ChemTexts (2015) 1:16

DOI 10.1007/s40828-015-0015-z

LECTURE TEXT

Basic interfacial thermodynamics and related mathematical

background
Gy} o G. Láng1
oz}

Received: 31 May 2015 / Accepted: 11 August 2015 / Published online: 8 September 2015
 Springer International Publishing 2015

Abstract This study is intended to give the reader a brief Introduction

overview of the mathematical background to the thermody-
namic theory of surfaces and interfaces. Some general aspects Despite the advent of surface-sensitive techniques, ther-
of the thermodynamics of systems with interfaces are dis- modynamic measurements remain a valuable tool for the
cussed, and a concise treatment of interfaces within the investigation of surfaces and interfaces.
framework of classical thermodynamics is provided. First, to Many (heterogeneous) systems of practical importance
introduce the reader to the topic, a reasonably simple ther- can only be described by complex thermodynamic models
modynamic treatment of interfaces, together with a brief including interface(s). For instance, electrodes are, in fact,
description of the models widely used in the literature, is capillary systems, because the interactions between the
presented, the characteristics of the Gibbs ‘‘dividing plane’’ different phases take place via the surface region. Thus, the
model and the Guggenheim ‘‘interphase’’ model are outlined, understanding of the thermodynamics of these interfaces is
and a derivation of the Gibbs adsorption equation is given. In of importance to all surface scientists and electrochemists.
the subsequent sections, several important mathematical The aim of the present study is to give a brief overview
concepts (e.g., theory of homogeneous functions and partly of the mathematical background, some important mathe-
homogeneous functions, Euler’s theorem and the Gibbs– matical concepts, and various functional relationships
Duhem equation, Legendre transformation) and various underlying the thermodynamic theory of interfaces (theory
functional relationships of the thermodynamics of surfaces and of homogeneous functions and partly homogeneous func-
interfaces are summarized, with particular attention to some tions, Euler’s theorem and the Gibbs–Duhem equation,
alternative formulations. Some of the mathematical principles Legendre transformation). The mathematical principles are
and methods are explained using illustrative examples. explained and illustrated with some typical examples.
First, to introduce the reader to the topic, a reasonably
Keywords Interfacial thermodynamics  Gibbs model  simple thermodynamic treatment of interfaces, together
Guggenheim model  Homogeneous functions  Partly with a brief description of the models widely used in the
homogeneous functions  Euler’s theorem  Legendre literature, is presented (more detailed discussions can be
transformation  Gibbs–Duhem equation found in several reviews and research papers [1–16]).

Electronic supplementary material The online version of this

article (doi:10.1007/s40828-015-0015-z) contains supplementary Basic concepts and notions related
material, which is available to authorized users.
to the thermodynamics of interfaces
& Gy}oz}
o G. Láng
[email protected] Models of the interfacial region
1
Institute of Chemistry, Laboratory of Electrochemistry and
Interfacial thermodynamics is the study of the application
Electroanalytical Chemistry & Department of Physical
Chemistry, Eötvös Loránd University, Budapest Pázmány of thermodynamics to interfacial phenomena, addressing
P. s. 1/A, Budapest 1117, Hungary topics, including adsorption, interfacial energies, interfacial

123
16 Page 2 of 17 ChemTexts (2015) 1:16

tension, and superficial charge, and about relations among surface [19]. Gibbs found it mathematically convenient to
them [see e.g., 1–18]. Adsorption of one or more of the consider an idealized system depicted in Fig. 1b, with
components, at one or more of the phase boundaries of a properties identical with those of the whole real system.
multicomponent, multiphase system, is said to occur if the The ‘‘surface of discontinuity’’ or ‘‘dividing surface’’ in the
concentrations in the interfacial layers are different from idealized system is a two-dimensional region whose posi-
those in the adjoining bulk phases. Consequently, the tion is determined by the requirements that the property
overall stoichiometry of the system deviates from that under consideration should maintain a uniform value in
corresponding to a reference system of (hypothetical) each bulk phase right up to the dividing surface. This
homogeneous bulk phases whose volumes and/or amounts corresponds to equating the two shaded areas in Fig. 1b. A
are defined by suitably chosen dividing surfaces, or by a disadvantage of this approach is that the position of the
suitable algebraic method (see later). dividing surface alters according to the property
The classic work is that of Gibbs [19]; a paper by considered.
Guggenheim and Adam [20] discusses the physical inter- In the Guggenheim model, two dividing surfaces, one at
pretation of surface excesses, and Guggenheim [21] has each boundary, are employed (Fig. 1c). It is assumed that
given a good summary of interfacial thermodynamics there is a ‘‘surface’’ or ‘‘interfacial’’ layer of finite thickness
emphasizing a viewpoint somewhat different from that of (s) bounded by two appropriately chosen surfaces parallel to
Gibbs. the phase boundary, one in each of the adjacent homoge-
In many studies, the plane ideally marking the boundary neous bulk phases. A layer of this kind is sometimes called a
between two phases is called the interface. Although Guggenheim layer or ‘‘interphase.’’ A disadvantage is that
interfaces are always dealt with from a thermodynamic terms dependent on surface volume are present in the
point of view, if attention is actually focused on only one of equations, but it is difficult to assign values to these terms. (It
the two phases, the plane marking the boundary between should be noted that for very highly curved surfaces, i.e.,
the phase and the environment is often called the surface of when the radius of curvature is of the same magnitude as s,
the phase (see e.g., [22]). The two words (i.e., ‘‘interface’’ the notion of a surface layer may lose its usefulness.)
and ‘‘surface’’) are often used synonymously, although Given a system, subsystems consisting of a segment of
interface is preferred for the boundary between two con- the interface and finite volumes of the adjacent phases can
densed phases and in cases where the two phases are be selected. In principle, these subsystems should not be
named explicitly, e.g., the solid/gas interface [23]. geometrically regular in shape; however, the rectangular
The region between two phases where the properties parallelepiped-shaped domain is usually the most expedi-
vary between those in the bulk is the ‘‘interfacial’’ or ent selection. In two dimensions, the macroscopic subsys-
‘‘interface’’ region, and sometimes it is called the tem selected for investigation is represented by the ABCD
‘‘interphase.’’ rectangle (Fig. 1).
An interface or interphase does not exist in isolation, Usually, the thickness of the interface or local values of
and valid thermodynamic conclusions can only be drawn physical quantities (parameters) cannot be measured. That
by considering the system, namely the interface and the is the reason why integrated quantities (which are acces-
two bordering regions, as a whole. Provided that the radius sible experimentally, or can be calculated from experi-
of curvature is large, the interface/interphase may be mental data) are used for the thermodynamic
regarded as a plane and its energy then differs from that of characterization of interfaces. Generally, these quantities
a bulk phase by a term expressing the contribution of are given by the expression:
changes of energy due to a change of the area of contact. Zbb
Edge effects can be eliminated by considering a section of r
W ¼ Y ðnÞ dn; ð1Þ
an interface in a larger system. There is no clear boundary
aa
between the interfacial region and the bulk of the phases,
so that the thickness of the interphase depends on the where n is the coordinate perpendicular to the plane of the
model chosen to describe this region. The geometric area is interface, Y is the function of n, Wr is the integrated
represented by the product of the length and breadth of a physical quantity, and aa and bb are the two adjacent
rectangle enclosing part of a surface. Many properties of a (homogeneous) phases.
system, for example, concentration of a particular species, Let the area of the surface or interface in the system
vary as a function of the distance perpendicular to the defined according to the above concepts be denoted by A,
surface, as shown in Fig. 1. and the internal energy by U. The volume V of the system
The classical Gibbs approach is based on a model in is the sum of the volumes of the two phases aa and bb, and
which a real interface layer is replaced by a dividing the volume of the inhomogeneous region is as follows:

123
ChemTexts (2015) 1:16 Page 3 of 17 16

Fig. 1 A schematic representation of the interfacial region, the Gibbs interface, and c the Guggenheim model of the interface. On the right-
‘‘dividing surface’’ (‘‘surface of discontinuity’’ or ‘‘mathematical hand side: the macroscopic subsystems selected for investigation are
plane’’) and the ‘‘interfacial layer’’ concept (‘‘interphase’’) proposed represented by the ABCD rectangles
by Guggenheim. a The real system, b the Gibbs model of the

V ¼ V aa þ V bb þ V inh : ð2Þ the (arbitrary) choice of the dividing surface(s). In the

Guggenheim model, the volume Vr of the interfacial layer is
The internal energy can be given as
V r ¼ sA: ð4Þ
aa bb inh
U¼U þU þU : ð3Þ
The Gibbs dividing surface (or Gibbs surface) is a
Of course, this division is completely arbitrary, since the geometrical surface chosen parallel to the interface and
values on the right-hand sides of Eqs. (2) and (3) depend on used to define the volumes of the bulk phases. That is

123
16 Page 4 of 17 ChemTexts (2015) 1:16

V ¼ V aa þ V bb : ð5Þ
This means that the volume of the ‘‘surface phase is’’
Vr : 0.

Adsorption

As already discussed above, the Gibbs interface is a two-

dimensional homogeneous phase without thickness (i.e.,
the interface is regarded as a mathematical dividing sur-
face). In Guggenheim’s approach, the interface is consid-
ered to be a surface phase with finite thickness and volume
treated in a way analogous to bulk phases, except that the
thermodynamic equations contain terms related to the
contributions of changes of energy due to changes of area
and electrical state of the interface.
The two apparently different approaches can be essen-
tially characterized by the following procedure:
Fig. 2 Scheme of the ‘‘real system’’ with the inhomogeneous
a) There is an idealized surface or surface phase sepa- ‘‘interfacial region’’ (inh) and the ‘‘model system.’’ In the Gibbsian
rating two homogeneous bulk phases (see Fig. 1). model, there is a hypothetical ‘‘dividing surface’’ (r) (in the
The bulk phases are in equilibrium with the surface Guggenheim model a hypothetical, three-dimensional surface phase,
phase. see Fig. 1) which is separating two homogeneous bulk phases aa and
bb (that are in equilibrium with the surface region). The two reference
b) Two separated reference systems a and b thought to systems are a and b
be noninteracting homogeneous bulk phases have to
be chosen (see Fig. 2), the conditions of temperature,
pressure, and composition being identical to those in reference amounts is not necessarily equal to the volume of
the adsorption equilibrium. Both reference phases the real system. It is even not necessary that the corre-
consist of suitably defined amounts of the compo- sponding phases are effectively present in their chosen
nents. Each of the chosen reference amounts is reference states within the real system. In principle, this is
characterized by its respective molar or specific why the Gibbs and the Guggenheim approaches can be
properties. considered as equivalent. Nevertheless, there is an impor-
c) Any extensive property of the reference systems is tant restriction in the Guggenheim approach replacing the
simply the sum of the contributions from the condition of equivalent volumes in the Gibbs method: the
reference amounts, without any contributions from reference systems must be chosen in such a manner that the
interactions with the interfacial layer in the real remaining ‘‘surface phase’’ has a constant thickness. Thus,
system. this restriction essentially affects the choice of the geo-
metrical shape of the reference systems. However, since
The surface excess quantities are then the respective the reference systems are homogeneous bulk phases, their
differences between the real system and the chosen refer- thermodynamic properties are independent of the shape.
ence systems (or reference phases). As explained above, For this reason, a set of appropriate reference systems can
this can be done, e.g., in the Gibbs sense of the total be always selected without loss of generality. This con-
extensive quantity minus its amount residing in hypothet- sideration determines implicitly the selection of thermo-
ical bulk phases that are uniform up to a mathematical dynamic systems ‘‘with cylindrical shape’’ [24], a
dividing surface, or in the Guggenheim sense of excesses ‘‘parallelepiped’’ [25], or simply as a ‘‘section’’ of the
over the average bulk amount in a boundary zone (‘‘surface interface cut out by perpendicular planes [26–28].
phase’’) of finite but small thickness. Obviously, if the The surface excess amount or Gibbs adsorption of
Gibbs model is used for the definition of surface excesses, component i is nri , which may be positive or negative, and
the reference amounts in the two reference phases are is defined as the excess of the amount of this component
thought to be contained in and making up the volume of the actually present in the system over that present in a refer-
actual real system, but can equally well be thought to be ence system of the same volume as the real system and in
quite independent and spatially apart one from the other. which the bulk concentrations in the two phases remain
On the other hand, however, the volume of the chosen uniform up to the Gibbs dividing surface.

123
ChemTexts (2015) 1:16 Page 5 of 17 16

nri ¼ ni  nai  nbi ¼ ni  na xai  nb xbi Therefore, according to Euler’s theorem (see ‘‘Homo-
b bb geneous functions’’ section and Example #4), and in the
¼ ni  na xaa
i  n xi ; ð6Þ
framework of the Gibbs model
where ni is the total amount of component i in the ‘‘real’’ X
U r ¼ T r Sr þ cA þ lri nri ; ð15Þ
system, xai and xbi are the mole fractions in phases a and b, i
respectively, and na and nb are the total amounts of the
components (‘‘total number of moles’’) in the reference where c is the intensive (interfacial) parameter conjugate to
systems. It is clear from Eq. (6) that the surface excess the extensive variable A.
amount is well defined only when na and nb are fixed. It Due to the thermodynamic equilibrium
can be also seen that with different na and nb values we T r ¼ T a ¼ T b ¼ T aa ¼ T bb ¼ T; ð16Þ
have different values for nri .
According to the above considerations, the surface and
excess Xr of any extensive property X is calculated as lri ¼ lai ¼ lbi ¼ laa bb
i ¼ li ¼ li : ð17Þ
r a b
X ¼XX X ; ð7Þ According to the equations like the two above, it is not
where X denotes the value of the extensive property for the necessary to use superscripts to distinguish T, l1 … lm, in
whole system and Xa and Xb are the values for the refer- the different equilibrium phases because these must have
ence systems. uniform values throughout a, b, aa, bb, and r (due to the
The relation that gives the internal energy U as a equilibrium assumptions).
function of the extensive parameters is a fundamental In the two reference phases, the following relationships
relation. If the fundamental relation of a particular system are valid:
X
is known, all conceivable thermodynamic information U a ¼ TSa  pV a þ li nai ; ð18Þ
about this system can be ascertained [29]. The internal i
energies of the reference phases are given by
  and
U a ¼ U a Sa ; V a ; na1 . . .nam ð8Þ X
U b ¼ TSb  pV b þ li nbi : ð19Þ
and i
 
U b ¼ U b Sb ; V b ; nb1 . . .nbm : ð9Þ According to Eq. (15), the intensive parameter (c) is
defined by
The internal energy (U) of the system depends on the  r
oU
entropy (S), volume (V), amounts n1…nm of the compo- c¼ : ð20Þ
oA Sr ;nr ...nrm
nents 1,…m, and the surface area (A): 1

  Although this expression is mathematically correct, it is

U ¼ U S; V; A; n1 . . .nm : ð10Þ
not really useful for practical purposes. Equation (15)
The excess of the internal energy is given by expresses the dependence of the energy U on the basis of
U ¼ U  Ua  Ub;
r
ð11Þ independent variables Sr, A, nr1 … nrm. This set of inde-
pendent variables is not by any means the most convenient.
and the excess of the entropy is It is usually preferable to use T as an independent variable
instead of S. If the experiment is such that the external
Sr ¼ S  Sa  Sb : ð12Þ
conditions are constant temperature and constant pressure,
The excess internal energy function the most convenient potential function to use is the Gibbs
  free energy function, G(T,p,n1 … nm), obtained from
U r ¼ U r Sr ; V r ; A; nr1 . . .nrm ð13Þ
U(S,V, n1 … nm) by two subsequent Legendre transfor-
is a homogeneous function of degree one with respect to all mations (see ‘‘Adsorption’’ section and especially Example
variables (see ‘‘Homogeneous functions’’ section, espe- #12):
cially Examples #1 and #4), if Vr : 0 (Gibbs model), or Ga ¼ U a þ pV a  TSa ð21Þ
Vr = As (Guggenheim model), since it is evident that
    and
U r kSr ; kV r ; kA; knri . . .knrm ¼ kU r Sr ; V r ; A; nri . . .nrm
ð14Þ Gb ¼ U b þ pV b  TSb : ð22Þ

for all k [ 0 real numbers. Consequently

123
16 Page 6 of 17 ChemTexts (2015) 1:16

X X
Ga ¼ li nai ð23Þ dGr ¼ Sr dT þ cdA þ li dnri : ð31Þ
i i

and The differential of Eq. (25) is

X X X
b
G ¼ li nbi : ð24Þ dGr ¼ cdA þ Ad c þ li dnri þ nri dli : ð32Þ
i i i

The excess Gibbs free energy function is given as

X There are thus two (general) expressions for dGr
r
G ¼ cA þ li nri ; ð25Þ (Eqs. 31 and 32), both of which are correct. This can only
i be the case if
X
and c is defined by Sr dT þ Adc þ nri dli ¼ 0: ð33Þ
 r i
oG
c¼ : ð26Þ
oA T;nr ...nrm Equation (33) is the so-called Gibbs–Duhem equation
1
for interfaces (see ‘‘The Gibbs–Duhem equation’’ section).
Unfortunately, this definition of c is still not appropriate At constant temperature
for experimental studies or to confirm experimental results X
Adc ¼ nri dli : ð34Þ
since Gr(T, A, nr1 … nrm) remains ill-defined and arbitrary
i
(because nr1 … nrm clearly depend on the selection of the
reference systems). The Gibbs free energy function for the Dividing both sides of Eq. (33) by A yields
whole system can be expressed as X nr X
i
X X b X dc ¼ dli ¼ Ci dli ; ð35Þ
G ¼ cA þ li nai þ li ni þ li nri i
A i

Xi  i  i where Ci is the surface excess concentration of species

¼ cA þ li nai þ nbi þ nri : ð27Þ
i. Equation (35) is commonly called the Gibbs adsorption
i
equation.
This means that c can also be defined in terms of the In the case of liquid/liquid interfaces, the interfacial
Gibbs free energy function of the whole system as intensive parameter (c) can be identified with the interfacial
 
oG tension. (Note that in case of solid/liquid interfaces there is
c¼ ; ð28Þ some controversy in the literature concerning the correct
oA T;p;n ...nm
1
name of c, e.g., it is sometimes called ‘‘specific surface
or in terms of the Helmholtz (free) energy function as energy’’ or ‘‘surface stress’’ [1, 16, 26, 30, 31]).
  There are two important points that should be addressed
oF
c¼ ; ð29Þ here:
oA T;V;n1 ...nm
1. In the case of ionic components (charged species),
when the Helmholtz energy or ‘‘free energy’’ function is ‘‘electrochemical potentials’’ (l~i ) may be used instead of
defined as the Legendre transform of the internal energy ‘‘chemical potentials’’ in the corresponding equations.
function (F = U - TS). On the other hand—still 2. It follows from Eq. (6) (which is the definition
remaining in the framework of the Gibbs model—it should equation of the surface excess amounts) that the Ci
be noted that since no volume term appears in Eq. (15), values are uncertain, since they depend on the arbitrary
there is no distinction between the surface Helmholtz and selection of na and nb.
Gibbs free energies.
According to the above discussions, Gr is a partly Nevertheless, for the analysis of the experimental data
homogeneous function of degree one (see ‘‘The Gibbs– we need measurable physical quantities that do not depend
Duhem equation’’ section) in the variables A and nr1 … nrm. on the size of the reference phases.
The expression for the total differential of Gr is The following procedure can be used for this purpose.
 r  r At constant T and p, the Gibbs–Duhem relationships for the
oG oG
dGr ¼ dT þ dA two reference bulk phases are
oT A;nr ...nrm oA T;nr ...nrm X
X oGr  xai dli ¼ 0
1 1
ð36Þ
þ dnri : ð30Þ i
i
oA T;A;nr
j6¼i
 r   and
r oGr X
Taking into account that oGoT A;nr1 ...nrm ¼ S , oA T;nr1 ...nrm
 r xbi dli ¼ 0: ð37Þ
¼ c, and oG
oA T;A;nr ¼ li , Equation (30) can be written as i
j6¼i

123
ChemTexts (2015) 1:16 Page 7 of 17 16

Using the above two relationships, it is possible to Therefore, the C’i values can be determined as
   
express dl1 and dl2 (i.e., the differential changes of the 0 oc 1 oc
Ci ¼  ¼ ð47Þ
chemical potentials of two selected components) as a oli T;p;lj6¼i RT o ai T;p;aj6¼i
function of the other dli values and the mole fractions at  
constant temperature and pressure: (or more exactly C0i ¼  dl o c 1
¼  RT
  iði6¼1;2Þ
lj6¼i
oc
xa X xa o aiði6¼1;2Þ ), where ai denotes the relative activity of
dl1 ¼  2a dl2  i
a dli : ð38Þ T;p;aj6¼i
x1 x
i6¼1;2 1 component i.
Equation (45) (Gibbs adsorption isotherm also called
and the Gibbs adsorption equation) is one of the most impor-
xb1 X xb tant results from interfacial thermodynamics, and it is used
i
dl2 ¼  b
dl1  b
dli : ð39Þ all the time in physical chemistry and surface science.
x2 i6¼1;2 x2
In the following sections, we will briefly review the
Combining Eq. (35) with Eqs. (38) and (39), we obtain relevant mathematical background necessary for some of
! the derivations presented above.
1X x a b
x  x b a
x x b a
x  x a b
x
dc ¼ nr þ 2 ib 2 i r
n1 þ 1 ib 1 i r
n2 dli
A i6¼1;2 i xa1 x2  xa2 xb1 xa1 x2  xa2 xb1
ð40Þ The mathematical background of interfacial
thermodynamics
or
! Homogeneous functions
X xa2 xbi  xb2 xai xb1 xai  xa1 xbi
d c ¼ Ci þ C1 þ C2 dli :
i6¼1;2 xa1 xb2  xa2 xb1 xa1 xb2  xa2 xb1 Definition
ð41Þ
Let f (x1, x2, …, xm) be a real function of variables x1, x2,
By taking into account that …, xm. The function f is a homogeneous function if for all
1  values of the factor k [ 0
Ci ¼ ni  na xai  nb xbi ; ð42Þ
A f ðkx1 ; kx2 ; . . .; kxm Þ ¼ sðkÞf ðx1 ; x2 ; . . .; xm Þ; ðM:1Þ
we have
where the function s(k) is usually called the scaling func-
!
1X xa2 xbi  xb2 xai xb1 xai  xa1 xbi tion, and is given by s(k) = kn.
dc ¼ ni þ n1 þ n2 dli In other words, a homogeneous function is a function of
A i6¼1;2 xa1 xb2  xa2 xb1 xa1 xb2  xa2 xb1
one or several variables that satisfies the following condi-
ð43Þ tion: when all independent variables of a function are
simultaneously multiplied by the same (arbitrary) factor,
or
! the value of the function is multiplied by some power of
X xa2 xbi  xb2 xai xb1 xai  xa1 xbi this factor. That is, if
dc ¼ Ci þ C1 þ C2 dli :
i6¼1;2 xa1 xb2  xa2 xb1 xa1 xb2  xa2 xb1 f ðkx1 ; kx2 ; . . .; kxm Þ ¼ kn f ðx1 ; x2 ; . . .; xm Þ ðM:2Þ
ð44Þ for all k [ 0, then f is said to be a homogeneous function of
degree n. The degree n can take on any value (positive,
Equation (44) can be written in the simpler form:
X negative, or zero). A function f is linearly homogenous if it
dc ¼ C0i dli ; ð45Þ is homogeneous of degree 1.
i6¼1;2 If for a function f the equation
0
where Ci denotes the (relative) surface excess of compo- f ðkx1 ; . . .; kxm ; y1 ; . . .; yw Þ ¼ kn f ðx1 ; . . .; xm ; y1 ; . . .; yw Þ
nent i with respect to the two selected components, and ðM:3Þ
xa2 xbi  xb2 xai xb1 xai  xa1 xbi is true, then we say that this function is homogeneous of
C0i ¼ Ci þ C1 þ C2 : ð46Þ
xa1 xb2  b
xa2 x1 xa1 xb2  xa2 xb1 degree n in the variables x1, x2, …, xm. Such functions are
0 called partly (or partially) homogeneous functions [1]. (It
It is clear that the Ci values do not depend on the should be noted that it is possible for functions to be
selection of the reference systems (that is, on the selection homogeneous of different degree in different variables, but
of na and nb).

123
16 Page 8 of 17 ChemTexts (2015) 1:16

here we restrict our attention to functions for which This means that f is represented by
Eq. (M.2) or Eq. (M.3) holds).  
n x2 xm
f ¼ x1 g ; . . .; ðM:10Þ
x1 x1
Some remarks to the definition of homogeneous functions
with some function g. Since, conversely, every function
Remark #1 A homogeneous function of degree n gives f formed by means of an appropriate function g of m - 1
rise to a set of derivative functions that are homogeneous in variables satisfies the condition of homogeneity, the
the same set of variables and of degree n - 1, that is, expression (M.10) represents the totality of homogeneous
partial derivatives of a homogeneous function of degree functions of degree n.
n are homogeneous functions of degree n - 1.
Euler’s theorem
Proof Differentiating both sides of Eq. (M.2) with respect
to xi (for i = 1,…, m), we get
Euler’s Theorem states that the differentiable function f of
m variables is homogeneous of degree n, then the following
of ðkx1 ; . . .; kxm Þ of ðx1 ; . . .; xm Þ identity holds:
k ¼ kn ; ðM:4Þ
okxi oxi Xm
of
nf ðx1 ; x2 ; . . .; xm Þ ¼ xi : ðM:11Þ
and then dividing both sides of Eq. (M.4) by k we obtain ox i
i¼1
of ðkx1 ; . . .; kxm Þ of ðx1 ; . . .; xm Þ
¼ kn1 : ðM:5Þ Proof Let f be a homogeneous function of degree n such
okxi oxi that
Hence the derivatives of f are homogeneous of degree
n - 1. f ðj^
x1 ; j^ xm Þ ¼ jn f ðx^1 ; x^2 ; . . .; x^m Þ;
x2 ; . . .; j^ ðM:12Þ
For example, in case of a homogeneous function of the
1
first degree f ðkx1 ; kx2 ; . . .; kxm Þ ¼ kf ðx1 ; x2 ; . . .; xm Þ the par- and k 6¼ 0; k ¼ j; xi ¼ 1k x^i ¼ j^
xi ; kxi ¼ x^i .
Evidently
tial derivative with respect to xi is of ðkxokx 1 ;...;kxm Þ okxi
i oxi ¼
f ðj^
x1 ; j^ xm Þ ¼ f ðx1 ; x2 ; . . .; xm Þ
x2 ; . . .; j^
k of ðx1ox;...;x mÞ
, and therefore of ðkx1 ;...;kxm Þ
okxi ¼ of ðx1 ;...;xm Þ
oxi .  n
i 1
¼ f ðkx1 ; kx2 ; . . .; kxm Þ; ðM:13Þ
Remark #2 Suppose that the domain of definition of the k
function f lies in the first quadrant, x1 [ 0,…, xm [ 0, and
and
contains the whole ray (kx1,…, kxm), k [ 0, whenever it
contains (x1,…, xm), i.e., it is assumed that for every kn f ðx1 ; x2 ; . . .; xm Þ ¼ f ðkx1 ; kx2 ; . . .; kxm Þ
point (x1,…, xm) in the domain of f, the point (kx1,…, - ¼ f ðx^1 ; x^2 ; . . .; x^m Þ; ðM:14Þ
kxm) also belongs to this domain for any k [ 0. Then f is
i.e., f(x1, x2, …, xm) is also a homogeneous function of
homogeneous of degree n if and only if there exists a
degree n.
function g of m-1 variables defined on the set of points of
Differentiating each side of formula (M.12) with respect
the form (x2/x1,…, xm/x1) such that for all (x1,…, xm) in the
to j, we have the following relationship:
domain of definition [see for example refs. 32, 33]:
of ðj^
x1 ; j^ xm Þ
x2 ; . . .; j^ of ðj^ x1 ; j^ xm Þ
x2 ; . . .; j^
x^1 þ x^2 þ   
f ðx1 ; . . .; xm Þ ¼ xn1 gðx2 =x1 ; . . .; xm =x1 Þ: ðM:6Þ oj^x1 oj^x2
of ðj^x1 ; j^ x2 ; . . .; j^xm Þ
Proof Let f (x1, x2, …, xm) be a homogeneous function of þ x^m ¼ njn1 f ðx^1 ; x^2 ; . . .; x^m Þ:
oj^ xm
degree n of variables x1, x2, …, xm such that for all k [ 0 ðM:15Þ
Introducing j ¼ 1k ; xi ¼ j^
xi ; x^i ¼ kxi ; Eq. (M.15)
f ðkx1 ; . . .; kxm Þ ¼ kn f ðx1 ; . . .; xm Þ: ðM:7Þ
combined with Eq. (M.14) can be rewritten as
Obviously of ðx1 ; x2 ; . . .; xm Þ of ðx1 ; x2 ; . . .; xm Þ
  kx1 þ kx2 þ   
x2 xm ox1 ox2
f ðx1 ; x2 ; . . .; xm Þ ¼ f x1 ; x1 ; . . .; x1 : ðM:8Þ  n1
x1 x1 of ðx1 ; x2 ; . . .; xm Þ 1
þ kxm ¼ n kn f ðkx1 ; kx2 ; . . .; kxm Þ:
If we set k = 1/x1, we have oxm k
  ðM:16Þ
n x2 xm
f ðx1 ; x2 ; . . .; xm Þ ¼ x1 f 1; ; . . .; : ðM:9Þ
x1 x1

123
ChemTexts (2015) 1:16 Page 9 of 17 16

This means that tn f ðtx1 ; . . .; txm Þ  f ðx1 ; . . .; xm Þ ¼ 0; ðM:26Þ

X of ðx1 ; x2 ; . . .; xm Þ
m
xi ¼ nf ðx1 ; x2 ; . . .; xm Þ; ðM:17Þ and so
oxi
i¼1 f ðtx1 ; . . .; txm Þ ¼ tn f ðx1 ; . . .; xm Þ ðM:27Þ
or obviously for all t [ 0, which means that f is homogeneous of degree
X
m
of ðx^1 ; x^2 ; . . .; x^m Þ n.
x^i ¼ nf ðx^1 ; x^2 ; . . .; x^m Þ; ðM:18Þ (Alternative proof: Let g(t) = f(tx1, …, txm). After dif-
i¼1
o^
xi
ferentiation with respect to t and by taking into account
which is exactly Euler’s theorem. Eq. (M.11), we see that
(Alternatively, by differentiating Eq. (M.2) with respect ogðtÞ of ðtx1 ; tx2 ; . . .; txm Þ of ðtx1 ; tx2 ; . . .; txm Þ
to k we get ¼ x1 þ x2
ot otx1 otx2
o o of ðtx1 ; tx2 ; . . .; txm Þ
f ðkx1 ; kx2 ; . . .; kxm Þ ¼ kn f ðx1 ; x2 ; . . .; xm Þ ðM:19Þ þ    þ xm
ok ok otxm

and thus 1 of ðtx1 ; tx2 ; . . .; txm Þ of ðtx1 ; tx2 ; . . .; txm Þ
¼ tx1 þ tx2 :
X
m t otx1 otx2
of ðkx1 ; kx2 ; . . .; kxm Þ
xi ¼ nkn1 f ðx1 ; x2 ; . . .; xm Þ: of ðtx1 ; tx2 ; . . .; txm Þ
i¼1
okxi þ    þ txm
otxm
ðM:20Þ n n
¼ f ðtx1 ; tx2 ; . . .; txm Þ ¼ gðtÞ
Then, setting k = 1 t t
ðM:28Þ
Xm
of ðx1 ; x2 ; . . .; xm Þ
xi ¼ nf ðx1 ; x2 ; . . .; xm Þ ðM:21Þ This means that
i¼1
oxi
dgðtÞ n
which was to be proved). ¼ gðtÞ: ðM:29Þ
dt t
We can show that the converse theorem also holds, that
is if the function f of the real variables x1, x2, …, xm After integration, we obtain for any t [ 0
satisfies the identity (M.11), then the function f is homo- lnjgðtÞj ¼ n ln t þ c; ðM:30Þ
geneous of degree n.
where c is an integration constant, which is independent on
Let us fix (x1, x2, …, xm) and define the function g of a
t. This means that
single variable t as
gðtÞ ¼ tn f ðtx1 ; . . .; txm Þ  f ðx1 ; . . .; xm Þ: ðM:22Þ gðtÞ ¼ ec tn : ðM:31Þ

Differentiating each side of this equation with respect to Choosing t = 1, we see that g(1) = ec and consequently
t, we get g(t) = g(1)tn, i.e., Eq. (M.27) holds).

dgðtÞ Some remarks to Euler’s theorem

¼ ntn1 f ðtx1 ; . . .; txm Þ
dt
Xm
df ðtx1 ; . . .; txm Þ
þ tn xi : ðM:23Þ Extensive variables in thermodynamics are those that
i¼1
dtxi depend linearly on the size of the system. This means that
By Euler’s theorem, we have if a system is composed of several subsystems, the value of
the extensive variable (‘‘extensive quantity’’) for the
X
m
of ðtx1 ; . . .; txm Þ
txi ¼ nf ðtx1 ; . . .; txm Þ; ðM:24Þ composite system is calculated by summing over the sub-
i¼1
otxi systems. As a consequence, extensive thermodynamic
functions are homogeneous functions of degree n = 1
so that
(homogeneous linear functions) with respect to their
ogðtÞ 1 extensive arguments, i.e.,
¼ ntn1 f ðtx1 ; . . .; txm Þ þ tn nf ðtx1 ; . . .; txm Þ
ot t f ðkx1 ; . . .; kxm Þ ¼ kf ðx1 ; . . .; xm Þ; ðM:32Þ
¼ 0:
ðM:25Þ and so
Xm
of
Thus g(t) is constant for all t. It is clear that g(1) = 0, xi ¼ f ðx1 ; . . .; xm Þ: ðM:33Þ
and therefore g(t) = 0 for all t, and with Eq. (M.22) we get i¼1
oxi

123
16 Page 10 of 17 ChemTexts (2015) 1:16

According to Eq. (M.5), partial derivatives of a homo- The Gibbs–Duhem Equation

geneous linear function are homogeneous functions of
degree n = 0 (homogeneous function of 0th degree), i.e., If the function f (x1, x2, …, xm) is homogeneous of degree
f ðkx1 ; . . .; kxm Þ ¼ f ðx1 ; . . .; xm Þ ðM:34Þ n = 1 with respect to the variables x1, x2, …, xm, then one
has the identity (M.2):
and
f ðkx1 ; kx2 ; . . .; kxm Þ ¼ kf ðx1 ; x2 ; . . .; xm Þ:
Xm
of of
xi ¼ 0: ðM:35Þ Let us set ox ¼ p1 , …, oxofm ¼ pm , etc., and apply Euler’s
oxi 1
i¼1 theorem to the function f.
Example #1 Let us consider the following function: We will obtain
f ¼ x 1 p1 þ x 2 p2 þ    þ x m pm : ðM:36Þ
3
x It results from this formula that the functions
f ðx; y; zÞ ¼  : ðE:1:1Þ
yz p1, p2, …, pm are quantities (functions) of the same type as
Since the quotient of energy or work by charge, mass, etc., and
 3 hence these are quantities (functions) of the same type as a
ðkxÞ3 x potential, e.g., in thermodynamics, if f is the internal
f ðkx; ky; kzÞ ¼  ¼k  ¼ kf ðx; y; zÞ;
ðkyÞðkzÞ yz energy function (U, see ‘‘Models of the interfacial region’’
ðE:1:2Þ and ‘‘Adsorption’’ sections), and the xi-s are the amounts of
substances, we can call them partial molar internal ener-
the function f (x,z,y) is homogeneous of the first degree in gies of the constituents 1, …, m in the system (thermody-
the variables x, y, and z. The partial derivatives are namic or chemical potentials).
of 3x2 of x3 of x3
ox ¼  yz , oy ¼ y2 z, and oz ¼ yz2 . According to the considerations outlined above, the func-
Thus, applying Euler’s theorem tions p1, p2, …, pm are homogeneous functions of degree zero
of of of in the variables x1, x2, …, xm. To each of these functions, we
f ðx; y; zÞ ¼ xþ yþ z can apply Euler’s theorem, and we will find the identities
ox
 oy  oz   3
3x2 x3 x x3 op1 op1 op1
¼  xþ 2 yþ 2
z¼ : x1 þ x2 þ    þ xm ¼0
yz y z yz yz ox1 ox2 oxm
ðE:1:3Þ .. ðM:37Þ
.
Example #2 We know that partial derivatives of a opm opm opm
homogeneous function of degree n are homogeneous x1 þ x2 þ    þ xm ¼ 0:
ox1 ox2 oxm
functions of degree n - 1.
The partial derivative of the function f (which is a The identities
homogeneous function of degree 1, as defined by opi opj
¼ ; ðM:38Þ
Eq. (E.1.1)) with respect to x is oxj oxi

of 3x2 which result from the definition of the functions (the mixed
fx ¼ ¼ : ðE:2:1Þ
ox yz second partial derivatives are equal), permit the substitu-
tion of equations
The function fx is a homogeneous function of degree 0 in
the variables x, y, and z, since op1 op2 opm
x1 þ x2 þ    þ xm ¼0
ox1 ox1 ox1
3ðkxÞ2 3x2 ..
fx ðkx; ky; kzÞ ¼  ¼ ¼ fx ðx; y; zÞ: ðE:2:2Þ . ðM:39Þ
ðkyÞðkzÞ yz
2 op1 op2 opm
The partial derivatives are ofoxx ¼  6x
yz ,
ofx
oy ¼ 3x
y2 z , and
x1 þ x2 þ    þ xm ¼ 0;
ofx 3x2
oxm oxm oxm
oz ¼ yz2 , that is
and therefore
   2  2
6x 3x 3x
 xþ 2 yþ z¼0 ðE:2:3Þ x1 dp1 þ x2 dp2 þ    þ xm dpm ¼ 0: ðM:40Þ
yz y z yz2
This relation is known as Gibbs–Duhem equation or
in accordance with Euler’s theorem. Gibbs–Duhem relation.

123
ChemTexts (2015) 1:16 Page 11 of 17 16

Historical outlook Duhem was a great admirer of Gibbs, X

m

and extended a number of Gibbs’ ideas. According to x1 dp1 þ x2 dp2 þ    þ xm dpm ¼ xi dpi ¼ 0; ðM:48Þ
i
Miller [34] most probably Duhem was the first to use
Euler’s theorem explicitly to prove the Gibbs–Duhem which is the Gibbs–Duhem equation).
equation. The first references appear in his book ‘‘Le
Example #3 Consider the function
Potentiel Thermodynamique’’ which was published in 1886
[35]. It is altogether fitting that his name be appended to the
Gibbs–Duhem equation. x3
f ðx; y; zÞ ¼  ; ðE:3:1Þ
yz
(An alternative derivation of the Gibbs–Duhem equa-
tion: consider an arbitrary function of which is homogeneous of degree 1 (see Example #1).
x1 ; x2 ; . . .; xm : f ðx1 ; x2 ; . . .; xm Þ: ðM:41Þ The partial derivatives with respect to x, y, and z are
of 3x2 of x3 of x3
The total derivative (full derivative) of f(x1, x2, …, xm) fx0 ¼ ¼ ; fy0 ¼ ¼ 2 ; andfz0 ¼ ¼ 2 :
ox yz oy y z oz yz
with respect to x1 is
0 0 0

df of of dx2 of dxm The partial derivatives of fx, fy, and fz with respect to x
¼ þ þ  þ can be given as
dx1 ox1 ox2 dx1 oxm dx1
dx2 dxm
¼ p1 þ p2 þ    þ pm ; ðM:42Þ ofx0 6x ofy0 3x2 of 0 3x2
dx1 dx1 ¼ ; ¼ 2 and z ¼ 2
of of ox yz ox y z ox yz
where p1 ¼ ox 1
, p2 ¼ ox 2
, pm ¼ oxofm , etc.
Multiplying both sides of the equation by the differential In accordance with Eq. (M.39)
dx1: 6x 3x2 3x2 6x2 þ 3x2 þ 3x2
x þy 2 þz 2 ¼ ¼ 0: ðE:3:2Þ
of of of yz y z yz yz
df ¼ dx1 þ dx2 þ    þ dxm
ox1 ox2 oxm
Alternatively, we can formally write
¼ p1 dx1 þ p2 dx2 þ    þ pm dxm : ðM:43Þ
The result will be the differential change df in the 6xyz  dx þ 3x2 z  dy þ 3x2 y  dz
dfx0 ¼ ðE:3:3Þ
function f. The differential of the form y2 z2
Xm 3x2 y2 z  dx  2x3 yz  dy  x3 y2  dz
df ¼ pi ðx1 ; x2 ; . . .; xm Þdxi ðM:44Þ dfy0 ¼ ðE:3:4Þ
y4 z2
i¼1

is called the total differential or the exact differential of the and

function f.
3x2 yz2  dx  x3 z2  dy  2x3 yz  dz
According to Eq. (M.33) dfz0 ¼ : ðE:3:5Þ
y2 z4
Xm
pi xi ¼ f ðx1 ; . . .; xm Þ: ðM:45Þ Thus,
i¼1
6x2 y2 z2  dx þ 3x3 yz2  dy þ 3x3 y2 z  dz
xdfx0 þ ydfy0 þ zdfz0 ¼
In order to get an expression for df /dx1 from (M.45) z3 y 3
comparable with that in (M.42), we must differentiate 3x y z  dx  2x3 yz2  dy  x3 y2 z  dz
2 2 2

(M.45) ‘‘generally,’’ that is þ

z3 y 3
df op1 ox2 op2 oxm 3x y z  dx  x yz  dy  2x3 y2 z  dz
2 2 2 3 2
¼ p1 þ x 1 þ p2 þ x2 þ    þ pm þ ¼0
dx1 ox1 ox1 ox1 ox1 z3 y 3
opm ðE:3:6Þ
þ xm :
ox1
in accordance with the Gibbs–Duhem equation (M.40).
ðM:46Þ
Example #4 Consider the internal energy function U de-
There are thus two (general) expressions for df /dx1,
fined by
both of which are correct. This can only be the case if
op1 op2 opm
x1 þ x2 þ    þ xm ¼0 ðM:47Þ U ¼ U ðS; V; n1 ; . . .; nm Þ; ðE:4:1Þ
ox1 ox1 ox1
or

123
16 Page 12 of 17 ChemTexts (2015) 1:16

 
where S is the entropy, V is the volume, and ni (i = 1,…,m) kz1 kzw
is the chemical amount of component i. Since U is a f^ðkx1 ; . . .; kxm ; kz1 ; . . .; kzw Þ ¼ f kx1 ; . . .kxm ; ; . . .;
kx1 kx1
homogeneous function of degree 1 with respect to all of its   :
z1 zw
variables ¼ f kx1 ; . . .kxm ; ; . . .;
x1 x1
U ðkS; kV; kn1 ; . . .; knm Þ ¼ kU ðS; V; n1 ; . . .; nm Þ; ðE:4:2Þ ðM:52Þ
The temperature (T), pressure (p), and the chemical On the other hand, f is a homogeneous linear function
potentials li (i = 1,…,m) of the components are with respect to (x1, x2, …, xm). Thus
  oU   
T ¼ oU , p ¼  , l ¼ oU
re-    
oS V;n1 ;...;nm oV S;n1 ;...;nm i oni S;V;ni6¼m
z1 zw z1 zw
f kx1 ; . . .; kxm ; ; . . .; ¼ kf x1 ; . . .; xm ; ; . . .; :
spectively. (In the above equations, ni denotes the set x1 x1 x1 x1
[n1,…,nm] (i = 1,…,m) and nj=i denotes all the elements ðM:53Þ
(variables) in [n1,…,nm] except for the ith.) According to
By taking into account the definition of f^;
Euler’s theorem
X f^ðkx1 ; . . .; kxm ; kz1 ; . . .; kzw Þ ¼ kf^ðx1 ; . . .; xm ; z1 ; . . .; zw Þ
U ¼ TS þ ðpVÞ þ li ni : ðE:4:3Þ
i ðM:54Þ
The Gibbs–Duhem relation is written as which was to be proved.
SdT þ Vdp þ n1 dl1 þ    þ nm dlm ¼ 0: ðE:4:4Þ The partial derivatives of f and f^ with respect to xi, yi,
and zi are, respectively, given by
It follows that T, p, and li cannot be independently     !
variable, i.e., the intensive variables are not independent. If of y of o ^
f
x
fi ¼ ; fi ¼ ; f^i ¼
x
;
we know m - 1 of them, the value of the mth can be oxi xj6¼i ;yj oyi xj ;yj6¼i oxi
! xj6¼i ;zj
determined from the Gibbs–Duhem equation. It is partic- ^
o f
ularly useful in its application to changes at constant f^iz ¼
temperature and pressure, and it may be written as ozi
xj ;zj6¼i
Xm ðM:55Þ
ni dli ¼ 0: ðE:4:5Þ
i¼1 For i = 1,…,m, the partial derivatives of f^ are given as
Partly homogeneous functions of degree 1 of^ of
f^ix ¼ ¼ ¼ fix ðM:56Þ
oxi oxi
A function f is called ‘‘partly homogeneous’’ of degree 1 in
and for j = 1,…,w
terms of m among m ? w variables if
f ðkx1 ; . . .; kxm ; y1 ; . . .; yw Þ ¼ kf ðx1 ; . . .; xm ; y1 ; . . .; yw Þ of^ of 1 1
f^jz ¼ ¼  ¼ fjy  : ðM:57Þ
ðk [ 0Þ; ozj oyj x1 x1
ðM:49Þ Thus, the total derivative of f^ is
   
i.e., the function f is homogeneous with respect to certain of^ of of z1 of z2 of
variables (x1, x2, …, xm), but not homogeneous with respect ¼ þ  2
þ  2
þ  þ
ox1 ox1 oy1 x1 oy2 x1 oyw
to all of the variables. These functions are important as  
zw
they are frequently encountered in thermodynamics [36].  : ðM:58Þ
x21
Let us introduce new variables
zi ¼ x1  yi ðM:50Þ Using Eqs. (M.50) and (M.55, M.56), Eq. (M.58) can be
rewritten in the form
and the function f^ as      
^x x y y1 y y2 y yw
  f1 ¼ f1 þ f1  þ f2  þ    þ fw 
z1 zw x1 x1 x1
f^ðx1 ; . . .; xm ; z1 ; . . .; zw Þ ¼ f x1 ; . . .; xm ; ; . . .; :
x1 x1 ðM:59Þ
ðM:51Þ
and with Eq. (M.57)
We can prove that f^ is homogeneous of the first order X
w
with respect to all of its variables. According to Eq. (M.51) f^1x ¼ f1x  f^iz yi : ðM:60Þ
i¼1

123
ChemTexts (2015) 1:16 Page 13 of 17 16

Since f^ is homogeneous of the first degree with respect to and

all of its variables, and its partial derivatives are homoge- of x2 w
neous of degree zero, we can apply Euler’s theorem. Hence fy0 ¼ ¼ 2 : ðE:5:5Þ
oy y
Xm X
w
f^ ¼ xi f^ix þ zj f^jz ðM:61Þ Thus, applying Euler’s theorem
i¼1 j¼1    2 
2xw 5 x w x2 w
f ðx; y; u; wÞ ¼  þu xþ y ¼  þ u5 x:
and y y2 y
X
m X
w ðE:5:6Þ
0¼ xi df^ix þ zj df^jz : ðM:62Þ
i¼1 j¼1 The partial derivatives with respect to u and w are
of
With Eqs. (M.51) and (M.56–M.62), we have fu0 ¼ ¼ 5u4 x ðE:5:7Þ
! ou
y
Xw
f j
Xm Xw
fjy
f ¼ x1 f1x  yj þ xi fix þ x1 yj ; ðM:63Þ and
x
j¼1 1 i¼2 j¼1
x1
of x2
fw0 ¼ ¼ : ðE:5:8Þ
and ow y
!
X
w
fjy X
m X
w
fjy We can formally write
0 ¼ x1 d f1x  yj þ xi dfix þ x1 yj d :
x1 x1 2wyx  dx  2x2 y  dw þ 2x2 w  dy
j¼1 i¼2 j¼1
x  dfx0 ¼ þ 5u4 x  du;
ðM:64Þ y2
ðE:5:9Þ
Consequently
Xm Xm   and
of
f ¼ xi fix ¼ xi ðM:65Þ
ox 2xwy2  dx þ x2 y2  dw  2x2 wy  dy
i¼1 i¼1 i xj6¼i ;yj
y  dfy0 ¼ : ðE:5:10Þ
y3
and
Similarly
X
m X
w
0¼ xi dfix  fiy dyi : ðM:66Þ fu0  du ¼ 5u4 x  du; ðE:5:11Þ
i¼1 j¼1
x2
The latter relation is the Gibbs–Duhem equation for fw0  dw ¼   dw: ðE:5:12Þ
y
‘‘partly homogeneous functions.’’
Thus,
Example #5 The function
xdfx0 þ ydfy0  fu0 du  fw0 dw
2wyx  dx  2x2 y  dw þ 2x2 w  dy
x2 ¼ þ 5u4 x  du
f ¼  w þ u5 x ðE:5:1Þ y2
y
2xwy2  dx þ x2 y2  dw  2x2 wy  dy
is not fully homogeneous, since þ
y3
 2 
f ðkx; ky; ku; kwÞ 6¼ kn f ðx; y; u; wÞ: ðE:5:2Þ 4 x
 5u x  du    dw ¼ 0
y
(If the powers of x, y, u, and w are added, the first term
on the right-hand side of the expression yields 2, and the ðE:5:13Þ
second term 6.) in accordance with the Gibbs–Duhem equation for ‘‘partly
However, if u and w are constant, the sum of the powers homogeneous functions’’ (Eq. M.66).
of x and y for each term is 1. Therefore, the function is
partly homogeneous (with respect to x and y), so that Example #6 The function
f ðkx; ky; u; wÞ ¼ kf ðx; y; u; wÞ: ðE:5:3Þ
x2
The partial derivatives with respect to x and y are f ¼ w þ u3 x ðE:6:1Þ
y
of 2xw
fx0 ¼ ¼ þ u5 ðE:5:4Þ is not fully homogeneous, since f(kx, ky, ku, kw) =
ox y
knf(x, y, u, w).

123
16 Page 14 of 17 ChemTexts (2015) 1:16

X
Let us introduce new variables a = xu and b = xw, SdT  Vdp þ ni dli ¼ 0: ðE:7:7Þ
i.e., u = a / x and w = b / x, and the function f^ as i

x2 b a3 xb a3 Legendre transformation

f^ ¼  þ x¼ þ 2: ðE:6:2Þ
y x x y x
(The transform is named after the French mathematician
Since
Adrien-Marie Legendre (1752–1833).)
 
kx  kb ðkaÞ3 x  b a3 Let f (x1, x2, …, xm) be an arbitrary analytic function of
f^ðkx; ky; ka; kbÞ ¼  þ ¼ k  þ
ky ðkxÞ2 y x2 variables x1, x2, …, xm. The differential of f is
¼ kf^ðx; y; a; bÞ; of of of
df ¼ dx1 þ dx2 þ    þ dxm
ðE:6:3Þ ox1 ox2 oxm
¼ p1 dx1 þ p2 dx2 þ    þ pm dxm : ðM:67Þ
it is obvious that f^ is a homogeneous function of degree 1
with respect to all of its variables (x, y, a, b). Consider a new function g of the variables p1 and x2,
x3,…,xm:
Example #7 Let us consider the Gibbs free energy func-
gðp1 ; x2 ; . . .; xm Þ ¼ f ðx1 ðp1 Þ; x2 ; . . .; xm Þ  p1 x1 ðp1 Þ;
tion G defined by
ðM:68Þ
of
G ¼ GðT; p; n1 ; n2 ; . . .; nm Þ; ðE:7:1Þ where p1 ¼ ox 1
. A necessary condition is the existence of a
one-to-one relation between p1 and x1, that is, the function
where T is the temperature, p is the pressure, and ni
p1(x1, x2, …, xm) can be inverted to give x1(p1) (This
(i = 1,…,m) is the chemical amount of component i. At
means that p1 is bijective).
given (constant) T and p, the function G is partly homo-
The new function g(p1, x2, …, xm) is called the Legen-
geneous of degree one in terms of the variables n1,…,nm.
dre transform of the function f(x1, x2, …, xm). A Legendre
We thus have
transform converts from a function of one set of variables
GðT; p; kn1 ; . . .; knm Þ ¼ kGðT; p; n1 ; . . .; nm Þ: ðE:7:2Þ to another function of a ‘‘conjugate’’ set of variables. In
general, this is a special transformation that allows us to
The total differential of G is given as
    replace variables in a function in a consistent manner.
Xm  
oG oG oG The differential of g(p1, x2, …, xm) is
dG ¼ dT þ dp þ dni :
oT p;ni op T;ni i¼1
oni T;p;nj6¼i dg ¼ df  x1 ðp1 Þdp1  p1 ðx1 ; x2 ; . . .; xm Þdx1 : ðM:69Þ
ðE:7:3Þ
With Eq. (M.67)
The entropy (S), volume (V), and the chemical potentials of of
  dg ¼ x1 ðp1 Þdp1 þ dx2 þ    þ dxm : ðM:70Þ
li (i = 1,…,m) of the components are S ¼  oG oT p;ni , ox2 oxm
   
V ¼ oGop , and li ¼ on oG
i
, respectively. Thus, Formally, the total differential of g is
T;ni T;p;nj6¼i
Eq. (E.7.3) can be rewritten as og og og
dg ¼ dp1 þ dx2 þ    þ dxm : ðM:71Þ
Xm op1 ox2 oxm
dG ¼ SdT þ Vdp þ li dni : ðE:7:4Þ og of og
i¼1
Comparing Eqs. (M.70) and (M.71) x1 ¼  op , ¼ ox
1 ox2 2
,
of og
ox3¼ ox3 , etc. Let us take a look at some simple examples to
According to Euler’s theorem
X see how this works.
G¼ li ni : ðE:7:5Þ
i Example #8 One-dimensional Legendre transformation
Consider an arbitrary function of x : f (x). We know that
In order to obtain an expression for dG from (E.7.5) locally the slope of this curve is precisely its derivative
comparable with that in (E.7.3), we must differentiate with respect to x, so the change in the function f (x) at the
(E.7.5). This gives point x for a small change in the argument dx is
X m Xm
dG ¼ li dni þ ni dli ; ðE:7:6Þ of
df ¼ dx  pð xÞdx; ðE:8:1Þ
i¼1 i¼1 ox
and by comparing Eq. (E.7.6) with Eq. (E.7.4) we obtain where p ¼ ofoxðxÞ ¼ f 0 ð xÞ, as usual. Now suppose that we
the following important relation (i.e., the Gibbs–Duhem want to find a function that reverses the roles of the slope
equation): and infinitesimal, i.e., a function g(p) such that

123
ChemTexts (2015) 1:16 Page 15 of 17 16

dg = x dp (where we now view x as a function of p defined f ? g is actually a function of either p or x but not both,
by the inverse of p = f’(x)). We can see that the function because one variable implicitly depends on the other via a
Legendre transform.
g ¼ f ð pÞ  pxð pÞ ðE:8:2Þ
Usually, the definition given in Eq. (M.68) is preferred
(the Legendre transform) has the desired property:
in thermodynamics.
dg ¼ df  xdp  pdx ¼ xdp: ðE:8:3Þ
Example #9 Consider the function
Notice that since g(p) is a function of p only, we must
have
f ðx; zÞ ¼ x2 þ y2 : ðE:9:1Þ
dgð pÞ
¼ x: ðE:8:4Þ of
The partial derivatives are ox ¼ 2x ¼ p1 , of
¼ 2y ¼ p2 ,
dp p1 p2
oy
x ¼ 2 , and y ¼ 2 .
Consequently, we have a pair of functions f(x) and The Legendre transforms of f are
g(p) related in the following way:
df ð xÞ p22 p22 p2
f ð xÞ ! gð pÞ ¼ f ðxðpÞÞ  xð pÞp . . . xð pÞ : ¼ p; gðx; p2 Þ ¼ x2 þ y2  p2  y ¼ x2 þ  ¼ x2  2 ;
dx 4 2 4
ðE:9:2Þ
dgð pÞ
gð pÞ ! f ð xÞ ¼ gðpðxÞÞ  pð xÞx . . . pð x Þ : ¼ x; p21 p2 p2
dp gðp1 ; yÞ ¼ x2 þ y2  p1  x ¼ þ y2  1 ¼ y2  1 ;
4 2 4
ðE:8:5Þ
ðE:9:3Þ
where f(x) and g(p) are Legendre transforms of each other.
p 2 p2 p2 p2
There is a certain symmetry here, and the same transfor- gðp1 ; p2 Þ ¼ x2 þ y2  p1  x  p2  y ¼ 1 þ 2  1  2
4 4 2 2
mation takes us back and forth. 1 2 2
Consider the function f(x) = ax2. In this case ¼  ðp1 þ p2 Þ:
4
p ¼ dfdðxxÞ ¼ 2ax, and x ¼ 2a
p
. The Legendre transform of f is ðE:9:4Þ
p2 p2 p2 Example #10 Consider the function
L½f ðxÞ ¼ f ðxÞ  p  x ¼ ax2  2ax2 ¼  ¼
4a 2a 4a
¼ gðpÞ: x3
f ðx; y; zÞ ¼  : ðE:10:1Þ
ðE:8:6Þ yz
of 2 of 3
On the other hand, gð pÞ ¼  4a
2
p dgð pÞ p
, dp ¼  2a ¼ x, and The partial derivatives are ox ¼  3xyz ¼ fx , oy ¼ yx2 z ¼ fy ,
x3
the Legendre transform of g(p) is given as and of
oz ¼ yz2 ¼ fz .
x3=2
p2 p 2 p2 The variable y can be expressed as y ¼ 1=2 .
L½gðpÞ ¼ gðpÞ  p  ðxÞ ¼  þ ¼ ¼ ax2 z1=2 fy
4a 2a 4a Define the new function f2 (x, fy, z) as follows:
¼ f ðxÞ:
ðE:8:7Þ   x3 x3 z1=2 fy1=2 x3=2
f2 x; fy ; z ¼   yfy ¼   f
1=2 y
yz zx3=2 z1=2 fy
Remark To avoid the minus sign in Eq. (E.8.4), the
Legendre transform can alternatively be defined as 2x3=2 fy1=2
¼ : ðE:10:2Þ
z1=2
g ¼ px  f ðE:8:8Þ It is obvious from this definition that f2 is the Legendre
transform of f with respect to y. Taking into account that
(for functions of several variables: g(p1, x2, …, xm) = p1-
 
x1(p1) - f(x1(p1), x2, …, xm)), in which case the Legendre of2 3=2 1=2 1 3=2 x3=2 fy1=2
¼ 2x fy  z ¼ 3=2 ¼ fz ; ðE:10:3Þ
transform is its own inverse, since f = px - g, rather than oz 2 z
being the negative of its inverse. This opposite-sign alter-
native definition has the advantage that it gives rise to the and
symmetric identity f ? g = px which in words says that xfy1=3
the sum of a function and its Legendre transform equals the z¼ 2=3
; ðE:10:4Þ
fz
product of the conjugate pair of variables. It is worth to
emphasize the dimensional consistency of this identity. the Legendre transformation of f2 (with respect to z) yields

123
16 Page 16 of 17 ChemTexts (2015) 1:16

   
f3 x; fy ; fz ¼ f2 x; fy ; z  zfz Example #12 The Gibbs free energy function
2x3=2 fy1=2 1=3 xfy1=3 (G(T, p, n1, …, nm)) is obtained from the internal energy
¼  f
1=6 z
 f
2=3 z
¼ 3xfy1=3 fz1=3 : function (U = U(S, V, n1, …, nm)) via appropriate
x1=2 fy fz
Legendre transformations as
ðE:10:5Þ
Alternatively, Eq. (E.10.5) can be derived directly by G ¼ U  TS þ pV; ðE:12:1Þ
performing two successive Legendre transformations
  where S is the entropy, V is the volume, p is the pressure,
f3 x; fy ; fz ¼ f ðx; y; zÞ  yfy  zfz ðE:10:6Þ T is the temperature, and ni is the chemical amount of
with f ðx; y; zÞ ¼  xyz, of3
x3 of 3
¼ yzx 2 ¼ fz , y = component i. p and T are given, respectively, as p ¼  oU
oy ¼ y2 z ¼ fy , oz oV
and T ¼ oU . Note that the function G is a typical example
xf-2/3
y f1/3
z , and z = xf1/3
y fz
-2/-3
. oS
of a partly homogeneous function (see ‘‘The Gibbs–Duhem
Thus
equation’’ section).
x
f ðx; y; zÞ  yfy  zfz ¼ 2=3 1=3 1=3 2=3
 xfy2=3 fz1=3
xfy fz  xfy fz Acknowledgments The financial support from the Hungarian Sci-
1=3 2=3 entific Research Fund (Grants No. K 109036) is gratefully
 fy  xfy fz  fz
acknowledged.
1=3 1=3
¼ 3xfy fz :
ðE:10:6Þ
References

1. Lang GG, Barbero C (2012) Laser techniques for the study of

We remark that while f is a homogeneous function of the electrode processes. Springer, Heidelberg
first degree in the variables x, y, and z (see Example #1), 2. Koenig FO (1934) J Phys Chem 38:111–128
the Legendre transforms f2(x, fy, z) and f3(x, fy, fz) are only 3. Koenig FO (1934) J Phys Chem 38:339–363
partly homogeneous (the first with respect to x and z, and 4. Grahame DC, Whitney RB (1942) J Am Chem Soc
64:1548–1552
the second with respect to only x). 5. Graham DC (1947) Chem Rev 41:441–501
6. Mohilner DM (1962) J Phys Chem 66:724–726
Example #11 Consider the function
7. Hansen RS (1962) J Phys Chem 66:410–415
8. Soffer A (1972) J Electroanal Chem 40:153–165
9. Mohilner DM (1966) The electrical double layer. In: Bard AJ (ed)
x2 w
f ðx; y; u; wÞ ¼  þ u5 x: ðE:11:1Þ Electroanalytical chemistry, vol 1. Marcel-Decker, New York
y 10. Defay R, Prigigine I, Bellemans A, Everett DH (1966) Surface
of 2 tension and adsorption. Longmans, Bristol
Since fy ¼ oy ¼ xy2w, the Legendre transformation of 11. Goodrich FC (1968) Trans Farad Soc 64:3403–3415
f with respect to y yields 12. Parsons R (1984) Thermodynamic methods for the study of
  interfacial regions in electrochemical systems. In: Bockris JO’M,
Ly ½f ðx; y; u; wÞ ¼ f2 x; fy ; u; w ¼ 2xw1=2 fy1=2 þ u5 x: Conway BE, Yeager E (eds) Comprehensive treatise of electro-
ðE:11:2Þ chemistry, the double layer, vol 1. Plenum Press, New York
13. Everett DH (1987) Pure Appl Chem 59:45–52
Note that f (x,y,u,w) is a partly homogeneous function of 14. Girault HHJ, Schiffrin DJ (1989) Electrochemistry of liquid/liq-
uid interfaces. In: Bard AJ (ed) Electroanalytical Chemistry, vol
degree one in the variables x and y. The Legendre 15. Marcel Dekker, New York
transform is also a partly homogeneous function, but only 15. Damaskin BB, Petrii OA (2011) J Solid State Electrochem
with respect to x. 15:1317–1334
Consider the function 16. Gutman EM (2014) J Solid State Electrochem 18:3217–3237
17. Láng G, Heusler KE (1995) J Electroanal Chem 391:169–179
x2 18. Láng GG, Seo M, Heusler KE (2005) J Solid State Electrochem
f ðx; y; u; wÞ ¼  þ u5 x þ w 2 : ðE:11:3Þ 9:347–353
y
19. Gibbs JW (1948) Collected works, vol I. Yale University Press,
of 2
Since fy ¼ oy ¼ xy2 , and therefore y ¼ x
1=2 , the Legendre New Haven, pp 219–237
fy 20. Guggenheim EA, Adam NK (1933) Proc R Soc Lond
transformation of f with respect to y yields A139:218–236
  21. Guggenheim EA (1950) Thermodynamics, 2nd edn. Interscience
  x2 fy1=2 5 2 x x2 fy Publishers Inc, New York
f2 x; fy ; u; w ¼  þ u x þ w  1=2 22. Trasatti S, Parsons R (1986) Pure Appl Chem 58:437–454
x fy x2
23. Everett DH (1972) Manual of symbols and terminology for
¼ 2xfy1=2 þ u5 x þ w2 : ðE:11:4Þ physicochemical quantities and units, appendix ii, definitions,

123
ChemTexts (2015) 1:16 Page 17 of 17 16

terminology and symbols in colloid and surface chemistry, part I. 33. Courant R, Hilbert D (1966) Methods of mathematical physics,
Butterworths, London partial differential equations, vol II. Interscience Pubs., John
24. Eriksson JC (1969) Surf Sci 14:221–246 Wiley & Sons, New York
25. Eriksson JC (1965) Ark Kemi 25:331–341 34. Miller DG (1963) J Chem Educ 40:648–649
26. Linford RG (1978) Chem Rev 78:81–95 35. Duhem P (1886) Le Potentiel Thermodynamique et Ses Appli-
27. Randles JEB, Behr B (1972) J Electroanal Chem 35:389–404 cations à La Mécanique Chimique et à L’Etude des Phénomènes
28. Guggenheim EA (1940) Trans Farad Soc 36:397–412 Electriques. A. Hermann, Paris
29. Callen H (1985) Thermodynamics and an introduction to ther- 36. Annamalai K, Puri IK (2002) Advanced thermodynamics engi-
mostatistics, 2nd edn. Wiley, New York neering, CRC series in computational mechanics and applied
30. Láng G, Heusler KE (1994) J Electroanal Chem 377:1–7 analysis. CRC Press, Boca Raton
31. Heusler KE, Láng G (1997) Electrochim Acta 42:747–756
32. Kudryavtsev LD (2002) Homogeneous function. In: Hazewinkel
M (ed) Encyclopaedia of mathematics. Springer, Berlin

123