If you are referring special characters such as universal code(unicode), then you will have to use nvarchar, or nchar which take twice the disk space as usual characters, non-unicode.
I am sure sorting will not be that faster using clustered indexes too and might have too many acans on your execution plans