05/17/2006
Weighing Small - Scanning Large
Last week, I wrote about Michael Brown at Caltech and his discoveries of objects rivaling or exceeding the size of the planet (or non-planet) Pluto. Another Michael at Caltech pursues a smaller objective. Michael Roukes hopes some day to measure an object weighing about one yoctogram! You say, “What the devil is a yoctogram?” That’s what I said when I saw the term in an article by Robert Service in the May 5 issue of Science. A yoctogram is 0.000000000000000000000001 gram (10 to the minus 24th grams). A hydrogen atom weighs 1.66 yoctograms.
Roukes and co-workers, in the April 4 issue of Nano Letters, report the world’s most sensitive mechanical scale. They have weighed something as light as 7 zeptograms. (A zeptogram is a thousand yoctograms or 0.000000000000000000001 grams.) How do they do it? Consider a string in an instrument such as a guitar or violin. Pluck the string and it vibrates at a specific frequency or pitch. The thicker the string, the lower the pitch. Shorten the string by placing a finger on it and the frequency goes up, a higher pitch. The frequency depends on the weight of the part of the string that’s vibrating.
During a concert, the string may shed or pick up bits of dust. The change in weight should affect the frequency. However, the dust weighs so little compared to the weight of the string that you don’t detect any significant change in pitch. But suppose you make a super thin, super short string and place it in a vacuum chamber. Now pluck the string. If the string is of a nano size, it will vibrate at a very high frequency you can’t hear, nearly microwave frequency. While it’s vibrating, squirt in a spray of relatively heavy xenon atoms or nitrogen molecules. If you keep your string cold, some of the atoms or molecules sit down on the string. For our miniscule string, the extra weight lowers the frequency enough that the change is detected by your sensitive frequency-measuring apparatus.
This is essentially what Roukes and his colleagues did. By measuring how much the frequency changes, they can calculate the extra weight. Their limit of detection of 7 zeptograms is equivalent to being able to detect about 30 xenon atoms. This is also about the mass of a typical protein molecule. The “strings” in their experiments were tiny rods of silicon carbide a micron long and a hundred or so nanometers in width. (I just measured a hair on my arm as between 30 and 50 microns. A micron is 1,000 nanometers.) They “plucked” the silicon carbide rods electronically by placing them in a magnetic field and passing current through them. The sideways force in the magnetic field simulated the plucking.
Roukes and Kamil Ekinci, a former postdoc and now a professor at Boston University, and their colleagues are working to get smaller slivers and to improve the sensitivity of their frequency measuring apparatus. They are optimistic that they will be able to weigh that individual hydrogen atom. They also foresee the possibility that in the future assemblies of these tiny vibrating tuning forks will be used to weigh and determine the presence of various proteins in applications such as identifying proteins associated with cancers or other maladies.
There’s another type of effort going on at nearby Stanford University that sort of falls between the Caltech projects of the very large and the very small. This effort involves taking something very large and condensing it into something very small. In the May 14 New York Times Magazine section, the feature article “Scan This Book” by Kevin Kelly is a provocative treatment of the projects currently underway to create a “library of libraries”. This utopian library would involve scanning all the world’s books and incorporating them into a digital library that could be housed on a relatively small number of disks, perhaps even in the iPod of the future. The ultimate library would, of course, contain not only all the books but also all the TV programs and movies and all the songs ever recorded plus all the art ever painted, as well as all the Web pages ever published.
According to the Times article, in the course of human history there have been “published” over 32 million books, 25 million songs, half a million movies, 3 million TV shows, videos and short films, half a million images and a hundred billion public Web pages. Let’s just worry about the books. I visualized the scanning operation as being a painful process requiring a person to turn each page, flatten it on the scanner and push the button. Wrong. Stanford University is scanning its collection of some 8 million books utilizing a robotic scanner made by a Swiss company, 4DigitalBooks.
Go to the company’s Web site (4DigitalBooks.com) and you can watch pages being turned. The machine turns pages at a rate of 1,500 to 3,000 pages an hour without the need for a human attendant, except to place and remove the books! The robot is also delicate enough to handle older rare books and can handle both hard and soft cover books. It’s only costing Stanford $30 a book, according to the Times article.
China is also scanning. A Beijing company called Superstar has already scanned all the books from 900 universities in China; well over a million different titles. In China, the cost is only $10 a book. Do you smell outsourcing? You bet. In 2004, Prof. Raj Reddy of Carnegie Mellon University gathered 30,000 books from the Carnegie Mellon and Carnegie libraries and shipped them off to China. Carnegie Mellon’s “Million Book Project” is a multinational effort involving universities and institutions from China, India and the U.SA. The goal is to finish scanning a million books by 2007. As of last November, over 600,000 books had been scanned, according to Carnegie Mellon’s Million Book Web site.
When the Million Book Project is finished the million books will take 50 petabytes of disk space. (I had to look up petabyte – it’s a quadrillion (10 to the 15th) bytes.) A double-sided DVD can store 17 gigabytes. I figure it would take 3 million DVDs to store the million books. I’m skeptical that the ultimate library will ever fit into an iPod but I could be wrong.
To make the ultimate library worthwhile, you want to be able to search the library for various titles or topics, just as you can do now on the Internet. To do this, the scanned pages in the library can’t just be photographs of the pages. The letters and words have to be identified as what they are. An “a” must be recognized as an a, etc. Enter the problem of Optical Character Recognition. The scanning software has to recognize the letters and words as such. In older books especially, the images might be poor and 100% accuracy is impossible.
Of the over 600,000 books scanned in the Million Book Project, only about 135,000 were in English. The rest were in Chinese, “Indian”, French, and other languages. India has over a hundred spoken languages and 17 different scripts. The number of characters the scanning software must recognize is in the hundreds. Not only that, but in a couple of Indian languages the characters in a word are connected by a line that makes separation of the characters very difficult. It’s a huge task!
A parting thought. When that ultimate library gets shrunk down to the size of a few disks, NASA clearly must send a spacecraft containing copies of these disks out into space beyond our solar system. Zillions of years from now, some unsuspecting creature from another universe may find the disks and be able to read this column and watch reruns of Seinfeld. I’d love to see his or her or it’s reaction!
Allen F. Bortrum
|