Several $B$ and $B_s$ decays have been observed that have been cited as evidence for exchange ($E$), penguin annihilation ($PA$), and annihilation ($A$) processes, such as $\overline{b}d\to \overline{u}u$, $\overline{b}s\to \overline{u}u$, and $\overline{b}u\to W^{*}\to \overline{c}s$, respectively. These amplitudes are normally thought to be suppressed, as they involve the spectator quark in the weak interaction and thus should be proportional to the $B$-meson decay constant $f_B$. However, as pointed out a number of years ago, they can also be generated by rescattering from processes whose amplitudes do not involve $f_B$, such as color-favored tree amplitudes. In this paper we investigate a number of processes such as $B^0\to K^{+}{K}^{-}$, $B_s\to {\pi }^{+}{\pi }^{-}$, and $B^{+}\to D_s^{+}\varphi$, and identify promising states from which they can be generated by rescattering. We find that $E$-and $PA$-type processes are characterized, respectively, by amplitudes ranging from 5% to 10% and from 15% to 20% with respect to the largest amplitude from which they can rescatter. Based on this regularity, using approximate flavor SU(3) symmetry in some cases and time-reversal invariance in others, we predict the branching fractions for a large number of as-yet-unseen $B$ and $B_s$ decays in an extensive range from order ${10}^{-9}$ to ${10}^{-4}$.

• Received 3 December 2012

DOI:https://doi.org/10.1103/PhysRevD.87.036008