ATL-PHYS-PROC-2008-029 07 October 2008 esrmn fbt h velocity the precision both of measurement MEASUREMENT applied VELOCITY usually 1.2. requirements quality the fail to chance trigger. the of and level high flipping, at charge consequent with materials, has average lower detector sleptons much produced a is models two to GMSB effect the due ( in This of problematic Typically velocities one CMS. algorithms. different least Trigger to with compared Level at produced dimensions High that larger be are or the to Level-1 sleptons to fail two the or due by crossing detector the ATLAS bunch imposed the wrong cuts for the quality important in respect with of reconstructed more time be because of either out all would system at therefore the reconstructed and reach may relativistic light typical of the speed the to than smaller significantly velocity with ”massive” to similarly TRIGGER behave 1.1. stable that more or momentum, one transverse of existence high few the detectors. and to the is LHC up velocity processes of the cross-sections these low framework in with all muons the LHC, with for particles the in signature particles, stable at independent or charged -like produced model models, copiously also massive largely Split-SUSY be and particles, with can in common -like collisions, and A predicted to p-p models, nb. are addition SUGRA in R- of In pairs LHC. scenario slepton in fb. LSP at produced the 20-30 produced directly of of are be mass order can the KK quantum the (R-hadrons) same of partner. on the LHC Model with depending Standard at dimensions, Standard each extra pb cross-section the for in as where 1 state (UED), spin (KK) to model Kaluza-Klein and Dimensions supersymmetric so-called fb numbers Extra corresponding heavier Universal 100 a of the exists from by chains predicted decay ranging Model also via are LHC mainly leptons at Stable proceeds with R- LHC cross-section particle. weakly example the couples typical slepton, for at with stau sleptons like the particles, of number, Breaking typically Production quantum (NLSP), LSP. Mediated global particle gravitino Gauge supersymmetric conserved, in the lightest almost example the for or when appear conserved, models, sleptons charged (GMSB) new heavy a Stable carry KK-parity. or states parity new more or one PARTICLES META-STABLE CHARGED 1. CMS and ATLAS in particles long-lived for Search h dnicto fahaysal hre atcerle ntepeiedtriaino t astruhthe through mass its of determination precise the on relies particle charged stable heavy a of identification The However muon. a as reconstructed being of probability high a has particle stable heavy lepton-like a level trigger At which in [1], Model Standard the beyond physics of models many by predicted are particles charged stable Heavy M osac o hs inl,wt atclrepai npsil hlegst h rge n eetroperations. and detector ATLAS and by trigger proposed the to strategies minimum challenges the possible to reviewed on volume, emphasis are particular detector Here with unique signals, the momentum. their these within high for exploiting search anywhere and by to velocity particles CMS production low Model jets Standard with beyond and/or from particles physics distinguished signal multi-leptons ionizing might be from which can particles ranging states long-lived heavy new signatures, for Such search to Model. used Standard be the can detectors CMS and ATLAS The collaborations CMS and ATLAS the of behalf on ainaUniversit Sapienza S.Giagu β sn ieo-ih ehius xliigteeclettm eouino h unsses hscan This systems. muon the of resolution time excellent the exploiting techniques, time-of-flight using iRm n NNRm,015Rm,IT 00185-Roma, Roma, INFN and Roma di a ` β n h atcemmnu.ALSadCSaeal omauewith measure to able are CMS and ATLAS momentum. particle the and β n h osblt feprecn utpenceritrcin ihthe with interactions nuclear multiple experiencing of possibility the and , > β 0 β . n9%o h vns -arn nta r uhmore much are instead R-hadrons events. the of 99% in 7 ,adwt the with and ), β itiuinpaiga ihvle,so values, high at peaking distribution 6 4 Selection and results with early data -1 -1 Tk Tk

! 1.9 ! 2.2 10 CMS Preliminary CMS Preliminary 104 1.8 2 ~ 1.7 3 10 1.8 1.6 1

1.5 1.6 102 1.4 1.4 10-1 1.3 10 1.2 1.2

1 1 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 1.2 1.4 1.6 1.8 2 2.2 !-1 !-1 DT DT

1 1 Figure 1: Distribution of β− versus β− for background (left) and for the signal sample t˜1 with −1 Tk DT Figure 1: CMS: distributionmass 500of β GeV (right).as measured from dE/dx (y axis) and from time-of-flight (x axis), for background (left) and for the signal sample t˜ (right). selection). No background events pass the above selection in a fully-simulated MC sample cor- responding to 1 fb 1. The efficiency of the above selection for signal samples is reported in ∼ − Table 3. be done offline and also at trigger level, exploiting the fast time response of the Resistive Plate Chamber systems,

available in both ATLASIn the and case CMS where experiment. the tails of the two In additionβ distributions to theare higher time-of-flight when they are determination, measured on CMS performs also an independent measurementdata than what exploiting is expected the from specific simulation, ionization a tighter selection ( can) measured be used. The in goal the is centralto tracking system. β 1 dE/dx have less than one background event expected for 1 fb− . Several additional cuts can be used The usage of dE/dx allowswhich also drastically to reduce reduce theto abackground, negligible but level have thesmall background effects on the signal. due to They cosmic cannot muonsbe or to muons from different bunch crossings.properly Figure investigated 1 shows on the background distribution for the of availableβ−1 as statistics measured since fromthe simpledE/dx selectionversus the determination proposed above leaves no events. from time-of-flight, for muon background events and for a possible R-hadron signal (stop squark of 500 GeV mass). A clear correlation between the two measurement is visible in the case of the signal. 4.2 Tracker standalone selection A standalone selection using only the tracker for the β measurement is the following: 1 reconstructed muon with p > 100 GeV • T 1.3. DISCOVERY POTENTIALβ < 0.8. • tk The muon system is still used but no time-of-flight information is extracted from it. This anal- ysis has been tested on a sample of events expected to be selected by the muon trigger and cor- 1 In order to maximizeresponding the background to about 1 fbrejection,− of integrated CMS luminosity. uses a combined The resulting selection number of based selected on events the two measurements of β, reducing the backgroundis reported to a in negligible Table 4. The level efficiency without on signal significant samples is slightly loss of higher signal. than Using that of thisthe com- technique CMS is able to bined analysis given that the DT coverage, which is not required for the standalone selection, −1 keep the expected backgroundis limited to below the barrel one region event of CMS. for an integrated luminosity of 1 fb , maintaining reasonable signal efficiencies. The CMS integrated luminosity needed to observe 3 events in the signal region, for different models is 4.3 Systematic uncertainty shown in Figure 2. Heavy Stable Charged Particles can be discovered with early data for different models and in The first source of systematic uncertainty is the trigger efficiency for late particles. The final −1 different mass regions.muon The triggerstable settings for search time gates with and 1 synchronization fb is sensitive can change to gluino the efficiency masses for trigger-above 1 TeV and the GMSB scenarios with stable stauing on can a late be particle discovered in the correct with bunch a few crossing. 100 Thispb− can1 [2]. easily Similar change the sensitivities trigger efficiency are obtained also by the by 50% and can shift the spectrum of the recorded HSCP towards higher values of β, further ATLAS collaboration [3]. 8 4 Selection and results with early data -1 3 CMS Preliminary CMS Preliminary 103 2.5 102 2 Particles in 1 fb 10

) to observe 3 events 1.5 -1

(pb 1 int Gluino 1 L Stop -1 10 GMSB stau 0.5 KK 0 200 400 600 800 1000120014001600 0 200 400 600 800 1000 1200 Mass (GeV) Mass (GeV)

1 Figure−1 2: The left plot shows the integrated luminosity (pb− ) needed for 3 events, for the four Figure 2: CMS: integrated luminosity (pbsignal) needed models (gluino to observe full circles, 3 signal stop events, full squares, for the KK four tau empty signal circles, models stau (gluino, empty squares)stop, KK τ and stau) as a function of the particle mass.as a function The error of HSCP bands mass. correspond The right to plot a systematic shows the reconstructeduncertainty of mass 50% distribution on the estimated with 1 trigger efficiency. 1 fb− for two of the lowest cross section samples (300 GeV KK tau and 800 GeV stop).

would be recomputed using more data.

4.4 Discovery and exclusion In the following, we compute the luminosity needed to observe 3 events in the signal region for different models as an estimate for the integrated luminosity needed for exclusion/discovery when no background events are expected. If an excess is observed it is also possible to perform several cross checks. First, one can check the distribution of the reconstructed mass (as shown in the right plot of Figure 4.4).

The Poisson probability to observe no events when 3 are expected is 5%. Therefore, we can claim a 95% C.L. exclusion if no events are observed when the luminosity for 3 events has been accumulated.

Figure 4.4 shows the required luminosity to select three Heavy Stable Charged Particles for different signal samples. The error bars correspond to a systematic uncertainty of 50% on the trigger efficiency. For the tracker standalone analysis, the luminosity needed for a 95% C.L. exclusion has been computed with a likelihood ratio method [14], and the result is shown in Figure 4.4.

4.5 Conclusions Heavy Stable Charged Particles can be discovered with early data for different models and in 1 different mass regions. The stable gluino search with 1 fb− is sensitive to gluino masses above 1 1 TeV and the GMSB scenarios with stable stau can be discovered with a few 100 pb− . 2. NEUTRAL LONG-LIVED PARTICLES

A number of extensions of the result in particles that are neutral, weakly-coupled, and have macroscopic decay lengths that can be comparable with LHC detector dimensions. A large class of such models are represented by the Hidden Valley models [4], where the SM is extended by a hidden sector, the v-sector for short, and a communicator (or communicators) which interacts with both sectors. A barrier (perhaps the communicators high mass, weak couplings, or small mixing angles) weakens the interactions between the two sectors, making production even of light v-sector particles (v-particles) rare at low energy. At the LHC, by contrast, production of v-particles, through various possible channels, may be observable. The communicator can be any neutral particle or combination of particles, including the Higgs , the Z’ , or the SUSY LSP for example. Here we present the results of 0 a first study of the ATLAS detector performance for the Higgs decay h πvπv , where πv is a new massive neutral → particle with long lifetime (1.5 m in this study), that decay mainly in a pair of bottom .

2.1. DETECTOR SIGNATURES AND TRIGGER SELECTION

Figure 3: ATLAS event display for the process h → πvπv. One of the πv decays in the muon spectrometer, while the other decays in the hadron calorimeter.

0 ¯ ¯ A simulation of a typical h πvπv bbbb event in the ATLAS detector is shown in Figure 3. Due to the displaced → → vertices with tracks non pointing to the interaction region, and to the low Higgs mass (140 GeV in this simulation), the standard ATLAS triggers [5] are able to select only a very small fraction of these events (typical Level-1 trigger efficiencies are smaller than 5%). A signature driven trigger strategy is therefore required. We focused on two detector regions to illustrate the trigger signatures of Hidden Valley particles: decays in the muon spectrometer, and decays in the calorimeters. These are both the most challenging from the trigger point of view and the ones giving the most striking experimental signatures for a possible early discovery. Decays occurring near the end of the hadron calorimeter and before the first muon trigger plane result in a large number of hadrons traversing a narrow (η, φ) region of the muon spectrometer. The Level-1 muon trigger will return several clustered muon candidates, as shown in Figure 4, where the Level-1 muon candidates contained in a cone of radius ∆R = 0.4 around the πv line of flight, as a function of the πv radial decay distance, are plotted. As the πv decay vertex approaches the end of the hadron calorimeter (4500 mm), the average number of muon candidates contained in the cone plateaus at 3.5 until the πv ∼ decays close to the first trigger plane (7000 mm), at which point the charged hadrons are not spatially separated enough to give multiple unique muon candidates. This signature can be used as a stand-alone Level-2 trigger object to select with good efficiency these late decays. Events with decays near the outer edge of the electromagnetic calorimeter and in the hadron calorimeter, are characterized by jets with few or no tracks and unconventional energy distributions (jets with more energy deposited in the hadron calorimeter than in the electromagnetic one). The logarithm of the hadronic to electromagnetic energy ratio for jets from πv decays as a function of the πv decay distance can be seen in Figure 4. As the πv decays closer to the end of the electromagnetic calorimeter (2200 mm), the ratio changes from a characteristic negative to a positive value. Displaced decays in the outer part of the inner detector or inside the electromagnetic calorimeter result in low tracking efficiency, because tracking requires seed hits in the pixel and silicon strip layers. This suggests that a jet with no tracks reconstructed in the inner detector may be used to select πv decays in the electromagnetic calorimeter. In this case, to reduce QCD background, a Level-1 muon candidate contained in a cone of radius ∆R = 0.4 around the jet axis is required, which selects a semileptonic decay of one of the two b daughter of the πv particle. ) EM /E HAD Log(E

V-Sector Signal

!v radial decay distance (mm)

Figure 4: ATLAS: Left: average number of Level-1 muon candidates contained in a cone of ∆R = 0.4 around the πv line of

flight vs πv radial decay distance. Right: log10(EHAD/EEM ) vs πv decay distance.

2.2. TRIGGER PERFORMANCE

In Table I the ATLAS trigger acceptances for the new signature based triggers described above are shown. With 0 these new triggers ATLAS will be able to select about 20% of events with displaced decays from h πvπv. Standard → Model QCD processes are a potential source of significant background at the trigger level. The same trigger objects have been applied to fully simulated di-jet samples, resulting in a negligible (6 nb) cross-section acceptance at Level- 2. In conclusion long lived neutral particles predicted by Hidden Valley models, can be successfully collected by implementing dedicated signature based triggers, that allow to increase the selection efficiency with a negligible background rate from Standard Model processes.

Table I: Hidden Valley specific triggers efficiency, normalized to the whole sample.

log10(EHAD/EEM ) Trackless Jet Muon Cluster Total HV Triggers Total All Triggers 5% 3.8% 9% 15.7% 18.5 %

Acknowledgments

The author wish to thank Matt Strassler (Rutgers University) for invaluable contributions in many stages of the work related to the Hidden Valley scenario.

References

1 See for a review: M. Fairbain et Al., Phys. Rept. 438, 1 (2007). 2 The CMS Collaboration, CMS PAS EXO-08-003 (2008). 3 The ATLAS Collaboration, CERN-OPEN-2008-020, Geneva, 2008, to appear. 4 M. Strassler and K. Zureck, Phys. Lett.B, 661,263 (2008); M. Strassler and K. Zureck, Phys. Lett.B, 651,374 (2007). 5 The ATLAS Collaboration, G. Aad et al., 2008 JINST 3 S08003.