It is difficult to even comprehend the scale of modern photolithography. Current-generation manufacturing processes are described with terms like "2 nanometer" and "18 angstrom". Those numbers undersell the actual size of the logic gates, but the point is, we're working at a scale so small that the wavelengths of light used for the etching have an impact on the maximum resolution of the etching process. Current-gen chips use "extreme ultraviolet," or EUV, but that's only going to get us so far.
The current plan is to extend the use of EUV lithography by increasing the numerical aperture, or NA value, of the lithographic process. This increases resolution by focusing the light ever more carefully using lenses and mirrors, but making these hyper-precise optics gets very expensive and punishingly difficult with rapidity. What if we instead used "sharper" light? It sounds silly, but that's exactly what researchers from a large group led by scientists at John Hopkins University have proposed.
The concept of using "Beyond EUV" light with a shorter wavelength -- also called "Soft X-Rays" due to their extremely short wavelength -- isn't exactly novel, but there are multiple problems with "B-EUV." One of the biggest is that it doesn't affect the chemicals used in photolithography the same way as EUV light. This sent the research group looking for alternatives to the "resists" currently used in chip etching. The laboratory in Baltimore discovered that Zinc is very good at absorbing B-EUV light, and they've now proved that they can use that discovery to make "amorphous zeolitic imidazolate frameworks" (aZIFs) that can in turn be used for photolithography.
That's one major issue down, but there are still critical problems to solve with the idea of B-EUV photolithography. For one thing, we don't actually have a standardized method of safely and efficiently producing B-EUV radiation. For another thing, there don't exist mirrors that are designed to reflect this kind of radiation with a high efficiency, either. Both of these issues are massive barriers still standing in the way of commercializing this technology for mass production.
Still, this is a legitimate breakthrough. The lead researcher, Michael Tsapatsis, told Cosmos that he believes this technology will find its way into processor manufacturing within the next ten years. As he says, "companies have their roadmaps of where they want to be in 10 to 20 years and beyond." It will be fascinating to see just how long advancements like this can keep microprocessor scaling going.