Early age and long‐term properties of mortars containing metakaolin and limestone powder as SCMs

In this study we focus on the development of plastic shrinkage and capillary pore pressure on mortars containing metakaolin and limestone powder as cement replacement. Isothermal calorimetry was also used to compliment the early age properties of the investigated mortars. Strength development and phase assemblage using thermal analysis were also investigated. Five different replacement levels of metakaolin and LSP and two different amounts of SP were used. Our data show that the shrinkage strain values increased as the amount of SCMs increased, while capillary pressure showed a mixed trend. The hydration kinetics did not appear to be influenced much from the SCMs used. Strength development improved in general when metakaolin was used while decreasing when LSP was added, as already expected. The TGA data show an increase in the consumption of portlandite as the amount of SCMs was increased.


Introduction
The cement industry alone accounts for 6 -8 % of CO2 emissions globally [1].Due to current sustainability requirements, many approaches for the reduction of these emissions arising from cement production are currently under investigation.Some of these approaches are discussed by Gartner and Hirao in a review paper [2].They have listed four options that the cement industry could utilise to face these sustainability requirements: the use of alternative fuels, the use of low-carbon supplementary cementitious materials (SCMs) to partially replace cement clinker, the development of low-carbon binders not based on cement clinker and capturing the CO2 produced from cement plants.Of these four options, the use of SCMs (either traditional or alternative) is probably the one where researchers are focused the most.Being by-products from other industries, or naturally occurring materials, SCMs can be secured relatively easily (although their general availability is of concern).Together with their potential in reducing CO2 emissions and the fact that they are simply added to Portland cement are most likely the reason why so much research has been focused on them.However, care has to be taken when incorporating SCMs into the mix design, since they tend to modify the fresh and hardened properties of concrete.Having high fines and very high specific surface areas, their influence on fresh properties could be significant.For instance, the rheology of a Portland cement paste differs widely from that of a blend of calcined clay, limestone powder and cement [3].It was found out that the increase in the SCMs content resulted in an increase in plastic viscosity, yield stress, cohesion and adhesion in these blends.Although some focus on the fresh properties of the systems containing SCMs is currently afforded, this still leaves plenty of gaps in understanding their role in these systems.As of the time of writing this paper, to the best knowledge of the authors, the influence of SCMs on early-age shrinkage, has not been extensively investigated.Early age shrinkage or plastic shrinkage (as it is often referred to since it happens during the plastic state) is caused by the development of capillary pressure in the fresh material due to the evaporation of water [4].The development of this capillary pressure depends not only on the water evaporation but also on the particle size distribution, as already reported by Slowik et al. [5].The fine particles of SCMs would modify the PSD of the blends influencing thus the early age shrinkage through capillary pressure development.Therefore, it became crucial that we include early-age shrinkage and capillary pressure investigation in the other investigation methods for this study.

2
Materials and methods

Raw materials and mix design
The mortars investigated in the study were composed of Portland cement (PC) of type CEM I 42.5 R and a fine aggregate with a maximum grain size of 1mm.The paste content of the reference mortars was 35%.Five replacement levels for both, metakaolin and limestone powder were used (5 -25%, with a 5% increment step).The particle size distribution of the raw materials is given in Figure 1.To improve the workability of the mortars, two amounts of polycarboxylate ether-based superplasticizer of type Melflux 2651 F (BASF Polymers, Germany) were additionally incorporated into the mix design.The SP amounts were 0.5% and 1% (by cement wt.) and were kept constant at these two percentages for comparison purposes.
The water-to-binder ratio (with both SCMs considered as binders in this context) was set to 0.35 and was not altered.All mortars were prepared in a Hobart mixer with a bowl capacity of 5l and two grading speeds, with the lowest one being 62 ±5 rpm and the highest one 125 ±5 rpm (specifications taken from DIN EN 196-1).Before mixing with water, all constituents (cement, aggregate, SCMs) including SP (since the type used in the study is supplied as a powder) were dry-mixed for two minutes to obtain a homogenous mass.Once water was added to the dry mix, the mass was stirred for 60 seconds at the lowest speed after which it was stopped (for 30 seconds approx.) to scrape the walls and the bottom of the bowl.The stirring was then resumed at the highest speed for 90 seconds.

Early age shrinkage
The early age shrinkage of the mortars was followed for at least 24 hours via LVDT transducers.The fresh material was poured into the so-called shrinkage drain produced by Schleibinger (Germany).The shrinkage drain is made of a U-shaped stainless-steel profile with dimensions 40 × 60 × 250 mm.Two removable anchors are mounted on each end of the channel.In one end the removable anchor is fixed, while in the other the anchor is mounted to a plate which is allowed to slide.The movable plate is connected to the LVDT transducer via a steel shaft which slides through three wheels before contacting the transducer.To avoid friction between the mortars and walls of the drain channel, the drain wall is covered with a neoprene sheet.
Once the fresh material was placed inside the drain channel, the measurement was started and the motion of the movable side was recorded through the transducer.The data obtained were registered and stored in the data logger as standard ASCII text files.These files were then further manipulated through other data processing software.

Capillary pore pressure
Capillary pore pressure was measured using a pressure sensor of type AP-10S (Keyence, Japan) capable of detecting pressure in a range of ±100 kPa.The sensor is equipped with a stainless-steel pressure element (diaphragm) that deflects when pressure is applied to it.The deflection of the pressure element causes an electrical signal which is converted into pressure values from an amplifier.This amplifier sends then the values to a data acquisition system (NR-1000 from Keyence) where the data were stored.A hard-plastic tube with an open conical tip (2 mm diameter) was connected to the pressure sensor.This tube was filled with deaerated water and care had been taken so that the connections were completely airtight.Once the mortars were placed in the drain channel, the plastic tube was positioned vertically inside the mortar at a depth of about 20 mm.The pore pressure was then recorded for at least 24 hours, with data collected every 30 seconds.

Isothermal calorimetry
The hydration kinetics of the investigated mortars were quantified through isothermal calorimetry.Immediately after mixing, sample holders with approx.10g of fresh material were each filled and inserted inside the calorimeter which contained three conduction channels.Three sample holders were filled from the same composition and inserted inside the calorimeter to be tested.Their average heat of hydration was then plotted as a single curve for the corresponding mortar compositions.

Strength development
To follow the strength development of the investigated mortars, the fresh material was cast into standard strength moulds (40 mm x 40 mm x 160 mm) as per DIN EN 196-1.The moulds are then placed inside a humidity chamber with 100% RH for 24 hours.After that, the samples are de-moulded and stored underwater before being tested at different curing ages (1d, 7d and 28d).

Thermogravimetric analysis (TGA)
TGA was used as a broad overview of the phase assemblage and also for the detection of the pozzolanic reactivity.Around 15 mg of ground mortar was placed inside an alumina crucible and heated up in a TG-50 device (Schimadzu, Japan).The temperature ranged from 30 to 1000 °C with a heating rate of 10 °C/min under N2 atmosphere.

Early age shrinkage and capillary pore pressure
The influence of metakaolin on early age shrinkage and capillary pore pressure when the SP amount was 0.5% and 1% is given in Figure 2. As can be seen from the graph, replacing cement with parts of MK increased the shrinkage strain values considerably.and d).Note that the data of capillary pore pressure after the pressure release were removed for better readability of the curves.
Interestingly enough, the highest shrinkage values are seen for the lowest replacement rates of Portland cement with MK (namely 5 and 10% of MK) when the SP amount was 0.5%.As the amount of MK was further increased, the shrinkage values started dropping down, with the highest substitution level of MK showing very close shrinkage values to the reference sample.A clear trend was seen also in the corresponding values of the capillary pore pressure.
As the amount of MK is increased, the capillary pressure breakthrough (the point where air enters the pore system and a release of pressure is seen) was observed earlier.
This was in good agreement with the onset of the shrinkage strain in the corresponding shrinkage curves.When looking at the maximum pressure values, however, a small disagreement with the shrinkage values was observed.The reference sample showed quite a high value of the maximum pressure, while the shrinkage strain values were fairly low.
When the amount of SP was increased to 1%, the influence of MK showed the opposite effect on early-age shrinkage.
In this case, the shrinkage strain values increased with the increase of the substitution rate.A slight delay in the onset of the shrinkage strain when the amount of SP was higher could also be seen.This could be related to the delay of the hydration reactions caused by the presence of admixtures in general [cite something here].As for the capillary pore pressure, also in this case a discrepancy with the shrinkage curves was observed.The pore pressure values when an amount of 20% of MK was used did not seem to fit the general trend since the corresponding shrinkage curve has the highest values, thus the same was expected for the capillary pressure values.The breakthrough times, however, appeared to be in good agreement with the corresponding shrinkage curves.
The results when LSP was used to partially replace the cement are shown in Figure 3. Same as in the case of MK, two amounts of SP were used (0.5 and 1%).Also, here, the shrinkage strain values were considerably higher than the reference when LSP was added.When the SP amount was 0.5%, the highest shrinkage values were seen for the lowest replacement rates as well.As the amount of LSP was increased the shrinkage of the mortars was mitigated.
The development of the capillary pore pressure followed the same trend more or less, except the values for the reference sample and the case when 15% of LSP was used.In these two cases, the pressure values were higher than expected.
For the case where the SP content was increased to 1%, the highest shrinkage values were seen for the lowest LSP replacement rates, the same as the case where the SP amount was 0.5%.Same as in the case of MK, the data points of capillary pore pressure after the pressure release, were removed for better readability.
This is in contrast to the results when MK was used in combination with 0.5% SP, as already described above.However, the total shrinkage for this particular case did not appear to be considerably influenced by the amount of LSP used.Furthermore, the shrinkage values were quite lower than the case when SP was 0.5%.This was also in contrast to when MK was used to replace the cement.The corresponding capillary pressure values, on the other hand, varied considerably, not following the same trend as the shrinkage results.The onset of the pressure breakthrough appeared to fit the shrinkage onset in general, with some exceptions still.

Isothermal calorimetry
Figure 4 summarizes the calorimetry results of the mortars containing both, MK and LSP at the two different SP levels used in the study.Overall, the addition of the SCMs did not appear to influence considerably the hydration kinetics.However, some minor effects from each SCM could still be depicted.For instance, when MK was present, the heat flow values were almost always lower than the reference sample when both amounts of SP were used.When the SP amount was 0.5%, the addition of MK appeared to slightly retard the main hydration peak in some cases, with this effect being more significant when 10 and 15% MK was used.The opposite was seen in the case of SP 1%, where a slight acceleration was noticed as the amount of MK increased.Although it has been reported that SCMs tend to broaden the main hydration peak due to …[find the article and cite], the opposite was seen in our case.
The situation appeared slightly different when LSP was used to replace the Portland cement.In case when an SP amount of 0.5% was used, both acceleration and retardation of the main peak could be seen from the addition of LSP.Whereas just a retardation effect from the addition of LSP was seen when the SP amount was increased.However, as already mentioned earlier, the effect from the addition of both SCMs did influence considerably the heat of hydration of the investigated mortars.The minor effect seen in the results could be just a consequence of the socalled filler effect that has been reported for a while now [6], [7], and not so much from the reactivity of the SCMs themselves which is usually negligible during the first of hydration, as already stated by Lothenbach et al. [8].This however is not conclusive since no other investigation techniques were used in this study to confirm the filler effect and the lack of reaction from the SCMs during the first day of hydration.

Strength development and TGA
The commonly used strength activity index is probably one of the easiest ways to confirm the presence of pozzolanic reactivity of SCMs.Several other tests (see a review paper from Juenger and Sidiqque on the matter [9]), some of which are of a chemical nature, are also often used as an easy assessment of pozzolanicity.These tests, however, could be related more to the short-term pozzolanic activity rather than the long-term one [10].The benefit from the pozzolanic activity in the strength development of the study mortars was significant in our case when MK was used as an SCM (see Figure 5).After one day of curing, the substitution of Portland cement by MK did not appear to improve the strength of either, mortars containing 0.5% SP or mortars containing 1% SP.As the amount of MK was increased, a general decrease in strength development could be seen, probably due to some dilution effect at this very early age.After 7 days of curing, however, a significant improvement of strength when MK was present could be seen for all the substitution levels.The maximum value of strength in the case of 0.5% SP was seen when MK equalled 10%.The strength of the mortars increased further after 28 days, following always the same trend as the 7 days values.Same as for the 7 days curing age, the highest value was seen for the 10% substitution rate, a value seen also in another study which used the same type of MK [11].Beyond this substitution value, it appeared as if the mortar was saturated in MK, so no further benefit from the substitution could be seen.In contrast to that, when SP was increased to 1%, a progressive increment of the compressive strength as the amount of MK was increased could be seen.The highest strength values, in this case, were seen for the highest amount of MK, namely 25%, for the curing ages of 7 and 28 days.The strength values for the SP amount of 1% were however lower than those of the SP amount of 0.5%.Normally the addition of admixtures improves the strength of concrete as already reported by other investigators [12], [13], but in the case of the increased value of SP, the strength decreased probably because the mortars were already saturated on SP.
The compressive strength of the mortars containing LSP as an SCM is shown in Figure 6.The situation appeared differently in this case, as already expected.The strength of the mortars decreased in all cases as the amount of LSP increased.This is because LSP is an inert powder and does not take part in any pozzolanic activity.Therefore, its addition to Portland cement will essentially decrease strength through what is known as the dilution effect.The strength values of the compositions containing less amount of SP were higher, as in the case when MK was used.Figure 7 shows the Portlandite amount of the investigated mortars at 1 and 7 days of curing.The data were extracted from the TGA analysis, which can also be used to detect pozzolanic activity through the consumption of Portlandite.Indeed, when MK was present, some consumption of portlandite could be seen even after 7 days of curing for all substitution levels when the SP amount was 0.5%.When the SP amount was increased, pozzolanicity could be seen from TGA only for the substitution levels of 10% and above.In the case of LSP, no consumption of CH could be seen, as already expected.

Conclusions
In this study, the effect of metakaolin and limestone powder on some early age and long-term properties of mortars with two different levels of superplasticizers was investigated.The early age properties included early age shrinkage and the development of capillary pore pressure, together with the hydration kinetics followed through isothermal calorimetry.The long-term properties included strength development at different curing ages and the presence of pozzolanic activity investigated through thermogravimetric analysis.The following conclusions could be drawn from the study: 1.Early age shrinkage was considerably influenced by the addition of MK and LSP.High shrinkage values were seen for the lowest amounts of MK in the case when SP was 0.5% and the shrinkage values decreased as the MK amount increased.The opposite was seen when the SP amount was increased.In case of LSP, for both amounts of SP, the shrinkage values were higher for the lowest substitution values and started decreasing as the amount of LSP increased.2. The development of the negative capillary pore pressure as the main factor causing shrinkage during the plastic stage, was not entirely in agreement with the shrinkage data, although very often similar trends were seen.This could be because capillary pore pressure is a sensitive investigation method and could potentially be easily influenced by many factors.
3. The hydration kinetics of the mortars did not appear to be significantly influenced from the addition of the SCMs.The minor differences observed in the calorimetry curves could arise from the filler effect of SCMs reported also from other investigators.4. The strength development was significantly improved after 7 and 28 days in all cases when MK was used to substitute Portland cement, whereas LSP showed adverse effects as expected.The strength improvement in the case of MK was related to pozzolanic reactivity, confirmed also from the consumption of CH from the TGA data, whereas the decrease of strength when LSP was added was attributed to the dilution effect.
From a future perspective, it can be said that the combination of the two, MK and LSP to study their combined effect on the early age and long-term properties of the same mortars would be a point of interest.The study could be broadened to include other properties during the early age and long term.

Figure 1
Figure 1 Particle size distribution of the raw materials used in the study.

Figure 2
Figure2Influence of MK on the shrinkage and capillary pressure of mortars containing 0.5% amount of SP (a and b) and 1% SP (c and d).Note that the data of capillary pore pressure after the pressure release were removed for better readability of the curves.

Figure 3
Figure3 Influence of LSP on the shrinkage and capillary pressure of mortars containing 0.5% amount of SP (a and b) and 1% SP (c and d).Same as in the case of MK, the data points of capillary pore pressure after the pressure release, were removed for better readability.

Figure 4
Figure 4 Isothermal calorimetry results for MK mortars when SP amount was 0.5% (a) and 1% (b) and for LSP mortars for the same amount of SP (c and d).

Figure 7
Figure 7Portlandite consumption at different curing ages taken from thermogravimetric analysis (TGA).Diagrams a) and b) belong to the mortars containing MK at the given substitution levels when SP was 0.5 and 1%, respectively.Diagrams c) and d) belong to the mortars containing LSP when SP was 0.5 and 1%, respectively.