I Calculating the specific heat capacity for the 2D Ising model

julian
Science Advisor
Gold Member
Messages
857
Reaction score
361
TL;DR Summary
I'm calculating the specific heat capacity. I nearly get the right answer. Is there a typo in the book?
So I'm looking at the book "Equilibrium Statistical physics" by Plischke and Bergersen. I'm doing the calculation of the specific heat of the 2D Ising model. I can't seen to quite get out the same expression as in the book - there are a coupe of minus signs that are different. I don't know if I have made a mistake or if the book has a typo. So here's the calculations that I did:

We have the identity

\begin{eqnarray}
\frac{d K_1 (q)}{d q} = \dfrac{E_1 (q)}{q (1 - q^2)} - \frac{K_1 (q)}{q} .
\nonumber
\end{eqnarray}

We now substitute in the explicit expression for q into this. The expression for q is

$$
q (K) = \dfrac{2 \sinh 2 K}{\cosh^2 2 K}
$$

implying

$$
q (1 - q^2) = \dfrac{2 \sinh 2 K}{\cosh^2 2 K} \left( 1 - \dfrac{4 \sinh^2 2 K}{\cosh^4 2 K} \right) = \dfrac{2 \sinh 2 K (1 - \sinh^2 2 K)^2}{\cosh^6 2 K}
$$

so

\begin{eqnarray}
\frac{d K_1 (q)}{d q} &=& \dfrac{E_1 (q)}{q (1 - q^2)} - \frac{K_1 (q)}{q}
\nonumber \\
&=& \dfrac{\cosh^6 2 K}{2 \sinh 2 K (1 - \sinh^2 2 K)^2} E_1 (q) - \dfrac{\cosh^2 2 K}{2 \sinh 2 K} K_1 (q)
\nonumber \\
\end{eqnarray}

We then have

\begin{eqnarray}
\frac{d K_1(q)}{d K} &=& \frac{d q}{d K} \frac{d K_1(q)}{d q}
\nonumber \\
&=& \dfrac{4 (1 - \sinh^2 2 K)}{\cosh^3 2 K} \left[ \dfrac{\cosh^6 2 K}{2 \sinh 2 K (1 - \sinh^2 2 K)^2} E_1 (q) - \dfrac{\cosh^2 2 K}{2 \sinh 2 K} K_1 (q) \right]
\nonumber \\
&=& \dfrac{2 \cosh^3 2 K}{\sinh 2 K (1 - \sinh^2 2 K)} E_1 (q) - \dfrac{2 (2 - \cosh^2 2 K)}{\cosh 2 K \sinh 2 K} K_1 (q)
\nonumber \\
&=& \dfrac{2 \coth 2 K \cosh^2 2 K}{1 - \sinh^2 2 K} E_1 (q) + 4 \coth 2 K (\tanh^2 2 K - 1) K_1 (q) + 2 \coth 2 K K_1 (q)
\nonumber \\
&=& \coth 2 K \left[ \dfrac{2 \cosh^2 2 K}{1 - \sinh^2 2 K} E_1 (q) + 2 (2 \tanh^2 2 K - 1) K_1 (q) \right]
\nonumber \\
\end{eqnarray}

The specific heat is given by

\begin{eqnarray}
\frac{1}{k_B} c (T) &=& \beta^2 \dfrac{\partial^2 \beta g(T)}{\partial \beta^2}
\nonumber \\
&=& \beta^2 \dfrac{\partial u(T)}{\partial \beta}
\nonumber
\end{eqnarray}

where

##
u (T) = - J \coth 2 K \left[ 1 + \dfrac{2}{\pi} (2 \tanh^2 2 K - 1) K_1 (q) \right]
##

Using this in the above expression for ##\frac{1}{k_B} c(T)##

\begin{eqnarray}
&\;& \frac{1}{k_B} c (T) = \beta^2 \frac{\partial}{\partial \beta} u (T)
\nonumber \\
&=& \beta^2 J \frac{d}{d K} \left( - J \coth 2 K \left[ 1 + \frac{2}{\pi} (2 \tanh^2 2 K - 1) K_1 (q) \right] \right)
\nonumber \\
&=& - \beta^2 J^2 \frac{d}{d K} \coth 2 K \left[ 1 + \frac{2}{\pi} (2 \tanh^2 2 K - 1) K_1 (q) \right]
\nonumber \\
&\;& - \beta^2 J^2 \coth 2 K \left[ \frac{8}{\pi} K_1 (q) \tanh 2 K \frac{d}{d K} \tanh 2 K \right]
- \beta^2 J^2 \coth 2 K \left[ \frac{2}{\pi} (2 \tanh^2 2 K - 1) \frac{d K_1 (q)}{d K} \right]
\nonumber \\
&=& - 2 \beta^2 J^2 (\coth^2 2 K - 1) \left[ 1 + \frac{2}{\pi} (2 \tanh^2 2 K - 1) K_1 (q) \right]
\nonumber \\
&\;& - \beta^2 J ^2 \coth 2 K \left[ \frac{16}{\pi} \tanh 2 K (1 - \tanh^2 2 K) K_1 (q) \right]
\nonumber \\
&\;& - \beta^2 J^2 \coth 2 K \left[ \frac{2}{\pi} (2 \tanh^2 2 K - 1) \coth 2 K \left( \dfrac{2 \cosh^2 2 K}{1 - \sinh^2 2 K} E_1 (q) + 2 (2 \tanh^2 2 K - 1) K_1 (q) \right) \right]
\nonumber \\
&=& - 2 \beta^2 J^2 (\coth^2 2 K - 1)
\nonumber \\
&\;& - \frac{4}{\pi} \beta^2 J^2 (\coth^2 2 K - 1) (2 \tanh^2 2 K - 1) K_1 (q)
\nonumber \\
&\;& - \frac{4}{\pi} (\beta J \coth 2 K)^2 (1 - \tanh^2 2 K) 4 \tanh^2 2 K K_1 (q)
- \frac{4}{\pi} (\beta J \coth 2 K)^2 (2 \tanh^2 2 K - 1)^2 K_1 (q)
\nonumber \\
&\;& - \frac{4}{\pi} (\beta J \coth 2 K)^2 \dfrac{(2 \tanh^2 2 K - 1) \cosh^2 2 K}{1 - \sinh^2 2 K} E_1 (q)
\nonumber \\
&=& - \frac{4}{\pi} (K \coth 2 K)^2 (1 - \tanh^2 2 K) \frac{\pi}{2}
\nonumber \\
&\;& - \frac{4}{\pi} (K \coth 2 K)^2 (1 - \tanh^2 2 K) (2 \tanh^2 2 K - 1) K_1 (q)
\nonumber \\
&\;& - \frac{4}{\pi} (K \cosh 2 K)^2 K_1 (q) + \frac{4}{\pi} (K \cosh 2 K)^2 E_1 (q)
\nonumber \\
&=& \frac{4}{\pi} (K \cosh^2 2 K)^2 \left\{ - K_1 (q) + E_1 (q) - (1 - \tanh^2 2 K) \left[ \frac{\pi}{2} + (2 \tanh^2 2 K - 1) K_1 (q) \right] \right\}
\nonumber
\end{eqnarray}

But the book says the answer is:

##
\frac{4}{\pi} (K \cosh^2 2 K)^2 \left\{ K_1 (q) - E_1 (q) - (1 - \tanh^2 2 K) \left[ \frac{\pi}{2} + (2 \tanh^2 2 K - 1) K_1 (q) \right] \right\}
##

Have I made a mistake or is there a typo in the book? (I think I have already found a couple of typos in the book).
 
Last edited:
Physics news on Phys.org
Have I written down the wrong formula for the specific heat? I think it should be:

##
\dfrac{1}{k_B} c (T) = \dfrac{1}{k_B} \dfrac{\partial }{\partial T} u (T) = \dfrac{1}{k_B} \dfrac{\partial \beta}{\partial T} \dfrac{\partial }{\partial \beta} u (T) = - \dfrac{1}{k_B} \dfrac{1}{k_B T^2} \dfrac{\partial }{\partial \beta} u (T) = - \beta^2 \dfrac{\partial }{\partial \beta} u (T) .
##

So that my expression becomes:

\begin{align*}
\dfrac{1}{k_B} c (T) = \frac{4}{\pi} (K \cosh^2 2 K)^2 \left\{ K_1 (q) - E_1 (q) + (1 - \tanh^2 2 K) \left[ \frac{\pi}{2} + (2 \tanh^2 2 K - 1) K_1 (q) \right] \right\}
\end{align*}

So there is still a different sign from the book's expression, but in front of the ##(1 - \tanh^2 2 K)## instead!

I think you need a positive sign in front of ##K_1 (q)## to get a positive value for the specific heat (calculated near the critical temperature, ##T_c##, defined by ##\sinh \dfrac{2J}{k_B T_c} = 1##) :

The Elliptic function of the first kind has the asymptotic behaviour:

##
K_1 (q) \sim - \frac{1}{2} \ln | 1 - q | \quad \text{as } q \rightarrow 1^-
##

Say ##C## is defined by ##\sinh C = 1##, then we have the expansions:

\begin{align*}
\sinh (C + x) &= \sinh C+ x \cosh C + \dfrac{x^2}{2!} \sinh C + \cdots = 1 + \sqrt{2} x + \dfrac{x^2}{2} + \cdots
\nonumber \\
\cosh (C + x) &= \cosh C + x \sinh C + \dfrac{x^2}{2!} \cosh C + \cdots = \sqrt{2} + x + \dfrac{1}{\sqrt{2}} x^2 + \cdots
\end{align*}

so that

\begin{align*}
q (C + x) &= \dfrac{2 \sinh (C+ x)}{\cosh^2 (C + x)}
\nonumber \\
&= \dfrac{2 (1 + \sqrt{2} x + x^2 + \cdots )}{(\sqrt{2} + x + \dfrac{1}{\sqrt{2}} x^2 + \cdots)^2} = \dfrac{1 + \sqrt{2} x + \cdots }{1 + \sqrt{2} x + \cdots}
\nonumber \\
&= (1 + \sqrt{2} x + \cdots ) (1 - \sqrt{2} x + \cdots )
\nonumber \\
&= 1 - 2 x^2 + \cdots
\end{align*}

Using this we approximate ##q (K)## for temperatures, ##T##, near to the critical temperature, ##T_c##. We have the definition ##2 K = \dfrac{2J}{k_B T}##, so that:

##
q (K) = q \left( \dfrac{2 J}{k_B T_c} \dfrac{1}{1 + \dfrac{T - T_c}{T_c}} \right) \approx q \left ( \dfrac{2 J}{k_B T_c} \left( 1 - \dfrac{T - T_c}{T_c} \right) \right) \approx 1 - 2 \left( \dfrac{2 J}{k_B T_c} \right)^2 \left( 1 - \dfrac{T}{T_c} \right)^2
##

where we assumed that ##T > T_c##.

##
q (K) = q \left( \dfrac{2 J}{k_B T_c} \dfrac{1}{1 - \dfrac{T_c - T}{T_c}} \right) \approx q \left ( \dfrac{2 J}{k_B T_c} \left( 1 + \dfrac{T_c - T}{T_c} \right) \right) \approx 1 - 2 \left( \dfrac{2 J}{k_B T_c} \right) \left( 1 - \dfrac{T}{T_c} \right)^2
##

where we assumed that ##T < T_c##. Using this in the asymptotic expression for ##K_1 (q)## we have

##
K_1 (q) \sim - \frac{1}{2} \ln | 1 - q | \sim - \ln \left| 1 - \dfrac{T}{T_c} \right|
##

and we have for the specific heat

##
\dfrac{1}{k_B} c (T) \approx - \dfrac{2}{\pi} \left( \dfrac{2J}{k_B T_C} \right)^2 \ln \left| 1 - \dfrac{T}{T_c} \right| + const
##

Which is the expression given in the book for the asymptotic behaviour of ##\dfrac{1}{k_B} c (T)##. Importantly it is positive.
 
Last edited:
Hi. I have got question as in title. How can idea of instantaneous dipole moment for atoms like, for example hydrogen be consistent with idea of orbitals? At my level of knowledge London dispersion forces are derived taking into account Bohr model of atom. But we know today that this model is not correct. If it would be correct I understand that at each time electron is at some point at radius at some angle and there is dipole moment at this time from nucleus to electron at orbit. But how...
Back
Top