The United States has used fewer formal collaborative activities in SCE, and these have been more limited in their styles, but informal collaborations are much more common than in Japan. Also, in the early years of the four small SCE companies that have blazed the trail in the United States (Conductus, ISC, SCT, and STI), their activities were in some ways similar to centralized collaborative activities. Research staff and management were drawn from companies, universities, and federal laboratories. Funding came from venture capital, corporate investment, public investment through the sale of stock, and government agencies. (Government funding has been up to about 50% of the total in some cases). The broad objective was initially to develop a viable HTS technology, and interaction with the research community was strong. AMTEL in particular resembles both STI and Conductus in their first five years. With their stronger (exclusive, in some cases) focus on wireless products of the past few years, the four companies can no longer be regarded as so strongly linked to the research community, but over the first 5 years of HTS work, they had a unique role in the U.S. research program in HTS.
In the view of many in the field, the first and second High Temperature Superconducting Space Experiments, early joint ventures, were almost solely responsible for giving a focus to HTS radio frequency (rf) and microwave activities in the United States. In the first of the two planned satellite experiments, HTSSE I, relatively simple HTS devices, the majority being passive components made from a single HTS film, were to be placed in orbit to investigate the durability of HTS in space. Although the satellite was lost during launch, the benefits of the project are clear. A large number of groups had to make decent films, integrate them into a space-qualified package, and, most importantly, deliver them on time (Mitschang 1995). Much of the capability that allowed them to succeed was developed with funding from the Defense Advanced Research Projects Agency (DARPA).
The second experiment, HTSSE II, which is still waiting for its scheduled launch, includes a smaller number of components, but they are more complex, being roughly small subsystems (Kawecki et al. 1996). Again, the majority are rf and microwave applications, e.g., channelizers, receivers, and a SQUID-based digital multiplexer. Groups that have developed these packages had to produce more than one device and provide more sophisticated packages.
There is no doubt that the leadership position in rf and microwave applications of superconductivity enjoyed by the United States at present is very largely due to the learning experience of HTSSE I and II. Although much of the component development was funded by Department of Defense sources, the HTSSE "vision" galvanized a broad sector of the HTS electronics R&D community to deliver advanced technology. It is most unfortunate that the United States did not continue with HTSSE III, which was tentatively planned as a communications satellite using hybrid SCE and semiconductor technology. The high level of funding needed for such a satellite (or other system) will be obtained only with substantially stronger customer pull than currently exists.
The Consortium for Superconducting Electronics, involving Bell Labs, IBM, MIT, and MIT Lincoln Laboratory (LL), had characteristics of both a joint venture and a distributed activity. There were several small independent projects in materials and devices at MIT, Boston University, Cornell University, and SUNY Stony Brook. As application concepts matured, the circuit programs drew most of the resources. IBM pursued SQUID development independently. Work on LTS digital circuits was a collaboration between IBM and Lincoln Laboratory. The largest program, that on rf and microwave applications, began as a joint venture between Bell Labs, MIT, and MIT LL, and soon expanded with the addition of Conductus and CTI. Approximately 50% of the cost of the CSE was contributed by DARPA, with cost sharing from the member companies comprising the balance. After the end of the 7-year program in 1996, Bell Labs, Conductus and MIT LL continued their joint work in a follow-on program to develop wireless systems, with partial funding from DARPA.
The University Research Initiative in the SCE field is a distributed activity that takes place at the universities of California at Berkeley, New York at Stony Brook and Rochester, and at Stanford University. The URI has had as its focus the development of Josephson LTS digital technology with an emphasis on rapid flux single quantum (RSFQ) circuits. The programs in each university are independent of each other, and Hypres was used heavily as a foundry for circuit fabrication. The advances in RSFQ circuits have been impressive, and it appears likely that the program, at least at Stony Brook, will continue with funding under the Petaflop Computer project. This represents a milestone, being the first time that superconducting digital electronics has been recognized and funded as a competitive electronics technology, rather than as a superconducting device and circuit program.
For the past 5 years, Conductus, Stanford University, the National Institute of Standards and Technology (NIST), TRW and the University of California at Berkeley have been engaged in an Advanced Technology Program joint venture to develop hybrid system technology. (Hewlett Packard Research Laboratory was also a member of the team for the first year.) The majority of the LTS chips are made at Hypres. The work is carried out in the separate laboratories but has a common objective. Initially, this was to develop a digital signal processing system, but after about 2 years this was changed to demonstrating an optical switch system. The system, expected to be complete in the summer of 1997, includes optical fiber inputs and outputs from 300 K; a semiconductor receiver, 4 x 4 RSFQ switch, clock recovery circuit, and Josephson junction amplifier, all at 4 K; a transmit laser at 300 K, and a semiconductor amplifier at 70 K. The planned system operating speed was 10 Gbit/s, and achieved system performance is 8 Gbit/s. Cooling is by a Boreas refrigerator, with conduction cooling of the SCE chips and semiconductor components. This is the most complex integrated hybrid SCE digital system built to date, in terms of the variety of technologies incorporated.
The 32 x 32 Switch Project, sponsored by the U.S. government, is a joint venture targeted at a high performance computing environment, that is, multiple processors in a shared memory configuration. The government sponsor conceived the architecture and designed the circuits for a high-throughput crossbar switch with input/output ports at room temperature. The government has coordinated the development and demonstration of Nb-based technology, with MCC, Hypres, Tektronix, TRW, and MIT-LL participating at various stages.
Performance goals of the 32 x 32 switch project include the following:
Tests of a fully populated 20-chip 128 x 128 crossbar using a 5-metal-level multichip module (MCM) were conducted in 1997. The channel data rate did not reach its goal, but full-path, bidirectional data flows were demonstrated. The system used eight 74-trace ribbon cables to connect the room temperature electronics to the MCM in its cryostat.
The present phase of the effort will produce a fully functional 128 x 128, 2.5 Gbit/sec/port crossbar. It will use 24 chips flip-chip-mounted on a 13-metal-level ceramic-polyimide MCM that will be cooled by a Boreas refrigerator. The 8 connecting cables to room temperature will contain 1,000 signal/power lines.
The Vapor Phase Manufacturing activity to develop thin film deposition methods for the manufacture of large-area HTS films falls between a joint venture and a distributed activity. Conductus and STI are independently investigating film manufacture by co-evaporation and metallorganic chemical vapor deposition (MOCVD), respectively, but they are working closely with a number of other companies, universities, and national laboratories on issues such as deposition rate monitoring, precursor development, cost modeling, etc. The two HTS wireless companies also compare performance data of their films by exchange of samples and through measurements made at NIST.
There are a number of other collaborations in the United States that cannot be called projects, in that the funding of the work of the participants is provided separately and sometimes from different sources. There are many such informal interactions in the United States, often between only two or a few scientists, generally initiated without any input from or to the funding agency. This network of collaborations represents one of the striking differences between the United States and Japan. In Japan, collaborations are generally formal. The WTEC panel noted many opportunities for informal collaborations during our week in Japan, but these do not occur.
A somewhat more formal type of collaboration in the United States is nicknamed the "Big Three." With funding that is independently negotiated, Conductus, Northrop Grumman, and TRW have under strong government urging shared the results of their efforts to improve the uniformity of HTS SNS junctions. Initially, all three companies investigated the cobalt-doped YBCO barriers pioneered by K. Char and his colleagues at Conductus. The spreads of critical currents were reduced to one sigma of about 12 to 15%. Recently B. Moeckly and K. Char have invented an alternative type of barrier for the junctions that, in early results, indicates a one sigma spread below 10%. It is likely this new junction technology will also be investigated by the other two companies and by other labs.