Sponges, Skyscrapers, And Low-K

New strategies for using low dielectric constant materials.


A sponge is a porous structure. So is a skyscraper. These two very different images exemplify the materials being considered for advanced low dielectric constant (κ) materials.

Most porous dielectrics that have been tested up to this point resemble sponges. As Intel’s David Michalak explained at this month’s Materials Research Society (MRS) Spring Meeting, these materials consist of a backbone polymer and a sacrificial poragen. Process details vary, but typically a mixture of the two materials is spin-coated onto a wafer, and the poragen removed by either heat treatment or chemical action. The backbone polymer is left behind and forms the mechanical structure of the layer.

The challenges presented by this type of material are well known. Typically, some kind of protective coating or surface treatment is needed to keep contaminants from infiltrating the pores. The backbone must be chemically stable under process conditions, and must be rigid enough to maintain its structure after the poragen is gone.

An even more fundamental challenge, however, is inherent in the sponge-like structure of these materials. For any given backbone material, reducing the dielectric constant will always involve increasing the porosity of the sponge. Unfortunately, systems with random porosity are constrained by the percolation threshold.

Percolation theory appears in models of dielectric breakdown, where it describes the formation of a connected failure path through the dielectric layer. Similarly, the percolation threshold in a porous dielectric is the maximum porosity beyond which the film is no longer continuous. (A basic simulation of behavior at the percolation threshold can be found here.) The threshold for a specific system depends on pore size and distribution, but is generally in the vicinity of 50% to 60% or so. Below this threshold, there exists a continuous structure connecting the individual backbone molecules to each other and to the edges of the film. Above it, there does not. Without a continuous structure, mechanical rigidity is lost and the film collapses.

Ordered structures, in contrast, can achieve much higher levels of porosity. Consider the exposed girders of a building under construction. The framework surrounds enormous amounts of open space, but the structural loads from the upper floors are easily transmitted to the lower floors.

This is the approach seen in metal-organic frameworks (MOFs), also discussed at this year’s MRS Spring Meeting. A metal-organic framework is composed of metallic units, tied together with organic “linker” elements. The result resembles a child’s construction toy or a classic ball-and-stick crystal lattice structure.

Metal-organic frameworks have already found numerous applications in catalysis, gas storage, and other fields. They offer a wide range of pore sizes, with a wide selection of metallic and organic elements. Moreover, since they are ordered structures, MOFs are amenable to the full range of materials science simulation tools. It’s possible, in principle at least, to make exact theoretical predictions for a given set of structural parameters. Most commercial MOFs are powders, though. Is it possible to make continuous films of these materials?

It turns out, according to Christof Woell of the Karlsruhe Institute of Technology, that the answer is yes. The Karlsruhe group created highly oriented, layered structures by alternating metal and organic depositions. Both layers appear to be self-terminating, giving designers a great deal of control over the final structure. For example, the composition of an exposed surface could be modified to decrease porosity, or to increase resistance to chemical attack. This was a proof-of-concept study, and so much work remains to be done to achieve process-worthy materials. However, the group did demonstrate creation of a metal-free polymer gel with rigid structure and high porosity by adding a secondary “crosslink” polymer component and then removing the metal units.

The first time the semiconductor industry tried to incorporate porous low-κ dielectrics, it didn’t turn out so well. Even materials that survived integrated circuit fabrication typically collapsed during device assembly. But as often happens in IC manufacturing, the need to reduce dielectric constant never went away. Armed with new materials and a better understanding of the stresses to which devices are subjected, manufacturers hope for a better outcome this time around.

  • Chung Jen Lee

    History is repeating its self: we are still at the days when SELETE given the low k award to DSI in 2003 again?