United States
Environmental Protection
Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park NC 27711
Research and Development
 EPA/600/S2-88/037 Nov. 1988
 Project  Summary

 Effect of Relative  Humidity and
 Additives on the Reaction of
 Sulfur Dioxide with  Calcium
 Hydroxide
 Rosa Ruiz-Alsop and Gary T. Rochelle
  Previous results  with  flue gas
desulfurization by spray drying   of
Ca(OH)2 show that a significant
amount of SOj is removed in the bag
filters used to collect the solids. This
research program investigates the
reaction of SOz with Ca(OH)2  at
conditions similar to  those  of
commercial scale bag filters: 19-
74%  relative humidity, 30.4-95 °C,
and 300-4000 ppm SO2. This  study
was carried out in  a bench-scale
fixed-bed  reactor, with powder
reagent Ca(OH)2 dispersed in silica
sand. The gas phase was  a mixture
of NZ, SOz, and water vapor. The
effects of  Ca(OH)2 loading, tem-
perature, relative humidity, Inlet SOj
concentration, and  additives were
investigated. Of the  additives tried
(buffer acichf, and organic and
inorganic  deliquescents),  only the
deliquescent salts improved Ca(OH)2
reactivity  toward  SO2-  The im-
provement depends on the type and
amount of salt and  on the relative
humidity. The experimental data were
modelled by a shrinking core model
with zero order kinetics in SOz, using
an empirical correlation to account
for shape and  surface roughness  of
the Ca(OH)2 particles. The diffusion
coefficient  of the SOj through the
product layer (De) increases linearly
with relative humdity  and the amount
of additive, and the  kinetic rate
constant (kg) increases exponentially
with relative humdity  and the amount
of additive. De values ranging from
 0.75E-9  to 1.20E-6 cm2/sec and ks
 values  ranging  from 1.0E-9 to
 8.23E-9 cm4/gmol sec in the model
 simulated the experimental results.
  This Project Summary was devel-
 oped by EPA's Air and Energy Engi~
 neering Research Laboratory. Re-
 search Triangle Park, NC, to announce
 key findings  of the research project
 that is fully documented in a separate
 report of the same title (see Project
 Report ordering Information at back).
 Introduction
  Flue  gas desulfurization by  spray
 drying of a Ca(OH)2 slurry has become
 increasingly important in recent years as
 an alternative to the more traditional wet
 lime or limestone scrubbing.  During
 spray drying,  the SO2-containing flue
 gas is contacted in the dryer with a finely
 atomized aqueous solution or slurry of an
 alkali (typically slaked lime or soda ash),
 which absorbs and neutralizes the SC>2.
 Simultaneously, the water is evaporated
 from the slurry droplets leaving a solid
 material which can be  collected  using
 conventional solids collection equipment
 such as bag   filters  or electrostatic
 precipitators. Bag filters are  preferred
 because the unreacted  Ca(OH)2 in the
 solids reacts with SO2 in the bag niters,
 producing  additional  SO2  removal. The
 reaction between lime  solids and S02
 taking place in the ducts and bag filters
 of a spray  dryer system is the subject of
 the present research.

-------
  Most of the information available in the
literature for  this reaction  is  reported
results of SO2  removal across the bag
filters of pilot and demonstration spray
dryer plants. These results are difficult to
interpret because the conditions of the
flue gas entering the bag filters depend
on the  spray absorber behavior. Any
variable changes that affect the SOs
removal across  the spray  absorber will
change  the concentration  of  SO2
entering the bag  filters. Therefore, two
variables  will  have effectively  been
changed in the  bag  filters and the
contribution of each cannot be isolated.
  Two bench scale studies regarding the
reaction of interest have boen reported:
one  at  the Lund  Institute  of  Tech-
nology-Sweden, and the other at EPA.
However,  Vie characterization  of this
reaction is far from complete.
  The present study was carried out in a
bench-locale  fixed-bed reactor,  oper-
ated  at conditions similar to those found
in bag filters of  commercial spray drying
systems,  and  using  powder reagent
Ca(OH)2 as the sorbent. The effects  of
Ca(OH>2  loading,  temperature,  SOa
concentration, and relative humidity were
studied.  Also,  additives  that  improve
lime's   reactivity  toward  SO2  were
identified.

Experimental Methods
  The general design of the experimental
apparatus  is given  in  Figure  1.  A
simulated flue gas was synthesized by
combining  Ng and  SOg  from gas
cylinders. The gas flow  rates were
measured using rotameters.  Water was
added  to the system  using   a syringe
pump (Sage  Instruments,  Model  341 A)
and evaporated  at  120*C in a stainless
steel chamber before mixing with the gas
stream.
  The  reactor  was glass, 4   cm  in
diameter, and 12 cm tall. The reactor
was packed with a mixture of silica sand
and Ca(OH)2 reactant in a weight ratio of
40:1. Sand avoids the channeling caused
by lime agglomeration. The silica sand
(between  SO  and  115  mesh) was
obtained from  Martin Marietta  Aggre-
gates.
  The reactor was immersed in  a water
bath that maintained system temperature
within 0.1 "C. Tubing upstream from the
reactor was heated to  prevent the
condensation of moisture on the walls.
Before  being analyzed,  the gas was
cooled and the  water  vapor then
condensed out  by cooling water and an
ice bath. The gas was analyzed for SO2
using » pulsed fluorescent SOg analyzer
                                                       Ice Bath
 Figurt 1.   Experimental apparatus.
(Thermoelectron  Corporation Model 40),
and the  SO?  concentration  was
continuously recorded. The SO2 analyzer
was calibrated  using a  calibration
standard (a mixture  of 2000 ppm  SOg
and N2). The reactor was equipped with
a bypass, to allow  the bed to be  pre-
conditioned and the gas flow stabilized at
the desired S02 concentration  before
beginning the experimental run.  Before
each  experimental run,  the bed  was
humidified by flushing with pure N2, first
at a relative humidity of about 98% for 10
minutes, and then  for 8 minutes at the
relative humidity at which the experiment
was to be performed. This humidification
simulated moisture  conditions encoun-
tered in the bag  filters where the solids
are originally slurry droplets.
  The reaction time was normally 1 hr.
The raw data from each experimental run
were  curves  of SOa concentration
leaving  the reactor  versus time.  These
curves were produced by the recorder of
the S02 analyzer.
  The raw data from the  experiments
were SO? concentrations from the reactor
as  a  function of time. By integration of
the SOa concentration over time  and a
mass balance on the reactor, the average
fraction  of Ca(OH)2  converted was
calculated at each time. As a backup, the
reacted solids were analyzed for  sulfite
and hydroxide using acid/base and iodine
titrations.
  Most of the experimental work involved
reagent grade Ca(OH)2  as a  reactant.
Two batches of Ca(OH)2,  identified  as
Lime  O and Lime  A, were  used.  These
two batches of Ca(OH)g  differ slightly in
particle size and BET surface area. The
slurrying  and drying process caused a
slight decrease in  the surface  area  of
Ca(OH)2.
  An  aqueous solution  containing the
desired additive  was  prepared. This

-------
solution (5 ml) was than added to 1 g of
Ca(OH)2 and slurried. The sample was
placed in  an oven to dry at 75°C for
about 14 hr, and then sieved to separate
the individual Ca(OH)2 particles  prior to
being mixed with the silica and placed in
the reactor.
Model
  The  equations  that  describe the
absorption or  a component from  a
moving gas stream by a fixed solid in a
packed bed consist of two  differential
equations  obtained  from material bal-
ances in the gaseous and solid phases.
  Schematically:
                  dz
                    r»~    z=L

                         7"\   Cso,
                         I '
         •hi.
Assuming uniform  SOz  and  Ca(OH)2
concentrations in the r and 0 directions,
and  neglecting  SOg diffusion  and
dispersion in  the z direction  as well as
the  time  derivative  of  the  SOg
concentration, the following equations are
obtained:
     Vmdcso2

       Adz
              = rrt
with boundary and initial conditions:
                                  (1)
                                  (2)
                                   (3)
  Att =

where:

vm

Cso.

A
         volumetric  flow  rate  of  gas,
         cm^/sec
         concentration  of  SO2 in  the
         gas phase, gmol/cm3
         cross  sectional area  of  the
         reactor, cm2
                                        z     =  length of the reactor, cm
                                        rSO2  =  rate of disappearance of
                                                 gmol/cm3 sec
                                        Clime =  Ca(OH)2 concentration, gmo!/
                                                 cm3
                                        t     =  time, sec

                                        The rate equation for rso2 depends on
                                        the model selected to represent the
                                        kinetics of the reaction.
                                          A shrinking core model with kinetics of
                                        zero order in S02 was chosen  to fit the
                                        experimental data. The  shrinking  (or
                                        unreacted) core model assumes that the
                                        reaction  takes place  at  the  exterior
                                        surface  of the  particle. As the  reaction
                                        proceeds, the surface of reaction moves
                                        into the  interior of the solid,  leaving
                                        behind a layer of  inert  product.  The
                                        external radius of the particle remains the
                                        same, assuming no shrinkage or swelling
                                        of the product layer.
                                          The  shrinking core  model was orig-
                                        inally  developed for the isothermal re-
                                        action  of spherical  solid  particles.  The
                                        Ca(OH)2 particles are  non-spherical and
                                        have a  rough surface, so their surface
                                        area is much higher than that of spherical
                                        particles  of the same  volume.  An  em-
                                        pirical  expression  was introduced  to
                                        account for the decrease in roughness as
                                        the reaction progresses. A dimensionless
                                        roughness parameter was defined as:
                                               o =
                                                                          (4)
                                        where:
A  = actual  surface area of the lime,
      mz/g
A0  = surface area of spherical particles
      of equal mass, m/g

When the lime is  unreacted, o can be
estimated as the ratio of the BET surface
area of  the  lime and  the surface  area
calculated from  the Coulter  Counter
particle  size  distribution,  assuming
spherical particles. As  the  reaction
progresses, o should decrease and ap-
proach the limit o = 1.0 when all the lime
has reacted.
  An empirical expression  of the form:
                                             o =
               «*P(aXn   +b)   (5)
                                                               Ume
where  Xi,me  is  the fraction  of lime
unreacted,  was used to describe the
change  of  roughness with the  reaction.
Equation (5) must satisfy the condition:
                                            At X,.    = 1.0  =
                                                 lime      o
                                                   o  =  BET area/A
                                                    o               o
                                   (6)
                                          To force a to decrease more rapidly at
                                        high values of X|ime and to simulate the
                                        experimental  results, the  following
                                        additional condition was imposed:
                                           At X..   = 0.8  o =  2.0
                                                lime
                                   (7)
                                          For the two batches of Ca(OH)2 used
                                        in the experiments (Lime O and Lime A),
                                        the constants a and b in Equation (5) took
                                        slightly different values (a =  12.878 for
                                        Lime O and a =  13.428 for Lime A, and
                                        the  values  of b  were -10.302 and  -
                                        10.742. respectively).
                                          Using  the  roughness parameter o,
                                        Equations (8) and (9) can be obtained for
                                        the  rate  of  disappearance of  SC>2. as-
                                        suming  that  chemical  reaction (or S02
                                        diffusion through the gas film and product
                                        layer) is the rate controlling step.
                                               _ _j_

                                            32~V
                     lime s  lime
                                                                                                                   (8)
                                                                                                                   (9)
1
V

v-1/3 .
Alim« '
r-g
S02
[

1
                                                                                      4nD NR(oo
                                                                                           e      o
                                                                                                   1/2
                                                                                                         4nR2k No
                                                                                                               f   o
where:

V     = reactor volume, cm3
R     = radius of the particle, cm
N     = number of Ca(OH)2 particles
Plime  = Ca(OH)2  molar  density,
         gmol/cm3
       = kinetic rate constant, cm/sec
  'SO2 = S02 concentration at  the  gas
         bulk gmol/cm3
De     = diffusivity  of SO2  through
         product, cm2/sec
kg     = mass  transfer  coefficient,
         cm/sec

If the chemical reaction is slow,  the rate
of disappearance of S02 will be given by
Equation (8).  If the chemical reaction is
fast, all of the SOg  that reaches  the
surface of the core  will be immediately
consumed and the concentration of S02

-------
 at the surface of the unreacted core will
 become zero. At these conditions the
 rates of diffusion of  the SOg through the
 gas film and product layer will become
 the  limiting steps, and Equation  (9) will
 become important.
   Of Equations (8) and (9), the one that
 gives  the  lower   rate  of SOg  dis-
 appearance  will  determine  the  overall
 kinetic rate. The parameters of the model
 are  kg, De, and ks. Equations (1)  and (2)
 -with the  rate  of  disappearance  of
 SOa given  by Equation (8) and  (9)--
 were integrated by assuming an average
 particle size of the  lime particles which
 gave the  same surface  area  as the
 measured  particle  size  distribution,
 assuming spherical  particles. The mass
 transfer coefficient, kg, was not used as
 an adjustable parameter because, at the
 conditions at which the experiments were
 performed, gas film diffusion is not  likely
 to be important.  The only effect of gas
 film  diffusion is  that  it  limits  the rate
.expression  (9) at the beginning of the
 reaction where product layer resistance
 is zero,  as  there is  no product formed.
 The value  of kg  (544  cm/sec)  was
 estimated using a Sherwood number of
 2, corresponding to  mass  transfer  from
 spherical particles in  a stagnant fluid.
   Thus Oe and ks were the only  param-
 eters used to fit the experimental  data
 Depending on whether mass transfer  or
 chemical reaction is  the controlling  step,
 only one of these  parameters may be
 important.
   A  computer program, with variable
 step size in time and in distance along
 the  reactor,  was written to  model the
 reaction. This program uses the IMSL
 routine  DGEAR  to  carry out the  inte-
 gration, and the IMSL routine ICSCCU to
 provide interpolated  values of lime con-
 version needed for each time integration
 step.
 Results
  The  effects  of relative  humidity,
 temperature,  inlet ?Oa concentration,
 and the amount of lime in the reactor on
 the  reaction  of SO2  with  powdered
 reagent Ca(OH)2 were studied.  The
 experimental  conditions  are listed  in
 Tabtel.
 Relative  Humidity
  Relative  humidity was found to have a
 dramatic effect on the rate of reaction of
 SC>2  with  Ca(OH)2 as illustrated by
 Rgure 2. The full lines correspond  to
 experimental results at 2000 ppm inlet
      and 66°C using 4  g  of  Lime  0.
Table 1. Experimental Conditions

 Relative Humidify:     17-90°.b


                    500-4000 ppm
 SO2 Inlet
 Concentration:
 Reactor
 Temperature:        30.5-95°C

 Nitrogen Flow Rate:   4600 cm3:min (O°C,
                   1 aim)

 Amount of Lime:     1.0-4.0 g
                         Figure 2  shows  that, for all  relative
                         humidities, 100% of the SOa entering the
                         reactor is being removed  during the first
                         1  or 2 min. of reaction, then the reaction
                         rate (represented  by  the slope of the
                         conversion  versus time  curves)  de-
                         creases quickly at low relative humidities
                         but more slowly at high  relative humid-
                         ities.
                           The broken  lines in  Figure 2   corre-
                         spond to the model prediction for these
                         experiments.   Figure  3  shows  the
                         dependence of the rate constant  and
                         solid diffusion  coefficient  on  relative
                         humidity. At high  relative humidity, the
                         chemical  reaction  is the controlling step
                         and  ks determines the rate of reaction.
         25
         20
      •8  15
      r
  4 gLimeO
    66°C
2000 ppm SO2
 4.6 I/min H
                                              D.>8.£-8
                                              k, = 1.8E-9
                                                 Experimental

                                              "  Model
                     10
              20       30
                 Time (minutes/
                                                 40
                                                           50
                                                                    60
Figure 2.   Effect of relative humidity on reaction rate. Lime 0.
                                                         100.00
                                                                                                             B
                         20    30    40   50    BO    70

                               Relative Humidity l%l

Figure 3.   Effect of relative humidity on model parameters.

-------
Any value of  De greater than  8.0E-8
cm 'sec  will give essentially the same
result. SOa diffusion through  the product
layer  becomes more  important  as  the
relative humidity decreases.  At 19%
relative humidity, Oe determines the rate
of adsorption  of S02.  At 19% relative
humidity,  any  value of ks greater than
1.0E-9 cm/sec will give essentially  the
same  results. Both  De and ks  are
affected by the relative humidity but. in
the range of relative humidities studied,
ks increases approximately linearly while
De increases exponentially. Because the
SOa diffusion  coefficient  changes more
rapidly than the  kinetic  constant,  the
controlling  mechanism changes when
the relative humidity is increased. The
strong effect  of  relative humidity on
reaction rate has also  been reported by
other researchers.
Inlet SO2 Concentration
  The  effect  of  the inlst SOa concen-
tration  on the reaction  rate was found to
depend on the relative humidity. Figures
4 and 5 illustrate the effect of inlet SOa
concentration at 70 and 50%  relative
humidity, respectively. The full  lines in
these figures correspond to experimental
results, and the broken lines to the model
predictions. At 70% relative humidity, the
Ca(OH)2 conversion  was  practically
independent of the  inlet  SOa con-
centration as can be seen in Figure 4. At
lower relative humidity the reaction rate
is not affected by the SOa concentration
if  the SOa  concentration is  high.
However, at  lower  levels of SOa  tne
reaction rate is affected  by the  SOa
concentration as illustrated by Figure 5.
The observed  effect  of  SOa  c°n~
centratipn can be explained by assuming
that the reaction rate  has  zero  order
kinetics in  SOa, but  at  low relative
humidity and/or  low SOa concentration,
SOa diffusion   (instead of  chemical
reaction) becomes the controlling step.
As  can be  seen from  Figures 4  and 5,
the model predicts the SOa effect with
reasonable accuracy.
                                                   25
           20
      a
               70% RH
              4 g Lime 0
             4.61/minNi
                66°C
                                     2150 ppm
                                                      1500 ppm
                                        1060 ppm
                                          Experimental
                                          Model:         De > r.4£-8
                                                         k, = 1.5E-9
             0        10        20       30       40

                                    Time (minutes!

 Figure 4,   Effect of SO2 concentration, high relative humidity.
                                                      50
                                                                      60
          25
          20
a
5    10
I
                50% RH
               4 gLimeO
               4.61/minNt
                 66°C
                                             4000 ppm
                                               2050 ppm   _ ^. *•



                                                 1100 ppm
                                                    D. = 1.4E-8
                                                     , = 7.55-9
             0        10       20        30       40

                                   Time fminutesl


 Figure 5.    Effect of SOi concentration, moderate relative humidity.
                                                     50
                                                              60
Amount of CafOHfe In the
Reactor
  The   effect on the average Ca(OH)a
conversion  of changing the  amount of
Ca
-------
keeping all other variables constant. The
relative humidity  was 74%. so  in this
region the reaction is expected  to be
kinetically controlled.
  The broken  lines in Figure 7 are the
model predictions for the  experiments
run at the two different temperatures. At
the conditions at which the experiments
were performed,  the reaction rate is
kinetically controlled, so  ks is the only
important adjustable parameter  in the
model. By using the values of ks  given
by  the model, an  apparent activation
energy of 2.9 kcal/gmol can be estimated
for  Ca(OH)2.  This value of activation
energy is somewhat lower than the value
of 6 kcal/gmol  reported by other sources
for this reaction.
  A very  weak  dependence of the
reaction  rate with temperature was also
reported by other researchers.
Additives
  Two organic  acids  (adipic acid  and
glycolic  acid)  and three organic  de-
liquescents  (ethylene  glycol. triethyl-
ene glycol, and  monoethanolamine) were
selected  as  test additives for Ca(OH)2.
All of them proved to be detrimental to
the reaction of SOa with Ca(OH)2-
  A number of deliquescent  salts  were
also tested as additives at 74 and  54%
rela'ive humidity. The beneficial effect of
the salts depends on the  type   and
amount of salt and the  relative humidity.
At high  relative humidity (74%),  all the
deliquescent salts tried v,ere  successful
in increasing  the reactivity  of  the
Ca(OH)2  toward 862. At a lower relative
humidity  (54%), some of the salts did not
perform  as  well, and  some, such as
Ca(N03>2, did  not have any beneficial
effect.
  The water  activity  over  saturated
solutions of the salts (70°C and 1 atm) is
about equal  to the fractional  relative
humidity  of  the  gaseous  phase   that
would be in equilibrium with a saturated
solution of the salt at that temperature. If
one  of  these  deliquescent salts is
contacted with  a gaseous  phase of
relative humidity greater than the water
activity, the  salt will capture water  from
the gas  phase  and become  a solution.
This tendency to capture water has  been
extensively documented in the literature
by studies  of  the   growth  of   salt
containing aerosols as  a function of the
atmospheric relative humidity. From  data
in the literature, it is clear that (based on
deliquescence alone) most of the  salts
tested, specifically NaNO3 and all the
chlorides except  LiCI,  should not  have
any  beneficial  effect  at 54%  relative
      I
      n

      !
    20


    18


    16


    14


    12


    10


     8


     6


     4


     2


     0
€
c
.0
5

I
                  1000 ppm SO.
                     50% RH
                   4.6 1/minNi
                      66°C
 Figure 6.
       0        10       20       30

                             Time Iminutesl

      Effect of Ca(OHl2 loading, 1000 ppm SO*
                                                            SO
                                                                      60
          25
          20
          IS
      r
               74% RH
             500 ppm SOi
              1 g Lime A
             4.6 I/mm Ni
                   64.4°C
                                            Experimental

                                            Model:  D« > 8.E-8
                      1O       20       30        40

                                    Time (minutes)
                                                       SO        60
Figure 7.    Effect of reactor temperature. Lime A.
humidity. Nevertheless,  these  salts are
among the ones that behave the best at
that  moderate  relative humidity. A
possible explanation for this finding
would  be a  hysteresis  phenom-enon;
this will be discussed later.
  The  report also  shows the values of
the diffusion coefficient  of SO2 through
the product  layer  (De),  and the  kinetic
constant (ks)  that  can be  used  to
simulate the  experimental runs using salt
additives. De values  ranging from 1.5E-
                                   B to 120E-8 (cm2/sec) and  ks values
                                   from 1.5E-9 to 8.23E-9 (cm4/gmol sec)
                                   were used  in the simulation of the salt
                                   experiments. Depending on tha amount
                                   and type of salt added, chemical reaction
                                   or SOg  diffusion  can  become  the
                                   controlling  step. A  reasonably  good
                                   agreement  was found  between  the
                                   predictions of  the  model  and  the
                                   experimental  results. The  maximum
                                   percentage  of  error  between the
                                   experimental data and  predicted values

-------


_
I
'L
c |5

                                                                +
                                                                iu
                                                                j£
                                                                c •
                                                                is
                    0123456789   10

                                 Mole % NaCI
 Figun 9.    Effect of NaCI on model parameters.
was below 10% In most cases; \n a few
exceptions, it was below 20%.
  The in'Lance of the salt concentration
on the 802 reaction rate is illustrated by
Figure 6. The salts used were NaCI and
NaNOa in concentrations ranging from 1
to IS mole%. The  experiments  were
carried out at a relative humidity of 54%,
and a  reactor  temperature  of 66°C.
Figure 8 shows  that the  conversion
increases with increasing concentration
of additive until about 10 mole%. After
this the curve levels  off.  The optimum
concentration of additive is then about 10
mole% for 1:1 electrolytes like NaCI and
NaN03.
  The  deliquescent salts affected the
model parameters similarly to the relative
humidity:  the  diffusion coefficient
increased  more rapidly  than the kinetic
constant  by  the addition  of  the  salts.
Figure  9 illustrates the effect of  NaCI on
the model  parameters.  The  kinetic
constant  increased  linearly when   NaCI
concentration was increased from 0 to 10
mole%. while the diffusion  coefficient
increased exponentially after  a sharp
increase from 0 to 1 mole% NaCI.
  As  mentioned earlier,  before  each
experimental  run  the  fixed bed  was
prehumidified by flushing with pure N2 at
a relative humidity of about  98% for 10
minutes before flushing with Ng at the
relative humidity at which the experiment
was to be performed.
  This prehumidification could  be  why
some of the salts were still effective at a
relative humidity  lower than  the  one
predicted from equilibrium consider-
ations. Due to hysteresis it  is  possible
that,  when  the relative humidity  was
lowered to the experimental conditions
after the prehumidification, some excess
water  remained in  the solids. Strong
hysteresis effects have been reported in
NaCI aerosols.
  To check if hysteresis was responsible
for the beneficial effect  of some salts at
low relative humidities, experimental runs
were made omitting the  prehumidification
step.  Table 2 shows  the results obtained
at 54 and  17.4%  relative humidity with
and without prehumidification of the bed
at 98% relative humidity. The  additives
used  were  NaCt.  NaNO3,  and  KCI.  At
54%  relative humidity, even  when some
decrease of the Ca(OH)2 conversion was
found  without the  prehumidification, the
results with  additives were  still  far
superior to those with the  pure  Ca(OH)£.
Hysteresis then, cannot  explain  all of the
beneficial effect observed at  54% relative
humidity. At  17.4% relative  humidity,  all
the beneficial effect with NaCI appears to
be due to  prehumidification of  the  bed;
i.e., due to a hysteresis phenomenon.
Conclusions
  As shown earlier, the relative humidity
of  the  gaseous  phase  is the  most
important variable in the reaction of SO?
with Ca(OH)2  solids. This result agrees
with results reported ii the  literature for
S02 removal in the bag filters of  spray
dryer pilot and  commercial plants. The
other variables tested  (i.e., temperature,
amount  of Ca(OH>2,  and  SO;  con-
centration)  have less  impact on the
reaction rate. The different effect of SOg
concentration  at low and high relative
humidities c?n be explained by assuming
that the reaction has zero order kinetics
in S02 and that, at  low relative humidity,
the  reaction  rate is  mass  transfer
controlled  while at high  relative
humidities the reaction is controlled by
reaction kinetics.

-------
         Effect of Prehuman/cation of the Bed at 98% flH on CafOH)2 Reactivity

                  500 ppm SOj, 1.0 g Ca(OH)2 A, 4.6 llmin (O'C, 1 aim) W2

                                    Average Ca(OH)2 Conversion after 1 hr 1%)
       Additive (Mole%)
                                    54% AH
                                      66'C
                                  Prehumidified
                                Yes
                                             No
                   17.4% RH
                     95'C
                  Prehumiaified
                                                         Yes
None
70S NaCI
10% N«M03
fOKKC/
11.2
27.0»
27.2s
37.3
-
23.2
23.7
19. 3
4.0
9.7
779
3.4
-
4.0
-
—
humidity: the diffusion  coefficient  in-
creases  more rapidly  than the rate
constant. When  increasing amounts  of
the same salt (NaCI)  were  added,  ks
increased  linearly  and  De  nearly ex-
ponentially.  Depending  on the  amount
and  type  of  salt  additive,  chemical
reaction  or gas  diffusion can  be the
controlling mechanism.
  The kinetic constant was a very weak
function  of  temperature. The estimated
activation energy was 2.9 kcal/gmol.
 Average of two nxperimsntal runs.
  Most  of  the  deliquescent  salts tried
effectively  increased Ca(OH)a reactivity
toward SOj. The extent of the beneficial
effect was a function of the type of salt.
the salt concentration, and  the  relative
humidity. Some salts are effective  at a
lower relative  humidity  than would be
predicted  from   their  deliquescent
properties. Hysteresis due to pre-
humidification of the bed appears to be
partially responsible for this behavior, but
it cannot explain  all of the  reactivity
improvement observed  at 54%  relative
humidity.  An  alternate explanation
proposed  is that  the chlorides  and
NaNOa  modify the properties  of the
product  CaSOs 1/2H2O layer that  is
formed  as the reaction  takes   place.
thereby  facilitating the access of the SO?
to the unreacted Ca(OH)z which remains
in the interior  of the particle.  NaCI and
CaCfe have been  reported  to enhance
the SOz reactivity of limestones  in
fluidized-bed combustion by affecting
the pore structure  of  the lime  during
calcination, which  then increases the
extent of sulfation of the limestone.
  The only previous modelling effort for
this reaction used an integral shrinking
core model with only reaction kinetics  to
explain  the dependence of reaction rate
on  lime  conversion.  That effort's
experimental  data  fit this integral
shrinking core model only after a certain
lime conversion had been reached. The
sharp decrease in reaction rate observed
at initial times  was  attributed  to  a
decrease in surface roughness,  but no
attempt  was  made to  correlate  this
decrease in surface roughness with  lime
conversion. The  model  neglected the
effects  of  SOz diffusion through the
product layer, and the SOg concentration
and  Kme concentration profiles  in the
fixed bed reactor. All of these factors will
be more important at initial times, when
the SOa removal is  higher,  so it is  not
surprising  that  the  experimental data
could be fit only at later times.
  The simple  model  presented here
seems able to predict  with reasonable
accuracy the effect  of  all the process
variables tested and explain the  trends
observed in the experimental  data. The
experimental data was estimated to have
±11% experimental  error, so  that most
predictions of the model are well within
the range  of  experimental error.  The
values of the diffusion coefficient used in
the modelling (from  0.75E-9  to  1.20E-
6} seem  reasonable for  diffusion of SOg
in a solid, as they are of the same order
of magnitude of the diffusivities of gases
in polylmers.
  The relative humidity  affects both  the
diffusion coefficient of SO2  and  the
kinetic constant. In the  range  of relative
humidity studied (19  to 74%), the  kinetic
constant  increased linearly with relative
humidity, while the diffusion  coefficient
increased exponentially. This depend-
ence  of the parameters on  the relative
humidity  leads to  a  change in  the
reaction controlling  mechanism as  the
relative humidity decreases.  At high
relative  humidity  and/or  high SOa
concentration, the  reaction rate   is
kinetically controlled and the reaction
rate is independent on the SOg level. At
low relative humidity and/or  low SOz
concentration, the controlling step is  the
diffusion of  the  S02  through  the
CaS03-1/2H2O  product  layer.  At  these
conditions  the  overall reaction rate
becomes  affected   by  the   SO?
concentration which is the driving force
for diffusion.
  The addition  of  deliquescent  salts
increases  the  diffusion  coefficient and
the kinetic  constant similarly to relative

-------
 R. Ruiz-Alsop and  G.  T. Rochelle are with the University of Texas at Austin,
   Austin,  TX 78712.
 Charles B. Sedman is the EPA Project Officer (see below).
 The complete report, entitled "Effect of Relative Humidity and Additives on the
   Reaction of Sulfur Dioxide with Calcium Hydroxide," (Order No. PB 88-234
   1741 AS; Cost: $25.95, 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:
    Air and Energy Engineering Research Laboratory
    U.S. Environmental Protection Agency
    Research  Triangle Park, NC27711
United States                   Center for Environmental Research
Environmental Protection         Information
Agency                         Cincinnati OH 45268
Official Business
Penalty for Private Use $300

EPA/600/S2-88/037

-------