I am looking for references where people were able to collect 128-256 channel EEG data during face perception task and the map it onto FFA or OFA using source localization algorithms?
The N170 (the face selective component of the brain) is commonly mapped indeed to OFA or FFA. It's trivial to find examples for this, e.g.: this, this, this, this, this. Many more studies can be found by searching the relevant data bases, e.g. like this. Some of these include co-registered EEG/fMRI or MEG. There is some uncertainty of where exactly it originates from - OFA, FFA, both.
Note that dipole fits can be seen as leading to oversparse solutions - they might locate the N170 to one out of OFA or FFA, when it is actually projected by a larger are including OFA, FFA, and possibly even more of the cortex.