United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S7-83-043 Nov. 1983
&ER& Project Summary
Limestone Dissolution in Flue
Gas Desulfurization Processes
Gary T. Rochelle, Pui K. R. Chan, and Anthony T. Toprac
Dissolution rates of reagent CaCOs
and commercial limestones (9 types/
19 grinds) have been measured at
constant pH and solution composi-
tion by batch titration with HCI. Condi-
tions were selected to simulate flue gas
desulfurization.
A mass transfer model has been
developed which includes theoretical
effects of particle size and equilib-
rium acid/base reactions. The cumu-
lative rate of mass transfer is calcu-
lated by integrating over a particle
size distribution obtained by Coulter
Counter and screening measure-
ments. The mass transfer model pre-
dicts measured dissolution rates with
a standard deviation of 30 percent,
without any allowance for limestone
type. Therefore, particle size distribu-
tion was found to bathe most significant
factor governing limestone reactivity.
The mass transfer model accurately
predicted the effects of solution com-
position and temperature, at pH 4 to 7,
25 to 55°C, 0 to 20 mM organic acid, 0
to 1 atm CO2. and 0 to 0.1 M Ca++. The
dissolution rate is a strong function of
pH and a weak function of temperature.
Buffers, such as adipic acid and low
concentrations of su Ifrte, enhance mass
transfer by increasing acidity transport
to the limestone surface. Mn+2, Fe+2,
Mg+2, and SOf inhibit limestone
dissolution, probably by formation of
adsorption surface layers.
This Project Summary was devel-
oped by EPA's Industrial Environmen-
tal Research Laboratory, Re-
search Triangle Park, NC, to an-
nounce key findings of the research
project that is fully documented in a
separate report of the same title
(see Project Report ordering informa-
tion at back).
Introduction
The rate of limestone (CaCOs) dissolu-
tion directly affects the overall perform-
ance of flue gas desulfurization (FGD)
i processes based on scrubbing with lime-
stone slurry. In combination with gas/
liquid mass transfer and calcium sulfite
(CaSOs) dissolution/crystallization, the
rate of CaCC"3 dissolution determines
the relationship of SO2 removal and
CaCC>3 utilization. It also affects the
potential for scaling by CaSC>3 and cal-
cium sulfate (CaSC>4) in the scrubber.
The operating pH of a CaCOs slurry
scurbber is a tradeoff of better S02 re-
moval at higher pH and improved lime-
stone utilization at lower pH. The rate of
limestone dissolution is also significantly
affected by dissolved C02. sulfite/bisul-
fite, and other buffer components in the
solution. In addition to quantifying these
effects of solution composition, it is also
important to predict the effects of varia-
tions in the type and grind of limestone.
Previous investigators in geochemistry
developed the pH-stat to measure
CaCOs dissolution rates. They con-
cluded that rates were controlled by H+
diffusion below pH 5.0 and by surface
reaction kinetics above pH 5.0. Previous
work in FGD has concentrated primarily
on methods of getting relative reactivity
as a function of type and grind.
This project adapted the pH-stat
method to measure absolute dissolution
rates at FGD conditions as a function of
solution composition and limestone type
and grind. Particle size distributions of
the CaCOs samples were measured by
a Coulter Counter. A mass transfer
-------
model was developed which includes
effects of particle size, diffusion of H+
with equilibrium buffer reactions, and the
finite-rate reaction of C02 in the mass
transfer boundary layer. With no allow-
ance for surface reaction kinetics, this
model accurately predicts limestone dis-
solution rates over the entire range of
experimental data, from pH 4 to 7, from
1 to 100 fjm particle size, with nine
sources of CaCOs. Particle size distribu-
tion is shown to be the primary factor
determining reactivity of naturally occur-
ring limestone.
The results of this project have also
been reported in theses and papers on
effects of solution composition and on
effects of type and grind.
Mass Transfer Model
Several computer programs were devel-
oped using mass transfer theory to pre-
dict the dissolution rate of limestone as
a function of particle size. The dissolu-
tion rate (cm3/sec) of a single spherical
particle of diameter dp (cm) and volume
Vp (cm3) is given by the mass transfer
expression:
dV
(1)
dt p ™
where k|_ = mass transfer coefficient
( £/cm2-sec),
AC = effective concentration driv-
ing force (M), and
Pm = molar density of limestone
(0.0271 gmol/cm3).
For particles from 1 to 100 urn, the
mass transfer coefficient was taken to
be the sum of one term representing
diffusion in stagnant solution and one
term representing the effect of agitation.
The rate was correlated in terms of the
constants K(cm2/sec) and B(cnrr1):
dVp
dt = — K(Vp1/3 + BVn2/3). (2)
The constant K was found to be 1.5
times its theoretical value:
1.52(67T2)1/3DAC
K = - (3)
Pm -1000
where D = effective diffusivity (cm2/sec).
This empirical adjustment of 1.5 is prob-
ably a correction for nonspherical shape
of the particles.
CaCOs mass transfer is enhanced by
acid/base reactions. Therefore, the effec-
tive DAC was calculated by numerical
solution of equilibria and material bal-
ances in the boundary layer for the spe-
cies H+. OH', Ca++, COJ, HCOs, SOs,
HSO^, A", HA~, and H2A (adipic acid
or other buffer). In the simpliest case, it
was assumed that HCOs does not react
with H+ in the boundary layer.
This theoretical calculation of K as-
sumes that the CaCOs solid is in equi-
librium with the solution at its surface.
The reaction of CaCOs so^ w'tn aque-
ous solution is assumed to be instan-
taneous.
The constant B should be independent
of particle size and was given by:
B =
O-167
u j -2/3
D
where e = agitation power (cm2/sec3),
u — kinematic viscosity of solution
(cm2/sec), and
p— density of solution (g/cm3).
The adjustable constant, 0.167, based
on data from this project was within 50
percent of that predicted by literature
correlations. At typical levels of agitation
with mass transfer controlled by H+ dif-
fusion the value of B was 400 cnr1 .
Eq. (2) gives the dissolution rate of a
single particle of size Vp. In order to
model dissolution rates in a batch reac-
tor at constant solution composition, Eq.
(2) was integrated to give the fraction
of undissolved CaCOs as an implicit
function of time. The total fraction
CaCOs remaining was obtained by sum-
ming over the fraction remaining in each
size fraction of the initial particle size
distribution obtained by a Coulter Coun-
ter. Predictions of limestone utilization
in a scrubber system were made by an
additional integration over a stirred tank
residence time distribution. These calcu-
lations were implemented on a com-
puter.
Experimental Methods
The absolute dissolution rates of 9
limestone types and 19 grinds were
successfully measured by the pH-stat
method. A batch of 0.5 g limestone
sample was dissolved in 1 liter of agi-
tated solution at constant pH. The reac-
tor was sparged with N2 or C02 to
maintain constant dissolved C02 and
constant dissolution stoichiometry. The
pH was controlled at values from 4.0 to
7.0 by titration with hydrochloric acid.
The cumulative dissolution of CaCOs
was obtained as a function of time di-
rectly from the titration volume of HCI
by the stoichiometry:
CaC03(s) + 2HCI - Ca++
+ CO2(g) + 2 CI~+H2O.
Constant concentrations of other com-
ponents were obtained by initially add-
ing CaCl2, Na2SO4, Na2S03, organic
acids, and other soluble salts. This pro-
cedure was precisely performed with an
automatic digital pH titrimeter. However,
satisfactory results were also obtained
with manual pH control by titration from
a burette.
Particle size distributions from 0.7 to
160 }im were obtained with a Coulter
Counter. These data were necessary to
use the mass transfer model, but not to
determine reactivity of the samples. A
method was developed to use a two-
parameter log gamma size distribution in
the model, so that simple techniques
such as screening could be used with
the mass transfer model.
Results
The mass transfer model correlated all
of the measured effects of solution com-
position and limestone type and grind
on the CaCOs dissolution rate. The two
adjustable constants, K and B, were
found to be about 50 percent greater
than the theoretical or predicted values. If
surface reaction kinetics were significant,
K and B would be less than their predicted
values. Since mass transfer is controlling,
the primary effect of limestone type and
grind results from the particle size distri-
bution of the ground stone.
Effects of Solution Composi-
tion
Figure 1 illustrates typical results ob-
tained with reagent CaCOs in 0.1 M
CaCl2 solution. The rate constant, k,
was obtained from experimental data and
from the model by neglecting the second
term of Eq. (2). Because of the small ef-
fect of agitation, it is equal to 1.25 times the
constant K. The curves in Figure 1 were
calculated by the mass transfer
model.
The rate data show the strong effect of
pH. With no dissolved sulfite and N2
sparging at 25°C, the dissolution rate
increases from 1.4 x 10"10 cm2/sec at
pH 6.0 to 2 x 10'9 cm2/sec at pH
4.5. CaCOs dissolves faster at lower pH
because there is a proportionately larger
driving force for H+ diffusion from bulk
solution to the limestone surface.
Solution components that buffer be-
tween the pH of the bulk solution and the
pH of the limestone surface (typically 5.5
to 8.0) enhance the effective H+ transport.
Figure 1 shows that at pH 5.0, 1 mM of
SOsYHSOg" buffer enhances the dissolu-
tion rate by a factor of 3.0. Similar but less
dramatic effects were obtained with ace-
tate, adipate, and other organic acid buf-
fers.
-------
The equilibrium at the CaCOs solid/solu-
tion interface is modified by the adsorption
of ionic species on the limestone surface.
Figure 1 shows that higher concentrations
of SOj/HSOs inhibit CaCOs dissolution.
This effect was successfully modeled by
assuming that the solubility of CaCOs is
depressed by the presence of dissolved
CaSOa at the CaCOs surface. The inhibit-
ing effect of sulfite depends primarily on
the CaSOs saturation in the solution. At
extreme values of CaSOs saturation,
CaCOs dissolution stopped completely,
probably because of irreversible crys-
tallization of CaSOs on tne CaCOs surface.
Similar inhibiting effects were measured
with Mn++, Fe++, and polyacrylic acid.
Dissolved C02 inhibits CaCOs dissolu-
tion when the bulk solution is nearly
saturated to CaC03_ However, at lower
pH, dissolved CO2 can enhance the dis-
solution rate by acting as a buffer (CO2/
HCOs)to carrV acidity. Unlike most other
buffers, C02 reacts with I-^O at a finite
rate to produce H+ and HCO^. Therefore
the contribution of CO2 is important only
with large particles (>50/jm) under condi-
tions where other buffers, such as SOT/
HSO3~, are not present. The effect of the
CO2 reaction was observed as an empir-
ical adjustment in the constant B when
using CC>2 sparging.
Effects of Type and Grind
The effect of type and grind was deter-
mined by 31 experiments with 9 lime-
stone types and 19 different particle size
distributions. Experiments were per-
formed at 25°C in 0.1 M CaCIo at pH 4 or
5 with sparging by N2 or C02-The results
are presented in Table 1 as the time requir-
ed to dissolve 50 percent (150) or 80 per-
cent (tgo) °f the CaCOs in the sample.
The data were correlated by Eq. (2) with
integration over time and summation over
particle size. With N 2 sparging the value of
B was found to be 400 cnrr1. Because
C02 enhances or inhibits dissolution more
with coarse particles, the value of B with
C02 sparging was found to be 880 cm"1
at pH 5 and 260 cnrr1 at pH 4. Table 1
shows that the predicted values of 150
agree with measured values within a stan-
dard deviation of less than 30 percent. This
close agreement is maintained for samples
with values of 150 varying over two orders
of magnitude.
For limestone sources of reasonable puri-
ty (85 percent CaCOs), the particle size
distribution of the ground sample is the
primary factor determining reactivity, rather
than the limestone type or composition.
With lower purity or greater than 90 per-
10'
1
10'
10 15
Total Sulfite, mM
20
25
30
Figure 1.
Effect of sulfite, NX sparging, 25 and 55°C. (Curves calculated from transfer model
using CaCO°3 /CaSO 3 solid solution.)
cent utilization, the dissolution of dolomite
or other impurities may have more pro-
nounced effects on the rate of CaCOs
dissolution.
Experiments with pure dolomite (Mg
COs-CaCOs) established that it dissolves
3 to 10 times slower than calcite (CaCOs).
Therefore, the dissolution rate of dolomite
is controlled by surface reaction kinetics
rather than mass transfer. If dolomite is
present in a limestone sample, it will dis-
solve slower than calcite, but it will still dis-
solve with some release of Mg++ into
the solution.
Design Implications
In terms of limestone dissolution, a slurry
scrubbing system can be defined as a con-
tinuous stirred tank reactor (CSTR) with a
residence time, r, equal to the molar ratio
of calcium solids inventory and CaCOs
feedrate. Given a particle size distribution,
the computerized mass transfer model can
calculate a relationship of stoichiometric
ratio (SR) and relative reactivity (1 /KrSR),
as shown in Figure 2 for six limestone
grinds. The constant K can be determined
experimentally or by the computer model
as a function of solution composition,
shown in Figure 1.
In a typical scrubber hold tank, the lime-
stone has a residence time, r, of 10 hours.
In the absence of equilibrium limitations.
Figure 2 shows that K can easily obtain a
value approaching 10"9 cm2/sec at pH
5.5 with dissolved sulfite.With a stoichi-
ometric ratio near 1.0, these values of K
and T give relative reactivity of 3 x 10"*
pm"2, suggesting an actual stoichiometry
from Figure 2 of 1.05 to 1.1 depending on
the limestone grind. In practice, stoichi-
ometric ratios of 1.2 to 1.5 give pH values
of 5.5 to 6.0. Therefore, hold tanks may
operate near equilibrium with respect to
CaCO3 or the CaCO3-CaS03 adsorption
layer. If the hold tank is operating at near
equilibrium, its volume will have only a
small effect on system performance.
In a typical scrubber, the limestone resi-
dence time is as much as 100 times less
than in the hold tank, for example 0.1 hours.
However, the pH is typically much lower,
giving a faster rate of CaCOs dissolution.
At pH 4.5 with dissolved sulfite and in the
-------
Table 1. Summary of Measured and Predicted Values of Time Required to Dissolve 50% (tKn) and 80% (ton) of the Initial
Available CaCO3 °
Limestone Type
Grind (um) ^90/^50 pH Sparge meas(min) meas/calc meas(min) meas/calc
Ash Grove
%A = 96.7 a
Brassfield
%A = 84.0
Fredonia
%A = 95.0
Georgia Marble
%A = 96.5
Longview
%A = 95.0
Maysville
%A = 98.4
Pfizer
%A = 98.7
Reagent Grade
%A = 100
Stoneman
%A = 86.0
-325
120-200
Coarse
Extra Coarse
Feedbelt
Fine
100-140
120-200
170-270
200-325
Coarse
Fine
Coarse
Fine
-325
120-200
200-325
Coarse
Fine
12.7
100
38.1
40.3
8.0
130
125
66
66.4
42.0
24.0
17.9
15.0
8.3
19.8
21.8
11.4
40.3
22.0
2.30
1.35
3.73
4.02
8.51
1.20
1.24
1.36
1.95
2.86
2.06
8.60
3.63
4.31
2.82
2.34
1.35
2.90
2.30
5
4
4
4
5
5
4
4
5
5
4
4
4
5
5
4
4
5
5
4
5
4
4
4
4
4
5
5
4
5
5
CC-2
CC-2
CC-2
C02
A/2
A/2
C02
A/2
C02
A/2
C02
CC-2
A/2
CC-2
A/2
C02
CC-2
CC-2
A/2
C02
CC-2
CC-2
CC-2
CO2
CC-2
A/2
C02
A/2
CC-2
CC-2
A/2
7.88
39.00
6.95
6.10
3.910
6.1
54.2
40.0
2080
494°
63.0
14.6
13.9
81.5
145
22.0
11.9b
47.0
27.0C
2.800
7.88b
0.6750
3.7 Qd
4.60d
1.26
1.17
8.39
10.7
6.9C
14.9
20.3C
0.99
0.98
1.16
0.91
0.92d
1.37
0.93
0.95
1.05
1.03
1.65^
0.71
0.87
1.01
1.00
0.9Qd
1.34
1.17
0.95
1.31
0.87
1.01
1.48
1.63
1.07
1.07
1.09
1.08
0.90
1.01
0.94
32.6
85.1
57.1
38.7
60.2
72.8
107
893
131
44.1
48.90
108
24.9
3.54
20.3
23.6
81.4
1.17
1.08
1.33
0.79
1.32
0.98
0.94
1.18
1.68
0.87
1.34
1.21
0.980
0.794
1.92
2.02
1.34
a . %A = percent available for dissolution (wt % CaCC-3 +
° - average value from several runs.
c - obtained in early apparatus with magnetic stirrer.
d - 700 rpm.
absence of equilibrium limitations. Figure
1 shows that K can easily be as large as 5 x
10'9 cm2/sec.The combination of large K
and small r still gives a relative reactivity
approaching 4x10"^ ^im"^ and a stoichi-
ometry of 1.5 to 2.0 with fine grind lime-
stones. Therefore, it is conceivable that a
significant fraction of the limestone could
dissolve in the scrubber rather than the
hold tank.
Conclusions
1. The pH-stat method is effective for de-
termining absolute and relative reactivity
of limestone samples.
2. CaCOs dissolution is controlled by dif-
fusion of H+, OH", and buffer species, not
by surface reaction kinetics.
3. Dissolution rates of relatively pure lime-
stones do not depend on limestone type or
source, but on particle size distribution.
4. Particle size distributions can be mea-
sured by a Coulter Counter. Other methods
can be used to give approximate results.
5. Sulfite/bisulfite, adipic acid, and other
buffers enhance limestone dissolution.
6. Sulfite, Fe++, Mn++, Mg++, and poly-
acrylic acid inhibit CaCOs dissolution,
possibly by reducing the effective solubil-
ity of CaCOs.
7. The dissolution rate of limestone is rel-
atively fast; therefore, a significant fraction
of limestone can dissolve in the scrubber
itself, and solution in the hold tank will be
near equilibrium with the CaC03 solids.
Recommendations
1 . Dissolution rates should be measured
with 1 to 1 5 percent solids concentration.
2. Diffusivities of H+, OH', HCOsf, SOJ,
and Ca++ should be measured in solutions
of 0.1 to 2.0 M CaCI2, MgSO^ and
3. Careful measurements of CaCOs dis-
solution rates near equilibrium pH are
4
-------
10'
85
K
t 10'
10"
1.0
1.2
2.2
2.4
1.4 1.6 1.8 2.0
Stoichiometric Ratio ISR)
Figure 2. CSTR mass transfer model dissolution rates for various limestone types and grinds.
needed to quantify effects of SOs, metal
ions, and Mg"*"1" on the effective solubility
of CaCOs-
4. Careful measurements of limestone re-
activity at high utilization are needed to
establish any second order effects of lime-
stone type and impurities.
Gary T. Rochelle. PuiK. R. Chan, and Anthony T. Toprac are with the University of
Texas, Austin. TX 78712. J. David Mobley is the EPA Project Officer (see
below).
The complete report, entitled "Limestone Dissolution in Flue Gas Desulfurization
Processes, "(Order No. PB 83-252 833; Cost: $13.00, subject to change) will be
available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
U.S. GOVERNMENT PRINTING OFFICE: 1983-759-102/0792*
------- |