Noncatalytic Heterogeneous Kinetics in the Engel-Precht Potassium

Carbonate Process
Gene L. Smithson
Saskatchewan Research Council, Saskatoon, Canada S7N OX 7

Narendra N. Bakhshi'
Department of Chemistry and Chemical Engineering, University of Saskatchewan, Saskatoon, Canada S7N OW0

The Engel-Precht process of K2CO3 manufacture was abandoned in Germany for economic and technical rea-
sons. This study attempts to identify and solve the process control problems and provide the kinetic data neces-
sary for converting the original batch process into a continuous flow process. The two reactions investigated
here are the formation of Engel salt (KHC03.MgC03.4H20) and the decomposition of this insoluble salt into a so-
lution of KHC03 plus solid MgCO3-3H20.The kinetics and mechanism of the Engel salt formation reaction (KCI,
CO2, and MgC03.3H20 reacted in an aqueous slurry) have been established along with the optimum conditions
for the reaction. Similarly, the kinetics, mechanism, and optimum conditions for the decomposition of Engel salt
have also been determined. This paper completes a study of all reactions in the Engel-Precht process and to-
gether with the results obtained in two previous publications (the hydration of MgO and the carbonation of MgO)
presents all the data required for the design of an efficient, continuous flow process.

Introduction formation of Engel salt. The present study was intended to

The Engel-Precht process of manufacturing K2C03consists supply the required information on the kinetics and mecha-
nisms of these reactions and to establish the optimum con-
of the reaction of Con, KC1, and MgC03-3H20 in an aqueous
ditions for each reaction.
slurry to form the insoluble double salt (KHC03.MgC03.
4&0, Engel salt) which can be separated and decomposed Experimental Section
in water to produce a solution of KHC03 or K2C03and solid
magnesium carbonate. As long as the magnesium carbonate A. Formation of Engel Salt. The MgO used to prepare the
is recovered as MgC03.3H20 it can be recycled in the process. MgC03-3Hz0 required in reactions 3 and 4 was sample no. 3
This method of K2C03manufacture was utilized in Germany of the MgO described before (Smithson and Bakhshi, 1969).
from the turn of the century until 1938 when it was superceded The normal method of preparing MgC03.3H20 was to add the
by the Formate process (Wiedbrauck,1942).The main reasons required amounts of MgO and water (usually 10 g of MgO, 250
for its demise were problems in process control and the high ml of H20) to the reactor and bubble in COz gas through a
operating costs of the batch process. The purpose of this study sintered glass dispersion tube while stirring a t 1600 rpm with
was to identify and solve these processing problems and to a four-blade propeller stirrer. This reactor has previously been
provide sufficient kinetic data for the design of a continuous described (Smithson and Bakhshi, 1973). The reaction was
flow process. continued for 1h at which time the appropriate amount of KC1
The chemical reactions encountered in the Engel-Precht was added and the CO2 flow and stirring were continued re-
process are as follows: sulting in the formation of Engel salt. Temperature control
in all Engel salt formation and decomposition reactions was
MgO + H20 Mg(OH)2 maintained by immersing the reactor to within 1-2 cm from
its top in a constant temperature bath. The CO:! flow rate was
followed by
adjusted so that there was a slight froth produced on the
+ COS + 2H20 MgC03.3H20
Mg(OH)2 F! (1) surface of the slurry. Preliminary runs using this method re-
sulted in severe thickening and frothing when KCl was added.
MgO + COn + 3H20 s MgCOy3H20 (2) This was found to be caused by the fine particle size of the
3(MgC03*3H20)+ CO2 + 2KC1 freshly formed MgC03.3H20. By allowing the MgC03 pre-
cipitate to digest in the reactor (maintaining reduced CO2 flow
2(KHC03*MgC03*4H20)+ MgC12
F! (3) and stirring) for 6-12 h, large crystals developed which did not
2(MgC03*3H20)+ 2KC1+ MgO + 2C02 cause thickening problems when KC1 was introduced. All
+ 3H20 s 2(KHC03*MgC03*4H20)+ MgCl2 (4)
subsequent Engel salt production runs were performed with
these digested MgC03.3H20 crystals.
KHCOyMgC03.4H20 s KHC03 The first series of Engel salt production runs was designed
MgC03.3H20 + + H2O (5) to determine the effect of temperature and initial KC1 con-
centration on the equilibrium conversion of KCl. In each case
The results of our studies of the first two reactions have the reaction was allowed to proceed to near equilibrium (124
been published previously (Smithson and Bakhshi, 1969, h). The extent of reaction was determined by measuring the
1973). This paper deals with reactions 3,4, and 5 . These re- amount of unreacted K+ remaining in solution. The reaction
actions have been investigated by Bayliss and Koch (1952) and slurry was centrifuged and the solids washed with cold water.
Bahavnagary (1962) with respect to phase equilibria, the The filtrate and washings were combined, diluted to a definite
equilibrium yield of Engel salt, and methods of decomposing volume, and analyzed for potassium by flame emic-:on spec-
Engel salt. There are no data available on the kinetics of the trometry. From the amount of KCl added a t the start of the

450 Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976
reaction and the amount remaining a t the end, it is possible titer X 0.025 N 256.3 X volume filtrate
X
to calculate the conversion to Engel salt. 1000 weight E. salt X volume aliquot
Once these preliminary runs had established the optimum
range of experimental conditions the method of sampling was X 100% = % decomposition (8)
changed so that reliable rate curves could be obtained. Sample The acid titration for carbonate-bicarbonate content is
aliquots were withdrawn from the reacting slurry a t definite preferred to the flame photometric determination of potas-
time intervals for analysis. A clear sample of the reacting so- sium as a means of calculating percent decomposition because
lution was obtained by immersing the sintered glass tip of a of the higher speed and precision of the titration.
gas dispersion tube in the slurry and drawing into it ap- In several experimental runs, Engel salt was decomposed
proximately 1 ml of liquid. A medium porosity gas dispersion at elevated temperatures under a positive pressure of COz in
tube filters out all crystalline solids. Because of its sensitivity a Parr low-pressure, catalytic reactor. Constant temperature
and hence the requirement of very small sample volumes, the operation was maintained by circulating water from a constant
flame photometric determination of potassium was used to temperature bath through a copper coil surrounding the glass
follow the course of the reaction. I t was also necessary to ac- reactor bottle. The reaction was carried out for a definite time
curately estimate the total volume of the solution in the slurry interval after which the COS pressure was released and the
so that the potassium concentration could be used to calculate slurry centrifuged and analyzed as before.
the amount of unreacted potassium. This meant taking into The kinetics of the decomposition reaction were followed
account all additions Gr removals of water and dissolved ionic by sampling the solution at time intervals and analyzing for
species. For reaction 3 there is no uptake or release of water COB- and HC03- or K+. The reactor and sampling procedure
in the reaction. The major changes to be considered are the were the same as those used for continuously monitoring the
solution volume changes in going from KCl to MgC12, the Engel salt formation reaction. The COz flow rate was main-
volume due to the dissolved magnesium carbonate, and the tained at a level just sufficient to produce mild bubbling at the
volume of solution removed with each sample aliquot. From surface of the slurry (approximately 200 ml/min) and the
these corrections an equation was derived (Smithson, 1973) stirring rate was 1600 rpm. Temperature control was main-
which has been used to calculate the fractional conversion of tained by placing the reactor in a constant temperature bath.
KC1 to Engel salt from the measured concentration of Kf in To calculate the percent decomposition of Engel salt it is
solution and the original volume as corrected for previously necessary to know the concentration of KHC03 in solution
removed aliquots. and the total volume of solution. The total volume can be
calculated from the initial volume of water by correcting for
(xE)n =
X,,Wo - VACn
[ Wo - cn[ vo - u(n - 1) - J ] all additions and removals. These volume corrections are for
dissolved KHC03 and MgCOr3HzO and the volume removed
with each sample aliquot. In addition a correction is required
for the amount of KHCOs removed with the sample aliquots.
These corrections have been incorporated into an equation
which gives the fractional decomposition of Engel salt
(Smithson, 1973).

The term Cn [ V O- u ( n - 1) - J ] is the correction for volume where

changes due to dissolved magnesium bicarbonate and for the
water removed in each sample aliquot. The term

is the correction for the amount of dissolved KC1 removed in The term V, is the correction for the volume increase due to
previous sample aliquots. dissolved MgCO3-3H20. Strictly speaking this correction in-
B. Decomposition of Engel Salt. Decomposition reactions creases gradually for the first 10-20% reaction and then de-
were performed in the same reactor used for the preparation creases through the remainder of the reaction as the increasing
of Engel salt. Measured amounts of Engel salt, water, and any concentration of KHC03 suppresses the amount of dissolved
decomposition additive required were placed in the reactor MgC03-3H20. Rather than introducing this complex correc-
and stirred at the maximum stirrer speed (1600 rpm). The tion, a constant value of V, is used throughout the reaction.
decomposition of Engel salt is normally carried out as in re- The error introduced by this simplification is less than 1%over
action 5, but when MgO is used as a decomposition agent the the range 0 I X D I1.00.
reaction proceeds as follows.
Results and Discussion
A. Preparation of Engel Salt. Equilibrium Conversion.
From the phase data compiled and expanded upon by Bayliss
and Koch (1952) it is possible to calculate the equilibrium
The decomposition was continued for a definite time period conversion of KC1 to Engel salt. These calculated conversions
and then the slurry was centrifuged and the solids were are plotted in Figure 1 for the KC1 concentration range 0-90
washed with cold water. After measuring the volume of filtrate mo1/1000 mol of HzO at both 18 and 25 "C. To confirm these
plus washings, an aliquot was taken for analysis. The car- values and to check on the extent of the equilibrium conver-
bonate and bicarbonate content were determined by titrating sion achieved in 24 h and over, three runs were performed a t
with 0.025 N HzS04 to the phenolphthalein and methyl purple KC1 concentrations ranging from 29 to 72.5 mo1/1000 mol of
end points. The total titer (methyl purple end point) was HzO. The temperature was 25 "C. These values are plotted
calculated as HC03- which was used to calculate the extent as the broken line in Figure 1. The close agreement between
of decomposition as follows. the two 25 "C curves indicates that near equilibrium has been

Ind. Eng. Chem., Process Des. Dev., Vol. 1 5 , No. 3, 1976 451
 $60
L

2 40
8

20

'0 10 20 K) 40 M 60 70 80
Inifla1 KCI COncenfrafiOn - moler/1000 moles H,O TEMPERATURE- 'C
Figure 1. Conversion of KC1 to Engel salt vs. initial KC1 concentration Figure 2. Conversion of KC1 to Engel salt vs. temperature (1atm of
at 18 and 25 "C, 1atm of COP:0,calculated from Bayliss and Koch's COz; initial KC1, 50 mo1/1000 mol of H20).
phase data; +, experimental data this study (24-25 "C).

achieved in the experimental reaction period. From these

calculated and experimental curves it is seen that the highest
equilibrium yield is obtained a t an initial KC1 concentration
of 50 mo1/1000 mol of H20 (at both 18 and 25 "C).
Another series of runs was performed a t 12,18, and 25 "C
to determine the effect of temperature on the equilibrium
conversion of KC1. These results, obtained a t the optimum
KC1 concentration of 50 mo1/1000 mol of H20, are shown in
Figure 2. The equilibrium conversion of KC1 is seen to de-
crease with increasing temperature. This is a result of the
thermal instability of Engel salt a t the higher temperatures.
Consequently, the reaction should be conducted a t the lowest
temperature which still provides an acceptable rate of reac-
tion.
In addition to reaction temperature and the initial con- 0 5 10 15 20 25
centration of KCl, a third variable, the pressure of COz on the COS Pressure -atm
reacting system, has a pronounced affect on equilibrium
conversion. As a suitable pressurized reactor was not available Figure 3. Influence of CO2 pressure on the maximum conversion of
KC1 to Engel salt (18 "C).
for this study, it was necessary to calculate the expected
equilibrium yields using the phase data and method of Bayliss
and Koch (1952). They used the equilibrium relationship for stirrer shafts are normally designed for pressures under 20
eq 3 atm. For this reason and the small increase in conversion
above 10 atm it is probable that the optimum pressure range
would be 10-20 atm of COz. An added benefit of operating a t
positive COz pressure should be an increase in reaction rate
and the equilibrium equation for the dissolution of MgC03- due to the increased concentration of Mg(HC03)2 in solution.
3Hz0 by COz From eq 12 it has been calculated that a tenfold increase in
+
(MgCO3.3HZO COz F! Mg2+ + 2HC03- + 2H20) COz pressure would approximately double the Mg(HC03)2
concentration. The increase in reaction rate produced by this
namely doubled concentration would depend on the nature of the
rate-controlling process (mass transfer or chemical' reaction).
(12) From the foregoing calculations and experimental results
it may be concluded that the optimum range of conditions for
and the equation relating ionic equilibrium between the four the production of Engel salt via eq 3 would be: a temperature
species of 18 "C or lower, an initial concentration of KCl of 50 mol/
1000 mol of HzO or higher, and a pressure of 10-20 atm of COz.
[Cl-] + [HC03-] = [K+] + 2[Mg2+] (13)
Temperatures below 18 "C will produce higher conversions
These three equations are solved simultaneously by trial and of KCl but lower rates of reaction. Conversely, concentrations
error and the value of [K+] thus obtained was then used to of KCl greater than 50 mo1/1000 mol of H20 will result in lower
calculate the conversion of KC1 to Engel salt. Details of the conversions of KCl but higher rates of reaction. An increase
calculation method are given by Smithson (1973). These cal- in COZ pressure should result in an increase in both the rate
culations have been performed for COz pressures of 2,3,5,10, of reaction and the conversion of KCl.
and 20 atm a t a temperature of 18 "C and are plotted in Figure Rate of Engel Salt Formation. The effect of temperature,
3. The equilibrium conversion increases with increasing COZ initial KC1 concentration, and the method of adding MgC03.
pressure but levels off rapidly a t pressures above 10 atm. 3H20 on the rate of Engel salt formation were investigated in
Stirred tank reactors with mechanical pressure seals on the a series of six experimental runs (Figure 4). In all cases, the

452 Ind. Eng. Chem., Process Des. Dev., Vol. 15,No. 3, 1976
 TIME - hours
"-
Y

TIME - hours
Figure 4. Conversion of reactable KC1 to Engel salt vs. time: 0,
MgC0~3H20added gradually throughout reaction; +, all MgCO? Figure 5. Comparison of several Engel salt formation rate curves with
3H20 added at start of reaction (1atm of COz in all runs; initial KC1, eq 15; 18" rate curve: 0 , MgCOr3HzO added gradually; 25" rate
50 mol/lOOO mol of H20 unless otherwise noted). curves: +, MgC0~3H20all at start, 21.75 mol of KClilOOO mol of H20;
0,MgCOr3HzO all at start, 50 mol of KC1/1000 mol of HzO.
percent conversion is expressed as a percentage of the equi-
librium conversion. A common characteristic of these rate reaction rate of 80% is observed while the increase is 60% a t
curves is the existence of an induction period prior to the 18 "C for the gradual addition of MgC03.3H20.
formation of Engel salt. This induction period varies from The increase in reaction rate with decreasing temperature
approximately 1.5 h a t 12 "C to 4 h a t 25 "C. The rate of re- is the result of two factors. The first is that as the temperature
action increases with increasing KC1 concentration and with is lowered the concentration of dissolved HC03- increases
decreasing temperature. The rate of reaction is also increased with a corresponding increase in the rate of Engel salt for-
by adding the MgCOY3H20 gradually throughout the reaction mation. The second is the result of the thermal instability of
rather than all a t the beginning. Engel salt. A t lower temperatures (12-18 "C) the rate of de-
The cause of the induction period prior to the formation of composition of Engel salt is much less than its rate of forma-
Engel salt is the slow rate a t which nuclei of this solid are tion. As the temperature is increased the rate of the chemically
generated. Formation of crystalline Engel salt requires the controlled decomposition reaction increases faster than the
simultaneous incorporation of K+, Mg2+,HC03-, C03- and rate of the formation reaction resulting in a net decrease in the
4H20 into the crystal lattice. The probability of spontaneous rate of Engel salt production. At some temperature slightly
nucleation in the solution will be very low. I t is much more above 27 "C the rates of the formation and decomposition
likely that nucleation begins on the surface of MgC03.3Hz0 reactions become equal resulting in zero net production of
by interaction with K+ and HCO3-. Once formed these nuclei Engel salt.
can continue growing by uptake of all four ionic species and Rate Equation for Engel Salt Formation. The shape of
water molecules. It has previously been noted that Engel salt the rate curves obtained in the conversion of KCI to Engel salt
formation occurs only when MgC03.3HzO is used as the is typical of reactions which are rate controlled by nucleation
magnesium carbonate reactant (Precht, 1900). This would be and crystal growth. In such a reaction the rate of formation
explained if the MgCO:l.3H20 crystal lattice is topotaxically of the precipitated solid will depend on its own surface area
compatable with the Erigel salt lattice. The induction period and the concentration of the reacting species. For the frac-
is reduced as the concentration of KC1 is increased because tional conversion XBthe remaining reactant will be propor-
of the increase in the production and growth of nuclei with tional to 1 - XBand the surface area of solid will be propor-
increased K+ concentration. The induction period also de- tional to x B 2 / 3 . From this, the rate of reaction can be ex-
creases as the temperature is lowered. This is most likely due pressed as
to the increase in dissolved HCOB- with decreasing temper-
ature and the effect of this increased concentration on the rate rate = k (1 - x B ) x B 2 ' 3 (14)
of nuclei formation and growth.
Integration of this expression results in the following inte-
From the two lower rate curves a t 25 "C it is seen that the
grated rate equation.
rate of reaction increases approximately 70% for a 2.3-fold
increase in KC1 concentration. I t is most probable that the
reaction is first order with respect to [K+] but this is not ob-
served here because of the inhibition of diffusion caused by 1
the high solids content of the slurry. The inhibition of mass + tan-' - = k t
4
(15)
transfer produced by a high level of MgC03-3H20 is readily
observed in the upper two 25 "C rate curves and the two 18 "C Plots of X B vs. t/tl,z for the experimental rates curves were
rate curves. In each case the lower curve is the result of adding compared to the corresponding plot for eq 15 in Figure 5. Al-
all the MgC03.3H20 a t the start of the reaction, whereas the though the experimental rate curves tend to bracket eq 15,
upper curve is observed when the MgC03-3HzO is added there is no general agreement with this relationship.
gradually throughout the reaction, At 25 "C an increase in A second relationship was derived in which the effects of

Ind. Eng. Chem., Process Des. Dev., Vol. 15,No. 3, 1976 453
 10
nuclei production and nuclei growth were introduced. Instead
of the simple proportionality between rate and surface area
(rate a x B 2 i 3 ) the rate was assumed to be proportional to the 08
total number of nuclei, A , plus a simple function of the surface
area, B X B . The term X B is used instead of XBzi3 because it
has been shown that where a solid has a log-normal size dis- 06
tribution the surface area is more closely proportional to XB
than X J ~ "(Smithson,
~ 1973).Substituting the expression A x,
+ BXB for XBzi3 eq 14 results in 04

rate = k (1- XB)(A BXB) + (16)

02
Integration of this expression results in the following inte-
grated rate equation.

In
A + BXB - In A = ht 0
0 10 20 30

1-xB
Plots of X B vs. t/tliz for eq 17 with various ratios of BIA are Figure 6. Variation of XB vs. t / t l / z for the equation In [ ( A+ BXB)/(1
shown in Figure 6. For small ratios of BIA, that is when a large - XB)] - In A = kt with various ratios of B / A
number of nuclei are present initially or the rate of nuclei
production is very rapid, eq 17 reduces to
- In (1 - XB) = ht (18)
In this case the rate of reaction is proportional to the con-
centration of reactant B in the solution. As the ratio BIA in-
creases the rate curves become more S-shaped approaching
a step function a t infinitely large values of BIA. This final
condition represents the case of a super-cooled melt which
undergoes sudden, complete crystallization when nucleation
is induced.
The most appropriate value of BIA for a particular set of
rate data can be determined by comparing the plot of X B vs.
tltllz with the curves in Figure 6 and selecting the ratio cor- 2 4 6 8 IO 12 14 16
responding to the closest match. Once BIA has been deter- TIME - hours
mined it is possible to calculate the left-hand side of eq 17 for
each value of X B and plot this value vs. time. A straight line
plot indicates good agreement between the experimental data
and the derived rate equation. The linear plots observed in
Figure 7 indicate reasonable agreement between the 18 and
25 "Crate data and eq 17. The 12 "C rate curve does not cor-
respond to a single value of B / A . The initial portion of this rate
curve is best fitted by BIA = 10 000 (nucleation control)
whereas the mid and latter portions correspond to BIA <
0.001 (crystal growth controlling). This may be the result of
both nucleation and crystal growth being limited by solution
mass transfer after a certain number of nuclei have been
generated. A t Bo, the rate curve corresponding to the addition -
TIME hours
of all MgC03-3H20 a t the start of the reaction has a smaller +
Figure 7. Plots of In [ ( A BXB)/(1 - XB)] - In A vs. time for the
value of B / A than when the M g C 0 ~ 3 H 2 0is added gradually, experimental Engel salt formation rate data. Selected best f i t of B / A
indicating a greater dependence on crystal growth or mass indicated on each curve.
transfer. Likewise, a t 25 "C the rate curves for the low KC1
concentration and for the complete, initial addition of The observed rate data are plotted in Figure 8. From these rate
MgC03.3HzO have lower values of B / A than the gradual ad- curves it is seen that the induction period has been completely
dition rate curve confirming the reduction in mass transfer eliminated at all three temperatures. That is, nucleation is no
under these conditions. longer a rate-determining factor. Consequently, eq 18 should
It is evident from these results that as the temperature is be applicable under these circumstances. Plots of -In (1 -
lowered or the concentration of KC1 is increased or the addi- Xp,) vs. time for the observed rate curves are shown in Figure
tion of MgC03.3Hz0 is regulated to increase the rate of re- 9. The excellent linear agreement indicates that eq 18 is
action, nucleation of Engel salt becomes rate limiting. Under obeyed. Equation 18 is a special case of the general eq 17 in
these conditions it should be possible to further increase the which surface area is so large that mass transfer of reactant
rate of reaction by adding sufficient seed crystals of Engel salt B to this surface becomes rate limiting. The small increase in
a t the start of the reaction. the rate constant in going from 12 to 18 "C (Figure 9) corre-
Seeded Engel Salt Formation Reaction. A set of exper- sponds to an activation energy of approximately 2 kcal/g-mol.
imental runs were performed a t 12,18, and 25 "C to observe This confirms that mass transfer is the rate-controlling pro-
the effect of adding seed crystals of Engel salt a t the start of cess for the formation of Engel salt under these conditions.
the reaction. In each run, 10 g of finely ground Engel salt At higher temperatures (i.e., 25 "C) the thermal decomposi-
(approximately 100 mesh) was added to 250 ml of 2.6 M KC1 tion of Engel salt which is a chemically controlled reaction
and the appropriate amount of MgC03-3H20 (depending on causes a reduction in the net formation of Engel salt. As pre-
temperature) was added gradually throughout the reaction. viously mentioned, a temperature is eventually reached (27

454 Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976
 TIME - hours

Figure 8. Conversion of reactable KCI to Engel salt vs. time (seeded

reaction).

2
Temperature - "C
Figure 10. Rate of formation of Engel salt vs. temperature (initial KCl
concentration, 50.2 mo1/1000 mol of H20).

2
Therefore, this simple comparison is greatly biased in favor
of the unseeded reactions.
In summary, the production of Engel salt can be optimized
with respect to the maximum conversion of KC1 and the
I
maximum rate of reaction by the following means. A signifi-
-
X"
cant increase in the rate of reaction is accomplished by adding
-
I
MgC03.3Hz0 gradually throughout the reaction rather than
-
c
I
all a t the beginning. Two other means of increasing the rate
I of reaction are to increase the concentration of KC1 and op-
erate the reaction a t a positive pressure of COS.Increasing the
concentration of KCl shifts the reaction from the maximum
conversion value but the decrease in conversion is small in
( comparison to the increase in rate obtainable. An increase in
KCl concentration from 50 to 80 mo1/1000 mol of HzO would
decrease the conversion of KC1 by 4%but should increase the
rate of reaction by up to 50%. Operating a t a positive pressure
of CO2 will increase the concentration of HC03- and thereby
1 2 3 4 5 6 7 8 increase the rate of reaction. At 10 atm of CO2 the concen-
TIME -hours
tration of HC03- is more than double that a t 1 atm of COz.
Figure 9. Integrated rate equation, -In (1 - X B ) vs. time for the Increasing the pressure of COn has the added benefit of in-
seeded, Engel salt formation reaction.
creasing the conversion of KC1 as well as the rate of reaction.
In going from 1 atm of CO2 to 20 atm of COz the conversion
"C) a t which the decomposition reaction exceeds the forma- of KCl a t 18 "C is increased from 69% to 89%. Also, the con-
tion reaction. At or above this temperature Engel salt cannot version of KC1 can be increased by lowering the temperature
be produced by the foregoing reaction scheme. from 18 to 12 "C. At 18 "C, 1atm of COz the conversion is 69%
A comparison of the reaction rates for the seeded and un- whereas a t 1 2 "C, 1 atm of COz it increases to 78%.
seeded reactions can be made on the basis of the maximum Using the foregoing optimized conditions it should be
slope of each rate curve after correcting for the total amount possible to increase the rate of reaction 2.5-fold, or from 1
of KCl converted. The maximum slopes for the curves in g-mol/l.-h to 2.5 g-moi/l.-h. In addition, the conversion of KCl
Figure 4 and 8 have been computed in terms of moles/(liter- could probably be increased from 69% to approximately 90%.
hour) and are plotted in Figure 10. The rate of the seeded re- B. Decomposition of Engel Salt. Decomposition
action increases rapidly from 25 to 18 "C and then decreases Method. The two most commonly referred to methods of
gradually a t temperatures below 17 "C. The gradual increase decomposing Engel salt are by heating or adding MgO to a
in reaction rate in going from 12 to 18 "C corresponds to the slurry of the crystals in water (Bhavnagary, 1962; Bayliss and
mass transfer control region. Above 18 "C, the chemical Koch, 1952; French Patent, 1931; Heimann, 1955; Precht,
rate-controlled decomposition reaction begins to limit the rate 1900; Precht, 1901; Weidbrauck, 1973). All attempts a t using
of reaction. The unseeded reaction rate increases gradually either of these methods or variations thereof resulted in fail-
with decreasing temperature but must level off below 12 "C ure. In all cases the M g C 0 ~ 3 H 2 0was contaminated with basic
as the rate of the unseeded reaction cannot exceed that of the magnesium carbonate and in many cases only basic magne-
seeded reaction. In addition, it must be remembered that the sium carbonate was produced. Other methods of decomposing
induction periods have not been taken into account when the double salt were attempted, such as the addition of KOH
determining these maximum rates for the unseeded reactions. or NH40H to an aqueous Engel salt slurry. Neither of these

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976 455
methods was successful. Another attempted method was the
application of vacuum to the slurry to cause decomposition
by removal of COz.
2(KHC03.MgC03-4HzO) + 2(MgC03*3H~O)
+ KzC03 + coz + 3H20 (19)

This method resulted in the immediate formation of basic

magnesium carbonate which caused severe thickening of the
slurry and a maximum decomposition of 42%.
From these preliminary experiments it became evident that
the recovery of M g C 0 ~ 3 H 2 0was favored by low temperatures
and low pH'i. On careful examination of Bayliss and Koch's
phase diagram for this system (Figure 11)the reason for the
requirement of a low pH became apparent. The region in
which MgC03.3HzO occurs as a stable solid phase lies close I
L
/
/
I
i - 4
1

$';
to the HC03- axis of the phase diagram. A high ratio of
HC03- to COS- necessitates a low pH. If this ratio is decreased
by increasing the pH, decomposition will take place in the
region where basic magnesium carbonate is the stable solid I
phase. The easiest method of maintaining a low pH and high
HC03- concentration is to saturate the solution with COS.
A few trial runs were performed in which a positive pressure
of COz (1-2 atm) was applied to an aqueous slurry of Engel
salt heated in a Parr low-pressure hydrogenation apparatus. Figure 11. Janecke phase diagram for the system K+-Mg2+-
For all temperatures up to 65 "C the only solid decomposition HC03--C0s2- at 25 "C from Bayliss and Koch.
product was MgC03-3H20. Above this temperature the
M g C 0 ~ 3 H 2 0was contaminated with basic magnesium car-
bonate. In all cases a solution of KHC03 was obtained as the
other decomposition product. Because of the time required
to assemble and disassemble the Parr reactor, it was not
possible to obtain accurate measurements of the rate of de-
composition. Consequently, rate measurements were per-
formed in an open stirred tank reactor as described in the
Experimental Section. The ratio of Engel salt to water was
selected such that there was just sufficient water t o allow all
Engel salt to decompose. The concentration of this equilib-
rium solution changes with temperature and was determined
by slurrying an excess of Engel salt in water a t the desired
temperature and analyzing the composition of the final so-
lution. The concentration of KHCO3 in these solutions at 31,
41, 51,61, and 71 "C is shown in Figure 12. The rate curves
obtained for the decomposition of Engel salt under these
conditions are shown in Figure 13. 0
3
'
0 4
I
0 v
I
) 6
I
0 7
I0 8 0
I 9
I
0 100
Normally, it will be desirable to decompose the Engel salt TmprOtW -Y
a t the highest temperature practical as this gives the maxi- Figure 12. Temperature dependence of the maximum solution con-
mum rate of decomposition and the highest concentration of centration of KHC03 attainable from the decomposition of Engel salt.
KHC03 in solution. The upper limit will be dictated by the
maximum temperature a t which MgC03.3HzO can be pro-
duced free of basic magnesium carbonate. In the trial runs I I I f

with the Parr reactor, pure MgCOy3H20 was produced a t 65 -

16-
"C (2-h reaction period) but a trace of basic magnesium car- 60q Enqcl 5011

bonate was formed at 70 "C (2-h reaction period). In the open

stirred tank runs no basic magnesium carbonate was evident
up to 71 "C. The difference between these two cases is that the
reaction time was only 20 min in the open stirred tank reactor.
Therefore, it may be concluded that a decomposition tem-
perature of 70 "C may be used as long as the reaction time is
not much over 20 min.
Decomposition of Engel salt a t 70 "C gives a maximum
KHC03 concentration of 14% when saturated with COz and
15% when no COz is applied (Figure 12). This difference is
caused by the higher level of dissolved Mg(HC03)2 present
in a solution saturated with COS. The dissolved Mg(HC03)z
I I I I
suppresses the solubility of KHC03. It is possible to circum- 10 20 30
vent this problem by saturating the Engel salt slurry with COz TIME- min
a t the start of the reaction and then cease further addition of Figure 13. Decomposition of 60 g of Engel salt in that volume of water
COz. Once the decomposition is started in the region where which produces the maximum attainable KHC03 concentration a t
MgC03.3HzO is the stable phase, the KHCO.1released in so- 31,41,51,61, and 71 OC.

456 Ind. Eng. Chem.,Process Des. Dev., Vol. 15, No. 3, 1976
 lrition will maintain the concentration of HC03- within the IO
desired region.
9
R a t e Equation for Engel Salt Decomposition. The rate
of decomposition of Engel salt could be rate controlled by ei- 8
ther a chemical reaction at the surface of the solid or by mass
transfer into solution. In the first case the rate of reaction 7
would be mainly dependent on the surface area of the solid.
For a solid composed of a distribution of particle sizes the 6
integrated rate equation would be (Smithson and Bakhshi, -EL
I-x.
5
1969)
-In (1 - XB) = k t (20) 4

In order to make plots involving XBit is necessary to convert 3

the amount reacted in Figure 13 to a fraction of the maximum
2
equilibrium concentration. Plots of -In (1 - XB) vs. time,
using these fractional equilibrium conversions, are not linear. I
Therefore, surface chemical reaction control may be rejected.
Mass transfer rate control for the dissolution or decompo-
'0 I 2 3 4 5 6 7 8 9
sition of a granular solid is dependent on the surface area of
~

TIME- min
the solid and the concentration differential of the dissolving
species between the solid surface and the bulk solution con- F i g u r e 14. Integrated rate equation plots [X,/'(l - XB)]
= k t for the
Engel salt decomposition rate data.
centration. Surface area is proportional to (1 - Xg)2'3 but
when a distribution of particle sizes exists, the change in
surface area is more closely proportional to (1 - X B ) . The
difference in concentration of the diffusing species a t the
surface of the solid and in the bulk of the solution will be
proportional to (1 - X B ) . From these two proportionalities
the rate may be expressed as
rate = k ( 1 - x B ) ( 1 -XB) (21)
Integratinn of eq 21 results in the following integrated rate
equation.

(22)

Plots of XB/(1 - X,) vs. time for the fractional equilibrium

conversions calculated from Figure 13 are shown in Figure 14.
The straight line agreement of the experimental data in these
plots indicate that mass transfer is the probable rate-con-
I I I I I
trolling process. -20
28 29 30 31 32 33
It is possible to confirm mass transfer control if the acti- pT x 10'
vation energy of the reaction is below 6 kcal/g-mol. An Ar-
rhenius plot of the rate constants determined in the Figure F i g u r e 15. Arrhenius plot of rate constants, k , from the Engel salt
14 plot is presented in Figure 15. A straight line relationship decomposition reaction (Figure 14).
exists from 41 to 71 "C and results in an activation energy of
2.3 kcal/g-mol. This definitely confirms mass transfer as the
rate-controlling process in the decomposition of Engel salt at I ' I 'I I f I I
temperatures above 40 "C. Below 40 "C the activation energy W. W 70. BO' 50. 40' W 20' 10. QC

appears to increase dramatically. It is in this region that the

decomposition reaction becomes chemical rate controlled.
This has been pointed out previously in the formation of Engel
salt that as the temperature is increased above 18 "C the de-
composition of Engel salt (a chemically controlled reaction)
increases rapidly, reducing the net rate of its formation. At
some temperature above 27 "C the rate of the decomposition
reaction exceeds that of the formation reaction. As the tem-
perature is increased further, the rate of the chemically con-
trolled decomposition reaction increases until it is limited by
mass transfer (above 40 "C). The relationship between the
mass transfer formation region, the chemical reaction con-
trolled decomposition region, and the mass transfer controlled
decomposition region are illustrated in Figure 16.
In summary, the optimum conditions for the decomposition
of Engel salt are 100 g of Engel salt/243 ml of water heated to I I I I 1 I I I I I
70 "C. The water must be saturated with COn a t the start of
the reaction to ensure the formation of MgC03-3H20 kT 10'
throughout the decomposition. No further C02 need be added F i g u r e 16. Descriptive representation of the various reaction rate
during the course of the decomposition as the dissolution of control regimes in the formation and decomposition of Engel salt.

Ind. Eng. Chern., Process Des. Dev., Vol. 1 5 , No. 3, 1976 457
KHC03 will maintain the required HC03- concentration. The C, = the concentration of KC1 in solution a t the time of the
time required for complete decomposition is from 20 to 30 min removal of the nth sample aliquot, g/ml
and results in a solution of 15 wt % KHCOs. J = the correction for the decrease in volume near the end
of the reaction when all solid MgC03.3HzO has been con-
Conclusions sumed and the concentration of Mg(HC03)z in solution
begins to decrease; J = L ( X E - X E ~ ml;
) , J = 0 when X E
1. The rate of formation and the yield of Engel salt pro- -XE1<o
duced from a slurry of MgC03.3HzO in a C02 saturated so- L = the volume contribution due to dissolved Mg(HC03)2,
lution of KC1 are favored by: a low solids (MgC03.3H20) ml
content, a high level of KC1, a positive pressure of COz, the u = the volume of each sample aliquot, ml
presence of seed crystals of Engel salt, and a low temperature vk = the increase in solution volume due to the complete
(12-18 "C). Under optimized conditions it should be possible dissolution of the KHC03 contained in the initial Engel salt,
to,achieve a rate of production of 2.5 g-mol/l.-h and a 90% ml
conversion of KCl to Engel salt. The seeded Engel salt for- ( VmIn = the increase in solution volume due to dissolved
mation reaction is mass transfer controlled a t temperatures Mg(HC03)Z a t the time of the removal of the n t h sample
aliquot, ml
up to 18 "C. Above 18 "C, the rate of reaction becomes limited V , = the total volume of solution present when the nth ali-
by the chemically controlled thermal decomposition of Engel quot is removed from the reacting slurry, ml
salt. The rate of thermal decomposition increases with tem- ( V s ) , = the total volume of solution a t the time of the re-
perature causing the net rate of Engel salt formation to de- moval of the nth sample aliquot, ml
crease to zero a t approximately 27 "C. V , = the initial volume of water, ml
2. The thermal decomposition of an aqueous slurry of Engel Va = the volume change in going from the initial KC1 solu-
salt is rate controlled by a chemical reaction up to 40 "C and tion to the final equilibrium solution of MgC12 and KC1, ml
by mass transfer above this temperature. Mass transfer rate ( W K ) O= the weight of KHC03 contained in the initial Engel
control above 40 "C is confirmed by an activation energy of salt, g
Wo = the initial weight of KC1, g
2.3 kcal/g-mol. Optimum decomposition conditions are 100
X B = the fractional conversion of component B
g of Engel salt/243 ml of water a t a temperature of 70 O C re- (XDIn = the fractional decomposition of Engel salt at the
sulting in a 15% (w/v) solution of KHC03. By initially satu- time of the removal of the n t h sample aliquot
rating the decomposition slurry with COz the solution com- ( X E ) =~ the fraction of X,,achieved a t the time of the re-
position is maintained in the phase region where M g C 0 ~ 3 H 2 0 moval of the n t h sample aliquot
is the only stable solid. Once the decomposition is initiated X E =~ X,,minus the value of X E which is stoiciometrically
in this phase region the addition of COz can be discontinued equivalent to the amount of dissolved Mg(HC03)z
as the production of dissolved KHC03 maintains the solution X,, = the fraction of the initial KC1 which is converted to
composition within the desired region. Engel salt a t equilibrium

Acknowledgment Literature Cited

Bhavnagary, H. M., Chem. Age India, 19, 110 (1962).
This project was carried out in the laboratories of the Sas- Bayiiss, N. S., Koch, D.F. A , , Aust. J. Appl. Sci., 3,237 (1952).
katchewan Research Council. The authors thank Brad Gunn French Patent 715 007 (1931).
Heimann, H. H., Bull. Res. Council Israel, 5c, 113 (1955).
and J. H. Hudson for their interest and time spent in dis- Precht, H., German Patent 143 408 (1900).
cussing the interpretation of data in this study. Precht, H., British Patent 2 343 (1901).
Smithson, G. L., Bakhshi, N. N., Can. J. Chem. fng., 47,508 (1969).
Smithson. G. L.. Bakshi, N. N., lnd. Eng. Chem., Process Des. Dev., 12,99 (1973).
Nomenclature Smithson, G.L., P h D Thesis, University of Saskatchewan, 1973.
Weidbrauck, E., Chem. Tech. (Berlin), 15, 188 (1942).
A = a factor proportional to the total number of Engel salt
nuclei present a t a given time
B = a factor proportional to the surface area of Engel salt Receiued for reuieu: July 11,1975
Accepted M a r c h 2 7 , 1976
crystals a t a given time
(CK), = the concentration of dissolved KHC03 a t the time Presented a t the VIth Interamerican Congress of Chemical E n g i -
of the removal of the n t h sample aliquot, g/ml neering, Caracas, Venezuela, July 13-16, 1975.

458 Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976