10.1351 Pac196102010207

IRREVERSIBLE THERMODYNAMICS
KENNETH s. PITZER
University of California, Berkeley, California, U.S.A.
There are alternative sets of postulates which yield entirely equivalent

results for several important theories. In such cases the choice of a parti-
cular set of postulates may be based upon mathematical simplicity, close
relationship to experiments or to the nature ofthe theory, or merely personal
taste. While the postulate of conservation of energy is usually presented
in similar form for thermodynamics or statistical mechanics, the form of
the postulate comprising the second law is commonly very different. In
thermodynamics the postulate may be in terms of the impossibility of an
isothermal engine producing work at the expense of heat while in quantum
statistical mechanics one postulates equal a priori probability for allindividual
quantum states. It is the purpose of the present paper to point out that
there are such alternatives for the additional postulate necessary to derive
the Onsager equations for the irreversible thermodynamics of processes in
the linear rate law region near equilibrium.
The heat engine, which was the primary subject of interest during the
original development of classical thermodynamics, comprises both reversible
and irreversible processes. In such machines the separation of the revers-
ible changes of the working fluid from the frictional effects and the irrevers-
ible heat leaks is practically self-evident. Thus in his dassie paper " On
the dynamical theory of heat ", Kelvin 1 assumes this separation of pro-
cesses to be obvious in ordinary heat engines \-vhere he simply specifies
" perfect engine ", but he discusses the separation at length for thermo-
electric phenomena as a separate postulate.
The magnificient work of Gibbs, van't Hoff, and others, showing the
applicability of thermodynamics to physicochemical equilibria distracted
attention for a time from steady-state problems. But in 1926 Eastman 2
applied Kelvin's method to thermo-osmosis, the Soret effect, and the
properties of thermocells. W agner 3 and London 4 used similar methods soon
thereafter. In these treatments it is assumed that the irreversible heat
flow constitutes a process which is independent of the processes transferring
the material components. Then the second law ofthermodynamics requires
the entropy tobe a maximum with respect to any transfer ofmatter and thus
determines the steady-state distribution.
The new formulation of the near-equilibrium problems in 1931 by Onsager 5
in terms of linear rate laws and his derivation of the reciprocal relationships
gave new impetus to theoretical work and de Groot 6 , Prigogine 7 and others
soon applied this new formulation to a wide variety of phenomena. It was
implied in some of these treatments that the new methods were more
rigorous in spite of the fact that in so far as the problems had been treated
previously by Kelvin's methods, the results were identical. However, the
P.A.C. (V. 2. NOS. 1-2)-14 207
KENNETH S. PITZER
new formulation is more convenient in many instances. Onsager's deriva-

tion of his reciprocal relationships is based upon the postulate of microscopic
reversibility and upon statistical arguments concerning the decay of fluctua-
tions. Among the several other statistical derivations which have been
given is that by Parlin, Marcus and Eyring8 in terms of absolute reaction
rate theory.
While the convenience and power of the Onsager formulation is a great
advantage, this may be attained from the Kelvin postulate of independent
processes, as has been demonstrated by Li 9 • lndeed, the present author
believes that this form of the basic postulate is most appropriate to thermo-
dynamics as weil as having a historical precedent. A simple derivation of
the Onsager relationships from the Kelvin postulate is given below. This
presentation is essentially similar to that of Li but is more concise and avoids
one possibly misleading implication *.
Following the usual definitions, one selects co-ordinates ai which measure
the amounts ofvarious components of matter, of heat, etc., transferred from
one portion of the system to another. The zeros of these co-ordinates are
selected at the state of equilibrium. Hence the entropy change from
equilibrium, to the second order, is
f:..S = - i .L
i, j
giJaiaJ (I)
The rate of change of a co-ordinate, äi is a flux which is commonly given the

symbol ]i· Also the restoring force or potential difference associated with
the i-th co-ordinate is defined by the equation
(2)
The rate of increase of entropy, within the region of validity of equation

(I), is
(3)
where we note the particularly simple form of the expression as a sum of

terms ]iXi.
In practice it is usually easier to write an expression for the rate of entropy
increase directly in terms of defined fluxes and differences of temperature,
pressure, concentration, etc., within the system. Then the potential con-
jugate to a given flux may be found by identifying the complete factor
which, multiplied by ]i, constitutes a term in the equation for S.
* I t should be noted that there is no limit upon the number of independent microscopic
processes. For example in a thermo-osmosis system there might be several different mem-
branes and each would have its characteristic heat of transfer for passage of matter. How-
ever, the number of fluxes required to describe the macroscopic system is fixed by the
number of components, etc. While Li's resultis identical with that given here, his use of a
square transformation matrix implies an unnecessary Iimitation of equal numbers of fluxes
in the two sets.
208
IRREVERSIBLE THERMODYNAMICS
At equilibrium all gradients Xi are zero and by the.rmodynamics all
ßuxes must be zero. Also for systems not at equilibrium S must be positive
regardless of the signs of the Xi. Consequently the ]i must change sign
with the xi and one expects a linear dependence of ]t on xi to be a good
approximation for small xi.
An arbitrary selection of ßuxes cannot be expected to yield those micro-
scopic processes which are actually independent of one another but we
postulate that such non-interacting processes exist in the region offirst-order rate laws.
Forthisspecial set ofßuxes, ]j, each ßux depends only upon its corresponding
potential gradient x;, hence
]j = L)Xj (4)
where Lj is the linear rate constant.

The arbitrary ßuxes ]i can be expressed in terms of the non-interacting
ßuxes Jf by a linear transformation
]i = 2 j
CXij]J (5)
and this transformation may be substituted into equation (3) for entropy
production
(6)
The coefficient of ]j in equation (6) is
x; = 2 cxijxii
(7)
Su bsti tu tion of these results and the rate eq ua tions (4) for the independen t
processes into equations (5) yields
(8)
and if we define a new rate constan t Lik :
(9)
Ltk = 2 j
CXij(XkjLj (10)
But it is evident from equation (10) that
Lik = Lkt ( 11)
which, for various i, k, constitute the Onsager reciprocal relationships.

209
KENNETH S. PITZER
References
1 W. Thomson (Lord Kelvin). Math. Phys. Papers, Val. I, Art. XLVIII, The University
Press, Garnbridge (1882); reprinted from articles in Trans. Roy. Soc. Edinburgh,
1851-1854
2
E. D. Eastman. J. Am. Chem. Soc., 48, 1482 (1926); 49, 794 (1927); 50,283,292 (1928)
3 C. Wagner. Ann. Physik., 3, 629 (1929); 6, 370 (1930)
4 H. London. Proc. Roy. Soc. London, A 171, 484 (1939)
6 L. Onsager. Phys. Rev., 37, 405 (1931); 38,2265 (1931)
6
S. R. de Groot. Thermodynamics oj Irreversible Processes, North Holland Publishing Co.,
Amsterdam ( 1951)
7 I. Prigogine. Introduction to Thermodynamics rif Irreversible Processes, C. C. Thomas, Publisher,
Springfield, Ill. ( 1955)

8 R. B. Parlin, R. J. Marcus and H. Eyring. Proc. Nat. Acad. Sei., U.S., 41, 900 (1955)
9 J. C. M. Li. J. Chem. Phys., 29, 747 (1958)
210

10.1351 Pac196102010207

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10.1351 Pac196102010207

Uploaded by

Copyright:

Available Formats

IRREVERSIBLE THERMODYNAMICS

University of California, Berkeley, California, U.S.A.

There are alternative sets of postulates which yield entirely equivalent

new formulation is more convenient in many instances. Onsager's deriva-

The rate of change of a co-ordinate, äi is a flux which is commonly given the

The rate of increase of entropy, within the region of validity of equation

where we note the particularly simple form of the expression as a sum of

where Lj is the linear rate constant.

The coefficient of ]j in equation (6) is

and if we define a new rate constan t Lik :

But it is evident from equation (10) that

Lik = Lkt ( 11)

which, for various i, k, constitute the Onsager reciprocal relationships.

Springfield, Ill. ( 1955)

You might also like