Microstructurally-designed cement pastes: A mimic strategy to determine the relationships between microstructure and properties at any hydration degree

Cement and Concrete Research 71 (2015) 66–77

Contents lists available at ScienceDirect

Cement and Concrete Research journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Microstructurally-designed cement pastes: A mimic strategy to determine the relationships between microstructure and properties at any hydration degree Pipat Termkhajornkit ⁎, Rémi Barbarulo, Gilles Chanvillard Lafarge Research Centre, 95 Rue de Montmurier, 38290 St Quentin Fallavier, France

a r t i c l e

i n f o

Article history: Received 23 October 2014 Accepted 20 January 2015 Available online xxxx Keywords: Hydration Compressive strength Pore size distribution Microstructure Mimic cement paste

a b s t r a c t This paper proposes a new strategy to study the relationships between cement paste microstructure and its properties. In this perspective, microstructurally-designed cement pastes are produced by replacing a specific part of the actual binder by inert particles of similar fineness. This strategy is referred to as ‘mimic’ in this paper. It is shown that, after complete hydration of the reactive part, the microstructure obtained, in which the inert particles play the role of unhydrated binder particles, exhibits similar properties as a cement paste at a lower hydration degree. The concept is tested and validated on pore profile measured by mercury intrusion porosimetry and compressive strength. The same concept could be applied to other properties. In particular, the obtained materials are fully hydrated, which allows performing time-consuming testing (such as e.g. creep and drying-shrinkage tests) on microstructures equivalent to low degrees of hydration, which would not be possible on the hydrating material counterparts. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction In order to develop innovative binders and concretes, it is essential to master the link between their composition and the performance they can develop. The following two types of concrete properties can be considered: • On the one side, some properties are time-dependent (for instance rheology and compressive strength) but can be easily characterized with simple and rapid tests. • On the other side, some properties cannot be characterized rapidly compared to the rate of hydration (for instance transport or creep properties), and it becomes difficult, if not impossible, to estimate how hydration modifies these properties.

Moreover, many phenomena can occur simultaneously, making it even more difficult to estimate the role of each individually. For instance, measuring drying shrinkage on a fully reacted concrete sample is obvious, but interpreting the results of a drying shrinkage experiment in which the binder is still hydrating is not an easy task. However, this is closer to real conditions, in which concrete is cured for less than one day before being exposed to drying conditions when the formworks are removed. ⁎ Corresponding author. Tel.: +33 4 74 82 83 87; fax: +33 4 74 82 80 11. E-mail address: [email protected] (P. Termkhajornkit).

http://dx.doi.org/10.1016/j.cemconres.2015.01.020 0008-8846/© 2015 Elsevier Ltd. All rights reserved.

In order to explore how hydration modifies these properties that need long experiments to be evaluated, one could think of stopping hydration (for example by freeze-drying). This method is commonly used to measure the hydration degree at given times, but cannot be considered for properties such as shrinkage, creep or transport properties which are closely linked to the water saturation of the porous medium. A novel strategy is proposed in the present paper, with which it should be possible to study properties that need long experiments to be evaluated, depending on the hydration degree, and by such offer new possibilities to better understand coupled phenomena. This strategy is referred to as ‘mimic’ in this paper. Rather than trying to stop hydration, the idea is to design a mix so that once fully reacted, its microstructure is equivalent to the microstructure of interest. Since this sample does not evolve anymore, it is easy to evaluate its properties such as shrinkage, creep, transport properties and others. The novel concept of microstructurally-designed cement pastes or mimic microstructure is described, and the equivalence of microstructure (porosity and distribution) and properties (compressive strength) is presented. The strategy is first applied on Portland cement, and then extended to blended cements. 2. Description of the mimic strategy 2.1. Defining mimic strategy for Ordinary Portland Cement (OPC) The assumption behind the mimic cement paste concept is to consider that two materials having the same volume fractions of hydrates,

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

67

porosity and unreacted products would demonstrate similar properties. We assume that, as it has been shown for compressive strength [1–3], the impact of properties of the unreacted products is of second order, so that OPC could be indifferently replaced by an equivalent volume of inert mineral, as shown in Fig. 1. The hydration degree of OPC is defined as the volume fraction of reacted OPC over the initial volume fraction of OPC. One can also define the average hydration degree as the volume fraction of reacted OPC over the initial volume fraction of cement, where OPC and fine inert material will be considered as cement Fig. 2 illustration of the evolution of the OPC hydration degree and the average hydration degree for a mimic cement paste as a function of time. In this study, the amounts of replacements of fine quartz were chosen so that the average hydration degree of the fully hydrated mimic paste can mimic the hydration degree of interest for the pure OPC system. 2.2. Defining mimic strategy for blended cements As for OPC, mimic cement pastes can be useful to study the relationships between cement paste microstructure and its properties when the binder is not just OPC but a blend of OPC and a supplementary cementitious material (SCM) such as fly ash or slag. The main difference originates from the fact that SCMs react with different (usually slower) kinetics than OPC. In order to capture the effects of SCMs on microstructure and related properties, the volume fractions of the phases in a fully hydrated mimic cement paste with SCMs should be representative of the state of hydration of the blended system of interest at a given state of hydration. For the sake of simplicity, it is easier to decouple the kinetics of OPC and SCM, a first stage before SCM has started to dissolve (Fig. 3), and then a second stage when SCM is partially hydrated (Fig. 4). In both cases, similarly to what was proposed for unhydrated OPC, the volume of unhydrated SCM is replaced by the same volume of inert material. The first stage is similar to the pure OPC case, for which the dilution of OPC with inert material allows controlling the final average hydration degree. The second stage is more complex due to the fact that the hydrates in the system to be mimicked come from both OPC and SCM. It is thus necessary to introduce SCM in the mimic paste, or at least a reactant that will produce similar hydrates. Moreover, in order to achieve rapidly a fully hydrated mimic paste (typically in about

Fig. 1. Mimic cement paste concept for OPC system.

Fig. 2. Illustration of the evolution of the OPC hydration degree and the average hydration degree for a mimic cement paste as a function of time.

3 months), it was decided to use ultrafine ground slag to mimic slag blends and silica fume to mimic fly-ash blends.

3. Experimental methods 3.1. Cement pastes For cement paste, all materials and apparatuses were pre-conditioned at 20 °C. Water, OPC and fine quartz were mixed with a Waring blender for 60 s at 3000 rpm. Paste spread on the mixer bowl was put back in the mix. Then cement paste was further mixed for 60 s at 3000 rpm. The fresh cement paste was poured into 7.4 mm diameter and 22.7 mm height waterproof molds. To prevent segregation, the molds were continuously rotated at about 5 rpm for 24 h at 20 °C. After that, the samples were demolded. The specimens were cured under water at 20 °C until the required age. To stop hydration of cement paste, the samples were broken at a given age in pieces of maximum 5 mm and were put in acetone for 15 min. Then the specimens were removed from acetone and vacuum-dried by a pump with a pressure of

Fig. 3. Mimic cement paste concept for blended cement with supplementary cementitious materials, before SCM starts to hydrate.

68

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

starting from mixing process until the first hydrostatic weight was about 20 min. The pycnometer was then weighted regularly until 91 days. A schematic figure of the equipment is shown in Fig. 5. The hydration degrees of OPC and of SCM were estimated by the combination of chemical shrinkage and image analysis. A near-linear relationship was found between chemical shrinkage and degree of hydration of Portland cement measured by QXRD [4]. The results reported by Kocaba [5] show that there is a good correlation between chemical shrinkage of slag and hydration degree of slag quantified by image analysis. Accordingly, in this study, it is assumed that the hydration degree of OPC is proportional to the contribution of OPC to chemical shrinkage and likewise for SCM. In blended cement with SCM paste, chemical shrinkage is a result due to the reaction of OPC and of SCM. The chemical shrinkage of SCM is defined here as the chemical shrinkage of OPC + SCM paste minus the chemical shrinkage of OPC + fine quartz at the same replacement ratio of SCM or fine quartz. The hydration degrees of SCMs at 91 days were estimated by image analysis: Spectral–spatial image processing strategies for classifying multispectral SEM–EDS X-RAY maps [6]. The conversion factor is the ratio of the hydration degree and the chemical shrinkage of SCM at 91 days.

Fig. 4. Mimic cement paste concept for blended cement with supplementary cementitious materials, when SCM is partially hydrated.

0.002 mbar for 24 h. Porosity was measured by mercury intrusion porosimetry (MIP). Remaining fragments of stopped specimens age 91 days were ground below 63 μm (determined by sieving) and were used to measure the hydration degree by XRD/Rietveld and thermal gravimetric analysis (TGA). XRD scans were collected with CuKα radiation over the range 2θ = 5°–65° (2θ = 1°/min). Panalytical Highscore Plus version 2.2.2 software was used to run Rietveld refinements. The total amorphous content in a sample was quantified indirectly by using an internal standard with 10% corundum. Non-evaporable water was measured by TGA. The hydration degree estimated by XRD/Rietveld coupled to TGA is determined as follows: 

 W ðt Þ 1−W wn ðt Þ α c ¼ 1− W ðt ¼ 0Þ

ð1Þ

where αc is the hydration degree of OPC, W(t = 0) and W(t) are the mass fractions of the OPC in the initial paste (before hydration) and in the stopped hydrated paste, respectively. W(t = 0) is equal to 100% when there is only OPC as a binder. Wwn(t) is the mass fraction of non-evaporable water in a stopped hydrated paste. W(t) was measured by XRD/Rietveld analysis. Wwn(t) was measured by TGA (mass loss between 20 °C and 550 °C). The average hydration degree is calculated as follows: α av ¼

αc Xc þ αs Xs Xc þ Xs

3.2. Mortars For mortar, the water to binder ratio (W/B) was 1.60 by volume. Sand to binder mass ratio was 3 to 1. The samples were prepared according to European standard EN 196-1. The fresh mortar was poured into 40 mm × 40 mm × 160 mm prism steel molds. The samples were demolded after 24 h. The specimens were cured under water. At required age, the compressive strengths of mortar were measured after flexural test. The results are the average value from six different measurements.

4. Applying the mimic strategy for OPC 4.1. Experimental program In this study, two different OPCs and two quartzes were used. The quartz is considered as an inert material. The two quartzes have the quantities of SiO2 (98.86%). The properties of OPCs and quartzes are shown in Table 1. The cements differ by both their mineralogy and their fineness. The quartzes differ by their fineness, so as to evaluate the impact of the inert particles in the mimic concept. Indeed, one could suggest that, as cement grains dissolve, they become finer, and that the inert should reduce in size accordingly to be representative of the residual unhydrated OPC.

ð2Þ

where αav, αc and αs are the average hydration degree, hydration degrees of OPC and of supplementary materials, respectively. Xc and Xs are the volume fraction of OPC and supplementary materials, respectively. The supplementary materials could be fine quartz, fly ash, slag or silica fume. If it is the fine quartz, αs is equal to 0. The chemical shrinkage of cement pastes was measured under water. 5 g of fresh cement paste was poured into 25 ml Gay-Lussac pycnometer. The thickness of paste in the pycnometer is about 3 mm. The pycnometer was filled with degassed water. Then pycnometer was vacuumed to extract air bubble. Next the pycnometer was weighed with hydrostatic system in water bath controlled at 20 °C. The time

Fig. 5. Measurement of chemical shrinkage.

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

69

Table 1 Mineralogical and physical properties of OPCs and fine quartzes. Name

Density g/cm3

Blaine surface area cm 2/g

Alite %

Belite %

Ferrite %

Aluminate ortho %

Aluminate cub %

Chaux %

Periclase %

Anhydrite %

Hemihydrate %

Gypsum %

Calcite %

Portlandite %

Quartz %

OPC A OPC B Quartz A Quartz B

3.14 3.16 2.66 2.65

3561 6271 5548 3673

60.20 53.80 – –

19.20 24.40 – –

10.20 10.30 – –

1.00 0.90 – –

4.20 5.40 – –

1.00 0.30 – –

0.00 0.00 – –

0.00 0.10 – –

1.60 2.30 – –

1.90 0.80 – –

0.70 0.70 – –

0.10 0.90 – –

0.00 0.00 – –

Cement paste specimens were used to measure hydration degree, chemical shrinkage and porosity. Mortar specimens were used to measure flexural and compressive strengths. In this paper only the compressive strengths are reported. The W/B ratio was 1.60 by volume for both cement paste and mortar. Here, binder is defined as the blend of OPC and fine quartz. The mix designs of cement pastes and mortars are given in Table 2. There are two experiment plans. The purpose of plan I is to evaluate the kinetics of hydration of OPC with or without fine quartz. To do so, degree of hydration and porosity were measured on cement paste. Compressive strengths were measured on mortars. Based on the results of degree of hydration of OPC in plan I, mimic systems were designed so as to reproduce equivalent systems (plan II). Compressive strength was measured on mortars. In plan II, it was assumed that the maximum degree of hydration of OPC is 0.95. The average hydration degrees of mimic samples were calculated by Eq. (2).

4.3. Porosity Fig. 8 shows porosities of MIP measurements for pure OPC paste and OPC + Quartz paste at various ages (Plan I Table 2) as a function of the average hydration degree. It is considered that hydration degree of OPC reaches a maximum value at 91 days. All results fall on the same master curve, with a linear relationship between the porosity and the average hydration degree (provided that the initial water/binder volume ratio is constant). Porosity at null hydration degree is coherent with the 1.60 water/solid volume ratio of all the systems tested. Moreover, the results also allow concluding that fineness of OPC or the presence of fine quartz does not modify the relationship between the average hydration degree and porosity. At this stage, the analysis confirms that the mimic cement paste strategy could be used to mimic the porosity of hydrated OPC: for a well-chosen replacement ratio, after full hydration, the microstructure of mimic pastes will be comparable to the porosity of pure OPC paste at the hydration degree of interest.

4.2. Degree of hydration 4.4. Intrudable pore profile by MIP Degree of hydration was estimated by a combination of chemical shrinkage evolution over time and XRD/Rietveld analysis as explained in Section 3.1. The results (Fig. 6) show that the hydration degree of OPC is slightly affected when OPC is partly substituted by fine quartz, as already reported elsewhere [7,8,5]. At first order, quartz can be considered as inert from the point of view of hydration. Fig. 7 shows the average hydration degree as defined in Section 3.1. As expected, the average hydration degree decreases as the replacement ratio of the fine quartz increases. A replacement ratio with fine quartz can be chosen so that the average hydration degree of the fully hydrated mimic paste will mimic the hydration degree of interest for the pure OPC system.

At this stage, we have validated that for the same average hydration degree, the porosity measured by MIP is comparable. It becomes then possible to identify couples of results with or without the presence of quartz with similar porosities, so as to compare their intrudable pore profiles, as described hereafter. Fig. 9 shows intrudable pore profiles as a function of average hydration degree of cement paste with or without the presence of quartz. It confirms the general trend of porosity refinement as hydration progresses, and although the main pore size evolution is not absolutely monotonic, there is no systematic difference between pure OPC (C) samples and samples with quartz (Q). Intensity of pore with smaller diameter increases when average hydration degree increases.

Table 2 Mix design of cement paste and mortar. Name

OPC A

Quartz B

A

Equivalent age (days) Average hydration degree Paste Mortar Age

100 90 55 40 – – – 28 41 47 54 56 71 73 84

– – – – 100 70 40 – – – – – – – –

0 10 45 60 – – – 72 59 53 46 44 29 27 16

Sand:Binder for mortar

B

% by volume OPC A OPC A + Quartz A 10% OPC A + Quartz A 45% Plan I OPC A + Quartz A 60% OPC B OPC B + Quartz B 30% OPC B + Quartz B 60% Mimic OPC 1 Mimic OPC 2 Mimic OPC 3 Plan II Mimic OPC 4 Mimic OPC 5 Mimic OPC 6 Mimic OPC 7 Mimic OPC 8

W/B

days – – – – 0 30 60 – – – – – – – –

– – – – – – – – ~1 – – ~3 – ~7 ~28

– – – – – – – 0.26 0.39 0.44 0.52 0.54 0.67 0.69 0.80

x x x x x x x

% by volume % by weight

x x x x

1, 3, 7, 28, 91 1.60

x x x x x x x x

91

3:1

70

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

Fig. 6. Hydration degree of OPC as a function of time.

To get more detail, intrudable pore profiles for a given average hydration degree are extracted and presented in Figs. 10–12. The corresponding porosity by MIP, hydration degree of OPC and average hydration degree are presented in Tables 3–5, respectively. Fig. 10 compares the intrudable pore profiles of normal fineness OPC at 91 days and fine OPC at 7 days, both at an average hydration degree around 0.88–0.89. Both intrudable pore profiles are very close, allowing concluding that the cement fineness does not modify the intrudable pore profile when comparing similar degrees of hydration and the same initial water/solid volume ratio. Furthermore, Fig. 11 shows that the addition of 10% fine quartz to fine cement does not modify this conclusion: systems with an average hydration degree around 0.79–0.80 present remarkably similar intrudable pore profiles. Fig. 12 shows that mixes with different fine quartz replacement ratios, different quartz fineness and different cement fineness, will exhibit similar intrudable pore profiles provided they are compared at the same average hydration degree (here around 0.52–0.56). The results above validate the concept of mimic cement pastes from the intrudable pore profile point of view. It can be concluded that at a certain average hydration degree, whatever the initial mix but at a

constant water/solid ratio by volume, the developed microstructure of mimic and pure OPC systems is close in terms of porosity and intrudable pore profile. 4.5. Compressive strength The mimic concept was applied to mortars. Compressive strengths are measured at 1, 3, 7, 28 and 91 days on mixes from experimental plan I with OPC A (Table 6) and at 91 days for mixes in experimental plan II (Table 7). As expected, at 91 days, the compressive strengths of mortar decreased when the replacement ratio of fine quartz increased. There is a good correlation between compressive strength of real mortar and of mimic mortar (Fig. 13). The results imply that the properties of the unreacted products (either anhydrous OPC or fine quartz) on compressive strength are of second order. The results can be analyzed by gel–space ratio (GSR) descriptor proposed by Powers [9] as computed as follows: GSR ¼

V hydrates V hydrates þ V φ

Fig. 7. Average hydration degree as a function of time.

ð3Þ

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

Fig. 8. Porosity (MIP) for pure OPC and OPC+ Quartz (1, 3, 7, 28, 91 days) as a function of the average hydration degree.

where Vhydrates is the volume fraction of hydrates and Vφ is the volume fraction of capillary porosity. There is no parameter concerning the properties of anhydrous cement in the above equation. GSR can be rewritten [10] as:

GSR ¼

V c K v αc wo þ V c α c

Fig. 10. Intrudable pore profiles for an average hydration degree of 0.88–0.89.

cement, which equals 2.06 [10]. The GSR concept was also applied for the system with limestone [11]. Then a power-law relationship is usually assumed between GSR and compressive strength: n

ð4Þ

where Vc is the initial volume of OPC, wo is the initial volume of water and air, αc is the hydration degree of OPC, and Kv is the ratio between the volume fraction of hydrates over the volume fraction of dissolved

71

CS ¼ CS0 ðGSRÞ

ð5Þ

where CS is the compressive strength, and CS0 and n are the empirical constants. Fig. 14 shows compressive strength of mortar as a function of GSR. As for the specimens in Table 2 plan I, the hydration degrees of OPC in Fig. 6 were used as an input to calculate GSR by Eq. (4). It is worth

Fig. 9. Intrudable pore profiles as a function of average hydration degree.

72

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77 Table 3 Porosity, hydration degree of OPC and average hydration degree corresponding to Fig. 10.

OPC B 7 days OPC A 91 days

Porosity by MIP (%)

Hydration degree of OPC

Average hydration degree

27.43 27.51

0.89 0.88

0.89 0.88

Table 4 Porosity, hydration degree of OPC and average hydration degree corresponding to Fig. 11.

OPC B 3 days OPC A + Quartz A 10% 28 days

Fig. 11. Intrudable pore profiles for an average hydration degree of 0.79–0.80.

mentioning that the hydration degree was measured on paste, while the compressive strength was measured on mortar. By comparing GSR to compressive strength, it is assumed that the hydration degree in the mortar is the same as in paste. As for mimic OPC specimens, GSR was calculated assuming that the hydration degree of OPC at 91 days was 0.95 or 1.00 despite the fact that the hydration degree of OPC in OPC A + Quartz A hardly reaches this ultimate value. For a given GSR, GSR of mimic OPC calculated with 1.00 hydration degree gives the strengths closer to the strength of OPC A and OPC A + Quartz A than those calculated with 0.95 hydration degree, nevertheless, the small discrepancy between the two hypotheses does not appear as critical.

Porosity by MIP (%)

Hydration degree of OPC

Average hydration degree

32.28 32.98

0.79 0.89

0.79 0.80

The results show that there is only one master curve of the relationship between GSR and compressive strength of mortar indicating that the impact of properties of anhydrate cement, either anhydrous OPC or fine quartz, on compressive strength are negligible. In addition, for W/B ratio of interest, the micromechanical analysis gives the same conclusion [2]. It should be noticed that thanks to the mimic strategy, the investigation of properties of early age is possible. The example of compressive strength has been given in this study. The lowest equivalent hydration degree shown here is 0.26 (Table 2). Based on the experimental results on OPC, it is concluded that the mimic strategy can be used to mimic porosity, intrudable pore profile of cement paste and compressive strength of mortar. After complete hydration of the reactive part, the microstructure obtained, in which the inert particles play the role of unhydrated binder particles, exhibits similar properties as a cement paste at a lower hydration degree.

5. Applying the mimic strategy for OPC–SCM systems In the previous section, the concept of mimic cement paste has been applied and tested on OPC. It was shown that it allowed to reproduce, by replacing unreacted cement by fine quartz, the behavior of the system in terms of porosity, intrudable pore profile and compressive strength. The same concept is tested hereafter to mimic systems containing SCMs (fly ash and slag). As explained above, silica fume was used to mimic fly ash, and ultrafinely ground slag to mimic slag.

5.1. Experimental program Table 8 shows the mineralogical and physical properties of SCMs used in this study. Table 9 shows the mix design of blended cement paste and mortar with SCMs. Chemical shrinkage was measured on paste specimens until 91 days. The hydration degree of SCMs in paste was estimated

Table 5 Porosity, hydration degree of OPC and average hydration degree corresponding to Fig. 12.

Fig. 12. Intrudable pore profiles for an average hydration degree of 0.52–0.56.

OPC B + Quartz B 30% 3 days OPC A + Quartz A 45% 91 days

Porosity by MIP (%)

Hydration degree of OPC

Average hydration degree

39.07 41.56

0.80 0.95

0.56 0.52

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

73

Table 6 Compressive strength of mortar of OPC and of OPC + Quartz. Age (days)

Plan I

1 3 7 28 91

Compressive strength of mortar (MPa) OPC A

OPC A + Quartz A 10%

OPC A + Quartz A 45%

OPC A + Quartz A 60%

20 37 46 61 68

18 32 42 57 61

8 16 20 29 33

5 9 11 16 19

Table 7 Compressive strength of mortars of mimic OPC specimens.

Name

Compressive strength of mortar (MPa)

Mimic OPC 1 Mimic OPC 2 Mimic OPC 3 Mimic OPC 4 Mimic OPC 5 Mimic OPC 6 Mimic OPC 7 Mimic OPC 8

10 21 27 33 34 47 49 55

91 days

by combination of chemical shrinkage and image analysis [6]. Mortar specimens were used to measure compressive strength.

5.2. Degree of hydration Figs. 15 and 16 show the chemical shrinkage of blended cement pastes with 45% slag and fly ash, respectively. The contribution of SCM hydration to chemical shrinkage was calculated as the difference between the chemical shrinkage of the blended mix and that of a reference mix where SCM is replaced by fine quartz. Until about 7 days for slag, and 14–28 days for fly ash, the chemical shrinkage of the blended mix and of the reference mix are superimposed, confirming that chemical shrinkage is mainly the result of OPC hydration in this phase. At 91 days, the chemical shrinkage stops to increase in the reference mix with fine quartz, showing that hydration of OPC is mostly complete. On the contrary, in both cases, SCM hydration is not complete yet at 91 days. In order to convert chemical shrinkage into hydration degree, a convert factor for SCM is needed. The hydration degree of SCMs was

Fig. 14. Compressive strength of mortar as a function of gel–space ratio (GSR).

measured by image analysis with spectral–spatial image processing strategies for classifying multi-spectral SEM–EDS X-RAY maps [6]. Hydration degrees of slag and fly ash at 91 days are 0.62 and 0.34, respectively. The values found in this study are of the same magnitude as those reported elsewhere for slag [5,12] and for fly ash [12–15]. As for OPC, considering proportionality between hydration degree of SCM and SCM contribution to chemical shrinkage, one can plot hydration degree as a function of time (Fig. 17 for slag, Fig. 18 for fly ash). Slow hydration of SCMs could be a blocking point for developing the mimic cement pastes concept with SCMs. Indeed, it is expected to get a stabilized hydration after a reasonable curing time otherwise it would not be possible to proceed to time-consuming measurements on a still evolving microstructure. In order to overcome this difficulty, fastreacting products leading to similar reactions as slag or fly ash, respectively, were chosen as follows: - Finely ground slag for slag - Silica fume for mimicking the pozzolanic reaction of dissolved fly ash and portlandite though we are aware that fly ash and silica fume are not completely the same materials.

Fig. 13. Comparison between compressive strength or real mortar and those of mimic mortar of OPC.

74

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

Table 8 Chemical and physical properties of SCMs. Density g/cm3

Name

Fly ash Slag Slagμ Silica fume

2.09 2.93 2.92 2.34

BET m2/g

1.31 – 2.75 8.8

Blaine surface area cm2/g

3686 3400 10,400 28,570

Amorphous %

80.50 93.90 99.10 –

XRF (%) SiO2

Al2O3

Fe2O3

CaO

MgO

K2O

Na2O

SO3

TiO2

Mn2O3

P2O5

ZrO2

54.70 35.82 35.42 91.4

23.28 11.16 11.02 0.47

3.82 0.27 0.22 0.28

10.92 42.29 41.97 0.06

1.08 8.31 8.32 0.05

0.84 0.34 0.31 0.03

3.15 0.18 0.18 0.04

0.16 1.99 1.87 0.05

0.67 0.47 0.47 0.05

0.06 0.22 0.23 0.00

0.08 0.00 0.00 0.36

0.05 0.02 0.02 0.68

Table 9 Mix design of blended cement paste and mortar with SCMs. Name

OPC A

Slag

Fly ash

Paste

Mortar

% by volume OPC A + slag 45% OPC A + fly ash 45%

55 55

45 –

45

x x

x x

Age

W/B

days

% by volume

1, 3, 7, 28, 91

1.60

Sand:Binder for mortar % by weight 3:1

In order to confirm that the finely ground slag reaction and the silica fume reaction reach a stable stage, two pastes were prepared for chemical shrinkage measurement. The following mixes were compared: • 55% OPC + 45% fine quartz (volume %) • 55% OPC + 35% fine quartz + 10% ultrafinely ground slag or silica fume (identified as ‘mimic slag 4’ and ‘mimic fly ash 5’ in Table 10 for more details)

Figs. 19 and 20 show the measured chemical shrinkage, and the contributions of ultrafinely ground slag and silica fume to chemical shrinkage to that of slag and fly ash, respectively. The results indicate that at 91 days the contributions of ultrafinely ground slag and silica fume to chemical shrinkage are almost completed. It is then possible to use these products to produce stabilized – if not fully hydrated – hydrated cement pastes that can potentially reproduce the phase assemblage and microstructure of OPC–SCM systems. Hydration degrees of SCMs and OPC in blended cement pastes were used to define OPC–SCM mimic mortars that would have the same volume fraction of dissolved OPC and SCM as in the systems to be mimicked. In order to define these volume fractions, one has to assume

Fig. 16. Chemical shrinkage of blended cement with fly ash paste.

Fig. 15. Chemical shrinkage of blended cement with slag paste.

Fig. 17. Hydration degree of OPC and slag in blended cement with slag paste.

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

75

Fig. 19. Chemical shrinkage of “Mimic slag 4” paste and contribution of ground slag on chemical shrinkage.

Fig. 18. Hydration degree of OPC and fly ash in blended cement with fly ash paste.

the hydration degrees of ultrafinely ground slag and silica fume in the mimic systems as follows: • Based on SEM observations showing no unreacted slag, the hydration degree of ultrafinely ground slag was assumed to be equal to 1.0 at 91 days. • Unreacted silica fume could not be observed by SEM or detected by XRD/Rietveld. Thus the hydration degree of silica fume was assumed to be equal to 0.7 at 91 days, close to the value reported by Poulsen [12].

It should be noted that the terminology “average hydration degree” is not used for OPC–SCM system. The word “average hydration degree” can be used to mimic OPC but it is less suitable for OPC + SCM because SCM can react as well. It is possible to have the same average degree of hydration coming from different couple of degrees of hydrations of OPC and those of SCM. 5.3. Compressive strength Compressive strength was measured on mortars of compositions detailed in Tables 9 and 10.

Fig. 20. Chemical shrinkage of “Mimic fly ash 5” paste and contribution of silica fume on chemical shrinkage.

Table 10 Mix design of mimic blended cement with SCMs. Name

Equivalent age

Equivalent hydration degree of OPC

Equivalent hydration degree of SCM

OPC A

Slag m

days Mimic slag 1 Mimic slag 2 Mimic slag 3 Mimic slag 4 Mimic slag 5 Mimic slag 6 Mimic fly ash 1 Mimic fly ash 2 Mimic fly ash 3 Mimic fly ash 4 Mimic fly ash 5 Mimic fly ash 6

~3 ~7 – ~28 ~91 ~3 ~7 ~28 – ~91

Silica fume

Quartz A

Paste

Mortar

% by volume 0.48 0.70 0.81 – 0.92 0.95 0.48 0.70 0.81 0.92 – 0.95

0.00 0.02 0.16 – 0.42 0.62 0.00 0.00 0.00 0.07 – 0.34

28 41 47 55 53 55 28 41 47 53 55 55

– 1 7 10 19 28 – – – – – –

– – – – – – – – – 4 10 22

72 58 46 35 28 17 72 59 53 42 35 23

x

x

x x x x x x x x x x x x

Age

W/B

Sand:Binder for mortar

days

% by volume

% by weight

91

1.60

3:1

91

76

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

ash and slag system lie in between, showing the contribution of SCMs to strength depending on time. Figs. 22 and 23 show comparison between compressive strength of real mortar and of mimic mortar for blend cement with slag and for blend cement with fly ash, respectively. The results are encouraging. The compressive strengths of mimic samples are not far from those of real samples. The results show that the mimic concept can be potentially used for blended cement paste. 6. Conclusions

Fig. 21. Compressive strength of mortar of blended cement system.

Fig. 21 shows the evolution of compressive strength plot as a function of time for mixes with 45% volume replacement by respectively inert quartz, fly ash and slag. As expected, pure OPC gives the highest strength level while the mix with 45% fine quartz gives the lowest. Fly

The final goal of this research is to propose a strategy that enables to evaluate properties that necessitate a long measurement time (such as water vapor sorption isotherm or shrinkage) at any stage of hydration of a given system. This is usually not feasible since the measurement time is longer than the kinetics of hydration. The proposed strategy is based on a new concept, defined as ‘microstructurally designed cement paste’ (mimic), whose main assumption is that unreacted OPC or SCM can be replaced by a chosen inert mineral (quartz in this study) without affecting the measured properties. The first evaluations of the concept are presented focusing on hydration degree, porosity and compressive strength, first on OPC, then with SCMs. The conclusions are summarized below. For OPC system: • Pure OPC and mimic systems with the same average degree of hydration show the same porosity (MIP). Moreover, the fineness of the quartz used to replace unreacted cement has no impact on porosity. The intrudable pore profiles of pure OPC or mimic specimens of equivalent average hydration degree are very similar as well. As for porosity, the impact of cement fineness and that of quartz is negligible.

Fig. 22. Comparison between compressive strength of real mortar and those of mimic mortar of OPC A + 45% slag.

Fig. 23. Comparison between compressive strength or real mortar and those of mimic mortar of OPC A + 45% fly ash.

P. Termkhajornkit et al. / Cement and Concrete Research 71 (2015) 66–77

• The same compressive strengths are observed when comparing pure OPC and mimic systems of similar average hydration degree. It is important to note that in this study the experiments were designed so that when the average hydration degrees are equal, the volume fraction of hydrates, the porosity and the unreacted cement volume fraction are equal. The unreacted cement can be either fine quartz or anhydrous cement. This indicates that the role of unreacted cement and fine quartz can be considered equivalent at first order. Moreover, pure OPC and mimic systems all follow the same compressive strength/GSR relationship. For OPC–SCM systems: • The extension of the mimic concept to blended cements shows that there is no significant impact of unreacted cement grains, unreacted SCM grains and fine quartz grains on compressive strength of mortars. • More research is necessary to fully validate the mimic cement paste concept for OPC–SCM systems, but the satisfactory behavior for compressive strength is promising. As a conclusion, this study shows that the proposed mimic cement paste strategy has proven its relevance for some microstructural characterizations (porosity and intrudable pore profile), and a major property of cement-based systems, compressive strength. These good results let us think that using the mimic cement paste strategy should enable to evaluate properties that necessitate a long measurement time (such as hydric isotherm or shrinkage) on stable systems (mimic) that simulate any stage of hydration of real systems, especially at early age, which was not possible until now. This strategy opens new fields of research for generating relevant experimental data for modeling the relationship between microstructure and properties of cement-based materials, and modeling coupled phenomena involving simultaneous hydration and externallyinduced evolutions of the material, such as drying shrinkage, creep or chemically-induced alterations.

77

References [1] B. Pichler, C. Hellmich, Upscaling quasi-brittle strength of cement paste and mortar: a multi-scale engineering mechanics model, Cem. Concr. Res. 41 (2011) 467–476. [2] B. Pichler, C. Hellmich, J. Eberhardsteiner, J. Wasserbauer, P. Termkhajornkit, R. Barbarulo, G. Chanvillard, Effect of gel–space ratio and microstructure on strength of hydrating cementitious materials: an engineering micromechanics approach, Cem. Concr. Res. 45 (2013) 55–68. [3] P. Termkhajornkit, Q.H. Vu, R. Barbarulo, S. Daronnat, G. Chanvillard, Dependence of compressive strength on phase assemblage in cement pastes: beyond gel–space ratio — experimental evidence and micromechanical modelling, Cem. Concr. Res. 56 (2014) 1–11. [4] J. Parrott, M. Geiker, W.A. Gutteridge, D. Killoh, Monitoring Portland cement hydration: comparison of methods, Cem. Concr. Res. 20 (1990) 919–926. [5] V. Kocaba, E. Gallucci, K.L. Scrivener, Methods for determination of degree of reaction of slag in blended cement pastes, Cem. Concr. Res. 42 (2012) 511–525. [6] S.P. Meulenyzer, J. Chanussot, S. Crombez, J.J. Chen, Spectral–spatial image processing strategies for classifying multi-spectral SEM–EDS X-RAY maps of supplementary cementitious materials, 14th Euroseminar on Microscopy Applied to Building Materials, June 2013 (Helsingør, Denmark). [7] P. Lawrence, M. Cyr, E. Ringot, Mineral admixtures in mortars: effect of inert materials on short-term hydration, Cem. Concr. Res. 33 (2003) 1939–1947. [8] M. Cyr, P. Lawrence, E. Ringot, Mineral admixtures in mortars — quantification of the physical effects of inert materials on short-term hydration, Cem. Concr. Res. 35 (2005) 719–730. [9] T.C. Powers, Physical properties of cement paste, Proceedings of the Fourth International Symposium on the Chemistry of Cement, 1960, pp. 577–613 (Washington, DC, (Paper V-1)). [10] A.M. Neville, Properties of Concrete, 3rd edn. ELBS with Longman, London, 1981. 257–279. [11] D.P. Bentz, E.F. Irassar, B.E. Bucher, W.J. Weiss, Limestone fillers conserve cement part 1: an analysis based on Powers' model, Concr. Int. (2009) 41–46. [12] S.L. Poulsen, Methodologies for measuring the degree of reaction in Portland cement blends with supplementary cementitious materials by 29Si and 27Al MAS NMR spectroscopy(Thesis) Aarhus University, Department of Chemistry and Interdisciplinary Nanoscience Center, 2009. [13] M. Ben Haha, K. De Weerdt, B. Lothenbach, Quantification of the degree of reaction of fly ash, Cem. Concr. Res. 40 (2011) 1620–1629. [14] K. De Weerdt, M. Ben Haha, G. Le Saout, K.O. Kjellsen, H. Justnes, B. Lothenbach, Hydration mechanisms of ternary Portland cements containing limestone powder and fly ash, Cem. Concr. Res. 43 (2011) 279–291. [15] F. Deschner, B. Münch, F. Winnefeld, B. Lothenbach, Quantification of fly ash in hydrated, blended Portland cement pastes by backscattered electron imaging, J. Microsc. 251 (2013) 188–204.