At LithoVision 2014, Dr. Mark Phillips, Intel Senior Principal Engineer, showed what was needed to enable a full 2x density shrink on second generation TriGate technology using 193 immersion (193i) lithography. The number-one challenge of using 193i for 14 nm was edge placement error (EPE). He explained that pitch division and pattern splits kept resolution within 193i capabilities at the cost of more masks, making it imperative that all contributors to edge placement error, not just overlay, had to improve commensurately. The second main challenge for 14 nm production was focus control and Phillips reported that “Pattern splits are not a panacea.” He explained that spacer based pitch division (SBPD) may require printing of more isolated patterns with reduced depth of focus (DOF), while focus variations cause CD variations that contribute to the EPE budget. He emphasized that the focus budget is now nearly as critical as the overlay budget (Figure 1A).
The third key challenge was rapid problem resolution. With those tight EPE and focus budgets and a short time for readying production, the scanner must maximize its performance on product wafers. Additionally, chip makers are faced with challenges that weren’t anticipated when the scanner was initially designed and specifications were established. Therefore, suppliers must have the technical capabilities to quickly understand the issues and the willingness to quickly address them. The 14 nm node using 193i was ultimately made possible through 1) intrinsic tool capability, 2) performance on product wafers, and 3) metrology/data analysis tools to assist process development.
 |  |
| Figure 1A. Phillips emphasized that the focus budget is now nearly as critical as the overlay budget (left image). Figure 1B. 25 wafer mix-and-match (MMO) 3σ overlay equaling 3.57 nm in X and 3.22 nm in Y was reported. | |
Looking first at intrinsic tool capability, data was shown for NSR-S621D single machine overlay (SMO) 3σ across 25 wafers of X = 1.56 nm, Y= 1.84 nm (after linear error was removed). In addition to 25 wafer mix-and-match (MMO) 3σ overlay equaling 3.57 nm in X and 3.22 nm in Y (Figure 1B). Phillips subsequently described the new autofocus sensor system incorporated in the NSR-S622D generation scanner. The narrower sensor pitch and improved edge mapping provided enhanced fundamental tool capabilities with focus uniformity 3σ of all shots reduced from 15.5 nm to 9.8 nm as measured by phase shift focus monitor (PSFM). AF sensor fluctuation was also markedly reduced from 9.5 nm (worst sensor on the grid plate) to 6.6 nm, and focus stability was improved as well (Figure 2A). Beyond these results on flat wafers, optimal performance on product is vital. Autofocus data taken after the modifications demonstrated AF sensor dependency on the process was reduced to a quarter of original levels (Figure 2B). The S622D design also successfully decreased the magnitude of focus corrections by minimizing occurrences when AF sensors were “misled” by the process.
 |  |
| Figure 2A. AF sensor fluctuation was markedly reduced and focus stability was improved as well (left image). Figure 2B. Optimal performance on product is vital, and autofocus data taken after the modifications demonstrated AF sensor dependency on the process was reduced to a quarter of original levels. | |
Phillips then turned to the topic of on-product focus control and bringing focus metrology into an advanced process control (APC) loop. While overlay control incorporates product and layer-specific APC, historically there have been few options for on-product focus control. On-product focus metrology requires fast setup, compatibility with full product stacks and design rules, precision of about 5 nm (today), and preferably the ability to utilize standard targets. Phillips described an effective exposure tool management process using the Nikon NFDM-3500 system, which employs device patterns instead of specialized targets. The NFDM captures the diffraction image of an entire developed wafer in a single frame. The diffraction images are then used to calculate dose and focus maps within 10 minutes, and that information is sent back to the scanner. On-product focus performance was reported at 42 nm 3σ for all shots using a 22 nm scanner, with 14 nm node scanner results reduced to 25 nm 3σ. Additionally, 14 nm node scanner on-product focus control surveyed by the NFDM showed all shots 3σ of 26 and 27 nm respectively for two different S622D scanners. Phillips announced that the S622D autofocus system combined with full-wafer scanner focus mapping provides excellent edge field performance and the lowest process dependency (Figure 3A). Additionally, data analysis tools recently made available to the customer, such as the AF Viewer, allow on-call engineers to quickly debug problems with incoming wafers.
Phillips summarized his invaluable presentation by reminding the audience that for 14 nm the intrinsic tool capabilities must meet very challenging edge placement error and focus control requirements on product wafers, and that use of external metrology and data analysis tools are a means to that end.
 |  |
| Figure 3A. Phillips announced that the S622D autofocus system combined with full-wafer scanner focus mapping provides excellent edge field performance and the lowest process dependency (left image). Figure 3B. The newly launched NSR-S630D delivers world-class mix-and-match overlay and focus control with “Sustained Stability” to enable the 10/7 nm node. | |
In complementary presentations at LithoVision and SPIE Advanced Lithography, Yuichi Shibazaki, Nikon Technical System Director, and Hiroyuki Egashira, Nikon Development Engineer, provided updates on next-generation immersion scanner innovations. The Nikon specialists announced that even tighter budgets and enhanced stability are required beyond 14 nm, and revealed how the newly launched NSR-S630D delivers world-class mix-and-match overlay and focus control with “Sustained Stability” to enable the 10/7 nm node (Figure 3B). It was explained that the NSR-S630D incorporates newly designed optics that deliver multiple levels of active control, while the Multipoint High Speed PMI system enables adjustment of the lens at intervals to reduce wavefront and distortion aberrations. Enhanced tuning capabilities enable the S630D projection lens to have wavefront rms levels ~20% lower than the previous generation system, and single nanometer distortion values have been achieved.
Beyond imaging, as Phillips and others had already mentioned, overlay and focus control are the critical performance factors for sub-14 nm. Egashira reported that the newly designed S630D reticle stage uses an encoder servo system to increase accuracy. In addition, the wafer stage delivers improved temperature control, coupled with structural and water management enhancements to improve stability, while very low distortion is a central factor in optimizing mix-and-match overlay. He revealed that the S630D has already demonstrated (Avg.+3σ) single-machine overlay below 1.4 nm across the lot, and across lot S622D/S630D mix-and-match overlay below 2.5 nm has also been achieved (Figure 4A). Further improvements to autofocus performance were also verified with uniformity 3σ data below 9 nm including edge shots and 5.9 nm for full field shots alone. Intrinsic CD uniformity results below 0.69 nm were also demonstrated for 41 nm lines on a 90 nm pitch.
 |  |
| Figure 4A. Across lot S622D/S630D mix-and-match overlay below 2.5 nm (Avg. +3σ) has been achieved (left image). Figure 4B. Further improvements to autofocus performance were verified with S630D 3σ data below 9 nm including edge shots. | |
Shibazaki highlighted that at the most advanced nodes, tool stability and process robustness become increasingly critical. Additional calibrations will help, but may affect productivity. Therefore, long-term inherent tool stability and process robustness must be maintained. The S630D has demonstrated five lot SMO data below 1.7 nm (Avg. + 3σ) across a ten-day period (Figure 5A), and SMO performance (Avg. + 3σ) below 1.4 nm across the lot has been achieved with both hydrophobic and hydrophilic processes. Additionally, a two week focus stability range of only 5.3 nm max/min was attained (Figure 5B).
 |  |
| Figure 5A. The S630D has demonstrated five lot SMO data below 1.7 nm (Avg. + 3σ) across a ten day period (left image). Figure 5B. A two week focus stability range of only 5.3 nm max/min was attained on the NSR-S630D. | |
Shibazaki ended his presentation with a discussion on 450 mm technology. He commented that while some have questioned the litho benefits from the wafer size transition, Nikon believes there are many possible prospects for productivity and performance innovation at 450 mm, and is taking full advantage of this opportunity to build the next-generation tool platform. Applicable advances in 450 tooling development will also benefit 300 mm scanner design as Nikon immersion extension will enable production beyond the 7 nm node, satisfying process node requirements in 300/450 mm. In closing, Shibazaki announced that Nikon 450 mm lithography tooling will be ready when the industry decides to make the transition.