Looking at the graphs I think we may both be making the same mistake in interpretation, but in different cases. I was reading the graphs as if the radial distance indicates worst to best as you go outwards, but if it instead means lowest to highest the contradictions disappear. It only matters which it is in terms of cost, and BU doesn't say which it is .JRP3 wrote:Battery University isn't always an up to date resource, but even they show NCA as lower cost than LiFePO4, and though their graph shows LiTiO as least expensive, further down in their summary table they say the following:
Long life, fast charge, wide temperature range but low specific energy and expensive.
Another, more detailed source I have at home, probably dating from 2012 as it mentions the Roadster, LEAF and Volt as being available (Unfortunately while I usually write the name, date and author(s) on a xerox when I make it if it isn't already present, I didn't in this case), does say LTO is more expensive, which agrees with you and BU and also with my thought that titanium is expensive.
I suspect, as BU says, that much of the switch from LMO towards NMC is driven as much by longevity as it is by higher capacity. The same source I mentioned above lists the various Li-ion types, and after giving their voltages and specific capacities (in mAh/g rather than Wh/kg.), leaving out Li-Co-O2 here's what it says of each in the comments section:
LiMN2O4: Most commonly used in automobile, low cost, acceptable rate capability, poor cycle and calendar life.
LiFEPO4: Low cost, improved abuse tolerance, good cycle life, and power capability, but low capacity and calendar life.
NMC: Lowest cost, high capacity, life is less than NCA.
NCA: Highest capacity, low cost, but safety concerns.
Graphite: Most commonly used in all applications, low cost.
LTO: Highest cycle and calendar life, but costly and low in energy density.
Silicon: Still in research stage, high energy [3,700 mAh/g!, versus 372 for graphite and 168 for LTO], but large volume expansion during charging [elsewhere stated as up to 270%] need[s] to be solved.
Another academic source, which I believe was in the same collection of articles I copied the first one from contradicts some of the above. It rates each type (not including NMC) by energy and power density, maturity, life expectancy, cost, safety, operating temp characteristics, usable charge range, and application. As it quotes a battery manufacturing cost study from 2010 it must post-date that. Anyway, here's what it has to say, for each at that time (there've undoubtedly been changes in the interim, especially with NCA and LMO). My comments are in [ ]:
Energy density: NCA 170; LMO 150; LMO/LTO 150; LFP 140.
Power density: NCA Highest; LMO Good; LMO/LTO Good; LFP fair.
Maturity: NCA Most proven[?]; LMO Safety and durability needs proving; LMO/LTO safety and durability need proving; LFP electric monitoring needs proving.
Life expectancy: NCA Good, demonstrated 15 years and 350,000 cycles; LMO OK, capacity fading concerns; LMO/LTO Very good; LF Very good.
Cost: NCA High (cobalt and nickel); LMO High; LMO/LTO Highest; LFP lowest.
Safety: NCA Least thermal stability, high charge thermal runaway; LMO better than NCA, some thermal instability expected; LMO/LTO better than LMO and NCA; LFP Best, least risk of overcharge.
Op. Temp characteristics: NCA [blank]; LMO Capacity fades above 40 deg. C, poor charging at low temp; LMO/LTO Capacity fades above 40 deg. C; LFP Poor cold temperature performance.
Usable charge range: NCA Degrade at high charge, approximately 30-70%; LMO Capacity fades with cycling, approximately 30-70%; LMO/LTO 100% charge; LFP 10-100% charge.
Application: NCA Power-assisted HEVs, because of high power density and cycle life; LMO HEV; LMO/LTO HEVs, because of long life, low energy but high power; LFP PHEV and BEV.
What's of additional interest in this last source is that it includes the most comprehensive cost breakdown for Li-ion battery manufacturing I've ever seen, with three bar graphs which divide costs into 'Materials', 'Manufacturing', and 'Other' [Profit, Warranty, Transportation, Marketing, R&D, Corporate O/H], and the total at that time seems to be about $1,100/kWh if I'm reading it right. Each larger category is further divided into cell, module and pack sub-levels, and each sub-level even more specifically in each stage - for instance, the materials cost graph lists costs for cathodes as Active materials, Binder (PVFD), Additives (Carbon), Aluminum foil (if I've got the divider in the right place) then the anode; Graphite, Binder (PVFD), copper foil here or in the next category; Lithium salt, Polyethylene, Tabs/terminals, Container, Yield adjustment, Enclosure (I think this is at the module level), Terminals, Enclosure (pack level?), Connections, Control/Safety circuitry. Similar divisions are present in the other two graphs.
The imprecision is due to the fact that the original source from which the graphs were taken was probably in color but was printed in this book in black & white, making many of the divisions difficult or impossible to determine, even if I weren't trying to read them from a Xerox.
This kind of data is undoubtedly considered a proprietary trade secret by the companies, so it's not surprising that precise cost data is scarce. One of the most surprising statements to me, and what may well be the largest source of cost reduction since then, even more than economies of scale, is that the article states:
"What stands out immediately from Fig. 4.7 [the bar graphs] is the enormous scrap rate, which is estimated to contribute about 25% of total pack cost [ Note, I believe that this refers to the portions of the graphs labeled 'yield adjustment', as they represent the largest single category that are under the control of the manufacturer]. To help put this measure of wasted material in context, while automotive industry standards require scrap rates below 0.1%, battery manufacturers have been found to operate with scrap rates of up to 60%!"