Dear Vishal,

to better understand your problem please let me ask a few questions first:

The PAB procedure for PMMA is usually not very critical as the baking step is only intended to remove the solvent from the reist layer and has no significant influence on the clearin dose, the contrast or other key FoM for the process. Furthermore, no PEB is needed for PMMA. So basically something like "spin coat first layer (in your case the 495k PMMA), PAB (5-10min @180°C or something similar), spin coat second layer (the 950k PMMA), PAB with the same parameters as before" should work just fine. The main issue you might have to face is the charging of the quartz sample which might make the definition of a high resolution 20nm CD feature with a resist profile suitable for lifting of 140nm of Si a challenging task.

Best regards,

 Jens

I am planning to use 15 to 20 Kev. The charging problem is handle by a conducting polymer which I applied after PMMA coating. Developing solution is MIBK:IPA (1:3). 

Now, since I am depositing amorphous Si (X shape 140 nm height with feature size of 20 nm) on top of Quartz sample. After patterning, I will go for lift-off. So here I want to know few things.

1. What should be the baking timing after first layer of PMMA 495 and 950. Some people heat the substrate for 10 minute, some of them suggesting the 30 minute at 180 degree Celsius. How much time this Photo-resist take to stick with the quartz. 

2. What thickness I should choose for lift-off. I am spinning PMMA at 5000 rpm for both 495 and 950K.  I am planning to lift-off with gold (80nmm). Is this sufficient ? 

As mentioned before, the baking is not very critical, ~10min @ 180°C for both layers should be just fine. What I still don't get is your overall process flow. Did I understand you correctly: You plan to deposit a layer of 140 nm of Si, then do the eBeam litho using the bi-layer resist stack, deposit ~80nm gold, do a lift off and then use the gold as an etch mask for the Si?

The thickness ofthe PMMA will depend on the exact type of PMMA you use, both 495k and 90k PMMA come in very different concentrations (A2,A3,...) which yield very different layer thickneeses for identical spinning parameters. Usually the thickness of a lift-off resist layer should at least be three times as thick as the layer which has to be lifted, so in your case a reist mask of ~300nm should be sufficient.

Yes, you got the process flow correctly. The PMMA that I am using is 4% concentration for 495K and 950 K. Thank you for your valuable reply. 

Hi Vishal, I have one question. When you do the bi layer resist, do you expect any mixing of PMMA 495 and 950? Lets say you intend to put 200nm 495 and 200 nm of 950, total would be 400 nm. But if there is mixing of two layer resist happening, then this thickness will be less and non uniform over the surface. 

Well, you question was specifically directed to Vishal, so I hope giving you my opinion is ok as well. ;-)

From my experience using a bi layer PMMA stack results in a quite uniform resist thickness. If you cleave an exposed and developed structure, e.g. to investige the undercut of a profile for a stack like the one Vishal described, you can see a relatively sharp transition between both resists, even for quite thin (~50nm) layers of the top resist.

The same is true if you use such a bi layer stack "the other way round", e.g. to create a T-gate structure. If there is any mixing of the two resist variants it is limited to just a few nm  and non-observable in a standard SEM inpsection after development. Resist unifmority is more or less as good as it is for each of the two individual coating processes, the combination of the two, at least for our processes, doesn't create any additional non uniformity.

@S M Naeemul Islam, Hello, we are using prebaking and postbaking for two minute at 180 degree Celsius. So I don not expect any mixing of two layer. Actually, it was too old discussion, the problem was associated with non-uniform deposition of a-Si which I realized after a lot of experiment. So now it's almost solved. 

Thanks for advice @Jens Bolten.