Measurement of the fraction of $t\overline{t}$ production via gluon-gluon fusion in $p\overline{p}$ collisions at $\sqrt{s}=1.96\mathrm{TeV}$

We present a measurement of the ratio of the $t\overline{t}$ production cross section via gluon-gluon fusion to the total $t\overline{t}$ production cross section in $p\overline{p}$ collisions at $\sqrt{s}=1.96\mathrm{TeV}$ at the Tevatron. Using a data sample with an integrated luminosity of $955{\mathrm{pb}}^{-1}$ recorded by the CDF II detector at Fermilab, we select events based on the $t\overline{t}$ decay to $\mathrm{\text{lepton}}+\mathrm{\text{jets}}$. Using an artificial neural network technique we discriminate between $t\overline{t}$ events produced via $q\overline{q}$ annihilation and $gg$ fusion, and find $G_f=\sigma (gg\to t\overline{t})/\sigma (p\overline{p}\to t\overline{t})<0.33$ at the 68% confidence level. This result is combined with a previous measurement to obtain the most stringent measurement of this quantity by CDF to date, $G_f={0.07}_{-0.07}^{+0.15}$.

Figure 1

Distributions of $\cos {\theta }^{*}$ for events with one $b$ tag in data, expected $t\overline{t}$ and background. The distribution of $t\overline{t}$ plus background uses the ratio of $gg$ to total $t\overline{t}$ obtained from the fit of $G_f^{\mathrm{fit}}=-0.075$. We also show the expected distributions for $gg$-only and $q\overline{q}$-only hypotheses where background is included. The error bars on the total $t\overline{t}+\mathrm{\text{background}}$ includes the statistical uncertainty from Poisson statistics.

Figure 2

Feldman-Cousins bands for the combination of the analyses with statistical and systematic uncertainties for 68% C.L. and 95% C.L. In the data, we find $G_f^{\mathrm{fit}}=0.073$, which yields $G_f={0.07}_{-0.07}^{+0.15}$ and $G_f<0.38$ at the 95% C.L.