The 2,162nd Meeting of the Society

May 2, 2003

A New Spin on Electronics

Stuart Wolf

Defense Advanced Research Projects Agency

About the Lecture

This talk will focus on describing projects that are exploiting the spin degree of freedom of electrons for future generations of electronics. Spintronics, as this field has begun to be called, (this was the name of the original DARPA project) involves developing a host of new materials, devices and techniques that rely on the spin instead of the charge to store, manipulate and communicate information. There are several projects that have evolved from the original spintronics project to develop sensors and memory. These new projects are now exploring enhanced logic devices such as spin transistors and spin-FETs, spin optical devices like spin-LEDs and high speed optical switches and finally spin quantum devices for quantum computation and communication. I will describe where we are and where we hope to be at the end of the first decade of the 21st century.

About the Speaker

Stu Wolf is currently both a Research Professor at the University of Virginia, and a consultant in the Defense Sciences Office at the Defense Advanced Research Projects Agency (DARPA). At DARPA he has conceived and initiated several projects on functional materials that have the goal of pushing the frontiers of materials science for electronics. These programs include: 1) "Spintronics" whose goal is the development of non-volatile, high density, high speed magnetic memory, and now has expanded to include Spins IN Semiconductors (SPINS) which hopes to develop a new paradigm in semiconductor electronics based on the spin degree of freedom of the electron in addition to or in place of the charge, 2) Quantum Information Science and Technology (QuIST) which aims to develop communication and computing systems and architectures based on the principles of quantum coherence and entanglement, and 3) Molecular Observation, Spectroscopy And Imaging using Cantilevers (MOSAIC) which has the goal of 3D imaging of molecules and nano-structures with atomic scale resolution. Dr. Wolf coined the term Spintronics in 1996. He has an AB from Columbia College (1964) and an MS (1966) and PhD (1969) from Rutgers University. He was a Research Associate at Case Western Reserve University (1970-73) and a Visiting Scholar at UCLA (1981-82). He is a Fellow of the APS (1984), and was a Divisional Councilor for the Condensed Matter Division (1990-91) and is currently the Divisional Councilor for the Forum on Industrial and Applied Physics. He has authored or co-authored two books, over 300 articles, and has edited numerous conference proceedings.


The 2162nd of the Philosophical Society of Washington was called to order at 8:15 PM in the Powell auditorium of the Cosmos Club. President Haapala was in the chair. The president introduced Stewart Wolf, the speaker for the evening. Mr. Wolf is a research professor at the University of Virginia. He is also a consultant in the Defense Sciences Office of the Defense Advanced Research Projects Office (DARPA) where he has conceived and initiated projects on functional materials with the goal of pushing the frontiers of materials science in electronics. Among these are the development of nonvolatile memory based on principles of quantum coherence and entanglement. Another of his program initiatives is the development of molecular observation, spectroscopy and imaging using cantilevers, the goal of which is three-dimensional imaging of molecules and nano-structures with atomic scale resolution. Mr. Wolf coined the term Spintronics in 1996 and has authored numerous publications on the subject. In the 1990's the objective was to create a new class of electronics based on the spin degree of freedom to produce a non-volatile, radiation hard memory chip with speeds of less than 3 nano-seconds. The chip would have a density of dynamic random access memory of up to 4 gigabytes, low power draw, low cost of manufacture and be infinitely cycleable plus having a high sensitivity with a small overall size. Mr. Wolf illustrated his explanations of how this works with graphics showing materials classified as metals, semiconductors or insulators. The valence and conduction electron positions were full in all classes of materials but the electron band was partially filled in metals but empty in the semiconductors and insulators. Take Copper for example, this conductor has equal numbers of up and down spins. Ferromagnetic metals however have an unequal number of spins. In the late 1980's it was discovered that a thin sandwich of these metals produced a spin bottleneck magneto-resistance effect. Because electrical resistance through the sandwich depends on the direction of the magnetic layers, one could achieve more than 100 % change in resistance through a combination of multiple layers. IBM introduced a new hard disk read head sensor based on this effect, which is now known as GMR technology. The Travelstar disk drive can store 4.1 gigabytes of information per square inch and the market for this technology is now around $100 billion per year. Another item of GMR technology is the digital isolator sensor. This new magneto-resistive sensor, 2 mm on a side in used in laptop PCs. It represents a factor of 5 improvement over the previously used HP optio-isolator. Another major development is in memory chips. The speaker illustrated the key attributes of memory with another graphic showing the relationship between high speed, non-volatility, high density, no refresh requirement, non-destructive reads and low voltage operation. Motorola has employed this technology to make memory cells using 10 Angstroms thick insulator layers on an 8 inch square wafer. The manufacturing process must maintain absolute control of the thickness of the insulating layer. The device can flip the memory state with current in two lines and read the memory state by measuring the resistance across the layers. Motorola currently employs these devices embedded on a processor chip to provide more than one megabyte of memory. Within the structure the transistors take up a very small part, the rest of the space is wires. Another technology program is coherent control by pulse shape tailoring. This work is being conducted by the National Institute of Standards and Technology in order to develop a competitive form of SRAM. Shaping the pulse into two peaks, which has the effect of taking the ringing out of the circuit, allows memory to be switched in less than 1 nanosecond. Currently both Motorola and IBM are investing more research money in these technologies than DARPA is investing. Richard Feynman issued a challenge in 1959 to (1) Take spin beyond ferromagnetic materials, (2) to make computers with wires not wider than 100 atoms and (3) To make a microscope to view individual atoms. The third challenge has been met. New directions in optical effects in semiconductors have been demonstrated, for example using circularly polarized light in GaMnAs were reported in the Scientific American issue on spintronics. One Illustration showed how a puddle of magnetism could be moved through a semiconductor electronically. This effect, although demonstrated at 4 K may soon be operating at temperatures higher than 180 Kelvin. Polarized light exerts a torque on the electrons inducing a spin coherent state in quantum dots of CdSe. This effect lasted tens of nanoseconds at room temperature. The current research looks at quantum spin electronics, spin field effect transistors, light emitting diodes, modulateable spin lasers and coherent spin electronics. The reported slowing of light is actually a transfer of coherence in GaAs. What needs to be done next includes development of flows of coherent information across the hetero-interface of dissimilar metals such as the semiconductor GaAs or ZnSe. By applying a voltage to the semiconductor with layers having different g-factors, the first electrons through, precess with the frequency and sign of the g-factor in the layer they have crossed into, whereas the electrons that are driven across the boundary precess with the frequency and sign of the g-factor of the semiconductor they originated in. Driving the spin through a junction from a metal to a semiconductor, you want to be able to flip the spin with a very small magnetic field. Creating a gradient of the Aluminum content of GaAlAs varies the g-factor. It is possible to make a parabolic quantum well using evaporated thin Ti/Au as the front gate. A few volts on the gate dramatically effect the spin. Phase shifts can be controlled with phase locked microwave inputs. G-tensor modulation resonance has been demonstrated using the aforementioned parabolic quantum wells, the effective magnetic field can be modulated with an appropriate electric field. Electrons flip spins when the system comes into spin resonance. Thus you could do ESR with an electric field. Quantum spintronics uses the superposition state, which has statistical probability of being in any state. This technology could produce computers, which could factor numbers faster then the best classical computers. Quantum computers could be based on quantum dots in GaAs. They would optically excite electrons perpendicular to a magnetic field. The current DARPA project is called QuIST for Quantum Information Science and Technology. QuIST, among other methods, uses the spin states in quantum dots to form a quantum network. This leads to a form of reversible computing. The speaker kindly agreed to take questions from the floor. Questions included: Q- You showed a 1 megabyte memory embedded in a microprocessor. Must it be embedded? A- Motorola is making perhaps 16 megabyte memory blocks now embedded on a microprocessor. They have a customer for this. They will be moving ahead with DRAM and Flash RAM soon. IBM is working to develop 256 megabyte memory chips. Q- What is an electron spin? I think of an electron as a form of a wave. A- Call it angular momentum. You get a magnetic moment from an electron. What is actually going on is unknown. Q- In quantum entanglement we don't know which state a function is in until we measure it. That gives a probability function. The state with the higher probability is what you are apt to observe. Is it a dead cat or a live cat? A- It is really a dead and a live cat at the same time but it could be spending more time alive than dead. Q- In the slide of polarized light on a junction. Enigmatic mode - the band structure must be preventing or excluding something? A- The first few electrons across diffuse. Later ones are really driven across and become ballistic. They accelerate through and are not scattered by the lattice. Only when they are scattered do the electrons realize they are in ZnSe. Q- is any of this related to electrical energy applications? A- Light is easy to propagate but can not be stored. Electrons on the other hand can be stored but do not propagate well. Q- How soon do you expect a commercial spin transfer device using spin coherence? A- Four to five years. This will be 2 to 3 generations of the technology down the road. It may be an extremely compact optical delay line. These are now done with 100-foot coils of optical fiber. An optical delay line, which stores the spin would be preferable. President Haapala thanked the speaker on behalf of the Society and presented him with a one year complementary membership. The president then made the usual parking and beverage control announcements and invited the members and guests to the 72nd annual Joseph Henry Lecture to be delivered on the 16th of May. The topic will be Astrobiology and Life in Space. President Haapala also urged guests to consider joining the Society and for Members to contribute for its continued ability to present programs. An informal communication as given by Carl Vershaluef who noted that he had met W. Paul Gorley, an international law scholar in environmental law and had been extrememly impressed with the man. Paul died five weeks ago and Mr. Vershaluef wished to speak to anyone who had know Mr. Gorley. Attendance: 50 Temperature: 24.4° C Weather: Warm and humid Respectfully submitted, David F. Bleil Recording Secretary