Specific Process Knowledge/Etch/DRIE-Pegasus/Pegasus-2/ORE with Al2O3 mask

Feedback to this page: click here

By Maria Farinha @Nanolab Internal, Jan 2023 if nothing more is stated

Al₂O₃ mask

Scheme of the sample fabrication using Al₂O₃ mask. These samples were coated with either 50 or 100 nm of Al₂O₃, followed by 65 nm of BARC and 750 nm of DUV resist. Not at scale.
Scheme: Maria Farinha, @DTU Nanolab

In these samples, the hard mask, Al₂O₃, was coated in 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:

1. 1 µm width pillars with 1 µm pitch (50 nm Al₂O₃ + 65 nm BARC + 750 nm DUV resist).
2. 1 µm width holes with 1 µm pitch (50 nm Al₂O₃ + 65 nm BARC + 750 nm DUV resist).
3. 200 nm width nanoholes with 400 nm pitch (100 nm Al₂O₃ + 65 nm BARC + 750 nm DUV resist).

The Si / Al₂O₃ / BARC / DUV stack was first processed in Pegasus 2 to etch the BARC layer (recipe named "barc gentle etch", ER=5.42nm/min), and the alumina was etched in the III-V ICP, using a BCl₃ / Ar recipe, which has an ER = 0.45 nm/min. With the pillars and holes samples, the BARC was also etched for 12 min on Pegasus 2, followed by 90 minutes on the III-V ICP.
For the nanoholes, the Al₂O₃ layer was etched for 260 minutes on the III-V ICP. The next thing would be to etch the BARC layer on Pegasus 2, 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, having no need to do the dedicated "barc gentle etch" recipe.

Samples profile before going for the ORE process:

	After Al2O3 etch	Comments
1 µm pillars		Visible that after the 90 min Al2O3 etch, the mask was correctly etched away confirmed by reaching the silicon (etched ~30nm of it).
1 µm holes		Visible that after the 90 min Al2O3 etch, the mask was correctly etched away confirmed by reaching the silicon (etched ~16nm of it).
200nm nanoholes		Visible that after the 260 min Al2O3 etch, the mask was correctly etched away confirmed by reaching the silicon (etched ~20nm of it). Resist visible on top of the mask since it was etched during the ORE process.

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 etched correctly, without the unwanted bias.

CORE process graphics and recipes: a) unwanted DC bias during the E-step; b) E-step separated into two steps, E₁ decreased the bias.
Scheme: Maria Farinha @DTU Nanolab, March 2022

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 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.

Pillars with different etching times.
Photo: Maria Farinha @DTU Nanolab, March 2022

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 was achieved maintaining a straight profile. When going for deeper profiles, 17 μm was also achieved, but the profile started 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.

Holes profile. #144 cycles for 60 min and #288 cycles for 120 min.
Photo: Maria Farinha @DTU Nanolab, March 2022

1μm holes recipe

	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. The removal power was varied, and the achieved ER for 40W was 48.5 nm/min.

Nanoholes profile when varying the removal power.
Photo: Maria Farinha @DTU Nanolab, April 2022

200nm nanoholes recipe, varying R-power

	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.

Nanoholes profile when varying the ramping E-time .
Photo: Maria Farinha @DTU Nanolab, July 2022

200nm nanoholes recipe, varying E-ramping time

	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	0
E2-step	*	220 mTorr	0	350	100-300

*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. Even though the etching ramping times may appear similar, there are differences in the feature profile. The ER for 0-15s and 1-15s is around 52 nm/min and for 5-15s and 15s is approximately 48 nm/min.

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 2 or 3 times 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. Trying to oxide the sample before doing the isotropic etch can help to protect the top part, maintaining a more straight profile. However, after trying to do an oxidation step on Pegasus 2, with 3% valve control. 200 sccm O₂ and 2000W of coil power, the oxide layer was only 3.2 nm.

x2 anisotropic + isotropic etch.
Photo: Maria Farinha @DTU Nanolab, June 2022

Table: nanoholes + isotropic etch

		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

x3 anisotropic + isotropic etch.
Photo: Maria Farinha @DTU Nanolab, June 2022

Previous work on Peg2

• SF6/O2 Plasma Etching: Optimizing the CORE process and going into 3D Engineering_November22 by Maria Farinha@DTU Nanolab