Fundamental role of arsenic flux in nanohole formation by Ga droplet etching on GaAs(001)
© Fuster et al.; licensee Springer. 2014
Received: 5 May 2014
Accepted: 11 June 2014
Published: 18 June 2014
Skip to main content
© Fuster et al.; licensee Springer. 2014
Received: 5 May 2014
Accepted: 11 June 2014
Published: 18 June 2014
Nanoholes with a depth in the range of tens of nanometers can be formed on GaAs(001) surfaces at a temperature of 500°C by local etching after Ga droplet formation. In this work, we demonstrate that the local etching or nanodrilling process starts when the Ga droplets are exposed to arsenic. The essential role of arsenic in nanohole formation is demonstrated sequentially, from the initial Ga droplets to the final stage consisting of nanoholes surrounded by ringlike structures at the surface and Ga droplets consumed. The kinetics of local etching depends on the arsenic flux intensity, while the ringlike structures are basically the same as those formed underneath the droplets in the absence of arsenic. These structures show motifs with well-defined crystalline facets that correspond to those expected from surface energy minimization. These experimental results are qualitatively analyzed for a better understanding of the nanohole formation underlying processes.
Semiconductor quantum dots (QDs) have been extensively studied in the last years. The quantum confinement effect of these structures allows the design of novel devices related to a wide range of applications in electronics and optoelectronics [1, 2]. Self-assembled QDs have been successfully fabricated by the epitaxial growth of a layer in a lattice-mismatched III-V semiconductor system through the well-established Stranski-Krastanov (SK) process. Although a lot of fundamental physical understanding and a variety of applications have been realized using this kind of QDs, custom design of the shape and size of the nanostructures is seriously constrained by the self-assembling processes. The droplet epitaxy (DE) technique is another way to obtain QDs with some advantages over the SK mode . For example, QDs of lattice-matched materials (as GaAs/AlGaAs) can be formed by DE. A variety of shapes have been obtained by this technique: dots, rings, concentric double-ring structures, dot pairs [4–6]. Several nanostructures fabricated by DE have been implemented in devices as lasers, detectors, single-photon emitters, and solar cells [7–11]. An effect of the DE technique is the ‘nanodrilling’ of the substrate or local etching by the droplets, which has been recently studied as an in situ procedure to obtain nanoholes [12–14]. The nanodrilling process has its origin in the etching of a semiconductor by a liquid metal [15–17]. For Ga droplets on GaAs(001), we have observed the etching process for substrate temperatures ≥450°C. The nanoholes formed by DE provide cleaner interfaces than those formed by any other ex situ lithographic techniques without any need of special treatments for further regrowth processes. By depositing a III-V semiconductor of lower bandgap, the nanoholes can be refilled and QDs are formed at the nanoholes. The density of the holes determines the density of the QDs and their size depends on the amount of deposited material to form them, being relatively easy to tune the emission wavelength independently of the density . The optical properties of these QDs are also influenced by the characteristics of the nanoholes. For example, the depth and shape of the nanoholes are determinant in obtaining GaAs/AlGaAs QDs with narrow line shape and null fine structure splitting . Moreover, the kind of QD/nanohole interface would be in the origin of the charge exciton species predominant in the micro-PL spectra of InAs/GaAs QD  and in the formation of QD molecules instead of single QD . In order to take advantage of all the potential of droplet epitaxy as a nanopatterning technique, a complete understanding of the mechanisms of nanohole formation is mandatory.
A lot of experimental and theoretical work has been reported (, Chap. 3 and references therein, [22, 23]) to explain the droplet crystallization evolution at a low temperature (<300°C, where nanoholes are not observed). Although some works have also been dedicated to model local droplet etching [24, 25], experimental results showing step by step the full process would be of great help for a deeper understanding.
In this work, we monitor the hole formation process during the transformation of Ga droplets into nanoholes on GaAs(001) surfaces at substrate temperature TS = 500°C. This process takes place when Ga droplets are exposed to arsenic. The essential role of arsenic in nanohole formation is demonstrated sequentially, from the initial Ga droplets to the final stage consisting of nanoholes at the surface and Ga droplets completely consumed. For this purpose, we have grown samples at different stages of the local etching process under several annealing conditions, and we have studied the dependence of the depth of the nanoholes with arsenic flux and annealing time. The experimental results are qualitatively analyzed for a better understanding of the processes underlying the nanohole formation.
The samples under study were grown on GaAs(001) substrates by molecular beam epitaxy (MBE) in two different reactors: a homemade MBE system and a RIBER (Paris, France) Compact 21E MBE system. Previous to any growth process, oxide from the GaAs(001) surface was thermally removed and a 150-nm-thick GaAs buffer layer was grown at a growth rate of 0.5 monolayer (ML) per second at substrate temperature TS = 580°C. The droplets were formed by depositing at TS = 500°C 4 ML of Ga at 0.04 ML/s, denoted in equivalent monolayers of GaAs on GaAs(001). For ensuring a minimal As background pressure in the MBE reactor before Ga is deposited, we follow specific procedures in the different MBE systems. In the RIBER Compact 21E MBE, once the As cell valve is closed, we wait until the background pressure reading is lower than 3 × 10−9 Torr. In the homemade MBE system, we need to cool down the As cell besides closing its valve, to achieve a final background pressure reading lower than 1 × 10−9 Torr. With these procedures, reproducible results are obtained independently on the system where the samples were grown. After droplet formation, the surface was annealed either under As4 flux or in the absence of arsenic during different times. The different As fluxes used in this work are also indicated in equivalent ML/s, 1.40, 0.70, and 0.08 ML/s, and were measured by monitoring the specular beam RHEED oscillations during GaAs growth limited by V element . The samples annealed under arsenic flux were cooled down in the presence of arsenic before taken out from the MBE chamber.
The morphology of Ga droplets and nanoholes was measured by atomic force microscopy (AFM) in a Nanotec (Tres Cantos, Spain) and/or a Veeco Dimension Icon (Plainview, NY, USA) scanning probe microscopy system, using Nanosensors silicon cantilevers (K = 40 to 50 N/m, Neuchatel, Switzerland) with small radius tips (≤7 nm) in tapping mode. For AFM data analysis, the free Gwyddion software was employed.
Contrary to the previously published works [12–14], our results show that in the absence of arsenic, the Ga droplets formed at TS = 500°C remain at the GaAs(001) surface after growth interruptions (at TS = 500°C) ranging from 5 to 30 min. Under these experimental conditions, no nanoholes appear across the surface. An actual low As pressure in the system background is the key point for reproducing this result. In fact, in our homemade MBE system, nanoholes appear (results not shown) if the As cell is not cooled down, besides being fully closed, previously to Ga deposition for droplet formation, in complete agreement with the experimental results reported by other authors up to date.
When the Ga droplets are removed by HCl chemical etching (Figure 1b), the surface shows ≈ 2-nm-deep flat depressions in the areas previously occupied by the droplets. These depressions are caused by the dissolution of the GaAs substrate by metallic Ga droplets, incorporating As atoms from the substrate until a stable composition is reached. The composition of the resulting alloy is limited by the arsenic solubility in Ga at 500°C , being Ga-rich enough to be etched by HCl. The observed depressions are surrounded by GaAs ringlike structures whose diameter is similar to that of the corresponding Ga droplet. A similar phenomenology was observed in Ga droplets formed at TS = 350°C  and in ten times larger Ga droplets created by annealing a GaAs(001) substrate at 670°C, above the surface congruent evaporation temperature .
These depressions show a quasi-square shape with their sides along <110 > directions. They are surrounded by GaAs ringlike structures with four sectors (one for each side of the depression) aligned along <110 > directions. Among the four sectors of the ring, three are similar in height (≈5 nm). The other one is higher (≈8 nm) and always appear along one of the  sides; from this point on, this sector will be referred as the main sector.
Figure 3b shows a plane view of a three-dimensional (3D) representation of an AFM image with the edges marked by dashed lines to clearly distinguish the facets; a scheme of the proposed structure with the corresponding facets labeled is depicted on Figure 3c. The facets forming the main sector correspond to the family planes that are obtained by surface energy minimization calculations [30–32] for the equilibrium shape of GaAs crystals. So, we can conclude that this faceted structure is a minimum energy state of the GaAs grown from Ga coming from the droplet and As coming from the substrate (in the absence of arsenic) and also from the incoming arsenic flux when the As cell valve is opened.
The above described results point out the similarities of the nanorings formed at the surface when the Ga droplets are exposed to arsenic and below the Ga droplets in the absence of arsenic. But there is a fundamental difference between both results: nanoholes only appear if the droplets are exposed to arsenic.
These results show that the nanohole formation process is activated when Ga droplets are exposed to arsenic, while in the absence of arsenic, only flat depressions beneath the Ga droplets are observed. Arsenic exposure also leads to the consumption of the Ga droplets.
It is well known that As supply to Ga droplets triggers the onset of different processes [4, 21–23], in particular a change in Ga droplet composition due to the incoming arsenic diffusion through metallic Ga, driving the Ga droplet arsenic content out of the equilibrium value at the corresponding temperature. In order to restore the arsenic equilibrium composition, Ga atoms belonging to the substrate would migrate towards the Ga droplet, if kinetics is not inhibited, with the subsequent enhancing of local substrate dissolution and the onset of the nanohole formation process. This explains why nanoholes penetrating in the substrate only appear in the presence of arsenic at high enough substrate temperatures. Simultaneously to the nanodrilling effect, GaAs is forming around and at the edge of the Ga droplet as has been previously reported [6, 23], leading to its consumption at a rate that will depend on TS and As flux. In this way, there is a competition between Ga coming from the substrate that incorporates at the Ga droplet and droplet consumption by forming GaAs. Altogether, a Ga droplet under As gives rise to ringlike nanostructures surrounding a deep and narrow hole that can penetrate up to tens of nanometers into the substrate.
These processes are closely related to the Ga-assisted vapor-liquid-solid growth of nanowires, where the incorporation of Ga and As and the GaAs crystallization take place below and around the Ga droplet , being in our case the source of Ga is the GaAs substrate instead of an incoming Ga flux.
In view of our results, we can outline the following processes running during the annealing of Ga droplets under As exposure, which are associated to the characteristic evolution rates: local etching by the metallic Ga droplets (I) active until the Ga droplets are consumed by GaAs growth (II) and evolution of nanoholes to shallower structures (III).
In this context, it can be explained that the annealing time for reaching the nanohole maximum depth by nanodrilling beneath the Ga droplet (process I) depends on As flux, as the consumption rate of the droplet by GaAs formation (process II) depends on As flux in MBE growth under growth conditions limited by V element . Once the etching is over by consumption of the Ga droplets (nanohole maximum depth achieved), a further annealing time under As flux leads to a reduction of the nanohole depth due to the incorporation of Ga atoms at B-type walls coming from the lateral movement of Ga surface atoms during the annealing process, a behavior observed in any patterned surface at high temperature .
In this work, we have studied the formation of nanoholes on GaAs(001) substrates produced after Ga droplet epitaxy at TS = 500°C. Our results show that nanodrilling of the GaAs(001) substrate is only possible in the presence of arsenic. We have identified three processes that take place when Ga droplets are exposed to an arsenic flux: (I) local etching by the metallic droplet, (II) GaAs growth by consumption of the Ga droplet under As supplied, and (III) evolution of nanoholes to shallower structures. In this picture, the key role of arsenic flux would be the reactivation of dissolution of the GaAs substrate by the metallic Ga droplets and further GaAs growth, processes that are also in the origin of the well-known flat depressions beneath the Ga droplets in the absence of an arsenic flux. Actuation on the kinetics of the processes involved in nanohole formation may facilitate obtaining nanoholes under design, which ultimately will influence the optical properties of the nanostructures formed inside.
We want to acknowledge the financial support from the Spanish MINECO through grants TEC2011-29120-C05-01/04, ENE2012-37804-C02-02, and AIC-B-2011-0806. We also want to acknowledge Raquel Álvaro from the Micro- and Nano-fabrication service (MiNa) at IMM for the AFM measurements. We acknowledge the support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.