Extending Lithography Capability with Specialty Materials, Part 1 of 2

April 05,2016

In this, the 51st anniversary year of Moore’s Law, it is now evident that further dimension scaling of devices can no longer be accomplished solely by improving the patterning resolution capability of lithography technology. This two-part series offers an overview of the challenges presented by further scaling and how extensions to lithography capability allow for patterning of smaller features.

In this, the 51st anniversary year of Moore’s Law, it is now evident that further dimension scaling of devices can no longer be accomplished solely by improving the patterning resolution capability of lithography technology. This two-part series offers an overview of the challenges presented by further scaling and how extensions to lithography capability allow for patterning of smaller features.

Critical dimensions (CD) of integrated circuit (IC) devices continue to shrink, while the raw patterning capability is constrained by the resolution limitations of the lithographic exposure tools using the 193 nm wavelength. Below the 22 nm node, typical device CDs require complex multi-patterning process integration schemes using sequential deposition steps combined with specialty sacrificial thin-films to create pitch-splits. Such complexity tends to increase the cost of high-volume manufacturing, especially when patterning across complex 3D topographies such as those created by finFETs.

193 nm immersion (193i) lithography remains the proven technology used throughout the industry, since this is the smallest optical wavelength transmittable through quartz lenses. Post-optical lithography technology such as extreme ultraviolet (EUV) is still working its way through process development, and more time is required to make it fully cost-effective in high-volume manufacturing. Meanwhile the industry is seeking cost-effective ways to pattern ever-smaller structures by extending tried and true 193i processes.

Thin-film vacuum processes such as plasma-etch and atomic layer deposition (ALD) have been used to shrink photoresist features in R&D settings, but can add substantial cost and complexity to the overall process flow. To reduce direct cost as well as fab turnaround time it is desirable to have spin-on specialty shrink chemicals that can be processed in industry standard lithography track tools. Chemicals have been developed to shrink lithographic features at the same pitch using 193i extension materials including the following:

  • 5th generation spin-on contact hole shrink materials, and
  • 3rd generation spin-on line-trimming materials.

Figure 1: 193i Extension Materials

Figure 1 illustrates why the 5th generation of hole shrink chemistry is not only cost-effective, but provides superior final results compared to an ALD approach. ALD provides perfectly conformal sidewall coatings inside contact holes, but resolution-limited contacts tend to print in photoresist with non-circular shapes such that smoothing is desired instead of perfect conformality. The shrink material has been formulated such that surface-chemistry effects round off the hole to correct for the resolution limitations of 193i. Tests on dense contact hole arrays with 3 sigma variation of 5.1 nm after initial development showed slight reduction to 4.4 nm using ALD, or a near 40% reduction to 3.1 nm using the 5th generation shrink chemistry.

Contact holes starting with CD of 55 nm can be controllably shrunk by 25 nm to 30 nm using Dow's shrink materials. Contacts printed on the same grid pitch but with CD of 42 nm can be likewise shrunk by 25 nm to just 17 nm in diameter. While the chemistry has been tuned to smooth sidewalls, it is still somewhat conformal such that trench or oval contracts retain their unique shapes and are not simply rounded. The shrink amount is determined by the formulation, allowing for consistent shrink across multiple feature sizes and pitches, which is ideal for OPC modeling.

While shrinking a hole requires the addition of material, shrinking a post or a line requires the removal of material. Plasma etching removes material with excellent control, but separate, expensive tools are required. Chemical etching in some controllable manner provides the needed material removal with a similar level of LWR (Line Width Roughness) improvement, and can be done in standard litho track systems, provided that chemistry families are compatible. With decades of experience in R&D for lithography materials, Dow’s Electronic Materials team is able to ensure chemical compatibility from the beginning to the end.

Dow’s spin-on chemical trim overcoat material extends 193i lithography by improving not just CD uniformity (CDU) but also critical parameters such as depth of focus (DOF), line width roughness (LWR), and defects. All the chemical trim overcoat materials, photoresists, and other coating formulations studied here were prepared by Dow Electronic Materials using proprietary materials.

In one experiment starting with 45 nm 1:1 Line/Space (L/S) pattern of a commercial photoresist, a 600Å film was formed by spin-coating the chemical trim overcoat material at 1500rpm. The wafer was then developed with 2.38% TMAH developer. The resulting line CDs and LWRs were measured on Hitachi CG4000, while cross-section SEM (XSEM) images were collected on a Hitachi S4800, as shown in Figure 2.

  Without Overcoat With Overcoat
Average Line CD 46.3 nm 30.4 nm
CDU Sigma 0.42 nm 0.24 nm
LWR 3.5 nm 2.6 nm

Figure 2: Lithographic Performance Comparison With and Without Chemical Trim Overcoat

The extent of CD trim depends on the diffusion of acid from the chemical trim overcoat into the photoresist. The speed of acid diffusion depends on the time and temperature that the overcoat is exposed to the photoresist during what would be termed a “soft-bake.” The size of the overcoat acid molecules also determines how fast and thus how far into the resist they diffuse for any given time and temperature. Control of acid diffusion is, of course, critical to the functioning of the photo-acid generator (PAG) in modern photoresist, so we can use our expertise in molecular engineering to target specific acid sizes. Chemical trim overcoat bake temperature provides a very controllable and convenient route towards tuning in the exact level of shrink desired, or to optimize compatibility with the underlying resist composition.

Update: Part two of this interview was published on April 25, 2016, in which we discuss the unique challenges associated with removing photoresist from advanced finFET structures post-ion-implantation. Be sure to see part two for the conclusion to this discussion.

Editor’s Note
This article was originally published in Semi Manufacturing China. To read the original version of the article in Simplified Chinese, view the article on the SEMI China website (please note, registration is required for access).