A Path to Nanolithography
- PDF / 4,199,808 Bytes
- 7 Pages / 576 x 777.6 pts Page_size
- 106 Downloads / 177 Views
Nanolithography F. Cerrina and C. Marrian
Parallel or Sequential? It is yet unclear what type of new devices, if any, will replace the metaloxide field effect transistor in the sub0.1-ju.m domain. Since in any case, the development of quantum-effect devices requires smaller and smaller dimensions for operation above temperatures of a few milliKelvin, we can safely assume that high-resolution patterning steps will always be required to manufacture the devices themselves. Alternative approaches (such as the use of self-assembling systems) have not yet reached a convincing level of demonstration. Furthermore we can assume that the complexity of the circuits will continue to increase because this is the true driving force of miniaturization. In order to process large amounts of information in a short amount of time, the processing circuit must be correspondingly complex. Hence the future quantum devices will continue the development pattern of modern electronics, leading to the fabrication of large chips with very small devices—that is, exceedingly large processing power. The apparent insatiability of our appetites for more memory and processing makes this prediction an almost certain evolution of the current technology. The fabrication of integrated circuits is in many ways similar to the modern printing process. The press artwork (the mask in semiconductor lithography) is prepared with painstakingly complex procedures creating the copper plate engraved with the text (pattern) to be printed. This plate is then used in the high-volume printing presses (steppers) yielding accurate color overlay, without misalignment of the image levels that would compromise the quality of the printed page. The production volume of the printing industry is such that it would be folly to think of generating the same amount of material by writing each page independently and sequentially. Indeed one can observe the semantic difference between printing—that is, the
56
large-volume replication of a preprepared page—with writing—that is, the direct production of the final page, letter by letter. The same paradigm applies to the electronics industry where the volume of production is too overwhelmingly large to allow direct writing of circuits. The same relation continues to exist between the preparation of the original artwork and its high-volume replication. Even considering one of today's advanced products, say a 256-Mbit dynamic memory, we are faced with daunting production volumes. A factory may have 1000 wafer starts a day, each wafer carrying 16 chips. Each chip carries about 300 million devices, with features as small as 0.25 ^m written on a chip that is 25 mm on a side. If the chip is represented as a bitmap, it will contain perhaps 1010 pixels. In reality the "writing" tool must have a finer resolution so that the bitmap may be 100 times as large— that is, 1012 excels (exposure pixels). These need to be transferred to the recording material in just 1 second—that is, with a throughput of 1 THz. As the successive generations increase
Data Loading...
