Database Software

Source Database: AMERICAN_AMERICA.US7ASCII
Target Database: AMERICAN_AMERICA.AL32UTF8

From the source database, the chinese characters are stored in some schema table. From the csscan result, there are convertiable, truncate, data lossy character. So, I have tried to use exp/imp for the conversion. However, all chinese characters are invalided and cannot be read anymore. How can I convert them from US7ASCCI to UTF8 database?

Also, I have tried build up another database with AMERICAN_AMERICA.ZHT16MSWIN950. The exp/imp is used for conversion again. The chinese characters are readable in AL32UTF8 database.

- source database (US7ASCII)
export NLS_LANG=AMERICAN_AMERICA.US7ASCII
export LANG=AMERICAN_AMERICA.US7ASCII
exp userid='/ as sysdba' file=export.dmp full=y

- target database (AL32UTF8)
export NLS_LANG=AMERICAN_AMERICA.US7ASCII
export LANG=AMERICAN_AMERICA.US7ASCII
imp userid='/ as sysdba' file=export.dmp full=y ignore=y


Result:
from US7ASCII to AL32UTF8:
the chinese characters cannot be read


from US7ASCII to ZHT16MSWIN950:
the chinese characters cannot be read


from ZHT16MSWIN950 to AL32UTF8:
the chinese characters can be read

How can I convert the chinese character from US7ASCCI to UTF8 database?