At LithoVision earlier this year, Dr. Mark Phillips, Intel Fellow and Director of Lithography Hardware and Solutions, provided an in-depth “Chipmaker View of Lithography for the Next Node.” He reported that there has been two years of solid progress on EUVL, and believes that its introduction in production is a question of “when” rather than “if”. He reminded the audience of the Intel position that EUVL is highly desirable for the 7 nm node, but it will only be used when it is ready. EUV insertion will also be different in that it won’t simply replace existing technology for critical layers. Availability, stability, and operating costs were identified as on-going concerns, and Phillips reported that it is time to also consider the implications of EUVL insertion on other lithography tooling and infrastructure so those do not gate high volume manufacturing.
Even at full productivity EUV will remain expensive per wafer pass—with tool cost double and productivity half that of 193i lithography. Phillips cautioned that EUVL must be used carefully, and that it will be necessary to replace three or more 193i masks plus other process steps used for multiple patterning in order for EUVL to be cost effective (Figure 1A). Reduction of mask count can’t be done indiscriminately; e.g. one can’t combine a 193i via mask and a 193i metal mask into one pattern. Phillips stressed that 193i will continue to be employed wherever possible to keep wafer cost in control. The number of 193i layers in process at the introduction of EUV will remain high, and the total mask layers will essentially be flat compared to the previous node. Multiple patterning techniques will continue to be utilized wherever possible. Methods developed to extend 193i are well understood, and employed in design rules and layouts to maximize device performance and density. Complementary EUV/193i patterning will be used at introduction, and 193i + multiple patterning will continue to be utilized whenever at least three 193i masks cannot economically be replaced by one EUV mask.
Figure 1A. Phillips reported that even at full productivity, EUV will remain expensive. 193i will continue to be used wherever possible (left image). 1B. Total edge placement error is the biggest technical obstacle to scaling, and it is necessary to reduce EPE contributions across all process steps.
Resolution is not the only challenge at the next nodes. Continued device scaling requires innovation in transistor architecture and interconnects, which challenges technology development outside of litho and etch. Phillips believes that total edge placement error (EPE) is the biggest technical obstacle to scaling, and that it is necessary to reduce EPE contributions across all process steps in order to take full advantage of the resolution benefits possible with EUVL (Figure 1B). He noted that the additional wafer cost makes it hard to absorb cost increases from lithography. Aggressive scaling is needed to control cost, which drives demand for continued improvements in EPE contributions from 193i, OPC, etch, metrology, etc.
Phillips highlighted that total device EPE is vital and that its budget should drive development priorities and choices regarding tools, metrology, applications, and patterning options. Enhancing 193i scanner overlay and focus is an essential part of reducing the budget, and he reported that the latest generation NSR-S631E immersion scanner single machine overlay (SMO) is already outperforming that of the previous generation toolset (Figure 2A). Phase Shift Focus Monitor (PFSM) performance is highly correlated to focus performance (and yield) on product wafers, and Phillips showed all-shot 3σ results reduced from 11.8 nm for the S622D to 5.6 nm for the S631E. In addition, Reticle Bending allows field curvature compensation without inducing astigmatism, with negligible stress-induced birefringence. He reported that the Nikon Reticle Bending function allows correction for a wide range of blank shapes with maximum focus impact at the wafer reduced from ~11 nm to ~2 nm (Figure 2B).
The importance of enhancing metrology using site-by-site metrics as compared to RMSE was discussed as well. This will aid in achieving the necessary overlay accuracy and stability, and also flag systematic issues. Phillips explained that the industry must continue to focus metrology work inwards – on tool, site/wafer-level analysis and use site-by-site metrics to test against process sensitivities. In closing, he emphasized the continued pervasiveness of 193i and multiple patterning, and the criticality of 193i scanner enhancements that will benefit edge placement error and productivity. Given the limits that EPE places on scaling, effective budgets and modelling are imperative. He concluded that the industry must work to drive down edge placement errors and their contributors (Figure 3A).
In a complementary LithoVision presentation, Ryoichi Kawaguchi, Nikon Corporation Technology Solutions Sector Manager discussed the importance of on-product performance (OPP) in enabling next-generation scaling. He showed the evolution of the Nikon Streamlign platform and continuous scanner innovations that enhance manufacturing performance and productivity (Figure 3B). He cautioned that as devices continue to scale, defectivity becomes increasingly significant and even the smallest immersion defects can be problematic. Kawaguchi-san reported that Nikon nozzle technology progressions are enabling more stable water control to prevent immersion defects at the increased scan speeds necessary to maximize productivity.
He also showed a complex representation of the countless contributors to on-product overlay (OPO). OPO has contributions from single machine overlay and mix-and-match overlay (MMO) components, as well as process-related difficulties and tool stability detractors (Figure 4A). Nikon is addressing many manufacturing-related performance challenges through alignment and autofocus system improvements, enhanced wafer holder and temperature control, as well as new aberration compensation capabilities (Figure 4B), and advanced software Turnkey solutions.
During his presentation, Kawaguchi-san provided a thoughtful summary of the various 7 nm node requirements (Figure 5A). He reported that the NSR-S631E (announced in February 2016) is meeting all of its overlay, focus, and throughput specifications. The S631E also provides a number of innovative functions and capabilities that satisfy the on-product performance requirements necessary to meet next-generation scaling challenges. In closing, Kawaguchi-san highlighted that Nikon Turnkey solutions further enhance process windows through collaboration with Metrology and EDA suppliers.
At SPIE Advanced Lithography later that week, Nikon Corporation Performance Development Section Manager Yasushi Yoda, detailed a number of the S631E advancements that enable 7 nm manufacturing performance. The S631E uses an innovative alignment module with a new light source, improved image sensor, and optimized heat control. It delivers enhanced detection, signal intensity, and measurement repeatability for on-product alignment marks. Yoda-san showed across lot SMO results Mean + 3σ ≤ 1.25 nm, as well as MMO results below 2 nm (Figure 5B). This data was collected under 250 wafers per hour throughput conditions and included edge shots. Focus uniformity is also critical to on-product performance and Yoda-san reported that the S631E autofocus system incorporates a better autofocus (AF) light source as well as improved glass coatings. These advances reduce variability for product materials, and overall on-product focus errors for the S631E were 30% lower than the previous generation S630D. S631E 3σ focus uniformity was below 6 nm for all shots across the wafer using PSFM.
The S631E also delivers lower residual thermal aberrations compared to the S630D due to high-order thermal aberration correction capabilities with the MDDM/TAAAF systems. This helps optimize overlay even under extreme production conditions. Yoda-san also described the various XY, Distortion, and AF Zeroing functions, which are automated applications that further enhance focus and overlay performance and stability during high volume manufacturing. These detailed LithoVision and SPIE presentations provided valuable insight on next-generation lithography scaling challenges and solutions.