First, we have explored a possibility of HNLs with rather large masses, larger than 10 GeV and up to 1000 GeV, and short lifetime, meaning that these particles would decay very close to their production point. To make this possible, we deviated from the traditional approach of looking at fully visible in a detector HNL decays, and pioneered new particle search in this mass range with a partially reconstructed final state. For this, we looked at the events with three leptons (electrons or muons), and constructed various kinematic variables to differentiate between the known standard model processes and signal we were looking for. To make sure we are able to discover new particles with such strategy, we performed a signal injection test, where we added events with expected HNL distributions to the evaluated background processes, and proved that we would be able to see their presence. With the data recorded by the CMS detector in 2016, however, we have not seen any hints of HNLs, but we have improved the boundaries on their properties for the first time since more than 20 years. A follow-up, more traditional, analysis exploiting fully visible HNL final state was able to place stricter limits on very heavy particles.
After that, we concentrated on the challenging search for HNLs which fly some distance in a detector before their decay producing so-called displaced signature. The detector design and standard particle reconstruction algorithms are not optimized for the best performance in such signatures. Therefore, we developed dedicated techniques for the best identification of displaced leptons produced in long-lived HNL decays, targeting the decays happening within the CMS tracking detector volume. With novel signatures, we needed to also develop new calibration procedures which ensure that we understand the detector performance, and to come up with new methods for standard model background estimation, as previously there was no searches performed in similar parameter space. The efforts paid off by achieving sensitivity to HNL mixing parameter which is up to two orders of magnitude better than previous state-of-the-art. In this search, we haven't observed hints of new particles yet, but we have paved the way towards new exploration with a larger dataset to be recorded in the future.
