science/technology
AN INNOVATION ENGINE FOR LUCENT
lion, nearly 12% of which was plowed back into R&D at Bell Labs, its "innovation engine." The scientists of Bell Labs, who number 26,000 in the U.S. and 19 other countries, are reenergized, confident, and blazing new trails. At Bell Labs' headquarters in Murray Hill, N.J., long-time members of the re search staff still remember the day the split was announced. "It was shocking be cause it was unexpected," says a chemist who's worked there for 23 years. Today, research in the physical scienc es (defined here as chemistry, physics, ond marriage. The calm has returned after photonics, silicon electronics, and materi Ron Dagani a tumultuous period that began in 1984 als science) no longer predominates as it C&EN Washington with the court-ordered breakup of AT&T's once did, but is balanced more evenly he people who helped transform the Bell System monopoly, creating seven re with work on mathematical and software sciences, according to James W. Mitchell. 20th century with the transistor, the gional telephone companies. solar cell, the laser, communications During the next decade, AT&T under A chemist, Mitchell heads the Materials, satellites, long-distance TV transmission, went other upheavals. Fundamental re Reliability & Ecology Research Lab where and cellular telephony—among many oth search at Bell Labs was scaled back, and most of the chemically oriented research er bold technological innovations—are greater emphasis was placed on applied is done. Plans call for the software, mathe ready to take on the next century. research and on meeting business objec matics, and communications sciences sec The next 100 years may well be shaped tives. Then, in 1996, AT&T spun off its tor to expand to more than 50% of the Bell by newer wonders recently fashioned by computer business (NCR) and its systems Labs research pie, he adds. the scientists and engineers of Bell And although fundamental reLaboratories, the research and devel | search still goes on at Bell Labs, opment arm of Lucent Technolo % most projects are now more closegies. Consider, for example: is ly tied to Lucent's business inter^ ests in communications and net• A technology that promises § working. For many researchers, the efficient delivery of high| that's fulfilling because they can quality speech and music over | see the impact of their work on Luthe Internet. ί cent's bottom line much sooner. • An electron-beam lithogra S Lucent also seems to care more phy system, four generations * about the research—and the reahead of technology currently 1 searchers—at Bell Labs than AT&T used to manufacture computer ever did. It's a closer and warmer chips, that can make transistors relationship, according to the lab's and integrated circuits with fea employees, who speak of the emer tures just 250 atoms wide. gence of "a new spirit," a sense • Plastic transistors that are far that "the excitement is coming less expensive to make than cur back." Indeed, says chemical engi rent silicon transistors and could neer Jorge Valdes, head of the Pro be used in products such as flexi cess & Chemical Engineering Re ble computer screens and creditsearch Department, "Lucent is the card-sized smart cards. best thing that ever happened to • Experimental "bow-tie" microBell Labs." lasers so small that hundreds would Chemist Glen Kowach Inspects a sample of zirconium fit on the head of a pin and emit tungstate, a ceramic that shrinks when heated, after Lucent is a high-technology com highly directional beams of light sintering It In a high-temperature furnace. pany, and it recognizes that funda with more than 1,000 times the mental research is necessary to power of conventional, disk-shaped micro- and technology business (Lucent). Most of drive innovations. "If you're not doing re lasers. These high-power microlasers Bell Labs was carried off as part of Lucent, search, you can't be a high-tech company could increase the speed of voice, video, and AT&T became largely a provider of 10 or 20 years down the road," says chem Internet, and other data transmission via long-distance telephone, data transport, ist Eisa Reichmanis, who heads the Poly existing fiber-optic networks or could be and wireless services. mer & Organic Materials Research Depart come the basis of entirely new architec Today, two years after that split, Lucent ment. The commitment at Bell Labs to tures for local-area optical networks. is the largest communications equipment fundamental research is stronger now This is just a sampling of the many in company in the world, with 136,000 em than ever, she adds. At the same time, novations that Bell Labs, once a part of ployees in more than 90 countries. In fis notes physicist Cherry A. Murray, direc AT&T Corp., is now making available to cal 1998, which ended Sept. 30, Lucent tor of the Physical Research Lab, "You Lucent in what appears to be a sunny sec generated revenues of more than $30 bil have to talk to people in the business
In both fundamental and applied research, Bell Labs forges ahead with a new spirit and an eye on the bottom line
T
24 NOVEMBER 30, 1998 C&EN
the Computing Sciences Research requires high temperatures, stringently Center, sees Bell Labs as a special controlled environments, and precision place where "magical results" hap- optics. Furthermore, organic transistors ofpen. "People arc absolutely excellent fer mechanicalflexibility,easier tunability, in individual areas of science, and and good compatibility with a variety of we're encouraged to work together," substrates, includingflexibleplastics, all of which will likely open up new markets for she says. "It's a rare environment." The center in which Wright works them. is part of Computing & Mathematical At Bell Labs, as well as other labs Sciences, one of the three large re- around the world, researchers have for search divisions that together define years been exploring ways to make thinthe scope of work at Bell Labs. A sec- film transistors using an organic semiconond research division, Communica- ductor instead of silicon. According to tions Sciences, focuses on wireless chemist Howard E. Katz, the most promcommunications technology, net- ising hole-transporting and electronworking, and multimedia and soft- transporting alternatives to silicon are ware systems. certain polythiophenes and fluorinated The thirdresearchdivision, Physi- phthalocyanines, respectively. cal Sciences & Engineering, is home Substituted polythiophenes with a very to most of the chemists, chemical en- regular ordering of the monomer units gineers, physicists, electrical engi- have become "workhorse" semiconducneers, and materials scientists at Bell tors because they can be applied to surfacLabs—550researchersin all. A large es from solution in a continuous manner, Physicist Cherry Murray, director of the Physical fraction of the projects in this division Katz points out. This capability allowed Research Lab, Is the first woman In the history of chemist Zhenan Bao, postdoctoral fellow Bell Labs research to reach the director level, the are materials oriented. Materials of all kinds—organic, inorganic, and com- Yi Feng, physicists Ananth Dodabalapur organization's third-highest management rank. posites—with potentially useful elec- and Andrew J. Lovinger, and chemical enunits to find out what they will need in tronic, magnetic, optical, and mechanical gineer V. Reddy Raju to create "the first properties are being studied with an eye high-performance plastic transistor" in the future." Two labs, for example, are structured toward future applications in communica- which all the essential components are to support specific business units: The tions, data storage and transmission, and printed directly onto a plastic substrate [Chem. Mater., 9, 1299 (1997)]. They Silicon Electronics Research Lab, direct- computing. ed by physicist David J. Eaglesham, is One of the main thrusts in materials sci- used a multistep screen printing process tied to the microelectronics business; the ence is the quest for organic or polymeric (a low-resolution approach) to sequentialPhotonics Research Lab, led by physicist materials that couldreplacethe inorganic ly lay down the electrodes, the polythioAlastair M. Glass, supports Lucent's opti- materials that are the mainstay of today's phene, and the other components. The cal networking business unit. Other labs, electronic and photonic technologies. At smallest feature size achievable with this though, provide more general support to Bell Labs, organic materials are being ex- approach is 75 urn, Bao says. a number of business units. ploited for such applications Xina Quan, a chemical engineer who as thin-film transistors, holoheads the Polymer & Chemical Engineer- graphic storage media, and ing Research Department (part of the Pho- optical fibers. tonics Research Lab), says there is a good Researchers who are relationship between the labs and the bust working on the "plastic" or ness units. Because of these connections, re- organic transistor do not exsearchers in some departments are used to pect it will make the conjuggling long-term projects that arc more ventional silicon transistor fundamental in nature with short-term obsolete any time soon. As a projects that may involve solving a prob- switching device, the silicon lem with a product that is not performing transistor appears to have a as well as expected in the field. secure long-term future in At the same time, the work environ- high-performance or highment at Bell Labs has an academic flavor, density devices such as but the research is much more inter- and computers. multidisciplinary. "You don't have all the But for many less detalent that's necessary in any one particu- manding applications, such lar shop," Mitchell points out. Most re- as smart cards or electronic search projects are "a joint effort requiring labels, organic transistors expertise across a number of disciplines." promise some striking adIn addition, he says, Bell Labs researchers vantages. Organic devices have always collaborated with universities, would be easier and less ex- Clockwise from top, chemists Zhenan Bao and Howard and those ties are being strengthened. pensive to make than silicon Katz and device physicist Ananth Dodabalapur look at Margaret Wright, a mathematician in devices, whose fabrication test results on an organic semiconductor. NOVEMBER 30, 1998 C&EN 25
science/technology More recently, the Bell Labs team has tions for commercial systems because of its tem to demonstrate—for the first time, been exploring higher resolution meth- physical properties, processibility, and cost. they believe—high-capacity holographic With the recent advent of polymeric re- storage of digital data in photopolymer ods for fabricating plastic transistors. The impetus has been to reduce the feature cording media, much progress has been films of the thickness required for high size of the device because that will in- made in high-density holographic storage storage densities [Opt. Lett., 23, 7910 crease its switching speed and decrease in the past three years, Mitchell says. Still, (1998)]. The team included physical he adds, "we're a long ways from having chemists Lisa Dhar, William L. Wilson, and its power consumption. Harris; organic chemist Marcia Schilling; Earlier this year, chemist John A. Rog- anything that's commercializable." Undeterred by the challenges, physical electrical engineer Kevin Curtis; and other ers, working with Bao and Raju, combined screen printing with a micromolding ap- chemist Mark J. Cardillo, head of the Pho- coworkers. The recording film they used proach developed in chemistry professor tonic Materials Research Department, and was prepared from a mixture of a difuncGeorge M. Whitesides' lab at Harvard Uni- physical chemist Alex L. Harris, head of tional acrylate oligomer, the monomers versity to produce a transistor with fea- the Materials Chemistry Research Depart- ./V-vinylcarbazole and isobornyl acrylate, tures as small as 1 urn [Appl. Phys. Lett., ment, and their colleagues have been and a photoinitiator. 72, 2716 (1998)]. Compared with the fast- working on a holographic system for perThey were able to achieve high storest commercially available silicon tran- manent storage of data—the so-called age density by recording multiple pages sistors, which have features as small as WORM system (Write Once, Read Many). of digital data (480 kilobits per page) in 0.25 pm, the organic transistor's l-um fea- The system uses thick (250- to 1,000-um) the same small volume of material. In tures are still relatively large. Nevertheless, photopolymer films, which are prepared this process, known as multiplexing, the says Rogers, "1 urn is about as small as by compressing an oligomer-monomer holographic structure of each page is inyou'd ever want to go with these organic mixture between two flat pieces of glass termixed with that of the other pages. systems. It gives you just about everything and partially polymerizing it under illumi- The pages were then retrieved with low you want in terms of performance for the nation. The remaining unreacted mono- bit-error rates. mer is consumed during two subsequent sort of applications we're interested in." In other experiments, Curtis, Dhar, and Rogers, Bao, and coworkers also are polymerization steps, which accomplish coworkers achieved a storage density equivalent to 5 gigabytes of data on a 5V4working on a high-resolution printing the recording. technique that may be more suitable for During recording, the interference pat- inch disc—about the same amount of incommercial production of organic micro- tern from two interfering laser beams (one formation that can now be stored on the electronic devices. And Katz is investigat- of which carries the information) causes 5V4-inch DVD (digital versatile disc). "Our ing spraying and casting processes using some of the free monomers to undergo goal is to store 100 to 200 GB on a 5%thiophene oligomers. The key point, gelation in a spatially modulated pattern. inch disc," Harris says. That's the storage Katz emphasizes, is that these experi- When the residual unreacted monomers capacity that would be commercially intermental processes do not involve high- are allowed to diffuse to other sites and esting. To achieve this, they will have to vacuum, high-temperature, or other then are locked into place by polymeriza- move to new materials and new designs steps that would interrupt a continuous tion, an irreversible refractive index grat- for the optics system. In their latest results, printing process. ing is created that encodes the informa- which will be revealed this week at the Plastic transistors won't necessarily per- tion in the polymer. A laser can then Materials Research Society meeting in Boston, a new family of materials has allowed form better than their inorganic counter- "read" the encoded information. parts, scientists say, but they will offer atEarlier this month, Bell Labs research- them to store the equivalent of about tractive new options. For example, Doda- ers reported that they had used this sys- 50 GB on a 5^-inch disc. balapur and coworkers recendy integrated a polythiophene-based transistor with an organic light-emitting diode (LED), creating "the first organic smart pixel" [Appl Phys. Lett., 73, 142 (1998)]. Such pixels could be the basis of new types of luminescent displays. For example, because organic transistors can be made transparent i and bendable, they could be incorporated I into an automobile's windshield along with the LEDs that form the display. "Plastic electronics will extend the realm of electronics into new areas and lead to lots of new applications that we don't have at present," Dodabalapur tells C&EN. Polymeric materials also promise to transform the realm of photonics. Scientists have been trying for decades to develop systems for storing information in three-dimensional holograms. The traditional medium of study has been lithium From left, ceramic scientist David Johnson, chemist James Mitchell, and chemical engineer niobate (IiNbOj), but it poses serious limita- Jorge Valdes take a break on the roof of Bell Labs' Murray Hill, N.J., headquarters. 26 NOVEMBER 30, 1998 C&EN
I
Chemical engineer XIna Quan's department Is working to develop plastic optical fibers.
The use of such photopolymer mate dream, she tells C&EN, is to go one day to rials offers several important advantages Home Depot andfindplastic optical fiber over lithium niobate: They are much for sale. Before that can happen, though, more sensitive to light than the inorgan researchers will have to find polymer ic material and can be designed to give blends that offer the right properties and larger refractive index contrasts, allow are compatible with a continuous manu ing faster writing and reading of data. In facturing process, rather than the fiberaddition, photopolymers are inexpensive drawing process currently used to make and can be shaped and processed more optical fiber. easily than inorganic crystals. And writ Not only would plasticfiberbe less ex ing a hologram in lithium niobate tends pensive to install than silica, it would also to erase any previously recorded holo be more flexible, Quan notes. And one gram in the multiplexing process, mak could make thicker fibers, which would ing the crystal unsuitable for permanent make their interconnection easier and less data storage, Harris points out. expensive. Because of some of these same advan It will probably be a while before plas tages, polymers also are being looked at as tic opticalfiberis manufactured in quanti possible replacements for fused silica ties large enough to make producers of sil (S1O2) glass used in opticalfibers.The in icafibertake notice. After all, it's estimat formation-carrying capacity offiberoptics ed that silica optical fiber is produced is so great, according to Glass of the Pho worldwide at the rate of about 2,000 miles tonics Research Lab, that all of today's In offibereach hour, according to David W. ternet traffic could be accommodated on a Johnson Jr., who heads the Ceramics & single optical fiber. The technology has Metallurgy Research Department. enormous potential but its use has been The technology for manufacturing opti limited because of the high cost of some cal fiber is well established. Basically, it of the optical components. The biggest involves preparing a glass "preform"— challenge, Glass noted at a recent confer a cylinder of high-purity glass (the core) ence, is cutting the cost so that homes and surrounded by a "cladding" and an "overoffices can be "wired" with optical fiber cladding" layer of glass, each of which can the way they are wired today with con have a slightly different chemical composi ventional telephone or cable TV lines. tion. This preform is then heated and Chemical engineer Quan and her co drawn out to produce the opticalfiber.In workers are studying amorphous fluorinat- thisfiber,the cladding serves to keep the ed polymers as possible replacements for light confined inside the glass core, which silica in the short lengths of optical fibers is where it actually travels. The overclad(up to 1 km) that would go into homes. ding provides about 90% of the fiber's They are focusing onfluorinatedpolymers mass, making thefibereasier to handle. because replacing carbon-hydrogen bonds The preform is prepared using a va with carbon-fluorine bonds eliminates por deposition technique. As an alterna much of the absorption losses. Quan's tive, several groups have explored the
potential of the well-known sol-gel pro cess to produce the preform at signifi cantly lower cost. And earlier this year, Johnson, materials scientist John B. MacChesney, and their Bell Labs coworkers described a sol-gel process that may be come a less expensive commercial pro cess for producing the overcladding, which is not the most important part op tically but is quite expensive to make [/. Non-Cryst. Solids, 226, 232 (1998)]. In the Bell Labs process, colloidal sili ca particles, so-called fumed silica, are dispersed in water at high pH by rapid mixing. Particles of zirconia and other contaminants, which would lead to flaws and fiber breakage» are removed by centrifugation. An ester, typically methyl for mate, is added to the milky sol, which is then pumped into the cavity of a tubular mold. The gradual hydrolysis of the ester causes chemical changes that destabilize the sol and lead to the formation of a gel within a few minutes. The gel body is removed and carefully dried for about six days. Then it is subject ed to a series of heat treatments to further purify the glass. Finally, the material is sin tered, which yields a hard, transparent tube about 115 cm long. "No one has been as successful as us in making such large bodies of optical-fiberquality silica glass" by the sol-gel route, Johnson points out. The inner part of the preform— prepared by vapor deposition—is insert ed into this sol-gel-derived overcladding tube, which is then heated on a glassworking lathe to shrink it down so it bonds to the inner rod, producing a thicker solid rod. The opticalfiberspro duced by drawing out this hybrid pre form are "virtually indistinguishable" from conventionally produced fibers in their strength and optical properties, Johnson says. He believes this process could replace the process now used to make the outer part of the preform. Another inorganic material that has be come the focus of investigation in John son's department is zirconium tungstate (ZrVi^Og), an astonishing ceramic that shrinks—rather than expands—when heated. Scientists have known about this material's "negative thermal expansion" (ΝΤΕ) for 30 years. But their interest has perked up in the past few years because of new revelations about ZrW208's behav ior. Specifically, the magnitude of the con traction was found to be relatively large and to occur over a very wide tempera ture range, from 0.3 Κ up to the material's decomposition temperature of about 1,050 Κ (777 °Q. Furthermore, the shrinkNOVEMBER 30, 1998 C&EN 27
science/technology age occurs uniformly along all three di is a measure of how much charge sepa constant lower than Si02's but higher than air's value of 1. Such materials not mensions at lower temperatures, so the ration a material can accommodate. Microelectronics researchers have only are necessary for high-speed chips crystals remain cubic (C&EN, April 8, 1996, page 9). Previously discovered ΝΤΕ been driven to create smaller components but also reduce cross talk (electrical in materials were less impressive because with enhanced performance. One prob terference) between closely spaced met they either shrink less, shrink over only a lem area involves capacitive components, al lines, making smaller chips more effi narrow temperature range, or shrink in which form the basis of many memory de cient. At Bell Labs, scientists are looking vices. The dielectric insulators used in for suitable low-dielectric-constant mate only one or two directions. One immediately apparent applica these components are amorphous silicon rials among porous forms of silica, fluorition for zirconium tungstate and a related dioxide (S1O2) and silicon nitride (S13N4), nated polymers, and organic-inorganic material, hafnium tungstate, is that they which have dielectric constants of 4 and composites. Also of interest to Bell Labs research might be used in composites to compen 7, respectively. Such low dielectric con sate for the expansion of other materials, stants limit how small a practical device ers are a wide variety of other materials, Johnson points out. The result might be can be made—a limit that is now being including lead-free solders, piezoelectric a zero-expansion material or a material approached. Thus, researchers are moti materials, lithographic chemicals, mag that undergoes a precisely engineered vated to find materials with significantly netic materials, and superconductors. change in its dimensions during heating higher dielectric constants so that the Some of their investigations into these or cooling. Such composites might be march toward miniaturization can proceed. and other materials will undoubtedly used in components for next-generation The best alternative materials currently lead to technological advances and com fiber-optic systems for optical network being considered for integrated-circuit ca mercial products. Other investigations ing, for example. pacitors are barium-strontium titanates, may lead not to an immediate practical Lucent is currently evaluating a zirconi (Ba,Sr)Ti03, which boast dielectric con- benefit but to a Nobel Prize, as happened recendy when Bell Labs physicist um tungstate composite developed ο Horst L. Stormer and two former by Bell Labs ceramic engineer Deb§ Bell Labs physicists—Robert B. ra A. Fleming and chemist Glen R. % Laughlin and Daniel C. Tsui—were Kowach as a potential packaging I awarded the 1998 Nobel Prize in material for a refractive index grat S Physics (C&EN, Oct. 19, page 14). ing used in glass opticalfiber.The I That prize brought to 11 the total material's unique shrinkage proper °- number of Nobel Laureates (all in ties would compensate exactly for physics) who did their prize-winning variations in the refractive index work at Bell Labs. and dimensions of the glassfiberas the temperature changes, Bell Labs A further indication of Bell Labs' scientists explain. Otherwise, the stature in the world of physical sci wrong wavelength would be re ences research is the degree to flected by the grating. which the lab's published papers are cited in the scientific literature. The origin of the negative ther Lynn Schneemeyer (left), Robert Fleming (right rear), According to the November/ mal expansion phenomenon at and Bruce van Dover use a combinatorial approach to December 1997 issue of Science the atomic level is of keen interest develop thln-fllm dielectrics. Watch, a publication of Philadel to scientists. They believe that the material's Zr-O-W linkages flex like el stants in the range of 200 to 800. But mak phia's Institute for Scientific Information, bow joints when heated. This pulls the ingfilmsof these materials requires a high Bell Labs received by far the most citations metal atoms closer together, causing the deposition temperature (typically over (18,840) of any research organization in the 1990-97 period. structure to contract. 650 °Q. Researchers at Bell Labs and at Johns As Bell Labs gets ready to celebrate its Using a combinatorial chemistry ap Hopkins University have studied the vibra proach, Bell Labs chemist Lynn F. 75th anniversary in 2000, an obvious tional modes of this process using neutron Schneemeyer and physicists R. Bruce van question to ask is: Can Bell Labs under scattering. Earlier this month, they pub Dover and Robert M. Fleming recently Lucent continue this splendid record? lished theirfindings,which were obtained found a family of amorphous zirconiumWhen Francis J. Di Salvo was asked this in part using large single crystals of tin-titanium oxides that can be deposited question, he fired back: "I hope so," add ZrW208 grown by Kowach [Nature, 396, as thin-film dielectrics by sputtering at ing with a laugh, "I don't have a crystal 147 (1998)]. The groups results indicate 200 °C [Nature, 392, 162 (1998); C&EN, ball." Di Salvo, a materials chemistry pro that the atomic vibrational modes occur at March 16, page 9; IEEE Electron Device fessor at Cornell University, worked at Bell unusually low frequencies. This provides Lett., 19, 329 (1998)]. This would make Labsfrom1971 to 1986 and still maintains important information that eventually the oxide's deposition more compatible close ties to the facility. "Bell Labs went might be exploited to make other ΝΤΕ with current fabrication technology for through some hard times" during the materials more easily and economically, silicon integrated circuits. The dielectric AT&T breakup, he says, "but I think says team member Arthur P. Ramirez, a constants of the Zr-Sn-Ti-O materials they've recovered largely." Although it has physicist at Bell Labs. are 50 to 70—still a major improvement changed, it's still an "exceptional" place, he opines, and "they still take only the top The element zirconium also pops up over current dielectric materials. For other device applications, scien people." in other Bell Labs research that is aimed at finding new materials with a high di tists are going in the opposite direction: That would seem to bode well for its electric constant. The dielectric constant searching for materials with a dielectric future.^ 28 NOVEMBER 30, 1998 C&EN
