Specific Process Knowledge/Etch/DRIE-Pegasus/processA

Process A

Process A is labelled Large trench (80μm wide) 150μm depth. In the acceptance test the process was run on a 150 mm SPTS wafer with 12-13 % etch load.

Process A specifications

Parameter	Specification	Average result
Etch rate (µm/min)	> 15	18.9
Etched depth (µm)	150	189.1
Scallop size (nm)	< 800	718
Profile (degs)	91 +/- 1	91.1
Selectivity to AZ photoresist	> 150	310
Undercut (µm)	<1.5	0.84
Uniformity (%)	< 3.5	3.0
Repeatability (%)	<4	0.43

The process developed by SPTS that fulfilled these criteria had the following parameters:

Process A recipe

	Step 1		Step 2
Parameter	Etch	Dep	Etch	Dep
Gas flow (sccm)	SF6 350 (1.5 s) 550	C4F8 200	SF6 350 (1.5 s) 550	C4F8 200
Cycle time (secs)	7.0	4.0	7.0	4.0
Pressure (mtorr)	25 (1.5 s) 90 >> 150	25	25 (1.5 s) 150	25
Coil power (W)	2800	2000	2800	2000
Platen power (W)	120 >> 140 (1.5) 45	0	140 (1.5) 45	0
Cycles	11 (keep fixed)		44 (vary this)
Common	Temperature 20 degs, HBC 10 torr, Short funnel, with baffle & 5mm spacers

Process A performance

The perfomance of Process A has been investigated as a function of feature size and etch load.

Experiment

A number of wafers are patterned with the travka masks in AZ photoresist or 600 nm oxide. The wafers are then etched (batch recipe with 5 minute TDESC interstep cleans) using two different durations of process A in the DRIE-Pegasus.

Wafers Wafer number Mask Mask material Process A duration C01548.01 Travka 05 AZ photoresist 55 cycles, 10:05 mins C01548.02 Travka 10 AZ photoresist 55 cycles, 10:05 mins C01548.03 Travka 20 AZ photoresist 55 cycles, 10:05 mins C01548.04 Travka 35 AZ photoresist 55 cycles, 10:05 mins C01548.05 Travka 50 AZ photoresist 55 cycles, 10:05 mins

Wafers Wafer number Mask Mask material Process A duration C01549.01 Travka 05 600 nm oxide 110 cycles, 20:10 mins C01549.02 Travka 10 600 nm oxide 110 cycles, 20:10 mins C01549.03 Travka 20 600 nm oxide 110 cycles, 20:10 mins C01549.04 Travka 35 600 nm oxide 110 cycles, 20:10 mins

Results: Optical images

• Optical images of the C01548 batch that is processed 10:05 mins.
• 
  Wafer C01548.01: 5 % exposed area
• 
  Wafer C01548.02: 10 % exposed area
• 
  Wafer C01548.03: 20 % exposed area
• 
  Wafer C01548.04: 35 % exposed area
• 
  Wafer C01548.05: 50 % exposed area
• File:Without pre-etch measurements of the thicknesses of the photoresist it is not possible to determine the resist etch rate. However, it looks very uniform.

• Optical images of the C01549 batch that is processed 20:10 mins.
• 
  Wafer C01549.01: 5 % exposed area. Some 190 nm of 600 nm oxide remains.
• 
  Wafer C01549.02: 10 % exposed area. Some 122 nm oxide remains.
• 
  Wafer C01549.03: 20 % exposed area. In the centre some 90 nm oxide remains.
• 
  Wafer C01549.04: 35 % exposed area. In the centre some 90 nm oxide remains.
• 
  Wafer C01549.05: 35 % exposed area (close-up). The oxide has disappeared leaving the Si exposed.

The etching of silicon releases energy. This means that the faster the etch is, the more heat needs to be dissipated. The consequence is the same if a larger percentage of the wafer is etched. Process A is the fastest etch and as seen above, the exposed area plays an important role. As the exposed area increases the oxide mask erosion is more and more pronounced.

If a larger area is to be etched, the cooling must be made more efficient, either by lowering the platen temperature or with increasing the pressure of He on the backside of the wafer. This will be investigated soon.

Results: Etched depths in trenches of different widths

• Graphs
• 
  Process time 10:05
• 
  Process time 20:10