Microstructural Investigation of High-Volume Fly Ash Composites Containing Nano-Calcium Silicate Hydrate Crystals

The cement and fly ash used in this research were supplied by Cement Australia (Darra, Queensland, Australia). Table 1 gives information about the chemical composition of cement and fly ash, and Figs. 1 and 2 illustrate the particle size distributions of cement and fly ash. The $D_{V50}$, the size where half of the particles reside below this point and half reside above this point in volume distribution, for fly ash and cement, is 16 μm and 18 μm, respectively. This indicates that the median size of fly ash particles is a bit smaller than, but overall similar to, that of cement particles. The fly ash is sourced from Gladstone (Cement Australia, Darra, Queensland, Australia) and is classified as Class F fly ash because the composition of fly ash satisfies ${\mathrm{SiO}}_2{+\mathrm{Al}}_2{\mathrm{O}}_3{+\mathrm{Fe}}_2{\mathrm{O}}_3\ge 70\%$ and $\mathrm{CaO}<10\%$ (ASTM 2005). The superplasticizer used in the experiment is MasterGlenium ACE 8710, provided by BASF Australia (Southbank, Victoria, Australia). It is polycarboxylic-based superplasticizer (SP) for retaining the workability of concrete. The recommended dosage of the MasterGlenium ACE 8710 is about 600–1200 ml per 100 kg of cementitious material (BASF 2020a). As it was found that excessive use of SP could adversely affect the compressive strength of concrete (Alaka et al. 2016), the SP added in the mix design in this research (Table 2) is within the recommended dosage range. Note in Table 2 that the dosages of SP used in the FA pastes are much lower compared with the OPC paste. This result happens because the FA particles have spherical shape, which could improve the workability of mixes (Thomas 2007), reducing the usage of SP. The Master X-Seed 1500 (BASF Australia, Southbank, Victoria, Australia) is used in this research as the suspension of nano-CSH crystals (BASF 2020b). Master X-Seed 1500 is a type of white liquid with the suspension of 200-nm CSH nano-particles, which have the specific gravity of $1.14\pm 0.02\mathrm{g}/{\mathrm{cm}}^3$. The solid content of nano-CSH crystals is 77% in the Master X-Seed 1500 (BASF 2015). If handled or stored properly, the suspension is stable and there is no hazardous material inside. The dosages of Master-X Seed 1500 were selected to include both within and outside of the recommended range. The 0.5% nano-CSH crystals is within the recommended dosage, and the use of 1% and 1.5% nano-CSH crystals is outside of the recommended dosage.

Compression tests were conducted on all concrete samples to evaluate the effect of nano-CSH crystals on the early-age strength development. Moreover, microstructural analyses were also performed for the pastes to investigate the underlying mechanism and the physicochemical properties. These were determined using TGA, FTIR, and SEM.

The mix proportions of concrete samples are presented in Table 2 with the number behind the abbreviation of substance standing for the mass ratio of the substance in the binder. For example, FA59.5CSH0.5 represents that the FA takes up 59.5% of the binder, and the nano-CSH crystals occupies 0.5% of the binder. The mix design of pastes for microstructural investigations takes the same proportions as the corresponding concrete samples in Table 2, except that the aggregates are excluded. The calculated replacement rate of nano-CSH crystal should take into account that the nano-CSH crystal is the solid content of the Master X-Seed 1500 and occupies 77% of the total mass of the Master X-Seed 1500.

The OPC- and FA-based samples were characterized for their elemental contents using wavelength dispersive X-ray fluorescence (WD-XRF), which was conducted on a Bruker S4 pioneer X-ray spectrometer (Bruker, Karlsruhe, Germany). The samples were first calcined under air at 700°C to remove any residual organics and to ensure that all material constituents were in the oxidized form. Although such precautions were not essential for the samples used in this particular study, this standard procedure was followed to ensure the highest accuracy was achieved in the X-ray fluorescence (XRF) measurements. The XRF instruments are used to determine heavy element content within a sample and are usually inaccurate or completely blind to low molecular weight elements such as nitrogen and carbon. Therefore, it is necessary to remove organic materials within a sample to enhance the accuracy of the metal and metal oxide content measurements with this method. Calcining is mostly performed in air to make complete combustion of the organics possible, and to minimize the presence of residual carbon in the final product. Given that organic materials can degrade above 500°C (Tomczyk et al. 2020), a high calcination temperature is necessary to ensure that all organic materials are removed from the sample prior to XRF analysis (Ahmad Fara and Abdullah 2016: Xiao and Pang 2017). The samples were first calcined under air at 700°C to remove any residual organics and to ensure that all material constituents were in the oxidized form. The samples were characterized in the powder mode using He gas (over vacuum) and undergoing oxides evaluation. The samples were placed on a 6 μm Mylar pad, which is transparent in XRF measurements. Any data that did not provide material content within two decimal places and had $>10\%$ error at the same time were excluded as part of the analysis.

The TGA was performed by using PYRIS TGA machine (PerkinElmer, Shelton, Connecticut). Prior to the commencement of TGA testing and after the pastes were cured for 28 days, the pastes were crushed and milled to transform to powder, which was required to be fine enough to pass a sieve of 63 μm. On the testing day, the hydration reaction of the pastes was stopped first, which was achieved by placing the powder of each mix on a filter paper above a vacuum pump, with the powder entirely immersed in acetone, and dried by the vacuum pump repetitively 5 times. For each mix, powder with the weight of 5–12 mg was loaded into the machine and then heated from 30°C to 1,000°C under ${\mathrm{N}}_2$ atmosphere, with the heating rate set constantly to be $10°\mathrm{C}/\min$. The TG curve plotted was based on the mass of the sample subject to heating in the percentage of original sample. The decrease in the mass of mix samples is caused by the decomposition under heating. Deboucha et al. (2017) suggested the temperature ranges for the decomposition of different compounds in the pastes, and these ranges were adopted in this research. The temperature range of 105°C–400°C mainly corresponds to the mass loss due to dehydration of C-S-H (Ldh). The range of 400°C–600°C is associated with the decomposition of $\mathrm{Ca}{(\mathrm{OH})}_2$, called dehydroxylation (Ldx). The mass loss between 600°C and 800°C is related to the decarbonation of ${\mathrm{CaCO}}_3$ (Ldc). The actual temperature ranges may be subject to minor changes with regard to the first derivative of TG curves (DTG). The calcium hydroxide (CH) content in each paste is calculated by Eq. (1) (Klimesch and Ray 1996)

The chemically bound water ($W_B$) is obtained using the Eq. (2), as follows (Bhatty 1986):

The FTIR test was conducted on pastes cured for 28 days. The machine used was a Perkin Elmer Spectrum 100 Fourier transform infrared spectrometer (PerkinElmer, Akron, Ohio) equipped with a universal attenuated total reflection (ATR) accessory. The procedure of controlling the software could be found in a user’s guide (PerkinElmer 2005). FTIR shares the same processes as TGA in terms of preparing samples.

The method used to observe the microstructure of pastes at 28 days was scanning electron microscopy (SEM) equipped with backscatter electron detector under the voltage of 15 V. The energy-dispersive X-ray spectroscope (EDS) was attached to the SEM apparatus to conduct elemental analysis in the pastes. Regarding the sample preparation for each mix, a small piece of paste was mounted in a cylinder mold filled with resin. The mold had a diameter of 25 mm and a height of 10 mm. The sample piece is a random fragment from the sample crushed by the compressive strength testing machine. As long as the sample is smaller than the mold, the sample could be placed into it. The surface of sample piece was polished smooth enough for the SEM test.

Fig. 3 shows the compressive strength development of concrete with different mix designs. Observe that the compressive strengths of FA60 were 55.9% and 80.8% of OPC at 1 and 28 days, respectively. By increasing the replacement ratio of fly ash to 70%, the compressive strengths of FA70 were slightly reduced to 54.2% and 79.5% of OPC at 1 and 28 days, respectively. The slow development of HVFA concrete was mainly due to the postponed hydration of fly ash. It was found that Class F fly ash mixed in OPC could hydrate to produce CSH after about a week from the time of mixing (Neville 2011). Fly ash can participate in the chemical reaction only after the pH value in the pore solution reaches 13.2, as a result of the accumulation of ${\mathrm{Ca}(\mathrm{OH})}_2$ generated from OPC hydration (Neville 2011). The rate of strength development of HVFA concrete was slower at an early age and accelerated at a later age, compared with OPC.

Adding nano-CSH crystals could increase the early-age compressive strength of concrete, but the extent of increase in the compressive strength depends on the dosage of nano-CSH crystals. It can be observed that for a replacement ratio of fly ash of 60%, an increase in the content of nano-CSH crystals from 0.5% to 1% increased the compressive strength, but a further increase from 1.0% to 1.5% decreased the strength at 1, 3, 7, and 28 days. The decrease of strength could be due to the reduction of cementitious material. The higher the content of nano-CSH crystals, the lower the content of cement contained in the concrete. The nano-CSH crystal itself does not have any cementitious property. It needs to interact with cement to stimulate the formation of more hydration product. Therefore, there should be an optimal dosage of nano-CSH crystals that could make concrete reach the highest strength. For the concrete with fly ash replacement ratio of 70%, adding nano-CSH crystals demonstrated a different trend in the compressive strength development. When the content of nano-CSH crystals was 0.5%, the compressive strength of FA69.5CHS0.5 was decreased as compared with FA70 without nano-CSH crystals.

The compressive strength of concrete increased with increasing content of nano-CSH crystals from 0.5% to 1.5% at 28 days. Szostak and Golewski (2018) used 4% nano-CSH crystals in OPC and concrete with 20% fly ash and also found an increase in the strength of concrete both with and without fly ash (Szostak and Golewski 2018). The mechanism could be that the nano-CSH crystals stimulated the hydration of OPC to produce more hydration products involving CSH and ${\mathrm{Ca}(\mathrm{OH})}_2$. The formation of more ${\mathrm{Ca}(\mathrm{OH})}_2$ could further facilitate the hydration of fly ash to generate C-S-H. Because C-S-H was the compound responsible for the majority of concrete properties, including compressive strength (Nonat 2010), the strength of concrete was increased due to the formation of CSH. This conclusion will be further verified in the microstructural investigation.

Table 3 shows the XRF data for selected samples at 28 days. Z stands for the atomic number of an element. Observe that the chemical compositions of HVFA samples were significantly different from those of OPC. The ${\mathrm{SiO}}_2$ content in OPC was 16.81%, and it was more than 25% in HVFA samples. The ${\mathrm{Al}}_2{\mathrm{O}}_3$ content in OPC was only 3.73%, but it was greater than 10% in HVFA concrete. This finding was attributed to the fact that ${\mathrm{SiO}}_2$ and ${\mathrm{Al}}_2{\mathrm{O}}_3$ are the major components of fly ash. Moreover, OPC samples contained fewer ${\mathrm{Fe}}_2{\mathrm{O}}_3$ and more CaO contents than HVFA concrete due to the difference in the major components of OPC and fly ash. However, the differences among HVFA mixes with and without nano-CHS crystals were insignificant. The contents of ${\mathrm{Al}}_2{\mathrm{O}}_3$ and ${\mathrm{SiO}}_2$ in FA59CSH1 and FA68.5CSH1.5 were slightly smaller than those in FA60 and FA70 due to the replacement of fly ash with nano-CSH crystals.

The TG and DTG curves for all of the pastes are in Figs. 4–8. The curves for all of the pastes demonstrated reasonable trends that the loss of weight increases with the increase of age, as the hydration product increases over time. Three major sags could be seen in each DTG curve corresponding to the mass loss due to Ldh, Ldx, and Ldc.

From Table 4, where Ldh, Ldx, and Ldc values of the mixes are presented, note that FA59CSH1 and FA68.5CSH1.5 have higher Ldh values compared with FA60 and FA70, respectively. Ldh is the chemically bound water from C-S-H. It indicates that adding nano-CSH crystals facilitates the formation of C-S-H in the HVFA pastes. This result agrees well with the compressive strength results that higher strength could be seen in the HVFA mixes with nano-CSH crystals. Comparing Ldc as the chemically bond water from ${\mathrm{CaCO}}_3$ among different mixes, the pure HVFA mixes have higher Ldc values compared with the mixes with nano-CSH crystals. The ${\mathrm{CaCO}}_3$ content is from the carbonation of ${\mathrm{Ca}(\mathrm{OH})}_2$ during hydration. This could reflect that the pure HVFA mixes without nano-CSH crystals are more prone to carbonation. The Ldx value represents the decomposition of ${\mathrm{Ca}(\mathrm{OH})}_2$ in the mixes. However, the actual amount of ${\mathrm{Ca}(\mathrm{OH})}_2$ in the mixes are composed of the weight both from Ldx [decomposition of ${\mathrm{Ca}(\mathrm{OH})}_2$] and Ldc (decomposition of ${\mathrm{CaCO}}_3$), as ${\mathrm{Ca}(\mathrm{OH})}_2$ carbonate to form ${\mathrm{CaCO}}_3$. The actual ${\mathrm{Ca}(\mathrm{OH})}_2$ content for the mixes are calculated and presented in Fig. 10.

Chemically bound water and ${\mathrm{Ca}(\mathrm{OH})}_2$ content values are calculated from the Ldh, Ldx, and Ldc of the mixes. Fig. 9 shows the chemically bound water in the pastes. Among all pastes without nano-CSH crystals, OPC shows the highest $W_B$, and FA70 demonstrates lowest $W_B$ at all of the ages of 1, 3, 7, and 28 days. Therefore, increasing the content of FA decreased the hydration products in the pastes. This could be explained by the postponed hydration of FA compared with OPC (Lothenbach et al. 2011; Neville 2011).

After the addition of 1% nano-CSH seeds, FA59CSH1 had a higher content of chemically bound water than FA60 at 1, 3, 7, and 28 days. Similarly, FA68.5CSH1.5 contained more chemically bound water than FA70. Adding nano-CSH seeds could stimulate the hydration of HVFA pastes containing 60% and 70% fly ash. Furthermore, the effect of nano-CSH seeds on FA60 was more significant than it was on FA70, in particular when the age was less than 3 days.

Fig. 10 shows the ${\mathrm{Ca}(\mathrm{OH})}_2$ content in OPC, FA60, FA70, FA59CSH1, and FA68.5CSH1.5 pastes. Note that the ${\mathrm{Ca}(\mathrm{OH})}_2$ content decreased with the increase of fly ash dosage in cement pastes at 1, 3, 7, and 28 days when comparisons were made among OPC, FA60, and FA70 pastes. With less portland cement in FA60 and FA70 pastes, less ${\mathrm{Ca}(\mathrm{OH})}_2$ could be generated from hydration. After adding nano-CSH crystals, FA59CSH1 and FA68.5CSH1.5 had higher ${\mathrm{Ca}(\mathrm{OH})}_2$ content than FA60 and FA70 pastes at the ages of 1, 3, and 7 days, respectively. However, the ${\mathrm{Ca}(\mathrm{OH})}_2$ content of FA59CSH1 and FA68.5CSH1.5 pastes on the 28th day became lower than those of FA60 and FA70, respectively. It is known that the hydration of OPC generates ${\mathrm{Ca}(\mathrm{OH})}_2$, whereas the hydration of fly ash consumes ${\mathrm{Ca}(\mathrm{OH})}_2$.

The variation of ${\mathrm{Ca}(\mathrm{OH})}_2$ in the pastes indicated that the formation of ${\mathrm{Ca}(\mathrm{OH})}_2$ from OPC hydration exceeded the consumption of ${\mathrm{Ca}(\mathrm{OH})}_2$ by the hydration of fly ash in FA59CSH1 and FA68.5CSH1.5 pastes at the early ages of 1, 3, and 7 days. On the 28th day, the ${\mathrm{Ca}(\mathrm{OH})}_2$ consumed by fly ash hydration in FA59CSH1 and FA68.5CSH1.5 finally exceeded the ${\mathrm{Ca}(\mathrm{OH})}_2$ generated from OPC hydration. Adding nano-CSH crystals facilitated the consumption of FA in HVFA mixes, or the pozzolanic reaction between FA and ${\mathrm{Ca}(\mathrm{OH})}_2$ to form C-S-H. Improved pozzolanic reaction of FA concrete by the addition of other nano-particles, mostly nano-silica and ${\mathrm{nano}\text{-}\mathrm{CaCO}}_3$, can be found in the previous literature (Said and Zeidan 2009; Supit and Shaikh 2014; Liu et al. 2020).

Fig. 11 shows the FTIR results for all samples at the ages of 1, 3, 7, and 28 days. The broad bands present in the range of $\mathrm{3,648}–\mathrm{3,650}{\mathrm{cm}}^{-1}$ of all of the samples is associated with the overlapping stretching vibration of the $-\mathrm{OH}$ groups of ${\mathrm{Ca}(\mathrm{OH})}_2$ from the tricalcium silicates (${\mathrm{C}}_3\mathrm{S}$) and dicalcium silicates (${\mathrm{C}}_2\mathrm{S}$) hydration and the free $-\mathrm{OH}$ groups of water molecules in the mixes (Biricika and Sarierb 2014). Note that the band diminished with the increase of hydration time from 1 to 28 days, indicating the decrease of free water as a result of CSH formation. Comparisons among the $-\mathrm{OH}$ groups of different mixes suggest that OPC showed the largest decrease of this band from 1 to 28 days, but FA60 and FA70 had the least significant decrease. The hydration of OPC generated more CSH than HVFA mixes because the reaction of fly ash is slower than OPC, especially at an early age (Bremseth 2009; Neville 2011). Compared to HVFA pastes FA60 and FA70, FA59CSH1 and FA68.5CSH1.5 had a more remarkable decrease in the intensity of the $-\mathrm{OH}$ groups between 1 and 28 days. The decrease of $-\mathrm{OH}$ groups in HVFA pastes with nano-CSH crystals could be induced by either the consumption by fly ash or less free water present, which has been demonstrated by TGA results. The TGA results show that adding nano-CSH crystals increased the content of CSH, so less free water was available. Note that less ${\mathrm{Ca}(\mathrm{OH})}_2$ could be found in the HVFA pastes with nano-CSH crystals compared with pure HVFA pastes.

The bands between $949\mathrm{and}719{\mathrm{cm}}^{-1}$ correspond to the bending-in-plane vibrations of the Si–O bonds in ${\mathrm{C}}_3\mathrm{S}$ and ${\mathrm{C}}_2\mathrm{S}$. The intensity of the bands decreased with the formation of CSH over time. For HVFA pastes with nano-CSH crystals, the peak intensity of Si–O bonds decreased more substantially than HVFA paste, indicating that adding nano-CSH crystals increased the content of CSH after the hydration of the pastes.

The bands in the wave number range of $1410–1510{\mathrm{cm}}^{-1}$ are associated with the C–O functional group due to the carbonation by the reaction between ${\mathrm{Ca}(\mathrm{OH})}_2$ and ${\mathrm{CO}}_2$ in the air (Chang and Chen 2006; Biricika and Sarierb 2014). At 28 days, the peak intensity of FA59CSH1 and FA68.5CSH1.5 was smaller compared with FA60 and FA70. This result matches with the TGA findings. It indicates that the HVFA mixes with nano-CSH crystals has a lesser degree of carbonation compared with pure HVFA mixes.

Fig. 12 shows the SEM images of OPC and HVFA pastes. Observe from the images with a magnification of 2,000× that round fly ash particles could be seen in HVFA pastes but not in OPC. Compared with FA60 and FA70, fewer fly ash particles were observed in FA59CSH1 and FA68.5CSH1.5 because fly ash was consumed by adding nano-CSH crystals. This observation agrees well with the TGA results. Existing research (Lau 2011) demonstrates denser hydration product if incorporating nano-CSH crystals in OPC mix. Besides being effective in OPC mix as previously mentioned (Lau 2011), adding nano-CSH crystals in HVFA mixes could densify the hydration product, as well. In the images with a magnification of $\mathrm{10,000}\times$ times, where C-S-H could be observed more clearly, the hydration product that formed surrounding the nano-CSH crystals for FA59CSH1 and FA68.5CSH1.5 mixes were denser compared with pure HVFA pastes. This result could be closely associated with the results revealed in the previous sections, showing improved compressive strength and carbonation resistance of HVFA pastes with nano-CSH crystals.