Hello Liang, I think Gareth is right, it looks like a lot of thickness (50um) and too much short development time, especially if you take into account the sieves dimensions (around 20um). It will be difficult develop the unsposed parts... or perhaps impossible at all: You have light scattering so it is possible that at any depth some photons are reaching the SU-8 under de sieve mask, so could be polymerised and so it will very difficult develop it. I really don´t understand exactly your channel-sieve sistem, especially the arc- sieve geometry, but looks like something like a block with "holes" with arc shape. You want the SU-8 just as a mould to get the chanel with "walls" with arc shape in PDMS. You can´t get holes from one side to other in su-8, so your walls in PDMS are not so high as you want, and cells flow away freely, isn´t it?Well, why not make the channel directly in SU-8? The only thing you will need is invert the mask, leaving the sives clear and the channel dark. I hope this idea works well for you. regards Eowin --- Den ons 2008-10-29 skrev Liang Zhao: FrÃ¥n: Liang Zhao Ãmne: Re: [mems-talk] SU-8 microfabrication problem Till: "General MEMS discussion" Datum: onsdag 29 oktober 2008 11.00 thanks, The channel was transparent and patterns of seives was opaque arcs which located in the channel. That is to say, by using SU-8, the channel patterns was a positive mold and with a very small holes (the seives). Our obstacle was how to make the holes thoughly from top to bottom? The unefficient development or unefficient exposure? I am not sure if I explain it clear. Now, we still working on it. We tried different exposure time but cannot solve such a problem. the exposure dosage we tried was: 400mJ/cm2, 200mJ/cm2, 100mJ/cm2, 50mJ/cm2, but no one works. could any one help me? Thanks.