Metakaolin-based inorganic polymer composite: Effects of fine aggregate composition and structure on porosity evolution, microstructure and mechanical properties

Cement & Concrete Composites 53 (2014) 258–269

Contents lists available at ScienceDirect

Cement & Concrete Composites journal homepage: www.elsevier.com/locate/cemconcomp

Metakaolin-based inorganic polymer composite: Effects of fine aggregate composition and structure on porosity evolution, microstructure and mechanical properties Elie Kamseu a,⇑, Maria Cannio a, Esther A. Obonyo b, Fey Tobias c, Maria Chiara Bignozzi d, Vincenzo M. Sglavo e, Cristina Leonelli a a

Department of Materials and Environmental Engineering, University of Modena and Reggio Emilia, Via Vignolese 905, 41125 Modena, Italy M. Rinker School of Building and Construction, University of Florida, RNK304/115703, FL32611-5703 Gainesville, FL, USA c Department Werkstoffwissenschaften Lehrstuhl fur Glas Und Keramik, Universtat Erlangen-Nurnberg, Martensstr. 5, D-91058 Erlangen, Germany d Department of Civil, Environmental and Materials Engineering, University of Bologna, Via Terracini 28, 40131 Bologna, Italy e Department of Materials Engineering and Industrial Technologies, University of Trento, Via Mesiano 77, 38050 Trento, Italy b

a r t i c l e

i n f o

Article history: Received 6 September 2012 Received in revised form 18 June 2014 Accepted 10 July 2014 Available online 21 July 2014 Keywords: Inorganic polymer Microstructure Pore size distribution Fines Flexural strength

a b s t r a c t This paper examines the phase transformation, pore evolution, microstructural and mechanical changes that occur in inorganic polymer cement (IPC) in the presence of three different grade of fine aggregates (a < 100 lm) of ladle slag, nepheline syenite and quartz sand. Experimental results indicate that polycondensation was enhanced in nepheline syenite based specimens, compared to quartz sand, due to the increase in HAMAAAS phases in relation to the dissolution and interaction of amorphous/disordered fraction of aggregates. HACAS and HACAAAS with HAMAAAS phases were identified in the ladle slag based specimens. The formation of these new phases reduced both the cumulative pore volume and pores size. The apparent increase in volume of capillary pores in ladle slag based specimens was explained by the residual bubbles from the carbonates included in raw slag. The flexural strength of the inorganic polymer cement increases from 4 MPa to 4.2, 4.8 and 6.8 MPa with the addition of 20 wt% of quartz sand, nepheline syenite and ladle slag respectively. These values increase significantly between 28 and 180 days of curing (9.1 MPa for ladle slag and 10.0 MPa for nepheline syenite). It was concluded that fines can be used to remove the HM and poorly bounded alumina oligomers in metakaolin based inorganic polymer matrices and improve the interfacial zone for the design of an optimum grade and highperformance composites. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Reactions between strong alkaline solution and amorphous to metastable aluminosilicate materials produce an inorganic binder, which is also known as inorganic polymer cement (IPC) or geopolymers [1]. These reactions have being investigated for the development of alternatives materials to conventional cements and concretes based on klinker. Researchers and industries are motivated by the global warming concerning the carbon dioxide emission derived from klinkerization process combined to its the high energy consumption [1]. Metakaolin [2], industrial waste, industrial by-products, including coal fly ash and metallurgical slag [3], and natural volcanic ash [4] based IPC have been projected as ⇑ Corresponding author. Tel.: +39 0592056230; fax: +39 0592056243. E-mail address: [email protected] (E. Kamseu). http://dx.doi.org/10.1016/j.cemconcomp.2014.07.008 0958-9465/Ó 2014 Elsevier Ltd. All rights reserved.

alternative binders due to the amorphous content of the respective aluminosilicate easily dissolvable in high alkaline solution. An ideal binder should have a pH, reactivity and microstructure that enable the design of dense, stable and high strength mortar and concrete. White et al. [5] indicated that the subtle structural changes, observed up to 17 h after the mix of solid aluminosilicate with alkaline solution, are predominantly related to the dissolution of the initial metakaolin precursor prior to the formation of gel. When considering IPC and concretes, this dissolution would include not only metakaolin (or fly ash, . . .) but also the amorphous fraction of aggregates used. In the presence of soluble silica, reactive aluminum is able to substitute for silicon in many of the oligomers anions that occurs and displays more polymerized structures [2]. Aggregates have in their structure various allotropic forms of silica that directly affect their behavior in mortar or concrete. In a highly

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

259

Nomenclature H M A

H2O Na or K Al2O3

concentrated alkali media (case of the pore solution), the amorphous silica is highly reactive compared to metastable silica, b and a-cristobalite and quartz. From the foregoing, it is clear that the alkali content of the binder as well as the amount of aggregates, the nature of the silica in the aggregate and the total proportion of the amorphous or metastable phases as their distribution will govern the interface behavior between the IPC paste and the aggregate and by the way the design of the durable and high strength mortar and concrete. Moreover, It is recognized that the presence of sodium silicate in the bulk composition of inorganic polymer cements is one of the preoccupation for this new class of materials to be defined as ‘‘green’’ [6,7]. Successful formulations can be design with fine aggregates capable to easily react with alkalis and form gel type binder which can completely or partially replace the sodium silicate and improve the sustainability of the final porous matrices [6,7]. During the dissolution/polycondensation the cross-linking and densification is enhanced by the residual amorphous silica present in the matrix and its transformation in HAMAAAS gel phase, acting as bonds. This gel is characterized by colloidal-sized, globular units closely bonded together at their surfaces, whereas the high silica concentration in a silicate-activated system enables a more homogeneous gelation process throughout the interparticle volume [8,9]. The silica to alumina ratio (in the amorphous or reactive fraction of solid precursors) and the alkali to alumina ratio (in the pore solution) directly affect the engineering properties as the stability of geopolymer cement. It is then evident that several groups of geopolymer systems should be considered. Sagoe-Crentsil et al. [10,11] classified geopolymer pastes with low and high Si/Al. When the Si/Al is low (<1), it is expected that the excess of aluminate oligomers formed will combine with the soluble silica at the surface of aggregate. New HAMAAAS phases should be formed alongside the alkaline silicate gels resulting from the reaction of the residual alkali with silica from aggregate. The additional HAMAAAS phases and the alkaline silicate gel formed could promote efficient packing and crosslinking between IPC and aggregate, developing a continuous and homogeneous strengthening phase at the interfacial zone. Dissolved silica from the surface of aggregates can then react easily with the gel modifying their structure and probably acting on the existing pores (meso and macro) and phases. Larger capillary pores from the processing of inorganic polymer cement are generally filled with residual alkali, hence the presence/addition of fine aggregates, also known as fines, will provide materials to form binder phases capable to close those pores. The grow up of additional phases, able to pack more closely against the aggregate surface, will reduce the width between the paste and will contribute to better stress transfer between aggregate and inorganic polymer cement paste. It is our main objective to study the properties of inorganic polymer cement with ground aggregates (fines) with the aim to see the effectiveness of the addition of fines to improve the mechanical properties of IPC composites. Grounded silica sand, nepheline syenite and calcium-rich aluminosilicate-namely ladle slag from steel industry-aggregates have been here used with particles size below 100 lm. The aggregates content was varied so that to understand the extent of their dissolution and polyconden-

S C

SiO2 CaO

sation in inorganic polymer matrix. The calcium hydrate is the major binding phase within Portland cement system that could be compared to HAMAAAS of inorganic polymer cement system. Further comparison present MAH to substitute CAH since MAH is generally observed as residual product of hydration of geopolymer. When in contact with aggregate, the geopolymer paste will firstly be available with its fluid highly concentrated in MAH. It is expected that the behavior of the system MAH and fine aggregate will provide more homogeneous and stable matrix capable to improve the interface behavior of concrete or mortar. The interfacial zone in binders, mortars and concretes can be improved by monitoring the bulk chemical composition and physico-mechanical parameters (particles size and distribution, compressive compaction and diffusion phenomenon). These investigations contribute to a comprehensive understanding of the possibility to improve the interfacial zone between inorganic polymer cement and aggregates, which is important for the design and production of dense, resistant and durable composites (mortars and concretes). Determination of the effects of phases transformation on the volume and spatial distribution of pores, the mechanical properties and microstructure constitute the main objective of the study. The empirical data were generated using mechanical testing, Environmental Scanning Electronic Microscopy (ESEM), Mercury Intrusion Porosimetry and Microtomography. Improved understanding of IPC-Fines systems is essential for their use in developing high performance binders, mortars and concretes. 2. Materials and experimental procedures 2.1. Materials and preparation Metakaolin (MK) was used as amorphous aluminosilicate for the production of inorganic polymer cement, obtained by the thermal treatment at 700 °C for 6 h of a standard kaolin from Cameroon [12]. MK was milled to obtain particle with size below 60 lm; the surface area value of the resultant powder was 37 m2/g (Brunauer–Emmet–Teller, method by nitrogen absorption on a Micormetitics GEMINI 2360 instrument).

Fig. 1. Particle size distribution of the 3 different fines: ladle slag, quartz sand, nepheline syenite.

260

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

Three aggregates were collected and ground finely (/ < 100 lm) to produce fines: (i) quartz-sand, from Concretes Industry, Bologna, Italy; (ii) nepheline syenite type of feldspathic sand from MIPROMALO, Cameroon; and (iii) calcium-rich ground ladle slag from steel making plant ‘‘Accieria di Rubiera’’, Reggio Emilia, Italy. The fines were prepared by grinding the respective aggregates to a particle size less than 100 lm, the D50 diameter varied from .5 to 15 lm (Fig. 1). The surface area of the three fines, determined as described above, were 51.22 m2/g, 21.98 m2/g and 12.45 m2/g respectively for ladle slag, nepheline syenite and quartz sand. Table 1 reports the chemical composition and mineral phases of these raw materials as obtained by XRF. Alkaline solution used as activator was prepared by mixing a 8 M solution of sodium hydroxide (NaOH, 99.9%, Sigma–Aldrich) and potassium (KOH, 98.9%, Sigma–Aldrich) hydroxide with sodium silicate solution (SiO2/Na2O = 3.1, L.O.I = 60 wt%, Ingessil, Verona, Italy). The volume ratio of the NaOH:KOH:Na2SiO3 was 1:1:2. The mixed NaOH and KOH solutions were prepared 48 h before their addition to the solid aluminosilicate (MK). Alkaline solution was added to MK powder with a L/S (liquid to solid ratio) of 0.36. Metakaolin based geopolymer (GPM) was obtained as inorganic polymer cement paste with Si/Al and Na/Al molar ratios of 1.23 and 0.88 respectively. The obtained paste was ball-milled in porcelain jar for 10 min then fines were added and stirred to obtain perfect homogeneous matrices. GC20, GC40 and GC60 represent the specimens of inorganic polymer composites with 20, 40 and 60 wt% of calcium-rich fines, similarly geopolymers added with 20, 40 and 60 wt% of quartz sand based fines were indicated as GS20, GS40 and GS60 and when 20, 40 and 60 wt% of nepheline syenite fines were added, GF20, GF40 and GF60 were obtained. The different viscous pastes obtained were poured in TeflonÒ molds with dimensions of 140 mm  10 mm  10 mm and sealed in plastic bag for the first 72 h. All the specimens, tested after 28 days, already presented a constant weight after the third week of curing at ambient temperature (22 ± 2 °C, 54% of humidity).

2.2. Characterization techniques Mineralogical analysis of the inorganic polymer cement and composites were carried out with an X-ray powder diffractometer, XRD, (PW3710, Phillips) Cu Ka, Ni-filtered radiation (the wavelength was 1.54184 Å). The radiation was generated at 40 mA and 40 kV. The analysis was performed on fine grains of ground samples. Specimens were step-scanned as random powder from 5° to 70°, 2h range, and integrated at the rate of 2s per step. Fourier transformed infrared spectroscopy, FT-IR, (Avatar 330 FTIR, Thermo Nicolet) was performed on each sample analyzing surface and bulk areas. A minimum of 32 scans between 4000 and 500 cm 1 were averaged for each spectrum at intervals of 1 cm 1. Phases distribution, morphology and porosity were evaluated using an optical microscope, Leica DMI 5000 M. An Autopore IV 9500, 33000 psi (228 MPa) Mercury Intrusion Porosimeter (MIP) covering the pore diameter range from approximately 360–0.005 lm having two low-pressure ports and one high-pressure chamber was used for the pores analysis. Pieces

were prepared from the bulk of each sample with specimens of 1 cm3 of volume for the MIP. A recent developed synchrotron-based imagining technique, Xray micro tomography (lCT) was used for pores with diameters P20 lm, computerized measurements were made using a Skyscan 1172, (Skyscan B.V., Leuven, Belgium), at 80 kV with 100 lA, with no additional filtering and an image pixel size of 10.1 lm. Pores size were evaluated with a CT-Analyzer (CTan), 1.10.0, (Skyscan B.V., Leuven, Belgium), on a minimum of 250 slices for each sample. Visualization of the scanned images was performed using Amira 5.3.2 software (Visage Imaging GmbH, Berlin, Germany) with Voltex displaying mode. Scanning and reconstruction time was 90 min for each sample on a quad core E9500 PC with 8 GB RAM. The 3D evaluation took between 1 and 4 h per sample depending on the pore volume fraction with a total amount of 60 GB data. The MIP and micro tomography techniques together with the optical microscope (Leica DMI 5000M) allow to investigate the polycondensation process of the inorganic polymer composites in terms of cumulative pore volume and spatial pore distribution. The two latter values allowed to determine the spectrum of pore-size distribution distinguishing the ‘‘gel’’ pores corresponding to the very fine size (/ < 15 nm) and capillary pores: fine capillary pores (15 nm < / < 20 lm) and larger capillary pores (/ > 20 lm).

3. Results 3.1. Phases evolution Fig. 2a shows the XRD patterns of the inorganic polymer cement with different fraction of ladle slag based fines. The XRD pattern of reference inorganic polymer cement (GPM) shows a diffuse peak with a maximum intensity at about 2h = 26°; small peak of unreacted quartz is also present. These features are characteristic of the disordered alkali aluminosilicates structures [13,15] since metakaolin based inorganic cement, like others inorganic binders hardened at room temperature, remain predominantly X-ray amorphous [5,8,15]. Broad diffuse halo in XRD patterns corresponds to three-dimensional networks for low angle compositions (2h < 30°) while the high angle compositions (2h > 30°) is generally attributed to low molecular weight silicate (dimer, monomer) [5,16]. Therefore the basic geopolymer formulation, GPM, prepared only with metakaolin, corresponds to a typical three dimensional structure of inorganic polymer cement (Fig. 2a, bottom curve). The structure remains disordered with addition of ladle slag and the broad peak characteristic of disordered structure decreased in intensity and shifts slightly towards lower 2h indicating the effectiveness in reactivity between the inorganic polymer cement and the calcium-rich fines. In addition, there is a series of small peaks ranging between 29° and 55° which intensities become more significant with 60 wt% of calcium-rich fines with the possible interpretation that MH, (Na, K) aluminosilicate, gels decrease in intensity, while C–S–H and (Na–Ca) gels appeared. Newly formed crystalline phases among which laumontite (A4SC6H) are also generally described in these type of formulations [11,12]. A geopolymer based on ladle slag with addition of 30 wt% of metakaolin

Table 1 Overall chemical composition and mineral phases of metakaolin and fines. Samples

SiO2

Al2O3

CaO

Fe2O3

TiO2

Na2O

K2O

Phases

MK Quartz sand Nepheline syenite Lsadle slag

51.88 98.07 53.91 15.57

39.62 0.43 20.81 11.47

0.09 0.04 0.96 42.03

1.95 0.69 2.05 7.95

5.15 0.34 0.90 0.18

0.12 0.31 10.6 0.45

0.96 0.26 5.22 0.25

Amorphous + quartz + traces of spinel Quartz + trace of amorphous Nepheline syenite + amorphous Amorphous + calcium aluminosilicate

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

261

Fig. 2a. XRD patterns of the inorganic polymer cement with various amount of ladle slag: S = silica phase (quartz), CASAH = calcium silicate hydrate and NaACaAS = NaAcalcium silicate gels. Fig. 3a. FT-IR spectra of the inorganic polymer cement with various amount of ladle slag: OH is associated to the presence of water, C = carbonate phases, and SAOAT are HAMAAAS phases developed during geopolymerization with T which can be Al or Si.

Fig. 2b. XRD patterns of the inorganic polymer cement with various amount of nepheline syenite: S = silica phase (quartz), N = nepheline syenite phases.

Fig. 3b. FT-IR spectra of the inorganic polymer cement with various amount of nepheline syenite: OH is associated to the presence of water, C = carbonate phases, and SAOAT are HAMAAAS phases developed during geopolymerization with T which can be Al or Si.

Fig. 2c. XRD patterns of the inorganic polymer cement with various amount of quartz sand: S = silica phase (quartz).

have been reported [16] and the XRD patterns suggest the possibility of form C–(A)–S–H solid solution type phases. The product could contain potential aluminate oligomers poorly or not even bounded to the inorganic polymer matrix that might easily engage themselves in more stable polycondensation reaction in presence of additional silicate oligomers from soluble silica. The high content of amorphous silica and carbonated phases makes ladle slag one of the most reactive solid silicate in high alkaline media. The evolution of the IR bands at 3420 and 1650 cm 1 with the addition of ladle slag fines suggests the decrease in OH content with increasing aggregate content (Fig. 3a). The 3D phases (HAMAAAS) characterized by the bands at 1000 and 510 cm 1 are

Fig. 3c. FT-IR spectra of the inorganic polymer cement with various amount of quartz sand: OH is associated to the presence of water, C = carbonate phases, and SAOAT are HAMAAAS phases developed during geopolymerization with T which can be Al or Si.

262

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

affected by new bands formed as results of the effective reactivity of the inorganic polymer cement and the calcium-rich aggregate. Peaks at 1420, 875, 712 and 692 cm 1 have been attributed to the calcium-rich phases [2,10]. With the increase in calcium-rich fines, the principal band characteristic of the formation of sorosilicates [2,10] type units shifts from 1004.78 cm 1 (GPM) to 999.8, 993 and 989.8 cm 1 for 20, 40 and 60 wt% respectively. Bands at 1420 cm 1 and at 875 cm 1 seem to increase with the ladle slag content implying the increase of possible unreacted fines (Fig. 3a). Based on the foregoing, a limited consumption of ladle slag by the GPM matrix was inferred. The X-ray diffractions of GF series (Fig. 2b) show that the addition of nepheline syenite fines at relative low content (20 wt%, GF20) into IPC matrix does not modify the nature of the broad halo pattern. However, peaks of nepheline syenite appear. The maximum of the halo peak is centered at about 26°. The evolution of the X-ray patterns of GF series suggests a possible reaction of dissolution and polycondensation of amorphous fraction of nepheline syenite with the inorganic polymer cement. In contrast to Ca-content fines, unreacted nepheline syenite did fix carbonates (compare Figs. 3a and 3b). The soluble siliceous phase present in the nepheline syenite fines tends to enhance the polycondensation of IPC matrices by participating to the formation of HAMAAAS and CA(A)ASAH phases and to their cross-linking. With the increase in fine’s content, the crystalline phases dominated the disordered nature of the matrix due to the relative low content of amorphous phase in nepheline syenite. As underlined with IR spectra, the addition of nepheline syenite fines contributes to an increase in the intensity of the 3D aluminosilicate network (HAMAAAS). The principal band shifts to 989.64 cm 1 with 20 wt% of aggregate confirming the hypothesis of reaction deduced from XRD analysis: GF20 remains essentially disordered with a diffuse halo at relative low 2h values. The alkali attack to the feldspar crystalline structure is relatively slow and involves only a very small fraction of this aluminosilicate solid. On the other hand, the ions diffusion into the fluid in which Na+ and K+ are available permits the formation of additional phases as well as the improvement in Si/Al and Si/Na ratios, important fact for the chemical stability of IPC formed. In the case of nepheline syenite, all chemical constituents are elements already present in the geopolymer paste. This can explain the absence of effective new phase formation but only an increase in intensity of HAMAAAS due to the increase in Q3 and Q2 sites in the GF series [13,14]. As it can be observed in Fig. 3b, the intensity of peaks at 3420 and 1650 cm 1 decrease progressively with the addition of fine particles of nepheline syenite [17,18]. The peaks at 1420 and 875 cm 1, that seem to increase with the content of ladle slag fines, are completely inexistent in GF series. Even the crystalline quartz sand used as fines in the third family of concretes is disordered at its surfaces due to the unsatisfied OA surface charges and forms acidic „SiAOH (silanol) groups with water. This means that slow reaction in high alkaline media is expected. However, the X-ray patterns of the matrices with quartz-sand as fines do not give prove of any important reactivity (Fig. 2c). The broad halo patterns disappears progressively with the increase in quartz sand content and the maximum of the peak shifts to higher values of 2h indicating a low amounts of Q3 sites and the formation of dimers and monomeric silicates with the decrease in aluminum content. As in the case of ladle slag and nepheline syenite additions, the FT-IR peaks (Fig. 3c) at 3420 and 1650 cm 1 decrease progressively with the addition of fine particles of sand. The peaks at 1420 and 875 cm 1 are absent. No significant increase in intensities of HAMAAAS was observed due to the poor dissolution of quartz sand.

3.2. Cumulative pore volume and pore-size distribution GPM, a typical metakaolin based inorganic polymer cement, presents an important fraction of capillary pores due to the action of air bubbles that remain entrapped in the paste during the hydration and polycondensation phenomena [12]. The addition of calcium-rich fines contributes to a decrease of the cumulative pore volume from 340 mm3/g for GPM to 304 mm3/g (GC20), 271 mm3/g (GC40), 230 mm3/g (GC60) for 20, 40 and 60 wt% of addition respectively (Fig. 4a). The average pores size, investigated using MIP, decreases from 0.031 lm to 0.022, 0.021 and 0.018 lm

Fig. 4a. Cumulative pore volume vs pore diameter for inorganic polymer cement with various amount of ladle slag.

Fig. 4b. Cumulative pore volume vs pore diameter for inorganic polymer cement with various amount of nepheline syenite.

Fig. 4c. Cumulative pore volume vs pore diameter for inorganic polymer cement with various amount of quartz sand.

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

respectively for 20 (GC20), 40 (GC40) and 60 (GC60) wt% of ladle slag added (Fig. 5a). The threshold of pores changed from 0.042 to 0.029 lm for GC20 and GC40 and to 0.027 lm for GC60. The average capillary pore size increases slightly from 120 to 130 lm at 20 wt% of addition (GC20) before decreases to 90 and 50 lm respectively for 40 and 60 wt% of addition (Fig. 5b). The decreases in size of larger capillary pores (diameter P 20 lm) can be attributed to the increase in densification of the inorganic polymer matrix which conducts to the compression of air pockets. However, some of capillary pores increases with the increase of the amount of ladle slag [19,20] and is due to the action of the residual carbonates in the ladle slag. The decrease in cumulative pore volume for the nepheline syenite based specimens is relatively important as evidenced in Fig. 4b: 287 mm3/g (GF20), 233 mm3/g and 221 mm3/g (GF60) for 20, 40 and 60 wt% of fines respectively. The average size of fine capillary pores does not change and the value remains at

263

0.031 lm with 20 wt% of nepheline syenite (Fig. 6a). The threshold remains at 0.045 lm. With 40 wt% of nepheline syenite added, the average pore size decreases to 0.022 lm and the threshold to 0.033 lm. The average size of capillary pores (a P 20 lm) decreased under 100 lm with 20 wt% of fines (GF20) and 95 lm with 40 wt% of fines (GF40) as it can be observed in Fig. 6b. The increase of nepheline syenite to 40 wt% increase slightly the size of capillary pores to 80 lm. At 60 wt% of aggregates addition, the larger capillary pores increased in size to 180 lm. However, a global trend was the significant reduction of cumulative pore volume and high compact of the final matrix. Wolff and Wolff & Toney [17,18] describing the mineral chemistry of several nepheline syenite blocks from the deposits of Tenerife evidenced the presence of glass that reach 3.8 wt% for some samples. Enclosed crystals are always euhedral toward the glass indicating that the glass is a true quenched residual liquid and not the result of partial melting due to heating of xenoliths upon immersion in the host magma.

Fig. 5. Spatial distribution of pores for inorganic polymer cement with various amount of ladle slag: (a) fine pores; (b) capillary pores.

Fig. 6. Spatial distribution of pores for inorganic polymer cement with various amount of nepheline syenite: (a) fine pores; (b) capillary pores.

Fig. 7. Spatial distribution of pores for inorganic polymer cement with various amount of quartz sand: (a) fine pores; (b) capillary pores.

264

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

Conversion of highly-differential phonotic magma to solid nepheline syenite is seem to been achieved between 770 and 680 °C [17,18]. The presence of glass in nepheline syenite rocks justifies their partial corrosion in high alkaline solution. So increasing the amount of nepheline syenite contributes to increase the content of soluble silicate that forms additional HAMAAAS and soluble alkali silicates that can combine with the existing phases and reduce the pores size. The strengthening effects seem to be delayed with respect to the action of the alkali silicate solution. The kinetic is very slow with regard to the curing time necessary to achieve significant changes. Above 40 wt% of addition, an important amount of fines remains not actively transformed into the HAMAAAS phase and contributes to change the capillary pores size (Fig. 6b). The addition of quartz sand resulted in a slight reduction in pore volume (Fig. 4c) but the threshold for the series remains around 0.040 lm and the average pore size at 0.033 lm. In the formulation with 20 wt% of quartz rich fines the cumulative pore volumes remain >300 mm3/g which suggests the poor capacity of this type of fines to be dissolved in alkaline media and form the silicate gel related to the formation/appearance of newly HAMAAAS phases. This behavior is explained by the low amount of reactive elements in crystalline quartz that characterizes the sand. The very low amount of defects (amorphous or metastable phase) reduces the amount of soluble silica available for the formation of additional HAMAAAS capable on enhancing the densification and reduce porosity (Fig. 7a). For all the specimens with the addition of the quartz sand, the larger capillary pores size increase (Fig. 7b) and reach 190, 160, and 140 lm for GS20, GS40 and GS60 respectively. It was noted that for the same amount of fines added, with the same granulometry (a < 100 lm), quartz sand based samples produced paste with the lower viscosity compared to nepheline syenite while the paste with ladle slag presented the higher viscosity. This suggests that the evolution of capillary pores in the samples with quartz sand may not be exclusively linked to the coarsening of the microstructure but includes also the action of the water that is not absorbed by aggregate but remains at their surface and modify the pores formation.

changed with the curing time. With the 20 wt% of ladle slag, the three points flexural strength increased from 3.99 to 6.80 MPa (Fig. 8a) after 28 days of curing. Further addition of ladle slag (40 and 60 wt%) tends to decrease slightly the flexural strength

Fig. 8b. Three-point bending strength (MPa) with various content of nepheline syenite vs curing time.

3.3. Mechanical characterization GPM, the standard metakaolin inorganic matrix presented a value of flexural strength of 3.99 MPa [10], value that did not

Fig. 8c. Three-point bending strength (MPa) with various content of quartz sand vs curing time.

Fig. 8a. Three-point bending strength (MPa) with various content of ladle slag vs curing time.

265

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

at 28 days of curing: 6.43 and 5.90 MPa respectively for GC40 and GC60. These results can be explained by the fact that despite the decrease of pores size with the addition of slag, the volume of capillary pores increases [19]. Between 28 days and 90 days of curing, the mechanical properties of GC20, GC40 and GC60 increases from 6.80 to 9.01 MPa; 6.43 to 9.77 MPa and 5.9 to 8.80 MPa. Between 90 and 180 days, the mechanical strength remains constant. The actual results agree with the findings of many authors [19–23] on the positive effects on the physical properties and durability of combined HAMAAAS and CASAH phases. The presence of highly reactive CaO and amorphous silica resulting from the thermal history of the ladle slag allows the formation of additional phases (HAMAAAS and MACa silicate gels) that contribute to increase the densification with beneficial effects on the pore size and mechanical properties. For geopolymers with 20 wt% of nepheline syenite (GF20), the value of the three points flexural strength moves from 3.99 to 4.76 MPa after 28 days of curing. Changes are not as important as with ladle slag and can be explained by the relatively low reactivity of nepheline syenite (only the small glassy fraction reacts) with respect to ladle slag. The increase of nepheline syenite enhances connectivity in the matrix with consequence reduction of pores size and the cumulative pore volume (Fig. 4b). The three-points flexural strength increases to 5.6 and 7.35 MPa after 28 days of curing (Fig. 8b). By increasing the curing time to 90 days, the threepoints flexural strength reach 7.01, 7.76 and 9.94 MPa respectively for GF20, GF40 and GF60. These values did not change significantly with further increase of curing time to 180 days (Fig. 8b). The continuous increase of the mechanical strength with the addition of nepheline syenite contrast with the optimum reach with ladle slag yet at 20 wt% of addition. The difference in the behavior is discussed here with the effective amount of material that pass into the pore solution. Excess of ladle slag that does not participate to the reaction is harmful for the strength and stability of the matrix. Considering the addition of quartz sand as fine aggregate (Fig. 8c) we can compare to the samples 20 wt% (GS20) to nepheline syenite containing ones: the variation of the mechanical properties indicates the limited action of the quartz sand in terms of improvement of polymerization and cross-linking. However, the increase in curing time, from 28 to 90 days contributed to improve the flexural strength that reached 5.5 MPa. This value does not change with further increase of curing time to 180 days suggesting some pozzolanic reactions or more condensation between 28 and 90 days that contribute to enhance the strength of the matrix. The flexural strength reaches the value of 7 MPa with 60 wt% of quartz sand when 90 days of curing were considered. As already

(a) GPM

noted with nepheline syenite, the continuous increase of flexural strength with the addition of quartz sand contrast with ladle slag with which saturation was reached as with 20 wt% of addition. 3.4. Microstructure and Fracture surface observations Fig. 9 shows the morphology of the matrices of GPM, metakaolin based inorganic polymer cement, with the evolution associated to the additions of ladle slag and nepheline syenite. The GPM sample micrograph (Fig. 9a) shows typical amorphous nature of any metakaolin based IPC. The structural changes observed with the addition of ladle slag (Fig. 9b) and nepheline syenite (Fig. 9c) fines are directly linked to the reactive processes at the molecular level that take place between the geopolymer cement paste (GPM) and the fines. Larger capillary pores and micro cracks were observed in GPM. From Fig. 9a, it can observed that once a defect origin develops into a micro cracks line, the component’s stresses are easily redistributed with actives wing cracks. The reduction of the pore size and the greater densification of the matrices with the addition of fines suggest that with the presence of sufficient and controlled amount of reactive phase (amorphous silica) the rearrangement to form an extensively cross-linked network become more significant. The poorly bonded alumina oligomers and the low degree of polymerization of aluminosilicates in standard metakaolin based inorganic polymer cement is consistent with others findings [21,22]. A high-silica crystalline phase, analcine was identified in the sample with low silica available alone with zeolite [22]. Introducing fine powders of ladle slag, nepheline syenite or, in some extend, silica sand, the behavior is primary function of the size of their particle and the related specific area that govern the densification and reduction of pore size and cumulative pore volume. The increase in capillary pores that contrast with the reduction of size for pores <20 lm in GC series is indicative of the attention that should be paid on the amount of these class of materials in mortars and concretes although their greater reactivity. Ladle slag contains amorphous silica which in pore solution dissolves to enhance the formation of HAMAAAS phases. Additionally, the reactive CaO present in the aggregate tends to form HACAS and NaACa silicate gels. The combination of these phases has as effects the increase in polycondensation and improvement of strength (Figs. 9–11). These results achieved with ladle slag can be also linked to the presence of C2S (belite) whose reactivity is known to enhance the hydration of cements [24]. The mineral admixture with its reactivity leads to the better particle packing, good compaction and improvement of the interfacial zone that finally explain the increase in mechanical properties.

(b) GC60

(c) GF60

Fig. 9. Fracture surface (ESEM images) of inorganic polymers after the three points flexural strength: (a) without aggregates (GPM), (b) with 60 wt% of ladle slag (GC60) and (c) with 60 wt% of nepheline syenite (GF60).

266

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

(a)

(c)

(b)

Fig. 10. ESEM micrographs of inorganic polymer composites (40 wt% fine aggregates) showing: (a) dense matrix with open porosities of GC40; b) fully dense matrix of GF40 and (c) dense matrix of GS40 with microcracks at the interface IPC-quartz grain.

(a) C20

(b) F20

(c) Fig. 11. ESEM micrographs of inorganic polymer cement showing the improvement of the polycondensation of the gels with the presence of (a) 20 wt% of ladle slag and (b) 20 wt% of nepheline syenite; (c) EDS compositions of the two gels.

In the nepheline syenite based composites, the glassy phase dissolved in the pore solution has bulk composition similar to that of the inorganic polymer cement (GPM). The soluble silica produced in the pore solution is essential in combining the residual alumina ologomers, increases the Si/Al and consequently the nucleation sites of HAMAAAS phases together with the degree of connectivity within the matrix: the effects are the increase in polycondensation and the reduction of both pore size and the cumulative pore volume. The glass content of nepheline syenite is described to be relatively low around 3 or 4 wt% [19,20]. It is

expected that the nepheline syenite acts actively in the composite matrix without any preoccupation on the proportion since the remaining fraction >95 wt% are dense and stable crystalline allowing their use for high strength and durable matrices. The addition of quartz sand enhances the densification of the IPC matrices. However, the poor reactivity of the material from the poor dissolution in alkaline solution modifies the interface behavior between the inorganic polymer cement and aggregates with consequence of micro cracks intensively present in the matrix (Fig. 9c).

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

4. Discussion 4.1. Correlation between microstructure and properties The surface morphological features of fresh fractured ladle slag and nepheline syenite based inorganic polymer composites showed that fines with relative reactivity (ladle slag and nepheline syenite) tend to be completely surrounded by the HAMAAAS precipitated gels. At lower magnification, the fractured surfaces appeared frosted (Figs. 9 and 10) due to the presence of relatively intense hills and valleys on the scale of 1–2 lm. The bulk composition of the homogeneous matrix that results present typical HAMAAAS gels for nepheline syenite and both HAMAAAS and HACAS gels for ladle slag (Figs. 2a–3c). These results demonstrate that if we identify the ideal grade of fines and inorganic polymer cement paste, if we can master the level of reactivity in order to predict the chemical equilibrium for inorganic polymer cement, mortars or concretes, we can achieve high strength and durable inorganic polymer composites. Using the optimal fine particles size of aggregates could therefore be possible engineering solution for addressing the need to reduce or eliminate early age cracking (both micro and macro) in concretes and mortars. The observed reduction of larger capillary pores with the densification of the matrices is expected to contribute to the long term durability of IPC composites. While inorganic polymer composites are presented as innovative materials for the future, theoretical and experimental models for predicting matrices with definite chemico-physical and mechanical properties remain a key need. In this work, fracture surfaces observations revealed curved grooves or cracks, slightly deep in most cases when the fine powders of ladle slag or nepheline syenite are used. These raised blocky areas with concavities in some places (Figs. 9–11) suggest that there were conchoidal fractures arising from intergranular cracks. The bond strength developed in these cases was significant. From the observed micrographs (Fig. 11), it was concluded on the good coexistence of HAMAAAS and HACAS phases in dense and homogeneous matrix. No phase separation was observed for ladle slag (20 wt%) and nepheline syenite (up to 60 wt%) based inorganic polymer composites. The increase of the amount of ladle slag above 20 wt% (GC40 and GC60) tends to create in the matrices a new class of pores with size comprised between the existing fine and larger size described in GPM. The objectives of this work were to investigate on the possibilities to control the alkali residues generally present in inorganic polymer cement as well as poorly bounded alumina oligomers identified in the standard metakaolin based IPC. Results showed that the elimination of MAH gels and formation of more HAMA(A)AS gels, permitted to avoid the crystallization of MAH phases and the reduction of the gradient of concentration of alkalis sources of micro cracks generally act as failure origin. When the MAH gel is consumed by the amorphous aluminosilicates, high SiO2/Na2O molar ratio silicate gels are formed; they are capable of enhancing the polycondensation of the inorganic polymer composites and also reducing the pore size (Figs. 4a–4c). The formation of high strength inorganic polymer composites (binders, mortars and concretes) is correlated to a good understanding of the extend of dissolution and polymerization in various systems. For a given volume of pore solution, the amount of the reactive materials produced is linked to the amount available to pass into the reactive system, here the amorphous or disordered phases. Ladle slag reacts according to the pH of the pore solution and as amorphous silica and calcium pass into the reactive system, the pH decreases with the decrease in alkaline elements. In the case of nepheline syenite, the amount of amorphous phase available is limited but sufficient to achieve the results described in this study. On the contrary, the

267

quasi absence of amorphous phase limits the action of quartz. Lee and Deventer [22,23] investigated the interval between mineral aggregates and inorganic polymer cements and noted the importance of the availability of soluble silicate in improving and strengthening the interface. They demonstrated that when soluble silica is available, the interface between paste and aggregate are completely similar to that of binder. Soluble silica were effective in reducing alkali saturation in the concrete pore solution even when highly alkali-concentrated activating solution was used [26]. The experimental model from this work has established that when well designed, taking in account the bulk composition, particles size and particles distribution, the paste promotes greater interparticles bonding with aggregates resulting in a mortar or concrete with continuous structure having the space between paste and aggregate reduced to the minimum. Minimum capable to promote the crystallization of alkali and others gels prompt to alter the properties of the end products. The particles size distribution along with curing conditions can be used to monitor the porosity (especially the connected accessible volume of pores) [24,25]. 4.2. Strengthening mechanisms It is significant to note that the production of inorganic polymer systems in form of binders, mortars and concretes is supported by the same mechanism of polymerization of inorganic polymer cement. The Si/Al molar ratio, the Na/Al and the SiO2/Na2O are all important parameters which, when under control (Si/Al = 2–3, Na/Al  1 and SiO2/Na2O P 3) contribute to develop a chemicophysical equilibrium. The physical aspect includes undoubtedly the particles packing and homogeneous phases distribution. Particle size distribution focused on the dissolution of reacting grains and on the formation of a shell of reaction products around these grains. In the case of ladle slag and nepheline syenite, we observed the formation of HACAS and HAMAAAS gels capable on embedding the particles in continuous matrices with the difficulty to identify the interface transition between paste and aggregate (Fig. 11). The formation of HACAS, HAMAAAS and additional siliceous gels contributed to act as new nucleating sites for the formation and consolidation of strength enhanced system. The results of this study indicate that the phases evolution and the spatial pore distribution as well as the cumulative pore volume of inorganic polymer composites can be monitored for the better mastering/tailoring of the key parameters (pH of the bulk matrix, porosity and microstructure of the interface zone between paste and aggregates, etc.) in the design of optimum binders, mortars and concretes. By using calcium-rich amorphous aluminosilicates (case of ladle slag), the disordered structure of the matrix was maintained and the relatively low content of HAMAAAS phase was due to the important formation of HACAS and HACAAAS as indicated by XRD and IR (Figs. 2a–3c). Alonso and Palomo [26,27] described the nucleation of both HAMAAAS and HACAS gels in alkali-actived aluminosilicates. Therefore the use of a higher amount of this type of fines decreases, as described in this work, the HAMAAAS phases with the increase of CA(A)ASAH phase. The residual carbonates in the ladle slag contributes to enhance the number of air pockets formed. The quartz sand is composed of more stable crystalline phase of quartz, while nepheline syenite is a feldspathic phase with relative disordered and amorphous phases similar to glassy silica. These disordered to amorphous phases are responsible for the reactivity of nepheline syenite which appeared significant but limited. In the long term schedule (complete reaction), the action of aggregates in the inorganic polymer composites decreases the M/Si ratio of HAMAAAS and HAMAS phases as well as M/Al and thereby improves its chemico-physical equilibrium.

268

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

The chemico-physical equilibrium here is, as already described, related to the absence of ions exchange and ions transport hence, to the stability of the matrix composite. The formation of new phases and their nature, the pore volume evolution, the changes in pore size distribution demonstrate that the nature of aggregate used is important for the characteristics of the inorganic mortars or concretes. From the above mentioned, an ideal inorganic polymer concrete is characterized by the aggregate, the continuous inorganic polymer cement paste/matrix and the interfacial zone capable on preserving the continuity between paste and aggregate. The formation of additional HAMAAAS phase when using nepheline syenite, the HACAS phases in calcium-rich fines coupled with the reduction of pore volume and pore size, and the compatibility between the new gels and the metakaolin based inorganic polymer cement in the case of non-calcium fines are promising in the design of inorganic polymer concretes with optimum grade. 4.3. Effects of particles size of aggregates The particle size distribution of the three aggregates used in this work (Fig. 1) is closed to that presented by cement described by Bentz [25] who explained the reduction in connected accessible pores when the changes is operated from coarser cement to cement with higher fineness [24]. Very fine particles (27, 39 and 49 nm in diameter) are observed in ladle slag, quartz sand and nepheline syenite ground aggregates with, in the same order, 40, 59 and 58 vol% of particles with a < 10 lm. This particle size distribution which should affect the hydration process is used to support the understanding of the changes observed in the porosity, mechanical properties and microstructure of the inorganic polymer composites. Considering the specimens with the effective reactivity between the IPC and the fines (case of GC and GF series), the connected pore volume is expected to decrease within the complete hydration phenomena (70–120 h). The important increase in mechanical properties between 28 and 90 days of curing (Figs. 8 and 4) corresponds to the period in which the connected pore volume decreases considerably knowing their negative effects on the properties of final products [24,25]. Between 90 and 180 days, the variation in the connected pore volume is limited with the hydration kinetic and the reaction of polycondensation already in the last step of their activities. The fact that nepheline syenite contains just a small amount of glass (amorphous phase) conducts to the leaching of a reasonable amount into the pore solution while the particles of crystalline fraction with incongruent dissolution easily insert themselves as fillers and participate on the strengthening action reacting at their surfaces. The process is slow but effectively positive for the final properties of the inorganic polymer composites. 5. Conclusion Ladle slag, nepheline syenite, and quartz sand based aggregates have been added, in form of fine powder, to fresh metakaolin based inorganic polymer cement in different proportion (20, 40, 60 wt%). Investigations based on IR and XRD revealed that HAMAAAS phases increase in GF based compositions compared to GS based compositions indicative of the degree of dissolution and interaction of amorphous to disordered silica of the aggregates. The significant reactivity of the glass fraction of nepheline syenite aggregate contributes to the reduction of both fine and capillary pores together with the pore size. Moreover, the coarsening microstructure observed with the increase in aggregate content affects the size of capillary pores. In calcium-rich aggregate (ladle slag), HACAS and HACAAAS accompanied HAMAAAS phases and the significant decrease in content of HAMAAAS demonstrates the high reactivity

of the CG based aggregates in the inorganic polymer cement. The formation of these new phases reduces the cumulative pore volume and the particle size for both fine and capillary pores. The increase in CG content affects the capillary pore volume, while the size of fine pores decreases due to an increase of air-bubbles from the slag. Additionally the following conclusive remarks can be drawn:  The particle size, specific surface area and reactivity of fines results important parameters in the packing and consolidation processes of IPC binders, mortars or concretes.  The improvement in mechanical properties with the curing time suggests reductions in connected accessible pore volume, which is an undesirable form of porosity in IPC.  It was concluded that fine powder of aggregates can be used to remove the MH and poorly bounded aluminate oligomers [Al(OH)4] in metakaolin based inorganic polymers. This suggests the possibility to improve the interface zone for the design of mortars and concretes with optimum performance. The simultaneous evaluation of the porosity with MIP and Micro tomography enable coverage of a broad range of pore sizes and confirmed the pore structure of the IPC: high proportion of nanopores and fine capillary pores and limited amount of larger capillary pores. Additionally the experimental results here reported demonstrated that it is possible to act on both fine and larger capillary pores for the optimization of the performance of inorganic polymer composites. Acknowledgment Ingessil Srl is acknowledged to have provided sodium silicate used for these investigations. References [1] Van Deventer JSJ, Provis JL, Duxson P. Technical and commercial progress in the adoption of geopolymer cement. Miner Eng 2012;29:89–104. [2] Davidovits J. Geopolymers – inorganic polymeric new materials. J Therm Anal 1991;37(8):1633–56. [3] Duxson P, Fernandez-Jimenez A, Provis JL, Luckey GC, Palomo A, van Deventer JSJ. Geopolymer technology: the current state of the art. J Mater Sci 2007;42(9):2917–33. [4] Kamseu E, Leonelli C, Perera DS, Melo UC, Lemougna PN. Investigation of volcanic ash based geopolymers as potential building materials. Interceram 2009;58(2–3):136–40. [5] White CE, Provis JL, Llobet AL, Proffen T, van Deventer JSJ. Evolution of local structure in geopolymer gels: an in situ neutron pair distribution function analysis. J Am Ceram Soc 2011;94(10):3532–9. [6] Rodriguez Erich D, Bernal Susan A, Provis John L, Paya Jordi, Monzo Jose M, Borrachero Maria Victoria. Effect of nanosilica-based activators on the performance of an alkali-activated fly ash binder. Cem Concr Compos 2013;35:1–11. [7] Habert G, d’Espinose de Lacaillerie JB, Roussel N. An environmental evaluation of geopolymer based concrete production: reviewing current research trends. J Cleaner Prod 2011;19:1229–38. [8] Lloyd RR, Provis JL, van Deventer JSJ. Microscopy and microanalysis of inorganic polymer cements 1: remnant fly ash particles. J Mater Sci 2009;44:608–19. [9] Lloyd RR, Provis JL, van Deventer JSJ. Microscopy and microanalysis of inorganic polymer cements 2: the gel binder. J Mater Sci 2009;44:620–31. [10] Weng L, Sagoe-Crentsil K. Dissolution processes, hydrolysis and condensation reactions during geopolymer synthesis: part I – low Si/Al ratio systems. J Mater Sci 2007;42:2997–3006. [11] Weng L, Sagoe-Crentsil K. Dissolution processes, hydrolysis and condensation reactions during geopolymer synthesis: part II – high Si/Al ratio systems. J Mater Sci 2007;42:3007–14. [12] Kamseu E, Leonelli C, Chinje Melo UF, Perera DS, Lemougna LN. Polysialate matrixes from Al-rich and Si-rich metakaolins: polycondensation and physicochemical properties. Interceram 2011;60(1):25–31. [13] Gaboriaud F, Chaumont D, Nonat A, Hanquet B, Craievich A. Study of the influence of alkaline ions (Li, Na and K) on the structure of the silicate entities in silico alkaline sol and on the formation of the silico-calco-alkaline gel. J Sol– Gel Sci Technol 1998;13(1,2,3):353–8. [14] Brough AR, Katz A, Sun G-K, Struble LJ, Kirkpatrick RJ, Young JF. Adiabatically cured, alkali-activated cement-based wasteforms containing high levels of fly

E. Kamseu et al. / Cement & Concrete Composites 53 (2014) 258–269

[15] [16]

[17] [18]

[19] [20]

ash: formation of zeolites and Al-substituted C–S–H. Cem Concr Res 2001;31(10):1437–47. Prince W, Castanier G, Giafferi JL. Similarity between alkali-aggregate reaction and natural alteration of rocks. Cem Concr Res 2001;31:271–6. Bignozzi MC, Manzi S, Lancellotti I, Kamseu E, Barbieri L, Leonelli C. Mix-design and characterization of geopolymer system based on metakaolin and ladle slag. Appl Clay Sci 2013;73:78–85. Wolff JA. Crystallisation of nepheline syenite in a subvolcanic magma system: Tenerife, Canary Islands. Lithos 1987;20:207–23. Wolff JA, Toney JB. Trapped liquid from a nepheline syenite: a re-evaluation of Na-, Zr-, F-rich interstitial glass in a xenolith from Tenerife, Canary Islands. Lithos 1993;29:285–93. Hajimohammadi A, Provis JL, van Deventer JSJ. The effect of silica availability on the mechanism of geopolymerisation. Cem Concr Res 2011;41:210–6. White CE, Provis JL, Llobet A, Proffen T, van Deventer JSJ. Evolution of local structure in Geopolymer Gels: an in situ Neutron pair distribution function analysis. J Am Ceram Soc 2011;94(10):3532–9.

269

[21] Fukuda K, Ito S. Improvement in reactivity and grindability of belite-rich cement by remelting reaction. J Am Ceram Soc 1999;82(8):2177–80. [22] Lee WKW, van Deventer JSJ. Chemical interactions between siliceous aggregates and low-Ca alkali-activated cements. Cem Concr Res 2007;37:844–55. [23] Lee WKW, van Deventer JSJ. The interface between natural siliceous aggregates and geopolymers. Cem Concr Res 2004;34:195–206. [24] Bentz DP. Capillary porosity depercolation/repercolation in hydrating cement pastes via low-temperature calorimetry measurements and CEMHYD3D Modeling. J Am Ceram Soc 2006;89(8):2606–11. [25] Bentz DP. Influence of cement particle-size distribution on early age autogeneous strains and stresses in cement-based materials. J Am Ceram Soc 2001;84(1):129–35. [26] Alonso S, Palomo A. Alkaline activation of metakaolin–calcium hydroxide solid mixtures: influence of temperature, activator concentration and metakaolin/ Ca(OH)2 ratio. Mater Lett 2001;47:55–62. [27] Alonso S, Palomo A. Calorimetric study of alkaline activation of calcium hydroxide–metakaolin solid-mixtures. Cem Concr Res 2001;31:25–30.