The Schwinger Effect

Introduction

One of the most dramatic predictions of relativistic quantum field theory is that a sufficiently strong electric field can create particles from the vacuum. This is the Schwinger effect: the spontaneous production of charged particle-antiparticle pairs by a classical background electric field.

That sentence is easy to say and easy to misunderstand.

The effect is not a classical instability of point particles. It is not a correction that appears in ordinary single-particle quantum mechanics. It is a genuine quantum-field-theoretic effect. The vacuum of the Dirac field is not an inert empty stage. In the presence of a constant electric field, the structure of the quantum field evolves in such a way that the state that began as vacuum can turn into a state containing electron-positron pairs.

The lecture introduces the effect in a very concrete way. It says that solving the Schrödinger equation with the Dirac Hamiltonian in a constant classical electric field shows that particles are continuously created from the vacuum. In terms of the point charges that fill the quantum vacuum, one can picture the electric field as pulling correlated positive and negative charges apart, producing a net current. The lecture also notes the historical fact that the problem was first solved and explained by Julian Schwinger in 1951. :contentReference[oaicite:1]{index=1}

But the real point of the lecture is the derivation. Rather than starting from a formal effective action result, it builds the process explicitly from the Dirac Hamiltonian.

The strategy is:

1. choose a gauge for a constant electric field,
2. rewrite the Dirac Hamiltonian in that background,
3. show that the full evolution factorizes into independent problems for each momentum-spin sector,
4. reduce each such sector to a 4-state fermionic system,
5. identify the vacuum/pair sector as a 2-state avoided-crossing problem,
6. map it to the standard Landau–Zener model,
7. interpret the Landau–Zener transition probability as pair production,
8. count how many modes undergo the transition per unit time,
9. derive the pair-production rate.

This is why the lecture is so valuable. It makes the Schwinger effect look like a concrete dynamical mechanism rather than a mysterious formula.

Learning Objectives

• Explain why a constant classical electric field can create charged particle-antiparticle pairs from the vacuum in QFT.
• Write the Dirac Hamiltonian in a constant background electric field using a convenient gauge choice.
• Understand why the Hamiltonian remains translationally symmetric while becoming explicitly time-dependent.
• Derive the mode decomposition into independent $(k,s)$ sectors.
• Explain why each $(k,s)$ sector is only a 4-state fermionic system.
• Identify the vacuum/pair subspace and reduce it to a 2-state dynamical problem.
• Map that 2-state system to the standard Landau–Zener avoided-crossing problem.
• Interpret the Landau–Zener transition probability as the probability of creating an electron-positron pair.
• Understand why pair creation occurs only during a finite “time of decision” for each mode.
• Explain why the pair-production probability is independent of longitudinal momentum $k_3$.
• Derive the qualitative form of the Schwinger production rate and understand its non-perturbative character.

Prerequisite Knowledge

• Dirac quantum field theory
• Fermionic mode expansion of the Dirac field
• Electron/positron operators and vacuum structure
• Classical electromagnetic backgrounds
• Gauge choice in electrodynamics
• Two-level quantum systems
• Basic idea of the Landau–Zener avoided-level-crossing problem
• Momentum-space mode counting

1. Physical statement of the effect

The lecture begins with the central claim:

Solving the Schrödinger equation with the Dirac Hamiltonian in a constant classical electric field reveals that particles are continuously created from the vacuum.

It then gives an intuitive picture in terms of the point charges filling the quantum vacuum: the electric field can pull correlated positive and negative charges apart, producing a net current. The lecture explicitly says that the derivation presented is more explicit than the ones it found in standard texts, while agreeing with the standard result. :contentReference[oaicite:2]{index=2}

That framing matters. The lecture is not using perturbation theory around a particle picture. It is using the full quantum field Hamiltonian in a background field.

2. Dirac Hamiltonian in a constant electric field

The lecture then writes the Dirac Hamiltonian in a classical background electric field. The field is chosen to be constant and directed along the $z$-axis. The convenient gauge choice is

$$\vec A = - c E\,\hat z\, t, \qquad A_0 = 0.$$

With this choice,

$$\vec E = -\dot{\vec A} - \nabla \Phi = E\,\hat z.$$

This gauge makes the Hamiltonian explicitly time-dependent, but it preserves translation symmetry in space. That is a very useful tradeoff. The lecture emphasizes exactly that point: we get a time-dependent Hamiltonian with translation symmetry. :contentReference[oaicite:3]{index=3}

The free Dirac Hamiltonian is then modified by the coupling to the background field. The lecture isolates the extra term involving the current operator $\hat J_3$, the $z$-component of the Dirac current. So the field couples directly to the charge flow along the field direction.

3. Free Hamiltonian plus current coupling

The lecture rewrites the Hamiltonian as the free Dirac Hamiltonian plus a time-dependent coupling proportional to the electric field and the current operator $\hat J_3$. It also gives the free-field Hamiltonian in momentum space:

$$\hat H_{\text{free}} = \sum_s \int d^3k\, \sqrt{(Mc^2)^2 + (\hbar c k)^2} \left( \hat b_s^\dagger(k)\hat b_s(k) + \hat d_s^\dagger(k)\hat d_s(k) \right),$$

up to the lecture’s exact notation conventions. :contentReference[oaicite:4]{index=4}

The nontrivial work is now to understand the current operator $\hat J_3$ in terms of the free electron and positron mode operators.

4. Evaluating the current operator

The lecture states that after a direct computation using the explicit gamma matrices and the $u$- and $v$-spinors, one obtains an expression for

$$\hat J_3 = \int d^3r\, \hat\psi^\dagger \gamma^0 \gamma^3 \hat\psi.$$

The result contains two kinds of terms:

• number-operator-like terms involving electron and positron occupations,
• and mixing terms that create or destroy an electron-positron pair.

The slide then explains that by choosing a new, $k$-dependent spin quantization axis, one can rotate the spin structure into a simpler form. If one does not insist on keeping track of spin directions relative to one fixed external frame, this simplifies the algebra drastically. :contentReference[oaicite:5]{index=5}

This is an important move. The pair-production probability will not depend on the fine details of how spin is labeled in a fixed lab frame. So the lecture adopts the basis that makes the mode dynamics simplest.

5. Simplified Hamiltonian and the irrelevant $C$-number

After the simplification, the lecture notes that one term $C$ arises from the operator change. In a finite system with discrete $k$, $C$ is just 1. In an infinite system it is formally infinite. But it is a pure $C$-number, so it only contributes the same time-dependent phase to every quantum state. Therefore it has no effect on observables, and the lecture sets $C=0$. :contentReference[oaicite:6]{index=6}

This is a familiar QFT lesson:

• not every infinity matters,
• $C$-number shifts that only change the overall phase of the state are physically irrelevant.

The lecture also introduces the shorthand

$$E_k = \hbar \omega, \qquad \omega = \sqrt{\left(\frac{Mc^2}{\hbar}\right)^2 + c^2k^2},$$

to keep the symbol $E$ from being confused with the electric field strength. :contentReference[oaicite:7]{index=7}

6. Factorization into independent $(k,s)$ sectors

The lecture then reaches one of the most important structural simplifications. The Hamiltonian can be written as

$$\hat H = \sum_s \int d^3k\, \hat H_{k,s}(t),$$

with different $(k,s)$ sectors commuting:

$$[\hat H_{k,s}(t), \hat H_{k',s'}(t)] = 0$$

for distinct sectors. Therefore the full time evolution factorizes into a tensor product over the solutions in each $(k,s)$ subspace separately. The lecture says explicitly:

So this small subspace problem is now our only problem. :contentReference[oaicite:8]{index=8}

This is the big reduction. Instead of solving one enormous field-theory problem, we solve one small problem over and over again, once for each momentum-spin pair.

7. Finite volume and the fermionic 4-state system

To make the mode algebra concrete, the lecture temporarily considers a finite volume, so the momentum labels become discrete rather than continuous. Then each $(k,s)$ sector is built from one electron mode and one positron mode, each fermionic. That means each sector has only four basis states:

• $|00\rangle$: no electron, no positron,
• $|10\rangle$: one electron only,
• $|01\rangle$: one positron only,
• $|11\rangle$: one electron and one positron.

The lecture defines these explicitly and notes that because the field is fermionic, this is the whole subspace: occupation numbers are 0 or 1 only. It also points out that we are always free to redefine the overall phase of a quantum state, and chooses phases in a way that removes an inconvenient $\pm_s$ factor from the Hamiltonian. This does not change any observable result but makes the equations cleaner. :contentReference[oaicite:9]{index=9}

This is where the field-theory problem becomes an ordinary few-state quantum-mechanical problem.

8. Structure of the 4-state Hamiltonian

In this basis, the lecture shows that the Hamiltonian acts diagonally on the one-particle states $|10\rangle$ and $|01\rangle$, while it mixes $|00\rangle$ and $|11\rangle$. So the four-state problem splits into:

• a two-dimensional subspace with one-particle states that remain one-particle states,
• and a two-dimensional vacuum/pair subspace where nontrivial dynamics happens.

The lecture computes the instantaneous eigenvalues of the full 4-state Hamiltonian and notes that the one-particle sector corresponds to the middle doubly-degenerate eigenvalue. These states may represent sectors where an electron or positron is already present, but they are not the initial vacuum state. :contentReference[oaicite:10]{index=10}

So if we want to understand pair creation from vacuum, only the $|00\rangle$-$|11\rangle$ subspace matters.

9. The vacuum/pair 2-state system

The lecture then states that the state which is initially the vacuum must be of the form

$$|\Psi(t)\rangle = \alpha(t)|00\rangle + \beta(t)|11\rangle.$$

Within this 2-state subspace, the time-dependent Schrödinger equation becomes a $2\times 2$ matrix equation. The explicit matrix elements depend on $k_3$, the electric field $E$, and the transverse momentum combination

$$\sqrt{\omega^2 - c^2k_3^2},$$

again up to the lecture’s notation conventions. :contentReference[oaicite:11]{index=11}

The lecture then performs a basis change and an overall phase redefinition:

$$\alpha(t)\to A(t),\qquad \beta(t)\to B(t),$$

with momentum-dependent coefficients chosen so that the Hamiltonian takes the standard avoided-crossing form.

It emphasizes that this transformation is only a basis change plus an overall phase shift.

10. Mapping to the Landau–Zener problem

After the basis change, the lecture says: How does the Schrödinger equation look in this basis? The Landau–Zener problem! :contentReference[oaicite:12]{index=12}

It introduces parameters

$$\nu^2 := -\,\frac{cqE}{\hbar}, \qquad t_0 := -\frac{\hbar k_3}{qE}, \qquad \gamma := \sqrt{\omega^2 - c^2k_3^2},$$

up to sign conventions chosen to match the standard textbook form. The resulting effective Hamiltonian is

$$\frac{\hat H(t)}{\hbar} = \nu^2 (t-t_0)\,\sigma_z + \gamma\,\sigma_x,$$

which is exactly the canonical Landau–Zener Hamiltonian for an avoided level crossing. :contentReference[oaicite:13]{index=13}

This is the central mathematical result of the lecture.

The Schwinger effect, mode by mode, is a Landau–Zener crossing problem.

11. Avoided crossing and adiabatic states

The lecture then interprets the two-level problem geometrically. For large positive and negative times, the Hamiltonian has simple adiabatic eigenstates. The energy levels would cross linearly if $\gamma=0$, but because $\gamma\neq 0$, the crossing is avoided. The lecture presents the standard avoided-crossing picture explicitly and identifies the exact instantaneous eigenvalues as

$$\Omega_\pm(t) = \pm \sqrt{\nu^4 (t-t_0)^2 + \gamma^2}.$$

The plotted figures on the slides show the avoided crossing and compare the asymptotic Landau–Zener prediction with the exact probability for different values of the adiabaticity parameter. :contentReference[oaicite:14]{index=14}

This is the physical heart of the mechanism:

• far from $t_0$, the vacuum-like and pair-like adiabatic states are well separated,
• near $t_0$, the level gap is smallest,
• the system can make a non-adiabatic transition,
• that transition is precisely pair creation.

12. The Landau–Zener transition probability

The lecture states the standard Landau–Zener asymptotic result: two orthogonal exact solutions behave at early and late times like adiabatic eigenstates, with exponentially small or large transition coefficients governed by the Landau–Zener exponent. The probability to end in the opposite adiabatic state is given by the famous exponential

$$P_{\text{LZ}} = e^{-\pi \gamma^2/\nu^2},$$

up to the lecture’s parameter conventions and factors of two in definition. The slides then plot exact probabilities for several values of the dimensionless parameter $\epsilon$, showing excellent agreement with the Landau–Zener prediction in the relevant limit. :contentReference[oaicite:15]{index=15}

For the Schwinger effect, this is the pair-creation probability for a given $(k,s)$ mode pair.

13. What do the adiabatic states mean physically?

The lecture then returns from mathematics to physics and asks: What are these states, in terms of physical electrons or positrons?

It answers:

• the lower adiabatic state is the physical vacuum, the instantaneous ground state,
• the upper adiabatic state is the physical particle-antiparticle pair, the instantaneous highest excited state of the given $(k,s)$ mode.

It says explicitly that any detector passive enough to count particles rather than act as a strong perturbation must register the lower state as “no particles” and the upper state as “one positron-electron pair.” :contentReference[oaicite:16]{index=16}

That interpretation is critical. The avoided-crossing transition is not just a basis change; it is the actual creation of a real electron-positron pair in that mode.

14. The time of decision for each mode

The lecture then explains the dynamics of when a given mode can produce a pair.

For any mode $(k,s)$, there is a special crossing time

$$t_0 = -\frac{\hbar k_3}{qE}.$$

Away from that time, the mode remains adiabatically in its vacuum state. Only during a finite interval around $t_0$ — the avoided-crossing region — is there any significant chance of transition. The lecture describes this as the “time of decision” for that $\pm k,s$ pair. The duration of that interval is set by the Landau–Zener parameters. :contentReference[oaicite:17]{index=17}

So pair creation is not a continuous smooth build-up for each individual mode. For each mode, it is localized in time near its own crossing.

15. Why the production probability is independent of $k_3$

The lecture then observes that the Landau–Zener probability depends on $\gamma$ and $\nu$, but not on $k_3$ separately. Since

$$\gamma^2 = \omega^2 - c^2k_3^2$$

reduces to the transverse-momentum and mass contribution, the longitudinal momentum $k_3$ only determines when the avoided crossing happens, not the probability that it produces a pair. The lecture says this explicitly: the probability of a pair appearing is independent of $k_3$. :contentReference[oaicite:18]{index=18}

That fact is what makes the total production rate constant in time.

16. Constant production rate

Suppose the system is in the vacuum at $t=0$, and choose axes so that the electric field points in the convenient positive direction. Then modes whose crossing times $t_0$ lie in the past have already “missed their chance” and stay empty forever, while at any given time there is a whole slice of modes whose $t_0$ is near the present time and are currently in the non-adiabatic window. Because the transition probability is independent of $k_3$, this moving slice of modes produces pairs at a constant average rate. :contentReference[oaicite:19]{index=19}

This is the physical origin of the constant pair-production rate in a constant field.

17. Counting modes and deriving the Schwinger rate

The lecture then counts how many modes cross their avoided-crossing time within a time interval $T$ in a finite volume $V=L_1L_2L_3$. There is:

• a factor of 2 from the two spin states,
• a density of longitudinal $k_3$-modes passing through their crossing time during $T$,
• and an integral over transverse momenta.

Each mode contributes with the Landau–Zener probability. Summing/integrating them gives the average number of created pairs in time $T$ and volume $V$. Finally, dividing by $VT$ yields the pair-production rate per unit volume in a constant electric field. The last slide emphasizes two points:

• for electrons and positrons, one gets the standard Schwinger exponential suppression factor,
• and that exponential is non-perturbative in the electric charge $q$. Pair production is therefore, in a very strict sense, a purely quantum electrodynamic effect. :contentReference[oaicite:20]{index=20}

The standard form of the leading exponential suppression is

$$\Gamma/V \propto E^2 \exp\!\left(-\frac{\pi M^2 c^3}{\hbar q E}\right),$$

with the precise prefactor depending on conventions and units. The lecture’s main emphasis is the exponential suppression and its non-perturbative nature.

Worked Examples

Example 1: Why the vacuum/pair sector is only two-dimensional

For each fixed $(k,s)$, there is one electron mode and one positron mode, both fermionic. So the entire sector has only four basis states:

$$|00\rangle,\quad |10\rangle,\quad |01\rangle,\quad |11\rangle.$$

The electric-field-driven Hamiltonian leaves the one-particle states decoupled from the vacuum/pair sector, while it mixes

$$|00\rangle \leftrightarrow |11\rangle.$$

Therefore vacuum pair creation is governed by a $2\times 2$ problem, not the full field Hilbert space.

Example 2: Why $k_3$ determines the crossing time but not the transition probability

The effective Landau–Zener Hamiltonian takes the form

$$\hat H/\hbar = \nu^2 (t-t_0)\sigma_z + \gamma \sigma_x,$$

with

$$t_0 = -\frac{\hbar k_3}{qE}, \qquad \gamma = \sqrt{\omega^2 - c^2k_3^2}.$$

So $k_3$ shifts the center time $t_0$ of the avoided crossing, but the transition probability only depends on the Landau–Zener combination $\gamma^2/\nu^2$, which depends only on mass and transverse momentum. Hence every longitudinal mode has the same chance to pair-produce when its crossing time arrives.

Intuition

This lecture gives a very concrete picture of the Schwinger effect.

Each momentum-spin mode of the Dirac field in a constant electric field behaves like a tiny two-level system:

• the lower adiabatic state is “vacuum” for that mode,
• the upper adiabatic state is “electron-positron pair” for that mode.

As time passes, the electric field sweeps the mode through an avoided crossing. Most of the time nothing interesting happens; the mode just stays adiabatic. But near one special time $t_0$, the gap is smallest and the mode can make a non-adiabatic jump. That jump is exactly pair creation.

So the vacuum is not exploding all at once. Mode after mode, each at its own crossing time, gets a chance to turn vacuum into pair. Because a constant electric field sweeps an endless stream of modes through this crossing at a constant rate, the average production rate is constant.

The Landau–Zener mapping is what makes this picture sharp instead of vague.

Common Mistakes

• Thinking the Schwinger effect is a single-particle quantum-mechanics problem. It is not; it is a QFT vacuum-instability problem.
• Confusing the time-dependent gauge choice with loss of translation symmetry. In this gauge the Hamiltonian is time-dependent but still translationally invariant.
• Forgetting that the full field problem factorizes into independent $(k,s)$ sectors.
• Missing why fermionic statistics make each $(k,s)$ sector only 4-dimensional.
• Thinking the one-particle states are the relevant vacuum states for pair creation. They are not.
• Missing that only the $|00\rangle$-$|11\rangle$ subspace matters for vacuum pair creation.
• Treating the Landau–Zener mapping as only qualitative. In the lecture it is the actual exact reduction.
• Forgetting that the longitudinal momentum $k_3$ changes the crossing time $t_0$, not the pair-production probability.
• Missing the meaning of the adiabatic states: lower means physical vacuum, upper means physical pair.
• Forgetting that the Schwinger exponential is non-perturbative in the electric charge.

Short Summary

The lecture studies the Dirac field in a constant classical electric field and shows that the vacuum becomes unstable to electron-positron pair creation. Choosing the gauge

$$\vec A = -cEt\,\hat z,\qquad A_0=0,$$

makes the Hamiltonian time-dependent while preserving translation symmetry. Rewriting the current operator in terms of Dirac creation and annihilation operators and choosing a convenient spin basis reduces the full problem to a sum over independent $(k,s)$ sectors. Each sector is only a 4-state fermionic system, and the physically relevant vacuum/pair dynamics lies in the 2-state subspace spanned by $|00\rangle$ and $|11\rangle$. After a basis change and phase redefinition, the Schrödinger equation for this subspace becomes the standard Landau–Zener avoided-crossing problem:

$$\frac{\hat H(t)}{\hbar} = \nu^2(t-t_0)\sigma_z + \gamma\sigma_x.$$

The lower adiabatic state is interpreted as the instantaneous physical vacuum and the upper adiabatic state as the instantaneous electron-positron pair state. Therefore the Landau–Zener transition probability is the probability of pair creation for that mode. Since the probability is independent of the longitudinal momentum $k_3$, while $k_3$ only determines the crossing time $t_0$, a constant electric field produces pairs at a constant average rate as successive modes pass through their avoided crossings. Counting these modes gives the Schwinger pair-production rate per unit volume, with its characteristic non-perturbative exponential suppression

$$\exp\!\left(-\frac{\pi M^2 c^3}{\hbar q E}\right).$$

The lecture thus presents the Schwinger effect as a precise non-adiabatic transition problem in quantum field theory, not just as a formal vacuum-instability slogan. :contentReference[oaicite:21]{index=21}

Practice Problems

1. Why is the Schwinger effect inherently a quantum-field-theoretic phenomenon rather than a one-particle relativistic wave-equation effect?

2. What is the advantage of choosing the gauge

$$\vec A = -cEt\,\hat z,\qquad A_0=0$$

for a constant electric field?

3. Why does the full Dirac-field evolution factorize into independent $(k,s)$ sectors?

4. Why is each $(k,s)$ sector only a 4-state system?

5. Which two states form the physically relevant vacuum/pair subspace, and why?

6. Why can the 2-state problem be mapped to the Landau–Zener Hamiltonian?

7. What do the lower and upper adiabatic eigenstates mean physically in the Schwinger problem?

8. Why does the longitudinal momentum $k_3$ determine the time of the avoided crossing but not the pair-production probability?

9. What is meant by the “time of decision” for a given $(k,s)$ mode?

10. Why does a constant electric field produce pairs at a constant average rate?

11. Why is the Schwinger exponential suppression called non-perturbative in the electric charge?

12. Explain in your own words how the Landau–Zener picture turns “vacuum pair creation” into a concrete dynamical process.