Jump to content

Specific Process Knowledge/Lithography/Aligners/Aligner: Maskless 03 processing: Difference between revisions

Taran (talk | contribs)
Taran (talk | contribs)
Line 156: Line 156:
[[Image:MLA3_5214E_3umLS_bias.JPG|400x400px|thumb|The effect of applying CD Bias during conversion on Aligner: Maskless 03]]
[[Image:MLA3_5214E_3umLS_bias.JPG|400x400px|thumb|The effect of applying CD Bias during conversion on Aligner: Maskless 03]]


The size after development of the structures printed on Aligner: Maskless 03 is not necessarily 1:1 with the design. Tests have shown that 3µm lines of resist printed in positive resist using quality mode on Aligner: Maskless 03 resist are 0.4-0.8µm too narrow, i.e. the exposed and subsequently developed area is wider than in the design. This bias is a function of the dose, and gets bigger with higher dose. It is also different for different resists. At the dose that appears to give the best resolution for AZ 5214E, the bias is 0.7µm, while it is approximately 0.45µm for MiR 701.  
The size after development of the structures printed on Aligner: Maskless 03 is not necessarily 1:1 with the design. Tests have shown that 3µm lines of resist printed in positive resist using quality mode on Aligner: Maskless 03 resist are 0.4-0.8µm too narrow, i.e. the exposed and subsequently developed area is wider than in the design. The bias is a function of the dose, and gets bigger with higher dose. It is also different for different resists. At the dose that appears to give the best resolution for AZ 5214E, the bias is 0.7µm, while it is approximately 0.45µm for MiR 701.  


The conversion manager software allows for correction of CD bias during the design conversion process.  
The conversion manager software allows for correction of CD bias during the design conversion process. This correction is separate for the X- and Y-direction, but only in 200nm steps, and only negative values, i.e. decreasing the exposed area. In the X-direction, the step size changes with the exposure mode, so CD bias correction probably only makes sense for quality mode exposures. From tests, see also figure to the right, it seems that applying a -200nm CD bias makes a printed resist line 0.2µm wider, so the CD bias correction apparently works by decreasing all structures by one sub-pixel (which for MLA3 in quality mode corresponds to 100nm, as it prints 5-over using 500nm pixels).  


<br clear="all" />
<br clear="all" />