Since we provide a IRIS ADO provider, that should be used rather than using Microsoft's ODBC provider.  On Unix, Microsoft is linked to a version of unixODBC driver manager that is compiled to use 2 byte Unicode and therefore you must also use a ODBC driver that matches how unixODBC is built.  The Microsoft ODBC Provider is not flexible with how it retrieves Unicode, so UTF-8 and UTF-32 drivers that we have don't work for their DotNet implementation.  If you use our IRIS provider, there are no issues fetching Unicode data, and you avoid all the extra overhead of jumping between managed (DotNet) code and unmanaged code (ODBC driver manager and driver).