Basics of Irreversible Thermodynamics

Mukul Agrawal December 23, 2005



A general theory of non-equilibrium thermodynamics is not existing. Its a heavily researched area. Sometimes fudamental relation and the form of σ − U is considered as an additional assumption of statistical mechanics. It does give some insight into dynamics also but in very restrictive sense. So that assumption should still remain true even in dynamic situation. practically speaking. • Fundamental assumption of statistical mechanics still makes sense. only time reversibility assumption is needed apart from the two basic postulates of conventional statistical mechanics (ie fundamental assumption and the assumption about the profile of σ − U curve). If V is treated as a variable parameter for a system in equlibrium then it would remain so even in non-equilibrium. I think ateast three more postulates are needed :. monotonic. But from thermodynamic point of view. Its called time symmetry of physical laws. But almost all the processes we are mostly interested are irreversible processes and system is not in equilibrium state most the time when system is evolving. Most of the theory is dedicated to the macroscopic state reprenentation of a system in equilibrium state and then explaining the realtionship between various intensive and extensive equilibrium-state-variables. from statistical mechanical point of view. differentiable. one should note that its not possible for some variables to become independent or constrained when one moves in or out of equilibruim.1) knowing the intial equilibrium state and the final equilibrium state one can authoritatively say whether change from initial state to final state is possible spontaneously or not. But the thermodynamics of steady state processes is relatively well established atleast when the syatem does not deviate from equilibrium subtantially. and 2) if the change from initial state to final state is so slow that process can be assumed to be proceeding trough a series of closely spaced quasi-equilibrium states then such a process is called reversible process and the entire time evolution of each of the state variables can be obtianed from the conventional theory of thermodynamics. metals and other solid state systems. zero-starting function of extensive variables of the system. Hence instantaneous.CONTENTS Mukul Agrawal The conventional thermodynamic theory is. Hence the fundamental relation would remain same as that in equilibrium situtaion of the same infintesimaly small volume.Local Equilibrium. Time Reversibility. I am briefly writing about the basic structure of this theory below as its quite important for device physicsists to have a solid understanding of steady state irreversible process (like electric current and heat conduction) in semiconductor. • Localy and instantaneously its possible to represent the state of system using the same state variables as we did in equilibrium. local entropy can always be written as a single valued continuous. What I mean is this. Mostly the dynamic insight comes in two forms . a static theory. Moreover. • For studying irreversible processes we need atleast one more postulate. All laws of physics remain unaltered if we replace List of All Articles 2 . Basic Postulates and Definitions It seems. and form of Phenomenological Equation.

p is a tensor and Fp and jk are affinity and flux vectors respectively. – So for small perturbations one can write jk = ∑ Lk. In the markoffian system instantaneous local fluxes depends only on the instantaneous local affinities (and coefficient of propotinality would depend on instantaneous local intensive properties). • Affinities are defined as the gradient of all the intesive parameters ( actualy the gradients of coefficients in the expansion of d σ ) in the entropy representation. • Further it assumed that system is a Markoffian system. Fluxes are defined as the flux of flow density vector of any of the extensive property. q. But remeber that one type of flux can in general depend on all affinities and not only on its own affinity. This postulate can be used to prove that Li j = L ji where Lnm is the phemenological tensor coefficient. V. Entropy flux density vector and continuity equation for entropy Let us assume that the system has the following fundamental relation :U = U (σ. 1 µ p φ d σ = dU − dN − dV − dq + etc τ τ τ τ List of All Articles 3 . p Fp p Where Lk. Further for small perturbation theories one can assume the relationship to be linear.CONTENTS Mukul Agrawal t by −t everywhere. for generality. to depend on all types of affinities. N . dU = τd σ + µdN + pdV + φdq + etc This equation can be integrated to obtain :U = τσ + µN + pV + φq + etc Which can be inverted to obtain :µ p φ 1 σ = U − N − V − q + etc τ τ τ τ Using this we can define what we mean by entropy flux density vector :jσ = 1 µ p φ jU − jN − jV − jq + etc τ τ τ τ Also one invert the differential relation to obtain. etc) Hence. Note that the one type of flux is assumed.

jU + ∇. particles.∇( ) − jN .∇( ) − jN . Which contains the summation of dot products of all fluxes and corresponding affinities. Further Resources • Author’s Home Page • Other Articles List of All Articles 4 . d σ 1 dU µ dN p dV φ dq = − − − + etc dt τ dt τ dt τ dt τ dt Using the continuity equations for energy. volume and charge we can obtain (for all quantities writen as per unit volume). jσ = + jU . dσ 1 µ p φ + ∇.∇( ) − jq .∇( ) − jV .∇( ) − jq .∇( ) + etc τ τ τ τ is called the entropy production function.CONTENTS Mukul Agrawal Hence we can write. 1 µ p φ Φ = jU . jV + ∇.∇( ) + etc dt τ τ τ τ Which can be recognized as the continuity equation for entropy. jq + etc dt τ τ τ τ Which is same as. 1 µ p φ dσ = − ∇. jN + ∇.∇( ) − jV .