Joint friction has a significant influence on the dynamics and motion control of manipulators. However, the friction effect is often omitted or simplified in previous studies. In this paper, we establish the dynamics model of a single joint in a rotating industrial manipulator taking detailed friction effects into consideration and propose a new control algorithm for the friction compensation purpose. Firstly, the manipulator dynamics modeling is carried out employing a recently-proposed extended static friction model, which depicts load and temperature influence on Coulomb, Stribeck and viscous terms. Moreover, based on the established dynamics model, the paper presents a new adaptive fast nonsingular terminal sliding mode (AFNTSM) controller. The proposed approach has the advantages of continuous control inputs, fast convergence rate, no singularity and great robustness against disturbances. Furthermore, its adaptive property does not require any prior knowledge of the upper bound of the uncertainties. Finally, the proposed controller is applied to the manipulator joint trajectory tracking problem with varying friction subject to load and temperature changes. The numerical simulation verifies the effectiveness of our proposed method and its advantages over other controllers.