|Views: 152|Likes: 1

Published by Angel Martorell

Phys. Rev. Lett. 111, 021801 (2013)

While a nonzero photon mass has been under experimental and theoretical study

for years, the possible implication of a ﬁnite photon lifetime lacks discussion. The tight experimental upper bound of the photon mass restricts the kinematically allowed ﬁnal states of photon decay to the lightest neutrino and/or particles beyond the Standard Model. We discuss the modiﬁcations of the well-measured cosmic microwave background spectrum of free streaming photons due to photon mass and lifetime and obtain model-independent constraints on both parameters—most

importantly a lower direct bound of 3 yr on the photon lifetime, should the photon mass be at its conservative upper limit. In that case, the lifetime of microwave photons will be time-dilated by a factor order 1015.

.

While a nonzero photon mass has been under experimental and theoretical study

for years, the possible implication of a ﬁnite photon lifetime lacks discussion. The tight experimental upper bound of the photon mass restricts the kinematically allowed ﬁnal states of photon decay to the lightest neutrino and/or particles beyond the Standard Model. We discuss the modiﬁcations of the well-measured cosmic microwave background spectrum of free streaming photons due to photon mass and lifetime and obtain model-independent constraints on both parameters—most

importantly a lower direct bound of 3 yr on the photon lifetime, should the photon mass be at its conservative upper limit. In that case, the lifetime of microwave photons will be time-dilated by a factor order 1015.

.

Phys. Rev. Lett. 111, 021801 (2013)

While a nonzero photon mass has been under experimental and theoretical study

for years, the possible implication of a ﬁnite photon lifetime lacks discussion. The tight experimental upper bound of the photon mass restricts the kinematically allowed ﬁnal states of photon decay to the lightest neutrino and/or particles beyond the Standard Model. We discuss the modiﬁcations of the well-measured cosmic microwave background spectrum of free streaming photons due to photon mass and lifetime and obtain model-independent constraints on both parameters—most

importantly a lower direct bound of 3 yr on the photon lifetime, should the photon mass be at its conservative upper limit. In that case, the lifetime of microwave photons will be time-dilated by a factor order 1015.

.

While a nonzero photon mass has been under experimental and theoretical study

for years, the possible implication of a ﬁnite photon lifetime lacks discussion. The tight experimental upper bound of the photon mass restricts the kinematically allowed ﬁnal states of photon decay to the lightest neutrino and/or particles beyond the Standard Model. We discuss the modiﬁcations of the well-measured cosmic microwave background spectrum of free streaming photons due to photon mass and lifetime and obtain model-independent constraints on both parameters—most

importantly a lower direct bound of 3 yr on the photon lifetime, should the photon mass be at its conservative upper limit. In that case, the lifetime of microwave photons will be time-dilated by a factor order 1015.

.

See More

See less

Phys. Rev. Lett.

111

, 021801 (2013)

How Stable is the Photon?

Julian Heeck

Max-Planck-Institut f¨ ur Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany

Yes, the photon. While a nonzero photon mass has been under experimental and theoretical studyfor years, the possible implication of a ﬁnite photon lifetime lacks discussion. The tight experimentalupper bound of the photon mass restricts the kinematically allowed ﬁnal states of photon decay tothe lightest neutrino and/or particles beyond the Standard Model. We discuss the modiﬁcationsof the well-measured cosmic microwave background spectrum of free streaming photons due tophoton mass and lifetime and obtain model-independent constraints on both parameters—mostimportantly a lower direct bound of 3yr on the photon lifetime, should the photon mass be at itsconservative upper limit. In that case, the lifetime of microwave photons will be time-dilated by afactor order 10

15

.

PACS numbers: 14.70.Bh, 12.20.-m, 12.60.Cn, 98.80.-k

Classical electrodynamics, as encoded in Maxwell’sequations, can be readily extended to allow for a nonzerophoton mass; the resulting Proca equations[1] then de-scribe the behavior of a massive spin-1 ﬁeld, and havesince been used to set an impressive upper limit on thephoton mass of

m <

2

×

10

−

54

−

18

eV inthe natural units used in this Letter (

=

c

=

k

B

= 1).A nonzero photon mass is often dismissed on theoret-ical grounds, as the insertion of a mass term to theLagrangian of quantum electrodynamics (QED) breaksgauge invariance and might therefore spoil renormaliz-ability, i.e. the consistencyof the theoryat quantum level.This is, however, not the case as the Proca Lagrangiancan be viewed as a gauge-ﬁxed version of the St¨uckelbergLagrangian[3], which restores gauge invariance. For anexhaustive review we refer to[4]. To the point: gaugebosons of abelian symmetries are permitted a mass bymeans of the St¨uckelberg mechanism—retaining gaugeinvariance, unitarity, and renormalizability.The question of a photon mass in QED is then purelyexperimental, as there is no theoretical prejudice againsta small

m

over

m

= 0.

1

However, we already knowthat QED is just the low-energy approximation of theGlashow–Weinberg–Salam model of electroweak inter-actions, so our above motivation for the photon massmight be in danger. Fortunately, the electroweak gaugegroup

SU

(2)

L

×

U

(1)

Y

still features an abelian factor—the hypercharge

U

(1)

Y

—that can be used in a St¨uck-elberg mechanism. The resulting mass for the hyper-charge gauge boson eventually generates again a mas-sive photon [6].

2

A detailed discussion of this procedureand its implications can be found in Ref.[4]. Since theSt¨uckelberg mechanism only works for abelian groups,the grand uniﬁcation of the Standard Model (SM) gauge

1

A small

m

m

.

2

U

(1)

Y

itself results from thebreakdown of

SU

(2)

R

×

U

(1)

B

−

L

: A St¨uckelberg mass of the

B

−

L

boson trickles down and makes the photon massive.

group

SU

(3)

C

×

SU

(2)

L

×

U

(1)

Y

into a simple nonabeliangroup like

SU

(5),

SO

(10), or

E

6

would necessarily re-sult in a truly massless photon[8]. Turning this around,the discovery of a massive photon would exclude a hugenumber of grand uniﬁed theories—and, obviously, be aspectacular ﬁnding in its own right.Let us now move on to the key point of this Letter:If one can constrain the mass of a photon, one shouldalso be able to constrain its lifetime. Massless photonsin QED are stable purely due to kinematical reasons,there are no additional quantum numbers that forbid adecay. Recalling the tight upper bound on the photonmass though, there are not many possible ﬁnal states—indeed, only one known particle could be even lighterthan the photon: the lightest neutrino

ν

1

. This is be-cause current neutrino-oscillation experiments can onlyﬁx the two mass-squared diﬀerences ∆

m

231

=

m

23

−

m

21

and ∆

m

221

=

m

22

−

m

21

of the three neutrinos, so the abso-lute mass scale is not known as of yet[9]. Kinematically,this opens up the possibility of a decay

γ

→

ν

1

ν

1

—should

m

1

< m/

2 hold.

3

This loop-suppressed process can becalculated in the SM (using e.g. a seesaw mechanism tomake neutrinos massive in a renormalizable way), and isof course ridiculously small[12]—being suppressed by thesmall photon mass, the heavy particles in the loop andmaybe the smallest neutrino mass, depending on the op-erator that induces this decay. We also note that one of the side eﬀects of a massive hypercharge boson—besidesa massive photon—are tiny electric charge shifts of theknown (chiral) elementary particles [4,6]. The neutrino
then picks up a charge

Q

ν

∝

em

2

/M

2

W

, which gives riseto a correspondinglysmall tree-level decay rate

γ

→

ν

1

ν

1

.Still, unmeasurable small SM rates never stopped anyone

3

The naive prototype model—augmenting the SM by only tworight-handed neutrinos (SM+2

ν

R

)—is problematic, as the ini-tially massless

ν

1

will unavoidably pick up a ﬁnite mass at looplevel [10], which can be too large for our purposes[11]. Fine-
tuned solutions aside, we can obtain a simple valid model byimposing a

B

−

L

symmetry on the SM+2

ν

R

, resulting in twoDirac neutrinos and one exactly massless Weyl neutrino.

2from looking for a signal, as it would be a perfect signfor new physics.Particles beyond the SM could not only increase therate

γ

→

ν

1

ν

1

, but also serve as ﬁnal states themselves,as some extensions of the SM feature additional (close to)massless states; examples include sterile neutrinos, hid-den photons, Goldstone bosons and axions. These weaklyinteracting sub-eV particles [13] are less constrained thanneutrinos, and photon decay might be an indirect eﬀectof these states. Although mainly of academic interest, wealso mention that a massive photon provides the possibil-ity of faster-than-light particles—and a decaying photoneven predicts them. The question of photon decay istherefore obviously relevant even if the lightest neutrinoturns out to be an inaccessibly heavy ﬁnal state.Following the above motivation, we set out to ﬁndlimits on the photon mass

m

and lifetime

τ

γ

as model-independent parameters. Most importantly, we do notcare about the daughter particles for now. Because of the small allowed values for

m

, all measurable photonsaround us are highly relativistic, making a decay hardto observe because of time dilation. Correspondingly,a good limit on

τ

γ

needs a large number of low-energyphotons from well-known far-away sources. Seeing as wehave access to very accurate measurements of the cosmicmicrowave background (CMB)—consisting of the oldestphotons in the visible Universe—we will take

m

and

τ

γ

as parameters that will modify the blackbody radiationlaw—given by the Planck spectrum—and ﬁt the CMBspectrum to obtain bounds on both parameters. Simi-lar analyses have been performed to obtain a limit onthe neutrino lifetime in the channels

ν

i

→

γν

j

[14,15].
In our case, we are, however, not looking for a spectralline on top of the CMB, but rather a diminished overallintensity and change of shape.Before delving into the details, let us present a back-of-the-envelope estimate: CMB photons with low ener-gies around meV have a lifetime

τ

=

γ

L

τ

γ

that is in-creased by a relativistic Lorentz factor

γ

L

=

E/m

≃

1meV

/

10

−

18

eV = 10

15

. This lifetime has to be com-pared to the age of the Universe

t

0

≃

13

.

8

×

10

9

yr (orthe corresponding comoving distance). Seeing as an im-proved accuracy

A

in the measurements will increase thebound, we can estimate

τ

γ

t

0

/γ

L

A

. We therefore ex-pect a lifetime constraint in the ballpark of years fromthe very precise CMB measurements (

A

≃

10

−

4

), whichwill be conﬁrmed by the more reﬁned analysis below.The photon mass changes the spectral energy densityof blackbody radiation to

ρ

(

T,E

)d

E

=1

π

2

E

3

d

E e

E/T

−

1

1

−

m

2

E

2

,

(1)because of the modiﬁed dispersion relation

p

2

=

E

2

−

m

2

,but it is unclear how to include the decay width. Theexpansion of the Universe also needs to be taken intoaccount, as the blackbody spectrum no longer stays inshape for

m

= 0. Let us therefore give a brief deriva-tion of the energy spectrum of massive unstable photonsduring cosmic expansion.Ignoring the width for a moment, the number densityof massive photons right after decoupling (at the time of last scattering

t

L

≃

n

0

(

p,t

)d

p

=

a

(

t

L

)

a

(

t

)

3

n

0

(

p

L

,t

L

)d

p

L

=4

πgp

2

d

p/

(2

π

)

3

exp

p

2

+

m

2

a

(

t

L

)

a

(

t

)

2

/T

−

1

,

(2)where

p

=

p

L

a

(

t

L

)

/a

(

t

) is the redshifted momentum,

T

the temperature at time

t

, and

g

the number of spinstates. We take

g

= 2, because only the transverse modesare excited before decoupling (this implicitly constrains

m

, as discussed below). The chemical potential of mass-less photons is zero, and since we assume that as ourinitial condition at

t

L

, we set it to zero in all our calcu-lations.Including the width, we can write down the diﬀerentialequation for the time evolution of the number densitydd

tn

(

p,t

) =dd

tn

0

(

p,t

)

−

Γ(

p

)

n

0

(

p,t

)

.

(3)The ﬁrst term on the right-hand side describes the num-ber density dilution due to the expansion of the Universe,while the second one is due to photon decay. The widthcan be obtained from the rest-frame width Γ

0

= 1

/τ

γ

by a Lorentz boost: Γ(

p

)

≃

Γ

0

m p

. We use the bound-ary condition

n

(

p,t

L

) =

n

0

(

p,t

L

) and obtain the numberdensity today

n

(

p,t

0

) =

n

0

(

p,t

0

)

−

Γ

0

t

0

t

L

m pn

0

(

p,t

)d

t.

(4)The integral can be evaluated to

t

0

t

L

m pn

0

(

p,t

)d

t

=

m p

L

n

0

(

p

L

,t

L

)

t

0

t

L

a

(

t

L

)

a

(

t

)d

t

=

m pn

0

(

p,t

0

)

d

L

,

(5)with the comoving distance of the surface of last scatter-ing

d

L

=

t

0

t

L

a

(

t

0

)

/a

(

t

)d

t

≃

47 billion lightyears. Overallwe have:

n

(

p,t

0

)

≃

n

0

(

p,t

0

)

1

−

Γ

0

m pd

L

≃

n

0

(

p,t

0

)exp

−

Γ

0

m pd

L

.

(6)The energy density relevant for the CMB spectrumis then obtained by multiplying

n

(

p,t

0

) with

E

=

p

2

+

m

2

:

ρ

(

E,T

)d

E

≃

1

π

2

E

3

d

E e

√

E

2

−

m

2

/T

−

1

1

−

m

2

E

2

×

exp

−

Γ

0

mE d

L

,

(7)

3

0.5 1.0 1.5 2.0 2.502.

10

13

4.

10

13

6.

10

13

8.

10

13

Energy

meV

Ρ

e V

3

Figure 1: CMB spectral distribution for Γ

0

t

0

m

= 0 (gray),Γ

0

t

0

m

= 2

×

10

−

5

eV (dashed red line) and Γ

0

t

0

m

= 10

−

4

eV(dotted blue line) using Eq. (7), as well as the COBE data(error bars multiplied by 1000 to be visible). In all cases themass is

m <

10

−

6

eV and has no visible eﬀect.

where we approximated

p

2

+

m

2

a

(

t

L

)

a

(

t

)

2

≃

E

2

−

m

2

(8)because

a

(

t

L

)

/a

(

t

0

)

≃

8

×

10

−

4

. Because of this approx-imation, the limit

ρ

(

E

→

m,T

) is nonzero, which is,however, of no importance for the CMB analysis in thisLetter.Equation (7) is the key equation of this Letter and willnow be used to set constraints on

m

and Γ

0

. For illus-trative purposes we show the spectrum for various valuesin Fig.1.As expected from time-dilation arguments,the low-energy part of the spectrum shows the strongestdeviations, which fortunately also features the smallesterror bars.Using the COBE (COsmic Background Explorer) dataset of the CMB[17] we can construct a simple

χ

2

4

The best ﬁt values areat

m

= 0 = Γ

0

, so we can only obtain exclusion ranges,shown in Fig.2.The limit on the photon mass is notcompetitive with other experiments—

m <

3

×

10

−

6

eV—but for the photon width we ﬁnd the only existing (andmodel-independent) bound

τ

γ

>

2

×

10

−

10

m

10

−

18

eV

t

0

(9)at 95% C.L. This would correspond to a photon lifetimeof only three years, should the photon mass be close toits current bound. Another useful form of the constraintis given by

m

10

−

18

eV

Γ

0

7

.

5

×

10

−

24

eV

<

1

.

(10)

4

Ground-based and balloon experiments probe the CMB downto energies

∼

10

−

6

eV, which typically have much larger errors.Additionally, there is an excess at low energies that is not un-derstood yet [18], so we do not include those data.

Excluded at 95

C.L.

7.5

7.0

6.5

6.0

5.5

5.0

4.5

10.0

9.5

9.0

8.5

8.0

7.5

7.0

6.5

Log

10

m

eV

L o g

1 0

m

e V

Τ

Γ

t

0

Figure 2: Constraints on photon mass

m

and lifetime

τ

γ

fromthe CMB spectrum.

For two-particle fermionic ﬁnal states

X

, the decay rate

γ

→

XX

from (eﬀective) interactions like

gXγ

µ

XA

µ

willbe of the form Γ

0

∼

g

2

m/

4

π

g

0

.

03

e

, which corresponds to a very largeeﬀective electric charge and is excluded by other exper-iments[19].

5

In particular, ﬁnal state neutrinos are farbetter constrained by their electric properties (see, e.g.,Ref. [15] for a recent review) to be relevant in photondecay. Our complementary and model-independent ap-proach should be interesting nonetheless, as it constitutesthe only direct constraint on the photon lifetime as of yet.Let us make a couple more comments to illustrate thatour analysis above is somewhat inconsistent. Our ap-proach basically assumed a vanishing or negligible num-ber density of St¨uckelberg scalars

φ

and daughter parti-cles

X

prior to photon decoupling. To ensure this,

m

andΓ

0

need to be small:

φ

has only the interaction

mA

µ

∂

µ

φ

,so for small mass

m

, it will not be in equilibrium with therest of the SM. The creation rate of

φ

via

eγ

↔

eφ

is pro-portional to

α

2

m

2

/T

, which has to be smaller than theexpansion rate of the Universe

H

(

T

)

≃

T

2

/M

Pl

—at leastbefore weak decoupling around

T

≃

1MeV—in order tonot put

φ

in thermal equilibrium at Big Bang nucleosyn-thesis (BBN). For

m <

10

−

3

eV, only the transversal po-larizations of the photon are excited, making it okay totreat the photon as massless before BBN. For the initialcondition of our blackbody calculation however, we needto ensure that only the two transverse degrees of freedomof the photon are excited at the surface of last scatteringat

T

≃

0

.

25eV. This requires

m <

5

×

10

−

13

eV, mak-ing our approach a little inconsistent, because at theselow masses the primordial plasma—consisting mainly of partly ionized hydrogen and helium—cannot be ignored.We will remark on this below.On to the daughter particles: the interaction rate of photons with their will-be daughter particles at tempera-ture

T

will be something like Γ

0

T/m

, as it should be ﬁnitein the limit

m

→

0. This rate has to be smaller than theexpansion rate of the Universe at BBN—unless the ﬁnal

5

It is of course trivial to reinterpret bounds on millicharged par-ticles[19] in terms of photon decay.