Numerical modeling of arsenic mobility - In-Situ Remediation of Arsenic-Contaminated Sites - page 46

2.3.4 Mineral dissolution and precipitation
Besides redox and sorption-desorption reactions, precipitation and dissolution of minerals are
additional processes which can have a significant effect on the fate of arsenic. Dissolution reac-
tions involve the erosion of the structure of minerals. Trace elements such as arsenic, uranium
or lead, which can sometimes be present as major constituents of minerals or, more frequently,
as impurities in various minerals can be released to groundwater as the host mineral dissolves.
Prominent examples of arsenic release have been associated with the dissolution of sulfide miner-
als, most notably As-rich pyrite (FeS 2 ), which has been reported to contain up to 10 wt% arsenic
(Price and Pichler, 2007), arsenopyrite (FeAsS), realgar (AsS) and orpiment (As 2 S 3 ). These min-
erals are generally regarded as a primary source for As under oxidizing conditions (Plant et al. ,
2007; Sracek et al. , 2004; Welch et al. , 2000), while, under reducing groundwater conditions, the
reductive dissolution of Fe-oxides can be an important release mechanism for As (Burnol, 2007;
Dixit and Hering, 2003; Smedley and Kinniburgh, 2002).
At the same time, (co)-precipitation can, sometimes very strongly, mitigate the mobilization
of arsenic. For instance, the precipitation of As-bearing sulfides under reducing conditions, or
the precipitation of ferrihydrite under oxidizing conditions may reduce dissolved arsenic con-
centrations through incorporation of arsenic into the mineral structure at it forms (O'Day et al. ,
2004; Rittle et al. , 1995; Root et al. , 2009; Saunders et al. , 2008; Stollenwerk, 2003; Wolthers
et al. , 2008). Precipitation-dissolution and adsorption-desorption reactions are therefore not
independent and occur simultaneously.
In reactive transport models, dissolution-precipitation of minerals can be represented as equi-
librium or kinetic reactions, or both. Strictly speaking, mineral equilibrium should only be invoked
in cases where the local equilibrium assumption (LEA) applies, that is, where the timescale of the
precipitation-dissolution reaction is relatively fast compared to the transport timescale. Where this
is not the case, i.e., where mineral reactions are slow and mineral equilibria are not attained within
the time that corresponds to a travel-distance equivalent to the grid cell size imposed by model
discretization, the reactions should be simulated as kinetically controlled reactions. Reaction rate
equations exist for many common mineral dissolution and/or precipitation reactions that occur in
typical groundwater systems. These rate formulations can be incorporated into geochemical and
reactive transport models such as PHREEQC (Parkhurst and Appelo, 1999) and associated models
such as PHAST (Parkhurst et al. , 2010) and PHT3D (Prommer et al. , 2003) to simulate kinetically
controlled precipitation/dissolution reactions. Similarly kinetic source terms for arsenic release
or uptake (e.g., Lasaga, 1998; Matsunaga et al. , 1993; Schreiber and Rimstidt, 2013) can also
be formulated. Where more detailed reaction rate studies are unavailable, a commonly applied
reaction rate formulation for dissolution and precipitation of minerals is (e.g., Lasaga, 1998):

k k 1

IAP
K SP

R k = − (2.7)

where k k is an effective reaction rate constant and IAP / K SP is the saturation ratio, i.e., the ratio of
the ion activity product and the solubility product constant for the mineral.
The study by Wallis et al. (2010) is an example where kinetically controlled dissolution reactions
were used to define an arsenic source. This study presented simulations of an aquifer storage,
transfer and recovery (ASTR) operation in the Netherlands where oxygenated water was injected
into a reducing, pyritic aquifer. During injection, arsenic associated with pyrite, was found to
become unstable under the progressively more oxidizing conditions and thereby released into the
aqueous solution. Reactions for pyrite oxidation by oxygen and nitrate were therefore included in
the reactive transport model, based on previously proposed and applied rate expressions (Eckert
and Appelo, 2002; Prommer and Stuyfzand, 2005; Williamson and Rimstidt, 1994):

( C 0 . 5
O 2

10 − 10 . 19 A pyr
V

C
C 0

0 . 67

f 2 C 0 . 5

NO 3 ) C − 0 . 11

r pyr = + (2.8)

H +

pyr