---
title: Invisible K_S⁰ Decays & BSM Searches
url: https://www.emergentmind.com/topics/invisible-decays-of-k_s-0
type: topic
---

# Invisible K_S⁰ Decays & BSM Searches

Searching arXiv for relevant papers on invisible decays of neutral kaons and related theoretical/experimental context.
Invisible decays of \(K_S^0\) denote decay or disappearance processes in which the short-lived neutral kaon produces no detectable final-state activity in an ordinary detector. In the Standard Model, the benchmark invisible channel \(K_S^0\to \nu\bar\nu\) is extraordinarily suppressed: for massless neutrinos it is kinematically forbidden by angular momentum conservation, while for massive neutrinos it remains helicity suppressed, with a predicted branching fraction below \(10^{-16}\) [2501.06426]. Consequently, any experimentally observable \(K_S^0\) invisible signal at current sensitivities would indicate physics beyond the Standard Model [2501.06426, 1409.2288]. In recent literature, the topic has developed along two complementary lines: a model-specific mirror-matter prediction of an invisible branching fraction \(B_\text{inv}(K^0_S)\simeq 1.8\times 10^{-6}\) [2006.10746], and the first direct experimental search, which found no significant signal and set \(\mathcal{B}(K_S^0\to \text{invisible})<8.4\times 10^{-4}\) at \(90\%\) confidence level [2501.06426].

## 1. Definition and conceptual scope

“Invisible” is used in two distinct but related senses in the literature. Experimentally, it refers to decays whose products generate no reconstructed charged tracks and no significant neutral energy deposits attributable to the kaon [2501.06426]. In this operational sense, the decay is inferred from kinematic recoil or tagging, rather than by direct observation of final-state particles.

The theoretical meaning depends on the underlying model. In generic new-physics searches, the invisible final state may consist of neutrinos, dark-sector states, or other neutral weakly interacting particles [2501.06426, 1409.2288]. By contrast, in the mirror-matter framework of “Invisible decays of neutral hadrons” [2006.10746], an “invisible decay” of \(K^0_S\) is not a Standard Model neutrino mode. It is the disappearance of an ordinary neutral hadron through oscillation into its mirror counterpart, after which the mirror hadron decays entirely within the mirror sector and is therefore undetected in an ordinary detector [2006.10746].

The literature also uses two notational conventions. The mirror-oscillation paper consistently writes \(K^0_S\) [2006.10746], whereas the BESIII experimental search uses \(K_S^0\) [2501.06426]. Both designate the short-lived neutral kaon state.

## 2. Standard-Model suppression and beyond-Standard-Model motivation

The Standard-Model expectation for invisible neutral-meson decays is negligible on experimental scales. For the family \(M^0=\pi^0,\eta,\eta',K_S,K_L\), the decay rate to neutrinos is parametrically suppressed as
\[
\Gamma(M^0\to \nu\overline{\nu}) \sim \Bigl(\frac{m_\nu}{m_{M^0}}\Bigr)^2 \lesssim 10^{-16},
\]
as quoted in the SPS proposal paper [1409.2288]. The BESIII search states specifically that for \(K_S^0\to \nu\bar\nu\), massless neutrinos are excluded by angular momentum conservation and massive-neutrino contributions remain below \(10^{-16}\) [2501.06426].

This extreme suppression gives invisible \(K_S^0\) decays unusual discovery value. The search is sensitive to any process yielding no visible detector activity, including dark-sector or feebly interacting particles in generic invisible-decay frameworks [2501.06426]. The SPS proposal further cites possibilities such as a new gauge boson with nonuniversal couplings to quarks or transitions into a hidden sector, while deliberately keeping the experimental target broad rather than tied to a single model [1409.2288].

A second motivation is neutral-kaon precision phenomenology. The Bell–Steinberger relation connects CP and CPT violation in the \(K^0-\overline{K}^0\) system to amplitudes of all decay modes [1409.2288]. Since invisible \(K_S\) and \(K_L\) channels had not been directly tested when that proposal was written, they represented a potential loophole in high-precision Bell–Steinberger analyses [1409.2288]. The BESIII result is therefore relevant not only as a null search for new physics but also as a direct empirical input to CPT tests [2501.06426].

## 3. Mirror-matter interpretation and the \(K^0_S\) branching-fraction prediction

The most explicit quantitative prediction for invisible \(K^0_S\) decays in the supplied literature is given in “Invisible decays of neutral hadrons” [2006.10746]. In that framework, ordinary neutral hadrons can oscillate into mirror partners because spontaneous mirror-symmetry breaking induces tiny ordinary-mirror mixing. For a neutral particle in vacuum, neglecting decay or on timescales short relative to the lifetime, the oscillation probability is written as
\[
P(t)=\sin^2(2\theta)\sin^2\!\left(\frac{1}{2}\Delta t\right),
\]
where \(\theta\) is the ordinary-mirror mixing angle and \(\Delta\) is the mass splitting [2006.10746].

For decaying neutral hadrons, the oscillation probability is averaged over the exponential decay law. The paper gives
\[
\lambda_{hh'}=\frac{1}{\tau^2}\int_0^{\infty}P(t)\exp(-t/\tau)\,dt
\]
and hence
\[
\lambda_{hh'}=\frac{1}{2\tau}\sin^2(2\theta)\frac{(\Delta\tau)^2}{1+(\Delta\tau)^2},
\]
from which the invisible branching ratio follows as
\[
B_\text{inv}=\frac{1}{2}\sin^2(2\theta)\frac{(\Delta\tau)^2}{1+(\Delta\tau)^2}.
\]
This lifetime dependence is central to the kaon case, because \(K^0_S\) and \(K^0_L\) are assigned the same kaon-sector oscillation parameters in the model, so the difference in predicted invisible branching fractions arises primarily from their very different lifetimes [2006.10746].

The decisive model assumption is stated explicitly as the equivalence of the \(CP\) violation and mirror symmetry breaking scales for neutral kaon oscillations [2006.10746]. It is implemented as
\[
\Delta_{K^{0}K^{0'}} = \Delta_{K^{0}_LK^{0}_S} = 3.484(6)\times 10^{-6}\ \text{eV}
\]
and
\[
\sin\theta_{K^{0}K^{0'}} = \sin\theta_{K^{0}_LK^{0}_S} = 2.228(11)\times10^{-3},
\]
corresponding to
\[
\sin^2(2\theta_{K^0K^{0'}})\simeq 2\times 10^{-5}.
\]
With the \(K^0_S\) lifetime
\[
\tau(K^0_S)=8.954\times 10^{-11}\ \text{s},
\]
the paper obtains
\[
B_\text{inv}(K^0_S)\simeq 1.8\times 10^{-6}.
\]
For comparison, the same framework gives
\[
B_\text{inv}(K^0_L)\simeq 9.9\times 10^{-6},
\]
with the larger value attributable to
\[
\tau(K^0_L)=5.116\times 10^{-8}\ \text{s},
\]
which allows the oscillation to develop more fully during free propagation [2006.10746].

The same paper embeds this calculation in a broader mirror-matter scheme containing a universal relative mass-splitting scale
\[
\frac{\Delta}{m}\simeq 7\times 10^{-15},
\]
and a quark-level mixing structure encoded in an extended quark mixing matrix with cross-sector elements \(V_{qq'}\) [2006.10746]. The hadron-level mixing strength is estimated by
\[
\sin^2(2\theta)\simeq \prod_i |2V_{q_iq'_i}|^2,
\]
and for kaons specifically,
\[
\sin^2(2\theta_{K^0K^{0'}})=|2V_{dd'}|^2|2V_{ss'}|^2.
\]
Using \( |V_{dd'}| \simeq 0.063\) and \( |V_{ss'}| \simeq 0.018\), the model recovers the kaon mixing strength \(\simeq 2\times10^{-5}\) [2006.10746].

A plausible implication is that the quoted \(K^0_S\) prediction is numerically precise only within the internal logic of that mirror-breaking ansatz. The same paper emphasizes that the result depends on the assumed identification of mirror-breaking parameters with the measured neutral-kaon \(CP\)-violating scale, on the extension of \(\Delta/m\simeq 7\times10^{-15}\) across neutral hadrons, and on a quark-factorized estimate of hadron mixing rather than a derivation from QCD [2006.10746].

## 4. Bell–Steinberger relation and indirect bounds

The Bell–Steinberger relation is a central theoretical reason why invisible \(K_S\) decays matter even in the absence of a positive signal. In the Wigner–Weisskopf approximation, the neutral-kaon time evolution is written as
\[
i\frac{d\Phi(t)}{dt}=H\Phi(t)=\Bigl(M-i\frac{i}{c}\Gamma\Bigr)\Phi(t),
\]
with \(M\) and \(\Gamma\) as \(2\times2\) Hermitian matrices in \(K^0-\overline{K}^0\) space [1409.2288]. The Bell–Steinberger relation given in the proposal is
\[
\Bigl(\frac{\Gamma_S+\Gamma_L}{\Gamma_S-\Gamma_L}+i\rm{tan}\phi_{SW}\Bigr)\Bigl(\frac{Re(\epsilon)}{1+|\epsilon|^2}-i\rm{Im}(\delta)\Bigr)
=\frac{1}{\Gamma_S-\Gamma_L}\sum_F A_L(f)A_S^*(f),
\]
where the sum runs over all accessible final states [1409.2288].

The importance of invisible modes follows directly: if \(K_S\) or \(K_L\) has an unmeasured invisible branching fraction, it can contribute to the right-hand side of the Bell–Steinberger relation and thereby affect CPT/CP analyses [1409.2288]. The 2014 proposal inferred indirect constraints from the sums of visible branching fractions:
\[
\sum_i Br_i(K_S \to visible) = 1.00068 \pm 0.00048,
\]
\[
\sum_i Br_i(K_L \to visible) = 1.00032 \pm 0.00058,
\]
leading to
\[
Br(K_S \to invisible) < 1.1 \times 10^{-4}\qquad (95\%~\text{C.L.}),
\]
\[
Br(K_L \to invisible) < 6.3 \times 10^{-4}\qquad (95\%~\text{C.L.}).
\]
These are indirect allowances rather than direct search bounds [1409.2288].

For the invisible contribution to Bell–Steinberger, the same paper uses
\[
|\alpha_{inv}|^2=\tau_{K_S}\tau_{K_L}\,Br(K_L\to invisible)Br(K_S\to invisible),
\]
with
\[
\tau_{K_S}=0.08958\pm 0.00006~\text{ns},\qquad \tau_{K_L}=51.16\pm 0.2~\text{ns},
\]
to obtain
\[
|\alpha_{inv}| < 2.8 \times 10^{-5}\qquad (95\%~\text{C.L.}).
\]
The paper argues that channels with
\[
Br(K_S \to f)\gtrsim 10^{-5}
\]
and
\[
Br(K_L \to f)\Gamma_L/\Gamma_S \gtrsim 10^{-5}
\]
are within current Bell–Steinberger precision, so the indirect \(K_S\) invisible bound at \(1.1\times10^{-4}\) was not yet sufficiently strong [1409.2288]. This is why direct searches were motivated independently of any specific BSM model.

The BESIII search later noted that summing known \(K_S^0\) decay modes leaves an indirect allowance at the level of \(\mathcal{O}(10^{-4})\), reinforcing the need for a direct constraint [2501.06426]. This continuity between the 2014 conceptual argument and the 2025 measurement is one of the notable features of the subject.

## 5. Experimental strategies for direct searches

Two distinct direct-search strategies appear in the literature.

The first, proposed for CERN SPS, is a fixed-target missing-energy experiment using charge exchange,
\[
K^- + p \to \overline{K}^0 + n,\qquad K^+ + n \to K^0 + p,
\]
to produce a forward neutral kaon carrying nearly the full beam energy [1409.2288]. The signal is the complete disappearance of the beam energy after a tagged interaction in the target, summarized as
\[
S_{K^0}={\rm T\cdot \overline{V1\cdot V2\cdot HCAL}},
\]
with an apparatus consisting of beam counters, an upstream momentum spectrometer, an active target, an ECAL veto, forward veto counters, a decay volume, and a hermetic HCAL of total thickness \(\simeq 28\lambda_{int}\) [1409.2288]. With \(10^{12}\) incident kaons, a background-free assumption, \(\zeta\simeq0.9\), \(\epsilon_{tag}\gtrsim90\%\), and \(n^{inv}_{90\%}=2.3\) for a 90% C.L. no-signal upper limit, the projected sensitivity was
\[
Br(K_S\to \text{invisible})\lesssim 10^{-8},
\]
quoted under the one-at-a-time assumption that \(n^{inv}_{K_L}=0\) when deriving the \(K_S\) reach [1409.2288].

The second strategy, and the first actually realized, is the BESIII recoil-tagging method in
\[
J/\psi \to \phi K_S^0 K_S^0
\]
using one visible \(K_S^0\to\pi^+\pi^-\) decay as a tag and testing the recoiling kaon for invisible decay [2501.06426]. The analysis uses
\[
(1.0087\pm0.0044)\times10^{10}
\]
\(J/\psi\) events collected with the BESIII detector at BEPCII [2501.06426]. The mode is chosen because a dominant would-be background,
\[
J/\psi \to \phi K_S^0 K_L^0,
\]
is forbidden by \(C\)-parity conservation [2501.06426].

The signal candidate is defined by a reconstructed \(\phi\to K^+K^-\), one visible \(K_S^0\to\pi^+\pi^-\), exactly four charged tracks, a recoil mass consistent with the \(K_S^0\) mass,
\[
\left| M_{\rm recoil}(\phi S)-m_{K_S^0} \right| < 40~\text{MeV}/c^2,
\]
and little extra neutral energy as measured by
\[
E_{\rm EMC},
\]
the sum of energies of EMC showers not associated with charged tracks [2501.06426]. The recoil mass against the \(\phi\) is also required to satisfy
\[
M_{\rm recoil}(\phi) > 1.08~\text{GeV}/c^2
\]
to suppress
\[
J/\psi\to \gamma\phi\phi,\qquad \phi\to K_S^0K_L^0
\]
background [2501.06426].

A notable methodological feature is the internal normalization strategy. Rather than rely directly on the poorly known branching fraction of \(J/\psi\to \phi K_S^0K_S^0\), BESIII defines a control sample in which the recoiling \(K_S^0\) decays to anything other than \(\pi^+\pi^-\), the “non-\(\pi^+\pi^-\)” sample [2501.06426]. This makes many systematics cancel in the ratio used to extract \(\mathcal{B}(K_S^0\to\text{invisible})\).

## 6. Backgrounds, statistical treatment, and first direct limit

In the BESIII analysis, the principal backgrounds in the \(E_{\rm EMC}\) fit are: true
\[
J/\psi\to \phi K_S^0K_S^0
\]
events where the recoiling kaon decays visibly, especially \(K_S^0\to\pi^0\pi^0\); non-\(\phi\) backgrounds from
\[
J/\psi\to K^+K^- K_S^0(\text{tag}) K_S^0
\]
and
\[
J/\psi\to K^+K^- K_S^0(\text{tag}) K_L^0;
\]
and smaller contributions from unrelated \(J/\psi\) decays and continuum production [2501.06426]. The \(K_S^0\to\) anything background shape is taken from MC validated with a \(K_S^0\to \pi^0\pi^0\) control sample, while the non-\(\phi\) background shape is obtained from the \(K^+K^-\) sideband in data [2501.06426].

The normalization sample yield is extracted from a binned maximum-likelihood fit to \(M(\pi^+\pi^-)\) using a double Gaussian for signal and a second-order Chebyshev polynomial for non-peaking background. The preliminary fitted yield is
\[
N=(1.535\pm0.004)\times10^5.
\]
After subtracting
\[
(3.457 \pm 0.049)\times 10^4
\]
for non-\(\phi\) background and
\[
1022 \pm 260
\]
for four-pion contamination, the final normalization yield becomes
\[
N_{\rm non-\pi\pi} = (1.179 \pm 0.007)\times 10^5
\]
[2501.06426].

The signal yield is obtained from a binned maximum-likelihood fit to \(E_{\rm EMC}\). The fit returns
\[
N_{\rm sig}=56\pm201,
\]
which is consistent with zero [2501.06426]. The efficiencies, excluding subdecay branching fractions, are
\[
\varepsilon_{\rm non-\pi\pi}=16.02\%,\qquad \varepsilon_{\rm sig}=17.14\%.
\]
The branching fraction is derived from
\[
N_{\rm non-\pi\pi}=2\times N_{\rm tagmode}\times \mathcal{B}(\phi\to K^+K^-)\times \mathcal{B}(K_S^0\to\pi^+\pi^-)\times \left(1-\mathcal{B}(K_S^0\to\pi^+\pi^-)\right)\times \varepsilon_{\rm non-\pi\pi},
\]
\[
N_{\rm sig}=2\times N_{\rm tagmode}\times \mathcal{B}(\phi\to K^+K^-)\times \mathcal{B}(K_S^0\to\pi^+\pi^-)\times \mathcal{B}(K_S^0\to\text{invisible})\times \varepsilon_{\rm sig},
\]
and therefore
\[
\mathcal{B}(K_S^0 \to \text { invisible}) = \frac{N_{\rm sig}}{N_{\rm non-\pi\pi}} \left(\frac{\varepsilon_{\rm non-\pi\pi}}{\varepsilon_{\rm sig}}\right) \left(1-\mathcal{B}(K_S^0\to\pi^+\pi^-)\right).
\]
Because no significant signal is seen, BESIII uses a modified frequentist approach based on toy MC samples and sets
\[
\mathcal{B}(K_S^0\to \text{invisible}) < 8.4\times 10^{-4}\qquad (90\%~\text{C.L.}),
\]
the first direct experimental limit on the process [2501.06426].

This limit is stronger than the earlier indirect allowance at the \(\mathcal{O}(10^{-4})\) level only in the sense of being direct and assumption-light; numerically it remains above the mirror-model benchmark \(\mathcal{O}(10^{-6})\) cited by BESIII [2501.06426]. The paper explicitly notes that the result does not yet probe the mirror-model expectation and remains far from Standard-Model sensitivity [2501.06426].

## 7. Comparative status, misconceptions, and open issues

The present state of the subject is defined by a large separation between theoretical possibilities and experimental reach. The main quantitative landmarks are summarized below.

| Quantity | Value | Source |
|---|---:|---|
| Predicted \(B_\text{inv}(K^0_S)\) in mirror model | \(1.8\times10^{-6}\) | [2006.10746] |
| Indirect bound from visible-mode sums | \(<1.1\times10^{-4}\) (95% C.L.) | [1409.2288] |
| First direct BESIII bound | \(<8.4\times10^{-4}\) (90% C.L.) | [2501.06426] |
| Proposed SPS sensitivity | \(\lesssim10^{-8}\) | [1409.2288] |

Several recurring misconceptions are clarified by the literature.

First, “invisible” is not synonymous with \(K_S^0\to\nu\bar\nu\). The experimental searches are designed to be sensitive to any final state that leaves no detector activity, and the mirror-matter scenario does not involve neutrino final states at all [2006.10746, 2501.06426, 1409.2288].

Second, the absence of a direct signal does not make invisible decays irrelevant for kaon phenomenology. The Bell–Steinberger relation depends on contributions from all common final states, so even null searches matter by constraining a previously open channel in CPT/CP tests [1409.2288, 2501.06426].

Third, the mirror-matter branching-fraction prediction should not be read as model independent. Its precision is conditional on specific assumptions: equivalence of the ordinary neutral-kaon \(CP\)-violation scale and the mirror-symmetry-breaking scale, a universal relative mass splitting \(\Delta/m\simeq 7\times10^{-15}\), and a quark-factorized estimate of hadronic mixing [2006.10746].

The main open issue is sensitivity. BESIII established the first direct limit, but the gap between \(8.4\times10^{-4}\) and the mirror-model estimate \(1.8\times10^{-6}\) is substantial [2501.06426, 2006.10746]. The SPS concept, with projected reach near \(10^{-8}\), was designed precisely to close and surpass that gap [1409.2288]. Whether such sensitivities can be realized in practice depends on the control of beam-tail backgrounds, neutral-hadron punch-through, and detector hermeticity to the levels assumed in the proposal [1409.2288].

Taken together, the literature establishes invisible decays of \(K_S^0\) as a technically specialized but conceptually rich subject at the intersection of kaon spectroscopy, hidden-sector phenomenology, oscillation-based exotic decay mechanisms, and precision CPT tests. The current direct experimental limit excludes only comparatively large invisible branching fractions, while the most explicit BSM benchmark remains at the \(10^{-6}\) level and the most ambitious dedicated experimental concept targets the \(10^{-8}\) regime [2501.06426, 2006.10746, 1409.2288].

Source: https://www.emergentmind.com/topics/invisible-decays-of-k_s-0