Physical regularization for the spin-1/2 Aharonov-Bohm problem in conical space

F. M. Andrade,1, ∗ E. O. Silva,2, † and M. Pereira1, ‡

1
Departamento de Matemática e Estatı́stica, Universidade Estadual de Ponta Grossa, 84030-900 Ponta Grossa-PR, Brazil
2
Departamento de Fı́sica, Universidade Federal do Maranhão,
Campus Universitário do Bacanga, 65085-580 São Luı́s-MA, Brazil
(Dated: February 24, 2012)
We examine the bound state and scattering problem of a spin-one-half particle undergone to an
Aharonov-Bohm potential in a conical space in the nonrelativistic limit. The crucial problem of
the δ-function singularity coming from the Zeeman spin interaction with the magnetic flux tube is
solved through the self-adjoint extension method. Using two different approaches already known in
the literature, both based on the self-adjoint extension method, we obtain the self-adjoint extension
arXiv:1112.0265v2 [quant-ph] 23 Feb 2012

parameter to the bound state and scattering scenarios in terms of the physics of the problem. It is
shown that such a parameter is the same for both situations. The method is general and is suitable
for any quantum system with a singular Hamiltonian that has bound and scattering states.

PACS numbers: 03.65.Ge, 03.65.Db, 98.80.Cq, 03.65.Pm

Singularities are very common in quantum mechanics ing from a nontrivial conical geometry) through the met-
and already have a long history [1]. The first work with ric. From these studies, such materials could be analyzed
δ-like singularities was in the Kronig-Penny model [2] for through theoretical models allowing to include point in-
the description of the band energy in solid-state physics. teractions able to reproduce AB-like effects.
In addition, point interactions [3–5] have been of great In quantum mechanics, singularities and pathological
interest in various branches of physics for their relevance potentials, in general, are dealt with some kind of reg-
as solvable models [6]. In the Aharonov-Bohm (AB) ef- ularization procedure. A common approach to ensure
fect [7] of spin-1/2 particles [8–10] a two-dimensional δ that the wave function in the presence of a singularity is
function appears as the mathematical description of the square-integrable (and therefore might be associated to
Zeeman interaction between the spin and the magnetic a bound state) is to force it to vanish on the singularity.
flux tube [11, 12]. Hagen [9] argued that a δ-function con- More appropriately, an analysis based on the self-adjoint
tribution to the potential can not be neglected when the extension method [19], broadens the boundary condition
system has spin, having shown that changes in the am- possibilities that still give bound states. The physics of
plitude and scattering cross section arise when the spin the problem determines which of these possibilities is the
of the particle is considered. Point interactions usually right one, leaving no ambiguities [8, 20]. This method
appear in quantum systems in the presence of topological has been applied by many authors, in particular, for AB-
defects. A simple but nontrivial example is the case of a like systems [8, 12, 21–24]. However, the results obtained
cone rising from an effective geometry immersed in sev- in these works present the most important results (e.g.,
eral physical systems, such as cosmic strings [13], defects energy spectrum, phase shift, S matrix) in terms of an
in elastic media [14], defects in liquid crystals [15], and arbitrary real parameter, the so called self-adjoint exten-
so on. In such systems, although the particle does not sion parameter.
have access to the core (defect) region, its wave function In this article, we describe a general regularization
and energy spectrum are truly influenced by it. procedure to obtain the self-adjoint extension parame-
ter, based on the physics of the spin-1/2 AB system in
Recently, a device was proposed that would detect
(1 + 2)-dimensional conical space for both bound and
microstresses in graphene [16] based on a scanning-
scattering scenarios. We take as a starting point the
tunneling-microscopy setup able to measure AB interfer-
works of Kay-Studer (KS) [25] and Bulla-Gesztezy (BG)
ences at the nanometer scale. In this setup a δ-function
[26], both based on the self-adjoint extension method.
scattering potential was considered in the continuum
limit [17]. In Ref. [18] it was considered a topological The topological defect considered here is a linear quan-
insulator nanowire with a magnetic field applied along tity that appears embedded in the metric system ds2 =
its length, focusing on the AB conductance oscillations dr2 + α2 r2 dϕ2 , where r ≥ 0, 0 ≤ ϕ < 2π, and α is the
arising from the surface states. The Dirac Hamiltonian parameter which effectively introduces an angular excess
of this model takes into account the spinorial connection or deficit, identified by 2π(1 − α). The above metric has
that allows us to incorporate topological defects (aris- a conelike singularity at r = 0. In other words, the cur-
vature tensor of this metric, considered as a distribution,
12
is given by R12 = R11 = R22 = 2π(α − 1)δ(r)/α, where
δ(r) is the two-dimensional δ function in flat space [27].
∗ Electronic address: [email protected] This implies a two-dimensional conical singularity sym-
† Electronic address: [email protected] metrical in the z axis, which characterizes it as a linear
‡ Electronic address: [email protected] defect.
 2

In order to study the dynamics of the particle in B along the z-axis, i.e., a spin-1/2 AB problem in the
a nonflat spacetime, we should include the spin con- conical space. Let us consider a conical defect with a
nection in the differential operator and define the re- nucleus with radius r0 , so it is suitable to write Ushort (r)
spective Dirac matrices in this manifold. This system as [9, 11]
is governed by the modified Dirac equation in curved
space [iγ µ (∂µ + Γµ ) − qγ µ Aµ − M ]ψ(x) = 0, where q qsφ δ(r − r0 )
U short (r) = , (7)
is the charge, M is mass of the particle, ψ(x) is a four- 2M α r0
component spinorial wave function and Γµ is the spin and, at the end, the limit r0 → 0 is taken. Although
connection. The only nonvanishing spin connection in the functional structure of Ushort and U short are quite
this case is Γϕ = i(1 − α)σz /2, while the Dirac matrices different, as discussed in [9], we are free to use any form
are conveniently defined as αi = γ 0 γ i , β = γ 0 [28, 29]. of potential provided that only the contribution of the
The magnetic flux tube in the background space de- form (6) is excluded.
scribed by the metric above considered is related [29] to The operator H0 , with domain D(H0 ), is self-adjoint
δ(r)
the magnetic field sB = s(∇ × A) = sφ α r ẑ (where if D(H0† ) = D(H0 ) and H0† = H0 . For smooth functions,
φ = φ/2π is the flux parameter), while the vector poten- g ∈ C0∞ (R2 ) with g(0) = 0, we should have Hg = H0 g,
φ and hence it is reasonable to interpret the Hamiltonian
tial in the Coulomb gauge is Aϕ = αr ϕ̂, with s = ±1
being twice the spin projection parameter. The parame- (4) as a self-adjoint extension of H0 |C0∞ (R2 \{0}) [30–32].
ter s implies that the Dirac equation describes the planar In order to proceed to the self-adjoint extensions of (5),
motion (in the absence of the z coordinate) of the particle we decompose the Hilbert space H = L2 (R2 ) with re-
having only one projection of three-dimensional spin vec- spect to the angular momentum H = Hr ⊗ Hϕ , where
tor. To examine the physical implications of these equa- Hr = L2 (R+ , rdr) and Hϕ = L2 (S 1 , dϕ), with S 1 de-
∂2
tions, we consider their nonrelativistic limit. In this con- noting the unit sphere in R2 . The operator − ∂ϕ 2 is es-
text, writing ψ = (Φ, X)T e−iMt the Schrödinger-Pauli 2 1
sentially self-adjoint in L (S , dϕ) [19] and we obtain the
equation is HΦ = i∂t Φ, with operator H0 in each angular momentum sector. Now, us-
2 ing the unitary operator V : L2 (R+ , rdr) → L2 (R+ , dr),
qφ 1 − α

1 1 qsφ δ(r) given by (V g)(r) = r1/2 g(r), the operator H0 becomes
H= ∇α − + σz − , (1)
2M i αr 2αr 2M α r  2   
−1 1 d 2 1 1
∂ 2
1 ∂ 1 ∂2 h0 = V H0 V =− + j − , (8)
where ∇2α = ∂r 2 + r ∂r + α2 r 2 ∂ϕ2 is the Laplacian op- 2M dr2 4 r2
erator in the conical space, and σi = (σr , σϕ , σz ) are the which is essentially self-adjoint for |j| ≥ 1, while for
Pauli matrices in cylindrical coordinates. |j| < 1 it admits a one-parameter family of self-adjoint
For this system the total angular momentum operator, extensions [19], H0,λj , where λj is the self-adjoint exten-
Jˆ = −i∇ϕ + σz /2, commutes with the effective Hamilto- sion parameter. To characterize this family, we will use
nian. So, the solution to the Schr¨dinger-Pauli equation the KS [25] and the BG [26] approaches, both based in
can be written in the form boundary conditions.

f1 (r)ei(m−s/2)ϕ
 In the KS approach, the boundary condition is a match
−iEt
Φ(t, r, ϕ) = e (2) of the logarithmic derivatives of the zero-energy solutions
f2 (r)ei(m+s/2)ϕ
for Eq. (3) and the solutions for the problem H0 plus
with m = n + 1/2, n ∈ Z. At the same time, the radial self-adjoint extension. In the BG approach, the bound-
equation for f1 (r) becomes ary condition is a mathematical limit allowing divergent
solutions of the Hamiltonian (5) at isolated points, pro-
Hf1 (r) = Ef1 (r), (3) vided they remain square integrable.
Now, the goal is to find the bound states for the Hamil-
where tonian (4). Following [25], we temporarily forget the δ-
function potential and find the boundary conditions al-
H = H0 + Ushort , (4) lowed for H0 . But the self-adjoint extension provides
infinity possible boundary conditions, so that it cannot
1

d2 1 d j2
 give us the true physics of the problem. Nevertheless,
H0 = − + − , (5) once the physics at r = 0 is known [8, 33, 34], it is pos-
2M dr2 r dr r2
sible to determine any arbitrary parameter coming from
the self-adjoint extension, so that it is possible to obtain
qsφ δ(r) a complete description of the problem. Since we have a
Ushort = , (6) singular point, we must guarantee that the Hamiltonian
2M α r
is self-adjoint in the region of motion. Note that even if
with j = α1 (m − 2s − qφ + 1−α
2 ). The Hamiltonian in Eq. H0† = H0 , their domains could be different.
(4) governs the quantum dynamics of a spin-1/2 charged We must find the deficiency subspaces, N± , with di-
particle in the conical spacetime, with a magnetic field mensions n+ and n− , respectively, which are called de-
 3

ficiency indices of H0 [19]. A necessary and sufficient spectrum

condition for H0 being essentially self-adjoint is that   1/|j|
φ |j|
n+ = n− = 0. On the other hand, if n+ = n− ≥ 1, 2  Γ(1 + |j|)  1 + α|j| + 2
then H0 has an infinite number of self-adjoint extensions E =−  . (13)
M r02 Γ(1 − |j|) 1 − φ
− |j|
parametrized by unitary operators U : N+ → N− . α|j| 2
Next, we substitute the problem in Eq. (3) by H0 f̺ = Notice that there is no arbitrary parameters in the above
Ef̺ , with f̺ labeled by a parameter ̺ which is related to equation.
the behavior of the wave function in the limit r → r0 . But The above approach has the advantage of yielding the
we cannot impose any boundary condition (e.g. f = 0 self-adjoint extension parameter in terms of the physics
at r = 0) without discovering which boundary conditions of the problem, but it is not appropriate for dealing
are allowed to H0 . Then, from Eq. (5) we achieve the with scattering problems. On the other hand, the BG
modified Bessel equation (κ2 = −2M E, E < 0) method [26] is suitable to address both bound and scat-
tering scenarios, with the disadvantage of allowing ar-
d2 j2
  
1 d
2
+ − + κ2 f̺ (r) = 0. (9) bitrary self-adjoint extension parameters. Now, we ap-
dr r dr r2 ply the BG approach to solve bound and scattering
problems. By comparing the results of these two ap-
Now, in order to find the full domain of H0 in proaches for bound states, the self-adjoint extension pa-
L2 (R+ , rdr), we have to find its deficiency subspace. To rameter can be determined in terms of the physics of the
do this, we solve the eigenvalue equation problem. Here, all self-adjoint extensions of H0,λj are
parametrized by the boundary condition at the origin
H0† f̺± = ±if̺± , (10) [26] (g0 (r) = limr→0+ r|j| g(r))
 
where H0 is given by Eq. (5). The only square-integrable 1 ′ 1
g0 (r) = λj lim |j| g(r) − g0 (r ) |j| . (14)
functions that are solutions
√ of Eq. (10) are the modified r→0+ r r
Bessel functions K|j| (r ∓ε), with ε = 2iM . These func-
tions are square integrable only in the range |j| < 1, for The solutions for H0 f1,j = k 2 f1,j (k 2 = 2M E) for r 6= 0,
which H0 is not self-adjoint. The dimension of such de- can be written as (ρ = 2ikr)
ficiency subspace is (n+ , n− ) = (1, 1). Thus, D(H0 ) in ρ 1
f1,j (r) = Aj e− 2 ρ|j| 1 F1


L2 (R+ , rdr) is given by the set of functions [19] + |j|, 1 + 2|j|, ρ
2
√ √  ρ 1
+Bj e− 2 ρ−|j|


f̺ (r) = f1,j (r) + C K|j| (r −ε) + ei̺ K|j| (r ε) , (11) 1 F1 − |j|, 1 − 2|j|, ρ , (15)

2
where 1 F1 (a, b, z) represents the confluent hypergeomet-
where f1,j (r), with f1,j (r0 ) = f˙1,j (r0 ) = 0 (f˙ ≡ df /dr), is
ric function, and Aj , Bj are the coefficients of the regular
the regular wave function when we do not have U short (r). and irregular solutions, respectively. By implementing
The last term in Eq. (11) gives the correct behavior for Eq. (15) into the boundary condition (14), we derive the
the wave function when r = r0 . The parameter ̺(mod2π) following relation between the coefficients Aj and Bj :
represents a choice for the boundary condition. As we
λj k 2
 
shall see below, the physics of the problem determines −2|j| 2−2|j|
such a parameter without ambiguity. In fact, ̺ describes λj Aj = (2ik) Bj 1 + lim r . (16)
4(1 − |j|) r→0+
the coupling between U short (r) and the wave function.
Thus, it must be expressed in terms of α, the defect In the above equation, the coefficient of Bj diverges as
core radius r0 and the effective angular momentum j. limr→0+ r2−2|j| , if |j| > 1. Thus, Bj must be zero for
The next step is to find a fitting for ̺ compatible with |j| > 1, and the condition for the occurrence of a singular
U short (r). In this sense, we write Eq. (3) for E = 0, solution is |j| < 1. So, the presence of an irregular solu-
implying the zero-energy solution, Hf0 = 0. Now, we tion stems from the fact the operator is not self-adjoint
require the continuity for the logarithmic derivative for |j| < 1, and this irregular solution is associated with
a self-adjoint extension of the operator H0 [35, 36]. In
f˙0  f˙̺  other words, the self-adjoint extension essentially consists
=  , (12) in including irregular solutions in D(H0 ), which allows us
f0 r=r0 f̺ r=r0

to select an appropriate boundary condition for the prob-
where f̺ (r) comes from Eq. (11). However, since r0 ≈ 0, lem.
the right-hand side of the Hamiltonian (12) is calculated In the present system the energy of a bound state has
using the asymptotic representation for Eq. (11) in the to be negative, so that k is a pure imaginary, k = iκ.
limit r → 0. The left-hand side of Eq. (12) is achieved Thus, with the substitution k → iκ, we have (ρ′ = −2κr)
integrating the equation Hf0 = 0, from 0 to r0 , which B ρ′ |j| 1
(r) = Aj e− 2 ρ′ + |j|, 1 + 2|j|, ρ′


yields the parameter ̺ in terms of the physics of the f1,j 1 F1
2
problem, i.e., the correct behavior of the wave functions ρ′ −|j| 1
+Bj e− 2 ρ′ − |j|, 1 − 2|j|, ρ′ . (17)


for r → r0 . By solving Eq. (12) for E, we find the energy 1 F1
2
 4

For Eq. (17) representing a bound state, the solution Since the δ is a short range potential, it follows that the
B
f1,j (r) must vanish for r → ∞, i.e., it must be nor- behavior of f1,j for r → ∞ is given by
malizable. By using the asymptotic representation of r
1 F1 (a, b, z) for r → ∞, the normalizability condition 2  1 π λs 
f1,j (r) ∼ cos kr − |m|π − + δj j (k, φ) , (23)
yields the relation πkr 2 4
Γ(1 + |j|) λs
Bj = −16|j| Aj . (18) where δj j (k, φ) is a scattering phase shift. The phase
Γ(1 − |j|) shift is a measure of the argument difference to the
asymptotic behavior of the solution J(|m|, kr) of the
From Eq. (16), for |j| < 1 we have Bj = λj (−2κ)2|j| Aj ; radial- free equation that is regular at the origin. By us-
and by using Eq. (18), the bound state energy is ing the asymptotic behavior of J(|j|, kr) and Y (|j|, kr)
1/|j| for r → ∞ in Eq. (21), and comparing it with Eq.
λs

2 1 Γ(1 + |j|) (23), similarly as done in [37], we found that δj j (k, φ) =
E=− − . (19)
M λj Γ(1 − |j|)
∆m (φ) + θλsj , where ∆m (φ) = π2 (|m| − |m + φ|), and
λs
This coincides with Eq. (3.13) of Ref. [22] for α = 1, θλsj = arctan (µj j ). Therefore, the expression for the S
i.e., the spin-1/2 AB problem in Euclidean space with matrix is
the spinorial connection. By comparing Eq. (19) with  s 2|j|
λj k Γ(1 − |j|)ei|j|π + 4|j| Γ(1 + |j|)

Eq. (13), we find S=e 2i∆m (φ)
.
  λsj k 2|j| Γ(1 − |j|)e−i|j|π + 4|j| Γ(1 + |j|)
φ |j|
1 1  1 + α|j| + 2  (24)
= − 2|j| . (20) In accordance with the general theory of scattering, the
λj r0 φ
1 − α|j| − |j|
2 poles of the S matrix in the upper half of the complex
plane [38] [these poles occur in the denominator of (24)
We have thus attained a relation between the self-adjoint with the replacement k → iκ] determines the positions of
extension parameter and the physical parameters of the the bound states in the energy scale, Eq. (19). From this,
problem, j and r0 . It should be mentioned that some we have λsj = λj , with λj given by Eq. (20), and the self-
relations involving the self-adjoint extension parameter adjoint extension parameter for the scattering scenario
and the δ-function coupling constant were previously ob- being the same as that for the bound state problem. This
tained by using Green’s function in Ref. [23] and the is a very interesting result that has not been described
renormalization technique in Ref. [21], being both, how- in the literature yet, as far as we know. Thus, we also
ever, deprived from a clear physical interpretation. obtain the phase shift and the scattering matrix in terms
Once the bound energy problem has been examined, let of the physics of the problem. If λsj = 0, we achieve the
us now analyze the AB scattering scenario. In this case, corresponding result for the pure AB problem with the
the boundary condition is again given by Eq. (14) but Dirichlet boundary condition; in this case, we recover
now with the replacement λj → λsj , where λsj is the self- the expression for the scattering matrix found in Ref.
adjoint extension parameter for the scattering problem.
[39], S = e2i∆m (φ) . If we make λsj = ∞, we get S =
In the scattering analysis it is more convenient to use
the solution of the equation H0 f1,j = k 2 f1,j , in terms of e2i∆m (φ)+2iπ|j| .
Bessel functions In this article, we have presented a general regulariza-
tion method to address a system endowed with a singular
f1,j (r) = Cj J(|j|, kr) + Dj Y (|j|, kr), (21) Hamiltonian (due to localized fields sources or quantum
confinement). Using the KS approach, the bound states
with Cj and Dj being constants. Upon replacing f1,j (r) were determined in terms of the physics of the problem,
in the boundary condition (14), we obtain λsj Cj ξk |j| = in a very consistent way and without any arbitrary pa-
Dj ζk −|j| − λsj (ηk |j| + ζγk −|j| limr→0+ r2−2|j| ) , where rameter. In sequel, we employed the BG approach; by
 
|j| comparing the results of these approaches, we have deter-
ξ= 1
2|j| Γ(1+|j|)
,ζ = −2 Γ(|j|)
π , η = − cos(π|j|)Γ(−|j|)
π2|j|
and mined the value of the self-adjoint extension parameter
2
γ = k for the bound state problem, which coincides with the
4(1−|j|) . As
in the bound state calculation, when-
ever |j| < 1, we have Dj 6= 0; this means that there one for scattering problem. We thus obtain the S matrix
arises again the contribution of the irregular solution Y in terms of the physics of the problem, as well. A natu-
at the origin when the operator is not self-adjoint. Thus, ral extension of the problem studied here, amongst many
for |j| < 1, we obtain λsj Cj ξk |j| = Dj (ζk −|j| − λsj ηk |j| ), possible options, is the inclusion of the Coulomb poten-
and by substituting the values of ξ, ζ and η into above tial, which naturally appears in two-dimensional systems,
λs such as graphene [40] and anyons systems [41, 42]. Re-
expression we find Dj = −µj j Cj , where sults in this respect will be reported elsewhere.
The authors would like to thank M. G. E. da Luz,
λs λsj k 2|j| Γ(1 − |j|) sin(|j|π) D. Bazeia, E. R. Bezerra de Mello, C. R. de Oliveira
µj j = . (22)
λsj k 2|j| Γ(1 − |j|) cos(π|j|) + 4|j| Γ(1 + |j|) and M. M. Ferreira Jr. for their critical reading of the
 5

manuscript, and for helpful discussions and encourage- FAPEMA, CAPES-(PNPD) and M. Pereira acknowl-
ment. E. O. Silva acknowledges research grants from edges research grants from Fundação Araucária.

[1] K. M. Case, Phys. Rev. 80, 797 (1950). (World Scientific, Singapore, 1995).
[2] R. de L. Kronig and W. G. Penney, Proc. R. Soc. Lond. [22] D. K. Park and J. G. Oh, Phys. Rev. D 50, 7715 (1994).
A 130, 499 (1931). [23] D. K. Park, J. Math. Phys. 36, 5453 (1995).
[3] A. G. M. Schmidt, B. K. Cheng, and M. G. E. da Luz, [24] D. K. Park and S.-K. Yoo, Ann. Phys. (N.Y.) 263, 295
Phys. Rev. A 66, 062712 (2002). (1998).
[4] F. M. Zanetti, J. Khun, G. J. Delben, B. K. Cheng, and [25] B. S. Kay and U. M. Studer, Commun. Math. Phys. 139,
M. G. E. da Luz, J. Phys. A 39, 2493 (2006). 103 (1991).
[5] J. Kuhn, F. M. Zanetti, A. L. Azevedo, A. G. M. Schmidt, [26] W. Bulla and F. Gesztesy, J. Math. Phys. 26, 2520
B. K. Cheng, and M. G. E. d. Luz, J. Opt. B: Quantum (1985).
Semiclassical Opt. 7, S77 (2005). [27] D. D. Sokolov and A. A. Starobinski, Sov. Phys. Dokl.
[6] S. Albeverio, F. Gesztesy, R. Hoegh-Krohn, and 22, 312 (1977).
H. Holden, Solvable Models in Quantum Mechanics [28] K. Bakke, J. R. Nascimento, and C. Furtado, Phys. Rev.
(AMS Chelsea Publishing, Providence, RI, 2004), 2nd D 78, 064012 (2008).
ed. [29] E. R. Bezerra de Mello, J. High Energy Phys. 2004, 016
[7] Y. Aharonov and D. Bohm, Phys. Rev. 115, 485 (1959). (2004).
[8] P. d. S. Gerbert, Phys. Rev. D 40, 1346 (1989). [30] F. Gesztesy, S. Albeverio, R. Hoegh-Krohn, and
[9] C. R. Hagen, Phys. Rev. Lett. 64, 503 (1990). H. Holden, J. Reine Angew. Math. 1987, 87 (1987).
[10] C. R. Hagen, Int. J. Mod. Phys. A 6, 3119 (1991). [31] L. Dabrowski and P. Stovicek, J. Math. Phys. 39, 47
[11] C. R. Hagen, Phys. Rev. D 48, 5935 (1993). (1998).
[12] C. R. Hagen and D. K. Park, Ann. Phys. (N.Y.) 251, 45 [32] R. Adami and A. Teta, Lett. Math. Phys. 43, 43 (1998).
(1996). [33] C. Filgueiras, E. O. Silva, W. Oliveira, and F. Moraes,
[13] A. Vilenkin and E. P. S. Shellard, Cosmic Strings and Ann. Phys. (N.Y.) 325, 2529 (2010).
Other Topological Defects (Cambridge University Pres, [34] C. Filgueiras and F. Moraes, Ann. Phys. (N.Y.) 323,
Canbridge, 2000). 3150 (2008).
[14] M. Katanaev and I. Volovich, Ann. Phys. (N.Y.) 216, 1 [35] J. Audretsch, U. Jasper, and V. D. Skarzhinsky, J. Phys.
(1992). A 28, 2359 (1995).
[15] C. Sátiro and F. Moraes, Eur. Phys. J. E 20, 173 (2006). [36] F. A. B. Coutinho, Y. Nogami, and J. Fernando Perez,
[16] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Phys. Rev. A 46, 6052 (1992).
Y. Zhang, S. V. Dubonos, I. V. Grigorieva, and A. A. [37] C. R. de Oliveira and M. Pereira, J. Phys. A 43, 354011
Firsov, Science 306, 666 (2004). (2010).
[17] F. de Juan, A. Cortijo, M. A. H. Vozmediano, and [38] K. Bennaceur, J. Dobaczewski, and M. Ploszajczak,
A. Cano, Nature Physics 7, 810 (2011). Phys. Rev. C 60, 034308 (1999).
[18] Y. Zhang and A. Vishwanath, Phys. Rev. Lett. 105, [39] S. N. M. Ruijsenaars, Ann. Phys. (N.Y.) 146, 1 (1983).
206601 (2010). [40] A. H. Castro Neto, F. Guinea, N. M. R. Peres, K. S.
[19] M. Reed and B. Simon, Methods of Modern Mathemati- Novoselov, and A. K. Geim, Rev. Mod. Phys. 81, 109
cal Physics. II. Fourier Analysis, Self-Adjointness. (Aca- (2009).
demic Press, New York - London, 1975). [41] F. Wilczek, Phys. Rev. Lett. 49, 957 (1982).
[20] M. Alford, J. March-Russell, and F. Wilczek, Nucl. Phys. [42] Y.-H. Chen, F. Wilczek, E. Witten, and B. I. Halperin,
B 328, 140 (1989). Int. J. Mod. Phys. B3, 1001 (1989).
[21] R. Jackiw, Diverse topics in theoretical and mathemat-
ical physics, Advanced Series in Mathematical Physics