The idea of circuits that can be printed over large areas and onto flexible substrates has gained momentum in the past decade. From the materials standpoint, the most convenient class of semiconductors to form the basis for such circuitry is organic or polymeric semiconductors. Such materials have properties that are inherently suited to this kind of processing; indeed, some of the best polymeric semiconductors are cast from solution at room temperature by ink-jet printing, resulting in devices that have performance characteristics akin to those of amorphous Si transistors.
Close on the heels of individual transistor demonstrations has been the development of circuits and prototype systems that employ organic transistors. This includes electronic paper, in which an array of organic transistors constitutes the backplane that drives display elements such as electrophoretic cells or liquid crystals. Another potential application for organic transistors is in radio-frequency identification (RFID) tags, which are circuits that store small amounts of information that can be read remotely. Both RFID tags and display backplanes involve circuitry that possesses thousands of transistors, so it is not just the performance of individual transistors that is important but how the circuits are designed.
Early Si field-effect transistor (FET) circuits (such as the Intel 4004) utilized only p-channel transistors. Earlyp-FET circuits were followed by n-channel transistor circuits and, eventually, by the hugely successful CMOS (complementary metal oxide semiconductor) FET circuits, which utilize both n- and p-channel transistors. Si CMOS has been successful for several reasons, with the most important one being power dissipation. CMOS circuits dissipate power mostly while the constituent transistors are switching from one state to the other. There is very little static power dissipation, which means that CMOS circuits consume much less power than either n-MOS or p-MOS circuits.
In organic circuits the same holds true, with both measurements and simulations showing that complementary circuits require much less power. Other advantages include ease of circuit design, greater speed, better immunity to noise effects, and greater tolerance of variability in transistor operating characteristics. As an example, in RFID tag circuits, the total power budget needs to be minimized. If complementary circuits are used, then the tag can be read from a greater distance. This in turn will enhance the acceptance of the technology in the marketplace. For the same mobility, complementary circuits are about ten times faster than p-channel FETs and require fewer transistors. This means that the clock rate of display drivers can be increased by a factor of ten, making them acceptable for a wider range of applications. With all these advantages, one would imagine that the case for organic CMOS would be open and shut. However, there are tradeoffs involved. Semiconductor materials for n-channel transistors tend to be more environmentally sensitive than p-channel materials. The past few years have seen advances in improving both the charge carrier mobility and stability of n-channel FET materials. And then there is the complexity of having to pattern two different semiconductor materials in a circuit. This must be taken in the context of the principal advantage of organic transistor circuits: low cost. Anything that adds to the complexity of processing adds to cost. As with many technological systems, there is a cost-performance tradeoff, and organic CMOS, although potentially better performing, will be more expensive to make than p-channel transistor-based circuits.
There have been many attempts to make the processing of organic complementary circuits easy. The traditional complementary circuit design paradigm, which has evolved from Si technology, involves different compositions for the n- and p-channel devices. However, in organics, it is possible to create a device that functions as a p- or n-channel device depending upon the electrical bias voltage. This is done by mixing up the active semiconductor to have two components: one electron transporting and the other hole transporting. Such mixing can be done at the molecular level by using donor-acceptor combinations akin to those employed in solar cells, or at a spatially coarser level. It is also possible to have two thin layers instead of a mixture. All these approaches have been shown to work in laboratory demonstrations. It remains to be seen if these will translate into practical manufacturing processes.
The future will require many more advances from materials chemists and device engineers before organic CMOS becomes a practical reality. New materials have to be synthesized and simple fabrication techniques found to minimize the processing disadvantage and maximize the performance advantage. I am one among those who are betting that one day organic CMOS-based products will be practical!
[1] Ananth Dodabalapur is a professor in the Microelectronics Research Center at The University of Texas at Austin.