Welcome to LKO

INSIDE R&D ALERT
JUNE 8, 2001

SUB-100-NM MASK IS REPAIRABLE
Appl. Phys. Lett. Djenizian, Santinacci, Schmuki

Materials scientists in Germany tell us they have a mask that could put it into direct confrontation with conventional e-beam lithography as we move into sub-100-nm technology for device fabrication. The big plus for the University of Erlangen-Nuremberg technology: It's repairable. You can inspect the surface and where it's not what you want, you can place a repair with extremely high precision.

The researchers have already been approached by several European companies and are open to an industrial collaboration. It might be worth looking into. At the same time, they have done something which startles other scientists. They have placed a carbon layer less than 1 nm thick on a substrate, and it has blocked electron transfer and thus electrochemical reactions to an electrolyte. You would expect electron tunneling through such a thin insulating layer. You would not expect it to block much of anything. But that's what it does, and that's why you have a mask. They deposited carbon patterns on silicon by electron beam-induced contamination decomposition. Under optimized electrochemical conditions, they found that electrodeposition of gold can be blocked selectively by single-line carbon deposits only 1 nm thick. Lateral resolution of the negative patterning process is in the sub-100 nm range.

What we have is high definition patterning of semiconductor surfaces by selective electrodeposition. It's a sub-100-nm technology that could go head-to-head with e-beam, ion-beam x-ray and scanning probes in the next
generation of chipmaking. The Erlangen scientists work with an SEM (scanning electron microscope),
using it to write the selective contamination that suppresses the electrodeposition. This gives them a negative patterning method in the submicron range.

Hydrogen passivated samples of n-type silicon with a resistivity of 1-10 ohms were treated with the single line mode in an SEM. Increasing exposure time increases height and width of the lines drawn. The height increases linearly with time, but the width increases only weakly with time. After electrochemical deposition of gold, the surface is covered by a gold deposit everywhere except for the carbon lines. In the carbon-free region, a continuous, remarkably smooth gold film is left. Even the finest carbon line, less than 1 nm high, is clearly resolved and does not suffer from overgrowth. Exploring the lower size limit of the deposition process, the researchers got coherent, completely separated features where two parallel lines lead to a confined gold deposit line 70 nm wide. It may be that even smaller structures could be deposited, but their minimum experimental e-beam displacement was 100 nm.

Masks less than 1 nm thick can be produced, and even this low-layer thickness is enough to efficiently and selectively block charge transfer with the electrolyte and thus prevent an electrochemical deposition reaction. Generally it was found that higher cathodic potential steps for a short period of time give better results than longer exposure at less cathodic values. This creates more islands, which coalesce at an earlier growth stage, providing a finer grain size and better lateral resolution. The technology should be applicable to a large range of materials that can be electrodeposited from an aqueous environment (i.e., metals, semiconductor, polymers) so it should give you a lateral resolution in the 100 nm range. Any conductive substrate could be used, so it will be useful to locally functionalize different substrate materials. And it is repairable. After inspection of a surface in the SEM mode, an insulating carbon layer can be placed precisely where you want. This should be particularly attractive in making exploratory devices. Will carbon masking get into the big chipmaking race? Writing speed and best achievable lateral resolution, both still undetermined, will decide. The researchers know they must put more carbon in the vacuum to get up to maximum writing speed, and they are investigating intentional spiking of the vacuum with various carbon precursor gas molecules. Lateral resolution is still limited by their apparatus. If it comes to scale-up to production levels, doing it shouldn't be a problem. They would just use the e-beam writing systems for entire wafers already commercially available. Patenting is being considered.