I can't be the only one who's ever thought of this, but obviously it hasn't caught on:
In terms of energy density, fossil fuels are the best thing around short of enriched uranium (and, flammability aside, a lot safer). The biggest two problems with using fossil fuel as an energy source are the low thermal efficiency of combustion, and the production of CO2 gases. The second problem is the more publicized, but it can be mitigated to a very large extent by solving the first.
One idea, which may catch on in industrial power generation, is the "carbon-conversion fuel cell", which at least in the lab gives a thermal efficiency of about 80% of the theoretical energy potential of the fuel. However, the energy density (1kW of generation capacity per square meter of fuel cell surface area) and the operating temperature (750*C) make it impractical at the small scale.
However, we currently have technologies that can improve efficiency of combustion engines, we just aren't using them. ICEs typically top out at about 30% thermal efficiency; the rest is lost, primarily as heat, some as kinetic energy of expanding gas. We even go so far as to build cars with water coolant loops and mufflers to more efficiently waste that heat and kinetic energy. Well, what if we instead harnessed that energy to do more work for us before it leaves the car? Imagine a parallel hybrid with two additional generators in addition to the one powered directly off the drivetrain of the engine. One would be powered by steam from the superheated coolant (anyone who's opened the radiator cap on a hot engine will tell you this system is perfectly capable of producing copious amounts of pressurized steam), and the other would work much like a turbocharger, using the pressurized exhaust to turn a turbine, but instead of an intake compressor the turbine would power another generator.
To my knowledge, no car manufacturer has ever fielded a production passenger vehicle that utilizes these additional energy sources to charge the battery system. But these technologies are decades old; one would think that we have the technology to pass the breakeven point on power to weight with these systems. Does anyone have documentation showing these additional avenues of energy capture have at least been considered, and if so, why they're not being pursued more aggressively?