Revision as of 11:31, 31 January 2023

Feedback to this page: click here

THIS PAGE IS UNDER CONSTRUCTION

By Maria Farinha @Nanolab Internal, Jan 2023

Al₂O₃ mask

In these samples, the hard mask, Al₂O₃, was coated in the 150 mm Si wafers using the ALD 1 tool. These sets of samples also had the lithography process done in the DUV stepper to deposit 65 nm of BARC and 750 nm of DUV resist. Three different patterns were created after the development:

Pillars of 1 µm with 1 µm pitch (50 nm Al2O3 + 65 nm BARC + 750 nm DUV).
Holes of 1 µm with 1 µm pitch (50 nm Al2O3 + 65 nm BARC + 750 nm DUV).
Nanoholes of 200 nm with 400 nm pitch (100 nm Al2O3 + 65 nm BARC + 750 nm DUV).

The Si/Al₂O₃/BARC/DUV stack was first processed in Pegasus 2 to etch the BARC layer, and the alumina was etched in the III-V ICP, using a BCl₃/Ar recipe. With pillars and holes, the BARC was etched for 12 min on Pegasus 2, followed by 90 minutes on the III-V ICP. The nanoholes had two different approaches, starting with 260 minutes on the III-V ICP. However, it was seen that after the III-V ICP process, the leftover resist on the sample was removed during the CORE process on Pegasus 2.

ORE process

Important! The pressure settings used below may no longer be permitted, always check with the Dry etch group.

As stated by Nguyen et al., the SF₆ and O₂ fluxes are only separated after 4 s during the C-step. Shorter time steps than that do not execute the function of the clearing. After testing recipes with only 2 s of clearing and with no clear step, it was understood it could be eliminated, showing more depth as well as less undercut when going for shorter cycles. From then on, not CORE but ORE recipes were applied D/E = O/RE.
Moreover, during the E-step, the MFC (mass flow controller) presented a delay to read the pressure when compared with the power, creating unwanted bias. To fight it, the E-step was divided into E₁ and E₂. The E₁ of only 2 s is enough to stabilize the pressure, and E₂ the profile is etch correctly, without the unwanted bias.

Going further, the process is set with 4 steps, Oxidize, Removal, E₁ and E₂ steps. In the E₁, the pressure will rise, without power applied, so that for the E₂ the pressure is already rising and the power starts to be applied.

Pillars

Using patterned samples of 1 μm pillars with 50 nm of Al₂O₃. After the tests, approximately 32nm of Al₂O₃ were still intact. During the recipe, Pegasus 2 conditions were: Outer EM=10A, T=20°C, no clamping, no He BGC, no coil power and all heaters OFF.

1μm Pillar recipe from March 2022 - #244 = 122 min
	Time (s)	Pressure (valve control)	O₂ flow (SCCM)	SF₆ flow (SCCM)	Platen power (W)
O-step	10	3%	200	0	40
R-step	10	100%	0	5	40
E₁-step	2	4%	0	350	40
E₂-step	7	4%	0	350	300

* Pillars profile effect when varying the etch time.

Holes

Using patterned samples of 1 μm holes with 50 nm of Al₂O₃. After the tests, approximately 28nm of Al₂O₃ were still intact. When using this recipe, by adjusting the number of cycles, approximately 10 μm were achieved maintaining a straight profile. When going for deeper profiles, 17 μm were also achieved, but the profile starts to get positive. Further work may solve the issue. During the recipe, Pegasus 2 conditions were: Outer EM=10A, T=20°C, no clamping, no He BGC and all heaters OFF.

1μm holes recipe from March 2022
	Time (s)	Pressure	O₂ flow (SCCM)	SF₆ flow (SCCM)	Platen power (W)	Coil power (W)
O-step	10	220 mTorr	200	0	40	0
R-step	10	100%	0	40	40	0
E₁-step	2	220 mTorr	0	1200	0	0
E₂-step	1-5	220 mTorr	0	1200	1	2000

* For #288 cycles, 120 min, the profile depth reaches 17.6 μm. For #144 cycles, 60 min, the profile depth reaches 10.1 μm. Width variation presented in the pictures.

Nanoholes

The nanoholes are 200nm wide, with 400nm pitch with 100 nm Al₂O₃ mask. During the recipe, Pegasus 2 conditions were: Outer EM=10A, T=20°C, no clamping, no He BGC and all heaters OFF.

200nm nanoholes recipe from April 2022
	Time (s)	Pressure (valve control)	O₂ flow (SCCM)	SF₆ flow (SCCM)	Platen power (W)
O-step	10	220 mTorr	200	0	40
R-step	10	100%	0	40	*
E₁-step	2	220 mTorr	0	350	0
E₂-step	0-15	220 mTorr	0	350	100-300

*The R-power value was changed and observed.

The most suitable removal power was 40 W, removing the bottom of the profile correctly without damaging the top part of the profile.

200nm nanoholes recipe from July 2022
	Time (s)	Pressure (valve control)	O₂ flow (SCCM)	SF₆ flow (SCCM)	Platen power (W)
O-step	10	220 mTorr	200	0	40
R-step	10	100%	0	40	40
E₁-step	2	220 mTorr	0	350	40
E₂-step	*	220 mTorr	0	350	300

*The E₂-time value was changed and observed.

Regarding the measurements of the profiles, in a) the depth reached 3.17 µm with 0-15s of ramping, b) reached 3.21 µm with 1-15s of ramping, c) reached 2.87 µm whit 5-15s of ramping and d) reached 3.63 µm with 15 s without ramping.

Isotropic etch

Some isotropic etches were performed, intercalated with anisotropic etches. The nanoholes are 200nm wide, with 400nm pitch with 100 nm Al₂O₃ mask. The recipe presented in the table was repeated either 2 or 3 times in order to achieve the picture results. During the recipe, Pegasus 2 conditions were: Outer EM=10A, T=20°C, no clamping, no He BGC and all heaters OFF.

**Table: nanoholes + isotropic etch from June 2022**
		Time (s)	Pressure	O₂ flow (SCCM)	SF₆ flow (SCCM)	Platen power (W)	Coil power (W)
Nanoholes etch	O-step	10	220 mTorr	200	0	40	0
	R-step	10	100%	0	0	40	0
	E₁-step	2	220 mTorr	0	350	0	0
	E₂-step	0-15	220 mTorr	0	350	100-300	0
Isotropic etch	O-step	30	200 mTorr	200	0	0	2000
	R-step	30	100%	0	40	40	0
	E-step	20	200 mTorr	0	1200	0	2000

@@ Line 7: / Line 7: @@
 =Al<sub>2</sub>O<sub>3</sub> mask=
-[[File:esq met Al2O3 only (1).png|400px|thumb|'''''Scheme of the sample fabrication using Al<sub>2</sub>O<sub>3</sub> mask. The alumina oxide samples had layers with different thicknesses, of 50 or 100 nm of Al<sub>2</sub>O<sub>3</sub>, followed by 65 nm of BARC and 750 nm of DUV resist. No at scale. From @Nanolab internal, August 2022''''']]
+[[File:esq met Al2O3 compact.jpg|400px|thumb|'''''Scheme of the sample fabrication using Al<sub>2</sub>O<sub>3</sub> mask. The alumina oxide samples had layers with different thicknesses, of 50 or 100 nm of Al<sub>2</sub>O<sub>3</sub>, followed by 65 nm of BARC and 750 nm of DUV resist. No at scale. From @Nanolab internal, August 2022''''']]
 In these samples, the hard mask, Al<sub>2</sub>O<sub>3</sub>, was coated in the 150 mm Si wafers using the ALD 1 tool. These sets of samples also had the lithography process done in the DUV stepper to deposit 65 nm of BARC and 750 nm of DUV resist. Three different patterns were created after the development: