1: \documentclass[aps,prl,twocolumn,groupedaddress]{revtex4}
2:
3: \usepackage{graphicx}
4:
5: \bibliographystyle{apsrev}
6:
7: \begin{document}
8: %\draft
9: %\preprint{{\bf ETH-TH/98-??}}
10:
11: \title{Commensurate-incommensurate transition of cold atoms in an
12: optical lattice}
13:
14: \author{H.P.\ B\"uchler}
15: \author{G.\ Blatter}
16: \author{W.\ Zwerger}
17: \altaffiliation{ Present and permanent address: Sektion Physik,
18: Universit\"at M\"unchen, Theresienstr.\ 37, D-80333 M\"unchen,
19: Germany.}
20: \affiliation{Theoretische Physik, ETH-H\"onggerberg, CH-8093 Z\"urich,
21: Switzerland}
22:
23:
24: \date{\today}
25:
26: \begin{abstract}
27: An atomic gas subject to a commensurate periodic
28: potential generated by an optical lattice undergoes a
29: superfluid--Mott insulator transition. Confining a strongly
30: interacting gas to one dimension generates an instability
31: where an arbitrary weak potential is sufficient to pin the
32: atoms into the Mott state; here, we derive the corresponding
33: phase diagram. The commensurate pinned state may be detected
34: via its finite excitation gap and the Bragg peaks in the static
35: structure factor.
36: \end{abstract}
37:
38: %\pacs{03.75.Fi, 05.30.Jp, 32.80.Pj}
39:
40:
41: \maketitle
42:
43: Atomic gases are developing into the ultimately tunable laboratory
44: system allowing to study complex quantum phenomena
45: \cite{anglin02}. Recently, subjecting an atomic Bose-Einstein
46: condensate to an optical lattice, Greiner {\it et al.}
47: \cite{greiner02} have succeeded in tuning the system through a
48: quantum phase transition separating a superfluid (S) from a Mott
49: insulating (MI) phase. This 3D bulk transition involves weakly
50: interacting bosons and is well understood within the Bose-Hubbard
51: description \cite{fisher89,jaksch98}: the system turns insulating when the
52: on-site interaction energy $U$ becomes of the order of the hopping
53: energy $J$. This strong coupling transition is a result of
54: quenching the kinetic energy by a strong lattice potential.
55: Amazingly, by confining the atomic gas to one dimension, the
56: strong coupling limit can be reached {\it without} the optical
57: lattice: in 1D, the ratio $\gamma$ between the interaction- and
58: kinetic energies per particle scales inversely with the density
59: $n$ and thus it is the {\it low}-density limit which is
60: interacting strongly (Tonks gas) \cite{petrov00}. A new
61: instability then appears in the strongly interacting 1D quantum
62: gas: the superfluid groundstate in the homogeneous system turns
63: insulating in the presence of an arbitrarily weak commensurate
64: optical lattice and the S--MI transition changes to a transition of
65: the incommensurate--commensurate type. In this letter, we
66: analyze this new instability and derive the complete phase diagram
67: for the S--MI transition in the limits of both weakly and strongly
68: interacting gases. Remarkably, this goal can be achieved by a
69: mapping to two classic problems, the Bose-Hubbard model
70: (introduced in Ref.\ \onlinecite{jaksch98})
71: and the sine-Gordon problem describing the weakly- and strongly
72: interacting limits of the atomic gas, respectively. Below, we
73: first summarize the main results providing us with the phase
74: diagram shown in Fig.\ 1; we then proceed with a detailed analysis
75: of the strongly interacting Bose gas subject to a weak optical
76: potential, leading us to a proper understanding of the new
77: instability.
78:
79: %
80: \begin{figure}[hbtp] \label{Phasedia}
81: \includegraphics[scale=0.32]{PhaseDia_7s.eps}
82: \caption{Left: Schematic phase diagram illustrating the
83: superfluid and Mott insulating phases versus parameters
84: $\gamma$ (interaction), $V$ (optical potential), and $Q$
85: (commensuration).
86: Right: Critical amplitude $V_{c}$ versus
87: interaction $1/\gamma$ for the commensurate
88: situation with $Q=0$. Below $1/\gamma_{c}$,
89: an arbitrary weak potential $V$ drives the superfluid into
90: the pinned insulating state. The dashed line denotes the
91: asymptotic behavior near the critical point $1/\gamma_{c}$ as
92: determined from the sine-Gordon model,
93: while the dashed-dotted line derives from the Bose-Hubbard
94: criterion $U/J|_{\rm\scriptscriptstyle S-MI} \approx 3.84$.}
95: \end{figure}
96: %
97:
98: A weakly interacting atomic gas subject to an optical lattice is
99: well described by the Bose-Hubbard model, which starts from a
100: tight-binding model and takes the interaction between bosons into
101: account perturbatively; the hopping amplitude $J(V)$ and the
102: on-site interaction energy $U(V,\gamma)$ follow from the
103: underlying parameters of the atomic gas, the dimensionless
104: interaction parameter $\gamma$ and the amplitude $V$ of the
105: optical lattice. The phase diagram of the Bose-Hubbard model is
106: well known \cite{fisher89} and involves insulating Mott-lobes
107: embedded in a superfluid phase. In 3D, the mean-field analysis for
108: densities commensurate with the lattice provides the critical
109: parameter $U/J|_{\rm\scriptscriptstyle S-MI} \approx 5.8\, z$, in
110: good agreement with the experimental findings of Greiner {\it et
111: al.} \cite{greiner02} (here, $z$ denotes the number of nearest
112: neighbors). Going to 1D, fluctuations become important and
113: appreciably modify the mean-field result: numerical simulations
114: \cite{kuhner98,rapsch99} place the transition at the critical
115: value $U/J|_{\rm\scriptscriptstyle S-MI} = 2C \approx 3.84$. This result
116: is easily transformed into the $\gamma$--$V$ phase diagram of the
117: weakly interacting atomic gas, once the relations $J(V)$ and
118: $U(V,\gamma)$ to the Mott-Hubbard parameters are known; combining
119: $J \propto \exp[-2(V/E_r)^{1/2}]$ as derived from a
120: WKB-calculation, with the on-site repulsion $U \propto \gamma$,
121: the transition line $V_c(\gamma)$ (see Fig.\ 1) then derives from
122: the implicit equation
123: %
124: \begin{equation}
125: 4V/E_{r} = \ln^{2}\bigl[ 4\sqrt{2} \pi C
126: \left(V/E_{r} \right)^{1/2}/\gamma\bigr].
127: \end{equation}
128: %
129: Here, $E_{r}= \hbar^{2}k^{2}/2m$ is the recoil energy with $m$ the
130: boson mass and $k=2\pi/\lambda$ the wave vector of the light
131: generating the optical lattice $V(x) = V \sin^2(kx)$.
132: The dimensionless interac\-tion strength
133: $\gamma$ is defined via $\gamma= m g/\hbar^{2}n$, with $n$ the
134: density and $g$ the strength of the $\delta$-function interaction
135: potential; $g$ is related to the 3D scattering length $a_{s}$ and
136: the transverse confining frequency $\omega_{\perp}$ via $g = 2
137: \hbar \omega_{\perp} a_{s}$ \cite{olshanii98}.
138:
139: Increasing the interaction strength $\gamma$, the critical
140: amplitude $V_c$ of the optical lattice triggering the S--MI
141: transition decreases, see Fig.\ 1; the description of the atom gas
142: in terms of the Bose-Hubbard model breaks down and we have to look
143: for a new starting point. For a weak optical potential, a natural
144: choice is the 1D Bose gas with $\delta$-function interaction,
145: which resides in the strong coupling regime at small densities
146: $n$ \cite{petrov00}; the presence of the optical lattice is
147: taken into account perturbatively. The homogeneous 1D Bose gas
148: with $\delta$-function interaction has been solved exactly by Lieb
149: and Liniger \cite{lieb63}; the corresponding low-energy
150: physics is properly described in terms of a Luttinger liquid with
151: a parameter $K(\gamma)$ derived from the exact solution, see
152: below. Adding the optical potential $\propto V$, we arrive at the
153: sine-Gordon model; the critical value $K(\gamma,V) = 2(1+V/4E_r)$
154: separating the Mott insulating phase from the superfluid one
155: determines the phase line $V_c/E_{r} = (\gamma^{-1}-
156: \gamma_{c}^{-1})/5.5$, with the critical value $\gamma_{c}\approx
157: 3.5$. Combining the results of the Bose-Hubbard- and sine-Gordon
158: models we can complete the phase diagram for the commensurable
159: situation as shown in Fig.\ 1: most remarkable is the appearance
160: of a critical interaction strength $\gamma_c$ above which an
161: arbitrary weak optical lattice is able to pin the system into a
162: Mott insu\-lator state. The presence of this instability is due to
163: the closeness of the dilute 1D Bose liquid to
164: Wigner-crystallization. Tuning the system away from
165: commensurability with $Q \equiv 2\pi(n-2/\lambda) \neq 0$ the Mott
166: insulator survives up to a critical misfit $Q_c(V,\gamma)$, see
167: Fig.~1. The corresponding physics is similar to that of the
168: com\-men\-su\-rate-in\-com\-mensurate transition of adsorbates on
169: a per\-iodic substrate as studied by Pokrovsky and Talapov
170: \cite{pokrovsky79}.
171:
172: For a qualitative understanding of the transition in the strongly
173: interacting 1D Bose gas, it is useful to consider the limit
174: $\gamma \gg 1$, where the behavior is essentially that of an ideal
175: Fermi gas \cite{girardeau60}. A weak periodic potential
176: $V(x)=V\sin^{2}(k x)$, with a lattice constant such that an
177: integer number $i=1,2,\ldots$ of particles will fit into one unit
178: cell, gives rise to a single-particle band structure in which the
179: $i$ lowest bands are completely filled. The ground state for
180: noninteracting Fermions is then a trivial band insulator,
181: separated from excited states by an energy gap which scales like
182: $(V/E_r)^i$ in the limit $V \ll E_{r}$. Similar to the Mott phase
183: in the Bose-Hubbard model, the insulating state has a fixed
184: integer density, commensurate with the lattice. It remains locked
185: in a finite regime of the chemical potential, characterizing an
186: incompressible state \cite{fisher89}. Clearly, for a weak periodic
187: potential, the lowest energy gap with $i=1$ is much larger than
188: the higher-order ones. In the following, we will thus confine
189: ourselves to studying the commensurate-incommensurate transition
190: near integer filling $i=1$, where the commensurate phase has
191: maximal stability.
192:
193: For a quantitative theoretical analysis, it is convenient to use
194: Haldane's description of the interacting 1D Bose gas in terms of
195: its long wave length density oscillations \cite{haldane81}.
196: Introducing the two fields $\phi(x)$ and $\theta(x)$ describing
197: phase and density fluctuations, the Hamiltonian of the homogeneous
198: gas (without longitudinal confining trap and periodic potential)
199: is a quadratic form involving kinetic and interaction energies,
200: %
201: \begin{equation}
202: H_{0} = \frac{\hbar}{2\pi}
203: \int dx \left[ v_{J} (\partial_{x}\phi)^{2}
204: + v_{N} (\partial_{x}\theta- \pi n)^{2}\right].
205: \label{hamiltonian}
206: \end{equation}
207: %
208: Here, $v_{J}=\pi \hbar n/m$ is the analog of the bare `Fermi'
209: velocity at given average density $n$ while $v_{N}=\partial_{n}
210: \mu/\pi \hbar$ is determined by the inverse compressibility,
211: giving a sound velocity $v_{s}=\sqrt{v_{J} v_{N}}$ consistent with
212: the standard thermodynamic relation $mv_{s}^{2} = n \partial_{n}
213: \mu$. For short-range repulsive interactions, the dimensionless
214: ratio $K=\beta^{2}/4\pi= v_{J}/v_{s}$ is larger than one,
215: approaching $\infty$ in the noninteracting case and unity in the
216: hard core limit \cite{haldane81}. In the idealized model with
217: $\delta$-function interactions, the exact solution by Lieb and
218: Liniger shows that $K$ is a monotonically decreasing function of
219: the ratio $\gamma= m g/\hbar^{2}n$ between the interaction and
220: kinetic energies; the limiting behavior for small values of
221: $\gamma$
222: %
223: \begin{equation}
224: K(\gamma \rightarrow 0)
225: = \pi [\gamma-(1/2 \pi) \gamma^{3/2}]^{-1/2}
226: \label{kgamma}
227: \end{equation}
228: %
229: follows from the Bogoliubov approximation in 1D. Surprisingly,
230: this result remains quantitatively correct for $\gamma$ values up
231: to $10$ \cite{lieb63}. At large $\gamma>10$, the asymptotic
232: behavior is $K(\gamma\rightarrow \infty)=(1+2/\gamma)^2$.
233:
234: For our subsequent analysis, it is convenient to introduce the
235: conjugate fields $\Theta(x)= (2/\beta)\left[\theta(x) - \pi n
236: x\right]$ and $\Pi(x)=- \hbar (\beta/2 \pi) \partial_{x} \phi(x)$
237: obeying the commutation relation $\left[ \Theta(x), \Pi(x')\right]
238: = i \hbar \delta(x-x')$. Using these fields,
239: %
240: \begin{equation}
241: H_{0} = \frac{\hbar v_{s}}{2}
242: \int dx \left[ \left(\Pi/\hbar\right)^{2}
243: + \left(\partial_{x} \Theta \right)^{2}\right]
244: \label{string}
245: \end{equation}
246: %
247: takes the form of the Hamiltonian for a 1D harmonic string with a
248: linear spectrum $\omega = v_{s} q$ describing the long wave length
249: density modulations of the interacting Bose gas. The assumption of
250: a linear spectrum inherent in the simple form (\ref{string}) is
251: valid only below a momentum cutoff $1/a \sim \pi n$
252: \cite{lieb63}; the choice of the length scale $a$ fixes the
253: energy scale of $H_0$. The Boson density operator $n(x)$ is
254: related to the field $\Theta(x)$ by \cite{haldane81}
255: %
256: \begin{equation}
257: n(x) = \biggl[n+\frac{\beta}{2 \pi} \partial_{x}\Theta\biggr]
258: \biggl[1 + 2 \sum_{l =1}^{\infty}
259: \cos\biggl( \frac{l\beta \Theta}{2}+l \pi n x\biggr)\biggr];
260: \label{densityoperator}
261: \end{equation}
262: %
263: the last factor accounts for the discrete nature of the particles.
264: Adding an external periodic potential with amplitude $V/2$ and
265: period $\lambda/2$ gives rise to the perturbation
266: %
267: \begin{equation}
268: H_{\rm \scriptscriptstyle V}
269: =\frac{V}{2} \int dx\: n(x) \cos \frac{4 \pi x}{\lambda}.
270: \end{equation}
271: %
272: As noted already by Haldane, insertion of the Fourier expansion
273: (\ref{densityoperator}) generates terms of the type appearing in
274: the quantum (1+1)-dimensional sine-Gordon theory
275: \cite{coleman75,gogolin98}. Close to commensurability the dominant
276: term arising from the lowest harmonic in (\ref{densityoperator})
277: has the conventional sine-Gordon form \cite{fisher89}
278: %
279: \begin{equation}
280: H_{\rm \scriptscriptstyle V} = \frac{V\:n}{2}\,
281: \int dx\,\cos \left[ \beta \Theta + Q x\right]
282: \end{equation}
283: %
284: with coupling parameter $\beta = 2 (\pi K)^{1/2}$ and a twist
285: $Q=2\pi(n-2/\lambda)$. The strength of the nonlinear $\cos \beta
286: \Theta$- perturbation is conveniently expressed through the
287: dimensionless parameter $u=\pi a^{2} n V/2\hbar v_{s}$ which
288: naturally involves the cutoff parameter $a$ \cite{kehrein99_01}.
289: The twist $Q$
290: vanishes at commensurability; away from commensurability the
291: finite twist $Q$ acts as a chemical potential for excitations and
292: is preferably incorporated into the free Hamiltonian
293: (\ref{string}) via the replacement $\partial_{x} \Theta
294: \rightarrow \partial_{x} \Theta-Q/\beta$.
295:
296: At fixed potential $V$, the quantum sine-Gordon model describes
297: the competition between the preferred average inter-particle
298: distance at given density due to the repulsive interaction and the
299: period imposed by the external potential. A perturbative
300: calculation (see Ref.~\onlinecite{gogolin98} for a review) tells
301: that for $\beta^{2}/4 \pi =K>2$ a weak periodic potential is
302: unable to pin the density; hence for $K>K_{c}=2$
303: ($\gamma<\gamma_{c} \approx 3.5$) the ground state remains gapless
304: and superfluid in the presence of a small-amplitude lattice
305: potential. In the strong coupling regime $K<2$, however, the atoms
306: are locked even to a weak lattice, as long as the twist $Q$ is
307: less than a critical value $Q_{c}$. Beyond that, there is a finite
308: density of `solitons' (or domain walls for adsorbates on a
309: periodic substrate, see Ref.\ \onlinecite{pokrovsky79}), which
310: interpolate between minima of the external potential, relieving
311: the frustration present at incommensurate densities $Q\neq0$. The
312: `solitons' behave like relativistic particles with energy $E_{q} =
313: \hbar v_{s} \sqrt{q^{2}+M^{2}}$ and reestablish the superfluid
314: response. The `mass' $M$ determines the excitation gap in the Mott
315: insulating state, which translates into a jump $\Delta \mu$ in the
316: chemical potential at the commensurate density: given that an
317: additional/missing atom involves $K$ solitonic excitations with
318: energy $E_{q=0}=\hbar v_{s} M$ \cite{japaridze84} one obtains
319: %
320: \begin{equation}
321: \Delta\mu =\frac{2\pi\hbar^{2}n}{m}\, M;
322: \end{equation}
323: %
324: furthermore, this mass is also simply related to the critical
325: twist via $Q_{c} =2 K^{2}M$. The precise numerical value of $M$
326: depends on the high momentum cutoff $1/a$ via the dimensionless
327: amplitude $u$ of the lattice potential. The free-fermion limit
328: $K=1$ fixes this cutoff at $1/a=\pi n$, resulting in the simple form
329: $u= V/4 E_{r}$; we ignore small corrections arising due to a
330: possible modification in the cutoff away from $K=1$. The dependence
331: of the mass $M$ on $u$ can be obtained from a recent nonperturbative
332: renormalization group analysis of the quantum sine-Gordon model by
333: Kehrein \cite{kehrein99_01}; for small values $V \ll E_{r}$
334: and $K$ away from $K_c$, the gap in the chemical potential takes
335: the form
336: %
337: \begin{equation}
338: \Delta\mu = 2 E_{r}\,\left[\frac{V}
339: {(2-K)4 E_{r}}\right]^{1/(2-K)}.
340: \label{mu_K}
341: \end{equation}
342: %
343: In the limit $K\rightarrow 1$, the size of the gap approaches
344: $V/2$, in agreement with the above fermionic picture of the Tonks
345: gas limit, where the appearance of an insulating ground state,
346: even for a weak periodic potential, is due to the opening of a
347: single-particle band gap at the Fermi energy. In the practically
348: accessible regime of large but finite $\gamma>\gamma_{c}$ the gap
349: depends on $V$ in the more complicated manner as given by
350: (\ref{mu_K}) and vanishes exponentially as $K$ approaches
351: $K_{c}=2$ \cite{kehrein99_01}, see Fig.\ 2(a). Similarly,
352: the density range $n-2/\lambda= \pm Q_{c}/2\pi$ over which the
353: ground state remains locked approaches zero as $K\rightarrow K_c$.
354: Finally, for $K>2$, the dependence of the critical interaction
355: parameter $K_c$ (or $\gamma_c$) on the lattice amplitude $V$
356: follows easily from the Kosterlitz-Thouless nature of the scaling
357: flow near $K_{c}=2$: to lowest order in $u$, $K_{c}(u)= 2(1+u)$.
358: Combining this result with (\ref{kgamma}) it is straightforward to
359: determine the line $V_c(\gamma)$ separating the gapped insulating
360: regime from the superfluid at small but finite values of $u$, see
361: Fig.\ 1.
362: %
363: \begin{figure}[hbtp]
364: \includegraphics[scale=0.38]{excitationgaps.eps}
365: \caption{(a) Size of the gap $\Delta \mu$ versus interaction
366: strength $\gamma$ for a fixed amplitude $V = E_{r}/2$.
367: For $\gamma \rightarrow \infty$ the gap assumes the
368: free fermion limit $V/2$, while it vanishes exponentialy
369: for $\gamma \rightarrow \gamma_{c}$.
370: (b) Fraction of atoms in the Mott insulating phase. The inset
371: shows the density distribution $n(x)$ with the Mott phase
372: characterized by a locked commensurate density in the trap
373: center, surrounded by a superfluid region.
374: \label{excitationgap}}
375: \end{figure}
376: %
377:
378: In order to analyze the consequences of the
379: commensu\-rate-incommensurate transition for cold atoms in a trap,
380: we consider a 1D Bose gas in the presence of a weak longitudinal
381: confining potential $V(x) = m \omega^{2}x^{2}/2$. Provided the
382: associated oscillator length $l = (\hbar/m\omega)^{1/2}$ is
383: much larger than the inter-particle distance, the density profile
384: in this inhomogeneous situation may be obtained from the Thomas-Fermi
385: approximation \cite{dunjko01}
386: %
387: \begin{equation}
388: \mu\left[n(x)\right] + V(x) = \mu[n(0)]
389: \label{TF}
390: \end{equation}
391: %
392: where $\mu[n]$ is the chemical potential of the homogeneous
393: system. The central density $n(0)$ and the associated radius
394: $R=\left(2\mu[n(0)]/m\omega^{2}\right)^{1/2}$ of the cloud is
395: obtained from the normalization condition
396: %
397: \begin{equation}
398: N= \int_{-R}^{R} dx \: n(x) = 2 R \int_{0}^{n(0)} dn
399: \sqrt{1-\frac{\mu[n]}{\mu[n(0)]}}
400: \label{Nvsmu}
401: \end{equation}
402: %
403: with $N$ the particle number, provided the relation $\mu[n]$ is
404: known explicitly.
405:
406: In the absence of an optical lattice the density profiles are
407: smooth functions of the coupling $\gamma$ \cite{dunjko01}. In the
408: limit $\gamma\gg1$, $\mu[n] \rightarrow \mu_{\rm\scriptscriptstyle
409: F} [n]=(\hbar \pi n)^{2}/2m$ approaches the Fermi energy of an
410: ideal Fermi gas with density $n$, resulting in a profile
411: $n(x)=(2N/\pi R)\sqrt{1-(x/R)^{2}}$ with radius $R = (2 N)^{1/2}
412: l$. Adding now a periodic potential which is nearly
413: commensurate with the density in the trap center opens a gap
414: $\Delta\mu$ in the chemical potential. The density profile will
415: then exhibit a flat, incompressible regime in the trap center. In
416: order to find the detailed shape of the density profile, we
417: approximate the sine-Gordon model in its strong coupling, gapped
418: phase by a gas of noninteracting relativistic Fermions of mass
419: $M$; in the relevant regime $1<K<2$, this approximation is known
420: to work extremely well \cite{kehrein99_01}. The chemical potential as
421: a function of density can then be calculated explicitly,
422: %
423: \begin{equation}
424: \mu[n]\approx \mu_{\rm\scriptscriptstyle F}[n]
425: +\frac{\Delta\mu}{2}
426: f\biggl(\frac{4KE_{r}}{\Delta\mu}\frac{\delta n}{n_c}\biggr)
427: \label{chempot}
428: \end{equation}
429: %
430: with $\delta n=n-n_{c}$ ($n_{c} = 2/\lambda$). The dimensionless
431: function
432: %
433: \begin{equation}
434: f(z)=\pm\left( 1+z^{2}\right)^{1/2}-z
435: \end{equation}
436: %
437: is discontinous at $z=0$ and incorporates, via (\ref{chempot}),
438: the jump $\Delta\mu$ in the chemical potential at $n_{c}$
439: associated with the incompressible commensurate
440: state. Using this approximation, the density profile and the
441: fraction of particles participating in the commensurate phase
442: derive from an integration of (\ref{Nvsmu}) with (\ref{chempot}),
443: see Fig.~\ref{excitationgap}(b).
444: Knowledge of the locked fraction
445: $\Delta N/N$ plays an important role in the experimental detection
446: of a commensurate Mott phase. This can be achieved by measuring
447: the excitation gap through a phase gradient method as done
448: previously for the Bose-Hubbard transition \cite{greiner02}.
449: Alternatively, it should be possible to directly observe the
450: increase in the long-range translational order in the Mott phase
451: via Bragg diffraction \cite{birkl95,weidemuller95};
452: in either case the fraction $\Delta
453: N/N$ determines the experimentally available signal. The latter
454: can be further enhanced by generating an array of parallel `atom
455: wires' with the help of a strong 2D optical transverse lattice.
456: Using numbers similar to those in the recent experiment by Greiner
457: {\it et al.} \cite{greiner02}, it is possible to generate several
458: thousand parallel 1D wires with a transverse confining frequency
459: $\nu_{\perp}=20$ kHz. A longitudinal harmonic trap with
460: frequency $\nu=40$ Hz then
461: encloses $N \approx 50$ atoms in each 1D wire; the associated
462: central density in the absence of a longitudinal periodic
463: potential is $n(0)=2~\mu\hbox{m}^{-1}$ for $\gamma \gg 1$,
464: commensurate with the lattice constant $\lambda/2 \approx 0.5~\mu$m
465: of a typical optical lattice \cite{greiner02}. A weak periodic
466: potential will then lead to an incompressible Mott state in the
467: center of the cloud, provided the parameter $\gamma={2a_{s}}
468: /{n(0)l_{\perp}^{2}}$ is larger than the critical value
469: $\gamma_{c} \approx 3.5$. For $^{87}$Rb with a scattering
470: length $a_{s}\approx
471: 5$ nm, the resulting $\gamma$ is equal to one, i.e., not quite in
472: the required range. As noted already by Petrov {\it et al.}
473: \cite{petrov00}, however, larger and in particular tunable values
474: of $\gamma$ may be realized by changing $a_{s}$ via a Feshbach
475: resonance as present, e.g., in $^{85}$Rb.
476:
477: In conclusion we have shown that a commensurate Mott state can be
478: realized in dilute 1D BEC's already with an arbitrary weak lattice
479: potential, provided that the ratio $\gamma$ between the
480: interaction and kinetic energies is larger than a critical value
481: $\gamma_{c}\approx 3.5$. This instability provides a new and
482: experimentally accessible tool for the quantitative
483: characterization of 1D atomic gases in the strongly correlated
484: `Tonks gas' limit. Also, the observation of a Mott state in a
485: regime where the atoms are not confined to discrete lattice sites
486: would give direct evidence for the granularity of matter in strongly
487: interacting dilute gases.
488:
489: It is a pleasure to acknowledge fruitful discussions with I.\
490: Bloch, T.\ Esslinger, M.\ Greiner and S.\ Kehrein. This work was
491: supported by the DFG Schwerpunkt {\it Ultrakalte Quantengase}.
492:
493: \begin{thebibliography}{22}
494: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
495: \expandafter\ifx\csname bibnamefont\endcsname\relax
496: \def\bibnamefont#1{#1}\fi
497: \expandafter\ifx\csname bibfnamefont\endcsname\relax
498: \def\bibfnamefont#1{#1}\fi
499: \expandafter\ifx\csname citenamefont\endcsname\relax
500: \def\citenamefont#1{#1}\fi
501: \expandafter\ifx\csname url\endcsname\relax
502: \def\url#1{\texttt{#1}}\fi
503: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
504: \providecommand{\bibinfo}[2]{#2}
505: \providecommand{\eprint}[2][]{\url{#2}}
506:
507:
508:
509: \bibitem[{\citenamefont{Anglin and Ketterle}(2002)}]{anglin02}
510: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Anglin}} \bibnamefont{and}
511: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
512: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{416}},
513: \bibinfo{pages}{211} (\bibinfo{year}{2002}).
514:
515: \bibitem[{\citenamefont{Greiner et~al.}(2002)\citenamefont{Greiner, Mandel,
516: Esslinger, H{\"a}nsch, and Bloch}}]{greiner02}
517: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Greiner {\it et al.}}},
518: % \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Mandel}},
519: % \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Esslinger}},
520: % \bibinfo{author}{\bibfnamefont{T.~W.} \bibnamefont{H{\"a}nsch}},
521: % \bibnamefont{and} \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Bloch}},
522: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{415}}, \bibinfo{pages}{39}
523: (\bibinfo{year}{2002}).
524:
525: \bibitem[{\citenamefont{Fisher et~al.}(1989)\citenamefont{Fisher,
526: Weichman,
527: Grinstein, and Fisher}}]{fisher89}
528: \bibinfo{author}{\bibfnamefont{M.~P.~A.} \bibnamefont{Fisher {\it et
529: al.}}},
530: % \bibinfo{author}{\bibfnamefont{P.~B.} \bibnamefont{Weichman}},
531: % \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Grinstein}},
532: % \bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.~S.}
533: % \bibnamefont{Fisher}},
534: \bibinfo{journal}{Phys.\ Rev.\ B}
535: \textbf{\bibinfo{volume}{40}}, \bibinfo{pages}{546}
536: (\bibinfo{year}{1989}).
537:
538: \bibitem[{\citenamefont{Jaksch et~al.}(1998)\citenamefont{Jaksch, Bruder,
539: Cirac, Gardiner, and Zoller}}]{jaksch98}
540: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Jaksch {\it et al.}}},
541: % \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Bruder}},
542: % \bibinfo{author}{\bibfnamefont{J.~I.} \bibnamefont{Cirac}},
543: % \bibinfo{author}{\bibfnamefont{C.~W.} \bibnamefont{Gardiner}},
544: % \bibnamefont{and} \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Zoller}},
545: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{81}},
546: \bibinfo{pages}{3108} (\bibinfo{year}{1998}).
547:
548: \bibitem[{\citenamefont{Petrov et~al.}(2000)\citenamefont{Petrov, Shlyapnikov,
549: and Walraven}}]{petrov00}
550: \bibinfo{author}{\bibfnamefont{D.~S.} \bibnamefont{Petrov}},
551: \bibinfo{author}{\bibfnamefont{G.~V.} \bibnamefont{Shlyapnikov}},
552: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~T.~M.}
553: \bibnamefont{Walraven}}, \bibinfo{journal}{Phys.\ Rev.\ Lett.}
554: \textbf{\bibinfo{volume}{85}}, \bibinfo{pages}{3745} (\bibinfo{year}{2000}).
555:
556: \bibitem[{\citenamefont{K{\"u}hner and Monien}(1998)}]{kuhner98}
557: \bibinfo{author}{\bibfnamefont{T.~D.} \bibnamefont{K{\"u}hner}}
558: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Monien}},
559: \bibinfo{journal}{Phys.\ Rev.\ B} \textbf{\bibinfo{volume}{58}},
560: \bibinfo{pages}{R14741} (\bibinfo{year}{1998}).
561:
562: \bibitem[{\citenamefont{Rapsch et~al.}(1999)\citenamefont{Rapsch,
563: Schollw{\"o}ck, and Zwerger}}]{rapsch99}
564: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Rapsch}},
565: \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{Schollw{\"o}ck}},
566: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Zwerger}},
567: \bibinfo{journal}{Europhys.\ Lett.} \textbf{\bibinfo{volume}{46}},
568: \bibinfo{pages}{559} (\bibinfo{year}{1999}).
569:
570: \bibitem[{\citenamefont{Olshanii}(1998)}]{olshanii98}
571: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Olshanii}},
572: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{81}},
573: \bibinfo{pages}{938} (\bibinfo{year}{1998}).
574:
575: \bibitem[{\citenamefont{Lieb and Liniger}(1963)}]{lieb63}
576: \bibinfo{author}{\bibfnamefont{E.~H.} \bibnamefont{Lieb}} \bibnamefont{and}
577: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Liniger}},
578: \bibinfo{journal}{Phys.\ Rev.} \textbf{\bibinfo{volume}{130}},
579: \bibinfo{pages}{1605} (\bibinfo{year}{1963});
580: \bibinfo{author}{\bibfnamefont{E.~H.} \bibnamefont{Lieb}},
581: \bibinfo{journal}{Phys.\ Rev.} \textbf{\bibinfo{volume}{130}},
582: \bibinfo{pages}{1616} (\bibinfo{year}{1963}).
583:
584: \bibitem[{\citenamefont{Pokrovsky and Talapov}(1979)}]{pokrovsky79}
585: \bibinfo{author}{\bibfnamefont{V.~L.} \bibnamefont{Pokrovsky}}
586: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.~L.}
587: \bibnamefont{Talapov}}, \bibinfo{journal}{Phys.\ Rev.\ Lett.}
588: \textbf{\bibinfo{volume}{42}}, \bibinfo{pages}{65} (\bibinfo{year}{1979});
589: %\bibitem[{\citenamefont{Schulz}(1980)}]{schulz80}
590: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Schulz}},
591: \bibinfo{journal}{Phys.\ Rev.\ B} \textbf{\bibinfo{volume}{22}},
592: \bibinfo{pages}{5274} (\bibinfo{year}{1980}).
593:
594: \bibitem[{\citenamefont{Girardeau}(1960)}]{girardeau60}
595: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Girardeau}},
596: \bibinfo{journal}{J.\ Math.\ Phys.} \textbf{\bibinfo{volume}{1}},
597: \bibinfo{pages}{516} (\bibinfo{year}{1960}).
598:
599: \bibitem[{\citenamefont{Haldane}(1981)}]{haldane81}
600: \bibinfo{author}{\bibfnamefont{F.~D.~M.} \bibnamefont{Haldane}},
601: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{47}},
602: \bibinfo{pages}{1840} (\bibinfo{year}{1981}).
603:
604: \bibitem[{\citenamefont{Coleman}(1975)}]{coleman75}
605: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Coleman}},
606: \bibinfo{journal}{Phys.\ Rev.\ D} \textbf{\bibinfo{volume}{11}},
607: \bibinfo{pages}{2088} (\bibinfo{year}{1975}).
608:
609: \bibitem[{\citenamefont{Gogolin et~al.}(1998)\citenamefont{Gogolin, Neresyan,
610: and M.\ Tsvelik}}]{gogolin98}
611: \bibinfo{author}{\bibfnamefont{A.~O.} \bibnamefont{Gogolin}},
612: \bibinfo{author}{\bibfnamefont{A.~A.} \bibnamefont{Neresyan}},
613: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{M.\
614: Tsvelik}}, \emph{\bibinfo{title}{Bosonization and Strongly Correlated
615: Systems}} (\bibinfo{publisher}{Cambridge University Press},
616: \bibinfo{year}{1998}).
617:
618: \bibitem[{\citenamefont{Kehrein}(1999)}]{kehrein99_01}
619: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Kehrein}},
620: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{83}},
621: \bibinfo{pages}{4914} (\bibinfo{year}{1999});
622: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Kehrein}},
623: \bibinfo{journal}{Nucl.\ Phys.\ B} \textbf{\bibinfo{volume}{592}},
624: \bibinfo{pages}{512} (\bibinfo{year}{2001}).
625:
626: \bibitem[{\citenamefont{Japaridze and Nersesyan}(1984)}]{japaridze84}
627: \bibinfo{author}{\bibfnamefont{G.~I.} \bibnamefont{Japaridze}}
628: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.~A.}
629: \bibnamefont{Nersesyan}}, \bibinfo{journal}{Nucl.\ Phys.\ B}
630: \textbf{\bibinfo{volume}{230}}, \bibinfo{pages}{511} (\bibinfo{year}{1984}).
631:
632: \bibitem[{\citenamefont{Dunjko et~al.}(2001)\citenamefont{Dunjko, Lorent, and
633: Olshanii}}]{dunjko01}
634: \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Dunjko}},
635: \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Lorent}}, \bibnamefont{and}
636: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Olshanii}},
637: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{86}},
638: \bibinfo{pages}{5413} (\bibinfo{year}{2001}).
639:
640: \bibitem[{\citenamefont{Birkl et~al.}(1995)\citenamefont{Birkl, Gatzke,
641: Deutsch, Rolston, and Phillips}}]{birkl95}
642: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Birkl {\it et al.}}},
643: %\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Gatzke}},
644: %\bibinfo{author}{\bibfnamefont{I.~H.} \bibnamefont{Deutsch}},
645: %\bibinfo{author}{\bibfnamefont{S.~L.} \bibnamefont{Rolston}},
646: %\bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.~D.}
647: %\bibnamefont{Phillips}},
648: \bibinfo{journal}{Phys.\ Rev.\ Lett.}
649: \textbf{\bibinfo{volume}{75}}, \bibinfo{pages}{3823} (\bibinfo{year}{1995}).
650:
651: \bibitem[{\citenamefont{Weidem{\"u}ller
652: et~al.}(1995)\citenamefont{Weidem{\"u}ller, Hemmerich, G{\"o}rlitz,
653: Esslinger, and H{\"a}nsch}}]{weidemuller95}
654: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Weidem{\"u}ller} {\it et al.}},
655: \bibinfo{journal}{Phys.\ Rev.\ Lett.}
656: \textbf{\bibinfo{volume}{75}}, \bibinfo{pages}{4583} (\bibinfo{year}{1995}).
657:
658: \end{thebibliography}
659:
660: %\bibliographystyle{/home/buechler/Refbib/apsrev}
661: %\bibliography{/home/buechler/Refbib/journals,/home/buechler/Refbib/ref}
662:
663: %\end{multicols}
664: \end{document}
665: