Photolithography Oxide Masking 1957 Soft Bake Photoresist
# Soft-Bake the Photoresist: Driving Off Solvent Without Driving Off Sensitivity
## 1. Why Two Different Reactions Share the Same Oven and Disagree About the Temperature
This step places the freshly spun wafer on a hotplate well below the temperatures any later furnace step will use, and holds it there just long enough to drive most of the casting solvent out of the resist film — firming it, improving its grip on the oxide beneath it, and stabilizing the thickness Step 3 worked to control. The difficulty is that solvent evaporation is not the only thing happening to this film as it warms. The same heat that drives off solvent also, more slowly, begins to decompose the resist's light-sensitive compound — the one chemical property the next two steps depend on entirely — and the two processes do not share a convenient temperature where one finishes cleanly before the other starts:
where $C_s$ is residual solvent concentration, $k_s$ and $k_d$ are rate constants, and $E_s$ and $E_d$ the activation energies of solvent loss and of photoactive decomposition respectively, with $E_s$ comfortably below $E_d$. That gap between the two activation energies is the entire reason a soft-bake window exists at all: at a temperature high enough to clear solvent at a useful rate, decomposition is still slow enough to be negligible — but the margin is narrower than it looks, and it closes from both directions at once.
## 2. Real Diagram: One Bad Wafer, One Ruined Mask, Many Ruined Wafers After It
An under-baked film is tacky rather than firm. When Step 5 presses a photomask into near-contact with it, the mask lifts resist away with it on separation — destroying the pattern on that wafer, and leaving resist residue on the mask itself. Every wafer that mask touches afterward inherits that contamination as a printed defect, so the damage from one under-baked wafer does not stop at one wafer.
## 3. Why This Step Has No Equivalent Anywhere Earlier in the Project
Neither the 1954 diffused-base process nor the 1956 drift-transistor process used a true photoresist, so neither ever needed a step whose entire purpose was to thermally condition a temporary auxiliary material before using it. Every furnace cycle in those series acted directly on the device — growing an oxide, driving in a dopant, annealing a contact. This bake does none of that. It prepares a tool. That is a genuine marker of how much more process machinery 1957 requires to get the same silicon to the same destination, and it will not be the last step of its kind in this series; by the time this account reaches Step 8's hard-bake, two of the first eight steps will exist purely to manage a material that gets discarded before the transistor is finished.
Step 4 does not change anything about the device waiting beneath this film; it decides whether the film survives contact with a mask it has not met yet, and whether it still remembers how to respond to light when that contact is over.