Site: NEC Kansai Electronics Research Laboratory NEC Kansai, Ltd.
9-1, Seiran 2-Chome, Otsu
Shiga 520, Japan

Date Visit: December 15, 1994

Report Author: L. Coldren

ATTENDEES

JTEC:

L. Coldren
G. Day
M. DeHaemer
P. Shumate
M. Warren

HOSTS:

Dr. Hiroyuki Abe
General Manager, Kansai Electronics Research Lab, NEC Corporation
Noriyoshi Kitagawa
Senior Vice President, NEC Kansai, Ltd.
Sho Iwayama
Manager, Engineering, Compound Semiconductor Device Dept., 2nd Semi. Div., NEC Kansai, Ltd.
Hiroaki Ishiuchi
Engineering Manager, Compound Semi. Dev't. Dept., 2nd Semi. Div., NEC Kansai, Ltd.
T. Uji
Senior Manager, Photonic Device Research, Kansai Electronics Research Lab, NEC Corp.

BACKGROUND

The previous NEC-Tsukuba report outlines the operations of NEC and some of the relationships to the Kansai Electronics Research Lab (part of the Research and Development Group of NEC, like the OE Research Labs in Tsukuba) and NEC Kansai, Ltd., the wafer fabrication subsidiary. The Kansai Electronics Research Lab was formed in 1991 with 50 researchers from the OE Research Labs under Dr. Hiroyuki Abe, 50 from the ULSI Development Lab, and 40 other research engineers. The move to Kansai allowed them to carry out research using manufacturing-line equipment, thus effectively integrating previous R&D with the manufacturing of compound semiconductor devices.

DISCUSSION

First to speak was Noriyoshi Kitagawa of NEC Kansai, Ltd. NEC Kansai has 4,000 employees, and 1993 sales reached $1.2 billion. All products that the lab manufactures are sold to NEC; some go to other NEC subsidiaries for packaging, and so forth. Compound semiconductor devices moved to Kansai very recently and account for 7% of the factory's output. Besides compound semiconductor devices (microwave and optical devices), the rest of the production includes CMOS, DRAM, zener diodes, SBT (power), tantalum capacitors, EL displays, cordless telephones, and tuners. In the semiconductor area, the factory generally generates wafers or chips. These are then processed into final products elsewhere. Most of the compound semiconductor products are microwave devices: low-noise FETs for low-noise preamps for satellites (production of 2 million/month); power FETs for portable telephones (production of 1.2 million/month); MMICs (production of 50 thousand/month); and Schottky or Varactor diodes for mixers, tuners, and oscillators.

Optical devices include (1) 1.3 microns and 1.5 microns DFB and FP lasers (production of 30 thousand/month) for applications in fiber communications and analog and optical measurement; (2) visible (red) LDs (production of 30 thousand/month) for applications in bar-code readers, pointers, distance measuring, and optical disks; (3) long-wavelength LEDs for applications in fiber communications, data links, and LANs; (4) short-wavelength LEDs for applications in photo couplers, interrupters, and remote controls; and (5) APD and pin detectors for communications applications.

Next Dr. Abe reviewed the R&D activities of NEC, pointing out the parallel charter of Kobayashi's lab at Tsukuba and his lab. The Kansai Electronics Research Lab focuses on ultrahigh-speed compound semiconductor devices and semiconductor photonic devices. The primary application at present for the photonic devices explored in this lab is for access network (FTTH). Dr. Abe also pointed out the importance of the co-location of the ULSI Device Development Labs to produce engineering prototypes suitable for manufacture. Generally, there is an overlap of activities between the Kansai Electronics Research Lab and the ULSI Labs in developing new devices. Thus, a smooth technology transfer from research through development is assured. Also, the same holds for the transition from the ULSI labs to NEC Kansai, Ltd., the manufacturing subsidiary. To have close coupling between R&D and production, the labs have monthly R,D&P meetings that also include engineers from Tsukuba and the Kanagawa Central Research Labs. An example of the overlap of activities between the OERL (in Tsukuba and Kanagawa) was the simultaneous work on long-wavelength LDs. On the other hand, Tsukuba is researching long-distance LDs, while Kansai is working on subscriber-loop lasers. An outline of these interactions is shown in Figure NEC-Kansai.1.

Dr. Abe also reviewed the status and plans for dedicated communication networks between the many NEC sites for the purpose of supporting good communications between them. The labs plan to use 5 Mbit/s multimedia channels between plants (labs) multiplexed onto a 250 Mbit/s backbone. Figure NEC-Kansai.2 shows this R&D group network.


Fig. NEC-Kansai.1. R&D and production of compound semiconductor devices.


Fig. NEC-Kansai.2. NEC R&D group network.

T. Uji, a member of Dr. Abe's Lab, introduced some of the recent work on devices for optical access networks. Ease of use, low cost, and transmission capability were key elements that he stressed. As indicated in Figure NEC-Kansai.3, such devices must be cooler-less, control-free, be easy to align, and deliver the needed speed, power, and reliability. He presented record results for 1.3 microns edge-emitting laser diodes. His results included dynamic tests of control-free diodes that operated at zero bias over -40 to 85 deg. C at 1 Gbit/s with 30 mA. Passive alignment using Si V-groove and premarked diode locators gave coupling losses < 5 dB from reproducible submicron alignments. The researchers also seek MTTF of 100 kh at 85 deg. C. Figure NEC-Kansai.4 shows the basic structure of this device.


Fig. NEC-Kansai.3. Technology requirements for fiber-optic subscriber networks.


Fig. NEC-Kansai.4. Control-free LDs for optical subscriber loops.

Analog lasers using NEC-Kansai's partially corrugated waveguide technology showed high yield and low distortion. These are in production. The laboratory uses a "partial corrugation" DBR-like design to reduce the effect of unpredictable facet phase. This appears to give better uniformity of the optical field. In the array area, Uji showed results from a data link that used a 12-channel 1.3 microns LED source array and a matching InGaAs pin diode receiver array, operating at 155 Mbit/s. He also showed his scientists' results using TBA and TBP, nontoxic MOVPE sources, and a sub-mA threshold 1.3 microns LD that used an all-MOVPE process. His work measured a threshold current of 0.4 mA with an output of 0.8 mW at 10 mA. Finally, he outlined a full wafer E-beam-RIE process that led to very high diode wavelength predictability. Multiwavelength arrays are desired for dense WDM applications.

During the lab tour the JTEC panel learned that NEC Kansai's chips are sent to QDK (a subcontractor of NEC in Kyushu Island) for assembly and packaging. QDK is well known for its large and automated Si-IC line. The short-wavelength LED line at QDK is completely automated also.

SUMMARY

NEC-Otsu is a no-nonsense, generate-product operation. The most impressive aspect is that world-class researchers are working to produce practical, low-cost products. This operation is an excellent model for U.S. companies (or any companies) that really want to have an impact in this field. The JTEC panel recommends that people seeking success in the optoelectronics field visit NEC-Otsu. Although it was clear that technology was being transferred from other NEC locations to the factory at NEC Kansai, Ltd., it appeared that the transfer might be more efficient between Dr. Abe's lab and the factory co-located in Otsu. It seems likely that good, low-cost sources for the access (FTTH) network will soon be forthcoming.


Published: February 1996; WTEC Hyper-Librarian