Laboratory synthesis of C3A on the kilogram scale: Preliminary results

This paper focuses on tricalcium aluminate (C3A) synthesis on the kilogram scale in laboratory conditions. The proposed design aims to produce more than 97 wt% of C3A. The remaining three wt% could be associated with residual lime, CaO, which is preferable over mayenite (C12A7). The maximally used temperatures for the synthesis were 1350 °C and the borderline temperature of 1400 °C, beyond which C3A and liquid phase can be formed. The preliminary results showed that using the CaO:Al2O3 ratio of 62.5:37.5 and a temperature of 1400 °C, it was possible to achieve pure cubic C3A of the desired quality. Moreover, it was shown that the time for which was the maximal temperature kept before removing the samples from the furnace was an essential parameter that influenced most significantly the amount of mayenite in the structure of C3A.


Introduction
Tricalcium aluminate (C3A) is one of the main components of ordinary Portland cement (OPC), which makes up 15 wt% of its composition [1].Its uncontrolled rapid reaction with water leads to a fast setting (flash set) during which initial precipitation of solid hydration products, namely calcium hydroaluminate (AFm) phases, takes place.Subsequently, katoite can be formed during this early age.Consequently, fresh concrete loses its fluidity and workability, resulting in its irreversible poor strength development [2].To retard these fast reactions, a small amount of solid calcium sulfate (usually 2-5 wt%) is added to clinker during cement production [3].Interestingly, when Ca(OH)2 and calcium sulfate are present together, the hydration of C3A is more retarded than by calcium sulfate alone, as further discussed, e.g., in [4].Despite the importance of controlling the behavior of cement-based materials immediately after mixing with water, the main progress in understanding hydration processes has usually been made with a focus on the dominating phase of tricalcium silicate (C3S, alite) [5], whereas the hydration mechanisms of C3A have not been studied to this extent.Pure C3A is generally cubic.In OPC, alkalis are also incorporated into the C3A structure (for example, Na2O), which can change the symmetry of C3A from cubic to less reactive orthorhombic, tetragonal, and monoclinic [4].Similarly, like clinker production, the synthesis of C3A is done at elevated temperatures.To achieve the desired composition of the resulting product in the system CaO-Al2O3, a knowledge of the relevant high-temperature phase equilibria [2], as displayed in figure 1, is necessary.Pure C3A contains 62.3% CaO and 37.7% Al2O3 [2].Based on figure 1, it can also be deduced that C3A can be produced at temperatures above 1000 °C up to 1400 °C, beyond which C3A and liquid phase start to form.Additionally, if the mix composition moves slightly to the left from the line defining pure C3A (more CaO in the mix design), residual lime (CaO) occurs as a side product.If the mix contains a higher proportion of Al2O3, besides C3A, also mayenite C12A7 is generated.Its presence is not acceptable, as based on Quennoz [6], this phase has a substantial impact on the kinetics of the C3A-gypsum reaction leading to a flash set.And thus, its formation should be avoided in order not to modify the mechanisms of the studied processes once the synthesized pure C3A is added to more complex systems.It is preferable to be on the side with the formation of C3A and a few percent of residual CaO.Based on [2].
Various attempts have been made to prepare the mixture for the C3A synthesis.For example, Zunino and Scrivener [7] used calcium carbonate CaCO3, keeping the stoichiometric CaO content of the raw mix at 62.5% to prevent the formation of C12A7.The sintering temperature was chosen as 1450 °C and was maintained for 4 hours; thereafter, samples were air quenched.Cubic C3A was obtained with 0.4-0.7%content of CaO.Quennoz [6] increased the CaO content up to 62.8%.Consequently, higher amounts of CaO (1-2.5%) were observed besides pure C3A compared to [7].Whereas Zunino and Scrivener [7] used a heating rate of 7 °C/min with no plateau during heating, Quennoz [6] used a heating rate of 3 °C/min, and after reaching 1000 °C, this temperature was kept for 8 hours in order to ensure complete decarbonation of the CaCO3.The samples were then heated to 1400 °C, and the maximal temperature was kept for another 5 hours.After quenching, the second firing process was needed because some undesired amorphous Al-rich phase was formed during the first synthesis.This might be caused by the shape of the samples and molds (small pellets were used), whereas an easier approach [7] could have been used when the samples stood on the platinum open crucibles.
As can be seen from the comparison of these two relatively recent works, the laboratory synthesis of C3A is not straightforward, and a general protocol that could be easily adapted worldwide has been missing.In this paper, these two approaches are combined with the intention of looking for an effective and easily reproducible procedure for the laboratory synthesis of C3A, which would be possible to produce on the kilogram scale within a workweek.

Raw materials and the synthesis process
For the synthesis of C3A, calcium carbonate precipitated A.G., aluminium oxide A.G. (both Penta Chemicals Unlimited), and distilled water were used for the sample preparation.The mix design was chosen based on the previously mentioned studies [6,7], and it is summarized in table 1.After several trials of the sample preparation, the amount of distilled water was adjusted to achieve yogurt-like consistency.The desired amounts of powders and water were put in a ball milling jar with a volume of 3 L, and it was homogenized using a rolling machine for 24 hours.After that, the liquid mixture was filtered using a salad spinner to separate the material from the milling balls.The samples were then poured into cylinder molds.The quantity was kept approximately the same for each cylinder.Finally, the samples were put in an oven and were dried for 48 hours at 60 °C, and then, the temperature was increased to 105 °C until constant weight was achieved.Dried samples were placed into the platinum crucible and were heated (one sample per firing) from room temperature up to 1350 °C or 1400 °C with a heating rate of 7 °C /min.The maximal temperature was kept for 0.5, 1, and 2 hours.Then, the sample was carefully removed from the oven and was quenched using forced air. Figure 2 displays selected samples immediately after they were removed from the oven.It can be seen that the samples cracked during the burning but remained compact and stable, allowing safe manipulation.The cooling process was finished after approximately 15-20 minutes.Each sample yielded 180-200 g of C3A, which means that by using this setting, it is possible to produce a kilogram of C3A per week.Since C3A is a very reactive material, the synthesized samples were put in vacuum-tight plastic bags until they were crushed in an agate bowl (reaching a size of approximately 5-10 mm per piece).Then, the samples were milled into a fine powder using a disc mill twice for 30 s.The fine powder was put in a vacuum-tight plastic bag until further analyses were done.The samples were labeled as follows: Design-maximal temperature-time for which the temperature was kept.

Methods
The mineral composition of synthesized C3A was analyzed by means of X-ray powder diffraction (PANalytical Aeris).The diffractometer was equipped with a CoKα source operating at 7.5 mA and 40 kV.The incident beam path consisted of an iron beta-filter, Soller slits 0.04 rad, and divergence slit 1/2°.The diffracted beam path was equipped with a 9 mm anti-scatter slit and Soller slits 0.04 rad.A PIXcel1D-Medipix3 detector was used with an active length of 5.542°.The collected patterns were evaluated by Profex software [8].
The microstructural morphology and surfaces of the synthesized particles were studied by scanning electron microscopy (SEM) using a Phenom XL electron microscope with a CeB6 source operating at 5 kV with a BSE detector.The observations were done on the coated samples.

Purity of synthesized C3A
The XRD patterns of all studied samples are summarized in figure 3.At first sight, it can be seen that the main difference in the mineral composition of all synthesized C3A lies in the presence or absence of the significant pattern appearing approximately around 2θ = 21°, which is associated with mayenite.
It was included in all samples prepared at the lower temperature of 1350 °C, regardless of how long the maximal temperature was kept.When the sintering temperature was increased up to 1400 °C, mayenite still appeared in the samples held at this temperature for 0.5 h, whereas it was almost negligible after 1 h, and finally, its patterns disappeared after holding the maximal temperature for 2 h.The mineral composition of all synthesized C3A was quantified, and the estimated results are displayed in figure 4. All samples prepared at 1350 °C contained relatively low amounts of pure cubic C3A (approximately 87-92 wt%; figure 4A).The crystalline phases of A-1400-0.5 h consisted of only ~80 wt% of C3A, whereas when a longer sintering time was applied, it was possible to obtain the desired amounts above 97 wt%.All samples, except for A-1400-0.5 h, consisted of < 1.2 wt% unreacted CaO (figure 4B).Samples that were prepared at 1350 °C contained 7.7-11.5 wt% of mayenite (figure 4C).The additional time only slightly helped in decreasing its amount.Based on our results, we assume that using this sample design and temperature of 1350 °C, mayenite would always occur in the structure even if the firing time would be significantly prolonged.The application of a higher temperature of 1400 °C significantly reduced the mayenite content: it reached nearly the exact value (7.6 wt%) already after 0.5 h at 1400 °C, when compared to 2 h at 1350 °C.The mayenite content was decreased to 1.7 wt% and 0.6 wt% after 1 h and 2 h at 1400 °C.

Morphology
The microstructure of the selected C3A samples, as analyzed with the help of SEM, is shown in figure 5.The upper part of figure 5 is dedicated to the C3A pieces; the powdered samples are displayed below.Two extremes were compared: the samples sintered at 1350 °C for 0.5 h and at 1400 °C for 2 h.The microstructure of both samples was porous.In the A-1350-0.5h, smaller grains were connected; in the case of A-1400-2 h, the microstructure was seemingly more compact, consisting of larger coral-like objects.When the point analysis and mapping were done (not shown here), it was confirmed that the distribution of Ca and Al elements was homogeneous.The observations done on the powdered samples also showed that the A-1350-0.5h samples had generally smaller grains when compared to the sample sintered at a higher temperature, crushed, and milled in the same way.It was expected since the higher temperature led to the formation of a more solid microstructure.It is worth mentioning that while the higher temperature of 1400 °C generally led to a higher yield of C3A and lower contents of unreacted CaO and mayenite were formed, the resulting larger pieces of C3A, which were challenging to break due to a very condensed structure of these samples, could negatively influence the reactivity of synthesized C3A due to the lower surface area if these are not removed from the powder, e.g., by sieving.When these results are compared to the literature, the newly proposed setting could lead to the faster production of pure cubic C3A.More specifically, 97 wt.% of C3A could be synthesized after ~5:15 h (~4:15 h if less than 1 wt% of mayenite is not a problem), which is a significant improvement compared to Quennoz [6], who synthesized C3A for almost 20 h and consequently for another 12 h to ensure the mayenite would be removed from the pellets.Compared to Zunino and Scrivener [7], a lower temperature of 50 °C was needed, and the time was decreased by approximately 2-3 h.Nevertheless, further study is required to determine the reactivity and other essential properties of the C3A.

Conclusions
In this paper, the preliminary results of the C3A synthesis on the kilogram scale in laboratory conditions are discussed.The aim was to find an easily reproducible set-up to achieve approximately 97 wt% of the pure cubic mineral.The obtained results could be concluded as follows:  The mix design consisting of the 62.5% CaO and 37.5% Al2O3 generally yielded a certain amount of mayenite (over 7 wt%), especially when the samples were sintered at a lower temperature of 1350 °C for 0.5, 1, and 2 h. The sintering temperature of 1400 °C significantly reduced the mayenite content to 1.7 wt% and 0.6 wt% after 1 h and 2 h. The resulting microstructure of C3A was compact but porous, regardless of the temperature and time of sintering.Whereas the samples prepared at lower temperatures consisted of smaller grains connected, the C3A synthesized at 1400 °C had a seemingly more solid structure containing larger coral-like objects. When the sintered pieces were milled using a disc mill, the larger pieces of C3A prepared at 1400 °C were challenging to break due to a very condensed structure.These larger grains could negatively influence the reactivity of synthesized C3A due to their lower surface area, which will be further analyzed.Each studied factor (mixture design, maximal temperature, and the sintering time) influenced the final composition and morphology of C3A, but it seems that the sintering time was an essential parameter in order to influence the amount of mayenite formed in the structure of C3A.Compared to the literature, the proposed sintering process could lead to a faster production of C3A on the kilogram scale within a workweek.

Figure 1 .
Figure 1.The system CaO-Al2O3 modified by the presence of minor contents of H2O and O2.Based on[2].

Figure 4 .
Figure 4.Estimated mineral composition of studied samples based on XRD results in wt%.