I have heard virtual particles pop in and out of existence all the time, most notable being the pairs that pop out beside black holes and while one gets pulled away. But wouldn't this actually violate the conservation of energy?

  • 8
    $\begingroup$ I believe it does violate conservation of energy, but such a violation is allowed during short enough timescales due to the energy-time version of the Heisenberg uncertainty principle. This should best be answered by a physicist. $\endgroup$ Commented May 19, 2015 at 23:34
  • $\begingroup$ Possible duplicate: physics.stackexchange.com/q/147096/2451 Related: physics.stackexchange.com/q/143038/2451 $\endgroup$
    – Qmechanic
    Commented May 20, 2015 at 23:46
  • 1
    $\begingroup$ For a thorough discussion of misconceptions about virtual particles (including precise definitions and references) see my article physicsforums.com/insights/misconceptions-virtual-particles $\endgroup$ Commented Apr 18, 2016 at 18:36
  • $\begingroup$ There is nothing unreal/virtual about the amount of energy or momentum transferred by the virtual particles--especially if we look in to Feynman's diagrams.The only point is that they lie off the mass shell and hence remain unobserved. $\endgroup$ Commented Apr 9, 2017 at 5:26
  • $\begingroup$ There is nothing unreal/virtual about the amount of energy or momentum transferred by the virtual particles--especially if we look in to Feynman's diagrams.The only point is that they lie off the mass shell and hence remain unobserved except for the particle that receives the energy and momentum). The non conservation part is accounted for by Heisenberg's uncertainty principle[as stated in the first comment] $\endgroup$ Commented Apr 9, 2017 at 5:54

5 Answers 5


Ever since Newton and the use of mathematics in physics, physics can be defined as a discipline where nature is modeled by mathematics. One should have clear in mind what nature means and what mathematics is.

Nature we know by measurements and observations. Mathematics is a self consistent discipline with axioms, theorems and statements having absolute proofs, mathematically deduced from the axioms. "Existence" for physics means "measurable", for mathematics "possible to be included in the self consistent theory.

Modern physics has used mathematical models to describe the measurements and observations in the microcosm of atoms, molecules, elementary particles, adding postulates that connect the mathematical calculations with the physical observables

The dominant mathematical model is the field theoretical model that simplifies the mathematics using Feynman diagrams

These diagrams represent terms in an expansion of the desired solution, each term has a diminishing contribution to the cross section of the interaction. The diagram below would be the dominant term, as the next one would be more complicated and therefore smaller by orders of magnitude.

feynman diagram

To each component of the diagram there corresponds one to one a mathematical formula twhich integrated properly will give a prediction for a measurable quantity. In this case the probability of repulsion when one electron scatters on another.

This diagram for example, has as measurable quantities the incoming energy and momentum of the electrons ( four vectors) and of outgoing four vectors . The line in between is not measurable, because it represents a mathematical term that is integrated over the limits of integration, and within the integral energy and momentum are independent variables. The line has the quantum numbers of the photon though not its mass , and so it is called a "virtual photon". It does not obey the energy momentum rule which says that :

$$\sqrt{P\cdot P} = \sqrt{E^2 - (pc)^2} = m_0 c^2$$

The photon has mass zero.

Through the above relation which connects energy and momentum through the rest mass, the un-physical mass of the virtual line depends on one variable, which will be integrated over the diagram; it is often taken as the momentum transfer.

Quantum number conservation is a strong rule and is the only rule that virtual particles have to obey.

There are innumerable Feynman diagrams one can write, and the internal lines considered as particles would not conserve energy and momentum rules if they were on mass shell. These diagrams include vacuum fluctuations that you are asking about, where by construction there are no outgoing measurable lines in the Feynman diagrams describing them. They are useful/necessary in summing up higher order calculations in order to get the final numbers that will predict a measurable value for some interaction.

Thus virtual particles exist only in the mathematics of the model used to describe the measurements of real particles . To coin a word virtual particles are particlemorphic ( :) ), having a form like particle but not a particle.

  • 2
    $\begingroup$ I haven't read particle physics yet,but Eugene Hecht(optics) says electrons exchange virtual photons when interacting and via virtual photons they can exchange momentum which we call as force.so how this is possible if they exist only in mathematics? $\endgroup$
    – Paul
    Commented May 21, 2015 at 13:22
  • $\begingroup$ @Paul It is the Feynman diagrams where the exchange of quantum numbers identifies, photons, gluons , W and Z by their quantum numbers, not by their mass. The experimental fact is that an electron transfers momentum to an electron, in the diagram above. $\endgroup$
    – anna v
    Commented May 21, 2015 at 14:52
  • $\begingroup$ continued. It is a simple picture, but the true mathematics has many complicated higher order exchanges, as it is a perturbative expansion. It is simple to think as if "virtual" is like the real, except that one falls into contradictions like energy conservation, and that one can never do an experimental measurement with a virtual photon. Virtual is not real, that is why the adjective is necessary. $\endgroup$
    – anna v
    Commented May 21, 2015 at 14:53
  • 2
    $\begingroup$ @jameslarge There is no reason that we should accept that virtual particles are real. Quantum field theory never asserts that virtual particles are real. To put it simply, virtual particles are just factors that look mathematically like particles but don't behave quite like particles. But, the theory does not make the logical jump to state that VPs actually exist. These VPs are a mathematical way of quantifying the interaction between real particles. In that sense, your idea of an "isomorphism" is more correct since the diagrammatic calculation keeps track of interactions through VPs. $\endgroup$
    – Ian
    Commented Jul 1, 2015 at 17:54
  • 2
    $\begingroup$ The internal particle corresponding to a propagator with pole at $m^2$ won't always satisfy $p^2 = m^2$ - it is off shell - but energy-momentum is still conserved. $\endgroup$
    – innisfree
    Commented Jul 1, 2015 at 18:14

Energy and momentum are conserved at every vertex of a Feynman diagram in quantum field theory. No internal lines in a Feynman diagram associated with a virtual particles violate energy-momentum conservation. It is true, however, that virtual particles are off-shell, that is, they do not satisfy the ordinary equations of motion, such as $$E^2=p^2 + m^2.$$

There is an added complication. A process might have a definite initial and final state, but an "intermediate state" between the two is in a linear superposition of possible states - in this case, a linear superposition of Feynman diagrams - that interfere with each other. We cannot speak of what particles are in this intermediate state, let alone what their momentum is.

But in spite of that complication, I don't think it's ever justifiable to claim that energy-momentum conservation may be violated briefly because of an uncertainty relation. See e.g. this question for a discussion about the interpretation of $\Delta E \Delta t$.


To understand this one shall take in quantum-mechanical approximation method namely perturbation theory into account. In perturbation theory, systems can go through intermediate virtual states which often have energies different from that of the initial and final states. This is because of time energy uncertainty principle.

Consider an intermediate state with a virtual photon in it. It isn't classically possible for a charged particle to just emit a photon and remain unchanged itself. The state with the photon in it has too much energy, assuming conservation of momentum. However, since the intermediate state lasts only a short time, the state's energy becomes uncertain, and it can actually have the same energy as the initial and final states. This allows the system to pass through this state with some probability without violating energy conservation.

  • $\begingroup$ "Consider an intermediate state with a virtual photon in it. It isn't classically possible for a charged particle to just emit a photon and remain unchanged itself. The state with the photon in it has too much energy" But this isn't possible in quantum mechanics either, and doesn't happen that way as far as I understand. When the photon is emitted the electron loses an amount of energy exactly equal to the photon's energy--it does not remain unchanged as you suggest. It regains the energy later when the photon gets absorbed. $\endgroup$ Commented Dec 4, 2015 at 15:43
  • $\begingroup$ See physics.stackexchange.com/questions/221842/… $\endgroup$ Commented Dec 4, 2015 at 15:44

I think one has to be very careful when talking about "particles popping in and out of existence".

This interpretation is only sort of fine in flat-spacetime QFT, where the Minkowski metric is time-invariant, so has a global timeline Killing vector. The definition of a particle depends on the notion of there existing time invariance! Since black hole solutions are static and asymptotically flat, "particles popping in and out" are sort of okay there too.

BUT, quantum field theory is not a theory of particles, it is a theory of fields. So, "particles popping in and out of existence" is based on a naive "particle-interpretation" of QFT, which is not quite accurate for the following reasons (see also the book by Wald, QFT in Curved Spacetime)

Consider a two-level quantum mechanical system which is coupled to a Klein-Gordon field, $\phi$ in a Minkowski spacetime, for simplicity. The combined system will have a total Hamiltonian of the form

$\mathcal{H} = \mathcal{H}_{\phi} + \mathcal{H}_{q} + \mathcal{H}_{int}$,

where $\mathcal{H}_{\phi}$ is the Hamiltonian of the free Klein-Gordon field. We will consider the quantum mechanical system to be an unperturbed two-level system with energy eigenstates $| x_{o} \rangle$ and $|x_{1} \rangle$, with energies $0$ and $\epsilon$ respectively, so we can define

$\mathcal{H}_{q} = \epsilon \hat{A}^{\dagger} \hat{A}$,

where we define

$\hat{A} |x_{0} \rangle = 0, \quad \hat{A} |x_{1} \rangle = |x_{0} \rangle$.

The interaction Hamiltonian is defined as

$\mathcal{H}_{int} = e(t) \int \hat{\psi}(\mathbf{x}) \left(F(\mathbf{x}) \hat{A} + o\right) d^{3}x$,

where $F(\mathbf{x})$ is a spatial function that is continuously differentiable on $\mathbb{R}^{3}$ and $o$ denotes the Hermitian conjugate. One then calculates to lowest order in $e$, the transitions of a two-level system. In the interaction picture, denoting $\hat{A}_{s}$ as the Schrodinger picture operator, one obtains

$\hat{A}_{I}(t) = \exp(-i \epsilon t) \hat{A}_{s}$.

Therefore, we have that

$(\mathcal{H}_{int})_{I} = \int \left(e(t) \exp(-i \epsilon t) F(\mathbf{x}) \psi_{I}(t,\mathbf{x}) \hat{A}_{s} + o\right) d^{3}x$.

Using Fock space index notion, we can then consider for some $\Psi \in \mathbb{H}$, where $\mathbb{H}$ is the associated Hilbert space, and note that the field is in the state

$|n_{\Psi} \rangle = \left(0, \ldots, 0, \Psi^{a_{1}} \ldots \Psi^{a_{n}}, 0, \ldots \right)$.

The initial state of the full system is then given by

$|\Psi_{i} \rangle = | x \rangle |n_{\Psi} \rangle$.

One then obtains the final state of the system as being

$|\Psi_{f} \rangle = |n _{\Psi} \rangle |x \rangle + \sqrt{n+1} \| \lambda \| (\hat{A} |x \rangle) |(n+1)^{'}\rangle - \sqrt{n} (\lambda, \Psi) (\hat{A}^{\dagger} |x\rangle) |(n-1)_{\Psi}\rangle$,

where $| (n+1)^{'} \rangle$ is defined as in Eq. (3.3.18) in Wald, and $\lambda$ is defined as in Eq. (3.3.15) in Wald.

The key point is that if $|x \rangle = |x_{0} \rangle$, that is, the system is in its ground state, the derivation above shows explicitly that this two-level system can make a transition to an excited state, and vice-versa. Note that the probability of making a downward transition is proportional to $(n+1)$, and even when $n = 0$, this probability is non-zero. This in the \emph{particle interpretation} is interpreted as saying that the quantum mechanical system can spontaneously emit a particle. However, the above calculation in deriving explicitly shows that it is the interaction of the quantum mechanical system with the quantum field that is responsible for the so-called spontaneous particle emission. This misleading picture of the vacuum state is precisely promoted by the particle interpretation of quantum field theory. As the work above also shows, this is not spontaneous particle emission from ``nothing'' in any sense of the word. One must have both a well-defined quantum mechanical system interacting with a well-defined vacuum state for such spontaneous emission to occur, I emphasize that these are not nothing!

The more important point is perhaps that in general curved space times such as the FLRW class of metrics that describe our universe, one can never talk about particles popping in and out of existence, because in general curved space times, there exist no timelike Killing vectors, no Poincare symmetries, no way of defining a covariant ground state, and hence the concept of "particles" has no meaning.

  • 2
    $\begingroup$ So, I tried to follow your logic and even spent significant time tidying up the formulas... until I understood that this makes no sense. You provide a messy and complicated textbook computation and conclude that this overthrows the QFT. What is this field $\phi$ you invent? How does it fix the fundamental Heisenberg-principle-style consideration that allows to find a particle in a vacuum for a short period of time? Do you imply that interaction with gravitational field or a definition of vacuum in curved space plays a role? How does it work in a low energy/Minkowski limit that we live in? $\endgroup$ Commented Jun 15, 2016 at 23:13

Under the context and physical definition that theoretically these particles cannot be ever detected directly because their unstable nature and very tiny lifetime which is within the Heisenberg Uncertainty Principle that makes theoretically impossible their direct detection then,

of course they are physical (life and kicking :), there is undeniable experiment proof collected over the last 40 yrs with the discovery of the W and Z massive bosons virtual particles considered by that time at 1983 and the Higgs Boson virtual particle at 2012 by CERN. Also another recent proof are the latest 2023 run #2 official published results of the Fermilab g-2 muon experiment that prove the interaction of the "virtual" particles with the standard model known matter particles (i.e. fermions).

Currently there is no consensus on this matter if they should be regarded as physical or not and you can easily see this tension among the science community by reading the related wikipedia article.

However, these counter arguments are based solely on the existing theoretical formalism and don't take into account the experiment data and I believe they cannot hold any longer under the new overwhelming experimental evidence. They just hide behind the fact that these are unstable particles with lifetime shorter than the limit set by HUP and therefore cannot have any physical meaning in the formalism (i.e. mathematics) of the theory.

I remind here another concept which was used only as a "computational tool" for describing physics and that was proved finally that it was actually physical, namely the Einstein's cosmological constant Λ which was after confirmed by astronomical observations of the expansion of the universe.

IMO, virtual particles are just particles consisting the vacuum zero-point energy and its quantum fluctuations similar to the known SM particles consisting matter and light. Thhat's all. The second, matter and light particles we know very well and are easier to detect the first are more elusive we have not yet a very good theoretical background for them (see dark energy, dark matter) and they are much harder or even impossible by our current knowledge to directly detect.

In other words, among the three protagonists in our universe matter, light and the vacuum we know very well the first two but very little we know about the vacuum.

Also there is a deeper philosophical question but with physical implications namely the question is the vacuum a property of the matter and light or is the vacuum the source and origin of matter and light? Or in other words, during the very first fraction of the picosecond of the Big Bang which came first the vacuum or matter?


Not the answer you're looking for? Browse other questions tagged or ask your own question.