Chern-Simons Term and Charged Vortices in

Abelian and Nonabelian Gauge Theories

arXiv:hep-th/9505043v1 6 May 1995

Avinash Khare
Institute of Physics, Sachivalaya Marg,
Bhubaneswar 751 005, India.
January 16, 2014

Abstract
In this article we review some of the recent advances regarding
the charged vortex solutions in abelian and nonabelian gauge theories
with Chern-Simons (CS) term in two space dimensions. Since these
nontrivial results are essentially because of the CS term, hence, we
first discuss in some detail the various properties of the CS term in
two space dimensions. In particular, it is pointed out that this parity
(P) and time reversal (T) violating but gauge invariant term when
added to the Maxwell Lagrangian gives a massive gauge quanta and
yet the theory is still gauge invariant. Further, the vacuum of such a
theory shows the magneto-electric effect. Besides, we show that the
CS term can also be generated by spontaneous symmetry breaking as
well as by radiative corrections. A detailed discussion about Coleman-
Hill theorem is also given which aserts that the parity-odd piece of the
vacuum polarization tensor at zero momentum transfer is unaffected
by two and multi-loop effects. Topological quantization of the coeffi-
cient of the CS term in nonabelian gauge theories is also elaborated
in some detail.
One of the dramatic effect of the CS term is that the vortices of
the abelian (as well as nonabelian) Higgs model now acquire finite
quantized charge and angular momentum. The various properties of
these vortices are discussed at length with special emphasis on some
of the recent developments including the discovery of the self-dual
charged vortex solutions.

1
1 INTRODUCTION
In 1957, Abrikosov [1] wrote down the vortex solutions in Ginzburg-Landau
(GL) theory which is the mean-field theory of superconductivity. Subse-
quently, these vortices were experimently observed in type-II superconduc-
tors. In 1973, Nielsen and Olesen [2] rediscovered these solutions in the
context of the abelian Higgs model which is essentially a relativistic gener-
alization of the GL theory. These people were looking for string like objects
in field theory. It turns out that these vortex solutions have finite energy
per unit length (i.e. finite energy in 2+1 dimensions as the vortex dynamics
is essentially confined to the x-y plane), quantized flux but are electrically
neutral and have zero angular momentum.
In 1975, Julia and Zee [3] showed that the SO(3) Georgi-Glashow model
which admits t Hooft-Polyakov monopole solution also admits its charged
generalization i.e. the dyon solution with finite energy and finite, nonzero
charge. It was then natural to enquire if the abelian Higgs model which
admits neutral vortex solutions with finite energy (in 2+1 dimensions), also
admits charged vortex solutions with finite charge and energy in 2+1 dimen-
sions. Julia and Zee showed in the appendix of the same paper [3] that the
answer to the question is no. More than ten years later Samir Paul(then
my Ph.D. student) and myself showed [4] that the Julia-Zee result can be
bypassed by adding the CS term [5, 6] to the abelian Higgs model in 2+1
dimensions. In particular, we showed that the abelian Higgs model with CS
term in 2+1 dimensions admits charged vortex (soliton to be more precise)
solutions of finite energy, nonzero finite charge and flux. As an extra bonus,
one found that these vortices have nonzero angular momentum which is in
general fractional. This strongly suggested that these objects could infact be
charged anyons [7] i.e. the objects which are neither bosons or fermions but
which obey statistics which is interpolating between the two. Subsequently,
Frohlich and Marchetti [8] have shown using axiomatic field theory that these
objects are indeed charged anyons.
There is one question that has remained unanswered though i.e. can
one overcome Julia-Zee objection [3] in 3+1 dimension itself? Recently we
have answered the question in the affermative. In particular we [9] have
been able to construct self-dual topological as well as nontopological charged
vortex solutions of finite energy per unit length in a generalized abelian Higgs
model with a dielectric function and a neutral scalar field. The interesting

2
point is that in this case the Bogomolnyi bound on energy per unit length is
obtained as a linear combination of the magnetic flux and the electric charge
per unit length.
By now our work on the CS vortices has been extended in several direc-
tions. Most of the developments till 1988 have been well summarized in my
earlier review article on this subject [10]. Howevere, several new advances
have taken place since that time and one purpose of this article is to discuss
some of those developments. Since the key role in this game is played by the
CS term, it is only proper that at first I discuss the various properties of this
term.
The plan of the paper is as follows: In Sec. II, the role of the CS term is
discussed in the context of both abelian and nonabelian gauge theories. The
key point to note is that whereas the Chern-Pontryagin term is topological
but has no dynamics (being a total divergence), the CS term is topological
and also contributes to equations of motion and hence has nontrivial dynam-
ics. In particular, it is pointed out that in 2+1 dimensions, because of this
term, one has at the same time a massive gauge field and a gauge invariant
action. Quantum electrodynamics with CS term has some interesting and
unusual properties which are discussed here. For example, the vacuum po-
larization tensor has an extra piece which is odd in both P and T . Various
properties of this P and T-odd piece including Coleman-Hill (CH) theorem
[14] and magneto-electric effect [15] are discussed at length. In particular,
it is emphasised that contrary to the claim of CH, not only fermions but
any P and T violating interaction including even scalar [16] or vector [17]
particles can give nonzero contribution to the P and T-odd part of vacuum
polarization tensor at zero momentum transfer. We also point out that the
theorem is also valid in case the gauge symmetry is spontaneously broken
and the theorem is stated in terms of effective action rather than vacuum
polarization tensor [16]. In the nonabelian case, it turns out that the theory
is not well defined unless the coefficient of the CS term is quantized [6]. Var-
ious issues in this regard are discussed at length. In particular, it is pointed
out that the tree level quantization continues to remain valid at one [18] and
higher loops [19] in the case of pure gauge theories. Further, it is also valid
at one loop in the presence of matter fields. Finally, it is argued that the
quantization of the coefficient of the CS term is also valid at one loop for an
unbroken nonabelian gauge group in a theory where larger nonabelian gauge
symmetry is spontaneously broken to that of the smaller gauge group [20].

3
 We also point out that atleast in the nonabelian case, adding the CS term
to the action is not really a luxary in a sence one is forced to add it since
even if one does not add it at the tree level, it is automatically induced by
radiative corrections [21]. Finally, it is pointed out that a la gauge field mass
term the abelian CS term can also be generated by spontaneous symmetry
breaking [22]. This is possible because in 2+1 dimensions, even for scalar
particles one can introduce a Pauli type nonminimal interaction term. Some
other implications of this term are also discussed in Sec. II.
In Sec. III we discuss one of the most dramatic consequence of the CS
term in 2+1 dimensions i.e. the existance of charged evortex solutions [4].
Only salient features of this solution are discussed here since the details are
already contained in my previous review article [10]. In Sec. IV, we discuss
some of the recent developments in this field. In particular, we discuss the
self-dual charged vortex solutions in both relativistic [11] and nonrelativistic
[12] theories as well as semi-local charged vortex solutions [13].

2 Some Important Properties of the Chern-

Simons Term
It is well known that there is a gauge singlet (popularly known as axial)
anomaly in any even dimension 2n so that the divergence of the gauge singlet
axial current is proportional to the corresponding Chern-Pontryagin density
P2n in that dimension which in turn can always be written as a total diver-
gence i.e.
P2n = ; = 0, 1, 2, ......n 1 (1)
The object , for a particular value of (say 0 ) naturally lives in (n 1)
dimensions (excluding that particular dimension 0 ) and is known as the
Chern-Simons density. For example the gauge singlet anomaly in 3+1 di-
mensional quantum electrodynamics is given by

e2
j5 = F F (2)
2
e2
= ( A F ) (3)

4
so that the abelian CS term in 2+1 dimensions is given by
e2
3 = 3 A F (4)

which clearly lives in 2+1 dimensions. In the nonabelian case, the CS term
has an extra piece i.e.
2
3 T r(F A A A A ) (5)
3
where A and F are matrices

A = gT a Aa ; F = gT a F
a
= A A + [A , A ] (6)

Here we have used the representation matrices of the group (for SU(2): T a =
a /2i)
[T a , T b ] = f abc T c . (7)
Let us first discuss the properties of the abelian CS term as given by eq.
(4).
P and T Violation: It is easily seen that the abelian (as well as the
nonabelian) CS term is not invariant under P and T seperately even though
it is invariant under PT as well as charge conjugation C. It is worth noting
here that in 2+1 dimensions, even the fermion mass term m is odd under
P as well as T and infact this is the underlying reason as to why CS term
can be generated in perturbation theory by integrating over fermions in a
massive fermionic theory [21].
CS Term as Gauge Field Mass Term: Let us consider electrodynam-
ics in the presence of the CS term
1
L = F F + F A (8)
4 4
where F = A A . Note that the corresponding equation of motion

F + F = 0 (9)
2
as well as the action are invariant under U(1) gauge transformation
1
A A (10)
2

5
The field eq. (9) can also be written as
1
(g + ) F = 0 (11)

where F is the dual field strength which is a vector in three dimensions

1
F = F ; F = F (12)
2
On operating by (g /) to eq. (11) we have

( + 2 ) F = 0 (13)

which clearly shows that the gauge field excitations are massive. This re-
markable property of having a gauge invariant mass term for the gauge field
in the action itself is very special to 2+1 dimensions. In all other dimensions
one has to take recourse to the Higgs mechanism (or one could have dynam-
ical symmetry breaking as in the 1+1 dimensional Schwinger model). It is
woth noting here though that unlike in other dimensions, in this case, both
massless and the CS-mass photon has only one degree of freedom. Further,
whereas the massless photon in 2+1 dimensions has spin zero, the CS-mass
photon has spin 1 (-1) if > (<)0. Note that because of the CS term, one
has necessarily a P and T violating theory. On the otherhand, the normal
massive photon has two degrees of freedom and both spins 1 are present as
they should be in a parity conserving theory.
Coleman-Hill Theorem: It turns out that because of the P and T
violating but gauge invariant CS term, the most general form for the vacuum
polarization tensor (consistent with Lorentz and gauge invariance) is more
general than in other dimensions i.e.

(k) = (k 2 g k k )1 (k 2 ) i k 2 (k 2 ) (14)

Notice that the second term on the r.h.s. is P and T odd. It is clear that
any P and T violating interaction will contribute to 2 (k 2 ). For example
the fermion mass term which in 2+1 dimensions break both P and T, does
contribute to 2 (k 2 ) at one loop level. Remarkably enough, it was discovered
that at two loops, though, there is no contribution to 2 (0) [23]. Inspired
by this result, Coleman and Hill [14] have infact proved under very general

6
conditions that 2 (0) receives no contributions from two and higher loops
in any gauge and Lorentz invariant theory including particles of spin one or
less. In particular, they have emphasized that their result is valid even for
nonrenormalizable interactions in the presence of gauge and Lorentz invari-
ant regularizations. These authors also claimed that at one loop the only
contribution to 2 (0) can come from fermion loop. This is however not true.
In particular, there is no reason why P and T violating interactions involving
spin 0 or 1 particles should not contribute to 2 (0) at one loop. Indeed, Ha-
gen et al. [17] as well as we [16] have shown that nonrenormalizable spin one
and spin zero interactions respectively do contribute to 2 (0) at one loop.
It might be added here that apart from these situations which were over-
looked by Coleman-Hill, there are other situations where the initial assump-
tions of the theorem are not satisfied and where 2 (0) does get further radia-
tive corrections. One such situation is if there are massless particles present
in which case infrared divergences spoil the proof of the theorem [24]. An-
other case is if Lorentz or gauge invariance is not satisfied, a situation found
in the nonabelian case. A third case is that of spontaneously broken scalar
electrodynamics [25] where the term quadratic in the gauge field explicitly
violates one of the assumption of Coleman and Hill. However, even in this
case we have recently shown [16] that if the theorem is formulated in terms of
effective action rather than vacuum polarization tensor then the coefficient
of the CS term in the effective action does not receive a radiative correction
at one loop.
Magnetoelectric Effect: There are many crystals in nature like cromium
oxide which show this effect i.e. they get magnetized in an electric field and
electrically polarized in a magnetic field [26]. It is well known that this ef-
fect depends upon having a T-assymmetric medium. In this case the usual
relation between D ~ and E ~ as well as between H~ and B ~ is modified to
(e) (em)
Di = ij Ej + ij Bj (15)
(m) (me)
Hi = ij Bj + ij Ej (16)
Since CS term violates P and T, it is natural to ask if the vacuum of the 2+1
dimensional QED with CS term also shows the magnetoelectric effect or not.
We have shown [15] that indeed the vacuum in such a theory does show this
(em) (me)
effect and both i and i are proportional to ki 2 (k 2 ).

7
 CS Term by Spontaneous Symmetry Breaking: We have seen
above that the CS term provides mass to the gauge field. Now usually the
gauge field mass is generated by spontaneous symmetry breaking hence it
is worth enquiring if the CS term can also be generated by this mechanism.
The answer to the question turns out to be yes. This is because as has been
shown by us [22], unlike other dimensions, 2+1 dimensions offer more general
possibility of a covariant derivative. For example, it is possible to introduce
a Pauli type nonminimal coupling for even scalar particles. In particular,
notice that
D = ( + ieA + ig A ) (17)
also behaves like a covariant derivative since the nonminimal term by itself is
gauge covariant. As a result one now finds that the matter field kinetic energy
term 21 (D ) (D ) has a piece eg||2 ( A )A . Thus if acquires a
nonzero vacuum expectation value then the (abelian) CS term is generated.
Clearly a similar mechanism should also work for the nonabelian case, but
technically it is a tougher problem since one also has to generate the nonlinear
triple gluon coupling term. So far as I know, till today it is an open problem.
The resulting nonminimal theory has been shown to have some very in-
teresting properties in case the magnetic coupling constant g (called critical
magnetic moment) acquires a special value [27, 29]. In particular, this theory
gives rise to an effective action which is renormalizable at one loop [28] and
has similar properties to the one describing ideal anyons upto an additive
contact term. It is worth adding here that this nonminimal magnetic cou-
pling can be induced by radiative corrections even if it is not present at the
tree level [30].
Yet another remarkable property of the CS term is that in this case the
Lorentz invariance of the action automatically follows from gauge invariance.
In particular, whereas for the Maxwell case the most general gauge invariant
Lagrangian is
L=E ~ 2 + aB~2 (18)
it is only the demand of Lorentz invariance which fixes a to be -1. On the
other hand, in the CS case the demand of gauge invariance automatically
fixes the form of the CS term.
Let us now discuss some properties which are unique to the nonabelian
CS term.

8
 Quantization of the CS Mass: In the nonabelian gauge theory with
CS term in 2+1 dimensions one finds that the gauge field is again massive
and that the theory is well defined only if this mass is infact quantized. The
gauge field Lagrangian is
1 2
L= 2
T r(F F ) 2 T r(F A A A A ) (19)
2g 2g 3

where A and F are matrices as defined by eq. (6) The field equation which
follows from here

D F + F = 0 (20)
2
where
D = + [A , ] (21)
is gauge covariant. As in the abelian case it immediately follows
R 3
that the
gauge field has mass . Now notice that the action ICS = d xLCS even
though invariant under small gauge transformations, is not invariant un-
der homotopically-nontrivial gauge transformations [6]. In particular, if the
gauge group G is such that
3 (G) = Z (22)
where Z is the group of integers (note in particular that eq. (22) is true for
any gauge group of which SU(2) is a subgroup), then under these so called
large gauge transformations the action transforms as

8 2
Ics Ics + m (23)
g2
where m is an integer. Now in the path integral formulation, the action
itself may or may not be gauge invariant but what is required is that the
exponential of the action should atleast be gauge invariant. We thus conclude
that the nonabelian gauge theory with CS term does not make sence in 2+1
dimensions unless the CS mass is quantized in units of g 2 /4 i.e.

8 2 g2
= 2n or = n , n = 0, 1, 2, ... (24)
g2 4
Please note that in 2+1 dimensions the gauge coupling g is not dimensionless
1
but rather has dimension of (mass) 2 .

9
 Parity Anomaly: Someone might wonder as to why is one considering
models with CS term in the first place since afterall this term violates both
P and T. The answer to that is (atleast in nonabelian gauge theories) even if
one does not add CS term to the action at the tree level, it is still generated
by the radiative correctionsthe effect due to the so called parity anomaly
[21]. In particular, even though the action
1
Z
I[A , ] = d3 x[ T rF F + i D ] (25)
2g 2
is invariant under both gauge transformations and P and T, the effective
action Ief f [A] obtained by integrating out the fermionic degrees of freedom
must violate one of the two symmetries in the case of odd number of mass-
less fermions. In other words, there is no regularization which can simul-
taniously maintain the invariance of Ief f under parity as well as under large
gauge transformations. In particular, under large gauge transformations with
winding number n, Ief f transforms as

Ief f [A] Ief f [A] n n = 0, 1, 2, .... (26)

Since gauge invariance is usually regarded as more important than parity,

one can maintain gauge invariance at the cost of parity by adding the P
and T-odd CS term in the action (or by regulating the theory in such a
way that the CS term is automatically produced say by using Pauli-Villers
regularization). In this way one finds that because of the parity anomaly the
CS term is induced by radiative corrections even if it is absent at the tree
level. This is very similar to the way the Chern-Pontryagin term is induced
in even dimensions due to the gauge singlet (chiral) anomaly. Note however
that whereas the anomaly in even dimensions is in the divergence of the
current, in 2+1 case there is no anomaly in the divergence of the current.
The anomaly is in the current itself in that there is a piece with wrong parity.
Also notice that the parity anomaly is only in the nonabelian gauge theories
and not in the abelian U(1) case since in this case there are no large gauge
transformations in the first place!
Finally, since the CS action depends only on tensor and not on the
metric g hence the gauge field action with only nonabelian (or even abelian)
CS term is an example of topological field theory [31].

10
3 Charged Vortex Solutions
Let us consider the abelian Higgs model with CS term as given by the La-
grangian
1 1 C2 2
L = F F + ( ieA ) ( +ieA )C4 (| 2 | ) + A F
4 2 2C4 4
(27)
Following the neutral vortex case [2], let us consider the following n-vortex
ansatz

~ , t) = g(r) n
A(~ e0 C0 , t) = C0 ein f (r), A0 (~
, (~ , t) = C0 h(r) (28)
r
where s
r C2
= , C0 = (29)
eC0 2C4
We have rescaled the lengths and the fields so that one can work in terms
of the dimensionless variables. It turns out that the dynamics essentially
depends on two dimensionless parameters and defined by
q
= (8C4 /e2 ); = /eC0 (30)

The field equations which follow from here are [4]

1
g (r) g (r) gf 2 = rh (r) (31)
r
1
h (r) + h (r) hf 2 = g (r) (32)
r r
2 2
1 g f
f (r) + f (r) 2 + f (1 f 2 ) = h2 f (33)
r r 2
while the field energy can be shown to be [4]
1 dg df dh g 2 f 2 2
Z  
En = C02 rdr 2 ( )2 +( )2 +( )2 +h2 f 2 + 2 + (1f 2 )2 (34)
0 r dr dr dr r 4
Several remarks are in order at this stage.
(i) As expected, in the limit Ao = 0 (i.e. h = 0) and = 0 (i.e. = 0)
the field equations reduce to those of the neutral vortex case [2]. From the

11
Gauss law eq. (32) it also follows that if is nonzero then Ao must also be
nonzero.
(ii) The boundary conditions for finite energy solutions are

lim : f (r) = 1, h(r) = 0, g(r) = 0

r
(35)

lim : f (r) = 0, h(r) = , g(r) = n (36)

r0

with being an arbitrary number.

(iii) A la neutral vortex case, in this case also one has flux quantization
since the boundary conditions are again such that 1 (U(1)) = Z
Z
2 2 Z 1 dg 2
Bd x = rdr = n (37)
e 0 r dr e
From the Gauss law eq. (32) it then follows that these vortices also have
nonzero and finite quantized charge [4]
2
Z
Q e2 hf 2 d2 x = = n (38)
e
thereby bypassing the Julia-Zee objection. As far as I am aware off, this
is probably the first time that the quantization of the Noether charge has
followed from purely topological considerations. In a way the relation (38)
can be looked upon as the 2+1 analog of the Witten effect [32]. One can
also compute the magnetic moment of these vortices and show that whereas
the neutral vortices have it equal to the flux (= 2n/e), the charged ones
have an extra piece
2n
Z Z
Kz = (~r ~j)z d2 x = + h(r)d2 r (39)
e e 0

(iv) As an extra bonus, one also finds that unlike the neutral vortices,
the charged vortices have nonzero angular momentum which is ingeneral
fractional and quantized in units of Q/2e i.e.
nQ
Z
J d2 xij xi Toj = (40)
2e
This strongly suggests that the charged vortices could infact be charged
anyons [7]. By explicit construction of a quantum one vortex operator,

12
Frohlich and Marchetti [8] have rigorously shown that this is indeed the
case. Thus charged vortices provide us with a relativistic field theory model
of extended charged anyons.
So far, no analytic solution has been obtained to the field eqs. (31) to
(33). However, it is easily seen that for large r, the asymptotic values of the
gauge and Higgs field are reached exponentially fast

g(r) = remv r (41)

h(r) = emv r (42)
r
f (r) = 1 + er (43)
where , are dimensional constants while
s
2
mv = + e2 C02 . (44)
4 2
Naively, another solution with
s
2
mv = + e2 C02 + , (45)
4 2
is also possible but as has been shown in [33], such a solution does not exist
for all r. It is worth noting here that because of the P and T violating CS
term, the gauge field, after the Higgs mechanism, propagates two modes each
with one degree of freedom and with J = 1(-1) if > (<)0 [18, 34]. It is
easily seen that the field equations are invariant under r r so that the
behaviour of the fields around r=0 is given by

g(r) = n + r 2 + 0(r 4 ) (46)

r2
h(r) = + + 0(r 4 ) (47)
2
f (r) = 2 r |n| + 0(r |n|+2) (48)
The qualitative behaviour of the charged vortex is as follows: the magnetic
field B decreases monotonically from its nonzero value at the core of the
vortex (r=0) to zero at r = with penetration length 1/mv while the

13
scalar field increases from zero at origin to its vacuum value at infinity with
coherence length 1/ms . Finally, the electric field E vanishes at both r=0
and r = reaching the maximum in between at some finite r. It is worth
pointing out that as in the Hall effect, for the charged vortex solutions too
~ (= E ) is at right angles to ~j (=j ) and both in turn are at right angles to
E
B. It is also worth pointing out here that according to the presently accepted
explanation, the quasi-particles responsible for the fractionally quantized Hall
effect are the charged vortices.

4 Recent Advances
After our discovery of the charged vortex solutions in 1986 [4], in last ten
years this work has been extended in several directions. I shall only briefly
discuss the developments till 1988 which are already contained in my previous
review article on this subject [10] while will discuss in detail some of the latter
developments.
Charged Vortex-Vortex Interaction: Perhaps the most interesting
question is if we can directly observe the charged vortices in some condensed
matter system. In this context recall that the neutral vortices with one unit
of vorticity have been seen in type-II superconductors. However, none has
been seen in type-I superconductors. This can be understood from the fact
that in the neutral case, whereas the vortex-vortex interaction is attractive
in type-I region ( < 1), it is repulsive in type-II region ( > 1) [35]. It is
thus of great interest to study charged vortex-vortex interaction and see as
to when is it repulsive. This has been done by us [36] using perturbation
theory in CS mass as well as by a variational calculation. For example, when
CS mass is small, one can expand the charged vortex fields in terms of the
neutral vortex fields plus corrections in powers of CS mass . In particular,
it has been shown that to O( 2), the charged n-vortex fields are given by
[36, 33]
r22
g(, ) = g0 () + g0 () + 0( 4 ) (49)
8

h(, ) = g0 () + 0( 3 ) (50)
2
f (, ) = f0 () + 0( 4 ) (51)

14
where go and fo are the solutions to the corresponding neutral vortex solu-
tions in the absence of the CS term. On substituting the solution as given by
eqs. (49) to (51) in the expression for the field energy as given by eq. (34)
one can show that [36]

n2 2
Encha (, ) = Enneu () + + 0( 4 ) (52)
4
It is worth noting that the O( 2 ) correction is positive, proportional to n2
and independent of . From this equation it immediately follows that

(n2 n) 2
Encha (, ) n E1cha (, ) = Enneu () nE1neu () + + 0( 4 ) (53)
4
so that the charged vortex-vortex interaction is more repulsive than the cor-
responding neutral case with the extra repulsion coming from the electric
field of the charged vortex. We have also performed a variational calculation
(which is more reliable than the perturbative calculation for larger values of
) and find that even in this case the same picture continues to hold good.
For example, for = 0.5 the charged vortex-vortex interaction is repulsive
even for > 0.45 (note that in the neutral case the vortex-vortex interaction
is repulsive only when > 1).
A word of caution is in order here. Our analysis is only valid in the case
of superimposed vortices. The problem of charged vortex-vortex interaction
when the vortices are seperated by distance d, is still an open unsolved
problem.
Pure CS vortices: Can one obtain charged vortex solutions in abelian
Higgs model with pure CS term (i.e. no Maxwell kinetic energy term)? This
question is particularly sensible in the condensed matter context since in the
long wave length limit the CS term dominates over the Maxwell term. In this
context note that the Higgs mechanism is operative even in the absence of
the Maxwell term and even in this case one obtains both massive gauge and
scalar fields [37]. The question of the charged vortices in the abelian Higgs
model with pure CS term was addressed by us [38] and we showed that the
charged vortex solutions are indeed possible in this case. Their properties
are almost same as those of the Maxwell-CS charged vortices except that the
magnetic field is now zero at the core of the vortex and is concentrated in
a ring surrounding the vortex core [38]. It is worth noting here that in the

15
absence of the Maxwell term, the gauge field eqs. (31) and (32) are already
of first order
gf 2 = rh (r) (54)

hf 2 = g (r) (55)
r
while eq. (33) remains unaltered and is still a coupled second order equation.
The obvious interesting question is if one can also write it as a coupled
first order equation so that a la Bogomolnyi [39] one could obtain self-dual
charged vortex solutions. This question was raised by us [38] but we were
unable to obtain the first order equation. It was left to two other groups
[11] to make this important breakthrough. They showed that in addition to
dropping the Maxwell term one also has to replace the usual 4 potential
with the following 6 -type potential

e4
V (| |) = | 2 | (| 2 | v 2 )2 (56)
82
so as to obtain the self-dual equations. In that case, the self-dual equations
turn out to be
1
f = fg (57)
r
1 dg 1
B = f 2 (1 f 2 ) (58)
r dr 2
while
1
h = (1 f 2 ) (59)
2
One can infact decouple the eqs. (57) and (58) and obtain the following
uncoupled second order equation in f

1 f 2 1 3
f (r) + f (r) + f (1 f 2 ) = 0 (60)
r f 2
These self-dual equations are quite similar to those of the corresponding
neutral case (at = 1) which are given by
1
f = gf (61)
r

16
 1 dg 1
B = (1 f 2 ) (62)
r dr 2
and hence the uncoupled second order equation in that case is

1 f 2 1
f (r) + f (r) + f (1 f 2 ) = 0 (63)
r f 2
Further, whereas the Lagrangian for the self-dual neutral vortex case is the
bosonic part of a N = 1 supersymmetric theory [40], the Lagrangian for the
charged self-dual vortex case is the bosonic part of a N = 2 supersymmetric
theory [41].
Before we discuss the solutions to the self-dual eqs. (57) and (58) it may
be worthwhile to point out that the usual 4 potential and the 6 potential
as given by eq. (56) represent very different physical situations [42]. Whereas
the 4 potential in eq. (27) corresponds to the case of second order phase
transition with T < TcII , the 6 potential given above corresponds to the
case of first order transition with T = TcI . One way to understand as to why
6 -type potential is required for the CS vortices while 4 potential is required
for the neutral vortex case is that whereas in four space-time dimensions the
coefficient of the 4 -term is dimensionless, it is the coefficient of the 6 -term
which is dimensionless in 2+1 dimensions.
It turns out that the self-dual eqs. (57) and (58) admit both topological
and nontopological self-dual charged vortex solutions. Let us first discuss the
topological solutions.
Topological Self-dual Solutions: The topological solutions satisfy the
boundary conditions as given by eqs. (35) and (36) with = 1/2 (for n >
(<)0 respectively). The flux, the charge and the angular momentum of these
vortices are as given by eqs. (37) to (40) respectively and the energy is v 2 |n|.
Infinite number of sum rules have been derived for these vortices [43] (as well
as the neutral self-dual vortices [44]) and using these we have shown that
the magnetic moment of the self-dual pure CS vortex is 2n(n + 1)2 /e3 v 2 .
It is worth emphasizing that the self-dual solutions have been obtained not
only with the cylindrical ansatz but also with an arbitrary ansatz. Further,
following the work of Taubes for the neutral self-dual vortices [45], Wang
[46] has given rigorous argument for the existance of self-dual charged vortex
solutions even when the vortices are not superimposed on each other but lie
at arbitrary positions in the plane.

17
 Nontopological Self-dual Solutions: Since the potential as given by
eq. (56) has degenerate minima at = 0 as well as at | |= v, one finds
that apart from the topological, one also has nontopological self-dual charged
vortex solutions provided one chooses the following boundary conditions [43,
47]
lim : f (r) = 0, g(r) = , > 0 (64)
r

lim : f (r) = 0, g(r) = n, if n 6= 0 (65)

r0

lim : f (r) = , g(r) = 0 if n = 0 (66)

r0

where +() is for n > (<)0. The flux, the charge and the angular momen-
tum of these vortices can be shown to be (n > 0)
2
= (n + ); Q = ; J = 2 (2 n2 ); E = v 2 (n + ) (67)
e e
It is worth pointing out that the finiteness of energy requires that > 1 but
otherwise is arbitrary. However, again in this case infinite number of sum
rules have been derived by us [44] using which we have been able to show that
must infact satisfy the lower bound (n + 2) [48]. Further, from these
sum rules it also follows that the magnetic moment of these nontopological
vortices is 2( + n)( n 1)2 /e3 v 2 . As far as I am aware off, this is the
first instance when both topological and nontopological self-dual solutions
simultaneously exist in a given model. So far as the decay to charged scalar
mesons is concerned, these nontopological solutions are at the edge of their
stability [49]. Finally, Spruck and Yang have rigorously shown the existance
of self-dual nontopological CS vortices even when they are not superimposed
on each other but lie at arbitrary positions in the plane [50].
Various Other Self-dual Solutions: Once the self-dual CS vortices
were discovered in 1990, there has been a flurry of activity and several people
have obtained various other self-dual solutions. While it is clearly impossible
to mention all these developments, we shall try to note atleast some of them.
For example, since 4 and 6 -type models correspond to very different phys-
ical situations, hence it is of interest to enquire if self-dual charged vortex
solutions can also be obtained in the original 4 model [4] itself. This has
been done and Lee et al. [51] have shown that such solutions can also be
constructed in the 4 model [4] provided one adds a neutral scalar field in the
model. Further, self-dual solutions have also been constructed with unusual

18
properties [52] by essentially multiplying the Maxwell and/or the CS term
by a dielectric function (which in almost all cases is assumed to be a function
of the scalar field alone) and also by including the nonminimal interaction
term. For example, a class of self-dual solutions have been obtained which are
degenerate in energy but have different flux which is not quantized. Further,
nonabelian self-dual CS vortices have also been obtained [53].
Nonrelativistic Self-dual Vortices: In an interesting paper, Jackiw
and Pi [12] started with the abelian Higgs model with pure CS term and the
6 potential as given by eq. (56) (which has both topological and nontopo-
logical self-dual charged vortex solutions) and considered its nonrelativistic
limit. In particular, they showed that to leading order in the velocity of light
c, this model reduces to
1 1
LN R = F A + i (t + iA0 ) (Di ) (Di ) + ( )2
4 2m 2mc | |
(68)
where represents the particle part of the mode expansion of the scalar field
(the anti-particle part having been put equal to zero) and m is the mass
of the field . This nonrelativistic model can be looked upon either as a
classical field theory or as a second quantized N-body problem with 2-body
attractive -function interaction. These authors were able to obtain self-dual
nontopological charged vortex solutions with zero energy in the above model
and showed that these vortex solutions are precisely the nonrelativistic limit
of the corresponding nontopological charged vortex solutions. In fact, the
flux, the charge and the angular momentum of these objects turn out to be
the same as given by eq. (67) but where = (n + 2) i.e. the lower bound on
[44] is saturated in the nonrelativistic case. By now, this work has been
extended in several directions. Mention may be made of the time dependent
solutions by Ezawa et al. [54] and the nonrelativistic Maxwell-CS vortices
[55]. Further, nonrelativistic limit of the nonabelian self-dual CS vortices
has also been considered and interesting connections with integrable models
have been discovered [56].
Interaction Between Self-Dual CS Vortices: Following the work
of Manton [57] for the case of monopole and neutral vortices, recently Kim
and Min [58] have considered the slow motion of two well seperated CS
vortices and have shown that the effective Lagrangian (with finite degrees
of freedom) has a statistical interaction term and that this term reflects the

19
anyonic nature of the CS vortices. This analysis has recently been extended
to the nonrelativistic case [59].
Semi-Local Self-Dual CS Vortices: Recently, semi-local neutral vor-
tex solutions have been obtained in an abelian Higgs model with SU(N)global
U(1)local symmetry. The key point of the argument is that even though topo-
logically trivial, these solutions are stable under small perturbations due to
the gradient energy term [60]. These semi-local solutions gave some ini-
tial hope of finding energatically stable solutions in the Weinberg-Salam
model. However, subsequent analysis has shown that these semi-local vor-
tex solutions are always unstable for the realistic values of the Weinberg
angle. Inspired by this work, we have constructed [13] semi-local self-dual
CS vortex solutions in an abelian Higgs model with pure CS term and with
SU(N)global U(1)local symmetry. Most of the results for the CS vortices can
be easily extended to this case.
Charged Vortices of Finite Energy Per Unit length in 3+1 Di-
mensions: So far we have shown that the Julia-Zee objection [3] can be
bypassed in 2+1 dimensions by adding CS term to the action. However, the
question remains if one can also overcome their objection in 3+1 dimensions
itself (remember that their original argument is infact in 3+1 dimensions)?
Recently, we [9] have been able to do that. In particular, we showed that
if one generalizes the abelian Higgs model by adding a dielectric function
and a neutral scalar field then one can obtain self-dual topological as well as
nontopological charged vortex solutions with finite energy per unit length.
In particular, we considered the following generalized abelian Higgs model
1 1 1
L = G(| |)F F + | ( ieA ) |2 + G(| |) N N
4 2 2
2
e e2
(| |2 v 2 )2 N 2 | |2 (69)
8G(| |) 2

where N is the neutral scalar field. We showed that if the dielectric function
G(| |) is chosen to have the form

G(| |) = (70)
| |2

then the self-dual equations of the model can essentially be mapped to those
of the purs CS vortices thereby explaining as to why one has both topological

20
as well as nontopological charged vortex solutions. However, unlike in that
case, the Bogomolnyi bound on energy is expressed not only in terms of the
flux but also the vortex charge per unit length.

21
References
[1] A.A. Abrikosov, Sovt. Phys. JEPT 5 (1957) 1174.

[2] H.B. Nielsen and P. Olesen, Nucl. Phys. B 61 (1973) 45.

[3] B. Julia and A. Zee, Phys. Rev. D 11 (1975) 2227.

[4] S.K. Paul and A. Khare, Phys. Lett. B 174 (1986) 420; B 182 (1986)
E 414.

[5] W. Siegel, Nucl. Phys. B 156 (1979) 135 ; J. Schonfeld, Nucl. Phys. B
185 (1981) 157.

[6] S. Deser, R. Jackiw and S. Templeton, Ann. of Phys. 140 (1982) 372.

[7] J.M. Leinaas and J. Myrheim, Nuo. Cim. B 37 (1977) 1, For a recent
review of this field see A. Khare, Current Sc. (India) 61 (1991) 826.

[8] J. Frohlich and P.A. Marchetti, Comm. Math. phys. 121 (1989) 177.

[9] P.K. Ghosh and A. Khare, Bhubaneswar preprint IP/BBSR/94-14, hep-

th 9404015.

[10] A. khare, Forts. der phys. 38 (1990) 507.

[11] J. Hong, Y. Kim and P.Y. Pac, Phys. Rev. Lett. 64 (1990) 2330 ; R.
Jackiw and E.J. Weinberg, Phys. rev. Lett. 64 (1990) 2334.

[12] R. Jackiw and S-Y. Pi, Phys. Rev. Lett. 64 (1990) 2969 ; Phys. Rev. D
42 (1990) 3500.

[13] A. Khare, Phys. Rev. D 46 (1992) R 2287.

[14] S. Coleman and B. Hill, Phys. Lett. B 159 (1985) 184.

[15] A. Khare and T. Pradhan, Phys. Lett. B 231 (1989) 178.

[16] A. Khare, R.B. MacKenzie and M.B. Paranjape, Phys. Lett. B (1995)
In Press.

22
[17] C. R. Hagen, P.K. Panigrahi and S. Ramaswami, Phys. Rev. Lett. 61
(1988) 389.

[18] R.D. Pisarski and S. Rao, Phys. Rev. D 32 (1985) 2081.

[19] G. Giavarini, C.P. Martin and F. Ruiz Ruiz, Nucl. Phys. B 381 (1992)
222.

[20] A. Khare, R.B. MacKenzie, P.K. Panigrahi and M.B. Paranjape, Univ.
de Montreal preprint UdeM-LPS-TH-150, hep-th/9306027. This paper
had raised this interesting question but because of numerical mistakes
in computation they arrived at wrong conclusion. Recently the correct
answer to the issue has been given by L. Chen, G. Dunne, K. Haller and
E. Lim-Lombridas, Univ. of Connecticut preprint UCONN-94-8, hep-th
9411062.

[21] A.N. Redlich, Phys. Rev. D 29 (1984) 2366.

[22] S.K. Paul and A. Khare, Phys. Lett. 193 (1987) 253.

[23] Y. Kao and M. Suzuki, Phys. Rev. D 31 (1985) 2137 ; M. Bernstein

and T. Lee, Phys. Rev. D 32 (1985) 1020.

[24] V.P. Spiridonov, JETP Lett. 52 (1990) 513 ; V.P. Spiridonov and F.V.
Tkachov, Phys. Lett. B 260 (1991) 109.

[25] S. yu. Khlebnikov, JETP Lett. 51 (1990) 81 ; V.P. spiridonov, Phys.

Lett. B 247 (1990) 337.

[26] L.D. Landau and E.M. Lifshitz, Electrodynamics of Continuous Media,

Second Edition, Pergamon Press (1963) ; T.H. ODell, The Electrody-
namics of Magneto-Electric Media, North Holland (1970).

[27] J. Stern, Phys. Lett. B 265 (1991) 119.

[28] M.E. Carrington and G. Kunstatter, Phys. Lett. B 321 (1994) 223.

[29] Y. Georgelin and J.C. Wallet, Int. J. Mod. Phys. A7 (1992) 1149.

[30] I.I. Kogan, Phys. Lett. B 262 (1991) 83.

23
[31] E. Witten, Comm. Math. Phys. 121 (1989) 351 ; For the abelian case
see, M. Bos and V.P. Nair, Phys. Lett. B223 (1989) 61.

[32] E. Witten, Phys. Lett. B 86 (1979) 283.

[33] V.I. Inozemstsev, Euro. Phys. Lett. 5 (1988) 113 ; G. Lozano, M.V.
Manias and F.A. Schaposnik, Phys. Rev. D 38 (1988) 601.

[34] S.K. Paul and A. Khare, Phys. Lett. B 171 (1986) 244.

[35] L. Jacobs and C. Rebbi, Phys. Rev. B 19 (1979) 4486.

[36] L. Jacobs, A. khare, C.N. Kumar and S.K. Paul, Int. J. Mod. Phys. A
6 (1991) 3441.

[37] S. Deser and Z. Yang, Mod. Phys. Lett. A 4 (1989) 2123.

[38] D.P. Jatkar and A. Khare, Phys. Lett. B 236 (1990) 283.

[39] E.B. Bogomolnyi, Sov. J. Nucl. Phys. 24 (1977) 449.

[40] P. di Vecchia and S. Ferrara, Nucl. Phys. B 130 (1977) 93.

[41] C. Lee, K. Lee and E. Weinberg, Phys. Lett. 243 (1990) 105.

[42] S.N. Behera and A. Khare, Pramana (J. Phys., India) 15 (1980) 245.

[43] A. Khare, Phys. Lett. B 255 (1991) 393.

[44] A. Khare, Phys. Lett. B 277 (1992) 123.

[45] C. Taubs, Comm. Math. Phys. 72 (1980) 277.

[46] R. Wang, Comm. Math. Phys. 137 (1991) 587.

[47] R. Jackiw, K. Lee and E. Weinberg, Phys. Rev. D 42 (1990) 3488.

[48] A. Khare, Phys. Lett. B 263 (1991) 227.

[49] D.P. Jatkar and A. Khare, J. Phys. A 24 (1991) L1001 ; D. Bazeia,

Phys. Rev. D 43 (1991) 4074.

[50] J. Spruck and Y. Yang, Comm. math. Phys. 149 (1992) 361.

24
[51] C. Lee, K. Lee and H. Min, Phys. Lett. B 255 (1990) 79.

[52] P.K. Ghosh, Phys. Lett. B 326 (1993) 264; Phys. Rev. D 49 (1994)
5458 ; J. Lee and S. Nam, Phys. Lett. B 261 (1991) 79 ; M. Torres,
Phys. Rev. D 49 (1994) R 2295.

[53] K. Lee, Phys. Rev. Lett. 66 (1991) 553 ; L.F. Cugliandolo et al., Mod.
Phys. Lett. A 6 (1991) 479.

[54] Z. Ezawa, N. Hotta and A. Iwazaki, Phys. Rev. D 44 (1991) 452.

[55] G. Dunne and C. Trugenberger, Phys. Rev. D 43 (1991) 1332.

[56] B. Grossman, phys. Rev. Lett. 65 (1990) ; G. Dunne, R. Jackiw, S-Y.

Pi and C. Trugenberger, Phys. Rev. D43 (1991) 1332.

[57] N. Manton, Phys. Lett. B 110 (1982) 54 ; Phys. Lett. B 154 (1985)
397 ; P.I. Ruback, Nucl. Phys. B 296 (1988) 669 ; T.M. Samols, Phys.
Lett. B 244 (1990) 285.

[58] S.K. Kim and H. Min, Phys. Lett. B 281 (1992) 81.

[59] L. Hua and C. Chou, MIT preprint MIT-CTP No. 2064 (1992).

[60] M. Hindmarsh, Phys. Rev. Lett. 68 (1991) 1263 ; G.W. Gibbons, M.E.
Ortiz, F. Ruiz Ruiz and T.M. Samols, Nucl. Phys. 385 (1992) 127 .

25