Multicolor lithography

For decades, the predominant strategy for maintaining the course of Moore's law in semiconductor lithography (that the density of transistors in integrated circuits doubles every two years) has been to find ways to perform photolithography with ever shorter wavelengths of light. Each reduction in wavelength brings its own set of challenges in optics and materials chemistry. However, as the wavelength of light reduces, generating, propagating, and manipulating light becomes increasingly difficult and expensive, and it becomes more challenging to develop suitably robust materials. Consequently, it is not clear how much longer the strategy of wavelength reduction can be sustained.

Recently, a new class of approaches to high-resolution photolithography has been developed that relies on the use of two or more colors of visible (or near-IR or near-UV) light.^{1, 2} These techniques, which were inspired by stimulated emission depletion (STED) fluorescence microscopy,³ use one or more colors of light to expose a photoresist and one or more colors of light to counteract the exposure. Each color of light can have a different spatial pattern, which makes it possible to create features that are far smaller than the wavelengths of light employed.

To understand how two-color approaches to photolitho graphy work, consider the example of STED for deactivation of exposure of the photoresist. One color of light is first used to excite the photoinitiator (PI) molecules in the photoresist in a desired pattern (see Figure 1). A second color of light, which is of a longer wavelength than the first color of light, is used to drive the PI molecules back to the ground state before they have had the chance to cause chemistry to occur within the photoresist. The pattern of the deactivation is typically complementary to the pattern of excitation, such that the photoresist ends up being exposed effectively only in the regions in which the intensity of the deactivation beam is at a minimum. We have investigated a number of different deactivation mechanisms for such two-color lithography.¹ The current record for feature size is 9nm using 800nm light for two-photon excitation and 375nm light for deactivation.⁴

Figure 1. (a) Spatial intensity pattern (top) and the resultant pattern in the exposed and developed photoresist (bottom) with a conventional, one-color approach. (b) In a two-color approach, a pattern in a second color of light is used to reverse the effects of excitation, leading to a considerably smaller effective exposure region and thereby a finer feature size.

One limitation of two-color schemes is that, although features can be much smaller than either of the wavelengths of light employed, the minimum pitch (i.e., the distance between two features) is determined by the wavelength used for deactivation. Thus, multiple patterning steps are required to obtain densely packed features. (Note that these multiple patterning steps are all performed on the same photoresist film without removing it from the patterning tool, and so should not be confused with current multipatterning schemes.) However, in two-color schemes deactivation and chemistry both occur from the same state, and so these processes necessarily compete with one another: see Figure 2(a). As a result, it is never possible to completely deactivate a region of the photoresist that has been excited. This phenomenon leads to a degradation of the resolution as the number of patterning steps increases.

Figure 2. Energy-level diagrams for photoinitiator (PI) molecules. (a) In two-color approaches, chemistry and deactivation are driven from the same state of the PI, leading to an inevitable competition between the two processes. (b) In this three-color approach, deactivation occurs from a ‘pre-activated’ state. Only after the desired regions have been deactivated is the PI transferred to the activated state from which chemistry occurs, eliminating the competition between these processes.

One way to address the competition between deactivation and initiation is to perform these two processes from different energy states: see Figure 2(b) for one possible scheme to achieve this end. Light of one color takes the PI to a ‘pre-activated’ state. A second color of light can deactivate this state. Molecules that have not been deactivated can then be transferred to an activated state with a third color of light, after which chemistry occurs. This three-color approach not only eliminates the competition between deactivation and initiation, but also allows all three colors of light to be patterned to attain even higher resolution.

In summary, multicolor approaches to lithography offer many potential advantages over single-color schemes. Visible light can be generated, propagated, and manipulated readily and at relatively low cost. It is also much gentler on materials than is short-wavelength radiation, offering many more potential approaches to photoresist chemistry. However, there are also many challenges that still must be addressed in the development of multicolor approaches that will be attractive for industry. For instance, on the materials side, most work to date has involved materials for direct-write, two-photon lithography. We are working to develop efficient photoresist chemistry that can be implemented in thin films for large-area, linear exposure and is compatible with high-fidelity pattern transfer. On the hardware side, our collaborators at Periodic Structures Inc. are developing high-throughput tools that implement highly reproducible multiple patterning steps with multiple colors of light. Considerable headway has been made on both fronts, and with continued progress multicolor approaches have the potential to be a game changer for industrial photolithography.