5.2 Resonances Coupling to Third Generation Quarks
147
mediators in s-channel processes mediating tt production [819]. Future searches for
these non-resonant BSM models will need dedicated techniques to constrain the SM
tt background in the tails of the distributions in order to achieve optimal sensitivity.
5.2.2 t b Resonances
Heavy charged gauge bosons W
, decaying to tb (and charge conjugates), can be
searched for in +jets and all-hadronic final states. Single top quark production in the
s-channel constitutes the same signature and is part of the non-resonant background
of this search. In fact, for a W
boson with left-handed couplings there is interference
between SM s-channel tb production and tb production through an intermediate
W
. This interference needs to be taken into account, by simulating resonant and
non-resonant tb production simultaneously. In the leptonic channel, with a branching
fraction of 25.3%, a lepton is produced in close angular proximity to the b jet from
the t decay. The resulting lepton-jet-p
miss
T
system is balanced by a high- p T b jet. The
νbb final state can also be produced by tt and W +bb production, which constitute
irreducible backgrounds in this search. In the all-hadronic channel, with a branching
fraction of 67.4%, the t and b jet pair resembles a high-mass dijet system. Hence,
multijet production is the largest background in this channel, with an irreducible
component from tt production.
The first searches for W
→ tb have been performed with 7 TeV data in the +jets
channel by ATLAS [820] and CMS [821]. These early searches considered isolated
leptons and one or two b-tagged jets. Masses below about 1.85 TeV could be excluded,
considering right handed couplings and a branching fraction into tb of one. Similar
analyses on 8 TeV data by ATLAS [822] and CMS [823] improved this limit to
2.05 TeV. It should be noted that the ATLAS analysis requires two b-tagged jets,
which results in a much smaller signal efficiency for high resonance masses, such
that the ATLAS upper cross section limits are about a factor of 7 weaker than the
ones by CMS. Analyses in the all-hadronic final state have been made possible with
the advent of top tagging algorithms, and have been performed by ATLAS [824]
and CMS [825] on 8 TeV data. ATLAS uses a top tagging algorithm based on the
splitting scale
√
d 12 , and the N -subjettiness ratios τ 32 and τ 21 . No selection on the
jet mass is applied. This choice results in a t-tagging efficiency of about 50% with a
suboptimal misidentification rate of about 10% for light quark, b and gluon jets. The
background is estimated by a fit of a parametric function to data, where an exponential
function has been found to describe the shape of the background distribution. The
CMS analysis uses the CMSTT with τ 32 < 0.55 and subjet b tagging. The chosen
working point results in a t-tagging efficiency of about 25% with a misidentification
rate of 0.3%, which is more than an order of magnitude better than in the ATLAS
analysis. The background is estimated from data in sideband regions, obtained by
inverting individual steps of the substructure selection. The final m tb distribution is
described well by the predicted backgrounds from multijet, tt and to a lesser degree,
single t production. Due to the better t tagging algorithm, the background is about
147
mediators in s-channel processes mediating tt production [819]. Future searches for
these non-resonant BSM models will need dedicated techniques to constrain the SM
tt background in the tails of the distributions in order to achieve optimal sensitivity.
5.2.2 t b Resonances
Heavy charged gauge bosons W
, decaying to tb (and charge conjugates), can be
searched for in +jets and all-hadronic final states. Single top quark production in the
s-channel constitutes the same signature and is part of the non-resonant background
of this search. In fact, for a W
boson with left-handed couplings there is interference
between SM s-channel tb production and tb production through an intermediate
W
. This interference needs to be taken into account, by simulating resonant and
non-resonant tb production simultaneously. In the leptonic channel, with a branching
fraction of 25.3%, a lepton is produced in close angular proximity to the b jet from
the t decay. The resulting lepton-jet-p
miss
T
system is balanced by a high- p T b jet. The
νbb final state can also be produced by tt and W +bb production, which constitute
irreducible backgrounds in this search. In the all-hadronic channel, with a branching
fraction of 67.4%, the t and b jet pair resembles a high-mass dijet system. Hence,
multijet production is the largest background in this channel, with an irreducible
component from tt production.
The first searches for W
→ tb have been performed with 7 TeV data in the +jets
channel by ATLAS [820] and CMS [821]. These early searches considered isolated
leptons and one or two b-tagged jets. Masses below about 1.85 TeV could be excluded,
considering right handed couplings and a branching fraction into tb of one. Similar
analyses on 8 TeV data by ATLAS [822] and CMS [823] improved this limit to
2.05 TeV. It should be noted that the ATLAS analysis requires two b-tagged jets,
which results in a much smaller signal efficiency for high resonance masses, such
that the ATLAS upper cross section limits are about a factor of 7 weaker than the
ones by CMS. Analyses in the all-hadronic final state have been made possible with
the advent of top tagging algorithms, and have been performed by ATLAS [824]
and CMS [825] on 8 TeV data. ATLAS uses a top tagging algorithm based on the
splitting scale
√
d 12 , and the N -subjettiness ratios τ 32 and τ 21 . No selection on the
jet mass is applied. This choice results in a t-tagging efficiency of about 50% with a
suboptimal misidentification rate of about 10% for light quark, b and gluon jets. The
background is estimated by a fit of a parametric function to data, where an exponential
function has been found to describe the shape of the background distribution. The
CMS analysis uses the CMSTT with τ 32 < 0.55 and subjet b tagging. The chosen
working point results in a t-tagging efficiency of about 25% with a misidentification
rate of 0.3%, which is more than an order of magnitude better than in the ATLAS
analysis. The background is estimated from data in sideband regions, obtained by
inverting individual steps of the substructure selection. The final m tb distribution is
described well by the predicted backgrounds from multijet, tt and to a lesser degree,
single t production. Due to the better t tagging algorithm, the background is about