MEASUREMENT OF HIGH-ALUMINA CEMENT- CALCIUM CARBONATE REACTIONS USING DTA

Clay Minerals (1984) 19, 857-864 MEASUREMENT OF HIGH-ALUMINA CEMENT- CALCIUM CARBONATE REACTIONS USING DTA H. G. MIDGLEY Ilminster Cement Research, 24 Summerlands Park Drive, llminster, Somerset TA19 9BN (Received 7 November 1983; revised 19 May 1984) A B S T R A C T: Hydrating high-alumina cement will react with calcium carbonate to form the complex mineral calcium carboaluminate hydrate, 3CaO. AI203. CaCO 3 9 12H20. This mineral is reported to be capable of providing strength in concrete and so may provide an alternative to the minerals normally found in the hydration of high-alumina cement, which may under certain conditions convert to other minerals with a loss in strength. Some doubt has been cast on the stability of calcium carboaluminate hydrate and it has been found that in hydrated high-alumina cement, calcium carboaluminate hydrate decomposes at temperatures in excess of 60~ Cube compressive strength tests on high-alumina cement and high-alumina cement-calcium carbonate pastes have shown that the latter have a lower strength than pastes made with high-alumina cement alone. When cured at 50~ the high-alumina cement-calcium carbonate pastes show a loss in strength with curing time. Cements made with the high-alumina cement-calcium carbonate mixture always have a lower strength than those made with high-alumina cement alone and so no advantage is gained from their use. Calcium carboaluminate hydrate has been suggested as a mineral which would replace the metastable mineral mono-calcium aluminate decahydrate as the primary hydration product in hydrating high-alumina cement if calcium carbonate powder is added (Cussino et al., 1980; Negro et al., 1976; Negro et al., 1982). The first product of hydration of high-alumina cement is mono-calcium aluminate decahydrate which with time will convert to the more stable minerals tri-calcium aluminate hexahydrate and alumina trihydrate. If the concrete made with high-alumina cement has too high a water:cement ratio and is cured at too high a temperature, usually in excess of 25~ then the conversion may be accompanied by a loss in strength. If calcium carboaluminate hydrate is stable and does not decompose with a concomitant loss of strength, then a cement based on high-alumina cement-calcium carbonate (limestone) would provide the Construction Industry with a new cement. The present paper describes two series of experiments carried out to test the hypotheses of Negro et al. (1982). The first was to investigate the thermal stability of calcium carboaluminate hydrate as formed in hydrating high-alumina cement-calcium carbonate when cured at temperatures in the region 20 to 80~ The second followed the strength development of such cements and equivalent pastes made with high-alumina cement alone when cured at 30 and 50~ ~) 1984 The Mineralogical Society

858 H. G. Midgley MATERIALS AND METHODS The materials used were high-alumina cement (Ciment Fondu Lafarge as supplied by LAC of Grays, Essex), calcium carbonate (AnalaR laboratory reagent) and mains water. For the stability series, 10-g batches of cement were prepared with 7 g of the high-alumina cement and 3 g of calcium carbonate. The mixture was tumble-mixed for at least 1 h to homogenize as fully as possible. For each experiment the cement powder and the water were brought to the temperature of the test before mixing. To prepare the paste, 2.7 ml water was added to the powder and hand-mixed for two minutes, using the method given in Midgley & Midgley (1975), before placing the paste in sealed glass containers over water to maintain a moist atmosphere. The container was immediately placed in a thermostat maintained at the appropriate temperature to + 1 ~ Samples of the set paste were removed for examination after 2, 3, 6, 9, 12, 24, 48, 168, and 240 h. These were crushed and washed in methanol and then air-dried at 30~ Methanol washing was used to remove the 'free' water; if this was not done the sample suffered hydrothermal reactions during heating in the thermal analyser (Fig. 1). The DTA apparatus used was constructed by the author and consisted of a simple non-inductively-wound tube furnace heated by a programmed energy regulator (15~ _+l~ The sample crucibles were ceramic and temperatures were measured with a I I I 1 O0 200 300 Sample temperature ~ FIG. 1. Effect of removal of'free water' by methanol treatment. (a) air-dried; (b) methanol dried.

DTA of high-alumina eement-caco 3 reactions 859 chromel-alumel thermocouples. The temperature of the furnace was measured in the test sample; the inert reference material was alpha-alumina. In all tests, a 700 mg sample was used. Calibration of DTA equipment The DTA system was cafibrated with pure synthetic samples of mono-calcium aluminate decahydrate and tri-calcium aluminate hexahydrate, and with a natural gibbsite containing 98% of alumina trihydrate as determined by quantitative X-ray diffraction. No pure sample of calcium carboaluminate hydrate was available so the calibration was carried out by determinations on a thermobalance (Stanton TG 750). The water loss was determined on three samples of hydrated high-alumina cement-calcium carbonate pastes known to contain calcium carboaluminate hydrate from XRD analyses. From the weight loss it was assumed that the calcium carboaluminate hydrate contained 12 moles of water and thus the weight of mineral in the sample was established. These samples were then used to calibrate the DTA apparatus. The method used for determining amounts of the minerals present in the set pastes was that described by Midgley (1982). The calibration factors used were (peak height in mm per mg mineral present): mono-calcium aluminate decahydrate 0.9; alumina trihydrate 1.0; tri-calcium aluminate hexahydrate 1.0; calcium carboaluminate hydrate 1.2. These minerals gave approximate peak temperatures of 150, 275, 310 and 200~ respectively. In the samples examined, no di-calcium aluminate octahydrate, which has a peak temperature at ~230 ~ C, was detected. An extra peak at about 130 ~ C frequently occurred and this was considered to be due to the removal of inter-particulate water (Fig. 2). J water CAHlo C3AH6 1 O0 1 200 300 Sample temperature ~ FIG. 2. Typical DTA curve for hydrated HAC-CaCO 3 paste.

860 H. G. Midgley THERMAL STABILITY OF CALCIUM CARBOALUMINATE HYDRATE Samples of high-alumina cement-calcium carbonate pastes were prepared and cured at 20, 50, 60, 65, 70 and 80~ Quantitative estimates in these products of the phases mono-calcium aluminate decahydrate, calcium carboaluminate hydrate, tri-calcium aluminate hexahydrate and gibbsite as determined by DTA are given in Figs 3, 4, 5 and 6, respectively. From these data it can be seen that at all temperatures calcium carboaluminate hydrate was formed; at 20~ formation was delayed until 9 h, the quantity then increased rapidly until 72 h, but after this time amounts remained essentially constant. At 50~ calcium carboaluminate hydrate was detected after 2 h curing in quantities greater than for samples cured at 20~ however, by 48 h the amount found was the same for both samples. For samples cured at 60~ the calcium carboaluminate hydrate formed rapidly in the first two hours but amounts did not increase with further curing up to 240 h. At both 65 and 70~ calcium carboaluminate hydrate formed rapidly in the first two hours but then decreased; after 24 h, however, amounts remained essentially constant. At 80~ calcium carboaluminate hydrate was detected after two hour's curing but none was found after 6 h. From these data it would appear that there are two competing reactions: first, formation of calcium carboaluminate hydrate from the anhydrous phases and, second, decomposition of the calcium carboaluminate hydrate. From the data shown in Fig. 4 it can be concluded that at 80~ the rate of decomposition of the calcium carboaluminate hydrate is greatly in excess of its rate of formation. At 70 and 65 ~ C the decomposition balanced formation; at 60 ~ C the position is not clear but it is possible that the calcium carboaluminate hydrate was decomposing slowly. At 50 and 20~ there was no evidence for the decomposition of the calcium 25 ~-IO 20~ 20 ~ 15 ~r... O (HAC at 20~ i I...,"... i... t......4k (HAC at 50~... f.... o I.... X..'7" i I 10 i! I I 2 3 s t I J! I I I 6 9 12 I,~...?--q' 50oc 24 48 168 240 In time hours FIG. 3. Monocalcium aluminate decahydrate found after curing for different times at different temperatures.

DTA of high-alumina cement-caco 3 reactions 861 r ~t, 15 ~ 4r -~ "~'~-~,~ 50oc ~20oc p.... d" l l!. I I d 9 10 El-- "13 "'f ;J'~"\ x." 9 l i l '., 9 h3-... t3 60~ "u-..~ # " O. " "'0,. 03, ~'\ " ""... --0 65~ v. 8o~ o-...:~, c-.s... :27...,~ 7ooc "'-..~. I I I I... ~ I I I I I 1 2 3 6 9 12 24 48 168 240 In time hours Fla. 4. Calcium carboaluminate hydrate found after curing for different times at different temperatures. 25 20 o~..,o--...... _O1........ 15,>::'.~:.:-:::@"- ~-... ::7;:~:::::::~1:..-7..::;:::::.=... 70, 65oc o...~ ~.,.,, sa-... ~... ~-......... ~..~..~... ::~ 60oc zr... ~"....'0"-... --0,-..~.--4> 50~ 10 <::f - ~ -I1--- ~ ~ 20~ ~-----.:... x (.AC ~t 50~ 5 ~ ~O" ~'"~... O (HAC at 20~ 1 I I Y I 1 I [ I 2 3 6 12 24 48 168 240 In time hours F1G. 5. Gibbsite found after curing for different times at different temperatures. carboaluminate hydrate. The conclusion from this is that calcium carboaluminate hydrate decomposes at temperatures in excess of 60~ The quantities of the other hydrate minerals (Figs 3, 5 and 6) were as would be expected (Midgley & Midgley, 1975). Mono-calcium aluminate decahydrate decomposed at temperatures >20~ The amounts of gibbsite and tri-calcium aluminate hexahydrate

862 H. G. Midgley 25 20 15 v...~.o ":::='r'-';~'''~... -o-... o-_.:...,~.-r..~ -... 9 -. ~x 70~ 9 " ~.... ~' -... 0 65~ o ~ ~... ~ '~.'.'.~ ~:. --13-............. t3-... 60~ 10 ~"~"~" (HAC at 50~.... _._.-O 50~ r... ~... ~x...>~.~._..~.;;.~-;~t. zooc 5,,it ~ 0..?.:, ::i~... o (HAC at 20~ I k,---- -"~ I I I I I 2 3 6 9 12 24 168 240 In time hours FIG. 6. Tri-calcium aluminate hexahydrate found after curing for different times at different temperatures. formed were temperature-sensitive--the higher the temperature the greater the amounts formed. STRENGTH DEVELOPMENT OF HIGH-ALUMINA CEMENT- CALCIUM CARBONATE CEMENTS Experimental Two cements were investigated: a normal high-alumina cement and a mixture of this cement with calcium carbonate powder. Neat cement pastes were prepared from both materials. A water :cement ratio of 0.27 was used; the pastes were hand-mixed for about 2 min, hand-tamped into a brass mould with 12.5 mm sides, cured for 24 h at 100% r.h., then placed in water at 20 and 50~ Results and discussion Fig. 7 gives the cube crushing strength (average of 6 cubes) for the various periods of cure. It can be seen that for high-alumina cement alone, curing at 20~ produced a gain in strength up to 90 days; curing at 50~ showed a gain in strength during the first 7 days, followed by a reduction between 7 and 14 days, which was then followed by a slight gain in strength. The loss in strength due to curing at elevated temperatures was ~53% at 90 days. This behaviour of high-alumina cement pastes is well known and is called conversion (Midgley & Midgley, 1975). The strength results for high-alumina cement-calcium carbonate mixes cured at 20~ were always less than the strength for high-alumina cement alone--for example at 90 days the reduction in strength was ~25%. Since the high-alumina cement-calcium carbonate mixes contained 20% calcium carbonate, the loss in strength could be accounted for by a

DTA of high-alumina cement-caco 3 reactions 863 160... --~.~_... Lx... HAC at 20~ 120... ~ HAC-CaCO 3 at 20~ ~ '13-... HAC at 50oC 40 ~. ~ ~ HAC CaCO 3 at 50~ I I I I 7 14 28 90 In age (days) FIG. 7. Compressive strength development curves for HAC and HAC-CaCO different temperatures. 3 pastes cured at simple dilution effect, which would seem to indicate that for curing at 20~ calcium carboaluminate hydrate contributed nothing to the compressive strength. Curing at 50~ produced a different pattern: the high-alumina cement-calcium carbonate mixes showed a loss in strength of ~40% which was less than the loss in strength for neat high-alumina cement pastes (53%). However, by 90 days the loss in strength was 50% of the initial strength at 1 day which was similar to the loss in strength of high-alumina cement pastes alone. The explanation for these effects is found in the changes in mineralogy. Fig. 3 shows that at 20~ mono-calcium aluminate decahydrate was found in large quantities, while at 50 ~ C none was found. For high-alumina cement alone cured at 50~ there was little or no change in the quantity of this phase detected up to 7 days, but by 14 days it was decomposing to form the more stable gibbsite and tri-calcium aluminate hexahydrate. It was this change in mineralogy which accounted for the lower strengths found. However, the loss in strength in terms of load capacity was smaller for high-alumina cement-calcium carbonate mixes (50 Pa) than for the high-alumina cement alone (80 Pa), although on a percentage basis the loss in strength was similar (55 and 50%, respectively). The important fact is that the loss in strength occurs earlier for the high-alumina cement-calcium carbonate mixes--7 days compared with 28 days. Another important point is that there was an earlier recovery in strength (between 7 and 90 days) but the rate of strength recovery was asbout the same for both types of cement. The over-riding fact is that high-alumina cement-calcium carbonate mixes always produce lower strengths than high-alumina cement alone, usually by a factor of 25-30%. From this it is concluded that although calcium carbonate reacts with high-alumina cement to produce calcium carboaluminate hydrate, this confers no advantage to its performance as a cement. ACKNOWLEDGMENTS The author would like to thank Dr C. Fentiman of LAC for the sample of high-alumina cement and for many helpful discussions on the subject. Thanks are also due to the Director of the Hatfield Polytechnic, Dr J. M. Illston, for permission to use the DTG apparatus and for facilities to determine the compressive strengths, and to Mr T. Grounds for considerable assistance in the strength determinations.

864 H. G. Midgley REFERENCES CUSSINO L. & NEGRO A. (1980) Hydration of aluminous cement in the presence of silicic and calcareous aggregates. Proe. 7th Int. Cong. on Chemistry of Cement, Paris, v62-v67. MIDGLEY H.G. (1967) The mineralogy of set high-alumina cement. Trans. Brit. Ceram. Soc. 66, 161-187. MIDGLEY H.G. (1982) The relationship between hydrate mineral content and compressive strength of set high-alumina cement. Proc. Int. Seminary on Calcium A luminate, Turin, 1982, 314-324. MIDGLEY H.G. & MIDGLEY A. (1975) The conversion of high-alumina cement. Mag. Concr. Res. 27, 59-77. NEGRO A., MURAT M. & CUSSINO L. (1976) Carbonate reactions of calcium aluminates. Proc. Syrup. on Carbonation of Concrete, Theme 1, paper 8. Cement and Concrete Association, Wexham Springs, Buckinghamshire, UK. NEGRO A., BACHIORRINI A. & MURAT M. (1982) Interaction in aqueous medium between calcium carbonate and mono-calcium aluminate at 5, 20 and 40~ Bull. Mineral. 105, 284-290.