Contact Hole Area Dwarfs Corner Rounding in Reticle Considerations
Aaron Hand, Managing Editor -- Semiconductor International, 2/1/2001
 |
The fact is that masks are not perfect. Like the designs copied onto the wafers, reticles are often printed with rounded corners. But the question is: How much should a chipmaker be concerned with mask imperfections? Is it worth spending the dollars necessary to achieve less corner rounding on the masks?
Chris A. Mack, director of research for the FINLE division of KLA-Tencor (Austin, Texas) and vice president of technology for KLA-Tencor's Lithography Module Solution Group, studied the effects of wafer pattern infidelity on device performance. Presenting his findings at Interface 2000 in November, he showed that — at least in some cases — corner rounding has little effect on pattern fidelity. For small contact holes, for example, the area of the mask pattern in relation to the final pattern plays a much bigger role.
Looking at conventional and attenuated phase shifted contact holes in a variety of pitches, Mack used theory and simulation to look at the key parameters that contribute to corner rounding on a wafer.
Theory shows that some degree of corner rounding on the wafer is inevitable because higher spatial frequencies are lost when a pattern passes through a lens. The degree to which that corner rounds depends on various factors. In Mack's example, an isolated corner imaged with coherent light would have a corner radius of about 0.36 l/NA. But the amounts of rounding will be affected by how close that corner is to other features. For two corners next to each other (to print the end of a line), if the linewidth is more than twice the corner rounding radius, each corner only slightly influences the other. But if the linewidth is smaller than about 0.7 l/NA, the corners' rounding will overlap, resulting in line end shortening.
Contact holes or small islands face even more rounding, Mack explained, because they require the interaction of four corners. It's inevitable that small contact holes — despite being square on the reticle — will be round when printed. But because it is such a common occurrence, it is not considered a problem, he said. In this case, the area of the hole is the more important consideration.
This panned out in aerial image simulations, performed with PROLITH/3D using the High NA Scalar model and a 248 nm wavelength, a 0.7 NA, no aberrations or flare, a s of 0.4 or 0.7, and 4× reduction. All mask contact hole patterns had an area of 150 × 150 nm, but the corner rounding of the mask varied from square to round, and the pitch varied among dense, semi-dense and isolated. The resulting aerial image in all these cases varied only slightly. Mack also found little effect of corner rounding when running resist simulations on contact holes.
Mack performed similar simulations to look at line end shortening. As mentioned earlier, a line whose width is near the resolution limit will generally exhibit line end shortening because of the interaction of the two corners that the line is made up of. This makes line end shortening a good measure of the effects of corner rounding. In an example, Mack noted that a 180 nm isolated line imaged at 248 nm with a 0.688 NA lens and with a partial coherence of 0.5 will produce an aerial image with almost 50 nm of line end shortening.
Because the corning rounding is a strong function of k1, the line end shortening is significantly affected by the NA. A higher NA will produce less line end shortening. In the example above, changing the NA to 0.5 extends the line end shortening to about 130 nm. Conversely, increasing the NA to 0.8 would decrease the shortening to about 40 nm. The effects of partial coherence, on the other hand, are less predictable. But both image- and process-related factors can cause significant line end shortening.
Overall, corner rounding on a reticle has relatively little impact on final resist CD, Mack said, but it should still not be considered negligible. Corner rounding uniformity should certainly be considered in the mask making process. However, the effect of the area of the contact hole on the mask should be a much bigger consideration.
For additional information on lithography, go to www.semiconductor.net/lithography
