Low-k damage reduction using a modified plasma strip process

05/01/2010

Executive OVERVIEW As feature sizes decrease for each succeeding technology node, scaling of the inter-layer dielectric (ILD) k-value enables chipmakers to achieve the required signal speed and to control RC delay. This scaling is usually accomplished by increasing the porosity and carbon content of the ILD low-k material. Minimizing plasma-induced damage from etch and strip is an important integration challenge for advanced damascene interconnect structures, especially with ultra-low-k (ULK) materials (k < 2.4). This article describes a new plasma strip method that minimizes damage to the low-k film, while maintaining the required post-etch polymer removal efficiency.

Rajesh Mani, Bing Ji, Steve Sirard, Andrew Bailey, Lam Research Corp., Fremont, CA, USA

Reducing low-k damage due to plasma strip is critical because this step causes the majority [1,2] of the low-k damage during the etch–strip process sequence (Fig. 1). The damage is due largely to the fact that the low-k film is directly exposed to the plasma during strip/over-strip and the plasma chemistry is typically oxidizing and rich in radicals.

Figure 1. Progression of low-k film damage during a typical etch–strip process sequence, with the highest damage occurring during the over-strip step.

Low-k damage concerns

Exposure of low-k films to plasma during etch and strip processes can change the fundamental structure of the low-k film. These changes may include densification of the low-k film matrix (decreasing the porosity), absorption of moisture into pores, and/or removal of carbon from the film. This low-k damage causes an undesirable increase in the k-value of the ILD material, and ULK films are particularly susceptible due to the higher porosity.

In addition to lower k-values, the reduced CD spacing at advanced technology nodes further drives the need to control low-k damage from oxidizing strip chemistries. Low-k damage can lead to unacceptable changes in CD spacing. For example, widening via dimensions may lead to metal-to-metal leakage issues after copper fill and degradation of device reliability (time-dependent dielectric breakdown, TDDB). Furthermore, plasma-induced damage may cause an undercut between the mask/cap and low-k films, which may result in voids after metal fill, thereby affecting overall device reliability.

Trade-off between damage and polymer removal

Strip-based low-k damage is influenced primarily by the plasma radical density and the extent of radical diffusion into the low-k film (Fig. 2). Strip processes typically use oxidizing chemistries to improve the efficiency of post-etch polymer removal, leading to oxygen radical diffusion into the low-k film, which can cause damage [3]. Plasmas used during low-k strip processes have evolved from pure O₂-based chemistries to CO₂-based chemistries to ensure good polymer removal, while simultaneously achieving reduced oxygen radical diffusion. CO₂-based chemistries have been used effectively for the 65nm and 45nm nodes. This approach, however, faces issues in terms of controlling damage to the lower k-value films used at the 32nm node and beyond.

Figure 2. Diffusion of oxygen radicals (O*) into a low-k film causes damage by converting SiCOH into SiO2. The amount of damage can be measured by dipping the structure in dilute HF, which removes the SiO2-like layers.

It is expected that at the 28nm technology node, there will be wide-spread adoption of a metal hardmask (MHM)-based "all in one" (AIO) integration scheme—in which the via etch, strip, trench etch, and barrier open are all performed sequentially in the same etch chamber. The advent of the MHM AIO scheme could eliminate one of the root causes for low-k damage, namely post-trench strip. However, low-k damage will continue to be an issue due to post-via strip, which is still required because existing post-via wet clean methodologies are not capable of achieving the desired polymer removal efficiency by themselves. Additionally, via etch processes tend to use a more polymerizing process chemistry than trench etch processes, so there is more polymer to be removed after etch. As a result, a longer strip process is needed, thereby increasing the risk of low-k damage.

Approaches to minimizing low-k damage

Low-k repair and damage reduction are two approaches that have been pursued to address plasma-induced damage with different degrees of success. Low-k repair has traditionally involved the use of vapor-based surface treatment methods. In this approach, the low-k ILD film is exposed to a gas such as a silylating gas (e.g., HMDS, hexamethyldisilazane), to increase the carbon content in the film's matrix (depleted during plasma etch or strip). Such repair methods show advantages in terms of recovering the k-value of damaged low-k films [4]. However, it appears that these effects are transitory and the benefits of repair at one metal level are lost after the next metal level has been processed. In addition, repair methods add extra costs to the device manufacturer because of the need for specialized repair chambers. The replenishment of carbon in the low-k film matrix may be achieved in a simpler fashion through the use of post-etch treatments within the etch chamber itself, for example, by using methane (CH₄) as a post-etch treatment gas.

The other approach, damage reduction, aims to minimize low-k damage during the etch/strip processes themselves. This targets the root cause for plasma-induced damage, namely the effects of oxygen radical diffusion during strip, and has the added benefit of utilizing existing etch equipment. A new plasma strip method, Hybrid Strip, has been developed by Lam Research to slow the diffusion of oxygen-based radicals into low-k films, while maintaining polymer removal efficiency during strip.

A modified plasma strip process

The Hybrid Strip process involves a mix of etch, polymer deposition, and strip steps. This differs from a traditional process sequence (Fig.1), where etch steps (main etch, over-etch) are carried out first, followed by strip steps (strip, over-strip). In the new method, the process sequence is changed after the main etch: short process time etch and strip steps are performed one after the other repeatedly, with a polymer deposition step interspersed between.

The modified process sequence—which involves rapid changes between etch, polymer deposition, and strip—has the effect of minimizing oxygen radical diffusion into the low-k ILD matrix, thus reducing low-k damage. The deposition step helps in protecting the low-k film by creating a physical barrier against oxygen radicals during the etch and strip steps. A significant portion of the post-etch polymer, as well as some of the damaged low-k layers, are removed during the modified process. The remaining post-etch polymer residues, including the polymer deposited by the new process, are easily removed during subsequent wet clean steps.

Figure 3. Progressive polymer removal from a k = 2.2 via structure using the Hybrid Strip process.

The modified process ensures effective polymer removal by allowing the continued use of CO₂-based chemistries (Fig. 3) while minimizing the associated risk of oxygen radical diffusion and low-k damage. ILD films with k-values as low as k = 2.2 have been processed with the Hybrid Strip method, and significant improvements in low-k damage control have been achieved (Fig. 4). Damage evaluation carried out by dipping the low-k film into a dilute HF solution to remove damaged material has shown as little as ~3nm per side of physical damage. Electrical tests have shown up to a 7% improvement in RC delay performance for the new strip process vs. conventional CO₂ strip.

Figure 4. Comparison of low-k damage for a k = 2.2 film using a) a conventional CO2 strip (sidewall damage ~7-8nm per side) vs. b) Hybrid Strip (sidewall damage ~3-4nm per side).

Conclusion

Integration of ULK films is a continuing challenge due to plasma-induced low-k damage issues during etch and strip processes. In particular, strip processes, which involve the use of an oxygen radical rich chemistry to remove post-etch polymer residues, account for a majority of the plasma-induced low-k damage. The damage is due to increased radical diffusion into the exposed dielectric film. To allow the integration of ULK films into advanced dual damascene structures, strip processes must balance removing post-etch polymer with minimizing damage to the low-k film. A new strip method has been developed to achieve both of these requirements simultaneously. This method has shown improved results versus conventional strip for films with k-values as low as k = 2.2. In addition, the modified process offers a lower cost option vs. low-k repair techniques since existing etch equipment can be used.

Acknowledgements

The authors would like to acknowledge Gerardo Delgadino, Robert Hefty, Kenji Takeshita, Maryam Moravej, Daniel Le, Jungmin Ko, Jim Bowers, and Bi-Ming Yen from Lam Research Corp., for their assistance in carrying out the experiments and in the interpretation of the results.

References

1. S. Satyanarayana, R. McGowan, B. White, S.D. Hosali, "Damage Mechanisms in Porous Low-k Integration," Semiconductor International, June 2005.

2. A.M. Urbanowicz, V.V. Talanov, M. Pantouvaki, H. Struyf, S. De Gendt, M.R. Baklanov, "Evaluation of Plasma Damage in Blanket and Patterned Low-k Structures by Near-field Scanning Probe Microwave Microscope: Effect of Plasma Ash Chemistry," IITC 2009 Proc., 134-136.

3. S.K. Singh, A.A. Kumbhar, R.O. Dusane, "Resisting Oxygen Plasma Damage in Low-k Hydrogen Silsesquioxane Films by Hydrogen Plasma Treatment," Materials Letters 60 (2006) 1579–1581.

4. T. Rajagopalan, B. Lahlouh, J.A. Lubguban, N. Biswas, S. Gangopadhyay, J. Sun, et al., "Investigation on Hexamethyldisilazane Vapor Treatment of Plasma-damaged Nanoporous Organosilicate Films," Applied Surface Science 252 (2006) 6323–6331.

Biographies

Rajesh Mani received his B.Tech in materials science and engineering from the Indian Institute of Technology, Madras, and MS in materials science and engineering from the Georgia Institute of Technology, and is a product marketing manager for dielectric etch at Lam Research Corporation, 4650 Cushing Parkway, Fremont, CA 94538, USA;  [email protected]

More Solid State Technology Current Issue Articles
More Solid State Technology Archives Issue Articles