The Regulating Effects of Ice-Templated Directional Microchannels on Surface Micro-Ceramicization Strengthening of Cement Paste Containing TiB₂

Abstract

Cementitious materials prepared by the ice-templating method appear to have difficulty simultaneously possessing good mechanical properties and an oriented microstructure with microchannels. Surface micro-ceramicization of TiB₂ and the decomposed products of cement hydrates at high temperatures can be regarded as in situ solid–solid reactions involving oxygen, thereby enhancing mechanical properties. This study investigates the mechanical property changes in cement paste with different water-to-cement ratios containing 25% TiB₂ micron powder before and after high-temperature treatment. Cementitious samples are prepared using both freeze-casting (F-CAST) and regular casting (R-CAST) methods with and without the heating post-treatment. The average compressive strength of samples with a W/C of 0.65 prepared by the freeze-casting method at −60 °C with a heating post-treatment is much larger than that of samples prepared by the regular casting method with and without the same heating process. The freeze-casting process for preparing cementitious composites with TiB₂ not only reorders the distribution of water molecules but also redistributes the concentrations of the TiB₂ particles and the main hydrates in the frozen samples. Due to the concentration increase near ice crystal channels within the samples, led by the freeze concentration effect, the new products are formed and cover the channel surfaces after high-temperature treatment. This enhances both the overall and internal properties of the cement-based TiB₂ composite material. The variation in TiB₂ content within the specimens is of paramount importance.

1. Introduction

The ice templating method, also known as the freeze-casting method, has emerged as a novel and versatile processing technique, making its mark in numerous fields such as high-strength and high-toughness structural composite materials, thermal management, and energy storage [1]. By replicating the structure of ice crystals, it enables the creation of multi-scale hierarchical structures ranging from macroscopic to micro- and nano-scales, along with the corresponding unique structure-derived multi-functions. Therefore, the ice templating method is generally considered as a physical method [2], which involves water-ice phase transition, ice crystal growth, freeze–thaw concentration, and ice sublimation or thaw. However, some chemical reactions could occur during the above-mentioned processes, even in the ice growth assembly processes. In other words, introducing chemical reactions can bring about new results, microstructures and physical properties. As a result, the freeze-casting method can be developed beyond a pure physical approach by introducing more chemical reactions in the process.

Portland cement, which is a typical material that can react with water in a wide temperature range [3,4], is the most widely used cementitious material in the world due to its excellent plasticity, accessibility, stress resistance property, and economic benefits. Thus, Portland cement could be a good choice for introducing chemical reactions in different steps of the freeze-casting method, such as the ice crystal growth and thaw processes. Nevertheless, the hardened Portland cement-based materials are inherently brittle [5,6] because of the disordered and random distribution of hydration products and pores, which exhibit low tensile strength and significant volume shrinkage, and start thermal decomposition from 100 °C [7,8,9]. Therefore, it is difficult to strengthen the hardened or dried cementitious matrix (skeletons) directly through the commonly used post-treatment and assistant processing technology, for example, thermal processing. In the previous studies for the enhancement of toughness and strength of the cementitious materials prepared by the ice-templating method, the focus was on introducing organic polymers with good flexibility into the pores and microchannels in the cementitious matrix (skeletons) [10,11,12]. The additional polymers also weaken the fire resistance and the thermal insulation properties of the cementitious materials prepared by the freeze-casting method. Hence, it will be helpful if the cementitious materials can be applied to the thermal processing for post-treatment after freeze-casting steps.

According to the authors’ previous works [13,14], a surface micro-ceramicization mechanism is established for enhancing the mechanical properties after high-temperature treatment of Portland cement paste by adding TiB₂ micron powders in the slurry. The maximum temperature of the heating protocol is suggested to be higher than 750 °C for the optimum enhancement [13]. With the involvement of air during the heating process, the TiB₂ reacts with the decomposed products of Portlandite and C-S-H gels, which are the main hydrates of cement, forming a thick glassy layer on the surface of the matrix and significantly improving the compressive strength of the composites. However, the volume shrinkage caused by the large thermal stress of the chemical reactions is observed in the samples containing too much TiB₂ powder prepared by the regular casting method. Forming more pores or channels in the hardened cement-TiB₂ composites is the easiest way to release the excessive thermal stress of the chemical reactions during the surface micro-ceramicization.

Therefore, the TiB₂-assisted enhancement during heating treatment seems a promising post-treatment solution for the cementitious materials prepared by the ice-templating method, which may result in a much stronger cementitious skeleton with stable volume stability. This study aims to verify the possibility of the combination of the freeze-casting method and the surface micro-ceramicization enhancement led by a high-temperature process to obtain a stronger cementitious composite with a stable volume. Since the water-to-solid ratio is an important factor for the freeze-casting process, three different water-to-cement ratios of cement paste containing 25 wt.% of TiB₂ micron powders were designed in this work. The compressive strength values of the samples prepared by freeze-casting and regular casting methods with and without post-treatment via a heating process were measured. The changes in chemical phases, porosity and morphologies are also discussed for better understanding the synergistic reaction effects.

2. Materials and Methods

2.1. Experimental Materials

White Portland cement (P.W-1 52.5) with an average particle size (d50) of 14.2 μm was selected, and the chemical composition was measured by X-ray fluorescence (XRF, Panalytical B.V., NL, Almelo, The Netherlands) and is listed in

Table 1. Chemical composition of white Portland cement.

Compound	CaO	SiO2	MgO	SO3	Al2O3	Na2O	K2O	Fe2O3	MnO	P2O5	Others
Content (%)	61.04	24.87	6.05	3.48	2.79	0.46	0.57	0.31	0.03	0.03	0.37

. TiB₂ powder was provided by the Jiamai (Liaoning) New Materials Co., Ltd. (Jinzhou, China) with an average particle size (d50) of 3.5 μm, a purity of 99.5%, and a bulk density of 4.52 kg/m³. Powdered hydroxypropyl methylcellulose (HPMC) with a viscosity of 200,000 Pa·s was purchased from Shanghai Chenqi Chemical Technology Co., Ltd. (Shanghai, China). Tap water was used for sample preparation. Isopropyl alcohol with a concentration of 99.7% was purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Figure 1 shows the particle size distribution of white cement and TiB₂ recorded by a laser particle size analyser (Mastersize 3000 Malvern Panalytical, Ltd., Malvern, UK).

2.2. Experimental Setup for Freeze-Casting Method

As shown in Figure 2, a self-made mould made of polytetrafluoroethylene (PTFE) sides and a copper subface was used for preparing samples by freeze-casting. The internal dimensions of the square holes in the mould were 20 × 20 × 20 mm³, and the PTFE sides were fixed to the copper subface using copper screws. Before using the mould, coat its interior with a water-based release agent that does not affect the mould. As shown in Figure 3, the freezing system consists of four copper pillars (40 × 40 × 100 mm³), one copper plate (350 × 350 × 15 mm³) with a temperature sensor in the centre, a heating sheet, a power supply, and an insulation box.

2.3. Sample Preparation

As shown in

Table 2. Mix proportions and casting methods in each group.

Group Name	Water (kg/m3)	Cement (kg/m3)	TiB2 (kg/m3)	Casting Method	Heating Treatment
0.5 R	0.5	1	0.25	R-CAST	No
0.65 R	0.65	1	0.25	R-CAST	No
0.8 R	0.8	1	0.25	R-CAST	No
0.5 F	0.5	1	0.25	F-CAST	No
0.65 F	0.65	1	0.25	F-CAST	No
0.8 F	0.8	1	0.25	F-CAST	No
0.5 RT	0.5	1	0.25	R-CAST	Yes
0.65 RT	0.65	1	0.25	R-CAST	Yes
0.8 RT	0.8	1	0.25	R-CAST	Yes
0.5 FT	0.5	1	0.25	F-CAST	Yes
0.65 FT	0.65	1	0.25	F-CAST	Yes
0.8 FT	0.8	1	0.25	F-CAST	Yes

The HPMC-to-powder ratios were between 0 and 1.0 wt.% for regulating the workability of slurries.

, slurry samples with three water-to-cement (W/C) mass ratios (0.5, 0.65, and 0.8) were designed in this study. For each group, the TiB₂-to-cement (TiB₂/C) mass ratio was maintained at a constant value of 0.25. HPMC powders were used to adjust the workability of the slurry. The ratios of HPMC-to-powder ranged from 0 to 1.0 wt.%. All slurries were prepared at a room temperature of (20 ± 2) °C.

For the regular-cast (R-CAST) samples, the preparation process is as follows: all powdered raw materials, such as cement, TiB₂, and HPMC, were thoroughly mixed in a mixing pot for dry mixing of 60 s. Then, the mixing water was added to the mixing pot containing the powders, and the mixture was stirred for another 180 s. The slurry was manually vibrated for 30 s to expel the air bubbles introduced during the mixing process. The slurry was poured into a mould with an internal size of 20 mm × 20 mm × 20 mm (as shown in Figure 2), and then the fresh samples were cured in a curing box with RH > 95% and a temperature of (20 ± 1) °C for 24 h before demoulding. Next, the samples were put back into the curing box to cure for 7 days.

For the freeze-casting (F-CAST) samples, the slurry preparation process was the same as that for R-CAST samples, with differences only in the casting and thawing processes. After that, the slurry was poured into the homemade moulds, as shown in Figure 2. Then, the filled moulds were placed on a cold table for casting. No anti-sticking agent was used in the PTFE-copper moulds. During the freeze-casting processes, the cold table was submerged in liquid nitrogen, as shown in Figure 3. Before mould placement, the cold plate (copper plate) was maintained at a set temperature (−60 ± 1) °Cvia liquid nitrogen and an electric heating piece, monitored by a temperature sensor placed inside the cold plate. After mould placement, liquid nitrogen was replenished while adjusting the electric heating piece to maintain the temperature of the cold plate. After complete freezing (approximately 30 min), the mould was removed from the freeze-casting box, and the obtained frozen slurry sample was demoulded and placed in a 5 °C freezer for thawing and hydration for another 72 h. It was then placed in the curing box for 4 days to obtain the freeze-casting samples. The total time for sample thawing and curing was 7 days.

After drying the regular-cast and freeze-casting samples in an oven for 24 h at 50 °C, the samples were placed in a muffle furnace. The samples were placed in the chamber of the muffle furnace in an air atmosphere. The muffle furnace was heated at a constant rate of 10 K/min to 750 °C, held at that maximum temperature for 120 min, and then naturally cooled to room temperature. The temperature protocol set in the muffle furnace is shown in Figure 4.

2.4. Characterisations

The morphology of the samples was measured by a scanning electron microscope (S-4800 II, Hitachi Ltd., Tokyo, Japan) equipped with an energy-dispersive spectrometer with an acceleration voltage of 20 kV. The pore size distribution curves of the samples were measured by a mercury porosimeter (AutoPore V, Micromeritics Instrument Corporation, Norcross, GA, USA). The matrix density values of the samples were measured using a specific gravity flask (AccuPyc II 1340, Micromeritics Instrument Corporation, USA). Before the test, the block samples were cut into small pieces with a size of less than 10 mm. For the SEM analysis, the tested samples were coated with a thin layer of platinum for better conductivity.

Thermogravimetric analysis of samples was performed using a thermal analyser (TGA/DSC3+, Mettler-Toledo, Greifensee, Switzerland) from 60 °C to 1000 °C at a heating rate of 10 K/min in atmospheres of nitrogen and air, respectively. Chemical phase analysis of the samples was performed using an X-ray diffractometer (SmartLab9, Rigaku, Tokyo, Japan). Parameters were set at 10°/min, a step size of 0.02, and a scanning range of 5° to 90°. The Cooper target was set at 40 kV. Before the TGA and XRD tests, cubic samples in each group were crushed into powders with a particle size of less than 75 μm and dried in a vacuum oven at 50 °C for 24 h before the test.

A universal testing machine (CMT5105, Shenzhen Xinsansi Materials Testing Co., Ltd., Shenzhen, China) with a force rate of 500 N/s was used to measure the compressive strength values of cubic samples. Six parallel cubic samples were prepared in each group for compressive strength measurement. The change rate of compressive strength values was calculated by the following Equation (1):

R_cc = (f₂ − f₁)/f₁ × 100%

(1)

where R_cc refers to the compressive strength change ratio; f₁ is the average compressive strength value of a cubic sample; and f₂ is the average compressive strength of a cubic sample after high-temperature treatment.

3. Results

3.1. Compressive Strength

Figure 5 shows the compressive strength values of samples in each group before and after high-temperature treatment. In Figure 5a, before heating treatment, the compressive strength values of samples in each group decrease gradually with the increase in W/C ratio, which is attributed to the decrease in density between hydration products and the loosening of structure with a larger W/C ratio [3]. With the same W/C ratio, the compressive strength of samples prepared by the F-CAST method was lower than that of the R-CAST method, possibly due to the larger porosity of samples prepared by the freeze-casting method. After the heating treatment, the compressive strength values of the samples are significantly enhanced because of the surface ceramicization between TiB₂ and cement hydrates at high temperatures [14]. However, compared to the samples with the same W/C ratio and different casting methods, it can be found that for the samples with a W/C of 0.65, the compressive strength values of samples obtained through freeze-casting (0.65FT group) after heating treatment are much higher than those of samples obtained through regular casting (0.65RT group). For the samples with W/C of 0.5 and 0.8, the compressive strength values of the samples obtained through freeze-casting after heating treatment are smaller than those of the samples obtained through regular casting. The latter phenomenon is easier to understand if the surface ceramicization is the only strength mechanism during the heating. Therefore, it can be inferred that there may be another strength mechanism induced by the freeze-casting process.

As shown in Figure 5b, at W/C values of 0.5, 0.65 and 0.8, respectively, the compressive strength values of samples of FT groups increased by 127.43%, 178.12% and 60.33%, respectively, and those of RT groups increased by 39.52%, 17.99% and 62.03%, respectively. When the W/C ratio is 0.5, the R_cc of samples in the 0.5FT group is approximately 3 times that in the 0.5RT group. In contrast, the R_cc of samples in the 0.65FT group is approximately 10 times that in the 0.65RT group when the W/C ratio increased to 0.65. While at a W/C ratio of 0.8, there is no significant difference in R_cc values between samples prepared by the two casting methods. The compressive strength values of samples with a W/C ratio of 0.8 prepared by freeze-casting and regular casting methods before heating are very close, which may be because the pore volumes caused by both preparation methods are similar. Moreover, for the samples obtained through regular casting, the R_cc values first decrease and then increase with the W/C ratios, which is the opposite of the rule that changes for the samples prepared by the freeze-casting method.

Thus, the high-temperature micro-ceramicization strengthening effect of TiB₂ on cementitious materials is not only influenced by the W/C ratios but also by the casting methods. The W/C ratio mainly affects the density of the solid parts of the hardened cementitious materials, while the freeze-casting method mainly affects the microchannels and pores in the composites. The following sections will discuss the density and porosity features of samples in each group.

3.2. Matrix Density

The matrix density values of the samples are shown in Figure 6. Before heating treatment, the matrix density values of samples prepared by the regular casting method slightly increase with the increase in the W/C ratio. While for the samples prepared by the freeze-casting method, the matrix density values first slightly decrease and then increase with the increase in W/C ratios before heating treatment. After heating treatment, the matrix density values of all samples increase significantly compared with the original samples. For the samples in the 0.65F group with a W/C ratio of 0.65, the matrix density increased by 47.7% after high-temperature treatment. Before heating treatment, the matrix density values of the samples in the 0.65F group with the W/C ratio of 0.65 are smaller than those in the 0.65R group. After heating treatment, the matrix density values of the samples in the 0.65F group are much larger than those in the 0.65R group. On the contrary, when the W/C ratio of samples is 0.5, before heating treatment, the matrix density values of the samples in the 0.5F group are larger than those in the 0.5R group.

However, after heating treatment, the matrix density values of the samples in the 0.5F group are smaller than those in the 0.5R group. This may be attributed to the higher solid phase content at a W/C ratio of 0.5. The pronounced freezing-concentration effect and the filling effect of TiB₂ micron particles may result in a higher matrix density for the samples in the 0.5F group compared to those in the 0.5R group. For the samples with a W/C ratio of 0.8, there is no significant change in the matrix density values of the samples in the 0.8R and the 0.8F groups after the heating treatment. The cause of this change may be attributed to an excessively high water-cement ratio, where the freezing concentration effect was not pronounced. This resulted in minimal change to the matrix density of the samples in the 0.8F group after the freeze-casting process. These results suggest that the chemical reaction degrees of samples with the same W/C ratios prepared by the freeze-casting and regular casting methods may be different during the heating treatment.

3.3. Distribution of Pores and Microchannels

Figure 7 and Figure 8 show the pore size distribution curves and the cumulative mercury intake curves of samples at 28 days old in each group measured by MIP. To observe the changes in pore structure of the cementitious samples, the measured pore diameters are classified into three categories: mesopores (2–50 nm), macropores (50–7500 nm), and mega-macropores (>7500 nm) [15]. Due to the limitations of MIP measurement, the recorded pores of the samples are mainly in the range of mesopores and macropores.

In Figure 7a, it is found that there are two main peaks on all curves. As the W/C ratio increases, the first peak values of samples made using R-CAST (in 0.5R, 0.65R and 0.8R groups) around the 80 nm gradually rise, and the second peaks move towards larger pore size ranges. For the F-CAST samples in 0.5F, 0.65 and 0.8 groups, the first peaks appear when the pore size is around 50 nm and are much smaller than those samples with the same W/C ratio and made by the regular casting method. Meanwhile, the second peaks of the F-CAST samples in 0.5F, 0.65 and 0.8 groups move to larger pore size around 1000 nm. As shown in Figure 7b, after heating treatment with a maximum temperature of 750 °C, the peaks of the pore size distribution curve of all samples shifted to a wider range of sizes, with a significant change in the pore size range of macropores and a significant increase in the peak values. Larger pores may be attributed to the decomposition of cement hydrates caused by high temperatures. As the W/C ratio increases, more hydrates are decomposed at high temperatures in samples, causing the peaks of both curves of samples made using the R-CAST and F-CAST methods to show a tendency to shift towards larger size ranges. In addition, with a constant W/C ratio, the two peak pore sizes of samples prepared by the freeze-casting method tend to be smaller compared to the samples prepared by the regular casting method, while the macropore volume is smaller. By comparing the pore size distribution curves before and after high-temperature treatment, it was evident that the pore size of the sample increases while its volume decreases. Correlated with the aforementioned increase in matrix density after high-temperature treatment, it can be concluded that the new substances formed during high-temperature treatment cause the density of the sample matrix to rise.

Figure 8a shows that before heating treatment, when the W/C is a constant value, the freeze-casting process mainly affects the volume of pores between 50 and 7500 nm. For example, the cumulative pore volume of the sample in the 0.65F group between 50 and 7500 nm is much larger than that in the 0.65R group. In Figure 8b, after heating treatment at a maximum temperature of 750 °C, for the sample with the same W/C ratio, the cumulative pore volumes of samples prepared by the freeze-casting method are much smaller than those of samples prepared by the regular casting method. Moreover, the cumulative pore volume curves of each sample significantly vary in the pore size diameter range from 50 to 7500 nm. As a result, the macropore size distribution and volume are affected greatly by the freeze-casting process.

Due to the measurement limitations of MIP analysis, the pores with a diameter larger than 7500 nm can hardly be recorded, as shown in Figure 8. For a better understanding of the changes in megapore samples prepared by the F-CAST method and heating treatment, the morphology analysis of hardened samples is necessary. Figure 9 shows the SEM images of samples before and after heating treatment at low magnifications (×100 to ×250). The red lines in Figure 9 are the indication lines for the direction of the micro channels in the cementitious matrix. In Figure 9, the width values of the microchannels in samples of each group are between 50 and 100 µm, and are much larger than 7500 nm. The microchannels are distributed evenly in the vertical sections of the samples in each group.

Figure 10 shows the SEM images of samples in each group after casting before heating treatment. As shown in Figure 10a,c,e, distinct pore structures with microchannels are detected in the samples prepared by the freeze-casting method (in 0.5F, 0.65F and 0.8F groups). With the increase in W/C ratio, the width of the microchannels becomes larger, and the microstructure becomes looser, which is caused by different contents of solid phases. It can also be observed that the matrix near the microchannel edges is denser, with a large amount of hexagonal calcium hydroxide (CH), calcium silicate hydrate (C-S-H) gels, and TiB₂ particles squeezing and enveloping each other, accompanied by a small amount of needle-rod-like ettringite (AFt), which may be because of the ice crystal compression during the freeze-casting process. For the samples obtained through regular casting method, as shown in Figure 10b,d,f, the matrix gradually shifts from dense to incompact as the W/C ratio increases. Compared with the samples prepared by the regular casting method, the samples obtained through freeze-casting with the same original W/C ratio contain more directional pores and channels, resulting in a weak compressive strength. Those phenomena of cement hydrates in hardened samples can explain the results of the changing compressive strength values described in Section 3.1.

Figure 11 shows the SEM images of samples after heat treatment with a maximum temperature of 750 °C. It can be observed in Figure 11a,c,e that the directional micropores and microchannels of the samples remain after heating treatment. Here, it needs to be pointed out that the products after heating reactions (ceramic-like phases) seem much denser around the walls of micropores and microchannels, which becomes more obvious in the sample with a larger original W/C ratio. By comparing the images in Figure 11a,b with those in Figure 11c,d, the microstructure in the pore wall area of the samples prepared by freeze-casting with W/C ratios of 0.5 and 0.65 is more compact compared to that prepared by the regular casting method. This might be the main reason why the samples formed by freeze-casting show a more obvious strengthening effect after heating treatment. Comparing Figure 11a,c, in the area near the microchannels of the samples with a low W/C ratio of 0.5, the high-temperature oxidation reaction of TiB₂ is not sufficient because lots of Ti elements can be observed according to the EDS data. In the area around the channel with a higher W/C ratio of 0.65, due to the large amount of high-temperature product generation and the absence of the Ti element, it can be inferred that the chemical reaction is sufficient. This may be an important reason why the samples with a W/C ratio of 0.65 obtain the best compressive strength.

The above results suggest that the freeze-casting process with a certain W/C ratio (or liquid-to-solid ratio) promotes TiB₂ and cement particles to concentrate near ice crystals. After the ice crystals melt, the solid particles form the walls of the micropores and microchannels in the hardened samples. To verify this speculation, line-scanning EDS is used to detect the elements (Ti: yellow line, Ca: white line, Si: blue line, O: red line, Mg: green line) distribution across the micropores and microchannels in samples after heating treatment. The line-scanning EDS results of samples from 0.5F, 0.65F and 0.8F groups are shown in Figure 12. As shown in Figure 12a–c, in all the samples prepared by the freeze-casting method after heating treatment, a special phenomenon occurred. In Figure 12b, in the area near the micropores and microchannels, the content of Ti element increases, and the content of Ca element decreases. When the detection area is far away from the micropores and microchannels, the content of the Ca element increases, and the content of the Ti element decreases.

These results prove that during the freeze-casting process, the growth of ice crystals promotes the concentration of Ti elements rather than Ca elements, which means the TiB₂ particles are easier to concentrate in the walls of microchannels rather than the cement particles. The surface charge of the ice/solution interface at a pH of about 12 is negative [16,17], with a zeta potential value of about −30 mV, while the surface charges of TiB₂ particles are also negative [18,19], and the surface charges of cement particles are positive [20]. Thus, during the freeze-casting process at an ultra-low temperature of −60 °C, the negative TiB₂ particles are squeezed to the edge areas of the ice crystals, while the positions of positive cement particles containing calcium ions are relatively stable in the slurry. Furthermore, the distributions of TiB₂ and main cement hydrates are greatly influenced by the freeze-casting and curing processes before heating treatment process. Therefore, after high-temperature treatment, the walls of micropores and microchannels in samples, where TiB₂ is enriched, become much stronger because of the sufficient reactions between TiB₂ and the decomposition products of cement hydrates. A stronger skeleton means a higher anti-compressive property.

4. Discussions

4.1. Products of Cement Hydration and Ceramicization

Figure 13a shows the XRD patterns of the samples before heating treatment. As shown in Figure 13a, the main mineral phases of samples prepared by the F-CAST and R-CAST methods are similar. The main cement hydrates are CH and AFt. The minerals in the hardened samples are C₃S and C₂S, and the added TiB₂. Thus, the freeze-casting method does not affect the main hydrates and minerals in cementitious materials.

Figure 13b shows the XRD patterns of the samples after heating treatment. In Figure 13b, the peaks of new crystalline phases, such as TiO₂ (Rutile), Perovskite, Takedaite, Titanite and Sibirskite, are detected in the pattern curves of all samples. Here, it needs to be pointed out that no CaO peaks are detected in the XRD pattern curves of all samples after heating treatment, indicating that the decomposed Ca(OH)₂ fully participated in the high-temperature chemical reactions without residue. Moreover, the XRD patterns of the samples in the 0.5FT group with a W/C ratio of 0.5 were different from those of the other groups, which had peaks of TiB₂. In comparison, there is barely any TiB₂ detected in the other groups. These results indicate that TiB₂ particles in samples of the 0.5FT group are excessive for the chemical reactions of high-temperature ceramicization, but for samples in other groups with larger original W/C ratios, all added TiB₂ particles participate in the chemical reactions. Under the effect of high-temperature treatment, by the reaction among TiB₂ particles, the cement hydrates and oxygen, the products form a denser layer on the sample surface, resulting in the inability of TiB₂ in the sample to react with the outside air. Compared the XRD patterns of samples in the 0.5RT and 0.5FT groups, the main peak values of TiB₂ of the samples in the 0.5FT group are lower than those in the 0.5RT group.

These phenomena indicate that the extent of the chemical reactions between TiB₂ and the decomposition products of hydrates during heating treatment is different for samples prepared by freeze-casting and regular casting.

Figure 14a,b show the TG-DTG curves of samples before heating treatment under pure nitrogen and air atmospheres, respectively. For a typical TG curve of cementitious materials measured in pure nitrogen, an obvious mass loss occurs at the beginning of 80 °C due to evaporation of free water in cementitious materials [21,22], a significant mass loss is associated with Ca(OH)₂ decomposition between approximately 420 °C and around 550 °C, and the mass loss between 650 °C and 900 °C is caused by the decomposition of CaCO₃ [23,24]. As seen in Figure 14a, an ascending section of the TG curves of all samples has occurred when the temperature is higher than 700 °C, referring to a mass increase in the tested samples. Based on the results in XRD patterns shown in Figure 13a, no CaCO₃ is detected in the samples before heating treatment. Therefore, this may represent the typical phenomenon of trace oxygen permeation in non-hermetic systems under nitrogen atmosphere testing, enabling partial chemical reactions between TiB₂ particles, CaO, and air (oxygen). The relevant reaction Equations (2)–(5) were discussed in detail in the authors’ previous study [13,14]. As shown in Figure 14b, a much stronger mass increase section of the TG curve of samples in each group occurs when the temperature is between 700 °C and 900 °C. These mass increase sections represent the formation of new products during heating with the participation of the oxygen in air. By comparing the mass changes in the samples after 700 °C under the two atmospheres, it was evident that oxygen plays a crucial role in the reaction involving TiB₂ during high-temperature exposure.

$${\mathrm{C}\mathrm{a}\left(\mathrm{O}\mathrm{H}\right)}_2\stackrel{~460°\mathrm{C}}{\to }\mathrm{C}\mathrm{a}\mathrm{O}+{\mathrm{H}}_2\mathrm{O}$$

(2)

$${\mathrm{T}\mathrm{i}\mathrm{B}}_2+{\mathrm{O}}_2\stackrel{~650°\mathrm{C}}{\to }\mathrm{T}\mathrm{i}{\mathrm{O}}_2\left(\mathrm{R}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{l}\mathrm{e}\right)+{\mathrm{B}}_2{\mathrm{O}}_3$$

(3)

$$\mathrm{T}\mathrm{i}{\mathrm{O}}_2+\mathrm{C}\mathrm{a}\mathrm{O}\stackrel{~650°\mathrm{C}}{\to }\mathrm{C}\mathrm{a}\mathrm{T}\mathrm{i}{\mathrm{O}}_3$$

(4)

$${\mathrm{B}}_2{\mathrm{O}}_3+\mathrm{C}\mathrm{a}\mathrm{O}\stackrel{~650°\mathrm{C}}{\to }{\mathrm{C}\mathrm{a}}_3{\mathrm{B}}_2{\mathrm{O}}_6$$

(5)

The above analysis proves that the freeze-casting process does not affect the patterns of cementitious materials containing TiB₂ before heating treatment, but significantly affects the contents and distribution of main cement hydrates in the hardened samples. The follow-up heating ceramicization effect is significantly influenced, resulting in the larger compressive strength values of samples in the 0.65FT group, as mentioned in Section 3.1.

4.2. Ceramicization Enhancement Assisted by Freeze Concentration

The freeze concentration technology takes advantage of the physical property differences between the solvent (usually water) and solute molecules at ultra-low temperatures. By ultra-low temperature freezing, water in the solution is precipitated in the form of pure ice crystals. Then, the ice crystals are separated and removed through melting or sublimation, ultimately increasing the concentration of solutes in the remaining solution [25,26,27]. From the perspective of thermodynamics, the core of the freeze concentration effect lies in the preferential crystallisation of water molecules and the repulsion effects among solutes. That is, water molecules will combine through hydrogen bonds to form regular ice crystals at low temperatures, while solute molecules, due to their larger size than the lattice gaps of ice crystals, are difficult to enter the lattice structures of ice crystals and can only remain in the unfrozen liquid around the ice crystals [28,29]. As shown in Figure 15, the distributions of cement and TiB₂ particles in the samples are totally different during the two casting methods. The solid particles are distributed evenly and randomly in the samples by the regular casting method. During the freeze-casting process, as ice crystals grow, cement and TiB₂ particles are redistributed and arranged in an orderly manner.

Thermodynamic parameters determine the efficiency and feasibility of freeze concentration [30], mainly influenced by the type of solvents, solvent-to-solid ratios, cooling conditions and particle sizes in slurry. The decrease in the solvent-to-solid ratio of the slurry will increase its viscosity, leading to an increase in the frictional resistance on the solid particles and the sizes of ice crystals, forming a relatively dense and uniform structure. However, if the solvent-to-solid ratio in a slurry is either too high or too low, it will intensify the sedimentation of solid particles during freeze-casting, causing the microstructure of the frozen sample in the heat transfer direction to become more uneven, and thereby affecting the hardened compressive strength [31,32]. This classical freeze-casting mechanism can well explain the decrease in compressive strength values of cementitious materials with the increase in W/C ratios, as mentioned in Section 3.1.

According to the usual logical inference, after high-temperature ceramicization, the samples in each group become stronger with a higher compressive strength. However, the changing rules of compressive strength values with the original W/C ratios would be similar to those of the samples before heating treatment. Based on the results in Section 3.1, when the preparation method is the same, the compressive strength values of samples after heating treatment decrease with the increase in the initial W/C ratios. However, the unexpectedly high compressive strength values of samples with W/C of 0.65 prepared by F-CAST are recorded after heating treatment, whose values are approximately two times higher than samples prepared by R-CAST with the same W/C ratio. Therefore, the effect of enhancing high-temperature ceramicization on the compressive strength of samples should be closely related to the distribution of TiB₂ particles and cement hydrates in the hardened samples.

As mentioned in Section 3.2, the distributions of TiB₂ and cement hydrates are greatly influenced by the freeze-casting process. The hydration reactions between cement particles and water occur during the thawing and curing processes. As shown in the SEM images of Figure 9 in Section 3.2, the cement hydrates, such as CH crystals and C-S-H gels, tend to grow into the microchannels and micropores formed after the ice melts. Thus, besides the initial W/C ratios in slurry, the liquid-to-solid ratios in the areas surrounding the ice crystals induced by the freeze concentration effect during the freeze-casting processes also affect the high-temperature ceramicization of TiB₂ on cementitious matrix. A larger liquid-to-solid ratio in the areas surrounding the ice crystals indicates denser wall areas of the microchannels and micropores in the hardened cementitious samples containing TiB₂ before and after heating treatment. Moreover, the synergistic effects of the initial W/C ratio and the freeze-casting method led to a regional liquid-to-solid ratio that dominates the contents and distributions of main cement hydrates, which are involved in the high-temperature chemical reactions of ceramicization. The distribution of TiB₂ particles in the frozen samples, which is greatly influenced by the freeze concentration process, also affects the efficiency of the ceramicization effect. Figure 16 shows the potential enhancement mechanisms of the TiB₂–cement composite prepared by the freeze-casting method with a heating post-treatment process. As shown in Figure 16, the cement hydrates and TiB2 particles are redistributed after the freeze-casting, thawing and curing processes. After the heating treatment, the cementitious skeleton is covered by the new reaction products, such as rutile, Takedaite, perovakite and titanite. These newly produced phases in samples are much stronger than the initial cement hydrates, resulting in a higher compressive strength value. Therefore, both the initial water-to-solid ratio and the freeze concentration process affect the formation and distribution of micropores and microchannels in the cementitious composites. The freeze concentration effect on the slurry with a proper initial water-to-solid ratio can effectively enhance the ceramicization effect during heating treatment, leading to a much stronger TiB₂-modified cementitious composite.

5. Conclusions

This work investigated the changes in mechanical properties and densities of Portland cement paste containing TiB₂ micron powders before and after high temperature under different water-to-cement ratios prepared by freeze-casting and regular casting methods. The influences of the freeze concentration effect on cementitious composites were discussed through the analysis of microstructure and the study of chemical reactions. The main findings can be summarised as follows.

(1)

The ceramicization effects induced by the TiB₂ oxidation at high temperatures can effectively improve the compressive strength of cementitious samples prepared by the freeze-casting method. The compressive strength value of the cement paste containing 25 wt.% of TiB₂ micron powders prepared by the freeze-casting method after heat treatment is much higher than that of the sample prepared by the freeze-casting method without heat treatment. Moreover, when the sample containing 25 wt.% of TiB₂ with a water-to-cement ratio of 0.65, the compressive strength of samples prepared by the freeze-casting method with heating treatment is higher than that of samples prepared by the regular casting method with and without heating treatment.

(2)

The distributions of TiB₂ particles and the pores in the frozen cementitious materials affect ceramicization reactions. With the growth of ice crystals in slurry, the TiB₂ and cement particles are re-distributed by the swelling stress and the coulomb forces caused by the different surface charges on the different particles and the ice crystal. The negatively charged TiB₂ particles become concentrated at the wall areas of microchannels, and the positively charged cement hydrates grow into the microchannels from the walls as detected in the line-scanning SEM-EDS results.

(3)

The freeze concentration effects induced by the freeze-casting method can improve the enhancement of the high temperature ceramicization among TiB₂, air and decomposed products of cement hydrates. The enhancement effects of high-temperature ceramicization are closely related to the initial water-to-cement (or solid) ratios and the regional water-to-solid ratios near the ice crystals in the frozen samples. After heating treatment, the walls of microchannels and micropores in cementitious-TiB₂ composites, which are prepared by the freeze-casting method, are covered by a layer of thick glassy products, leading to a stable matrix and higher compressive strength.