OPTOELECTRONIC COMPONENTS: Researchers integrate multistable devices

The development of optical bistable devices, such as self-electro-optic-effect devices (SEEDs), led to a flurry of activity in optical logic during the early 1980s. This work continues, but its practicality has been questioned: optoelectronic devices have traditionally offered very little computing functionality to warrant their high power consumption. To combat this problem, researchers at Kyoto University (Japan) have been developing integrated components with more complex operational capabilities. Recently, for instance, they demonstrated tri-, tetra- and pentastable devices that can be set and reset optically. Though they have not yet used these components in a system, the researchers claim that both the devices and the philosophy that went into making them could have a significant impact on optical processing.

The multistable devices work through the coupling of smaller components that are integrated together. For a tristable device, the Kyoto team used four heterojunction phototransistors (HPT-A, -B, -C, and -D) and a laser diode (LD). In this system, the placement of each component relative to the others is as critical as the wiring for more conventional circuits.¹ Light falling on HPT-A is used to set the device. The photocurrent that flows when HPT-A is illuminated also flows through the laser diode. Because the laser is located directly underneath HPT-A, spontaneous emission caused by the photocurrent becomes optical feedback: the light goes into HPT-A, which creates more photocurrent, which causes more spontaneous emission.

If the light falling on HPT-A is strong enough (20 µW), then it turns the device into the first on-state. The optical feedback is now strong enough for self-oscillation to occur even when the light source is taken away, so the system is latched. Because the device is symmetrical, the spontaneous emission reaching HPT-A also reaches and causes optical feedback in HPT-B. If the incoming set light reaches 40 µW, then HPT-B is also turned on and self-oscillates with the laser diode even though it has never been illuminated from outside the device.

To reset the device, HPT-C is illuminated. The HPT-C and HPT-D sites are located slightly away from the LD stripe, so almost no coupling occurs between them. Current flowing through HPT-C simply causes a voltage drop across the laser. If this drop is big enough, it will prevent self-oscillation at HPT-B and eventually at HPT-A.

Advanced structures

For tetra- and pentastable devices, the Kyoto researchers use a six-HPT, two-LD system with more complex light/electronic couplings.² An AB pair that feeds back to one laser diode is coupled with an EF pair that feeds back to another. C and D HPTs are use for resetting.

These optical multistable devices are fabricated using a conventional InGaAsP/InP semiconductor structure. According to researcher Susumu Noda, even relatively simple components can be used to create useful optoelectronic components. By integrating them carefully, he says, you get "not only the combined characteristics of the constituent devices, but also new functions due to the mutual interaction between them." This leads to the more general philosophy of Noda`s group, led by Akio Sasaki—the higher the degree of integration, the higher the functionality of the device.

Using this approach, the group has developed many other optoelectronic components, including switches, amplifiers, and multistable flip-flops. Work continues to reduce the optical power needed to activate the devices and to incorporate surface-emitting, rather than edge-emitting, laser diodes into the design.

REFERENCES

1. S. Noda et al., IEEE J. Quantum Electron. 31(8), 1465 (Aug. 1995).

2. V Ahmadi et al., Solid-State Electronics 38(3), 551 (1995).