To begin  you  should  avoid  copper  hardmask so as to get smooth structures. You deal  with  60  to  100  microns  deep or nm as you wrote?

The etch depths are in nanometers. 

Yes, the copper mask does appear troublesome with some of the etching formulas I have tried. I used copper because I initially thought it would not react with the CF4 etching gas. What metals do you suggest?

Thanks!

Aluminum  can be a good  alternative,  nevertheless polysilicon remains the Best as regards to roughness reduction in Si02 trench bottom. And using CHF3 / CF4 chemistry with a fine Tuning of the gas ratio will allow you to get an interesting selectivity as regards to these hardmask materials reducing material sputtering & redeposition ( avoiding / reducing thus Micromasking effet at the origin of roughness apparition ). 

Regards Mohamed

Ps : The problem of roughness Can be due to the fact that Copper is sputtered and redeposited on oxide surface (because of no volatil byproducts) creating a micromasking effect .. . BUT it  Can  also  be  due  to  copper  residues  remaining  on   SiO2  surface  after  mask  opening...By  the  way ,  How  do  You  etch  this  Copper  layer ? 

Mohamed,

Thank you for your helpful response. Your explanation of the material sputtering and redeposition makes a lot of sense with what I am seeing. I have attached an optical microscopy image. My preliminary fabrication scheme is that I make a thin film pattern with copper on qaurtz via shadow masking and thermal evaporation. This leaves some areas covered in copper and others exposed SiO2. Etching of these areas is the step I am currently working on...

From the image and your knowledge on redeposition, does it seem  possible that the Cu is "filling in" the smaller, web-like apertures in the pattern?

I think my next step will be to attempt the SiO2 etching with an Al hardmask instead of Cu.

Thanks again for your time,

Shaina

Dear Shaina

About your question : «, does it seem possible that the Cu is " filling in " the smaller, web-like apertures in the pattern? » my answer is YES. YES ( First point ) because independently fromCopper when the CD ( critical dimension ) decreases, etching ismore difficult and thus redeposition becomes predominant. The etching is difficult as CD decreases & Hardmask Aspect ratio is important. ( What is the Copper thickness? Smallest dimension ( trench ) seems to be less or equal to 1micron! !?.... The second point is that Copper has no volatile by-product and thus etching needto be physical ( Argon assistance or CHF3plasma with high Bias ). The third point (Themost iimportant ) is a question : are You sure the smallest Trench are initially opened after Shadow Mask deposition? If not sure a FIB cross-section can be helpfull.

Regards/ Mohamed

Mohamed,

Thank you again for your reply. I am new to this sort of fabrication style and appreciate the help. 

Yes, the Cu mask (thin film) is very thin; it is 20- 30 nm. Also, you are correct that the smallest dimensions are less than 1 um; some of the smallest trenches are 200 - 500 nm wide.

In regard to your third point, yes, the smaller trenches are initially open after shadow mask deposition. I have inspected the samples with an SEM before etching procedures and confirmed this.

So from what I gather from your responses, it appears that minimizing redeposition is the way to deal with properly etching the small CD features. As a review, to etch small CDs and help eliminate redeposition I should:

1a.) try using a mask that achieves volatile byproducts (neither Cu or Al do this with flourine gases alone); OR 1b.) use an etching formula (like incorporating Argon) that does produce volatile byproducts.

Are there any other techniques to minimize the predominance of redeposition over etching? Overall, would you still recommend Al over Cu? I will start to look more into the redeposition vs etching  issue in the literature...

Thanks again,

Shaina

Shaina

According to what you have written about Copper CD and Thickness (and that small apertures are well-opened) -,I will advice you to simply keep the Cu layer & try to increase the physical aspect of Etching by adding Ar and increasing bias voltage. This is the easiest way to try to improve the etching without changing the whole process& Flow . Other point,concerning your special Target(special design& dimension) I am not sure that Al Can be a solution if my advice about Copper fells ...in this last case a poly-si mask can be THE solution.

Regards / Mohamed.

As I was reading again you first description I noted that you have an RIE plasma reactor and also a plassys one (PECVD) . If it is not already the case, You should try this last one (which is an ICP reactor ) because of the decoupling system between Ion Energy(Bias voltage ) & Ion density ( source power ). With the ICP Tool the physical aspect of Etching should be better controlled. 

Thank you for the suggestion. I will try the etching process with the PECVD!

~ Shaina