Elsevier

Journal of Crystal Growth

Volume 511, 1 April 2019, Pages 33-41
Journal of Crystal Growth

Analysis of GaAsBi growth regimes in high resolution with respect to As/Ga ratio using stationary MBE growth

https://doi.org/10.1016/j.jcrysgro.2019.01.010Get rights and content

Highlights

  • Methodology to determine flux and temperature distributions for stationary MBE.

  • GaAsBi growth regimes determined in high resolution with respect to the As/Ga ratio.

  • Strong variation of Bi incorporation and structural properties.

  • Material optimization by fine-tuning the As/Ga ratio.

Abstract

The control of Bi incorporation and material properties in III-V-Bi alloys has proved challenging due to their high sensitivity to the epitaxial growth parameters. Here, we present a methodology for determining the variation in the Ga, As, and Bi fluxes and the temperature across a stationary substrate in molecular beam epitaxy. By correlating the flux distributions with material properties, we identify distinct regimes for epitaxy of GaAsBi. In particular, we devise a detailed image of the interplay between Bi incorporation and structural properties of a bulk GaAs0.96Bi0.04 layer grown on GaAs(1 0 0) with respect to the As/Ga ratio. The influence of As/Ga is analyzed with high resolution over the important stoichiometric range (i.e. As/Ga = 0.6–1.6). Growth outside the near-stoichiometric As/Ga regime leads to decreased Bi incorporation, decreased structural quality and the formation of Ga, Ga/Bi or Bi droplets. On the other hand, growth at As/Ga = 1.00–1.17 leads to maximized material quality. For this regime, the surface roughness is further optimized by fine-tuning the As/Ga ratio to suppress surface mounding to a value of 0.5 nm. The results reveal the extreme sensitivity of GaAsBi growth to small variations in the As/Ga ratio, and demonstrate the applicability of stationary growth in studying these effects.

Introduction

Epitaxy of III-V alloys with desired properties requires careful control of the growth parameters, including the temperature, the growth rate, and the ratios of the fluxes impinging on the substrate. To this end, using molecular beam epitaxy (MBE) is seen as advantageous owing to the possibility to independently control many of the parameters with good accuracy. Yet, in some cases, even very small variations in these parameters can cause disproportionally large changes in material properties. This is particularly valid for metastable III-V-Bi alloys, such as GaAsBi, GaSbBi and InAsBi, which have received increasing attention in recent years due to their advantageous properties in optoelectronic applications [1], [2], [3], [4]. The sensitivity to growth parameters arises from weak III-Bi bonding, which necessitates non-equilibrium growth conditions governed by kinetics to facilitate Bi incorporation. For example, reaching desired Bi concentrations of several at.-% in GaAsBi, while maintaining high material quality, has proved to be a major challenge that requires the use of low growth temperatures (<400 °C) to prevent segregation and desorption of Bi, as well as low As/Ga flux ratios due to the preferential As-Ga bonding [5], [6], [7].

To guide the MBE process, models relating the growth parameters to the resulting Bi concentration have been proposed [5], [6]. However, due to the large number of interplaying parameters and complexity of the system, their applicability is limited to specific areas in the growth parameter space based on the inherent assumptions in the models [1]. For example, the model by Lu et al. [5] assumes an As-rich surface, whereas the model by Lewis et al. [6] assumes a Ga-rich surface, both of which are applicable to different As/Ga regimes. Furthermore, in addition to reaching the desired Bi concentration, the ability to control the structural, optical and electrical properties to meet the needs of specific applications is required. To this end, the interplay between the material properties and growth conditions has to be considered. For example, X-ray diffraction (XRD) measurements indicated inhomogeneous Bi incorporation for the samples used in the study by Lu et al. [5]. Moreover, while GaAsBi layers with Bi concentrations up to 22% were achieved within the framework of Lewis' model [6], the layer surfaces were covered in droplets; we also note that neither of these studies reported photoluminescence (PL) emission, which requires good structural quality and low defect densities. In general, GaAsBi growth has been shown to be highly prone to inhomogeneous Bi incorporation [5], [7], [8], [9], clustering [10], composition modulation and ordering [11], [12], [13], surface segregation [14], high point defect concentrations [15], [16], [17] and broad PL emission spectra [7], [9], [18]. Several reports have also demonstrated the accumulation of excess Ga or Bi in the form of droplets on the layer surface. These droplets have been shown to consist either of Ga/Bi, Ga, or Bi [19], [20], [21], [22], [23], [24].

Attempts to elucidate the relation between growth parameters and material properties have been relatively scarce and generally targeted to specific parameter ranges. Ptak et al. [25] studied the interplay between the surface roughness and growth rate of GaAsBi, finding that smooth surfaces can be obtained by choosing an optimum growth rate depending on the desired Bi concentration; however, no verification of the structural quality within the layers or of the optical properties was provided. The only report addressing the effect of the As/Ga ratio on GaAsBi growth was given by Masnadi-Shirazi et al. [26], who attributed improved structural quality to growth with the Bi-induced (2 × 1) reconstruction observed near the stoichiometric As/Ga ratio, while samples grown with the (1 × 3) reconstruction at higher As/Ga showed degraded XRD features. These observations suggest that a systematic study of the material properties over the critical stoichiometric As/Ga ratio is necessary. Such studies have been carried out for GaAs in As-rich conditions while grown at low temperatures (LT-GaAs), similar to those required by GaAsBi [27], [28]. When grown at above-stoichiometric As/Ga ratios, LT-GaAs contains a high concentration of point defects. However, the structural quality can be improved and defect concentrations greatly reduced by carefully controlling the As/Ga ratio in the near-stoichiometric range in order to minimize the amount of excess As [27], [28]. Significant point defect concentrations have been reported for GaAsBi as well [15], [16], despite the surfactant effect associated with Bi [29]. Studies on MBE-grown GaAsBi samples typically state growth close to or slightly above the stoichiometric As/Ga ratio, without giving quantitative values and often not reporting the method of calibration. This is likely due to the difficulty in measuring the flux ratio accurately, limiting the available options to qualitative determination of the stoichiometric ratio with methods such as observing RHEED transitions [26] or diffuse light scattering [18]. As we will demonstrate, optimization of material quality requires control of the flux ratio with greater accuracy that may be achievable by conventional methods. Evidently this could have contributed to the large variation in the reported material properties of GaAsBi layers.

Considering the challenges outlined above, it is clear that further development of GaAsBi alloys requires mapping of more comprehensive parameter spaces while also considering the effects of small parameter deviations, which can cause significant variation in material properties. Conventionally, such a study would require a very large amount of growths, while simultaneously the effect of very small parameter variations could be masked by the uncertainty in controlling them. For example, typical group III fluxes in MBE, such as Ga, can be measured with better than 1% accuracy by an ionization gauge, whereas for the more volatile group V species, such as As, the uncertainty can be as high as 10% [30] due to the non-unity sticking of As on the growth chamber surfaces and exacerbated by the relative complexity of a typical two-zone As cracker source.

In this report, we present a combinatorial [31] methodology in which we utilize stationary (non-rotating substrate) MBE to address the effect of flux variations in GaAsBi growth. In typical MBE, the substrate is continuously rotated to average out the non-uniform molecular beams [32]. Conversely, by stopping the rotation flux gradients will form over the substrate. Myers [33] used this principle to study the effect of As/Ga flux ratio on the properties of GaMnAs by using experimentally derived flux distributions for Ga, As and Mn. Li et al. [34], Wood et al. [35] and Collar et al. [24] utilized it for GaAsBi growth by using theoretically calculated flux distributions, where the material sources were assumed to behave as point sources with a radial flux emission. In practice, flux distributions at the substrate location are affected by system-specific aspects such as the shape of the material container and distribution of material within, and can be distorted by material deposits in the beam path [36]. Thus, to accurately determine the flux distributions each constituent has to be experimentally measured. Here, this methodology is established for Ga, As and Bi.

Another potential source of uncertainty are growth temperature variations across the substrate arising from the radiative heating designs used in MBE [30]. To address this, we determine the range of temperature variation over the substrate to assess its effect on the results. Using stationary growth, we show how the structural properties of GaAsBi vary with respect to small changes in the As/Ga ratio near the stoichiometric condition, leading to clearly defined regions with respect to the material properties.

Section snippets

Fabrication and analysis method

All samples were grown on 2 in. diameter GaAs(1 0 0) substrates. Conventional effusion cells were used for Ga and Bi, and a two-zone cracker for As2. For all samples except the As calibration sample, the substrate was heated for 10 min at 620 °C to remove the native oxide, after which a GaAs buffer layer with a thickness of 150 nm was deposited at 580 °C. The growth rates were 0.5 µm/h for the GaAsBi sample and 0.4 µm/h for the AlAs/GaAs calibration sample. The layer thicknesses and growth

Results

The stationary growth method was utilized in the study of a GaAsBi layer with a nominal thickness of 160 nm and a Bi concentration of 4.2% grown at a temperature of 220 °C. Position-dependent XRD, SEM, AFM and PL measurements were performed over the analysis direction. The XRD rocking curves are shown in Fig. 6, where the data is centered to the narrow GaAs(0 0 4) peaks at zero arcseconds arising from the substrate and buffer layers. The GaAsBi peaks to the side of the GaAs peak have clearly

Discussion

As shown above, regime S is characterized by the lack of droplets and a decreasing surface roughness, as well as maximized Bi incorporation and high structural quality. The degradation of XRD features at the boundaries of this regime is likely a result of the finite width of the XRD beam. The 2 mm beam width corresponds to an As/Ga variation of ±2.7%. This means that the signal for these points originates partially from regimes L or H with degraded structural quality. We assume As/Ga to equal 1

Conclusions

We have investigated the incorporation of Bi and the resulting material properties for GaAsBi growth with respect to varying flux ratios. The study was performed using stationary MBE growth with experimentally determined flux distributions for each constituent. This allowed the characterization of material properties with arbitrarily small step size as a function of the As/Ga ratio using a sample from a single growth run while eliminating the effect of unintentional growth-to-growth variations.

Acknowledgements

This work was supported by the Academy of Finland projects HIGHMAT (No. 259111) and Transphoton (No. 284686), and the European Research Council (ERC AdG AMETIST, #695116).

References (57)

  • E.S. Tok et al.

    Arsenic incorporation kinetics in GaAs(001) homoepitaxy revisited

    Surf. Sci.

    (1997)
  • C.R. Tait et al.

    Spontaneous nanostructure formation in GaAsBi alloys

    J. Cryst. Growth

    (2018)
  • R.R. Wixom et al.

    Sb and Bi surfactant effects on homo-epitaxy of GaAs on (0 0 1) patterned substrates

    J. Cryst. Growth

    (2004)
  • F. Bastiman et al.

    Non-stoichiometric GaAsBi/GaAs (100) molecular beam epitaxy growth

    J. Cryst. Growth

    (2012)
  • A. Duzik et al.

    Surface morphology and Bi incorporation in GaSbBi(As)/GaSb films

    J. Cryst. Growth

    (2014)
  • O. Delorme et al.

    In situ determination of the growth conditions of GaSbBi alloys

    J. Cryst. Growth

    (2018)
  • I.P. Marko et al.

    Progress toward III–V bismide alloys for near- and midinfrared laser diodes

    IEEE J. Sel. Top. Quantum Electron.

    (2017)
  • O. Delorme et al.

    GaSbBi/GaSb quantum well laser diodes

    Appl. Phys. Lett.

    (2017)
  • I.C. Sandall et al.

    Demonstration of InAsBi photoresponse beyond 3.5 μm

    Appl. Phys. Lett.

    (2014)
  • X. Lu et al.

    Effect of molecular beam epitaxy growth conditions on the Bi content of GaAs1−xBix

    Appl. Phys. Lett.

    (2008)
  • R.B. Lewis et al.

    Growth of high Bi concentration GaAs 1–x Bi x by molecular beam epitaxy

    Appl. Phys. Lett.

    (2012)
  • D.F. Reyes et al.

    Bismuth incorporation and the role of ordering in GaAsBi/GaAs structures

    Nanoscale Res. Lett.

    (2014)
  • A.R. Mohmad et al.

    Bismuth concentration inhomogeneity in GaAsBi bulk and quantum well structures

    Semicond. Sci. Technol.

    (2015)
  • S. Imhof et al.

    Clustering effects in Ga(AsBi)

    Appl. Phys. Lett.

    (2010)
  • A.G. Norman et al.

    Atomic ordering and phase separation in MBE GaAs1−xBix

    J. Vacuum Sci. Technol. B, Nanotechnol. Microelectron.: Mater. Process. Measur. Phenom.

    (2011)
  • M. Wu et al.

    Observation of atomic ordering of triple-period-A and -B type in GaAsBi

    Appl. Phys. Lett.

    (2014)
  • E. Luna et al.

    Spontaneous formation of nanostructures by surface spinodal decomposition in GaAs1−xBix epilayers

    J. Appl. Phys.

    (2015)
  • E. Luna et al.

    Spontaneous formation of three-dimensionally ordered Bi-rich nanostructures within GaAs1− xBix/GaAs quantum wells

    Nanotechnology

    (2016)
  • Cited by (12)

    • Low bandgap GaAsNBi solar cells

      2024, Solar Energy Materials and Solar Cells
    • Molecular beam epitaxy growth of GaSb<inf>1-x</inf>Bi<inf>x</inf> without rotation

      2019, Vacuum
      Citation Excerpt :

      Some growth parameter, e.g. Sb/Ga flux ratio, has big influence on the Bi incorporation and formation of surface droplets [13,14], and finely tuning the parameters by a small step is needed to map the interrelations. The positions of source cells with respect to the substrate holder in an MBE system provide a natural fine flux variation over the sample surface without sample rotation [15,16]. For example, by properly choosing both Sb and Ga flux, it is possible to achieve Sb-rich growth condition in one end and the Ga-rich growth condition in the other end.

    • Epitaxial phases of high Bi content GaSbBi alloys

      2019, Journal of Crystal Growth
      Citation Excerpt :

      It involves stationary growth of structures similar to typical growth rate calibration samples and their ex-situ spatial characterization. These structures were chosen as in our previous work [29], i.e. (i) an AlAs\GaAs heterostructure, (ii) Bi droplet epitaxy on GaAs(0 0 1), and (iii) an Sb-cap deposited at growth temperature of Tg < 40 °C for the Ga, Bi and Sb flux distributions, respectively. For example for the Ga flux distribution, the top GaAs layer of the AlAs\GaAs heterostructure was grown stationarily, producing a thickness gradient (proportional to the Ga flux variation) over the wafer, which was then measured with high resolution x-ray diffraction (HR-XRD).

    View all citing articles on Scopus
    View full text