I had this exact same problem. If you're doing the process on glass then
that's convenient because after you sandwich the slides and squish them
together you should be able to align the device with your mask and
expose it so that the SU8 in the channels gets developed out. It can
take a very long time to develop if the channels are very clogged (I've
left devices in developer for a week before they were done) but it will
eventually open up.

hope that helps,
Joe

Jiang Ziling wrote:

>  I tried to make channel by bonding two Glass wafers, one with a 50um
>exposed Su-8 layer with channel on it, another with a 50um unexposed
>Su-8 layer. According to some research, at 75 degree centigrade the
>unexposed Su-8 can reform shape and bond with the exposed layer when
>certain pressure is applied. But I tested and found that the 100um
>wide channels on the exposed su-8 surface are always clogged by the
>soft Su-8 flowing from the unexposed layer during pressing-bonding.
>However, if I don't press, there will be no bonding. Does anyone have
>experience in solving this clogging problem?