Solar cell researchers have been talking about roll to roll processes for at least two decades. There are startups [1] but so far, the regular IC wafer processes still seem to be winning on cost. Here's a study of roll to roll processes.[2]

Many battery parts are printed by a roll to roll process. OLEDs have been produced that way. Flexible displays have been made, but not sold in volume.

This is a real technology, but so far, it hasn't been able to beat out the mature technologies with which it competes. Anything it can do can be done with a regular IC fab, one several generations old because this is a low-resolution technology. So it's all about cost per unit.

[1] https://www.infinitypv.com/ [2] http://energy.gov/sites/prod/files/2015/02/f19/QTR%20Ch8%20-...

I wonder if the reason roll-to-roll processes haven't caught on is that there is so much capital investment in IC/wafer processes at scale that the unit economics just don't add up (until/unless very large capital investments are made to scale up and refine margins on roll-printed circuits).

If this is the case, are there areas to attack the market? Any smaller markets which don't attract the attention of major electronics fab and might be delved to help bootstrap roll-printing? Any old/displaced technology areas that no longer have the advantage of large fabs/infrastructure investments where roll-to-roll processes could come online to replace them at competitive prices?

Any idea what it'd take, in terms of unsolved R&D and in terms of capital investment, to create a roll-to-roll process capability of printing some circuit that would be cost-competitive in some area like solar cells or maybe simple ICs?

Low-end ICs are so cheap it's not worth the trouble.

Solar cells are the most promising area. There have been roll to roll processes. Applied Materials worked on one. Energy Conversion Devices (bankrupt 2012) had a production roll to roll process for flexible solar cells. But it wasn't cheaper than silicon wafers from China.

This is good to see, since machines that build complex things is one of the requirements for the scifi-style economy of surplus. The ideal is a machine or collection of machines that can build anything you'd need, including all the parts needed to duplicate themselves, starting with raw materials like rocks, sand, plant matter and sea water. Electronics has always been difficult, but these machines can build components directly.

e-Ink promised to make huge walls of displays, but couldn't really make the economics work due to the cost of the electronics behind it. I wonder if this could solve that issue.

"...recently fabricated a flexible transistor that operates at 110 gigahertz"

I know there have been other transistors from R&D projects which operate at 690+ ghz as well. But does anyone with much better EE and chip manufacturing experience than me know how this 110 ghz transistor might translate to actual chip/IC speeds, if this thing were scaled up and produced?

I'm not an EE, but if that 110 GHz is the transistor's Ft or gain-bandwidth product, then I think that's a respectable but not exceptional transistor speed— something you'd get from a silicon-wafer process a few generations ago. Which is actually pretty nice, because it means you could use this flexible process to produce actual computational elements, not just displays and things.

It seems to me that the only advantage here is that the substrate is flexible, instead of rigid (like a wafer). Or is there more to it?

In this instance, appending the subtitle would be helpful:

"Printing with conventional rotary presses will create cheaper electronics"

Thanks, we updated the title from ”On a roll”.