06/27/2002
Solubility
After last week''s column on colliding branes and the cosmic consequences, I find that only a couple of weeks ago there was almost a collision that would have really rocked our little speck of the cosmos. An asteroid "the size of a soccer field" sailed by our planet only 75,000 miles away from us. That''s over three times closer to us than the moon! To me, the disturbing thing was that we didn''t even know the asteroid was in our area until after it had passed us by! Brian Trumbore''s Week in Review for last week also expressed concern about this near catastrophe.
Let''s turn away from such monumental topics to the other extreme, the miniaturization of the transistors on the silicon chip. A few weeks ago, I wrote about the centennial meeting of The Electrochemical Society (ECS) in Philadelphia. The next ECS meeting is in Salt Lake City, a venue I first visited over 50 years ago. Four of us graduate students at Pitt drove to California and back and we stopped to swim (actually, float) in the Great Salt Lake. I came out covered with salt and thought that the lake must be a saturated solution of sodium chloride in water. A saturated solution is one in which a solvent (water in this case) has dissolved as much as it can of a solute (salt here).
A year before the Salt Lake float, I was engaged in an ill-fated Master''s degree project to measure the solubility of barium sulfate in water at different temperatures. Barium sulfate is the compound used in so-called "barium" enemas. Aptly, my aspirations for a Master''s degree were flushed away by the failure of that project. Later, I received my Ph.D. for another totally different kind of project. Ironically, at Bell Labs I would be best known for my work on solubility.
At Bell Labs, the solvents I studied were solids such as silicon. How can silicon be a solvent? Well, just as we have liquid solutions involving solvents like water, we have solid solutions. In 1960, I published a paper titled "Solid Solubilities of Impurity Elements in Germanium and Silicon". Recently, a friend asked if I had seen my picture in the May 2002 issue of the Journal of The Electrochemical Society (ECS). The article, by Howard Huff, is a retrospective of activities in electronic materials and devices over the past 50 years and cited my 1960 paper.
The photo of yours truly followed pictures of such eminent figures as Bardeen, Brattain and Shockley gathered around the first transistor, of Jack Kilby, recent Nobel Prize winner for his invention of the integrated circuit, and of Gordon Moore of Moore''s Law fame. My likeness in such stellar company certainly puffed up my ego. However, I was quickly depuffed upon reading an article in the April 2002 issue of Nature by P. M. Voyles and coworkers at Bell Labs and Agere Systems.
This article did not include my photo or even mention my name. What was worse, they were talking about silicon that contained ten times more antimony than I said was soluble in my 1960 paper. How could they get more antimony into silicon than that? Could it be that my data were wrong?
Let''s not panic, I tell myself. Let''s go back to Salt Lake. The solubility of common table salt, sodium chloride, is 36 grams in a hundred grams of water at room temperature. For most salts, as with antimony in silicon, the solubility goes up with temperature. For sodium chloride, the solubility at the boiling point of water is almost 40 grams in 100 grams of water. Let''s take a solution of 40 grams of salt in 100 grams of boiling water and cool it down to room temperature. The excess 4 grams of salt will precipitate out of solution until we have only 36 grams in solution.
But, if we use a scratch-free beaker and keep all dust and other dirt out of the solution, we might just be able to cool that solution down to room temperature and keep the excess 4 grams of salt in solution. Without anything to "seed" precipitation, we might maintain this "supersaturated" solution for some time. However, it won''t take much to get precipitation started and supersaturated solutions aren''t typically very stable.
Let''s turn to silicon and make a silicon layer saturated with antimony at a thousand degrees. When we cool the silicon down to room temperature, where our silicon chip operates, the solid solubility of antimony is much less. We expect the antimony, like the salt in water, to precipitate out, forming clumps of antimony. But silicon is a solid and water is a liquid. Big difference! When you cool down your salt solution, it''s still a liquid and the sodium and chloride ions can move around and get together with others to form precipitates pretty readily. With silicon, the antimony atoms sit down in their places (sites) in the silicon crystal lattice and are "frozen" in those sites as the silicon cools down. Their ability to move is not very great and it might take many years, perhaps centuries, before enough antimony atoms can get together to form a precipitate. We have a supersaturated solid solution that is stable enough in our lifetimes for the silicon chips in our computers to perform quite nicely.
I know many, perhaps most of you are saying, "Bortrum, who cares and what about miniaturization of devices that you mentioned?" Well, solubility and supersaturated solutions may play an important role in the fate of Moore''s Law, which has "driven" the semiconductor industry for some 37 years. There is concern that the number of transistors on a silicon chip soon may not double every couple of years. This in an era when security concerns require increased computer capabilities in fields such as facial or speech recognition, faster analysis of e-mail or other communications, etc. Obstacles to a continuation of Moore''s Law include such things as just the manufacturing problems in making these devices any smaller. The "wires" connecting these devices on the chip also must be made smaller. And when the transistors get too tiny they may not act normally but weird quantum effects will take over. Just a few of the problems.
Where does solubility come in? There is also a need to make the silicon more electrically conducting as the devices get smaller. One means of doing this is to put more electrons into the silicon. The more electrons, the more current the silicon can carry. Antimony is what is known as a "donor" impurity - it has an extra electron that it can donate to the silicon. Stick in one antimony atom and you get one more electron. Stick in 10 antimony atoms and you get 10 electrons. To really jazz up the conductivity, add lots of antimony!
But Voyles and coworkers are talking about amounts of antimony that are over ten times greater than I said was soluble at any temperature! They''ve had to use trickery that fools the silicon into accepting lots more antimony atoms than should dissolve in silicon. One method they''ve used is "molecular beam epitaxy". In this method a layer of silicon doped with antimony is laid down atom by atom on a silicon wafer. Apparently, if the temperature is low enough, the antimony atoms don''t know that there are more of them being incorporated into the silicon layer than there should be. As with our supersaturated solutions, they get frozen in, so to speak.
But trickery only goes so far. It turns out that, at these very high concentrations, they don''t get one electron for every antimony atom. Electrical measurements indicate that there are only about 70 percent of the electrons there should be. In other words, about 30 percent of the antimony atoms are slackers and not doing their share. Upon reading this, I felt better about my 1960 data and that these antimony atoms were trying to precipitate out from solution.
Well, Voyles and colleagues were quite clever and managed to use a sophisticated form of electron microscopy to actually image individual antimony atoms within the devices. Sure enough, it seems that the antimony atoms were trying to get together to form a precipitate. What the electron microscope studies showed was that there were clusters of antimony atoms forming. The studies indicated that about 30 percent of the antimony atoms were in clusters of just two antimony atoms, possibly combined with what is known as a vacancy, a missing silicon atom. Now, 30 percent is the just the number of missing electrons.
While I really don''t hope that solubility is the problem that brings down Moore''s Law, I admit that it''s satisfying to see that people haven''t been able to flout the concept of solubility with total abandon. Nature can only be tested so far! Now I can get back to worrying about those wandering asteroids.
Allen F. Bortrum
|