There is a very interesting device called the aluminum-air fuel cell: a consumable aluminum anode (-1.662 V to the oxide, but the relevant oxidation here is the oxidation to the hydroxide at -2.31 V), separated from a porous carbon cathode by a thin porous insulating material soaked with an electrolyte such as sodium chloride. With a caustic potash electrolyte this produces 1.2 V per cell, but table salt provides a wholly acceptable 0.7 V.
Aluminum’s atomic weight of 26.981 538 4(3), its generous three electrons per atom, the electron’s charge of 1.602 176 620 8 × 10⁻¹⁹ C, and Avogadro’s number of 6.022 140 9 × 10²³ atoms / mole together give us 10.727 928 available coulombs per kilogram of aluminum, or about 7 or 8 MJ/kg at 0.7 V. This is a respectable fraction of aluminum’s energy density as a fuel, 31 MJ/kg! (When burned in oxygen.) It’s probably as good as you could expect from fueling a steam turbine from aluminum and air, say. This is astonishing because batteries normally don’t come anywhere close to heat-engine energy-density territory.
Of course you can scale it down in a way that you can’t scale down a steam turbine: a gram of aluminum should provide you with 7 or 8 kJ, and only 13 mg of aluminum is necessary to provide 100 J.
Amateur aluminum-air batteries commonly use a copper current-collector grid on the carbon cathode rather than nickel, but I suspect that will suffer anodic corrosion to copper chloride over time. Replacing the copper wires when you replace the aluminum anode should not be too hard, but neither is plating them in nickel, if you have some; maybe lead and/or tin would also work.
Removing the gelatinous hydrated aluminum hydroxide may be more difficult; maybe some sodium fluoride or monosodium phosphate would work for that if they don’t corrode the aluminum fuel itself, but then they become additional consumables. (There’s also the possibility that tridentate citrate complexes might help.) Mixing some ethanol or isopropanol into the electrolyte might encourage the hydroxide to de-gel without creating too much toxicity or fire hazard.
I’ve been trying to figure out what kind of small generator would work to provide, say, a laptop with long autonomy; I’ve been looking at model-airplane two-stroke diesel glow engines and things like that, since those seem to be the smallest heat engines around, but it’s hard to find information about their efficiency, and they’re also messy and noisy and don’t scale down to low power. In Dercuano I concluded that under ten milliwatts on average, during use, was adequate for a responsive interactive computing experience, mostly to update the screen. 12 hours a day of usage for a month at 10 mW works out to 13 kJ, which is only about 1.6 kg of aluminum, so maybe the aluminum fuel cell can solve my problem. It certainly seems to scale down better than heat engines do.
It would probably be difficult and somewhat dangerous to get high current output from such a battery, for all that aluminum is easily available in 10-μm-thick foil. I think you’d have to finely powder it, with all the risk of class-D fires and possible waste via air oxidation that that would entail. But you could likely get it to work.
Magnesium is another possible anode fuel for such a battery; GE produced such a device in the 1960s, using the same NaCl electrolyte. I think it does not have the problem of fouling the anode with a sticky gelatinous hydroxide, but I think its specific energy is lower, because a magnesium atom has but two electrons to give for its battery; but the voltage is higher, which might compensate.