Practically every linear IC does bad stuff if the ESD diodes actually conduct any current. And the datasheets delicately avoid mentioning the issue.
Example: the AD7699, an 8-channel, 16-bit ADC. The ESD diodes are rated to 130 mA, pretty impressive. The datasheet talks about how nicely they clamp the analog inputs. What they don't say is that a little negative ESD diode current on one channel trashes the accuracy of other channels. So we're faced with the perennial problem of coming up with a signal clamp that limits the voltage at the inputs to, say,
-0.2 volts, has nanoamps of leakage at 0 volts and up, and does all this over temperature. Oh, we'd like the eight clamp circuits to take no more than, say 20 times the PCB area of the ADC itself.
I did NOT get nailed on this one this time. I tested for this before the board layout was done. Fool me once, etc.
The other cute thing about this "bipolar" "differential" ADC is that one side of the diff input must be held to exactly 2.048 volts or the ADC TC goes to hell.
Grrrrrr.
John