CERN Accelerating science

ATLAS and CMS experiments looking forward to Run 3

On the 4th of July of 2022 we celebrated the 10th anniversary of the end of decades-long search for an experimental confirmation of the BEH mechanism, and the start of a new era. With the discovery of the Higgs boson, the last unknown fundamental parameter of the Standard Model was finally learned the Higgs boson mass. 125 GeV, was “a gift of nature” for experiments as it maximised the number of decay modes in which the higgs boson could be measured, it provided a new key observable for the precision global Electroweak fit of the Standard Model, and failed to provide hints for a scale at which the Standard Model should break down either to preserve the vacuum stability or to avoid a Landau pole (also known as the “triviality” bound). One of the highest priorities of experimental physics today is to measure all the properties of the Higgs boson with precision, using it as the gateway to reach beyond the Standard Model (SM) of particle physics or even to renew our conception of the universe altogether. 

If the first run of the LHC (2010-2012, with proton-proton collisions at a center-of-mass energy of 7 and 8 TeV) was driven by the discovery of this new particle and the measurement of its mass, during the second run (2016-2018, with collisions a center-of-mass energy of 13 TeV) the ATLAS and CMS collaborations measured its coupling to fermions and established the existence of Yukawa interactions. The main production and decay modes of the boson are now observed, and yield precise measurements that test our understanding of the Higgs sector. Through new ideas and technical advances the perspective on what is achievable at the LHC has changed dramatically in the last years: a first measurement of the total width, evidence of the decay to rare processes, an in-depth exploration of CP violation in the Higgs sector, etc. without forgetting of course the progress towards observing the production of Higgs pairs and measuring the Higgs self-coupling. 

The data collected in the first run of the LHC (2010-2012, with proton-proton collisions at a center-of-mass energy of 7 and 8 TeV) was sufficient for the ATLAS and CMS collaborations to independently observe a new particle compatible with the elusive Higgs boson. Searches for a resonance decaying to pairs of vector bosons -photons, W bosons, Z bosons- identified a new particle with zero charge, a mass close to 125 GeV and production and decay rates compatible with the predictions for the SM Higgs boson. These are particularly clear experimental signatures: well-defined signals of high-momentum photons or leptons, with excellent experimental resolution, and a small or easy to model contamination coming from other events of the standard model with similar topologies. The new ‘Higgs-like’ boson started being studied in detail, still with large statistical uncertainties. 

The final run 1 measurements of ATLAS and CMS were combined for a first global picture of Higgs boson production at the LHC. The combination was based on the analysis of the main production processes at the LHC, gluon fusion (GGF), vector boson fusion (VBF), and associated production with a W or a Z boson or a pair of top quarks, and six decay modes, the mentioned decays to vector bosons H → ZZ, WW, and γγ , and three decay channels to pairs of fermions, H → ττ, bb, and μμ (more on them later). The combined production rate relative to the Standard Model prediction was measured to be in excellent agreement with the expectation, and known already to the 10% level. A similar combination of the measurement of the mass in the diphoton and ZZ 4l final states, in which it is possible to reconstruct a narrow Higgs peak as the invariant mass of the decay products, yielded a combined measurement of 125.09 ± 0.21 ± 0.11 GeV – already at the two per mil level with this early dataset.  The spin-zero nature of the new particle was also determined early on: the decay in two photons, predicted in the SM through loops of fermions or W bosons, implied that the spin of this candidate to Higgs boson could not be one, and further studies in the decay to vector bosons firmly established its scalar nature.  

Even with the astounding success in characterizing the new particle so early into the LHC era, many questions on its nature remained open. By then, the new particle was generally acknowledged as the ‘Higgs boson’, but the question of whether the BEH mechanism is as described in the standard model or it is part of an extended sector with significantly different Higgs fields remained (and remains) open. Furthermore, having found a SM-like Higgs boson at the LHC and so far nothing else is a deep puzzle that studying this unique particle in detail can help solve.  

Higgs Physics in run 2

At the end of run 1, the possibility of different mechanisms regulating the masses of fermions and bosons was not yet excluded. In the SM, the Higgs boson is expected to decay to pairs of fermions with a yukawa coupling strength proportional to the mass of the fermion, and therefore largest for third generation leptons and quarks. Despite their relatively large branching ratios, decays to fermions present larger backgrounds and more complicated object identification, and are, in general, more challenging experimentally than the less abundant but simpler experimentally signatures coming from photon and boson pairs. The discovery itself already implied the coupling of the Higgs boson to fermions. The production rate as observed in the LHC matched the SM predictions, which establishes that the dominant production mode for the Higgs boson at the LHC is gluon fusion (ggH), which involves a loop including top quarks. And combining experiments or channels, the first run of the LHC did provide the observation of the Higgs decays to fermions of the third generation. However, establishing independently the coupling to fermions of the third generation had to wait until the second run of the LHC. 

The heaviest charged leptons, tau leptons, have a mass of 1.77 GeV, and a mean lifetime of 10-13 s. They travel in the detector before decaying into collimated jets of hadrons or lighter leptons, always accompanied by one or more neutrinos. The techniques for the reconstruction of tau leptons through the identification of the decay products inside jet, employing state of the art machine learning techniques, have been significantly improved since the discovery. The run 1 studies provided the first evidence of the decay to tau leptons in the ATLAS and CMS combination. During run 2, a clear independent observation of the coupling was obtained already with a partial 13 TeV dataset. The results have long moved from search mode to the precision regime. The full dataset collected by each experiment has yielded precise measurements of inclusive production cross sections, in good agreement with the SM, and a first differential measurement in this decay mode. One of the consequences of the improvements in hadronic tau reconstruction at high transverse momentum is the precision achievable in the ggF production mode, now on par with the VBF mode. 

The observation of the decay of the Higgs boson to pairs of b quarks is experimentally challenging, despite the very large decay probability of this mode. In the main Higgs production mode at the LHC, gluon fusion, the footprint that a gg→H→bb event leaves in the detector is simply a b-quark - anti b quark pair.  B-quarks subsequently hadronize in the detectors in jets of particles, and the resulting B hadrons lifetime, ~10-12, is large enough to show a secondary displaced decay vertex that can be used to discriminate against light jets. The production in the SM of processes yielding pairs of b quarks is orders of magnitude larger than Higgs production. Differentiating the signal is challenging and requires advanced identification techniques. Originally, the solution was thought to be to focus on the less abundant production of a Higgs boson alongside a vector boson (W or Z). With the vector boson decay as a handle to identify the event, the contamination from other standard model processes can be suppressed. Individual observation of VH, H→bb production was announced in 2018  (a, b), with a measurement of the signal strength at the 20% level. The most recent results have moved beyond observation and focus on characterization of the production, including measuring the cross section in specific bins of phase space or fiducially. During run 2, technical advances in jet reconstruction and flavor identification techniques have brought on  a change in perspective for ggF production at very high transverse momentum. The H→bb decay mode has emerged as the dominant channel for measurements for Higgs bosons produced at a high lorentz boost, particularly interesting as it is very sensitive to BSM effects. In this case, the two b quarks are collimated in a single, large radius jet, and substructure techniques can be exploited to discriminate the Higgs boson from other processes as described here and here. Finally, VBF production (inclusive and photon-tagged) are also becoming sensitive (a, b). The measurements are currently statistically limited, and offer great promise for future runs, including the high luminosity runs.  

The top quark is a special case that can only be probed through in production due to its large mass. The associated production of Higgs bosons with a top quark, anti-top quark pair presents a very complicated phenomenology. Each one of the top quark decays to a W boson and b quark, with the W boson decaying hadronically (67%) or to charged lepton- neutrino pairs, and the b quark hadronization as described above. The Higgs boson can of course decay to any of the possible final states described already. The diphoton final state is particularly remarkable due the recognisability of the signal and its higher signal over background ratio. Amongst the other decay modes, the decay to b quark pairs is characterized by the more abundant signal but more complicated background, and the combination of final states involving leptons (WW, ττ, ZZ) by relatively large samples paired with clean reconstruction. The combination of the different decays results in very complicated topologies with a large number of jets, many of them coming from b quarks. Large backgrounds with similar topologies to the signal complicate the measurement, chiefly those involving top-antitop production associated with additional b-anti b pairs. The experimental efforts in devising strategies, relying on Machine Learning techniques to disentangle interesting events from the large background while preserving signal efficiency, have far surpassed expectations. The first observation of ttH production and the coupling of the Higgs boson to the top quark was obtained with a partial run 2 dataset (read more here and here). Current measurements with the full run 2 dataset already have combined uncertainties on the cross section at the 20% level (a, b, c, d). Finally, the sign of the top yukawa can be accessed through the associated production of the Higgs boson with a single top quark, with measurements in agreement with the SM but statistically limited.

Combining the individual measurements in fermion and boson final states described for a global perspective,  ATLAS and CMS paint a compelling image of Higgs boson production and decay. The Higgs has been observed to couple to bosons and fermions following the predictions of the SM. The main Higgs production modes (ggF, VBF, VH, ttH) have all been observed independently, with measurements of the cross sections with precisions at the 10-20% level.

Figure 1: Signal strength parameters per individual production mode and decay channel μif, and combined per production mode μi and decay channel μif as measured by CMS. Nature 607 (2022) 60-68.

Figure 1 shows a summary of the signal strengths (observed rate relative to the SM prediction) as measured by CMS with the full run 2 dataset. Furthermore, cross section measurements are not only performed inclusively and oriented to confirm agreement with the SM: they have been expanded to focus on obtaining a thorough description of Higgs kinematics, with fiducial and differential, even doubly differential, results becoming the norm. As an example, Figure 2 summarizes the Higgs production cross sections measured by ATLAS with the full run 2 dataset in different kinematic bins, in the  “Simplified Template Cross Sections (STXS)” schema

Figure 2: Observed and predicted Higgs boson production cross sections measured by ATLAS, in different kinematic regions (Simplified Template Cross Sections Schema). Nature 607 (2022) 52–59

The Higgs couplings to the SM particles is well established for the main decay modes (bb, WW, ττ, ZZ, γγ) and for the top yukawa coupling, The results obtained so far are in very good agreement with the SM, though uncertainties are still large (5-10% in most cases), as shown in Figure 3. Future data is necessary to increase this precision to the level where we are sensitive to BSM deviations with respect to the predictions. The full HL-LHC dataset, 3000 fb-1 per experiment or 20 times more than the one collected until now, will bring the precision to the few percent level for the main decay modes, with a sensitivity dominated by theoretical uncertainties. The evolution of the sensitivity to the kappa parameters with the LHC runs is also shown in Figure 3. Furthermore, the decays to SM particles we have observed so far are only a fraction of the full spectrum of possible decays. The sensitivity is either at evidence level (eg: 𝜇𝜇) or getting closer to it for several of the statistically dominated ones. The third run of the LHC will be fundamental in their study, and they are discussed in the next section.

Figure 3: Left: Reduced Higgs boson coupling strength modifiers and their uncertainties with the full run2 LHC dataset collected by ATLAS. Two fit scenarios with κc = κt (coloured circle markers), or κc left free-floating in the fit (grey cross markers) are shown. Nature 607 (2022) 52–59. Right: Observed and projected values resulting from the fit in the κ-framework in different data sets collected by CMS: at the time of the Higgs boson discovery, using the full data from LHC run 1, the full run2 data set , and the expected uncertainty at the HL-LHC for  L= 3000 fb−1. Nature 607 (2022) 60-68

The characterization of the Higgs boson continued, with increased precision in the measurement of its properties. Starting with the Higgs mass: in only a decade the measurement of this fundamental parameter of the SM has gone from being unknown to being measured at the level of 0.1% (125.38 ± 0.11(stat) ± 0.08(syst), CMS; 124.94 ± 0.17(stat) ± 0.03(sys) GeV, ATLAS. On the other hand the total width of the boson, connected to its lifetime, cannot be easily measured experimentally in ATLAS and CMS. The Higgs boson is an unstable particle, which exists for a very brief period before disintegrating to lighter particles, 1.6 × 10−22 s as predicted in the SM. This small lifetime translates to a  total width of the resonance of ΓH,SM = 4.1 MeV,  much smaller than what is directly measurable from the resonance line shape in ATLAS and CMS. Even exploiting the clean mass resolution of H → ZZ events, the current direct measurements set upper limits at the GeV level. ZZ events are also used in an alternative approach that compares the production of Higgs bosons with a mass close to the nominal value (on-shell) and far away from it (off-shell) to indirectly measure the total width.  Analyzing with this method the full dataset collected in run 2 a 50% precision on the total width  (ΓH = 3.2+2.4−1.7 MeV) is obtained.

The measurement of the charge-parity properties of the Higgs boson couplings are consistent with a CP-even scalar particle, (JPC=0++), as predicted in the SM. This was already established in run 1 for the couplings of the higgs boson to  massive gauge bosons. CP-odd contributions in Higgs-Gauge interactions occur through higher-orders while in the Yukawa interaction of the Higgs boson with fermions, CP-odd contribution can be present at the tree level. At run 2, with the observation of the Higgs couplings to fermions also the CP properties of these couplings have been studied for the top quark through the ttH production and to the tau through the use of tau-polarisation sensitive observables. The sensitivity achieved in the latter channel came as a surprise and underscores the rapid progress made in tau reconstruction in ATLAS and CMS.  

While the observed boson exhibits a behavior compatible with the predictions of the SM, there is room for BSM properties both in production and in decay. A hypothetical coupling of the Higgs to dark matter candidates is motivated in many models. This can be searched for experimentally by looking at ‘invisible’ decays of the Higgs, possible in the SM through the H→ZZ→4𝜈 decay but extremely rare, and characterized by signatures with high missing transverse momentum. The sensitivity of the searches is driven by VBF production mode. Branching ratios of Higgs to invisible larger than 16% [CMS] and 13% [ATLAS] at 95%CL are already excluded. They can be translated to weak interacting massive particle (WIMP)-nucleon scattering cross sections, and place limits for masses below 10 GeV. 

Other BSM Higgs decays that are searched for directly include signatures involving flavor violating yukawa couplings (h ll’ and FCNC top decays) and chain decays to new scalars and pseudoscalars (h aa/ss), prompt or long lived. In parallel to studying Higgs decays, ATLAS and CMS have an extended program of searches for Extended Higgs sectors at higher and lower masses. And further beyond, the Higgs boson is a critical tool in searches for new physics in a variety of scenarios, for example through the decay of exotic new states to higgs bosons. No signs of new physics has been found yet, but many searches of this kind require large datasets and innovative techniques to access challenging phase-spaces, and represent an exciting opportunity for the future. Complementarily, the combination of the measurements of the decays to SM particles and invisible decays can be used to place limits on other unobserved decays (BSM or simply rare). The current upper limits, 12% at 95% CL, leave a large phase space to probe in the future.

Higgs physics opportunities at run 3

With its successes at run1 and run 2, Higgs physics at LHC has by far exceeded expectations. The achievements of Higgs physics at run 2 have relied greatly on the detailed studies of run 1. For many channels, run 1 has provided invaluable intermediate goals that have immensely enriched the run 2 physics program. Similarly, defining intermediate goals at run 3 is essential to the development of the richest and broadest Higgs physics program at the HL-LHC.  

During its first decade of running, the machine performance culminated in 2018: during that year a dataset corresponding to an integrated luminosity of approximately 60 fb-1 was delivered to the experiments. At this pace, reaching the overall project’s goal of 3 ab-1 per experiment would require 50 full years of running. To reach the necessary performance, during the second Long Shutdown of the LHC (LS2), the collider complex and the experiments underwent major upgrades to get ready for the High-Luminosity phase (HL-LHC). These upgrades are however only a first step towards the necessary improvements to reach the performance required for this demanding HL-LHC phase. Another round of upgrades will be necessary during the LHC third long shutdown (LS3) which is now scheduled to take place from 2026 until 2029. With the new extended schedule, the conservative target luminosity per experiment is 250 fb-1, the foreseen estimate is approximately 290 fb-1. The center-of-mass energy for the run 3 will increase to 13.6 TeV, which brings a non-negligible increase in Higgs production cross section with 7% for ggH, 11% for HH and 13% for ttH. 

The flagship Higgs physics result expected at run 3 is the observation of the Higgs Boson decays to muons, which would provide direct evidence for the Yukawa coupling of the Higgs boson to fermions of the second generation. With the outstanding performance of the LHC at run 2, and the very high quality of the data collected by ATLAS and CMS, first evidence for this important decay was observed. The ATLAS experiment observes an excess corresponding to 2.0σ (with an expectation of 1.7σ) and CMS observes an excess of 3.0σ (with an expectation of 2.5σ). The precision of this result is currently limited by the statistics of the data sample; when extrapolating to run 3 conditions, both experiments fall slightly short of an observation significance. Their combination should provide an unambiguous discovery of this decay mode; however it is a goal for both experiments to strive to improve and optimize the analyses to reach an observation sensitivity independently. Every gain in sensitivity will be a very useful step towards the highest possible precision at HL-LHC. 

Until recently, the muon Yukawa coupling to the Higgs boson was thought to be the only second generation Yukawa reachable at the LHC. A very recent result of the search for Higgs boson decays to charm quarks at the LHC has provided new hopes for this channel. Given the challenges of cornering the small charm Yukawa coupling several methods have been developed and are described in ESPP-2019. These include the search for decays of the Higgs boson to a charmed meson J/ψ and a photon, measuring the charge asymmetry in the WH production, as well as constraining the coupling through constraints on the total width of the Higgs boson. The latter requires assumptions on the other couplings of the Higgs boson and can be done in several different ways. The strongest constraint was obtained through the measurements of the Higgs boson couplings through a global fit and assuming that all couplings to other particles have Standard Model couplings and the charm Yukawa is constrained from a measurement of the total width. This measurement yields a limit on an anomalous charm Yukawa coupling of 1.7 times the SM Yukawa at 95% CL at HL-LHC. In comparison the limits from the direct search in the associated production channel with a vector boson (VH), yielded a limit of 2.2. In contrast the latest result from CMS using the run 2 dataset which is twenty times smaller than that of the HL-LHC, already has a sensitivity of 3.4 times the Standard Model coupling! CMS obtained this result through the full exploitation of flavour tagging capabilities and the reconstruction of the Higgs boson through boosted large R-jet using state-of-the-art machine learning techniques. The CMS collaboration has extrapolated their result and assuming the run 2 performance obtains a limit on the production rate of 1.6 times the Standard Model for 3ab-1 as presented during Snowmass21.  With a combination with ATLAS and LHCb, the inclusion of other channels such as the boosted inclusive Higgs and the VBF production modes, and some improvements a first direct evidence for the Yukawa coupling of the Higgs boson to charm quarks seems to be at reach at HL-LHC. It is therefore extremely important as an intermediate goal of run 3 that progress is shown by all experiments in improving their sensitivity to this very difficult channel. 

An additional first evidence has been reached at run 2 in the yy* channel by ATLAS. The good sensitivity obtained by ATLAS was due to a new reconstruction technique of two close-by-electrons which allows to reach lower y* invariant masses. Although an excess of 3 standard deviations was observed, the sensitivity was only 2.1 standard deviations. Further analysis improvements and possibly a combination with CMS  will be needed to reach an unambiguous observation at run 3 and is therefore among its landmark intermediate goals. 

Another very interesting decay mode of the higgs boson is the Zy decay channel as it is a field-tensor coupling that hasn’t been measured yet. The expected branching fraction is very small (only 2.3% of the diphoton channel); however both ATLAS and CMS have observed excesses in this search at the 2 standard deviations level while their sensitivity are both 1.2 standard deviations [a, b]. A strong first evidence at run 3 is an important intermediate goal.

Guaranteed deliverables are key for our large scale science experiments. A key landmark result at HL-LHC is the observation and measurement of the di-Higgs production, through which direct constraints on the Higgs boson self  coupling can be imposed. The current observed upper limits on the HH production cross section are 2.4 and 2.5 times the Standard Model prediction for ATLAS and CMS respectively (2.9 and 3.4 expected), as shown in Figure 4. For the European Strategy for Particle Physics, the projections using first complete analyses of a partial run 2 datasets have shown that the combination of ATLAS and CMS could lead to a sensitivity of 4 standard deviations on the HH production, thus very close to the observation sensitivity. The precision obtained on the Higgs self coupling was approximately 50%. This was a major change in the prospects for Higgs physics at HL-LHC and was made possible by the intermediate goal of searching for this process in all possible modes at run 2. The analysis of the full run 2 data showed further progress in sensitivity, improving on the mere addition of data. These improvements were incorporated to the HL-LHC analyses prepared for the 2021 US Snowmass exercise, and show that individual projections could already be improved by almost a 20%, though the combined ATLAS+CMS sensitivity was not rederived. With these improvements the combined sensitivity of ATLAS and CMS at run 3 should be close to 2 standard deviations. It is key that analyses are further improved to best take advantage of the additional data and push the limits of sensitivity in this channel as much as possible.

Figure 4: Left: Observed and expected 95% CL upper limits on the signal strength for double-Higgs production from the bbbb, bbττ and bbγγ decay channels, and their statistical combination, with the ATLAS full run2 dataset. Right: Combined expected and observed 95% CL upper limits on the HH production cross section for different values of κλ, using the full data from LHC run 2 collected by CMS, assuming the SM values for the modifiers of Higgs boson couplings to top quarks and vector bosons. The result corresponds to the combination of bbZZ, multilepton (WWWW, WWττ and ττττ), bbττ, bbγγ and bbbb final states. Nature 607 (2022) 60-68.

In addition to the above intermediate goals which are all landmark results, run 3 should also provide a fertile ground to improve the numerous ancillary measurements which are key to improve the precision of the measurement of the Higgs boson properties in the main channels, including measurements of top production to further constrain PDFs, the measurement of vector-boson and jets to improve on the modeling of the overarching background, the measurement of diboson production, the measurement of ttV and tt-HF, and many more. The ensemble of measurements made at the LHC are invaluable for the further development of precise theoretical Standard Model predictions which are a cornerstone in reaching new heights in precision Higgs physics. 

Looking forward

The LHC and its experiments are wonderful machines which have pushed the limits of our understanding of particle physics into a new frontier. From the discovery of the Higgs boson to the establishment of its coupling to fermions, through the precise measurements of many of its properties, and in parallel with the study of its potential in the search for new phenomena, the decade since the Higgs discovery has been paved with highlights and never-befores. ATLAS and CMS have far surpassed expectations, scrutinizing the behavior of the only fundamental scalar particle known to date. The job is not done: with only 5% of the full LHC explored so far,  new results with unprecedented precision await us. The road ahead has clear guidelines and milestones, including the measurement of the self-coupling, and the possibility of exciting surprises hidden in the new data. Higgs physics will continue to drive the LHC effort and particle physics as a whole into the future.