Chemistry and Processing of DUV Chemically Amplified Photoresists


1. Basic Chemistry of DUV (Chemically Amplified) Photoresists  

In the 1980’s, printed critical dimensions at and below 300nm began appearing on device technology roadmaps for memory and advance logic integrated circuits. Research and development engineers with the world’s top manufacturers of exposure systems quickly realized that their current state of the art 365nm laser based scanning tools would not be capable of providing the resolution and process windows required to make volume production at these CD’s viable. In consideration of the Rayleigh criteria: 



it became obvious that shorter wavelength exposure energy would be required and KrF based lasers emitting photons at 248nm became the next logical choice for the illuminators in their next generation exposure systems. 

With the technology roadmap in place for the exposure tools, photoresist chemists soon realized that the workhorse DNQ/novolak platform that had served the IC industry so well through so many decades and technology nodes, had finally reached the limits of its’ technical extendibility. First, since novolak resin is highly absorbing at wavelengths below 300nm, an entirely new polymer platform with high transparency at 248nm would be required. Next, due to the potential for thermal degradation of exposure optics caused by absorption, the new exposure systems required very short exposure times. The target dose for imaging a 248nm photoresist was around 40mJ/cm2 maximum with the ideal being closer to 30mJ/cm2! And finally, because the shorter wavelength exposure energy would significantly reduce depth of focus, the new resists had to be much thinner than the DNQ’s but still block RIE etches and ion implant energy.

Poly-hydroxy-styrene (PHS) was soon identified as a suitable resin polymer for addressing the transparency and etch resistance issues. However, since DNQ type dissolution inhibition imaging systems did not work well with PHS and did not provide the quantum efficiency required to meet the low exposure dose requirement, an entirely new photo-chemical approach to imaging was required.   

To achieve imaging with high quantum efficiency, researchers at IBM (Willson, Ito et al.) attached pendant groups to the polymer backbone which rendered the PHS insoluble in photoresist developer. These particular “blocking groups” were known to convert to highly soluble species in the presence of acid so the final piece of the puzzle was generating the needed acidic molecules in the exposed area of the photoresist film with very high photo-chemical efficiency. Additives known as PAGs (photo-acid generators) which produce the required acids upon exposure to 248nm radiation were developed. This imaging mechanism can be depicted schematically as shown below. 





Note that the acid molecules simply catalyze the reaction that converts the blocking groups from insoluble to soluble (i.e. they are not consumed) hence one acid molecule can “de-block” several pendant sites along the polymer chain via diffusion aided by the post exposure bake (PEB). This feature of the imaging scheme led to the term “Chemically Amplified” and it provides the quantum efficiency required for a very fast (i.e. low exposure dose) photoresist.


2. Processing of Chemically Amplified Photoresists 

Fortunately, processing DUV photoresist is very similar to processing standard DNQ based materials with a few key exceptions. Substrate preparation guidelines are essentially the same since DUV resist requires hydrophobic substrates for adequate adhesion. Likewise, spin coating process variables and optimization techniques are substantially similar between chemically DUV resists and DNQ/novolak resists.

The photoresist bake steps are where the key differences in processing and process control lie. Because the activation energy associated with the de-blocking reaction and the diffusion length of the acid molecules during PEB are both critical to a particular DUV resist’s performance, the manufacturer’s recommended bake temperatures and times must be strictly adhered to in almost all cases. It is worth mentioning again, unlike with DNQ resists, the PEB step is not optional when processing chemically amplified materials. This step is critical to the de-blocking reaction.

The most significant difference between CA and DNQ resists however is the environmental stability of both the coated photoresist film (post soft bake) and the latent image in the exposed resist (pre-develop). Because CA resists rely on small acid molecules to de-block the insoluble polymer, any chemical base contamination of the coated film will neutralize some or all of the acid near the surface of the exposed areas and the imaging results will be very poor (possibly no image at all). Airborne amines from process chemicals, construction materials, etc. are common in lithography environments and will act as bases to poison CA photoresists if their concentration is too high and/or if the delays between coat, soft bake and expose are excessive. Additives used in modern CA materials reduce their sensitivity to air born amines to do not eliminate it entirely. Special air filters or a “mini-environment” enclosing the lithography process chambers may be required. Where mini-environments or air filtering is not practical, A DUV Top Anti-Reflective Coating (AZ® Aquatar VIII for example) may be used as an environmental barrier against airborne amines.



After exposure, acids may tend to migrate into the unexposed areas of the photoresist film causing instability in the printed CD. This issue is most problematic for small isolated features surrounded by large areas of exposed resist (i.e. lots of acid). Migrating acid molecules may de-block some of the resist in the unexposed area at the exposed/unexposed boundary and cause shrinkage of the printed CD. This effect is commonly referred to as “line slimming” and is controlled by minimizing delays between expose, PEB and develop. Base additives used in modern CA resists work to quench acids at the exposed/unexposed boundary and are effective at minimizing line slimming but do not eliminate it entirely. 



Lastly, due to their high transparency and lower film thickness, CA resists can be much more sensitive to substrate reflectivity and thin film interference effects than DNQ’s. Because of this, the use of a BARC (Bottom Anti-Reflective Coating) layer is almost universal when the CA photoresist film being imaged is less than about 0.8µm thick.