VACUUM RADIATION, ENTROPY AND THE ARROW OF TIME

JEAN E. BURNS
Consciousness Research
1525 - 153rd Avenue
San Leandro, CA 94578

Abstract

The root mean square perturbations on particles produced by vacuum radiation must be
limited by the uncertainty principle, i.e., < δ x2 >1/2 < δ p x >1/2 = h / 2 , where
2

< δ x > and < δ p x > are the root mean square values of drift in spatial and
2 1/2 2 1/2

momentum coordinates. The value < δ x 2 >1/2 = ( h t /m ) , where m is the mass of

1/2

the particle, can be obtained both from classical SED calculation and the stochastic
interpretation of quantum mechanics. Substituting the latter result into the uncertainty
principle yields a fractional change in momentum coordinate, < δ p x >1/2 / p , where p is
2

the total momentum, equal to 2-3/2 ( h /Et ) , where E is the kinetic energy. It is shown
1/2

that when an initial change < δ p x > is amplified by the lever arm of a molecular
2 1/2

interaction, < δ p x >1/2 / p > 1 in only a few collision times. Therefore the momentum
2

distribution of a collection of interacting particles is randomized in that time, and the action
of vacuum radiation on matter can account for entropy increase in thermodynamic systems.
The interaction of vacuum radiation with matter is time-reversible. Therefore
whether entropy increase in thermodynamic systems is ultimately associated with an arrow
of time depends on whether vacuum photons are created in a time-reversible or irreversible
process. Either scenario appears to be consistent with quantum mechanics.

1. Introduction

In this paper we will see that entropy increase in thermodynamic systems can be accounted
for by vacuum radiation, and then discuss the relationship between vacuum radiation and
the arrow of time.
The problem in accounting for entropy increase has always been that dynamical
interactions which occur at the molecular level are time -reversible, but thermodynamic
processes associated with entropy increase, such as diffusion and heat flow, only proceed in
one direction as time increases. In the past it was often held that entropy increase is only a
491
R.L. Amoroso et al (eds.), Gravitation and Cosmology: From the Hubble Radius to the Planck Scale, 491-498.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands.
492 J. E. BURNS
macroscopic phenomenon, which somehow appears when a coarse-grain average is taken
of microscopic processes. But no averaging of time-reversible processes has ever been
shown to account for phenomena which are not time-reversible.[1]
Nowadays entropy increase is often viewed as coming from effects of the
environment, such as walls of a container or thermal radiation, not taken into account in the
description of a system. Unruh and Zurek [2] have given examples in which entropy
increase is produced in this way.
However, the second law of thermodynamics specifies that entropy increase must
also occur in an isolated system. So if we are to hold that entropy increase is produced by a
physical process at the microscopic level, we must also understand how it can be produced
in this way in an isolated system.
Any explanation must satisfy the basic assumptions of statistical mechanics.
Classical statistical mechanics has only one assumption:

At equilibrium it is equally probable that the system will be in any (classical)

state which satisfies the thermodynamic constraints.

Quantum statistical mechanics has two basic assumptions. The first is essentially the same
as for classical, except that states are now counted quantum mechanically. Thus:

At equilibrium it is equally probable that the system will be in any (quantum)

state which satisfies the thermodynamic constraints.

The second assumption of quantum statistical mechanics is:

At equilibrium the relative phases of the eigenvectors describing the system

are random.

Once these fundamental assumptions are made, one can then define entropy as
klog(number of states), where k is Boltzmann's constant. It is always also assumed that the
number of molecules, and therefore the number of states, is extremely large. One can then
develop the physics of the microcanonical ensemble in the usual way, by requiring that
different parts of an isolated sys tem be in equilibrium with each other at temperature T. By
placing the system in equilibrium with a heat bath one can then derive the physics of the
canonical ensemble, and so forth.[3]
In order to talk about entropy, we must specify the context in which we refer to the
ensemble of all possible states. In the coarse-grain view we would use an ensemble of
states with all possible initial conditions, and then argue that because the number of states is
very large, the only states we are apt to see are the most probable ones (and not ones in
which all molecules are clustered in a corner of a box, for instance). Thus equilibrium
merely refers to the most probable state in a large collection of systems. In the view in
which entropy is produced at the microscopic level, we start with a single system which has
 VACUUM RADIATION, ENTROPY AND TIME 493
specified initial conditions (classical or quantum mechanical) and look for a process which
produces many random perturbations and by this means places the individual system into
its most probable state.
In order to inquire about an isolated system, let us consider the system to be
comprised of not only the interacting molecules under consideration, but also the walls of
their container, any heat bath surrounding them, and all the thermal radiation which might
affect them. It would seem that we have taken into account all interactions which could
possibly affect the system. What then could serve as an "environment" which would
account for entropy increase?
Let us ask if an interaction could take place within the limits of the uncertainty
principle which would affect molecules randomly? If this interaction could randomize the
momentum of each molecule and (when quantum mechanical description is needed)
randomize the quantum phases of the eigenvectors describing the system, this process
would then account for entropy increase. Yet the interaction itself could not be detected in
measurements of the system.
Vacuum radiation acts at the limits of the uncertainty principle, and clearly it would
perturb molecules in a random way. But are these effects large enough? A thermodynamic
system goes to equilibrium in a few molecular collision times.[3] So in order to account for
entropy increase, vacuum radiation would have to randomize the momentum of a system
and the quantum phases of its eigenvectors in that short time. Let us first take up the
question of momentum.

2. Randomization Of Momentum By Vacuum Radiation

2.1. DRIFT IN SPATIAL COORDINATE

It has been shown by Rueda [4] in a classical stochastic electrodynamics (SED) calculation
that the coordinate drift produced on a free particle by vacuum radiation can be described
by diffusion constant D = h /2m , where m is the mass of the particle. A quantum
mechanical calculation of this effect of vacuum radiation has not been done. However,
when only energy and momentum transfer are involved and not anything specifically
quantum about the nature of the radiation involved, it is reasonable that an SED calculation
will give the same result as a quantum mechanical one.[5,6]
Rueda showed that vacuum radiation moves electrons in a random walk at
relativistic speeds and that this motion accounts for nearly all of their mass, with step length
varying from the Compton wavelength to the de Broglie wavelength. The radiation acts on
hadrons at the quark level and moves the hadrons at sub-relativistic velocity.[4]
We note that the stochastic interpretation [7] of the Schrödinger equation, which has
no direct connection to vacuum radiation, but attributes a quantum brownian motion to
particles, yields the same diffusion constant. In a similar vein, the stochastic action of
particles, with the same range of step lengths as above, can be derived directly from the
494 J. E. BURNS
uncertainty principle in the following way. Suppose that we have an ensemble of particles,
labeled 1, 2, .... Each is subject to a series of position measurements at equal time intervals.
Particle 1 is measured with resolution ∆ x 1 , particle 2 with resolution ∆ x 2 , and so forth,
with ∆ x 1 > ∆ x 2 > .... According to the uncertainty principle, as measurement resolution
becomes increasingly fine, particle momentum is increasingly more uncertain, and the path
is more erratic. Using this point of view, a particle can be described as following a
continuous, non-differentiable path of fractal dimension two, which corresponds to
brownian motion.[8] Further analysis shows that the step lengths vary from the Compton
wavelength to the de Broglie wavelength.[9]
The above diffusion constant yields a root mean square spatial drift < δ x 2 >1/2 =
(2Dt )1/2 [10], so

 ht 
1/2

<δ x >2 1/2

=  .
 m 
(1)

The above result can be confirmed experimentally using a tightly collimated beam
of low energy electrons. For instance, if a beam of 100 ev electrons has vy/vx = 10 -5 (where
x is the forward direction of travel), the spread in beam width due to the above process will
be larger than the spread due to diffraction in the first 19.5 cm of travel.[11] This
experiment has not presently been done, however.

2.2. RANDOMIZATION OF MOMENTUM

Vacuum radiation acts at the limits of the uncertainty principle, so we write

< δ x2 >1/2 < δ p x >1/2 = h / 2 , where < δ p x > 1/2 is the root mean square shift in
2 2

momentum component of the particle produced by vacuum radiation. It is then easily

found that

< δ p 2x > 1/2 1  h 

1/2

= 3/2  ,
 E t 
(2)
p 2

where p is the total momentum of the particle and E = p2/2m is the energy. We see that
< δ p x >1/2 is proportional to t-1/2 , so momentum is conserved as time becomes large.
2

Perturbations in momentum of a particle will change its original value, and when
< δ p x >1/2 / p > 1, momentum has been completely randomized. We wish to know
2

how long this will take. In order to have a concrete example, let us start with air at standard
conditions. At the end of one collision time (i.e., the time to travel a mean free path),
< δ p x >1/2 / p = 1.186 x 10-3.[11] However, any change in momentum is multiplied by
2

a lever arm A = λ/r , where λ is the mean free path and r the molecular radius, during the
 VACUUM RADIATION, ENTROPY AND TIME 495
next collision.[11] In air at standard conditions A = 1.005 x 104.[11] Therefore, the
momentum distribution of the molecules has been randomized in two collision times.
The product A < δ p x >1/2 / p is proportional to (kT ) / ( σ P1/2 ) .[11]
2 1/4

Therefore, momentum is randomized in a few collision times for all gases except those at
very high pressures (> 100 atm, or higher if the temperature is substantially more than
300 K). In solids and liquids many particles interact simultaneously, so it is reasonable to
suppose that momentum will randomize within a few collision times in these also.[11]

3. Randomization Of The Phases Of The Eigenvectors

In order to fulfill the second fundamental assumption of quantum statistical mechanics, it is

necessary to show that vacuum radiation can randomize the relative phases of the
eigenvectors describing the system within a few collision times. We make no calculation
here, but simply show that this is likely to be the case.
First, we note that perturbation theory tells us that components of eigenvectors
added to a system because of a perturbation are out of phase with the original state
vector.[12] Furthermore, because vacuum radiation will produce many small, independent
effects, we can see by considering either a coordinate or a momentum representation of the
eigenvectors that these effects would affect different eigenvectors differently. So we would
expect the relative phases of the eigenvectors to be randomized.
The above does not tell us how quickly this randomization would occur. However,
Unruh and Zurek [2] have shown in various examples that when an environment perturbs a
system, the off-diagonal elements of the density matrix go to zero in a much shorter time
scale than effects involving spatial and momentum distributions. Thus it seems likely that
vacuum radiation can diagonalize the density matrix in a shorter time than it takes to
randomize momentum.

4. The Arrow Of Time

The dynamical laws of physics are time reversible, i.e., for any given trajectory described
by them, the time reversed trajectory is also a solution of the equations. And in nearly all
cases, both the process described by these equations as time moves forward and the process
described when time is reversed can be observed to occur. But curiously, there are a few
exceptions to this rule. The decay of K-mesons violates CP and therefore (assuming CPT
holds) is not time symmetric. Electromagnetic waves emanate from a source out to infinity,
but do not converge from infinity to a source. Collapse of the wave function is a one -way
process.[13,14] And as Prigogine and co-workers have shown, in systems which are so
unstable that they cannot be described analytically in an ordinary dynamical framework,
process can go in only one direction.[15] Such processes can be called irreversible, and
they are accounted for by saying they are governed by an arrow of time . It is not known
496 J. E. BURNS
what an arrow of time is, what it has to do with the rest of physics, or whether any of the
above arrows of time have anything to do with each other.
It has been shown herein that entropy increase in thermodynamic systems is
produced by the interaction of vacuum radiation with matter. This interaction is time
reversible. However, we can go back a step and ask how vacuum radiation is produced.
Whether an arrow of time is ultimately involved in entropy increase depends on the answer
to this question, as we will see.
In examining this issue, let us start with a classical (SED) analysis. Puthoff [16] has
shown that if vacuum radiation with its frequency-cubed spectrum once exists, then random
interactions with matter in which radiation is absorbed and matter accelerates and reradiates
maintain this frequency-cubed spectrum indefinitely. From this perspective, the random
nature of the interaction of vacuum radiation with any given particle is caused by the
random distribution in position and momentum of other particles the radiation previously
interacted with. All interactions are time-reversible, and it is not necessary to invoke an
"arrow of time" to explain entropy increase in thermodynamic systems.
In quantum mechanics photons exist in quantized units of energy hν . However,
the average energy per photon of vacuum radiation is 1/2 hν . For that reason it is
commonplace to explain the average energy by supposing that photons spontaneously and
causelessly arise out of the vacuum, exist for the time allotted by the uncertainty principle,
and then annihilate themselves back into the vacuum. In this scenario information
describing the state of the newly created vacuum photon arises from nothing, the photon
interacts with matter and modifies the information describing its state according to this
interaction, and this modified information is then destroyed when the photon annihilates
itself.
The dynamical information which is introduced in the creation of virtual photons is
purely random. However, the information which is removed is no longer random (or
potentially is not because the virtual photons could have interacted with an ordered system).
Thus the beginning and end points are inherently different, and an arrow of time is defined.
According to this view, entropy increase is therefore ultimately associated with an arrow of
time.[11]
On the other hand, it would seem that quite different views of the arising and
disappearance of photons are possible. The basic equations of QED and quantum field
theory do not tell us how vacuum photons (or other virtual particles) arise. And creation
and annihilation operators, although they have evocative names, simply describe mappings
from one state to another in Hilbert space, the same as any other operators. The idea that
vacuum photons arise spontaneously out of the vacuum is basically a pictorial device to
account for the average energy per photon of 1/2 hν . Alternatively, one can conceive
that, comparably to the classical picture, vacuum photons arise and disappear through
constructive and destructive phase interference of a large number of photons traveling in
different directions. To be consistent, one would have to view all other virtual particles as
also arising and disappearing through constructive and destructive interference of quantum
phase, perhaps through interaction with negative energy particles. But the appearance and
 VACUUM RADIATION, ENTROPY AND TIME 497
disappearance of virtual particles could perhaps occur in this way. Another possibility is
that the seemingly random appearance and disappearance of virtual particles comes about
through interactions in the extra dimensions provided by string theory. In each of these
cases processes would be entirely time-reversible, and no arrow of time would be involved.
We can put this issue another way by asking: Is the universe a continuo us source of
random dynamical information, creating virtual particles which can interact with matter and
then return some of the previous dynamical information describing this matter to the
vacuum? Or does the universe merely transform dynamical information, with virtual
particles arising and disappearing through a process such as the above? At present there is
no answer to these questions and, given quantum indeterminacy within the limits of the
uncertainty principle, there may never be any conclusive answer.

5. Conclusion

As vacuum radiation interacts with particles, it exchanges momentum with them. The
fractional change in momentum of a particle < δ p x >1/2 / p after one collision time,
2

when multiplied by the lever arm of succeeding molecular interactions, becomes greater
than one in only a few collision times. Therefore, particle momentum is randomized during
that time, and vacuum radiation can account for entropy increase in thermodynamic
systems.
Vacuum radiation interacts with matter in a time-reversible process. Therefore,
whether entropy increase in thermodynamic systems should be viewed as ultimately
connected with an arrow of time depends on whether the arising and disappearance of
vacuum photons should be considered as a time-reversible or irreversible process. Either
possibility appears to be consistent with quantum mechanics.

References

1. Zeh, H.-D. (1989) The Physical Basis of the Direction of Time, Springer-Verlag, New York.
2. Unruh, W.G. and Zurek, W.H. (1989) Reduction of a wave packet in quantum Brownian motion, Phys.
Rev. D 40(4), 1071-1094.
3. Huang, K. (1963) Statistical Mechanics, Wiley, New York.
4. Rueda, A. (1993) Stochastic electrodynamics with particle structure, Part I: Zero-point induced brownian
behavior, Found. Phys. Lett. 6(1), 75-108; (1993) Stochastic electrodynamics with particle structure,
Part II: Towards a zero-point induced wave behavior, Found. Phys. Lett. 6(2), 139-166.
5. Milonni, P.W. (1994) The Quantum Vacuum: An Introduction to Quantum Electrodynamics, Academic,
New York.
6. SED calculations are known to give the same result as quantum mechanical ones for the Casimir effect,
van der Waals forces, the shape of the blackbody spectrum, and the Unruh-Davies effect. See Ref. 5.
7. Chebotarev, L.V. (2000) The de Broglie-Bohm-Vigier approach in quantum mechanics, in S. Jeffers, B.
Lehnert, N. Abramson, and L. Chebotarev (eds.), Jean-Pierre Vigier and the Stochastic Interpretation of
Quantum Mechanics, Apeiron, Montreal, pp. 1-17.
498 J. E. BURNS
8. Abbott, L.F. and Wise, M.B. (1981) Dimension of a quantum-mechanical path, Am. J. Phys. 49(1), 37-39;
Cannata, F. and Ferrari, L. (1988) Dimensions of relativistic quantum mechanical paths, Am. J. Phys.
56(8), 721-725.
9. Sornette, D. (1990) Brownian representation of fractal quantum paths, Eur. J. Phys. 11, 334-337.
10. Haken, H. (1983) Synergetics, Springer-Verlag, New York.
11. Burns, J.E. (1998) Entropy and vacuum radiation, Found. Phys. 28(7), 1191-1207.
12. Peebles, P.J.E. (1992) Quantum Mechanics, Princeton University Press, Princeton, NJ.
13. Penrose, R. (1994). Shadows of the Mind. New York: Oxford University Press, pp. 354-359.
14. It should be noted that not all interpretations of quantum mechanics assume there is such a thing as
collapse of the wave function. See, e.g., Ref. 7.
15. Prigogine, I. (1997) From Poincaré's divergences to quantum mechanics with broken time symmetry,
Zeitschrift für Naturforschung 52a, 37-47; Petrosky, T. and Rosenberg, M. (1997) Microscopic non-
equilibrium structure and dynamical model of entropy flow, Foundations of Physics 27(2), 239-259.
16. Puthoff, H.E. (1989) Source of vacuum electromagnetic zero-point energy, Phys. Rev. A 40(9), 4857-4862;
(1991) Reply to "Comment on 'Source of vacuum electromagnetic zero-point energy'", Phys. Rev. A 44(5),
3385-3386.