Micropore Structures in Cenosphere-Containing Cementitious Materials Using Micro-CT

Cenospheres have been recently applied to increase the volume of uniform micropores in hardened cementitious materials. Therefore, application of micro-CT to cenosphere-containing binders will help better understand the micropores formed by cenospheres in the hardenedmaterials. Accordingly, the present study prepared Portland cement paste, alkali-activated fly ash/silica fume, and alkali-activated fly ash with 60%weight replacement by cenospheres and reconstructed their micropore structures using micro-CT. From the pore structure, individual micropores were extracted and analyzed using the principal moment ratios (I11/I33 and I22/I33). Based on the moment ratios, the representative pore shapes were determined in the different pore-volume ranges. Four-factor pore compliance contribution (4-factor PCC) model was then applied to predict the influences of the micropores on the elastic moduli of the micropore/matrix composites.


Introduction
Fly ash, a waste material produced by coal-fired power plants, has been applied to cement and concrete technologies to replace Portland cement [1][2][3].Fly ash particles have three main morphologies: solid sphere, cenosphere, and plerosphere [4,5].Among these, cenospheres can be separated due to having the lowest specific gravity; because of the hollow structure, they are substantially less dense than solid spheres and plerospheres.Therefore, cenospheres have been applied to fabricating lightweight concrete [6][7][8][9].Furthermore, the presence of cenospheres in concrete structures can produce more uniform micropores, allowing these structures to achieve greater energy efficiency with a minimum reduction in strength [9,10].This is because cenospheres increase the volume of uniform micropores in the hardened materials.
As cenosphere particles range in size from 10 to 400 m [11], they contribute to the formation of micropores.However, because of cenospheres' low reactivity, their micropore structures will differ depending on the cementitious material, such as geopolymers and Portland cement.Therefore, it is required to study more detailed pore shapes and the influences of micropores on the mechanical properties of cenosphere-containing cementitious pastes.Recent research on the application of micro-CT to cement and concrete has shed light on the detailed pore structures of cementitious materials [12][13][14][15].In addition, an analysis method developed by Drach et al. [16], called the four-factor pore compliance contribution (4-factor PCC) model, allows us to predict the influence of individual pores on elastic moduli using the Mori-Tanaka method and statistical methods.Therefore, now is an ideal time to look at the detailed pore structures formed by cenospheres and their influence on elastic moduli, using micro-CT technology.
In the present study, we first constructed 3D micropore structures in cenosphere-containing cementitious matrixes using micro-CT.For this, three samples were prepared: alkali-activated fly ash-silica fume, alkali-activated fly ash, and Portland cement paste with cenospheres comprising 60% of binders by weight.We used a Ca(OH) 2 -Na 2 CO 3 activator because a recent study showed that this activator is less expensive and less dangerous than NaOH without compromising the compressive strength [17].Furthermore, 2 Advances in Materials Science and Engineering  the sample with silica fume was included because McBride et al. reported that the addition of silica fume improves the compressive strength of cenosphere-containing concrete [7].The 3D micropores obtained from micro-CT were grouped on the basis of individual pore volumes (<10 −4 mm 3 , 10 −4 ∼10 −3 mm 3 , 10 −3 ∼10 −2 mm 3 , and ≥10 −2 mm 3 ).Based on these groups, each micropore was individually extracted; then the principal moments of inertia ( 11 ,  22 , and  33 ) were calculated to obtain the contour maps of  11 / 33 and  22 / 33 ratios.These ratios also provided numeric values to identify which representative pore shapes were most frequently observed.Lastly, we applied the 4-factor PCC model to predict the influence of the micropores on the elastic moduli of the cementitious composites.

Experimental Details
The present study used ASTM Type I Portland cement known as ordinary Portland cement (OPC) and silica fume marketed by Elkem AS under the product name Elkem-Microsilica5 940U.The cenospheres (product name QK150) and the fly ash, which is the original fly ash of cenospheres, were obtained from Spheretek Ltd. (Hong Kong).Wang et al. reported that cenospheres have substantial amount of amorphous phase and crystalline phases, such as mullite and quartz [8].The particle size distribution of the cenospheres is presented in Figure 1, showing a range of 10 −10 to 10 −1 mm 3 .We prepared three samples: alkali-activated fly ash-silica fume, alkali-activated fly ash, and Portland cement paste.All three samples contained cenospheres at 60% binders by weight.Cenospheres are considered as a binder component, because they contribute to long-term strength.The fly ashsilica fume and fly ash samples were activated using Ca(OH) 2 (Junse Chemical, Japan) and Na 2 CO 3 (Sigma Aldrich, USA).The water-to-solid ratio was maintained at 0.35 for all samples because it was the lowest mixable ratio.The detailed mixture proportion and sample labels are provided in Table 1.The samples were prepared with cylinder molds of 100 mm in diameter and then cured at room temperature (20∼22 ∘ C) in 99% relative humidity for 28 days.Micro-CT scans were performed with a SkyScan 1173 CT scanner at 130 kV and 60 A.The 2D image data had an image resolution of 18.11 m/pixel.Exposure time per projection was 500 ms with a 0.2 ∘ step angle.

4-Factor PCC Model
Drach et al. developed the 4-factor PCC model to evaluate the influence of micropores on elastic moduli including Young's modulus (), bulk modulus (), and shear modulus () [16,18].4-factor PCC is a statistical model to give a prediction from input variables obtained from variation in irregular pore geometries.Finite-element analysis was performed on the basis of a large set of irregular pores.This process constructs a statistical formula to relate pore geometry parameters and their contribution to effective elastic properties.The 4-factor PCC model is the statistical formula as follows:   where MID is a midrange value,   is surface area-to-volume ratio, and V is Poisson's ratio.The dimensionless parameters ( Ẽ , K, G) of the 4-factor PCC model are the influential degree of micropores as follows: where  is a volume fraction of pores and  0 ,  0 , and  0 are the elastic moduli of the matrix.For a detailed discussion of the 4-factor PCC model, see [16].cenospheres could vary, depending on different cementitious materials.This is because each binder has a different condition, which can lead to different micropore structure with cenospheres.We divided the pore-volume ranges into four groups (<10 −4 mm 3 , 10 −4 ∼10 −3 mm 3 , 10 −3 ∼10 −2 mm 3 , and ≥10 −2 mm 3 of individual pore volumes) and analyzed them separately.

Micropores in FASF Sample.
Figure 3 shows the 3D reconstruction of the micropore structures obtained from micro-CT data.In Figure 3, the image on the far left shows the entire structure of the micropores, while the four images to the right show the separated micropores based on the pore-volume ranges.One interesting observation is that there were no <10 −2 mm 3 micropores near other pores that were larger than 10 −2 mm 3 (circled area in Figure 3).This might be because the large pores merged with smaller pores.Despite the visual investigation, the large number of variations in pore shapes made it difficult to define the characteristics of the shapes.Therefore, we calculated the moments of inertia of individual pores and then the eigenvalues of the moments.These eigenvalues were the principle moments ( 11 <  22 <  33 ), and we obtained the contour maps on the left side of Figure 4 using the ratios of  11 / 33 and  22 / 33 (moment ratios).4.3.Micropores in FA Sample.Figure 5 shows the 3Dreconstructed micropore structure of the FA sample and four groups of pore-volume ranges.In contrast to the FASF sample, the FA sample had many pores measuring <10 −3 mm 3 near large pores.Figure 6 shows the contour maps of the moment ratios and the representative pore shapes on the marked areas.In Figure 6(a), the micropores with volumes measuring ≥10 −2 mm 3 have two areas of moment ratios.The  11 / 33 and  22 / 33 ratios of area 1 marked in Figure 6(a) are close to 1.00, meaning that the representative shape of the micropores in that area is a round sphere.However, the  11 / 33 ratios of area 2 marked in Figure 6(a) range between 0.16 and 0.32, and  22 / 33 is close to 1.00.With these moment ratios, the representative shape of the micropores is two connected spheres.The micropores in the volume range of 10 −3 ∼10 −2 mm 3 have three main areas in the contour map of Figure 6(b).Areas 1 and 2 have similar  11 / 33 ratios and pore shapes.However, the pore shape in area 3 is more irregular than those in areas 1 and 2.   of 10 −4 ∼10 −3 mm 3 and <10 −4 mm 3 .The pore shapes in these ranges are elliptical spheres, which are close to the shapes in the FASF sample in identical pore-volume ranges.

4.4.
Micropores in PC Sample.Figure 7 shows the micropores in the PC sample.Large pores with volumes measuring ≥10 −3 mm 3 were not frequently observed in the PC sample.Furthermore, the pore size distribution of the PC sample was more uniform than that of the FASF and FA samples.The large pores (≥10 −3 mm 3 ) in area 1 of Figure 8(a) have irregular shapes, and the  11 / 33 ratios mostly range from 0.14 to 0.42.These low  11 / 33 ratios originate from the elongated pore shapes in the PC sample.It is noteworthy that these low  11 / 33 ratios were also observed for two connected spheres in the FA sample.Furthermore, in Figure 8(b), the representative pore shape in area 2 is more irregular than in area 1, with a lower  11 / 33 ratio in area 2.

Predicted Influences of Micropores on Elastic Moduli.
The material properties influenced by pore structures are elastic moduli of hardened materials, such as Young's modulus, bulk modulus, and shear modulus.The application of the 4-factor PCC model to micropores formed by cenosphere can show the relation between their  11 / 33 and  22 / 33 ratios and elastic moduli.Figure 9 shows the dimensionless parameters ( Ẽ1 , Ẽ2 , Ẽ3 , K, G) of the predicted elastic moduli, which represent changes in elastic moduli induced by pore shape.In the FASF and PC samples, all five dimensionless parameters had similar values regardless of the pore-volume ranges, as the mean values and the gaps between the first and third quantiles were similar.This is because the mean values of the  11 / 33 and  22 / 33 ratios obtained from the micropores in the FASF and PC samples were similar.In the FA sample, the mean values of the dimensionless parameters were close together regardless of the pore-volume ranges, but the variations in the five parameters of the micropores measuring ≥10 −2 mm 3 were higher than those of the other volume ranges.This indicates that the connection of two pores can introduce a high variation in elastic moduli.A comparison between the FASF and FA samples showed similar influences of micropores on the elastic moduli.Furthermore, the micropores measuring ≥10 −3 mm 3 in the PC sample could introduce a greater variation in Ẽ1 than the FASF and FA samples due to the elongated pore shapes having low  11 / 33 ratios.

Conclusions
The main purpose of the present study was to understand the structure of micropores in different cementitious pastes and their influence on the elastic moduli.We obtained the 3D morphologies of micropores from Portland cement paste, Ca(OH) 2 -Na 2 CO 3 -activated fly ash-silica fume, and Ca(OH) 2 -Na 2 CO 3 -activated fly ash with cenospheres of all samples at 60% of binders by weight.The following results were observed: (i) The PC sample (cenosphere-containing Portland cement paste) had a less-broad size distribution of micropores compared to the FASF and FA samples (Ca(OH) 2 -Na 2 CO 3 -activated fly ash-silica fume and Ca(OH) 2 -Na 2 CO 3 -activated fly ash with cenospheres).

Figure 2 :
Figure 2: Pore-volume distributions of FASF, FA, and PC samples obtained from micro-CT data.

Figure 5 :
Figure 5: Reconstructed 3D micropore structure of the FA sample (separated micropores based on pore-volume ranges).

Figure 7 :
Figure 7: Reconstructed 3D micropore structure of the PC sample (separated micropores based on pore-volume range).

Figure 9 :
Figure 9: Dimensionless parameters ( Ẽ1 , Ẽ2 , Ẽ3 , K, G) of the elastic moduli predicted by the 4-factor PCC model: (a) four pore-volume ranges in the FASF sample, (b) four pore-volume ranges in the FA sample, and (c) three pore-volume ranges in the PC sample.

Table 1 :
Mixture proportions of the samples: w/s denotes water-to-solid ratio (in grams).