295
J. Phys. Chem. 1982, 86, 295-297
Solid-state Reactivity in the System Na,CO,-CaCO, P. K. Gallagher' and D. W. Johnson, Jr. Bell Laboratories, Murray Hlll, New Jersey 07974 (Received: March 6, 1981; I n Flml Fwm: August 27, 1981)
The extent of formation of NazCa(C03)zwas measured by differential scanning calorimetry. Comparison was made between isothermal experiments and experiments in which the temperature was cycled over ranges which included phase transitions within NazCa(C03)z.It w& concluded that there is a slight increase in the extent of the formation reaction under the cyclic conditions which is interpreted in favor of the existence of the Hedvd effect. Thermogravimetrywas used to follow the thermal decomposition of CaC03,NazCa(C03)2 and an equimolar physical mixture of NaZCO3and CaC03. The latter was found to decompose at the lowest temperature. This too was interpreted as an enhanced reactivity caused by the structural disruption induced by the reaction to form Na2Ca(C03)z.
Introduction In his classic review article on the reactivity of solids, Hedvalll concluded that "every interference with the crystal lattice which sufficiently decreases its interior stability, increases the mobility of the units of the lattice, and therefore more or less stimulated the crystal to react with another crystal". He went on to illustrate a variety of mechanisms which induce this enhanced reactivity, e.g., (1)heating, (2) deformation, (3) impurities in the lattice, (4) radiation, and (5) changes in crystal structure accompanying phase transitions and decompositions. This latter mechanism has become closely associated with him and is usually refereed to as the Hedvall effect. He expressed it as "crystal which a t a given temperature sustains a change of phase in the form of a polymorphous transition or in any other form, ought to show, during this process when the particles have an abnormally great freedom of movement, an abnormally great reactivity when mixed with another reactive substances, whether solid, liquid, or gaseous". This definition, however, leaves some ambiguity concerning the time scale of the event. Does it pertain only to the momentary destruction of the original crystal lattice or does it include the entire time from between the disappearance of the initial lattice and the formation of the stable equilibrium form of the product? The extant view of the Hedvall effect is the former; the enhanced activity exists only during the changes in state. This time frame is at most the observed period of a peak in DTA or DSC. But the examples described by Hedvall suggest that it is the latter interpretation. Included in this broad interpretation would be the effects of any residual strains or increased surface area (microcracksor pores) resulting from the transformation. Differences introduced by the changes in structure and bonding between the two phases which are reflected in changes of activation enthalpy and entropy that persist well beyond the transition zone are not part of the Hedvall effect. The observed magnitude of the effect is not great, invariably less than an order of magnitude enhancement in the rate. Besides the examples given by Hedvall,' the self-diffusion of P b in PbSiO, is an example of this effect where the diffusion constant is significantly increased during the first order crystallographic transition around 585 0C.2 The effect has also been observed during second-order transitions such as magnetic alignment. Simple (1)J. A. Hedvall, Chem. Reu., 16, 139 (1934). (2) R. Linder, Acta Chem. Scand., 6, 735 (1951). 0022-3654/82/2086-0295$01.25/0
enhancement has been observed during the reduction of NiO by HZ3or anomalous behavior has been noted for the oxidation of Fe3ok4 There is far from unanimous agreement on the existence of the Hedvall effect; however, see ref 5 and the numerous references contained therein. Recently studies were made5 in the system Na2C03-CaCO, which claim to have rigorously disproven the concept. That work, and the works cited therein, are based on the narrower concept-the first cited above. The specific studies were based upon the formation of the mixed carbonate Na2Ca(C03)a.This reaction appears to be not only a very suitable model system for such a study but it is also of considerable importance to the glass industrf and in the utilization of oil shale.' The evidence put forth to disprove the existence of the Hedvall effect involved a comparison of the extent of reaction 1which took place while cycling over the range of NazC03 + CaC03 NazCa(C03)2 (1)
-
600-770 K with that which occurred during an isothermal reaction for the same length of time at 770 K.5 No specific data were given and a qualitative statement was made that the extent of reaction was greater in the isothermal case. It was concluded5that since the range of the cycle included several phase transitions of NazCa(C03)2,7it should show greater reaction if the Hedvall effect were real. This seems unreasonable in view of the large difference in temperature, 85 K, between the isothermal experiment and the average of the cycle. Assuming, as is most often the case, that the Arrhenius equation is applicable, one can explain differences in rates by factors of 5-55 (assuming activation energies of 80-200 kJ/mol by a temperature difference of 85 K. Under those circumstances it would seem that the Hedvall effect is real if the rates are anywhere close to equal. Isothermal methods would seem preferable to the dynamic studies; however, the system does not lend itself to accurate isothermal techniques, i.e., mass change. Even then there is always the uncertainty in what proportion of the time the sample is in the state of transformation. Acknowledging that this should be a good system to evaluate the Hedvall effect, we have conducted experi(3) V. V. Boldyrev, M. Bulens, and B. Delmon, 'Studies in Surface Science and Catalysis",Elsevier, Amsterdam, 1979,Chapter 111. (4) P. K. Gallagher, E. M. Gyorgy,and H. E. Bair, J . Chem. Phys., 71, 830 (1979). (6)P. D. Garn and T. S. Habash, J. Phys. Chem., 83, 229 (1979). (6) J. Mukeji, k K. Nandi, and K. D. Sharma, Am. Ceram. SOC.Bull., 69, 790 (1980). (7)J. W.Smith, D. R. Johnson, and W. A. Robb, Thermochim. Acta, 2, 306 (1971).
0 1982 American Chemical Soclety
296
Gallagher and Johnson
The Journal of Physical Chemistty, Vol. 86, No. 2, 1982
n
-
05
-
04
I-
I
0 W
5
/
+----7-
0
/ 0.3-
o
w
/
LT
5 t;
02
/
/
/
1 /
/
/
/'
TEMPERATURE ( K )
Figure 2. Fraction reacted as a function of thermal treatment: (0) isothermal, 24 h, N,; (-) temperature range of cycle, 24 h, 20 K min-', N,.
f
0 1K
i I
I
5 00
I
ui
1
I
600 TEMPERATURE ( K )
Figure 1. DSC scan of Na,Ca(CO,),,
I
I
700
12.10 mg, 20 K min-', in N,.
menta, similar to previous worker^,^ in a more quantitative manner using much narrower ranges of temperature to reduce the uncertainties. As an extension of the reactivity study, the decomposition of a physical equimolar mixture of Na2C03and CaC03was compared with that of the pure mixed compound, Na2Ca(C03)2and CaC03 in order to determine if the formation reaction (eq 1) influenced the thermal decomposition (eq 2 and 3). CaC03 CaO + C02 (2)
-
-
Na2Ca(C03)2 CaO + Na2C03+ C02
(3)
Experimental Procedures and Results Baker reagent grade CaC03 and Fisher reagent grade Na2C03were used. An equimolar mixture was prepared by moderate grinding and a portion of this was converted to the mixed carbonate by repeated grinding and firing at 800 K in air until only the X-ray pattern' of the mixed carbonate was discernible. The surface areas were 0.74 m2 g-' for CaC03 3.4 m2 g-' for Na2Ca(C03)2,and 1.0 m2 g-' for the physical mixture CaC03/Na2C03. A Dupont Model 900 DSC unit was used to measure the extent of the reaction. Figure 1 shows the DSC scan for the mixed compound. A heating or cooling rate of 20 K min-' was used in an atmosphere of flowing N2 (700 mL min-l). The sample size was 12 f 0.1 mg and sealed A1 pans were used. An identical weight of A1203was used as the reference. Three peaks are evident. The small shoulder observed by Smith et al.' on the low temperature side of the second major peak is not resolved herein. The combined area under the three peaks is used as the standard for complete reaction. Separate physical mixtures were given various thermal treatments and then
scanned under conditions identical with those used for Figure 1. The areas for the three peaks were then compared with that in Figure 1 and the fraction reacted, a, equated to the fractional area. The isothermal treatments consisted of 24-h holds at 633,663,683, and 703 K. Dynamic treatments consisted of repeated heating-cooling (20 K min-l) cycles over the ranges of 603-663 and 663-703 K for 24 h. These ranges were selected to be as narrow as possible and still completely include the two major regions of transitions. The larger span needed to encompass the lower transition is merely indicative of a more sluggish transformation on either heating or cooling (Figure 1). The fraction reacted is presented in Figure 2 as a function of the thermal treatment. The points are for the isothermal experiments and the bars represent the cyclic experiments. The dashed line is an arbitrary choice through the isothermal points for illustrative purposes. A Perkin-Elmer TGS-1 system which had been modified to take data in a digital manner* was used to study the thermal decomposition at higher temperatures. Magnetic standards were used for temperature calibration? Samples, 5.2-5.5 mg, were heated at 2.5, 10, and 40 K min-' in flowing N2 (40 mL min-l). Computer-generated plots of the rate of weight loss as a function of temperature are presented in Figure 3 for the experiments at 10 K mi&. Kinetic analysis'O of the data was performed and results are summarized in Table I. Consistent results were obtained with the three different numerical approaches employed a differential method,l' an integral method12and one based upon a differencedifferential te~hnique.'~Data used in the kinetic analysis covered the range of fraction reacted, a,from 0.1 to 0.9.
Discussion Inspection of Figure 2 indicates that the reaction had proceeded further in the cyclic cases than it had in the isothermal experiments at the average temperature of the (8) P. K. Gallagher and D. W. Johnson, Jr., Thermochim. Acta, 4,283 (1972). (9) P. K.Gallagher and F. Schrey, Thermochim. Acta, 1,465 (1970). (10) D.W.Johnson, Jr., and P. K. Gallagher, J.Phys. Chem., 75,1179 (1971). (11) B. N.N. Achar, G. W. Brindley, and J. H. Sharp, "Proceedings of the Clay Conference", Jerusalem, Vol. 1, 1966, p 67. (12) A. W. Coats and J. P. Redfern, Nature (London), 201,68 (1964). (13) E. S. Freeman and B. Carroll, J. Phys. Chem., 62, 394 (1958).
The Journal of Physical Chemism, Vol. 86, No. 2, 1982 297
Solid-State Reactivtty of Na2C03-CaC0,
- Na2C03/Coco3 -- Na2Ca( ~ 0 3 ) 2 ---- coco3
TEMPERATURE ( K )
Figure 3. DTG curves at 10 K min-' In N2: (a) CaCO,, 5.49 mg; (b) NalCa(CO&, 5.28 mg; (c) Na&0,/CaC03 (physical mixture), 5.40 mg.
TABLE I: Results of Dynamic Kinetic Experimentsa 40 K min-'
K
10 K min-'
min"
226 1x 1 O ' O 1031 0.15
247 2 x 10" 971 0.043
2.5
CaCO, A H , kJ/mol A , s" TmaxR,
Rmax(da/dmh)
201 3 x lo* 1104 0.45
A H , kJ/mol A , s-* TmaxR>
Na,Ca( CO, ), 151 167 5 X 10' 6 X lo6 1068 1000
Rm,(da/dmh)
0.41
0.13
184 6 X 10' 943 0.034
Na,CO,/CaCo,c A H , kJ/mol A , s-l
TmaxR,
R,,(d01
/dmh)
205 5 X lo9 1045 0.42
197 2 X lo9 978 0.12
205 6 X lo9 852 0.036
a A H , is the apparent activation energy;A, the preexponential Arrhenius term; T m a x the ~ temperature at maximum rate of weight loss; and R,, the rate of change of 01 at TmaxR. Best-fitting rate law (0.1 < 01 Q 0.9) was the contracting area. Best-fitting rate law (0.1 < 01 Q 0.9) was the first-order rate law.
cycle. In neither case, however, had the mixture reacted as extensively as in the isothermal experiment at the maximum temperature of the cycle consistent with the earlier work.6 However, it must be remembered that the experimentallyobserved Hedvall effect is generally at most a factor of 4 in the enhancement of the rate at the specific transition temperature2+J4-16and hence an enhancement
factor of 1.2-2 at the average temperature observed in Figure 2 is reasonable for a 40-60-K cycle. Admittedly this is slightly in error because the rate is not a simple linear function of temperature. Nevertheless, the results more confirm rRther than deny the existence of a Hedvall effect. At higher temperatures CaC03 and Na2Ca(C03)2decompose according to eq 2 and 3. Table I and Figure 3 show that the physical mixture decomposes at the lowest temperature and CaC03 at the highest temperature. Among the mathematical models used for comparison, the decomposition of the pure compounds can best be described by the contracting area rate equation, eq 4,while the physical mixture best follows a first order rate law, eq 5, is a rate constant. The rate law and Arrhenius paramk t = 1 - (1- a p 2 (4) k t = - In (1- a) (5) eters for CaC03agree very well with previous work; however, conformation of a specific mechanism requires more varied and detailed work. The trend toward higher apparent activation energy with decreasing heating rate observed for the two compounds has been described as indicative that physical process as well as chemical factors influence the reaction rate." The earlier decomposition of Na2Ca(C03)2compared to CaC03 might be due to some nonstoichiometry but more likely arises from its greater surface area. For the physical mixture, the different rate law and absence of the dependence of apparent activation energy upon heating rate clearly imply that different factors are involved in its thermal decomposition. A likely explanation for its decomposition at lower temperatures involves an enhanced reactivity, again due to the disruptive influence of the reaction upon the decomposition. In this case it would be the simultaneous course of eq 1 with either eq 2 or 3 rather than the phase transition commonly invoked in the Hedvall effect. The decomposition reactions take place before the incongruent melting temperature of Na2Ca(C03)2(1086 K) or the eutectic mixture of Na&03-Na2Ca(C03)2(1053 K)! Hence, the presence of a liquid phase is not a potential explanation for this phenomena. Discussed herein have been two cases, the first being the narrow sense of the Hedval effect where the effect is confined to the brief period of time incorporating the actual structural rearrangement due to a phase transition. The DSC data herein have lent support to the existence of such an effect. The second case involves the more extended interpretation where the effect includes the broader time range to incorporate enhanced reactivity from residual effects of transitions and decompositions such as lattice strain, etc. This is closer to the original intent of Hedvall, as shown by some of the examples he chose. The present TG data also support the existence of the Hedvall effect in this broader sense. ~~
~~
~
~
~~
(14)G. Parravano, J . Chem. Phys., 20, 342 (1952). (15)T. Kawai, K. Kunimoki, T. Kondow, T. Onishi, and K. Tamaru, 2.Phys. Chem. (Frankfurt Am. Main), 86,268 (1973). (16)B.C.Scales and M. B. Maple, Phys. Rev. Lett., 39,1636(1977). (17) P.K. Gallagher and D. W. Johnson, Jr., Thermochim.Acta, 6,67 (1973).
