767

Stueckelberg mechanism for

antisymmetric tensor fields
S.V. Kuzmin and D.G.C. McKeon
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

Abstract: It is shown how vector Stueckelberg fields can be introduced to ensure gauge
invariance for mass terms for an antisymmetric tensor field. Scalar Stueckelberg fields allow
one to have gauge invariance for these vector fields. Both the Abelian and non-Abelian cases
are considered. Fully antisymmetric rank-three tensor fields and symmetric rank-two tensor
fields are also discussed.
PACS No.: 11.15-1

Résumé : Nous montrons comment l’introduction de champs vectoriels de Stueckelberg peut

assurer l’invariance de jauge pour les termes de masse dans le cas d’un champ tensoriel
antisymétrique. Les champs scalaires de Stueckelberg permettent d’avoir l’invariance de jauge
pour ces champs vectoriels. Nous étudions les cas Abélien et non Abélien. Nous analysons
For personal use only.

aussi des champs tensoriels de rang trois complètement antisymétriques et des champs
symétriques de rang deux.
[Traduit par la Rédaction]

1. Introduction
Gauge invariance is the property of massless vectors that permits one to renormalize models in
which they mediate interactions. By introducing auxiliary “Stueckelberg” fields [1,2], one can retain
gauge invariance even if the fields become massive; renormalizability is present in the Abelian case [3],
but not in the non-Abelian case [4]. Stueckelberg fields can also be used in conjunction with Abelian
Chern–Simons theory [5] and the U (1) sector of the standard model [6]. It has also recently proved
possible to consider a vector Stueckelberg field when considering massive gravitons [7,8].
The antisymmetric tensor field φµν has a gauge invariance

φµν → φµν + ∂µ fν − ∂ν fµ (1)

provided one deals with the Lagrangian [9–13]

1
2
L0 = − ∂µ φνλ + ∂λ φµν + ∂ν φλµ (2)
6

Received 15 August 2001. Accepted 15 February 2002. Published on the NRC Research Press Web site at
http://cjp.nrc.ca/ on 28 June 2002.
S.V. Kuzmin. Department of Physics and Astronomy, University of Western Ontario, London, ON N6A 5B7,
Canada.
D.G.C. McKeon.1 Department of Applied Mathematics, University of Western Ontario, London, ON N6A 5B7,
Canada.
1
Corresponding author (e-mail: [email protected]).

Can. J. Phys. 80: 767–779 (2002) DOI: 10.1139/P02-038 © 2002 NRC Canada
 768 Can. J. Phys. Vol. 80, 2002

It is possible to couple φµν to an Abelian gauge field Aµ in a way that preserves the symmetry of (1)
 
1 1
L = L0 + m ∗ Fµν φµν − Fµν Fµν Fµν = ∂µ Aν − ∂ν Aµ , ∗ Fµν = µνλσ Fλσ (3)
4 2
One can now show either by rearranging the degrees of freedom in (3) [12] or by direct calculation of
the propagator for Aµ [14] that the vector field Aµ acquires a mass proportional to m.
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

Modifying the action so that the gauge symmetry of (1) becomes non-Abelian is not straightforward.
Simply converting ordinary derivatives in (1) and (2) to gauge covariant derivatives does not leave us
with a gauge-invariant Lagrangian. One must introduce auxiliary fields if (2) is to survive as a kinetic
Lagrangian for φµν ; this has been done in a variety of ways [15–18]. It is also possible to discard having
a kinetic term for φµν and keep it as an auxiliary field that couples to a dual field strength [19,20].
These models are equivalent to either a massive scalar or a Stueckelberg model for a massive vector in
the Abelian case, or a nonlinear sigma model or a Kunimasa–Goto model for a massive vector in the
non-Abelian case. (It is also possible to consider such models in two and three dimensions [21].)
In this paper, we show how a vector Stueckelberg field can be used as an auxiliary field to generalize
the derivatives in (1) and (2) to gauge covariant derivatives. Furthermore, we can use additional scalar
Stueckelberg fields to ensure ordinary gauge symmetry for these vector fields. The Stueckelberg fields
can also be used to construct gauge-invariant mass terms. (In fact, it proves possible to introduce a
pseudoscalar mass for φµν .) Details of these models are presented in the subsequent sections. We also
briefly consider gauge invariance for a three-component antisymmetric tensor field κµνλ and for a
For personal use only.

symmetric tensor field hµν .

2. Massive tensor fields

We first note that the “field strength”

µνλ = ∂µ φνλ + ∂ν φλµ + ∂λ φµν (4)

and the quantity

1

φ̃µν = φµν + √ ∂µ Wν − ∂ν Wµ (5)
2µ
are both invariant under the gauge transformations

φµν → φµν + ∂µ fν − ∂ν fµ (6a)

√
Wµ → Wµ − 2µfµ (6b)

The field Wµ can be considered as being a “vector Stueckelberg field”. Using (4) and (5), we can
construct the gauge-invariant classical Lagrangian
1 1 1 ∗
Lcl = − µνλ µνλ − µ2 φ̃µν φ̃µν − λµφ̃µν φ̃µν (7)
6 2 2
To fix the gauge invariance of (6) in (7), we supplement Lcl with a gauge-fixing Lagrangian
  2
1 µβ β
Lβ = − ∂µ φµν + √ Wν + ∂ν σ (8)
β 2 2

A second auxiliary field (a scalar Stueckelberg field) σ has been introduced into the Lagrangian (8). As
a consequence, even though the gauge invariance of (6) is broken by (8), there still resides a residual
©2002 NRC Canada
 Kuzmin and McKeon 769

gauge invariance

Wµ → Wµ + ∂µ  (9a)

σ → σ − µ β (9b)

To break this gauge symmetry we introduce a second gauge-fixing Lagrangian

Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

1  2
Lγ = − ∂ · W + µγ βσ (10)
2γ
Collecting terms, we find that

Leff = Lcl + Lγ + Lβ
1

 1 1
= − ∂µ φνλ ∂µ φνλ − µ2 φµν φµν − λµφµν ∗ φµν
2  2 2
1

 1 1
− 1− ∂µ φνλ ∂ν φλµ − Gµν Gµν − µ2 βW 2
β 4 2
1 λ 1 1
− γ (∂ · W )2 − √ φµν ∗ Gµν − (∂µ σ )2 − µ2 γβσ 2 (11)
2 2 2 2
with Gµν = ∂µ Wν − ∂ν Wµ . In the gauge in which γ = β = 1, Leff reduces to
For personal use only.


  
1
 ∂ 2 − µ2 µνλσ − 21 λµµνλσ − √λ µνρβ ∂ρ φλσ
Leff = φµν , Wα
2 2 2
2 − √λ λσρα ∂ρ ∂ − µ δαβ Wβ
2
1
+ σ ∂ 2 − µ2 σ (12)
2
where
1

αβ,γ δ = δαγ δβδ − δαδ δβγ (13)
2
In Appendix A, the inverse of the matrix in (12) is worked out; it is shown that the propagator for φ is

1 λ2 ∂ 2
< φµν φλσ >=  µν,λσ +
2
 µν,λσ
∂ 2 − µ2 + λ2 ∂ 2 − µ2 ∂ 2 − µ2 + λ2

2 
λµ ∂ − µ2
+ µνλσ
2
∂ 2 − µ2 2 + λ2 ∂ 2
∂ 2 − µ2 + λ2 

µλ3 ∂ 2 L R
+
2

 Zµν,λσ + Zµν,λσ (14)
∂ 2 − µ2 +λ ∂
2 2 ∂ −µ ∂ −µ +λ
2 2 2 2 2

where
1

µν,λσ = δµλ ∂ν ∂σ + δνσ ∂µ ∂λ − δµσ ∂ν ∂λ − δνλ ∂µ ∂σ (15a)
2∂ 2
L 1

Zµν,λσ = 2 µνλρ ∂ρ ∂σ − µνσρ ∂ρ ∂λ (15b)
2∂
R 1

Zµν,λσ = 2 λσ µρ ∂ρ ∂ν − λσ νρ ∂ρ ∂µ (15c)
2∂

©2002 NRC Canada

 770 Can. J. Phys. Vol. 80, 2002

It is evident that to avoid tachyons, we must have

λ 2 < µ2 (16)

Similarly, the propagator for Wα is given by

 
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

µ2 /λ2 1 − µ2 /λ2 µ2
< Wα Wβ >= + −
2 δαβ
∂ 2 − µ2 ∂ 2 − µ2 + λ2 ∂ 2 − µ2
λ2 ∂ 2 ∂α ∂β
+
2
 2 (17)
∂ 2 − µ2 ∂ 2 − µ2 + λ2 ∂

We now consider coupling of φµν , Wα to a gauge field Aµ .

3. Gauge couplings
One way to couple an Abelian gauge field Aµ with field strength Fµν = ∂µ Aν − ∂ν Aµ is to simply
supplement Leff in (11) with a direct coupling [12,14] between ∗ Fµν and φµν as in (3); in this way the
invariances of (6) and (9) are not broken and one also retains the gauge invariance

Aµ → Aµ + ∂µ θ (18)
For personal use only.

A second approach that is considered here is to complexify the “matter” fields φµν , Wα and replace
ordinary derivatives ∂µ by gauge-covariant derivatives Dµ = ∂µ +iAµ . To do this without compromising
the invariances of (6) and (9) requires some care, as the quantity
(0)
µνλ = Dµ φνλ + Dν φλµ + Dλ φµν (19)

does not possess the invariance

φµν → φµν + Dµ fν − Dν fµ (20)

which is the natural generalization of (6a). (The gauge function fµ has been complexified in addition
to φµν and Wα .) However, since

Dµ , Dν = iFµν (21)

we see that the gauge invariance of (20) is preserved by

(1) (0) i

µνλ = µνλ + √ Fµν Wλ + Fνλ Wµ + Fλµ Wν (22)
2µ
if, as in (6b)
√
Wµ → Wµ − 2µfµ (23)

Thus, it is possible to generalize the transformation of (6). Similarly

(0) 1

φ̃µν = φµν + √ Dµ Wν − Dν Wµ (24)
2µ
is invariant under the transformations of (20) and (23).

©2002 NRC Canada

 Kuzmin and McKeon 771

Furthermore, if we wish to generalize the invariance of (9) when there is coupling to a gauge field
Aµ , we must further modify (22) and (24). By taking

(1) i

µνλ = µνλ + √ Fµν Dλ ρ + Fνλ Dµ ρ + Fλµ Dν ρ (25a)
2µ2
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

and

(0) i
φ̃µν = φ̃µν +√ Fµν ρ (25b)
2µ2

we find µνλ and φ̃µν are invariant under not only (20) and (23) but also under the transformations

Wµ → Wµ + Dµ  (26a)
ρ → ρ − µ (26b)

A gauge-invariant Lagrangian, constructed in analogy with (7), is given by

1 1
Lcl = − Fµν Fµν − ∗µνλ µνλ − µ2 φ̃µν
∗ ∗ ∗
φ̃µν − λµφ̃µν φ̃µν (27)
4 3
It possesses invariance under the transformation of (20), (23), and (26) as well as
For personal use only.

Aµ → Aµ + ∂µ θ (28a)
−iθ −iθ −iθ
φµν → e φµν , Wµ → e Wµ , ρ→e ρ (28b)

An appropriate gauge-fixing term associated with the invariances of (20) and (23) is
 2
2 µβ 1
Lβ = − Dµ φµν + √ Wν + Dν ρ (29a)
β 2 µ

This does not break the gauge invariance of (26); to do this we introduce

1 √ 2
Lγ = − Dµ Wµ + 2βγρ (29b)
γ

Lastly, we break the symmetry of (28) by introducing the usual gauge-fixing term

1
Lα = − (∂ · A)2 (29c)
2α
The gauge-fixing choices in (29) do not alter the form of the propagators in (14) and (17).
It is now possible to generalize these considerations to deal with the non-Abelian situation. The
a , W a , ρ a , and Aa are all now in the adjoint representation of some compact group whose
fields φµν µ µ
structure constants are cabc . The appropriate generalization of (27), supplemented by a direct coupling
between φµν a and ∗ F a , is given by
µν

1 a a 1 1 λµ a ∗ a
Lcl = − Fµν Fµν − aµνλ aµνλ + mφµν
a ∗ a a
Fµν − µ2 φ̃µν a
φ̃µν − φ̃ φ̃ (30)
4 6 2 2 µν µν
where

Dµab = ∂µ δ ab + capb Apµ

©2002 NRC Canada
 772 Can. J. Phys. Vol. 80, 2002

ab
Dµ , Dν = capb Fµν
p

aµνλ = Dµab φνλ

b
+ Dνab φλµ
b
+ Dλab φµν
b
      
cabc b c 1 cd d b c 1 cd d b c 1 cd d
+√ Fµν Wλ + Dλ ρ + Fνλ Wµ + Dµ ρ + Fλµ Wν + Dν ρ
2µ µ µ µ
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

and
a a 1 1
φ̃µν = φµν +√ Dµab Wνb − Dνab Wµb + √ cabc Fµν
b c
ρ
2µ 2µ 2

The Lagrangian of (30) possesses the invariances

a a
φµν → φµν + Dµab fνb − Dνab fµb + cabc φµν
b c
θ (31a)
a a ab b
√ a abc b c
Wµ → Wµ + Dµ  − 2µfµ + c Wµ θ (31b)
a a a abc b c
ρ → ρ − µ + c ρ θ (31c)
Aaµ → Dµab θ b (31d)

necessitating the introduction of gauge fixing (much as in (29)) given by

For personal use only.

  2
1 µβ 1
Lβ = − Dµab φµν
b
+ √ Wνa + Dνab ρ b (32a)
β 2 µ
1 √ 2
Lγ = − Dµab Wµb + 2βγρ a (32b)
2γ
1
Lα = − (∂ · A)2 (32c)
2α
Determining the contribution of the ghost fields associated with these gauge-fixing terms is a delicate
issue and will not be examined here. These considerations can now be extended to the situation in which
there is a rank-three antisymmetric tensor field κµνλ .

4. Rank-three tensor field

If one has a rank-three antisymmetric tensor field κµνλ , then the field strength

Kµνλσ = ∂µ κνλσ − ∂ν κλσ µ + ∂λ κσ µν − ∂σ κµνλ (33)

is invariant under the transformation

κµνλ → κµνλ + ∂µ λνλ + ∂ν λλµ + ∂λ λµν (34)

and hence one can form an invariant Lagrangian

1
L = − Kµνλσ Kµνλσ (35)
8
By introducing a tensor Stueckelberg field ζµν , it is possible to form, in analogy with (5), an invariant
quantity
1

κ̃µνλ = κµνλ + √ ∂µ ζνλ + ∂ν ζλµ + ∂λ ζµν (36)
3µ
©2002 NRC Canada
 Kuzmin and McKeon 773

provided
√
ζνλ → ζνλ − 3µλνλ (37)
when κµνλ undergoes the transformation of (34). The Lagrangian of (35) can now be supplemented by
the mass term
µ2
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

L=− κ̃µνλ κ̃µνλ (38)

2
no analogue of the pseudoscalar mass term in (7) can be defined.
Coupling to a gauge field Aµ can be effected in a manner similar to the coupling of (27). (A
direct coupling between κµνλ and ∗ Fµν does not appear to be possible in four dimensions, although in
five dimensions one could consider the gauge-invariant direct coupling µνλσρ κµνλ Fσρ . We could also
couple κµνλ to a scalar a by considering aµνλσ Kµνλσ .) The gauge-invariant Lagrangian we introduce
is
1 1 ∗ ∗
L = − Fµν Fµν − Kµνλσ Kµνλσ − µ2 κ̃µνλ κ̃µνλ (39)
4 4
where now
Kµνλσ = Dµ κ̃νλσ − Dν κ̃λσ µ + Dλ κ̃σ µν − Dσ κ̃µνλ (40a)
1
κ̃µνλ = κµνλ + √ Dµ ζ̃νλ + Dν ζ̃λµ + Dλ ζ̃µν (40b)
For personal use only.

3µ
1
ζ̃µν = ζµν + √ Dµ W̃ν − Dν W̃µ (40c)
2µ
1
W̃µ = Wµ + Dµ ρ (40d)
µ
The invariances possessed by the Lagrangian (39) are
Aµ → Aµ + ∂µ θ (41a)



κµνλ , ζµν , Wµ , ρ → eiθ κµνλ , ζµν , Wµ , ρ (41b)

κµνλ → κµνλ + Dµ λνλ + Dν λλµ + Dλ λµν (42a)

√
ζµν → ζµν − 3µλµν (42b)

ζµν → ζµν + Dµ fν − Dν fµ (43a)

√
Wµ → Wµ − 2µfµ (43b)
and
Wµ → Wµ + Dµ  (44a)
ρ → ρ − µ (44b)
These invariances clearly generalize the invariances of (20), (23), and (28), which are associated with
the Lagrangian of (27). Gauge fixings analogous to (29) now take the form
1 3 µδ 2
Lgf = − (∂ · A)2 − Dµ κµνλ + √ ζ̃νλ
2α δ 3
2 µβδ 2 1 2
− Dµ ζµν + √ W̃ν − Dµ Wµ + 2µβδ 2 ρ (45)
β 2 γ

©2002 NRC Canada

 774 Can. J. Phys. Vol. 80, 2002

The terms bilinear in the fields occurring in L + Lgf are by (39) and (45)

1
2 1
L(2) = − ∂µ Aν − ∂ν Aµ − (∂ · A)2
4 2α
1 2
− ∂µ κνλσ − ∂ν κλσ µ + ∂λ κσ µν − ∂σ κµνλ
4
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

1
2
− µ2 κµνλ + √ ∂µ ζνλ + ∂ν ζλµ + ∂λ ζµν
3µ
 2
3 µδ 1
− ∂µ κµνλ + √ ζνλ + √ (∂ν Wλ − ∂λ Wν )
δ 3 2µ
 2
2 µβδ 1 1 2
− ∂µ ζµν + √ Wν + ∂ν ρ − ∂µ Wµ + µβδ 2 ρ (46)
β 2 µ γ

which is diagonal in each of the fields. If α = β = γ = δ = 1, then (46) reduces to

1
2
L(2) = − ∂µ Aν − |∂µ κνλσ |2 − µ2 |κµνλ |2
2
− |∂µ ζνλ |2 − µ2 |ζµν |2 − |∂µ Wν |2 − µ2 |Wµ |2 − |∂ν ρ|2 − |µ2 ρ|2 (47)
For personal use only.

It is quite straightforward to find a non-Abelian generalization of (39) and (32) in the same manner
that (30) and (32) were obtained from (27) and (29).
We now consider a symmetric tensor field hµν .

5. Symmetric rank-two tensors

The symmetric rank-two tensor field hµν is naturally associated with the graviton [22]. Coupling
hµν to the vector gauge field Aµ by complexifying hµν , even with the assistance of a vector Stueckelberg
field Vµ , does not appear to be feasible as there is no field strength that is invariant under a spin-two
gauge transformation [23]. However, one can use the Stueckelberg field Vµ to introduce a mass term
for hµν that does in fact preserve spin-two gauge invariance [7,8]. We will briefly illustrate this.
√
The form of the free spin-two Lagrangian obtained from an expansion of the combination gg µν
√ µν
about a flat background (viz. gg = δµν + hµν ) to lowest order is [24]

1 1
L = − hµν ∂ 2 hµν + h∂ 2 h + hµλ ∂µ ∂ν hνλ (h ≡ hλλ ) (48)
2 4
this possesses the gauge invariance

hµν → hµν + ∂µ φν + ∂ν φν − δµν ∂ · φ (49)

The so-called Fierz–Pauli mass term [22,25]

m2
Lm = hµν hµν − h2 (50)
2
breaks the symmetry of (49). This is restored by introducing a vector Vµ , which, when hµν transforms
according to (49), undergoes the change
√
Vµ → Vµ − 2mφµ (51)
©2002 NRC Canada
 Kuzmin and McKeon 775

This means that

    
m2 1
 2 2∂ · V 2
Lm = hµν + √ ∂µ Vν + ∂ν Vµ − δµν ∂ · V − h− √
2 2m 2m
 
m2 1
 1
2
= h2µν − h2 + √ 4hµν ∂µ Vν + 2h∂ · V + 2 ∂µ Vν − (∂ · V )2 (52)
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

2 2m m

is gauge invariant. We break this gauge invariance by the gauge-fixing Lagrangian

 2
1 1 √
Lgf = ∂µ hµν + ∂ν h − 2mαVν (53)
2α 2

As a result we obtain
   
1 1 1
L + Lm + Lgf = − hµν ∂ − m hµν + h ∂ 1 −
2 2 2
− 2m h
2
2 4 2α
 
1 1 1
+ hµλ ∂µ ∂ν hµν 1 − − h∂µ ∂ν hµν + Vµ δµν ∂ 2 − ∂µ ∂ν − 2m2 αδµν Vν (54)
2α 2α 2

The full bilinear Lagrangian appearing in (54) has an associated propagator that is discussed in the
For personal use only.

Appendix A.
The form of the contribution of the vector field Vµ to (53) is that of a Proca field. One can restore
gauge invariance to this contribution by using a scalar Stueckelberg field χ [8]; Vµ is replaced by
∂µ χ
Vµ + and a gauge-fixing term
m
1
Lgf = (∂ · V + 2mαβχ )2 (55)
2β

is used to break the invariance

Vµ → Vµ + ∂µ ω (56a)
χ → χ − mω (56b)

The fields Vµ and χ decouple and the action for χ is that of a massive scalar. The five degrees of freedom
normally associated with a massive spin-two field have been redistributed; there are two helicity states
associated with the spin-two gauge field, hµν , two with the spin-one gauge field Vµ , and one with the
scalar field χ . If matter fields couple exclusively to hµν and do not couple to Vµ and χ , the discontinuity
at m2 = 0 found in refs. 4 and 26 disappears.

6. Discussion
In this paper, we have demonstrated that gauge invariance can be preserved in models involving
antisymmetric tensor fields using a generalization of the approach employed by Stueckelberg for a
U (1) vector theory. Not only kinetic but also mass terms for these models appear if one chooses suitable
gauge-fixing terms.
Of course, it is necessary, in general, to employ more than gauge-fixing terms to quantize these
models [27]; ghost fields must be introduced in general. A full treatment of the quantized model likely
would entail use of the Batalin–Fradkin–Vilkovisky formalism for quantizing constrained systems [28]
as is the case with the Freedman–Townsend model [29].
©2002 NRC Canada
 776 Can. J. Phys. Vol. 80, 2002

Coupling to the metric field by replacing the flat metric with a general metric and by using covariant
derivatives is also possible. We also note that antisymmetric tensor fields naturally occur in extended
supergravity models [30]. It should also be possible to arrange that the spin-3/2 graviton becomes
massive using auxiliary spin-1/2 Stueckelberg fields in supergravity models.
The occurrence of interaction terms with couplings inversely proportional to the mass parameter (as
in (27)) likely jeopardizes renormalizability, although possibly the high degree of symmetry in these
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

models may render the models renormalizable. This merits further investigation. One might begin by
examining the divergences in the model
1 e 2 1 2 1
L=− Dµ Wν − Dν Wµ + i Fµν φ − µ 2 Wµ + Dµ φ − Fµν Fµν (57)
2 µ µ 4
which possess the invariances
 
1
Aµ → Aµ + ∂µ ω (58a)
e



Wµ , φ → eiω Wµ , φ (58b)

Wµ → Wµ + Dµ  (59a)
φ → φ − µ (59b)
For personal use only.

These two-gauge invariances can be most conveniently broken by the gauge-fixing terms
1
Lα = − (∂ · A)2 (60a)
2α
1
Lβ = − |Dµ Wµ + µβφ|2 (60b)
β
The Faddeev–Popov procedure in conjunction with (59) and (60b) leads to the ghost Lagrangian
 
Lgh = c D 2 − µ2 β c (61)

This model resembles that of (27) and (29). We also note that the usual gauge coupling to a charged scalar
 can be supplemented by the gauge-invariant interaction exp(iφ/µ) + H.C. provided  → ei .
We have also shown how the Stueckelberg formalism can be used to have gauge-invariant mass terms
for the spin-two gauge field [7,8]. This may be of interest especially in 1 + 1-dimensional spin-two
models as such a mass term could provide dynamics for the spin-two field in 1 + 1 dimensions [31].
This would be similar to what has been shown to happen when a Stueckelberg mass term has been used
in conjunction with a U (1) Chern–Simons model in 2 + 1 dimensions [5].
It is worth examining where the physical degrees of freedom reside in the models of massive two-
forms φµν and three-forms κµνλ considered above. For the real massive two-form φµν , the equation of
motion


∂µ ∂µ φνλ + ∂ν φλµ + ∂λ φµν = µ2 φνλ (62)

implies the constraint ∂µ φµλ = 0 if µ2 = 0; this has the effect of reducing the anticipated number of
degrees of freedom for the antisymmetric field φµν from six to four. When gauge invariance is restored
by complexifying φµν and using Stueckelberg fields as in (27), then Reφµν and Imφµν retain only one
degree of freedom each; of the three other degrees of freedom, associated with Reφµν , two reside in
the massless vector field ReWµ and one in the scalar field Reρ (and similarly for Imφµν ). For the real
©2002 NRC Canada
 Kuzmin and McKeon 777

three-form κµνλ dealt with in Sect. 5, the massive field again has four degrees of freedom. However,
when gauge invariance has been restored, as in (46), these degrees of freedom are entirely absorbed by
the forms ζµν , Wµ , and ρ.
In principle, one could employ the approach to introducing Stueckelberg p-forms considered here
in higher dimensions. In general, in d dimensions gauge-invariant Lagrangians for p-forms with p ≤ d
can be defined; mass terms can be inserted without breaking this gauge invariance by use of a (p − 1)
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

form. Gauge breaking can be implemented without losing an additional gauge invariance associated
with this (p − 1) form by introduction of a further (p − 2) form, which acts as another Stueckelberg
field; this process can be repeated until finally one has a Stueckelberg scalar.
String models and supergravity have been responsible for recent interest in p-forms. Both type
IIA and type IIB strings have an antisymmetric two-form in their bosonic spectrum, while a type IIA
string also has a three-form and a type IIB string has a self-dual four-form. We also note the presence
of a three-form in N = 1 supergravity in eleven dimensions. It would be of interest to investigate
the possibility of generating masses for these p-forms using a supersymmetric generalization of the
Stueckelberg mechanism outlined in this paper. We currently are examining the Stueckelberg mechanism
in conjunction with supersymmetric quantum electrodynamics when supersymmetry is broken by a
Fayet–Iliopolis D-term.

Acknowledgements
F.A. Dilkes had helpful comments. NSERC provided financial support. R. and D. MacKenzie made
For personal use only.

a useful suggestion.

References
1. E.C. Stueckelberg. Helv. Phys. Acta, 30, 209 (1957).
2. T. Kunimasa and T. Goto. Prog. Theor. Phys. 37, 452 (1967); A.A. Slavnov and L.D. Faddeev. Theor.
Math. Phys. 3, 312 (1971); Yu and N. Kafiev. Nucl. Phys. B201, 341 (1982).
3. P.T. Matthews. Phys. Rev. 76, 1657 (1949).
4. H. van Dam and M. Veltman. Nucl. Phys. B22, 397 (1970); A.I. Vainshtein and I.B. Khriplovich. Sov. J.
Nucl. Phys. 13, 111 (1971).
5. F.A. Dilkes and D.G.C. McKeon. Phys. Rev. D: Part. Fields, 52, 4668 (1995).
6. S.V. Kuzmin and D.G.C. McKeon. Mod. Phys. Lett. A16, 747 (2001).
7. P.A. Grassi and P. van Niewenhuizen. Phys. Lett. 799A, 174 (2001).
8. F.A. Dilkes, M.J. Duff, J.T. Liu, and H. Sati. Phys. Rev. Lett. 87, 041301 (2001).
9. V.I. Ogievetsky and I.V. Polubarinov. Yad. Fiz. 4, 210 (1968).
10. K. Hayashi. Phys. Lett. 44B, 497 (1973).
11. M. Kalb and P. Raymond. Phys. Rev. D: Part. Fields, 9, 2273 (1974).
12. E. Cremmer and J. Scherk. Nucl. Phys. B72, 117 (1974).
13. Y. Nambu. Phys. Rep. 23, 250 (1976).
14. T.J. Allen, M.J. Bowick, and A. Lahiri. Mod. Phys. Lett. A6, 559 (1991); Phys. Lett. 237B, 47 (1989).
15. L. Smolin. Phys. Lett. 137B, 379 (1984).
16. L. Baulieu and J. Thierry-Mieg. Phys. Lett. 144B, 221 (1984).
17. J. Thierry-Mieg and Y. Ne’eman. Proc. Nat. Acad. Sci. 79, 7068 (1982).
18. D.S. Hwang and C.Y. Lee. Int. J. Mod. Phys. A, 10, 3649 (1995).
19. D.Z. Freedman and P.K. Townsend. Nucl. Phys. B177, 282 (1981).
20. S. Deser and E. Witten. Nucl. Phys. B178, 491 (1981).
21. D.G.C. McKeon. Int. J. Mod. Phys. 7, 2005 (1992).
22. M. Fierz and W. Pauli. Proc. R. Soc. London A, 173, 211 (1939).
23. D.G.C. McKeon. Can. J. Phys. 57, 2096 (1979).
24. D.M. Capper, G. Leibbrandt, and M. Ramon-Medrano. Phys. Rev. D: Part. Fields, 8, 4320 (1973).
25. M. Fierz. Helv. Phys. Acta, 12, 3 (1939).
26. V.I. Zakharov. JETP Lett. 12, 312 (1970).

©2002 NRC Canada

 778 Can. J. Phys. Vol. 80, 2002

27. L.D. Faddeev and V.N. Popov. Phys. Lett. 25B, 29 (1967); B.S. DeWitt. Phys. Rev. 162, 1195 (1967);
162, 1239 (1967); R.P. Feynman. Acta Phys. Pol. 24, 697 (1963).
28. I.A. Batalin and G.A. Vilkovisky. Phys. Rev. D: Part. Fields, 28, 2567 (1983); E.S. Fradkin and G.A.
Vilkovisky. Phys. Lett. 55B, 224 (1975); I.A. Batalin and G.A. Vilkovisky. Phys. Lett. 69B, 309 (1977);
I.A. Batalin and E.S. Fradkin. Phys. Lett. 122B, 157 (1983).
29. M. Leblanc, D.G.C. McKeon, A. Rebhan, and T.N. Sherry. Mod. Phys. Lett. A6, 3359 (1991).
30. E. Cremmer, B. Julia, and J. Scherk. Phys. Lett. 76B, 409 (1978).
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

31. D.G.C. McKeon and T.N. Sherry. J. Phys. 23, 4691 (1990).
32. P. van Niewenhuizen. Nucl. Phys. B60, 478 (1973).

Appendix A: Derivation of equations

In this appendix we first consider inversion of the matrix M occurring in (12). We start by noting
that if A, B, C, and D are block matrices, then
 −1  
A B E −1 −E −1 BD −1
= (A.1a)
C D −D −1 CE −1 D −1 + D −1 CE −1 BD −1

where E = A − BD −1 C. Similarly, we have

 −1  
A B A−1 + A−1 BF −1 CA−1 −A−1 BF −1
= (A.1b)
C D −F −1 CA−1 F −1
For personal use only.

where F = D − CA−1 B.
With the quantities defined in (15) we can show that 2 = , 2 = ,  2 = 4,  = 2Z L ,
Z = 2, Z R = 2( − ). Furthermore, we have
L



µνρβ ∂ρ π τ κβ ∂κ = 2∂ 2 ( − ) (A.2a)


 L
µνρβ ∂ρ δπβ ∂τ − δτβ ∂π = −2∂ 2 Zµνπτ (A.2b)


λσρα ∂ρ (λσ πτ ) = 0 (A.2c)


λσρα ∂ρ (λσ πτ ) = −2 (δπ α ∂τ − δτ α ∂π ) (A.2d)

 L
λσρα ∂ρ Zλσ πτ = δατ ∂π − δαπ ∂τ (A.2e)

 R
λσρα ∂ρ Zλσ πτ = 0 (A.2f)

If in (12) we make the identifications

1
A = ∂ 2 − µ2 µν,λσ − λµµνλσ (A.3a)
2
λ
B = − √ µνρβ ∂ρ (A.3b)
2
λ
C = √ λσρα ∂ρ (A.3c)
2
D = ∂ 2 − µ2 δαβ (A.3d)

then one can easily show that

 
−1 λ2 ∂ 2 1 λ2 ∂ 2
A − BD = ∂ 2 − µ2 + µν,λσ − λµµνλσ − 2 µνλσ (A.4)
∂ 2 − µ2 2 ∂ − µ2
©2002 NRC Canada
 Kuzmin and McKeon 779

and hence, by (A.1a), we obtain the propagator of (14) for φµν .

Similarly, we can show that

2 


−1 ∂ − µ2 ∂ 2 − µ2 + λ2 −λ2 ∂ 2 ∂ 2 − µ2 ∂α ∂β
D − CA B =
2 δαβ +
2 (A.5)
∂ 2 − µ 2 − λ 2 µ2 ∂
2
∂ 2 − µ 2 − λ 2 µ2
from which, by using (A.1b), we can derive the propagator for Wµ given by (17).
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13

To determine the propagator for the field hµν appearing in (54), we rewrite the terms in this La-
grangian that are bilinear in the fields hµν as
    2   
(2) p 2 + m2
(2) (1) p + m
2 1 1
L = hµν P +P − 1− p2
2 2 2 2α
 2    
(0) p + m
2 3 1
+ Ps − p 1−
2
+ 2m 2
2 4 2α
 2       
p + m2 1 1 1 1 2
+ Pw(0) − p2 1 − + 2m2 − 1 − p2 + p
2 4 2α 2α 2α
√  (0)  1   1
 
1 2

(0)
+ 3 Psw + Pws − p2 1 − + 2m2 + p hλσ (A.6)
4 2α 2α
Here we use the operators [32]
For personal use only.

(2) 1
 1
Pµνλσ = θµλ θνσ + θµσ θνλ − θµν θλσ (A.7a)
2 3
(1) 1

Pµνλσ = θµλ ωνσ + θνσ ωµλ + θµσ ωνλ + θνλ ωµσ (A.7b)
2
1
Ps(0) = θµν θλσ (A.7c)
µνλσ 3
Pw(0) = ωµν ωλσ (A.7d)
µνλσ
(0) 1
Pws = √ ωµν θλσ (A.7e)
µνλσ 3
(0) 1
Psw = √ θµν ωλσ (A.7f)
µνλσ 3
where p = −i∂, and
θµν = δµν − ωµν (A.8a)
ωµν = pµ pν /p 2
(A.8b)
(0) (0)
The operators P (2) , P (1) , Ps , and Pw comprise a complete set of orthogonal projection operators;
(0) (0) (0) (0) (0) (0) (0) (0) (0) (0)
Psw and Pws satisfy Psw Pws = Ps , Psw Psw = 0, Ps Psw = Psw plus other analogous equations.
As a result we see that
 
αP (2) + βP (1) + γ Ps(0) + δPw(0) +  Psw (0) (0)
+ Pws
 
1 (2) 1 (1) δ (0) γ (0)  (0) (0)
P + P + P + P − P + P = 1 (A.9)
α β γ δ − 2 s γ δ − 2 w γ δ − 2 sw ws

From (A.9), we can invert the operator appearing in (A.6). In particular, the spin-two contribution to
the propagator (coming from P (2) , the spin-two projection operator) is proportional to p2 +m
2
2.

©2002 NRC Canada

 This article has been cited by:

1. T. J. MARSHALL, D. G. C. MCKEON. 2008. RADIATIVE PROPERTIES OF THE STUECKELBERG MECHANISM.

International Journal of Modern Physics A 23:05, 741-748. [CrossRef]
2. HENRI RUEGG, MARTÍ RUIZ-ALTABA. 2004. THE STUECKELBERG FIELD. International Journal of Modern Physics
A 19:20, 3265-3347. [CrossRef]
Can. J. Phys. Downloaded from www.nrcresearchpress.com by University of British Columbia on 06/04/13
For personal use only.