Download as pdf
Download as pdf
You are on page 1of 8
Ne NS Title no. 88-M9 RNA TECHNICAL PAPER Sulfate Attack on Concrete — Research Needs by Menashi D. Cohen and Bryant Mather ‘The mechanism of sulfate attack on portland cement concrete is not well understood. This has limited the confidence that can be placed in and hence the reliability of existing standard tests and models t0 predict performance and service life of concrete subjected to sulfate ‘attack, A systematic research effort on sulfate attack is necessary t0 establish five criteria. This would lead both to better understanding of the sulfate atiack mechanism and to developing more reliable standard tests and better models to predict performance and service We. The objective ofthis paper isto present and discuss these five ei teria. In addition, an approach Jor asessing the influence of pozz0- lans, or other cementing materials, on concrete performance and service ife in sulfate environment is presented. Keywords: accelerated tests; cement pastes; concretes; duraily etringtes {allre riteria; gypsum; magnesium sulfate; porzolans sodium sulfate; sulfate tae; transition zone, Sulfate attack on concrete is a complex process. Nu- merous important structures have deteriorated because the concrete lacked the necessary resistance to sulfate attack. Many of these failures are attributed not to im- properly selected construction practice, but to improp- erly selected concrete materials and mixture propor- tions due to a lack of complete understanding of the nature and consequences of sulfate attack and to an in- ability to accurately predict performance and service life of concrete. ‘The U.S. Nuclear Regulatory Commission (NRC), in conjunction with the National Institute of Standards and Technology (NIST), has invested significant ef- forts to develop strategies for storing low-level radio- active waste in underground concrete vaults (Clifton and Knab 1989). These vaults are expected to have service lives of at least 500 years. Independent research is being conducted on this topic at the National Re- search Council of Canada, NRC (Feldman and Beau- doin 1987). Durability of concrete in a sulfate environ- ment is one of the most important factors influencing performance and service life that is currently being studied by NIST, NRC, Purdue University, U.S. Army Corps of Engineers, and several other public agencies and research universities. More reliable standard tests and models are needed to predict performance and service life of several hun- 62 dred to several thousand years for concrete exposed to sulfate ions. It has been suggested (Clifton and Knab 1989) that some of the existing models, such as the one developed by the U.S. Bureau of Reclamation (Kalou- sek et al. 1972) which is based on expansion, are not reliable to predict long-term performance. The reliabil- ity of another method developed by the Building Re- search Establishment in England (Harrison and Teychenne 1981), which is based on depth of visible deterioration, has not been well-established either (Atkinson et al. 1985; Clifton and Knab 1989). ‘There is a pressing need expressed by worldwide public concern to develop concrete containment vaults for storing radioactive and hazardous wastes that will be durable for centuries or millenia. The current need for reliably predicting long-term durability perform- ance and service life that are based on short-term accel- erated tests is more than ever justified. Many current methods are more of art than science and it is difficult to convince the public and public interest groups of the safety of concrete containment vaults designed on the basis of current state-of-the-art information on sulfate attack. In short, the current state of the art is inade- quate. ‘A systematic research effort on sulfate attack is nee- essary to establish five major criteria and to incorpo- rate findings in appropriate standards and predictive models. This would lead to a better understanding of the process of sulfate attack and confidence in con- structed structures. These five criteria are: 1, Establishing methods for accelerating sulfate at- tack. 2. Establishing appropriate measured properties or indicators. 3. Establishing failure criteria. 4, Establishing a standard methodology for system- atic data analysis and development of hypotheses. “ACT Materils Journal, V-88, No, 1, January-February 1991, Aeceived Sep. 29, 1989, and reviewed under Institute publication polices, Copyright © Yoo1, American Concrete Instat. Al ight reserve including ‘fetmeking of copie unless permission i obaied from the copyright propt ‘dors, Pertinent dcusion wil be pubshed in the November December 1991 ‘CT Moteriats Journal irrecesved by AUB. 1, 1931 ACI Materials Journal | January-February 1991 | ‘ACI member Menashi D. Cohen is Associate Profesor of Civil Engineering at Purdue University. He is Chairman of ACI Committees 231, Properties of Concrete at Early Ages, and ACI 208-1, Creep and Shrinkage: and Co-har ‘man of ACI Committee 125-5, Lunar Concrete. ‘ACT Honorary Member and past present Bryant Mather is Che, Strutures Laboratory, US. Army Engineer Waterways Experiment Station, Vicksburg ‘Misssipp, and hasbeen with that federal agency since 191 5. Establishing relationship between paste and con- crete durability. Based on these criteria, an approach to assessing the influence of pozzolans, or other cementing materials, on concrete performance and service life in sulfate en- vironment will also be presented. CRITERION 1: ESTABLISHING METHODS FOR ACCELERATING SULFATE ATTACK Most laboratory methods aimed at studying the sul- fate attack mechanism are based on accelerated tests. In most practical situations, sulfate solution concentra- tion in soil and groundwater are low (< 1000 ppm SO,). It is impractical to use such low-level concentrations in the laboratory because deterioration will not occur within a reasonable period of time. Accelerating sulfate attack is a necessity. Thus, development of models to predict long-term performance and service life must be based on short-term data obtained under accelerated testing conditions. Numerous methods have been developed to acceler- ate sulfate attack in concrete. ASTM E 632-82, ‘‘Stan- dard Practice for Developing Accelerated Tests to Aid Prediction of the Service Life of Building Compo- nents,”” discusses the significance and importance of accelerated tests. One of the more popular methods for accelerating sulfate attack is to increase the sulfate concentration to a level much higher than that to be encountered in the field.* Although a popular way of accelerating attack, this method has been criticized. The main concern is the change in attack mechanism as a result of increased sulfate concentration. This, there- fore, suggests that laboratory results obtained under accelerated tests by increasing sulfate concentration cannot be used to predict actual behavior in the field. Biczok (1967) has conducted an elaborate review of sulfate attack mechanism. He states that the intensity of sulfate attack corrosion increases with increase in sulfate concentration. At the same time, the type or mechanism of corrosion undergoes a change. "The assumption that sulfate concentrations inthe id are low aay be ge erally correct but fy no means universally so, Much sulfate etack involves Calcium sulfate of wiih a saturated solution only contains about 0. 100m ‘S compared with sodium or magnesium sulfates which may conai 101070 4/100 mi. The sulfate concentration required to do har to nonsulae es fant eoncete s 150 poe of 30, (aoe: to convert SO, concentration 10 30, ncsnration, multiply 80, by 0.833, “severe attack only takes place When {he SO, conatration exces 1500 ppm (ACI 201). Water with a qute low {pesetaon of aula ton ca poi loons very Mah contain Pi Splashes on or drops gn a urace rapeates, Bale Ups quany of Bre ‘ipiated sits that then doves or arty doles to form'a saturated or ‘arly saturated solution. Such conditions an ex on wharves, the tops of ‘dams where seepage rom an abutment ows onto the surface, of Ob the exe= ‘or surface of doling tower, ACI Materials Journal / January-February 1991 Regarding sodium sulfate attack, the corrosive mechanism is divided into two parts, depending on so- dium sulfate concentration. At low concentrations (less than 0.12 percent, i.e., SO, concentration of less than 830 mg/L — this limit changes with change in C,A content of portland cement) he suggests that the mech- anism causing damage to concrete is by ettringite for- mation. When SO, concentration attains a high value, the mechanism is dominated by gypsum formation (acid attack — generally all sulfate solutions are some- what acidic). The transition between sulfate attack and gypsum attack is not abrupt, and there is a range where the two overlap. Expansion, and subsequent microcracking, is gener- ally attributed to ettringite formation. However, it has been suggested that not all types of ettringite are ex- pansive (Mehta 1983), and in order for it to induce ex- pansion, it has to have certain morphological charac- teristics. Unlike ettringite, the specific consequence of gypsum formation is not well established. There is no clear documentation that gypsum formation leads to expan- sion, Softening has also been attributed to eypsum for- mation. If gypsum is expansive, the source of expan- sion should be due to volume increase upon transfor- mation of solid calcium hydroxide to gypsum by reaction with sulfate ions. However, this explanation is perhaps valid only when gypsum crystals form in situ or on surfaces of calcium hydroxide crystals; i.e., by to- pochemical reaction mechanism (Hansen 1963). This doesn’t apply to concrete, since the mode of formation of gypsum is by precipitation in the solution away for calcium hydroxide particles; i.e. by through-solution mechanism. Hansen's suggestion that topochemical formation of crystals (gypsum) causes expansion by crystallization pressure was based on an almost com- pletely impossible hypothetical mechanism, “‘a crystal resting on the floor of a container with another crystal resting on it under an applied load in a solution satu- rated with respect to the bottom crystal”” (Hansen 1963, p. 934). Hansen continued this argument and con- cluded that, if, under this scenario a force is exerted, it isn’t crystal growth, it is hydraulic pressure (Hansen 1963, p. 935). At this stage, it is suggested that precip- itation and crystallization of gypsum from solution, which is what happens in concrete, is not expansive. Regarding magnesium sulfate attack, the mechanism is divided into three parts. At the lower concentration level of about 3200 mg/L SO, or less (corresponding to less than 0.48 percent magnesium sulfate), the attack is dominated by ettringite formation. The second mecha- nism occurs at between 0.48 and 0.75 percent magne- sium sulfate concentration, and it is controlled by a combined ettringite-gypsum formation. The third and most advanced stage is controlled by magnesium at- tack, and this occurs when magnesium sulfate concen- tration exceeds 0.75 percent. Magnesium attack occurs when there is reaction be- tween magnesium sulfate solution and hydrated cal- cium silicates to form gypsum, magnesium hydroxide 63 (which is relatively insoluble as compared with calcium hydroxide), and silica gel (B. Mather 1966). Subse- quently, magnesium hydroxide can react with the silica gel to form water and hydrated magnesium silicates (B. Mather 1966), which is noncementitious (Cole and Hueber 1957; B. Mather 1966; Biczok 1967; Lea 1971; ‘Cohen and Bentur 1988). The process of decomposi tion of cementitious hydrated calcium silicates to a non- cementitious magnesium silicate hydrate can be accom- plished by magnesium sulfate and no other form of sulfate. Ettringite cannot form under high concentration of magnesium sulfate because the hydroxide-ion concen- tration is reduced by the resulting gelatinous magne- sium hydroxide to below the value necessary for ettrin- gite formation (Biczok 1967). Interestingly, yet surprisingly, the effects of sulfate concentration on the corrosive mechanism have not been studied much by researchers and practitioners al- though it is critical. Most of this work was done by V.V. Kind, and his publications were reviewed and dis- cussed in detail by Biczok (1967). Other accelerating methods of sulfate attack have been developed. One must, however, recognize the possibility that what may lead to accelerated attack may have a profound effect on the attack mechanism itself. ‘This may impede the practicality of correlating labora- tory data to field performance. Yet one can argue that such information is needed to better analyze and pre- dict field performance of concrete exposed to sulfate solutions and to see where the breakoff point is: ‘The most important rapid test methods are as fol- lows (Biczok 1967): 1. Increasing reaction surface (j.e., small specimens/ large surface areas). 2. Increasing concentration of solution. 3. Increasing crystal pressure (i.e., continuous wet- ting-drying cycles). 4. Raising the temperature of solution. Items 1 and 2, increasing reactive surface and con- centration, are the most practical and widely used methods. Item 2 has been used mainly for reason of accelerating attack but not so much as to study the na~ ture and mechanism of attack as a function of sulfate solution concentration, Item 3, increasing crystal pressure, has been used but has not received wide recognition since other disruptive mechanisms may be involved that would affect inter- pretation of results. Item 4, raising the temperature, may have some difficulties associated with the tech- nique. Increased temperature would have different ef- fects depending on mixture composition and specimen size. Higher temperature could also lead to significant modifications in the chemistry and microstructure of the particular system being studied. ‘A method of accelerated sulfate attack developed by Mehta (1975) received considerable attention. The method tests the resistance of cement paste to sodium sulfate solution held at a constant pH of 6.2 by repeat- edly adding acid to the solution. Although the method 64 accelerates attack, it is possible that sulfate attack mechanism becomes limited to gypsum action. How- believed that this test can provide valuable information in that one can analyze effects of gypsum formation and distinguish it from ettringite formation. Other methods used to accelerate sulfate attack in- clude using high w/c mortar and addition of stet gyp- sum to the mixture. The first method, high w/c, makes the mortar more permeable and thus more prone to sulfate attack at an earlier age. However, it generally fails to provide reproducible results (Wolochow 1952; K. Mather 1978). On this basis, K. Mather used a lower level w/c (by mass: 0.485 for non-air-entraining port- land cement, and 0.460 for air-entraining portland ce- ment). This makes concrete less permeable and the at- tack less rapid, but the results are more reproducible. Expansion measurement is conducted for one year to allow for adequate penetration of sulfate ions to the interior of concrete. This recommendation now forms the basis of ASTM C 1012-89, which is the standard test method for determining length change of hydrau- lic-cement mortars exposed to a sulfate solution. The test is suitable for evaluating sulfate resistance of blended cements as well as portland cements. ‘The latter, addition of stet gypsum, appears to be an effective way for accelerating sulfate attack. This method requires adding gypsum to the mixture in an amount sufficient to provide a total of 7 percent SO; by mass of portland cement when the SO, already present in the cement is also considered. This method forms the basis of ASTM C 452-89, which is the standard test method for determining the potential expansion of portland cement mortars exposed to sulfate. Although results are reproducible and can be obtained within 14 days, this test is not suitable for blended cements or blends of portland cement with slag or pozzolan. ASTM C 1012 is suitable for blends because it provides for substantial hydration. CRITERION 2: ESTABLISHING APPROPRIATE MEASURED PROPERTIES OR “INDICATORS” What property, or “‘indicator,”” best represents the nature and mechanism of sulfate attack? What are the failure criteria for sulfate resistance? These are impor- tant questions raised repeatedly in the literature (Thor- valdsen et al. 1932; Hansen 1966; Cement Research In- stitute of India 1981; Bentur et al., 1987; Cohen and Bentur 1988). According to reports published by the Cement Research Institute of India (1981) and Wong and Poole (1988), different parameters or indicators ive different information. The indicator may be a de- crease in a measured engineering property of concrete (e.g., a change in length, mass, compressive strength, flexural strength, or modulus of elasticity) which relate to the functional behavior of structures. Measurements ‘of compressive and flexural strengths are adequate in this case, but these measurements may not reflect the arly stages of the softening-spalling type of sulfate at- tack. Also, under some conditions, sulfate attack may produce hydration products that would fill the pores in ACI Materials Journal / January-February 1991 concrete and actually lead to an increase in strength (Kalousek and Benton 1970). Under this condition, change in strength and the actual destructive nature of sulfate attack may not be correlated. Although each in- dicator may have some advantages, none is ideal for all considerations. Hansen (1966), in reviewing the sulfate durability data of Polivka and Brown (1959) for concrete, indi- cated that loss in mass appears to be the best indicator of the degree of deterioration — as opposed to changes in length and compressive strength, visual observa- tions, and photographic records. Data in the literature, in general, point to the fact that cement paste (and concrete) respond differently to the different nature of sulfate attack. For example, they respond to ettringite formation by expanding. A drop in strength, however, does not always correspond to et- tringite formation because the paste can often with- stand relatively large local tensile strains (approxi- mately 150 microstrains) before cracking. Thus, the paste may well be at the advanced stage of sulfate at- tack without showing any sign of strength drop. Another example is gypsum formation. This action leads to softening and mushiness of concrete. When the concrete is wet, the gypsum can be removed easily. Thus, loss of mass can be a measure of sulfate attack due to gypsum formation. Expansion is not expected to accompany gypsum formation, but Mehta’s work (1983) suggests otherwise. The matter is therefore com- plicated due to the fact that it is not certain whether strength drop, if it accompanies expansion, signifies gypsum or ettringite formation. Mehta’s method (1975) of accelerating sulfate attack which limits attack to gypsum formation, may shed some light to this ques- ion. As discussed earlier, it is yet not clear whether gypsum causes expansion, and this uncertainty needs to be investigated. CRITERION 3: ESTABLISHING FAILURE CRITERIA There also is no consensus as to what values of loss of mass, expansion, and strength drop best represent failure. To complicate things, the values are different for pastes, mortars, and concretes. Based on previous research (Cohen and Bentur 1988) the following dividing lines were suggested for pastes. The values were selected on the basis of experience, vi- sual observation, and literature data: * Loss of mass: 5 percent (beam), 2.5 percent (cube) * Expansion: 0.4 percent © Strength drop: 25 percent Values greater than these indicated failure and those less than these indicated satisfactory performances. The value of loss in mass for a beam (5 percent) was twice that for a cube (2.5 percent), simply because of the higher surface-to-volume ratio (approximately 2:1). ACI Materials Journal | January-February 1991 These values correlated well with the appearance and integrity of the specimens. The classification or failure criterion value used for expansion was based on works of McMiltan et al. (1949) on mortars as quoted by Ne- ville (1973), K. Mather’s on mortar (1980), and the U.S. Bureau of Reclamation’s on concrete. McMillan used a failure criterion value of 0.1 percent for expan- sion at 28 days after exposure for mortars, a similar value to K. Mather’s. The U.S. Bureau of Reclamation used a value of 0.2 percent at 28 days for concrete. A value of 0.4 percent was used in the recent work (Co- hen and Bentur 1988) taking into account that expan- sion in paste is considerably greater than in mortars and coneretes. K. Mather (1982) and Patzias (1987) have suggested limits based on length increase of mortar bars using ASTM C 1012. Ouyang (1989) has established, using a damage model, that the values suggested agree well with those obtained using his model. The strength classification was after Mehta (1986). Reduction in strength for pastes of less than 25 percent indicated satisfactory performances, and those above, failure (ie., poor or very poor performance according to Mehta). Perhaps it is necessary not only to look at changes in strength, expansion, and mass but also at other prop- ‘erties. Recent unpublished experimental works on ex- pansive cements by one of the authors (M. D. Cohen), suggest that changes in bulk density and total porosity can indicate the extent of damage due to ettringite for- mation and expansion. The development of expansion due to the use of expansive cement is a form of con- trolled sulfate attack. Changes in bulk density and to- tal porosity also correlate well with changes in strength and dynamic modulus values measured during expan- sion. Measurements of changes in porosity, as deter- mined by the amount of evaporable water at 105 C of saturated-surface-dry specimens, can also reveal the ex- tent of damage. To establish failure criteria for sulfate resistance one must therefore measure a wide spectrum of properties. These must be monitored periodically to examine and assess durability. The properties include: 1. Visual appearance of test specimens. 2. Change in mass, length, and volume, bulk den- sity, total porosity. 3. Change in compressive strength, flexural strength, and elastic modulus. ‘A special visual rating system (Item 1) should be de- veloped to identify progress, type, and magnitude of sulfate deterioration. Although such a method is sub- jective, the mean value of ratings obtained by different observers for a certain experiment can give results in agreement with strength and modulus of elasticity val- ues (Biczok 1967). ‘The physical and mechanical properties, Items 2 and 3, respectively, are the most important factors that should be investigated. These include compressive and flexural strength, length change, and static and dy- namic Young’s modulus of elasticity. 65 nay expansion Fig. 1 — Percentage of change in length (% AL/L,) versus duration of exposure (tag) Fig. 2 — Percentage of change in mass (% Am/tn,) versus duration of exposure (ta) deat ei oo Fig. 3 — Percentage of change in bulk density (% p/p) versus duration of exposure (tay) men NO tee m o Fig. 4 — Percentage of change in total porosity (%% A/n) versus duration of exposure (ta) ete ° ead comeina sen cop Fig. 5 — Percentage of change in compressive strength (% Af! /f1) versus duration of exposure (tax) 66 sgn oop Fig. 6 — Percentage of change in flexural strength (% Af /f.) versus duration of exposure (ay) Fig. 7 — Percentage of change in elastic modulus (% AE/E,) versus duration of exposure (tap) CRITERION 4: ESTABLISHING A STANDARD ‘METHODOLOGY FOR SYSTEMATIC DATA ‘ANALYSIS AND DEVELOPMENT OF HYPOTHESIS a. Obtain quantified relationships Basically, the following quantified relationships should be obtained easily for concrete specimen after exposure to the sulfate solutions: 1. Percentage of change in length (% AL/L,) versus duration of exposure (tag). 2. Percentage of change in mass (% Am/m,) versus duration of exposure (t2,). 3. Percentage of change in bulk density (% p/p.) versus duration of exposure (ten) 4, Percentage of change in total porosity (% An/n,) versus duration of exposure (fay). 5. Percentage of change in compressive strength (% Aff’ e) versus duration of exposure (t..)- 6. Percentage of change in flexural strength (% Afr/fre) versus duration of exposure (Fa). 7. Percentage of change in elastic modulus (% AE/E,) versus duration of exposure (te,). All changes are with respect to original values as in- dicated in the denominators. b. Expected relationships For each of the relationships just listed, the expected behavioral patterns are as follows in Fig. 1 through 7. c. Identification of mechanism of sulfate attack and development of hypothesis While physical and mechanical studies will reveal only behavioral patterns, such as normal and delayed ‘ACI Materials Journal / January-February 1991 behaviors, the chemical and microstructural studies will assist in identifying the mechanisms of sulfate attack and specifically the events that trigger the delayed ac- tivities (for example delayed expansion, strength drop, etc.) that are common in sulfate attack. These studies will also help develop new hypotheses or confirm ear- ier ones. ‘The different types of relationships obtained in the laboratory and in the field are described in the preced- ing figures in part (b) which are based on physical and mechanical tests. Based on data obtained one must se- lect appropriate specimens for further testing. Identical specimens must be prepared and tested in-depth for their chemical and microstructural features. The chem- ical study should involve the use of XRD and EDXA for phase and elemental analyses, TGA/DTA for cal- cium hydroxide and other hydrates. Microstructural studies will involve the uses of SEM, MIP, image anal- ysis, and backscatter electron analysis methods. Delayed expansion and delayed strength and modu- lus decreases have been subjects of interest and contro- versy among scientists and practitioners (Kerdegari 1978; Cohen 1981; Cohen and Richards 1982; Mehta 1983), and it is not clear which action causes the trig- gering of delayed expansion, which almost always oc- curs simultaneously with delayed strength and modulus drops. It is also uncertain which of these (delayed ex- pansion or delayed strength and modulus drops) is the ‘cause or the result. With the data obtained, one could attempt to find out the sequence and consequences of events, such as the timing of ettringite formation, gyp- sum formation, and consequently their effects on length, mass, density, porosity, compressive and flex- ural strengths, and modulus. The effects of a pozzolan cn these events can be clarified. Researchers have yet to address these issues in such a comprehensive and sys- tematic research program as described previously. Chemical and microstructural analyses must also be conducted periodically on various layers of the speci ‘mens, starting at the exposed surface. These would give the rate, intensity, and depth of penetration of sulfate attack. The morphologies and quantities of gypsum and ettringite developed at each level should be determined. Different methods have been developed and used to determine the quantities of gypsum (Cohen and Bentur 1988) and ettringite (Uchikawa and Uchida 1974; Co- hen 1981; Cohen and Richards 1982; Odler and Abdul- Maula 1984) formed during sulfate attack. For the de- termination of quantity of ettringite the methods of Uchikawa and Uchida (1974), Odler and Abdul-Maula (1984), and recently by Lobo and Cohen* can be con- ducted and compared with standards to determine which of the three methods is more appropriate. These quantitative results can be tied in with SEM observations to assess the relationship between ettrin- ite morphology and expansion, which has long been a controversial issue (Mehta 1973, 1983; Cohen, 1983, “Lobo, C. and Cohen, M. D., unpublisid work on sulfate attack and ex: pankve cements ACI Materials Journal / January-February 1991 1984; B. Mather, 1984, CRITERION 5: ESTABLISHING RELATIONSHIP BETWEEN PASTE AND CONCRETE DURABILITY It has been shown (Bentur et al. 1987; Bentur and Cohen 1987) that neat cement paste and concrete are two different things that can’t be compared due to the transition zone effect which is present in concrete but absent in paste. Researchers have been arguing that it is not appropriate to use paste as a representative model to explain concrete behavior. Earlier Winslow and Liu (1990) and more recently Winslow and Cohen" at Pur- due University have shown that the microstructures of ure paste and paste within the concrete are two differ- ent things because the effects of transition zones extend far beyond the aggregate surface and individual effects of surfaces in concrete overlap, and there is little or no paste in concrete that isn’t affected. This is further confirmed in the earlier experimental investigation of Diamond et al. (1982). Bentur and Cohen (1987), how ever, have shown that paste located 50 um or more away from the aggregate surface are not affected by the aggregate. The transition zone appears to be limited to 50 wm from the aggregate surface. Thus, with calcu- lated modifications, it is possible to use paste as a model to explain concrete behavior. This is a crucial degree.of freedom since it will allow one to tie in re- sults with the earlier literature. Use of paste as a con- crete model was taken for granted in the earlier litera- ture dealing with modeling of creep, shrinkage, strength, and modulus behaviors of cement pastes and coneretes. It should be added that in mortars and concretes containing silica fume, SEM study indicated absence of a transition zone (Bentur and Cohen 1987) and that there appeared no microstructural difference between the transition zone and bulk pastes. This leads one to suggest that pure paste and pastes in mortars and con- cretes become more identical in their microstructural and engineering properties in the presence of silica fume. The influence of the transition zone in concrete on sulfate durability and performance can be studied by comparing paste and concrete data. A recent report (Bentur and Cohen 1987) revealed filling of the transi- tion zone due to silica fume addition. This filling proc- ess was attributed to the small sizes of silica fume par- ticles and their ability to pack the interface and reduce bleeding. The consequence of packing is the minimiz- ing or elimination of water-filled cavities at the inter- face. Interfacial filling and packing and consequently im- proved impermeability are expected to improve dura- bility. But recent work (Bentur and Cohen 1987; Co- hen and Bentur 1988) indicate that this expectation is not necessarily valid when magnesium sulfate attacks iow, D. Nand Cohen, M.D., unpublished work on poe structure of sea fume const 67 portland cement-silica fume concrete. This unexpected performance must be clarified. ASSESSING THE ROLE OF POZZOLANS ‘The great public awareness in the early 1980s of the deteriorated infrastructure in the United States has cre- ated an expectation that any new civil engineering ma- terial ought to be extremely durable and free from re- quirements for major maintenance and rehabilitation costs. Unfortunately, our understanding of fundamen- tal mechanism of physico-chemical actions of pozzo- lan, specifically silica fume, in portland-cement con- crete is inadequate. Research and field data on the effects of pozzolan on durability and service life are in- complete, or at best inconclusive. ‘The five research criteria described in this paper would enable one to characterize influence of pozz0- lan, or other cementing materials, on sulfate-attack du- rability. In so far as pozzolans are concerned, one can compare the durability in sulfate solution of CSH pro- duced by pozzolanic reaction and the CSH produced by portland-cement hydration. Pozzolanic CSH gel is sus- pected to behave differently due to its differences in atomic structure and arrangement (Buck and Burkes 1981; Regourd et al. 1983; Cheng-yi and Feldman 1985; Cohen and Bentur 1988). Basically, three steps are necessary to obtain infor- mation on sulfate attack mechanism in portland-poz- zolan cement concrete. These are: 1, Identify the accelerated test with cement pastes and concrete without pozzolan. Subsequently, define failure criteria and threshold concentration of sulfate for different w/c. 2. Examine effect of gypsum and ettingite according to the chosen failure criteria. This step would shed light ‘on the mechanism of sulfate attack. 3, Examine effects with pozzolan and compare with reference paste and concrete according to the failure criteria and the sulfate attack mechanism. Monitor changes in microstructure of cement with pozzolan in the presence of the sulfate solution. Clearly, the interpretation of data depends on defi- nition of “reference” paste and concrete. The method- ology of data interpretation, as indicated previously, is based on the reference being paste and concrete with- ‘out pozzolan. One must also interpret the data on the basis that the reference is paste and concrete with poz- zolan. The conclusions that are based on these two in- terpretations must be compared so that the significance of reference can be examined and established. SUMMARY AND CONCLUSIONS Research is necessary to establish five criteria as dis ‘cussed in this paper. This would lead both to better un- derstanding of sulfate attack mechanism and to devel- oping more reliable standard tests and models in order to predict better the performance and service life of concrete. Furthermore, the findings would help lay the groundstones for obtaining answers to numerous fun- 68 damental questions that have remained unanswered even after decades of research. Some of these questions that need to be specifically investigated relate to the following list: 1. Consequences of gypsum and ettringite forma- tion, 2. Correlation between morphology and amount of ettringite formed and expansion. 3. Causes, effects, and mechanics of delayed expan- sion, mass loss, density drop, pore formation, com- pressive strength drop, flexural strength drop, and modulus drop. 4, Failure criteria. 5. Performance classification. 6. Effect of sulfate solution concentration on the sulfate attack mechanism. 7, Role of concrete impermeability in durability. 8. Importance of the “transition zone’” on strength and durability. 9. Validity of using paste as a representative model for concrete. 10. Identification of an accelerated test that would not alter the mechanism of sulfate attack. 11. Identification of the threshold concentration of sulfates above which concrete deterioration is inevita- ble and below which there would be no deterioration. REFERENCES ‘Atkinson, Aj Goult, D. J.; and Hearne, J. Au, 1985, “An Assess- rent of the Long-Term Durability of Concrete in Radioactive Waste Repositories,” Proceedings, Materials Research Society Symposium ‘Basis for Nuclear Waste Management IX, V. 50, Stockholm, pp. 239- 246. Bentur, A., and Cohen, M. D., 1987, “Effect of Condensed Silica Fume on the Microstructure of the Interfacial Zone in Portland Ce- ‘ment Mortars,” Journal of the American Ceramic Society, V.10, No. 10, pp. 738-743. Bentur, A.; Goldman, A.; and Cohen, M. D., 1987, *'Contribu- tion of the Transition Zone to the Strength of High Quality Silica Fume Coneretes,” Proceedings, Materials Research Society Sympo- sium on Bonding in Cementitious Composites, V. 114, Boston, pp. 97-108. Biezok, Imre, 1967, Concrete Corrosion Concrete Protection, ‘Chemical Publishing Company, Inc., N.Y., $43 pp. Buck, A. D., and Burkes, J. P., 1981, Characterization and Reac- tivity of Sica Fume, Final Report, Misc. paper MP SL-81-13, U.S. ‘Army Engineer Waterways Experiment Station, Vicksburg, 10 pp. Cement Research Institute of India, 1981, Evaluation of Sulphate Resistance of Concrete, Report No. RB-20-1981, New Delhi, 28 pp. Clifton, James R., and Knab, Lawrence I., 1989, Service Life of Concrete, Report from U.S. Department of Commerce, National In- stitute of Standards and Technology, NISTIR 89-4086, p. 119. ‘Cohen, M, D., 1981, Micromechanics of Expansive Mechanism in Expansive Cement Coneretes, Ph.D. dissertation, Department of Civil Engineering, Stanford University, Stanford. Cohen, M. D., 1983, “Theories of Expansion in Sulfosluminate- type Expansive Cements: Schools of Thought,”” Cement and Con- crete Research, V. 13, NO. 6, pp. 809-818. ‘Cohen, M. D., 1984, A reply to a discussion by B. Mather of the paper, “Theories of expansion in sulfoaluminate-type expansive ce ‘ments: Schools of thought," Cement and Concrete Research, V. 14, No. 4, pp. 610-612. Cohen, M, D., and Bentur, A., 1988, “Durability of Portland Ce- ‘ment-Silica Fume Pastes in Magnesium Sulfate and Sodium Sulfate Solutions,” ACI Materials Journal, V. 85, No. 3, May-June 1988 pp. 148-157. ‘ACI Materials Journal / January-February 1991 Cohen, M. D., and Richards, C. W., 1982, “Effects of the Parti- cle Sizes of Expansive Clinker on Strength-Characteristcs of Type K Expansive Cements," Cement and Concrete Research, V. 12, No. 6, pp. 717-725 Cole, W. F., and Hueber, H. V., 1957, “Hydrated Magnesium Silicates and Aluminates Formed Synthetically and by the Action of the Sea Water on Concrete,” Silicates Industiels (Mons), V. 22, No. 2, pp. 75:85. Also, Nature (London), V. 1971, 1983, p. 354. Diamond, S., Mindess, S., and Lovell, J., “On the Spacing be- tween Aggregate Grain in Concrete and the Dimensions ofthe Aure- ‘ole de Transition,” Liasons Pates de Ciment Materieux Associes, ‘Communications, Colloque International, (RILEM), Toulouse, Nov. 17-19, 1982, pp. C. 42-C. 46. Feldman, R. F., and Beaudoin, J. J., 1987, Concrete for Low-level Radioactive Waste Repository — Five-Hundred Year Life Span, Na: tional Research Council Canada, Institute for Research in Construc 3m, Ottawa, 49 pp. Hansen, W. C., 1963, “Crystal Growth as a Source of Expansion in Portland-Cement Concrete,” Proceedings, American Society for ‘Testing and Materials, V. 63, pp. 932-945 Hansen, W. C., 1966, “Attack on Portland Cement Concrete by Alkali Soils and Waters — A Critical Review," Highway Research Record, 73, No. 113, pp 1-32. Harrison, W. H., and Teychenne, D. C., 1981, Sulfate Resistance of Buried Conerete, Second interim report on long term investigation at Northwick Park, Building Research Establishment, Department of Environment, Garston, Published by Her Majesty's Stationary Of- fice, 165 pp. Huang, Cheng-yi, and Feldman, R. F., 1985, “Influence of Silica, Fume on the Microstructural Development in Cement Mortar,” Ce- ‘ment and Concrete Research, V. 1S, No. 2, pp. 285-294 Kalousek, G. L. and Benton, E. J., 1970, “Mechanism of Seawa- ter Attack on Cement Paste,” ACI JOURNAL, Proceedings V. 67, No. 2, Feb. 1970, pp. 187-192. Kalousek, G. Lu Porter, L, Cand Benton, E. J., 1972, “Con- crete for Long-Term Service in Sulfate Environment,” Cement and Concrete Research, V.2, No. 1, pp. 19-89. Kerdegari, A., 1978, ‘Role of Gypsum in Portland Cement and Expansive Cements,” Ph.D. dissertation, Department of Civil E neering, Stanford University, Stanford, Lea, F. M., 1971, Chemistry of Cement and Conerete, 3rd Edi tion, Chemical Publishing Co., New York, 727 pp. Mather, B., 1966, “Effects of Seawater on Concrete, Highway Re. search Record, No. 113, pp. 33-42, Highway Research Board, NAS- NRC, Washington, D.C. Mather, B., 1984, Dicussion of paper, “Theories of Expansion in Sulfoaluminate-Type Expansive Cements: Schools of Thought.” by M. D. Cohen, Cement and Concrete Research, Vol. 14, No. 4, pp. 603-608. Mather, K., 1978, “Tests and Evaluation of Portland and Blended Cements for Resistance to Sulfate Attack,” Cement Standards, Evo- Iuion and Trends, American Society for Testing and Materials, STP 663, pp. 1486, Mather, K., 1980, “Factors Affecting Sulfate Resistance of Mor- tats,” Proceedings, ‘7th Interational Congress on the Chemistry of ‘Cement, V. IV, Paris, pp. 580-58. Mather, K., 1982, “Current Research in Sulfate Resistance at the ACI Materials Journal / January-February 1991 Waterways Experiment Station," Sulfate Resistance of Concrete, SP- 77, American Coficrete Institute, Detroit, pp. 63-74. MeMMillan, F. R.; Stanton, T. E.; Tyler, I L.: and Hansen, W. C., 1949, ““Long-Time Study of Cement Performance in Concrete, ‘Chapter $: Concrete Exposed to Sulfate Soils,” Special Publication, American Conerete Institute, Detroit, 64 pp. Mehta, P. K., 1973, “Mechanism of Expansion Associated with Ettringte Formation,” Cement and Concrete Research, V. 3, No. l, pp. 1-6. Mehta, P. K., 1975, “Evaluation of Sulfate Resisting Cements by New Test Method," ACI JOURNAL, Proceedings V. 72, No. 10, Oct 1975, pp. 573-575. Mehta, P. K., 1983, “Mechanism of Sulfate Attack on Portland ‘Cement Concrete — Another Look,’ Cement and Concrete Research, V.13, No. 3, pp, 401-406. Mehta, P. K., 1986, “Effect of Fly Ash Composition on Sulfate Resistance of Cement, ACI JOURNAL, Proceedings V. 83, No. 6, Nov, Dee. 1986, pp. 994-1000. Neville, A. M., 1973, Properties of Concrete, 2nd Edition, Pitman Publishing Co., Marabfield, 779 pp. Odler, I. and Abdul-Maula, S., 1984, “Possibilities of Quantita- tive Determination of the AFt- (Ettringite) and AFm- (Monosul- phate) Phases in Hydrated Cement Paste,” Cement and Concrete Research, V. 14, No. 1, pp. 133-141. Ouyang, C., 1989, “Damage Model for Sulfate Attack of Cement Mortars,” Cement, Concrete, and Aggregates, CCA GDP, V. 11, No. 2, pp. 9299. Patzias, T., 1987, “Evaluation of Sulfate Resistance of Hydraulic- ‘Cement Mortars by the ASTM C 1012 Test Method,"" Concrete Du- ‘ability, SP-100, American Concrete Institute, Detroit, pp. 2101-2120. Polivka, M., and Brown, E. H., 1959, “Influence of Various Fac- {ors on Sulfate Resistance of Concrete Containing Pozzolans,” Pro: ceedings, American Society for Testing and Materials, V. 58, pp. 1077-110. Regourd, M.; Mortureaux, B; and Hornain, H., 1983, “Use of, Condensed Silica Fume as Filler in Blended Cements,” Fly Ash, Sil- ica Fume, Slog and Other Mineral By-Products in Concrete, Ed. V. M. Malhotra, SP-79, American Conerete Institute, Detroit, pp. 2.62 ‘Thorvaldsen, T.; Wolochow, D.; and Vigfusson, V. A., 1932, “Studies on the Action of Sulfates on Portland Cement,” Canadian Journal of Research, V. 6, pp. 487-517, Uchikawa, H., and Uchida, S., 1974, “Analysis of Ettringte in Hardened Cement Paste," Cement and Concrete Research, V. 4, No, 5, pp. 821-834. Winslow, D. N., and Liu, Ding, “Pore-Structure of Concrete,” Cement and Concrete Research, V. 20, No. 2, 1990, pp. 227-235, Wolochow, D., 1952, “Determination of the Sulfate Resistance of Portland Cement, Report of Committee C-1 on Cement, Appendix,” Proceedings, V. 52, American Society for Testing and Materials, pp. 250-363; and “'Lean Mortar Bar Expansion Test for Sulfate Resis- tance of Portland Cements, Appendix A,"" Proceedings, V. 52, ‘American Society for Testing and Materials, 1952, pp. 264-265 Wong, G. Sam, and Poole, Toy, 1988, Sulfate Resistance of Mor- tars Made Using Portland Cement and Blends of Portland Cement ‘and Pozzolan or Slag, Technical Report SL-88-34, U.S. Army Engi- reer Waterways Experiment Station, Vicksburg, 143 pp. 69

You might also like