Applied Physics Letters | 2021
Quantum efficiency and oscillator strength of InGaAs quantum dots for single-photon sources emitting in the telecommunication O-band
Abstract
We demonstrate experimental results based on time-resolved photoluminescence spectroscopy to determine the oscillator strength (OS) and the internal quantum efficiency (IQE) of InGaAs quantum dots (QDs). Using a strain-reducing layer (SRL) these QDs can be employed for the manufacturing of single-photon sources (SPS) emitting in the telecom O-Band. The OS and IQE are evaluated by determining the radiative and non-radiative decay rate under variation of the optical density of states at the position of the QD as proposed and applied in J. Johansen et al. Phys. Rev. B 77, 073303 (2008) for InGaAs QDs emitting at wavelengths below 1 μm. For this purpose, we perform measurements on a QD sample for different thicknesses of the capping layer realized by a controlled wet-chemical etching process. From numeric modelling the radiative and nonradiative decay rates dependence on the capping layer thickness, we determine an OS of 24.6 ± 3.2 and a high IQE of about (85 ± 10) % for the long-wavelength InGaAs QDs. Introduction Semiconductor quantum dots (QDs) are well-established solid-state nanocrystalline structures that can be used as close-to-ideal two-level quantum emitters applicable for instance in quantum communication [1, 2, 3] or quantum computing [4, 5]. In recent years, performance of singlephoton emission based on InGaAs QDs could be pushed towards the physical limits for instance in terms of photon indistinguishability up to 99.7% [6] for QD emitting in the 930-950 nm range. Moreover, by material engineering the range of applications could be extended to the telecom O-band at 1.3 μm [7, 8, 9, 10] which enables long-distance fiber-based quantum communication [11]. A very promising way to achieve the required redshift of QD emission is the introduction of a strain-reducing layer of lower Indium content on top of the QD layer [12] maintaining the high quality of emission in terms of multi-photon suppression [9]. An additional advantage of this technology route is the usage of a mature and well-established GaAs-based technology platform which facilitates the fabrication of advanced quantum light sources with enhanced photon extraction efficiency [13, 14, 15] and the possibility of spectral fine-tuning by external strain fields [16]. The QDs used in this work have been carefully optimized for the manufacturing of single-photon sources emitting in a wavelength range around 1.3 μm at temperatures up to 40 K as detailed in previously published works [15, 16, 17]. An important intrinsic property of any QD emitter is the internal quantum efficiency (IQE) which describes the ratio of quantum conversion from exciton formation into single-photon emission. The IQE plays a major role in the total quantum efficiency of a fully-processed device and directly impacts the photon extraction efficiency being a key parameter of quantum light sources. Thus, knowing and increasing the IQE is crucial for device optimization. Another important QD parameter is the oscillator strength (OS) of the optical transition. The OS is directly related to the radiative decay time of the emitter and determines the light-matter coupling strength of QD devices using effects of cavity quantum electrodynamics [18]. The exciton decay dynamics and IQE of In(Ga)As QDs emitting in the 920 to 1030 nm range have been experimentally examined by several groups [19, 20, 21, 22] with a method introduced by Johansen et al. [19] which allows for a direct measurement of the QD properties without being influenced by QD density or relying on assumptions based on theoretical considerations of the exciton radius and the dimensions of the QDs [23, 24]. This method was also used to investigate an InGaAs quantum dot quantum well system [25]. So far however, to our best knowledge, no such measurements have been performed on InGaAs QDs emitting in the telecom O-band, so that reliable information on their IQE has not been available. In order to determine the IQE and OS of telecom wavelength InGaAs QDs their total decay rate Γ(ω, z) at a frequency ω and at z position below the sample surface is determined experimentally. The associated decay rate is given by the following relation: Γ(ω, z) = Γnrad(ω) + Γrad (ω) ρ(ω,z) ρhom(ω) (1) where Γnrad(ω) represents the nonradiative decay rate and Γrad (ω) is the radiative decay in a homogeneous medium, ρ(ω, z) is the projected local density of states (LDOS) divided by the LDOS in a homogenous medium ρhom(ω) (the GaAs surrounding the QD). The total decay rate Γ(ω, z) in dependance from z can be measured experimentally by means of time-resolved micro-photoluminescence (μPL) under variation of the capping layer thickness of the QD sample. Subsequently, the experimental components Γrad(ω) and Γnrad hom (ω) are obtained by fitting parameters of a curve that has been derived from simulated values of the LDOS in dependance from z to the experimental data. The IQE can then be calculated as: IQE = Γrad hom Γnrad+Γrad hom , (2) and the OS can easily be derived from fosc(ω) = 6meε0πc0 3 q2nω2 Γrad (ω) (3) where the electron mass me, the vacuum permittivity ε0, the vacuum speed of light c0, the elementary charge q and the refractive index n of the surrounding GaAs is given [9]. Sample Preparation The purpose of this work is to obtain a quantitative understanding of 1.3 μm InGaAs QDs, which were optimized for the development of quantum light sources emitting in the telecom Oband, in terms of their IQE and the OS. Using metal-organic chemical vapor deposition (MOCVD) the corresponding sample was prepared in the following way. First, a 300 nm GaAs buffer is grown on n-doped GaAs (100) substrate and followed by 1 μm thick Al0.90Ga0.10As layer. Then a 2 μm thick GaAs layer and another 100 nm thick Al0.90Ga0.10As layer is deposited. These three layers are originally designed as sacrificial layers to be removed during a postgrowth flip-chip processing, as described in more detailed way in Refs. [15, 16]. The final 879 nm thick layer of GaAs formerly designed to function as active device membrane includes a single InGaAs QD layer together with the SRL. The QD layer with a density of 5x10 cm is located 637 nm above the second etch stop layer and is formed by 1.5 monolayers of In0.7Ga0.3As followed by GaAs flush corresponding to a nominal thickness of half a monolayer. The subsequent 5.5 nm thick InGaAs SRL has a gradual decrease of the In-content from 30% to 10% over the first 3.5 nm. This layer is followed by a GaAs capping layer with a thickness of 236 nm. An overview of the whole sample structure is given in Fig. 1(a). Further information on the QD growth and optimization can be found in the supplementary information. In order to vary the thickness of the caping layer as required for extracting the radiative and non-radiative emission rate of the QDs the grown sample was prepared as described in the following. First, the surface of the sample was cleaned thoroughly and the photo resist AZ 701 MIR was applied using spin coating. UV lithography was used to pattern the surface with several slim rectangular structures to form a height reference to determine the capping layer thickness after the upcoming etching process. The sample was cut into 25 pieces to undergo wet-chemical etching after thorough cleaning with isopropanol. The etching solution [26] was prepared with the following ingredients: Hydrobromic acid (HBr), hydrogen peroxide (H2O2) and water (H2O) in a volumetric ratio of 2:1:60. The solution was continuously agitated with a magnetic stirrer and left to rest for 600 seconds after adding the H2O2. The etching rate had been previously determined to be 0.5 nm/s by etching and measuring a set of test sample pieces of GaAs. The 25 pieces were put together into the solution and subsequently removed piece after piece after a set time interval to etch away the desired amount of GaAs capping and put into clean water to stop the etching process. The remaining thickness of the GaAs separating the QD layer from the sample surface was determined with comparative measurement in a stylus profiler and an atomic force microscope (AFM) setup. Both measurements agree within a margin of less than 10 nanometers. Because of local deviations in the etching rate the exact position of the photoluminescence lifetime measurement is later marked on an optical microscopic image and the local top GaAs layer thickness remeasured via AFM. These values, as given in Fig. 1(b), are used in later analysis of the μPL data. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 0 50 100 150 200 250 (b) D is ta n c e Q D -s u rf a c e ( n m )