Arabic_CI_AS, Arabic_100_CS_AS_SC_UTF8 to name two. Any collation starting with Arabic should do.
I have not tested, but I would expect an Arabic collation to handle English correctly. Since the two language uses different scripts, there is not really any conflicts, so Latin characters will be sorted and handled in a "standard" (i.e. English) way. The only reason it would not be that way, would be if Arabic imposes any particular order on Latin letters. (Which I would not expect, but I don't know Arabic.)
I would recommend that you use nvarchar for string data, but varchar should do as well, at least to 99.5% for English. English has words like café, résumé, naïve, rôle, coöperate, but these spellings are not common.