Download as pdf or txt
Download as pdf or txt
You are on page 1of 39

Wang Liyang (Orcid ID: 0000-0002-8063-354X)

Liu Jiankun (Orcid ID: 0000-0001-5113-8374)


Wang Tengfei (Orcid ID: 0000-0003-4079-0687)

A simplified model for the phase composition curve of saline soils considering

the second phase transition

Liyang Wang1,2,3; Jiankun Liu1,4; Xinbao Yu3; Tengfei Wang5; Ruiling Feng2

1
School of Civil Engineering, Sun Yat-sen University, Zhuhai 519082, China
2
School of Civil Engineering, Beijing Jiaotong University, Beijing 100044, China.
3
Department of Civil Engineering, The University of Texas at Arlington, Arlington, TX 76019,
United States.
4
Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai, 519082,
China
5
MOE Key Laboratory of High-Speed Railway Engineering, Southwest Jiaotong Univ., Chengdu
610031, China;

Liyang Wang (16115267@bjtu.edu.cn, Orcid ID: 0000-0002-8063-354X)


Jiankun Liu (liujiank@mail.sysu.edu.cn, Orcid ID:0000-0002-3064-8575)
Xinbao Yu (xinbao@uta.edu, Orcid ID: 0000-0002-5681-0390)
Tengfei Wang (w@swjtu.edu.cn, Orcid ID:0000-0003-4079-0687)
Ruiling Feng (rlfeng@bjtu.edu.cn)

Corresponding authors: Jiankun Liu (liujiank@mail.sysu.edu.cn); Xinbao Yu (xinbao@uta.edu)

Key Points:
 The correlation between the matrix effect and the freezing point depression is simplified.
 A new conceptual model for the phase composition curve of saline soils is proposed and
validated.
 The phase composition characteristic of the pore solution of saline soils are analyzed.

This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1029/2020WR028556

©2020 American Geophysical Union. All rights reserved.


Abstract
Phase composition curves are the basis for the thermo-hydraulic analysis of saline soil in cold
regions. Two types of phase transition, namely the water-ice and solution-crystal phase
transitions, would occur in saline soils due to the variation in temperature and migration of
moisture and solute, and the occurrence order of the two types of phase transition is governed by
the initial salt and water content. This study proposed a conceptual model for the water phase
composition curve in saline soils by simplifying the correlation between pore solution properties
and soil grain. This model was subsequently applied to the analysis of phase composition
characteristics and explained the occurrence of a second phase transition in saline soil. The
model was validated by the experimental data from known soils and showed both good results
and ease of use when compared to three previous models. The conceptual model provides a
convenient and intuitive way to understand the phase composition characteristics of pore solution
in saline soil in cold regions, and is also easy to use in the thermo-hydraulic analysis such as the
prediction of unfrozen water content and solute concentration.

1 Introduction
Frozen soil is widely distributed in cold regions perennially or seasonally (Zhang et al.,
1999; Zhang et al., 2003). The frozen layer retains water in cold seasons and releases water in
warm seasons (Gao et al., 2018), which directly affects the local hydrological cycle. The
occurrence of the frozen layers also inhibits rainfall infiltration (Hayashi et al., 2003), while the
degradation of frozen layer contributes to the growth of thermokarst lakes (Grosse et al., 2013).
In addition to the impact on the local environment, structures built on frozen soils are also
subjected to frozen-soil-related problems, including frost heave (Tai et al., 2018), thaw
settlement (Tai et al., 2020), frost jacking (Guo et al., 2016), and creep (Qi et al., 2013).
Unfrozen water content fundamentally affects the hydromechanical behavior of frozen soil (Chen
et al., 2006; Qi et al., 2013; Teng et al., 2020), and serves as an indicator of potential ice
formation and moisture migration.
In many cases, pore water in the soil is not completely frozen at sub-zero temperatures
due to the soil grain matrix effect (Andersland & Ladanyi, 2004), which includes the surface
tension of pore water, the size of pore, and the adsorption of soil grain to water (Lu, 2020;
Scherer, 1999). The correlation between unfrozen water content and temperature is described by
the phase composition curve (Anderson & Tice, 1972; Liu & Yu, 2013; Lovell, 1957). Soluble
salts (e.g. 𝑁𝑎2 𝑆𝑂4, 𝑁𝑎𝐶𝑙, and 𝑁𝑎2 𝐶𝑂3) are common components in the natural soils in the
cold and arid regions of China and elsewhere (Han et al., 2018; Wang et al., 2020; Xu, 1993),
and the freezing points of pore water solutions are also depressed by the solutes (Banin &
Anderson, 1974; Bouyoucos & McCool, 1916; Wan & Yang, 2020; Xiao et al., 2018). The
solubility of the salts is governed by the temperature, and the solute will precipitate from the
pore solution as hydrate crystal when the solute concentration approaches or exceeds the
solubility, which also consumes the unfrozen water (Pronk, 2006; Steiger & Asmussen, 2008;
Wan et al., 2017; Xu et al., 2001; Zhang, 1991).

©2020 American Geophysical Union. All rights reserved.


In this study, a temperature range of +30 ℃ to -30 ℃, is assumed to cover the ambient
temperature variation in cold regions (Moiwo et al., 2011; Tai et al., 2018; Tai et al., 2020). For
non-saline soil, only one phase transition point can be observed in the phase composition curve
in the given temperature range, which indicates the beginning of the water-ice phase transition.
However, for saline soil, there are frequently two phase transition points in the phase
composition curve, which indicate the occurrences of water-ice and solution-crystal phase
transitions. For a given soil texture and type of salt, the phase composition curves of different
initial conditions tend to be very similar after reaching the second phase transition point (Wan et
al., 2015; Zhang, 1991). Other factors that can affect the phase composition curve include initial
pore saturation (Watanabe & Wake, 2009), pressure (Low et al., 1968) and freeze-thaw cycles
(Spaans & Baker, 1996), although these factors have limited impact (Anderson & Tice, 1972;
Zhou et al., 2020).
The phase composition curve of pore water is critical to analyze the thermo-hydraulic
processes of frozen soil in cold regions (Teng et al., 2020; Zhou et al., 2020). Based on the phase
composition curve, several heat-moisture coupled models have been developed for hydrologic
research in cold regions (Harlan, 1973; Taylor & Luthin, 1978; Wu et al., 2015). In addition,
these coupled models are used to study infrastructure performance in cold regions (Liu et al.,
2020; Tai et al., 2020). Phase composition curve can be obtained by experimental methods
(Anderson & Morgenstern, 1973), such as time/frequency domain reflectometry (Yu et al., 2010),
nuclear magnetic resonance (Li et al., 2020), and calorimetry (Kozlowski, 2009). However, since
it is time-consuming and costly to determine phase composition curves by means of laboratory
tests, the development of robust phase composition curve models would facilitate soil stability
studies in cold regions (Dillon & Andersland, 1966).
Several theoretical and empirical phase composition curve models for pore waters of soils
in cold regions (Anderson & Tice, 1972; Dillon & Andersland, 1966; Michalowski & Zhu, 2006;
Teng et al., 2020; Wang et al., 2017), some of which take the solute compositions into account
(Banin & Anderson, 1974; Wan & Yang, 2020; Xu et al., 1995; Yong et al., 1979; Zhou et al.,
2020). However, these models only include the water-ice phase transition in the pore solution,
and none of them consider the solution-crystal phase transition. As mentioned above, the
solution-crystal phase transition also significantly affects the unfrozen water content and cannot
be ignored. Some of these previous theoretical models are based on capillary freezing theory
(Liu & Yu, 2013; Wan & Yang, 2020; Wang et al., 2017; Zhou et al., 2018), which assumes the
pore water is frozen in pores of various sizes. The calculation processes are complex and include
many parameters, such as pore size distribution, which impedes the wide utilization of these
models (Chai et al., 2018). Some empirical models fit the trend of experimental phase
composition curve well and are easy to use (Anderson & Tice, 1972; Michalowski & Zhu, 2006),
however, they generally fail to interpret the development of phase composition characteristics of
saline soil (Teng et al., 2020).
The objectives of this study are to analyze and discuss the phase composition
characteristics of saline soils, simplify the correlation between the matrix and the freezing point

©2020 American Geophysical Union. All rights reserved.


of pore water (pore solution), and to develop and validate a simple but practical conceptual
model to describe and interpret the phase composition characteristics of saline soil over the given
temperature range. In addition, the impact of initial solute concentration and water content on the
phase composition behavior is analyzed. This proposed conceptual model will assist in the
understanding of phase composition characteristics of saline soil, and it can be easily employed
to calculations and simulations related to frozen soil with or without salts.

2 Background

2.1 Water freezing in non-saline soil


For a given non-saline soil, the unfrozen water content decreases quickly as the
temperature drops to the initial freezing point 𝑇𝑓0 as shown in Fig.1. The decrease rate of
unfrozen water content also drops quickly as the temperature decreases due to the enhancing
matrix effect, which has been widely reported and discussed (Andersland & Ladanyi, 2004;
Dillon & Andersland, 1966; Xu et al., 2001). Below the initial freezing point of the given soil, it
is the icing stage indicating the presence of water-ice phase transition.

2.2 Water freezing in saline soil

(1) Water freezing in a salt-water system


The presence of a solute significantly influences the physicochemical behavior of water.
Fig.2 shows a typical phase diagram for an inorganic salt-water system (Pronk, 2006). The phase
of a salt-water system is governed by temperature and solute concentration. The freezing point of
water is depressed by solute, and the relationship between solute concentration and freezing
point is plotted as ice line, while the solubility line shows the relationship between solute
capacity of system and temperature. The intersection of the ice line and solubility line is named
the eutectic point (Pronk, 2006), where ice and salt crystal occur simultaneously. For a system
with an initial state above the ice line and solubility line, the system is an aqueous solution. If the
initial state point is above the ice line, the state point of the given system moves vertically
downward until it reaches the ice line during cooling. After the state point reaches the ice line,
the system capacity for liquid water decreases in the aqueous solution as the temperature
decreases, and, the icing process begins to occur. The solution becomes more concentrated due to
the progressive loss of solvent during the icing process, and the state point of the system moves
along the ice line until it reaches the eutectic point during cooling. The system will change into a
totally solid phase when the temperature is lower than that of the eutectic point. On the other
hand, if the initial state point is above the solubility line, the movement of state point of the
system shows a similar pattern but follows the solubility line during cooling. It can be inferred
from the phase diagram that the initial solute concentration governs the movement of state point
during cooling, and systems with various initial solute concentrations will reach the same
endpoint at the eutectic point.

©2020 American Geophysical Union. All rights reserved.


(2) Pore water freezing in saline soil
The phase composition curve of saline soil is more complicated than that of non-saline
soil due to the presence of salt. The formation of the hydrated salts mainly depends on the types
of solute and the temperature. For instance, as per Pronk 2006, the product of the solution-crystal
phase transition of a 𝑁𝑎𝐶𝑙 aqueous solution is solid 𝑁𝑎𝐶𝑙 when the temperature is above 0.1
℃, while the product is 𝑁𝑎𝐶𝑙 ∙ 2𝐻2 𝑂 when the temperature is below 0.1 ℃. In this paper, we
mainly focus on the cases which generate hydrated salts after the solution-crystal phase
transition. Besides the occurrence of water-ice phase transition, the solution-crystal phase
transition (i.e., the precipitation of the salt from the solution) also significantly influences the
phase composition curve. The initial solute concentration governs the order in which the two
transitions occurs. In this study, when the water-ice phase transition occurs earlier than the
solution-crystal phase transition during cooling, the soil is recognized to have a low initial solute
concentration; otherwise, the soil is recognized to have a high initial solute concentration.
The typical water-ice phase transition of a soil with a low initial solute concentration
(Fig. 3a) starts at the initial freezing point, 𝑇𝑓0 , and the unfrozen water content gradually
decreases with decreasing temperature. Due to the low initial solute concentration, the pore
solution remains undersaturated at the freezing point, and the solute is concentrated due to the
loss of solvent in the icing stage. When the pore solution is sufficiently concentrated and the
temperature reaches the second phase transition point 𝑇𝑠 , the solution-crystal phase transition
occurs as the second phase transition in soil, termed the icing-precipitating stage in this paper.
The newly occurred solution-crystal phase transition accelerates the consumption of unfrozen
water together with the water-ice phase transition. It is noted that the unfrozen water content
drops rapidly with decreasing temperature at the beginning of the icing-precipitating stage, and
then the phase composition curve quickly stabilizes. As shown in the phase diagram of a salt-
water system in Fig. 2, both the ice line and solubility line are relatively smooth curves. It
indicates that the water-ice phase transition and the solution-crystal phase transition tend to be
gradual processes as the system is equilibrated, and the unfrozen water content should be
consumed gradually after the second phase transition begins. However, the observed phase
composition curve in the rapid phase transition zone seems not to match the gradual tendency of
the water-ice phase transition or solution-crystal phase transition at the equilibrium state.
Although the exact solute concentration of the pore solution in the rapid phase transition zone
was not measured, we suppose the pore solution is unequilibrated in the rapid phase transition
zone based on the tendency of phase composition curve.
For a given saline soil with a high initial solute concentration, the solubility of solute in
aqueous solution usually drops with decreasing temperature and part of the solute which exceeds
the solubility would precipitate as crystals. Some of the liquid water may also combine with the
solute and form hydrated crystals (e.g. 𝑁𝑎2 𝑆𝑂4 ∙ 10𝐻2 𝑂) in the precipitating stage. Generally,
the concentration in the pore solution is regarded to be equal to the saturated concentration at an
equilibrium state and drops with decreasing temperature after the temperature is below the initial
precipitation point 𝑇𝑐0 , as shown in Fig.3 (b). The presence of solute depresses the freezing

©2020 American Geophysical Union. All rights reserved.


point of pore water, while the decreasing concentration attenuates the magnitude of the
depression. When the temperature reaches the second phase transition point 𝑇𝑠 , the water-ice
phase transition occurs as the second phase transition, and the phase composition curve also
develops into the icing-precipitating stage. The newly occurred water-ice phase transition
accelerates the consumption of unfrozen water together with solution-crystal phase transition.
The pore solution of a soil with a high initial solute concentration is supposed to be
unequilibrated in the rapid phase transition zone due to the same reason for the soil with a low
initial solute concentration.

3 Conceptual models for non-saline soil and saline soil


In this section, the conceptual models for non-saline soil and saline soil are developed
with the following assumptions and simplifications.
(1) The model simplifies the structure of saline soil as a spherical particle that is covered
by the unfrozen water film.
(2) The soil grain and pore water are assumed as homogeneous materials, and the density
of liquid water is assumed to be 1 𝑔/𝑐𝑚3 .
(3) The pore solution is assumed to follow the solubility line of the salt-water system
when the temperature is below the second phase transition point.
(4) The relationship between the freezing point depressed by the soil matrix and the
thickness of the unfrozen water film is assumed as a power function of the thickness of the
unfrozen water film.

3.1 Non-saline soil


In non-saline soil, the freezing point of pore water 𝑇𝑓 is depressed by soil grain matrix
effect and expressed as follows.
𝑇𝑓 = 𝑇𝑓∗ + ∆𝑇𝑓𝑚 (1)

where 𝑇𝑓∗ is the freezing point of pure water at a reference state, and ∆𝑇𝑓𝑚 the depression

induced by soil matrix effect. In this study, the freezing point of free pure water at atmospheric
pressure is selected as a reference state, and, thus, 𝑇𝑓∗ is 0 ℃.

As discussed previously (Nagare et al., 2012; Watanabe & Wake, 2009; Xu et al., 2001),
pore saturation (as a function of initial water content and dry density of soil) has a minor
influence on phase composition curve compared to temperature and soil texture. Therefore, the
influence of pore saturation is neglected during the modeling progress. To establish a conceptual
model in Fig.4, soil grain is assumed as spherical, and unfrozen water covers it as a film. The
unfrozen water film in the conceptual model comprises of capillary water and bound water. 𝑟𝑚

©2020 American Geophysical Union. All rights reserved.


is the radius of the soil grain, and 𝑟 denotes the distance from the center of soil grain to the
interface between ice and liquid water. Thus, the distance from the water molecule at the
interface between ice and liquid water to the soil grain boundary, 𝑥, is calculated by 𝑥 = 𝑟 −
𝑟𝑚 . As reported by Scherer (1999), the depression in freezing point induced by soil matrix,
∆𝑇𝑓𝑚 , is a kind of power function of the pore size. Therefore, the relationship between the
distance 𝑥 and the magnitude of depression ∆𝑇𝑓𝑚 is simply described by a power function and
expressed as follows.
𝑎 𝑎 𝑚
∆𝑇𝑓𝑚 /𝑇 ∗ = 𝑓(𝑥) = − 𝑏 = − =−
𝑥 (𝑟 − 𝑟𝑚 ) 𝑏 𝑟 𝑏 (2)
(𝑟 − 1)
𝑚

where 𝑎 and 𝑏 are fitting parameters that reflect the influence of matrix effect on freezing
𝑎
point of pore water, 𝑇 ∗ = 1 ℃, and 𝑚 = 𝑟 𝑏 . The utilization of 𝑇 ∗ aims to nondimensionalize
𝑚

the value of temperature. Hence, Eq. (1) can be rewritten as follows.


1/𝑏 1/𝑏
𝑟 𝑎/𝑟𝑚 𝑏 𝑚
=( ) +1=( ) +1 (3)
𝑟𝑚 −∆𝑇𝑓𝑚 /𝑇 ∗ −∆𝑇𝑓𝑚 /𝑇 ∗

The gravimetric content of unfrozen water 𝑤𝑢 is then calculated by


4
𝜋(𝑟 3 − 𝑟𝑚 3 ) (𝑟/𝑟𝑚 )3 − 1
𝑤𝑢 = 3 × 100% = × 100% (4)
4 3×𝐺 𝐺𝑠
𝜋𝑟
3 𝑚 𝑠

where 𝐺𝑠 is the specific gravity of soil grain. Substituting Eqs. (1) and (3) into Eq. (4), the
gravimetric unfrozen water content can be calculated by

1 3
1 −
(𝑚𝑏 × (−∆𝑇𝑓𝑚 /𝑇 ∗ ) 𝑏 + 1) − 1
𝑤𝑢 = × 100%
𝐺𝑠
(5)
1 3
1 −
(𝑚𝑏 × (−𝑇𝑓 /𝑇 ∗ ) 𝑏 + 1) − 1
= × 100%
𝐺𝑠

As per Zhou et al. (2020), 𝑇𝑓 can be replaced with 𝑇 at phase equilibrium state. Thus,
Eq. (5) is rewritten as follows.
1 1 3
(𝑚𝑏 × (−𝑇/𝑇 ∗ )−𝑏 + 1) − 1
(6)
𝑤𝑢 = ∗ 100%
𝐺𝑠

©2020 American Geophysical Union. All rights reserved.


3.2 Saline soil

For saline soil, the freezing point of pore water is affected by the matrix effect and the
osmotic effect, where the osmotic effect represents the impact of solute. Considering the
influence of matrix and osmotic effects, the freezing point 𝑇𝑓 is calculated by (Xiao et al., 2018;
Xu et al., 1995; Zhou et al., 2020)

𝑇𝑓 = 𝑇𝑓∗ + ∆𝑇𝑓𝑚 + ∆𝑇𝑓𝑠 (7)


where ∆𝑇𝑓𝑠 is the depression induced by solute. Fig.5 shows a conceptual diagram of phase
compositions for saline soil.

For a soil containing a single type of soluble salt, its initial water content and salt content
are 𝑤0 and 𝑠0 . The initial mass ratio of salt to water is 𝑐0 = 𝑠0 /𝑤0 ∗ 100%. At a given
temperature 𝑇, the solute concentration 𝑐 is expressed as follows.
𝑠𝑢
𝑐= ∗ 100% (8)
𝑤𝑢

where 𝑠𝑢 is the dissolved salt content, [%], and 𝑤𝑢 the liquid or unfrozen water content [%].
The freezing point depression induced by solute is correlated with the concentration of solute
(Bing & Ma, 2011; Xu et al., 1995; Zhou et al., 2019). Thus, it is defined as
∆𝑇𝑓𝑠 /𝑇 ∗ = 𝜑(𝑐) (9)

The relationship between freezing point depression induced by solute and concentration
is the ice line of a salt-water system. Given 𝑇𝑓∗ = 0℃, Eq. (7) can be rewritten as follows.

∆𝑇𝑓𝑚 = 𝑇𝑓 − ∆𝑇𝑓𝑠 (10)

Assuming the density of solution is equal to free water, the gravimetric unfrozen water
content is given by substituting Eqs. (9) and (10) into Eq. (5) as follows.
1 3
1 −
∗ 𝑏
(𝑚𝑏 × (−𝑇𝑓 /𝑇 + 𝜑(𝑐)) + 1) − 1 (11)
𝑤𝑢 = × 100%
𝐺𝑠

Similarly to Eqs. (5) and (6), at phase equilibrium state, Eq. (11) is rewritten as follows.

1 3
1 −

(𝑚𝑏 × (−𝑇/𝑇 + 𝜑(𝑐)) 𝑏 + 1) − 1
(12)
𝑤𝑢 = × 100%
𝐺𝑠

©2020 American Geophysical Union. All rights reserved.


As discussed by Steiger (2006), the surface tension of pore solution is influenced by the
solute and so the matrix effect on the freezing point depression is affected by solute
concentration and constituents, As a result, and the fitting parameters 𝑚 and 𝑏 of saline soils,
which are both positive values, reflect the influence of the matrix effect on freezing point
depression at a given pore fluid composition that evolves during the freezing process.

(1) Precipitating stage

For saline soil at the precipitating stage, the unfrozen water content decreases with
decreasing temperature due to the solution-crystal phase transition, and the water-ice phase
transition has not occurred. In this model, the solubility of salt is assumed to be solely governed
by the temperature 𝑇, and the solubility line is expressed as
𝑐𝑠𝑎𝑡 = 𝑔(𝑇/𝑇 ∗ ) (13)

When the soil temperature 𝑇 is higher than its freezing point but lower than its initial
precipitation point 𝑇𝑐0 , which is calculated by 𝑇𝑐0 /𝑇 ∗ = 𝑔−1 (𝑐0 ), the solute will precipitate
from the solution. The initial water and salt content are expressed as follows.
𝑤0 = 𝑤𝑢 + 𝑤𝑐 (14)
𝑠0 = 𝑠𝑢 + 𝑠𝑐 (15)
where 𝑤𝑐 is the water content possessed by salt crystals, and 𝑠𝑐 the salt content possessed by
salt crystals. In the precipitation, the mass ratio of solid solute to solid water in salt crystals is
defined as
𝑠𝑐
𝑘= (16)
𝑤𝑐

Assuming the pore solution is equilibrated, the pore solution is saturated during this
stage. Thus, the pore solution can be calculated as
𝑠𝑢
𝑐= × 100% = 𝑐𝑠𝑎𝑡 (17)
𝑤𝑢

Substituting Eqs. (15), (16), and (17) into Eq. (14), the gravimetric unfrozen water
content in the precipitating stage is calculated as
𝑠0 − 𝑘𝑤0
𝑤𝑢 = × 100% (18)
𝑐𝑠𝑎𝑡 − 𝑘 × 100

(2) Icing stage

For saline soil at the icing stage, the unfrozen water content decreases with decreasing
temperature, but the solute remains dissolved in pore solution since it is unsaturated. The solute
concentration in the icing stage is calculated as follows.

©2020 American Geophysical Union. All rights reserved.


𝑠0
𝑐𝑖 = × 100% (19)
𝑤𝑢

Substituting Eq. (19) into Eq. (11), the unfrozen water content in the icing stage can be
calculated by
1 3
1 −
bi ∗ bi
(𝑚i × (−𝑇𝑓 /𝑇 + 𝜑(𝑐𝑖 )) + 1) − 1 (20)
𝑟 3 − 𝑟𝑚 3
𝑤𝑢 = 3 × 100% = × 100%
𝑟𝑚 × 𝐺𝑠 𝐺𝑠

where 𝑚i and bi are parameters in the icing stage.

At phase equilibrium state, Eq. (20) is rewritten as follows.

1 1 3
∗ −
bi
(𝑚i × (−𝑇/𝑇 + 𝜑(𝑐𝑖 )) bi + 1) − 1
(21)
𝑤𝑢 = × 100%
𝐺𝑠

(3) Icing-precipitating stage

When the temperature is low enough, icing and precipitating will occur simultaneously.
According to the conceptual model, at a given freezing point 𝑇𝑓 , the solute concentration can be
calculated by Eq. (13) since it is saturated and phase equilibrated. As a result, the freezing point
depression induced by the solute is given by Eq. (9), and the depression induced by soil matrix is
determined by Eq. (7). As parameters 𝑚 and 𝑏 are related to the soil matrix and the matrix
effect on the freezing point of the pore solution is also influenced by the solute concentration
(Steiger, 2006), 𝑚 and 𝑏 are determined for given soil texture and solute concentration.
According to the solubility assumption, the solute concentration can be determined for a known
temperature in the icing-precipitating stage, and the state point of pore solution will move along
the solubility line in this stage during freezing (Fig.2). Therefore, for a given soil texture and salt
type, soils with various initial salt and water contents will have the same values of 𝑚 and 𝑏 in
the icing-precipitating stage. As a result, the freezing point depression induced by the matrix
effect and the thickness of unfrozen solution film can be determined at different conditions. This
helps explain the observation that after the occurrence of second phase transition in soil, phase
composition curves of soils with various initial saline and moisture states tend to be very similar
for given soil texture and salt type, and this stage of the curve is termed the residual phase. The
stage of the phase composition curve can be calculated using Eq. (22), in which fitting
parameters are marked as 𝑚r and 𝑏r .

©2020 American Geophysical Union. All rights reserved.


1 3
1 −
br
(mr br × (−𝑇𝑓 /𝑇 ∗ + 𝜑(𝑐𝑠𝑎𝑡 )) + 1) − 1
𝑤𝑢 = × 100%
𝐺𝑠
3
(22)
1
1 −
br
(mr br ∗ (−𝑇𝑓 /𝑇 ∗ + 𝜑 (𝑔(𝑇𝑓 /𝑇 ∗ ))) + 1) − 1
= × 100%
𝐺𝑠
At phase equilibrium state, Eq. (22) can be rewritten as follows.
1 3
1 −
br ∗ ∗ ))) br
(mr × (−𝑇/𝑇 + 𝜑(𝑔(𝑇/𝑇 + 1) − 1 (23)
𝑤𝑢 = × 100%
𝐺𝑠

4 Model analysis

4.1 Non-saline soil

As described by Eq. (2), the fitting parameters 𝑚 and 𝑏 determine the magnitude and
degree of attenuation of the freezing point depression with increasing distance 𝑥, respectively. In
order to study the impact of the fitting parameters on the phase composition characteristics,
various values of 𝑚 and 𝑏 are assigned to Eqs. (2) and (6). Fig.6 plots the calculation results
of the relationship between freezing point depression and fitting parameters. As shown in Fig.6
(a) and (b), 𝑟/𝑟𝑚 is the relative distance from the unfrozen water molecular to the center of
spherical soil grain. The freezing point depression and its rate of attenuation rapidly decrease
with increasing 𝑟/𝑟𝑚 , which indicate that the pore water is more vulnerable to be freezing in
larger pores. As the distance from water molecule to soil grain boundary increases, the freezing
point depression induced by matrix effect gradually decreases, and the freezing point gradually
approaches 0℃, which is the freezing point of free water. As shown in Fig.6 (c) and (d), the
phase composition curve is more sensitive to the value of 𝑏 rather than 𝑚.

4.2 Saline soil

(1) Low initial solute concentration

The initial solute concentration is a key factor in determining the order of water-ice phase
transition and solution-crystal phase transition during cooling, while the matrix effect has only a
minor effect on it. As described above, the water-ice phase transition occurs earlier than the
solution-crystal phase transition if the initial solute concentration is low.

In order to illustrate the phase composition characteristics of saline soil using this
conceptual model, a soil containing 𝑁𝑎𝐶𝑙 is assumed and the phase characteristics of the given

©2020 American Geophysical Union. All rights reserved.


soil are calculated under various initial conditions. Assuming the influence of solute on ∆𝑇𝑓𝑚 is
negligible (Zhou et al., 2020), the values of 𝑚 and 𝑏 are constant along the entire phase
composition curve. Thus,
𝑚 = 𝑚0 = 𝑚𝑖 = 𝑚𝑟 (24)
𝑏 = 𝑏0 = 𝑏𝑖 = 𝑏𝑟 (25)

This assumption assures that the intersection point of Eqs. (23) and (21) (also known as
the second phase transition point) is saturated at the end of icing stage.

As per Rumble et al. (2019), the ice line and solubility line of a 𝑁𝑎𝐶𝑙 − 𝐻2 𝑂 system are
expressed as follows.
5

∆𝑇𝑓𝑠 /𝑇 = 𝜑(𝑐) = ∑ 𝑧𝑗 (100𝑐)𝑗



(26)
𝑗=0
𝑐𝑠𝑎𝑡 = 𝑔(𝑇/𝑇 ∗ ) = (0.2503 × 𝑇/𝑇 ∗ + 35.66) × 100% (27)

where 𝑧0 = 0, 𝑧1 = -0.5697, 𝑧2 = -1E-3, 𝑧3 = -1E-4, 𝑧4 = 2E-6, and 𝑧5 = -8E-8.

Below 0.1 ℃, 𝑁𝑎𝐶𝑙 (aq) will react with 𝐻2 𝑂 (aq) and generate 𝑁𝑎𝐶𝑙 ∙ 2𝐻2 𝑂 (s) as
precipitation when the concentration of 𝑁𝑎𝐶𝑙 exceeds its solubility in aqueous solution
(Rumble et al., 2019). The reaction is expressed as follows.
𝑁𝑎𝐶𝑙 (𝑎𝑞) + 2𝐻2 𝑂 (𝑎𝑞) → 𝑁𝑎𝐶𝑙 ∙ 2𝐻2 𝑂 (𝑠) (28)

Thus, the ratio of 𝑁𝑎𝐶𝑙 to 𝐻2 𝑂 in crystal, 𝑘, is calculated by


𝑀(𝑁𝑎𝐶𝑙)
𝑘= = 1.625 (29)
2𝑀(𝐻2 𝑂)

where 𝑀(𝑁𝑎𝐶𝑙) the molar mass of 𝑁𝑎𝐶𝑙, and 𝑀(𝐻2 𝑂) the molar mass of 𝐻2 𝑂.

Assuming 𝐺𝑠 = 2.8, 𝑚 = 4𝐸 − 2 and 𝑏 = 2, and substituting Eqs. (24), (25), (26),


(27), and (29) into Eqs. (23) and (21), the phase composition characteristics of the assumed soil
under various initial solute concentrations and water contents are calculated and then plotted in
Fig.7. As shown in Fig.7 (a) and (b), the phase composition curve of saline soils possesses two
characteristic points, including the initial freezing point and second phase transition point. While
the phase composition curve of salt-free soil (𝑐0 = 0%, 𝑤0 = 30%) only has one characteristic
point, which is the initial freezing point. The initial freezing point separates the icing stage from
the unfrozen stage, while the second phase transition point separates the icing-precipitating stage
from the icing stage.

Above the initial freezing point, there is no unfrozen water loss or concentration of the
pore solution (Fig.7 (c) and (d)). In the icing stage, the pore solution is concentrated due to the

©2020 American Geophysical Union. All rights reserved.


solvent loss (Fig.7 (c) and (d)), and the osmotic effect on freezing point depression is
strengthened (Fig.7 (e) and (f)). The state point of pore solution gets farther from the ice line of
𝑁𝑎𝐶𝑙 − 𝐻2 𝑂 system due to the diminishment of the unfrozen solution film and increase of
matrix effect during freezing. The concentrated pore solution and diminished unfrozen water film
thickness inhibit the rate of unfrozen water decrease due to the decrease in temperature. The pore
water begins freezing at a lower initial freezing point but at a higher second phase transition
point with increasing initial solute concentration 𝑐0 , while the initial freezing point and second
phase transition point both decrease with decreasing initial water content 𝑤0 . These predicted
effects are supported by the experimental data from several sources (Bing & Ma, 2011; Xiao et
al., 2018; Xu et al., 2001; Zhang, 1991; Zhou et al., 2020). At the end of the icing stage, the state
point of the pore solution reaches the solubility line and then moves along it in the icing-
precipitating stage during cooling. Therefore, the solute concentration and ∆𝑇𝑓𝑠 drop with
decreasing temperature and unfrozen water content in this stage (Fig.7 (e) and (f)). After
reaching a second phase transition point, phase composition curves with various initial
concentrations of pore solution and water contents will be very similar and fall on the residual
phase composition curve, which is also called as residual curve for short in this study.

(2) High initial solute concentration

In case of a high initial solute concentration, the precipitating stage occurs earlier than the
icing-precipitating stage. Soil containing 𝑁𝑎2 𝑆𝑂4 is selected as the research object.

In ideal dilute solution, the freezing point depression induced by solute of a salt-water
system is nearly linear correlated with concentration (Banin & Anderson, 1974; Zhou et al.,
2020). As per Tanaka et al. (1992) and Wan et al. (2015), the ice line and solubility line of a
𝑁𝑎2 𝑆𝑂4 − 𝐻2 𝑂 system are expressed as follows.
∆𝑇𝑓𝑠 /𝑇 ∗ = 𝜑(𝑐) = −30𝑐 (30)
−2504.1
7.89 × 10^ ( + 6.9336)
∗) 𝑇/𝑇 ∗ + 273.16
𝑐𝑠𝑎𝑡 = 𝑔(𝑇/𝑇 = × 100% (31)
−2504.1
1 − 10^ ( + 6.9336)
𝑇/𝑇 ∗ + 273.16

The precipitation reaction of 𝑁𝑎2 𝑆𝑂4 in aqueous solution generates 𝑁𝑎2 𝑆𝑂4 ∙ 10𝐻2 𝑂
and is expressed as follows.
𝑁𝑎2 𝑆𝑂4 (𝑎𝑞) + 10𝐻2 𝑂 (𝑎𝑞) → 𝑁𝑎2 𝑆𝑂4 ∙ 10𝐻2 𝑂 (𝑠) (32)

Thus, the ratio of 𝑁𝑎2 𝑆𝑂4 to 𝐻2 𝑂 in crystal is calculated by


𝑀(𝑁𝑎2 𝑆𝑂4 )
𝑘= = 0.789 (33)
10𝑀(𝐻2 𝑂)

©2020 American Geophysical Union. All rights reserved.


Assuming 𝐺𝑠 = 2.8, 𝑚 = 4𝐸 − 2 and 𝑏 = 2, and substituting Eqs. (24), (25), (30),
(31), and (33) into Eqs. (23) and (21), the phase composition characteristics of the assumed soil
with high initial solute concentration are calculated and plotted in Fig.8. As shown in Fig.8 (a)
and (b), the phase composition curve of soil with high initial solute concentration exhibits two
characteristic points, including the initial precipitating point and the second phase transition
point. In the precipitating stage, the solubility of 𝑁𝑎2 𝑆𝑂4 in the pore solution and the freezing
point depression induced by the osmotic effect, ∆𝑇𝑓𝑠 , decrease with decreasing temperature
(Fig.8 (c), (d), (e), and (f)). The initial precipitation point is solely governed by the solute
concentration, and, the paths of state point movement of soils with the same initial solute
concentration will be identical (Fig.8 (c) and (d)) during cooling. Soils containing a higher initial
solute concentration will have a higher initial precipitating point and a lower second phase
transition point, while the second phase transition point increases with increasing initial water
content.

5 Model validation and discussion

5.1 Non-saline soil

Experimental results of four types of soils (bentonite, silty clay, kaolin clay, and sandy
clay) are cited from Akimov (1978) and fitted by this model for validation. For non-saline soils,
the freezing point of pore water is solely governed by the matrix effect and Eq. (5) is used to fit
the phase composition data. The performance and robustness of this model was evaluated by
comparing these results to the results from three other widely used models. These models
include:

(1) Anderson and Tice (1973)


𝑤𝑢 = 1.299𝑆𝑎 0.552 (−𝑇)−1.449𝑆𝑎^(−0.264) (34)

where 𝑆𝑎 is the specific surface area of soil grains [𝑚2 /𝑔].

(2) Michalowski and Zhu (2006)

𝑤𝑢 = 𝑤𝑟 + (𝑤0 − 𝑤𝑟 )𝑒 𝑎(𝑇𝑓0 −𝑇) (35)


where 𝑤𝑟 is the unfreezable water content, and 𝑎 the fitting parameter.

(3) Kozlowski (2007)


𝑇𝑓0 −𝑇 0.37
[−3.35(
𝑇−𝑇𝑚
) ] (36)
𝑤𝑢 = 𝑤𝑟∗ + (𝑤0 − 𝑤𝑟∗ )𝑒 , 𝑇𝑚 < 𝑇 < 𝑇𝑓0
where 𝑇𝑚 = −12℃, and 𝑤𝑟∗ = 0.042𝑆𝑎 + 3.

As for non-linear regression, 𝑅𝑛𝑒𝑤 is used to evaluate the goodness of fitting and
expressed as follows.

©2020 American Geophysical Union. All rights reserved.


∑(𝑦̂𝑖 − 𝑦𝑖 )2
𝑅𝑛𝑒𝑤 = 1 − √ (37)
∑ 𝑦𝑖 2
where 𝑦̂𝑖 denotes the predicted value, and 𝑦𝑖 the measured value.

The parameters used for these models are listed in Table 1, and Fig.9 plots the
calculated/fitted curves. The models proposed by Anderson and Tice (1973) and Kozlowski
(2007) are prediction models, while the model proposed by Michalowski and Zhu (2006) is a
fitting model. The values of 𝑅𝑛𝑒𝑤 obtained by this model are all higher than 0.96 for all the four
tested soils, indicating a high performance in fitting. According to the comparison results, the
performance of this model is obviously higher than other three models. Therefore, as a two-
parameter model, this conceptual model is feasible and convenient to fit the phase composition
curves of non-saline soils.

5.2 Saline soil

(1) Low initial solute concentration

Experimental data for three soils containing 𝑁𝑎𝐶𝑙 (Lanzhou sand, Inner Mongolia clay,
and Lanzhou silt) were used for model validation, and these data were obtained by a Praxis PR-
103 Nuclear Magnetic Resonance analyzer (Xu et al., 2001; Zhang, 1991). Since there are no
other models that consider the second phase transition of saline soil, no performance comparison
among the models was possible. The experimental data ranging in the icing stage are fitted by
Eq. (21), while the experimental data ranging in the icing-precipitating stage except the rapid
phase transition zone are fitted by Eq. (23) as the model of residual curve is an equilibrium state
based model. The fitting parameters and 𝑅𝑛𝑒𝑤 are summarized in Table 2, and the fitting results
are plotted in Fig.10. As shown in Table 2 and Fig.10, the model fits experimental results very
well at the icing stage, and almost all the values of 𝑅𝑛𝑒𝑤 are higher than 0.96. At the icing-
precipitating stage, the residual curve also fits the measured data well except in the rapid phase
transition zone, and the values of 𝑅𝑛𝑒𝑤 are above 0.84. However, our model is developed based
on the equilibrated state of pore solution, and the real state of pore solution may be
supersaturated in the rapid phase transition zone. There is a deviation between the predicted and
measured second phase transition point, where the predicted point is the cross point of the fitted
curve at the icing stage and the residual curve at the icing-precipitating stage. The predicted
temperature of the second phase transition point is 0.96 ℃ higher than the measured temperature
on average. The deviation may be caused by the insufficient degree of supersaturation to drive
the salt crystallization at the predicted second phase transition point.

The phase composition curves of various initial states tend to be very similar in the icing-
precipitating stage, which indicates that the pore solutions are in the same state at a given
temperature. Although there are not enough data to establish that the solute concentrations of

©2020 American Geophysical Union. All rights reserved.


saline soil in the icing-precipitating stage follow the solubility line, good fitting results are
obtained by Eq. (23), which suggests that the assumption is reasonable.

As mentioned above, the fitting parameters 𝑚 and 𝑏 are influenced by the initial state
of saline soil in the icing stage. As shown in Fig.11, if the values of 𝑚 and 𝑏 are assumed to be
consistent among different stages and the values of salt-free state are used in the calculation, the
calculated results show some deviation from the experimental data. This indicates that the
influence of solute concentration on the values of fitting parameters cannot be ignored in some
cases. However, the correlation between the fitting parameters and the solute concentration is
still unknown due to insufficient experimental data.

As shown in Fig.12, the experimental data of solute concentration generally exhibits a


similar pattern to that in Fig.7 (c) and (d), which supports the usefulness of this model in the
icing stage. The experimental solute concentration slightly exceeds the solubility line at the end
of icing stage due to supersaturation. According to Jie (2019), the spontaneous formation of salt
crystals is initiated and driven by a sufficient degree of supersaturation in a solution system.
Thus, the experimental second phase transition point is postponed.

(2) High initial solute concentration

As mentioned above, the precipitating stage occurs before the icing-precipitating stage
for soils with high initial solute concentrations. Experimental data for Lanzhou silt containing
𝑁𝑎2 𝑆𝑂4 are taken from Xu et al. (2001) by a Praxis PR-103 Nuclear Magnetic Resonance
analyzer. The phase composition curves are calculated by Eq. (18) in the precipitating stage,
while the phase composition curves are fitted by Eq. (23) in the icing-precipitating stage. As
shown in Fig.13 (a), the calculated data generally agree with the measured data in the
precipitating stage with 𝑅𝑛𝑒𝑤 ≥ 0.9, and the residual curve shows a good agreement with the
experimental data in the icing-precipitating stage with 𝑅𝑛𝑒𝑤 = 0.94. The predicted temperature
of the first phase transition point is 6.91 ℃ higher than the measured point on average, and the
predicted unfrozen water content of the second phase transition point is 1.03% lower than the
measured point on average, where the predicted second phase transition point is the cross point
of the calculated curve at the precipitating stage and the residual curve. The measured data
indicate that the pore solution is supersaturated in the precipitating stage, which may be the
reason for the deviation of the modeled data from the experimental data. In addition, the
deviation may also be due to the unfrozen pore solutions of samples not being equilibrated
during analysis (Fig.13 (b)).

6 Concluding remarks

This study presents a conceptual model for predicting the phase composition curve for
saline soils. This model simplifies the pore water to be a water film covering the soil grain, and
the freezing point depression induced by the matrix effect is assumed to be a power function of

©2020 American Geophysical Union. All rights reserved.


the distance between water molecule and soil grain boundary. The model analyzes and discusses
the correlation between soil grain and pore solution during freezing from a new perspective
which is different from other models (e.g. Liu and Yu (2013) and (Wang et al., 2017)) using the
capillary freezing theory. The model is based on a power function, which is similar to Anderson
and Tice (1972). However, this model is more intuitive, which is capable of interpreting the
freezing behavior of saline soil during cooling, especially the second phase transition. The model
can easily be applied in the calculation of phase composition curves without requiring the pore
size distribution data or grain size distribution data. Some conclusions are drawn as follows.

(1) The initial solute concentration significantly influences the development of the phase
composition curve of soil due to the water-ice phase transition. For non-saline soils, an icing
stage occurs in the evolution of the phase composition curve during cooling. For saline soils with
a low initial solute concentration in the pore solution, the icing stage occurs initially due to the
water-ice phase transition, followed by the icing-precipitating stage due to the solution-crystal
phase transition. For saline soils with a high initial solute concentration, the precipitating stage is
subsequently followed by the icing-precipitating stage.

(2) The ice line of a salt-water-soil system is lower than it of a salt-water system, and the
influence of matrix effect on the ice line of saline soil increases with decreasing unfrozen water
film thickness.

(3) Phase composition curves of a given soil texture containing different initial water
contents and solute concentrations tend to become very similar in the icing-precipitating stage.
This experimentally observed phenomenon indicates the same correlation between solute
concentration and temperature for saline soils with various initial conditions in the icing-
precipitating stage. Although there is no experimental datum directly proving the assumption that
the correlation of pore solution and temperature follows the solubility line of a salt-water system
in the icing-precipitating stage, the assumption is observed in the model results. Further study
should focus on the measurement of the solute concentration of pore solution to support or
modify the assumption.

(4) Solute characteristics have an impact on the matrix effect, and the impact is
significant in some cases. However, the detailed mechanism is still uncertain due to insufficient
experimental data. The relationship between the fitting parameters (m and b) and the soil texture
is still unclear. More laboratory tests should be conducted to clarify the relationship in the future.

(5) The model performs well in the calculation of the phase composition curve in the
icing stage and icing-precipitating stage for soil containing a single type of soluble salt, while the
occurrence of supersaturation of pore solution would degrade the performance of the model since
the model assumes the components are in an equilibrium state. Since natural soils may contain
various types of salts, further study can be conducted based on this work.

©2020 American Geophysical Union. All rights reserved.


Nomenclature

The following symbols are used in this paper.


𝑎, 𝑏, and 𝑚 = Fitting parameters
𝑏𝑖 and 𝑚i = Fitting parameters at the icing stage
𝑏r and 𝑚r = Fitting parameters for the residual curve
𝑐 = Solute concentration, [%]
𝑐0 = Initial mass ratio of salt to water, [%]
𝑐𝑖 = The solute concentration in the icing stage, [%]
𝑐𝑠𝑎𝑡 = Saturated concentration of pore solution, [%]
Function for the depression of freezing point induced by the soil
𝑓(𝑥) =
matrix
𝑔(𝑇/𝑇 ∗ ) = Function for the solubility line, [%]
𝐺𝑠 = Specific gravity of soil grain
𝑘 = Mass ratio of solid solute to solid water in salt crystals
Distance from the center of soil grain to the interface between
𝑟 =
ice and liquid water
𝑟𝑚 = Radius of the soil grain
𝑠0 = initial salt content, [%]
𝑠𝑐 = salt content possessed by salt crystals, [%]
𝑠𝑢 = dissolved salt content, [%]
𝑇 = Temperature, [℃]
𝑇𝑐0 = Initial precipitation point of pore solution, [℃]
𝑇𝑓 = Freezing point of pore water, [℃]
𝑇𝑓0 = Initial freezing point of pore solution, [℃]
𝑇𝑠 = Second phase transition point of pore solution, [℃]
𝑇∗ = Unit degree Celsius, [℃]
𝑇𝑓∗ = Freezing point of pure water at a reference state, [℃]

𝑤0 = initial water content, [%]


𝑤𝑐 = water content possessed by salt crystals, [%]
𝑤𝑢 = Unfrozen water content, [%]
𝑥 = Thickness of the unfrozen water film
Calculation parameters for the ice line of a 𝑁𝑎𝐶𝑙 − 𝐻2 𝑂
𝑧𝑗 =
system
Depression of freezing point induced by the soil matrix effect,
∆𝑇𝑓𝑚 =
[℃]
∆𝑇𝑓𝑠 = Depression of freezing point induced by the solute, [℃]
𝜑(𝑐) = Function for the ice line

©2020 American Geophysical Union. All rights reserved.


Acknowledgments
This work was supported by the Natural Science Foundation of China (Grant Nos. 41731281,
41801055, 41901073), the Technology Research and Development Program of China Railway
(Grant No. 2017G002-S), and the State Key Laboratory of Frozen Soil Engineering (No.
SKLFSE201706). Datasets for this research are included in these papers: [Akimov, 1978], [Xu et
al., 2001], and [Zhang, 1991].

References
Akimov, Y. P. (1978). Sravnitelnaya ocenka metodov opredelenya soderzanya niezamierzszey
vody v merzlyh gruntah (Assessmentof methods to determine the unfrozen water content in
frozen soils, in Russian). Merzlotnye Issledovania, 17, 190–195.
Andersland, O. B., & Ladanyi, B. (2004). Frozen ground engineering (2nd). Hoboken, N.J.,
Chichester: Wiley.
Anderson, D. M., & Morgenstern, N. R. (1973). Physics, chemistry and mechanics of frozen
ground: A review. In North American Contribution to the Proceedings of the Second
International Conference on Permafrost.
Anderson, D. M., & Tice, A. R. (1972). Predicting unfrozen water contents in frozen soils from
surface area measurements. Highway Research Record, 393, 12–18.
Anderson, D. M., & Tice, A. R. (1973). The Unfrozen Interfacial Phase in Frozen Soil Water
Systems. In A. Hadas, D. Swartzendruber, P. E. Rijtema, M. Fuchs, & B. Yaron (Eds.),
Ecological Studies, Analysis and Synthesis, 0070-8356: Vol. 4. Physical Aspects of Soil Water
and Salts in Ecosystems (Vol. 4, pp. 107–124). Berlin, Heidelberg: Springer.
https://doi.org/10.1007/978-3-642-65523-4_12
Banin, A., & Anderson, D. M. (1974). Effects of Salt Concentration Changes During Freezing on
the Unfrozen Water Content of Porous Materials. Water Resources Research, 10(1), 124–128.
https://doi.org/10.1029/WR010i001p00124
Bing, H., & Ma, W. (2011). Laboratory investigation of the freezing point of saline soil. Cold
Regions Science and Technology, 67(1-2), 79–88.
https://doi.org/10.1016/j.coldregions.2011.02.008
Bouyoucos, G. J., & McCool, M. M. (1916). Freezing Point Method As a New Means of
Measuring the Concentration of the Soil Solution Directly in the Soil. Tech. Bul., 24.
Chai, M., Zhang, J., Zhang, H., Mu, Y., Sun, G., & Yin, Z. (2018). A method for calculating
unfrozen water content of silty clay with consideration of freezing point. Applied Clay Science,
161, 474–481. https://doi.org/10.1016/j.clay.2018.05.015
Chen, X., Liu, J., Liu, H., & Wang, Y. (2006). Frost action of soil and foundation engineering.
Beijing: Science Press.

©2020 American Geophysical Union. All rights reserved.


Dillon, H. B., & Andersland, O. B. (1966). Predicting Unfrozen Water Contents in Frozen Soils.
Canadian Geotechnical Journal, 3(2), 53–60. https://doi.org/10.1139/t66-007
Gao, B., Yang, D., Qin, Y., Wang, Y., Li, H., Zhang, Y., & Zhang, T. (2018). Change in frozen
soils and its effect on regional hydrology, upper Heihe basin, northeastern Qinghai–Tibetan
Plateau. The Cryosphere, 12(2), 657–673. https://doi.org/10.5194/tc-12-657-2018
Grosse, G., Jones, B., & Arp, C. (2013). Thermokarst Lakes, Drainage, and Drained Basins. In J.
F. Shroder (Ed.), Treatise on geomorphology (pp. 325–353). London, Waltham, MA: Academic
Press. https://doi.org/10.1016/B978-0-12-374739-6.00216-5
Guo, L., Xie, Y., Yu, Q., You, Y., Wang, X., & Li, X. (2016). Displacements of tower foundations
in permafrost regions along the Qinghai–Tibet Power Transmission Line. Cold Regions Science
and Technology, 121, 187–195. https://doi.org/10.1016/j.coldregions.2015.07.012
Han, Y., Wang, Q., Kong, Y., Cheng, S., Wang, J., Zhang, X., & Wang, N. (2018). Experiments
on the initial freezing point of dispersive saline soil. CATENA, 171, 681–690.
https://doi.org/10.1016/J.CATENA.2018.07.046
Harlan, R. L. (1973). Analysis of coupled heat-fluid transport in partially frozen soil. Water
Resources Research, 9(5), 1314–1323. https://doi.org/10.1029/WR009i005p01314
Hayashi, M., van der Kamp, G., & Schmidt, R. (2003). Focused infiltration of snowmelt water in
partially frozen soil under small depressions. Journal of Hydrology, 270(3-4), 214–229.
https://doi.org/10.1016/S0022-1694(02)00287-1
Jie, W. (2019). Principle and Technology of Crystal Growth. Beijing: Science Press. Retrieved
from http://find.nlc.cn/search/showDocDetails?docId=-
3697057150522872852&dataSource=ucs01
Kozlowski, T. (2007). A semi-empirical model for phase composition of water in clay–water
systems. Cold Regions Science and Technology, 49(3), 226–236.
https://doi.org/10.1016/j.coldregions.2007.03.013
Kozlowski, T. (2009). Some factors affecting supercooling and the equilibrium freezing point in
soil–water systems. Cold Regions Science and Technology, 59(1), 25–33.
https://doi.org/10.1016/j.coldregions.2009.05.009
Li, Z., Chen, J., & Sugimoto, M. (2020). Pulsed NMR Measurements of Unfrozen Water Content
in Partially Frozen Soil. Journal of Cold Regions Engineering, 34(3), 4020013.
https://doi.org/10.1061/(ASCE)CR.1943-5495.0000220
Liu, J., Wang, T., Tai, B., & Lv, P. (2020). A method for frost jacking prediction of single pile in
permafrost. Acta Geotechnica, 15(2), 455–470. https://doi.org/10.1007/s11440-018-0711-0
Liu, Z., & Yu, X. (2013). Physically Based Equation for Phase Composition Curve of Frozen
Soils. Transportation Research Record: Journal of the Transportation Research Board, 2349(1),
93–99. https://doi.org/10.3141/2349-11

©2020 American Geophysical Union. All rights reserved.


Lovell, C. W. (1957). TEMPERATURE EFFECTS ON PHASE COMPOSITION AND
STRENGTH OF PARTIALLY-FROZEN SOIL. Highway Research Board Bulletin. (168).
Retrieved from https://trid.trb.org/view/128123
Low, P. F., Hoekstra, P., & Anderson, D. M. (1968). Some Thermodynamic Relationships for
Soils at or Below the Freezing Point: 2. Effects of Temperature and Pressure on Unfrozen Soil
Water. Water Resources Research, 4(3), 541–544. https://doi.org/10.1029/WR004i003p00541
Lu, N. (2020). Unsaturated Soil Mechanics: Fundamental Challenges, Breakthroughs, and
Opportunities. Journal of Geotechnical and Geoenvironmental Engineering, 146(5), 2520001.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0002233
Michalowski, R. L., & Zhu, M. (2006). Frost heave modelling using porosity rate function.
International Journal for Numerical and Analytical Methods in Geomechanics, 30(8), 703–722.
https://doi.org/10.1002/nag.497
Moiwo, J. P., Yang, Y., Tao, F., Lu, W., & Han, S. (2011). Water storage change in the Himalayas
from the Gravity Recovery and Climate Experiment (GRACE) and an empirical climate model.
Water Resources Research, 47(7). https://doi.org/10.1029/2010WR010157
Nagare, R. M., Schincariol, R. A., Quinton, W. L., & Hayashi, M. (2012). Effects of freezing on
soil temperature, freezing front propagation and moisture redistribution in peat: laboratory
investigations. Hydrology and Earth System Sciences, 16(2), 501–515.
https://doi.org/10.5194/hess-16-501-2012
Pronk, P. (2006). Fluidized bed heat exchangers to prevent fouling in ice slurry systems and
industrial crystallizers. Delft University of Technology, Netherlands. Retrieved from
http://resolver.tudelft.nl/uuid:e2cf9ec0-fedc-4480-a8ab-ddf7a3a7afea
Qi, J., Wang, S., & Yu, F. (2013). A Review on Creep of Frozen Soils. In Q. Yang (Ed.), Springer
Series in Geomechanics and Geoengineering. Constitutive modeling of geomaterials: Advances
and new applications (pp. 129–133). Heidelberg, New York: Springer.
https://doi.org/10.1007/978-3-642-32814-5_13
Rumble, J. R., Lide, D. R., & Bruno, T. J. (2019). CRC Handbook of chemistry and physics: A
ready-referance book of chemical and physical data (100th edition). Boca Raton: Taylor &
Francis Group.
Scherer, G. W. (1999). Crystallization in pores. Cement and Concrete Research, 29(8), 1347–
1358. https://doi.org/10.1016/S0008-8846(99)00002-2
Spaans, E. J. A., & Baker, J. M. (1996). The Soil Freezing Characteristic: Its Measurement and
Similarity to the Soil Moisture Characteristic. Soil Science Society of America Journal, 60(1),
13–19. https://doi.org/10.2136/sssaj1996.03615995006000010005x
Steiger, M. (2006). FREEZING OF SALT SOLUTIONS IN SMALL PORES. In S. P. Shah & M.
S. Konsta-Gdoutos (Eds.), Measuring, monitoring and modeling concrete properties (pp. 661–
668). Amsterdam: Springer. https://doi.org/10.1007/978-1-4020-5104-3_80

©2020 American Geophysical Union. All rights reserved.


Steiger, M., & Asmussen, S. (2008). Crystallization of sodium sulfate phases in porous materials:
The phase diagram Na2SO4–H2O and the generation of stress. Geochimica Et Cosmochimica
Acta, 72(17), 4291–4306. https://doi.org/10.1016/j.gca.2008.05.053
Tai, B., Liu, J., & Chang, D. (2020). Experimental and numerical investigation on the sunny-
shady slopes effect of three cooling embankments along an expressway in warm permafrost
region, China. Engineering Geology, 269, 105545. https://doi.org/10.1016/j.enggeo.2020.105545
Tai, B., Liu, J., Yue, Z., Liu, J., Tian, Y., & Wang, T. (2018). Effect of sunny-shady slopes and
strike on thermal regime of subgrade along a high-speed railway in cold regions, China.
Engineering Geology, 232, 182–191. https://doi.org/10.1016/j.enggeo.2017.09.002
Tanaka, Y., Hada, S., Makita, T., & Moritoki, M. (1992). Effect of pressure on the solid-liquid
phase equilibria in (water + sodium sulfate) system. Fluid Phase Equilibria, 76, 163–173.
https://doi.org/10.1016/0378-3812(92)85085-M
Taylor, G. S., & Luthin, J. N. (1978). A model for coupled heat and moisture transfer during soil
freezing. Canadian Geotechnical Journal, 15(4), 548–555. https://doi.org/10.1139/t78-058
Teng, J., Kou, J., Yan, X., Zhang, S., & Sheng, D. (2020). Parameterization of soil freezing
characteristic curve for unsaturated soils. Cold Regions Science and Technology, 170, 102928.
https://doi.org/10.1016/j.coldregions.2019.102928
Wan, X., Hu, Q., & Liao, M. (2017). Salt crystallization in cold sulfate saline soil. Cold Regions
Science and Technology, 137, 36–47. https://doi.org/10.1016/j.coldregions.2017.02.007
Wan, X., Lai, Y., & Wang, C. (2015). Experimental Study on the Freezing Temperatures of
Saline Silty Soils. Permafrost and Periglacial Processes, 26(2), 175–187.
https://doi.org/10.1002/ppp.1837
Wan, X., & Yang, Z. (2020). Pore water freezing characteristic in saline soils based on pore size
distribution. Cold Regions Science and Technology, 173, 103030.
https://doi.org/10.1016/j.coldregions.2020.103030
Wang, C., Lai, Y., & Zhang, M. (2017). Estimating soil freezing characteristic curve based on
pore-size distribution. Applied Thermal Engineering, 124, 1049–1060.
https://doi.org/10.1016/j.applthermaleng.2017.06.006
Wang, L., Liu, J., Feng, R., Zhang, X., & Liu, Z. (2020). Experimental Study on Salt Expansion
Characteristics of Coarse-Grained Sulfate Soils. Journal of Cold Regions Engineering, 34(2),
4020004. https://doi.org/10.1061/(ASCE)CR.1943-5495.0000210
Watanabe, K., & Wake, T. (2009). Measurement of unfrozen water content and relative
permittivity of frozen unsaturated soil using NMR and TDR. Cold Regions Science and
Technology, 59(1), 34–41. https://doi.org/10.1016/j.coldregions.2009.05.011
Wu, D., Lai, Y., & Zhang, M. (2015). Heat and mass transfer effects of ice growth mechanisms in
a fully saturated soil. International Journal of Heat and Mass Transfer, 86, 699–709.

©2020 American Geophysical Union. All rights reserved.


https://doi.org/10.1016/j.ijheatmasstransfer.2015.03.044
Xiao, Z., Lai, Y., & Zhang, M. (2018). Study on the freezing temperature of saline soil. Acta
Geotechnica, 13(1), 195–205. https://doi.org/10.1007/s11440-017-0537-1
Xu, X., Wang, J., & Zhang, L. (2001). Frozen soil physics. Beijing: Science Press. Retrieved
from
http://find.nlc.cn/search/showDocDetails?docId=3316943824434124485&dataSource=ucs01
Xu, X., Zhang, L., & Wang, J. (1995). Establishment of model for predicting unfrozen water
content in saline soils. Crysphere, 57(1).
Xu, Y. (1993). Saline Soil Foundation. Beijing: China Architecture & Building Press. Retrieved
from http://find.nlc.cn/search/showDocDetails?docId=500246379725247007&dataSource=ucs01
Yong, R. N., Cheung, C. H., & Sheeran, D. E. (1979). Prediction of salt influence on unfrozen
water content in frozen soils. Engineering Geology, 13(1-4), 137–155.
https://doi.org/10.1016/0013-7952(79)90027-9
Yu, X., Zhang, B., Liu, N., & Yu, X. (2010). Comparison Study of Three Common Technologies
for Freezing-Thawing Measurement. Advances in Civil Engineering, 2010, 1–10.
https://doi.org/10.1155/2010/239651
Zhang, L. (1991). The law of unfrozen water content change in frozen saline (NaCl) soils.
International Symposium on Ground Freezing, 113-119. Retrieved from https://pascal-
francis.inist.fr/vibad/index.php?action=getRecordDetail&idt=6544918
Zhang, T., Barry, R. G., Knowles, K., Heginbottom, J. A., & Brown, J. (1999). Statistics and
characteristics of permafrost and ground‐ice distribution in the Northern Hemisphere 1. Polar
Geography, 23(2), 132–154. https://doi.org/10.1080/10889379909377670
Zhang, T., Barry, R. G., Knowles, K., Ling, F., & Armstrong, R. L. (2003). Distribution of
seasonally and perennially frozen ground in the Northern Hemisphere. In Proceedings of the 8th
International Conference on Permafrost. Zürich, Switzerland: AA Balkema Publishers Zürich.
Zhou, J., Liang, W., & Wei, C. (2019). Phase Equilibrium Condition for Pore Hydrate:
Theoretical Formulation and Experimental Validation. Journal of Geophysical Research: Solid
Earth, 124(12), 12703–12721. https://doi.org/10.1029/2019JB018518
Zhou, J., Meng, X., Wei, C., & Pei, W. (2020). Unified soil freezing characteristic for variably
saturated and saline soils. Water Resources Research. Advance online publication.
https://doi.org/10.1029/2019WR026648
Zhou, J., Wei, C., Lai, Y., Wei, H., & Tian, H. (2018). Application of the Generalized Clapeyron
Equation to Freezing Point Depression and Unfrozen Water Content. Water Resources Research,
54(11), 9412–9431. https://doi.org/10.1029/2018WR023221

©2020 American Geophysical Union. All rights reserved.


Fig.1. A typical phase composition curve of non-saline soil.

©2020 American Geophysical Union. All rights reserved.


Fig.2. A typical phase diagram of salt-water system.

©2020 American Geophysical Union. All rights reserved.


Fig.3. Typical phase composition curves for saline soil with different initial solute
concentrations. (a) Low initial solute concentration. The water-ice phase transition occurs earlier
than the solution-crystal phase transition during cooling. (b) High initial concentration. The
water-ice phase transition occurs later than the solution-crystal phase transition during cooling.

©2020 American Geophysical Union. All rights reserved.


Fig.4. A conceptual model diagram for non-saline soil. The soil grain matrix is assumed to be a
homogeneous sphere and covered by unfrozen water film. At the interface between the ice and
unfrozen water film, the freezing point of pore water is depressed by the soil matrix only.

©2020 American Geophysical Union. All rights reserved.


Fig.5. A conceptual model diagram for saline soil. The soil grain matrix is assumed to be a
homogeneous sphere and covered by unfrozen solution film. At the interface between the
ice/precipitation and unfrozen solution film, the freezing point of pore water is depressed by both
the soil matrix and the solute.

©2020 American Geophysical Union. All rights reserved.


Fig.6. Correlation between fitting parameters and phase composition characteristics of non-saline
soil. The influences of the magnitudes of 𝑚 (Figure (a)) and 𝑏 (Figure (b)) on the variation of
∆𝑇𝑓𝑚 vs. 𝑟/𝑟𝑚 are calculated by Eq. (2). The influences of the magnitudes of 𝑚 (Figure (c))
and 𝑏 (Figure (d)) on the variation of unfrozen water content vs. temperature are calculated by
Eq. (6).

©2020 American Geophysical Union. All rights reserved.


Fig.7. Phase composition characteristics of soils with low initial solute concentrations. Curves in
red plot the calculated results of assumed soil with 𝑁𝑎𝐶𝑙 varying in the initial solute
concentration (𝐺𝑠 = 2.8, 𝑤0 = 30%, 𝑚 = 4𝐸 − 2, and 𝑏 = 2), while blue curves show the
results of assumed soil with 𝑁𝑎𝐶𝑙 varying in the initial water content (𝐺𝑠 = 2.8, 𝑐0 = 15%,
𝑚 = 4𝐸 − 2, and 𝑏 = 2). Figures (a) and (b) show the variation of unfrozen water content vs.
temperature using Eqs. (21) and (23). Figures (c) and (d) show the movement of state point of
the pore solution system during cooling using Eqs. (19) and (27). Figures (e) and (f) show the
correlation between the unfrozen water content and freezing point depression using Eqs. (2) and
(26).

©2020 American Geophysical Union. All rights reserved.


Fig.8. Phase composition characteristics of soils with high initial solute concentrations. Curves in
red plot the calculated results of assumed soil with 𝑁𝑎2 𝑆𝑂4 varying in the initial solute
concentration (𝐺𝑠 = 2.8, 𝑤0 = 30%, 𝑚 = 4𝐸 − 2, and 𝑏 = 2), while blue curves show the
results of assumed soil with 𝑁𝑎2 𝑆𝑂4 varying in the initial water content (𝐺𝑠 = 2.8, 𝑤0 = 30%,
𝑚 = 4𝐸 − 2, and 𝑏 = 2). Figures (a) and (b) show the variation of unfrozen water content vs.
temperature using Eqs. (18) and (23). Figures (c) and (d) show the movement of state point of
the pore solution system during cooling using Eq. (31). Figures (e) and (f) show the correlation
between unfrozen water content and freezing point depression using Eqs. (2) and (30).

©2020 American Geophysical Union. All rights reserved.


Fig.9. Comparison between various models. (a) Bentonite. (b) Silty clay. (c) Kaolin clay. (d)
Sandy silt.

©2020 American Geophysical Union. All rights reserved.


Fig.10. Comparison between experimental and fitting results of saline soils with 𝑁𝑎𝐶𝑙. (a)
Lanzhou sand varying in initial solute concentration. (b) Lanzhou sand varying in initial water
content. (d) Inner Mongolia clay varying in initial solute concentration. (e) Inner Mongolia clay
varying in initial water content. (f) Lanzhou silt varying in initial solute concentration.

©2020 American Geophysical Union. All rights reserved.


Fig.11. Comparison between experimental and calculated results with constant values of 𝑚 and
𝑏. The agreement between the calculated and measured results seems unsatisfactory.

©2020 American Geophysical Union. All rights reserved.


Fig.12. Correlation between solute concentration and temperature calculated by the measured
data using Eq. (19) in the icing stage. The concentration of pore solution slightly exceeds the
saturated concentration at the end of the icing stage because of the supersaturation of pore
solution.

©2020 American Geophysical Union. All rights reserved.


Fig.13. Phase composition characteristics of Lanzhou silt containing 𝑁𝑎2 𝑆𝑂4. (a) The
relationship between the unfrozen water content and the temperature. (b) The measured
movement of state point of pore solution system during cooling.

©2020 American Geophysical Union. All rights reserved.


Table 1. Parameters of the fitting by the selected models

Anderson and Tice


Our model Michalowski and Zhu (2006) Kozlowski (2007)
(1973)
Soil
𝑇𝑓0 , 𝑤0 , 𝑤𝑟 , 𝑆𝑎 , [𝑚2 / 𝑇𝑓0 , 𝑤0 , 𝑤𝑟∗ ,
𝑚0 𝑏0 𝐺𝑠 𝑅𝑛𝑒𝑤 𝑆𝑎 , [𝑚2 /𝑔] 𝑅𝑛𝑒𝑤 𝑎 𝑅𝑛𝑒𝑤 𝑅𝑛𝑒𝑤
[℃] [%] [%] 𝑔] [℃] [%] [%]

1.46E-
Bentonite 6.59 2.8 0.99 382 0.89 2.02 -0.12 80.8 21.2 0.83 382 -0.12 80.8 19.04 0.91
04

1.41E-
Silty clay 2.95 2.7 0.98 111 0.70 1.32 -0.15 27.1 6 0.86 111 -0.15 27.1 7.66 0.91
03

Kaolin 7.33E-
0.98 2.6 0.97 13 0.54 1.57 -0.15 30.7 0.8 0.92 13 -0.15 30.7 3.55 0.66
clay 02

2.95E-
Sandy silt 2.77 2.63 0.96 68 -0.09 2.48 -0.15 17.7 3.8 0.93 68 -0.15 17.7 5.86 0.51
04

©2020 American Geophysical Union. All rights reserved.


Table 2. Parameters of fitting for saline soils at various stages

Icing-precipitating stage
Icing stage (except the rapid phase
Soil 𝐺𝑠 𝑤0 , [%] 𝑐0 , [%] transition zone)

𝑚𝑖 𝑏𝑖 𝑅𝑛𝑒𝑤 𝑚𝑟 𝑏𝑟 𝑅𝑛𝑒𝑤

9.4 0 2.65E-04 2.16 0.88 –* – –

9.4 5 2.30E-02 1.52 0.96 – – –

9.4 10 4.19E-02 1.47 0.96

Lanzhou sand 2.63 9.4 15 1.93E-03 2.64 0.96

9.4 20 2.39E-01 1.26 1.00 1.20E-06 3.02 0.84

4.54 15 1.21 0.29 0.98

7.25 15 3.94E-02 1.35 0.98

21.19 0 3.15E-02 2.06 0.98 – – –

21.19 5 4.37E-02 2.06 0.98 – – –

21.19 10 3.62E-02 2.21 0.98 – – –

12.18 20 4.06E-02 2.02 0.99 – – –

Inner Mongolia
2.72 16.23 20 6.52E-03 2.83 0.99 – – –
clay

21.19 15 6.80E-03 2.94 0.97

21.19 20 2.00E-04 4.38 0.99


2.48E-02 2.15 0.97
30.53 20 3.78E-02 1.78 0.99

34.39 20 8.33E-01 0.14 1.00

Lanzhou silt 2.84 15.22 0 2.06E-02 1.56 0.96 – – –

Copyright © 2020 John Wiley & Sons, Ltd.


15.22 5 3.50E-03 2.22 0.96 – – –

15.22 10 2.26E-03 2.59 0.97

15.22 15 3.01E-03 2.77 0.99


2.06E-02 1.56 0.96
15.22 20 9.19E-05 3.95 0.99

15.22 25 9.19E-05 3.95 0.99

Note:
*
– denotes the unavailability of data.

©2020 American Geophysical Union. All rights reserved.

You might also like