Search for Heavy QCD Axion-like Particles at BELLE-II

Sumitted to PubDB: 2026-08-21

Category: Bachelor Thesis

Tags: Performance

Principal Authors Aman Sahoo
Date 2026-04-12
Belle II Number BELLE2-UTHESIS-2026-001
Abstract We present a Monte Carlo feasibility study for searching for heavy QCD axion-like particles (ALPs) at the Belle II experiment via the decay $B^+ \to K^+ a$, $a \to \eta\pi^+\pi^-$, $\eta \to \gamma\gamma$. Heavy QCD axions arise in extensions of the Peccei-Quinn mechanism where additional hidden-sector contributions decouple the axion mass from the canonical QCD relation, permitting masses in the MeV-GeV range while simultaneously resolving the strong CP problem and the axion quality problem. The flavor-changing neutral current transition $b \to sa$ is induced at two loops through the axion-gluon coupling, yielding an estimated branching fraction $\text{BR}(B \to Ka) \sim 10^{-6}$-$10^{-7}$ for $f_a \sim \text{TeV}$. Signal Monte Carlo samples were generated using the basf2 framework at the KEK Computing Centre across the kinematically allowed ALP mass range of $0.85$-$3$ GeV, the upper bound being imposed by dense charmonium resonances that preclude reliable signal extraction above this threshold. Standard Belle II selection criteria were applied alongside figure-of-merit-optimized cuts on photon energy ($E_\gamma > 150$ MeV) and the $\eta\gamma\gamma$ opening angle ($\cos\theta > 0$), achieving reconstruction efficiencies of approximately 10-13\% across the signal mass points. Standard Model control channels ($B \to K D^0$ and $B \to K\eta'$) sharing the same final-state topology were used to validate the reconstruction strategy. Continuum background from $e^+e^- \to q\bar{q}$ processes was suppressed using Boosted Decision Trees trained on event-shape variables including thrust, Fox-Wolfram moments, and Kakuno-Super-Fox-Wolfram moments, achieving continuum rejection rates exceeding 90\% with signal loss below 7\%. The energy difference $\Delta E$ is established as the primary fitting observable for eventual signal extraction from data.
Link to LaTeX code https://github.com/aman-sas/bachelor_thesis_latex/
Institute IISc

Files