B-meson physics & mixing

Neutral-meson oscillation is one of the strangest predictions of quantum mechanics made routine: a particle produced with a definite flavour turns, all by itself, into its own antiparticle and back, ticking at a fixed frequency. The kaons showed it first, but the heavy neutral $B$ mesons — $B^0=(\bar b d)$ , $B^0_s=(\bar b s)$ — turn the phenomenon into a precision instrument. Their mixing is driven by a single second-order weak process, the box diagram, whose two $W$ bosons and virtual top quark carry the hard-to-reach CKM element $V_{td}$ straight into an observable oscillation frequency. The $B$ system is simpler than the kaon system where it counts. Because the $B^0$ and $\bar B^0$ are heavy ( $m_B\approx5.28\ \mathrm{GeV}/c^2$ ), they scatter their decays across more than four hundred channels, almost none shared by both flavours. The absorptive part of the mixing, $\Gamma_{12}$ , which for kaons splits the $K_S$ and $K_L$ lifetimes by a factor of 570, is here negligible. What survives is the dispersive part $M_{12}$ alone — pure box diagram — so the two mass eigenstates $B_L$ and $B_H$ have essentially equal lifetimes and differ only by the tiny mass gap $\Delta m_d=2|M_{12}|$ and a phase. That mass gap is astonishingly small — $\Delta m_d\approx3.3\times10^{-4}\ \mathrm{eV}$ , one part in $10^{14}$ of the meson's rest mass — yet it is measured to sub-percent precision because it appears as a frequency. The asymmetric $e^+e^-$ B-factories BaBar and Belle, running on the $\Upsilon(4S)$ resonance, timed that oscillation, extracted $|V_{td}|$ , and used the same mixing to expose CP violation in the interference between mixing and decay — the $\sin2\beta$ measurement that confirmed the Kobayashi–Maskawa picture and won them the 2008 Nobel Prize.

Key takeaways

The neutral $B^0$ – $\bar B^0$ system mixes through a weak box diagram dominated by the virtual top quark, because the loop function grows with mass and $|V_{tb}|\approx1$ while $|V_{td}|$ is tiny.
With the absorptive part $\Gamma_{12}\approx0$ , the mass eigenstates $B_L,B_H$ have equal lifetimes and $|q/p|=1$ , splitting only by $\Delta m_d=2|M_{12}|$ ; they differ from the flavour states by the pure phase $\xi=e^{-2i\beta}$ .
A tagged $B^0$ oscillates with $P_{\text{unmix}}\propto\cos^2(\tfrac12\Delta m_d\,t)$ and $P_{\text{mix}}\propto\sin^2(\tfrac12\Delta m_d\,t)$ ; measured $\Delta m_d=(0.5065\pm0.0019)\ \mathrm{ps}^{-1}$ (period $12.4$ ps), $x_d=\Delta m_d\,\tau=0.77$ .
The strange partner mixes almost maximally, $\Delta m_s=17.77\ \mathrm{ps}^{-1}$ (35× faster, $x_s\approx27$ , $\chi_s\to\tfrac12$ ); the ratio cancels hadronic uncertainty and gives $|V_{td}/V_{ts}|=0.206$ .
CP violation shows up not in the mixing ( $|\xi|=1$ ) but in the interference between mixing and decay: $B\to J/\psi\,K_S$ measures $\sin2\beta=0.70$ .
Key idea: a $B^0$ produced as a definite flavour is a superposition of two mass eigenstates that evolve at slightly different rates; their interference makes the meson periodically reappear as its antiparticle, and the oscillation frequency $\Delta m_d=2|M_{12}|$ directly measures the magnitude of the otherwise-inaccessible CKM element $V_{td}$ .

The core insight

A $B^0$ produced as a definite flavour state is a superposition of two mass eigenstates that evolve at slightly different rates; their interference makes the meson periodically reappear as its antiparticle, and the oscillation frequency $\Delta m_d=2|M_{12}|$ directly measures the magnitude of the otherwise-inaccessible CKM element $V_{td}$ .

Where this lives today

$B$ -meson mixing has matured from discovery into a precision industry that overconstrains the Standard Model. LHCb at CERN measures the strange oscillation to a stunning $\Delta m_s=(17.7656\pm0.0057)\ \mathrm{ps}^{-1}$ — a flavour that flips 3 trillion times a second — and $\Delta m_d$ is known to sub-percent precision; together they pin one side and, through $\sin2\beta=0.699\pm0.017$ ( $\beta\approx22^\circ$ ), one angle of the unitarity triangle, whose sides and angles must all close if three-generation CKM is the whole story — and so far they do. Belle II at SuperKEKB is now accumulating tens of billions of $B\bar B$ pairs to push the other angles ( $\alpha,\gamma$ ) and hunt for tiny mixing-induced asymmetries where new heavy particles could hide in the box loop alongside the top. The same box governs the ultra-rare decay $B_s\to\mu^+\mu^-$ , measured near $3\times10^{-9}$ in agreement with prediction, and feeds the still-simmering $b\to s\ell\ell$ anomalies that may — or may not — be the first cracks in the CKM edifice. And it connects to the deepest question the portrait raises: the CKM phase these oscillations measure is the Standard Model's only established source of CP violation, yet it is orders of magnitude too small to explain why the universe is made of matter and not antimatter. $B$ mixing measures both the triumph of the Kobayashi–Maskawa mechanism and the size of the mystery it leaves unsolved.

The physical picture

A $B^0$ cannot decay into a $\bar B^0$ at first order — that would change bottom flavour by two units, and no single $W$ can do it. But at second order the weak interaction sews the transition together: two $W$ bosons and a pair of virtual up-type quarks form a closed box that turns a $(\bar b d)$ into a $(b\bar d)$ . Every up-type quark runs in the loop, but the amplitude is weighted toward the top: the loop function climbs with the internal mass, and the top's CKM couplings $V_{tb}\approx1$ , $V_{td}$ dwarf the GIM-suppressed light-quark pieces. The first figure draws that box and reads off its consequence — the off-diagonal mass $M_{12}$ and, from it, the physical splitting $\Delta m_d=2|M_{12}|$ .

Figure

The second-order box diagram driving B⁰–B̄⁰ mixing: two W bosons and a virtual up-type quark loop, overwhelmingly the top, convert a B⁰ = (b̄ d) into a B̄⁰ = (b d̄). The off-diagonal mass is M₁₂ ∝ (V_tb V_td*)² S₀(x_t), so the physical mass splitting equals twice its magnitude, Δm_d = 0.507 ps⁻¹, with x_t = m_t²/M_W² ≈ 4.6 and Inami–Lim function S₀(x_t) ≈ 2.5

Because $\Gamma_{12}\approx0$ , diagonalising the mixing Hamiltonian leaves the two mass eigenstates with the same lifetime and only a tiny mass difference. The physical content is therefore a single number, $\Delta m_d$ , seen as an oscillation: a meson tagged as $B^0$ at birth sloshes into $\bar B^0$ and back at angular frequency $\Delta m_d$ . The second figure is what a B-factory actually reconstructs — the probability of finding the same flavour versus the opposite flavour as a function of proper time — carrying the measured frequency, the $12.4$ -ps period, and the contrast with the strange partner that oscillates thirty-five times faster.

Figure

The flavour oscillation of a tagged B⁰ versus proper time t: the probability it is still a B⁰, cos²(½ Δm_d t) in blue, and the probability it has become a B̄⁰, sin²(½ Δm_d t) in amber, for Δm_d = 0.507 ps⁻¹. The period is 2π/Δm_d = 12.4 ps and the first complete flavour flip is at t = π/Δm_d = 6.2 ps; the strange partner B_s oscillates 35× faster at Δm_s = 17.77 ps⁻¹

The core equations

The whole system is two states and a $2\times2$ matrix. A neutral $B$ at rest lives in the space spanned by $|B^0\rangle$ and $|\bar B^0\rangle$ , and the weak interaction — which lets each decay and lets one turn into the other — is packaged into a non-Hermitian effective Hamiltonian, a Hermitian mass matrix $M$ minus $\tfrac{i}{2}$ times a Hermitian decay matrix $\Gamma$ .

i\,\frac{\mathrm d}{\mathrm dt}\begin{pmatrix} a \\ b \end{pmatrix} = \left(M-\frac{i}{2}\Gamma\right)\begin{pmatrix} a \\ b \end{pmatrix},\qquad M=\begin{pmatrix} M & M_{12} \\ M_{12}^* & M \end{pmatrix},\ \ \Gamma=\begin{pmatrix} \Gamma & \Gamma_{12} \\ \Gamma_{12}^* & \Gamma \end{pmatrix} \qquad (1)

where: $a,b$ — amplitudes to be $B^0,\bar B^0$ ; $M,\Gamma$ — common diagonal mass and width (equal by CPT); $M_{12}$ — dispersive off-diagonal mass from the box; $\Gamma_{12}$ — absorptive off-diagonal width from shared decay channels. The diagonal governs decay, the off-diagonal governs mixing. Diagonalising gives two eigenstates $|B_{L,H}\rangle=p|B^0\rangle\pm q|\bar B^0\rangle$ . Their complex eigenvalues fix the mass difference $\Delta m$ and the width difference $\Delta\Gamma$ in one square root; for the $B$ system the absorptive term is negligible and the expression collapses.

\Delta m-\frac{i}{2}\Delta\Gamma = 2\sqrt{\Big(M_{12}-\tfrac{i}{2}\Gamma_{12}\Big)\Big(M_{12}^*-\tfrac{i}{2}\Gamma_{12}^*\Big)} \ \xrightarrow{\ \Gamma_{12}\to0\ }\ \Delta m_d=2|M_{12}|,\quad \frac{q}{p}=e^{-2i\beta} \qquad (2)

where: $\Delta m_d>0$ — the mass splitting, the measured oscillation frequency; $\Delta\Gamma\approx0$ — equal $B_L,B_H$ lifetimes; $q/p$ — the flavour-to-mass phase, of unit modulus, set by $\arg M_{12}$ with $V_{td}=|V_{td}|e^{-i\beta}$ . Because $|q/p|=1$ there is no CP violation in the mixing itself — only a rotation by $\xi=e^{-2i\beta}$ . The magnitude of $M_{12}$ is a pure box calculation. Summing the loop with the Inami–Lim function $S_0(x_t)$ and keeping only the top gives the master formula tying the measurable $\Delta m_d$ to $|V_{td}|$ and the hadronic factor $f_B^2\hat B_B$ .

\Delta m_d=\frac{G_F^2}{6\pi^2}\,\eta_B\,m_{B_d}\,f_{B_d}^2\hat B_{B_d}\,M_W^2\,S_0(x_t)\,\bigl|V_{td}V_{tb}^*\bigr|^2 \qquad (3)

where: $G_F$ — Fermi constant; $\eta_B\approx0.55$ — QCD short-distance correction; $f_{B_d}\sqrt{\hat B_{B_d}}\approx210$ MeV — lattice decay-constant/bag factor; $S_0(x_t)$ — loop function, $x_t=m_t^2/M_W^2$ . The frequency is proportional to $|V_{td}|^2$ , so timing the oscillation weighs the CKM element. Evolving a state born as a pure $B^0$ gives the two oscillation probabilities the second figure plots — an exponential decay modulating a cosine of the mass splitting.

P_{B^0\to B^0}(t)=\tfrac12\,e^{-\Gamma t}\bigl[1+\cos(\Delta m_d\,t)\bigr],\qquad P_{B^0\to\bar B^0}(t)=\tfrac12\,e^{-\Gamma t}\bigl[1-\cos(\Delta m_d\,t)\bigr] \qquad (4)

where: $\Gamma=1/\tau$ — common decay rate; the two probabilities sum to $e^{-\Gamma t}$ , the ordinary decay law. The flavour content oscillates as $\cos^2$ and $\sin^2$ of the half-angle $\tfrac12\Delta m_d\,t$ under a decaying envelope. Two derived numbers organize the phenomenology — the dimensionless mixing parameter $x=\Delta m/\Gamma$ and the time-integrated mixing probability $\chi$ — while the ratio of the down and strange frequencies is the cleanest CKM output, because the poorly known hadronic factors nearly cancel.

x_d=\Delta m_d\,\tau,\quad \chi_d=\frac{x_d^2}{2(1+x_d^2)},\qquad \left|\frac{V_{td}}{V_{ts}}\right|^2=\frac{\Delta m_d}{\Delta m_s}\,\frac{m_{B_s}}{m_{B_d}}\,\xi_{sd}^2 \qquad (5)

where: $\chi_d$ — time-integrated mixing probability; $\xi_{sd}=f_{B_s}\sqrt{\hat B_{B_s}}/(f_{B_d}\sqrt{\hat B_{B_d}})\approx1.21$ — the $SU(3)$ -breaking ratio, close to one. Measuring both frequencies turns the hadronic uncertainty into the small, well-computed factor $\xi_{sd}$ . Finally, the physics the B-factories were built for: CP violation appears as a time-dependent asymmetry between $B^0$ and $\bar B^0$ decays to a common CP eigenstate, whose sine coefficient is a clean CKM angle.

A_{CP}(t)=\frac{\Gamma_{\bar B^0\to f}(t)-\Gamma_{B^0\to f}(t)}{\Gamma_{\bar B^0\to f}(t)+\Gamma_{B^0\to f}(t)}=S_f\sin(\Delta m_d\,t)-C_f\cos(\Delta m_d\,t),\qquad S_{J/\psi K_S}=\sin2\beta \qquad (6)

where: $S_f$ — mixing-induced CP asymmetry; $C_f$ — direct CP asymmetry (near zero for $J/\psi K_S$ ); $\beta$ — the same angle as in equation (2). The interference between decay-without-mixing and decay-after-mixing writes $\sin2\beta$ directly onto a decay-rate asymmetry.

Derivation ADiagonalising the mixing Hamiltonian and the origin of Delta m

The goal is to reduce the two-state effective Hamiltonian of the neutral- $B$ system to its physical content: the two mass eigenstates $B_L$ and $B_H$ , their mass splitting $\Delta m_d$ , and the phase relating flavour to mass. We start from the most general CPT-respecting $2\times2$ Hamiltonian, diagonalise it exactly, and then impose the $B$ -system simplification $\Gamma_{12}\approx0$ to show that the splitting collapses to $\Delta m_d=2|M_{12}|$ and that the eigenstates differ only by the pure phase $\xi=e^{-2i\beta}$ . This is where "the oscillation frequency equals twice the off-diagonal mass" is actually proven.

Step 1. Write $H=M-\tfrac{i}{2}\Gamma$ . CPT forces the diagonal entries equal, $H_{11}=H_{22}=M-\tfrac{i}{2}\Gamma$ , while the off-diagonals are $H_{12}=M_{12}-\tfrac{i}{2}\Gamma_{12}$ and $H_{21}=M_{12}^*-\tfrac{i}{2}\Gamma_{12}^*$ . Why: CPT guarantees a particle and its antiparticle have identical mass and lifetime, so all the mixing physics lives off the diagonal.

Step 2. Diagonalise. A $2\times2$ matrix with equal diagonal $D$ and off-diagonals $H_{12},H_{21}$ has eigenvalues $D\pm\sqrt{H_{12}H_{21}}$ , so $\mu_{L,H}=\big(M-\tfrac{i}{2}\Gamma\big)\mp\sqrt{H_{12}H_{21}}$ with eigenvectors $p|B^0\rangle\pm q|\bar B^0\rangle$ and $q/p=\sqrt{H_{21}/H_{12}}$ . Why: the equal diagonal makes the mixing symmetric, so the eigenstates are equal-weight superpositions up to the phase carried by $q/p$ .

Step 3. Separate real and imaginary parts. Writing $\mu_{L,H}=m_{L,H}-\tfrac{i}{2}\Gamma_{L,H}$ , the splitting is $\Delta m-\tfrac{i}{2}\Delta\Gamma=2\sqrt{H_{12}H_{21}}$ . Why: one complex square root delivers both observables — its real part is the mass gap that drives oscillation, its imaginary part the lifetime gap that drives the kaon-style $K_S/K_L$ split.

Step 4. Impose the $B$ -system limit. With $\Gamma_{12}\approx0$ the root becomes $2\sqrt{M_{12}M_{12}^*}=2|M_{12}|$ , giving $\Delta m_d=2|M_{12}|$ , $\Delta\Gamma\approx0$ , and $q/p=\sqrt{M_{12}^*/M_{12}}=e^{-i\arg M_{12}}$ of unit modulus. Since $M_{12}\propto(V_{tb}V_{td}^*)^2$ and $V_{td}=|V_{td}|e^{-i\beta}$ , this phase is $\xi=e^{-2i\beta}$ . Why: dropping $\Gamma_{12}$ is the single feature that makes the $B$ system cleaner than the kaon — equal lifetimes, no CP violation in mixing, a mass gap that is purely the box. *The key results of this derivation: the eigenvalues are $D\pm\sqrt{H_{12}H_{21}}$ ; taking $\Gamma_{12}\to0$ gives $\Delta m_d=2|M_{12}|$ and $|q/p|=1$ ; and $\arg(q/p)=-2\beta$ follows from $V_{td}=|V_{td}|e^{-i\beta}$ .*

Go deeper · Derivation B · 1 minFrom eigenstate evolution to the oscillation probability

Derivation BFrom eigenstate evolution to the oscillation probability

The aim is to turn the two mass eigenstates into the observable oscillation: starting from a meson tagged as a pure $B^0$ at $t=0$ , we propagate each eigenstate by its own phase and lifetime, project back onto the flavour basis, and square to get the probabilities $P_{\text{unmix}}\propto\cos^2$ and $P_{\text{mix}}\propto\sin^2$ that a detector measures. Integrating over the exponential decay then gives the time-integrated mixing probability $\chi_d$ . The whole calculation is quantum interference between two amplitudes of slightly different frequency — the neutral-meson analogue of a two-level beat.

Step 1. Express the flavour state in the mass basis: inverting $|B_{L,H}\rangle=p|B^0\rangle\pm q|\bar B^0\rangle$ gives $|B^0\rangle=\tfrac{1}{2p}(|B_L\rangle+|B_H\rangle)$ . Why: only the mass eigenstates have simple evolution — each carries a single phase $e^{-imt}$ and decay $e^{-\Gamma t/2}$ — so we resolve the flavour state into them first.

Step 2. Evolve each eigenstate with $e^{-im_{L,H}t}e^{-\Gamma t/2}$ (equal widths), giving $|B^0(t)\rangle=\tfrac12 e^{-\Gamma t/2}(g_+|B^0\rangle+\tfrac{q}{p}g_-|\bar B^0\rangle)$ with $g_\pm=e^{-im_L t}\pm e^{-im_H t}$ . Why: the two eigenstates drift out of phase at rate $m_H-m_L=\Delta m_d$ ; that relative phase is the entire source of oscillation.

Step 3. Square the amplitudes. Using $|g_+|^2\propto[1+\cos\Delta m_d t]$ , $|g_-|^2\propto[1-\cos\Delta m_d t]$ , and $|q/p|=1$ : $P_{B^0\to B^0}=\tfrac12 e^{-\Gamma t}[1+\cos\Delta m_d t]$ and $P_{B^0\to\bar B^0}=\tfrac12 e^{-\Gamma t}[1-\cos\Delta m_d t]$ . Why: the cross term between the two eigenstate phases is the $\cos\Delta m_d t$ — interference, not classical transition, makes the flavour reappear.

Step 4. Integrate over time: $\chi_d=\int_0^\infty P_{\text{mix}}\,\mathrm dt/\int_0^\infty(P_{\text{mix}}+P_{\text{unmix}})\,\mathrm dt$ . With $\int_0^\infty e^{-\Gamma t}\cos\Delta m_d t\,\mathrm dt=\Gamma/(\Gamma^2+\Delta m_d^2)$ this gives $\chi_d=x_d^2/[2(1+x_d^2)]$ , $x_d=\Delta m_d/\Gamma$ . Why: the observable in a time-blind experiment is $\chi_d$ , saturating at $\tfrac12$ for fast mixing ( $x\gg1$ , the $B_s$ case) and small for slow mixing. *The key results of this derivation: $P_{\text{unmix}}+P_{\text{mix}}=e^{-\Gamma t}$ (probabilities sum to the ordinary decay law); $\chi_d=x_d^2/[2(1+x_d^2)]=0.186$ at $x_d=0.769$ ; and $\chi\to\tfrac12$ as $x\to\infty$ .*

Go deeper · Derivation C · 1 minThe ratio method that turns two frequencies into V_td over V_ts

Derivation CThe ratio method that turns two frequencies into V_td over V_ts

The goal is to show how measuring both $\Delta m_d$ and $\Delta m_s$ yields the CKM ratio $|V_{td}/V_{ts}|$ with far smaller theoretical error than either frequency alone can give $|V_{td}|$ . The obstacle is the hadronic factor $f_B^2\hat B_B$ in equation (3), known from lattice QCD to only a few percent; the trick is that dividing the two frequencies turns it into a ratio of very similar quantities that $SU(3)$ flavour symmetry makes close to one and computable to high accuracy. This is why the $B_s$ measurement was so eagerly awaited — it sharpened one whole side of the unitarity triangle.

Step 1. Write the master formula for both systems: $\Delta m_s\propto m_{B_s}f_{B_s}^2\hat B_{B_s}|V_{ts}V_{tb}^*|^2$ and $\Delta m_d\propto m_{B_d}f_{B_d}^2\hat B_{B_d}|V_{td}V_{tb}^*|^2$ , with the same $G_F^2M_W^2\eta_B S_0(x_t)/6\pi^2$ prefactor. Why: the box is identical apart from the spectator quark ( $d$ versus $s$ ) and the CKM element, so the common electroweak physics divides out exactly.

Step 2. Take the ratio; the prefactor, $|V_{tb}|^2$ and $S_0(x_t)$ all cancel: $\dfrac{\Delta m_s}{\Delta m_d}=\dfrac{m_{B_s}}{m_{B_d}}\,\dfrac{f_{B_s}^2\hat B_{B_s}}{f_{B_d}^2\hat B_{B_d}}\,\dfrac{|V_{ts}|^2}{|V_{td}|^2}$ , with $\xi_{sd}^2\equiv f_{B_s}^2\hat B_{B_s}/f_{B_d}^2\hat B_{B_d}$ . Why: the top-loop function and $W$ mass, the least certain inputs, appear identically in both and vanish in the quotient.

Step 3. Solve for the CKM ratio: $\left|\dfrac{V_{td}}{V_{ts}}\right|^2=\dfrac{\Delta m_d}{\Delta m_s}\dfrac{m_{B_s}}{m_{B_d}}\xi_{sd}^2$ . In the $SU(3)$ limit $\xi_{sd}=1$ ; the lattice value is $\xi_{sd}=1.21\pm0.03$ . Why: replacing an absolute factor known to $\sim5\%$ by a symmetry-protected ratio known to $\sim2\%$ is the entire gain.

Step 4. Insert numbers: with $\Delta m_d=0.5065$ , $\Delta m_s=17.77\ \mathrm{ps}^{-1}$ , $m_{B_s}/m_{B_d}=1.017$ , $\xi_{sd}=1.21$ , $|V_{td}/V_{ts}|^2=(0.02850)(1.017)(1.464)=0.0424$ , so $|V_{td}/V_{ts}|=0.206$ . Why: this fixes one side of the unitarity triangle and, with $|V_{ts}|\approx|V_{cb}|=0.041$ , gives $|V_{td}|\approx8.3\times10^{-3}$ — far less model-dependent than the direct extraction. *The key results of this derivation: $|V_{td}/V_{ts}|^2=(\Delta m_d/\Delta m_s)(m_{B_s}/m_{B_d})\xi_{sd}^2=0.0424$ ; $|V_{td}/V_{ts}|=0.206$ ; $|V_{td}|=8.3\times10^{-3}$ .*

Limiting cases

Case	Result	Meaning
$\Gamma_{12}\to0$ ( $B_d$ system)	$\Delta\Gamma\approx0$ , $\|q/p\|=1$	equal lifetimes for $B_L,B_H$ ; no CP violation in mixing
Kaon system (contrast)	$\Gamma_{12}$ comparable to $M_{12}$	huge lifetime split $\tau_{K_L}/\tau_{K_S}\approx570$ ; several quarks contribute
$x_d\sim1$ ( $\Delta m_d\,\tau=0.77$ )	partial oscillation per lifetime	most $B_d$ decay before a full flip; $\chi_d=0.186$
$x_s\gg1$ ( $\Delta m_s\,\tau\approx27$ )	many oscillations per lifetime	$B_s$ mixes almost maximally, $\chi_s\to\tfrac12$ ; needs ps resolution
Top-loop dominance	$S_0(x_t)\propto x_t$ at large $x_t$	GIM broken by the heavy top; $\Delta m$ measures $\|V_{td}\|$ (or $\|V_{ts}\|$ )
$\Gamma_{12}$ small but finite ( $B_s$ )	tiny $\Delta\Gamma_s\approx0.08\ \mathrm{ps}^{-1}$	measurable at LHCb; a rare window on the absorptive part

Common trapAssuming B mixing reveals CP violation the way kaon mixing does

Because the interference term $\Gamma_{12}$ is negligible, the mixing phase has unit modulus: $|\xi|=|e^{-2i\beta}|=1$ exactly. That means $P(B^0\to\bar B^0)\approx P(\bar B^0\to B^0)$ — the forward and backward mixing rates are equal — so CP violation in the mixing itself is minuscule (the semileptonic asymmetry $a_{sl}$ is below $10^{-3}$ ). The large, celebrated CP violation of the $B$ system lives instead in the interference between two paths to a common final state: a $B^0$ decaying directly to $J/\psi K_S$ , and a $B^0$ that first oscillates into a $\bar B^0$ and then decays. That interference carries the phase $2\beta$ and shows up as the time-dependent asymmetry $S_{J/\psi K_S}=\sin2\beta$ of equation (6). So the box supplies the clock ( $\Delta m_d$ ) and the phase ( $\xi$ ), but you read the phase by watching decays interfere — never by comparing forward and backward mixing rates, which are equal.

Worked examples

1 · The oscillation clock: x_d, chi_d, and the period (fully worked)

Given $\Delta m_d=0.5065\ \mathrm{ps}^{-1}$ and the $B^0$ lifetime $\tau=1.519$ ps, find the mixing parameter $x_d$ , the time-integrated mixing probability $\chi_d$ , the oscillation period, and the time of the first complete flavour flip.

Step 1. The mixing parameter compares oscillation frequency to decay rate: $x_d=\Delta m_d\,\tau=(0.5065)(1.519)=\mathbf{0.769}$ . Since $x_d<1$ , the meson typically decays before completing even one radian of oscillation — the down system is a slow mixer.

Step 2. From equation (5): $\chi_d=\dfrac{x_d^2}{2(1+x_d^2)}=\dfrac{0.592}{2(1.592)}=\mathbf{0.186}$ . So $18.6\%$ of neutral $B$ 's tagged as $B^0$ at birth decay as $\bar B^0$ — the ARGUS-era number that first revealed the mixing is large.

Step 3. The period is $T=2\pi/\Delta m_d=6.283/0.5065=\mathbf{12.4\ ps}$ , over eight lifetimes. The first moment the meson is certainly a $\bar B^0$ (first maximum of $\sin^2\tfrac12\Delta m_d t$ ) is at $\tfrac12\Delta m_d t=\tfrac\pi2$ , i.e. $t=\pi/\Delta m_d=\mathbf{6.20\ ps}\approx4.1\tau$ . x_d = 0.77, chi_d = 0.186, period 12.4 ps, first flip at 6.2 ps.

2 · Extracting |V_td / V_ts| from the two frequencies (fully worked)

Using $\Delta m_d=0.5065\ \mathrm{ps}^{-1}$ , $\Delta m_s=17.77\ \mathrm{ps}^{-1}$ , masses $m_{B_d}=5.280$ , $m_{B_s}=5.367$ GeV, and $\xi_{sd}=1.21$ , determine $|V_{td}/V_{ts}|$ and hence $|V_{td}|$ .

Step 1. Form the frequency ratio: $\dfrac{\Delta m_d}{\Delta m_s}=\dfrac{0.5065}{17.77}=\mathbf{0.02850}$ . The strange system oscillates $35\times$ faster — a direct reflection of $|V_{ts}|\gg|V_{td}|$ .

Step 2. Apply the ratio formula: $\left|\dfrac{V_{td}}{V_{ts}}\right|^2=(0.02850)(1.017)(1.464)=\mathbf{0.0424}$ , so $\left|\dfrac{V_{td}}{V_{ts}}\right|=\mathbf{0.206}$ .

Step 3. With $|V_{ts}|\approx|V_{cb}|=0.041$ , $|V_{td}|=0.206\times0.041=\mathbf{8.4\times10^{-3}}$ , matching BaBar and Belle's $(8.4\pm0.6)\times10^{-3}$ but now carrying only the $\sim2\%$ error of $\xi_{sd}$ rather than the $\sim5\%$ of an absolute $f_B^2\hat B_B$ . |V_td/V_ts| = 0.206, giving |V_td| = 8.4×10⁻³ from the theoretically clean ratio.

3 · Why the strange partner mixes maximally

For the $B_s$ system, $\Delta m_s=17.77\ \mathrm{ps}^{-1}$ and $\tau_{B_s}=1.51$ ps. Find $x_s$ and $\chi_s$ .

Step 1. $x_s=\Delta m_s\,\tau_{B_s}=(17.77)(1.51)=\mathbf{26.8}$ — the $B_s$ completes about $27/2\pi\approx4$ full oscillations within one lifetime, the fastest flavour oscillation ever resolved.

Step 2. $\chi_s=\dfrac{x_s^2}{2(1+x_s^2)}=\dfrac{719}{2(720)}=\mathbf{0.4993}\approx\tfrac12$ . A B_s is essentially as likely to decay as a B̄_s as itself — maximal mixing — which is why resolving Δm_s demanded picosecond vertexing and defeated every experiment until CDF and then LHCb.

Physics of the portrait

$B$ -meson mixing is the experimental stage on which the Kobayashi–Maskawa idea was proved. Two theorists, writing in 1973 when only three quarks were known, realised that a third generation would give the quark-mixing matrix an irremovable complex phase — and that this phase would show up, decades later, in the oscillations and decays of the $B$ mesons.

Origins · 1 minKobayashi and Maskawa and how this was discovered

Kobayashi and Maskawa

Kobayashi b. 1944 · Maskawa 1940–2021 · the CKM origin of CP violation

In 1973, Makoto Kobayashi and Toshihide Maskawa, working at Kyoto University, asked how the Standard Model could accommodate the CP violation that Cronin and Fitch had discovered in kaons in 1964. They proved a theorem of counting. With two generations of quarks the Cabibbo mixing matrix is a single real rotation angle — no complex phase, and therefore no CP violation possible from mixing. With three generations the $3\times3$ unitary matrix has four physical parameters: three angles and one irreducible complex phase — exactly what CP violation requires. It was audacious: at the time only up, down, and strange were known; charm was confirmed a year later, the bottom in 1977, the top only in 1995. Their matrix, now the Cabibbo–Kobayashi–Maskawa (CKM) matrix, predicted CP violation wherever all three generations interfere — and the cleanest such place turned out to be the $B$ system. The whole apparatus of this article is the CKM phase made visible: the box's $M_{12}\propto(V_{tb}V_{td}^*)^2$ carries the phase into the mixing, and the interference of $B^0\to J/\psi K_S$ with and without mixing reads it out as $\sin2\beta$ . When BaBar and Belle measured a large, nonzero $\sin2\beta$ in 2001, they confirmed that a single number written down in 1973 governs CP violation across the quark sector. The two shared the 2008 Nobel Prize in Physics.

History

1973

— Kobayashi and Maskawa show a third quark generation lets the CKM matrix carry a CP-violating phase

1987

— ARGUS at DESY discovers large B⁰–B̄⁰ mixing, evidence for an unexpectedly heavy top quark

2001

— BaBar and Belle observe CP violation in the B system, measuring sin2β

2006

— CDF resolves the fast B_s oscillation frequency Δm_s

2008

— Kobayashi and Maskawa share the Nobel Prize in Physics

Going further

Texts. Thomson, Modern Particle Physics, §14.6 (B-meson physics), §14.6.1–14.6.4 (mixing, oscillations, BaBar and Belle, CP violation) — the source this article grounds · Griffiths, Introduction to Elementary Particles, 2nd ed., §4.4 and §12.3 (neutral-meson mixing and CP violation) · Halzen & Martin, Quarks and Leptons, §12 (weak interactions and the CKM matrix) · Perkins, Introduction to High Energy Physics, 4th ed., Ch. 11 · Bigi & Sanda, CP Violation, Chs. 6–10; PDG Review of Particle Physics. The key numerical anchors for this article are: $\Delta m_d=2|M_{12}|$ and $|q/p|=1$ as $\Gamma_{12}\to0$ ; $x_d=\Delta m_d\tau=0.769$ , $\chi_d=0.186$ ; $|V_{td}/V_{ts}|^2=(\Delta m_d/\Delta m_s)(m_{B_s}/m_{B_d})\xi_{sd}^2=0.042$ ; $x_s=\Delta m_s\tau_{B_s}=26.8$ , $\chi_s\to\tfrac12$ ; $S_0(x_t)\approx2.3$ – $2.5$ at $x_t\approx4.6$ . Constants used throughout: $\Delta m_d=0.5065\ \mathrm{ps}^{-1}$ , $\Delta m_s=17.77\ \mathrm{ps}^{-1}$ , $\sin2\beta=0.70$ , $\beta\approx22^\circ$ .

Prerequisites. The Neutral Kaon System · The CKM Matrix · Weak Interactions of Quarks · Strangeness Oscillations. Leads to. CP Violation in the Standard Model · The Matter–Antimatter Asymmetry · CP Violation in Kaons.

Ask the tutor: Ask me to diagonalise the mixing Hamiltonian and pull out $\Delta m_d=2|M_{12}|$ step by step, or to reproduce the $\sin2\beta$ measurement from the interference of $B^0\to J/\psi K_S$ with and without mixing.

Why does the heaviest, rarest quark dominate the mixing loop?

It looks paradoxical, but two effects both point to the top. First, the box is a virtual loop, so the top need not be produced on shell — it only has to circulate, which heavy virtual particles are perfectly allowed to do. Second, the Inami–Lim loop function $S_0(x_t)$ grows with the internal quark mass (roughly linearly at large $x_t=m_t^2/M_W^2\approx4.6$ ), because the GIM cancellation that suppresses light quarks is badly broken by the huge top mass. Add the CKM factors — the top vertices carry $V_{tb}\approx1$ , and the light-quark alternatives carry more suppressive combinations — and $M_{12}\propto(V_{tb}V_{td}^*)^2 S_0(x_t)$ is effectively pure top, precisely opposite to the kaon system where several flavours contribute comparably.

How does B mixing differ from kaon mixing?

Both are two-state oscillations from a $2\times2$ effective Hamiltonian, but three things differ. Which quark drives it: the top dominates the $B$ box, whereas charm and up contribute comparably in the kaon box. The absorptive part: the heavy $B^0,\bar B^0$ share almost no decay channels, so $\Gamma_{12}\approx0$ and $B_L,B_H$ have equal lifetimes; the kaons share the dominant $\pi\pi$ and $\pi\pi\pi$ channels, so $\Gamma_{12}$ is large and $K_S,K_L$ differ in lifetime by a factor of 570. Where CP violation shows: for kaons the small $|q/p|\ne1$ produces $\varepsilon_K$ in mixing, while for $B$ the phase is essentially pure ( $|\xi|=1$ ) and the large CP violation surfaces in the interference between mixing and decay. The $B$ system is cleaner precisely because $\Gamma_{12}\approx0$ .

Why does Δm_d measure V_td and not V_tb, when both appear at the vertices?

Both appear — $M_{12}\propto(V_{tb}V_{td}^*)^2$ — but $|V_{tb}|$ is already known to be extremely close to $1$ from three-generation unitarity, so $|V_{tb}|^2\approx1$ is effectively a constant. The only genuinely unknown CKM magnitude in the down-system box is $|V_{td}|$ , and timing the oscillation gives $\Delta m_d=2|M_{12}|\propto|V_{td}|^2$ , from which $|V_{td}|$ is extracted given the hadronic factor $f_B^2\hat B_B$ . The element $|V_{td}|$ is otherwise very hard to reach — it governs no clean tree-level decay — so this second-order loop is one of the few windows onto it. In the strange system the same logic isolates $|V_{ts}|$ from $\Delta m_s$ .

What is "flavour tagging," and how did BaBar and Belle do it?

To see an oscillation you must know the meson's flavour at two times: at birth and at decay. The B-factories produced $B\bar B$ pairs from the $\Upsilon(4S)$ in a coherent state, so the two mesons stay perfectly anti-correlated until one decays. When one decays in a flavour-specific way — most cleanly semileptonically, where a positive lepton tags a $B^0$ and a negative one a $\bar B^0$ — it instantly tags the other meson as the opposite flavour at that instant. That second meson evolves and decays a measurable time later; comparing its decay flavour (or its decay to a CP eigenstate) against the tag reveals whether it oscillated. Crucially the beams were asymmetric in energy (for PEP-II, $9\ \mathrm{GeV}\,e^-$ on $3.1\ \mathrm{GeV}\,e^+$ ), so the $B$ 's are boosted and travel a resolvable $\sim250\ \mu$ m before decaying.

Is CP violation in B mixing the source of the matter–antimatter asymmetry?

It is the same mechanism — the CKM phase — but far too small to be the whole answer. The Kobayashi–Maskawa phase that $\sin2\beta$ measures is genuine, established CP violation, and any baryogenesis scenario must include it. But quantitative estimates show the CKM contribution, filtered through the required out-of-equilibrium dynamics of the early universe, falls short of the observed baryon-to-photon ratio by roughly ten orders of magnitude. So $B$ mixing confirms that CP violation exists in the quark sector while proving the Standard Model alone cannot account for our matter-dominated universe. That gap is one of the strongest arguments for new sources of CP violation — in the lepton sector or beyond the Standard Model — and is exactly what precision $B$ -physics hunts for in the loop.

What does the angle β actually mean physically?

The angle $\beta$ is one corner of the unitarity triangle, a geometric picture of the CKM relation $V_{ud}V_{ub}^*+V_{cd}V_{cb}^*+V_{td}V_{tb}^*=0$ . Because this is a sum of three complex numbers equal to zero, plotting them tip-to-tail closes a triangle in the complex plane, and $\beta$ is the angle at the vertex opposite the side built from $V_{td}$ ; in the standard convention $V_{td}=|V_{td}|e^{-i\beta}$ . That same phase is what the box stamps onto the mixing, $\xi=e^{-2i\beta}$ , so when a $B^0$ oscillates and then decays to a CP eigenstate like $J/\psi K_S$ , the interference between mixed and unmixed paths oscillates as $\sin2\beta$ in time. Measuring $\sin2\beta=0.70$ (hence $\beta\approx22^\circ$ ) therefore reads a fundamental phase of the CKM matrix straight off a decay-rate asymmetry.