Linear Spectroscopy

The goal of this section is to show that the linear absorption spectrum $ \sigma(\omega)$ of a system is determined by the regression of spontaneous polarization fluctuations at equilibrium as follows:
$\displaystyle \sigma(\omega) = \beta 2 \omega^2 f_0^2 \bar{\varepsilon}^2 \int_{0}^{\infty} <\delta A(0) \delta A(t')>$   cos$\displaystyle (\omega t') dt',$ (450)
where $ A(t)$ is the time-dependent polarization
$\displaystyle A(t) = \sum_j p_j \langle \phi_j(t) \mid \hat{A} \mid \phi_j(t) \rangle,$ (451)
where $ \phi_j(t)$ evolves according to the unperturbed system Hamiltonian $ H_0$ as follows:
$\displaystyle i \hbar \frac{\partial}{\partial t} \phi_j(t) = \hat{H}_0 \phi_j(t).$ (452)


In order to derive Eq. (450), consider that the system is perturbed by a monochromatic electric field,

$\displaystyle f(t)=f_0 \bar{\varepsilon} (e^{-i \omega t} + e^{i \omega t}),$ (453)


where $ f_0$ is the amplitude of the field and $ \bar{\varepsilon}$ is the unit vector along the spatially uniform electric field.

In the linear regime, the interaction between the system and the field is

$\displaystyle \Delta \hat{H}(t) = - f(t) \hat{A}.$ (454)
The total energy of the system is
$\displaystyle E(t) = \sum_j p_j \langle \phi_j(t) \mid \hat{H}_0 + \Delta \hat{H}(t) \mid \phi_j(t) \rangle,$ (455)


and the differential change of energy per unit time is

$\displaystyle \dot{E}(t) = \sum_j p_j \langle \phi_j(t) \mid \frac{\partial \Delta \hat{H}(t)}{\partial t} \mid \phi_j(t) \rangle = - \dot{f(t)} A(t),$ (456)


since according to Eq. (452),

$\displaystyle \langle \frac{\partial \phi_j(t)}{\partial t} \mid \hat{H}_0 + \D......at{H}_0 + \Delta \hat{H}(t) \mid \frac{\partial \phi_j(t)}{\partial t} \rangle.$ (457)


Therefore, the total energy change $ \sigma$ due to the interaction with the external field for time T is

$\displaystyle \sigma = \frac{1}{T} \int_0^T (-\dot{f}(t)) A(t),$ (458)


where, according to Eq. (453),

$\displaystyle \dot{f}(t) = -i f_0 \bar{\varepsilon} \omega [e^{-i \omega t} - e^{+i \omega t}].$ (459)


Substituting Eq. (459) into Eq. (458) we obtain,

$\displaystyle \sigma = - \frac{i \omega}{T} \int_0^T f_0 \bar{\varepsilon} [ e^...... t} - e^{-i \omega t} ] (<A> + \int_{-\infty}^{\infty} \chi (t-t') f(t') dt' ).$ (460)


Note that

$\displaystyle \int_{-\infty}^{\infty} \chi (t-t') f(t') dt' = \int_{-\infty}^{\infty} \chi(-t') f(t'-t) dt' = \int_{-\infty}^{\infty} \chi(t') f(-t'-t) dt',$ (461)


thus Eq. (460) becomes

$\displaystyle \sigma = - \frac{i \omega}{T} \int_0^T dt f_0 \bar{\varepsilon} [......a t} - e^{-i \omega t} ] (<A> + \int_{-\infty}^{\infty} \chi(t') f(-t'-t) dt').$ (462)


In order to simplify Eq. (462), we note that

$\displaystyle \frac{1}{T} \int_0^T e^{int'} dt' = \begin{cases}1, \hspace{.2cm}......hen} \hspace{.2cm} \text{n=0} \\ 0, \hspace{.2cm} \text{otherwise}, \end{cases}$ (463)


therefore, Eq. (462) becomes

$\displaystyle \sigma = - \frac{i \omega}{T} \int_0^T dt f_0 \bar{\varepsilon} [......i(t') f_0 \bar{\varepsilon} [e^{+i \omega (t+t')} + e^{-i \omega (t+t')} ] dt',$ (464)


or

$\displaystyle \sigma = - i \omega f_0^2 \bar{\varepsilon}^2 \int_{-\infty}^{\infty} \chi(t') [e^{+i \omega t'} - e^{-i \omega t'} ] dt'.$ (465)


Therefore,

$\displaystyle \sigma = 2 \omega f_0^2 \bar{\varepsilon}^2 \int_{-\infty}^{\infty} \chi(t')$   sin$\displaystyle (\omega t') dt'.$ (466)


Substituting Eq. (449) into Eq. (466) we obtain

$\displaystyle \sigma = -\beta 2 \omega f_0^2 \bar{\varepsilon}^2 \int_{0}^{\infty} \frac{d}{dt} <\delta A(0) \delta A(t')>$   sin$\displaystyle (\omega t') dt'.$ (467)


Finally, integrating Eq. (467) by parts we obtain Eq. (450), since Eq. (467) can be written as $ \int dt' u(t') dv/dt' = u(t') v(t') - \int dt' v(t') du/dt'$, with $ u(t') =$   sin$ (\omega t')$ and $ v(t')= <\delta A(0) \delta A(t')> $.