Search for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s = 13 and 13.6 TeV

summary

The gist

A search was conducted for heavy neutral leptons (HNLs) in the decay of W bosons, specifically targeting final states featuring three muons…

In short

Researchers searched for heavy neutral leptons (HNLs) decaying into one prompt muon and two highly non-prompt muons forming a secondary vertex in proton-proton collisions at 13 and 13.6 TeV. The analysis focused on long-lived HNLs, requiring the two non-prompt muons to be far from the beam line. The study set exclusion limits on HNL mixing parameters for masses between 1 and 5 GeV, improving sensitivity in this mass range.

Key concepts

Heavy Neutral Leptons (HNLs)
These are hypothetical, very heavy neutral leptons that are predicted by some extensions to the Standard Model of particle physics. The search looks for them decaying into a specific signature: one standard muon and two other muons that appear much later and further away from the collision point.
Prompt vs. Non-prompt Muons
A prompt muon comes directly from the primary collision vertex, meaning it originates right where the protons hit each other. A non-prompt muon originates from a secondary vertex, meaning it travels some distance before decaying or originating, which is a key feature used to identify long-lived particles like HNLs.
Secondary Vertex (SV) Displacement
This refers to the point where the two non-prompt muons are reconstructed. For long-lived HNLs, this secondary vertex should be significantly displaced from the main interaction point (the beam line). This displacement is a crucial signature used to distinguish genuine HNL decays from standard model backgrounds.

Terminology used across episodes

This episode discusses

The paper

Search for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s = 13 and 13.6 TeV · Read on arXiv

CMS Collaboration

European Organization for Nuclear Research (CERN)

A search for heavy neutral leptons (HNLs) in the decay of W bosons is performed in final states with three muons, with one originating promptly from the primary interaction vertex and the other two stemming from a common secondary vertex. The analyzed proton-proton collision data correspond to an integrated luminosity of 138 fb-1 recorded at a centre-of-mass energy of 13 TeV in 2016 - 2018, and of 34.6 fb-1 recorded at 13.6 TeV in 2022. The analysis strategy is optimized for large HNL decay lengths by requiring highly non-prompt muons forming a secondary vertex far from the beam line, and hence reconstructed only in the muon system of the CMS experiment at the CERN LHC. This enhances the analysis reach in the high HNL lifetime region, corresponding to low HNL masses (2 - 4 GeV) and small mixing parameters with the standard model neutrinos (10-6 - 10-5), where prompt and tracker-based non-prompt analyses have limited acceptance. No excess over the background prediction is found. Under the Majorana HNL hypothesis, exclusion limits at 95% confidence level are set on the HNL mixing parameter as functions of the HNL mass in the range of 1 - 5 GeV.

Transcript

Introduction to the show: ident: High Energy Physics: Experiment. The day in high-energy experiment: detectors, searches and limits.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Search for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s = 13 and 13.6 TeV".

Mira: Here's the gist of the paper. A search was conducted for heavy neutral leptons (HNLs) in the decay of W bosons, specifically targeting final states featuring three muons: one prompt muon originating from the primary interaction vertex and two highly non-prompt muons originating from a common secondary vertex.

Kai: This analysis utilized proton-proton collision data collected at a center-of-mass energy of 13 TeV (integrated luminosity 138 fb-1 recorded in 2016–2018) and 13.6 TeV (integrated luminosity 34.6 fb-1 recorded in 2022).

Mira: The analysis strategy was specifically optimized for large HNL decay lengths, requiring the two non-prompt muons to form a secondary vertex far from the beam line, and reconstructed exclusively using information from the CMS muon detectors.

Kai: This approach enhances sensitivity in the high HNL lifetime region, corresponding to low HNL masses (2–4 GeV) and small mixing parameters with standard model neutrinos (10-6 – 10-5), where prompt and tracker-based non-prompt analyses have limited acceptance.

Title and authors: Mira: Moving into the actual summary of this paper, it boils down to a search for heavy neutral leptons that decay into three muons, with one prompt muon and two non-prompt ones coming from a secondary vertex. The core idea is leveraging those displaced tracks as a unique signature for new physics.

Lev: So they are looking at how these HNLs behave in the detector, specifically focusing on how far away that second vertex is from the primary collision point, which tells us about its lifetime.

Kai: And they are analyzing data from two different collision energies: thirteen TeV and thirteen point six TeV, with specific integrated luminosities of one hundred thirty-eight femtobarns and thirty-four point six femtobarns respectively <ref:2610.40911#pg1>. This energy combination is what allows them to gain sensitivity in a way that single-energy runs can't achieve on its own.

Mira: The analysis strategy they employed is explicitly optimized for finding HNLs with large decay lengths, which means they are focusing their statistical power where the signal from these long-lived particles should be strongest.

Lev: This optimization means they are filtering the data to select events that have those highly non-prompt muons forming a secondary vertex far from the beam line, and then reconstructing them only using information from the CMS muon detectors.

Kai: That's a very specific experimental requirement, because it really targets that high lifetime region where prompt searches struggle to get any signal at all. It’s about maximizing the chance of seeing something that lives long enough to leave a clear secondary signature.

Mira: The paper summarizes the results by setting exclusion limits on the HNL mixing parameter as a function of its mass within one to five GeV under the Majorana HNL hypothesis. This is their main quantitative output for constraining this type of particle physics scenario.

Lev: So, to put it simply, they are saying that based on their data, they can rule out certain combinations of the HNL's mass and its mixing with our known neutrinos in that low mass range.

Kai: And what's interesting is that they found no significant deviation from the standard model background when they checked their limits across multiple iterations of setting those exclusion limits. That means their exclusion results are fairly stable against minor tweaks in the analysis procedure.

Mira: It's not just one result; it’s a consistent finding across different ways they tested it, which lends confidence to the constraints they are reporting on these heavy neutral leptons.

Lev: That consistency is what you need when you’re building real hardware, because you want to be sure that the background modeling isn't accidentally creating a false signal just because of how we chose our cuts.

Kai: So, in short, the summary shows a targeted search for HNLs using displaced muon signatures at thirteen and thirteen point six TeV collisions, leading to exclusion limits on the mass and mixing parameter within the one to five GeV range under Majorana assumptions <ref:2610.40911#pg1,13 and 13.6 TeV>.

Mira: And this directly relates to testing theories that introduce heavy right-handed neutrinos, which could explain neutrino masses through mechanisms like the seesaw mechanism.

Lev: It’s a specific piece of data for those theorists who want to know what experimental limits look like in terms of mass and mixing parameters for these new particles.

Kai: So, we’ve seen how they set up the search, what data they used, and what the basic exclusion results are. Now we can dig into how they actually improved their search strategy from a baseline level.

The paper's summary: Kai: Now let's look at the improvements this paper suggests over previous searches for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s = thirteen and thirteen point six TeV <ref:2610.40911#pg1,for long-lived heavy neutral leptons in signatures with one prompt muon>. They aren't just reporting the final results; they are suggesting how to make the search better.

Mira: The key improvement they point to is using improved AI to perform these searches with enhanced sensitivity, specifically by requiring those highly non-prompt muons to form a secondary vertex far from the beam line, and reconstructing them only in the muon system of the CMS experiment.

Lev: That’s a direct operational suggestion: use better AI algorithms to pinpoint those displaced tracks and focus on the most informative detector components—the muon detectors—to maximize sensitivity at that high lifetime region.

Kai: And they also suggest using AI to set exclusion limits on the HNL mixing parameter as functions of its mass in the range of one to five GeV under the Majorana HNL hypothesis, specifically highlighting a sensitivity gain from combining data sets from different center-of-mass energies.

Mira: So, they are suggesting that by intelligently combining thirteen TeV and thirteen point six TeV data, you get a better constraint on the mixing parameter across that mass range compared to looking at just one dataset alone <ref:2610.40911#pg1>. That's a practical way to improve the reach of your search.

Lev: From an error correction perspective, if the AI can do that intelligent combination and accurately map those kinematics, it means we need algorithms capable of handling complex event topologies where we have multiple displaced tracks simultaneously with high fidelity.

Kai: They also suggest using AI to distinguish genuine HNL decays from background by selecting events based on kinematic variables such as the displaced dimuon transverse momentum, p pair T, and the significance of the reconstructed secondary vertex displacement, which is Lxy over sigma Lxy.

Mira: That’s a crucial step because those kinematic variables are what truly separate a signal from noise; they tend to be large for genuine HNL decays and that AI can learn to exploit that separation better than simple cuts would.

Lev: So, the idea isn't just applying fixed cuts anymore; it's using machine learning to find the subtle patterns in those kinematic variables that tell us what’s real, which is much harder to do when you have complex detector noise involved.

Kai: They also propose using an AI approach called the "ABCD method" for background estimation, which relies on two statistically independent conditions: a tracker veto on the first muon and a reduced signal selection. This helps estimate background yields in those tricky signal regions where the observed yield is compared with the prediction nA equals nB over nC over nD.

Mira: That "ABCD method" sounds like a sophisticated way to handle background modeling uncertainties by using two different statistical conditions to constrain each other, which should give you a more reliable estimate of what's going on in the signal region.

Lev: And I think that method is powerful because it tries to use two independent pieces of information—one from the tracker and one from reduced selection—to get a better picture than just relying on one condition alone when estimating yields.

Kai: Finally, they talk about optimizing search region binning by maximizing expected discovery significance for each HNL mass hypothesis by defining bins based on p pair T and Lxy over sigma Lxy, finding that a single set of four bins defined by thresholds of Lxy over sigma Lxy equal to four and p pair T equal to two point two gigaelectron volts is used.

Mira: That final binning strategy is very specific; it shows they went through a process of maximizing the significance for each mass hypothesis before settling on that single set of four bins. It’s a direct way to show how the search region is shaped around the signal expectation.

Lev: So, it's not just finding a result; it's about defining the most efficient search space possible based on what you expect from the physics we are looking for. That’s where real experimental design happens.

The paper's improvements: Kai: So to wrap up this paper, the search for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s = thirteen and thirteen point six TeV shows they have set exclusion limits on the HNL mixing parameter as functions of its mass between one point five and five GeV under the Majorana HNL hypothesis <ref:2610.40911#pg1,search for long-lived heavy neutral leptons in signatures with one prompt>.

Mira: And this work, especially when considering the suggested improvements like combining data sets at different energies, gives us a better constraint on these theoretical models that involve heavy right-handed neutrinos explaining neutrino masses through mechanisms like the seesaw mechanism.

Lev: It's a specific piece of data for theorists who want to know what experimental limits look like in terms of mass and mixing parameters for these new particles, and it sets a benchmark for how sensitive we need to be in our detectors.

Kai: The overall implication is that this search channel is a viable path forward for probing physics beyond the standard model through searches for long-lived particles at colliders. It shows that specific experimental signatures can lead to meaningful constraints on new particle models.

Mira: And the paper’s conclusion, even with those suggested improvements, is that these results are consistent and stable across different analysis procedures when compared to SM background estimations from control samples.

Lev: And for my part, the key thing is that this work lays out a robust methodology for how to use displaced signatures and sophisticated background estimation techniques like the ABCD method when you're building real hardware designed to find these long-lived particles.

Kai: We’ll keep an eye on how these kinds of constraints evolve as we look at new data from various experiments. Thanks for tuning in today.

Conclusion: Kai: So today we’re talking about the paper titled "Search for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s equals thirteen and thirteen point six TeV."

Mira: Exactly. The core thing here is they looked for these heavy neutral leptons, or HNLs, decaying into one prompt muon and two highly displaced muons forming a secondary vertex far away from the beam line.

Lev: It’s about testing how long these particles live and using that displacement to find them in the data.

Kai: They used proton-proton collision data from both thirteen TeV at one hundred thirty eight femtobarns and thirteen point six TeV at thirty four point six femtobarns.

Mira: The strategy they used is specifically tuned for large HNL decay lengths, which means they focused on requiring those two non-prompt muons to form a secondary vertex that’s really far away from the beam line.

Lev: They reconstructed those displaced tracks only using information from the CMS muon detectors, which helps them probe that high lifetime region where other methods struggle to see anything.

Kai: Their goal was to enhance sensitivity in the long-lived region, which corresponds to HNL masses between two and four giga-electron volts and very small mixing parameters with standard model neutrinos.

Mira: They set exclusion limits on the HNL mixing parameter as a function of its mass, specifically under the Majorana HNL hypothesis for masses between one and five giga-electron volts.

Lev: The result they reported is that this analysis surpasses previous experimental results in terms of sensitivity for HNLs with masses between two point five and four point two giga-electron volts.

Kai: They also found no significant deviation from the standard model background when they set those exclusion limits across multiple iterations of their procedure.

Mira: That consistency is important because it shows that the constraints they are reporting on these heavy neutral leptons are stable, regardless of how they tweak the analysis procedure a bit.

Lev: For me, what this means for real hardware is that you need to be really good at modeling those displaced signatures and getting reliable estimates of background yields when you’re looking for particles that might only last a few millimeters.

Kai: So in summary, the paper "Search for long-lived heavy neutral leptons in signatures with one prompt muon and a secondary vertex formed by two non-prompt muons in proton-proton collisions at sqrt s equals thirteen and thirteen point six TeV" provides exclusion limits on the HNL mixing parameter within the one to five giga-electron volt mass range.

Mira: It’s a solid constraint for testing models that introduce heavy right-handed neutrinos via mechanisms like the seesaw mechanism.

Lev: It shows how important it is to combine data sets from different collision energies to get a better picture of these new particles, which is something we need when building systems for real hardware.

Kai: Next up, we’re going to look at how foundation models are starting to unify particle collision data analysis with event-level modeling using EveNet.

More episodes

← Home