Why SAMs?

 

Koomey’s Law observes that the power efficiency of computation doubles every 18 months. While this trend drives down the costs of data centers and makes mobile and embedded devices more practical, it ignores the separate and enormous energy-cost of production. Williams et al.1,2 calculated that manufacturing a 32 MB chip with a mass of 2 g (and an area of 1.24 cm2) consumes 1.6 kg of chemicals. Furthermore, the energy cost of producing an average desktop computer is four times as much as the computer consumes over three years of continuous use, casting the same carbon footprint as a refrigerator does over ten years of continuous use. This footprint is largely materials-based; of the 803 kWh of energy that goes into producing a silicon wafer, 540 kWh is consumed just in purifying the silicon. An average semiconductor fabrication facility then consumes 2 kW/m2 to fabricate an integrated circuit from that wafer. For comparison, an average household consumes < 10 W/m2. Academic facilities consume the same energy per-area, but the lower volume of production translates into a far greater per-unit energy-cost. In 2005, one cleanroom fabrication facility at Harvard University consumed 26, 715, 045 kWh/yr, which is 5.2% of the total energy consumption of the entire university (∼ 500, 000, 000 kWh/hr). Of that energy consumption, 66% powered electrical tools, 95% of which comprised tools for conventional (e.g., photo and electron-beam) lithography. The total carbon footprint of a modern microelectronic device is two tons of CO2 per kilogram of silicon. For comparison, gasoline produces about two kilograms of CO2 per liter.

Moore’s Law observes the doubling of the density of transistors every 18 months. This doubling is driven by reducing feature-sizes and parallelization (e.g., multi-core processors), but smaller features demand higher voltages to compensate for noise/crosstalk. These higher voltages generate more heat, which is exacerbated by the increasing density of transistors. The inability to dissipate this heat at a sufficient rate is a major contributor to the prediction that Moore’s Law will be violated by 2020. Thermal limitations become even more restrictive in mobile/embedded devices that rely on passive cooling. And since clock (switching) speeds saturated about ten years ago, the only place to turn to continue the increase in computing power that drives Information Technology—and in turn many aspects of modern civilization—is new materials. The recent explosion in research into graphitic materials reflects this need, but even if graphene-based semiconductors become a reality before 2020, they will rely on same top-down lithography techniques and will suffer from the intrinsic energy- costs of modern semiconductor fabrication.

Another milestone in 2020 is the prediction that the manufacturing of feature-sizes below 10 nm will become a reality. This size-regime marks the upper limit for molecular devices, which are based on electron tunneling and therefore do not suffer from the same problems of heat dissipation, cross-talk, and low signal-to-noise as feature-sizes are scaled down—Molecular Electronics (ME) has the potential to carry Information Technology past the physical limitations of modern semiconductors by combining organic synthesis, nano-fabrication, and self-assembly. Moreover, molecular systems that are designed to self-assemble in a way that defines the nanometer-scale features of molecular-electronic devices eliminate the high-energy-cost steps of conventional lithography, enabling the development of ME as a sustainable technology. While molecules may not replace microprocessors in the near future, the speed of digital data transmission scales with switching speed. The fiber optic backbone of the Internet is driven by simple opto-electric switches where molecular systems can be employed as drop-in replacement.

Often dubbed “unconventional nano-fabrication,” the tools that we use to electrically address molecules involve no photolithography, clean rooms, and very little specialized equipment. As such, they are inherently environmentally friendly and the materials used are non-toxic. (Nanoskiving does not even require electricity.) Self-assembled—i.e., SAM-based—molecular-electronic devices invert the top-down paradigm; while 800 g of chemicals is consumed for every 1 g of silicon, a SAM of 1 km2 can be formed from 1 kg of a molecule. The largest mass of chemicals consumed in industrial syntheses is solvents, which—depending on the reaction—can be recycled with near 100% efficiency. The largest energy-cost of the production of pharmaceuticals (i.e., complex organic molecules) is actually heating, ventilation and air conditioning (HVAC), which leaves considerable room for improvement. As such, the consumption of energy by the pharmaceutical industry has, in aggregate, fallen over the past five years, even though production has increased; by contrast, the consumption of the semiconductor industry has grown.

Over a century ago, Tyndall, Fourier, and Arrhenius recognized that the booming technologies of their time, driven by the Industrial Revolution, would have unintended consiquences for life on Earth by changing the composition of the atmosphere. We are now seeing transformations occurring to the technologies bequeathed to us by the Industrial Age in the form of electric cars, solar-thermal power plants, wind farms, etc. In our group, we see the combination of unconventional nano-fabrication, and the departure from top-down photolithography, with tunneling junctions based on SAMs as a nacent technologies that will be part of a similar transformation of Information Technology into a sustainable part of society as we move into the next Age.

 

Thanks to Darren Lipomi for providing a lot of the data/references.

 

 

[1] Williams, E. D.; Ayres, R. U.; Heller, M. Environ. Sci. Tech. 2002, 36, 5504–5510.

[2] Williams, E. Environ. Sci. Tech. 2004, 38, 6166–6174.

 

 

 

Tags: 

Zircon - This is a contributing Drupal Theme
Design by WeebPal.