Using Light to Make Processors with Features Smaller than It's Wavelength?

[peter.ell]

I've been reading up on photolithography trying to understand how processors with sub-wavelength features can be made through photolithography, but I just don't get it.

Can someone please help me by explaining in a conceptual way how light with a wavelength of, say, 200nm can be used to make a processor with 65nm features? Even with a large lens used to focus the light, you still can't make a point smaller than the wavelength of the light being focused, right?

Thank you so much!

Also, what is meant by numerical aperture in regards to a lens? Wikipedia didn't help me understand this.

 

[Andy Resnick]

The photolithography guys use a lot of tricks- using a phase mask (as opposed to intensity masks) is one technology, near-field contact mode is another. There are more exotic technologies under development. The mechanisms by which these technologies work are quite complex.

Loosely speaking, numerical aperture is a relationship between the focal length and diameter of a lens. Recall that diffraction occurs as a result of limiting the spatial extent of a wavefront. Collecting more of the wavefront (by increasing the diameter of the lens) means that diffraction effects are reduced, leading to a smaller spot size.

 

[JeffKoch]

Basically, use a big lens, plus some other tricks, for projection lithography. There is nothing limiting a focal spot size to the dimension of a wavelength, though in practice it becomes increasingly challenging to do this.

 

[chrisbaird]

When you go to sizes smaller then a light beam's wavelength, the electromagnetic fields of the light do not just disappear and become unusable. Rather, at sub-wavelength sizes, our neat intuitive picture of light being a bundle of rays traveling in straight lines and bouncing off things like billiard balls breaks down. For example, the air molecules in the sky are sub-wavelength to most of sunlight's spectrum, yet we have no problem seeing the blue sky. All that is required at sub-wavelength sizes is to return to full-wave electromagnetic solutions to Maxwell's equations and avoid assumptions or mental constructs from the world of optics (easy to do in principle, hard to do practice). When you go sub-wavelength, you have to really know what the electromagnetic fields are doing (typically using numerical em codes), and not just ray-trace.