Thanks for the response, Chris.

I was able to solve my problem. I eventually started looking at each string in the process to trace the resulting encoding using a hex editor to see what characters were actually there.

The root causes of my issues are as follows:

1) In the string storing the template email file being read, I was noticing some extra bytes being inserted in the UTF-8 characters, such as "83 c2". This led me to an article outlining how the string had likely been "double-UTF8 encoded," which led me to test and see that the UTF8 file being loaded was seen by PowerBuilder as ANSI encoded. My temporary solution for testing is to read the file into a Blob (which PB thinks is ANSI encoded) and then reading that Blob to string as if it were already UTF8 encoded:

        string ls_email_template_path
        integer li_FileNum
        long ll_file_bytes
        blob lb_ansi
        ls_email_template_path = "c:\path\to\template.html"
        li_FileNum = FileOpen(ls_email_template_path, TextMode!, Read!, Shared!)
        ll_file_bytes = FileReadEx(li_FileNum, lb_ansi)
        //FORCE TO UTF8 as the file is being read as ANSI when it is actually UTF8 encoded
        ls_email_body = string(lb_ansi, EncodingUTF8!)
        FileClose(li_FileNum)

The long-term solution is to instead ensure the file templates are UTF8-BOM encoded to remove ambiguity when reading them.

2) The next issue was that while all of the characters were encoded consistently, they were also displayed incorrectly in the resulting email. I believe this issue is related to our SMTP pipeline, and as a workaround I am actually "double-encoding" the HTML message text in UTF8.

        //Workaround - double-encode the text as UTF8
        blob lb_msg
        lb_msg = Blob(ls_email_body, EncodingUTF8!)
        ls_email_body = String(lb_msg, EncodingANSI!)

Since it works, it tells me that somewhere in the pipeline the message text is being re-encoded incorrectly, but that is an issue for another day.