Solutions-based approach to developing next-generation CVD precursors
06/01/2004
Providers of new low-k technologies have typically attempted to develop a complete integrated solution on their own by using proprietary technology and hoping for universal adoption in the market. The problem with this approach is that the resources required for any one company are large, and the financial risks are very high. A better way is to collaborate with other companies, where each party brings a particular area of expertise to the table, so that a variety of possible solutions can be pursued simultaneously, while minimizing the risk of investing in a losing technology.
To be successful in developing and commercializing new materials technology, companies must be willing to bet on multiple possible solutions. Focusing on a single proprietary solution to today's materials challenges not only limits innovation, but entails an unacceptable risk for technology providers. A recent article published by the Semi Chemical and Gas Manufacturers Group (CGMG) said that the cost of new materials development in some cases may exceed the potential size of the market [1]. Clearly, a different approach is needed.
A multipronged approach is especially important in the development of next-generation CVD precursor solutions. Using molecular engineering as a foundation, it is possible to develop broad families of materials that can be used for advanced low-k interconnect dielectric, copper barrier, hardmask, and low-temperature applications. Compared to a single material, families of materials offer the industry a flexible chemistry set that users can tailor to address specific needs, such as dielectric performance, mechanical integrity and chemical resistance.
The implementation of this approach at Dow Corning has led to the development of a number of materials suitable for use in advanced semiconductor manufacturing. One particular family of CVD precursors has been developed for interconnect applications requiring a k value in the 2.2–2.6 range.
There are two main approaches to lowering the dielectric constant of the interconnect layer. The first is to incorporate increasing levels of porosity into a matrix material to lower the overall density, and thus the permittivity. This method has several drawbacks, including a significant drop in the mechanical strength of the films and difficulty in sealing the pores during subsequent integration steps.
 |
A second approach to lowering permittivity is to change the structure of the material by incorporating atomic species such as carbon and fluorine, which have lower polarizability compared to silicon and oxygen, into standard SiO2 dielectric layers. This method has problems as well, including lower stability of the overall film to additional thermal processing, and poor electrical leakage performance.
Focusing on molecular structure/composition
To overcome the drawbacks of the typical approaches discussed above, our recent work has focused on molecular structure and composition changes that reduce the density of the film and polarizability of the bonds, thus reducing permittivity. At the same time, these changes minimize porosity and improve chemical stability.
We began by choosing molecular structures with specific bond strain and asymmetry. Through careful chemical synthesis, specific carbon-to-silicon ratios were established in the precursor molecules. Adjustment of the CVD process parameters of temperature, pressure, power, and gas feed enabled control of the decomposition characteristics in the plasma. The combination of precursor design and process selection allows for tailoring of the film properties between an Si-O and an Si-C character. Based on a model, several molecules have been identified, synthesized in high purity, and evaluated as PECVD precursors for advanced low-k dielectric applications (see figure). Materials A–D are individual precursor materials that have been evaluated as part of this effort.
The new PECVD process chemistries have demonstrated that it is possible to reduce the relative permittivity to the range of 2.5–2.2 with specifically designed precursors. In addition, the k value of the thin films appears to be tunable over about a 10% range. The typical properties of these advanced dielectric films are shown in the table. In addition to the low-k property, these films also have good electrical isolation characteristics and show thermal stability to 400°C.
Achieving success requires collaboration across the industry to round up the needed expertise to make a new material work. While materials companies have a great deal of expertise in key technologies such as silicon chemistry, materials synthesis and high-volume chemical manufacturing, solving the problem also requires expertise in processes such as etching, CMP, stripping, and cleaning to ensure materials integration into the finished device. To pioneer new technology, everyone — the materials suppliers, process equipment makers, and device manufacturers — needs to step up and share both the risks and rewards.
Conclusion
By focusing on families of materials and using molecular modeling to predict performance, it is possible to develop materials needed for future generations of semiconductor devices. The commercialization of these advancements into high-volume manufacturing requires the additional collaboration of industry suppliers and device manufacturers willing to take the next step in the continued drive for higher performance and lower cost devices. Offering multiple solutions and collaborating with the rest of the industry to demonstrate performance raises the probability of commercial success for everyone, without burdening any one company with the entire resource load.
Reference
- "Chemical and Gas Suppliers Speak Out," CGMG article published on the Semi web site, March 2, 2004; go to http://www.semi.org to view article.
Phil Dembowski is the global market manager for Semiconductor Fabrication Solutions at Dow Corning Corp.; e-mail [email protected].
B.K. Hwang is a senior thin-film specialist in the Thin Film Technology Platform at Dow Corning Corp.
Mark Loboda is an electronics industry scientist and R&D manager for the Compound Semiconductor Business at Dow Corning Corp.