Casimir Effect Lectures on Vacuum Energy

The Casimir effect and the physical vacuum
Lectures given at the intensive course

“Advances in Strong-Field Electrodynamics”
G. Takács
BUTE Department of Theoretical Physics

and
MTA-BME “Momentum” Statistical Field Theory Research Group
Bolyai College, February 3-6, 2014

Outline
1 Introduction: QED and the Casimir effect

2 Realistic cases I: temperature and material dependence
3 Realistic cases II: geometry dependence
4 Comments on Casimir force and zero-point energy
5 Time dependent boundaries
6 Gravitational aspects
7 Some related topics
Recommended reading
1 K.A. Milton: The Casimir Effect, World Scientific, 2001.

2 J. Phys. A41 No. 16, Special Issue: Proceedings of
QFEXT07, 2008.
3 M. Bordag, U. Mohideen and V.M. Mostepanenko: New
Developments in the Casimir Effect, Phys.Rept. 353: 1-205,
2001. [quant-ph/0106045]
4 G.L. Klimchitskaya, U. Mohideen and V.M. Mostepanenko:
The Casimir force between real materials: experiment and
theory, Rev. Mod. Phys. 81:1827-1885, 2009.
[arXiv:0902.4022]
5 I. Brevik, J.S. Høye: Temperature Dependence of the Casimir
Force, Eur. J. Phys. 35: 015012, 2014. [arXiv:1312.5174]
Outline

QED
Casimir effect: discovery and simple derivation
A physical derivation: from momentum flow
Some other cases: massive scalar, EM field, fermions
The myth of a mysterious force between ships at sea
Outline

QED
Quantum theory of the electromagnetic field
Relativistic quantum electrodynamics (QED)
1948: Feynman, Schwinger, Tomonaga (Nobel prize: 1965)
1
L = − Fµν F µν + ψ̄ i γ µ (∂µ + ieAµ ) − m ψ Fµν = ∂µ Aν − ∂ν Aµ
4
Theory of the photon and the electron/positron field
(Origins: Dirac, Pauli, Weisskopf, Jordan; 1927-)
Experimental confirmation of QED
e2
α= fine structure constant
4πε0 h̄c
e− anomalous magnetic moment : 1/α = 137.035999710(96)

Nuclear recoil: 1/α = 137.03599878(91)
Hyperfine splitting in muonium: 1/α = 137.035994(18)
Lamb shift: 1/α = 137.0368(7)
Quantum Hall effect: 1/α = 137.0359979(32)
QED: „quod erat demonstrandum”

– the most precisely validated physical theory!
Outline

QED
The discovery of the Casimir effect
Attractive force between two perfect conductor plane in

vacuum (Casimir, 1948)
F h̄c π 2
=−
A 240a4
A macroscopic prediction of QED:
1 µ m distance: 8.169 × 10−3 Pa
Lamoreaux,
1996:
experimental
verification
within 5%
Naive derivation: from vacuum energy I
Scalar field with Dirichlet BC (units: h̄ = 1 = c)
φ (z = 0) = φ (z = a) = 0
r
1 ∞ d 2k
Z nπ 2
1
E = ∑ h̄ω = ∑ k 2+
2 2 n=1 (2π )2 a
This is divergent, but we can use dimensional regularization.
Using
Z ∞ Z π d/2
dt −n −zt 2
t e = Γ(−n)z n d d k e −tk =
0 t t
we can write
ddk
Z Z
∞ dt
1 2 2 2 2 1
E = ∑ d
t −1/2 e −t(k +n π /a )
2 n (2π ) 0 t Γ(−1/2)
Z ∞
1 1 dt −1/2−d/2 −tn2 π 2 /a2
=− √ ∑
4 π (4π )d/2 n 0 t
t e
Naive derivation: from vacuum energy II
Z
∞ dt
1 1 2 2 2
E =− √ d/2 ∑ t −1/2−d/2 e −tn π /a
4 π (4π ) n 0 t

1 1 π 1+d d +1
=− √
4 π (4π )d/2 a
Γ −
2 ∑ nd+1 Re d < −1
n
1 1 π 1+d d + 1
=− √ Γ − ζ (−d − 1) Re d < −2
4 π (4π )d/2 a 2
= ∞·0 for d positive odd integer
Physical: d ∈ N → analytic continuation is needed!
z
−z/2 1−z
Γ ζ (z)π =Γ ζ (1 − z)π −(1−z)/2
2 2

1 1 d π2 1
E = − d+2 d/2+1 d+1 Γ 1 + ζ (2 + d) → −
2 π a 2 d=3 1440 a3
∂E π2 1
Pressure: F = − =− EM: 2×
∂a 480 a4
Outline

QED
A physical derivation: from momentum flow I
Energy-momentum tensor
Tµν = ∂µ φ (x)∂ν φ (x) − ηµν L (x)
1
L (x) = ∂µ φ (x)∂ µ φ (x)
2
Left plate at z = 0: what we want is
F = hTzz iz>0 − hTzz iz<0
How do we compute? From QFT

T φ (x)φ (x ′ ) = −iG (x, x ′ )
Now
−∂ 2 G (x, x ′ ) = δ (x − x ′ )
d d k i~k·(~x−~x ′ ) d ω −i ω (t−t ′ )
Z Z
G (x, x ′ ) = e e g (z, z ′ |~k, ω )
(2π )d 2π
2
∂
− − λ g (z, z ′ ) = δ (z − z ′ )
2
λ 2 = ω2 − k2
∂z2
g (0, z ′ ) = g (a, z ′ ) = 0
A physical derivation: from momentum flow II
Internal contribution
1
gint (z, z ′ ) = − sin λ z< sin λ (z> − a)
λ sin λ a
⇓
int = 1 i
tzz ∂z ∂z ′ gint (z, z ′ )|z→z ′ =0 = λ cot λ a
2i 2
so
ddk
Z Z
dω i
Fint = d
λ cot λ a
(2π ) 2π 2
1
Z d
d k
Z
dζ
=− κ coth κ a divergent!
2 (2π )d 2π
p
with ω → i ζ λ → iκ = i k2 + ζ 2
Outer contribution
1
gout (z, z ′ ) = sin λ z< e ikz>
λ
out = 1 ∂ ∂ ′ g (z, z ′ )| 1
tzz z z out z→z ′ =0 = λ
2i 2
A physical derivation: from momentum flow III
d d~k κddκ
Z Z Z ∞
1 dζ κ
F =− κ (coth κ a − 1) = −Ωd+1
2 (2π )d 2π 0 (2π )d+1 e 2κ a − 1
Angular integral
Z Z d
d −~x 2 −ξ 2
d xe = dξ e = π d/2
2π d/2
Z
2 Γ(d/2)
= Ωd x d−1 e −x dx = Ωd ⇒ Ωd =
2 Γ[d/2]
Use
22z−1/2 y s−1
Z ∞
Γ(2z) = √ Γ(z)Γ(z + 1/2) Γ(s)ζ (s) = dy
2π 0 ey − 1
to get
Γ(1 + d/2)ζ (d + 2) ∂
F = −(d + 1)2−d−2 π −d/2−1 d+2
= − E (a)
a ∂ a
1 1 d
with E (a) = − d+2 d/2+1 d+1 Γ 1 + ζ (2 + d)
2 π a 2
Outline

QED
Massive scalar field
Massive scalar field
1 1
L (x) = ∂µ φ (x)∂ µ φ (x) − m2 φ 2
2 2
(∂µ ∂ µ + m2 )φ = 0
−(∂ 2 + m2 )G (x, x ′ ) = δ (x − x ′ )
d d k i~k·(~x−~x ′ )
Z Z
d ω −i ω (t−t ′ )
G (x, x ) = ′
e e g (z, z ′ |~k, ω )
(2π )d 2π
2
∂
− − λ 2
g (z, z ′ ) = δ (z − z ′ ) λ 2 = ω 2 − k 2 − m2
∂z2
g (0, z ′ ) = g (a, z ′ ) = 0
√
κddκ
Z ∞
κ 2 + m2
F = −Ωd+1 √
0 (2π )d+1 e 2a κ 2 +m2 −1
Massive scalar field II; EM field; fermions
1 1
Z ∞ √
2 2 2
E = dt t d log 1 − e 2 t +m a
ad+1 2d+1 π (d+1)/2 Γ( d+1
2 ) 0
ma d/2+1 1 ∞ 1
= −2
4π ∑ nd/2+1 Kd/2+1 (2n ma)
ad+1 n=1
r
π −x
Kn (x) ∼ e 1 + O(x −1 )
2x
so the effect decays exponentially with ma.
For the EM field between perfectly conducting planes one needs to
consider 2 independent polarizations: 2× the result for scalar with
Dirichlet BC.
For fermions
L = ψ̄γ µ ∂µ ψ
proper BC is that no conserved current flows out (bag model):
(1 +~n · ~γ )ψ |S = 0
Result for planar BC: 7/4 of the scalar force.

Outline

QED
A popular myth: mysterious force between ships at sea
Popular myth: ships attract at heavy

swell due to smaller wave pressure in
between.
The two situations were messed up:

Caussée claimed attraction in calm
sea (below), not in a swell (above)!
P. C. Caussée:
Nature, doi:10.1038/news060501-7 The Album of the Mariner
(1836)
Outline

Temperature effects
Material effects
Dependence on the fine structure constant
Outline

Temperature effects
Material effects
Temperature dependence
Matsubara formalism
1
Z = Tr e −β H β=
T
Z φ (~x,t2 )=φ2 (~x) R t2 R
dt d d xL
hφ2 (~x)|e −i (t2 −t1 )H |φ1 (~x)i = [d φ ] e i t1
φ (~x,t1 )=φ1 (~x )
⇓ τ = it LE = −L |t→−i τ
Z Rβ R
d τ d d xLE
Z= [d φ ] e − 0
φ (~x,β )=φ (~x ,0)
Due to PBC in τ , the Euclidean frequencies are quantized

2π n π (2n + 1)
ζ → ζn = fermions: APBC ζn =
β β
Z
dζ 1
2π
→
β ∑
n
s 2
d d~k
Z
1 κn 2π n
FT = −
β (2π )d ∑ e 2κn a − 1 κn = k2 +
β
n
High-temperature limit is classical
T → ∞: only n = 0 term

d d~k
Z
k d d +1
FT = −T = −T √ Γ ζ (d + 1)
(2π )d e 2ka − 1 (2 π a)d+1 2
Classical free energy
F = −T log Z = T ∑ log(1 − e −β |~p | )

~p
Z
d d~kπ ∞ √
−β ~k 2 +n2 π 2 /a2
= TV
(2π )d+1 a ∑ log 1 − e
n=−∞
d s

For T → ∞ expand exponential and use log ξ = ds ξ s=0
s
d d~k ∞
1 2s n2 π 2
Z
1 d
+ k2

F ∼ TV
2a ds (2π )d+1 ∑ β 2
n=−∞ 2 a
s=0

1 d +1
= −TV √ Γ ζ (d + 1)
(2 π a)d+1 2
High-temperature limit is classical
s
d d~k ∞
n2 π 2
Z
1 d 1 2s
∑ + k2

F ∼ TV β
2a ds (2π )d+1
n=−∞ 2 a2
s=0
Now do the momentum integral, perform the summation using
ζ -function and use
d 1
= −1
ds Γ(−s) s=0
So the free energy is

1 d +1
F = −TV √ Γ ζ (d + 1)
(2 π a)d+1 2
Now the pressure is
∂F ∂ 1 ∂
F =− V = Aa ⇒ =
∂V ∂V A ∂a
and this gives the same result

d d +1
FT = −T √ Γ ζ (d + 1)
(2 π a)d+1 2
Low-temperature limit
This is much more complicated: the result is not analytic in T .

The leading correction is
Γ(1 + d/2)ζ (d + 2)
F ≈ −(d + 1)2−d−2 π −d/2−1
ad+2
d+2 !
1 2a
× 1+
d +1 β
but there are also corrections of the form

...
a
e −...πβ /a
β
For details cf. Milton’s book.

Outline

Temperature effects
Material effects
Material dependence
Lifschitz theory for dielectrics

in planar geometry
Z∞ Z
1 1 1
F T =0
=− d ζ d 2~k2κ3 + ′
16π 3 0 d d
κ3 + κ1 κ3 + κ2 2κ3 a
TE: d = e −1 TM: d ′ = d(κ → κ /ε )
κ3 − κ1 κ3 − κ2
κ 2 = k 2 + εζ 2 (ζ = i ω )
Finite temperature:
2π n
ζ → ζn =
β
Z ∞ ∞
dζ 1
0 2π
→
β ∑′ (n = 0 with half weight)
n=0
Controversy over thermodynamics
One can rewrite the force (ε1 = ε2 = ε and ε3 = 1)
 
∞  A e −2qa Bn e −2qa 
Z ∞
T ′ n
FT = − q 2 dq 
 
π ∑ ζn
+ 
 1 − An e −2qa 1 − Bn e −2qa 
n=0 | {z } | {z }
TM mode TE mode
ζn = 2π nT
2
εp − s 2 p−s
An = Bn =
εp + s p+s
q
s 2 = ε − 1 + p2 p=
ζn
Limit of ideal metal: ε (i ζn ) → ∞ . However, in the zero-frequency

TE mode, the limits do not commute:
first ε → ∞ then ζ → 0 : B0 → 1
first ζ → 0 then ε → ∞ : B0 → 0
Reflectivity of metals
In terms of reflectivity
(1) (2) (1) (2)
An = rTM (i ζn ,~k⊥ )rTM (i ζn ,~k⊥ ) Bn = rTE (i ζn ,~k⊥ )rTE (i ζn ,~k⊥ )
Ideal metals ε = ∞
rTM (ω ,~k⊥ ) = 1 rTE (ω ,~k⊥ ) = −1
so An = Bn = 1 for all n.
For real metals ε < ∞
rTM (0,~k⊥ ) = 1 rTE (0,~k⊥ ) = 0
so B0 = 0, and stays so in the limit ε → ∞.
Casimir free energy per unit surface
 
∞ Z ∞
T 
∑′ qdq log 1 − An e −2qa + log 1 − Bn e −2qa 

F=
2π n=0 ζn | {z } | {z }
TM mode TE mode
∂F
FT = −
∂a
Ideal metal
Ideal metal: An = Bn = 1 for all n. Then

π2 1 4
FT = − 1 + (2aT ) aT ≪ 1
240a4 3
Casimir free energy per unit surface

 
π2  45ζ (3) 
(2aT )3 4

F =− 1 + −(2aT )  ζ (3) ≈ 1.2
720a3  | π 3
{z } 
requires special care
Entropy
∂F 3ζ (3) 2 4π 2 a 3
S =− = T − T aT ≪ 1
∂T 2π 45
This is fine: S(T → 0) = 0.
Modified ideal and Drude metals
Drude model
2
ωplasma
ε (i ζ ) = 1 +
ζ (ζ + ν )
very good model for many metals in optical experiments for
ζ < 2 · 1015 Hz
(e.g. gold: ωp = 9.03 eV , ν = 0.0345 eV ). Whenever
lim ζ 2 (ε (i ζ ) − 1) = 0
ζ →0
the zero-frequency TE mode does not contribute, i.e. B0 = 0:

T π2 1 4 T
F =− 4
1 + (2aT ) + ζ (3) aT ≪ 1
240a 3 8π a3

π2 45ζ (3) 3 4 T
F =− 3
1+ 3
(2aT ) − (2aT ) + ζ (3)
720a π 16π a2
3ζ (3) 2 4π 2 a 3 ζ (3)
S= T − T − !!! violates Nernst theorem
2π 45 16π a2
Proposed solutions
Mostepanenko, Geyer: abandon Drude model.

Low frequency ⇒ wave-length long, field constant inside plate⇒
cannot exist, leads to charge separation
However: why to give up a successful description of materials,

when there are other ways to avoid the problem.
E.g. if resistivity does not simply go to 0 at T = 0, i.e.
ν (T → 0) 6= 0
Additional physical effects:

1. Spatial dispersion
ε (ω ,~k)
Only ε (0, 0) would be infinite, but that is zero measure in ~k space.
Proposed solutions II
2. Anomalous skin effect: mean free path of electrons becomes

longer than field penetration depth near T = 0.
Again, no contribution from TE zero mode found.
3. Large separation: result for Casimir effect same as for large T ,

i.e. classical. It turns out TE modes do not contribute in this limit
and
ζ (3)T
F =− a→∞
8π a3
and this precisely agrees with the Drude prediction.
Future experiments will decide which scenario is valid (possibly

dependent on material).
Present experimental situation seems inconclusive to me.

Repulsive Casimir forces
One way: measure inside fluid,
suitably chosen dielectric constant
⇒ Lifshitz theory predicts repulsion.
J.N. Munday, F. Capasso, and V.A.
Parsegian:
Nature 457: 170–173, 2009.
Gold sphere - gold plate, in bromobenzene:

150 pN at 20 nm separation
Other way: coat surfaces of appropriate

(meta)materials
e.g. εleft = ∞ and µright = ∞
or negative refraction (cloaking)
(KK: only in limited freq. range!)
Analysis: K.A Milton et al, J. Phys. A45

374006, 2012. [arXiv:1202.6415]
Puzzle
What do you get if you lay an an invisibility cloak on the floor?
A flying carpet!
Outline

Temperature effects
Material effects
Where has α gone?
QED effect: would expect appearance of fine structure constant,
but it is nowhere to be found...
Actual metals: frequency-dependent dielectric constant and
conductivity. Drude model:
4π e 2 n
σ (ω ) = 0 ω 2 > ωplasma
2
=
m
For ω < ωplasma : penetration length (skin depth)
2πω |σ | ne 2
δ −2 = σ=
c2 m(γ0 − i ω )
Typically ω ≫ γ0 (damping)
c
δ≈√
2ωplasma
Frequencies dominating Casimir effect: c/d ⇒ perfect conductor
approximation means
c mc
≪ ωplasma α≫
d 4π h̄nd 2
Where has α gone? II
Typically: d < 0.5µ m. Copper:
mc 1
2
≈ 10−5 ≪ α ≈
4π h̄nd 137
Casimir force is α → ∞ limit!!!
α → 0 limit:
h̄2 1
aBohr = 2
∝
me α
and so n ∝ α 3 ⇒ ωplasma ∝ α 2 : for any fixed separation d,
Casimir effect goes away.
Also δ → ∞: separation d becomes ill-defined.
For more details

cf. R.L. Jaffe: The Casimir effect and Quantum Vacuum,
hep-th/0503158.
Radiative corrections: Schwinger’s method
Schwinger’s approach: consider the vacuum persistence amplitude
in the presence of sources and boundaries
Z R
e iW [K ] = h0|e −iHT |0i = DΦe i (S[Φ]+ K Φ)
Z
1
W [K ] = dxdx ′ K (x)G (x, x ′ )K (x ′ )
2
Effective field
Z
φ (x) = dx ′ G (x, x ′ )K (x ′ )
Z
K (x) = dx ′ G −1 (x, x ′ )φ (x ′ )
Altering the geometry (e.g. moving boundaries adiabatically)

Z
1
δ W [K ] = dxdx ′ K (x)δ G (x, x ′ )K (x ′ )
2
Z
1
=− dxdx ′ φ (x)δ G −1 (x, x ′ )φ (x ′ )
2
Casimir energy from response of Green’s function
Now
Z
1 R
dxK (x)φ (x) 1
e iW [K ] = e 2 i = ··· − dxdx ′ φ (x)K (x)K (x ′ )φ (x ′ )
2
i.e. changing boundaries is equivalent to a new two-particle source

iK (x)K (x ′ ) eff = −δ G −1 (x, x ′ )
Z Z
i ′ ′ i−1 ′
δW = dxdx G (x, x )δ G (x, x ) = − dxdx ′ δ G (x, x ′ )G −1 (x, x ′ )
2 2
Z
i i
=− dxdx ′ δ log G (x, x ′ ) = − δ Tr log G
2 2
so
i
E = lim (Tr log G − Tr log Gref )
T →∞ 2T
where Gref is the value at some reference state (e.g. with bodies
infinite distance apart).
Radiative correction for electromagnetic field
Use perturbative form of G with Π as polarization
G = G0 (1 + ΠG0 + . . . )
Result for parallel plates
E π2 απ 2
E = =− + + O(α 2 )
A 720a3 2560me a4
This is suppressed by
α me−1
a
and is inobservable in practice
me−1 = λCompton ≈ 2.43 · 10−12 m

1
α≈
137
Outline

Proximity force approximation
The method of Green’s dyadic
Lateral Casimir force
Casimir force between compact bodies
Outline

Novel measurement methods
Figure : Mohideen et al.
Figure : Bell Labs
AFM (Atomic Force Microscope),

sensibility in principle can be 10−17 N
Torsion balance (reached: 10−13 N)
(Capasso, Harvard) Si-plate: dielectric constant can be
modulated by laser
(U. Mohideen et al., UC Riverside)
Proximity force approximation; special geometries
Simplest way to account for geomery dependence:
Proximity Force Theorem
Sphere and plate, R ≫ d: every element of sphere is approximately
parallel to plate
Z π Z R
V (d) = 2π R sin θ Rd θ E (d + R(1 − cos θ )) = 2π R dxE (d + R − x)
0 −R
R Z
∂V dE (d + R − x)
F =− = 2π R dx
∂d −R dx
= 2π R (E (d) − E (d + 2R)) ≈ 2π RE (d)
Lamoreaux: 5% → Mohideen & Roy: 1%→ Bell Labs 0.5%

Need to include: finite conductivity corrections, surface roughness.
Other calculations: sphere - plate, cylinder - plate, concentric

spheres, coaxial cylinders.
(K.A.Milton: The Casimir effect, World Scientific, 2001.)
Outline

Green’s dyadic
Green’s dyadic: response of EM field to polarization
Z
Ei (x) = d 4 x ′ Γij (x, x ′ )Pj (x ′ )
Z
Hi (x) = d 4 x ′ Φij (x, x ′ )Pj (x ′ )
Static situation: frequency decomposition

Z
d ω −i ω (t−t ′)
Γij (x, x ′ ) = e Γij (~x,~x ′ ; ω )
2π
Z
d ω −i ω (t−t ′)
Φij (x, x ′ ) = e Φij (~x,~x ′ ; ω )
2π
Maxwell’s equations (ε0 = µ0 = 1)
~
~ = −∂H
rotE ⇒ εijk ∂j Γkl = i ω Φil
∂ t
∂ E ~ +P~
~ =
rotH ⇒ − εijk ∂j Φkl − i ω Γil = i ωδil δ (~x −~x ′ )
∂t
~ =0
divH ⇒ ∂i Φij = 0
Solving for Green’s dyadic
Redefining Γ:
Γ′il = Γil + δil δ (~x −~x ′ ) ⇒ ∂i Γ′ij = 0
Taking the rotation of Maxwell’s equations, we get

∇2 + ω 2 Γ′ij = −(∂i ∂j − δij ∇2 )δ (~x −~x ′ )

∇2 + ω 2 Φij = i ωεikj ∂k δ (~x −~x ′ )
This has to be solved with boundary conditions:

e.g. for a conducting boundary, tangential electric field vanishes on
the surface
εijk nj Γ′kl (~x,~x ′ ; ω )~x∈Σ = 0
Main advantage of method: explicit gauge invariance.
Computing the Casimir stress
The two-point functions of fields are

Ei (x)Ej (x ′ ) = −i Γij (x, x ′ )

1
Hi (x)Hj (x ′ ) = i 2 εikl ∂k εjmn ∂k Γmn (x, x ′ )
ω
(from εikl ∂k El (x) = i ω Hi (x) )
and the Maxwell stress tensor is
1 ~2 1 ~2
Tij = Ei Ej − δij E + Hi Hj − δij H
2 2
⇒ Casimir stress on the surface.
E.g. for a perfectly conducting sphere of radius a

1 ∂E
F = hTrr (r = a − 0)i − hTrr (r = a + 0)i = −
4π a2 ∂a
and the self-energy from Casimir stress is (Boyer)
0.092353
E= (h̄ = 1 = c)
2a
Outline

Lateral force
PFA: averaging over surface roughness. Condition: λc ≫ zA , zero

lateral force.
F. Chen and U. Mohideen, Phys. Rev A66: 032113, 2002.
Use of Casimir force in micromachines
Standard worry: Casimir force would

make nanobots stick.
Idea: exploit Casimir force to produce

motion.
T. Emig: Casimir force driven

ratchets
Phys. Rev. Lett. 98:160801, 2007
[cond-mat/0701641] A Casimir ratchet producing lateral
motion by vibrating separation
With typical parameters hv i ∼ mm/s
Other similar effect: Casimir torque (for asymmetric bodies)
Not yet observed!
Outline

Arbitrary compact bodies
Emig, Graham, Jaffe & Kardar ’2007

Z
i i
Z [C ] = Tr e− h̄ HC T = [DΦ]C e h̄ S[Φ]
Φ(~x, t + T ) = Φ(~x, t)
and Φ|C = 0
1 1 [
Tr e− h̄ HC Λ → e− h̄ E0 [C ]Λ + . . . C= Σα
Λ→∞
α
h̄ Z [C ] h̄
⇒ E [C ] = lim ln = ∑ (ωn − ωn,∞ )
T →−i ∞ |T | Z∞ n 2
Suppose C is time-independent: we can Fourier expand in time
Z Z
[DΦ]C → ∏ [D φn (~x)]C
n
Φ(x) = ∑ φn (~x)e 2π int/T
n
Fluctuating surface charges
So we get
Z R 2

i Th̄ d~x ( 2cTπ n ) |φn (~x)|2 −|∇φn (~x)|2
log Z [C ] = ∑ log [D φn (~x)]C e
n
Z ∞
cT
(T → ∞) = dk log ZC (k)
π 0 Z R
i
d 3~x (k 2 |φ (~x,k)|2 −|∇φ (~x ,k)|2 )
ZC (k) = [D φ (~x , k)]C e h̄ T
Now putting T = −i Λ/c, Wick rotating k = i κ

Z ∞
h̄c ZC (i κ )
E [C ] = − d κ log
π 0 Z∞ (i κ )
Z R
T
d 3~x (κ 2 |φ (~x,i κ )|2 +|∇φ (~x,i κ )|2 )
ZC (i κ ) = [D φ (~x , i κ )]C e− h̄
Implement Dirichlet BC with Lagrange multipliers:

Z Z Z R
T
d 3~x (ρα (~x)∗ φ (~x)+c.c.)
[D φ (~x )]C = [D φ (~x )] ∏ [D ρα (~x)D ρα∗ (~x)] ei h̄ Σα
α
| {z }
functional Dirac delta
Performing the Φ integral
So
Z Z
i
ZC (k) = [D φ (~x , k)]C ∏ [D ρα (~x)D ρα∗ (~x)] e h̄ T S̃(φ ,ρ )
α
Z
S̃(φ , ρ ) = d ~x k 2 |φ (~x, k)|2 − |∇φ (~x, k)|2
3
Z
+ d 3~x (ρα (~x)∗ φ (~x , k) + c.c.)
Σα
Idea: integrate out Φ from quadratic functional integral → classical

solution + fluctuations.
(∇2 + k 2 )φcl (~x, k) = 0 x ∈

/ Σα
∆φcl (~x, k) = 0 x ∈ Σα
∆∂n φcl (~x, k) = ρα (x) x ∈ Σα
Integrating out fluctuations
Z
φcl (~x) = ∑ d~x ′ G0 (~x,~x ′ , k)ρβ (~x ′ )
β Σβ
′
e ik|~x −~x | (1)
G0 (~x,~x ′ , k) = = ik ∑ jl (kr< )hl (kr> )Ylm (x̂ ′ )Ylm (x̂)∗
4π |~x −~x ′ | lm
Put now φ = φcl + δ φ

Z
i
ZC (k) = ∏ [D ρα (~x)D ρα∗ (~x)] e h̄ T S̃cl (ρ )
α
Z R
T
d 3~x (k 2 |δ φ (~x,k)|2 −|∇δ φ (~x,k)|2 )
× [D δ φ (~x , k)] ei h̄
| {z }
unconstrained
Z
fluctuations: cancel out with denominator
S̃cl (ρ ) = d 3~x (ρα (~x)∗ φ (~x, k) + c.c.)
Σα
Also note that φcl = ∑φβ , where φβ is sourced by ρβ .

β
Interaction terms
Z
" #
′ (1)
φcl (~x) = ∑ d~x ik ∑ jl (kr< )hl (kr> )Ylm (x̂ ′ )Ylm (x̂)∗ ρβ (~x ′ )
β Σβ lm
Interaction terms (α 6= β ): in terms of multipoles

Z
∗
Qβ ,lm = d~xβ jl (krβ )Ylm (x̂β )ρβ (~xβ )
Σβ
(1)
φβ (~xβ ) = ik ∑ Qβ ,lm hl (krβ )Ylm (x̂β )
lm
αβ (1)
φβ (~xα ) = ik ∑ Qβ ,lm ∑ Ulm,l ′ m′ hl ′ (krα )Yl ′ m′ (x̂α )
lm l ′ m′
αβ
Ulm,l ′ ,m′ : translation coefficients, depending on Σα and Σβ
Z
S̃αβ (ρ ) = d 3~x ρα (~x)∗ φβ (~x, k) + c.c.
Σα
1
αβ
= ik ∑ ∑ Qα∗ ,l ′ m′ Ul ′ m′ ,lm Qβ ,lm + c.c
2 lm l ′ m′
Self-interaction terms
Z
1
S̃αα (ρ ) = d 3~x (ρα (~x)∗ φα (~x, k) + c.c.)
2 Σα
Field inside Σα is regular Helmholtz solution, outside general
φin,α (~x) = ∑ φα ,lm jl (kr )Ylm (x̂) φout,α (~x) = φin,α (~x) + ∆φα (~x)
lm
!
(1)
∆φα (~x) = ∑ χα ,lm jl (kr )Ylm (x̂) + ∑ Tl ′αm′ lm (k)hl ′ (kr )Yl ′ m′ (kr )
lm l ′ m′
where Tl ′αm′ lm (k)

is from ∆φα (~x)|Σα = 0. But the out field is
regular at infinity ⇒ χα ,lm = −φα ,lm . So
(1)
φout,α (~x) = − ∑ φα ,lm ∑ Tl ′αm′ lm (k)hl ′ (kr )Yl ′ m′ (kr )
lm l ′ m′
Z
(1)
but it is also = d~x ′ G0 (~x,~x ′ )ρα (~x ′ ) = ik ∑ Qα ,l m hl ′ ′ ′ (kr )Yl ′ m′ (x̂)
Σα l ′ m′
so that ikQα ,l ′ m′ = ∑ φα ,lm Tl ′αm′ lm (k)
lm
φα ,lm = −ik ∑ [T α (k)]−1
l m lm Qα ,l m
′ ′ ′ ′
l ′ m′
Integrating over charge fluctuations
The final form for the self-interaction is
ik
S̃αα (ρ ) = −
2 ∑ Qα ,lm [T α (k)]−1
l m lm Qα ,l m + c.c.
′ ′ ′ ′
l ′ m′
and we are left with the functional integral

Z
ZC (k) = ∏ [D ρα (~x)D ρα∗ (~x)]
α
(
k
exp ∑ ∑ Qα∗ ,lm Tα−1 lm,l ′ m′ Qα ,l ′ m′
2 α lm,l ′ m′
)
k ∗

− ∑ ∑ Qα ,lm Uαβ lm,l ′ m′ Qα ,l ′ m′ − c.c.
2 α 6=β lm,l ′ m′
Z Z
∗
= Jacobian × ∏ dQα ,lm dQα ,lm exp {. . . }
α ,l.m
Jacobian is independent of functional integration variables (Q − ρ

relation linear) and drops out with denominator.
Casimir force: averaged interaction between fluctuating
charges
The end result is:
Z ∞
h̄c det MC (i κ )
EC = − d κ ln
π 0 det M∞ (i κ )
   
T−
1
1
U12 ··· U1N T−
1
1
0 ··· 0

 U21 T−
2
1
··· U2N 


 0 T−
2
1
··· 0 

M(k) =  .. .. .. ..  M∞ (k) =  .. .. .. .. 
 . . . .   . . . . 
UN1 UN2 ··· T−
N
1
0 0 ··· T−
N
1
For two bodies:

Z ∞
h̄c
E12 (C ) = − d κ Tr ln 1 − T1 U12 T2 U21
π 0
Note: this is entirely finite, convergent and physically meaningful.

General formula for planar situations
In one space dimension it is easy to derive the Casimir interaction
with other methods:
Z
h̄c ∞ h i
E12 (L) = − d κ log 1 − e−2κ L R1 (i κ )R2 (i κ )
π 0
where R1,2 (ω ) is the reflection coefficient of the mode ω on the
boundaries and
e−2κ L = e2i ω L = e2i |k|L , ω = |k|
So here:
T1 = R1 (ω ) T2 = ei ω LR2 (ω )
U12 = U21 = e2i ω L
which looks really sensible.
This also extends to planar situations
Z Z q
h̄c ∞ −2L κ 2 +~
k 2 +m2
E12 (L) = − d κ d~k⊥ log 1 − e ⊥ R1 (i κ ,~k⊥ )R2 (i κ ,~k⊥ )
π 0
(Bajnok, Palla & Takács, hep-th/0506089).
Outline

Zero-point energy
Why does the ZPE derivation work?
Casimir force and van der Waals interaction
Outline

Zero-point energy
Does the Casimir force originate from zero-point energy?
Mystery: a naive consideration of zero modes leads to a huge
vacuum energy density.
Quantum field
d d~k
Z
1 ~k)e −i ω (~k)t+i~k·~x + a† (~k)e +i ω (~k)t−i~k·~x
Φ(~x, t) = q a(
(2π )d/2 2ω (~k)
Z Z
1 1 2
H = d d~xT00 = d d~x (∂t Φ)2 + ~∇Φ
2 2
Z d ~ h i
=
d k
ω (~k) 1 a† (~k)a(~k) + a(~k)a† (~k)
(2π )d 2
d d~k d d~k 1 ~
Z Z
= d
ω (~k)a† (~k)a(~k) + ω (k)δ (0)
(2π ) (2π )d 2
With δ (0) = (2π )d V , d = 3 and a high energy cutoff Λ we get an

energy density
Z Λ
E0 1
= k 2 dk k ∝ Λ4
V 0 2
The naive vacuum energy density and the QFT Hamiltonian
QFT (Standard Model) valid at least up to Λ ∼ 1 TeV: EV0 ∼ 1047 mJ3
If Λ = MPlanck ∼ 1019 GeV : EV0 ∼ 10110 mJ3
How comes the Casimir force is such a small effect?
Crucial observation: quantum Hamiltonian is not uniquely
fixed!
E.g.: why is the standard mass point Hamiltonian
p̂ 2
Ĥ = + V (q̂)
2M
Explanation: this comes from correspondence principle
d i
Ô = [Ĥ, Ô] [q̂, p̂] = i h̄
dt h̄
d p̂ d
q̂ = p̂ = −V ′ (q̂)
dt M dt
h̄ → 0: q̂,p̂ commute ⇒ simultaneously diagonalizable ⇒
eigenvalues obey classical equations of motion.
The naive vacuum energy density and the QFT Hamiltonian
A perfectly good Hamiltonian for QFT is given by
Z Z
1 1 2
H = d d~xT00 = d d~x : (∂t Φ)2 + ~∇Φ :
2 2
Z d
d k ~ 1 †~ ~
Z
d d~k
= ω (~
k) : a (k)a(k) + a(~
k)a † ~
(k) := ω (~k)a† (~k)a(~k)
(2π )d 2 (2π )d
Moral: QFT does not predict vacuum energy density! Some other
interaction is needed ⇒ gravity.
Einstein’s “greatest mistake”:
1 8π G
Rµν − gµν R + λ gµν = Tµν
2 c4
(λ )ν c 4λ ν
Tµ =− g = E gµν
8π G µ
Cosmological constant: p = −E . Present concordance cosmology
(ΛCDM):
J
E ∼ 5.4 × 10−10 3
m
Outline

Zero-point energy
Why does the zero-point energy derivation work?
Energy of a point charge
e 1 ~2 e2
E= ⇒ E = ε 0 E =
4πε0 r 2 2 32π 2 ε0 r 4
Field energy:
e2
Z ∞
4π r 2 E dr =
r0 8πε0 r0
r0 = 0: divergent! Renormalization:
e2
mphys c 2 = m0 c 2 +
8πε0 r0
mphys : physical mass: the only observable.
Radius of the electron
Physical mass
e2
mphys c 2 = m0 c 2 +
8πε0 r0
m0 = 0: classical electron radius
r0 ∼ 10−15 m
Present experiments: r0 < 10−18 m
QED self-energy:
! !
2
λCompton
2 2 3α 1 2
m0 c = mphys c 1− log + + O(α )
4π r02 2
λCompton = 2.4263102175(33) × 10−12 m
r0 ∼ 10−18 m : 5% correction.
Theoretical limit: m0 > 0 → r0 > 10−136 m
Two point charges
Figure : Two point charges with distance d
~ =E
E ~1 + E
~2 → ~2
E = 21 ε0 E
Z
E(d) = d 3~xE
still divergent for r0 = 0

e1 e2 1 1
but: E(d1 ) − E(d2 ) = − finite!
4πε0 d1 d2
e1 e2
Interaction energy: Eint (d) =
4πε0 d
This works because
Z
WLorentz = − d 3~x∆E
Outline

Zero-point energy
Casimir effect and and van der Waals interaction
van der Waals force = interaction between fluctuating dipols
~
d1 · ~
d2 r 2 − 3(~d1 ·~r )(~d2 ·~r)
Hint =
r5
h0|Hint |mihm|Hint |0i
Veff = ∑ ∝ r −6
m6=0 E0 − E m
Original problem investigated by Casimir &

Polder: retardation effects on vdW force
Dielectric ball: Casimir self-stress ≡ vdW

forces
Casimir effect = relativistic vdW
Outline

Velocity dependence of Casimir force
Dynamical Casimir effect: particle creation
Outline

Moving boundary
Plates: K : x 3 = 0 and K ′ : x 3 = vt.
Solve for Dirichlet Green’s function (scalar field):

∂t2 − ∇2 G (x, x ′ ) = −δ (x − x ′)
G (x, x ′ ) = 0 x, x ′ ∈ K or K ′
Energy density
1 3 i 3
h0|T00 (x)|0i = ∑ h0|∂k Φ(x)∂k Φ(x)|0i = lim ∑ ∂k ∂k′ G (x, x ′ )
2 k=0 2 x ′ →x k=0
Solution in x 3 < 0: using method of images

> ′ i 1 1
G (x, x ) = 2 −
4π (x − x ′ )2 (x − SK x ′ )2
 
1 0 0 0
 0 1 0 0 
SK =  0 0 1 0 

0 0 0 −1
Moving boundary II
Solution for x3 > vt: use Lorentz transform to get into system of
K ′ , find image, transform back.

> ′ i 1 1
G (x, x ) = 2 −
4π (x − x ′ )2 (x − SK ′ x ′ )2
 
cosh s 0 0 − sinh s
 0 1 0 0  c −v
SK ′ = 
 0
 s = log
0 1 0  c +c
sinh s 0 0 − cosh s
Solution in between: infinitely many images

∞
i 1
G in (x, x ′ ) = 2 ∑ (−1)m ′ )2
4π m=−∞ (x − xm
′
x2m = (SK SK ′ )m x ′ ′
x2m−1 = SK (SK SK ′ )m x ′
′
x−2m = (SK ′ SK )m x ′ ′
x−2m−1 = SK (SK ′ SK )m x ′
Moving boundary III
Renormalization: eliminate vacuum contribution, which is the term
i
G0 =
4π 2 (x − x ′ )2
in all three domains.

Force per unit area:
Z
∞
d
F (a(t)) = − dx 3 h0|T00 (x)|0i a(t) = vt
d(vt) −∞
4
π2 8 v 2 v
=− 4
1+ +O
480a(t) 3 c c4
Electromagnetic case:
4
π2 10 2 v 2 v
F (a(t)) = − 4
1+ 2
− +O v ≪c
240a(t) π 3 c c4

3 (c 2 − v 2 )2 (c 2 − v 2 )4
=− 2 1 + + O v ≪c
8π a(t)4 16c 4 c8
Outline

Dynamical Casimir effect
Scalar field in 2d
∂t2 Φ − c 2∂x2 Φ = 0
Take an interval (0, a(t)), where a(t) = a0 for t < 0. The field is

(−) (+)
Φ(t, x) = ∑ χn (t, x)an + χn (t, x)an†
n
(±) 1 π nx cπn
χn (t ≤ 0, x) = √ e ±i ωn t sin ωn =
πn a 0 a0
r
(−) 1 a0 π kx
χn (t > 0, x) = √ ∑ Qnk (t) sin
πn k a(t) a(t)
∗
(+) (−)
χn (t > 0, x) = χn (t > 0, x)
Initial conditions
′
Qnk (0) = δnk Qnk (0) = −i ωn δnk
Equation of motion
Field equation gives
′′
Qnk (t) + ωk2 (t)Qnk (t)
" #
′
=∑ hkj 2ν (t)Qnj (t) + ν ′ (t)Qnj (t) − ν (t)2 ∑ hjl Qnl (t)
j l
cπk a′ (t)
ωk (t) = ν (t) =
a(t) a(t)
2kj
hkj = −hjk = (−1)k−j 2 j 6= k
j − k2
Suppose that a(T ) = a0 after some time T ⇒
t > T : Qnk (t) = αnk e −i ωk t + βnk e i ωk t

(−) (+)
Φ(t, x) = ∑ φn (t, x)bn + φn (t, x)bn†
n
(±) 1 π nx cπn
φn (t, x) = √ e ±i ωn t sin ωn =
πn a0 a0
Bogolyubov transform
r
k ∗ †

bk = ∑ αnk an + βnk an
n n

Unitarity: ∑ k |αnk |2 − |βnk |2 = n
k
In- and out-vacuum:
ak |0iin = 0 bk |0iout = 0
Number of created particles:
∞
1
nk = in h0|bk† bk |0iin = k ∑ n |βnk |2
n=1
∞
N= ∑ nk
k=1
Enhancing effect: parametric resonance. E.g.
a(t) = a0 [1 + ε sin(2ω1 t)]
cπ
ω1 =
a0
Particle creation
Solution is long, but result is that only odd modes are populated
and
n1 (t) ≈ τ 2 τ ≪1
4
n1 (t) ≈ 2 τ τ ≫1 τ = εω1 τ
π
1
E (t) = ω1 ∑ knk (t) = ω12 sinh2 (2τ )
k 4
Typical values for photons in cm cavity ω1 ∼ 60 GHz

maximum endurance for wall materials εmax ∼ 3 × 10−8
dn1 4
≈ 2 εmax ω1 ∼ 700 s−1
dt π
Total number created is typically thousands of photons per second.
Effects to take into account: finite wall reflectivity, detector
interaction.
Nonzero temperature: factor ∼ 103 at room temperature.
Experiments
MIR (Motion Induced Radiation, Padova) :(
Microwave line modulated by a SQUID: success!
C.M. Wilson et al., 2011

Nature 479: 376-379
Microwave line: 100 µ m
“Mirror motion”: ∼ nm
Outline

Local effects: the energy-momentum tensor
How does Casimir energy fall?
Cosmological constant from Casimir energy of extra dimensions
Non-Newtonian gravity
Outline

Energy density
Scalar field, Dirichlet plates: Green’s function of a given mode
1
gint (z, z ′ ) = − sin λ z< sin λ (z> − a)
λ sin λ a
⇓ λ 2 = ω2 − k2
1 2 2
Z d ω d 2k
hT00 i = (∂t Φ) + (∇Φ) = ht00 i
2 (2π )3
1
ht00 i = ω 2 + k 2 + ∂z ∂z ′ gint (z, z ′ )|z=z ′
2i
1 2
=− ω cos λ a − k 2 cos λ (2z − a)
2i λ sin λ a
Wick rotate ω → i ζ , λ → i κ and use polar coordinates ζ = κ cos θ ,
k = κ sin θ :
Z ∞ Z π /2
"
1 2 sin θ
hT00 i = − 2 κdκ dθκ cos2 θ cosh κ a
4π 0 0 sinh κ a
#
+ sin2 θ cosh κ (2z − a)
Energy density II
!
1 e 2κ z + e 2κ (a−z)
Z ∞
1 3 1
hT00 i = − 2 d κκ + +
6π 0 e 2κ a − 1 2 e 2κ a − 1
The second term is the vacuum constant, to be discarded. The

result is
hT00 i = u + g (z)
π2
u=−
1440a4
1 1
Z ∞ yz y (1−z/a)
3e +e
g (z) = − 2 dyy
6π 16a4 0 ey − 1
1
=− [ζ (4, z/a) + ζ (4, 1 − z/a)]
16π 2 a4
∞
1
ζ (s, z) = ∑ Hurwitz zeta
n=0 (n + a)s
Energy density III
g (z) diverges at z = 0, a. Fortunately 15
Z a h i 1 2κ a 10
dz e 2κ z + e 2κ (a−z) ==
log H-g Hz LL
e −1
0 κ 5
so, although its integral is divergent, 0
it is also a-independent and does not 0.0 0.2 0.4 0.6 0.8 1 .0
z a
contribute to the force.
Similar calculation gives Txx , Tyy , Tzz
   
1 0 0 0 1 0 0 0
 0 −1 0 0   0 −1 0 0 
hT µν i = u 
 0 0 −1 0  + g (z) 
  
0 0 −1 0 
0 0 0 3 0 0 0 0
Energy-momentum tensor
The energy-momentum tensor is not unique: instead of canonical
we may use the conformal one
1 µ ν
T̃ µν = T µν − ∂ ∂ − g µν ∂ 2 Φ2
6
for which
µ
T̃µ = 0
Then
 
1 0 0 0
 0 −1 0 0  π2
hT µν i = u  
 0 0 −1 0  u=−
1440a4
0 0 0 3
Casimir pressure and energy density
p = −3u
e=u
Outline

Equivalence principle
Binding energy: mass defect

Chemical bonds: ∆m/m = 10−9
⇓
The equivalence principle is valid
for EM energy with at least 10−3
precision!
Between parallel plates
 
1
 −1 
hT µν i = u   θ (z)θ (a − z)
 −1 
3
π 2 h̄c z= 0 z= a
u=−
1440a4
Remarks:
1. Volume divergence („ZPE”) trivially eliminated.
d 3~k ~
Z
h̄
u0 = c k
2 (2π )3
2. Surface divergence ∝ z −4 ⇒ renormalizing mass of plates.

Equivalence principle holds!
Gravitation energy in weak field limit:
Z
Eg = − d 3~x hµν (~x)T µν (~x)
Problem: Eg is not gauge invariant!

Z
hµν → hµν + ∂µ ξν + ∂ν ξµ : ∆Eg = 2 d 3~x ξµ ∂ν T µν
Why? ∂ν T µν 6= 0: there is a force on the plates!

Solution: Use locally inertial coordinates (K.A. Milton et al.):
Fermi coordinates: gij quadratic in distance from origin. Locally
h00 = −gz h0i = hij = 0

Eg = gz0 uAa + const = gz0 ECasimir + const
which is just right!

A full analysis: K.A. Milton et al: How does Casimir energy fall?
IV, arXiv:1401.0784
Outline

Compact extra dimensions
Compact extra dimensions: Kaluza-Klein theory, later resurrected
by string theory.
Space-time: M = M4 × K
Λ µν
hT µν i = −u(a)g µν = − g
8π G
Case of a sphere: K = S N
Casimir energy of free massless scalar, for odd N
Z ∞
1 2π
u(a) = − 2 4
Re dy [y 2 − i (N − 1)y 2 ]D(iy ) 2π y
64π a 0 e −1
(2l + N − 1)(l + N − 2)!
Dl =
(N − 1)!l !
3ζ (5) 5 × 10−5
N = 1 : u(a) = − ≈ −
64π 6 a4 a4
For even N u(a) is logarithmically divergent; cutoff is necessary:
1 h a i
u(a) = αN log + const
a4 bZ
1 ∞ dt
αN = Im 2π t − 1
[(N − 1)it − t 2 ]2 D(it)
16π 2 0 e
b: frequency cut-off, presumably Planck scale. For large extra
dimensions a/b ∼ 1016 : logarithmic term sufficient for estimate.
Estimate for size of extra dimensions
Cosmological constant (ΛCDM concordance cosmology)
GeV
Λ ∼ ρc ∼ 10−5
cm3
Maximum value for coefficient
10−3
u(a) ∼
a4
Restoring units using h̄c = 2 × 10−14 GeV cm we find
cm3
a4 ∼ 102 h̄c ∼ 10−12 cm4
GeV
a ∼ 10 µ m
Such a compact dimension would lead to non-Newtonian gravity on

a submm scale.
Outline

Non-Newtonian gravity experiments
E.g. searching for a correction of the form
e−r /λ
V (r ) = α
r
Presently: extra dimensions with size around 100 µ m are ruled out.
Outline

Vacuum birefringence
Axions
Sonoluminescence
Outline

Axions
Sonoluminescence
2
1 ~ 2 ~ 2 ξ ~ 2 ~ 2
2
~ ~
Leffective = E −B + E −B +7 E ·B
2 2
h̄e 4
ξ= 4 7
∆n ∼ 4 × 10−24 (Bext /1 Tesla)2
45π m c
PVLAS (Polarizzazione del Vuoto con LASer, INFN, Padova)
G. Zavattini et al, QFEXT11, arXiv:1201.2309
Factor of 104 needed to reach sensitivity to QED: no signal yet!

→ can still look for axion signal
Outline

Axions
Sonoluminescence
Axions
1 1 1 ~ 2 ~ 2 ~ ·B
~
L = ∂µ a∂ µ a − ma2 a2 + ε E − B − ga aE
2 2 2
Axions induce vacuum birefringence
PVLAS had a signal, turned out to be detector effect on reanalysis
(2008 exclusion plot)

Shining light through walls
It is possible to shine light through walls using e.g. axions.
Standard modell contributions
4
L 4

Graviton conversion very weak: P(γ → g → γ ) ∼ 10−83 B
1T 1m
Neutrino conversion is even weaker:

Shining light through walls: beyond the standard model
(a) Axions (b) Hidden sector γ (c) Hidden γ enhanced by MCP

(MCP: milli-charged particles)
ALP experiment (DESY), using HERA magnet

So far no signal...
J. Redondo and A. Ringwald: Light shining through walls,
arXiv:1011.3741.
Outline

Axions
Sonoluminescence
Sonoluminescence
Collapsing bubble emits flash of light
a ∼ 10−3 cm, overpressure ∼ 1 atm,
f ∼ 104 Hz, Etot ∼ 10 MeV
Schwinger: divergent bulk contribution

4π a3 d 3~k 1
Z
1
Ebulk = k 1−
3 (2π )3 2 n
Schwinger estimate (adiabatic approximation):

a3 K 4 1
Ebulk ∼ 1− √
12π ε
Putting in a ∼ 4 × 10−3 cm, cutoff K ∼ 2 × 105 cm−1 (UV),

√
ε ∼ 4/3:
Ec ∼ 13 MeV
Casimir calculations
Casimir energy for dielectric sphere (renormalized by bulk
subtraction, equal to vdW!)
23
E= (ε − 1)2 (|ε − 1| ≪ 1)
1536π a
Experiment: ai ∼ 4 × 10−3 cm to af ∼ 4 × 10−4 cm
∆E ∼ −10−4 eV
Dynamical Casimir effect? Radiated energy spectrum: T ∼ 104 K.
Simple estimate using results from Unruh effect:
h̄A a
Unruh temperature: T = Acceleration: A ∼
2π c τ2
we get τ ∼ 10−15 s which is way too short!
Experiment: collapse time scale 10−4 s, emission 10−11 s.
Best present explanation: towards end of bubble collapse
T ∼ 104 K, ionized noble gas radiates.
K.A. Milton, arXiv:hep-th/0009173

Casimir Effect Lectures on Vacuum Energy

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Casimir Effect Lectures on Vacuum Energy

Uploaded by

Copyright:

Available Formats

The Casimir effect and the physical vacuum

Lectures given at the intensive course

BUTE Department of Theoretical Physics

Bolyai College, February 3-6, 2014

1 Introduction: QED and the Casimir effect

1 K.A. Milton: The Casimir Effect, World Scientific, 2001.

1 Introduction: QED and the Casimir effect

1 Introduction: QED and the Casimir effect

Relativistic quantum electrodynamics (QED)

1948: Feynman, Schwinger, Tomonaga (Nobel prize: 1965)

e− anomalous magnetic moment : 1/α = 137.035999710(96)

QED: „quod erat demonstrandum”

1 Introduction: QED and the Casimir effect

Attractive force between two perfect conductor plane in

1 Introduction: QED and the Casimir effect

1 Introduction: QED and the Casimir effect

Result for planar BC: 7/4 of the scalar force.

1 Introduction: QED and the Casimir effect

Popular myth: ships attract at heavy

The two situations were messed up:

1 Introduction: QED and the Casimir effect

1 Introduction: QED and the Casimir effect

Due to PBC in τ , the Euclidean frequencies are quantized

Classical free energy

F = −T log Z = T ∑ log(1 − e −β |~p | )

This is much more complicated: the result is not analytic in T .

but there are also corrections of the form

For details cf. Milton’s book.

1 Introduction: QED and the Casimir effect

Lifschitz theory for dielectrics

Limit of ideal metal: ε (i ζn ) → ∞ . However, in the zero-frequency

Casimir free energy per unit surface

the zero-frequency TE mode does not contribute, i.e. B0 = 0:

Mostepanenko, Geyer: abandon Drude model.

However: why to give up a successful description of materials,

Additional physical effects:

2. Anomalous skin effect: mean free path of electrons becomes

3. Large separation: result for Casimir effect same as for large T ,

Future experiments will decide which scenario is valid (possibly

Present experimental situation seems inconclusive to me.

Gold sphere - gold plate, in bromobenzene:

Other way: coat surfaces of appropriate

Analysis: K.A Milton et al, J. Phys. A45

What do you get if you lay an an invisibility cloak on the floor?

1 Introduction: QED and the Casimir effect

Also δ → ∞: separation d becomes ill-defined.

For more details

Altering the geometry (e.g. moving boundaries adiabatically)

Result for parallel plates

me−1 = λCompton ≈ 2.43 · 10−12 m

1 Introduction: QED and the Casimir effect

1 Introduction: QED and the Casimir effect

Figure : Mohideen et al.

Figure : Bell Labs

AFM (Atomic Force Microscope),

Lamoreaux: 5% → Mohideen & Roy: 1%→ Bell Labs 0.5%

Other calculations: sphere - plate, cylinder - plate, concentric

1 Introduction: QED and the Casimir effect

Static situation: frequency decomposition

Γ′il = Γil + δil δ (~x −~x ′ ) ⇒ ∂i Γ′ij = 0

Taking the rotation of Maxwell’s equations, we get

This has to be solved with boundary conditions: