Influence of calcined clay on micromechanical properties and creep of hardened cement paste

A highly interesting approach to lower CO2 emissions in cement and concrete industry is to reduce the cement clinker content while using supplementary cementitious materials (SCMs) such as calcined clays. Whereas for classical Portland cement clinker, comprehensive studies are already available on the microstructural development during hydration as well as on the resulting macromechanical properties such as compressive strength or long‐term deformation behavior, such understanding is missing for calcined clays. In this paper, systematic micromechanical investigations using microindentation of hardened cement pastes with various amounts of calcined clay at ages up to 28 days are presented. The micromechanical results indicate a faster strength development as well as a lower overall porosity of the classical clinker system compared to mixtures containing calcined clay. Looking to creep, the indentation creep modulus is higher for pure OPC mixtures compared to calcined clay systems up to seven days, indicating less pronounced creep. Beyond this age, however, the differences in creep behaviour became less pronounced. The outcomes suggest that the reactivity of the calcined clay seems to be decisive when it comes to mechanical properties and creep behaviour rather than the rate of substitution.


Introduction
Considering the efforts of the building materials and construction industry to meet CO2 reduction targets, the substitution of ordinary Portland cement (OPC) by secondary cementitious materials (SCM's) is an important pathway [1].SCM's such as fly ash or blast furnace slag are in use since many years, however, with their availability declining despite increasing demand [1][2].As already shown in numerous investigations, calcined clays (cc) have turned out to be one of the promising alternatives due to their pozzolanic activity and high availability worldwide [3][4][5][6].Several studies reveal promising material properties for a use as SCM's in concrete [5][6].The existing studies primarily focus on pure metakaolin systems, investigating e.g.their effect on strength or durability.Investigations on systems with reduced metakaolin content -such as common in northern Europe -and regarding more complex concrete properties such as creep and shrinkage however are still rare [7][8].Further, the clays available in northern Europe show higher fluctuations in materials properties than e.g.pure metakaolin systems.An important pathway in coping with such fluctuations therefore consists in understanding the chemical and physical interactions during hydration and their effect on the resulting microstructure.For classical OPC, comprehensive investigations on the structure formation in the course of hydration are already available [9][10].This process is characterized by the dissolution of the cement clinker phases (primarily C3S and C2S as well as C3A) and the precipitation of hydrate phases in particular calcium silicate hydrate (C-S-H), portlandite (CH) and ettringite.The C-S-H formed during the hydration of OPC has a decisive influence on the strength and the mechanical properties in general [11][12].Investigations by Jennings [10], among others, show that two different densely packed phases of C-S-H are formed during hydration: the so called high density (HD-) C-S-H which is characterized by a high density and which is formed preferentially in areas where hydrate phase growth is restricted due to lack of available space [9].The other phase, referred to as low density (LD-) C-S-H, has a lower density and is formed preferably between the clinker particles [10].The hydration and structure formation in presence of calcined clays is more complex and characterized by a different phase formation [13], depending on the type of calcined clay as well as on the cement substitution rate, the calcination temperature,

Abstract
A highly interesting approach to lower CO2 emissions in cement and concrete industry is to reduce the cement clinker content while using supplementary cementitious materials (SCMs) such as calcined clays.Whereas for classical Portland cement clinker, comprehensive studies are already available on the microstructural development during hydration as well as on the resulting macromechanical properties such as compressive strength or long-term deformation behavior, such understanding is missing for calcined clays.In this paper, systematic micromechanical investigations using microindentation of hardened cement pastes with various amounts of calcined clay at ages up to 28 days are presented.The micromechanical results indicate a faster strength development as well as a lower overall porosity of the classical clinker system compared to mixtures containing calcined clay.Looking to creep, the indentation creep modulus is higher for pure OPC mixtures compared to calcined clay systems up to seven days, indicating less pronounced creep.Beyond this age, however, the differences in creep behaviour became less pronounced.The outcomes suggest that the reactivity of the calcined clay seems to be decisive when it comes to mechanical properties and creep behaviour rather than the rate of substitution.
and composition of the clays and further additives [5,[13][14].In addition to the main hydrate phases (C-S-H and CH) formed during OPC hydration, a pozzolanic reaction occurs in which the tempered aluminosilicatic clays react to varying degrees with the CH formed during cement hydration.This results in the formation of additional calciumaluminate-silicate-hydrate C-(A)-S-H phases, which incorporate an increased content of aluminum [15] as well as additional calcium aluminate hydrate C-A-H phases.Here, not only the content of these newly formed phases seems to be highly relevant but also the whereabouts these phases form in, eventually leading to a clogging effect of pores, thus micromechanically stiffening the structure [16][17].Further, a decrease in CH content in the paste is observed which also alters the microstructure [18].Consequently, the reaction kinetics change which leads to different microstructural and micromechanical properties of the binder paste.As already examined in various studies, metakaolin is further known for its filler effect [5].Combined with OPC, this leads to an acceleration of the hydration process due to additional nuclei growth as well as a refinement of porosity [5].Compressive strengths of metakaolin blended cements e. g. show best results for a clinker replacement rate of approx.5 % to 15 % by mass [18].Regarding the creep behaviour, it is known that there is a strong correlation between microstructure and phase composition as well as between porosity and creep behavior [7].Clinker replacement by metakaolin has shown to result in reduced creep of concrete specimen, especially for high replacement levels greater 10 % metakaolin by mass of OPC [19].Further investigations on cement pastes with calcined clay combined with limestone powder show a great potential for reducing the basic creep compliance [7].As outlined before, for common clays such as in northern Europe, with higher availability as kaolinite, the hydration and reaction kinetics significantly differs from high grade kaolinitic clay sources used in literature data [7,20].These differences primarily result in a large extent of the mineralogy and content of the included clay minerals (other than metakaolin).Investigations using such common clays show that below 30 mass percent of kaolinite content in the clay, the reactivity of clay minerals like illite, smectite or mixed-layer minerals becomes highly significant [20][21][22].
For determining micromechanical properties of cementbased materials in the last two decades, investigations using indentation techniques have proven to be a very efficient analysis methods [24][25]12].Various investigations using nanoindentation have been implemented to analyze the micromechanical properties of multiphasic hardened cement pastes and their material properties [11][12]24].Using nanoindentation, the microstructural model of Jennings [10], which introduced two types of C-S-H, could be confirmed and the mechanical properties of these C-S-H types could derived from the packing density of C-S-H nanoparticles/foils [25][26].Haist et al. were able to confirm a direct logarithmic relationship between the packing density of the C-S-H nanoparticles and the micromechanical properties of hardened cement paste, and here especially of the creep modulus [12].The investigations indicate that the deformation of the paste is directly related to the C-S-H particles and creep occurs by sliding between the individual C-S-H particles [11][12].Furthermore, it is well established that creep is influenced by porosity whereas creep increases as porosity increases, by simultaneously decreasing of material stiffness and strength [26].
In this study, investigations using the microindentation technique were performed at different ages and varying ratios of OPC to calcined clay to determine the general composite response of this combination.The micromechanical properties and creep response were investigated as well as the characteristics of porosity.Altogether, the interactions between those parameters were examined to gain a better understanding of micromechanical and creep behavior when using calcined clay as an SCM.

2
Materials and Methods

Raw Materials and mixing regime
For this investigation, cement pastes with different substitution rates of clinker with calcined clay (OPC:CC = 80:20 and 60:40) were prepared (Tab.1).All cement paste samples were created with a water to binder ratio w/b of 0.40 (calcined clay was fully accounted as binder) with demineralized water with a temperature of 20 ± 1 °C.

Sample preparation
Using the freshly prepared cement paste, cylindrical PVC tubes (diameter 20 mm; height 60 mm) were filled and compacted for 30 s with a shaker (IKA Vortex 2).Subsequently, the samples were sealed and stored under isothermal conditions at 20 ± 1 ° C. All samples were demolded after different ages of 1, 3, 7 and 28 day.The specimens were subjected to a seven-day solvent exchange with isopropanol, as to arrest hydration (compare [30]).After this storage, the cylindrical samples were cut into discs with a diameter of 20 mm and height of 10 mm and finally dried for 24 hours under vacuum at 5 mbar and 20 ± 1 °C.Finally, the surface of the slices were polished (MicroCut P1200, P2500 and P4000 from Buehler).This procedure is decisive to avoid interfering impacts while using indentation techniques due to surface morphology, for instance, a large heterogeneity or coarseness or rather roughness of the material.

Methodology
In order to determine the micromechanical properties of hardened cement paste, statistical nano-and microindentation have proven to be highly suitable techniques [31].
Primarily, the indentation modulus M (corresponding to the Youngs modulus) and hardness H can be determined from the indentation deformation curve [32].By incorporating a phase with constant loading, also the indentation creep modulus C can be determined following [11].In this study, microindentation tests were carried out on hardened cement paste samples.All measurements were performed using a micro indenter (MCT 3 , Anton Paar) with a 4-sided Vickers tip.On each sample, 15 by 15 = 225 indentations were performed.These points are regularly distributed in a quadratic grid over the sample surface in order to be able to measure the inhomogeneity of the hardened cement paste surface [33].All indents were performed in a force-controlled manner as detailed in Figure 1, consisting of a loading phase with loading rate of 6 N/min a 1 min (creep) holding phase with constant force and an unloading phase with a loading rate of 6 N/min (also the maximum force applied).The resulting deformation time and load-deformation patterns are exemplarily shown for one indent in Figure 1.From the unloading part of the load-deformation curve (Fig. 1), the indentation modulus M and indentation hardness H can be calculated.From the deformation-time pattern the indentation creep behaviour can be derived and used to calculate the indentation creep modulus.This is measured during the holding phase at maximum load (cf.Fig. 1).
In this study, microindentation measurements were performed at 1, 3, 7 and 28 days of age.The relation between the deformation at maximum load and deformation at complete unloading can be analyzed for the plastic and elastic fractions of the cement paste deformation behavior [32].Figure 2 shows an indented specimen with the typical indentation grid arranged over the surface with a lateral spacing of the identation points of 400 µm.This spacing provided enough distance between the indentation spots to prevent interaction between the individual indents.The data evaluation was performed according to the methodology of Oliver and Pharr which is based on an 3-sided Berkovich indenter [31][32].The slightly different geometries of Berkovich and Vickers indenter are taken into account in this study and refer to the geometry of the loaded area.To investigate the influence of the calcined clays on the deformation behavior, the indentation hardness H, the indentation modulus M and the creep modulus C were determined.To determine the contact area A for calculating the indentation hardness H, the contact depth during application of force is required (Eq.1).
The indentation hardness H is defined as the ratio between the maximum force Pmax during loading by the indenter and the contact area A between the indenter tip and the indentated surface calculated according to Eq. 2. The hardness H correlates with the strength of the indented sample.
The indentation modulus M describes the elasticity of the material and is defined according to Eq. 3.
Here, E stands for the young's modulus and  2 is the Poisson ratio of the sample (value of  2 = 0.30 is used in this study, cf.[34]).The indentation modulus is determined by the unloading slope S during indentation measurement (Eq.4) and the radius au of the indenter with Eq. 5.
To evaluate creep deformation, a defined dwelling time of 60 s under constant maximum load (6 N) was performed during the loading phase and the temporal evolvement of deformation was measured.This deformation-time-behaviour denotes as Δℎ(), can be described by Eq. 6.Following [11], the parameters x1 to x4 were determined using nonlinear least-squares optimization.
Δℎ() =  1 ln( 2 t+1)+ 3 t+ 4 Further investigations show that for indentation creep, the logarithmic creep rate can be simplified to Eq. 7 where  1 describes the creep compliance rate in Eq. 8 with   standing for the contact radius of the indenter while unloading and   is the maximum load during dwelling time (compare to [11]).
Following this, the logarithmic creep was calculated with Eq. 9.
In order to analyze the pore system of different mixtures and, consequently, the progress of hydration and densification of the structure, measurements with mercury intrusion porosimetry (MIP) were implemented to examine the total porosity as well as variances in pore size distributions of the hardened cement pastes.For these investigations, the (hydration-stopped) sample discs with different curing times were crushed into pieces of 2-4 mm of size.Thereupon, sample masses of approximately 1,4 g were weighed and measured with a Micromeritics Autopore III using a maximal intrusion pressure of 400 MPa. and a mercury contact angle of 140° [35].

Experimental Results
In this study, investigations with microindentation were performed at different ages and varying amounts of calcined clay.Here, the focus is on mechanical behavior like indentation hardness and modulus as well as creep modulus.Additionally, the porosity of the different mixtures was investigated.

Fresh paste properties
As described in Section 2, the different mixtures were prepared with the mixing regime in Table 4. Figure 5 shows the slump flow behaviour of the investigated mixes whereas the desired slump flows were defined within the range of 20 -25 cm.The OPC sample was sufficiently workable as to cast samples.An addition of superplasticizer was not necessary.For the calcined clay samples, superplasticizier was used to ensure flowability of the mixtures.

Indentation modulus M and hardness H
Figures 3 and 4 show the results of indentation grids with the distribution of indentation hardness H over the sample surface of pure OPC samples and samples with a replacement rate of 20 % (20cc) at 7 days of sample age.The OPC sample (cf.Fig. 3) shows a slightly higher hardness with maximum values up to approximately 400 MPa compared to the sample with calcined clay (cf.Fig. 4), which shows values up to 350 MPa.Beyond that, no significant differences in spatial distribution are observed.Figure 6 presents the mean value of the indentation modulus M of the different mixtures as a function of time.As can be clearly seen, the indentation modulus M increases with advancing age for all mixtures, with the pure OPC sample showing a higher stiffness at all times.With increasing substitution rate of calcined clay, the indentation modulus M is reduced for all points of time.In principle, the same behaviour can be observed in Figure 7 for the indentation hardness H. Here, the hardness H measured on samples 20cc show a very similar behaviour as OPC but with systematically slightly lower values.From this it can be concluded, that the indentation hardness H seems to be less affected by the replacement of OPC with calcined clay compared to the indentation modulus M. Furthermore, it becomes apparent from the curves in Figure 6 that the data follow a logarithmic trend, with a quite rapid increase up to 7 days and a steady increase from this age on.In particular, a significant rise of the slope and, consequently of the stiffness, can be noted between 1 and 3 days in all mixtures.During this interval, the hydration is still processing remarkably and the development of strength increases notably so the formed structure leads to a higher stiffness regardless of its composition.

Indentation creep modulus
Figure 8 shows the mean indentation creep modulus C of the clay mixtures and OPC as a function of time.Between 1 and 3 days, there is a strong increase in C for all mixtures, which decreases slightly between 3 and 7 days.After that, no significant increase of creep modulus occurs between 7 and 28 days.At ages up to 7 days, C is higher for OPC compared to calcined clay within error bars.After 7 days, the creep modulus of the 20cc and the pure OPC sample seem to be identical, whereas the 40cc sample shows a slightly lower creep modulus, however still within error bars.As compared to the previous illustration, Figure 9 shows the creep modulus C of the different clay mixtures illustrated as boxplots for the interval from 1 to 28 days of age.Here, Figure 9 shows the statistical distribution of creep data as well as outliers.As described before, the boxplots show increasing indentation creep values up to 7 days apart from sample mixture.The samples with an age of one day show a higher amount of outliers compared to the older samples, which is explained by the higher heterogeneity of the young sample surface and its differences in hydration progress.At other sample ages, the distribution of the data points and the amount of outliers are almost comparable.From the comparison of the values for the indentation hardness H and indentation modulus M with the results for the indentation creep modulus C, it becomes obvious that despite the significant increase in both, H and M beyond the age of 7 days, no such change can be observed for the creep modulus C beyond 7 days, hinting to the fact that after 7 days, the creep behaviour is not significantly altered by the progressing microstructure formation.Further, it can be seen that despite a significant influence on the indendation modulus M, the effect of calcined clay on the creep modulus C is of minor importance.This indicates, that the addition of calcined clay leads to a reduction in creep compared to ordinary OPC mixes, with the calcined clay obviously being able to compensate for the lack of OPC in the mix, thus reducing overall creep.To interpret the results, the different pore volumes and pore radii can be assigned to various types of pores at different dimensions.At small dimensions up to approximately 10 nm, pores can be classified as gel pores [36].
Larger pore entry radii between approximately 10 and 100 nm can be determined as interhydrate pores [37] followed by capillary pores at larger ranges [37].Nevertheless, the transitions between pore types is smooth [38].At the age of 7 days (cf.Fig. 10), the samples with calcined clay show a higher volume of gel pores compared to OPC whereas 40cc has the highest amount.This trend continues up to larger pore radii where a great slope is identifiable between 8 and 10 nm where the pore volume rises significantly at all mixtures.The 40cc sample shows a notably higher pore volume while 20cc and OPC feature rather related values.This increase occurs approximately between 10 and 50 nm of pore radius, which can be interpreted as interhydrate pores.Pore sizes larger than 50 nm can be clearly identified as capillary pores [37].At this scale, OPC shows a higher content of capillary pores compared to clay mixtures.The 20cc und 40cc sample differ very slightly from another.In the period between 7 and 28 days, the rate of gel pores increases significantly while the general trend remains that 40cc has the highest amount of gel pores followed by 20cc and OPC (cf.Fig. 11).At the dimension of interhydrate pores with higher pore entry radii, this trend reverses where 20cc shows a slightly larger pore volume compared to 40cc.OPC continues with the lowest values at this range.Furthermore, OPC shows higher amounts of capillary pores in relation to the clay mixtures.The sample with 40cc indicates a slightly higher volume of capillary pores compared to 20cc.The total porosity (cf.Tab. 5) of all mixtures increases with advancing age from 7 to 28 days.Thereby, the OPC shows the lowest porosity which changes from 25% to 22% of total porosity followed by the 20cc sample which indicates values from 27% to 25%.The 40cc sample shows the highest values with a decreases in total porosity from 30% to 28%.

Conclusion and Outlook
In this paper, the influence of calcined clay on the micromechanical properties and creep response was investigated.In addition, the influence of calcined clay substitution on porosity has been analyzed at varying samples ages.From the results it is obvious that the sample age and content of calcined clay have significant influences on the hardness, stiffness as well as the creep behavior of the hardened cement paste.The results show lower values of hardness and indentation modulus in the presence of calcined clay up to 28 days.These outcomes correspond to the total porosity in the presence of calcined clay that exhibits higher values compared to OPC.However, after 28 days of hydration, a clear refinement of the porous structure different from OPC is visible and simultaneously, a moderate reduction of the total porosity indicates the precipitation of additional hydrate phases resulting from the pozzolanic reaction of the clays.This results in a decrease of capillary porosity and an increase of gel porosity.The investigated creep properties show an increasing trend from 1 to 7 days of sample age whereas beyond this age, no significant further increase was found.The samples containing calcined clay differ only slightly from the OPC samples.The results indicate that the addition of calcined clay leads to an increase in creep modulus, which compensates the reduction in creep modulus due to the reduced OPC content.In summary, the results indicate a slower early strength development in the presence of calcined clay due to the slower pozzolanic reaction of calcined clay as well as a presumably lower reactivity of the clay in general.The effect of the clay type and its mineralogical composition is considered to play a significant role in the resulting properties, which however could not be investigated in this study.This refers to the metakaolin content and, for instance, inert components with lower reactivity potential as quartz or feldspar.Here, the calcined clay used in this study can be rated as less reactive with a low metakaolin content of 32 mass %.With regard to the interpretation of the creep behaviour, additional investigations on nanoscale are needed to assess correlations between porosity of the structure and slippage effects between nanoparticles when using calcined clays.Here, calcined clays seem to significantly hinder such slippage.Such studies should be combined with chemical and mineralogical characterization to explain material behavior in the presence of calcined clay.This is primarily linked to e.g.C-A-S-H phases and their impact on the clogging and "reinforcement" of pores within the hardened cement paste.Finally, despite the purely microstructural nature of this paper, the main intention of replacing OPC by calcined clay should be kept in mind: mixtures with calcined clay show a comparable performance compared to OPC especially regarding the creep behavior with a CO2 reduction of approximately 17% when substituting 20% calcined clay and 34 % when using 40% of calcined clay in relation to the OPC system (emission values obtained from [39]).[16] Chen, J. J.; Sorelli, L.; Vandamme, M.; Ulm, F.

Figure 1 Figure 2 Figure 5
Figure 1 Schematic depiction of the deformation-time curve during microindentation (modified after Oliver and Pharr [31-32]) Figure 2 Indentation grid with 225 indents allocated over the sample surface

Figure 6 Figure 3 Figure 4
Figure 6Trend of indentation modulus M of samples with OPC and varying clay substitution rates (20% and 40%) at different ages Between 3 and 7 days, the gradient of stiffness of the clay samples increase faster compared to the OPC.The pozzolanic reaction appears to be increasing in this period and provides a structuring of texture.After 7 days, the slope of stiffness is rising slower due to a slower strength development in the course of hydration at all mixtures.

Figure 9
Figure 9 Indentation creep modulus C illustrated as boxplots of samples with OPC and varying clay substitution rates (20% and 40%) at ages of 1, 3, 7 and 28 days

Figure 7 Figure 8
Figure 7 Trend of indentation hardness H of samples with OPC and varying clay substitution rates (20% and 40%) at different ages Figure 8 Indentation creep modulus C as a function of time of OPC and varying clay substitution rates (20% and 40%) at ages of 1, 3, 7 and 28 days

Figure 10 Figure 11
Figure 10 Pore size distribution (first derivative of the cumulative pore volume) vs. pore entry radius measured by MIP for different clay substitution rates (0%, 20% and 40%) and hydration/curing time of 7 days Figure 11 Pore size distribution (first derivative of the cumulative pore volume) vs. pore entry radius measured by MIP for different clay substitution rates (0%, 20% and 40%) and hydration/curing time of 28 days

Table 1
Composition of hardened cement pastes with w/b = 0.40 [27]his context, a Portland cement CEM I 42.5 R (Wittekind Hugo Miebach Söhne KG) according to EN 197-1[27]and a calcined clay (Liapor GmbH & Co. KG) were used.The physical properties of the constituents are summarized in Table2.The oxidic composition of both, OPC and calcined clay, are detailed in Table.3.The cement, calcined clay and water were mixed with an blender (IKA Eurostar 40) according to the mixing regime detailed in Table.4.

Table 2
Physical properties of binder systems (data of CEM I 42,5 R are provided by Wittekind; data of the calcined clay originates

Table 3
[28]ic composition of OPC and calcined clay (data of CEM I 42,5 R provided by Wittekind; data of the calcined clay originates from[28]) [29]all mixtures containing calcined clay, a superplasticizer based on polycarboxylithic ether (PCE; VP 2018/13.1,MasterBuildersSolutions) was added to ensure sufficient flowability of the fresh pastes.The added amount of superplasticizer varied depending on the mixture composition (see Tab. 1).Immediately after mixing, the Haegermann slump flow without shocks according to DIN EN 1015-3[29]was determined (see section 3.1).

Table 4
Mixing regime

Table 5
Total porosity measured by MIP for different clay substitution rates (0%, 20% and 40%) and hydration/curing times of 7 and 28 days