Al2O3 etching to make contact

Al2O3 etching to make contactEdit

Old Plassys

Etch parameters: Vbeam = 532V, Vacc = -32V, Ibeam = 5mA

- first etch test, the wafer is evaporated with 100nm, no oxidized, then etched.
The center circle appears at ~9min
No picture taken, so it is difficult to compare with next results, but we can infer a etch rate of Al > 8nm/s for sure
(technically it is here 11nm, but the time recorded is approx, and also, maybe it is a heat activated process)

- second etch: wafer evaporated with 15nm (+1 or 2), oxidized over night
Center circle appears at 6min30 (some modification already visible at 5min)

- third etch: wafer evaporated with 50nm, oxidized over night
Center circle appears at 10min

The time determination is very difficult (because the contrast is not the same on the 2 wafers)
So we will only put boundaries. Also we impose the rate of Al2O3 to be 0.13 the rate of Al (we believe the tables at 500V)

Solving :
Reduce[tAl1 + tAlO == 52 && tAl1/RAl + tAlO/RAlO == 10 && tAl2 + tAlO == 16 && tAl2/RAl + tAlO/RAlO ==6.5 && RAl > 7 && RAlO == 0.13 RAl && RAlO > 0 && tAlO > 0 && tAl1 > 0 && tAl2 > 0, {RAlO, tAlO, RAl}]

Leads to :
RAlO == 1.33714 && tAlO == 7.59934 && RAl == 10.2857
-> 7.6/1.34 = 5min40 to remove Al2O3

Solving :
Reduce[tAl1 + tAlO == 52 && tAl1/RAl + tAlO/RAlO == 9.5 && tAl2 + tAlO == 16 && tAl2/RAl + tAlO/RAlO ==6 && RAl > 7 && RAlO == 0.13 RAl && RAlO > 0 && tAlO > 0 && tAl1 > 0 && tAl2 > 0, {RAlO, tAlO, RAl}]

Leads to :
RAlO == 1.33714 && tAlO == 6.83087 && RAl == 10.2857
-> 6.83/1.34 = 5min06 to remove Al2O3

Test performed on WJOX1_6 shows that 5'30 is quite well to take a contact on oxidized Al layer, if sample is positionned at the center of the ion beam

New Plassys

Etch parameters: Vbeam = 415V, Vacc = -107V, Ibeam = 127mA

Pictures are much more difficult to take so showing a film is not possible
The homogeneity is much better than in MEB350 (below, 100nm oxidized wafer after 3' etch at 130mA 500V, planar rotation 16°/s, diameter is 3cm)

 

Determination of etch rates:

• First test 50nm:

- evaporate 50nm  on WALOX3 and let it oxidize for a few hours
- meanwhile pump WALOX4, evap 50nm and etch right away
- then etch WALOX3

results:
- oxidized wafer WALOX3: etch steps of 15".
1'15 slight change of color, 1'30 completely gone through layer -> keep 1'20
After 2' etching, the diameter is 3.5cm

- unoxidized wafer WALOX4: steps of 30".
Was the first wafer etched so the procedure of picture taking was not operative.
after 1'30 the diameter is the same as WALOX3 1'30. After 2'30 the circle diameter is ~3.5cm, at 3' it is 4.5cm
-> keep between 1'15 and 1'20 => rate Al = 40nm/min

• same steps with 100nm:

- first evaporate film WALOX5, let it oxidize
- second evaporate film WALOX6, etch it right away
- third, in a row, etch WALOX5 ->

results:
- WALOX5 (night ox): etch steps of 15".
circle appears between 2'15 (nothing) and 2'30 -> keep 2'30

- WALOX6 (no ox): etch steps of 15".
circle appears between 2' (nothing) and 2'15 (1cm diameter)-> keep 2'10 => rate Al = 46nm/min


I get the same behavior as before: the oxidized film is etched faster. I suspect that the way I etch it is responsible. Indeed, in both cases:

-for both unoxidized wafers I pumped loadlock through chamber, waited down to 1e-7mb, evaporated, closed the valve, then etched
- for both oxidized wafers I did not pump the loadlock, and only waited for the process parameter i.e. P_ll < 8e-6mb

Solving for :
Reduce[tAl1 + tAlO == 51 && tAl1/RAl + tAlO/RAlO == 1.33 &&  tAl2 + tAlO == 101 && tAl2/RAl + tAlO/RAlO == 2.5 && RAl > 0 &&  RAlO > 0 && RAlO == 0.5 RAl && tAlO > 0 && tAl1 > 0 &&  tAl2 > 0, {RAlO, tAlO, RAl}]

Yields:
RAlO == 21.3675 && tAlO == 5.83761 && RAl == 42.735

Solving for :
Reduce[tAl1 + tAlO == 51 && tAl1/RAl + tAlO/RAlO == 1.33 &&  tAl2 + tAlO == 101 && tAl2/RAl + tAlO/RAlO == 2.5 && RAl > 0 &&  RAlO > 0 && RAlO == 0.13 RAl && tAlO > 0 && tAl1 > 0 &&  tAl2 > 0, {RAlO, tAlO, RAl}]

Yields:
RAlO == 5.55556 && tAlO == 0.872286 && RAl == 42.735

Changing the rates ratio in the solve does not change Al rate, which is constrained by the time difference between the 2 layers only (1'20 for 50nm, 2'30 for 100nm), if we assume the Al2O3 thickness is the same in both cases.
42nm/min is consistent with the bare etch rate, so I guess this hypothesis is valid.
So it seems the Al2O3 etch rate is greatly enhanced when pumping is coarse. The tabulated Al2O3 etch rate (=0.13 Al etch rate) gives completely inconsistent Al2O3 thickness.

The tabulated value should give 5.5nm/min, so 1min20 (7.6nm, from previous tests) etching should be OK to recover contact.

=> I have to perform the test again with well enough pumping.

• Walox 7, 50nm 1st part oxidized 3h:

Measure step height:
 

• Walox 8, 30nm 1st part oxidized a few hours (to check in notebook):

• Walox 8, 30nm 2nd part oxidized over the week end (to check in notebook):

2nd part is slightly longer to etch, maybe 5"-10" difference