Thinking about it, one could use a similar, but simpler technique that is also completely reversible:
Send the mono signal through a very short delay like Scooter and OtheroneK mention (too short to be percieved as echo - the Hass region) to a stereo return, hard-panned left/right, then invert polarity of one of the return channels. Mix the stereo return in at a very low level with the mono signal to taste. Because the delayed material has identical levels but inverse polarity in each channel, the delay signal cancels completley if summed back to mono. You can EQ the delayed stereo contribution if you like and the EQ will also cancel when summed, as long as the EQ applied is the same for both sides [edit- must be applied prior to inverting the polarity on one side] and is not applied to the mono signal.
Clean and mono preserving, but with some stereo width.