Advanced Thermodynamics-IV: A Guide to Non-Equilibrium States
Alright, let's dive into one of the most interesting parts of thermodynamics! Classical thermodynamics, which you've studied so far, is fantastic... but it has a huge limitation: it only describes systems at equilibrium. The real world, from a chemical reaction to a living cell, is almost never at equilibrium. Things are constantly flowing, changing, and reacting.
So, how do we describe these non-equilibrium states? That's what this entire topic is about. We're going to build a framework to describe the *rates* of processes, not just their start and end points.
Tutor's Note: ⚡ Think of classical thermodynamics as a perfect, still photograph. It's beautiful and precise, but it can't show you the *motion*. Non-equilibrium thermodynamics (NET) is the video. It describes the blur, the flow, and the direction of change. It's the "how fast" and "why" of spontaneous processes.
The First Big Idea: The Postulate of Local Equilibrium
The first problem we face is that for a system not at equilibrium, temperature ($T$) and pressure ($P$) might not be uniform. So how can we even use our old thermodynamic functions like $S$, $U$, or $G$?
The solution is a clever assumption: the postulate of local equilibrium.
We imagine breaking our large non-equilibrium system into millions of tiny (infinitesimal) volume elements.
We assume that each tiny cell is, by itself, in a state of local equilibrium.
This means that within one tiny cell, we *can* define variables like temperature $T(\mathbf{r}, t)$, pressure $P(\mathbf{r}, t)$, and chemical potential $\mu_k(\mathbf{r}, t)$. Notice they now depend on position ($\mathbf{r}$) and time ($t$).
This assumption holds surprisingly well for most systems, as long as the changes are not *too* fast or *too* steep (like in a shockwave).
Because we assume local equilibrium, we can also assume that the fundamental Gibbs equation, which you know from equilibrium, still holds for each tiny cell.
The Entropy Balance: $dS = d_eS + d_iS$
This is the most important new equation. The total change in entropy ($dS$) in a system is split into two parts:
$d_eS$ (Entropy Flow): This is the entropy that crosses the boundary of the system. It's exchanged with the surroundings. For example, if heat ($dQ$) flows into your system at temperature $T$, then $d_eS = dQ/T$. This term can be positive, negative, or zero.
$d_iS$ (Entropy Production): This is the entropy *created* inside the system by irreversible processes. Think of friction, diffusion (mixing), a chemical reaction, or heat flowing across a temperature gradient *within* the system.
The Second Law of Thermodynamics, in its most powerful form, is a statement about $d_iS$:
For a reversible process: $d_iS = 0$
For an irreversible (real) process: $d_iS > 0$
It is *impossible* for $d_iS < 0$. You cannot destroy entropy.
We are almost always interested in the *rate* of change. So we write this equation in terms of time:
$$ \frac{dS}{dt} = \frac{d_eS}{dt} + \frac{d_iS}{dt} $$
Where $\frac{dS}{dt}$ is the total rate of entropy change, $\frac{d_eS}{dt}$ is the rate of entropy flow, and $\frac{d_iS}{dt}$ is the rate of internal entropy production. This last term, $\frac{d_iS}{dt}$, is the star of the show. We often call it $\sigma$ (or $\sigma_V$ if we're talking about per-unit-volume).
Finding the Entropy Production: Forces and Fluxes
Our main goal is to find a mathematical expression for $\sigma = \frac{d_iS}{dt}$. The methodology is to combine the Gibbs equation with the balance equations for energy and mass. This is a bit long to derive fully, but the result is *always* of the same beautiful, simple form:
$\sigma = \sum_k J_k X_k$
This is the general expression for the rate of entropy production (per unit volume). It's a sum of products of conjugate pairs:
$J_k$ (Fluxes): These are the *rates* of the irreversible processes. They represent a flow of some quantity per unit area per unit time. Examples: heat flow ($J_q$), matter diffusion flow ($J_k$), or electric current ($J_e$).
$X_k$ (Forces): These are the *driving forces* that *cause* the fluxes. They are typically gradients of thermodynamic potentials.
The whole trick of NET is to correctly identify these conjugate pairs. Let's look at the most common examples.
1. Entropy Production in Heat Flow
Flux ($J_q$): The flow of heat (energy) per unit area per time.
Force ($X_q$): The derivation shows this is the gradient of the *inverse* temperature: $X_q = \nabla \left(\frac{1}{T}\right) = -\frac{1}{T^2}\nabla T$.
Since $\sigma$ must be positive, this shows that $J_q$ (heat flow) and $\nabla T$ (the temperature gradient) must have opposite signs. This is a mathematical proof of what we know intuitively: heat flows from hot to cold.
2. Entropy Production in Matter Flow (Diffusion)
Flux ($J_k$): The flow of component $k$ (in moles per unit area per time).
Force ($X_k$): The force is the gradient of the chemical potential (at constant $T$): $X_k = -\nabla \left(\frac{\mu_k}{T}\right)$. At constant $T$, this simplifies to $X_k = -\frac{1}{T}\nabla \mu_k$.
This proves that matter flows from high chemical potential to low chemical potential.
3. Entropy Production in Chemical Reaction
This one is a bit different because it's a "scalar" process (no spatial direction).
Flux ($J_r$): The "flux" is the reaction rate, $v = \frac{d\xi}{dt}$ (where $\xi$ is the extent of reaction).
Force ($X_r$): The "force" is the Chemical Affinity, $A$, divided by temperature. Affinity is defined as $A = -\sum_j \nu_j \mu_j$. (Remember, $\Delta_r G = \sum_j \nu_j \mu_j$, so $A = -\Delta_r G$).
Since $\frac{d_iS}{dt}$ must be positive, this means the reaction rate ($J_r$) and the Affinity ($A$) must have the same sign. If a reaction is spontaneous ($A > 0$, meaning $\Delta_r G < 0$), then the rate must be positive ($J_r > 0$). This elegantly connects the thermodynamic driving force ($A$) with the kinetic rate ($J_r$).
4. Entropy Production in Electrical Conductors
Flux ($J_e$): The electric current density.
Force ($X_e$): The gradient of the electrical potential, $-\nabla \phi$, (which is the electric field $E$) divided by $T$. $X_e = -\frac{1}{T}\nabla \phi = \frac{E}{T}$.
Using Ohm's Law ($E = \rho J_e$, where $\rho$ is resistivity) and $R = \rho l/A_{area}$, $I = J_e A_{area}$, $V = El$, you can show this for a whole resistor: $\frac{d_iS}{dt} = \frac{I^2 R}{T}$. This is the entropy production associated with Joule heating.
The Linear Regime and The Phenomenological Laws
We've identified the forces ($X_k$) and fluxes ($J_k$). Now, how are they related?
In general, this relationship is very complex. But... for systems "close" to equilibrium, the forces ($X_k$) are small. In this case, we can assume the simplest possible relationship: a linear one. This is the linear regime.
We state that "every flux is a linear function of *all* the forces." These are the Linear or Phenomenological Laws.
For a system with two forces, $X_1$ and $X_2$, and two fluxes, $J_1$ and $J_2$:
$$ J_1 = L_{11} X_1 + L_{12} X_2 $$
$$ J_2 = L_{21} X_1 + L_{22} X_2 $$
The $L_{ik}$ are called the phenomenological coefficients.
$L_{11}$ and $L_{22}$ are the "main" coefficients. $L_{11}$ relates flux $J_1$ to its *own* force $X_1$. (e.g., $L_{11}$ would be related to thermal conductivity, $L_{22}$ to a diffusion coefficient).
$L_{12}$ and $L_{21}$ are the "cross" or "coupling" coefficients. $L_{12}$ describes how force $X_2$ can cause flux $J_1$! This is the mathematical description of coupled processes.
Tutor's Note: ⚡ This is the really cool part! A temperature gradient (a "heat" force $X_q$) can cause a flow of matter (a "diffusion" flux $J_d$). And a concentration gradient (a "diffusion" force $X_d$) can cause a flow of heat (a "heat" flux $J_q$). This is not an assumption; it's a measured fact, and these equations give us a way to handle it.
Onsager's Reciprocal Relations
In 1931, Lars Onsager proved something truly remarkable about the cross-coefficients, which won him the Nobel Prize. Based on the principle of microscopic reversibility (at the molecular level, physics equations work the same forwards and backwards in time), he showed that:
$L_{ik} = L_{ki}$
(This only holds if you've chosen the conjugate forces and fluxes correctly from the $\sigma$ equation).
This is a profound statement of symmetry in nature. In our 2x2 example, it means $L_{12} = L_{21}$. The effect of force 2 on flux 1 is *identical* to the effect of force 1 on flux 2. This halves the number of independent coefficients we need to measure and connects two seemingly different coupled phenomena.
Applications of Coupled Flows
1. Thermoelectric Effects (Coupling of Heat and Electricity)
Let $J_e$ be the electric flux and $J_q$ be the heat flux.
$$ J_e = L_{ee} X_e + L_{eq} X_q $$
$$ J_q = L_{qe} X_e + L_{qq} X_q $$
With Onsager's relation: $L_{eq} = L_{qe}$.
Seebeck Effect: We create a temperature gradient ($X_q \neq 0$) but allow no current to flow ($J_e = 0$). The first equation becomes $0 = L_{ee} X_e + L_{eq} X_q$, which rearranges to $X_e = -\frac{L_{eq}}{L_{ee}} X_q$. This means a temperature gradient *creates* an electrical potential gradient (a voltage). This is the principle of a thermocouple.
Peltier Effect: We pass a current ($X_e \neq 0$) in an isothermal system ($X_q = 0$). The second equation becomes $J_q = L_{qe} X_e$. This means the electric current *causes* a flow of heat. This is the principle of thermoelectric cooling.
2. Thermodiffusion (Coupling of Heat and Matter)
Let $J_d$ be the diffusion flux and $J_q$ be the heat flux.
$$ J_d = L_{dd} X_d + L_{dq} X_q $$
$$ J_q = L_{qd} X_d + L_{qq} X_q $$
With Onsager's relation: $L_{dq} = L_{qd}$.
Soret Effect (Thermodiffusion): We take a uniform mixture ($X_d = 0$ initially) and apply a temperature gradient ($X_q \neq 0$). The first equation says $J_d = L_{dq} X_q$. This means the temperature gradient will *cause* a flow of matter. One component will move to the hot side and the other to the cold side, creating a concentration gradient ($X_d \neq 0$).
Dufour Effect: We take an isothermal system ($X_q = 0$ initially) and create a concentration gradient ($X_d \neq 0$). The second equation says $J_q = L_{qd} X_d$. This means the diffusion of matter will *cause* a flow of heat, creating a temperature gradient. This is the reciprocal of the Soret effect and is much harder to observe.
Non-Equilibrium Stationary States (NESS)
Finally, what happens to a non-equilibrium system over time?
An isolated system will evolve until it reaches equilibrium. At this point, all forces and fluxes are zero ($X_k=0, J_k=0$) and entropy production ceases ($\sigma = 0$). Entropy is maximized.
An open system (exchanging energy/matter with its surroundings) may evolve to a Non-Equilibrium Stationary State (NESS).
A NESS (or steady state) is a state where the macroscopic properties (like $T$ or concentration) are constant in time, but the fluxes and forces are non-zero.
Example: A metal rod held at $T_1 = 100^\circ$C on one end and $T_2 = 0^\circ$C on the other. Heat is constantly flowing ($J_q \neq 0$) driven by a force ($\nabla T \neq 0$), so $\sigma > 0$. But after a while, the temperature at any point *in* the rod becomes constant. This is a classic NESS.
Prigogine's Principle of Minimum Entropy Production
Ilya Prigogine (another Nobel winner) proved a key principle for these stationary states. He showed that:
For a system in the linear regime ($J=LX$) and subject to fixed boundary conditions, the NESS it evolves to is the state of minimum entropy production.
This means that out of all possible ways the system *could* be steady, it "chooses" the one that dissipates the least amount of energy (produces the least entropy). This principle is a powerful stability criterion for non-equilibrium systems, showing that nature finds an "optimal" path of least resistance, even when it's not at equilibrium.
Summary: The Methodology of NET
Start with the Postulate of Local Equilibrium.
Use the Gibbs equation and balance laws to derive the Entropy Production rate $\sigma = \sum_k J_k X_k$.
This derivation automatically identifies the conjugate Forces ($X_k$) and Fluxes ($J_k$).
Assume a Linear Regime (close to equilibrium) and write the Phenomenological Laws ($J_i = \sum_k L_{ik} X_k$).
Apply Onsager's Reciprocal Relations ($L_{ik} = L_{ki}$) to relate the cross-coefficients.
Use this set of equations to describe coupled processes and stationary states (NESS), which are governed by the Principle of Minimum Entropy Production.
I hope this detailed walkthrough helps you master these concepts. It's a very elegant and powerful extension of the thermodynamics you already know! ⚡
Answers to Advanced Thermodynamics-IV Questions
Here are the detailed, step-by-step explanations for each of the questions on your Advanced Thermodynamics-IV syllabus. Let's walk through them like a tutorial.
1. Gibbs Equation for Non-Equilibrium Open Systems
This is the starting point for almost everything in NET. Here's how we get there:
The Problem: In a non-equilibrium system, temperature ($T$), pressure ($P$), and chemical potential ($\mu$) aren't uniform. So, how can we even use $dU = TdS - PdV + \sum \mu_k dn_k$?
The Solution (Postulate): We use the Postulate of Local Equilibrium. We imagine our big system is made of tiny (infinitesimal) cells. We *assume* that each tiny cell is itself in equilibrium. This means we can define local variables like $T(\mathbf{r}, t)$ (temperature at position $\mathbf{r}$ and time $t$), $P(\mathbf{r}, t)$, etc.
Adapting the Gibbs Equation: Because we assume local equilibrium, we also assume the Gibbs equation holds *locally*. It's more convenient to use specific quantities (per unit mass) or densities (per unit volume). Let's use $s$, $u$, and $v$ as the entropy, internal energy, and volume per unit mass (specific quantities). The Gibbs equation for a local cell is:
$$ du = Tds - Pdv + \sum_k \mu_k dc_k $$
where $c_k$ is the mass fraction of component $k$.
The Fundamental Form: By rearranging this, we get the local form of the Gibbs equation for entropy:
$$ ds = \frac{1}{T}du + \frac{P}{T}dv - \sum_k \frac{\mu_k}{T} dc_k $$
Tutor's Note: ⚡ This equation is our "Rosetta Stone." It connects our equilibrium state functions ($s, u, \mu, T$) to each other, even in a non-equilibrium system. Our next step in any derivation is to take the time derivative of this equation ($\frac{ds}{dt}$). We then plug in "balance equations" (for $\frac{du}{dt}$ and $\frac{dc_k}{dt}$) which describe how energy and mass flow. The algebra that follows will magically separate all the terms into two groups: one for entropy *flow* and one for entropy *production*. The entropy production part gives us the forces and fluxes.
2. Postulates and Methodology of Linear NET
The entire methodology of Linear Non-Equilibrium Thermodynamics (LNET) rests on a few key ideas:
Postulate 1: Local Equilibrium (As above). All our equilibrium thermodynamic relationships are assumed to hold locally.
Postulate 2: The Linear Regime (Linear Laws). This is the core assumption. We assume the system is "close" to equilibrium. This means the thermodynamic forces ($X_k$) driving the processes are *small*.
If the forces are small, we can assume the simplest possible relationship between a flux ($J_i$) and the forces: a linear one.
We state that any flux $J_i$ is a linear combination of *all* the forces $X_k$ present in the system.
Mathematically: $J_i = \sum_k L_{ik} X_k$
These are the phenomenological laws. The $L_{ik}$ are the phenomenological coefficients (like conductivity, diffusivity).
Postulate 3: Onsager's Reciprocal Relations. This is a third, crucial postulate (covered in Q8). It states that the matrix of coefficients $L$ is symmetric: $L_{ik} = L_{ki}$.
The Methodology is:
Use the Gibbs equation (from Q1) and conservation laws to derive the entropy production $\sigma = \sum_k J_k X_k$.
This derivation identifies the conjugate forces ($X_k$) and fluxes ($J_k$).
Assume the system is in the linear regime and write the phenomenological laws ($J_i = \sum_k L_{ik} X_k$).
Apply Onsager's relations ($L_{ik} = L_{ki}$) to simplify the equations and relate coupled processes.
3. The Entropy Balance Equation
This is one of the most important equations in this field. Let's derive it.
Total Entropy: The total entropy $S$ of a system is the integral of the local entropy density $s_v$ (entropy per unit volume) over the total volume $V$.
$$ S = \int_V s_v dV $$
Rate of Change: The rate of change of total entropy is the time derivative:
$$ \frac{dS}{dt} = \int_V \frac{\partial s_v}{\partial t} dV $$
Local Continuity Equation: The change in entropy density at a point ($\frac{\partial s_v}{\partial t}$) can only happen for two reasons: entropy flows in or out (a flux, $\mathbf{J}_s$), or entropy is created on the spot (a production source, $\sigma$). This is a standard continuity equation:
$$ \frac{\partial s_v}{\partial t} = - \nabla \cdot \mathbf{J}_s + \sigma $$
(The $-\nabla \cdot \mathbf{J}_s$ term is the divergence, which measures the net "flow out" of that point).
Substitute and Integrate: We plug this back into our integral:
$$ \frac{dS}{dt} = \int_V (- \nabla \cdot \mathbf{J}_s + \sigma) dV = - \int_V \nabla \cdot \mathbf{J}_s dV + \int_V \sigma dV $$
Gauss's Divergence Theorem: This theorem is a mathematical tool that relates a volume integral of a divergence to a surface integral of the flux over the boundary ($A$) of that volume.
$$ \int_V \nabla \cdot \mathbf{J}_s dV = \int_A \mathbf{J}_s \cdot d\mathbf{A} $$
The term $\int_A \mathbf{J}_s \cdot d\mathbf{A}$ is the total entropy flux *leaving* the system through its boundary.
Final Equation: Substituting this back gives:
$$ \frac{dS}{dt} = - \int_A \mathbf{J}_s \cdot d\mathbf{A} + \int_V \sigma dV $$
We define the entropy flow term $\frac{d_eS}{dt}$ as the entropy exchanged with the surroundings (flow *into* the system):
$$ \frac{d_eS}{dt} = - \int_A \mathbf{J}_s \cdot d\mathbf{A} $$
And we define the entropy production term $\frac{d_iS}{dt}$ as the total entropy created by irreversible processes *inside* the system:
$$ \frac{d_iS}{dt} = \int_V \sigma dV $$
This gives the final, famous entropy balance equation:
$$ \frac{dS}{dt} = \frac{d_eS}{dt} + \frac{d_iS}{dt} $$
Proof of Positivity: The Second Law of Thermodynamics is the statement that entropy can be exchanged or created, but *never destroyed*. This means the internal production term $\sigma$ must *always* be non-negative ($\sigma \ge 0$). Since $\frac{d_iS}{dt}$ is the integral of this non-negative quantity over the whole volume, it must also be non-negative:
$$ \frac{d_iS}{dt} \ge 0 $$
It is zero only for a perfectly reversible process. For any real, spontaneous, irreversible process, $\frac{d_iS}{dt} > 0$.
4. Entropy Production in Heat Flow
To find $\sigma$ for a specific process, we start from the full (and complex) derivation of $\sigma$, which comes from combining the Gibbs equation (Q1) with conservation laws. That derivation gives:
For pure heat flow, we assume there is no matter flow ($\mathbf{J}_k = 0$ for all $k$). The energy flux $\mathbf{J}_u$ is then just the heat flux $\mathbf{J}_q$.
The general expression for $\sigma$ simplifies to just the first term:
$$ \sigma = \mathbf{J}_q \cdot \nabla\left(\frac{1}{T}\right) $$
This is already in our desired $\sigma = J \cdot X$ form!
Flux ($J_q$): The heat flux vector $\mathbf{J}_q$.
Force ($X_q$): The conjugate force is $\mathbf{X}_q = \nabla\left(\frac{1}{T}\right)$.
Alternative Form: We can use the chain rule to expand the force:
$$ \mathbf{X}_q = \nabla\left(\frac{1}{T}\right) = \frac{d(1/T)}{dT} \nabla T = -\frac{1}{T^2} \nabla T $$
Plugging this in gives the entropy production in its most common form:
$$ \sigma = \mathbf{J}_q \cdot \left( -\frac{1}{T^2} \nabla T \right) $$
Tutor's Note: ⚡ Since we know $\sigma$ must be positive, and $T^2$ is positive, this equation $\sigma = -(\mathbf{J}_q \cdot \nabla T)/T^2$ mathematically *proves* that the heat flux vector $\mathbf{J}_q$ and the temperature gradient vector $\nabla T$ must point in opposite directions. This is the formal proof of Fourier's Law: heat flows from hot to cold.
5. Entropy Production in Matter Flow (Diffusion)
We start from the same general expression for $\sigma$:
$$ \sigma = \mathbf{J}_u \cdot \nabla\left(\frac{1}{T}\right) - \sum_k \mathbf{J}_k \cdot \nabla\left(\frac{\mu_k}{T}\right) $$
For pure diffusion, we make two key assumptions:
Isothermal System: The temperature is uniform, so $\nabla T = 0$. This makes the first term $\nabla(1/T)$ equal to zero, and the entire energy flux term vanishes.
Isobaric System: We can also assume $\nabla P = 0$.
The expression for $\sigma$ simplifies to just the matter flow terms:
$$ \sigma = - \sum_k \mathbf{J}_k \cdot \nabla\left(\frac{\mu_k}{T}\right) $$
Since we are in an isothermal system, $T$ is a constant, so we can pull it out of the gradient:
$$ \sigma = - \sum_k \mathbf{J}_k \cdot \left( \frac{1}{T} \nabla \mu_k \right) $$
Rearranging to get the standard $\sigma = \sum J \cdot X$ form:
$$ \sigma = \sum_k \mathbf{J}_k \cdot \left( -\frac{1}{T} \nabla \mu_k \right) $$
This immediately identifies the conjugate pairs:
Flux ($J_k$): The diffusion flux of component $k$, $\mathbf{J}_k$.
Force ($X_k$): The conjugate force is $\mathbf{X}_k = -\frac{1}{T} \nabla \mu_k$.
Tutor's Note: ⚡ This proves Fick's Law. Since $\sigma$ must be positive, $\mathbf{J}_k$ and $-\nabla \mu_k$ must point in the same direction. This means $\mathbf{J}_k$ and $\nabla \mu_k$ point in opposite directions. Matter flows from a region of high chemical potential to low chemical potential.
6. Entropy Production in Chemical Reaction
This process is different because it's scalar (it happens *at* a point, not *across* a point). There are no gradients ($\nabla$).
Let's analyze a closed system at constant $T$ and $P$. For any change, the Gibbs free energy change is $dG = dH - TdS$.
For an irreversible process, $TdS = dQ_{irrev} + d_iS \cdot T$. Since $dH = dQ_{irrev}$ (at const $P$), we get $TdS = dH + d_iS \cdot T$.
Substitute this into the $dG$ equation: $dG = dH - (dH + d_iS \cdot T) = -T d_iS$.
This gives us a crucial link: $d_iS = -\frac{dG}{T}$ (at constant $T, P$).
Now, let's look at the change $dG$ during a single reaction:
$$ jA + kB \rightleftharpoons lC + mD $$
The change in $G$ is $dG = \sum_i \mu_i dn_i$.
The change in moles $dn_i$ is related to the extent of reaction ($\xi$) by the stoichiometric coefficient $\nu_i$ (negative for reactants, positive for products): $dn_i = \nu_i d\xi$.
The term in parentheses is $\Delta_r G$, the reaction Gibbs free energy. We define a new term, the Chemical Affinity ($A$), as the negative of this:
$$ A = - \Delta_r G = - \sum_i \nu_i \mu_i $$
So, $dG = -A d\xi$.
Now put it all together: $d_iS = -\frac{dG}{T} = -\frac{(-A d\xi)}{T} = \frac{A}{T} d\xi$.
Finally, we look at the rate of entropy production by taking the time derivative:
$$ \frac{d_iS}{dt} = \frac{A}{T} \left( \frac{d\xi}{dt} \right) $$
We define the reaction rate as a "flux" ($J_r$) and the affinity term as the "force" ($X_r$):
Flux ($J_r$): The reaction rate, $J_r = \frac{d\xi}{dt}$.
Force ($X_r$): The chemical affinity, $X_r = \frac{A}{T}$.
This gives the final form: $\frac{d_iS}{dt} = X_r \cdot J_r$.
7. Phenomenological Laws and Coupled Processes
This is where we connect the forces and fluxes.
Phenomenological Laws: These are laws that relate fluxes to forces. Simple examples are Fourier's Law ($J_q = -\kappa \nabla T$) and Fick's Law ($J_d = -D \nabla c$). You can see these are laws of the form $J = L \cdot X$.
The Linear Generalization: In the linear regime (see Q2), NET generalizes this. It says that any flux ($J_i$) can be driven by any force ($X_k$) present in the system.
$$ J_i = \sum_k L_{ik} X_k $$
Example (Two Coupled Processes): Let's say we have heat flow (flux $J_1$, force $X_1$) and matter flow (flux $J_2$, force $X_2$) happening at the same time. The laws are:
$$ J_1 = L_{11} X_1 + L_{12} X_2 $$
$$ J_2 = L_{21} X_1 + L_{22} X_2 $$
Defining the Coefficients:
$L_{11}, L_{22}$ (Main Coefficients): These relate a flux to its *own* conjugate force. $L_{11}$ relates $J_1$ (heat) to $X_1$ (temp. gradient). It's the thermal conductivity. $L_{22}$ relates $J_2$ (matter) to $X_2$ (conc. gradient). It's the diffusion coefficient.
$L_{12}, L_{21}$ (Cross-Coefficients): These are the interesting ones! They describe the coupling.
$L_{12}$: Describes how force $X_2$ (a concentration gradient) can cause flux $J_1$ (a heat flow). This is the Dufour Effect.
$L_{21}$: Describes how force $X_1$ (a temperature gradient) can cause flux $J_2$ (a matter flow). This is the Soret Effect.
8. Onsager's Reciprocal Relations
This is a Nobel Prize-winning discovery by Lars Onsager, and it's a cornerstone of NET.
The Statement: For any system in the linear regime, if the forces and fluxes are chosen *correctly* from the entropy production equation ($\sigma = \sum J_k X_k$), then the matrix of phenomenological coefficients $L$ is symmetric.
$$ L_{ik} = L_{ki} $$
The Microscopic Principle: The proof is very advanced, but it's based on microscopic reversibility and the statistical mechanics of fluctuations.
Microscopic Reversibility: At the molecular level, the laws of physics (like $F=ma$ or the Schrödinger equation) work the same forwards and backwards in time. A video of two molecules colliding looks just as valid if played in reverse.
Onsager's Insight: He considered how a tiny, random fluctuation (like a momentary concentration difference) decays back to equilibrium. He argued that on average, the *macroscopic* decay of this fluctuation (described by $J=LX$) must follow the same time-symmetric laws as the *microscopic* particle motions. By correlating fluctuations at two different times, he proved that for this to be true, the $L$ matrix *must* be symmetric.
Significance:
It's a "law of nature." It shows a fundamental symmetry in transport processes.
It reduces the work: In our 2x2 example from Q7, we don't need to measure 4 coefficients ($L_{11}, L_{12}, L_{21}, L_{22}$). Since $L_{12} = L_{21}$, we only need to measure 3.
It connects phenomena: It proves that the Soret effect ($L_{21}$) and the Dufour effect ($L_{12}$) are not independent. The coefficient describing how a $\nabla T$ causes matter flow is *identical* to the one describing how a $\nabla c$ causes heat flow.
9. Thermoelectric Effects (Seebeck and Peltier)
This is the classic example of coupled flow: the coupling between electricity (flux $\mathbf{J}_e$) and heat (flux $\mathbf{J}_q$).
The phenomenological equations are (using a practical form):
$$ \mathbf{E} = \rho \mathbf{J}_e + \epsilon \nabla T $$
$$ \mathbf{J}_q = \Pi \mathbf{J}_e - \kappa \nabla T $$
Where $\mathbf{E}$ is the electric field, $\rho$ is resistivity, $\epsilon$ is the Seebeck coefficient, $\mathbf{J}_q$ is heat flux, $\Pi$ is the Peltier coefficient, and $\kappa$ is thermal conductivity. Onsager's relation for this system (the Kelvin Relation) proves that $\Pi = \epsilon T$.
The Seebeck Effect
What it is: A temperature gradient creates a voltage. This is the principle of a thermocouple.
How it works:
Take a circuit of two different metals, (A-B). Make one junction hot ($T_H$) and one cold ($T_C$). This creates a $\nabla T \neq 0$.
Measure the voltage with a voltmeter. A voltmeter has very high resistance, so no current flows ($J_e = 0$).
Look at the first equation:
$$ \mathbf{E} = \rho (0) + \epsilon \nabla T \implies \mathbf{E} = \epsilon \nabla T $$
This shows that the temperature gradient has *caused* an electric field $\mathbf{E}$. Integrating this field around the loop gives the open-circuit voltage $V_{Seebeck}$, which is proportional to the temperature difference.
The Peltier Effect
What it is: An electric current causes heating or cooling at a junction. This is the principle of a thermoelectric cooler.
How it works:
Take a junction of two different metals (A-B). Keep the whole thing at a constant temperature ($\nabla T = 0$).
Use a battery to force a current $\mathbf{J}_e$ to flow through the junction.
Look at the second equation:
$$ \mathbf{J}_q = \Pi \mathbf{J}_e - \kappa (0) \implies \mathbf{J}_q = \Pi \mathbf{J}_e $$
This shows that the electric current $\mathbf{J}_e$ *causes* a heat flow $\mathbf{J}_q$ at the junction. To keep the temperature constant, heat must be actively pumped in or out. This is the Peltier heat. Depending on the direction of the current, the junction will either get hot (releasing heat) or cold (absorbing heat).
10. Entropy Production in an Electrical Conductor (Joule Heating)
Let's derive the entropy production from Joule heating.
Starting Point: We are looking at a charged species (electrons) moving in a conductor. From Q5, the force for matter flow is $X_e = -\frac{1}{T}\nabla \tilde{\mu}_e$, where $\tilde{\mu}_e$ is the electrochemical potential: $\tilde{\mu}_e = \mu_e + zF\phi$. Here $z=-1$ for electrons.
So, $\mathbf{X}_e = -\frac{1}{T} \nabla(\mu_e - F\phi)$.
Let's assume the conductor is uniform (so $\nabla \mu_e = 0$) and isothermal ($\nabla T = 0$, so $T$ is a constant).
The force simplifies: $\mathbf{X}_e = -\frac{1}{T} \nabla(-F\phi) = \frac{F}{T} \nabla\phi$.
The flux of electrons is $\mathbf{J}_e$. The electric current density is $\mathbf{J}_{el} = -F\mathbf{J}_e$. So, $\mathbf{J}_e = -\mathbf{J}_{el}/F$.
The entropy production $\sigma = J_e \cdot X_e$ is:
$$ \sigma = \left(-\frac{\mathbf{J}_{el}}{F}\right) \cdot \left( \frac{F}{T} \nabla\phi \right) = - \frac{1}{T} \mathbf{J}_{el} \cdot \nabla\phi $$
We know the electric field is the negative gradient of the potential: $\mathbf{E} = -\nabla\phi$.
Substitute this in: $\sigma = - \frac{1}{T} \mathbf{J}_{el} \cdot (-\mathbf{E}) = \frac{\mathbf{J}_{el} \cdot \mathbf{E}}{T}$.
This is the entropy production per unit volume.
Relation to Ohm's Law:
Ohm's Law is $\mathbf{J}_{el} = \kappa \mathbf{E}$ (where $\kappa$ is conductivity).
The phenomenological law is $J_k = L_{kk} X_k$. Here, $J_k = \mathbf{J}_{el}$. The conjugate force is $X_k = \mathbf{E}/T$.
So, $\mathbf{J}_{el} = L_{ee} (\mathbf{E}/T)$. Comparing with Ohm's Law, we see the phenomenological coefficient is $L_{ee} = \kappa T$.
Final Form: Substitute Ohm's law into our $\sigma$ expression:
$$ \sigma = \frac{(\kappa \mathbf{E}) \cdot \mathbf{E}}{T} = \frac{\kappa E^2}{T} $$
Or, using $\mathbf{E} = \rho \mathbf{J}_{el}$ (where $\rho = 1/\kappa$ is resistivity):
$$ \sigma = \frac{\mathbf{J}_{el} \cdot (\rho \mathbf{J}_{el})}{T} = \frac{\rho J_{el}^2}{T} $$
For a whole resistor of volume $V=A \cdot L$, the total entropy production is $\frac{d_iS}{dt} = \sigma \cdot V = (\frac{\rho J_{el}^2}{T}) (AL)$.
Using $I = J_{el} A$ and $R = \rho L/A$, this becomes:
$$ \frac{d_iS}{dt} = \frac{I^2 R}{T} $$
The power dissipated as heat is $P = I^2 R$. The entropy production is this dissipated power divided by temperature, $T\sigma = P$. This is the formal thermodynamic description of Joule heating.
11. Non-Equilibrium Stationary State (NESS) and Prigogine's Principle
Non-Equilibrium Stationary State (NESS)
A NESS (or steady state) is the non-equilibrium equivalent of an equilibrium state, but for open systems.
It is a state where the macroscopic properties (like $T(\mathbf{r})$, $c(\mathbf{r})$) are constant in time ($\frac{\partial T}{\partial t} = 0$, etc.).
Unlike equilibrium, a NESS is characterized by:
Non-zero forces (e.g., a permanent $\nabla T$).
Non-zero fluxes (e.g., a constant heat flow $\mathbf{J}_q$).
A constant, positive rate of entropy production ($\frac{d_iS}{dt} > 0$).
In a NESS, the total entropy of the system is constant ($\frac{dS}{dt} = 0$) because the internal production ($\frac{d_iS}{dt} > 0$) is perfectly balanced by an entropy flow *out* of the system ($\frac{d_eS}{dt} < 0$).
Example: A metal rod with its ends held at $T_H = 100^\circ$C and $T_C = 0^\circ$C. After a while, the temperature at any point in the rod is constant, but heat is flowing continuously. This is a classic NESS.
Prigogine's Principle of Minimum Entropy Production
Statement: For a system that obeys the linear laws ($J=LX$) and is subject to fixed boundary conditions (like the $T_H, T_C$ in the example), the non-equilibrium stationary state that the system evolves to is the one that has the minimum possible total entropy production ($\frac{d_iS}{dt} = \text{minimum}$) for those given constraints.
Explanation: The system "chooses" the most "efficient" stable state, the one that dissipates the least energy. It's a stability criterion for the linear regime. The NESS is the stable state because any fluctuation would (in the linear regime) increase the entropy production, and the system would relax back to the state of minimum production.
Conditions for Validity: This principle is powerful but only* applies if:
The system is in the linear regime ($J=LX$).
Onsager's relations ($L_{ik} = L_{ki}$) hold.
The boundary conditions are fixed and time-independent.
12. Thermodynamic Criteria: Equilibrium vs. Non-Equilibrium
Criteria for Equilibrium
Macroscopic: The system is stable, and its properties are uniform in space and constant in time. No flows occur.
Forces and Fluxes: All thermodynamic forces and fluxes are zero. ($X_k=0, J_k=0$).
Entropy Production: The rate of internal entropy production is zero. $\frac{d_iS}{dt} = 0$.
Thermodynamic Potentials: The state is defined by an extremum (a minimum or maximum) of a thermodynamic potential:
Isolated System (const $U, V$): Entropy $S$ is maximized.
Const $T, P$ System: Gibbs Free Energy $G$ is minimized.
Const $T, V$ System: Helmholtz Free Energy $A$ is minimized.
Criteria for Non-Equilibrium States
Macroscopic: Properties are *not* uniform in space. They may or may not be constant in time.
Forces and Fluxes: At least one force and one flux are non-zero. ($X_k \neq 0, J_k \neq 0$).
Entropy Production: The rate of internal entropy production is strictly positive. $\frac{d_iS}{dt} > 0$.
Thermodynamic Potentials (The Adaptation): We cannot use $S, G, A$ for the *whole* system as criteria. Instead, we use the Postulate of Local Equilibrium.
We assume that *local* thermodynamic functions $s(\mathbf{r},t), g(\mathbf{r},t)$ exist.
We assume the *local* Gibbs equation ($Tds = du - ...$) is valid.
The total entropy $S$ is the integral of the local entropy density: $S(t) = \int_V s_v(\mathbf{r},t) dV$.
Criteria for NESS (a special case): If the system is in the linear regime and has fixed boundary conditions, the stable NESS is characterized by the minimum rate of entropy production ($\frac{d_iS}{dt} = \text{minimum}$) consistent with the constraints.
13. Soret and Dufour Effects
These are the two coupled phenomena describing the link between heat flow and matter flow, as introduced in Q7.
Where $\mathbf{X}_d$ is the diffusion force (from $\nabla c$) and $\mathbf{X}_q$ is the heat force (from $\nabla T$). Onsager's relation states $L_{dq} = L_{qd}$.
Soret Effect (Thermodiffusion)
What it is: A temperature gradient causes matter flow (diffusion).
The Process:
Start with a uniform mixture (e.g., salt in water). This means the diffusion force is zero: $\mathbf{X}_d = 0$.
Apply a temperature gradient (make one side hot, one cold). This means $\mathbf{X}_q \neq 0$.
Look at the first equation:
$$ \mathbf{J}_d = L_{dd}(0) + L_{dq} \mathbf{X}_q \implies \mathbf{J}_d = L_{dq} \mathbf{X}_q $$
This shows that the temperature gradient *causes* a diffusion flux $\mathbf{J}_d$. The salt (or one component) will begin to migrate, say, to the cold side, and the water to the hot side.
This migration *creates* a concentration gradient ($\mathbf{X}_d \neq 0$), which in turn drives a normal diffusion flux in the *opposite* direction.
A NESS is reached when the two fluxes cancel out, and the net matter flux is zero ($\mathbf{J}_d = 0$). At this point:
$$ 0 = L_{dd} \mathbf{X}_d + L_{dq} \mathbf{X}_q \implies \frac{\mathbf{X}_d}{\mathbf{X}_q} = - \frac{L_{dq}}{L_{dd}} $$
The final, stable concentration gradient is proportional to the applied temperature gradient. The ratio $S_T = - L_{dq} / (L_{dd} T)$ is the Soret coefficient.
Dufour Effect (Diffusion-Thermo)
What it is: A concentration gradient causes heat flow.
The Process:
Start with an isothermal system ($\mathbf{X}_q = 0$).
Create a concentration gradient (e.g., let two different gases diffuse into each other). This means $\mathbf{X}_d \neq 0$.
Look at the second equation:
$$ \mathbf{J}_q = L_{qd} \mathbf{X}_d + L_{qq}(0) \implies \mathbf{J}_q = L_{qd} \mathbf{X}_d $$
This shows that the diffusion process *causes* a heat flux $\mathbf{J}_q$. This flow of heat will create a small, transient temperature gradient in the system.
This effect is the exact reciprocal of the Soret effect (and $L_{qd} = L_{dq}$). It's much harder to measure and is typically only significant in gases.
14. Entropy Production in Multiple Chemical Reactions
This is a direct extension of Q6. Now, we have $R$ different reactions happening at the same time.
Each reaction $j$ (where $j = 1, ..., R$) has its own Affinity $A_j$ and its own Reaction Rate $J_{r,j}$.
$A_j = - \sum_k \nu_{kj} \mu_k$ (where $\nu_{kj}$ is the stoich. coeff. of species $k$ in reaction $j$).
$J_{r,j} = d\xi_j / dt$ (where $\xi_j$ is the extent of reaction $j$).
The total change in moles of species $k$, $dn_k$, is the sum of its changes from *all* reactions:
$$ dn_k = \sum_{j=1}^{R} \nu_{kj} d\xi_j $$
The total change in Gibbs free energy (at const $T, P$) is $dG = \sum_k \mu_k dn_k$.
Swap the order of summation: $dG = \sum_j \left( \sum_k \mu_k \nu_{kj} \right) d\xi_j$.
Recognize the inner sum: $\sum_k \mu_k \nu_{kj} = -A_j$.
This gives $dG = \sum_j (-A_j) d\xi_j$.
From Q6, we know $\frac{d_iS}{dt} = - \frac{1}{T} \frac{dG}{dt}$ (at const $T, P$).
Take the time derivative of $dG$: $\frac{dG}{dt} = \sum_j (-A_j) \frac{d\xi_j}{dt} = \sum_j (-A_j J_{r,j})$.
Finally, substitute this into the entropy production equation:
$$ \frac{d_iS}{dt} = - \frac{1}{T} \left( \sum_j -A_j J_{r,j} \right) = \sum_{j=1}^{R} \frac{A_j J_{r,j}}{T} $$
This is in our canonical form $\sum J_k X_k$.
Fluxes: The $R$ reaction rates, $J_{r,j}$.
Forces: The $R$ conjugate affinities, $X_{r,j} = A_j / T$.
The phenomenological laws would be: $J_{r,i} = \sum_j L_{ij} (A_j/T)$. The $L_{ij}$ terms ($i \neq j$) represent chemical coupling, where the affinity of one reaction can drive the rate of another.
15. The Phenomenological Law and Methodology of NET (Summary)
This question asks for the overall "game plan" of NET. Here it is, step by step.
Tutor's Note: ⚡ Think of this as your "lab manual" for analyzing any non-equilibrium process.
Step 1: Make the Fundamental Assumption (Local Equilibrium).
Postulate that your non-equilibrium system can be treated as a collection of tiny equilibrium cells. This allows you to use the standard Gibbs equation ($Tds = du - ...$) in a local form.
Step 2: Find the Entropy Production ($\sigma$).
This is the main mathematical step. Combine the time-derivative of the local Gibbs equation with the conservation laws (for mass, energy, etc.). After rearranging, you will isolate an equation for the entropy production rate, $\sigma$. This derivation will *always* result in the form:
$$ \sigma = \sum_k J_k X_k $$
Step 3: Identify the Conjugate Forces ($X_k$) and Fluxes ($J_k$).
The previous step *gives* you the correct pairs. You don't guess. The derivation rigorously identifies the flux (e.g., $\mathbf{J}_q$) and its one-and-only true conjugate force (e.g., $\nabla(1/T)$).
Step 4: Write the Linear Laws (The Phenomenological Law).
Now, you *assume* the system is "close" to equilibrium (the linear regime). The phenomenological law is the generalized linear equation: "any flux is a linear combination of all forces."
$$ J_i = \sum_k L_{ik} X_k $$
Step 5: Apply Symmetry (Onsager's Relation).
Invoke Onsager's reciprocal relation, $L_{ik} = L_{ki}$, which is based on microscopic reversibility. This simplifies your set of equations and reveals hidden connections between coupled processes.
Step 6: Analyze the System.
With this complete, predictive model ($J=LX$ with a symmetric $L$), you can now describe real-world phenomena:
Coupled Processes: Analyze effects like thermoelectricity (Q9) or thermodiffusion (Q13), where $L_{ik} \neq 0$.
Stationary States (NESS): Find the stable operating point of an open system by applying Prigogine's Principle (Q11) to find the state of minimum entropy production.
Dissipation: Calculate the rate of energy dissipation (Joule heating, friction, etc.) using $T\sigma$.
Time Remaining: 900s
Enter Password
Enter Your Details
90:00
0% Complete
Your Results
Detailed Report
Fullscreen Required!
Please stay in fullscreen mode to continue the quiz.