Irreversible Thermodynamics

• Classical thermodynamics deals with transitions from one equilibrium

state to another and since it does not analyse the changes between state
points it could be called thermostatics.
• The term thermodynamics will be reserved, in this chapter, for
dynamic non-equilibrium processes.
• In previous chapters, phenomenological laws have been given which
describe irreversible processes in the form of proportionalities, e.g.
Fourier's law of heat conduction, Ohm's law relating electrical current
and potential gradient, Fick's law relating flow of matter and
concentration gradient etc.
• When two of these phenomena occur simultaneously they interfere,
or couple, and give rise to new effects. One such cross-coupling is the
reciprocal effect of thermoelectricity and electrical conduction: the
Peltier effect (evolution or absorption of heat at a junction due to
the flow of electrical current) and thermoelectric force (due to
maintenance of the junctions at different temperatures).
• It is possible to examine irreversible phenomena by statistical mechanics
and the kinetic theory but these methods are on a molecular scale and do
not give a good macroscopic theory of the processes.
• Another method of considering non-equilibrium processes is based
on 'pseudo-thermostatic theories'. Here, the laws of thermostatics are
applied to a part of the irreversible process that is considered to be
reversible and the rest of the process is considered as irreversible
and not taken into account.
• Thomson applied the second law of thermostatics to thermoelectricity
by considering the Thomson and Peltier effects to be reversible and
the conduction effects to be irreversible.
• Systematic macroscopic and general thermodynamics of irreversible
processes can be obtained from a theorem published by Onsager
(1931a,b).
• This was developed from statistical mechanics and the derivation will
not be shown but the results will be used.
• The theory, based on Onsager's theorem, also shows why the
incorrect thermostatic methods give correct results in a number of cases.
 Definition of irreversible or steady state
thermodynamics
• All previous work on macroscopic 'thermodynamics' has been related to
equilibrium.
• A system was said to be in equilibrium when no spontaneous process took place and
all the thermodynamic properties remained unchanged.
• The macroscopic properties of the system were spatially and temporally invariant.

Fig. Steady state conduction of heat along a bar

• Consider the system shown in Fig., in which a thermally insulated
rod connects two reservoirs at temperatures T2 and T1 respectively.
• Heat flows between the two reservoirs by conduction along the rod.
• If the reservoirs are very large the conduction of energy out of, or
into, them can be considered not to affect them.
• A state will be achieved when the rate of heat flow, dQ/dt, entering
the rod equals the rate of heat flow, -dQ/dt, leaving the rod.
• If a thermometer were inserted at any point in the rod the reading
would not change with time, but it would be dependent on position.
• If the rod were of uniform cross-section the temperature gradient
would be linear. (This is the basis of the Searle's bar conduction
experiment.)
• The temperature in the bar is therefore a function of position but is
independent of time.
• The overall system is in a 'stationary' or 'steady' state but not
in 'equilibrium', for that requires that the temperature be
uniform throughout the system.
• If the metal bar were isolated from all the influences of the
surroundings and from the heat sources, i.e. if it is made an isolated
system, the difference between the steady state and equilibrium
becomes obvious.
• In the case where the system was in steady state, processes would
occur after the isolation (equalisation of temperature throughout the
bar); where the system was already in equilibrium, they would not.
• Entropy flow and entropy production
• Still considering the conduction example given above.
• If the heat flows into the left-hand end of the bar due to an
infinitesimal temperature difference, i.e. the process is reversible, the
left-hand reservoir loses entropy at the rate

• Similarly the Right-hand reservoir gains entropy at the rate

• Thus, the total change of entropy for the whole system is

• Now T2 > T1 and therefore the rate of change of entropy, dS/dt > 0.
• To understand the meaning of this result it is necessary to consider a
point in the bar.
• At the point l from the left-hand end the thermometer reading is T.
This reading is independent of time and is the reading obtained on the
thermometer in equilibrium with the particular volume of the rod in
contact with it.
• Hence, the thermometer indicates the 'temperature' of that volume of
the rod. Since the temperature is constant the system is in a 'steady
state', and at each point in the rod the entropy is invariant with
time.
• However, there is a net transfer of entropy from the left-hand
reservoir to the fight-hand reservoir, i.e. entropy is 'flowing' along
the rod. The total entropy of the composite system is increasing
with time and this phenomenon is known as 'entropy production'.
 Thermodynamic forces and thermodynamic
velocities
• It has been suggested that for systems not far removed from equilibrium the
development of the relations used in the thermodynamics of the steady state should
proceed along analogous lines to the study of the dynamics of particles, i.e. the laws
should be of the form
Fick's Law for the diffusion due to a concentration gradient is, in one dimension
 Onsager’s reciprocal relation
• If two transport processes are such that one has an effect on the other, e.g. heat
conduction and electricity in thermoelectricity, heat conduction and diffusion
of gases, etc., then the two processes are said to be coupled. The equations of
coupled processes may be written

It is obvious that in this equation the basic processes are defined by the
diagonal coefficients in the matrix, while the other processes are
defined by the off-diagonal terms.
Consider the situation where diffusion of matter is occurring with
a simultaneous conduction of heat. Both of these processes are
capable of transferring energy through a system.
• The diffusion process achieves this by mass transfer, i.e. each
molecule of matter carries some energy with it.
• The thermal conductivity process achieves the transfer of heat
by the molecular vibration of the matter transmitting energy
through the system.
• Both achieve a similar result of redistributing energy but by different
methods.
• The diffusion process also has the effect of redistributing the matter
throughout the system, in an attempt to achieve the equilibrium state
in which the matter is evenly distributed with the minimum of order
(i.e. the maximum entropy or minimum chemical potential).
• It can be shown, by a more complex argument, that thermal
conduction will also have an effect on diffusion. First, if each
process is considered in isolation the equations can be written
• Now, the diffusion of matter has an effect on the flow of energy
because the individual, diffusing, molecules carry energy with
them, and hence an effect for diffusion must be included in the
term for thermal flux, J1. Hence

where L12 is the coupling coefficient showing the effect of mass

transfer (diffusion) on energy transfer.
In a similar manner, because conduction has an effect on
diffusion, the equation for mass transfer can be written