Question on Gilbert cell mixer, HELP!!!!!

For a 0.18um MOS Gilbert cell mixer, simulations show me that the drain current of the switching pair is not a sqare wave, but something like a half sinusoidal curve, the peak is not flat, no matter how i increase the LO power. Therefore it consists of not only the odd order harmonics, but also even order harmonics. i want to know how to use functions to represent the switching pair drain current, what is the main reason to cause the current more like a half sinusoidal curve, is it due to the short channel effect? anything else?

This document may help you , I hope..

THere is a agilent seminar

hxxp://eesof.tm.agilent.com/adsdemo/seminar4/extras.html#workbook

it may help