Leading twist nuclear shadowing: uncertainties, comparison to experiments and higher twist effects

Using the leading twist approach to nuclear shadowing, which is based on the relationship between nuclear shadowing and diffraction on a nucleon, we calculate next-to-leading order nuclear parton distribution functions (nPDFs) and structure functions in the region $0.2 > x > 10^{-5}$ and $Q^2 \geq 4$ GeV$^2$. The uncertainties of our predictions due the uncertainties of the experimental input and the theory are quantified. We determine the relative role of the small ($\sim Q^2$) and large ($\gg Q^2$) diffractive masses in nuclear shadowing as a function of $x$ and find that the large mass contribution, which is an analog of the triple Pomeron exchange, becomes significant only for $x \le 10^{-4}$. Comparing our predictions to the available fixed-target nuclear DIS data, we argue, based on the current experimental studies of the leading twist diffraction, that the data at moderately small $x\sim 0.01$ and $Q^2 \sim 2$ GeV$^2$ could contain significant higher twist effects hindering the extraction of nPDFs from that data. Also, we find that the next-to-leading order effects in nuclear shadowing in the ratio of the nucleus to nucleon structure functions $F_2$ are quite sizable. Within the same formalism, we also present results for the impact parameter dependence of nPDFs. We also address the problem of extracting of the neutron $F_{2n}(x,Q^2)$ from the deuteron and proton data. We suggest a simple and nearly model-independent procedure of correcting for nuclear shadowing effects using $F_2^A/F_2^D$ ratios.