Planar Process 1959 Etch Contact Windows Field Oxide Intact

# Etch Only the Contact Windows, Leaving the Field Oxide Intact Everywhere Else: The Same Acid, a Much Smaller Target

## 1. Why the Old Undercut Math Now Has to Share the Margin With Something New

This step immerses the wafer in buffered hydrofluoric acid — the identical etch chemistry the 1957 series used to open its own diffusion windows at Step 10 — and removes oxide only inside the small contact windows Step 3 defined, leaving the field oxide standing everywhere else on the wafer, including all the way out past the covered junction edge Step 2 identified. The etch is isotropic exactly as before, undercutting the resist edge by roughly one oxide thickness on each side, and that undercut was already a known cost the 1957 series budgeted for against its own window-width tolerance. What is new here is that this undercut now has to share a design margin with a second requirement Step 3 introduced — staying clear of the covered junction — so the budget this step has to satisfy stacks both terms together rather than accounting for either alone:

$$w_{\text{window, final}} = w_{\text{contact}} + 2x_{\text{ox}} < w_{\text{diffused region}} - 2y_{\text{lateral}}$$

where $w_{\text{contact}}$ is the resist-defined window width from Step 3, $x_{\text{ox}}$ the oxide thickness, and $y_{\text{lateral}}$ the covered-junction margin from Step 2. The etch undercut and the junction-coverage margin are now competing for the same physical space on the wafer surface, which means a thicker oxide — still desirable for a more reliable diffusion mask back in Step 1 — now also costs more of the margin this step depends on to stay clear of the junction, a trade-off this series inherits directly from 1957's Step 10 but applies to a target this series alone needs protected.

Two Margins, Now Competing for the Same Space etch undercut and junction-coverage margin, stacked in one budget TOP VIEW, ONE CONTACT WINDOW diffused region resist window final etched window, wider by 2xox remaining margin remaining margin wwindow = wcontact + 2xox < wdiffused − 2ylateral — undercut now eats into the junction's own safety margin a thicker oxide helps step 1's diffusion mask and costs this step's margin, at the same time

## 2. Real Diagram: Field Oxide, Untouched, Standing Guard the Whole Way Out

Away from the small contact window, this etch does nothing at all — the field oxide over the covered junction edge and everywhere beyond it never sees the acid. The only oxide this step removes is the oxide this series never needed to protect anything in the first place.

One Small Opening, Everything Else Untouched cross-section after this etch, field oxide intact out past the covered junction covered junction covered junction field oxide, never attacked by this etch, standing all the way to the wafer's edge the smaller the window this series dares open, the more oxide stays protecting the junction

## 3. Why the Smallest Window in This Project's History Is This Series' Biggest Advantage

Every earlier masking window this project has documented — in 1954, 1956, 1957, and the 1958 production line — was sized to admit a dopant or define a device boundary, and larger windows there were often acceptable or even necessary. This step inverts that incentive completely: because this window exists only to let a metal contact through, this series wants it as small as reliably possible, since every bit of unneeded window width is oxide that did not need to be removed and margin that did not need to be spent. No earlier series had a reason to minimize a window purely to maximize how much oxide stayed in place; this one does, because the oxide staying in place is the entire point of the series.

Step 4 does not open the smallest window this project could etch for its own sake; it opens the smallest window this series can get away with, because every square micron this window does not cover is a square micron the covered junction gets to keep.

Take planar process 1959 etch contact windows field oxide intact further

Ask the copilot about this term, or have our engineers assess it against your process.

Ask CFSGPT about this → Request a technical assessment → Open the CFS app