The (long and winding) road to the golden Higgs channel

8 March 2010

The latest LHC schedule implies that complete results from the Higgs boson searches may take a while. Still, the Higgs hunters will have a lot to do with early collision data. An example is the roadmap for the Higgs boson search in the Higgs decay channel into four leptons.

When I was asked a few days ago to write this article, my first reaction was: “Oh, with pleasure! Hmm, but … Who wants to read today about the remote future plans for the Higgs boson search, now that the first ATLAS results with real collision data are starting to appear?”

Indeed, following the LHC schedule over the last year or so, it became clear that the exclusion of any of the Higgs boson masses predicted by the Standard Model may still take a while.The most sensitive Higgs mass region between 162-66 GeV has already been excluded at the 95% CL at the Tevatron. An integrated luminosity of about 300 pb-1 at a centre-of-mass collision energy of 7 TeV, using the Higgs boson decays into two W bosons could reinforce that confidence level. For all other Higgs decay channels to contribute to this exclusion will only be possible when more than 1 fb-1 of data is collected, i.e. probably only at the end of the next year or later. Combining different channels, such as H→WW(*), H →ZZ(*), H → γγ, and H → ττ will significantly improve the sensitivity at high and low masses. Furthermore, ATLAS and CMS combined could perhaps discover a Higgs with a mass close to 160 GeV by the end of 2011, if we are lucky enough to hav a Higgs at that mass. However, at the beginning, the spotlights will be shining on the more fruitful detector performance and Standard Model physics studies.

Nevertheless, the Higgs hunters are not sitting idly in the dark and waiting their turn. On the contrary, they are eagerly paving the road to the first Higgs physics results with their various performance and background studies for early collision data. It goes without saying that this is done in tight collaboration or even completely merged with the relevant ATLAS performance and physics working groups.

As an example, let me take you down the road of the Higgs search in the four-lepton decay channel. This so-called "golden channel" - a Higgs boson decaying into two Z bosons, which subsequently both decay into two leptons - owes its name to a very clean, almost background-free signature in the detector. The reducible Z + jets and tt backgrounds are suppressed by the lepton isolation and impact parameter criteria. The remaining ZZ background is irreducible, but it is expected to be well below the Higgs signal for most of the mass range of interest.

This channel is in fact so clean that there is not much to see in early collision data. There are at most 2 or 3 signal events expected at an integrated luminosity of 1 fb-1 in the most sensitive Higgs mass region around 200 GeV. Also the total background contribution is small - there are only about 5 events surviving in the full mass range (compared to what would be expected with 30 fb-1 at 14 TeV as shown in the figure.).  Loosening the selection criteria does not increase significantly the number of events. Already the requirement of four leptons in the final state reduces the event rates to very small numbers.

(Left) Simulated Higgs boson decay into two electrons and two muons in the Vector Boson Fusion channel where a Higgs is produced in association with two jets. (Right) Reconstructed four-lepton invariant mass for signal and background processes, in the case of a 130 GeV Higgs boson, normalized to an integrated luminosity of 30 fb-1 at a 14 TeV centre-of-mass energy. The event yield at 1 fb-1 and 14 TeV is 30 times smaller. Estimate for 7 TeV: scaling down by a factor of ~3 to 4.

Precise knowledge of the signal selection efficiency, needed for reliable exclusion limits, can be obtained from the measurement of single-lepton reconstruction efficiencies. In addition, since the signal yield scales with the fourth power of the single-lepton efficiency, even a small optimization of electron and muon identification significantly improves the exclusion reach. An improved matching between electron clusters and inner detector tracks is just one example here. All of these tasks have been well integrated within the corresponding performance working groups for quite some time.

The contribution of the Standard Model background processes can also be evaluated with early data, using final states with two or three leptons. In order to reproduce as much as possible the original analysis selection criteria, the invariant mass of the two leptons is usually required to be close to the Z boson mass. Different measurements are then performed for reducible and irreducible backgrounds.

The reducible Z + jets background contribution, caused by two fake or non-isolated leptons originating from the jets, is expected to be very small after all analysis cuts. Nonetheless, this must be proved and optimized with real data. Dedicated control data samples (dijets, Z + light-jets, Z + b-jets) are therefore defined to study the identification of the jet-initiated leptons, separately for the jets from light and from heavy quarks. Obtained results are then extrapolated to the signal topology.

The remaining irreducible ZZ(*) background contribution cannot be determined from the fit to the side-bands, due to the small number of events. Instead, one has to rely on the Monte Carlo predictions. The dominant uncertainty on the background normalization is initially given by the measured luminosity. This uncertainty cancels out if the number of ZZ events is estimated from the measured number of Z boson events. The ratio of these two numbers, corrected for the single-lepton lepton efficiencies, is well known from theoretical calculations.

It is worthwhile mentioning that the ZZ(*) background in the Higgs analysis channel is almost identical to the ZZ(*) signal for the Standard Model cross-section measurements. Special effort has therefore been put into defining a set of baseline selection criteria common to both the Standard Model and the Higgs analysis. This allows for a straightforward comparison of results and thus a better understanding of the ZZ(*) process.

This roadmap towards the first exclusion limits in the golden Higgs channel has been developed during the last year and a half using simulated Monte Carlo data. As more and more collision data start to arrive, the dream should now be turning into reality. Let us just hope that this road we have built will safely and without many windings lead us to success!




Sandra Horvat

Max Planck Institute