e-beam bilayer PMGI / PMMA

e-beam bilayer PMGI / PMMAEdit

Outline: I. Goals    II. Typical parameters    III. Precise receipe example    IV. Notes and Issues

I. Goals

Control undercut with two orthogonal post-exposure development solvents.
Increase undercut speed by UV flood of PMGI
Avoid craks in PMMA mask by using thick PMMA
Avoid charging effect by evaporating 7 nm of Al

II. Typical parameters

Deposition:

• PMGI-SF8 between 2500 and 4500 rpm  x 60" + bake 150-200°C (max 250°C - Tg=190°C)
  2s @ 500rpm + 60s @ 2500 rpm + 2s @ 8000rpm  + bake 5min @ 190 °C => 670 nm
  or
  2s @ 500rpm + 60s @ 1000 rpm + 2s @ 8000rpm  + bake 3min @ 170 °C => 1040 nm
  2s @ 500rpm + 60s @ 2500 rpm + 2s @ 8000rpm  + bake 3min @ 170 °C => 666 nm
• optional: UV flood the PMGI to
• PMMA 950k A6,  6500 tr/min 60 " + bake 150°C 10 min => 390-420 nm (less brittle but less resistant to Al etchants)
  PMMA 950k A6,  6500 tr/min 60 " + bake 170°C 10 min => 390-420 nm (more brittle but more resistant to Al etchants)
  PMMA 950k A3,  4000 tr/min 60 " + bake 150°C 5-10 min => 110 nm
• optional: 7 nm Al evaporated

e-Beam patterning:
Dose@ 25 kV: 250 µC/cm2 x 1.3 (for 660 nm PMGI + 400 nm PMMA)
Dose@ 10 kV: ?? µC/cm2

Development:

• Al removal : XOH or MF319 or commercial Al etch (See note below about exposed PMMA etching rate)
• Developer PMMA: MIBK 1 Vol + IP 3 Vols
• Developer PMGI: Nano PMGI 101 developper OR 60% solution of CD26 in water.

Lift-off:
PMGI remover ultrasonic bath 60°C 10 min => 30 nm/min
Resist removal in asher 20 ccO2 - 0.2mbar - 300 W => 40nm/min

III. Precise example with thick PMMA, UV flood and Al (KJ + DV 12/2013)

• Si wafer
• Measure SiO2 with white light interferometer =>500 nm
• Deshydrate 1min @ 150°C, cooldown
• PMGI SF8 2s @ 500rpm + 60s @ 2500 rpm + 2s @ 8000rpm  + bake 5min @ 190 °C => interferometer : 670 nm
• UV flood 7000s x 4.5 mW/cm2 in MJB3 broadband UV (documentation recommends 1-5 J of deep UV, unavailable in our lab)
  
• PMMA 950k A6,  2s @ 500rpm +6500 tr/min 60 " + 2s @ 8000rpm + bake 170°C 10 min => interferometer : 400 nm
• Al evaporation 7nm @ 1nm/s.
• Scribing
• e-beam exposure 25 kV x 250 µC/cm2 (x 1.0-1.3 depending on feature size)
• Al etch in old KOH 200g/L x 15 s (7nm in 7s + 7s overetch) / 2s rinse in water 1 / 2s rinse in water 2 /15s rinse in IPA / dry.
• optional: check in microscope => PMMA is slightly developped by KOH
• PMMA development in MIBK (3)-IPA(1) during 50s / 15s rinse in IPA / dry
• optional: check in microscope => PMMA fully developped down to PMGI. PMGI not impacted
• PMGI development and undercut in PMGI developer 101 / rinse water /rinse IPA
• observe and measure undercut in the microscope (x 250) every 15s and stop when satified => 0.8 µm undercut in 45s with relative dose 1.3

Optical microscopy at x250 showing the mask at different development steps: after Al removal , PMMA development, 15s PMGI development, and 45s PMGI development (from left to right).

IV. Precise example with 250 nm PMMA (HLS+ SS 01/2014)

• Si wafer doped 10Ω.cm, SiO2 500nm
• Bilayer spin:
  

  spin PMGI SF8 @ 3000rpm 60" -> 613 +/-15 nm (cleavage and angle observation, See Preliminary tests)
  bake hot plate 170°C, 5'
  spin PMMA A6 @ 6000rpm 60" -> 253 +/- 21nm (cleavage and angle observation, depends on position on the wafer)
  bake hot plate 170°C, 10'
  spin UV III @ 4000rpm 60" (protection for dicing)
  bake hot plate 140°C, 60"

• Exposure: 300µC/cm2 (polygon) or 1200pC/cm (path -> equivalent dose of a 40nm polygon) @ 30keV in XL30 (WD=17mm). 
  Correct dose factor:
  x1.1 for large patterns,
  x1.2 for patterns in the range 70nm - 1µm,
  x1.3 and x1.4 for patterns of 50nm (single path or two path spaced by 20nm - not polygon)
• Development
  

  MIBK+IPA (1+3) 1'30
  rinse IPA 30"
  no dry
  rinse ODI 15"
  MIF726 or MFCD26 1'
  rinse ODI 30"
  no dry
  rinse ethanol 15"
  very very gently dry N2 (should take several seconds)
  

  Note the relevance of the last two steps was not proven. It was first intended to avoid PMMA mask deformation of very thin high aspect ratio bridges (80nm wide x 3µm long, see pictures below), but the problem was solved by widening the lateral openings so that the bridge cannot stick to an adjacent wall.
  
• PRELIMINARY TESTS:
  Exposure of lines on a whole x2000 field, development, cleavage, sputter 15" Au/W, angle observation (75°)
  - PMGI thickness: 613 +/- 15nm
  - PMMA thickness: 253 +/- 21nm
  - undercut vs. dose for various designed linewidth (different colors), and 1' MF-CD26 (plain lines) or 30" MF-CD26 (dashed) at room temp.
  
  - comparison mask opening / double angle evaporation Au 30nm +/-22°
  
• Evaporation (vieux canon)
  Small junctions design and very high aspect ratio PMMA bridges
  

  ion mill 500V, 3mA, 2x10" @ +/- 22°
  Al 30nm @ 1nm/s (Pchamber = 4.3e-7mb, Psas = 8e-7mb) @ +22°
  Dynamic oxidation, 100µb (8e-2 Torr), 5' 
  Al 60nm @ 1nm/s (Pchamber = 4.3e-7mb, Psas = 9e-7mb) @ -22°
  

•  

V. Notes and Issues

1) PMMA dissolution in Al etchants
Unexposed PMMA does not dissolves but exposed one does !
The etching rate is lower with KOH than with commercial acid Al etch.
The etched depth is much larger when an Al discharge layer covers PMMA.

2) PMGI SF8 dissolution in PMGI developer 101
PMGI dissolution rates depends dramatically on the exposure to electrons and deep UV:

3) Cracks on PMMA mask
The PMMA mask tends to crack when bridges narrower than 400 nm are patterned. Using 400 nm thick PMMA seem to solve the problem, which could reappear with narrower bridges.

4) Angle evaporation through thick PMMA mask
The width of a pattern is narrowed by a quantity between 0 and thickness x Tan(angle) because the PMMA mask walls are not perpendicular to the mask surface. So geometry has to be determine by trial and error.

5) Problems on saphire subrate
Many irreproducible problems have been observed with this process when using a saphire (Al2O3) substrate instead of Si: fast etching rate of PMMA in KOH, cracks in PMMA before PMGI dissolution, PMGI flow below the PMMA mask when immersed in PMGI developer, etc
We attribute them to uncorrect bake of the resists on saphire and are working to solve them.