Frost resistance improvement of recycled powder concrete by chemical admixtures

Abstract

The sustainable recycling of construction waste is a critical challenge in civil engineering. Recycled powder (RP) shows great potential as a supplementary cementitious material, but the deterioration in frost resistance it causes limits the application of recycled powder concrete (RPC) in cold regions. This study screened the polycarboxylate superplasticizer (PC) with optimal compatibility with RP through adsorption tests and systematically compared the improvement effects of antifreeze water-reducer (AR), air-entraining agent (AE), and antifreeze agent (AF) on frost resistance. Macroscopic properties were evaluated via freeze–thaw cycle tests, and microscopic mechanisms were analyzed using scanning electron microscopy (SEM), super depth of field (SDF) imaging, and mercury intrusion porosimetry (MIP). Results indicated that after 200 freeze–thaw cycles, the frost resistance improvement ranked as 1% AF > AR > 5% AE. Quantitative analysis revealed that AF optimized the pore structure, reducing the proportion of harmful pores > 200 nm by 8.73%, significantly increasing hydration products, and effectively inhibiting frost heave damage. The study confirms that AF effectively enhances the durability of RPC, providing technical support for the resource utilization of construction waste in concrete engineering in cold regions.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-35840-8.

Introduction

The construction industry globally faces the pressing challenge of managing enormous volumes of waste, with annual concrete and masonry debris exceeding 2 billion tons in China alone^1–3. While European nations achieve construction waste recycling rates of approximately 70%⁴, China’s rate remains below 10%⁵, highlighting a significant gap in sustainable resource utilization. A promising pathway involves processing this waste into recycled powder (RP) for use as a supplementary cementitious material^6,7, capitalizing on its filler effect and potential pozzolanic activity^8–11. Existing research confirms the feasibility of incorporating RP^12–14, demonstrating its ability to refine microstructure¹⁵ and influence hydration kinetics¹⁶. However, a critical synthesis of the literature reveals a predominant focus on mechanical properties and hydration characteristics, often overlooking a key durability aspect essential for cold-region applications: frost resistance.

Although RP can be used as a concrete admixture, the inclusion of RP also raises the water requirement for the concrete, which adversely affects the frost resistance of the concrete. By choosing a suitable polycarboxylate superplasticizer (PC), this impact can be significantly alleviated. The molecular structure of PC has a comb-like characteristic, then it adsorbs onto the surface of cement particles, reduces the energy at the solid–liquid interface, and blocks the agglomeration of particles to achieve the dispersion of cement^17–21. There have been relatively few studies on the adsorption of PC by RP, and these have mainly been carried out in the laboratory. Frost resistance, as the core of the durability of concrete, is directly proportional to its ability to withstand cold weather for a long time^22,23. Chai W et al. showed that water-reducing agents reduce the internal pores and lower the water content of concrete, thereby enhancing durability²⁴. Zhang H et al. demonstrated that air-entraining agent (AE) alleviates the structural damage to concrete caused by freeze–thaw cycles (FTCs) through the introduction of uniformly distributed small air bubbles²⁵. Of course, the addition of RP also has a certain impact on the frost resistance of cement-based materials. Hou S et al. showed that 30% recycled brick powder had a negative effect on frost resistance, but using it in combination with RP can improve the results²⁶. Furthermore, combining RP with slag and silica fume can substantially enhance the frost resistance of concrete²⁷. The existing literature is characterized by a primary emphasis on macroscopic performance. There is a notable lack of in-depth mechanistic investigations that link these macroscopic improvements to microstructural evolution in RPC under freeze–thaw conditions.

In summary, the use of a water-reducing agent is an efficient method to decrease water usage. In this paper, PC types with better compatibility with RP were selected through adsorption research and comparison. Based on this admixture system, the improvement effects of antifreeze water-reducing agent (AR), air-entraining agent (AE), and antifreeze agent (AF) on the antifreeze performance of RPC were systematically compared. The improvement effects of the three methods were compared and analyzed. Finally, the frost resistance of RPC was analyzed at the microscopic level using SEM, Super Depth of Field (SDF), and Mercury Intrusion Porosimetry (MIP). Recycled concrete prepared from RP not only brings economic and environmental benefits, achieving “harmlessness” and “resource utilization”, but also conforms to the current sustainable development strategy.

Materials and methods

Materials

Cement and recycled powder

The cement used in the test was PO 42.5 cement produced by Beijing Jinyu Company; the recycled powder (RP) from demolished brick-concrete structures was produced by Jianyang Wuxiong Building Materials Co., Ltd. The chemical composition of the cement and RP is shown in Table 1. The particle size distribution parameters of the cement and RP are shown in Table 2. The specific surface areas of the cement and RP as shown in Table 3.

Table 1.

Chemical composition of cement and recycled powder(%).

Powder	SiO₂	CaO	Al2O3	Fe2O3	SO3	MgO	Other
Cement	20.68	60.09	7.76	3.26	3.31	3.01	1.89
RP	51.89	15.93	15.07	9.31	1.15	0.99	5.56

Table 2.

Cement and recycled powder particle size distribution parameters.

Powder	Diameter (μm)	Particle size distribution (μm)			Consistency	Specific surface area (m2/kg)
		D10	D50	D90
Cement	3.243	3.52	40.1	133	1.038	372.8
RP	3.515	2.45	24.2	87.4	1.201	722.8

Table 3.

Determination of specific surface area of recycled powder by Blaine gas permeability and nitrogen adsorption(m2/kg).

Method	Powder	Specific surface area
Blaine gas permeability	Cement	375
	RP	745.6
Nitrogen adsorption	Cement	449.1
	RP	4316.2

The microstructure of the RP was observed using SEM in Fig. 1. As can be seen in Fig. 1a, b, the RP surface is rough, with large pores and an angular, flaky, or irregular shape. Therefore, when incorporated into concrete, it requires a higher water demand.

Fig. 1.

SEM images of recycled powder.

Admixture

1. Polycarboxylate superplasticizer (PC): The PC used in the test was synthesized by China Construction Western Building Materials Science Research Institute Co., Ltd. Taking the monomer type, the acid-ether ratio, and the polyether molecular weight as variables, nine polycarboxylate superplasticizers were designed. Its numbers and parameters are shown in Table 4.

2. Antifreeze water-reducing agent (AR): The HZ-2 antifreeze water-reducing agent, provided by Hebei Hezhong Building Materials Co., Ltd., is a compound water-reducing agent.

3. Air-entraining agent (AE): An alkylbenzene sulfonate air-entraining agent, provided by Dushi Lvyuan Environmental Technology Co., Ltd., is a translucent liquid

4. Antifreeze agent (AF): The HZ-D antifreeze agent, provided by Hebei Hezhong Building Materials Co., Ltd., is a compound inorganic salt antifreeze agent.

Table 4.

Parameters of polycarboxylate superplasticizer for test.

No	Monomer type	Acid-ether ratio	Polyether molecular weight	Solid content
V4-2	VPEG	4:1	2400	41%
V4-3	VPEG	4:1	3000	41%
V4-4	VPEG	4:1	4000	41%
V4-5	VPEG	4:1	5000	41%
V8-2	VPEG	8:1	2400	41%
V8-3	VPEG	8:1	3000	41%
V8-4	VPEG	8:1	4000	41%
V8-5	VPEG	8:1	5000	41%
H4-2	HPEG	4:1	2400	50%

Aggregate

The aggregates utilized in the experiment were natural gravel and natural river sand produced by Beijing Yugong Co., Ltd. Their basic performance indicators and chemical compositions are shown in Table 5.

Table 5.

Main performance indicators of aggregate.

Type	Apparent density (g·cm−3)	Bulk density (g·cm−3)	Moisture content(%)	Water absorption rate(%)	Crushing index (%)
Coarse aggregate	2825	1610	0.08	0.45	9
Fine aggregate	2618	1765	0.3	0.7	–

Water

Laboratory tap water.

Methods

Test method for properties of recycled powder

The methods for testing the properties of RP are in accordance with Chinese standard JG/T 573–2020²⁸.

Sample preparation

The RPC prepared in this experiment was manufactured and cured according to the requirements of the Chinese standard GB/T 50081–2019²⁹.

Slump

Control the slump to 220–240mm. The slump test method is in accordance with Chinese standard GB/T 50080–2016³⁰.

Compressive strength

The YA-3000 produced by Sansi Zongheng Machinery Manufacturing Co., Ltd. was used in accordance with Chinese standard GB/T 50081–2019²⁹.

Freeze–thaw cycle (FTC)

The FTC test was carried out in accordance with the Chinese standard GB/T 50082–2024³¹.

Microstructure analysis

The XRD sample is a block sample. The mineral composition of the sample is determined using a Rigaku Ultima IV from Japan. The SEM sample is a block sample. After the sample is gold spraying treatment, it is examined using a ZEISS Gemini 300 scanning microscope. XRF uses a Rigaku Ultima IV to determine the chemical composition of the sample; the laser particle size analysis sample is a powder sample, and the particle size is analyzed using a Malvern 3000 Malvern laser particle size analyzer; the nitrogen adsorption method employs a Micromeritics ASAP2420 analyzer to characterize the pore structure model and evaluate the internal pore structure of the powder sample.

Mix proportion design

The mix ratio of RPC is shown in Table 6.

Table 6.

Mix ratio of recycled powder concrete.

Admixture type	No	W/B	Material amount(kg·m−3)					Sand ratio (%)	PC (%)	AR (%)
			Cement	RP	Fine aggregate	Coarse aggregate	Water
Control group	A30	0.56	242	104	1006	823	194	45	0.28	–
	B30	0.41	306	131	1000	750	181	43	0.57	–
AR	A30-AR	0.56	242	104	1006	823	194	45	–	3.93
	B30-AR	0.41	306	131	1000	750	181	43	–	5.19
PC + AE	A30-AE5	0.56	242	104	1006	823	194	45	0.26	–
	B30-AE4	0.41	306	131	1000	750	181	43	0.57	–
	B30-AE5	0.41	306	131	1000	750	181	43	0.57	–
	B30-AE6	0.41	306	131	1000	750	181	43	0.57	–
PC + AF	B30-AF1	0.41	306	131	1000	750	181	43	0.57	–
	B30-AF2	0.41	306	131	1000	750	181	43	0.57	–
	B30-AF3	0.41	306	131	1000	750	181	43	0.57	–

W/B 0.56 and 0.41 were selected, and the RP dosage was 30%. The slump was maintained within 220–240 mm by adjusting the dosages of PC and AR. The frost resistance was improved by three improvement methods: AR, AE, and AF. The control dosage of the AE group was set to make the RPC gas content 4%, 5% and 6%, respectively. The dosage of AF was 1%, 2%, and 3% respectively.

Results and discussion

Adsorption of polycarboxylate superplasticizer by recycled powder

The amount of water-reducing agent adsorbed by RP particles and their hydration products is calculated using the TOC adsorption test method. The acid-ether ratio of PC is different, the main chain has different macromonomers, and the polyether molecular weight is different. The adsorption amount and adsorption rate are different, as shown in Table 7.

Table 7.

Adsorption amount and adsorption rate of polycarboxylate superplasticizer under three situations.

Situation		No	Adsorption amount(g/L)	Adsorption rate(%)
Acid-ether ratio	4: 1	V4-2	1.68	8.20
	8: 1	V8-2	0.8	3.84
	4: 1	V4-3	7.64	29.81
	8: 1	V8-3	5.02	15.47
	4: 1	V4-4	2.20	14.63
	8: 1	V8-4	3.88	10.42
	4: 1	V4-5	2.02	12.38
	8: 1	V8-5	1.70	7.79
Main chain macromonomer	VPEG	V8-2	0.8	3.84
	HPEG	H4-2	1.30	8.48
Polyether molecular weight	2400	V4-2	1.68	8.20
	3000	V4-3	7.64	29.81
	4000	V4-4	2.20	14.63
	5000	V4-5	2.02	12.38
	2400	V8-2	0.8	3.84
	3000	V8-3	5.02	15.47
	4000	V8-4	3.88	10.42
	5000	V8-5	1.70	7.79

Table 7 shows that the RP and PC have the best compatibility when the acid-ether ratio is 4:1, the polyether molecular weight is 3000, and the main chain macromonomer is VPEG. When the acid-ether ratio and the type of main chain macromonomer are different, the adsorption rate shows a large difference. The adsorption of RP to the main chain macromonomer HPEG is stronger, which is more than twice the adsorption rate of VPEG, but both are less than 10%. As the polyether molecular weight of PC increases, the adsorption rate first increases and then decreases.

Studies have shown that the higher the adsorption rate of RP to PC, the better the dispersing effect of PC. This may be because the molecular chain of PC interacts with the active or adsorption sites on the surface of RP through its hydrophilic groups to form an adsorption layer. The Ca²⁺ on the RP surface forms a complex with groups such as the carboxyl group in the PC molecule, further enhancing the stability of the adsorption layer. The side chains provide steric hindrance, which is conducive to dispersion and has a good water-reducing effect^32–34. Based on the above test results, the next test selected to add V4-3 PC.

Improvement of frost resistance of recycled powder concrete

Antifreeze water-reducing agent

Mechanical property The comparison of the compressive strength of the control group and the AR group is shown in Fig. 2.

Fig. 2.

Strength comparison of recycled powder concrete.

As shown in Fig. 2, after the addition of AR, the early compressive strength increases slightly, but as the age increases, the compressive strength of the AR group decreases instead. This may be due to the fact that AR has a lower water-reducing rate and a higher water consumption than PC, resulting in an increase in the W/B. And the strong retarding effect of AR may prolong the setting time of RPC, resulting in insufficient hydration reaction, which affects the generation and distribution of hydration products in RPC, increasing the intercrystalline pores and making the network structure more loose, thus affecting the strength of pieces³⁵.

Frost resistance The apparent morphology of the control group and the AR group after FTCs is shown in Fig. 3. After the same number of FTCs, the frost resistance of the RPC had a marked increase after the addition of AR, and the surface of the test piece was more complete, with no obvious aggregate exposure or mortar layer peeling.

Fig. 3.

Apparent morphological comparison of improved frost resistance of recycled powder concrete.

The mass loss rate (MLR) and relative dynamic modulus of elasticity (RDM) of the control group and the AR group are shown in Figs. 4 and 5.

Fig. 4.

Mass loss rate.

Fig. 5.

Relative dynamic modulus of elasticity.

The frost resistance of the AR group showed substantial improvement, with a significant reduction in both the MLR and the loss of RDM. As shown in Fig. 4, the RPC with a W/B of 0.41 exhibited minimal mass loss, which subsequently increased over time. Throughout the testing, the AR group consistently demonstrated lower MLR compared to the control group. Figure 5 shows that the RDM of A30, A30-AR, B30, and B30-AR decreased slightly before 100 FTCs, and the decrease began to increase after 100 FTCs. However, the RDM of the AR group decreased relatively little.

After 100 FTCs, the surface porosity of group A30 decreased. RP filled the capillary pores, forming a dense surface layer with anti-spalling properties, avoiding bubble defects, optimizing ITZ, and inhibiting the generation of cracks. In the A30-AR group, the unhydrated particles in RP adsorbed AR, resulting in bubble coarsening, intensifying surface spalling, increasing mass loss, and the coarse bubbles merged and expanded under freeze–thaw pressure, making it more likely for cracks to occur along the ITZ. The RPC in group B30 is denser, reducing water migration. However, RPC is also more prone to brittle damage, and ITZ generates local stress concentration and microcracks. In the B30-AR group, due to the dense matrix, the microbubbles within it are difficult to participate in the failure, and the internal stress is reduced, thereby decreasing the generation of cracks³⁶.

The above results show that compared to PC, AR can significantly reduce the porosity and harmful pores of RPC, improve its compactness and overall strength, while reducing the formation of micro-cracks, improving the internal stress distribution of RPC, enhancing its toughness, and making the internal structure of RPC less damaged after FTCs³⁷.

Air-entraining agent

Mechanical property The compressive strength of RPC specimens with varying gas contents (4%, 5%, and 6%) is presented in Fig. 6 for W/B 0.41; the compressive strength comparison of RPC with a gas content of 5% is shown in Fig. 7.

Fig. 6.

Compressive strength of different air-entraining agents.

Fig. 7.

Comparison of compressive strength of different W/B.

Figure 6 shows that as the gas content of the RPC increases, the compressive strength of the RPC tends to decrease, and the decrease increases with the increase of the dosage. This is because with the increase of the AE gas content and the addition of AE, a substantial amount of small pores appear in the RPC, forming some stable microbubbles inside the RPC, which increases the porosity of the RPC³⁸.

Figure 7 shows that the laws of change in compressive strength of the two are basically the same, with a smaller early decrease and a larger late decrease. This is because there is more water at W/B 0.56, and the porosity increases. These pores become stress concentrations, making RPC more likely to break when subjected to pressure. As the age increases, the cement in the RPC gradually hydrates and forms hydration products, but the distribution of hydration products is uneven³⁹. Although AE can improve the performance of RPC by introducing tiny bubbles, poor or uneven bubble stability can also lead to uneven stress distribution inside the RPC, thereby reducing its compressive strength^40,41.

Frost resistance The apparent morphology of the control group and the AE group after FTCs is shown in Fig. 8.

Fig. 8.

Comparison of the apparent morphology of recycled powder concrete with different air-entraining agent content.

Figure 8 shows that after 200 FTCs, most of the mortar layer on the surface of the control group has peeled off, exposing the aggregate. The apparent form of the RPC is relatively complete when the air content is 4% or 5%, with slight peeling on the surface of the piece and the outer loosened, and holes appeared on the surface of B30-AE4 with a gas content of 4%; the surface of the RPC with a gas content of 6% fell off more seriously, similar to that of the control group, so the addition of AE can improve the frost resistance of RPC.

The MLR and RDM of the control group and the AE group are shown in Figs. 9 and 10.

Fig. 9.

Mass loss rate.

Fig. 10.

Relative dynamic modulus of elasticity.

As shown in Fig. 9, the mass of both the control and AE groups decreased after FTCs, with significant loss observed after 150 and 200 cycles. The incorporation of AE markedly reduced the MLR. After 150 FTCs, the MLR of A30 reached 10.71%, exceeding the 5% limit, whereas with AE addition, it was reduced to 3.43%. After 200 cycles, B30-AE4 and B30-AE5 exhibited lower MLRs than B30, with the most significant improvement at an air content of 5%. Figure 10 indicates that the RDM of the AE group decreased at a slower rate compared to the control. However, when the air content increased to 6%, the RDM decline became more pronounced, but its MLR exceeds 5%.

This is because AE blocks the moisture channels inside the RPC by introducing tiny air bubbles, reducing migration and ice expansion during FTCs and improving its frost resistance. After 100 FTCs, the increase in porosity is very small. The stable structure formed by the introduced bubbles inside the RPC can absorb and disperse the expansion pressure caused by the freezing of moisture, preventing damage to the internal structure of the RPC⁴². Overall, RPC with a 5% air content has the best frost resistance.

Antifreeze agent

Mechanical property The compressive strength of W/B 0.41 of RPC mixed with different amounts of AF (1%, 2%, 3%) is shown in Fig. 11.

Fig. 11.

Compressive strength of recycled powder concrete antifreeze agent with different dosages.

Figure 11 shows that the compressive strength of RPC at all ages is significantly improved after the addition of AF, and the improvement increases with the increase of AF dosage, especially the early strength improvement is obvious. The improvement of late strength is not as obvious as that of early strength. This is due to the early-strength component of AF accelerating the hydration process in RPC during the initial stages. It encourages the development of C–S–H gel, which efficiently fills micropores and decreases porosity, leading to a more tightly packed and denser RPC structure⁴³.

Frost resistance The RPC with W/B 0.41 was selected for the experiment to improve frost resistance (quick-freezing method).

The apparent morphology of the control group and the AF group after FTCs is shown in Fig. 12.

Fig. 12.

Comparison of apparent morphology of recycled powder concrete compound antifreeze agent.

Figure 12 shows that the frost resistance of RPC is significantly improved after the addition of AF. Compared with the control group, the surface mortar layer of the test piece peels off less, and the surface of the test piece is relatively complete, without peeling corners, exposed aggregate and other phenomena; however, the surface conditions of different AF dosages are not significantly different.

The MLR and RDM of the control group and the AF group are shown in Fig. 13 and 14.

Fig. 13.

Mass loss rate.

Fig. 14.

Relative dynamic modulus of elasticity.

Figure 13 shows that the frost resistance of RPC was enhanced following the addition of AF. Initially, during the early FTCs, the MLR of the AF group was higher than that of the control group. However, after 150 FTCs, the MLR of the AF group increased at a slower pace compared to the control group. Notably, when the AF content was 1%, the MLR remained lower throughout all FTCs. As shown in Fig. 14, after 100 FTCs, the ranking of the RDM is as follows: B30-AF2 > B30-AF1 > B30 > B30-AF3. Beyond 100 FTCs, the RDM of B30 declines rapidly.

During the initial phase of the FTCs, the influence of the AF group was minimal and, in some cases, even had a negative effect. However, as the number of cycles increased, the beneficial effects of AF became more evident. AF lowers the freezing point of water in RPC, allowing it to remain at least partially liquid and thereby reducing the stress damage caused by ice formation within the material. In addition, it can increase the surface area of the cement and improve the contact area between the cement and the aggregate⁴⁴. After 100 FTCs, the hydration reaction of RPC is complete, a large amount of free water is consumed, and a dense hydration product is formed. The framework structure formed by the interaction between the aggregate and the slurry can resist part of the expansion pressure caused by freezing⁴⁵. Overall, it can be seen that the frost resistance of RPC is better when 1% AF is added.

Comparison of frost resistance improvement

The RPC’s MLR and RDM of the three improvement methods are shown in Figs. 15 and 16.

Fig. 15.

Mass loss rate.

Fig. 16.

Relative dynamic modulus of elasticity.

Figure 15 shows that initially, the improved group experienced a higher mass loss compared to the control group. However, as the number of FTCs increased, the mass loss in the control group rose significantly. With the exception of B30-AE6, the other formulations effectively reduced the mass loss of RPC. In contrast, the AR group exhibited the lowest mass loss. As shown in Fig. 16, At the start of the FTCs, the RDM of all groups, except B30-AE6 and B30-AF3, was higher than that of the control group. After 200 FTCs, the RDM for all test groups exceeded that of the control group. In comparison, the AR group exhibited the smallest loss in RDM. In summary, for RPC, the AR and AF dosages of 1% and the air content of 5% have a better effect on improving the frost resistance of RPC. The AR dosage of 1% is better than the AR dosage of 1% and the air content of 5%.

Microstructure analysis

SEM analysis

The RPC cross-section samples of B30, B30-AR, B30-AE and B30-AF were prepared, and the interface transition zones and hydration products of each group were magnified and shown in different magnifications in Figs. 17 and 18, Figs. 19 and 20, Figs. 21 and 22, Figs. 23 and 24, respectively.

Fig. 17.

SEM images of B30 interface transition zone.

Fig. 18.

SEM images of B30 hydrated product.

Fig. 19.

SEM images of B30-AR interface transition zone.

Fig. 20.

SEM images of B30-AR hydrated product.

Fig. 21.

SEM images of B30-AE interface transition zone.

Fig. 22.

SEM images of B30-AE hydrated product.

Fig. 23.

SEM images of B30-AF interface transition zone.

Fig. 24.

SEM images of B30-AF hydrated product.

Figure 17 shows that the pore distribution of the interface is uneven, with many obvious large pores and cracks (Fig. 17a), and the distribution of hydration products near the interface transition zone (Fig. 17b) is uneven. The bond stress between the aggregate and the paste is poor, and there are wide cracks. Due to the existence of these cracks, the RPC is divided into several independent blocks. During the FTC, water can easily enter the interior of the RPC, and the cracks at the joints will expand into wider and deeper cracks, resulting in a decrease in the frost resistance of the RPC⁴⁶.

As shown in Fig. 18a, there was less hydration product formation inside the RPC, and there were more crystals as shown in Fig. 18c. The results of the energy spectrum showed that the crystal was probably 3CaO SiO₂ that had not undergone a hydration reaction. Therefore, it is speculated that the high dosage of RP resulted in incomplete cement hydration, which in turn led to a decrease in compressive strength and frost resistance.

As shown in Fig. 19, the AR group had fewer surface pores and cracks and a denser structure in the interfacial transition zone than the control group.

As shown in Fig. 20, the amount of hydration product formation increased, there were fewer macropores, and at high magnification, a large number of neatly arranged worm-like C-S-H gels and a small amount of AFt were visible. In addition, hexagonal sheet-like Ca(OH)₂ crystals were encapsulated by C-S-H gels to form a uniform, continuous, and densely bonded body⁴⁷. Therefore, after the addition of AR, the frost resistance of RPC was improved.

As shown in Fig. 21a, after the addition of AE, a large number of uniform and enclosed tiny bubbles are introduced inside the RPC, which can cut off the pore connection channels, inhibit the formation of harmful macropores, and increase the surface area of the entire porous system of the RPC.

As shown in Fig. 22, although there are a few cracks, the hydration products are relatively dense, with fibrous C-S-H gels with high crystallinity and neatly arranged worm-like gels, completely encapsulating Ca(OH)₂ crystals and filling the internal pores, which is very beneficial for improving the frost resistance of RPC⁴⁸.

As shown in Fig. 23, after the addition of AF, the harmful macropores on the surface of the RPC matrix are significantly reduced, the matrix structure is more compact, the aggregate and the paste are tightly bonded, and the macroscopic performance is improved in terms of frost resistance and compressive strength.

As shown in Fig. 24, the number of hydration products increased significantly, and there were more needle-like Aft crystals (Fig. 24b). Due to the bonding between Aft crystals and the mechanical interlocking between radial Aft crystal clusters⁴⁹, their formation can promote the early strength of cement and improve the early strength of RPC. Therefore, after the addition of AF, the early strength of RPC has been significantly improved. At high magnification (Fig. 24c), the C-S-H gel aggregates together, and the pores in the middle are filled with fine Aft crystals⁵⁰. The hydration is complete, the interface is dense, and therefore its compressive strength and frost resistance have been significantly improved.

To sum up, the B30 group has more harmful macropores and cracks on its surface. The weak interface transition zone and the loose connection between the mortar and the aggregate lead to wide cracks. The amount of hydration product formation is low, so the macroscopic performance is a decrease in compressive strength and frost resistance. The B30 group was mixed with AR, AE, or AF, respectively, the connection between the slurry and the aggregate was tight, the number of cracks and uneven pores was significantly reduced, and the evenly distributed small pores introduced after mixing with AE could be clearly seen. The number of hydration products increased and filled the pores in the matrix.

Image analysis of super depth of field (SDF)

The cross-section of the RPC specimen was magnified 100 times, and the SDF of RPC was processed and analyzed based on professional image processing software Image Pro Plus (IPP). The internal pore conditions of the RPC for different methods of improving frost resistance were analyzed, and the red area in the figure indicates the pores. Specimens of B30, B30-AR, B30-AE, and B30-AF were prepared, and the pore distribution of RPC was observed at 100 times. The pore distribution and data of RPC are shown in Fig. 25.

Fig. 25.

Pore distribution and data of recycled powder concrete.

As shown in Fig. 25, the total number of pores in the B30 sample is 667,172, and the total number of pores in B30-AR and B30-AF is 567,585 and 421,306, respectively, a decrease of 14.9% and 36.9%, respectively. After improvement by different methods, the RPC pore structure can be refined and the number of tiny pores can be reduced. It can be seen from the pore distribution that the red area in the B30 group is large, with a large number of pores. After the addition of AR or the addition of AF, the red area is significantly reduced; although the addition of AE has a large number of pores, it can be clearly seen that the bubbles inside are of uniform size. It is confirmed at the microscopic level that the addition of admixtures can significantly improve the frost resistance of RPC.

Mercury intrusion porosimetry (MIP)

Among the three improvement methods, the effect of AF is the best, so RPC with W/B 0.41 was selected. Samples were prepared from the control group and the AF group, and MIP was performed to analyze the pore structure. The pore structure parameters of RPC are shown in Table 8. The proportion of multi-harmful pores decreased, while the proportion of less harmful pores increased.

Table 8.

Pore structure parameters of recycled powder concrete.

No		B30	B30-AF
Total porosity(%)		15.30	17.98
Total pore volume(ml/g)		0.09	0.13
Maximum pore size(nm)		50.36	40.28
Average pore size(nm)		50.21	38.10
Pore size distribution (%)	< 20nm	13.07	12.28
	20–50nm	17.38	18.92
	50–200nm	7.69	15.67
	> 200nm	61.86	53.13

The pore size distribution of the B30 group is compared and analyzed with that of the B30-AF group as shown in Fig. 26 and Fig. 27, and the volume distribution of cumulative pores is shown in Fig. 28.

Fig. 26.

Differential curve of B30 pore size distribution.

Fig. 27.

Differential curve of B30-AF pore size distribution.

Fig. 28.

Volume distribution of cumulative pores.

As shown in Fig. 26 and Fig. 27, the total porosity of B30 is 15.30%, and the total porosity of B30-AF is 17.98%. Although the total porosity increases after the addition of AF, the porosity is high, and the frost resistance is not necessarily poor. This is because the frost resistance is related to the pore distribution inside the RPC⁵¹. It can be seen that the average pore diameter and the most probable pore diameter of B30-AF are reduced, and its pore structure may be more uniform and smaller. As shown in Fig. 28, the critical pore size of B30-AF is significantly reduced, indicating that the number of large pores inside the RPC has decreased, which helps to reduce the damage to the RPC structure caused by the expansion of water ice at low temperatures. The internal pore structure has been optimized, so its frost resistance is better.

Conclusions

In this paper, a PC with better compatibility with RP was selected as the base admixture. On this basis, the frost resistance was improved by three improvement methods: adding AR, AE, and AF. Finally, the frost resistance of RPC was analyzed from a microscopic mechanism perspective. The detailed conclusions are summarized as follows:

1. A systematic comparison reveals that AF at a 1% dosage is the most effective strategy, outperforming the AR and AE. The improvement is quantitatively demonstrated by the superior retention of the relative dynamic modulus of elasticity and the lowest mass loss after 200 freeze–thaw cycles.

2. The superior performance of AF is mechanistically explained by a significant refinement of the pore structure. MIP analysis provides the critical quantitative evidence, showing that AF incorporation reduces the proportion of harmful pores larger than 200 nm by 8.73%, while increasing the volume of harmless and less harmful pores. This microstructural densification is the primary mechanism for suppressing frost heave damage.

3. The research confirms the feasibility of producing durable, frost-resistant concrete using construction waste solid powder. This provides a viable and sustainable technological pathway for the resource utilization of construction waste in cold regions, contributing directly to the goals of a circular economy in the construction industry.

4. This study focused on evaluating individual admixtures at their practical optimum dosages. A key direction for future work is to employ a controlled factorial experimental design to systematically investigate the interaction effects between AR, AE, and AF, then to determine the performance ceiling of higher AF dosages.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Can Yang: Writing—original draft, Visualization, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation. Wenjuan Zhou: Writing—review & editing, Supervision, Resources, Project administration, Funding acquisition, Formal analysis, Conceptualization. Handi Zhao: Visualization, Software, Investigation. Mingli Zhou: Visualization, Software, Investigation.

Funding

This research was supported by the Key Research and Development Project of the 14th Five-year Plan of China (2022YFC3803403-02).

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.