C.E. Buchanan, Jr.¹

^{Rapid Determination of the Predominant Form of Calcium Sulfate Found in Portland Cement and Its Effect on Premature Stiffening}

REFERENCE: Buchanan, C. E., Jr., "Rapid Determination of the Predominant Form of Calcium Sulfate Found In Portland Cement and Its Effect on Premature Stiffening," Cement, Concrete, and Aggregates, CCAGDP, Vol. 2, No. 2, Winter 1980, pp. 84-88.

ABSTRACT: The object of this research program was to enable us to predict very rapidly the performance of portland cement in the field in relationship to abnormal setting characteristics. A cement water leach is prepared, and the filtrates are analyzed for calcium and sulfate. A sodium (DI) ethylenediaminetetraacetate acid (EDTA) titration is used for the calcium and a turbidimetric method is used for the sulfate. Based on the results obtained with this chemical procedure, quality control methods that fairly accurately predict whether a cement contains an excessive amount of either anhydrite or plaster of paris have been prepared.

KEYWORDS: cements, calcium sulfates, portland cements, stiffening

The data presented in this paper are limited to premature stiffening caused either by gypsum precipitation or by rapid hydration of the alumina phases. To clarify this further, "false set" is used to mean the abnormal setting characteristic of Portland cement occurring when a hydrated form of calcium sulfate having less than two molecules of water of crystallization precipitates quickly to form gypsum with little or no heat evolution and the cement can be restored to its initial plasticity without the addition of water. "Flash set" is defined as that setting phenomenon of Portland cement in which the tricalcium aluminate hydrates directly with considerable heat evolution and the subsequent cement cannot be restored to a plastic condition without additional water.

Gypsum has been used as a retarder probably since close to the inception of the Portland cement industry. The early experimenters found that Portland cement clinker when ground by itself would hydrate rapidly and that calcium sulfate was excellent for slowing this reaction to provide sufficient time for manipulation and placement and produced better strength.

This addition, however, presented some other problems, in that a false set could occur. False set does not present a major difficulty to the Portland cement industry as long as extended mixing is used, which usually occurs in a ready-mix truck. In the late 1930s, however, it became a problem when relatively large dams were constructed in the western part of the United States. The concrete, batched at the job site, was dumped almost immediately into forms with little or no manipulation. As a result of this, revised federal specifications included a test designed to identify cements having this characteristic. Other than such an application, false set did not really become a major problem until slip-form pavers came into common use. Some of this cement was mixed only 45 s from the time of water-cement contact to placement.

Researchers at Penn-Dixie, O. J. Glantz, F. M. Parise, and the author, became acutely aware of this false-set problem, especially in the Midwest, and expended considerable research money to develop a rapid method for the prediction of potentially false-setting cement. In the preliminary work, samples of natural terra alba gypsum were heated to a variety of temperatures, that is, 125, 175, 300, and 600°C (257, 347, 572, and 1112°F). Some samples were kept at each of these temperatures for 24 h to assure the proper amount of dehydration. The samples were then immediately removed and placed in sealed containers. It was assumed that after the dehydration procedure the samples were composed of plaster of paris, soluble anhydrite, a mixture of soluble and insoluble anhydrite, and insoluble anhydrite, respectively.

A 1-g sample of each material was then shaken with 100 g of water for periods of 1, 5, 15, and 60 min, immediately filtered, and then a sulfur trioxide determination was made. These results, showing sulfur trioxide solubility, are shown in Fig. 1.

Note that gypsum solubility is essentially a straight line with time. Soluble anhydrite or plaster of paris (the 125°C [257°F] and 175°C [347°F] samples) started off at high solubilities (in excess of 50 mmols sulfur trioxide per litre) but at the end of 60 min went down to slightly above the gypsum line. The most interesting sample is the one heated or formed at 300°C (572°F); it apparently held (at least through 60 min) a higher solubility than is expected from gypsum. The 600°C (1112°F) material started off with an extremely low sulfur trioxide solubility (similar to natural anhydrite), but by 60 min sufficient gypsum had been formed so that a solubility approaching that of gypsum resulted.

As a sidelight, the optimum amount of sulfur trioxide was determined with these various forms of calcium sulfate by using a sample of high-alkali high-tricalcium aluminate Type-I clinker. The clinker was ground and blends were made at 0.3% sulfur trioxide increments. The one-day compressive strengths were determined in accordance with ASTM Test for Compressive Strength of Hydraulic Cement Mortars (Using 2-in. or 50-mm Cube Specimens) (C 109) and are shown in Fig. 2.

It can be noted that gypsum gave an optimum of 3.6% sulfur trioxide while the 125°C (257°F) temperature material gave optimum results at 2.8%, the 300°C (572°F) material at 2.9%, and the 600°C (1112°F) material at 3.8%. Note also that the strength differences were fairly marked with the same clinker being able to produce a one-day compressive strength between 7.23 and 12.26 MPa (1050 and 1780 psi), depending on which type and the amount of calcium sulfate used.

Based on these results, a review was made of Hansen's data [1] on the solubility of calcium oxide, sulfate, sodium, and potassium hydroxide in moles per litre. Liberty has been taken in rearranging these data to plot the solubility product of calcium and sulfate versus the total alkali content of the solution. This information is presented in Fig. 3, which shows an excellent logarithmic relationship.

Based on this figure it was apparent that once the alkali content of the solution was determined, it was possible to ascertain if the cement in question was supersaturated or undersaturated with respect to gypsum simply by noting where the solubility product fell on this regression line (above or below).

A rapid procedure (Appendix I) was then developed to determine the calcium and sulfate concentrations. Basically, a cement is leached with water for a given time.² The sample is rapidly filtered and acidified and aliquots are taken for lime and sulfate determinations. The lime (Appendix 2) is titrated with sodium (DI) ethylenediaminetetraacetate (EDTA) acid and hydroxy naphthol blue (Calcium Indicator) is used as the indicator. This test has an excellent end-point and is rapid and very precise. The sulfur trioxide (Appendix 3) is determined with the turbidimetric equipment available at most cement laboratories, different standards are used than for portland cement analyses and the microammeter is modified slightly.

The normal procedure is to leach for 1, 3, and 5 min and use these data as indications of the potential false-setting characteristics of the cement. A 60-min leach is used to determine the saturation level. The 60-min leach is a more rapid and probably a more accurate method of determining the gypsum saturation than determining the water soluble alkali content. The 1-, 3-, and 5-min data are divided by the 60-min leach, and these ratios are used to express the degree of either supersaturation or undersaturation of the particular cement.

In Figs. 4 and 5, data are shown in which the leach test results are plotted against laboratory air exposure for two cements. It is interesting to note that neither of the cements was false-setting upon receipt, but on exposure they both developed false-setting characteristics as determined by two physical methods of measuring premature stiffening and by the leach test. Cement A (Tables 1 and 2 ) kept its false set for six days. However, at the end of two days the false set of Cement B (Tables 3 and 4) had disappeared as determined by ASTM Test for Early Stiffening of Portland Cement (Paste Method) (C 451), and by six days it had disappeared as determined by ASTM Test for Early Stiffening of Portland Cement (Mortar Method) (C 359). The contradiction between one method showing false set and the other one not could be explained by the leach data and their relationship to the different mixing times used by the two methods.

Parenthetically, measurements of the resistance of Portland cement to fluid flow were made on these samples; as they became more false-setting, the resistance first increased and then decreased with further exposure.

Our normal control parameters are to keep the leach ratios around 1 1/2 at 1, 3, and 5 min. If this is done, there is sufficient calcium and sulfate in solution to prevent flash-setting but not enough to induce false set. When anhydrite is used to control false set, it is extremely important not to let the saturation ratio level go much below 0.9 at any time. If this drop does occur, the cement will probably not flash set but will become very sensitive to admixtures, as they apparently tie up with the calcium sulfate in some manner or make the tricalcium aluminate more reactive.

Presented at the Symposium on Nonstandard Test Methods Useful in Determining Physical-Chemical Characteristics of Cements, held in San Diego on 12 Dec. 1979 by ASTM Committee C-1 on Cement and chaired by Albert W. Isberner of the Portland Cement Association.
¹Vice president of operations, Penn-Dixie Industries, Inc., P.O. Box 162, Nazareth Pa. 18064. Member of ASTM.
²Private communication with Dr. Vance Dodson, W. R. Grace Co., Cambridge, Mass.
 
 

TABLE 1-Premature stiffening and resistance to fluid flow of Cement A.  

TABLE 2-Leach ratios for Cement A.  

TABLE 3-Premature stiffening and pack-set of Cement B.  

TABLE 4-Leach ratios for Cement B.  

APPENDIX 1

Determination of the Ca⁺⁺ and S0₄⁼ Ions

Place 100 g of cement in a 500-mL Erlenmeyer flask. Add 75 mL of water, stopper, and shake for the leach time desired. (If a magnetic stirrer is available, stirring may be substituted for shaking.) Immediately pour into a Buchner funnel that has been fitted with a prewetted filter paper and is under vacuum to a filtering flask. To the filtrate obtained, add one or more drops of concentrated hydrochloric acid, the amount being just sufficient to dissolve any precipitate formed.
 
 

APPENDIX 2

Determination of Calcium

EDTA Solution

Dissolve 3.7222 g of EDTA in water and dilute to 1 L. When a 10-mL sample of filtrate is taken, 1 mL of EDTA is equivalent to 1 mmol (0.001 mol) of calcium per litre of original solution.
 
 

Potassium Hydroxide Solution

Dissolve 250 g of potassium hydroxide water and dilute to 1 L.
 
 

Hydroxy Naphthol Blue (Calcium Indicator)

Place exactly 10 mL of the leach solution in a 250-mL beaker. Add 2 mL of 1:1 hydrochloric acid and bring to a boil. Dilute to approximately 100 mL with hot water, add 20 mL potassium hydroxide solution and approximately 0.2 g of hydroxy naphthol blue. Titrate to the complete disappearance of the red color. The amount of EDTA (in millilitres) used is equivalent to the millimoles of calcium per litre of original solution.

The pH should be above 11.5 at all times during the titration. The concentration of the potassium hydroxide solution used can be adjusted to provide these limits.
 
 

APPENDIX 3

Determination of Sulfate

Dissolve 1.7427 g of reagent-grade potassium sulfate in water and dilute to 1 L. This is a 0.01-molar solution. Withdraw 5 mL of this solution, place in a sulfur trioxide turbidimeter jar, add 5 mL of 1:1 hydrochloric acid, and dilute to the reference mark. Place the jar in the turbidimeter with the filter up and set the needle at approximately 45, noting the exact reading. Add the required quantity of barium chloride crystals (approximately 1 g), remove from the turbidimeter, oscillate for 90 s, replace in the turbidimeter, and lower the filter. If the needle goes off scale, repeat the entire procedure lowering the initial needle setting until a reading on scale is determined. Repeat this procedure with the arrived-at setting with 10, 15, 20, 25, 30, 40, and 50 mL of the potassium sulfate solution. The amount (in millilitres) of potassium sulfate used is then plotted against the sulfur trioxide turbidimeter readings. With the Penn-Dixie research microammeter, it was found that approximately 3000 0 resistance had to be placed in series with the microammeter to maintain accurate readings over the entire range.

Place 10 mL of the leach solution in the turbidimeter jar and repeat the entire procedure, observing the turbidimeter curve. The amount (in millilitres) of the potassium sulfate stock solution used in calibration is read from the curve and this is the value in millimoles per litre of sulfate present in the original leach solution.
 
 

Reference

[1] Hansen, W. C., "Quick and False Set in Portland Centents," Materials Research and Standards, Vol. 1, No. 10, Oct. 1961, pp. 791-797.