Comparison of Two Metakaolins and a Silica Fume Used as Supplementary Cementitious Materials

Transcription

1 Article to be published in: Proc. Seventh International Symposium on Utilization of High-Strength/High Performance Concrete, to be held in Washington D.C., June 20-24, Comparison of Two Metakaolins and a Silica Fume Used as Supplementary Cementitious Materials J.M. Justice, L.H. Kennison, B.J. Mohr, S.L. Beckwith, L.E. McCormick, B. Wiggins, Z.Z. Zhang, and K.E. Kurtis Synopsis: The performance of two metakaolins as supplementary cementitious materials (SCMs) was evaluated at 8% by weight cement replacement. The metakaolins varied by their surface area (11.1 vs m 2 /g). Performance of metakaolin mixtures was compared to control mixtures at water-to-cement ratios of 0.40, 0.50, and 0.60 where no SCM had been used and to mixtures where silica fume had been used as partial replacement for cement. In both mixtures containing metakaolins, compressive, splitting tensile, and flexural strengths increased, as well as elastic modulus, as compared to control mixtures. Setting time was reduced in the pastes with both metakaolins. Additionally, considering durability, both metakaolins reduced rapid chloride ion permeability and expansion due to alkali-silica reaction when compared to control and silica fume mixtures. In general, the finer of the two metakaolins proved more effective in improving concrete properties, although both performed superior to silica fume. Keywords: admixture; concrete; durability; fineness; metakaolin; pozzolan; SCM; silica fume; strength

2 ACI member Kimberly E. Kurtis is Assistant Professor in the School of Civil and Environmental Engineering at the Georgia Institute of Technology. She is Chairman of ACI Committee E802: Teaching Methods and Educational Materials, Secretary of Committee 236: Materials Science of Concrete, and Associate Member of Committee 201: Durability of Concrete. Joy M. Justice is a second-year graduate student at the Georgia Institute of Technology. She completed her undergraduate studies at the California Institute of Technology. Joy is advised by Dr. Kimberly Kurtis and is currently pursuing a Master's degree in Materials Science and Engineering. Luke H. Kennison earned a B.S. in Civil and Environmental Engineering from the Georgia Institute of Technology in 2003, followed by an M.S. in Luke was advised by Dr. Kimberly Kurtis. He is currently employed by Hazen and Sawyer, P.C. in Atlanta, Georgia. ACI Member Benjamin J. Mohr is a Ph.D. candidate at the Georgia Institute of Technology where he obtained his M.S. in Civil Engineering. He also obtained his B.S. in Civil Engineering from the University of Delaware. His research interests include fiber-cement materials, cement chemistry, and microstructure of cement-based materials. Staci L. Beckwith earned her Bachelors degree in Civil and Environmental Engineering from the Georgia Institute of Technology in 2004 and is currently employed by Power Engineers in Norcross, Georgia. Lauren E. McCormick is a fourth-year undergraduate student in Civil and Environmental Engineering at the Georgia Institute of Technology. She is advised by Dr. Kimberly Kurtis. ACI member Billy Wiggins is a professional engineer employed by Thiele Kaolin Company. He holds a Bachelors degree is in Civil Engineering and a Masters degree in Building Construction. Z.Z. Zhang obtained his M.S. and Ph.D. from Purdue University. Currently he is a senior research scientist and group leader of the materials research group at the R&D Department of Thiele Kaolin Company. INTRODUCTION Metakaolin differs from other supplementary cementitious materials (SCMs), like fly ash, silica fume, and slag, in that it is not a by-product of an industrial process; it is manufactured for a specific purpose under carefully controlled conditions. 1 Metakaolin is produced by heating kaolin, one of the most abundant natural clay minerals, to temperatures of C. This heat treatment, or calcination, serves to break down

3 the structure of kaolin. Bound hydroxyl ions are removed and resulting disorder among alumina and silica layers yields a highly reactive, amorphous material with pozzolanic and latent hydraulic reactivity, suitable for use in cementing applications. 2,3 Metakaolin reacts with portlandite (CH) to form calcium-silicate-hydrate (C-S- H) supplementary to that produced by portland cement hydration. This reaction becomes important within the interfacial transition zone (ITZ) located between aggregate and paste fractions. This region typically contains a high concentration of large, aligned CH crystals, which can lead to localized areas of increased porosity and lower strength. 4,5 Metakaolin can react with some of the CH produced by cement hydration, thereby densifying the structure of the hydrated cement paste. The rates of pozzolanic reaction and CH consumption in metakaolin systems have been shown to be higher than in silica fume systems, indicating a higher initial reactivity. 6 Because this reaction with CH occurs early and rapidly, metakaolin incorporation may contribute to reduced initial and final set times. 3,4 In addition, this refinement in the ITZ can result in increased strength in metakaolin concrete. 2,6 As portlandite in the ITZ and elsewhere in the paste is water soluble and is susceptible to deterioration in aggressive chemical environments, metakaolin has great potential for improving concrete durability. 4 Also, because the supplementary C-S-H formed during the pozzolanic reaction with metakaolin has a lower Ca/Si ratio than ordinary C-S-H, these products are believed to be better able to bind alkali ions from the pore solution, thus reducing concrete's susceptibility to alkali-silica reaction (ASR). 1,3,7 This potential beneficial use of metakaolin is particularly relevant, as silica fume agglomerates have been shown to contribute to ASR expansion in some cases. 8 Metakaolin has also been shown to decrease concrete permeability, which in turn increases its resistance to sulfate attack and chloride ion ingress Additionally, metakaolin may reduce autogenous and drying shrinkage, which could otherwise lead to cracking. 13 Thus, when used as a partial replacement for portland cement, metakaolin may improve both the mechanical properties and the durability of concrete. In general, metakaolin offers a set of benefits similar to those imparted by silica fume, including comparable strengths, permeability, chemical resistance, and drying shrinkage resistance. Physically, metakaolin particles measure approximately one-half to five micrometers across, making them an order of magnitude smaller than cement grains and an order of magnitude larger than silica fume particles. Both metakaolin and silica fume are typically used to replace 5 to 20 weight % of the cement. Metakaolin is white in color, whereas standard silica fume ranges from dark grey to black (although white silica fume is available at higher cost). This makes metakaolin particularly attractive in colormatching and other architectural applications. For these reasons, metakaolin is increasingly used in the production of high-performance concrete. 9,14 Here the influence of metakaolin fineness is investigated and the performance of concrete containing one of two metakaolins of different fineness are compared to silica fume concrete and ordinary concrete. Measurements of fresh properties, including

4 slump, setting time, and shrinkage (autogenous, chemical, and free shrinkage) are examined, as well as mechanical performance (compressive strength, splitting tensile strength, flexural strength, elastic modulus) and certain aspects of durability (shrinkage, chloride permeability, sulfate resistance, alkali reactivity). MATERIALS The two metakaolins used in this study, MK235 (Kaorock) and MK349 (Kaorock F), were provided by Thiele Kaolin Company in Sandersville, GA. These metakaolins differ primarily in their fineness, with MK349 having a smaller particle size and greater surface area (25.4 vs m 2 /g). Physical characteristics of the metakaolins are shown in Table 1. Commercially available Type I cement was used for all paste and concrete mixtures. Aggregates were #67 3/4"- (19 mm) MSA crushed stone and 2.38 fineness modulus natural sand, as well as alkali-reactive Jobe sand for the ASR testing. Commercially available silica fume and superplasticizer, which conforms to the ASTM C 494 Type F designation, were also used. Chemical analyses for silica fume and cement are shown in Table 2. EXPERIMENTAL METHODS General In this study, the early age properties of fresh concrete and mechanical performance and durability of hardened concrete were examined. All tests were conducted using the following four sample groups: (1) an ordinary cement paste or concrete, (2) pastes or concrete substituted with 8% MK235 by mass, (3) pastes or concrete substituted with 8% MK349 by mass, and (4) pastes or concrete substituted with 8% silica fume by mass. Pastes, mortars, or concretes were prepared at three water-to-cementitious materials ratios (w/cm) 0.40, 0.50, and 0.60 for each of the above sample groups, unless ASTM tests made specific requirement for w/cm. Concrete raw materials were batched and mixed for approximately 15 minutes in accordance with ASTM C 192 using a 2.5 ft 3 - capacity (71 L) Lancaster Counter Current Batch mixer, according to the mixture designs given in Table 3. Mortars and pastes were mixed with a Hobart mixer per ASTM C 305 for at least five minutes after the addition of water. SCMs were incorporated into mixtures concurrently with or immediately following cement after aggregates, but prior to water.

5 Concrete samples were removed from plastic-covered molds 24 hours after casting and placed in a 23 C fog room for the remainder of the active testing period. Mortar and cement paste samples, with the exception of the Vicat samples, were demolded at 24 hours and placed in a 23 C limewater curing tank. A summary of the tests conducted on fresh and hardened samples, their corresponding ASTM standards, and the dimensions of samples used for each is shown in Table 4. Early Age Properties Early age properties of pastes and concrete, including slump, unit weight, and setting time were measured. Slump was measured according to ASTM C 143. Superplasticizer was used as necessary in order to achieve a target slump of 3-4" ( mm) for all mixtures. Unit weight was calculated based on an average of five 3 6" ( mm) concrete cylinders. Time to initial and final set was measured using a Vicat apparatus according to ASTM C 191. Three samples were used for each measurement. Because pastes made with metakaolin required a higher water content to become workable, setting time tests were conducted both at a normal consistency (varying w/cm) as determined by ASTM C 187, as prescribed by the standard, and at a constant w/cm of This was the w/cm necessary for MK349 to reach normal consistency the highest value determined by ASTM C 187 (Table 5). In order for the other three experimental groups to approach normal consistency (approximately 15 mm penetration with the Vicat needle), these pastes were allowed to remain in the mixing bowl as necessary before placing in the ring molds and moist cabinet. Shrinkage Three types of shrinkage were monitored: autogenous, chemical, and free shrinkage. All examinations were conducted at a water-to-cementitious materials ratio of 0.40, and SCMs were used at 8% replacement. Autogenous shrinkage was measured on four replicate samples using rigid corrugated polyethylene tubes sealed on both ends to prevent loss of moisture to the environment. Linear deformation of these tubes was monitored according to the well-accepted technique described by Jensen and Hansen. 15 Chemical shrinkage was evaluated on four replicate samples by a method modified from Geiker and Knudsen. 16,17 This method involves introducing a known volume of cement paste (approximately 10 g, sample thickness < 10 mm) into a small glass vial fitted with a graduated pipette, thereby allowing volume change to be measured over time. Both autogenous and chemical shrinkage samples were kept in an environmental chamber at 20 C and 50% relative humidity. Data was recorded daily.

6 Free, or bulk, shrinkage was measured according to ASTM C 157 on three or four replicate samples, for concrete and mortar prisms, respectively. A sand-to-cement ratio of 2.25 was used to cast mortar prisms of " ( mm), and the mixture design shown in Table 3 was scaled down to make " ( mm) concrete prisms. These were removed from molds 24 hours after casting and allowed to cure in limewater for the remainder of one week. At that point, samples were moved to an environmental chamber at 23 C and 50% relative humidity to evaluate drying shrinkage. Measurements were recorded on days 1, 3, 5, 7, 10, and 14, and then every seven days for the following six weeks. Mechanical Properties Compressive strength was measured on three replicate samples according to ASTM C 39. Cylinders of 3 6" ( mm) were compressed at a rate of 20,000 lb/min (1480 N/s). Splitting tensile strength was also measured on three replicate 3 6" ( mm) cylinders loaded at a rate of 5,000 lb/min (370 N/s), as outlined in procedure ASTM C 496. Compression and tension tests were conducted on 1, 3, 7, 28, and 90 days of age using an 800,000 lb-capacity (3600 kn) compression machine with a digital indicator. Modulus of rupture was evaluated at ages of 1, 3, 28, and 90 days using a 400,000 lb-capacity (1800 kn) universal testing machine. Prisms were cast at " ( mm), and ASTM C 78 (third-point loading) was followed, using steel supporting rods and rubber pads. Three samples were tested in flexure for each condition. Modulus of elasticity was determined per ASTM C 469, using 6 12" ( mm) cylinders and a compressometer. This test was conducted on three replicate samples per mixture at day 28 of age on the 400,000 lb-capacity (1800 kn) load frame. Durability The rapid chloride permeability test (RCPT), as described in ASTM C 1202, was performed on three replicate specimens 4" (102 mm) in diameter and 2" (51 mm) thick. These samples were cut from 4 8" ( mm) concrete cylinders at 28 days of moist curing. To determine sulfate resistance, six replicate " ( mm) mortar bars were prepared and measured for expansion according to ASTM C In addition to controls, two sets of samples were cast for each of the three SCMs, one at 8% and one at 15% replacement. ASTM C 1012 was modified slightly by preparing all

7 mortar bars using a single w/cm of and adding superplasticizer to those mixes with metakaolin and silica fume replacements to achieve suitable workability. Samples were exposed to a 33,800 ppm sulfate solution (50 g/l sodium sulfate) at room temperature and length change was recorded weekly. Potential for alkali reactivity was measured according to the accelerated mortar bar method (ASTM C 1260). This method has been shown reliable for evaluating the effectiveness of SCMs in suppressing ASR. 18 As with sulfate testing, a single water-tocementitious materials ratio (0.47) was used and both 8% and 15% replacement levels were examined for samples (six replicates) containing SCMs. Gradation information for the alkali-reactive Jobe sand, which is slightly modified from the standard, and mortar mixture designs are shown in Tables 6 and 7, respectively. Mortar bars ( " or mm) were stored in a 1 N sodium hydroxide solution at 80 C, and length change data was collected for 28 days. RESULTS AND DISCUSSION General Metakaolin addition proved beneficial, yielding concrete with considerably higher strengths and greater durability than the controls (those without SCMs). In general, the finer MK349 appeared to be more effective in improving concrete properties than the coarser MK235, although both were more effective than silica fume. Key results regarding early age properties, shrinkage, mechanical properties, and durability are presented herein. Early Age Properties In order to achieve a target slump of 3-4" ( mm), superplasticizer was required for all concrete mixtures, with the exception of the control mixtures at a w/cm of 0.50 and When using SCMs, the necessary superplasticizer dosage increased with decreasing w/cm. MK235 typically required twice the amount used with silica fume, and MK349 required three times that value. Both metakaolins produced concrete with unit weights similar to control samples, while silica fume yielded concrete of 1% lower unit weight than controls on average over the three w/cm. The water-to-cementitious materials ratios used to produce pastes of normal consistency determined by ASTM C 187 are shown in Table 5, and the initial and final setting times measured are shown in Figure 1. In general, final set occurred approximately 30 minutes after initial set, which was generally shorter than expected. For the normal consistency pastes, the MK235-containing samples had longer initial and

8 final setting times than the control sample (initial set at 155 minutes), while the paste containing the finer MK349 had shorter setting times. The paste containing silica fume had the lowest setting times approximately 135 minutes for initial set. The initial and final setting times determined for pastes at a constant water-tocementitious materials ratio of 0.34 are shown in Figure 2. In this scenario, all pastes containing SCMs had shorter times to initial and final set than the control sample (initial set at 305 minutes) at this w/cm. Both metakaolin pastes exhibited faster setting times than the silica fume pastes, with the paste containing MK349 showing the fastest setting times of all four paste types at 145 minutes. Shrinkage Autogenous shrinkage, which occurs due to the lowering of cement paste relative humidity, increased with the addition of metakaolin. Both metakaolin samples experienced greater autogenous shrinkage than controls, with MK349 pastes showing the most shrinkage (-2100 µstrain at 28 days). Values were -750, -200, and +200 µstrain for MK235, silica fume, and control pastes, respectively, as shown in Figure 3. These results correspond to results for setting time at a constant w/cm. That is, pastes experiencing greater autogenous shrinkage had shorter times to initial and final set. Pastes containing metakaolin showed greater chemical shrinkage than controls or silica fume pastes, with the MK235 paste showing the most chemical shrinkage approximately 8.5 ml/100 g at 28 days of curing. The MK349 paste was expected to show the greatest chemical shrinkage because of its high surface area; the lower chemical shrinkage as compared to the coarser MK235 paste could be related to morphology and stacking behavior, but further investigation is necessary to isolate the potential causes for these observations. At 28 days, control samples had experienced the least chemical shrinkage: approximately 4 ml/100 g at 28 days. However, around seven weeks, silica fume samples actually started to expand, and continued to do so for the duration of the four-month testing period, such that they showed the least overall chemical shrinkage (Figure 4). This expansion is likely due to ASR involving silica fume agglomerations. Metakaolin incorporation had varying effects on free shrinkage. Mortar bars made with MK235 shrank approximately 0.053% during the first two weeks of drying. Overall, these samples showed the least bulk shrinkage over the duration of the six-week testing period. Prisms made with silica fume experienced the most shrinkage. Length change results are shown in Figure 5a. Mass loss followed the same trend as shrinkage: samples that shrank the most also generally lost the most mass during the testing period. Mass change is shown in Figure 5b. Concrete prisms made with MK235 also showed the least bulk shrinkage and the least mass loss during the first six weeks of drying, while prisms incorporating silica fume experienced the most shrinkage and mass loss (Figure 6).

9 Mechanical Properties A significant increase in compressive strength as compared to the ordinary concrete controls was observed for both metakaolin samples at 8% replacement for cement, with the finer MK349 having a more pronounced effect. Compressive strength increased with decreasing w/cm, reaching a value of nearly 11,000 psi (75 MPa) for the MK349 concrete with w/cm=0.40 at 28 days. Strength increases due to MK349 fineness were less apparent are higher w/cm likely because ample water was available for hydration and particle surface area became less critical at higher w/cm. Silica fume addition resulted in concrete of strength comparable to the control mixtures, which was unexpected. As a result, all the mixtures containing silica fume were recast using a new supply of the material; however, results from this second round of testing were similar. As shown in Figure 7, results were inconsistent across the three w/cm, but generally strength did not develop until later ages. The cause(s) for these lower-than-anticipated strengths with silica fume concrete are not clear, but similar results have been recently reported in the metro-atlanta region 19 and in the literature. 6,20 Agglomerations were not readily apparent when examining these samples by optical microscopy, thus it is assumed that adequate silica fume dispersion was achieved in these mixtures. Splitting tensile strength results also generally showed increases with metakaolin use as compared to ordinary concrete controls. Splitting tensile strengths for the metakaolin mixtures generally ranged between 3 and 4 MPa for all three w/cm at 28 days (Figure 8). Both MK235 and MK349 improved concrete performance in this test, but neither was dominant, and standard deviations for all test results were relatively large. The silica fume used did not greatly affect splitting tensile strength, as compared to controls. Metakaolin incorporation generally increased flexural strength when concrete prisms were subjected to four-point bending. There was an increase of 1-2 MPa associated with the use of both metakaolins, although there was not a clear trend indicating that one was superior. Prisms cast with silica fume as a partial replacement for cement showed a slightly higher modulus of rupture than the control at w/cm=0.40, though mixtures at higher w/cms did not differ from controls. At w/cm=0.40, MK349 and silica fume prisms reached 600 psi (4.1 MPa) at one day of age, control and MK235 prisms at three days. At higher w/cms, MK349 samples reached 600 psi (4.1 MPa) by day three, while other mixtures did not reach this value until 28 days of age. These results are shown in Figure 9. As with compression testing, silica fume addition yielded unexpectedly low elastic modulus values. However, when recast and cured, new silica fume samples showed a modulus of elasticity higher than controls and both metakaolin samples for all water-to-cementitious materials ratios, as depicted in Figure 10. The effect was most

10 pronounced at w/cm=0.40, with silica fume yielding an elastic modulus of 37 GPa versus 34 GPa for the metakaolins and 30 GPa for controls. Durability The metakaolin mixtures showed markedly lower permeability than controls, as measured by RCPT (Figure 11). For all w/cm, the control samples were considered to have a high permeability (above 4000 Coulombs). MK235 proved to be the most effective in reducing charge passed, with values in the low or very low (below 2000 Coulombs) range for all water-to-cementitious materials ratios. Concrete cast with MK349 produced RCPT results which were not much higher than the MK235 samples, with results ranging from very low at w/cm=0.40 to moderate for Silica fume addition also reduced permeability as compared to the control, with values in the low to moderate range. However, the silica fume did not produce reductions in permeability as great as either of the metakaolins. Sulfate resistance was measured by ASTM C Several months of measurements are reported in Figure 12. With the data to date, the effect of the metakaolins and silica fume is not clearly evident, as the mortar bars are only starting to show evidence of sulfate-induced expansion. Data collection is ongoing, but the results are expected to indicate that mixtures resistant to sodium sulfate attack can be produced using either of the metakaolins examined or silica fume at either 8% or 15% weight replacement. Results from ASR testing by ASTM C 1260 are presented in Figure 13. According to ASTM C 1260, expansion of less than 0.10% at 14 days of age indicates acceptable performance, and expansion of greater than 0.20% indicates unacceptable performance. Based upon these criteria, the 15% MK235 mixture passed, the 8% MK235 and 15% MK349 specimens fell into the intermediate range (showing between 0.10% and 0.20% expansion at 14 days), and all other mixtures failed. These results show that both metakaolins reduce expansion due to ASR and to a greater extent than silica fume at the same rate of addition. Additionally, 15% replacement with either metakaolin sample produced greater reductions in expansion than 8% replacement, and MK235 was more effective than MK349 in mitigating expansion due to ASR. CONCLUSIONS The performance of two metakaolins, which varied primarily in their fineness, was examined and compared to the performance of ordinary and silica fume pastes, mortars, and concretes. The following conclusions may be drawn:

11 1. With regard to workability and setting time, both metakaolins examined generally required more superplasticizer and shortened setting time of pastes as compared to control mixtures and companion silica fume mixtures. 2. Greater shrinkage, both autogenous and chemical, was observed in mixtures containing SCMs as compared to ordinary cement and concrete control mixtures. Of the two metakaolins, the finer material produced greater autogenous shrinkage, while the relatively coarser material produced more chemical shrinkage. Free shrinkage was generally greatest for the silica fume mortars and concrete. Of the two metakaolins, the coarser metakaolin (MK235) exhibited the least free shrinkage, even less than the control samples. The variation in shrinkage behavior (considering autogenous, chemical, and free shrinkage) warrants further examination. 3. Increased concrete strength, as compared to both control and silica fume mixtures, was measured for concretes produced with both metakaolins. However, the finer metakaolin, MK349, yielded the highest compressive and splitting tensile strengths and moduli. The positive influence of the metakaolin fineness was more apparent at lower w/cm. Concretes produced with the finer metakaolin also exhibited an increased rate of strength gain. 4. With regard to the durability tests reported here, concretes produced with metakaolin at 8% by weight cement exhibited reduced permeability, as measured by RCPT, with very low measurements at w/cm of 0.40, low measurements at 0.50, and moderate measurements at The coarser metakaolin generally produced greater reductions in permeability at all three w/cms examined. In accelerated alkali-silica reaction tests (ASTM C 1260), the best performance was achieved in mortars with 15% by weight replacement of cement with metakaolin, with the coarser material resulting in the least expansion in this test. REFERENCES 1. Ramlochan, T.; Thomas, M.; and Gruber, K. A., "The Effect of Metakaolin on Alkali-Silica Reaction in Concrete," Cement and Concrete Research, V. 30, 2000, pp Bensted, J., and Barnes, P., "Structure and Performance of Cements," Spon, New York, 2002, 565 pp. 3. Kostuch, J. A.; Walters, V.; and Jones, T. R., "High Performance Concretes Incorporating Metakaolin: A Review," Concrete 2000: Economic and Durable Concrete through Excellence, R. K. Dhir and M. R. Jones, eds., E&FN Spon, London, 1993, pp Wild, S., and Khatib, J. M., "Portlandite Consumption in Metakaolin Cement Pastes and Mortars," Cement and Concrete Research, V. 27, 1997, pp

12 5. Bentz, D. P., and Garboczi, E. J., "Simulation Studies of the Effects of Mineral Admixtures on the Cement Paste-Aggregate Interfacial Zone," ACI Materials Journal, V. 88, No. 5, Sept.-Oct. 1991, pp Poon, C.; Lam, L.; Kou, S. C.; Wong, Y.; and Wong, R., "Rate of Pozzolanic Reaction of Metakaolin in High-Performance Cement Pastes," Cement and Concrete Research, V. 31, 2001, pp Aquino, W.; Lange, D. A.; and Olek, J., "The Influence of Metakaolin and Silica Fume on the Chemistry of Alkali-Silica Reaction Products," Cement and Concrete Composites, V. 23, 2001, pp Diamond, S.; Sahu, S.; and Thaulow, N., "Reaction Products of Densified Silica Fume Agglomerates in Concrete," Cement and Concrete Research, V. 34, 2004, pp Boddy, A.; Hooton, R. D.; and Gruber, K. A., "Long-Term Testing of the Chloride- Penetration Resistance of Concrete Containing High-Reactivity Metakaolin," Cement and Concrete Research, V. 31, 2001, pp Asbridge, A. H.; Chadbourn, G. A; and Page, C. L., "Effects of Metakaolin and the Interfacial Transition Zone on the Diffusion of Chloride Ions through Cement Mortars," Cement and Concrete Research, V. 31, 2001, pp Khatib, J. M., and Wild, S., "Sulphate Resistance of Metakaolin Mortar," Cement and Concrete Research, V. 28, 1998, pp Courard, L.; Darimont, A.; Schouterden, M.; Ferauche, F.; Willem, X.; and Degeimbre, R., "Durability of Mortars Modified with Metakaolin," Cement and Concrete Research, V. 33, 2003, pp Brooks, J. J., and Megat Johari, M. A., "Effect of Metakaolin on Creep and Shrinkage of Concrete," Cement and Concrete Composites, V. 23, 2001, pp Ding, J., and Li, Z., "Effects of Metakaolin and Silica Fume on Properties of Concrete," ACI Materials Journal, V. 99, No. 4, July-Aug. 2002, pp Jensen, O. M., and Hansen, P. F., "A Dilatometer for Measuring Autogenous Deformation in Hardening Cement Paste," Materials and Structures, V. 28, No. 181, 1995, pp Geiker, M., and Knudsen, T., "Chemical Shrinkage of Portland Cement Pastes," Cement and Concrete Research, V. 12, 1982, pp Knudsen, T., and Geiker, M., "Obtaining Hydration Data by Measurement of Chemical Shrinkage with an Archimeter," Cement and Concrete Research, V. 15, 1985, pp Bérubé, M.; Duchesne, J.; and Chouinard, D., "Why the Accelerated Mortar Bar Method ASTM C 1260 is Reliable for Evaluating the Effectiveness of Supplementary Cementing Materials in Suppressing Expansion due to Alkali-Silica Reactivity," Cement, Concrete, and Aggregates, V. 17, 1995, pp Wolfe, J. E., personal communications with K.E. Kurtis, Apr.-June Curcio, F.; DeAngelis, B. A.; and Pagliolico, S., "Metakaolin as a Pozzolanic Microfiller for High-Performance Mortars," Cement and Concrete Research, V. 28, 1998, pp

13 Table 1 Physical characteristics of the two metakaolin samples examined. Characteristics MK235 MK349 Oxide Analysis (%) SiO Al 2 O TiO Fe 2 O Sedigraph PSD (%) < 2.0 µm < 1.0 µm < 0.5 µm 9 53 < 0.2 µm 4 4 Surface Area (m 2 /g) Bulk Density (lb/ft 3 ) (kg/m 3 ) Table 2 Chemical oxide analysis, weight %, for Type I cement and silica fume and Bogue potential compositions for the cement. Component Cement Silica Fume SiO Al 2 O Fe 2 O CaO MgO Na 2 O K 2 O TiO MnO P 2 O SrO BaO SO Loss on Ignition Insoluble Residue 0.11 N/A Moisture N/A 0.43 C 3 S 55.2 N/A C 2 S 19.3 N/A C 3 A 7.4 N/A C 4 AF 9.5 N/A

14 Table 3 Concrete mixture designs for control mixtures and mixtures with SCMs at 8% by weight replacement for cement. Amounts shown are required to produce one cubic yard or one cubic meter of concrete. 1 yd 3 or 1 m 3 Nominal w/cm Water Coarse (SSD) Fine (SSD) Cement SCM Density lb kg lb kg lb kg lb kg lb kg lb/yd 3 kg/m w/ SCM w/ SCM w/ SCM Table 4 Tests conducted and cast specimen dimensions. Description Designation Dimensions in mm Slump C 143 N/A N/A Setting time C 191 N/A N/A Compressive strength C Splitting tensile strength C Modulus of rupture C Modulus of elasticity C Autogeneous shrinkage N/A [14] Chemical shrinkage N/A [15, 16] Free shrinkage C , , Chloride permeability C Sulfate resistance C Alkali-silica reaction C

15 Table 5 Normal consistency of pastes, as determined by ASTM C 187. Sample w/c or w/cm 0.27 MK MK Silica Fume 0.28 Table 6 ASR testing, Jobe aggregate gradation. Sieve Mass retained, g Mass retained, % # # # # # Total Table 7 Mortar mixture designs for ASTM C 1260 (ASR test). Mixture Water (g) Aggregate (g) Cement (g) SCM (g) N/A 8% %

16 Initial Set Final Set Time (minutes) MK235 (8%) MK349 (8%) SF (8%) Sample Figure 1 Vicat initial and final setting times at normal consistency with varying w/cm Initial Set Final Set Time (minutes) MK235 (8%) MK349 (8%) SF (8%) Sample Figure 2 Vicat initial and final setting times at constant w/cm (0.34).

17 Autogenous deformation (microstrain) MK235 MK349 SF Age (days) Figure 3 Autogenous deformation of cement paste at w/cm=0.40, 8% replacement. Chemical shrinkage (ml/100 g cementitious material) MK235 MK349 SF Age (days) Figure 4 Chemical shrinkage of cement paste at w/cm=0.40, 8% replacement.

18 (a) Length change (%) MK235 MK349 SF (b) Mass change (%) Age (days) MK235 MK349 SF Age (days) Figure 5 Free shrinkage of mortar prisms at w/cm=0.40, 8% replacement: (a) length change and (b) mass change.

19 (a) Length change (%) MK235 MK349 SF Age (days) (b) Mass change (%) MK235 MK349 SF Age (days) Figure 6 Free shrinkage of concrete prisms at w/cm=0.40, 8% replacement: (a) length change and (b) mass change.

20 (a) Strength (MPa) MK235 MK349 SF SF Redo 10 0 (b) Strength (MPa) Age (days) MK235 MK349 SF SF Redo 10 0 (c) Strength (MPa) Age (days) MK235 MK349 SF SF Redo 0 90 Age (days) Figure 7 Average peak compressive strength versus concrete age for (a) w/cm=0.40, (b) w/cm=0.50, and (c) w/cm=0.60.

21 (a) 6 5 Strength (MPa) (b) 5 Strength (MPa) MK235 MK349 SF 90 MK235 MK349 SF Age (days) 0 (c) 4 Strength (MPa) MK235 MK349 SF Age (days) 0 90 Age (days) Figure 8 Average peak splitting tensile strength versus concrete age for (a) w/cm=0.40, (b) w/cm=0.50, and (c) w/cm=0.60.

22 (a) 9 8 Strength (MPa) Strength (MPa) (b) MK235 MK349 SF MK235 MK349 SF Age (days) 1 Strength (MPa) 0 (c) MK235 MK349 SF Age (days) Age (days) Figure 9 Average peak flexural strength (modulus of rupture) versus concrete age for (a) w/cm=0.40, (b) w/cm=0.50, and (c) w/cm=0.60.

23 E (GPa) MK235 MK349 SF SF Redo w/cm Figure 10 Modulus of elasticity, E, at 28 days of age. Charge passed (Coulombs) w/cm Figure 11 Rapid chloride permeability results at 28 days of age. MK235 MK349 SF HIGH MODERATE LOW VERY LOW

24 Length change (%) % MK235 8% MK349 8% SF 15% MK235 15% MK349 15% SF Age (days) Figure 12 Mortar bar expansion due to sulfate exposure, w/cm= Length change (%) % MK235 8% MK349 8% SF 15% MK235 15% MK349 15% SF FAIL 0.1 PASS Age (days) Figure 13 Expansion due to alkali-silica reaction, w/cm=0.47.

25 Evaluation of Thiele Metakaolin for Applications in Concrete Phase II Final Report September 12, 2006 V. Garas, F. Lagier, and K.E. Kurtis School of Civil and Environmental Engineering Georgia Institute of Technology Atlanta, Georgia

26 Table of Contents INTRODUCTION AND OVERVIEW OF THE LITERATURE... 1 OBJECTIVES... 3 MATERIALS... 3 EXPERIMENTAL METHODS... 4 General... 4 Design Approaches for Ternary Blends... 4 Particle Packing Optimization... 4 Isothermal Calorimetry... 5 Trial Batching... 6 Chloride Content of Metakaolin... 6 Plastic Properties... 6 Early Age Shrinkage... 6 Compressive Strength... 7 Durability... 7 RESULTS AND DISCUSSION... 7 Plastic Properties... 7 Chloride content of Metakaolin...7 Shrinkage... 8 Mechanical Properties... 8 Results from Ternary Blends... 8 Analysis of Particle Packing in Ternary Blends... 9 Influence of Early Curing Conditions... 9 Durability... 9 COST ANALYSIS CONCLUSIONS REFERENCES

27 INTRODUCTION AND OVERVIEW OF THE LITERATURE Metakaolin (MK) is produced by heating kaolin (natural clay) to temperature of o C to break down the kaolin s structure and create an amorphous material with pozzolanic and latent hydraulic reactivity, suitable for use in cementing applications. 1,2 In combination with Portland cement, metakaolin reacts with portlandite (CH) to form supplementary calcium silicate hydrate (C-S-H), the primarystrength giving phase in Portland cement concrete; various calcium aluminate products (e.g., C 4 AH 13, C 3 AH 6, C 2 ASH 8 ) may also result from the reaction of metakaolin in these types of systems. The formation of supplementary C-S-H, however, is particularly important within the interfacial transition zone (ITZ), which is the region of the paste immediately surrounding the coarse aggregate. This region typically is more porous (due to poor particle packing at the aggregate surface and one-sided growth) and contains a higher concentration of larger, aligned CH crystals. As a result, the ITZ is considered to be the weak link in concrete. 3,4 Because metakaolin reacts with CH produced by cement hydration and the supplementary C-S-H formed can fill existing porosity, the inclusion of metakaolin in concrete can densify the microstructure of the hydrated cement paste and potentially lead to greater strength and impermeability. 1,5 Compared to most other supplementary cementitious materials (SCM) used in concrete, metakaolin s reaction rate is rather rapid, owing to its high surface area and amorphous structure. Of the SCMs commonly used, silica fume is the only other material with a similar reaction rate. Fly ash is, like metakaolin, an SCM which is used as a partial replacement for portland cement in concrete. But, unlike metakaolin which is specially processed for use in concrete and which possesses latent hydraulic properties, fly ash is the largely pozzolanic by-product of industrial processes such as coal-burning power stations. Fly ash consists of spherical particles composed primarily of silica, alumina and iron; it s reaction rate is rather slow, 6,7,9 and depends upon the relative amount of crystalline (or nonreactive) material present, as well as on the calcium content of the fly ash. Whiles its rate of reaction is relatively slow, the use of fly ash can improve workability, 1 lead to higher later age strength, 6 and provide superior resistance sulfate attack, alkali silica reaction, and carbonation. 6,8 The difference between fly ash and Portland cement becomes apparent under a microscope. ASTM 618 describes two classes of fly ash: Class F, with SiO 2 + Al 2 O 3 + Fe 2 O 3 > 70% Class C, with SiO 2 + Al 2 O 3 + Fe 2 O 3 > 50% The ASTM specification indicates, then, that higher CaO-content fly ashes be classified as Class C. The recently revised Canadian Standards Association (CSA) specification for fly ash, CSA A23.5, divides fly ash into three classes depending on its calcium content in recognition of the difference in behavior between low and high lime fly ashes. These classes are as follows: Type F, low calcium, < 8% CaO Type CI, intermediate calcium, 8 20% CaO Type CH, high calcium, > 20% CaO Low-CaO fly ashes (Class F) generally provide good resistance to alkali-silica reaction (ASR). However, strength development at early ages is typically slower than that at conventional Portland cement, especially at higher levels of replacement. High-CaO fly ashes (Class C), on the other hand, are less efficient in suppressing chemical deterioration of concrete, but generally react faster than low-cao fly ashes and have less negative impact on the early strength of concrete and are less sensitive to inadequate curing. 10 Most fly ashes, regardless of composition, tend to reduce the water demand of concrete, while improving cohesiveness (i.e., resistance to segregation, such as excessive bleeding). The beneficial effects of fly ash on permeability and diffusivity tend to become more apparent with time, especially in the case of the more slowly reacting, low-cao fly ashes. 11 Research during the past decade has shown some additional improvements in concrete properties due to partial replacement of Portland cement with fly ash. In addition to increasing the long-term 1

28 compressive strength, 6,8,12 the long-term bond strength has been found to increase, up to 30% wt. fly ash. 13 Fly ash also was used to optimize the chloride resistance of typical structural concrete, as the use of fly ash resulted in up to 2 4 times better performance than corresponding Portland cement concrete at a 30 wt% replacement level. In addition, relatively low levels of moderate-calcium fly ash reduced the potential of alkali reactivity of aggregates as it reduced the expansion of the ASTM C 1260 test mortar bars at 14 days below the critical limit of 0.10%. 14 Thus, the benefits of using metakaolin or fly ash separately in concrete as partial replacement for Portland cement are fairly well-established, especially for fly ash. However, because the cost of metakaolin is about 6-7 times the cost of ordinary Portland cement in the United States, thus using metakaolin alone as a supplementary cementitious material (SCM) may not be cost effective. On the other hand, the slow reaction rate of fly ash can make its use impractical when rapid early strength development is required. However, use of these materials in combination as a ternary blend has the potential to overcome the higher cost associated with metakaolin concrete and the slower strength development associated with fly ash concrete. 16 Recent studies of ternary blends, which contain cement and two supplementary materials (SCMs), have shown improvements in economy 17, early and late strength 6,18,19,20, durability 8,18,19,20, and also decrease the heat of hydration 15 as compared to ordinary concrete or binary blends. 6,8 Ternary cementitious blends of Portland cement, silica fume, and fly ash offered significant advantages over binary blends and even greater enhancements over plain Portland cement, as the silica fume improves the early age performance of concrete, with the fly ash continuously refining the properties of the hardened concrete as it matured. 8,20 In addition, the shortfalls of high CaO fly ash in terms of controlling ASR resistance could be compensated for by the incorporation of relatively small quantities of other SCMs like silica fume. Such combinations produced concrete with generally excellent properties and offset the problems associated with using the increased amounts of high CaO fly ash or silica fume required when these materials are used individually. In terms of durability (chloride diffusion, ASR and sulfate resistance), such blends were vastly superior to plain Portland cement concrete, although it was not clear how the two materials worked together to improve the durability. 8 However, the same study suggested that it was possible that many of these benefits were attributed to reductions in permeability and ionic diffusivity in the system. Another study focused on accelerated cured ternary cement concrete showed that ternary systems (8% Silica fume + 25% slag) provided 18 h strengths exceeding 40 MPa. So, concrete mixtures containing 8% SF and 25% slag appear to have good potential for use in precast operations employing accelerated curing 19. In addition, the combined utilization of silica fume, fly ash and slag was found to be beneficial to the rheological properties of the high performance concrete (HPC). A recent study focused on studying the effects of different supplementary cementitious materials on strength and durability of concrete cured for a short period of time (2 weeks only), showed that replacing Portland cement with a combination of 10% silica fume, 25% slag, and 15% fly ash, also showed increase in compressive strength. 19 Only a few research studies have examined the incorporation of metakaolins in ternary blend systems, resulting in a body of knowledge which is much less complete compared to the literature available for fly ash, silica fume, and slag ternary blend systems. Reductions in free drying shrinkage, restrained shrinkage cracking width, and chloride diffusion rate have been reported when metakaolin is used in combination with silica fume, as compared to concrete where these SCMs have been used alone. 21 Another study showed that when metakaolin is combined with fly ash, the effects of MK and fly ash on the temperature-rise tend to compensate for one another. For example, the temperature-rise for a 10% MK 10% fly ash mortar is the same as that of the plain cement control. 8 For water-cured concrete made with Portland cement, fly ash, and metakaolin, increasing the MK content enhanced the 28-day compressive strength and reduced sorptivity to values below that of the control, whereas the sorptivities of fly ash concrete exceeded that of the control. 18 2

29 Thus, it is believed that a combination of metakaolin and fly ash in a ternary cement system (i.e., Portland cement being the third component) should result in a number of synergistic effects, some of which may include: Fly ash increases long-term strength development of metakaolin concrete. Fly ash offsets increased water demand of metakaolin. Fly ash compensates for higher heat release from metakaolin cement. The relatively low cost of fly ash offsets the increased cost of metakaolin. Metakaolin compensates for low early strength of concrete with fly ash (binary blend of cement and fly ash). Metakaolin reacts with CH to produce C-S-H, thus potentially improving the behavior of higher CaO fly ash for reduces the normally high levels of high CaO fly ash required for ASR preventio Thus significant improvements in mechanical and durability properties could be achieved upon replacing some of the cement with metakaolin, and fly ash. That will be the main focus of this study. OBJECTIVES The principal objective of this research plan was to build upon the prior research in Phase I, which examined the effect of metakaolin addition rate on compressive and flexural strength development, plastic concrete properties, shrinkage, permeability, and durability to address issues not previously considered in Phase I and to examine with further testing those results from Phase I which were inconclusive. A major focus of this continued research is the development and characterization of metakaolin/fly ash ternary-blended concrete. The specific studies to be performed, then, under this Phase II research plan are: Development of ternary blends containing Class C fly ash and metakaolin Assessment of plastic concrete properties Measurements of chloride content in both metakaolin samples Measurement of very early age (< 24 hour) concrete shrinkage Assessment of ternary blends for mitigation of expansion by alkali-silica reaction Assessment of metakaolin for mitigation of sodium sulfate and magnesium sulfate attack. In addition, a detailed cost analysis - based on the concrete strength at different ages - was also done to assess the suitability and practicality of using metakaolin as a supplementary cementitious material in ternary blended concrete. The cost of metakaolin based upon different locations of the United States and on the amount ordered was considered. Additional testing to characterize the freeze/thaw resistance and contribution to mitigation of corrosion afforded by the use of metakaolin was performed at Texas A&M University; those results are reported elsewhere. MATERIALS The two metakaolins used in this study, MK 235 (Kaorock) and MK 349 (Kaorock F), were provided by Thiele Kaolin Company in Sandersville, GA. These metakaolins differ primarily in their surface area, with MK 349 having the greater surface area (25.4 vs m 2 /g). Physical characteristics of the two metakaolins are shown in Table 1. Commercially available Type I/II cement and Class C fly ash were used for all paste, mortar, and concrete mixtures. Oxide analyses for cement and fly ash are shown in Tables 2 and 3. Aggregates were #89 1/2"- (12.5 mm) MSA crushed stone and 2.38 fineness modulus natural sand, as well as alkalireactive Jobe sand for the ASR testing. A carboxylated polyether-type superplasticizer (Grace ADVA 100) was also used. 3

30 EXPERIMENTAL METHODS General Pastes, mortars, or concretes were prepared at three water-to-cementitious materials ratios (w/cm) 0.30, 0.40, and 0.50 unless ASTM specifications for a particular test required a different w/cm. Concrete raw materials were batched and mixed for approximately 15 minutes in accordance with ASTM C 192 using a 2.5 ft 3 -capacity (71 L) Lancaster Counter Current Batch mixer, according to the mixture designs given in Table 4. Mortars were mixed with a Hobart mixer per ASTM C 305 for at least five minutes after the addition of water. SCMs were incorporated into mixtures concurrently with or immediately following cement after aggregates, but prior to water. The plastic-covered concrete were kept in their molds at ambient conditions or in insulated curing boxes for the first 24 hours and then demolded and placed in a 23 C fog room for the remainder of the active testing period. A summary of the tests conducted on fresh and hardened samples, their corresponding ASTM standards, and the dimensions of samples used for each is shown in Table 5. Design Approaches for Ternary Blends Three mix deign approaches were evaluated for the design of ternary blends containing Class C fly ash and metakaolin: 1- particle packing optimization, 2- isothermal calorimetry, and 3- trial batching at three water/cementitious materials (w/cm) of 0.30, 0.40, and 0.50, where the initial cement weight in the control mixes was replaced by either 25% FA only, or 25% FA + 3% or 5% or 8% of one of the metakaolins. Particle Packing Optimization The commercially available computer particle packing model LISA2 was used to optimize packing of the particles (i.e., cement, SCMs, aggregates) used to produce concrete. LISA2 is shareware, provided by the silica fume-supplier Elkem, that calculates and displays the particle size distribution of a mixture of components by creating a library of particle size distributions for different materials, determining the particle size distribution for any combination of these materials, and then comparing the resulting particle distribution graph to the corresponding graph produced by a model developed by Andreassen in 1931 (Figure 1). 22 Andreassen suggested that optimal packing occurs when the particle size distribution can be described by: CPFT = (d/d) q (1) where CPFT is the Cumulative (Volume) Percent Finer Than, d is the particle size, D is the maximum particle size, and q is the distribution coefficient. This model (eq. 1) assumes that the smallest particles would be infinitesimally small, which is not realistic for cement-based materials. A later, modified version of the model considers the minimum particle size distribution; this called the Modified Andreassen Model : 23 CPFT = (d q -dm q /D q -dm q )....(2) where CPFT is the Cumulative (Volume) Percent Finer Than, d is the particle size, dm is the minimum particle size of the distribution, D is the maximum particle size, and q is the distribution coefficient (qvalue) which increases upon increasing the amount of coarse materials and decreases upon increasing the amount of fine materials. A more detailed description of the two models and the software algorithms is given in the LISA2 user manual. 22 4

31 For this study, the Modified Andreassen Model was used. Prior research 23 has shown that a q coefficient value of will result in high strength concrete, and that q values < 0.23 give less workability mixtures. This approach was applied to results reported by Wiggins 24 for the strength of mortar mixtures containing metakaolin and fly ash (Table 6). However, no correlation was found between the distribution coefficient, q, and the compressive strength of the mortars (Figure 2). Thus, other approaches to mix design and optimization were explored. Isothermal Calorimetry Recent, not yet published work by Hansen s group at University of Michigan suggests that relationships can be developed to link heat evolved in pastes, measured through isothermal, to concrete compressive strength. In an attempt to understand the influence of metakaolin on the reaction of pastes containing Class C fly ash and, ultimately, to use this knowledge to optimize mix proportions, calorimetry was performed on plain cement pastes and binary and ternary blends. Ternary blends containing 3 and 5% metakaolin were examined based upon economic considerations. Heat of hydration was measured via isothermal calorimetry using a Thermometric TAM Air eight-channel heat conduction calorimeter maintained at 25 C. Pastes contained 200 g total cementitious material were mixed using a Sunbeam hand mixer. Polyethylene ampoules were filled with approximately 20 g of paste, with empty ampoules serving as references. All experiments were conducted in replicates of three and data was collected for at least 24 h. The heat of hydration of the control, binary, ternary blends at w/cm of 0.40 were measured using Type I/II cement, with 25% fly ash, and 3 and 5% metakaolin, following the mixture proportions in Table 4. Additionally, the influence of increasing fly ash content was examined in 5% metakaolin pastes by examining Class C fly ash dosages of 25, 35, 40, and 50%. Figure 3a shows the rate of heat evolution and Figure 3b shows the cumulative heat evolved for the 25% fly ash cases. Data in both is normalized per gram of cementitious material. As expected, the lowest and slowest heat evolution occurred in the 25% fly ash paste, with the MK pastes (at 3 and 5%) generating less heat than the control, but more than fly ash alone. Fly ash incorporation reduced the total heat evolved (Figure 3b) relative to the control, due to the replacement of the highly exothermic cement with the fly ash. Additionally, upon incorporating each of the metakaolins in ternary blends with cement and fly ash, a significant increase in the total heat evolved was observed as compared to the binary blend (cement + 25% FA). However, the increase of the MK content from 3% to 5% resulted in only small increases in the cumulative heat of hydration. While no clear influence of metakaolin surface area is found in the cumulative heat of hydration data, of the blended pastes, the finer MK 349 showed generally greater peaks in heat evolution when used at a replacement level of 5%. This specimen produced approximately 2.5 wm/g cement, which occurred at just over six hours of age, and coincided with the second peak in hydration, which is typically associated with the reaction of calcium aluminate phases. This effect was also noted when examining cements of varying composition (Table 7). In all cements examined, an acceleration in the reaction rates and an increase in the rate of heat evolution were observed in the presence of metakaolin, suggesting that they have a catalyzing effect on the cement hydration (Figure 5). In addition, for those cements with moderate and high alkali contents (i.e., Cements 2 and 5 in Table 7), increased heat associated with the second peak was noted. This effect may be important for massive concrete construction, where early heat development must be controlled, or in construction when longer setting times are necessary. That the MK 349 may be more reactive than MK 235 is also suggested by the data in Figure 4, where the metakaolins are combined at 5% by weight of cement with 25, 35, 40, and 50% fly ash. Here, MK 349 blends resulted in both higher peak heats of hydration and slight increases in the cumulative heat evolved when compared to MK 235 blends at the same fly ash content. 5

32 Figure 4 also shows that increasing the fly content decreases both the cumulative amount and the rate of heat evolved, as expected. This will likely result also in a decrease in the strength of concrete especially in early ages. Overall, these results suggest that metakaolin replacements of 5% or higher are needed in order to overcome the hydration delay that resulted form incorporating fly ash at a replacement level of 25% of the mass of the cement in the control. Trial Batching Based upon isothermal calorimetry results, ternary blends (Portland cement, fly ash Class C, and metakaolin) were developed at three w/cm s , 0.40, and 0.50 as shown in Table 4. Guidelines in ACI 211.4R-93 Guide for High Strength Concrete Mixture Proportioning were used to develop these mix proportions. A target slump of 3-4 was chosen to match Phase I. Chloride Content of Metakaolin The chloride content of raw metakaolin materials was measured according to FM Florida Method of Test for Determining Low-Levels of Chloride in Concrete and Raw Materials, using a Metrohm Autotitrator (798 MPT Titrino) (Figure 6). Three gm samples were used for each of the two metakaolins. Each sample was prepared by adding 35 ml of a 1:12 nitric acid solution to g of the raw material and heating at ~ 250 o C until boiling. The samples were allowed to boil for 2 to 4 minutes. Samples were then filtered through a Whatman No. 41 filter paper into 100 ml volumetric flasks and allowed to cool to room temperature. The titration process then started by placing the sample on the magnetic stirrer of the autotitrator, and the titrant (0.01N AgNO3 solution) was dispensed until the end point is reached. The chloride content is internally calculated by the titrator according to the following equation: Cl - (ppm) = (1,000VxMxN)/(W)...(3) where: N = normality of the AgNO 3 solution V = corrected end point in milliliters M = molecular weigh of chloride W = weight of sample in grams Plastic Properties Slump of the concrete mixtures prepared according to Table 4 was measured according to ASTM C 143. Superplasticizer dosage was adjusted as necessary in order to achieve a target slump of 3-4" ( mm) for all mixtures, and the amount required for each mixture is recorded in Table 4. Unit weight was calculated based on the average of five hardened 3 6" ( mm) concrete cylinders. Early Age Shrinkage Free shrinkage of the mixtures shown in Table 8 were examined according to ASTM C 157 with the first 24 hours shrinkage measured using embedded waterproof, low-modulus strain gages (TML KM- 30) with 31 mm gage length. Each shrinkage prism 3 x3 x11.25 ( mm) mold was first halffilled with concrete, then a strain gage was embedded and centered in the mold using two inverted U shape steel seats (Figure 7), then the second half was filled. Temperature at the center of the specimen was measured simultaneously with strain, using an OMEGA Type K thermal probe embedded vertically in the specimen immediately after casting. Strain and temperature measurements started at ~15 minutes after mixing was finished. Data was recorded every 1 min., using data acquisition systems. 6

33 Compressive Strength Compressive strength of ordinary concrete and binary and ternary blends shown in Table 4 was measured on three replicate samples according to ASTM C 39. Compression tests were conducted on 1, 3, 7, and 28 days of age using an 800,000 lb-capacity (3600 kn) compression machine with a digital indicator. Cylinders of 3 6" ( mm) were compressed at a rate of 15,000 lb/min (1110 N/s). Durability To determine sulfate resistance of mixture proportions investigated in Phase I (Table 9), four replicate " ( mm) mortar bars were prepared and measured for expansion according to ASTM C In addition to controls, two sets of samples were cast for each of the three SCMs (i.e., MK 235, MK 349, and silica fume), one at 8% and one at 15% replacement by wt. cemetn. ASTM C 1012 was modified slightly by preparing all mortar bars using a single w/cm of and adding superplasticizer to those mixes with metakaolin and silica fume replacements to achieve suitable workability. Samples were exposed to a 33,800 ppm sulfate solution (50 g/l sodium sulfate), at room temperature and length change was recorded on weekly basis for the first 3 weeks, and then monthly until the age of 1 year. Additionally, to examine magnesium sulfate attack, ASTM C 1012 was repeated using MgSO 4 also at a sulfate concentration of 33,800 ppm ( g/l magnesium sulfate). Use of ternary blends for control of expansion of alkali-reactive was measured according to the accelerated mortar bar method (ASTM C 1260). This method has been shown reliable for evaluating the effectiveness of SCMs in suppressing ASR. 25 A single water-to-cementitious materials ratio (0.47) was used and 3, 5, and 8% metakaolin replacement levels were examined with 25% Class C fly ash. Gradation information for the alkali-reactive cherty sand (Jobe, TX), which is slightly modified from the standard, and mortar mixture designs are shown in Tables 10 and 11 respectively. Mortar bars ( " or mm) were stored in a 1 N NaOH solution at 80 C, and length change data was collected for 28 days, which is longer than the 14-day period specified in the standard. RESULTS AND DISCUSSION Metakaolin addition proved beneficial, yielding concrete with considerably higher strengths and improved durability over the controls and fly ash binary blends. In general, the finer MK 349 appeared to be more effective in improving concrete properties than the coarser MK 235. Plastic Properties The use of fly ash generally decreased unit weight particularly at lower w/cm. Unit weight decreases were typically between 0.5% and 1.0% as compared to the control mixes. Mixtures with metakaolin and fly ash (ternary blends) showed further decrease in unit weight, typically between 0.25% and 1.70%, as compared to the control mixes. No superplasticizer was needed to achieve a target slump of 3-4" ( mm) for cement or fly ash binary blend at 0.50 w/cm, but the admixture was needed for all other concrete mixtures, as shown in Table 4. As expected, higher dosages were required as the w/cm decreased. The amount of superplasticizer needed generally increased upon using metakaolin and generally decreased upon using fly ash. MK 349 required from 15% to 400% more superplasticizer than MK 235. When using fly ash at a replacement level of 25% by mass of cement, a decrease of 32% to 60% in the necessary superplasticizer dosage was observed to achieve the target slump, when compared to the corresponding control mixes with same the w/cm (Table 4). Chloride content of Metakaolin The average chloride content measured for raw MK 235 and MK 349 was ppm ( %), and ppm ( %) respectively (Table 12). This shows that the finer metakaolin (MK 349) contained about twice the chloride in MK 235. Although FM5-516 does not provide a specific 7

34 classification of raw materials according to their chloride content, it is recommended that Thiele Kaolin consider including this test as part of their ordinary QA/QC procedures if repeated tests are necessary to maintain approval within the state of Florida Shrinkage Phase I shrinkage studies on MK concretes showed that metakaolin replacement resulted in increased autogenous and chemical shrinkage, measured in cement paste. However, reduced free shrinkage was observed in mortars and concretes. The inconsistency of these results prompted further investigation to better understand the shrinkage behavior of metakaolin concretes. Results from very early age (< 24 hours) shrinkage tests, performed with concrete, are shown in Figure 8. These results show that during the first 24 hours, the metakaolin-containing concretes experienced less shrinkage or comparable shrinkage to the control concretes. This supports the results obtained on concrete in Phase I. The maximum shrinkage measured was % for the control, and % for the MK 235 concrete and % for the MK 349 concrete. This represents 7.5%, and 55.5% reduction in shrinkage for the metakaolin concretes, as compared to the control. The reduction in early shrinkage upon incorporating metakaolin could be related to the very low permeability (and reduced water loss) in metakaolin concretes, as reported in Phase I. However, storage in high RH environment during the testing suggests that another mechanism is more likely responsible for the behavior. For example, the decreased shrinkage during the first 24 hours could be related to the slower reaction rate of the metakaolin, relative to the cement it replaces. Also, it could be that any shrinkage at < 24 hours is partially or in the case of MK 349 completely offset by expansion. Data in Figure 8 shows that the MK 349 sample actually exhibited a net expansion; this could be thermal expansion due the higher internal temperature measured in this concretes, as compared to the control and the MK 235 concretes. In addition, any expansion may also be related to greater early ettringite formation, as suggested by the exaggeration of the second peak (associated with reaction of aluminates) in calorimetry. This could result from the higher alumina content in metakaolin when compared to portland cement, and, hence, the greater alumina content in the metakaolin/cement pastes (Tables 1 and 2). Mechanical Properties Results from Ternary Blends Significant increases in compressive strength as compared to the ordinary concrete controls and the fly ash binary blends were observed for both metakaolins at 8% replacement for cement at low w/cm (i.e. w/cm=0.30) (Figure 9). The finer MK 349 had a more pronounced effect, particularly at ages of three days or more. In general, the compressive strengths of all mixtures increased with decreasing w/cm, reaching a value of nearly 12,400 psi (85.5 MPa) for mix 8 (25% FA + 8% MK 349) at 28 days. As in Phase I, strength increases due to metakaolin were less apparent at higher w/cm. This is likely because ample water was available for hydration and particle surface area and size became less critical at higher w/cm. Higher rates of strength gain than the control concretes were apparent for the 5% and 8% MK 349 ternary blends at one day for a w/cm of 0.30 and at seven days for a w/cm of By the age three days, 8% metakaolin ternary blends strengths exceeded the control at w/cm of 0.3. For w/cm or 0.40, this occurred at 28 days of age (Figure 10). 8

35 Analysis of Particle Packing in Ternary Blends Results from the Phase II (ternary blend) strength characterization were analyzed using the modified Andreassen model, described previously in the section Particle Packing Optimization, to determine if a relationship between the model coefficient (q) and the measured strength of these mixtures could be found. First, as expected, the modified Andreassen model coefficient for these mixtures decreased as the amount of fine materials (i.e., metakaolin) in the mix increased. However, while a q coefficient of between 0.25 and 0.30 has been associated with high performance concrete, 23 a strong correlation between q value and strength was not apparent in the Phase II strength data, as was also the case for the Wiggins data presented earlier (Figure 2). For example, while Figures show that ternary blends with 5% and 8% MK showed significantly higher compressive strength (and other data, presented subsequently in Figures 16 and 17 show improvements in durability), almost all the q values for these mixes were below This is likely because the modified Andreassen model does not adequately account for the faster and greater chemical reactivity of the metakaolin. Furthermore, this result suggests that - for highly reactive materials like metakaolin - concrete performance is more significantly affected by the chemical reactivity of the material not the optimum packing of particles. A relationship was found, however, to exist between q value and superplasticizer dosage required to achieve the target slump. Upper limits of 0.23 and 0.25 have been set by researchers 23 and the LISA2 software developers, respectively, to maintain high flowability in high performance concrete. Figure 14 shows that mixtures with q values lower than these limits required more superplasticizer, indicating that they had a lower workability than mixtures with higher q value. These results are counter to the expected and suggest that the faster rate of chemical reactivity and higher surface area (in addition to particle size) of highly reactive materials like metakaolin should be also accounted for as while modeling the rheology of concretes incorporating such materials. Influence of Early Curing Conditions Additionally, to replicate conditions at center of large concrete sections or produced by accelerated curing practices, control and 5% MK ternary blends specimens at w/cm=0.30 were cast and kept for the first 24 hours in an insulated curing box and were subsequently cured in a fog room. Results from these samples which were initially accelerated cured were compared to those cured entirely by the ASTM standard curing regime (Figure 19). Similar trends are apparent for control and ternary concretes. Initial accelerated curing resulted in ~70% increase in one-day compressive strength, as compared to standard curing. The one-day strength is ~87% of the 28-day compressive strength, indicating a higher initial reaction rate due to the higher temperatures in the insulated curing box. Strengths at later ages (i.e. 7 and 28 days) were similiar to those subjected to ASTM-fog room curing. The significant increase in the early strength is of particular importance to prestressing industry where early high strength is needed in order to apply the prestressing forces without causing damage to the concrete section. Durability Sulfate resistance to both sodium and magnesium sulfate - afforded by the use of metakaolin was measured by procedures in ASTM C The results shown in Figure 16 show that specimens exposed to MgSO 4 showed less expansion than specimens exposed to Na 2 SO 4. This may be due to the formation of a brucite layer near the sample surface with MgSO 4 exposure; formation of relatively insoluble brucite may slow sulfate ingress. As a result, it is difficult to distinguish among the expansion results for the various mixtures in the magnesium sulfate environment. Although ASTM C 1012 provides no strict limit on expansion, it has been suggested that expansion of 0.05% to 0.10% at 180 days would indicate moderate sulfate resistance and expansion of less than 0.05% at 180 days would indicate high sulfate resistance during ASTM C 1012 exposure. 20 9

36 According to this criteria, all mixtures investigated fell into the passing range (below 0.05% expansion at 180 days). The low C 3 A content (< 8%) of the Type I/II cement used likely contributed to the sulfate resistance of the mortars. However, the type and addition rate of the SCMs examined clearly influenced the expansion behavior in the standard Na 2 SO 4 solution. Results in Figure 16 show that when used at 8% by weight of cement, metakaolin produces greater reductions in expansion than silica fume. In fact, while further reductions in expansion are noticed when the silica fume dosage is increased from 8% to 15%, the behavior of the metakaolin mortars at both of these rates are nearly the same in the sodium sulfate environment. This indicates that the metakaolin may be used at lower cost, assuming silica fume and metakaolin are comparably priced, to provide equivalent sulfate resistance in concrete. In addition, the performance of ternary blends for mitigation expansion by alkali-silica reaction was examined by ASTM C 1260 (Figure 17). According to ASTM C 1260, expansion of less than 0.10% at 14 days of age indicates acceptable performance, and expansion of greater than 0.20% indicates unacceptable performance. Based upon these criteria, the 8% MK 235 ternary blend, the 3%, 5%, and 8% MK 349 ternary blends specimens fell into the intermediate range (showing between 0.10% and 0.20% expansion at 14 days), while all other mixtures failed. These results show that although both metakaolins significantly reduce expansion due to ASR, MK 349 was found to be more efficient, perhaps due to its larger surface area and greater reactivity. Additionally, 8% replacement with MK 235 produced greater reductions in expansion than 3% and 5% replacements, while increasing replacement with MK 349 from 5% to 8% did not produce significant reduction in expansion due to ASR. Further testing, by the longer term ASTM C 1293, is ongoing. COST ANALYSIS A cost analysis was performed based on the price lists received from Thiele Kaolin in August 2006 (Table 13) and the compressive strength results for control, binary, and ternary blends investigated in Phase II (Table 4). Figures 18-21, 22-25, 26-29, and show analysis results when Thiele s metakaolin is ordered in different amounts for use in the state of Georgia, the east coast, west of the Mississippi river, and in California, respectively. These results show that the 3%, and 5% MK ternary blends showed cost savings (in terms of cost/psi) at low w/cm (i.e. w/cm = 0.30) at 28 days when compared to the control mix (PC concrete). The magnitude of cost reduction depended on the amount of the metakaolin ordered. For use in Georgia, the 3% MK 235, 3% MK 349, and 5% MK 349 ternary blends showed a maximum of 7.0 %, 3.5 %, and 4.7 % reduction in the cost/psi at 28 days, when compared to the w/cm=0.30 control mix. For the east cost, the 3% MK 235, 5% MK 235, 3% MK 349, and 5% MK 349 blends showed maximum reductions in cost/psi at 28 days of 6.3 %, 0.5 %, 2.9% and 3.5 % as compared to the w/cm=0.30 control mix. In the area west of the Mississippi River, the 3% MK 235, 3% MK 349, and 5% MK 349 blends showed a maximum reduction in cost/psi at 28 days of 5.03 %, 1.6 %, and 1.6 % compared to the w/cm=0.30 control mix. In California and the west coast, only the 3% MK 235 ternary blend showed a reduction (3.4%) in the cost/psi when compared to the w/cm=0.30 control mix at 28 days; this is due to the higher freight rates for shipping metakaolin further distances. However, the initial cost of materials is but one parameter in determining the lifecycle cost of a structure or project. Thus, potential improvements in durability and resulting increased service life and decreased maintenance costs must also be considered. By taking into account the enhanced durability performance afforded by the use of metakaolin, as compared to control and fly ash binary blends investigated (refer to the Durability section, and Phase I report), a long-term cost effectiveness could be 10

37 achieved upon using metakaolin concretes. This is particularly true when longer service life is required and/or when construction is to occur in aggressive environments. CONCLUSIONS The performance of two metakaolins, which varied primarily in their fineness, was examined and compared to the performance of plain concrete and in concrete containing ternary blends of the metakaolin and Class C fly ash. Three mix design approaches for the development of metakaolin/fly ash ternary blends were examined - particle packing optimization, isothermal calorimetry, and trial batching. The particle packing approach was found to be unsatisfactory, likely due to lack of consideration of the constituent s chemical reactivity. Additionally, calorimetry was used to examine the influence of metakaolin addition on the early age reaction in ternary systems. While the fineness of the metakaolin was found to have some influence on accelerating the early reactions (when comparing the two metakaolin samples), these result overall suggest that metakaolin replacements of 5% or higher are needed in order to overcome the hydration delay that resulted form incorporating fly ash at a replacement level of 25% of the mass of the cement in the control. Thus, ternary blended concretes were developed using ACI guidelines, with additional investigation by calorimetry. From this investigation, the following observations for ternary blended concretes were made: When compared to companion control and fly ash concrete at the same w/cm, mixtures incorporating metakaolin as a third cementing material required more superplasticizer to achieve the same workability. Increased concrete strength, as compared to both control and fly ash mixtures, was measured for ternary blended concretes produced with both metakaolins. The finer metakaolin, MK 349, generally yielded the highest compressive strength. The positive influence of the metakaolin fineness on strength was more apparent at lower w/cm. Concretes produced with the finer metakaolin also exhibited an increased rate of strength gain when compared to the controls and the fly ash binary blends. In accelerated mortar tests (ASTM C 1260), both metakaolins significantly reduced expansion due to alkali-silica reaction when combined with 25% Class C fly ash. The finer MK 349 was found to be more efficient, mitigating expansion of a highly reactive cherty aggregate at 5% replacement, as compared to the 8% required for MK 235. The cost effectiveness in terms of (cost/psi) increased with time for all ternary blends. Results show that ternary blends of both metakaolins investigated showed cost savings (in terms of cost/psi) at w/cm of 0.30 at 28 days, with 3% MK ternary blends being the most cost effective. Farther cost savings may achieved in the log term taking into account the enhanced durability performance of MK concretes. In addition, continued examination of binary blends of metakaolin-cement systems yielded the following observations: With regard to durability, resistance to sodium sulfate attack was more improved upon incorporating metakaolin, with 8% metakaolin generally producing greater reductions in expansion than 8% silica fume. 11

38 Concrete containing 8% MK 235 experienced very early age shrinkage (<24 hours) comparable to the ordinary cement concrete control. Expansion was noted in the first 24 hours in the 8% MK 349 concrete, perhaps due to thermal effects or early ettringite formation. Finally, the chloride content of both metakaolins was measured. While both contained very small amounts of chloride (< 100 ppm), the finer metakaolin (MK 349) had twice as much chloride. It is recommended that Thiele Kaolin include this test as part of their ordinary QA/QC procedures, if repeated tests are necessary to maintain approval within the state of Florida. REFERENCES 1. Bensted, J., and Barnes, P., "Structure and Performance of Cements," Spon, New York, 2002, 565 pp. 2. Kostuch, J.A.; Walters, V.; and Jones, T.R., High Performance Concretes Incorporating Metakaolin: A Review, Concrete 2000; Economic and Durable Concrete through Excellence, R.K. Dhir; and M.R. Jones, eds., E & FN Spon, London, 1993, pp Wild, S.; and Khatib, J.M., Portlandite Consumption in Metakaolin Cement Pastes and Mortars, Cement and Concrete Research, Vol. 27, 1997, pp Bentz, D.P.; and Garboczi, E. J., Simulation studies of the effects of Mineral Admixtures on the Cement Paste-Aggregate Interfacial Zone, ACI Materials Journal, Vol.88, No.5, Sept.-Oct.1998, pp Poon, C.; Lam, L.; Kou, S.C.; Wong, Y.; and Wong, R., Rate of Pozzolanic Reaction of Metakaolin in High-Performance Cement Pastes, Cement and Concrete Research, Vol. 31, 2001, pp Antiohos, S.; Maganari, K.; and Tsimas, S., Evaluation of Blends of High and Low Calcium Fly Ashes for Use as Supplementary Cementing Materials, Cement and Concrete Research, Vol. 27, 2005, pp Hwang, K.; Noguchi, T.; and Tomosawa, F., Prediction Model of Compressive Strength Development of Fly-Ash Concrete, Cement and Concrete Research, Vol. 34, 2004, pp Thomas, M.D.A.; Shehata, M.H.; Shashiprakash, S.G.; Hopkins, D.S.; and Cail, K., Use of Ternary Cementitious Systems Containing Silica Fume and Fly Ash in Concrete, Cement and Concrete Research, Vol. 29, 1999, pp March, Thomas, M.D.A.; Shehata, M.H.; and Shashiprakash, S.G., The Use of Fly Ash in Concrete: Classification by Composition, Cement, Concrete, and Aggregates, Vol. 12, No. 2, 1999, pp Thomas, M.D.A.; and Matthews, J.D., Chloride penetration and Reinforcement Corrosion in Marine- Exposed Fly Ash Concretes, Proceedings 3rd CANMET/ACI International Conference on Concrete in a Marine Environment, V.M. Malhotra (Ed.), ACI SP-163, American Concrete Institute, Detroit, 1996, pp Han, Sang-Hun; Kim, Jin-Keun; and Park, Yon-Dong, Prediction of Compressive Strength of Fly Ash Concrete by New Apparent Activation Energy Function, Cement and Concrete Research, Vol. 33, 2003, pp Li, G., A New Way to Increase the Long-Term Bond Strength of New-to-Old Concrete by the Use of Fly Ash, Cement and Concrete Research, Vol. 33, 2003, pp Bektas, F.; Turanli, L.; Topal, T.; and Goncuoglu, M.C., Alkali Reactivity of Mortars Containing Chert and Incorporating Moderate-Calcium Fy Ash, Cement and Concrete Research, Vol. 34, 2004, pp Bai, J.; and Wild, S., Investigation of the Temperature Change and Heat Evolution of Mortar Incorporating PFA and Metakaolin, Cement and Concrete Research, Vol. 24, 2002, pp

39 16. Dhir, R.K.; and Jones, M.R., Development of Chloride-Resisting Concrete Using Fly Ash, Fuel, Vol. 78, 1999, pp Shehata, M.H.; and Thomas, M. D.A., Use of Ternary Blends Containing Silica Fume and Fly Ash to Suppress Expansion due to Alkali Silica Reaction in Concrete, Cement and Concrete Research, Vol. 32, 2002, pp Bai, J.; Wild, S.; and Sabir, B.B., Sorptivity and Strength of Air-Cured and Water- Cured PC PFA MK Concrete and the Influence of Binder Composition on Carbonation Depth, Cement and Concrete Research, Vol. 35, 2005, pp Hooton, R.D.; and Titherington, M.P., Chloride Resistance of High-Performance Concretes Subjected to Accelerated Curing, Cement and Concrete Research, Vol. 34, 2004, pp Lane, D.S.; and Ozyildirim, C., Preventive Measures for Alkali-Silica Reactions (Binary and Ternary Systems), Cement and Concrete Research, Vol. 29, 1999, pp Khatib, J.M. and Wild, S., Pore size distribution of Metakaolin Paste, Cement and Concrete Research, Vol.26, No.10:. 1996, pp Kumar V, S.,and Santhanam, M, Particle Packing Theories and their Application in Concrete Mixtrure Proportioning: A Review, The Indian Concrete Journal, September 2003, pp Wiggins, B., Cement/Fly Ash/Metakaolin Ternary Blends: Effects on Compressive Strengths of Mortars, submitted for publication. 25. Bérubé, M.; Duchesne, J.; and Chouinard, D., "Why the Accelerated Mortar Bar Method ASTM C 1260 is Reliable for Evaluating the Effectiveness of Supplementary Cementing Materials in Suppressing Expansion due to Alkali-Silica Reactivity," Cement, Concrete, and Aggregates, Vol. 17, 1995, pp Taylor H.F.W., "Cement Chemistry" Second Edition, Thomas Telford Publishing, Frías, M., Sánchez, M.I., Cabrera, J., The Effect that the Pozzolanic of Metakaolin Has on the Heat Evolution in Metakaolin-Cement Mortars, Cement and Concrete Research, Vol. 30, 2000, pp Hewlett, P.C., Lea s Chemistry of cement and concrete, Fourth edition, Butterworth-Heinemann,

40 Table 1. Physical characteristics of the two metakaolin samples examined. Characteristics MK 235 MK 349 Oxide Analysis (%) SiO 2 Al 2 O 3 Fe 2 O 3 CaO MgO Na 2 O K 2 O TiO 2 MnO 2 P 2 O 5 SrO BaO SO 3 Sedigraph (PSD) % <2 µm % <1 µm % <0.5 µm % <0.2 µm Loss on Ignition Moisture, as received Surface Area (m 2 /g) Bulk Density (lb/ft 3 )

41 Table 2. Chemical oxide analysis, weight %, for Type I/II cement and Bogue potential compositions for the cement. Component Weight % Oxide analysis SiO 2 Al 2 O 3 Fe 2 O 3 CaO MgO Na 2 O K 2 O TiO 2 Mn 2 O 3 P 2 O 5 SrO BaO SO 3 Loss on Ignition Moisture N/A Bouge potential composition C3S C2S C3A 7.66 C4AF Table 3. Chemical oxide analysis, weight %, for class C fly ash Component Weight % Sum of SiO 2, Al2O 3, and Fe 2 O CaO MgO 5.85 Na2O 1.88 K 2 O 0.51 TiO MnO P 2 O SrO 0.38 BaO 0.71 SO Moisture content 0.10 Loss on ignition 0.43 Density

42 MK Type Table 4. Concrete mixture designs for control mixtures and mixtures with SCMs. 1 yd 3 or 1 m 3 FA Fine Agg. Coarse Agg. Cement Water Metakaolin (%) Fly ash Superplasticizer MK (%) Mixture w/cm lb kg lb kg lb kg lb kg lb kg lb kg lb kg

43 Table 5. Tests conducted and cast specimen dimensions. Description Designation Dimensions in mm Chloride content of raw materials FM5-516 N/A N/A Slump C 143 N/A N/A Early age free shrinkage C 157 3x3x11.25 prisms 76x76x286 prisms Compressive strength C 39 3x6 cylinders 76x152 cylinders Sulfate resistance C x1x11.25 prisms 25x25x286 prisms Alkali-silica reaction C x1x11.25 prisms 25x25x286 prisms Table 6. SCMs used in Wiggins Mixtures SCM Type Replacement Level Mix 1 Mix 2 Mix 3 Mix 4 Mix 5 Mix 6 Mix 7 Mix 8 Mix 9 Fly ash Metakaolin

44 Table 7. Chemical oxide analysis, weight %, cement and Bogue potential compositions for the five cements different in chemical composition. Oxide Analysis Weight % Cement 1 Cement 2 Cement 3 Cement 4 Cement 5 Component Low C 3 A Moderate C 3 A/Mod and Na2Oeq High C 3 A Low Na 2 Oeq High Na 2 Oeq Low Mod High Low High SiO Al 2 O Fe 2 O CaO MgO SO Na 2 O K 2 O NaO 2 eq P 2 O TiO SrO Mn 2 O Cr 2 O LOI LSF Silica Ratio Aluminum Ratio Bogue potential composition C 3 S C 2 S C 3 A C 4 AF NaO 2 eq

45 Mixture w/cm MK Type Table 8. Mix designs for early free shrinkage tests 1 yd 3 or 1 m 3 MK Fine Agg. Coarse Agg. Cement Water Metakaolin (%) lb kg lb kg lb kg lb kg lb kg Table 9. Concrete mixture designs for control mixtures and mixtures with SCMs at 8% by weight replacement for cement. Amounts shown are required to produce one cubic yard or one cubic meter of concrete. 1 yd 3 or 1 m 3 Theoretical w/cm Water Coarse (SSD) Fine (SSD) Cement SCM Density lb kg lb kg lb kg lb kg lb kg lb/yd 3 kg/m w/ SCM w/ SCM w/ SCM Table 10. ASR testing Jobe aggregate gradation. Sieve Mass retained (gm) % (from total) # # # # # Total

46 Mixture w/cm Table 11. Mortar mixture designs for ASTM C 1260 (ASR test). Amounts / 4 mortar bars MK Fine aggregates Cement Water % MK % FA Type (gm) (gm) (gm) MK (gm) FA (gm) ASR ASR ASR ASR ASR ASR ASR ASR Table 12. Chloride content test results Type of Metakaolin Specimen Cl - (ppm) Cl - (%) Specimen Specimen MK 235 Specimen Average Standard deviation Specimen Specimen MK 349 Specimen Average Standard deviation

47 Table 13. Price List of Metakaolin Amount of MK ordered 20 tons tons tons > 700 tons Cost ($)/ton Item West of Georgia East coast California Mississippi Material Price packaging Freight Total Material Price packaging Freight Total Material Price packaging Freight Total Material Price packaging Freight Total Table 14. Price List of Other Concrete Materials Material Cost ($)/ton Portland Cement Type I/II 90 Fly ash Class C 37 Superplasticizer 4100 Coarse Aggregates 11 Fine Aggregates 13 21

48 Figure 1. Particle size distribution of LISA2 where the actual particle distribution in a mixture is compared to the Modified Andreassen Model. 22

49 Wiggins Mixes Compressive Strength (psi) (a) % FA, 0% MK 15% FA, 0% MK 30% FA, 0% MK 0% FA, 5% MK 0% FA, 10% MK 15% FA, 5% MK 15% FA, 10% MK 30% FA, 5% MK 30% FA, 10% MK Mixtures Wiggins Mixes Coeffcient q (b) % FA, 0% MK 15% FA, 0% MK 30% FA, 0% MK 0% FA, 5% MK % FA, 10% MK % FA, 5% MK % FA, 10% MK % FA, 5% MK % FA, 10% MK- 235 Mixtures Figure 2. (a) Compressive strength and (b) q coefficient predicted by the Modified Andreassen Model for Wiggins s mixtures. 23

50 Figure 3. Isothermal calorimetry results showing the effect of metakaolin content on (a) rate of heat evolution and (b) cumulative heat evolved per gram of cementitious material. 24

51 Figure 4. Isothermal calorimetry results showing the effect of MK type on the rate of heat evolution, and the cumulative heat evolved per gram of cementitious material at w/cm or 0.40 and different fly ash contents. 25

52 4,5 4 (a) Cement Low C3A (Cemex-Calif T5) Cement Low C3A (Cemex-Calif T5) & 8% MK 235 Cement Low C3A (Cemex-Calif T5) & 8% MK ,5 200 Rate of heat evolution (mw/g) 3 2,5 2 1, Cumulative Rate evolution (J/g) , Age (h) 4,5 4 (b) Cement Mod C3A & NaO2 eq (Holcim-Ada,OK) Cement Mod C3A & NaO2 eq (Holcim-Ada,OK) & 8% MK 235 Cement Mod C3A & NaO2 eq (Holcim-Ada,OK) & 8% MK ,5 200 Rate of heat evolution (mw/g) 3 2,5 2 1, Cumulative Rate evolution (J/g) , Age (h) 0 26

53 4,5 4 (c) Cement High C3A (Holcim-Artesia) Cement High C3A (Holcim-Artesia) & 8% MK 235 Cement High C3A (Holcim-Artesia) & 8% MK ,5 200 Rate of heat evolution (J/g) 3 2,5 2 1, Cumulative Rate evolution (J/g) 0, Age (h) 0 4,5 4 (d) Cement Low (Na2.O)eq (Cemex-Clinchfield) Cement Low (Na2.O)eq (Cemex-Clinchfield) & 8% MK 235 Cement Low (Na2.O)eq (Cemex-Clinchfield) & 8% MK ,5 200 Rate of heat evolution (mw/g) 3 2,5 2 1, Cumulative Rate evolution (J/g) , Age (h) 0 27

54 4,5 4 (e) Cement High (Na2.O)eq (Lehigh-Allenown) Cement High (Na2.O)eq (Lehigh-Allenown) & 8% MK 235 Cement High (Na2.O)eq (Lehigh-Allenown) & 8% MK ,5 200 Rate of heat evolution (mw/g) 3 2,5 2 1, Cumulative Rate evolution (J/g) , Age (h) Figure 5. Rate of evolution and the cumulative heat involved for (a) Cement with low proportion in C 3 A, (b) Cement with moderate proportion in C 3 A and Na 2 O eq, (c) Cement with high proportion in C 3 A, (d) Cement with low proportion in (Na 2 O) eq, and (e) Cement with low proportion in (Na 2 O) eq 0 Figure 6. Test setup for measuring chloride content of metakaolin 28

55 Strain gage Thermal probe Shrinkage mold End pin Seats Figure 7. Early Shrinkage test setup Strain 1.E-04 8.E-05 6.E-05 4.E-05 2.E-05 0.E+00-2.E-05-4.E-05-6.E-05-8.E-05-1.E Time (Hrs) (Strain) MK 235 (Strain) MK 349 (Strain) (Temp.) MK 235 (Temp.) MK 349 (Temp.) Temperaturechange from the initial (oc) Figure 8. Early shrinkage of concrete measured using embedded low modulus strain gages 29