Actual etch rate test
Reactive Sputtering: Physical Vapour Deposition of TiN
It was noticed that the type of material deposited could be discerned more confidently by viewing the colour of the plasma surrounding the Titanium target within the chamber (at higher N2 flow rates more pink, as well as at lower DC Powers more white). Pure Argon or low N2 flow rates produced blue plasmas, signifying Titanium deposition - which is also intermittently necessary for cleaning the target to prevent target poisoning.
The calculated deposition rates for the first tranche of depositions was deemed to unrealible after noticing the substrate shutter does not fully close. For accuracy timing the entire run or completing the deposition in one shot might be necessary in future. Alternatively new equipment could be used (also capable of supporting the desired substrate temperature 600ºC).
Barring that, here is a table summary of the settings used:
--
Wet etching: For tailoring thicknesses of TiN base layers or patterning top layers.
To measure etching rates two experimental methods were attempted. Initially three samples of various thicknesses were prepared via reactive sputtering
The first sample did not visibly respond after 10 minutes of agitated exposure to 30% wt. Hydrogen Peroxide heated to 50ºC (and 60ºC in the latter part of the experiment).
At 52±3nm the intention was to remove the sample from the etchant every 60 seconds - rinse in DI and Nitrogen blow-dry - and visibly check when the material had been fully removed, to get an upper limit estimate on etch rate for each.
Following this the 70nm (clear) sample was omitted and different (red) sample not original intended for experiment was used simply to check whether any etching could be achieved at all, which fortunately was the case.
A sample originally measuring 127±5nm (green) was etched for 15 minutes and will be measured using Atomic Force Microscopy (a duplicate from the same batch was retained for comparison including of roughness/uniformity).
The second method involved first spin-coating a protective layer of photoresist, baking for 60 seconds, and then removing the resist from regioned to be etched (located next to regions covered with kapton during depositon). The original intention had been to create 'steps' to represent 60 seconds worth of etching, for indication of any etch rate differences over the full time.
Another (red) sample, this one confidently measured to be 228±4nm in thickness, was etched for 14 minutes. The intention was for 15 minutes, however after around 12 minutes it was clear that the photoresist used (Ma-N 405) was beginning to react with the H2O2. This reacted photoresist material subsequently became too difficult to remove with the commercially recommended chemical.
Using the Bruker Profiler this etched region of the sample was observed to have similar, if not better, roughness than before (±3nm) . Although this would be confirmed and enumerated using AFM. The thickness could be estimated to be 122.5 ± 10.6 nm (or 0.91 uniformity Figure of Merit). This indicates an etch rate of 7nm/minute ± 12%. For reference 40nm/min was reported for others using a similar method [doi: 10.1063/5.0179924 , https://doi.org/10.1557/PROC-477-447].