Nanomagnet Logic (NML) - Field-Coupled Nanocomputing: Paradigms, Progress, and Perspectives - page 25

Fig. 4. Experimental demonstration of an NML three-input majority-logic gate. (Source: Imre
et al., Ref. [ 13 ].)

Thus far [ 16 - 18 ], our work has provided a proof-of-concept demonstration of
NML - i.e. the feasibility of performing digital logic with physically-coupled nano-
magnets. However, so far we have used MFM to read the state of the dots, and
externally generated magnetic fields to switch the dots. Of course, this is not practical
for real applications. Below, we describe on-going work to develop electronic input
and output (I/O) and mechanisms for generating local magnetic fields for ''on-chip''
clocking.
Before proceeding, we want to emphasize that our experimental proof-of-concept
demonstrations satisfy five ''tenets'' that are considered essential for a digital system:
(1) NML devices have non-linear response characteristics due to the magnetic hys-
teresis loop. (2) NML can deliver a functionally complete logic set enabled by the
3-input majority gate and the NOT operation naturally achieved by the antiferro-
magnetic dot-to-dot coupling. (3) Signal amplification/gain greater than 1 has been
experimentally demonstrated by showing the feasibility of 1:3 fanout [ 19 ], where the
energy for the gain is provided by an external clocking field discussed further below.
(4) The output of one device can drive another as the fringing fields from individual
magnets can bias a neighbor. (5) Unwanted feedback is preventable through clocking.

4 Nanomagnet Logic: Towards System Integration

Recent and ongoing work addresses electronic means for both NML I/O and clocking.
Our approach leverages existing MRAM technologies for READ/WRITE operations.
After all, setting an input for NML, i.e. setting the state of an input magnet, is similar
to writing the state of an MRAM bit. Similarly, reading the state of an NML output dot
is just like reading an MRAM bit. These similarities to MRAM suggest that an NML
circuit is analogous to a patterned ensemble of the free layers in an MRAM stack.