QuantumEspresso was born at a time when norm conserving pseudopotentials were definitely more reliable. For this reason certain codes were developed for norm-conserving pseudopotentials only. At the moment, there is little interest to extend some postprocessing code that composes the QE suite to USPP as there are norm-conserving pseudos that do not require large cut-offs.

So to answer your question, you need to use nc pseudopotentials for doing optics, and I indeed would recommend using NC normally.

Cheers,

Roberto

Thank you. I have find nc pseudopotential files from the researchgate comments you added. And quantum espresso code run it with these pseudopotentials. Many thanks again Roberto D'Agosta