Influence Mechanism of Accelerator on the Hydration and Microstructural Properties of Portland Cement

Abstract

Shotcrete is one of the most important types of concrete used in engineering construction, and its properties are significantly influenced by accelerators. This study investigates the effects of aluminum sulfate series alkali-free accelerator (AKF) and alkali accelerator (ALK) on the strength, hydration process, characteristic hydration products, and microstructure properties of shotcrete. Techniques such as setting time measurement, isothermal calorimetry, simultaneous thermal analysis, scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM–EDS), and mercury intrusion porosimetry (MIP) were utilized. The results indicate that both ALK and AKF significantly accelerate and increase the early hydration heat release rate and cumulative hydration heat of Portland cement, producing the characteristic hydration products hexagonal plate AFm and rod AFt, respectively. This acceleration notably speeds up the setting process of Portland cement. ALK negatively impacts the later-stage microstructural development and pore structure filling of hardened cement paste, leading to average reductions of 15.3% and 19.9% in flexural and compressive strengths at 28 days, respectively. Specifically, compared to ALK, AKF shows a faster hydration heat release rate during the induction period and a more significant increase in cumulative hydration heat during the hydration process; the cumulative hydration heat is on average 18.2% higher than AKF. Furthermore, AKF does not hinder the subsequent C₃S hydration and C-S-H gel densification process. After 28 days of curing, EDS analysis indicates an average Ca/Si ratio of 1.171 for the AKF-treated shotcrete; the average Ca/Si ratio shows minimal variation from the reference group and is classified as the same type of C-S-H gel as the reference group. Therefore, the strength of hardened cement paste with AKF continues to increase steadily in the later stages. At 28 days, the average flexural strength increased by 10.2%, while the compressive strength decreased by only 3.0%. These findings suggest that AKF enhances the microstructural development and strength of shotcrete, making it a more effective accelerator for engineering applications.

1. Introduction

Concrete is one of the most widely used construction materials globally, widely recognized for its strength, durability, structural integrity, and versatility [1,2]. It forms a composite material capable of withstanding various environmental loads and stresses, making it essential in various engineering applications [3,4]. Recent advancements in technology have led to specialized types of concrete designed to meet specific engineering requirements [5,6,7]. Shotcrete is one such innovation, a type of concrete that is rapidly sprayed onto the sprayed surface and instantaneously compacted through an inflatable hose or pipeline under the action of air pressure [8]. Shotcrete is widely used in railway, highway, water conservancy, and mining engineering due to its rapid final setting time, fast hydration and hardening speed, high early-age strength, and efficient application process [9]. Accelerating agents, which are admixtures that enhance the setting and hardening speeds of shotcrete, play a crucial role in shotcrete applications. These agents not only accelerate the setting and hardening processes but also increase the build-up thickness and reduce the interval time between the sprayed layers [10,11]. They are essential components in shotcrete construction [12]. Accelerating agents are primarily categorized into two types: alkali-based and alkali-free agents. Aluminate alkali accelerators, such as sodium aluminate (NaAlO₂) and potassium aluminate (KAlO₂), are currently the most commonly used accelerators in the market [13]. However, there is a general belief that alkali accelerators may negatively impact shotcrete strength. Research by Renan P. S. indicates that NaAlO₂ rapidly consumes gypsum in cement, leading to the formation of AFm and C-A-H hydration products [14]. These products can block the dissolution sites of Alite and inhibit its hydration, which may hinder strength development over time. Han Jianguo observed that, while NaAlO₂ significantly improves the hydration heat release rate during the pre-induction period, it reduces the rate of heat release after the acceleration period. The addition of NaAlO₂ in the early stage of hydration (≤24 h) refines the pore structure of the cement paste and increases the pore content in the 5–30 nm range [15]. This effect is similar to that observed with nanomaterials, which refine the pore structure in cementitious composites, but the mechanism of action is distinct. NaAlO₂ accelerates the reaction with gypsum in cement, causing the rapid depletion of gypsum. Once the gypsum is depleted, it promptly reacts with Ca(OH)₂ in the solution to form AFm, which reduces the Ca²⁺ concentration in the solution. This process promotes the hydration of silicate minerals in cement, particularly C₃S, resulting in the formation of C-S-H gel and facilitating early pore refinement of the cement paste. Nanomaterials, due to their small size and high surface energy, have the potential to act as nucleation seeding sites for the hydration products [16]. During the hydration process of cement-based materials, the nanomaterials achieved through acceleration in the C-S-H gel formation restrain the growth of the CH crystals and modify their orientation index to refine the microstructural environment [17,18].

Furthermore, N. Smaoui’s research suggests that NaAlO₂ hydrolysis results in NaOH, which produces porous hydration products that are detrimental to compressive strength development [19].

In contrast, alkali-free accelerators, such as aluminum sulfate Al₂(SO₄)₃, are increasingly replacing alkali-based accelerators due to their high long-term strength retention; absence of alkalis and chlorides; and their safety, environmental benefits, and durability.

Research by Paglia suggests that the addition of an Al₂(SO₄)₃ accelerator promotes the formation of numerous prismatic AFt crystals that overlap on the surface of cement particles, which promotes rapid cement setting [20,21,22]. Renan P. S. found that Al₂(SO₄)₃ delays the hydration reaction of C₃A, avoids early AFm formation, and facilitates unimpeded hydration of alite, resulting in higher compressive strength compared to NaAlO₂ [23,24]. Dinoia and Sandberg used calorimetry to study the effect of an Al₂(SO₄)₃ accelerator on the hydration of different types of Portland cement [25]. The results showed that Al₂(SO₄)₃ promotes C₃A hydration but inhibits Alite hydration. Han JG et al. investigated the effects of aluminum sulfate (AS) on cement hydration products, pore structure, and strength, discovering that AS promotes calcium aluminate hydration but inhibits calcium silicate hydration [26]. Maltese et al. studied the effects of Al₂(SO₄)₃ and different gypsum types on cement setting and hydration products, concluding that the setting effect of Al₂(SO₄)₃ is influenced by gypsum dissolution rates, with slower dissolution rates improving setting performance [27].

Although significant research has been conducted, most studies focus on the formulation and synthesis of accelerators and their effects on early hydration reactions and setting mechanisms. There is limited research on how accelerators affect the microstructure of shotcrete, and the mechanisms underlying accelerators’ impact on cement hydration remain controversial. This study examines commonly used aluminum sulfate alkali-free accelerators and sodium aluminate alkali accelerators available in the market. It systematically studies their impact on cement hydration and microstructure, analyzing their hydration processes, characteristic hydration products, pore structures, and matrix microstructures. The results aim to provide a theoretical foundation for understanding the hydration characteristics and strength development of shotcrete.

2. Materials and Methods

2.1. Raw Material

The primary material employed in this study was Portland cement (P·I 42.5, conforming to the Chinese National Standard GB 8076, equivalent to CEMI 42.5). The detailed chemical composition and physical properties are provided in

Table 1. Chemical and mineral composition of cement clinker (%).

SiO2	Al2O3	Fe2O3	CaO	MgO	SO3	Na2O	f-CaO	C3S	C2S	C3A	C4AF
22.64	4.68	3.57	64.88	2.94	0.35	0.49	0.91	56.53	21.94	6.44	10.40

and

Table 2. Cement physical properties.

Fineness 0.08 mm (%)	Density/ (g∙cm−3)	Specific Surface Area (m2·kg−1)	Setting time (Minutes)		Flexural Strength (MPa)		Compressive Strength (MPa)
			Initial	Final	3 d	28 d	3 d	28 d
1.2	3.16	340	159	214	5.7	8.2	26.2	50.4

. The accelerators used were SBT^®-N (I) alkali liquid accelerator (referred to as ALK) and SBT^®-N (II) alkali-free liquid accelerator (referred to as AKF).

The concentration of ALK and AKF is expressed as a weight percentage of the cement in the cement paste. For the setting time tests, concentrations of 2%, 4%, 6%, and 8% ALK and 4%, 6%, 8%, 9%, 10%, and 12%AKF were used. For the hydration heat tests, 2% and 4% ALK and 4% and 8% AKF were applied. For the thermal analysis and scanning electron microscopy tests, 4% ALK and 8% AKF were utilized. All tests were compared with a reference group that did not contain any accelerators. The mix ratios are shown in

Table 3. Mix proportion for test.

Cement	ALK	AKF	Water
1.000	—	—	0.350
1.000	0.020	—	0.340
1.000	0.040	—	0.330
1.000	0.060	—	0.320
1.000	0.080	—	0.310
1.000	—	0.040	0.330
1.000	—	0.060	0.320
1.000	—	0.080	0.310
1.000	—	0.090	0.305
1.000	—	0.100	0.300
1.000	—	0.120	0.290

.

2.2. Experimental Method

In this experiment, the effects of ALK and AKF on the hydration and microstructural properties of Portland cement were analyzed by the setting time, the compressive and flexural test, hydration heat thermos gravimetric analysis, the porosity test, and the scanning electron microscopy test.

Table 4. Grouping of hydration and microstructural properties test.

Performance Index	Specimen Size	Quantity
setting time	—	—
compressive and flexural strength	160 mm × 40 mm × 40 mm	45
hydration heat	—	—
thermos gravimetric	40 mm × 40 mm × 40 mm	12
porosity	40 mm × 40 mm × 40 mm	12
scanning electron microscopy	40 mm × 40 mm × 40 mm	6

shows the grouping of the test, including the size and number of each test and specimen.

2.2.1. Setting Time

The setting time was tested according to GB/T 35159 (flash setting admixtures for shotcrete) [28]. The water–cement ratio used for determining the setting time of the cement paste was 0.35, and the average value was taken from three experiments in each group.

2.2.2. Strength

The hardened cement paste specimens, measuring 40 mm × 40 mm × 160 mm, were selected for strength testing. The tests were conducted at specified ages of 1, 3, and 28 days in accordance with (GB/T 17671—2021, equivalent to ISO 679:2009) [29]. The loading rates applied for measuring the flexural and compressive strengths were 0.05 kN/s and 2.4 kN/s, respectively.

2.2.3. Hydration Heat

The hydration heat was measured using a TAM AIR isothermal calorimeter (TA Instruments, New Castle, Delaware, American). The test was conducted at a temperature of 25 °C, with a temperature fluctuation range of less than 0.02 °C and a measurement accuracy of ±20 μW. The duration of the test was 4 days.

2.2.4. Thermos Gravimetric Analysis

A STA-449F3 simultaneous thermal analyzer (NETZSCH Group, Selb, Germany) was employed, with the test sample weighing approximately 15 mg. The heating rate was controlled at 10 °C/min, and the maximum temperature was raised to 1000 °C. The DTG curves for each group of samples were obtained.

2.2.5. Porosity Test

The porosity and pore size distribution of the hardened cement paste were analyzed using the mercury intrusion method. Small-size paste specimens were prepared for standard curing (20 °C ± 2 °C, RH > 95%) [30]. After curing for 1, 3, 7, and 28 days, the hydrated samples were broken using pliers for testing. The porosity was measured using a Poremaster-60T automatic mercury porosimeter(Anton Paar QuantaTec Inc., Boynton Beach, FL, USA)with an aperture measurement range of 3.5 nm to 360,000 nm.

2.2.6. Scanning Electron Microscopy Test

The microstructure of the cement paste samples was observed using a Quanta FEG 250 field emission environmental scanning electron microscope (FEI Ltd., Hillsboro, OR, USA). After curing to the specified age, the samples were broken with pliers, and hydration was stopped before testing [31]. The complete experimental procedures and research methodology are shown in Figure 1.

3. Experiment Results and Analysis

3.1. Effect of Accelerator on Setting Time

Figure 2 shows the effect of accelerator content on setting time. The ALK accelerator exhibits a three-stage change, while the AKF accelerator demonstrates a two-stage change. Both accelerators cause a rapid reduction in setting time at low dosages, with the change stabilizing at higher dosages.

Figure 2a illustrates the effect of ALK on the setting time of cement paste. The initial and final setting times of the reference sample (REF) are 98 and 157 min, respectively. When the ALK dosage reaches 2%, the initial and final setting times decrease significantly to 10.6 and 17.1 min, respectively.

This indicates that a low concentration of ALK significantly accelerates the setting process. As the accelerator dosage increases, it enhances the chemical reactions responsible for setting, leading to a faster transition from a liquid to a solid state. In the second stage, between 2% and 4% dosage, the setting time decreases more slowly. This suggests that, while additional amounts of ALK continue to accelerate the setting process, the rate of acceleration diminishes. The cement paste reaches a point where further increases in the dosage have a reduced impact on the setting time. This behavior can be attributed to the diminishing returns of accelerator effectiveness as the reaction reaches a more complete state. Beyond the 4% dosage, the setting time changes little, indicating that the setting process is near completion, and additional increases in ALK concentration do not significantly affect the setting time. At this stage, the cement paste’s chemical system has reached a saturation point where the accelerator’s impact is minimal.

Figure 2b shows the effect of AKF on the setting time of cement paste. When the dosage of AKF reaches 8%, the initial and final setting times decrease rapidly to 4.9 and 10.2 min, respectively. Similar to ALK, a lower concentration of AKF significantly accelerates the setting process. The rapid decrease in setting time with increasing dosage reflects the accelerator’s effectiveness in speeding up the cement hydration process. However, once the dosage exceeds 8%, the setting time remains relatively stable despite further increases in the accelerator dosage. This pattern suggests that AKF accelerates the setting time effectively up to a certain point, beyond which additional amounts do not contribute significantly to a further reduction in setting time. This stabilization indicates that the maximum effect of AKF on setting time has been achieved, and additional quantities do not enhance the process.

3.2. Effect of Accelerator on Strength

Figure 3 illustrates the effect of accelerators on the flexural strength of hardened cement paste. The flexural strength of cement paste with the ALK accelerator shows an increasing trend in the early stages but a decreasing trend in the later stages at different dosages, with higher dosages not conducive to strength improvement. In contrast, the AKF accelerator significantly enhances flexural strength at all ages, with higher dosages leading to more pronounced strength improvements.

Figure 3a presents the effects of alkaline accelerators on the flexural strength. It can be observed that ALK-2% and ALK-4% exhibit higher flexural strengths than the reference sample (REF) at 1 day. However, after 1 day, the rate of increase in flexural strength for ALK-2% and ALK-4% is notably slower than the REF group. After 28 days, the flexural strengths of ALK-2% and ALK-4% decrease by 9.49% and 21.08%, respectively, compared to REF. This suggests that increasing the dosage of alkaline accelerators hinders the development of flexural strength, with ALK-4% consistently showing lower flexural strength than ALK-2% across different curing periods.

Figure 3b demonstrates the impact of alkali-free accelerators on flexural strength. It can be observed that, across various curing ages, AKF-4% and AKF-8% achieve higher flexural strengths than REF, with the strength increasing as the dosage of the accelerator rises. At 1 day, AKF-4% and AKF-8% exhibit significantly higher flexural strengths than REF, with increases of 91.21% and 95.40%, respectively. Although the rate of increase in flexural strength for AKF-4% and AKF-8% begins to decelerate after 3 days compared to REF, their flexural strengths remain substantially higher. After 28 days, the flexural strengths of AKF-4% and AKF-8% are 8.13% and 12.35% higher than REF, respectively.

Figure 4 illustrates the impact of accelerators on the compressive strength of hardened cement paste. The ALK accelerator improves the early compressive strength of cement paste, but the later strength decreases obviously, while the AKF accelerator improves the compressive strength in the early and late stages, especially in the later stage, and the strength performance is better than the ALK accelerator.

Figure 4a reveals that increasing the dosage of alkaline accelerators significantly enhances the early compressive strength, with ALK-4% showing a 33.70% increase compared to the reference sample (REF) at 1 day. However, after 1 day, the rate of strength growth for the different dosages of alkaline accelerators slows compared to REF. At 28 days, the compressive strengths of ALK-2% and ALK-4% decrease by 24.75% and 15.00%, respectively, compared to REF.

As can be observed in Figure 4b, increasing the dosage of alkali-free accelerators also improves early compressive strength. Unlike the pattern observed with alkaline accelerators, the growth rate of compressive strength for AKF-4% and AKF-8% does not show a significant reduction compared to the control group between 1 and 3 days. After 3 days, the growth rate of compressive strength for AKF-4% and AKF-8% becomes lower than REF. At 28 days, the compressive strengths of AKF-4% and AKF-8% are reduced by 4.46% and 1.47%, respectively, compared to REF. In contrast, the use of alkali-free accelerators is more beneficial for the development of later strength in the hardened cement paste.

3.3. Effect of Accelerator on Hydration Process

Figure 5 shows the hydration heat release curve of cement paste mixed with ALK. The initial ALK accelerates the release of hydration heat and then enters the stationary phase, but it is still higher than the reference group, and the acceleration period is lower than the reference group. ALK accelerates heat evolution during the pre-induction and induction periods but reduces it during the acceleration period, with greater dosages leading to more significant effects.

Figure 5a illustrates the effect of ALK on the hydration heat release rate of cement paste. After the cement is in contact with water, the cement undergoes a vigorous exothermic reaction. The hydration heat release rate for mixtures with ALK accelerators is significantly higher than that of the reference group, and it increases with higher accelerator dosages. However, after this rapid heat release, the hydration process enters a dormant or induction period. During this phase, the heat release rate for the ALK mixtures decreases significantly, though it remains higher than the reference group. Following the induction period, the hydration rate enters the acceleration phase, where the exothermic rate of the ALK mixtures is lower than that of the reference group and decreases further with increased dosage.

Figure 5b depicts the cumulative hydration heat for cement paste mixed with ALK accelerators. From the figure, it can be observed that the heat of hydration of the ALK mixtures exhibit a three-stage development pattern. In the first 3 h, the accumulated hydration heat of the ALK mixtures increases significantly compared to the reference group, indicating a higher early heat evolution rate. Between 3 and 24 h, the rate of heat accumulation for the ALK mixtures is lower than that of the reference group, with a more pronounced effect at higher dosages. After 24 h, the accumulated heat increase rate for the ALK-2% group aligns with the reference group, while the ALK-4% group shows significantly higher accumulated heat than both the reference and ALK-2% groups.

Figure 6 shows the hydration heat release curve of cement paste mixed with AKF. Both ALK and AKF accelerators enhance the hydration heat release rate during the pre-induction and induction periods of cement hydration. AKF, in particular, demonstrates a faster rate of heat release during the induction period compared to ALK. During the acceleration period, both types of accelerators exhibit a lower heat release rate than the reference group. After 24 h, the accumulated hydration heat is increased with both ALK and AKF, with AKF showing a more pronounced growth in accumulated heat.

Figure 6a illustrates the impact of AKF on the hydration heat release rate of cement paste. The addition of AKF accelerators notably increases both the heat evolution rate and the accumulated hydration heat during the pre-induction period, with the effect intensifying as the dosage of AKF increases. This indicates that AKF significantly enhances the initial hydration reactions. Throughout the induction period, the hydration heat release rate for the AKF mixtures remains markedly higher than that of the reference group. However, once the hydration enters the acceleration period, the heat release rate of the AKF mixtures decreases and becomes lower than that of the reference group, reflecting a reduced exothermic activity during this stage.

Figure 6b illustrates the impact of AKF on the hydration heat of cement. The addition of AKF accelerators significantly increases the heat evolution rate of the cement paste during the first 4 h, indicating a rapid initial reaction. From 4 to 24 h, the heat evolution rate for AKF is lower than that of the reference group, suggesting a slower rate of heat release during this period. However, after 24 h, the growth rate of heat evolution for the AKF mixtures significantly surpasses that of the reference group, with minimal differences observed between various AKF dosages.

3.4. Effect of Accelerator on Characteristic Hydration Products

Figure 7 shows the thermal analysis curves of different mixing ratios after 1 day. It can be observed in Figure 7 that DTG shows three typical endothermic peaks in the temperature ranges of 80–100, 130, and 440–450 °C. According to previous studies, these three temperature ranges correspond to the decomposition peaks of Aft, AFm, and Ca(OH)₂, respectively [32,33]. After 1 day of hydration, the 4% ALK mixture exhibits both AFt and AFm endothermic peaks, whereas neither the reference group nor the 8% AKF mixture shows an AFm peak. Notably, the AFt moisture mass loss in the 4% ALK mixture is significantly lower than that in the reference group and 4% AKF. This suggests that ALK accelerates the conversion of AFt to AFm, reflecting a more advanced stage of hydration product transformation [34]. The relevant cement hydration reaction equations are outlined in Equations (1)–(7).

Reference group cement hydration reaction equations:

$${\mathrm{C}}_3\mathrm{S}/{\mathrm{C}}_2\mathrm{S}+\mathrm{H}\to \mathrm{C}\mathrm{S}\mathrm{H}+\mathrm{C}\mathrm{H}$$

(1)

$${\mathrm{C}}_3\mathrm{A}/{\mathrm{C}}_4\mathrm{A}\mathrm{F}+\mathrm{C}\stackrel{-}{\mathrm{S}}+\mathrm{H}\to \mathrm{A}\mathrm{F}\mathrm{t}(\stackrel{-}{\mathrm{S}}/\mathrm{A}\ge 3)$$

(2)

$$\mathrm{A}\mathrm{F}\mathrm{t}+{\mathrm{C}}_3\mathrm{A}\to \mathrm{A}\mathrm{F}\mathrm{m}(1<\stackrel{-}{\mathrm{S}}/\mathrm{A}<3)$$

(3)

After adding ALK:

$$\mathrm{N}\mathrm{a}\mathrm{A}\mathrm{l}{\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}\to \mathrm{N}{\mathrm{a}}^{+}+\mathrm{A}\mathrm{l}{(\mathrm{O}\mathrm{H})}_4^{-}$$

(4)

$$\mathrm{A}\mathrm{l}{(\mathrm{O}\mathrm{H})}_4^{-}+\mathrm{C}{\mathrm{a}}^{2+}+\mathrm{O}{\mathrm{H}}^{-}+\mathrm{S}{\mathrm{O}}_4^{2-}+{\mathrm{H}}_2\mathrm{O}\to \mathrm{A}\mathrm{F}\mathrm{t}/\mathrm{A}\mathrm{F}\mathrm{m}(\mathrm{d}\mathrm{e}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{s}\mathrm{o}\mathrm{n}\stackrel{-}{\mathrm{S}}/\mathrm{A})$$

(5)

$${\mathrm{C}}_3\mathrm{A}+\mathrm{C}\mathrm{H}+12\mathrm{H}\to {\mathrm{C}}_4\mathrm{A}{\mathrm{H}}_{13}$$

(6)

$${\mathrm{C}}_4\mathrm{A}{\mathrm{H}}_{13}+\mathrm{C}{\mathrm{a}}^{2+}+\mathrm{O}{\mathrm{H}}^{-}+\mathrm{S}{\mathrm{O}}_4^{2-}+{\mathrm{H}}_2\mathrm{O}\to \mathrm{A}\mathrm{F}\mathrm{t}/\mathrm{A}\mathrm{F}\mathrm{m}(\mathrm{d}\mathrm{e}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{s}\mathrm{o}\mathrm{n}\stackrel{-}{\mathrm{S}}/\mathrm{A})$$

(7)

It is evident that the water loss of AFt for the 8% AKF mixture is significantly higher compared to the reference group and the 4% ALK mixture. This suggests that the formation of AFt is notably greater with the 8% AKF than with the 4% ALK and the reference group. This enhanced AFt formation is attributed to the aluminum sulfate (AS) in the AKF, which actively participates in the hydration reactions to produce more Aft. The reaction is shown in Equation (8):

$$\mathrm{A}{\mathrm{l}}_2{(\mathrm{S}{\mathrm{O}}_4)}_3+\mathrm{C}\mathrm{a}{(\mathrm{O}\mathrm{H})}_2+{\mathrm{H}}_2\mathrm{O}\to \mathrm{A}\mathrm{F}\mathrm{t}+\mathrm{A}\mathrm{l}{(\mathrm{O}\mathrm{H})}_3$$

(8)

The chemical compositions of AFt, AFm, and Ca(OH)₂ are consistent. Moreover, they have corresponding characteristic temperature decomposition ranges. So, the higher the percentage of water loss, the larger the amount of characteristic hydration products, such as AFm, AFt, and Ca (OH)₂, is generated.

Table 5. Water mass loss of Aft, AFm, and Ca(OH)₂ with different curing ages.

Curing Age (Days)	Water Mass Loss of AFt (%)			Water Mass Loss of AFm (%)			Water Mass Loss of Ca(OH)2 (%)
	REF	ALK-4%	AKF-8%	REF	ALK-4%	AKF-8%	REF	ALK-4%	AKF-8%
1	4.06	4.55	7.82	—	2.51	—	3.74	4.15	2.54
3	4.24	3.80	7.59	—	3.18	—	4.56	4.88	3.53
7	4.36	4.41	9.19	1.05	3.85	—	5.24	5.11	3.74
28	4.55	4.45	8.89	2.29	4.57	1.64	5.84	5.09	4.32

illustrates the variation in the mass loss of AFt, AFm, and Ca(OH)₂ with age for REF, ALK-4%, and AKF-8%. As shown in Table 5, the ALK-4% mixture generates AFm early and continues to increase its AFm content over time. In contrast, the REF mixture produces AFm only after 7 days, while the AKF-8% mixture generates AFm later, at 28 days, and produces a lower overall quantity of AFm. Additionally, both the ALK and AKF mixtures promote early cement hydration, leading to the formation of characteristic hydration products and a reduction in Ca(OH)₂ content. With increasing age, ALK-4% exhibits the formation of AFm products from the first day of hydration, with the amount of AFm continuing to rise over time. In contrast, REF shows the formation of AFm products only after 7 days, while AKF-8% does not produce AFm until 28 days and in much smaller quantities compared to REF and ALK-4%. This suggests that AFm is the early characteristic hydration product for ALK-4%, while the AKF group primarily forms AFt. The inclusion of AKF facilitates the formation of AFt but delays the formation of AFm. Between 1 and 3 days of hydration, the amount of Ca(OH)₂ produced with ALK-4% is only slightly higher than that of the reference group, whereas the Ca(OH)₂ content with AKF-8% is considerably lower. This indicates that both ALK and AKF accelerate cement hydration and the formation of characteristic hydration products early in the hydration process. However, they do not significantly increase the Ca(OH)₂ content, implying that both ALK and AKF directly react with Ca(OH)₂ to form AFm and AFt, respectively, thereby reducing the Ca(OH)₂ content in the paste.

4. Influence Mechanism of the Accelerator on Shotcrete Microstructure

4.1. Pore Structure Character Parameter

As two key parameters of pore structure, porosity and the most probable pore diameter have a significant effect on the matrix properties [35,36].

Figure 8 illustrates the impact of ALK and AKF on the porosity and pore structure of the hardened cement paste. Alkaline accelerator has a more significant effect on the pore structure of shotcrete, affecting the early and long-term porosity changes. With the increase in curing age, the total porosity and the most probable pore size of the reference group, 4% ALK, and 8% AKF decreased. Among them, at 1 day, the porosity and most probable pore size of 4% ALK and 8% AKF were significantly lower than those of the reference group. However, after 28 days, the porosity and most probable pore size of the reference group were lower than those of the ALK and AKF groups. Specifically, the porosity and most probable pore size of the 4% ALK group were significantly higher than those of the reference group and the 8% AKF group.

As shown in Figure 8, the total porosity and most probable pore size of the reference group and the samples containing 4% ALK and 8% AKF gradually decreased over time. At 1 day, the porosity and most probable pore size of the 4% ALK and 8% AKF samples were significantly lower than those of the reference group; this was attributed to the addition of ALK and AKF, which accelerated the early hydration, promoted the formation of more hydration products, and effectively filled the internal voids in the cement paste. However, after 28 days, the porosity and the most probable pore size of the reference group were lower than those of the ALK and AKF groups, indicating that the presence of the accelerator affected the refinement process of the pore structure during a longer curing period. In particular, the porosity and most probable pore size of the 4% ALK group were significantly higher than those of the reference group and the 8% AKF group, indicating that although ALK accelerates early hydration, it may lead to less fine pore structure in the long-run compared with AKF.

4.2. Pore Structure Fractal Dimension

The literature indicates that fractal dimensions based on thermodynamic relationships are more accurate compared to those derived from the Menger sponge model [37,38]. Consequently, this study calculates the fractal dimension of the pore structure for different mix proportions using a thermodynamic-based fractal model.

When mercury porosimetry is used to measure the relationship between pore volume and pore diameter of porous materials, the work done by the external environment on mercury is equal to the increased surface energy of mercury liquid entering the pores. The pressure P (Pa) applied to mercury and the mercury injection volume V (m³) are related by the following equation:

$${\displaystyle {\int }_0^{\mathrm{V}}\mathrm{P}\mathrm{d}\mathrm{V}=-}{\displaystyle {\int }_0^{\mathrm{S}}\mathsf{\sigma }\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{\theta }\mathrm{d}\mathrm{S}}$$

(9)

where $\mathsf{\sigma }$ is the surface tension of mercury (N/m), $\theta$ is the contact angle between mercury and the tested material (°), and S is the pore surface area of the tested material (m²).

Through dimensional analysis, the pore diameter (r) and V are related to the fractal scale of S. An expression for the fractal dimension Dm can be derived from dimensional analysis. Considering mercury injection, the equation can be approximated in a discrete form [39], as shown in Equation (10):

$${\displaystyle \sum _{i=1}^n\stackrel{—}{{\mathrm{P}}_i}}\mathsf{\Delta }{\mathrm{V}}_i=\mathrm{C}{\mathrm{r}}_n^{2-{\mathrm{D}}_{\mathrm{s}}}{\mathrm{V}}_n^{{\mathrm{D}}_{\mathrm{s}}/3}$$

(10)

where $\stackrel{—}{{\mathrm{P}}_i}$ is the mean pressure of the mercury injection operation at the ith cycle (Pa); $\mathsf{\Delta }{\mathrm{V}}_i$ is the mercury injection volume of the ith operation (m³); n is the number of intervals used to exert pressure during the mercury injection operation; ${\mathrm{r}}_n$ is the corresponding pore diameter of the nth mercury injection (m); V is the accumulative mercury injection volume of pressure intervals numbered 1-n (m³); and C is a constant.

Equation (10) can be organized as shown in Equation (11):

$${\displaystyle \sum _{i=1}^n\stackrel{—}{P_i}}\mathsf{\Delta }{\mathrm{V}}_i=\mathrm{C}{\mathrm{r}}_n^2{\left({\displaystyle \frac{{\mathrm{V}}_n^{1/3}}{{\mathrm{r}}_n}}\right)}^{D\mathrm{s}}$$

(11)

If ${\mathrm{W}}_n={\displaystyle \sum _{i=1}^n\stackrel{—}{{\mathrm{P}}_i}}\mathsf{\Delta }{\mathrm{V}}_i$, $Q_n={\mathrm{r}}_n^{2-D_{\mathrm{s}}}{\mathrm{V}}_n^{D_{\mathrm{s}}/3}$, Equation (11) can be obtained as follows:

$$\mathrm{l}\mathrm{g}\left({\displaystyle \frac{{\mathrm{W}}_n}{{\mathrm{r}}_n^2}}\right)=D_{\mathrm{s}}\lg {\mathrm{Q}}_n^{\prime }+{\mathrm{I}}_n\mathrm{C}$$

(12)

Using Equation (12), the fractal dimension ${\mathrm{W}}_n/{\mathrm{r}}_n^2$,${\mathrm{Q}}_n^{\prime }$ was calculated from the MIP data, where the logarithms of ${\mathrm{W}}_n/{\mathrm{r}}_n^2$ were used for regression analysis, as depicted in Figure 9. The slope of these regression lines provides the fractal dimension D_s. The correlation coefficients for the linear fits of $\mathrm{l}\mathrm{g}({\mathrm{W}}_n/{\mathrm{r}}_n^2)$ vs. $\lg {\mathrm{Q}}_n^{\prime }$ curves as functions were all greater than 0.999, indicating that the pore surface area for different mix proportions exhibits clear fractal characteristics. This confirms that the fractal model based on thermodynamics effectively describes the pore structure.

Pores with larger fractal dimensions exhibit more complex spatial geometrical characteristics and a greater capacity for pore filling. Based on the fractal model of thermodynamic relationships, the fractal dimensions of different mix proportions at various ages are calculated and presented in

Table 6. Fractal dimensions with different curing ages.

Age (d)	Fractal Dimension
	REF	ALK-4 (%)	AKF-8 (%)
1	2.3970	2.5729	2.4425
3	2.5294	2.7084	2.5870
7	2.6449	2.7297	2.6626
28	2.8945	2.8298	2.8965

. The fractal dimensions of different mix proportions increase with age. At early hydration stages, the fractal dimensions of 4% ALK and 8% AKF are higher than those of the reference group. This indicates that the addition of ALK and AKF significantly enhances the early hydration rate of cement, leading to the generation of more hydration products that fill the internal spaces of the paste.

However, at later stages of hydration (3–28 days), the growth rate of the fractal dimension for 4% ALK and 8% AKF is lower than that of the reference group. After 28 days, the fractal dimension of the reference group is higher than that of 4% ALK and 8% AKF. This can be attributed to the fact that, with increased curing time, the hydration rate of 4% ALK and 8% AKF slows down. Consequently, the newly generated hydration products are less effective in filling the hardened cement paste, leading to a slower increase in gel pore content compared to the reference group. This aligns with the observation that a higher gel pore content corresponds to a larger fractal dimension [40].

In summary, while ALK and AKF increased the early hydration rate of cement and improved early compressive strength, the addition of ALK delayed cement hydration, especially C₃S, hindered the C-S-H gel densification process, slowed down the pore structure refinement degree of hardened cement paste, and made the later strength lower than the reference group. In contrast, AKF was more conducive and favorable to the development of flexural and compressive strength.

4.3. Matrix Microstructure

Figure 10 illustrates the microstructure of cement paste samples after 1 day of hydration. The AFt crystals in the reference group are needle-like and randomly distributed. In contrast, a significant number of plate-like AFm crystals form in the matrix doped with 4% ALK, indicating that ALK accelerates the hydration of C₃A and promotes the conversion of AFt to AFm. However, the samples containing 8% AKF exhibit shorter, denser clusters of rod-like AFt crystals. This suggests that AKF affects the crystallization process, resulting in a more compact and clustered arrangement of AFt crystals.

As depicted in Figure 10a, the reference group features needle-like AFt crystals randomly distributed and overlapping on the C-S-H gel matrix. This typical arrangement is indicative of standard hydration processes. In contrast, the matrix with the 4% ALK admixture, shown in Figure 10b, reveals the formation of numerous plate-like hydration products. Energy-dispersive spectroscopy confirms these products as AFm, consistent with the thermal analysis results [41]. The significant presence of AFm indicates that ALK accelerates the hydration rate of C₃A and facilitates the conversion of AFt to AFm. Figure 10c displays the microstructure of the sample with 8% AKF after 1 day of hydration. Here, a large number of rod-shaped AFt crystals are observed. These crystals are notably shorter and smaller compared to those in the reference group, and they appear densely packed, growing in clusters.

Figure 11 displays the microstructure of hardened cement paste after 28 days of curing for various mixing ratios. As shown in Figure 11a, the reference group exhibits a dense matrix with numerous well-formed, continuous clusters of C-S-H gel. Energy-dispersive spectroscopy analysis indicates an average Ca/Si ratio of 1.140, reflecting balanced and stable C-S-H structures. Figure 11b,c illustrate the microstructures of samples with 4% ALK and 8% AKF after 28 days of curing, respectively. For the ALK sample (Figure 11b), the C-S-H gel displays a honeycomb-like, relatively loose structure. The energy-dispersive spectrum shows a higher average Ca/Si ratio of 1.674 compared to the reference group. This elevated ratio suggests that ALK accelerates the consumption of gypsum by its main component, NaAlO₂. This rapid consumption leads to a sulfate-deficient environment, which, in turn, stimulates the formation of AFm phases. The presence of AFm occupies the dissolution sites of C₃S [42], hindering its subsequent hydration and leading to a less dense C-S-H gel structure [43,44]. Additionally, the hydrolysis product NaOH from ALK contributes to a more porous and less dense matrix, adversely affecting the development of the microstructure and strength of the shotcrete.

In contrast, the matrix with 8% AKF (Figure 11c) exhibits a more fully developed and dense C-S-H gel structure. The energy spectrum analysis shows an average Ca/Si ratio of 1.171, indicating that AKF does not impede the growth and densification of the C-S-H gel. AKF continuously provides SO₄²⁻ ions to the solution while promoting the hydration of C₃A, thereby maintaining a stable Ca²⁺/SO₄²⁻ ratio. This contributes positively to the microstructural development and strength of the shotcrete [45,46], enhancing its overall performance.

5. Conclusions

This study evaluated the impact of ALK and AKF accelerators on several critical properties of cement paste, including the setting time, hydration heat, characteristic hydration products, pore structure, and matrix microstructure. The primary findings are summarized as follows:

• Both ALK and AKF accelerators significantly accelerate the early hydration of cement, leading to a faster setting time for shotcrete. AKF also accelerates setting time up to an 8% dosage, after which it stabilizes. AKF shows a faster increase in hydration heat release during the early induction period compared to ALK, enhancing early hydration more substantially.
• ALK and AKF influence the hydration of C₃A differently. ALK accelerates gypsum consumption, primarily producing hexagonal plate-like AFm phases. In contrast, AKF speeds up C₃A hydration and maintains a stable Al³⁺/SO₄²⁻ ratio, resulting in the formation of predominantly rod-shaped AFt phases.
• ALK and AKF significantly accelerate the early hydration of cement. Hence, the porosity and the most probable pore size of hardened cement paste at 1 d are significantly smaller than those of the reference group, which contributes to the rapid strength improvement of early hydration. However, ALK byproducts lead to a more porous and less refined pore structure. AKF, on the other hand, does not impede the hydration of C₃S or the densification of C-S-H gel, thereby fostering a more refined pore structure and better overall development of the shotcrete matrix.