WEBVTT 00:00:00.000 --> 00:00:03.200 Right now, sitting inside the device you are 00:00:03.200 --> 00:00:06.900 using to listen to this is a microscopic chemical 00:00:06.900 --> 00:00:09.240 pressure vessel. Yeah, it's basically a tiny 00:00:09.240 --> 00:00:13.460 liquid -filled metal can that is slowly and inevitably 00:00:13.460 --> 00:00:16.339 evaporating. Exactly. And if it gets too hot, 00:00:16.780 --> 00:00:19.399 it dies faster. Plus, if you were to somehow 00:00:19.399 --> 00:00:21.480 open your device and plug that tiny cylinder 00:00:21.480 --> 00:00:24.140 in backward, it would literally explode. Which 00:00:24.140 --> 00:00:26.879 is wild to think about. It really is. So today 00:00:26.879 --> 00:00:29.820 we are cracking open the hidden, highly volatile 00:00:29.820 --> 00:00:32.299 world of the electrolytic capacitor. It's just 00:00:32.299 --> 00:00:34.859 entirely invisible to most people, yet our entire 00:00:34.859 --> 00:00:37.740 modern infrastructure is completely dependent 00:00:37.740 --> 00:00:39.820 on the electrochemistry happening inside those 00:00:39.820 --> 00:00:42.280 tiny metal cans. Totally. And our mission today 00:00:42.280 --> 00:00:44.659 is to figure out why. Our source material for 00:00:44.659 --> 00:00:47.679 this deep dive is a massive, incredibly comprehensive 00:00:47.679 --> 00:00:50.439 Wikipedia article simply titled Electrolytic 00:00:50.439 --> 00:00:53.939 Capacitors. It's dense. But there is so much 00:00:53.939 --> 00:00:56.100 good stuff in there. Oh for sure. Okay. Let's 00:00:56.100 --> 00:00:58.380 unpack this Why should you the listener care 00:00:58.380 --> 00:01:00.719 about a tiny cylinder soldered to a motherboard? 00:01:00.820 --> 00:01:03.600 Well without them nothing works, right? These 00:01:03.600 --> 00:01:06.219 components are the sole reason our gadgets can 00:01:06.219 --> 00:01:08.560 store the energy needed to be so incredibly small 00:01:08.560 --> 00:01:12.159 and fast. But, as we will see, they are also 00:01:12.159 --> 00:01:14.859 delicate chemical bombs that have literally caused 00:01:14.859 --> 00:01:17.400 global tech meltdowns. What's fascinating here 00:01:17.400 --> 00:01:20.159 is that this isn't just a story of modern, cutting 00:01:20.159 --> 00:01:22.579 -edge tech. I mean, the foundational science 00:01:22.579 --> 00:01:24.840 for this component stretches all the way back 00:01:24.840 --> 00:01:27.760 to the 1850s. Wait, the 1850s? Yeah, exactly. 00:01:28.060 --> 00:01:30.620 We are relying on the bizarre, almost magical 00:01:30.620 --> 00:01:33.420 quarks of a few specific metal - discovered back 00:01:33.420 --> 00:01:35.379 then to keep the internet running today. That 00:01:35.379 --> 00:01:39.359 is insane. But before we get to the explosions 00:01:39.359 --> 00:01:42.099 and the 1850s science experiments, we need to 00:01:42.099 --> 00:01:44.540 understand the physical reality of what this 00:01:44.540 --> 00:01:47.019 thing actually is. Right, like how do you pack 00:01:47.019 --> 00:01:49.079 so much electricity into such a ridiculously 00:01:49.079 --> 00:01:51.859 tiny space? Exactly. At its most basic level, 00:01:51.959 --> 00:01:54.120 a capacitor's job is just to store electrical 00:01:54.120 --> 00:01:56.799 energy. It is a polarized component, meaning 00:01:56.799 --> 00:01:59.299 it has a positive side, the anode, and a negative 00:01:59.299 --> 00:02:02.409 side, the cathode. And between those two sides 00:02:02.409 --> 00:02:03.810 you have the dielectric and kind of think of 00:02:03.810 --> 00:02:06.790 the dielectric like a microscopic impenetrable 00:02:06.790 --> 00:02:09.349 rubber wall separating two powerful magnets. 00:02:09.889 --> 00:02:11.550 Okay I like that visual. Yeah so the positive 00:02:11.550 --> 00:02:13.750 and negative charges desperately want to reach 00:02:13.750 --> 00:02:16.969 each other but the wall stops them. So the energy 00:02:16.969 --> 00:02:19.789 just sits there stored in that tension waiting 00:02:19.789 --> 00:02:21.770 to be released when the circuit needs it. And 00:02:21.770 --> 00:02:25.139 in a standard old school capacitor you might 00:02:25.139 --> 00:02:28.020 use like a physical sheet of ceramic or plastic 00:02:28.020 --> 00:02:31.240 film to act as that wall between the positive 00:02:31.240 --> 00:02:34.080 and negative plates. Exactly. But an electrolytic 00:02:34.080 --> 00:02:36.439 capacitor throws the plastic sheet out the window. 00:02:36.639 --> 00:02:39.060 Right. The positive plate is made of a very specific 00:02:39.060 --> 00:02:42.099 type of metal. When the manufacturer dunks that 00:02:42.099 --> 00:02:44.659 metal into a chemical bath and applies a voltage, 00:02:45.080 --> 00:02:47.860 the metal literally grows a microscopic layer 00:02:47.860 --> 00:02:50.460 of rust over its surface. An oxide layer, technically. 00:02:50.560 --> 00:02:52.979 Yeah, an oxide layer. And that layer of rust 00:02:52.979 --> 00:02:55.659 basically becomes the impenetrable rubber wall. 00:02:55.780 --> 00:02:58.620 It becomes the insulator. And because it is grown 00:02:58.620 --> 00:03:01.759 chemically through a controlled reaction, That 00:03:01.759 --> 00:03:05.099 oxide layer is unbelievably thin. We are talking 00:03:05.099 --> 00:03:07.439 nanometers. Which is mind -blowing. It really 00:03:07.439 --> 00:03:09.300 is. And the thinner the wall, the stronger the 00:03:09.300 --> 00:03:10.819 attraction between the positive and negative 00:03:10.819 --> 00:03:13.460 charges and the more energy you can store. But 00:03:13.460 --> 00:03:16.639 the real secret to their massive power capacity 00:03:16.639 --> 00:03:19.159 is the surface area of the metal itself, right? 00:03:19.379 --> 00:03:23.520 Yes, exactly. The volumetric efficiency or CV 00:03:23.520 --> 00:03:26.300 product relies heavily on that surface area. 00:03:26.419 --> 00:03:29.919 I want you to picture a standard flat sheet of 00:03:29.919 --> 00:03:32.819 printer paper. If you lay it on a desk, it takes 00:03:32.819 --> 00:03:35.439 up a specific footprint. Right, like eight and 00:03:35.439 --> 00:03:38.419 a half by 11 inches. Yeah. Take that same piece 00:03:38.419 --> 00:03:41.180 of paper and tightly crumple it up into a tiny 00:03:41.180 --> 00:03:43.639 ball. That crumpled ball takes up a fraction 00:03:43.639 --> 00:03:46.699 of the space, but the actual physical surface 00:03:46.699 --> 00:03:49.080 area of the paper hasn't changed. Which is a 00:03:49.080 --> 00:03:51.939 great analogy because manufacturers do exactly 00:03:51.939 --> 00:03:54.340 this to the metal inside these capacitors. They 00:03:54.340 --> 00:03:56.780 microstop it, crumple it. Basically, yeah. Through 00:03:56.780 --> 00:03:59.259 chemical processes called electrochemical etching 00:03:59.259 --> 00:04:01.900 or sintering, they eat away at the flat metal 00:04:01.900 --> 00:04:04.379 until the surface is incredibly rough and porous. 00:04:04.520 --> 00:04:06.580 So if you look at it under an electron microscope, 00:04:06.680 --> 00:04:09.639 it looks like a jagged microscopic sponge. Exactly. 00:04:09.900 --> 00:04:12.039 And by microscopic crumpling the metal like that, 00:04:12.259 --> 00:04:15.020 they can fit up to 200 times more surface area 00:04:15.020 --> 00:04:19.639 into the exact same volume. Wow. 200 times. So 00:04:19.639 --> 00:04:21.800 in the world of energy storage, more surface 00:04:21.800 --> 00:04:24.879 area combined with an incredibly thin insulator 00:04:24.879 --> 00:04:27.860 equals an enormous capacity to hold an electrical 00:04:27.860 --> 00:04:32.019 charge. Right. But to make that crumpled surface 00:04:32.019 --> 00:04:35.199 actually work presents a massive engineering 00:04:35.199 --> 00:04:37.399 hurdle. Because it's so jagged. Yeah, you have 00:04:37.399 --> 00:04:41.319 this jagged microscopic mountain range of oxidized 00:04:41.319 --> 00:04:44.100 metal acting as your positive plate. You need 00:04:44.100 --> 00:04:47.149 to get the negative plate. the cathode, to perfectly 00:04:47.149 --> 00:04:50.170 touch every single microscopic nook, cranny, 00:04:50.290 --> 00:04:52.649 and valley of that rough sponge. Which you obviously 00:04:52.649 --> 00:04:54.569 can do by just pressing another flat piece of 00:04:54.569 --> 00:04:56.310 solid metal against it. No, it would only touch 00:04:56.310 --> 00:04:58.230 the highest peaks and completely miss all the 00:04:58.230 --> 00:05:00.089 valleys. So the engineering work around here 00:05:00.089 --> 00:05:02.329 is just brilliant. The cathode isn't a solid 00:05:02.329 --> 00:05:04.290 piece of metal at all. It is the electrolyte 00:05:04.290 --> 00:05:06.490 itself. Yeah, it's a conductive liquid, a gel, 00:05:06.610 --> 00:05:08.930 or even a solid polymer that flows entirely into 00:05:08.930 --> 00:05:12.529 the sponge, perfectly molds to that complex microscopic 00:05:12.529 --> 00:05:15.120 surface. So the liquid physically becomes the 00:05:15.120 --> 00:05:17.920 negative half of the component. Exactly. And 00:05:17.920 --> 00:05:20.120 there's another deeply elegant detail about the 00:05:20.120 --> 00:05:23.079 manufacturing process. The voltage the factory 00:05:23.079 --> 00:05:25.879 applies during that initial chemical bath literally 00:05:25.879 --> 00:05:28.339 dictates the thickness of the oxide layer. Wait, 00:05:28.339 --> 00:05:31.259 really? It's that precise? Yeah. If they want 00:05:31.259 --> 00:05:33.220 a capacitor rated for 10 volts, they apply a 00:05:33.220 --> 00:05:36.339 little over 10 volts. And the rust grows exactly 00:05:36.339 --> 00:05:38.800 to the thickness needed to hold back that specific 00:05:38.800 --> 00:05:41.259 pressure. That is so cool. They basically custom 00:05:41.259 --> 00:05:44.399 grow the insulation atom by atom. They do. But 00:05:44.399 --> 00:05:46.439 you can't just use any random metal you have 00:05:46.439 --> 00:05:49.079 lying around to achieve that perfect uniform 00:05:49.079 --> 00:05:51.660 layer of rust. Right. You need specific materials 00:05:51.660 --> 00:05:54.379 that behave in very strange ways under electrical 00:05:54.379 --> 00:05:57.620 pressure. Enter the valve metals. Yeah. Early 00:05:57.620 --> 00:05:59.839 researchers coined the term valve metals because 00:05:59.839 --> 00:06:02.360 these specific elements form an oxide layer that 00:06:02.360 --> 00:06:05.120 behaves physically like a one -way gate for electrons. 00:06:05.500 --> 00:06:07.699 Meaning it blocks electrical current from flowing 00:06:07.699 --> 00:06:10.220 in one direction, but allows it to flow freely 00:06:10.220 --> 00:06:12.759 in the other. Exactly. Just like a physical valve 00:06:12.759 --> 00:06:15.250 in a plumbing system. And the three main families 00:06:15.250 --> 00:06:18.269 we utilize for this are aluminum, tantalum, and 00:06:18.269 --> 00:06:20.189 niobium. Which brings us back to the history. 00:06:20.790 --> 00:06:22.889 The source notes that a German physicist named 00:06:22.889 --> 00:06:26.790 Johann Buff first documented this weird one -way 00:06:26.790 --> 00:06:30.410 valve quark in aluminum way back in 1857. Yeah, 00:06:30.550 --> 00:06:34.730 1857. And then, fast forward to 1896, and Charles 00:06:34.730 --> 00:06:37.009 Pollack realizes that this oxide layer stays 00:06:37.009 --> 00:06:39.430 perfectly stable even after you turn the power 00:06:39.430 --> 00:06:41.910 off. So he patents the first liquid capacitor 00:06:41.910 --> 00:06:44.610 based on this chemistry. Right. And that patent 00:06:44.610 --> 00:06:46.790 basically laid the groundwork for the very first 00:06:46.790 --> 00:06:48.910 industrial applications. Which were pretty clunky 00:06:48.910 --> 00:06:51.470 at first. In the early 1900s, early telephone 00:06:51.470 --> 00:06:54.509 exchanges were plagued by massive audio noise 00:06:54.509 --> 00:06:56.629 issues on their power lines. Oh, the static must 00:06:56.629 --> 00:06:58.610 have been terrible. Yeah. So to filter out that 00:06:58.610 --> 00:07:01.259 electrical static, engineers built the very first 00:07:01.259 --> 00:07:04.399 wet aluminum electrolytic capacitors, but calling 00:07:04.399 --> 00:07:06.939 them components feels like a huge stretch. Oh 00:07:06.939 --> 00:07:08.860 yeah, they were massive. We're talking about 00:07:08.860 --> 00:07:12.079 literal jars or heavy metallic boxes filled with 00:07:12.079 --> 00:07:14.620 a sloshing solution of borax and water. With 00:07:14.620 --> 00:07:17.220 just a folded aluminum plate dropped inside, 00:07:17.680 --> 00:07:20.019 imagine trying to run a telecommunications network 00:07:20.019 --> 00:07:22.500 where your power filters are essentially giant 00:07:22.500 --> 00:07:25.459 open buckets of chemical water. It's honestly 00:07:25.459 --> 00:07:28.259 hilarious. They were heavy. prone to spilling 00:07:28.259 --> 00:07:30.800 and entirely impractical for anything moving 00:07:30.800 --> 00:07:34.120 or consumer facing. Right. So the shift to modern 00:07:34.120 --> 00:07:36.379 electronics required getting rid of the sloshing 00:07:36.379 --> 00:07:40.839 jar. In 1925, an inventor named Samuel Rubin 00:07:40.839 --> 00:07:43.379 teamed up with Philip Mallory, who was the founder 00:07:43.379 --> 00:07:45.420 of the company that would eventually become Duracell, 00:07:45.579 --> 00:07:49.319 actually. Exactly. And Rubin invented the dry 00:07:49.319 --> 00:07:53.040 aluminum capacitor, though dry is a bit of a 00:07:53.040 --> 00:07:55.180 relative term here. True. It isn't bone dry. 00:07:55.529 --> 00:07:59.069 Rubin basically took two long thin strips of 00:07:59.069 --> 00:08:01.689 aluminum foil and separated them with a highly 00:08:01.689 --> 00:08:04.480 porous paper spacer. Then he soaked that paper 00:08:04.480 --> 00:08:07.120 in a thick, non -aqueous chemical paste, which 00:08:07.120 --> 00:08:08.959 is the electrolyte. Yeah. And then he rolled 00:08:08.959 --> 00:08:10.899 the whole sandwich up tightly into a small cylinder 00:08:10.899 --> 00:08:13.220 and sealed it in a metal can. Suddenly, you have 00:08:13.220 --> 00:08:15.399 a compact stacked component that doesn't leak 00:08:15.399 --> 00:08:17.639 when you tilt it. And that specific invention 00:08:17.639 --> 00:08:20.779 is what made early, affordable home radios possible. 00:08:21.259 --> 00:08:23.500 But the real evolutionary leap happened in the 00:08:23.500 --> 00:08:26.519 1950s, entirely driven by the transistor revolution. 00:08:26.680 --> 00:08:28.500 Right. Because William Shockley and the team 00:08:28.500 --> 00:08:30.819 at Bell Labs had just invented the transistor. 00:08:31.050 --> 00:08:33.769 And transistors were microscopic compared to 00:08:33.769 --> 00:08:36.690 the old vacuum tubes they replaced. But the aluminum 00:08:36.690 --> 00:08:38.909 capacitors needed to support those transistors 00:08:38.909 --> 00:08:41.480 were still comparatively bulky. So Bell Labs 00:08:41.480 --> 00:08:43.480 had to figure out a way to shrink the energy 00:08:43.480 --> 00:08:46.399 storage to match the tiny new switches. They 00:08:46.399 --> 00:08:48.980 abandoned aluminum and turned to tantalum. Right. 00:08:49.240 --> 00:08:52.759 In 1950, researchers took pure tantalum powder, 00:08:53.279 --> 00:08:55.519 pressed it into a tiny pill shape, and baked 00:08:55.519 --> 00:08:58.320 it at roughly 2000 degrees Celsius in a vacuum. 00:08:58.539 --> 00:09:00.179 Which is that sintering process we mentioned 00:09:00.179 --> 00:09:03.879 earlier. It fused the powder into a highly porous 00:09:03.879 --> 00:09:06.600 solid metal sponge that was incredibly dense 00:09:06.600 --> 00:09:08.539 with surface area. But they still had the liquid 00:09:08.539 --> 00:09:10.259 problem, didn't they? They had this amazing They 00:09:10.259 --> 00:09:12.279 were using new solid sponge, but they were still 00:09:12.279 --> 00:09:14.440 soaking it in liquid electrolytes. Yeah, which 00:09:14.440 --> 00:09:17.779 wasn't ideal for the new era of pristine solid 00:09:17.779 --> 00:09:21.679 -state electronics. But by 1952, Bell Labs cracked 00:09:21.679 --> 00:09:24.559 it. They figured out how to use manganese dioxide 00:09:24.559 --> 00:09:27.440 as a completely solid electrolyte. Wow. So they 00:09:27.440 --> 00:09:29.840 created a capacitor with no liquid paste, no 00:09:29.840 --> 00:09:32.559 evaporation, and insanely high reliability. All 00:09:32.559 --> 00:09:34.759 shrunk down to a fraction of the size of an aluminum 00:09:34.759 --> 00:09:37.039 can. Okay, I need to stop here because logic 00:09:37.039 --> 00:09:39.960 dictates that if Bell Labs figured out how to 00:09:39.960 --> 00:09:43.200 make a leak -proof, highly conductive, incredibly 00:09:43.200 --> 00:09:47.379 reliable, completely solid capacitor in the 1950s... 00:09:47.379 --> 00:09:49.139 The liquid -filled aluminum one should have gone 00:09:49.139 --> 00:09:52.059 extinct. Yes, exactly. Yet the source material 00:09:52.059 --> 00:09:54.840 states that today, the vast majority of consumer 00:09:54.840 --> 00:09:59.039 electronics still rely on wet, non -solid aluminum 00:09:59.039 --> 00:10:02.419 electrolytic capacitors. Why are we still putting 00:10:02.419 --> 00:10:05.240 liquid paste inside our modern flat screen TVs? 00:10:05.740 --> 00:10:08.220 If we connect this to the bigger picture, engineering 00:10:08.220 --> 00:10:10.379 is always a brutal compromise between physics 00:10:10.379 --> 00:10:13.039 and economics. First off, aluminum is one of 00:10:13.039 --> 00:10:15.480 the most abundant elements on earth. It is incredibly 00:10:15.480 --> 00:10:18.000 cheap. Whereas tantalum is rare, right? Yeah. 00:10:18.080 --> 00:10:20.659 Its extraction is complex and its global market 00:10:20.659 --> 00:10:22.740 price is highly volatile. So we use aluminum 00:10:22.740 --> 00:10:24.600 because it keeps the cost of a laptop charger 00:10:24.600 --> 00:10:27.000 under 50 bucks. Makes sense. But cost is just 00:10:27.000 --> 00:10:29.740 the primary driver. Wet aluminum capacitors also 00:10:29.740 --> 00:10:32.259 possess a massive technological advantage when 00:10:32.259 --> 00:10:34.759 dealing with the messy, unpredictable reality 00:10:34.759 --> 00:10:37.500 of city power grids. They can handle power surges 00:10:37.500 --> 00:10:41.720 better. Exactly. They are remarkably resilient 00:10:41.720 --> 00:10:45.220 to sudden voltage spikes. If a rogue surge of 00:10:45.220 --> 00:10:48.259 electricity hits a wet aluminum capacitor, it 00:10:48.259 --> 00:10:50.460 acts a bit like a pressure relief valve on a 00:10:50.460 --> 00:10:53.610 steam pipe. Oh, so it briefly lets a little Current 00:10:53.610 --> 00:10:56.090 leak through the oxide layer safely absorbing 00:10:56.090 --> 00:10:58.309 the blow and then the chemistry seals itself 00:10:58.309 --> 00:11:02.009 right back up Yes, whereas the solid tantalum. 00:11:02.429 --> 00:11:04.470 Let me guess they don't handle surge as well. 00:11:04.769 --> 00:11:08.129 Not at all Solid tantalum capacitors are incredibly 00:11:08.129 --> 00:11:10.830 reliable under perfect conditions, but they have 00:11:10.830 --> 00:11:13.809 zero tolerance for electrical abuse Yicks if 00:11:13.809 --> 00:11:16.149 they get hit with a sudden high -energy voltage 00:11:16.149 --> 00:11:19.690 spike the solid dielectric shatters. It creates 00:11:19.690 --> 00:11:22.210 a dead short circuit And because the solid manganese 00:11:22.210 --> 00:11:24.809 dioxide electrolyte contains oxygen, the massive 00:11:24.809 --> 00:11:26.669 heat from the short circuit can literally cause 00:11:26.669 --> 00:11:28.789 the component to ignite and aggressively burn 00:11:28.789 --> 00:11:30.789 right on the circuit board. So we essentially 00:11:30.789 --> 00:11:33.309 trade the risk of a literal miniature fire for 00:11:33.309 --> 00:11:35.230 a cheaper component that can handle a bit of 00:11:35.230 --> 00:11:37.269 rough electrical weather. Basically, yeah. But 00:11:37.269 --> 00:11:40.370 because we are forced by economics to put these 00:11:40.370 --> 00:11:43.250 wet aluminum capacitors into nearly every device 00:11:43.250 --> 00:11:46.350 we own, we have to deal with their fatal flaw. 00:11:46.669 --> 00:11:49.450 The liquid is trying to escape. They really are 00:11:49.450 --> 00:11:52.639 ticking clocks. To understand why they inevitably 00:11:52.639 --> 00:11:55.120 die, we have to look at how they operate in a 00:11:55.120 --> 00:11:56.580 circuit. Because they aren't just sitting there 00:11:56.580 --> 00:11:59.299 holding a static charge, right? Right. They are 00:11:59.299 --> 00:12:02.240 constantly, rapidly charging and discharging 00:12:02.240 --> 00:12:05.220 to smooth out the flow of electricity. This alternating 00:12:05.220 --> 00:12:07.639 current flowing in and out is called ripple current. 00:12:07.740 --> 00:12:09.919 But nothing in the physical world is perfectly 00:12:09.919 --> 00:12:12.779 efficient. The materials inside the capacitor 00:12:12.779 --> 00:12:15.320 have a natural resistance to that constant flow 00:12:15.320 --> 00:12:17.799 of electricity. Engineers call this equivalent 00:12:17.799 --> 00:12:21.600 series resistance, or ESR. And when the ripple 00:12:21.600 --> 00:12:24.139 current pushes through the materials, that internal 00:12:24.139 --> 00:12:27.139 resistance generates physical friction. And friction 00:12:27.139 --> 00:12:29.879 generates heat. Which is the absolute enemy of 00:12:29.879 --> 00:12:31.779 the liquid electrolyte. Think of it like the 00:12:31.779 --> 00:12:34.019 oil in your car engine. The ripple current is 00:12:34.019 --> 00:12:37.240 the friction of the moving Tistons. The ESR determines 00:12:37.240 --> 00:12:40.320 how severe that friction is. And the heat generated 00:12:40.320 --> 00:12:43.700 by that friction slowly, relentlessly boils off 00:12:43.700 --> 00:12:46.720 the engine oil. Exactly. Inside the capacitor, 00:12:47.159 --> 00:12:50.179 the liquid electrolyte is the oil. As it gets 00:12:50.179 --> 00:12:53.100 hot, the liquid turns into a gas and slowly pushes 00:12:53.100 --> 00:12:55.480 its way out through the rubber seals at the bottom 00:12:55.480 --> 00:12:58.360 of the tiny metal can. And as that liquid evaporates, 00:12:58.799 --> 00:13:00.919 the internal resistance of the capacitor actually 00:13:00.919 --> 00:13:03.879 goes up. So higher resistance creates even more 00:13:03.879 --> 00:13:07.059 heat, which evaporates even more liquid. It is 00:13:07.059 --> 00:13:10.399 a vicious, accelerating cycle. Yep. Eventually 00:13:10.399 --> 00:13:13.419 the capacitor simply dries out, loses its ability 00:13:13.419 --> 00:13:16.259 to store energy, and the device it powers refuses 00:13:16.259 --> 00:13:19.259 to turn on. You know, this makes me think of 00:13:19.259 --> 00:13:21.419 how companies like Apple will sometimes design 00:13:21.419 --> 00:13:24.679 incredibly thin laptops with minimal venting. 00:13:24.799 --> 00:13:27.220 Oh, basically gluing the aluminum chassis shut. 00:13:27.299 --> 00:13:29.850 Yeah. They prioritize silence and sleekness, 00:13:29.870 --> 00:13:31.789 but they are trapping the internal heat. They 00:13:31.789 --> 00:13:33.750 aren't just making the computer run warm. They 00:13:33.750 --> 00:13:35.950 are literally evaporating the lifespan of their 00:13:35.950 --> 00:13:38.149 own internal component. Which is absolutely true. 00:13:38.450 --> 00:13:41.330 Engineers actually map this life cycle out mathematically 00:13:41.330 --> 00:13:43.970 using the bathtub curve. The bathtub curve? Yeah. 00:13:44.169 --> 00:13:46.370 If you graph the failure rate of millions of 00:13:46.370 --> 00:13:48.169 components, it looks like a bathtub. You have 00:13:48.169 --> 00:13:50.490 a steep drop at the very beginning. Those are 00:13:50.490 --> 00:13:52.629 the units that fail immediately due to factory 00:13:52.629 --> 00:13:55.929 defects. Okay, then what? Then, the curve flattens 00:13:55.929 --> 00:13:59.049 out into a long, stable plateau. That is the 00:13:59.049 --> 00:14:02.289 reliable working life of the device. But eventually, 00:14:02.529 --> 00:14:05.350 with wet aluminum capacitors, you reach the other 00:14:05.350 --> 00:14:07.649 end of the bathtub. Where the curve spikes sharply 00:14:07.649 --> 00:14:10.509 upward again. Exactly. That is the unavoidable 00:14:10.509 --> 00:14:13.210 wear -out phase, where the liquid has simply 00:14:13.210 --> 00:14:15.450 evaporated away. And the speed at which you hit 00:14:15.450 --> 00:14:18.230 that wall is governed by a physical law called 00:14:18.230 --> 00:14:20.879 the Arrhenius rule. which engineers casually 00:14:20.879 --> 00:14:23.580 refer to as the 10 degree rule. It is a stark 00:14:23.580 --> 00:14:26.419 reminder of how heat accelerates chemical reactions. 00:14:26.639 --> 00:14:29.200 I mean, heat is just atomic motion. So the hotter 00:14:29.200 --> 00:14:31.879 the liquid inside the capacitor gets, the faster 00:14:31.879 --> 00:14:34.259 its molecules vibrate and the faster they push 00:14:34.259 --> 00:14:36.559 their way past the rubber seals. Right. The rule 00:14:36.559 --> 00:14:38.740 states that for every 10 degrees Celsius you 00:14:38.740 --> 00:14:41.080 drop the operating temperature, the lifetime 00:14:41.080 --> 00:14:44.669 of the capacitor literally doubles. That is massive. 00:14:45.090 --> 00:14:48.009 So if a capacitor is rated to survive for 2 ,000 00:14:48.009 --> 00:14:51.850 hours at a boiling 105 degrees Celsius and you 00:14:51.850 --> 00:14:54.029 instead run it at a comfortable well -ventilated 00:14:54.029 --> 00:14:56.429 45 degrees Celsius, you double that lifespan 00:14:56.429 --> 00:15:00.330 six times over. That measly 2 ,000 hours mathematically 00:15:00.330 --> 00:15:04.090 scales to over 128 ,000 hours. You can take a 00:15:04.090 --> 00:15:06.090 component destined to die in a few months and 00:15:06.090 --> 00:15:08.929 make it last 15 years purely by keeping it cool. 00:15:09.200 --> 00:15:11.620 It perfectly demonstrates why thermal management 00:15:11.620 --> 00:15:14.440 is the most critical aspect of modern electronics 00:15:14.440 --> 00:15:17.769 design. Interestingly, there is also a truly 00:15:17.769 --> 00:15:20.970 bizarre chemical quirk to how these oxide layers 00:15:20.970 --> 00:15:23.750 age when they aren't subjected to heat. You mean 00:15:23.750 --> 00:15:25.509 when they are just sitting completely unused? 00:15:25.710 --> 00:15:27.909 Yeah. Oh, the self -healing phenomenon. Exactly. 00:15:28.429 --> 00:15:30.730 If you take a wet aluminum capacitor and leave 00:15:30.730 --> 00:15:32.750 it sitting in a drawer for five years with no 00:15:32.750 --> 00:15:34.970 power running through it, the chemical aggressiveness 00:15:34.970 --> 00:15:37.470 of the electrolyte paste will slowly start to 00:15:37.470 --> 00:15:40.070 eat away at its own delicate oxide layer. So 00:15:40.070 --> 00:15:42.529 the internal rubber wall starts to rot and thin 00:15:42.529 --> 00:15:44.190 out. Right. And if you were to pull it out of 00:15:44.190 --> 00:15:45.919 that drawer, and immediately slam it with full 00:15:45.919 --> 00:15:48.399 power, that weakened wall would shatter, causing 00:15:48.399 --> 00:15:50.860 a short circuit. But if you apply a positive 00:15:50.860 --> 00:15:53.960 voltage very, very slowly, like just a gentle 00:15:53.960 --> 00:15:56.259 trickle of electricity. The capacitor will use 00:15:56.259 --> 00:15:59.120 that current to trigger the original electrochemical 00:15:59.120 --> 00:16:02.240 reaction. The electricity physically pulls the 00:16:02.240 --> 00:16:05.059 oxygen atoms out of the liquid and bonds them 00:16:05.059 --> 00:16:07.779 to the bare aluminum. Regrowing the oxide molecules 00:16:07.779 --> 00:16:09.860 and patching the microscopic cracks in the wall 00:16:09.860 --> 00:16:11.759 while it sits right there in the circuit. Yeah. 00:16:12.090 --> 00:16:15.450 The electricity literally rebuilds the rust barrier. 00:16:15.710 --> 00:16:17.690 It is a self -repairing mechanism built into 00:16:17.690 --> 00:16:19.470 the physics of the metal. That's incredible. 00:16:19.669 --> 00:16:23.529 It is. But as we established early on, they don't 00:16:23.529 --> 00:16:25.830 always quietly dry out and they don't always 00:16:25.830 --> 00:16:29.120 peacefully heal themselves. Sometimes the pressure 00:16:29.120 --> 00:16:32.159 vessel violently fails. Because we cannot ignore 00:16:32.159 --> 00:16:34.700 the physical reality that these are sealed metal 00:16:34.700 --> 00:16:37.379 cans containing highly reactive chemistry. So 00:16:37.379 --> 00:16:39.820 what actually triggers a catastrophic failure? 00:16:40.100 --> 00:16:42.080 Gas. It always comes down to gas. Like from a 00:16:42.080 --> 00:16:44.419 power surge. Yeah. For instance, if a power surge 00:16:44.419 --> 00:16:47.259 subjects the capacitor to severe overvoltage, 00:16:47.480 --> 00:16:49.419 pushing vastly more electrical pressure into 00:16:49.419 --> 00:16:51.379 it than the thin oxide layer was going to handle, 00:16:51.980 --> 00:16:55.059 the current spikes, generating massive instantaneous 00:16:55.059 --> 00:16:58.259 heat. And then the liquid electrolyte rapidly 00:16:58.259 --> 00:17:00.860 decomposes into hydrogen gas. Exactly. Hydrogen 00:17:00.860 --> 00:17:04.160 gas rapidly expanding inside a sealed rigid metal 00:17:04.160 --> 00:17:07.299 cylinder. That sounds bad. It is. The internal 00:17:07.299 --> 00:17:09.700 pressure builds exponentially until the aluminum 00:17:09.700 --> 00:17:13.819 casing simply bursts open. The same exact physical 00:17:13.819 --> 00:17:16.279 mechanism happens with thermal runaway, where 00:17:16.279 --> 00:17:18.759 the ripple current generates heat faster than 00:17:18.759 --> 00:17:21.059 the surrounding air can cool it. But the absolute 00:17:21.059 --> 00:17:23.339 fastest way to detonate one of these is reverse 00:17:23.339 --> 00:17:26.299 polarity, right? Absolutely. Because, like we 00:17:26.299 --> 00:17:29.240 established, valve metals act as a one -way physical 00:17:29.240 --> 00:17:32.420 gate for electrons. The oxide layer is literally 00:17:32.420 --> 00:17:35.180 designed to block current flowing in one specific 00:17:35.180 --> 00:17:38.279 direction. So if a manufacturer accidentally 00:17:38.279 --> 00:17:41.539 solders a capacitor onto a board backward, connecting 00:17:41.539 --> 00:17:43.519 the positive voltage to the negative terminal, 00:17:43.779 --> 00:17:46.000 the electricity violently attacks the oxide layer. 00:17:46.160 --> 00:17:48.900 It rips the microscopic wall apart almost instantly. 00:17:49.039 --> 00:17:52.000 And with the barrier gone, you get an unrestricted 00:17:52.000 --> 00:17:54.440 massive flow of current directly through the 00:17:54.440 --> 00:17:57.220 liquid electrolyte. It boils into gas in milliseconds, 00:17:57.980 --> 00:18:00.740 the internal pressure spikes violently, and the 00:18:00.740 --> 00:18:04.710 casing detonates. It often fires the metal cannon 00:18:04.710 --> 00:18:07.690 into the air like a bullet and sprays boiling 00:18:07.690 --> 00:18:10.470 chemical fluid across the entire circuit board. 00:18:10.710 --> 00:18:13.670 That is why every single electrolytic capacitor 00:18:13.670 --> 00:18:16.650 has a massive brightly colored warning stripe 00:18:16.650 --> 00:18:18.769 pinned down its side pointing to the negative 00:18:18.769 --> 00:18:20.650 leg. Yeah, you really don't want to get that 00:18:20.650 --> 00:18:23.109 backward. But sometimes the explosion has nothing 00:18:23.109 --> 00:18:25.910 to do with heat or voltage spikes or someone 00:18:25.910 --> 00:18:28.970 plugging it in backward. Sometimes the component 00:18:28.970 --> 00:18:31.569 is doomed from the moment the chemicals are mixed. 00:18:32.250 --> 00:18:34.630 Here's where it gets really interesting. The 00:18:34.630 --> 00:18:37.549 source material details a massive decade -long 00:18:37.549 --> 00:18:40.150 global crisis known as the capacitor plague. 00:18:40.509 --> 00:18:43.630 It is a profound case study in how a single microscopic 00:18:43.630 --> 00:18:46.450 error in chemistry can scale up to bring down 00:18:46.450 --> 00:18:49.490 global infrastructure. So in the late 1990s, 00:18:49.849 --> 00:18:51.769 researchers in Japan, specifically at a manufacturer 00:18:51.769 --> 00:18:53.849 called Rubicon, were trying to solve the internal 00:18:53.849 --> 00:18:55.710 resistance problem. They wanted to lower the 00:18:55.710 --> 00:18:57.789 ESR to reduce the friction and heat. And they 00:18:57.789 --> 00:19:00.170 discovered that using a highly purified water 00:19:00.170 --> 00:19:02.670 -based electrolyte was incredibly effective. 00:19:03.089 --> 00:19:05.910 Because water is remarkably cheap, it flows perfectly 00:19:05.910 --> 00:19:09.430 into the microscopic metal sponge, and it dramatically 00:19:09.430 --> 00:19:11.750 improved the electrical conductivity. But water 00:19:11.750 --> 00:19:14.980 poses a severe chemical threat to aluminum. Left 00:19:14.980 --> 00:19:17.740 to its own devices, water will react aggressively 00:19:17.740 --> 00:19:21.140 with the aluminum anode, producing aluminum hydroxide 00:19:21.140 --> 00:19:24.519 and massive amounts of hydrogen gas. Which means 00:19:24.519 --> 00:19:27.539 gas buildup. So to prevent the capacitor from 00:19:27.539 --> 00:19:30.019 instantly destroying itself, the scientists had 00:19:30.019 --> 00:19:32.880 to formulate highly complex, closely guarded 00:19:32.880 --> 00:19:35.400 chemical stabilizers. Basically inhibitors that 00:19:35.400 --> 00:19:37.920 stop the water and aluminum from reacting. Right. 00:19:38.089 --> 00:19:41.589 But then, through corporate espionage, a rogue 00:19:41.589 --> 00:19:44.069 scientist stole the baseline recipe for this 00:19:44.069 --> 00:19:46.990 highly conductive, water -based electrolyte. 00:19:47.069 --> 00:19:50.029 Sounds like a movie plot. It totally does. The 00:19:50.029 --> 00:19:52.390 thieves took the stolen formula and sold it to 00:19:52.390 --> 00:19:55.390 massive component manufacturing plants in Taiwan. 00:19:56.009 --> 00:19:58.410 These factories immediately began pumping out 00:19:58.410 --> 00:20:01.109 millions of these incredibly cheap, high -performance 00:20:01.109 --> 00:20:03.430 capacitors, selling them to every major tech 00:20:03.430 --> 00:20:05.849 brand on the planet. But the thieves made a catastrophic 00:20:05.849 --> 00:20:08.279 oversight. The stolen formula was incomplete. 00:20:08.799 --> 00:20:11.279 Exactly. They managed to copy the water base, 00:20:11.779 --> 00:20:14.259 but they completely missed the proprietary chemical 00:20:14.259 --> 00:20:17.539 stabilizers. They were essentially building tiny 00:20:17.539 --> 00:20:20.400 sealed bombs. This raises an important question 00:20:20.400 --> 00:20:23.079 about the extreme fragility and interconnectedness 00:20:23.079 --> 00:20:25.420 of our global supply chains. It really does, 00:20:25.500 --> 00:20:29.119 because that incomplete, highly volatile chemical 00:20:29.119 --> 00:20:32.140 formula was replicated and soldered onto millions 00:20:32.140 --> 00:20:34.940 of motherboards, power supplies, and the network 00:20:34.940 --> 00:20:37.480 switches worldwide. And when consumers brought 00:20:37.480 --> 00:20:40.000 these computers home and turned them on, the 00:20:40.000 --> 00:20:42.940 unstabilized water inside the capacitors immediately 00:20:42.940 --> 00:20:45.180 began reacting with the aluminum. But it was 00:20:45.180 --> 00:20:47.460 a slow burn. It took about a year or two for 00:20:47.460 --> 00:20:49.539 the hydrogen gas pressure to build up inside 00:20:49.539 --> 00:20:52.519 the cans. And then, around the early 2000s, the 00:20:52.519 --> 00:20:54.980 tech world started popping. Literally. Dell, 00:20:55.240 --> 00:20:58.099 HP, Apple networking giants. No one was immune 00:20:58.099 --> 00:21:00.279 because they all unknowingly sourced their base 00:21:00.279 --> 00:21:02.480 components from the same compromised factories. 00:21:03.299 --> 00:21:05.779 Thousands of network servers crashed. Millions 00:21:05.779 --> 00:21:08.339 of home computers simply refused to boot. Motherboards 00:21:08.339 --> 00:21:10.759 were found covered in hardened, leaked electrolyte 00:21:10.759 --> 00:21:12.900 fluid, with the tops of the capacitors blown 00:21:12.900 --> 00:21:15.619 completely open. It perfectly illustrates how 00:21:15.619 --> 00:21:18.480 deeply reliant our macro -level digital society 00:21:18.480 --> 00:21:22.380 is on invisible, highly precise microscopic chemistry. 00:21:23.180 --> 00:21:25.660 One missing ingredient in a chemical paste brought 00:21:25.660 --> 00:21:28.319 the hardware world to its knees. Yeah, and it 00:21:28.319 --> 00:21:31.140 created an entire cottage industry of technicians 00:21:31.140 --> 00:21:34.160 who did nothing but painstakingly solder new, 00:21:34.579 --> 00:21:37.519 properly stabilized capacitors onto dead motherboards 00:21:37.519 --> 00:21:39.680 just to resurrect them. So what does this all 00:21:39.680 --> 00:21:41.640 mean for you listening right now? I want you 00:21:41.640 --> 00:21:44.039 to just look around the room you are in. Look 00:21:44.039 --> 00:21:46.500 at your TV screen, your laptop charger. the power 00:21:46.500 --> 00:21:48.420 brick for your phone, the router blinking in 00:21:48.420 --> 00:21:51.059 the corner. Exactly. Every single one of those 00:21:51.059 --> 00:21:53.920 devices relies on the weird physics of valve 00:21:53.920 --> 00:21:56.660 metals and perfectly balanced electrochemistry. 00:21:56.859 --> 00:21:59.759 It is a direct unbroken lineage of scientific 00:21:59.759 --> 00:22:02.859 iteration from Johann Buff's 1850s experiments 00:22:02.859 --> 00:22:06.140 to Samuel Rubin's paper rolls to Bell Labs right 00:22:06.140 --> 00:22:08.380 to the gadget sitting on your desk. And every 00:22:08.380 --> 00:22:10.460 single one of those components is currently subject 00:22:10.460 --> 00:22:12.980 to the relentless laws of physics. They are all 00:22:12.980 --> 00:22:14.900 generating tiny amounts of internal friction. 00:22:15.279 --> 00:22:17.599 producing heat, and slowly evaporating their 00:22:17.599 --> 00:22:19.839 liquid lifespans. So I want to leave you with 00:22:19.839 --> 00:22:22.079 a final thought to ponder that builds on everything 00:22:22.079 --> 00:22:25.319 we've explored today. We now know that these 00:22:25.319 --> 00:22:28.200 wet capacitors have a strict heat -dependent 00:22:28.200 --> 00:22:31.079 lifetime. Right, they have an eventual mathematically 00:22:31.079 --> 00:22:33.880 unavoidable wear -out phase at the end of that 00:22:33.880 --> 00:22:37.140 bathtub curve. Exactly. And right now, the entire 00:22:37.140 --> 00:22:39.920 world is aggressively transitioning to green 00:22:39.920 --> 00:22:41.859 energy infrastructure. We are filling our homes 00:22:41.859 --> 00:22:44.599 with smart meters, our driveways with electric 00:22:44.599 --> 00:22:47.180 vehicle chargers, and our global power grids 00:22:47.180 --> 00:22:49.859 with millions of massive solar inverters and 00:22:49.859 --> 00:22:53.059 wind turbine controllers. We are deploying billions. 00:22:53.039 --> 00:22:56.839 of new highly complex electronic devices globally 00:22:56.839 --> 00:23:00.619 in a very condensed, rapid window of time. Devices 00:23:00.619 --> 00:23:03.539 that universally rely on massive banks of electrolytic 00:23:03.539 --> 00:23:05.960 capacitors to manage that heavy electrical load. 00:23:06.460 --> 00:23:08.680 So what happens in 15 or 20 years when all those 00:23:08.680 --> 00:23:11.160 microscopic chemical clocks deployed at the exact 00:23:11.160 --> 00:23:13.339 same time hit the steep end of their bathtub 00:23:13.339 --> 00:23:15.640 curve simultaneously? Food for thought. Thank 00:23:15.640 --> 00:23:17.880 you for joining us on this deep dive. Till next 00:23:17.880 --> 00:23:18.059 time.