1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: \usepackage{graphicx}
3:
4: \begin{document}
5:
6: \title{Magneto-optical rotation of spectrally impure fields and its nonlinear dependence on optical density}
7: \author{G. S. Agarwal and Shubhrangshu Dasgupta}
8: \affiliation{Physical Research Laboratory, Navrangpura, Ahmedabad-380 009, India}
9: \date{\today}
10:
11: \begin{abstract}
12: We calculate magneto-optical rotation of spectrally
13: impure fields in an optical thick cold atomic medium. We show that the
14: spectral impurity leads to nonlinear dependence of the rotation angle on optical density. Using our calculations, we provide a quantitative analysis of the
15: recent experimental results of
16: G. Labeyrie {\it et al.\/} [Phys. Rev. A {\bf 64}, 033402 (2001)] using cold
17: Rb$^{85}$ atoms..
18: \end{abstract}
19:
20: \pacs{33.55.Ad, 32.80.Pj}
21:
22: \maketitle
23:
24: \section{Introduction}
25: A very useful way to get important spectroscopic information is by measuring
26: magneto-optical rotation (MOR) of a plane polarized light propagating through a
27: medium \cite{rmp}. Clearly it is desirable to obtain as large an angle of rotation
28: as possible \cite{conn,gsa}. It is known that the angle of rotation is proportional
29: to the density of the medium. Thus an increase in density will help in achieving
30: large rotation angles. Recently, very large rotation angles in a cold sample
31: have been reported \cite{kaiser}. In this experiment, optical densities of the order of $10-20$ were
32: achieved. This experiment also reported a very interesting result, viz., the
33: departure from the linear dependence of the rotation angle on the optical
34: density. This departure has been ascribed to the nonmonochromatic nature of the
35: input laser. The findings of the experiment warrant a quantitative analysis of the
36: dependence of the rotation angle on the spectral profile of the input laser. We
37: present a first principle calculation of this dependence. Note that a nonlinear
38: dependence on optical density cannot result from a simple argument based on the
39: standard formula for the rotation angle $\theta$:
40: \begin{equation}
41: \label{eq1}\theta=\pi kl\;\textrm{Re}(\chi_--\chi_+)\;,
42: \end{equation}
43: where, $k$ is the wave-number of the electric field, $l$ is the length of the
44: medium, and $\chi_\pm$ represent the linear susceptibilities of the medium for right
45: or left circularly polarized components of the input field. Since $\chi_\pm$
46: are proportional to the number density, the rotation angle becomes proportional to the
47: optical density $\alpha$ defined by
48: \begin{equation}
49: \label{optden}\alpha=\frac{3\lambda^2}{2\pi}Nl\;,
50: \end{equation}
51: where, $\lambda=2\pi/k$ is the wavelength of the input field, $N$ is
52: the number density of the atomic medium.
53: If one were to argue that spectrally impure laser field would replace $\chi_\pm$
54: by their averages over the width of the laser, then $\theta$ would continue to
55: be proportional to the optical density $\alpha$. A more quantitative analysis of the rotation angle is thus warranted.
56:
57: The organization of the paper is as follows. In Sec.~\ref{sec:basic}, we recapitulate the
58: relevant equations for studies of MOR in atomic medium.
59: In Sec.~\ref{sec:three}, we describe a simple three-level atomic configuration in the context of
60: MOR and discuss the effect of laser line-shape and magnetic field in a thick
61: atomic medium. In Sec.~\ref{sec:general}, we consider experimental configuration used in \cite{kaiser}.
62: We show, how in an optically thick medium, one deviates from the linear dependence of rotation on optical density as a result of
63: the spectral impurity of the laser field. We give a quantitative analysis of the
64: experimental data.
65:
66:
67:
68: \section{\label{sec:basic}Basic equations}
69: Let us consider that an atomic medium of length $l$ is resonantly excited
70: by a monochromatic $\hat{x}$-polarized electric field
71: \begin{equation}
72: \vec{E}(z,t)=\hat{x}{\cal E}e^{ikz-i\omega t} +\textrm{c.c.}\;,
73: \label{field}
74: \end{equation}
75: where, ${\cal E}$ is the field amplitude and $k=\omega/c=2\pi/\lambda$ is the
76: wave number of the field, $\omega$ and $\lambda$ being the corresponding
77: angular frequency and wavelength. The field is propagating in $z$-direction.
78: Clearly, we can resolve the amplitude of the electric field into its two
79: circular components as
80: \begin{equation}
81: \hat{x}{\cal E}\equiv\hat{{\epsilon}}_+{\cal E}_++\hat{{\epsilon}}_-{\cal E}_-\;,
82: \label{circular}
83: \end{equation}
84: where, ${\cal E}_{\pm}={\cal E}/\sqrt{2}$ are the amplitude components along two
85: circular polarization $\hat{\epsilon}_{\pm}=(\hat{x}\pm i\hat{y})/\sqrt{2}$.
86:
87: While passing through the atomic medium these two circular components
88: behave differently due to the anisotropy of the medium. Let $\chi_{\pm}$ be the
89: susceptibilities of the medium corresponding to the two circular components. The
90: electric field at
91: the exit face of the medium can be written as
92: \begin{eqnarray}
93: \vec{E}(l,t)&=&\vec{\cal E}_le^{ikz-i\omega t}+\textrm{c.c.}\;,\nonumber\\
94: \label{fieldout}\vec{\cal E}_l&=&\left[\hat{\epsilon}_+{\cal E}_+e^{2\pi ikl\chi_+}+\hat{\epsilon}_-{\cal E}_-e^{2\pi ikl\chi_-}\right]\;,
95: \end{eqnarray}
96: where, we have assumed that the medium is dilute so that
97: $|4\pi\chi_{\pm}|\ll 1$. In MOR, the polarization
98: direction of the input electric field gets rotated due to the difference in
99: their dispersions (phase shifts) in a non-attenuating medium. The electric
100: field however, remains linearly polarized after passing through the medium. In
101: the present case, because the atom interacts with near-resonant electric
102: field, the two circular components suffer attenuation (given by the imaginary
103: part of $\chi_{\pm}$) while propagating through the medium. Thus the medium,
104: concerned here, is responsible for both dispersion and attenuation. We term
105: the medium to be both circularly birefringent and circularly dichroic. The
106: output electric field becomes elliptically polarized under the action of such a medium. Thus to
107: fully characterize the polarization state of the output field,
108: one has to use the Stokes parameters \cite{born}. The four Stokes parameters for
109: an electric field are designated by $S_{\alpha}$ ($\alpha =0,1,2,3$) and can
110: be defined as follows:
111: \begin{subequations}
112: \begin{eqnarray}
113: \label{s0}S_0&=&I_{\parallel}+I_{\perp}\;,\\
114: \label{s1}S_1&=&I_{\parallel}-I_{\perp}\;,\\
115: \label{s2}S_2&=&I_{45^o}-I_{-45^o}\;,\\
116: \label{s3}S_3&=&I_{\sigma_+}-I_{\sigma_-}\;,
117: \end{eqnarray}
118: \label{stokes}
119: \end{subequations}
120: where, $I_{\hat{n}}$ is the measured intensity along the polarization direction
121: $\hat{n}$. Then the output polarization state
122: can be characterized by the following three quantities:
123: \begin{subequations}
124: \begin{eqnarray}
125: \label{polar}P&=&\frac{\sqrt{S_1^2+S_2^2+S_3^2}}{S_0}\;,\\
126: \label{theta}\tan{2\theta}&=&\frac{S_2}{S_1}~~~~(0\leqslant\theta < \pi)\;,\\
127: \label{phi}\tan{2\phi}&=&\frac{S_3}{S_0P}~~~~(-\pi/4 <\phi \leqslant\pi/4)\;,
128: \end{eqnarray}
129: \label{defination}
130: \end{subequations}
131: where, $P$ is the degree of polarization, i.e., the ratio of the intensities of the
132: polarized component to the unpolarized one, $\theta$ is the Faraday rotation
133: angle of the input field and is measured between the major axis of the
134: ellipse and the $x$-axis, $\phi$ provides the ellipticity of polarization
135: through the relation $e=\tan{\phi}$.
136:
137: From Eq.~(\ref{fieldout}) one can express the output intensities along
138: different polarization directions in the following way:
139: \begin{subequations}
140: \begin{eqnarray}
141: I_{\parallel}(\omega)&=&|\hat{x}.\vec{\cal E}_l|^2=\frac{I_o}{4}|e^{2\pi ikl\chi_+}+e^{2\pi ikl \chi_-}|^2\;,\\
142: I_{\perp}(\omega)&=&|\hat{y}.\vec{\cal E}_l|^2=\frac{I_o}{4}|e^{2\pi ikl\chi_+}-e^{2\pi ikl\chi_-}|^2\;,\\
143: I_{\pm 45^o}(\omega)&=&\left|\frac{\hat{x}\pm\hat{y}}{\sqrt{2}}.\vec{\cal E}_l\right|^2\nonumber\\
144: &=&\frac{I_o}{8}|(1\pm i)e^{2\pi ikl\chi_+}+(1\mp i)e^{2\pi ikl\chi_-}|^2\;,\\
145: \label{s3a}I_{\sigma_\pm}(\omega)&=&|\hat{\epsilon}_\pm .\vec{\cal E}_l|^2=\frac{I_o}{2}\exp{[-4\pi kl.\textrm{Im}(\chi_\pm)]}\;,
146: \end{eqnarray}
147: \label{intensity}
148: \end{subequations}
149: where, $I_o=|{\cal E}|^2$ is the input intensity of laser field.
150:
151: Note that all the measured quantities defined by Eq.~(\ref{intensity}) are
152: functions of the frequency of the exciting field. If the exciting field is
153: spectrally impure, then the Stokes parameters $\langle S_\alpha\rangle$ are
154: to be
155: obtained by averaging over the spectrum $S(\omega)$ of the laser field. Thus,
156: $I$'s in Eq.~(\ref{stokes}) are to be obtained from
157: \begin{equation}
158: \langle I_{\hat{n}}\rangle=\frac{1}{I_o}\int_{-\infty}^{\infty}d\omega I_{\hat{n}}(\omega)S(\omega)\;.
159: \label{avg}
160: \end{equation}
161: For simplicity, we can adopt, say, a Lorentzian line-shape for the input
162: \begin{equation}
163: S(\omega)\equiv I_o\frac{\gamma_c/\pi}{\gamma_c^2+(\omega-\omega_l)^2}\;,
164: \label{lineshape}
165: \end{equation}
166: where, $\omega_l$ is the central frequency of the laser field and $2\gamma_c$ is the full width at half maximum. We will demonstrate how the fluctuations of the
167: input field leads to the nonlinear dependence of the rotation angle on
168: optical density.
169:
170: \begin{figure}
171: \scalebox{0.75}{\includegraphics{fig1.eps}}
172: \caption{\label{fig1}Level diagram for a three-level configuration. The
173: excited levels $|\pm\rangle$ ($m_e=\pm 1$) are Zeeman shifted from the level $m_e=0$ by an
174: amount $s$. The detuning $\delta$ is defined between the levels $m_e=0$ and
175: $m_g=0$.}
176: \end{figure}
177:
178: \section{\label{sec:three}A simplified atomic model}
179: We first consider a three-level atom in $V$ configuration (see Fig.~\ref{fig1}) in order to uncover the effect of optical density on MOR. The levels $|\pm\rangle$
180: ($J_e=1, m_e=\pm 1$) are coupled to the ground state $|g\rangle$
181: ($J_g=0, m_g=0$) by two circular components $\sigma_\pm$ of $\hat{x}$-polarized
182: electric field [Eq. (\ref{field})]. The excited level degeneracy
183: has been removed by an uniform magnetic field $\vec{B}$ applied in the
184: direction of propagation of the applied electric field. The levels
185: $|e_\pm\rangle$ are shifted about line-center by an amount of $\mp \mu_BB/\hbar$
186: ($\mu_B$=Bohr magneton). The field $\vec{E}$ is detuned from the
187: line center by an amount $\delta=\omega_{+g}(B=0)-\omega$, $\omega_{+g}(B=0)$
188: being the atomic transition frequency in absence of the magnetic field.
189:
190: The susceptibilities of the $\sigma_\pm$ components inside the medium can be
191: written as
192: \begin{equation}
193: \chi_\pm=\frac{N|\vec{d}|^2}{\hbar\gamma}\rho_\pm;~~\rho_\pm=\frac{i\gamma}{\gamma+i(\delta\mp s)}\;,
194: \label{succep}
195: \end{equation}
196: where, $2\gamma=4|\vec{d}|^2\omega^3/3\hbar c^3$ is the
197: spontaneous decay rate of the levels $|\pm\rangle$, $|\vec{d}|$ is the
198: magnitude of the dipole moment vector for the transitions $|\pm\rangle\leftrightarrow |g\rangle$,
199: $N$ is the atomic number density, and $s=\mu_BB/\hbar$ is the Zeeman
200: splitting of the excited levels. Using these $\chi_\pm$, we can now write the
201: field amplitude from Eq.~(\ref{fieldout}) as
202: \begin{equation}
203: \vec{\cal E}_0=\left[\hat{\epsilon}_+{\cal E}_+e^{i\frac{\alpha}{2}\rho_+}+\hat{\epsilon}_-{\cal E}_-e^{i\frac{\alpha}{2}\rho_-}\right]\;,
204: \label{eout1}
205: \end{equation}
206: where, $\alpha=4\pi klN|\vec{d}|^2/\hbar\gamma=(3\lambda^2/2\pi)Nl$ is the optical density of the medium.
207:
208: \begin{figure}
209: \scalebox{0.4}{\includegraphics{B2Three.eps}}
210: \caption{\label{fig2}Variation of MOR angle $\theta$ with optical density
211: $\alpha$ for $s=2\gamma$ and different laser line-widths $\gamma_c=0.1\gamma$ (solid line),
212: $\gamma_c=0.5\gamma$ (dashed line), $\gamma_c=\gamma$ (dot-dashed line), and $\gamma_c=2\gamma$
213: (long-dashed line). We have chosen $\lambda=422.67$ nm corresponding to Ca$^{40}$ 1S$_0 \leftrightarrow 1$P$_1$ transitions. Note the nonlinear dependence of $\theta$ on $\alpha$ for
214: larger $\gamma_c$.}
215: \end{figure}
216:
217: In what follows, we will assume that $\omega_l=\omega_{+g}(B=0)$. We calculate $\langle I_{\hat{n}}\rangle$'s using Eqs.~(\ref{lineshape}) and
218: (\ref{avg}) numerically for different values of $\gamma_c$ and $s$. We
219: show the results in Fig.~\ref{fig2}. We clearly see
220: that for $\gamma_c\ll s$, the rotation angle $\theta$ is
221: linearly proportional to $\alpha$. But for $\gamma_c\gtrsim s$, this variation
222: deviates from linearity in large $\alpha$ domain. This behavior can be explained
223: in terms of the off-resonant components which dominate for the large $\gamma_c$
224: and large $\alpha$.
225:
226: In order to understand the numerical results, we first consider the limit of
227: small optical densities whence
228: \begin{subequations}
229: \begin{eqnarray}
230: S_1&=&\frac{1}{2}[2+\alpha\textrm{Re}\{i(\rho_++\rho_-)\}]\;,\\
231: S_2&=&\frac{\alpha}{2}\textrm{Re}(\rho_--\rho_+)\;.
232: \end{eqnarray}
233: \label{small1}
234: \end{subequations}
235: Thus, the departure of the rotation angle from the linearity has to do with the
236: averages of the exponentials appearing in $I$'s [Eq.~(\ref{intensity})]. If
237: one were to make the approximation of replacing all $\chi$'s in Eq.~(\ref{intensity})
238: by their averages, i.e.,
239: \begin{equation}
240: \langle \exp{[2\pi ikl\chi_\pm]}\rangle\equiv \exp[2\pi ikl\langle\chi_\pm\rangle]\;,
241: \label{approx}
242: \end{equation}
243: then the Stokes parameters $\langle S_1\rangle$ and $\langle S_2\rangle$ would be
244: \begin{subequations}
245: \begin{eqnarray}
246: \langle S_1\rangle&=&e^{-\frac{\alpha}{2}(\langle\rho_2^+\rangle+\langle\rho_2^-\rangle)}\cos{\left[\frac{\alpha}{2}(\langle\rho_1^+\rangle-\langle\rho_1^-\rangle)\right]}\;,\\
247: \langle S_2\rangle&=&-e^{-\frac{\alpha}{2}(\langle\rho_2^+\rangle+\langle\rho_2^-\rangle)}\sin{\left[\frac{\alpha}{2}(\langle\rho_1^+\rangle-\langle\rho_1^-\rangle)\right]}\;,
248: \end{eqnarray}
249: \label{small3}
250: \end{subequations}
251: where, $\langle\rho_\pm\rangle=\langle\rho_1^\pm\rangle +i\langle\rho_2^\pm\rangle$ and thus
252: \begin{figure*}
253: \scalebox{0.9}{\includegraphics{fig16.eps}}
254: \caption{\label{fig16}Level diagram for the $F_e=4 \leftrightarrow F_g=3$
255: transition. The numbers at the top of the figure indicate the magnetic
256: quantum numbers of the sublevels. The relevant Clebsch-Gordan coefficients for the corresponding
257: transitions are given by $a_1=-1/\sqrt{42}, a_2=-\sqrt{5}/3\sqrt{14}, a_3=-1/2\sqrt{21}, a_4=-\sqrt{5}/2\sqrt{21}, a_5=-1/6\sqrt{7}, a_6=-1/2\sqrt{3}, a_7=-1/3, a_8=-1/6, a_9=-1/\sqrt{21}, a_{10}=-\sqrt{15}/6\sqrt{7}, a_{11}=2/3\sqrt{7}$.
258: The Zeeman splitting of the various sublevels are not shown.}
259: \end{figure*}
260: \begin{equation}
261: \langle\theta\rangle=\frac{\alpha}{4}(\langle\rho_1^-\rangle-\langle\rho_1^+\rangle)=\frac{1}{2}.\frac{\alpha\gamma s}{(\gamma+\gamma_c)^2+s^2}\;.
262: \label{small4}
263: \end{equation}
264: Clearly, the absorption does not contribute to the rotation angle. We have again
265: recovered the linear dependence of $\theta$ on $\alpha$, provided the approximation
266: (\ref{approx}) is valid. Thus, any departure in linearity of $\theta$ with
267: respect to $\alpha$ indicates the breakdown the approximation (\ref{approx}).
268: The numerical results of Fig.~\ref{fig2} clearly show the breakdown of
269: the
270: mean field description obtained by replacing $\chi$'s by their average values.
271:
272: From the Eq.~(\ref{small4}), we readily see that in the low
273: $\alpha$-domain, by increasing $s$ (or $\gamma_c$) while keeping $\gamma_c$ (or $s$)
274: constant, the slope of $\theta$ with $\alpha$ decreases. This is clear from
275: the numerical results of Fig.~\ref{fig2}. Also, for larger
276: values of $\gamma_c$, variation of $\theta$ with $\alpha$ is deviated from
277: linearity.
278: Linear variation of $\theta$ with $\alpha$ is attributed to the monochromatic laser field.
279: If the electric field is spectrally impure, then the off-resonant components
280: also contribute to $\theta$, through the relations (\ref{avg}), (\ref{stokes}), and (\ref{defination}).
281: Thus, $\theta$ starts
282: varying with $\alpha$ linearly in low $\alpha$ limit, then saturates, and finally decreases
283: to zero to change the direction of rotation, for larger $\gamma_c$ (see Fig.~\ref{fig2}). But for smaller
284: $\gamma_c$, the linear behavior is retained even for larger $\alpha$, as
285: the off-resonant components are not dominant in this parameter zone.
286: \begin{figure*}
287: \begin{tabular}{cc}
288: \scalebox{0.4}{\includegraphics{B2.eps}}&
289: \scalebox{0.4}{\includegraphics{B8.eps}}
290: \end{tabular}
291: \caption{\label{gen}Variation of magneto-optical rotation angle $\theta$ with
292: $\alpha$ for magnetic field (a) 2 Gauss ($\equiv 2\pi\times 2.8$ MHz) and (b) 8 Gauss ($\equiv 2\pi\times 11.2$ MHz) for laser line-widths
293: $2\gamma_c=2\pi\times 0.5$ MHz (solid line), $2\gamma_c=2\pi\times 1$ MHz (dashed line), $2\gamma_c=2\pi\times 3$ MHz (dot-dashed line), and $2\gamma_c=2\pi\times 5$ MHz (long-dashed line). The dot-dashed curves correspond to the width of the laser used in the experiment \cite{kaiser}. Note that the line-width of the D$_2$ line is $2\pi\times 5.88$ MHz.}
294: \end{figure*}
295:
296: Next we consider the variation of the degree of polarization $P$ and the ellipticity
297: $e$ with $\alpha$. We have noticed that $P$ decreases from unity for increasing
298: $\alpha$. This means that the output field no longer remains fully polarized,
299: rather it becomes partially polarized.
300:
301: Again, from the Eqs.~(\ref{s3}), (\ref{s3a}), and (\ref{succep}), it is clear
302: that an integration over the entire range of detuning $\delta$ would yield
303: $\langle S_3\rangle=0$, as the integrand is an odd function of $\omega$.
304: Thus the ellipticity $e$ becomes zero. This means that
305: the polarized part of the output field remains linear.
306: \begin{figure*}
307: \begin{tabular}{cc}
308: \scalebox{0.4}{\includegraphics{B2P.eps}}&
309: \scalebox{0.4}{\includegraphics{B8P.eps}}
310: \end{tabular}
311: \caption{\label{figP}Variation of degree of polarization $P$ with $\alpha$ is shown for magnetic field
312: (a) 2 Gauss and (b) 8 Gauss, for laser line-widths $2\gamma_c=2\pi\times 0.5$ MHz (solid line), $2\gamma_c=2\pi\times 1$ MHz (dashed line), $2\gamma_c=2\pi\times 3$ MHz (dot-dashed line), and $2\gamma_c=2\pi\times 5$ MHz (long-dashed line). The dot-dashed curves correspond to the width of the laser used in the
313: experiment \cite{kaiser}.}
314: \end{figure*}
315:
316: From the above discussion, it is clear that the output field gets
317: rotated as a manifestation of cumulative effect of optical density, magnetic
318: field, and laser line-width. Also it becomes partially polarized with
319: no ellipticity.
320:
321: \section{\label{sec:general}Quantitative modeling of experimental results of
322: Labeyrie \lowercase{{\it et al.\/}} for MOR in spectrally impure fields}
323:
324: We now extend our understanding of resonant MOR as described in the previous
325: section to explain the experimental data of Labeyrie {\it et al.\/}. In
326: their experiment, a cold atomic cloud of Rb$^{85}$ is subjected to a
327: static magnetic field. The laser probe beam passing through the medium in
328: the direction of the magnetic field is tuned to the D$_2$ line of the atoms
329: (2S$_{1/2}\leftrightarrow$ 2P$_{3/2}$; $\lambda$=780.2 nm). They have measured the
330: intensities of outputs with different polarizations, as function of laser
331: detuning and also at different values of optical density. They have found a
332: nonlinear dependence of the
333: MOR angle $\theta$ on optical density. They found for larger magnetic field that
334: the linear behavior is recovered.
335:
336: To explain these observations, we consider the relevant energy levels of
337: Rb$^{85}$ as used in the experiment (see Fig.~\ref{fig16}). The $\hat{x}$-polarized electric field
338: (\ref{field}) is applied to cold Rb$^{85}$ medium near resonantly. The
339: medium is subjected to uniform magnetic field $\vec{B}$ applied in the $z$-direction,
340: i.e., along the direction of propagation of (\ref{field}).
341:
342: \subsection{Calculation of $\chi_\pm$ and optical density}
343:
344: The circular components $\sigma_\pm$ of the input electric field (\ref{field})
345: interact with the transitions $m_e\leftrightarrow m_g=m_e-1$ and
346: $m_e\leftrightarrow m_g=m_e+1$, respectively. We assume that the electric field is weak enough
347: so that it is sufficient to use the linear response of the system to the laser
348: field. We neglect the ground-state
349: coherences. As we are considering the cold atoms, we neglect the collisional
350: relaxations and Doppler broadening of the sublevels. We also assume that the atomic
351: population is equally distributed over all the ground sublevels.
352:
353: Using all these assumptions, we can write the susceptibilities $\chi_\pm$ for the
354: $\sigma_\pm$ components as the sum of the susceptibilities of all the relevant
355: $m_e\leftrightarrow m_g$ transitions in the following way:
356: \begin{equation}
357: \label{chis}\chi_\pm\equiv\sum_{m_e,m_g}\frac{1}{7}\frac{N|\vec{d}_{m_e,m_g}.\hat{\epsilon}_\pm|^2}{\hbar}.\frac{i}{\Gamma_{m_e,m_g}+i(\delta+ s_{m_e,m_g})}\;,
358: \end{equation}
359: where, $s_{m_e,m_g}=(g_gm_g-g_em_e)s$ is the relative amount of
360: Zeeman shift of the excited sublevel $m_e$ with respect to the
361: Zeeman shifted ground sublevel $m_g$, $g_g=1/3$ and $g_e=1/2$ are the Land\'e-g
362: factors of the ground and excited levels, respectively.
363: The factor $1/7$ comes into the expression (\ref{chis}) as we have assumed equal population distribution in all the $(2F_g+1)=7$ ground sublevels.
364: The coherence
365: relaxation rate $\Gamma_{m_e,m_g}$ in Eq.~(\ref{chis}) is given by
366: \begin{equation}
367: \Gamma_{m_e,m_g}=\frac{1}{2}\sum_{k}\gamma_{k,m_e}\;,
368: \end{equation}
369: where, $\gamma_{i,j}$ is the spontaneous relaxation rate from the
370: sublevel $j$ to $i$. Here we have assumed that there is
371: no spontaneous relaxation from the ground sublevels. The terms $\vec{d}_{m_e,m_g}$ and $\Gamma_{m_e,m_g}$'s can be
372: calculated from the relevant Clebsch-Gordan coefficients (see Fig.~\ref{fig16})
373: \cite{sobel}. The Einstein's A coefficient for the D$_2$ line is known to be
374: \begin{widetext}
375: \begin{equation}
376: A=\frac{4\omega^3}{3\hbar c^3}\frac{|(J=\frac{3}{2}\parallel D\parallel J'=\frac{1}{2})|^2}{4}=\frac{4\omega^3}{3\hbar c^3}\frac{|(\frac{3}{2}, \frac{5}{2}, 4\parallel D\parallel \frac{1}{2}, \frac{5}{2}, 3)|^2}{9}\;,
377: \end{equation}
378: \end{widetext}
379: where, $(~\parallel D\parallel ~)$ represents the reduced matrix element of the dipole moment vector $\vec{d}_{m_e,m_g}$. The three symbols $3/2$, $5/2$, and $4$ correspond to $J$, $I$, and $F$ values respectively of the upper levels.
380: Thus all $\Gamma_{m_e,m_g}$'s in (\ref{chis}) are found to be equal to $(4\omega^3/3\hbar c^3)|(\frac{3}{2}, \frac{5}{2}, 4\parallel D\parallel \frac{1}{2}, \frac{5}{2}, 3)|^2/2$.
381:
382: We calculate the optical density $\alpha$ of the medium, when the input light
383: field is resonant with $m_e=0\leftrightarrow m_g=0$ transition ($\delta=0$) in the
384: absence of any magnetic field ($B=0$). For this, we first obtain the total output
385: intensity from Eq.~(\ref{s0})
386: averaged by a very narrow laser line-shape, i.e., in the limit $\gamma_c\rightarrow 0$. Using Eq.~(\ref{chis}), we thus find that the transmittivity of the
387: medium becomes
388: \begin{equation}
389: T=\frac{1}{I_o}\langle S_0\rangle_{\gamma_c\rightarrow 0}=\frac{1}{I_o}S_0|_{\delta=0}=\exp(-\alpha)\;,
390: \end{equation}
391: where, $\alpha=(3/7).(3\lambda^2/2\pi) Nl$.
392: It should be borne in mind that it is different from the definition in Sec.~\ref{sec:three}.
393:
394:
395: \subsection{Discussions}
396:
397: Using the above expressions of $\chi_\pm$ [Eq.~(\ref{chis})] and Eq.~(\ref{avg}),
398: we calculate the averaged intensities $\langle I_{\hat{n}}\rangle$ in different polarization directions. The Stokes parameters
399: $S_\alpha$, degree of polarization $P$, and the Faraday rotation $\theta$ are
400: calculated using the relations (\ref{defination}).
401: In Fig.~\ref{gen}, we show how the Faraday angle $\theta$ varies with the
402: optical density $\alpha$ for different values of $\gamma_c$ and $B$. Clearly,
403: for $\gamma_c\ll s$, the rotation angle $\theta$ varies linearly with $\alpha$.
404: But for larger $\gamma_c$ ($\gtrsim s$), the variation of $\theta$
405: with $\alpha$ deviates from linearity in large $\alpha$. This is because the
406: off-resonant components contribute to the output intensity. Also, note that
407: for a given value of
408: $\gamma_c$, if $s$ is increased, the linearity is maintained even in the large $\alpha$-domain.
409: This is because for larger $s$, the off-resonant components do not contribute
410: much to the output intensity. The resonant frequency component is always
411: dominant in the optical density range considered. We also note that, as
412: $\gamma_c$ increases, the linear slope of $\theta$ with $\alpha$ decreases in the small $\alpha$ domain.
413:
414: In Fig.~\ref{figP}, we show the variation of degree of polarization $P$ with $\alpha$ for various values of the $B$ and $\gamma_c$. These results reveal that
415: with increase in $\alpha$, the degree of polarization deviates from unity,
416: i.e., the
417: output electric field not only rotates in polarization, but also it becomes
418: {\it partially\/} polarized. However, the ellipticity of the output field still
419: remains zero as we have argued in Sec.~\ref{sec:three}.
420:
421: \section{conclusions}
422:
423: In summary, we have given a quantitative analysis of magneto-optical rotation of
424: spectrally impure fields in optically thick cold Rb$^{85}$ atomic medium. We have shown that the dependence of rotation
425: on the optical density of the medium deviates from linearity
426: due to the finite laser linewidth. Using our
427: model, we explain the experimental results of Labeyrie {\it et al.\/}.
428:
429: \begin{thebibliography}{999}
430:
431: \bibitem{rmp}
432: D. Budker, W. Gawlik, D. F. Kimball, S. M. Rochester, V. V. Yashchuk, and A. Weis, Rev. Mod. Phys. {\bf 74}, 1153 (2002).
433:
434: \bibitem{conn}
435: J.-P. Connerade, J. Phys. B: At. Mol. Phys. {\bf 16}, 399 (1983).
436:
437: \bibitem{gsa}
438: G. S. Agarwal, P. Anantha Lakshmi, J.-P. Connerade, and S. West, J. Phys. B: At. Mol. Opt. Phys. {\bf 30}, 5971 (1997).
439:
440: \bibitem{kaiser}
441: G. Labeyrie, C. Miniatura, and R. Kaiser, Phys. Rev. A {\bf 64}, 033402 (2001).
442:
443: \bibitem{born}
444: M. Born and E. Wolf, {\it Principles of Optics\/}, 7th Ed. (Cambridge University Press, Cambridge, 1999), p. 630.
445:
446: \bibitem{sobel}
447: I. I. Sobel'man, {\it Atomic Spectra and Radiative Transitions\/}, 2nd Ed. (Springer Verlag, Berlin, 1992), Sec. 9.3.6.
448:
449: \end{thebibliography}
450:
451: \end{document}
452: