by Rud Istvan
This post addresses issues related to ‘vehicular decarbonization’. It is an energy storage insider’s narrative of how tough a slog developing some of the requisite applied science technologies has been over the past decades. It is a saga of research twists and turns, abject failures, near misses, and ‘before its time’ inventions.
In the US (e.g. California ZEV credits) and Europe (e.g. hybrid incentives) there has been a push to electrify automobiles to reduce CO2 emissions. As an extreme example, Germany’s Bundesrat just passed a resolution that would eliminate all fossil fueled new cars by 2030. Whether or not Germany’s upper house has lost all sense of Mercedes/BMW/VW reality, the push to ‘decarbonize vehicles’ clearly exists independent of whether it is necessary for climate’s sake. Is it even practically feasible? Tesla’s Elon Musk thinks so, based on $4.5 billion of Tesla investment plus another $3.9 billion from Panasonic for his Gigafactory. But Teslas are expensive, slow charging (~4 hours at 240V), and range limited compared to ordinary cars. They are mainly a rich person’s virtue signalling toy. Tesla has consumed $8.4 billion of capital with no financial return in sight yet.
There are three vehicle electrification steps moving away from conventional internal combustion engines (ICEs): hybridization (e.g. the Prius), range extended electric vehicles (e.g. the Chevy Volt), and full EVs (e.g. Tesla or the new Chevy Bolt)
Full hybridization like the Toyota Prius works in several ways. Engine idle-off saves 5-8% depending on urban traffic. Regenerative braking saves 9-10%. The additional power of the electric motor enables two further major changes. First, the ICE can be downsized, saving both weight and fuel. Second, it can be converted from the Otto cycle to the Atkinson cycle. Atkinson cycle ICE saves about 15% in fuel economy, but with a significant torque loss. That doesn’t matter in a full hybrid; the electric motor provides missing torque. The 2016 Prius couples a 121 hp 1.8 liter Atkinson I4 with a 71 hp electric motor for 192 hp combined. Its new lithium ion battery (LIB) is just 0.75Kwh and weighs only 54#. Recharging is from the alternator or regenerative braking while driving. Prius comfortably seats 5 along with 24.6 cubic feet of cargo space (65 cubic feet with the rear seat folded down). Range is 633 miles from ~52 mpg. 2016 price is >$24,200. From launch in 2010 through yearend 2015 Toyota has unsurprisingly sold ~1,170,000 Prius hybrids.
The 2016 Chevy Volt is powered by two electric motors providing only 149 hp fed from a 18.4 Kwh LIB providing a marketed ~50 mile electric range, up from 40 in the previous version. The original all-electric range was chosen because about 2/3 of US urban trips are under 40 miles. With a 240 volt charger, Volt recharging takes 4.5 hours (with 120 volt charging, it takes 13 hours). The battery is warrantied for 8 years or 100,000 miles. The LIB battery weights 405# (189kg) and is a 5.5 foot long T shaped monster. The range extending gasoline engine is a 1.5 liter 101 hp I4 driving an onboard 54 Kw generator. With a full tank of gas and a fully charged battery, Volt range is ~408 miles. Seating is essentially only 4, and cargo capacity is only 10.6 cubic feet. For those middling vehicle values the price is >$33170. Unsurprisingly, Chevy has only sold about 117,000 Volts since 2010 (the same launch year as Prius, so a fair comparison).
The MY2017 Chevy Bolt is a pure EV. The marketed range is 238 miles from a 60Kwh LIB (actual is supposedly ~208). The battery weights 960# (435kg) and its 285 liter volume covers the entire floor of the car. Each hour of 240 volt charging provides 25 miles of range. The warranty is Volt-like. Bolt uses a single 200hp (150kw) electric motor. Seating is essentially 4 and cargo capacity is only 16.9 cubic feet. Price is >$37495.
The preceding examples illustrate three fundamental facts. (1) Full hybridization makes commercial sense, and sells well to fuel economy minded consumers. (2) The extra weight and cost of a conventional engine and generator to solve EV range anxiety is a poor Volt tradeoff. (3) The Bolt’s newest chemistry LIB is still insufficiently capable and too expensive to provide a competitive range at a competitive price.
This post identifies technical solutions to issues (2) and (3) that might also make commercial sense. A bit technical, but perhaps some important insights. Intentionally lots of google click bait for those wishing to learn more. Just the usual links/images for those who don’t.
Technology 1 for fact (2): FPLG
Free piston linear generators (FPLG) are not a new idea. The first patent was issued in 1934 as US2362151. The basic idea is simple. Remove unnecessary ICE connecting rods, crankshaft and bearings, at least some valves plus cams and/or rockers, the necessary connecting rod/crank/bearing/valve oil lube system, and the conventional rotating generator attached to the crank. Just have a magnetic piston in a combustion chamber energize a linear generator as it travels back and forth. Simple, small, light, cheap, and efficient.
But FPLG was not technically feasible until the development of powerful rare earth permanent magnets (for the piston component) beginning in 1966. A patented idea before its time. These permanent magnets still weren’t commercially viable until rare earth magnets improved in the 1990’s for portable electronics hard disk drive and speaker applications. The past decade has seen significant further price/performance improvement driven mainly by wind turbines and hybrid vehicles. FPLG are now conceptually commercially viable.
In the conventional ICE/Generator market, there is no need for FPLG. There is no size/weight/efficiency constraint on gensets that must run outdoors.
But for range extended EV’s like Chevy Volt, there definitely is—a new potential market for an old but only recently feasible technology. It is not surprising that among several entities now working on FPLG, Toyota already presented a working prototype in 2014.
Toyota’s prototype has a single fairly complex combustion chamber with a gas spring piston return. The 24” long by 8” wide prototype generates 10 Kw with 42% fuel efficiency. A typical vehicle ICE is 26-28% efficient, but with parasitic losses (e.g. transmission, oil pump, engine friction) only 15- 20% energy efficiency is delivered to the wheels.
Launched in 2014, Israeli startup Aquarius Engines has developed a simpler (than Toyota) single piston dual combustion chamber FPLG with ‘novel airflow for reduced emissions and enhanced cooling’.
The pictured Aquarius FPLG production prototype develops 35Kw from just 70kg at a demonstrated fuel efficiency of ~40%. Peugeot is planning several range extended EV prototypes using Aquarius FPLG for initial 2017 road testing.
FPLG works technically. The remaining automotive grade reliability, emissions, and control issues look more like engineering developments than further technology breakthroughs. FPLG might solve the fundamental size/weight/cost range extending genset problem evidenced by the Chevy Volt.
Technology 2 for fact (3): LIC
Lithium ion capacitors (LIC) are also not a new idea. In 2007 and 2008, Subaru presented (at the 17th-18th International Seminars on DLC and Hybrid Energy Storage Devices) working asymmetric prototypes combining features of supercapacitors and LIB into a single electrochemical device with very attractive hybrid properties.
Some simplified energy storage basics in the following two paragraphs for interested denizens, which otherwise can be skipped.
Capacitors store charge electrostatically. There are two basic technologies. The common one is an insulator sandwiched between two conducting plates. This is the basic technology of all electronic capacitors, with a technology lineage dating back to the Leyden jar in 1745. They are very fast, can be made very small and cheap, and can last trillions of cycles—but they only store tiny amounts of electricity. The other technology is supercapacitors (aka ultracapacitors or EDLCs) based on Helmholtz double layer capacitance, the electrostatic mechanism producing thunderstorm lightning. Supercaps can store thousands of Farads per cell, charge or discharge in ~1.5 seconds, and last over 1 million cycles. Energy stored is a function of voltage squared, and for reasons beyond the scope of this post aprotic supercaps are limited to about 2.8 V; LIB is ~3.5 V. The best current supercaps have about 20 times the power density of current LIB, but only about 1/50 the energy density. Their commercial advantage is very long cycle life at very high power density; the market is ~ $1 billion.
Batteries store electricity in Faradic electrochemical reactions. There are again two basic technologies. With pseudocapacitance there is no chemical species change; in true battery chemistries there is on at least one electrode (e.g. lead to lead sulfate and back in PbA). LiB is the most energy dense second type commercially available, but has an inherently limited life to at most a few thousand charge/discharge cycles (Chevy Volt is ~ 5000 cycles from 80% SoC to 35% SoC at a 1C rate).
Subaru was looking for a replacement to standard lead acid batteries (PbA) that would have a significantly enhanced cycle life and more energy density without excessive cost. Subaru’s motivation was an under hood battery for mild hybridization (just engine idle off and regen braking, for example the Valeo system) that did not also automatically kill battery life. They used a standard activated carbon for the cathode, lithiated graphite for the anode (with a very clever first charge lithiation scheme using lithium metal mesh), and standard LIB LiPF6 as the electrolyte salt. The result was a 3.8 volt device (better than LIB) with a demonstrated 20,000 cycles (95%SoC to 45%SoC at a 40C rate) at 80°C! But, Subaru then decided LIC enabled mild hybridization did not make commercial sense (added cost > ~15% fuel saving). So they licensed their LIC technology to JM Energy. The product is sold as the Ultimo and is used in specialty applications like heavy duty UPS (backup/reactive power/peak support in heavy manufacturing). A near miss despite Dr. Hatozaki’s R&D success.
The energy density limitation of supercapacitors that LIC seeks to overcome is directly related to the effective surface (per gram or cc) upon which the Helmholtz double layer can form, and to the voltage at which they can operate. Activated carbons have high total surface areas, but surprisingly low effective surfaces. (Full disclosure. My NanoCarbons invention cost effectively increases activated carbon effective surface about 50% using three patented tricks. And will likely be rendered much less valuable by what follows.)
Growth of vertically aligned closely spaced multiwall carbon nanotubes on a metal current collector via CVD provides very high effective surface (an MIT Ph.D thesis), but is difficult to scale and very expensive.
The 2009 MIT spinout company that attempted to develop this technology for EV’s has received tens of $millions in DARPA and DOE grants, but has struggled to get beyond very high priced very small niche specialty markets. It survives, barely, on continued government R&D support rather than product sales.
When Geims got the 2010 Nobel Prize for discovering graphene, it was surmised by many that graphene based structures could solve the effective surface problem more easily and cheaply than vertically aligned carbon nanotubes. Graphenes are essentially single atom sheets of carbon (like an ‘unrolled’ single wall nanotube, only with greater XY area). They are extremely strong, highly conductive, and fairly easy to make. Graphene Energy (spun out of Ruoff’s group at U. Texas Austin) investigated this energy storage possibility. Ruoff converted graphite oxide to graphene in an aqueous solution using acid. Their problem was that the resulting graphenes clump thanks to Van der Waals force, and the effective clump surface was no better than NanoCarbon but much more expensive. Graphene Energy failed and folded.
What this failed company’s research suggested was that some inexpensive way to make a robust unclumped graphene structure might be a path forward.
Given that background, imagine my shock (reading last week) that Henrick Fisker has just founded a new electric vehicle company plus a new ‘battery’ subsidiary, Fisker Nanotech, claiming >400 mile battery range plus very rapid charge time in a hybrid lithium battery/graphene capacitor device. The HOLY GRAIL according to MSM PR! For those denizens who do not know about him, Henrick Fisker is a famous supercar designer (Aston Martin DB8 of James Bond movie fame, amongst others). He started an electric supercar company before Tesla. Alas, the sourced batteries exploded over 100 times in his Karma cars (really bad karma). Then his LIB supplier A123 Systems (spun out of MIT) imploded into bankruptcy losing $250 million of US subsidies and grants plus $100 million for investors, after being sold to China for ~$200 million. Fisker Automotive quickly followed, whose investors lost an additional $1.4 billion.
Can there be any credence to Fisker’s newly announced phoenix like rise from his EV ashes? He has funding, so somebody believes. But then, many somebodies believe Elon Musk. The credibility question requires untangling a fascinating technology development web that leads to a new LIC technology. The patent application for Fisker’s PR’d low cost graphene ‘machine’ is not yet published (but when it does, it surely won’t be the linked MIT CVD approach that has already failed with vertically aligned carbon nanotubes—more below). The related predecessor graphene/MnO2 hybrid device providing the necessary Fisker clues just published as US20160148759. It provides very informative clues, but is of no direct vehicle significance because only a 2 volt system.
In what follows we deduce what Fisker is up to. There are several demonstrated subparts, producing a combined plausible breakthrough. Each is another self-contained energy storage R&D mini-saga.
Thread one is the invention of laser scribed graphene (LSG) in 2012. Then UCLA Ph.D student El-Kady in Prof. Kaner’s nanotech lab made the LSG breakthrough. He took ordinary graphite oxide, coated it onto an ordinary DVD disk using water, then ran the dried DVD disk through an ordinary commercial HP DVD Lightscribe. (Lightscribe used a 780nm (infrared) 5 mW LED laser to inscribe a DVD label/illustration onto a DVD surface coated with heat sensitive dye, each scribe track about 20 microns wide, total full disk pass for a full ‘label’ about 20 minutes. HP has since stopped selling the technology because it is monochromatic and not durable. A near miss.) The LSG process produces about 8μ thick 3D graphene structures in DVD sized sheets via simple laser heat reduction of graphite oxide to graphene. These graphene films are extremely mechanically robust because of 3D interlinking. He further showed that six passes of the Lightscribe laser (each ~20 minutes per dvd) improved conductivity many fold. He made a high effective surface, mechanically robust, highly conductive graphene structure for supercaps. Ph.D granted along with a major Nature paper. This was well reported and discussed at the ISDLC conference in 2012. We ‘experts’ discounted it 4 years ago, because the Nature paper showed the electrode thickness was only ~8 microns and the reported supercap energy density was nothing exceptional in the aqueous phosphoric acid electrolyte at maximum 1 V.
Thread two is the subsequent 2015 El Kady and Kaner development of an asymmetric hybrid device based on LSG. Their new hybrid combined LSG graphene carbon supercapacitance with (subsequently electrodeposited nanoparticle) MnO2 pseudocapacitance. Total voltage 2 V, up from 1 V. Still not a lot of stored energy, but perhaps interesting for specialized niche applications like transdermal drug delivery via electroporation according to the UCLA PR. Yawn.
Thread three is from recent LIB research. Lithium titanate has been an object of intense study for several years as a safer, energy denser alternative to traditional intercalating graphite for LIB anodes. There is a big problem. The material’s conductivity is very poor, so its power density is inadequate, and the charging time far too long, even for cell phones and laptops. Graphene is extremely conductive. So this research focused on somehow incorporating conductive graphene into the bulk of lithium titantate at a nano-level in order to improve anode conductivity. There have been two recent seminal research ‘breakthroughs’. Both use nanotechnology and the idea of graphite oxide plus chemical precursors to lithium titanate, with the final material mix formed in a single heat treatment synthesis. One paper used an aerosol process. The other paper used a sol gel process. [Guo et. al., Electrochemica Acta 109: 33-38 (2013), available outside paywall via google as an MIT.edu posting.] These newish papers present two different lithium titanate precursor mixes together with graphite oxide for simple subsequent heat synthesis.
Fisker Nanotech has not said anything specific about their ‘battery’ other than it uses graphene and lithium (their new patent applications are not yet published). My SME supposition is that they have a new hybrid asymmetric LIC. A mechanically robust LSG graphene cathode plus a mechanically robust hybrid graphene/lithium titanate anode synthesized in one step from triple precursors using an LSG analog process rather than the literature’s sol gel or aerosol. Much easier and cheaper than Subaru’s graphite lithiation. And likely still ~20000 cycle life at a 40C charge rate for much faster EV charging while still meeting vehicle life equivalent ‘battery’ life.
Fisker says they also have a patent pending machine to make 1000 Kg (/day?) of graphene electrode at $0.10/Kg. That may be a bit hyped, but is not implausible by simply ‘thought experiment’ reengineering of LSG in light of the two LIB lithium titanate anode papers already cited. The commercial Lightscribe 780nm 5mW laser has a track width of 20 microns. It took 6 20 minute disk spins to reach optimal graphene conductivity. Fine for simple lab proof of principle for a Ph.D thesis. Not fine for volume production. But there are cheap commercial solid-state diode 780nm lasers with up to 2 watts (2000 mW) power each. Rather than a lens concentrating the laser power as in the Lightscribe, it could be a lens dispersing 2000mW over a larger area with enough power for 1 pass heat treatment as in the sol gel and aerosol papers. Lightscribe hit 20 microns track width with 5mW 6 times for perhaps a millisecond each for an optimal graphene electrode; that is a total of 30mW on 20 microns for ~6 milliseconds. A 2000mW 780nm laser could hit a 1.3 millimeter stripe with the same total power at the same scan speed. Or an even wider track with a slower scan rate (as is likely for a bulk production machine).
Imagine a paper machine like system. The furnish box equivalent is continuously spreading a water based graphite oxide plus 2 lithium titanate precursors slurry onto a rapidly moving continuous plastic support belt equivalent to the dvd. First step beyond the furnish box, evaporate the furnish water with radiant heat and fans. Second step, IR heat nanosynthesis by powerful 2W spread focus 780nm lasers to convert GO to graphene and the lithium titanate precursors to interspersed lithium titanate nanocrystals. This finished material is still supported by the rapidly moving continuous plastic belt. Third step, peel off and spool up a finished continuous electrode sheet as wide and long as wished as the support belt turns under at the end of the machine for its return trip. Imagine a second identical machine making the graphene only electrode by simply leaving out the lithium titanate precursors. Big rolls, made very fast and very cheap. Spooling up very thin electrodes, but made continuously in bulk. Not aerosol or sol gel or CVD small batches.
Imagine assembly of Volt like prismatic pouch ‘battery’ cells. Cut the electrode materials to size before or after stacking as many layers as wanted; they are very conductive so simple contact likely suffices. No backing metal current collector is needed like for LIB and supercaps (a cost and weight saving). Attach a current collector to one end (the hybrid MnO2 patent application describes simple silver soldering at the connection point). Place a standard LIB separator. Now place as many of the other electrode layers as wanted. Attach another simple current collector. Encapsulate in pouch, fill with electrolyte, seal—just like Chevy Volt cells.
Form a battery pack similar to Volt/Bolt with interleaved aluminum heat extraction plates. Done except for the ‘battery’ control electronics.
The basic cell and battery production steps have already been developed by GM. Continuous sheet electrode production is analogous to conventional papermaking, just substituting purpose build evaporation/ LSG for the draining mesh belt/calendaring of papermachines. Every other needed technology element has been shown in the lab. Thanks to optics and LED infrared lasers, scale up appears to be a matter of straightforward engineering rather than more invention.
The revolutionary new LIC supposed here and strongly hinted by Fisker last week does not have to enable a 400 mile EV range. Perhaps the disappointing Chevy Volt points to a different ‘best’ commercial path. Provide say a 100 or 150 mile EV only LIC range covering most trips. That minimizes battery cost, size, and recharging time. Use inexpensive, small, lightweight FPLG to provide extended range capability to 400 miles as with conventional car equivalents. All existing vehicle electrification tradeoffs would need to be fundamentally re-evaluated.
Fisker lost $1.4 billion on his Karma. The successful key to his new venture is probably not his new supercar design. That is the sizzle. It is likely a revolutionary new super ‘battery’ LIC. That is the steak. And every step of the way has already been successfully technically demonstrated.
Moderation note: As with all guest posts, please keep your comments civil and relevant.