Photoluminescence investigation of GaAs 1 - x Bi x /GaAs heterostructures Vaidas Pačebutas, Renata Butkutė , Bronius Čechavičius, Julius Kavaliauskas, Arūnas Krotkus Department of Optoelectronics, Center for Physical Sciences and Technology, Vilnius, Lithuania abstract article info Article history: Received 24 November 2011 Received in revised form 1 June 2012 Accepted 11 June 2012 Available online 16 June 2012 Keywords: Bismides Molecular beam epitaxy Photoluminescence In this work the impact of doped GaAs buffer and cap layers on the bismide photoluminescence (PL) spectra has been studied. For this study three types of heterostructures were grown: a thin GaAsBi layer deposited directly onto the nominally undoped GaAs buffer layer, a GaAsBi layer grown onto the GaAs:Be layer, and a GaAsBi layer deposited onto the GaAs:Be layer and capped with the GaAs:Be layer. It has been demonstrated that p-type doping of the GaAs buffer and cap layers is resulting in a signicant increase of the PL intensity. This enhancement was explained by a better photoexcited electron and hole connement in the GaAsBi layer. Published by Elsevier B.V. 1. Introduction Bismuth containing semiconductor alloys are attracting an in- creasing interest because the Bi incorporation into GaAs results in a strong band gap reduction of 70 to 90 meV per % of Bi [1,2]. This fact and a weaker band gap temperature dependence than in tradi- tional semiconductor alloys makes GaAs 1 - x Bi x an interesting active layer material for GaAs-based laser diodes emitting in the telecom wavelength range [3]. One of the most important methods of charac- terizing materials for light emitting semiconductor devices is the measurement of their photoluminescence spectra that give informa- tion on the structural quality and the efciency of the radiative pro- cesses in these materials. PL of GaAs 1 - x Bi x was investigated in a number of articles [46]. It has been found in [5] that the efciency of the photoemission is increasing monotonically up to Bi-content of 6%. At higher Bi-contents the PL amplitude drops drastically in a GaAs 1 - x Bi x layer with x = 0.1 that corresponds to the band gap wavelength of ~ 1.55 μm, PL amplitude is nearly by three orders of magnitude smaller than that for the composition when it is at the maximum [6]. One possible explanation of this reduction of the PL efciency is larger density of the non-radiative recombination centers and shorter carrier lifetimes in GaAs 1 -x Bi x layers with larger x and larger lattice mismatch the with GaAs substrate [7]. Additional complication could arise due to the photoexcited electron and hole separation at the GaAs 1 -x Bi x /GaAs heterostructure. All PL measurements up now were performed on GaAs 1 -x Bi x layers grown on unintentionally doped GaAs buffer. It has been shown in [8] that depending on the arsenic species used in the molecular-beam-epitaxy (MBE) reactor this buffer can be either slightly p-type (for As 4 sources) or slightly n-type (for As 2 source) doped. The conduction band off-set in the GaAs 1 -x Bi x /GaAs heterostructure is also undened yet; its value changes from 0% (when it is assumed that all the band structure variation is caused by the valence band anti- crossing effect [9]) to ~30% (theoretical value evaluated in [10]). In the present work, a study of the PL effect in GaAs 1 -x Bi x layers with x ~ 5.5% grown on differently doped GaAs buffers was performed. It has been shown that p-type doping of GaAs buffer and cap layers is resulting in a signicant increase of the PL intensity. 2. Layer growth and structural characterization The samples were grown in a SVT-A MBE deposition system equipped with conventional Knudsen effusion cells for metallic Ga, Bi and Be, and a two-zone arsenic cracker source to produce As 2 .A semi-insulating undoped (100)-oriented GaAs was selected as a sub- strate. Three types of heterostructures were grown: (A) a thin GaAsBi layer deposited directly onto the nominally undoped GaAs buffer layer, (B) a thin GaAsBi layer deposited onto the GaAs:Be layer, and (C) a thin GaAsBi layer deposited onto the GaAs:Be layer and capped with the GaAs:Be layer. The substrate temperature was monitored by thermocouple read- ings with an accuracy of 1 °C. Beam equivalent pressures (BEP) of Ga, As 2 , and Bi were measured with the retractable ion gage. The cleaning of wafers and growth was in-situ monitored by reection high- energy electron diffraction. Prior to the growth of samples, the GaAs wafers have been heated up to 565 °C for 10 min under arsenic ux to remove a native oxide. After deoxidation procedure, the substrate was cooled down to ~530 °C. A 50 nm thick-GaAs buffer layer (undoped or Be-doped) was grown rst under standard GaAs growth conditions keeping high substrate temperature T ~ 530 °C and the beam equivalent pres- sure ratio As 2 /Ga from 7 to 10 with the 0.2 μm/h growth rate. The Thin Solid Films 520 (2012) 64156418 Corresponding author. E-mail address: renata@p.lt (R. Butkutė). 0040-6090/$ see front matter. Published by Elsevier B.V. doi:10.1016/j.tsf.2012.06.047 Contents lists available at SciVerse ScienceDirect Thin Solid Films journal homepage: www.elsevier.com/locate/tsf