Materials

The materials used for this process include a photoacid generator (PAG), an inert polymer to serve as planarizing layer and to disperse the PAG, and a silicon-containing monomer to form the graft.  Because the functional requirements of photoresists have been split between the three materials, each one can be separately chosen and/or designed to meet these needs.

PAG

Because the cationic graft process includes an oxygen reactive ion etch step, the materials used must not sputter in the etcher.  This constraint does not affect the polymer or monomer, but does preclude some photoacid generators, such as triphenylsulfonium hexafluoroantimonate (the first candidate).  Other PAGs do not produce acid strong enough (or enough acid) to initiate the graft reaction.  The PAG we are currently using is triphenylsulfonium tris(trifluoromethylsulfonyl) methide, courtesy of Dr. William Lamanna at 3M.  It is entirely organic, avoiding etcher problems, and easily initiates the graft reaction.

Monomer

To ensure good throughput, the monomer must be highly reactive to cationic initiators.  It must also contain enough silicon so that the final graft will resist oxygen etching.  For ease of vapor-phase processing, the monomer should have a reasonable vapor pressure at process temperatures.  The graft must maintain the desired feature shape as defined by the mask; thus, it must not have a low glass transition temperature (Tg).  Given that silicon-containing polymers typically have low Tgs, this means that the ideal monomer for this should cross-link.  Many vinyl ethers and epoxides have been custom synthesized and evaluated for this process.  The current monomer of choice is bis(vinyloxymethyl)dimethylsilane (or DMDVS, see Figure 1). 

Figure 1. ; Structure of DMDVS
Polymer
The polymer that forms the base layer must be inert, in order to avoid monomer deposition on all areas of the wafer.  It must also have a high glass transition temperature to avoid flow of the features formed at process temperatures.  Finally, the polymer must not sorb the monomer excessively nor irreversibly, as this, too, would degrade the contrast between exposed and unexposed areas.  Although this requirement involves both the monomer and polymer, we will address it via the polymer as it has fewer requirements in terms of composition and reactivity - it does not require silicon and needs no reactive moieties.  Early trials with this process involved poly(p-methoxy-styrene) (PMOST).  While this polymer, paired with DMDVS, met the solubility requirements, its Tg was too low to form reasonable features.  Another polymer, poly(p-methoxystyrene-co-N-p-methoxyphenylmaleimide) (PMMI), has been synthesized and tested, as maleimides are known to have high Tgs.  While it has led to the best images yet, it has too high an affinity for the monomer, as shown in figure 2.  We are currently in the process of evaluating a series of other polymers that may meet the criteria outlined here.
(a)
(b)                                    (c)
Figure 2.  (a) PMMI structure.  (b) 200 nm (nominal half-pitch) 1:1 lines imaged at International SEMATECH using 248nm exposure at 35 mJ/cm2.  DMDVS grafted onto PMMI.  (c) Cross-sectional decoration etch (oxygen RIE) of large feature edge.  Note the depth of the graft is about 1/3 the depth of the initial film.  Also note the thin layer that resisted the etch in the unexposed area (left side).  This is likely background silylation by sorbed monomer.