Photomask renaissance predicted at bacus 96

11/01/1996

Photomask renaissance predicted at BACUS `96

In his keynote address for the 16th annual BACUS Symposium on Photomask Technology and Management, held in September in Redwood City, CA, KLA Instruments chairman and CEO Ken Levy predicted a new era for the maskmaking industry, one that would reclaim the grandeur of the past.

Before the development of the wafer stepper, mask fabrication was the high-tech segment of the semiconductor industry, with wafer production relying on crude contact printers, said Levy. The reduction stepper brought on a dark age, with photomask industry revenues declining to 1% of the semiconductor industry, consolidation, and only Asia supporting the development of new tools. Now, with DUV photomasks, optical proximity correction (OPC) and phase-shifting masks (PSMs) required for semiconductor industry progress, the stage is set for a rebirth of mask making. However, maskmakers and their equipment suppliers may now be too debilitated to renew themselves without help.

The symposium highlighted the challenges of achieving the CD control and defect levels required by the SIA roadmap. Larry Weins of SEMATECH reported on a benchmark study of six mask writer lithography systems. While half beat the 70-nm registration roadmap requirement for 350-nm ground-rules, none of the fabricated masks met the 50-nm CD uniformity spec. While conference participants suspected that the CD variation reflected processing variability rather than intrinsic mask-writer performance, no one could deny the difficulty of achieving the 35 nm or better CD control needed for low k1 wafer lithography.

Sheldon Kugelmass of Lepton Inc. reported <40 nm (mean+3s) CD variation, but highlighted the importance of including the sample size in any spec of the range of CD variation. Christofer Braun of Lucent Technologies reported achieving similar results for a full year`s production of a MEBES-4-TFE, monitored and optimized by on-product metrology. Rik Jonckheere of IMEC, Leuven, Belgium, reported 20-nm CD control with a Leica EBMF 10.5 vector-scan exposure tool, dry etching of MoSi films and CAPROX e-beam proximity correction. The capability to meet CD control specs thus appears to exist, if such innovations as dry etching (with etch resistant resists), on-product monitoring, proximity correction, and process optimization are adopted.

The technology for avoiding and correcting defects may not be so certain. Dry etching (which improves CD control) appears to introduce more pinholes than conventional wet etching, and no consensus appeared as to defect printability. Franklin Kalk of DuPont reported that defect size was not crucial for estimating printability. A tiny pinhole that would not print when isolated produced horrific effects when near a circuit element. A study of clear and partly-opaque programmed defects using the KLA Starlight inspection system revealed that it could find 100% of clear pinholes larger than 150 nm and 100% of 32% transmitting defects larger than 400 nm. Thus, while defect size did not matter, energy did. Kalk proposed a new defect specification based on printability, to be calibrated in "Wileys," in honor of mask industry figure Jim Wiley, with every printable defect being >1 Wiley. Unfortunately, no algorithm for predicting printability yet exists.

Yoon Hoon Kim of Samsung presented a paper showing just how difficult it might be to achieve mask CD control sufficient to meet a 10% wafer CD specification in the 256-Mbit DRAM era. Using a 4? programmed-defect contact mask with a nominal CD of 310 nm (at 1?), the Samsung team showed that any anomaly that caused a 3% difference in transmitted energy would cause the wafer feature to miss spec. A 40-nm shift of one edge, or a 230-nm chrome or clear extension was sufficient, neither of which can be detected with current inspection equipment.

To minimize contamination due to handling and storage, Michael Kling of Motorola advocated improved and standardized shipping/storage containers for reticles. Present systems produce abrasions, deposition, and particulates due to vibration, and require manual (rather than robotic) transfer. With standardization, contamination and damage can be reduced. Other innovations reported included gas-assisted focused ion beam repair for opaque defects, dry etching with an inductively coupled plasma source for improved uniformity and inspection tools based on differential interference contrast.

The 16th BACUS Symposium provided a suitable conclusion to a year devoted to photomask defects. However, to achieve the renaissance predicted by Ken Levy, the mask-making community will have to learn to do new things, not just reduce defects in current technology and overcome the memory of the recent "dark-ages."

M.D.L.