I am looking for a way to power green LED (like GNL-3012GD) from MFC which provides 0.2mA @ 0.6V for about 1hr (there should be some drop off period then; 2..3hrs to restore). I understand that the amount of power is too small but what about the energy storage to drive the LED for a while (from a few seconds)? Any hints or schematics are highly welcomed.
Okay, here are some considerations.
Your source has the energy storage of 0.6V* 0.2 mA * 1hr = 120 uW-h.
If you get some newer LEDs, say LTST-C193TGKT-5A chip LED from Lite-ON, it has a fairly high efficiency, about 100 mcd at 5 mA., not the miserable 7 mcd @ 20 mA as for GNL-3012GD, or 50 times more efficient. I just run a test, so the light is pretty visible at 16 uA (micro Amp !!!), with forward voltage about 2.3V. This equates to power consumption at just 37 uW, while you can see the light with no mistake. If I take a proportion of 100 mcd and 16/5000, this is about 0.32 mcd of light, and there are many LEDs with 0.1 - 0.3 mcd light output. So the 0.32 mcd is pretty visible.
Now, if you find some "energy-harvesting" IC operating at 500-600 mV, with output boost to 2.3-2.5 V, something like the module from Advanced Linear Devices, or LTC3108, the LTST-C193TG LED will be glowing (at 0.32 mcd) for up to 3 hours.
The hints here are (1) high-efficient LED, and (2) energy-harvesting IC. Would it fit your requirements?
It may be done using no electronic DC/DC converter at all. Just use an electromechanic voltage multiplier.
You need 4 or 5 super capacitors connected in parallel by a multipole switch. Charge the capacitors for one hour. Then operate the switch by hand connecting the capacitors in series and to the LED. May be a resistor in series to the LED for current limiting.
The switch should have a pair of dual throw switches for each capacitor.
There will be no electronic losses, just the very small leakage currents of the capacitors and the switches. Electronic analog switches should not be used, their leakage current is too high.
There will be an RC time constant when charging and also when discharing the supercaps. During charge, the internal resistance of the MFC will be part of the R(C), not only the wiring resistance. The internal resistance of such a low power source may be quite high. About 3 RC time constants will be needed for a 95 % charge. If the capacity of the supercaps is too big, charging will take a very long time, but there is more energy stored to operate the LED longer.
Of course the effective capacity of 4 supercaps in parallel is 4 times the capacity of a single one. The effective capacity of 4 supercaps in series is 1/4 of a single one.
Ran across a Texas Instruments Ultra Low-Power Boost Converter With Battery Management for Energy Harvester Applications. It appears to be an excellent and well documented solution for your project.
End of Update
1 hour run time and 2-3 hr restore time is currently not possible.
Theoretically, with 100% efficiencies, it would take at least 16 hours to harvest enough energy to light an LED for 1 hour.
A harvesting module at this input power would be 25-50% efficient best case.
It appears the MFC cannot sustain the stated 0.6V @ 200 µA and needs few hours to charge its energy capacity to deliver 1.2 mW.
It may be possible if you have a couple of days to harvest the required energy.
You need more voltage. Voltage must be more than the LED's forward voltage.
You need more power, like at least 2 mW, you have 0.12 mW. Even if you were to boost the voltage there would not be enough power.
Harvesting the power would take more than 2-3 hours. You need 16x more power than you have.
Your LED is too inefficient. A 18,000 mcd QT-BrightekQBL7IG30C would do better but still not enough voltage or power.
Not sure what you mean... that is neither enough voltage or current to directly drive ANY led. You could possibly use a boost IC like This To boost the voltage up to power the led then slowly charge a capacitor from that and switch on the led once it is charged. I don't believe that the switching process could be automated as I believe that even a 555 timer would use up more power than that MFC could supply.