Hydration and Properties of Cement in the Belite-Ye′elimite-Ternesite System

Energy consumption and carbon emissions are lower in the production of belite-ye′elimite-ternesite (C2S-C4A3$-C5S2$, BYT) clinker than Portland cement (PC) clinker. BYT cement can combine the early strength of CSA cements and the durability of belite cements. X-ray diffraction, mercury intrusion porosimetry, isothermal conduction calorimetry and scanning electron microscope were conducted to investigate the hydration process of BYT cement. The hydration products of BYT cement include mainly ettringite, strätlingite and some amorphous AH3 (aluminum hydroxide). Ternesite did prove an early reactivity in BYT cement. The reaction of ternesite with AH3 occurs on the surface of ternesite. Ternesite delays the second heat flow peak of ye′elimite. The strength of BYT cement containing 10% ternesite in the prepared clinker exceeds that of other cement at all ages.


Introduction
The cement industry is facing increasing pressure due to the depletion of natural fuel resources, shortage of raw materials, continuously increasing cement demand and concerns about climate-linked environments [1]. Calcium sulfoaluminate (CSA) clinker has a lower calcium content and therefore releases less CO 2 from the decomposition of calcium carbonate during the clinker calcination process than that of PC clinkers [2,3]. CSA clinker can be produced within a temperature range of 1250-1350 • C, which is 100-150 • C lower than that in the production process of PC clinker; therefore, fossil fuel consumption is decreased [4,5]. In addition, power consumption is reduced significantly to produce CSA cement, which has a good characteristic of grindability [6]. CSA cement has some advantages, such as high early strength, rapid setting, sulfate resistance and shrinkage compensating properties [7,8]. Hence, CSA cement is considered a promising alternative to PC cement. The main phases in the CSA clinker include belite, ye'elimite and smaller amounts of ferrite and aluminate. The fast hydration of ye elimite results in higher early strength of CSA cement, and belite provides power for improving the later strength [9,10]. In the past, ternesite was considered a hydrated inert mineral. However, Bullerjahn [6,11] et al. demonstrated that ternesite reacted already at early hydration ages. In later work, they revealed that the reaction was linked to the presence by AH 3 . Ternesite bridges the reactivity gap between the rapid aluminate reaction and the late strength contribution of belite [12,13]. Belite-ye elimite-ternesite (BYT) cement combines the early strength of CSA cements and the durability of belite cements [13]. The contributions of major clinker minerals to the evolution of the early performance and properties of BYT cement are in the following order: ye elimite >> ternesite > belite [11].
Ye elimite is generally considered to react ideally in the presence of water to form monosulphate (AFm) and microcrystalline gibbsite (AH 3 ) according to reaction Equation (1) [14,15]. Bullerjahn and co-authors demonstrated that, indeed, the "ideal" reaction does not occur, which is linked to the formation of a metastable phase assemblage C 4 A 3 $ + 18H → C 6 A$H 12 + 2AH 3 (1) The hydration behavior of belite in PC mainly occurs at later ages. The hydration products of belite are mainly amorphous calcium silicate hydrate (C-S-H) gel and portlandite (CH) (Equation (3)) [24,25]. The hydration process of belite changes due to the formation of AH 3 from ye elimite in the ye elimite-rich cements [26]. The hydration product of belite is mainly strätlingite in the presence of AH 3 (Equation (4)) [27].
The hydration of ternesite must be stimulated by AH 3 in pores solution according to Equation (5) [28]. Montes et al. noted that all aluminates (C 3 A, C 12 A 7 , C 4 A 3 $ and CA) could activate the hydration of ternesite and that the activation degree of aluminate to ternesite is C 12 A 7 ≈ CA > C 3 A >>>> C 4 A 3 $ [29,30]. The reactivity of ternesite also depends on the curing temperature and humidity, particle fineness and size distribution [31].
Bullerjahn believed that the hydration of ternesite could only occur when gypsum was completely consumed [11]. The amount of gypsum addition mainly depends on the content of ye elimite. This paper studies the hydration process of a BYT cementitious system in the presence of gypsum. It is necessary to incorporate a large amount of gypsum, which can control the type and quantity of hydration products, inhibit the formation of AFm and promote the formation of ettringite. The purpose of this study is to understand the hydration and performance of the BYT cementitious system, including the evolution of the phase combination with hydration time and the mechanical and microstructural properties. The system contains 5%, 10%, 15% 20% and 25% ternesite. The key effect and hydration activity of ternesite are discussed.

Raw Materials and Sample Preparation
The raw materials used in the experiment include calcium carbonate (CaCO 3 ), silicon dioxide (SiO 2 ), gypsum (CaSO 4 ·2H 2 O) and aluminum oxide (Al 2 O 3 ), which were all analytical grade reagents. Raw materials were weighed accurately according to the stoichiometric C 2 S, C 4 A 3 $ and C 5 S 2 $. The designed phase composition of clinkers and amount of raw materials are shown in Table 1.  45 25 Clinker preparation: the raw materials were mixed in a planetary ball mill for 12 h to ensure raw mix homogeneity. A pressure of 20 MPa was applied to prepare 4 cm diameter cylindrical pellets. All pellets were dried at 110 • C. The dried pellets were calcined at 900 • C with a heating rate of 10 • C/min and held for 0.5 h to completely decompose the calcium carbonate. Then, BYT-1, BYT-2 and BYT-3 were heated to 1180 • C, and BYT-4 and BYT-5 were heated to 1210 • C for 1 h to obtain BYT clinkers. Each calcined sample was quenched by air flow.
Cement preparation: BYT cements were obtained by adding gypsum into BYT clinkers. The amount of gypsum added was determined according to Equation (6) [27,32].
C G is the ratio of gypsum to clinker, A C is the mass fraction of ye elimite formed in clinker, S is the mass fraction of SO 3 in gypsum, M is the gypsum coefficient (M = 2), and ye elimite is hydrated together into ettringite. The content of clinker and gypsum in each group of cement is listed in Table 2. Hydration sample preparation: BYT cement sample was mixed with water according to the mass ratio (w/c) of 0.5. The slurries were stored in a plastic centrifugal tube at a temperature of 20 ± 1 • C and 95% RH. Hydrated samples were cut into 3 mm discs with a precision cutting machine at different ages. The discs were immersed in anhydrous ethanol for one week to stop hydration. After stopping hydration, parts of the discs were ground into powder for X-ray diffraction (XRD) and thermogravimetric-differential scanning calorimetry (TG-DSC) analysis, and the other discs were used for mercury intrusion porosimetry (MIP) analysis.

X-ray Fluorescence (XRF) Analysis
X-ray fluorescence (XRF) spectrometry was used to measure the bulk chemical composition. Measurements were performed using a Philips PW2400 XRF spectrometer (Philips, Amsterdam, The Netherlands), and the data were analyzed with the Uniquant ® software suite (Version: 8, Thermo Scientific, Waltham, MA, USA). The XRF data were used to calculate the mass attenuation coefficients (µ m ) of the clinker samples used in the quantitative X-ray diffraction. Table 3 presents the chemical composition and µ m value of the five clinkers. The XRD data were collected by a Miniflex600 X-ray diffractometer (Rigaku, Matsumoto, Japan) at 20 ± 1 • C. The diffractometer generator was fitted with a Cu Kα X-ray source (k = 1.54059 Å) operating at 40 kV and 15 mA. The range of 2 theta is 5 • to 70 • with a step size of approximately 0.01 • . HighScore Plus software (version 5.1, Panalytical, Almelo, The Netherlands) was applied for the phase identification and quantification.
In order to obtain an absolute quantification of the crystalline phase content, the external standard approach was used. The Rietveld refinements were carried out based on the structure models shown in Table 4. The refined parameters included background coefficients, cell parameters, peak shape parameters and phase scales. The background was fitted graphically with the shifted Chebyshev polynomial using 10 coefficients. The peak shapes were fitted by a Gauss-Lorentz function. In this article, R-weighted pattern (Rwp) is less than 10. For major phases (more than 20 wt.%), the percent error ranges are less than 5%. The external standard approach was used to obtain an absolute quantification of the crystalline-phase content. The diffractometer constant G was determined by standard NIST Al 2 O 3 . This G factor was used to determine the mass concentration of each mineral phase of the BYT clinker and hydration products of the BYT cement through the XRD Rietveld quantification method using Equation (7): where ρ α is the density of phase α, V α is the unit-cell volume of phase α, W α is the weight fraction of phase α, and µ α is the mass attenuation coefficient of the quantified sample, which is provided in Table 3 and calculated by the content of each oxide in the sample [33]. The textural properties are as important as the phase contents in the aspect of interpreting the reactivities. Therefore, a thorough textural characterization was carried out. The fired clinker was crushed and ground. Particle size of BYT-1~5 cement ranged from 0.89 µm to 271.4 µm, and the mean particle size (D 50 ) was 16.96, 15.56, 20.17, 14.27, 28.53 µm, respectively ( Figure 1). Powders were first dried at 110 • C for 1 h, and then, the fineness of five BYT cements was measured in a Blaine fineness apparatus (SBT-127) according to JB-T 8074-2008. Their Blaine specific surface area was about 420 ± 10 m 2 /kg.

Isothermal Conduction Calorimetry
Isothermal conduction calorimetry (8-channel TAM AIR calorimeter, Thermometric AB, Sweden) was used to measure the hydration heat liberation of BYT clinkers during the first 72 h at a constant temperature of 20 °C. To measure the initial thermal response of the samples, an internal stirring method was used in this experiment. An amount of 4 g of BYT clinker was weighed into a flask, and 2 g of water was added to an injector, which was then placed in the measuring position of the instrument. Before testing, the samples rested for 45 min to stabilize their temperature and the water temperature.

Thermogravimetric Analysis-Differential Scanning Calorimetry
To determine the cement hydration products, a TGA/DSC 1/1600LF synchronous thermal analyzer (Mettler toledo, New York, United states) was used to perform a TG-DSC analysis of the hydration samples. Weight loss was measured by heating samples from room temperature to 1000 °C at a uniform heating rate of 10 °C/min in a nitrogen atmosphere.

Electron Microscopy
Microstructure and morphology of pastes were identified using a Zeiss-Supra 55 Scanning Electron Microscopy (SEM) (Carl Zeiss, Oberkochen, Germany) with an acceleration voltage of 15 kV. The sample was sputtered with gold in the vacuum to improve electrical conductivity. The backscattered scanning electron (BSE) mode was chosen, and the accelerating voltage was 15 kV. The elemental composition of the pastes was determined by energy dispersive spectroscopy (EDS) using an X-max 150 X-ray energy spectrometer (Oxford instruments, London, England).

Mercury Intrusion Porosimetry
Porosity and pore size distribution were measured for the specimen fragments using mercury intrusion porosimetry (MIP, Quanta chrome PoreMaster 60 GT). A major advantage of MIP is that it has a wide range of pore size characteristics. With increasing pressure, the mercury breaks the interfacial tension and intrudes into the pores of cement paste. The applied pressure and the amount of mercury intrusion were used to determine the corresponding pore size and quantity. The pressure applied by the intrusion porosimeter ranged from 0 to 206 MPa, and the pore size in the range of 0.007 μm to 230 μm was measured.

Compressive Strength
BYT cement was used to test the compressive strength with a w/c ratio of 0.5. The slurry was stabilized in a 20 × 20 × 20 mm test mold, and the slurry was cured in a curing box with a temperature of 20 ± 0.1 °C and RH 95% for 24 h. Then, the test block was placed

Isothermal Conduction Calorimetry
Isothermal conduction calorimetry (8-channel TAM AIR calorimeter, Thermometric AB, Stockholm, Sweden) was used to measure the hydration heat liberation of BYT clinkers during the first 72 h at a constant temperature of 20 • C. To measure the initial thermal response of the samples, an internal stirring method was used in this experiment. An amount of 4 g of BYT clinker was weighed into a flask, and 2 g of water was added to an injector, which was then placed in the measuring position of the instrument. Before testing, the samples rested for 45 min to stabilize their temperature and the water temperature.

Thermogravimetric Analysis-Differential Scanning Calorimetry
To determine the cement hydration products, a TGA/DSC 1/1600LF synchronous thermal analyzer (Mettler toledo, New York, NY, USA) was used to perform a TG-DSC analysis of the hydration samples. Weight loss was measured by heating samples from room temperature to 1000 • C at a uniform heating rate of 10 • C/min in a nitrogen atmosphere.

Electron Microscopy
Microstructure and morphology of pastes were identified using a Zeiss-Supra 55 Scanning Electron Microscopy (SEM) (Carl Zeiss, Oberkochen, Germany) with an acceleration voltage of 15 kV. The sample was sputtered with gold in the vacuum to improve electrical conductivity. The backscattered scanning electron (BSE) mode was chosen, and the accelerating voltage was 15 kV. The elemental composition of the pastes was determined by energy dispersive spectroscopy (EDS) using an X-max 150 X-ray energy spectrometer (Oxford instruments, London, UK).

Mercury Intrusion Porosimetry
Porosity and pore size distribution were measured for the specimen fragments using mercury intrusion porosimetry (MIP, Quanta chrome PoreMaster 60 GT). A major advantage of MIP is that it has a wide range of pore size characteristics. With increasing pressure, the mercury breaks the interfacial tension and intrudes into the pores of cement paste. The applied pressure and the amount of mercury intrusion were used to determine the corresponding pore size and quantity. The pressure applied by the intrusion porosimeter ranged from 0 to 206 MPa, and the pore size in the range of 0.007 µm to 230 µm was measured.

Compressive Strength
BYT cement was used to test the compressive strength with a w/c ratio of 0.5. The slurry was stabilized in a 20 × 20 × 20 mm test mold, and the slurry was cured in a curing box with a temperature of 20 ± 0.1 • C and RH 95% for 24 h. Then, the test block was placed in water at 20 ± 0.1 • C to continue the curing process. The compressive strength was tested at 1 d, 3 d, 7 d, 28 d, 56 d, 90 d and 180 d.

The Degree of Hydration
The degree of hydration (DoH) was calculated from the result of XRD-Rietveld refinement, according to Equation (1) where the w t0 (C 2 S), w t0 (C 4 A 3 $), w t0 (C 5 S 2 $) represent the initial content of minerals phases in anhydrous BYT cement, and w t (C 2 S), w t (C 4 A 3 $), w t (C 5 S 2 $) represent the content of minerals in hydrated pastes at t. The content of minerals was normalized by the weight loss obtained from TG-DSC.

Phase Composition of BYT Clinker
The XRD patterns of the synthesized BYT clinkers are shown in Figure 2. BYT clinkers were synthesized at different temperatures, mainly because the content of ternesite was sensitive to temperature. The previous investigation claimed that the calcined temperature should be at 1180 • C for BYT clinkers with less than 15% ternesite and at 1210 • C for BYT clinkers with more than 15% ternesite. The XRD patterns mainly contain diffraction peaks of belite, ye elimite, ternesite and quartz for all the clinkers. Two polymorphs of ye'elimite were formed, including orthorhombic and cubic crystal forms. The phase compositions quantified through Rietveld analysis are shown in Table 5. The results show that the three targeted phases (belite, ye elimite and ternesite) formed adequately, and no free CaO was detected in the synthesized clinkers. The results indicated that these BYT clinkers were synthesized successfully.

The Degree of Hydration
The degree of hydration (DoH) was calculated from the result of XRD-Rietveld refinement, according to Equation (1) where the w C S , w C A $ , w C S $ represent the initial content of minerals phases in anhydrous BYT cement, and w C S , w C A $ , w C S $ represent the content of minerals in hydrated pastes at t. The content of minerals was normalized by the weight loss obtained from TG-DSC.

Phase Composition of BYT Clinker
The XRD patterns of the synthesized BYT clinkers are shown in Figure 2. BYT clinkers were synthesized at different temperatures, mainly because the content of ternesite was sensitive to temperature. The previous investigation claimed that the calcined temperature should be at 1180 °C for BYT clinkers with less than 15% ternesite and at 1210 °C for BYT clinkers with more than 15% ternesite. The XRD patterns mainly contain diffraction peaks of belite, ye′elimite, ternesite and quartz for all the clinkers. Two polymorphs of ye'elimite were formed, including orthorhombic and cubic crystal forms. The phase compositions quantified through Rietveld analysis are shown in Table 5. The results show that the three targeted phases (belite, ye′elimite and ternesite) formed adequately, and no free CaO was detected in the synthesized clinkers. The results indicated that these BYT clinkers were synthesized successfully.    Figure 3 shows the heat evolution curves and cumulative heat of the BYT cements. Only two main peaks were observed in the curves of all BYT cements, and the induction period between the two exothermic peaks was very short. Previous studies have shown that the first exothermic peak occurs directly when the water is added and can be attributed to the solution heat and the rapid hydration of ye elimite. The first peak is higher for the BYTC-2 and BYTC-4 in comparison with the other cements because the contents of ye elimite are the highest. For the BYTC-1, BYTC-3 and BYTC-5, their first exothermic peaks are similar due to the same content of ye elimite in these samples. It was reported that the second exothermic peaks occur after the massive dissolution of ye elimite (+gypsum) and precipitation of ettringite (+AH 3 ) [43]. The second peak of cement BYTC-1, BYTC-2, BYTC-3, BYTC-4 and BYTC-5 appeared at approximately 1.5 h, 1.7 h, 1.9 h, 1.9 h and 2.1 h, respectively. It seems that the time when the second exothermic peak appears is related to the content of ternesite. The higher the content of ternesite in the system, the later the second exothermic peak appears. It demonstrated that ternesite delays the second exothermic peak of ye elimite. After 10 h hydration, the heat flow is close to zero for all the samples. It can be seen from the hydration heat release diagram that the accumulated heat release of BYTC-4 cement is the highest and can reach 205 J/g at approximately 30 h hydration, mainly because the content of ye elimite in this cement is the highest. Through the hydration heat analysis of different BYT cement compositions, when the hydration process reached 30 h, the hydration curve basically remained stable, indicating that the hydration process entered a stable period and that hydration was completely controlled by the diffusion rate.  Figure 3 shows the heat evolution curves and cumulative heat of the BYT cements. Only two main peaks were observed in the curves of all BYT cements, and the induction period between the two exothermic peaks was very short. Previous studies have shown that the first exothermic peak occurs directly when the water is added and can be attributed to the solution heat and the rapid hydration of ye′elimite. The first peak is higher for the BYTC-2 and BYTC-4 in comparison with the other cements because the contents of ye′elimite are the highest. For the BYTC-1, BYTC-3 and BYTC-5, their first exothermic peaks are similar due to the same content of ye′elimite in these samples. It was reported that the second exothermic peaks occur after the massive dissolution of ye′elimite (+gypsum) and precipitation of ettringite (+AH3) [43]. The second peak of cement BYTC-1, BYTC-2, BYTC-3, BYTC-4 and BYTC-5 appeared at approximately 1.5 h, 1.7 h, 1.9 h, 1.9 h and 2.1 h, respectively. It seems that the time when the second exothermic peak appears is related to the content of ternesite. The higher the content of ternesite in the system, the later the second exothermic peak appears. It demonstrated that ternesite delays the second exothermic peak of ye′elimite. After 10 h hydration, the heat flow is close to zero for all the samples. It can be seen from the hydration heat release diagram that the accumulated heat release of BYTC-4 cement is the highest and can reach 205 J/g at approximately 30 h hydration, mainly because the content of ye′elimite in this cement is the highest. Through the hydration heat analysis of different BYT cement compositions, when the hydration process reached 30 h, the hydration curve basically remained stable, indicating that the hydration process entered a stable period and that hydration was completely controlled by the diffusion rate.

Phase Composition of the Hydrated Cement Paste
The XRD spectra of hydration of BYT cements at different ages are listed in Figure 4. The major diffraction peak varies within the range of 2 theta between 5 • and 40 • . The variation of phase compositions was similar with hydration age for all the samples. At the early hydration (from 1 d to 28 d), the phase assemblages included the remained phases (ye elimite, belite); neither portlandite nor C-S-H was observed. The absence of AH 3 in XRD patterns is related to its amorphous nature [44]. The diffraction peak of ye elimite basically disappeared in the first 7 d of hydration, corresponding to the rapid weakening of the gypsum diffraction peak and the rapid strengthening of the ettringite diffraction peak. The strätlingite diffraction peak is observed only in the diffraction pattern of BYTC-2 cement, which is related to the rapid consumption of gypsum in the paste. The peak intensity of ternesite decreased in the first 28 d of hydration, but part of the peak remained. Gypsum from the hydration of ternesite was not found, probably because it was involved in the formation of ettringite. These results indicated that ternesite can be activated in the presence of ye elimite. of the gypsum diffraction peak and the rapid strengthening of the ettringite diffraction peak. The strätlingite diffraction peak is observed only in the diffraction pattern of BYTC-2 cement, which is related to the rapid consumption of gypsum in the paste. The peak intensity of ternesite decreased in the first 28 d of hydration, but part of the peak remained. Gypsum from the hydration of ternesite was not found, probably because it was involved in the formation of ettringite. These results indicated that ternesite can be activated in the presence of ye′elimite. The main hydration products were further determined by TG-DSC analysis. TG-DSC curves of BYTC-2 cement with different hydration ages are shown in Figure 5. The two endothermic peaks of the hydrated sample are located at 120 • C and 250 • C. The presence of the first endothermic peak was due to the thermal decomposition of ettringite, leading to the greatest weight loss. Weight loss at around 200 • C was related to the decomposition of strätlingite. The third endothermic peak was due to the dehydration of AH 3 , which was formed during the hydration of ye elimite and was partly consumed during the hydration of belite and ternesite. AH 3 is not observed in the XRD patterns, which may be due to its lower weight percentage and poor crystallinity. The total weight loss percentages of the hydrated samples on 1 d, 3 d and 7 d were 22.71%, 25.07%, 28.76%, respectively. of the first endothermic peak was due to the thermal decomposition of ettringite, leading to the greatest weight loss. Weight loss at around 200 °C was related to the decomposition of strätlingite. The third endothermic peak was due to the dehydration of AH3, which was formed during the hydration of ye′elimite and was partly consumed during the hydration of belite and ternesite. AH3 is not observed in the XRD patterns, which may be due to its lower weight percentage and poor crystallinity. The total weight loss percentages of the hydrated samples on 1 d, 3 d and 7 d were 22.71%, 25.07%, 28.76%, respectively.

The Hydration Degree of BYT Cement
The XRD Rietveld quantification method was used to determine the content of belite, ye′elimite and ternesite at different ages ( Figure 6). The content of ettringite increases particularly rapidly in the first 7 d of hydration, which corresponds to the sharp decrease in the ye′elimite content. Afterward, the content of ettringite increases slowly, which is related to the crystallinity improvement. It should be noted that because ettringite (as the main hydration product) is combined with a large amount of water, the mass fraction of various phases can be diluted to varying degrees in different BYT cements and hydration ages. Therefore, three-mineral phase content is normalized by the anhydrous sample mass. The early hydration activity of belite was higher in BYT cement than in PC cement. The initial contents of belite in BYTC-2 and BYTC-3 were similar. After 28 d of hydration, the content of belite in BYTC-3 changed very slowly, while that in BYTC-2 slowed down after 90 d. The difference in the content of ternesite may contribute greatly to the activity of belite. Ye′elimite is almost completely consumed after 7 d of hydration, which corresponds to its high hydration activity. The content of ternesite decreased significantly in the first 7 d because the hydration activation of ternesite was stimulated in the presence

The Hydration Degree of BYT Cement
The XRD Rietveld quantification method was used to determine the content of belite, ye elimite and ternesite at different ages ( Figure 6). The content of ettringite increases particularly rapidly in the first 7 d of hydration, which corresponds to the sharp decrease in the ye elimite content. Afterward, the content of ettringite increases slowly, which is related to the crystallinity improvement. It should be noted that because ettringite (as the main hydration product) is combined with a large amount of water, the mass fraction of various phases can be diluted to varying degrees in different BYT cements and hydration ages. Therefore, three-mineral phase content is normalized by the anhydrous sample mass. The early hydration activity of belite was higher in BYT cement than in PC cement. The initial contents of belite in BYTC-2 and BYTC-3 were similar. After 28 d of hydration, the content of belite in BYTC-3 changed very slowly, while that in BYTC-2 slowed down after 90 d. The difference in the content of ternesite may contribute greatly to the activity of belite. Ye elimite is almost completely consumed after 7 d of hydration, which corresponds to its high hydration activity. The content of ternesite decreased significantly in the first 7 d because the hydration activation of ternesite was stimulated in the presence of ye elimite. The hydration activity of the three minerals is inconsistent in different design compositions, which also indicates that appropriate composition design may result in better synergy in the hydration process of the three minerals. Figure 7 shows the hydration degree (DoH) of BYT cement at all hydration ages. The hydration degree of cement is mainly related to the content of ye elimite. Due to the high hydration activity of ye elimite, when the content of ye elimite increases, the hydration degree of cement is higher. BYTC-4, which has the highest ye elimite content (60 wt.% in clinker), has the highest hydration degree, followed by BYTC-2 (50 wt.% in clinker). It is noteworthy that while the ye elimite contents in BYTC-1, BYTC-3 and BYTC-5 are comparable, the contents of belite and ternesite are different, and the hydration degree of samples with more ternesite is higher, which indicates that the hydration activity of ternesite is higher than that of belite. of ye′elimite. The hydration activity of the three minerals is inconsistent in different design compositions, which also indicates that appropriate composition design may result in better synergy in the hydration process of the three minerals. Figure 6. Evolution of belite, ye′elimite, ternesite and ettringite obtained by Rietveld refinement (measurement error is ±2%). Note: the y-axis scale is different. Figure 7 shows the hydration degree (DoH) of BYT cement at all hydration ages. The hydration degree of cement is mainly related to the content of ye′elimite. Due to the high hydration activity of ye′elimite, when the content of ye′elimite increases, the hydration degree of cement is higher. BYTC-4, which has the highest ye′elimite content (60 wt.% in clinker), has the highest hydration degree, followed by BYTC-2 (50 wt.% in clinker). It is noteworthy that while the ye′elimite contents in BYTC-1, BYTC-3 and BYTC-5 are comparable, the contents of belite and ternesite are different, and the hydration degree of samples with more ternesite is higher, which indicates that the hydration activity of ternesite is higher than that of belite.    Figure 7 shows the hydration degree (DoH) of BYT cement at all hydration ages. The hydration degree of cement is mainly related to the content of ye′elimite. Due to the high hydration activity of ye′elimite, when the content of ye′elimite increases, the hydration degree of cement is higher. BYTC-4, which has the highest ye′elimite content (60 wt.% in clinker), has the highest hydration degree, followed by BYTC-2 (50 wt.% in clinker). It is noteworthy that while the ye′elimite contents in BYTC-1, BYTC-3 and BYTC-5 are comparable, the contents of belite and ternesite are different, and the hydration degree of samples with more ternesite is higher, which indicates that the hydration activity of ternesite is higher than that of belite.   Figure 8 shows the typical microstructure of hardened BYT cement systems hydration for 7 d. Most areas of hardened slurry are morphologies as shown in Figure 8a. A great quantity of needle and columnar ettringite appeared in the hydrated sample, as well as a large amount of amorphous gels. Amorphous gels are attached to ettringite surfaces and interstices. Song [45] et al. proved that the amorphous gels were mainly composed of AH 3 precipitated onto ettringite surface by EDS and TEM. Padilla-Encinas [46,47] et al. considered that amorphous gels and low crystallinity AH 3 , which found it easier to fill the pores between the ettringite needles, could improve the matrix properties. The results of EDS on ettringite surface show a higher content of aluminum than stoichiometric ettringite, which can be explained by the hydration equation of ye elimite (Equation (2)). Compared with stoichiometric ettringite, the co-existence of ettringite and AH 3 leads to a higher proportion of Al element. In addition, hydration product strätlingite of belite and ternesite also affects the ratio of the elements. considered that amorphous gels and low crystallinity AH3, which found it easier to fill the pores between the ettringite needles, could improve the matrix properties. The results of EDS on ettringite surface show a higher content of aluminum than stoichiometric ettringite, which can be explained by the hydration equation of ye′elimite (Equation (2)). Compared with stoichiometric ettringite, the co-existence of ettringite and AH3 leads to a higher proportion of Al element. In addition, hydration product strätlingite of belite and ternesite also affects the ratio of the elements. SEM images (Figure 9a,b) show the morphology of ternesite at 7 d of hydration. The etch pits are present in abundance on the surface of ternesite, and there are abundant needle ettringite in the depressions. The proper explanation is with the reaction between AH3 and ternesite. Ternesite gradually loses part of its structure, which leads to nucleation of small pits on the surface. Needle ettringite on the surface can be regarded as the hydration product of ternesite in an Al-rich pore solution [13,28]. The selected BSE area of ternesite is shown in Figure 9c, and elemental maps are depicted in Figure 9d. It can be observed that Ca and Si are enriched in the ternesite-rich zone, while Al is basically outside the ternesite rich zone. S is distributed in all areas, but it is extremely abundant at the ternesite boundary. This may be caused by the accumulation of ettringite on the surface of ternesite. SEM images (Figure 9a,b) show the morphology of ternesite at 7 d of hydration. The etch pits are present in abundance on the surface of ternesite, and there are abundant needle ettringite in the depressions. The proper explanation is with the reaction between AH 3 and ternesite. Ternesite gradually loses part of its structure, which leads to nucleation of small pits on the surface. Needle ettringite on the surface can be regarded as the hydration product of ternesite in an Al-rich pore solution [13,28]. The selected BSE area of ternesite is shown in Figure 9c, and elemental maps are depicted in Figure 9d. It can be observed that Ca and Si are enriched in the ternesite-rich zone, while Al is basically outside the ternesite rich zone. S is distributed in all areas, but it is extremely abundant at the ternesite boundary. This may be caused by the accumulation of ettringite on the surface of ternesite.

Proposed Hydration Mechanisms of Ternesite
Based on the analysis above, it can be stated that the reaction between ternesite and AH3 mainly occurs on the surface of ternesite. The prerequisite of reaction is that there is enough AH3 around the ternesite. The hydration process on the surface of ternesite follows the steps shown in Figure 10. Firstly, ternesite reacts with AH3 in direct contact to form strätlingite and CaSO4 by means of ion exchange (Equation (5)). Meanwhile, pits will be formed on the surface of ternesite, and the concentration of AH3 will also decrease. The CaSO4 formed by hydration of ternesite will continue to react with AH3 near the pits to

Proposed Hydration Mechanisms of Ternesite
Based on the analysis above, it can be stated that the reaction between ternesite and AH 3 mainly occurs on the surface of ternesite. The prerequisite of reaction is that there is enough AH 3 around the ternesite. The hydration process on the surface of ternesite follows the steps shown in Figure 10. Firstly, ternesite reacts with AH 3 in direct contact to form strätlingite and CaSO 4 by means of ion exchange (Equation (5)). Meanwhile, pits will be formed on the surface of ternesite, and the concentration of AH 3 will also decrease. The CaSO 4 formed by hydration of ternesite will continue to react with AH 3 near the pits to form ettringite, which also requires the participation of Ca 2+ diffused from other areas. The needle ettringite accumulates on the surface of ternesite and inhibits ion migration, thus restricting further hydration of ternesite.

Proposed Hydration Mechanisms of Ternesite
Based on the analysis above, it can be stated that the reaction between ternesite and AH3 mainly occurs on the surface of ternesite. The prerequisite of reaction is that there is enough AH3 around the ternesite. The hydration process on the surface of ternesite follows the steps shown in Figure 10. Firstly, ternesite reacts with AH3 in direct contact to form strätlingite and CaSO4 by means of ion exchange (Equation (5)). Meanwhile, pits will be formed on the surface of ternesite, and the concentration of AH3 will also decrease. The CaSO4 formed by hydration of ternesite will continue to react with AH3 near the pits to form ettringite, which also requires the participation of Ca 2+ diffused from other areas. The needle ettringite accumulates on the surface of ternesite and inhibits ion migration, thus restricting further hydration of ternesite.   Figure 11 shows the porosity distribution of BYTC-2 hydration at 3 d and 28 d. According to the effect of different pore sizes on cement performance and hydration, the pore size can be divided into large pores (>0.5 µm), medium pores (0.01-0.5 µm) and gel pores (<0.01 µm). The pore volume of BYT cement after hydration is higher on day 3 than on the other days, and the pore has double peaks or a wide peak, which indicates that the pore is widely distributed. After hydration for 7 d, the pore distribution of the paste is concentrated in the medium pores, and the number of large pores and gel pores decreases. The growth of ettringite transforms the large pores into medium pores, making the cement paste denser. On 28 d, a large number of pores larger than 0.1 microns in diameter were filled by hydration products, which reduced the pore volume, and the pores larger than 0.1 microns in the BYT cement samples basically disappeared. The appearance of pore peaks is related not only to the concentrated distribution of pore sizes but also to the strength of cement paste. The pore volume of the BYTC-2 sample is basically lower than that of the other samples, which is consistent with its good strength. Figure 12 shows the compressive strength development of BYT cement. The compressive strength of the BYT sample within 56 d of hydration age is greatly affected by the content of ternesite. The higher the content of ternesite, the weaker the early strength. After hydration for more than 56 d, the strength still grows slowly, and there is no strength decrease phenomenon. This can be explained by the continued hydration of ternesite and belite in the system. It can be seen from the figure that with the extension of hydration time, the compressive strength of each sample presents a slightly upward trend. The highest compressive strength for each age is observed with the BYTC-2 sample, of which compressive strengths of 90 d and 180 d reach 51 MPa and 52 MPa, respectively. Figure 13 shows the compressive strength of the BYT cement relation with the ternesite content in the clinker after 180 d of hydration. As the content of ternesite samples reached 10%, the compressive strength performance was better than that of the other samples after 180 d of hydration. pore is widely distributed. After hydration for 7 d, the pore distribution of the paste is concentrated in the medium pores, and the number of large pores and gel pores decreases. The growth of ettringite transforms the large pores into medium pores, making the cement paste denser. On 28 d, a large number of pores larger than 0.1 microns in diameter were filled by hydration products, which reduced the pore volume, and the pores larger than 0.1 microns in the BYT cement samples basically disappeared. The appearance of pore peaks is related not only to the concentrated distribution of pore sizes but also to the strength of cement paste. The pore volume of the BYTC-2 sample is basically lower than that of the other samples, which is consistent with its good strength.  Figure 12 shows the compressive strength development of BYT cement. The compressive strength of the BYT sample within 56 d of hydration age is greatly affected by the content of ternesite. The higher the content of ternesite, the weaker the early strength. After hydration for more than 56 d, the strength still grows slowly, and there is no strength decrease phenomenon. This can be explained by the continued hydration of ternesite and belite in the system. It can be seen from the figure that with the extension of hydration time, the compressive strength of each sample presents a slightly upward trend. The highest compressive strength for each age is observed with the BYTC-2 sample, of which compressive strengths of 90 d and 180 d reach 51 MPa and 52 MPa, respectively.

Conclusions
In this paper, the hydration process and hydration products of BYT cement clinker and the compressive strength of BYT cement were studied. The synergistic effect among belite, ye′elimite and ternesite was discussed, and the following conclusions were drawn:

Conclusions
In this paper, the hydration process and hydration products of BYT cement clinker and the compressive strength of BYT cement were studied. The synergistic effect among

Conclusions
In this paper, the hydration process and hydration products of BYT cement clinker and the compressive strength of BYT cement were studied. The synergistic effect among belite, ye elimite and ternesite was discussed, and the following conclusions were drawn: (1) The presence of ternesite in cement delays the early hydration of ye elimite, resulting in a low rate of early hydration. The higher the content of ternesite in the system, the later the second hydrating exothermic peak. (2) Ternesite showed good hydration activity in the hydration process of BYT cement. The presence of ternesite promotes belite hydration activity. With increasing hydration age, the content of ettringite increased, and the crystallinity improved, so the intensity of the diffraction peak of ettringite gradually strengthened. AH 3 precipitated in an amorphous form and was not detected by XRD. The hydration of ternesite mainly occurs on the surface. The presence of ettringite on the surface of ternesite may hinder further hydration of ternesite. (3) With the increase in hydration time, the pore volume of cement paste decreased in each sample, which promoted and improved the strength of the cement sample. The compressive strength of the BYT cement samples did not reverse when the hydration age exceeded 56 d but still grew slowly, which was due to the involvement of ternesite in the system during hydration. The strength of BYTC-2 (containing 10% ternesite in the prepared clinker) exceeds that of other cements at all ages.