Technical Blog
Deep dive into the physics and integration logic of semiconductor manufacturing
SiO2 Bonding: Counting Shared Oxygen in a Solid Network
In an ordinary solid silicon-dioxide film, SiO2 describes composition, rather than a collection of isolated O=Si=O molecules.
Die Yield, Defect Density and Critical Area: Read the Model
Use a dimensionless Poisson example to distinguish defect counts, critical area and die yield, and identify when the assumptions fail.
How CMOS Image Sensors Work: From Light to Pixel Readout
A CMOS image sensor converts absorbed light into charge, stores that charge during exposure and reads an electrical signal from each pixel.
Semiconductor Manufacturing Process: From Devices to Interconnect
Semiconductor manufacturing builds devices and their connections on a wafer through repeated cycles of adding material, defining patterns, removing…
Why Post-Bond Annealing Cannot Be Judged Without the Earlier Copper State
Post-bond annealing acts on the copper and interfaces inherited from earlier processing.
Why Replacement Metal Gate Starts with a Dummy Gate
Understand how a dummy gate preserves geometry and what must remain intact when the final metal gate replaces it.
Why STI Can Be Recessed After Filling and CMP
See why STI filling, CMP and later recess can serve different geometric goals in a device integration sequence.
Why expose a 7nm gate contact twice?
LELE divides a target pattern between two lithography-and-etch passes. Each exposure handles a less crowded subset, but overlay and edge placement matter when the subsets are combined. The CON/CB gate-contact sequence in this 7nm Flow contains E1 and E2 imaging and memorization-layer transfer, even though its step names do not spell out LELE.
Color filters, clear layers and microlenses: three different jobs in a BSI pixel
Color filters select the light reaching a pixel, a clear base prepares the surface above the filter array, and a microlens redirects light toward the sensing region. Reading these structures as three interchangeable “sensitivity layers” misses why the process needs all three.
Why conformal coverage is not complete gap fill: liner versus FCVD
Conformal deposition describes coverage along an existing surface. Gap fill asks whether material occupies the interior volume. Films can grow uniformly on both walls yet meet at the entrance before the interior is filled. FCVD uses a mobile depositing network to improve filling, while introducing its own conversion and material-stability requirements. A conformal liner remains useful because it has a different job in the same trench.
Hardmask vs Etch-Stop Layer: Opening vs Depth
Positioned above the target, a hardmask sets lateral openings while a selective stop layer below bounds depth. Shared SiN never implies one job.
Photodiode, transfer gate and FD: from charge to readout
A pixel voltage is the result of charge generation, collection, storage, transfer, and conversion. Calling every n-type region photosensitive confuses the photodiode with its readout node. Treating the surface p-type pinning region as extra storage volume also misses its purpose.