Page History

AZ 1512 test results (restricted access)

14 Feb 2020 

• In a first attempt on this test we discovered that the lights of the room are not appropriately filtered. Until the filters are replaced, we will keep the lights off while handling samples in the room. Lights can be turned on after loading sample and closing the window.

25 Feb 2020 

• In a second attempt, a sample was prepared according to the parameters above. Best dose and defocus was determined to be 120 mJ/cm² and +1, respectively. Defocus of +2 was similarly good, with no clear difference with respect to +1.
  • Even thought this dose was enough to open features down to 2 µm completely, it still shows signs of underexposure on the edges of stitching fields. We are still searching for the cause of these defects.

02 Mar 2020 

• After some more tests it was determined that 110 mJ/cm² with a defocus of +9 gives better results when developing for longer periods of time (90 s to 120 s).
  • This has to be further confirmed since the latests tests have also shown that both LEICA microscopes expose the resist while inspecting. This seems to be worsened by increasing the magnification, or at least the effect becomes concentrated enough to be easily noticed.

09 Mar 2020 

• Running another dose and defocus test we observed that the best parameters differ depending on total area exposed. Images are here.
  • For small area test the best parameters were dose = 110 mJ/cm² and defocus = +5.
  • For large area test the best parameters were dose = 90 mJ/cm² and defocus = +7.
• Gustavo de Oliveira Luiz sent a question to Heidelberg to determine if this is an expected behavior or if we should change something.
• The two testes were exposed in the same same, first the small area then the large area. All patterns were developed at the same time, after the large area had been exposed.
• Development standardized to be 90 s.

03 Apr 2020 

• Testing on a borofloat sample with Cr/Au (36/183 nm).
  • Defoc fixed at +5
  • Dose varied from 40 mJ/cm² to 150 mJ/cm² in step of 10 mJ/cm²
  • Best dose: 80 mJ/cm²

09 Apr 2020 

• Patterning Emil Varga's sample (quartz with Cr/Au) to dose had to be higher.
  • The films are thinner than on my test sample of 03 Apr 2020, clearly partially transparent.
  • 100 mJ/cm² seems to have worked fine.

27 Apr 2020 

• Test optical autofocus
  • Tested for both 375 nm and 405 nm lasers
  • 375 nm: best defoc = 0
  • 405 nm: best defoc = -2

01 May 2020 

• Tested autofocus:
  • pAF, 405 nm: best defoc = -3
  • pAF, 375 nm: best defoc = -1
  • oAF, 405 nm: best defoc = -3
  • pAF, 375 nm: best defoc = -1

17 Sept 2021 

• Tested dose for both HQ and Fast modes
  • HQ: Best dose found at 130 mJ/cm² (a ~20% increase compared to previous tests)
  • Fast: Best dose found at 150 mJ/cm² (not even close to the 2.5× suggested by Heidelberg)
• Tested Defoc @110mJ/cm² in HQ mode to make sure that the above results are not due to bad focus
  • There was no change from the previous nest focus value

10 Nov 2021 

• Dose test for HQ mode:
  • Best dose found at 100 mJ/cm²
  • At 130 mJ/cm², previous best dose (before HIMT service) still produces good results, but small isolated negative features (area around is exposed) vanished

AZ 1529 test results (restricted access)

25 Feb 2020 

10 Nov 2021 

• Dose test for HQ mode:
  • Best dose found at 290 mJ/cm²
  • Used design largeareacalib (10 mm × 10 mm) to test for bubbles when exposing @290 mJ/cm². No bubbles were observed on a 50.4 mm × 50.4 mm area.
  • Series with 25 copies in a 5 × 5 matrix with 10.1 mm pitch in both directions

AZ 3330-F test results (restricted access)

28 Jun 2021 

• Cut one wafer with each filme thickness in quarters to avoid double development (this caused issues before and we cannot afford it when doing dose tests).

  • 1500 rpm wafer
    • Test: 100:10:250 mJ/cm2
      • 250 mJ/cm2 was underdosed
  • 2000 rpm
    • Test: 100:10:250 mJ/cm2
      • 240 mJ/cm2 looks good
  • 3000 rpm
    • Test: 100:10:250 mJ/cm2
      • 200 mJ/cm2 looks good

  Defoc for all tests: -6.

  All exposed with 405 nm laser.

  Development: 2 min in AZ 400k 1:4, rinse with DI-water for ~30 s and dry with N2 gun.

02 Jul 2021 

Continue test on 1500 rpm and 2000 rpm to obtain best results

• 1500 rpm
  • Test: 200:20:350 mJ/cm2
    • 310 mJ/cm2 looks good

Upon further inspection of all tests above, it was determined that the focal position was not optimal. A test on a piece of mask showed that the nozzle had to be adjust (best defoc was -10). This could have been Gustavo de Oliveira Luiz mistake the last time it was adjusted.

The nozzle for pAF was adjusted and all recipes/files updated with new values. New best defoc for pAF is -1 for 405 nm and +1 for 375 nm.

• 1500 rpm
  • test 100:10:340 mJ/cm2
    • 340 mJ/cm² is underdosed
    • Could be old developer
  • Test: 160:10:400 mJ/cm²
    • 290 mJ/cm² looks good
    • Developer likely the cause of the previous result
• 2000 rpm:
  • Test: 160:10:400 mJ/cm²
    • 270 mJ/cm² looks good

All of the above still had a trapezoidal shape, such that the very top looks overdosed (wider openings) and the bottom looks lightly underdosed (narrower openings). Is this something we observe on AZ 1529? We should check.

Dose test on the 3000 rpm is pending (did not have time to finish).

27 Jul 2021 

Completed test with the film spun at 3000 rpm

• Test 100:10:250 mJ/cm² (defoc = -1)
  • 200 mJ/cm² looks best, albeit a little overdosed

20 Sept 2021 

• Did a dose test using design "calibchk_SU8", since it covers a larger area increasing the probability of seeing bubbles, on a 2 μm thermal oxide wafer
  • With single pass: best dose @290 mJ/cm², but some bubbles were observed
  • With double pass the same total dose was deposited but no bubbles were observedwas deposited but no bubbles were observed
    • We still need to make sure that this is due to exposure and not lack of baking

10 Nov 2021 

• Dose test for HQ mode:
  • Best dose found at 300 mJ/cm²
  • Used design largeareacalib (10 mm × 10 mm) to test for bubbles when exposing @300 mJ/cm². No bubbles were observed on a 50.4 mm × 50.4 mm area.
    • Series with 25 copies in a 5 × 5 matrix with 10.1 mm pitch in both directions
    • We still need to make sure that this is due to exposure and not lack of baking

LOR 5B / AZ 1512 bilayer test results (restricted access)

22 May 2020 

• Dose test with design "largeareacalib"
  • Dose range: 90:10:140 mJ/cm² (Defoc: -3)
  • Develop for 90 s with AZ developer 1:1
  • Hard to determine dose as both resists are developed and the LOR 5B may lift the AZ1512 off.
  • Not much undercut observed on the edges of the exposed areas.
  • Microscope inspection damaged resist, which made the MF-319 undercut etch useless.
• On a second sample another dose range was tested
  • Dose range: 80:10:130 mJ/cm² (Defoc: -3)
  • Develop on AZ Developer 1:1 for 90 s and MF-319 for 5 s, without inspecting in between
  • 110 mJ/cm² seems a little overdosed, similar to AZ1512 recipe, and 80 mJ/cm² is clearly underdosed

24 Jun 2020 

• Pattern sample for liftoff
• Recipe used (same as AZ1512, except developer):
  • Dose: 110 mJ/cm²
  • Defoc: -3
  • Develop: AZ Developer 1:1 for 90 s
  • Didn't follow with MF-319 for undercut
    • Tom Jones was reporting that the undercut was larger than anticipated, perhaps the first developer was causing some initial undercut.
  • Under optical microscope it seemed to have a small undercut, but hard to estimate the depth.

09 Jul 2020 

• After testing liftoff on the above sample it was determined that the undercut was probably too small
  • The liftoff worked, but the edges are too rough.
• Undercut process:
  • Tried 15 min in Remover PG with sonication → didn't work well.
  • Left sample in Remover PG overnight → worked with rough edges
• More tests will be done to determine a working recipe.

20 Jul 2020 

• Undercut tests (images here, here and here)
  • 5 s in pure MF-319: ~1.3 µm undercut
  • Diluted MF-319 in DI water (1:1): ~0.3 µm in 60 s
  • From the cross section SEM images, it's clear that for Al sputtering on Floyd (standard recipe) we need at least about 0.5 µm of undercut — this is how far under the resist the metal can reach.

AZ 5214-E test results (restricted access)

09 Mar 2020 

• Effect of hard bake temperature (100 °C and 110 °C) and time (1 min, 1.5 min, 2 min and 2.5 min) tested. It is possible to overbake or underbake! Optical microscope images here.
• Best results obtained with 50 mJ/cm² and defocus of +5. Hard bake for 1 min at 110 °C, flood expose for 60 s with either Bert or Grover (~60 mW/cm²). Develop in MF-319 for 30 s and stop process with DI-water. Dry with N₂  gun.
  • 2 µm lines opened nicely. Below that seems to be a matter of luck.
• Imaging with SEM will tell quality of undercut.

24 Jun 2020 

• Pattern sample for liftoff test
• Used recipe:
  • Dose: 50 mJ/cm²
  • Defoc: -3
  • Post-bake: 100 °C for 60 s ( WHY 100 )
  • Flood expose for 60 s
  • Develop: MF-319 for 30 s

09 Jul 2020 

• After testing the above recipe with liftoff of Al, it was determined that this recipe doesn't work properly. The undercut is not deep enough, requiring a longer bake time.
• Undercut process:
  • Tried 15 min in Acetone with sonication → didn't work well.
  • Left sample in Acetone overnight → didn't work
• More tests will be done to determine a working recipe.

20 Jul 2020 

• Patterned another sample with dose test and followed standard recipe for post-bake and flood exposure.
• Sample coated with Al for imaging and lift off test.

02 Oct 2020 

• Patterned 2 samples with "largeareacalib_grid":
  • Sample 1: Dose kept at 50 mJ/cm²; series defoc: -5:2:1.
  • Sample 2: defoc kept at -3; series dose: 30:10:60 mJ/cm².
• Both samples went to the same PEB process:
  • Bake at 120°C for 120 s.
  • Flood expose on Grover (Avg. power: 42.1 mW/cm² as of Aug. 11) for 30 s.
  • Develop on MF-319 for 60 s (+ DI water rinse + N2 dry).
• Images can be found here.
  • Best setting for this run seems to be 50 mJ/cm² with Defoc set to +1.
    • This resulted in approximately 100 nm of undercut.
    • However, 40 mJ/cm² is the dose that resulted on dimensions closest to nominal.
  • Dose seemed to make very little difference to the undercut.
  • Defoc seems to have some effect on undercut.
  • Run a test with 40 mJ/cm² and Defoc of +1 with longer development times.

04 Dec 2020 

• After obtaining information from Eric Milburn (he found this information on MIT's website) that we should use only 10-18 mJ/cm² to expose this resist on MLA150, this was tested.
  • Sample was coated and soft baked at 110°C for 50 s on 26 Aug 2020 (following datasheet recipe)
  • Dose test: 12:2:18 mJ/cm²
  • Defoc: 0 (forgot to change)
  • PEB: 120°C for 120 s (following datasheet)
  • Flood expose on Grover (posted intensity of 72.8 mW/cm²): 30 s
  • Develop on MF 319 for 30 s
    • Since the flood exposed part should be removed I don't expect this time to be different from the standard recipe using masks.
  • It looks good. Left with Tim for depositing Ti (~100 nm) using non-conformal e-beam evaporation.
• Tests on T-shape process
  • Flood exposed 4 samples
    1. 0.2 s
    2. 0.3 s
    3. 0.4 s
    4. 0.5s
  • Then baked all at 120°C for 120 s
  • Exposed as usual, using a lower dose, as suggested by Eric Milburn
    • Dose: 14 mJ/c² (mean of 10 mJ/cm² and 18 mJ/cm²)
    • Defoc: -3 
    • PEB: 120°C for 120 s (nothing visible after this step on any sample)
    • Flood expose on Grover for 30 s (nothing visible after this step on any sample)
    • Develop on MF 319
      • Sample 1 was developed for 30 s → A shadow of the pattern became visible, but it seems to be under a thin film
      • Sample 1 was developed for another 30 s → Same as before
      • Sample 1 was developed for another 60 s → Same as before.
      • Sampes 2 to 4 were all developed for 2 min → Not even a shadow visible.
    • It is clear that 0.2 s on Grover is too much for this process, let alone longer exposures.
    • Try again with short (0.1s to 0.3 s) exposures on Sunny. The slightly weaker lamp may help us out.

26 Jan 2021

• Tested exposure on 3 samples.
  • All three had the same dose array: 12:2:18 mJ/cm²
  • PE processes done following our standard recipe: 2 min PEB at 110°C + 60 s flood exposure on one of the ABM aligners
  • Each sample was developed at different times: 30 s, 60 s and 90 s
  • Samples coated with ~100 nm Ti and cut through the grid lines for cross section imaging with SEM. Images here.
  • It is clear that larger doses reduce the size of the undercut, as well as shorter development times.
  • Best result overall was at 14 mJ/cm² and 90 s development. However 12 mJ/cm² and 90 s development should be considered if larger undercuts are necessary (e.g. sputtering).
  • Images after lift-off (25-30 min sonication in acetone) are here.
    • Residue is probably resist that could be cleaned with O2 plasma or organic solvent (piranha can be used if the rest of the substrate is safe).
    • Perhaps an overnight soak in acetone could result in a cleaner process.
    • Dimensions don't always match the SEM images very well. This is probably because the samples were not always cut exactly in a straight line.

SU-8 2050 test results (restricted access)

07 Apr 2020 

• Developer 2 min + IPA for 30 s + N2 dry
  • Most of the resist was still in place
• Developer 1 min + IPA for 30 s + N2 dry
  • Some of the resist was still in place
• Developer 1 min + IPA for 30 s + N2 dry
  • Seems to be clear when inspecting by eye
  • Under optical microscope the structures are not well exposed. Even large areas seem to have suffered a lot of loss during development.
• Checking the resist template list saved on MLA150, there are two SU-8 recipes for films with 100 µm and 150 µm thickness.
  • The suggested dose is 5000 mJ/cm² for 150 µm and 8000 mJ/cm²  for 100 µm.
  • I have to study these recipes more carefully to understand why the higher dose was set for the thinner resist.
  • I'll run another test with a coarser dose range to cover these huge values.

20 May 2020 

• Created a new design called calibchk_SU8. Scaled up the original by a factor 2
• Exposed the new design in 2 series jobs
  1. Dose: 500:50:1500 mJ/cm²; Defoc = -1 (trying to minimize standard DoF effects )
  2. Dose: 1500:500:20000 mJ/cm²; Defoc = -1
• Followed PEB and development from standard recipe:
  • PEB at 65°C for 1 min followed by 95°C for 4 min
  • Develop in SU-8 Developer for 4 min (SU-8 is a high contrast resist, so dark erosion should be negligible).
  • Images can be found here.
  • Best dose seems to be 600 mJ/cm².
• Next test will verify best dose for Large and Extra Large DoF.

AZ-P4620 test results (restricted access)

15 Mar 2021 

• Dose test @ 405 nm
  • Best dose between 500 mJ/cm² and 550 mJ/cm²
  • Used Defoc = -3
• Dose test @375 nm
  • Best dose between 850 mJ/cm² and 900 mJ/cm²
• Tried refining the dose test with 10 mJ/cm2 between the best before.
  • Had some problems with development. Apparently the developer is turning bad after a single sample is processed

18 Mar 2021 

• Tested some more to make sure if the developer was the problem
  • It was. A fresh bath was able to develop properly after the first one failed 
  • Try using fresh developer for each sample
• Tested defoc series to determine if "slanted" exposure was caused by wrong focus
  • It didn't make any difference when exposing between ±10

23 Mar 2021 

• Refine dose test with defoc of -3 and @ 405 nm
  • Dose range: 450:10:500 mJ/cm2
  • Best: 560 mJ/cm2 (best to clearance after 2 min development, but there is a ~1µm bias)
• Defoc test:
  • Range: -10:1:10
  • Best: -3 (as expected)

Photomasks

• Substrate:  5" × 5" or 7" × 7" photomask blank from Nanofilm
  • 0.090" thick soda-lime photomask blank with 100 nm low reflective chrome and 0.530 µm AZ 1518 resist
• Job conversion:
  • Use high-quality mode
  • If orientation is RRCD, mirror design (at x or y); if RRCU, do not mirror.
  • Invert polarity if specified in photomask submission.