Suceptibility

Given a free energy functional, \(\mathcal{F} = \mathcal{F}_0 + \mathcal{F}_f\) compare to Thermodynamics: \(\tilde{\mathcal{F}} = \mathcal{F} - fx\) with a generalized force and displacement \(f,x\). So, \(\mathcal{F}_f = \int d\vec{x}f(\vec{x},t)s(\vec{x},t)\).

Examples:

\(f\) \(S\)
\(H\) \(M\)
\(E\) \(P\)

For \(t'

Fourier transforming this \(t\to\omega,x\to k\). \(\tilde{s}(\vec{k},\omega) = \tilde{\chi}(\vec{k},\omega)\tilde{f}(\vec{k},\omega)\).

\(\chi(\vec{k},\omega) = \int d\vec{x}\int dt\exp(i\omega t)\exp(-i\vec{k}\cdot\vec{x})\chi(\vec{x},t)\) note the signs only are due to us wanting the plane wave to look like \(\exp i(\omega t-kx)\). Then we get, \(\tilde{\chi}(\vec{k},\omega) = \chi'(\vec{k},\omega) + i\chi''(\vec{k},\omega)\). So, \(\tilde{\chi}^* = \tilde{\chi}(-\vec{k},-\omega)\).

A system with inversion symmetry, \(\vec{x}=-\vec{x}\), \(\tilde{\chi}^*(\vec{k},\omega) = \tilde{\chi}(\vec{k},-\omega)\). \(\chi'\) is even in \(\omega\) and \(\chi''\) is odd in \(\omega\).

Then, \(\tilde{\chi}(\vec{k},\omega) = \int d\vec{x}\int dt\exp(i\omega t)\exp(-i\vec{k}\cdot\vec{x})\chi(\vec{x},t)\).

C.f. the AC suceptibility \(\tilde{P}(\vec{k},\omega) = \tilde{\chi}(\vec{k},\omega)\tilde{E}(\vec{k},\omega)\) and \(\tilde{M}(\vec{k},\omega) = \tilde{\chi}(\vec{k},\omega)\tilde{H}(\vec{k},\omega)\).

Obtaining the work, \(W = \int \vec{F}\cdot d\vec{S} = \int \vec{F}\frac{\partial\vec{s}}{\partial t}dt = \int\vec{F}\cdot\vec{v}dt\). Power: \(P = \lim_{T\to\infty}\frac{1}{T}\int_0^Tf(t)\frac{\partial s}{\partial t}dt = \lim_{t\to\infty}\frac{1}{T}\left\{\int_0^T\left(-s(t)\frac{\partial f}{\partial t}dt\right) - s(t)f(t)|_0^T\right\}\), with our generalized force and displacement. The boundary terms go to zero, \(f(t) = \Re\{f\omega\exp(i\omega t)\} = \frac{1}{2}\left(f_\omega\exp(-i\omega t) + f^*_\omega\exp(i\omega t)\right) = \Re f_\omega\cos\omega t + \Im f_\omega\sin\omega t\). \(f(t) = \cos\omega_0 t, f(\omega) = \pi(\delta(\omega-\omega_0) + \delta(\omega+\omega_0))\Rightarrow f(t) = \frac{1}{2\pi}\int f(\omega)\exp(-i\omega t)d\omega\). Then, \(s(\omega) = \tilde{\chi}(\omega)f(\omega)\). \(s(t) = \frac{1}{2\pi}\int s(\omega)\exp(-i\omega t)d\omega = \frac{1}{2}\left(\exp(i-\omega_0 t)\chi(\omega_0) + \exp(i\omega_0 t)\chi(-\omega_0)\right) = \frac{1}{2}\left(\exp(i\omega_0 t)(\chi'(\omega_0) + i\chi''(\omega_0)) + \exp(i\omega_0t)(\chi'(\omega_0)-i\chi''(\omega_0))\right) = \chi'(\omega_0)\cos(\omega_0 t) + \chi''(\omega_0)\sin(\omega_0 t)\). Then the power becomes, \(P = \lim_{T\to\infty}\frac{1}{T}\int_0^Tdt s(t) \frac{1}{2}(i\omega) (f_\omega\exp(-i\omega t) - f_\omega^*\exp(i\omega t))dt\). \(s(t) = \int_{-\infty}^t dt'\chi(t-t') f(t')\). Since there is no response from the future, \(\chi(t-t')= 0\) for \(t \begin{align} p(\omega) &= \int_{-\infty}^\infty d\tau \frac{i\omega\tau}{4}\lim_{T\to\infty}\frac{1}{T}\int_0^Tdt\left(f_\omega^2\exp(-i\omega(2t-\tau)) - f_\omega^{*2}\exp(i\omega(2t-\tau)) + |f_\omega^*|^2\left(\exp(-i\omega\tau) - \exp(i\omega\tau)\right)\right) \\ &= \int_{-\infty}^\infty d\tau \frac{i\omega\tau}{4}\lim_{T\to\infty}\frac{1}{T}(-1)\int_0^Tdt\left(|f_\omega^*|^2\left(\exp(i\omega\tau) - \exp(-i\omega\tau)\right)\right) \\ &= \frac{\omega}{2}|f_\omega|^2\int d\tau \chi(\tau)\sin(\omega\tau) \\ &= \frac{\omega}{2}|f_\omega|^2\Im\{\mathcal{F}(\chi(\tau))\} \\ &= \frac{\omega}{2}|f_\omega|^2\Im\{\tilde{\chi}(\tau)\} \\ &= \frac{\omega}{2}|f_\omega|^2\tilde''{\chi}(\tau) > 0. \end{align}

Then, \(\omega\chi''(\omega)> 0\) which means each one must be greater than zero. Know that \(\chi''(\omega)\) is odd.

Note: \(\chi(k,\omega) = \chi(\omega)\) shorthand was used.

Example: Calculate the current through a wire. Suppose we have \(V(t) = V_\omega\cos(\omega t)\). \(\langle p\rangle = \frac{\langle V^2\rangle}{R} = \frac{V_\omega^2\langle\cos^2\omega t\rangle}{R} = \frac{V_\omega^2}{2R}\). Suppose we have a wire of length \(L\) and cross-section \(A\), conductivity \(\sigma = \frac{1}{\rho}\). \(R = \frac{\rho L}{A} = \frac{L}{\sigma A}\) The force (electric field) due to the voltage is then \(E(t) = E_\omega\cos\omega t = \frac{V_\omega}{L}\cos\omega t\). The power density, power per unit volume, \(p = \frac{\langle p\rangle}{V} = \frac{V_\omega^2}{2R(L\cdot A)} = \frac{V_\omega^2}{2\rho L^2} = \frac{V_\omega^2\sigma}{2L^2} = \frac{1}{2}\sigma E_\omega^2 = \frac{1}{2}\omega |f_\omega|^2\chi''(\omega)\). Thus, the suceptibity is \(chi''(\omega) = \frac{\sigma}{\omega}\) and \(|f_\omega|^2 = E_\omega^2\). \(\alpha''(\omega)\), the polarizability is, \(\alpha'' = \frac{\sigma}{\omega}\). \(p = \lim_{\omega\to0} p(\omega) \Rightarrow \sigma = \lim_{\omega\to 0}\omega\alpha''(\omega)\) gives the DC conductivity on the left and the AC polarizability on the right.

Suppose we have an order parameter \(s_{macro}(\vec{x},t) = \overline{s} - \int d\vec{x}'G(\vec{x}-\vec{x}',t) (s_{macro}(\vec{x},t) - \overline{s})\). Transforming this, \(\tilde{s}_{macro}(\vec{k}',\omega) = \overline{\tilde{s}} - \int d\vec{k}'\tilde{G}(\vec{k}-\vec{k},\omega)(\tilde{s}_{macro}(\vec{k}',\omega)-\overline{\tilde{s}})\)

Compare if we have a Schrodinger equation and a propogator: \(H\varphi = -\frac{i}{\hbar}\frac{\partial}{\partial t}\varphi\Rightarrow \varphi(\vec{x},t) = \int dt'\int d\vec{x}' U(\vec{x}-\vec{x}',t-t')\varphi(\vec{x}-\vec{x}')\).

Static Susceptibility

Removing the time dependence.

\(\chi_0(\vec{r})\). The response to this continuous force (a lean rather than a kick) is then \(s(\vec{x}) = \int d\vec{x}'\chi_0(\vec{x}-\vec{x}')f(\vec{x}')\).

We will get: \(\tilde{\chi}(\vec{k}) = \frac{\partial\langle s_k\rangle_{eq}}{\partial f_k}|_{f=0}\).

\(F_f(t) = -\int d\vec{x}f(\vec{x},t)s(\vec{x},t)\). For a static system, \(F_f = -\vec{x}f(\vec{x})s(\vec{x})\). Then, \(\langle \tilde{s}_k\rangle_{eq} = \frac{\text{Tr}(\tilde{s}_k\exp(-\beta(F_0 - F_f)))}{\text{Tr}(\exp(-\beta(F_0 - F_f)))} = \frac{1}{\beta V}\frac{\partial \ln Z}{\partial f_k}\). \(\tilde{\chi}_0(\vec{k}) = \frac{\partial\langle s_k\rangle_{eq}}{\partial f_k}|_{f=0} = \beta\hat{C}(\vec{k},0)\). The hat here means just a fourier transform for space.

For the uniform case, thermodynamically, \(\chi = \langle(s-\langle s\rangle)^2\rangle\).

Fluctuation-Dissipation Theorem

\(p(\omega) = \frac{\omega|f_\omega|^2}{2}\chi''(\omega)\). Want \(\chi\leftrightarrow C\), connect fluctuations.

Cases: \(f(\vec{x},t) \to \chi(\vec{x},t)\) and \(f(\vec{x})\to\chi(\vec{x})\).

Starting from forced equilibrium at \(t=0\) due to \(f(\vec{x})\). Now we turn off the force. So, \(f(\vec{x},t>0) = 0\) so what is our \(\chi\)?

From our response, \(s(\vec{x},t-t_0) = \int d\vec{x} G(\vec{x}-\vec{x}',t-t_0)s(\vec{x}',t_0)\). For \(t_0 = 0\), \(s(\vec{x},t) = \int d\vec{x} G(\vec{x}-\vec{x}',t)s(\vec{x}',0)\). Then, \(s(x,t) = \int d\vec{x}\int_{-\infty}^t dt'\chi(\vec{x}-\vec{x}',t-t')f(\vec{x},t')\). We choose, \(f(\vec{x},t) = f(\vec{x})\) for \(t<0\) and 0 otherwise. Then, \(s(x,t\geq 0) = \int d\vec{x}\int_{-\infty}^0 dt'\chi(\vec{x}-\vec{x}',t-t')f(\vec{x})\). Let \(\tau=t-t'\), \(dt = -d\tau\), \(t'\in\{-\infty,0\}\) so \(\tau\in\{+\infty, t\}\). Then, \(s(\vec{x},t) = \int dx f(\vec{x})\int_t^\infty d\tau\chi(\vec{x}-\vec{x}',\tau)\) [2b].

For the ideal gas, \(\chi_0(\vec{r}) = \frac{\delta(\vec{r})}{\alpha}\). Then, inserting this, \(s(x,0) = \rho(\vec{x},0) = \frac{f(\vec{x})}{\alpha}\). Which gives, \(\rho(\vec{x},t) = \int d\vec{x}'\frac{f(\vec{x}')}{\alpha}G(\vec{x}-\vec{x}',t)\). \(C^{id} (\vec{x},t) = \frac{G(\vec{x},t)}{\beta\alpha}\). Then, \(\rho(\vec{x},t) = \int d\vec{x}' \frac{f(\vec{x}')}{\alpha}\beta\alpha C^{id}(\vec{x}-\vec{x}',t) = \int d\vec{x}' f(\vec{x}')\beta C^{id}(\vec{x}-\vec{x}',t)\). Comparing this to the last statement in the previous paragraph [2b], \(\beta C^{id}(\vec{x}-\vec{x}',t)\int_t^\infty d\tau\chi(\vec{x}-\vec{x}',\tau)\). Then, \(\beta\frac{\partial C^{id}}{\partial t} = \chi(\vec{x},\infty) - \chi(\vec{x},t) = -\chi(\vec{x},t)\). Which gives us the FD theorem in the time domain, for \(t>0\),

\begin{align} \chi(\vec{x},t) = -\beta\frac{\partial C^{id}(\vec{x},t)}{\partial t} \end{align}

And is zero for \(t<0\).

Note that \(C(\vec{x},t) = C(\vec{x},-t)\). Now we can calculate \(\chi''(\omega)\). \(\overline{\chi}(\vec{r},\omega) = \tilde{\chi}(\omega) = \int_{\pm\infty}dt\chi(t)\exp(i\omega t)\) which is zero for \(t<0\). So, \(\tilde{\chi}(\omega) = \int_0^\infty dt\chi(t)\exp(i\omega t) = -\beta\int_0^\infty \frac{\partial C(t)}{\partial t}\exp(i\omega t)\). Integrating by parts, \(= -\beta\left(\left.C(t)\exp(i\omega t)\right|_0^\infty - \int_0^\infty dt C(t)(i\omega)\exp(i\omega t)\right) = \tilde{\chi}' + i\tilde{\chi}''\). The first term is real, thus, \(\tilde{\chi}'' = \Im \beta\int_0^\infty dtC(t)(i\omega)\exp(i\omega t) = \beta\omega\int_0^\infty dtC(t)\cos(\omega t)\). Since \(C(t) = C(-t)\) and \(\cos(t) = \cos(-t)\), thus the integrand is even. Thus, \(\tilde{\chi}'' = \frac{\beta\omega}{2}\int_{-\infty}^\infty dt C(t)\cos\omega t = \frac{\beta\omega}{2}\int_{-\infty}^\infty dt C(t)\cos\omega t + i\frac{\beta\omega}{2}\int_{-\infty}^\infty dt C(t)\sin\omega t = \frac{\beta\omega}{2}\int_{-\infty}^\infty dt C(t)\exp(i\omega t) = \frac{\omega\beta}{2}\tilde{C}(\omega)\).

Then we get the FD theorem in frequency domain for classical systems,

\begin{align} \tilde{\chi}''(\omega) = \frac{\omega\beta}{2}\tilde{C}(\omega) \end{align}

Then, \(p(\omega) = \frac{\beta\omega^2|f_\omega|^2}{4}\tilde{C}(\omega)\). So in the continuum limit for matter, we get density fluctuations. For solids (crystals), this would be lattice vibrations.

The QM version,

\begin{align} \tilde{\chi}(\vec{k},\omega) = \frac{1}{2\hbar}(1-\exp(-\beta\hbar\omega))\tilde{C}(\vec{k},\omega). \end{align}

Starting point is QM version of linear response, Kudo formula, \(\langle \hat{A}(t)\rangle = \langle \hat{A}\rangle_0 - i\int_{t_0}^t dt'\langle[\hat{A}(t),\hat{V}(t)]\rangle\), \(\langle\cdot\rangle_0\) with respect to equilibrium and \(\langle\hat{A}\rangle = \frac{\text{Tr}[\hat{\rho}\hat{A}]}{\text{Tr}[\hat{\rho}]}\).

Kramers-Kronig Relations

We have, and abbreviate, \(\chi(\vec{r},t) \equiv \chi(t)\), real and \(t>0\). Then, \(\vec{\chi}(\vec{r},\omega) = \chi'(\omega) + i\chi''(\omega)\). Then, \(\tilde{\chi}(-\omega) = -\tilde{\chi}''(\omega)\). 2 real valued functions on half line, \(\omega > 0\).

Let \(\chi(\omega) \to \chi(u+iv) = \int_0^\infty dt\chi(t)\exp(iut)\exp(-vt)\). The zero coming from the causality condition. From Cauchy’s theorem for a function analytic in a region and a contour in the region, \(\oint_C f(z')dz = 0\). Cauchy’s integral formula for a simple pole is \(\oint_c\frac{f(z')}{z'-z}dz' = 2\pi i f(z)\).

Then, \(\oint_c \frac{\chi(\omega')}{\omega'-\omega}d\omega' = 0\) by making a loop that goes near \(\omega\) but cuts it out with a small half circle. Taking the off-axis loop to infinity, we get zero. The integral on the axis, \(\lim\limits_{\varepsilon\to0} = \left(\int_{-\infty}^{\omega-\varepsilon} + \int_{\omega+\varepsilon}^\infty\right)\frac{\tilde{\chi}(\omega')d\omega'}{\omega' - \omega} = PV\int \frac{\chi'(\omega')}{\omega' - \omega}d\omega\) The semicircle integral, \(\int\frac{\tilde{\chi}(\omega')}{\omega'-\omega}d\omega' = \int\frac{\tilde{\chi}(\omega)}{\omega'-\omega}d\omega' = \chi(\omega)\ln(\omega-\omega)|_{\omega+\epsilon\exp(i\pi)}^{\omega+\epsilon\exp(i0)} = -i\pi\tilde{\chi}(\omega)\). Then, the principle value of the integral is \(i\pi\tilde{\chi}(\omega)\). Hence, \(\tilde{\chi}(\omega) = \frac{1}{\pi i}\int_{-\infty}^\infty \frac{\tilde{\chi}(\omega)}{\omega'-\omega}d\omega'\). So, \(\tilde{\chi}'(\omega) = \Re\{\tilde{\chi}(\omega)\} = \frac{1}{\pi}\int_{-\infty}^\infty \frac{\chi''(\omega')}{\omega'-\omega}d\omega'\).

Since \(\tilde{\chi}(\omega) = -\tilde{\chi}^*(-\omega)\), \(\tilde{\chi}''(\omega) = -\tilde{\chi}''^*(-\omega)\). So, \(\tilde{\chi}'(\omega) = \frac{1}{\pi}\int_0^\infty\chi''(\omega) \left(\frac{1}{\omega'-\omega} - \frac{1}{-\omega'-\omega}\right) = \frac{2}{\pi}\int_0^\infty\frac{\chi''(\omega)\omega'}{\omega'^2-\omega^2}d\omega'\).

Alternatively, \(\tilde{\chi}''(\omega) = \frac{-2\omega}{\pi}\int_0^\infty \frac{\chi'(\omega)}{\omega'^2-\omega^2}d\omega'\).

Author: Christian Cunningham

Created: 2024-05-30 Thu 21:17

Validate