Formation Of Ettringite From Monosubstituted Calcium Sulfoaluminate Hydrate And Gypsum.pdf

Journal

J. Am. Ceram. Soc., 82 [10] 2900–905 (1999)

Formation of Ettringite from Monosubstituted Calcium Sulfoaluminate Hydrate and Gypsum Boyd A. Clark* and Paul W. Brown* Department of Materials Science and Engineering, Pennsylvania State University, University Park, Pennsylvania 16802

The formation of ettringite (3CaOⴢAl2O3ⴢ3CaSO4ⴢ32H2O) from monosulfate (3CaOⴢAl2O3ⴢCaSO4ⴢ12H2O) and gypsum (CaSO4ⴢ2H2O) was investigated by isothermal calorimetry and X-ray diffraction (XRD) analyses. Hydration was carried out at constant temperatures from 30° to 80°C using deionized water and 0.2M, 0.5M, and 1.0M sodium hydroxide (NaOH) solutions. Ettringite was found to be the dominant crystalline phase over the entire temperature range and at all sodium hydroxide concentrations. A sodium-substituted monosulfate phase was formed as a hydration product in the 1.0M sodium hydroxide solution regardless of temperature. XRD and calorimetry demonstrate that hydration in increasing sodium hydroxide concentrations decreases the amount of ettringite formed and retards the rate of reaction. The apparent activation energy for the conversion of the monosulfate/gypsum mixture to ettringite was observed to vary depending on the sodium hydroxide concentration. Ettringite formation was observed to depend upon the concentration of calcium in solution; thus the formation of calcium hydroxide and sodium-substituted monosulfate phase competes with ettringite formation. I.

extensively.5–9 The formation of these sulfoaluminates occurs during normal curing, but can also be caused by interaction with a variety of environments, including intrusion of solutions containing high concentrations of sulfate and/or other ionic species. Thus the interaction of various ionic species and their effects on ettringite and monosulfate formation have also been studied extensively.10 Concrete is a hydrated, porous medium and a certain amount of “free” water is usually present in the pores. This porosity allows solutions to move through concrete and this may result in sulfoaluminate compound formation. There is also evidence in the literature that the interaction of ionic species (most notably alkali ions) initially present may cause subsequent sulfoaluminate compound formation.11 The presence of various ionic species can cause a conversion of monosulfate into other AFm phases. One such phase is a sodium-substituted AFm phase referred to as the “U-phase.” Such a sodium-substituted AFm phase (4CaO⭈0.9Al2 O3 ⭈ 1.1SO3⭈0.5Na2O⭈16H2O) has been observed when C3A hydration was carried out in high concentrations of sodium and sulfate.12 Recent studies indicate that the U-phase also forms in ordinary portland cement and its formation results in expansion, implying a deleterious effect in hardened concrete.4,13 Although much work has been done to study the formation of ettringite from C3A followed by its conversion to monosulfate, we are unaware of any studies which have considered the conversion of monosulfte to ettringite. Therefore, the mechanism and kinetics of ettringite formation from monosulfate are investigated in the present study. Deionized water and various concentrations of sodium hydroxide solutions were used to hydrate a monosulfate/gypsum mixture. Isothermal calorimetry and X-ray diffraction (XRD) were used to characterize the rates of hydration and the crystalline phases formed. Analysis of the results obtained by these methods provides insight into the kinetics of the hydration processes involved and the stability of the various phases formed.

Introduction

E

TTRINGITE is a hydrated sulfoaluminate compound (3CaO⭈Al2O3⭈3CaSO4⭈32H2O) and is commonly formed in concrete during curing. Monosulfate (3CaO⭈Al2O3⭈CaSO4⭈ 12H2O) is a related sulfoaluminate compound. In cement nomenclature ettringite is also referred to as AFt and monosulfate as AFm. Ettringite formation while the concrete is in a plastic state contributes to the initial “setting” of concrete. Alternatively, ettringite formation after concrete has “set” into a rigid, solid mass can be deleterious and can cause the loss of structural integrity. The reason for this dual character is due to the solid specific volume increase accorded to ettringite formation. In the literature, expansion due to ettringite formation is attributed to its needlelike structure.1 Monosulfate incorporates only 1 mol of sulfate and exhibits a platelike microstructural morphology.2 The formation of monosulfate in hardened concrete is generally not considered deleterious, because its formation causes less expansion than does that of ettringite. There is, however, evidence in the literature which indicates that monosulfate formation also causes expansion.3,4 The formation of ettringite and monosulfate from cement phases (tricalcium aluminate and gypsum) has been studied

II.

Methods and Materials

(1) Monosulfate Synthesis The precursor monosulfate powder was prepared by mixing a slurry of tricalcium aluminate and gypsum under an inert gas (nitrogen) using a modification of the method outlined by Taylor14 (CaCl2 or soda lime were not used for drying). The slurry was filtered and dried, again under nitrogen, and then mixed with gypsum powder at a molar ratio of 1:2.5. (2) Experimental Procedure Isothermal calorimetry was performed using the experimental setup described by TenHuisen et al.15 Approximately 1.00 g of the monosulfate/gypsum mixture and 3.0 mL of solution were used for each sample run. Solutions used were deionized water, and 0.2M, 0.5M, and 1.0M NaOH. The solid sample and a syringe containing the solution were allowed to reach equilibrium at each experimental temperature before injection. Sample runs were carried out at 30°, 40°, 50°, 60°, 70°, and 80°C.

C. M. Jantzen—contributing editor

Manuscript No. 190392. Received February 9, 1998; approved April 24, 1998. Supported by the National Science Foundation under Grant No. CTS 93-09528 and by the R. J. Lee Group, Inc. *Member, American Ceramic Society.

2900

October 1999

Formation of Ettringite from Monosubstituted Calcium Sulfoaluminate Hydrate and Gypsum

2901

As the solution was inoculated onto the solid sample, heat output was measured by thermopiles, which completely enclosed the sample cup. Data points were collected at 0.1 s intervals. When further heat output could not be observed, the solids present were removed from the calorimeter, washed with acetone, and allowed to air dry before XRD characterization. These solids were hand ground in an agate mortar and pestle to a fineness of approximately 325 mesh (45 ␮m diameter particulate). A Scintag 3100 system powder diffraction unit was used to examine the powders. All samples were scanned at 2°/min between 5° and 55° 2␪, using CuK␣ radiation. Analysis of the XRD data is complicated in that the ettringite pattern has diffraction X-ray peaks which overlap the primary diffraction peaks for monosulfate, calcium hydroxide, and the U-phase. This problem was overcome by examining the ratio of the primary (monosulfate, calcium hydroxide, or U-phase) peak and primary ettringite peak intensities and then determining where the ratio varies. Quantitative analysis of the hydration products was performed by determining the heights of the primary diffraction peaks for each crystalline phase. The primary diffraction peak for each hydrated phase is listed in Table I. III.

Results

Figures 1(a) and (b) show the rates of reaction and heat output for reaction between monosulfate and gypsum hydrated in deionized water between 30° and 80°C. In general, the rate of hydration is observed to increase with increasing temperature in any given solution. The hydration reactions are, at all temperatures, characterized by a single major peak. The duration of the peak is less than 1 h for every hydration temperature. The total heat evolved (Fig. 1(b)) reaches a maximum value of approximately 140 kJ/(mol of Al2O3) at 70° and 80°C and is indicative of complete hydration when ettringite is the final product. The kinetics of reaction are apparent in Fig. 1(b) and indicate only small differences with hydration temperature. Specimens hydrated at 30° through 60°C evolve less total heat than those at 70° and 80°C, evolving approximately 115 kJ/ (mol of Al2O3). Figures 2(a) and (b) also show rates of reaction and heat output curves for specimens hydrated in 0.2M NaOH solutions between 30° and 80°C. The rate curves (Fig. 2(a)) are again characterized by single peaks which reach maxima before 1 h for hydration temperatures greater than 50°C and at approximately 2.5 h for hydration at 30° and 40°C. The reaction rates are observed to increase as the temperature is increased. The total heat evolved reaches a maximum value of approximately 120 kJ/(mol of Al2O3) for hydration at 80°C; these values are approximately 110, 95, and 85 kJ/(mol of Al2O3) for hydration at 70°, 50°, and 60°, and 40° and 30°C, respectively. The rate curves shown in Fig. 3(a) for specimens hydrated in 0.5M NaOH are more complex than those observed for the deionized water and 0.2M solutions. There is still a single major peak observed for each specimen, but the peaks are not as distinct as those observed in less concentrated solutions. A distinct endothermic reaction is observed in specimens hydrated at 30°, 40°, and 50°C. The results of integrating the rate curves in Fig. 3(a) are shown in Fig. 3(b). The reaction rates are, as in the 0.2M solution, observed to increase with increasing temperature. The total heat evolved is approximately 80

Table I.

Fig. 1. (a) Rates of heat evolved when monosulfate and gypsum are hydrated in deionized water. (b) Total heats evolved when monosulfate and gypsum are hydrated in deionized water.

kJ/(mol of Al2O3) for samples hydrated between 30° and 60°C, while 140 and 150 kJ/mol are evolved for samples hydrated at 80° and 70°C, respectively. Reaction rate curves for monosulfate and gypsum hydrated in 1.0M NaOH are very similar to those hydrated in 0.5M NaOH, shown in Fig. 4(a). The rate curves all show distinct endothermic reactions, except for the sample hydrated at 80°C. As with hydration in 0.5M NaOH solution, the rate curves are very diffuse without apparent, distinct hydration peaks. The total heat evolved, shown in Fig. 4(b), reaches a maximum of approximately 100 kJ/(mol of Al2O3) when hydration is carried out at 70°C. Samples hydrated in deionized water generated approximately the same amounts of heat between 30° and 60°C, while more heat was liberated from samples hydrated at 70° and 80°C. Because all of the experiments in deionized water were carried out using approximately 1 g of mixed precursor, these

Primary Diffraction Peak Listing

Hydrated phase

Angular position 2␪

Position d-spacing

hkl indices

Crystal structure

JCPDS Ref.

Monosulfate Ettringite Calcium hydroxide U-Phase Gypsum

9.93 9.08 34.08 17.67 11.58

8.90 9.72 2.63 5.01 7.63

003 100 101 006 020

Hexagonal Hexagonal Hexagonal Hexagonal Monoclinic

18–275 41–1451 4–733 44–272 33–311

2902

Journal of the American Ceramic Society—Clark and Brown

Vol. 82, No. 10

Fig. 2. (a) Rates of heat evolved when monosulfate and gypsum are hydrated in 0.2M NaOH. (b) Total heats evolved when monosulfate and gypsum are hydrated in 0.2M NaOH.

Fig. 3. (a) Rates of heat evolved when monosulfate and gypsum are hydrated in 0.5M NaOH. (b) Total heats evolved when monosulfate and gypsum are hydrated in 0.5M NaOH.

results indicate the extent of hydration to be affected by a change in ettringite morphology and/or an unidentified change in the hydration reaction. The total heat evolved varies as a function of temperature for hydration in a given sodium hydroxide concentration; in general the introduction of higher sodium hydroxide concentrations causes the rate of the reaction to be retarded and the total heat evolved to be reduced. Analyses of the XRD patterns shown in Figs. 5, 6, 7, and 8 indicate that ettringite, gypsum (CaSO4⭈2H2O), calcium hydroxide (Ca(OH)2), and a sodium calcium aluminum sulfate hydrate phase (3CaO⭈Al2O3⭈CaSO4⭈0.5Na2SO4⭈15H2O, designated “U-phase”) are the crystalline phases observed. The amounts of these phases vary with hydration temperature and hydration solution. No residual monosulfate was observed in any of the samples. In general, in each sodium hydroxide concentration, the amount of ettringite is observed (by XRD) to increase with increasing temperature. However, the extent to which increased ettringite formation is observed decreases as the concentration of sodium hydroxide is increased. Ettringite and gypsum are the only crystalline phases observed, over the entire temperature range, in samples hydrated in deionized water, and 0.2M and 0.5M sodium hydroxide solutions. Ettringite, the U-phase, and calcium hydroxide are observed, over the entire temperature range, in samples hydrated in 1.0M sodium hydroxide. The primary U-phase XRD peak is also observed to increase with temperature and substantially at 80°C, in the 1.0M solution.

The primary calcium hydroxide peak also increases with temperature above 40°C, in the 1.0M solution. Small XRD peaks observed in samples hydrated at 50°, 60°, and 70°C are also attributable to calcium hydroxide, but their intensities are only slightly above those observed in samples hydrated in 0.2M sodium hydroxide. Again, in general, the amount of gypsum detected by XRD decreases as the sodium hydroxide solution concentration increases. Gypsum detected in the XRD patterns is due to an excess of gypsum present in the precursor mix. As the sodium hydroxide concentration and temperature increase, more of gypsum is observed to go into solution. In the 0.2M solution gypsum is observed to substantially decrease in samples hydrated between 60° and 80°C (indicating a reaction occurring at approximately 50°C). In the 0.5M solution gypsum is observed in samples hydrated at 40° and 50°C, but is not present in samples hydrated between 60° and 80°C (again indicating a reaction is occurring at 40° to 50°C). Small amounts of gypsum are observed in the sample hydrated in 1.0M sodium hydroxide, and at 30°C, none of the other samples hydrated in 1.0M solution show any indication of gypsum. The conditions under which gypsum was not completely consumed or where Ca(OH)2 precipitated were calculated using the aqueous geochemical code PhreeqC.16 These calculations were carried out as a function of temperature (30°, 50°, or 70°C) and NaOH concentration (H2O, 0.2M NaOH, and 1.0M NaOH). Table II summarizes the output from the calculation reporting the pH, the saturation indices for calcium hydroxide

October 1999

Formation of Ettringite from Monosubstituted Calcium Sulfoaluminate Hydrate and Gypsum

2903

Fig. 6. XRD patterns obtained when monosulfate and gypsum are hydrated in 0.2M NaOH at various temperatures.

Fig. 4. (a) Rates of heat evolved when monosulfate and gypsum are hydrated in 1.0M NaOH. (b) Total heats evolved when monosulfate and gypsum are hydrated in 1.0M NaOH.

Fig. 7. XRD patterns obtained when monosulfate and gypsum are hydrated in 0.5M NaOH at various temperatures.

Fig. 5. XRD patterns obtained when monosulfate and gypsum are hydrated in deionized water at various temperatures.

Fig. 8. XRD patterns obtained when monosulfate and gypsum are hydrated in 1.0M NaOH at various temperatures.

(CH) and gypsum (a value of zero indicates saturation), and the moles of solid (gypsum or CH). These calculated results are consistent with the XRD analyses.

tion. Increasing the sodium hydroxide concentration increases the extent of gypsum dissolution via the double decomposition reaction:

IV.

Discussion

The first observation to be discerned from a review of the data concerns the amount of gypsum dissolved into each solu-

CaSO4⭈2H2O(s) + NaOH(aq) ⳱ Ca(OH)2(s) + Na2SO4(aq) Thus sodium hydroxide promotes gypsum dissolution and the formation of calcium hydroxide. Our studies have shown the

2904

Journal of the American Ceramic Society—Clark and Brown Table II.

PhreeqC Output Summary Saturation index

Solution

Temp (°C)

Vol. 82, No. 10

pH

CH†

Gypsum

Amount in solid (mol) Gypsum

CH†

DI H2O

30 50 70

7.0 6.8 6.6

−10.7 −9.8 −9.0

0.0 0.0 0.0

0.064 0.064 0.065

0.000 0.000 0.000

0.2M NaOH

30 50 70

12.6 11.9 11.4

0.0 0.0 0.0

0.0 0.0 −0.1

0.006 0.000 0.000

0.061 0.069 0.072

1.0M NaOH

30 50 70

13.7 13.1 12.6

0.0 0.0 0.0

−2.4 −2.7 −2.9

0.000 0.000 0.000

0.080 0.080 0.080

†

Calcium hydroxide.

dissolution of gypsum in 1.0M sodium hydroxide to be endothermic. The dissolution of gypsum influences both the total heat evolved and the rate of the reaction. This is why the total heat output is observed to decrease with increasing sodium hydroxide concentrations and why samples hydrated in 1.0M NaOH solution undergo a reaction which is initially endothermic. The XRD and calorimetry data indicate the monosulfate/ gypsum mixture converts to ettringite at all temperatures in deionized water or in solutions with low concentrations of sodium hydroxide. This conversion indicates ettringite is the stable phase over this temperature range. The similarity of the hydration rates of all samples in deionized water, regardless of temperature, is consistent with prior work indicating a diffusion-controlled process.17 While the hydration rates of all samples in deionized water are similar, the total heats evolved are slightly different. These differences may be due to morphological and/or crystallinity differences in the final products. Similar small discrepancies between the total heats of samples hydrated in NaOH solutions are also observed. The maximum total heat occurs at 70° or 80°C, evolving 20 to 80 kJ/(mol of Al2O3) more than at lower temperatures. The presence of sodium hydroxide retards the monosulfate/ gypsum to ettringite conversion and lowers the total heat output. For samples hydrated in sodium hydroxide solutions, the hydration rates show a dependence on temperature and sodium hydroxide concentration. This suggests that the hydration process is not solely diffusionally controlled, but may also be controlled by an interfacial process. Small U-phase diffraction peaks are observed in the XRD patterns for the samples hydrated in 0.5M NaOH. The calorimetry data also provides indications (reduced heat output, irregularly shaped rate curves) of a second reaction taking place. When monosulfate is hydrated in 1.0M sodium hydroxide, both calcium hydroxide and the U-phase form; these results are in agreement with recent studies by Li, Bescop, and MoranvilleRegourd.18 U-phase formation, in 1.0M sodium hydroxide, is consistent with the double decomposition reaction CaSO4⭈2H2O(s) + NaOH(aq) ⳱ Ca(OH)2(s) + Na2SO4(aq) As gypsum dissolves in NaOH solution, both the calcium and sulfate ion concentrations increase and calcium hydroxide precipitates. The formation of calcium hydroxide therefore competes with ettringite formation. Thus the amount of calcium hydroxide formed limits ettringite formation and allows for the formation of other phases. The U-phase is formed due to the depletion of calcium from solution due to the common ion effect. Thus calcium is replaced in the AFm structure by sodium or more accurately the calcium sulfate in the structure is replaced by sodium sulfate. To establish whether the hydration reactions fit an Arrhenius model (k ⳱ exp−(Ea /RT ) ), the apparent activation energy for each hydration reaction was calculated. Figure 9 plots the logarithm of the reaction rate versus the inverse temperature for

Fig. 9. Plot showing apparent activation energy for monosulfate hydration to ettringite at various NaOH concentrations.

each of the hydration solutions. The plot shows that the reaction rates are in general nonlinear over the entire temperature range; at higher temperatures (70° and 80°C) the reaction rates decrease significantly in deionized water, and in 0.2M NaOH and 0.5M NaOH solutions. The plot also shows that sodium hydroxide significantly decreases the rate of reaction, although the apparent activation energy, the slope of the linear region along each curve, increases with low concentrations of sodium hydroxide. To determine the apparent activation energy, the slope was taken of each curve from 30° to 60°C in Fig. 7; the apparent activation energies vary as shown in Table III. The apparent activation energies reflect the various phase changes produced during hydration of the monosulfate/gypsum mixture: the dissolution of gypsum and formation of ettringite, calcium hydroxide, and U-phase. In deionized water only ettringite is produced and residual gypsum remains. In 0.2M and 0.5M NaOH solutions ettringite is produced, but the apparent activation energy increase indicates the effect of these solutions on ettringite formation. The slight decrease in the activation energy between the 0.2M and 0.5M solutions indicates the

Table III.

Apparent Activation Energies

Hydration solution

Apparent activation energy (kJ/(mol of Al2O3))

Hydration phases observed by XRD

Deionized water 0.2M NaOH 0.5M NaOH 1.0M NaOH

12.7 68.0 56.5 25.2

AFt AFt AFt, little CH AFt, CH, U-phase

October 1999

Formation of Ettringite from Monosubstituted Calcium Sulfoaluminate Hydrate and Gypsum

formation of calcium hydroxide. Ettringite, calcium hydroxide, and U-phase are all produced in the 1.0M NaOH solution; the apparent activation energy is lower due to the formation of the two additional phases. Examination of the heat evolution curves and the Arrhenius plot suggests the rate-controlling mechanism for each reaction. For samples hydrated in deionized water hydration reaction rates show little change with temperature, indicating a diffusion-controlled process. For samples hydrated in 0.2M and 0.5M NaOH solutions the rates show substantial changes with temperature, indicating an interfacially controlled process. For samples hydrated in 1.0M NaOH solution the reaction rates are again greatly influenced by temperature, but the analysis is complicated by calcium hydroxide and U-phase formation. The reaction kinetics can also be evaluated from the above data. Between 20° and 70°C samples hydrated in deionized water and 1.0M NaOH are consistent with an Arrhenius model; this indicates a first-order reaction. For samples hydrated in 0.2M and 0.5M NaOH the Arrhenius plot (Fig. 9) is nonlinear, indicating the reactions are now dependent on more than one parameter due to competing reactions of ettringite and U-phase formation. V.

Conclusions

The results presented indicate that ettringite is the stable phase, in the reactions examined, over a wide temperature range. Complete hydration produces 140 kJ/(mol of Al2O3) of heat output when ettringite is the only crystalline product. Hydration of the monosulfate/gypsum mixture in increasingly concentrated sodium hydroxide solutions was observed to retard ettringite formation. Competitive reactions of U-phase and calcium hydroxide formation occur as the sodium concentration is increased, decreasing the total heat output and the total amount of ettringite produced. The consequences of these observations could be the formation of U-phase, which can be expansive, in a portland cement or the delayed formation of ettringite, both of which could lead to crack formation in a hardened concrete structure.

2905

References 1 I. Odler and I. Jawed, “Expansive Reactions in Concrete”; pp. 221–47 in Materials Science of Concrete II. Edited by J. Skalny and S. Mindess. American Ceramic Society, Westerville, OH, 1991. 2 H. F. W. Taylor, Cement Chemistry; p. 167. Academic Press, New York, 1990. 3 S. K. Chatterji, “Mechanism of Sulfate Expansion of Hardened Cement Paste”; pp. 336–41 in Proceedings of the 5th International Symposium on Chemistry of Cement (Tokyo, Japan, 1968), Vol. 3, 1969. 4 G. Li, P. Le Bescop, and M. Moranville, “Expansion Mechanism Associated with the Secondary Formation of the U-Phase in Cement-Based Systems Containing High Amounts of Na2SO4,” Cem. Concr. Res., 26 [2] 195–201 (1996). 5 N. Tenoutasse, “The Hydration Mechanism of C3A and C3S in the Presence of Calcium Chloride and Calcium Sulfate,” Proc. Int. Symp. Cem., 5th, 2, 372–78 (1968). 6 J. Pommersheim and J. Chang, “The Kinetics of Hydration of Tricalcium Aluminate,” Cem. Concr. Res., 16, 440–50 (1986). 7 K. L. Scrivener and P. L. Pratt, “Microstructural Studies of the Hydration of C3A and C4AF Independently and in Cement Paste,” Br. Ceram. Proc., 35, 207–19 (1984). 8 P. W. Brown and P. LaCroix, “The Kinetics of Ettringite Formation,” Cem. Concr. Res., 19, 879–84 (1989). 9 P. W. Brown, “Kinetics of Tricalcium Aluminate and Tetracalcium Aluminoferrite Hydration in the Presence of Calcium Sulfate,” J. Am. Ceram. Soc., 76 [12] 2971–76 (1993). 10 D. Damidot and F. P. Glasser, “Thermodyanmic Investigation of the CaO– Al2O3–CaSO4–H2O System at 25°C and the Influence of Na2O,” Cem. Concr. Res., 23, 221–38 (1993). 11 P. W. Brown and J. V. Bothe Jr., “The Stability of Ettringite,” Adv. Cem. Res., 5 [18] 47–63 (1993). 12 W. Dorsch and H. zur Strassen, “An Alkali-Containing Calcium Aluminate Sulfate Hydrate,” Zem.-Kalk-Gips, 20, 392–401 (1967). 13 G. Li, P. Le Bescop, and M. Moranville, “The U-Phase Formation in Cement-Based Systems Containing High Amounts of Na2SO4,” Cem. Concr. Res., 26 [1] 27–33 (1996). 14 H. F. W. Taylor, Cement Chemistry; p. 193. Academic Press, New York, 1990. 15 K. S. TenHuisen, “The Formation of Biocomposites at Physiological Temperature”; Master of Ceramic Science Thesis. Pennsylvania State University, University Park, PA, 1992. 16 D. L. Parkhurst, “User’s Guide to PhreeqC—A Computer Program for Speciation, Reaction-Path, Advective-Transport, and Inverse Geochemical Calculations,” Water Resources Investigative Rept. No. 95-4227, U.S. Geological Survey, Denver, CO, 1995. 17 E. Grusczscinski, P. W. Brown, and J. V. Bothe, Jr., “The Formation of Ettringite at Elevated Temperature,” Cem. Concr. Res., 23, 981–87 (1993). 18 G. Li, P. Le Bescop, and M. Moranville, “Synthesis of the U Phase (4CaO⭈0.9Al2O3⭈1.1SO3⭈0.5Na2O⭈16H2O),” Cem. Concr. Res., 27 [1] 7–13 (1997). 䊐