Ed’s Threads 080407
Musings by Ed Korczynski on April 7, 2008

CNT and graphene dreams may be real
Carbon nano-tubes (CNT) are the only viable (pun-intended) new materials being developed to replace copper as the electrical interconnects for future ICs. There are no known room-temperature superconductors, and optical interconnects require relatively slow and expensive lasers and detectors, and CNTs are the future. The theory and practice of growing CNTs was thoroughly reviewed at this spring’s Materials Research Society (MRS) meeting, and the applications as electronic IC interconnects will be seen at the International Interconnect Technology Conference (IITC) to be held in Burlingame, California in June. The deadline for submitting late news to IITC is this Friday.

Carbon can form an amazing variety of stable crystals and molecules based on different bond energies and angles between atoms. In crystalline form, sp2 electron orbitals can form 2D planes of graphite or sp3 electron orbitals can form 3D tetrahedral of diamond. The 2D form of solid carbon shows very interesting properties when reduced down to less than a few atomic layers.

Graphene is one or two atomic layers only, which results in geometrically induced electron energy-band modification and the ability to form semiconducting devices. Graphene is a great potential “long-shot” technology first reported in January 2006 Solid State Technology…sure to generate many Ph.D. theses and likely to benefit DARPA programs…but still quite a way away from proven as commercially manufacturable. As Gordon Moore reminds us in this recent interview, “The actual idea of an MOS transistor was patented in the mid-'20s,” though it was not until over 40 years later that Intel started making a business out of it.

Take 60 carbon atoms and you can coax them together into a cage-like spheroid called a “buckyball” or fullerene (C60)—initially predicted by R. Buckminster Fuller based on the potential for stable bond-angles in regular polyhedra—which has the same 2D form as graphene. Larger and more complex carbon cage molecules can be formed, and seem to be formed naturally by stars in space. Take a continuous supply of carbon atoms and you can coax them together using a catalyst particle into growing as a nano-tube with that same basic 2D form. You can grow both single-walled CNT (SWCNT) and multi-walled CNT (MWCNT). Both grow off of metal catalyst particles, which must somehow first be deposited in the bottom of vias to form interconnects between lines; making the connection on the top side seems like it will be inherently a bit tricky.

At IITC this year, researchers from MIRAI-Selete and Waseda University (Japan) will show actual integration results for CNT in 160nm diameter vias at temperatures as low as 365°C. The team will report that the CNT fabrication process didn’t degrade a fragile low-k (2.6) dielectric and that the vias sustained a current density as high as 5.0 MA/cm2 at 105°C for 100 hours with no deterioration.

SEM cross-sections of 160nm-diameter CNT vias fabricated with growth temperatures of (a) 450°C and (b) 400°C (IITC2008 Paper #12.4, “Robustness of CNT Via Interconnect Fabricated by Low Temperature Process over a High-Density Current,” A. Kawabata et al.)

One of the reasons that MRS meetings are exciting for materials scientists and engineers is that truly leading results are shown. Oleg Kuznetsov et al.—from Honda Research Institute in Columbus OH (USA) and Goteborg University (Sweden) and Duke University (USA)—presented information on the size-dependence peculiarities of small catalyst clusters and their effect on SWCNT growth. Though exact mechanisms are not fully understood yet, we know that nano-scale catalysts particles play key roles in growth, and that sizes alter growth properties. The general background assumption is a vapor-liquid-solid (VLS) model for growth: carbon in the vapor phase is absorbed into the catalyst particle as a liquid from which solid SWCNT grows out. An observed ‘paradox’ is that with decrease of catalyst size from 3nm to 1nm the required minimum temperature for SWCNT growth increases. Molecular dynamics simulations revealed that reducing the catalyst particle size reduces its solubility of carbon atoms and thereby requires higher temperature for SWCNT growth.

Since the researchers used Fe as the catalyst for SWCNT growth, their rigorous modeling work included a re-working of the classic Fe-C phase diagram where they showed that SWCNTs grow in a liquidous region above the Eutectic point. The Fe-C phase diagram is arguably the foundation of modern materials engineering, since it shows how to make the varieties of steel which are the physical backbone of construction in our age, and is taught in all undergraduate materials science courses. While I haven’t been looking very hard, but this is the first time I’ve seen something new in a Fe-C phase diagram since I left MIT in 1984.

—E.K.