Semiconductors are relentlessly marching down Moore's curve. I just read the announcement about Intel's new packaging architecture, the Bumpless Build-Up Layer (BBUL), allowing 20GHz processors in 5 years. However, on the mechanical side of things, miniaturization is also following a steadfast path to the micro scale. Soon we may even see the nanoscale.
On the microscale side of things, we have micro electro-mechanical systems (MEMS). These generally are built from the top using common lithographic techniques, such as optical aligners, CVD, Etch, etc. However, nanoscale devices, while they may evolve from top-down process in the short term, must eventually use bottom-up processes.
Tops down processes describe the conventional semiconductor manufacturing processes. These are used today in the processing of MEMS devices. One of the most popular processes is the multi-user MEMS process (MUMPS) process developed through work done by the Berkeley Sensors and Actuators Center at the University of California. This is a 3-layer process that uses silicon dioxide and polysilicon to create pumps, motors, sensors, actuators, and other useful machines. A complete description can be found at:
Cronos is a spin of from MCNC and was acquired by JDS Uniphase in April 2000.
This basic scenario has been extended to 5 layers. While this complication necessitates a more complex design cycle and lengthier prototyping, the versatility of 5 layers is much better than for 3 layers. A description of the 5-layer process, called Summit V, can be found at:
These kinds of simple machines will quickly find their way into our electronic systems, the integration of which is a simple extension of common semiconductor processes. Nanoscale systems, however, will require a change in paradigm.
MEMS engineering is a boon to mankind. We will see a proliferation of these devices in the next few years. Nanoscale devices, however, will probably arrive in 10 years or so. These devices are envisioned to work on the scale of one or two nanometers - about the size of a few large molecules. Using light, or even electron beams, from above cannot get the resolution necessary for that precision. For instance, the resolution of a modern TEM at relatively high energy is about 0.2 nm. Since we need positional accuracy of better than 10%, this type of beam would be able to pattern 2 nm or greater line geometries. However, at higher beam currents, this resolution decreases considerably. To make devices at high throughputs, we need high beam currents, so this technique will run out of steam at several nanometers, far beyond the realm in which applications are being contrived.
In this month's issue of Scientific American, the magazine is devoted to nanotechnology. In particular, there is a debate between two scientists regarding the viability of machines that can operate at the nano scale. These scientists are K. Eric Drexler (pro) and Richard E. Smalley (con). Dr. Drexler is the author of Engines of Creation and Nanosystems: Molecular Machinery, Manufacturing, and Computation. He is also the chairman of the Foresight Institute and research fellow of the Institute for Molecular Manufacturing. Dr. Smalley received the Nobel Prize in Chemistry in 1996 for the discovery of fullerenes. He is the Gene and Norman Hackerman Professor of Chemistry and Physics at Rice University.
Drexler envisions an "assembler" that will be able to arrange atoms or molecules into definite patterns under program control. These assemblers will also be able to make copies of themselves which is an absolute requirement for volume manufacturing as we will see below. In the article, he does not deal with the various mechanisms by which these assemblers might be made, but refers instead to his 1992 text, Nanosystems. He does offer a defense against the critics of nanophase manufacturing as proffered by scientists that include Smalley to be described below.
However, in another article in the Scientific American issue, several ways of building small are covered in "The Art of Building Small," by George Whitesides and Christopher Love. Whitesides, a professor of chemistry at Harvard, and Love, one of his graduate students, work on unconventional methods of nanofabrication. Aside from conventional methods, as mentioned above, they describe several other top-down methods that may be used in future nanoscale devices:
Whiteside and Love also describe some bottom-up methods. They feature two of the most prominent methods being investigated, employing nanotubes and quantum dots.
Nanotubes represent a large molecule, by nanophase standards, that can be manipulated easily. Tremendous research is being done on these structures which can be made as single-walled nanotubes (SWNTs) or multi-walled nanotubes (MWNTs). It has been demonstrated that they can be picked up, moved, bent, stretched, and broken by using AFM probe technology. An excellent write-up of this research is found at:
Zyvex's objective, as found at their website, is to hire bright people to conduct research, develop technology, and design products related to the mission of being the industry leader in adaptable, affordable, molecularly precise manufacturing. Check out their webpage - they have an intimidating array of industry talent working for or with the company.
Quantum dots are crystals that contain only a few hundred atoms. According to quantum mechanics, the smaller the spatial confinement, the wider the separation between allowed energy states. Therefore, quantum dots only emit one wavelength when excited - i.e., the energy peak is very sharp. They can be programmed to emit a given wavelength by controlling the size. Thus, they can find many applications from markers in biomedical applications to quantum lasers. Quantum Dot Corporation has several illustrations - look at:http://www.qdots.com/new/homeB.html
There are two ways to manufacture quantum dots, (1) masking and etching and (2) chemically growing nanocrystals. For the first method, using e-beam sensitive resist, small arrays on the order of 500Å can be patterned on semiconductor material with a buried layer of quantum well material. For the chemical method, cadmium selenide (CdSe), cadmium telluride (CdTe), or indium arsenide (InAs) crystals are formed in a solution containing a polymer surfactant. The organic compound prevents the crystals from clumping and it controls the growth and size of the crystalline particles. In both cases, the applications utilize the quantum affects of the crystals by taking advantage of the quantum tunneling properties.
Aside from nanotubes and quantum dots, there is another bottom-up technology that Drexler and many others envision. It is finding appropriate molecules that can perform definite function in the building of devices. A common example is to look at how proteins are formed and encoded by messenger RNA and ribosomes. By engineering molecules that can capture molecules, rotate, add, twist, and displace, it should be possible to make anything from scratch.
We can see that there are many approaches to actualizing nanoscale technology. We can see a path, but how far will the path take us? This where Professor Smalley comes in.
Smalley points out that making anything on the atomic scale will need to deal with Avagadro's number of atoms - 6.23 X 1023 atoms. If we could attain the prodigious rate of a billion operations per second, it would still take a nanobot 19 million years to make a mole of anything. Of course, if we could have the assemblers replicate themselves, then we could attain incredible rates of production - assuming that the self-replicated assemblers are in some useful array. Given that they are feasible, Smalley opines his concern that there is a probability that they could run amuck. How could we control them? But isn't this just some futuristic trepidation? It does not really address the feasibility.
More to the point, Smalley addresses the forces and complications that arise from moving stuff around at the nanoscale. Grabbing an atom or molecule will necessarily run into the forces acting from everything to which the particle is connected. This could be 20 other atoms or molecules. Such complications may require that the assembler have multiple degrees of freedom. In addition, the "hands" will have to overcome tremendous forces whether ionic, van der waals, or covalent. Once having grabbed the particle, how would it then release it?
More importantly, since the manipulators must be comprised of atoms, there is a certain minimal size to them. This, in turn, limits the smallness of the atom or molecule that must be positioned.
Drexler deals with these objections in his article. His defense, he states, is contained in his books and in the calculations of scientists too bulky to include in a magazine article. Whether his final vision can be realized or not is not of great concern. Something will be realized. In its inexorable march to smallness, our technology will find a way to pay for itself. In each step of the path, applications will fall out. The state of nanotechnology may be quite different that what is envisioned by Drexler, futurists, and science fiction writers. But it is only in beginning the journey that one gets anywhere, n'est-ce-pas, mes amis.
Copyright October 2001 by REM Enterprises. Inc.Return to REM home page.
Send mail to REM Webmaster with questions or comments about this web site.