Tracking missing energy

27 April 2010


A precise measurement of missing transverse energy ETmiss is essential for many physics studies with the ATLAS detector. It is particularly crucial for some new physics searches, such as supersymmetry or extra dimensions, where a large
ETmiss is a key final state signature. The ATLAS calorimeters play a vital role in the ETmiss measurement and have been designed to provide an excellent resolution at the LHC energy scale. However, several effects could either fake large ETmiss or degrade its resolution. Among them the presence of dead or noisy read-out channels, dead detector regions, poorly estimated inactive material, escaping particles or mis-measurement of their energies. This could make the estimation and the rejection of the background from physics processes with no real intrinsic transverse missing energy more challenging as the measured ETmiss would not reflect the one originating from the non-interacting particles.


To better evaluate and check the quality of the missing transverse energy measured by the calorimeters, one needs an independent measurement of a similar quantity using another sub-detector. The high performance ATLAS Inner Detector, with its three components: the pixel detector, the silicon micro-strip detector (SCT) and the Transition Radiation Tracker (TRT), offers such a possibility as it reconstructs the tracks of charged particles produced in the collision allowing the measurement of their momentum. We have used this feature to measure a new quantity, the missing transverse momentum
pTmiss, which is strongly correlated to the missing transverse energy in a proton-proton collision with non interacting particles such as neutrinos or neutralinos. The absence of such correlation in events with either fake or poorly measured ETmiss will allow rejecting them. This is of particular importance right from the beginning of the ATLAS data taking, as many Standard Model processes, such as W  → e/μ/τ + ν or t-quark physics, rely heavily on the rejection of QCD multi-jet events for early measurements. In addition,
pTmiss measurement from tracks offers other benefits which make it crucial for both low and high luminosity regimes:


  • Tracking and vertexing algorithms can associate tracks to each collision vertex. The
    pTmiss variable is calculated for each identified collision vertex. This feature is very important in the case of pile-up as many collision vertices may exist in one proton-proton bunch crossing event. Thus, vertex-by-vertex estimation of pTmiss is more accurate than the calorimeter-based ETmiss where all interactions are overlaid together in one single measurement.

  • pTmiss has a better sensitivity than the calorimeter-based ETmiss to reject cosmic-ray muons and beam halo background, since most of their tracks have different time and geometry signatures compared to tracks from the collisions.

However, its performances are mainly limited by the Inner Detector smaller geometrical coverage and its inability to measure neutral particles momentum. The presence of direct photons or neutral pions would leads to a larger pTmiss than ETmiss for the same event. Hence the correlation between the two variables will be somehow weaker, but not destroyed, as only a fraction of the energy of the event is not measured with the Inner Detector while it is still seen in the calorimeters. This means that even in this case, the
pTmiss-ETmiss correlation can still be used, although with a lower efficiency, to reject background with fake ETmiss.

After implementing this new method and some feasibility tests with ATLAS Monte Carlo events, the real test was to check its performance with collision data. We have used the 900 GeV proton-proton collisions delivered by the LHC in December 2009 for the first studies. It was a great achievement that the ATLAS Inner Detector delivered high quality reconstructed tracks and collision vertices for all these early data. Events were selected according to all ATLAS data quality criteria in order to have the best and most stable operating environment and the cleanest detector responses. For each event, only high quality tracks reconstructed with both the pixel and the SCT detectors, which ensure a good resolution on their measured momentum, are kept for the pTmiss measurement. The presence of a well reconstructed primary vertex is also required for an event to be accepted. The missing transverse momentum x and y components are then calculated as:

equation

The figure below shows the total missing transverse momentum calculated from the momentum of the tracks reconstructed with the ATLAS Inner Detector for proton-proton collision at 900 GeV. Comparisons to expectations from the ATLAS detector simulation with minimum bias events using the PYTHIA Monte Carlo generator show a good agreement at this low energy. It is a very promising result and it should lead to a better understanding of the pTmiss performances in the 7 TeV collision data and its use in ATLAS early data physics analysis.


pTmissdistribution measured by the ATLAS Inner Detector. Only tracks that pass track quality cuts are used in its calculation. All distributions are normalized to the number of events in data.



 

 

Rachid Mazini

Academia Sinica