See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/284338655

Derivations of the Young-Laplace equation

Research · November 2015

DOI: 10.13140/RG.2.1.4485.5768

CITATIONS READS
6 26,680

1 author:

Svein Magne Skjæveland

University of Stavanger
143 PUBLICATIONS 1,729 CITATIONS

SEE PROFILE

All content following this page was uploaded by Svein Magne Skjæveland on 22 November 2015.

The user has requested enhancement of the downloaded file.

 Derivations of the Young-Laplace equation

Leiv Magne Siqveland Svein Magne Skjæveland

June 9, 2014

Abstract
The classical Young-Laplace equation relates capillary pressure to
surface tension and the principal radii of curvature of the interface
between two immiscible uids.
In this paper the required properties of space curves and smooth
surfaces are described by (1) dierential geometry, and by (2) linear
algebra.
The equilibrium condition is formulated by a (1) force balance, and
(2) minimization of surface energy.

1 Introduction
The Young-Laplace equation (Young, 1805; Laplace, 1806)

1 1
 
pc = σ + , (1)
R1 R2
gives an expression for the capillary pressure pc , i.e., the pressure dierence
over an interface between two uids in terms of the surface tension σ and the
principal radii of curvature, R1 and R2 . This expression is often encountered
in the literature covering the concepts of capillary pressure and wettability
since it is quite general.
The expression in parenthesis in Eq. 1 is a geometry factor. At equilib-
rium, each point on the interface has the same geometry factor.
It will be shown that this simple expression reects the fact that for
arbitrary, smooth surfaces and curves (Shifrin, 2013), the curvature at any
point may be dened by assigning two radii of curvature, R1 and R2 , in two

1
normal planes that cut the interface along two principal curvature sections.
These two normal planes are also normal to each other and their line of
intersection is the surface normal at the chosen point. Also, the curvature
of an arbitrary normal section may be expressed in terms of the principal
curvatures.
With sucient knowledge of the mathematical properties of surfaces, the
Young-Laplace equation may easily be derived either by the principle of min-
imum energy or by requiring a force balance.
The properties of surfaces necessary to derive the Young-Laplace equation
may be found explicitly by dierential geometry or more indirectly by linear
algebra. The combination of these two approaches gives insight into the
properties of smooth space surfaces that are required for the simple form of
Young-Laplace equation.

2 Space curves by dierential geometry

Most of this section follows the exposition of space curves in the textbook
by Tambs Lyche (1962).
r
Let denote the radius vector from the origin of the Cartesian coordinate
ijk
system (x, y, z) with unit vectors ( , , ). A surface S may be dened by the
vector equation
r = f(u, v) = ϕ(u, v)i + ψ(u, v)j + χ(u, v)k, (2)
or in parameter form
x = ϕ(u, v), y = ψ(u, v), z = χ(u, v), (3)
where ϕ, ψ and χ are functions of the two parameters u and v . If the two rst
equations in Eq. 3 are solved for u and v and substituted in the third equation,
we get z expressed as a function of x and y , the usual way to represent a
surface. However, the parameter form is a very useful representation of a
surface to describe curvature characteristics.
r f
If we set u = u(t) and v = v(t) we get the vector equation = (t) for a
space curve on the surface, or in parameter form:
x = ϕ(t), y = ψ(t), z = χ(t),
where t is a parameter. By assumption, all functions are twice dierentiable
with continuous second order derivatives. A curve or surface represented by
functions fullling this requirement is said to be smooth.

2
Denitions
Arc Length f
If (t) is dierentiable with continuous derivative in the in-
terval [a,b], then the arc length L is dened by

ḟ
Z b
L= | (t)|dt,
a

where the dot denotes dierentiation with respect to t. If t ∈ [a, b] and we

ḟ
set s = at | (t)|dt, we get the arc dierential ds = | (t)|dt = ±|d |. Then s is
R
ḟ r
a continuous function of t that increases from 0 to L when t increases from a
to b. Instead of t, we can use s as a parameter to represent the curve. By this
arc length form, many formulas are especially simple, e.g., | 0 | = |d /ds| = 1. r r
Tangent to a Curve t r
The vector = d /ds = 0 is dened as the tangent r
r f t
vector of the space curve = (t). Since | | = 1, is a unity vector along the t
tangent of the curve.

Curvature The curvature K of a curve is dened by K = |dt/ds| =

|d r/ds | = |r |, or simply K = |f (s)|, the curve being on the arc length
2 2 00 00

form.

Radius of Curvature The radius of curvature R of a space curve C is

dened by R = 1/K .

Principal Normal to a Curve The principal normal h of a curve is

dened by h = r /|r | = r /K . Since r = 1 it follows that r · r = 0,
00 00 00 02 0 00

and hence h is normal to t (and the curve).

Normal of a Surface The surface normal n to a surface at a point is

dened by n = r × r /|r × r |. Here r and r denotes partial derivatives
of r with respect to u and v , cf. Eq. 2. The total dierential dr is given by
u v u v u v

dr = r du + r dv, u v

and for the space curve on the surface, u = u(t) and v = v(t). From the
t r t
denition of , d is along , and it is easily seen that d · = 0. That is, r n n
is normal to all curves on the surface drawn through the selected point.

3
Normal Plane and Normal Section A plane through the normal to a
surface, i.e., the normal is lying in the plane, is called a normal plane The
cut between a normal plane and the surface is a curve on the surface and is
called a normal section.

Curvature of a Normal Section

r f r f
Again, let = (u, v) a surface S and = (u(t), v(t)) a space curve C on S.
h t
From the denitions, we have K = d /ds. Multiplying by gives n
dt
ds
n = K cos θ,
where θ is the angle between the principal normal to C and the surface normal
at the chosen point P, Fig. 1.

t
P
C

S
θ
h
n

t
Figure 1: Surface S, curve C through point P, tangent to the curve, n
h
surface normal, and principal normal to the curve.

Since n · t = 0, we get by dierentiation

n dt + dn t = 0,
ds ds
and thereby
K=−
1 d nt
=−
1 d ·d
.
n r
cos θ ds cos θ ds2

4
 n
From the denition of , we have r n = 0, r n = 0.
u v Dierentiating with
respect to u and v , we get

r n + r n = 0, r n + r n = 0,
r n + r n = 0, r n + r n = 0.
u u uu v u uv
u v uv v v vv

Since
n n du + n dv, dr = r du + r dv,
d = u v u v

we have

n r
d ·d rn rn rn
= u u du2 + ( u v + v u )du dv + v v dv 2 rn
r n r n
= −( uu du2 + 2 uv du dv + vv dv 2 ), r n
and we get
1 L du2 + 2M du dv + N dv 2
K= , (4)
cos θ E du2 + 2F du dv + G dv 2
when

r r
ds2 = d 2 = ( u du + v dv)2 r
r
= 2u du2 + 2 u v dudv + rr r dv
2
v
2

and

E= r,
2
F= rr, G= r, 2

r n, r n, r n. (5)
u u v v
L= uu M= uv N= vv

We note that the quantities E, F, G, L, M, N only depend on properties

of the surface S with no reference to the space curve C on the surface. For
all curves C that start out from point P in the same direction, determined
by the ratio dv : du, the angel θ is the same according to Eq. 4. Conversely,
t
all space curves through P with the same and has the same curvature at h
P.
If we choose θ = 0, K is the curvature of a normal section, i.e., the
principal normal of the curve coincides with the normal to the surface,

L du2 + 2M du dv + N dv 2
K= . (6)
E du2 + 2F du dv + G dv 2

5
Principal Curvature Sections
If K is known, Eq. 6 is a quadratic equation for the ratio dv : du, and may
be written

(L − EK) du2 + 2(M − F K) du dv + (N − GK) dv 2 = 0. (7)

If this equation has two distinct roots, there will be two normal sections with
curvature K . If it has only one root, there exist only one normal section with
the given curvature, and if there are no roots, no normal section exists with
curvature K . To discern these alternatives, we consider the expression

(M − F K)2 − (L − EK)(N − GK)

that is under the square root sign when solving Eq. 7. This expression is
generally equal to zero for two values of K , the principal curvatures K1 and
K2 . The corresponding normal sections are called the principal curvature
sections.
After simplifying the last expression, we have to investigate the roots of

(EG − F 2 )K 2 − (EN − 2F M + GL)K + (LN − M 2 ) = 0. (8)

Solving this equation we have to nd the square root of

(EN − 2F M + GL)2 − 4(EG − F 2 )(LN − M 2 ).

As will be shown, this expression is never negative. Let us assume chosen

values for E, F, G, L, N such that the last expression is a function of M ,
denoted by ϕ(M ). It is a polynomial of second degree with the derivative

ϕ0 (M ) = −4F (EN − 2F M + GL) + 8(EG − F 2 )M,

and ϕ0 (M ) = 0 for M = M1 = F (EN +GL)/2EG. Then ϕ00 (M ) = 8EG > 0,

from the denition of E and G, i.e. ϕ(M ) has a minimum at M = M1 , and
after some calculation
(EG − F 2 )(EN − GL)2
ϕ(M1 ) = ≥ 0,
EG
rr rr
since EG − F 2 = 2u 2v − ( u v )2 = ( r ×r ) 2
≥ 0. Actually, we will assume
r or r
u v
that EG − F 2 > 0 since otherwise u v is the null vector or they are

6
parallel. Then ϕ(M ) can only be zero if EN = GL and M = M1 , i.e.
GM = F N . We then have
L N M
= = ,
E G F
and from Eq. 6 the curvature K is independent of du and dv and equal to
L/E . A point where the curvature is the same for all normal sections is
called a umbilical point of the surface.
For a point P on the surface that is not a umbilical point, Eq. 8 will have
two distinct roots, K1 and K2 , as postulated above.

Principal Curvature Sections are Orthogonal

Substitution of K = K1 or K = K2 into Eq. 7 results in a quadratic expres-
sion of the general form (Adu + Bdv)2 , since the equation has single roots
for these values of K . Its derivative with respect to dv then has to be zero
for the same values of K , that is

(M − F K)du + (N − GK)dv = 0,

or
M du + N dv
K= .
F du + Gdv
Substituting this expression into Eq. 7, we get

(EM − F L)du2 + (EN − GL)dudv + (F N − GM )dv 2 = 0.

From this equation we get the two directions dv1 : du1 and dv2 : du2 (or the
inverted ratios if F N − GM = 0), for the two principal curvature sections.
Using rules for the sum and product of the roots of a quadratic equation, we
get
dv1 dv2 EN − GL dv1 dv2 EM − F L
+ =− , = .
du1 du2 F N − GM du1 du2 F N − GM
We also have

d r 1 = r du + r dv , dr
u 1 v 1 2 = r du + r dv ,
u 2 v 2

7
and hence

d r · dr
1 2 = r" du du + r r (du !dv
2
u 1 2 u v 1 2
#
r
+ du2 dv1 ) + 2v dv1 dv2
dv1 dv2 dv1 dv2
= E+F + +G du1 du2
du1 du2 du1 du2
EN − GL EM − F L
 
= E−F +G du1 du2
F N − GM F N − GM
E(F N − GM ) − F (EN − GL) + G(EM − F L)
= du1 du2
F N − GM
= 0,

i.e. the principal curvature sections are orthogonal. (One can easily show
that this is the case also for F N − GM = 0).

A Theorem of Euler
A theorem of Euler (Weatherburn, 1947) states that the curvature of an
arbitrary normal section may be expressed by the curvatures of the principal
sections. Let ds1 and ds2 be the arc dierentials of the two principal sections
and ds the arc dierential in a normal section at an angle α with ds1 , Fig. 2.

R
ds
ds 2

α ds 1
P Q

Figure 2: Arc dierentials along a normal section and the two principal
curvature sections.

Generally, if Φ(u, v) is a function of u and v , we have

Φ(R) − Φ(P) = Φ(R) − Φ(Q) + Φ(Q) − Φ(P),

or
Φ(R) − Φ(P) Φ(R) − Φ(Q) ds1 Φ(Q) − Φ(P) ds2
= + ,
ds ds1 ds ds2 ds

8
and letting ds1 and ds2 approach zero,
dΦ dΦ ds1 dΦ ds2 dΦ dΦ
= + = cos α + sin α.
ds ds1 ds ds2 ds ds1 ds2
We now apply this general expression to r and n and get
dr
t =
ds
= t cos α + t sin α
1 2

dn dn dn
= cos α + sin α,
ds ds1 ds2
and by scalar multiplying these two expressions,

−K = t ddsn
dn n dn dn
t ds t t ds sin α cos α + t ds
!
2 d 2
= 1 · cos α+ 1 + 2 · sin 2 α
1 ds2 1 2
dn dn
sin α + t +t
!
= −K1 cos2 α − K2 2
sin α cos α.
1 2
ds2 ds1
Since n·t 1 = n · t = 0, we get
2

dn t = 0, t dn + n t
t ds +n
1
ds 2 ds
1
ds 2
2
1
2

1
= 0.

The curves C1 and C2 are embedded in two orthogonal planes, 1 · 2 = 0, t t

t t
and 1 is independent of s2 . Therefore d 1 /ds2 = 0 and likewise d 2 /ds1 = 0, t
and we get Euler's result
K = K1 cos2 α + K2 sin2 α. (9)
Let us now choose another normal section at an angle α + π/2 with ds1 and
denote the corresponding arc dierential by ds⊥ since it is at an angle π/2
with ds. For the corresponding curvature K⊥ we get from Eq. 9
K⊥ = K1 cos2 (α + π/2) + K2 sin2 (α + π/2)
= K1 sin2 α + K2 cos2 α.
By summation, we the get
K + K⊥ = K1 + K2 , (10)
that is, the sum of the curvatures of two orthogonal normal sections is con-
stant, equal to the sum of the curvatures of the principal sections.

9
3 The Young-Laplace Equation
The Young-Laplace equation may be derived either by minimization of energy
or by summing all forces to zero. We will do both here although the concept
of force in connection with surface tension may be somewhat obscure. The
force approach follows the derivation of Defay and Prigogine (1966) and the
energy approach is taken from the book by Landau and Lifshitz (1987). In
both cases it is assumed that the interface is without thickness and that the
interfacial tension is constant.

Force Balance
Consider a point P on the surface, Fig. 3, and draw a curve at a constant
distance ρ from P. This curve forms the boundary of a cap for which we shall
nd the equilibrium condition as ρ tends to zero.

P
B ρ ρ
δl
C A
σδl
R1
R2

Figure 3: Equilibrium of a nonspherical cap.

Through P we draw the two principal curvature sections AB and CD on

the surface. Their radii of curvature at P are R1 and R2 . At the point A, an

10
element δl of the boundary line is subjected to a force σδl whose projection
along the normal PN is
ρ
σδl sin φ ' σφδl = σ δl,
R2
since φ by assumption is small.
If we consider four elements δl of the periphery at A, B, C, and D, they
will contribute with a force
1 1
 
2ρσδl + .
R1 R2
Since this expression by Euler's theorem, Eq. 10, is independent of the choice
of AB and CD, it can be integrated around the circumference. Since four
orthogonal elements are considered, the integration is made over one quarter
of a revolution to give
1 1
 
2
πρ σ + .
R1 R2
The force on the surface element caused by the pressure dierence over
the surface is given by (p1 − p2 )πρ2 , and equating the last two expressions
Laplace's equation follows.

Minimum Energy
Let the surface of separation undergo an innitesimal displacement. At each
point of the undisplaced surface we draw the normal. The length of the seg-
ment of the normal lying between the points where it intersects the displaced
and undisplaced surfaces is denoted by δζ . Then a volume element between
the two sufaces is δζdf , where df is a surface element. Let p1 and p2 be the
pressures in the two media, and let δζ be positive if the displacement of the
surface is towards medium 2 (say). Then the work necessary to bring about
the change in volume is Z
(−p1 + p2 )δζdf.
The total work δW in displacing the surface is obtained by adding to
this the work connected with the change in area of the surface. This part of
the work is proportional to the change δf in area of the surface, and is σδf ,
where σ is the surface tension. Thus the total work is
Z
δW = − (p1 − p2 )δζdf + σδf. (11)

11
The condition for thermodynamical equilibrium is, of course, that δW be
zero.
Next, let R1 and R2 be the principal radii of curvature at a given point
of the surface. We set R1 and R2 as positive if they are drawn into medium
1. Then the elements of length (the arc dierentials) ds1 and ds2 on the
surface in its principal curvature sections are increased to (R1 + δζ)ds1 /R1
and (R2 + δζ)ds2 /R2 when the angles ds1 /R1 and ds2 /R2 remain constant,
i.e., an expansion normal to the surface (ds1 is the arc length of a circle
with radius R1 , and correspondingly for ds2 ). Hence the surface element
df = ds1 ds2 becomes, after displacement,
ds1 (1 + δζ/R1 )ds2 (1 + δζ/R2 ) ∼
= ds1 ds2 (1 + δζ/R1 + δζ/R2 ),
i.e. it changes by δζdf (1/R1 + 1/R2 ). Hence we see that the total change in
area of the surface of separation is
1 1
Z  
δf = δζ + df. (12)
R1 R2
Substituting these expressions in Eq. 11 and equating to zero, we obtain the
equilibrium condition in the form
1 1
Z   
δζ (p1 − p2 ) − σ + df = 0.
R1 R2
This condition must hold for every innitesimal displacement of the surface,
i.e. for all δζ . Hence the expression in braces must be identically equal to
zero and Young-Laplace's equation follows.

4 Space curves by linear algebra

We assume the space surface dened by z = f (x, y) to be smooth (Shifrin,
2013). A Taylor expansion around a point (a, b) gives an approximation to
the surface around (a, b),
∂f (a, b) ∂f (a, b)
f (x, y) = f (a, b) + (x − a) + (y − b)+
∂x ∂y
1 ∂ 2 f (a, b) ∂ 2 f (a, b)
"
2
2
(x − a) + 2
(y − b)2 +
2! ∂x ∂y
2
∂ 2 f (a, b)
#
∂ f (a, b)
(x − a)(y − b) + (y − b)(x − a) + · · · . (13)
∂x∂y ∂y∂x

12
 Z
z (a,b)

Y
X

Figure 4: Surface in space.

A new coordinate system (XY Z) is now introduced with origin in (a, b)

and the (XY ) plane dened as the tangent plane to the surface at (a, b).
This gives

f (0, 0) = 0,
∂f (0, 0)
= 0, (14)
∂X
∂f (0, 0)
= 0.
∂Y
Since the surface is smooth, the order of dierentiation is arbitrary, and the
crossterms may be added to render
1 ∂ 2 f (0, 0) 2 ∂ 2 f (0, 0) 2 ∂ 2 f (0, 0)
" #
Z = f (X, Y ) ≈ X + Y + 2 XY ,
2! ∂X 2 ∂Y 2 ∂X∂Y
1h i
= fXX X 2 + fY Y Y 2 + 2fXY XY .
2
This may be reformulated as the matrix product
! !
1  fXX fXY X
f (X, Y ) ≈ X Y . (15)
2 fXY fY Y Y
The 2 × 2 matrix in Eq. 15 formed by the partial derivatives of f is symmet-
rical, the matrix can be diagonalized with orthogonal eigenvectors (Howard,

13
1984), and the surface may be approximated by the matrix product
! !
1  α 0 ξ
f (X, Y ) ≈ g(ξ, η) = ξ η ,
2 0 β η
1 2 1 2
= αξ + βη , (16)
2 2
Here ξ and η are the coordinates along the new unit vectors and α and β
the corresponding eigenvalues. This is equivalent to a rotation of the tangent
plane around the Z -axis with the two new unit vectors (1 0)t and (0 1)t
in the (ξη)-plane.
In the (Zξ)-plane (η = 0), the function g(ξ, 0) in Eq. 16 will form the
parabola
1
Z = g(ξ, 0) = αξ 2 . (17)
2
Approximating the parabola with a circle of radius Rα , we get

ξ 2 + (Z − Rα )2 = Rα2 ,
ξ 2 + Z 2 − 2ZRα + Rα2 = Rα2 ,
ξ 2 + Z 2 − 2ZRα = 0, (18)

and solved with respect to Z ,

q
Z = Rα ± Rα2 − ξ 2 ,
v
ξ2
u
u
= Rα (1 ± t1 − ),
Rα2
ξ2
= Rα (1 ± [1 + − . . .]).
2Rα2
Selecting the minus sign and deleting higher order terms, we arrive a the
simple expresson

ξ2
Z ≈ . (19)
2Rα
By comparing Eq. 17 and Eq. 19 we nd

αξ 2 ξ2
Z= = ,
2 2Rα

14
and
1
Rα = . (20)
α
The curvature κ of a space curve at a point is dened in Sec. 2 as the
inverse of the radius of curvature at the point. The curvature of the parabola
Z = g(ξ, 0) is therefore
1
κ=α= .
Rα
Considering instead the (Zη)-plane (ξ = 0), the curvature of the parabola
Z = g(0, η) is given by
1
κ=β= .
Rβ

A arbitrary plane normal to the tangent plane at the point (a, b), see Sec. 2,
will cut the tangent plane (ξη) along a straight line l : ξ = Kη where K is
a constant as shown i Fig. 5 The distance λ between the point (a, b) and a

ξ
ξ=Kη

(ξ,η)
(a,b) l
η

Figure 5: Arbitrary line l through point (a, b) in the tangent plane.

point (ξ, η) on l may then be expressed by

λ2 = ξ 2 + η 2 ,
2 2
= K

η + η2 ,
= K 2 + 1 η2. (21)

15
Solved with respect to η 2 we get
λ2
2
η = 2 . (22)
K +1
The cut of the normal plane and the surface f (ξ, η) is then
1 2 1 2
f (ξ, η) = αξ + βη ,
2 2
1 1
= αK 2 η 2 + βη 2 ,
2 2!
2 2
λ2
!
1 K λ 1
= α + β ,
2 K2 + 1 2 K2 + 1
K2
" #
1 1
= α+ 2 β λ2 . (23)
2 K2 + 1 K +1
The curvature of the space curve dened by the cut between the normal plane
touching the surface in (a, b) and the surface f (ξ, η) is then, as shown above,
given by
K2 1
κ1 = 2 α+ 2 β. (24)
K +1 K +1
The expression for the curvature is a weigthed average between α and β . The
value of κ1 lies between α and β , i.e., between the largest and the smallest
curvature. Hence the two normal planes containing the space curve with the
largest and smallest curvature are normal to each other. These two space
curves are called the principal curves.
A normal plane that cuts the tangent plane along the line ξ = Kη will
cut the surface along a curve on the surface, the normal section, see Sec. 2,
with curvature κ1 . Another normal plane that is normal to the rst one will
cut the tangent plane in the line ξ = −η/K and have a normal section with
curvature
1 K2
κ2 = 2 α+ 2 β. (25)
K +1 K +1
Adding the two curvatures from Eqs. 24 and 25, vi get

κ1 + κ2 = α + β. (26)

The sum of the curvatures of two normal sections in planes also normal to
each other is constant and equal to the sum of the curvatures of the principal
curves.

16
5 Surface energy and the Young-Laplace equa-
tion
dl'1
dS'
dl'2
(a,b)
δζ
dS dl2
dl1
R2
θ2

R1

θ1

Figure 6: Displaced surface between two phases.

We now consider the surface between two phases to be intesimally dis-

placed by δζ . The volume element between the two surfaces is δζ · dS where
dS is the surface element. Let P1 and P2 denote the pressures in the two
phases. The work done by the volume change is
Z
δWp = (−P1 + P2 ) δζdS. (27)

The total work of the displacement also includes the work of changing the
surface area by δS and is given by

δWσ = σδS, (28)

where σ is the surface tension. And the total work is

Z
δW = (P2 − P1 ) δζdS + σδS. (29)

17
At thermodynamic equilibrium this work is equal to zero.
It remains to express the surface area change δS ,
δS = dS 0 − dS, (30)
in terms of the displacement δζ and the curvatures of the principal normal
sections, 1/R1 and 1/R2 . The surface areas before and after the displacement,
dS and dS 0 , respectively, are equal to the product of the length elements along
the principal normal sections since, as shown earlier, the normal sections are
normal to each other. Then,
dS = dl1 dl2 ,
dS 0 = dl10 dl20 , (31)
where dl1 , dl2 and dl10 , dl20 are the length elements long the principal normal
sections before and after the displacement, respectively. The length elements
dl1 and dl10 may be written as
dl1 = R1 θ1 ,
dl10 = (R1 + δζ)θ1 ,
where θ1 is the angle shown in Fig. 6. Then
dl10 R1 + δζ
=
dl1 R1
δζ
= 1+ .
R1

Similar expressions are valid for dl2 and dl20 . We substitute for dl10 and dl20 in
Eq. 31 and get
dS 0 = dl10 dl20 ,
! !
δζ δζ
= 1+ dl1 1 + dl2 ,
R1 R2
! !
δζ δζ
= dl1 dl2 1 + 1+ ,
R1 R2
δζ 2
!
δζ δζ
= dS 1 + + + ,
R1 R2 R1 R2
!
δζ δζ
≈ dS 1 + + , since δζ 2 << R1 R2 . (32)
R1 R2

18
If this expression is substituted into Eq. 30 we get

δS = dS 0 − dS,
!
δζ δζ
= dS 1 + + − dS,
R1 R2
!
δζ δζ
= dS + . (33)
R1 R2

This expression is inserted for δS in Eq. 29 to give

1 1
Z Z  
δW = (P2 − P1 ) δζdS + σ + δζdS
R1 R2
Z 
1 1
 
= −Pc + σ + δζdS,
R1 R2
= 0, for alle δζ, (34)

where Pc = P1 − P2 is the capillary pressure. Then, according to the funda-

mental lemma of calculus of variations Papatzacos (1989),

1 1
−Pc + σ( + ) = 0,
R1 R2
and rearranged we get
1 1
 
Pc = σ + , (35)
R1 R2
which is the Young-Laplace equation.

6 Conclusion
The classical Young-Laplace equation has been derived by detailing the neces-
sary properties of space curves combined with force balance and minimization
of surface energy.

Nomenclature
[a, b] = interval

19
 )
EF G
= parameters dened by Eq. 5.
LM N
f = area, m2
K = curvature, m−1
L = arc length, m
dl1 , dl2 = length elements, m
dl10 , dl20 = length elements, m
p = pressure, Pa
R = radius of curvature, m
R1 , R2 = principal radii of curvature, m
s = arc length parameter, m
t = parameter, dimensionless
u = parameter, dimensionless
v = parameter, dimensionless
(x, y, z) = Cartesian coordinates
W = work, J
α = angle, radians
δ = dierential operator
δζ = innitesimal displacement of surfaces, m
σ = surface tension, N/m
θ = angle, radians
ς = length element along normal, m
ϕ = function of (u, v)
ψ = function of (u, v)
χ = function of (u, v)
ρ = radius of cap, m
φ = angle, radians

Subscripts
c = capillary
u = partial derivative with respect to u
v = partial derivative with respect to v
α = constant
Γ = adsorption (kg surfactant/kg rock)
γ = interfacial tension, N/m

20
 Vectors
f = vector function, m
r = radius vector, m
t = tangent vector, dim.less
i, j, k = unit vectors, dim.less
h = principal normal to a curve, dim.less
n = surface normal, dim.less

References
Defay, R., Prigogine, I., 1966. Surface Tension and Adsorption. Longmans,
Green & Co Ltd, London. pp. 68.

Howard, A., 1984. Elementary Linear Algebra. John Wiley and Sons.

Landau, L., Lifshitz, E., 1987. Fluid Mechanics. Pergamon Press, Oxford.
pp. 23839.

Laplace, P., 1806. Supplement to the tenth edition. Méchanique céleste 10.

Papatzacos, P., 1989. Matematisk modellering. Kompendium ved Høgskolen

i Stavanger. In Norwegian.

Shifrin, T., 2013. Dierential Geometry: A First Course in Curves and

Surfaces. University of Georgia. Preliminary Version.

Tambs Lyche, R., 1962. Matematisk Analyse II. Gyldendal Norsk Forlag,
Oslo. In Norwegian.

Weatherburn, C., 1947. Dierential Geometry. Cambridge.

Young, T., 1805. An essay on the cohesion of uids. Phil. Trans. , pp 65.

21

View publication stats