High-$x$ structure function of the virtually free neutron

The pole extrapolation method is applied to the semi-inclusive inelastic electron scattering off the deuteron with tagged spectator protons to extract the high-$x$ structure function of the neutron. This approach is based on the extrapolation of the measured cross sections at different momenta of the spectator proton to the nonphysical pole of the bound neutron in the deuteron. The advantage of the method is in the possibility of suppression of the nuclear effects in a maximally model-independent way. The neutron structure functions obtained in this way demonstrate a surprising $x$ dependence at $x\ge 0.6$ and $1.6\le Q^2\le 3.38{\mathrm{GeV}}^2$, indicating a possible rise of the neutron-to-proton structure functions ratio. If the observed rise is valid in the true deep inelastic region then it may indicate new dynamics in the generation of high-$x$ quarks in the nucleon. One such mechanism we discuss is the possible dominance of short-range isosinglet quark-quark correlations that can enhance the $d$-quark distribution in the proton.

Figure 1

IA contribution to $h_1+A\to h_2+C+X$ reaction, with $C$ acting as a spectator.

Figure 2

IA (a) and FSI (b) contributions to reaction (1).

Figure 3

Ratio $R$ of the BONuS data to a plane-wave model (see Ref. [21] for details) as a function of spectator $cos{\theta }_s$ compared to our VNA IA (black dashed curve) and FSI (dotted blue curve) calculation for $E_{\text{beam}}=4.23$ GeV. An overall normalization factor was fitted for each $p_s$ value in the model to the FSI calculation; see text for details. The IA calculation is shown using the same normalization factor.

Figure 4

As Fig. 3 but for $E_{\text{beam}}=5.27$ GeV.

Figure 5

$F_{2n}$ to $F_{2p}$ ratio obtained using the pole extrapolation applied to the renormalized BONuS data. Systematic errors are depicted as open error bars. The dotted black and dashed red curves show the ratio obtained with $F_{2n}$ parametrization of Ref. [26]. The $F_{2p}$ values are estimated using the fit of Ref. [27].