Thermodynamics of Information Processing
20.1 Landauer Bound in Practice
The minimum energy dissipation per irreversible bit operation depends on the physical implementation:
- CMOS transistor (2000s-era): per switch (vastly above the Landauer limit)
- Modern CMOS (7 nm node): — per switch
- Adiabatic / reversible logic proposals: — per operation (approaching the limit)
The gap between theory ( eV at 300 K) and practice (— fJ per switch) spans 5—6 orders of magnitude. Closing this gap requires fundamentally different computing paradigms.
20.2 Bennett”s Clock and Reversible Computing
Bennett (1982) showed that a computer can be made logically reversible at every step if it never erases information. Such a computer dissipates energy only during the initialisation of bits and during optional output, not during computation.
A logically reversible computation can be embedded in a thermodynamically reversible process by driving the system slowly enough that it remains near equilibrium at all times. The energy cost is then:
For a quasi-static process: (minimum possible).
Fredkin and Toffoli gates are examples of logically reversible logic gates. Any computation can be made reversible by saving all intermediate results and running the computation in reverse to restore the input tape.
Worked Examples
Example 1: Carnot efficiency
Problem. A Carnot engine operates between and . Find the maximum efficiency.
Solution. .
Example 2: Entropy change
Problem. Find the entropy change when of ice at melts ().
Solution. .
Summary
- First law: ; conservation of energy.
- Second law: ; entropy of an isolated system never decreases.
- Carnot efficiency: (maximum possible for given temperatures).
- Statistical mechanics: ; Boltzmann distribution: .
Cross-References
| Topic | Site | Link |
|---|---|---|
| [Thermal Physics] | A-Level | View |
| [Thermal Physics] | IB | View |
| [Thermal Physics] | DSE | View |
| [Thermal Physics] | University | View |