WEBVTT 00:00:00.000 --> 00:00:08.140 Processors have come a really long way over the past few decades, but one thing that's remained constant is the fact that they're based on silicon wafers. 00:00:09.300 --> 00:00:12.660 But it turns out we don't have to use silicon substrates. 00:00:12.660 --> 00:00:20.160 In fact, if we print transistors on another type of material, we could get processors that are not only cheaper, but bendable. 00:00:22.100 --> 00:00:29.000 That's right. While everyone is talking about foldable phone screens, scientists have been working on a flexible ARM-based processor 00:00:29.240 --> 00:00:35.800 printed on plastic. But how the heck do you put transistors on plastic and why would we want a flexible CPU anyway? 00:00:35.800 --> 00:00:41.060 So it turns out that we've actually been putting transistors onto materials other than silicon for a long time now. 00:00:41.420 --> 00:00:46.720 Specifically, I'm talking about oxide thin film transistors, better known as TFTs. 00:00:47.000 --> 00:00:50.440 These have been used for quite a while in run-of-the-mill LCD screens. 00:00:50.480 --> 00:00:57.880 The idea is that instead of silicon, the transistors are printed onto some kind of non-conducting substrate, such as glass. 00:00:57.920 --> 00:01:04.920 But even though we've been using this tech for displays for a long time, full-fledged CPUs are a much more complicated ballgame, 00:01:04.920 --> 00:01:08.640 kind of like when you get a double reverse triple-out-of-bounder in the 8th inning. 00:01:10.320 --> 00:01:17.320 And the bendable CPU we're talking about today is notable because not only is it, well, a full-fledged CPU, 00:01:17.440 --> 00:01:24.040 but it uses an existing architecture that's already been put into tons of devices, the ARM Cortex M0. 00:01:24.320 --> 00:01:32.400 The new chip is essentially an M0 core, plus a small amount of memory on a plastic substrate, printed using photolithography, 00:01:32.600 --> 00:01:39.240 meaning the printing process is similar to how conventional CPUs are made, which you can learn lots more about up in this video. 00:01:39.400 --> 00:01:44.200 The transistors themselves can be made from indium-gallium zinc oxide, better known as 00:01:44.800 --> 00:01:48.720 IGSO, again commonly used in TFT displays at a low cost. 00:01:48.720 --> 00:01:57.560 Of course, since we're talking about cheap production methods, you might think that the plastic ARM, as it's called, isn't very computationally powerful, and... 00:01:58.360 --> 00:02:04.600 Well, you'd be right. It runs at only 29 kilohertz and is built on an 800 nanometer process, 00:02:04.600 --> 00:02:09.320 which is the same process the original Pentium's from 1993 used. Great year. 00:02:09.320 --> 00:02:15.600 And the thing isn't even very power efficient. Despite the fact it uses 21 milliwatts of power, which seems low, 00:02:15.600 --> 00:02:19.240 it turns out that 99% of that is lost to waste heat. 00:02:19.320 --> 00:02:21.960 So yeah, it's slow and inefficient. 00:02:22.440 --> 00:02:28.880 So what's the point? So, remember how we said the Cortex M0 is in a huge number of embedded devices already? 00:02:29.040 --> 00:02:34.920 Putting the M0 on flexible plastic means a chip that's complex enough to handle more advanced functions 00:02:34.960 --> 00:02:40.680 could go into many more products, especially in cases where silicon is too expensive or too brittle. 00:02:40.680 --> 00:02:45.000 Even though the plastic ARM's implementation of the M0 is quite slow, as we said, 00:02:45.080 --> 00:02:50.600 it should still be around 12 times as powerful as previous integrated circuits based on plastic. 00:02:50.600 --> 00:02:57.240 Plastic ARM would have enough computational muscle to be connected to environmental sensors and alert users to real-time conditions. 00:02:57.240 --> 00:03:04.440 Think about a plastic ARM chip inside food packaging that could tell if the food inside was spoiling instead of going by an expiration date. 00:03:04.440 --> 00:03:08.520 Or how about a bandage that could keep an eye on how well your cuts are healing? 00:03:08.520 --> 00:03:15.880 It's like a little tiny doctor on your finger. Although we do already have microchips that can accomplish some tasks that plastic ARM is envisioned for, 00:03:15.880 --> 00:03:21.080 plastic ARM could enable them to be deployed in many more environments for much less money. 00:03:21.080 --> 00:03:28.440 But I wouldn't expect to see it in the immediate future, as the plastic ARM prototype that was just developed can only execute a few hard-coded programs. 00:03:28.440 --> 00:03:31.720 And scientists still need to make the chip more power efficient. 00:03:31.720 --> 00:03:39.000 However, as time goes on, ARM believes we could see chips like these in more than a trillion objects over the course of 10 years. 00:03:39.000 --> 00:03:44.040 Hopefully, some of them will end up in lottery tickets, so I can be disappointed as soon as the drawing happens.