The semiconductor industry has at last started upgrading to extreme ultraviolet lithography. The delay was in large part occasioned by the difficulty of building a suitable EUV light source, and looking at the source now used – a multistep process involving tin droplets – it is easy to see why it was difficult; those machines are among the most delicate and complex things our species has ever built.

As I understand it, synchrotrons are a reliable and much more direct way to generate large quantities of coherent extreme ultraviolet light. Why are they not used? According to some of the quotes linked at https://www.researchgate.net/publication/222511431_Extreme_ultraviolet_EUV_sources_for_lithography_based_on_synchrotron_radiation

High energy accelerators are however larger then would be desired for a clean room or sub-facility environment.

Okay, true, but why is this a crippling problem? Why is 'larger than would be desired' so big a disadvantage that the tin-droplet method is preferable?

The candidate enabling source for a technology transition is an optical one because it must not only meet technical requirements, but also be accessible in terms of cost of ownership and have a revenue-effective duty cycle. These requirements automatically rule out large facilities like the synchrotron's free electron lasers (FEL)

Why do these requirements automatically rule out synchrotrons? I understand synchrotrons are expensive. But the current EUV lithography equipment costs billions of dollars. How expensive is a synchrotron, that it is automatically ruled out even at that level? Or put another way: people have been building and operating synchrotrons for decades. What is it about them, that makes them so expensive that they are still uncompetitive even in this sort of cost environment? How is it possible for them to have existed for that long yet still be that expensive?

  • \$\begingroup\$ What happens when the synchrotron is misaligned - will it spray particles everywhere and thus you'll need a tonne of shielding? Is it a safety issue? \$\endgroup\$
    – Andy aka
    Jan 15 at 11:50
  • \$\begingroup\$ I love 'why not ... ' questions, especially 'why not just ... ' questions, and, full disclosure, I've asked some doozies in my time. See Dunning-Kruger. \$\endgroup\$
    – Neil_UK
    Jan 15 at 12:18
  • \$\begingroup\$ Since most of the major fabs are in Taiwan, space is a tad limited. Plus synchrotrons are power hungry - you need to pump a silly amount of power into them. For me, this is not an issue - I just go visit my physicist friend at the local synchrotron up the road when I want to do the odd bit of photolithography, xray crystallography or to reset the lock bits on a microcontroller. \$\endgroup\$
    – Kartman
    Jan 15 at 12:42
  • 1
    \$\begingroup\$ But the current EUV lithography equipment costs billions of dollars. I disagree, ASML sells their EUV machine for around $ 140 Million: cnbc.com/2021/12/10/…. I also invite you to watch: youtube.com/watch?v=5Ge2RcvDlgw there it was briefly mentioned that the getting the required EUV output power at high enough efficiency was one of the reasons to go for the tin-droplets solution. \$\endgroup\$ Jan 15 at 12:56
  • 1
    \$\begingroup\$ but you need a bunch of them per fab, don't you? Yes/No. It depends. In theory, you can make a chip factory based around one ASML EUV wafer stepper. However, that's not economical, a company will not do that as they're not going to earn enough money with only one machine. \$\endgroup\$ Jan 16 at 16:04