=chemistry =batteries =energy
Why do I like the battery chemistries I do?
background
A battery is a chemical reaction
separated by a barrier that only allows ions and inert molecules to pass.
This couples the reaction to transfer of electric charge. Most batteries use
a liquid electrolyte that dissolves ions but not reactants. For example, an
alkaline battery
uses zinc metal and manganese oxides, and neither dissolves in alkaline
water, but zinc ions do.
If there's leakage, even at 1/1000th the
maximum discharge rate, that's a big problem. The reactants must have
negligible solubility in the electrolyte, and usually that's only the case
for solids, because low solubility implies strong self-interaction which
usually implies high melting points, so electrodes are generally solid. The reactants must also have
some electrical conductivity, even if they're small particles embedded in a
conductive matrix.
Lithion-ion batteries typically use a
polypropylene mesh as
separator, and the battery layers are so thin that even that mesh is a
nontrivial fraction of the battery cost. It's possible to make polymer
membranes that are selective for positive or negative ions. This involves a
complex
internal microstructure. Such membranes are obviously much more
expensive than a mesh and liquid electrolyte - too expensive for batteries
for electric cars. For grid energy storage, the charge/discharge rates are much lower,
so membranes are closer to economic viability, but still too expensive and/or not
durable enough. (Also, electricity is worth more in an electric car than as
grid power.) Ion-selective polymer membranes also tend to have some
leakage, which can permanently reduce battery capacity for some chemistries.
Cost also depends on voltage: higher voltage is better. Form Energy has a
low cell voltage, and the membranes are too expensive. So, ion-selective
polymer membranes are rejected.
A
ceramic
electrolyte is obviously much more expensive than a flexible polymer
membrane, which is already too expensive. It's hard to make lots of thin
ceramic without cracks that stuff can leak through or metal dendrites can
form through. Ceramic electrolytes can work at high temperatures and make
things like
sodium-sulfur batteries work, but they're too expensive. So, ceramic
electrolytes are rejected.
Dendritic deposition is a big problem for batteries. Alkaline batteries
can't be recharged mainly because the zinc metal forms needles because a
shorter path has less resistance. Lithium metal does the same thing, which
is why
lithium complexed with graphite is used despite being more expensive and
having less capacity. The electrolyte needs to reach everywhere in the
graphite, but it also has to all be connected to be conductive. That
requires somewhat expensive processing.
There are rechargeable zinc
flow batteries. If electrolyte flows over the zinc surface fast enough, it
prevents dendritic deposition. However, doing this is more expensive.
There are now many startups pursuing lithium-metal batteries with
ceramic electrolytes to block dendrites. Thin ceramic electrolytes without
cracks are too expensive for this, so those startups will fail. Another
problem is getting good contact, which usually requires high pressure (which
can cause ceramics to crack) or a
liquid on the lithium-metal side. Having a liquid there leads to another
problem: resistance. In lithium-ion batteries, a
solid-electrolyte
interphase (SEI) forms on the particles of lithium x graphite, from
decomposition of the electrolyte. This SEI increases resistance and thickens
over time, reducing capacity, but it protects the Li x graphite from further
reaction. With a lithium-metal battery, the SEI forms on a flat sheet, so
the surface area is lower, so the resistance is higher.
sodium-ion
Lithium is somewhat expensive.
How about a lithium-ion battery, but with sodium instead? Sodium gives only
0.3v lower voltages than lithium, which is good. The extra mass of Na vs Li
isn't a big problem, and neither is diffusion rate through electrolyte. The
problem is with the electrodes.
Na doesn't complex with graphite as
well as Li does. You need a lot more graphite to do the same thing. So, the
negative electrodes for Na-ion batteries are a problem.
Li is a
smaller ion than Na, which means it fits into crystals better. For example,
the structure of LiFePO4 is more similar to FePO4 than NaFePO4 is. That
means Na gives higher resistance and faster degradation of positive
electrodes. So, the negative electrodes for Na-ion batteries are also a
problem. There are some organic molecules that work with sodium, but they're
too expensive, degrade over time, and have poor specific energy.
magnesium
Magnesium is cheap, has good
theoretical capacity, and gives the same voltage as sodium. The main problem
is that it has a much higher charge density than lithium, so it has stronger
interactions with everything, which increases the energy barrier of it
moving between things. This is especially a problem in the SEI.
Li-ion batteries use electrolytes like LiPF6, LiBF4, and LiClO4. These form
SEIs that have low but adequate Li+ ion conductivity. If you try to make a
Mg-ion battery with the corresponding Mg salts, you get a SEI of MgO or MgF2, and those
aren't conductive at all.
what's the goal?
Suppose you replace the Li of a
Li-ion battery with Na or Mg and get the same performance. This is good
because it reduces material costs, but the cost of Li is only a fraction of
the battery cost, so the improvement is limited. If the positive electrode
gets more expensive, or performance is worse, then it's probably not worth
it.
The positive electrode of Li-ion batteries is expensive because:
- it might contain cobalt and/or nickel
- it has to be porous with
electrolyte contacting small particles
- it needs to be continuously
connected with conductive material
These complex requirements require somewhat expensive processing, and that's part of what I wanted to avoid.
sulfur
What if you used sulfur <->
sulfide as a reaction? Solid sulfur isn't electrically conductive, but
polysulfides are soluble in, for example, water and molten sulfur. Sulfur
particles can be reduced at their surface, dissolving as polysulfides that
are reduced to sulfides.
Of course, you then have a liquid containing
polysulfides, which would react with the negative electrode. The
polysulfides must be kept separate from it to avoid permanent degradation,
which is hard. This requires a ceramic membrane or a very high-performance
polymer membrane, which is too expensive.
There are startups working
on
lithium-sulfur batteries, but they have no solution for this problem, so
they will fail. Sodium-sulfur batteries work, but again, the ceramic
electrolyte is too expensive.
bromine
Zinc-bromine
batteries have been made. Bromine, conveniently, can be stored as a
liquid at room temperature, and it's easy to convert between Br- and Br2 so
electrode overpotential is low.
A membrane is obviously needed to
prevent Br2 from just reacting with zinc directly. You might note that Br2
is not charged, which means ion-selective membranes can't reject it very
effectively. How, then, do zinc-bromine batteries prevent Br2 from crossing
over? The answer is, they don't. Lifetimes are short and continuous
electrolyte regeneration to put the bromine back was too expensive. You also
still need a cation-exchange membrane because bromine forms Br3- ions, and
again, that's somewhat expensive.
Anyway, the specific energy of
zinc-bromine batteries is too low for good electric cars.
what options remain?
Suppose we want to make a
low-cost battery with abundant materials that's suitable for electric cars.
- Lithium,
cobalt, nickel, vanadium are rejected because they're expensive and limited.
- Palladium, platinum, and rhodium are too expensive and rare to use even as
catalysts.
- Ceramic electrolytes are too expensive. Polymer membranes
are too expensive.
- Current lithium-ion anode and cathode materials are
rejected for being too expensive.
- Because lithium is rejected, current
lithium-ion cathode materials that don't work well for other ions are
rejected.
- Chemistries with specific energy too low for electric cars
are rejected.
What is left?
The anode
must be solid metal. That means the SEI must have lower resistance than
current ones. It also means a method to prevent dendritic deposition is
needed.
The only cathode options left are:
- things that
are too expensive
- manganese oxide
- things that aren't conductive
While it's not a problem for
non-rechargeable alkaline batteries, manganese oxide cathodes tend to have
relatively poor cycle life. When used with Mg, there's a tendency to form
non-conductive MgO layers.
If the active material at the cathode
isn't conductive, then it must react and dissolve at the surface, and it
must be inside a conductive liquid.
If the cathode has a conductive
liquid (CL), then there must be a separate electrolyte layer to prevent
self-discharge. Membranes are too expensive, so that must be another
electrolyte liquid (EL). For it to be a separate layer, CL and EL must be
immiscible. To prevent battery degradation, the active material must have
negligible solubility in EL. For acceptable conductivity and self-discharge,
the conductive material must have much higher solubility in CL than EL,
>1000x higher.
For all this to be the case, CL and EL must be
maximally different types of liquid, but both must conduct ions.
The
biggest difference between CL and EL you can get while conducting ions
through both is low-polarity ethers vs something high-polarity.
High-polarity stuff tends to react with a metal anode, so EL must be
low-polarity, and CL must then be high-polarity.
High polarity means
strong interactions, but CL also needs to be liquid and have low viscosity,
which are conflicting requirements.
So, those are the requirements I tried to meet:
- nonpolar EL
- very polar CL
- very polar cathode active material
- solid metal
anode
- low-resistance SEI
- no dendritic deposition
The result worked as far as I was able to test it, but that required several new elements, which seems to make designs hard to understand.