Two-state Bose-Hubbard model in the hard-core boson limit
Phase transition into the phase with Bose-Einstein (BE) condensate in the two-band Bose-Hubbard model with the particle hopping in the excited band only is investigated. Instability connected with such a transition (which appears at excitation energies δ <|t′₍₎|, where |t′₍₎| is the particle hopp...
Gespeichert in:
| Datum: | 2011 |
|---|---|
| Hauptverfasser: | , |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Інститут фізики конденсованих систем НАН України
2011
|
| Schriftenreihe: | Condensed Matter Physics |
| Online Zugang: | https://nasplib.isofts.kiev.ua/handle/123456789/119899 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Zitieren: | Two-state Bose-Hubbard model in the hard-core boson limit / I.V. Stasyuk, O.V. Velychko // Condensed Matter Physics. — 2011. — Т. 14, № 1. — С. 13004: 1-14. — Бібліогр.: 40 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
nasplib_isofts_kiev_ua-123456789-119899 |
|---|---|
| record_format |
dspace |
| spelling |
nasplib_isofts_kiev_ua-123456789-1198992025-02-09T15:06:12Z Two-state Bose-Hubbard model in the hard-core boson limit Двостанова модель Бозе - Хаббарда в границi жорстких бозонiв Stasyuk, I.V. Velychko, O.V. Phase transition into the phase with Bose-Einstein (BE) condensate in the two-band Bose-Hubbard model with the particle hopping in the excited band only is investigated. Instability connected with such a transition (which appears at excitation energies δ <|t′₍₎|, where |t′₍₎| is the particle hopping parameter) is considered. The reentrant behaviour of spinodales is revealed in the hard-core boson limit in the region of positive valuesof chemical potential. It is found that the order of the phase transition undergoes a change in this case and becomes the first one; the re-entrant transition into the normal phase does not take place in reality. First order phase transitions also exist at negative values of δ (under the conditionδ > δcrit ≈ −0.12|t′₍₎|). At μ <0the phase transition mostly remains to be of the second order. The behaviour of the BE-condensate order parameter is analyzed, the (Θ, μ) and(|t′₍₎|, μ) phase diagrams are built and localizations of tricritical points are established. The conditions are found at which the separation on the normal phase and the phase with the BE condensate takes place. Дослiджено перехiд у фазу з бозе-конденсатом у двозоннiй моделi Бозе-Хаббарда з переносом частинок лише у збудженiй зонi. Розглянуто нестiйкiсть, пов’язану з таким переходом, який вiдбувається при енергiях збудження δ <|t′0| (де |t′0| – параметр перескоку частинок). У границi жорстких бозонiв виявлено зворотнiй хiд спiнодалей в областi додатних значень хiмiчного потенцiалу. Встановлено, що рiд фазового переходу у цьому випадку змiнюється на перший i повернення до нормальної фази насправдi не вiдбувається. Переходи першого роду iснують i при вiд’ємнiй енергiї δ(за умови δ > δ crit ≈ −0.12|t′0|). При μ<0 фазовий перехiд залишається переважно переходом другого роду. Проаналiзовано хiд параметра порядку бозе-конденсату, побудовано фазовi (Θ,μ) i (|t′0|, μ) дiаграми, встановлено локалiзацiю трикритичних точок. Знайдено умови, при яких для фiксованої концентрацiї частинок наступає розшарування на нормальну фазу i фазу з бозе-конденсатом. 2011 Article Two-state Bose-Hubbard model in the hard-core boson limit / I.V. Stasyuk, O.V. Velychko // Condensed Matter Physics. — 2011. — Т. 14, № 1. — С. 13004: 1-14. — Бібліогр.: 40 назв. — англ. 1607-324X PACS: 03.75.Hh, 03.75.Lm, 64.70.Tg, 71.35.Lk, 37.10.Jk, 67.85.-d DOI:10.5488/CMP.14.13004 arXiv:1103.5662 https://nasplib.isofts.kiev.ua/handle/123456789/119899 en Condensed Matter Physics application/pdf Інститут фізики конденсованих систем НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| description |
Phase transition into the phase with Bose-Einstein (BE) condensate in the two-band Bose-Hubbard model with the particle hopping in the excited band only is investigated. Instability connected with such a transition (which appears at excitation energies δ <|t′₍₎|, where |t′₍₎| is the particle hopping parameter) is considered. The reentrant behaviour of spinodales is revealed in the hard-core boson limit in the region of positive valuesof chemical potential. It is found that the order of the phase transition undergoes a change in this case and becomes the first one; the re-entrant transition into the normal phase does not take place in reality. First order phase transitions also exist at negative values of δ (under the conditionδ > δcrit ≈ −0.12|t′₍₎|). At μ <0the phase transition mostly remains to be of the second order. The behaviour of the BE-condensate order parameter is analyzed, the (Θ, μ) and(|t′₍₎|, μ) phase diagrams are built and localizations of tricritical points are established. The conditions are found at which the separation on the normal phase and the phase with the BE condensate takes place. |
| format |
Article |
| author |
Stasyuk, I.V. Velychko, O.V. |
| spellingShingle |
Stasyuk, I.V. Velychko, O.V. Two-state Bose-Hubbard model in the hard-core boson limit Condensed Matter Physics |
| author_facet |
Stasyuk, I.V. Velychko, O.V. |
| author_sort |
Stasyuk, I.V. |
| title |
Two-state Bose-Hubbard model in the hard-core boson limit |
| title_short |
Two-state Bose-Hubbard model in the hard-core boson limit |
| title_full |
Two-state Bose-Hubbard model in the hard-core boson limit |
| title_fullStr |
Two-state Bose-Hubbard model in the hard-core boson limit |
| title_full_unstemmed |
Two-state Bose-Hubbard model in the hard-core boson limit |
| title_sort |
two-state bose-hubbard model in the hard-core boson limit |
| publisher |
Інститут фізики конденсованих систем НАН України |
| publishDate |
2011 |
| url |
https://nasplib.isofts.kiev.ua/handle/123456789/119899 |
| citation_txt |
Two-state Bose-Hubbard model in the hard-core boson limit / I.V. Stasyuk, O.V. Velychko // Condensed Matter Physics. — 2011. — Т. 14, № 1. — С. 13004: 1-14. — Бібліогр.: 40 назв. — англ. |
| series |
Condensed Matter Physics |
| work_keys_str_mv |
AT stasyukiv twostatebosehubbardmodelinthehardcorebosonlimit AT velychkoov twostatebosehubbardmodelinthehardcorebosonlimit AT stasyukiv dvostanovamodelʹbozehabbardavgranicižorstkihbozoniv AT velychkoov dvostanovamodelʹbozehabbardavgranicižorstkihbozoniv |
| first_indexed |
2025-11-27T04:19:53Z |
| last_indexed |
2025-11-27T04:19:53Z |
| _version_ |
1849915814845612032 |
| fulltext |
Condensed Matter Physics 2011, Vol. 14, No 1, 13004: 1–14
DOI:10.5488/CMP.14.13004
http://www.icmp.lviv.ua/journal
Two-state Bose-Hubbard model
in the hard-core boson limit
I.V. Stasyuk, O.V. Velychko
Institute for Condensed Matter Physics of the National Academy of Sciences of Ukraine, 1 Svientsitskii Str.,
79011 Lviv, Ukraine
March 14, 2011
Phase transition into the phase with Bose-Einstein (BE) condensate in the two-band Bose-Hubbard model
with the particle hopping in the excited band only is investigated. Instability connected with such a transition
(which appears at excitation energies δ < |t′
0
|, where |t′
0
| is the particle hopping parameter) is considered.
The re-entrant behaviour of spinodales is revealed in the hard-core boson limit in the region of positive values
of chemical potential. It is found that the order of the phase transition undergoes a change in this case and
becomes the first one; the re-entrant transition into the normal phase does not take place in reality. First order
phase transitions also exist at negative values of δ (under the condition δ > δcrit ≈ −0.12|t′
0
|). At µ < 0
the phase transition mostly remains to be of the second order. The behaviour of the BE-condensate order
parameter is analyzed, the (Θ, µ) and (|t′
0
|, µ) phase diagrams are built and localizations of tricritical points
are established. The conditions are found at which the separation on the normal phase and the phase with
the BE condensate takes place.
Key words: Bose-Hubbard model, hard-core bosons, Bose-Einstein condensation, excited band
PACS: 03.75.Hh, 03.75.Lm, 64.70.Tg, 71.35.Lk, 37.10.Jk, 67.85.-d
1. Introduction
During the recent years Bose-Hubbard model (BHM) is proved to be a valuable tool in the
theory of systems of strongly correlated particles. The model achieves a wide recognition due to a
successful description of thermodynamics and dynamics of ultracold Bose atoms in optical lattices
where a phase transition to the phase with the Bose-Einstein (BE) condensate (so-called Mott
insulator (MI) – superfluid state (SF) transition) occurs at very low temperatures. Experimental
evidence of BE condensation in optical lattices was found for the first time in works [1, 2] while
theoretical predictions of such an effect were given earlier [3]. Starting from the 90-ies of the past
century, a series of papers were devoted to the theory of this phenomenon. Among the first key
articles on the subject one should mention the work [4] where BHM was studied in the mean
field approximation. The calculated therein phase diagrams demonstrate that in the simplest case
(i.e., hopping of Bose particles in the presence of a single-site Hubbard repulsion) the MI-SF
transition is of the second order. Moreover, it is supposed that particles reside in the ground
state of local potential wells in the lattice. Forthcoming theoretical investigations in this field were
performed with the use of various techniques, e.g., the random phase approximation (RPA) in the
Green function method [5, 6], a strong-coupling perturbation theory [7, 8], the dynamical mean
field theory (Bose-DMFT) [9, 10] as well as quantum Monte-Carlo calculations [11, 12] and other
numerical methods.
The Bose-Hubbard model is also intensively used for a theoretical description of a wide range
of phenomena: quantum delocalization of hydrogen atoms adsorbed on the surface of transition
metals [13, 14], quantum diffusion of light particles on the surface or in the bulk [15, 16], thermo-
dynamics of the impurity ion intercalation into semiconductors [17, 18].
In the last mentioned applications, there is usually a restriction on the position occupation
number (ni 6 1), which corresponds to the limit of an infinite Hubbard repulsion for the consid-
ered model. Such a model of hard-core ions (where particles are described by the Pauli statistics)
c© I.V. Stasyuk, O.V. Velychko 13004-1
http://dx.doi.org/10.5488/CMP.14.13004
http://www.icmp.lviv.ua/journal
I.V. Stasyuk, O.V. Velychko
is also known as the fundamental one for the investigation of a wide range of problems, e.g., super-
conductivity due to a local electron pairing [19] or ionic hopping in ionic (superionic) conductors
[20, 21].
The study of a quantum delocalization or diffusion reveals an important role of excited vibra-
tional states of particles (ions) in localized (interstitial) positions with a much higher probability of
ion hopping between them [15, 22, 23]. A similar issue of a possible BE condensation in the excited
bands in optical lattices is also considered but the condition of their sufficient occupation due to
the optical pumping (see, e.g. [24]) is imposed. An orbital degeneration of the excited p-state is
accompanied by anisotropy of hopping parameters and causes the appearance of variously polar-
ized bands in the one-particle spectrum. Such bands correspond by convention to different sorts
(so-called “flavours”) of bosons and their number correlates with the lattice dimensionality. In the
framework of the necessary generalization of the Bose-Hubbard model, a possibility of the MI-SF
transition to the phase with BE condensate in the pumping-induced quasi-equilibrium long-living
state of the system has been established [25].
In the equilibrium case, the issue of BE condensation involving the excited states in the frame-
work of ordinary Bose-Hubbard model was not considered in practice. The exception is the system
of spin-1 bosons [26, 27] where a hyperfine splitting gives rise to multiplets of local states resulting
in closely-spaced excited levels. As demonstrated in [28, 29], the MI-SF phase transition could be
of the first order when a single-site spin interaction is of the “antiferromagnetic” type. A simi-
lar change of the phase transition order also takes place for multicomponent Bose systems in the
optical lattices [30].
In the present work we consider an equilibrium thermodynamics of the Bose-Hubbard model
taking into account only one nondegenerated excited state on the lattice site besides the ground
one. On the one hand, such a model corresponds to 1D or strongly anisotropic (quasi-1D) optical
lattice, and on the other hand, it is close to a situation that is characteristic of a system of light
particles adsorbed on the metal surface. For example, the excited states of hydrogen atoms on the
Ni(111) surface are sufficiently distant [22] so only the lowest one could be taken into account. We
shall investigate a condition of instability of a normal state of the Bose system with respect to BE
condensation considering a criterion of divergence of the susceptibility (χ ∼ 〈〈cl|c+p 〉〉ω |q=0,ω=0)
characterizing the system response with respect to the field related to a spontaneous creation or
annihilation of particles. We shall also study the behaviour of the order parameter 〈c0〉 (〈c+0 〉) as
well as the grand canonical potential in the region of the MI-SF transition and shall build relevant
phase diagrams. Special attention will be paid to a change of the phase transition order and
localization of tricritical points at different values of excitation energy, particle hopping parameter
and temperature.
We shall limit ourselves to the hard-core boson (HCB) limit where a limitation on occupation
numbers is present: no more than one particle per site regardless of the state (excited or ground)
occupied by it. Thus, the single-site problem is a three-level one (contrary to the two-level ordi-
nary HCB case). For this reason, it is convenient to use the formalism of Hubbard operators [31]
(standard basis operators [32]).
2. Two-state Bose-Hubbard model in RPA: normal phase
The Bose-Hubbard model is used for description of the system of Bose particles which are
located in a periodic field and can reside in lattice sites. Taking into account only the ground
and the first excited vibrational levels in the potential well on the site, one can express the model
Hamiltonian as:
Ĥ = (ε− µ)
∑
i
b+i bi + (ε′ − µ)
∑
i
c+i ci +
Ub
2
∑
i
nb
i(n
b
i − 1) +
Uc
2
∑
i
nc
i (n
c
i − 1) + Ubc
∑
i
nb
in
c
i
+
∑
ij
tbijb
+
i bj +
∑
ij
tcijc
+
i cj +
∑
ij
tbcij (b
+
i cj + c+i bj), (2.1)
13004-2
Two-state Bose-Hubbard model in the HCB limit
where bi and b+i (ci and c+i ) are Bose operators of annihilation and creation of particles in the
ground (excited) state, ε and ε′ are respective energies of state and µ is the chemical potential
of particles. Such a Hamiltonian includes the single-site Hubbard repulsions with energies Ub, Uc
and Ubc as well as the particle hopping between ground (tb), excited (tc) and different (tbc) states.
Hereinafter we assume Ub = Uc = Ubc for simplicity.
Let us define a single-site basis |nb
i , n
c
i〉 ≡ |i;nb
i , n
c
i 〉 (which is formed by particle occupation
numbers in the ground and in the excited states, i.e., eigenvalues of operators nb
i = b+i bi and
nc
i = c+i ci) as well as introduce Hubbard operators (standard basis operators)
Xn,m;n′,m′
i ≡ |i;n,m〉〈i;n′,m′|. (2.2)
Annihilation and creation Bose operators may be written as
bi =
∑
n
∑
m
√
n+ 1Xn,m;n+1,m
i , b+i =
∑
n
∑
m
√
n+ 1Xn+1,m;n,m
i ;
ci =
∑
n
∑
m
√
m+ 1Xn,m;n,m+1
i , c+i =
∑
n
∑
m
√
m+ 1Xn,m+1;n,m
i . (2.3)
Corresponding occupation numbers look as follows
nb
i =
∑
n
∑
m
nXn,m;n,m
i , nc
i =
∑
n
∑
m
mXn,m;n,m
i , (2.4)
where summation indices n,m = 0, . . . ,∞ in both (2.3) and (2.4) formulae.
In the X-operator representation, the single-site part of Hamiltonian (2.1) can be written as
Ĥ0 =
∑
i
∑
n
∑
m
λnmXn,m;n,m
i , (2.5)
where
λnm = n(ε− µ) +m(ε′ − µ) +
U
2
(n+m)(n+m− 1). (2.6)
Terms describing an inter-site transfer in Hamiltonian (2.1) are transformed in a similar way.
Our primary goal is to calculate the two-time temperature boson Green’s functions 〈〈b|b+〉〉 and
〈〈c|c+〉〉, which describe an excitation spectrum and make it possible to investigate the conditions
of the system’s instability with respect to the spontaneous symmetry breaking and the appearance
of a BE condensate. As follows from definitions (2.3)
〈〈bl|b+p 〉〉ω =
∑
nm
∑
rs
√
n+ 1
√
r + 1 〈〈Xn,m;n+1,m
l |Xr+1,s;r,s
p 〉〉ω ,
〈〈cl|c+p 〉〉ω =
∑
nm
∑
rs
√
m+ 1
√
s+ 1 〈〈Xn,m;n,m+1
l |Xr,s+1;r,s
p 〉〉ω . (2.7)
We will use the equation-of-motion method for the evaluation of X-operator Green’s functions.
For the first one, from relations (2.7) one could write
~ω〈〈Xn,m;n+1,m
l |Xr+1,s;r,s
p 〉〉ω =
~
2π
〈Xn,m;n,m
l −Xn+1,m;n+1,m
l 〉δlpδnrδms
+ 〈〈[Xn,m;n+1,m
l , Ĥ]|Xr+1,s;r,s
p 〉〉ω . (2.8)
13004-3
I.V. Stasyuk, O.V. Velychko
Let us write the commutators
[Xn,m;n+1,m
l , Ĥ0] = (λn+1,m − λn,m)Xn,m;n+1,m
l , (2.9)
[Xn,m;n+1,m
l , b+i ] = δli
√
n+ 1
(
Xn,m;n,m
l −Xn+1,m;n+1,m
l
)
, (2.10a)
[Xn,m;n+1,m
l , bi] = δli
(√
n+ 2Xn,m;n+2,m
l −
√
nXn−1,m;n+1,m
l
)
, (2.10b)
[Xn,m;n+1,m
l , c+i ] = δli
(√
mXn,m;n+1,m−1
l −
√
m+ 1Xn,m+1;n+1,m
l
)
, (2.10c)
[Xn,m;n+1,m
l , ci] = δli
(√
m+ 1Xn,m;n+1,m+1
l −
√
mXn,m−1;n+1,m
l
)
. (2.10d)
The latter are originated from the commutation of an initial X-operator with the inter-site transfer
terms of the Hamiltonian, thus producing the higher-order Green’s functions
〈〈X ···
l bj|Xr+1,s;r,s
p 〉〉ω , 〈〈X ···
l b+j |Xr+1,s;r,s
p 〉〉ω , . . . , (2.11)
where X ···
l stands for operators on the right-hand side of expressions (2.10a)–(2.10d).
Decoupling of functions (2.11) in the random phase approximation (RPA) is performed in the
following way:
〈〈X ···
l bj |Xr+1,s;r,s
p 〉〉ω ≈ 〈X ···
l 〉〈〈bj |Xr+1,s;r,s
p 〉〉ω + 〈bj〉〈〈X ···
l |Xr+1,s;r,s
p 〉〉ω . (2.12)
In the case of the normal phase (which will be studied herein) 〈bj〉 = 〈b+j 〉 = 0. Thus, retaining
only the averages 〈X ···
l 〉 of diagonal X-operators we have
[Xn,m;n+1,m
l , Ĥ ] ≈ ∆nmXn,m;n+1,m
l +
√
n+ 1Qnm
∑
j
tljbj +
√
n+ 1Qnm
∑
j
t′′ljcj (2.13)
and equation (2.8) can be rewritten as
〈〈Xn,m;n+1,m
l |Xr+1,s;r,s
p 〉〉ω =
~
2π
δlpδnrδms
Qnm
~ω −∆nm
+
√
n+ 1Qnm
~ω −∆nm
∑
j
tlj〈〈bj |Xr+1,s;r,s
p 〉〉ω +
√
n+ 1Qnm
~ω −∆nm
∑
j
t′′lj〈〈cj |Xr+1,s;r,s
p 〉〉ω . (2.14)
The following notations are introduced
Qnm = 〈Xn,m;n,m
l −Xn+1,m;n+1,m
l 〉, ∆nm = λn+1,m − λn,m , (2.15)
for the occupation difference of adjacent levels and the related transition energies when the number
of Bose particles in the ground state (with the energy ε) on the site increases by one.
Proceeding from X-operators in equation (2.14) to the Bose operators b and b+ according to
definition (2.3) we obtain
〈〈bl|b+p 〉〉ω =
~
2π
δlpg0(ω) + g0(ω)
(∑
j
tlj〈〈bj |b+p 〉〉ω +
∑
j
t′′lj〈〈cj |b+p 〉〉ω
)
, (2.16)
where the function
g0(ω) =
∑
nm
Qnm
~ω −∆nm
(n+ 1) (2.17)
has the meaning of the unperturbed Green’s function for bosons residing in the single-site ground
state.
Equations of motion for “mixed” Green’s functions 〈〈c|b+〉〉 are obtained in the way similar to
the above described scheme. Using decoupling (2.12) one can write
[Xn,m;n,m+1
l , Ĥ] ≈ ∆′
nmXn,m;n,m+1
l +
√
m+ 1Q′
nm
∑
j
t′′ljbj +
√
m+ 1Q′
nm
∑
j
t′ljcj , (2.18)
13004-4
Two-state Bose-Hubbard model in the HCB limit
which results in the equation
〈〈cl|b+p 〉〉ω = g′0(ω)
(∑
j
t′′lj〈〈bj |b+p 〉〉ω +
∑
j
t′lj〈〈cj |b+p 〉〉ω
)
. (2.19)
Here, similarly to (2.15) and (2.17)
Q′
nm = 〈Xn,m;n,m
l −Xn,m+1;n,m+1
l 〉, ∆′
nm = λn,m+1 − λn,m , g′0(ω) =
∑
nm
Q′
nm
~ω −∆′
nm
(m+ 1),
(2.20)
and the function g′0(ω) is the unperturbed Green’s function for bosons residing in the excited state.
By means of the Fourier transform
〈〈bl|b+p 〉〉ω =
1
N
∑
q
eiq(Rl−Rp)〈〈b|b+〉〉q,ω , (2.21)
one can proceed to the momentum representation obtaining a system of equations
〈〈b|b+〉〉q,ω =
~
2π
g0(ω) + g0(ω)tq〈〈b|b+〉〉q,ω + g0(ω)t
′′
q 〈〈c|b+〉〉q,ω ,
〈〈c|b+〉〉q,ω = g′0(ω)t
′′
q 〈〈b|b+〉〉q,ω + g′0(ω)t
′
q〈〈c|b+〉〉q,ω , (2.22)
where tq , t
′
q and t′′q stand for the Fourier transforms of hopping parameters.
A pair of equations for other Green’s functions are obtained in a similar way
〈〈b|c+〉〉q,ω = g0(ω)tq〈〈b|c+〉〉q,ω + g0(ω)t
′′
q 〈〈c|c+〉〉q,ω ,
〈〈c|c+〉〉q,ω =
~
2π
g′0(ω) + g′0(ω)t
′′
q 〈〈b|c+〉〉q,ω + g′0(ω)t
′
q〈〈c|c+〉〉q,ω . (2.23)
Solutions of equations (2.22) and (2.23) are as follows:
〈〈b|b+〉〉q,ω =
~
2π
1
Dq(ω)
g0(ω)(1− g′0(ω)t
′
q),
〈〈c|c+〉〉q,ω =
~
2π
1
Dq(ω)
g′0(ω)(1− g′0(ω)tq),
〈〈c|b+〉〉q,ω =
~
2π
1
Dq(ω)
g0(ω)g
′
0(ω)t
′′
q = 〈〈b|c+〉〉q,ω , (2.24)
where
Dq(ω) = 1− g0(ω)tq − g′0(ω)t
′
q + g0(ω)g
′
0(ω)
[
tqt
′
q − (t′′q )
2
]
. (2.25)
The equation Dq(ω) = 0 gives the excitation spectrum which is obtained here in the RPA. On
the other hand, the divergence of boson Green’s functions (2.24) at the zero values of wave vector
and frequency is the criterion of instability with respect to BE condensation [5, 33], thus giving
the following condition
Dq=0(ω = 0) = 0, (2.26)
which can be rewritten in the explicit form
1− g0(ω)tq − g′0(ω)t
′
q + g0(ω)g
′
0(ω)
[
tqt
′
q − (t′′q )
2
]
= 0, (2.27)
where
g0(0) = −
∑
nm
Qnm
(n+m)U − µ
(n+ 1), g′0(0) = −
∑
nm
Q′
nm
(n+m)U + δ − µ
(m+ 1), (2.28)
and δ = ε′ − ε is the excitation energy.
13004-5
I.V. Stasyuk, O.V. Velychko
We should point out that divergence of the 〈〈b|b+〉〉0,0 function correlates with the appearance
of the BE condensate in the ground state while at the divergence of the 〈〈c|c+〉〉0,0 function, BE
condensation takes place in the excited state. In general, both condensates appear simultaneously
except the case t′′q = 0 (e.g. due to symmetry reasons) where these effects become independent and
only the one type of condensate arises in the instability point.
Equation (2.27), mutually relating the chemical potential, hopping parameters and temperature,
allows us to construct spinodal surfaces (or lines) in the above mentioned coordinates and to find
the temperature of the phase transition to the phase with BE condensate (so-called SF phase)
where such a transition is of the second order. Below, this problem (especially the issue of the
phase transition order) will be investigated more in detail.
3. NO phase instability in HCB limit
Let us consider now a simple special case of the HCB limit when occupation numbers in the
|n,m〉 state are restricted by a condition n + m 6 1. In the framework of the model, it formally
means U → ∞.
In this case, the model becomes a three-level one with the local energies
λ00 = 0, λ01 = δ − µ, λ10 = −µ (3.1)
and the following transition energies
∆00 = −µ, ∆′
00 = δ − µ. (3.2)
Thus, equation (2.27) can be rewritten in the form
1− Q00
µ
t0 −
Q′
00
µ− δ
t′0 +
Q00Q
′
00
µ(µ− δ)
[
t0t
′
0 − (t′′0 )
2
]
= 0, (3.3)
where
Q00 =
1− eβµ
1 + eβµ + eβ(µ−δ)
, Q′
00 =
1− eβ(µ−δ)
1 + eβµ + eβ(µ−δ)
(3.4)
in the zero approximation with respect to hopping.
-0.9 -0.6 -0.3 0.0 0.3
0.0
0.1
0.2
0.3
δ = 0.9
δ = 0.5
T
em
pe
ra
tu
re
Θ
Chemical potential µ
δ = 0.1
Figure 1. Lines of the NO phase instability
(spinodals) with respect to the appearance of
BE condensate in the (Θ, µ) plane in the HCB
limit at various excitation energies (t0 = 0,
|t′0| = 1, t′′0 = 0).
We take into account (according to esti-
mations made in [15, 25]) that boson wave
functions in adjacent potential wells overlap in
greater extent in the excited states compared to
the ground ones. Accordingly, we shall put here
t0 = 0. For a centrosymmetric lattice and in
the case of different parity of wave functions of
ground and excited states we have also t′′0 = 0.
Finally, we follow a usual convention of the BH
model for optical lattices taking t′0 < 0. In this
way equation (3.3) can be reduced to
|t′0|
δ − µ
1− eβ(µ−δ)
1 + eβµ + eβ(µ−δ)
= 1. (3.5)
Its solutions determine the stability region
boundaries of the normal (NO) phase. Respec-
tive lines of spinodals are numerically calcu-
lated and presented in figure 1 (here and below
the energy quantities are given in units of |t′0|).
As illustrated in figure 1, at δ < |t′0|
spinodals surround an asymmetric area in the
13004-6
Two-state Bose-Hubbard model in the HCB limit
(Θ, µ) plane which is located between the points µ = δ − |t′0| and µ = 0 of the abscissa axis. In
this region, the NO phase is unstable; this is connected with the appearance of BE condensate. At
δ < |t′0|/2 and µ > 0 the backward path of spinodal is observed and a lower temperature of the
NO phase instability appears, thus suggesting a possibility of the SF phase existence in the inter-
mediate temperature range (so-called “re-entrant transition”). However, as will be shown further,
in the mentioned region a real thermodynamic behaviour is even more complicated. The order of
the NO-SF transition can change to the first one and the SF-phase remains stable up to the zero
temperature.
4. Phase diagrams in MFA
For a more detailed treatment of the NO-SF transition issue, let us study the thermodynamics
of the considered system in the HCB limit, thus reducing the problem to a three-state model with
the Hamiltonian
Ĥ =
∑
ip
λpX
pp
i +
∑
ij
tijX
10
i X01
j +
∑
ij
t′ijX
20
i X02
j +
∑
ij
t′′ij(X
10
i X02
j +X20
i X01
j ), (4.1)
where the shorthand notations are used
|0〉 ≡ |00〉, |1〉 ≡ |10〉, |2〉 ≡ |01〉; λ0 = λ00, λ1 = λ10, λ2 = λ01 . (4.2)
Possibility of BE condensation will be studied in the MFA. Average values of creation (annihi-
lation) operators for Bose particles in the ground or excited local state
η = 〈X10
i 〉 = 〈X01
i 〉 (≡ 〈bi〉), ξ = 〈X20
i 〉 = 〈X02
i 〉 (≡ 〈ci〉) (4.3)
play the role of order parameters for the SF-phase. Hence, the mean-field Hamiltonian is as follows:
ĤMF = −N(t0η
2 + t′0ξ
2 + 2t′′0ηξ) +
∑
ip
λpX
pp
i
+
∑
i
[
t0η(X
10
i +X01
i ) + t′0ξ(X
20
i +X02
i ) + t′′0ξ(X
10
i +X01
i ) + t′′0η(X
20
i +X02
i )
]
. (4.4)
Self-consistency equations for parameters η and ξ
η = Z−1 Sp[X10
i exp(−βĤMF)], ξ = Z−1 Sp[X20
i exp(−βĤMF)] (4.5)
are equivalent to the condition of minimum of the grand canonical potential Ω = −Θ lnZ, where
Z = Sp exp(−βĤMF).
Limiting our consideration to the case of particle hopping only through excited states (t′0 6= 0,
t0 = t′′0 = 0) we can diagonalize Hamiltonian (4.4) by a rotation transformation
|0〉
|1〉
|2〉
=
cosϑ 0 − sinϑ
0 1 0
sinϑ 0 cosϑ
|0̃〉
|1̃〉
|2̃〉
, (4.6)
where
cos 2ϑ =
λ2 − λ0√
(λ2 − λ0)2 + 4(t′0ξ)
2
, sin 2ϑ =
2|t′0|ξ√
(λ2 − λ0)2 + 4(t′0ξ)
2
(4.7)
and λ2 − λ0 = δ − µ. In terms of operators X̃rs = |r̃〉〈s̃|
ĤMF = N |t′0|ξ2 +
∑
ip
λ̃pX̃
pp
i . (4.8)
13004-7
I.V. Stasyuk, O.V. Velychko
New energies of single-site states are
λ̃0,2 =
δ − µ
2
∓
√(
δ − µ
2
)2
+ (t′0ξ)
2, λ̃1 = −µ. (4.9)
In the new basis
X02
i +X20
i = −(X̃22
i − X̃00
i ) sin 2ϑ+ (X̃20
i − X̃02
i ) cos 2ϑ, (4.10)
which yields after averaging
ξ =
1
2
〈X̃00
i − X̃22
i 〉 sin 2ϑ. (4.11)
Taking into account that 〈X̃pp〉 = Z−1 exp(−βλ̃p), Z =
∑
p exp(−βλ̃p) we come to the equation for
-1.0 -0.5 0.0 0.5 1.0
0.0
0.1
0.2
0.3
0.4
0.5
δ = 0.1
Θ = 0.2
Θ = 0.1
O
rd
er
p
ar
am
et
er
ξ
Chemical potential µ
Θ = 0.0
Figure 2. Dependences of the order parameter
ξ on the chemical potential µ for the reduced
three-level (HCB) model at various tempera-
tures indicating the possibility of the first or-
der phase transition at low enough temperatures
(δ = 0.1, |t′0| = 1).
the order parameter ξ:
ξ =
1
Z
|t′0|ξ√
(δ − µ)2 + 4(t′0ξ)
2
(
e−βλ̃0 − e−βλ̃2
)
.
(4.12)
Solution ξ = 0 corresponds to the NO phase. A
nonzero solution describing the BE condensate
is obtained from the equation
1
Z
|t′0|√
(δ − µ)2 + 4(t′0ξ)
2
(
e−βλ̃0 − e−βλ̃2
)
= 1.
(4.13)
In the limit ξ → 0 this equation determines the
line where the order parameter for the SF phase
tends to zero. One can readily see that it coin-
cides with spinodal equation (3.5) thus defining
the line of the second order NO-SF phase tran-
sition (when just the transition of such an order
takes place).
Numerical solutions of equation (4.13) make
it possible to study the behavior of the order
parameter ξ depending on chemical potential µ at various temperatures as illustrated in figure 2. In
the main, at negative values of chemical potential the parameter ξ changes smoothly and the phase
transition to the SF phase is of the second order. But at µ & 0 and low enough temperatures, the
0.4 0.6 0.8 1.0
0.0
0.1
0.2
0.3
0.4
0.5
δ = 0
Θ = 0.001
Θ = 0.01
O
rd
er
p
ar
am
et
er
ξ
Chemical potential µ
Θ = 0.0
0.3 0.6 0.9
0.0
0.1
0.2
0.3
0.4
0.5
δ = −0.05
Θ = 0.07
Θ = 0.15
O
rd
er
p
ar
am
et
er
ξ
Chemical potential µ
Θ = 0.0
Figure 3. Low-temperature behaviour of the order parameter ξ for the reduced three-level (HCB)
model at zero and negative excitation energies δ and various temperatures (|t′0| = 1).
13004-8
Two-state Bose-Hubbard model in the HCB limit
-1.0 -0.5 0.0 0.5
0.0
0.1
0.2
0.3 1st order PT
2nd order PT
δ = 0.6
δ = 0.4
T
em
pe
ra
tu
re
Θ
Chemical potential µ
δ = 0.1
0.0 0.3 0.6 0.9
0.00
0.04
0.08
0.12
0.16
1st order PT
2nd order PT
spinodal line
T
em
pe
ra
tu
re
Θ
Chemical potential µ
δ = 0.01
Figure 4. Lines of the NO-SF phase transi-
tion in the (Θ, µ) plane at various excitation
energies δ (|t′0| = 1).
Figure 5. An illustration of discrepancy between
the spinodal curve and the real line of the first
order phase transition for δ = 0.01 (|t′0| = 1).
ξ(µ) dependence has an S-like bend. In this case, the first order phase transition with an abrupt
change of the parameter ξ takes place. This phase transition occurs at a certain value of the
chemical potential which could be calculated using the Maxwell rule or considering the minimum
of the grand canonical potential Ω(µ) as a function of the chemical potential (see below). Obviously,
the point of ξ nullification does not anymore correspond here to the phase transition.
Similar behaviour of the parameter ξ holds even at zero excitation energy (δ = 0) where the
first order phase transition remains for nonzero temperatures whereas at T = 0 its order changes
to the second one (figure 3). At negative values of δ (which corresponds to inversion of ε and ε′
levels and to hopping between ground states) the second order of the transition is preserved in the
low-temperature region close to T = 0 transforming to the first order transition at the temperature
increase and recovering henceforth (figure 3).
Changes in the NO-SF phase transition order and localization of the corresponding tricritical
points are depicted in figure 4, where phase diagrams are given for various values of the excitation
energy δ. At temperatures lower than tricritical, spinodal lines and phase transition curves come
apart as one can see comparing figures 1 and 4. At small values of δ, the discrepancy is quite
0.4 0.6 0.8 1.0
0.00
0.04
0.08
0.12
0.16
δ = −0.1
1st order PT
2nd order PT
δ = 0.01 δ = −0.05T
em
pe
ra
tu
re
Θ
Chemical potential µ
δ = 0
-4 -2 0 2
1
3
5
Θ = 0.8
Θ = 0.5
Θ = 0.1
1st order PT
2nd order PT
H
op
pi
ng
e
ne
rg
y
|t 0'|
Chemical potential µ
Θ = 0
Figure 6. Appearance of two tricritical points
at zero and negative values of excitation en-
ergy δ (|t′0| = 1).
Figure 7. Lines of the NO-SF phase transition
in the (|t′0|, µ) plane at various temperatures Θ
(energy quantities are given in units of δ).
13004-9
I.V. Stasyuk, O.V. Velychko
significant (figure 5). In the case of δ < 0, two critical points appear at a certain distance; the
latter tends to zero at δ = δcrit ≈ −0.12|t′0| and the first order phase transitions at a further
increase of δ (figure 6) is suppressed.
Phase diagrams in the (|t′0|, µ) plane at various temperatures for δ > 0 are depicted in figure 7
with indication of tricritical points. In distinction to the standard two-level HCB model [34] (where
the SF phase transition is of the second order) the diagrams are asymmetric. In the limit T → 0
for µ > 0 the first order transition occurs at µ = (
√
δ −
√
|t′0|)2 (see the next section) whereas for
µ < 0 they are of the second order on the line µ = δ − |t′0|.
5. Phase separation at fixed boson concentration
Let us consider now the thermodynamics of the model at a fixed concentration of Bose particles.
We will utilize a connection between the concentration and the chemical potential of bosons which
can be established using its definition in such a form
n ≡ 〈nb
i + nc
i 〉 = 〈X11
i +X22
i 〉 (5.1)
or basing on the relationship
n = −∂(Ω/N)
∂µ
. (5.2)
In the first case similarly to equality (4.10) one can obtain a relation
X11
i +X22
i = X̃11
i + X̃00
i sin2 ϑ+ X̃22
i cos2 ϑ+ (X̃02
i + X̃20
i ) sinϑ cosϑ (5.3)
which results in
n = 〈X̃11
i 〉+ 〈X̃00
i 〉 sin2 ϑ+ 〈X̃22
i 〉 cos2 ϑ =
=
1
Z
{
e−βλ̃1 +
[
1
2
− δ − µ
2
√
(δ − µ)2 + 4(t′0ξ)
2
]
e−βλ̃0 +
[
1
2
+
δ − µ
2
√
(δ − µ)2 + 4(t′0ξ)
2
]
e−βλ̃2
}
.
(5.4)
In the second case, taking into account that
Ω/N = |t′0|ξ2 −Θ lnZ, Z = eβµ + e−β(δ−µ)/2 coshβ
√(
δ−µ
2
)2
+ (t′0ξ)
2 (5.5)
and differentiating with respect to µ, one can come to the same expression as (5.4).
There are different relationships between n and µ in NO and SF phases; in the last case, a
nonzero value of ξ (a solution of equation (4.13)) should be substituted into expression (5.4). Order
parameter ξ has a jump at the first order phase transition, so a stepwise change of concentration
n takes place. In the n = const regime (at the value of n in the region of step) it means a
phase separation into two phases with different concentrations: the NO phase (ξ = 0 and a larger
concentration of bosons) and the SF phase (ξ 6= 0 and their smaller concentration).
The above described situation is illustrated in figure 8, where the numerically calculated (Θ, n)
phase diagrams are presented. At δ > 0, phase separation region spans up to tricritical tempera-
tures. When δ goes to zero and finally reverses its sign, the shape of the separation region changes
in a peculiar way moving off abscissa axis (figure 8). Now the phase separation begins at nonzero
temperatures and vanishes at δ < δcrit; the line of the second order phase transition remains
only. At the further increase of |δ| (in the µ < 0 region) the (Θ, n) diagram becomes more and
more symmetric, approaching by its shape the diagram known for the usual HCB model [35] (see
also [36]).
Phase diagrams in the (|t′0|, n) coordinates are given in figure 9 where the regions of NO, SF
and separated phases are shown at various temperatures.
13004-10
Two-state Bose-Hubbard model in the HCB limit
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.1
0.2
0.3 δ = 0.1
δ = 0.6
2nd order PT
PS boundaries
T
em
pe
ra
tu
re
Θ
Concentration n
δ = 0.4
0.80 0.85 0.90 0.95 1.00
0.00
0.04
0.08
0.12
0.16
δ = 0.01
δ = 0
2nd order PT
PS boundaries
T
em
pe
ra
tu
re
Θ
Concentration n
δ = −0.1
Figure 8. Lines of the NO-SF phase transition and the phase separation region in the (Θ, n)
plane at various excitation energies δ including the case of small, zero and negative values of δ
(|t′0| = 1).
The case of the zero temperature can be studied more in detail in a pure analytic way. In this
limit there are three branches of order parameter ξ as a function of the chemical potential (see
figure 2):
(1) : ξ =
1
2|t′0|
√
|t′0|2 − (µ− δ)2,
(2) : ξ =
√
µδ/|t′0|,
(3) : ξ = 0. (5.6)
After elimination of ξ parameter, one can obtain the grand canonical potential Ω as follows:
(1) : Ω/N =
(µ− δ + |t′0|)2
4|t′0|
,
(2) : Ω/N = (δ/|t′0| − 1)µ,
(3) : Ω/N =
{
0, µ < 0,
−µ, µ > 0.
(5.7)
0.0 0.2 0.4 0.6 0.8 1.0
1
3
5
Phase
separation
NO
Θ = 0.3
2nd order PT
PS boundaries
H
op
pi
ng
e
ne
rg
y
|t 0'|
Concentration n
SF
0.0 0.2 0.4 0.6 0.8 1.0
1
3
5
Θ = 0.5
Θ = 0
Θ = 0.3
2nd order PT
PS boundaries
H
op
pi
ng
e
ne
rg
y
|t 0'|
Concentration n
Figure 9. Phase diagram with the indication of possible phases (above) and lines of the NO-SF
phase transition in the (|t′0|, µ) plane at various temperatures Θ (energy quantities are given in
units of δ).
13004-11
I.V. Stasyuk, O.V. Velychko
Differentiating expressions (5.7) with respect to µ we have
(1) : n =
1
2
+
µ− δ
2|t′0|
,
(2) : n = 1− δ/|t′0|,
(3) : n =
{
0, µ < 0,
1, µ > 0.
(5.8)
At the first order phase transition from the SF phase to the NO phase, the order parameter ξ
jumps from branch (1) to branch (3). This occurs at the µ = µ∗ ≡ (
√
|t′0| −
√
δ)2 value given by
equality of respective grand canonical potentials Ω(1) = Ω(3). Then boson system separates into
SF and NO phases with concentrations of bosons:
nSF =
1
2
+
µ∗ − δ
2|t′0|
,
nNO = 1. (5.9)
6. Discussion and conclusions
As was shown in this work, the transition to the SF phase (the phase with BE condensate) in
the Bose-Hubbard model with two local states (the ground and excited ones) on the lattice site
can be of the first order in the case, when the particle hopping takes place only in the excited
band. Calculations and estimates for optical lattices give evidence of significant distinction be-
tween hopping parameters t0 and t1 in the ground and excited bands, respectively. It follows from
estimates [25] that t1/t0 ≈ 30−50 depending on depth V0 of local potential wells (one can produce
effect on V0 changing the intensity of laser beams which create an optical lattice). Similar results
are obtained in the studies of quantum delocalization of the adsorbed hydrogen atoms. One can
see from calculations [22, 23] of energy spectrum of the H-atom subsystem on the nickel surface
that the ground-state band has a negligible bandwidth. At the same time, for excited bands, the
bandwidth varies in the range from 15 to 45 meV (depending on the excited state symmetry and
on the crystallographic orientation of metal surface), being mostly of the order of half the corre-
sponding excitation energy ∆εα = εα − ε0 . There are, however, the cases of strong delocalization
(e.g. H on the Ni(110) surface) where the excited bands overlap, and the width of the lowest one
is of the same order as ∆εα [22].
The values of hopping parameters greatly increase at the decrease of V0; the distance between
the local energy levels becomes smaller in this case (see [37, 38]). It is one of the possible ways
of changing the relation between the hopping parameters and excitation energy (|t′0| and δ in
our model). Another possibility (discussed in [39]) is connected with an essential reduction of the
energy gap between local s- and p-levels due to sufficiently strong interspecies Feshbach resonance
in the presence of Fermi atoms added to the Bose system in optical lattice.
Along with investigation of BE condensation in the excited band (or bands px, py (px, py, pz)
in two- (three-) dimensional case) on condition that certain concentration of Bose-atoms has been
created in the band by optical pumping [25, 38], an attempt was made in [40] to study the effect of
excited bands on the physics of BE condensation in the lowest (s-) band (when the s-band hopping
is taken into account). The case of finite values of the one-site interaction U was considered.
The possibility of the re-entrant behaviour of the MI-SF transition was claimed. However, the
order of phase transition was not investigated; the consideration was restricted to the case of zero
temperature. As we show in this work, re-entrant type dependence on T or µ takes place only for
spinodals and the return to the initial MI phase from the SF phase could be possible only in the
case of the second order phase transitions. In reality, the order of phase transition changes to the
first order in this region. In the HCB limit (no more than one particle per lattice site), it takes
place mainly at positive values of chemical potential of particles; at µ < 0, the transition remains,
for the most part, of the second order. The region of existence of SF phase is restricted, as a
13004-12
Two-state Bose-Hubbard model in the HCB limit
whole, to the interval −|t′0| < µ < |t′0|, while excitation energy should obey the inequality δ < |t′0|.
We have constructed the corresponding phase diagrams and established localization of tricritical
points, where the order of phase transition changes. The separation on SF and NO phases at the
fixed particle concentration is investigated; the conditions of the appearance of phase-separated
state are analyzed.
It should be mentioned that phase diagrams in figures 2–9 are close by their shape to the
diagrams obtained in the framework of Bose-Hubbard model for Bose atoms with spin S = 1 in
optical lattices [29]. The excited levels are formed in that case by the higher spin single-site states
and corresponding interactions of the “ferromagnetic” or “antiferromagnetic” type (the Hund-rule-
like splitting), while the hopping parameter is taken the same for all bands. The similarity of the
mentioned diagrams points out to the fact that the role of the excited states in the change of the
phase transition order in going to the phase with the BE condensate is the same in both cases.
Distinction, however, consists in another genesis of the single-site spectrum. In our model, in the
limiting case of HCB there are no effects connected with the level splitting due to interaction; the
excited single-particle states are taken by us into account instead.
The consideration developed in this work can be extended to the systems with the close or
degenerate excited local levels. Generalization of the model by adding inter-site interactions is
also important. It could even make it possible to take into consideration other phases (density-
modulated or supersolid) besides NO and SF ones.
We finally emphasize that the hopping parameter t′ij in the excited band can be positive; in
particular, this concerns the p-bands [39]. In such a situation, the condensation takes place into
states with wave vector ~Q on the boundary of the Brillouin zone, while the order parameters 〈cQ〉,
〈c+Q〉 describe the modulated condensate. Since t′Q = −t′0 , the results obtained in this work are
also valid (with |t′Q| in place of |t′0|) in that case.
References
1. Greiner M., Mandel O., Esslinger T., Hänsch T.W., Bloch I., Nature, 2002, 415, 39;
doi:10.1038/415039a.
2. Greiner M., Mandel O., Hänsch T.W., Bloch I., Nature, 2002, 419, 51; doi:10.1038/nature00968.
3. Jaksch D., Bruder C., Cirac J.I., Gardiner C.W., Zoller P., Phys. Rev. Lett., 1998, 81, 3108;
doi:10.1103/PhysRevLett.81.3108.
4. Sheshadri K., Krishnamurthy H.R., Pandit R., Ramakrishnan T.V., Europhys. Lett., 1993, 22, 257;
doi:10.1209/0295-5075/22/4/004.
5. Konabe S., Nikuni T., Nakamura M., Phys. Rev. A, 2006, 73, 033621;
doi:10.1103/PhysRevA.73.033621.
6. Ohashi Y., Kitaura M., Matsumoto H., Phys. Rev. A, 2006, 73, 033617;
doi:10.1103/PhysRevA.73.033617.
7. Freericks J.K., Monien H., Europhys. Lett., 1994, 26, 545; doi:10.1209/0295-5075/26/7/012.
8. Iskin M., Freericks J.K., Phys. Rev. A, 2009, 79, 053634; doi:10.1103/PhysRevA.79.053634.
9. Byczuk K., Vollhardt D., Phys. Rev. B, 2008, 77, 235106; doi:10.1103/PhysRevB.77.235106.
10. Anders P., Gull E., Pollet L., Troyer M., Werner P., Phys. Rev. Lett., 2010, 105, 096402;
doi:10.1103/PhysRevLett.105.096402.
11. Batrouni G.G., Scalettar R.T., Phys. Rev. B, 1992, 46, 9051; doi:10.1103/PhysRevB.46.9051.
12. Capogrosso-Sansone B., Söyler Ş.G., Prokof’ev N., Svistunov B., Phys. Rev. A, 2008, 77, 015602;
doi:10.1103/PhysRevA.77.015602.
13. Astaldi C., Bianco A., Modesti S., Tosatti E., Phys. Rev. Lett., 1992, 68, 90;
doi:10.1103/PhysRevLett.68.90.
14. Nishijima M., Okuyama H., Takagi N., Aruga T., Brenig W., Surface Science Reports, 2005, 57, 113;
doi:10.1016/j.surfrep.2005.03.001.
15. Reilly P.D., Harris R.A., Whaley K.B., J. Chem. Phys., 1991, 95, 8599; doi:10.1063/1.461239.
16. Ignatyuk V.V., Phys. Rev. E, 2009, 80, 041133; doi:10.1103/PhysRevE.80.041133.
17. Velychko O.V., Stasyuk I.V., Condens. Matter Phys., 2009, 12, 249.
18. Mysakovych T.S., Krasnov V.O., Stasyuk I.V., Ukr. J. Phys., 2010, 55, 228.
19. Micnas R., Ranninger J., Robaszkiewicz S., Rev. Mod. Phys., 1990, 62, 113;
doi:10.1103/RevModPhys.62.113.
13004-13
http://dx.doi.org/10.1038/415039a
http://dx.doi.org/10.1038/nature00968
http://dx.doi.org/10.1103/PhysRevLett.81.3108
http://dx.doi.org/10.1209/0295-5075/22/4/004
http://dx.doi.org/10.1103/PhysRevA.73.033621
http://dx.doi.org/10.1103/PhysRevA.73.033617
http://dx.doi.org/10.1209/0295-5075/26/7/012
http://dx.doi.org/10.1103/PhysRevA.79.053634
http://dx.doi.org/10.1103/PhysRevB.77.235106
http://dx.doi.org/10.1103/PhysRevLett.105.096402
http://dx.doi.org/10.1103/PhysRevB.46.9051
http://dx.doi.org/10.1103/PhysRevA.77.015602
http://dx.doi.org/10.1103/PhysRevLett.68.90
http://dx.doi.org/10.1016/j.surfrep.2005.03.001
http://dx.doi.org/10.1063/1.461239
http://dx.doi.org/10.1103/PhysRevE.80.041133
http://dx.doi.org/10.1103/RevModPhys.62.113
I.V. Stasyuk, O.V. Velychko
20. Mahan G.D., Phys. Rev. B, 1976, 14, 780; doi:10.1103/PhysRevB.14.780.
21. Stasyuk I.V., Dulepa I.R., J. Phys. Studies, 2009, 13, 2701 (in Ukrainian).
22. Puska M.J., Nieminen R.M., Surface Science, 1985, 157, 413; doi:10.1016/0039-6028(85)90683-1.
23. Brenig W., Surface Science, 1993, 291, 207; doi:10.1016/0039-6028(93)91492-8.
24. Müller T., Fölling S., Widera A., Bloch I., Phys. Rev. Lett., 2007, 99, 200405;
doi:10.1103/PhysRevLett.99.200405.
25. Isacsson A., Girvin S.M., Phys. Rev. A, 2005, 72, 053604; doi:10.1103/PhysRevA.72.053604.
26. Demler E., Zhou F., Phys. Rev. Lett., 2002, 88, 163001; doi:10.1103/PhysRevLett.88.163001.
27. Krutitsky K.V., Graham R., Phys. Rev. A, 2004, 70, 063610; doi:10.1103/PhysRevA.70.063610.
28. Kimura T., Tsuchiya S., Kurihara S., Phys. Rev. Lett., 2005, 94, 110403;
doi:10.1103/PhysRevLett.94.110403.
29. Pai R.V., Sheshadri K., Pandit R., Phys. Rev. B, 2008, 77, 014503; doi:10.1103/PhysRevB.77.014503.
30. Chen G.-H., Wu Y.-S., Phys. Rev. A, 2003, 67, 013606; doi:10.1103/PhysRevA.67.013606.
31. Hubbard J., Proc. R. Soc. Lond. A, 1965, 285, 542.
32. Haley S.B., Erdös P., Phys. Rev. B, 1972, 5, 1106; doi:10.1103/PhysRevB.5.1106.
33. Ohliger M., Pelster A., Green’s function approach to the Bose-Hubbard model, Preprint
arXiv:0810.4399v1 [cond-mat.stat-mech], 2008, 4 p.
34. Schmid G., Todo S., Troyer M., Dorneich A., Phys. Rev. Lett., 2002, 88, 167208;
doi:10.1103/PhysRevLett.88.167208.
35. Pedersen M.H., Schneider T., Phys. Rev. B, 1996, 53, 5826; doi:10.1103/PhysRevB.53.5826.
36. Stasyuk I.V., Mysakovych T.S., Condens. Matter Phys., 2009, 12, 539.
37. Bloch I., Dalibard J., Zwerger W., Rev. Mod. Phys., 2008, 80, 885; doi:10.1103/RevModPhys.80.885.
38. Scarola V.W., Das Sarma S., Phys. Rev. Lett., 2005, 95, 033003; doi:10.1103/PhysRevLett.95.033003.
39. Liu W.V., Wu C., Phys. Rev. A, 2006, 74, 013607; doi:10.1103/PhysRevA.74.013607.
40. Larson J., Collin A., Martikainen J.-P., Phys. Rev. A, 2009, 79, 033603;
doi:10.1103/PhysRevA.79.033603.
Двостанова модель Бозе-Хаббарда
в границi жорстких бозонiв
I.В. Стасюк, О.В. Величко
Iнститут фiзики конденсованих систем НАН України, вул. I. Свєнцiцького, 1, 79011 Львiв, Україна
Дослiджено перехiд у фазу з бозе-конденсатом у двозоннiй моделi Бозе-Хаббарда з переносом
частинок лише у збудженiй зонi. Розглянуто нестiйкiсть, пов’язану з таким переходом, який
вiдбувається при енергiях збудження δ < |t′
0
| (де |t′
0
| – параметр перескоку частинок). У
границi жорстких бозонiв виявлено зворотнiй хiд спiнодалей в областi додатних значень хiмiчного
потенцiалу. Встановлено, що рiд фазового переходу у цьому випадку змiнюється на перший i
повернення до нормальної фази насправдi не вiдбувається. Переходи першого роду iснують i при
вiд’ємнiй енергiї δ (за умови δ > δcrit ≈ −0.12|t′
0
|). При µ < 0 фазовий перехiд залишається
переважно переходом другого роду. Проаналiзовано хiд параметра порядку бозе-конденсату,
побудовано фазовi (Θ, µ) i (|t′
0
|, µ) дiаграми, встановлено локалiзацiю трикритичних точок. Знайдено
умови, при яких для фiксованої концентрацiї частинок наступає розшарування на нормальну фазу i
фазу з бозе-конденсатом.
Ключовi слова: модель Бозе-Хаббарда, жорсткi бозони, бозе-конденсацiя, збуджена зона
13004-14
http://dx.doi.org/10.1103/PhysRevB.14.780
http://dx.doi.org/10.1016/0039-6028(85)90683-1
http://dx.doi.org/10.1016/0039-6028(93)91492-8
http://dx.doi.org/10.1103/PhysRevLett.99.200405
http://dx.doi.org/10.1103/PhysRevA.72.053604
http://dx.doi.org/10.1103/PhysRevLett.88.163001
http://dx.doi.org/10.1103/PhysRevA.70.063610
http://dx.doi.org/10.1103/PhysRevLett.94.110403
http://dx.doi.org/10.1103/PhysRevB.77.014503
http://dx.doi.org/10.1103/PhysRevA.67.013606
http://dx.doi.org/10.1103/PhysRevB.5.1106
http://dx.doi.org/10.1103/PhysRevLett.88.167208
http://dx.doi.org/10.1103/PhysRevB.53.5826
http://dx.doi.org/10.1103/RevModPhys.80.885
http://dx.doi.org/10.1103/PhysRevLett.95.033003
http://dx.doi.org/10.1103/PhysRevA.74.013607
http://dx.doi.org/10.1103/PhysRevA.79.033603
Introduction
Two-state Bose-Hubbard model in RPA: normal phase
NO phase instability in HCB limit
Phase diagrams in MFA
Phase separation at fixed boson concentration
Discussion and conclusions
|