The Donut Lab Solid-State Battery Situation is Crazy — Transcript

Full transcript

1. 0:00In case you haven't heard, a company
2. 0:02called Donut Lab claimed to have solved
3. 0:04a trillion dollar problem with their
4. 0:06miracle battery. It promises to charge
5. 0:09fast, store huge amounts of energy,
6. 0:12require no critical materials, and last
7. 0:15literal generations, all whilst being
8. 0:18cheaper than anything on the market. If
9. 0:21you think that sounds too good to be
10. 0:22true, then you're not alone. And over
11. 0:25the last 10 months since the
12. 0:27announcement, Donut Lab haven't done
13. 0:29much to settle the debates. In fact, my
14. 0:32last video on Donut Lab outlined how
15. 0:35data from their previous test reports
16. 0:37match commercially available lithium ion
17. 0:40cells. And now, in a strange twist of
18. 0:44events, they've started testing a
19. 0:46completely different set of battery
20. 0:48cells, and they can't even prove that
21. 0:50those are rechargeable. So, let's see
22. 0:53how Donut Lab responded to my last
23. 0:55video, the outcome of some recent legal
24. 0:58proceedings, and why their new test data
25. 1:01is even more questionable than before.
26. 1:04I'm Ryan MᶜNellis, and this is a Zero to
27. 1:06Deep Dive.
28. 1:10Let's start by quickly going over Donut
29. 1:12Lab's response to my last video, because
30. 1:14they made some really good technical
31. 1:16counterpoints to my claims.
32. 1:19Oh, wait. No, they didn't. They instead
33. 1:21went after me as an individual and
34. 1:23didn't address a single one of my
35. 1:25points. Like the voltage curve matching
36. 1:28that of a lithium ion cell, or the cell
37. 1:30expanding in the same way a lithium ion
38. 1:33cell with a graphite anode would expand.
39. 1:36Publicly, they've said that I'm ignoring
40. 1:38the facts, but my points were based on
41. 1:41their own test data and peer-reviewed
42. 1:43publications. I will say that a
43. 1:46highlight for me in Donut Lab's response
44. 1:48was this meme posted by a friend of
45. 1:50their CEO, which is truly terrifying.
46. 1:53So, instead of addressing any of my
47. 1:55concerns, Donut Lab have just said that
48. 1:58my criticism comes from having worked
49. 2:00with their competitors, including CATL,
50. 2:04BYD, Panasonic, and BMW. I have worked
51. 2:07with these companies and will hopefully
52. 2:09work with them again because they give
53. 2:11me real test data, show me real
54. 2:13demonstrations, and have patents that
55. 2:16back up their products that they even
56. 2:18ship with warranties.
57. 2:20>> [music]
58. 2:20>> And either way, none of that changes
59. 2:22anything about the points I made in my
60. 2:24previous video. None of those points
61. 2:27depend on who I've worked with, and
62. 2:29anyone can check the data for
63. 2:31themselves. One thing that does concern
64. 2:34me, though, is what I see as an attempt
65. 2:36by Donut Lab to silence its critics
66. 2:38through the court system via a lawsuit
67. 2:41brought about by its subsidiary Nordic
68. 2:43Nano.
69. 2:45Nordic Nano Group, which is now owned by
70. 2:47Donut Lab, has just lost much of a legal
71. 2:50battle against Lauri Peltola, who is one
72. 2:53of the co-founders. I've got the court
73. 2:55documents here directly from the court,
74. 2:57though I can't share them publicly. It
75. 2:59turns out that after Lauri started
76. 3:01talking to the press, Nordic Nano Group
77. 3:03went to the Helsinki District Court
78. 3:05asking for an order to stop him from
79. 3:08publishing any of their confidential
80. 3:09information or the information of their
81. 3:12partners and suppliers backed with fines
82. 3:15of up to 100,000 euros. The court
83. 3:17actually granted an emergency but
84. 3:20temporary gagging order in May to stop
85. 3:22Lauri from talking to the press.
86. 3:24However, after hearing his side of the
87. 3:26story, they lifted that order and
88. 3:28refused the rest of Nordic Nano Group's
89. 3:30demands to stop Lauri talking to the
90. 3:32press, finding them too vague to
91. 3:34enforce. Nordic Nano Group did win back
92. 3:37Lauri's work laptop and phone, and the
93. 3:39court rejected Lauri's claims that the
94. 3:42case was an abusive lawsuit.
95. 3:44>> [music]
96. 3:44>> Interestingly, though, Nordic Nano Group
97. 3:46wanted the whole case itself to be kept
98. 3:49secret. It initially asked for the
99. 3:51entire application, including the
100. 3:54request to stop Laurie publishing any
101. 3:56more, to be sealed for 25 years.
102. 3:59That raises a question.
103. 4:02If Donut Lab says Laurie was never in on
104. 4:05its battery team and has no up-to-date
105. 4:07knowledge of the technology, then
106. 4:09[music] why did its own subsidiary go to
107. 4:12court to stop him speaking publicly? If
108. 4:15their public message is true, in which
109. 4:17they say that Laurie knows nothing, then
110. 4:20the aim of the court case couldn't be to
111. 4:23protect genuine secrets, because he
112. 4:25wouldn't have had any. It seems to me
113. 4:28more about keeping the fraud allegations
114. 4:30quiet, especially because they wanted
115. 4:32the whole trial to be kept secret, too.
116. 4:35I will say there are still ongoing court
117. 4:37proceedings relating to this, but it
118. 4:39doesn't look good for Donut Lab and the
119. 4:42team. Okay, so that's the legal stuff
120. 4:44out of the way, and now let's move on to
121. 4:46the new data published by Donut Lab. And
122. 4:49honestly, I've lost count of all the
123. 4:51different cells Donut Lab have now
124. 4:53tested. So, let's look at that after a
125. 4:56quick message from today's sponsor,
126. 4:58Mammoth AI. Instead of spending $20 on a
127. 5:01long list of different AI subscriptions
128. 5:04like OpenAI, Gemini, and Grok, Mammoth
129. 5:07brings them all together in one low-cost
130. 5:09subscription, saving you time and money.
131. 5:13The user interface is really clean, and
132. 5:14a feature I like is the re-prompting,
133. 5:17which allows you to run the same prompt
134. 5:19on multiple AI models, meaning I can
135. 5:22quickly compare a response or image from
136. 5:25one model to the next, making sure I
137. 5:27always get the best results, as there
138. 5:30really is no single perfect model. As a
139. 5:33European company, they also follow
140. 5:35strict privacy laws such as GDPR, making
141. 5:38sure that the data you share is secure
142. 5:40and never stored or used for AI
143. 5:42training. And you have full control to
144. 5:45delete your history at any time. Also,
145. 5:47if you have a task you do regularly,
146. 5:49like improving the clarity of an
147. 5:51explanation in this example, you can
148. 5:54create a custom Mammoth that's tuned to
149. 5:56nail the response every time. Check out
150. 5:59Mammoth today at mammoth.ai or use our
151. 6:02link in the description or the on-screen
152. 6:04QR code. Now, let's look at the other
153. 6:07new cells Donut Lab are testing and
154. 6:09apologies if you're losing track of all
155. 6:11of them, I am too.
156. 6:13Firstly, there was the one that fast
157. 6:15charged and then got damaged after being
158. 6:17exposed to high temperatures, puffing
159. 6:19and losing over a half its usable
160. 6:21capacity. They've since introduced
161. 6:23another cell which is apparently version
162. 6:261.5.
163. 6:27Donut Lab have now admitted that the
164. 6:29first version couldn't actually reach
165. 6:32their claimed 400 watt hours per
166. 6:33kilogram energy density, but this new
167. 6:36version has been verified at 409 watt
168. 6:40hours per kilogram with no other tests
169. 6:42performed. This energy density is in
170. 6:45line with commercially available high
171. 6:47energy lithium ion cells. They also
172. 6:49claim that there's a new new version
173. 6:52being released soon which I think is
174. 6:53version two and that will also hit some
175. 6:56of their other claims like fast
176. 6:58charging. This strongly suggests the
177. 7:01cell tested at 409 watt hours per
178. 7:03kilogram is not able to fast charge, but
179. 7:06the new new one will be able to. This is
180. 7:09convenient timing when other companies
181. 7:11like Amprius and BEI have recently
182. 7:14released high energy density cells that
183. 7:16can also fast charge, both based on
184. 7:19lithium ion chemistries. However, for
185. 7:21me, the strangest part relates to a new
186. 7:24set of cells that Donut Lab have started
187. 7:27testing. And yes, there are now more
188. 7:29cells which are completely different in
189. 7:32both function and design to the other
190. 7:35ones they've tested.
191. 7:36The claim is that these new cells use a
192. 7:39bipolar stack design and from
193. 7:42third-party test data, it's clear that
194. 7:45there are indeed stacks of anode and
195. 7:47cathode layers, and the cell produces a
196. 7:50voltage. But, there's more to this than
197. 7:52meets the eye. The idea of bipolar cells
198. 7:55is actually a well-researched field. In
199. 7:58a bipolar cell, a shared conductive
200. 8:01plate has anode material on one face and
201. 8:05cathode material on the other. If you
202. 8:07stack these plates together, the voltage
203. 8:10adds up in series, all neatly inside one
204. 8:13sealed cell. This means you can save
205. 8:16space compared to having lots of
206. 8:19individual cells stacked together.
207. 8:21Despite the possible advantages, the
208. 8:23cells have some challenges, and a lot of
209. 8:25them relate to lifespan. Meaning, the
210. 8:28cell won't be able to charge or
211. 8:30discharge many times before it becomes
212. 8:33unusable. But, Donut Lab have done one
213. 8:36better than this.
214. 8:37Their cells don't appear to be able to
215. 8:40charge or discharge at all. Yes, you
216. 8:44understood that right. It appears the
217. 8:46new cells from Donut Lab aren't even
218. 8:49rechargeable batteries. The test report
219. 8:52from Intertek said this, "Intertek have
220. 8:55not verified that the device under test
221. 8:58is a functioning battery cell which can
222. 9:00be electrically discharged and charged."
223. 9:03The Donut Lab CEO also actively refused
224. 9:07to comment on if the cells could be
225. 9:09recharged, and the company never gave me
226. 9:11a response.
227. 9:12I did track down another company
228. 9:14involved called Future Motive by seeing
229. 9:17who commissioned the latest Donut Lab
230. 9:18reports. And when I phoned them up, they
231. 9:21were also unable to confirm whether the
232. 9:23latest cells were rechargeable. The
233. 9:26reason it's important that the cell
234. 9:28doesn't appear rechargeable is because
235. 9:30non-rechargeable bipolar cells are
236. 9:33nothing new. In fact, the earliest
237. 9:36battery, Volta's pile from around 1800,
238. 9:40was a bipolar stack. I've literally got
239. 9:43a bipolar stack here. It's got layers of
240. 9:46copper and zinc, and then felt in
241. 9:48between it, and if you add an
242. 9:49electrolyte like salt water, it will
243. 9:52conduct and work as a bipolar stack.
244. 9:55Each pair gives around 0.8 volts, so if
245. 9:58I was to stack five of them together,
246. 10:00I'd get about 4 volts.
247. 10:02Um so, I think maybe I have a
248. 10:04billion-dollar battery company. Like
249. 10:06with all electrochemical systems, the
250. 10:09voltage of my bipolar cells are
251. 10:11determined by the materials. If you
252. 10:13change the materials, you change the
253. 10:15voltage.
254. 10:17Which is how we know that the latest
255. 10:19bipolar cell from Donut Lab uses
256. 10:21different materials to the high-energy
257. 10:23density or fast-charging ones they've
258. 10:25tested before. The voltages we've seen
259. 10:28are 12.3 volts, 7.8 volts, 5 volts, and
260. 10:333.3 volts. If these are stacks of layers
261. 10:36in series, then we can assume they are
262. 10:39eight, five, three, and two layers,
263. 10:41which give 1.54 volts, 1.56 volts, 1.67
264. 10:47volts, and 1.65 volts per layer for each
265. 10:51of these battery cells. I picked these
266. 10:53number of layers as they are the whole
267. 10:55numbers that give a consistent result,
268. 10:58so do treat this as a hypothesis.
269. 11:00But that voltage range we're seeing of
270. 11:031.5 to 1.65
271. 11:06volts per layer is typical of a
272. 11:08zinc-based chemistry, like a cell using
273. 11:11zinc and manganese dioxide.
274. 11:14One using lithium in each layer would
275. 11:16sit much higher, probably around 3 volts
276. 11:19or more.
277. 11:20This graph shows the electrode
278. 11:22potentials for different material
279. 11:24combinations, where the difference in
280. 11:26voltage between the anode and cathode
281. 11:28material dictates the cell voltage. You
282. 11:31could pair different materials with
283. 11:33lithium to get lower voltages. But, as
284. 11:36the bipolar cell was shown to have no
285. 11:37lithium, and its voltage matches with a
286. 11:41zinc-based cell, that would be my best
287. 11:43guess. This is funny to me because
288. 11:46you're probably more familiar than you
289. 11:48think with non-rechargeable
290. 11:50zinc-manganese-based cells. As they are
291. 11:53the conventional AA batteries that you
292. 11:56can find in your TV remote. Guess I'm
293. 11:58sat on another billion-dollar idea.
294. 12:01It's not impossible that the cell
295. 12:03they're testing is, in fact,
296. 12:05rechargeable. They do exist. But, it's a
297. 12:09wildly different cell to the others they
298. 12:11have tested. This brings up the
299. 12:13question, if you have a cell that you
300. 12:16say doesn't contain any lithium, why
301. 12:18would you test a different cell for its
302. 12:21lithium content? This also makes the
303. 12:24nail penetration and chemical analysis
304. 12:26test that Donut Lab performed pretty
305. 12:29useless, as they didn't do it on the
306. 12:31high-energy-density cell that they
307. 12:33tested, but instead the small,
308. 12:36non-rechargeable bipolar prototype.
309. 12:38Donut Lab didn't need to do a chemical
310. 12:41analysis on this prototype cell to show
311. 12:43it had no lithium or sodium. The
312. 12:45voltages already suggested that.
313. 12:48Though, I would be far more interested
314. 12:50if it didn't contain any zinc. As a side
315. 12:52note, non-rechargeable zinc-based
316. 12:54batteries are actually really cool. You
317. 12:57can even buy commercial zinc-air
318. 13:00batteries that reach around 400 to 500
319. 13:03watt-hours per kilogram, and they're
320. 13:05commonly used in hearing aids, though
321. 13:07they are single use. I feel like a
322. 13:09broken record, but the same questions
323. 13:12remain 10 months after the original
324. 13:15announcement. Where is the cell that can
325. 13:18fast charge, hold 400 watt-hours per
326. 13:20kilogram, and last 100,000 cycles, all
327. 13:24without lithium? Instead of running the
328. 13:26tests that critics have asked for, they
329. 13:29keep pulling new cells out of nowhere
330. 13:31and testing those instead.
331. 13:33I'm pretty confident that this will be
332. 13:35the last video that I make on Donut app,
333. 13:37but people wanted to know what had
334. 13:39happened since I made my last video, so
335. 13:41there it is.
336. 13:43They've been testing a range of
337. 13:44different cells, losing lawsuits, and
338. 13:47making memes.
339. 13:48So, if you've enjoyed this video, please
340. 13:50consider subscribing. It's free and
341. 13:52helps the channel a lot. You might also
342. 13:54like some of our other videos, like this
343. 13:56one on Tesla's new electric motor.
344. 13:59And as always,
345. 14:01thanks for watching.