In our What is an Electronic Grade Chemical? post, we discussed some of the quality aspects of electronic grade chemicals: metals content, particle content to name but two. In this post we will discuss how some chemicals can have those carefully obtained particle counts ruined by storage time and temperature. Because this post will focus on photoresists, we will discuss the particular way a photoresist can age with respect to particles but also with respect to a performance criteria that are critical to how the photoresist is designed to function.
Photoresists are used to transfer a master pattern onto a wafer (or similar substrate). There are two key aspects of photoresists that can make or break a semiconductor process: they must achieve certain performance qualities (e.g. photo-speed, coated thickness, contrast or wall angle, and particle count) and even more to the point these properties must be as close to identical, resist batch after batch throughout the life of each bottle of resist.
There are in general four things that happen to a positive tone photo-resist as it ages: coated thickness increases, photo-speed increases, wall angles decrease, and particle count increases. The thickness increases can come from simple evaporation of the photoresist casting solvent, directly leading to a thicker film thickness. (To some degree thermally-induced polymer cross-linking can also do this.) This is straightforward and really doesn’t require further discussion. But the other aging changes derive from a curious aspect of photoresists: the same desired photo-reaction (referred to as the Wolff Rearrangement) that takes place during a normal Production process can also occur as a simple thermal reaction. That is to say, the random banging about that molecules do at some finite rate (as a function of concentration and temperature) can cause the components of a resist formulation to react in the absence of light – the desired trigger used in the normal Production process!
Let’s review for a moment a simple g-line or i-line resist. The standard positive mode g- or i-line resist consists of a novolac resin, a photoactive component (or PAC), some minor additives to improve coating, and a casting solvent. Prior to exposure to light, the PAC interacts with the resin in a manner that creates a neutral complex; such a complex being non-acidic prevents it from dissolving into the alkaline developer that the resist film will be immersed in during the develop step (that comes after exposing the resist film to light shown through the master pattern referred to above). The PAC will change dramatically after exposure to light of the proper wavelength during the exposure step; the PAC/resin complex collapses resulting in an acidic resin and acidic by-products of the rearranged PAC. All components can now dissolve in the developer.
Those areas of the resist thus exposed will no longer prevent the resin from dissolving into the developer. The key point for this post is that the PAC will also thermally convert at a certain rate for a given temperature. That means, storing your resist at a certain temperature will after a given amount of time be the same thing as exposing your resist – only will do so without the benefit of using the desired master pattern. It will be the same as a blanket exposure.
However, thermal exposure is not the same as a blanket exposure at the Production exposure dose. Instead it will be some fractional portion of that dose. The end result will be that for a certain time and temperature, partial conversion will have occurred. The following graph shows how the photo-speed changes as a function of time and temperature.
Photoresist Photo-speed Change as a Function of Time and Temperature
Thus, if the Production dose for the above process is 100 mJ/cm2 and if the specification for this process is +/- 2.5% variation from normal, you can readily see that this resist when stored at 5oC will be within spec for up to nine months (i.e. shelf life); but only for five months when stored at 20oC. That means, the manufacturer that has stored this resist in the clean room (i.e. 20oC) may have to change out the resist bottle while there is remaining material still in the bottle (if his usage rate takes more than 4.5 months) – clearly this is not economical. Just as clearly, there are compelling reasons for keeping the resist in cool storage as long as possible prior to taking it into the clean room to be used in the manufacturing process.
(Note: the recommended storage temperature will vary with resist and manufacturer.)
However, the “litho” changes that take place over time can be more subtle than an exposure dose change. There will be changes in the contrast of the resist resulting in a degradation of the imaged resist wall profile; increases in unexposed film loss, reduced process latitude and worsened resolution. There will be increased particle counts (see the following chart) that will reduce even more the already degraded resist performance. The bottom-line is each manufacturer needs to carefully consider how they store their resists prior to use and how their supplier stores and ships the resist prior to taking possession of that resist. The resist manufacturer will supply via the certificate of analysis that comes with each batch of resist the expected shelf life (SL) for that resist. Use that SL information plus the temperature tracking information during shipment, and finally how you store your resist on-site as you consider how to maximize your resist’s performance and reduce your costs.
Photoresist Particle Count Change as a Function of Time and Temperature.