Need your help. BGR test result analysis

[rock_zhu]

Hi all.
My BGR was fabed out recently. The spec output voltage is 1.2V . But the results show Vref distribution as gaussion from 1.0V to 1.4V. This BGR was fabed in a new process which combines with a cmos process and some 5V transsitors. This BGR's architecture was verified by other process and shows better gaussion from 1.1~1.3V.
Is the circuit problem or process variation problem?
How can I analysis thedistribution problem like this?

Thanks in advance.

 

[dhasmana]

check the Avt parameter (pelgrom coefficient) for the transistors in the original models and the extracted parameters (if wafer health report is available). This increase in variation should be because of the increase in mismatch.

Alternatively the current mirrors might not be operating in sufficient inversion , resulting in a larger mismatch effect (even with same mismatch parameter of the transistors).
Hope this helps !