Besides what Suhas mentioned, your optimization idea seems to based on the notion of tuple-wise join algorithms.
That's not how joins are processed with column store tables.
The join algorithms and their implementation are developed to optimally support the requirements of typical SAP HANA applications, such as BW on HANA or S/4 HANA.
Choosing a data type that is somewhat "more internal" won't speed up anything really.
- Lars