Response to “Comment on ‘Fundamental limits of energy dissipation in charge-based computing’ ” [Appl. Phys. Lett. 98, 096101 (2011)]

Zhirnov, Cavin, and co-workers ZC presented in Ref. 1 an analysis of the fundamental limits of a binary switch, represented by a double well system, which purported to show that such a switch must dissipate kBT log 2 because the switching event must dissipate an energy equal to the barrier height. We showed in Ref. 2 that their argument was incorrect because by switching the system smoothly using the adiabatic paradigm of Landauer and Bennett, dissipation much lower than kBT log 2 could be achieved. The barrier height must indeed be much larger than kBT log 2 when the device holds a bit of information but can be lowered in the switching process in such a way that much less energy needs to be dissipated. ZC argued in Refs. 4 and 5 that the apparent energy saved by adiabatic switching would in fact be dissipated in the electrodes that create the adiabatically changing barriers. In an argument in Ref. 5, invoking “Cavin’s demon” to charge a capacitor, they claimed that even smoothly adiabatically charging and discharging the capacitance representing the barrier-created electrodes would inevitably dissipate more than kBT log 2 . Because Landauer’s principle LP clearly contradicts this, they argue in Ref. 5 that LP only applies to “hypothetical” systems that are “perfectly isolated from the external environment.” Our results in Ref. 6 show by direct measurement that a capacitor can in fact be charged and discharged adiabatically. The energy stored on the capacitor can be much larger than kBT log 2 while the dissipated energy is much less than kBT log 2 . This precisely contradicts the “Cavin’s demon” argument of Ref. 5 and supports LP. To argue as they do in their Comment that this is somehow off-point because it focuses on charging and discharging a capacitor is incorrect. We quote their paper third page as follows: