Comment by ChuckMcM | Hacker News Reader

Comment by ChuckMcM | Hacker News Reader

Parent Full thread

Having read through a lot of battery breakthrough reports, I tend to zero in on the 'hard bit' which in this particular case is here: U-M researchers solved this problem with chemical and mechanical treatments that provide a pristine surface for lithium to plate evenly, effectively suppressing the formation of dendrites or filaments.

My reading of that was if you create this pristine surface, you get a wonder battery. On the other hand, if the surface has a defect in it anywhere you get dendrites, and the subsequent shorts they cause.

When I read that part I said "Ok, so this will make great bespoke batteries that the military can use for their human carried gear, but they will be too expensive to compete with the existing lithium ion technology."

I really hope that I'm wrong here, and there is some process that can reliably make massive numbers of pristine plated lithium / ceramic nodes but until we see that I don't see this breakthrough making it out of the lab.

Is “pristine ceramic” that much harder to create than e.g. “pristine silicon”? Cilicon chips seem relatively cheap these days.

That is an excellent restatement of the core question. Can the process that makes "pristine ceramic" be scaled. (harder is a dependent variable on scaling difficulty.)

I don't know how much harder it is to make ceramics pristine. I am inferring that it is the thing in the article that wasn't already part of making batteries so the 'new' bit. And after seeing many, many breakthroughs where the 'new' bit was doable in small scales at great expense in a laboratory, but impossible to do at scale in an automated process, I've watched those breakthroughs go from 'world changing' to 'laboratory curiosity.'

So when Nature Materials publishes the paper on the process to repeatedly make tons of pristine ceramic material, then I will be excited for the new batteries that my devices will surely get in an upgrade.

That's interesting perspective.

> And after seeing many, many breakthroughs where the 'new' bit was doable in small scales at great expense in a laboratory, but impossible to do at scale in an automated process,

Can you elaborate on that, maybe give a few examples? I always thought that inventions were mainly judged by their merit (efficiency), I guess I was blind to the "scaling" (i.e. "actual industry") variable, and just assumed that there must be some way to automate and streamline almost everything, as long as the demand is there (as Elon Musk has recently been showing, with batteries/electric cars and rockets).

I have found that many times when people are "talking past" each other it is because there is a belief in there somewhere that one holds one way and the other holds a different way. In this case you have illuminated it nicely, which (slightly edited) is this:

I ... assumed that there must be some way to automate and streamline almost everything, as long as the demand is there.

I have experienced many things over my life where this assumption was invalid. The most common reason I have seen that this assumption breaks down is that "demand" is expressly tied to "price." And more specifically when the price goes up, demand goes down (in the economic sense, the number of people willing to pay that price for a widget) and at some price the demand goes to zero. Thus "not possible" which is not literally true, it is only actually true.

Do a search on "Battery breakthrough" (Google Scholar will give you better results but the web search works too) and pick any one where the improvement is >= 50% improvement in any axis (cost, charging, capacity, weight, what ever). This works best for articles that are 5 or more years old (you have to allow some time for them to come up with a way to do it), and then try to find that breakthrough used in any battery today. Sometimes you will find a 'why we couldn't do it' article but mostly it will just not be there. Even though batteries that were 50% better on any axis would seem to have a market right? And then read through to find the 'hard bit' in the break through and go back to Google Scholar and see if you can find papers related to that bit. You will find "the rest of the story" as it is sometimes said where some thing prevents wider adoption than laboratory curiosity.

Here is a simpler technology, silicon anodes (announced in 2010) reviewed in 2018 [1], maybe hitting the market if they can go to production [2]. This is a simpler process and as it has matured its impact has gone down to 10 - 15%.

[1] https://sci-hub.tw/10.1016/j.nanoen.2016.07.023

[2] [BMW] plans to incorporate Sila Nano batteries into vehicles by 2023 to boost energy capacity by 10 to 15 percent. --https://venturebeat.com/2018/08/16/sila-nanotechnologies-rai...

> The most common reason I have seen that this assumption breaks down is that "demand" is expressly tied to "price."

I knew this, but reading it again in this context makes many business decisions very clear.

>I knew this, but reading it again in this context makes many business decisions very clear.

Yeah. You can get this very clear in your mind by considering a few examples. Here's one.

How much demand is there for legal, genuine, clean $2 bills that are legal currency and have no meaningful restrictions (won't get you in trouble, nothing special about it)?

Barring anything I can think of, just your personal demand for such bills should be at least a billion of them... at the price point of $1.99 each. (For a profit of 1 billion * $0.01 = $10 million USD. There are some logistics involved, according to Google a U.S. dollar bill weighs about a gram, so a billion grams weigh in at around 1102 tons. So that is approximately $9074 per ton. But for $10 million you'll figure the logistics out.)

And now compare your personal demand for them at $2.00. That's probably about 0 in number. Why would you want even one of them, they're less useful than the $2 in other currency you bought it for.

So just your personal demand for a legal $2 bill is either a billion of them or 0 of them, depending on whether we're talking about the price point of $2.00 or $1.99.

Demand is at a given price point.

Right, but that the demand/supply curve isn't really my point. (Maybe we're still "talking past" each other.) Nobody wants to buy 2GB of memory at a "640 kB ought to be enough for anybody" price point; that's fairly obvious. My point is, human civilisation has proven time and again, especially when it comes to micro/nano-electrical/chemical technology (batteries, transistors, screens, chips, ...), that it is possible to optimize, improve, reduce price and make more efficient, production (while making the product better) beyond what anyone thought possible (well except Gordon Moore).

Why would batteries be different? If I find some way to make a battery better, 50% more efficient and/or 50% lighter, why can't I just dump $1-10bn into research/production and then dominate the market (likely worth 10-100x more)? Is there a fundamental technological reason (for some technologies/inventions at least), or is it just a matter of economics/management (too risky)?

I'm guessing this comment reads fairly unclear (my thoughts aren't particularly clear on this matter either), but hopefully it gets some of my point across.

Edit: petra gives a good example below - scaling of biological processes often fails; which makes sense because biology - we don't understand how life works, and it changes/evolves/moves by itself! But it seems that non-biological processes should be manageable, at least at this scale (i.e. not quantum/high-energy processes).

> (Maybe we're still "talking past" each other.)

Yes and no, I understand your point. If I can restate it for you, your point is this "Anything that is invented can be improved with a variable amount of effort."

And I don't disagree with that point at all. Where we part ways is that I also believe that there are inventions that cannot be improved enough to actually be worth producing. They are "dead ends" in the parlance when the amount of effort needed to produce the thing (even with the improved process) is insufficient to make that thing worth doing.

My interest in batteries and battery chemistry comes from my interest in mobile robotics (from roomba sized to battlebot sized). So I've been following the market pretty closely since 1986 or so when I joined the Homebrew Robotics Club in the Bay Area. I have watched many battery "breakthroughs" go no where. And it isn't because the people didn't want them to go no where, they lost a lot of money that way[1].

I will grant you that these companies "improved" what they had done in the lab, but they never improved enough to actually be able to sell batteries and keep the company going. Which is what I am defining as "not possible."

If you would like to explore this topic further there is even a book on it [3]. Salient comment from that article "While countless breakthroughs have been announced over the last decade, time and again these advances failed to translate into commercial batteries."

[1] For instance, only 36 battery startups received more than $500,000 in funding since 2000, and of these only two returned more money than was invested into them --https://spectrum.ieee.org/energywise/consumer-electronics/po...

[2] Company: Aquion Energy

Select VC investors: Bill Gates, CapX Partners, Constellation Technology Ventures

Total disclosed funding: $196.6M

Company CEO, Scott Pearson, commented: “Creating a new electrochemistry and an associated battery platform at commercial scale is extremely complex, time-consuming, and very capital intensive. Despite our best efforts to fund the company and continue to fuel our growth, the Company has been unable to raise the growth capital needed to continue operating as a going concern.” -- https://www.cbinsights.com/research/biggest-startup-failures...

[3] https://www.technologyreview.com/s/534866/why-we-dont-have-b...

Ok, I think I get it... mostly.

I'm reading a book about rockets right now ("Ignition"). It actually contains many funny stories about inventions and dead ends; how some substance seemed very promising, but then it turned out it corrodes steel very slowly; or how they abandoned a better fuel for a slightly worse one that was more practical (e.g. wouldn't freeze in the winter). I guess in general, we only hear/see the success stories, i.e. technology that is better, and could be put into production and scaled. As a software developer I might be a bit blind to that ;)

Chemistry(and biology) is one such field: the chemical/biological process behaves very differently whether it's don't in small quantities of the lab or in the huge volumes required for a commercial effort.

So think of that large($100mm) factory as a research process , that can often fail.

Example: commercializing biotech often fails because it's not possible to control contamination and genetic evolution of the bacteria you grow in huge processes .

> and just assumed that there must be some way to automate and streamline almost everything, as long as the demand is there

Scaling a solution to the travelling salesman problem would solve many other problems. The demand is definitely there. But many people believe it cannot be done.

That's a fundamentally different problem. "Scaling" it requires figuring out how to solve a different concrete problem every time. Scaling e.g. battery production requires figuring out how to solve the same real-world problem very efficiently.

No, you could (possibly) solve TSP once by providing an algorithm that scales well just like inventing an industrial process. My point was that it looks very much like some problems are fundamentally difficult and have no scalable solutions.

But most silicon isn't pristine, that's why we have chip binning.

Here'a post a made on reddit a few years ago about this very problem. I think its pretty relevant here and just about anywhere else that does cutting edge research:

"I want to expand on this because it is a point so often overlooked with all these "holy shit we just created a new tech that is 10000% better" articles. I'm gonna use a loose analogy to try and explain it.

Say you made a perfect cupcake. It is insanely fucking delicious, and will revolutionize the idea of cupcakes. You spent years researching the art of cupcakes, and now have written down the perfect combination of all cupcake factors to make this divine cupcake. Your milk must come from a special dairy who set aside the .02% of the milk that meets your needs, your flour must come from one certain granary using a special indoor grown grain, the eggs from your own 3 specially bred super pampered chickens. The mixing takes place over eight unique specialized mixing procedures, with varying steps in between. The cupcake paper is made from the wood of an endangered Guatemalan tree (requiring federal clearance to obtain 10 sheets of it) and the baking tray from a steel alloy made by one company in Japan, and it's not reusable. The baking process has been tailored to the unique characteristics of your $20,000 1'x1'x1' oven. It takes 3 hours to bake with precise temperature control and positioning adjustment needed throughout the process. The icing also requires it's own set of unique ingredients and precise procedures. To make one of these cupcakes takes 14 hours and costs $2600. If any corners are cut in the process the whole thing falls apart. But done right it is indeed perfect.

A business man finds out about the cupcake that makes people cry tears of joy. He comes to you and needs you to make 35,000 of them a day to sell for $10 each in order to be profitable. Logistically this is simply impossible. No equipment on earth exists that can follow such precise steps for such large quantities, much less a company capable of making it. Your oven costs 20 grand and can bake one cupcake in 3 hours, you need 5000 in 3 hours. Your ingredients come from sources that produce only a minute fraction of what you would need. Never mind that they are scattered around the country making same day delivery impossible short of having your own air fleet. Or that there are only 3 chickens on earth that produce the right eggs. Or that the rare beetle poop in the icing comes from an entomology professor in Peru. Even if you could do all this you still need to adjust the recipe to account for being handled by machines and conveyors. And packaging. And shipping. Oh and slash the cost by a factor of 260.

Since this is all obviously impossible at the time (and for the foreseeable future), you instead just submit the recipe to a food journal and let cooking magazines and TV shows have a field day with sensational articles about the coming cupcake jesus. You become well known and grant money for your legendary lasagna project pours in. There never was any intention to bring this to market, it was just a proof of concept with the added benefit of getting your name out there. It's up to someone else down the road to pick it up and figure out mass production.

If anything people should get more hyped about advancements in production techniques. That's when you will actually get new shit in your hands."

This happened to me in real life. I made a circuit board by hand in the 1980's called the Amiga 14. Basicly if you had the board and a 16 MHz 68000 it double the speed of your Amiga 1000.

I demo it at the World of Commodore in Toronto (I was a member of the Amiga Developer Group that had a booth) the following week I go a call from a gentleman, he wanted to buy 5,000 of them if I made them cheap enough.

But when I designed the circuit I just wanted something to speed up my computer, I never planned a design for mass production. I tried for weeks to change the design for mass production but I could not do it.

Today I know how to get the job done, but back then I did not have the knowledge to mass produce anything I designed.

That wasn't an impossible task back then, you could've partnered with a good EE and got it shipped.

Here's the original reddit comment: https://np.reddit.com/r/technology/comments/2bvpzq/new_recha...

A bit off topic, but your talk of industrially-produced perfect cupcakes made me remember this old story about just such a thing:

http://aliendovecote.com/uploads/twine/sauna.html

Ok, that is just perfect. Thanks.

Yes, but can this sort of lengthy, complicated comment be posted every time one of these pie in the sky technology articles comes out?

Doubtful. :-)

"Let's say you found the recipe for the perfect cupcake, but you need an egg so perfect that only one per day is laid in the whole world. No matter how much demand there is, you can't scale up production, since you depend on that single egg."

This reminds me of an article I once read about perfect fruit sold as gifts in Japan. 12 strawberries to a box for about $82. But it is an industry!

https://www.bbc.com/news/world-radio-and-tv-17352173

Except what you have here is a completely contrived example that shows you how wrong it is in the ridiculous lengths you go to try to make your point.

> the eggs from your own 3 specially bred super pampered chickens

I mean, sure, whatever.

Reply on news.ycombinator.com