High performance concrete containing lower slag amount: A complex view of mechanical and durability properties

Construction and Building Materials 23 (2009) 2237–2245

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

High performance concrete containing lower slag amount: A complex view of mechanical and durability properties Eva Vejmelková a, Milena Pavlíková a, Zbyneˇk Keršner b, Pavla Rovnaníková c, Michal Ondrácˇek d, Martin Sedlmajer d, Robert Cˇerny´ a,* a

Department of Materials Engineering and Chemistry, Faculty of Civil Engineering, Czech Technical University in Prague, Thákurova 7, 166 29 Prague 6, Czech Republic Institute of Structural Mechanics, Faculty of Civil Engineering, Brno University of Technology, Veverˇí 95, 602 00 Brno, Czech Republic c Institute of Chemistry, Faculty of Civil Engineering, Brno University of Technology, Zˇizˇkova 17, 602 00 Brno, Czech Republic d Institute of Technology of Building Materials and Components, Faculty of Civil Engineering, Brno University of Technology, Veverˇí 95, 602 00 Brno, Czech Republic b

a r t i c l e

i n f o

Article history: Received 10 November 2008 Accepted 28 November 2008 Available online 6 January 2009 Keywords: Concrete Slag Mechanical properties Fracture-mechanical properties Durability properties Hydric properties Thermal properties Chloride binding

a b s t r a c t A wide set of parameters of concrete containing 10% of ground granulated blast furnace slag as Portland cement replacement involving basic material characteristics, mechanical and fracture-mechanical properties, durability characteristics, hydric and thermal properties and chloride binding characteristics is determined and compared with the parameters of reference Portland cement concrete with otherwise the same composition. The experimental results show that the replacement of Portland cement by even such a low amount of ground granulated blast furnace slag as environmental more friendly and still valuable alternative binder either affects positively or at least does not worsen in a significant way the substantial properties of hardened concrete mix. The mechanical and fracture-mechanical properties are found to be very similar as compared to the reference mix, the liquid water transport parameters of the mix containing slag are significantly better, the basic durability characteristics such as the frost resistance and corrosion resistance similar and very good, the resistance against de-icing salts slightly worse. These findings may be significant for the future use of slag in the countries where its available amount is decreasing and its more efficient use as a binder than it was common to date can appear necessary. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction The application of slag as binding agent for composite building materials has a rather long tradition. Ground granulated blast furnace slag (GGBFS) has been used since 1940s either as a component of blended cement or partial replacement of Portland cement [1–8]. Alkali activators stimulating the latent pozzolanic properties of GGBFS made also possible its utilization as a sole binder [9–16] instead of cement. The motivations for using GGBFS in concrete production are mostly economical and ecological. Nowadays, the cement producers in Europe are supposed to meet the tightened up ecological legislative, the restrictions upon emission limits in particular. The increasing prices of fuel for burning the clinker are another burden. GGBFS is one of the ways how to deal with the increasing ecological and economical requirements. It is a waste material, thus cheaper than cement, and its use instead of a part of cement decreases the overall CO2 consumption. In addition, GGBFS increases the workability of concrete, improves strength, reduces the hydra* Corresponding author. Tel.: +420 2 2435 4429; fax: +420 2 2435 4446. ˇ erny´). E-mail address: [email protected] (R. C 0950-0618/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2008.11.018

tion heat, permeability, porosity and alkali–silica expansion [17–20] which is another bonus. The optimum amount of GGBFS is usually in the range of 40–60% of the total mass of binder; its higher content can already impair the 28-days compressive strength of hardened concrete [5,6,20]. So, a relatively high amount of cement can be safely replaced by slag which leads to a tendency to use as much slag in concrete as possible. However, the sources of GGBFS are not unlimited. In some countries, there is already shortage of high-quality slag which is necessary for concrete production. In Czech Republic, since 2008 no GGBFS is available on the free market. Cement producers make use of all slag produced in the country and still complain of its lack. Therefore, a necessity of an even more efficient use of slag as a binder than it was common to date can appear in the near future. One of the possibilities is to utilize slag for the production of high performance concrete (HPC) as it was indicated already some years ago [21]. Slag may also no longer be considered just as a waste material which is supposed to get rid of but rather a worthy binding agent improving the properties of Portland cement based concretes. This can lead to its use in lower amounts which are just necessary to obtain the denser microstructure of cement matrix; the structure compacting which accompanies the use of slag as

2238

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

partial Portland cement replacement is a consequence of the secondary pozzolanic reaction resulting in the consumption of Ca(OH)2 and C–S–H structures formation [22]. The majority of investigators was concentrated on studies of the effect of higher amount of slag on concrete properties until now, driven by the above mentioned motivation to use as much slag as possible. The analyses mostly began at 20% of the total mass of the binder. The properties of concrete with lower amount of slag as cement replacement were studied much less often so that they were not so thoroughly explored. This paper intends to contribute to the extension of knowledge of the concrete mixes containing lower amount of slag. A representative set of parameters of concrete containing 10% of slag as Portland cement replacement including basic material characteristics, mechanical and fracturemechanical properties, durability characteristics, hydric and thermal properties and chloride binding characteristics is investigated and compared with the parameters of corresponding Portland cement concrete. The set of measured parameters is sufficiently wide to make possible to perform even very complex computational service life analyses where they can be used as input data. 2. Experimental methods 2.1. Basic material characteristics The consistence of fresh concrete mixtures was analyzed by the slump test ˇ SN EN 12350-2 [23], using a conical mould (upper diameter according to C 100 mm, lower diameter 200 mm, height 300 mm). The result of the slump test (in mm) was the difference between the height of the mould and the uppermost point of the specimen after the test. As fundamental physical material characteristics, bulk density qb (kg m3), open porosity (Vol. %) and matrix density qmat (kg m3) were determined using the water vacuum saturation method [24]. Each sample was dried in a drier to remove majority of the physically bound water. After that the samples were placed into a desiccator with deaired water. During three hours air was evacuated with vacuum pump from the desiccator. The sample was then kept under water not less than 24 h. From the mass of the dry sample md, mass of water saturated sample mw, and mass of immersed water saturated sample ma, the volume V of the sample was determined from the equation

V¼

mw  ma

qw

ð1Þ

where qw is the density of water. The open porosity, bulk density and matrix density were calculated according to the equations

mw  md V qw m q¼ d V md qmat ¼ Vð1  w0 Þ w0 ¼

ð2Þ ð3Þ ð4Þ

Characterization of the pore structure of studied materials was performed by mercury intrusion porosimetry. This well known method is based on intrusion of mercury to the porous sample by gradually increasing intrusion pressure while mercury penetrates to smaller pores. The experiments were carried out using the instruments PASCAL 140 and 440 (Thermo Scientific). The range of applied pressure corresponds to pore radius from 2 nm to 2000 lm. Since the size of the specimens is restricted to the volume of approximately 1 cm3 and the studied materials contained some aggregates about the same size, the porosimetry measurements were performed on samples without coarse aggregates. The matrix density was determined by the helium pycnometry as well, for the sake of comparison. The method is based on measurement of the real volume of a sample using helium which has very small atoms easily penetrating into a porous system. The measurements were performed by Pycnomatic ATC equipment (Porotec, Germany). 2.2. Mechanical and fracture-mechanical properties The measurement of compressive strength and bending strength was done using the hydraulic testing device VEB WPM Leipzig 3000 kN having a stiff loading frame with the capacity of 3000 kN. The compressive strength was tested according ˇ SN EN 12390-3 [25]; a constant loading rate of 0.2 MPa/s was imto the standard C posed on the specimens. The bending strength was determined using the procedure described in CˇSN EN 12390-5 [26], with the loading rate of 0.04 MPa/s. The basic

tests were performed after 28 days of standard curing. The details of the specific tests where mechanical properties were used as criteria for durability assessment are given in Section 2.3. The effective fracture toughness was measured using the effective crack model [27] which combines the linear elastic fracture mechanics and crack length approaches. A three-point bending test of a specimen having a central edge notch with a depth of about 1/3 of the depth of the specimen was used in the experiment. The loaded span was equal to 300 mm. A continuous record of the load–deflection (F–d) diagram was used for the calculation of effective fracture toughness. An estimate of fracture energy was obtained from the F–d diagram according to the RILEM method (work-of fracture). 2.3. Durability tests ˇ SN 73 1322/Z1:1968 [28]. Frost resistance tests were carried out according to C The samples were tested after 28 days of concrete maturing and standard curing. The total test required 100 freezing and thawing cycles. One cycle consisted of 4 h freezing at 20 °C and 2 h thawing in 20 °C warm water. Frost resistance coefficient K was determined as the ratio of bending or compressive strength of specimens subjected to 100 freezing and thawing cycles to the strength of reference specimens which did not undergo the frost resistance test. The resistance of studied concrete against de-icing salts was measured according to CˇSN 731326/Z1:1984 [29]. The tested specimens were saturated by water and put into a bath with 3% NaCl solution. Then, freeze/thaw cycles were applied. In one cycle the tested specimen was cooled at first in an automatic conditioning device from 20 to 15 °C during 45 min, then it was left at 15 °C for 15 min, subsequently heated to 20 °C during 45 min and left 15 min at that temperature. After every 25 cycles the specimens were removed from the bath, their mass loss due to spalling of particles on the surface was determined, the NaCl solution replaced and specimens put into the bath again. The test was supposed to be finished either after the prescribed number of cycles or after the mass loss exceeded 1000 g/m2. The corrosion resistance in various environments was tested according to the procedure developed at the Brno University of Technology. The specimens were prepared in 100  100  400 mm molds and placed into a climatic chamber with 100% relative humidity environment. After 24 h they were demolded and stored in the same environment for another 27 days. Then, the specimens were cut to 100  100  50 mm blocks and put in groups of three into the corrosion environments specified in Table 1. One set of specimens was just after the 28-days curing subjected to the compressive-strength test to obtain reference strength value. Test of concrete carbonation was performed in a desiccator where the CO2 concentration was kept at 65 ± 5 vol. % (the concentration was measured by an IR probe). The carbonation took place in an environment above saturated KNO3 solution (85 ± 5% relative humidity). The specimens denoted as ‘‘air” in Table 1 were stored in common laboratory conditions at 21 ± 2 °C and 45 ± 5% relative humidity; those marked ‘‘distilled water” were in distilled-water bath which was replaced every 10 days. The duration of the corrosion test was 60 days. Then, the specimens were subjected to the compressive-strength test. The way of loading the specimens is shown schematically in Fig. 1 as the test was not quite a standard one. The coefficient of corrosion resistance Kcr was then determined as the ratio of the compressive strength after 60 days in a corrosion environment and compressive strength after 60 days in laboratory conditions. All the specimens were water-leached after the compressive-strength test and the pH value was determined by potentiometry. X-ray diffraction analysis (Bruker D8 Advance device) in the range of H-angle of 5–80° was done as well to test the possible appearance of new phases. The specimens exposed to the CO2 action were after the compressive-strength test subjected to the phenolphthalein test where 1% solution of phenolphthalein in 70% ethanol was spread on the fracture area. The violet coloring gave evidence that the pH value of the pore solution in the cement gel was higher than 9.5. 2.4. Hydric properties The wet cup method and dry cup method [24] were employed in the measurements of water vapor transport parameters. The specimens were water- and vaporproof insulated by epoxy resin on all lateral sides, put into the cup and sealed by technical plasticine. The impermeability of the plasticine sealing was achieved by heating it first for better workability and subsequent cooling that resulted in its

Table 1 Corrosion environments used in the tests. Environment

Concentration

Air Distilled water MgCl2 (g L1) NH4Cl (g L1) Na2SO4 (g L1) HCl (mol/L) CO2 (vol. %)

– – 17.76 2.97 14.79 103 65 ± 5

2239

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

lengths were measured using a contact comparator having the accuracy of 0.001 mm.

F

2.5. Thermal properties Thermal conductivity and specific heat capacity were measured using the commercial device ISOMET 2104 (Applied Precision, Ltd.). ISOMET 2104 is equipped with various types of optional probes, needle probes are for porous, fibrous or soft materials, and surface probes are suitable for hard materials such as concrete in this paper. The measurement is based on the analysis of the temperature response of the analyzed material to heat flow impulses. The heat flow is induced by electrical heating using a resistor heater having a direct thermal contact with the surface of the sample.

h

direction of compacting

2.6. Chloride binding

b a Fig. 1. Scheme of the compressive-strength experiment for the corrosion-resistance test. hardening. In the wet cup method the sealed cup containing saturated K2SO4 solution (the equilibrium relative humidity above the solution was 97.8%) was placed into an air-conditioned room with 30% relative humidity and weighed periodically. The measurements were done at 25 ± 1 °C in a period of 4 weeks. The steady state values of mass loss determined by linear regression for the last five readings were used for the determination of water vapor diffusion coefficient. In the dry cup method the sealed cup containing dried CaCl2 (the equilibrium relative humidity above the desiccant was 5%) was placed in an air-conditioned room with 30% relative humidity. Otherwise, the measurement was done in the same way as in the wet cup method. The water vapor diffusion coefficient D (m2 s1) was calculated from the measured data according to the equation

D¼

Dm  d  R  T S  s  M  D pp

ð5Þ

where Dm (kg) is the amount of water vapor diffused through the sample, d (m) the sample thickness, S (m2) the specimen surface, s (s) the period of time corresponding to the transport of mass of water vapor Dm, Dpp (Pa) the difference between partial water vapor pressure in the air under and above specific specimen surface, R the universal gas constant, M the molar mass of water, T (K) the absolute temperature. On the basis of the diffusion coefficient D, the water vapor diffusion resistance factor l was determined

l¼

Da D

ð6Þ

where Da is the diffusion coefficient of water vapor in the air. The liquid water transport was characterized by the water absorption coefficient and apparent moisture diffusivity. The specimen was water- and vapor-proof insulated on four lateral sides and the face side was immersed 1–2 mm in the water. Constant water level in the tank was achieved by a Marriott bottle with two capillary tubes. One of them, inside diameter 2 mm, was ducked under the water level. The second one, inside diameter 5 mm, was above water level. The automatic balance allowed recording the increase of mass. The water absorption coefficient A (kg m2 s1/2) was calculated using the formula

i¼A

pffiffi t

japp 

A wc  w0

2 ð8Þ

where wc (kg m3) is the saturated moisture content and w0 (kg m3) the initial moisture content.The hygric strain eu () was determined using a comparative technique according to the equation

eu ¼

Dl l0;u

Cb ¼

M Cl VðC 0  C 1 Þ W

ð10Þ

where MCl is the molar mass of chlorine (g/mol), V the volume of the solution (ml), C0, C1 the initial and equilibrium concentrations, respectively, of chloride solution (mol/l), and W the mass of the dry sample (g), which can be calculated from the difference in mass of the sample dried in a desiccator at 11% relative humidity and in an oven at 105 °C. The free chloride content Cf (mol/l), corresponding to the value of Cb calculated from Eq. (10) was given by

Cf ¼ C1

ð11Þ

By performing the experiment with different values of the initial salt concentration C0, a point wise function Cb = Cb(Cf) can be obtained, which is the ion binding isotherm. The possible shortage of the method of Tang and Nilsson [31] is that the original measuring procedure is developed for crushed samples of cement paste and it is assumed that the content of cement is the main criterion for a possible recalculation to real concrete or cement mortar. In fact, the method is certain idealization of the

Table 2 Chemical composition of cement. Component

Amount (%)

SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3

21.89 5.60 3.75 62.33 1.04 0.92 0.11 0.30 0.17 2.88

ð7Þ

where i (kg/m2) is the cumulative water absorption, t the time from the beginning of the water absorption experiment.The water absorption coefficient was then used for the calculation of the apparent moisture diffusivity in the form [30]



Chloride adsorption isotherms were measured by a modification of the method by Tang and Nilsson [31] based on the adsorption from solution. In the original version of the method a crushed sample of cement mortar or cement paste dried at 11% relative humidity was put into a cup, then the cup was vacuumed in a desiccator for 2 h, before being filled with a specific concentration NaCl solution saturated with Ca(OH)2. The volume of the solution inside the cup was calculated from the increment of the mass of the cup and the density of the solution. The cup was covered and stored at 20 °C to reach equilibrium. Adsorption equilibrium was typically achieved after 7 days for 25 g samples. Then the inside solution was pippetted to determine the chloride concentration by potentiometric titration using 0.01 N AgNO3 and a chloride selective electrode. The bound chloride content Cb (mg/g) was calculated from the equation

ð9Þ

where l0,u is the length at the reference moisture content, Dl the difference between the length at actual moisture content and length at reference moisture content. The

Table 3 Chemical composition of slag. Component

Amount (%)

SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O SO2 4

34.74 5.91 0.39 40.27 9.60 0.405 0.288 0.39

2240

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

binding problem assuming that chlorides can get in direct contact with every small grain of hydrated cement. However, in a real concrete specimen (even in cement paste) the interior pore surface where the cement binder is accessible to the chlorides is certainly smaller than the total surface of a crushed specimen. The chloride binding capacity can be then affected by many other factors such as the change in the porous structure and pore distribution due to the application of different aggregates, the presence of various admixtures etc. Therefore, the result obtained by the Tang and Nilsson method can be considered as a certain upper limit to the real chloride binding capacity. The modification of the Tang and Nilsson adsorption method used in this paper consisted in using the specimens of more realistic dimensions instead of crushed specimens [32].

3. Materials and samples The high performance concrete mix studied in the paper was prepared with Portland cement CEM I 42.5 R (its chemical composition is shown in Table 2, its specific surface area was 341 m2/kg) as the main binder. A part of cement was replaced by ground granulated blast furnace slag (Kotoucˇ Štramberk, Ltd., CZ). The chemical composition of slag is shown in Table 3, its grain size distribution in Table 4. The specific surface area of slag was 380 m2/kg. The phase analysis showed the presence of merwinite Ca3Mg(SiO4)2, akermanite Ca2MgSi2O7 and augite (Ca,Na)(Mg,Fe,Al,Ti)(Si,Al)2O6. The composition of the HPC mix (denoted as BS) which was studied in all the experiments in this paper is shown in Table 5. The slag was used in the amount of 10% of the mass of cement (the reasons for this choice were explained in the Introduction section). For the sake of comparison, also a reference mix BR with only Portland cement as the binder but all other components the same as in BS was investigated. The total mass of binder in the reference mix was the same as in BS (Table 5). The measurement of all parameters described in Section 2 took place in a conditioned laboratory at the temperature of 22 ± 1 °C and 25–30% relative humidity. The following specimens were used in the experiments: basic physical properties – 6 specimens 50  50  25 mm, freeze/thaw resistance, de-icing salts resistance – 3 specimens 100  100  400 mm, corrosion resistance – 3 specimens 100  100  50 mm, bending strength – 3 specimens 100  100  400 mm, compressive strength – 3 specimens 150  150  150 mm and 6 specimens 100  100  100 mm (remainders after the bending test), fracture-mechanical properties – 3 specimens 100  100  400 mm, water vapor transport properties – 6 specimens 50  50  25 mm, water transport properties – 5 specimens 50  50  20 mm, thermal properties – 3 specimens 70  70  70 mm, chloride binding isotherms – 2 specimens 40  40  10 mm.

4. Experimental results and discussion 4.1. Basic material characteristics The slump of the tested concrete mix BS was 130 mm which was lower than for the reference mix BR (150 mm). However, the workability was still satisfactory. Table 6 shows that the open porosity of the BS mix measured by the water vacuum saturation method was 20% lower than for the reference mix. The results of XRD analysis in Figs. 2 and 3 (the lowest curves) reveal that the amount of Ca(OH)2 in the BS mix was significantly lower than in the reference BR mix. This means that the slag pozzolanic reaction resulted in the consumption of a part of the originally formed Ca(OH)2 and formation of additional C–S–H structures which partially filled the former pore space. The matrix density of BS decreased by 4%, however, so that after the pozzolanic reaction it contained lower-density components than the matrix of the BR mix produced using only Portland cement. The bulk densities of both mixes were very similar; the 2% difference is on the edge of the error range of the experimental method. The matrix densities measured by helium pycnometry (Table 7) agreed within an approximately 2% margin with the values measured by the water vacuum saturation method (Table 6). The global pore space parameters determined by mercury porosimetry (Table 7) showed that the total pore volume of the BS mix was about 20% lower as compared to the reference mix BR. This is in a very good agreement with the water vacuum saturation measurements. The bulk density of the gel (the porosimetric measurements were performed on samples without coarse aggregates) was lower than that of the whole samples but it was an expected outcome [22]. The open porosities calculated using the total pore volume and bulk density agreed within a 5% margin with the water vacuum saturation measurements. Fig. 4 shows that the majority of pores (60– 70% by volume) was in the range of 10–100 nm for both materials. Another important pore range (15–20% by volume) was 100 nm to 1 lm. The amount of other-size pores was much lower. These findings indicate that both materials were mixed successfully to obtain not only high strength concrete but also high performance concrete with supposed good durability properties. The decrease of porosity due to the replacement of 10% of Portland cement by GGBFS was mainly due to the reduction of the volume of pores in the range of 10–100 nm which indicated compaction of concrete microstructure. This compaction is a typical result of pozzolanic reactions leading to additional C–S–H structures formation [8,33,34]. For higher GGBFS dosage it was observed for instance in a SEM analysis in [35]. 4.2. Mechanical and fracture-mechanical properties

Table 4 Slag granulometry. Sieve residue 0.045 mm (%) 12.4

0.09 mm (%) 1.9

The basic compressive strength test for 150  150  150 mm specimens after 28 days (Table 8) showed for the BS mix approximately 3% lower strength value than for the reference mix. However, both materials safely met the basic criterion of the compressive strength of 60 MPa to be considered high performance materials. The compressive strengths determined on the fragments of specimens after the bending test were for both BS

Table 5 Composition of studied concretes (kg m3). Component

BR

BS

CEM I 42.5 R Mokrá Aggregates 0–4 mm Aggregates 8–16 mm Plasticizer Mapei Dynamon SX Ground granulated blast furnace slag Water

484 812 910 5.3 – 188

440 812 910 5.3 44 188

Table 6 Basic physical properties of studied concretes measured by water vacuum saturation method. Material

Bulk density (kg m3)

Matrix density (kg m3)

Open porosity (% m3 m3)

BR BS

2380 2334

2715 2602

12.3 9.7

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

2241

35

Fig. 2. X-ray diffraction analysis of BS specimen.

and BR almost the same and about 10% higher than for the cubic specimens. This was due to the size effect; the specimens’ area was larger (approximately 200  100 mm) than the actual loading area of 100  100 mm. The bending strengths differed within the range of 5%, with BS achieving lower value. So, the replacement of Portland cement by 10% of GGBFS did not affect the mechanical properties of the studied concrete mix in a significant way; the differences were rather low, on the edge of the error range of the experimental methods. These results are in a qualitative agreement with the experimental work done by other investigators for higher (up to 40–50% of Portland cement replacement) GGBFS dosage where in some measurements the mixes with slag achieved slightly higher 28-days strength values [4,8,20], in others slightly lower values [36,37] than for the reference mixes. The effective fracture toughness and fracture energy of the analyzed mix BS (Table 9) were almost the same as for the reference mix BR; its effective toughness value was about 15% lower. In general, the values of fracture parameters were quite high for both materials; the analyzed composites were very tough which was promising from the point of view of their expected durability. 4.3. Durability tests Table 10 shows that both the analyzed mix BS and the reference material BR could be characterized as frost resistant. The frost resistance coefficient K calculated as the ratio of both compressive and bending strengths was the same for BS and BR and always higher than 0.75. The reference mix BR achieved better resistance

against de-icing salts than the main investigated mix BS (Fig. 3). However, both BR and BS were safely within the limit of 1000 g m2 after 100 cycles defined in [29] as satisfactory de-icing salts resistance. The compressive strength after 28-days curing and 60 days in laboratory conditions defining the relative value of the corrosion resistance coefficient Kcr = 1 was for BR 93.6 MPa, for BS 92.1 MPa. This gives evidence of the continuing strength increase of both materials also after 28 days of curing (cf. the data in Table 8). However, the rate of this increase was the same for both materials so that the pozzolanic effect of using slag as partial Portland cement replacement was not clearly manifested, contrary to the properties of mixes with higher slag content [4,8,20,36]. The corrosion resistance of the BS mix was in distilled water, Na2SO4 and HCL better, in MgCl2, NH4Cl and CO2 worse in a comparison with BR but the differences were not very high (Table 11). The pH values of water leaches after 60 days in corrosion environments were for both BR and BS similar (Table 12), for BS slightly lower than for BR except for CO2. The results of XRD analysis in Figs. 2 and 3 did not show any corrosion products for both BS and BR and all studied environments. This means that both materials were very compact; the corrosive substances could react just on the surface and could not penetrate deeper into the porous structure. 4.4. Hydric properties The results of measurements of water vapor diffusion resistance factor l of the analyzed composites are presented in Table 13.

2242

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

35

Fig. 3. X-ray diffraction analysis of BR specimen.

Comparing the data measured for the two studied materials, we can see that the l values of BR were in the ranges of both higher and lower relative humidity lower than for BS. This agrees in a qualitative way with the porosity results in Tables 6 and 7 and Fig. 4. The water absorption coefficient and apparent moisture diffusivity of the material BS (Table 14) were almost two times lower than for BR. This is again in a good qualitative agreement with the porosimetry measurements; the material with lower open porosity achieved the lower values of liquid water transport parameters. As the low liquid water transport parameters are the most important from the point of view of achievement of good durability because harmful substances can penetrate slower into the porous structure, the partial replacement of Portland cement even by an amount as low as 10% of slag can be considered as a successful solution. Reduction of water- and chloride permeability was observed also for higher (up to 40–50% of Portland cement

Table 7 Basic physical properties of studied concretes measured by helium pycnometry (matrix density) and mercury porosimetry (all other parameters). Material

Matrix density (kg m3)

Pore volume (mm3 g1)

Pore surface (m2 g1)

Bulk density (kg m3)

Open porosity (% m3 m3)

BR BS

2651 2604

59.4 46.8

3.0 2.5

2060 2190

12.2 10.2

50 45 BR BS

35

3

Pore volume [mm /g]

40

30 25 20 15 10 5 0 0.001-0.01 0.01-0.1

0.1-1

1-10

10-50

50-100

100-2000

-6

Pore radius [10 m] Fig. 4. Pore distribution of studied concretes.

Table 8 Mechanical properties of studied concretes. Material

Compressive strength – 150  150  150 mm (MPa)

Compressive strength – fragments after the bending test, loading area 100  100 mm (MPa)

Bending strength (MPa)

BR BS

77.4 75.0

85.2 84.7

12.9 12.3

2243

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

replacement) GGBFS dosage [35,38–41]. The results obtained in this paper showed that a lower slag amount can be effective as well in that respect (see Fig. 5).

Table 9 Fracture mechanical properties of studied concretes. Material

Effective fracture toughness (MPa m1/2)

Effective toughness (N m)

Specific fracture energy (J m2)

BR BS

1.97 2.02

85 72

381 369

Fig. 6 shows that higher hygric strain exhibited the material BS containing slag but the difference was not very high, typically 10%. The hygric strain vs. moisture content functions of both materials had concave character. So, the hygric expansion coefficient decreased with increasing moisture content. This finding is in a general agreement with the basic physical considerations on the interaction of water molecules with the porous matrix of concrete. The highest hygric strains are typical for lower moisture content when the van der Waals forces between the water molecules and C–S–H structures of the concrete matrix are responsible for the formation of the surface phase of water [22].

Table 10 Frost resistance of studied concretes.

BR BS

BS

As the ratio of compressive strengths ()

As the ratio of bending strengths ()

1 1

0.8 0.8

BR

Air Distilled water MgCl2 NH4Cl Na2SO4 HCl CO2

500 400 300 200 100

Table 11 Coefficient of corrosion resistance Kcr of studied concretes in various environments. Environment

BR

600

Frost resistance coefficient K

Loss of mass [g m-2]

Material

700

0 25

BS

1.00 0.85 0.84 0.87 0.88 0.83 1.21

50

75

100

Freeze/thaw cycles

1.00 0.98 0.82 0.84 0.93 0.96 1.17

Fig. 5. Loss of mass of studied concretes due to the de-icing salts action.

6.E-04

Table 12 Measured pH values of water leaches of studied concretes submerged for 60 days in various environments. Environment

BR

BS

Air Distilled water MgCl2 NH4Cl Na2SO4 HCl CO2

12.33 12.40 12.44 12.27 12.40 12.40 11.82

12.20 12.15 12.18 12.02 12.04 11.99 11.97

Hygric strain [-]

5.E-04

4.E-04

3.E-04

2.E-04

BR BS

1.E-04

0.E+00 0.000

0.010

0.020

0.030

0.040

0.050

0.060

Moisture content [kgkg-1] Fig. 6. Hygric strain of studied concretes.

Table 13 Water vapor transport properties of studied concretes. 2.5

BR BS

97/25%

5/25%

Water vapor diffusion coefficient (m2 s1)

Water vapor diffusion resistance factor ()

Water vapor diffusion coefficient (m2 s1)

Water vapor diffusion resistance factor ()

3.63E06 2.61E06

6.60 8.99

1.50E06 1.31E06

15.8 17.7

Table 14 Water transport properties of studied concretes. Material

Water absorption coefficient (kg m2 s1/2)

Apparent moisture diffusivity (m2 s1)

BR BS

0.0099 0.0057

7.15E09 3.77E09

Thermal conductivity [W m-1K-1]

Material

2

1.5

BS BR

1 0

1

2

3

4

Moisture content [%kgkg-1] Fig. 7. Thermal conductivity of studied concretes.

5

2244

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245

4.5. Thermal properties

50

BR

Bound chlorides [mg/g-sample]

The thermal conductivity of the material BS (Fig. 7) was in the whole range of moisture content approximately 10–15% higher than for the reference mix BR which is in an agreement with its lower porosity. The specific heat capacity of BR (Fig. 8) was about 5–10% lower than for BS in the whole range of moisture content. However, the observed differences were within the error range of the measuring method being ±10%. The values of thermal diffusivity (Fig. 9) were very similar for both materials in whole the moisture range. This reflects the mutual compensation of the differences in thermal conductivity and specific heat capacity of the investigated materials.

40 BS 30

20

10

0

4.6. Chloride binding

0

10000

20000

30000

40000

Free chlorides [mg/l-solution]

The chloride binding capacity of the material BS (Fig. 10) was 5– 15% higher as compared to the reference mix BR. The differences were more remarkable in the range of high chloride concentration. The higher chloride binding capability of BS may be related to the slower course of pozzolanic reaction in comparison with Portland cement hydration. The chloride binding test was taking much more time than the other types of tests, approximately 6 months. Therefore, the pozzolanic reaction might already be almost completed after that time – contrary to the measurements of most other parameters which were done 28 days after mixing.

-1

Specific heat capacity [J kg-1K ]

850

800

750

700 BS BR

650

600 0

1

2

3

4

5

Moisture content [%kgkg-1] Fig. 8. Specific heat capacity of studied concretes.

1.2

-6

2 -1

Thermal diffusivity [10 m s ]

1.4

Fig. 10. Chloride binding isotherms of studied concretes.

5. Conclusions The experimental results presented in this paper showed that the replacement of Portland cement by a lower amount of ground granulated blast furnace slag as environmental more friendly and still very valuable alternative binder either affected positively or at least did not worsen the substantial properties of hardened concrete mix. With the same water/binder ratio, the porosity of the BS mix containing slag was 20% lower than of the composite containing only Portland cement. This had a positive effect on the durability properties of the mix. The liquid water transport parameters decreased to almost one half of the values measured for the reference Portland cement mix BR which is a very substantial improvement because the rate of water transport is the critical parameter for penetration of harmful substances into the bulk material. Water vapor transport parameters of the BS mix were also lower than of the reference BR mix. The frost resistance and corrosion resistance was for BS similarly high as for BR, the resistance against de-icing salts was for BS worse than for BR but it still safely met the basic durability criteria. The mechanical and fracture-mechanical properties of both BS and BR achieved high values typical for high performance concrete and were very similar, mostly within the error range of the experimental methods. The thermal properties differed typically in the range of 5–15% and followed the difference in porosity. Taking into account the results summarized above, it can be concluded that the replacement of Portland cement by GGBFS can be considered an effective solution even if the amount of slag is much lower than the usual 40–60%. This may be significant for the future use of slag in the countries where its available amount is decreasing and its more efficient use as a binder than it was common to date can appear necessary. Acknowledgement

1

This research has been supported by the Czech Science Foundation, under Project No 103/07/0034. BS

0.8

References

BR

0.6 0

1

2

3

Moisture content [%kgkg-1] Fig. 9. Thermal diffusivity of studied concretes.

4

5

[1] Roy DM, Idorn GM. Hydration, structure, and properties of blast furnace slag cement, mortars, and concrete. ACI J 1982;79:444–57. [2] Swamy RN, Bouikni A. Some engineering properties of slag concrete as influenced by mix proportioning and curing. ACI Mater J 1986;87:551–76. [3] Miura T, Iwaki I. Strength development of concrete incorporating high levels of ground granulated blast-furnace slag at low temperatures. ACI Mater J 2000;97:66–70.

E. Vejmelková et al. / Construction and Building Materials 23 (2009) 2237–2245 [4] Khatib JM, Hibbert JJ. Selected engineering properties of concrete incorporating slag and metakaolin. Constr Build Mater 2005;19:460–72. [5] Guo LP, Sun W, Zheng KR, Chen HJ, Liu B. Study on the flexural fatigue performance and fractal mechanism of concrete with high proportions of ground granulated blast-furnace slag. Cement Concr Res 2007;37:242–50. [6] Oner A, Akyuz S. An experimental study on optimum usage of GGBS for the compressive strength of concrete. Cement Concr Compos 2007;29:505–14. [7] Hale WM, Freyne SF, Bush Jr TD, Russell BW. Properties of concrete mixtures containing slag cement and fly ash for use in transportation structures. Constr Build Mater 2008;22:1990–2000. [8] Chidiac SE, Panesar DK. Evolution of mechanical properties of concrete containing ground granulated blast furnace slag and effects on the scaling resistance test at 28 days. Cement Concr Compos 2008;30:63–71. [9] Douglas E, Bilodeau A, Malhotra VM. Properties and durability of alkaliactivated slag concrete. ACI Mater J 1992;89:509–16. [10] Wang SD, Pu XC, Scrivener KL, Pratt PL. Alkali-activated slag cement and concrete: a review of properties and problems. Adv Cement Res 1995;7:93–102. [11] Roy DM. Alkali-activated cements – opportunities and challenges. Cement Concr Res 1999;29:249–54. [12] Bakharev T, Sanjayan JG, Cheng YB. Effect of admixtures on properties of alkaliactivated slag concrete. Cement Concr Res 2000;30:1367–74. ˇ erny´ R. Properties of alkali activated [13] Zuda L, Pavlík Z, Rovnaníková P, Bayer P, C aluminosilicate material after thermal load. Int J Thermophy 2006;27:1250–63. ˇ erny´ R. Thermal properties of alkali activated [14] Zuda L, Rovnaník P, Bayer P, C slag subjected to high temperatures. J Build Phy 2007;30:337–50. [15] Pacheco-Torgal F, Castro-Gomes J, Jalali S. Alkali-activated binders: a review. Part 1. Historical background, terminology, reaction mechanisms and hydration products. Constr Build Mater 2008;22:1305–14. ˇ erny´ R. Mechanical and hydric properties of [16] Zuda L, Bayer P, Rovnaník P, C alkali-activated aluminosilicate composite with electrical porcelain aggregates. Cement Concr Compos 2008;30:266–73. [17] Bijen J. Benefits of slag and fly ash. Constr Build Mater 1996;10:309–14. [18] Aldea CM, Young F, Wang K, Shah SP. Effects of curing on properties of concrete using slag replacement. Cement Concr Res 2000;30:465–72. [19] Park CK, Noh MH, Park TH. Rheological properties of cementitious materials containing mineral admixtures. Cement Concr Res 2005;35:842–9. [20] Atis CD, Bilim C. Wet and dry cured compressive strength of concrete containing ground granulated blast-furnace slag. Build Environ 2007;42:3060–5. [21] Shi C, Qian J. High performance cementing materials from industrial slags — a review. Resources. Conserv Recycl 2000;29:195–207. ˇ erny´ R, Rovnaníková P. Transport processes in concrete. London: Spon Press; [22] C 2002.

2245

ˇ SN EN 12350-2. Testing of fresh concrete – Part 2: slump test. Prague: Czech [23] C Standardization Institute; 2000. [24] Roels S, Carmeliet J, Hens H, Adan O, Brocken H, Cˇerny´ R, et al. Interlaboratory comparison of hygric properties of porous building materials. J Therm Envelope Build Sci 2004;27:307–25. ˇ SN EN 12390-3. Testing of hardened concrete – Part 3: compressive strength. [25] C Prague: Czech Standardization Institute; 2002. ˇ SN EN 12390-5. Testing of hardened concrete – Part 5: bending strength. [26] C Czech Standardization Institute, Prague; 2007. [27] Karihaloo BL. Fracture mechanics of concrete. New York: Longman Scientific & Technical; 1995. ˇ SN 73 1322/Z1:1968. Concrete testing – hardened concrete – frost resistance. [28] C Prague: Czech Standardization Institute; 2003. ˇ SN 731326/Z1:1984. Determination of the resistance of the surface of [29] C concrete against water and de-icing salts. Prague: Czech Standardization Institute; 2003. [30] Kumaran MK. Moisture diffusivity of building materials from water absorption measurements. J Therm Envelope Build Sci 1999;22:349–55. [31] Tang L, Nilsson LO. Chloride binding capacity and binding isotherms of OPC pastes and mortars. Cement Concr Res 1993;23:247–53. ˇ erny´ R. Chloride binding in building materials. J Build Phys [32] Jirˇicˇková M, C 2006;29:189–200. [33] Swamy RN. Cement replacement materials concrete technology and design, vol. 3. London: Surry University Press; 1986. [34] Bentz DP. Influence of water–cement ratio on hydration kinetics: simple models based on spatial considerations. Cement Concr Res 2006;36:238–44. [35] Binici H, Aksogan O, Gorur EB, Kaplan H, Bodur MN. Performance of ground blast furnace slag and ground basaltic pumice concrete against seawater attack. Constr Build Mater 2008;22:1515–26. [36] Bougara A, Lynsdale C, Ezziane K. Activation of Algerian slag in mortars. Constr Build Mater 2007;23:542–7. [37] Oh BH, Cha SW, Jang BS, Jang SY. Development of high-performance concrete having high resistance to chloride penetration. Nucl Eng Design 2002;212:221–31. [38] Roy DM, Jiang W, Silsbee MR. Chloride diffusion in ordinary, blended, and alkali-activated cement pastes and its relation to other properties. Cement Concr Res 2000;30:1879–84. [39] Leng F, Feng N, Lu X. An experimental study on the properties of resistance to diffusion of chloride ions of fly ash and blast furnace slag concrete. Cement Concr Res 2000;30:989–92. [40] Yeau KY, Kim EK. An experimental study on corrosion resistance of concrete with ground granulate blast-furnace slag. Cement Concr Res 2005;35:1391–9. [41] Cheng A, Huang R, Wu JK, Chen CH. Influence of GGBS on durability and corrosion behavior of reinforced concrete. Mater Chem Phys 2005;93: 404–11.