1: \documentclass[twocolumn,showpacs,aps,prl,a4paper]{revtex4}
2: \usepackage{amsmath,graphicx}
3:
4: \begin{document}
5:
6: %\documentstyle[aps,twocolumn]{revtex}
7: %\documentstyle[aps,preprint]{revtex}
8: \title{Jaynes-Cummings dynamics with a matter wave oscillator}
9: \author{Klaus M\o lmer}
10: \affiliation{ QUANTOP, Danish National Research Foundation Center for
11: Quantum Optics,
12: \\Department of Physics and Astronomy,
13: University of Aarhus \\
14: DK-8000 \AA rhus C, Denmark}
15:
16: \begin{abstract}
17: We propose to subject two Bose-Einstein condensates to a periodic potential,
18: so that one condensate undergoes the Mott insulator transition to a state
19: with precisely one atom per lattice site. We show that
20: photoassociation of heteronuclear molecules within each lattice site is
21: described by
22: the quantum optical Jaynes-Cummings Hamiltonian. In analogy with
23: studies of this Hamiltonian with cavity fields and trapped ions,
24: we are thus able to engineer quantum optical states of
25: atomic matter wave fields and we are able to reconstruct these states
26: by quantum state tomography.
27: \end{abstract}
28:
29: \pacs{03.75.Gg, 03.75.Mn, 42.50.Fx}
30:
31: \maketitle
32:
33: The Jaynes-Cummings model was introduced to describe the
34: resonant interaction between
35: an atom and a single mode of the quantized radiation field
36: \cite{Cummings}, and with
37: incorporation of pumping and relaxation mechanisms, it constitutes
38: the corner stone for the quantum treatment of maser and laser action
39: \cite{ScullyLambHaken}. The unitary Jaynes-Cummings dynamics has
40: been succesfully demonstrated in the micro maser \cite{Micromaser},
41: in which atoms in Rydberg excited
42: states are injected into a high-Q cavity with a single mode which is
43: resonant with an atomic transition.
44: This system has been used to demonstrate fundamental
45: aspects of quantum theory such as the oscillatory exchange of energy
46: between the atom and the field, sub-poissonian photon statistics,
47: entanglement, and decoherence \cite{QED,ENS-cat}.
48: Another implementation of the Jaynes-Cummings Hamiltonian is in the
49: description of the quantized motion of an ion in a harmonic trap
50: \cite{Ions}, for which a classical laser field
51: drives transitions in which the ion is excited and
52: simultaneously one quantum of energy is extracted from the motional state.
53: The ion system is characterized by very long lifetimes of the oscillator
54: state, and hence the Jaynes-Cummings dynamics has been studied in great
55: detail, and numerous examples of state synthesis and detection have been
56: presented \cite{Meekhof,Leibfried}.
57:
58: In a quantum optical perspective, what characterizes a Bose-Einstein
59: condensate is its single-mode character and coherence properties
60: \cite{Atomcoh}.
61: The bosonic commutator relation applies for atoms as well as for
62: photons and phonons, and if one neglects interactions,
63: many features of single mode
64: fields are thus common for atoms, light and harmonic motion.
65: This has of course led to
66: various proposals for generation of atomic states with properties
67: similar to the ones of light, e.g., phase and number squeezed
68: states and Schr\"odinger cat states.
69: By clever use of the atomic interactions as an adjustable
70: non-linear element, strong quantum effects are foreseen
71: \cite{Anders}, and in combination with the long storage
72: times for atoms, these effects are potentially much more useful,
73: for atoms than for light, and indispensable, e.g., for the
74: improvement of atomic clocks \cite{Clocks}.
75:
76: In this Letter we present a quantum optical analysis of
77: the dynamics of a pair of atomic condensates which have been
78: trapped in a periodic potential, so that one species has experienced
79: the Mott insulator transition to a state with exactly one atom
80: per lattice site \cite{Jaksch98,Munichdiff}.
81: The other condensate is assumed to have a different
82: interaction strength, and hence it may still be in
83: the superfluid phase with Poissonian number fluctuations on each site,
84: or it may have reduced fluctuations due to the on site repulsion.
85: A detailed presentation of this process was
86: given recently \cite{Hetero}.
87: When the optical lattice potential is raised the atomic wave functions
88: become more localized in the potential minima, and thus the
89: collisional interaction between atoms increases and the tunneling rate
90: between wells decreases. From the
91: time dependendent shape of the potential for the atoms one determines
92: the on-site interaction strengths
93: $U_a,\ U_b$ and $U_{ab}$ between atoms of the
94: same and different species $A$ and $B$
95: and the interwell tunneling couplings $J_a,\ J_b$, but for simplicity
96: we shall just assume
97: the values shown in the upper panel of Fig.1.
98: The ratio of approximately 0.6 between $U_a$ and $U_b$
99: corresponds to the ratio between the free space s-wave scattering
100: lengths of Rb and K atoms. The tunneling coupling of both species
101: are taken to follow the same lower curve.
102: These curves are in qualitative agreement with the
103: dependence of the interaction strength and tunnel coupling for
104: a single species evaluated in \cite{Jaksch02}, where the
105: energy and frequency unit $J_0$, on the order of a tenth of the atomic
106: recoil shift, is introduced ($\hbar=1$).
107: Using a symmetry breaking Gutzwiller ansatz, we assume that the
108: full many body system is in a product state over all sites, where each
109: site is described by the state vector
110: $|\psi\rangle=\sum_{n_A,n_B} a_{n_A,n_B}|n_A,n_B\rangle$,
111: with $n_i$ the number of atoms of species $i=A,B$.
112: We have solved the self-consistent equations for
113: the amplitudes $a_{n_A,n_B}$, and we have
114: verified that the two species undergo the Mott insulator transition
115: at different times, see also \cite{Hetero}. The lower panel of
116: Fig. 1. shows typical results of a numerical solution of the problem.
117: The solid lines
118: show the variance of the populations (the mean
119: values are unchanged and equal to the initial variance
120: where we assume Poisson statistics). The dashed curves are obtained
121: by neglecting the cross coupling between the two species, and
122: as in \cite{Hetero} we verify that this coupling has
123: little effect on the population statistics of the two species in the
124: lattice wells.
125:
126: \begin{figure}[htbp]
127: \includegraphics[width = 80 mm]{gutz1.eps}
128: \caption{Mott insulator transition for two atomic species.
129: The upper panel shows
130: the time dependent coupling constants and
131: the tunneling coupling, all in units of the constant $J_0$.
132: The lower panel shows the variance of populations of the two atomic
133: species. The dashed curves are obtained by neglect of the
134: cross coupling term $U_{ab}$.
135: .}
136: \end{figure}
137:
138: Numerous studies have addressed the
139: prospects of forming molecules from ultra-cold atoms both theoretically
140: and experimentally \cite{Molecules}.
141: It has been
142: suggested to trap the atoms in lattices with low occupation number,
143: and either make explicit use of the Mott insulator transition
144: \cite{Jaksch02,Hetero}, or simply work with low mean occupation of each
145: site \cite{Esslinger} to form the molecules in an optimally controlled
146: manner.
147: We suggest to build on these proposals and to drive
148: the photoassociation process on each
149: lattice site to form a heteronuclear molecule with a single atom
150: from each of the two species. We are more interested in the process than
151: in the final molecular state, and hence a sufficiently long lived excited
152: vibrational state will suffice as molecular state of the process.
153: The coherent photoassociation process is described by
154: the effective, second quantized, Hamitonian
155: \begin{equation}
156: H= \chi a_Aa_Ba^\dagger_M + \chi^*a_A^\dagger a_B^\dagger a_M
157: \label{ham}
158: \end{equation}
159: where $a_i$ is the annihilation operator of atoms and molecules $(i=A,B,M)$.
160: When restricted to the states with initially precisely one B atom,
161: we can rewrite this Hamiltonian as
162: \begin{equation}
163: H= \chi a_A\sigma^\dagger + \chi^*a_A^\dagger \sigma,
164: \label{hamjc}
165: \end{equation}
166: where the Pauli spin lowering operator $\sigma=a^\dagger_Ba_M$ represents the
167: two-state transition operator
168: $|1_B,\ 0_M\rangle\langle 0_B,\ 1_M|$, ($\sigma$ turns an AB molecule
169: into a B atom, i.e. , it does not conserve the number of A atoms,
170: and it appears only in combination with the creation operator
171: $a^\dagger_A$). The Hamiltonian couples pairs of levels
172: \begin{equation}
173: |n_A,\ 1_B,\ 0_M\rangle \leftrightarrow
174: |(n-1)_A,\ 0_B\ 1_M\rangle,
175: \label{pa}
176: \end{equation}
177: and it thus follows that there will be an oscillatory exchange of the
178: single B atom between being part of the molecule and being an
179: independent atom, in complete analogy with the Jaynes-Cummings
180: transfer of excitation between a two-level system and an oscillator
181: mode.
182:
183: This oscillation
184: will undergo collapses and revivals, because the coupling strength
185: betwen the two mentioned states is proportional to $\sqrt{n_A}$,
186: and hence the different number state components of A atoms
187: will give away an atom to the association process
188: at different frequencies. An experiment
189: that counts the total number of B atoms as a function of duration
190: of the photoassication process, can thus teach us about the number
191: statistics of the A atoms!
192:
193: So far, the Mott insulator transition has been demonstrated
194: experimentally by the loss of interference between different
195: lattice sites, and by a gap in the excitation spectrum
196: \cite{Munichdiff}. The Jaynes-Cummings dynamics presents a unique
197: application of a quantum optical method for the study of this
198: many-body problem. We suggest to turn on slowly the
199: lattice potential, so that the B atoms undergo the transition to the Mott
200: phase at a time where the A atoms are still in an unknown state. At
201: this time we hold the lattice potential and we turn on the
202: photoassociation process, i.e., the dynamics
203: driven by the Jaynes-Cummings Hamiltonian (\ref{hamjc}).
204: The probability to find a B atom on any lattice site
205: is given by the square of the Rabi oscillation amplitudes, weighted by the
206: $n_A$ probability distribution $p(n_A)$:
207: \begin{equation}
208: \langle n_B\rangle =\sum_{n_A} p(n_A) \cos^2(\chi\sqrt{n_A} t)
209: \label{collapse}
210: \end{equation}
211: This function is illustrated in Fig. 2. which shows the mean value of
212: $n_B$ as a function of time in case of coupling to a Poissonian
213: distribution with on average two A atoms per lattice site (upper panel),
214: a sub-Poissonian distribution of A atoms achieved at $J_0t=-50$, cf. Fig 1,
215: during application of the periodic potential (middle panel), and an
216: $n_A=2$ (Mott insulator) number state (lower panel).
217: For time sequences like the ones in Fig.2, one can use
218: Eq.(\ref{collapse}) to obtain a fit, which produces the complete
219: $p(n_A)$ distribution.
220: By counting the total number of B atoms, without requiring the
221: experimental ability to detect atoms with high efficiency on a
222: microscopic spatial scale, we thus obtain the precise
223: number statistics at each site of the A atoms.
224: A similar procedure was initially applied for trapped ions \cite{Meekhof},
225: where the internal state of the ion was used to monitor the population
226: of the harmonic oscillator levels of motion. Many repeated experiments
227: on the same system were necessary to probe the state of the ion at
228: different durations $t$ of the dynamics and with good precision.
229: We also have to operate the process many times with different durations
230: of the photoassociation process, but since the experiment 'repeats
231: itself' in parallel in every lattice site,
232: determination of the total number of B atoms in
233: the entire system suffices to give the mean value of $n_B$.
234: To take into account both deviations from the
235: Lamb-Dicke limit and decoherence, in Ref.\cite{Meekhof} the ideal
236: Jaynes-Cummings dynamics (\ref{collapse}) was replaced by
237: a more general formula
238: \begin{equation}
239: \langle n_B\rangle =\sum_{n_A} p(n_A) \cos^2(\Omega_{n_A} t)
240: e^{-\gamma_{n_A} t}
241: \label{collapseW}
242: \end{equation}
243: with fitting parameters $p(n_A)$, $\Omega_{n_A}$ and $\gamma_{n_A}$.
244: Similar extra handles on the theory may be useful in our matter-wave
245: analog where, e.g., atom-atom interactions may cause an
246: $n_A$-dependence of the atomic wave function and hence of the
247: photo-association coupling rate, and where various
248: mechanisms, such as inhomogeneities over the size of the lattice,
249: may cause decoherence.
250:
251: \begin{figure}[htbp]
252: \includegraphics[width = 80 mm]{gutz2.eps}
253: \caption{Time dependence of the mean number of B atoms per
254: lattice site during photoassociation/-dissociation. The upper panel
255: shows the results for a Poissonian
256: distribution with an average of two A atoms per site; the middle
257: panel shows the results for the distribution of A atoms at
258: $J_0t=-50$ in Fig. 1, where the Mott insulator transition
259: has not yet taken fully place for the A atoms; and, the lower panel
260: shows the results for an $n_A=2$ number state.}
261: \end{figure}
262:
263: In case one can only effectively control the Mott-insulator transition
264: of one species (the A atoms), we note that for a sufficiently low
265: density of B atoms, every lattice site will be occupied by a single or
266: none of these atoms,
267: and hence the mean value of $n_B$ will ocillate in the same
268: way as in (\ref{collapse},\ref{collapseW}) but multiplied with the
269: mean number of B atoms per site.
270:
271: As mentioned in the introduction of the Letter, the Jaynes-Cummings
272: Hamiltonian has been applied extensively in quantum optics, and the
273: range of interesting phenomena studied is very wide.
274: Let us turn to the possibilities for synthesis and studies of
275: particular quantum
276: states of matter by means of the effective Jaynes-Cummings Hamiltonian.
277: Studies of the collapse and revival of the diffraction pattern of
278: lattice trapped atoms have already been made \cite{Munichrevival,
279: Yalediff},
280: and we propose to use the Jaynes-Cummings dynamics to produce a
281: Schr\"odinger cat like state of the A atoms, and to probe its coherence
282: properties. The photoassociation process can be driven off resonance
283: with a detuning $\delta$ in which case
284: the two-state transition (\ref{pa}) does not take place, but the coupling
285: (\ref{hamjc}) perturbs the energy of the state
286: $|n_A, 1_B\rangle$ by the amount $\Delta E_{n_A}=
287: |\chi|^2n_A/\delta$. If the A atoms are in a coherent state
288: $|\alpha\rangle$, this state will
289: undergo a simple phase rotation,$|\alpha\rangle \rightarrow
290: |\alpha e^{i\phi}\rangle$
291: due to this coupling. Imagine now
292: that the B atom has a level structure, so that it can be initally
293: prepared in a superpostion of the active state $|B\rangle$
294: which experiences the
295: coupling discussed above, and a passive state $|B'\rangle$
296: for which the photoassociative coupling vanishes completely. This
297: implies that the Jaynes-Cummings Hamiltonian produces a material
298: Schr\"odinger cat like state of the two species,
299: \begin{equation}
300: |\Psi\rangle = a|\alpha e^{i\phi}\rangle\otimes |B\rangle
301: +b|\alpha\rangle\otimes|B'\rangle,
302: \label{cat}
303: \end{equation}
304: very analogous to the production of a similar state in cavity QED
305: experiments \cite{ENS-cat}.
306: By driving the transition between the active and the passive level
307: in the B atom, we can subsequently monitor the interference of
308: the two coherent components. The production of a mesoscopic superposition
309: state can thus be verified experimentally.
310: In a gauge invariant formulation of the problem, there is of course not
311: a coherent state and mean atomic field in a single site, but a
312: total number state of
313: the entire atomic ensemble may be well approximated by a coherent state,
314: and when this is split into a product of coherent states on each
315: lattice site, a total number projection operator produces a very entangled
316: state of the atomic components on each site \cite{Esslinger}, but physical
317: observables will be correctly handled by the coherent state ansatz
318: \cite{Klaus}.
319:
320: The cavity QED implementation of the Jaynes-Cummings
321: Hamiltonian allows for classical excitation of both the
322: two-level system and the oscillator, and it operates
323: experimentally with the injection of a beam of
324: excited two-level atoms, who can give off their excitation energy to
325: the oscillator degree of freedom. In the ion trap
326: implementation, both the oscillator and the two-level
327: system can be excited independently, and by changing the coupling
328: frequency an ``anti" Jaynes-Cummings Hamiltonian
329: $\chi a \sigma + \chi^* a^\dagger \sigma^\dagger$, and a
330: ``two-phonon" Hamiltonian $\chi a^2 +\chi^* (a^\dagger)^2$ can be
331: implemented to produce various non-classical states \cite{Meekhof}.
332: Tomographic reconstruction of the motion of a trapped
333: ion has been demonstrated in \cite{Leibfried}. That measurement scheme
334: made use of a classical driving to displace the oscillator by the complex
335: amplitude $\alpha$ and subsequently the populations of the various
336: oscillator states were probed by the two-level population oscillations
337: (\ref{collapse},\ref{collapseW}) as described above.
338: The position-momentum Wigner function was then obtained by an
339: analytical formula \cite{Leibfried} from the
340: populations recorded with different displacement $\alpha$.
341:
342: It will add an experimental level of difficulty to
343: couple atoms coherently from a reservoir condensate into the
344: lattice potentials, but with the current progress in transfer and loading
345: of degenerate gasses into microfabricated structures \cite{Micro}, some
346: possibilities are certainly open.
347: Without continuous loading of the lattice, there is another possibility
348: to mimic the quantum optical effects driven by external, classical fields.
349: We illustrate this with a protocol for full quantum state tomography
350: of the matter wave field of the A atoms:
351: Since we cannot accomplish coherent displacements of the atomic field
352: we suggest to use A atoms with two internal states, the active state
353: $|A\rangle$ and a passive state $|A'\rangle$ which is not affected
354: by the photoassication process. We can then use the $|A'\rangle$ population
355: as a 'local oscillator', and as long as the A'-population is large,
356: coherent transfer of atoms between the two atomic
357: states is well described by a coherent displacement of the A atom oscillator.
358: In fact, this internal state
359: coupling is correctly described by means of a rotation of the Dicke
360: collective spin of the atoms, and the detection of the population
361: statistics of the $|A\rangle$-state atoms by means of the
362: Jaynes-Cummings dynamics can be used for perfect collective spin
363: tomography: If there is a total of $N$ atoms populating states
364: $|A\rangle$
365: and $|A'\rangle$ in every lattice site, it takes $4N+1$ effective
366: spin rotations followed by population measurements to reconstruct
367: the full collective spin density
368: matrix, following the protocol applied to large hyperfine
369: manifolds in Ref.\cite{Jessen}.
370:
371: In summary, we have suggested how to realize a purely atomic
372: Jaynes-Cummings Hamiltonian and how to use it both to
373: generate quantum optical states of matter and to study various
374: many-body properties of the system. To isolate the
375: different effects and mechanisms, we did not present studies of the
376: joint effect of interactions, inter-well tunneling and the
377: photoassociation process. It is clear that these effects in
378: combination may provide wide possibilities
379: for detection and production of quantum states.
380: It has been estimated in \cite{Jaksch02, Hetero} that one can reliably
381: transfer atoms coherently between the trapped ground states of the
382: system, but it is of course also an interesting possibility to use
383: the resonance conditions for the photoassociation lasers to
384: excite and study collective excitations of the atoms.
385: We have focussed on lattice trapped atoms, where the
386: coherent Jaynes-Cummings dynamics is most readily realized, and
387: where the
388: existence of many replica of the same quantum systems provides effective
389: means for detection of generic quantum optical behavior of the matter wave
390: fields. Within this framework, one can also consider multiple species
391: and multi-level systems in analogy with multi-mode and multi-level
392: generalizations of the Jaynes-Cummings Hamiltonian.
393: The dynamics is not restricted to lattices, and as an extension
394: to the work we imagine that a collection of distinguishable, e.g.,
395: spatially localized B atoms immersed in a larger trapped condensates
396: of A atoms can serve to
397: induce a host of interesting effects in analogy with the more
398: general interaction of light and two-level atoms. An appealing
399: possibility would, for example, be to look for a
400: matter wave analogue of the Mollow fluorescence triplet,
401: when an atomic laser beam interacts with an ensemble of B
402: atoms and molecules can dissociate to atomic states with higher
403: and lower energy than the energy of the incident beam.
404:
405: Comments on the manuscript by Michael Budde are gratefully acknowledged.
406:
407: \begin{thebibliography}{99}
408:
409: \bibitem{Cummings} F. W. Cummings, Phys. Rev. {\bf 140}, A1051 (1965).
410:
411: \bibitem{ScullyLambHaken} M. Sargent III, M. O. Scully and W. E. Lamb, Jr,
412: {\it Laser Physics}, Addison-Wesley, Reading, Massachussetts, (1974).
413:
414: \bibitem{Micromaser} D. Meschede {\it et al}, Phys.
415: Rev. Lett. {\bf 54}, 551 (1985); P. Filipowicz {\it et al},
416: Phys. Rev. A {\bf 34}, 3077 (1986).
417:
418: \bibitem{QED} G. Rempe {\it et al},
419: Phys. Rev. Lett.
420: {\bf 58}, 353 (1987); M. G. Raizen {\it et al}, Phys. Rev. Lett. {\bf
421: 63}, 240 (1989); M. Brune {\it et al}, Phys. Rev. Lett. {\bf 65}, 976
422: (1990).
423:
424: \bibitem{ENS-cat} M. Brune {\it et al}, Phys. Rev. Lett. {\bf 77}, 4887
425: (1996).
426:
427: \bibitem{Ions} J. I. Cirac, {\it et al},
428: Phys. Rev. Lett. {\bf 70}, 762 (1993).
429:
430: \bibitem{Meekhof} D. M. Meekhof, {\it et al}, Phys. Rev. Lett. {\bf 76},
431: 1796 (1996).
432:
433: \bibitem{Leibfried} D. Leibfried, {\it et al}, Phys. Rev. Lett. {\bf
434: 77}, 4281 (1996).
435:
436: \bibitem{Atomcoh} M. Kasevich, Science {\bf 298}, 136, november (2002).
437:
438: \bibitem{Anders} A. S\o rensen {\it et al}, Nature {\bf 409}, 63 (2001).
439:
440: \bibitem{Clocks} G. Santarelli, {\it et al}, Phys. Rev. Lett. {\bf 82},
441: 4619 (1999).
442:
443: \bibitem{Jaksch98} D. Jaksch et al., Phys. Rev. Lett. {\bf 81}, 3108
444: (1998).
445:
446: \bibitem{Munichdiff} M. Greiner, {\it et al},
447: Nature, {\bf 415}, 39-44 (2002).
448:
449: \bibitem{Hetero} B. Damski {\it et al}, cond-mat/0208375.
450:
451: \bibitem{Jaksch02} D. Jaksch, et al.,
452: % V. Venturi, J. I. Cirac, C. J. Williams, and P. Zoller,
453: Phys, Rev. Lett. {\bf 89}, 040402 (2002).
454:
455: \bibitem{Molecules} R. Wynar, {\it et al},
456: Science {\bf 287}, 1016 (2000);
457: M. Mackie et al., Phys. Rev. Lett. {\bf 84}, 3803
458: (2000); J. Javanainen and M. Mackie, Phys. Rev. Lett. {\bf 88}, 90403
459: (2002);
460: C. McKenzie et al. Phys. Rev. Lett. {\bf 88}, 120403
461: (2002);
462: S.J.J.M.F. Kokkelmans, and M.J. Holland, cond-mat/0204504;
463: T. Koehler, T. Gasenzer, and K. Burnett, cond-mat/0209100;
464: E. Donley et al, Nature {\bf 417}, 529 (2002).
465:
466: \bibitem{Esslinger} T. Esslinger and K. M\o lmer,
467: cond-mat/0210324.
468:
469: \bibitem{Klaus} K. M\o lmer, Phys. Rev. A {\bf 55}, 3195 (1997);
470: J. Mod. Opt. {\bf 44}, 1937 (1997).
471:
472: \bibitem{Munichrevival} M. Greiner et al,
473: Nature, {\bf 419}, 51-54 (2002).
474:
475: \bibitem{Yalediff} C. Orzel et al., Science {\bf 291}, 2386 (2001).
476:
477: \bibitem{Micro} W. H\"ansel et al, Nature {\bf 413}, 498 (2001);
478: H. Ott, et al, Phys. Rev. Lett. {\bf 87},
479: 230401 (2001); A. E. Leanhardt, Phys. Rev. Lett. {\bf 89}, 040401 (2002).
480:
481: \bibitem{Jessen} G. Klose et al, Phys. Rev. Lett. {\bf 86}, 4721, (2001).
482:
483: \end{thebibliography}
484:
485: \end{document}
486: