Optoelectronic Devices Cross Significant Lithography Hurdle

More often than not, this space is devoted to the issues surrounding the cutting edge of microlithography — with its shrinking feature sizes and remaining next-generation lithography (NGL) hurdles. But it's easy to forget that what may be considerably low-tech for microprocessors or memory devices can present its own challenges for other devices. This is certainly the case for the optoelectronics realm, which has until recently relied on contact printing of large devices such as waveguides.

Albeit on a different scale, dimensions of optoelectronic devices are also shrinking, and CD control is tightening, with requirements beginning to be beyond the capability of contact lithography. Accordingly, i-line steppers are fairly standard, with a move toward deep ultraviolet (DUV) scanners entering the picture. However, although feature dimensions are smaller, field sizes remain large — some devices longer than 50 mm — requiring a stitching of multiple reticle fields that was not required with contact printers.

At this year's SPIE Microlithography conference, researchers from Nikon Precision Europe GmbH (Livingston, Scotland) and Compugraphics International Ltd. (Glenrothes, Scotland) presented a poster detailing experimental results from stitching multiple fields on i-line and DUV tools.

Going forward, more manufacturers will make the move to DUV lithography for printing some of the smaller features in their optoelectronic devices. DUV tools have better resolution, tighter CD control, and are able to print steeper sidewall angles (important for etch pattern transfer in optical devices) than their i-line predecessors. Still, a cost-effective printing solution will typically involve continued use of i-line tools for the larger features, presenting a mix-and-match situation in the fab. Many fabs use 5× i-line steppers and 4× DUV scanners, further complicating field stitching.

The researchers explored the feasibility of stitching 4× to 4× fields, 5× to 5× fields, and 4× to 5× fields. They found that, not only is it possible to stitch fields together from the various tools studied, the stitching accuracy is comparable to the accuracy of stitching fields together within a single machine.

The field size of the 5× i-line tool (a Nikon NSR2205i14TFH) was 22 mm², and the field size of the 4× DUV tool (a Nikon NSR-S204B) was 25 × 33 mm, so the researchers settled on a common reticle/tool field size of 20 mm². Wafer alignment marks were incorporated into the design, with 10 measurement points along the stitch boundary enabling stitch accuracy to be measured after exposure. Compugraphics manufactured the reticles — essentially the same, except scaled for 4× and 5× reduction — with an Etec ALTA 3000 laser writing tool.

For stitching, the fields were butted together, with no overlap. To join the measurement marks at the stitch point, half of the measurement mark was cut out from the 20 mm² field, with the other half falling slightly outside the field area. In the experiments involving machine-to-itself stitching (4× to 4× and 5× to 5×), no wafer alignment was necessary; for 4× to 5× stitching, the wafer and reticle were re-aligned on the second tool.

The mean and 3s results were taken over three wafers, with 16 stitched fields measured across each wafer, and 10 measurement points along each stitch — resulting in 480 measurement points for each condition. According to measurements on a KLA5200 overlay tool, measuring 20/10 µm box-in-box structures, a very tight stitching accuracy of 20 nm or less for 3s was achieved in both X and Y for i-line 5× stitched to itself. Stitch accuracy was even better for the 4× tool stitched to itself — 3s repeatability of m15 nm. Even matching between tool types managed a tight repeatability, with a 3s of 22 nm in the X direction, and 15 nm in the Y. In addition to checking stitch results with the overlay tool, the researchers performed measurements with lithography tool self-alignment, noting a close relationship between the two measurement methods.

The researchers also explored the effects of field magnification and field rotation, and found a linear relationship between the mean stitch in X and the applied magnification, and between the mean stitch in Y and the applied rotation. These two effects would enable quick and easy adjustment of the alignment between tools, they noted.

For additional information on lithography, go to www.semiconductor.net/lithography.