You are on page 1of 12

View Article Online / Journal Homepage / Table of Contents for this issue

Reaction between gaseous sulfur dioxide and solid calcium oxide


Mechanism and kinetics

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

Deborah Allen and Allan N. Hayhurst*


Department of Chemical Engineering, University of Cambridge, Pembroke Street, Cambridge, U K
CB23RA

Sulfur dioxide has been reacted with tiny (diameter 35-85 pm) particles of porous CaO in a thermogravimetric balance. Kinetic
measurements, together with infrared studies of the products, indicate that two reactions, viz: CaO + SO, CaSO, and CaO
SO, + jCaS0, + aCaS occur. In the presence of both SO, and 0, , when CaSO, and CaS both oxidise to CaSO,, these two
reactions of SO, provide the rate-determining steps for the initial stages of reaction for such tiny particles. Thus, the conditions
used in this study succeed in preventing the rates of either of these reactions being controlled by the diffusion of SO,. The rates of
both reactions per unit surface area of CaO have the general form: k([SO,] - [SO,],), where [SO,], is the concentration of SO,
for the reaction concerned being at equilibrium and k is the rate constant. Measurements of initial rates indicate that
k = 7.2 x
exp(-1443/T) m s - l and 1.2 x
exp(-481/T) m s-l for the two reactions, respectively, correct to 25%.
However, CaSO, is unstable above ca. 1123 K, so that only the slower (second) reaction occurs above ca. 1123 K. That the rates
of these reactions are independent of the concentration of oxygen confirms that SO, plays no part in them. Measurements for
large extents of reaction showed that the diffusion coefficient of SO, through the solid products of reaction (mainly CaSO,) was
D, = 1.9 & 0.5 x 10- l4 m2 s- '. However, at high temperatures (> 1160 K) when only CaSO, and CaS are produced, D , falls to
2.3 x lo-'' m2 s-'.
--f

Sulfur dioxide, along with sulfur trioxide, is a major pollutant


of the atmosphere. It is produced e.g. whenever coal burns in
excess oxygen. One way of minimising emissions of these
oxides of sulfur is to absorb them on solid calcium oxide, produced by calcining limestone, i.e. in CaCO, + CaO + CO, .
The porous solid, CaO, can then react with SO, in the presence of oxygen (e.g. in a fluidised-bed coal combustor) to give
calcium sulfate, via
CaO

+ SO, + 4 0 2 = C a S 0 ,

(2)

for which AH' = -220.8 kJ mol-' and AGO = -46.4 kJ


mol- at 1000 K. In the presence of 0, , subsequent oxidation
of the sulfite to sulfate can occur in:

'

CaSO,

+ $0,

CaSO,

(3)

with AH' = -268.1 kJ mol-' and AGO = -178.4 kJ mol-'


at lo00 K. There are thus no thermodynamic constraints on
these exothermic reactions. The second mechanism has oxidation of SO, to SO, as the initial step in
SO,

+ $O,*SO,

(4)

followed by production of the sulfate in


CaO + S O , e C a S O ,

CaSO,

(1)

but the actual mechanism is still the subject of disagreement.


This arises from there being at least two mechanisms by which
sulfation can take place. The first possible route involves a
direct reaction between CaO and SO, to give CaSO, :
CaO + SO, e C ~ S O ,

extent of sulfation at a temperature ca. 1100 K. However,


opinion differs as to the correct interpretation of the observed
temperature dependence of sulfur capture. Lyngfelt and
Leckner, claim that this temperature dependence is not seen
in laboratory experiments and must therefore be associated
with conditions in a fluidised bed. They followed Jonke et a/.'
and Fieldes6 in proposing a 'reduction theory', which suggests
that CaSO, can be reduced by C O

(5)

It is unlikely that either of these two schemes fully describes


the situation; thus other routes are discussed below, e.g. one
involving CaS as an intermediate, which is later oxidised to
CaSO,. The relative likelihood of each pathway will now be
discussed, in conjunction with a review of previous work.
It is generally agreed'., that the reaction between CaO and
SO,, in the presence of 0, , is first order with respect to SO,.
There is also c o n ~ e n s u son
~ *there
~
being a maximum in the

+ CO e CaO + SO, + CO,

(6)

The proposal is that the extent of the regeneration of CaO, via


reaction (6), increases with increasing temperature, thus effectively reversing the sulfation reaction (1). An alternative
explanation, of the temperature dependence of sulfur capture
is the 'oxygen depletion' idea, whereby combustion of volatiles in a fluidised bed burning coal creates a depletion of 0,
in the particulate phase at higher temperatures and in the
absence of O,, SO, reacts only slowly with CaO. This explanation implies that the rate of sulfation becomes dependent
upon the concentration of 0, in the bed, which at higher concentrations of 0, is no longer true, as discussed below; in fact
the order of the overall reaction (1) is normally zero with
respect to ~ x y g e n .Other
~
workers2**have observed a temperature dependence in sulfur capture when using a thermogravimetric balance or a fixed bed, indicating that the
maximum in the uptake of sulfur arises from the fundamental
nature of the reaction between CaO and SO, and does not
occur only within a fluidised bed. In accordance with this,
other theories include those of 'structure impairment '. These
either propose changes'.'' in the structure of the limestone or
invoke rapid blocking of the pores when the temperature
Both these explanations have been shown to be
increases.'
~ n l i k e l y . ~The
~ ' ~main evidence is that higher temperatures
during calcination result in a more porous structure in the
product, i.e. a more suitable pore-size distribution for sulfur
uptake and a larger surface area over which reaction can take
place. Lyngfelt and Leckner l4 also showed that decreasing the
temperature, from above 1100 K, results in an increase in rate.
This would not be expected if permanent inactivation of the
'9''

J . Chem. Soc., Faraday Trans., 1996,92(7), 1227-1238

1227

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

View Article Online

limestone, arising from rapid pore blocking, had taken place


at high temperature.
Another explanation for the temperature dependence of
sulfur capture is the reaction rate theory," which interprets
the phenomenon as a switch between two different reaction
pathways. Before further comment can be made on this
approach, it is necessary to understand the actual mechanism
of the reaction between CaO and SO,. First, the thermodynamics of reaction (2) should be considered. Values of the mole
fraction of SO, for reaction (2) at equilibrium have been calculated for the temperature range 800-1200 K and a total
pressure of 1 atm. It should be noted that SO, can only be
absorbed by CaO when the partial pressure of SO, exceeds
that for equilibrium at a particular temperature. Typical mole
fractions for SO, during coal combustion are 3 x lo-, to
1 x lod2 mol%. The thermodynamic calculations indicate
that the reverse of reaction (2), i.e. the decomposition of
CaSO, to CaO and SO,, will become significant above 1000
K. The thermodynamic instability of CaSO, above 1000 K
was noted by Dennis and Hayhurst,, who observed a decrease
in SO, retention by limestone above this temperature.
IR spectroscopy16 has shown that CaSO, is indeed a
product of the reaction between CaO and SO, below 923 K,
with possible conversion to CaSO, if the temperature
increases beyond the thermodynamic limits for CaSO, being
stable. Tarradellas and Bonnetain' concluded that CaSO,
decomposes at these temperatures ( T z 1000 K) to give
CaSO, and CaS via a succession of gaseous intermediates,
SO, and S, , in

'

2CaS0,

-+

+ 3s02
2CaO + $S,

6CaS0,

2 C a 0 + 2S0,
6CaS0,

-+ 2CaS

(7)

+ $S,

+ SO,

(9)

with a net reaction


8CaS0,

e 6CaS0,

+ 2CaS

Calcium sulfate and sulfide, but not CaSO,, were also


identified" as products at temperatures above 773 K. Thus,
there appears to be some route by which CaSO, and CaS can
be formed from CaO and SO, without CaSO, as an intermediate. The overall reaction is
4Ca0 + 4s0,

e 3CaS0,

+ CaS

(11)

for which AHo = - 1007.6 kJ mol-' and AGO = -295.5 kJ


mol-' of CaS at 1000 K. Indirect evidence for reaction (11)
has been produced recently:I9 mixtures of CaS and CaSO,
were found to give a
above 1100 K, accompanied by
the production of CaO and SO, in the reverse of (11). It is
thus expected that the forward reaction in (11) will proceed if
the partial pressure of SO, exceeds that for equilibrium at the
given temperature.,' Reaction (1 1) was also reported by
MOSS,^' who was studying the problem from the viewpoint of
the regeneration of limestone from sulfated calcium species.
Moss" concluded that there was a change of mechanism at
1100 K. He proposed that at temperatures below 1100 K, sulfation occurs mainly with the formation of CaSO, , as in reaction (2). At higher temperatures he suggested that sulfation
proceeds uia the slower reaction (9,with SO, as the intermediate species; this would give rise to almost a step change in
the observed rate at 1100 K. However, since the formation of
SO, uia reaction (4) requires oxygen to be present, the role of
SO, in the sulfation of CaO must now be considered.
In the presence of O , , it would be expected, and has been
shown,22that both CaS and CaSO, are oxidised to the thermodynamically stable sulfate. However, if oxygen is present
throughout the sulfation process, then there is also the possibility of reactions (4) and (5), i.e. the oxidation of SO, to SO,
followed by uptake of the SO, by Ca0.23*24
This is an alter1228

J . Chem. Soc., Faraday Trans., 1996, Vol. 92

native explanation' for the observed continuation of sulfation above a temperature of 1 1 0 0 K, where CaSO, is no
longer thermodynamically stable. In this instance involving
SO, as an intermediate, it would be expected that the rate of
reaction would show some dependence on the amount of 0,
present. Kinetic studies3v8indicate that there is no dependence
on O,, even down to zero concentrations of 0,. The experiments of Simons et aL7 suggested that there might be some
dependence with large 0, concentrations, but nevertheless the
effect of 0, on the rate would be small compared with the
effect of temperature.
To summarize, there are, at least in principle, three possible
routes by which sulfation of CaO can occur. These are: (a) by
the initial formation of the sulfite [i.e. reaction (2) followed by
reaction (3)]; (b) by a gas-phase reaction to form SO,, [i.e.
reaction (4) followed by reaction ( 5 ) ] ; and finally (c), via CaS
in reaction ( l l ) , followed by the oxidation of CaS to
CaS0,.25 In addition, there is the apparent anomaly of a
decrease in the extent of sulfur capture at temperatures greater
than 1100 K. Infrared spectroscopy, kinetic and thermodynamic considerations indicate that CaSO, is the major intermediate of the reaction at temperatures below 1 1 0 0 K and
with SO, concentrations typical of those in coal combustion.
At temperatures above ca. 1100 K, where CaSO, is thermodynamically unstable, a change of mechanism can be postulated to one of the other two alternatives. This is also used to
explain the decrease in SO, uptake at the higher temperatures.' This work thus involves a thermodynamic and
kinetic study of the reactions between CaO and SO,, both in
the absence and presence of 0, , in an attempt to elucidate the
major aspects of the reaction. The kinetics of the reactions
operating will be discussed in some detail.

Experimental
A thermogravimetric balance was used to study the reaction
between CaO and a gas mixture containing e.g. SO, and N,
initially in the absence of 0,. Such a thermogravimetric
balance continuously measures the mass of a solid sample,
while it undergoes reaction with a gas, at increasing times.
The microbalance (C. I. Robal) is shown in Fig. 1. It has a
digital output, recorded at 1 min intervals on a computer.
From one end of the balance is suspended a quartz hangwire
(length 300 mm), which carries a small quartz bucket [internal
diameter (id) 4 mm], in which the solid sample (CaO) is
placed. The whole is surrounded by a quartz tube as shown in
Fig. 1, and is situated within a Stanton Redcroft electric
furnace. One aspect of the quartz tube (id 40 mm) which
should be noted is that the reacting gases enter at a point
approximately 200 mm above the quartz bucket and are then
directed down an internal tube to a position adjacent to the
sample. Fig. 1 also shows a small quartz shelf positioned just
above the quartz bucket to ensure that the reacting gas must
pass over the sample before escaping up the tube to the
exhaust. The temperature of the furnace is controlled thermostatically, and to gain an accurate measurement of the temperature near the sample, a type K chromel-alumel
thermocouple was placed as shown in Fig. 1. The thermocouple indicated that the gas entering the quartz tube has been
heated to the temperature of the furnace by the time it
reached the sample. Typically, the furnace was operated
within the range 873-1 173 K.
The balance was calibrated using a 100 or 10 mg weight. In
a typical experiment, ca. 20 mg of solid CaO (AnalaR from
BDH; see below for more details) were weighed into the
quartz bucket, which contained a small amount of quartz
wool. The effect of the quartz wool is to separate the particles
of CaO and also to distribute them throughout the bucket;
the significance of this experimental detail is discussed below.
Nitrogen was continuously passed over the top of the balance

View Article Online


balance head

quartz casing

Mercury porosimetry

I
-

thermocouple

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

counter balance

quartz bucket
filled with quartz wool

Fig. 1 Schematic diagram of the thermogravimetric balance (not to


scale) with its weighing bucket filled with quartz wool to separate the
particles of CaO in the sample

The porosity of these particles was determined, using a Carlo


Erba (Model 220) mercury porosimeter, to be 0.38 & 0.03 in
the following way. The output is a measured cumulative pore
volume plotted against pore diameter, as shown in Fig. 2. In
fact, the total pore volume per gramme, Vp, was measured to
be 190 f 18 mm3 g-' and the density of the solid, p s , has a
valuez7 of 3.3 x
g mmP3. The porosity, E, can be
expressed as the product of the bulk density, p, (i.e. the density
inclusive of both solids and voids) and Vp, so E = pVp. The
bulk density, however, is p = (1 - ~ ) p , so
, E becomes E = Vp(1
- &)ps,which gives E = 0.38 & 0.03. This now allows a value
for the bulk density to be calculated. Solid, non-porous CaO
has a density of ps = 3.3 x lop3 g rnrn-,; therefore a sample
with porosity of 0.38 will have a bulk density of
(1 - 0.38) x 3.3 x lo3 = 2.2 x lo-, g rnrn-,. The implication
of a porosity as high as 0.38 is that there is probably a large
internal surface area, so that reaction within the pores is
likely. This is important, because if reaction can take place in
the pores, there is considerably more surface area available for
reaction than if it were to occur only on the external surface of
a particle (the BET analysis gave the ratio of internal to external surface area as 397). Also, it would require that the nature
of diffusion within the pores be characterised. Both of these
points are considered below.

Results
as a purge gas, and also over the sample, until constant mass
and temperature had been achieved. The decrease in mass
(typically ca. 5 mg) observed while heating to the required
temperature is presumably due to the thermal decomposition
of hydrated Ca(OH), or CaCO, to yield CaO. The flow of N,
over the sample was then stopped, the reactant gas was introduced over the sample and the change in mass of the solid
recorded at 1 min intervals. The error in a measurement of the
mass of the reacting solid was k0.3 mg. An experiment
usually lasted 60 min, at the end of which the flow of reacting
gas was stopped and the sample was cooled under nitrogen.
The sample was then usually analysed by IR spectroscopy.
Gases were supplied from cylinders often containing known
mixtures of e.g. SO, in N,. Their flow rates into the thermogravimetric balance were controlled by needle valves and
rotameters. The total flow rate was maintained at 2 ml s- ; it
was established that the mass reading was not sensitive to the
flow rate of gas.
IR spectroscopy of the solid product was carried out using
a Perkin-Elmer 882 spectrometer, with the sample pressed
into a KBr disc. This produced spectra in the range 400-4000
cm - covering the characteristic absorption frequencies of
most polyatomic inorganic compounds. Reference spectra
were obtained directly for CaO, CaSO, and CaS; for CaSO, ,
CaS,O, , CaS,O, and CaS,O, characteristic absorption frequencies were taken from Miller and Wilkins.26

Experimental characterisation of calcium oxide


particles

It was mentioned above that quartz wool was used to distribute the solid within the quartz bucket of the thermogravimetric balance. It was found that when using quartz wool, a
linear increase in mass of the sample with time was obtained
for the initial stages of the reaction. This is illustrated in Fig.
3, which gives the mass of the sample us. time for which the
sample was exposed to SO,. Fig. 3 has two plots: one without
quartz wool and the other with quartz wool used to separate
each particle of CaO. Clearly Fig. 3 shows that quartz wool
causes a linear mass us. time plot. This linearity suggests that
distributing the particles in this way removes any interparticle

Q)

BET surface analysis

AnalaR CaO particles (from BDH Chemicals) were used and


for these experiments were sieved to have diameters of 75 to
94 pm, Using a Micrometrics ASAP 2000 BET instrument,
with N, as the analysis gas, these particles were found (after
several determinations) to have a BET surface area of
14.3 & 0.6 m2 g-I. Further BET measurements also gave
surface areas in pores of different diameters; these results are
presented below.

pore diameterhm

Fig. 2 Cumulative pore volume, as measured by mercury porosimetry, us. pore diameter for an unreacted sample of pure calcium
oxide

J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

1229

View Article Online


2.4

with quartz
wool

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

no quartz

"V
I

wool

10

20

0.9

30

time / min
Fig. 3 Mass of sample at increasing times, when 1 vol.% of SO, in
N, was contacted with CaO at 1073 K. Plots are presented for the
particles of CaO distributed over quartz wool (0)
and without quartz
wool (0).

resistance to diffusion of SO, and furthermore that the formation of product on the surface of the CaO does not introduce
a resistance to reaction over this range of time. This in turn
implies that the depth of the product layer for this extent of
reaction is small. This is a probable result for the particles
seen above to have a porosity as high as 0.38 and a large ratio
of internal to external surface area of 397.
That reaction takes place even in the smallest pores, as proposed above, was verified by Fig. 4. This gives the cumulative
pore area per gramme of solid us. pore diameter, as determined from BET measurements. The samples were unreacted
CaO, and CaO reacted with SO, (0.27 vol.% in N,) for times
of 15, 30 and 45 min. If reaction is occurring in even the smallest pores, it would be expected that the area within these
pores would be reduced. This can be seen to be so from Fig. 4,
since the amount of surface area due to small pores decreases
with increasing times of exposure to SO,. The trend of less
area in small pores after increasing reaction times, even up to
45 min, also shows that few of the large pores have been
reduced substantially in size over this time. The conclusion is
that reaction takes place within even the smallest pores, so
that the entire internal surface area is available, without significant resistance to diffusion of SO, in a pore. This, considered with the linearity of Fig. 3, suggests that reaction at
the solid surface will probably be the rate-determining step,

0.41

30

40

50

60

80

70

90

average particle diameter /pn

Fig. 5 Normalised value of the initial rate of reaction as measured


for differently sized particles of CaO reacting with sulfur dioxide (0.27
vol.% in N,) at temperatures: (H) 973; (0)
1073; and (A)1173 K;
error bars correspond to uncertainties of 20%

i.e. the reaction will be kinetically controlled. This conclusion


gains further support from experiments in which the effect of
particle size on the rate of reaction was determined. Calcium
oxide particles were sieved into four different size ranges
(<53, 53-57, 57-74 and 74-95 pm) and a sample from each
range was reacted in the thermogravimetric balance with SO,
(0.27 vol.% in N,) at 1023 K. The plots of mass against time
are all linear, as expected from Fig. 3. Their slopes (dmldt)
divided by the initial mass (mo)are plotted against mean particle diameter in Fig. 5, from which it is clear that there is no
apparent dependence on particle size. This result can be more
readily interpreted by considering Table 1, which gives the
surface area (BET with N,) of the particles within each size
range and shows that for these particle sizes there is very little
variation in BET surface area. Table 1 also gives the external
surface area per gramme of solid, calculated assuming the particles to be spherical, of diameter d and density, 2.0 x lo3 kg
m-3.
Since the external surface area does vary with size, it can be
concluded that for the normalised reaction rate in Fig. 5 to
show no dependence on particle diameter over these size
ranges, the reaction must be taking place over the total BET
surface area of the particle, which is also invariant from Table
1. It must now be shown that within the pores no significant
concentration gradients are initiated, so that the ratedetermining step can be considered to be a chemical reaction

Table 1 Surface area measurements for different size ranges of


calcium oxide particles

20

40

60

pore diarneterhrn

Fig. 4 Cumulative pore area (measured by BET) us. pore diameter


for CaO reacted with sulfur dioxide (0.27 vol.% in N,) for: (0)
0, ( 0 )
15, (m) 30, and (0)
45 min

1230

J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

CaO particle
diameter
range/pm

BET surface
area/m g - '

average particle
diameter

IW

external surface
area per gramme
of solid/m2 g-'

c 53
53-57

13.1 f 0.6
12.8 f 0.6

55
66

0.055
0.046

57-74
74-95

14.3 f 0.6
15.8 +_ 0.6

85
99

0.036
0.030

View Article Online

at the solid surface of the pores. Note that Fig. 5 suggests the
rate of reaction decreases with temperature; this is discussed
further below.

section, the problem will now be approached from a theoretical viewpoint.

External mass transfer and diffusion coefficients

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

Diffusion and reaction within a pore


It will be demonstrated that even within the smallest pores of
the CaO, the concentration of SO, is in effect uniform, i.e. the
reaction has no significant resistance from diffusion within the
pores. In this case the findings of the previous section can be
interpreted as implying kinetic control. If the pore diameter is
greater than the mean free path of the SO, molecules, their
diffusion will be similar to that in the gas phase (i.e. molecular
diffusion) adjacent to the exterior of a CaO particle. Thus it is
important to consider the case when the pore size is less than
the mean free path, so there is Knudsen diffusion, determined
by collisions with the walls of the pore. For diffusion into a
cylindrical pore (closed at one end), of length L and radius r,
an expression can be derived for the Thiele modulus, 4, i.e. the
ratio of the maximum rates of surface reaction and diffusion of
gas into the pore as
$2

2nrLkC -2L2k
=
nr2D(C/L) - rD

where C is the concentration of gas at the mouth of the pore,


D is the effective diffusion coefficient of the gas into the pore
and k is the rate constant per unit surface area for the reaction
between the gas the the solid surface (assumed to be first
order). For 4 < 1, the concentration everywhere is C, i.e. the
effectiveness factor (see below), q = 1 and the reaction is
kinetically controlled. When 4 > 1 , large concentration gradients are set up within the pore and q < 1. So for the situation when a pore is beginning to generate concentration
gradients within it, 4 = 1 , so that 2L2k = rD. Combining this
with the expression28 for the Knudsen diffusion coefficient Dk
for a small pore :

-"(")"'

k-3

7tM

Fig. 3 can now be discussed in terms of a single isolated spherical particle of CaO reacting with SO,, in the absence of 0, ,
in the forward step of reaction (2) only. It will be assumed that
all the SO, reacts inside the porous CaO particle, because Fig.
3 and 4 suggest that the large total internal surface area of the
particle is available for reaction. In addition, the reaction will
be taken to be first order in SO,, and it will be assumed that
a layer of CaSO, of thickness 6 has built up. In this case,
where the ratio of internal to external surface area is much
greater than unity, the rate of reaction (kmol s - ') of SO, per
particle is
nd2C

where d is the diameter of a particle of CaO and S (m2 kg- ')


is its BET area; C is the concentration of SO, in the inlet gas,
k , is the rate constant (in m s-') of reaction (2) per unit
surface area of CaO, D, is the diffusivity of SO, through the
solid product CaSO, and k, is the mass-transfer coefficient (m
s-') for diffusion of SO, from the bulk of the gas to the
exterior of the particle. The terms in the denominator of eqn.
(11) represent, respectively, the resistances from diffusion and
reaction within the pores, and diffusion through the product
layer and external mass transfer. The effectiveness factor, q, is
the ratio of the actual reaction rate within a pore to the rate of
reaction if diffusional effects are negligible inside the pore.
Thus q = 1 describes a situation where external diffusional
resistance is probably small, all the internal surface area of the
particle is available for reaction and the reaction is kinetically
controlled. More quantitatively, q is characterised by the
Thiele modulus, 4, which for a first-order reaction in a sphere
is given by

gives

(x)

2r2 8 R T
2L2k = 3

'I2

with R , T and M being, respectively, the ideal gas constant,


the temperature at which reaction is taking place and the
molar mass of the diffusing gas. Now consider the length of a
pore to be half the diameter, d , of a CaO particle, so
d2k

2r2 8 R T

'I2

T=&)
For a particle of diameter 85 pm (i.e. a large one in Fig. 5),
reacting with SO, at 1073 K, the rate constant is measured
(see below) to be 1.2 & 0.3 x lop6 m s-'. Substituting these
values into eqn. (I) yields a value for r of 3.6 nm. This can be
compared with a mean free path for SO, at this temperature
of 164 nm and thus confirms the initial assumption of
Knudsen diffusion. The significance of this quantitative treatment is that for all pores of radius greater than 3.6 nm, there
will be no concentration gradient within the pore, i.e. 4 < 1
and the reaction will be kinetically controlled. From Fig. 2 it
can be calculated that this will describe the situation for over
96.5% of the volume of the pores. For the smaller particles of
Fig. 4 reaction will occur in even narrower pores. Obviously it
is extremely important to determine whether a reaction is
kinetically or diffusion controlled if the correct interpretation
of the experimentally measured rates is to be made. To
support the experimental findings of this and the previous

where D, is the effective diffusivity of SO, within the pores of


the particle of CaO. For q = 1.0, Levenspie128 shows that
4 < 0.4 is a required condition. The CaO particles used in this
work are so small that it is probable that 4 < 0.4, so that
q = 1.0 and consequently the only significant term in eqn. (11)
will be that for reaction in the pores. It will be assumed for the
present that this is so and eqn. (11) becomes
r = ( 7nd) k , S p C

For an initial mass of particles m, this yields


dm- 64m, Sk, C
dt

with the factor of 64 arising from an increase of 64 kg kmol- ',


when CaO gives CaSO,. For this to occur, a linear plot as
shown in Fig. 3 is expected, provided S does not vary with
time; in fact, the constant slope corresponds to 64m, Sk, C . In
Fig. 3 the slope is 0.17 mg min-', which for m, = 21.1 mg,
yields a value for the fractional rate, i.e. (l/m,)(dm/dt) of
1.34 x
s-'.
With C = 0 . 1 2 mol rn-,
and
S = 14.3 & 0.6 x l o 3 m2kg- ', all this gives the first-order rate
constant k , = 1.2 f 0.3 x
m s-'. Substitution into eqn.
(111) now gives 4 = 0.07, i.e. much less than 0.4, thus confirming the initial supposition that q = 1.0. For this calculation D,
was taken2' to be D,E' and D, the diffusivity of SOz in N, , to
J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

1231

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

View Article Online

be 1.04 x
m2 s-' at 273 K (Ref. 30). To estimate D, at
was
higher temperatures, the relationship3' D, cc T1.75,
assumed. Values of E (porosity) = 0.38 and p = 2.0 x lo3 kg
rn-, were used. The conclusion is that SO, rapidly diffuses
into these particles and reacts in pores wider than the mean
free path of 164 nm. This is also true (i.e. 4 < 0.4), for
Knudsen diffusion in pores of radius greater than 1.0 nm with
the smallest (d = 35 pm) particles in Fig. 5. Of course, Fig. 4
shows that most of the internal area is initially in larger pores.
Thus the conclusion is that all the internal surface area of one
of these small particles of CaO is available for reaction; this
can also be checked by comparing the magnitudes of the three
resistances in the denominator of eqn. (11). The term, 6 / D , is
for diffusion of SO, through the product layer of CaSO, and
is difficult to quantify, because the mechanism by which SO,
diffuses through CaSO, is unknown. Bhatia and Perlmutter3'
estimated D, to be 6.9 x lo-', m2 s-' at 1253 K and noted
that this is comparable to values for ionic diffusion. Hajaligol
et a/.,,, however, report a higher value of lo-'' m2 s-',
although it should be noted that this was derived for the sulfation of limestone, i.e. CaCO, rather than CaO and there is
thus a two-way process of gas diffusion taking place: SO, in
and CO, out of a pore. Here, the linearity of Fig. 3 shows that
diffusion through the product layer is not presenting a significant resistance to the rate of reaction, i.e. 6 / D , must be negligible compared with l/k,. For the 25% increase in mass
shown in Fig. 3, 6 can be taken3, to be ca. 13 nm, which is
equivalent to almost 32 molecular layers. This requires (for
k, = 1.2 x
m s-') a value of D,,necessarily larger than
1.6 x 10- l4 m2 s-' for 6/D, < l/k, , as implied by experiment,
In fact, both of the above estimates of D, exceed 1.6 x lo-',
m s-l, again indicating purely kinetic control for the initial
stages of the reaction considered in Fig. 3. Finally, k, can be
estimated using a Sherwood number (= k, d/D,) of two. This
gives l/k, = d/2D, = 85 x 10-6/(2 x 1.14 x lo-,) = 0.4 s
m-', which is clearly much less than 6/Spdk, = 2020 s m-'.
This indicates that the rate-determining step is a reaction at
the internal surface of the particle.
The significance of this is that diffusional effects (externally,
inside the pores and also for transport through the product
layer), which are difficult to quantify, have been rendered negligible by distributing tiny particles in the manner described
over quartz wool and considering only initial rates of sulfation, as in Fig. 3. Furthermore, it has been shown that the
area over which reaction can take place is fairly accurately
described by the BET surface area. This allows a value of the
rate constant to be readily derived; this is carried out below.

IR spectroscopy
Fig. 6 shows the IR spectra of the products of the reaction
between CaO and SO, (0.27 vol.% in N,) for the temperatures
873, 973, 1073 and 1773 K. The spectra were taken over a
wavelength range 0-4000 cm-', but only the absorptions in
the range 400-1600 cm-' are shown in Fig. 6, because this is
the region in which characteristic absorption of calcium
species occurs. Below 1123 K, it can be seen that the following
anions are present: 0,-, SO,,-, SO,,- and S2-. The first
three show strong absorptions, with S2- present as a weak
band only. Above 1123 K, 0,-, SO,,- and S2- are still
found, but the S2- band has become stronger. However, the
SO, - absorption is no longer present. These spectroscopic
observations agree with the thermodynamic information
above, suggesting that CaSO, is a product of the reaction
between CaO and SO, below a temperature of 1123 K only,
whereas CaSO, and CaS are products of the reaction over the
full temperature range, i.e. reactions (2) [with possibly (lo)]
and (11) can take place at temperatures below 1123 K, but
only reaction (11) appears to occur above this limit. It is
important to note that reaction (1l), although giving the same
1232

J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

a73 K

1073 K

973 K

,
20

1173 K

I
I

wavenumberkm-'

Fig. 6 IR absorption spectra of the solid produced by reacting sulfur


dioxide (0.27 vol.% in N2) with calcium oxide. Spectra for different
temperatures of reaction are shown, together with the species
responsible for each absorption.

products as reaction (2) followed by reaction (lo), represents a


mechanism which does not involve CaSO, as an intermediate.
This is because reaction (11) is seen to occur at temperatures
at which CaSO, is no longer stable.
Low et a1.,16 when analysing infrared spectra of the products of the reaction between CaO and SO,, found, in addition to the absorptions corresponding to 0,-, SO,'-, SO,,and S2-, some traces of various S,O,"- species. It would be
expected that S,,O,"- species might be oxidised on standing in
air, and furthermore, detailed analysis shows that the presence
of S,0,2- and S,OS2- would be masked by strong absorptions of SO,,- and SO,,-. Therefore the fact that these peaks
were not detected in this work is not conclusive evidence for
them not being present.
Kinetics of reaction of sulfur dioxide with calcium oxide
This is an analysis of the thermogravimetric experiments
described for Fig. 3, in which SO, concentrations ranged from
0.135 to 1.0 vol.% in N,, and temperature from 873 to 1173
K. Fig. 3 gave the mass of a sample of CaO, while it increased
with time, due to reaction with SO,. The slope of the best
linear fit was taken from each graph, divided by the initial
mass, m,, to give a normalised rate, which was then plotted
against the corresponding concentration of SO, for each temperature studied. The results are presented in Fig. 7. It can be
seen that a good linear fit can be made to each set of measurements, except those at 1123 K ; the implications of this for the
kinetics of the reaction between CaO and SO, will now be
discussed.
First, we consider the expected kinetics of a reaction in
which solid and gas react to produce a different solid, e.g.
ki

CaO+SO,

CaSO,

(2)

k-2

where k, and k - denote the forward and back rate constants


as shown. Let the rate of absorption of SO, by unit surface
area of solid be
rsot = k2CSO2I - k - 2

(IV)

where [SO,] is the concentration of SO, at the surface of the


solid CaO. The rate of decomposition of CaSO, has been
taken to be zero order in CaSO,, whereas the rate of absorption of SO, is independent of the fraction of the CaO surface
covered by CaSO,. In this case equilibrium is characterised
by
0 = k,[SOJ,

- k-2

View Article Online


I

.-C

m
I

0
7

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

% 4

E
z.

0.5

1 .o

(Sod (vol.% in N2)


Fig. 7 Normalised values of the initial rate of reaction us. the concentration of sulfur dioxide in N,. Temperatures: (0)
873; (V) 973;
(A) 1073; ( 0 )1123; and (m) 1173 K.

where [SO,], is that for equilibrium of reaction (2). This gives


the rate of absorption per unit surface area to be
rso2 = k2(CS021 - Cs0,le)

(11). The measurements for 1123 K fall on two straight lines,


as shown in Fig. 7. At a temperature of 1123 K and SO, concentrations above 0.41 vol.% in N, {i.e. [SO,], for reaction
(2)}, the more rapid formation of sulfite via reaction (2) can
take place, whereas at the same temperature, but with SO,
concentrations below 0.41 vol.% in N, at 1123 K, reaction
occurs via reaction (11) only. This is in agreement with the
infrared spectra for this temperature, not shown in Fig. 6, in
which CaSO, was not observed when the concentration of
SO, was 0.27 vol.% in N2. From this is can be deduced that
the temperature of the maximum rate of sulfation will increase
with SO, concentration. This was shown to be so by Simons
et
from their studies of the effect of SO, concentration on
the temperature at which the maximum in sulfur capture
occurs. In addition, it should be noted in Fig. 7, that the best
linear fits do not pass through the origin. Instead, they show
that the rate of reaction will be reduced to zero below a
certain threshold concentration of SO,. This follows from
eqn. (V), which is in agreement with Dennis and H a y h u r ~ t , ~
and their treatment of the kinetics of CaO + CO, + CaCO,.
Thus for reaction (2) or (11) to occur, [SO,] must exceed its
value for the particular reaction being at equilibrium.
It was concluded above that the diffusional resistance terms
in eqn. (11) may be fieglected. Therefore, the rate of reaction
between CaO and SO,, in the absence of 0, and at temperatures below 1123 K, may be written as:

and for temperatures above 1123 K, as

(V)

This is in accord with the equilibrium concentrations, [SO,], ,


being independent of the surface concentrations of the two
solids. Also, the rate of formation of CaSO, can be expressed
without k - , ; in fact, the rate of reaction is linear in ([SO,]
- [SO,],). When the measured rate, rsoz, is plotted against
[SO,], as in Fig. 7, eqn. (V) predicts a straight line with an
intercept on the horizontal axis equal to [SO,],. This is
observed in Fig. 7, except at 1123 K.
That Fig. 7 is linear signifies a reaction order, with respect
to SO, concentration, of unity, as assumed in eqn. (IV) above.
This is generally
but with Borgwardt and Bruce',
reporting a reaction order, with respect to SO,, of 0.62.
However, they were considering a regime in which external
diffusional resistances were significant and attributed this
unusual reaction order to diffusional effects. Also, it can be
seen in Fig. 7 that there is a significant difference between the
gradients of the best-fit lines for temperatures above and
below 1123 K. This can be interpreted in the light of the conclusions drawn from the above infrared spectroscopy. The
spectra in Fig. 6 indicated that below 1123 K, the possible
products of the reaction are CaSO,, CaSO, and CaS, presumably from reactions (2) and (11). It should be noted that
reaction (10) does not lead to a change in mass, so its
occurrence cannot be detected in the thermogravimetric
balance. However, reaction (10) is thermodynamically favourable (i.e. exoergic) over the temperature range being considered, so where reaction (2) forms CaSO, , it can be assumed
that some decomposition of the sulfite might occur via reaction (10). Above 1123 K, which is the thermodynamic limit of
stability for CaSO, for the [SO,] used here, the species
CaSO, and CaS were still evident in the spectra, but CaSO,
could no longer be detected, i.e. presumably only reaction (1 1)
was taking place. Therefore in Fig. 7, the steeper gradient at
the lower temperatures reflects a combination of the rates of
reactions (2) and (11) and at the higher temperature of 1173 K
the shallow gradient indicates a slower rate, solely for reaction

where, S is the BET surface area (m2 kg- l), 64 g mol- is the
increase in mass per mole of sulfur dioxide reacted, and subscripts 2 and 11 refer to reactions (2) and (Il), respectively.
Eqn. (VI) and (VII) show that the gradient of the lines in Fig.
7 will be proportional to ( k , + k , , ) and k , , , respectively,
according to the temperature. From the plot at 1173 K, the
value of k , , is estimated to be 2.8 f 0.7 x l o w 7m s-'. If a
small activation energy is assumed for reaction (11) then a
rough estimate of the rate constant for reaction (2) can be
made. This yields values of the order of 1.2 f 0.5 x
m
s- over the temperature range 873 to 1073 K. Note that this
value is of the same order of magnitude as those reported by
others.36 The intercept on the horizontal axis of Fig. 7 is characterised by
(k2

+ k11)CS021 = k2[SO,I,e + kllCSO2llle

(VIII)

at temperatures below 1123 K, and by [SO,] = [SO,],,,


above 1123 K. The values of the measured intercept are presented graphically in Fig. 8 for each temperature. Also shown
are the values of [SO,], calculated for reactions (2) and (11)
separately from thermodynamic data.,'-,'
From Fig. 8 it can
be seen that as temperature increases, the experimentally
derived intercepts deviate away from those predicted for reaction (2) to those for reaction (11). This again suggests that
both reactions occur at low temperatures, with reaction (1 1)
tending to dominate as the temperature increases.
Extent of reaction
Experiments were carried out in the thermogravimetric
balance, where 1.0 vol.% SO, in N, was passed over a sample
of CaO, for the temperature range 873-1173 K, until no
further mass change was recorded. Usually the time for this to
happen was of the order of 100 min. Fig. 9 shows the results
plotted as the ratio of the mass of the sample at time t, to its
J . Chem. SOC., Faraday Trans., 1996, Vol. 92

1233

View Article Online

controlled by both chemical kinetics and diffusion (most probably of SO,) through the layer of product of thickness 6. In
eqn. (IX), k , has been replaced by the overall rate constant k ;
in fact, below ca. 1123 K it is likely that k = k , + k l l , but at
higher temperatures k = k l l . If a fractional mass increase is
defined as

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

then eqn. (IX), after integration from time t = 0 (when F = 0),


becomes

a73

923

973

1023

1073

1123

1173

temperature/K
Fig. 8 Calculated concentrations of sulfur dioxide for reactions (2)
and (1 l),respectively, at equilibrium; (m) intercepts from Fig. 7

initial mass, m o , against time. It can be seen that Fig. 9 is


initially, as expected, an extension of Fig. 3, followed by an
abrupt termination of the reaction. As in Fig. 7, there is
apparently a faster reaction rate in Fig. 9 at lower temperatures, with a decrease in rate after 1123 K, when the more
rapid formation of CaSO, uia reaction (2) is becoming less
viable. Assuming that CaSO, is the major product and thus
100% conversion would correspond to a final-to-initial mass
ratio of 2.14, it is calculated that the extent of conversion at
lower temperatures becomes CQ. 80%. A conversion of 80% is
considerably greater than most reports in the literature.
Dennis and Hayhurst, observed up to 40% conversion and
Simons et al., saw no more than 20%, although O'Neill et aL8
did find that with extended calcination times for the limestone,
they could realise almost 90% conversion. It should be noted
that the increase in mass shown in Fig. 9 is approximately
linear in the initial stages of reaction. This is again synonymous with no significant diffusional limitation until appreciable extents of reaction have been reached.
Fig. 9 can be analysed by neglecting l/k, in the denominator of eqn. (11), giving
dm
dt

-=

where pp is the density of the product of reaction, most probably CaSO, at the lower temperatures. Thus, provided S does
not change with time, a plot of F against ( t / F ) should be
linear, with a slope characterised by D,and an intercept (when
F = 0) determined, among other things, by k. Fig. 10 gives
such a plot for the results presented in Fig. 9. Examination of
Fig. 10 reveals that the expected behaviour is found at 1173 K
only. At lower temperatures there is at low F an initial linear
part, followed by a region when the mass of sample and F are
hardly changing with time. This second region must correspond to some new factor controlling the rate of reaction; it
could be that the layer of product is so thick that the pores
are beginning to block at their entrances. This situation has
been considered previou~ly.~'The initial slopes in Fig. 10
have been used to determine D, (using pp = 2800 kg rn-,);
their values are listed in Table 2, together with k from the
intercept on the horizontal axis. Below 1160 K there is no
significant change of D, with temperature: D, = 1.9 & 0.5
x
m2 s-' from 973 to 1160 K. Such a magnitude is
lower than some measured for this system. For example
Bhatia and Perlmutter3' report a value of 6.9 x lo-', m2
s-'. Also, at 1098 K, Dennis and Hayhurst4' found D, = 7.6

0.E
1 60 K

64m,S(C - C,)

0.4
k

where C, is the value of C (the concentration of SO,) for equilibrium. Eqn. (IX) holds provided q = 1 and that reaction is

0.2

t
0
time/min.

200

I
400

/
I

600
( t / F )/min

I
800

I
1001

Fig. 9 Measured mass of the sample for extended reaction between


CaO and 1.0vol.%)SO, in N,. Temperatures: (m) 973; (A)1073; (0) Fig. 10 Plot af F, the fractional increase in mass of sample, us. t/F
for the results in Fig. 9. Temperatures (K) are as shown.
1123;(A) 1148;( 0 )1160;and (0)1173 K.

1234

J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

View Article Online

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

Table 2 Values of k and D, for initial stages of reaction between SO,


and CaO, as derived from Fig. 10
temperature/K

k/m s-l

973
1073
1123
1148
1160
1173

1.7 x lop6
1.3 x
1.7 x
9.4 x 10-7
4.9 x 10-7
3.0 x 10-7

2.0 x
1.7 x
2.2 x
1.5 x
1.9x
2.3 x

10-14
10-14
10-14
10-14
10-14
10-15

Reaction between sulfur dioxide and calcium oxide in the


presence of oxygen

- a
Od

I
I

1123K
I

1
60

120

180

time / min

Two sets of experiments are described. In the first, 0, was


introduced to the system after CaO had reacted with SO, and
the second set involves the simultaneous introduction of SO,
and 0, over solid CaO. The experimental procedure for the
first experiments was as follows. A sample of CaO was heated
under N, to constant mass and temperature, at which point
SO, (0.27 vol.% in N,) was introduced over the CaO, for
approximately 60 min. The flow of SO, was then stopped and
0, (10.5 vol.% in N,) was passed over the sample until no
further change in mass appeared to be taking place. These
experiments were carried out for well defined constant temperatures in the range 873-1173 K. At the end of each experiment the sample was cooled under N, and analysed
spectroscopically. The products of the initial reaction with
SO, were those identified above, i.e. CaS, CaSO, and CaSO,
at temperatures below 1123 K, but only CaS and CaSO, at
higher temperatures. It would therefore be expected that
introducing 0, to the system would result in the oxidation of
CaSO, in reaction (3) and of CaS in
+ CaSO,

1023K

DJm's-'

x lo-', m2 s-', whereas the results of Mulligan et a1.36 correspond to 5.8 x lo-', m2 s - l at 1098 K, but both these
studies used natural limestones rather than AnalaR CaO. It
should be noted that D, does not vary systematically with
temperature from 973 to 1160 K (see Table 2). This lack of an
activation energy suggests that the diffusional process could
involve gaseous SO, being transported through the solid
product (mainly CaSO,) rather than the diffusion of ions,
which would probably have36 an associated activation energy.
Table 2 shows that D, is decreased by a factor of 10 when the
temperature falls from 1160 to 1173 K, suggesting that SO, is
diffusing through a different solid (CaSO, + CaS) at the
highest temperature. Finally, it might be noted that these
values of D, are in agreement with Fig. 3 being linear for low
extents of reaction, thereby confirming the above analysis. The
values of k in Table 2 agree well with those from the initial
slope of Fig. 3 discussed above.

CaS + 2 0 ,

(12)

The infrared spectra of the products of these experiments


involving 0, show that most of the CaSO, and CaS has been
oxidised to CaSO,. It was noted3, that the atmospheric oxidation of CaS at room temperature produces CaSO, , presumably as an intermediate in the oxidation of CaS to CaSO,.
The thermogravimetric results can now be interpreted. Fig. 11
depicts typical mass us. time plots for such an experiment at
temperatures of 1023 K [Fig. ll(a)] and 1123 K [Fig. ll(b)].
Fig. ll(a) shows almost a step change in mass on the introduction of 0, to the system, whereas Fig. ll(b) illustrates a
much more gradual increase in mass for the same occurrence.
This suggests that the oxidation of CaSO, is more rapid than
the oxidation of CaS, the latter being the only reaction taking
place at 1123 K. In other words, it appears that the rate of
reaction (3) is considerably faster than that of reaction (12).
On this basis, a rough estimate of the rate coefficients for reactions (3) and (12) can be made. It will be assumed that the

Fig. 11 Plots of the mass of CaO against time, showing when SO,
(0.135vol.% in N,) was admitted and also when both 0, (10.5vol.%
in N,) was introduced and SO, removed. Results are for temperatures
of (a) 1023 and (b) 1123 K, as shown.

increase in mass in Fig. ll(a) on the introduction of 0, relates


principally to the oxidation of CaSO, , whereas the increase in
mass due to oxidation in Fig. ll(b) corresponds to the oxidation of CaS. In addition, the first-order dependence with
respect to both 0, and the available surface area for reaction
is assumed. Thus the rate of increase of mass of the sample in
Fig. 1l(a) is

and similarly in Fig. 1l(b) is

where k , and k , , denote the rate coefiicients of the forward


steps of reactions (3) and (12), respectively, SCaS03and SCaSrepresent the total specific surface area of CaSO, or CaS available for reaction and the numbers 16 and 64 are the increases
in mass (kg kmol-') of CaSO, formed from CaSO, and CaS,
respectively. It must be stressed that only an approximate
value of either k , or k , , can be derived from Fig. 11, because
the exact values of SCaS03and SCaSare not known. Also, in Fig.
ll(a), it can be seen that the majority of the oxidation of
CaSO, has taken place after only 5 min, but no further information regarding the actual time for oxidation can be
obtained from this particular plot, i.e. it is possible that the
mass increase took place over a considerably shorter time
interval than 5 min. Estimates of the surface area available for
reaction were made from the results shown in Fig. 4, in which
a sample of CaO was reacted with SO, (0.27 vol.% in N,) for
increasing times and the BET surface area was then measured.
This gave the BET surface area of a sample of CaO, after
reaction with SO, for 45 min as 8.3 f 0.6 m2 g-'. Thus,
minimum values of k , and k , , are estimated to be
J . Chem. SOC., Faraday Trans., 1996, VoZ. 92

1235

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

View Article Online

6.7 & 2.0 x lo-' m s-' and 8.8 f 2.7 x lo-' m s - l at 1123
K, respectively. Recently, k , , has been measured,, to be
2.8 x lov6 m s-' at 1123 K, i.e. 32 times larger than the
minimum value derived here for k , , .
These experiments, in which a sample of CaO was first
reacted with SO, and afterwards with O,, were investigated
further. If, as is indicated by the infrared spectra, CaSO, and
CaS are being oxidised to CaSO,, the ratio of the number of
moles of SO, absorbed initially by the solid to the number of
moles of 0, absorbed later should be 2. This does not depend
on whether the sequence of reactions is (2) followed by (3), or
(11) followed by (12). The mass increases in the two stages of
reaction (see Fig. 11) always gave3, a ratio of 2 in the temperature range 873-1173 K. To summarise, CaSO, and CaS
are oxidised in the presence of 0, to CaSO,; the rate of oxidation of CaSO, appears to be considerably faster than the
rate of oxidation of CaS.
The second set of experiments involved passing SO, and 0,
(0.14 vol.% and 10.5 vol.%, respectively, balance N,) simultaneously over a sample of CaO in the thermogravimetric
balance for the temperature range 873-1173 K. Again, infrared spectra were taken of the products of reaction. As
expected, CaSO, is the major product with weak absorptions
of CaSO, and CaS also evident over this temperature range.
That traces of both the sulfite and sulfide can be detected in
the products of the reaction at both low and high temperatures is particularly noteworthy. First, this indicates that
at 1173 K, CaSO, is either stabilised by the presence of 0, ,
or it is an intermediate in the oxidation of CaS to CaSO,.
Secondly, if the reaction occurred only as reaction (9,with
SO, as an intermediate, CaS and CaSO, would not be
formed. This can be used to interpret the thermogravimetric
results. If the reaction were to involve SO,, then some dependence of the rate on 0, concentration would be expected.,'
However, if SO, is not a major participant, then it is unlikely
that the rate of reaction will depend on the concentration of
0,. If it is assumed that with 0, present, reaction is mainly
by the relatively slow steps (2) and (11) followed by rapid oxidation of sulfide and sulfite, then the fractional increase in
mass can be expressed as :
1 dm
--m, dt

80Sko(C - C,)

where 80 is the mass increase, (g mol- ') of CaSO, produced


and k , is the overall rate coefficient measured in the presence
of oxygen. If reaction (1) is taken to be the overall reaction,
then the value of C,, the equilibrium concentration of SO,, is
calculated,, to have a negligibly small value. Thus, C, can be
omitted from eqn. (XI). For each experiment in the presence of
oxygen, a plot of mass against time, such as in Fig. 3, was
obtained and the value of k , ( = k 2 + k , , ) was derived using
eqn. (XI). The values of k , are presented as an Arrhenius plot
in Fig. 12, along with the values of ( k , + k , , ) , derived above
from the slopes of Fig. 7. The rate constants for the reaction
between CaO and SO,, in both the absence and presence of
0,, are presented together in Fig. 12. It can be seen that the
plots show no trend with 0, concentration. This shows that
the reaction is zero order in 0, concentration, as suggested by
fluidised-bed experiments, and other ~ t u d i e s .It~ would be
expected that, if SO, were playing a major role in the sulfation of CaO, the reaction rate would show some dependence
on oxygen concentration. This is evidently not the case and it
may therefore be concluded that sulfation of CaO occurs
without the formation of SO, as a necessary intermediate.
That is, the rate-determining steps in the sulfation of CaO,
even in the presence of 0, , will be reactions (2) and (1 1) below
1123 K and reaction (11) above 1123 K. Oxidation of the
sulfite and sulfide will then occur if 0, is present, with the rate
of oxidation of the sulfite being faster than that of the sulfide.
1236

J . Chem. SOC.,Faraday Trans., 1996, Vol. 92

-13
0

-14

-15

Fig. 12 Arrhenius plot of the natural logarithm of the overall rate


constant for the reaction between SO, and CaO, in both the presence
and absence of 0,. In each case the concentration of SO, was 0.135
vol.% in N,; concentrations of 0, are: (a)zero; (A) 1.2;(A) 2.5; (0)
5.0; and (m) 10.5 vol.%.

In addition, Fig. 12 clearly shows that there is almost a step


change in the rate of sulfation both in the absence and presence of O,, i.e. the mechanism of the reaction is not dependent on the presence of 0,. Since the formation of SO,
requires that 0, be present, this further substantiates the
claim that sulfation of calcium oxide can proceed without the
formation of SO, to any significant extent. The scatter of the
points in Fig. 12 makes analysis of the rate coefficients difficult. The values of In k , for each temperature were averaged
and analysed as follows. Assuming that both k , and k , , are of
the simplest Arrhenius form, then
ln(k,

+ k , , ) = ln[A,
+A,,

exp(-E,/RT)
exP(-E,,/RT)I

(XW

Eqn. (XII) describes the points plotted as in Fig. 12 for the


lower temperatures, whereas at 11 73 K, the value of k , corresponds to k , , alone. This enables the values of A , , E , , A , ,
and E l , to be determined by an interative process. The
expressions for the rate constants thus derived are k , = 7.2
x lop6exp(-12000/RT)
m sC1 and k , , = 1.2 x lop6
exp(-4000/RT) m s-l, with the activation energies having
units of J mol-'; errors of approximately 25% are estimated
for both rate coefficients.

Discussion
The above results will now be discussed, with a view to presenting a comprehensive picture of the sulfation of CaO. First,
as for the reaction between CaO and SO, in the absence of
0, , infrared spectroscopic studies showed that the products of
the reaction, at temperatures below 1123 K, are CaSO,,
CaSO, and CaS, whereas above 1123 K, only CaSO, and CaS
are produced. That is, below 1123 K, reaction can occur in
both (2) and (11). However, above 1123 K, only reaction (11)
takes place. It is also possible that CaSO, can be converted to
CaSO, and CaS in reaction (lo), but there is no change of
mass associated with this reaction, so its occurrence cannot be
detected by weighing.Thermogravimetric experiments gave the
rates of reactions (2) and (11) for each temperature. Fig. 7
shows that the rate of reaction below 1123 I( is considerably
faster than that above 1123 K ; this can be correlated with the
infrared results (i.e. CaSO, is only observed below 1123 K), if
the rate of reaction (2) is greater than that for reaction (1 1).

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

View Article Online

This was supported by thermodynamic calculations, which


pared with a porosity of 38% for the pure sample of CaO used
showed that CaSO, is unstable with respect to CaO and SO,
here. Therefore, that the rate of sulfation in these experiments
at temperatures above ca. 1100 K, i.e. reaction (2) will not
was not as high as that of eg. Borgwardt' might be explained
proceed in the forward direction above this temperature, with
by the purity of the CaO sample used. A pure sample was
normal concentrations of SO,. That the fast reaction (2) takes
chosen deliberately to eliminate factors which could not be
place up to ca. 1100 K only, also explains the observed
readily quantified.
maximum in sulfur capture,., by the sorbent CaO; the
Intrinsic rate constants derived from measured rates of suldecrease above 1100 K arises from reaction occurring only via
fation show considerably more variation. It was justified
the slow process (11). Lyngfelt and Leckner, suggested that
above that by using quartz wool in this work to distribute the
the maximum in the conversion of CaO to the sulfate
CaO, and by using very small porous particles, diffusional
occurred only in a fluidised-bed combustor and, along with
resistances become negligible and thus the observed rate of
, ~
and Dennis and H a y h ~ r s t they
, ~ proJonke et ~ l . Fieldes6
sulfation could be directly related to the chemical rate conposed various interpretations of the phenomenon in terms of
stant. This yielded values for k , of ca. 2.0 & 0.5 x lop6 m s - l .
conditions within the fluidised bed. However, this study shows
These compared well with those of Mulligan et
who also
that the maximum in the extent of sulfur capture can be
used a thermogravimetric balance, but are lower by at least
explained solely by the mechanism of reaction between CaO
two orders of magnitude than those of Dennis and Hayand SO,. This is in agreement with Moss," who also used a
hunt.,' However, it was noted above that the actual rates of
reaction rate theory to explain the effect of temperature on the
sulfation do not differ to the same extent and the apparent
sulfation of CaO. Moss'5 proposed that reaction above 1100
discrepancy in the rate coefficients can be attributed to the
K was by the oxidation of SO, to SO,, with the latter comassumptions made in their calculation. It was shown above
bining with CaO to give CaSO,. Since the maximum rate of
that, for the sample of CaO used in this work, the reaction
SO, removal was observed in the absence of 0, (as well as the
with SO, is kinetically controlled, and thus the rate coefficient
presence of O,), it can be concluded that the sulfation of CaO
was derived assuming that the whole of the BET surface area
is not by the route proposed by Moss," involving SO,. The
is available for reaction. Dennis and Hayhurst,' made the
assumption that the sulfation of CaO is diffusion controlled
reaction between SO, and CaO will be reported in a later
and therefore reaction initially only occurs at the exterior of a
p~blication.~,
Experiments to investigate the reaction between CaO and
CaO particle with diameter of e.g. 0.78 mm. Energy dispersive
SO, in the presence of 0, showed no dependence of the rate
X-ray analysis (EDXA) was therefore carried out4' on sulfated
on 0, concentration; this agrees with previous ~ o r k . ~This
, ~ , ~ CaO particles. It was claimed that these show that only the
conclusion establishes that the principal mechanism of suloutermost layer of CaO particle takes up sulfur, but the depth
fation does not involve the oxidation of SO, to SO,, as sugof the sulfated layer accounts for almost 20% of the particle. It
could be that the discrepancy between the values reported
gested before,' 5,24,25 since the rate of oxidation of SO, to SO,
would show a dependence on 0, concentration.
here and those by Dennis and Hayhurst21 for the rate coefficients of the reaction between SO, and CaO arise from the
The overall reaction for the absorption of SO, in the presence of 0, is reaction (1) and the formation of 1 mol sulfate
assumptions made concerning the surface area available for
reaction. If up to 20% of the total surface area was available,
from 1 mol oxide is therefore equivalent to the absorption by
rather than just the exterior area, the rate coefficient measured
the CaO of 1 mol SO,. Thus actual measured (initial) rates of
by Dennis and Hayhunt,' decreases to ca. 8 x lo-' m s-' at
sulfation, expressed in terms of mol SO, absorbed per gram of
1148 K, i.e. very much in line with those measured here.
CaO per second, will be considered and compared with the
Simons et aL2 also noted a wide variation in rate constants,
mol g-' s-' at a temvalue from this work of 4.4 x
despite similarity in rates of sulfation. They attributed this priperature of 1100 K and SO, concentration equal to 0.6 vol.%
marily to estimations of the diffusion coefficients involved in
in N,. For these conditions, examples of reported values are
cases where diffusional effects cannot be ignored. This adds
2.1 x lo-' by Borgwardt,' 2.4 x lo-' by Dennis and
further justification to the methods used in this work, which
Hayhurst,' and 5.2 x lop6 mol SO, per gramme of CaO per
have been shown to render diffusional effects negligible, even
second by Mulligan et ~ 1 The. variation
~ ~ in these values was
a t t r i b ~ t e d , ~ .to~ ,the specific surface area of the CaO availto a higher extent of conversion of the CaO.
Values of the activation energy for the sulfation of CaO in
able. To illustrate this point, Borgwardt and Harvey4, report
the presence of 0, also reflect this variation in rate constants.
values for the initial rate of sulfation, at a temperature of 1253
Dennis and Hayhurst,' reported an activation energy of 38 kJ
K and 0.3 vol.%SO,, of 2.8 x l o p 7 and 1.4 x lo-' mol SO,
mol-', which is in the range quoted by Borgwardt and
per gram of CaO per second, for particles of mean diameter
Harvey4, of 34 to 76 kJ mol-'. However, more recently,
1.3 and 0.25 mm, with the corresponding effectiveness factors
Borgwardt and BruceI3 report a value of 9 kJ mol- ' for small
calculated to be 0.17 and 0.85, respectively. Another feature
CaO particles in the region of kinetic control. This is more in
contributing to the variation in reported rates of reaction
line with Simons et al.,, who quote 17 kJ mo1-' and also
could be impurities within limestone. The CaO used in this
Hajaligol et al.,,, who found an activation energy of 8 kJ
study was a pure chemical sample, as opposed to calcined
mol-'. The latter three activation energies given are in fairly
natural limestone, which typically contains some magnesium
good agreement with that found here of 12 f 2 kJ mol-' for
and other metal oxides. Yang et ~ 1 found
. ~an enhanced
~
rate
the major step (2) in the sulfation of CaO. Previous authors
of SO, uptake on adding Fe,O, to CaO and attributed this to
do not appear to have observed that reaction (11) occurs in
the catalytic effect of Fe,O, for the reaction. other^^^,^^ have
parallel with (2).
added sodium salts to CaO and found an increase in the sulRecently, the importance of reaction (11) has been
fation rate, probably due to the sodium compounds modifying
confirmed' 9 , 2 5 by investigating its reverse reaction as follows.
the pore structure of the calcine by increasing e.g. the mean
Mixtures of CaSO, and CaS powders were made into a
pore size. This was also used to explain the effect of magnesium salts on sulfation. Mulligan et
compact pellet (diameter 5 mm) and heated in pure nitrogen
found a high rate of
up to 1OOO"C. The evolution of SO, was observed above
sulfation for limestones with a high magnesium impurity level,
900C. After cooling, the pellet was set in epoxy resin, secbut, since separate attempts to sulfate magnesium compounds
tioned, polished and then examined in a scanning electron
were unsuccessful, they concluded that the enhanced sulfation
microscope. It is clear that there were regions in the solid
rate was due to the magnesium producing a more porous
where melting and agglomeration had occurred, as suspected
structure. Furthermore, extended calcination times of the
previously.20 EDXA established a deficiency in sulfur
limestone can result34 in a porosity as high as 60%, as comJ . Chem. SOC., Faraday Trans., 1996, Vol. 92

1237

Published on 01 January 1996. Downloaded by University of Texas Libraries on 19/08/2013 18:19:02.

wherever there had been melting; there was also a local


increase in the amount of calcium present. This suggests a loss
of sulfur as SO, formed in the backwards step of reaction (1l),
accompanied by the formation of CaO. Thus a critical temperature (probably2' ca. 830C) is required to produce
melting, so that the solids CaSO, and CaS can contact one
another and react. Fig. 12 shows no discontinuity at 80090O0C, so it appears that the appearance of a melt has no
effect on the kinetics of SO, absorption. Nevertheless, these
melts merit further investigation, in particular because they
appear at ca. 830C, which is also the temperature at which
CaSO, decomposes thermally in the reverse of reaction (2).
This could be a coincidence or merely a manifestation that the
reverse of reaction (11) involves the backward step of reaction
(10). Note that the plot of Fig. 7 for 1123 K comprises two
straight lines; this confirms that two reactions are occurring
when SO, reacts with CaO. On balance, the evidence points
to these being reactions (2) and (1l), but whether reaction (10)
participates or not remains unclear.

Conclusions
Using quartz wool in the thermogravimetric balance to distribute very small calcium oxide particles throughout the
available volume effectively removes the interparticle diffusional resistance to SO, reacting with CaO. This is shown by
the linearity in Fig. 3. In addition, BET experiments indicate
that reaction can take place within even the smallest pores
and thus the entire internal surface area of these tiny particles
is available for reaction, i.e. each reaction is initially
kinetically controlled. Thus accurate determinations of rate
constants could be made. Furthermore, infrared spectroscopy
and kinetic analysis enabled identification of the reaction
mechanism. At temperatures below 1123 K, sulfation of CaO
is predominantly via the formation of the sulfite and at temperatures above this, sulfation directly gives the sulfate and
sulfide, because the sulfite is no longer thermodynamically
stable. This also accounts for the sudden drop in sulfation rate
observed at temperatures ca. 1123 K, since the rate of formation of CaSO, is up to five times faster than the rate of formation of CaSO, and CaS. As temperature increases, so does the
observed rate up to 1123 K, but after this, sulfation can only
proceed by a slower pathway. This effect occurs both in the
absence and presence of 0, , explaining the mechanism of sulfation of SO, in both cases. The rate of sulfation by SO, is
shown to be independent of 0, concentration, which eliminates SO, as a necessary intermediate in the reaction, since its
formation would depend on the concentration of 0, present.
Rate coefficients were measured for the reactions involved and
the diffusion coefficient for SO, through a layer of the product
of reaction was determined
The authors are grateful to the former CEGB for support as a
CASE studentship to D. A.

View Article Online


5 A. A. Jonke, G. J. Vogel, E. L. Carls, D. Ramaswami, L. Anastasia, R. Jarry and M. Haas, AlChE, Symp. Ser., 1972, 68, (126),
241.
6 R. B. Fieldes, PhD Thesis, University of Cambridge, 1979.
7 G. A. Simons, T. E. Parker and J. R. Morency, Combustion and
Flame, 1988, 74, 107.
8 E. P. ONeill, D. L. Keairns and W. F. Kittle, Thermochim. Acta,
1975, 14, 209.
9 R. T. Yang, C. R. Krishna and M. Steinberg, ind. Eng. Chem.,
1977, 16, 465.
10 E. T. Turkdogan, Physical Chemistry of High Temperature Technology, Academic Press, London 1980.
11 N. A. Burdett, J. inst. Energy, 1983,56, 198.
12 J. E. Stantan, in Fluidised Beds: Combustion and Applications, ed.
J. R. Howard, Applied Science Publishers, London, 1983, ch. 5.
13 R. H. Borgwardt and K. R. Bruce, AZChE J., 1986,32,239.
14 A. Lyngfelt and B. Leckner, Chem. Eng. J., 1989,40,59.
15 G. Moss, inst. Fuel Symp. Ser., 1975, 1, (1) D2.
16 M. J. D. Low, A. J. Goodsel and N. Takezawa, Enuiron. Sci.
Technol., 1971, 5, 1191.
17 J. Tarradellas and L. Bonnetain, Bull. SOC. Chim. Fr., 1973, 6,
1903.
18 J. Zawadski, 2. Anorg. Allg. Chem., 1932,205, 180.
19 N. H. Davies and A. N. Hayhurst, Combus. Flame, in the press.
20 A. N. Hayhurst and R. F. Tucker, J. lnst. Energy, 1992,65, 166.
21 J. S. Dennis and A. N. Hayhurst, Chem. Eng. Sci., 1990,45, 1175.
22 T. R. Ingraham and P. Marier, J. Air Poll. Control Assoc., 1971,
21, 347.
23 K. Wickert, Energie Miinchen, 1960, 12,240.
24 P. Marier and H. P. Dibbs, Thermochim. Acta, 1974,8, 155.
25 N. H. Davies, K. M. Laughlin and A. N. Hayhurst, in 25th Sym26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44

References
1 R. H. Borgwardt, Enuiron. Sci. Technol., 1970,4, 59.
2 G. A. Simons, A. R. Garman and A. A. Boni, AIChE J., 1987,33,
21 1.
3 J. S. Dennis and A. N. Hayhurst, in 20th Symposium

(international) on Combustion, The Combustion Institute, Pittsburgh, 1984, p. 1347.


4 A. Lyngfelt and B. Leckner, J. Inst. Energy, 1989,62,62.

1238

J . Chem. SOC., Faraday Trans., 1996, Vol. 92

45
46

posium (international) on Combustion, The Combustion Institute,


Pittsburgh, 1994, p. 21 1.
F. A. Miller and C. H. Wilkins, Anal. Chem., 1952,24, 1253.
Handbook of Chemistry and Physics, ed. R. C. Weast, Chemical
Rubber Co., Cleveland, OH, 61st edn., 1980.
0. Levenspiel, Chemical Reaction Engineering, Wiley, New York,
2nd edn., 1972.
N. Wakao and J. M. Smith, Chem. Eng. Sci., 1962,17,825.
E. L. Cussler, Digusion: Mass Transfer in Fluid Systems, Cambridge University Press, Cambridge, 1984.
S. K. Bhatia and D. D. Perlmutter, AiChE J., 1981,27, 247.
M. R. Hajaligol, J. P. Longwell and A. F. Sarofim, Znd. Eng.
Chem. Res., 1988, 27, 2203.
D. Allen, P h D Thesis, Cambridge, 1990.
G. A. Simons, A. R. Garman and A. A. Boni, AiChE J., 1987,33,
21 1.
J. S. Dennis and A. N. Hayhurst, Chem. Eng. Sci., 1987,42,2361.
T. Mulligan, M. Pomeroy and J. E. Bannard, J. inst. Energy,
1990,62,40.
J. M. Smith and H. C. Van Ness, introduction to Chemical Engineering Thermodynamics, McGraw-Hill, New York, 3rd edn., 1975.
R. H. Perry and C. C. Chilton, Chemical Engineers Handbook,
McGraw-Hill, New York, 5th edn., 1973.
D. Cubicciotti, A. Sanjurjo and D. L. Hildenbrand, J. Electrochem. SOC.,1977,124,933.
I. Barin and 0. Knacke, Thermochemical Properties of Inorganic
Substances, Springer-Verlag, Berlin, 1973.
J. S. Dennis and A. N. Hayhurst, Chem. Eng. Sci., 1986,41, 25.
D. Allen and A. N. Hayhurst, J. Chem. SOC.,Faraday Trans.,
1996,92, 1239.
R. H. Borgwardt and R. D. Harvey, Enuiron. Sci. Technol., 1972,
6, 350.
R. T. Yang, M. Shen and M. Steinberg, Enuiron. Sci. Technol.,
1978, 12,915.
I. Johnson et al., Annual Report, ANL/CEN/FE-79-10, Argonne
National Laboratory, Argonne, Illinois, 1979.
J. A. Shearer, J. F. Lenc, I. Johnson and C. B. Turner, Annual
Report, ANL/CEN/FE-79- 11, Argonne National Laboratory,
Argonne, IL, 1979.

Paper 5/05815E; Received 4th September, 1995

You might also like