Molecular Interconnects: A Bottom-Up Paradigm for Signal Propagation

Electron scattering stands as a major obstacle to the extendibility of Moore's Law. As reported by researchers at the University of Albany, College of Nanoscale Science and Engineering (CNSE), electron scattering is generated from boundaries of conductors as their cross-section approaches the mean free path for electron scattering (39.3 nm in copper).¹ This scattering leads to signal delay and performance degradation.² The impact could become significant as early as the 32 nm node.
 
To address resulting issues, CNSE researchers are pursuing a molecular interconnect paradigm that exploits ballistic electron transport to eliminate limitations imposed by incoherent (non-elastic) electron scattering.³

Coherent scattering, while contributing to line resistance, does not generate Joule heating in single-quantum channels. This feature holds true even in multichannel systems, provided there is no inter-channel coupling. Datta notes that this effect has allowed experimentalists to significantly increase current density through small conductors.⁴

Therefore, an ideal molecular conductor with optimal signal transmission via ballistic electron transport must display periodicity and electronic overlap between adjacent atomic sites. These characteristics are essential to the formation of extended electron states that constitute quantum conductance channels.

Moreover, the channels must be spatially separated to reduce or eliminate Joule heating via inter-channel coupling. This requirement is difficult to achieve in conventional materials and favors "nanomaterials," where separate quantum channels can be designed with atomic precision.

In addition, it is desirable that the quantum conductor possesses electronic energy levels that can be tuned to match those of conventional metal contacts to ensure they can be integrated into prevailing IC architectures. In terms of fabrication strategy, the reliable and reproducible formation of functional structures requires sequential and hierarchical "bottom-up" fabrication methodologies. A schematic conceptualization of such methodology is illustrated in Figure 1 , which depicts the two distinct self-assembly steps that underline the bottom-up, three-dimensional fabrication strategy.

1. Schematic conceptualization of a “bottom-up” 3-D fabrication methodology depicting two distinct self-assembly steps.

The first step consists of intermolecular self-assembly of basic molecular constituents, displayed as colored tiles in Figure 1a, to produce atomically precise interconnect building blocks that guarantee monodispersity and eliminate dimensional variance on the nanometer scale. The building blocks shown in Figure 1b consist of two components: charge carrier groups and direct attachment functionality groups.

The second step involves intermolecular assembly of interconnect building blocks through molecular-scale recognition mechanisms. This assembly propagates the dimensional stability of building blocks to form 3-D structures by exploiting directed assembly on prepatterned wafers, as indicated in Figure 1c .

The exploitation of molecular recognition paradigms, combined with intrinsically monodisperse molecular fabrication routes, could eliminate dimensional variance at the atomic scale, while maintaining the material and process flexibility required for integration with base templates. The latter could be formed using conventional lithographic techniques.

Pioneering research has shown that ballistic electron transport is achievable in a range of organic systems, including aromatic hydrocarbons (polyphenylene, 4-thioacetylbiphenyl, oligothiophenes, oligocenes, etc.) and conductive polymers.

Corresponding nanoscale device demonstrations using novel electron transport or switching mechanisms have been reported by Aviram and Ratner,⁵ Heath et al.,⁶ and Want et al.⁷ These demonstrations form a strong foundation for the knowledge necessary for successful moletronics applications.

One common feature in virtually all of this work is the dominant role that local molecular ordering plays in electronic transport. Innovative strategies were proposed for enhancing the local molecular order in molecular structures, including thermal treatment.^8,9

Similarly, Liu and coworkers¹⁰ have cited the importance of adjusting quantum channel design to maintain ballistic transport in the presence of electron-phonon effects and avoid loss of ballistic transport in organic systems, as proposed by Avouris and coworkers.¹¹ However, surface and interfacial effects will dominate electronic transmission in organic structures in the deep nanometer regime. These effects are further compounded by structural disruptions in the local organic crystal bonding, thus requiring a fabrication strategy that maintains local molecular order. Similarly, molecular disorder in quantum conductance channels will introduce incoherent scattering centers detrimental to electronic transport.

These considerations result in the emergence of a nanoscale fabrication strategy for molecular systems with three necessary requirements to ensure maximum electron transport efficiency and reproducible manufacturability.

These requirements are:

• Nanoscale structural rigidity of the base molecular unit to secure the attachment of ordered organic groups for ballistic electron transport.
• Intermolecular recognition protocols for atomically precise and reproducible directed assembly.
• Compatibility with prevailing lithographically driven directed assembly to ensure high-volume processing on large wafer platforms.

Flynn et al.¹² and Scheibel et al.¹³ have demonstrated the effectiveness of biologically inspired molecular recognition for nanowire metallization utilizing, respectively, naturally occurring virus assemblies and amyloid fibers. However, to fully exploit the capabilities of these biologically derived molecular constructs, a true "bottom-up" approach is required that permits atomic-level design of customized electronic functionality in addition to intermolecular recognition for high-precision directed assembly.

The CNSE team devised a three-phase molecular interconnect development and integration roadmap (Fig. 2 ). The strategy is driven by the projection that molecular interconnects could be required as early as the 32 nm technology node.

2. The University of Albany, College of Nanoscale Science and Engineering (CNSE) molecular interconnect development and integration roadmap.

Phase I (2003-2006) screens molecular wire concepts to select best candidates, explores self-assembly protocols, and establishes performance metrics. Phase II (2007-2010) develops and demonstrates polylithic (hybrid) integration pathways of molecular interconnects with silicon CMOS. Phase III of the program (2010-2013) identifies and deploys monolithic (singular) "nanochip" solutions consisting entirely of molecular interconnects and devices.

Figure 3a displays a schematic diagram of the three basic building blocks of the molecular interconnect system under development at CNSE. The highly ordered linear arrangement is intrinsically adaptive to coherent electronic transport through quantum channels. Also, the increased molecular weight of the β-sheet stabilizes the structure against ordering defects that have typically plagued molecular crystals consisting of low-molecular-weight organic materials.

3. Schematic diagram of the three basic building blocks of CNSE’s molecular interconnect system (a) and atomic force microscopy (AFM) of a polypeptide nanowire network formed of an aligned linear array of individual self-assembled nanowires on a graphite substrate (b).

Base polypeptide template

At the core of the building blocks of Figure 3a is a flexible, monodisperse and scalable base motif, made of a single molecular "pleated" β-sheet formed from a folded polypeptide chain, synthesized using genetic engineering protocols. The CNSE team has previously reported on a conceptual driver for a one-dimensional crystalline organic motif for use as a base template for "construction" of molecular interconnects.¹

More recently, the CNSE team has successfully implemented genetic engineering protocols to demonstrate controlled self-assembly of the base motif into nanoscale wiring networks. Figure 3b displays an atomic force micrograph of a base template consisting of a polypeptide nanowire network on a graphite substrate. Spectroscopic and morphological analyses reveal that the network is formed of an aligned linear array of individual self-assembled nanowires.

Each nanowire consists of a bilayer b-sheet, with the axis of the sheet parallel to the nanowire axis. The width of individual nanowires is ~25 nm, with an average wire separation of 90 nm. The nanowires are chemically bonded to the nickel substrate through hexahistidine end groups, thus creating a mechanically robust structure stable for subsequent processing.

Controlled self-assembly was achieved through an established processing technique; namely, molecular plating of the purified polypeptide material from an aqueous solution suitable for assembly on metallic or dielectric surfaces. The use of conventional fabrication methodology is designed to ensure ease of integration into prevailing semiconductor process flows.

The remaining two building blocks of the molecular interconnect system consist of predesigned functionality groups. The first is selected to ensure directed self-assembly through preferential (selective) attachment on specific surfaces, and lend added mechanical and structural rigidity to the entire molecular interconnect system. As shown in Figure 3a , silane or thiol groups could be employed for selective covalent attachment to, respectively, SiO₂ or metal surfaces.

The second consists of complexes for charge transport that exploit appropriately designed aromatic complexes (Fig. 3a ) or the high electronic density of metal-containing moieties, such as gold, silver or palladium. Such moieties can be incorporated into the polypeptide backbone during or after directed self-assembly of the base template.

A proof-of-concept study by CNSE researchers employed gold nanoparticles because of their ability to act as spatially identifiable "markers" that can be detected through molecular-resolved scanning probe microscopy. Nanowires were fabricated by molecular plating using an aqueous solution consisting of a mixture of the purified polypeptide material and gold nanoparticles of 3-4 nm in diameter.

Selected results are illustrated in Figure 4 , which displays an atomic force micrograph from the successful attachment of gold nanoparticles to the base nanowire template and not the substrate. These findings confirm that post-assembly metallization can be combined with genetic engineering to form nanoscale interconnect networks with the desired architecture and functionality.

4. Scanning probe microscope images of gold nanoparticle attachment to β-sheet nanowires. No nanoparticle attachment to the substrate was observed.

Electrical tests

Test structures for electrical testing of molecular nanowires were fabricated from a 10 nm thick nickel film grown on SiO₂ to minimize surface roughness. Nickel was selected as the electrode surface to allow covalent attachment of the polypeptide functional end moieties.

Conventional lithography created a pattern consisting of a central contact pad connected via nickel "legs" to five peripheral contact pads. The pattern is shown in Figure 5a , where the nickel legs are labeled "a" through "e."

5. Test structures for electrical testing of molecular nanowires (a), and atomic force micrograph of a molecular nanowire spanning a 90 nm trench in a test structure (b).

Two-terminal electrical test structures were fabricated on each "leg" via ion-beam "milling" to form 90-200 nm thick trenches spanning the leg width. A representative 90 nm trench is shown in Figure 5b after controlled self-assembly of molecular nanowires. A single polypeptide nanowire can be seen spanning the trench. Analysis by non-contact atomic force microscopy revealed that the nanowire was continuous across the two-terminal test structure.

All test structures were electrically characterized prior to processing to confirm open circuit conditions. Electrical characterization of the polypeptide backbone was then performed using scanning tunneling microscopy along a conduction path normal to the plane of the backbone (transverse direction), as shown in Figure 6a . The transverse I-V curve yielded an insulating behavior with a dielectric strength exceeding 1 MV/cm. The nonlinear I-V response was indicative of tunneling across a monomolecular dielectric thin film.

Electrical characterization along a conduction path parallel to the plane of the backbone revealed a non-ohmic response (Fig. 6b ). This response was caused by electron tunneling across the two-terminal test structure, and was most likely mediated by polypeptide material spanning the trench.

6. I-V curves measured via scanning tunneling microscopy normal to the plane of the polypeptide backbone reveal a dielectric strength exceeding 1 MV/cm (a). Electrical characterization along a conduction path parallel to the backbone reveals a non-ohmic response caused by electron tunneling across the two-terminal test structure (b).

Conclusions

Although moletronics is still in its infancy, proof-of-concept studies by CNSE researchers support its potential to manipulate electrical conductors at the atomic scale, leading to the formation of nanometer-scale IC building blocks

that employ novel signal-propagation schemes. Resulting architectures could provide innovative pathways to extending Moore's Law beyond the gigascale regime. The next few years promise to bring exciting prospects for the eventual viability of moletronics at maintaining the historical rate of progress of the IC industry.