WorksPhysics and Astronomy
The receding contact line cools down during dynamic wetting
Abstract
When a contact line (CL) – where a liquid-vapor interface meets a substrate – is put into motion, it is well known that the contact angle differs between advancing and receding CLs. Using non-equilibrium molecular dynamics simulations, we reveal another intriguing distinction between advancing and receding CLs: while temperature increases at an advancing CL – as expected from viscous dissipation, we show that temperature can drop at a receding CL. Detailed quantitative analysis based on the macroscopic energy balance around the dynamic CL showed that the internal energy change of the fluid along the pathline induced a remarkable temperature drop around the receding CL, in a manner similar to latent heat upon phase changes. This result provides new insights for modeling the dynamic CL, and the framework for heat transport analysis introduced here can be applied to a wide range of nanofluidic systems.
Introduction
Wettings are ubiquitous in our daily life, in nature and in various scientific and engineering fields. In particular, the behavior of the contact line (CL), where a liquid-vapor interface meets a solid surface, has long been a topic of interest because it plays a key role in wetting properties.[1, 2, 3, 4] For static wetting without CL motion, a common measure of wettability at the macroscopic scale is the contact angle (CA), described by Young’s equation,[5] which was first proposed in 1805 based on a balance between solid-liquid, solid-vapor and liquid-vapor interfacial tensions. These interfacial tensions originate from the microscopic molecular interaction forces, and recent molecular simulation studies have provided significant advance in understanding static wetting.[6, 7] [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24]
The situation is more complex at the dynamic CL (DCL)—appearing typically during droplet spreading or moving on a substrate, where the advancing and receding CAs are different. To model the CA difference, numbers of theoretical, computational and experimental studies about the DCL have been carried out and have indicated that this dynamic effect is induced by the viscosity and friction in the vicinity of the DCL;[1] [25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48] however, the governing principle of the DCL motion still remains unclear, mainly due to the lack of detailed information on the nanoscale thermal and flow fields around the DCL, and it is considered to be one of the long-standing unsolved problems of fluid dynamics.
In this article, we show a unique thermal phenomenon around the DCLs, cooling as well as heating at the DCLs. To elucidate its mechanism, we analyze the heat flow field around the DCL using molecular dynamics (MD) simulations of a quasi-2D system with liquid–solid–vapor CLs, consisting of a Lennard-Jones (LJ) fluid between parallel solid walls moving in opposite directions as shown in the top panel of Figure 1.

Figure 1. Top: quasi-2D Couette-type flow system of a Lennard-Jones liquid confined between two solid walls. Middle: density and velocity distributions calculated by volume average; the macroscopic streaming velocity is denoted by black arrows. Bottom: temperature distribution of the fluid. (Partly from Kusudo, H., Omori, T., Yamaguchi, Y., J. Chem. Phys., 155 (2021), 184103; licensed under CC BY)
To that aim, we have developed a heat transport analysis methodology applicable in multi-component MD systems.
Methodology
Prior to the analysis, we first need to construct a methodology to calculate heat flows based on the method of planes (MoP),49–55 which defines surface-averaged field values on a finite control plane so that obtained values satisfy the continuum conservation laws described by the Reynolds transport theorem for arbitrary control volume (CV) surrounded by finite control planes.56 Specifically, in this article, we extend the formulation proposed for single-component fluid systems by Todd and Davis51 to the heat flow in multi-component systems with a solid wall. Energy conservation in the presence of an external force writes
where , and denote the density, velocity and specific total energy of the fluid—defined by the sum of the specific internal energy and the specific convective kinetic energy , whereas , and denote the heat flux, stress tensor and external force per unit mass, respectively. For this energy conservation law, we treat the fluid–fluid intermolecular interaction force as the stress while we treat the fluid–solid one as the external force.57,58 (1) can be integrated for an arbitrary CV, and by applying Gauss’ theorem to the advection and stress work terms on the right hand side (RHS), one obtains:
meaning that the fluid energy change in the CV in the left hand side (LHS) balances the heat production/absorption, the work of the external body force on the fluid in the CV, and the macroscopic energy advection and stress work through its surrounding surface in the RHS. The divergence of the heat flux term, which corresponds to the heat production/absorption value in the CV, can be rewritten as
meaning that the heat flow from the CV is obtained by integrating the energy change in the CV and the energy advection and stress work on the surface of the CV —obtainable by the MoP, and by integrating the work by the external body force on fluid in the CV. In this article, we calculated the first term in the RHS by integrating the energy flux through the whole surrounding surface of the CV, see detail in the supplementary materials (SM). Note that we adopted the volume-averaged fluid velocity in the second term of the RHS of Eq. (3) because we take its inner product with the body force as the volume-averaged intermolecular force exerted on the fluid by the solid.
System
The top panel of Figure 1 shows the MD simulation system of a quasi-2D Couette-type flow, where the basic setups are the same as in our previous study.56 The fluid–fluid and fluid–solid interactions were modeled by the 12-6 LJ potential
where is the distance between the particles and , while and denotes the LJ energy and length parameters, respectively. Quadratic functions were added to this LJ potential so that the potential and interaction force smoothly vanished at a cut-off distance of .59 We used the following parameters for fluid–fluid (ff) and fluid–solid (fs) interactions: nm, J, nm, J. The atomic masses of fluid and solid particles were u and u, respectively. Finally, the equations of motion were integrated using the velocity-Verlet algorithm, with a time step of 5 fs.
Periodic boundary conditions were set in the - and -directions, and 20,000 LJ particles were confined between two parallel solid walls (dimension of nm) at a distance of nm, so that the LJ fluid formed two quasi-2D menisci with CLs on the walls upon the preliminary equilibration at a control temperature K without shear. The static CA on both top and bottom walls was deg. After the equilibration, further relaxation runs to achieve a steady shear flow with asymmetric menisci were carried out for 10 ns by moving the particles in the outmost layers of both walls with opposite velocities of m/s in the -direction.
After the relaxation run, the main calculation was conducted for an average time of 400 ns. We calculated the external body force and volume averaged velocity in the RHS of Eq. (3) using cuboid bins of size
3.92 nm, while we calculated the energy flux, velocity, stress and the specific energy in the RHS of Eq. (3) using the MoP with the faces of each local bin. Regarding the calculation of the energy flux and the specific energy, see details in SM.
Results and discussion
The middle panel of Figure 1 shows the density distribution and velocity field obtained by the volume average. Due to the shear applied by the wall, a caterpillar-like flow was induced, and DCLs, i.e., advancing and receding CLs with different CAs, appeared. In addition to the CA difference, we showed the stress inhomogeneity in the bulk liquid induced by this flow in our previous study.56 In the present study, we report a distinct thermal difference in the DCLs as shown in the temperature distribution in the bottom panel of Figure 1: temperature rises around the advancing CLs (bottom right and top left), and temperature drops around the receding CLs (bottom right and top left), and temperature drops around the receding CLs (bottom left and top right). Quantitatively, in the bulk liquid away from the interfaces, the temperature is around 86.5 K, which is slightly higher than the control temperature of the wall due to viscous dissipation, whereas that around the advancing CLs is about 2 K higher and that around the receding CLs is about 2 K lower than the bulk, as shown in Figure 1. The cooling at the receding CLs is especially intriguing, because viscous dissipation can only induce temperature rise through heat production.
To elucidate the mechanism of the heat production/absorption around the DCLs, we conducted a heat flux analysis. The top panel of Figure 2 shows the heat flux field superimposed on the temperature distribution. Note that the heat flux field was depicted only for the fluid sufficiently away from the wall, where the effect of potential field from the wall on the fluid was negligibly small; the MoP methodology of the heat flux calculation is shown in SM. Heat flow from the high temperature area to the low temperature area can be observed, meaning that the heat produced around an advancing CL induces temperature rise there, and flows to the cold neighboring receding CL due to the heat absorption. To quantitatively evaluate the heat production/absorption around the DCLs and in the liquid area sufficiently away from the CLs, we set three control volumes (CVs) as shown with magenta rectangles in Figure 2; a CV surrounding the receding CL, a CV surrounding the advancing CL, and a CV between them.

Figure 2. Volume integral of (top) the heat flux divergence, (middle) the inner product of stress tensor and tensor derivative of velocity vector and (bottom) the Lagrange derivative of internal energy on three CVs; left: surrounding the receding CL, right: surrounding the advancing CL, middle: between the left and right CVs. The top panel also shows the distributions of temperature and heat flux field. The bottom panel also shows the internal energy distribution and the velocity field.
We integrated the divergence of the heat flux in each CV and the corresponding values are shown inside the CVs with a unit of mW/m, calculated as the heat production/absorption rate divided by the system depth, in the top panel of Figure 2. According to Figure 2, heat is produced and absorbed at the CVs surrounding the advancing (97 mW/m) and receding ( mW/m) CLs, respectively, and the absolute values are approximately twice as large as the heat production 42 mW/m in the middle CV even though its volume is twice as large as the others. From this, we see that viscous dissipation is not the main cause of heat production/absorption at the DCLs.
To elucidate this, one can rewrite the energy conservation in Eq. (1) as follows (see details in SM):
where denotes the specific internal energy defined by
and and denote the Lagrangian derivative and the inner product of a second order tensor, respectively. Note that the specific internal energy includes not only the fluid–fluid interaction potential but also the fluid–solid one. (4) indicates that the heat production/absorption arises from two mechanisms: 1. the inner product of stress tensor and velocity gradient, which corresponds to the viscous dissipation in bulk; 2. the internal energy change along the pathline (identical to the streamline in a steady flow). The work done by the solid–fluid interaction force seems to be absent from Eq. (4) because it does not contribute to the internal energy change but to the convective kinetic energy change, see details in SM. We rewrite the first term on the RHS of Eq. (4) as
where the momentum conservation with external force is applied for the second equality and macroscopically steady-state system is assumed for the third equality. The rightmost-hand side (HS) can be directly integrated for the CV with MoP by applying Gauss’ theorem while the numerical differentiation is essential to integrate the LHS. The middle panel of Figure 2 shows the integral of this stress term by using Eq. (6), and this indicates that this stress work is the main cause of the heat production in the middle CV without CL, but is not remarkably large in the CV surrounding the DCLs. Note that the stress term contains not only the viscous dissipation but also the work by the pressure or interfacial tensions so that it is not always positive, specifically around the advancing CL. The other factor of the heat production/absorption is the internal energy change in the RHS of Eq. (4), and we rewrite it as
where the mass conservation is applied for the first equality, Eq. (5) is applied for the second equality, and the convective kinetic energy is assumed to be constant over time. Note that the steady state is not assumed for the first term in the rightmost-HS because it depends on the microscopic configuration difference between the start and end time of the sampling interval, and it is not negligibly small especially around the CLs, which are microscopically fluctuating.56 Also for this Eq. (7), the rightmost-HS can be directly integrated for the CV with the MoP by applying Gauss’ theorem, and the integral of the first term in the rightmost-HS is obtained with the energy flux through the whole surrounding surface of the CV, see detail in SM. The integral values of this internal energy change for the CVs are shown in the bottom panel of Figure 2. The large absolute values of −100.2 and 101.9 mW/m in the two CVs indicate that this term is the main cause of the heat production/absorption around the DCLs, which is small in the middle CV without CL.
We also show the internal energy distribution and the velocity field as the background of bottom panel of Figure 2, and one can observe that the internal energy changes along the streamline specifically near the DCLs, where the heat is produced/absorbed. At the advancing CL, heat is produced when the fluid flows from the solid-vapor and liquid-vapor regions to the solid-liquid region, whereas at the receding CL heat is absorbed when the fluid flows from the solid-liquid region to the solid-vapor and liquid-vapor regions. During these processes, the internal energy of the fluid changes due to the surrounding density change as well as due to the potential field induced by the solid surface, and it leads to the cooling and heating at the DCLs. This phenomenon is analogous to latent heat, which induces the heat production/absorption upon the phase change.
Therefore, it is expected that this cooling and heating effect should be increased with the flow rate around the DCLs, i.e., the faster wall speed. Here, we additionally conducted the heat analysis for the CVs with various wall speeds: , 2.5, 5.0, 7.5, and 12.5 m/s (the density and velocity fields and temperature distribution with each condition are shown in SM). Top, middle and bottom panels of Figure 3 show the volume integral values of the heat flux divergence, stress work term in Eq. (6) and internal energy change along the streamline in Eq. (7), respectively. Blue and red ones denote the values of CV including the receding CL (RCL) and the advancing CL (ACL), and green one denotes the CV between them. Note that the CV arrangement for all wall speeds is same as that in Figure 2. Similar to Figure 2, the internal energy change is the main part of heat production/absorption in CVs including DCLs while the stress work term is dominant in the middle CV (referred to as "Bulk" in Figure 3). The internal energy change appears to be proportional to the wall speed, implying that the spatial distributions of the density and the specific energy do not largely change due to the wall speed. On the other hand, the stress work term appears to be proportional to the square of the wall velocity in the middle CV since the shear stress, i.e., viscous stress, is proportional to the shear rate in the bulk where the shear rate can be roughly proportional to the wall velocity. Also in the CVs including the DCLs, the work done by the solid–fluid interaction force in Eq. (6) should largely depend on the wall speed because that frictional force is supposed to be proportional to the slip velocity. Under the present wall velocities where the steady-state caterpillar-like flows with DCLs are achieved, the internal energy change is always dominant over the stress work term in the CVs including DCLs: and thus the temperature rise/drop near the DCLs should always exist. In addition, we observed this cooling/heating phenomena at the DCLs induced by the same mechanism also on less wettable walls as shown in SM. Note that this quasi-latent heat around the DCLs is not a dissipation energy, meaning that it cannot be included in the dissipation terms of existing macroscale DCL models.[1] [42] However, it indeed induces temperature changes in the vicinity of the DCLs, which should be included in the DCL models.

Figure 3. Volume integral values of (top) the heat flux divergence, (middle) the inner product of stress tensor and derivative of velocity vector and (bottom) the Lagrangian derivative of internal energy in Eq. (7) on the three CVs for various wall speed . The locations of the CVs are the same as in Figure 2.
Conclusion
In this article, we have presented a heat transport analysis methodology applicable in multi-component MD systems, which we have used to investigate the heat transport features of the DCL. The heat analysis revealed that heat is not only produced but also absorbed around DCLs, mainly due to the quasi-latent heat induced by the internal energy change of fluid along the pathline, when the fluid moves among the interfaces, which is accompanied by a change in fluid–fluid and fluid–solid interaction energy. In addition, this latent heat is not a dissipation energy, thus almost the same heat is absorbed and produced at receding and advancing CLs, respectively, while heat is only produced in bulk liquid due to viscous dissipation. Overall, these results provide new insights into the molecular mechanisms controlling the dynamics of the CL. Moreover, the framework for analyzing the heat transport at the molecular scale should be useful for investigating various nanoscale systems such as the flow in carbon nanotubes or in nanoporous media.60–67
Supplementary Material
The supplementary material contains the calculation methods of the energy density, energy flux and heat flux by the method of plane and the derivation of the Lagrangian derivative of the internal energy in Eq. (4). We also show therein the density, velocity and temperature distributions with various wall speeds on the lyophilic walls corresponding to Figure 3, and the density and temperature distributions around the DCL on the lyophobic walls.
Acknowledgments
HK, TO and YY are supported by JSPS KAKENHI Grant Nos. JP23KJ0090, JP23H01346 and JP22H01400, respectively. YY is also supported by JST CREST Grant No. JP-MJCR18I1, Japan. Numerical simulations were performed on the Supercomputer system "AFI-NITY" at the Advanced Fluid Information Research Center, Institute of Fluid Science, Tohoku University.
Conflict of Interest Statement
The authors have no conflicts to disclose.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Supplementary Materials – The receding contact line cools down during dynamic wetting
Hiroki Kusudo; Takeshi Omori; Laurent Joly; Yasutaka Yamaguchi
Energy density calculation by the method of plane
The energy density in the right hand side of Eq. (3) can be obtained by applying the method of plane (MoP) formulation shown in our previous study[1] as follows:
where and denote the molecular energy defined by Eq. (10) and the -component of the velocity vector of fluid particle , respectively. The angular bracket denotes the ensemble average, and the summation is taken for every fluid particle passing through within a time interval of , which is replaced by the time increment for the numerical integration.
The energy flux calculation by the method of plane
We calculated the first term in the RHS of Eq. (3) by integrating the energy flux through the whole surrounding surface of the CV, which is essential for the depiction of the heat flux, while the integrated value is technically same as the total molecular energy change in the CV during the simulation. The energy flux in the MoP formulation[2, 3] writes
where denotes the fluid molecular energy defined by
and denotes the relative position of -th molecule from -th. In addition, is the intermolecular energy transport from -th to -th defined by
where denotes the intermolecular force exerted on -th from -th molecule. This is based on the mechanism that molecular kinetic energy changes due to the work done by the intermolecular force. In the velocity-Verlet algorithm, the change of molecular energy at from time writes
where and denote the total intermolecular force exerted on molecule at time and , respectively. We can interpret that the time derivative of kinetic energy at equals the inner product of the force and velocity, which are defined by
and
respectively. To satisfy the relationship of molecular energy transport in the energy flux calculation in Eq. (9), we rewrite Eq. (11) as
where we adopted for the intermolecular force
This assignment of the force and velocity guarantees that the energy change in the control volume (CV) equals to the integral of the energy flux for the whole surface of the CV calculated by Eq. (9).
Heat flux calculation away from the wall
The heat flux field in the top panel of Figure 2 was obtained by using the MoP. The energy conservation without the external force writes
and in this article it holds for the fluid sufficiently away from the wall, where the effect of the potential field from the wall on the fluid was negligibly small. The energy change in the left hand side (LHS) in the control volume (CV) balances the energy flux through the surrounding surface of the CV. Therefore, Eq. (17) writes
where denotes the energy flux. By assuming that the energy flux in the LHS is comprised of the macroscopic advection term, heat flux and stress work terms in the RHS, the heat flux through the bin face with its normal vector pointing to the -th Cartesian direction can be calculated from the other three terms in Eq. (18) obtainable via the Method-of-Plane as
where , , and denote the energy flux[2, 3, 4, 5] through the bin face , the energy density and the averaged velocity[1] on the bin face and the stress in the -direction exerted on the bin face , respectively. Note that the third term on the RHS of Eq. (19) denotes the inner product of the stress and the velocity vector on the bin face , i.e., the Einstein notation is used for the dummy index .
Derivation of the Lagrangian Derivative of the Internal Energy
In the main text, we analyzed the heat transport around the DCLs based on the the energy conservation in the Lagrangian form of the internal energy in Eq. (4). Here, we show the derivation from the equations of mass and momentum conservation and Eq. (5) as the definition of the internal energy (see also (see also Ref. 6). The mass conservation writes
and the momentum conservation with the presence of the external force writes
By introducing the Lagrangian derivative
and by inserting mass conservation in Eq. (20), the momentum conservation in Eq. (21) writes
By taking the inner product of Eq. (23) with , the Lagrangian derivative of the convective kinetic energy is derived as follows:
where denotes the inner product of a second order tensor. This equation means that the inner product of stress tensor with the velocity gradient does not contribute to the change of kinetic energy because it corresponds to the viscous dissipation in the bulk, whereas the work done by the external force contributes to the change of kinetic energy.
By transposing , in Eq. (1) as well, the Lagrangian form of the energy conservation writes
meaning that the fluid energy change is induced by the heat flux divergence, the stress work and the work done by the external force. By inserting Eq. (5), it follows
By further inserting Eq. (24), the Lagrangian of the internal energy in Eq. (4) is derived:
From this equation, it is shown that the work done by the external force is implicit, i.e., it does not contribute to the change of internal energy but to the change of convective kinetic energy.
Density, velocity and temperature distributions with various wall speed on the lyophilic wall
We conducted the heat flow analysis of the DCLs with various wall speeds: , 2.5, 5.0, 7.5, 10.0, and 12.5 m/s on the lyophilic wall. The density and velocity fields and temperature distribution with each condition are shown in Figure S1. With the increase of the wall speed, the speed of the caterpillar-like flow became larger and the meniscus shape got more skewed, resulting in a larger difference in the CAs between the advancing and receding CLs. As easily expected, the bulk temperature gets higher with the faster wall speed due to the viscous dissipation as shown in Figure S2.

Figure S1. Density and velocity fields with various wall speeds: , 2.5, 5.0, 7.5, 10.0, and 12.5 m/s on the lyophilic wall.

Figure S2. Temperature distribution with various wall speeds: , 2.5, 5.0, 7.5, 10.0, and 12.5 m/s on the lyophilic wall.
Heat transport around the dynamic CL on the lyophobic wall
The same temperature trend around the dynamic CL shown in Figure 1 with the lyophilic wall in the main text is also observed in a system with lyophobic wall. Specifically, we set a smaller Lennard-Jones energy parameter between solid and fluid molecules and a faster wall speed m/s, whereas the other setting is the same as the system in Figure 1. The static contact angle on such a wall was deg.7
Figure S3 shows the density, velocity and temperature distributions. The curvature of the menisci is positive unlike the system in Figure 1 due to the low wettability, and this results in a positive pressure in the liquid phase. The bulk liquid temperature is about 106 K which is much higher than the control temperature 85 K due to the high shear rate in bulk liquid, whereas that around the advancing CL is about 3 K higher and that around the receding CL is about 4 K lower. This indicates that the cooling due to the internal energy change around the dynamic CL is induced even on the lyophobic wall, and it is more remarkable than that in Figure 2 due to the faster flow.

Figure S3. (Top) Density distribution and flow field in a system with a lyophobic wall. (Bottom) temperature distribution.
References
- [1]H. Kusudo, T. Omori, and Y. Yamaguchi, “Local stress tensor calculation by the method-of-plane in microscopic systems with macroscopic flow: A formulation based on the velocity distribution function,” J. Chem. Phys. 155, 184103 (2021).DOI
- [2]B. D. Todd and P. J. Davis, Nonequilibrium Molecular Dynamics: Theory, Algorithms and Applications (Cambridge University Press, 2017).
- [3]T. Ohara, “Intermolecular energy transfer in liquid water and its contribution to heat conduction: A molecular dynamics study,” J. Chem. Phys. 111, 6492–6500 (1999).DOI
- [4]D. Torii, T. Nakano, and T. Ohara, “Contribution of inter- and intramolecular energy transfers to heat conduction in liquids,” J. Chem. Phys. 128, 044504 (2008).DOI
- [5]B. D. Todd, P. J. Davis, and D. J. Evans, “Heat flux vector in highly inhomogeneous nonequilibrium fluids,” Phys. Rev. E 51, 4362–4368 (1995).pubmed.ncbi.nlm.nih.gov/9963149
- [6]D. Evans and G. Morriss, Statistical Mechanics of Nonequilibrium Liquids, 2nd ed. (Cambridge University Press, 2008) pp. 71–72.
- [7]Y. Yamaguchi, H. Kusudo, D. Surblys, T. Omori, and G. Kikugawa, “Interpretation of Young’s equation for a liquid droplet on a flat and smooth solid surface: Mechanical and thermodynamic routes with a simple Lennard-Jones liquid,” J. Chem. Phys. 150, 044701 (2019).