Spheroidal and conical shapes of ferrofluid-filled capsules in magnetic fields

Christian Wischnewski; Jan Kierfeld

doi:10.1103/PhysRevFluids.3.043603

Wischnewski, Christian;Kierfeld, Jan

2018-03-07 00:00:00

Spheroidal and conical shapes of ferro uid- lled capsules in magnetic elds Christian Wischnewski and Jan Kierfeld Department of Physics, TU Dortmund University, 44221 Dortmund, Germany (Dated: April 16, 2018) We investigate the deformation of soft spherical elastic capsules lled with a ferro uid in external uniform magnetic elds at xed volume by a combination of numerical and analytical approaches. We develop a numerical iterative solution strategy based on nonlinear elastic shape equations to calculate the stretched capsule shape numerically and a coupled nite element and boundary element method to solve the corresponding magnetostatic problem, and employ analytical linear response theory, approximative energy minimization, and slender-body theory. The observed deformation behavior is qualitatively similar to the deformation of ferro uid droplets in uniform magnetic elds. Homogeneous magnetic elds elongate the capsule, and a discontinuous shape transition from a spheroidal shape to a conical shape takes place at a critical eld strength. We investigate how capsule elasticity modi es this hysteretic shape transition. We show that conical capsule shapes are possible but involve diverging stretch factors at the tips, which gives rise to rupture for real capsule materials. In a slender-body approximation we nd that the critical susceptibility above which conical shapes occur for ferro uid capsules is the same as for droplets. At small elds capsules remain spheroidal, and we characterize the deformation of spheroidal capsules both analytically and numerically. Finally, we determine whether wrinkling of a spheroidal capsule occurs during elongation in a magnetic eld and how it modi es the stretching behavior. We nd the nontrivial dependence between the extent of the wrinkled region and capsule elongation. Our results can be helpful in quantitatively determining capsule or ferro uid material properties from magnetic deformation experiments. All results also apply to elastic capsules lled with a dielectric liquid in an external uniform electric eld. I. INTRODUCTION Elastic capsules consist of a thin elastic shell enclosing a uid inside. Elastic microcapsules are found in nature, for example, as red blood cells or virus capsids. They can also be produced arti cially by various methods, for example, interfacial polymerization at liquid-liquid interfaces or multilayer polyelectrolyte deposition [1]. Arti cially produced microcapsules are attractive systems for encapsulation and transport, for example, in delivery and release systems. Their overall shape is often nearly spherical, and the shell can be treated as a two-dimensional elastic solid with a curved equilibrium shape. In experiments and for applications, elastic properties of capsules can be tuned by varying size, thickness, and shell materials. For applications involving delivery by rupture of capsules it is necessary to understand and characterize the mechanical properties and elastic instabilities of capsules. The mechanical properties of elastic capsules are governed by the elastic shell, which is curved (typically spherical) in its equilibrium shape. This gives rise to dierent characteristic instabilities in response to external forces [1{3]. Thin elastic membranes bend much more easily than stretch. This protects a curved equilibrium shape against deformations changing its Gaussian curvature and that is the reason for the stability of capsules under uniform compression. In contrast to uid drops, elastic capsules under uniform compression fail in a buckling instability below a critical volume or critical internal pressure [4{9]. Buckling-type instabilities can also be triggered by external forces, for example, in electrostatically driven buckling transitions of charged shells [10] or in hydrodynamic ows [11]. Under point force loads, for example, exerted by atomic force microscopy tips, elastic capsules indent linearly at small forces and assume buckled shapes in the nonlinear regime at higher forces [2, 12, 13]. As opposed to uid droplets, elastic capsules can also develop wrinkles upon deformation [14{17] if compressive hoop stresses arise. Microcapsules can be manipulated and deformed in hydrodynamic ow [14, 18, 19], by micromanipulation using an atomic force microscope [2, 3] or micropipettes or capillaries [16, 17]. Another promising route to exert mechanical forces and to actuate elastic capsules in a noninvasive manner is via magnetic or electric elds [20, 21]. For magnetic elds this requires the presence of magnetizable material either in the shell or in the capsule interior. The whole capsule then acquires a magnetic dipole moment, which can be manipulated in external magnetic elds. For actuation by electric elds the capsule has to contain polarizable dielectric material such that the capsule acquires an electrostatic dipole moment, which can be manipulated by an electric eld. Homogeneous elds orient dipole moments but also christian.wischnewski@tu-dortmund.de jan.kierfeld@tu-dortmund.de arXiv:1803.02607v2 [cond-mat.soft] 13 Apr 2018 2 induce capsule deformations, which increase the size of the dipole moment after orientation. Therefore, homogeneous elds always lead to stretching and elongation of the capsule. Inhomogeneous elds can also exert a net force on the capsule and induce directed motion at xed magnetic dipole moment along the eld gradient. In the following we focus on spherical elastic capsules that are lled with a (quiescent) magnetic uid and deformed in homogeneous external magnetic elds. As magnetic uid we consider a ferro uid, which is a liquid that is magnetizable by external magnetic elds because it consist of ferromagentic or ferrimagnetic nanoparticles suspended in a carrier uid. Because of the small particle size, ferro uids are stable against phase separation and show superparamagnetic behavior [22]. Ferro uids are used in technical and medical applications [23{25]. All our results also apply to elastic capsules lled with a (quiescent) dielectric uid which are placed in a homogeneous external electric eld. The problem of ferro uid droplets in uniform external magnetic elds has already been theoretically studied in the literature. Also, a spherical ferro uid droplet is elongated in the direction of the magnetic eld for increasing eld strength; the resulting elongated shape was observed to be nearly spheroidal [26]. Bacri and Salin [27] used the assumption of a spheroidal shape for a quite precise approximation of the elongation by minimizing the total energy. Although the droplet is only elongated by the eld, an abrupt shape transition is possible [27]: Beyond a threshold magnetic eld strength the spheroidal droplet becomes unstable and elongates discontinuously into a shape with conical tips. The conical shape is stabilized by a positive feedback between shape and magnetic eld distribution: A sharp tip gives rise to a diverging eld strength at the tip, which in turn generates strong stretching forces stabilizing the sharp tip. The mechanism of forming sharp tips is reminiscent of the normal eld instabilities (Rosensweig instabilities) of free planar ferro uid surfaces in a perpendicular homogeneous magnetic eld, which were rst described by Cowley and Rosensweig [28] and later extended by a nonlinear stability analysis to study subsequent pattern formation [29]. The discontinuous shape transition to a conical shape exhibits hysteresis and only occurs above a critical suscepti- bility of the ferro uid. In Refs. [30, 31] a value ' 16:59 was found below which no conical shape can exist; a c c slender-body approximation in Ref. [32] gives ' 14:5. Using the approximative energy minimization for spheroidal shapes of Bacri and Salin [27] gives ' 19:8. The jump in droplet elongation at the transition to a conical shape depends on the magnetic susceptibility: Large elongation jumps are possible for high susceptibilities. This behavior was investigated in more detail in several numerical studies [33{35]. Apart from free ferro uid droplets, the deforma- tion behavior of sessile droplets on a plate [36] or sedimenting ferro uid drops in external elds [37] have also been investigated for homogeneous external magnetic elds. Dielectric droplets in a homogeneous external electric eld exhibit the same shape transition from a spheroidal to a conical shape. For the electric eld, however, free charges exist, and conducting droplets are easily realized experimentally. In fact, the rst experimental observations of conical droplet shapes were made for water droplets [38] and soap bubbles [39]. In Ref. [31] it was shown that also a conducting liquid droplet surrounded by an outer conducting liquid in a homogeneous electric eld exhibits conical shapes above a corresponding critical conductivity ratio = = 1 + ' 17:59. In the limit of an ideally conducting droplet with in nite susceptibility or in nite out c conductivity (both resulting in zero electric eld inside the droplet), the conical solutions in Refs. [30{32] approach Taylors cone solution with a half opening angle ' 49:3 [40]. Both for liquid metal (i.e., ideally conducting) and dielectric droplets, the conic cusp formation has been studied dynamically and dynamic self-similar solutions have been obtained [41, 42]. Fluid droplets, which are neither perfect conductors nor perfect insulators, disintegrate at higher external electric elds by emitting jets of uid at the tip, from which small droplets pinch o. Also for this process, scaling laws for droplet sizes could be theoretically obtained [43, 44]. In a ferro uid- lled elastic capsule the ferro uid drop is enclosed by a thin elastic membrane, which will modify the transition from a spheroidal to a conical shape observed for droplets. Such ferro uid- lled capsules have already been realized experimentally. Neveu-Prin et al. [45] encapsulated ferro uids by polymerization and analyzed the magnetization behavior of the magnetic capsules. Degen et al. [20] investigated experimentally elastic capsules lled with a magnetic liquid in an external magnetic eld. They used a magnetic liquid consisting of micrometer-sized magnetic particles that do not show the special properties of ferro uids but form long chains in the presence of external magnetic elds. These magnetite- lled elastic capsules could be actuated to deform in a magnetic eld. A quantitative theoretical description of their deformation is still missing. In Ref. [21], capsules lled with a dielectric liquid in an external electric eld were investigated experimentally and theoretically with a focus on small deformations. We will describe the elastic shell by a nonlinear elastic model based on a Hookean elastic energy density for thin shells, assume axisymmetric capsules, and calculate the shape at force equilibrium by solving shape equations as they have been derived in Refs. [7, 17]. As stated above, homogeneous magnetic elds acting on ferro uid- lled capsules give rise to stretching and elongation of the capsule in order to increase the total dipole moment. Therefore, stretching tensions are dominant in the elastic shell. This is why we will consider the limiting case of vanishing bending modulus and bending moments for most of the present work, which is commonly called the elastic membrane limit (as opposed to the elastic shell case). In our numerical approach, the magnetic eld inside the capsule is calculated using a coupled nite element and 3 boundary element method. The capsule shape provides the geometric boundary for the eld calculation. Vice versa, the magnetic eld distribution couples to the shape equations via the magnetic surface stresses. We solve the full coupled problem numerically by an iterative method. We combine this numerical approach with several analytical approaches to investigate the capsule deformation in a homogeneous magnetic eld as a function of the magnetic eld strength and Young modulus of the capsule material. First we characterize the linear deformation regime of spheroidal capsules for small elds both numerically and analytically. Then we answer the question to what extent the elastic shell will suppress the discontinuous spheroidal to conical shape transition of a ferro uid droplet and whether elastic properties such as the Young modulus of the shell material can be used to tune and control the instability. We show that conical shapes can also occur for capsules with nonlinear Hookean membranes but require diverging strains at the conical tips. As a real elastic material is not able to support arbitrarily high strains, we expect that diverging local stretch factors at the capsule poles indicate that real capsules tend to rupture close to the poles as soon as the conical shape is assumed. Then the existence of a sharp discontinuous shape transition into a conical shape provides an interesting route to trigger capsule rupture at the poles at rather well-de ned magnetic (for ferro uid- lled capsules) or electric (for dielectric- lled capsules) eld values. The subsequent rupture process has some analogies to the onset of the disintegration of droplets in electric elds [43, 44], but our static approaches based on nonlinear Hookean material laws are not suited to model the rupture process itself. We nd that the discontinuous shape transition between spheroidal and conical shapes with hysteresis eects and shape bistability is also present for elastic ferro uid- lled capsules. Numerically, we obtain a complete classi cation of the shape transition in the parameter plane of dimensionless magnetic eld strengths (magnetic Bond number) and the dimensionless ratio of the Young modulus of the shell material and the surface tension of the ferro uid. These ndings are partly corroborated by an analytical approximative energy minimization extending the spheroidal shape approximation of Bacri and Salin [27] to ferro uid- lled capsules. For conical shapes we generalize the slender-body approximations of Stone et al. [32], which allows us to quantify the divergence of local stretch factors at the capsule poles and to show that the same analytic formula as for ferro uid droplets governs the dependence of the cone angle on the magnetic susceptibility (or the dielectric susceptibility "=" 1 for a dielectric droplet with dielectric out constant " in a surrounding liquid with " ). In particular, we predict the critical susceptibility , above which out c the hysteretic shape transition between spheroidal and conical capsule shapes can be observed, to be identical to the critical value for ferro uid or dielectric droplets. We also nd that, for elastic capsules, magnetic stretching can give rise to wrinkling along the capsule equator region. We predict the parameter range for the appearance of wrinkles and the extent of the wrinkled region on a spheroidal capsule depending on its elastic properties and its elongation. II. THEORETICAL MODEL AND NUMERICAL METHODS We start with a ferro uid drop suspended in an external nonmagnetic liquid of the same density as the ferro uid, which eliminates gravitational forces. Thus the drop is force-free except for the surface tension , which forces the drop to be spherical and is balanced by internal pressure. If the drop is enclosed by an elastic shell, for example, after a polymerization reaction at the liquid-liquid interface, we have a spherical elastic capsule. We assume that the relaxed rest shape of this capsule is spherical with a rest radius R , which is given by the xed volume V = 4R =3 0 0 of the droplet or capsule. After applying a uniform magnetic eld H e in the z direction, the resulting shape of the capsule becomes stretched 0 z in the z direction, but the capsule shape and magnetic eld distribution remain axisymmetric around the z axis. A uniform external magnetic eld causes mirror-symmetric forces on the capsule, resulting in a shape with re ection symmetry with respect to the plane z = 0 (see Fig. 1). A. Geometry We describe the axisymmetric shell using cylindrical coordinates r, z, and '. The capsule's shell is thin compared to its diameter, so we consider the shell to be a two-dimensional elastic surface. Because of the axial symmetry, we only need the contour line r(z) to describe the whole capsule shape. For our calculations, we parametrize the surface by the arc length s of the undeformed spherical contour with s 2 [0; L = R ], starting at the lower apex and ending at the upper apex. Using the re ection symmetry, we only 0 0 0 need half of that interval, s 2 [0; L =2], to describe the capsule's shape completely. In addition to the coordinates 0 0 r(s ) and z(s ), we de ne a slope angle (s ) by the unit vector e following the contour line via e = (cos ; sin ). 0 0 0 s s 4 FIG. 1. Illustration of the parametrization in cylindrical coordinates (r; z; ') and the contour line with arc length s. The complete capsule is obtained by revolution of the red contour line, while the angle describes its slope. This contour line is calculated numerically. The polar radius is called a, while b denotes the equatorial radius. B. Magnetostatics 1. Forces by the ferro uid In order to calculate the shape of the capsule in an external magnetic eld, we have to take the magnetic forces that are caused by the ferro uid on the capsule surface into account. Because we are interested in a static solution, we can assume that the uid is at rest. Then the uid can only exert hydrostatic forces normal to the surface, while tangential components are zero. In order to calculate the normal magnetic force density f (r; z) on the surface, we use the magnetic stress tensor by Rosensweig, [22] H(r;z) f (r; z) = M (r; z)dH (r; z) + M (r; z): (1) m 0 Here M = jMj is the absolute value of the magnetization and M = M n its normal component (n is the outward unit normal to the capsule surface). Magnetization M and magnetic eld H are taken on the inside of the capsule surface. We assume a linear magnetization law M = H (2) with a susceptibility for the ferro uid ( = 1 in terms of its magnetic permeability ), which is justi ed for small elds H M =3, where M is the saturation magnetization of the ferro uid. References [46, 47] studied the s s behavior of drops with a nonlinear Langevin magnetization (polarization) law. The saturation of the magnetization or polarization forbids sharp tips and leads to more rounded drops. It was shown, on the other hand, that the linear law is a very good approximation for small and even medium elds. This typically requires the maximum magnetic ux density B = H to be in a range of 50 100 mT, depending on the speci c uid [36, 48]. For a linear max 0 max magnetization we can rewrite Eq. (1) as 2 2 f (r; z) = H (r; z) + H (r; z) (3) (assuming = 0 for the external non-magnetic liquid or using = = 1 in terms of the magnetic permeabilities out out of the ferro uid and the of the external liquid), where H = jHj and H is the normal component of the magnetic out n eld. We will use this position-dependent normal magnetic force density to modify the pressure in our elastic equations in Sec. II C 1. 5 2. Calculation of the magnetic eld To calculate the total magnetic eld, i.e., the superposition of the external uniform eld and the eld from the ferro uid magnetization, we use the fact that ferro uids are generally non-conducting [22]. Then Maxwell's equations give r H = 0, which allows us to introduce a scalar magnetic potential u with ru = H. From Maxwell's equation r B = r (H + M) = 0 we get Poisson's equation in magnetostatics r u(r; z) = r M(r; z): (4) For the linear magnetization law (2), Poisson's equation simpli es to the Laplace equation r u(r; z) = 0. For the numerical solution of this partial dierential equation we use a coupled axisymmetric nite element { boundary element method [49{52] with a cubic spline interpolation for the boundary [53]. This combination of methods was also used by Lavrova et al. for free ferro uid drops [34, 54, 55] and earlier for electric drops, e.g., by Harris and Basaran [56]. The nite element method (FEM) is used to solve Eq. (4) in the magnetized domain inside the capsule and the boundary element method (BEM) for the nonmagnetic domain outside. Both domains are coupled by the continuity conditions of magnetostatics for u and its normal derivative on the boundary of the capsule, @u @u in out u = u ; = ; (5) in out @n @n with = 1 for the external nonmagnetic liquid. Both the FEM and BEM exploit axial symmetry and eec- out tively operate in the two-dimensional rz plane, where the axisymmetric capsule shape is described by a contour line (r(s); z(s)). For the FEM we use a standard Galerkin method with linear elements on a triangular two-dimensional grid in the rz plane that is created with a Delauney triangulation using the Fade2D software package [57], where we set a xed number of grid points on the capsule's boundary. In the BEM we express solutions u(r ) of the Laplace equation r u = 0 for r on the outside or the boundary 0 0 of the capsule in terms of integrals over the boundary of the capsule. Using fundamental solutions with rotational symmetry [58], we have to solve a set of one-dimensional integrals over the whole boundary of the capsule @u (r ;r) @u(r) ax cu(r ) u(r) u (r ;r) rds = z : (6) 0 0 0 ax @n @n Here u (r ;r) u (r ;r)d' is the axially symmetric fundamental solution of Laplace's equation, which is 0 0 ax obtained from the fundamental solution u (r;r ) = 1=4jr r j of Laplace's equation, u (r;r ) = (r r ). In 0 0 0 0 the integral equation (6), u and its normal derivative are evaluated on the outside of the capsule surface. The point r is the point where u is to be calculated, while the integrals are taken over points r(s) on the capsule contour. Both r and r lie in the same rz-plane. For the geometric factor c, we have c = 1=2 for points r on the boundary and 0 0 c = 1 for points r in the exterior domain. The vector n denotes the outward unit normal vector and z describes 0 0 the z-component of r . On the right-hand side of (6), z can be interpreted as the potential of the external electric 0 0 eld. For numerical evaluation, the integrals in Eq. (6) are discretized by a point collocation method and solved by applying Gaussian quadrature for nonsingular integrands and a midpoint rule for weakly singular integrands. The FEM and BEM are coupled at the boundary by the continuity conditions (5). The FEM provides values for u on every nite element grid point inside the capsule including values u on the inner side of the boundary; in in addition, the normal derivatives @u =@n on the inside of the discretized capsule boundary are needed for the FEM in but remain a priori unknown. Values for these normal derivatives on the boundary points of the FEM grid are obtained by the BEM method. Our BEM uses linear interpolation for u between the discretized boundary points. We use the continuity conditions (5) to write the boundary integral equation (6) in terms of quantities on the inner capsule boundary. Using one BEM equation (6) for each boundary point (with c = 1=2), we obtain a set of equations that allows us to calculate the unknown derivatives @u =@n for given u and to get a closed system of equations for u in in everywhere inside the capsule. After solving the resulting system of FEM equations we know u everywhere inside the capsule. For the calculation of u inside the capsule and thus for the calculation of the magnetic force density f (r; z) acting on the capsule using (3), which is also calculated with the magnetic eld on the inside, it is not necessary to calculate u in the entire external domain explicitly. This is done implicitly by the BEM. If needed (for example, in order to calculate the eld in the exterior regions in Fig. 2), u can be calculated by solving (6) for points r in the exterior with c = 1. In a ferro uid capsule or drop with sharp edges, very high eld strengths can arise [see Fig. 2(c)]. Also eld gradients can be large, which makes pointed shapes prone to discretization errors caused by the grid. This eect can be countered to some degree by placing more FEM grid points at the tip in order to improve the precision there, 6 which is, however, limited by the BEM part of the solution scheme: The collocation points must not come too close to the symmetry axis because the weakly singular integrals become strongly singular on the z axis [59]. This leads to massively increasing numerical errors near the axis and a decrease of the overall precision. Overall, our numerical scheme to calculate singular BEM integrals is not the most advanced as a trade-o for simplicity. There are more elaborated schemes for the integration of singular integrals as, for example, developed over many years by Gray et al. [60, 61], which could provide a more elegant way to deal with the problem. We use the following compromise for the discretization: We place N = 250 elements on the boundary such that the length L of the ith boundary element (beginning at the equator) is given by i 1 L = c exp ln(l ) : (7) i 0 0 The constant c is chosen in order to obtain the correct total arc length L, which is given by the meridional stretch N L =2 factors = ds=ds of the deformed capsule [see Eq. (11) below], L = L=2 = ds . We choose l = 0:1 s 0 i s 0 0 i=1 (l = 1 gives a constant element length and l < 1 leads to a higher element density at the capsule's tip). Increasing 0 0 N beyond 250 does not improve the precision signi cantly. A higher density of points at the capsule's tip (lower l ) leads to stronger oscillations in the iterative solution scheme (see Sec. II D below). Sphere (a) Spheroid (b) Conical shape (c) 8.000 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0.125 FIG. 2. Numerical results for the magnetic eld distribution and capsule shape (two-dimensional projection) for a capsule lled with a ferro uid with a susceptibility of = 21. The ratio of Young's modulus and surface tension is Y = = 100. The 2D external magnetic eld H is uniform and points in the upward direction. Arrows indicate the local direction of H ; the color codes for the absolute value of H in units of H . The (a) spherical capsule and the (b) spheroidal capsule have uniform elds inside, while the eld in the (c) conical-shaped capsule increases strongly in the tips. The elongations a=b (ratio of the polar radius to the equatorial radius) are (a) a=b = 1, (b) a=b = 2:26, and (c) a=b = 5:38. The magnetic Bond numbers B [see de nition in Eq.(23)] are (a) B = 0 , (b) B = 262:4, and (c) B = 702:2. m m m 3. Electric elds and dielectric liquid Our approach to elastic capsules lled with a ferro uid in a magnetic eld also applies to capsules lled with a dielectric uid in an electric eld. The generic situation for a capsule lled with a uid with dielectric constant " is to be suspended in a dielectric liquid with a dierent " 6= ", which does not equal unity " 6= 1. Then the dielectric out out 7 force density in a linear medium is " " " 0 out " 2 2 f (r; z) = E (r; z) + E (r; z) ; 1; (8) e " " 2 " out which is completely analogous to (3) with playing the role of the susceptibility . For the general case, Poisson's equation becomes r (r; z) = r P(r; z); (9) with the electric potential and the polarization P. For a linear polarization law, it simpli es to the Laplace equation r (r; z) = 0. C. Equilibrium shape of the capsule 1. Elasticity and shape equations The capsule is deformed by the normal magnetic stresses f from the ferro uid. We have to calculate the resulting deformed equilibrium shape, where all elastic stresses, surface tension and magnetic stress are balanced everywhere on the capsule. Every point of the reference shape [r (s ); z (s )] is mapped onto a new point [r(s ); z(s )]. The 0 0 0 0 0 0 deformed shape [r(s ); z(s )] is calculated by solving shape equations, which are derived from nonlinear theory of 0 0 thin shells [7, 17, 62, 63]. We use a Hookean elastic energy density with a spherical rest shape. The Hookean elastic energy density (de ned as energy per undeformed unit area) is given by 1 Y 1 2D 2 2 2 2 w = (e + 2e e + e ) + E (K + 2K K + K ): (10) s s ' B s ' s ' s ' 2 1 2 Here e and e are meridional and circumferential strains that contain the stretch factors and : s ' s ' ds r e = 1; e = 1 ; = ; = : (11) s s ' ' s ' ds r 0 0 Here and in the following, quantities with subscript 0 refer to the undeformed spherical reference shape and quantities without 0 describe the deformed shape. Analogously, the bending strains K and K are generated by the curvatures s ' and : s ' d sin K = ; K = ; = ; = : s s s s0 ' ' ' ' s ' ds r In the elastic energy (10), Y is the two-dimensional Young modulus governing stretching deformations, E is the 2D B bending modulus, and is the two-dimensional Poisson ratio. Elastic properties are usually only weakly dependent; we use = 1=2, which is the typical value for an incompressible polymeric material. The arc length of the deformed capsule's contour is given by L = ds , while L = R is the xed arc length of the undeformed spherical s 0 0 0 capsule. In experiments, the capsule's shell is constructed by polymerization on the surface of a drop. Therefore, the undeformed reference shape, which is spherical in the absence of gravity, is also a solution of the Laplace-Young equation ( + ) = p; (12) s ' where is the surface tension of the droplet. The solution of the Laplace-Young equation will be discussed in detail in Sec. II C 4 below. In the following, we will neglect the bending energy, which means we set E = 0. The characteristic length scale of the problem is the radius R of the undeformed sphere, such that the neglect of the bending energy corresponds to the limit of large F oppl-von K arm an numbers Y R =E . This is the limiting case of an elastic Hookean FvK 2D B membrane and is a good approximation for two reasons. First, we will only consider capsules with thin shells as they were prepared in experiments [17, 20]. The shell thickness D is very small compared to the capsule size, D R . 3 2 With Y / D and E / D it follows that (R =D) 1 and stretching energies are typically larger 2D B FvK 0 than bending energies. The second argument is that the homogeneous magnetic eld acting on the ferro uid- uid 8 capsule predominantly stretches and elongates the capsule in order to increase its total dipole moment. This increases stretching energies, whereas the capsules develop high curvatures only at the conical tips. However, we show below that stretch factors diverge at conical tips, so the stretching energy dominates over the bending energy associated with these high curvatures also in the tip regions. Elastic tensions in the shell (de ned as force per deformed unit length) derive from the surface elastic energy density by 1 @w Y s 2D = = [( 1) + ( 1)] ; s s ' @e (1 ) ' s ' (13) 1 @w Y s 2D = = [( 1) + ( 1)] : ' ' s @e (1 ) s ' s Although we use a Hookean elastic energy density, the constitutive relation (13) is nonlinear because of the additional 1= factors, which arise for purely geometrical reasons: The Hookean elastic energy density is de ned per undeformed unit area such that @w =@e is the force per undeformed unit length, whereas the Cauchy stresses and are s s s ' de ned per deformed unit length. In addition to the elastic tensions and , there is also a contribution from an isotropic eective surface tension s ' between the outer liquid and the capsule. Such a contribution arises either as the sum of surface tensions of the liquid outside with the outer capsule surface and the liquid inside with the inner capsule surface or, if the capsule shell is porous such that there is still contact between the liquids outside and inside the capsule, with additional contributions from the surface tension between outside and inside liquids. In the absence of elastic tensions, the surface tension also gives rise to the spherical rest shape of the capsule. For macroscopic capsules the surface tensions should be negligible, but for microcapsules with weak walls they should not be neglected. We expect the eective surface tension to be somewhat smaller than typical liquid-liquid surface tensions, which are around = 50 mN=m; we will use = 10 mN=m below. The equilibrium of forces in the deformed elastic membrane is described by 0 = + + ( + ) p; (14) s s ' ' s ' cos 1 d(r ) 0 = ; (15) r r ds where Eq. (14) describes the normal force equilibrium and Eq. (15) tangential force equilibrium (in the s direction, equilibrium in the ' direction is always ful lled by axial symmetry). In the presence of magnetic forces, the pressure p(s) = p + f (s) (16) 0 m is modi ed by the magnetic stress f , which is a position-dependent normal stress pointing outwards and thus stretching the capsule and given by the magnetic eld at the capsule surface [see Eq. (3)]. It is important to note that magnetic forces are always normal to the surface such that they do no enter the tangential force equilibrium (15). The (homogeneous) pressure p is the Lagrange multiplier for the volume constraint V = V = 4R =3. 0 0 The equations of force equilibrium and geometric relations can be used to derive a system of four rst-order dierential equations with the arc length s of the undeformed spherical contour as an independent variable, which are called shape equations in the following: 0 0 r (s ) = cos ; z (s ) = sin ; 0 s 0 s (s ) = [ ( + ) + p(s )] ; 0 ' ' 0 (17) cos (s ) = ( ): 0 s ' s In these shape equations, the surface tension gives an isotropic and constant stress contribution, in addition to the elastic stresses and . This is because we assume that the undeformed rest state, where the elastic stresses and s ' s vanish, is identical to the shape of a ferro uid droplet of surface tension . We neglect that could change during capsule preparation and during elastic deformation. The system of shape equations is closed by the constitutive relation (13) for and the relations r sin = (1 ) ( 1) + 1 with = ; = ; s ' ' ' ' Y r r 2D 0 where the rst relation derives from the constitutive relation (13) for and the second relation is geometrical. For further details on the derivation of the shape equations, we refer the reader to Refs. [7, 17, 62]. 9 2. Numerical solution of the shape equations The system of shape equations (17) has to be solved numerically. The integration starts at the pole with s = 0 and runs to the capsule's equator at s = L =2. To integrate the four rst-order dierential equations we have three 0 0 boundary conditions at s = 0: r(0) = 0; z(0) arbitrary; (0) = 0: (18) The condition for r(0) follows from the absence of holes in the capsule. We can choose z(0) arbitrarily because the external magnetic eld does not depend on the z coordinate. The boundary condition (0) = 0 at the pole seems to exclude possible conical capsule shapes with (0) > 0. We discuss this issue below in Sec. III C and in Appendix C 3. There we derive the boundary condition (0) = 0 for nite stretches and at the poles. The boundary condition s ' (0) = 0 also arises if the magnetic forces f remain nite at the poles such that the normal force equilibrium requires nite curvatures at the poles. Conical shapes, however, have divergent stretches and and divergent magnetic s ' normal forces f at their conical tips. In the numerical calculation of capsule shape and magnetic eld we have to discretize the capsule surface such that divergences are cut o (this numerical issue is discussed in more detail also in Appendix D) and the boundary condition (0) = 0 for nite stretches and or nite magnetic force f is s ' m appropriate. Then the right-hand side of the shape equation for in (17) vanishes, (0) = 0 for s = 0 [see also Eq. s 0 (C14)], which can be used to start the integration at the pole. A priori, a fourth boundary condition for the tension (0) at the pole is unknown. On the other hand, we have (L =2) = =2 as a matching condition at s = L =2 s 0 0 0 to prevent kinks there. With the help of this matching condition, we can use a shooting method to determine (0). To increase numerical stability, we expand the shooting method to a multiple shooting method, where we use several integration intervals with several matching points. To keep the volume of the capsule constant, we have to use the internal pressure p as the Lagrange multiplier, which is adjusted during the calculation. In order to do so, p becomes another shooting parameter with V V as 0 0 the corresponding residual. In this work, we use a fourth order Runge-Kutta scheme with a step size of s = 10 in the rst integration interval starting at the apex and s = 10 in all other intervals, while there is a total of 250 integration intervals. 3. Wrinkling A ferro uid- lled capsule is stretched in a uniform external magnetic eld in the direction of the magnetic eld. As opposed to a ferro uid droplet, a capsule can develop wrinkles if circumferential compressive stresses arise as a result of this stretching. Because of volume conservation, the circumferential radius of the capsule has to decrease in the equator region giving rise to compression with < 1 in this region and a region of negative elastic stress < 0 develops. In ' ' contrast to a droplet with a liquid surface and constant surface tension > 0, regions of negative total hoop stress + < 0 can develop for capsules if the negative elastic hoop stress exceeds the surface tension. Then the elastic shell can reduce its total energy by developing wrinkles in the circumferential direction (see Fig. 3 for illustration). These wrinkles cost stretching energy in the meridional s direction and bending energy, but this is compensated by a release of compressional stresses and a reduction of elastic compression energy in the ' direction. Strictly speaking, + < 0 is only an approximation neglecting the bending energy, which will also increase upon wrinkling, and the negative stress has to exceed a small Euler-like threshold value. We expect the wrinkles to occur in a region near the capsule equator. Thus they will be roughly parallel to the external magnetic eld and therefore we assume that they do not eect the magnetic properties of the capsule. In order to introduce wrinkling in the shape equations, we will use the same approach that has been used for pendant capsules in Ref. [17]. The wrinkles will break the axial symmetry. In the wrinkled regions, where + < 0, we approximate the shape by an axisymmetric pseudomidsurface (r(s ); z(s )) for which we use modi ed axisymmetric 0 0 shape equations, where we set + = 0. This condition states that the total circumferential hoop stress is completely relaxed by fully developed wrinkles [64]. This leads us to a new set of equations (see also Ref. [17]), which read cos 0 0 0 r (s ) = cos ; z (s ) = sin ; (s ) = p; (s ) = ( + ): (19) 0 s 0 s 0 0 s s + r We also have to introduce a modi ed eective surface tension = = , because the real surface area exceeds the ' ' pseudosurface area, and we have to model this increase of E by increasing instead. This new system of dierential equations is closed by the relations + Y r s ' 2D = ; = : s ' Y r 2D 0 In order to calculate , the circumferential stretch factor of the real, wrinkled surface has to be calculated via the constitutive relations (13). To calculate wrinkled capsule shapes we start to solve the shape equations (17) as described before. As soon as the condition + < 0 is valid, we continue the calculations by solving the modi ed system (19). By following the solution of the modi ed system, we can calculate the length L of the wrinkled region L = ds: (20) + <0 At this point, it is also possible to calculate the wavelength of the wrinkles using the same methods as in Ref. [17]. Here we will mainly be interested in the extent L of the wrinkled region. FIG. 3. Three-dimensional illustration of a wrinkled capsule. The length L of the wrinkles is measured as the length of the region, where + < 0. The wrinkling wavelength is not determined explicitly here. 4. Ferro uid droplet The special case Y = 0 describes a ferro uid droplet without an elastic shell and has been treated in the literature 2D before. The balance of forces on the surface is given by the Laplace-Young equation (12). Using the de nitions of and , this equation can be translated into d p sin = : ds r In order to have a parametrization in the reference arc length s and a xed integration interval, we introduce a constant stretch factor , which is adjusted as a shooting parameter. The boundary and matching conditions are the same as in the case of the elastic shape equations. Together with the already known geometrical relations for r and z we get a system of three shape equations for a droplet: p + f sin 0 m 0 0 0 r (s ) = cos ; z (s ) = sin ; (s ) = : (21) 0 s 0 s 0 s This system is solved in the same way as the shape equations for elastic capsules in the previous sections. The basic shooting parameters are given by and p . Our solution scheme for the Laplace-Young equation is chosen such that s 0 it is completely analogous and comparable to the elastic shape equations. There are several other ways to solve this equation with a volume constraint, for example, by employing nite elements [65]. 11 D. Iterative numerical solution of the coupled problem The magnetostatic and the elastic problem are coupled: The capsule shape determines the boundary conditions for the magnetic eld via the continuity conditions (5), while the normal magnetic force density f (r; z) acting on the capsule surface [see Eq. (3)] enters the shape equations (17) via the pressure [see Eq. (16)]. To nd a joint solution we use an iterative numerical solution scheme. We start with the reference shape and calculate the corresponding magnetic eld H(r; z) for a given external eld H . Then, we can calculate a deformed shape of the capsule using this magnetic eld. Now we recalculate the magnetic eld and so on until the iteration converges. At this xed point, the solution of the shape equations and the magnetic eld are self-consistent. This iterative coupling of elastic shape equations to an external eld calculated by a boundary element method is similar to the iterative scheme used in Ref. [11] to calculate the shape of sedimenting capsules in an external ow eld. For the problem of ferro uid droplets, an analogous iterative strategy has been introduced in Refs. [34, 54, 55, 59]. The iteration can cause numerical problems in the solution of the the nonlinear elastic shape equations. If the capsule shape changes rapidly during the iteration, the shooting method used to solve the shape equations does not nd a solution. This problem can be reduced by slowing down the iteration. To solve the elastic shape equations in the nth step, we use a convex linear combination of the updated magnetic eld H and the magnetic eld H from n1 the previous iteration step instead of H itself [11, 59]: H = H + (H H ): (22) n n1 n1 The parameter ranges between 0 and 1 and has to be lowered in situations of quickly changing shapes of the capsule. Finally, it is switched back to 1 to ensure real convergence. To track a solution as a function of the magnetic eld strength, it is helpful to increase the external magnetic eld H in small steps H and let the capsule's shape 0 0 converge after each step. This slows down the calculation speed drastically but increases numerical stability and helps to track a speci c branch of stable solutions (see Sec. IV C 4). A problem with the iterative solution scheme can arise if the capsule shape becomes nearly conical with a very sharp tip of high curvature. Then the numerical error in the calculation of the magnetic eld (see Sec. II B 2), makes it dicult or even prohibitive to reach a xed point of the iterative scheme. Instead the iteration gives oscillations of the capsule shape around the required xed point, which worsens the quality of the results. The iterative strategy used here directly converges to stationary shapes without simulation of the real dynamics. An alternative to our iterative scheme is to directly simulate the dynamics for the uid from the electromagnetic, elastic, and hydrodynamic forces. Then the uid motion is simulated over time until it reaches a steady state. This method was used by Karyappa et al. for elastic capsules in electric elds [21]. For liquid droplets, there are comparable problems with sharp tips and numerical singularities, where the full dynamics could by solved to great accuracy, such as the emission of uid jets at the tip of drops in electric elds [43], pinch-o dynamics [66], and coalescence phenomena [67]. The errors of the eld calculation with nite elements at such sharp tips can also be reduced by using advanced mesh algorithms, such as the elliptic mesh generation [68]. E. Control parameters and non-dimensionalization In order to identify the relevant control parameters and reduce the parameter space, we introduce dimensionless quantities. We measure lengths in units of the radius R of the spherical rest shape, energies in units of R , i.e., tensions in units of the surface tension of the ferro uid, and magnetic elds in units of the external eld H . The problem is then governed by essentially three dimensionless control parameters. The magnetic Bond number B , R H 0 0 B ; (23) is the dimensionless strength of the magnetic force density. With this dimensionless number, the Laplace-Young equation (12) for a ferro uid droplet can be written in dimensionless form 2 2 ~ ~ ~ + ~ = pe + B H + H ; s ' 0 m with H H=H , ~ R , and p ~ pR = . The scaled droplet shape described by this Laplace-Young equation then 0 0 0 only depends on the two dimensionless parameters B and . The dimensionless Young modulus Y = is the control parameter for elastic properties of the capsule shell. Another 2D dimensionless control parameter for elastic properties is Poisson's ratio , which is set to = 1=2 and thus xed 12 throughout this paper. The limit Y = = 0 describes a droplet without an elastic shell while Y = 1 describes a 2D 2D system dominated by the shell elasticity. The three dimensionless parameters B , Y = , and the magnetic susceptibility of the ferro uid uniquely deter- m 2D mine the capsule shape (apart from its overall size R ). In the following we consider Bond numbers B between 0 0 m and 10 (see Sec. IV). For a typical ferro uid- lled capsule with = 21, R = 1 mm [21, 69], and = 0:01 N=m, these Bond numbers correspond to magnetic eld strengths H between 0 and about 500 kA/m (or elds B = H between 0 and 0:5 T). We consider dimensionless Young moduli Y = from 10 (nearly no elasticity) to 100 (elastically 2D dominated) and the purely elastic limit Y = = 1 (where the de nition of B is not useful anymore). 2D m For the analogous problem of a dielectric droplet in an external electric eld E we can introduce a dielectric Bond number B by B = " " R E =2 , where is the analog of the magnetic susceptibility and has been de ned e e 0 out 0 " " in (8). III. ANALYTICAL APPROACHES In this section we introduce three approximative analytical approaches to the problem, which describe ferro uid- lled elastic capsules in three dierent deformation regimes. The rst approach is the analysis of the linear response of the capsule to small magnetic forces. The second approach applies to spheroidal shapes at moderate magnetic forces and is an approximative minimization of the total magnetic and elastic energy under the assumption of a spheroidal shape and uniform stretch factors. This extends the approximative energy minimization of Bacri and Salin [27] for ferro uid droplets to capsules. Finally, we investigate conical capsule shapes as they can arise under strong magnetic forces. We investigate the existence of conical shapes and derive the governing equations in a slender-body approximation by extending the approach of Ref. [32] from conical droplets to conical capsules. A. Linear shape response at small elds In this section we derive the linear response of the spherical capsule shape to small magnetic forces. In particular, we derive the elongation a=b of the capsule, where a denotes the capsule's polar radius and b its equatorial radius (see Fig. 1). Details of the derivation are given in Appendix A; here we present the main results. At small elds displacements change linearly in the magnetic force density f . Therefore, radial and tangential displacements u () and u () (using spherical coordinates with a polar angle and assuming axisymmetry) are of O(H ). In order to calculate the displacements we consider the force equilibria in normal direction, i.e., the Laplace- Young equation (14), and in tangential direction, i.e., Eq. (15). For a liquid ferro uid droplet with an isotropic surface tension both force-equilibria give equivalent results. Expanding to linear order in the displacements around the spherical shape, we obtain two coupled dierential equations for the functions u and u . These linearized force-equilibrium equations can be solved exactly. The solution takes the form u = A + B cos ; u = C sin cos ; (24) 2 2 2 where A, B, and C are determined in Appendix A explicitly. We nd B = (5 + ) H R =8[Y + (5 + ) ] from 0 2D the normal force equilibrium, and C = 2(1 + )B=(5 + ) from the tangential force equilibrium, and the pressure is adjusted such that A = B=3 in order to ful ll the volume constraint. The functional form u = A + B cos of the normal displacement leads to a spheroidal shape in linear response. For a spheroid we can use the relation H = 3H (3 + ) and obtain B = R B 9(5 + )=4(3 + ) [Y = + (5 + )]. 0 0 m 2D The linear response approach remains valid as long as A; B; C R or B =[Y = + (5 + )] (3 + ) = . 0 m 2D From the displacement u () we can calculate its elongation a u (0) u (=2) B R R 1 + = 1 + b R R 0 0 in linear order in the displacement. For a ferro uid droplet with surface tension and without any elastic tensions, i.e., Y = = 0, we get, for the elongation a=b in linear order [see Eq. (A10)], 2D a 9 R 0 0 = 1 + H : b 8 (3 + ) For the general case Y = > 0, we nd [see Eq. (A16)] 2D a 9 R (5 + ) 9 B 0 0 m = 1 + H = 1 + ; (25) 2 2 b 8[Y + (5 + )](3 + ) 4 (3 + ) Y = (5 + ) + 1 2D 2D 13 which gives a precise prediction of the capsule's elongation for small elds, as a comparison with the numerical results in Fig. 4 shows. To leading order in B Eq. (25) agrees with the results from a similar small deformation approach in Ref. [21] for capsules lled with a dielectric liquid in electric elds. 1.6 1.5 1.4 Y / = 0 2D Y / = 0.1 2D 1.3 Y / = 1 2D Y / = 10 2D 1.2 Y / = n.w. 2D Y / = 2D 1.1 linear approx. 1.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 B /(Y / + (5 + )) m 2D FIG. 4. Elongation a=b of a capsule lled with a ferro uid with = 21 as a function of B =[Y = + (5 + )] for dierent m 2D values of Y = in the region of small deformations. The solid line describes the linear approximation from Eq. (25). The 2D best agreement between the numerical data and the linear approximation is given for a purely elastic system without wrinkling eects (closed purple circles). Wrinkling eects lead to considerable deviations (squares). B. Approximative energy minimization for spheroidal shapes In this section we derive an analytical approximation for the elongation a=b of the capsule at moderate magnetic forces by minimizing an approximative total energy, which assumes a spheroidal shape for magnetic and elastic contributions. For ferro uid droplets, the spheroidal approximation is based on the experimental observation that the droplet shape in uniform magnetic elds is very similar to a prolate spheroid [26, 27, 35] for suciently small magnetic Bond numbers before a transition into a conical shape can take place. Our numerical results show that this behavior remains qualitatively unchanged with an additional elastic shell (see Sec. IV A). Therefore, we consider a capsule with prolate spheroidal shape. Analogously to Bacri and Salin [27], we use an energy argument by minimizing the total energy of the capsule at xed volume V = (4=3)ab = V . The total energy consists of three dierent contributions. First is the surface energy E , which is caused by the surface tension . It is proportional to the surface area A and given by b 1 E = A = 2ab + arcsin ; (26) 2 2 where 1 b =a is the eccentricity. The second energy contribution is the magnetic eld energy E . According to Ref. [70], E can be written as mag mag E = H (27) mag 2 1 + n 2 2 3 for = 1 and with the demagnetization factor n = (b =2a ) [2 + ln ((1 + )=(1 ))]. out The third energy contribution is the elastic stretching energy E , which we construct by taking the energy density el w from Sec. II C, Z Z 1 Y 2D 2 2 E = w dA = (e + 2e e + e )dA ; el s 0 s ' 0 s ' 2 1 with e = 1 and e = 1, as de ned in Sec. II C 1. At this point, the stretch factors and are unknown s s ' ' s ' and we need further approximations. An acceptable approximation for spheroidal shapes, which is checked below by a/b 14 comparison with the numerics (see Fig. 6) is constant stretch factors throughout the shell, i.e., ; = const, which s ' leads to 1 Y 2D 2 2 E = (e + 2e e + e )A : (28) el s ' 0 s ' 2 1 We approximate the circumferential stretch factor by the stretching of a ber at the capsule equator and set = : In meridional direction we approximate by taking the ratio of the perimeter P of the corresponding ellipse, s ellipse which generates the prolate spheroid by rotation, and the perimeter P = 2R of a great circle on the initial circle 0 sphere. The perimeter of the ellipse is given by an elliptic integral. Therefore, we use Ramanujan's approximation [71], which leads us to P a + b 3 ellipse = 1 + ; P 2R circle 0 10 + 4 3 with (b a)=(b + a). As the last step, we have to minimize the total energy E = E + E + E with respect to the elongation ratio tot mag el a=b at xed volume V = (4=3)ab = V in order to get the equilibrium elongation as a function of the magnetic Bond number B for spheroidal shapes. Details of the calculation are presented in Appendix B. We obtain a closed but quite complicated analytical expression for the inverse relation B = g(b=a), i.e., the magnetic Bond number B as a function of the inverse elongation b=a < 1 for spheroidal shapes in Eq. (B1). The function g(k) in Eq. (B1) still depends on three dimensionless parameters: the susceptibility , the dimensionless Young modulus Y = , 2D and Poisson's ratio . This relation reduces to the results of Bacri and Salin [27] for ferro uid droplets in the limit Y = 0. 2D C. Conical membrane shapes with normal magnetic forces For ferro uid- lled droplets a shape transition into a stable conical shape with (0) > 0 is possible above a critical susceptibility and at high magnetic elds [27, 30, 32, 72]. We want to show that a conical shape with a strictly conical tip can also exist for an elastic capsule with spherical rest shape and normal magnetic stretching forces if the constitutive relation is of the nonlinear form (13). Details of the argument are presented in Appendix C. The existence of sharp cones in deformed membranes is an important issue in deformations of membranes with planar rest shape [73]. A membrane of thickness D prefers bending deformations (energy proportional to D ) over stretching deformations (energy proportional to D). If external forcing or constraints are such that stretching can be avoided, the membrane responds by pure bending. Any deformation of such an unstretched membrane has to preserve the metric and thus the vanishing Gaussian curvature of a plane. This results in so-called developable cones, which have zero Gaussian curvature everywhere except at the tip of the cone. Cones only develop in response to external forces or constraints, typically under compressional constraints or forcing as in the crumpling of paper. Then unstretched membranes develop folds or wrinkles around the developable cones in order to accommodate the excess area that occurs under compression [73{75]. Our ferro uid elastic membranes dier in several respects. The magnetic forces are always stretching forces and they are always normal to the surface such that the tangential force equilibrium (15) only involves internal stresses of the membrane. Under stretching forces the membrane cannot respond by pure bending and changes in the metric are unavoidable. However, the forcing depends on the magnetic eld distribution [see Eq. (3)] and becomes concentrated in points of high elds, which are typically points of high curvature. This establishes a positive feedback between shape and magnetic eld distribution that can stabilize conical tips. Moreover, we consider membranes with spherical rest shape and, thus, non-zero Gaussian curvature K = 1=R . This is another reason why deformation into a cone with K = 0 is impossible without stretching. Similar conditions (normal forces and spherical rest shape) are ful lled for spherical shells under point forces, where conical solutions have also been obtained [13] and to which most of our results regarding the existence of conical shapes should also apply. The tangential force equilibrium (15) has to be ful lled in the vicinity of the conical tip and is independent of the stretching magnetic forces, which are always normal. In combination with the nonlinear constitutive relations (13) this requires that the stretching tensions remain nite and isotropic at the conical tip, i.e., (0) = (0) > 0 s ' at s = 0. From the constitutive relations then also follows the isotropy of the stretches (0) = (0) at the tip. 0 s ' However, stretches are not necessarily nite at a conical tip. 15 For nite isotropic stretches (0) = (0) < 1 at the pole, l'H^ opital's rule applied at s = 0 gives (0) = s ' 0 ' (0) cos[ (0)] [see Eq. (C2)]. Then isotropy requires (0) = 0 and it follows that a sharp conical tip with (0) > 0 is impossible if stretches remain nite at the tip. Finite isotropic stretches at the pole thus always lead to at tips with (0) = 0 as for the spheroidal shapes. For diverging and asymptotically isotropic stretches (s ) (s ) const s ; (29) s 0 ' 0 with an exponent > 0; however, l'H^ opital's rule does not apply at s = 0. Then we nd instead that isotropy of the diverging stretches requires a conical tip with the relation = cos[ (0)] 1 = sin 1 (30) between the exponent and the half opening angle = =2 (0) of the conical tip [see Eq. (C4)]. This result can be obtained from a modi ed l'H^ opital's rule or directly from analyzing stretches for a deformation into a conical tip under the constraint of isotropy of the stretches at the tip [see Eq. (C12)]. For the nonlinear constitutive relation (13) diverging and isotropic stretches are still compatible with nite and isotropic tensions, which approach (0) = (0) = Y =(1 ), see Eq. (C5), at the tip. Moreover, > 1 according to (30) and, therefore, the divergence is ' 2D such that the elastic energy [the energy density (10) integrated over the tip area] remains nite. Any numerical approaches to capsule shell mechanics and the calculation of the magnetic elds rely on discretization. In the numerical solution of axisymmetric shape equations the arc length s is discretized. After discretization in the numerics, stretches necessarily remain nite at potential conical tips at the apices. Then our results for nite stretches apply, and we have to choose a boundary condition (0) = 0. Also, for the calculation of the magnetic elds, we discretize the boundary of the capsule [see Eq. (7)]. Therefore, also magnetic elds remain nite at conical tips. Then also the normal magnetic forces remain nite and can only support nite curvatures at the tip of the conical shape. This leads to a rounding of conical tips and, thus, also requires (0) = 0. This implies that, in the numerical calculations, all shapes of ferro uid capsules will have rounded tips with (0) = 0; the rounding of a conical tip for these numerical reasons will happen on the scale of the discretization of the problem. A boundary condition (0) for the numerical solution of the shape equations [see Eq. (18)] has also been used in Refs. [34, 54, 55, 59] for ferro uid droplet shapes. D. Slender-body approximation for conical capsules For ferro uid droplets, the conical shape could be investigated analytically using a slender-body approximation [32], which we want to adapt for conical shapes of the ferro uid- lled capsule. We have shown that conical shapes can also exist for ferro uid- lled capsules but they involve diverging isotropic stretches at the conical tip. Tensions are isotropic, remain nite at the conical tip and approach the limiting values (0) = (0) = Y =(1 ) [see Eq. s ' 2D (C5)]. The capsule shape is described by a function r(z) in cylindrical coordinates. In a slender-body approximation, we assume @ r 1; for a conical tip with half opening angle = =2 (0), we have @ r tan in the vicinity of the z z tip. Then we can neglect small radial eld components and approximate the magnetic eld as parallel to the z axis, H = H (z)e . The eld H (z) is determined by ln A 2 2 H = H (z) @ r (z)H (z) ; (31) where A is the aspect ratio of the slender shape, which can be expressed in terms of the half opening angle, A = 1= tan , for a conical shape [32]. This relation is unchanged as compared to uid droplets as it is a result of the slender shape and magnetic boundary conditions only and independent of the surface elasticity underlying the shape. In the slender-body approximation we also assume @ r 1=r such that the meridional curvature is small 1=r(z). Then the Laplace-Young equation describing normal force equilibrium becomes f [r(z)] + g = p + f : (32) ' 0 m r(z) This relation diers from the corresponding relation for uid droplets by the appearance of the additional elastic tension = (r). As shown in Appendix C 3, tangential force equilibrium is ful lled in the vicinity of the conical ' ' tip if stretches are diverging, and the resulting circumferential tension is h i 2D 1= sin 1= sin 1 (r) = 1 2R sin (a tan ) r (33) ' 0 1 16 [see Eq. (C14)] in the vicinity of the conical tip. Note that a still denotes the polar radius. In Appendix C 2 we also outline how the tension (r) could be calculated for a general shape r(z), in principle. The Laplace-Young equation (32) with an elastic tension (33) and the slender-body eld equation (31) provide two coupled equations for r(z) and H (z). The pressure p has to be chosen such that the resulting shape r(z) ful lls the volume constraint V = r (z)dz: (34) The three equations (31), (32), and (34) governing slender (and, in particular, conical) shapes of a ferro uid- lled capsule only dier in the appearance of the additional elastic tension = (r) from the corresponding equations ' ' for ferro uid droplets from Ref. [32]. They can be also be solved analogously as for ferro uid droplets, in principle. IV. RESULTS A. Spheroidal capsule shapes While the capsule is spherical at B = 0, it becomes elongated for increasing magnetic eld or Bond number B m m similarly to a ferro uid droplet. We can quantify the elongation by the ratio of capsule length a in the z direction and capsule diameter b at the equator, a=b. At small or moderate magnetic elds ferro uid capsules assume a prolate spheroidal shape to a very good approximation; one example is shown in Fig. 2(b). For small elds we calculated the linear response of the capsule exactly in Sec. III A and Appendix A and found displacements (24), which describe a prolate spheroid with an elongation a=b > 1 given by Eq. (25). This analytical result is in excellent agreement with numerical results for small elds (see Fig. 4). The linear response regime is valid as long as a=b 1 1 or B [Y = (5 + ) + 1](3 + ) = according to Eq. (25). m 2D Small magnetic elds are easily accessible and for many ferro uids, susceptibilities are rather small (for example, ' 0:36 in Ref. [36]). Therefore, spheroidal shapes in the linear response regime are experimentally easily accessible. Then the linear response relation (25) can be used as experimental method to deduce unknown capsule material properties, for example, Young's modulus Y if the magnetic properties of the ferro uid are known. eD At moderate magnetic elds, the capsule shape remains very similar to a prolate spheroid for all elongations a=b . 3, which was one basic assumption of the approximative energy minimization in Sec. III B. Figure 5 demonstrates this for shapes with a=b = 2. The spheroidal approximation works better for systems dominated by the surface tension, i.e., for small ratios Y = . For xed Bond number B and susceptibility the elongation decreases with increasing Y = 2D m 2D because of the additional stretching energy of the shell as compared to a droplet, so a ferro uid droplet (Y = = 0) 2D always shows the highest elongation. For small elds, this trend can be quanti ed with the linear response relation (25). For smaller elongations, the spheroidal approximation tends to work better. The other assumption in the approximative energy minimization in Sec. III B was constant stretch factors throughout the shell, i.e., ; = const (and thus constant elastic tensions and ). Also this approximation works very well s ' ' s for spheroidal shapes with elongations a=b . 3, as the numerical results in Fig. 6 for a=b = 2 (left scale, red line) show. As a result, the approximative energy minimization in Sec. III B gives very good results for moderate magnetic elds, i.e., for all elongations a=b . 3, where we always nd prolate spheroidal shapes, as the comparison with numerical results in Fig. 7 shows. B. Conical capsule shapes and capsule rupture For large magnetic elds or Bond numbers B and at suciently high susceptibilities , ferro uid capsules can also assume conical shapes, such as the shape in Fig. 2(c), which have also been found for ferro uid droplets [30, 32]. We investigated the possibility of conical shapes for elastic capsules with normal magnetic forces above in Sec. III C and found that stretch factors have to diverge at the conical tips, (s ) (s ) const s [see Eq. (29], with an s 0 ' 0 exponent = sin 1, which is determined by the half opening angle = =2 (0) of the conical tip [see Eqs. (30) and (C4)]. This behavior is con rmed by our numerical results in Fig. 6 (left scale, blue line). The stretch factors diverge but are asymptotically isotropic at the tips. The nonlinear constitutive relations (13) then result in nite and isotropic tensions (0) = (0) = Y =(1 ) [see Eq. (C5)]. s ' 2D Diverging stretch factors cannot be realized in an actual material without rupture. Typical alginate capsule materials can only resist stretch factors of ' 1:2 before rupture; highly stretchable hydrogels can resist stretch factors up 17 0.8 Y / = 100, 2D with wrinkling 0.7 Y / = 100, 2D no wrinkling 0.6 spheroid, a/b = 2 0.5 160 180 200 220 240 260 2.2 0.4 2.1 0.3 2.0 0.2 1.9 0.1 1.8 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 z/R FIG. 5. Comparison of numerically calculated r(z) contour of a capsule with Y = = 100, and = 21, for a value B chosen 2D m such that the elongation is a=b = 2 (the inset shows the location of the pictured shapes in the B -a=b plane) with a spheroid. The shape calculated without wrinkling (blue solid line) shows very good agreement with a spheroid of the same volume and elongation (red dashed line). Taking wrinkling into account leads to visible deviations (green dotted line). 1.5 1.5 = 2, spheroid (a) (b) (0) = 1178 3 3 = 5.34, conical 10 10 b 1.4 1.4 0.562 0.651s z(r)/R (cone) 1.3 1.3 2 2 10 10 1.2 1.2 3.1 1 1 10 10 1.1 1.1 3.0 0 0 1.0 10 1.0 10 5 4 3 2 1 3 (c) 0 10 10 10 10 10 0.0 0.1 8 4 2 s /R r/R 0 0 0 s /R 0 0 FIG. 6. (a) Stretch factors in the meridional direction (s ) following the whole contour line from the south pole (s = 0) to s 0 0 the equator (s =R = =2) for Y = = 100 and = 21. The left scale (red dashed line) gives almost constant stretch factors 0 0 2D for a spheroidal shape with a=b = 2. The right scale (blue solid line) gives diverging stretch factors for a conical shape with a=b = 5:34. (b) Logarithmic plot of (s ) near the tip for s =R < 10 . The function (s ) = const s [see Eq. (29)] was s 0 0 0 s 0 tted to the data of the conical shape, which gave = 0:562, corresponding to an angle = 25:98 in Eq. (30). (c) Zoom in to the tip of the contour line z(r) for the conical shape; the half opening angle is 25 . to 20 [76]. Therefore, a real capsule should rupture at the poles at the transition into a conical shape and we conclude that investigations of conical shapes are primarily of theoretical interest. Such rupture events have actually been observed in Ref. [21] for capsules lled with a dielectric liquid in external electric elds. We expect that the nonlinear Hookean material law will become invalid at such high stretch factors prior to rupture. Then constitutive relations which are more realistic for high strains should be used. Nevertheless, the appearance of large stress factors is a robust feature of the conical shape independently of the material law. Conical shapes cannot be described quantitatively by the approximative energy minimization from Sec. III B as spheroidal shapes with a large elongation a=b, which is clearly shown by the deviations between numerical results (data points) and the approximative energy minimization from Sec. III B (solid lines) for the conical shapes in Fig. 7. For ferro uid droplets, conical shapes can be described by a slender-body theory [32], which we generalized in Sec. III D to ferro uid- lled capsules. The three governing equations (31), (32), and (34) from Sec. III D can be used to r/R a/b z/R 0 18 describe conical capsule shapes quantitatively. As pointed out above, the tensions remain nite and isotropic at the conical tip, i.e., (r) (r) Y =(1 ) s ' 2D for small r [see Eq. (C5)]. Then the slender-body equation (32) from normal force balance actually becomes identical to the corresponding equation for a droplet from Ref. [32], however, with an eectively increased surface tensions = + (0). Also the other two equations (31) and (34) are identical such that we obtain very similar slender e ' conical shapes for capsules and droplets, which can be mapped onto each other by a simple shift of the surface tension. The mechanism underlying the stabilization of the conical shape is analogous to ferro uid droplets because tensions remain nite and isotropic at the conical tip. A sharp conical tip with curvatures / 1=r gives rise to diverging 1=2 magnetic elds H / r and normal magnetic forces 2 1 f / H / r ; (35) both for ferro uid droplets and capsules. These strong magnetic stretching forces stabilize the conical tip against high elastic restoring forces. The normal component of the elastic force is mainly due to the nite circumferential tension + (0) acting along the high circumferential curvature / 1=r at the conical tip, resulting in an elastic force ' ' f / [ + (0)] / r with the same divergence. Magnetic and elastic normal forces balance in the Laplace-Young el ' ' 1=2 equation (32) in the slender-body approximation. The magnetic eld exponent H / r is identical for capsules and droplets, as long as the elastic tensions at the conical tip are nite. This exponent determines the critical susceptibility above which a shape transition into conical shapes is possible and therefore we also nd the identical for capsules c c and droplets as discussed in the following section. C. Spheroidal-conical shape transition of capsules Upon increasing the magnetic eld or the magnetic Bond number B at xed capsule elasticity Y = > 0 and for m 2D a suciently large and xed ferro uid susceptibility , we nd a discontinuous shape transition from spheroidal to conical capsule shapes, similar to what has been found for ferro uid droplets (Y = = 0) [27, 30, 32]. One of our 2D main results is the diagram of capsule elongation a=b as a function of Bond number B in Fig. 7 for dierent values of elasticity parameters Y = and for = 21, where a lower spheroidal branch and an upper conical branch and a 2D discontinuous transition between both branches can be identi ed. In the following sections we will discuss dierent aspects of this shape transition in more detail. 1. Critical susceptibility For ferro uid droplets, a discontinuous shape transition was observed in experiments [27, 77] and numerical simulations [34, 35] only for susceptibilities > , i.e., above a critical susceptibility . In Ref. [30] a value c c = = 1 ' 16:59 was found below which no conical shape can exist; the slender-shape approximation for c c out droplets from Ref. [32], which we generalized to elastic capsules in Sec. III D, gives = 16e=3 ' 14:5. The ap- proximative energy minimization of Bacri and Salin [27], which we generalized to elastic capsules in Sec. III B, gives ' 19:8 for ferro uid droplets. Numerically, a range of ' 19 to ' 19:5 is observed [72]. The question arises c c c whether a critical susceptibility can also be found for the existence of a discontinuous spheroidal-conical transition for ferro uid- lled elastic capsules. For given and half opening angle of the conical shape electromagnetic boundary conditions determine the divergence H / r of the eld via the equation [30, 31] 0 0 P (cos )P ( cos ) + ( + 1)P ( cos )P (cos ) = 0: (36) Because of the nite elastic tension (0) at the conical tip, the magnetic eld at the tip of a conical capsule diverges with the same = 1=2 [see Eq. (35)] as for a conical droplet. Therefore, we nd the same critical susceptibility ' 16:59, above which a conical solution can exist, for both capsules and droplets. In the slender-body approach, Eq. (31) determines and applies unchanged to both slender conical droplets and 1=2 ferro uid- lled capsules. Also the magnetic eld divergence H / r is identical in both cases, so the analysis of Eq. (31) predicts the same critical value = 16e=3 ' 14:5 for ferro uid- lled capsules as for ferro uid droplets. In particular, both the analysis of Eq. (36) and the slender-body approach predict that the value for to be independent of the Young modulus Y of the capsule. This result is corroborated by our numerics for = 21, where 2D we always observe a spheroidal-conical shape transition, even for Y = ! 1 [see Eq. (7)]. 2D This result is in contrast, however, to what we nd using the approximative energy minimization for spheroidal shapes from Sec. III B. Analyzing Eq. (B1), B = g(k) = g(b=a), for the saddle points of the function g(k) gives the m 19 7 7 Y / = 0, (0) = 70.2 2D Y / = 0 2D 6 6 Y / = 0.01 2D (c) Y / = 0.1 2D 5 5 Y / = 1 2D Y / = 10 2D 4 4 conical Y / = 100 2D spheroidal 3 3 Y / = , 2D no wrinkling Y / = , 2 2 2D (b) with wrinkling 1 1 0 2 4 6 8 0.62 0.64 0.66 B /(Y / + (5 + )) B /(Y / + (5 + )) m 2D m 2D FIG. 7. Elongation a=b of a capsule lled with a ferro uid with = 21 as a function of magnetic Bond number B for dierent values of the dimensionless elastic parameter Y = . The magnetic Bond number is rescaled by Y = + (5 +), which 2D 2D is motivated by the small eld behavior [see Eq. (25)]. The solid lines describe the theoretical results from approximative energy minimization (see Sec. III B). Open (closed) symbols denote numerical data for increasing (decreasing) B . The agreement is good for small elongations; the approximation fails for higher elongations, especially at the shape transition (close-up in the right diagram), where a=b jumps for small changes of B . Hysteresis eects are clearly visible in that area. There are two sets of numerical data for Y = = 1: Square data points are based on the modi ed shape equations that take wrinkling into 2D account, while diamonds are calculated without wrinkling. There are also two sets of data without elasticity: The upper data points (black) describe a droplet with a real conical tip with a cone angle of (0) = 70:2 , as it was given in Ref. [32]; for the lower points (blue) we used the boundary condition (0) = 0. Dashed lines indicate the position of shape transitions. Above these lines, shapes are conical, while they are spheroidal below. The markers (b) and (c) correspond to the shapes in Fig. 2. 0 00 critical value of the susceptibility [the two equations g (k) = 0 and g (k) = 0 determine two critical parameter values k = k and = ]. Using this approach, we nd a , which is strongly increasing with the Young modulus Y = , c c c 2D such that we nd > 21 already for Y = > 0:015, which clearly disagrees with all our numerical and analytical c 2D results. The reason for this disagreement is the failure of the approximative energy minimization to correctly describe conical shapes as discussed in Sec. IV B. It is interesting to consider the robustness of our result of a Y -independent that is identical to the for 2D c c ferro uid droplets with respect to the constitutive relation. We used the nonlinear Hookean constitutive relation (13), which can only support nite tensions at a conical tip, even for diverging stretches (see Sec. III C). A simple linear Hookean constitutive relation [missing the 1=-factors in Eq. (13)] behaves dierently and exhibits diverging tensions r with > 0 at a conical tip. Then tangential force equilibrium (C1) also requires r but with an ' s ' anisotropy = = 1 . With the linear constitutive relation this in turn leads to stretches r with an ' s s ' anisotropy = = (1 )=(1 + ) or () = (1 2)=(1 + ) for a Poisson ratio = 1=2. Requiring ' s this anisotropy in Eq. (C10) at a conical tip with half opening angle leads to a modi ed dierential equation (C11) 11=() sin and a divergence r . Consistency with r then requires s ' s ' 1 1 + 1 = 1 = 1; () sin 1 2 sin which determines the divergence = () of tensions r as a function of the opening angle . At the s ' conical tip we have now curvatures / 1=r in combination with circumferential tensions r such that normal ' ' 2 1 force balance also requires magnetic forces f / H / r [cf. Eq. (35)]. Thus, we have to use = 1() instead of = 1=2 in H / r in Eq. (36) and obtain a modi ed equation for the cone angle as a function of the parameter . This equation has a solution only above ' 40:5 and thus the critical value is strongly increased for a strictly c c linear Hookean constitutive relation. Our numerical results corroborate this result as we nd only spheroidal capsule shapes for a strictly linear constitutive relation at a susceptibility = 21. This shows that the value of is very sensitive to changes in the constitutive relation and a measurement of allows us to draw conclusions about the constitutive relation of the capsule material. a/b a/b 20 2. Critical Bond numbers Our numerical solutions of the shape equations show that the discontinuous spheroidal-conical shape transition that exists for ferro uid droplets [27, 30, 32] persists for ferro uid- lled elastic capsules and shows qualitatively similar features. Both for droplets and for capsules, the driving force of the shape transition is the lowering of the magnetic eld energy in the conical shape. Above an upper critical Bond number B the spheroidal shape becomes unstable m;c2 and the droplet or capsule deforms into a much more elongated, conical shape. This shape transition is discontinuous, i.e., the deformation into the conical shape is associated with a jump in a=b. The discontinuous transition between spheroidal to conical shapes also exhibits hysteresis: Lowering the Bond number starting from values B > B , m m;c2 the conical shape becomes unstable at a lower critical Bond number B with B < B . The discontinuous m;c1 m;c1 m;c2 spheroidal-conical transition only exists above the critical susceptibility . In other words, both droplets and capsules exhibit a line of discontinuous shape transitions in the -B plane for > , which terminates at a critical point m c located at = . The lines B () and B () are the limits of stability (spinodals) of this shape transition and c m;c1 m;c2 meet in the critical point. Figure 7 shows the capsule elongation with respect to B for dierent values of the dimensionless elastic parameter Y = of the capsule. We choose = 21, which is only slightly above . This ensures that we have a shape 2D c transition for a ferro uid droplet (corresponding to the limit Y = = 0), on the one hand, and relatively small 2D and thus numerically more stable elongations in the conical shape, on the other hand. Figure 7 clearly shows a discontinuous jump in elongation and hysteresis eects also for capsules with Y = > 0. 2D 3. Stretch factors as order parameter The discontinuous jump in the elongation ratio a=b at the spheroidal-conical transition is dicult to localize for larger values of Y = , as Fig. 7 shows. More suitable order parameters for the spheroidal-conical transition are the 2D stretch factors and . Because the stretch factors diverge at the tips of the conical shape (the divergence is only s ' limited by numerical discretization eects), whereas they stay nite at the poles of spheroidal shape (see Fig. 6 and our above discussion), we can directly employ the stretch factor (s = 0) at one of the poles as a convenient order s 0 parameter. For = 21 and Y = = 100, the shape transition occurs where a=b has a rather small jump from about 5.2 to 2D 5.35 for increasing Bond number B , whereas the stretch factor (s = 0) exhibits a much bigger jump by a factor m s 0 of more than 10, as demonstrated in Fig. 8. Also the shape hysteresis at the spheroidal-conical shape transition can be clearly seen for the order parameter (s = 0). s 0 Using this order parameter, we can detect the spheroidal-conical shape transition of ferro uid- lled capsules by the criterion lim j (s = 0; B ) (s = 0; B + B )j > 0; (37) s 0 m s 0 m m B !0 where we use values B = 0:005 for Y = < 1 up to values B = 0:5 for Y = = 100 in practice [B and m 2D m 2D m;c1 B grow approximately linearly with Y = (see Fig. 9 below) such that larger values B can be used for larger m;c2 2D m Y = ; smaller values of B give more precise results]. For ferro uid droplets, i.e., in the limit Y = 0, we 2D m 2D still have to use jumps in the elongation a=b for small changes B in the magnetic Bond number to detect the spheroidal-conical shape transition. We note that the discretization problem at the sharp conical tip mentioned above causes high relative errors in the numerical values of stretch factors in the tip area. Therefore, our numerical results for the diverging stretch factors at the tips of conical capsule shapes cannot be numerically exact. The detection of a divergence in at the poles, which we use to detect the transition into a conical shape, is, however, still possible even in the presence of numerical errors. 4. Shape hysteresis In order to track the range of elastic control parameters Y = , where a discontinuous shape transition with 2D hysteresis can be observed (for xed = 21), we use the stretch factor (s = 0) as the order parameter and the s 0 criterion (37) to determine B and B . We determine B by increasing the Bond number in small steps m;c1 m;c2 m;c2 B > 0 to locate the jump in the stretch factor (s = 0) at the pole, when the spheroidal shape becomes m s 0 unstable. Analogously, we determine B by decreasing the Bond number in small steps B < 0 to locate the m;c1 m jump in (s = 0), when the conical shape becomes unstable (see Fig. 8). s 0 21 B increasing B decreasing 655 660 665 670 675 680 685 690 FIG. 8. Meridional stretch factor at the capsule pole s = 0 as a function of Bond number B for Y = = 100 and = 21. s 0 m 2D The stretch factor clearly exhibits a jump at the location of the discontinuous shape transition and hysteretic behavior. Repeating this procedure for increasing values of the elastic control parameter Y = , we obtain the location and 2D size of the hysteresis loop B < B < B for a xed susceptibility as a function of Y = (see Fig. 9). We see m;c1 m m;c2 2D that B and B increase (approximately linear) for increasing Y = because of the increasing elastic energy m;c1 m;c2 2D needed for the same deformation. Note that the absolute numerical values of B and B cannot be considered m;c1 m;c2 exact as they are depending on the discretization of the magnetic eld calculation (see also Appendix D). The approximative energy minimization for spheroidal shapes from Sec. III B can be used to calculate approximative values for B and B from Eq. (B1), B = g(k) = g(b=a) [the two equations g (k) = 0 and B = g(k) determine m;c1 m;c2 m m the critical Bond numbers B = B and a corresponding critical inverse aspect ratio k = k ]. We nd that the m m;c1=2 c hysteresis loop closes already for Y = > 0:015 for = 21 (see Fig. 9), which is equivalent to our above nding 2D (see Sec. IV C 1) that > 21 for Y = > 0:015 in the approximative energy minimization. Comparison with our c 2D numerical results in Fig. 9 shows that the approximative energy minimization gives quite accurate results for the upper critical Bond number B , i.e., the stability limit of the spheroidal shape. It fails completely to predict the m;c2 lower critical Bond number B , i.e., the stability limit of the conical shape, because it is not able to describe conical m;c1 shapes quantitatively (see Sec. IV B). The numerical calculation shows hysteresis behavior for all values of Y = (see Fig. 9). Only the relative size of the 2D hysteresis loop, B 2(B B )=(B + B ), decreases slightly for increasing Y = in the numerical m;c m;c2 m;c1 m;c2 m;c1 2D results. 0.040 3.75 spheroidal approx. energy minimization 0.035 3.70 numerical data 0.030 3.65 0.025 m, c2 3.60 0.020 3.55 0.015 m, c1 3.50 0.010 0.005 3.45 (a) (b) 0.000 3 2 1 0 1 2 3 0.00 0.01 0.02 0.03 0.04 0.05 10 10 10 10 10 10 10 Y / Y / 2D 2D FIG. 9. (a) Critical Bond numbers B (lower data points) and B (upper data points) for varying Y = with = 21. m;c1 m;c2 2D The solid lines describe the prediction by the approximative energy minimization for spheroidal shapes. Both critical Bond numbers increase for increasing Y = . In the region B < B < B there are hysteresis eects in the spheroidal-conical 2D m;c1 m m;c2 shape transition. (b) Relative size B of the hysteresis area for a wider range of Y = . m;c 2D (s = 0) s 0 m, c 22 D. Wrinkling 1. Wrinkled shapes As opposed to liquid droplets, elastic capsules can develop wrinkles if a part of the shell is under compressive stress [14{17]. Wrinkles have also been considered for the equivalent problem of capsules lled with a dielectric liquid in an external electric eld in Ref. [21]. As it was stated in Sec. II C 3, wrinkles appear if the total hoop stress becomes compressive, + < 0. Then we have to use modi ed shape equations (19) in the numerical calculation of the shape. As can be seen in Fig. 7, taking wrinkling into account has a visible eect on the capsule's elongation for higher values of Y = . If wrinkling is taken into account capsules elongate because wrinkling reduces the compressional 2D stretch energy, which is stored near the equator. This elastic energy gain can be used for a further elongation of the capsule at the same eld strength to lower the magnetic energy. This also results in stronger deviation from the spheroidal shape. To visualize this eect, Fig. 5 shows the projection of the contour line of the upper right quadrant of capsules with and without wrinkling using the same elongation a=b = 2. While the shape is indistinguishable from a spheroid without wrinkling, the wrinkled shape deviates from a spheroid. Also in the presence of wrinkling, the discontinuous spheroidal-conical shape transition where the elongation in- creases persists. In the following, we will focus on the eect of wrinkles on the spheroidal branch of shapes. 2. Extent of wrinkled region In order to characterize the wrinkling tendency of spheroidal capsules we calculate the extent of the wrinkled region L [cf. Eq. (20) and Fig. 3], which can easily be measured in experiments. First we use the wrinkle criterion + < 0 to calculate the extent of the wrinkled region in the linear response regime for small magnetic elds as outlined in Sec. III A and Appendix A. In the linear response regime, we calculate the deviation from a sphere with radius R to leading order. We can characterize the size of the wrinkled region in terms of the polar angle as < < where is the smallest polar wrinkle where wrinkles appear, w w w ( ) + = 0. This angle is related to the length L of the wrinkled region by L = R ( 2 ): An angle of ' w w w 0 w = =2 implies the absence of wrinkles, while = 0 means that the wrinkles extend from pole to pole. Using Eq. w w (A17) for , we nd 5 R 5 + 5 4(3 + ) 1 + (5 + ) =Y 0 2D cos = = : (38) 9 Y B 3 9 27 B 2D m Interestingly, is universal and given by cos = 5=9 for purely elastic capsules ( =Y = 0), i.e., it does not depend w w 2D on the magnetic eld or capsule elongation. This is also the limiting result for large values of B =[1 + (5 + ) =Y ] m 2D (see Fig. 10). We note, however, that linear response theory is only applicable if B =[1 + (5 + ) =Y ] Y = . m 2D 2D For small magnetic elds, the results for from the linear response prediction (38) agree well with numerical results, as Fig. 10 shows. Now we address the extent of the wrinkled region beyond linear response and calculate numerically the relative extent of the wrinkled region, L =L. A value L =L = 0 means that there are no wrinkles, while L =L = 1 describes w w w a system where wrinkles extend from pole to pole. In Fig. 11, we change B and calculate L =L for dierent values m w of the capsule elongation a=b in the spheroidal shape, i.e., for a=b < 5. We use = 21 and consider several values of the elastic parameter Y = . 2D As Fig. 11 shows, there are no wrinkles for thin stretchable capsules, i.e., wrinkles only occur above a critical value of the dimensionless elastic parameter for 2D > 8:93 for = 21: (39) This result is only very weakly dependent on : We nd Y = > 9:03 for = 1 and Y = > 8:87 for = 100. For 2D 2D small Y , wrinkles are energetically unfavorable, i.e., the reduction of stretching energy E by wrinkles is smaller 2D el than the increase of E due to the increase of the surface area. Slightly above the critical value (39), wrinkles can only occur for capsules with elongations a=b ' 2:4. Further increasing Y (or shell thickness), the wrinkles become 2D longer and appear for a wider range of elongations. The extent of wrinkling is still limited by two eects. At the lower elongation a=b, where L =L = 0, a certain elongation is needed to create a sucient compressional stress at the equator to overcome the surface tension. The upper elongation a=b, where L =L = 0, is the point where the w 23 1.6 Y / = 20 2D Y / = 50 1.4 2D Y / = 100 2D theory /Y > 0 2D 1.2 theory /Y = 0 2D 1.0 0.8 0.6 0 10 20 30 40 50 60 70 80 B /(1 + (5 + ) /Y ) m 2D FIG. 10. Extent of the wrinkled region represented by the polar angle as a function of B =[1 + (5 +) =Y ]. The lines are w m 2D the linear response result (38), crosses and stars are numerical data points for dierent values of Y , which all collapse to the 2D linear response result. The red (dashed) line gives the asymptotic result cos = 5=9 for large values of B =[1 + (5 +) =Y ] w m 2D and for purely elastic capsules ( =Y = 0). 2D capsule is elongated so much that the transverse strain, which is related to Poisson's number and tends to shrink the capsule in the circumferential direction, counteracts any energy gain by the wrinkles. The wrinkles' length L =L for dierent elongations a=b turns out to be almost independent of the susceptibility . In systems completely dominated by the elasticity and with negligible surface tension, there are wrinkles for almost all elongations. The wrinkle length quickly rises to a maximum and then slowly decreases due to the transverse strain. 0.5 Y / = 2D Y / = 100 2D 0.4 Y / = 50 2D Y / = 20 2D Y / = 12 0.3 2D Y / = 11 2D Y / = 10 2D 0.2 Y / = 9 2D Y / = 8.93 2D 0.1 0.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 a/b FIG. 11. Relative wrinkle length L =L as a function of elongation a=b for spheroidal capsules with xed = 21 and dierent values of Y = . There are no wrinkles (L =L = 0) for Y = . 8:93. The range of L =L > 0 and the extent of wrinkles 2D w 2D w increase with Y = until they converge to an asymptotic curve for thick shells. 2D V. DISCUSSION AND CONCLUSION Magnetic or electric elds provide an interesting and fairly easily realizable route to the manipulation of elastic capsules if capsules can be lled with ferro uids or dielectric substances. In this work we investigated the deformation of ferro uid- lled capsules with thin elastic shells in uniform external magnetic elds numerically and using several analytic approaches. Our results apply unchanged to elastic capsules lled with a dielectric liquid in an external uniform electric eld (see Sec. II B 3). L /L w 24 Numerically, we obtained equilibrium shapes by solving the coupled elastic and the magnetostatic problems in an iterative manner. To calculate the magnetic eld, we used a combination of the nite element method and the boundary element method for a given capsule shape. The elastic capsule was described by nonlinear shell theory with a Hookean elastic law. By neglecting the bending rigidity we had to solve a system of four shape equations describing the force equilibrium in the absence of wrinkling and modi ed shape equations to take the eect of wrinkling into account. In addition to the dimensionless control parameters, the magnetic Bond number B and susceptibility , that characterize ferro uid drops, we used the dimensionless ratio Y = as an elastic control parameter. 2D As for ferro uid droplets, we found spheroidal shapes at small and moderate magnetic elds, conical shapes at high magnetic elds, and a discontinuous shape transition between spheroidal and conical shapes. The general behavior of ferro uid- lled capsules is comparable to drops but higher Bond numbers B are needed to reach the same elongation due to the additional elastic forces. For small elds, the capsule shape is exactly spheroidal and its elongation is very well described by a linear response theory, which is in good agreement with our numerical results (see Fig. 4). The small eld regime is easily accessible in experiments and our result (25) for the elongation a=b can be used to determine the Young modulus Y of 2D the capsule material from elongation measurements if the magnetic properties of the ferro uid are known. Also at moderate magnetic elds, capsule shapes with elongations a=b . 3 are prolate spheroids to a very good approximation and can be well described by an approximative energy minimization, as Fig. 7 shows. For high elds a conical shape is possible. Capsules in a conical shape must have nite isotropic tensions and diverging isotropic stretches at the conical tip [see Eq. (29)] with a divergence exponent, which is given by the half opening angle of the conical tip [see Eqs. (30) and (C4)]. The niteness of tensions at the tip is a consequence of the nonlinear constitutive relations (13). An important consequence of the divergence of stretches at the tips of a conical shapes is that conical shapes are probably not observable experimentally because the high stretch factors give rise to rupture close to the capsule tips. Another consequence of such high stretch factors is that the nonlinear Hookean material law will become locally invalid. A real elastic capsule material will show plastic behavior for high stretches, followed by strain hardening and, nally, the material's destruction [78]. Our results can explain experimental observations of rupture of capsules lled with a dielectric liquid in external electric elds, where the capsules' shells were destroyed near the tip [21]. Then the existence of the sharp discontinuous shape transition into a conical shape can provide an interesting tool to trigger capsule rupture at rather well-de ned magnetic (for ferro uid- lled capsules) or electric (for dielectric- lled capsules) eld values in future applications of such capsules as delivery systems. Capsule rupture at the tips has some analogies with the disintegration of droplets in electric elds by emitting uid jets at the tip [43, 44]. Real uid drops, which are not perfect conductors or perfect insulators, disintegrate at higher external electric elds by emitting jets of uid at the tip. This is known from experiments [39] as well as quite precisely understood in theory [43, 44]. In our setup of a uid inside an elastic shell, the emission of a uid jet is prevented by the shell at rst. However, the tangential stresses at the tip that lead to the formation of a uid jet may support the destruction of the shell near the tip. Once the shell is broken, a jet can be emitted. The rupture process itself cannot be described by our numerical approach and is an interesting topic for future work. Our elastic shape equation approach provides a very precise tool to solve the static elastic part of the problem, as long as nonlinear Hookean elasticity can be used. Also the generalization to other material laws, which are more appropriate for large strains, is possible [79]. Breaking of axisymmetry and topology changing rupture events cannot be easily incorporated into the shape equation approach, however. Also the magnetic eld calculation should be improved if rupture is addressed, in particular in the capsule's tip region by using, for example, an elliptic mesh generation for the nite element method. One idea for a future improved simulation method that captures possible rupture processes at the tip is a dynamic simulation, where the magnetohydrodynamics of the uid and the viscoelastic dynamics of the capsule shell including rupture processes can be calculated explicitly, similarly to what has been achieved for droplets in electric elds [43, 44]. We presented a complete shape diagram in Fig. 7 and characterized the discontinuous shape transition between spheroidal and conical shapes. The slender-body theory predicts that this discontinuous shape transition only exists above the same critical value as for ferro uid droplets, which was predicted to lie between ' 14:5 [32] and c c ' 16:59 [30]. It also predicts that is independent of the Young modulus Y of the capsule. We predict that c c 2D the critical will be very sensitive to the constitutive relation of the material. A strictly linear constitutive relation, for example, could give rise to diverging tensions at a conical tip, resulting in much higher values for . We used the meridional stretch factor at the pole as a suitable order parameter to detect the spheroidal-conical transition, because stretches diverge at the tip of conical shapes but remain nite for spheroidal shapes, resulting in a pronounced jump of the stretch factor in the numerical calculations. The spheroidal-conical transition exhibits hysteresis eects in an interval B < B < B between two critical Bond numbers, which are the limit of m;c1 m m;c2 stability of the spheroidal and conical shapes. In the hysteresis interval both types of shapes are metastable. The interval has its maximum size for ferro uid droplets and decreases slightly with increasing Young's modulus of the 25 elastic shell. In the numerical calculations for = 21, we observe hysteresis eects for all Y = , which shows that, 2D indeed, < 21 for all the Young moduli. It turned out that the formation of wrinkles is an important eect in systems with low surface tension . It has a visible eect on the elongation and the speci c shape. Wrinkles appear for the rst time for Y = & 8:93 (for 2D = 21), and are almost always present for systems with lower surface tension, even at very low elongations. Using this knowledge, it is possible to determine, for example, Y = in experiments by a simple measurement of the wrinkle 2D length L , which should be easy to perform in practice. Appendix A: Linear response at small magnetic elds In this appendix we derive the linear response of the capsule elongation a=b for small applied magnetic elds. Without applied eld, the capsule is spherical with a rest radius R . In the presence of a surface tension , this also requires an internal pressure p = 2 =R (Laplace-Young equation). If a small magnetic eld is applied the additional 0 0 position-dependent normal magnetic force density f = O(H ) [see Eq. (3)] acts on the spherical surface, resulting in normal displacements u ()e and tangential displacements u ()e , where we use spherical coordinates with the R R polar angle (i.e., = 0 at the upper pole and = =2 at the equator) and the spherical coordinate unit vectors e and e . Because of axisymmetry the displacements do not depend on the azimuthal angle ' and there is no displacement in direction e . The deformed capsule surface is parametrized as r(; ') = [R + u ()]e (; ') using ' 0 R R polar and azimuthal angles and '. The new equilibrium shape has small displacements u ; u = O(H ) and ful lls force equilibrium in two independent directions on the surface. We will consider normal force equilibrium as described by the Laplace-Young equation [see Eq. (14)] and tangential force equilibrium [see Eq. (15)]. We start with the Laplace-Young equation ( + ) + ( + ) = p + f ; (A1) s s ' ' 0 m 2 2 where is a surface tension, and are elastic tensions, and f = ( =2)[H + (n H) ] is the small normal s ' m 0 magnetic force density (3) causing small displacements. The pressure will change to linear order in the displacements p = 2 =R + O(u ; u ) to ensure a xed volume. In spherical coordinates and in linear order in the displacements, 0 0 R p p the stretch factors can be calculated using r = g = j@ rj (r = R sin ) and ds = g d = j@ rjd (ds = R d): '' ' 0 0 0 0 ds j@ rj 1 = = = 1 + (u + @ u ) ; s R ds R R 0 0 0 r j@ rj 1 = = = 1 + (u + u cot ) : ' R r R sin R 0 0 0 In linear order in the displacements the constitutive relations (13) can then be written as [80] 2D = Y ( 1) = (u cot + u ); (A2) ' s 2D ' R 2D = Y ( 1) = (@ u + u ) : (A3) s ' 2D s R Elastic tensions are small for small magnetic elds, ; = O(u ; u ) = O(H ), whereas the uid surface tension s ' R cannot be considered small. Therefore, we also need to consider curvature corrections up to linear order O(u ; u ) in Eq. (A1): 2 1 + 2u @ u + @ u cot : s ' R R R R R On the right-hand side of Eq. (A1), we can use n = e for the outward unit normal to O(H ). This results in the following normal force balance to linear order in the displacements, i.e., to O(H ): 2u @ u + @ u cot + R ( + ) R R R 0 s ' 2 2 2 2 = (p R 2R ) + H R (1 + cos ): (A4) 0 0 0 0 We rst solve this equation for a ferro uid droplet (Y = = 0), where the elastic stresses and thus u are zero. 2D Boundary conditions are @ u (0) = @ u (=2) = 0 and u (0) = u (=2) = 0 to avoid kinks (we are not considering R R conical shapes in the linear response) or holes in the shape. Then u = 0 and an ansatz u = A + B cos (A5) R 26 leads to a solution 0 0 2 2 2 2 2 B = H R H R (A6) 0 0 0 8 8 (3 + ) 1 p R 0 0 0 2 2 A = 2 R 1 + H R : (A7) 2 4 2 To leading order in u = O(H ), the ansatz (A5) describes a spheroid such that we can replace the magnetic eld H in (A6) by the analytically known value for a eld inside a spheroid [70], H = H =(1 + n); (A8) where n denotes the demagnetization factor. To leading order O(H ) it is also correct to use the result n = 1=3 for a sphere (A8). Moreover, volume conservation requires A = B=3; (A9) which determines the pressure correction p = 2 =R + O(H ) from Eq. (A7). For the deformation a=b we nd, to 0 0 leading order in u = O(H ), a R + u (0) B 9 R 0 R 0 0 = 1 + = 1 + H : (A10) b R + u (=2) R 8 (3 + ) 0 R 0 For a ferro uid- lled elastic capsule we also need to consider the force equilibrium in the tangential direction because the total tensions + 6= + become anisotropic now (for a liquid interface with = = 0 the force equilibrium s ' s ' in tangential direction becomes exactly equivalent to the normal force equilibrium, i.e., the Laplace-Young equation). The tangential force equilibrium (15) can be written as = @ (r ) = + r@ = + : ' r s s r s s @ r Using r = j@ rj = sin (R + u + u cot ) and Eq. (A3) for the elastic stresses, the tangential force equilibrium ' 0 R becomes 2D = (u cot @ u ) ' s (1 + )R Y u 2D = @ = tan @ u + @ u + (1 ) tan @ u : (A11) r s R (1 )R cos sin For the ferro uid capsule, the two force equilibria (A4), where and have to be expressed in terms of the s ' displacements using the constitutive relations (A3), 2D + = (2u + u cot + @ u ) ; s ' R (1 )R and Eq. (A11) have to be solved for the deformed capsule shape. Boundary conditions are @ u (0) = @ u (=2) = 0 R R and u (0) = u (=2) = 0. For the uid limit Y = = 0, we derived an exact solution above. For the ferro uid 2D capsule, we make an ansatz u = A + B cos ; u = C sin cos ; (A12) which still describes a spheroid to leading order in the displacements because u 6= 0 only generates an additional tangential displacement. Then the tangential force equilibrium gives 2(1 + ) C = B: (A13) 5 + For the ferro uid capsule, the normal force equilibrium (A4) gives (5 + ) 2 2 2 B = H R ; (A14) 8[Y + (5 + ) ] 2D 1 p R C 0 0 0 2 2 A = 2 R + H R 1 + (A15) 2 Y Y 4(1 )Y 2 1 + 2D 2D 2D 27 and the relation A = B=3 [see Eq. (A9] from the xed volume constraint determines the pressure p . For the deformation a=b we nd, to leading order in u = O(H ), a R + u (0) B 9 R (5 + ) 0 R 0 0 = 1 + = 1 + H : (A16) b R + u (=2) R 8[Y + (5 + )](3 + ) 0 R 0 2D The criterion for wrinkling is + < 0, where Y 1 5 2D = [u cot + (1 + )u + @ u ] B + 3 cos (A17) ' R (1 )R 5 + 3 from Eq. (A3) and using Eq. (A12) with Eqs. (A13) and (A9). Appendix B: Approximative energy minimization for spheroidal shapes In this appendix we derive an analytical approximation for the elongation a=b of the capsule at moderate magnetic forces by minimizing an approximative total energy, which assumes a spheroidal shape for magnetic and elastic contributions and constant elastic stretch factors throughout the shell. We minimize the total energy, the sum of surface, magnetic, and elastic energies with respect to the inverse elongation ratio k b=a < 1 at xed volume V = (4=3)ab = V (quantities :::j are at xed volume V ): 0 V dE j dE j dE j V mag V el V 0 = + + : dk dk dk 2 2 For xed volume V = (4=3)ab = (4=3)R = V , we have 2=3 1=3 aj = R k ; bj = R k : V 0 V 0 The surface energy (26), which is proportional to the surface area A at xed volume, can then be written as 1 1 1=3 E j = Aj with Aj = A k k + arcsin ; V V V 0 where = (k) 1 k is the spheroid's eccentricity and A = 4R the area of the undeformed sphere. The 0 0 magnetic energy (27) is given as V 1 0 0 E j = H = A B ; mag V 0 m 2 1 + n 3(1 + n) where n is the demagnetization factor k 1 + (k) n = n(k) = 2(k) + ln 2 (k) 1 (k) and B = R H =2 is the Bond number. m 0 0 Finally, we calculate the elastic stretch energy (28) via 2D 2 2 E j = A (e j ) + 2e j e j + (e j ) el V 0 s V s V ' V ' V 2(1 ) using the approximation of constant e and e . At xed volume, we nd s ' P a + b 3 ellipse e = 1 1 + 1; P 2R circle 0 10 + 4 3 2=3 2 k (1 + k) 3 (k) e j = 1 + 1; s V 10 + 4 3 (k) 1=3 e = 1 ; e j = k 1; ' ' V 0 28 with = (k) (b a)=(b + a) = (k 1)=(k + 1). Now we can nd the elongation k that minimizes the total energy at xed volume; k can only be determined implicitly as a function of the magnetic eld H by the following relation between the Bond number B = R H =2 and a 0 m 0 0 complicated function g(k) of the elongation k, which also depends on the susceptibility , the dimensionless Young modulus Y = , and Poisson's ratio : 2D R H 0 0 B = = g(k) with 2D c (k) + c (k; ) 1 2 2 2 (1 ) g(k) 3 + n(k) ; (B1) c (k) where 1 dAj c (k) ; A dk de j de j de j de j s V s V ' V ' V c (k; ) 2e j + 2 e j + e j + 2e j ; 2 s V ' V s V ' V dk dk dk dk dn 3k k 3k 1 + (k) c (k) = + + ln : 4 3 5 dk (k) (k) 2 (k) 1 (k) The functions c (k) and c (k) from surface and magnetic energies depend on the inverse elongation ration k = b=a < 1 1 3 only, whereas the function c (k; ) from the elastic energy also depends on Poisson's ratio (which is set to = 1=2 and thus xed throughout this paper). This relation reduces to the results of Bacri and Salin [27] for ferro uid droplets in the limit Y = 0, where the function c (k; ) drops from Eq. (B1). The solid lines in Fig. 7 show plots of 2D 2 1=k = a=b versus B as given by the relation B = g(k). m m Appendix C: Conical shapes for elastic membranes with spherical rest shape 1. Stretches and tensions at a conical tip with normal magnetic forces In this appendix we show that a conical shape, as it is observed for ferro uid drops at a critical eld strength, is also possible for an elastic capsule with a spherical rest shape and stretched by normal magnetic forces but requires diverging and asymptotically isotropic stretches with an exponent determined by the opening angle of the cone, whereas elastic tension have to remain nite and isotropic at the tip of the cone. A sharp conical tip implies a non-zero slope angle (s = 0) > 0, where = =2 (0) is half of the opening angle of the cone. In contrast to a ferro uid droplet with constant and isotropic surface tension , an elastic capsule develops additional elastic tensions and , which depend on the state of stretching, i.e., the stretches and s ' s ' with respect to the spherical rest shape via the nonlinear constitutive relations (13), and which have to ful ll an additional tangential force equilibrium (15) that we rewrite as = @ (r ) = + r@ : (C1) ' r s s r s It is important to note that the tangential force equilibrium does not contain external magnetic forces, which are always normal to the surface [see Eqs. (16) and (14)]. The internal tangential force equilibrium has to be compatible with the deformation into a conical tip. First we show that (0) and (0) have to remain nite at the tip at s = 0 (corresponding to r = 0). The reason s ' 0 for a divergence of one of the tensions can only be a divergence of one or both of the stretches. According to the nonlinear constitutive relations (13), only one of the tensions can exhibit a divergence ( = and = cannot both s ' ' s diverge). Then it is easy to verify that a single divergent tension at r = 0 contradicts the force equilibrium (C1). Therefore, both tensions have to remain nite at s = 0 (or r = 0). Next we show that niteness of the tensions at the conical tip necessarily leads to tension isotropy (0) = (0) at s ' the tip. Because magnetic forces are stretching forces, both tensions are equal and stretching, (0) = (0) > 0. If s ' (0) 6= (0), the tangential force equilibrium (C1) immediately leads to @ [ (0) (0)]=r for small r, resulting s ' r s ' s in a logarithmically diverging / ln r for small r contradicting niteness. The equality (0) = (0) at the tip also leads to isotropy of the stretches (0) = (0) at the tip because of s ' s ' the constitutive relations (13), however, not necessarily to niteness of the stretches at the tip. Therefore, we have to discuss the cases of nite and diverging stretches = at the conical tip separately. s ' 29 We start with nite isotropic stretches, (0) = (0) < 1. Then we can apply l'H^ opital's rule at the tip s = 0: s ' 0 r r cos[ (0)] (0) = lim = lim = = (0) cos[ (0)] (C2) ' s s !0 r s !0 r cos[ (0)] 0 0 0 0 [where we used (0) = 0 for the spherical rest shape]. Equality of the stretches (0) = (0) then leads to the 0 s ' conclusion (0) = 0, i.e., a sharp conical tip is impossible if stretches remain nite at the tip. L'H^ opital's rule can no longer be applied if the stretches diverge at the tip (remaining asymptotically isotropic), i.e., (s ) (s ) const s (C3) s 0 ' 0 for s 0 with an exponent > 0. Because of = r = cos , this requires r(s ) const s =(1 ) cos (0) for 0 s 0 s 0, whereas r (s ) = R sin(s =R ) s for the spherical rest shape. Then Eq. (C2) is replaced by 0 0 0 0 0 0 0 r consts 1 r cos[ (0)] (s ) = lim = = lim = (s ) ' 0 s 0 s !0 s !0 0 r (1 ) cos (0) 0 1 r 1 for s 0. The equality (s ) (s ) necessarily leads to the condition 0 s 0 ' 0 = cos[ (0)] 1 = sin 1 (C4) between the exponent of the divergent stretches and the half opening angle = =2 (0) of the conical tip. In conclusion, a deformation of the spherical rest shape into a sharp conical tip with (0) > 0 is only possible if stretches are asymptotically isotropic and diverge as (s ) (s ) s with an exponent , which is related s 0 ' 0 by Eq. (C4) to the opening angle 2 of the cone. Because of the nonlinear constitutive relation (13), diverging and isotropic stretches are compatible with nite and isotropic tensions at the tip with 2D (0) = (0) = : (C5) s ' Note that away from the tip (s > 0), tensions and stretches feature anisotropic corrections. 2. Governing equations for stretches and tensions in a conical shape with spherical rest shape In this section we present how to systematically calculate stretches and elastic tensions in a deformation from a spherical rest shape into a conical shape by deriving the governing equations. This is the basis of the generalization of the slender-body theory of Stone et al. from ferro uid conical droplets to capsules. We assume that the conical shape is given by a function r(z), where z runs from the bottom of the cone at z = a to its top at z = a. We will show that, if the conical shape r(z) is known, we can calculate all stretches and tensions 2 2 1=2 in this shape. The rest shape is spherical and parametrized analogously by a function r (z ) = (R z ) with 0 0 0 0 z 2 [R ; R ]. For the following it is advantageous to replace z and z by coordinates d = a + z measuring the 0 0 0 0 distance from the lower conical tip and d = R + z measuring the distance from the corresponding south pole of 0 0 0 the sphere. This geometry is illustrated in Fig. 12. Given a conical shape r(d) and the spherical rest shape 2 1=2 r (d ) = (2R d d ) ; (C6) 0 0 0 0 we want to show how the function d(d ) describing the stretching in the z direction can be calculated systematically from the tangential force equilibrium (C1) or (15) and the constitutive relations (13). If the conical shape r(d) and the function d(d ) are given [and the spherical rest shape r (d )] the meridional and hoop stretches can be calculated 0 0 0 as a function of d by r(d(d )) r(d(d )) 0 0 = = ; 2 1=2 r (d ) (2R d d ) 0 0 0 0 (C7) 1=2 0 2 2 1=2 1 + r (d(d )) ds 0 1=2 (2R d d ) 0 0 0 0 2 0 0 = = d (d ) = 1 + r (d(d )) d (d ); s 0 0 0 1=2 ds 2 R 0 [1 + r (d ) ] 0 0 30 (a) (b) z z r r z = 0 z = 0 (0) z = a z = R FIG. 12. Illustration of the geometry at the capsule's south pole (not true to scale) for (a) a conical tip and (b) the spheroidal reference shape. where d (d ) = dz=dz is the stretch in the z direction. z 0 0 If both stretches are known then the constitutive relations (13) can be used to express tensions and as s ' algebraic functions of the stretches and from Eq. (C7) and thus as functions of d , the conical shape r(d), and s ' 0 the unknown function d(d ) and its derivative. These tensions have to ful ll the tangential force equilibrium (C1), which we rewrite in terms of stretches using the constitutive relations (13) , = + r@ ; ' s r s 3 2 2 (1 + ) = (1 + ) + r f (@ ) (@ ) [ (1 + )]g : (C8) ' s ' s ' r s r ' s ' ' s Plugging in the stretches from (C7) and using 1 1 @ = @ = @ ; r d d 0 0 0 0 @ r r (d(d ))d (d ) d 0 0 we obtain a complicated nonlinear dierential equation for the unknown function d(d ) and its derivative (d ) = 0 z 0 d (d ). If this dierential equation can be solved, all stretches and tensions arising from the deformation from r (d ) 0 0 0 into r(d) are determined, in principle. Unfortunately, this equation cannot be solved in general. In the next section we obtain features of a solution close to the conical tip. 3. Stretches and tensions in the vicinity of a conical tip for a spherical rest shape In the vicinity of a the conical tip the conical shape r(d) with a half opening angle becomes strictly conical, and we can use r(d) = d tan ; (C9) resulting in stretches tan = d(d ); ' 0 1=2 (2R d d ) 0 0 (C10) 2 1=2 1 (2R d d ) 0 0 0 0 = d (d ): s 0 cos R Close to the conical tip, and are diverging and asymptotically equal according to Appendix C 1. Requiring s ' = for small d gives a dierential equation s ' 0 d (d ) = sin d(d ); (C11) 0 0 2R d d 0 0 0 31 which is solved by (sin )=2 (sin )=2 d d 0 0 (sin )=2 d(d ) = a a / d ; 2R d 2R 0 0 0 where we use a boundary condition d(R ) = a resulting from the conservation of the mirror symmetry plane at z = z = 0. This results in (sin )=2 r(d(d )) = tan [d(d )] a tan 0 0 2R and, using (C10), (sin 1)=2 11= sin a d a tan r = tan = : (C12) s ' 2R 2R 2R a tan 0 0 0 Noting that d R [1 cos(s =R )] s =2R for the spherical rest shape, the exponent in (C12) is exactly equivalent 0 0 0 0 0 to our above result (C4), = 1 sin , for the relation between the exponent of the divergent stretches (s ) s 0 (s ) s and the half opening angle of the conical tip. ' 0 Away from the tip, the stretches and tensions acquire anisotropic corrections. Therefore, we start with an ansatz ~ ~ = br + b r ; = br + b r ; s s ' ' (C13) 1+ = 1= sin 1; b (a tan ) =2R for small r in the vicinity of the conical tip, where < . We use this ansatz in the tangential force balance relation (C8) derived in Appendix C 2. First we obtain the tensions, which are isotropic and in agreement with (C5) to leading order but also acquire anisotropic corrections 1 b b 1 + (1 + )b b (b + b ) ~ ~ ~ ~ s ' s ' s ' 2 2 2 = 1 + + r r + r r ; 2 2 Y b b b b 2D 1 b b 1 + (1 + )b b (b + b ) ~ ~ ~ ~ ' s ' s s ' 2 2 2 = 1 + + r r + r r ; 2 2 Y b b b b 2D 3 2 neglecting terms O(r ). These expression are used in the tangential force balance relation (C8), = r@ , ' s r s in which we compare coecients order by order in r in order to determine the exponent and the coecients b and b . If we assume > 0 the leading order terms are O(r ), and comparing coecients gives a contradictory relation 2 = < 0. It follows that = 0; i.e., the leading anisotropic corrections in the stretches (C13) are constant. ~ ~ Continuing with = 0, terms O(r ) and O(r ) are of equal order and comparing all coecients gives (1 + ) b b = > 0; s ' 2 + i.e., the anisotropy close to the tip is such that > and > . For the tensions this results in s ' s ' Y 1 1 ~ 2D = 1 r ; 1 b 1 + =2 (C14) Y 1 1 2D ~ = 1 r ; 1 b 1 + which speci es the leading anisotropic corrections to Eq. (C5). Finally, we can compare coecients of all terms O(r ) for = 0 to obtain 2 2 ~ ~ b b + (1 + )(b b ) = 2 (1 + )b 2 b (b + b ) ' s s ' s ' ' s which can be used to go on and determine both b and b if needed. s ' 32 Appendix D: Discretization errors To observe the transition to a conical shape, it is necessary to have a high resolution for the nite element-boundary element method in the tip of the capsule. If we consider the number of boundary elements to be xed to N = 250, we can vary the density of elements near the tip by changing the parameter l [see Sec. II B 2 and Eq. (7)]. For dierent values of l , we see a quite dierent numerical behavior. Every result in the text above is calculated with l = 0:1. 0 0 For signi cantly smaller values of l , we cannot calculate conical shapes. The problem is that our shooting method for the elastic shape equations does not nd solutions anymore due to extremely high and rapidly changing stretch factors at the tip [ (s = 0) > 10 ]. On the other hand, with constant element density (l = 1), a shape transition s 0 0 cannot be found anymore; the capsule's shape stays rounded. This indicates that the numerical calculation of the shape transition is prone to changes of l . An example of this phenomenon can be seen in Fig. 13, which is identical to Fig. 9 but with additional data for l = 0:2. Lowering the elements' density at the tip leads to slightly dierent values for the critical Bond numbers and lowers the relative sizes of the hysteresis loops, especially for higher values of Y = . 2D 0.040 3.75 spheroidal approx. energy minimization 0.035 3.70 numerical data, l = 0.1 0.030 numerical data, l = 0.2 3.65 0.025 m, c2 3.60 0.020 3.55 0.015 m, c1 3.50 0.010 0.005 3.45 (a) (b) 0.000 3 2 1 0 1 2 3 0.00 0.01 0.02 0.03 0.04 0.05 10 10 10 10 10 10 10 Y / Y / 2D 2D FIG. 13. Comparison of data from Fig. 9 for l = 0:1 (blue) with data for l = 0:2 (red). There is an increasing deviation for 0 0 higher values of Y = . 2D [1] M. P. Neubauer, M. Poehlmann, and A. Fery, \Microcapsule mechanics: From stability to function," Adv. Colloid Interface Sci. 207, 65{80 (2014). [2] A. Fery, F. Dubreuil, and H. M ohwald, \Mechanics of arti cial microcapsules," New J. Phys. 6, 18 (2004). [3] O. I. Vinogradova, O. V. Lebedeva, and B.-S. Kim, \Mechanical behavior and characterization of microcapsules," Annu. Rev. Mater. Res. 36, 143{178 (2006). [4] C. Gao, E. Donath, S. Moya, V. Dudnik, and H. M ohwald, \Elasticity of hollow polyelectrolyte capsules prepared by the layer-by-layer technique," Eur. Phys. J. E 5, 21 (2001). [5] S. Sacanna, W. T. M. Irvine, L. Rossi, and D. J. Pine, \Lock and key colloids through polymerization-induced buckling of monodisperse silicon oil droplets," Soft Matter 7, 1631 (2011). [6] S. S. Datta, S.-H. Kim, J. Paulose, A. Abbaspourrad, D. R. Nelson, and D. A. Weitz, \Delayed buckling and guided folding of inhomogeneous capsules," Phys. Rev. Lett. 109, 134302 (2012). [7] S. Knoche and J. Kierfeld, \Buckling of spherical capsules," Phys. Rev. E 84, 046608 (2011). [8] S. Knoche and J. Kierfeld, \The secondary buckling transition: Wrinkling of buckled spherical shells," Eur. Phys. J. E 37, 62 (2014). [9] S. Knoche and J. Kierfeld, \Secondary polygonal instability of buckled spherical shells," EPL 106, 24004 (2014). [10] V. Jadhao, C. K. Thomas, and M. O. de la Cruz, \Electrostatics-driven shape transitions in soft shells," Proc. Natl. Acad. Sci. USA 111, 12673 (2014). [11] H.-H. Boltz and J. Kierfeld, \Shapes of sedimenting soft elastic capsules in a viscous uid," Phys. Rev. E 92, 033003 (2015). [12] C. I. Zoldesi, I. L. Ivanovska, C. Quilliet, Wuite G. J. L., and A. Imhof, \Elastic properties of hollow colloidal particles," Phys. Rev. E 78, 051401 (2008). m, c 33 [13] D. Vella, A. Ajdari, A. Vaziri, and A. Boudaoud, \The indentation of pressurized elastic shells: from polymeric capsules to yeast cells." J. R. Soc. Interface 9, 448{55 (2012). [14] H. Rehage, M. Husmann, and A. Walter, \From two-dimensional model networks to microcapsules," Rheol. Acta 41, 292{306 (2002). [15] D. Vella, A. Ajdari, A. Vaziri, and A. Boudaoud, \Wrinkling of Pressurized Elastic Shells," Phys. Rev. Lett. 107, 174301 (2011). [16] E. Aumaitre, S. Knoche, P. Cicuta, and D. Vella, \Wrinkling in the de ation of elastic bubbles." Eur. Phys. J. E 36, 22 (2013). [17] S. Knoche, D. Vella, E. Aumaitre, P. Degen, H. Rehage, P. Cicuta, and J. Kierfeld, \Elastometry of de ated capsules: Elastic moduli from shape and wrinkle analysis," Langmuir 29, 12463{12471 (2013). [18] G. Pieper, H. Rehage, and D. Barth es-Biesel, \Deformation of a capsule in a spinning drop apparatus," J. Colloid Interface Sci. 202, 293{300 (1998). [19] D. Barth es-Biesel, \Modeling the motion of capsules in ow," Curr. Opin. Colloid Interface Sci. 16, 3{12 (2011). [20] P. Degen, S. Peschel, and H. Rehage, \Stimulated aggregation, rotation, and deformation of magnetite- lled microcapsules in external magnetic elds," Colloid Polym. Sci. 286, 865 (2008). [21] R. B. Karyappa, S. D. Deshmukh, and R. M. Thaokar, \Deformation of an elastic capsule in a uniform electric eld," Phys. Fluids 26, 122108 (2014). [22] R. E. Rosenweig, Ferrohydrodynamics (Cambridge University Press, Cambridge, 1985). [23] P.A. Voltairas, D.I. Fotiadis, and C.V. Massalas, \Elastic stability of silicone ferro uid internal tamponade (s t) in retinal detachment surgery," J. Magn. Magn. Mater. 225, 248 (2001). [24] D. L. Holligan, G. T. Gillies, and J. T. Dailey, \Magnetic guidance of ferro uidic nanoparticles in an in vitro model of intraocular retinal repair," Nanotechnology 14, 661 (2003). [25] X. Liu, M. D. Kaminski, J. S. Rie, H. Chen, M. Torno, M. R. Finck, L. Taylor, and A. J. Rosengart, \Preparation and characterization of biodegradable magnetic carriers by single emulsion-solvent evaporation," J. Magn. Magn. Mater. 311, 84 (2007). [26] V.I. Arkhipenko, I. D. Barkov, and V. G. Bashtovoi, \Shape of a drop of magnetized uid in a homogeneous magnetic eld," Magnetohydrodynamics 14, 373 (1979). [27] J. C. Bacri and D. Salin, \Instability of ferrouid magnetic drops under magnetic eld," J. Phys. Lett. 43, 649 (1982). [28] M. D. Cowley and R. E. Rosensweig, \The interfacial stability of a ferromagnetic uid," J. Fluid Mech. 30, 671 (1967). [29] A. G. Boudouvis, J. L. Puchalla, L. E. Scriven, and R. E. Rosensweig, \Normal eld instability and patterns in pools of ferro uid," J. Magn. Magn. Mater. 65, 307 (1987). [30] H. Li, T. C. Halsey, and A. Lobkovsky, \Singular shape of a uid drop in an electric or magnetic eld," EPL 27, 575 (1994). [31] A. Ramos and A. Castellanos, \Conical points in liquid-liquid interfaces subjected to electric elds," Phys. Lett. A 184, 268{272 (1994). [32] H. A. Stone, J. R. Lister, and M. P. Brenner, \Drops with conical ends in electric and magnetic elds," Proc. Royal Soc. A 455, 329 (1999). [33] O. E. S ero-Guillaume, D. Zouaoui, D. Bernardin, and J. P. Brancher, \The shape of a magnetic liquid drop," J. Fluid Mech. 241, 215 (1992). [34] O. Lavrova, G. Matthies, V. Polevikov, and L. Tobiska, \Numerical modeling of the equilibrium shapes of a ferro uid drop in an external magnetic eld," Proc. Appl. Math. Mech. 4, 704 (2004). [35] S. Afkhami, A. J. Tyler, Y. Renardy, T. G. St. Pierre, R. C. Woodward, and J. S. Rie, \Deformation of a hydrophobic ferro uid droplet suspended in a viscous medium under uniform magnetic elds," J. Fluid Mech. 663, 358 (2010). [36] G.-P. Zhu, N.-T. Nguyen, R. V. Ramanujan, and X.-Y. Huang, \Nonlinear deformation of a ferro uid droplet in a uniform magnetic eld," Langmuir 27, 14834 (2011). [37] M. S. Korlie, A. Mukherjee, B. G. Nita, J. G. Stevens, A. D. Trubatch, and P. Yecko, \Modeling bubbles and droplets in magnetic uids," J. Phys. Condens. Matter 20, 204143 (2008). [38] J. Zeleny, \Instability of electri ed liquid surfaces," Phys. Rev. 10, 1 (1917). [39] C. T. R. Wilson and G. I. Taylor, \The bursting of soap-bubbles in a uniform electric eld," Math. Proc. Cambridge Philos. Soc. 22, 728 (1925). [40] G. Taylor, \Disintegration of water drops in an electric eld," Proc. R. Soc. London Ser. A 280, 383 (1964). [41] N. M. Zubarev, \Formation of conic cusps at the surface of liquid metal in electric eld," JETP Lett. 73, 544 (2001). [42] N. M. Zubarev, \Self-similar solutions for conic cusps formation at the surface of dielectric liquids in electric eld," Phys. Rev. E 65, 055301 (2002). [43] R. T. Collins, J. J. Jones, M. T. Harris, and O. A. Basaran, \Electrohydrodynamic tip streaming and emission of charged drops from liquid cones," Nat. Phys. 4, 149 (2008). [44] R. T. Collins, K. Sambath, M. T. Harris, and O. A. Basaran, \Universal scaling laws for the disintegration of electri ed drops," Proc. Natl. Acad. Sci. USA 110, 4905 (2013). [45] S. Neveu-Prin, V. Cabuil, R. Massart, P. Escare, and J. Dussaud, \Encapsulation of magnetic uids," J. Magn. Magn. Mater. 122, 42 (1993). [46] A. G. Boudouvis, J. L. Puchalla, and L. E. Scriven, \Magnetohydrostatic equilibria of ferro uid drops in external magnetic elds," Chem. Eng. Commun. 67, 129 (1988). [47] Osman A. Basaran and Fred K. Wohlhuter, \Eect of nonlinear polarization on shapes and stability of pendant and sessile drops in an electric (magnetic) eld," J. Fluid Mech. 244, 1 (1992). 34 [48] R. Chantrell, J. Popplewell, and S. Charles, \Measurements of particle size distribution parameters in ferro uids," IEEE Trans. Magn. 14, 975{977 (1978). [49] M. Costabel, \Symmetric methods for the coupling of nite elements and boundary elements (invited contribution)," in Mathematical and Computational Aspects , edited by C. A. Brebbia, W. L. Wendland, and G. Kuhn (Springer, Berlin, 1987) pp. 411{420. [50] W. L. Wendland, \On asymptotic error estimates for combined bem and fem," in Finite Element and Boundary Element Techniques from Mathematical and Engineering Point of View , edited by E. Stein and W. Wendland (Springer, Vienna, 1988) pp. 273{333. [51] D. N. Arnold and W. L. Wendland, \On the asymptotic convergence of collocation methods," Math. Comput. 41, 349 (1983). [52] D. N. Arnold and W. L. Wendland, \The convergence of spline collocation for strongly elliptic equations on curves," Numer. Math 47, 317 (1985). [53] J. A. Ligget and J. R. Salmon, \Cubic spline boundary elements," Int. J. Numer. Methods Eng. 17, 543 (1981). [54] O. Lavrova, V. Polevikov, and L. Tobiska, \Equilibrium shapes of a ferro uid drop," Proc. Appl. Math. Mech. 5, 837 (2005). [55] O. Lavrova, G. Matthies, T. Mitkova, V. Polevikov, and L. Tobiska, \Numerical treatment of free surface problems in ferrohydrodynamics," J. Phys. Condens. Matter 18, S2657 (2006). [56] M. T. Harris and O. A. Basaran, \Capillary electrohydrostatics of conducting drops hanging from a nozzle in an electric eld," J. Colloid Interface Sci. 161, 389 { 413 (1993). [57] B. Kornberger, \Fade2D Delauney Trianglulation library," (2016), http://www.geom.at/products/fade2d/. [58] L. C. Wrobel, The Boundary Element Method: Applications in Thermo uids and Acoustics, Vol. 1 (Wiley, New York, 1985). [59] O. Lavrova, Numerical methods for axisymmetric equilibrium magnetic- uid shapes, Ph.D. thesis, Otto-von-Guericke Uni- versity Magdeburg (2006). [60] L. J. Gray, L. F. Martha, and A. R. Ingraea, \Hypersingular integrals in boundary element fracture analysis," Int. J. Numer. Methods Eng. 29, 1135 (1990). [61] L. J. Gray, J. M. Glaeser, and T. Kaplan, \Direct evaluation of hypersingular galerkin surface integrals," SIAM J. Sci. Comput. 25, 1534 (2004). [62] A. Libai and J. G. Simmonds, The Nonlinear Theory of Elastic Shells (Cambridge University Press, Cambridge, 1998). [63] C. Pozrikidis, Modeling and Simulation of Capsules and Biological Cells (Chapman and Hall/CRC, Boca Raton, 2003). [64] B. Davidovitch, R. D. Schroll, D. Vella, M. Adda-Bedia, and E. A. Cerda, \Prototypical model for tensional wrinkling in thin sheets," Proc. Natl. Acad. Sci. USA 108, 18227 (2011). [65] R. A. Brown and L. E. Scriven, \The shape and stability of rotating liquid drops," Proc. Royal Soc. London Ser. A 371, 331 (1980). [66] R. Suryo and O. A. Basaran, \Local dynamics during pinch-o of liquid threads of power law uids: Scaling analysis and self-similarity," J. Non-Newton. Fluid Mech. 138, 134 (2006). [67] C. R. Anthony, P. M. Kamat, S. S. Thete, J. P. Munro, J. R. Lister, M. T. Harris, and O. A. Basaran, \Scaling laws and dynamics of bubble coalescence," Phys. Rev. Fluids 2, 083601 (2017). [68] K. N. Christodoulou and L. E. Scriven, \Discretization of free surface ows and other moving boundary problems," J. Comput. Phys 99, 39 (1992). [69] E. Zwar, A. Kemna, L. Richter, P. Degen, and H. Rehage, \Production, deformation and mechanical investigation of magnetic alginate capsules," J. Phys. Condens. Matter 30, 085101 (2018). [70] J. A. Stratton, Electromagnetic Theory (McGraw-Hill, New York, 1941). [71] S. Ramanujan, \Modular equations and approximations to ," Q. J. Math. 45, 350 (1914). [72] F. K. Wohlhuter and O. A. Basaran, \Shapes and stability of pendant and sessile dielectric drops in an electric eld," J. Fluid Mech. 235, 481 (1992). [73] T. A. Witten, \Stress focusing in elastic sheets," Rev. Mod. Phys. 79, 643{675 (2007). [74] M. Ben Amar and Y. Pomeau, \Crumpled paper," Proc. Royal Soc. A 453, 729{755 (1997). [75] E. Cerda and L. Mahadevan, \Conical Surfaces and Crescent Singularities in Crumpled Sheets," Phys. Rev. Lett. 80, 2358{2361 (1998). [76] J.-Y. Sun, X. Zhao, W. R. K. Illeperuma, O. Chaudhuri, K. H. Oh, D. J. Mooney, J. J. Vlassak, and Z. Suo, \Highly stretchable and tough hydrogels," Nature (London) 489, 133 (2012). [77] V. G. Bashtovoi, S. G. Pogirnitskaya, and A. G. Reks, \Determination of the shape of a free drop of magnetic uid in a uniform magnetic eld," Magnetohydrodynamics 23, 248 (1987). [78] R. Mercad e-Prieto, R. Allen, Z. Zhang, D. York, J. A. Preece, and T. E. Goodwin, \Failure of elastic-plastic coreshell microcapsules under compression," AIChE J. 58, 2674 (2012). [79] J. Hegemann, S. Knoche, S. Egger, M. Kott, S. Demand, A. Unverfehrt, H. Rehage, and J. Kierfeld, \Pendant capsule elastometry," J. Colloid Interface Sci. 513, 549{565 (2018). [80] L. D. Landau and E. M. Lifshitz, Theory of Elasticity (Pergamon, Oxford, 1970).

http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Condensed Matter arXiv (Cornell University)

http://www.deepdyve.com/lp/arxiv-cornell-university/spheroidal-and-conical-shapes-of-ferrofluid-filled-capsules-in-4OO2ru33gz

Spheroidal and conical shapes of ferrofluid-filled capsules in magnetic fields

Loading next page...

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

ISSN: 2469-990X
eISSN: ARCH-3331
DOI: 10.1103/PhysRevFluids.3.043603
Publisher site: See Article on Publisher Site

Abstract

Journal

Condensed Matter – arXiv (Cornell University)

Published: Mar 7, 2018

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Spheroidal and conical shapes of ferrofluid-filled capsules in magnetic fields

Spheroidal and conical shapes of ferrofluid-filled capsules in magnetic fields

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Spheroidal and conical shapes of ferrofluid-filled capsules in magnetic fields

Spheroidal and conical shapes of ferrofluid-filled capsules in magnetic fields

References

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies