Specific Process Knowledge/Etch/DRIE-Pegasus/Pegasus-4/Silicon Nitride Etch

Feedback to this page: click here

THIS PAGE IS UNDER CONSTRUCTION

Silicon Nitride Recipe

Parameter	Recipe name: Slow Etch	Recipe name: Slow Etch2
Coil Power [W]	350	200
Platen Power [W]	25	50
Platen temperature [oC]	20	20
H2 flow [sccm]	15	15
CF4 flow [sccm]	30	30
Pressure [mTorr]	3	10

Etch rate & uniformity

Recipe: Slowetch 2

SRN (LPCVD) etch uniformity on 6″ wafer

Process: No mask, 100 % load, etching was done after 10 mins chamber clean + 5 mins TDESC clean (for all wafers)

Note: etch rate for SRN with Slowetch 2. The 4 wafers were processed on 4 different days, the variation suggests process instability or chamber memory effects influencing etch behavior across runs.

Note: Compared to the first dataset (with chamber cleaning), the second set shows higher variability in etch performance despite identical recipes. Wafer 5, without conditioning, had the highest etch rate (11.5 nm/min) but poor uniformity, while Wafer 6 showed lower etch rate with improved uniformity. This confirms chamber conditioning significantly stabilizes SRN etch rates and enhances wafer-to-wafer uniformity.

Recipe: Slowetch

SRN (LPCVD) etch uniformity on 6″ wafer

Process: No mask, 100 % load, etching was done after 10 mins chamber clean + 5 mins TDESC clean (for all wafers)
Note: The slowetch recipe has faster etch recipe significantly improves etch rate (~30 nm/min) but leads to reduced uniformity (~2.5%) compared to the slower recipes (<1% uniformity). This highlights a trade-off between throughput and etch uniformity in SRN processing.

With DUV mask on ~300 nm SRN layer

Process: Slowetch recipe for 10 mins, DUV mask (with Barc) ~750 nm, etching was done after 10 mins chamber clean + 5 mins TDESC clean (for all wafers)
Note: The SEM images illustrate the etched SRN profiles across the wafer, progressing from one edge (left) through the center chip (middle) to the opposite edge (right), the etch rate is ~28 nm/min.

*The process was repeated twice, but only one set of data is presented here.