Chloride Ion Transport Properties in Lightweight Ultra-High-Performance Concrete with Different Lightweight Aggregate Particle Sizes

In this paper, the microstructure and resistance to chloride ion penetration of ultra-high-performance concrete (UHPC) prepared from lightweight aggregate (LWA) were studied through simulation and experiment. The effects of LWA with different particle sizes on the chloride ion transport properties of lightweight ultra-high-performance concrete (L-UHPC) were discussed through simulation test results. The random delivery model of LWA in L-UHPC was established by MATLAB, and the model was introduced into COMSOL. Through the comparative analysis of experimental data and simulation results, the repeatability of the proposed model and the simulation accuracy were verified. The results show that when the LWA particle size changes from 0.15–4.75 mm to 0.15–1.18 mm, the width of interfacial transition zone (ITZ) and the overall porosity of L-UHPC decrease. This is because the large particle size LWA has more open pores with larger pore diameters and related interconnections, which are potential channels for chloride ion transport. Therefore, the chloride ion transport properties in L-UHPC are inhibited, which is manifested by the “tortuosity effect” of the LWA.


Introduction
Reinforcement corrosion is one of the main reasons for the reduction in durability of reinforced concrete (RC) structures in service environment, and chloride contamination is the main factor leading to reinforcement corrosion [1]. Concrete is a three-phase system consisting of solid-phase mortar, aggregates, and liquid-phase water and gas-phase air. Liquid-phase water and gas-phase air are present in pores and cracks that provide channels for chloride ion transport [2]. Thus, reinforcement corrosion in RC is highly correlated with the porous nature of concrete [3][4][5][6]. As one of the important components of concrete, aggregate is usually dispersed in the concrete non-uniformly to form the skeleton of concrete, so aggregate directly affects the properties of concrete [7][8][9][10]. Though the traditional aggregate itself is relatively dense and chloride ions cannot be transported through it, the interfacial transition zone (ITZ) formed around the aggregate provides a channel for the transport of chloride ion [11]. The effect of aggregate on chloride ion transport can be attributed to the following four effects, namely dilution effect, tortuosity effect, ITZ and percolation effect [12]. When the free chloride ion concentration on the surface of steel bars exceeds 3.0%, the internal structure of RC will be destroyed, reducing the service life of RC [13]. At present, the methods of studying chloride ion transport in concrete are mainly macroscopic The composition of L-UHPC cementitious material is cement: SF: FA = 1:0.26:0.21; the particle size distribution of LWA is calculated by the modified Andreasen-Andersen method. The LWA particle sizes used in the test were 0.15-4.75 mm, 0.15-2.36 mm, and 0.15-1.18 mm. The mass ratio in each interval of the three LWA particle sizes is shown in Table 2.  33 29 Based on the above-mentioned particle gradation composition of cementitious material and LWA, the benchmark mix proportion of L-UHPC was determined by adjusting the binder-aggregate ratio and the admixtures, as shown in Table 3.

Preparation and Curing of L-UHPC
Firstly, LWA was mixed according to particle gradation requirements and then put into mesh bags and immersed in water for pre-wetting. After the LWA was soaked for  24 h, the desired pre-wetted lightweight aggregate (PLWA) was obtained by paving it until the surface was dry. The cement, SF, and FA weighed according to the mix proportion were poured into the concrete mixer and mixed for 1 min. After the cementitious materials were evenly mixed, water and SP were added in and mixed for 2 min until a uniform flow was obtained. Then, the PLWA was added while stirring for 5 min, then molding and forming was performed. The surface of the forming mold was covered with plastic wrap and then placed in the standard curing box for one day to be taken out. After the mold was removed, the samples were maintained in the standard curing box until the specified age. The samples for resistance to chloride ion penetration were the standard samples specified in GB/T50082-2009 "Standards for Long-term Performance and Durability of Ordinary Concrete".

Basic Mechanical Properties of L-UHPC
The apparent density and working performance of L-UHPC were tested according to GB/T 50080-2016 "Standard for test method of performance on ordinary fresh concrete". The strength of L-UHPC was tested according to the national standard GB/T 31387-2015 "Reactive powder concrete". It can be observed from Table 4 that due to the apparent density of LWA being low, the apparent density of L-UHPC is less than 2000 kg/m 3 , and it decreases with the decrease in the LWA particle size. With the decrease in the LWA particle size, the slump and expansion of L-UHPC both increased first and then decreased. The compressive strength of the L-UHPC gradually increased with the decrease in the LWA particle size, and the flexural strength increased slightly. This is because the closest packing is used as the design principle of L-UHPC. The smaller the LWA particle size, the better the filling effect, and the overall porosity of L-UHPC is reduced. Therefore, the mechanical property of L-UHPC is improved, and the slump and expansion of L-UHPC are reduced. According to GB/T50082-2009 "Standards for Long-term Performance and Durability Test of Ordinary Concrete", the corrosion behavior was monitored by using rapid chloride migration method (RCM) and electric flux method. The specific test method is shown in Figure 1. Steel fibers were not incorporated in the test samples, and only the permeability of the cement matrix material was examined. The specific test method is shown in Table 5.
(2) Micromorphology ZEISS GeminiSEM 500 SEM (Oberkochen, Germany) was used to observe the microscopic morphology of the samples. The low vacuum degree was 1-210 Pa, the voltage was 0.5-32 kV, the magnification was 18-300,000 times, and the resolution was 4.0 nm.
(3) The width of ITZ EDS line scan is used to analyze the element distribution of LWA and cement paste to judge the width of the ITZ, as shown in Figure 1. Figure 1 shows the element distribution of LWA-ITZ-cement paste. The left side is the cement paste element distribution, the middle is the ITZ, and the right side is the LWA. It can be observed from Figure 2 that the Si concentrations were observed in the ITZ similar to that in the paste but was lower than in the LWA; the Ca element is mainly distributed in the paste; the Al element is mainly distributed in the LWA, less in the paste and the ITZ. Therefore, the ITZ width of LWA can be calculated by Si, Ca and Al elements [28,32,33].

(4) Microhardness
For this experiment, a DHV-1000 microhardness tester was used to test the microhardness value of L-UHPC. After preparing the 40 × 40 × 40 mm samples, after curing to the age, a precision cutting machine was used to cut the samples, and the middle width of the samples with a thickness of 1 cm was selected, which were put in an alcohol solution for immersion and saved, then taken out for drying before testing. During the test, the LWA was taken as the first test point, and then one point was taken every 20 µm for testing, and a total of 10 points were taken. This was repeated 10 times and the average value was taken as the data value for each test point.
For this experiment, according to the ASTM C1585-2013 standard, the capillary water absorption of L-UHPC was determined by the capillary water absorption method. The MesoMR23-60 H medium-sized MRI analyzer (Niumag, Suzhou, China) was used to test the pore structure of the L-UHPC. Samples of 40 × 40 × 40 mm were used for testing. When the sample reached the curing age, it was soaked in alcohol and dried before the test. Before the LF-NMR test, samples were treated for water retention.
Micromorphology ZEISS GeminiSEM 500 SEM (Oberkochen, Germany) was used to observe the microscopic morphology of the samples. The low vacuum degree was 1-210 Pa, the voltage was 0.5-32 kV, the magnification was 18-300,000 times, and the resolution was 4.0 nm.
(3) The width of ITZ EDS line scan is used to analyze the element distribution of LWA and cement paste to judge the width of the ITZ, as shown in Figure 1. Figure 1 shows the element distribution of LWA-ITZ-cement paste. The left side is  It can be observed from Figure 2 that the Si concentrations were observed ilar to that in the paste but was lower than in the LWA; the Ca element is uted in the paste; the Al element is mainly distributed in the LWA, less i the ITZ. Therefore, the ITZ width of LWA can be calculated by Si, Ca a [28,32,33].

(4) Microhardness
For this experiment, a DHV-1000 microhardness tester was used t hardness value of L-UHPC. After preparing the 40 × 40 × 40 mm sample the age, a precision cutting machine was used to cut the samples, and th of the samples with a thickness of 1 cm was selected, which were put in an for immersion and saved, then taken out for drying before testing. Dur LWA was taken as the first test point, and then one point was taken ever ing, and a total of 10 points were taken. This was repeated 10 times and th was taken as the data value for each test point.

Model and Parameters
In this paper, an L-UHPC global model and a two-dimensional three-item random aggregate model are established, wherein the three items are cement paste, LWA and ITZ, as shown in Figure 3. Figure 3a is a two-dimensional L-UHPC global model, and Figure 2b is a two-dimensional three-term L-UHPC random aggregate model. The volume fraction of the aggregate is 42%, and the shape of the aggregate is a polygon with 8-12 sides. Each LWA is evenly wrapped with a layer of ITZ, and the thickness of the ITZ is set to be 0.15 times the LWA particle size. The proportion of each particle size is based on the design in Section 2.2.1, and the positions are randomly generated by MATLAB.
In this paper, the chloride ion transport simulation of L-UHPC simulation is carried out by using the dilute material transport plate in the COMSOL. Based on the experimental data and related literature [34][35][36][37][38][39], the specific parameters are shown in Tables 6-8. In this paper, the chloride ion transport simulation of L-UHPC simulation is carried out by using the dilute material transport plate in the COMSOL. Based on the experimental data and related literature [34][35][36][37][38][39], the specific parameters are shown in Tables 6-8.        Figure 4 shows the effect of the LWA particle size on the chloride ion transport properties in L-UHPC. It can be observed from Figure 3 that with the increase in the LWA particle size, the chloride ion diffusion coefficient of L-UHPC gradually increases, indicating that the LWA particle size has a great impact on the chloride ion transport properties of L-UHPC. Compared with traditional UHPC whose chloride ion diffusion coefficient is 0.69 × 10 −12 m 2 /s and electric flux range is 400 C [40], when using 0. 15 the chloride ion diffusion coefficient and electric flux of L-UHPC decreased by 44.9% and 28.5%, respectively. Figure 5 shows the distribution of chloride ions in L-UHPC after chloride ion erosion for 30 d and 360 d in the simulated state. Figure 6 shows the effect of LWA particle size on the chloride ion transport of L-UHPC at 30 d, 90 d, 180 d, and 360 d erosion ages. From the analysis of Figures 4 and 5, it can be observed that the erosion depth of chloride ions gradually decreases with the reduction in the LWA particle size. This is due to the following reasons: (1) the reduced LWA particle size improves the internal compactness of L-UHPC, which reduces the total porosity and the chloride ion transmission channel of L-UHPC; (2) with the reduction in the LWA particle size, the internal curing effect of the LWA is more prominent, which optimizes the structure of ITZ around the LWA and hinders the transmission of chloride ions. When the erosion period is 30 d, the erosion depth of L-UHPC is only about 4 mm; after 360 d erosion, the erosion depth is more than 10 mm. This is consistent with the variation trend of transmission depth in the UHPC seawater splash zone in the study of Ma [41]. Additionally, this is attributed to the increase in erosion age; chloride ions react with hydrated calcium aluminate and Ca(OH) 2 in the ITZ, resulting in the increase in cracks and pores in the ITZ [42,43], which further accelerates the transport rate of chloride ions in L-UHPC. Moreover, it can be seen from the analysis of Figure 6 that with the increase in the erosion time, the transmission rate of chloride ions of the specimen with a larger LWA particle size is significantly faster than that of the specimen with a smaller LWA particle size. This is because the pore size and porosity of LWA is proportional to its particle size; when the pore size of LWA increases, the chloride ion transport channel increases. Therefore, with the increase in LWA particle size, the internal structure of ITZ is more deteriorated.   Figure 6 shows the effect of LWA p on the chloride ion transport of L-UHPC at 30 d, 90 d, 180 d, and 360 d erosion the analysis of Figures 4 and 5, it can be observed that the erosion depth of ch gradually decreases with the reduction in the LWA particle size. This is due to ing reasons: (1) the reduced LWA particle size improves the internal compa UHPC, which reduces the total porosity and the chloride ion transmission ch UHPC; (2) with the reduction in the LWA particle size, the internal curing e

ITZ
The morphologies of the ITZ of L-UHPC with particle sizes of 0.15-4.75 mm, 0.15-2.36 mm, and 0.15-1.18 mm for the LWA are shown in Figure 7. It can be observed from Figure 7 that the widths of the ITZ of the L-UHPC with particle sizes of 0.15-4.75 mm, 0.15-2.36 mm, and 0.15-1.18 mm for the LWA are 67 μm, 65 μm, and 40 μm, respectively. With the decrease in the LWA particle size, the LWA and the hardened cement paste are closely combined, and there is no obvious interface. This is because the microstructure of the ITZ between the cement paste and the LWA depends largely on the properties of the LWA. Compared with the porous ITZ around the normal weight aggregate, LWA has no wall effect, and the microstructure of the ITZ formed on its surface is comparable to that of the cement paste [44][45][46][47]. Additionally, the surface of the LWA is mostly open pores, which can tightly wrap the cement paste [48]. During the curing process of concrete, the free water of PLWA is slowly released, and the interface area of LWA-cement paste is fully hydrated. Therefore, the ITZ is improved, and the tortuosity of the chloride ion transport path is increased. The LWA surface is rough and has open pores. When the LWA particle size is large, there are more open pores and connected pores on the surface of LWA, the water release rate increases, and the impact range is large, so the width of ITZ is large. The opposite is true when the LWA particle size is small.
It can be observed from Figures 7 and 8 that the LWA particle size is proportional to the ITZ width and inversely proportional to the microhardness value. It can be attributed to the openings on the surface of the LWA being large; the pre-wetted water will be released in advance, resulting in the deterioration of the ITZ. In the later curing, the water release rate is faster, and a water film is formed around the LWA, which causes the deterioration of the ITZ. This is not good for L-UHPC to resist the invasion of chloride ions [49,50]. When chloride ions migrate to the ITZ, they react with hydrated calcium aluminate and Ca(OH)2 in the ITZ, which further leads to the increase in microcracks in the ITZ, providing migration channels for chloride ions. Therefore, the LWA particle size has an influence on the resistance to chloride ion penetration of L-UHPC.

ITZ
The morphologies of the ITZ of L-UHPC with particle sizes of 0.15-4.75 mm, 0.15-2.36 mm, and 0.15-1.18 mm for the LWA are shown in Figure 7. It can be observed from Figure 7 that the widths of the ITZ of the L-UHPC with particle sizes of 0.15-4.75 mm, 0.15-2.36 mm, and 0.15-1.18 mm for the LWA are 67 µm, 65 µm, and 40 µm, respectively. With the decrease in the LWA particle size, the LWA and the hardened cement paste are closely combined, and there is no obvious interface. This is because the microstructure of the ITZ between the cement paste and the LWA depends largely on the properties of the LWA. Compared with the porous ITZ around the normal weight aggregate, LWA has no wall effect, and the microstructure of the ITZ formed on its surface is comparable to that of the cement paste [44][45][46][47]. Additionally, the surface of the LWA is mostly open pores, which can tightly wrap the cement paste [48]. During the curing process of concrete, the free water of PLWA is slowly released, and the interface area of LWA-cement paste is fully hydrated. Therefore, the ITZ is improved, and the tortuosity of the chloride ion transport path is increased. The LWA surface is rough and has open pores. When the LWA particle size is large, there are more open pores and connected pores on the surface of LWA, the water release rate increases, and the impact range is large, so the width of ITZ is large. The opposite is true when the LWA particle size is small.
It can be observed from Figures 7 and 8 that the LWA particle size is proportional to the ITZ width and inversely proportional to the microhardness value. It can be attributed to the openings on the surface of the LWA being large; the pre-wetted water will be released in advance, resulting in the deterioration of the ITZ. In the later curing, the water release rate is faster, and a water film is formed around the LWA, which causes the deterioration of the ITZ. This is not good for L-UHPC to resist the invasion of chloride ions [49,50]. When chloride ions migrate to the ITZ, they react with hydrated calcium aluminate and Ca(OH) 2 in the ITZ, which further leads to the increase in microcracks in the ITZ, providing migration channels for chloride ions. Therefore, the LWA particle size has an influence on the resistance to chloride ion penetration of L-UHPC.

Pore
PLWA provides additional water for the continuous hydration of cement in L-UHPC and the pozzolanic reaction of auxiliary cementitious materials, so ITZ is enhanced, which reduces the internal porosity and permeability of L-UHPC [51,52]. Meng [53] reported that the partial replacement of river sand with pre-wet fine sand in UHPC can effectively reduce the total porosity of UHPC. Pores are an important channel for chloride ion transport in L-UHPC. The pores in L-UHPC can be divided into two types: one is the pores of the LWA itself, and the other is the pores in the hardened cement paste. Generally speaking, the pore size range in cement paste is smaller than that of LWA [54]. It can be observed from Figure 9 that reducing the LWA particle size not only reduces the content of small pores and macropores but also accelerates the accumulation rate of small pores, effectively refines the internal pore structure of L-UHPC, and reduces the total porosity

Pore
PLWA provides additional water for the continuous hydration of cement in L and the pozzolanic reaction of auxiliary cementitious materials, so ITZ is enhanced reduces the internal porosity and permeability of L-UHPC [51,52]. Meng [53] r that the partial replacement of river sand with pre-wet fine sand in UHPC can eff reduce the total porosity of UHPC. Pores are an important channel for chlor transport in L-UHPC. The pores in L-UHPC can be divided into two types: one is th of the LWA itself, and the other is the pores in the hardened cement paste. Ge

Pore
PLWA provides additional water for the continuous hydration of cement in L-UHPC and the pozzolanic reaction of auxiliary cementitious materials, so ITZ is enhanced, which reduces the internal porosity and permeability of L-UHPC [51,52]. Meng [53] reported that the partial replacement of river sand with pre-wet fine sand in UHPC can effectively reduce the total porosity of UHPC. Pores are an important channel for chloride ion transport in L-UHPC. The pores in L-UHPC can be divided into two types: one is the pores of the LWA itself, and the other is the pores in the hardened cement paste. Generally speaking, the pore size range in cement paste is smaller than that of LWA [54]. It can be observed from Figure 9 that reducing the LWA particle size not only reduces the content of small pores and macropores but also accelerates the accumulation rate of small pores, effectively refines the internal pore structure of L-UHPC, and reduces the total porosity of concrete. Moreover, it can be observed from Figure 10 that reducing the LWA particle size reduces the primary and secondary adsorption coefficient of L-UHPC and effectively improves the capillary structure of L-UHPC. Liu [55] reported that under the same LWA content, the incorporation of LWA with a smaller particle size into ultra-high-strength concrete (UHSC) could increase the volume of protective slurry, thereby reducing the permeability of UHSC.
Materials 2022, 15, x FOR PEER REVIEW 12 of 18 of concrete. Moreover, it can be observed from Figure 10 that reducing the LWA particle size reduces the primary and secondary adsorption coefficient of L-UHPC and effectively improves the capillary structure of L-UHPC. Liu [55] reported that under the same LWA content, the incorporation of LWA with a smaller particle size into ultra-high-strength concrete (UHSC could increase the volume of protective slurry, thereby reducing the permeability of UHSC. (a) (b) (c) Figure 9. Effect of LWA particle size on the pore structure of L-UHPC. (a) Effect of LWA particle size on the T2 spectra of L-UHPC; (b) effect of LWA particle size on the pore size distribution of L-UHPC; (c) effect of LWA particle size on the cumulative pore size distribution of L-UHPC. (c) Figure 9. Effect of LWA particle size on the pore structure of L-UHPC. (a) Effect of LWA particle size on the T2 spectra of L-UHPC; (b) effect of LWA particle size on the pore size distribution of L-UHPC; (c) effect of LWA particle size on the cumulative pore size distribution of L-UHPC.
(a) (b) Figure 10. Effect of LWA with different particle sizes on water adsorption of L-UHPC; (a) initial adsorption; (b) secondary adsorption.
LWA have more pores, which can be divided into open pores and closed pores. Closed pores are relatively isolated and not connected to each other, which can effectively block the transmission path of chloride ions; however, open pores are usually larger in diameter and connected to each other, which may become potential channels for chloride ion transmission [15]. It can be observed from Figure 11 that the larger the LWA particle size, the more connected pores on the surface and inside the LWA, which provides a potential channel for chloride ion transport. In addition, the larger the LWA particle size, the more pores on the LWA surface. This leads to the fact that during the curing process of L-UHPC, the cementitious paste enters the interior of the LWA, and a hydration product C-S-H gel is formed at a special site, which can effectively improve the structure of the LWA and the ITZ. Additionally, it was also found that the larger the LWA particle size, the more C-S-H gel generated inside the LWA, and vice versa. When the LWA particle size is 0.15-1.18 mm, the pore structure inside the LWA is mostly isolated pores, which can block the transmission path of chloride ions. LWA have more pores, which can be divided into open pores and closed pores. Closed pores are relatively isolated and not connected to each other, which can effectively block the transmission path of chloride ions; however, open pores are usually larger in diameter and connected to each other, which may become potential channels for chloride ion transmission [15]. It can be observed from Figure 11 that the larger the LWA particle size, the more connected pores on the surface and inside the LWA, which provides a potential channel for chloride ion transport. In addition, the larger the LWA particle size, the more pores on the LWA surface. This leads to the fact that during the curing process of L-UHPC, the cementitious paste enters the interior of the LWA, and a hydration product C-S-H gel is formed at a special site, which can effectively improve the structure of the LWA and the ITZ. Additionally, it was also found that the larger the LWA particle size, the more C-S-H gel generated inside the LWA, and vice versa. When the LWA particle size is 0.15-1.18 mm, the pore structure inside the LWA is mostly isolated pores, which can block the transmission path of chloride ions. In conclusion, reducing the LWA particle size can refine the pore structure of L-UHPC, reduce the total porosity, and reduce the migration channel of chloride ions; in addition, as shown in Figure 12, for the pores of LWA itself, under the same pre-wetting condition, the water inside the LWA also becomes a diffusion medium for chloride ions, and it provides a certain potential channel for chloride ion transmission. However, for LWA with a small particle size, the pores are mostly isolated closed pores, which further hinders the transmission of chloride ions to a certain extent. In conclusion, reducing the LWA particle size can refine the pore structure of L-UHPC, reduce the total porosity, and reduce the migration channel of chloride ions; in addition, as shown in Figure 12, for the pores of LWA itself, under the same pre-wetting condition, the water inside the LWA also becomes a diffusion medium for chloride ions, and it provides a certain potential channel for chloride ion transmission. However, for LWA with a small particle size, the pores are mostly isolated closed pores, which further hinders the transmission of chloride ions to a certain extent.
In conclusion, reducing the LWA particle size can refine the pore structure UHPC, reduce the total porosity, and reduce the migration channel of chloride io addition, as shown in Figure 12, for the pores of LWA itself, under the same pre-we condition, the water inside the LWA also becomes a diffusion medium for chloride and it provides a certain potential channel for chloride ion transmission. Howeve LWA with a small particle size, the pores are mostly isolated closed pores, which fu hinders the transmission of chloride ions to a certain extent.

Influence Mechanism
The migration of chloride ions mainly takes water as the medium, enters the in through the surface defects of the concrete, and then utilizes the internal pores and m cracks in the concrete as the transmission channel for internal migration. For the p

Influence Mechanism
The migration of chloride ions mainly takes water as the medium, enters the interior through the surface defects of the concrete, and then utilizes the internal pores and microcracks in the concrete as the transmission channel for internal migration. For the porous and "fragile" ITZ, the transmission rate of chloride ions in it is about 6-12 times higher than that in cement paste [36,37]. Therefore, the ITZ and pore structure characteristics are the key factors affecting chloride ion transport. Table 9 shows the relevant parameters of L-UHPC with different particle sizes measured by each test method. It can be observed from Table 8 that as the maximum LWA particle size increased from 1.18 mm to 4.75 mm, the diffusion coefficient of L-UHPC increased by 21%, the width of the ITZ increased by 67.5%, the porosity increased by 62.8%, and the adsorption coefficient increased by 19.3%. From the analysis of Table 9, it can be observed that there were more openings on the surface with the increase in LWA particle size, which leads to the early loss of pre-wet moisture during the mixing process and deteriorates the L-UHPC. Not only does this increase the total porosity, the increase in capillary pores results in a coherent network of pores [55]. In addition, the increase in the LWA particle size also has a negative impact on the ITZ. The larger particle size of LWA releases water faster to form a water film on the surface, which increases the width of the ITZ and increases the cracks and pores in the ITZ [56]. Therefore, when chloride ions enter the interior of L-UHPC, the pores in the matrix serve as the transport channels, and the pores and cracks in the ITZ further expand the transport channels and accelerate the transport rate of chloride ions in L-UHPC. Reducing the LWA particle size can reduce the number and area of pores on the surface of the LWA, thus effectively reducing the negative impact of the mixing process and curing process. As a result, the density of the L-UHPC is improved, the width of the ITZ is narrower, and the strength is higher. Therefore, when chloride ions are transported inside L-UHPC, there are fewer pores in the matrix and fewer chloride ion transport channels, and the ITZ is no longer a weak region. Not only that, when the LWA particle size is small, the interior is made up of mostly closed and independent solitary pores, which can further hinder the transmission of chloride ions in L-UHPC. This is consistent with the variation law of compressive strength.

Meso-Scale
The simulation results of the L-UHPC random aggregate model prepared by LWA with different particle sizes are shown in Figure 13. Due to D ITZ being larger than D cp , the volume fraction of ITZ increases with the increase in LWA particle size, which provides a channel for the rapid transmission of chloride ions. Furthermore, when the volume fraction of LWA is the same, the number of LWA gradually increases with the decrease in LWA particle size, which increases the tortuosity inside the L-UHPC. In the same region, when chloride ions enter the interior of the L-UHPC through surface pores or microcracks, the increased tortuosity of the LWA lengthens the transport path of the L-UHPC. As the transmission distance increases, the chloride ion concentration per unit volume is continuously diluted, and the chloride ion transmission rate continue to decrease [54]. The LWA particle size is reduced, which can effectively hinder the transmission of chloride ions.

Conclusions
In this paper, the effect of the light-aggregate particle size on the microstructure and resistance to chloride ion penetration of L-UHPC was found through indoor immersion and simulation tests. The microstructure of L-UHPC is directly related to its resistance to chloride ion penetration. COMSOL was used as an efficient simulation method to study the long-term resistance to chloride ion permeation of L-UHPC. From the experimental results, the following conclusions can be drawn:

Conclusions
In this paper, the effect of the light-aggregate particle size on the microstructure and resistance to chloride ion penetration of L-UHPC was found through indoor immersion and simulation tests. The microstructure of L-UHPC is directly related to its resistance to chloride ion penetration. COMSOL was used as an efficient simulation method to study the long-term resistance to chloride ion permeation of L-UHPC. From the experimental results, the following conclusions can be drawn:

1.
With the reduction in the particle size range of LWA, the working performance of L-UHPC shows a trend of first improving and then weakening. When the particle size of the LWA is in the range of 0.15-2.36 mm, L-UHPC has the best working performance, and its expansion can reach 550 mm. 2.
The LWA particle size determines the ITZ structure of L-UHPC. When the maximum particle size of LWA decreased from 4.75 mm to 1.18 mm, the width of ITZ gradually decreased, and the microhardness gradually increased. The denser ITZ played a positive role in hindering the transport of chloride ions. 3. LWA with a smaller particle size can refine the L-UHPC pore structure more effectively.
As the LWA particle size decreased from 0.15-4.75 mm to 0.15-1.18 mm, the number of capillary pores inside L-UHPC decreased, the number of self-isolated pores increased, and the total porosity decreased from 12.96% to 7.96%. The reduction in pores reduces the chloride ion transport channels, thereby improving the resistance of L-UHPC to chloride ion penetration.

4.
The decrease in LWA particle size increases the internal tortuosity of L-UHPC. Since the small particle size of LWA prolongs the transmission path of chloride ions, the chloride ions are continuously diluted when they are transported inward, resulting in a gradual decrease in the migration rate of chloride ions in L-UHPC. Therefore, with the decrease in LWA particle size, the resistance to chloride ion penetration of L-UHPC gradually increased. When the particle size of LWA is 0.15-1.18 mm, the chloride ion diffusion coefficient of L-UHPC is only 0.38 × 10 −12 m 2 /s.