Thermal stability of potassium carbonate near its melting point

Article abstract of DOI:10.1016/S0040-6031(98)00289-5

The behavior of potassium carbonate was studied by thermal analysis near its melting point under 1 atm total pressure. Literature data are scant regarding the degree to which K₂CO₃ decomposes in sub- and pro-liquidus thermal environments, and reliable thermodynamic data are not available for the reaction K₂CO₃?K₂O+CO₂ in the temperature range of interest. TGA and DTA analyses indicate that significant volatilization occurs below the melting point and the logarithm of the rate of volatilization is a linear function of 1/T above the melting point. The effect of CO₂ partial pressure was highly significant in reducing, but not eliminating, volatilization near the melting point. Graphs of ln[P_CO2×weight loss rate] vs. (1/T) were linear, supporting a decomposition model rather than congruent volatilization. The melting point of K₂CO₃, as measured by DTA, is 905°C in CO₂ and 900°C in N₂.

Full text of DOI:10.1016/S0040-6031(98)00289-5

Thermochimica Acta 316 (1998) 1±9
Thermal stability of potassium carbonate near its melting point
a,*
Richard L. Lehman , Jefffery S. Gentry , Nick G. Glumac
1b
a
a
Rutgers University, Piscataway, NJ 08855, USA
b
RJ Reynolds, Inc., Winston-Salem, NC 27102, USA
Received 7 August 1995; received in revised form 8 January 1998; accepted 27 January 1998
Abstract
The behavior of potassium carbonate was studied by thermal analysis near its melting point under 1 atm total pressure.
Literature data are scant regarding the degree to which K CO decomposes in sub- and pro-liquidus thermal environments, and
reliable thermodynamic data are not available for the reaction K CO „K O‡CO in the temperature range of interest. TGA
2
3
2
3
2
2
and DTA analyses indicate that signi®cant volatilization occurs below the melting point and the logarithm of the rate of
volatilization is a linear function of 1/T above the melting point. The effect of CO partial pressure was highly signi®cant in
2
reducing, but not eliminating, volatilization near the melting point. Graphs of ln[PCO Âweight loss rate] vs. (1/T) were linear,
2
supporting a decomposition model rather than congruent volatilization. The melting point of K CO , as measured by DTA, is
2
3
9
058C in CO
2
and 9008C in N
2
. # 1998 Elsevier Science B.V.
Keywords: Potassium carbonate; K
2
CO ; Decomposition; Melting behaviour
3
1
. Introduction
Potassium carbonate is an important industrial com-
modity used in a wide variety of processes and pro-
atmosphere on the stability of K CO near the melting
2 3
point under various CO partial pressures and 1 atm
2
total pressure.
Of the many compounds that exist in the compre-
hensive K±H±C±O system, four anhydrous com-
pounds, K O, KHCO , KOH, and K CO are of
ducts. The speci®c behavior of K CO between 7008±
1
examined in the present work.
Although an enormous amount of literature exists
on the physical and chemical properties of potassium
carbonate, only a paucity of consistent data is avail-
able at elevated temperatures under the foregoing
conditions. Therefore, the present study was con-
ducted to elucidate the effect of temperature and
2
3
0008C and under various CO partial pressures was
2
2
3
2
3
principal interest under ordinary ambient environ-
ments. Potassium hydrogen carbonate, KHCO , forms
3
under certain conditions of H O and CO2 partial
2
pressures, but it decomposes rapidly between 1008
and 2008C. Potassium hydroxide, KOH, which forms
under conditions of suf®cient moisture and low CO2
partial pressure, melts at 3608C and boils at 13168C.
K CO occurs in two forms, g and b. The g phase
2
3
transforms to the b form at 4218C and the b form is
stable up to the melting point, near 9008C.
The carbonate thermally decomposes at high tempera-
*
Corresponding author. Tel.: 001 908 445 2317; fax: 001 908
4
0
45 5584; e-mail: rllehman@rci.rutgers.edu
1
Tel.: 910-741-4518.
040-6031/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved
P I I S 0 0 4 0 - 6 0 3 1 ( 9 8 ) 0 0 2 8 9 - 5

2
R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
tures according to the following reaction,
422Æ28C. The melting temperature reduced to 891±
8
968C, when tested under air instead of CO .
2
K CO „ K O ‡ CO
2
3
2
2
Of particular interest in the present study is the
degree to which K CO volatilizes near the melting
2 3
although considerable uncertainty is re¯ected in the
literature regarding the temperature and rate at which
this reaction proceeds. Some studies [1] indicate that
the reaction is negligible below the melting point of
point and whether the volatilization is congruent
evaporation or decomposition. As already indicated,
the literature is quite vague in this regard. Although it
9
and varying degrees of dissociation [4].
008C, whereas others note thermal instability [2,3]
is clear that, well above the melting point, K CO3
2
decomposes according to Eq. (1), the thermal onset of
this reaction, the degree to which it occurs, and the
inhibiting effect of CO overpressures have not been
Of the numerous studies on K CO , most are for
2
3
mixtures of salts [5,2,6,3], or for molten salts [7].
Several studies are devoted to crystallography and
phase transitions [8,9]. Only a few studies have
addressed the thermal-analysis behavior of pure
K CO . The study of K CO by Reisman [3] was part
2
well documented. The only data, found in the litera-
ture, which de®ne the vapor pressure over K CO are
2
3
given in Fig. 1 [10] and suggest a low rate of decom-
position near the melting point (13 mbar is achieved
only by 12008C). Lee and Ladd [10] assumed that the
decomposition reaction (Eq. (1)) was applicable.
2
3
2
3
of a more general investigation, and his work on
potassium carbonate remains the most comprehensive
examination of the high-temperature behavior of this
compound. Most of Reisman's thermal-analysis
experiments with the carbonates were done under
Across the entire temperature range, K CO3 has
2
low volatility rates compared to Na CO , a carbonate
2
3
salt known for its signi®cant volatility. The goal of the
present study was to re®ne our knowledge of the
decomposition behavior of K CO by thermal-analy-
1
atm CO because of ``the varying degrees of dis-
2
sociation they undergo at elevated temperatures.'' The
melting temperature of K CO was found to be
2
3
2
3
sis studies in the presence of varying CO /N atmos-
2
2
9
01Æ18C, with a single solid±solid transition at
pheres.
2
. Thermodynamic assessment of
K CO decomposition
2
3
Prior to commencing experimental studies, an
assessment of the behavior of K CO and related
2
3
compounds was conducted using thermodynamic pro-
perties from the literature. Considering the presence of
CO overpressure and the existence of small amounts
2
of ambient H O, ꢀ20 pertinent chemical equations
2
were identi®ed in the K±O±H±C system. Of these, the
following four were selected as relevant to the heating
of K CO under the experimental conditions:
2
3
K CO „ K O ‡ CO
K2O ‡ H2O „ 2KOH
K2CO3 ‡ CO2 ‡ H2O „ 2KHCO3
(1)
(2)
(3)
(4)
2
3
2
2
2
KOH ‡ CO2 „ K2CO3 ‡ H2O
The stability of K CO and the degree to which other
2
3
compounds listed in Eqs. (1)±(4) exist under given
conditions can best be identi®ed outside the laboratory
by evaluating thermodynamic quantities. To accom-
Fig. 1. Vapor pressure of alkali carbonates.

R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
3
plish this preliminary task, the free energies for each
reaction, Á G, were determined by calculations using
or,
r
0
Á GꢁT† ꢂ Á G ꢁT† � RT lnPH2OP
r
r
CO2
the most recent and reliable data from the literature
0
ˆ
� 115450 � 136:53T ‡ 88:06T
ꢁln T � 1† � 8:3144ꢁT†lnꢁP
[
lated in the standard way, viz.:
11]. Á G for reaction aA‡bBˆcC‡dD was calcu-
r
P
CO2 †
H2O
0
0
(10)
ÁrG ꢁ298 K†; l ˆ Áf G ꢁproducts†
0
�
ÁfG ꢁreactants†
(5)
At
Tˆ298 K,
and
1
PCO ꢂ PH2O ˆ 0:1 bar,
2
�
Á Gˆ� 21.7 kJ mol , thus indicating the reaction
r
The standard free energies of formation of principal
compounds addressed in this study are given in
is favorable for the forward direction under these
conditions. For general purposes, it is desirable to
know the locus of points in the parametric space of T,
PH O and PCO in which Á Gˆ0, thus de®ning
0
r
Table 1. To calculate ÁG for reactions at tempera-
tures other than 298 K, Eq. (6) was used:
2
2
r
T
the boundary between favorable forward- and
Z
0
0
0
back-reactions. The loci of Á Gˆ0 points, for the
ÁrG ꢁT† ˆ ÁrG ꢁ298 K† ‡
ÁCpdT � T
r
three important reactions illustrated in Fig. 2(A)±
(
2
98 K
C), are useful and signi®cant curves since they de®ne
T
Z
ÁCp
the regions above and below the line, where ÁG<0
and ÁG>0, and as such are regions in which products
or reactants will be observed. Using this type of
analysis, KOH and KHCO3 were precluded from
any further consideration as compounds present near
the melting point under the thermal conditions and
atmospheres of interest, as illustrated by Fig. 2(A)
and (B).
0
dT
(6)
0
T
2
98 K
where ÁC is calculated from the heat capacities of
p
the products and reactants. In practice, it was most
0
convenient to obtain Á G (T) from the following
r
polynomial expression,
0
ÁrG ꢁT† ˆ a ‡ bT ‡ cTꢁlnT � 1†
using the well-established coef®cients, a, b, and c,
(7)
Fig. 2(C) illustrates that Á G>0 for the K CO
3
r
2
decomposition reaction (1) for all temperatures shown
and, thus, the reaction will not proceed in the forward
direction. However, the thermodynamic data for reac-
tion (3) only apply below 8808C, since the published
from Table 1.
For a chemical reaction involving gas-phase species
not at equilibrium, the relation between the free
data for K O are not available for higher temperatures.
Extrapolation of the data in Table 1 to higher tem-
peratures leads to obvious error. For example, calcula-
2
energy for the reactions, Á G(T), the standard free
r
0
energy of the reaction, Á G , and the partial pressures
r
of gaseous reactants and products at temperature T is
tion for Á G for P
ˆ 1:3 mbar at 9008C yields
r
CO2
given by:
�
1
� 1
c
d
b
‡117 kJ mol and at 12008C, Á
r
Gˆ‡38 kJ mol
,
ꢁ
ꢁ
Pc=Pt† ꢁPd=Pt†
0
ÁrGꢁT† ˆ ÁrG ꢁT† ‡ RT ln
indicating non-spontaneous forward-reactions under
temperature conditions which surely promote the
forward reaction. Thus, the thermodynamic approach
to evaluating reaction (1) was limited to temperatures
a
P =P † ꢁP =P †
a
t
b
t
(
8)
where the pressures in parentheses represent the par-
tial pressure of each gaseous component, and P the
<
8808C and no estimate was obtained by thermo-
t
dynamic calculations in the temperature range of
interest.
total pressure. Given partial pressures and tempera-
ture, the free energy change of the reaction, Á G can
r
be calculated.
Consider the reaction, K CO ‡CO ‡H O„
2
3
2
2
3
. Thermogravimetric analysis (TGA)
2
KHCO as an example:
3
1
0
To further de®ne the behavior of K CO near the
2
ÁrGꢁT† ˆ ÁrG ꢁT† ‡ RT ln
(9)
3
2
PH OPCO2
melting point, thermogravimetric analysis was per-

4
R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9

R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
5
Fig. 2. (A) Regions of favorable forward- and back-reactions based on free energy calculations. K
of favorable forward- and back-reactions based on free energy calculations. 2KOH(liq)‡CO „K
forward- and back-reaction based on free energy calculations. K CO O‡CO
„K
2
CO
3
‡CO
2
‡H
2
O„2KHCO
3
. (B) Regions
2
2
CO
3
‡H
2
O. (C) Regions of favorable
2
3
2
2
.
2
formed with TGA equipment having 1 mg resolution,
2
were 30 mg. The geometric surface area of the sample
2
was 0.64 cm and the sample chamber was purged
mg sensitivity, Æ5 mg accuracy, and thermal preci-
sion of 0.58C. Thermal calibration was veri®ed using
the melting point of a silver standard (mpˆ961.98C).
Gravimetric calibration was veri®ed by calcium oxa-
late decomposition.
with gas continuously during the experiments at the
�
1
rate of 1 l min . Specimens were heated at a rate of
�
1
208C min to ꢀ8508C, where a 108C ramp was
initiated leading to a 45 min isothermal period near
the melting point. Successive isotherms at ꢀ9508C
and 10008C were reduced to 30 min duration to pre-
vent excessive volatilization and reduction of the
geometric surface area. Weight-loss curves were
recorded during these periods as shown in Table 2
and Fig. 3.
After careful calibration of the instrument, a series
of TGA runs were made in atmospheres ranging from
3
pure CO to pure N . Certi®ed K CO was placed on
platinum pans in the TGA and typical sample weights
2
2
2
3
2
TA Instruments, New Castle, Delaware, 302-427-4000, Model
The TGA curves were analyzed using the analysis
software provided with the thermal analysis equip-
TGA 951.
3
Fisher Scientific, Fairlawn, NJ 07410, 201-796-7100.

6
R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
�
1
Table 2
sphere and � 35611 K for CO atmosphere. Stan-
2
� 1
TGA data for K
mixed CO
2
CO tests at elevated temperatures in CO, N , and
atmospheres (total pressureˆ1.01 bar; geometric
3
2
� 1
and 3784 K ,
dard errors were 1391 K
respectively. Thus, activation energies, Q, for the
2 2
/N
2
surface area of sampleˆ0.64 cm )
volatilization processes were determined to be
� 1
CO
2
Temperature/
K
Sample
weight/
(mg)
Rate of
ꢀ
300 kJ mol
The effect of CO2 partial pressure was highly
.
pressure/
bar
weight loss/
�
1
C
(mg min
)
signi®cant in reducing the rate of volatilization in this
experiment, supporting the generally recognized nat-
ure of the volatilization process. At ꢀ10008C, the rate
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1154
1185
1210
1183
1221
1264
1186
1225
1268
881
912
937
910
948
991
913
952
995
41.982
41.982
41.982
42.045
42.045
42.045
42.484
42.484
42.484
3.67
11.53
24.16
10.14
28.41
80.05
9.52
�
1
of volatilization in pure N (90 mg min ) is nearly
2
�
1
four times the rate under pure CO (25 mg min ). The
relationship between the rate of weight loss and the
CO partial pressure is an important one since it sheds
2
2
32.50
90.19
light on whether the volatilization from K CO solid
2
3
or liquid is congruent sublimation or evaporation, or
whether the decomposition reaction is occurring. The
difference is important with regard to the chemical
behavior of the vapor phase in certain industrial
processes and products. If Eq. (1) represents the vola-
tilization process, then at equilibrium:
0
0
0
0
0
0
.4
.4
.4
.4
.4
.4
1181
1219
1264
1186
1225
1268
908
946
991
913
952
995
40.563
40.563
40.563
41.187
41.187
41.187
10.65
21.10
46.32
8.28
19.81
46.91
ꢀ
ꢁ
aK OPCO
0
2
2
�
ÁrG ˆ RT ln
(11)
0
0
0
0
0
.7
.7
.7
.7
.7
1187
1226
1226
1269
1270
914
953
953
996
997
41.133
41.133
40.519
41.133
40.519
6.07
13.50
9.93
aK2CO3
where aK O and aK CO represent the activities of those
phases. If the activity of K CO is constant and the
2 3
2
2
3
34.53
23.30
volatilized K O is part of the weight loss, then a plot of
2
ln[PCO2 Âweight loss rate] vs. [1/T] should be linear.
1
1
1
1
1
1
1
.0
.0
.0
.0
.0
.0
.0
1155
1175
1184
1185
1189
1201
1211
1222
1228
1231
1267
1272
882
902
911
912
916
928
938
949
955
958
994
999
41.364
27.521
38.847
41.364
41.222
27.521
41.364
38.847
41.222
27.521
38.847
41.222
0.91
3.94
5.16
2.97
3.03
6.26
4.67
8.78
9.85
11.31
19.49
25.14
Fig. 5 presents the CO -dependent weight loss data in
2
this form with a linear best-®t line. The quality of the
2
regression ®t, R ˆ0.91, provides support to the
decomposition model. Indeed, if congruent volatiliza-
tion were occurring, the rate of weight loss would be
independent of PCO . The slope of the line, an estimate
2
of � Á G /R, is � 32,433 K� 1
0
,
or Á G ꢃ270
0
1
1
1
1
1
.0
.0
.0
.0
.0
r
� 1
r
kJ mol
.
The issue of congruent or incongruent volatilization
of K CO could perhaps be best resolved by X-ray
2
3
diffraction analysis of the volatilization products.
Unfortunately, this approach presents serious experi-
mental obstacles which, which could not be overcome
ment. The natural logarithm of the weight loss,
�
1
expressed as mg min , was calculated and plotted
vs. 1/T in Fig. 4 for pure CO and pure N atmo-
in the present work. The high reactivity of K
moisture and collection substrate materials near
9008C, the dissociation of K O, and the reversible
2
O with
2
2
spheres. Graphical representation of this form permits
estimation of the activation energy of the process from
the slope. Indeed, the relationship was linear through-
out the temperature range studied and the slope of the
best ®t line was calculated by linear regression. The
slope, � Q/R, was � 40101 K� for nitrogen atmo-
2
nature of reaction (1) all confound the collection and
identi®cation of decomposition products. Given these
dif®culties, the authors believe the use of CO
pressures to be the most practical approach to probing
the thermal behavior of K CO at high temperatures.
over-
2
1
2
3

R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
7
2 3 2 2
Fig. 3. Thermogravimetric analysis of K CO under varied CO /N atmospheres.
2 3 2 2
Fig. 4. Volatile weight loss from K CO during thermal analysis at elevated temperatures under varied CO /N atmospheres.
4
. Differential thermal analysis [DTA]
ments were conducted to determine the melting point
under CO and N , as shown in Table 3. Rapid heating
2
2
�
1
Differential thermal analysis was conducted con-
currently with the TGA studies to determine the
melting point of K CO under N and CO atmo-
spheres. DTA equipment with an accuracy of Æ18C,
sensitivity of 0.58C, and thermal resolution of 0.18C
was calibrated with a silver standard mpˆ961.98C.
The sample size was ꢀ90 mg. A total of 16 experi-
rates of 208C min were used up to 8508C, where the
apparatus was held for equilibration. From this point
onward, heating was commenced at the speci®ed
heating rate to 9208C at which time the sample was
cooled at the same rate to 8808C. The endotherm and
exotherms of melting and freezing were measured and
the software onset utility was used to determine the
precise point of the melting or freezing process. The
values achieved during heating and cooling tend to
converge on a single value as the heating rate
2
3
4
2
2
4
Scanning calorimeter 912 equipped with 1600o C DTA cell. TA
Instruments, New Castle, DE, 302-427-4000.

8
R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
Fig. 5. Volatile weight loss from K
decomposition.
2 3 2
CO during thermal analysis at elevated temperatures. The effect of CO partial pressure on
Table 3
Experimental design for melting point determination
K CO below the melting point and a linear relation-
2
3
ship between the logarithm of weight loss and reci-
procal temperature in the temperature range studied.
Substitution of CO for N reduced the rate of vola-
Test Heating rate/
8C/min)
Direction Atmosphere
Replication
(
2
2
tilization but measurable losses still occurred at all
temperatures. Based on the temperature and CO2
partial pressure dependence of the weight loss, the
results support a K CO decomposition model rather
1.
2.
3.
4.
5.
6.
7.
8.
2
2
1
1
2
2
1
1
heating
cooling
heating
cooling
heating
cooling
heating
cooling
nitrogen
2
2
2
2
2
2
2
2
nitrogen
nitrogen
nitrogen
2
3
carbon dioxide
carbon dioxide
carbon dioxide
carbon dioxide
than a congruent sublimation/evaporation model. The
melting points of K2 O were 9058C in CO and
C
3
2
9008C in N2.
Acknowledgements
approaches zero. The differences in melting point
�
1
values obtained by heating and cooling at 18C min
The authors wish to thank Dr. Jie Li and Mr. Robert
Zolandz for assisting in the thermodynamic calcula-
tions and preparation of the manuscript, and to Paul
Bernasek for contributing to the thermal analysis
measurements. This work would not have been pos-
sible without the administrative and technical support
of Dr. David E. Townsend who provided essential
material and intellectual support.
were quite small, <18C, and these melting point values
were averaged to produce the reported results. The
melting points were determined to be 9058C under
CO and 9008C under nitrogen.
2
5
. Summary and conclusions
Thermal analysis of potassium carbonate was con-
ducted at 1 atm total pressure between 8808 and
0008C to elucidate the behavior of the solid and
the liquid, particularly with respect to volatilization.
Available thermodynamic data are not valid in this
temperature range and were of little use in predicting
behavior for this particular reaction. Thermogravi-
References
1
[
1] O. Knacke, O. Kubaschewski, K. Hesselmann (Eds.),
Thermochemical Properties of Inorganic Substances, I, 2nd
edn., Springer-Verlag, Heidelberg, 1991.
[
2] R.E. Mills, R.T. Coyle, Differential thermal analysis of phase
equilibria in mixtures of anhydrous carbonates. III. K
Ba , Thermochim. Acta 124 (1988) 89±98.
2 3
CO ±
metric analysis in pure N revealed weight loss from
2
C 3
O

R.L. Lehman et al. / Thermochimica Acta 316 (1998) 1±9
9
[
[
[
[
3] A. Reisman, Reactions of the Group VB pentoxides with
alkali oxides and carbonates. IX. A DTA study of alkali metal
carbonates, J. Am. Chem. Soc. 80 (1958) 3558±3561.
4] G.J. Janz, M.R. Lorenz, Solid±liquid phase equilibria for
mixtures of lithium, sodium, and potassium carbonates, J.
Chem. Eng. Data 6(3) (1961) 321±323.
[8] H.Y. Becht, R. Struikmans, A monoclinic high-temperature
modification of potassium carbonate, Acta Cryst. B32 (1976)
3344±3346.
2 3
[9] S.J. Schneider, E.M. Levin, Polymorphism of K CO , J. Am.
Ceram. Soc. 56(4) (1973) 218±219.
[10] W.H. Lee, J.F.C. Ladd, Supplement to Mellor's Comprehen-
sive Treatise on Inorganic and Theoretical Chemistry, Vol. II,
Supp. III, The Alkali Metals, Part 2, Section LVI: Potassium
carbonate, Longmans, Green and Co. Ltd, London, p. 1893,
1963.
5] S. Mukhopadhyaya, K.T. Jacob, Thermodynamic study of
mixed anionic solid solutions using gradient solid electro-
lytes, J. Electrochem. Soc. 140(9) (1993) 2629±2635.
6] M.S. Kumari, E.A. Secco, IV. Order-disorder transitions:
Solid state kinetics, Thermal analysis, X-ray diffraction and
[11] M.W. Chase, Jr., C.A. Davis, J.R. Downey, Jr., D.J. Frurip,
R.A. McDonald, A.N. Syverud, JANAF Thermochemical
Tables, 3rd edn., Supp. No. 1, Part I and II, J. Phys. Chem.
Ref. Data, vol. 14, 1985.
2 4 2 4
electrical conductivity studies in the Ag SO ±K SO system,
Can. J. Chem. 63 (1985) 324±328.
[
7] G. Moutiers, M. Cassir, J. Devynck, Oxygen reduced species
in molten Li CO CO (42.7‡57.3 mol%) at 6508C, J.
‡K
Electroanal. Chem. 324 (1992) 175±189.
2
3
2
3
[12] D.R. Lide (Ed.), Handbook of Chemistry and Physics, 71st
edn., CRC Press, 1990.