Interfaces endowed with non-constant surface energies revisited with the d'Alembert-Lagrange principle
aa r X i v : . [ c ond - m a t . s o f t ] N ov Interfaces endowed with non-constant surface energiesrevisited with the d’Alembert-Lagrange principle
Henri Gouin
University of Aix-Marseille and M2P2, C.N.R.S. U.M.R. 7340,Marseille, France
Abstract
The equation of motions and the conditions on surfaces and edges between fluidsand solids in presence of non-constant surface energies , as in the case of surfactants attached to the fluid particles at the interfaces, are revisited under the principle ofvirtual work. We point out that adequate behaviors of surface concentrations maydrastically modify the surface tension which naturally appears in the Laplace andthe Young-Dupr´e equations. Thus, the principle of virtual work points out a strongdifference between the two revisited concepts of surface energy and surface tension.
Key words:
Variational methods; capillarity; surface energy; surface tension.
PACS:
This paper develops the principle of virtual work due to d’Alembert-Lagrange [1] ( ) whendifferent phases of fluids are in contact through singular surfaces or interfaces. The studyis first presented without constitutive assumption for surface energies but the displacementfields are considered for a simple material corresponding to the first-gradient theory. Thed’Alembert-Lagrange principle allows us to obtain equation of motion and boundary con-ditions of mechanical nature and is able to be extended to more complex materials withmicrostructures [4] or to multi-gradient theories [5]. Here, we aim to emphasize the formu-lation of the principle of virtual work when the interfaces are endowed with non-constantsurface energies: the surfaces have their own material properties independent of the bulks andare embedded in the physical space, which is a three dimensional metric space. The surface Email address: [email protected]; [email protected] (Henri Gouin). The principle of virtual work is also referred to in the literature as the principle of virtual power while virtual displacements are called virtual velocities [2,3].
Preprint submitted to
MEMOCS , vol. , n o nergy density is taken into account and naturally comes into in the boundary conditions asthe Laplace and the Young-Dupr´e equations by using variations associated with the virtualdisplacement fields. To do so, it is necessary to propose a constitutive equation of the surfaceenergy; defining this is a main purpose of the paper. Such a presentation is similar to theone of the deformational and configurational mechanics [6]; the method is analogous withthe one employed in [2,3,4] but with powerful differential geometry tools as in [7]. However,the mathematical tools are adapted to the linear functional of virtual displacement fieldsand not to the integral balance laws over nonmaterial interfaces separating fluid phases asin [8].Consequently, the main result of the paper is to propose a general form of the linear func-tional with interfaces in first-gradient theory which points out the significance of constitutivebehaviors for the surface energies and highlights a strong difference between the notions ofsurface energy and surface tension. Fischer et al emphasized a thermodynamical definition ofsurface energy, surface tension and surface stress for which surface tension and surface stressare identical for fluids [9]. Our presentation is not the same: without any thermodynamicalassumption, the difference between surface energy and surface tension is a natural conse-quence of the virtual work functional and the d’Alembert-Lagrange principle. The surfaceenergy allows to obtain the total energy of the interfaces and the surface tension is directlygenerated from the boundary conditions of the continuous medium.In the simplest cases the two notions of surface energy and surface tension are mingled, butit is not generally the case when the surface energy is non-constant along the interfaces. Toprove this property, we first focus on the simplest case of Laplace’s capillarity and we obtainthe well-known equations on interfaces and contact lines.Surfaces endowed with surface matter as in the case of surfactants is a more complex case.The last decades have seen the extension of surfactant applications in many fields includingbiology and medicine [10]; surfactants can also be expected to play a major mechanical rolein fluid and solid domains. The versatility of surfactant mainly depends on its concentrationat interfaces. It experimentally appears that surfactant or surface-active agent is a substancepresent in liquids at very low concentration rate and, when surface mass concentration isbelow the critical micelle concentration, it is mainly absorbed onto interfaces and altersonly the interfacial free energies [11]. The interfacial free energy per unit area (generallycalled surface energy ) is the minimum amount of work required to create an interface at agiven temperature [12,13] and the fact that surfactants can affect the mechanical behaviorsof interfaces must be modelized in order to predict and control the properties of completesystems.In fact, our aim is not to study the general case of surfactants proposed in the literature butto focus on the virtual work method to prove that simple behaviors of the surface energydepending on the mass concentration can drastically change the capillary effects. So, theconcept of surface tension naturally appears in the equations on surfaces and on lines. Inthis paper, we call surfactant the matter distributed only on the interfaces: we consider thespecial case when surfactant molecules are insoluble in the liquid bulk (the surface mass con-centration is below the critical micelle concentration [10]) and are attached to fluid particlesalong the interfaces (without surface diffusion as in [14]).The manuscript is organized as follows: 2ection 2 briefly reminds some results formally presenting the principle of virtual work inits more general form by using the kinematics of a continuous medium and the notion ofvirtual displacement. The simplest example of the Laplace model of capillarity concludes theSection.Section 3 deals with the case when the interfaces are endowed with non-constant surfaceenergy, whereby we essentially focus on liquid in contact with solid and gas. The special caseof surfactants as interface matter attached to the fluid particles is considered. The surfaceenergy depends on the surface matter concentration. Such a property drastically changes theboundary conditions on the interface by using surface tension instead of surface energy.Section 4 deals with an explicit comparison between surface energy and surface tension onlywithin deformational mechanics.Section 5 is the conclusion in which some general extension can be forecasted.The main mathematics tools are collected in a large appendix so that the presentation ofthe text is not cluttered with tedious calculations. The main mathematical tool is Relation(15) which can be extended to more complex media. In continuum mechanics, motions can be equivalently studied with either the Newton modelof system of forces or the Lagrange model of the work of forces [2,3]. The Lagrange model doesnot derive from a variational approach but, at equilibrium, the minimization of the energycoincides with the zero value of a linear functional. Generally, the linear functional express-ing the work of forces is related to the theory of distributions; a decomposition theoremassociated with displacements (as C ∞ -test functions whose supports are compact manifolds)uniquely determines a canonical zero order form ( separated form ) with respect both to thetest functions and the transverse derivatives of the contact test functions [15]. In the sameway that the Newton principle is useless when we do not have any constitutive equation forthe system of forces, the d’Alembert-Lagrange principle is useless when we do not have anyconstitutive assumption for the virtual work functional.The equation of motion and boundary conditions of a continuous medium derives from the d’Alembert-Lagrange principle of virtual work , which is an extension of the same principle inmechanics of systems with a finite number of degrees of freedom: For any virtual displace-ment, the motion is such that the virtual work of forces is equal to the virtual work of massaccelerations [5].
The background of the principle of virtual work
The motion of a continuous medium is classically represented by a continuous transformation ϕ of a three-dimensional space into the physical set. In order to describe the transformationanalytically, the variables X = ( X , X , X ) which single out individual particles corre-spond to material or Lagrange variables; the variables x = ( x , x , x ) corresponds to Euler3ariables. The transformation representing the motion of a continuous medium is of the form x = ϕ ( X ,t ) or x i = ϕ i ( X , X , X , t ) , i ∈ { , , } , (1)where t denotes the time. At a fixed time the transformation possesses an inverse and contin-uous derivatives up to the second order except on singular surfaces, curves or points. Then,the diffeomorphism ϕ from the set D of the particle references into the physical set D isan element of a functional space ℘ of the positions of the continuous medium considered asa manifold with an infinite number of dimensions.To formulate the d’Alembert-Lagrange principle of virtual work in continuum mechanics, weremind the notion of virtual displacements . This notion is obtained by letting the displace-ments arise from variations in the paths of particles. Let a one-parameter family of variedpaths or virtual motions denoted by { ϕ η } , and possessing continuous partial derivatives upto the second order, be analytically expressed by the transformation x = Φ ( X ,t ; η ) (2)with η ∈ O, where O is an open real set containing 0 and such that Φ ( X ,t ; 0) = ϕ ( X ,t ) (thereal motion of the continuous medium is obtained when η = 0). The derivative with respectto η at η = 0 is denoted by δ . In the literature, derivative δ is named variation and the virtualdisplacement is the variation of the position of the medium [1] . The virtual displacementis a tangent vector to ℘ , functional space of positions, at ϕ ( δ ϕ ∈ T ϕ ( ℘ )). In the physicalspace, the virtual displacement δ ϕ is determined by the variation of each particle: the virtualdisplacement ζ of the particle x is such that ζ = δ x when at η = 0 , { δ X = 0 , δt = 0 , δη = 1 } and we associate the field of tangent vectors to D : x ∈ D → ζ = ψ ( x ) ≡ ∂ Φ ∂η | η =0 ∈ T x ( D ) , where T x ( D ) is the tangent vector bundle to D at x (Figure 1). x X D o D ϕ η ϕ S S o Physical space ζ ℘ ϕ = ϕϕ o η (C) δϕ functionalspace of positions Fig. 1. The boundary S of D is represented by a thick curve and its variation by a thin curve. Thevariation δ ϕ of the family { ϕ η } of varied paths belongs to T ϕ ( ℘ ), tangent space of ℘ at ϕ . The virtual work concept, dual of Newton’s method, can be written in the following form:
The virtual work δτ is a linear functional value of the virtual displacement, δτ = < ℑ , δ ϕ > (3)4 here < . , . > denotes the inner product of ℑ and δ ϕ , with ℑ belonging to the cotangentspace T ∗ ϕ ( ℘ ) of ℘ at ϕ .In Relation (3), the medium in position ϕ is submitted to covector ℑ denoting all the”stresses” in mechanics. In the case of motion, we must add the inertial forces, correspondingto the accelerations of masses, to the volume forces. The d’Alembert-Lagrange principle of virtual work is expressed as follows
For all virtual displacements, the virtual work is null . The principle leads to the analytic representation ∀ δ ϕ ∈ T ϕ ( ℘ ) , δτ = 0 Theorem : If expression (3) is a distribution expressed in a separated form [15], thed’Alembert-Lagrange principle yields the equation of motion and boundary conditions in theform ℑ = 0 .The virtual displacement is submitted to constraints coming from the constitutive equationsand geometrical assumptions such as the mass conservation. Consequently, the constraintsare not expressed by Lagrange multipliers but are directly taken into account by the varia-tions of the constitutive equations. The equation of motion and boundary conditions resultfrom the explicit expression of δτ associated with the considered physical problem. As a firstexample, the simplest case of theory of capillarity at equilibrium is considered. The classical Laplace theory of capillarity
Liquid-vapor and two-phase interfaces are represented by material surfaces endowed withan energy related to the Laplace free energy of capillarity. When working far from criticalconditions, the capillary layer has a thickness equivalent to a few molecular beams [16,17]and the interface appears as a geometrical surface separating two media, with its own char-acteristic behavior and energy properties [18]. The domain D of a compressible fluid (liquid)is immersed in a three Euclidian space. The boundary of the domain D is a surface S sharedin N parts S p of class C , ( p = 1 , ..., N ) (Fig. 2). We denote by ( R m ) − the mean curvatureof S ; the union of the limit edges Γ pq between surfaces S p and S q is assumed to be of class C and t p is the tangent vector to Γ p = S Γ pq , q = 1 , ..., N with q = p , oriented by the unitexternal vector to D denoted n p ; n ′ p = t p × n p is the unit external normal vector to Γ p inthe tangent plane to S p ; the edge Γ of S is the union of the edges Γ p of S p .To first verify the well-founded of the model, we consider the explicit expression of the func-tional δτ for compressible fluids with capillarity in non-dissipative case. The variation of thetotal energy E of such a fluid results from the variation of the sum of the local density ofenergy integrated on the domain D and the variation of the local density of surface energyintegrated on its boundary S ; to these variations, we must add the work of volume force ρ f in D , surface force T on S and line force L on Γ. Such an amount represents, for the domain5 ig. 2. The set D has a surface boundary S divided in several parts. The edge of S is denoted byΓ which is also divided into several parts with end points A m . D , the virtual work of forces of the compressible fluid with capillarity.The Laplace theory of capillarity introduces the notion of surface energy (or superficialenergy) on surfaces such that, for a compressible liquid with capillary effects on the wallboundaries, the total energy of the fluid writes in the form E = Z Z Z D ρ α ( ρ ) dv + Z Z S σ ds, with Z Z S σ ds ≡ N X p =1 Z Z S p σ p ds, where ρ is the matter density, α ( ρ ) is the fluid specific energy ( ρ α ( ρ ) is the volume energy)and the coefficients σ p are the surface energy densities on each surface S p represented -forthe sake of simplicity- by σ on S ( ). Surface integrations are associated with the metricspace. As proved in Appendix, the variation of the deformation gradient tensor F = ∂ x /∂ X (with components { ∂x i /∂X j } ) of the mapping ϕ combined with the mass conservation andthe variation of σ allow to obtain the variation − δE (see Eq. (27) in Appendix); then theindependent variables come from the position x of the continuous medium.The virtual work of volume forces defined on D is generally in the form Z Z Z D ρ f T ζ dv, with f = − grad U where U ( x ) is a potential per unit mass and superscript T denotes the transposition. Thevirtual work of surface and line forces defined on S and Γ are respectively, Z Z S T T ζ ds and Z Γ L T ζ dl. Consequently, the total virtual work of forces δτ is δτ = − δE + Z Z Z D ρ f T ζ dv + Z Z S T T ζ ds + Z Γ L T ζ dl. Our aim is not to consider the thermodynamics of interfaces. Consequently, α and σ are notconsidered as functions of thermodynamical variables such as temperature or entropy. δτ ≡ Z Z Z D (cid:16) − grad T p + ρ f T (cid:17) ζ dv + Z Γ (cid:16) L T − σ n ′ T (cid:17) ζ dl (4)+ Z Z S (cid:20) − δσ + (cid:26)(cid:18) p + 2 σR m (cid:19) n T + grad T σ (cid:16) − nn T (cid:17) + T T (cid:27) ζ (cid:21) ds where p ≡ ρ α ′ ( ρ ) is the pressure of the liquid [19], δσ denotes the variation of the surfaceenergy σ and denotes the identity tensor. When σ is constant we get δσ = 0; then, δτ ≡ Z Z Z D (cid:16) − grad T p + ρ f T (cid:17) ζ dv + Z Z S (cid:26)(cid:18) p + 2 σR m (cid:19) n T + T T (cid:27) ζ ds + Z Γ (cid:16) L T − σ n ′ T (cid:17) ζ dl. and the d’Alembert-Lagrange principle yields the equation of equilibrium on D , − p ,i + ρf i = 0 or − grad p + ρ f = 0 . (5)The condition on boundary surface S is, (cid:18) p + 2 σR m (cid:19) n i + T i = 0 or (cid:18) p + 2 σR m (cid:19) n + T = 0 , (6)where, for an external fluid bordering D , T = − P n , with P value of the pressure in theexternal fluid. On the lines, it is necessary to consider the partition of S such that the edgeΓ pq is common to S p and S q , σ p n ′ pi + σ q n ′ qi − L i = 0 or σ p n ′ p + σ q n ′ q − L = 0 . (7)Surface condition (6) is the Laplace equation and line condition (7) is the
Young-Dupr´e equa-tion with a line tension L .It is interesting to note that in [20], Steigmann and Li used the principle of virtual workby utilizing a system of line coordinates on boundary surfaces and lines. By introducing thefree energy per unit area of interfaces and the free energy per unit of contact curve, theyobtained Laplace’s equation and a generalization of Young-Dupr´e’s equation of equilibrium;moreover, by employing necessary conditions for energy-minimizing states of fluid systemsthey got a demonstration that the line tension associated with a three-phase contact curvemust be nonnegative.When σ is not constant but δσ = 0, we obtain the same equations for Eq. (5) and Eq. (7)but Eq. (6) on S is replaced by (cid:18) p + 2 σR m (cid:19) n + (cid:16) − nn T (cid:17) grad σ + T = 0 . (cid:16) − nn T (cid:17) grad σ = grad tg σ is the tangential part of grad σ to the surface S . This term corresponds to a shear stress necessarily balanced by the tangential componentof T . Such is the case when σ is defined on S image of S in the reference space D (then, σ = σ ( X )). We understand the importance of the surface energy constitutive behavior; thisquestioning is emphasized in the following section. We have seen in the previous section that the problem associated with the behavior of thesurface energy is the key point to obtain the boundary conditions on interfaces and contactlines bordering the fluid bulk. In this section we consider a very special case of surfactant:the interfaces are endowed with a concentration of matter which affects the surface energy.The surface matter is attached to the particles of the fluid such that they obey together tothe same Eq. (1) of motions and Eq. (2) of virtual motions. We consider a more general casethan in Section 2.2: we study the motion of the continuous medium with viscous forces. Thisviscosity affects not only the equation of motion but also the boundary conditions.
Geometrical description of the continuous medium
A drop of liquid fills the set D and lies on the surface of a solid. The liquid drop is alsobordered by a gas. All the interfaces between liquid, solid and gas are assumed to be regularsurfaces. We call σ S and σ S the values of the surface energies of S and S , respectively(see Fig. 3). These energies may depend on each point of the boundary of D . Afterwards, onthe domain S ′ , the surface energy between gas and solid is neglected [21]. The liquid dropis submitted to a volume force ρ f . The external surface force on D is modelized with twoconstraint vector fields T on the solid surface, S and T on the free surface, S . The linetension L is assumed to be null.By using the principle of virtual work, we aim to write the motion equation of the liquiddrop and the conditions on surfaces and line bordering the liquid drop. Surfactant attached to interfacial fluid particles
To express the behavior of the surface energy, we need to represent first the equation of thesurface matter density.By using the mapping ϕ , the set D of boundary S has the image D of boundary S . Weassume there exists an insoluble surfactant with a surface mass concentration c defined on D of image c in D [18,21,22]. Let us consider the case when the surfactant is attached to ig. 3. A liquid in drop form lies on a solid surface. The liquid is bordered with a gas and a solid; S is the boundary between liquid and solid; S ′ is the boundary between gas and solid; S is theinterface between liquid and gas ; n and n are the unit normal vectors to S and S , external tothe domain of liquid; the edge Γ (or contact line) is common to S and S and t is the unit tangentvector to Γ relative to n ; n ′ = n × t and n ′ = n × ( − t ) are the binormals to Γ relative to S and S , respectively. the fluid particles on the surface S , i.e. c = c ( X ) , where X ∈ S . (8)The mass conservation of the surfactant on the surface S requires that for any subset S ∗ of S , of image S ∗ subset of S , Z Z S ∗ c ds = Z Z S ∗ c ds . (9)Relation (9) implies c det F n T F − n = c with n T = n T F q ( n T F F T n ) , (10)where n denotes the unit normal vector to S . The proof of Rel. (10) is given in Appendix.From Rel. (8) and Rel. (10), we obtain:Firstly, the conservation of the surface concentration of the surfactant, dcdt + c (cid:16) div u − n T D n (cid:17) = 0 , (11)where u is the fluid velocity vector and D = (cid:16) ∂ u /∂ x + ( ∂ u /∂ x ) T (cid:17) denotes the rate of thedeformation tensor of the fluid. The term div u − n T D n expresses the tangential divergencerelative to the surface S .Secondly, the variation of the mass concentration of the surfactant, δc + c " div ζ − n T ∂ ζ ∂ x n = 0 . (12)9he proofs of Relations (11) and (12) are also given in Appendix. In the case when thesurface energy σ is a function of the surfactant concentration, σ = σ ( c ) , we deduce δσ = σ ′ ( c ) δc . If we denote γ = σ − c σ ′ ( c ) , (13)which is the Legendre transformation of σ with respect to c , then by taking Rel. (27) intoaccount, we obtain in Appendix, δE = − Z Z S (cid:20) γR m n T + grad T γ (cid:16) − nn T (cid:17)(cid:21) ζ ds + Z Γ γ n ′ T ζ dl. (14)As we shall see in Section 4, γ is the surface tension of the interface S . The variation δc of theconcentration has an important consequence on the surfactant behavior and the surfactantbehavior is essential to determine the virtual work of the liquid drop.Relation (13) can easily be extended to several surfactants: if σ = σ ( c , . . . , c n ) where c i , i ∈{ , . . . , n } are the concentrations of n surfactants, then γ = σ − n X i =1 c i ∂σ∂c i , corresponding to the Legendre transformation of σ with respect to c i , i ∈ { , . . . , n } andEq. (14) is always valid. Governing equation of motion and boundary conditions
As previously indicated, we do not consider the thermodynamical problem of interfaces, butfor example, when the medium is isothermal, α can be considered as the specific free energyof the bulk and σ the free surface energy of the interface.The use of virtual displacements yields a linear functional of virtual work, sum of severalpartial works. To enumerate the works of forces, we have to consider how they are obtainedin the literature [2,3,5]. The virtual work expressions of volume force ρ f , surface force T andliquid pressure p are the same as in Section 2.2.a) For fluid motions, the virtual work of mass impulsions is − Z Z Z D ρ a T ζ dv, where a is the acceleration vector.b) For dissipative motions, we must add the virtual work of viscous stresses − Z Z Z D tr τ v ∂ ζ ∂ x ! dv, τ v denotes the viscous stress tensor usually written in the form of Navier-Stokes [21].Taking account of the relation tr τ v ∂ ζ ∂ x ! = div( τ v ζ ) − (div τ v ) ζ , an integration by parts using Stokes’ formula in Eq. (14) for the virtual work of interfacialforces, and relations n ′ = n × t , n ′ = − n × t , allow to obtain the virtual work of forcesapplied to the domain Dδτ = Z Z Z D (cid:16) − grad T p + div τ v + ρ f T − ρ a T (cid:17) ζ dv + Z Z S " grad T γ (cid:16) − n n T (cid:17) + p + 2 γ R m ! n T − n T τ v + T T ζ ds + Z Z S " grad T γ (cid:16) − n n T (cid:17) + p + 2 γ R m ! n T − n T τ v + T T ζ ds + Z Γ (cid:16) γ n ′ T − γ n ′ T (cid:17) ζ dl, (15)where R m i denotes the mean radius of curvature of S i , γ i denotes the surface tension of S i and T i the surface force on S i , i ∈ { , } ; T = − P n , where P is the pressure in theexternal gas to the domain D .The field of virtual displacement x ∈ D −→ ζ ( x ) ∈ T x ( D ) must be tangent to the solid(rigid) surface S . The fundamental lemma of variational calculus yields the equation ofmotion associated to domain D , the conditions on surfaces S , S and the condition oncontact line Γ.Due to the fact that Eq. (15) is expressed in separate form in the sense of distributions [15],the d’Alembert-Lagrange principle implies that ∀ ζ ( x ) ∈ T x ( D ) tangent to S , each of thefour integrals of Eq. (15) is null. Then, we obtain equations on D , S , S and Γ, respectively. • We get the equation of motion in Dρ a + grad p = (div τ v ) T + ρ f . (16)Equation (16) is the Navier-Stokes equation for compressible fluids when τ v is written in theclassical linear form by using the rate of the fluid deformation tensor, τ v = λ (tr D ) + 2 µ D .We may add a classical condition for the velocity on the boundary as the adherence condition. • We get the condition on surface S The virtual displacement is tangent to S ; the constraint n T ζ = 0 implies there exists ascalar Lagrange multiplier x ∈ S −→ χ ( x ) ∈ ℜ , such that p + 2 γ R m ! n − τ v n + (cid:16) − n n T (cid:17) grad γ + T = χ n , (17)The normal and tangential components of Eq. (17) relative to S are deduced from Eq. (17),11 + 2 γ R m − n T τ v n + n T T = χ, (18) (cid:16) − n n T (cid:17) ( − τ v n + grad γ + T ) = 0 . (19)Following Eq. (18), we obtain the value of χ along the surface S . The scalar field χ corre-sponds to the unknown value of the normal stress vector on the surface S ; it correspondsto the difference between the mechanical and viscous normal stresses and a stress due to thecurvature of S taking account of the surface tension. Equation (19) represents the balancebetween the tangential components of the mechanical and viscous stresses and the tangentialcomponent of the surface tension gradient. • We get the condition on surface S p + 2 γ R m ! n − τ v n + (cid:16) − n n T (cid:17) grad γ − P n = 0 . (20)The normal and tangential components of Eq. (20) relative to S are deduced2 γ R m − n T τ v n + p = P, (21) (cid:16) − n n T (cid:17) ( − τ v n + grad γ ) = 0 . (22)Equation (21) corresponds to the expression of the Laplace equation in case of viscousmotions; the normal component of viscous stresses is taken into account. Equation (22) issimilar to Eq. (19) for the surface S but without component of the stress vector. • We get the condition on line ΓTo get the line condition we must consider a virtual displacement tangent to S and conse-quently in the form ζ = α t + β t × n , where α and β are two scalar fields defined on Γ. From the last integral of Eq. (15), we getimmediately:For any scalar field x ∈ Γ −→ β ( x ) ∈ ℜ , Z Γ βγ n ′ T ( t × n ) dl − Z Γ βγ n ′ T ( t × n ) dl = 0 , with n ′ = − t × n and n ′ = t × n and consequently, − γ − γ n T n = 0 . Denoting by θ the angle < n , n > , we obtain the well-known relation of Young-Dupr´e butadapted to γ and γ in place of σ and σ γ + γ cos θ = 0 . (23)12 .4 Remarks
For a motionless fluid, τ v = 0 and consequently:Equation (19) yields, grad tg γ = − T tg , where grad tg γ and T tg denote the tangential parts of grad γ and T , respectively. Thetangential part of the vector stress is opposite to the surface tension gradient. Therefore, atgiven value of T n = n T T , Eq. (18) yields the value χ corresponding to the normal stressvector to the surface S ,Equation (21) yields P = p + 2 γ R m corresponding to the classical equation of Badshforth andAdams [21] but with the surface tension γ instead of σ ,Equation (22) implies (cid:16) − n n T (cid:17) grad γ = 0. At equilibrium, along S , the surface tension γ must be uniform.In case of motion, Eq. (22) represents the Marangoni effect as proposed in [22,23] but withthe surface tension γ instead of σ . A surface tension must appear on the boundary conditions as a force per unit of line. TheLegendre transformation γ of σ with respect to c exactly corresponds to this property onthe contact line Γ; then, surface tension γ differs from the surface energy ; this importantproperty was pointed out by Gibbs [23] and Defay [22] by means of thermodynamical consid-erations. The fundamental difference between surface tension and surface energy, in presenceof attached surfactants, is illustrated in the following cases corresponding to formal behav-iors.- If σ is independant of c , then γ = σ : the surface tension is equal to the surface energy.This is the classical case of capillarity for fluids considered in Section 2.2 and Eq. (23) is theclassical Young-Dupr´e condition on the contact lines.- In fact σ is a decreasing function of c [21]; when c is small enough we consider the behavior σ = σ − σ c where σ > σ > , then, Eq. (13) implies γ = σ and surface tension and surface energy are different.- Now, we consider a formal case when the surface energy density model writes in the form σ = σ − σ c − σ c sin (cid:18) c (cid:19) σ > , σ > , σ >
0. Then, Eq. (13) implies γ = σ − σ cos (cid:18) c (cid:19) . (24)This case does not correspond to σ as a monotonic decreasing function of c . Nevertheless,when c → γ does not have any limit and we get γ ∈ [ σ − σ , σ + σ ] . The surface tension may have a large scale of values. When the concentration c is low,a variation of the concentration c may generate strong fluctuations of the surface tensionwithout significant change of the surface energy. Alternatively, the concentration behaviorstrongly affects the surface tension but not the surface energy. Relation (24) fits with thewell-known physical case of an hysteresis behavior for a drop lying on a horizontal plane (seefor example [24] and the literature therein). So, the surface roughness is not the only reasonof the hysteresis of the contact angle even if the surface energy is nearly constant. The principle of virtual work allows us to deduce the equation of motion and conditionson surfaces and line by means of a variational analysis. When capillary forces operate andsurfactant molecules are attached to the fluid molecules at the interfaces, the conditions onsurfaces and lines point out a fundamental difference between the concepts of surface energyand surface tension. This fact was thermodynamically predicted in [22,23]. Hysteresis phe-nomenon may appear even if surface energy is almost constant on a planar substrate whenthe surface tension strongly varies.In Eq. (23), γ and γ are not assumed to be constant, but are defined at each point of Γ.This expression of Young-Dupr´e boundary condition on the contact line Γ is not true in morecomplex cases. For example in the case when the surface tension is a non-local functional ofsurfactant concentration, the surface tension is no longer the classical Legendre transforma-tion of the surface energy relative to surfactant concentration and more complex behaviorscan be foreseen. These behaviors can change the variation of the integral of the free energyas in the case of shells or in second gradient models for which boundary conditions becomemore complex [3,25,26,27,28]. In a further article [29], we will see this is also the case whenthe surface energy depends on the surface curvature as in membranes and vesicles [30,31,32].14 Appendix - Geometrical preliminaries [33,34,35]
Expression of the virtual work of forces in capillarity
The hypotheses and notations are presented in the previous Section 2.2.
Lemma 1 : we have the following relations, δ det F = det F div ζ , (25) δ (cid:16) F − n (cid:17) = − F − ∂ ζ ∂ x n + F − δ n . (26)The proof of Rel. (25) comes from the so-called Jacobi identity δ (det F ) = det F tr (cid:16) F − δF (cid:17) and from δF = δ ∂ x ∂ X ! = ∂ ζ ∂ X . Then, tr (cid:16) F − δF (cid:17) = tr ∂ X ∂ x ∂ ζ ∂ X ! = tr ∂ ζ ∂ X ∂ X ∂ x ! = tr ∂ ζ ∂ x ! = div ζ . The proof of Rel. (26) comes as follow δ (cid:16) F − n (cid:17) = δ (cid:16) F − (cid:17) n + F − δ n But the implication, F − F = = ⇒ δ (cid:16) F − (cid:17) F + F − δF = 0 = ⇒ δ (cid:16) F − (cid:17) = − F − ∂ ζ ∂ X F − = − F − ∂ ζ ∂ x , yields Rel. (26). Lemma 2 : Let us consider the surface integral E = Z Z S σ ds. Then the variationof E is, δE = Z Z S (cid:20) δσ − (cid:18) σR m n T + grad T σ (cid:16) − nn T (cid:17)(cid:19) ζ (cid:21) ds + Z Γ σ n ′ T ζ dl. (27)Relation (27) points out the extreme importance to know the variation of δσ . The variation δE of E drastically changes following the different possible behaviors of the surface energy.15he proof can be found as follows: the external normal n ( x ) to S is locally extended in thevicinity of S by the relation n ( x ) = grad d ( x ) , where d is the distance of point x to S ; forany vector field w , we obtain [34,35]rot( n × w ) = n div w − w div n + ∂ n ∂ x w − ∂ w ∂ x n . From n T ∂ n ∂ x = 0 and div n = − R m , we deduce on S , n T rot( n × w ) = div w + 2 R m n T w − n T ∂ w ∂ x n . (28)Due to E = Z Z S σ det ( n , d x , d x ) where d x and d x are differentiable vectorsassociated with two coordinate lines of S, we get E = Z Z S σ det F det ( F − n , d X , d X ) , where d x = F d X and d x = F d X . Then, δE = Z Z S δσ det F det ( F − n , d X , d X ) + Z Z S σ δ (cid:16) det F det ( F − n , d X , d X ) (cid:17) . Due to Lemma 1 and to n T n = 1 = ⇒ n T δ n = 0, Z Z S σ δ (cid:16) det F det ( F − n , d X , d X ) (cid:17) = Z Z S " σ div ζ det( n , d x , d x ) + σ det ( δ n , d x , d x ) − σ det ∂ ζ ∂ x n , d x , d x ! = Z Z S div( σ ζ ) − (grad T σ ) ζ − σ n T ∂ ζ ∂ x n ! ds. Relation (28) yields div ( σ ζ ) + 2 σR m n T ζ − n T ∂σ ζ ∂ x n = n T rot ( σ n × ζ ) . Then,
Z Z S σ δ (cid:16) det F det ( F − n , d X , d X ) (cid:17) = Z Z S (cid:18) − σR m n T + grad T σ ( nn T − ) (cid:19) ζ ds + Z Z S n T rot ( σ n × ζ ) ds, where grad T σ ( nn T − ) belongs to the cotangent plane to S and we obtain Relation (27).16 .1.3 Variation of the internal energy
Let us note that δ Z Z Z D ρ α dv = Z Z Z D ρ δα dv where δα = ∂α∂ρ δρ .Due to the mass conservation, ρ det F = ρ ( X ) , (29)where ρ is defined on D , the differentiation of Eq. (29) yields, δρ det F + ρ δ (det F ) = 0and from Lemma 1, we get δ ρ = − ρ div ζ . Consequently, from p = ρ ∂α∂ρ and div( p ζ ) = p div ζ + (grad p ) T ζ , we get δ Z Z Z D ρ α dv = (30) Z Z Z D ρ ∂α∂ρ δρ dv = Z Z Z D − p div ζ dv = Z Z Z D (grad p ) T ζ dv − Z Z S p n T ζ ds. By taking account of Rel. (27), we immediately get Rel. (4).
Study of a surfactant attached to fluid particles
Proof of relation (10)
Under the hypotheses and notations of Section 3.2,
Z Z S ∗ c ds = Z Z S ∗ det( n c, d x , d x ) = Z Z S ∗ det( F F − n c, F d X , F d X )= Z Z S ∗ c (det F ) det( F − n , d X , d X ) = Z Z S ∗ c (det F ) n T F − n ds , where n T n = 1. Moreover, n T d x = 0 ⇒ n T F d X = 0, then n ′ T = n T F is normal to S ∗ andconsequently, n T = n T F q ( n T F F T n ) , n T = n T F − q ( n T F − ( F − ) T n )and from Rel. (9), c det F q n T F − ( F − ) T n = c . (31)17 .2.2 Proof of relations (11) and (12)
With the notations of Section 3.2, Rel. (31) yields dcdt = − c d (det F ) dt (det F ) r(cid:16) n T F − ( F − ) T n (cid:17) − c ddt (cid:16) n T F − ( F − ) T n (cid:17) F (cid:16) n T F − ( F − ) T n (cid:17) / . But, d (det F ) dt = (det F ) div u and ddt (cid:16) F − ( F − ) T (cid:17) = − F − D ( F − ) T . Then, dcdt + c (cid:16) div u − n T D n (cid:17) = 0 . The same calculation with δ in place of ddt yields immediately δc + c " div ζ − n T ∂ ζ ∂ x n = 0 . Proof of relation (14)
From Rel. (12) and σ = σ ( c ) we get, δσ = κ " div ζ − n T ∂ ζ ∂ x n with κ ( c ) = − c σ ′ ( c ) . Consequently,
Z Z S δσ = Z Z S div( κ ζ ) − grad T κ ζ − κ n T ∂ ζ ∂ x n ! ds. But Rel. (28) implies, n T rot( κ n × ζ ) = div( κ ζ ) + 2 κR m n T ζ − n T ∂ ( κ ζ ) ∂ x n and, n T ∂ ( κ ζ ) ∂ x n = ( n T ζ ) . (grad T κ n ) + κ n T ∂ ζ ∂ x n = grad T κ nn T ζ + κ n T ∂ ζ ∂ x n . Then,div( κ ζ ) − grad T κ ζ − κ n T ∂ ζ ∂ x n = − κR m n T ζ − grad T κ ( − nn T ) ζ + n T rot( κ n × ζ )Due to Z Z S n T rot( κ n × ζ ) ds = Z Γ κ n ′ T ζ dl, we get Z Z S − δσ ds = Z Z S (cid:20) κR m n T + grad T κ ( − nn T ) (cid:21) ζ ds − Z Γ κ n ′ T ζ dl, δE = − Z Z S (cid:20) γR m n T + grad T γ (cid:16) − nn T (cid:17)(cid:21) ζ ds + Z Γ γ n ′ T ζ dl. References [1] J. Serrin,
Mathematical principles of classical fluid mechanics , in Encyclopedia of PhysicsVIII/1, Ed: S. Fl¨ugge, pp. 125-263, Springer, Berlin, 1960.[2] P. Germain,
La m´ethode des puissances virtuelles en m´ecanique des milieux continus , J.M´ecanique , 235-274, 1973.[3] P. Germain, The method of the virtual power in continuum mechanics - Part 2 : microstructure ,SIAM, J. Appl. Math., , 556-575, 1973.[4] N. Daher and G. A. Maugin, The method of virtual power in continuum mechanics: applicationto media presenting singular surfaces and interfaces , Acta Mechanica, (3-4), 217-240, 1986.[5] H. Gouin, The d’Alembert-Lagrange principle for gradient theories and boundary conditions ,in: Ruggeri, T., Sammartino, M. (Eds.), Asymptotic Methods in Nonlinear Wave Phenomena,World Scientific, pp. 79-95, Singapore, 2007.[6] P. Steinmann,
On boundary potential energies in deformational and configurational mechanics ,Journal of the Mechanics and Physics of Solids, (3), 772-800, 2008.[7] R. Fosdick and H. Tang, Surface transport in continuum mechanics , Mathematics andMechanics of Solids, (6), 587-598, 2009.[8] P. Cermelli, E. Fried and M. E. Gurtin, Transport relations for surface integrals arising inthe formulation of balance laws for evolving fluid interfaces , J. Fluid Mechanics, , 339-351,2005.[9] F. D. Fischer, T. Waitz, D. Vollath and N. K. Simha,
On the role of surface energy and surfacestress in phase-transforming nanoparticles , Progress in Materials Science, (3), 481-527, 2008.[10] M. J. Rosen, Surfactants and interfacial phenomena , Wiley Interscience, New Jersey, 2004.[11] P. G. de Gennes, F. Brochard-Wyart and D. Qu´er´e,
Capillarity and wetting phenomena: drops,bubbles, pearls, waves , Springer, Berlin, 2004.[12] D. A. Edwards, H. Brenner and D. T. Wasan,
Interfacial transport processes and rheology ,Butterworth-Heinemann, Stoneham, 1991.[13] J. C. Slattery, L. Sagis and E. S. Oh,
Interfacial transport phenomena , 2nd edition, Springer,2007.[14] A. McBride, A. Javili, P. Steinmann and S. Bargmann,
Geometrically nonlinear continuumthermomechanics with surface energies coupled to diffusion , Journal of the Mechanics andPhysics of Solids, (10), 2116-2133, 2011.
15] L. Schwartz,
Th´eorie des distributions , Ch. 3, Hermann, Paris, 1966.[16] S. Ono and S. Kondo,
Molecular theory of surface tension in liquid , in Encyclopedia of Physics,X, Ed: S. Fl¨ugge, p.p. 134-280, Springer, Berlin, 1960.[17] C. Domb,
The critical point , Taylor & Francis, London, 1996.[18] V. Levitch,
Physicochemical hydrodynamics , Prentice-Hall, New Jersey, 1962.[19] Y. Rocard,
Thermodynamique , Masson, Paris, 1952.[20] D.J. Steigmann and D. Li,
Energy minimizing states of capillary systems with bulk, surface andline phases , IMA J. Appl. Math. , 1-17, 1995.[21] A. W. Adamson, Physical chemistry of surfaces , Interscience, New York, 1967.[22] R. Defay,
Thermodynamique de la tension superficielle , Gauthier-Villars, Paris, 1971.[23] J. W. Gibbs,
The scientific papers of J. Willard Gibbs , vol. 1, p. 234, Longmans, Green andCo, London, 1928.[24] H. Gouin,
The wetting problem of fluids on solid surfaces. Part 2: the contact angle hysteresis ,Continuum Mech. Thermodyn. , 597-611, 2003.[25] E. Cosserat and F. Cosserat, Sur la th´eorie des corps d´eformables , Hermann, Paris, 1909.[26] R. A. Toupin,
Elastic materials with couple stresses , Arch. Rat. mech. Anal. , 385-414, 1962.[27] W. Noll and E. G. Virga, On edge interactions and surface tension , Arch. Rat. Mech. Anal. , 1-31, 1990.[28] F. dell’Isola and P. Seppecher,
Edge contact forces and quasi-balanced power , Meccanica ,33-52, 1997.[29] H. Gouin, Vesicles: a capillarity model revisited (in preparation).[30] W. Helfrich,
Elastic properties of lipid bilayers: theory and possible experiments , Z. Naturforsch.C , 693–703, 1973.[31] U. Seifert, Configurations of fluid membranes and vesicles , Adv. Phys. , 13–137, 1997.[32] A. Agrawal and D.J. Steigmann, A model for surface diffusion of trans-membrane proteins onlipid bilayers , ZAMP , 549-563, 2011.[33] H. Gouin and W. Kosi´nski, Boundary conditions for a capillary fluid in contact with a wall ,Arch. Mech. , 907-916, 1998.[34] R. Aris, Vectors, tensors, and the basic equations of fluid mechanics , Dover Publications, 1989[35] S. Kobayashi, K. Nomizu,
Foundations of differential geometry , vol. 1, Interscience Publ., NewYork, 1963., vol. 1, Interscience Publ., NewYork, 1963.