The battery materials field needs to take open data seriously

Two years ago I discovered an attempt at fabrication and misrepresentation of data in a manuscript which had been given to me to read ahead of its planned submission. (I want to make clear at the outset that no staff or student of Uppsala University had any role in perpetrating this fraud.)

One of the corresponding authors of the manuscript asked me to read and give my opinion on it, since I had more experience on the specific topic (related to materials for a rechargeable battery). From what I knew about the work previously I suspected the electrochemical measurement (battery test) data as being a little "too good", even though it looked plausible. I had access to the raw data, so I analysed it myself. My instincts were correct, but I was staggered by just how "too good" the results were. The electrodes being made were not even close to the specification claimed in the experimental section, with capacities actually an order of magnitude lower than claimed. Data points were moved or removed to reduce the appearance of capacity fade. Data points were even added to make the tests look like they had run more cycles than they actually had. And there was no reproducibility in the experiments – the few tests presented were more or less the only ones that even worked.

Essentially, the results could not have been more misleading if they had been fabricated in their entirety. I raised this immediately with the author who had sent me the manuscript in the first place, and it soon became clear this was deliberate misconduct on the part of the first author, a PhD student. After that the matter was out of my hands – the student was at a different institution in a different country. The manuscript, of course, was now not going to be submitted. Had that data been published, it would almost certainly have got through peer review without suspicion.

So why am I writing about this now? Because I discovered recently that the work is now published in a journal of high repute in the battery materials field. It's changed a bit from the manuscript I read – all the battery test data is new, and there are some other new experiments. But it is, to a large extent, the same work with the same first author – now at a different institution – with a lot of the same data and text. On top of that, the original corresponding authors are not co-authors anymore.

What the original authors do about this is now for them to decide, but this leaves me wondering what I should do with this knowledge. None of the data I could prove was altered and misrepresented is in the published version (it is for this reason that I am not naming names or the paper, though I can still prove the original fraud). The data in the published version could well be acquired and presented honestly and accurately, for all I know.

But then this is the problem: I don't know, I have no way of knowing, and nor does anyone else who reads it. But what concerns me now is that this is only one bad symptom of a wider disease.

Many battery scientists would probably agree with me when I say that there are a lot of papers published which are missing important experimental details, making them impossible to reproduce properly. Raw data for battery tests is very rarely, if ever, made readily available. The practice of cherry-picking results from the cells which work best – and which are therefore not necessarily representative – is I suspect rather widespread.

Also problematic – although not any kind of misconduct – is a general lack of knowledge and experience with electrochemical methods for batteries and battery materials, and that this is probably part of the reason why important details are missing. For example, many researchers do not fully grasp the interplay of different, seemingly-minor variables – such as electrolyte volume, thickness of the lithium foil in half cells, electrode thickness, quality of electrode preparation, amount or type of binder, and more – on rate capability, capacity fade and so on. This routinely leads to conclusions which aren't supported by the data. This should get picked up in the course of peer review, but then papers are reviewed by other researchers with a similar lack of experience in the area, and very quickly poor practices propagate through the literature. This can be the case regardless of the reputation of the PI and the impact factor of the journals the work gets published in.

Couple this with the perverse incentives surrounding publishing already – publish or perish, the well-observed tendency for "high impact" journals to publish eye-catching but not necessarily rigorous results, and career or even financial rewards for publishing in such journals, and so on – and clearly we have a problem.

The sum of all this – unavailability of raw data, normalisation of bad practices, bad incentives – is that it is hard not only to spot manipulation of battery test data but to even separate it from mere poor practice. How widespread is data manipulation or other practices done in bad faith, when the incentives are there and the chances of getting away with it are high? What fraction of papers published in the literature are actually reproducible, and make conclusions which are truly warranted by the results?

This situation surely has to change for the field as a whole to retain credibility, and I am increasingly of the opinion that open data is required. This would make fraud such as the type I discovered more difficult, but also allow other researchers to verify for themselves the quality of results, as well as to explore the data more deeply than conventional selected plots can allow.

That is why I, with the support of my PhD student, will now commit to releasing openly all raw and processed battery test data at the time of publication (at least where I am corresponding author, and as far as possible otherwise), and I will work to encourage my colleagues to do the same elsewhere. I do not know yet the best way to do this, but I hope to think of a useful solution in the coming months. In the meantime, I welcome any feedback on how this might best be done.

I also hope that other researchers in the field carefully consider the importance of how battery data is treated and presented, as well as the necessity to be absolutely clear about all experimental parameters and the validity of control experiments - and how important it is as reviewers to ensure this. As a field, we should have better standards.

Follow this link to comment...

The Braga/Goodenough glass battery part 2: of course I'm still skeptical

The most recent post published on this topic is Part III, published on March 13th, 2020. You can find that post here.

tl;dr: The latest paper on the Braga and Goodenough "glass battery" in the Journal of the American Chemical Society is, I believe, objectionable for a lot of the same reasons as with their earlier work in Energy and Environmental Science. Although this paper is clearer than their 2017 work in terms of the clarity of experiments, the results show that the electrochemistry of their battery is rather inconsistent, and the analysis is deeply flawed and also inconsistent. The discussion of an unprecedented "self-charge", which supposedly gives rise to the ever-increasing capacity, seems to me to stem from a misunderstanding of the nature of double layer capacitance. As with their 2017 work, there is a lack of consideration of any unwanted chemical reactions which would better explain the electrochemistry they observe in their cells, and which would not be translated into practically usable batteries. But, with such limited characterisation, we can only guess.

Read my previous post on the Energy Environ. Sci. (2017) paper here. I'm sorry, this one's really long too (~5,000 words).

Glass battery, part two: the saga continues

Those following the story of the "glass battery" work by Maria Helena Braga and the battery world's most high-profile figure, John Goodenough, may well have noticed the report of a new battery system in a paper entitled "Nontraditional, Safe, High Voltage Rechargeable Cells of Long Cycle Life" in the Journal of the American Chemical Society in April 2018. This new battery system is based around the same "glassy" electrolyte presented in their earlier work in a 2017 issue of Energy and Environmental Science. That earlier work attracted plenty of skepticism from other battery scientists, not least me: I wrote a rather long post detailing my own considerable objections to the claims in that work.

This new paper, however, presents some new startling claims of its own. For example: we now have a claimed lifetime of 23,000 cycles, a capacity which seems to increase over time without a clear limit, and an ability to "self-charge". These properties, much with the huge energy density claims in the previous work, would be revolutionary if they could be reproduced in devices of the scale needed for, say, electric vehicles. Steve Levine at Axios has already reported on some of the skepticism to some of these claims, but as was the case with essentially all of the press on the previous paper last year, there has been no real analysis from a scientific point of view.

Gerbrand Ceder was quoted in that article as saying that this new work is "not what it is stated to be. Most of us have moved on from this saga." I do agree, but the problem is that Goodenough has a high enough reputation that the majority of people not in the know (i.e., the general public, potential funders, investors) will implicity trust that he really is onto something and "the skeptics" (e.g., me, and most other scientists with some experience in this area) maybe don't understand some new science, or are just wrong. It's fallacious reasoning - but the rest of the battery field cannot expect the general public to take "the skeptics" seriously without trying to explain what we find problematic in these works. Otherwise we might as well look like those who doubted people like Galileo or the Wright brothers. And it does not help matters that this new paper, unlike the last one, is paywalled.

The new battery - it's not that "nontraditional", and it's not "all-solid-state"

This new paper is, I think, rather more complicated than the 2017 one, but let's start with the basics: the construction of the battery.

Despite being described as "nontraditional" in the title of their paper, the battery chemistry Braga and Goodenough are reporting here is actually rather "traditional" compared to their 2017 work. Gone, essentially without explanation, is the controversial claim of extensive plating of lithium metal on the positive electrode at potentials far more positive than the thermodynamically expected potential; the positive electrode is now a relatively conventional lithium insertion electrode with the formula Li[LixNi0.5−yMn1.5−z]O4−x−δFx.

It isn't explained why this particular compound was chosen, but it is similar to a well-studied high voltage electrode material known as LNMO (LiNi0.5Mn1.5O4). The paper provides data for this material when made into a very traditional-looking electrode with carbon conductive additive and PVdF binder, then made into a traditional-looking cell paired a lithium metal negative electrode and a conventional liquid electrolyte - and it looks as I would expect it to, so there's nothing controversial there.

Then, in comes the "glass electrolyte". This is made in essentially the same way as in their 2017 work, so my doubts about its composition and behaviour as I detailed in my previous post still stand - I won't repeat these here. However, to make contact between the electrode and the electrolyte, the authors coat a relatively large amount (30% by weight) of a waxy electrolyte ("plasticizer") made from succinonitrile onto the electrode, and then they add "half a drop" of a liquid electrolyte to make the electrode and electrolyte stick.

It's impossible to know how much "half a drop" really is, but it is probably a very large amount compared to the amount of active material (the LNMO-like compound) on the electrode. The amount of electrode material used in their cells is teeny-tiny - stated to be only up to 0.35 mg cm-2, which gives their "traditional" liquid electrolyte cells a capacity of ~0.06 mAh per square centimeter of electrode area. Electrodes in "traditional" lithium ion batteries have around 2 mAh cm-2, or more. In such batteries the weight of the electrolyte is often maybe 30% of the weight of the active material, but "half a drop" might be 10 mg or more, which would then be maybe five times the weight of the active material in the cell. There are therefore three different electrolyte materials at the positive electrode/electrolyte interface. This complicates things a lot, and I will come back to this later.

In any case, it is simply false to describe the cell as "all-solid-state", as the authors do repeatedly.

So this battery increases in capacity over time? How?

Before I get to why the capacity might be increasing I have to back up to discuss an experiment which is presented in the paper but mostly skipped over. I've drawn below simple schematics of three different cell constructions which are presented. The "traditional" cell with the liquid electrolyte and the "non-traditional" cell with the waxy succinonitrile plasticiser and the "glass" electrolyte I have already mentioned. But, the authors also briefly tested a cell with an electrode containing the plasticiser along with a liquid electrolyte as an intermediate experiment.

Schematics of the different cells

The discussion about the properties of the succinonitrile component is not particularly detailed and, I think, mostly wrong. The conductivity of the succinonitrile layer was determined (Figure 2a in the paper) by impedance spectroscopy (which I know plenty about) and the authors found it to be lower than they were expecting from the literature. In their experimental section, they say:

A mixture of succinonitrile (SN) plasticizer and LiClO4 salt in a molar ratio of 20:1 was heated in an argon-filled glovebox (MBraun, Germany) to get a transparent solution. The LiClO4 was mixed into the SN, but the conductivity of the SN did not increase as much as expected8 owing to not melting the LiClO4 prior to mixing it with SN.

The transparent solution tells me that the salt dissolves just fine in molten succinonitrile, so why they think that because they didn't melt the salt as well is some sort of problem is beyond me. Actually, since LiClO4 is a strong (if relatively kinetically inert) oxidiser, I really would not recommend melting it and adding it to an also hot organic liquid because I'm not sure it wouldn't just burst into flames.

Anyway, the authors present impedance measurements on the cells as well (Figure 2b) and conclude that the addition of the succinonitrile:

...creates a quasi-blocking electrode characterized by a Warburg impedance element associated with a charge-transfer resistance and a double-layer capacitance.

This isn't supported by the results at all - there is no significant difference in the impedance of the cell with or without the plasticiser when using a liquid electrolyte - it is only when the "glass" electrolyte is introduced that the cells have a very high impedance (with internal resistances approaching 10 kΩ - close to 1,000 times higher than what I'd expect for most "traditional" batteries). Which is also an interesting observation in itself, considering the supposedly high conductivity of the "glass".

Anyway, the telling result comes in Figure 4c, where the authors test the cell with the liquid electrolyte and the plasticiser. The cell initially behaves quite similarly to the "traditional" cell without the plasticiser on charging the cell (the cells are assembled in a discharged state), but on the first discharge it shows about half the capacity, and on the second charge, the charging voltage drops considerably, and no end to the charge is shown. All the authors say about this test is this (emphasis mine):

The electrochemical cell with the organic-liquid electrolyte contacting the anode and a plasticizer contacting the cathode (c) showed a rapid reduction of the charging voltage and no discharge current by the second cycle, indicating metallic lithium was not plated on the anode from the mobile cations in the liquid electrolyte.

The conclusion that metallic lithium is somehow not plated back onto the Li electrode at the negative electrode/electrolyte interface during charging is completely without supporting evidence, and assumes that the positive electrode/electrolyte interface is still working perfectly. But - and it is a really big but - in this comparison they have only changed the materials at the positive electrode and the negative electrode/electrolyte interface is in principle exactly the same! That is to say, the authors have changed the materials of the positive electrode and appear to be arguing that as a result they have changed the properties of the negative electrode instead. How this is effect is supposed to happen is left completely without discussion.

If I obtained this result in the lab, I would automatically work on the basis that the plasticiser layer I had added was not actually electrochemically stable in the operational window of the battery. It could be the succinonitrile oxidising, it could be the propylene carbonate in the "half a drop" that is oxidising, it could be corrosion of the aluminium current collector because of the lithium perchlorate salt - it could be all three.

The fact that the authors do not discuss this observation any further is inexplicable and in itself undermines the credibility of the rest of the article. If there are side reactions associated with the plasticiser layer, then these can be expected to still be there when the "glass" electrolyte is added, and this has to be taken into account in any discussion of the efficiency or cycle life of the cell.

So what's going on in the "non-traditional cell"?

Compared to the 2017 work, there is a relatively large amount of data to consider, and to avoid any possible accusations of copyright infringement I would rather not reproduce plots here (my apologies to those who cannot access the article). So, I will summarise here some of my observations about the battery test data of the cells with the glass electrolyte before I talk about the authors' analysis of the data.

Firstly, compared to the 2017 paper, data from five or six different cells is presented (there is additional data in the Supporting Information). The presentation of the data is somewhat inconsistent and the test conditions are a little different in each case, but all the cells seem to have (what I would say are) significant differences in their behaviour (voltage, capacity etc.). This suggests to me that these cells are not that well-reproducible.

There is a good example of this if one compares Figures 4f and 5c, which show how the "middle voltage" changes on repeated cycling. The cells tested in these two figures are apparently identical, but tested at different applied currents.

At a slower rate of 23 mA g-1 (Figure 4f), the "middle voltage" stays approximately constant at 3.37 V while the capacity gradually increases by a factor of five over the first 250 cycles or so. Over the next 60 cycles, the middle voltage drops, and the rate of capacity increase starts to slow down (only 308 cycles are shown). In Figure 5c, at a rate of 153 mA g-1, the "middle voltage" starts off closer to 2.5 V, increases to around 3.3 V over the next 2,000 cycles, then suddenly jumps up to 3.7 V where it remains more or less constant for the next 13,000 cycles. I don't see that these observations are consistent and the authors don't seem to connect them.

Secondly, a large part of the analysis and the discussion of the "self-charge" mechanism appears to hinge on the observation that the cell shows coulombic efficiency (the ratio of the discharge capacity to the charge capacity on any given cycle) in excess of 100% for "many cycles". It is true, for many cycles the indicated coulombic efficiency is over 100%, but for the 15,000 cycle test I just mentioned, the coulombic efficiency for any given cycle usually falls anywhere in the range 95 - 105%, which I would say shows that the cell is "all over the place". Any normal well-functioning lithium-ion cell would have a coulombic efficiency consistent to within a small fraction of a percent, which will mostly depend on what instrument is being used for the measurement and how stable the temperature in the room is.

What >100% coulombic efficiency should tell you

The only thing that a coulombic efficiency of more than 100% indicates is exactly what the definition suggests: that more charge is passed during the discharge of the cell within the specified voltage window than is passed during the charge of the cell.

All lithium-ion batteries can be expected to have a coulombic efficiency of slightly less than 100% on any given full charge/discharge cycle. That is, for the same charge/discharge current, the charging time of the cell will be slightly longer than the discharge. This is mostly because of unwanted reactions which happen especially when the cell is charging, such as the formation of a solid-electrolyte interphase (SEI) at the negative electrode or oxidation of the electrolyte at the positive electrode. Because of this, it is possible in theory to put more charge into the cell than it can theoretically take. This also tells you that you have to put more charge into the cell to charge it than you can get out when you discharge it, so this also impacts on the energy efficiency. However, in any well-functioning Li-ion battery, the coulombic efficiency is very close to 100% (e.g., 99.99..%).

By this definition, it is possible to have a coulombic efficiency more than 100%, but this would usually tell you that you have more side reactions on the discharge of the cell compared to the charge. For example, instead of oxidising the electrolyte, you could be reducing it: different side-reactions of course, but side-reactions all the same.

Obviously, this means that you could get more energy from discharging the battery than you put into it when you charge it. But this doesn't violate thermodynamics: there is chemical energy stored in other materials besides the active materials, such as the electrolyte itself, which can be released if it is reduced (almost anything reacts with lithium to release energy). But, if this is the case, you will eventually run out of electrolyte (or lithium) to burn away and the cell will die. But, if there is a relatively large amount of electrolyte - like there often is in small test cells like these - this can take a long time. Many battery researchers are simply ignorant of the effect of electrolyte volume on the cycle life of small cells and this is part of the reason why claims of super-long cycle life in laboratory tests do not get repeated in the real world, where electrolyte volume has to be kept as low as possible to minimise dead weight.

But what is happening? Why is the capacity increasing?

Ok, ok - let's get back to the article. I will pick out a few choice quotes to try and approach the authors' argumentation, which is really not easy to get into. But bear with me...

Why the capacity of our cells increases with cycle number over an initial aging time appears to reflect primarily an increase with time in the dipole polarization contribution to the anode and Li+-glass electrolyte/plasticizer EDLCs.

The key word is "appears". There does not seem to be any data to support this, and the cells already undergo "one month of aging" before the cycling begins, and the practical effect of this one month aging on the cell behaviour is not discussed or shown. And since this will become more important later - EDLC means "electrochemical (or electric) double-layer capacitance".

The argument is essentially that double layer capacitance contributes to the charge storage of the whole cell, and that this capacitance can increase with time and therefore the capacity of the cell can increase with time. This is actually not unreasonable, but the problem is that DL capacitance is present in liquid electrolytes too and is very, very small compared to the capacity of the insertion electrodes themselves. Supercapacitors work on this principle, but they use extremely high surface area activated carbons as the electrodes. Even then, the energy they store is quite small compared to battery materials, which store charge in the bulk volume of a material and not just at its surface.

The origin of the reduced Coulombic efficiency in the first cycle of the all-solid-state cell (which may reflect structural changes in the active material) has not been totally determined, although it might be due to initial lack of polarization of the electrolyte and plasticizer and interface disorder. The subsequent steady increase in capacity with cycle number is typical of the self-charge capability due to spontaneous polarization of the Li+-glass ferroelectric electrolyte.5

Some more weak language with "may" and "might be", but now we have the argument that the increase in capacity is in fact "typical" of the electrolyte's "self-charge capability". I didn't remember this being mentioned in any of the previous work, but reference 5 is cited here. Wait, reference 5 is unpublished results?!?!

The increase of capacity, and therefore of the energy density, with increasing cycle number eventually gives a capacity that is greater than the theoretical capacity of the oxide host cathode particles. This extraordinary observation indicates that there must be a storage of charge in addition to that in the active particles.

This I can agree with. But...

The additional stored charge can only be electrostatic storage in an EDLC as in a supercapacitor.

No. As I have briefly mentioned here and also discussed in my previous post, there are a lot of possible side reactions which can contribute to charge storage in the cell. However, it is impossible to know, because I discussed several possibilities in the previous work and with the addition of the succinonitrile, liquid electrolyte and wider voltage window there are even more possibilities this time. In addition, the theoretical capacity of the cell based on the active material is so low now that it is very reasonable to expect that side reactions from the electrolytes - which make up the vast majority of the total material in the cell except for the cell housing - could dominate the behaviour. And it is entirely possible that the breakdown of the electrolyte can produce compounds which are redox active and effectively capable of storing charge on this scale. The authors simply do not seem to consider this.

What about the "self-charge"?

Before I get into this it's worth pointing out that we got a hint about what Goodenough means by "self-charge" in an interview with Bloomberg in November 2017. In that, he said:

She [Braga]'s been lighting an LED with this battery charging itself for two years. It runs on ambient heat.

At the time this caught my attention especially because there was the direct claim that their battery was literally charging itself using heat from the atmosphere around it. There was no mention of this in any of their previous literature, and there is no battery known that can pull heat out of the atmosphere to charge itself without any applied current. This is a huge claim, and now that it's actually in the published literature we can take a closer look at it.

As I noted previously, there is a mention of "typical" self-charge in relation to unpublished work, but there is a further discussion of the "nature of self-charge" in the Supporting Information. This discussion also cites a rebuttal to the formal comment on the 2017 paper published by Daniel Steingart and Venkat Viswanathan - a rebuttal which I would have expected to have been published at the same time as the comment but which I can find no trace of - so that's a second unpublished work which they cite in relation to this supposed self-charge effect.

Anyway - this discussion of the "nature of self-charge" starts by describing one "self-charge" process as being common to "all kinds of battery cells", which

happens due to the necessity to align each electrode’s Fermi level with the electrolyte’s Fermi level in contact with the electrodes at the electrode/electrolyte interface as demonstrated in ref. [S1].

(that ref [S1], incidentally is the rebuttal I can find no trace of).

This, as far as I can tell, is referring to bog-standard double layer capacitance, which is completely uncontroversial and has been standard theory in electrochemistry for more than a century. I have never heard of this being described as self-charge, certainly not the sort which charges your batteries for you while simultaneously cooling your apartment down. This double layer capacitance effect - the alignment of Fermi levels - occurs in every battery, on both electrodes - but there is no net change in the charge stored in the cell, so there is definitely no "self-charge".

The discussion seems to continue to discuss this effect in the context of their solid electrolyte. The difference, they argue, is this:

...we argue that this self-charge is the result of equalization of the Fermi levels across the negative-electrode/electrolyte interface and a time lag between the arrival at the interface of the fast-moving electrolyte cation and the slower-moving electrolyte electric dipoles. Equalization of the Fermi levels by the interface EDLC (Fig. S9) is retained on arrival of the electric-dipole charge by a plating of an electrolyte cation onto the negative electrode, i.e. a self-charge, that creates a negative charge in the electrolyte.

Ok, this is complicated stuff and it's not easy to read, but I think I have a handle on what they're trying to say. They have diagrams which I also don't think are that clear, so I've tried to draw my own to describe the interfaces formed between a Li electrode and a liquid or a solid electrolyte:

Double layer capacitance in the "glass" electrolyte

As the authors describe, the electrode and electrolyte have different potentials (Fermi levels in physics language) which causes charges to move in order to equalise this imbalance in energy. In a liquid electrolyte, both positive and negative charges (i.e., the ions in the electrolyte salt) can move, so they rearrange themselves at the interface so that positive charges form a layer along the nominally negatively charged Li surface. Then, the negative charges form a more loose layer on top of those positive charges. This is the "double layer". The potential difference is then "dropped" across this double layer - effectively, the energy change as a result of this process is stored in the "double layer capacitor" that forms.

In an ideal solid electrolyte, only the positive charges can move. The negative charges cannot, because they are fixed in structure of the solid. But, there is still an energy imbalance that needs to be corrected. The authors argue that to do this, lithium can be deposited out of the electrolyte and dipoles in the electrolyte can become polarised in order to correct for the charge imbalance. I think this is fine, because it still creates a double layer in my view, but this only needs roughly one monolayer of lithium to equalise the energy difference - and this is where I think the authors have got it wrong - for much the same reasons they did in their 2017 work.

Moreover since plating creates negative charge in the electrolyte on the anode side, the electric field in the direction of E’’’ increases again and more cations are attracted to the anode surface while the cation-deficiencies diffuse towards the cathode charging the electrolyte/plasticizer EDLC. Furthermore, charging the electrolyte negatively makes the Fermi level of the electrolyte increase, discharging the anode/electrolyte EDLC and charging the electrolyte/plasticizer EDLC, which will discharge during cell’s discharge and contribute with the corresponding extra capacity.

They seem to be arguing that because the negative charges do not move, the difference in energy is never equalised, it just ends up somehow increasing without any suggestion of a limit - lithium ends up getting dragged out of the positive electrode, through the electrolyte, and onto the negative electrode, which represents an actual "self-charge" of the cell. And that this happens no matter what you do - they do say that this process happens on discharge as well as charge.

As they say in the article, charge is conserved in the cell. But - energy can't be. Assuming this is all true, if you discharge the cell only at the same rate it self-charges, then you can draw energy out of it forever - you have a perpetual motion machine. So I'm pretty sure we're back at breaking the laws of thermodynamics all over again.

The authors do actually suggest though that there is a limit to all this, bizarrely at the end of the introduction:

The cell capacity increases with cycle number until it stabilizes (not shown) at a value higher than the theoretical capacity of the cathode host oxide.

"Not shown" I think sums up this work.


If I have been a bit more heavy handed in my analysis of this paper compared to the 2017 Energy Environ. Sci. work, it is because I am dismayed that papers on this topic have continued to be published from this group and continue to receive press exposure with very little serious critique, and moreover that none of the authors have made a substantial effort to engage constructively with "the skeptics". Instead, we are waved away, laughed off or even criticised for "pouring cold water" on the work and discouraging those who might invest or obtain a license for their patent. I have to question their priorities.

Goodenough is on record as saying that "the skeptics do not understand the properties of an electrode/electrolyte heterojunction", but without explaining exactly why the electrode/electrolyte heterojunction in these cells should behave so differently from any other similar system. It's not as if their electrolyte is the only solid electrolyte - there is a huge literature on solid electrolytes and solid state ionics, not just in batteries but also other electrochemical devices such as solid oxide fuel cells. But the authors don't engage with this either in this paper or their previous work.

And this, for me, is the biggest problem. In my previous post, I criticised the authors for not discussing essentially any of the wider literature in their 2017 paper, and this remains true now. It could be true that I don't understand the properties of a heterojunction - but the authors of this work do not show much hint that they understand that much electrochemistry, because they show zero consideration of any of the myriad ways in which electrochemical cells can misbehave and give you unexpected results. And if there's one thing I'm sure I understand, it's that. (Otherwise I might need to find a new career!)

It is also a major problem, I believe, that some of the most controversial arguments in this new paper reference unpublished work, and we have seen how apparently key details are "not shown". It is difficult to understand how any paper can make it through peer review on this basis, but this is not an isolated incident anymore - and this new paper is in the flagship journal of the American Chemical Society. What does this say about the editorial and peer review processes, not only of this journal but those in which their previous work on this electrolyte has been published?

I met up with a friend at a recent conference who explained to me that Goodenough is a great ideas man with a great grasp of physics, but he's not an experimentalist - and so he trusts that when he is brought experimental data that it is correct, the cell is working properly and he has to be able to explain it somehow. I don't how true this is. But it seems that no-one involved in the process of this work, from experimental design to publication, has given much consideration to any of the many ways in which the data might not show what they think it shows. And there is no reason - in my moderately humble but somewhat experienced opinion - for anyone to take any of these convoluted and unprecedented arguments seriously until they have proved beyond reasonable doubt that the chemistry of their cell is what they say it is.

As a final note: although I never received any direct questions or criticism on my decision to publish my previous post publicly on my own website rather than in any kind of peer-reviewed literature, I know others have been criticised for doing similar things (not just on this topic). To pre-empt any such questions: this research I've done in my spare time in the space of a couple of days, and I have no intention of wasting my time going through any sort of peer-review process, which I'm not confident is all that good a mark of quality anyway. I also do this because this is a direct response not just to the research itself but also the wide press it and the previous work has received - so I believe it's important that questions are raised in the public domain. It just seems that there are very few people prepared to do this. I'm happy to hold up my hand and admit I'm wrong if presented with a well-supported argument, and correct anything in this post accordingly. I'm also happy to answer questions and try and clarify anything if requested. The comments are open below!

Follow this link to comment...

On the skepticism surrounding the "Goodenough battery"

The most recent post published on this topic is Part III, published on March 13th, 2020. You can find that post here.

tl;dr: It is my belief that the claimed mechanism by which the "Goodenough battery" is said to work is not supported by the evidence given, and appears to violate the first law of thermodynamics. There is considerable reason (and precedent) to suspect that the energy stored and released by the cell is significantly influenced by decomposition reactions of the electrolyte and/or reactions of impurities in the electrolyte. Most of the claims related to the properties of this battery are also not supported by the available evidence.

Last month's announcement of a safe, cheap, fast-charging, all-solid-state and long-lived battery technology from John Goodenough's lab was received with much fanfare by the science and technology media. The huge interest in this technology is in no small part due to the reputation of Goodenough, perhaps the most high profile scientist in the Li-ion battery field for his role in the discoveries of both lithium cobalt oxide and lithium iron phosphate, two of the most important electrode materials in Li-ion batteries. Goodenough was widely tipped for the Nobel Prize in Chemistry last year for these findings, and is still active in the field at 94 years of age. It is naturally intriguing that one of the giants of the field — and at his age — might have "done it again" and at last invented the super-battery at a time when the spotlight is on energy storage like never before.

However, the reaction within the battery field itself has been one of quiet but nonetheless strong skepticism. Some of this skepticism has surfaced, but I'm not convinced that any of the coverage fully explains how significant — and problematic — the claims surrounding this newly christened "glass battery" or "Goodenough battery" truly are. The observation that this battery appears to break the laws of thermodynamics is the most significant and eye-catching aspect to this announcement, but this is merely the tip of the iceberg. My feeling is that someone not intimately familiar with the field may readily get the impression that "the battery works, but scientists don't know why" from the existing coverage — rather than "this may not be a battery at all, even though it may appear to work", which I would argue is more accurate.

I am also concerned that several major claims have been made in the press release and promoted in the media which are not actually supported by what Goodenough and his co-authors have shown in their published articles. This is by no means unique to this story, but is a common and serious problem with popular scientific journalism, at least when it comes to reporting on developments in battery science.

For these reasons, I feel obliged to make some effort to try and explain why I too am skeptical about the "Goodenough battery". My own thoughts on this have come after a considerable amount of reading of the original articles related to this work[1-4], as well as the previous literature in this area. I wish to stress I have no stake in this particular area — I am not working on any similar or competing topic, and I have never met John Goodenough or any of his co-authors. But this story has captured my curiosity.

I will try and keep my explanations as simple as possible, but I want to be scientifically unambiguous and objective. Hopefully, someone with the time and resources to repeat these experiments, confirm or disprove the "Goodenough battery", and publish it in a peer-reviewed journal will eventually do so. That person probably won't be me, because this is not my area. Until then, I'm happy to debate my analysis in the comments and update this post should any new information come to light.


Let's start with the most significant claim: the operating mechanism of the battery. Dan Steingart at Princeton has already explained this well and in detail in his own excellent post, but I will briefly describe this here anyway.

The first experiment in [1] describes a cell with a Li metal negative electrode, the solid electrolyte, and a positive electrode which was a mixture of sulfur and carbon. When the cell was discharged, the amount of charge passed was more than ten times what should theoretically be obtained by discharging the sulfur, as should be expected, but was 90% the theoretical capacity of the Li metal negative electrode. Immediately, it is concluded that:

[...] the sulfur acts as a redox center determining the voltage of the cell at which electrons from the anode reduce the Li+ at the electrolyte/cathode interface to plate lithium rather than reducing the sulfur, so long as the voltage remains above 2.34 V; below 2.34 V, the S8 molecules are reduced to Li2Sx (1 < x < 8 ) [...]

A battery operating in this way is without precedent. In terms of the half reactions of the battery, we have Li metal being oxidised (stripped) off the negative electrode:

$$ \text{Li} \longrightarrow \text{Li}^+ + \text{e}^-$$

and being redeposited back on the positive side:

$$ \text{Li}^+ + \text{e}^- \longrightarrow \text{Li}$$

Combining these two equations gives us no overall chemical reaction, but a battery voltage of ~2.5 V and a theoretical capacity determined only by the lithium electrode, for a total of 8,500 Wh/kg(!!). This would translate into perhaps ten times higher energy density than Li-ion batteries (a good rule of thumb is to take the energy density of the materials and divide by four — so in this case a cell level energy density of 2,000 Wh/kg would be a reasonable estimate), and more than any other battery system known to science.

But with no overall chemical reaction, where does the energy come from? In response to the public skepticism, John Goodenough himself provided an explanation to Computer World (my emphasis added):

The answer is that if the lithium plated on the cathode current collector is thin enough for its reaction with the current collector to have its Fermi energy lowered to that of the current collector, the Fermi energy of the lithium anode is higher than that of the thin lithium plated on the cathode current collector.

It is worth noting at the outset that this explanation is not given or even implied in the original paper, nor in Goodenough's more recent single-author paper in ACS Catalysis[5]. What is being described here, as far as I can tell, is underpotential deposition, which is well known. It is true that it can be more favourable for a metal ion (let's say, copper, as an example) to be deposited onto a different metal (such as platinum) than to deposit on itself. Copper will deposit onto platinum at a higher voltage than it will onto copper itself. However — once one monolayer, of copper atoms is deposited on the platinum, any further deposition of copper is onto copper, which occurs at the equilibrium voltage.

Much the same would be true for the lithium battery: a single layer of lithium atoms a fraction of a nanometer thick may indeed deposit onto a different substrate at a higher voltage (but 2.5 V higher is huge — lithium doesn't even alloy with any metals at such high voltages). At most, this effect might occur for a few monolayers, but still corresponding to a very small amount of charge. Once lithium is depositing onto itself, it will do so at the equilibrium voltage, and there will be no difference in the chemical potential of lithium on either side of the cell, and hence no voltage. There are countless examples of this behaviour, and why it should be any different in this situation is unexplained.

Is there even lithium metal deposition on the positive side?

The problem with getting stuck into a discussion about how this battery seems to violate the first law of thermodynamics is that it is easy to overlook a more fundamental issue: by any conventional standard, there is no evidence that lithium has been deposited onto the positive side of the cell in the first place. Again from [1]:

In order to verify this conclusion, we disassembled the cell of Fig. 1 and examined the electrodes with the naked eye and with SEM EDS analysis, as shown in Fig. 2, which indeed shows lithium plated on the cathode current collector and no evidence of metallic lithium remaining on the stainless steel at the anode or the anode side of the electrolyte after full discharge of the lithium anode.

Firstly, no EDS spectra are given in the paper (and there is no Supporting Information). In any case, EDS is a particularly poor technique for studying light elements such as lithium. Most electron microscopes are incapable of even detecting the element (although it seems like it can be done with the latest instruments), and if they could they would not be able to distinguish metallic lithium from any other material containing lithium. Neither the photographs nor the electron micrographs provided look like lithium metal (I say this as someone with experience of looking at lithium metal with an electron microscope), and there is no point of reference (a micrograph of a pristine or "re-charged" electrode, for example). I accept that it is probably not that easy to unambiguously confirm the presence of metallic lithium, but considering the significance of the claims being made, that confirmation is essential.

So what is happening?

Before I go on I want to calculate some numbers (capacities and currents) related to the battery tests made in the paper, because this is useful when assessing the cell "performance". There are three different "batteries" tested:

  • A lithium negative electrode and a sulfur/carbon positive electrode
  • A sodium negative electrode and a ferrocene/carbon positive electrode
  • A lithium negative electrode and a MnO2 positive electrode

Of these, I will only consider the first one; there is simply not enough information provided about the others to do any calculations with, however they all appear to be constructed and behave similarly. Comparable behaviour between the different batteries seems to be implied in the paper — the complete description of the sodium system, for example, is only two sentences long.

The "lithium-sulfur" cell is said to contain 1.99 mg of sulfur, and in the first experiment is discharged until the total charge is 10.8 times the theoretical capacity of the sulfur (which is 1,672 mAh g-1). I can estimate this as being 1.99 mg × 1.672 mAh mg-1 × 10.8 = 35.9 mAh. This is quite a lot of charge for a laboratory coin cell - I make decent (if I say so myself) lithium-sulfur batteries in coin cells of the same size and they typically have a capacity of 3 — 5 mAh.

The current though is quite low - this 35.9 mAh is discharged over 28 days, which implies a constant current of 35.9 mAh / 672 h = 53 µA through the whole cell. The current is also stated to be 30 mA/g — I assume based on the mass of sulfur — which comes out to be 30 µA/mg × 1.99 mg = 59.7 µA. These numbers are a bit different, but fairly close, so I seem to be in the right ballpark. The authors state that:

At voltages V > 2.34 V, the cell is rechargeable and the sulfur is not reduced.

A voltage of 2.34 V is reached after the cell has discharged about 8.5 times the theoretical capacity of the sulfur, so repeating the previous calculation, I understand that the reversible capacity of the cell should be 28.3 mAh. Still quite a lot.

The "rechargeability" of the cell is demonstrated in the next experiment. This time, the current is increased slightly to "40 mA/g", so based on the previous numbers I estimate the current is now at most 79.6 µA. The cell is discharged and charged for 10 hours each, so now the total charge passed through the cell on each charge/discharge cycle is 7.9 µA × 10 h = 0.796 mAh. This is just 2.8% of the reversible capacity of the cell, and how the cell behaves when discharged and charged more deeply (or more quickly) is not shown. It is perhaps worth noting at this point that with this narrow window, the 8,500 Wh/kg mentioned earlier becomes closer to 240 Wh/kg. Using my "divide by four" rule of thumb again, this might correspond to about 60 Wh/kg on the cell level, so quite a lot lower than today's Li-ion batteries. The maximum reversible capacity of the other batteries is impossible to estimate.

An inadvertent Li-air battery?

Regardless, something is charging and discharging, but if it is not lithium metal plating, then what? The suggestion that maybe oxygen leaked into the cell, inadvertently forming a lithium-air battery, was discussed in the Quartz article linked to previously. In the lithium-air battery, oxygen is reduced below approximately 2.7 V to form lithium peroxide. The question of whether a leak could be involved is a fair one: the authors themselves, in describing this experiment, say:

The charge and discharge voltages show a good coulombic efficiency over 1000 h; the cycling was continued beyond 46 cycles despite an imperfect seal of the cell.

However, the paper's first author, Maria Helena Braga, has rejected the explanation, being quoted in the same Quartz article as saying:

Well if we have a Lithium-air battery then we have a very good Lithium-air battery

Since the topic of the Li-air system has come up, it's worth remembering what has defined the research in that field over the last decade. K. M. Abraham reported the first lithium-air battery in 1996, and showed that the discharge product was lithium peroxide, Li2O2. These batteries used a polymer electrolyte and the reversible capacities were relatively low. About a decade later, other researchers did similar experiments using liquid electrolytes, showing much higher capacities, and demonstrating rechargeability over tens of cycles. But... they didn't check that Li2O2 was formed on discharge, or that it gave back the oxygen on recharge.

It was later found that only a very small amount of the charge passed on discharge went into the formation of Li2O2 — and no oxygen was reformed on charge. Even though the cell looked like a battery, almost all the charge passed went into burning off the electrolyte. The "cycle life" depended only on how much electrolyte was in the cell — nothing was reversible. For this reason, a lot of work has gone into developing better methods to determine what truly happens inside the Li-air battery during usage — both to check that the "correct" substances are formed and that the unwanted decomposition products are avoided — and to use these techniques to search for more stable systems. The recent history of the Li-air field should be a warning to all battery scientists to be careful about what unwanted reactions might be occurring.

But let's consider this a bit further. The positive electrode in the first experiment is 10% carbon by weight, and 47% sulfur. If there was 1.99 mg of sulfur in the positive electrode, then presumably there was 0.42 mg of carbon in the electrode. Dividing the 35.9 mAh discharge by this mass of carbon gives us a whopping 85,500 mAh per gram of carbon (mAh gC-1). The charge relative to the mass of carbon in the electrode has long been a common performance metric for Li-air batteries (because it's not easy to measure the mass of the oxygen itself), so we can make a comparison here. 85,500 mAh gC-1 is a lot — a few thousand is more reasonable for most Li-air battery electrodes. For this reason, I don't think an oxygen leak alone can account for this observation — but it is only one observation.

What about the glass electrolyte? Could it decompose?

I'll describe a little bit about the electrolyte first. It's based on a crystalline material, Li3OCl, which shows high ("superionic") conductivity at room temperature. Li3OCl belongs to a relatively new class of ion-conducting solids which have an "antiperovskite" crystal structure. With some amount of doping (exchanging of 2 Li+ for a single M2+, leaving a vacant site where the other atom should be), the authors claim to have created an amorphous material with an even higher conductivity.

In the four papers[1-4] published by these authors on the reported glass electrolyte, there is only a limited discussion of the stability of the electrolyte in[3]. The electrochemical stability is determined to be at least 8 V. I could question this number, because there seem to be some inconsistencies in the experiments and the results — but for the sake of this discussion let's go with it. An earlier paper, however, calculated the properties of crystalline Li3OCl from first principles and predicted that it should decompose into Li2O2 and LiCl above 2.55 V. This might be an oxidation of the form:

$$ 2 \text{ Li}_3\text{OCl} \longrightarrow \text{Li}_2\text{O}_2 + \text{LiCl} + 2 \text{ Li}^+ + 2 \text{ e}^- $$

occurring at a voltage similar to the charging voltage of the cell in [1]. Further oxidation into lithium perchlorate (LiClO4) was also predicted at higher voltages. It is also worth noting that another previous paper describes the synthesis of Li3OCl films by pulsed laser deposition (PLD) — and even the construction of batteries, with lithium cobalt oxide and graphite electrodes. In that paper, the batteries show the behaviour expected from the electrode materials, and have been charged and discharged for 20 cycles. But — there is certainly evidence of continuous over-charging, which might be related to oxidation of the electrolyte. Nothing is obvious here, but this is something that merits closer consideration.

Is the electrolyte impure?

At this point I have to make clear my lack of experience in solid electrolytes and this type of chemistry. I know, however, that some scientists have privately suggested that the conditions used to prepare the electrolyte are not sufficient to give a material with the claimed composition. Having read the four papers [1-4] in more detail, I think there are numerous reasons to suspect so, with little evidence available to determine the true composition of the electrolyte. Firstly, I think it is important to note that the electrolyte composition is given slightly differently in each paper:

  • In [1], it is A2.99B0.005O1+xCl1-2x
  • In [2], it is A2.99B0.005OCl1-x(OH)x
  • In [3], it is A2.99B0.005OCl
  • In [4], it is A2.99B0.005OCl.xH2O, (x < 1).

where A is Li or Na, and B is the dopant (e.g., Mg, Ca or Ba). Since [1] and [4] both reference paper [2] for the synthesis, it is strange that the composition should be given differently each time.

The original paper reporting the crystalline Li3OCl describes the synthesis as follows:

In a typical synthesis, 2.40 g of LiOH [...] and 2.12 g of LiCl [...] are ground together for several minutes with a mortar and pestle. The resulting paste is placed in a quartz tube and heated to 330−360 °C (past the melting point Tm = 282 °C of the product) under vacuum for several days. During heating, water is effectively removed with a condensation (liquid nitrogen) trap and a high-vacuum pump [...] At the end of the synthesis, the apparatus is flushed with a dry inert gas (e.g., Ar) and the very hygroscopic sample is never subsequently exposed to atmospheric moisture. Continuous heating (330−360 °C at melt) and removal of water under high vacuum drive the chemical equilibrium toward the formation of the Li3OCl product: $$ 2 \text{ LiOH} + \text{LiCl} \longrightarrow \text{Li}_3\text{OCl} + \text{H}_2\text{O}$$

Quick summary: heat above melting point, high-vacuum pump, liquid nitrogen trap, and never expose it to the atmosphere. In [3], the first of the papers by Braga et al., the reaction conditions are somewhat simpler: after mixing the same precursors and adding "a few drops" of water:

a paste was formed and introduced in a Teflon reactor, which was firmly closed. The reactor was heated at 220–240 °C for at least 4 days before it was opened to let the water evaporate for approximately 1 h. Then it was closed in glassware and allowed to cool to room temperature. A vacuum pump was used to dry the water out.

Again, to summarise - excess water is evaporated off at ambient pressure, and then vacuum dried at room temperature. This is less extreme than in the original paper, but Li3OCl can be obtained, but with unknown purity - there is at least one impurity phase the authors detect, Li5(OH)2Cl3, and there may conceivably be other amorphous impurities. I am not sure if even lithium hydroxide monohydrate (LiOH.H2O) can be dehydrated at 240 °C and atmospheric pressure, let alone condense it into the oxide. The vacuum pressure and duration are surely important, but no further details are given. It is also worth noting that the authors say that the added water is essential. In [2], the later paper, the conditions have been simplified even further.

The precursors were weighed and mixed in stoichiometric amounts for 25 g batches; 10 to 30 mL of deionized water was added to the powder mixture before the solution was enclosed in a teflon reactor that was heated to 230 to 250 °C for 2–3 h in a heated sand bath. The hot reactor was then opened to evaporate water and HCl from the glass/amorphous products at the heating temperature. A slurry was prepared by grinding the glass/amorphous product to a powder in liquid ethanol (99.9%, Merck) to prevent attack of the particle surfaces by humid air. Similar procedures have been described for glass-electrolyte experiments with gold blocking electrodes and alkali-metal electrodes.[2]

That reference at the end there is in fact the authors referring to their own previous work which we have just seen — [3] in my reference list below. Now, instead of a few drops of water, 10 — 30 mL is added, and the product is ground to a powder in a solvent which is also hygroscopic (that is, it absorbs water from the air). No further vacuum drying is mentioned — only a 130 °C heating step to remove the ethanol. For those who are unfamiliar, lithium salts in general are very, very hygroscopic, and many can keep absorbing water until they dissolve themselves into the water they've absorbed (I've seen LiCl do this in a closed container, for example).

The evidence for the formation of the desired product is provided by differential scanning calorimetry (DSC) and dielectric spectroscopy. Neither of these techniques provide information about chemical composition. I have little experience with the former and none with the latter, so hesitate to make my own interpretations from the data — but the data does not speak for itself and I am unconvinced by the arguments. A simple technique such as thermogravimetric analysis (TGA) would easily and clearly support claims of water loss at specific temperatures, but this is not included. On the available evidence, I think there are good reasons to expect a significant amount of hydroxides and water in the electrolyte. The presence of water, especially, could be a major factor in the unprecedented high conductivity.

The presence of water, as well, complicates the possible side-reactions in the cell enormously. Water can be electrolysed directly; either oxidised to oxygen gas and H+ or reduced to hydrogen gas and hydroxide, both with the transfer of 2 electrons per water molecule. Because water has such a low molecular mass, a lot of charge can be passed in its electrolysis: approximately 3 mAh is required to destroy one microlitre of water. Whether or not water is electrolysed directly in the "Goodenough battery" I couldn't say — in fact I would be surprised if it were the case — but water can cause all sorts of other reactions. I have already mentioned the possibility for formation of Li2O2 from electrolyte decomposition; this can react with water to form hydrogen peroxide and lithium hydroxide:

$$ \text{Li}_2\text{O}_2 + 2 \text{ H}_2\text{O} \longrightarrow \text{ H}_2\text{O}_2 + 2 \text{ LiOH}$$

Hydrogen peroxide might be oxidised or reduced by the applied current itself, or it would slowly decompose into water and oxygen. The consequences of oxygen being present in the cell I have already discussed. And if there is a leak in the cell, as has also been argued, then water as well as oxygen could be leaking in, fuelling these side-reactions further.

Unfortunately, at this point, it is all speculation. We will not know what is really happening until the original authors or another group repeat these experiments but with a much more detailed analysis of the chemistry.

1,200 cycles? Fast recharge?

To finalise this post, which has become more of a tome than I had originally anticipated, I want to address some of the claims that have been made and repeated in the media which I feel are either unsubstantiated or misleading. I will go through them (mostly) in the order they appear in the original press release.

A team of engineers led by 94-year-old John Goodenough [...] has developed the first all-solid-state battery cells that could lead to safer [...]

This battery is not the first all-solid-state battery, and it is hard to see on what basis this claim is made. Solid polymer electrolytes have been known for decades, and for years now, Bolloré have thousands of cars on French roads powered by an all-solid-state battery with a solid polymer electrolyte and a lithium metal anode. Solid electrolytes based on ceramics are also well-known, and Sakti3 — despite their controversies — were bought by Dyson a couple of years ago for $90m. I'm sure there are a number of other companies developing similar technologies. This isn't even the first report of a battery based on the Li3OCl electrolyte - at least one paper I mentioned previously demonstrates a working battery, even if it's a small, low power one.

The researchers demonstrated that their new battery cells have at least three times as much energy density as today’s lithium-ion batteries.

This I discussed before. Based on the claims in the paper, I would say about ten times — but maybe three was chosen to sound more reasonable. The claimed energy density of 8,500 Wh/kg is more than three times the theoretical energy density of lithium-sulfur batteries, which are themselves already close to twice the energy of Li-ion batteries, and thought to be capable of much more. But if we're talking about how much reversible capacity is actually demonstrated, then this number is much lower indeed.

The UT Austin battery formulation also allows for [...] a faster rate of recharge (minutes rather than hours).

This is unsubstantiated. The fastest charge/discharge rate in any of the "batteries" in [1] is four hours, and only a small percentage of the cell's claimed capacity is "recharged" at all.

Instead of liquid electrolytes, the researchers rely on glass electrolytes that enable the use of an alkali-metal anode without the formation of dendrites.

Protection against dendrites is one obvious advantage of a solid electrolyte over a liquid, but it is no guarantee. The ability of dendrites to grow through soft polymer electrolytes has been long known, and it did not take long for me to find a paper reporting the growth of dendrites through a ceramic electrolyte. A completely non-porous, grain boundary-free electrolyte is needed to completely protect against dendrite growth. In [1] and [2], Braga et al. claim that the 130 °C heating step after preparing the electrolyte film "reform[s] the solid glass electrolyte without grain boundaries". It is hard to see how this step, below the melting point and without any external pressure, and with the "glass" held inside a fibreglass or paper sheet, can accomplish this — no evidence is provided.

In experiments, the researchers’ cells have demonstrated more than 1,200 cycles with low cell resistance.

The largest number of cycles shown in the papers is 250. If the authors have observed cells cycling for 1,200 cycles, they have not shown it. Nor have they shown evidence of low cell resistance — only in symmetrical cells, which are not the same. The "batteries" show relatively large differences between charge and discharge voltage, which may in fact turn out to be a result of high cell resistance.

This is the first all-solid-state battery cell that can operate under 60 degree Celsius.

Again, it's hard to see on what basis this claim is made. Here's a solid polymer electrolyte battery at room temperature from 1990. Here's a more recent all-solid-state lithium-air battery, at room temperature. Even in my own research group, I have colleagues making polymer electrolyte-based batteries which operate at room temperature. I am sure I could find more examples if I looked.

And finally, let's consider the assertion that the battery would be safe, because of the non-combustible electrolyte. I am being silly here, but let's just take these claims to their logical conclusion. We have a battery which is said to deliver 8,500 Wh/kg on the materials level (let's call it 2,000 Wh/kg on the cell level), and can charge in minutes. It's safe to say if it can do that, it can also discharge in minutes, or less. So what happens if, God forbid, something goes wrong and the battery short-circuits, maybe because of damage? The cell will turn that 2,000 Wh/kg into heat as fast as it can, and start to get hot — really hot. I believe a good rule of thumb for the maximum temperature is about a 1 °C rise per Wh/kg of the battery energy density. In which case, the battery will be glowing red hot, melting, and even if it's not on fire itself, anything flammable in the vicinity will be. What really makes any high energy density battery safe?


I am deeply skeptical about the claims made about the "Goodenough battery". To accept the claimed mechanism of operation means accepting something at odds with two centuries of accumulated knowledge on electrochemistry and battery science, on the basis of evidence significantly below the generally accepted standard and with a large number of reasons to suspect an alternative, conventional explanation. There is no discussion in the paper [1] in which an alternative explanation is even considered, and there is almost no discussion of the related literature besides the authors' own work. It is remarkable that this, and many of the other points I and others have already publicly raised do not seem to have been addressed at the peer review stage. Both the most recent paper [1] and especially the previous paper [2] (which I have not discussed here in any detail) invite far more questions than they answer. The recent public responses by Braga and Goodenough have not — so far — clarified any of these concerns, but have rather created more questions.

Given the situation it's hard to avoid caveating these conclusions with "there's a small chance it may be right". Maybe it is — anyone can be wrong. However, given the available evidence, the scientist in me is insisting there is essentially no reason to believe any of the central claims, and every reason to suspect that the observed behaviour can be explained by other processes, including redox activity of these supposed "catalytic relays" (sulfur, MnO2, etc.). The electrolyte is not well characterised, and there is simply not enough evidence available in the authors' papers, or their patent application, to draw any firm conclusions about what is really happening. As a result, there is almost no foundation to any speculation about the potential applications of this new "battery". I want to stress something I mentioned at the top of this page, that the question surrounding this battery is not "the battery works, but we don't know why", but instead "this may not be a battery at all, even if it may appear to work". The difference is huge.

It is discomforting to write this post knowing that John Goodenough has written authoritatively about these developments in his own article[5] and is privately expressing confidence that these results will be confirmed. Goodenough is, after all, a decorated scientist who made significant contributions to human knowledge before even my parents were born. But, I know the skepticism in the field surrounding this paper is intense, and I think in this case it is very much in the public interest for it to be debated openly. I hope that this post will give some food for thought.


[1] M.H. Braga, N.S. Grundish, A.J. Murchison, J.B. Goodenough, Energy Environ. Sci. 2017, 10, 331
[2] M.H. Braga, A.J. Murchison, J.A. Ferreira, P. Singh, J.B. Goodenough, Energy Environ. Sci. 2016, 9, 3, 948
[3] M.H. Braga, J.A. Ferreira, V. Stockhausen, J.E. Oliveira, A. El-Azab, J. Mater. Chem. A 2014, 2, 15, 5470
[4] M.H. Braga, J.A. Ferreira, A.J. Murchison, J.B. Goodenough, J. Electrochem. Soc. 2017, 164, 2, A207
[5] J.B. Goodenough, ACS Catal. 2017, 7, 2, 1132

Update - 1/5/2017

As has been noted in the comments below, John Goodenough recently did an interview with Slashdot in which he answered a range of questions from the commenters there. One aspect of the operation of the cell which has been described in more detail relates to thickness of the lithium supposedly deposited at the positive electrode:

The key to the concept of a battery voltage that takes metallic lithium from the anode and plates it on the cathode is that a thin lithium (order of a micron thick) current collector is plated on a copper (or other) cathode lithium having a chemical potential over 3.5 V below that of metallic.

And again in another reply (my emphasis added):

...plating on the cathode from the anode can only give a voltage for a finite thickness of the plated material on the cathode side. We have not yet obtained a good measure of the thickness of the cathode plating that is viable, but it appears to be micro not nanometers thick. Optimizing the capacity will involve the ability to optimize the surface area of the cathode material. This optimization has yet to be performed, but we can plate sodium as well as lithium.

"Microns" would be much too thick for underpotential deposition, as I described earlier on, so whatever limits or determines this thickness remains a mystery. Most concerning, however, is that this directly contradicts one of the central claims of [1]:

...the ability to plate/strip an alkali-metal anode in contact with a Li-glass or Na-glass electrolyte allows a totally unconventional strategy for the design of a rechargeable battery in which reversible plating of an alkali metal from the anode onto the cathode current collector gives a battery cell having a capacity determined by the amount of alkali metal used as the anode rather than the solid-solution range of the working ion in a host cathode lattice.

The system cannot have a capacity limited both by the amount of metal at the negative electrode and by a maximum thickness that can be plated on the positive electrode. The latter is not described in the published literature, and is remarkable that such crucial information should be left out. Unfortunately, this interview does little to resolve any of my concerns about this work.

Follow this link to comment...

A half-solution for two (or more) y-axes with ggplot

I've been teaching R, and especially ggplot, to beginners in the language this week, and predictably the topic of how to put two separate y-axes (with a common x-axis) on the same plot came up.

Unfortunately, the answer is "not easily", since the inability to do this is on purpose (Hadley Wickham gives the reasons here, for example). Actually putting one y-axis on the left side of the graph and a different y-axis can be done, but requires some delving into the heart of ggplot which is a beyond my understanding at the moment.

What is easier - and in my opinion, preferable in most cases - is to use facetting or a package like gridExtra to have separate stacked panels. But gridExtra (specifically the grid.arrange() function) misaligns plots which have expressions (subscripts and superscripts) in the axis titles - and facetting by default doesn't make it easy to label axes the way I want (again, because I often need to add super/subscripts in labels), or rescale the y-axes of individual facets to values I want.

I had a think about it after the discussion we had in class, and managed to reach a reasonable compromise with the facetting approach, which is fairly straightforward and doesn't require any extra packages.

I'll demonstrate this with some arbitrary functions with very different ranges of y values:

x <- seq(from = -5, to = 5, by = 0.05)

df <- data.frame(
  x = x,
  fun_a = sin(x^2),
  fun_b = 50 * sin(x)

To make use of facet_grid() this data needs to be converted to "long" format, which is easily accomplished with tidyr::gather():


df2 <- df %>%
  gather(key = fun, value = y, -x)

Now we can ggplot() this data with the two functions in separate facets, making use of the scales = "free_y" argument:

ggplot(df2) +
  geom_path(aes(x = x, y = y, color = fun)) +
  facet_grid(fun ~ ., scales = "free_y")

plot of chunk unnamed-chunk-3

This is fine, but what if I want to plot data series which have different units? I'd prefer to have an axis title on the left for each facet. There's only one y-axis title here, and I can't easily change that - but what I can do instead is change the facet labels, move them and make them look like axis titles.

The easiest thing to do seems to be to change the column names with dplyr::rename() before gather(). To show the superscripts, etc, the column names have to have the form of expressions, like they would if you were to do the same thing with ylab().

df3 <- df %>%
  rename(`sin~(x^2)` = fun_a, `'50'~sin~(x)` = fun_b) %>%
  gather(key = fun, value = y, -x)

Now I can remake the plot with a couple of extra arguments to facet_grid(), and some theme() modifications to make the strip.text (facet label) look the same as the x-axis label.

ggplot(df3, aes(x = x, y = y, color = fun)) +
  geom_path() +
  facet_grid(fun ~ ., scales = "free_y", 
             labeller = label_parsed,
             switch = "y") +
  theme(strip.background = element_blank(),
        axis.title.y = element_blank(),
        strip.text = element_text(size = rel(1))) +
  guides(color = FALSE)

plot of chunk unnamed-chunk-5

Great! This is more or less what I'm after. I still have some grumbles though. One of the main ones is that I can't easily rescale the y-axis on an individual facet - I'm stuck with scales = "free_y".

The most I can do - as far as I know - is force it to rescale a bit outside the range of the data by making some dummy data which I include in the plot as an invisible geom_blank. Like this:

dummy <- data.frame(
  x = 0,
  y = c(-1.5, 2),
  fun = "sin~(x^2)"

ggplot(df3, aes(x = x, y = y, color = fun)) +
  geom_path() +
  geom_blank(data = dummy) +
  facet_grid(fun ~ ., scales = "free_y", 
             labeller = label_parsed,
             switch = "y") +
  theme(strip.background = element_blank(),
        axis.title.y = element_blank(),
        strip.text = element_text(size = rel(1))) +
  guides(color = FALSE)

plot of chunk unnamed-chunk-6

But I don't know if rescaling the y-axis of a single facet to a range within that of the data is easily achievable. I would also like to be able to change the facet heights manually - maybe that's possible with gtable, for example, but that's out of my expertise. Here's hoping for an easier implementation in a future version of ggplot!

Follow this link to comment...

Could wind + batteries really replace a nuclear power plant?

tl;dr: Probably not. The intermittency and seasonal variation of wind is so severe that even optimistically, it would be considerably more expensive and, if it was to use Li-ion batteries for storage, would require as many as have been produced in the entire world over the last four years.

The UK government's recent decision to delay the final decision on the planned Hinkley Point C nuclear power plant has somewhat rekindled the debate on whether the UK should have the plant at all. Some have gone as far as to suggest that given the projected cost, the guaranteed price of the energy produced and the timescale of the project, it should be scrapped and allowed to be replaced with a combination of renewable energy sources and with energy storage, both of which are dropping in cost.

[Jeremy Leggett, the founder of solar panel maker Solarcentury] is delighted that others are picking up on arguments he has been making for years. "Finally the message is getting through that Hinkley, and indeed nuclear, make no sense today simply because wind and solar are cheaper. If we accelerate renewables in the UK, we can get to 100% renewable power well before 2050," he says.

I'm sure we can all agree that cheap, low-carbon renewable electricity would be a great thing. But if you are proposing to eventually remove nuclear and fossil fuels entirely, are renewables still as "cheap" when we need to rely on them to maintain demand? From that same Guardian article:

The Economist believes improved electricity storage is a key answer to the frequently repeated criticism of wind and solar that it is intermittent, and points out that battery technology is fast improving.

First, let's be clear. It is not a mere "criticism" of wind and solar that it is intermittent - it is a cold, hard fact. It's physically impossible to generate solar power at night and to generate significant wind power when the wind's not blowing. In order for electricity generation to meet or follow demand, excess generation must be curtailed somewhere, or the energy stored. Similarly, insufficient generation must be supported by some other form of energy generation to avoid blackouts. At the moment, this role is largely provided by gas in the UK. But in the absence of other conventional means of generation, this role needs to be filled by some form of storage, perhaps the much vaunted batteries. This is especially important if neighbouring countries make similar moves towards renewables, since available wind speed and sunlight does not tend to vary much across neighbouring countries - those neighbours may not be able to export energy when generation is barely sufficient across an entire continent.

But how much storage would be needed, say, to effectively convert intermittent renewable power into providing baseload power equivalent to that which would be provided by Hinkley Point C - a constant 3.2 GW? Well, this can be estimated with some crude analysis of publicly available data. I was interested to see how it turned out, and I figured it was worth reproducing here. I did the analysis with R, and have included the code (except for the code generating the plots) and the data here so that it can be reproduced.


I'm going to use existing wind generation data for this analysis, since the UK already has a significant amount of wind power, and on the assumption that large-scale deployment of solar power would not be all that sensible for one of the darkest countries in the world. The data I've used is the energy production data for the UK for the entire year 2015 - from - which I've reuploaded to this website.

The data is in the standard csv format, and I use a couple of addon packages for analysis.

gridwatch <- read.csv("")


Let's check what it looks like:

##       id            timestamp demand frequency coal nuclear ccgt wind
## 1 377525  2015-01-01 00:00:04  28809    50.090 9079    8049 3360 5251
## 2 377526  2015-01-01 00:05:02  28645    50.092 8947    8053 3369 5254
## 3 377527  2015-01-01 00:10:02  28768    50.116 8843    8052 3372 5272
## 4 377528  2015-01-01 00:15:02  28917    50.045 8763    8047 3339 5303
## 5 377529  2015-01-01 00:20:02  28964    50.030 8818    8051 3386 5223
## 6 377530  2015-01-01 00:25:02  29055    50.006 8906    8055 3392 5189
##   french_ict dutch_ict irish_ict ew_ict pumped hydro oil ocgt other
## 1        582       900       -72   -136     15   443   0    0  1157
## 2        586       898      -100   -134      0   441   0    0  1157
## 3        586       898      -100   -134      0   440   0    0  1157
## 4        586       898      -100   -134      0   439   0    0  1155
## 5        586       898      -100   -134      0   440   0    0  1155
## 6        586       898      -100   -134      0   441   0    0  1155

The data is quite thorough, but all I really want for now is the data for wind. First I'm going to convert the timestamp to POSIXct date/time format with the appropriate function from the lubridate package, then I can select out the data I need.

gridwatch$timestamp <- ymd_hms(gridwatch$timestamp)
df1 <- select(gridwatch, timestamp, wind)

Check again:

##             timestamp wind
## 1 2015-01-01 00:00:04 5251
## 2 2015-01-01 00:05:02 5254
## 3 2015-01-01 00:10:02 5272
## 4 2015-01-01 00:15:02 5303
## 5 2015-01-01 00:20:02 5223
## 6 2015-01-01 00:25:02 5189

The wind column shows the power generated in units of MW.

plot of chunk unnamed-chunk-6

Straight away you can see the issue with intermittency. Wind production averages about 2.6 GW over the whole year, but this can be in excess of 6 GW during windy times, and almost nothing during some lulls in the summer. At the end of 2015, the UK had a total of 13.6 GW of capacity installed, indicating a capacity factor of 19%, which seems reasonable.

I'll make two new columns in this data frame - one for the time increment (in seconds) and then use that to integrate the power column to get the total energy generated in MWh.

df1$difftime <- c(0, diff(df1$timestamp))
df1$totwind <- cumsum(df1$wind * df1$difftime / 3600)

So, I want to see how a constant 3.2 GW baseload can be generated by wind, with excess energy stored and then released when the wind isn't sufficient. We can reasonably assume that with more turbines the power generated will scale linearly. We can make a new table from the same data, but adjust the wind so that the total energy generated by wind power throughout the year will be equal to 3.2 GW x 24 hours x 365 days.

df2 <- select(gridwatch, timestamp, wind)
df2$difftime <- c(0, diff(df2$timestamp))

df2$wind <- df2$wind * (3200 * 24 * 365 / (max(df1$totwind)))
df2$totwind <- cumsum(df2$wind * df2$difftime / 3600)

The average should come out to be about 3200 MW now, so let's check that's the case:

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   77.92 1506.00 2889.00 3201.00 4805.00 8022.00

For plotting purposes I'll include equivalent columns for the constant 3.2 GW.

df2$base <- 3200
df2$baseenergy <- cumsum(3200 * df2$difftime / 3600)

Now the table looks like this:

##             timestamp     wind difftime   totwind base baseenergy
## 1 2015-01-01 00:00:04 6294.875        0    0.0000 3200     0.0000
## 2 2015-01-01 00:05:02 6298.472      298  521.3735 3200   264.8889
## 3 2015-01-01 00:10:02 6320.050      300 1048.0443 3200   531.5556
## 4 2015-01-01 00:15:02 6357.213      300 1577.8120 3200   798.2222
## 5 2015-01-01 00:20:02 6261.309      300 2099.5878 3200  1064.8889
## 6 2015-01-01 00:25:02 6220.550      300 2617.9669 3200  1331.5556

Let's see how the total amount of energy generated from wind over the year would look like compared to a constant 3.2 GW.

plot of chunk unnamed-chunk-12

The energy generated by wind is shown in red. The difference between these two lines will show us roughly the difference between the energy generated by wind power and the energy consumed by the constant 3.2 GW we're looking for. So let's do that:

df2$diffpower <- df2$wind - df2$base
df2$diffenergy <- cumsum(df2$diffpower * df2$difftime / 3600)

The power requirements will look like this (positive values indicate wind in excess, so therefore the batteries would be charging. Negative values indicate that wind is insufficient, so the batteries need to take over to ensure 3.2 GW):

plot of chunk unnamed-chunk-14

The total energy stored or released looks like this:

plot of chunk unnamed-chunk-15

It's clear from this plot there is a huge seasonal variation in wind power, with greater generation in the winter, and the storage needed as backup in the summer. The difference between those minimum and maximum peaks (in March and October respectively) is the total amount of energy we would need to backup the wind power - therefore it's the capacity of the storage we would need to guarantee a constant 3.2 GW baseload without relying on other methods of generation. In this case, we can see it's of the order of about 4 TWh (that's TERAwatt-hours).

4 TWh is a colossal amount of energy - roughly the yield of 35 Trident nuclear missiles, roughly the total battery capacity of about 45 million Tesla Model S cars, and about 100 years worth of Li-ion batteries at the current rate of production.

More than that, at a target price of $100/kWh for batteries (=$100bn/TWh) this would cost of the order of $400 bn for the batteries alone. This price for usable battery storage is considerably cheaper than currently available for any of the available chemistries, even if you're Tesla.

Meanwhile, based on a capacity factor of about 20%, we would need wind power with a capacity of 16 GW, which based on estimates of $1.3m - $2.2m per MW would cost between $20.8 and $35.2bn. The cost of Hinkley Point C by comparison is estimated to be of the order of $24 bn. So yes, the wind turbines could potentially be cheaper to install than the nuclear power plant, but even then, they can't supply power on the same basis without additional - and possibly insane - storage capability.

More wind turbines?

Ok, maybe it's not viable to try and store every last bit of energy produced by the wind turbines. Perhaps we can just disconnect them during very windy periods and only store enough energy needed when wind isn't enough. How many batteries might we need then?

To estimate this I went back again and scaled up the wind further, so that the averege power is 20% over the 3.2 GW we actually want. The rest of the code is the same.

df4 <- select(gridwatch, timestamp, wind)
df4$difftime <- c(0, diff(df4$timestamp))

df4$wind <- df4$wind * (1.2 * 3200 * 24 * 365 / (max(df1$totwind)))
df4$totwind <- cumsum(df4$wind * df4$difftime / 3600)

df4$base <- 3200
df4$baseenergy <- cumsum(3200 * df4$difftime / 3600)

df4$diffpower <- df4$wind - df4$base
df4$diffenergy <- cumsum(df4$diffpower * df4$difftime / 3600)

If I remake the same plot and rescale it a bit:

plot of chunk unnamed-chunk-17

Then the difference between the minimum and maximum values here is the amount of energy that needs to be stored, which is about 2 TWh - halving the cost of batteries needed for just 20% more wind turbines. This seems more reasonable, so what if you keep adding more wind power?

Well, I won't reproduce the plot here - you can do it yourself - but the cost does keep shrinking. If you double the amount of wind power - to 32 GW, giving an average of 6.4 GW over the year, you can still see that there are periods of insufficient wind to meet 3.2 GW that still require about 200 GWh worth of storage. This is still a huge amount, equivalent to the total storage of a few million electric cars and about four years worth of production of Li-ion batteries. For comparison, the world's largest grid storage battery opened earlier this year in Japan - a sodium-sulfur battery with a capacity of 300 MWh (that's 0.3 GWh).

The cost of installing this would probably then be in the range:

  • Wind turbines: 32 GW capacity, $41.6 - $70.4bn
  • Battery storage: 200 GWh, ~$20bn

This would imply an installation cost of maybe three times as much as Hinkley Point C, for a system with a considerably shorter lifetime, and most likely with much more expensive electricity for the consumer. I doubt if I kept going that it would get considerably cheaper than this, since at this point the cost of the wind turbines is already rather in excess of the cost of the batteries.

So what can we conclude? I think the main thing this thought experiment shows is just how important it is to have a diverse mix of technologies in the energy generation mix. I think this fact is usually lost on people with stubbornly anti-nuclear and anti-fossil fuel views. I know as well that I have not considered other things like tidal or hydroelectric power here, but these are very much geography-dependent and not always a viable option.

I don't intend to say that there is no place for renewables at all in our energy mix, but I cannot see a target of 100% solar or wind grid production as being anything except ruinously expensive, and it seems wildly improbable that battery storage could make it work.

Bear in mind too, that this is just one power plant. Average demand in the UK was 32.8 GW in 2015!

The last word

At this point it is reasonable to point out that the seasonal trend in solar generation runs (very roughly) opposite to that of wind. That is, it is less windy in the summer, but it is sunnier in the summer, so a mix of solar and wind might reduce the requirement for storage. This is true, but some level of storage would still be required, and my intuition tells me that it would not be a significant difference (and other analyses done elsewhere suggest as much). At the moment I don't have data for UK solar generation to play around with, but this would be interesting for a future post.

Follow this link to comment...