STATE OF CALIFORNIA EARL WARREN. Governor DEPARTMENT OF NATURAL RESOURCES WARREN T. HANNUM. Director DIVISION OF MINES FERRY BUILDING, SAN FRANCISCO 11 OLAF P. JENKINS. Chief SAN FRANCISCO SPECIAL REPORT 27 JANUARY 1953 ALKALI -AGGREGATE REACTION IN CALIFORNIA CONCRETE AGGREGATES By RICHARD MERRIAM Digitized by the Internet Archive in 2012 with funding from University of California, Davis Libraries http://archive.org/details/alkaliaggregater27merr ALKALI-AGGREGATE REACTION IN CALIFORNIA CONCRETE AGGREGATES By Richard Merriam • OUTLINE OF REPORT m ay facilitate breakdown by other processes such as freez- a if e ing and thawing or wetting and drying. Introduction 3 ^ s minerals and rocks make up the bulk of the aggre- Alkali-aggregate reaction 3 gates involved, the subject of alkali-aggregate reaction is Evidence 3 of interest to geologists, particularly petrographers, who Causes Tr"""": t are the best fitted to deal with the problems of explaining Reactive aggregate constituents 5 , , , „ ,. . r , *.••*.* Reactive California aggregates-- 6 the phenomenon and ot predicting the reactivity ot un- Prevention of reaction 10 tried aggregates. Petrographic examination is the most References 10 rapid method of preliminary evaluation of aggregates. Numerous organizations have taken part in the investi- lllustrations ^ gation of alkali-aggregate reaction. Some of the first Figure 1. Graph showing expansion of mortar bars *4 research was done by the California Division of Highways 2. Photo of cracks in Parker Dam 4 Materials and Research Department in Sacramento under 3. Photo of cracks in pier 5 the direction of T. E. Stanton. This laboratory has made 4. Phot,, of popout around shale fragment.. .__ 5 numerous tests with a wide variety of aggregate materials 5. Photo of repaired section of bridge railing b „ ., „ , J po ° 6. Photo of pattern cracking in bridge pier 6 trom a11 Parts Ot the Mate. 7. Photo of popout around shale fragment 7 The U. S. Bureau of Reclamation is one of the most 8. Photo of cracking in bridge pier__ 7 active agencies investigating in this field. Studies are made 9. Photo of pattern cracking in concrete floor.. 8 in a Moratory maintained by the regional office in Sacra- 10. Photo of cracking in steps 8 ,,"'., „ . . T , • • t-w 11. Graph showing expansion plotted against percent mento and by the Engineering Laboratories in Denver, reactive aggregate 8 Reports by members of the staff of the Denver laboratories 12. Map of California showing distribution of reactive constitute one of the best sources of information on this rocks 9 and related subjects. In the course of planning and construction of various engineering structures the U. S. Corps of Engineers Pa- The principal cement-aggregate reactions occurring in California cific Division Testing Laboratory formerly in Sausalito are between cement alkali and opaline shale or intermediate to acid has studied many aggregates in the Southwest although volcanic rocks Expansion, cracking and deterioration usually result ]ittle f their data hag fe published . from such reactions. The reactivity of an aggregate can be estimated * by service history, mortar bars, chemical tests, and petrographic Acknowledgment . The writer is indebted to Dr. Rich- examination. Reaction may be inhibited by limiting the cement ard C. Miclenz, Head, Petrographic Laboratory TJ. S. alkalies or by adding a corrective pozzolan. Bureau of Reclamation, Denver, Colorado, who has kindly read and criticized the manuscript of this paper. INTRODUCTION * For more than a decade certain California concrete alkali -aggregate reaction aggregates have been known to be chemically reactive Evidence with high alkali cements. Consequently a great deal of Expansion. Expansion of concrete, one of the most research has been done on these aggregates to determine common manifestations of alkali-aggregate reaction, varies the exact nature of the reaction and to establish which in rate and quantity under varying conditions and ma- specific rock types enter into the reaction. Studies have terials. It is an approximately quantitative measure of been made to determine methods of preventing reaction reactivity, although anomalies are common. Unlike some and tests have been devised for estimating the suscepti- of the minor surface phenomena associated with reactivity, bility of an aggregate. Much of the resulting data is tin- expansion affects the entire mass of concrete. The results published and that which has been published is dissemi- depend upon the type of structure; commonly cracks de- nated through the literature. The purpose of this report velop by volume changes, forcing such engineering struc- is to bring this information together rather than to present tures as gates or generators out of alignment, and buckling new material or concepts. paving or curbing. The principal cement-aggregate reaction is termed Laboratory experiments with mortar bars have pro- alkali-aggregate reaction. It is a chemical process in which duced expansion in excess of 1 percent in 1 year. Experi- soluble alkalies (chiefly sodium and potassium) released ments have shown fhat expansion is dependent upon high by the hydration of high alkali cement react with siliceous alkali cement, reactive aggregate and available moisture, constituents of the aggregate to produce alkali silica gel. Figure 1 illustrates such experimental results. Expansion and deterioration of the concrete normally Popouts. Popouts are surface features formed by local- accompanies the process. Although the nature of deteriora- iz p d expansion which spalls off irregular- to conical-shaped tion depends in part upon the type of structure, it is gen- fragments an inch to several inches across. They usually erally characterized by cracking, and in extreme cases by develop in concrete containing aggregate having a small displacement of cracked blocks, and may necessitate ex- percentage of reactive material in the coarse fraction, pensive repairs or even complete replacement. Although Such large reaction particles act as centers of reaction and alkali-ag gregate reaction may not in itself be so serious, it expansion. •tiniv^itv „ f o„,,^ , ^ , -, • t „ , „ Pattern Cracking and Microfractures. The largest and university of Southern California. Los Angeles. Manuscript sub- . . - '. • • .u ,. c j. mitted for publication May 1952. most prominent fractures originate as the result ot stresses (3) Special Report 27 Figure 1. Graph showing expansion of mortar bars using high alkali cement and reactive California aggregates. Curve I — Friant andesite (from McConnell et al. 1950) ; Curve II — Saticoy aggre- gate (opaline shale) (from Stanton 19//2). Reaction Rims. The chemical reaction between alkalies and rock or mineral fragment proceeds inward from the surface of the fragment. In certain rock types the affected portion is distinct from the unaffected portion, the former being darker in reflected light and lighter in transmitted light. Such rims are most common in vitreous or hemi- crystalline andesites, dacites and rhyolites. As pointed out by McConnell et al. 1 all rims of this appearance are not due to alkali reaction ; they may be formed by weathering or some other geologic agent. Rims can be assumed to be indicative of reactivity only if they are much more com- mon in the concrete than in the unused aggregate. If they occur around rock or mineral fragments known to be reactive their significance is greater. Causes Chemistry of the Reactions. The exact nature of the chemical reaction involved is not known but it appears likely that sodium and potassium released during hydra- tion of the cement is concentrated as setting up occurs. They react with unstable siliceous materials such as vol- canic glass, opal, etc. to form alkali-silica gel. In at least some cases the reaction may release alkalies of the aggre- gate which in turn react with unaltered aggregate. Thus reactions may go on indefinitely, as for example at Stewart Mountain Dam, Arizona, where reaction has con- tinued for 20 years. 2 1 McConnell, D., Mielenz, R. C, Holland, W. Y., and Greene, K. T., Pet- rology of concrete affected by cement-aggregate reaction : Geol, Soc. America Mem. Berkey volume, pp. 234-235, 1950. 1 Mielenz, R. C, personal communication, 1952. ( arising from expansion, but others, shown in some struc- tures as a network of cracks, have a more obscure origin. The size of fractures and complexity of pattern varies widely ; in the more seriously affected structures the open- ings thus produced may be filled with secondary products of decomposition. Cracks of similar appearance may be produced by over finishing or excessive water but such cracks are generally limited to the surface. Cracking of concrete always results in lowered strength ; however, a weakening may occur without such obvious features as visible cracks. Stresses developed in the con- crete may be insufficient to cause rupture but enough to accelerate breakdown under such mechanical distortion as compression. Mortar bar tests have shown that expansion due to alkali-aggregate reaction is accompanied by a decline in strength and dynamic modulus of elasticity. In some tests tensile strength decreased more rapidly than compressive strength. Siliceous Gels. Siliceous gels also are good evidence of reaction between cement and aggregate. The gel, which forms on surfaces or in voids within the concrete, ranges in consistency from watery and gelatinous to rubbery. Desiccation or carbonation occurs upon exposure to the atmosphere, when the gel becomes hard and white. Chemi- cal analyses show that the gels consist chiefly of silica (50 to 80 percent) , alkalies (5 to 26 percent) , and water. / Figure 2. Parker Dam. Cracks shown are the result of aggregate reaction. Alkali-aggregate Reaction Figure 3. Cracks in pier, Sixth Street Bridge, Los Angeles. Mechanism of Reaction. Silica gel formed in aggregate particles is surrounded in part by cement paste which acts as a semipermeable membrane. If water is available it is absorbed osmotically by the gel. Osmotic pressures thus developed within the gel may exceed the strength of the enclosing concrete and result in fractures. More gel then accumulates in the fractures which in turn expand by inhibition of water. Pressure in excess of 500 P.S.I, has been produced experimentally. 3 Miscellaneous Factors. Some structures built with a given aggregate have been found to deteriorate whereas other structures made of the same aggregate show no signs of deterioration. The difference is believed to be due to differences in the alkali content of the cements. Similar differences have been observed in mortar bar experiments. Other conditions being equal, expansion is roughly equiva- lent to the alkali content of the cement. Although the total alkali content (calculated as Na 2 0) rarely exceeds 1.5 percent, reactions may take place where concentrations are as low as 0.3 percent. In general, however, an amount of 0.6 percent or less is thought to be safe. It is not defi- nitely known whether this permanently inhibits reaction or merely delays it. One of the first discoveries in experimental work showed that the amount of expansion is not necessarily propor- tional to the amount of reactive aggregate. For" each reactive material there is a certain percentage giving maximum expansion ; more or less than this amount will result in less expansion. In general the more reactive min- erals and rocks produce maximum expansion with smaller amounts of aggregate and the less reactive materials re- quire greater amounts for maximum effect. Figure 11 shows these relationships for some California aggregate minerals. A seemingly anomalous relationship exists between ex- pansion and size of aggregate particles. Very reactive * McConnell, et al., op. cit., p. 246. Figure 4. Popout around opaline shale fragment, Sixth Street Bridge, Los Angeles. Scale shown is 6 inches in length. materials such as opal show increased expansion with de- crease in particle size. Very reactive materials such as opal show increased expansion with decrease in particle size, whereas less reactive materials usually show more expan- sion with larger sizes. Climatic conditions may be responsible for continuing or accentuating a deterioration initiated by alkali-aggre- gate reaction. The expansion and cracking, although pos- sibly incipient and imperceptible, renders a concrete more susceptible to such common weathering processes as hydra- tion, solution, carbonation, and freezing and thawing. A slightly reactive aggregate may thus be satisfactory if used in a mild climate but unsatisfactory if used in a rigorous climate. Therefore in the testing and evaluation of aggregate the anticipated conditions of use should be considered before final recommendations are made. REACTIVE AGGREGATE CONSTITUENTS Knowledge of the reactivity of a wide variety of ma- terials has become available through practical experience and laboratory tests. Mielenz 4 has tabulated the rocks and minerals known to be deleterious with high alkali cement. Most common rock-forming minerals react to a very slight, insignificant extent. McConnell et al. 5 have found the following minerals to be essentially innocuous: quartz, microeline, albite, oligoclase, andesine, bytownite, nephe- line, hornblende, diopside, augite, olivine, biotite, phlogo- pite, muscpvite, vermiculite, prochlorite, almandite, kaolin, montmorillonite, serpentine, talc, analcite, stilbite, prehnite, epidote, dolomite, calcite, apatite, and collo- phane. 'Mielenz, R. C, Petrographic examination of concrete aggregates: Geol. Soc. America Bull., vol. 57, p. 312, 19 46. « McConnell, et al., op. cit, p. 238. Special Report 27 Table 1. Reactive rocks and minerals (from Mielenz). Reactive minerals Chemical composition Physical character Opal Chalcedony Tridymite SiOs.nHzO SiO. SiO* Amorphous Cryptocrystalline fibrous Crystalline Reactive rocks Reactive component Siliceous rocks : Opaline chert Chalcedonic chert Siliceous limestone Opal Chalcedony Chalcedony and/or opal Volcanic rocks :* Rhyolites and rhyolite tuff Dacite and dacite tuff Andesites and andesite tuff I Volcanic glass, devitrified glass and tridymite Metamorphic rocks : Phyllites Hydromica (?) Miscellaneous rocks : Any rocks containing vein- lets, inclusions or grains of the reactive rocks or minerals listed above. •The volcanic types listed are known to be reactive; basalts are known to be Innocuous ; data regarding trachytes, latites and phonolites are lacking. Reactive California Aggregates The following discussion does not attempt to include all kinds of alkali-aggregate reaction in California since many data of this type are not published. However, ex- amples of most of the deleterious rocks and minerals are given (see table 1). Opaline Shales and Related Bocks. Sedimentary rocks of this type are widespread in the Miocene of California. 9 Figure 6. Pattern cracking in pier, Sixth Street Bridge, Los Angeles. Figure 5. Repaired section of railing, Sixth Street Bridge, Los Angeles. Their lithology has been well described by Bramlette. 6 All rocks of this group are distinguished by a much higher SiOo content than characteristic of the average shale or mudstone. Many types of siliceous rocks occur in the group, there being a complete gradation between the various members. Some of the principal types are : rela- tively soft, diatomaceous rocks ; harder, flinty or porcel- laneous rocks ; and cherts or cherty shales. The diatomaceous group consists of diatomaceous shale, diatomaceous mudstone and siltstone, and diatomite. The reactive constituent is believed to be opal composing the diatoms ; it has a fairly uniform composition of about 89 percent Si0 2 and 11 percent H 2 (index of refraction 1.440). Many silica-cemented rocks which are less vitreous than chert have a dull luster similar to unglazed porcelain and show little or no lamination. These are termed porce lainite, and consist chiefly of clay or silt with a large amount of opaline silica. The term chert is used for dense, vitreous rocks made up largely of opaline silica with some chalcedony. It maj show thin banding but there is no tendency to separate along the bands except in the more impure varieties t( which the term cherty shale is applied. Limestone is rare in this group of rocks but impure cal careous or dolomitic rocks form thin beds. These rock: usually contain a small but variable amount of opalim silica and are highly reactive. This formation is widely distributed. In most areas i has been mapped as the Monterey formation but severa other names have been applied locally. The following tabu lar summary of names of siliceous shale formations h California suggests the wide variation of names used fo: units at least partly equivalent to the Monterey formation 8 Bramlette, M. N., The Monterey formation of California and th origin of its siliceous rocks : U. S. Geol. Survey, Prof. Paper 21$ 1946. Alkali-aggregate Reaction* Miocene Opaline Shale and Related Rocks. Locality Berkeley Hills San Pablo Bay Monterey County San Luis Obispo Santa Barbara County McKittrick Modelo Canyon Santa Monica Mts. Puente Hills Palos Verdes Hills Dana Point Kettleman Hills Maricopa Name Tice shale, Claremont shale Rodeo shale Monterey Monterey Monterey Monterey Modelo Modelo Puente Valmonte diatomite, Altamira shale Monterey McLure shale Maricopa These are the most widespread reactive rocks of the tate, consequently they are responsible for damage to aany concrete structures by alkali-aggregate reactions nd accompanying expansion. Extensive sections of high- pay in San Luis Obispo and Monterey Counties have re- hired repair or replacement because deleterious aggre- gate was used. In Los Angeles, the Sixth Street bridge crossing the Los kngeles River shows considerable damage. Portions of the 5epulveda Dam near Los Angeles are marked by a severe >opout condition. The deterioration of a concrete break- water at Santa Barbara is due at least in part to reactivity if opaline shale in the aggregate. Andesite from Friant and Vicinity. Certain andesite >ebbles known to be reactive have been found in the gravel »f the San Joaquin River near Fresno. Although not aiown in place, they are presumably debris that come rom the Sierra Nevada Tertiary lava to the east. Al- hough according to standard acceptance tests the rock has food physical properties, yet its chemical nature places it n the deleterious class. Both service history and labora- ory tests indicate its reactivity. riGUKE 7. Popout around opaline shale fragment, Sixth Street Bridge, Los Angeles. Figure 8. Cracking of pier, Sixth Street Bridge, Los Angeles. Petrographically the rock closely resembles the Parker Dam andesite, one of the first recognized reactive rocks. The average hand specimen is gray to purplish or brown and has small sparse phenocrysts of plagioclase, pyroxene, and hornblende. In thin-section it exhibits massive to distinctly fluidal hyalopilitic texture. Phenocrysts of fresh andesine ranging from 0.1 to 2 millimeters make up 10 percent of the rock. Phenocrysts of hypersthene and horn- blende are less abundant. Much of the latter mineral has been replaced by magnetite ; and in some sections the horn- blende is basaltic. The groundmass comprises 80 to 85 per- cent of the rock and is composed of small, thin laths of andesine and glass of intermediate composition. Feldspar is generally in excess over glass. Although information concerning the use of this rock in concrete is limited to aggregate from deposits along the San Joaquin River, nearly identical andesite is known in several areas along the east side of the Central Valley. Concrete in portions of the State Highway in Fresno has been damaged by severe pattern cracking. Certain sections made with low alkali cement show no cracking although the same andesitic aggregate was used. A few of the first blocks poured in Friant Dam showed minor effects of alkali-aggregate reaction. The later blocks using low alkali cement lack such features although both contain andesitic aggregate. Parker Dam Aggregate. Portions of Parker Dam, which lies on the southeast border of California exhibit marked effects of reactivity. Expansion of more than 0.1 percent has been measured. Nearby Gene Wash and Copper Basin Dams also show serious distress. The crown of the latter was deflected upstream 5 inches in 9 years owing to expansion. No evidence of reaction has been seen in certain related concrete structures in which low alkali cement was used. Although the aggregate was obtained from terrace de- posits on the Arizona side of the Colorado River, similar deposits and possibly some of the source rocks may be on the California side. Special Report 27 Figure 9. Pattern cracking in floor, Law Building, University of Southern California, Los Angeles. Petrographic studies revealed such reactive rocks as rhyolitic, dacitic, andesitic and trachytic lava and tuff, chalcedonic chert and chalcedonic limestone in the Parker Dam aggregate. Although a considerable variety of vol- canic rocks is present, the commonest types are hemi- crystalline, many with hyalopilitic groundmass. In such rocks the vitreous portion has undergone devitrification and the plagioclase microlites show incipient albitization. Ferromagnesian granules make up the remainder of the groundmass. Phenocrysts are oligoclase or andesine and hornblende. The oligoclase may be partially albitized and the andesine and hornblende are frequently bordered by magnetite. Reaction with the cement has produced clari- fied rims around most of the volcanic particles. Franciscan Chert. Chert of the Franciscan formation is widespread in the northern and central Coast Ranges, hence it is a prominent constituent in most gravel deposits in these areas. Although such gravel, when used as aggre- gate, has a satisfactory service history, laboratory tests have shown them to be reactive, probably because of their content of chalcedony and possibly opal. In thin-section the Franciscan chert is seen to consist of varying propor- tions of granular quartz and fibrous chalcedony. Some specimens are composed largely of an indeterminate, iso- tropic or nearly isotropic material having a refractive index of about 1.535. Its specific gravity, hardness, and index of refraction indicate that it is probably crypto- crystalline chalcedony rather than opal. A small but vari- able part of most of the chert is made up of radiolarian remains. In thin-section, areas showing an outline of these organisms are seen to be generally clear and composed of granular quartz. Disseminated fine ferruginous material is abundant in red phases of the chert. Anorthosite from the Western San Gabriel Mountains. Loughlin 7 has described the disintegration of cast stone 7 Loughlin, G. F., An interesting case of a dangerous aggregate : Amer- ican Concrete Inst. Proc, vol. 19, p. 142, 1923. ^ "> ) Figure 10. Cracking in steps, Law Building, University of Southern California, Los Angeles. and stucco using an aggregate of altered anorthosite from the western San Gabriel Mountains. It has been suggested that this disintegration may be due to the reactivity of the zeolite laumontite, one of the alteration products of the andesine in anorthosite. However, laumontite is not known to expand in mortar bars using high alkali cement. Pos- sibly deterioration resulted in part from base exchange processes involving the zeolite. A mixture of 5 parts altered anorthosite and one part calcium carbonate was proven "innocuous" in tests performed by Irving Sherman for K. V. Vail. 8 8 Written communication from K. V. Vail to California Division ol Mines, June 12, 1950. Figure 11. Graph showing percent expansion (vertical) plottet against percentage of reactive aggregate replacing inert aggregate Curve I — opal from Roseville, cement alkali 1.14 percent, age i months (from Stanton, 1942) ; Curve II — opaline chert, Montere; County, cement alkali 1.14 percent, age 8 months (from Stanton 1942) ; Curve III — Friant andesite, high alkali cement, age 6 month: (from McConnell et al. 1950) ; Curve IV — opal with low alkal cement, (from McConnell et al. 1950). Alkali-aggregate Reaction APPROXIMATE AREAS OF POTENTIALLY REACTIVE ROCKS IN CALIFORNIA VOLCANICS OPALINE SHALE Fioi;bk 12. Map of California showing approximate distribution of potentially reactive rocks. 10 Special Report 27 PREVENTION OF REACTION Testing of Material. As there are still some unknown and unpredictable factors it might be argued that the only indisputable, conclusive test is that of actual use. How- ever, information on the service history is available for relatively few aggregates; an aggregate must have been in use for several years as reaction is not always apparent during the first few years. Furthermore, the exact condi- tions of mixing, placing, and most important, the alkali content of the cement, are rarely known. The making of mortar bars constitutes one of the most informative laboratory tests. In this process cements of known composition are mixed with various proportions of aggregate and cast in 1- x 1- x 10-inch (or larger) bars. The bars are cured under conditions of controlled temperature and humidity. Their lengths are measured periodically for one or two years at least. Thus the slowness is a great objection to this test, as is the requirement of rather ex- tensive laboratory facilities. A test which was devised to determine reactivity of aggregate in a short time was described by Mielenz et al. 9 In brief, the material to be tested is digested in sodium hydroxide and filtered. The filtrate is analyzed for dis- solved silica and titrated against a standard acid. The amount of silica and the reduction in alkalinity are a measure of the reactivity. This test, too, requires consider- able laboratory work and equipment. As knowledge of the reactivity of minerals and rocks increases it is becoming possible to evaluate an aggregate by the more rapid method of petrographic study. Although the bulk of the petrographic classifications of constituents of a natural aggregate can be made with a hand lens, the most important determinations require a petrographic microscope. Examples are the distinction of crystalline, hemicrystalline, and glassy groundmasses ; of basalt and andesite ; quartz, chalcedony and opal. Chemical Control. Where the use of reactive aggregate is unavoidable, two procedures have been used to reduce or prevent deterioration by cement-aggregate reaction. The simplest method is that of limiting the alkali con- tent of the cement. As previously stated expansion is generally not serious if the total cement alkalies do not exceed 0.6 percent. The second preventive method consists of adding a cor- rective. Many, but not all, pozzolans are correctives. Pozzolans are defined by Mielenz et al. 10 as "natural or artificial siliceous and aluminous substances which are not cementicious themselves but which react with lime in the presence of water at atmospheric temperatures to produce cementitious compounds. ' ' Pozzolans when used as a partial replacement of the Portland cement aid in reducing reaction simply by re- ducing the amount of cement and its contained alkalies. However, to be a satisfactory corrective, a pozzolan must react chemically with or absorb alkalies released by hy- drating cement. The alkalies are thus made unavailable for reaction with the aggregate. The cement-pozzolan re- • Mielenz, R. C, Greene, K. T., and Benton, E. J., Chemical test for reactivity of aggregates with cement alkalies ; Chemical processes in cement-aggregate reaction : American Concrete Inst. Jour., vol. 3, 1948. 10 Mielenz, R. C, Greene, K. T., and Schieltz, N. C, Natural pozzolans for concrete : Econ. Geol., vol. 46, no. 3, 1951. action presumably takes place mostly before the concrete becomes rigid and, although alkali silica gel may form, it is so disseminated, because of the fine size of the particles, that osmotic pressures do not accumulate. Although there are certain advantages in addition to the corrective action of such pozzolans, some disadvantages may develop. They are : increase in water requirement, decrease in compres- sive strength, decrease in freezing and thawing durability, and increased drying shrinkage. The following pozzolans adequately control alkali-ag- gregate reaction : certain opals and rocks high in opaline silica; some silicic volcanic glasses; kaolinite (calcined to 1000-1600°F) ; some diatomaceous earth ; certain artificial glasses ; and montmorillonite-type clays such as beidellite (calcined to 1000 to 16O0°F) in which calcium is the ex- changeable cation. In California the principal pozzolans capable of con- trolling cement-aggregate reaction are opaline shales and porcelainities of the Monterey or equivalent Miocene for- mations. They are composed of varying proportion of opal and montmorillonite-type clay with insignificant quanti- ties of quartz, calcite, feldspar and other common min erals. The quality of this material is considerably improvec by calcining to 1400 °F. Structures in which this pozzolan has been used are : Davis Dam, San Jacinto Tunne Golden Gate and San Francisco Bay Bridges, Big Cree Dam No. 7, and some sections of highway paving. In addition to opaline shales, numerous other natura pozzolans have been used in California. However, none an completely satisfactory correctives ; in fact some aid rathei than suppress reaction. They include pumicite fron Friant used in the Friant Dam, and Pardee Dam ; altered rhyolite tuff from the Rosamond formation used in th( Los Angeles aqueduct; and argillaceous silt from Sar Francisco Bay used in Bonneville Dam. Mortar bars in which a mixture of altered anorthosit* and calcite was substituted for high alkali cement gave th< following results in one year : 15 percent substitution, ex pansion reduced 30 percent of expansion of control bar ; 2E percent, 57 percent reduction ; 40 percent, 72 percent re duction. 11 REFERENCES Bramlette, M. N., The Monterey formation of California and th origin of its siliceous rocks : U. S. Geol. Survey Prof. Paper 212 1946. Hanna, W. C, Unfavorable chemical reaction of aggregates in con Crete and a suggested corrective : American Soc. for Testing Ma terials, vol. 47, pp. 1-24, 1947. Loughlin, G. F., An interesting case of a dangerous aggregate American Concrete Inst. Proc, vol. 19, p. 142, 1923. McConnell, D., Mielenz, R. C, Holland, W. T., and Greene, K. Petrology of concrete affected by cement aggregate reaction : Geo! Soc. America Memoir, Berkey volume, pp. 234-235, 1950. Mielenz, R. C, Petrographic examination of concrete aggregates Geol. Soc. America Bull., vol. 57, p. 312, 1946. Mielenz, R. C, Greene, K. T., and Benton, E. J., Chemical test fo reactivity of aggregates with cement alkalies ; Chemical process in cement-aggregate reaction : American Concrete Inst. Jour vol. 3, 1948. Mielenz, R. C, Greene, K. T., and Schieltz, N. C, Natural pozzolan for concrete : Econ. Geology, vol. 46, no. 3, 1951. Stanton, T. E., Porter, O. J., Meder, L. C, and Nicol, Allen, Cal fornia experience with the expansion of concrete through reactio between cement and aggregate : American Concrete Inst. Jour vol. 13, no. 3, 1942. u Vail, K. V., Personal communication, 1952. O trinted in California state printing office 69028 9-52 2M