C3A Variation in Synthetic Model Cements ‐ Influence on Rheology and Reactivity

The amount of early ettringite crystallization is known to influence the rheology of cementitious materials. It is dependent on various factors, e.g., the reaction temperature, the chemical composition of the cement and the amount of PCE. In this study, we investigate the rheology and reactivity of synthetic model cements with different contents of C3S, C3A and β‐hemihydrate (β‐HH). Their reactivity is studied using in‐situ calorimetry and their rheology is studied using a rotational rheometer. Our results demonstrate that higher C3A contents in the clinker increase the early heat release of the cements which is associated with an increased formation of ettringite. Increased C3A contents from 2.5 wt.% to 12.5 wt.% enhance the early heat by four orders of magnitude. The static yield stress of the different cement pastes was determined with an oscillation based rheometry method 17 minutes after contact with water. Interestingly, the static yield stress increases exponentially with the heat of hydration. These findings have implications for the understanding of OPC hydration, as well as for chemical optimization and the design of OPC‐based materials.


Introduction
The hydration of ordinary Portland cement (OPC) is a complex process.It comprises many chemical reactions which involve the main clinker phases and minor components such as sulfate salts.Among the clinker phases, tricalcium aluminate (C3A) exhibits the fastest initial reactivity and, consequently, has a significant influence on the early properties of fresh cementitious materials, e.g., the setting time or rheological parameters like the yield stress [1,2].Portland cement typically contains 4-10 wt.% of C3A as cubic or orthorhombic polymorph or a mixture of both [3].The early hydration reactions of C3A depend on the presence or absence of a soluble sulfate source.The detailed hydration processes were recently reviewed by Hirsch et al. [4].In the absence of CaSO4, pure C3A and water react very rapidly, precipitating Hydroxy-AFm phases and resulting in a flashset.These precipitating phases compromise the workability of concrete.Therefore, setting retarders (i.e., CaSO4 phases like gypsum, anhydrite, or hemihydrate) are interground with cement clinker.The addition of the calcium sulfate source changes the hydration pathway of C3A, resulting in ettringite precipitation instead of AFm phases [5,6].Recent studies suggest that the adsorption of sulfates or calcium-sulfate ion pairs on an aluminium-rich layer of partially hydrated C3A is responsible for the dampening of the initial C3A reactivity [7][8][9][10].
It is widely agreed that calcium sulfate suppresses the very fast hydration of C3A, resulting in the slower formation of ettringite.Furthermore, ettringite formation significantly affects material properties during cement hydration, particularly the rheology of cement pastes and mortars.Gołaszewski varied the C3A content in cement and found that the yield value of mortars measured by rotational rheometry increased with the C3A content [2].By combining in-situ XRD and rheological experiments, Jakob et al. recently highlighted the importance of ettringite precipitation on the flow behavior of OPC pastes [11].As ettringite is expected to be the first hydrate phase precipitating during the early hydration of cement, it is of great importance for the structural build-up during this period.
To study the viscoelastic properties of cement pastes, such as structural build-up or yield stress, oscillatory shear testing can be performed as a type of rheological measurement.Here, small amplitude oscillatory shear tests (SAOS) allow the investigation of the structural build-up caused by the formation of hydrates during the early period of hydration in OPC pastes [12][13][14].

Correspondence
Prof. Dr. Torben Gädt Chair Chemistry of Construction Materials TUM School of Natural Sciences Technical University of Munich Lichtenbergstraße 4 85748 Garching Email: torben.gaedt@tum.de 1 Chair Chemistry of Construction Materials (TUM School of Natural Sciences), Garching, Germany Oscillation methods were also used to determine the static yield stress of cement pastes.The material is subjected to a gradually increasing oscillatory strain in this case.The linear viscoelastic region (LVE) is the range of strain over which the viscoelastic behavior of the material can be considered linear [15].Subsequently, the static yield stress is calculated from the storage modulus plateau in the LVE region and the detected critical strain [16].We were motivated to study the influence of C3A and a sulfate carrier on the static yield stress of model cement suspensions.
In this contribution, we study the effect of different model cement compositions on their rheological properties.The amounts of cubic C3A and hemihydrate in the synthetic model clinkers were varied.Therefore, we prepared 18 model cements by combining six clinkers that vary in their C3A content with three amounts of β-hemihydrate.By combining in-situ isothermal calorimetry with oscillatory strain sweep tests, the influence of the C3A content and the sulfate amount on the early heat of hydration and yield stress of the model cement pastes were investigated.

Preparation of model clinkers
This study aims to investigate the effect of C3A content in combination with varying amounts of β-hemihydrate (β-HH) on the reactivity of model cements.Therefore, the aim was to prepare six model clinkers containing different amounts of cubic C3A, each mixed with three different amounts of β-HH (3, 4, and 5 wt.%) to obtain 18 model cements in total.
Synthetic clinkers were prepared according to a preparation method for synthetic model clinkers described in the dissertation of Marchon [17].The appropriate amounts of fine quartz powder (SiO2), calcium carbonate (CaCO3), CaO, aluminium oxide (Al2O3), and magnesium carbonate (MgCO3) were homogenized with water (w/s 0.45) and filled into silicon molds.The resulting cylinders were dried and sintered at 1450 °C for 6 h.The sintered bodies were rapidly quenched with pressurized air, crushed, and ground in a ceramic ball mill.The obtained powders were sieved with a mesh size of 90 µm and stored in a desiccator over silica gel until further usage.The sample ID for the samples is used in the following manuscript as an abbreviation for clarity, with the corresponding expected C3A content as numbers.For example, C025 corresponds to the clinker with 2.5 wt.% cubic C3A.

Clinker characterization
All obtained model clinkers were characterized using powder X-ray diffraction (PXRD).Diffractograms for the six prepared samples were recorded on a Bruker D8 Advance diffractometer in Bragg-Brentano geometry (goniometer radius = 217.5 mm) with CuKα1,2 radiation (1.5406 Å).The device is equipped with a VÅNTEC-1 LPSD detector with a 12° 2Θ detector opening.All measurements were conducted at 32 kV and 40 mA.Primary and secondary soller slits were set to 2.5°, and the divergence slit was fixed to 0.2°.The diffraction data was collected between 7 and 70° 2Θ with a step size of 0.016° and a counting time of 0.5 s per step (40 min total scan time).XRD samples were prepared by the backloading technique and were rotated at 30 rpm during the measurements.
Particle size distributions of the six prepared laboratory clinkers were obtained via laser diffraction analysis on a CILAS 1064 Particle Size Analyzer.The powders were suspended in isopropanol and treated in an ultrasonic bath for 10 min before the measurements to avoid particle agglomeration.Additionally, the surface area of the clinkers was determined by Blaine analysis.

Model cement preparation
The laboratory-synthesized clinkers were blended with three amounts of β-hemihydrate (3, 4, and 5 wt.%).Therefore, the clinker powder and the respective amount of hemihydrate were weighed into a plastic bottle.The mix was homogenized in a shaker-mixer (Heidolph REAX 20) at 15 rpm for 4 h.All samples were stored in a desiccator for 14 days until further usage to obtain samples of equal constitution before calorimetric and rheological measurements.The same conventions for the abbreviation of samples are applied to the model cements, as already described in Section 2.1.C025_3, for example, encodes for the model cement sample with 2.5 wt.% C3A and an addition of 3 wt.%β-HH.

Rheological measurements
The rheological properties of the synthetic cements were characterized with a rotational rheometer (Anton Paar MCR 302e) at 20 °C.For all experiments, two serrated plates with a diameter of 25 mm (PP25/P2) were used at a measuring gap of 1 mm.The measurement profile was based on a strain sweep (strain γ increased from 10 -4 to 10 1 %) in oscillatory mode at a constant frequency (ω = 10 rad/s) [16].
The static yield stress (τsta), which describes the force required to initiate flow, was ascertained from the oscillation measurements.The critical strain values γcrit for the cement pastes are determined by selecting the tolerance range of 10 % deviation of the storage modulus around the plateau value in the LVE region.The static yield stress τsta was calculated according to Equation (1) as the product of the average value of the data points of the storage modulus G' plateau before reaching the targeted deviation and the observed critical strain γcrit.
For the rheological measurements, 10.00 g of each synthetic cement was weighed in a cup, 5.00 g of deionized water was added, and the mixture was stirred for 1 min at 1200 rpm using a mechanical stirrer equipped with a propeller-shaped paddle.The paste was left to rest under a wet towel for 13 min.Afterwards, the pastes were stirred for 1 min at 1200 rpm and applied to the serrated plates.The measurement typically started at 17 min 20 s after the first water contact of the cements.

Isothermal calorimetry
The reactivity of the cements was characterized using insitu mixing calorimetry.The experiments were conducted at 20 °C on an eight-channel calorimeter (TAM Air, TA Instruments).The samples were mixed inside the calorimeter using Admix ampoules supplied by the device manufacturer.14.2 g quartz sand (calculated according to Wadsö [22]) was used as reference material with equal heat capacity as the tested samples at w/c = 0.5.4.0000 ± 0.0005 g of the respective model cement was weighed into a glass vial, and 2.0000 ± 0.0005 g of deionized water was weighed into two syringes, and the syringes were mounted onto the Admix ampoule.The ampoules were placed in the calorimeter and were thermally equilibrated overnight.After 10 s of premixing the powder using an electrical motor, the water was added, and pastes were further continuously stirred for 1 min.Experiments were typically conducted for 72 h.The Tian correction according to Equation ( 2) was applied for all experiments (Pc = corrected heat flow, P = measured heat flow, τ = time constant) [23] to correct the measured data for the inertia of the calorimeter and the mixing cell.
The time constant τ for the applied in-situ mixing device was determined to be 329 s.The Tian correction mainly affects rapid heat-flow events and is, therefore, suitable for adjusting the heat-flow signal of the aluminate reaction.

Clinker characterization
The main goal of this study was to investigate the impact of C3A and calcium sulfate content on the reactivity and rheological behavior of model cement pastes.Therefore, six laboratory clinkers containing different amounts of cubic C3A are prepared in the first step.The C3A content increases linearly from 2.5 wt.% to 15.0 wt.% in 5 steps.
The phase composition was determined by Rietveld analysis for all six prepared model clinkers (Table 1).
The determined C3A contents are in good agreement with the expected values.Nevertheless, the C3A amounts slightly exceed the desired values for five prepared clinkers.In this study, the crystalline phase content was assumed to be 100%.Consequently, amorphous contents were not quantified.We speculate that this might explain the minor deviations from the expected values.With the applied cooling technique, low values for β C2S were achieved.Nevertheless, the quenching procedure needs to be improved to avoid the re-formation of C2S and to obtain clinkers only consisting of C3A/C3S mixtures.Additionally, the particle size distributions and Blainespecific surface were determined for the six model clinkers, and the results are summarized in Table 2.All the synthesized samples exhibit comparable values for the particle sizes and the determined Blaine surface.Consequently, the deviations in reactivity and the impact on the rheological behavior of the prepared model cement pastes can be attributed to the variations of the C3A and β-HH contents of the cements and not to differences in particle size [5,24,25].

Model cement reactivity
We determined the reactivity of all prepared model cements by in-situ isothermal calorimetry at w/s = 0.5.The initial heat flow of three cements prepared from the clinker with the lowest C3A content is shown in Figure 1.
Higher sulfate amounts increase initial heat flow, which is only clearly visible for cements with low C3A contents.The different amounts of sulfate do not influence the silicate reaction.For the lowest C3A content, increasing sulfate amounts enhance the initial aluminate peak (Figure 1 left).Hence, the formation of more ettringite can be assumed [11].The small peak to the left of the main hydration peak of the silicate phase can tentatively be ascribed to the rehydration of β-HH to gypsum [26].Without additional in-situ XRD experiments, this assignment cannot be proven.Additionally, increasing the amount of sulfate from 3 to 5 wt% does not change the main hydration peak of the silicate phase for the cements with the lowest C3A content.
To evaluate the effect of the increasing C3A contents in the cements at identical sulfate amounts, hydration curves of three cements with low, median, and high C3A amounts are shown in Figure 2. The increasing amount of C3A from 2.5 to 15 wt.% in the cements leads to an enhanced early heat flow maximum of almost 500 %.However, the increase in the aluminate peak is not proportional to the C3A content in the cement samples.Additionally, with the increase in the C3A amounts, a significant effect on the silicate reaction is found.Therefore, the addition of 5 wt.% β-HH is insufficient to retard the aluminate reaction in the cement with high C3A content of 15 wt.% [27,28].
Consequently, the sulfate depletion peak occurs before the silicate reaction resulting in an aluminate/silicate imbalance and a retarded silicate peak [29].The dissolution of C3A, the dissolution of the sulfate carrier, and the rapid precipitation of ettringite [30] affect the magnitude of the initial heat flow event.Therefore, we conclude that the initial heat flow maximum is driven by the ettringite formation, i.e., more C3A and sufficient sulfate lead to more ettringite and more heat.
We show the cumulative heat values after 15 min and 48 h as a function of the C3A content of the model cement (Figure 3).The initial aluminate reaction is completed after min.Hence, the cumulative heat after 15 min captures the entire early hydration heat maximum.The increasing C3A content increases heat formation in the early hydration period up to 12.5 wt.%C3A.The heat observed for the highest C3A (15 wt.%) amount is lower than the maximum at 12.5 wt.% of C3A.A similar trend is found for the heat values after 48 h.Interestingly, the highest C3A content in the cements decreased the measured heat after 48 h.Again, the early sulfate depletion peak resulted from undersulfation of these cements, which disturbs the aluminate/silicate balance and retards the silicate reaction [28].Additionally, the sulfate addition shows only a relatively small influence on the heat of hydration of the cements.Higher amounts of added β-HH slightly increase the heat values after 15 min and 48 h.In summary, larger C3A fractions in the model cements lead to higher heats of hydration.The correlation for the heat after 15 minutes is linear.The deviation from linearity at higher C3A contents is caused by undersulfation.

Rheometry of model cement pastes
We are interested in the effect of C3A and β-HH on the rheological behavior of the cement pastes.Therefore, the static yield stress was determined according to a protocol recently described by Ukrainczyk et al. [16].The three samples with the highest C3A content were undersulfated and were not investigated by oscillatory rheometry.Three examples of the obtained curves of the strain sweeps are shown in Figure 4.  Based on the results of the rheometric experiments, the values for τsta were determined and were related to the C3A content and sulfate amount added to the cements (Figure 5).An increased C3A content in the cement leads to higher values for τsta, where the values increase exponentially as the C3A amount increases.Again, the varying addition of β-HH has little to negligible effect on the values with no discernible trends.Gołaszewski also found that the increase of SO3 does not affect the yield stress value of mortars in rotational rheometry [2].Again, there is a clear dependence of the determined static yield stress on the C3A contents of the cements.

Correlation between heat of hydration and static yield stress
Both investigation techniques in this study demonstrated a relationship between the C3A content of the cements on the one hand and heat of hydration or the static yield stress on the other hand.We find a linear relationship between the heat of hydration after 15 min and the logarithm of the static yield stress τsta (Figure 6).
According to an additional study, we suggest that the experimental error for the heat of hydration after 15 min is in the range of ±1 J/g for the applied setup [31].An exponential relation between the heat of hydration and the static yield stress of the synthetic cement pastes is found.By comparing spread flow measurements with the hydration rate, Mantellato et al. also found a linear correlation between heat rate and the logarithm of the static yield stress at later ages [32].Due to the internal mixing of the pastes directly in the calorimeter, we can now demonstrate that the static yield stress increases exponentially with the measured heat.In addition, we can show that both determined values, the heat of hydration and the yield stress, are directly dependent on the C3A content in the synthetic model cements.Furthermore, it was found that increasing the addition of β-HH from 3 to 5 wt.% has a much smaller effect on the static yield stress τsta.In summary, we ascribe the increase in heat flow after 15 minutes to increased ettringite formation from C3A and sulfates.Consequently, the increased ettringite amount causes an exponential growth of the static yield stress.
Following the temporal phase development of the hydrating synthetic cements by in-situ XRD can quantify the forming hydrate phases, especially ettringite.This data can lead to a better understanding of the processes involved in yield stress evolution.

Conclusion
This study has shed light on the importance of C3A content on the hydration kinetics and rheology of model cement pastes.By combining in-situ mixing isothermal calorimetry with rheometric experiments, we have demonstrated that increasing the amount of C3A in the clinker leads to an accelerated hydration rate and enhanced static yield stress.Additionally, a correlation was found between the C3A content of the synthetic cements and the heat of hydration and between the C3A content and the static yield stress τsta.We assume that an enlarged formation of ettringite caused by the higher amount of C3A is the reason for the exponential increase in the static yield stress.Interestingly, the different amounts of β-HH have only small influence on the hydration characteristics of the model cements.
Nevertheless, an excessive amount of C3A in combination with insufficient sulfate addition resulted in undersulfation, and consequently, rheometric experiments with a high amount of C3A were not possible.Therefore, the optimization of C3A content is essential for the design of OPC-based materials.For future studies, sulfate content has to be increased for model cements with high C3A amount to avoid undersulfation and to allow investigations on the effect of systemic variation of ettringite-forming components in cement.Additionally, in-situ XRD measurements can be of great interest to combine the phase development during hydration with rheological properties in cements with a systematic variation of highly reactive C3A.

Figure 1
Figure 1 Exemplary calorimetric curves for the three cements with the lowest C3A content and increasing β-HH addition.Left: the aluminate peak, and right: the silicate peak

Figure 2
Figure 2 Comparison of cements with different C3A content and equal β-HH addition.The left panel shows the aluminate peak and the right panel shows the silicate peak

Figure 3
Figure 3 Cumulative heats after 15 min (bottom) and 48 h (top) depending on C3A content and sulfate addition of cements

Figure 4
Figure 4 Exemplary graphs demonstrating raw data of the rheometric experiments in oscillation mode of three cements with different C3A content and equal β-HH additionAs described before, higher amounts of C3A lead to increased reactivity in the early hydration period and, consequently, more ettringite.The increase in early ettringite formation increases the storage modulus by two orders of magnitude from the lowest to the highest C3A (2.5 and 12.5 wt%) content.

Figure 5
Figure 5 Static yield stress obtained by strain sweeps as a function of cements' C3A and sulfate content

Figure 6
Figure 6 Static yield stress (after 18 min) obtained from oscillatory rheometry as a function of the heat of hydration (after 15 min) obtained from isothermal calorimetry

Table 1
Phase contents of the six prepared model clinkers with the respective theoretical C3A content and Rietveld quantities of the crystalline clinker phases

Table 2
Particle size distributions and Blaine specific surface of synthesized model clinkers