Comparator offset problem

I was designing rail to rail latched comparator and when I measure the offset this histogram appear to me. I wanna know what may be the reason for this far occurrence between -20 and -30 and Is this due ti rail to rail and how can I solve this problem?

Thanks in advance
Regards,
Amr

You can solve the problem by

1) identifying the specific MC iteration(s) responsible

2) assert one specific iteration and rerun the analysis
(same MC mode / settings, set the index, run, check
that the same post-analysis retuens the same offending
result)

3) Inspect operation for what's whacked

This might be a gain shortfall at one end of the Leff
range (and you should then inspect the imposed Leff
position, relative to WAT spec), a failure to start up
in the master bias reference or a way-off bias level,
some gross mismatch in N vs P VT / IDsat (skewing
inverter thresholds and pushing back into the front
end), etc. You could speculate all day, so get on with
the task of making a single MC iteration repeatable
and accessible in a foreground simulation environment.