Measuring environmental innovation patents

Dear Stata experts,

For my master thesis, I'm figuring out how to analyse in order to measure environmental innovation patents.
I'm using IPC classes from the NBER dataset. But I struggle to prepare my data so if anyone can help me would be a blessing!

I want to keep a certain amount of different IPC classes which I use as environmental patents. But this means I need to get rid of all the other observations(patent types /IPC classes).
So I tried to do this with drop if, egen, bysort egen, but it all didn't work sadly. The tricky part is the IPC classes are string variables. I tried to use destring, but due to non-numeric observations (B01C example of a class) i can't use Destring or generate. I tried to fix this with force but this only gerenates dots, so I'm kinda lost how to proceed.

I need to sort and keep only a few specified IPC classes so I can merge that prepared data set with other data sets.

Thanks in advance and sorry if this is kind of a beginner/easy question.