Your own tests with conditions closely simulating the real world application are the only way you'll even get close to a reasonably accurate friction coefficient. I am familar with aluminum. Published values are 0.3 to 1.0 and larger. In tests, with loads representative of the real world application, I saw friction coefficients from 0.4 to 0.9; the geometry had only tiny variations (surface roughness was consistent, the geometry dimensions of the two bodies in contact were the same for all tests) and the only variations were on the contact load applied to the opposing cylindrical pads, and the fatigue load of the test coupon (varied from 16 to 28 ksi). With such variations, it is almost a requirement to do your own tests if friction coefficient has a substantial effect on your design (it might be that varying from 0.4 to 1 has little effect of course).