Higher–twist Contributions to Fl and F2 from Instantons ∗

We study the twist–4 contribution to the unpolarized nucleon structure functions, F L and F 2 , in an approach where the vacuum structure of QCD is described by a dilute medium of instantons (¯ ρ/ ¯ R ≪ 1). The dominant power corrections come from the twist–4 quark–gluon operators and are of the order 1/(¯ ρ 2 Q 2). The contributions from four–quark operators are suppressed. The approach predicts large 1/Q 2 –corrections to F L , in agreement with the results of QCD fits to the data. The operator product expansion of QCD explains not only the scaling behavior of structure functions at asymptotically large Q 2 — including the logarithmic corrections, which can be computed using perturbation theory — but allows to classify also the power (1/Q 2 –) corrections to scaling. The latter are determined by the expectation values of certain operators of non-leading twist (> 2) in the target hadron state 1. While a fully quantitative theory combining logarithmic and power corrections is presently out of reach, it is interesting to study these higher–twist matrix elements in their own right. These are " universal " measures of the non-perturbative correlations of the quark and gluon fields inside hadrons, which appear also in other contexts, e.g. in the description of " intrinsic " heavy quark contributions to light hadron properties 2. Knowledge of these matrix elements would be useful for a more precise extraction of the leading–twist parton distributions from the DIS data, and for a better understanding of parton–hadron duality 3. Two basic types of twist–4 operators appear in the 1/Q 2 –power corrections to the second moments of the unpolarized structure functions (for details, see Refs. 1,7). One are quark–gluon operators, which describe the influence in DIS