Study on the Adsorption Characteristics of Loess Influenced by Temperature Effects

Abstract

Loess, a typical unsaturated soil, is a Quaternary sedimentary deposit widely distributed across arid and semi-arid regions worldwide. In recent years, global climate change has led to significant temperature fluctuations in Northwest China, impacting loess properties and soil–water characteristic curves (SWCCs). This study investigated typical loess deposits in Mizhi County, Shaanxi Province, systematically analyzing their basic physical properties and microstructure. The SWCCs of the loess were measured at three temperature gradients (15 °C, 20 °C, and 25 °C) using the dynamic dew-point isotherm method to investigate the impact of temperature on SWCC hysteresis. The results showed that with increasing temperature, the SWCC exhibited increasing divergence. The magnitude of the water content change and the corresponding suction forces along the wetting and drying paths increased, leading to an enlargement of the hysteresis loop area. These findings indicate that temperature significantly affects the hysteresis behavior of loess, providing a certain basis and ideas for the study of the soil–water characteristic curves of unsaturated soils such as loess under the influence of temperature.

1. Introduction

Loess, a distinctive Quaternary sediment, is widely distributed in arid and semi-arid regions globally. China’s Loess Plateau is particularly notable, hosting the world’s thickest and most concentrated loess deposits and providing abundant material for research. Loess is a typical unsaturated soil characterized by unique physical properties, including a macroporous structure, weak interparticle cementation, and sensitivity to water content—particularly its collapsibility and water sensitivity [1]. These properties, along with well-developed vertical joints, control its macroscopic behavior, resulting in complex and unique engineering responses under hydraulic loading that differ significantly from other soil types, such as uneven foundation settlement, expansion of shrinkage cracks, and slope instability [2,3].

Soil adsorption has attracted much attention in unsaturated soil mechanics. A soil–water characteristic curve describes the constitutive relationship between soil matrix suction and water content. The hysteresis effect refers to the phenomenon where the relationship between moisture content and suction differs during drying and wetting processes. Specifically, under identical suction conditions, the moisture content on the drying path is higher than that on the wetting path. In the study of soil adsorption and the hysteresis effect, different scholars have conducted corresponding research work based on different models to study soil adsorption and the hysteresis effect. Some models focus on capillary water holding behavior [4,5,6], and the water holding capacity of soil under high suction has also attracted much attention. Many scholars have carried out tests related to suction. For example, the filter paper method is used to determine the matrix suction on the shear surface of undrained shear samples. Suction makes a significant contribution to the shear strength of expansive soil, mainly because it increases the effective stress between soil particles [7]. In addition, a triaxial shear test using expansive soil to control the suction has been carried out, leading to the conclusion that the shear strength of unsaturated plastic expansive soil increases with the increase in suction [8]. Some scholars have found that the liquid permeability coefficient is related to the matrix suction [9]. With the development of cities and towns, the heat conduction of underground pipe networks has a significant impact on the temperature field around the soil mass and also has a certain impact on the clay area [10,11]. At present, the determination of matrix suction is mainly based on the soil–water characteristic curve measured by specific soil samples, which reflects the change of matrix suction due to water content changes to a certain extent and also calibrates the water content under different suction conditions [12,13,14,15,16].

Temperature has a certain influence on matric suction [17]. Philip and de Vries initially studied the effect of temperature on the soil–water characteristic curve (SWCC) of unsaturated soils. Based on the Laplace equation, they assumed that temperature changes only affect surface tension and provided an expression and model for how suction changes with temperature [18]. Subsequently, Gardner tested the suction of various soils under varying temperature conditions. The results showed that in some soils, such as sand, matric suction decreases with increasing temperature [19]. Chahal tested the effect of temperature on matric suction under air-tight conditions and used thermodynamics to explain how the influence of thermal expansion on air might dominate the direction of suction changes [20,21]. Haridasan and Jensen tested two silty loams using both steady-state method and pressure plate outflow method. The results showed that the pressure plate outflow test effectively captured the effect of temperature on suction, while the steady-state test exhibited excessive variations, making it impossible to determine a direct effect of temperature on suction [22]. Hopmans and Dane demonstrated that the temperature dependence of the SWCC is primarily driven by changes in gas–liquid interfacial tension [23]. Nimmo and Miller first quantified the temperature dependence of the soil water hysteresis curve, identifying surfactant dynamics as the core mechanism by which temperature effects in natural soils far exceed theoretical predictions [24]. Constantz’s experiments show that changes in air pressure and temperature can lead to changes in matric suction [25]. She and Sleep [26], Bachmann [27], Wang [28], and others have conducted extensive research on the effects of temperature. Their results indicate that the influence of temperature on unsaturated soil suction cannot be attributed solely to the effect on surface tension. Grant and Salehzadeh modified the formula proposed by Philip and de Vries to include the effect of temperature on the wetting coefficient [29]. Their results showed that when both the effects of surface tension and the wetting coefficient on temperature are considered, the experimental results converge with the formula, but the specific relationship remains unestablished.

Recent years have witnessed an exacerbation of global climate change, with frequent extreme heat events drawing attention to the impact of temperature on geo-engineering bodies. Under global warming, the influence of temperature on the characteristics of unsaturated loess has become increasingly significant. According to the research on climate and temperature changes in Northwest China, the average temperature in Northwest China has increased at an average rate of 0.2 °C/10 a. The average temperature in Shaanxi Province is about 20–25 °C, which also provides a certain reference for the selection of the temperature gradient in this study [30]. Temperature variations can alter the interactions between loess particles, the physical properties of pore water, and moisture migration patterns, thereby significantly affecting hydraulic behavior. The specific mechanisms by which temperature influences the SWCC remain unclear and lack consensus within the field. Understanding the precise role of temperature in SWCC variations is an ongoing subject of discussion and research. This knowledge gap poses new challenges for geotechnical engineering design, construction, and long-term stability assessments, representing a critical scientific issue requiring in-depth investigation. Addressing this problem is essential not only for a more accurate understanding of loess engineering properties but also for providing a more scientific and reliable theoretical basis and technical support for related engineering projects.

This study investigated loess sourced from Mizhi County, Shaanxi Province, situated within the Loess Plateau. Utilizing a multi-scale analytical approach, we characterized the samples through conventional geotechnical property testing, scanning electron microscopy (SEM), X-ray diffraction (XRD), and X-ray fluorescence (XRF). The hydraulic behavior of the loess was specifically examined using the dynamic dew-point isotherm method under three distinct temperature gradients (15 °C, 20 °C, and 25 °C). The primary objectives were to quantify temperature-induced variations in the soil–water characteristic curves (SWCCs) and to assess the influence of temperature on the hysteresis effect observed between the wetting and drying paths.

2. The Study Area

Mizhi County is situated in the core area of the Northern Shaanxi Loess Plateau, within the middle reaches of the Wuding River, and possesses unique geographical characteristics. The region exhibits typical loess ridge–hill landforms. It lies between 34°30′ N to 36°30′ N and 104°30′ E to 107°30′ E. The geological structure is relatively simple, with low tectonic activity. The degree of crustal uplift in the western part of Mizhi County is significantly higher than in the east, resulting in an overall terrain sloping from west to east and north to south. Geomorphologically, the northwest is dominated by loess tablelands and loess hills, while the southeast primarily features loess ridges and hills. The region is dominated by fluvial and rainfall erosion, leading to extensive gully development, fragmented topography, generally steep slopes, and frequent geohazards. The increasing frequency of extreme climate events in recent years, coupled with these factors, provides the necessary topographic conditions and material basis for geohazard occurrence (Figure 1).

3. Materials and Methods

3.1. Materials

The study area is Mizhi County in Northern Shaanxi. Samples were collected from a loess profile (37°53.28′ N, 110°14.38′ E) at different elevations (H₁: 965.5 m, H₂: 955.2 m, H₃: 945.7 m, H₄: 935.9 m) to analyze the basic characteristics of undisturbed Mizhi loess from macro-, micro-, and compositional perspectives. Macroscopically, the loess exhibits a layered structure with paleosol layers of varying thicknesses interbedded, ranging in color from light yellow to brownish-yellow, reflecting environmental changes during deposition and influencing physical property distribution. Microscopically, loess particles display diverse morphologies (clay, silt, sand) with distinct pore structures; the distribution and size of these pores influence subsequent SWCCs and adsorption behavior. Compositionally, variations in mineral content affect the sensitivity to water and temperature, further impacting soil–water characteristics and adsorption effects.

3.2. Methods

According to the relevant characteristics of unsaturated soil, the following experimental scheme (Figure 2) was developed to measure the basic physical properties and mineral composition of loess. Finally, the adsorption characteristics were measured under different temperatures and the soil–water characteristic curve and hysteresis effect were explained and evaluated. The experimental protocol began with characterization of fundamental loess properties—particle size distribution, bulk density, and gradation characteristics—followed by comprehensive mineralogical analysis to quantify compositional abundance while eliminating confounding factors through standardized pretreatment and rigorous temperature control. Subsequently, vapor sorption analysis employing dynamic dew-point isotherm methodology was conducted to derive soil–water characteristic curves (SWCCs) under controlled thermal regimes.

3.2.1. Laser Diffraction Particle Size Analyzer Analyzes the Sample

Employing a coherent laser source at a fixed monochromatic wavelength, the angular distribution of diffracted and scattered radiant energy is governed exclusively by particle diameter. For polydisperse particulate systems, the population density of discrete size classes determines the radiant flux density detected at each scattering angle. The normalized radiant energy fraction at each angular position correlates directly with the relative abundance of corresponding particle size fractions. These fundamental principles were operationalized through laser diffraction particle analysis to quantitatively characterize the particle size distribution and particle content of undisturbed loess specimens from Mizhi County.

3.2.2. Compositional Analysis Using XRF and XRD

Mineral composition was analyzed using a Bruker S8 TIGER Series 2 X-ray Fluorescence Spectrometer (XRF) (Brooke AXS Company, Karlsruhe, Germany) [12,31] and a Bruker D8 Advance X-ray Diffractometer (XRD) (Brooke AXS Company, Karlsruhe, Germany) [32,33]. XRF enables simultaneous multi-element concentration analysis across diverse sample forms (powders, granules, liquids, thin films). Elemental quantification derives from the characteristic energy/intensity of X-ray fluorescence emitted under high-energy excitation. XRD determines loess mineral composition (clay minerals, quartz, feldspar) and the relative abundances, which is critical for elucidating loess provenance, evolution, and physical properties.

3.2.3. Microstructural Analysis Using Phenom Scanning Electron Microscope (SEM)

Microstructural characterization was conducted using a scanning electron microscope (SEM) (Zeptools, Shanghai, China). Imaging was performed at 500× magnification to systematically record particle morphologies (angular, platy, blocky, rod-like) and spatial arrangements. Pore network architecture was categorized by size and distribution patterns. Microstructural discontinuities were assessed, including desiccation cracks and stress-induced fissures.

3.2.4. Dynamic Dew-Point Isotherm Method

Soil–water characteristic curves (SWCCs) under high-suction conditions were determined for each loess specimen using a vapor sorption analyzer (VSA) (Aqualab, Pullman, WA, USA) via the dynamic dew-point isotherm method. By precisely controlling temperature and relative humidity (RH) within the sample chamber, gravimetric measurements of adsorption/desorption processes enabled calculation of corresponding matric suction values. During testing, samples were equilibrated in the chamber while RH was systematically adjusted in incremental steps. Mass variations at each RH setpoint were recorded, with subsequent error analysis eliminating outliers to derive validated SWCCs.

4. Results

4.1. Determination of Basic Physical Properties

Basic physical properties of the loess samples (collected via drilling, stored in sealed plastic tubes, and carefully transported to minimize disturbance) were determined using a laser particle size analyzer and a liquid–plastic limit combination tester. Properties measured included dry density, void ratio, specific gravity, Atterberg limits, grain size distribution (clay, silt, sand), uniformity coefficient, curvature coefficient, and specific surface area. Data for the four samples (H₁–H₄) are presented in

Table 1. Basic physical properties of undisturbed Mizhi loess samples.

Type		H1	H2	H3	H4
Depth (m)		965.5	955.2	945.7	935.9
Dry density (ρd, g/cm3)
Porosity ratio (e)
Specific gravity (G)
Atterberg limits (%)	(WL, %)	28.2	33.67	29.3	27.37
	(WP, %)	15.9	15.63	14.2	11.93
	(IP, %)	12.3	18.04	15.1	15.44
Atterberg limits (%)	(Clay, %)	6.98	5.43	4.92	5.26
	(Silt, %)	50.47	51.69	38.31	33.35
	(Sand, %)	42.55	42.88	56.77	61.39
Cu		8.68	6.46	6.83	7.44
Cc		2.16	1.75	1.64	2.16
SSA (m2/g)		0.521	0.396	0.410	0.417

.

Analysis showed that the samples are predominantly sandy loess (sand content: 42.55–61.39%, average 50.90%). Grain size distribution curves (Figure 3) generally exhibited bimodal characteristics for all layers, with peaks around 1–10 μm and 50–100 μm and a plateau around 1–2 μm, indicating fewer particles in that range. The median particle size was 23.4 μm, suggesting a wide but relatively well-sorted particle distribution. Compared to other areas of the Loess Plateau, the clay content was slightly higher, potentially due to source material differences and weathering. Grain size composition showed little vertical variation, indicating relatively stable depositional conditions.

Based on the analysis of particle size distribution and the resulting grain size classification triangle for Mizhi loess (Figure 4), the following observations are made:

(1)

For samples from the H₁ and H₂ strata, the sand content is less than 50%, while the silt content exceeds the sand content. This significantly affects pore connectivity and pore size distribution, which in turn has substantial implications for subsequent research on the soil–water characteristic curve (SWCC). When the pore size is smaller, the suction required for both the wetting and drying paths will correspondingly decrease and increase.

(2)

In contrast, undisturbed samples from the H₃ and H₄ strata exhibit a sand content exceeding 50%. This provides favorable conditions for the formation of a soil skeleton structure and creates inherent conditions for larger pores. Meanwhile, the clay content (approximately 25%) is nearly equivalent to the silt content (approximately 25%). During SWCC studies, adsorption in later stages tends to form smaller pores, requiring increased suction and making the formation of interconnected pathways more challenging.

(3)

The analysis of the grain size classification triangle for Mizhi loess depends not only on variations in the data but also on the sedimentary environment and depositional time. Notably, the deeper H₃ and H₄ Mizhi loess samples have been subjected to greater overburden pressure and have experienced distinct sedimentary environments. Consequently, pore sizes have also decreased, aligning with the characteristics of the sedimentary environment and the depositional sequence features of the Loess Plateau.

4.2. Sample Mineral Composition and Microstructure

Microstructure was examined using a Phenom SEM at magnifications of 500× (Figure 5). Observations revealed particles with diverse morphologies (angular, platy, blocky, rod-like) and varying arrangements (random to semi-ordered). (a), (b), (c), and (d) are SEM scanning images of loess samples from Mizhi County. From the scanning images, it can be seen that with the deepening of the sampling position, the pore size between the particles gradually decreases and the frequency and proportion of large pores also gradually decrease. The maximum pore diameter is about 50 microns, and the minimum is about 2 microns. These data will help us discuss the influencing factors after the subsequent measurement of the SWCCs.

The pore structure was complex, comprising large, medium, and small pores of different shapes and distributions. At 500× magnification, angular detrital particles dominated, with significant large inter-particle pores. Micro-cracks and fissures were also observed, likely formed by external stresses during loess formation and playing an important role in pore connectivity. These microstructural features significantly influence the physical and hydraulic properties of loess, particularly the wetting and drying processes under high suction.

XRF analysis (

Table 2. XRF chemical composition of Mizhi loess.

Sample Number	Average Content of Main Chemical Components (%)
	SiO2	Al2O3	Fe2O3	FeO	CaO	MgO	K2O	Na2O	TiO2	P2O5	MnO
H1	64.06	11.08	3.16	1.39	7.77	2.13	2.4	1.67	0.65	0.14	0.09
H2	63.61	11.07	1.32	6.37	2.25	2.14	1.91	0.59	0.13	0.066	3.75
H3	64.67	10.92	1.34	6.42	2.08	2.09	2.04	0.56	0.12	0.065	3.52
H4	65.45	11.02	1.20	6.03	1.97	2.17	2.07	0.56	0.14	0.067	3.50

) showed SiO₂ and Al₂O₃ as the predominant components, indicating quartz and feldspar as major constituents. According to Table 2, the average proportion of SiO₂ in loess at different depths is about 64.45% and the average proportion of Al₂O₃ is about 11.02%. The deviations of other elements are relatively small. For MnO, the sample at position H₁ has the lowest content. Judging from the local climatic conditions and surrounding environment, it is believed that the low MnO content in the surface soil is caused by surface runoff and rainfall erosion. The SiO₂/Al₂O₃ ratio (silica modulus, Ki) suggests varying degrees of oxidation during diagenesis, with samples H3 and H₄ (deeper) exhibiting slightly higher Ki values than samples H₁ and H₂ (shallower), indicating stronger chemical weathering under relatively humid conditions for H₁/H₂. Depletion of mobile elements (K, Na, Ca) in deeper layers suggests slow leaching of Ca²⁺ by groundwater. The Fe₂O₃ content exceeded the FeO content, indicating active oxidation at the sampled depths due to oxygenated precipitation.

XRD analysis of detrital minerals (

Table 3. Representative XRD pattern of Mizhi loess clay fraction.

Sample Number	Mineral Content (%)
	Quartz	Plagioclase	Potassium Feldspar	Calcite	Dolomite	Amphibole	Hematite	Amorphous Phase	TCCM
H1	43.8	18.5	5.9	7.9	0.5	2.3	/	/	21.1
H2	51.0	17.4	4.0	6.7	1.1	1.2	/	/	18.6
H3	50.4	20.5	6.2	8.0	1.2	0.9	/	/	12.8
H4	47.1	25.3	5.3	6.7	1.3	1.6	/	/	12.7

) revealed a diverse assemblage (“multi-mineralic”) including quartz (43.8–51.0%), plagioclase (total Plagioclase: 17.4–25.3%), calcite, dolomite (light minerals), and heavy minerals like Amphibole.

Stable and moderately stable opaque minerals were most abundant among the heavy minerals, while unstable minerals like hornblende decreased with depth and stable minerals increased, reflecting climate and weathering conditions during deposition. The XRD results show that the loess samples as a whole record a climate transition event from a warm and humid weathering period to an arid and high-energy sedimentation period, accompanied by fluctuations in the structure of the provenance area or the composition of the parent rock.

4.3. Determination of Soil–Water Characteristic Curves (SWCCs) Using Dynamic Dew-Point Isotherm Method

SWCCs at high suction ranges were obtained using a vapor sorption analyzer (VSA) based on the dynamic dew-point isotherm method (weight-based principle) [34,35,36]. Samples were placed in a temperature-controlled chamber and relative humidity (RH) was systematically varied. Mass changes during adsorption (wetting) and desorption (drying) cycles were recorded, allowing for the calculation of matrix suction. Tests were conducted at 15 °C, 20 °C, and 25 °C. Data points were subjected to error analysis, and outliers were removed before plotting the SWCCs [37]. The suction and moisture content in the process of moisture absorption and dehumidification are shown in

Table 4. Moisture content and suction under three temperature conditions.

Temperature (°C)	Experimental Time (Min)	Start Point of Moisture Absorption			Dehumidification Starting Point			Dehumidification End Point
		RH	$\mathit{w}$ (%)	$\mathit{\psi }$ (kPa)	RH	$\mathit{w}$ (%)	$\mathit{\psi }$ (kPa)	RH	$\mathit{w}$ (%)	$\mathit{\psi }$ (kPa)
15 °C	1485 (196)	0.0300	0.05	482,931	0.9509	5.41	6933	0.0163	0.01	566,622
20 °C	1959 (203)	0.0512	0.02	402,201	0.9473	5.56	7328	0.0253	0.01	497,532
25 °C	1651 (193)	0.0539	0.01	391,779	0.9456	5.68	7459	0.0857	0.02	328,758

Note: The end point of moisture absorption is the start point of dehumidification.

. According to the experimental data (Table 4), at the end of the dehumidification paths at 15 °C and 20 °C, the moisture content decreased compared to the starting point of the moisture absorption path, while the required suction force increased. At 25 °C, due to equipment limitations, the lowest moisture content in the dehumidification path was 0.02%. Comparing the suction force at the end of the dehumidification path at different temperatures reveals that the required matrix suction force decreases with increasing temperature, which also leads to an increase in the area enclosed by the moisture absorption and dehumidification paths, demonstrating the significant influence of reaction temperature on the hysteresis effect.

4.4. SWCC Characteristics Under Different Temperatures

The SWCC data for wetting and drying paths at 15 °C, 20 °C, and 25 °C exhibited systematic temperature dependence, displaying an “S” shape in a semi-logarithmic coordinate system. The data are plotted according to the contents in Table 4, and the results are shown as follows:

(1)

Temperature condition 15 °C

Water content decreased with increasing suction during drying and increased with decreasing suction during wetting. An inflection point occurred near RH = 0.8 (water content ~4–5%). Slope increased after the inflection point on the wetting path and decreased on the drying path, see Figure 6.

(2)

Temperature condition 20 °C

Similar “S” shape to 15 °C, but the inflection point shifted (occurring within RH = 0.8–1.0). The initial water content on the drying path was higher, and a more distinct separation between the wetting and drying paths was evident, see Figure 7.

(3)

Temperature condition 25 °C

Increased complexity was observed. The drying path exhibited a trend towards dual inflection points within RH = 0.8–1.0; the first inflection point shifted leftwards. After the first inflection point, the drying rate increased; it then decreased after the second point. The wetting path showed a more pronounced increase in rate at a low RH. Path separation was significantly more pronounced. The initial suction point on the drying path shifted leftwards, and the final water content was lower than the initial wetting point water content at the same level of suction, see Figure 8.

Increasing the temperature caused distinct separation of the wetting and drying paths, increased the range of water content change and the required suction along both paths, and altered the shape and position of inflection points. The drying path at 25 °C showed particularly complex behavior.

5. Discussion

5.1. Influence of Temperature on Hysteresis

The hysteresis loop area was calculated for each temperature condition (Figure 9). The results showed a clear increase with rising temperature:

Soil–water characteristic curves (SWCCs) measured at 15 °C, 20 °C, and 25 °C are presented in panels (a), (b), and (c), respectively. Experimental data points were fitted with continuous curves, and hysteresis loop areas were quantified through numerical integration. The results demonstrate a progressive increase in the hysteresis loop area with rising temperature, confirming the pronounced temperature dependence of hysteresis behavior in loess.

5.2. Role of Mineral Composition and Microstructure

Temperature increases enhance the activity of water molecules, accelerating diffusion and migration rates and leading to higher desorption rates at lower water contents. The clay mineral composition (dominated by illite, with kaolinite, montmorillonite, chlorite) forms microscopic, interconnected channels and pore spaces. Temperature variations affect the mobility of water molecules within these confined spaces. SEM observations confirmed the presence of small channels formed by clay aggregates. During wetting, the influx of numerous water molecules requires migration within the sample. Higher temperatures accelerate molecular motion but potentially also the filling of pores, requiring greater suction for external water molecules to enter the soil matrix, as reflected in the path separation.

At 15 °C, suction at the wetting start point was higher than at the drying end point. The residual water content was higher, indicating stronger water retention capacity at lower temperatures. At 20 °C and especially 25 °C, suction at the drying end point became significantly higher than at the wetting start point, and the final water content on drying was lower than the initial water content on wetting. This indicates faster vapor movement within pores under warmer conditions and stronger adsorption (binding) of water by the loess at 25 °C, which is less favorable for water retention.

5.3. SWCC Model Fitting

Based on the soil–water characteristic curves (SWCCs) measured under different temperatures, we fitted the data of the desorption path using the Van Genuchten model [6] (Equation (1)).

$${\displaystyle \frac{\theta -{\theta }_r}{{\theta }_s-{\theta }_r}}={\left[{\displaystyle \frac{1}{1+{\left(\alpha \psi \right)}^n}}\right]}^m$$

(1)

The fitting results are shown in Figure 10. The parameters of the curve at 15 °C are a = 2.29, n = 1.32, m = 0.12, and R² = 0.87; at 20 °C are a = 6.94, n = 0.91, m = 0.26, and R² = 0.83; and at 25 °C are a = 7.21, n = 0.67, m = 0.71, and R2 = 0.81. It is evident that as the temperature increases, parameters a and m show upward trends, while parameter n exhibits a downward trend. A downward bending occurs in the latter segment of the curve, which the author attributes to potential data acquisition issues or suboptimal fitting due to temperature effects.

Building on this fitting analysis, the author statistically examined the correlation between temperature and water content. The relationship is quantified in Equation (2). Using 20 °C as the baseline, water content values at 15 °C and 25 °C were calculated to infer the impact of temperature on the hysteretic behavior of loess. This approach provides a quantitative description of the relationship between temperature and water content.

$${\omega }_T={\omega }_{20}\times \left[1-0.03\times \left(T-20\right)\right]$$

(2)

6. Conclusions

The unique physical properties of loess govern its response to temperature and moisture under varying conditions. This study investigated undisturbed loess from Mizhi County, characterized by a stable depositional environment. Samples collected at different elevations underwent analysis of their basic physical properties, microstructures, and mineral compositions. SWCCs were determined using the dynamic dew-point isotherm method under three temperature conditions (15 °C, 20 °C, 25 °C), and the influence of temperature on hysteresis was analyzed. The key findings are

(1)

Temperature-Driven SWCC Modifications

Elevated temperatures induce progressive divergence between wetting and drying paths, amplify water content variations, and necessitate higher suction thresholds for moisture transitions. Distinct dual-inflection behavior emerges on the drying path at 25 °C, signifying fundamental alterations in moisture retention dynamics.

(2)

Hysteresis Intensification

The quantified hysteresis loop area nearly triples from 76 kPa (15 °C) to 215 kPa (25 °C). Concurrently, the water content on the drying path exceeds that at lower temperatures under equivalent suction, demonstrating temperature-dependent hydraulic irreversibility.

(3)

Microstructural Control Mechanism

Clay mineral assemblages (illite/kaolinite/montmorillonite/chlorite) form nanoconfined channels governing water mobility. Temperature accelerates molecular migration within these conduits while simultaneously strengthening water–matrix interactions during drying and weakening pore-filling efficiency during wetting, collectively explaining hysteresis enhancement.