Beruflich Dokumente
Kultur Dokumente
net/publication/293655512
CITATIONS READS
0 1,319
1 author:
Geoffrey D. Thyne
Aalborg University - Esberg
98 PUBLICATIONS 1,793 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Geoffrey D. Thyne on 12 October 2017.
October, 2007
2
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
TABLE OF CONTENTS
Computer Models 5
Review of the Geochemical Principles 10
Units of concentration – wt./wt., wt./vol., moles 10
Review of Basic Thermodynamics 11
Mass action equations 12
Saturation index 14
The PHREEQC Model 15
Getting Started 16
Equilibrium of water and calcite 17
Concepts of Aqueous Speciation and Solubility of Solids and Gases 20
Acid-base reactions 21
PHREEQC examples of solubility of calcite and atmospheric CO2 21
REACTION command (a reaction path) 23
SELECTED_OUTPUT command – first use 25
INCREMENTAL_REACTIONS command 26
Alkalinity 29
Fixed pH 31
The pH Sweep
Titration of AMD – Acidity
Introduction to KINETICS and the linked RATE commands 34
Activity and Concentration 36
Salt example 29
Databases 38
Mt. Edna Solicchiata example – how to use the other databases 39
Adding new species - Ground water with benzene 41
Metal database example
Managing the Input and Output 42
SOLUTION_SPREAD command 42
SELECTED OUTPUT example 42
Introduction to Redox 32
Basic Principles
Principles of AMD generation 65
Oxidation of AMD solution 35
Oxidation of organic matter 36
Specifying Redox - uranium and speciation of seawater 36
Example of passive treatment
Fixed Eh 38
Surface Reactions 44
Implicit Surfaces 44
Kd partitioning formulation (isotherms) 44
Sorption with Frendlich and Langmuir isotherms 46
Explicit Surfaces 48
SURFACE command 51
Specifying the surface 52
Organic surfaces 57
Reaction Path Models
Mixing
The command and examples
Water Compatibility
Mixing, Redox and Surfaces Combined-AMD mixing with stream water 55
3
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Evaporation
Inverse Modeling 60
Concepts of Inverse Modeling-Sierra springs example 60
Complex AMD Example 65
Transport –1-D Advection and Dispersion 73
Simulation of continuous landfill leachate leak 75
Simulation of continuous landfill leachate leak with sorption 79
Simulation of pulse landfill leachate leak with sorption 80
Simulation of AMD in a carbonate column
Uncertainty in Models
References 84
Appendix of Input Files 85
4
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Computer Models
Capabilities and Methods
The computer codes require initial input constraints that generally consist of water
chemistry analyses, units of the measurement, temperature, dissolved gas content, pH and
redox potential (Eh). The models work by converting the chemical concentrations,
usually reported in wt./wt. or wt./volume terms such as mg/kg or mg/L, to moles, and
then solving a series of simultaneous non-linear algebraic equations (chemical reaction,
charge balance and mass balance equations) to determine the activity-concentration
relationship for all the chemical species in the specified system. The models usually
require electrical balance, and will force charge balance with one of the components (that
can be designated), as they solve the matrix of non-linear equations. The capabilities of
modern codes include calculation of pH and Eh, speciation of aqueous species,
equilibrium with gases and minerals, oxidation and reduction reactions (redox), kinetic
reactions and reactions with surfaces.
The non-linear algebraic equations are solved using an iterative approach by the Newton-
Raphson method (Bethke, 1996). The equations to be solved are drawn from a database
that contains equations in the standard chemical mass action form. In theory, any
reaction such as sorption of solute to surface that can be represented in this form can be
incorporated into the model. These equations represent chemical interactions with
reactants on the right and products on the left. Reactions are assumed to reach
equilibrium (the point of lowest free energy in the system) when there is no change in
concentration on either side.
Note that the arrow in the calcite dissolution example above goes in both ways, that is the
reaction as written is reversible. Once a mineral reaches equilibrium with a solution,
adding more mineral will not increase the dissolved concentration since we have already
saturated the solution. But removing ions from the right side (e.g. lowering the
concentration by dilution with distilled water), will cause more solid to dissolve.
Kinetic reactions, those involving time, are included by assuming that the chemical
reaction will proceed to equilibrium, but at a specified rate. The available kinetic
reactions include mineral dissolution and precipitation, redox reactions and microbial
growth and metabolism of solutes. The rate laws used in the codes vary, but all codes
with kinetic capabilities include simple first order rate laws, and may include more
5
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
complex rate formulations such as cross-affinity, Michaelis-Menten and Monod
formulations (Bethke, 2002).
Geochemical computer models are divided into two basic types, speciation models and
reaction-path models. In both cases, the models are fundamentally static, that is there is
no explicit transport function, however some forms of transport can be simulated by
manipulation of the models. More complex reaction-transport models that explicitly
incorporate transport are briefly described below.
All the equilibrium models are speciation models in that they can calculate the speciation
(distribution) of aqueous species for any element or compound included in the database.
Speciation models calculate activities (chemically reactive concentration), species
distribution for elements in the database, mineral saturation indices, gas fugacities and
ion ratios at the specified conditions of pH and redox potential (ORP or Eh). Most of the
models allow selection of method of activity calculation (Davies, Debye-Huckel,
extended Debye-Huckel, Pitzer). Some models incorporate surface reactions such as
adsorption and multiple kinetic formulations. Only one model, PHREEQC, has the
inverse modeling option. This feature uses mass balance constraints to calculate the mass
transfer of minerals and gases that would produce an ending water composition given a
specified starting water composition (Garrels and Mackenzie, 1967). This method does
not model mass transport, only calculates and provides statistical measures of fit for
possible solutions to the mass balance between starting and ending water compositions.
6
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
To add components to the system, you use the Input tab. The components added to
define the system must be defined as SOLUTION_MASTER SPECIES. This can be a
little difficult at first when you are entering data from a water analysis that lists HCO3 for
inorganic carbon, while the PHREEQC input requires the inorganic carbon to be entered
as Alkalinity. However, the program can automatically convert the chemical analysis
parameter to the basis species if correctly input. By default, there is 1 kg of water in the
system automatically. All the programs work that way. The reason for using 1 kg is that
all the programs use concentrations in units of molal (moles/kg) even if you enter the
concentrations in other units.
This brings us to the subject of how the program actually works. Instead of calculating
the solution to a matrix with ninety species (90x90), the strategy is to calculate the
smallest possible matrix and then distribute the total amount of each basis species among
7
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
all possible related species. So Ca (really the total calcium in this system) is distributed
among Ca++, CaHCO3+, CaOH+ and CaCO3 after the interactions among the basis species
are accounted for. This strategy simplifies the calculations, but does confuse many new
users who struggle with the concept of a basis. The figure below is taken from Bethke
(1996), an excellent reference for the interested reader.
The next step in complexity is the reaction path (mass transfer) models. The reaction
path models use speciation calculation as starting point and then make forward
predictions of changes along the specified reaction path (specified change in T, P, pH,
addition of new reactants such as another fluid or solid). The program makes small
incremental steps with step-wise addition or removal of mass (dissolution or
precipitation), and can include changes in temperature or pressure along the reaction path.
Besides the theoretical problems with using models such as the assumption of
equilibrium in complex natural systems, there are practical limitations with employing
any of the models. The field input data may be corrupt with bad analysis, missing
parameter, or electrical imbalance. The databases are also a source of uncertainty. They
do not always contain all the elements or species of interest, the data has some
uncertainty, and some data may be inaccurate data (Drever, 1997). Some of the available
codes try and minimize this problem by including several of the most popular databases
such as the MINTEQ database (EPA-approved database specializing in metals), WATEQ
(USGS database specializing in minerals), and the LLNL database (the most complete
and internally consistent database available which is compiled and maintained by
Lawrence Livermore National Laboratory). For environmental applications, the limited
data for organic compounds remains a concern.
8
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Other limitations include the redox reactions that are of particular importance in metal
transport. These reactions are difficult to model correctly since redox reactions may have
different rates producing natural systems that are not in redox equilibrium (Lindberg and
Runnels, 1984). This problem can be addressed by modeling redox reactions as rate-
limited (kinetic) formulations if the data is available.
The most complex models explicitly incorporate transport and reaction. The codes
couple and solve both the partial differential equations of flow using the advective-
dispersion equation and the non-linear algebraic equations of chemical equilibrium. The
general approach is to solve for the reaction term in each cell using the chemical module
of the code, then separately solve for the effects of transport (split-operator approach).
The effects of adsorption are solved in the transport module using the retardation portion
of the equation (Appelo and Postma, 1993, Parkhurst and Appelo, 1999, Zhu and
Anderson, 2002). These models are much more complex than the reaction path models.
Presently there are only three commonly used models that have transport capabilities,
HYDROGEOCHEM2, PHREEQC (1D) and the related PHAST (3D) code, and
Geochemist’s Workbench® Professional.
The following table lists the most commonly used programs, their sources and some of
the most useful capabilities. The list is not meant to be exhaustive rather it offers an
overview to serve as a starting point for further investigation. Details of the capabilities
of each program can be found on the listed websites or in the manuals. More detailed
comparison of these and other models are available in related publications (Nordstrom et
al. 1979, Engesgaard and Christensen, 1988, Mangold and Tsang, 1991).
9
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Review of the Geochemical Principles
The earth is composed of numerous elements of which only a few are present in
significant amounts. In order, Si, Mg, Fe, Al, Ca, Na, Cr, Mn, P, K, Ti and Ni as oxides
form 98.8+% of the earth’s mantle, and is the source of surface rocks. These elements
are referred to as the major elements because of their abundance. The situation is
different in natural waters where the major elements are usually Na, Ca, K, Mg, C as
HCO3, S as SO4, Si, and Cl. These elements are the most soluble, where the more
common elements such as Si, Fe and Al have low solubility. Minor elements have, by
definition, concentrations below 5 ppm and may include Fe, B, F, P and N, while trace
elements are present in amounts less than 1 ppm ( Mn, Ni, Co, As, St, Li, etc.).
The elements are segregated in the earth’s crust by tectonic processes (heat and pressure)
into three general classes. The first class is the Lithophiles, which are concentrated in the
crust (Li, Be, B, Si, Na, Al, Ca, Sc, Ti, V, and Mn, and the large ion lithophiles (rock-
loving) or incompatibles that include K, Rb, Cs, Sr, Ba, Y, REE’s, Zr, Hf, Nb, Ta, W, Th
and U. These elements are concentrated in granites, and some basaltic to intermediate
rocks. The second class is the siderophiles (iron-loving) that include Fe, Co, Ni, Au and
Pt group elements and are found concentrated in ultramafic rocks. The third class is the
chalcophile (copper-loving) that include Cu, Zn, Ga, Ag, Cd, S As and Sb) and are
concentrated in basaltic and intermediate rocks. These classes are general and there are
notable deviations.
This situation is complicated during weathering when the different solubility of minerals
produces a fractionation, with the more mobile elements being separated from the less
mobile. Certain sedimentary rocks concentrate mobile elements as they form. For
instance, limestone concentrates Ca, Mg and Sr.
Example
One mole of H2O is (2 x 1g) for the two hydrogen atoms, and (1 x 16g) for the one
oxygen atom for a total of 18g/mole. One liter of water has about 55.8 moles of water. If
there is one mole of an element (Ca) dissolved in one liter of water, then the solution is
one molar Ca (1 mole/L) or 40.08g/L Ca. In some applications we are interested in the
amount of charge of an element, molecule or compound. In that case we calculate the
equivalents or moles x charge.
10
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Example
One mole of CaCl2 is equal to (40.08 g/mole x 1) for Ca and (35.45 g/mole x 2) for Cl for
a total of 110.98 g/mole. However, if 110.98 g of CaCl2 is dissolved in one liter of water
(40.08 g/l and 70.9g/l Cl), there are two equivalents of Ca (1 mole x 2 charges) and two
equivalents of Cl (2 moles x 1 charge) in solution.
The basic idea is that elements, molecules and compounds all contain some internal
energy and that all systems try to reach a state where that energy is minimized
(equilibrium). Natural systems, particularly low temperature systems, do not always
reach equilibrium, but they do move in that direction. The internal energy of an element,
molecule or compound is expressed as its enthalpy or internal heat. The “free energy of
an element is an element, molecule or compound is the sum of its internal heat and its
internal tendency toward disorder (entropy).
G = H – TS
Assuming constant T and P we get
ΔG = ΔH – T ΔS
The total free energy of a component in the system is dependent on this inherent energy
of an element, molecule or compound and the amount (concentration). When two or
more elements, molecules and compounds are combined, the result is a reaction that
minimizes the energy of the new system, lowering the ΔG. The free energy of a reaction
is calculated by:
ΔG0 rxn. = ∑ΔG0 products - ∑ΔG0 reactants
The ΔG0 values (standard free energy) for many compounds can be found in the back of
textbooks such as Drever. This way we can calculate a ΔG value for any reaction that for
which we can write a balanced chemical equation. The minimum energy state
(equilibrium) between the reactants and products is related to the ΔG value by:
11
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
where: R is the universal gas constant in kJ/mole
T is the temperature in Kelvin
And K is the equilibrium constant for the reaction
Note that the arrow in the calcite dissolution example goes in both ways, the reaction as
written is reversible. Once a mineral reaches equilibrium with a solution, adding more
mineral will not increase the dissolved concentration since we have already saturated the
solution. But removing ions from the right side (e.g. lowering the concentration by
dilution with distilled water), will cause more mineral to dissolve. If we add more ions to
solution, more calcite will precipitate until the balance between ions and solid is restored.
This is referred to as Le Châtelier’s Principle.
The reaction above is for one mole of calcite dissolving to form one mole of calcium
cation and one mole of carbonate anion. The reaction must be balanced with the same
number of moles of each element and the same moles of charge on each side of the
equation. The dissolved calcium and carbonate are often referred to as dissolved species.
This reaction is the dissolution of calcite (reactant(s) to product(s), left to right). We
express this in a mathematical form where, at equilibrium, the ratio of the concentration
of reactants (on the bottom) and products (on the top) is equal to K, known as the
proportionality constant or distribution coefficient or equilibrium constant.
The value for Kcal at 25°C is 10-8.48 in most books (Table 3.1 in Drever). We know that
the K values are dependent on temperature and pressure. For instance, as temperature
increases, calcite becomes less soluble (retrograde solubility) and the K value is smaller.
12
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Example of temperature-dependent equilibrium constants, modified from Drever, 1999.
The typical range of values for equilibrium constants, concentrations, etc. used in
geochemistry is very large (10+ orders of magnitude), so we often use the log of a value
and since many values are very small (10-5), we use the convention of p, where p = -log.
For example:
pH = -log a H+
pe = -log a e-
pK = -log K
In the case where the reaction is written as the dissolution of a solid (mineral), the K is
called the solubility product or Ksp.
13
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Saturation Index
One advantage of the mass action formalism is that we can express the degree of solution
saturation relative to the saturated condition as an index. In order to calculate the degree
of saturation, PHREEQC calculates the ion activity product (IAP) of the actual solution
(the numerator of the dissolution reaction expression) and uses that quantity and the
solubility constant (Ksp) to compute the SI.
For instance, in the case of calcite solubility we can see that if the activity of Ca 2+ x CO32-
in a solution is less than 10-8.48 at 25°C (see table above), then the solution is under-
saturated with respect to calcite. In PHREEQC the SI is reported in units of log SI, so SI
= 0 is saturated and SI >0 is super-saturated and SI<0 is under-saturated.
14
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
The PHREEQC Model
PHREEQC can perform batch reactions, which are mixing different waters or water and
minerals or water and gases, etc. together to create a system (the bucket). The model
calculates the distribution of aqueous, gaseous and solid species in the specified system.
PHREEQC includes adsorption, kinetics and transport capabilities for aqueous species
within the system. Access to the PHREEQC code and support on the internet can be
found at http://wwwbrr.cr.usgs.gov/projects/GWC_coupled/phreeqc/index.html. The
latest version of PHREEQC is 2.12.03, which is used in this course. Another useful
alternative is the batch version that allows users to execute large jobs or link PHREEQC
to other programs. Users can also download PhreeqcI, which has an interactive input
window. All the versions use the same basic code and solve problems the same way.
The PHREEQC model uses three types of keywords (commands) that are input using text
files. The first type is the “manager” commands such as:
KNOBS, PRINT, TITLE, USE, END, SAVE, SELECTED_OUTPUT,
SOLUTION_SPREAD, etc.
The third type is the batch-action or transport that will perform an action:
SOLUTION, REACTION, INCREMENTAL_REACTIONS, MIX, SURFACE,
ADVECTION, INVERSE_MODELING, EXCHANGE, EQUILIBRIUM_PHASES,
KINETICS, etc.
Command blocks may be strung together in an input deck to simulate complex situations.
Batch reactions are implicitly defined and executed whenever a solution and any one of
the keyword data blocks, EXCHANGE, EQUILIBRIUM_PHASES, GAS_PHASE,
KINETICS, MIX, REACTION, REACTION_TEMPERATURE, SOLID_SOLUTIONS,
or SURFACE, is also defined in the same simulation (not separated by END). There are
subcommands for each command keywords that further refine the calculations. For
instance, the command SOLUTION has subcommands such as temp, units, pe, and pH
that can be used to set these values.
Spreadsheet data make be imported directly and selected output values can be written to a
text file for import into spreadsheet programs. Graphs of data send to the
SELECTED_OUTPUT file can be plotted and the plot exported. The version of the
program we will use has a simple window interface that facilitates modeling, using
multiple databases and x-y plotting capabilities.
15
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Getting Started
The SOLUTION command word causes the chemical species listed to be dissolved in 1
kg of water. Note command words are always in capitals. The first input file will
illustrate the command (input file 1). Note you can select and copy the text below from
the manual and paste it directly into the area provided under the PHREEQC Input tab or
type it in. All the input files will be refereed to in italics and are listed in the appendix at
the end of the manual. There is also a menu of command words on the right panel in the
PHREEQC window. You can double click on the command to automatically place it in
the input window. An important note: PHREEQC will save/write any files created to
the default directory (c:/Programs Files/Phreeqc) unless you reset the directory. To do
so, choose Edit»Preferences»Miscellaneous, and reset to the folder you want. This is
an important feature that allows you to control your workflow.
(input file 1)
SOLUTION 1 Mt. Edna Solicchiata
units mg/l
pH 6.90
pe -0.2
density 1.00
temp 17.2
Al 0.01
C(4) 1180 as HCO3
Ba 0.0296
Ca 28.1
Cd 0.00026
Cl 238 charge
Cu 0.0018
Fe 0.150
K 30.1
Li 0.180
Mg 214
Mn 0.0277
Na 250
Pb 0.017
S(6) 128
Si 79.4 as SiO2
Sr 0.673
Zn 0.0604
END
The output file is printed and the program automatically shifts the window showing that
tab. A brief examination of the file shows a great deal of information presented in a
complex format.
16
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Briefly, the output file shows the input file, then the result with the following major
categories:
The input values are the analytical results of a water analysis. Analytical results for
elements are totals. Speciation calculations partition the total amounts into all the
potential forms (e.g. total dissolved Ca into the dissolved species Ca 2+, CaOH+ and
Ca(OH)2º) depending on the relevant mass action equations in the database. This is the
basic information calculated for the 1 kg of water and specified amount of other
components at thermodynamic equilibrium. Actual solutions may not be at equilibrium.
The following PHREEQC exercise and is designed to further familiarize you with the
program’s input and output by calculating the concentration of Ca and CO 3 in a solution
that is in equilibrium with calcite. The first simulation produces a system where calcite
and pure water are in equilibrium (input file 1a). As you look at the output you should
note that the pH changes from an initial value of 7 to about 9.9 and that additional
dissolved species are formed besides Ca and CO3 ions.
TITLE calcite solublity example
SOLUTION 1 pure water & calcite
pH 7
temp 25
EQUILIBRIUM_PHASES
Calcite 0.0 1
END
17
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
EQUILIBRIUM_PHASES
Calcite 0.0 1
END
TITLE
----------------------Saturation indices-----------------------
Phase SI log IAP log KT
Reaction step 1.
Using solution 1. pure water
Using pure phase assemblage 1.
-------------------------Phase assemblage------------------------
Moles in assemblage
Phase SI log IAP log KT Initial Final Delta
18
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Calcite 0.00 -8.48 -8.48 1.000e+00 9.999e-01 -1.227e-04
---------------------Solution composition-----------------------
Elements Molality Moles
C 1.227e-04 1.227e-04
Ca 1.227e-04 1.227e-04
---------------------Description of solution--------------------
pH = 9.910 Charge balance
pe = -3.597 Adjusted to redox equilibrium
Activity of water = 1.000
Ionic strength = 3.848e-04
Mass of water (kg) = 1.000e+00
Total alkalinity (eq/kg) = 2.454e-04
Total CO2 (mol/kg) = 1.227e-04
Temperature (deg C) = 25.000
Electrical balance (eq) = -1.082e-10
Percent error, 100*(Cat-|An|)/(Cat+|An|) = -0.00
Iterations = 16
Total H = 1.110124e+02
Total O = 5.550658e+01
-----------------------Distribution of species-------------------
Log Log Log
Species Molality Activity Molality Activity Gamma
-----------------------Saturation indices-----------------------
Phase SI log IAP log KT
In this case there are two major blocks of output. The first shows the solution (pure
water) speciated at the specified temperature. The second block brings the solutioninto
equilibrium with 10 moles of calcite, dissolving about 0.0001 moles to reach equilibrium
and calculating the change in dissolved concentrations and pH. Note there are four
species of dissolved calcium listed, Ca2+, CaOH+ and CaCO3º, and CaHCO3+. In fact
there are more species, but their concentration s are less than the default cutoff
CO2(g) ↔ CO2(aq)
H2CO3 ↔ H+ + HCO3-
HCO3- ↔ H+ + CO32-
The equilibrium constants for these reactions are usually labeled KCO2, K1, and K2,
respectively. The reaction of CO2(aq) to form H2CO3 is usually combined with the
CO2(g) to form CO2(aq) reaction and the sum of the reactions has the constant KCO2.
These equations, together with the dissolution of calcite equation if calcite is present, can
be used to describe the carbonate system.
The alkalinity value reported for water samples is determined from a titration of the water
sample with H2SO4, and is reported as meq/l or eq/l CO3, meq/l or eq/l HCO3 or meq/l or
eq/l CaCO3. Alkalinity is actually equal to total HCO3- +CO32- + any other weak acids
that are negatively charged at natural water pH values of 7 to 10 (usually boron and
organic acids). This value is calculated by measuring the number of equivalents of acid
needed to change the water sample (of known volume) from its starting pH to a pH of 4.
So eq/l of H+ used = eq/l of alkalinity in the water. Sometimes this data is further
converted to eq/l or mg/l of HCO3 and /or CO3 using the pH of the water and assuming
that the ratio of HCO3/CO3 is a function of pH (10-10.3 = CO32- * H+/HCO3-).
20
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Acid-Base Reactions
The concentration of H+ in solution is a result of all the reactions that consume or
produce protons. The final pH value of a solution is dependent on the free energy of each
reaction. One measure of the inherent or intrinsic strength of a reaction is represented by
the K value. Strong acids (often called mineral acids), such as sulfuric or hydrochloric
have K values greater than 1, whereas weak acids such as acetic (vinegar) have smaller K
values (10-6.4).
Strictly speaking, pH is the negative log of the activity of the hydrogen ion (H+), and is
measured with an electrode. The concentration of H+ in solution is a result of all the
reactions that consume or produce protons. The final pH value of a solution is dependent
on the free energy of each reaction. One measure of the inherent or intrinsic strength of a
reaction is represented by the K value. Strong acids (often called mineral acids), such as
sulfuric or hydrochloric have K values greater than 1, whereas weak acids such as acetic
(vinegar) have smaller K values (10-6.4).
Since the equations for calcite solubility involve protons (H+ + HCO3- = H2CO3, etc.),
they are considered acid-base equations. That is they involve protons (consumption or
production). Since the reactions in the calcite system involve protons they affect pH,
which is the concentration of H+ in solution or correctly, pH is the -log aH+, and is
measured with an electrode. The next exercise returns to calcite solubility to investigate
the effect of pH on solubility. We can vary pH by setting the value by adding an acid or
base to the system. We will add the weak acid carbon dioxide by adding more CO 2 gas
to the system than just equilibrium with the atmosphere. Note that gas concentrations are
input as log of the partial pressure.
Next we will add CO2 gas to the system and observe changes (input file 1b). We add
CO2 gas as an equilibrium phase to mimic contact of the solution with the atmosphere.
Finally, we can add oxygen gas to the equilibrium phases, which will affect the redox
potential (input file 1c).
PHREEQC example of solubility of calcite and atmospheric CO2 - file “input file 1b”
Note that gas concentrations are entered as log fugacity (partial pressure).
TITLE pH example
SOLUTION 1 pure water & calcite & CO2 gas (atmospheric value)
pH 7
temp 25
EQUILIBRIUM_PHASES
Calcite 0.0 1
CO2(g) -3.5
END
21
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Activity of water = 1.000
Ionic strength = 1.001e-07
Mass of water (kg) = 1.000e+00
Total alkalinity (eq/kg) = 1.082e-10
Total carbon (mol/kg) = 0.000e+00
Total CO2 (mol/kg) = 0.000e+00
Temperature (deg C) = 25.000
Electrical balance (eq) = -1.082e-10
Percent error, 100*(Cat-|An|)/(Cat+|An|) = -0.05
Iterations = 0
Total H = 1.110124e+02
Total O = 5.550622e+01
-------------Distribution of species----------------------------
Reaction step 1.
Using solution 1. pure water & calcite & CO2 gas (atmospheric value)
Using pure phase assemblage 1.
-------------Description of solution----------------------------
pH = 8.279 Charge balance
pe = -1.966 Adjusted to redox equilibrium
Activity of water = 1.000
Ionic strength = 1.463e-03
Mass of water (kg) = 1.000e+00
Total alkalinity (eq/kg) = 9.867e-04
Total CO2 (mol/kg) = 9.803e-04
Temperature (deg C) = 25.000
Electrical balance (eq) = -1.082e-10
Percent error, 100*(Cat-|An|)/(Cat+|An|) = -0.00
Iterations = 15
Total H = 1.110124e+02
Total O = 5.550867e+0
In the first calcite example, the initial solution had a pH value of 7 but became more basic
with addition of calcite to the system. Addition of CO2 gas into the system adds acid and
lowers the pH value. Carbon dioxide is considered a weak acid and its addition to the
system did not lower the final pH (after the batch reaction) very much. However, if we
add more CO2 to the system, the final pH should be lower than the first case (input file =
input file 1d).
TITLE pH example
SOLUTION 1 pure water & calcite & CO2 gas (>atmospheric value)
pH 7
temp 25
EQUILIBRIUM_PHASES
Calcite 0.0 1
CO2(g) -1.5
END
----------------Saturation indices-------------------------------
Phase SI log IAP log KT
We can also add calcite in a series of steps to the solution and see the changes for each
step. For this we use the REACTION command (input file 1e). This is the first example
of a simple reaction path calculation. A reaction path model uses speciation calculations
to make forward predictions of changes in water and rock (dissolution/precipitation)
along reaction path (specified change in T, P, pH, new reactants).
23
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
The figure below shows a conceptual diagram where the system can be altered by adding
or subtracting mass of heat, or composition or controlled by contact with another
reservoir that buffers change. Reaction path models operate by first calculating the
equilibrium condition of the specified system, then the program changes the condition
(e.g. temperature, composition) by a small increment and calculates the new equilibrium
condition. Then the next step uses the last step as the starting point until the end of the
path is reached. In the example below, the increments along the reaction path are
specified. A good way to visualize this is the addition of base to a solution (titration) to
change the initial acidic pH of 2 to a value of 12 with a calculation made for each drop of
acid added. During the process, the solution may have minerals precipitate or dissolve,
gas content increase or decrease, etc. This method is used to model many real world
processes of interest such as mixing of fluids, dissolution of minerals, changes in
temperature, Eh, pH, fugacity or concentration of a particular species. PHREEQC differs
from other reaction path models such as Geochemist’s Workbench and EQ6 in that does
not automatically allow mineral precipitation and dissolution along the reaction path
unless specified and then only those minerals specified by the user.
24
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
TITLE pH example
SOLUTION 1 pure water & calcite
pH 7
temp 25
REACTION
Calcite
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
SELECTED_OUTPUT
-file N1dout.csv
-molalities Ca+2 HCO3- CO3-2
END
Note we have added another new command, SELECTED_OUTPUT to the input file
above (input file 1e). This writes user-selected portions of the regular output file to a
comma-delimited text file. In this example we have named the output file ‘N1dout.csv’;
the suffix means comma-separated-values. This command allows us to manage the large
amount of data that is generated by multiple simulations. The output file still contains the
complete output of each simulation. The command is discussed in more detail below.
We can also start to use the grid function in PHREEQC. This function allows us to view
and plot the data in the selected output file from within the PHREEQC shell. This
command crates a comma separated value (CSV) file named N1dout.csv that can be
opened using the Grid tab. The gird output file should have been written to the default
folder that you set. Opening the file under the Grid tab will create a window that looks
like this:
The columns on the far right will contain the parameters you listed, in this case the
molalities for Ca+2, HCO3- and CO3-2. This feature can be very useful for exporting to a
spreadsheet program for plotting and analysis of the calculated values. You can also use
the Chart tab in PHREEQC to plot values and export the chart to another text or graphic
program. To plot the values from the Grid tab on the Chart tab, select the desired values
and use the right hand button to select Plot in Chart. The left hand column will be the x-
25
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
axis and the other parameters will be plotted on the y-axis. The chart format can be
edited to suit your taste and the entire chart exported by choosing Edit » Copy chart »
Copy as Bitmap or Copy as Metafile. Metafiles are vector based files that can be scaled
without loss of resolution.
This last simulation added calcite to pure water and reached saturation in the first step;
the addition of 0.1 moles to the 1 kg of solution completely saturates the solution. The
second step adds 0.2 moles of calcite (20 g) to a different 1 kg of solution and so on, in
effect making 10 different simulations. Sometimes you want to add increments of a
reactant to the same solution. In that case, use the INCREMENTAL_REACTIONS
command rather than REACTION. An example is considered below.
First, graph the results of the REACTION simulation after you add the subcommand
saturation_indices line to the SELECTED_OUTPUT command block. Use the input
menu to add the command by clicking on the plus by the SELECTED_OUTPUT
command on the right hand side of the window and by double-clicking on the term. This
will insert the term into the input file wherever the cursor is located. We can see that we
have reached saturation with respect to Calcite very quickly.
Suppose you want to approach saturation more slowly simulating a titration and watch
the change in pH as you add Calcite in increments to a single solution. You can first
reduce the amount of calcite added by changing the numbers specifying the amounts
added or by changing the units. For instance, we can and reduce the amount of calcite by
making the input mmoles rather than moles (the default).
TITLE pH example
SOLUTION 1 pure water & calcite
pH 7
temp 25
INCREMENTAL_REACTIONS true
REACTION
Calcite
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 mmoles
SELECTED_OUTPUT
-file N1dout.csv
-molalities Ca+2 HCO3- CO3-2
-saturation_indices Calcite
END
This simulation adds 1 mmole of calcite to a single kg of water in 0.1 mmole steps. We
can look at the grid file and see that the incremental simulation does not reach
equilibrium until the third step. That’s still pretty fast, so we could change the
INCREMENTAL_REACTIONS line back to false (the default) and give up or change
the amount again to a smaller value to make the change slower.
26
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
This brings up an important point about PHREEQC input files. If something is not
explicitly written, the program assumes that function is set to default. The
INCREMENTAL_REACTIONS command in is only meaningful when the command is
set to not be the default term (false). Another words, you can either remove the
INCREMENTAL_REACTIONS line or set it to false for the same effect. Sometimes if
your input file is not doing what you expected, you may have a default issue. All the
commands and sub-commands are defined in the PHREEQC manual under the
Description of Data Input chapter in the Keywords section. You can open the manual
from your input window under the help menu. It should look like the picture below. The
PDF manual is an essential resource for using this program. All the keywords are
described, the default conditions listed and examples of their use given.
At this point you can run the simulation and plot the output. The plot should look
something like the one below.
27
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Calcite and pure water
2
si_Calcite
Calcite log SI
0
-1
-2
9 10 11
pH
Next we will use an alternative formulation of the REACTION command to add the
reactant to the system in smaller steps (input file 1f).
TITLE pH example
SOLUTION 1 pure water & calcite
pH 7
temp 25
REACTION
Calcite
1.0 moles in 50 steps
SELECTED_OUTPUT
-file reaction1out.csv
-molalities Ca+2 HCO3- CO3-2
END
Finally we can add both calcite and CO2 gas to the system at the same time (input file
1g).
TITLE pH example
SOLUTION 1 pure water & calcite & CO2 gas
pH 7
temp 25
REACTION
Calcite
CO2(g)
1.0 moles in 50 steps
SELECTED_OUTPUT
-file reaction2out.csv
-molalities Ca+2 HCO3- CO3-2
END
The interactions between water (H2O) and dissolved calcite produced several other
aqueous forms of Ca and CO3 (species). The CO32- is related to HCO3- and H2CO3 by a
series of equilibrium equations. The total of dissolved carbonate species is referred to as
“carbonate alkalinity”. A complete description of the system requires us to include
additional chemical equations:
CO2(g) ↔ CO2(aq)
H2CO3 ↔ H+ + HCO3-
HCO3- ↔ H+ + CO32-
The equilibrium constants for these reactions are usually labeled KCO2, K1, and K2,
respectively. The reaction of CO2 (aq) to form H2CO3 is usually combined with the
CO2(g) to form CO2(aq) reaction and the sum of the reactions has the constant KCO2.
These equations, together with the dissolution of calcite equation if calcite is present, can
be used to describe the carbonate system.
Alkalinity
It is worthwhile to briefly discuss the concepts of alkalinity and pH at this point. The
alkalinity value reported for water samples is determined from a titration of the water
sample with H2SO4, and is reported as meq/l or eq/l CO3, meq/l or eq/l HCO3 or meq/l or
eq/l CaCO3. Alkalinity is actually equal to total HCO3- +CO32- + any other weak acids
that are negatively charged at natural water pH values of 7 to 10 (usually boron and
organic acids). This value is calculated by measuring the number of equivalents of acid
needed to change the water sample (of known volume) from its starting pH to a pH of 4.
So eq/l of H+ used = eq/l of alkalinity in the water. Sometimes this data is further
converted to eq/l or mg/l of HCO3 and /or CO3 using the pH of the water and assuming
that the ratio of HCO3/CO3 is a function of pH (10-10.3 = CO32- * H+/HCO3-).
29
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
SOLUTION 1 Simplified Mt. Edna Solicchiata
units mg/l
pH 9.90
pe -0.2
density 1.00
temp 17.2
C(4) 1180 as HCO3
Ca 28.1
Cl 238 charge
Fe 0.150
K 30.1
Mg 214
Mn 0.0277
Na 250
S(6) 20
Si 79.4 as SiO2
Sr 0.673
REACTION
H2SO4
0.02 moles in 50 steps
SELECTED_OUTPUT
-file titration1.csv
-reaction true
END
Alkalinity Titration
10
pH
8.4
6.8
pH
5.2
3.6
2
0 0.004 0.008 0.012 0.016 0.02
moles H2SO4 added
30
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Fixed pH
The EQUILIBRIUM_PHASES command below add the new “phase” to system where it
will reach a saturation index for H+ of 10-4.5 using HCl to achieve that saturation index by
adding up to 10 moles of HCl. This is a somewhat roundabout way to achieve the desired
result. Other programs such as MINTEQ and GWB allow the user to specify fixed pH
directly.
SOLUTION
the solution composition follows
EQUILIBRIUM_PHASES 1
Fix_H+ -4.5 HCl 10.0
END
The pH Sweep
In order to try out the pH sweep concept let’s make a plot of inorgainc carbonate
speciation with pH. This is also known as the Bjerrum plot. You can start with the input
file below (input file pH sweep).
TITLE pH Sweep example
SURFACE_SPECIES
SOLUTION 1
-units mmol/kgw
pH 8.0
Zn 0.0001
Na 1500 charge
Cl 2500
C(4) 250 as HCO3
#
# Model definitions
#
PHASES
Fix_H+
H+ = H+
log_k 0.0
END
SELECTED_OUTPUT
-file pHsweep
-molalities Zn+2 HCO3- CO3-2
31
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -5.0 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -5.25 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -5.5 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -5.75 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -6.00 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -6.25 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -6.50 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -6.75 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -7.00 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -7.25 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -7.50 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ -7.75 NaOH 10.0
END
USE solution 1
EQUILIBRIUM_PHASES 1
Fix_H+ 8.00 NaOH 10.0
END
32
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
After running the file make the plot below. This is going to take a few tries so take your
time and ask questions of those instructors. Hint: check the speciation of the inorganic
carbon species.
0.00070001
0.00060001
0.00050001
0.00040001
0.00030001
0.00020001
0.00010001
9.001E-5
8.001E-5
7.001E-5
6.001E-5
5.001E-5
4.001E-5
3.001E-5
2.001E-5
1.001E-5
9.01E-6
8.01E-6
7.01E-6
6.01E-6
5.01E-6
4.01E-6
3.01E-6
2.01E-6
1.01E-6
9.1E-7
8.1E-7
7.1E-7
6.1E-7
5.1E-7
4.1E-7
3.1E-7
2.1E-7
1.1E-7
1E-7
9E-8
8E-8
7E-8
6E-8
5E-8
4E-8
3E-8
2E-8
1E-8
9E-9
8E-9
7E-9
6E-9
5E-9
4E-9
3E-9
2E-9
1E-9
4 5 6 7 8 9 10 11 12
pH
Titration of AMD
This example covers the simulation of a standard cold titration for acidity when the
solution has low initial pH. An important feature to note is that a solution with a pH value
below 5 will not have little or no alkalinity as determined by the standard titration. That
is because the standard alkalinity titration measures HCO 3+CO3 and at pH 5 there is
almost no HCO3. That does not mean there is no inorganic carbon, but that the form
(species) is predominantly H2CO3 as we saw above. In the case of AMD or other acidic
solutions we are trying to measure acidity. The titration is usually done with a standard
basic solution such as 5N NaOH. The following input file will simulate the titration of an
acidic solution (input file AMD titration 1):
33
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Ba 0.054
Ca 5.16
Cl 0.994
Cu 0.041
F 0.142
Fe 23.19
K 0.624
Mg 1.227
Mn(2) 0.0098
Na 0.539
S(6) 82.65 charge
Zn 0.501
REACTION
NaOH
0.001 mole in 50 steps
SELECTED_OUTPUT
-file AMDtitration.csv
-reaction true
END
You can plot the change in pH with addition of the NaOH solution. There are many
options you can add to this simulation such as opening the system to the atmosphere,
letting mineral surfaces precipitate that will sorb solutes from solution, etc.
AMD Titration
10
8.6
7.2
pH
5.8
4.4
3
0 0.0002 0.0004 0.0006 0.0008 0.001
moles NaOH added
pH
34
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Introduction to Kinetics
All the batch reactions can be given a kinetic rate. This feature allows a forward model
that has an outcome depend on time. Kinetic models require two commands, KINETICS
and the linked RATE commands. The following example PHREEQC input file (Kin1)
shows the kinetically controlled dissolution of calcite. In this case we approach
saturation with respect to calcite in about ten days.
35
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
The KINETICS command specifies what reaction will be time dependent, the amount of
reactant, the total time and the number of steps that the model takes during that total time.
The m is the current number of moles of reactant, m0 is the initial number of moles. M is
usually equal to m0 by default. The steps subcommand refers to the time (in seconds)
over which the rate is integrated. In this example the first integration takes place over
10,000 seconds, then as an independent reaction, the next step takes place over 25,000
seconds. You can choose to make the time steps incremental by using the
INCREMENTAL_REACTION keyword. In the example that would make the reaction
run for 10,000 seconds and produce results and then more results after 25,000 seconds
and so on. The time step subcommand means that the first step of 10,000 seconds will
be divided into 100 smaller sections (100 seconds each) and the results integrated to
produce the result at 10,000 seconds.
The RATE command is the BASIC code block that enables the kinetic simulation. This
command block is taken from the examples at in the standard database. The example
BASIC rate command blocks can be found at the end of the database files. While this is
somewhat clunky, it does have the advantage of being almost infinitely flexible.
The kinetic values for the dissolution and precipitation reactions of common minerals are
the subject of extensive literature. In general, most experimental work produces rates
that are much faster than field-derived rates by a factor of as much as two orders of
magnitude. Therefore, the reality is that the kinetic rate is a major source of uncertainty
in modeling. A more complete discussion with some relevant examples will be covered
later in the manual.
Table 3. Rates of dissolution , surface area, effective surface area and surface area
modified values used in kinetic modeling. Effective surface area is relative measure.
36
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
All the reactions we have considered have assumed equilibrium. As discussed earlier,
many reactions rates are fast enough that this assumption is sufficient. However, some
near surface processes involve physical transport rates that are as fast, or faster than the
chemical reactions. When we reach this condition we need to describe the chemical
processes using kinetics, that is give the reaction a rate and fix the time period of the
simulation. PHREQC can incorporate several types of kinetic reactions including
dissolution/precipitation of minerals, redox reactions and microbial metabolism. You
access the kinetic functions through the KINETICS and RATE commands.
The kinetic option is also able to model multiple simultaneous reactions. Thus far we
have been expressing the rates using the built-in rate expressions. The most typical
formulation is written as,
r = As k+ [1-Q/K]
In this case, r is the reaction rate (mol/sec), As is the surface area in cm2, k+ is the rate
constant (mol/cm2 sec), Q is activity product and K is the equilibrium constant. This
formulation will work for dissolution or precipitation with rate linked to saturation state
(Q/K). This formulation is not temperature dependent. Alternatively, the user can supply
the activation energy and pre-exponential factor to create a temperature-dependent rate
based on the Arrhenius formulation.
Even this treatment of pyrite kinetics can be very simplistic. In fact, the reaction is pH
dependent as are rates for other common minerals such as calcite (see figure below).
Measured rates for most mineral reactions have reported rates that vary 10-50-fold
depending on a number of variables such as solution composition, pH and grain size. In
addition, mineral reaction rates measured in the field are usually several orders of
magnitude slower than laboratory rates. The figure below shows some experimental data
that can be used to formulate pH dependent rate expressions.
37
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Figure 9.1 Pyrite rate experiment figure modified from Singer and Stuum (1970), Calcite
rate experiment figure modified from Plummer and others (1978) showing the rate
dependence on pH.
More complex treatments for calcite dissolution include using rates fitted to the empirical
expression:
r = k (A/V) (1 – Q/K)n
where A is surface area, V is solution volume, and n and k are coefficients dependent on
solution composition. The coefficients are obtained by fitting experimental data.
Mechanistic models such as those of Plummer and others (1978) are more complex.
Their model used the expression:
where the first three terms are the forward rate and the last is the backward rate to explain
the experimental data with values (Appelo and Postma, 1993). The reaction is also
temperature dependent. Evaluation of the expression has shown that the first term is
dominant at pH values below 5. If we were interested in the rates of acidic solutions
only, we might implement the more complex formulation to more accurately model the
process.
Using the input file below (input file pyr_kin 1Y), the rate of reaction is a function of pH,
O2, etc. You can run this file and examine the grid file created. The file has a starting pH
of 7.3. Based on what we have seen about pH dependency, we expect little change over
the time. Alternatively, you can change the starting pH to 3.3. Running the acidic
simulation will produce essentially no change over the same time period. This brings up
38
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
the whole problem with kinetic values determined in laboratory conditions. In the case of
pyrite oxidation, the rapid process that is observed in the field is controlled by microbial
reactions that increase the rate of reaction several orders of magnitude. You can test the
sensitivity of rate by changing the value in line 20 from -10.19 to -6.19, a 10,000 fold
increase. Now there is a difference between the acidic and alkaline simulations, however
since the microbes in question are acidophiles (acid-loving), how can you justify using
the pH dependent rate law if the higher rate is only applicable in acidic environments.
This leads to yet more complex rate laws where rates are dependent on external variables
such as pH that are themselves a function of the reactions we want to model.
One approach to dealing with this is to examine the system to be modeled to determine if
the kinetic approach is necessary. For instance, we can need to decide if 1) the rate of
flow through a system is so fast that little reaction can occur. This rate of reaction versus
rate of transport can be quantified using the Damköhler and Peclet numbers. The
Damköhler number is the ratio of the rate of reaction versus the rate of transport and is
defined as:
Da = x S anH+ L/ ceq
And the Peclet number is a measure of the importance of advection versus dispersion +
diffusive transport and is defined as:
Pe = L/ Deff + D
39
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
40
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
TITLE reaction of pyrite
SOLUTION 1
units ppm
pe 5
temp 10
pH 7.3
S(6) 50 as SO4
Cl 450 charge
N(5) 100 as NO3
Fe 6
Zn 10 ppb
C(4) 100 as HCO3
Na 500
K 4
Mg 50
Ca 400
EQUILIBRIUM_PHASES
O2(g) -0.7
KINETICS 1
Pyrite
-tol 1e-8
-m0 0.01
-m 0.01
-parms -5.0 0.1 .5 -0.11
-steps 31536000 in 100 steps # 1 year
-step_divide 100000
INCREMENTAL_REACTIONS true
RATES
Pyrite
-start
1 rem parm(1) = log10(A/V, 1/dm) parm(2) = exp for (m/m0)
2 rem parm(3) = exp for O2 parm(4) = exp for H+
41
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Activity and Concentration
Now it is time to take up the concept of activity. In the next exercise will modify the
system by adding other solutes to the solution (input file 2). In this case we will add salt
(NaCl) as the mineral halite by using the same formulation as we did for calcite.
Look at the output fie. The addition of salt to the system has increased the apparent
solubility of calcite (there is more Ca and total carbonate in solution than the first
example), but calcite is still exactly saturated (SI = 0). The result in the model can be
duplicated in the lab. To allow for the effect of other ions in a solution in equilibrium
with calcite, we employ a concept called activity. The activity is defined as the
chemically reactive concentration and is related to the mass concentration by the activity
coefficient (see equation below).
[Ca2+] γ = aCa2+
Activity coefficients are usually less than one and get smaller with increasing ionic
strength (total ions in solution). The activity coefficient (γ) is dependent on the ionic
strength of a solution, where the ionic strength (IS) is the sum of all the charges in that
solution. There are several different equations that relate the activity coefficient, and thus
the activity of each species, to ionic strength such as Davies, Debye-Huckel, etc. Several
are available in PHREEQC. These formulations are relatively accurate until the ionic
strength of the solution exceeds about 100 molar (seawater is 10-0.22 M). At higher ionic
strengths, the measured solubility of many minerals can only be explained if the activity
coefficient is greater than 1. Higher ionic strength solutions are usually modeled using a
different activity concept such as the Pitzer method. PHREEQC has a Pitzer option.
Solid phases are generally given an activity of one (by definition) and gas concentrations
are expressed as fugacity, which are equal to the partial pressure of the gas. For example,
oxygen gas (O2) is 20% of the total atmosphere, thus PO2 = 0.2 or 10-0.7. The activity of
water is usually close to one, but there are exceptions.
The equilibrium constant equation for calcite dissolution using the full thermodynamic
formalism is:
43
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
factors. PHREEQC has a solid solution option that allows us to approximate the effect of
including one additional element in a pure mineral (e. g. Zn in FeS2, pyrite).
Databases
The PHREEQC program solves a set of simultaneous non-linear equations. These
equations are in the form of the mass action reactions. The set of equations available for
the program to use is compiled in the database (files under the calculation menu). If the
element that you want to model is not in the database, the program cannot model that
reaction (input file 4).
The equilibrium constants in the database for specific reactions are a source of potential
error in any computer model. The constants for a reaction are determined by laboratory
experiments or by calculation. The PHREEQC program allows the user to choose from
four databases when using the activity concept or a separate database for the Pitzer
activity model (used in high salinity problems). The four databases normally used are
PHREEQC, WATEQ4f, LLNL and MINTEQ. The database can be modified by
changing the constant value for a reaction, or by adding a new element and the reactions
44
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
for that element. The database file can be edited directly with a text editor. However,
PHREEQC has the useful ability to accept database input as part of the input file. This
way the syntax and the effect of the inclusion of new components can be evaluated
without changing the database. An example is seen in the example PHREEQC file Ex1,
which adds nitrate and ammonia, uranium and its dissolved species and the solid phase
uraninite (which are missing from the database) to the calculation in the input file.
45
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
U+4 + 5 H2O = U(OH)5- + 5 H+
log_k -13.147
delta_h 27.580 kcal
#secondary master species for U(5)
U+4 + 2 H2O = UO2+ + 4 H+ + e-
log_k -6.432
delta_h 31.130 kcal
#secondary master species for U(6)
U+4 + 2 H2O = UO2+2 + 4 H+ + 2 e-
log_k -9.217
delta_h 34.430 kcal
UO2+2 + H2O = UO2OH+ + H+
log_k -5.782
delta_h 11.015 kcal
2UO2+2 + 2H2O = (UO2)2(OH)2+2 + 2H+
log_k -5.626
delta_h -36.04 kcal
3UO2+2 + 5H2O = (UO2)3(OH)5+ + 5H+
log_k -15.641
delta_h -44.27 kcal
UO2+2 + CO3-2 = UO2CO3
log_k 10.064
delta_h 0.84 kcal
UO2+2 + 2CO3-2 = UO2(CO3)2-2
log_k 16.977
delta_h 3.48 kcal
UO2+2 + 3CO3-2 = UO2(CO3)3-4
log_k 21.397
delta_h -8.78 kcal
PHASES
Uraninite
UO2 + 4 H+ = U+4 + 2 H2O
log_k -3.490
delta_h -18.630 kcal
END
Using the rules specified in the PHREEQC manual, you can add any compound or
element. For instance, PHREEQC databases do not presently include NAPL’s such as
benzene. The following input will include benzene as a dissolved species (input file 4a).
You need to look up or calculate and include the correct log k for the dissolution reaction
based on the reported solubility.
SOLUTION 1 ground water with benzene
units ppm
pH 8.1
density 1.00
temp 25.0
As 0.010
Cu 0.010
Pb 0.0002
Zn 0.050
Fe 0.010
Na 22.92
C(4) 67.0 as HCO3
Cl 3.6
46
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
S(6) 40 charge
Ca 20.0
Mg 8.4
F 2.8
SOLUTION_MASTER_SPECIES
Benzene Benzene 0.0 78.0 78.0
SOLUTION_SPECIES
Benzene = Benzene
log_k 0
delta_h 0 kcal
PHASES
Benzene
Benzene = Benzene
log_k -1.64
delta_h 0 kcal
EQUILIBRIUM_PHASES
CO2(g) -3.50
O2(g) -2.0
Calcite 0.0 1.0
Benzene 0.0 1.0
END
You should notice that the benzene specified in this modified database is an “element” in
the sense that it is a basis species (SOLUTION_MASTER_SPECIES). It is not C 6H6.
This formulation is decoupled carbon; the benzene does not have any relationship to the
inorganic carbon or hydrogen in the model. This is the standard method to add organic
compounds into PHREEEQC. The other option is to explicitly couple the benzene by
giving it a chemical formula of C6H6. This is a non-trivial exercise, but can be important
if you are trying to calculate the effect of benzene degradation on alkalinity.
In the case of mining activities and related environmental problems the major limitations
of the standard PHREEQC database and most of the other databases are lacking in the
species of interest. In this case you can use the MINTEQ databases or add to your
database of choice in order to create a custom database.
For instance, you want to model a problem that involves cyanide. Try and run the input
file below using the default database to investigate the speciation of a metal-rich solution
in the presence of cyanide (CN-) (input file AMD solution). You can see that in the file
below there is no cyanide so you will have to add it to the input file. The first thing you
should do is see if the database includes cyanide either as a separate “element” or
explicitly as CN-. If you cannot find cyanide in the default database then check other
databases. Remember you can change databases using the Files command under the
Calculations menu. In fact, the only database that contains cyanide is the MINTEQ
database. In this case the cyanide is listed as an “element”, in effect it’s own basis
species. This means that all reactions between cyanide and other elements must have a
reaction in the database that has the formulation:
Ag+ + cyanide- = Agcyanide
47
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Or obeying the rules for writing reactions for the standard PHREEQC compatible
database:
AgCyanate
AgCyanate = Cyanate- + Ag+
log_k -6.6159
delta_h 13.175 kcal
This reaction has a log K (equilibrium constant) and a temperature dependency (delta_h)
term. Note it has to be charge balanced. If you look in the first set of datablocks in the
database you will find the basis species and see that the element silver is defined as Ag+.
This means that any reaction that includes silver has to be written using Ag + since the
program forms a computational matrix of basis species to solve the equilibrium condition
for the specified system. This is relevant when we are taking data for a reaction of
interest from one database and putting it into our custom database.
For instance, try and run the file below after adding 10 ppm cyanide using the default
database. You will get the message:
WARNING: Could not find element in database, Ag. Concentration is set to zero.
WARNING: Could not find element in database, As. Concentration is set to zero.
WARNING: Could not find element in database, Cyanide. Concentration is set to zero.
Now switch to the MINTEQ database and run the simulation. Now the program can’t
find Fe(OH)3, aka ferrihydrite. This problem is overcome by changing the name in the
SELECTED_OUTPUT block to ferrihydrite. Checking the output file you see that you
now have the speciation of cyanide and that the predominant species are hydrogen and
silver cyanide. However, when you show this triumph to the boss, he says “I wanted to
know about Moly as well”. Checking the database we see there is no Mo in the database.
Checking the other databases you find Mo is included in the llnl database, but that the
basis species for the electron is in terms of O2 not the e- that is in your custom database.
This means that the reaction you want:
The log K for the reaction from the llnl database is not the correct value for the reaction
you want although both describe the dissolution of the metal. This illustrates the
difficultly in taking data from one database and putting it into another. The solution is to
calculate the K for the reaction you want from the thermodynamic expressions as
described earlier.
48
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
TITLE example of AMD solution
SOLUTION 1
units ppm
pH 3.499
density 1.00
temp 25.0
Al 1.237
Ag 2.56
As 3.25
C(4) 5.87 as HCO3
Ba 0.054
Ca 5.16
Cl 0.994
Cu 0.041
F 0.142
Fe 23.19
K 0.624
Mg 1.227
Mn(2) 0.0098
Na 0.539
N(5) 0.02
Pb 0.0012
S(6) 82.65 charge
Sr 0.052
Zn 0.501
SELECTED_OUTPUT
-file cyanide.csv
-molalities Fe+2 Fe+3
-saturation_indices Fe(OH)3(a)
END
The example file (input file 4b) contains data from over 145 samples that were originally
in worksheet format and then saved as a text file. Coupled with the
SOLUTION_SPREAD command, users can rapidly input large sample sets and create
comma-separated output files of specified results for plotting and analysis.
49
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Output from PHREEQC calculations can also be modified using the USER_PUNCH and
USER_PRINT commands. The USER_PUNCH command allows further modification
of any value written to the SELECTED_OUPUT file using BASIC code. For instance,
molalities for solutes (e.g. Ca+2) can be converted to ppm and written to the output file, or
printed using the USER_PRINT command.
The following file (input file 4c) shows an example of changing the parameters written to
the selected output file. This is the same file that we used before (number 1g), but now
the selected output file will be different. Run both files and examine the grid files
produced to see that some parameters such as simulation, solution, etc., are no longer
written to the file, while temperature is written.
TITLE selected output example
SOLUTION 1 pure water & calcite & CO2 gas (>atmospheric value)
pH 7
temp 25
REACTION
Calcite
CO2(g)
1.0 moles in 50 steps
SELECTED_OUTPUT
file selout.csv
reset true
simulation false
solution false
temperature true
time false
reaction false
END
50
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Introduction to Redox
Basic Principles
The example with O2 gas in the system caused a change in the pe value, which is a
measure of the system redox state. The word redox refers to reduction-oxidation
reactions. These are reactions that involve the transfer of electrons. One common
example is:
Fe2+ ↔ Fe3+ + e-
The formalism is the same, an equilibrium constant:
K = aFe3+ ae-
aFe2+
The thermodynamic formalism for the elements that can have more than one valence
state under surface conditions (Fe, Mn, Cu, V, Se, As, etc.) is slightly different; using the
Nerst equation:
Eh = E0 + RT ln K
nF
where E0 is the standard potential of the reaction (found in the same tables as ΔG0
values). F is Faraday’s constant and n is the number of electrons in the reaction. Eh is
the oxidation potential, which can be measured with an electrode. Thus, the oxidation
/reduction potential (ORP) or Eh measurement has units of millivolts or volts. Positive
values are termed oxidizing and negative are reducing. Negative values indicate an
abundance of electrons, which can react to reduce oxidized compounds such as hematite
(Fe2O3).
Fe2O3 + 6H+ + 2e- ↔ 2Fe2+ + 3H2O
And
K = a(Fe2+)2 a(H2O)3
a(H+)6 a(e-)2 aFe2O3
a(e-)2 = a(Fe2+)2
K
where the activity of the electron ( or pe) is a function of the activity (concentration) of
Fe2+ in equilibrium with hematite (Fe2O3). Analogous reactions can be written for other
redox-sensitive elements. In PHREEQC, we use the pe value to express the Eh of a
solution. PHREEQC converts a pe value to an Eh value using the formula:
Eh = 2.303RT pe
F
51
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Generally, the most significant components in the control of redox are (the presence or
absence) oxygen and organic matter. Oxygen is the major electron acceptor in most
natural systems, that is dissolved oxygen gas (valence state of 0) is reduced by addition of
electrons, while another dissolved species is oxidized by giving up those electrons (e.g.
ferrous iron Fe2+ going to ferric iron, Fe3+ and precipitating as Fe(OH)3). The most
common manifestation of this coupled redox reaction is probably the oxidation of organic
matter to carbon dioxide where organic carbon (valence of 0) is oxidized to CO 2 (valence
of +4 for carbon) during aerobic decay. The presence of dissolved oxygen produces
positive valued of pe or an oxidizing environment, while environments that can exclude
or consume oxygen fasterthan it can be replaced have reducing conditions. The figure
shows common terrestrial environments in pe-pH space.
Therefore the redox potential of a natural system is often a result of several competing
reactions (mixed potential) that are not in equilibrium with each other and are trying to
reach equilibrium at different rates. This means the Eh measured with a platinum
52
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
electrode will not reflect a single redox pair’s equilibrium, but the sum of all the redox
reactions in solution. The figure below shows measured Eh versus that calculated from
the relative abundance of various redox pairs that are present in solution. Note the
measured Eh is usually not equal to the value calculated from a single pair.
PHREEQC allows the user to set the pe value and thus, control the speciation of the
redox-sensitive elements (RSE). Another option allows the speciation of RSE to be set
by using known values for a redox pair such as nitrate and ammonia that we assume is
dominant (controlling the other redox pairs). In most natural systems, the dominant
redox element is oxygen. The strength of the oxygen reaction is much greater than other
reactions, and until oxygen is removed from the system, other redox reactions are not
very effective.
The redox state of the system is particularly important in the behavior of metals. Many
metals have more than one oxidation state (e.g. iron, manganese, arsenic, selenium, Zinc,
etc.) and have very different solubility for the different oxidation states. A general rule is
that the reduced from is more soluble and thus, more mobile. Examples other than iron
include Mn, Pb, Zn, Cu, Co. However, sometimes the more oxidized form is more
soluble (mobile). Examples of elements that are more mobile in the oxidized form
include U, Mo, Se and As.
Iron redox behavior is also important because iron forms solid precipitates such as
Fe(OH)3, often called ferrihydrite or amorphous ferric hydroxide, when a reduced
53
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
solution is oxidized (usually by contact with the atmosphere). Ferrihydrite is a good
adsorption surface and will remove many metals from solution. Ferrihydrite is gradually
converted by de-hydration to limonite, goethite and finally, hematite.
Acid mine drainage (AMD) is the result of surface weathering of sulfide mineral
deposits. The sulfur in sulfide minerals such as chalcopyrite, sphalerite, galena, cinnabar,
pyrite etc. is reduced (-1), and can be oxidized to sulfate (+6) by atmospheric oxygen.
Reaction of sulfides, primarily pyrite, will generate an acidic solution:
The oxidation reaction of iron is slow enough (1000 days) and ferrous iron soluble
enough that the iron may be fairly mobile. There are additional reactions involved such
as:
FeS2 + 14Fe3+ + 8H2O ↔ 15Fe2+ + 2SO42- + 16H+
which is fast or biologically-mediated oxidation of iron that may be even faster. AMD
produces iron oxide/hydroxide deposits (rust coatings) and jarosite (MFe 3(OH)6(SO4)2,
where M = K, Na, NH4, Ag and Pb), which is a yellow to brown colored mineral.
Naturally occurring examples are found near hot springs and fumaroles.
One consequence of commercial value is the supergene enrichment where the local acidic
environment of pyrite oxidation can mobilize associated chalcophile elements such as Cu
and Ag. The metals in solution will be carried downward by rain toward the water table
and re-precipitate in un-weathered sulfide ore by replacement of Fe in pyrite with Cu and
Ag. The results is a redistribution of the Fe-Cu-Pb-Zn sulfide into a Fe-rich gossan cap
and with Cu, Ag, Pb and Zn concentrated under the gossan cap below the water table (Pb
and Zn form insoluble carbonate and sulfate salts).
In order to test the effect of oxygen on redox, we will use a reduced fluid, acid mine
drainage and react the fluid with oxygen gas. This exercise will simulate adding O 2 gas
to acid mine drainage (AMD), (input file 3). We do not specify an initial pe value for the
solution, so PHREEQC uses the default value of 4.
54
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
TITLE example of oxidation of AMD solution
SOLUTION 1
units ppm
pH 3.499
density 1.00
temp 25.0
Al 1.237
As 3.25
C(4) 5.87 as HCO3
Ba 0.054
Ca 5.16
Cl 0.994
Cu 0.041
F 0.142
Fe 23.19
K 0.624
Mg 1.227
Mn(2) 0.0098
Na 0.539
N(5) 0.02
Pb 0.0012
S(6) 82.65 charge
Sr 0.052
Zn 0.501
REACTION
O2(g)
0.001 mole in 50 steps
SELECTED_OUTPUT
-file oxexample.csv
-molalities Fe+2 Fe+3
-saturation_indices O2(g) Fe(OH)3(a)
END
Inclusion of organic matter into the system can be accomplished by adding organic
matter as a phase that can be used in inverse modeling reactions. In this approach
organic matter is represented by the simple formula CH2O. Note this reaction
formulation explicitly links organic matter degradation to production of protons and
electrons.
PHASES
CH2O
CH2O + H2O = CO2 + 4H+ + 4e-
log_k 0.0
Using this formulation we can evaluate the control of solution redox by organic matter.
In the formulation above, the organic C has a valance state of 0, while the carbon valance
in CO2 is +4. In this case the equilibrium constant of the reaction is set to 1 (log 1 = 0).
The input file (org deg example) specifies that the solution is first speciated, then two
phases, CO2 gas and organic matter are added to the system and equilibrated. We can
examine the effect of redox by noting the pe of the first solution, then the pe after the
reaction step or the distribution of the dissolved redox sensitive elements such as Fe or S
55
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
in both solutions and the effect on the saturation indices of minerals such as Fe(OH) 3 (a),
FeS (ppt) and pyrite.
Specifying Redox
Note that PHREEQC has the ability to calculate the Eh (pe) value if the concentrations
for both members of a redox pair such as Fe3+/Fe2+ are specified (using the redox
subcommand, see example below). The calculated pe value can be used to speciate any
or all the other redox sensitive elements. This can become very complex as seen in Ex1
from the input files supplied with PHREEQC.
56
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
log_k 119.077
delta_h -187.055 kcal
gamma 2.5000 0.0000
SOLUTION 1 SEAWATER FROM NORDSTROM ET AL. (1979)
units ppm
pH 8.22
pe 8.451
density 1.023
temp 25.0
redox O(0)/O(-2)
Ca 412.3
Mg 1291.8
Na 10768.0
K 399.1
Fe 0.002
Mn 0.0002 pe
Si 4.28
Cl 19353.0
Alkalinity 141.682 as HCO3
S(6) 2712.0
N(5) 0.29 as NO3
N(-3) 0.03 as NH4
U 3.3 ppb N(5)/N(-3)
O(0) 1.0 O2(g) -0.7
SOLUTION_MASTER_SPECIES
U U+4 0.0 238.0290 238.0290
U(4) U+4 0.0 238.0290
U(5) UO2+ 0.0 238.0290
U(6) UO2+2 0.0 238.0290
SOLUTION_SPECIES
#primary master species for U
#secondary master species for U+4
U+4 = U+4
log_k 0.0
U+4 + 4 H2O = U(OH)4 + 4 H+
log_k -8.538
delta_h 24.760 kcal
U+4 + 5 H2O = U(OH)5- + 5 H+
log_k -13.147
delta_h 27.580 kcal
#secondary master species for U(5)
U+4 + 2 H2O = UO2+ + 4 H+ + e-
log_k -6.432
delta_h 31.130 kcal
#secondary master species for U(6)
U+4 + 2 H2O = UO2+2 + 4 H+ + 2 e-
log_k -9.217
delta_h 34.430 kcal
UO2+2 + H2O = UO2OH+ + H+
log_k -5.782
delta_h 11.015 kcal
2UO2+2 + 2H2O = (UO2)2(OH)2+2 + 2H+
log_k -5.626
delta_h -36.04 kcal
3UO2+2 + 5H2O = (UO2)3(OH)5+ + 5H+
log_k -15.641
57
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
delta_h -44.27 kcal
UO2+2 + CO3-2 = UO2CO3
log_k 10.064
delta_h 0.84 kcal
UO2+2 + 2CO3-2 = UO2(CO3)2-2
log_k 16.977
delta_h 3.48 kcal
UO2+2 + 3CO3-2 = UO2(CO3)3-4
log_k 21.397
delta_h -8.78 kcal
PHASES
Uraninite
UO2 + 4 H+ = U+4 + 2 H2O
log_k -3.490
delta_h -18.630 kcal
END
In the input file, the pe value is specified in the first portion of the input. This input pe is
the default entry position. If no pe value is specified in this entry, then a default value of
4 is applied. Using the redox specifier overrides the input pe. The specified redox value
will now be used to distribute the other redox species (e.g. Fe). In the example the redox
is set to the dissolved oxygen/water couple. The concentration of dissolved oxygen can
be used to calculate a pe value based on the equation discussed above. This value is
further overridden by placing pe after the Mn entry. This will distribute the Mn species
based on the initial pe value (8.451) rather than the value calculated from the dissolved
oxygen concentration. Finally, the uranium redox species will be distributed based on the
redox value calculated from nitrate/ammonia ratio.
There are cases where the best treatment for alteration of water quality is to allow
reactions between the water to be treated and natural components. One such case is the
passive treatment strategy where acidic mine water is allowed to react with an aquifer or
wetland containing organic matter, pyrite and carbonate. These are typical mineral found
in a wetland where oxygen is generally consumed near the atmospheric-water interface.
In this case we have made the judgment that the rates of reaction versus transport are
similar and that there will be time-dependent reactions during transit. Therefore a kinetic
model is appropriate. This judgment can be based on a strict quantitative analysis using
the Damköhler and Peclet numbers or can be based on our experience. That is we know
from observation that there is some degree of treatment in a wetland, but we may wish to
develop a more rigorous estimate for regulatory review.
We expect that the oxidation of organic matter by microbes will lower the oxygen content
of the water and cause the oxidizing solution to become reducing. This will change the
redox state of the metals and probably cause precipitation, hence treatment. The reaction
with the carbonate (calcite) will change pH, which will also cause metals to precipitate as
hydroxides. We are making assumptions about the flow rates by using 1 kg of solution,
0.01 moles of calcite, and 1 mole of pyrite and organic matter. In effect we are saying
58
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
that 1 liter of water flows through a matrix in 600 days composed of equal parts organic
matter and pyrite with trace calcite. These types of constraints are usually based on
knowledge of aquifer composition and porosity, flow rates calculated from Darcy’s Law,
and composition of the fluid. The input file (input file passive treat) is an example of a
complicated kinetic simulation will take some time to run on the average computer (a few
minutes).
We can take some time to examine the output and modify the simulation to investigate
the factors that control the process.
TITLE Degradation of organic matter and reduction of redox sensitive elements (600 days)
SOLUTION 1 Leaching solution
units ppm
pe 4 O2(g) -0.7
temp 10
pH 2.3
S(6) 5000 as SO4 charge
Cl 450
F 1
N(5) 100 as NO3
U 40
Fe 600
Zn 100
As 2
Mn 20
Pb 0.2
Ni 5
Cu 3
C 1 # added for calcite since there is no carbon otherwise
Cd 1
Li 0.1
Na 500
K 4
Mg 50
Ca 400
Al 200
Si 50
SAVE solution 1
END
USE solution 1
KINETICS 1
Calcite
-tol 1e-8
-m0 1e-2
-m 1e-2
-parms 50 0.6
Organic_C
-formula CH2O
-tol 1e-8
-m0 1 # mol/kgw
-m 1
-steps 51840000 in 100 steps # 600 days
-step_divide 1000000
59
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Pyrite
-tol 1e-8
-m0 1
-m 1
-parms -5.0 0.1 .5 -0.11
RATES
Calcite
-start
1 rem parm(1) = A/V, 1/dm parm(2) = exponent for m/m0
5 if time > 8640000 then goto 200
10 si_cc = si("Calcite")
20 if (m <= 0 and si_cc < 0) then goto 200
30 k1 = 10^(0.198 - 444.0 / (273.16 + tc) )
40 k2 = 10^(2.84 - 2177.0 / (273.16 + tc) )
50 if tc <= 25 then k3 = 10^(-5.86 - 317.0 / (273.16 + tc) )
60 if tc > 25 then k3 = 10^(-1.1 - 1737.0 / (273.16 + tc) )
70 t = 1
80 if m0 > 0 then t = m/m0
90 if t = 0 then t = 1
100 moles = parm(1) * 0.1 * (t)^parm(2)
110 moles = moles * (k1 * act("H+") + k2 * act("CO2") + k3 * act("H2O"))
120 moles = moles * (1 - 10^(2/3*si_cc))
130 moles = moles * time
140 if (moles > m) then moles = m
150 if (moles >= 0) then goto 200
160 temp = tot("Ca")
170 mc = tot("C(4)")
180 if mc < temp then temp = mc
190 if -moles > temp then moles = -temp
200 save moles
-end
Organic_C
-start
10 if (m <= 0) then goto 200
20 mO2 = mol("O2")
30 mNO3 = tot("N(5)")
40 mSO4 = tot("S(6)")
# 50 rate = 1.57e-9*mO2/(2.94e-4 + mO2) + 1.67e-11*mNO3/(1.55e-4 + mNO3)
50 rate = 1.57e-7*mO2/(2.94e-4 + mO2)
# 60 rate = rate + 1.e-13*mSO4/(1.e-4 + mSO4)
60 rate = rate + 1.e-10*mSO4/(1.e-4 + mSO4)
70 moles = rate * m * (m/m0) * time
80 if (moles > m) then moles = m
200 save moles
-end
Pyrite
-start
1 rem parm(1) = log10(A/V, 1/dm) parm(2) = exp for (m/m0)
2 rem parm(3) = exp for O2 parm(4) = exp for H+
EQUILIBRIUM_PHASES
Al(OH)3(a) 0 0
Calcite 0 0
Coffinite 0 0
Jurbanite 0 0
Kaolinite 0 0
Pyrite 0 0
Uraninite(c) 0 0
SELECTED_OUTPUT
-file 3_degradation_organic_600days.csv
-totals Fe(2) Fe(3) U(6) As(3) As(5) Cu(1) Cu(2) Mn(2) Mn(3) Mn(6) Mn(7)
-molalities SO4-2 CaSO4 UO2+2 UO2SO4 CuCl2- Cu+ CuCl3-2 CuHCO3+ Cu+2 CuSO4
CuCO3
-saturation_indices Al(OH)3(a) Calcite Coffinite Jurbanite Kaolinite Pyrite
Uraninite(c)
-kinetic_reactants Organic_C Pyrite Calcite
END
61
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Surface Reactions
Sorption reactions, which include adsorption, desorption, chemisorption, etc. are a
surface effect that can be modeled as electrostatic attraction of ions in solution onto
charged surfaces, primarily clays and oxides. In clays, the surface charge is caused by
structural conditions where the substitution of Al for Si in the crystal lattice creates a
negative charge deficient on the surface. This charge deficient is thus intrinsic and
independent of solution chemistry. In contrast, oxides of Si, Al and Fe have ionized
surfaces depending on the pH of the solution.
OH Neutral
OH2+ Acidic
O- Basic
The neutral or isoelectric point is different for different oxide surfaces. Thus, some of the
surfaces in detrital sediment can be positively charged, while others are negatively
charged at the same solution pH.
Implicit Surfaces
There are two basic approaches to describing surface properties depending on the
material in question. The first is to use an empirical approach that describes the
interaction between solute and all the various surfaces in a sample, but does not explicitly
describe the electrical properties of the surface. This is the isotherm approach.
Kd = [solid]
[water]
where Kd is the distribution or partitioning coefficient with units of L/kg and the
concentration in the solid (Cs) is in units of mg/kg and the concentration in the solution
(Cw) is in units of mg/L.
Note this formulation is very similar to that of the equilibrium constant and like that
formulation we assume that sorption reactions reach equilibrium in natural systems. In
62
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
general, this is true as sorption reactions are relatively fast (hours to days). The
formulation is employed to provide an empirical description of the interaction of all the
surfaces in a sample and a single solute. The formulation is employed for both organic
and inorganic solutes. Strictly speaking the formulation does not accommodate multiple
solutes found in a natural system since the experiments are conducted for single solutes
only.
The relationship is measured by mixing a known solution with a known volume of solid,
and measuring the solution composition after equilibrium is reached (usually 12 to 48
hours). The simplest relationship between the amount of an element on the surface and
that in the solution is linear. The plot of Cs versus Cw is called an isotherm. The value
of Kd is temperature dependent. The slope of the plotted line is the Kd value.
An example of formulating a linear Kd and pH-dependent sorption reaction is shown
(taken from the PHREEQC website).
To write PHREEQC code that will adsorb HAsO4-2 dependent on the pH of the cell
solution and use a linear equation for Kd and pH with following form: Kd = -
.175*log(H+)+1.925 try the following.
The key is defining the right mass-action equation for your Kd. Assuming you are
sorbing HAsO4-2 and you want:
Kd = [HAsO4-2sorbed]/[HAsO4-2],
There are more complex formulations for isotherms including the Freundlich isotherm,
which has the form:
Kd = [solid]
[water]n
where n is a fitted parameter. Usually the n value ranges between 0.7 and 1.1. While
these formulations are good within the range of the data used to calculate the Kd, they are
difficult to extrapolate to higher values. Both the linear and Freundlich isotherms assume
that there are an infinite number of sorption sites, not the case natural surfaces. A more
realistic formulation is the Langmuir isotherm that assumes a finite number of sorption
sites.
[solid] = αβ[water]
1 + α [water]
where α is the sorption constant (L/mg) and β is the maximum amount of solute that can
be sorped by the solid (mg/kg), called the cation exchange capacity (CEC).
Again from the PHREEQC website:
Explicit Surfaces
The second approach is to surface reactions is to explicitly describe the surface and its
electrical properties, usually as a double layer (see Drever, 1999). Traditionally,
quantitative descriptions of adsorption are accomplished using empirical adsorption
models. Distribution coefficients, Kd, are often used to model sorption due to the
experimental simplicity (batch and column studies) and the ease of incorporation into
transport models. However, factors such as solute composition, pH, PCO2 and solid
characteristics may significantly impact Kd values (USEPA, 1999). Langmuir and
Freundlich isotherm constants will generally describe sorption over a wider concentration
range than Kd values but the constants are also system dependent (Koretsky, 2000).
Therefore, Kd values and isotherms derived from a specific set of conditions may not be
applicable to natural systems with different geochemical conditions. Thermodynamic-
based surface complexation models (SCMs) are considered an improvement over
66
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
empirical models due to the reduced system dependence of the intrinsic equilibrium
constants, Kint.
The fundamental difference between the isotherm and surface models is the explicit
description of the surface, and theoretically, the ability to model multi-solute interactions
with that surface. There are two main approaches used in surface complexation models
capable of describing sorption to heterogeneous sediments and soils. They are the
component additivity (CA) and generalized composite (GC) approaches (Kent et al.,
2000; Waite et al., 2000; Sanpawanitchakit, 2002; Davis et al., 2004).
67
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
experiments in their compilation determined the sorption constants using a background,
non-sorbing electrolyte, one sorbing ion, and temperatures from 20-30ºC. These single-
solute sorption constants have been incorporated in the most commonly available
computer codes including Geochemist’s Workbench (Bethke, 1992), MINTEQA2
(Allison et al., 1991), and PHREEQC (Parkhurst and Appelo, 1999). The codes predict
the distribution of sorbed and dissolved species by solving a set of simultaneous non-
linear equations, however, the applicability of these constants in modeling multi-
component and lower temperature systems has never been rigorously tested. When using
either approach, the equilibrium constants should be calibrated with field data from the
site being modeled.
Dzombak and Morel (1990) used the Generalized Two Layer Model, which is a
modification of the Diffuse Double Layer Model by adding surface precipitation and two
types of binding sites, strong and weak for sorption of cations. This basic approach can
be modified to add multiple surfaces with different properties to the model.
OH Na+
O-
+ Cl- Na+
OH2
Na+
Surface
OH
OH
+
Cl- Cl-
OZn
SO4-
OH
Surface Plane Diffuse Layer
Potential
Distance, x
68
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
The surface of the iron solid contains reactive surface sites that can be represented as
≡ FeOHº. On the surface of the bulk solid are sites that are protonated according to
Equations 6.1 and 6.2.
All protonated sites may also sorb cations or anions. Equations 6.3 and 6.4
provide examples, using ≡ FeOHº although the reactions can occur on the other two site
types as well.
Therefore, the compilation of K values usually includes two values for each cation, one K
for sorption to the strong sites and one K for sorption to the weak sites. Equations 6.5
and 6.6 show sorption of Zn to the strong and weak sites as in the GTLM.
The GTLM as well as the DDLM represents the surface of the reactive solid as having a
surface potential that is constant through a finite width of the surface plane. Beyond the
boundary of the plane, the surface potential gradually decreases in the bulk solution. In
this first plane, the protonation and deprotonation reactions occur, as well as the specific
sorption of cations and anions (inner sphere). Beyond the boundary of the first plane, and
in what is called the diffuse layer, non-specific sorption of cations and anions occur
through long range interactions with the surface sites (outer sphere). In the remaining
bulk solution, the background electrolytes have no specific orientation with respect to the
surface. The figure above shows the relationship of specifically (Zn 2+) and non-
specifically (SO42-) sorbed solutes to the reactive surface as well as the potential
relationships.
The PHREEQC model contains a pre-loaded ferrihydrite surface and additional surfaces
can be added as the user desires using either type of formulation (isotherm or electrical
69
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
surface). The more complex double layer model of Dzoback and Morel can be
configured with and without an explicitly calculated diffusion layer.
The standard PHREEQC manual is not very clear on using the SURFACE command.
The SURFACE command can be configured in one of two modes. The first is termed the
explicit mode in the standard manual (closed system). In this mode, the surface can be
used within a reaction step where the system includes the surface (surf_in _output1).
This creates and reports the composition of the surface in equilibrium with the solution
by redistributing the initial species specified in the solution between aqueous and solid
phases and corresponds to a closed system (effectively fixed water–solid ratio). The
amount of solid is set by the third number in Hfo_sOH line (89) which means 89 grams
of ferrihydrite are in the system.
SOLUTION 1 AMD
units ppm
pH 5
density 1.00
temp 25.0
Al 1.237
As 4.210
C(4) 5.87 as HCO3
Br 0.054
Ca 5.16
Cl 0.994
Cd 0.23
Cu 0.041
F 0.142
Fe(+3) 23.19
Fe(+2) 1.2
K 0.624
Mg 1.227
Mn(2) 0.0098
Na 0.539
N(+5) 0.02
N(+3) 0.001
P 2.1
Pb 0.0012
S(6) 82.65 charge
Si 12
Zn 0.501
O(0) 4.1
SURFACE 1
Hfo_sOH 0.005 600 89
Hfo_wOH 0.2
END
The other mode, termed implicit in the standard manual, is to first create a system (e. g.
solution) and then react that system with the surface, creating a new system that
represents equilibrium between the solution and surface without changing the solution
composition, effectively an open system where the water-solid ratio is very large. The
example below (surf_in _output2) will produce an output file that lists the surface
composition that is in equilibrium with the initial solution, but will not redistribute the
70
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
components in that solution when creating the surface. In this case the SURFACE
command is preceeded by the subcommand, equilibrate.
SOLUTION 1 AMD
units ppm
pH 3.499
density 1.00
temp 25.0
Al 1.237
As 4.210
C(4) 5.87 as HCO3
Br 0.054
Ca 5.16
Cl 0.994
Cd 0.23
Cu 0.041
F 0.142
Fe(+3) 23.19
Fe(+2) 1.2
K 0.624
Mg 1.227
Mn(2) 0.0098
Na 0.539
N(+5) 0.02
N(+3) 0.001
P 2.1
Pb 0.0012
S(6) 82.65 charge
Si 12
Sr 0.052
Zn 0.501
O(0) 4.1
SURFACE 1
-equilibrate with solution 1
Hfo_sOH 0.005 600 89
Hfo_wOH 0.2
END
The amount of surface can be specified or taken from a previous batch reaction step in
which the surface is gerneated. For instance, the AMD solution in the prior examples is
supersaturated with respect to ferrihydrite. If we allow the solution to equilibrate with
ferrihydrite, we can use that surface for a later step. The following input file (input file
surf_in_ouput 3) will again simulate an apparent open system, but use the ferrihydite
created in the prior batch reaction step (EQUILIBRIUM_PHASES command) as the
surface in the SURFACE command step. In this simulation the amount of ferrihydrite is
only 3.5e-4 moles. However, this input formulation will not produce an open system
result since any solution created by a batch reaction calculation (the
EQUILIBRIUM_PHASES command followed by the SURFACE command is considered
a single batch reaction caculation) will equilibrate with the surface changing the
composition of both (in effect negating the equilibrate subcommand). In order to have
the open system calculation for surf_in_output 3 input file you must separate the batch
71
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
reaction calculation into two parts by placing the END command between
EQUILIBRIUM_PHASES and SURFACE commands (input file surf_in_output 4).
SURFACE 1
Hfo_sOH 0.005 600 89
Hfo_wOH 0.2
END
Remember that if the grams of ferrihydrite are reduced to 0.089 (1000X), the number of
sites must be reduced proportionally to maintain the applicability of the Dzomback and
Morel model. The specific surface value remains the same.
SURFACE 1
Hfo_sOH 0.000005 600 0.089
Hfo_wOH 0.0002
SURFACE 1
Hfo_sOH Ferrihydrite equilibrium_phases 0.005 5.3e4
Hfo_wOH Ferrihydrite equilibrium_phases 0.2
In the example above (surf_in _output3), the solution was equilibrated with ferrihydrite
created using the EQUILIBRIUM_PHASES command (there was no initial ferrihydrite
in the system). In that example, the ferrihydrite precipitates and is saved for use in the
second batch-reaction, equilibration with the surface.
There are times when it is useful to examine the behavior of dissolved metals with pH in
the presence of a reactive surface. The most common type of exploratory simulation is
the plot of speciation over a relevant pH range. To make this type of simulation we
employ the pH sweep technique described above with suitable modifications, specifically
we have to include the surface. The following example (input file pH sweep surface) is
from the standard PHREEQC example problems.
The following figure shows the results of the simulation. You can see that most of the
Zn in the system is in solution at pH values less than 6. At more alkaline values the Zn
starts to be strongly sorped to the surfaces.
Zn distribution
1E-7
m_Zn+2
m_Hfo_wOZn+
8.000001E-8 m_Hfo_sOZn+
6.000002E-8
moles
4.000003E-8
2.000004E-8
5E-14
5 6 7 8 9 10
pH
76
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
More Reaction Path Models
As discussed earlier, the reaction path concept is simply a model that uses speciation
calculations to make forward predictions of changes in water and rock
(dissolution/precipitation) along reaction path (specified change in T, P, pH, new
reactants).
As the figure below shows, the system can be altered by adding or subtracting mass of
heat, or composition or controlled by contact with another reservoir that buffers change.
The model operates by first calculating the equilibrium condition of the specified initial
system. The program then changes the condition (e.g. temperature, composition) by a
small increment and calculates the new equilibrium condition. Then the next step uses
the last step as the starting point until the end of the path is reached. A good way to
visualize this is the addition of base to a solution (titration) to change the initial acidic pH
of 2 to a value of 12 with a calculation made for each drop of acid added. During the
process, the solution may have minerals precipitate or dissolve, gas content increase or
decrease, etc.
This method is used to model many real world processes of interest such as mixing of
fluids, dissolution of minerals, changes in temperature, Eh, pH, fugacity or concentration
of a particular species.
77
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
MIXING
Water Compatibility
Often the question of the compatibility of the formation water with potential sources of
water flood brines needs to be addressed. In this example we will use the formation
water from the example above and mix it with injection water. The injection water has a
lower salinity and may cause dissolution or precipitation of reservoir minerals. The
injection water chemistry is listed below (input file injection water).
SOLUTION 2
-units mg/l
-pH 7.25
-temp 40
-pe -6
Na 1200
K 70
Ca 600
Mg 120
Cl 200 charge
C(4) 750
S(6) 40
Fe 1.4
F 0.6
Ba 0.02
Si 50 as SiO2
END
In PHREEQC you can use the MIX command to mix any number of solutions together in
specified proportions. The example input below mixes solution 1, the formation water
with 100 volumes of the injection water. The first number in the lines under the MIX
command specifies the solution and the second that mixing fraction. So the same result
could have been achieved using mixing fractions of 0.1 and 100, or 0.01 and 10. In this
case we might be interested in the potential for mineral precipitation or dissolution.
In that case it would be most useful to have a output file that tracked the saturation index
of the reservoir minerals. Let’s assume that the reservoir is composed of quartz, calcite,
albite (feldspar), illite and kaolinite (clay minerals). In order to track the feldspar and
clay minerals we must include all the elemental components of the minerals in the water
chemistry (the system). This means we have to add aluminum to the water chemistry.
This brings up an important point for most engineering applications, that the water
analyses we commonly use to not include mineral components such as Al and Si. In
those cases we need to measure these parameters or make assumptions about the
dissolved concentrations. The usual assumption in the case of petroleum reservoirs is
that the Si is controlled by equilibrium with quartz and Al is controlled by equilibrium
with a feldspar or clay mineral. Therefore, the dissolved Si and Al concentrations can be
calculated by specifying mineral equilibrium as described above.
78
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
In practical terms we need some knowledge of the reservoir minerals. Let us specify that
the injection water source has quartz, kaolinite and calcite in the formation. The file
below sets equilibrium with those minerals (input file water compatibility). The initial Al
value is just a guess to start. We can also add a SELECTED_OUTPUT command to
make the output data easier to handle. From the output we can see that Calcite and
Kaolinite may precipitate.
79
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
MIX
1 1.0
2 100
END
81
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
pH 8.1
density 1.00
temp 25.0
As 0.010
Cu 0.010
Pb 0.0002
Zn 0.050
Fe 0.010
Na 22.92
C(4) 67.0 as HCO3
Cl 3.6
S(6) 40 charge
Ca 20.0
Mg 8.4
F 2.8
EQUILIBRIUM_PHASES
CO2(g) -3.5
O2(g) -0.1
MIX 1 mixing AMD and stream water #this simulates mixing of stream and AMD
1 1.0
2 1000.0
EQUILIBRIUM_PHASES
O2(g) -0.10
Ferrihydrite 0.0 0.0
SAVE solution 3 #this is mixture 1
SAVE equilibrium_phases # this command keeps the ferrihydrite created by mixing for later use
END
SURFACE 1
-equilibrate with solution 3
Hfo_sOH Ferrihydrite equilibrium_phase 0.1 1e5
Hfo_wOH Ferrihydrite equilibrium_phase 0.001
END
Evaporation
Evaporation is the removal of water from the initial system. During this process a typical
reaction path will involve the increases in concentration of solutes, ionic strength, and if
the process continues, some solid phases may become supersaturated and precipitate. In
order to simulate this process we can use the following input file (evap1). This is also
example 4a from the PHREEQC example files
.
TITLE Example 4a.--Rain water evaporation
SOLUTION 1 Precipitation from Central Oklahoma
units mg/L
pH 4.5 # estimated
temp 25.0
Ca .384
Mg .043
Na .141
K .036
Cl .236
C .1 CO2(g) -3.5
S(6) 1.3
N(-3) .208
82
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
N(5) .237
REACTION 1
H2O -1.0
52.73 moles
END
In this case the simulation is for the evaporation of rain water by a factor of
approximately 95% (52.73/55.5) since 1 kg of water is 55.5 moles. Note the “reactant” is
negative water, or the reaction path is the removal of 52.73 moles of water from the
initial system of 1 mole and solutes. Note that this simulation did not allow precipitation
of minerals. Examination of the output shows that in fact no minerals did become super-
saturated, so precipitation is not likely.
A more complex simulation would be the case where the water undergoing evaporation
has more solutes to begin. Let us use the AMD from the mixing with a stream example
(input file evap AMD).
In this case the results are very different. For instance, Barite starts out slightly super-
saturated but becomes over two orders of magnitude more super-saturated with
evaporation. Of course the model does not let any phases precipitate without an
EQUILIBRIUM_PHASES command. So we can continue to modify our simulation to
make it increasingly realistic by allowing phases to precipitate, perhaps specifying
equilibrium with the atmosphere as well.
83
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Inverse Modeling
This function in PHREEQC is unique among the available models. The user specifies the
initial and final solution chemistry (e.g. samples from upstream and downstream) and the
solid phases that are available to react with the initial solution to create the final solution.
The program calculates all the possible mass balance models that can create the final
solution from the initial solution and all possible combinations of available reactants.
This is particularly useful when modeling a well-constrained system such as a column
experiment where the initial and final solution chemistry is well known and the solid
phase can be well characterized. The classic application is the modeling of water-rock
interaction in the Sierra Nevada mountains in the PHREEQC examples (Ex16).
INVERSE_MODELING 1
solutions 1 2
uncertainty 0.025
range
phases
Halite
Gypsum
Kaolinite precip
Ca-montmorillonite precip
CO2(g)
Calcite
Chalcedony precip
Biotite dissolve
Plagioclase dissolve
balance
Ca 0.05 0.025
PHASES
Halite
NaCl = Na+ + Cl-
log_k 0.0
Biotite
KMg3AlSi3O10(OH)2 + 6H+ + 4H2O = K+ + 3Mg+2 + Al(OH)4- + 3H4SiO4
log_k 0.0
Plagioclase
Na0.62Ca0.38Al1.38Si2.62O8 + 5.52 H+ + 2.48H2O = \
0.62Na+ + 0.38Ca+2 + 1.38Al+3 + 2.62H4SiO4
log_k 0.0
END
There are several interesting features in this input file. The example uses the
SOLUTION_SPREAD command to define the two solutions. Additional solid phases
(Halite, Biotite and Plagioclase) are added to the system in the input. Note that the log K
84
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
for the reactions is zero. The INVERSE_MODELING feature does not use the
equilibrium constant of the mass balance equation, only the mole balance (stoichiometry)
of the reaction. Some of the phases are constrained to either precipitate or dissolve. If
unspecified, then the phases can either dissolve or precipitate (e.g. halite and gypsum).
Solution 2:
Input Delta Input+Delta
pH 6.800e+00 + -3.417e-03 = 6.797e+00
Al 0.000e+00 + 0.000e+00 = 0.000e+00
Alkalinity 8.950e-04 + -1.801e-06 = 8.932e-04
C(-4) 0.000e+00 + 0.000e+00 = 0.000e+00
C(4) 1.199e-03 + 0.000e+00 = 1.199e-03
Ca 2.600e-04 + 6.500e-06 = 2.665e-04
Cl 3.000e-05 + 0.000e+00 = 3.000e-05
H(0) 0.000e+00 + 0.000e+00 = 0.000e+00
K 4.000e-05 + 1.000e-06 = 4.100e-05
Mg 7.100e-05 + -9.005e-07 = 7.010e-05
Na 2.590e-04 + 0.000e+00 = 2.590e-04
O(0) 0.000e+00 + 0.000e+00 = 0.000e+00
S(-2) 0.000e+00 + 0.000e+00 = 0.000e+00
S(6) 2.500e-05 + 0.000e+00 = 2.500e-05
Si 4.100e-04 + 0.000e+00 = 4.100e-04
85
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Sum of residuals (epsilons in documentation): 5.576e+00
Sum of delta/uncertainty limit: 5.576e+00
Maximum fractional error in element concentration: 5.000e-02
Solution 2:
Input Delta Input+Delta
pH 6.800e+00 + -3.417e-03 = 6.797e+00
Al 0.000e+00 + 0.000e+00 = 0.000e+00
Alkalinity 8.950e-04 + -1.801e-06 = 8.932e-04
C(-4) 0.000e+00 + 0.000e+00 = 0.000e+00
C(4) 1.199e-03 + 0.000e+00 = 1.199e-03
Ca 2.600e-04 + 6.500e-06 = 2.665e-04
Cl 3.000e-05 + 0.000e+00 = 3.000e-05
H(0) 0.000e+00 + 0.000e+00 = 0.000e+00
K 4.000e-05 + 1.000e-06 = 4.100e-05
Mg 7.100e-05 + -9.006e-07 = 7.010e-05
Na 2.590e-04 + 0.000e+00 = 2.590e-04
O(0) 0.000e+00 + 0.000e+00 = 0.000e+00
S(-2) 0.000e+00 + 0.000e+00 = 0.000e+00
S(6) 2.500e-05 + 0.000e+00 = 2.500e-05
Si 4.100e-04 + 0.000e+00 = 4.100e-04
86
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Model contains minimum number of phases.
Summary of inverse modeling:
Number of models found: 2
Number of minimal models found: 2
Number of infeasible sets of phases saved: 20
Number of calls to cl1: 62
End of simulation.
The output is somewhat complicated with many parameters that are not always used.
The important point is the mole phase transfer values. Positive values indicate that the
phase was added and negative values show the phases precipitated. Adding the
SELECTED_OUTPUT command will create a table for the grid function in PHREEQC
or to be opened as an external spreadsheet. This is useful to evaluate which possible
models are geologically reasonable and which should be ignored. Insert the following
after the last solution and before END, and re-run the model.
SELECTED_OUTPUT
-file inverse1.csv
-inverse_modeling
Shale-hosted deposits have a relatively simple ore-mineral assemblage with the sulfides
sphalerite, galena, pyrite and marcasite. Gangue may include forms of silica, barite,
carbonates and several different clays. The problem is to see if you can formulate a
predictive model for the changes in stream water chemistry due to contact with the
mineralized rocks, a natural acid drainage. We have the background concentration
(stream water upgradient of mineralized areas), two samples from locations downstream
from the mineralized areas and the mineral assemblage including the secondary minerals
of the deposits to help formulate and constrain the model.
Three samples will be used for modeling. Sample DW002 was collected upgradient of
the mineralization, sample DW007 was collected downgradient of an area with minimal
mineralization, and sample DW010 was collected downgradient of an area of extensive
mineralization. We will use the PHREEQC commands INVERSE MODELING,
REACTION, EQUILIBRIUM PHASES, and SURFACE. First, we speciate the water
samples with the solution command. This is input file 7.
TITLE Background Speciation
87
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
SOLUTION 1 DW002
units ppm
pH 7.2
pe 12
temp 6
F 0.035
Cl 0.07
S(6) 12
N(5) 0.14
Al 0.07
Ca 9
Fe 0.042
K 0.1
Mg 1
Mn 0.014
Na 0.4
Si 0.7
Ba 0.039
Cd 0.0014
Cu 0.00063
Pb 0.00021
Sr 0.027
Zn 0.012
Alkalinity 12.7 as HCO3
O(0) 1.0 O2(g) -0.7
SOLUTION 2 DW007
units ppm
pH 7.0
pe 12
temp 9
F 0.09
Cl 0.11
S(6) 24
N(5) 0.14
Al 0.07
Ca 10
Fe 0.3
K 0.2
Mg 2
Mn 0.065
Na 0.4
Si 1
Ba 0.039
Cd 0.0014
Cu 0.002
Pb 0.00021
Sr 0.034
Zn 0.073
Alkalinity 10.5 as HCO3
O(0) 1.0 O2(g) -0.7
SOLUTION 3 DW0010
units ppm
pH 4.3
pe 12
temp 3
F 0.18
88
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Cl 0.11
S(6) 49
N(5) 0.14
Al 1
Ca 11 charge
Fe 3
K 0.3
Mg 1
Mn 0.48
Na 0.1
Si 0.7
Ba 0.032
Cd 0.006
Cu 0.007
Pb 0.005
Sr 0.013
Zn 1.4
Alkalinity 7 as HCO3
O(0) 1.0 O2(g) -0.7
END
89
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
pe 12
temp 9
S(6) 24
Al 0.07
Ca 10 charge
Fe 0.3
K 0.2
Mg 2
Si 1
Zn 0.073
Alkalinity 10.5 as HCO3
O(0) 1.0 O2(g) -0.7
INVERSE_MODELING 1 DW002 and DW007 using py and sph
solutions 12
uncertainty 0.05 0.05
phases
Quartz dissolve
Pyrite dissolve
Sphalerite dissolve
CH4(g)
Calcite dissolve
Chlorite(14A) dissolve
Illite
Ca-Montmorillonite
Fe(OH)3(a) precipitate
END
For both models, F, Cl, NO3, Na, Mn, Ba, Cd, Cu, and Sr were removed from the solution
inputs. The addition elements require a solid phase containing these elements in the
PHASES list. This constraint makes model convergence very difficult since these
elements are not major components in the water chemistry. Lead (Pb) was removed from
the first model because it was reported as non-detect in both samples and there was no
basis to specify a value. A lead value was included in the second model. As seen in the
input files, the phases quartz, pyrite, sphalerite, CH4(g), calcite, and illite are included in
both models. Illite was required to satisfy K, calcite for Ca, pyrite for Fe, sphalerite for
Zn, and quartz for Si. Methane, CH4(g) was required for the C(-4), to force the pe to
match the reported value.
This may seem strange as there is not likely to be any methane in the stream. However,
attempts to decouple all reactions involving CH4 and completely remove it from the
database, including as a master species produced an inequality error in the output. This
may be because CH4 is coupled to pe in the code of the program, and it may not be
possible to completely remove it. This situation points out how difficult it is to model Eh
in real world systems.
Chlorite and Ca-montmorillonite were added as phases in the first model to provide Mg
and Ca, respectively. The CO2, kaolinite, and galena were added in the second model for
C, Al, and Pb respectively. The data for galena was obtained from the Wateq4f database.
The second model was run with Chlorite and Ca-Montmorillonite.
Examination of your output shows there are 10 possible models. Phases were
sequentially removed to attempt to decrease the possible models and reach the simplest
set of models. Removing chlorite reduced the possible models to six, removing Ca-
montmorillonite reduced the models to 1. The inverse modeling shows that with the
91
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
correct set of reactions, it may be possible to explain the change in water chemistry from
background to downstream.
After successful completion of inverse modeling, forward modeling was attempted with
the REACTION command. This is a reaction path model that could be used to predict
the affects of mineralized deposits on stream chemistry. We know that the reaction of
stream water with the sulfide minerals should locally decrease the pH and Eh of water.
But in order to better match the actual data, we added calcite to increase the pH and
oxygen to adjust the pe. This is not an unreasonable situation, but we are now modeling
and should not forget that we have not verified these assumptions. This is input file 7c.
TITLE React DW002 Background with py and sph
SOLUTION 1 DW002
units ppm
pH 7.2
pe 12
temp 6
S(6) 12
Al 0.07
Ca 9 charge
Fe 0.042
K 0.1
Mg 1
Si 0.7
Pb 0.00021
Zn 0.012
O(0) 1.0 O2(g) -0.7
REACTION 1 DW002 to DW007
Pyrite 4.62
Sphalerite 0.94
Calcite 12.28
O2(g) 0.045
1.0e-06 moles
END
Calculated S is only about half of what was measured in the water samples. Therefore,
further modification to each model is necessary. Based on the supersaturated phases for
each water sample, and the kinetics involved in precipitating the species in natural
systems, the most likely phase to precipitate is ferrihydrite. Allowing precipitation of
ferrihydrite will remove Fe from the system. This will stimulate the dissolution of more
pyrite, which will increase sulfate. This is input file 7d on your disk.
TITLE React and precip DW002 Background to DW010 with py, sph and gln
SOLUTION 1 DW002
units ppm
pH 7.2
pe 12
temp 6
S(6) 12
Al 0.07
Ca 9 charge
Fe 0.042
92
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
K 0.1
Mg 1
Si 0.7
Pb 0.00021
Zn 0.012
O(0) 1.0 O2(g) -0.7
REACTION 1 DW002 to DW010
Pyrite 6.5
Sphalerite 2.125
Galena 0.00231
O2(g) 0.43
1.0e-05 moles
EQUILIBRIUM PHASES 1
Fe(OH)3(a) 0.0 0.0
END
In this model more pyrite was dissolved to reach the higher measured concentration, but
the pe dropped to negative values. At low pe, ferrihydrite was not supersaturated.
Therefore, it was necessary to add O2(g) in the REACTION datablock. Adding oxygen
raises the pe closer to the known value and allowed ferrihydrite to precipitate. In effect,
we are simulating a stream with active atmospheric exchange.
However, now the calculated pH is slightly lower than the measured value (3.9 vs 4.3).
Note that the SURFACE command was used to allow absorption to ferrihydrite. The
amount of ferrihydrite that precipitated from the EQUILIBRIUM_PHASES command
can be used as input for the datablock. The known Fe concentrations, recommended
number of moles of absorption sites per mole of Fe and default value of surface area are
used. Both strong and weak surface sites are modeled. This is input file 7e.
TITLE React, precip, and sorb DW002 to DW010
SOLUTION 1 DW002
units ppm
pH 7.2
pe 12
temp 6
S(6) 12
Al 0.07
Ca 9 charge
Fe 0.042
K 0.1
Mg 1
Si 0.7
Pb 0.00021
Zn 0.012
O(0) 1.0 O2(g) -0.7
REACTION 1 DW002 to DW010
Pyrite 6.5
Sphalerite 2.125
Galena 0.00231
O2(g) 0.43
1.0e-05 moles
EQUILIBRIUM_PHASES 1
Fe(OH)3(a) 0.0 0.0
93
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
SURFACE 1
Hfo_sOH 2.7e-07 600 1.3e-03
Hfo_wOH 1.1e-05 600 1.3e-03
END
The final model sorbed mainly OH to the strong and weak sites. This increased the pH
by 0.003 units. Sulfate was also sorbed to the surface, which decreased the sulfate in the
resulting solution. While Fe and Zn were sorbed as well, the amount was so small it was
insignificant. The small change in pH caused ferrihydrite to precipitate, removing Fe.
The concentrations of S and Pb also decreased. Since more sulfate than Fe was removed,
including sorption in the overall model made the difference between the calculated and
measured water chemistries greater. However, since even a small change in pH changes
the Fe speciation, properly modeling the pH may resolve the remaining differences in the
water chemistries.
This exercise shows just how difficult it is to generate a forward model. Reaction path
modeling can be difficult and time consuming. The best rule is start simple and gradually
increase complexity until you reach a reasonable solution that captures the essential
portions of the system you are trying to model.
Transport
The code can model one dimensional transport processes such as diffusion, advection,
advection and dispersion, advection and diffusion with diffusion into stagnant zones (dual
porosity). These options cover most simple systems in the environmental field. The
transport processes can be combined with equilibrium and kinetic chemical reactions
producing the capacity to emulate the simpler reactive-transport codes. The PHREEQC
code solves the advective-transport equation:
C C 2C q
v DL 2
t x x t
where v = advective velocity in the pores, q/porosity (m/sec), t = time (s), C =
concentration (mol/Kgw), DL = hydrodynamic dispersion coefficient (m2/sec) and q =
concentration in the solid phase (mol/kgw in pore space of the 1 kg of water). This
formulation is equivalent to a 1-D column experiment.
“For each time step, advective transport is calculated, then all equilibrium and kinetically
controlled chemical reactions, followed by dispersive transport, which is followed again
by calculation of all equilibrium and kinetically controlled chemical reactions. The
scheme differs from the majority of other hydrogeochemical transport models (Yeh and
Tripathi, 1989) in that kinetic and equilibrium chemical reactions are calculated both
after the advection step and after the dispersion step. This reduces numerical dispersion
and the need to iterate between chemistry and transport” – Parkhurst and Appelo, 1999.
94
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
The model can be envisioned as a series of cups (cells) filled with an initial fluid, which
then is shifted (transported) from the first cell into the adjacent cell and so forth down the
line of cells. In transport simulations the three most important variables are the time step,
shifts and cells. “The time step is the amount of time it takes for water to travel from one
cell to the next; the “shifts” define how many times water travels (shifts) from one cell to
the next; and the length(s) define the length of the cells. For each cell, the velocity is the
cell length divided by the time step. It takes the same length of time to traverse any of
the cells, but the cells may have different lengths; therefore the velocities may change as
you move through the column. It is easier to use a constant cell length and consequently
a constant velocity through the system. You need to run at least as many shifts as you
havecells for the front to move through the entire column”- Parkhurst, 2002.
In order to simulate the movement of groundwater (and solutes) we define the number of
shifts and the time steps. Shifts are the number of times that that the solution in cell one
is moved into the adjacent cells. In the example, six shifts will move the solution in cell
one all the way down the line to cell six (66 meters). The velocity is dependent on the
length of the “column” (six cells x 11m = 66m), and the time step, which is the time in
seconds for each shift (6 x 1.05e8 seconds = 6.3 x 10e8 seconds = 20 years) for a velocity
of 3.3m/year. Porosity is not specified explicitly, but is rather implicit in the velocity in
the model. The dispersion and diffusion coefficients are specified (usually derived from
literature values or adjusted to fit the data). For this exercise an arbitray value of 0.1 is
used. Note that the dispersivity value entered is in length (m) and is the α term in the
equation DL = αv + D* where v is advective velocity and D* is the diffusion coefficient
(m2/sec). For this example we are using an alpha value of 0.1 m. This value is a value
for each cell. The value for dispersivity is often expressed as a fraction of path length,
such as longitudinal dispersivity, L = 0.1L where L is the length of the flowpath.
However, in PHREEQC the cell length is used to estimate the value not the total path
length since the value will be applied to each cell. This also allows the user to assign
different values to each cell to represent heterogenities. The summary figure below
tabulates so of the dispersivity relationships that have been proposed and shows the scale
dependency of the value.
95
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Since the calculation is made for each cell and each time step, the amount of output can
be rather large. The output of transport models is managed with the
SELECTED_OUTPUT, and integral subcommands print and punch.
The example specifies that the results for the sixth cell will be written to the output file
after each shift, and the simulation will create a selected output file named grid.csv that
contains the amounts of Na, Cl and benzene for each step. This is input file 8a. Note you
will encounter a problem running this input file with the available databases. These
databases do not include benzene. You will have to add the benzene solute species to the
input file as per our prior example.
SOLUTION_MASTER_SPECIES
Benzene Benzene 0.0 78.0 78.0
SOLUTION_SPECIES
Benzene = Benzene
log_k 0
delta_h 0 kcal
PHASES
Benzene
96
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Benzene = Benzene
log_k -1.64
delta_h 0 kcal
Benzene(g)
Benzene = Benzene
log_k -0.9894
SOLUTION 0 Pulse solution leachate #solution injected
Units ppb
pH 7.0
density 1.00
temp 10.0
Na 39320
Cl 60660
Benzene 13.4
SOLUTION 1-6 Background solution initially filling column
Units ppb
pH 7.0
density 1.00
temp 25
Na 1966000
Cl 3033000
SELECTED_OUTPUT
-file grid.csv
-selected_out true
-high_precision false
# set value for all indentifiers to follow (lines 1 - 6)
-reset true
-simulation true
-state true
-solution true
-distance true
-time true
-step true
-percent_error true
-totals Cl Na Benzene
TRANSPORT Pulsing of Solution 0
-cells 6
-shifts 6
-time_step 1.05E8
-flow_direction forward
-boundary_conditions flux flux
-lengths 11.0
-dispersivities 0.1
-correct_disp true
-diffusion_coefficient 0.3e-9
-stagnant 0
-thermal_diffusion 1
-initial_time 0
-print_cells 6
-print_frequency 1
-punch_cells 123456
-punch_frequency 1
END
97
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
The output file can be fairly long even for this simple simulation. A better overview of
the simulation results is found in the grid.csv file that can be opened using the grid tab in
the PHREEQC window. The data was copied and pasted into MS Excel to create the plot
below. The figure below shows the results from input file 8a. You can see that the water
containing benzene is successively moved through the cells with the water advancing one
cell per shift until the benzene contaminated water reaches the last cell by twenty years.
2.5E-07
0y 3.3 y
6.6y 10 y
2.0E-07 13.2 y 16.7 y
20 y
Benzene (M)
1.5E-07
1.0E-07
5.0E-08
0.0E+00
0 10 20 30 40 50 60 70
distance (m)
If we want to look in greater spatial detail, we can increase the number of cells and
reduce the dimensions of each cell. We will need to increase the number of shifts
accordingly. The next input file increases the number of cells to 12 and reduces the
length of each cell by half (to 5.5 meters). Note that the number of shifts is increased so
the leachate reaches the end of the “column” and the time step is shortened (we have
more steps, if we left the time step unchanged, we would transport for 40 years not 20)
and the print parameters changed so all results for all the cells are printed. We also had
to change the initial solution modifer from 1-6 to 1-12 to fill all the cells with the initial
solution. This is input file 8b.
SOLUTION_MASTER_SPECIES
Benzene Benzene 0.0 78.0 78.0
SOLUTION_SPECIES
Benzene = Benzene
log_k 0
delta_h 0 kcal
PHASES
Benzene
Benzene = Benzene
log_k -1.64
delta_h 0 kcal
Benzene(g)
Benzene = Benzene
98
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
log_k -0.9894
SOLUTION 0 Pulse solution leachate #solution injected
units ppb
pH 7.0
density 1.00
temp 10.0
Na 3932000
Cl 6066000
Benzene 13.4
END
SOLUTION 1-12 Background solution initially filling column
units ppb
pH 7.0
density 1.00
temp 25
Na 1966000
Cl 3033000
END
SELECTED_OUTPUT
-file grid2.csv
-selected_out true
-high_precision false
# set value for all indentifiers to follow (lines 1 - 6)
-reset true
-simulation true
-state true
-solution true
-distance true
-time true
-distance true
-step true
-percent_error true
-totals Cl Na Benzene
TRANSPORT Pulsing of Solution 0
-cells 12
-shifts 12
-time_step 0.525E8
-flow_direction forward
-boundary_conditions flux flux
-length 5.5
-dispersivities 0.1
-correct_disp true
-diffusion_coefficient 0.3e-9
-stagnant 0
-thermal_diffusion 1
-initial_time 0
-print_cells 12
-print_frequency 1
-punch_cells 1 2 3 4 5 6 7 8 9 10 11 12
-punch_frequency 1
END
The figure below shows the results of the new simulation with increased spatial
resolution. Again the water containing benzene has successively moved through the cells
with the water advancing one cell per shift.
99
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
2.5E-07
0y 1.65 y 3.3 y
4.95 y 6.6 y 8.3 y
9.9 y 11.5 y 13.2 y
2.0E-07 14.8 y 16.7 y 18.2 y
20 y
1.5E-07
Benzene (M)
1.0E-07
5.0E-08
0.0E+00
0 10 20 30 40 50 60 70
distance (m)
SURFACE_MASTER_SPECIES
Surfa_s Surfa_sH2O
SURFACE_SPECIES
Surfa_sH2O = Surfa_sH2O
log_k 0.0
Surfa_sH2O + Benzene = Surfa_sBenzene + H2O
log_k 1.31
The leachate in the has higher chloride concentration than the groundwater (see input
file). We can use chloride as a conservative tracer and compare the chloride profile to
that of benzene.
100
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
1.E+00 6
0 y Cl
9.9 y - Cl 5
20 y - Cl
20 y Benzene
4
Benzene ppb
Chloride (M)
1.E-01 3
1
Initial benzene = 13.4 ppb
1.E-02 0
0 10 20 30 40 50 60 70
distance (m)
The plot shows that after 20 years of transport all the cells have been flushed with
leachate as indicated by the higher chloride profile, but that benzene has been retarded by
sorption on the organic matter in the aquifer, which has lowered aqueous concentrations
and limited travel to about 15 meters. The input file could be further modified to account
for the biodegradation of benzene using the KINETIC/RATE commands, which would
further reduce the concentration and could be used to evaluate the potential for natural
attenuation.
Another likely scenario is a spill where the contaminate addition is single event after
which fresh groundwater pushes the contaminated groundwater along. In order to
simulate this situation we can modify the input file as follows. First, we begin by making
sure that benzene the solute and an organic surface to sorb benzene is added to the model.
Then we will modify the spatial setting by making 20 cells 5 meters each for a total flow
path of 100 meters. Then we will assume that the groundwater velocity is 5 meters. The
first TRANSPORT block will move one slug of contaminated water 1 shift. This is
followed by 19 shifts using the original pore water for a total of 20 shifts for 20 cells, a
flowpath of 100 meters (input file 8d).
SOLUTION_MASTER_SPECIES
Benzene Benzene 0.0 78.0 78.0
SOLUTION_SPECIES
Benzene = Benzene
log_k 0
delta_h 0 kcal
PHASES
Benzene
Benzene = Benzene
log_k -1.64
delta_h 0 kcal
SURFACE_MASTER_SPECIES
101
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
Surfa_s Surfa_sH2O
SURFACE_SPECIES
Surfa_sH2O = Surfa_sH2O
log_k 0.0
Surfa_sH2O + Benzene = Surfa_sBenzene + H2O
log_k 1.31
SOLUTION 0 Pulse solution leachate #solution injected
units ppb
pH 7.0
density 1.00
temp 10.0
Na 3932000
Cl 6066000
Benzene 13.4
SOLUTION 1-20 Background solution initially filling column
units ppb
pH 7.0
density 1.00
temp 25
Na 1966000
Cl 3033000
SELECTED_OUTPUT
-file grid4.csv
-molalities Surfa_sBenzene
-selected_out true
-high_precision false
# set value for all indentifiers to follow (lines 1 - 6)
-reset true
-simulation true
-state true
-solution true
-distance true
-time true
-distance true
-step true
-percent_error true
-totals Cl Na Benzene
SURFACE 1-20
Surfa_sH2O 1 600 0.33
#the surface binding site name, sites in moles, specific area of surface
# in m2/g and the mass of surface in grams (per kilogram of solution)
END
TRANSPORT Pulsing of Solution 0
# simulation of 100 meter flowpath, cell length times cells
-cells 20
-shifts 1
-time_step 0.31536E8 # default in seconds
#shifts times time step gives total time of simulation
# velocity is total length/time, uniform cell length gives constant velocity
#for this example of 100 meters in 20 years, velocity is 5 m/y
-flow_direction forward
-boundary_conditions flux flux
-lengths 5
-dispersivities 0.1
-correct_disp true
-diffusion_coefficient 0.3e-9
102
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
-stagnant 0
-thermal_diffusion 1
-initial_time 0
-print_cells 20
-print_frequency 1
-punch_cells 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
-punch_frequency 1
END
SOLUTION 0 Background solution initially filling column
units ppb
pH 7.0
density 1.00
temp 25
Na 1966000
Cl 3033000
TRANSPORT Pulsing of Solution 0
-cells 20
-shifts 19
-time_step 0.31536E8
-flow_direction forward
-boundary_conditions flux flux
-length 5
-dispersivities 0.1
-correct_disp true
-diffusion_coefficient 0.3e-9
-stagnant 0
-thermal_diffusion 1
-initial_time 0
-print_cells 20
-print_frequency 1
-punch_cells 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
-punch_frequency 1
END
By setting the number of surface sites to zero in the SURFACE command block, we can
simulate the movement of benzene without the retardation caused by sorption. Then we
can reset the number of moles of surface sites back to 1 moles and re-run the simulation.
The result of the simulation shows that the initial introduction of the benzene creates a
slug in which the center of mass travels 100 meters in 20 years with spreading of solute
controlled by the dispersivity value and the groundwater velocity.
103
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
14
1 year
12 5 years
10 years
10 15 years
20 years
Benzene (ppm)
8
0
0 20 40 60 80 100
Distance (meters)
The simulation where benzene is retarded by sorption to the organic surface shows that
the dissolved benzene transport is very small since most of the organic solute is
partitioned onto the surface in the first cell lowering the dissolved content. The
continuous flushing of the first cell by more clean water gradually removes the sorbed
benzene (pump and treat), but even after 20 years the benzene on the surface has only
decreased by about half and the majority of the benzene has been moved into the surfaces
in the adjacent cells.
14
12 cell 1 solution
cell 1 surface
cell 2 surface
10
cell 3 surface
Benzene (ppm)
0
0 5 10 15 20
Time (years)
104
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
We can use the transport feature in PHREEQC to model treatment trains for acid
drainage as well. The following simulation is for the movement of acid drainage through
a channel lined with calcite. Since we know the reaction of calcite to neutralize the low
pH water is likely to be rapid, we will need a kinetic expression for calcite dissolution.
Examine the input file (input file treatment 1). You can see first the acid drainage
composition of solution 0, the “injected water”. Note that the amount of calcite to react is
set to 1 mole. This is simply set to a value that is sufficient and not meant to actually
represent the amount in the channel. That is, we assume calcite is in excess. We use the
KINETICS and RATE commands to make the calcite dissolution time-dependent and the
EQUILIBRIUM_PHASES command to make sure that any gypsum and ferrihydrite that
precipitates is quantified. Note that the initial amount of potential mineral precipitates is
set to zero meaning there is none initially. The precipitation is not time-dependent. This
may or may not be reasonable.
105
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
SOLUTION 1-20 surface water in carbonate channel
units ppm
pe 6.0
temp 10
pH 8
S(6) 14.3 as SO4
Cl 2.1
N(5) 0.5 as NO3
Fe 0.06
Na 5.8
K 1.5
Mg 3.5
Ca 36.6 charge
Si 3.64
C 130 as HCO3 CO2(g) -2.0
KINETICS 1-20
Calcite
-tol 1e-8
-m0 1
-m 1
-parms 50 0.6
RATES
Calcite
-start
1 rem parm(1) = A/V, 1/dm parm(2) = exponent for m/m0
10 si_cc = si("Calcite")
20 if (m <= 0 and si_cc < 0) then goto 200
30 k1 = 10^(0.198 - 444.0 / (273.16 + tc) )
40 k2 = 10^(2.84 - 2177.0 / (273.16 + tc) )
50 if tc <= 25 then k3 = 10^(-5.86 - 317.0 / (273.16 + tc) )
60 if tc > 25 then k3 = 10^(-1.1 - 1737.0 / (273.16 + tc) )
70 t = 1
80 if m0 > 0 then t = m/m0
90 if t = 0 then t = 1
100 moles = parm(1) * 0.1 * (t)^parm(2)
110 moles = moles * (k1 * act("H+") + k2 * act("CO2") + k3 * act("H2O"))
120 moles = moles * (1 - 10^(2/3*si_cc))
130 moles = moles * time
140 if (moles > m) then moles = m
150 if (moles >= 0) then goto 200
160 temp = tot("Ca")
170 mc = tot("C(4)")
180 if mc < temp then temp = mc
190 if -moles > temp then moles = -temp
200 save moles
-end
TRANSPORT 1-20
-cells 20 # 500m total length
-shifts 20 # 20 shifts means one pore volume
-time_step 25 # time in seconds per shift [1.0 m/s]
-length 25 # length per cell in m
-dispersivities 0.1 # dispersivity in m
-punch_cells 1-20 # Output: all cells
106
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
-punch_frequency 1 # Output: last time step
PRINT
-reset true
EQUILIBRIUM_PHASES 1-20
Gypsum 0 0 # gypsum precipitation without kinetics
CO2(g) -2.0 # 0.03 Vol% CO2
O2(g) -0.7 # 21 Vol% O2
Fe(OH)3(a) 0 0 # precipitate ironhydroxide
SELECTED_OUTPUT
-file treatment_1.csv
-totals Ca C Fe
-molalities SO4-2 CaSO4
-saturation_indices Gypsum Calcite Fe(OH)3(a)
-kinetic_reactants Calcite # how much calcite dissolves?
-equilibrium_phases Gypsum Fe(OH)3(a) # how much gypsum and Fe(OH)3 precipitates?
END
107
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.
List of References
Allison, J. D., D. S. Brown, and K. J. Novac-Gradac, 1991, MINTEQA2/PRODEFA2, a
geochemical assessment model for environmental systems: version 3.0 user's manual U.S.
EPA, Athens, GA, 104 p.
Appelo, C. A. J. and D. Postma, 1993, Geochemistry, groundwater and pollution. A. A.
Balkema, Rotterdam. 536 p.
Bethke, C. M., 1996, Geochemical Reaction Modeling – Concepts and Applications,
New York, Oxford University Press, 397p.
Drever, J. I., 1997, The Geochemistry of Natural Waters (3rd ed), New Jersey, Prentice-Hall Inc.
436p.
Dzomback, D. A., and Morel, F. M., 1990, Surface complexation modeling: New York, John
Wiley & Sons, 393 p.
Engesgaard, P., and T. H. Christensen, 1988, A review of chemical solute transport models.
Nordic Hydrology 19, (3): 183-216.
Garrels, R.M. and F. T. Mackenzie, 1967, Origin of the chemical compositions of some springs and lakes
in Equilibrium Concepts in Natural Waters, American Cancer Society, Washington D. C. p. 222-242.
Glynn, P. D., P. Engesgaard, K. L. Kipp, G. E. Mallard, and D. A. Aronson, 1991, Two
geochemical mass transport codes; PHREEQM and MST1D, their use and limitations at the
Pinal creek toxic-waste site; U.S. geological survey toxic substances hydrology program;
abstracts of the technical meeting, Monterey, California, March 11-15, 1991. Open-File
Report - U.S. Geological Survey OF 91-0088, (1991): 79p.
Greaser, K. K., 2004, Modeling Solid-Water Interactions in Mineralized Areas- An Example
from Red Dog, Alaska: Ph.D. Thesis, Colorado School of Mines, Golden, CO, 175 p.
Grove, D. B., and K. G. Stollenwerk, 1987, Chemical reactions simulated by ground-water-
quality models. Water Resources Bulletin 23, (4) (Aug): 601-615.
Güler C. and Thyne, G.D., 2004, Hydrologic and geologic factors controlling surface
and ground water chemistry in Indian Wells-Owens Valley area and surrounding
ranges, California, U.S.A. J. Hydrology 285, 177-198.
Lindberg, R. D, and D. D. Runnels, 1984, Ground water redox reactions: an analysis of
equilibrium state applied to Eh measurements and geochemical modeling: Science 225: 925-
927.
Mangold, D. C., and C-F. Tsang, 1991, A summary of subsurface hydrological and
hydrochemical models. Reviews of Geophysics 29, (1) (Feb): 51-79.
Nordstrom, D. K., L. N. Plummer, T. M. L. Wigley, T. J. Wolery, J. W. Ball, E. A. Jenne, and R.
L. Bassett, et al. 1979. Comparison of computerized chemical models for equilibrium
calculations in aqueous systems; chemical modeling in aqueous systems; speciation, sorption,
solubility, and kinetics. ACS Symposium Series 93: 857-892.
Parkhurst, D. and C.A.J. Appelo, 1999, USER’S Guide to PHREEQC (VERSION 2)—A
Computer Program for Speciation, Batch-Reaction, One-dimensional Transport and Inverse
Geochemical Calculations, Water-Resources Investigations Report 99-4259.
Zhu, C. and G. Anderson, 2002, Environmental Applications of Geochemical Modeling,
Cambridge University Press, 284p.
108
Prepared by Science Based Solutions LLC 2007.
All Rights Reserved.