Documente Academic
Documente Profesional
Documente Cultură
Lecture Summary:
This Lecture presented a review on the research progress of using molecular-based
devices to build logic gates as well as memory elements. These devices exhibit potential
to be utilized as an alternative to CMOS circuits at nano-scale.
chemically prepared systems. Second, in the process of chemical synthesis and assembly,
it is almost impossible to grow a perfect part with no defect components. This is due to
intrinsic finite chemical reaction yields, whereas any reliable computing machine requires
perfect complexity. As a result, any CAEN will have to tolerate an extremely large
number of defects. Defect and fault tolerance has to be incorporated into design of any
practical CAEN systems. The pioneering work of Teramac by HP lab has provided an
insight into the architectural direction of CAEN systems. Teramac manages to tolerate a
considerable amount of defect components by exploiting FAT-tree architecture.
In recent years, researchers have made impressive progress on the building block of logic
gates and memory elements with chemically prepared devices.
operation voltage and it provides enough margin to prevent molecular switches from
accidental reconfiguration.
Fig. 3 shows the voltage-current curve of molecular switches. Each switch is composed
of a monolayer of redox-active rotaxanes sandwiched between metal electrodes. There
are two states and the switches are irreversibly opened by applying an oxidzing voltage
across the device. The state of the switch is measured with monitoring current flow at
reducing voltage. In the closed state, the current is dominated by resonant tunneling
effect through the electronic states of the molecules. In the open state, the switch can
be seen as a resistor with extremely high resistance and the current flow is very small.
The data given are all measured in air at room temperature. For an open switch, the
measured current at 2V to 1.5V was much less (on the order of 15 to 30) than that of a
closed switch. The difference of high current level and low current level of these switches
is so large that they could be simplified as diodes in the design of logic gates. Fig. 4
shows a simple 2-input AND gate and a 2-/3-input OR gate based on molecular switches.
The implementation is in fact the same as diode-based wired-logic gates, with the
exception that they are electronically configurable and they are able to maintain a diodelike behavior over a wide voltage range.
In short, molecular switches have proved to be viable as building block of logic gates.
However, there are problems to be addressed. First, it is only one-time programmable and
thus cannot be used as reconfigurable logic. Second, there is static current for open
switches even though it is very small. The power dissipation due to static current has to
be studied. Third, molecular switches are inherently molecule-sized in only one
dimension, that is, the width of molecular layer. Fabrication technology has to be
developed to stack the switches in order to reduce area.
As far as system design with molecular devices is concerned, Goldstein and Rosewater
proposed a hybrid architecture which combines molecular self-assembly and
lithographically produced conventional CMOS structure. At nano or molecule scale, all
subprocesses are intrinsically self-directed and only at micron scale precise and reliable
lithographically generated circuitry is possible. Figure 6 shows a proposed nanoBlock
using molecular techniques. Basically MLA is a 2-D grid of reconfigurable molecular
diodes that can be configured into circuits based on diode/resistor threshold logic. Figure
2 is a nanoBlock that is configured into a 2-input AND gate. The reconfigurable nature
of this nanoBlock allows defect tolerance.
However, to build an operational circuit such as a sequential machine, both logic gates
and latches are required. A molecular latch is proposed for this purposed as shown in
figure 7 and 8. As it shows, the latch is load/drive ratio based and it requires pretty
Figure 7. (a) Simple molecular latch circuit. (b) Load-line diagram (c) Simulation results showing latching
and voltage
Figure credit: Digital Logic Using Molecular Electronics
Figure 8. (a) The complete molecular latch and (b) the clocking methodology
Figure credit: Digital Logic Using Molecular Electronics
Unlike in CMOS technology where 3-terminal transistors are easily configured as latches
to hold and restore a signal, molecular switches/diodes are 2-terminal devices and it is not
trivial to construct a suitable latch structure while maintaining low complexity and low
static current.
Molecule-Based Memory:
In addition to logic gates built with molecular devices, researchers are also investigating
memory counterpart with molecular techniques. For molecular memories to compete with
conventional microelectronics, they must be compatible with semiconductors under
extreme conditions required for processing and operating a practical device. Whats
more, they do not degrade under large number of read-write cycles.
Liu and Yasseri etc discovered porphyrin-based molecules bound to Si(100) which
exhibit redox behavior useful for information storage as well as meeting the challenge to
compete with conventional microelectronics. The figure blow presents the molecular
structure of this specific porphyrin.
Figure credit: Molecular Memories That Survive Silicon Device Processing and Real-World Operation, by
Z. Liu, etc.
Figure credit: Molecular Memories That Survive Silicon Device Processing and Real-World Operation, by
Z. Liu, etc.
Figure credit: Molecular Memories That Survive Silicon Device Processing and Real-World Operation, by
Z. Liu, etc.
Conclusion:
The advent of physical limit of CMOS technology has triggered research for molecular
alternative for microelectronics industry at nano-scale. The research has been divided into
two main parts. One is the search for molecular building blocks for logic gates and
latches; the other is for molecular memories. Molecular switches are constructed to build
logic gates and porphyrin has proved to be stable molecule for molecular memories.
However, there are still too many issues (such as silicon compatibility and molecular
manipulation) to be addressed before mass production of molecular chips, which will
certainly motivate enormous follow-up researches.