J. appl. Chem. Biotechnol. 1974,24,221-227

The Reactivity of Calcium Oxide Towards Carbon Dioxide and Its Use for Energy Storage

Ronald Barker

Electricity Council Research Centre, Capenhurst, Chester CHI 6ES (Paper received 25 January 1974, amendedpaper accepted 7 March 1974)

By using calcium carbonate powder of particle size - 10 nm (and therefore of high surface area) it has been shown that the reaction CaC03 d CaO+COz can be made almost wholly reversible. A reactivity of 93 % was achieved, (i.e. 93 % of the calcium oxide reacted with carbon dioxide) and this was maintained for 30 24-h decomposition-back reaction cycles at 629 "C with no detectable decrease. This material has an energy storage capacity of 200 Wh/lb, but only about 1 kWh/fL3 When this calcium carbonate was pressed, however, to in- crease its bulk density (from 0.1 to - 1 g/ml) there was a large decrease in its surface area and a smaller decrease in its reactivity on undergoing a few de- composition and back reaction cycles. Scaling up of the sample size from 20 mg to 10 g only had a small ( x 3) effect on the rates of decomposition and back reaction.

1. Introduction

It has previously been shown,' using calcium carbonate in the form of a powder of particle size 10 pm, that the reaction CaCO, # CaO + CO, is not wholly reversible. In addition, there is a continued loss of reversibility as the reaction is cycled, which is mainly due to the decrease in surface area of the oxide caused by the loss of small pores. It was suggested that if calcium carbonate (or oxide) in the form of a powder of particle size < 44 nm was used (so that the reaction of carbon dioxide with cal- cium oxide was not diffusion controlled) the reaction would be wholly reversible and there would be no decrease in reversibility as the reaction was cycled. This paper describes the results of experiments designed to test this view.

2. Experimental

A common method of obtaining sub-micron powders is by high temperature evapora- tion with suitably rapid quenching. Calcium carbonate of crystallite size < 10nm was obtained from A.E.R.E., Harwell; this had been prepared (as the oxide) by an electron beam process in an inert atmosphere., The oxide was reacted with carbon dioxide before exposure to the atmosphere and was found by thermogravimetry (t.g.) to contain - 94% by weight of calcium carbonate, the remainder presumably being calcium oxide. The specific surface area of this material, either carbonate or oxide, was quoted as being in excess of 200 m2/g by Hatwell. From this value and

221

R. Barker 222

assuming that the crystallite is a cube, we can calculate the particle size as being - lOnm, and transmission electron micrographs of replicates showed this to be correct.

Cycles of decomposition-back reaction were carried out on the Du Pont t.g. apparatus described previously, for small samples ( N 100 mg) of the high surface area calcium carbonate. The reversibility of the reaction for larger samples (up to 12g) was measured using an apparatus in which the pressure of carbon dioxide released by the carbonate on decomposition or taken up by the oxide on back reaction, was measured, This method of following the reaction is compared with the gravimetric method for "ordinary" calcium carbonate in Figure 1: it will be seen that the two methods give the same results to within the experimental error.

ol.d--A 0 2 4 6 8 10 12 14

No. of 24-h cycler

Figure 1. Comparison of reaction measured by t.g. and carbon dioxide pressure. ( x ) by weighing; ( 0 ) by pressure measurement.

Surface area measurements were carried out on the Perkin-Elmer Sorptometer as described previously.' Samples of high surface area calcium carbonate were pressed either in a hydraulic press using a metal die or in an isostatic press using a liquid for transmitting the pressure, which should therefore give more uniformly pressed samples. The electron microscope and mercury porosimeter used were as described previously. '

3. Results

A typical weight Y. time plot for the high surface area calcium carbonate is shown in Figure 2. The sample was decomposed at 629 "C under nitrogen and back reacted at 577 "C with carbon dioxide, In Figure 2 the weight of calcium oxide has been put at 100%: on this scale full back reaction to calcium carbonate should give a weight of 178.5%. The much longer induction period and slower rate of decomposition

Reactivity of calcium oxide towards carbon dioxide 223

compared with the results obtained previously for “ordinary” calcium carbonate’ arise from the lower temperatures used here. In fact the reciprocal of the induction period and the decomposition rate both fall on the appropriate’ Arrhenius plots obtained for “ordinary” calcium carbonate to within the experimental error, as shown in Figure 3.

I 8 0 I I I I I I

\

Figure 2. Weight of sample against time: decomposed under N2 at 629 “C, back reacted with C02 at 577 “C.

The back reaction here is different from that found for “ordinary” calcium car- bonate. It was found previously that there was a fast surface reaction followed by a much slower reaction controlled by diffusion of the carbon dioxide gas through the relatively impermeable layer of calcium carbonate and that the transition between these two reactions was quite marked (see the dotted line in Figure 2). This con- trasts with the behaviour of the high surface area material which exhibited no dis- continuity in the back reaction curve (Figure 2). For thc oxide derived from the ordinary calcium carbonate, the surface reaction decreased in extent with cycling with the eventual result that the reaction became more and more irreversible (Figure 2 of ref. 1). In comparison the high surface area material showed no such decay but exhibited a reversibility of reaction which did not alter as the reaction was cycled many times.

A plot of the calcium carbonate concentration in the sample as a function of the number of 24 h cycles is shown in Figure 4. The fraction of calcium carbonate in the sample remained at 0.93 throughout the experiment and the reaction was therefore almost completely reversible (93 %) and this reversibility showed no decrease with time. The extent of the fast reaction is also shown-after an initial decrease this remained constant for the duration of the experiment.

224 R. Barker

1.0 I I I I I 1 I I

CX " A X u T m k A A n n A A X A n

08 - n

-

8

6 -

-577 .c bdwk GI cydc

c

c a 0 6 : L m m w 0 4 Font noc)ion mly - I

0.2 - -

I I I 1 I I I

A sample of the high surface area calcium carbonate was decomposed and back reacted at 866 "C. After being maintained at this temperature overnight the sample showed behaviour very similar Lo that for "ordinary" calcium carbonate with a discontinuity between the fast and slow back reaction and a marked hysteresis

0

01 ; 10 - ) o a I2

I O'/ T

Figure 3. Arrhenius plots for the rate of decomposition and induction period. ( x ) rate of de- composition; ( 0 ) reciprocal of the induction period. H represents the high surface area material.

The other points are for "ordinary" calcium carbonate.

s 10 15 20 25 30 35 40

Figure 4. Reactivity against numbcr of cycles: decomposed under N, at 629 "C, back reacted with CO, at 577 "C.

Reasrtivlty of calcium oxide towards carbon dioxide 225

effect if none of the slow back reaction was allowed to occur. Stereoscan pictures showed that extensive sintering had taken place and the size of the smallest particle was then - 2 pm.

The bulk density of the high surface area calcium carbonate was very low (- 0.1 g/ ml) and samples were pressed to increase the density and therefore improve the energy storage per unit volume. The surface area, pore size distribution and reactivity of these samples were measured and the effect of increasing sample size over the range 10 mg to 12 g on the overall reactivity was examined. Some results which were

TABLE 1. Experiments with larger samples ~ _ _ _ _ _ _ ~~ ~ - ~~~

Density Specific surface area m'lg Applied Sample after -- No. pressure weight pressing After After of

Sample (kglcm') (9) (g/ml) pressing experiment cycles

1 280 6.6 .. . 44.4 9.4 8 2 314 6.1 0.76 39.0 1 .o 7 3 1082 9.9 0.96 38.4 0.5 18

obtained using larger samples (6 to 9 g) in the apparatus using the pressure method of following the reaction, are shown in Table 1. The specific surface area of the un- pressed calcium carbonate used here was 65 m'/g. After pressing to achieve a density of 1 g/ml the carbonate had a surface area of 40 mZ/g and on taking the carbonate through eighteen decomposition-back reaction cycles there was a further fall off in the specific surface area to - 0.5 m'/g. A plot of the variation of the reactivity of these three samples with the number of 24-h reaction cycles is shown in Figure 5. The pore size distribution of unpressed, pressed and pressed and reacted samples of calcium carbonate is shown in Figure 6. The unpressed material had a large volume, 7 ml/g, spread over a wide range of pore diameters from 0.02 to 30 pm, which ob- viously represents the spaces between the particles. The pressed sample had a much smaller pore volume (0.32 ml/g) arising from pores of diameter 0.02 pm (20 nm).

w" 0 2 4 6 B 10 12 14 16 18

No of 24-h cycles

Figure 5. Extent of reaction against the number of cycles. ( x ) sample 1 , (0) sample 2, (0) sample 3. (See Table 1).

226 R. Barker

The sample which had been pressed and reacted showed a further fall in pore volume and an enlarging of the pore sizes presumably caused by internal evolution of carbon dioxide gas during decomposition. When the sample mass was increased from 20 mg to 10 g the rates of decomposition and back reaction decreased by a factor of N 3.

L I

0 4

- Unprrnned Co CO,

-

wi by 314 kqlcrn' . .. -

-..--, f Cam3 p m m d by cocoaprnm 471 Q/cmZ and token through 7 c y c l m at 70072

- - - f I -- lo.,

Pore dlameter ( r m )

Figure 6. Mercury porosimctry of unpressed, pressed and reacted calcium carbonate.

4.1Dhcussion

Calcium carbonate of particle size 10 nm shows 93 % reactivity (i.e. on back reaction a mixture of 93% calcium carbonate and 7% calcium oxide is formed) and this reactivity can be maintained for at least 30 cycles at 629 "C with no perceptible decrease. This calcium carbonate does not exhibit the hysteresis effect of the low surface area calcium carbonate and the reaction can be stopped at any stage without overall loss of reactivity. It is not clear why only 93 % of carbonate is formed on the back reaction. Chemical analysis showed no impurities greater than 0.1 % by weight. This lowered reactivity is not present on the first cycle in the case of the large particle- sized calcium oxide where the surface area is overwhelmingly internal. The surface area of the fine-particle sized oxide is predominantly external but the packing to- gether of these oxide particles does not seem to be a factor in their reactivity as the value of 93% is not reduced on the first cycle when the sample is pressed.

With the material of 10 pm particle size, the reactivity of the calcium oxide formed, to carbon dioxide was always proportional to the surface area. In the case of the pressed high surface area calcium carbonate the relationship between reactivity and surface area seems to be much less straightforward. Considering sample No. 3 in Table 1 and Figure 5 : after eighteen cycles of the reaction the surface area fell from 38.4 to 0.5 m2/g. whereas the reactivity decreased from 1 to 0.47. Therefore, it seems that

Reactivity of calcium oxide towards carbon dioxide 227

the surface area of these samples of pressed calcium carbonate decreases quite quickly with the number of reaction cycles whereas the reactivity falls off much more slowly.

It has been pointed out previously' that the Tammann temperature (above which a solid would be expected to lose surface area by sintering, i.e. by lattice or bulk diffusion) is 533 "C for calcium carbonate, and sintering is therefore likely to be important under the conditions used here. There was no loss of reactivity of the unpressed powder over a period of 30 days at 629 "C but once the powder had been pressed (to at least 300 kg/cmZ), and of course this would increase the rate of sintering, there was a rapid loss of surface area and (to a lesser extent) of reactivity after a few decomposition-back reaction cycles.

5. Conclusions

The principle aim of the work described here and previously' was to establish whether the reaction CaCO, # CaO + COz was wholly reversible and whether an energy storage system based on this reaction would have an attractive energy storage density. It has been shown using calcium oxide of particle size - 10nm that the reaction can be made almost wholly reversible and that the reversibility does not decrease when the reaction is taken through many decomposition-back reaction cycles.

The energy storage capacity per unit weight of the unpressed calcium carbonate is - 200 Wh/g, which is attractively high, but the energy storage per unit volume is only 1 kWh/ft., This latter value can obviously be increased by pressing the material but it is unlikely to exceed 10 kWh/ft., without reducing the reversibility of the reaction. Calcium carbonate of this particle size could therefore form the basis of an energy storage system if weight was the over-riding parameter, but for electric storage heaters where energy storage per unit volume is also important, this material is not satisfactory.

Acknowledgement Thanks are due to Mr A. MacKay for experimental assistance.

References I . 2.

Barker, R . J. appl. Chem. Biotechnol. 1973, 23, 133. Ramsay, J . D. F. Private communication.