HINTEQA2 input file
HINTEQA2 output file
PHREEQE input file
PHREEQE output file
CZAD03MT.IN
CZAD03UT.OUT
PHREEQE input file
PHREEQE output file
UINTEQA2 input file
UINTEQA2 output file
Sensitivity Analyses

       A set of sensitivity  analyses was performed on two subsets of  the

pit wall dissolution models.   The first simulation modeled the effects

of variable pyrite dissolution from the pit wall  in the presence of

different host lithologies.   The second simulation modeled the

progression of anoxia in the  pit water.

       Pyrite dissolution:   The mass of pyrite dissolved from the pit

wall,  and hence the mass of iron and sulfate dissolved into the pit

water,  was varied in three  different simulations  depicting different

host  lithologies.  The mass of pyrite dissolved into the pit water was

increased incrementally in  ten steps from 0.0 to  0.1 moles/kilogram,  as

shown in Table 7-25.  In each model, pCO2 and pO2  were held constant  at

the levels determined in the  earlier pit water speciation models.

       In the "carbonate" scenario,  calcite and dolomite were defined as

reversible reactions, a situation already proven  as likely in prior

discussions.  In the "granite" scenario, reversible reactions were

specified for albite, anorthite,  microcline, and  amorphous silica, four

phases  that might appear in a variety of igneous  rocks,  including

granite.   In the "shale" scenario,  only amorphous silica was specified

-------
                                                                     168
as reversible.  The results are shown in Table 7-25, and Figure  7-2. and
the output files are:

            PIT-C03.0UT       Carbonate lithology
            PIT-SIL.OUT       Granite lithology
            PIT-SH.OUT        Shale lithology

Figure 7-2 shows that the carbonate system remains buffered at neutral
pH throughout the modeled range of pyrite dissolution.  The granite
system shows a drop at 10  moles pyrite dissolved,  but the  rate of
decline decreases and the system  remains generally buffered at a pH
around 5.0.  The shale system, with no buffering capacity available,
shows a dramatic pH drop at 10'1   moles pyrite dissolved, and continues
to decline to a pH less than 2.0, a scenario seen in some serious acid
mine drainage environments.
Table 7-25:  Evolution of pH am a function of pyrite dissolved and host
rock  (Reversible reactions defined for each host:  Shale; quartz.
Carbonate; calcite, dolomite.  Granite; aicrocline, albite, anorthita,
quarts).
Pyrite diolved
(log BO!)
none
-5.0
-4.5
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
Sh*l
best
7.67
7.67
7.66
7.61
7.46
3.90
2.51
2.00
1.59
1.22
Carbonat*
hot .. _
7.67
7.75
7.75
7.75
7.73
7.70
7.63
7.53
7.41
7.25
Granite
,.bo.t ._ .___. ....
7.67
7.66
7.65
7.61
7.46
5.82
5.35
5.25
5.18
5.09
       The  "granite"  scenario is unrealistic because of the low
 probability of  the pit water reaching equilibrium with any
 aluminosilicate phases.  Speciation of the Liberty pit water shows
 equilibrium with  a few silica phases, but no aluminosilicates.  The
 result is  insufficient buffering  capacity, and development of acidic

-------
                                                                      169
waters with high concentrations  of  metals.
                                         State
                non* -5.0 -4.5 -4.0 -3.$ -3.0 -2.5 -2.0 -1.5 -1.0
                        llbtt pyritt dissolved (log molts)
Figure 7-2  Evolution of pB  as a  function of pyrite dissolved and host
rock  (Reversible reactions defined for each hosts   Carbonate - ealcite,
dolomitei  Granite  - aicrocline, albite,  anorthite, quartz*   Shale -
quarts).
      From Figure  7-2, one can  infer son* interesting implications
regarding possible scenarios  for  future pit waters in Nevada.   Carbonate
hosted deposits, and even those with a minor siliceous component such as
Cortex, may remain buffered throughout their evolution.   However, pit
waters in more  siliceous host lithologies may develop acidic conditions
similar to those depicted in  Figure 7-2.  Since the granite scenario
depicted here represents the  best possible case (i.e. equilibrium with
aluminosilicate minerals), a  more likely outcome might be the
development of  pH  somewhere between the "granite" and "shale" scenarios.
      As stated earlier, over one-fourth of Revada'a precious metal
deposits  (approximately 30) are hosted in igneous rocks.   Of the three
that are currently known to be  water-filled, two (Ruth and Liberty) are
acid mine drainage situations.  The potential for the remaining 27

-------
                                                                     170
igneous-hosted deposits,  as well as sediment-hosted deposits with no
significant carbonate component, to develop similar conditions might be
a subject where future research should be emphasized.
      Anoxia progression:  In the second sensitivity analysis, anoxia
was simulated by removing oxygen from the system through incremental
reduction of pO2  (partial pressure  of  oxygen)  in the water.  The results
are shown in Table 7-26,  and can be found on datadisk 2  (CZRX03PH.OUT).
      Table 7-26 shows that if the system goes anoxic to the point at
which pe * -5.53, approximately 2.553E-04 moles of solid Fe(OH)} (0.0273
g/1) will dissolve,  thus  increasing the dissolved Fe concentration to
3.708E-04 moles/kg.  (21 mg/1).   The value of 2.553E-04 moles of solid
Fe(OH)3 dissolved represents approximately one-third of the total
amorphous ferric hydroxide that was precipitated in the MINTEQA2 model.
If that much Fe(OH)} actually dissolves,  up to one-third of the species
that were adsorbed on the FetOH), surface (i.e.,  trace metals)  could be
released back into solution,  resulting in a 'blowout* of trace metals to
the environment.
Table
7-26:
Simulated anoxia in Cortex pit lake.
Datadiak 2 file
CZRX03PH.OOT.









po,
-45
-SO
-55
-CO
-cs
-70

P PH
1.17 .07
O.C2 .07
-0.673 .11
-2.153 .34
-3.10 .74
-5.53 .22
Dissolved
Pe (total)
.082t-0t
.2931-07
.698E-OC
.165B-OS
.1S5B-04
.7081-04
re (OB),
Haas tranaCer '
-1.I12H-07
2.98SB-07
4.369K-06
2.6951-05
.31*1-05
9. 5531-04
      This scenario is highly improbable however, because a decline in
pe to a level as low as -5.53 is unlikely without other redox processes
intervening.  As anoxia progresses, sulfate reduction should begin at
about pe = -3, causing precipitation of sulfides and generation of

-------
                                                                     171
alkalinity (Drever, 1988),  which would preclude dissolution of iron
hydroxides and release of trace metals.  Only in the absence of organic
matter (an unlikely scenario) would the pe drop unimpeded to levels as
low as -5.53.  A more plausible scenario, and the situation which
presents the greatest concern in pit water chemistry at neutral pH,
involves the situation in which the pe drops to the range 0 to -3.0.
This would cause the system to move from the stability field of ferric
iron to that of ferrous iron (Figure 5-7). with possible destabilization
of ferric hydroxides and a subsequent increase in total dissolved iron
concentration.

-------
                                                                  172

                        8.   CONCLUSIONS
Pit water chemical modeling is difficult and  influenced by many
interrelated variables.

A good understanding of system geology  is  required for each step in
the model, from speciation through to inverse and forward models.
This information should be obtained from a combination of chemical,
XRD, and petrographic analyses of drill core  and pit wall rock.

A good understanding of geochemical and geologic process is necessary
for successful interpretation of models and elimination of the
implausible.  There is no substitute for geochemical "common sense."

Precipitation and adsorption will be important process removing ions
from solution, depending on system Eh and  pH.

The partitioning from solubility and adsorptive controls must be
considered in mass transfer models.

Rate-independent dissolution models will generate unrealistic mass
transfer calculations.  The likely result  will be an underestimation
of dissolution of important minerals such  as  pyrite.

Rate-dependent dissolution models will  produce the most realistic
mass transfer results.  Site-specific experimental data on reaction
kinetics are required for model input.

Speciation modeling is a vital part of  the overall pit water modeling
process  (MINTEQA2, HATBQ4F), by helping to guide interpretation of
inverse and forward modeling results.  .-..--=  - ^---,. ..JLC- -,-.;--.__.-.-  -----

This study has answered the question posed at the beginning of .the . ..
thesis, by proving that a combination of inverse'and forward'modeling'
methods using BALANCE, PHREEQE  (expanded), and MINTEQA2 is able to
duplicate real world conditions within  reasonable error.

This model can be easily applied to future pit waters after the
incorporation of experimental kinetic data.

-------
                                                                     173
                         9.   RECOMMENDATIONS

      This study has shown that the aqueous geochemistry of an existing
pit lake can be reasonably duplicated through a combination of inverse
and forward modeling methods.  However, predicting the future water
chemistry for a pit that does not even exist  will not have the luxury of
an inverse model for the particular site.  The results of an inverse
model from an existing site, such as the Cortez pit, may not be
applicable to other locations because many critical variables will
differ greatly among mines.  Many factors might be similar, such as host
lithology, mineralogy, and local evaporation,  but not likely identical.
      The question remains, can the model developed in this study be
applied to future pit water modeling situations? I believe the answer
to this is yes, but proof of that lies in further study, such as
laboratory experiments involving site-specific pit wallrock.  If the
mass transfer predicted by the model can be duplicated in laboratory
experiments that determine the dissolution'of minerals'from the actual
site wallrock, then'the model should bV applicable to" other sites using
the results for similar experiments on the local wall rock.
      The forward modeling techniques used in this study perform
reasonably well for the Cortez pit water simulations, and the
incorporation of site-specific mass transfer  data, obtained from
laboratory experiments, should give equally valid results for any site.
      None of the computer codes applied in this study are specifically
designed for the task of modeling pit water chemistry.  MINTEQA2 comes
the closest with its large trace metal database and adsorption modeling
capability.  Unfortunately, MINTEQA2 cannot model a system reaction path
as PHREEQE can, so the two codes are best applied in tandem in the pit
water simulation.
      To produce the most accurate and valid  model possible, the

-------
                                                                     174

following procedure is recommended:


1)  Obtain a high quality water analysis for the upgradient groundwater,
    and run through speciation model to ascertain the speciation and
    saturation state of the water.

2)  Determine amounts and proportions of all minerals in the pit
    wallrock.  Location and geometry of ultimate pit should be available
    from mine plans.  The best source for this data will be chemical
    analyses and/or XRD work performed on drill core samples.

3)  Conduct laboratory experiments to determine mass transfer that
    occurs from interaction between the upgradient groundwater and the
    pit wall, under a variety of conditions (anoxic vs. oxic, closed vs.
    open).

4)  Determine bulk chemistry of pit water by applying the mass transfer
    results obtained in the laboratory study (PHREEQE).  A mass transfer
    of solid, from a known volume of water passed through a known mass
    of wallrock can be extrapolated to the entire pit lake system.

S)  Determine the final pit lake chemistry by applying the precipitation
     and adsorption model (MIKTEQA2) .
Recommendations for Further Study


      There are many components of the overall pit water modeling

problem that are only partially understood, and for which existing

models or data may be suspect.  Until more is known about these aspects

of the problem, pit water modeling will be at best a collection of

educated guesses incorporating many assumptions.  Each of these aspects

is broad enough to be evaluated in a separate study, and better

information regarding each would greatly improve future pit water

models:

*  Behavior of iron hydroxide  (Eh/pH stability and control) and ita
   ability to scavenge metals  in pit lakes.

*  Behavior of trace metals such as arsenic, cadmium, mercury, and zinc
   in pit lake environments.   Are there any equilibrium solubility
   controls or is control entirely by adsorption?

*  Thermodynamic data for aluminosilicates and potential solubility
   control in pit lakes.

*  Reaction kinetics of sulfides and host rock minerals.

 *   Effect of armoring on acid generating vs. acid neutralizing
    minerals.

-------
                                                                     175



*  Limnology; will pit lakes turnover or not?



*  Role of organic matter and associated rates of microbial oxidation.


-------
                                                                     176

                           10.   REFERENCES
ALLISON J.D., BROWN D.S., and NOVO-GRADAC K.J.  (1991)
   MINTEQA2/PRODEFA2, A Geochemical Assessment Model for
   Environmental Systems:  Version 3.0 User's Manual.  U.S.
   Environmental Protection Agency, Environmental Research
   Laboratory, Athens, GA.

BACK, W.  (1966)  Hydrochemical Facies and Ground-Water Flow Patterns  in
   Northern Part of Atlantic Coastal Plain.  U.S. Geological Survey
   Professional Paper 498-A.

BACK, H.  (1961)  Techniques for mapping of hydrochemical facies.  U.S.
   Geological Survey Professional Paper 424-D:  380-382.

BALISTRIERI L.S. and MURRAY J.W.  (1982)  The adsorption of Cu, Pb, Zn,
   and Cd on goethite from major ion seawater.  Geochimica ec
   Cosmochimica Acta 46, 1253-1265.

BALISTRIERI L.S. and MURRAY J.W.  (1983)  Metal-solid interactions in  the
   marine environment:  estimating apparent equilibrium binding
   constants.  Geochimica et Cosmochimica Acta 47, 1091-1098.

BALL J.W., JENNE E.A., and NORDSTROM O.K.  (1979)  WATEQ2 - A
   COMPUTERIZED CHEMICAL MODEL FOR TRACE AND MAJOR ELEMENT SPECIATION
   AND MINERAL EQUILIBRIA OF NATURAL WATERS.  In Chemical Modeling in
   aqueous systems.  Spedation, Sorption, Solubility, and Kinetics  (ed.
   E.A. Jenne).  ACS Symposium Series 93, American Chemical Society,
   Washington D.C., p. 815-835.

BALL J.W., NORDSTROM O.K., JENNE E.A.  (1980)  ADDITIONAL AND REVISED
   THERMOCHEMICAL DATA AND COMPUTER CODE FOR HATBQ2 -- A COMPUTERIZED.
   CHEMICAL MODEL FOR TRACE AND MAJOR ELEMENT SPBCIATION AND MINERAL
   EQUILIBRIA OF NATURAL WATERS.  U.S. Geological Survey Water Resources
   Investigations WRI 78-116.

BALL J.W., JENNE E.A., and CANTRELL M.W.  (1981)  WATEQ3 - A
   GEOCHEMICAL MODEL WITH URANIUM ADDED:  U.S. Geological Survey Open-
   File Report 81-1183, 81 p.

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.
   U.S. Geological Survey Open-File Report 91-183.

BAUM D.,  and KNOX M.L.  (1992)  In Butte, Montana, A is for Arsenic, Z is
   for  Zinc.   Smithsonian 23, no. 8, 46-56.

BELL P.E., MILLS A.L. and HERMAN J.S.  (1987)  Biogeochemical conditions
   favoring magnetite formation during anaerobic iron reduction.
   Applied Environmental Microbiology 53, 2610-2616.

BENCALA K.E.,  MCKNIGHT D.M., and  ZELLWEGER G.W.  (1987)  Evaluation of
   natural tracers in an acidic and metal-rich stream.  Water Resources
   Research  23, 827-836.

-------
                                                                     177

BENJAMIN M.M. and LECKIE J.O.  (1980)  Adsorption of  metals  at  oxide
   interfaces:  Effects of the concentrations of adsorbate  and competing
   metals.  In Contaminants and Sediments, 2, R.A. Baker  ed.,  305-322,
   Ann Arbor Sci. Pub. Ann Arbor, Michigan.

BENJAMIN M.M. and LECKIE J.O.  (1981)   Multiple-site adsorption of Cd, Cu,
   Zn, and Pb on amorphous iron oxyhydroxide.  J. Colloid Interface
   Science 83, 410-419

BERNER R.A.  (1971)  Principles of Chemical Sedimentology, McGraw-Hill,
   240 pp.

BITTON G. and GERBA C.P. (1984)  Groundwater Pollution Microbiology.
   Wiley Interscience  377 pp.

BLM (1991)  U.S. Bureau of Land Management.  Final Environmental  Impact
   Statement, Betze Project.

BLOOMSTEIN E.I., MASSINGILL G.L., PARRATT R.L., and  PELTONEN D.R.  (1991)
   Discovery, Geology, and Mineralization of the Rabbit Creek  Gold
   Deposit, Humboldt County, Nevada.  In Geology and Ore  Deposits of the
   Great Basin  (Raines G.L., Lisle R.E., Schafer R.W., and  Wilkinson
   W.H., eds.), Symposium Proceedings, Geological Society of Nevada,
   April 1-5, 1991, 821-844.

BOCKRIS J.O.M., and REDDY A.K.N., (1970)  Modern Electrochemistry 1,
   Plenum Press, New York.

BONHAM H.F.,  Jr.  (1991)  Bulk-Mineable Precious Metal Deposits  and
   Prospects in Nevada.  Map 91;  Second Edition.  Nevada Bureau  of
   Mines and Geology, Mackay School of Mines, University  of Nevada,
   Reno.
BROOKINS D.G. (19881  Sh-pH Diagrams for Geochemistry.^.. Springer-Verlag.
   176 PP.          --.	;. -      	        ----- --'*  --

BUSENBERG E. and CLEMENCY C.V. (1976)   The dissolution kinetics of
   feldspars at 25C and 1 atm CO, partial pressure.  Geochimica et
   Cosmochimica Acta 40, 41-49.

CAMP DRESSER and McKEE  (1988)  Preliminary water balance for the
   Berkeley Pit and related underground mine workings.  EPA Contract
   68-01-6939.

CARDCCIO F.T., GEIDELL G. and SEHELL J.M. (1976)  The character of
   drainage as a function of the occurrence of framboidal pyrite and
   ground water quality in eastern Kentucky.  In 6th Sytnp. Coal Mine
   Drainage Res., Louisville, KY.   1-16.

CHAO T.T. and THEOBALD P.K. Jr. (1976)   The significance of secondary
   iron and manganese oxides in geochemical exploration.  Economic
   Geology 71, 1560-1569.

CHAPMAN B.M., JONES D.R. and JONG R.F.  (1983)  Processes controlling
   metal ion attenuation in acid mine drainage streams.  Geochimica et
   Cosmochimica Acta 47, 1957-1973.

DAMES & MOORE, Salt Lake City, UT.  Phase I IB Report, Hydrogeology
   Studies, Ruth Mine, Nevada, For Kennecott Nevada Mines Division,
   Feb. 9, 1990.

-------
                                                                     178

DAVIES C.W.  (1962)  Ion Association.  Butterworth, Washington D.C.
   190 pp.

DAVIS, A.  (1992, 93)  PTI Environmental Services.  Presentation at 1993
   Annual Meeting of the Society for Mining, Metallurgy, and
   Exploration; Reno, Nevada.

DAVIS A. and RUNNELLS D.D.  (1987)  Geochemical interactions between
   acidic tailings fluid and bedrock:  use of the computer model MINTEQ.
   Applied Geochemistry 2, 231-241.

DAVIS A. & ASHENBERG D. (1989)  The aqueous geochemistry of the Berkeley
   Pit, Butte, Montana," U.S.A.  Applied Geochemistry 4, 23-36.

DAVIS A., OLSEN R.L., and WALKER D.R. (1991)  Distribution of metals
   between water and entrained sediment in streams impacted by acid mine
   discharge, Clear Creek, Colorado, U.S.A.  Applied Geochemistry 6,
   333-348.

DAVIS G.B. and RITCHIE A.I.M.  (1986)  A model of oxidation in pyritic
   mine wastes:  part 1 equations and approximate solution.  Applied
   Mathematical Modelling, 10, 314-322.

DAVIS J.A. and LECKIE J.O.  (1980)  Surface lonization and Complexation
   at the Oxide/Water Interface.  3. Adsorption of Anions.  Journal of
   Colloid and Interface Science 74, No. 1, 32-43.

DAVIS J.A., FULLER C.C.,  and COOK A.D.  (1987)  A model for trace metal
   sorption processes at the calcite surface:  Adsorption of Cd2+ and
   subsequent solid solution formation.   Geochimica et Cosmochimica 51,
   1477-1490.

DAVISON W. and HOUSE W.A.  (1988)   Neutralizing strategies for acid
   waters:  Sodium and calcium products generate different acid
   neutralizing capacities.   Water Research 22, No. 5, 577-583.

DENVER KNIGHT PIESOLD Environmental Consultants,  Inc.  (1991)  Waste
   Characterization Plan,  Permit No. NEV 90002, Bullion Monarch Project,
   Eureka County, Nevada,  May 1991.

DOYLE W.S. (1976)  Recovery of Acid Strip Mine Lakes, In Strip Mining of
   Coal:  Environmental Solutions, Pollution Technology Review No. 27,
   Noyes Data Corp. New Jersey.

DREVER J.I.  (1988)  The Geochemistry of Natural Waters, 2nd edition.
   Prentice Hall.  437 pp.

DUAIME, T. (1992).  Hydrogeologist, Montana Bureau of Mines and Geology.
   Personal communication.

DZOMBAK, D.A. and MOREL M.M.F. (1990)  Surface Complex Modeling.
   Hydrous Ferric Oxide.   Wiley-Interscience, John Wiley & Sons.  393 pp.

ENGESGAARD P. and KIPP K.L.  (1992)  A Geochemical Transport Model for
   Redox-Controlled Movement of Mineral Fronts in Groundwater Flow
   Systems:  A Case of Nitrate Removal by Oxidation of Pyrite.  Water
   Resources Research 28,  no. 10, 2829-2843.

-------
                                                                     179

ENSR CONSULTING AND ENGINEERING, and DREVER J.I.  (1991)  Water Resources
   Technical Report for the Betze Project Environmental  Impact
   Statement.  Prepared for:  USDI Bureau of Land Management, Elko
   District Office, Elko, Nevada.

FELMY A.R., GIRVIN D.C., and JENNE E.A.  (1984)  MINTEQ:  A Computer
   Program for Calculating Aqueous Geochemical Equilibria.   (NTIS PB84-
   157148)  EPA-600/3-84-031, National Technical Information Service,
   Springfield, Virginia.

FETTER C.W.  (1988)  Applied Hydrogeology.  Second Edition.   Macmillan
   Publishing, New York.  592 pp.  .

FILIPEK L.H., NORDSTROM O.K., and FICKLIN W.H. (1987)  Interaction of
   Acid Mine Drainage with Waters and Sediments of West  Squaw Creek  in
   the West Shasta Mining District, California.  Environment Science and
   Technology 21, 388-396.

FITZGERALD W.F.  (1979)  Distribution of mercury in natural waters,   in
   The biogeochemistry of mercury in the environment  (Nriagu J.O., ed.).
   Elsevier.  696 pp.

FLEMING G.W., and PLUMMER G.W.  (1983)  PHRQINPT-AN INTERACTIVE COMPUTER
   PROGRAM FOR CONSTRUCTING INPUT DATA SETS TO THE GEOCHEMICAL
   SIMULATION PROGRAM PHREEQE.  U.S. Geological Survey Water-Resources
   Investigations Report 83-4236.

FLORENCE T.M. and BATLEY G.E.  (1980)  Chemical speciation in natural
   waters.  In Critical Reviews in Analytical Chemistry  (eds B. Campbell
   and L. Meites), v 219-296.  CRC Press.

FULLER C.C. and DAVIS J.A.  (1989)  Influence of coupling of  sorption and
   photosynthetic processes on trace element cycles in natural waters.
   Nature 340, 52-54."             '           --. .-- .v^r-.-.---'--.*

GARRELS R.M., and MACKENZIE F.T. (1967).  Origin of the  Chemical
   Compositions of Some Springs and Lakes.  Equilibrium  Concepts in
   Natural Water Systems.  In American Chemical Society  Advances in
   Chemistry Series 67, 222-242.

GARRELS R.M., and THOMPSON M.B.  (1960)  Oxidation of pyrite in ferric
   sulfate solution.  American Journal of Science 258, 57-67.

GARRELS R.M., and THOMPSON M.B.  (1962)  A Chemical Model for Sea Water
   at 25"C and One Atmosphere Pressure.  American Journal of* Science
   260, 57-66.

GERAGHTY t MILLER, INC., Environmental Services, Reno, NV (1991)
   Rydrogeologic  Investigation for Mill No. 1 Tailings Facility and
   Downgradient Drainage Area, Carlin, Nevada, NV02202.  Prepared for
   Newmont Gold Company.

GLYNN P.D., PARKHURST D.L., and PLUMMER L.N.  (1992)  Principles and
   Applications of Modeling Chemical Reactions in Ground Hater.  Short
   Course, International Groundwater Modeling Center; Colorado School of
   Mines, Golden, CO.

GRAY D.M., ed.  (1970)  Handbook on  the Principles of Hydrology.  Hater
   Information Center, Inc.  Canadian National Committee for the
   International  Hydrological Decade.  676 pp.

-------
                                                                     180

GUILBERT J.M., and PARK C.F., Jr.  (1986)   The Geology of Ore Deposits.
   W.H. Freeman and Company, New York.   985 pp.

HAMER W.J.  (1968)  Theoretical mean activity coefficients of strong
   electrolytes in aqueous solutions  from 0 to 100C:   U.S.  National
   Bureau of Standards, National Standard Reference Data Ser.   24,
   271pp.

HARRIS A.M. (1992)  Comstock Mining Services.   UNR Hydrology/
   Hydrogeology Graduate Seminar, Spring 1992.

HARVIE C.E., and WEARE J.H.  (1980)  The  prediction of  mineral
   solubilities in natural waters:   the Na-K-Mg-Ca-Cl-SO4-H,O system
   from zero to high concentration at 2SC.   Geochiotica et Cosmochimica
   Acta 44, 981-997.

HCI  (Hydrologic Consultants, Inc.)  (1992)   Hydrogeologic Investigation
   for Gold Quarry Mine.  HCI-768.  Prepared for Newmont Gold Company.

HCI  (Hydrologic Consultants, Inc.)  (1992a)   MINEDW - a finite-element
   program for three-dimensional simulation of mine dewatering:
   unpublished report, 79 pp.

HELGESON H.C.  (1968)  Evaluation of irreversible reactions in
   geochemical processes involving minerals and aqueous  solution -  I.
   Thermodynamic relations.  Geochimica  et Cosmochimica  Acta 32,
   853-877.

HELGESON H.C., CARRELS R.M. and MACKENZIE F.T.  (1969)  Evaluation of
   irreversible reactions in geochemical processes involving minerals
   and aqueous solution - II.  Applications.   Geochimica et  Cosmochimica
  \ Acta 33, 455-481.

HELGESON H.C.  (1969)  Thermodynamics  of  Hydrothermal Systems at  Elevated
   Temperatures and Pressures.  American J.  of Science 267,  729-804.

HEM  J.D.  (1985)   Study and Interpretation of the Chemical
   Characteristics of Natural Water.  U.S.  Geological  Survey Water-
   Supply Paper 2254, 263 pp.

HERCZEG A.L.,  and IMBODEN D.M.  (1988)  Tritium hydrologic  studies in
   four closed-basin lakes in the Great  Basin, U.S.A.  Limnol.
   Oceanography, 33(2), 157-173.

HERLIHY A.T.,  MILLS A.L. and HERMAN J.S.  (1988)  Distribution of reduced
   inorganic sulfur compounds in lake sediments receiving  acid mine
   drainage.  Applied Geochemistry 3, 333-344.

HUANG H. and TAHIJA K.  (1990)  Characteristics and treatment problems of
   surface and underground waters in  abandoned mines at  Butte, Montana.
   In Mining and Mineral Processing Wastes (Doyle,  F.M., ed.). Proceed-
   ings of  the Western Regional Symposium on Mining and Mineral  Process-
   ing Wastes, Berkeley, California,  May 30 - June 1,  1990,  261-270.

HURLBUT C.S. Jr., and KLEIN C.  (1977)  Manual of Mineralogy (after James
   D. Dana) .  19th Edition.  John Wiley  6 Sons, New York,  532 pp.

HUTCHINSON G.E.  (1957)  A Treatise on Limnology, v.l,  John Wiley fc Sons,
   New York, 1015 pp.

-------
                                                                     181

HYDRO-SEARCH,  INC.  (1991)   Predicted Post Mining Hydrogeochemistry of
   the Lone Tree Pit,  Valmy,  Nevada.  Hydro-Search, Inc., Reno, NV.
   Prepared for Santa Fe Pacific Gold.

INTERA Environmental Consultants Inc.   (1983)  EQ3/EQ6:  Geochemical
   Models Suitable  Performance Assessment of Nuclear Waste Storage:
   Comparison of PHREEQE and EQ3/EQ6.  ONWI-473.  Prepared for Battelle
   Memorial Institute,  Office of Nuclear Waste Isolation, Columbus,
   Ohio.

JAMES R.O. and HEALY T.W.  (1972)  Adsorption of hydrolyzable metal  ions
   at the oxide-water interface.  Journal of Colloid and Interface
   Science 40,  (1),  42-81.

KARLSSON S.,  ALLARD B.  and HAKANSSON K. (1988)  Characterization of
   suspended solids in a stream receiving acid mine effluents, Bersbo,
   Sweden.  Applied Geochemistry 3, 345-356.

KHARAKAY.K.,  ROBINSON S.W.,  LAW L.M., and CAROTHERS,  W.W. (1984)
   Hydrogeochemistry of Big Soda Lake, Nevada:  An alkaline meromictic
   desert lake.   Geochimica et Cosmochimica Acta 48, 823-835.

KHARAKA T.K.,  GUNTER W.D.,  AGGARWAL P.O.,  PERKINS E.H., and DeBRAAL J.D.
   (1988)  SOLMINEQ.88: A Computer Program for Geochemical Modeling of
   Water-Rock Interactions.   U.S. Geological Survey, Water-Resources
   Investigations Report 88-4227.  208 p.

KINNIBURG D.G.  and  JACKSON M.L.  (1981)  Cation adsorption by hydrous
   metal oxides and clay.  .In Adsorption of Inorganics at Solid-Liquid
   Interfaces (Anderson M.A.  and Rubin A.J., eds.),  91-160.  Ann Arbor
   Science Publishers,  Inc.

KNIGHT PIESOLO (1992) ..Site Characterization and Closure. Plan. Permit  -
  . No. NSV 90002, Bullion  Monarch Project, Eureka County,  Nevada,
   February 14,  1992.

LACY D.T. and LAWSON P. (1970)   Kinetics of the liquid-phase oxidation
   of acid ferrous  sulfate by the bacterium Thiobacillus ferrooxidans.
   Biotec. Bioeng.  12,  29-50.

LINDBERG R.D.  and RUNNELLS D.D.  (1984)  Ground-water redox reactions:
   An analysis of equilibrium state applied to Eh measurements and
   geochemical modeling.   Science 225, 925-927.

LUNDGRBN D.G.  and SILVER M.  (1980)   Ore leaching by bacteria,  Annual
   Reviev of Microbiology,  34,  263-283.

LYNCH D.C. (1988)   A review of the physical chemistry of arsenic as it
   pertains to primary metals production,  In Arsenic Metallurgy
   Fundamentals and Applications, R.G. Reddy, J.L.  Rendrix,  and P.B.
   Queneau, eds., AIME Proceedings, 3-33.

LYONS W.B., DOYLE G.A., PETERS EN R.C., and SWANSON E.E. (in press)   The
   limnology of future pit lakes in Nevada:   The importance of shape.

MACDONALD M.  SAUNDERS  (1992)   Hater Quality in Open Pit Precious Metal
   Mines.  Master's Thesis  (Professional Paper),  University of Nevada,
   Reno.

-------
                                                                     182

MACINNES D.A.  (1919)  The activities of the  ions  of strong electrolytes:
   American Chemical Society Journal 41,  1086-1092.

MANTOURA R.F.C., DICKSON A. and RILEY J.P.   (1978)   The complexation of
   metals with humic materials in natural waters.   Bstuar.  Coast. Marine
   Science , 387-408.

MASSCHELEYN P.H., DELAUNE R.D., and PATRICK  N.H..  Jr.   (1991)  Effect of
   redox potential and pH on arsenic speciation and solubility in a  con-
   taminated soil.  Environmental Science and Technology 25, 1414-1419.

MAZOR E. (1991)  Applied Chemical and Isotopic Groundwater Hydrology.
   Halsted Press, New York.

McBRIDE M.B.  (1980)  Chemisorption of Cd2* on calcite surfaces.  Soil
   Sci. Soc. Amer. J. 44, 26-28.

MCDONALD M.G. and HARBAUGH A.W. (1984)  A MODULAR THREE-DIMENSIONAL
   FINITE-DIFFERENCE GROUND-WATER FLOW MODEL (MODFLOW)   U.S. Geological
   Survey Open-File Report 83-875, 528 pp.

MILLERO F.J.  (1975)  The physical chemistry  of  estuaries,  in Church,
   T.M., ed., Marine Chemistry in the Coastal  Environment:  American
   Chemical Society Symposium Series 18, 25-55.

MILLERO F.J.  (1977)  Thermodynamics of the carbonate system in seawater:
   Geochimica et Cosmochimica Acta 43, 1651-1661.

MILLERO F.J., SCHREIBER D.R.  (1982)  Use of  the ion pairing model to
   estimate activity coefficients of the  ionic  components of natural
   waters.  American Journal of Science 282,  1508-1540.

MOREL M.M., BERING J.G.  (1993)  Principles and Applications of Aquatic
   Chemistry.  John Wiley fc Sons,  Inc. Publishers*.^.,  -    _ ..-i;vi.. -  V~i
MURRAY J.W., and BREWER P.O.  (1977)  The mechanisms  of removal of iron,
   manganese, and other trace metals from seawater.   In Marine Manganese
   Deposits  (G.P. Glasby, ed.)  Elsevier, Amsterdam,  391-426.

NDEP  (1992,  93)  NEVADA DIVISION OF ENVIRONMENTAL PROTECTION, Bureau of
   Mining Regulation and Reclamation.  Files.

NEVADA DIVISION OP WATER PLANNING  (1992)  Nevada  Water Facts.  State of
   Nevada, Department of Conservation and Natural Resources.

NGDC  (National Geophysical Data Center)   (1983)   Geothermal Resources of
   Nevada.   National Oceanic and Atmospheric Administration.  1:500,000
   scale Map.                          

NIENHUIS, P., APPELO T., and WILLEMSEN, G.  (1991)  Adaptation of Phreeqe
   for use in a Mixing - Cell Flowtube:  PHREEQM.  User Guide.

NORDSTROM O.K.  (1977)  Hydrogeochemical and microbiological factors
   affecting the heavy metal chemistry of an acid mine drainage system,
   Ph.D. Dissertation, Stanford University, 210 pp.


-------
                                                                     183

NORDSTROM O.K., PLUMMER L.N. WIGLEY T.M.L.,  WOLERY T.J.,  BALL J.W. ,
   JENNE E.A., BASSETT R.L. , CRERAR D.A. ,  FLORENCE T.M.,  FRITZ B. ,
   HOFFMAN M., HOLDREN G.R. Jr.,  LAFON G.M.,  MATTIGOD S.V.,  McDUFF R.E.,
   MOREL F., REDDY M.M. , SPOSITO  G.,  THRAILKILL J.  (1979a)   A Comparison
   of Computerized Chemical Models  for Equilibrium Calculations in
   Aqueous Systems. In Chemical Modeling in  Aqueous Systems  (E.A.  Jenne,
   Ed.)  ACS Symposium Series 93, American Chemical Society,  Washington
   D.C., 857-892.

NORDSTROM O.K., JENNE E.A., and BALL J.W.  (1979b)   Redox  Equilibria  of
   Iron in Acid Mine Waters.  In  Chemical  Modeling in Aqueous Systems
   (E.A. Jenne, Ed.) ACS Symposium  Series  93,  American Chemical Society,
   Washington D.C., 857-892.

NORDSTROM O.K. (1982)   The effect of  sulfate on aluminum  concentrations
   in natural waters:  some stability relations in the system A120,-SO,-
   HjO at  298K.   Geochimica et Cosmochimica Acta 46, 681-692.

NORDSTROM D.K., and BALL J.W. (1983)   Chemical models,  computer programs
   and metal complexation in natural  waters.   Unpublished manuscript.

NORDSTROM D.K. (1985)   The Formation  of Acid Mine Waters:  A Review.
   Proceedings of Hazardous Materials Management Conference,  HAZMAT
   WEST '85, p. 453-457.

NORDSTROM O.K., and BALL J.W. (1985)   Toxic  Element Composition of Acid
   Mine Waters From Sulfide Ore Deposits.  Mine Water,  Granada,  Spain,
   International Mine Water Association, 1985.

NORDSTROM D.K. and ALPERS C.N. (1990)   Geochemical  Evaluation of Acid
   Mine Waters at Iron Mountain. Shasta, California,  1990.   U.S.
   Geological Survey Administrative Report.   36 pp.
NORDSTROM O.K., PLUMMER L.N. LANGMUIRD,, BUSENBERQ, MAY-H.Mr>
   B.F. and PARKHURST D.L.  (1990)  Revised chemical equilibrium data for
   major water-mineral reactions and their limitations. In Chemical
   modeling of aqueous systems II  (Melchior D.C. and Bassett R.L. eds.) ,
   ACS Symposium Series 416, American Chemical Society, 398-413.

PANKOW J.F. (1991)  Aquatic Chemistry Concepts.  Lewis Publishers.
   683 pp.

PARKHURST D.L., THORSTENSON D.C., and PLUMMER L.N.  (1980)  PHREEQE - A
   COMPUTER PROGRAM FOR GBOCHEMICAL CALCULATIONS.  U.S.G.S. Water-
   Resources Investigations 80-96.  Revised and reprinted 1990.

PARKHURST D.L., PLUMMER L.N., and THORSTENSON D.C., (1982)  BALANCE - A
   COMPUTER PROGRAM FOR CALCULATING MASS TRANSFER FOR GBOCHEMICAL
   REACTIONS IN GROUND WATER.  U.S. Geological Survey Water-Resources
   Investigations 82-14.

PARSONS J.D. (1977)  Effects of acid mine wastes on aquatic ecosystems.
   Water, Air and Soil Pollution 7, no. 3, 333-354.

PERCIVAL T.J., BAGBY W.C., and RADTKE A.S. (1988)  Physical and Chemical
   Features of Precious Metal Deposits hosted by Sedimentary Rocks in
   the Western United States.  In Bulk Mineable Precious Metal  Deposits
   of the Western United States  (Schafer R.W., Cooper J.J., and Vikre
   P.G., eds.). Symposium Proceedings, Geological Society of Nevada,
   April 6-8, 1987, 11-34.

-------
                                                                     184

PETERSON S.R., HOSTETLER C. J. , DEUTSCH W. J. , and COWAN C.E.  (1987)
   MINTEQ User's Manual.  Pacific Northwest Laboratory.

PIERCE M.L. and MOORE C.B. (1982)  Adsorption of arsenite  and arsenate
   on amorphous iron hydroxide.  Water Research 16,  1247-1253.

PITZER K.S. (1973)  Thermodynamics of electrolytes  I:   Theoretical basis
   and general equations.  Journal of Physical Chemistry 77, 268-277

PITZER K.S. (1979)  Theory:   Ion interaction approach.   Activity
   Coefficients in Electrolyte Solutions, Volume i,  R.M. Pytkowicz, ed.,
   CRC Press,  Boca Raton, Fla., 157-208.

PITZER K.S. (1980)  Electrolytes.  From dilute solutions to fused salts.
   Journal of the American Chemical Society 102, 2902-2906.

PLANKEY B.P. PATTERSON H.H. and CRONAN C.S. (1986)   Kinetics of aluminum
   fluoride complexation in acid waters.  Environmental  Science and
   Technology 20, 160-165.

PLUMMER L.N. (1984).  Geochemical Modeling:  A Comparison  of Forward and
   Inverse Methods.  From The First Canadian/American  Conference on
   Hydrogeology, Practical Applications of Ground Water Geochemistry.

PLUMMER L.N.,  JONES B.F., and TRUESDELL A.H. (1976)  WATEQF - A FORTRAN
   IV VERSION OF WATEQ, A COMPUTER PROGRAM FOR CALCULATING CHEMICAL
   EQUILIBRIUM OF NATURAL WATERS.  U.S. Geological  Survey  Water-
   Resources Investigations 76-13.   70 p.

PLUMMER L.N.,  WIGLEY T.M.L. and PARKHURST D.L.  (1978).   The kinetics of
   calcite dissolution in CO, -water systems at  5 to  60C and 0.0 to 1.0
   atm CO,.  Am.  J.  Sci.,  278,  179-216.
PLUMMER L.N., PARKHURST D.L., and THORSTENSON  .C.  t!983K  Development -
   of reaction models for ground-water systems.--  Oeochiaica et;.
   Cosaiochifflica Acta 47, 665-686.

PLUMMER L.N., JONES B.F., and TRUESDELL A.H. (1984)  WATEQF - A FORTRAN
   IV VERSION OF WATEQ, A COMPUTER PROGRAM FOR CALCULATING CHEMICAL
   EQUILIBRIUM OF NATURAL HATERS (Revised) .  U.S. Geological Survey
   Water-Resources Investigations Report 76-13

PLUMMER L.N., PARKHURST D.L., FLEMING Q.W.. and DUNKLE S.A. (1988)  A
   COMPUTER PROGRAM INCORPORATING PITZER 'S EQUATIONS FOR CALCULATIONS OP
   GEOCHEMICAL REACTIONS IN BRINES.  U.S. Geological Survey Water-
   Resources Investigations Report 88-4153.  Version 0.2.

PLUMMER L.N., BUSBY J.P., LEE R.W. , and HANSHAW B.B. (1990).
   Geochemical Modeling of the Madison Aquifer in Parts of Montana,
   Wyoming, and South Dakota.  Water Res. Res. 26, No. 9, 1981-2014.

PLUMMER L.N., PRESTEMON B.C., and PARKHURST D.L.  (1991)  AN INTERACTIVE
   CODE  (NETPATH) FOR MODELING NET GEOCHEMICAL REACTIONS ALONG A FLOW
,   PATH.  U.S.G.S. Water-Resources Investigations Report 91-4078.

POTTER R.W. II and NORDSTROM D.K. (1977)  The weathering of sulfide ores
   in Shasta County, California, U.S.A.  2nd Int. Symp. Water-Rock
   Interactions, Strasbourg, France.  1-42 - 1-46.

-------
                                                                     185

PTI Environmental Services  (1992)  Chemogenesis of  the Gold Quarry Pit
   Lake, Eureka County, Nevada.  Prepared  for Newmont Gold Company.

RADTKE A.S., FOO A.T., and PERCIVAL T.J.  (1987)  Geological and  Chemical
   Features of the Cortez Gold Deposit, Lander County Nevada.  Bulk
   Mineable Precious Metal Deposits of  the Western  United States;
   Guidebook for Field Trips. 319-325.

RAMPE J.J. and RUNNBLLS D.D.  (1989)  Contamination  of water and  sediment
   in a desert stream by metals from an abandoned gold mine and  mill,
   Eureka District, Arizona, U.S.A.  Applied Geochemistry 4, 445-454.

ROBINSON R.A. and STOKES R.H.  (1955)  Electrolyte Solutions.
   Butterworths Scientific Publications, London.  559 pp.

RUNNELLS D.D. and LINDBERG R.D.  (1990)  Selenium in aqueous solutions:
   the impossibility of obtaining a meaningful Eh using a platinum
   electrode with implications for modeling of natural waters.   Geology
   18, 212-215

RUNNELLS D.D. and SKODA R.E.  (1990)  Redox modeling of arsenic in  the
   presence of iron:  applications to equilibrium computer modeling, in
   Proceedings:  Environmental research conference  on ground water
   quality and waste disposal.  EPRI Report EN-6749, p. 22-1 to  22-11.

SAWYER C.N., and McCARTY P.L.  (1978)  Chemistry for Environmental
   Engineering.  McGraw-Hill, 532 pp.

SCHECHER W.D. and McAVOY D.C.  (1991)  M1NEQL*:  A Chemical Equilibrium
   Program for Personal Computers.  User's Manual Version 2.1.
   Environmental Research Software/Proctor and Gamble Company.

SEYLER P. and MARTIN J.M. (1989)  Biogeochemical Processea Affecting..;...
   Arsenic Species Distribution in a Permanently Stratified-Lake.-  ,  -~-
   Environmental Science and Technology 23,  1258-1263.

SINGER P.C., and STDMM W. (1970)  Acidic Mine Drainage:  The Rate-
   Determining Step.  Science 167, 1121-1123.

SMITH R.H. and JENNB B.A. (1991)  Recalculation, Evaluation, and
  Prediction of Surface Complexation Constanta for  Metal Adsorption on
  Iron and Manganese Oxides.  Environ.  Sci.  Tech. 25, S2S-S31.

SRK  (STEFFEN ROBERTSON and KIRSTEN, B.C.,  Inc.) (1989).  Draft Acid Rock
   Drainage Technical Guide, Volume 1.  British Columbia Acid Mine
   Drainage Task Force Report.  Prepared in association with Norecol
   Environmental Consultants, and Gormely  process Engineering.

STUMM W. and MORGAN J.J.  (1981)  Aquatic Chemistry.  John Wiley  t Sons.
   780 pp.

SHALLOW K.C., HUME D.N. and MOREL F.M.M.  (1980)  Sorption of copper and
   lead by hydrous ferric oxide.  Environ. Sci. Tech. 14, 1326-1331.

TESSIER A., RAPIN, F. and CARIGNAN R.  (1985)  Trace metals in oxic lake
   sediments: possible adsorption onto  iron oxyhydroxides.  Geochimica
   et Cosmochiodca Acta 49, 183-194.

-------
                                                                     186

THORSTENSON D.C.  (1984)  The Concept of Electron Activity and Its
   Relation to Redox Potentials in Aqueous Geochemical Systems.   U.S.
   Geological Survey Open File Report.

TRUESDELL A.H. and JONES B.F.  (1969)  Ion association in natural brines.
   Chemical Geology 4, 51-62

TRUESDELL A.H. and JONES B.F.  (1974)  WATEQ, A Computer Program for
   Calculating Chemical Equilibria of Natural Waters:   Journal of
   Research, U.S. Geological Survey, v. 2, no. 2.

TUREKIAN K.K. (1977)  The fate of metals in the oceans.   Geochlmica et
   Cosmochimica. Acta 41, 1139-1144.

U.C.  (UNIVERSITY OF CALIFORNIA) BERKELEY MINING WASTE STUDY TEAM (1988)
   Mining Waste Study, Final Report, 416 pp.

U.S. ARMY (1986)  CE-QUAL-R1:  A numerical one-dimensional  model of
   reservoir water quality;  User's Manual.  U.S. Army Engineers
   Waterways Experimental Station, Vicksburg, MS.  Instruction Report IR
   E-82-1.

VAN BREEMEN, N.   (1973) .  Calculation of ionic activities  in natural
   waters.  Geochimica et Cosmochimica Acta 37, 101-107.

VOKT E. (1992)  Geological Engineer, Cortez Gold Mines.   Personal
   communication.

WANG H.F.  and ANDERSON M.P.  (1982)  Introduction to Groundwater
   Modeling, finite difference and finite element methods.   W.-H.  Freeman
   and Company,  New York.

HELLS J.D.,  STOISER L.R. and ELLIOTT J.E. (1969)  Geology and
   Geochemistry of the Cortez Gold Deposit, Nevada.   Economic Geology
   64, 526-537.

WELLS J.D. and MULLENS T.E.  (1973)  Gold-Bear ing Arsenian Pyrite
   Determined by Microprobe Analysis, Cortez and Carlin Gold Mines,
   Nevada.  Economic Geology 68, 187-201.

WETZEL R.G.   (1983)  Limnology, second edition.  Saunders  College
   Publishing.  767 pp.

WICKS C.M.,  HERMAN J.S., and MILLS A.L.  (1991)  Early diagenesis of
   sulfur in the sediments of lakes that receive acid mine  drainage.
   Applied Geochemistry 6, p. 213-224.

HICKS C.M.,  and GROVES C.Q.  (1993)  Acidic mine drainage  in carbonate
   terrains:  geochemical processes and rates of calcite  dissolution.
   Journal of Hydrology 146, 13-27.

WIGLEY T.M.L. (1977)  WATSPEC:  A computer program for determining
   equilibrium speciation of aqueous solutions:  British
   Geomorphological Research Group Technical Bulletin no. 20,  48  pp.

WILLIAMS D.I. and VON HERZEN R.P.  (1983)  On the Terrestrial Reatflow
   and Physical Limnology of Crater Lake, Oregon.  Journal  of
   Geophysical Research 88, B2, 1094-1104.

-------
                                                                     187

WOLERY T.J. (1992)  EQ3/6,  A Software Package for Geochemical Modeling
   of Aqueous Systems:  Package Overview and Installation Guide  (Version
   7,0).   Lawrence Liverraore National Laboratory.

WOODWARD-CLYDE CONSULTANTS  (1992)  Screencheck Environmental Assessment,
   Robinson Project,  prepared for Magma Copper Company, Tucson, Arizona,
   (Draft).

YEH G.T-  and TRIPATHI V.S.  (1989)  HYDROGEOCHEM:  A Coupled Model of
   Hydrologic Transport and Geochemical Equilibria of Reactive
   Multicomponent Systems.   Oak Ridge National Laboratory Report ORNL-
   6371,  Oak Ridge, TN

-------
                           APPENDIX A
                 Debye-Hiickel 4 and b parameters
                                                                   188
   Ion
H*
SrHCO,*
SrOH*
SrCO,
Cu(S4),-
                            lon
      9.0
      5.4
      5.0
      0.0
     23.0
     25.0
      6.5
S J"
 S*-
 
                                           8.0
                                          10.0
                                          12.0
                                          14.0
                                          22.0
                                          24.0
                                          15.0
    Ion
   Mg2*
   Ka*
   K*
   Cl"
 &

5.0
5.5
4.0
3.5
3.5
              Ion
0.165
0.20
0.075
0.015
0.015
soJ-
HCO,'
CO,"
H,CO,*
SrS.
5.0
5.4
5.4
0.0
5.26
-0.04
0.0
0.0
0.0
0.121

-------
                                                            189




                          APPENDIX B
Cortex pit water mass transfer models calculated in BALANCE
                       Model # 1  *******************
1:
2:
3:
4:
S:
6:
7:
8:
9:
10:
11:
12:
13:
14
15:
16:
17:
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
:

:

.

*
.

:
:








:
*
*
:
*

*
*
*



Cortez Pit water mass transfer
C 4626.570 1838.800
S 939.0208 385.200
AS 6.0904 .134
BA 33.680 .000
CA1132.740 1272.500
FE 524.110 6.088
K 299.250 127.900
MG 954.540 567.800
MN .03094 1.440
NA2985.240 1348.400
HG .00798 .000
ZN .32112 .3365
F 126.326 26.300
AL .000 .000
SI4785.780 1385.400
PB 1.63369 .000

CALCITE F CA
"CH20" +C
DOLOMITE +CA
FLUORITBF+CA
ArsenopyF+AS
BARITE F+BA
GOETHITE -FB
PYRITE F+FE
CinnabarF+HG
K-MICA F K
MAGNESIT MG
NaCl F+NA
Ca/Na EX NA
Galena F+PB
SphaleriF+ZN
Mn02 MN
ILLITB K
CO2 GAS F C
K-FBLDSPF+K
Gypsum CA
S102 +SI
SOH:Zn F ZN
KAOLINIT -AL
Ca-Mont -CA
K-Mont K
PlagioclF+AL
Mg/Na EX NA
Na-Mont -NA

1
1
1
1
1
1
1
1
1
1
1
1
2
1
1
1
0
1
1
1
1
1
2
0
0
1
2
0

.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.600
.000
.000
.000
.000
.000
.000
.167
.330
.500
.000
.330

C

MG
F
FB
S
RS
S
S
AL
C
CL
CA
S
S
RS
MG
RS
AL
S


SI
AL
AL
CA
Mg
AL

1.

1.
2.
1.
1.
3.
2.
1.
3.
1.
1.
-1
1.
1.
4.
0.
4.
1.
1.


2.
2.
2.
B
-1
2.

000

000
000
000
000
000
000
000
000
000
000
.000
000
000
000
250
000
000
000


000
330
330
500
.000
330

RS

C

S "
RS

RS

SI
RS





AL

SI
RS.



SI
SI
NA

SI

4.

2.

i.
6.

0.

3.
4.





2.

3.
6.



3.
3.
.

3.

000

000 RS 8 . 000 ,
- t "  -- - -
ooo" ~--~T:
ooo -":
'
000

000
000





300 SI 3.500

000
000



670
670
500 SI 2.500

670

-------
                                                                      190
Plagioclase +
Fluorite +
SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Kaolinite
Magnesit
MnO2
Plagioclase +
Fluorite +
=SOH : Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Gibbsite
Magnesit
Mn02
F
F
F
F
F
F
F
F
F.
F
F
F




F
F
F
F
F
F
F
F
F
F
F
F




3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3608.0547
108.0000
512.0656
33.6800
171.3500
-1207.0247
-511.5883
-2378.9350
386.7400
-1.4091
1154.5320
50.0130
- .0154
1.6337
.0070
5.9564
2656.4807
1059.5740
512.0656
33.6800
171.3500
-255.4507
-511.5883
-1903.1480
386.7400
-1.4091
Plagioclase +
Fluorite +
-SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Zllite
Rhodochr
Mg/Na EX
Plagioclase +
Fluorite +
=SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum *
Illite
Rhodochr
Dolomite +
F
F
F
F
F
F
F
F
F
F
F
F




F
F
F
F
F
F
F
F
F
F
F
F




3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3996.2038
2280.8535
512.0656
33.6800
1850.5982
-1207.0247
-511.5883
-2798.7471
-1.4091
-1086.4268
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
2909.7770
108.0000
512.0656
33.6800
1850.5982
...-2293,4515
-^:=-511-.S883
:~2-79S-3471
1.4091
1086.4268
Plagioclase +  F   3057.6800
Fluorite    +  F     50.0130
SOH:Zn        F     - .0154
Galena      +  F      1.6337
Cinnabar    +  F     , .0070
Arsenopy    +  F      5.9564
CO2 gas        F   2908.3680
NaCl        +  F    108.0000
Pyrite      +  F    512.0656
Barite      +  F     33.6800
K-Feldsp    +  F   1850.5982
Calcite        F  -1207.0247
Gypsum             -511.5883
Illite      -     -2798.7471
Magnesit           1086.4268
MnO2                 -1.4091
Plagioclase +  F
Fluorite    +  F
SOH:Zn        F
Galena      +  F
Cinnabar    +  P
Arsenopy
CO2 gas
NaCl
Pyrite
Barite
K-Feldsp
Calcite
Gypsum
Gibbsite
Rhodochr
Mg/Na Ex
F
F
F
F
F
F
F
 1154.5320
   50.0130
    .0154
    1.6337
     .0070
    5.9564
 3044.6298
 1833.0540
 .512.0656
   33.6800
  171.3500
 -255.4507
 -511.5883
-1903.1480
   -1.4091
 -386.7400

-------
191
Plagioclase +
Fluorite +
-SOH : Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite 
K-Feldsp +
Calcite .-
Gypsum
Kaolinite
Magnesit
Rhodochr +
Plagioclase +
Fluorite +
-SOH:Zn
Galena +
Cinnabar +
Arsenopy +
CO2 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Gibbsite
Magnesit
Rhodochr +
Plagioclase +
Fluorite +
SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Illite
Magnesit
Rhodochr +
F
F
F
F
F
F
F
F
F
F
F
F




F
F
F
F
F
F
F
F
F
F
F
F




F
F
F
F
F
F
F
F
F
F
F
F




3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3609.4638
108.0000
512.0656
33.6800
171.3500
-1207.0247
-511.5883
-2378.9350
386.7400
-1.4091
1154.5320
50.0130
- .0154
1.6337
.0070
5.9S64
2657.8898
1059.5740
512.0656
33.6800
171.3500
-255.4507
-511.5883
-1903.1480
386.7400
- -1.4091
3057.6800
50.0130
- .0154
1.6337
.0070
5 . 9564
2909.7770
108.0000
512.0656
33.6800
1850.5982
-1207..0247
-511.5883
-2798.7471
1086.4268
-1.4091
Plagioclase +
Fluorite +
SOH:Zn
Galena +
Cinnabar +
Arsenopy +
CO2 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Gibbsite
Rhodochr
Dolomite +
Plagioclase +
Fluorite +
>SOH:Zn
Galena +
Cinnabar +
Arsenopy +
CO2 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum.
Kaolinite
Rhodochr
Dolomite +
Plagioclase +
Fluorite +
SOH:Zn
Galena +
Cinnabar +
Arsenopy +
002 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Kaolinite
Rhodochr
Dolomite +
F
F
F
F
F
F
F
F
F
F
F
F




F
F
F
F
F
F
F
F
F
F
F
F














F
F




1154.5320
50.0130
- .0154
1.6337
.0070
5.9564
2657.8898
1059.5740
512.0656
33.6800
171.3500
-642.1907
-511.5883
-1903.1480
-1.4091
386.7400
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3996.2038
881.4800
512.0656
33.6800
171.3500
-1207.0247
-511.5883
.-2378.9350
r 1.4091
...386.7400
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3609.4638
108.0000
512.0656
33.6800
171.3500
-1593.7647
-511.5883
-2378.9350
-1.4091
386.7400

-------
o)gf3S??$?8rT22
M O 0-rJ MiihHowwBM
O N 0- CO O A H- H- M._  A B A
rr
A
    en
    H-
    A
H- M ft ft
rr p. A ~
A u
  TJ
 W
ti-M
o e P
sa out
N H-O
*!?M
    ft)
                                                     xooT?5s
                                          R
1 H* tt n A H- H-
 H- C H- M rr rr
 3 3 ft O. A A

 J?  *
 A
                                                                   o
                                                                       M 1] "O
                                                               O O* (U
                                                               en -d to
                                                                 < H
                                          H f 
                                          H- O
                                          rr o
                                          (0 M
                                             ftt
                                             w
                                             0

                                          -r +
                                                                                     MO O  _
                                                                                     o to M u o
                                           3   H-
                                           H-   a
                                           rr
                                           A
                                               ct
                                               A
    n H n to  ,. .
    s^-^sss:?
ss-**   s^^g
  a          < h
c P>
OIQ
1 H-
H- O
rr o
A H-

  u
  A
     H I i
Ul   to Ul 
                                             O
                                         imui
                                      H   O-J
                                                                                          i     i
                                                                                          M I H         Ul            u>
                                                                                     Ul   Ul U Ul H   Ul H Ot            O
                                                                                     0lvlHO-4UIHOO        IUIUI
                                                                                     aiHOOHUiHUiK>a>aui   M   o ^i
                                                                                     OtOUIOi^OOUIOi>Otv)UIUIUIO
     me
     H-B
             W9 55 O >OO 
             p> < PI O H H- * CO
          - H rr rf
         r to
           "d
                     jo
                      TJ
                               I MM
                                          9i
I H-O
ft O

  P)
  a
           O M H-M

           H.  ?
                                          ft
                                          A
  M H- W O X H- H-
     rr e H- M rt rr
         ft PI A A
          I (A
           TJ
                                                                i 1/1 p H1 O C 01
                                                                 (ft 5 (0 Jlj 0 iQ
                                                              IQ 3 pi 3  < H-
                                                                O &  N H- O
                                                               u no pi    3 rr o
                                                                 < H      AM

                                                                            to
                                                                                 M a o yt to w as
                                                                                 M*< PI I  PI < 0
                                                                                 MW M *i H H fl
                                                                                 H- 01 O A H- H- M
                                                                                 rr C H- M rr rr
                                                                                 A 3 ft p, A A
                                                                                     A U
                                                                                       TJ
                                                                                 S*1
                                                                    N
                                                                    a
                       ota

                       H- o
                       ft O
                       A M
                         PI
                         (0
                         A
                  M U)
                Ul OS O
   I  OHUI-IUIHUtifc
   HUIMUlHU*K)UU(
       l*
     tOUIMH
                      Ul
                                 H
                             I Ul Ul
                               o *
                                               I
                                               M  I
                                                   i
                                                   K)
           O  M I K) M       10
           o>  4Uitoa>   tn M u>
           Oil tOMtOUIUIHOO
                                  ui
                                  o
                               Ul Ul
                               O -J
                                                                                     H    I    I
                                                                                     O   K) I H H     K) Ul
                                                                                     09   x)UtK)   UIKItO
                                                                                     at  i toMouiuii-'a>o
                                                                                                              o
                                                                                                           in in
                                                                                                           o -J
*kOifrODUtUIOBO\UtK>UIOUtMHU
OU>OBOOOOUIOA
                                                                                                                             vo
                                                                                                                             K)

-------