EMISSIVE DISPLAYS

Scientists at IBM Thomas J. Watson Research Center (Yorktown Heights, NY) have demonstrated three-color emission from gallium nitride-based light-emitting diodes (LEDs) deposited on a silicon substrate. Each device was covered by a transparent layer containing organic dyes that allowed color conversion.

Aug 1st, 1998

EMISSIVE DISPLAYS

Full-color LED arrays on silicon show promise

Scientists at IBM Thomas J. Watson Research Center (Yorktown Heights, NY) have demonstrated three-color emission from gallium nitride-based light-emitting diodes (LEDs) deposited on a silicon substrate. Each device was covered by a transparent layer containing organic dyes that allowed color conversion.

The proof-of-principle experiment has encouraged the researchers to tackle some of the problems inherent to a device containing materials that are normally incompatible. These include such issues as lattice mismatch and thermal strain. The work is at an early stage, but success could result in development of extremely compact, bright, and efficient miniature displays.

Many researchers have been trying to develo¥LEDs that are compatible with silicon substrates. Most researchers have concentrated on using III-V materials such as gallium arsenide (GaAs) but have found device performance to be limited by the lattice mismatch between the two materials. At IBM, Supratik Guha and Nestor Bojarczuk looked to gallium nitride (GaN) based LEDs instead. Although these also suffered from a mismatch between the GaN and crystalline silicon structures, the consequent defects did not seem to have as great an effect on device performance--it was this robustness that the IBM team saw as crucial.

To deposit the LEDs in a way that was compatible with silicon processing, Guha and Bojarczuk decided to use molecular-beam epitaxy (MBE) instead of metal-organic chemical-vapor deposition. The latter is demonstrably more precise and produced better-quality devices, but it requires higher processing temperatures than MBE, which is a low-temperature fabrication technique. Although the resulting light emitters were of relatively low quality and contained many defects, their blue/ultraviolet (UV) output was visible under normal room-lighting conditions.

To create a multiple-colored output with as high an efficiency as possible, the IBM researchers chose a two-dye color-conversion process. In each case, a host captures the emitted light and the excitation is then passed on to the dopant dye, which emits a photon of the desired output color. This combination is necessary to combat the Forster transfer effect--a phenomenon that occurs where dye concentration is very high. In such circumstances, absorbed energy can be passed on to adjacent molecules until it has been quenched, instead of being reradiated. In the host-dopant case, the high density of the host can act to absorb as much incident light as possible, while the lower concentration of dopant dye makes sure that the energy "escapes" as light.

Color-conversion materials were applied to three LEDs on the substrate using a micropipette. The resulting three-color output included some shifting from the intended dye color--the red dye, for instance, produced an orange glow (see photo). Some of this was expected because the organic laser dyes used are known to exhibit different properties in liquid form from those they have when held in a solid matrix. The output spectrum also contained some blue/UV light that the converter had failed to absorb.

Ongoing research

The researchers will now attempt to optimize both the design and fabrication of these devices. According to Guha, making the MBE-grown devices efficient is an important challenge. Another is to minimize the defects that result from thermal strain caused by layers of varying materials expanding and contracting differently with temperature changes. A third area of research will be finding better combinations of organic dyes.

Guha and Bojarczuk also will be investigating how to minimize total internal reflection within the LED/ color-converter structure, which has been identified as a major source of light loss. One possibility might be to pattern the to¥of the color converter with a microlens. The team has done some initial testing, using a commercial blue-output LED and fluorescent dyes in a polymer matrix. In one experiment, index-matching a plano-convex lens to the to¥of a glass slide, with the LED index-matched underneath, increased the output light by 60%.

Sunny Bains

More in Research