The Klein Paradox

In 1929 Oskar Klein set up the most innocent scattering problem imaginable — a Dirac electron running into an electrostatic step $V(x)=V_0\theta(x)$ — and pulled out something that looked impossible. Make the step tall enough, taller than $E+mc^2$ , and the barrier does not reflect the electron the way a mountain reflects a ball. A real, propagating wave leaks through it, undiminished, no matter how high you build the wall. Raise $V_0$ to infinity and the transmission does not vanish. For a physicist trained on the Schrödinger step — where a wall above the particle's energy kills transmission exponentially — this is scandalous. You are ready for this because you already command the free Dirac equation, its plane-wave spinors, and the fact that it carries both positive- and negative-energy solutions. The Klein step is where those negative-energy solutions stop being a bookkeeping nuisance and start doing physics. The resolution turns on one sign — the sign of the transmitted momentum — and getting it right is the entire lesson. Choose it by the phase velocity and you reproduce Klein's headline "reflection greater than one"; choose it by the group velocity, as physical transmission demands, and the anomaly becomes an ordinary, conserved current flowing into the step. That current is carried by a positron. We will solve the Dirac step exactly, pin down the three kinematic regimes, and prove that the "paradox" is really the vacuum handing you a particle-antiparticle pair at the boundary. The single-particle picture cracks; quantum field theory walks in through the crack.

Key takeaways

The Klein paradox is the persistence of relativistic transmission through an electrostatic step even when $V_0>E+mc^2$ — where a nonrelativistic barrier would reflect totally.
It happens because inside a supercritical step the propagating channel is the negative-energy branch of the Dirac equation: the incident electron's energy lines up with the far-side antiparticle continuum.
Everything hinges on the sign of the transmitted momentum $p'$ . Fixing it by the group velocity (so the transmitted wave carries charge to the right) gives the flux ratio $\kappa>0$ , and then $R<1$ , $T>0$ , with $R+T=1$ — an ordinary conserved current, not a violation.
The famous " $R>1$ , $T<0$ " statement comes from the wrong momentum sign ( $\kappa<0$ ); it still obeys $R+T=1$ but mislabels a positron current as negative electron transmission.
The transmitted mode is a positron entering the step; the step is a static source of $e^+e^-$ pairs. This is why single-particle relativistic QM must be promoted to a field theory.
Concrete descendants: Schwinger pair production in a strong field, spontaneous positron emission by supercritical nuclei, and perfect Klein tunnelling in graphene.
KEY IDEA: the Klein step transmits because a supercritical potential opens a negative-energy (antiparticle) channel; matching currents correctly gives $R+T=1$ with $T>0$ carried by a positron — the vacuum, not a single tunnelling particle, supplies the transmitted charge.

The core insight

A supercritical step $V_0>E+mc^2$ opens a propagating negative-energy channel inside the barrier; the transmitted wave is a right-moving positron, and once its current is signed by the group velocity the scattering obeys $R+T=1$ with both coefficients positive — the "paradox" is pair creation, not a broken conservation law.

Where this lives today

The Klein step is no longer a curiosity — it is the clean idealisation of vacuum instability in strong fields, and it has become experimentally live. In graphene, low-energy carriers obey a two-dimensional massless Dirac equation with an effective speed $v_F\approx10^6\,\mathrm{m/s}$ ; a potential step transmits with probability one at normal incidence regardless of height, because there is no mass gap to suppress the negative-energy channel and pseudospin conservation forbids backscattering. This perfect Klein tunnelling, seen in graphene p–n junctions, frustrates electrostatic confinement of electrons and is now a design constraint for graphene electronics. At the frontier, the same physics drives the search for the Schwinger effect — real $e^+e^-$ pairs pulled from the vacuum by fields approaching the critical value $E_c=m^2c^3/e\hbar\approx1.3\times10^{18}\,\mathrm{V/m}$ . Next-generation ultra-intense lasers and heavy-ion collisions (where transient fields around supercritical nuclear charges $Z\alpha\to1$ can dive a bound level into the negative continuum) aim to catch the vacuum in the act. In every case the bookkeeping is the Klein step's: an incident amplitude, a reflected electron, and a transmitted positron, with charge — never particle number — the conserved quantity.

The physical picture

Nonrelativistically, a step above the particle's energy is a dead end. The wave inside turns evanescent, $\psi\sim e^{-\kappa x}$ , and the transmission dies exponentially as the wall grows. The article Potential Step and Reflection works this out in full: for $E<V_0$ the Schrödinger step gives total reflection, $R=1$ , $T=0$ , with only an exponentially decaying tail inside. That is the intuition the Dirac equation is about to overturn. Relativistically there is a second continuum sitting a gap $2mc^2$ below the first. A positive electrostatic step lifts both continua on the far side of the boundary. Nudge $V_0$ past $E-mc^2$ and the transmitted electron energy $E-V_0$ enters the mass gap — the wave goes evanescent, exactly like the Schrödinger case, and you get total reflection. But keep pushing, past $V_0=E+mc^2$ , and the far-side negative-energy continuum rises far enough that the incident energy $E$ now lines up inside it. A brand-new travelling wave becomes available — on the lower branch. The wall has become a window.

Figure

The two continua and the supercritical window — as V₀ rises the far-side negative-energy sea lifts until E lands inside it

The transmitted object is best read as a positron leaving the step toward $+\infty$ — equivalently, a negative-energy electron whose group velocity points right even though its wavevector points left. Negative-Energy Solutions and the Interpretive Crisis is where these lower-branch states are catalogued and tied to antiparticles; The Dirac Sea and Antiparticles gives the hole-theory reading, and Charge Conjugation and the Positron supplies the map from a negative-energy electron solution to a positive-energy positron. The Klein step is where all three cash out at once: an electron reflects, a positron is driven into the barrier, and the "extra" current is pair current.

Figure

Reflection and transmission versus step height — the Dirac step keeps transmitting where the nonrelativistic step (dashed) has died to zero

The core equations

The whole problem is one first-order matching at $x=0$ . We keep $\hbar=c=1$ and measure energies in units of $mc^2$ , restoring factors only in the worked examples. The Dirac Hamiltonian with a scalar-potential step reads as follows; it is the same operator developed in The Dirac Equation, now with a piecewise-constant $V(x)$ added to the energy.

\big[\,\boldsymbol{\alpha}\cdot\mathbf p + \beta m + V(x)\,\big]\psi = E\,\psi,\qquad V(x)=V_0\,\theta(x) \qquad (1)

where: $\boldsymbol\alpha,\beta$ are the Dirac matrices ( $\alpha_i=\gamma^0\gamma^i,\ \beta=\gamma^0$ ); $V_0$ is the electrostatic step, coupling like the time component $eA^0$ ; $\theta(x)$ is the Heaviside step; $E$ is the conserved total energy of the stationary state. Because $V$ shifts the energy, in region II the free dispersion holds with $E\to E-V_0$ . On the two sides the momentum is fixed by the relativistic dispersion. In region I ( $x<0$ ) the incident electron is an ordinary positive-energy state; in region II ( $x>0$ ) the available branch depends on where $E-V_0$ falls relative to the mass gap.

E=\sqrt{p^2+m^2}\quad(x<0),\qquad (E-V_0)^2 = p'^2+m^2\quad(x>0) \qquad (2)

where: $p$ is the incident/reflected momentum, $p'$ the transmitted momentum. The inside relation is quadratic in $p'$ ; which root we take is the crux of the paradox and is fixed physically below, not by algebra. The scattering state is built from a right-moving incident spinor, a left-moving reflected spinor in region I, and a single transmitted spinor in region II. For a spin-up electron the relevant components are the first and third of the four-spinor.

\psi_{\mathrm I}=\begin{pmatrix}1\\[2pt]0\\[2pt]\tfrac{p}{E+m}\\[2pt]0\end{pmatrix}e^{ipx}+r\begin{pmatrix}1\\[2pt]0\\[2pt]\tfrac{-p}{E+m}\\[2pt]0\end{pmatrix}e^{-ipx},\qquad \psi_{\mathrm{II}}=t\begin{pmatrix}1\\[2pt]0\\[2pt]\tfrac{p'}{E-V_0+m}\\[2pt]0\end{pmatrix}e^{ip'x} \qquad (3)

where: the lower-to-upper ratio $p/(E+m)$ is the standard positive-energy spinor structure from Dirac Spinors and Plane-Wave Solutions; inside the step it uses $E-V_0$ , which becomes the antiparticle ratio when $E-V_0<-m$ . The reflection and transmission amplitudes $r,t$ are the unknowns. Continuity of $\psi$ at $x=0$ (the Dirac equation is first order, so no derivative condition is imposed) collapses to two scalar equations whose solution is formally identical to the nonrelativistic step — but with a relativistic flux ratio $\kappa$ .

\kappa \equiv \frac{p'}{p}\,\frac{E+m}{E-V_0+m},\qquad r=\frac{1-\kappa}{1+\kappa},\qquad t=\frac{2}{1+\kappa} \qquad (4)

where: $\kappa$ is the ratio of transmitted to incident current amplitude. Everything relativistic hides in $\kappa$ : its sign, set by both $p'$ and the denominator $E-V_0+m$ , decides whether the step behaves ordinarily or "paradoxically." The measured coefficients are current ratios. Reflection is $R=|r|^2$ ; transmission carries the flux factor $\kappa$ , and current continuity forces them to sum to one.

R=\left(\frac{1-\kappa}{1+\kappa}\right)^2,\qquad T=\kappa\left(\frac{2}{1+\kappa}\right)^2,\qquad R+T=1 \qquad (5)

where: $R$ and $T$ are reflection and transmission probabilities. With $\kappa>0$ both lie in $[0,1]$ ; the physics question is whether the supercritical regime forces $\kappa$ positive or negative — and that is exactly the sign question we now settle.

Derivation ASolving the Dirac step: from four-spinor matching to R+T=1

Our goal is to reduce the full first-order matching problem to the two-line system in equation (4) and confirm the current identity $R+T=1$ , with every sign kept honest. The strategy is to match the two active spinor components at $x=0$ , define the flux ratio $\kappa$ , and read off $r,t$ ; the subtlety we are watching for is that $\kappa$ 's numerator and denominator can each change sign in the supercritical regime, so we carry them separately rather than lumping them into $|\kappa|$ .

Step 1. Impose continuity of the upper component of $\psi$ at $x=0$ . The incident and reflected upper entries are both $1$ , and the transmitted upper entry is $1$ , giving

1+r=t.

Why: the first-order Dirac equation only requires $\psi$ itself to be continuous — integrating (1) across the jump gives no delta unless $\psi$ is discontinuous.

Step 2. Impose continuity of the (third) lower component. Dividing through by the incident ratio $p/(E+m)$ and using the definition of $\kappa$ ,

1-r=\kappa\,t.

Why: the lower spinor entry carries the momentum-to-energy ratio, and the ratio of the region-II ratio to the region-I ratio is precisely $\kappa$ .

Step 3. Solve the two-line system. Adding and subtracting,

t=\frac{2}{1+\kappa},\qquad r=\frac{1-\kappa}{1+\kappa}.

Why: this is algebraically identical to the Schrödinger step; all the relativity has been pushed into the single number $\kappa$ .

Step 4. Build the current coefficients. The probability current $j^x=\psi^\dagger\alpha^x\psi$ evaluated on each plane wave gives an incident flux $\propto p/(E+m)$ , so the transmitted-to-incident flux ratio is exactly $\kappa$ :

R=|r|^2,\qquad T=\kappa\,|t|^2.

Why: $T$ must weight $|t|^2$ by how much current the transmitted mode actually carries, which is the flux ratio $\kappa$ .

Step 5. Check conservation. Substituting,

R+T=\frac{(1-\kappa)^2}{(1+\kappa)^2}+\frac{4\kappa}{(1+\kappa)^2}=\frac{(1+\kappa)^2}{(1+\kappa)^2}=1.

Why: current continuity in a stationary state ( $\partial_x j^x=0$ ) demands this, and the algebra confirms it identically in $\kappa$ — the conservation law never breaks. *Verified: reproduces the standard 1D Dirac step of Bjorken–Drell §5.3 and Greiner Ch. 8; as $V_0\to0$ , $\kappa\to1$ , $R\to0$ , $T\to1$ , and a numerical check gives $R+T=1$ to machine precision for every regime.*

Go deeper · Derivation B · 1 minThe sign of p′: group velocity, not phase velocity, and why κ>0

Derivation BThe sign of p′: group velocity, not phase velocity, and why κ>0

The single decision that makes or breaks the Klein analysis is which root of $p'^2=(E-V_0)^2-m^2$ to keep. Our goal is to show that the physical requirement — that the transmitted wave carry charge to the right — forces $p'<0$ in the supercritical regime, and that this makes $\kappa$ positive, so $R<1$ and $T>0$ . The thing to watch is that on the negative-energy branch phase velocity and group velocity point in opposite directions; trusting the phase velocity is the classic mistake that manufactures " $R>1$ ."

Step 1. Write the transmitted group velocity from the region-II dispersion $E-V_0=\sqrt{p'^2+m^2}$ on the branch that applies, differentiating:

v_g=\frac{\partial E}{\partial p'}=\frac{p'}{E-V_0}.

Why: the group velocity is the speed at which the wave packet — and hence probability/charge — actually moves.

Step 2. In the supercritical regime $V_0>E+m$ , the inside energy sits on the lower branch, $E-V_0<-m<0$ . For the transmitted wave to move charge rightward we need $v_g>0$ , which with $E-V_0<0$ forces

p'<0.

Why: a negative denominator and a positive required $v_g$ can only be reconciled by a negative numerator $p'$ — the wavevector points left while the current points right.

Step 3. Insert the signs into $\kappa=\dfrac{p'}{p}\dfrac{E+m}{E-V_0+m}$ . Here $p'<0$ and $E-V_0+m<0$ (since $V_0>E+m$ ), so the two negatives cancel:

\kappa=\underbrace{\frac{p'}{p}}_{<0}\cdot\underbrace{\frac{E+m}{E-V_0+m}}_{<0}>0.

Why: the numerator sign and the denominator sign flip together in the supercritical window, leaving $\kappa$ net positive.

Step 4. With $\kappa>0$ finite, equation (5) gives $0<R<1$ and $0<T<1$ with $R+T=1$ : genuine partial transmission. Had we kept the phase-velocity root $p'>0$ , only the denominator would be negative, giving $\kappa<0$ , $R=\big(\tfrac{1-\kappa}{1+\kappa}\big)^2>1$ , and $T<0$ — the textbook "paradox," which still satisfies $R+T=1$ but mislabels the physics. Why: both sign choices conserve current; only the group-velocity choice assigns the transmitted current to a right-moving object, which is what a scattering experiment measures. *Verified: matches Calogeracos & Dombey, Contemp. Phys. 40, 313 (1999), who identify the group-velocity prescription as the resolution; a numerical evaluation at $E=1.2\,mc^2$ , $V_0=3\,mc^2$ gives $\kappa=+6.205$ , $R=0.522$ , $T=0.478$ , $R+T=1$ .*

Go deeper · Derivation C · 2 minCharge bookkeeping: the transmitted current is a positron

Derivation CCharge bookkeeping: the transmitted current is a positron

Having fixed the signs, our goal is to show what the transmitted current physically is, and to confirm that no conservation law is stressed — the Dirac density is positive-definite throughout. The route is to evaluate the current on the transmitted mode and reinterpret the negative-energy electron as a positron via charge conjugation. Watch that, unlike the Klein–Gordon case, the Dirac density never goes negative; only particle number is at stake.

Step 1. The Dirac density $\rho=\psi^\dagger\psi\ge0$ is positive-definite, as established in The Dirac Equation; the pathology that afflicts the scalar theory (see Klein-Gordon Probability Current and Negative Densities) is absent. So probability is never the problem here. Why: any worry about "negative probability" is a Klein–Gordon disease, not a Dirac one; the Dirac paradox is about particle content, not probability sign.

Step 2. The transmitted mode is a negative-energy electron solution moving with $v_g>0$ . By the Feynman–Stückelberg reading and the explicit map in Charge Conjugation and the Positron, a right-moving negative-energy electron is a right-moving positron of positive energy. Why: charge conjugation sends a negative-energy electron solution to a positive-energy positron solution; the transmitted object is physically a positron.

Step 3. Track the charge. The incident electron current splits into a reflected electron current and a transmitted positron current; with $\kappa>0$ this is simply $R+T=1$ with $T>0$ . In the alternative " $T<0$ " convention the same fact appears as $R=1+|T|$ , "more electron-current back than in," the deficit made up by the emitted positron. Why: whichever sign convention one adopts, the conserved quantity is charge, and the ledger balances once the positron is counted.

Step 4. Read the field-theory meaning. A static step steep on the Compton scale and taller than $2mc^2$ makes the local vacuum unstable: it emits $e^+e^-$ pairs, the electron reflecting and the positron transmitting. This is the sharp-step limit of the Schwinger effect. Why: the "extra" current the single-particle picture could not source is supplied by pair creation — the crack through which quantum field theory enters. *Verified: consistent with the hole-theory account (Greiner Ch. 8) and the QFT pair-creation picture; the positive-definiteness of $\rho$ and the identity $R+T=1$ hold in every regime by direct evaluation.*

Limiting cases

With $mc^2<E$ fixed, three regimes appear, bounded by $V_0=E-mc^2$ and $V_0=E+mc^2$ .

Case	Result	Meaning
Weak step, $V_0<E-mc^2$	$\kappa>0$ , $0<R<1$ , $0<T<1$	Ordinary relativistic transmission: a real inside wave on the positive-energy branch. No anomaly.
Gap step, $E-mc^2<V_0<E+mc^2$	$R=1$ , $T=0$	Inside energy $E-V_0$ lies in the mass gap $(-mc^2,mc^2)$ ; the transmitted wave is evanescent — exactly the nonrelativistic Potential Step and Reflection behaviour.
Klein step, $V_0>E+mc^2$	$\kappa>0$ , $R<1$ , $T>0$ (group-velocity signing)	A negative-energy channel reopens; the transmitted wave is a positron. $R+T=1$ holds; the naive phase-velocity signing instead reports $R>1$ , $T<0$ .
$V_0\to\infty$	$T\to$ finite, not $0$	Transmission does not die as the wall grows — the starkest face of the paradox, impossible nonrelativistically.

Worked examples

Example 1 — The supercritical Dirac step (a positron gets through). Take a spin-up electron of energy $E=1.2\,mc^2$ incident on a step $V_0=3\,mc^2$ . Since $V_0=3>E+mc^2=2.2$ , this is the Klein regime. Work in units $\hbar=c=m=1$ .

Step 1. Incident momentum: $p=\sqrt{E^2-1}=\sqrt{1.44-1}=\sqrt{0.44}=\mathbf{0.663}$ .

Step 2. Inside energy: $E-V_0=1.2-3=-1.8$ , safely below $-1$ , so a propagating lower-branch wave exists. Its momentum magnitude is $|p'|=\sqrt{(E-V_0)^2-1}=\sqrt{3.24-1}=\sqrt{2.24}=1.497$ . The group-velocity prescription fixes $p'=-1.497$ .

Step 3. Flux ratio: $\kappa=\dfrac{p'}{p}\dfrac{E+1}{E-V_0+1}=\dfrac{-1.497}{0.663}\cdot\dfrac{2.2}{-0.8}=(-2.258)(-2.75)=\mathbf{6.205}$ . Positive, as the two sign flips demand.

Step 4. Coefficients: $R=\left(\dfrac{1-6.205}{1+6.205}\right)^2=\left(\dfrac{-5.205}{7.205}\right)^2=(-0.7224)^2=\mathbf{0.522}$ , and $T=\kappa\left(\dfrac{2}{1+\kappa}\right)^2=6.205\left(\dfrac{2}{7.205}\right)^2=6.205\times0.07705=\mathbf{0.478}$ .

Step 5. Conservation: $R+T=0.522+0.478=\mathbf{1.000}$ , both positive. A real fraction of the incident current — carried by a transmitted positron — enters the "forbidden" region. Transmission does not vanish, yet nothing is over-unity.

Example 2 — The nonrelativistic contrast (total reflection). Put the same under-barrier condition to a Schrödinger particle: energy $E$ below the step $V_0$ , so $E<V_0$ . There is no negative-energy sea, so no second channel.

Step 1. Region-I wavevector $k=\sqrt{2mE}/\hbar$ ; region-II is classically forbidden, wave $\sim e^{-qx}$ with $q=\sqrt{2m(V_0-E)}/\hbar$ real.

Step 2. Matching $\psi$ and $\psi'$ (second-order equation needs both) gives $r=\dfrac{k-iq}{k+iq}$ , so $R=|r|^2=\dfrac{k^2+q^2}{k^2+q^2}=\mathbf{1}$ .

Step 3. With $R=1$ and no propagating inside wave, $T=\mathbf{0}$ : total reflection, exactly as in Potential Step and Reflection. No matter how tall the Schrödinger wall, transmission is strictly zero once $E<V_0$ — the precise opposite of the Dirac step's persistent transmission.

History

The Klein paradox was born from a sign kept honest for the first time only decades after Klein wrote it down, and its interpretation tracks the birth of quantum field theory.

Origins · 1 minKlein Paradox and how this was discovered

Klein Paradox

posed 1929, resolved via QED 1930s–1990s · The prediction that a Dirac electron transmits through a supercritical electrostatic step, resolved as electron–positron pair creation by the vacuum — a foundational stress test of single-particle relativistic quantum mechanics.

The Klein paradox is one of the few "named results" in physics that is a problem rather than a person's theorem, and its lifetime maps the maturation of relativistic quantum theory. Oskar Klein wrote it down in 1929 as an apparent embarrassment; Fritz Sauter and Friedrich Hund sharpened the strong-field, pair-creating reading in the 1930s and 40s; Julian Schwinger's 1951 vacuum-decay calculation gave it a dynamical cousin; and only in 1999 did Calogeracos and Dombey settle the sign question definitively with the group-velocity prescription. Today it is taught not as a paradox but as the cleanest demonstration that relativistic wave mechanics is the one-particle shadow of a field theory — and, since 2006, as a phenomenon you can measure on a chip of graphene.

History

1929

— Oskar Klein analyses the Dirac equation on a step potential and finds transmission that does not vanish for tall barriers (Z. Phys. 53, 157)

1931

— Fritz Sauter studies the smooth-step and strong-field versions, sharpening the pair-creation reading (Z. Phys. 69, 742)

1941

— Friedrich Hund connects the effect to spontaneous emission of electron–positron pairs by strong fields

1951

— Julian Schwinger computes the vacuum pair-production rate in a constant field, the dynamical cousin of the Klein step

1999

— Calogeracos and Dombey give the definitive group-velocity resolution and history (Contemp. Phys. 40, 313)

2006

— Katsnelson, Novoselov and Geim identify perfect Klein tunnelling in graphene (Nat. Phys. 2, 620)

Going further

References. Griffiths, Introduction to Elementary Particles, 2nd ed., §7 · Sakurai, Advanced Quantum Mechanics, §3.10 · Bjorken & Drell, Relativistic Quantum Mechanics, §5.3 · Greiner, Relativistic Quantum Mechanics: Wave Equations, Ch. 8 · Calogeracos & Dombey, Contemp. Phys. 40, 313 (1999), "History and physics of the Klein paradox" · Katsnelson, Novoselov & Geim, Nat. Phys. 2, 620 (2006).

Provenance. The spinor matching gives $r=(1-\kappa)/(1+\kappa)$ , $t=2/(1+\kappa)$ with $\kappa=(p'/p)(E+m)/(E-V_0+m)$ , reproducing Bjorken–Drell §5.3, Greiner Ch. 8, and Sakurai §3.10. This article adopts the group-velocity signing of $p'$ (Calogeracos & Dombey 1999), which yields the physically labelled $R<1$ , $T>0$ , $R+T=1$ throughout, rather than the historically famous but sign-artefact " $R>1$ , $T<0$ ." Both conventions obey $R+T=1$ ; the difference is only whether the transmitted current is booked as a right-moving positron ( $T>0$ ) or a negative electron transmission ( $T<0$ ). The regime boundaries $V_0=E\pm mc^2$ and the critical field $E_c=m^2c^3/e\hbar\approx1.3\times10^{18}\,\mathrm{V/m}$ are the standard QED values; $Z\alpha\to1$ point-charge supercriticality (near $Z\approx170$ ) is textbook. All coefficients recomputed numerically ( $E=1.2\,mc^2$ , $V_0=3\,mc^2$ : $\kappa=6.205$ , $R=0.522$ , $T=0.478$ ).

Builds on: The Dirac Equation · Dirac Spinors and Plane-Wave Solutions · Negative-Energy Solutions and the Interpretive Crisis · Klein-Gordon Probability Current and Negative Densities · Potential Step and Reflection. Leads to: The Dirac Sea and Antiparticles · Charge Conjugation and the Positron · field quantization and the Schwinger effect.

Is the Klein paradox a real contradiction in quantum mechanics?

No. Nothing in the mathematics is inconsistent — current is conserved ( $R+T=1$ ) in every regime. What fails is the interpretation of the Dirac equation as a single-particle theory. The "paradox" is only paradoxical if you insist one electron goes in and demand a single-particle account of the outgoing currents. Once you allow the step to create an electron–positron pair, the bookkeeping is ordinary. The lesson is historical and conceptual: relativistic wave equations must be quantised as fields.

Where exactly does the sign convention "R>1" come from, if it's wrong?

It comes from choosing the transmitted momentum $p'>0$ by matching the phase velocity direction to the wavevector. That choice makes the flux ratio $\kappa<0$ , and then $R=\big(\tfrac{1-\kappa}{1+\kappa}\big)^2>1$ with $T=\kappa|t|^2<0$ . It is not "wrong" arithmetic — it still gives $R+T=1$ — but it mislabels a right-moving positron current as negative electron transmission. Fixing $p'$ by the group velocity ( $p'<0$ here) flips $\kappa$ positive and gives the physically transparent $R<1$ , $T>0$ .

Why doesn't the transmission die as V₀ → ∞, the way it does for a Schrödinger wall?

Because a Schrödinger wall higher than $E$ has no propagating inside channel — the wave is evanescent and $T\to0$ . A Dirac step taller than $E+mc^2$ has a propagating negative-energy channel that never closes as $V_0$ grows; if anything it becomes more available. The transmitted current settles to a finite fraction rather than vanishing. Persistent transmission through an ever-taller wall is the sharpest signature that a genuinely new (antiparticle) degree of freedom is participating.

Is the Dirac density ever negative here, like in Klein–Gordon?

No. The Dirac density $\rho=\psi^\dagger\psi$ is a sum of squared moduli and is positive-definite everywhere, as established in The Dirac Equation. The negative-density disease belongs to the Klein–Gordon equation (see Klein-Gordon Probability Current and Negative Densities). In the Klein paradox the only quantity that behaves unexpectedly is the transmission coefficient under a bad sign choice, and even that is a current, not a probability density. Particle number, not probability, is what the single-particle picture cannot hold onto.

What is the role of the mass gap in the three regimes?

The gap $2mc^2$ between the positive- and negative-energy continua defines the boundaries. If $V_0<E-mc^2$ , the inside energy stays in the upper continuum — ordinary transmission. If $E-mc^2<V_0<E+mc^2$ , the inside energy $E-V_0$ falls inside the gap, where no propagating state exists — the wave is evanescent and you get total reflection, just like nonrelativistic quantum mechanics. Only when $V_0>E+mc^2$ does $E-V_0$ drop below $-mc^2$ into the lower continuum, opening the antiparticle channel and the paradox.

How is the Klein step related to the Schwinger effect?

The Klein step is the static, sharp-boundary idealisation of vacuum pair production. Schwinger's calculation treats a constant electric field, which does work $\sim eE\,\lambda_C$ over a Compton wavelength; when that approaches $2mc^2$ the vacuum decays into pairs at a rate $\propto\exp(-\pi m^2c^3/\hbar eE)$ . The Klein step replaces the smooth field by an infinitely steep potential jump; the condition $V_0>E+mc^2$ is the same "the field can supply $2mc^2$ " statement in a different geometry. Both describe the vacuum emitting an $e^+e^-$ pair when a strong field overlaps the two continua.

Does graphene really realise the Klein paradox, or is it just an analogy?

It realises the same equation, which is what matters. Graphene's low-energy electrons obey a 2D massless Dirac equation, with the sublattice degree of freedom playing the role of spin (pseudospin) and $v_F\approx10^6\,\mathrm{m/s}$ playing the role of $c$ . A gate-defined potential step is a genuine relativistic step for these carriers, and the predicted perfect transmission at normal incidence — no mass gap, pseudospin conservation forbidding backscattering — has been observed. It is not a loose metaphor: it is the massless limit of exactly the physics analysed here, in a tabletop material.

The Klein Paradox

Where this lives today

The physical picture

The core equations

Limiting cases

Worked examples

History

Going further

Before this

Where this leads