03 September 2008

Nanoparticle LiFePO4 Batteries

The lithium-ion batteries based on phospho-olivine (i.e. LiMPO4, where M = {Mn,Co,Fe}) crystalline structure have been the subject of a great deal of research over the past decade. A recent paper in Nature Materials from Gibot et al. has demonstrated some of the developments in the area and I'd like to rehash them here [1]. The paper demonstrates fabrication of a single phase LiFePO4 with very small particle dimensions. It's not an, "Oh my god what an amazing engineering development paper," but rather one of scientific interest to elucidate the difference between two-phase and single-phase Li-ion batteries.

LiFePO4 thus far seems to be the most impressive performer, especially from a safety perspective. It is produced entirely in solution (e.g. a beaker) by a chemical recipe. I don't know what the yields are like but the nature of the production method implies that it can be undertaken in large vats on an industrial scale.

Unfortunately, it suffers from poor conductivity characteristics. Two main approaches have been made to improve the conductivity of LiFePO4: (1) to coat the particles with a thin layer of amorphous carbon, and (2) to manufacture the LiFePO4 in the form of small nanoparticles (~ 40 nm average axial dimension). Of course, both approachs can be combined.

Adding carbon improves the conductivity but adds an, "electrochemically inactive," layer to the cathode material, hence reducing performance by adding dead weight to the battery. One can imagine that when you take a mass of Li-ion particles and sinter them together to form the cathode, if they've all been coated with carbon then there's an electronically conductive pathway from any buried particle to the electrolyte.

For the nanoparticle approach, presumably the higher ratio of surface area to volume reduces the ion diffusion length, but the literature also suggests that the introduction of defects to the nanoparticles may also improve conductivity. I know from experience that stacking faults (such as twins) can act as diffusion pathways for reaction species in solid-state reactions.

Normally the cathode material is really LixFePO4, where x = 0.5 - 0.75. This is (at least partially) as a result of the boundry between crystallites being composed of an extremely lithium poor phase (x ~ 0.03). The primary advance shown in the Gibot paper is that they made the nanoparticles small enough that only a single phase is found in each particle. To explain, if you are familar with the difference between monocrystalline and polycrystalline silicon solar cells, the sub-40 nm LiFePO4 nanoparticles are monocrystalline. Particles in the range of 100 nm are polycrystalline and hence have the low lithium phases present at the boundaries of each crystallite. Note that there's no fundamental electrochemical advantage to the monocrystalline approach as far as I know.

Figure 1: Potential-capacity and capacity-power curves for nanoparticle LiFePO4 (reprinted from [1]). In the top figure, 'C' represents a charge curve and 'D' a discharge curve for carbon-coated LiFePO4 nanoparticles. The number after the letter is the number of hours the discharge took place over. I assume '2D' is the discharge curve for thirty minutes.

I infer from the paper that a big difference here seems to be in the lithium loading. Gibot showed by a variety of methods that their nanoparticle was loaded with more lithium (x = 0.82-0.92). However, their discharge performance curves aren't actually more impressive than existing LiFePO4 batteries with larger particles. Existing batteries have flatter discharge curves from what I've seen.

The real advantage for these monocrystalline nano-LiFePO4 is likely to be reversibility. As I've discussed previously, the volume of the crystal changes from lithuim insertion to deinsertion. This introduces strain into the crystal and after many cycles defects will form and degrade performance. However, in a monocrystalline material there's not a lot to break. The nano-LiFePO4 does have some substitution defects (Fe where Li should be and vice versa) but without the crystal boundries the defect density is likely to be lower overall.

Another potential advantage for the nanoparticle approach is that it requires less in the way of process temperature (108 °C versus 500 °C over 24 hours) compared to the traditional approach. That should make the manufacturing process less energy intensive and less expensive.

[1] P. Gibot et al., "Room-temperature single-phase Li insertion/extraction in nanoscale LixFePO4", Nature Mat 7 (2008), 741-747.


Fat Man said...

Am I correct in calculating that 160mAh/g is about 2MJ/kg?

Robert McLeod said...

Yes but that's probably for the anode material in isolation. I.e. no cathode, electrolyte or casing. The theoretical limit is around 170 mA*hr/g.

I can't read the article from home easily so I'll double check tomorrow.

Robert McLeod said...

Yeah those figures are electrode only.

Unknown said...

The carbon coated version might be better for high current applications and the higher internal resistance version better for specific energy and storage leakage.

Anonymous said...

Hey guys!
i just started blogging not that long ago and running across this blog it seemed a bit too interesting to only read the first paragraph. I kinda got confused in the middle of it but the end just made it all go together like a puzzle. Please, who ever wrote this, keep me updated!

JetForMe said...

Hi. I realize this comment isn't directly related to your posting, but given your knowledge of the subject, I wonder if you know about (or resource about) computer simulation models of LiFePo4 batteries?