Mechanical performances and microstructural characteristics of reactive MgO-carbonated silt subjected to freezing-thawing cycles

Guanghua Cai, Songyu Liu, Guangyin Du, Zhen Chen, Xu Zheng, Jiangshan Li

a College of Civil Engineering, Nanjing Forestry University, Nanjing, 210037, China

b Institute of Geotechnical Engineering, Southeast University, Nanjing, 211189, China

c State Key Laboratory of Geomechanics and Geotechnical Engineering,Institute of Rock and Soil Mechanics,Chinese Academy of Sciences, Wuhan,430071,China

d Department of Civil and Environmental Engineering, Washington State University, Pullman, WA99163, USA

Keywords:Reactive magnesia (MgO)Freezing-thawing (F-T) cycle Carbonated/stabilized silt Engineering performance Microstructural characteristics

ABSTRACT The characteristics of reactive magnesia(MgO)-carbonated silt in respect to long-term stability have not been well understood in severely cold climate despite the usage of reactive MgO in enhancing the engineering performances.Under the binder content of 15%and initial water content of 25%,MgO-admixed silt specimens were carbonized for 3 h and 6 h and then subjected to different numbers of freezingthawing (F-T) cycles. After different F-T cycles, the physico-mechanical properties of MgO- carbonated silt were analyzed in comparison with Portland cement (PC)-stabilized silt through physical and unconfined compression tests. Besides, a series of micro tests on MgO-carbonated specimens was performed including X-ray diffraction (XRD), scanning electron microscopy (SEM) and mercury intrusion porosimetry (MIP) tests. The results demonstrate that both mass change ratio and moisture content of carbonated/stabilized silt decrease,and these values of MgO-carbonated silt are significantly lower while the density is higher compared to PC-stabilized silt.The strengths and moduli of MgO-carbonated silt are still two times higher than those of PC-stabilized specimens and the strength change ratio of keeps above 0.8 after F-T cycles. There is no visible transformation between nesquehonite and dypingite/hydromagnesite,although the XRD peaks of nesquehonite decrease and the bonding and filling effects weaken slightly.After 6 and 10 F-T cycles,the pore-size characteristics changed from a unimodal distribution to a three-peak and bimodal distribution, respectively. The total, macro and large pore volumes increase obviously while the medium and small pore volumes decrease except for intra-aggregate pore. The findings show better F-T durability of MgO-carbonated silt, which would be helpful for facilitating the application of MgO carbonation in the soil treatment.

1. Introduction

Weak soils with high compressibility and low strength have caused many geotechnical problems, which need to be treated by appropriate solidification/stabilization techniques to improve the bearing capacity(Lorenzo and Bergado,2006;Shen et al.,2008;Lu et al.,2020;Rafiean et al.,2020).Ordinary Portland cement(PC)and lime are regarded as easy-to-use curing agents to treat weak soils owing to the low cost and good bonding effect (Zhang and Tao,2008; Liu et al., 2012; Du et al., 2014; Choobbasti and Kutanaei,2017; Oluwatuyi et al., 2020). However, such curing agents have brought serious environmental issues, such as large energy and resource consumption in cement production (kiln temperature>1450°C),as well as significant CO2emissions(Vandeperre and Al-Tabbaa,2007;Andrzej and Karin,2013;Wang et al.,2019;Lim et al.,2020). The above issues along with the slow strength growth and high alkalinity of PC have driven researchers to seek environmentally friendly alternatives from industrial by-product materials(e.g.slag,fly ash and carbide residue)(Du et al.,2016;Sudla et al.,2018;Zhao and Zhu, 2019; Zhu et al., 2020).

In recent years, as an alternative of low-carbon binder or activator, reactive magnesia (MgO) began to attract more and more attention due to its outstanding environmental advantages such as low energy consumption(calcining temperature<750°C)and high capacity in absorbing carbon dioxide (CO2) (Liska and Al-Tabbaa,2012;Unluer and Al-Tabbaa,2015).The carbonation of reactive MgO has become a feasible solution for addressing soft soil problems (Yi et al., 2013; Cai et al., 2015a, b). When subjected to the admixing of reactive MgO and subsequent CO2carbonation,the low-plastic soil could achieve higher unconfined compressive strength (UCS) than PC-solidified soil (28 d) (Yi et al.,2013,2016). The engineering performances of carbonated MgO-admixed silt (hereinafter referred to as “MgO-carbonated silt”) have been adequately explored under varying conditions of MgO dosage, moisture content, carbonation period and wetting-drying cycle(Cai and Liu,2017;Cai et al.,2019a,b). Moreover, the low liquid limit (or moisture content) of soil and high-activity MgO could significantly enhance the strength of MgOcarbonated soil(Cai et al.,2019a;Liu et al.,2020).

Lime and PC are commonly used binders in the weak subgrade treatment for decades, and the stabilized soils could achieve fairly good integrity and mechanical performances including strength and hydraulic permeability (Jamshidi et al., 2016). However, the engineering performances of the treated soil systems may be subjected to the varying degree of degradation under severe environmental conditions such as wetting-drying cycles, freezing-thawing (F-T)cycles and chemical attacks (Neramitkornburi et al., 2015; Wang et al., 2020). In most studies, the strength of materials was generally regarded as a highly relevant durability indicator to evaluate the integrity and stability of the subgrade or geotechnical engineering in severe environments(Shihata and Baghdadi,2001a;Du et al.,2016).In a cold climate, the consecutive F-T cycles caused the engineering performance degradation of the soils or stabilized soils(Shihata and Baghdadi,2001b;Júnior et al.,2018;Saleh et al.,2019;Jamshidi and Lake,2015).Cui et al.(2014)quantitatively analyzed the influence of F-T cycle on the microstructures of silty clay,and the result showed that the soil structure became looser and the pore orientation angles were unevenly distributed. The modulus and shear strength of PCtreated soils varied depending on soil types and F-T conditions(Simonsen et al., 2002; Zhang et al., 2016). Moreover, the shear strength of lime-treated soils gradually decreased with increasing FT cycles (Ma et al., 1999), but it could almost recover when the temperature became constant(Tebaldi et al.,2016).The soil friction angle increased while the cohesive force reduced after a string of F-T cycles (Wang et al., 2007). For PC-treated silty sand, the hydraulic conductivity increased by three orders of magnitude after 4 F-T cycles, while the strength decreased after 12 cycles (Jamshidi et al.,2015).

However,as an innovatively and effectively treated soil,it is not known whether the MgO-carbonated silt could be applied in the variably cold climate or not, and the effect of cyclic F-T on the engineering properties of MgO-carbonated silt has been rarely involved. Therefore, the F-T cycle tests of MgO-carbonated silt specimens were carried out to systematically study the changes of physical properties and explore the F-T durability compared with PC-solidified silt. Moreover, the microstructural tests including Xray diffraction (XRD), scanning electron microscopy (SEM) and mercury intrusion porosimetry (MIP) tests were performed to elucidate the transformations of mineral composition and microstructural morphology,and to probe into the evolution mechanism of micropores.The results will facilitate a better understanding for the F-T durability of MgO-carbonated silt and provide guidelines for designing cost-effective,reliable and durable soil treatment in cold climate zones.

2. Materials and methods

2.1. Materials

The soil used in this study was obtained from a subgrade site located in Suqian,China,and it was a kind of low-plastic silt as per ASTM D2487-11(2011).The physico-chemical properties of silt are listed in Table 1. The Atterberg limits were tested by the fall cone method, the pH value was tested by ASTM D4972-07 (2007), and the compaction characteristics were tested by ASTM D698-12(2012). The binders used in this study are the light-burned MgO and PC(#32.5),and their physical properties are shown in Table 2.The specific gravity values of silt, PC and MgO were measured according to ASTM D854-10 (2010), and their grain size characteristics were analyzed by Mastersizer 2000 laser diffractometry. It should be noted that the dehydrated alcohol was used in the analyses of grain size and the measurements of specific gravity to avoid any hydration of binders. The specific surface area (SSA) values of MgO and PC were tested by a Physisorption Analyzer (ASAP2020 HD88). The MgO activity was represented by the neutralization time between the acidic solution and MgO. The chemical compositions of MgO,PC and silt were analyzed by an X-ray fluorescence spectrometer (XRF) and their results are shown in Table 3. The high-concentration (99.9%) CO2was used in the carbonation experiment.

Table 1 Physico-chemical properties of silt.

Table 2 Physical properties of reactive MgO and Portland cement.

Table 3 Chemical compositions of the materials (by weight percentage).

2.2. Specimen preparation

The amount of commonly used binders, such as PC, usually ranges from 5% to 30% in the soil stabilization (Horpibulsuk et al.,2010; Liu et al., 2012), and the binder (PC and MgO) contents are recommended as 15% (as per the weight of dry soil) (Davis et al.,2007; Yi et al., 2016; Cai et al., 2019b). Moreover, considering that the binder-admixed soil specimen can be smoothly made and successfully demolded, the moisture content was determined as 25% (approximately the natural moisture content of 26.1%). A suitable amount of binder (PC and MgO) powder was first mixed with the dry soil according to the binder content of 15%. Subsequently,an amount of distilled water was added into the mixture as per the water content of 25%and subjected to uniform mixing.The MgO-admixed mixtures were covered by the plastic film and mellowed for at least one night (about 12-18 h) to achieve the moisture homogeneity.The mixtures were layered into a standard cylindrical mold (φ50 mm × 100 mm) and compacted using the hydraulic jack to suitable compaction degree. After achieving the maximum density,the compacted specimens were demolded with the help of a hydraulic jack. Three compacted specimens (weight error of ～3 g)were prepared for each F-T cycle test,and there were 123 specimens in total. The specimens were put into the carbonation chamber (Fig. 1), which was filled with high-pressure CO2(99.9%, 200 kPa) for accelerated carbonation of different periods(some specimens for 3 h and the rest for 6 h).The reason why two carbonation periods were chosen was that MgO-admixed specimen could reach the maximum strength after 3-6 h carbonation in previous study (Cai et al., 2015a). After carbonation, the hot specimens were taken out and placed into a curing room(20°C±2°C)for 24 h to cool down,since the carbonation was an obvious exothermic process(Cai et al.,2015a).It was worth noting that the carbonation in this study was much easier to conduct using a carbonation chamber rather than the modified triaxial carbonation instrument (Yi et al., 2013, 2016). Finally, to better evaluate the durability of MgO-carbonated specimens, some PC-stabilized specimens (a total of 63) used for comparison were also prepared and subjected to 28-d standard curing in a chamber (20°C ± 2°C and relative humidity of 95%).

2.3. Testing procedure of F-T durability

According to the F-T cycling specifications stipulated in ASTM D560/D560M-16 (2016), the F-T cycle experiments were performed on two types of carbonated/stabilized specimens. The specimens were placed onto the felt pads with the absorbing-water and spitting-water function,and the felt pads along with specimens were put into the plastic boxes without additional water.Then,the boxes were put into a cold chamber at a constant temperature of -23°C (-10°F) for 1-d freezing. The specimens were then removed out of the chamber and put into a standard curing room(20°C±2°C)for 1-d thawing.One complete F-T cycle included 1 d of freezing and 1 d of thawing. In this study, it was worth noting that the physical properties of the specimens as well as their mechanical performances became stable after 8 F-T cycles. Thus, the maximum number of F-T cycles was selected as 10. The mass, size and apparent characteristics of all specimens were measured and recorded before and after each freezing (or thawing). The mass change ratio of a specimen undergoing designated cycles was calculated as follows:

Fig.1. Test setup used for carbonation of reactive MgO-stabilized specimens.

where Rmis the mass change ratio(%),mnis the final weight after n F-T cycles (g), and m0is the initial weight before any F-T cycle (g).

The UCS, a key parameter for evaluating F-T resistance, should be conservatively tested under the thawing state after each F-T cycle, knowing that the strength of most soil materials after freezing is often much higher than that after thawing (Ma et al.,1999; Tebaldi et al., 2016). The UCS of all specimens after thawing were tested based on ASTM D4219-08(2008),and the vertical force was exerted at a speed of 1 mm/min. The variation coefficient of results was less than 7.6%,showing the high reproducibility of tests.The deformation modulus (E50) of specimens was determined according to the slope of the stress-strain curve at the half of the maximum stress.

After unconfined compression test, the moisture content of broken specimens was tested in a low-temperature oven(45°C)for drying over 2 d to achieve a steady weight.This temperature could prevent magnesium carbonates such as nesquehonite from losing the bound water (Unluer and Al-Tabbaa, 2013). Besides, the intrinsic mechanisms of the changes in the durability of MgOcarbonated soil (3-h carbonation) were explained through a series of SEM, MIP and XRD tests. Some fragments from broken specimens were first squashed into small pieces and then were soaked in liquid nitrogen. Finally, to remove the frozen water, the soil fragments were put into an F-T instrument with a vacuum chamber.Afterwards,one freeze-dried piece was used for SEM test after being coated with gold, and another piece was used for MIP test to provide the microstructural characteristics. Additionally, to detect any changes of crystalline phases,the powder from a freezedried piece(<75 μm)was used for XRD test under the circumstance of Cu-Kα radiation (30 mA/40 kV, 1.54059 ?). The powder was continuously scanned from 5°(2θ)to 50°at a scanning speed of 2°per minute and a step length of 0.02°.

3. Results and analyses

3.1. Physico-mechanical characteristics

Fig. 2 shows the appearances of MgO-carbonated and PCstabilized specimens after F-T cycles. There is no apparent change for both carbonated and stabilized specimens after 10 F-Tcycles and they still show good integrity except that there are only a few broken particles falling off from the ends of the specimens. Fig. 3 presents the mass change ratios of the two types of soil specimens after F-T cycles.Negative ratios represent the mass reduction compared to initial mass.It can be concluded that the mass change ratios gradually increase with more F-T cycles for both types of specimens. After 10 cycles, the mass change ratios of MgOcarbonated specimens (with 3 h and 6 h carbonation) are less than 2%,while those of PC-stabilized specimens are larger than 4%.The final fluctuating magnitudes of mass change ratio are 0.4%-0.5% and 1.4% for MgO-carbonated and PC-stabilized specimens,respectively,indicating that the mass changes of MgO-carbonated specimens are almost the same for 3 h and 6 h carbonation,and the change is far less than that of PC-stabilized specimens. The above results could be explained as follows:

Fig. 2. Photos of MgO-carbonated and PC-stabilized silt before and after F-T cycles: (a) NF-T = 0, and (b) NF-T = 10.

(1) When the freezing begins, the outside temperature of specimens is -23°C and the inner temperature is near initial room temperature (～20°C). As the freezing continues, the internal temperature gradually decreases until it is the same as the external temperature(-23°C),and the unfrozen water has a migration tendency from the high-temperature inside to the low-temperature outside under the temperature gradient, leading to water loss.

(2) On the contrary, when the thawing begins, the outside temperature of specimens is the room temperature(～20°C)and the inner temperature is near -23°C. As the thawing continues,the internal temperature gradually increases until the external room temperature (～20°C), and the moist water has a moving tendency from the high-temperature outside to the low-temperature inside under the temperature gradient, leading to the water absorption.

(3) There is a slight shedding phenomenon of small particles on the surface of specimens.

Fig. 4 depicts the moisture content changes of both MgOcarbonated and PC-stabilized specimens with F-T cycles. The moisture contents of MgO-carbonated and PC-stabilized silt decrease as the F-T cycle increases,and the moisture content of PCstabilized silt is significantly higher than that of MgO-carbonated silt. The MgO-carbonated silt without F-T cycle has a lower moisture content compared to PC-stabilized silt. This is due to the greater water consumption for MgO-carbonated soil during the exothermic hydration and carbonation of reactive MgO while the water consumption only occurs in the hydration process (or pozzolanic reaction) of PC (Cai et al., 2015b, 2019b).

The F-T cycles almost have no influence on the volumes of two types of silt specimens. Both the bulk density and dry density of specimens vary with mass and moisture content,as shown in Fig.5.The bulk density and dry density of MgO-carbonated silt keep the constant levels of 1.96-1.99 g/cm3and 1.74 g/cm3, respectively,after several F-T cycles, which are higher than the corresponding values of PC- stabilized silt (1.82-1.86 g/cm3and ～1.59 g/cm3,respectively).The bulk density and dry density values for 3-h and 6-h carbonated silt are very close after F-T cycles. Regardless of F-T cycles,the results of mass change ratio,water content,bulk density and dry density are almost the same after 3 h and 6 h carbonation,indicating that 3 h carbonation could achieve the stable physical properties. The great difference between the bulk density and dry density for carbonated/stabilized specimens is mainly due to the significant growth in weight and the decrease in moisture content of MgO-carbonated specimens. The very slight change in the bulk density or dry density with F-T cycles mentioned above is attributed to the small amount of clay particles in specimens.As revealed in previous studies,PC-stabilized clay showed much larger frozenheave rate compared to PC-stabilized silt or sand under the same moisture content (Wang et al., 2018; Fan et al., 2019).

Fig. 3. Mass change ratio of MgO-carbonated and PC-stabilized silt after F-T cycles.

Fig. 6 depicts the variations of strength for two types of silt specimens with increasing F-T cycles.As observed in Fig.6,the UCS of both MgO-carbonated (3-h and 6-h carbonation) and PCstabilized silt decreases with F-T cycles, especially after the second cycle in comparison with the standard curing. The strength values of MgO-carbonated silt,in the range of 4-5.3 MPa in spite of F-T cycles, are 2-3 times those of PC-stabilized silt (1.5-1.7 MPa),indicating that the MgO binder and its carbonation have a good effect in the silt treatment compared to PC solidification.This result could be explained by the difference in the treatment mechanisms.The PC-stabilized silt mainly gains the strength development through the pozzolanic reaction and weak ion-exchange reaction of PC. In contrast, the strength gain of the MgO-carbonated silt is attributed to the bonding and the pore filling of the hydration and carbonation products. It is noteworthy that the strength of 3-h carbonated specimens is higher than that 6-h carbonated specimens.This phenomenon may result from the following facts:(1)the good permeability makes the MgO-admixed silt specimens reach the maximum strength in the first 3 h effective carbonation;and (2) the long-period high-pressure exposure (>3 h) might impair the weak cementation amongst silt particles and adversely affect the mechanical strength of specimens (Cai et al., 2015a).

Fig. 4. Moisture content of MgO-carbonated and PC-stabilized silt after F-T cycles.

Fig. 5. Density of MgO-carbonated and PC-stabilized silt after F-T cycles.

Fig. 6. UCS of MgO-carbonated and PC-stabilized silt after F-T cycles.

Fig.8. Relationship between residual strength and mass loss ratio of MgO-carbonated and PC-stabilized silt after F-T cycles.

For analyzing the change reason of strength, the strength change ratios of carbonated/stabilized specimens after F-T cycles compared to the standard curing (qu(F-Tcycle)/qu(Standardcuring)) are shown in Fig. 7. The strength change ratios of MgO- carbonated specimens are 0.86-1, overall greater than those of PC-stabilized specimens, especially after 5 F-T cycles. Besides, the strength change ratios for all specimens become steady after 8 F-T cycles.Although both MgO-carbonated and PC-stabilized specimens maintain relatively good integrity and soundness after several F-T cycles (Fig. 2), there are still some broken particles falling off from specimens, resulting in a slight mass reduction, and thus affecting the residual strength. Fig. 8 shows the relationship between the residual strength and the mass loss ratio of carbonated/stabilized specimens after F-T cycles. Based on Fig. 8, the residual strength declines as the mass loss ratio increases.The relationship between the residual strength and the mass loss ratio can be appropriately represented by an exponential function.

Fig.7. Strength change ratio of MgO-carbonated and PC-st abilized silt after F-T cycles.

To show the resistance of specimens to deformation after F-T cycles, the deformation moduli (E50) of carbonated/stabilized specimens after F-T cycles are depicted in Fig. 9. As shown in this figure, the deformation moduli of carbonated specimens (～410-550 MPa) are about two times greater than those of stabilized specimens(～250 MPa).These results indicate that it is difficult for F-T cycles to affect the mechanical properties of both carbonated and stabilized specimens.

Fig.9. Deformation modulus of MgO-carbonated and PC-stabilized silt after F-T cycles.

3.2. Phase transformation analysis

Fig.10 shows the XRD results of carbonated silt after F-T cycles with the peaks of primary substance phases marked by vertical lines. The peak intensities in XRD patterns are set on a semilogarithmic scale to scrutinize the evolution of hydration and carbonation products.As shown in Fig.10,apart from some strong peaks of quartz at 2θ of 20.83°(4.26 ?), 26.59°(3.35 ?), 36.57°(2.45 ?),39.5°(2.28 ?)and 42.45°(2.13 ?),there are some brucite peaks at 2θ of 18.67°(4.75 ?),20.03°(4.43 ?)and 38.04°(3.63 ?)in all specimens regardless of F-T cycles. It is hard to detect the MgO peaks in these specimens,indicating that the MgO mixed with silt has been fully hydrated. Furthermore, it can also be observed that magnesium carbonates are found in all MgO-carbonated specimens despite F-T cycles such as nesquehonite (N), dypingite (D) and hydromagnesite(H).Nesquehonite is mainly located at 2θ of 13.65°(6.48 ?),23.08°(3.85 ?),34.24°(2.62 ?)and 29.45°(3.03 ?),while the dypingite/hydromagnesite is detected at 2θ of 15.11°(5.89 ?),15.29°(5.79 ?),30.48°(2.93 ?)and 41.58°(2.16 ?).As the number of F-T cycles increases,the peak intensities of nesquehonite have a certain decreasing trend,but there is no significant increase in the peaks of dypingite/hydromagnesite. Meanwhile, it is difficult to distinguish specific carbonation products at some overlapping peaks (such as dypingite/hydromagnesite, and MgO/magnesite)through the XRD analysis.

Fig.10. XRD of MgO-carbonated silt (3 h) after different F-T cycles.

3.3. Microstructure evolution analysis

Fig.11 shows some typical SEM images(magnified by 1000 and 5000 times) of MgO-carbonated silt after different F-T cycles. Previous studies have identified the existence of prismatic crystals of nesquehonite (N), and rosette-flaky crystals of dypingite/hydromagnesite(D/H)in MgO-carbonated specimens(Yi et al.,2013;Cai et al.,2015b,2019b).In all MgO-carbonated specimens,a plenty of carbonates, including nesquehonite and dypingite/hydromagnesite,are distinctly seen in Fig.11a-c.Before F-Tcycles,the silt particles are closely bound by the carbonation products to form a dense structure as well as a little number of interconnected pores(Fig.11a).After 6 F-T cycles,the linkage between prismatic crystals of nesquehonite begins to weaken apparently, and some small pores gradually expand compared to no F-T condition (Fig. 11b).After 10 F-T cycles,the soil structure becomes less compact,and the bonding between prismatic crystals is further loosened, as shown in Fig.11c. There is still abundant prismatic nesquehonite but less rosette-flaky dypingite/hydromagnesite in the microscopic morphology. Therefore, it is difficult to determine that the F-T cycles promote any transformation of carbonation products from the microscopic images.This finding is consistent with those obtained from the aforementioned physical,mechanical and XRD tests.

3.4. Micropore distribution characteristics

From the MIP results, Fig. 12 describes the micro-pore distribution characteristics of carbonated specimens after several F-T cycles. The cumulative pore volume is determined from the cumulative mercury intrusion volume. According to Fig. 12a, when the carbonated specimens experience 6 and 8 F-T cycles, their cumulative pore volumes change from 1.33 mL/g to 1.44 mL/g and 1.45 mL/g, respectively, but their macro dry density is kept at～1.74 g/cm3. The cumulative pore volume of MgO-carbonated silt first increases and then becomes stable after 6 F-T cycles. Fig.12b depicts the differential pore size distribution (dV/d(log10D)) of carbonated specimens before and after F-T cycles, where V and D are the cumulative pore volume and pore diameter, respectively.Based on Fig.12b, the pore size distribution of carbonated specimens before F-T cycle follows a unimodal pattern as frequently observed in coarse granular soil, and the peak of the unimodal curve appears at the diameter of 0.5 μm (Zhang and Li, 2010).However,after the MgO-carbonated silt has been subjected to 6 and 10 F-T cycles,the distribution pattern turns into a three-peak curve and a bimodal curve,respectively.The three peaks appear distinctly at the diameters of ～10 μm(primary peak),～1.05 μm(secondary peak) and 0.05 μm (third peak), while the primary and secondary peaks of the bimodal curve appear at the diameters of ～10.05 μm and ～0.3 μm,respectively.Previous studies have indicated that the bimodal pattern of pore size distribution is also observed in structured soil,residual soil,MgO-carbonated silty clay and most of the PC-stabilized soils(Delage et al.,2006;Zhang and Tao,2008;Li and Zhang, 2009; Lemaire et al., 2013; Du et al., 2014). The transformation from a single peak to multiple peaks indicates some refinement of the pore structures and an increase in macro pore volume to a large extent.

To show the evolution of various pores in detail, the pores are divided into macro pore(>10 μm), large pore(1-10 μm), medium pore(0.1-1 μm),small pore(0.01-0.1 μm)and intra-aggregate pore(<0.01 μm) according to pore diameter (Horpibulsuk et al., 2010).Fig.12c and d shows the comparison of various pore volume and incremental pore volume.As the F-T cycles increase,the volumes of macro and large pores as well as total pores increase remarkably while the volumes of medium and small pores decrease (Fig.12c and d). There is almost no change for the volume of the intraaggregate pores. The changes of the pore structure for MgOcarbonated silt after several F-T cycles should be primarily on account of the cracking of silt particles caused by the expansive force generated from the free water freezing (Wang et al., 2012). The overall effect of the freezing is increasing the large and macro pore volumes of carbonated soil but decreasing the volume of medium pores. In conclusion, the MIP experiments provide an in-depth insight into the great changes in the microstructure of MgOcarbonated silt under cyclic F-T from the perspective of the pore structure. The findings of pore distribution characteristics are basically the same as those of XRD and SEM.

Fig.11. SEM pictures of different MgO-carbonated silt(3 h)after different F-T cycles:(a)NF-T=0,(b)NF-T=6,and(c)NF-T=10(N:Nesquehonite;D:Dypingite;H:Hydromagnesite).

4. Discussion

In this study,the two sets of MgO-carbonated silt specimens for different periods (3 h or 6 h) showed similar physical properties(such as the mass change rate, density and moisture content) but slightly different mechanical properties (such as strength and moduli). The steady physico-mechanical characteristics indicate that the MgO-admixed specimens could achieve the complete carbonation after 3 h. Contrasted with the PC-solidified silt specimens,the MgO-carbonated silt specimens have the lower moisture contents, higher bulk densities (or dry densities) as well as higher strengths and deformation moduli. The above results can be explained as follows:(1)The water loss of MgO- admixed soil due to the hydration and carbonation reactions is far more than that of PC-stabilized soil because of the only hydration reaction(Fig.4);(2)Under the similar volume,the mass growth of MgO-carbonated soil from CO2absorption leads to the growth in the bulk density or dry density of specimens(Fig.5);and(3)The developments of strength for MgO-carbonated and PC-stabilized specimens come from the cementing and filling effects of MgO carbonation products (especially nesquehonite) and PC hydration products, respectively. The lower moisture content and porosity together with the higher dry density lead to the significant growth in strength and moduli.However, the unexpected reverse in mechanical strength was developed for 3-h and 6-h MgO-carbonated specimens. The possible reason was the high carbonation effectiveness and some adverse effects from long-period high-pressure exposure in the pressure chamber(Cai et al., 2015a, b).

Fig.12. Influence of F-T cycles on the porosity of MgO-carbonated silt: (a) Cumulative pore volume; (b) Pore-size distribution; (c) Comparison of various pore volumes; and (d)Incremental pore volume.

Fig.13. Schematic illustration of microstructural evolution in MgO-carbonated silt during freezing-thawing cycles: (a) First step; (b) Second step; and (c) Third step.

Cyclic F-T in a cold climate is a natural and unavoidable phenomenon affecting the soil stability (Wang et al., 2012). Previous studies have shown that the untreated silt specimens are more vulnerable to be affected by short-term F-T cycles compared with fine sand and clay (Cui et al., 2014). The reason is that the high capillarity under the specific pore sizes and sufficient permeability leads free water to the icing front and facilitates the generation of ice lens(Zwissler et al.,2016).When the MgO-carbonated and PCstabilized silt specimens have been subjected to several periodic FT cycles, there are different degrees of reduction in their mass change rates,moisture contents,UCSs and moduli.Compared with the 28-d PC-stabilized silt specimens, the MgO-carbonated silt specimens exhibit smaller changes in the mass change ratio and moisture content but slightly higher strength and moduli with increasing F-T cycles.According to Parsons and Milburn(2003),the mass loss ratio of PC-stabilized soil specimens(2%-7%)after 12 F-T cycles was the lowest compared to fly ash-treated soil and limestabilized soil (7%-19%). Therefore, it can be deduced that the F-T cycle exhibits the least influence on the mass loss ratio of MgOcarbonated specimens compared to PC-stabilized specimens.

Fig.13 shows a schematic illustration of microstructural evolution in MgO-carbonated silt specimens during the F-T cyclic process.Before every F-T cycle, the MgO-carbonated silt specimen forms a compact structure with higher strength through the cementing effect of hydrated magnesium carbonates (Fig. 13a). During the severe freezing period, the part of pore water is frozen to ice lens, and the associated volume expansion from the ice will exert stress on the surrounding soil structure(Zhang et al.,2016).If the expansion stress among particles exceeds the tensile strength of cementing products,some irreversible structural damage such as micro-cracking will occur(Fig.13b).When the ice thawed,the micro-cracking would not completely disappear, and the soil structure became disrupted,resulting in greater internal porosity and higher moisture content in the soil matrix (Zwissler et al., 2016). After repeated F-T cycles, the cracks of MgO-carbonated and PC-stabilized specimens would become larger (Fig.13c). However, the MgO-carbonated specimens are not damaged under exposure to cyclic F-T owing to the strong cementingeffectofmagnesiumcarbonates.Moreover,therepeatedFT cycles could alter the cumulative pore volume and pore size distribution,leading to the increase of the macro and large pore volumes as well as the reduction of the medium and small pore volumes(Fig.12d).Compared to common clay,cyclic F-T poses little effect on the soil structure of MgO-carbonated silt with very low moisture content and slightly larger pore size.The relatively fine particles could fill the intra-aggregate pores,which alleviates the destructive effect on the soil structure to some extent under extremely severe environmental conditions. There is no apparent phase transformation observed in the absence of high temperature,even though the micropore structure morphology and silt particle orientation may be altered on account of the swelling stress caused by the frozen water and ice.

The periodic F-T cycles have greater effects on soil aggregate stability under different conditions of clay minerals,organic matter,aggregate size, initial moisture content, number of F-T cycles and freezing temperature. Nevertheless, this investigation only studies the permanence of MgO-carbonated silt against cyclic F-T under a specific temperature range. Besides, other studies have not been addressed with regard to the effects of F-T cycles on the engineering properties such as compressibility, hydraulic permeability and leaching properties of MgO-carbonated soils. Further research will need to be performed about the durability of other MgO-carbonated soils under the coupling effects of F-T cycle and chemical attack.

5. Conclusions

Comprehensive investigations have been performed to reveal the effects of F-T cycles on the engineering permanence of MgOcarbonated silt. The main conclusions are summarized as follows:

(1) As the F-T cycle increases, the mass change ratios and moisture contents of MgO-carbonated and PC-stabilized specimens gradually decrease. Under the same F-T cycles,the mass change ratio and moisture content of MgOcarbonated specimens are significantly lower while the density is much higher compared with PC-stabilized specimens.

(2) The strength of MgO-carbonated and PC-stabilized specimens reduces to varying degrees especially after 2 F-T cycles,and their residual strength ratio keeps above 0.8. The strength of MgO-carbonated specimens slightly decreases with increasing F-T cycle, but the strength and moduli are still two times higher than those of PC-stabilized specimens,showing better F-T durability.

(3) There is a certain decrease in the peaks of main carbonation products. The bonding and filling effects of rod-like nesquehonite begin to weaken after F-T cycles, making the soil structure less compact and loosened.However,the peaks of dypingite/hydromagnesite decrease significantly, and there is no visible transformation between nesquehonite and dypingite/hydromagnesite detected.

(4) After 6 and 10 F-T cycles, the pore-size characteristics of MgO-carbonated silt change from a unimodal curve to a three-peak curve and a bimodal curve, respectively. Various kinds of pores underwent the remarkable evolution,and the total,macro and large pore volumes increase obviously while the medium and small pore volumes of decrease except for intra-aggregate pores.

(5) A microstructural mechanism of F-T cycles of MgOcarbonated silt is proposed and described schematically.The findings would be helpful for facilitating the application of the carbonation method in the soil treatment.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The authors are grateful to the support of the National Natural Science Foundation of China(Grant Nos.41902286 and 41972269),and Open Research Fund of State Key Laboratory of Geomechanics and Geotechnical Engineering,Institute of Rock and Soil Mechanics,Chinese Academy of Sciences (Grant No.Z019026).