Preparation of β-belite using liquid alkali silicates

P. Koutník*

Unipetrol Centre for Research and Education, (Ústí nad Labem, Czech Republic)

*petr.koutnik@unicre.cz

1. INTRODUCTIONTOP

Cement production is associated with high energy consumption and CO₂ emissions, therefore the cement industry is facing challenges to reduce them. One of the routes is the manufacturing of clinkers based on dicalcium silicate (C₂S–belite). Belite formation generates reduced amounts of CO₂ and also energy consumption is lower in comparison with tricalcium silicate (C₃S – alite), the main component of Portland cement (1,2).

Dicalcium silicate exists in several polymorphic forms at ordinary pressures: α, α´_H, α´_L, β and γ (3). The γ form is thermodynamically stable at room temperature, but it has a lower degree of hydraulicity than the other forms. α, α´_H, α´_L and β forms are not stable in the pure state at room temperature (4,5). Chemical stabilization of these reactive belite forms can be carried out by the addition of many elements (K, Na, Fe, Mg, S, Ba) in various compounds, e.g. BaCl₂, Na₂SO₄, NaF, K₂CO₃, Na₂CO₃, FeSO₄ (3,5,6–11). Stabilization of the reactive belite structure was also achieved by the rapid cooling of clinker (3,8).

Although β-belite hydratation is much slower than that of alite, the later strength of the belite-rich and alite-rich cement pastes can be similar (3,12). Lower reactivity of belite is enhanced by the production of belite-rich cements, also containing other hydraulic phases and/or pozzolans. Belite sulfoaluminate cements and belite sulfoferroaluminate cements can contain whichever of the phases C₂S, C₁₂A₇, CA, C₂A₃S, C₄A₃S, C₄AF, CS, depending on clinker chemical composition (3,12–16). Roman cements, key building material in the nineteenth century, are produced by burning of limestones rich in clay minerals, below their sintering point. Mineralogical compositions of Roman cements are variable, strongly depending on raw material composition. Roman cements typically contain phases C₂S, quartz, free lime and amorphous aluminosilicates. Other phases can be CS, C₃S₂, C₃A, C₂AS and C₄AF; C₃S is excluded because of the low temperature during burning. Currently, Roman cements are especially manufactured for repairing historical monuments (17–21).

Several methods of the preparation of belite cement or pure belite have been studied closely in laboratory scale. The most common process, corresponding to industrial production, involves the blending of powdered raw materials eventually followed by agglomeration, and final burning of the mixture (8,12,22–25). Better homogenization can be achieved by dispergation of the raw materials in water or ethanol (5,9–11), especially under ultrasound treatment (6,26). A hydrothermal method for belite cement or pure belite preparation is more sophisticated and is also often applied. This method is based on hydrothermal synthesis of hydrated precursors (one or more items among C₂SH, CSH, C₃ASH₄ etc.) and on their subsequent burning (1,7,9,27–36). The described conditions of hydrothermal synthesis (temperature, pressure and time) are various; extending from ambient temperature and pressure and a long time (up to one month) (7) to high pressure and high temperature at short times (17 bar, 250°C, 4 h) (28). The hydrothermal process can be carried out not only in water but also in alkaline solutions (NaOH, KOH) (1,9,31,36). Other methods for pure belite preparation are based on a sol-gel process, using calcium nitrate tetrahydrate and silica sol to produce gel. The gel is dried and calcined (37–43). Zeng (44) prepared belite by the co-precipitation of calcium silicate in CaCl₂-Na₂SiO₃-NaOH solutions and subsequent calcination of the formed hydrated precursor.

The present work is focused on the preparation of β-belite by a solid-state reaction, using raw materials produced in high volumes in industrial scale: powdered limestone, amorphous silica and liquid alkali silicates (water glasses). These alkali silicates were used here for their ability to bind effectively powdered raw materials, and as a natural source of reactive SiO₂ and of alkali ions, the latter having been reported as β-belite structure stabilizers. Systematic investigation was carried out to determine the influence of C/S ratio, reaction temperature and the sort of silica and alkali ions used on phase composition and free lime content.

2. MATERIALS AND METHODSTOP

2.1. MaterialsTOP

The raw materials used for dicalcium silicate preparation were partly powdered, such as synthetic silica (VP4; AV EKO-COLOR, s.r.o., Czech Republic), silica fume (České lupkové závody, a.s., Czech Republic) and limestone (Omyacarb 5VA; Omya), partly liquid, such as potassium silicate (Z. Ch. Rudniki S.A., Poland) and sodium silicate (Vodní sklo, a.s., Czech Republic).

The chemical compositions of powdered raw materials, determined by X-ray fluorescence, are presented in Table 1. The specific surface area obtained by the BET method, specific gravity and bulk density of powdered raw materials are shown in Table 2. The mineralogical compositions are given in XRD patterns in Figure 1. Particle size distributions are shown in Figure 2 and pore size distributions in Figure 3. The chemical compositions of liquid alkali silicates are presented in Table 3.

Analytical grade potassium hydroxide (Lachner) and distilled water were used to prepare a 25% KOH solution.

2.2. Analytical and testing methodsTOP

The chemical compositions of powdered raw materials were determined by X-ray fluorescence (BRUKER S8 Tiger).

A BRUKER D8 Advanced X-Ray diffraction system (XRD) equipped with BRUKER SSD 160 detector and operating with Cu-Kα radiation at 40 kV and 25 mA was used for analysis of raw materials and prepared dicalcium silicates. XRD scanning was taken at the 2θ = 0.02 step over an angular range from 5° to 70° with 1 s counting time.

A Mastersizer 2000 laser diffraction particle size analyser (MALVERN Instruments) was used to determine size distribution of powdered raw materials. Agglomerates were disrupted by ultrasound treatment.

Pore size distributions of powdered raw materials were determined using AutoPore 9510 mercury intrusion porosimeter (Micromeritics), which operates with pressures from 0.01 MPa to 414 MPa.

A gas sorption analyser Autosorb iQ from Quantachrome was used for the determination of specific surface area by the Brunauer-Emmett-Teller method (BET).

Free lime content in dicalcium silicate samples was determined by a glycerine-alcohol test.

The morphology of silica fume and synthetic silica was studied by a scanning electron microscope Mira 3 from (TESCAN).

An inductively coupled plasma optical emission spectrometer OPTIMA 8000 (Perkin Elmer) was used to determine the content of micro-elements and K/Na ratio in liquid alkali silicates. Total content of alkali metals (Na, K) and content of SiO₂ in alkali silicates were determined by conventional acid-base titration methods; the reason for their application being higher accuracy at higher concentrations compared with other methods.

Specific gravity was determined by the pycnometric method.

2.3. ProcedureTOP

Ten mixtures M1–M10 were prepared by hand stirring in a vessel. Compositions of these mixtures, which are given in Table 4, were determined by calculation according to the required C/S ratios. The M8 mixture was a plastic material that was divided into amounts of about 20 g and dried at 120 °C. The others mixtures were compacted into tablets (diameter 40 mm, weight 20 g) under 20 kN using a hydraulic press. All agglomerated mixtures were burnt at a designed temperature in an electric muffle furnace for the required time, and then allowed to cool very slowly (overnight) inside the furnace. The heating rates were the maximum possible, which comprised the interval 8–12 °C/min. Twenty β-belite samples B1–B20 were obtained under the conditions shown in Table 5. Chemical compositions of the β-belite samples were found out by the calculation are given in Table 6.

3. RESULTS AND DISCUSSIONTOP

3.1. Effect of CaO/SiO₂ ratioTOP

All the samples B1–B6 with C/S ratio from 1.6 to 2.1 were white. Figure 4 shows their XRD patterns. The sample B5 (C/S = 2) contained mostly β-belite (larnite, ICDD 330302) and a small amount of lime (ICDD 821690). Sample B4 (C/S = 1.9) also included C₃S₂ (rankinite, ICDD 701138), whose content increased (while β-belite content decreases) with decreasing C/S ratio. The rise of the C/S ratio above 2 led to the increase in lime content.

Free lime assessments (Figure 5) agreed with the results of XRD analysis described above. Free lime content was lower than 1 % up to C/S = 1.8; further increase of C/S ratio led to increasing free lime content. When the C/S ratio was 2, which is the stoichiometric ratio of these components in β-belite (sample B5), free lime content was 4.19 %, which means that approximately 93 % of CaO was bound.

3.2. Effect of burning temperatureTOP

The samples B5, B7, B8 were white in colour, while samples B9, B10 were light green. The sample B11 was amorphous green glass, obviously because of the applied temperature 1400 °C, which was higher than the melting point of mixture M5. The development of phase composition with burning temperature is shown in Figure 8. β-belite was present in all the samples burnt at temperatures from 800 °C to 1300 °C; its content rose up to 1100 °C, then held at 1300 °C. Another form of dicalcium silicate (ICDD 310297) was formed above 1250 °C. The sample B8, burnt at 950 °C also contained wollastonite (ICDD 431460) and probably pseudowollastonite (ICDD 896463), while the sample B7 burnt at 800 °C did not. β-belite is thus formed either by direct reaction of two molecules of CaO with an SiO₂ molecule (sample B7) or by reacting one molecule CaO with CaSiO₃ (wollastonite, sample B8), which is under certain conditions an intermediate of β-belite formation. Both of these mechanisms of β-belite formation were described (22,44) and the presence of wollastonite in the samples of β-belite or belite cements was documented (9, 22, 28, 44). The XRD patterns (Figure 6) did not present any α-belite and γ-belite peaks.

The effect of burning temperature on the free lime content is shown graphically in Figure 7. Free lime rapidly diminished to the value 4.19 % up to the temperature 1100 °C. Further temperature increase brought about only moderate decrease in the free lime content.

Typical morphologies of the samples prepared at different temperatures are shown in Figures 8–10.

The samples B7 and B8 are not homogenous. Spherical particles of silica fume are clearly visible in Figure 8. SEM morphologies of the samples B5, B9 and B10 are similar – materials are homogenous and particles are sintered (Figure 9). Figure 15 shows a glassy surface (sample B8), which was formed at temperatures exceeding the melting point of the mixture M5.

3.3. Effect of raw materialsTOP

The samples B5 and B12–B15 were prepared from various mixtures of raw materials, but the C/S ratio and the calcination procedure were kept the same. XRD patterns of these samples and the results of free lime determination, compared in Figure 11 and Figure 12 show differences in free lime content. The color of the sample B13 was light green, sample B14 was brownish and the rest of the samples were white.

The sample B5, prepared from a mixture of limestone, silica fume and potassium silicate had the lowest content of free lime. The replacement of potassium silicate with sodium silicate (B13) or potassium hydroxide (B15) led to a slight increase in free lime content. A significantly higher increase was caused by the replacement of silica fume with synthetic silica (B12). The highest lime content in the sample B14 demonstrates a significant influence of alkaline ions on the solid-state reaction rate. X-ray diffraction patterns are consistent with the results of free lime determination. All samples contained β-belite. The samples B15, B12 and B14 contained lime, the amounts of which increased respectively. The sample B13 probably contained sodium calcium silicate (ICDD 731726) and the sample B14 cristobalite (ICDD 391425). The diffraction pattern of the sample B12 explains why it contains more free lime than the sample B5, although synthetic silica has significantly higher specific surface area than silica fume (Table 2). The presence of rankinite in the sample of B12 indicates a local excess of SiO₂ (see Figure 4). The reason for a higher content of free lime in the sample B12 is thus sample inhomogeneity rather than low reactivity of synthetic silica. Synthetic silica is less miscible with other components of the mixture compared to silica fume, due to considerably lower bulk density (Table 2).

3.4. Effect of burning timeTOP

The samples B5 and B16–B20 were prepared from the same mixture M5 (C/S = 2) by burning at 1 100 °C for various times (0.5–16 h). All of these samples were white. The influence of burning time on mineralogical composition and free lime content are shown in Figure 13 and Figure 14. All samples contained predominantly β-belite. The presence of lime and rankinite was clearly visible in the diffraction patterns of samples burned for less than 4 hours. The results of free lime content determination agreed with the results of XRD analysis. Free lime content rapidly decreased within the time of calcination 2 h to 4.80 %. Further prolongation of burning time did not change free lime content significantly. The presence of free lime residues in all samples indicates that the cause was not an excess of lime in the mixture (see Figure 5), neither slow reaction rate (see Figure 7 and Figure 14), but only inhomogeneities in the mixture of raw materials.

4. CONCLUSIONSTOP

β-belite with low content of free lime (below 5 %) was prepared by burning a mixture of powdered limestone, silica fume and liquid potassium silicate (water glass) with a ratio C/S = 2 and K₂O content about 3,5 % at temperatures from 1 100 to 1 300 °C for more than 2 h. Increasing C/S ratio above 2 led to the an increase in free lime content. Rankinite was formed at a C/S ratio under 2; its content increased (while β-belite content decreases) with decreasing C/S ratio. A decrease in the burning temperature to 950 °C resulted in an increase in free lime content and formation of wollastonite – an intermediate of β-belite formation. Liquid potassium silicate in the reaction mixture was replaced with liquid sodium silicate or potassium hydroxide, without substantial deterioration in β-belite purity. The replacement of silica fume with synthetic silica with a low bulk density did not prove as expected because of low miscibility with other raw materials.

The described procedure for β-belite preparation is very simple and uses raw materials produced in large volumes. Therefore, it indicates high potential for industrial application, especially if impurity content is reduced by optimizing the amount of alkali metal in the reaction mixture.

ACKNOWLEDGMENTSTOP

The author thanks Karol Bayer (Faculty of Restoration, University of Pardubice) for providing SEM pictures.

The publication is a result of the project Development of the UniCRE Centre (project code LO1606) which was financially supported by the Ministry of Education, Youth and Sports of the Czech Republic under the National Sustainability Programme I.

REFERENCESTOP