AlGaAs Regrown On AiGaAs By MOCVD

Abstract

Epitaxial regrowth on existing epilayers is an important technique in obtaining complicated nonplanar device geometries requiring intermediary processing steps between epitaxial growths. Such processing is an important step in the realization of optoelectronic integrated circuits (OEIC's). However, one of the major difficulties encountered with regrowth on the AlGaAs system is that the episurfaces are exposed to atmosphere during processing steps. These surfaces oxidize, forming a layer of interface states which degrade device performance. In this study, several methods have been investigated to reduce these interface states, thereby improving device characteristics. Currently, no growth technique exists to completely eliminate surface states formed at the regrowth interface. Previously, two methods have been used to reduce the effects of these states. The first method is to remove the interface region with an in-situ etchant (usually HCI) before beginning the second growth. The second method is to passivate the exposed surface using an inorganic sulfide, which ties up the oxygen bonds so that the reactivity with the aluminum species is reduced. In this study, these two methods are investigated to determine their effect on the electrical and optical characteristics of regrown laser diodes. The samples were grown in a horizontal MOCVD reactor at 100 Torr. The growths were conducted at a temperature of 725" C, a V/III ratio of 50, and a growth rate of 500 &min. The sources used for AlGaAs were 100% arsine, trimethylgallium, and trimethylaluminum. Diethyltellurium (74 ppm in H2) was used as an n-type dopant and carbon tetrachloride (489 ppm in H2) was used as a p-type dopant in the laser structures. In the regular. single-growth structures, the (100) GaAs substrate cleaning procedure consists of successive rinses in trichloroethylene, acetone, isopropyl alcohol, and deionized water. The cleaning procedure is then followed with a light etch consisting of NQOH:H202:H20 (1:1:400) for 15 seconds. The wafers are then spun dry before being loaded into the reactor through a load-locked glove box. A hightemperature pregrowth heat treatment is usually conducted prior to the in-situ HCl etching step. The heat treatment consists of exposing the wafer to an elevated temperature of 780" C under an arsine overpressure for 5 minutes. For the regrown samples, the same cleaning and etching procedure was used, which is subsequently followed by spinning on the ammonium sulfide solution, ("4)2S. before loading into the reactor. The time between application of the (NH4)zS and the exposure to the oxygen-free reactor atmosphere was approximately 5 minutes. The time elapsed between the initial growth and the regrowth was approximately 24 hours, during which the AlGaAs layers were exposed to atmosphere under a laminar flow hood. Several problems with in-situ HC1 etching result from the fact that the interface region can be quite thick. A deep etch to remove the interface defects can significantly alter the non-planar structure created before the epitaxial regrowth. In addition, the depth of the etch can be hard to control and deep etches can ruin the surface morphology of the regrown layers. Hence, it is important to precisely control the etch depth to minimize these problems. To determine the etch depth for various HC1 exposure times, portions of a wafer were masked off with Si3N4 and subsequently cleaned and etched as described above. After loading the samples into the reactor, the normal growth routine of a heat treatment and HCI etch were conducted. The step between the masked and unmasked regions was then measured to give the etch depth. As shown in Fig. 1, the HC1 etch depth is a linear function in time with a constant substrate temperature and HCI flow. This relationship allows precise control of the etch depth. As expected, the GaAs etch rate is faster than AlGaAs with a decreasing etch rate as the aluminum content of the alloy is increased. Next, the concentration and thickness of the interface state layers were investigated with several test structures containing aluminum compositions of x=O.1 and x=0.4. The first growth consisted of a 3 p m AlXGal-,As layer ( n = l ~ l O ~ ~ cm9), a 2500 A AIxGal.As layer ( n = l ~ l O ~ ~ cm3), and a 50 8, undoped GaAs cap layer. Roughly 100 A of the test structure was then chemically etched with the 1:1:400 solution to expose the AI,Gal-,As oxidizing layer. After an exposure time of 24 hours, the samples were cleaned, light etched and the interface treatments were applied. The four interface treatments investigated were heat treatment only, heat treatment with ("4)2S surface passiLation, heat treatment with a 700 A HCI etch, and heat