Dear Mattia Antonini,

Have a look at the attached paper.

Varun Pratap Sharma thanks for the material, but I'm using a 100 substrate and the problem is related for a very short etch, no more than 200nm deep, with that etch rate is very difficult to control.

Also the bottom of the samples shown on this paper is very rough.

There are many ways to optimize KOH etching. Increasing the etch rate of the (111) plane isn't really one of them. Increasing concentration can decrease the rate of all the other planes witch should give you the same effect. You may also be having these issues because of your masking material, or the solution temperature, or the agitation method. I would try doing a parameter optimization process controlling each of the previously mentioned parameters. You can also try adding IPA to change the wetting characteristics of the etch solution. Start with 10% and work your way up to maybe 30% IPA to optimize your process. Good Luck!

Ron Reiserer, thanks for the answer. I have already tried different solutions.

Mask

Hard mask with Sio2

Mask with PMMA

Both withstand the KOH but the second melts if IPA is added.

KOH concentration

Higher concentration is better, a slow etch rate makes easier to have a shallow etch.

Temperature

Some studies show that with high temperature the surface is smoother, but I see that you can have the same effect with low temperature and higher concentration.

Stirring

Stirring at 150 rpm is good, going too fast is not giving a better surface at the end.

IPA

IPA makes the surfaces smoother, also the etch rate is affected.

All of these parameters affect the quality and the etch rate, but unfortunately, always the ratio between 100 and 111 does not change, or it does but it is so small that I couldn't see it, so all the time I cannot etch less than 250nm without having defects on the flank.

I will continue to investigate further.

I would like to suggest that the SiO2 mask is the problem. Since SiO2 has a significant isotropic etch rate in KOH, you will intermittently expose the (100) plane as the the SiO2 etches back. Si3N4 may be a better option. This also allows you to include the IPA if you like the results from that. Best of Luck!

Dear Mattia Antonini ,

Etching silicon with hot KOH leads always to nonflat surface. The surface will be always rough and the geometrical patterns are specific for the surface orientation. So one uses always a process called chamisal mechanical polishing CMP, where one uses KOH while mechanically polishing the wafers with very fine slurry of white ceramics. The resulting surface will be mirror like.

Best wishes

These two articles may help

A comprehensive review on convex and concave corners in silicon bulk micromachining based on anisotropic wet chemical etching

Micro and Nano Systems Letters volume 3, Article number: 6(2015)

https://mnsl-journal.springeropen.com/articles/10.1186/s40486-015-0012-4

Etching characteristics of Si{110} in 20 wt% KOH with addition of hydroxylamine for the fabrication of bulk micromachined MEMS

Micro and Nano Systems Letters 5(1), (2017), :

DOI : 10.1186/s40486-017-0057-7

Dear Mattia,

It is not a good idea to enhance the etch rate of (111) selectively. This is because KOH etching advances in a manner of step propagation, in particular, in the vicinity of (111) orientation. It is better for you to examine the following conditions in order to achieve better results.

(1) It seems that there was a large misalignment angle of the square mask aperture whose sides should be exactly aligned to direction. You should minimize the misalignment angle first.

(2) Your etching mask aperture may not be straight enough, or straight mask side has any defects in the middle of the side.

(3) Your silicon wafer may contain crystal defects like stacking fault due to improper heat treatment before etching process.

(4) Your Si wafer surface may be mechanically stressed and damaged, or be contaminated before etching.

Hopefully you find any appropriate reason for your problem, and get better result.

Best regards,

Kazuo Sato@

Dear Abdelhalim abdelnaby Zekry ,

thanks for your answer, but unfortunately I cannot use KOH on the polisher we have because I don't want to contaminate it.

Best wishes

Ibrahim M. Hamammu, thanks for the answer, the papers you shared are more focused on the convex rather than concave corners, but after reading I had some ideas. I can actually modify the structure I want or create some compensation in the mask.

Dear Kazuo Sato,

thanks for the answer, so far I can say that the alignment of the mask or its edges are not the problem.

After splitting my sample and etching them with two different time, I have got two different results.

Long etch gave me perfect 111 flanks, short etch gave me very bad flank.

So I guess that there should be no imperfections on the sample.

I think the only problem is that the etch rate of the 111 plane is slow so it take more time to form the concave corner and this lead to an increase of the roughness from the centre to the actual corner.

One idea I have now is to use SOI substrates with a very thin Si on top in order to have an etch stop on the direction so I can prolong the etch time to reach a better corner.

Another Idea is to create some corner compensation or to modify the shape of my structure.

Best regards

Dear Mattia Antonini ,

Sorry may be i miss understood your question!

Why did you use the KOH if contaminates your material?

Best wishes

Dear Mattia,

I am sure, if you can use the SOI, you will get an improved result.

I do not think you can get a better solution by a use of compensation mask pattern in this case, because actual mask profile inevitably contains dimensional error that causes steps in the square etch profile.

Best regards,@

Kazuo

Dear Abdelhalim abdelnaby Zekry ,

I'm using the KOH for CMOS process, I know it might sound weird, but after cleaning my samples with RCA I don't have any more K contamination on the samples and I can use other tools without any problems. But regarding the mechanical polishing, we don't have a tool that I can dedicate to KOH.

best wishes

Dear Kazuo Sato ,

I'm writing just to update on the situation, at the end the SOI works perfectly for my purpose. The flanks look smooth and the corners are formed. The only imperfections I see are clearly coming from impurities in the Silicon and, of course, as expected, I have a small "beak" of silicon at the bottom of the SOI, but this is not a big deal.

Best regards

Mattia

I wonder if you tried ultrasonic shakiing during KOH etching? It will increase the stability and decrease the roughness in the deep etching application.

Dear Jyh-Cheng Yu , thanks for the answer, but I'm not doing deep etch, I have to etch a very tiny amount of silicon. So far the use of SOI helps. In case of standard Silicon, I have to try the H2 anneal to see if I can make smoother the surfaces.