Lateral p-n Junctions: an Original Concept in Optoelectronic Devices




Introduction

Optoelectronic devices, such as light-emitting diodes, laser diodes, and photodetectors, are essential components of many appliances that we use in our daily lives. These devices contain one or more semiconductor materials that conduct electricity either by the movement of electrons (n-type material) or "holes" (lack of electrons that allow conduction of electricity in an otherwise bounded electronic structure) (p-type material). The conduction type of the material is determined by the incorporation of controlled amounts of impurities. Some impurities that are called amphoteric dopants can produce both n-type or p-type material under different incorporation conditions (i.e., different crystalline orientation of the surface during the growth of a semiconductor material). A piece of semiconductor material that has adjacent p-type and n-type regions is called a p-n junction, and it is the fundamental structure for the above-mentioned optoelectronic devices.

These p-n junctions are currently fabricated by two processes: 1. Piling p-type and n-type layers, and 2. Diffusion of p-type (n-type) impurities in an n-type (p-type) material until one region becomes p-type (n-type). We are developing a third way to make p-n junctions. This consists of growing a semiconductor material layer (gallium arsenide, GaAs) with an amphoteric dopant (silicon) on a patterned substrate that exposes more than one crystalline orientation on the surface. According to the surface orientation, some regions are n-type and other are p-type. A p-n junction is formed at the intersection of these surfaces.1

Lateral p-n junctions have particular properties that are promising for application Iin new devices. The junction area is controlled by the thickness of the layer with silicon impurities that can be as thin as one atomic single layer. Therefore, the junction area can be made orders of magnitude smaller than in conventional p-n junctions. The p-type and n-type regions are side by side on the sample surface; therefore, all the electrical contacts can be made on the same side of the sample using a coplanar geometry. Moreover, electrically insulating substrates can be used, simplifying electrical isolation of multiple devices made on the same substrate. Also, the junction can be clad between non-doped epitaxial layers that reduce leakage current and produce carrier confinement. Finally, carriers in a lateral p-n junction are transported in the plane of multilayer structures, where carriers have a high mobility due to quantum confinement produced by modulation doping in heterostructures. We proposed and demonstrated various devices which use these advantageous properties of lateral junctions.


Lateral Junction Devices

Conventional light-emitting diodes (LEDs) have a large portion of the light-emitting area covered by the top contact. This configuration obstructs light emission and reduces total efficiency. The situation becomes worse when very small devices are required, as is the case of LED arrays for printer applications. On the other hand, lateral-junction LEDs (LKJ-LEDs) have contacts placed on the sides of the light-emitting area, and light emission is not obstructed.

We fabricated LJ-LEDs with various active layer structures and compositions.2 A typical device structure is shown in Figure 1. The active layer includes one or more GaAs layers with silicon impurities and single-quantum or multiple-quantum well structures made of aluminum gallium arsenide (AlGaAs) or indium gallium aresenide (InGaAs). The energy bandgap of these semiconductor alloys and the quantum confinement effects allow control of the emitted light over a large wavelength range (between 650 nm and 1100 nm). In a typical device at room temperature, light emission intensity saturates and starts to decrease above 90 mA of current due to sample heating. These devices operate above room temperature without apparent degradation for current densities as high as 309 kA/cm2.

This p-n junction can be also used for making vertical-cavity as well as edge-emitting laser diodes. First, we made a lateral-junction vertical-cavity surface-emitting laser (LJ-VCSEL).3 the structure of the LJ-VCSEL is similar to the LJ-LED, with the addition of multilayer dielectric mirrors (distributed Bragg reflectors) on the top and bottom of the active layer, to make a Fabry-Perot cavity with its axis perpendicular to the substrate surface. The mirrors are composed of alternating layers of AlGaAs and AlAs deposited by molecular beam epitaxy. After the deposition, the AlAs layers are oxidized to increase the reflectance of the mirrors. Figure 2 shows a scanning-electron microscope picture of a completed device.

Laser emission at room temperature is observed for a current higher than 2.3 mA. We use pulsed current to avoid overheating the device. Figure 3 shows emission spectra below and above threshold current at room temperature. Below threshold, the spontaneous emission from the active layer that leaks through the Fabry-Perot cavity resonant mode is observed. Above threshold, single mode laser emission is observed at 942 nm with a full width at half maximum (FWHM) of 0.15 nm.

Electrons in lateral-junction quantum-well laser diodes are directly injected into the active region. In this case, the modulation bandwidth of the laser diodes can be greatly enhanced. Moreover, the applied bias required to inject a given current can be reduced, thereby improving the power efficiency of the device. The lateral junction area is orders of magnitude smaller as compared to vertical-junctions, with the consequent reduction of the device capacitance.

Another interesting application of the lateral junction is for photodiodes.4 The time required for photogenerated carriers to drift across the depletion layer and the capacitance of the device are the factors that limit the response speed of conventional photodiodes. Therefore, the depletion layer should be thin to obtain a short transit time and junction area should be small to have small capacitance. However, in conventional photodiodes, where the incident light and internal electric field are parallel, a very thin depletion layer will decrease the absorption of light, and thus diminish the quantum efficiency. Basides, a small junction area makes difficult light injection and device fabrication.

The situation is different when the device has a lateral p-n junction. The incident light and the internal electric field in this device are perpendicular to each other. Therefore, we can make the junction area much smaller than the incident-light spot size, with the consequent reduction of the device capacitance and we can also keep the depletion layer as thin as is compatible with the incident-light spot size.
We made a photodiode consisting of a GaAs layer with a thickness of 1 ƒÊm to absorb the incident light. This layer has a relatively low density of silicon impurities, to form a lateral p-n junction with a rather broad depletion region, where photogenerated carriers are quickly separated and wiped out by the junction's electric field. This layer has AlGaAs cladding layer on both sides to decrease the carrier recombination at the interfaces.

We measured the change of reflectivity of the sample due to photogenerated carriers using a pump-and-probe configuration to assess the highest limit of the response speed of the photodiode and found that the recovery time of the reflectivity is about 10 ps. The design of this device is not optimized for high-speed operation, therefore this figure will be even better if impurities concentration, layer structure and contact spacing are optimized. This concept-demonstration device shows very encouraging characteristics. The GaAs absorption region cannot detect light with wavelengths longer than 890 nm, while most of the devices used in optical-fiber communications operate at 1.3 ƒÊm or 1.55 ƒÊm of wavelength. In principle, this lateral-junction photodiode can be made on indium phosphide substrates using germanium or silicon as impurities, and it would be able to detect light with the wavelengths used in optical-fiber communications.


More Advanced Concepts

The radically distinct geometry of the lateral p-n junction opens the door to design devices that were unthinkable or very difficult to make using more conventional structures. An example is a device proposed by Fiore et al.5 that generates far-infrared laser light by mixing the light from two near-infrared lasers in a waveguide including GaAs and aluminum oxide layers for phase matching. By integrating laser diodes inside the waveguide, a monolithic device can be made. However, the electrically Insulating aluminum oxide layers prevent current injection in a conventional structure. Here, the lateral p-n junction is an elegant solution to this problem. A blue light-emitting device based on second-harmonic generation can be fashioned in a similar way, as long as blue-light absorption by the GaAs or AlGaAs layer can be kept to a reasonable level.

Another example is a device with short-pitch arrays of LJ-LEDs or LJ-VCSELs for application in printers. Under the current technology, in a 1200 dpi array, a good portion of the light-emitting area in conventional devices is covered by the top contact, decreasing appreciably their luminosity. On the other hand, the lateral-junction configuration would allow 2400-dpi or even 4800-dpi arrays with negligible shading by the contacts. Looking at the limitations of current optoelectronic devices surely will suggest many other applications for the unconventional lateral-junction diode structure.


Reference