Nothing remains on top (or at the bottom) forever. \u201cPico\u201d is three orders of magnitude smaller\nthan \u201cnano\u201d. A rule-of-thumb is that to\nbe able to manufacture or produce something, one must be able to measure to at\nleast an order of magnitude smaller and more likely two orders-of-magnitude\nsmaller. Why? In order to mass manufacture something, one\nneeds to be able to make it so that there is consistency from one item to\nanother. If pieces go together, they\nneed to have a tolerance so that the fit is acceptable. <\/p>\n\n\n\n
There is a story that circulated during the early\napplication of robots that the robots could not do what assembling humans were\ndoing. It turned out that the parts\nbeing assembled had a slight variation in the centering of the pieces. The human worker could make an adjustment,\nbut the robot could not make that adjustment. \nThe tolerances were changed and the robots were \u201chappy\u201d and assembled\nthe parts. The tolerances had to be made\nsmaller for the automated assembly to function properly. <\/p>\n\n\n\n
So what does that have to do with \u201cpico\u201d? Please bear with me on getting to the point why\n\u201cpico\u201d is needed. As the reader is\nprobably aware, the semiconductor industry has been steadily decreasing the\nsize of the size of the circuitry in the semiconductors. The reduction in the minimum feature size\ndimension has been reducing at of rate of 0.7 every roughly 18 months. (A 0.7 reduction in dimension results in an\napproximately 50% area reduction.) This\nobservation was first quantified by Gordon Moore of Intel and has become known\nas Moore\u2019s Law. The current \u201cnext\u201d\ngeneration being developed for semiconductors is below 10nm. <\/p>\n\n\n\n
Semiconductor manufacturing consists of processing many 10s\nof layers, each layer contains different portions of the final circuitry. The material on each of the layers is\nmodified by various process that create the desired features. The process involves coating a material,\nexposing the desired pattern, and then removing or adding specific\nmaterials. <\/p>\n\n\n\n
The features for the semiconductor are produced by\nilluminating a wafer with light that is modified by an imaging mask. The mask contains the features that will\nproduce the desired images. The\nprojection illuminates a resist to create a pattern that can be etched to\nremove the unwanted portion of the surface. \nTypically, the mask has features at a magnification of the final\nimage. The optical system reduces the\nimage to create the desired feature size. \n<\/p>\n\n\n\n
Roughly, the minimum \u201cachievable\u201d image spot definition produced\nby a light source is defined by the wavelength. \nThe wave nature of length produces diffraction patterns, which have\nalternating rings of light and dark. Lord\nRayleigh defined the minimum separation at which it is possible to identify\nseparate points is when the center of the one spot falls into the first dark\nring of the other spot. The actual\nseparation also includes a relationship to the diffraction limit of the angular\nillumination from the lens system. In\nthis discussion, we will use the wavelength divided by 4. (This would require extremely good\noptics.) The current semiconductor tools\nin widespread production employ 193nm wavelength sources. With that assumption, it would appear that\nspots of 50nm could be produced. But,\nthere are other influences that degrade the images. A key factor is aberrations\nor distortions introduced by the lenses. \nAnother issue is that the images produced need to have vertical\nsidewalls and not the slope of two overlapping images. Add to that the fact that resists do not\nproduce smooth patterns at the nanoscale size. \nThe molecules of the resist cause roughness. Then the actual lens system is also limited\nby the index of refraction of the lens material (for transmissive systems) and\nthe numerical aperture, which is a function of the focus angle of the lens\nsystem. The list of imperfections that\ndegrade the final image.<\/p>\n\n\n\n
How do we produce such small feature as are being done today?\u00a0 There are many factors.\u00a0 The application of immersion techniques to the 193nm systems has further reduced the feature size possible.\u00a0 The development and application of mathematical techniques that evaluate the interference of light provides the ability to actually use features on the mask to prevent portions of adjacent features from being imaged.\u00a0 (For a more in-depth understanding of these techniques, publications by Chris Mack are recommended.\u00a0 See reference 1.)\u00a0 These imaging systems, known as Lithography systems, are very large and quite expensive.\u00a0 There are many engineering innovations that have become part of the existing Lithography systems.<\/p>\n\n\n\n
The optics are the key to successful imaging. Surface variations in the lens cause defects\nin the image on the surface. The surface\nsmoothness of a lens from the theoretical design is measured and the resultant\nnumber is called the Root Mean Squared (RMS) error. RMS is basically an estimation of the\ndeviation from the perfect design. The\ncurrent state-of-the-art for high resolution camera lenses is roughly 200nm\nRMS. <\/p>\n\n\n\n
As presented in the Keynote Session presentation at the Advanced\nLithography Conference, the current RMS for the latest Lithography systems in\nproduction is less than 1nm. In this\npresentation, it was stated that the value for the coming image sizes requires\nthe Lithography systems\u2019 RMS to be 50pm. \nThat is picometers. One picometer\nis one thousandth of a nanometer. [Ref.\n2] This talk referenced a slide from\nWinfried Kaiser that showed the equivalent of 50pm corresponds to variations\nacross the length of Germany, across 850 Km, would need to be less than 100\nmicrometers. <\/p>\n\n\n\n
As the feature sizes continue to shrink, the variations in\nproperties will be in the picometer range. \nIt will probably start out as a decimal point and a nanometer\nreference. Nanometers were first\nreferred to as micrometers with a decimal point before the numbers. \u201cPico\u201d is coming to nanotechnology.<\/p>\n\n\n\n
References:<\/p>\n\n\n\n