I tried several Ecut values and realized at least 30 Ryd is needed for 'slow' convergence with 'Cu.pbe-d-rrkjus.UPF', which is pretty going beyond my expectation. The best value for Cu bulk I've got is 3.673, approximately 1.7% larger than that of experimental a0. I appreciate Eyvaz and Aloysius for their valuable comments. Sampyo