Effect of Freezing-Thawing Cycle on the Mechanical Properties and Micromechanism of Red Mud-Calcium-Based Composite Cemented Soil

+e environmental issues caused by solid waste have become increasingly serious. Adding additive is considered as an effective measure to improve the performance of the cemented soil. +erefore, the feasibility study of solid waste such as red mud and desulfurization gypsum used in composite cemented soil is in urgent demand. In this study, the mechanical properties and durability to freezing-thawing cycle of red mud-calcium-based composite cemented soil (RMCC) were analyzed through compressive strength test, resistivity test, and freezing-thawing cycle test. +e action mechanism of RMCCwas revealed through a series of X-ray diffraction (XRD) and scanning electron microscope (SEM) with energy dispersive spectrometer (EDS) test. +e results show that the optimal red mud content in RMCC is 12%. As the freezing-thawing cycle progresses, the difference in resistivity and pressure sensitivity of RMCC gradually weakens. When the freezing-thawing cycle reaches 7, the pressure sensitivity characteristic of RMCC is lost. +e change in resistivity and pressure sensitivity can be used to characterize the damage caused by the freezing-thawing cycle. Combined with XRD and SEM analysis, the presence of minerals such as K2Ca5(SO4)6·H2O and (Ca, Na)2(Si, Al)5O10·3H2O play a key role in fixation of alkali metal elements, and the coordination of CSH gel cementation effect and AFt filling effect has a significant impact on mechanical properties.+e study provides an effective way to the utilization of red mud and desulfurization gypsum in subgrade strengthening.


Introduction
As an effective measure for soil stabilization, cementedbased stabilization has been widely used in engineering such as base material for pavement, foundation stabilization, and subgrade strengthening [1]. However, cemented soil has disadvantages in strength and durability [2] and improving the performance of cemented soil is essential for its application. In cold regions, when temperature drops below 0°C in winter, the pore water freezes and causes volume expansion, which seriously endangers the safety of public transport [3]. As a byproduct of industrial production, solid waste has huge reserves and improper disposal of solid waste could cause critical environmental hazards. e utilization of solid waste in composite cemented soil and other civil engineering has great potential [4][5][6][7].
As a metallic element widely existing in nature, aluminum occupies a significant position in the process of industrialization. Red mud is the byproduct of extracting alumina from bauxite ore. e production of 1 t alumina will also produce approximately 1 t to 1.5 t of red mud [8]. Today, the main methods of red mud disposal are slurry disposal, dry stacking, and marine disposal [9]. ermal power generation still occupies a considerable share in China, and desulfurization gypsum is the byproduct produced in the process of flue gas desulfurization in coal-fired power plants [10]. e utilization of solid waste can not only consume a large amount of solid waste but also reduce the environmental burden.
Over the years, many attempts have been reported to explore the utilization of red mud as building materials [11], treatment of contaminated soils [12], and raw materials for the extraction of the metal elements [13]. In addition, there are many reports for the application of red mud in the construction for road traffic. Alam et al. [14] used alkaliactivated ground granulated blast furnace slag (GGBS) to stabilize red mud as the geotechnical material and tested its durability in the drying-wetting cycles. e leaching toxicity tests showed that the stabilized red mud was nontoxic. Zhang et al. [15] used red mud, blast furnace slag, cement clinker, and gypsum to develop new cement-based material. Mukiza et al. [16] added red mud and fly ash to the subgrade material. e research and development of the application of desulfurization gypsum is mainly concentrated on building materials and other fields [17,18].
For the red mud-cement stabilized soil, many research studies also have been reported. Mi et al. [19] studied the permeability of cemented soil mixed with red mud and pointed out that the additive of red mud greatly reduces the permeability coefficient. Tian et al. [20] investigated the effect of simulated acid rain on red mud-cement stabilized soil. Results showed that the permeability coefficient decreased with the increase of sulfuric acid concentration. Chen et al. [21] added a small amount of cement and red mud to strengthen the loess roadbed filling and analyzed the mechanical properties, dynamic properties, and microstructural properties of the stabilized soil. Salih et al [22] applied red mud, fly ash, and desulfurization gypsum to subgrade materials and studied the mechanical properties and durability to drying-wetting cycles.
ese provide a reference for the application of red mud in civil engineering.
Many researchers have explored the reductions of the mechanical properties of the stabilized soil after freezingthawing cycles and its mechanism. Jafari and Esna-Ashari [23] added waste tire to modified clay with lime and studied the mechanical properties of modified soil under freezingthawing cycle. Liu et al. [24] studied the dynamic properties of cement and lime modified clay soils used on high-speed railways in northern China after freezing-thawing cycle. Compared to unimproved soil, the critical dynamic stress of lime-modified soil and cement-modified soil increased by 60% and 200%, respectively. Aldaood et al. [25] added gypsum to lime stabilized soil and investigated the mechanical properties and microscopic mechanism during freezing-thawing cycle. e results showed that strength decreased after 5 freezing-thawing cycles. Liu et al. [26] used red mud and coal gangue as raw materials for road base and carried out freezing-thawing cycle tests.
As an inherent characteristic of rocks and soils, resistivity can be used to characterize the variation in internal structure. Zha et al. [27] studied the resistivity characteristics of contaminated soil solidified by cement and fly ash. e results showed that unconfined compressive strength and resistivity maintain linear relationships on a semilog scale. Mosavinejad et al. [28] used different kinds of materials as electrodes and tested the resistivity of composite cemented soil. e results showed that brass was more suitable as an electrode material than pure copper and steel. Dong et al. [29][30][31] did a lot of research studies on the relationship between unconfined compressive strength and resistivity of cemented soil under complex environments. Under the freezing-thawing cycle condition, the internal structure of the materials changes and the resistivity also changes accordingly. Wang et al. [32] used silica fume, fly ash, groundgranulated blast furnace slag, and steel slag as additives to ordinary Portland cement and test the resistivity under freezing-thawing cycles. Fu and Wang [33] tested the resistivity of silt clay from the Qinghai-Tibet Plateau under freezing-thawing cycles and indicated that the resistivity variation corresponded to soil deformation.
In spite of the mentioned studies, there are few studies on the mechanism of interaction between CSH gel and AFt in red mud-calcium-based composite materials and fixation of alkali metal elements in red mud-based materials. Furthermore, the pressure sensitivity of stabilized soil under freezing-thawing cycles has not been investigated yet. e objective of this study is to evaluate the feasibility of red mud and desulfurization gypsum in subgrade strengthening under freezing-thawing cycle condition and to reveal the action mechanism of CSH gel and AFt in RMCC from a microscopic perspective. Moreover, the mechanism of alkali metal elements fixation of RMCC is also studied. Firstly, the mechanical properties were assessed by compressive strength tests and a serious of freezing-thawing cycle tests were carried out. ereafter, the resistivity and pressure sensitivity of RMCC was studied. Finally, the microstructural properties were revealed by X-ray diffraction (XRD) and scanning electron microscope (SEM and EDS).
is paper provides a theoretical support for the utilization of red mud and desulfurization gypsum in subgrade strengthening.

Materials.
e red mud was excavated from Xiaoyi, Shanxi Province in China. It belongs to the Bayer method red mud. Figure 1 shows the SEM image and XRD pattern of red mud. e magnification in Figure 1(a) is 10,000 which indicates that the particle size is small and an aggregate microstructure composed of massive fine particles can be observed. e red mud contains CaCO 3 , Al 2 O 3 , Al(OH) 3 , and Calcium Silicate Hydrate. e specific surface area of red mud is 1217 m 2 /kg. e desulfurization gypsum (FGD) was excavated from Taiyuan Power Plant in Shanxi Province. Figure 2 shows the SEM image and XRD pattern of desulfurization gypsum. e desulfurization gypsum particles are regular in prismatic shape and the particle size is large. XRD Figure 3 shows the particle size distribution of red mud, desulfurization gypsum, and loess. Table 1 shows the basic physical properties of loess. e chemical composition of raw materials is listed in Table 2. Note that the alkali metal oxides are the source of alkalinity of red mud.

Sample Preparation.
Firstly, the loess, red mud, and desulfurization gypsum were crushed and dried. Loess and red mud were sieved through a 0.075 mesh-sized sieve. For RMCC mixture, the cement accounted for 10% of loess dry mass and the red mud accounted for 50% of desulfurization gypsum dry mass, respectively (Table 3). Since the moisture content has a great influence on freezing-thawing cycle tests, the moisture content was set at 29%. Secondly, according to the China standard JGJ/T70-2009 [35], the predetermined amounts of red mud, cement, desulfurization gypsum, and water were mixed to obtain a homogeneous mixture through concrete mixing device. Moreover, for cement-treated loess      Advances in Civil Engineering which will be the control group, the cement accounted for 10% and 15% of loess dry mass, respectively. irdly, the mixture was placed into a cube metal mold (70.7 mm inside length) to obtain the RMCC samples through concrete vibrating table. Finally, the metal molds were covered with plastic film and stripped after 24 hours. e samples were maintained in a standard curing room with constant temperature and humidity for 7, 14, and 28 days, respectively. e curing temperature was 19.5-21.5°C and the curing humidity was 95%.
Some specimens were obtained from samples after the corresponding curing period and freezing-thawing cycles for microstructure measurement. ese specimens were submerged in absolute ethanol to stop the hydration process and then placed in air to dry. Some of the dried specimens were polished to the dimensions of 1 cm × 1 cm × 0.5 cm for the SEM and EDS tests, and the other dried samples were ground into a powder for the XRD tests.

Compressive Test and Resistivity Test.
Compressive strength tests were conducted through a microcomputer controlled-electronic universal testing machine (WAW-2000) and the samples were pressed with vertical deformation rate of 1 mm/min. Resistivity tests were conducted through LCR digital bridge (TH2828A). First step in the resistivity tests were to smooth the upper and lower surfaces of the sample. en, graphite was applied to both sides of the sample to improve conductivity. Select a current frequency of 50 kHz as the analysis object [30]. e data obtained by the digital bridge was the impedance |Z|. e resistivity is calculated according to formula where ρ is the resistivity (Ω·m); |Z| is the impedance (Ω); S is the cross-sectional area of the simple (m 2 ); and L is the distance between the electrodes (m). For the samples that have completed the freezingthawing cycle tests, before the resistivity tests, the surface of the samples should be dry. For the stress-strain-resistivity Table 1: Basic physical properties of loess.
Maximum dry density ρ max (g·cm −3 ) Optimum moisture content ω opt (%) Plastic limit ω L (%) Liquid limit ω p (%) Plasticity index (I p )    tests, copper foils were placed on the upper and lower surfaces of the samples as electrodes in the process of compressive strength test. During the loading process, load value, position value, and impedance were recorded simultaneously. en, the corresponding data was calculated to obtain the stress-strain-resistivity data.

2.3.2.
Freezing-awing Cycle Test. In this paper, the freezing-thawing cycle tests were conducted by SRTDR-16 rapid concrete freezing-thawing testing machine. e lowest temperature was set at −15°C and the highest temperature was set at 5°C. e freezing-thawing cycle tests are following the China national standard GBT50082-200 [36]. In the first step, 3 samples were placed in a rubber cylinder. A total of 12 samples were tested for one freezing-thawing cycle. In the second step, the rubber cylinders were placed in the test equipment. e third step was to add water in the rubber cylinder and inserted the temperature sensor. After the equipment was started, the temperature was reduced to −15°C, and then the equipment began to heat up. When the temperature reaches 5°C, the test was finished for one freezing-thawing cycle. en, the samples were removed from the rubber cylinder and left at room temperature. After the corresponding freezing-thawing tests, the samples were placed for approximately 7-8 hours in room temperature before the compressive strength test. When freezing-thawing cycle reached 7 times, the tests finished [25,26].

Microscopic Test.
To further explore the micromechanism of RMCC, a series of microscopic test was performed. e XRD analysis was conducted by Ultima IV X-ray diffractometer. e powdery specimens were scanned ranging from 10°to 90°with a speed of 20°/min. e SEM and EDS analysis were conducted by using a JSM-IT200 scanning electron microscope. Figure 4 depicts the relationship between the compressive strength q u value and the curing time of RMCC. It exhibits a similar trend of q u with the different red mud contents (C R ). When a curing period is 14 days, the q u , respectively, reaches 2.32, 2.27, 2.64, 2.72, and 2.74 MPa with a red mud content of 3%, 6%, 9%, 12%, and 15%, respectively. It accounts for 81%, 74%, 82%, 83%, and 89% of q u after 28 days-curing. From the perspective of strength, in stage one, the first 14 days, the strength of RMCC has a faster growth rate; in stage two, from day 14 to day 28 of the curing period, the strength of RMCC has a slow growth rate. is result may be attributed to that the hydration reaction rate is relatively faster in the first stage of the curing period. Moreover, the gelling hydrates such as CSH gel generated from hydration reaction can bond red mud particles, desulfurization gypsum particles, and loess particles to form a stable and compact soil structure in the first stage [21]. However, the rate of hydration reaction in the second curing period is slowing down and leading to a slower growth in compressive strength. Figure 5 shows the effect of red mud content on the compressive strength q u value under different curing time periods. With the increase of the red mud content, the q u of RMCC shows an increasing trend. When the red mud content exceeds 12%, the q u shows a decreasing trend. e optimum red mud content of RMCC is 12%. When the red mud content increases from 3% to 6% after 7-day-curing and 14-day-curing, the q u shows a slight decreasing trend.

Strength Performance under Standard Curing.
is result may attribute to that when the red mud content is 6%, the content of desulfurized gypsum is relatively low and the AFt generated from pozzolanic reaction has a lower capacity to fill the pores so that there are more pores interior. However, when the curing time is 28 days, the hydration process is almost complete, and the structure is stable. Figure 6 shows the relationship between resistivity and red mud content under different curing time periods. With the increase of red mud content, resistivity first increases and then decreases. When the red mud content is 12%, the resistivity reaches its maximum value. e alkalinity of the RMCC system is low with the low red mud content, resulting in the loss formation of CSH gel and AFt.

Resistivity Performance under Standard Curing.
erefore, there are more pores in the interior which result in low resistivity. With an increase of the red mud content, the hydration reaction and pozzolanic reaction are facilitated, the internal structure tends to be stable, resulting in the reduction of pores and increasing of compressive strength and resistivity [29,37]. When the red mud content is 12%, the compressive strength and resistivity 5 1 0 1 5 Curing time (d) Advances in Civil Engineering reaches its maximum value, indicating stable and dense structure forms. Resistivity can be used to characterize the process of the reaction and the quantity of pores inside the samples.  [26], it is calculated according to the following formula:

Mechanical Properties of RMCC under
where BDR is the loss rate of compressive strength after the freezing-thawing cycle, f FT is the residual compressive strength after the freezing-thawing cycle, and f u is the initial compressive strength without the freezing-thawing cycle. Figure 7 shows the relationship between the loss rate of compressive strength (BDR) and red mud content under different freezing-thawing cycles (FTC). With the increase of the red mud content, the BDR is minimum when the red mud content is 12%. After 5th and 7th freezing-thawing cycles, the compressive strength suffered a large reduction and the two curves are relatively close to each other. It means that the sample has been severely damaged and the reduction of compressive strength value between 5th and 7th freezing-thawing cycle is low. As the control group, the samples of cement-treated loess are destroyed after 3 freezing-thawing cycles.
During the freezing-thawing cycles, the pore water freezes and expands 9% of its volume [38], damaging the structure of the specimen, and the peeling phenomenon will gradually become serious. Define the rate of mass loss to the following formula: where Δ n is the rate of mass loss after n freezing-thawing cycles. m n is the mass of the sample after n freezing-thawing cycles and m is the mass of the initial sample without the freezing-thawing cycle. Figure 8 shows the relationship between the mass loss rate and red mud content under different freezingthawing cycles. After 1 or 2 freezing-thawing cycles, the rate of mass loss of all specimens fluctuates around 1%. After 3 freezing-thawing cycles, the mass loss rate increases faster as red mud content is 3%. After 4-6 freezing-thawing cycles and if the red mud content is 3% and 15%, the mass loss rate is relatively large. At the same time, with the red mud content of 6%, 9%, and 12%, the value of mass loss shows a tendency to gradually increase, but the range of variation is within 1%. After 7 freezingthawing cycles, the mass loss rate shows a large difference.
is result may attribute to the exfoliation on surface of the samples, which means that the hydration products cannot cement the raw material particles, resulting in a large increase of the mass loss rate. Combined with the analysis of compressive strength, the largest reduction of mechanical properties indicates that the RMCC is losing the durability to freezing-thawing cycles. e variations at the rate of mass loss can characterize the durability of RMCC under freezing-thawing cycle condition [39]. By analyzing the effect of different freezingthawing cycles on the rate of mass loss, the 3rd freezingthawing cycle is an inflection point, and the 6th freezingthawing cycle is another inflection point. With the  Advances in Civil Engineering progression of the freezing-thawing cycles, the mass loss rate is lowest when the red mud content is 12%, indicating that the RMCC has the best durability to the freezing-thawing cycle at this content. Figure 9 shows the relationship between the resistivity and red mud content under freezing-thawing cycle condition. e curve of resistivity exists a remarkable peak value without the freezing-thawing cycle. After 1st or 3rd freezingthawing cycles, the peak value of resistivity weakens at the red mud content of 12%, but there still exists a peak value. After 5th or 7th freezing-thawing cycles, the peak value disappeared, and the variation of resistivity curves is similar to a straight line. Figure 10 shows the resistivity-strain 3D Ribbons with a red mud content of 12% under different freezingthawing cycles. It can be observed that the freezingthawing cycle has a large effect on 3D Ribbons and different Ribbons show large differences. After the 1st freezing-thawing cycle, the resistivity shows a rapidly decreasing trend with increasing of strain. After the 3rd and 5th freezing-thawing cycles, the resistivity decreases as the strain decreases, but the rate of decrease slows down. However, after the 7th freezing-thawing cycle, the resistivity-strain 3D Ribbon is close to flat, and the resistivity changes little with increasing strain, which shows that with the progression of freezing-thawing cycles, the pressure sensitivity of RMCC is gradually reducing.

Resistivity Performance under Freezing-awing Cycle.
Combined with the analysis of compressive strength, mass loss rate, and resistivity of specimens, RMCC has the lowest strength and the largest mass loss and very little variation in resistivity after 7 freezing-thawing cycles, indicating that the internal structure has been severely damaged. At this point, the ability of hydration products such as CSH gel cement the raw material particles is lost. e above analysis shows that, after 7 freezing-thawing cycles, the formation factor of the RMCC has little effect. At the same time, among the factors influencing pressure sensitivity characteristic of samples, the change in resistivity caused by pore water dominates, and the stress-strain variations cannot have a major effect on the resistivity. e variation in pressure sensitivity effectively reflects the damage in internal structure of RMCC under freezingthawing cycle condition [40].   Advances in Civil Engineering 7

X-Ray Diffraction Analysis.
e XRD analysis was conducted to explore the formation of compounds for RMCC. Figure 11 presents the XRD pattern of specimens with 7-day-curing time and 28-day-curing time. Moreover, a sample after a 56-day-curing was also tested in order to make the analysis more complete. By comparing the XRD patterns, the compounds are basically similar in the mineralogical level, but some peaks appear or disappear within a certain range of degrees. e XRD patterns of 7-day-sample and 28-day-sample show similarities, while the XRD pattern of 56-day-sample shows more significant differences. e comparison shows that the specimen of 56-day-sample has a disappearance peak near 10°.
is peak represents CaSO 4 ·2H 2 O, and the peaks which also represent CaSO 4 ·2H 2 O disappear in the range between 31.43°and 33.34°, indicating that CaSO 4 ·2H 2 O was consumed in pozzolanic reaction. e peak of Yugawaalite-Ca(Si 6 Al 2 ) O 16 ·4H 2 O can be observed at 18.86°in the pattern of 7-daysample, but the same peak cannot be observed in the pattern of 56-day-sample. e peak which represents CaSO 4 or Epistilbite-Ca 2 (Si 9 Al 3 )O 24 ·8H 2 O can be observed at 23.02°, but cannot be observed in pattern of 56-day-sample. ree types of minerals can be observed in the pattern of 7-daysample in the range between 27°and 28°. However, only one mineral can be observed in 56-day-sample in the same range, the peaks which represent Akdalaite-(Al 2 O 3 ) 4 ·H 2 O and Epistilbite-Ca 2 (Si 9 Al 3 )O 24 ·8H 2 O disappear. In the range between 34°and 38°, two peaks disappear in patterns of 28day-sample and 56-day-sample, which represent Prehnite-Ca 4 Al 2 O 7 ·19H 2 O and Calcium Aluminum Oxide Hydrate-Ca 4 Al 2 O 7 ·19H 2 O. e analysis of XRD reveals that, with increasing curing period, the types of reaction products in RMCC gradually decrease, and the components tend to be stable and CSH gel, Aft, and CaCO 3 can be observed in specimens of three different curing periods [15]. e formation of CaCO 3 is attributed to carbonation reaction [41]. e products of pozzolanic reaction, carbonation reaction, ion exchange reaction, and hydration reaction play a key role in the strength of RMCC.
Compared to the pattern of 7-day-sample, two minerals with different elemental compositions can be found in the pattern of 56-day-sample: Gorgeyite-K 2 Ca 5 (SO 4 ) 6 ·H 2 O and Rustumite-Ca 10 (Si 2 O 7 ) 2 SiO 4 C l2 (OH) 2 . e formation of two minerals can be explained by the differences between desulfurized gypsum and other gypsum such as analytical pure gypsum and construction gypsum, which is that desulfurization gypsum contains a small amount of ions such as Cl − , Mg 2+ , Na + , and K + [41,42]. In the hydration reaction, Ca 2+ reacts with Cl − and other anions, SO 2− 4 reacts with K + ions, and other cations, thus forming minerals such as Gorgeyite and Rustumite. is indicates that the addition of desulfurized gypsum can facilitate the formation of hydration products. Moreover, it can be observed that there are two peaks of minerals containing sodium and potassium elements in XRD patterns: Gorgeyite-K 2 Ca 5 (SO 4 ) 6 ·H 2 O and Gonnardit-(Ca, Na) 2 (Si, Al) 5 O 10 ·3H 2 O. e silicate hydrate formed by the hydration process in RMCC contains active Ca 2+ , which occurs as a replacement reaction with alkali metal elements in red mud to achieve the fixation of alkali metal elements, indicating that RMCC has alkaline binding capacity. Figure 12 shows the SEM images and corresponding EDS analysis of RMCC after 28day curing time with a red mud content of 12%. Figure 12 Compared with the edges and corners in Figures 12(a) and 12(b), it can be confirmed that Figure 12(b) shows the desulfurization gypsum particle. is indicates that the pores between desulfurization gypsum particles which have larger particle size are filled with fine particles, and all the raw material particles are cemented by the gelling products to form a stable and dense structure. In Figure 12(c), the microstructure is dense, and a large amount of gel hydrate can be observed. Figure 12(d) is the 20,000 magnified image of the circled area in Figure 12(c). In Figure 12(d), obvious plaque-shaped substances can be observed. Combined with the related research [43], it can be recognized as Ca(OH) 2 . At the same time, a regular hexagonal flake-shaped substance can be observed in the square area of Figure 12(d). Combined with the SEM analysis of the red mud material, it can be considered as unreacted red mud particles. e filling effect of fine red mud particles can enhance the mechanical properties of RMCC.

Microstructure Analysis.
Referring to the EDS analysis results (see Figures 12(f ) and 12(h)), it is evident that a large amount of hydration products can be observed such as CSH gel and AFt [44,45].

Advances in Civil Engineering
Moreover, the alkali metal element such as Na and K can be found in the analysis of elemental composition. Combined with the presence of Gorgeyite-K 2 Ca 5 (SO 4 ) 6 ·H 2 O and Gonnardit-(Ca, Na) 2 (Si, Al) 5 O 10 ·3H 2 O in XRD pattern, this is the further evidence that RMCC has ability of fixation to alkali metal elements. For the RMCC in freezing-thawing cycles with a red mud content of 12%, a series of images was captured by SEM with the magnification of 1000 because, under this magnification, the damage caused by the freeing-thawing cycles to the microstructure can be clearly and intuitively observed.
In Figures 13(a) and 13(c), it can be observed that there is a small amount of crack on the surface and the shape does not change. In Figure 13(b), there are less hydration products and more micropores compared with Figure 12(c), but the microstructure still maintains continuity after 1 freezing-thawing cycle. By comparing Figures 13(b) and 13(d), it can be seen that the microstructure became broken, and larger granular substances can be observed after 3 freezing-thawing cycles. Microstructure changed from relatively flat to uneven, but no obvious microcracks were observed in Figure 13(d). From a microscopic perspective, the soil microstructure still maintains relative continuity which results no significant detriment on the surface. However, the micropores has a greater impact on mechanical properties, so when the shape remains unchanged and mass loss is small, the compressive strength suffered a significant reduction.
Comparing Figure 13(c) with 13(e), the surface was severely damaged after 5 freezing-thawing cycles, and   obvious crack propagation can be observed. In Figure 13(f ), the microcracks shown in the elliptical frame cannot be observed in Figure 13(d). is indicates that the freezingthawing cycles result in higher degree of damage to microstructure. In Figure 13(g), the shape of sample changed after 7 freezing-thawing cycles showing the maximum damage caused by freeze-thaw cycles in macroscale. In Figure 13(h), a larger hole can be observed in the lower left corner. Compared with Figure 13(f ), the destruction of microstructure changed from microcracks into the larger hole means that cementation property between raw material particles was destroyed [25]. From a macroscopic perspective, the speed of mass loss increased significantly, and the mechanical properties suffered a significant reduction. In summary, with the progression of the freezingthawing cycles, the damage caused by the freezing process to the microstructure gradually became serious. e increase in porosity and microcracks leads to a sharp increase in damage caused by the expansion of ice lenses. After 5 freezing-thawing cycles, the RMCC do not be completely destroyed. However, most cement-treated soil in this paper was destroyed after 3 freezing-thawing cycles. erefore, RMCC exhibits better durability against freezing-thawing cycles.
3.4. Discussion on the Mechanism. In RMCC system, the hydrolysis of loess and red mud provides active ions. At the same time, the hydrolysis of red mud improves the alkalinity of the reaction system, and the hydrolysis of desulfurized gypsum provides enough Ca 2+ . Under the combined action of these three phenomena, the hydration process is facilitated [21,37]. e nucleation of Ca 2+ forms aggregates on the surface of raw material particles. ese aggregates are cemented by gelling products such as CSH gel and increase the interparticle interaction. Aggregates, fine particles, and hydration products attached to the surface of desulfurized gypsum particles reduce the dissolution of CaSO 4 ·2H 2 O and improve water stability of RMCC [46,47]. e CSH gel generated by hydration reaction has a cementation effect, while the AFt generated by pozzolanic reaction has a filling effect. erefore, when red mud content is low, the corresponding desulfurization gypsum content is also low, and the filling effect of AFt is not significant, resulting in a large quantity of micropores, low compressive strength, and poor durability against freezing-thawing cycles. When the red mud content is excessive, on the one hand, high alkalinity of the reaction system retards the hydration process. On the other hand, due to the large specific surface area of red mud [48], more gelling hydrates will be consumed in order to cement red mud particles. At the same time, high desulfurization gypsum content causes excessive AFt generated, and AFt is a mineral with expansive behaviour which significantly damages the microstructure. As a result, the mechanical properties of RMCC decline. When the red mud content is 12%, the cementation effect of CSH gel and the filling effect of AFt have the best coordination indicating the highest ability and stable microstructure of RMCC.

Conclusion
is paper provides a theoretical guidance to the use of red mud and desulfurization gypsum in subgrade strengthening which exhibits both environmental and economic benefits. Based on these results, the following conclusions are drawn: (1) e optimum red mud content of RMCC mixture is 12%. With this optimum content, RMCC has the highest strength and the best durability to freezingthawing cycles. (2) e increase of compressive strength can be divided into two stages, and the strength increases rapidly when the curing period is 14 days. With the increase of the red mud content, the strength and the resistivity of RMCC show the same variation trend, both are first increasing and then decreasing. At the optimum content of red mud, strength and resistivity reach the maximum. (3) Under freezing-thawing cycle condition, during the first 5 freezing-thawing cycles, the rate of compressive strength loss is faster, and between the 5th and 7th freezing-thawing cycles, the rate of compressive strength loss slows down. With the progression of the freezing-thawing cycle, the difference of variation in resistivity with different red mud contents decreases. At optimum red mud content, the pressure sensitivity of RMCC gradually reduces with the progression of freezing-thawing cycles. (4) In the RMCC system, the silicate hydrate contains active Ca 2+ , which can causes replacement reaction with alkali metal elements playing a key role in the fixation of the alkali metal elements. With the optimum content of red mud, the cementing effect of CSH gel and the filling effect of AFt are in the best coordination, and the interior pores are the lowest, which results in the highest compressive strength and the best durability to freezing-thawing cycles. e results demonstrate a great potential of RMCC used in subgrade strengthening.

Data Availability
All the data presented in the manuscript were obtained from laboratory tests at Taiyuan University of Technology in Shanxi China. All the laboratory testing data were presented in the Figures and tables in the manuscript. Additional data can be provided by the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.