b-bbar Dijet Production using SVT
Abstract

We present a b-bbar jet cross section measurement based on about 260 pb-1 of data, collected by CDF Run II until September 2004. The analysis strongly relies on the CDF detector good tracking capabilities both at trigger level, as data is selected requiring two displaced tracks at Level 2, and offline, since b-tagging is performed reconstructing secondary vertices inside the jet. Jets are reconstructed using a cone algorithm. The cross section is measured in the central region (|eta|< 1.2) as a function of leading jet Et, b-bbar pair invariant mass and azimuthal angle between the two jets. Results are corrected to the hadron level and compared to Leading Order Monte Carlos (Pythia and Herwig) as well as MC@NLO predictions.


Tagging efficiency

The events are selected using a trigger, specifically designed to select events rich in heavy flavour, which requires two high transverse energy jets (Et>20 GeV) associated to two displaced tracks (impact parameter |d0|>120 μm) reconstructed using the SVT at Level 2. The presence of two such tracks associated to two jets requires a careful study on how the particular trigger selection shapes the events. Offline, jets are tagged using the 2 dimensional SecVtx algorithm: the secondary vertex inside the jet is reconstructed and the jet is positively tagged if the distance of the secondary vertex to the primary vertex of the interaction is positive with respect to the jet axis.

We define "SVT b-tagged" jets in such a way that events with two such objects always pass the trigger. The advantage of such a choice is that it allows to reasonably neglect the problem of calculating the trigger efficiency, measuring instead trigger and b-tagging efficiency in one single step in the Monte Carlo and then rescale it to data. The main requirements that define the "SVT b-tagged" jet are:

  • Jet tagged by SecVtx algorithm with Et>30 GeV and |eta|<1.2
  • Jet associated to a SVT track with Pt>2 GeV and |d0|>120 μm
  • The efficiency for requiring two SVT b-tagged jets in the event is calculated in Pythia MonteCarlo and corrected for the scale factor of 1.029 +/- 0.009 (stat) +/- 0.034 (syst). The result is shown in the figures below:

    Efficiency as a function of Et

    Efficiency as a function of M

    Efficiency as a function of Delta phi

    Efficiency as a function of the leading jet Et (eps) Efficiency as a function of the di-jet invariant mass (eps) Efficiency as a function of the di-jet azymuth distance (eps)

    b-jets purity

    The invariant mass of the tracks associated to the secondary vertex can be used to separate the contribution due to heavy flavour jets from light quark and gluon jets. A full reconstruction of the invariant mass would allow a precise separation of the b-jet sample from the background but the presence of neutral particles and the ambiguities in associating tracks to primary or secondary verices that are due to finite detector resolution contribute in limiting the reconstruction of the original b mass. Nevertheless the shape of the mass distribution is different according to the flavour of the jet and mass templates can be built in the MonteCarlo and used to fit the data.

    The templates are chosen to be the sum of the two SVT b-tagged jets secondary vertex masses. A two components fit is performed, using a "signal" template describing the b-bbar case and a "background" template merging all the other possible contributions (cc, bc, cl, etc..) whose relative weight is taken from MonteCarlo (systematic uncertainties being assigned to this assumption).

    The picture below shows the fit over the full jet Et range (eps):

    Fit

    The resulting b-bbar purity for 2 SVT b-tagged jets is quite high as shown in the figures below:

    Purity as a function of Et

    Purity as a function of M

    Purity as a function of Delta phi

    bb jet purity as a function of the leading jet Et (eps) bb jet purity as a function of the di-jet invariant mass (eps) Purity as a function of the di-jet azymuth distance (eps)

    Systematic uncertainties

    Sources of systematic uncertainties are related to: luminosity (a standard 6% value is applied); jet energy corrections (the highest contribution, around 13-20%); SVT b-tagging efficiency and scale factor; b-jet purity determination (second highest contribution related to secondary vertex mass templates sensitivity to tracking inefficiency inside jets).

    Total systematic uncertainty as a function of Et

    Total systematic uncertainty as a function of di-jet invariant mass

    Total systematic uncertainty as a function of Delta phi

    Total systematic uncertainty as a function of the leading jet Et (eps) Total systematic uncertainty as a function of the di-jet invariant mass (eps) Total systematic uncertainty as a function of the di-jet azymuth distance (eps)

    Results

    The differential cross section unfolded to the hadron level is shown in the figures below as a function of the leading jet Et, the di-jet invariant mass and the azimuth distance between the two jets. The measurement is compared to Leading Order Monte Carlos (Pythia Tune A and Herwig) and to Next-to-Leading Order predictions from MC@NLO. The MC@NLO sample is generated using a minimum quark pt of 10 GeV in |eta|<1.75, CTEQ6.1M PDFs, and μR = μF = sqrt(pt^2 + m^2). MC@NLO and Herwig samples are produced with or without Jimmy to simulate multiple parton interactions. The total cross is reported in the table at the end.

    *** Differential cross section plots: Data is compared to Pythia Tune A, Herwig + Jimmy and MC@NLO + Jimmy

    Cross section as a function of Et

    Cross section as a function of M

    Cross section as a function of Delta phi

    Differential cross section as a function of the leading jet Et (eps) Differential cross section as a function of the di-jet invariant mass (eps) Differential cross section as a function of the di-jet azymuth distance (eps)

    *** Data / MC ratio as a function of the di-jet azymuth distance (eps):

    Cross section as a function of Delta phi


    *** Di-jet azymuth distance: additionnal comparisons

    Cross section as a function of Delta phi

    Cross section as a function of Delta phi

    Delta Phi: Data is compared to MC@NLO (eps) Delta Phi: Data is compared to Herwig (eps)

    Data/MC as a function of Delta phi

    Data/MC as a function of Delta phi

    Delta Phi: Data / MC@NLO ratio (eps) Delta Phi: Data / Herwig ratio (eps)

    *** Total cross section: Data is compared to Pythia (Tune A), MC@NLO + Jimmy and Herwig + Jimmy predictions

    Total cross section