The Development of Quantum-Dot Cellular Automata - Field-Coupled Nanocomputing: Paradigms, Progress, and Perspectives - page 6

dynamics getting caught, even temporarily, in a metastable state; this was the heart of
Landauer's objection. It retained the advantages of (local) ground-state mapping.
During switching each cell is always very close to its instantaneous ground state (the
definition of adiabaticity). Though we did not fully appreciate it at the time, this
essentially turned QCA into a concrete implementation of the gedanken experiments
which had led Landauer to conclude that there was no fundamental lower bound to the
energy that must be dissipated to compute a bit [ 22 ]. Clocking further allowed much
larger computational structures to be envisioned, combining memory and processing.

3.1 Semiconductor QCA

The Cavendish group of Smith et al. demonstrated QCA operation in GaAs/AlGaAs
heterostructures with confining top-gate electrodes [ 23 - 25 ], as originally envisioned
in the earliest QCA publications. The group of Kern et al. demonstrated a QCA cell in
silicon, using an etching technique to form the dots [ 26 - 29 ]. The group of Mitic et al.
used a novel method to form dots from small clusters of phosphorus donors in silicon
[ 30 ]. They succeeded in demonstrating QCA operation in that system. Interestingly,
their long-term goal is coherent quantum computing and they conceive QCA devices
as providing an ultra-low-power layer of interface electronics to connect a cryogenic
quantum computer to standard CMOS electronics [ 31 ].
The challenges of all semiconductor implementations have been two-fold. Firstly,
the lithographically accessible sizes for quantum dots are large enough that kink
energies are low and cryogenic operation is required. More importantly, the perfection
of the interface and electronic environment becomes an issue. While dots with tens of
electrons effectively screen small amounts of impurity and imperfections, in the limit
of single occupancy, semiconductor dots become very sensitive to the details of the
environment. Even MBE-grown samples have enough random imperfections that the
dot is often not exactly where one expects it to be based on lithography [ 32 ].

3.2 Metal-Dot QCA

Although electronic QCA has been demonstrated in a number of material systems,
metal dot implementations have proven to be the most successful, so far, building on
the fabrication techniques developed for single-electron transistors [ 33 , 34 ]. The
advantages of metal dots are that the fabrication yield is relatively high, and they are
electrically well-behaved, meaning that energy required to add each additional elec-
tron to the dot typically remains constant over the addition of many electrons. This
makes it easier to load the QCA cell with the proper number of electrons and to bias
the cell so that the two polarizations are energetically degenerate. However, in
semiconductor dots [ 33 , 34 ] the fabrication yield is low and the addition energy
typically differs for each additional electron, and the electrical behavior of the dot can
change from run to run, making it difficult to prepare the cell for proper operation.
This makes the metal dot an attractive option for QCA experiments. The main dis-
advantages of metal dots are background charge fluctuations [ 35 ], and low operating
temperature. Background charge fluctuations are caused by the random arrangement