Abstract
In this paper we study the double spin asymmetry $A_{LT}^{\cos(\phi_h-\phi_S)}$ of a single hadron production in semi-inclusive deep inelastic scattering (SIDIS) within the framework of transverse momentum dependent (TMD) factorization. The asymmetry is contributed by the convolution of the worm gear $g_{1T}$ and the unpolarized fragmentation function $D_1$. Similar to Sivers function case, the worm gear function in the coordinate space in the perturbative region can be represented as the convolution of the $C$-coefficients and the corresponding collinear correlation function. The Wandzura-Wilczek approximation is used to obtain this correlation function. We perform a detailed phenomenological numerical analysis of the $A_{LT}^{\cos(\phi_h-\phi_S)}$ asymmetry in the SIDIS process within TMD factorization at the kinematics of the HERMES measurements. It is found that the obtained $x_B$-, $z_h$- and $P_{h\perp}$-dependent asymmetries are basically consistent with the HERMES measurements.
Introduction
Mapping the three-dimensional internal structure of the nucleon is a primary goal of high-energy physics. While traditional collinear factorization models describe the longitudinal momentum fraction ($x$) of quarks, they are insufficient for explaining various azimuthal asymmetries observed in experiments. To address this, researchers utilize Transverse Momentum Dependent (TMD) factorization, which incorporates the quarks' transverse momentum ($k_\perp$).
This study focuses on the Kotzinian-Mulders (KM) effect, a subtle spin-dependent phenomenon characterized by the double spin asymmetry $A_{LT}^{\cos(\phi_h-\phi_S)}$. This effect involves the $g_{1T}$ function, often called the "worm-gear" function, which describes the probability of finding a longitudinally polarized quark within a transversely polarized nucleon. Despite being a fundamental "leading-twist" property, $g_{1T}$ has historically received less attention than the Sivers or Boer-Mulders functions.
Theoretical Framework and Methods
The authors employ TMD factorization to decompose the scattering cross-section into a convolution of a TMD Parton Distribution Function (TMD-PDF) and a TMD Fragmentation Function (TMD-FF). A critical component is the CSS (Collins-Soper-Sterman) evolution, which allows for the scaling of TMD functions from a low initial energy to the high energy scales ($Q^2$) typical of experiments. This evolution is governed by the Sudakov form factor, accounting for gluon radiation.
Since $g_{1T}$ has not been directly extracted from experimental data, the Wandzura-Wilczek (WW) approximation is used to relate it to the better-known quark helicity distribution ($g_1$). Calculations are performed in coordinate space ($b$-space) to simplify the complex momentum-space convolutions into products. For numerical implementation, the authors modified the QCDNUM evolution package to include the specific kernels required for the KM function.
Numerical Analysis and Results
The theoretical predictions were compared against experimental data from the HERMES and COMPASS collaborations. The study utilized the MSTW2008 set for unpolarized distributions and DSS sets for fragmentation functions.
A key finding is that TMD evolution significantly impacts the shape and magnitude of the KM asymmetry, particularly in the small-$x$ region. The calculated asymmetries for pions ($\pi^+, \pi^-, \pi^0$) and kaons ($K^+, K^-$) are found to be basically consistent with HERMES measurements.
As shown in the figures, the asymmetry tends to increase with the transverse momentum of the produced hadron ($P_{h\perp}$), reaching peaks around 0.1 in specific channels. Furthermore, the absolute maximum of the TMD distributions grows with the energy scale, highlighting the importance of using correct evolution equations when comparing different experimental results.
Conclusion
The success of the Wandzura-Wilczek approximation in matching experimental trends suggests it is a reliable proxy for the $g_{1T}$ function until more direct experimental extractions are available. However, the authors note that current data from HERMES and COMPASS still possess large statistical uncertainties, which prevents making strong conclusions. Future data from the Electron-Ion Collider (EIC) will provide the precision necessary to extract the $g_{1T}$ function directly and test these TMD factorization predictions to a much higher degree of accuracy.
Summary updated