Hm...

Last I checked....

I = C dv/dt

Can't touch I and can't touch C?

and you want to improve dV/dt?

If you stay with an OTA you are probably stuck. I or C need to change.

If you are willing to consider some non-linear systems that force current when the differential pair are not equal in input (not steady state for a high gain feedback system) there are some ways of throwing extra current in while slewing. An op-amp which goes to a voltage mode output throws extra current out all the time to maintain the desired voltage when slewing.

Also, consider a class AB type OTA structure. More power efficient.

Jerry

Thank you , Jerry and Yawei!

If i decrease the load cap, the loop phase margin decrease.

The system load step response may produce malfunction, and lower the system efficieny.

Now i use alternative methods to reduce hysteresis voltage, need not improve the SR of balanced OTA.

Also, look at the additive phases of all stages of the amplifier.

And - Can also drop gain of stages down a bit to maintain the same output capacitor.

Will 60db of open loop gain get it done instead of 90 db ? (Or whatever numbers are relative to what you are doing.)

Jerry

Jerry is correct, but it may be best to clarify.  It is the Gm of the diff pair that you want to change.  Changing the gain by lowering the output impedance won't give you the change you need.  

To lower the gm of the diff pair you need to use more overdrive or use some resistive degeneration.  Both eat up headroom.  If this is an issue use a separate tail current source for each diff pair "side," and put the degenerating resistance between the sources.  However, if you do this, be aware the noise from tail current sources is no longer a common mode signal.

Adaptive bias amplifiers also avoid slew rate limitations.