WEBVTT 00:00:00.000 --> 00:00:02.740 So usually when we look at our phones or our 00:00:02.740 --> 00:00:06.259 laptops, we just think of them as these clean, 00:00:06.480 --> 00:00:09.859 stable blocks of solid state magic. Right, yeah. 00:00:09.919 --> 00:00:11.919 It's a really common assumption. Like we imagine 00:00:11.919 --> 00:00:15.000 there are no moving parts, no messy chemicals 00:00:15.000 --> 00:00:18.059 involved, just pure, pristine silicon doing the 00:00:18.059 --> 00:00:20.719 math. Exactly. We expect the devices we carry 00:00:20.719 --> 00:00:23.699 around to just be, well, completely sanitized, 00:00:23.859 --> 00:00:26.239 inert pieces of tech. But if you actually break 00:00:26.239 --> 00:00:29.210 open almost any device you own, Tucked inside 00:00:29.210 --> 00:00:32.710 are these tiny volatile chemical silos. Oh yeah, 00:00:32.850 --> 00:00:35.369 very volatile. Little pressurized cans of fluid 00:00:35.369 --> 00:00:37.850 and metal that are constantly fighting to keep 00:00:37.850 --> 00:00:41.649 from literally exploding all while doing a job 00:00:41.649 --> 00:00:43.969 that makes modern electronics possible. It's 00:00:43.969 --> 00:00:45.729 wild when you really think about it. It really 00:00:45.729 --> 00:00:47.570 is. And you probably never know it, but there's 00:00:47.570 --> 00:00:50.210 this incredibly dense Wikipedia article we're 00:00:50.210 --> 00:00:52.869 looking at today that outlines a century -long 00:00:52.869 --> 00:00:55.289 war taking place inside your phone right now. 00:00:55.390 --> 00:00:58.090 Yes, the history of electrolytic capacitors. 00:00:58.429 --> 00:01:01.750 Right. So whether you are just insanely curious 00:01:01.750 --> 00:01:03.990 about how the physical world works, or maybe 00:01:03.990 --> 00:01:06.310 you're prepping for an engineering project, our 00:01:06.310 --> 00:01:09.230 mission for this deep dive is to shortcut past 00:01:09.230 --> 00:01:13.090 all the intense jargon. We are going to translate 00:01:13.310 --> 00:01:16.510 the fascinating reality of electrolytic capacitors. 00:01:16.829 --> 00:01:19.290 And, you know, to really grasp what's happening 00:01:19.290 --> 00:01:21.650 inside those little cylinders, we sort of have 00:01:21.650 --> 00:01:23.849 to establish a baseline of what they're actually 00:01:23.849 --> 00:01:26.530 designed to do. Okay, let's unpack this. What 00:01:26.530 --> 00:01:29.750 is the baseline? So, at its absolute core... 00:01:29.719 --> 00:01:32.319 An electrolytic capacitor is a component that 00:01:32.319 --> 00:01:34.480 stores massive amounts of electrical energy. 00:01:35.120 --> 00:01:37.739 It acts like a, well, like a temporary reservoir 00:01:37.739 --> 00:01:40.280 for power. So it's just smoothing out the electrical 00:01:40.280 --> 00:01:42.760 current so your device gets a steady supply of 00:01:42.760 --> 00:01:45.620 juice. Exactly. And it achieves this by using 00:01:45.620 --> 00:01:48.659 an incredibly thin oxide layer on a metal surface, 00:01:48.739 --> 00:01:51.299 which acts as an insulator. But as I was reading 00:01:51.299 --> 00:01:53.659 the source material, there's this distinct historical 00:01:53.659 --> 00:01:55.719 theme that pops up over and over. Attention, 00:01:55.780 --> 00:01:58.129 right. Yeah, this constant almost aggressive 00:01:58.129 --> 00:02:00.829 push and pull between the tech industry's drive 00:02:00.829 --> 00:02:04.590 to make devices smaller and, well, the fundamental 00:02:04.590 --> 00:02:06.930 laws of physical chemistry threatening to make 00:02:06.930 --> 00:02:09.289 them violently fail. It's the ultimate engineering 00:02:09.289 --> 00:02:11.349 challenge, shrinking things down without turning 00:02:11.349 --> 00:02:14.169 them into tiny bombs. Right. But before we get 00:02:14.169 --> 00:02:16.389 into the explosions, we kind of have to go backward 00:02:16.389 --> 00:02:18.729 to figure out the microscopic trick that makes 00:02:18.729 --> 00:02:21.009 storing energy in a tiny can possible at all. 00:02:21.150 --> 00:02:23.629 Yeah, because this isn't modern Silicon Valley 00:02:23.629 --> 00:02:27.319 tech. The foundational discovery goes way back 00:02:27.319 --> 00:02:29.879 to the mid -19th century. 1857, right? Yeah, 00:02:29.939 --> 00:02:33.639 1857. There was this German physicist and chemist 00:02:33.639 --> 00:02:36.240 named Johann Heinrich Buff. He was just experimenting 00:02:36.240 --> 00:02:38.800 in his lab, running electrical currents through 00:02:38.800 --> 00:02:41.479 different materials. Like you. Right, just 19th 00:02:41.479 --> 00:02:43.939 century science things. And he noticed something 00:02:43.939 --> 00:02:46.280 highly unusual happening with specific metals 00:02:46.280 --> 00:02:48.819 in an electrochemical bath. What were they doing? 00:02:49.199 --> 00:02:51.180 Well, these metals would naturally form a very 00:02:51.180 --> 00:02:53.949 thin oxide layer on their surface. But unlike 00:02:53.949 --> 00:02:56.490 normal rust, which, you know, just eats away 00:02:56.490 --> 00:02:59.729 and destroys the metal, this specific oxide layer 00:02:59.729 --> 00:03:03.150 had a totally unique property. It completely 00:03:03.150 --> 00:03:05.789 blocked electric current from flowing in one 00:03:05.789 --> 00:03:09.210 direction, but it allowed it to flow almost freely 00:03:09.210 --> 00:03:11.509 in the opposite direction. Oh, wow. So it's acting 00:03:11.509 --> 00:03:13.849 like a one -way street for electricity? Yes. 00:03:13.969 --> 00:03:16.469 Or I guess more like a physical plumbing valve, 00:03:16.710 --> 00:03:18.830 like it lets water push through one way, but 00:03:18.830 --> 00:03:21.030 if the water tries to reverse, it slams shut. 00:03:21.229 --> 00:03:23.330 Which is exactly how they were eventually named. 00:03:23.550 --> 00:03:27.629 By 1875, a French researcher, Eugene Ducrete, 00:03:28.270 --> 00:03:30.550 was studying the same thing, and he actually 00:03:30.550 --> 00:03:34.250 coined the term valve metals. Valve metals? Yeah. 00:03:34.409 --> 00:03:36.169 Because they're literally one -way valves for 00:03:36.169 --> 00:03:39.189 current. Exactly. And then a few decades later, 00:03:39.389 --> 00:03:42.430 in 1896, a guy named Charles Pollack took that 00:03:42.430 --> 00:03:44.949 concept further. He realized that if you have 00:03:44.949 --> 00:03:47.789 the stable, highly insulating oxide layer blocking 00:03:47.789 --> 00:03:50.550 the current, you can use it to build a polarized 00:03:50.550 --> 00:03:53.180 capacitor. which is basically a device that traps 00:03:53.180 --> 00:03:55.259 and holds an electrical charge on that barrier. 00:03:55.419 --> 00:03:57.360 You got it. So let's bring this into the physical 00:03:57.360 --> 00:04:00.120 world for a second. If I took a pair of pliers 00:04:00.120 --> 00:04:02.280 right now and just cracked open one of these 00:04:02.280 --> 00:04:04.659 modern cylindrical capacitors on a circuit board, 00:04:05.039 --> 00:04:07.360 what are the actual pieces I'm looking at? OK, 00:04:07.379 --> 00:04:10.120 so you'd essentially find three main layers rolled 00:04:10.120 --> 00:04:12.439 up inside. First, you have the anode, the positive 00:04:12.439 --> 00:04:14.840 electrode. This is usually made of a valve metal, 00:04:14.960 --> 00:04:17.740 like a strip of high purity aluminum foil. OK, 00:04:17.800 --> 00:04:20.240 aluminum foil. Got it. Second, you have the dielectric. 00:04:20.490 --> 00:04:23.670 This is that incredibly thin, insulating oxide 00:04:23.670 --> 00:04:26.250 layer that forms directly on the surface of the 00:04:26.250 --> 00:04:29.310 aluminum. The rust that isn't rust. Right. And 00:04:29.310 --> 00:04:31.430 third, you have the cathode, which is the negative 00:04:31.430 --> 00:04:34.310 electrode. But here is the really critical part 00:04:34.310 --> 00:04:37.610 about electrolytic capacitors. The true cathode 00:04:37.610 --> 00:04:39.689 isn't another piece of solid metal. Wait, it's 00:04:39.689 --> 00:04:42.730 not? No, it's actually a liquid or gel electrolyte 00:04:42.730 --> 00:04:45.550 that physically surrounds and touches every microscopic 00:04:45.550 --> 00:04:48.550 part of that oxide layer. Oh, so it's a wet component. 00:04:49.049 --> 00:04:52.009 But if they just used a flat strip of aluminum 00:04:52.009 --> 00:04:56.430 foil, the capacitor would have to be the size 00:04:56.430 --> 00:04:58.829 of a dinner plate to hold enough energy for a 00:04:58.829 --> 00:05:01.110 modern device. It would be huge, yeah. So the 00:05:01.110 --> 00:05:04.230 real engineering magic here is the scale. They 00:05:04.230 --> 00:05:06.769 don't leave the aluminum smooth. They etch it. 00:05:06.889 --> 00:05:09.350 Yeah, the etching process is where all the modern 00:05:09.350 --> 00:05:11.850 capacity really comes from. Before they ever 00:05:11.850 --> 00:05:15.610 form that oxide layer, manufacturers use an electrochemical 00:05:15.610 --> 00:05:18.449 process to eat away at the aluminum foil. So 00:05:18.449 --> 00:05:21.870 they're scarring it up. Exactly. Creating this 00:05:21.870 --> 00:05:25.550 deeply scarred, rough, mountainous topography 00:05:25.550 --> 00:05:28.329 at the microscopic level. We're talking about 00:05:28.329 --> 00:05:31.170 increasing the physical surface area of the foil 00:05:31.170 --> 00:05:36.480 by up to 200 times. 200 times. compared to a 00:05:36.480 --> 00:05:39.180 perfectly smooth piece of metal. Yeah, exactly 00:05:39.180 --> 00:05:42.579 the same dimensions, but vastly more surface 00:05:42.579 --> 00:05:44.699 area. I'm trying to visualize this. It's almost 00:05:44.699 --> 00:05:47.420 like taking a giant, highly textured sponge, 00:05:47.899 --> 00:05:50.120 something with just a massive, intricate amount 00:05:50.120 --> 00:05:52.339 of physical surface area, and cramming it down 00:05:52.339 --> 00:05:53.899 into a thimble. That's a perfect way to look 00:05:53.899 --> 00:05:55.759 at it. You are somehow fitting a football field 00:05:55.759 --> 00:05:59.040 where the microscopic metal surface into a can 00:05:59.040 --> 00:06:01.420 the size of a pencil eraser. And that's necessary 00:06:01.420 --> 00:06:03.800 because the capacitance, the actual volume of 00:06:03.800 --> 00:06:06.480 energy the component can store, is directly dictated 00:06:06.480 --> 00:06:08.980 by the surface area of the electrodes and the 00:06:08.980 --> 00:06:11.160 extreme thinness of the insulator between them. 00:06:11.319 --> 00:06:13.220 And the oxide layer is measured in nanometers, 00:06:13.379 --> 00:06:16.519 right? Literally nanometers. So when you combine 00:06:16.519 --> 00:06:19.199 a nanometer thin insulator with that massive 00:06:19.199 --> 00:06:22.439 sponge -like etched surface area, you get a staggering 00:06:22.439 --> 00:06:25.899 energy capacity inside a very tiny volume. I 00:06:25.899 --> 00:06:28.000 saw the source material refer to this as the 00:06:28.000 --> 00:06:30.519 CV product. I'm assuming CV product is basically 00:06:30.519 --> 00:06:32.779 just the ultimate engineering metric for how 00:06:32.779 --> 00:06:36.279 much juice fits into a tiny space. Spot on. Capacitance 00:06:36.279 --> 00:06:38.720 multiplied by voltage. It's the standard way 00:06:38.720 --> 00:06:41.339 engineers measure volumetric efficiency. The 00:06:41.339 --> 00:06:43.829 higher the CV product, the more more power you're 00:06:43.829 --> 00:06:46.550 packing into a smaller footprint. That microscopic 00:06:46.550 --> 00:06:51.110 sponge trick is honestly brilliant. Just squeezing 00:06:51.110 --> 00:06:53.889 maximum surface area into a tiny space. Yeah. 00:06:54.040 --> 00:06:56.000 But as we move forward into the 20th century, 00:06:56.319 --> 00:06:58.079 even that aluminum trick wasn't small enough, 00:06:58.079 --> 00:06:59.899 right? No, it really wasn't, because then the 00:06:59.899 --> 00:07:01.720 microchip gets invented. Right, and suddenly 00:07:01.720 --> 00:07:04.139 the entire hardware world is in a desperate race 00:07:04.139 --> 00:07:06.240 to shrink everything else down to match it. Yeah, 00:07:06.500 --> 00:07:08.399 because the physical realities of the early 20th 00:07:08.399 --> 00:07:11.759 century were incredibly bulky. Like in the 1920s, 00:07:12.139 --> 00:07:14.740 early wet aluminum capacitors were massive. How 00:07:14.740 --> 00:07:17.399 big are we talking? Well, they were used to filter 00:07:17.399 --> 00:07:20.100 out hum and background noise in large telephone 00:07:20.100 --> 00:07:24.740 exchanges, or to stabilize those giant furniture 00:07:24.740 --> 00:07:27.660 -sized wooden radio receivers people had. They 00:07:27.660 --> 00:07:30.959 were literally large metallic boxes or glass 00:07:30.959 --> 00:07:34.920 jars filled with liquid borax electrolyte. Definitely 00:07:34.920 --> 00:07:37.000 not something you can slide into your jeans pocket. 00:07:37.259 --> 00:07:40.060 Not at all. So the major historical pivot happened 00:07:40.060 --> 00:07:44.060 in the 1950s. Bell Labs had just invented the 00:07:44.060 --> 00:07:46.560 solid state transistor. OK, a huge breakthrough. 00:07:46.939 --> 00:07:49.800 Massive. Suddenly they had this revolutionary 00:07:49.800 --> 00:07:53.139 tiny component that could process signals, but 00:07:53.139 --> 00:07:55.920 the capacitors they needed to support the transistor's 00:07:55.920 --> 00:07:58.339 power supply were still relatively huge. So the 00:07:58.339 --> 00:08:00.399 supporting cast was dwarfing the star of the 00:08:00.399 --> 00:08:03.180 show. Exactly. So researchers at Bell Labs turned 00:08:03.180 --> 00:08:05.759 away from a and look to a different valve metal 00:08:05.759 --> 00:08:08.360 entirely tantalum. But wait, changing the metal 00:08:08.360 --> 00:08:10.199 doesn't automatically shrink the physical size 00:08:10.199 --> 00:08:12.399 of the component. What does tantalum do differently 00:08:12.399 --> 00:08:14.379 that allows the whole package to get smaller? 00:08:14.639 --> 00:08:17.480 It comes down to a property called relative permittivity, 00:08:18.199 --> 00:08:21.050 tantalum pentoxide. The oxide layer that forms 00:08:21.050 --> 00:08:23.529 on tantalum has a relative permittivity of about 00:08:23.529 --> 00:08:26.870 27, which is roughly three times higher than 00:08:26.870 --> 00:08:29.310 the aluminum oxide used in the older capacitors. 00:08:29.589 --> 00:08:31.410 Let me stop you there, because permittivity of 00:08:31.410 --> 00:08:34.649 27 sounds like pure physics jargon. Fair enough. 00:08:35.389 --> 00:08:38.610 To visualize this, I think of permittivity like 00:08:38.610 --> 00:08:41.309 the thickness of a crash mat absorbing a fall. 00:08:42.490 --> 00:08:46.529 So tantalum's mat absorbs and holds Three times 00:08:46.529 --> 00:08:48.889 the electrical field as aluminum does. That is 00:08:48.889 --> 00:08:50.909 a highly effective way to look at it, yes. Which 00:08:50.909 --> 00:08:53.549 means you need three times less physical material 00:08:53.549 --> 00:08:56.450 to do the exact same job. Exactly. Because of 00:08:56.450 --> 00:08:58.429 that higher permittivity, you can hold the exact 00:08:58.429 --> 00:09:01.009 same amount of energy in a component one -third 00:09:01.009 --> 00:09:04.190 the size. And the manufacturing process Bell 00:09:04.190 --> 00:09:06.269 Labs developed for this was a radical departure, 00:09:06.490 --> 00:09:08.110 too. They didn't use foil anymore, did they? 00:09:08.230 --> 00:09:10.549 No. Instead of a strip of foil, they ground the 00:09:10.549 --> 00:09:13.610 raw tantalum metal into a fine microscopic dust. 00:09:13.870 --> 00:09:16.950 Then they pressed that powder into a tiny cylindrical 00:09:16.950 --> 00:09:19.509 pellet, they called a slug, and baked it in a 00:09:19.509 --> 00:09:21.750 vacuum furnace at extreme temperatures. We're 00:09:21.750 --> 00:09:24.269 talking like what, 1 ,500 to 2 ,000 degrees Celsius? 00:09:24.730 --> 00:09:27.750 Yep, incredibly hot. That baking process is called 00:09:27.750 --> 00:09:30.389 sintering, right? Yes, sintering. It's like taking 00:09:30.389 --> 00:09:33.570 a snowball and squeezing it in your hands, just 00:09:33.570 --> 00:09:36.250 enough so the ice crystals melt and fuse together 00:09:36.250 --> 00:09:39.669 at the edges. But the snowball itself stays completely 00:09:39.669 --> 00:09:42.590 porous and full of tiny air pockets. Right. They 00:09:42.590 --> 00:09:45.250 melt the tantalum dust just enough to stick together 00:09:45.250 --> 00:09:48.230 into a solid structure, but it remains a highly 00:09:48.230 --> 00:09:51.450 porous microscopic sponge on the inside. So you're 00:09:51.450 --> 00:09:53.730 keeping the massive internal surface area, but 00:09:53.730 --> 00:09:56.190 now you are building it out of a vastly superior 00:09:56.190 --> 00:10:00.200 material. Exactly. And these solid Tantalum capacitors 00:10:00.200 --> 00:10:02.679 completely changed the trajectory of consumer 00:10:02.679 --> 00:10:06.100 electronics. They are what made the extreme miniaturization 00:10:06.100 --> 00:10:09.679 of modern laptops, cell phones, and medical implants 00:10:09.679 --> 00:10:12.080 physically possible. I do have to ask the obvious 00:10:12.080 --> 00:10:14.960 question here, though. If Tantalum is three times 00:10:14.960 --> 00:10:17.379 better at holding a charge and lets us shrink 00:10:17.379 --> 00:10:20.279 devices down so well, why isn't every single 00:10:20.279 --> 00:10:22.399 device in my house just packed entirely with 00:10:22.399 --> 00:10:24.940 Tantalum? Why does aluminum even still exist? 00:10:25.210 --> 00:10:27.789 Well, because material science always has to 00:10:27.789 --> 00:10:29.710 answer to the brutal realities of the global 00:10:29.710 --> 00:10:33.730 market. Economics. Always. Tantalum is a relatively 00:10:33.730 --> 00:10:36.750 rare material, and its supply chain is incredibly 00:10:36.750 --> 00:10:39.789 volatile. There was a massive price shock for 00:10:39.789 --> 00:10:42.970 raw tantalum in 1980. Oh, wow. And then right 00:10:42.970 --> 00:10:45.889 at the turn of the millennium, around 2000, 2001, 00:10:46.470 --> 00:10:49.450 the mobile phone industry absolutely exploded. 00:10:49.590 --> 00:10:51.649 Right. Everybody was buying cell phones. Exactly. 00:10:52.009 --> 00:10:55.450 The demand for tiny tantalum capacitors skyrocketed, 00:10:55.490 --> 00:10:58.210 which created a massive supply crunch and another 00:10:58.210 --> 00:11:01.289 huge price explosion. So it just became economically 00:11:01.289 --> 00:11:04.230 impossible to put them into everyday consumer 00:11:04.230 --> 00:11:06.669 electronics without driving the price of the 00:11:06.669 --> 00:11:08.970 gadgets through the roof. Exactly. The entire 00:11:08.970 --> 00:11:11.759 tech industry was forced to pivot. Manufacturers 00:11:11.759 --> 00:11:13.940 went back to the drawing board to improve the 00:11:13.940 --> 00:11:16.100 cheaper aluminum capacitors so they could actually 00:11:16.100 --> 00:11:18.639 use them for modern applications. Necessity breeds 00:11:18.639 --> 00:11:21.620 invention. It really does. The shortage also 00:11:21.620 --> 00:11:24.360 heavily spurred the development of niobium capacitors. 00:11:24.740 --> 00:11:27.460 Niobium is a sister metal to tantalum. It sits 00:11:27.460 --> 00:11:29.820 right near it on the periodic table and behaves 00:11:29.820 --> 00:11:31.940 almost identically in a chemical sense. But it's 00:11:31.940 --> 00:11:35.100 cheaper. Way cheaper and far more abundant in 00:11:35.100 --> 00:11:37.740 nature. It really highlights how the physical 00:11:37.740 --> 00:11:40.639 availability and costs of raw metals directly 00:11:40.639 --> 00:11:43.200 dictate the engineering of our tech. Okay, so 00:11:43.200 --> 00:11:45.080 we figured out how to make these energy silos 00:11:45.080 --> 00:11:48.019 microscopic and cheap using aluminum and clever 00:11:48.019 --> 00:11:50.799 engineering. But here's where it gets really 00:11:50.799 --> 00:11:53.659 interesting. When you squeeze that much raw power 00:11:53.659 --> 00:11:57.100 into a cheap microscopic space, the physics of 00:11:57.100 --> 00:11:59.259 the situation start fighting back. Oh, they fight 00:11:59.259 --> 00:12:01.679 back hard. Pushing these materials to their absolute 00:12:01.679 --> 00:12:04.840 limits comes with a massive, sometimes literal, 00:12:05.080 --> 00:12:07.620 cost. And that brings us to how these things 00:12:07.620 --> 00:12:10.139 violently fail. Yeah, catastrophic failures are 00:12:10.139 --> 00:12:12.539 a very real part of the engineering. The source 00:12:12.539 --> 00:12:15.100 material is adamant about this. Electrolytic 00:12:15.100 --> 00:12:17.659 capacitors are highly polarized. They are completely 00:12:17.659 --> 00:12:19.860 asymmetrical by design. Right, meaning electricity 00:12:19.860 --> 00:12:22.620 is only supposed to enter from one specific direction. 00:12:23.059 --> 00:12:25.059 They must be operated with a higher positive 00:12:25.059 --> 00:12:28.019 voltage on the anode terminal at all times. And 00:12:28.019 --> 00:12:30.110 what happens if you don't? If you make a mistake 00:12:30.110 --> 00:12:32.789 on a circuit board and reverse the polarity even 00:12:32.789 --> 00:12:36.169 by just 1 to 1 .5 volts for a sustained period, 00:12:36.909 --> 00:12:39.350 you will fundamentally destroy that nanometer 00:12:39.350 --> 00:12:42.149 thin oxide layer. And when that barrier is destroyed, 00:12:42.450 --> 00:12:44.190 we aren't just talking about your laptop screen 00:12:44.190 --> 00:12:46.830 quietly going black, are we? We're talking about 00:12:46.830 --> 00:12:49.669 an aggressive chemical breakdown inside the hardware. 00:12:49.929 --> 00:12:52.750 Definitely. The most common physical failure 00:12:52.750 --> 00:12:56.350 modes are overvoltage, thermal runaway from handling 00:12:56.350 --> 00:12:58.970 too much alternating current, and electrolyte 00:12:58.970 --> 00:13:02.230 decomposition. That sounds intense. It is. When 00:13:02.230 --> 00:13:04.649 these intrinsic components get too hot or the 00:13:04.649 --> 00:13:07.450 voltage spikes pass their rating, the liquid 00:13:07.450 --> 00:13:10.129 electrolyte inside the cylinder starts to literally 00:13:10.129 --> 00:13:12.710 boil and decompose, and that rapidly generates 00:13:12.710 --> 00:13:14.789 hydrogen gas. So you're generating expanding 00:13:14.789 --> 00:13:18.240 gas inside a tightly sealed metal cylinder. That 00:13:18.240 --> 00:13:20.320 is a bomb. The pressure builds up incredibly 00:13:20.320 --> 00:13:23.120 fast. Most modern aluminum capacitors have a 00:13:23.120 --> 00:13:25.419 tiny safety vent scored into the top of the metal 00:13:25.419 --> 00:13:27.720 can. Oh, is that what that is? If you look closely, 00:13:27.799 --> 00:13:30.320 it looks like a little K or an X stamped into 00:13:30.320 --> 00:13:32.519 the aluminum. Yep, that's exactly what that is. 00:13:32.620 --> 00:13:35.340 It's an engineered weak point designed to safely 00:13:35.340 --> 00:13:37.879 pop open and vent the gas before the whole thing 00:13:37.879 --> 00:13:40.639 goes. But what if it doesn't pop? Well, if the 00:13:40.639 --> 00:13:42.860 pressure builds too aggressively or that vent 00:13:42.860 --> 00:13:46.500 fails to open, the capacitor will outright explode 00:13:46.500 --> 00:13:49.799 like a firecracker. It shoots out hot electrolyte 00:13:49.799 --> 00:13:52.059 fluid and destroys the sensitive circuitry all 00:13:52.059 --> 00:13:54.240 around it. Which brings us to one of the most 00:13:54.240 --> 00:13:57.419 wild, chaotic stories in modern tech history. 00:13:57.690 --> 00:14:01.950 The capacitor plague. Oh, yes. From roughly 1999 00:14:01.950 --> 00:14:05.250 to 2010, millions of devices around the world 00:14:05.250 --> 00:14:07.789 just started detonating on people's desks. It 00:14:07.789 --> 00:14:10.490 was an absolute crisis for the industry, and 00:14:10.490 --> 00:14:12.850 the root cause traces back to manufacturers trying 00:14:12.850 --> 00:14:15.649 to lower something called equivalent series resistance, 00:14:15.730 --> 00:14:19.169 or ESR. Let's break down ESR. Why do manufacturers 00:14:19.169 --> 00:14:21.610 care so much about lowering resistance inside 00:14:21.610 --> 00:14:24.190 a capacitor? Because in electrical engineering, 00:14:24.730 --> 00:14:27.230 electrical resistance always creates heat. and 00:14:27.230 --> 00:14:30.570 heat is the absolute enemy of electronics. Manufacturers 00:14:30.570 --> 00:14:33.350 desperately wanted cheaper, highly conductive 00:14:33.350 --> 00:14:36.009 liquid electrolytes that would lower the internal 00:14:36.009 --> 00:14:39.750 resistance and run cooler. So they started experimenting 00:14:39.750 --> 00:14:42.350 with using water -based solutions instead of 00:14:42.350 --> 00:14:45.789 the traditional organic solvents. But wait, pouring 00:14:45.789 --> 00:14:48.710 water onto raw aluminum inside an electrical 00:14:48.710 --> 00:14:50.730 component sounds like a recipe for disaster. 00:14:50.909 --> 00:14:53.590 Doesn't water react aggressively with aluminum? 00:14:54.009 --> 00:14:56.250 Exceptionally aggressively. The chemical reaction 00:14:56.250 --> 00:14:58.789 between water and aluminum generates high heat 00:14:58.789 --> 00:15:01.710 and a massive amount of hydrogen gas. So how 00:15:01.710 --> 00:15:04.409 could they even consider using it? Well, to make 00:15:04.409 --> 00:15:07.210 a water -based electrolyte function safely inside 00:15:07.210 --> 00:15:10.789 a commercial capacitor, you need a very specific, 00:15:10.990 --> 00:15:14.309 painstakingly balanced chemical cocktail of stabilizing 00:15:14.309 --> 00:15:17.070 inhibitors. Those prevent that exact reaction 00:15:17.070 --> 00:15:18.669 from happening. And this is where the history 00:15:18.669 --> 00:15:20.840 reads like a corporate heist movie. According 00:15:20.840 --> 00:15:23.360 to the sources, a highly guarded recipe for a 00:15:23.360 --> 00:15:25.299 water -based electrolyte was actually stolen. 00:15:25.460 --> 00:15:27.559 That's the story, yeah. Someone lifted the formula 00:15:27.559 --> 00:15:29.759 and sold it to various component manufacturers 00:15:29.759 --> 00:15:32.759 in Taiwan to produce incredibly cheap capacitors. 00:15:33.480 --> 00:15:35.639 But the thieves messed up the theft. They really 00:15:35.639 --> 00:15:38.159 did. They essentially stole the recipe for a 00:15:38.159 --> 00:15:40.240 cake, but they forgot to write down the baking 00:15:40.240 --> 00:15:43.559 powder. They completely missed the crucial stabilizing 00:15:43.559 --> 00:15:46.240 inhibitors in the formula. And the omission of 00:15:46.240 --> 00:15:49.320 those inhibitors was disastrous. Over the next 00:15:49.320 --> 00:15:52.559 decade, millions of desktop computers, server 00:15:52.559 --> 00:15:55.279 power supplies and flat screen televisions were 00:15:55.279 --> 00:15:59.059 manufactured using these flawed, unstable components. 00:15:59.159 --> 00:16:01.779 Because they were so cheap. Right. And the water 00:16:01.779 --> 00:16:04.799 inside the capacitor slowly ate away at the aluminum 00:16:04.799 --> 00:16:08.850 anodes. constantly generating hydrogen gas, eventually 00:16:09.279 --> 00:16:11.440 The pressure just became too much. The metal 00:16:11.440 --> 00:16:14.320 cylinders would bulge. Yeah, they'd bulge ooze 00:16:14.320 --> 00:16:16.519 brown corrosive liquid onto the motherboards 00:16:16.519 --> 00:16:19.539 or just outright explode with loud popping sounds. 00:16:20.279 --> 00:16:21.960 Entire I .T. departments were suddenly dealing 00:16:21.960 --> 00:16:24.080 with thousands of dead machines. It actually 00:16:24.080 --> 00:16:26.539 spawned a massive cottage industry of computer 00:16:26.539 --> 00:16:29.419 repair shops whose only job was soldering new 00:16:29.419 --> 00:16:31.700 stable capacitors onto ruined circuit boards. 00:16:31.740 --> 00:16:33.820 It was a massive headache for everyone involved. 00:16:34.100 --> 00:16:36.220 It's incredible that a single missing chemical 00:16:36.220 --> 00:16:38.779 ingredient brought down millions of computers. 00:16:38.600 --> 00:16:41.899 But it makes me wonder, what does this mean for 00:16:41.899 --> 00:16:45.100 the devices that actually have good parts? Because 00:16:45.100 --> 00:16:48.679 even if a capacitor escapes a catastrophic, oozing 00:16:48.679 --> 00:16:52.480 explosion, it's not immortal. The source material 00:16:52.480 --> 00:16:54.480 describes them almost like they have a biological 00:16:54.480 --> 00:16:57.379 life cycle. Oh, absolutely. They age. They age. 00:16:57.500 --> 00:17:00.600 They experience a very slow, completely inevitable 00:17:00.600 --> 00:17:03.519 aging process, especially the wet aluminum types 00:17:03.519 --> 00:17:06.019 we've been discussing. Even under perfectly normal 00:17:06.019 --> 00:17:08.660 operating conditions, that liquid electrolyte 00:17:08.660 --> 00:17:11.839 slowly evaporates and dries out over time. So 00:17:11.839 --> 00:17:14.470 they basically dry out. Yes. Their entire lifespan 00:17:14.470 --> 00:17:17.210 is dictated by ambient heat governed by a principle 00:17:17.210 --> 00:17:19.930 called the 10 degree rule. This is based on the 00:17:19.930 --> 00:17:22.150 Arrhenius equation for chemical reaction rates. 00:17:22.430 --> 00:17:23.950 I read about this 10 degree rule and it sounds 00:17:23.950 --> 00:17:26.450 almost too simple to be true. Just cooling a 00:17:26.450 --> 00:17:28.910 device down drastically changes the hardware's 00:17:28.910 --> 00:17:31.650 lifespan. It's a remarkably consistent rule of 00:17:31.650 --> 00:17:34.589 thumb. For every 10 degrees Celsius drop in operating 00:17:34.589 --> 00:17:37.349 temperature, the electrolytic capacitor's lifespan 00:17:37.349 --> 00:17:41.690 literally doubles. Yeah. If an industrial capacitor 00:17:41.690 --> 00:17:44.410 is rated by the factory to last 2 ,000 hours 00:17:44.410 --> 00:17:47.829 at 105 degrees Celsius, running it at a much 00:17:47.829 --> 00:17:51.390 cooler 45 degrees extends its expected life to 00:17:51.390 --> 00:17:55.250 roughly 15 years. Heat accelerates the chemical 00:17:55.250 --> 00:17:57.369 evaporation, so keeping the environment cool 00:17:57.369 --> 00:17:59.069 is really the only way to keep the components 00:17:59.069 --> 00:18:01.829 alive. So keeping your laptop out of the hot 00:18:01.829 --> 00:18:04.390 sun actually prevents the internal chemistry 00:18:04.390 --> 00:18:07.339 from drying up. That is wild. Very true. But 00:18:07.339 --> 00:18:09.839 my absolute favorite detail from the source material 00:18:09.839 --> 00:18:12.960 is how these components handle internal physical 00:18:12.960 --> 00:18:15.480 damage. Because they have an incredible self 00:18:15.480 --> 00:18:17.859 -healing mechanism. They do, yeah. When positive 00:18:17.859 --> 00:18:20.559 voltage is applied, the chemical reaction inside 00:18:20.559 --> 00:18:23.380 actually generates new oxide. It actively seeks 00:18:23.380 --> 00:18:25.660 out weakened or cracked imperfections in the 00:18:25.660 --> 00:18:28.099 dielectric layer and patches them up on the fly. 00:18:28.859 --> 00:18:30.559 Wait, so are we saying old electronic hardware 00:18:30.559 --> 00:18:33.009 is actually like... alive enough to heal its 00:18:33.009 --> 00:18:34.950 own wounds like Wolverine? Well, I mean, it's 00:18:34.950 --> 00:18:37.210 a purely chemical process, but yeah, it's undeniably 00:18:37.210 --> 00:18:39.430 elegant. It's known as a continuous forming process. 00:18:39.490 --> 00:18:41.970 How does it work? When a microscopic crack develops 00:18:41.970 --> 00:18:44.710 in the oxide insulator, the electrical current 00:18:44.710 --> 00:18:47.170 naturally flows toward that weak point because 00:18:47.170 --> 00:18:49.730 it's the past, at least, resistance. Okay, that 00:18:49.730 --> 00:18:51.769 makes sense. And the moment that current hits 00:18:51.769 --> 00:18:54.009 the liquid electrolyte at the site of the crack, 00:18:54.549 --> 00:18:57.960 it instantly oxidizes the exposed aluminum, plugging 00:18:57.960 --> 00:19:00.500 the microscopic hole with fresh insulator. It's 00:19:00.500 --> 00:19:02.660 like a pothole filling itself with fresh asphalt 00:19:02.660 --> 00:19:06.160 the exact second a car drives over it. The mechanism 00:19:06.160 --> 00:19:09.799 itself acts as the repair. Exactly. The hardware 00:19:09.799 --> 00:19:12.539 maintains its own integrity as long as it has 00:19:12.539 --> 00:19:16.119 power. And interestingly, this self -healing 00:19:16.119 --> 00:19:19.180 chemistry is exactly why vintage electronics 00:19:19.180 --> 00:19:22.140 like an antique stereo receiver from the 1970s 00:19:22.140 --> 00:19:24.220 that has been sitting unplugged in a closet for 00:19:24.220 --> 00:19:28.119 30 years require a delicate process called preconditioning. 00:19:28.180 --> 00:19:29.980 Right, you can't just plug an antique radio into 00:19:29.980 --> 00:19:31.759 the wall, flip the switch and start listening 00:19:31.759 --> 00:19:34.359 to music. No, if you do, you risk blowing the 00:19:34.359 --> 00:19:37.019 internal components to pieces. After decades 00:19:37.019 --> 00:19:39.319 of sitting completely idle without any power, 00:19:39.880 --> 00:19:42.519 the nanometer thin oxide layer inside those wet 00:19:42.519 --> 00:19:45.640 capacitors chemically degrades and thins out. 00:19:45.940 --> 00:19:48.180 If you hit that degraded layer with full wall 00:19:48.180 --> 00:19:50.900 voltage immediately, the electrical current will 00:19:50.900 --> 00:19:52.980 punch right through the weakened dielectric, 00:19:53.339 --> 00:19:56.119 short out, and vaporize the electrolyte. Ouch. 00:19:56.490 --> 00:19:58.970 So what do you do? Restorers have to run the 00:19:58.970 --> 00:20:01.509 old radio on very low power, usually feeding 00:20:01.509 --> 00:20:04.230 it through a series resistor for an hour or more. 00:20:04.450 --> 00:20:06.730 So you have to slowly feed it a trickle of power 00:20:06.730 --> 00:20:09.670 so the chemical reaction has the time to naturally 00:20:09.670 --> 00:20:12.730 regenerate the oxide layer. Right. You are basically 00:20:12.730 --> 00:20:15.369 letting the capacitor heal itself before you 00:20:15.369 --> 00:20:17.809 demand it to handle a full electrical load. That 00:20:17.809 --> 00:20:20.470 is astonishing. It's a mandatory ritual for bringing 00:20:20.470 --> 00:20:23.349 vintage electronics back to life. You are patiently 00:20:23.349 --> 00:20:25.769 coaxing the internal chemistry back into a stable 00:20:25.769 --> 00:20:28.410 state. So we've gone from Johann Heinrich Buff 00:20:28.410 --> 00:20:31.009 messing around with one -way electrical currents 00:20:31.009 --> 00:20:35.630 in 1857 to Bell Labs baking cantalum dust to 00:20:35.630 --> 00:20:38.269 keep up with the microchip to corporate spies 00:20:38.269 --> 00:20:40.730 accidentally turning a generation of desktop 00:20:40.730 --> 00:20:45.049 computers into oozing time bombs. It really completely 00:20:45.049 --> 00:20:46.750 changes the way you look at the digital world. 00:20:47.049 --> 00:20:50.410 It's definitely not just pure clean data moving 00:20:50.410 --> 00:20:53.539 through solid silicon. No. Our devices only survive 00:20:53.539 --> 00:20:55.980 the day to day because these tiny, volatile, 00:20:56.240 --> 00:20:59.180 self -healing chemical silos are constantly managing 00:20:59.180 --> 00:21:01.259 the physics behind the scenes. And, you know, 00:21:01.460 --> 00:21:03.039 exploring that physical reality brings up one 00:21:03.039 --> 00:21:05.970 final... incredibly strange phenomenon to leave 00:21:05.970 --> 00:21:07.710 you with straight from the engineering manuals. 00:21:08.049 --> 00:21:10.750 It's a behavioral quirk of these components called 00:21:10.750 --> 00:21:13.630 dielectric absorption or soakage. Soakage, like 00:21:13.630 --> 00:21:15.869 a sponge soaking up water. Very similar, yeah. 00:21:16.029 --> 00:21:18.369 In theory, an ideal capacitor holds a charge, 00:21:18.529 --> 00:21:20.470 and when you discharge it, the voltage should 00:21:20.470 --> 00:21:23.730 drop exactly to zero and stay there. Right. But... 00:21:23.109 --> 00:21:25.569 Real -world electrolytic capacitors have a sort 00:21:25.569 --> 00:21:27.769 of physical memory. If they have been holding 00:21:27.769 --> 00:21:30.789 a high voltage for a long time, and you completely 00:21:30.789 --> 00:21:33.309 discharge them so your multimeter reads absolute 00:21:33.309 --> 00:21:36.970 zero volts, if you leave them sitting alone on 00:21:36.970 --> 00:21:40.009 a workbench, they can mysteriously develop a 00:21:40.009 --> 00:21:43.109 ghost voltage hours later. Wait, they recharge 00:21:43.109 --> 00:21:45.210 themselves while sitting completely disconnected? 00:21:45.559 --> 00:21:47.640 Where's the power coming from? It comes from 00:21:47.640 --> 00:21:50.019 the dielectric material itself. The molecules 00:21:50.019 --> 00:21:52.680 inside the insulator act like tiny internal magnets 00:21:52.680 --> 00:21:56.000 called dipoles. Under high voltage, these dipoles 00:21:56.000 --> 00:21:58.519 get physically twisted and aligned by the electrical 00:21:58.519 --> 00:22:00.740 field. When you turn the power off and empty 00:22:00.740 --> 00:22:04.220 the capacitor, the main charge is gone. But those 00:22:04.220 --> 00:22:07.549 dipoles slowly... over hours, untwist back to 00:22:07.549 --> 00:22:09.730 their resting state. And as they untwist, they 00:22:09.730 --> 00:22:11.829 release a time -delayed ghost charge back into 00:22:11.829 --> 00:22:14.230 the terminals. So the component is slowly bleeding 00:22:14.230 --> 00:22:16.829 energy. It had crapped in its own physical structure. 00:22:17.029 --> 00:22:20.009 Yes. It is such a recognized safety hazard that 00:22:20.009 --> 00:22:22.589 large industrial high -voltage capacitors actually 00:22:22.589 --> 00:22:24.650 have to be physically shipped with thick metal 00:22:24.650 --> 00:22:26.869 wires shorting their terminals together. Just 00:22:26.869 --> 00:22:29.619 to keep them empty. Exactly. This is to ensure 00:22:29.619 --> 00:22:31.740 that a capacitor that was completely emptied 00:22:31.740 --> 00:22:34.220 at the factory doesn't slowly build up enough 00:22:34.220 --> 00:22:36.980 ghost voltage during shipping to give a technician 00:22:36.980 --> 00:22:40.319 a lethal shock when they open the box. Wow. That 00:22:40.319 --> 00:22:42.900 is the perfect thought to end on. The next time 00:22:42.900 --> 00:22:45.019 you power down your laptop or turn off a piece 00:22:45.019 --> 00:22:47.740 of hardware, just remember the device isn't truly 00:22:47.740 --> 00:22:50.919 dead the second the screen goes black. Somewhere 00:22:50.919 --> 00:22:54.640 inside, a tiny metal cylinder is slowly untwisting, 00:22:54.700 --> 00:22:56.880 holding onto a ghost of the energy it used to 00:22:56.880 --> 00:22:59.299 carry, quietly maintaining the memory of the 00:22:59.299 --> 00:23:01.400 current. Thank you for joining us on this deep 00:23:01.400 --> 00:23:02.660 dive, and we'll catch you on the next one.