The concept of thermal efficiency of friction stir welding was first proposed by Schmidt and Hattel at 2008. The authors were trying to evaluate the real amount of heat that is conducted to the workpiece compared to the supposed total heat generation. For that purpose the authors used a formula to predict the thermal efficiency value and the value was ranging between 0.7-0.9.
You can read the following article to know more about how you can predict this value.
Thermal efficiency factor depends on your tool and workpiece materials thermo-physical properties. Nandan et al. 2008 a review paper published in progress in materials is clearly talked about this. If you are taking about steel tool and aluminum alloy workpiece alloy, it ranges between 0.6 to 0.8.
I think it will be removed automatically when you will perform next welding using same tool. Otherwise, you can apply some flux (liquid used during brazing for wetting) on the tool (where the aluminium stuck) and then perform the welding. I am not sure but this may work.
I don't know whether there is any chemical solution available or not for such use.