1

The stability of the QED vacuum in the temporal gauge

by

Dan Solomon

Rauland-Borg Corporation
3450 W. Oakton
Skokie, IL USA

Email: dan.solomon@rauland.com

August 7, 2004

PACS: 11.10.-z, 03.70.+k, 03.65.-w

 2

Abstract

The stability of the vacuum for QED in the temporal gauge will be examined. It

is generally assumed that the vacuum state is the quantum state with the lowest energy.

However, it will be shown that this is not the case for a system consisting of a fermion

field coupled to a quantized electromagnetic field in the temporal gauge. It will be shown

that for this situation there exist quantum states with less energy than the vacuum state.
 3

I. Introduction.

In this article the stability of the QED vacuum in the temporal gauge will be

examined. In the temporal gauge the gauge condition is given by the relationship A 0 = 0

[1,2,3,4] where A 0 is the scalar component of the electric potential. The advantage of

the temporal gauge is due to the simplicity of the commutation relationship between the

electromagnetic field quantities which are given below. In the coulomb gauge, for

instance, these are more complicated. Due to this fact the temporal gauge is particularly

useful in the treatments of QED which use the functional Schrödinger equation [3,4].

We will examine a field consisting of fermions coupled to a quantized

electromagnetic field. It is generally assumed that that there is a lower bound to the

energy of quantum states and that the quantum state with the lowest energy is the vacuum

state. We will show that that this is not the case for QED in the temporal gauge and that

there must exist quantum states with less energy than the vacuum state. Throughout this

discussion we will use = c = 1 . Also vectors are indicated by bold text.

We will work in the Schrödinger picture. In this case the field operators are time

independent and the time dependence of the quantum system is reflected in the state

vector Ω ( t ) which evolves in time according to the Schrödinger equation,

∂ Ω(t)
i ˆ Ω(t)
=H (1.1)
∂t

ˆ is [3],
where the Hamiltonian H

0,D + H 0,M − ∫ J ( x ) ⋅ A ( x ) dx
ˆ =H
H ˆ ˆ ˆ ˆ (1.2)

The quantities in the above expression are defined by,

 4

1  †
ˆ
H ˆ ( x ) , H 0,D ψ
0,D = ∫  ψ ˆ ( x )  dx ; H 0,D = −iα ⋅∇ + βm (1.3)
2 

ˆ
H
1 ˆ2 ˆ2
(
0,M = ∫ E + B dx ;
2
) Bˆ ( x ) = ∇ × A
ˆ (x) (1.4)

q †
Jˆ ( x ) =  ψ
ˆ ( x ) , αψ
ˆ ( x ) (1.5)
2 

In the above expressions m is the fermion mass, α and β are the usual 4x4 matrices, q is

ˆ
the electric charge, Ĥ 0,D is the Dirac Hamiltonian, H 0,M is the Hamiltonian for the

electromagnetic field, and Ĵ ( x ) is the current operator. The Schrödinger picture time

independent fermion field operators are ψ̂ ( x ) and ψ̂† ( x ) . The field operators for the

electromagnetic field are  ( x ) and Ê ( x ) . The electromagnetic field operators are real

ˆ † ( x) = A
so that A ˆ ( x ) and E
ˆ † ( x ) = Ê ( x ) .

The field operators obey the following relationships [2,3],

 ( )
ˆ i x , Eˆ j ( y )  = −iδ δ3 ( x − y ) ;  A
 ( )
A ˆ j ( y )  =  Eˆ i ( x ) , Eˆ j ( y )  = 0
ˆ i x ,A (1.6)
 ij   

and

{ψˆ †
a ( x ) , ψˆ b ( y )} = δabδ ( x − y ) ; {ψˆ †a ( x ) , ψˆ †b ( y )} = {ψˆ a ( x ) , ψˆ b ( y )} = 0 (1.7)

where “a” and “b” are spinor indices. In addition, all commutators between the

electromagnetic field operators and fermion field operators are zero, i.e.,

 ( x ) , ψ ( y )  = E ( x ) , ψ ( y )  =  A ( x ) , ψ ( y )  = E ( x ) , ψ ( y ) = 0
ˆ
A ˆ  ˆ ˆ  ˆ ˆ†  ˆ ˆ†  (1.8)

Define,

ˆ (x) = ∇ ⋅ E
G ˆ ( x ) − ρˆ ( x ) (1.9)
 5

where the current operator ρ̂ ( x ) is defined by,

q †
ρˆ ( x ) = ˆ ( x ) , ψˆ ( x ) 
ψ (1.10)
2 

The state vector Ω ( t ) must satisfy the gauss’s law constraint [3],

Ĝ ( x ) Ω ( t ) = 0 (1.11)

ˆ ( x )  = 0 therefore if (1.11) is valid at some initial time it

It can be shown [1] that  Ĥ, G 

will be valid for all time.

The energy of a normalized state Ω is given by,

E( Ω ) = Ω H
ˆ Ω (1.12)

where the normalization condition is Ω Ω = 1 . The commonly held assumption is that

there exists a state Ω vac , called the vacuum state, which is a state of minimum energy

so that all other states have an energy that is greater than the vacuum state, i.e.,

E ( Ω ) − E ( Ω vac ) > 0 for all Ω ≠ Ω vac (1.13)

We will show that the above statement cannot be true if there exists a normalized state

Ω1 which satisfies Gauss’s law and for which the divergence of the current expectation

value is non-zero, that is,

∇ ⋅ Ω1 J ( x ) Ω1 ≠ 0 in some region of space. (1.14)

Before proceeding we must ask the question “how do we know that a state Ω1 can be

found where the above condition holds?”. The answer is that if quantum mechanics is a

correct model of the real world then there must exist many states where the above
 6

condition holds because in the real world there are many examples where the divergence

of the current is non-zero over some region of space.

Now given some initial state new states can be defined by acting with operators

on the initial state [6]. With this in mind define the state,

ˆ
Ω 2 = e−iC Ω1 (1.15)

where the operator Cˆ is defined by,

C ˆ ( x ) ⋅∇χ ( x ) dx
ˆ = ∫E (1.16)

and where χ ( x ) is an arbitrary real valued function. Note that dual state is,

ˆ† ˆ
Ω2 = Ω1 e+ iC = Ω1 e+ iC (1.17)

ˆ † = Ĉ since Ê ( x ) and χ ( x ) are both real. From this we have that

where we have used C

Ω 2 Ω 2 = Ω1 Ω1 = 1 where we use the relationship,

ˆ ˆ
e +iC e −iC = 1 (1.18)

Now is Ω 2 a valid state, i.e., does it satisfy (1.11)? Based on the commutator

relationships (1.6) and (1.8) we see that the operator Ĉ commutes with both Ê ( x ) and

ˆ ( x ) Ω = e−iCˆ Ĝ ( x ) Ω = 0 so that Ω satisfies (1.11) since Ω

ρ̂ ( x ) . Therefore G 2 1 2 1

has been assumed to satisfy Ĝ ( x ) Ω1 = 0 .

Next we want to evaluate the energy of the state Ω 2 . To do this use (1.2) and

(1.12) to obtain,

E ( Ω2 )= ˆ
Ω2 H 0,D Ω 2 + Ω 2 H 0,M Ω 2 − Ω 2 ∫ J ( x ) ⋅ A ( x ) dx Ω 2
ˆ ˆ ˆ (1.19)
 7

Consider first the term Ω2 Ĥ 0,D Ω2 . To evaluate this use the fact that Ê ( x ) , and

thereby Cˆ , commutes with the fermion field operators ψ̂ ( x ) and ψˆ † ( x ) . Use this fact

along with (1.18) to obtain,

ˆ
Ω2 H ˆ
0,D Ω 2 = Ω1 H 0,D Ω1 (1.20)

Next consider the term Ω2 Ĥ 0,M Ω2 . From (1.6) we obtain,

 ( x ) , C  = −i∇χ ( x )
ˆ
A ˆ (1.21)

Use this result to obtain,

 Βˆ ( x ) , C
 ( x ) , C  = −i∇ × ∇χ ( x ) = 0
ˆ  = ∇ × A
ˆ ˆ (1.22)
 

Therefore Cˆ commutes with Ĥ 0,M so that,

ˆ
Ω2 H ˆ
0,M Ω 2 = Ω1 H 0,M Ω1 (1.23)

Now for last term in (1.19) use the fact that Cˆ commutes with Ĵ ( x ) to obtain,

Ω2 ∫ Jˆ ( x ) ⋅ A 2 1 ∫ (
ˆ ( x ) dx Ω = Ω Jˆ ( x ) ⋅ e+ iCˆ A
)
ˆ ( x ) e−iCˆ dx Ω
1 (1.24)

To evaluate the above expression further use the Baker-Campell-Hausdorff relationships

[7] which states that,

ˆ
ˆ e−Oˆ 1 = O
e+ O1 O ˆ ˆ  1  ˆ  ˆ ˆ 
ˆ + O
2 2  1 , O 2  +  O1 ,  O1 , O 2   + … (1.25)
2

ˆ and Ô are operators. Use this relationship along with (1.6) and (1.21) to
where O1 2

obtain,

ˆ ˆ ˆ
e+iC A ( x ) e−iC = Aˆ ( x ) − ∇χ ( x ) (1.26)

Use this result in (1.24) to obtain,

 8

Ω2 ∫ Jˆ ( x ) ⋅ A
ˆ ( x ) dx Ω = Ω Jˆ ( x ) ⋅ A
2 1 ∫ ( )
ˆ ( x ) − ∇χ ( x ) dx Ω
1 (1.27)

Use the above results in (1.19) to yield,

E ( Ω2 )= ˆ
Ω1 H( ˆ ˆ ˆ )
0,D + H 0,M − ∫ J ( x ) ⋅ A ( x ) dx Ω1 + ∫ Ω1 J ( x ) Ω1 ⋅∇χ ( x ) dx
ˆ

(1.28)

Next use (1.2) and (1.12) in the above and integrate the last term by parts, assuming

reasonable boundary conditions, to obtain,

E ( Ω2 ) = E ( Ω1 ) − ∫ χ ( x ) ∇ ⋅ Ω1 Jˆ ( x ) Ω1 dx (1.29)

Next subtract the energy of the vacuum state, E ( Ω vac ) , from both sides to obtain,

E ( Ω 2 ) − E ( Ω vac ) = ( E ( Ω1 ) − E ( Ωvac ) ) − ∫ χ ( x ) ∇ ⋅ Ω1 Jˆ ( x ) Ω1 dx (1.30)

(
Now in the above expression the quantities E ( Ω1 ) − E ( Ω vac ) ) and Ω1 Jˆ ( x ) Ω1

are independent of χ ( x ) . Recall that we have picked the quantum state Ω1 so that

∇ ⋅ Ω1 Jˆ ( x ) Ω1 is nonzero. Based on this we can always find a χ ( x ) so that

( E ( Ω2 ) − E ( Ωvac ) ) is a negative number. For example, let

χ ( x ) = λ∇ ⋅ Ω1 Jˆ ( x ) Ω1 where λ is a constant. Then (1.30) becomes,

) = ( E ( Ω1 ) − E ( Ωvac ) ) − λ ∫ ( ∇ ⋅ Ω1 Jˆ ( x ) Ω1 )
2
E ( Ω 2 ) − E ( Ω vac dx (1.31)

Now, since ∇ ⋅ Ω1 Jˆ ( x ) Ω1 is nonzero, the integral must be positive so that as λ → ∞

(
the quantity E ( Ω 2 ) − E ( Ω vac ) ) → −∞ . Therefore the energy of the state Ω 2 is less

than that of the vacuum state Ω vac by an arbitrarily large amount. Therefore there is no

lower bound to the energy of a QED quantum state in the temporal gauge.
 9

II. Interaction with Classical fields

In the previous section we have shown that if there exists a state Ω1 that

satisfies (1.14) then there exists a state Ω 2 whose energy is less that that of Ω1 by an

arbitrarily large amount. This suggests the possibility that it would be possible to extract

an arbitrarily large amount of energy from a quantum state through the interaction with

an external field. It will be shown in this section that this is theoretically possible.

In the absence of external interactions the energy of a quantum state remains

constant. In order to change the energy we must allow the field operators to interact with

external sources or fields. This is done by adding an interaction term to the Hamiltonian.

Let this term be,

ˆ = − ∫ S ( x, t ) ⋅ A
H ˆ ( x ) dx − Jˆ ( x ) ⋅ R ( x, t ) dx
∫ (2.1)
int

In the above expression S ( x, t ) is a classical field that interacts with the quantized

electromagnetic field and R ( x, t ) is a separate classical field that interacts with the

fermion current operator. It should not be assumed that the classical fields S ( x, t ) and

R ( x, t ) correspond to physical fields that actually exist. For the purposes of this

discussion these fields are fictitious. They have been introduced for the purposes of

perturbing the Hamiltonian in order to change the energy of some initial state. It will be

shown that for properly applied fields S ( x, t ) and R ( x, t ) an arbitrarily large amount of

energy can be extracted from some initial state. Therefore even though these fields do

not correspond to actual physical objects we believe that the following results are

mathematically interesting. The reason we pick these fields is because for particular
 10

values of the interaction we obtain an exact solution to the Schrödinger equation. This is

demonstrated in the following discussion.

When the interaction is included the Schrödinger equation becomes,

∂ Ω(t)
i ˆ Ω(t)
=H (2.2)
T
∂t

where,

ˆ =H
H ˆ +H
ˆ =H ˆ − ∫ S ( x, t ) ⋅ A
ˆ ( x ) dx − Jˆ ( x ) ⋅ R ( x, t ) dx
∫ (2.3)
T int

Now we will solve (2.2) for the following interaction,

R ( x, t ) = 0 for t < t1 ; R ( x, t ) = −g ( t ) ∇χ ( x ) for t1 ≤ t ≤ t 2 ; R ( x, t ) = 0 for t > t 2

and,

S ( x, t ) = 0 for t < t1 ; S ( x, t ) = g ( t ) ∇χ ( x ) for t1 ≤ t ≤ t 2 ; S ( x, t ) = 0 for t > t 2

where the double dots represent the second derivative with respect to time. In addition to

the above g ( t ) satisfies the following relationship at time t 2 ,

g ( t 2 ) = 0 and g ( t 2 ) = −1 (2.4)

According to the above expressions the interaction is turned on at time t1 and turned off

at time t > t 2 . During this time energy is exchanged between the quantized fermion-

electromagnetic field and the classical fields S ( x, t ) and R ( x, t ) . At some initial time

t i < t1 the state vector is given by Ω ( t i ) . We are interested in determining the state

vector Ω ( t f ) at some final time t f > t 2 . Based on the above remarks the state vector

Ω ( t ) satisfies,
 11

∂ Ω(t)
i ˆ Ω ( t ) for t < t
=H (2.5)
1
∂t

∂ Ω(t)  Ĥ − g ( t ) ∫ ∇χ ( x ) ⋅ Aˆ ( x ) dx 
i =  Ω ( t ) for t1 ≤ t ≤ t 2 (2.6)
∂t  + g ( t ) ∫ Jˆ ( x ) ⋅∇χ ( x ) dx 
 

∂ Ω(t)
i ˆ Ω ( t ) for t > t
=H (2.7)
2
∂t

Since these equations are first order differential equations the boundary conditions at t1

and t 2 are,

Ω ( t1 + ε ) = Ω ( t1 − ε ) and Ω ( t 2 + ε ) = Ω ( t 2 − ε ) (2.8)
ε→0 ε→0

The solution to (2.5) is,

ˆ ( t−t )
−iH
Ω(t) = e i
Ω ( t i ) for t < t1 (2.9)

It is shown in Appendix A that the solution to (2.6) is,

Ω ( t ) = e ( ) e ( ) e ( )e ( 1 ) Ω ( t1 ) for t 2 ≥ t ≥ t1
ˆ ig t D
ig t C ˆ iw t −iH
ˆ t −t
(2.10)

ˆ is defined by,
where the operator D

ˆ ( x ) ⋅∇χ ( x ) dx
D̂ = ∫ A (2.11)

and

 g ( t ′ )2
t 
w (t) = ∫  ( ) ( )∫
2 2
∫ ∇χ dx + g t ′ g t ′ ∇χ dx dt ′ (2.12)

t1  2 


The solution to (2.7) is,

ˆ ( t −t )
−iH
Ω ( tf ) = e f 2
Ω ( t 2 ) where t f > t 2 (2.13)

Use the boundary conditions (2.8) in the above to obtain,

 12

ˆ ( t − t ) ig( t )C
−iH ˆ ig( t )D
ˆ iw ( t ) −iH
ˆ ( t −t )
Ω ( tf ) = e f 2
e 2
e 2
e2
e 2 i
Ω ( ti ) (2.14)

Use (2.4) in the above to obtain,

ˆ ( t − t ) −iC
−iH ˆ iw ( t 2 )
Ω ( tf ) = e f 2
e e Ω0 ( t 2 ) (2.15)

where Ω0 ( t 2 ) is defined by,

ˆ ( t −t )
−iH
Ω0 ( t 2 ) = e 2 i
Ω ( ti ) (2.16)

Ω0 ( t 2 ) is the state vector that the initial state Ω ( t i ) would evolve into, by the time

t 2 , in the absence of the interactions. Use (2.14) in (1.12) to show that the energy of the

state Ω ( t f ) is,

( )
E Ω ( t f ) = Ω0 ( t 2 ) eiC He
ˆ
ˆ −iC Ω ( t )
0 2
ˆ
(2.17)

From the discussion leading up to equation (1.29) we obtain,

( )
E Ω ( t f ) = E ( Ω0 ( t 2 ) ) − ∫ χ ( x ) ∇ ⋅ Ω0 ( t 2 ) Jˆ ( x ) Ω0 ( t 2 ) dx (2.18)

Now, as before, assume that we select an initial state Ω ( t i ) so that

∇ ⋅ Ω0 ( t 2 ) Jˆ ( x ) Ω0 ( t 2 ) is non-zero. Recall that Ω0 ( t 2 ) is the state that Ω ( t i )

evolves into in the absence of interactions. Therefore E ( Ω0 ( t 2 ) ) = E ( Ω ( t i ) ) and

Ω0 ( t 2 ) is independent of χ ( x ) . The function χ ( x ) can take on any value without

affecting ∇ ⋅ Ω0 ( t 2 ) Jˆ ( x ) Ω0 ( t 2 ) . Let χ ( x ) = λ∇ ⋅ Ω0 ( t 2 ) Jˆ ( x ) Ω0 ( t 2 ) so that

(2.18) becomes,

( ) ( )
2
E Ω ( t f ) = E ( Ω ( t i ) ) − λ ∫ ∇ ⋅ Ω0 ( t 2 ) Jˆ ( x ) Ω0 ( t 2 ) dx (2.19)
 13

Define ∆E ext as the amount of energy extracted from the quantum state due to its

interaction with the classical fields. From the above equation,

( )
2
∆E ext = λ ∫ ∇ ⋅ Ω0 ( t 2 ) Jˆ ( x ) Ω0 ( t 2 ) dx (2.20)

Obviously as λ → ∞ then ∆E ext → ∞ .

In conclusion, it has been shown that there exist quantum states with less energy

than the vacuum state for QED in the temporal gauge. In fact there is no lower bound to

the energy of quantum states. If an initial state interacts with properly applied classical

fields then it is possible to extract an arbitrarily large amount of energy from the initial

state. The classical fields that were applied in this article are mathematical objects that

are not assumed to correspond to real physical objects. A possible next step in this

research would be to determine if these same results can be obtained through the

interaction of real existing fields.

Appendix A

It will be shown that (2.10) is the solution to (2.6). Take the time derivative of

(2.6) and multiply by “i” to obtain,

∂
i
∂t
(
Ω ( t ) = − gC )
ˆ + w Ω(t) − Ω + Ω
a b (A.1)

where,

ˆ
ˆ igDˆ eiw ( t )e−iHˆ ( t − t1 ) Ω ( t )
Ωa = eigC gDe (A.2)
1

and

Ω b = eigC eigD e ( ) He
ˆ −iHˆ ( t − t1 ) Ω ( t )
ˆ ˆ iw t
1 (A.3)

To evaluate (A.2) we will use the following relationships.

 14

ˆ
( ˆ ˆ
ˆ −igC eigC
ˆ = eigC De
eigC D ) ˆ
(A.4)

Use (1.6) and (1.25) to obtain,

ˆ
ˆ −igCˆ = D
eigC De ˆ ˆ ˆ
ˆ + ig C, 2
 D  = D − g ∫ ∇χ dx (A.5)

Use these results in (A.2) to yield,

(
ˆ − g ∫ ∇χ 2 dx Ω ( t )
Ωa = g D ) (A.6)

Next evaluate (A.3). Use (1.25) and the commutation relationships to obtain,

2
ˆ
ˆ −igDˆ = H
eigD He ˆ ˆ  g  ˆ  ˆ ˆ 
ˆ + ig  D,
 H  − 2  D,  D, H   (A.7)

where,

ˆ ˆ
 D, ˆ ˆ ˆ ˆ  ˆ
 H  = ∫  A, H  ⋅∇χdx = ∫  A, H 0,M  ⋅∇χdx = −iC (A.8)

and,

 D, ˆ   = −i  D,
ˆ ˆ 2
ˆ  D,
ˆ H  ˆ ˆ 
    C  = −i  ∫ A ⋅∇χdx, ∫ E ⋅∇χdx  = − ∫ ∇χ dx (A.9)

Therefore,

2
ˆ
ˆ −igDˆ = H
eigD He ˆ + g ∇χ 2 dx
ˆ + gC
2
∫ (A.10)

Use this in (A.3) to obtain,

ˆ  
2
ˆ + g ∇χ 2 dx  eigDˆ eiw ( t )e −iHˆ ( t − t1 ) Ω ( t )
ˆ + gC
Ω b = eigC  H
 2
∫  1 (A.11)
 

To evaluate this further use,

ˆ
ˆ −igCˆ = H
eigC He ˆ ˆ ˆ
ˆ + ig C, ˆ (
 H  = H + g ∫ J ⋅∇χdx ) (A.12)

where we have used,

 15

ˆ ˆ
 C, ˆ
 H  = −i ∫ J ⋅∇χdx (A.13)

and the fact that  C,

ˆ C,
ˆ Hˆ   = 0 . Therefore,
  

 g2 
Ωb ˆ (
ˆ ˆ)
=  H + g ∫ J ⋅∇χdx + gC +
 2
∫ ∇χ dx  Ω ( t )
2

(A.14)
 

Use this along with (A.11) and (A.6) in (A.1) to obtain,

 g2  
∂


ˆ
H + g ∫ (
ˆ
J ⋅∇χd x +) ˆ
gC +
2
∫
2 ˆ +w
∇χ dx  − gC
 ( )
i Ω(t) =    Ω(t) (A.15)
∂t  ˆ
( 2
−g D − g ∫ ∇χ dx ) 


Rearrange terms and do some simple algebra to obtain,

 H

(
ˆ + g ∫ Jˆ ⋅∇χdx − g ∫ Aˆ ⋅∇χdx ) 

∂
i Ω(t) =  g 2  Ω ( t ) (A.16)
∂t −  w − ∇χ
2
− ∇χ
2
2
∫ d x gg ∫ dx
 

  

Now let

g2 2 2
w= ∫ ∇χ dx + gg ∫ ∇χ dx (A.17)
2

to obtain

∂
i
∂t
Ω(t) = H (
ˆ + g ∫ Jˆ ⋅∇χdx − g ∫ A )
ˆ ⋅∇χdx Ω ( t ) (A.18)

which is (2.6) in the text. This completes the proof.

 16

References

1. G. Leibbrandt, Rev of Modern Phys, 59, 1067 (1987).

2. M. Creutz, Ann. Phys., 117, 471 (1979)

3. B. Hatfield. “Quantum Field Theory of Point Particles and Strings”, Addison-Wesley,

Reading, Massachusetts (1992).

4. C. Kiefer and A. Wiepf , Annals Phys., 236, 241 (1994).

5. W. Greiner, B. Muller, and T. Rafelski, “Quantum Electrodynamics of Strong Fields”,

Springer-Verlag, Berlin (1985).

6. S. Weinberg. The Quantum Theory of Fields. Vol. I. Cambridge University Press,

Cambridge, 1995.

7. W. Greiner and J. Reinhardt. “Field Quantization”, Springer-Verlag, Berlin (1996).