provided by: Semiconductor International 65 nm half-pitch (HP) lithography, traditionally called the 45 nm node, has been in full swing for over two years at most advanced wafer fabs. During this time, many lithography technology options have been on the table; however, in the end, most logic and memory companies decided to use a combination of ArF dry, ArF immersion, and/or some form of a dual hard mask strategy for their most challenging layers. As we sit poised to dive into 45 nm HP development strategies, we need to ask ourselves, what will change? Will, for example, extreme ultraviolet (EUV) be ready in time? Most experts in the field agree that it will not. Therefore, what additional options do we have? Two things are clear: Design rules will continue to shrink, and we won't be able to jump to a shorter wavelength of light in order to help us get there. So how do we do it?
As the extension of optical lithography becomes clearer, the lithography community has had to come up with many innovative solutions to ensure that we can meet the necessary shrinks without extending the wavelength.¹ While the emergence of immersion lithography has allowed us to imagine the possibility of extending ArF for a record fifth generation, much of this "imagination" has been forced on us because of timing and/or cost factors associated with the alternatives. Immersion lithography has allowed us to shatter the conventional numerical aperture (NA) barrier with the addition of water at the lens/wafer interface, and will, in turn, allow us to drive down HPs to <45 nm.
However, we can't get there for free. High cost of immersion scanners and advanced reticle sets are forcing us to look at ways to extend standard dry processing as long as we can. This article will first discuss techniques used to allow us to achieve the design criteria outlined by the next technology node with advanced materials such as chemical shrink and trilayer materials. We will then discuss the status of immersion materials today, followed by a review of 193 nm resist design and, finally, where double patterning fits into the picture.
Chemical shrink
One way to shrink future critical dimension (CD) requirements without going to a smaller wavelength is to do just that - shrink it. Chemical shrink technology allows lithographers to print CDs at a larger initial size and then shrink them by 10-30 nm.² Figure 1 shows an example of this, where we are printing contact holes at a 100 nm CD and then shrinking them to 80 nm. During the shrink process, we are able to maintain the edge roughness and circularity of the hole while reaching a smaller CD.
An additional benefit to chemical shrink processing is that it allows us to extend the life of exposure equipment without costly upgrades in NA to meet necessary design rules. The process of chemical shrink can also be done using a standalone track, which does not tie up processing time on a scanner. In a typical chemical shrink process flow (Fig. 2), the mechanism of chemical shrink works on the premise that the applied material cross-links with the resist structures in the presence of residual acid and heat. It is believed that the residual acid from exposure of the photoacid generator may contribute to the cross-linking reaction; however, this is just one possibility. After the development process, some acid is left on the sidewall of the existing resist structure - either on the inside of a contact hole or the sides of a resist line. This remaining acid cross-links with the chemical shrink material during the subsequent bake process, and non-cross-linked material is removed during a development process, leaving a smaller hole or trench.
Traditionally, chemical shrink has been used to improve process latitude for contact hole applications. However, as there is more pressure put on line/space layers, lithographers are applying chemical shrink techniques to trench applications as well (Fig. 3).
Finally, key chemical shrink characteristics that enable lithographers to reach necessary CDs for 45 nm HP are complex and require special attention to multiple parts of the process. There needs to be no width/length (X/Y) or pitch dependency during the shrink in order to avoid complications with optical proximity correction (OPC) on the mask. Good etch durability is important post-shrink to avoid any CD transfer issues. There must be no intermixing with the existing resist structure in order to maintain excellent defectivity post-shrink. Additionally, the ability to modulate the necessary shrink, depending on the bake temperature applied, is critical, as a controllable shrink rate is essential to avoid any additional CD variation specifically caused by the shrink process (<0.5 nm/°C).
Trilayer systems
Multilayer techniques were developed as early as 1979, and have recently been redefined as viable techniques for feature sizes approaching the 45 nm HP.3 A multilayer process can improve process margin and help extend the life of current workhorse technologies, such as 193 nm lithography. Multilayer techniques can also improve resolution, linewidth control and etch resistance through their unique use of multiple pattern transfers into highly selective films.
Multilayer lithography involves coating a thick planarizing underlayer followed by the application of a single layer containing both imaging and etch-resistant properties (one coating) or separate imaging and etch-resistant layers (two coating steps). The hard mask, etch masking or middle layer are applied either by chemical vapor deposition (CVD) or a spin-on process. The types of multilayer techniques available for 193 nm lithography can be grouped into three categories: multilayer resist (resist/BARC/SiON/SiC), trilayer imaging using CVD (resist/SiOCH/SiC) or spin-on (resist/SOG/underlayer), and bilayer lithography (Si-containing resist/underlayer).
On one hand, the implementation of multilayer resist has been slow because of cost and the complicated process involved in stack preparation. On the other hand, the bilayer lithography is more simple to implement but requires modification of the properties of the underlayer and silicon-containing resist, which often leads to a trade-off between imaging quality and etch selectivity. Trilayer imaging, which combines the properties of both bilayer and multilayer approaches, is preferred over other processes because of its flexibility in process and resist design and its extendibility to high-NA lithography.³ For example, a 0.93 NA immersion scanner can be used to image 80 nm contacts using an advanced ArF contact hole resist, a silicon-containing middle layer and a high-carbon-content underlayer (Fig. 4).
Some of the critical advantages of spin-on trilayer lithography include reduced process costs from CVD tooling, large etch selectivity with the underlying substrate, the ability to use cutting-edge resists at relatively low aspect ratios, planarization of the underlayer, and reflectivity control of the stack.
Polymers/oligomers or small molecules containing silicon are of particular interest for middle layers because of the formation of refractive oxide and its etch resistance to oxygen plasma. The most actively studied materials for multilayer lithography are organosilicon materials, such as silicon-containing polymethacrylates, silicon-containing polynorbornene-maleic anhydride alternating copolymers, and polysilsesquioxane-based resists.4-7 In summary, there have been several advancements in spin-on trilayer materials compared with older-generation middle and underlayers, which have lead to improved cost of ownership in the fab.
Status of immersion lithography
Immersion lithography is considered the most promising application to enable lithographers to use ArF for 45 nm HP technology. There has been significant effort put into establishing an immersion process for high-volume manufacturing for the 65 nm HP, including a large focus on tools, materials and process development.8 One of the early issues identified with immersion lithography has been defects in the form of watermarks, bubble defects, residue defects and general degradation of pattern profile caused by leaching of the water into the resist film. These defects have been one of the largest challenges for the materials community, and are being addressed mainly in two ways - immersion topcoats and advanced topcoat-less resists.
Immersion topcoat technology has made significant advances since its early introduction in 2004. Original concerns with immersion stemmed not only from defectivity, but also from resist leaching into the fluid and onto the scanner lens, causing potential issues with imaging and costly lens replacements. Over the past three years, materials suppliers have been able to provide advanced topcoat solutions that have not only quenched leaching concerns (Fig. 5), but have also reduced defectivity levels of immersion processes similar to levels comparable to those of dry processes (Fig. 6).8
From a design perspective, immersion topcoats were initially developed to be very hydrophobic in order to minimize mixing of the water and resist; however, as we learned more about the characteristics of topcoats, it has become clear that there needs to be a balance of surface characteristics between the resist, topcoat and immersion fluid. If the topcoat is too hydrophobic, there is a tendency to cause bubble defects when air becomes entrapped along the advancing angle of the immersion shower head. Additionally, if the surface is too hydrophobic, there is a tendency to leave behind residue defects during the development process, since the developer can have difficulty penetrating the very hydrophobic surface. The key to a successful topcoat is the balance of surface properties that lead to leaching and defect control.
The same surface property characteristics that are applied to advanced topcoats are also used in topcoat-less resist development. The main challenge in topcoat-less resists is increasing the hydrophobicity of the resist surface without causing too much inhibition at the resist surface, which can lead to bridging or residue defects. Topcoat-less resists may include additives that prevent water penetration, while increasing the solubility of the surface after exposure in order to prevent defects. The key to such a design is in the balance between leaching and dissolution control. Much like topcoat designs, it is critical that the surface of the resist not be too hydrophobic, which leads to the advancing contact angle being too high, causing bubble defects.
Advanced ArF resist designs
There are many stumbling blocks to 45 nm HP design with conventional chemically amplified resist systems. Advanced resist design for 32 nm with hyper-NA 193 nm immersion lithography must consider several problems, including resist/fluid interactions, acid diffusion and leaching, and molecular anisotropy caused by the mixing of polymers and additives. In addition, the index of refraction of the photoresist will become a significant factor with further increases in NAs.
The resist/fluid interactions have been addressed by fine-tuning the hydrophobic and hydrophilic properties of the components present in the resist. To prevent higher acid diffusion and leaching, new photoacid generators and polymeric photoacid generators have been proposed. In order to address the molecular anisotropy at smaller dimensions and enhance the resolution and linewidth control, molecular resists have been developed. A modest increase in the index of refraction of base polymer was achieved by both modification of the polymer structure and doping with additives.
Molecular glass resist, such as fluorinated cyclodextrin,9 POSS-based materials,10 cholate derivatives with or without silicon,11 and adamantane-based resist12 were proposed and developed for 193 nm lithography. A POSS-based molecular resist developed by IBM (White Plains, N.Y.) showed higher resolution, better dissolution characteristics and better line edge roughness.10 High-index resists are expected to improve contrast and process latitude in hyper-NA exposure tools, and the refractive indices of sulfur-containing polymers have been found to be greater than that of a standard ArF resist polymer.13 However, it should be noted that the above-mentioned approaches and resist designs require further improvements to be viable at 45 nm HP.
The key to ArF's extendibility for short-term development needs, and eventually 45 nm HP insertion, will be the continued focus on the key items that have been needed in previous nodes. These critical items include high contrast and resolution, low line edge roughness, and mask error factor (MEF) control. All of these items can be managed by proper acid diffusion control, distribution of the resist constituents in the film (both vertically and horizontally) and adjusting the activation energy of the resist. Figure 7 shows an example of extending ArF technology to <45 nm HP CD targets using our advanced methacrylate technology.
Double patterning
ArF double exposure/double patterning are being seen as possible technology alternatives for 45 nm HP. Various approaches to double patterning bring particular advantages, as well as technology hurdles, to the table. Although there are many approaches, this article focuses on three main types:
Double exposure/single develop - Featuring a simpler integration scheme, this approach uses two masks. The first exposure prints the densest pitch, involving an aggressive off-axis illumination (OAI) setting. The second exposure, with aggressive OPC illumination, is applied to the second mask for all the forbidden pitches and isolated features. Demand on the resist material called for a higher-contrast material.
In another scheme called "frequency doubling," on first exposure/PEB, the same layer of resist will be exposed again with a shift of mask alignment. This approach poses a technical challenge to the traditional linear behavior of acid generation of the photoresist and, therefore, some treatment of the resist layer on first exposure might be needed.14
Double exposure with one develop step creates issues such as flare (from the second exposure) and post-exposure delays (from the first exposure); however, the throughput is higher than all other methods.
Double exposure/double develop - One often discussed example among many involves patterning the first layer with negative tone resists and the second layer by positive tone resist. Challenges include limited depth of focus (DoF) because of topography for the second exposure, and intermixing or dissolution of patterned resist when coating the second layer of resist. Nevertheless, there are already some established material techniques that will help to overcome these hurdles. Many established chemical shrink techniques can be employed for double exposure/double develop strategies by forming a narrow semi-isolated trench during lithography, then cross-linking this pattern, followed by a second exposure and develop. Overall, some kind of "freezing" of the first layer is needed before the coating of a second layer of resist for double exposure with the two-develop step method.
Double exposure/double etch - This approach does not put as much demand on new material/resist innovation as the other two methods. However, there are still potential issues of compatibility of the two resists with different hard masks and DoF limitation caused by topography. The potential low throughput might also be a showstopper.
For double exposure techniques, higher contrast and good acid diffusion control are needed. Cross-linking agents, or freezing agents, are needed for double exposure with two develop steps. Resists must have good compatibility with diverse hard masks for double exposure/double develop/double etch approaches.
Conclusions
Chemical shrink, trilayer, immersion and double patterning are all viable options for getting to the 45 nm HP using the same wavelength we've been using for the past 3-4 generations. These options do not come free of process complexities, additional costs and unforeseen complications. However, they are ready for development today, and are much more cost-effective than wavelength-changing alternatives, such as EUV lithography.
Acknowledgements
We would like to thank the JSR Micro Inc. Technical Group for all of its contributions to the development done in the areas described in this paper, along with the JSR R&D group in Yokkaichi, Japan - particularly Tsutomu Shimokawa, Motoyuki Shima, Shiro Kusumoto, Junichi Takahashi and Yoshikazu Yamaguchi.
Mark Slezak is senior technical manager of JSR Micro Inc. Phone: (408) 543-8800 Email: mslezak@jsrmicro.com
Ramakrishnan Ayothi is a senior development engineer at JSR Micro.
Zhi Liu is technical supervisor at JSR Micro.
References:References
- Y. Borodovsky, "Marching to the Beat of Moore's Law," Proc. of SPIE, 2006, Vol. 6153.
- T. Kanda et al., "Advanced Microlithography Process With Chemical Shrink Technology," Proc. of SPIE, 2000, Vol. 3999, p. 881.
- J.H. Hah et al., "Spin-on Hard Mask With Dual BARC Property for 50 nm Devices," Proc. of SPIE, 2006, Vol. 6153.
- A.M. Douvas et al., "Partially Fluorinated, Polyhedral Oligomeric Silsesquioxane-Functionalized (Meth)Acrylate Resists for 193 nm Bilayer Lithography," Chemistry of Materials, August 2006, Vol. 18, p. 4040.
- R. Kwong et al., "Evolution of a 193-nm Bilayer Resist for Manufacturing," Proc. of SPIE, 2002, Vol. 4690, p. 403.
- H. Ito et al., "Silsesquioxane-Based 193 nm Bilayer Resists: Characterization and Lithographic Evaluation," Proc. of SPIE, 2005, Vol. 5753, p. 109.
- H. Sugita et al., "Spin-on-Glass (SOG) for the Trilayer Imaging Process," J. Appl. Polymer Sci., 2003, Vol. 88, p. 636.
- H. Nakagawa et al., "Improvement of Watermark Defect in Immersion Lithography: Mechanism of Watermark Defect Formation and its Reduction by Using Alkaline Soluble Immersion Topcoat," Proc. of SPIE, 2006, p. 6153.
- H. Kudo, N. Inoue, I. Nishimura and T. Nishikubo, "Novel Molecular Photo-resists Based on the Cyclodextrin Derivatives, Containing Fluorine Atoms and t-Butyl Ester Groups," Bulletin Chemical Society Japan, 2005, Vol. 78, p. 731.
- R. Sooriyakumaran et al., "Molecular Resists Based on Polyhedral Oligomeric Silsesquioxanes (POSS)," Proc. of SPIE, 2005, Vol. 5753, p. 329.
- D. Shiono et al., "Molecular Resists Based on Cholate Derivatives for Electron-Beam Lithography," Japanese J. Appl. Phy., 2006, Vol. 45, p. 5435.
- S. Tanaka and C.K. Ober, "Adamantane Based Molecular Glass Resist for 193 nm Lithography," Proc. of SPIE, 2006, Vol. 6153.
- W. Conley and R. Socha, "Everything You Ever Wanted to Know About Why the Semiconductor Industry Needs a High Refractive Index Photoresist: But Were Afraid to Ask, Part 1," Proc. of SPIE, 2006, Vol. 6153.
- M. Maenhoudt et al., "Double Patterning Scheme for sub-0.25 k1 Single Damascene Structures at NA=0.75, ?=193nm," Proc. of SPIE, 2005, Vol. 5754, p. 1508.
author: Mark Slezak, Ramakrishnan Ayothi and Zhi Liu, JSR Micro Inc., Sunnyvale, Calif., www.jsrmicro.com
Semiconductor International. Copyright © 2007 Reed Business Information, a division of Reed Elsevier Inc. All rights reserved.