Arduino overclocking with Liquid Nitrogen cooling (-196°C). Maximum stable frequency is 65.3Mhz at 7.5-8V supply voltage. Higher or lower voltages are unstable at this clock. Stability was tested by my custom stress test for more than 1 hour. During this overclocking test 3 liters of Liquid Nitrogen were consumed. At frequencies higher than 65.3Mhz it was failing at SRAM read/write test. There were a number of hardware issues with Arduino at cryogenic temperatures: capacitors loose most of their capacitance, brown-out detection was tripping, 3.3V linear regulator for LCD backlight supply was failing. They were solved by soldering capacitors which don't loose too much capacitance (NP0 and X7R dielectric materials), disabling brown-out detection and bypassing 3.3V linear regulator. Also, yellow LED became green (due to wider bandgap at lower temperature) and then stopped working at all - apparently there was insufficient voltage to light it. For comparison, at 5V supply and room temperature maximum stable frequency is 32.5Mhz, at 8V - only 37Mhz.