Etch First Wafer Effect Budget
# Etch First Wafer Effect Budget
An etch first wafer effect budget accounts for the difference in etch behaviour between the opening wafers of a lot and the steady state reached once the chamber walls are conditioned, and bounds how far those early wafers may deviate. It treats the chamber start state as part of the process, not a detail that averages out.
The inner surfaces of a plasma etch reactor accumulate a conditioning layer during processing that shifts the plasma chemistry toward a steady state. When a lot starts from a cleaned or idle chamber, the first wafers run against walls that have not reached that state, so their etch rate, profile or uniformity can differ. This budget does not restate the chamber-seasoning budget, which qualifies a reconditioned chamber; it focuses on the transient the opening wafers experience.
## See the early wafers settle
Etch behaviour can drift across the first wafers of a lot and then settle into a steady band as the walls condition. The opening wafers see a different wall state, and therefore different neutral and ion populations, than the wafers that follow.
The sketch shows an etch-rate trend that is off-target for the first few wafers and then settles. The trend is schematic and carries no measured values; the settling shape depends on the chamber, the idle time and the chemistry.
Reported work ties this directly to the walls. A study of wafer-to-wafer reproducibility states that reproducibility during plasma etching often depends on the conditioning of the reactor's inner surfaces, and that gradual deposition of etch products on those surfaces can drift etch rate, profile, selectivity and uniformity. A controls study of chlorine etching of polysilicon named the classic first wafer effect and showed it could be eliminated by holding the plasma density constant despite the changing wall state.
## Attribute the effect to the wall state
The early-wafer deviation is a wall-condition transient, not a random excursion. Because it comes from the surfaces reaching a steady coverage, it recurs after events that reset the walls, such as a chamber clean or an idle period, rather than appearing arbitrarily.
Distinguish a first wafer effect from a gradual rate drift or a one-off defect. The first wafer effect is tied to the start of processing after a reset; attributing an opening-wafer deviation to steady-state drift, or the reverse, misdirects the correction.
## Budget the opening wafers explicitly
Define the chamber start state the budget assumes: the idle time, the last clean, and the wall condition at lot start. The same recipe can give a different first-wafer result from a freshly cleaned chamber than from one that has been running.
Decide how the opening wafers are protected. Options include conditioning or season wafers run before product, a warm-up sequence, or real-time control that holds a plasma parameter constant so product starts in band. Each has a cost in throughput or complexity and must be qualified for the process.
Verify the opening wafers, not only mid-lot. A lot-average or a measurement taken after the chamber has settled can pass while the first wafers sit outside the limit, so sampling must include the start of the lot.
## Tie the budget to the product requirement
The first wafer effect matters only through the product, so the budget closes on the result. Specify the acceptable deviation for the opening wafers, the start-state assumptions, the conditioning or control method, and confirm the early wafers meet the limit rather than assuming they match the steady state.
No single remedy is universal. More conditioning wafers cost throughput, and a control scheme that stabilises one parameter may not cover every affected output, so the approach needs qualification on the real chamber and product.
## Close the budget with traceable evidence
Specify the start state, the opening-wafer deviations, the conditioning or control method, and the measured early-wafer results against the limit, at a named process state. Retain idle time, clean history and recipe revision so a later lot start can be compared with the earlier one.
A first wafer effect budget bounds the opening-wafer transient tied to chamber wall conditioning, separate from steady-state behaviour. Start state, conditioning method and early-wafer result are separate questions, and each needs its own evidence.
## Sources and scope
- [Wafer-to-wafer reproducibility during plasma etching ... reactor wall conditioning (J. Vac. Sci. Technol. A, Univ. of Michigan)](https://cpseg.eecs.umich.edu/pub/articles/jvsta_26_498_2008.pdf): states wafer-to-wafer reproducibility often depends on conditioning of the reactor inner surfaces, with wall deposition drifting etch rate, profile, selectivity and uniformity.
- [Real-time control of the first wafer effect in Cl2 etching of poly-Si (Klimecky, Grizzle, Terry; J. Vac. Sci. Technol. A)](http://www.eecs.umich.edu/~grizzle/papers/Klimecky_JVSTA_wfigs.pdf): reports that controlling plasma density to a constant value eliminated the classic first wafer effect in chlorine etching of polysilicon.
Original diagrams and explanatory synthesis. No universal deviation limit, conditioning count or qualified control method is implied.