The definite aid which chemical analysis may render in determining the quality of a cement is limited by the following considerations. It is not definitely known just what part is played by each of the compounds that go to make up commercial cement, and chemical analysis does not tell the manner of the occurrence of these compounds. A cement may have a chemical composition that is thought to be perfect, but if the burning has not been properly accomplished, it may be a dangerous product and analysis would show no defect. Some of the best authorities say that chemical analysis is useful principally in tracing the cause of defects which, by other tests, have been found to exist. However, there are some constituents which it is fairly well known a cement should not contain in any considerable quantities. An analysis may be of value in estimating quantitatively such constituents, while it may also be of service in detecting adulterations. It is not impossible, then, that chemical tests may yet play a more important role in cement testing, especially if the method of analysis can be made more simple and rapid, without too great a sacrifice of accuracy.
The proportion of lime in Portland cement may vary from 59 to 67 per cent. A much greater range than this is allowable in natural cement, the percentage usually being from 30 to 45, according to the amount and character of the other active constituents. An analysis of Portland cement which shows a percentage of lime far outside of the limits mentioned above, should be regarded with suspicion and submitted to very thorough tests before acceptance. As already stated, the ratio of the silica and alumina to the lime in a cement is called the hydraulic index. The value of this ratio is usually between .42 and .48 for Portland cement.
Cement mixtures containing a large percentage of lime require a high temperature for calcination, are difficult to grind, and yield a slow-setting product. The danger in highly limed cements is that they will not be properly calcined and a portion of the lime will be left in a free state. The demand for high strength in short-time tests has led manufacturers to make a heavily limed product, and in some cases the limits of safety have probably been overstepped. The introduction of the rotary kiln, however, has so improved the facilities for burning cement that a higher percentage of lime is now possible.
There is no method known at present for determining quantitatively the amount of free lime in a cement, and it seems doubtful whether its presence can be detected with certainty by chemical analysis. The method usually employed for this purpose depends on the hydration of the lime and subsequent absorption of carbonic acid.
The detection of magnesia in several concrete structures that had failed, led to the conclusion that magnesia, in quantities exceeding two or three per cent., was a dangerous element in Portland cement. In 1886-87 Mr. Harrison Hayter1 mentioned several failures of masonry and concrete which he considered were due to magnesia, and concluded that cement should not contain more than one per cent. Later investigations, however, indicated that such failures could be explained in other ways, and that the magnesia found in the failing structure had come from the sea water and replaced the lime in the cement. Mr. A. E. Carey2 has considered that "an excess of caustic lime or magnesia causes first, disintegration by expansion due to hydration, and second, being soluble, when conditions permit of their washing out, leave the concrete in a honeycombed state." Notice that this refers to caustic magnesia, and Prof. S. B. Newberry3 has stated that "it is doubtful if magnesia is ever combined in Portland cement. Our own experiments tend to confirm the opinion of many German authorities that magnesia remains free in cement and does not combine with the constituents of clay after the manner of lime".
1 Proc. Inst. C. E., Part 1, Session of 1886-87.
2 Ibid., 1891-92.
3 Municipal Engineering, October, 1896.
On the other hand, M. H. LeChatelier1 says that the "accidents occasioned by certain magnesian elements, and the similar results obtained in laboratory experiments, have been due to the employment of badly proportioned cements, containing free uncombined magnesia and too small a quantity of clay. Corresponding mixtures containing lime instead of magnesia would have caused still more serious accidents, yet it would not be concluded that there must be no lime in cement." Again, Dr. Erdmenger characterizes magnesia as an adulterant only, and considers that its effect is nil if a greater percentage of lime is added in the manufacture.
Some authoritative information on the amount of magnesia allowable in Portland cement is contained in the report of the magnesia commission of the Association of German Cement Makers, 1895: Three members of this committee, Messrs. Schott, Meyer and Arendt concluded that "the presence of magnesia up to ten per cent, causes no harmful expansion or cracking of the cement, even after several years." Mr. DyckerhofT, however, presented a minority report, in which he pointed out that while a large amount of magnesia, not sintered, may not have an injurious effect, yet a content of more than four per cent, of sintered magnesia, whether added or substituted for part of the lime, has an injurious effect after long periods. The committee continued the ruling of 1893 that "a magnesia content of five per cent, in burnt cement is harmless," but held the question open for further investigation, indicating that this limit might be raised.
In view of the disagreement among such eminent authorities it is impossible to arrive at a satisfactory conclusion, but if the effect of magnesia depends upon the manner of its occurrence, whether free or combined, sintered or unsintered, then chemical analysis can be of but limited value as a test of quality in this regard. Natural cements frequently contain large proportions of magnesia replacing lime, and in this case an analysis is of the same value as an analysis for lime.
The amount of alumina which a cement should contain is not well established. Its presence tends to facilitate the burning, and it renders the product quicker setting. Cements containing large percentages of alumina are inferior for use in air or sea water, and it is probable that the percentage of alumina should not exceed eight or ten to obtain the best results in all media. A slag cement may be detected by its large content of alumina. Oxide of iron acts as a flux in burning, but in the finished product is little more than an adulterant.
1 Trans. Amer. Inst. Mining Engrs., 1893.
French specifications say that Portland cements shall not contain more than one per cent, of sulphuric acid or sulphides in determinable proportions. This is doubtless intended for cement to be used in sea water. Adulterations with blast-furnace slag may sometimes be detected by the amount of sulphides present, but small quantities of sulphuric acid in the cement may be derived from the coke used in burning and have no injurious effect for use in fresh water. A content of 1.75 per cent, of sulphuric anhydride, S03, is now considered the maximum permissible. Sulphates mixed with the raw materials and burned with the cement may be harmless, while the same amount added after burning would not be permissible. [For tests on the effect of adding sulphate of lime to cement, see Art. 48].
The determination of these may give some idea of the deterioration of a product by storage, and they may also indicate defective burning. M. Candlot considers that in the case of Portland cement, a loss on ignition (water and carbon dioxide) exceeding three per cent, "indicates that the cement has undergone sufficient alteration to appreciably diminish its strength." Natural cements may, however, contain considerable proportions of these ingredients and still give good results.
Finally, then, the determination of silica, alumina, magnesia and lime may be of value, first, in classifying a product, and second, as indicating whether the proportions contained in it are such that if properly manufactured it is capable of giving good results. What these proportions should be for Portland cement has already been stated, § 9. The determination of certain injurious ingredients is also of some value, but it must be remembered that the dangerous elements most commonly occurring, namely, free lime and magnesia, are not determinable by chemical analysis. It has been stated by M. LeChatelier that "neither complete nor partial chemical analysis of the constituents of hydraulic materials can be ranked among normal tests. But chemical analysis may render real service in controlling the classification of a product concerning which there is reason to doubt the declaration of the manufacturer. Thus, a slag cement can be distinguished from a Portland by its tenor in alumina and water; certain natural cements, by their contents of sulphuric acid, etc." 1
The methods of analysis for Portland cement are given in considerable detail in a little book, "The Chemical and Physical Examination of Portland Cement," by Richard K. Meade. The method of analysis suggested by the New York Section of the Society of Chemical Industry is published in the Engineering Record of July 11, 1903, and in Engineering News of July 16, 1903.
1 "Tests of Hydraulic Materials," H. LeChatelier.