Integrating dielectric etching with 193nm resists
10/01/2002
overview The introduction of 193nm lithography into 130nm production lines has improved lithographic capability, but has also created significant integration issues at the etch steps. The culprits are inferior etch resistance, reduced mechanical stability of printed features, and a reduction in thickness relative to 248nm resists. The industry is responding with novel integration schemes such as metal hardmasks, new etch chemistries, and pre-etch treatments.
Integrating 193nm photoresists is considered to be one of the biggest challenges faced by the etch community. The transition from 248nm to 193nm radiation results in reduced depth-of-focus, and therefore requires a reduction in resist thickness, from 4500–6000Å to 3000–3500Å. Demands on the optical properties of 193nm photoresists have forced radical changes in polymer composition. For two of the most demanding processes, high-aspect-ratio contacts (HARC) and dual damascene (DD) patterning, where aspect ratios can be as high as 20:1 and 6:1, respectively, etch is facing complex manufacturing issues calling for new process conditions and the development of novel schemes.
Photoresist may be unstable in a plasma environment. In addition to being etched away, the resist may lose pattern integrity, becoming "wrinkled," during the etch process. As reactive and energetic plasma species interact with the resist, the surface and sidewalls of patterned features may roughen, leading to striations and pinholes in the dielectric film.
null
Compared to previous generations of resists, 193nm resists are displaying more pronounced and more rapid morphological changes in plasma etch environments. Compounding the problem is increased "line-edge roughness," which occurs after patterning but before etching (Fig. 1). During exposure and development, 193nm resists tend to produce roughened surfaces along the sidewalls of lines for trench and gate structures. Subsequent etching is performed using an already imperfect mask surface, neither straight nor flat. Thus, the etching industry is faced with two closely related problems —increased roughness on incoming material and a greater amount of roughness incurred during the etching process.
Etch resistance
The concept of etch resistance goes beyond "etch rate," the speed at which a material is etched. It embodies the more serious morphological changes in the photoresist, a phenomenon observed in blanket films, lines and spaces, and complex patterns. One characteristic of 193nm resists is noticeable shrinkage when exposed to the electron beams of scanning electron microscopes (SEMs). The resulting "line slimming" distorts top-down critical dimension (CD) measurements and sidewall profiles [1].
In a similar manner, exposure to plasma can roughen these nascent resists, affecting the integrity of the resist mask. Under etch conditions, which have little effect on 248nm resist, the 193nm resist may become severely deformed (Fig. 2). This mask degradation ultimately impacts the shape of the transferred pattern, appearing as pinholes in the remaining dielectric, sidewall striations, and twists and weaves on lines and spaces.
null
Roughening causes the resist thickness to vary widely. During etch, thinner regions may etch away faster, increasing the roughness and creating deep pits in the resist. One explanation for the enhanced etch rate in the "valley" regions is that charge accumulation on the surrounding "peaks" focuses and floods ions to that region. If the resist is completely removed in these regions, the underlying dielectric will be rapidly etched, opening tiny pinholes ~20nm in dia. (Fig. 3).
Striations appear as vertical grooves, cut into the sides of lines, vias, and contacts. They may be initiated by pre-existing or evolving irregularities in the resist mask, which are transferred to the underlying dielectric film during etch. Integration issues arise during CD measurements due to asymmetry dictated by the striations. In-line CD monitoring is either interrupted or the accuracy of the measurement impaired. Further, striations can create shorts between contacts and lines and affect resistance and capacitance, all of which can significantly impact yield. Additionally, striations may prevent complete surface coverage of subsequent depositions. In copper DD, for example, a barrier layer is deposited in vias, followed by a Cu seed layer. Incomplete coverage can leave behind voids or micro voids, leading to severe reliability problems.
Some of the worst integrity problems faced when etching 193nm resists are obtained with dense line patterns, as required for DD trench etch and gate etch. In particular, resist lines tend to fall over during etch, sometimes with a quasiperiodic, alternating pattern. This can create twisted or weaving lines and spaces in the dielectric film, or regions of incomplete etch where the dielectric becomes shaded (Fig. 4). The primary goal in the integration of 193nm resists with etching processes is a reduction in the severity of these morphological changes.
193nm resists
The move from 248 to 193nm was a major challenge for the photoresist sector. Currently available 193nm resists mainly utilize methacrylate or olefin-anhydride polymer platforms [2]. These resists lack the maturity of their 248nm counterparts, and suppliers continue their efforts to improve these formulations, striving to maintain good lithography process latitude while increasing etch resistance.
Both 193nm and 248nm lithographies employ chemically amplified resists, where a nonabsorbing resin, transparent to the exposure wavelength, is required. The light activates a low-molecular-.weight photo acid generator. During a subsequent bake step, the acid catalyzes deprotection reactions, making the resin soluble in developer. A common component of lithography resins is phenolic groups, which provide inherent etch resistance by reducing the amount of hydrogen in the film. However, phenolic groups absorb strongly in the UV region, preventing the necessary transparency of shorter-wavelength 193nm resists. Alternatives such as carbon-carbon double bond functionality face similar limitations.
 Figure 3. A 193nm resist mask, roughened during etch, transferred ~20nm pinholes into the top surface of the underlying organosilicate (OSG) dielectric layer. |
null
The primary strategy for boosting the carbon-to-hydrogen ratio and, correspondingly, etch resistance, is the use of multiringed aliphatic groups. Alicyclic groups, typically based on adamantane and norbornene, are incorporated into the polymer. Each ring is equivalent to eliminating two hydrogen atoms. This approach has some success in increasing etch resistance, but has not produced a result comparable to 248nm resists [3].
The first 193nm resists used the poly methyl methacrylate platform. They have achieved good lithographic performance, but have limited etch resistance. A newer family of polymers based on a rigid backbone platform of cyclo olefin-maleic anhydride copolymer are designed to have better etch resistance, but do not resolve all etch problems. Efforts continue to improve resin formulations, including a hybrid approach drawing on both classes of polymer. But while improvements in 193nm resists are underway, the industry must work with chemistries available today. A key issue is the stability of these resists in the etching environment.
HARC and DD etch applications
The formation of contacts with high aspect ratio (10:1 to 20:1) constitutes one of the most demanding applications in dielectric etch. Process windows tend to be smaller, leaving little leeway to compensate for resist constraints. Key objectives to etching this leading-edge structure are: vertical profiles, maintaining CDs, good selectivity to the stop layer, and striation-free sidewalls. Because HARC requires a deep oxide etch, selectivity to photoresist is one of the critical control parameters. Without adequate selectivity, top CDs will be poorly controlled. Mechanistically, a good HARC etch relies on tight system controls and optimized gas chemistries.
The critical issues and criteria in etching vias in DD patterning are similar to those of HARCs. While the aspect ratio for DD vias is smaller, typically <6:1, the challenges can still be significant. The dielectric being etched may consist of several different stacked components such as a hardmask (HM) or cap, trench dielectric, intermediate stop layer, via dielectric, and barrier film. Additionally, the dielectric films are often new and nascent materials intended for low-k application. This requires greater flexibility in process control and the ability to change processes rapidly as etching proceeds through one material, encountering another. Selectivity is also important, as is fidelity of the profile through the dielectric stack. Bowing and undercut must be minimized to ensure subsequent coverage of the metal barrier and seed, and prevent void formation during copper electroplating.
DD etch issues also cover trench-, trench-over-via, and barrier-opening steps. The waviness attributable to unstable 193nm resists can occur between trenches, where it can cause shorts between metal lines; increased capacitance due to reduced thickness; and poor copper filling and resulting reliability issues.
null
When etching a trench with vias present, the issues with 193nm resists are compounded. In the commonly used via-first DD scheme, the via is typically filled or partially filled with a polymer, whether from a bottom antireflective coating (BARC) layer or a specific process to fill and etch back. This plug material etches very slowly, especially if resist passivation techniques are employed, and can result in "fencing" where the plug shadowing prevents etching of dielectric just adjacent to the via hole inside the trench.
Typically, integration schemes use a recess etch (extended BARC etching) or tune down the selectivity to polymer material so that the trench dielectric etch rate and resist/polymer etch rates are more similar; clearly, the reduced thickness and increased etch rate of 193nm resists makes this problematic. Furthermore, any corner of any feature exposed to the plasma, whether photoresist or the top corner of a via inside a trench, can be potentially cut off, forming a chamfered surface or "facet."
The plasma etch environment
The plasma etch environment is characterized by reactive radical species, energetic ions, free electrons, and ultraviolet radiation. Each can interact with a material in a potentially harmful way. Typical dielectric etch process conditions apply thousands of watts of power to the plasma, producing a host of fluorocarbon radicals, ions with hundreds of electron-volts of energy, and radiation extending from the infrared to the soft x-ray. The same plasma radicals and ions required for etching may also roughen photoresist. The key to preventing resist damage is process control.
From the standpoint of etch, the most significant impact of the radically different 193nm chemistries is a measurable shift in the composition of the plasma, relative to previous-generation resists. This highlights the need to understand the changes in the plasma and determine how they will alter the etching process. For this reason, plasma diagnostics, such as optical emission spectroscopy and ultraviolet-visible absorption spectroscopy may be needed to study the influence of photoresist. These noninvasive probes measure concentrations of radical species in the plasma. As process requirements are continually tightened, spectroscopic techniques and other diagnostic methods are becoming necessary tools for etch integration.
When etching a substrate, the plasma predominantly interacts with photoresist; HARCs and vias typically comprise only a few percent of the total wafer area. Therefore, byproducts from the interaction of the plasma with the resist may have a profound influence. Results of an internal study revealing significant differences in the process environment are being used to tailor 193nm etch processes.
Process stability for 193nm resist
The resist's heightened sensitivity to the plasma environment can be exacerbated if process conditions drift, such as, for example, if residue on the chamber walls changes over time and interacts with the plasma. To address this issue, the etch system can be designed to operate so that walls stay residue-free and memory effects are minimized. This keeps the system centered on the limited process window for 193nm resist, and also allows the use of a variety of different plasma chemistries in one chamber, for example, to etch the complicated stacks required by some integration schemes. The ability to tightly control chamber conditions can ensure a consistent processing environment over long production runs, allowing the handling of thousands of wafers between preventive maintenance. When developing an etch process for this temperamental resist, which may require etching complex HMs or bi- and trilayer resists, flexible chemistries and process stability over months of operation become critical.
Improving etch resistance
The most direct way to improve the etching performance of 193nm resists is to modify the etch conditions. This may be a way to maintain current integration schemes and at the same time "fix" the process to improve results, a simpler path in terms of integration. One approach is to find process regimes where the etching of the underlayers can still be accomplished, but which are gentler to the resist. This would include hardware modifications and control of gas, power, temperature, and other parameters, typically to less energetic processes.
Another method is to passivate the resist surface, for example, by increasing the polymerization properties of the etch plasma. The resist is protected from the influence of the plasma and may become more resistant to etch. Ideally, this approach will not change the CD profile or other specifications of the etched features.
null
The choice of gas chemistry is critical for each approach, with the use of novel additives representing one of the most promising directions. The effect of process modification on etching trenches in an organosilicate (OSG) dielectric using 193nm resists was recently demonstrated (Fig. 4). In this study, two different etch recipes were compared, a standard 248nm process and one optimized to 193nm resists. By using an alternative gas chemistry as a means of controlling the photoresist, resist roughening and line weaving were significantly reduced.
The industry is also looking at pre-treatments, which are performed after resist patterning, separate from and prior to etch. A variety of techniques are currently in development that would ideally improve the etch properties of the resist without compromising CDs or pattern integrity. This approach is desirable from an etch standpoint because less modification of process conditions will be needed; it would add cost and complicate integration. In any case, an optimal pre-treatment strategy has not yet emerged.
Integration approaches
Integration schemes are being developed to transfer the pattern from resist into another layer that can better withstand the plasma environment. With both metal HMs and multilayer resists, an unstable 193nm resist mask is quickly transferred into another layer that is more robust to the plasma. The integration schemes applied to HARCs and vias using 193nm resists are designed to address high-aspect-ratio etching and related resist issues.
In DD patterning, HMs are already a reality. In particular, dual HMs have been implemented to handle organic low-k materials. Metal HM technology, however, is new to DD, driven by the arrival of 193nm resists. Metal HMs are typically thin (~15–50nm) so they can be quickly etched away in shorter etching times and with potentially lower energy plasmas, before the resist integrity is lost. Materials under consideration are titanium nitride (TiN) and tantalum nitride (TaN). A single- or dual-metal HM will likely be used in a via-first DD process (Fig. 5).
There are two basic approaches for processing metal HMs. The first uses a combination of etch systems. The metal HM is opened in a metal etch system and the subsequent etch steps are performed as usual in a dielectric etch system. The metal etch tool typically uses chlorine gas and a high-density plasma, and can quickly etch through the metal HM with good fidelity. The resist may then be stripped and a standard dielectric etch follows to complete etching of the via. Trench formation proceeds, repeating the process with the metal HM once again opened early on.
With an imperfect process, the presence of the metal can pose problems, with the byproduct of the metal etch at times redepositing onto the sidewalls of the resist. Following resist removal, residual fences may appear at the very top of the material, requiring a subsequent wet treatment for selective removal. Alternatively, the metal and dielectric etch modules may be combined on a single platform with wafer transfer occurring in situ. Maintaining a vacuum may prevent the reliability issues; however, these systems are still under evaluation.
The second approach to implementing the metal HM scheme performs both the metal and dielectric etches in a dielectric etching system. This is a newer and less developed approach, though rather intriguing from a cost perspective. However, there are concerns from a manufacturing standpoint, including questions about the impact of metal residues and byproducts after long-term usage. With metal HMs high on the list of integration options for 193nm resists, opening the metal mask with minimum contamination is a top priority.
Many of the major resist manufacturers are actively developing 193nm bilayer and trilayer resists. Classic multilayer systems separate the imaging and etch layers, the uppermost layer designed for optimum imaging properties and the underlying layer(s) for dry etch resistance. Despite its longevity, this technology has seen little commercial success because of the complexity of the process, defect levels, and cost. As linewidths continue to shrink, control issues in terms of iso/dense CD and lot-to-lot variations, become primary concerns. However, a potential door-opener for 193nm bilayer and trilayer resists is the current activity to extend 248nm lithography with 248nm bilayer chemistries. Integration solutions at 248nm can be passed on to the next generation [4].
Conclusion
As resist suppliers continue to improve 193nm resists, the focus in the etching community is on learning how to integrate current 193nm chemistries into the etching process, with a greater emphasis on understanding and developing the necessary etch processes and systems. While multilevel HM and other novel approaches may provide short-term solutions, they are costly and complex. The bilayer approach, if suppliers can demonstrate its manufacturability, may also play an important role in the future. Ultimately, more traditional integration schemes will be implemented as etch systems, processes, and the resists evolve and 193nm resists become more fully integrated into production lines.
References
- T. Kudo et al., J. Photopolymer Sci. and Tech., Vol. 14, No. 3, pp. 407–413, 2001.
- M. Rahman, J. Bae, et al., Advances in Resist Technology and Processing XVII, Proc. SPIE, Vol. 3999, p. 220, 2000.
- R. Kunz et al., SPIE, Vol. 2724, pp. 335–376.
- I. Pollentier et al., Proc. Interface 2000, pp. 265–283, Nov. 2000.
For more information, contact Steve Lassig at Lam Research Corp., 4650 Cushing Parkway, Fremont, CA 94538; ph 510/572-3787, e-mail [email protected].