Precision Measurement of the Neutron Spin Asymmetry$A_1^n$ and Spin-Flavor Decomposition in the Valence Quark Region

We have measured the neutron spin asymmetry $A_1^n$ with high precision at three kinematics in the deep inelastic region at $x=0.33$, $0.47$, and $0.60$, and $Q^2=2.7$, $3.5$, and $4.8(\mathrm{G}\mathrm{e}\mathrm{V}/c{)}^2$, respectively. Our results unambiguously show, for the first time, that $A_1^n$ crosses zero around $x=0.47$ and becomes significantly positive at $x=0.60$. Combined with the world proton data, polarized quark distributions were extracted. Our results, in general, agree with relativistic constituent quark models and with perturbative quantum chromodynamics (PQCD) analyses based on the earlier data. However they deviate from PQCD predictions based on hadron helicity conservation.

Figure 1

Our $A_1^n$ results compared with theoretical predictions and existing data obtained from a polarized ${}^3\mathrm{H}\mathrm{e}$ target [37, 38, 39]. Curves: predictions of $A_1^n$ from SU(6) symmetry (zero) [3], constituent quark model (shaded band) [6], and statistical model (long-dashed) [23]; predictions of $g_1^n/F_1^n$ from PQCD HHC based BBS parametrization (higher solid) [10] and LSS(BBS) parametrization (dashed) [11], bag model with the effect of hyperfine interaction but without meson cloud (dash-dotted) [20], LSS 2001 NLO polarized parton densities (lower solid) [21], and chiral soliton model (dotted) [22].

Figure 2

Results for $(\Delta u+\Delta \overline{u})/(u+\overline{u})$ and $(\Delta d+\Delta \overline{d})/(d+\overline{d})$ in the quark-parton model, compared with HERMES data [41], the RCQM predictions [6], predictions from LSS 2001 NLO polarized parton densities [21], the statistical model [23], and PQCD-based predictions incorporating HHC [11]. The error bars of our data include the uncertainties given in Table 2. The dark-shaded error band on the horizontal axis shows the uncertainty in the data due to neglecting $s$ and $\overline{s}$ contributions. The light-shaded band shows the difference between $\Delta q_V/q_V$ and $(\Delta q+\Delta \overline{q})/(q+\overline{q})$ that needs to be applied to the data when comparing with the RCQM calculation.