Ginzburg Criterion: Validating Mean-Field Theory

Introduction

In our previous lesson, we extended Landau's theory to include spatial variations in the order parameter, constructing the Landau-Ginzburg functional. By analyzing the quadratic fluctuations around the mean-field solution, we derived the Ornstein-Zernike form of the correlation function and calculated the correlation length, $ξ$ . This revealed that within the mean-field approximation, $ξ$ diverges at the critical point as $∣ T - T_{c} ∣^{- 1/2}$ , defining the mean-field critical exponent $ν = 1/2$ . However, this analysis was based on a Gaussian approximation, and as we've seen, experimental exponents often differ from these mean-field values.

This lesson addresses the crucial question that arises from this discrepancy: When is the mean-field theory valid? We will investigate the limits of this approximation by quantifying the impact of the very fluctuations we've been studying. Our objective is to derive the celebrated Ginzburg criterion, a condition that determines the temperature range where fluctuations become so large that they overwhelm the mean-field behavior and render the theory invalid. This criterion will not only delineate the domain of mean-field theory but also formally establish the concept of the upper critical dimension, above which mean-field predictions become exact.

Fluctuation Corrections and the Breakdown of Mean-Field Theory

The mean-field solution corresponds to the saddle-point of the functional integral for the partition function, where we consider only the single, most probable field configuration. The Gaussian approximation, which we used to find the correlation length, is the first correction beyond this simple picture. To test the consistency of the mean-field approach, we must now evaluate the contribution of these Gaussian fluctuations to a thermodynamic observable and compare it to the mean-field prediction. The specific heat is an excellent candidate for this comparison.

Recall that the mean-field specific heat exhibits a characteristic finite jump at the critical temperature. Let's calculate the correction to this result coming from the fluctuations. For simplicity, we'll work with a single-component order parameter ( $n = 1$ ). The total free energy density is the sum of the mean-field (saddle-point) part and a fluctuation contribution:

f = f_{M F} + f_{f l u c}

The fluctuation part arises from summing over all the quadratic fluctuation modes:

f_{f l u c} = \frac{k _{B} T}{2 V} q \sum ln [K (q^{2} + ξ^{- 2})] \to \frac{k _{B} T}{2} \int \frac{d ^{d} q}{( 2 π ) ^{d}} ln [K (q^{2} + ξ^{- 2})]

The singular part of the specific heat per unit volume is $C_{s in g} \propto - T \frac{\partial ^{2} f}{\partial T ^{2}}$ . Let's analyze the contributions from both parts near $T_{c}$ .

Mean-Field Contribution: For $T < T_{c}$ , the mean-field free energy density is $f_{M F} \approx - t^{2} / (4 u)$ , where $t \propto (T - T_{c})$ . The specific heat is therefore a constant, leading to the familiar jump $Δ C_{M F} \propto 1/ u$ .
Fluctuation Contribution: The singular behavior of the fluctuation correction, $δ C$ , comes from the temperature dependence of the correlation length $ξ$ . The calculation involves taking two derivatives of $f_{f l u c}$ with respect to temperature.

A detailed derivation is presented in the following resource, which builds directly upon our previous discussions.

[PDF] Chapter 2 Ginzburg-Landau Phenomenology

This text by Ben Simons provides a direct and rigorous derivation of the Ginzburg criterion by calculating and comparing the fluctuation correction to the specific heat with the mean-field result.

First, read Section 2.7, "Fluctuation Corrections to the Saddle-Point". This section starts by writing down the full free energy, including the fluctuation integral. Follow the derivation of the singular part of the specific heat in Eq. (2.17), which leads to the integral for the fluctuation correction, \delta C. The crucial step is the analysis of this integral. Pay close attention to how its behavior depends on the dimensionality d. This analysis introduces the concept of the upper critical dimension. Next, read Section 2.8, "Ginzburg Criterion". This section uses the result for \delta C to establish the criterion for the validity of mean-field theory by comparing the fluctuation and mean-field contributions. The final inequality is the Ginzburg criterion.

The Upper Critical Dimension and the Ginzburg Criterion

As the reading demonstrates, the analysis of the fluctuation integral for the specific heat, $δ C$ , reveals a profound dependence on spatial dimension $d$ :

δ C \propto \int \frac{d ^{d} q}{( q ^{2} + ξ ^{- 2} ) ^{2}} \propto ξ^{4 - d}

Since the mean-field correlation length diverges as $ξ \propto ∣ t ∣^{- 1/2}$ , where $t = (T - T_{c}) / T_{c}$ , the fluctuation correction behaves as:

δ C \propto ∣ t ∣^{- (4 - d) /2}

This leads to three distinct physical regimes:

For $d > 4$ : The exponent is negative, so $δ C \to 0$ as $t \to 0$ . The fluctuation correction is negligible compared to the finite mean-field jump $Δ C_{M F}$ . In high dimensions, fluctuations are irrelevant near the critical point, and mean-field theory provides a qualitatively correct description.
For $d < 4$ : The exponent is positive, causing $δ C$ to diverge as $t \to 0$ . This divergence completely overwhelms the finite mean-field jump, signaling a catastrophic failure of the mean-field approximation. The theory is internally inconsistent in this regime.
For $d = 4$ : This is the marginal case where the integral yields a logarithmic divergence. This dimension, where the nature of fluctuations fundamentally changes, is known as the upper critical dimension, $d_{c} = 4$ .

The Ginzburg criterion formalizes this observation. Mean-field theory is self-consistent only when the fluctuation correction is small compared to the mean-field value itself:

δ C ≪ Δ C_{M F}

Using our results $δ C \propto ∣ t ∣^{- (4 - d) /2}$ and $Δ C_{M F} \propto 1/ u$ , and reintroducing the microscopic parameters, we arrive at the criterion. Mean-field theory breaks down when the reduced temperature $∣ t ∣$ becomes smaller than a characteristic value, the Ginzburg temperature $t_{G}$ :

∣ t ∣ ≪ t_{G} \approx (\frac{u a ^{d}}{k _{B} T _{c} K ^{d /2}})^{2/ (4 - d)} \propto (\frac{a}{ξ _{0}})^{2 d / (4 - d)}

Here, $a$ is a microscopic length scale (like the lattice spacing) and $ξ_{0} = K / a_{0}$ is the "bare" correlation length away from $T_{c}$ . The Ginzburg criterion defines a critical region, $∣ t ∣ < t_{G}$ , where fluctuations dominate and mean-field theory fails.

A Physical Interpretation

The Ginzburg criterion can also be understood through a more direct physical argument, often called the Levanyuk-Ginzburg argument. Mean-field theory describes the system by a single value, the average order parameter $m ˉ$ . This can only be a good approximation if the thermal fluctuations of the order parameter are small compared to its average value. The most natural way to quantify this is to compare the average fluctuation within a correlation volume, $ξ^{d}$ , to the mean-field value squared.

The criterion is:

\frac{⟨( δ m ) ^{2} ⟩ _{in a volume ξ^{d}}}{m ˉ ^{2}} ≪ 1

The following video provides an excellent derivation based on this physical picture.

Landau Ginzburg theory of Phase Transitions

This video by Prof. Jos Thijssen provides an alternative and very physical derivation of the Ginzburg criterion, complementing the specific heat approach.

Watch the final part of the video, which is dedicated entirely to this topic. The derivation starts at this summary where he formulates the criterion as the condition that the average squared fluctuation, \langle (\delta m)^2 \rangle, must be much smaller than the mean-field order parameter squared, m_0^2. He then proceeds with a beautiful scaling argument. He relates the fluctuation of the local order parameter to the fluctuation of the global order parameter by considering the system as a collection of independently fluctuating patches of volume \xi^d. This argument runs from this point. Follow how he derives the scaling of the local fluctuation \langle \delta m(\mathbf{r})^2 \rangle \propto \xi^{2-d} \propto |t|^{-(2-d)/2}. Finally, from the final comparison, he compares this fluctuation to the mean-field order parameter \bar{m}^2 \propto |t| to arrive at the same condition on the dimensionality, d > 4.

This argument beautifully illustrates the physical content of the criterion. The average local fluctuation scales as $∣ t ∣^{(d - 2) /2}$ , while the order parameter itself vanishes as $∣ t ∣^{1/2}$ . The ratio of the fluctuation to the mean value squared therefore scales as $∣ t ∣^{(d - 4) /2}$ . This ratio becomes large as $t \to 0$ for any dimension $d < 4$ , signifying the inevitable breakdown of the mean-field approximation sufficiently close to the critical point.

The size of the Ginzburg temperature $t_{G}$ explains why mean-field theory works remarkably well for some systems, like conventional superconductors, but fails for others, like liquid-gas transitions or ferromagnets. For superconductors, the bare correlation length $ξ_{0}$ (the Cooper pair size) is very large ( $\sim 1000$ Å). This makes $t_{G}$ astronomically small ( $\sim 1 0^{- 16}$ ), meaning the critical region is experimentally inaccessible. For most other systems, $ξ_{0}$ is of atomic dimensions, making $t_{G}$ large enough ( $\sim 1 0^{- 1} - 1 0^{- 2}$ ) that experiments can easily probe the non-mean-field critical region.

Conclusion

This lesson marks the culmination of our study of Landau-Ginzburg theory and its limitations. By analyzing the role of fluctuations, we have established a clear criterion for the validity of the mean-field approximation that has been the cornerstone of our analysis so far.

Key Takeaways:

The validity of mean-field theory can be tested by comparing the contribution of thermal fluctuations to a thermodynamic quantity (like specific heat) with its mean-field value.
This comparison reveals the existence of an upper critical dimension, $d_{c} = 4$ , above which fluctuations are irrelevant near $T_{c}$ and mean-field exponents are exact. Below $d_{c}$ , fluctuations dominate and alter the critical behavior.
The Ginzburg criterion defines a temperature range close to $T_{c}$ , $∣ t ∣ < t_{G}$ , where mean-field theory necessarily fails for $d < d_{c}$ .
The size of the Ginzburg interval $t_{G}$ depends on microscopic parameters, primarily the ratio of the bare correlation length $ξ_{0}$ to the lattice spacing. This explains why mean-field theory provides an excellent description for some systems (like superconductors with large $ξ_{0}$ ) but not for others.

The breakdown of Landau theory inside the critical region is not a dead end. Instead, it serves as the crucial motivation for a more powerful theoretical framework designed specifically to handle the physics of strong fluctuations and scale invariance at the critical point: the Renormalization Group. You have now reached the frontier of this course and are well-prepared to explore that fascinating topic.

Recommended resources

Introduction

Fluctuation Corrections and the Breakdown of Mean-Field Theory

The Upper Critical Dimension and the Ginzburg Criterion

A Physical Interpretation

Conclusion