LFP vs NMC vs LCO: which lithium-ion cathode to pick?

At a glance (164 = one LFP research powder at 0.1 C; 140 = LCO at a 4.2 V cut-off; density kinds differ; blank = no material-level value)

Which lithium-ion cathode has the highest capacity?

On paper, NMC, Lithium nickel oxide (LNO) and Lithium cobalt oxide (LCO) sit close together at 275 mAh/g, 250 to 275 mAh/g and 274 mAh/g, and Lithium iron phosphate (LFP) is lowest at 170 mAh/g. These are calculated material-level figures that assume the lithium is removed as far as the formula allows, which real cells do not do. The NMC figure is a general number, not one per composition.

Delivered capacity is lower and depends on the voltage window. LCO delivers about 140 mAh/g at a 4.2 V upper cut-off, which a review puts at 51% of its theoretical capacity, and LFP gave 164 mAh/g at 0.1 C for one carbon-coated research material. NMC has no material-level delivered capacity. Its values are full-cell results per gram of NMC, in graphite cells at 1 C and 25 °C: NMC111 gave 162.8 mAh/g at a 4.4 V cut-off and 140.2 at 4.2 V, NMC622 gave 177.8 at 4.4 V and 155.4 at 4.2 V, and NMC811 gave 172.5 at 4.2 V and 131.9 at 4.0 V.

No like-for-like comparison exists: the LFP number is one research powder at a slow rate, the NMC numbers are full cells at 1 C, and the LCO number has no stated rate. NMC622 at 4.4 V (177.8) and NMC811 at 4.2 V (172.5) exceeded the LFP research figure (164 mAh/g); LCO (140), NMC111 (162.8 and 140.2) and NMC811 at 4.0 V (131.9) did not. So the higher-nickel NMC grades beat LFP only at higher cut-offs. Our data cannot separate NMC from LCO on energy; that choice rests on voltage window, cobalt exposure and heritage.

Which lithium-ion cathode has the highest voltage?

LCO and LNO run highest in our data, at 3.9 to 4.2 V vs Li/Li+ and 4 V vs Li/Li+, and LFP lowest at 3.4 V vs Li/Li+, but the numbers are different kinds. The LFP value is 3.4 V as quoted by a review. The LCO value is an operating range from a comparison table, and a half-cell measurement on nanostructured LCO gave a first plateau of 3.92 V. The LNO value is a review statement of about 4 V with no cell format or rate.

NMC has no sourced average voltage, so it is not on the chart, and "LCO and LNO run highest" is a reading of the listed values, not a ranking that includes NMC. The profile does hold the voltage at which charged NMC starts to release oxygen in a half cell at 25 °C: near 4.7 V for NMC111 and NMC622 and 4.3 V for NMC811. That is a limit on how far to charge, not a working voltage.

Cell voltages include the anode and are quoted differently from these material values: a Sony camera pack that blends LNO with LCO has an average operating voltage of 3.6 V, and NASA's sheet for an LFP cell lists a nominal 3.3 V at C/2 and a 2 to 3.6 V range. General engineering knowledge, not in our data: cell voltage is roughly the cathode voltage minus the anode voltage. Lithium titanate (LTO) works at 1.55 V vs Li/Li+, about 1.2 V above a carbon anode, which lowers cell voltage and energy density.

What do capacity and voltage give together, and what about density?

Energy per gram needs capacity and voltage together, and our data cannot give that product for all three cathodes: LCO has a theoretical 274 mAh/g at 3.9 to 4.2 V vs Li/Li+, LFP 170 mAh/g at 3.4 V vs Li/Li+, and NMC 275 mAh/g with no sourced voltage. We do not multiply them, because the voltages are not averages from a discharge curve.

Our data holds two cell-level specific energy figures and they are not comparable. The NASA A123 26650 LFP cell, an older design from a 2007 evaluation, works out to about 105 Wh/kg from a 2.3 Ah nameplate at 3.3 V and a mass of 72 g. The ISS 134 Ah LCO/graphite cell is listed at 155 Wh/kg on a NASA slide, but the slide is inconsistent: its own nameplate energy of 496 Wh and mass of 3.53 kg give about 140.5 Wh/kg. A small cylindrical cell and a large flight cell differ in design as well as cathode.

Density is the weakest comparison: the chart note lists its three kinds. NMC has no density, and no cathode has energy per litre.

Which cathode is safest when a cell is heated?

LFP's cell peaked lowest in both side-by-side heating tests in our data. In the first, a 2.2 Ah LCO/graphite cell peaked at 545.11 °C, an NMC811 cell (written NCM811 in the source) at 462.52 °C, an NCA cell at 491.84 °C and an LFP cell at 239.26 °C, all at full charge, and the paper ranks the runaway danger LCO above NMC811 and far above LFP. Our profile states no capacity for the NMC811, NCA or LFP cells, so this test is not normalised for stored energy either. In the second, 18650 cells: the 1.1 Ah LFP cell peaked at 259 °C and stayed below 3 °C/min, while the 2.2 Ah NMC111 cell reached 731 °C and 7580 °C/min. These are peak temperatures in an insulated calorimeter, which is not propagation between cells or fire behaviour.

Onset is a different story, so "peaked lower" describes severity once running away, not when it starts. In the small-cell test that ran LFP, NMC111 and an LMO (lithium manganese oxide) cell at the same threshold, onset was 90 °C for LFP and 91 °C for NMC111, and the authors say it was about 90 °C for all three, so early cell reactions set it. The LCO/graphite cell began at 111.78 °C, which its authors attribute to the film on the graphite. A 24 Ah prismatic LFP/graphite cell began at 116.743 °C, reached the runaway trigger at 170.265 °C, vented at 182.582 °C and peaked at 416.788 °C; that cell's vent gas was mostly hydrogen, carbon monoxide, ethylene and methane, so "more tolerant" does not mean non-flammable. A 250 mAh LFP/graphite pouch cell showed onset at 140 °C.

These onsets are not comparable one to one: cell size runs from 250 mAh to 24 Ah, detection thresholds differ (0.01, 0.02 and 0.03 °C/min), and heating steps, electrolytes and cathode blends are not matched. General engineering knowledge, not in our data: in an insulated test the peak temperature rises with the energy stored in the cell, and the NMC111 cell held twice the capacity of the LFP cell, so part of that gap is cell size. The one material-level figure for LFP, a threshold of 210 to 250 °C, is a second-hand review quote with no test method and sits above every measured cell onset, so it is not a safety margin.

For the other cathodes the evidence is about oxygen. Charged NMC releases oxygen from its surface, which can react with the flammable electrolyte. LCO starts to lose surface oxygen above about 4.3 V, with significant release at 4.5 V and faster loss above 60 °C. LNO has no cell-level runaway number; a comparison table gives it 190 to 210 °C with no stated definition, so it is indicative only.

Which cathode lasts the most charge-discharge cycles?

Our data cannot rank the cathodes on cycle life, because the tests are not alike, but it shows what each one's life depends on. A 250 mAh LFP/graphite pouch cell kept 91% of its capacity after about 275 cycles at C/3 and 40 °C. A review quotes more than 6000 cycles with no test conditions, so the two figures should not be set side by side.

For NMC, life depends on the upper cut-off. In graphite full cells at 1 C and 25 °C between cycle 5 and cycle 300, NMC111 kept 94% at 4.4 V, 93% at 4.2 V and 42% at 4.6 V; NMC622 kept 94%, 95% and 39% at the same three; NMC811 kept 66% at 4.2 V, 77% at 4.1 V and 90% at 4.0 V. The authors conclude stable cycling is possible up to 4.4 V for NMC111 and NMC622 but only 4.0 V for NMC811. At each grade's stable limit NMC811 delivered 131.9 mAh/g, below the 162.8 and 177.8 that NMC111 and NMC622 delivered at 4.4 V.

For LCO, the long counts are battery-level. The ISS batteries use 134 Ah cells charged to 3.95 V per cell, not 4.2 V, and had logged about 16,000 cycles on batteries 1A and 3A by 9 Oct 2019, a count to date and not to failure. Capacity of battery 1A1 went from 113.1 to 109.9 Ah between Jan 2017 and Feb 2019. A ground test of a 10 Ah cobalt-oxide battery passed 3441 cycles at 40% depth of discharge, and the Aquarius battery lists about 6500 cycles to date. These are the only flight counts for any cathode here.

For LNO, undoped material retained 69.2% of its capacity after 100 cycles at 0.5 C in a second-hand review figure with no voltage window or temperature, against 91.4% for a niobium-doped version. The profile ties the fade to nickel in lithium sites and to a damaging phase change once more than 75% of the lithium is removed. Cycles to a retained-capacity limit, cycles to date and cycles quoted without conditions are three different things.

What do the anode and electrolyte change?

They shape heat and life, so a cathode result only holds for its stated pairing. Graphite has a heat-release onset of 80 to 160 °C in a literature summary, with its protective film breaking down at 90 to 120 °C, which fits the roughly 90 °C cell onsets above. Fully lithiated graphite began releasing heat at 100 °C with 2700 J/g, against 125 °C and 383 J/g for lithiated LTO, whose advantage is mainly less heat, not a later onset. In the NMC set more than 90% of the CO and CO2 formed in high-voltage cycling came from released lattice oxygen reacting with the electrolyte.

Terms used

Theoretical and delivered capacity

theoretical capacity is the charge a material could store per gram if its lithium were removed as far as the formula allows. Delivered capacity is what a cell or electrode actually gives in a test; it depends on rate, temperature and voltage window.

Cut-off voltage

the voltage at which charging or discharging stops. A higher upper cut-off gives more capacity from layered oxides but stresses the cathode and the electrolyte.

Half cell and full cell

a half cell pairs the material with a lithium-metal electrode; a full cell pairs it with a real anode such as graphite. Full-cell numbers are closer to a product.

Material level and cell level

material-level numbers describe the powder or electrode. Cell-level numbers describe a finished cell with its own anode, electrolyte, size and case. The two should not be compared directly.

Thermal runaway and self-heating onset

thermal runaway is a cell heating itself faster than it can lose heat, often with venting of flammable gas. Self-heating onset is the lowest temperature at which a heated cell starts to warm itself above a stated detection rate, so it depends on the test.

Accelerating-rate calorimetry (ARC)

a test that holds a cell in an insulated chamber and heats it in steps, so the cell's own heat release is seen without heat loss.

State of charge (SOC)

how full the cell is. Every cell-level heating test cited here was run at 100% state of charge.

NMC111

grades of nickel manganese cobalt oxide; the digits are the nickel to manganese to cobalt ratio, so NMC811 has the most nickel.

LFP

lithium iron phosphate, lithium cobalt oxide and lithium nickel oxide (cathodes), and lithium titanate (an anode).

What to check before you order

1. Ask whether a capacity figure is per gram of cathode or per gram of cell, at what rate, in which voltage window, and in a half cell or a full cell.
2. Ask for the upper cut-off voltage and temperature behind any cycle-life claim, and the retention limit it was counted to. ISS cycle counts were at 3.95 V per cell, not 4.2 V.
3. Ask for accelerating-rate calorimetry data on the exact cell: state of charge, cell size and stored energy, detection threshold, self-heating onset, trigger temperature, peak temperature and peak heating rate.
4. For NMC, ask for the nickel-manganese-cobalt ratio (111, 622 or 811); for LFP, the carbon coating; for LCO, whether the particles are single-crystal or polycrystalline.
5. Ask which anode, salt, solvents and additives the data were taken with, and which density is quoted: tap, electrode or crystal.
6. Read the safety data sheet for the exact powder. PubChem notifications carry the same codes for LCO and NMC: H350 (may cause cancer), H360 (may damage fertility or the unborn child), H330 (fatal if inhaled), H334 and H372. Supplier sheets differ: the LCO sheet gives carcinogenicity category 1B and reproductive toxicity category 2, the NMC811 sheet H317 and H351. LNO notifications show H350, H317 and H372. The LFP powder sheets we hold list it as not hazardous.
7. If the cell will fly or sit in a closed volume, ask for vacuum, radiation, thermal-vacuum and venting-gas data. Our data holds none for LFP, NMC or LNO cells.

What do we not know yet?

• NMC has no sourced average voltage or density, and its delivered capacity is held only as full-cell values per gram of NMC.
• LNO has no cell-level thermal-runaway result, no delivered capacity, and a voltage that is a review statement with no cell format. The Sony pack that contains it blends it with LCO, and the sheet does not say whether the nickel oxide is pure.
• No test compares LFP, NMC and LCO cells at the same size, state of charge, detection threshold, rate and electrolyte. There are two heating tests with different cathode sets, and cycle-life figures are held under different rates, temperatures and windows.
• No vacuum, radiation or thermal-vacuum result is held for LFP, NMC or LNO cells. LFP has a NASA ground evaluation only, and LCO has battery-level flight records. Volume change and cold capacity are missing for LCO.
• Not covered: cost and supply beyond what the LCO profile says (cobalt is on the USGS and EU critical lists); state-of-charge estimation on LFP's flat voltage curve; energy per litre; calendar life; low-temperature charging beyond one LFP cell that could not be charged at 10C at -20 °C; and NCA and lithium manganese oxide cathodes. No usable data is held on these.

Questions people ask