Website by fretchen

Lecture 10 - Propagation of light in dielectric media

In this lecture we will study the propagation of light through a dielectric medium like atomic gases. We will see that it is characterized by the susceptibility and discuss the case of two-level atoms. This sets the stage for the laser.

Until now we focused on the properties of atoms and how can control them through external fields. In this lecture, we will focus much more on the properties of the light passing through a medium.

Introduction

We would like to study the propagation of a electric field through an ensemble of atoms as visualized in Fig. 1. We assume a mono-chromatic plane wave to come in, such that we can write down the electric field as:

Ein=E0ϵeikziωLt\vec{E}_{in}= E_0 \vec{\epsilon}e^{i kz -i\omega_L t}

This incoming field will polarize the gas of dipoles.

Propagation of a light field through a dielectric medium.

For the propagation we will do the following assumptions:

  • The atoms are independent.
  • We can describe them as small dipoles.
  • We can describe the light in the semi-classical approximation.

We have already employed this picture in in the slightly abstract formulation in Lecture 4, where we studied the evolution of the atoms in electric fields and in Lecture 6 concerning the transition rules in hydrogen. This allows us to calculate the expectation value of the dipole operator through:

D=ψDψ\langle \vec{D}\rangle = \left\langle\psi\right|\vec{D}\left|\psi\right\rangle

As already discussed in Lecture 6 we can then write it down as:

D=E0α\langle \vec{D}\rangle = E_0 \vec{\alpha}

We call α\alpha the polarizability. For a large gas with a constant density of dipoles nn, we obtain a macroscopic polarization of:

P=nD=nαE0\vec{P} = n \langle \vec{D}\rangle\\ = n \vec{\alpha} E_0

This leads us then to identify the susceptibility of the dielectric medium:

P=ϵ0χEχ=nαϵ0 \vec{P} = \epsilon_0 \chi \vec{E}\\ \chi = \frac{n \alpha}{\epsilon_0}

To notes to this relation:

  1. The linear relationship between polarization and electric field is only valid for weak electric fields. For stronger fields, higher order terms become important. They are the fundamental ingredient of non-linear optics. In general, we can write:
Pi=ϵ0jχij1Ej+ϵ0jkχijk2EjEk+... P_i = \epsilon_0 \sum_{j}\chi_{ij}^{1}E_j+\epsilon_0 \sum_{jk}\chi_{ijk}^{2}E_jE_k + ...
  1. Given that χ\chi and α\alpha are proportional to D\langle D \rangle, they can be complex. We will see that real and imaginary part have very different interpretations.

Propagation of light

At this stage we would like to understand the propagation of an electric field through such a polarized medium. The general Maxwell equation actually reads:

2E1c22Et2=1ϵ0c22Pt2 \nabla^2 \vec{E}-\frac{1}{c^2}\frac{\partial^2 \vec{E}}{\partial t^2}= \frac{1}{\epsilon_0 c^2}\frac{\partial^2 \vec{P}}{\partial t^2}

This equation can be massively simplified by only looking at a slowly-evolving envelope E(r,t)\mathcal{E}(r,t) and P(r,t)\mathcal{P}(r,t), which are defined through:

E=EeikziωLtP=PeikziωLt\vec{E} = \mathcal{E} e^{ikz-i \omega_L t}\\ \vec{P} = \mathcal{P} e^{ikz-i \omega_L t}\\

As shown in more detail in Chapter 4 of Lukin, the Maxwell equation reduces then to:

zE+1ctE=ik2ϵ0P\frac{\partial}{\partial z}\mathcal{E}+\frac{1}{c}\frac{\partial}{\partial t}\mathcal{E} = \frac{ik}{2\epsilon_0}\mathcal{P}

This equation becomes especially transparent, if we investigate it for very long times, such that we can perform a Fourier transformation and obtain:

zE=iωcE+ik2ϵ0P\frac{\partial}{\partial z}\mathcal{E}= i\frac{\omega}{c}\mathcal{E} +\frac{ik}{2\epsilon_0}\mathcal{P}

Finally, we can use the definition of the susceptibility to write:

zE=i(ωc+k2χ(ω))EE(ω,z)=E0ei(ωc+k2χ(ω))z\frac{\partial}{\partial z}\mathcal{E}= i\left(\frac{\omega}{c} +\frac{k}{2} \chi(\omega)\right) \mathcal{E}\\ \mathcal{E}(\omega, z) =E_0 e^{i\left(\frac{\omega}{c} +\frac{k}{2}\chi(\omega)\right)z}

Absorption and refraction

The meaning of the susceptibility becomes especially clear for a continuous wave, where ω0\omega\rightarrow 0 and we obtain:

E(ω0,z)=E0eikχ(0)2z\mathcal{E}(\omega\rightarrow 0, z) =E_0 e^{i\frac{k\chi(0)}{2} z}

We can then see that:

  • The imaginary part of the susceptibility leads to absorption on a scale l1=k2Im(χ(0))l^{-1} = \frac{k}{2}\text{Im}(\chi(0))

  • The real part describes a phase shift. The evolution of the electric field can be seen as propagating with a wavevector kk+k2Re(χ(0))k \rightarrow k +\frac{k}{2}\text{Re}(\chi(0)), so the dielectric medium has a refractive index n=1+Re(χ(0))2n = 1 + \frac{\text{Re}(\chi(0))}{2}

Dispersion

If the electric field has a certain frequency distribution, we might have to perform the proper integral to obtain the time evolution, i.e.:

E(t,z)=dωeiωtE(ω,0)ei(ωc+k2χ(ω))z\mathcal{E}(t, z) =\int d\omega e^{-i\omega t}\mathcal{E}(\omega,0) e^{i\left(\frac{\omega}{c} +\frac{k}{2}\chi(\omega)\right)z}

To solve the problem we can develop the susceptibility:

χ(ω)=χ(0)+dχdωω\chi(\omega) = \chi(0)+\frac{d\chi}{d\omega}\omega

And we obtain:

E(t,z)=eizkχ(0)/2E(tz/vg,0)vg=c1+ωL2dχdω\mathcal{E}(t, z) =e^{izk\chi(0)/2} \mathcal{E}(t-z/v_g, 0)\\ v_g = \frac{c}{1+\frac{\omega_L}{2}\frac{d\chi}{d\omega}}

So the group velocity is controlled by the derivative of the susceptibility !

Two level system

After this rather general discussion, we will now employ it to understand the action of two-level systems on the travelling beam. So we will now focus on the influence of the atoms on the field in comparision with the previous discussions. Further, we will have to take into account the finite lifetime of the excited states in a phenomenological manner. For a two level system with excited state e\left|e\right\rangle and groundstate g\left|g\right\rangle, we can directly write down the wavefunction as:

ψ=γgg+γee\left|\psi\right\rangle = \gamma_g\left|g\right\rangle+ \gamma_e\left|e\right\rangle

In this basis, the dipole element reads:

D=eDgγeγg=dσeg\langle D\rangle = \left\langle e\right|D\left|g\right\rangle \gamma_e^*\gamma_g\\ = d \sigma_{eg}

In the second line we introduced the notations:

  • d=eDgd = \left\langle e\right|D\left|g\right\rangle

  • The product γeγg\gamma_e^*\gamma_g can identified with the off-diagonal component of the density operator σ=ψψ\sigma=\left|\psi\right\rangle\left\langle\psi\right|. We will often call it coherence.

The Hamiltonian of this model reads then in the rotating wave-approximation:

H^=0gg+δee+Ω[eg+ge]Ω=dE/\hat{H} = 0\left|g\right\rangle\left\langle g\right|+\hbar\delta \left|e\right\rangle\left\langle e\right| + \hbar\Omega\left[\left|e\right\rangle\left\langle g\right|+\left|g\right\rangle\left\langle e\right|\right]\\ \Omega = d E/\hbar

This is exactly the model that we discussed in the lectures 3 and 4 [@Jendrzejewskib; @Jendrzejewskia]. We then found that the time evolution might be described via:

iγ˙g(t)=Ωγeiγ˙e(t)=δγe+Ωγgi\dot{\gamma}_g(t) = \Omega \gamma_e\\ i\dot{\gamma}_e(t) = \delta \gamma_e +\Omega \gamma_g\\

We can combine them to the components of the density operator, which then read:

σij=cicj\sigma_{ij} = c_{i}^*c_j

From these coefficients, we can now obtain the evolution of the populations:

N˙g=σ˙gg=γ˙gγg+γgγ˙g=iΩ(σegσge)N˙e=N˙g\dot{N}_g = \dot{\sigma}_{gg} = \dot{\gamma}_{g}^*\gamma_g+ \gamma_{g}^*\dot{\gamma}_g\\ = i\Omega(\sigma_{eg}-\sigma_{ge})\\ \dot{N}_e = -\dot{N}_g

So the total number of atoms stays automatically conserved. As for the coherences we obtain:

σ˙eg=γ˙eγg+γeγ˙g=iδσeg+i(NgNe)Ωσ˙ge=iδσgei(NgNe)Ω\dot{\sigma}_{eg} = \dot{\gamma}_{e}^*\gamma_g+ \gamma_{e}^*\dot{\gamma}_g\\ = i\delta \sigma_{eg}+i (N_g-N_e)\Omega\\ \dot{\sigma}_{ge}= -i\delta \sigma_{ge}-i (N_g-N_e)\Omega

This density operator approach allows us to introduce spontaneous decay in a very straight-forward fashion:

  • The time evolution of the excited state gets an additional term ΓNe-\Gamma N_e.

  • Atoms coming from the excited state relax to the ground state, so we add a term ΓNe\Gamma N_e.

  • The coherence decays also through a term Γ2σge-\Gamma_2 \sigma_{ge}. We will use here for simplicity the limit of Γ2=Γ/2\Gamma_2 = \Gamma/2

So the full equations read now:

N˙g=iΩ(σegσge)+ΓNeσ˙ge=iδσgei(NgNe)ΩΓ2σge\dot{N}_g = i\Omega(\sigma_{eg}-\sigma_{ge})+\Gamma N_e\\ \dot{\sigma}_{ge}= -i\delta \sigma_{ge}-i (N_g-N_e)\Omega-\Gamma_2\sigma_{ge}

At this stage we can find the steady-state solutions by setting N˙g=σ˙ge=0\dot{N}_g = \dot{\sigma}_{ge} = 0. This leads too:

Ne=12Ω2Γ2Γ(ω0ωL)2+Γ22+Ω2Γ2Γ N_e = \frac{1}{2}\frac{\Omega^2 \frac{\Gamma_2}{\Gamma}}{(\omega_0-\omega_L)^2+\Gamma_2^2+\Omega^2\frac{\Gamma_2}{\Gamma}} σge=iΩ2Γ2i(ωLω0)Γ22+(ω0ωL)2+Ω2Γ2/Γ \sigma_{ge} = i\frac{\Omega}{2}\frac{\Gamma_2-i(\omega_L-\omega_0)}{\Gamma_2^2+(\omega_0-\omega_L)^2+\Omega^2\Gamma_2/\Gamma}

In the next lecture we will employ those results to study the laser.