1: \documentclass[aps,twocolumn,superscriptaddress,showpacs]{revtex4}
2: %\documentclass[preprint,onecolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: %\documentclass[epsfig,preprint,aps]{revtex4}
4: %\documentclass[twocolumn,pra]{revtex4}% Physical Review A
5: \usepackage{graphicx}% Include figure files
6: %\usepackage{slashbox}
7:
8: \begin{document}
9:
10: \title{Dipolar effect in coherent spin mixing of two atoms in a single optical lattice site}
11:
12: \author{B.~Sun}
13: \affiliation{School of Physics, Georgia Institute of Technology,
14: Atlanta, Georgia 30332, USA.}
15:
16: \author{W.~X.~Zhang}
17: \affiliation{Ames Laboratory, Iowa State University, Ames, Iowa
18: 50011, USA.}
19:
20: \author{S. Yi}
21: \affiliation{Institute of Theoretical
22: Physics, The Chinese Academy of Sciences, Beijing 100080, China}
23:
24: \author{M. S. Chapman}
25: \affiliation{School of Physics, Georgia Institute of Technology,
26: Atlanta, Georgia 30332, USA.}
27:
28: \author{L.~You}
29: \affiliation{School of Physics, Georgia Institute of Technology,
30: Atlanta, Georgia 30332, USA.}
31:
32: \date{\today}
33:
34: \begin{abstract}
35: We show that atomic dipolar effects are detectable in the system
36: that recently demonstrated two-atom coherent spin dynamics within
37: individual lattice sites of a Mott state. Based on a two-state
38: approximation for the two-atom internal states and relying on a
39: variational approach, we have estimated the spin dipolar effect.
40: Despite the absolute weakness of the dipole-dipole interaction, it
41: is shown that it leads to experimentally observable effects in the
42: spin mixing dynamics.
43: \end{abstract}
44:
45: \pacs{03.75.Lm, 03.75.Gg, 03.75.Mn, 34.50.-s}
46:
47: \maketitle
48:
49: The recent successes of coherent spin mixing
50: \cite{MSZhang,HS,Bloch1} inside Bose condensed atoms, driven by
51: reversible collisions between pairs of atoms
52: $|M_F=0\rangle+|M_F'=0\rangle\leftrightarrow
53: |-1\rangle+|1\rangle$, have generated significant interest in the
54: quantum dynamics of atomic spins \cite{jason,machida,law,kurn}.
55: These experiments cover a broad limit from a condensate with large
56: number of atoms \cite{MSZhang,HS} to many identical trapping sites
57: as in an optical lattice each containing two atoms \cite{Bloch1},
58: and raise significant hope for the long discussed applications of
59: atomic quantum gases to the emerging field of quantum information
60: science. As observed in these experiments, the spin coherence time
61: rivals the best motional state coherence ever achieved on neutral
62: atoms \cite{eric} and is ideally suited for quantum information
63: processing applications \cite{zoller,Bloch}.
64:
65: In this paper, we reveal an interesting observation: the atomic
66: spin dipolar effects seem to be detectable in the system that
67: recently demonstrated two-atom coherent spin mixing \cite{Bloch1}.
68: Dipolar interactions are ubiquitous in atomic systems. It offers a
69: unique playing field because of the two significant differences
70: from the the nominal $s$-wave short-ranged interactions: it is
71: anisotropic and long-ranged \cite{dipyou}. These interesting
72: properties have stimulated intensive research of dipolar
73: condensates \cite{yi1,goral1,santos,giov,pfau}.
74:
75: %The successful observation of a Cr dipolar condensate has
76: %generated renewed interest \cite{}.
77:
78: Our model system consists of two spinor atoms ($^{87}$Rb)
79: inside a harmonic trap. Corrections due to the anharmonic trapping
80: potential of a standing wave optical
81: lattice \cite{artur} will be addressed elsewhere.
82: We will consider the simple case of a cylindrical harmonic trap
83: instead of the spherical harmonic trap
84: as in the experiment \cite{Bloch1},
85: because the dipolar interaction is known to display a sensitive dependence on
86: the trap aspect ratios \cite{yi1}. The spherical harmonic
87: trap minimizes the dipolar effect and thus provides an excellent
88: calibration for the experimental systems,
89: particularly the small spin exchange interaction \cite{MSZhang,artur}.
90: Following the usual procedure of separating the total motion
91: into the center of mass (CM) and the relative (rel) motion,
92: the system Hamiltonian becomes
93: $H = H_{\mathrm{CM}} + H_{\mathrm{rel}}$ with
94: \begin{eqnarray}
95: H_{\mathrm{CM}} &=& -{\hbar^2 \nabla^2_{\vec{R}}\over 2M} + {1\over
96: 2}M\omega_\rho^2(X^2+Y^2+\lambda^2Z^2),\\
97: H_{\mathrm{rel}} &=& H_0 + H_s + H_{dd}+ H_{B},
98: \end{eqnarray}
99: for the CM- and rel-motion, respectively, with coordinates
100: $\vec{R} = {(\vec{r}_1 + \vec{r}_2) / 2}$ and
101: $\vec{r} = \vec{r}_1 -\vec{r}_2 $. $M=2m$ is the total
102: mass, while $\mu={m/ 2}$ is the reduced mass.
103: $H_{\mathrm{rel}}$ contains several parts as outlined below,
104: \begin{eqnarray}
105: H_0 &=& -{\hbar^2\nabla^2_{\vec{r}}\over 2\mu} + {1\over 2}\mu\omega_\rho^2(x^2+y^2+\lambda^2z^2), \\
106: H_s &=&(c_0+c_2\vec{F}_1\cdot\vec{F}_2)\delta^{({\rm reg})}(\vec{r}), \\
107: H_{dd} &=& {c_d\over r^3} \left[\vec{F}_1\cdot\vec{F}_2
108: -3(\vec{F}_1\cdot\hat{r})\cdot (\vec{F}_2\cdot\hat{r})\right], \label{dip}
109: \end{eqnarray}
110: where $H_0$ describes the rel-motional,
111: and $H_s$ accounts for the regularized s-wave contact
112: interactions between two spin-1 atoms with $\delta^{({\rm reg})}(\vec{r})
113: \equiv \delta(\vec{r})({\partial/
114: \partial r})r$. $H_{dd}$ denotes the spin dipolar interaction.
115: The first
116: order Zeeman effect does not contribute due to a zero
117: magnetization, while the second order Zeeman effect
118: $H_{B}\approx \pm 72(\vec F\cdot\vec B)^2$ (Hz/Gauss$^2$)
119: in reference to the $M_F=0$ state with the `+' and `$-$' signs
120: for $F=1$ and $F=2$ respectively. The various interaction
121: coefficients are listed below
122: \begin{eqnarray}
123: c_0 &=& {4\pi \hbar^2\over m} {a_0 + 2a_2\over 3}, \\
124: c_2 &=& {4\pi \hbar^2\over m} {a_2 - a_0\over 3}, \\
125: c_d &=& {\mu_0\over 4\pi} g_F^2 \mu_B^2.
126: \end{eqnarray}
127: $a_{0(2)}$ is the scattering length for the combined channel of total $F=0(2)$.
128: $g_F$ is the Land$\acute{e}$ g-factor for the hyperfine spin state of $F$,
129: and $\mu_B$ is the Bohr magneton.
130:
131: Before presenting our theoretical analysis, we comment on the
132: strength of spin dipolar interactions \cite{pu}. In a scalar
133: condensate, the dipolar effect is usually calibrated against the
134: nominal $s$-wave interaction, i.e., one simply compares $|c_d|$
135: with $|c_0|$ and uses their ratio as a parameter. This ratio can
136: be increased by decreasing $c_0$ perhaps through a Feshbach
137: resonance \cite{jila}. For spin dipoles, however, the spin
138: exchange interaction coefficient $c_2$ also needs to be compared.
139: The extremely small $c_2$ for $^{87}$Rb serves to enhance the spin
140: dipolar effect because the spinor nature of the ground state
141: energy is determined by $c_2$ not by the much larger $c_0$. Using
142: $|c_d|/|c_2|$ as a parameter, $^{87}$Rb could be viewed as a
143: stronger spin dipolar condensate than Cr. Although in a polarized
144: condensate, Cr atoms $(F=3)$ with a dipole moment of 6$\mu_B$
145: enhance significantly the dipolar interaction. When the spinor
146: nature of the dipole is of interest \cite{pu,dipsantos,ueda}, the
147: dipolar effect for Cr is weaker because of the large spin exchange
148: interactions \cite{pu,ho}.
149:
150: We first will estimate the relative strength of different
151: interaction terms for the situation as in Ref. \cite{Bloch1}. For
152: an optical lattice $\propto
153: V_0[\sin^2(kx)+\sin^2(ky)+\lambda^2\sin^2(kz)]$ of a depth $V_0=s
154: E_r$ in units of the recoil energy $E_r$ for the lattice laser,
155: each single lattice site is approximated like a harmonic trap with
156: a radial frequency $\omega_\rho=\sqrt{ 2s E_r k^2 / m}= {2\sqrt{s}
157: E_r / \hbar}$
158: %\begin{eqnarray}
159: %\omega_\rho=\sqrt{ 2s E_r k^2 \over m}= {2\sqrt{s} E_r \over
160: %\hbar},
161: %\end{eqnarray}
162: and a Gaussian relative motional ground state
163: \begin{eqnarray}
164: \phi_0(\vec{r}) = {\lambda^{1/4} \over \pi^{3/4} a_\rho^{3/2}}\,
165: \exp\left({-{\rho^2\over 2\,a_\rho^2}- {\lambda z^2\over 2 a_\rho^2} }\right),
166: \label{gw0}
167: \end{eqnarray}
168: with a radial width $a_\rho=\sqrt{\hbar/(\mu\omega_\rho)}$
169: ($\rho=\sqrt{x^2+y^2}$). The center of mass motional ground state
170: $\Phi_0(\vec R)$ takes the same form except for the width
171: $A_\rho=\sqrt{\hbar/(2m\omega_\rho)}$. For a wavelength of $840$
172: nm, we find $\omega_\rho \sim (2\pi) 41,126.3$ Hz, and
173: $a_\rho=0.0752$ $\mu$m for $s=40$. We further note that $a_0=
174: 101.8a_B $ and $a_2 =100.4a_B$, with $a_B$ being the Bohr radius
175: \cite{coef}. These give rise to typical density interaction $c_0
176: \langle n\rangle_0 /\hbar\sim (2\pi)6,589$ Hz, spin exchange
177: interaction $c_2 \langle n\rangle_0 /\hbar\sim (2\pi)(-30)$ Hz,
178: and spin dipolar interaction $c_d \langle n\rangle_0/\hbar \sim
179: (2\pi)2.71$ Hz for the spherically symmetric case of $\lambda=1$,
180: all much less that the trap level spacing. Thus, a reasonable
181: estimate of the motional state would be the non-interacting ground
182: state $\phi_0(\vec{r})$ from which the averaged density becomes
183: \begin{eqnarray}
184: \langle n \rangle_0=2\int |\phi_0(\vec r)|^4 d\vec{r}
185: ={\sqrt{\lambda}\over \sqrt{2}\,\pi^{3/2} a_\rho^3}.
186: \end{eqnarray}
187: We further note the relative strength of $c_2/c_0 \sim
188: -0.00462657$ and $c_d/c_2 \sim -0.0902$, which encourages a
189: Gaussian variational ansatz \cite{yi1}.
190:
191: %To understand the spin mixing dynamics, we rely on an approximate
192: %picture based on the estimates of the various types of previously
193: %discussed interaction strength.
194: The dominant mixing interaction is spin exchange that couples the
195: two-atom internal state $|M_F=0,M_F'=0\rangle$ to $|1,-1\rangle$.
196: The spin dipolar interaction is about 5 times smaller and averages
197: to vanishingly small net effect for a spherical symmetric motional
198: state. We therefore will limit our discussions to the above two
199: internal states, a picture uniformly adopted by the
200: experimentalists \cite{MSZhang,HS,Bloch1}. We have further carried
201: out numerical simulations at $B=0$ of the full system dynamics,
202: including all other spin states that are coupled by the dipolar
203: term, i.e., with the complete Hilbert space of spin degree of
204: freedom: $|M_F=-1,0,1;M_F'=-1,0,1\rangle$. For a spherical trap
205: with the same (averaged) frequency, the probability of atoms in
206: other spin states that do not conserve the total magnetization
207: ($|1,1\rangle$, $|-1,-1\rangle$, $|0,1\rangle$, and
208: $|0,-1\rangle$), which are responsible for the dipolar relaxation,
209: is found to be less than $10^{-6}$ in the first oscillation period
210: of coherent spin mixing; even for a trap with $\lambda=3$, this
211: probability remains negligible, and is in fact only several times
212: enhanced. Thus the negligibly small dipolar spin relaxation in
213: $^{87}$Rb \cite{coef} makes the two-state model an excellent
214: approximation for our system. At finite values of the B-field,
215: except for accidental resonances when other spin states in higher
216: motional states are shifted into near resonance with the two-state
217: doublet in ground motional state, the linear Zeeman effect
218: generally leads to large detunings, also validates the
219: approximation. Even at accidental resonances, the total population
220: out of the two-state doublet is found to be only $\sim 10^{-3}$
221: for a spherical trap. For Cr atoms, however, more effort is needed
222: to understand the conditions for spin mixing dynamics due to the
223: much enhanced dipolar relaxation \cite{sr}.
224:
225: The two-atom wave function is then approximated as
226: \begin{eqnarray}
227: \alpha_{0,0}|0,0\rangle\psi_{0,0}(\vec r_1,\vec r_2)
228: +\alpha_{1,-1}|1,-1\rangle\psi_{1,-1}(\vec r_1,\vec r_2). \hskip
229: 12pt
230: \end{eqnarray}
231: This leads to a spin mixing matrix element of
232: \begin{eqnarray}
233: {1\over 2}
234: \hbar\Omega = \int d\vec r_1\int\!\!&d\vec r_2&\!\!
235: \psi_{0,0}^*(\vec r_1,\vec r_2)
236: \langle 0,0|(H_s\nonumber\\
237: && +H_{dd})
238: |1,-1\rangle\psi_{1,-1}(\vec r_1,\vec r_2). \hskip 18pt
239: \label{g0}
240: \end{eqnarray}
241:
242: As a first approximation, the relative motion is
243: simply taken to be the ground state of the harmonic trap, i.e.,
244: $\psi_{M_F,M_F'}(\vec r_1,\vec r_2)=\Phi_0(\vec{R})\phi_{M_F,M_F'}(\vec{r})$
245: with $\phi_{0,0}(\vec{r})=\phi_{1,-1}(\vec{r})=\phi_{0}(\vec{r})$.
246: This leads to
247: \begin{eqnarray}
248: {1\over 2} \hbar\Omega &=& \int d\vec r
249: \phi_{0,0}^*(\vec{r})\langle 0,0|H_s+H_{dd}
250: |1,-1\rangle\phi_{1,-1}(\vec{r}). \hskip 12pt
251: \label{feq}
252: \end{eqnarray}
253:
254: An improved approximation is the variational calculation, labelled as $\phi^{(v)}$,
255: for the relative motional state including
256: the dipolar interaction as have been used extensively in the past \cite{yi1}.
257: We take a Gaussian ansatz with its widths $w_{\rho/z}$ as variational
258: parameters \cite{yi1}
259: $$\phi_{1,-1}(\vec{r}) = {1\over \pi^{3/4}(w_{\rho}^2 w_z)^{1/2}}
260: \,\mathrm{exp} \left(- {x^2+y^2\over 2 w_{\rho}^2} -
261: {z^2\over 2 w_z^2} \right),
262: $$
263: the relative energy functional
264: then becomes
265: \begin{eqnarray}
266: {E\over \hbar\omega_{\rho}} &=& {1\over 4}\left(2{a_\rho^2 \over
267: w_{\rho}^2} +{a_\rho^2 \over w_{z}^2} \right) + {1\over
268: 4}\left(2{w_{\rho}^2 \over a_\rho^2} + \lambda^2 { w_{z}^2 \over
269: a_{\rho}^2} \right)\nonumber\\
270: && + {c_0-c_2\over
271: \pi^{3/2}w_{\rho}^2w_{z}\hbar\omega_{\rho}}
272: -{2\over 3\sqrt{\pi}}{c_d \over
273: w_{\rho}^2w_{z}\hbar\omega_{\rho}} \chi(\kappa), \hskip 18pt
274: \label{Ey}
275: \end{eqnarray}
276: with $\chi(\kappa)=[2\kappa^2+1-3\kappa^2H(\kappa)]/[
277: 2(\kappa^2-1)] + (\kappa^2-1) H(\kappa)$ and $H(\kappa)=
278: {\mathrm{tanh}^{-1}\sqrt{1-\kappa^2}/ \sqrt{1-\kappa^2} }$. $a_{z}
279: = \sqrt{\hbar/ (\mu \omega_{z})}$ is the axial width of the trap,
280: and $\kappa=w_{\rho}/w_z$ is the aspect ratio of the variational
281: ground state, non-spherical (or $\kappa\neq 1$) even in a
282: spherical harmonic trap with $\lambda=\omega_z/\omega_\rho=1$
283: because of the dipolar interaction \cite{yi1}.
284: $\phi_{0,0}(\vec{r})$ is obtained from the result of
285: $\phi_{1,-1}(\vec{r})$ by excluding the dipolar interaction or
286: taking $c_d=0$ and adjusting to its own $s$-wave scattering
287: strength by taking $c_2=0$.
288:
289: A mean field approach is sometimes used in the
290: literature where the two-atom motional state
291: $\psi_{M_F,M_F'}(\vec r_1,\vec r_2)$ is approximated by
292: $\phi_{c}^{M_F,M_F'}(\vec{r}_1)\phi_{c}^{M_F,M_F'}(\vec{r}_2)$
293: as for a two-atom condensate with
294: $\phi_{c}^{M_F,M_F'}(\vec{r})$ obtained from the corresponding
295: Gross-Pitaevskii equation
296: \begin{eqnarray}
297: \left[-{\hbar^2\nabla_{\vec{r}}^2\over 2m}+{m\omega_\rho^2\over
298: 2}(\rho^2+\lambda^2 z^2) +V_{\rm
299: int}\right]\phi_{c}(\vec{r})=\mu_c \phi_{c},
300: \end{eqnarray}
301: with $V_{\rm int}=c_0|\phi_{c}(\vec{r})|^2$, or
302: \begin{eqnarray}
303: V_{\rm int}&=&(c_0-c_2)|\phi_{c}(\vec{r})|^2\nonumber\\
304: &&-c_d\int d\vec r'{1\over |\vec r-\vec r'|^3}(1-3\cos^2\theta)|\phi_{c}(\vec{r}')|^2, \hskip 12pt
305: \end{eqnarray}
306: respectively, for the internal states $|0,0\rangle$ or
307: $|1,-1\rangle$. $\mu_c$ is the chemical potential, and $\theta$ is
308: the angle between the $z$-axis and $\vec r-\vec r'$. A Gaussian
309: variational approach is proven to be adequate within the parameter
310: regions of interest \cite{yi1,santos,blume}. Like the $\phi_0$
311: approximation, atom-atom correlation \cite{ceder} is neglected
312: because of the use of product motional states here.
313:
314: We have used the momentum space pseudo-potential \cite{Derevianko}
315: $v(\vec{k},\vec{k}') = -{\hbar^2 /(2\pi^2
316: m)}a_{sd}'\sqrt{5}[P_2(\mathrm{cos}\theta_{k'}) + ({k/k'})^2
317: P_2(\mathrm{cos}\theta_{k}) ]$ with $a_{sd}' = {\sqrt{2}mc_d/(
318: 12\sqrt{5}}{\hbar^2})$ for a more accurate evaluation of the
319: dipolar term in Eq. (\ref{Ey}), which can differ upto 50\% for the
320: parameter range reported here. Because the complete motional wave
321: function is Gaussian shaped, we find that $\langle
322: \phi_{1,-1}|H_{dd}|\phi_{0,0}\rangle$ can be evaluated
323: analytically. In the results shown below for two
324: $^{87}\mathrm{Rb}$ atoms, the spin mixing effective Rabi frequency
325: is defined as $\Omega_{\rm eff}=\sqrt{\Omega^2+\Delta^2}$, and
326: $f_{\rm eff}=\Omega_{\rm eff}/2\pi$. $\Delta$ is the bare energy
327: difference: $\Delta\equiv (E_{1,-1}-E_{0,0})/\hbar$.
328:
329: \begin{figure}[h] \centering
330: \includegraphics[width=3.00in]{fig1.eps}
331: \caption{(Color online) The aspect ratio dependence of $f_{\rm
332: eff}$ computed with the three approximation schemes as labelled.
333: The three solid lines include dipolar interaction, while the
334: dot-dashed lines are the corresponding ones without dipolar
335: interaction. Here we fix
336: $(\omega_\rho^2\omega_z)^{1/3}=(2\pi)41.1$(kHz).} \label{fig1}
337: \end{figure}
338:
339: In Fig. \ref{fig1}, we show the dependence of $f_{\rm eff}$ on the
340: trap aspect ratio $\lambda$ at an external magnetic field of
341: $B=0.1$ (Gauss). To facilitate a fair comparison, we have fixed
342: the geometric average of the trap frequencies
343: $(\omega_\rho^2\omega_z)^{1/3}$ to the spherical trap of
344: $\omega_{\rho}=\omega_z=(2\pi)41.1$(kHz) \cite{Bloch1}. Somewhat
345: surprising at first sight is the noticeable quantitative
346: differences (within experimental sensitivity) among the different
347: approximations. At $\lambda=1$ for a spherical trap, our results
348: also differ from the experimental observations \cite{Bloch1}. We
349: note that the harmonic approximation of $V_0\sin^2(kx)$ by
350: $m\omega_\rho^2x^2/2$ introduces about a $5\%$ error \cite{artur}.
351: Although the spread among the different approximation schemes
352: calls for a more accurate treatment for the motional state, the
353: dipolar effect due to the $H_{dd}$ term in Eq. (\ref{feq}) is
354: quite accurately reproduced to less than $1\%$ \cite{note}. An
355: improved treatment of the relative motional wave function is also
356: needed inside a cylindrical trap, if spin mixing is used to
357: calibrate atomic interactions like what has been accomplished for
358: a spherical harmonic trap \cite{artur}.
359: \begin{figure}[h]
360: \centering
361: \includegraphics[width=3.00in]{fig2.eps}
362: \caption{(Color online) The B-field dependence of $f_{\rm eff}$
363: computed within $\phi^{(v)}$ approximation for $\lambda=3$. Solid
364: line includes dipolar interaction while the dot-dashed line
365: excludes dipolar interaction. Here, unlike in Fig. \ref{fig1}, we
366: fix $\omega_{\rho}=(2\pi)41.1$(kHz).} \label{fig2}
367: \end{figure}
368:
369: Based on the current experimental sensitivity, dipolar effects
370: should be detectable for $\lambda>3$ and can constitute a more
371: than $10\%$ increase in $f_{\rm eff}$. They are minimized for a
372: spherical trap as shown in Fig. \ref{fig2} with an actual shift of
373: about $10^{-4}$ or less for the experiment of Ref. \cite{artur}.
374:
375: Also we have studied the $F=2$ manifold \cite{zgm}, where spin
376: mixing dynamics is generally dominated by three two-atom internal
377: states $|0,0\rangle$, $|1,-1\rangle$, and $|2,-2\rangle$ at zero
378: magnetization \cite{Bloch1,artur}.
379: We find that spin dipolar effect remains observable in the
380: frequencies of the various mixing channels.
381: More interesting is the two-state mixing
382: channel of $|-1,-1\rangle\leftrightarrow |0,-2\rangle$ with a
383: nonzero magnetization as shown in Fig. \ref{fig3}. Inside an
384: oblate trap, dipolar interaction is dominantly repulsive in state
385: $|-1,-1\rangle$ in contrast to the attractive state $|1,-1\rangle$
386: of the $F=1$ case. At a weak magnetic field and for $\lambda=3$,
387: we find the dipolar interaction constitutes a 25 Hz downward shift
388: computed within the $\phi^{(v)}$ approximation.
389:
390: \begin{figure}[h]
391: \centering
392: \includegraphics[width=3.00in]{fig3.eps}
393: \caption{(Color online) The same as in Fig. \ref{fig2} except for
394: spin mixing between $|-1,-1\rangle$ and $|0,-2\rangle$ for $F=2$.
395: } \label{fig3}
396: \end{figure}
397:
398: Dipolar interaction is normally stronger by a factor of two
399: in magnitude along the direction of the dipoles
400: (favored by the $\lambda<1$ geometry)
401: in comparison to the perpendicular direction (favored when $\lambda>1$).
402: Our results on the spin mixing frequency, however, reveals a
403: completely opposite trend; we find relatively larger (smaller)
404: dipolar effects for $\lambda>1$ ($<1$).
405: This can be easily understood. For the $F=1$ case,
406: the dipolar term $H_{dd}$ is positive (negative)
407: for $\lambda<1$ ($>1$), thus destructively (constructively) add to the
408: (negative) $c_2$ term in Eq. (\ref{feq}). For the $F=2$ case,
409: the same reasoning applies despite of the opposite dipolar
410: interaction in state $|-1,-1\rangle$. The dipolar effect becomes
411: constructively enhanced for $\lambda>1$ because the $c_2$
412: is positive in $F=2$.
413:
414: Before conclusion, we note that we have also calculated the spin
415: mixing frequency for two $^{52}$Cr atoms with $F=3$. In the limit
416: of very large ($\lambda =10$) and very small aspect ratios
417: ($\lambda =1/10$), dipolar interaction causes about an 8\%
418: difference in the spin mixing frequency, which is about four times
419: smaller than in $F=1$ of $^{87}$Rb at $\lambda =10$, and close to
420: each other at $\lambda =1/10$. Similar conclusions hold for
421: $^{87}$Rb in $F=2$. It is in this sense we say that the dipolar
422: effect of $^{52}$Cr is weaker than $^{87}$Rb. The absolute
423: frequency difference due to dipolar interaction, however, is
424: larger in $^{52}$Cr because of the faster spin dynamics involved
425: due to the large spin exchange term in $^{52}$Cr.
426:
427: In summary, we have studied dipolar effects in spin mixing between
428: two atoms trapped in a single optical lattice site.
429: While this effect is small, and can be ignored completely
430: for spherical harmonic traps, we find it
431: is observable inside cylindrical harmonic traps,
432: especially for oblate shaped traps with $\lambda>3$.
433: We hope this study will
434: stimulate experimental efforts
435: aimed at observing dipolar effect in spin mixing.
436:
437: We thank Drs. K. Bongs, T. Pfau, L. Santos, and K. Sengstock for enlightening
438: discussions. This work is supported by CNSF and NSF.
439:
440:
441: \begin{thebibliography}{99}
442:
443: \bibitem{MSZhang}
444: M.-S. Chang {\it et al.}, Nature Phys. {\bf 1}, 111 (2005).
445:
446: \bibitem{HS}
447: J. Kronj\"ager {\it et al.}, Phys. Rev. A \textbf{72}, 063619 (2006).
448:
449: \bibitem{Bloch1}
450: A. Widera {\it et al.}, Phys. Rev. Lett. {\bf 95}, 190405 (2005);
451: F.Gerbier {\it et al.}, Phys. Rev. A {\bf 73}, 041602(R) (2006).
452:
453: \bibitem{jason}
454: T.-L. Ho, Phys. Rev. Lett. {\bf 81}, 742 (1998).
455:
456: \bibitem{machida}
457: T. Ohmi and K. Machida, J. Phys. Soc. Jpn. {\bf 67}, 1822 (1998).
458:
459: \bibitem{law}
460: C. K. Law {\it et al.}, Phys. Rev. Lett. {\bf 81}, 5257 (1998).
461:
462: \bibitem{kurn}
463: D. M. Stamper-Kurn {\it et al.}, Phys. Rev. Lett. {\bf 80}, 2027
464: (1998); J. Stenger {\it et al.}, Nature (London) {\bf 396}, 345
465: (1998).
466:
467: \bibitem{eric}
468: D. S. Hall {\it et al.}, Phys. Rev. Lett. {\bf 81}, 1543 (1998).
469:
470: \bibitem{zoller}
471: D. Jaksch {\it et al.}, Phys. Rev. Lett. {\bf 81}, 3108 (1998).
472:
473: \bibitem{Bloch}
474: M. Greiner {\it et al.}, Nature (London) {\bf 415}, 39 (2002).
475:
476: \bibitem{dipyou}
477: M. Marinescu and L. You, Phys. Rev. Lett. {\bf 81}, 4596 (1998).
478:
479: \bibitem{yi1}
480: S. Yi and L. You, Phys. Rev. A {\bf 61}, 041604(R)
481: (2000); {\it ibid} {\bf 63}, 053607 (2001).
482:
483: \bibitem{goral1}
484: K. Goral {\it et al.}, Phys. Rev. A {\bf 61}, 051601(R) (2000);
485: J.-P. Martikainen {\it et al.}, {\it ibid} {\bf 64}, 037601
486: (2001).
487:
488: \bibitem{santos}
489: L. Santos {\it et al.}, Phys. Rev. Lett. {\bf 85}, 1791 (2000); {\it ibid}
490: {\bf 88}, 139904(E) (2002).
491:
492: \bibitem{giov} S. Giovanazzi {\it et al.}, Phys.
493: Rev. A {\bf 63}, 031603(R) (2001).
494:
495: \bibitem{pfau}
496: J. Stuhler {\it et al.}, Phys. Rev. Lett. {\bf 95}, 150406 (2005).
497:
498: \bibitem{artur}
499: A. Widera {\it et al.}, cond-mat/0604038.
500:
501: \bibitem{pu}
502: S. Yi {\it et al.}, Phys. Rev. Lett. {\bf 93}, 040403 (2004).
503:
504: %\bibitem{ho}
505: %R. B. Diener and T.-L Ho, Phys. Rev. Lett. {\bf 96}, 190405
506: %(2006).
507:
508: \bibitem{jila}
509: In this case, it is a challenge to calibrate exactly where the
510: zero of the scattering length is. (E. Cornell and C. Wieman,
511: private communications, 2000).
512:
513: \bibitem{dipsantos}
514: L. Santos and T. Pfau, Phys. Rev. Lett. {\bf 96}, 190404 (2006).
515:
516: \bibitem{ueda}
517: Y. Kawaguchi {\it et al.}, Phys. Rev. Lett. {\bf 96}, 080405
518: (2006).
519:
520: \bibitem{ho}
521: R. B. Diener and T.-L Ho, Phys. Rev. Lett. {\bf 96}, 190405
522: (2006).
523:
524: \bibitem{coef}
525: J. P. Burke, Jr. and J. L. Bohn, Phys. Rev. A {\bf
526: 59}, 1303 (1999); N. N. Klausen {\it et al.},
527: {\it ibid} {\bf 64}, 053602 (2001); E. G. M. van Kempen {\it et al.}, Phys. Rev.
528: Lett. {\bf 88}, 093201 (2002).
529:
530: \bibitem{sr}
531: S. Hensler {\it et al.}, Appl. Phys. B {\bf 77}, 765 (2003); Z.
532: Pavlovic {\it et al.}, Phys. Rev. A {\bf 71}, 061402(R) (2005).
533:
534: \bibitem{blume}
535: D. C. E. Bortolotti {\it et al.}, cond-mat/0604432.
536:
537: \bibitem{ceder}
538: S. Klaiman {\it et al.},
539: Phys. Rev. A {\bf 73}, 013622 (2006).
540:
541: \bibitem{Derevianko}
542: A. Derevianko, Phys. Rev. A \textbf{67}, 033607 (2003); {\it ibid}
543: {\bf 72}, 039901(E) (2005).
544:
545: \bibitem{note}
546: We have also compared these variational calculations
547: with a more complete approach using an
548: exact solution basis for two atoms interacting through a contact
549: pseudo-potential inside a harmonic trap \cite{Wilken,cd}.
550:
551: \bibitem{zgm}
552: H.-J. Huang and G.-M. Zhang, cond-mat/0601188.
553:
554: \bibitem{Wilken}
555: T. Busch {\it et al.}, Found. Phys. \textbf{28}, 549 (1998).
556:
557: \bibitem{cd}
558: Z. Idziaszek and T. Calarco, Phys. Rev. A \textbf{71}, 050701(R)
559: (2005).
560:
561: \end{thebibliography}
562:
563:
564:
565: \end{document}
566: