L. M. Abreu, C. de Calan, A. P. C. Malbouisson, J. M. C. Malbouisson, and A. E. Santana
Citation: Journal of Mathematical Physics 46, 012304 (2005); doi: 10.1063/1.1828589 View online: http://dx.doi.org/10.1063/1.1828589
View Table of Contents: http://scitation.aip.org/content/aip/journal/jmp/46/1?ver=pdfcov Published by the AIP Publishing
Critical behavior of the compactified
4theory
L. M. Abreu and C. de CalanCentre de Physique Théorique, Ecole Polytechnique, 91128 Palaiseau, France A. P. C. Malbouisson
CBPF/MCT, Rua Dr. Xavier Sigaud, 150, Rio de Janeiro RJ, Brazil J. M. C. Malbouissona)
Department of Physics, University of Alberta, Edmonton, AB, Canada, T6G 2J1 A. E. Santana
Instituto de Física, Universidade de Brasília, 70910-900, Brasília-DF, Brazil
(Received 30 June 2004; accepted 30 September 2004; published online 5 January 2005) We investigate the critical behavior of the N-component Euclidean4 model, in the large N limit, in three situations: confined between two parallel planes a dis-tance L apart from one another; confined to an infinitely long cylinder having a square transversal section of area L2; and to a cubic box of volume L3. Taking the mass term in the form m02=␣共T−T0兲, we retrieve Ginzburg–Landau models which
are supposed to describe samples of a material undergoing a phase transition, respectively, in the form of a film, a wire and of a grain, whose bulk transition temperature 共T0兲 is known. We obtain equations for the critical temperature as
functions of L and of T0, and determine the limiting sizes sustaining the
transition. © 2005 American Institute of Physics. [DOI: 10.1063/1.1828589]
I. INTRODUCTION
Models with fields confined in spatial dimensions play important roles both in field theory and in quantum mechanics. Relevant examples are the Casimir effect and superconducting films, where confinement is carried on by appropriate boundary conditions. For Euclidean field theories, imaginary time and the spatial coordinates are treated exactly on the same footing, so that an extended Matsubara formalism can be applied for dealing with the breaking of invariance along any one of the spatial directions.
Relying on this fact, in the present work we discuss the critical behavior of the Euclidean4 model compactified in one, two, and three spatial dimensions. We implement the spontaneous symmetry breaking by taking the bare mass coefficient in the Lagrangian parametrized as m02 =␣共T−T0兲, with␣⬎0 and the parameter T varying in an interval containing T0. With this choice,
considering the system confined between two parallel planes a distance L apart from one another, in an infinitely long square cylinder with transversal section area A = L2, and in a cube of volume V = L3, in dimension D = 3, we obtain Ginzburg–Landau models describing phase transitions in samples of a material in the form of a film, a wire and a grain, respectively, T0 standing for the bulk transition temperature. Such descriptions apply to physical circumstances where no gauge fluctuations need to be considered.
We start recapitulating the general procedure developed in Ref. 1 to treat the massive共4兲
D
theory in Euclidean space, compactified in a d-dimensional subspace, with d艋D. This permits to extend to an arbitrary subspace some results in the literature for finite temperature field theory2 and for the behavior of field theories in the presence of spatial boundaries.3,4We shall consider the vector N-component共4兲
DEuclidean theory at leading order in 1 / N, thus allowing for
nonper-a)Permanent address: Instituto de Física, Universidade Federal da Bahia, 40210-340, Salvador, BA, Brazil.
46, 012304-1
turbative results, the system being submitted to the constraint of compactification of a d-dimensional subspace. After describing the general formalism, we readdress the renormalization procedure we use treating the simpler situation of d = 1, which corresponds to the system confined between parallel planes(a film), analyzed in Ref. 5 for the case of two components, N=2. We then focus on two other particularly interesting cases of d = 2 and d = 3, in the three-dimensional Eu-clidean space, corresponding, respectively, to the system confined to an infinitely long cylinder with square transversal section(a wire) and to a finite cubic box (a grain). Extending the inves-tigation to these new cases demands further developments in the subject of multidimensional Epstein functions.
For these situations, in the framework of the Ginzburg–Landau model we derive equations for the critical temperature as a function of the confining dimensions. For a film, we show that the critical temperature decreases linearly with the inverse of the film thickness while, for a square wire and for a cubic grain, we obtain that the critical temperatures decrease linearly with the inverse of the side of the square and with the inverse of the edge of the cube, respectively, but with larger coefficients. In all cases, we are able to calculate the minimal system size (thickness, transversal section area, or volume) below which the phase transition does not take place.
II. THE COMPACTFIED MODEL
In this section we review the analytical methods of compactification of the N-component Euclidean4 model developed in Ref. 1 We consider the model described by the Hamiltonian density, H =1 2a a+ 1 2m¯0 2 aa+ N共aa兲 2, 共1兲
in Euclidean D-dimensional space, confined to a d-dimensional spatial rectangular box of sides Lj,
j = 1 , 2 , . . . , d. In the above equation is the renormalized coupling constant, m¯02 is a boundary-modified mass parameter depending on兵Li其 i=1,2, ... ,d, in such a way that
lim 兵Li其→⬁
m¯02共L1, . . . ,Ld兲 = m0
2共T兲 ⬅␣共T − T
0兲, 共2兲
m02共T兲 being the constant mass parameter present in the usual free-space Ginzburg–Landau model. In Eq.(2), T0represents the bulk transition temperature. Summation over repeated “color” indices
a is assumed. To simplify the notation in the following we drop out the color indices, summation over them being understood in field products. We will work in the approximation of neglecting boundary corrections to the coupling constant. A precise definition of the boundary-modified mass parameter will be given later for the situation of D = 3 with d = 1, d = 2, and d = 3, corresponding, respectively, to a film of thickness L1, to a wire of rectangular section L1⫻L2and to a grain of volume L1⫻L2⫻L3.
We use Cartesian coordinates r =共x1, . . . , xd, z兲, where z is a 共D−d兲-dimensional vector, with
corresponding momentum k =共k1, . . . , kd, q兲, q being a 共D−d兲-dimensional vector in momentum space. Then the generating functional of correlation functions has the form
Z =
冕
D†Dexp冉
−冕
0L
ddr
冕
dD−dzH共,ⵜ兲冊
, 共3兲 where L =共L1, . . . , Ld兲, and we are allowed to introduce a generalized Matsubara prescription,冕
dki 2→ 1 Lin兺
i=−⬁ +⬁ , ki→2ni Li , i = 1,2, . . . ,d. 共4兲A simpler situation is the system confined simultaneously between two parallel planes a distance L1apart from one another normal to the x1axis and two other parallel planes, normal to the x2axis
separated by a distance L2(a “wire” of rectangular section).
We start from the well-known expression for the one-loop contribution to the zero-temperature effective potential,6
U1共0兲 =
兺
s=1 ⬁ 共− 1兲s+1 2s 关120 2兴s冕
d Dk 共2兲D 1 共k2+ m2兲s, 共5兲where m is the physical mass and0is the normalized vacuum expectation value of the field(the
classical field). In the following, to deal with dimensionless quantities in the regularization pro-cedures, we introduce parameters
c = m 2, bi= 1 Li , g = 424−D, 0 2 = 0 2 D−2, 共6兲
whereis a mass scale. In terms of these parameters and performing the replacements (4), the one-loop contribution to the effective potential can be written in the form
U1共0,b1, . . . ,bd兲 =Db1¯ bd
兺
s=1 ⬁ 共− 1兲s 2s 关12g0 2兴s兺
n1,. . .,nd=−⬁ +⬁冕
dD−dq⬘
共b1 2n 1 2+ ¯ + b d 2n d 2+ c2+ q⬘
2兲s, 共7兲 where q⬘
= q / 2 is dimensionless. Using a well-known dimensional regularization formula7 to perform the integration over the共D−d兲 noncompactified momentum variables, we obtainU1共0,b1, . . . ,bd兲 =Db1¯ bd
兺
s=1 ⬁ f共D,d,s兲关12g02兴sAdc2冉
s −D − d 2 ;b1, . . . ,bd冊
, 共8兲 where f共D,d,s兲 =共D−d兲/2共− 1兲 s+1 2s⌫共s兲⌫冉
s − D − d 2冊
共9兲 and Adc2共;b1, . . . ,bd兲 =兺
n1,. . .,nd=−⬁ +⬁ 共b1 2n 1 2+ ¯ + b d 2n d 2+ c2兲− = 1 c2+ 2兺
i=1 d兺
ni=1 ⬁ 共bi 2 ni2+ c2兲− + 22兺
i⬍j=1 d兺
ni,nj=1 ⬁ 共bi 2n i 2+ b j 2n j 2+ c2兲−+ ¯ + 2d兺
n1,. . .,nd=1 ⬁ 共b1 2n 1 2+ ¯ + b d 2 nd2+ c2兲−. 共10兲Next we can proceed generalizing to several dimensions the mode-sum regularization pre-scription described in Ref. 8. This generalization has been done in Ref. 1 and we briefly describe here its principal steps. From the identity
1 ⌬= 1 ⌫共兲
冕
0 ⬁ dt t−1e−⌬t, 共11兲and using the following representation for Bessel functions of the third kind, K,
2共a/b兲/2K共2
冑
ab兲 =冕
0
⬁
dx x−1e−共a/x兲−bx, 共12兲 we obtain after some rather long but straightforward manipulations,1
Adc2共;b1, . . . ,bd兲 = 2−共d/2兲+12−共d/2兲 b1¯ bd⌫共兲
冋
2−共d/2兲−1⌫冉
−d 2冊
共2c兲 d−2 + 2兺
i=1 d兺
ni=1 ⬁冉
ni 2cbi冊
−共d/2兲 K−共d/2兲冉
2cni bi冊
+ ¯ + 2d兺
n1,. . .,nd=1 ⬁冉
1 2c冑
n12 b12+ ¯ + nd2 bd2冊
−共d/2兲 K−共d/2兲冉
2c冑
n1 2 b12+ ¯ + nd2 bd2冊
册
. 共13兲 Taking= s −共D−d兲/2 in Eq. (13) and inserting it in Eq. (8), we obtain the one-loop correction to the effective potential in D dimensions with a compactified d-dimensional subspace in the form (recovering the dimensionful parameters)U1共0,L1, . . . ,Ld兲 =
兺
s=1 ⬁ 关12g0 2兴sh共D,s兲冋
2s−共D/2兲−2⌫冉
s −D 2冊
m D−2s +兺
i=1 d兺
ni=1 ⬁冉
m Lini冊
共D/2兲−s K共D/2兲−s共mLini兲 + 2兺
i⬍j=1 d兺
ni,nj=1 ⬁冉
m冑
Li2ni2+ Lj2n2j冊
共D/2兲−s K共D/2兲−s共m冑
Li2n2i+ L2jnj2兲 + ¯ + 2d−1兺
n1,. . .,nd=1 ⬁冉
m冑
L12n12+ ¯ + Ld2nd2冊
共D/2兲−s K共D/2兲−s共m冑
L12n12+ ¯ + Ld2nd2兲册
, 共14兲 with h共D,s兲 = 1 2D/2+s−1D/2 共− 1兲s+1 s⌫共s兲 . 共15兲Criticality is attained when the inverse squared correlation length,−2共L1, . . . , Ld,0兲, vanishes
−2共L 1, . . . ,Ld,0兲 = m¯0 2+ 12 0 2+ 24 L1¯ Ldn1,. . .,n
兺
d=−⬁ ⬁冕
dD−dq 共2兲D−d ⫻ 1 q2+冉
2n1 L1冊
2 + ¯ +冉
2nd Ld冊
2 +−2共L1, . . . ,Ld,0兲 , 共16兲where 0 is the normalized vacuum expectation value of the field (different from zero in the
ordered phase). In the disordered phase,0vanishes and the inverse correlation length equals the
physical mass, given below by Eq.(18). Recalling the condition,
冏
20
2U共D,L1,L2兲
冏
0=0
= m2, 共17兲
where U is the sum of the tree-level and one-loop contributions to the effective potential (remem-bering that at the large-N limit it is enough to take the one-loop contribution to the mass), we obtain m2共L1, . . . ,Ld兲 = m¯0 2共L 1, . . . ,Ld兲 + 24 共2兲D/2
冋
兺
i=1 d兺
ni=1 ⬁冉
m Lini冊
共D/2兲−1 K共D/2兲−1共mLini兲 + 2兺
i⬍j=1 d兺
ni,nj=1 ⬁冉
m冑
Li2ni2+ Lj2n2j冊
共D/2兲−1 K共D/2兲−1共m冑
Li2n2i+ Lj2n2j兲 + ¯ + 2d−1兺
n1,. . .,nd=1 ⬁冉
m冑
L12n12+ ¯ + Ld2nd2冊
共D/2兲−1 K共D/2兲−1共m冑
L12n12+ ¯ + Ld2nd2兲册
. 共18兲 Notice that, in writing Eq.(18), we have suppressed the parcel 2−共D/2兲−1⌫关1−共D/2兲兴mD−2from its square brackets, the parcel that emerges from the first term in the square bracket of Eq.(14). This expression, which does not depend explicitly on Li, diverges for D even due to the poles of thegamma function; in this case, this parcel is subtracted to get a renormalized mass equation. For D odd,⌫关1−共D/2兲兴 is finite but we also subtract this term (corresponding to a finite renormalization) for sake of uniformity; besides, for D艌3, the factor mD−2does not contribute in the criticality.
The vanishing of Eq. (18) defines criticality for our compactified system. We claim that, taking d = 1, d = 2, and d = 3 with D = 3, we are able to describe, respectively, the critical behavior of samples of materials in the form of films, wires, and grains. Notice that the parameter m on the right-hand side of Eq. (18) is the boundary-modified mass m共L1, . . . , Ld兲, which means that Eq.
(18) is a self-consistency equation, a very complicated modified Schwinger–Dyson equation for the mass, not soluble by algebraic means. Nevertheless, as we will see in the next sections, a solution is possible at criticality, which allows us to obtain a closed formula for the boundary-dependent critical temperature.
III. CRITICAL BEHAVIOR FOR FILMS
We now consider the simplest particular case of the compactification of only one spatial dimension, with the system confined between two parallel planes a distance L apart from one another. This case, which has already been considered in Ref. 5 concerning with the two-component model, is reanalyzed here to set the required renormalization procedure in the proper large-N grounds and, also, for the sake of completeness. Thus, from Eq.(18), taking d=1, we get in the disordered phase
m2共L兲 = m¯02共L兲 + 24 共2兲D/2
兺
n=1 ⬁冉
m nL冊
共D/2兲−1 K共D/2兲−1共nLm兲, 共19兲 where L共=L1兲 is the separation between the planes, the film thickness. If we limit ourselves to theneighborhood of criticality共m2⬇0兲 and consider L finite and sufficiently small, we may use an
asymptotic formula for small values of the argument of Bessel functions,
K共z兲 ⬇1 2⌫共兩兩兲
冉
z 2冊
−兩兩 共z ⬇ 0兲, 共20兲and Eq.(19) reduces, for D⬎3, to
m2共L兲 ⬇ m¯02共L兲 + 6
D/2LD−2⌫
冉
D
2 − 1
冊
共D − 2兲, 共21兲where共D−2兲 is the Riemann zeta-function, defined for Re兵D−2其⬎1 by the series
共D − 2兲 =
兺
n=1
⬁ 1
nD−2. 共22兲
It is worth mentioning that for D = 4, taking m2共L兲=0 and making the appropriate changes 共L
→,→/4!兲, Eq. (21) is formally identical to the high-temperature (low values of) critical equation obtained in Ref. 9, thus providing a check of our calculations.
For D = 3, Eq. (21) can be made physically meaningful by a regularization procedure as follows. We consider the analytic continuation of the zeta-function, leading to a meromorphic function having only one simple pole at z = 1, which satisfies the reflection formula
共z兲 = 1 ⌫共z/2兲⌫
冉
1 − z 2
冊
z−1/2共1 − z兲. 共23兲
Next, remembering the formula
lim
z→1
冋
共z兲 −1
z − 1
册
=␥, 共24兲where␥⬇0.5772 is the Euler–Mascheroni constant, we define the L-dependent bare mass for D ⬇3, in such a way that the pole at D=3 in Eq. (21) is suppressed, that is we take
m¯02共L兲 ⬇ M − 1 共D − 3兲
6
L, 共25兲
where M is independent of D. To fix the finite term, we make the simplest choice satisfying(2),
M = m02共T兲 =␣共T − T0兲, 共26兲
T0being the bulk critical temperature. In this case, using Eq.(25) in Eq. (21) and taking the limit
as D→3, the L-dependent renormalized mass term in the vicinity of criticality becomes
m2共L兲 ⬇␣共T − Tc共L兲兲, 共27兲
where the modified, L-dependent, transition temperature is given by
Tc共L兲 = T0− C1
␣L, 共28兲
C1=6␥
⬇ 1.1024. 共29兲
From this equation, we see that for L smaller than
Lmin= C1
␣T0
, 共30兲
Tc共L兲 becomes negative, meaning that the transition does not occur.
5
IV. CRITICAL BEHAVIOR FOR WIRES
We now focus on the situation where two spatial dimensions are compactified. From Eq.(18), taking d = 2, we get(in the disordered phase)
m2共L 1,L2兲 = m¯0 2共L 1,L2兲 + 24 共2兲D/2
冋
兺
n=1 ⬁冉
m nL1冊
共D/2兲−1 K共D/2兲−1共nL1m兲 +兺
n=1 ⬁冉
m nL2冊
共D/2兲−1 K共D/2兲−1共nL2m兲 + 2兺
n1,n2=1 ⬁冉
m冑
L12n12+ L22n22冊
共D/2兲−1 K共D/2兲−1共m冑
L12n12+ L22n22兲册
. 共31兲 If we limit ourselves to the neighborhood of criticality, m2⬇0, and taking both L1and L2finite and
sufficiently small, we may use Eq.(20) to rewrite Eq. (31) as
m2共L1,L2兲 ⬇ m¯02共L1,L2兲 + 6 D/2⌫
冉
D 2 − 1冊
冋
冉
1 L1D−2+ 1 L2D−2冊
共D − 2兲 + 2E2冉
D − 2 2 ;L1,L2冊
册
, 共32兲 where E2关共D−2兲/2;L1, L2兴 is the generalized (multidimensional) Epstein zeta-function defined byE2
冉
D − 2 2 ;L1,L2冊
=n1兺
,n2=1 ⬁ 关L1 2 n12+ L22n22兴−关共D−2兲/2兴, 共33兲 for Re兵D其⬎3.As mentioned before, the Riemann zeta-function 共D−2兲 has an analytical extension to the whole complex D-plane, having a unique simple pole (of residue 1) at D=3. One can also construct analytical continuations (and recurrence relations) for the multidimensional Epstein functions which permit to write them in terms of Kelvin and Riemann zeta-functions. To start one considers the analytical continuation of the Epstein–Hurwitz zeta-function given by8
兺
n=1 ⬁ 共n2+ p2兲−= −1 2p −2+冑
2p2−1⌫共兲冋
⌫冉
− 1 2冊
+ 4兺
n=1 ⬁ 共pn兲−1/2K−1/2共2pn兲册
. 共34兲 Using this relation to perform one of the sums in(33) leads immediately to the question of which sum is first evaluated. As it is done in Ref. 10, whatever the sum one chooses to perform first, the manifest L1↔L2symmetry of Eq.(33) is lost; to overcome such an obstacle, in order to preservethis symmetry, we adopt here a symmetrized summation generalizing the prescription introduced in Ref. 1 for the case of many variables.
To derive an analytical continuation and symmetrized recurrence relations for the multidimen-sional Epstein functions, we start by taking these functions defined as the symmetrized summa-tions
Ed共;L1, . . . ,Ld兲 = 1 d!
兺
n兺
1=1 ⬁ ¯兺
nd=1 ⬁ 关1 2n 1 2+ ¯ + d 2n d 2兴−, 共35兲wherei=共Li兲, withrunning in the set of all permutations of the parameters L1, . . . , Ld, and the
summations over n1, . . . , ndbeing taken in the given order. Applying(34) to perform the sum over
nd, one gets Ed共;L1, . . . ,Ld兲 = − 1 2d
兺
i=1 d Ed−1共; . . . ,Lˆ, ... 兲 +i冑
2 d⌫共兲⌫冉
− 1 2冊
兺
i=1 d 1 LiEd−1冉
− 1 2; . . . ,Lˆ, ...i冊
+ 2冑
d⌫共兲Wd冉
− 1 2,L1, . . . ,Ld冊
, 共36兲where the hat over the parameter Liin the functions Ed−1means that it is excluded from the set
兵L1, . . . , Ld其 (the others being the d−1 parameters of Ed−1), and
Wd共;L1, . . . ,Ld兲 =
兺
i=1 d 1 Lin1,. . .,n兺
d=1 ⬁冉
ni Li冑
共¯ + Li 2 ni2 ˆ + ¯ 兲冊
K冉
2ni Li冑
共¯ + Li 2n i 2 ˆ + ¯ 兲冊
, 共37兲 with 共¯+Lˆi2ni2+¯兲 representing the sum 兺dj=1L2jn2j− Li2ni2. In particular, noticing that E1共; Lj兲 = Lj−2共2兲, one finds E2冉
D − 2 2 ;L1 2,L 2 2冊
= −1 4冉
1 L1D−2+ 1 L2D−2冊
共D − 2兲 +冑
⌫冉
D − 3 2冊
4⌫冉
D − 2 2冊
冉
1 L1L2 D−3+ 1 L1D−3L2冊
共D − 3兲 +冑
⌫冉
D − 2 2冊
W2冉
D − 3 2 ;L1,L2冊
, 共38兲which is a meromorphic function of D, symmetric in the parameters L1 and L2 as Eq. (33)
suggests.
Using the above expression, Eq.(32) can be rewritten as
m2共L1,L2兲 ⬇ m¯02共L1,L2兲 + 3 D/2
冋
冉
1 L1D−2+ 1 L2D−2冊
⌫冉
D − 2 2冊
共D − 2兲 +冑
冉
1 L1L2D−3+ 1 L1D−3L2冊
⌫冉
D − 3 2冊
共D − 3兲 + 2冑
W2冉
D − 3 2 ;L1,L2冊
册
. 共39兲 This equation presents no problems for 3⬍D⬍4 but, for D=3, the first and second terms between the square brackets of Eq.(39) are divergent due to the-function and⌫-function, respectively. We can deal with divergences remembering the property in Eq.(24) and using the expansion of ⌫关共D−3兲/2兴 around D=3,⌫
冉
D − 3 2冊
⬇2
D − 3+⌫
⬘
共1兲, 共40兲⌫
⬘
共z兲 standing for the derivative of the ⌫-function with respect to z. For z=1 it coincides with the Euler digamma-function共1兲, which has the particular value共1兲=−␥. We notice however, that differently from the case treated in the preceding section, where a renormalization procedure wasneeded, here the two divergent terms generated by the use of formulas(24) and (40) cancel exactly between them. No renormalization is needed. Thus, for D = 3, taking the bare mass given by m¯02共L
1, L2兲=␣共T−T0兲, we obtain the renormalized boundary-dependent mass term in the form
m2共L
1,L2兲 ⬇␣共T − Tc共L1,L2兲兲, 共41兲
with the boundary-dependent critical temperature given by
Tc共L1,L2兲 = T0− 9␥ 2␣
冉
1 L1 + 1 L2冊
− 6 ␣W2共0;L1,L2兲, 共42兲 where W2共0;L1,L2兲 =兺
n1,n2=1 ⬁再
1 L1 K0冉
2 L2 L1 n1n2冊
+ 1 L2 K0冉
2 L1 L2 n1n2冊
冎
. 共43兲The quantity W2共0;L1, L2兲, appearing in Eq. (42), involves complicated double sums, very
difficult to handle for L1⫽L2; in particular, it is not possible to take limits such as Li→⬁. For this
reason we will restrict ourselves to the case L1= L2. For a wire with the square transversal section,
we have L1= L2= L =
冑
A and Eq. (42) reduces toTc共A兲 = T0− C2
␣
冑
A, 共44兲where C2is a constant given by
C2= 9␥ + 12 n1
兺
,n2=1 ⬁ K0共2n1n2兲 ⬇ 1.6571. 共45兲We see that the critical temperature of the square wire depends on the bulk critical temperature and the Ginzburg–Landau parameters ␣ and (which are characteristics of the material consti-tuting the wire), and also on the area of its cross section. Since Tc decreases linearly with the
inverse of the side of the square, this suggests that there is a minimal area for which Tc共Amin兲
= 0, Amin=
冉
C2 ␣T0冊
2 ; 共46兲for square wires of the transversal section areas smaller than this value, in the context of our model the transition should be suppressed. On topological grounds, we expect that(apart from appropri-ate coefficients) our result should be independent of the transverse section shape of the wire, at least for transversal sectional regular polygons.
V. CRITICAL BEHAVIOR FOR GRAINS
We now turn our attention to the case where all three spatial dimensions are compactified, corresponding to the system confined in a box of sides L1, L2, L3. Taking d = 3 in Eq.(18) and
using Eq.(20), we obtain (for sufficiently small L1, L2, L3and in the neighborhood of classicality,
m2共L1,L2,L3兲 ⬇ m¯0 2共L 1,L2,L3兲 + 6 D/2⌫
冉
D − 2 2冊
冋
兺
i=1 3 共D − 2兲 LiD−2 + 2i⬍j=1兺
3 E2冉
D − 2 2 ;Li,Lj冊
+ 4E3冉
D − 2 2 ;L1,L2,L3冊
册
, 共47兲 where E3共; L1, L2, L3兲=兺n⬁1,n2,n3=1关L1 2n12+ L22n22+ L23n32兴−and the functions E2are given by Eq.(38).
The analytical structure of the function E3关共D−2兲/2;L1, L2, L3兴 can be obtained from the
general symmetrized recurrence relation given by Eqs.(36) and (37); explicitly, one has
E3
冉
D − 2 2 ;L1,L2,L3冊
= − 1 6i⬍j=1兺
3 E2冉
D − 2 2 ;Li,Lj冊
+冑
⌫冉
D − 3 2冊
6⌫冉
D − 2 2冊
兺
i,j,k=1 3 共1 + ijk兲 2 1 Li ⫻E2冉
D − 2 2 ;Lj,Lk冊
+ 2冑
3⌫冉
D − 2 2冊
W3冉
D − 3 2 ;L1,L2,L3冊
, 共48兲 whereijkis the totally antisymmetric symbol and the function W3is a particular case of Eq.(37).Using Eqs.(38) and (48), the boundary dependent mass can be written as
m2共L1,L2,L3兲 ⬇ m¯0 2共L 1,L2,L3兲 + 6 D/2
冋
1 3⌫冉
D − 2 2冊
兺
i=1 3 1 LiD−2共D − 2兲 +冑
6 共D − 3兲 ⫻兺
i⬍j=1 3冉
1 LiD−3Lj + 1 LD−3j Li冊
⌫冉
D − 3 2冊
+ 4冑
3 i⬍j=1兺
3 W2冉
D − 3 2 ;Li,Lj冊
+ 6共D − 4兲⌫冉
D − 4 2冊
i,j,k=1兺
3 共1 + ijk兲 2 1 Li冉
1 LD−4j Lk + 1 LkD−4Lj冊
+2 3 i,j,k=1兺
3 共1 + ijk兲 2 1 Li W2冉
D − 4 2 ;Lj,Lk冊
+ 8冑
3 W3冉
D − 3 2 ;L1,L2,L3冊
册
. 共49兲 The first two terms in the square brackets of Eq.(49) diverge as D→3 due to the poles of the ⌫ and-functions. However, as it happens in the case of wires, using Eqs.(24) and (40) it can be shown that these divergences cancel exactly one another. After some simplifications, for D = 3, the boundary dependent mass(49) becomesm2共L1,L2,L3兲 ⬇ m¯02共L1,L2,L3兲 +6
冋
␥ 2兺
i=1 3 1 Li +4 3i⬍j=1兺
3 W2共0;Li,Lj兲 + 18i,j,k=1兺
3 共1 + ijk兲 2 Li LjLk +2冑
3 i,j,k=1兺
3 共1 + ijk兲 2 1 Li W2冉
− 1 2;Lj,Lk冊
+ 8 3W3共0;L1,L2,L3兲册
. 共50兲 As before, since no divergences need to be suppressed, we can take the bare mass given by m¯02共L1, L2, L3兲=␣共T−T0兲 and rewrite the renormalized mass as m2共L1, L2, L3兲⬇␣共T
− Tc共L1, L2, L3兲兲. The expression of Tc共L1, L2, L3兲 can be easily obtained from Eq. (50), but it is a very complicated formula, involving multiple sums, which makes almost impossible a general
analytical study for arbitrary parameters L1, L2, L3; thus, we restrict ourselves to the situation
where L1= L2= L3= L, corresponding to a cubic box of volume V = L3. In this case, the boundary
dependent critical temperature reduces to
Tc共V兲 = T0− C3
␣V1/3, 共51兲
where the constant C3is given by[using that K−1/2共z兲=
冑
/ 2ze−z]C3= 1 + 9␥ + 12 n1
兺
,n2=1 ⬁ e−2n1n2 n1 + 48 n1兺
,n2=1 ⬁ K0共2n1n2兲 + 48 n1,n兺
2,n3=1 ⬁ K0共2n1冑
n2 2 + n3 2兲 ⬇ 2.7657. 共52兲 One sees that the minimal volume of the cubic grain sustaining the transition isVmin=
冉
C3 ␣T0冊
3 . 共53兲 VI. CONCLUSIONSIn this paper we have discussed the spontaneous symmetry breaking of the 共4兲
D theory
compactified in d艋D Euclidean dimensions, extending some results of Ref. 1. We have param-etrized the bare mass term in the form m02共T−T0兲, thus placing the analysis within the Ginzburg–
Landau framework. We focused on the situations with D = 3 and d = 1 , 2 , 3, corresponding(in the context of condensed matter systems) to films, wires, and grains, respectively, undergoing phase transitions which may be described by(mean-field) Ginzburg–Landau models. This generalizes to more compactified dimensions of previous investigations on the superconducting transition in films, both without5 and in the presence of a magnetic field.11 In all cases studied here, in the absence of gauge fluctuations, we found that the boundary-dependent critical temperature de-creases linearly with the inverse of the linear dimension L: Tc共L兲=T0− Cd/␣L, where␣and are
the Ginzgurg–Landau parameters, T0is the bulk transition temperature, and Cdis a constant equal
to 1.1024, 1.6571, and 2.6757 for d = 1(film), d=2 (square wire), and d=3 (cubic grain), respec-tively. Such behavior suggests the existence of a minimal size of the system below which the transition is suppressed. It is worth mentioning that having the transition temperature scaling with the inverse of the relevant length L for all the cases analyzed (films, wires, and grains) is in accordance with what one learns from finite-size scaling arguments.12
These findings seems to be in qualitative agreement with results for the existence of a minimal thickness for disappearance of superconductivity in films.13–16 Experimental investigations in nanowires searching to establish whether there is a limit to how thin a superconducting wire can be, while retaining its superconducting character, have also drawn the attention of researchers; for example, in Ref. 17 the behavior of nanowires has been studied. Similar questions have also been raised concerning the behavior of superconducting nanograins.18,19 Nevertheless, an important point to be emphasized is that our results are obtained in a field-theoretical framework and do not depend on microscopic details of the material involved nor account for the influence of manufac-turing aspects of the sample; in other words, our results emerge solely as a topological effect of the compactification of the Ginzburg–Landau model in a subspace. Detailed microscopic analysis is required if one attempts to account quantitatively for experimental observations which might deviate from our mean field results.
ACKNOWLEDGMENTS
This work was partially supported by CAPES and CNPq (Brazilian agencies). L.M.A. and A.P.C.M. are grateful for kind hospitality at Centre de Physique Théorique/Ecole Polytechnique, and J.M.C.M. acknowledges the hospitality at Department of Physics/University of Alberta, where part of this work has been done.
1
A.P.C. Malbouisson, J.M.C. Malbouisson, and A.E. Santana, Nucl. Phys. B 631, 83(2002).
2
C. de Calan, A.P.C. Malbouisson, and N.F. Svaiter, Mod. Phys. Lett. A 13, 1757(1998).
3
A.P.C. Malbouisson and J. M.C. Malbouisson, J. Phys. A 35, 2263(2002).
4
L. Da Rold, C.D. Fosco, and A.P.C. Malbouisson, Nucl. Phys. B 624, 485(2002).
5
A.P.C. Malbouisson, J.M.C. Malbouisson, and A.E. Santana, Phys. Lett. A 318, 406(2003).
6
C. Itzykson and J-B. Zuber, Quantum Field Theory(McGraw-Hill, New York, 1980), p. 451.
7
J. Zinn-Justin, Quantum Field Theory and Critical Phenomena(Clarendon, Oxford, 1996).
8
A. Elizalde and E. Romeo, J. Math. Phys. 30, 1133(1989).
9
L. Dolan and R. Jackiw, Phys. Rev. D 9, 3320(1974), Eq. (3.19b).
10
K. Kirsten, J. Math. Phys. 35, 459(1994).
11
L.M. Abreu, A.P.C. Malbouisson, J.M.C. Malbouisson, and A.E. Santana, Phys. Rev. B 67, 212502(2003).
12
See Ref. 7, Chap. 36 and references therein.
13
J. Simonin, Phys. Rev. B 33, 7830(1986).
14
J. H. Quateman, Phys. Rev. B 34, 1948(1986).
15
M. S. M. Minhaj, S. Meepagala, J. T. Chen, and L. E. Wenger, Phys. Rev. B 49, 15 235(1994).
16
A. V. Pogrebnyakov et al., Appl. Phys. Lett. 82, 4319(2003).
17
A. Bezryadin, C.N. Lau, and M. Tinkham, Nature(London) 404, 971 (2000); N. Markovic, C.N. Lau, and M. Tinkham,
Physica C 387, 44(2003).
18
D.C. Ralph, C.T. Black, and M. Tinkham, Phys. Rev. Lett. 76, 688(1996); 78, 4087 (1997).
19