Starting materials
Four raw clays from different deposits in Central Germany were studied. Two of the clays, designated I1 and I2, can be classified as brick clays, i.e. they fall in the compositional range of clays used for the production of bricks [19] (see below). The other two clays, designated K1 and K2, contained more kaolinite, but were used without a purification process, i.e. they can be classified as low-grade kaolinitic clays [19] (see below). All clays were preconditioned by drying at 120 °C for 48 h and grinding in an impact mill until passing through a 2-mm sieve.
Portland cement (designation PC), CEM I 42.5 R, with a sodium oxide equivalent of Na2Oeq = 0.88% was used to produce blended cement pastes and tested as a reference. Quartz powder (designation Q) was used as a filler in blended cement pastes for comparison with the calcined clays.
The chemical compositions and relevant physical properties of the starting materials are shown in Table 1.
Table 1 Chemical composition and physical properties of the Portland cement, the quartz powder and the raw clays
The phase assemblages of the clays, as determined by XRD (Fig. 1; for experimental conditions see Sect. 2.6), differed substantially between the illitic clays (brick clays) and the low-grade kaolinitic clays. Both illitic clays, I1 and I2, exhibited prominent reflections of illite (PDF # 00-026-0911) and kaolinite (PDF # 01-074-1784). In the diffractogram of I2, a reflection at a d-spacing of 14.3 Å (~ 6.2° 2θ) can be discerned in addition, which is assigned to smectite [23]. A small hump around d = 12.2 Å (~ 7.3° 2θ) in the same diffractogram is consistent with minor amounts of smectite with only one water layer in the interlayer space [23], possibly caused by partial dehydration during drying and grinding of the clay. Accessory minerals in both illitic clays were quartz (PDF # 00-046-1045), anatase (PDF # 00-021-1272) and feldspar (orthoclase, anorthoclase); in I2, hematite (PDF # 00-033-0664), calcite (PDF # 01-086-0174) and possibly traces cristobalite (PDF # 00-039-1425) were identified in addition.
The diffractograms of clays K1 and K2 exhibited a considerably higher intensity of the basal reflection of kaolinite (PDF # 01-074-1784) relative to the basal reflection of illite (PDF # 00-026-0911); indications of smectite were not found. Quartz (PDF # 00-046-1045) was an accessory mineral in both kaolinitic clays, while anatase (PDF # 00-021-1272) and minor amounts of gibbsite (PDF # 00-033-0018) were identified only in K1.
Rietveld quantitative phase analyses (RQPA) were performed using montmorillonite as the smectite mineral. The analyses showed that none of the clays contained significant amounts of amorphous phase(s); the resulting abundances of the major crystalline phases are listed in Table 2.
Table 2 Phase abundances of the raw clays as determined by RQPA (in wt%)
The TGA/DTG curves of the clays are shown in Fig. 2 (for experimental conditions and computation see Sect. 2.7). The fractions of kaolinite that were computed from the TGA curves were 17% for I1, 13% for I2, 56% for K1, and 61% for K2. These results are in good agreement with the RQPA results, though the relative order of the amounts of kaolinite in I1 and I2 was not consistent between the two methods.
Both brick clays exhibited a shoulder/peak in the DTG curves at around 600–700 °C, assigned to the dehydroxylation of illite, which was more pronounced for I2, in accord with the RQPA results. However, evaluation of the TGA curves with the tangent method yielded similar illite contents of ~ 23% for both clays.
The mass loss up to ~ 200 °C, seen in all clays, but most pronounced for I1 and I2, can be assigned to loss of adsorbed water and to loss of interlayer water from smectite-type clay minerals. However, for all clays this mass loss was low (< 1%), due to the drying of the clays at 120 °C before grinding and characterization. The similarity of the TGA/DTG curves of I1 and I2 in the temperature range < 200 °C may suggest that I1, as well, contained clay minerals with hydrated interlayers. However, smectite was identified only in I2 by XRD (Fig. 1; Table 2), and in line with this, only I2 exhibited a discernible peak in the DTG curve at ~ 800–900 °C.
Calcination of the clays
The thermal activation of the clays was carried out by batch calcination, each batch ~ 50 g of raw clay, in platinum crucibles in a muffle furnace. The calcination temperatures were 650, 850 and 900 °C. The clays were heated at a rate of 10 °C/min and then soaked at the specified temperature for 3 h. Subsequently, samples were removed from the furnace and left to cool on a lab bench for 30 min, and then stored in a desiccator at 23 °C with silica gel to avoid excessive absorption of water and potential rehydroxylation [24].
Below, calcined clays are designated by the name of the raw clay, followed by an underscore and the calcination temperature in °C.
Blended cement pastes
For the preparation of blended pastes, the Portland cement and calcined clay or quartz powder were first manually mixed and homogenised for 30 s. Two substitution rates were employed: 15 v/v% and 30 v/v%. These volume-based substitution rates are equivalent to mass-based substitution rates of ~ 13 wt% and ~ 26 wt%, respectively (the apparent densities of the calcined clays were 2.53–2.68 g/cm3).
The Portland cement and the blended cements were mixed with deionised water at a water-to-solids ratio (w/s) of 1.26 by volume (equivalent to a w/s of ~ 0.42 by weight for the blended cement pastes; w/s = 0.40 by weight for the plain Portland cement paste). The resulting pastes for strength testing and microstructural characterisation were mixed in a planetary centrifugal mixer at a rotational speed of 1250/min for 4 min. Subsequently, the pastes were cast into 20 mm × 20 mm × 20 mm cube moulds and cured for 1, 7 or 28 days at 20 °C and ~ 99% relative humidity (above water). The pastes for isothermal calorimetry were mixed for 4 min at 1600 rpm using a Janke & Kunkel IKA-WERK mechanical mixer and then ~ 15 g of the paste immediately placed in plastic ampoules and inserted in the calorimeter cell.
Superplasticizers were not applied. Thus, workability differed between pastes, depending on the calcined clays and the substitution rate, but all pastes were workable enough to be produced and cast without the occurrence of excessive air voids.
Designations of blended cement pastes below are composed of PC (for Portland cement), the substitution rate in v/v% and the supplementary material (either Q for quartz powder or the designation of the calcined clay), separated by an underscore.
Isothermal calorimetry
The heat release during hydration at 20 °C (heat of hydration) of the plain Portland cement paste and the blended cement pastes was measured up to 28 days in a TAM Air isothermal calorimeter. The time between addition of water to the cement pastes and the start of the measurements was recorded and the obtained calorimetry data adjusted accordingly.
Compressive strength testing
Compressive strength testing was performed on 16 hydrated blended cement pastes. Blends with brick clays (i.e. clays I1 and I2) were tested for calcination temperatures of 650 and 850 °C, and the low-grade kaolinitic clays (i.e. clays K1 and K2) were tested for calcination temperatures of 650 and 900 °C. Compressive strength testing of the hardened cubes was done on a 10-kN testing machine at a loading rate of 240 N/s, equivalent to 0.6 MPa for the chosen sample dimensions.
After strength testing, specimens were ground manually using mortar and pestle for subsequent XRD analyses, and the resulting powders stored in glass vials in a desiccator (flushed with argon to remove the air) until required for testing.
X-ray diffraction
Rietveld quantitative phase analysis (RQPA) was performed for the raw clays. The NIST Standard Reference Material 660c (LaB6) was used to determine the instrumental parameters of the employed XRD device; zincite (ZnO; AppliChem, grade p.a., min. 99.5%) was used as internal standard for the measurements of the clay samples. The samples were ground in a McCrone ‘micronizing’ mill with 10% zincite in propan-2-ol for 5 min, and the resulting slurries were dried at 40 °C overnight. The dry powders were filled into purpose-built sample holders by side-loading. XRD patterns were recorded in Bragg–Brentano geometry on a Rigaku Ultima IV diffractometer under the following conditions: Cu Kα radiation (λ = 1.541874 Å); tube operating at 40 kV, 40 mA; sampling interval: 0.01° 2θ; scan rate: 0.2° 2θ min−1; scanning range: 5°–150° 2θ; divergence slit: in-plane 1/6°, axial 10 mm; strip detector D/teX Ultra with 5° Soller slits. Phase identification was done with Match! version 3.5 (Crystal Impact, Germany); Rietveld analyses were performed with TOPAS version 5 (Bruker AXS, Germany).
XRD patterns for qualitative phase analysis were recorded for the calcined clays and the hydrated cement pastes. The sample powders were filled into sample holders by top-loading. Measurements were done on the same device as the measurements for RQPA, but employing the following conditions: CuKα radiation (λ = 1.541874 Å); tube operating at 40 kV, 40 mA; sampling interval: 0.02° 2θ; scan rate: 0.5° 2θ min−1; scanning range: 5°–65° 2θ; divergence slit: in-plane 0.5°, axial 10 mm; strip detector D/teX Ultra with 5° Soller slits. The sample holders were spun at 15 rpm during the measurements. Phase identification was performed with Match! version 3.5 (Crystal Impact, Germany).
Thermogravimetric analysis
Thermogravimetric analysis (TGA) measurements were performed on raw clays, calcined clays as well as hydrated cement pastes using a Mettler Toledo TGA/DSC3+STARe. Sample masses of ~ 10 mg were heated in Al2O3 crucibles first to 40 °C and kept at that temperature for one hour, then the samples were heated to 1000 °C (clays) or 900 °C (cement pastes) at a heating rate of 5 °C/min; the nitrogen flux during the measurements was 80 ml/min. From the TGA results, the derivative thermogravimetry (DTG) curves were calculated.
The dehydroxylation of the kaolinite, illite and smectite occurs at temperature intervals that partially overlap, but may allow their differentiation in DTG curves (see e.g., Refs. [4, 25]). Under the present experimental conditions, the mass losses due to dehydroxylation of kaolinite, illite and smectite were found at approximately 400–600, 600–700 and 800–900 °C, respectively. Where a distinctive feature could be identified in the DTG curves, the corresponding mass loss (\(m_{{{\text{H}}_{2} {\text{O}}}}\)) was determined by the tangent method from the TGA curve.
The mass fractions of the individual clay minerals (fclaymin) in the clays was calculated from the determined mass losses according to
$$f_{{{\text{claymin}}}} = m_{{{\text{H}}_{2} {\text{O}}}} \times \left( {\frac{{M_{{{\text{claymin}}}} }}{{n_{{{\text{H}}_{2} {\text{O}}}} \times M_{{{\text{H}}_{2} {\text{O}}}} }}} \right)$$
(1)
where Mclaymin and \(M_{{{\text{H}}_{2} {\text{O}}}}\) are the molar masses of the clay mineral and water, respectively, and \(n_{{{\text{H}}_{{2}} {\text{O}}}}\) is the number of moles of water released on dehydroxylation of the clay mineral. For the numerical values of \(n_{{{\text{H}}_{{2}} {\text{O}}}}\) and Mclaymin, the following idealised stoichiometries were assumed for the dehydroxylation of kaolinite, illite and smectite (montmorillonite), respectively:
$${\text{Al}}_{2} \left( {{\text{Si}}_{2} {\text{O}}_{5} } \right)\left( {{\text{OH}}} \right)_{4} \to {\text{Al}}_{2} {\text{O}}_{3} \cdot 2{\text{SiO}}_{2} + 2{\text{H}}_{2} {\text{O}}$$
(2)
$${\text{KAl}}_{2} \left( {{\text{AlSi}}_{3} {\text{O}}_{10} } \right)\left( {{\text{OH}}} \right)_{2} \to 0.5{\text{K}}_{2} {\text{O}} \cdot 1.5{\text{Al}}_{2} {\text{O}}_{3} \cdot 3{\text{SiO}}_{2} + {\text{H}}_{2} {\text{O}}$$
(3)
$$\begin{aligned} & {\text{Na}}_{0.3} \left( {{\text{Al}}_{1.7} {\text{Mg}}_{0.3} } \right)\left( {{\text{Si}}_{4} {\text{O}}_{10} } \right)\left( {{\text{OH}}} \right)_{2} \cdot 2{\text{H}}_{2} {\text{O}} \to 0.15{\text{Na}}_{2} {\text{O}} \cdot 0.3{\text{MgO}} \cdot 0.85{\text{Al}}_{3} {\text{O}}_{3} \cdot 4{\text{SiO}}_{2} \\ & \quad + 2{\text{H}}_{2} {\text{O}}\,\left[ {{\text{dehydration}}} \right] + \,{\text{H}}_{2} {\text{O}}\,\left[ {{\text{dehydroxylation}}} \right] \\ \end{aligned}$$
(4)
The degree of dehydroxylation (DTG) of kaolinite and the 2:1 minerals (illite and smectite, considered together) in the calcined clays was done by comparing the respective mass losses in the TGA/DTG curves of the raw clays and the calcined clays:
$$D_{{{\text{TG}}}} = \left( {\frac{{f_{{{\text{claymin}}}} - f_{{\text{claymin,calcined}}} }}{{f_{{{\text{claymin}}}} }}} \right) \times 100\%$$
(5)
where fclaymin,calcined is the mass fraction of clay mineral remaining after calcination, obtained from Eq. (1) applied to the TGA curve of the calcined clay.
The amount of bound water in the hydrated cement pastes was defined as the mass loss in the temperature interval 50–500 °C (thus excluding the weight loss due to decomposition of carbonates). The portlandite content of the pastes was determined by the tangent method from the step in the TGA curves at about 400–450 °C.