PX Jobs can use any delimiter; I think that the issue you are seeing is that you are putting the delimiter character in the wrong location in the sequential file definitions. I'm not at a client now so can't do an exact walkthrough, but if you start with a new sequential file stage (to make sure that other changes you might have made don't affect the outcome) and just change the field delimiter from a pipe to your u-umlaut and post the error message, if any.
Use "od" to see what the delimiter character in the source file really is after the transfer to UNIX, then specify that character. This is the probable cause. To test delimeters, just choose a letter or displayable character in your file and set that as delimiter, your error message should now move to another column.
I'll try this out but, in all honesty, I don't think it is the file. I can view the file with no issue in a DataStage server job. If the delimiter was incorrect it would not view correctly in the Server job either.
Are you using NLS? It is just that I have used odd delimiters without issue in both Server and Parallel jobs, thus I think your error is not something inherent in DataStage but in your definitions or environment somewhere.
Could you try using a letter that is found in the file as a delimiter to see if your error shifts from the first column to the second (or subsequent) columns?
</a>