Discover millions of ebooks, audiobooks, and so much more with a free trial

Only $11.99/month after trial. Cancel anytime.

Molecular Beam Epitaxy: From Research to Mass Production
Molecular Beam Epitaxy: From Research to Mass Production
Molecular Beam Epitaxy: From Research to Mass Production
Ebook2,461 pages32 hours

Molecular Beam Epitaxy: From Research to Mass Production

Rating: 0 out of 5 stars

()

Read preview

About this ebook

This multi-contributor handbook discusses Molecular Beam Epitaxy (MBE), an epitaxial deposition technique which involves laying down layers of materials with atomic thicknesses on to substrates. It summarizes MBE research and application in epitaxial growth with close discussion and a ‘how to’ on processing molecular or atomic beams that occur on a surface of a heated crystalline substrate in a vacuum.MBE has expanded in importance over the past thirty years (in terms of unique authors, papers and conferences) from a pure research domain into commercial applications (prototype device structures and more at the advanced research stage). MBE is important because it enables new device phenomena and facilitates the production of multiple layered structures with extremely fine dimensional and compositional control. The techniques can be deployed wherever precise thin-film devices with enhanced and unique properties for computing, optics or photonics are required. This book covers the advances made by MBE both in research and mass production of electronic and optoelectronic devices. It includes new semiconductor materials, new device structures which are commercially available, and many more which are at the advanced research stage.
  • Condenses fundamental science of MBE into a modern reference, speeding up literature review
  • Discusses new materials, novel applications and new device structures, grounding current commercial applications with modern understanding in industry and research
  • Coverage of MBE as mass production epitaxial technology enhances processing efficiency and throughput for semiconductor industry and nanostructured semiconductor materials research community
LanguageEnglish
Release dateDec 31, 2012
ISBN9780123918598
Molecular Beam Epitaxy: From Research to Mass Production

Related to Molecular Beam Epitaxy

Related ebooks

Physics For You

View More

Related articles

Reviews for Molecular Beam Epitaxy

Rating: 0 out of 5 stars
0 ratings

0 ratings0 reviews

What did you think?

Tap to rate

Review must be at least 10 words

    Book preview

    Molecular Beam Epitaxy - Mohamed Henini

    USA

    Chapter 1

    Molecular beam epitaxy

    fundamentals, historical background and future prospects

    Secondo Franchi

    CNR-IMEM Institute, Parco delle Scienze, Parma, Italy

    Chapter Outline

    1.1 Introduction

    1.2 Basics of MBE

    1.3 The technology of MBE

    1.3.1 MBE machines

    1.3.2 MBE growth chambers

    1.3.3 Sources of molecular beams

    1.3.3.1 Thermal-evaporation cells

    1.3.3.2 Valved cracker cells

    1.3.4 Variants of the MBE process

    1.3.4.1 Gas source MBE, metalorganic MBE and chemical beam epitaxy

    1.3.4.2 Nitride MBE and reactive MBE

    1.3.4.3 Group-IV MBE

    1.3.4.4 Migration enhanced epitaxy (MEE) and atomic layer MBE (ALMBE)

    1.3.4.5 Droplet epitaxy

    1.3.5 Measurements of molecular beam fluxes

    1.3.6 Control of composition and doping profiles along growth direction

    1.4 Diagnostic techniques available in MBE systems

    1.4.1 RHEED and surface reconstructions

    1.4.2 STM and surface reconstructions

    1.4.3 Other techniques (AES, SIMS, XPS and UPS)

    1.5 The physics of MBE

    1.5.1 Group-V and group-III species on III–V surfaces

    1.5.2 2D layer-by-layer growth mechanism

    1.5.2.1 Monte Carlo simulations of the MBE 2D growth process

    1.5.2.2 Evidence by RHEED of the 2D layer-bylayer growth mechanism

    1.5.2.3 Other evidences of the 2D layer-by-layer two-dimensional growth mechanism

    1.5.2.4 Consequences of the layer-by-layer growth mechanism

    1.5.2.5 Effects of interface roughness on device properties

    1.5.2.6 Layer-by-layer growth mechanism in lattice-matched or lowly latticemismatched structures

    1.5.3 Three-dimensional growth mechanism

    1.5.3.1 Three-dimensional growth of selfassembled nanoislands

    1.5.3.2 Strain relaxation as the driving force for 3D growth

    1.5.3.3 Metamorphic structures and quantum dot strain engineering

    1.5.4 MBE-assisted growth of one-dimensional structures

    1.6 Historical background

    1.7 Future prospects

    1.8 Conclusions

    References

    1.1 Introduction

    In the last four decades, it has been largely proved that epitaxial technologies for material growth have unique advantages over simpler counterparts, in spite of their generally higher technological costs. Interesting epitaxial technologies developed since the end of the 1960s are vapour-phase epitaxy (VPE), metalorganic VPE (MOVPE) and molecular beam epitaxy (MBE), along with its metalorganic variant (MOMBE).

    In this chapter it will be shown that MBE and variants are particularly well suited for the growth of advanced epitaxial structures; indeed, MBE allows for the growth of (i) materials with reduced concentrations of thermodynamical defects, due to the relatively low growth temperatures; (ii) structures where composition or doping profiles in the growth direction can be modulated in abrupt or continuous ways, a feature that has opened the way to (a) the preparation of new epitaxial structures, (b) the fabrication of innovative devices, (c) researches that were awarded the 2000 Nobel prize in physics (e.g., Alferov [1]; Nobel lecture and Kroemer [2]; Nobel lecture); and, finally, (iii) quantum structures, where engineered composition and doping profiles confine carriers in two- or three-dimensional regions with sizes smaller or comparable to the de Broglie wavelength of carriers (Esaki [3]; von Klitzing [4]; Tsui [5] and Störmer [6]; Nobel lectures).

    In this chapter it will be emphasised that MBE growth takes place according to two mechanisms that allow the fabrication of structures where carriers may undergo two- (2D) or three- (3D) dimensional quantum confinement; indeed, most of quantum structures (such as quantum wells, superlattices, selectively doped heterostructures, structures for resonant tunnelling, as well as quantum dots) and related devices have been demonstrated by MBE.

    Many classes of materials have been prepared by MBE: semiconductors, such as III–Vs, II–VIs, IV–VIs and IV–IVs, and also oxides, magnetic materials and metals; however, most of the work on the development of technology and physics of MBE reviewed in this chapter has been done on III–Vs.

    In this chapter, after the short introduction (Section 1.1), the basic elements of MBE technology are considered in Section 1.2. Section 1.3 deals with different aspects of MBE such as (i) layouts of MBE machines used for research purposes (Section 1.3.1), (ii) growth chambers for research (Section 1.3.2), and (iii) sources of molecular beams (Section 1.3.3); in Section 1.3.4 a description is given of the numerous variants of MBE, developed either for growing different materials or for taking advantages of particular features of MBE growth mechanisms. Then, two points are considered, which are particularly relevant for the growth of structures with engineered composition and doping profiles: (i) the measurement of molecular beam fluxes (Section 1.3.5) and (ii) the accurate control of such profiles (Section 1.3.6). One of the most interesting features of MBE is the availability of diagnostic techniques in growth or in analysis chambers, all connected under ultra-high vacuum (UHV). The use of these techniques, that in a few instances are available in-situ (in the growth chamber) and in real time (during the growth), accounts for the deep understanding of growth mechanisms, which, in turn, explains why MBE processes can be controlled very accurately and with high yields. Techniques such as reflection high-energy electron diffraction (RHEED), scanning tunnelling microscopy (STM), Auger electron spectroscopy (AES), secondary ion mass spectrometry (SIMS), X-ray photoemission spectroscopy (XPS) and ultraviolet photoemission spectroscopy (UPS) are briefly reviewed in Section 1.4. Section 1.5.1 deals with surface phenomena of group-III and group-V species on III–Vs substrates, which determine the kinetic mechanisms of MBE growth of III–Vs and which have been supposedly extended also to other systems; in Section 1.5 the 2D layer-by-layer growth mechanism is discussed, which results in the growth of interfaces between lattice-matched (or lowly mismatched) materials smooth on the atomic scale; in particular, proofs of the occurrence of such a mechanism are considered, which were obtained both experimentally and by Monte Carlo simulation of the growth process. Section 1.5.3, instead, deals with the 3D growth of self-assembled nanoislands, which takes place when the materials in structures are lattice-mismatched by more than 2–3%; it will be shown that this mechanism is the natural candidate for the preparation of quantum dot structures, characterised by the 3D confinement of carriers; in Section 1.5.3 it is shown how strain can be considered as a degree of freedom to engineer the band structure of metamorphic structures. In Section 1.5.4 a short consideration is given to the MBE-assisted growth of nanowires based on the vapour–liquid–solid mechanism. A résumé of the historical background of MBE is given in Section 1.6, while in Section 1.7 some prospects of MBE are presented; finally, in Section 1.8 a few conclusions are drawn.

    1.2 Basics of MBE

    Molecular beam epitaxy (MBE) is an epitaxial process by which growth of materials takes place under UHV conditions on a heated crystalline substrate by the interaction of adsorbed species supplied by atomic or molecular beams [7]. The layers or deposits have: (i) the same crystalline structure of the substrate or a structure with a similar symmetry and (ii) a lattice parameter differing from that of the substrate by no more than ∼10%. The beams generally have thermal energy and are produced by evaporation or sublimation of suitable materials contained in ultra-pure crucibles.

    In the archetypal case of III–V semiconductors grown on III–V substrates, such as InGaAsP on GaAs, adsorbed group-III atoms and group-V molecules migrate on the heated substrate surface until they interact in proximity of suitable vacant lattice sites, where they are incorporated into the solid phase (Section 1.5.1); in InGaAsP solid solutions, In and Ga, as well as As and P, are randomly distributed in the group-III and group-V sublattices, respectively. Atomic beams of elements such as Si and Be provide n-type and p-type doping in III–V compounds.

    As it will be discussed in more detail below, beams may also be produced by injection in the growth environment of gaseous species, such as AsH3 and PH3 (gas source MBE (GSMBE)) or of volatile metalorganic compounds carried by hydrogen flows (metalorganic MBE, (MOMBE)).

    Cells producing ionised beams with non-thermal energy have been proposed: they may generate Zn-ion beams to dope III–Vs [8] or As- and Sb-ion beams for Si ([9] [11], respectively) (Section 1.3.4.3); the ions are accelerated by electric fields (a few hundreds eV) towards growing layers in order to improve the sticking of dopants. Another example of beams with non-thermal energy is that of supersonic beams (a few tens eV) which are used to enhance surface migration of adsorbed species and, then, to improve the morphology of deposits [12].

    Interesting examples of epitaxial relationships between layers and substrates are those between (i) III–V (100) zinc-blende layers and substrates (such as GaAs/GaAs, AlGaAs/GaAs, InGaAs/GaAs, InGaAsP/InP and low N-content InGaAsN/GaAs); (ii) (0001) wurtzite InGaN layers on (111) Si, a-SiC and c-sapphire substrates; and (iii) (0001) wurtzite CdS on (111) InP, with the (11–20) CdS prismatic planes parallel to the (1–10) InP ones [13].

    As mentioned above, important requirements of MBE growth are (i) the use of substrates heated at temperatures that depend on materials to be grown (Section 1.6.2) and (ii) UHV conditions in the growth environment. As for (i), the relatively high substrate temperatures (500–600 °C for GaAs) activate the efficient migration of adsorbed species on the growth surface, which is necessary for the formation of an ordered lattice; this topic and the existence of optimum growth-temperature windows will be discussed more thoroughly while presenting the kinetic models of MBE growth (Section 1.6.2). On the other hand, UHV conditions are required for (i) minimising the incorporation of unintentional impurities from growth environment and, then, obtaining high-purity materials and (ii) optimising the surface morphology, which could be affected by surface contamination by specific impurities such as C [7].

    As for impurity incorporation, let us consider the case of GaAs, which in many cases of basic and technological interest should have a content of unintentional dopants of 10¹⁴ cm–3 or lower. The concentration of an impurity i incorporated into the solid during growth is given by the ratio c = (σi Ni)/(σGa NGa), where Nk is the number of atoms k (k = i, Ga) per unit surface impinging on the surface and σk is the incorporation coefficient of the species k; σk, in turn, is given by the ratio between numbers of atoms k incorporated and that of atoms impinging on the surface. While according to the kinetic mechanisms of MBE growth of GaAs (Section 1.5.1) NGa is related to the growth rate, from the kinetic theory of gases it follows that

    where pi, Mi, kB, NA and T are the partial pressure and the molecular weight of the species i, the Boltzmann’s constant, the Avogadro’s one and the absolute temperature, respectively [14]. In the case of incorporation of C, an impurity deriving from CO – a typical MBE background species – MCO = 28 and σC ∼ 1 × 10−3. Since T = 300 K and σGa = 1 (Section 1.5.1), it follows that, for GaAs growth rates of 1 ML/s (∼1 μm/h), in order to have C concentrations of 1 × 10−9, it is required that the CO backpressure is pCO = 1.6 × 10−12 Torr, a value which fully justifies the requirement of UHV conditions.

    A very interesting consequence of the UHV growth environment is that MBE growth takes place in a molecular regime, as opposed to a viscous one, typical of vapour-phase technologies such as MOVPE, which is an epitaxial technologies competing with MBE [15]. The molecular regime [14] is characterised by mean free paths between collisions of atoms and molecules in the beams larger or comparable to critical lengths of the growth system, such as distances between cells and substrates (<0.2 m). Under this regime, atoms and molecules basically do not interact during their paths and, then, mechanical beam shutters can be used to switch on and off the beams directed towards the substrate; in such a way, the composition of the nourishing phase can be abruptly changed in times given by the actuation times of shutters, that is, in the order of 0.1 s. Since the MBE growth rate generally is, at most, in the ML/s range, the thickness of interfaces between layers with different composition or doping can be in the order of or smaller than tenths of an ML (1 ML = 0.28 nm for (100) GaAs).

    In order to confirm that MBE growth generally takes place under a molecular regime, it is useful to recall that according to the kinetic theory of gases the mean free path λ between collisions of atoms or molecules at a pressure p is given by

    where D is the diameter of atoms or molecules in the beam [14]. Therefore, for typical beams and MBE or MOMBE operating pressures of 10−6–10−4 Torr, it follows that λ = 5–0.05 m, respectively (Sections 1.2 and 1.3.4.1).

    It is interesting to compare these values with those of MOVPE that is characterised by operating pressures in the 10–760-Torr range; consequently, λ is shorter than 50 μm and, then, the regime is viscous. Therefore, any surface immersed in the gas flow is surrounded by a boundary layer (BL or stagnant layer) due to the finite viscosity of gases; BLs are defined as regions close to any surface where the tangential component of gas velocity rapidly varies from vanishingly small values at the surface to values determined by the main gas flow [15].

    The existence of BLs has significant effects on the properties of epitaxial layers. In particular, since mass transport in BLs takes place by diffusion and since different chemical species may have significantly different diffusion coefficients, after any abrupt change in the vapour-phase composition, different species reach the growing surface at different times; therefore, layers of uncontrolled composition and doping are grown, until the new steady state is reached. In other words, BLs actually act as temporary sinks or reservoirs of the nourishing species which have been switched on or off, respectively. Since BLs exist not only around the substrate, but also close to the reactor walls and in the gas piping, times as long as minutes may be required to reach steady-state compositions of vapour phase; therefore, the existence of BLs may result in broadening of interfaces between materials with different composition and/or doping; the broadening can extend over hundreds of MLs or more, unless special precautions are taken in the design and the operation of epitaxial reactors, such as low-pressure operation [15].

    Instead, it is worth recalling that the UHV conditions of MBE growth environment (i) result in the absence of BLs and (ii) allow the use of mechanical shutters; both these features are instrumental in achieving interfaces abrupt on the atomic scale, when no hindrance is set by kinetic growth mechanisms (Section 1.5.1). The sharpness of interfaces is a prerequisite for the growth of nanostructures where effects of quantum confinement may show up.

    1.3 The technology of MBE

    1.3.1 MBE machines

    MBE machines may have different layouts, depending on (i) whether they are used for research or production, (ii) the materials to be grown and (iii) the specific variant of MBE technology that is actually implemented.

    In this section we will briefly deal with MBE machines for research, which are better suited for a discussion on the fundamentals of MBE technology. More details are given in review papers and books [7,14,16–21].

    Modern machines generally have a modular design, where each module is optimised for a definite process. Figure 1.1 shows a typical MBE machine for research on the growth of archetypal III–Vs; it consists of (i) two growth chambers; (ii) an analysis chamber fitted with different analytical techniques (Section 1.4); and (iii) a preparation chamber for in-situ processes such as etchings, metallisations or insulator depositions. The wafers are introduced in and extracted from the system through individually pumped introduction and extraction chambers, respectively, so that growth, analysis and preparations chambers, as well as transfer modules, are always under UHV. The transfers are carried out by means of suitable translators. The availability of diagnostic techniques in both growth and analysis chambers accounts for both the accurate control and the deep understanding of the MBE process (Section 1.4). Such an availability is strictly related to the relatively small chemical aggressiveness of MBE growth environment.

    FIGURE 1.1 Schematic view of a modular multichamber MBE machine.

    The use of multiple growth chambers is of interest for both: (i) increasing the yield of a single machine that can simultaneously carry out different growth runs and (ii) preparing heterostructures consisting of layers with significantly different compositions, which can be grown in different chambers to avoid any cross-contamination. A well-known example is that of structures made of III–V and II–VI layers, since group-II and -VI species act as unintentional dopants in III–Vs, as do III- and V-group atoms in II–VIs.

    Introduction chambers are provided with heaters to thoroughly outgas wafer holders before their insertion into growth chambers. In case of III–V substrates, the outgassing process is carried out at temperatures of hundreds of °C for tens of minutes.

    1.3.2 MBE growth chambers

    The basic elements of an MBE growth chamber are (Figure 1.2) (i) a stainless steel vessel, that in case of research machines generally has a diameter smaller than 20 inches, but that is much larger in case of production systems, (ii) cells to produce molecular beams, which will be described in the next sections, (iii) shutters to switch molecular beams on and off, (iv) a substrate holder heatable up to temperatures of several hundreds of °C and (v) a load-lock system which allows for the introduction of wafers into the growth chamber and for their extraction without breaking UHV in the growth environment. An electron gun is generally provided to study electron diffraction (RHEED) from growing surfaces in real time (Section 1.4.1)

    FIGURE 1.2 Schematic view of an MBE growth chamber.

    As discussed above, to limit the incorporation of contaminants which may affect material quality and morphology, the chambers are evacuated to base pressures of a few 10–11 Torr. For solid source MBE (SSMBE), this is generally achieved using a combination of ion, titanium and closed-cycle helium pumps, while for GSMBE and MOMBE (Section 1.3.4.1) the higher gas load (mainly H2) on the system is handled by trapped diffusion or turbomolecular pumps. Extensive use of cryopanels cooled with liquid nitrogen is made around the substrate holder and the cells, in order to remove from growth environment condensable contaminants such as CO, CO2 and H2O. A quadrupole mass analyser is used to monitor background species.

    The cryopanels also provide thermal insulations among different cells, that may have largely different temperatures (200–1100 °C) and that are situated at very close distances (smaller than few tens of cm). As it will be shown in Section 1.3.6, the accurate and independent control of beam fluxes generally requires the accurate control of cell temperatures.

    As shown in Figure 1.2, the substrate holder is mounted on a wafer manipulator which can rotate around an axis perpendicular to the axis of the cell flange, so that either the substrate or an ion gauge can be faced to the beams for growing or for measuring beam fluxes (Section 1.3.5), respectively. The substrate holder can be heated at temperatures sufficiently high to activate surface migration of the species adsorbed on the growing surface (Section 1.5.2). The substrate heating assembly generally consists of (i) a molybdenum block radiatively heated by a high-purity tantalum ribbon, which, in turn, is heated by Joule effect and (ii) a thermocouple positioned behind the block, close to the rotating substrate. The temperature of growing surface can be directly read by an optical pyrometer through a heated viewport. Absolute calibrations of substrate temperatures can be done by either observing changes in surface reconstructions (Section 1.4.1) or melting of suitable eutectic alloys deposited on sacrificial substrates.

    Since beams must converge towards the substrate, the cell axes are also convergent, as shown in Figure 1.2; therefore, in order to improve the homogeneity of beam fluxes on the substrate, the substrate holder is kept in continuous azimuthal rotation (Section 1.3.3.1) [22]. In production machines, where growth proceeds simultaneously on a number (for instance, up to 7) of substrates (with diameters of up to 6 inches, in case of III–V compounds), the substrates are mounted on platens, which rotate around the axis of the cell flange.

    In front of each cell a mechanical shutter, made of a high-purity material such as refractory metal or pyrolitic boron nitride, is generally provided; the shutter is used to switch on and off the beams in times in the order of tenths of seconds. As discussed above, the availability of shutters allows sudden changes of composition of the nourishing phase and is instrumental in the achievement of abrupt composition and doping interfaces.

    A load-lock system is used to transfer wafers into and out of the growth chamber without impairing the UHV conditions, which would otherwise require days or weeks to recondition the growth chamber.

    1.3.3 Sources of molecular beams

    The sources of molecular beams are among the most important components of an MBE system, since they should operate at high temperatures (up to 1200 °C or more) and generate molecular beams of high purity and time stability, with high uniformity over the substrate surfaces. A few types of sources have been developed to growth different materials by means of different variants of MBE; they are described in the present section and in Section 1.3.4, which deals with variants of MBE.

    1.3.3.1 Thermal-evaporation cells

    The most used cells generate beams by thermal evaporation or sublimation of suitable materials, which can be elements (such as Ga, As, Zn, Si and Be) or compounds (e.g., CdS and InP). The source materials are contained in crucibles generally made of high-purity pyrolitic boron nitride or graphite but also of quartz or stainless steel in cases where large volumes of evaporants (such as As and P) are used at low temperatures. The crucibles may have volumes of few cm³ up to several litres, depending on whether the evaporant is a dopant in research machines or a constituent atom to be supplied in relative excess (such as As or P in III–Vs, Section 1.5.1) in production machines. The crucibles are radiatively heated by Ta ribbons or wires, which, in turn, are heated by Joule effect. The temperature of crucibles is measured by W–Re thermocouples in contact with cell bottoms or sidewalls. Radiation shields made of Ta foils surround crucibles in order to provide thermal insulation of cells, required for minimising the thermal cross-talk between adjacent cells and for improving their thermal efficiency. Figure 1.3 schematically shows a typical thermal-evaporation cell.

    FIGURE 1.3 Schematic cross-sectional view of a conventional solid source MBE effusion cell.

    Cell temperatures are accurately set by closed-loop controllers which assure (i) stabilities in the order of tenths of °C and (ii) fast temperature variations, which – in any case – are affected by the thermal inertia of cells. In order to calculate the fluxes of beams impinging on substrates, an evaporation cell can be initially approximated by a Knudsen cell of species effusing in UHV from an orifice with area Ae is given by [7,14]

    where NA, kB, M and T are the Avogadro’s and Boltzmann’s constants, the molecular weight of the evaporating species and its absolute temperature, while p(T) is the equilibrium pressure of the evaporant at a temperature T.

    that reaches a point P on a substrate (i) situated at a distance rP from the orifice and (ii) following a path at an angle (θ + φ) with the substrate axis (φ and θ being the angles between cell and substrate axes and between cell axis and path, respectively (Figure 1.4)), is given by [14]:

    Since Knudsen cells have orifices too small to give fluxes resulting in reasonable growth rates (in the order of 1 μm/h or ML/s), open-ended crucibles with relatively large apertures are generally used; the fluxes from such crucibles cannot be predicted by the above equations, based on the equilibrium between solid (or liquid) and vapour; on the contrary, they largely depend on geometrical details of crucibles, such as shapes (e.g., whether conical or cylindrical), height-to-diameter ratios and relative position of crucibles and substrates [14]. Therefore, real fluxes are generally deduced from measurements of (i) beam equivalent pressures (BEPs) on beams, by using ion gauges (Section 1.3.5) or (ii) growth rates, obtained by the analysis of RHEED intensity oscillations (Section 1.4.1).

    FIGURE 1.4 Mutual positions of effusion cells and substrates in an MBE growth chamber. The dashed and dotted lines represent the axes of cells and substrates, respectively. Not to scale (For colour version of this figure, the reader is referred to the online version of this book).

    The above considerations on the angular dependence of effusion patterns account for the reasons why substrates and platens holding substrates are kept in continuous azimuthal rotation (Section 1.3.2) [22]. However, by stopping substrate rotation, the angular distribution of beam fluxes along substrate diameters can be used to have a reproducible and continuous distribution of (i) amounts of deposited material (coverages), (ii) alloy compositions and (iii) doping levels. For instance, this possibility allowed the accurate study of the critical coverage at which the 2D–3D transition (Section 1.5.3.1) occurs in the growth of self-assembled quantum dots [25].

    As shown before, the dependence of fluxes on cell temperatures is mainly determined by the temperature dependence of equilibrium pressures of evaporants, which can be described with good accuracy by Arrhenius laws with constants that depend on materials; these constants can be either found in the literature (e.g., Honig and Kramer [26]) or experimentally deduced by calibration runs.

    Other evaporation sources may simply consist of wires or ribbons of materials to evaporate (e.g., C or Ti) which are directly heated by Joule effect.

    1.3.3.2 Valved cracker cells

    In order to effectively modulate evaporant fluxes generated by crucibles with large capacity and, then, high thermal inertia, valved cells have been developed, where fluxes pass through properly designed needle valves, made of refractory metals, which can be actuated by means of computer-controlled stepping motors. In such a way, valved cells produce molecular beams with really fast time response and with very accurate beam intensity. A typical cell for species such as As and S is shown in Figure 1.5; it consists of (i) a heated crucible, (ii) a needle valve with an actuator and (iii) a conduction tube protruding through the UHV flange into the growth chamber. Generally, such As cells have also cracking stages where the evaporated As tetramers are partially converted into dimers [27]. Cracking is achieved by interactions of tetramers with surfaces at several hundreds of °C; the temperature of the cracking stage determines the relative amounts of As4 and As2 supplied to the growing surface. These cells are termed as valved cracker cells. The use of group-V dimers, instead of tetramers, was considered advantageous, thanks to the supposedly more favourable surface kinetics of As2 as compared to that of As4 (Section 1.5.1), that more recently was questioned by the results of Tok et al. [28].

    FIGURE 1.5 Schematic cross-sectional view of an As valved cracker cell (For colour version of this figure, the reader is referred to the online version of this book).

    To overcome the disadvantages of using highly toxic gaseous phosphine (PH3) as a source of P for the GSMBE of III-phosphides (Section 1.3.4.1), the concept of valved cracker cells has been extended [29]. P has two allotropic forms, the white and the red ones; the former is suitable for evaporation, since it provides stable and reproducible fluxes, but it is volatile and flammable in air; on the other hand, the latter has a lower vapour pressure and, then, requires relatively higher cell temperatures, but can be more safely handled during charge loading. Baillargeon et al. [10] proposed a P valved cracker cell to in-situ convert the red- to the white-P to be evaporated. The cell consists of (i) a crucible containing red-P, which can be easily loaded in air, (ii) a white-P cooled reservoir, (iii) a needle valve with an actuator and (iv) a cracking stage (Figure 1.6). During the evaporation of red-P, the P4 vapours are condensed as white-P in the cooled reservoir; when a sufficient quantity of P4 is converted, only the white-P is heated to obtain a stable P4 flux; then, tetramers pass through the needle valve and, finally, P4 may be cracked in the cracker stage and P2 species may effuse towards the growing surface. The availability of reliable and safe P solid sources has greatly facilitated the MBE growth of the InGaAsP alloys for optoelectronics and photonics in the 1.3-μm and 1.55-μm windows of lightwave communications.

    FIGURE 1.6 Schematic cross-sectional view of a P valved cracker cell (For colour version of this figure, the reader is referred to the online version of this book).

    1.3.4 Variants of the MBE process

    A number of variants of solid source MBE have been developed since its original proposal to fully exploit the MBE features in the growth of different class of materials and types of epitaxial structures. In this section we will briefly review (i) gas source MBE (GSMBE), (ii) metalorganic MBE (MOMBE), (iii) migration-enhanced epitaxy (MEE) (also termed as atomic layer MBE (ALMBE)) and (iv) droplet epitaxy.

    1.3.4.1 Gas source MBE, metalorganic MBE and chemical beam epitaxy

    GSMBE was proposed by Panish [30] in order to face two problems in the preparation of group-III phosphides such as InGaAsP. The problems are both related to the relatively high vapour pressure of P, which result in (i) frequent openings of growth chambers to the atmosphere to recharge P cells and (ii) the use of high-capacity P crucibles, which have large thermal inertia. The problems were addressed also by the development of group-V valved cells (Section 1.3.3.2) for SSMBE few years after the proposal of GSMBE.

    In the case of MBE of III–V compounds, As, P and Sb are admitted to growth chambers as gaseous hydrides (AsH3, PH3 and SbH3), the flows of which are set by gas-handling systems and electronically driven controllers. The hydrides are injected into the growth chamber and, then, are cracked in cracker cells by interaction with surfaces at several hundreds of °C; cracking results in the formation of tetramers and dimers and, as a by-product, of hydrogen. A schematic of a hydride cracker cell is given in Figure 1.7. An additional advantage of GSMBE is that the group-V beam fluxes and, then, the composition of III–V–V’ solid solutions can be controlled more precisely than in SSMBE, since in GSMBE group-V fluxes depend linearly on hydride flows, while in SSMBE they are related to group-V cell temperatures by Arrhenius laws.

    FIGURE 1.7 Schematic cross-sectional view of a hydride cracker cell.

    In MOMBE [31], termed also as chemical beam epitaxy (CBE) [32,33], the nourishing species are supplied as metalorganics; a typical example is the preparation of III–Vs by using metalorganic compounds of group-III species, such as triethylgallium (TEGa), trimethylindium (TMIn) and trimethylaluminium (TMAl). Group-III metalorganics are cracked directly on growing surfaces; group-III metalorganics, which are generally liquids contained in bubblers, saturate hydrogen flows which feed the growth chamber; therefore, the cell is simply a gas injector consisting of tubes with capillary leaks. Metalorganic reagents, such as triethylphosphine (TEP) and trimethylarsine (TMAs), can be used as sources of group-V elements [33,34], with additional advantages and disadvantages on hydrides, which are the lower toxicity and purity, respectively. Group-V alkyls, transported into the growth chamber through calibrated leaks by hydrogen flows, are thermally cracked by hot Ta or Mo surfaces at 950–1200 °C in cracker cells; the resulting group-V species, which effuse towards the growing surface, are prevalently dimers.

    GSMBE and MOMBE machines generally differ from those for SSMBE only in the cells; in the case of machines for III–V compounds, the conventional group-III sources are substituted by gas injectors for group-III metalorganic compounds (MOMBE), while group-V species can be supplied by means of either hydride crackers (GSMBE and MOMBE) or group-V metalorganic crackers (MOMBE).

    The piping of gases in GSMBE and MOMBE machines is accurately designed in order to minimise dead volumes and boundary layers that would (i) slow down the time response of the system to rapid variations of composition of the gaseous phase and, hence, (ii) broaden the interfaces between different layers.

    The operating pressures of GSMBE and MOMBE are in the 10–6–10–4 Torr range, due to (i) the relatively high H2 backpressure related to decomposition of hydrides and (ii) the use of H2 to transport metalorganic species; it may be of interest to recall (Section 1.2) that in this pressure range (i) the regime is still molecular, (ii) the atoms or molecules in the beams still have almost collision-free paths towards substrates and, then, (iii) shutters can be used to abruptly change the composition of nourishing species impinging on the substrate, so that (iv) the interfaces can be sharp.

    1.3.4.2 Nitride MBE and reactive MBE

    In the MBE growth of a number of materials, a few species may be too stable to participate in surface reactions resulting in deposition. A typical example is given by group-III nitrides, which have recently gained a widespread interest for optoelectronic and photonic applications, and also for wide-gap microelectronics; group-III nitrides have fundamental bandgaps spanning a wide region (corresponding to ∼0.2–1.9 μm at room temperature) [35], which is wider than that of InGaAsP. Preparation and properties of III-nitrides are reviewed in Henini [36], Morkoç [37] and, more recently, by Guina [38], Bensaoula [39] and Lischka and As [40].

    In the MBE of nitrides, N2 could be used as the source of N; however, N2 is a particularly stable molecule (with a molecular cohesive energy of 9.8 eV) which cannot be thermally cracked. The more reactive N atoms can be obtained by fragmenting N2 molecules directly in the growth environment by means of plasmas, produced either by electron cyclotron resonance (ECR) sources or by radio-frequency (RF) ones. Then, the deposition of GaN takes place by the interaction on growing surfaces of atomic N and of Ga deposited by a molecular beam. The MBE variant where a plasma is used to prepare the species for growth is called also as plasma-assisted MBE (PAMBE) and is used for a number of materials.

    The N plasma is generally contained in a UHV nipple separated from the main chamber by an all-metal gate valve [41]; this solution lets (i) to stabilise the plasma well before (tens of minutes) N species are admitted into growth chambers and (ii) to suppress unintentional incorporation of N in layers that should be N-free. Indeed, it has been shown that the use of conventional shutters alone to intercept N beams results in incorporation of significant amounts of N [43].

    RF-plasma cells are fed by N2 flows in the order of 1 sccm s−1; the N2 molecules are cracked in atomic N and in N radicals and ions; a disadvantage of this approach is the production of ionised species that are accelerated by electric fields against growing layers, where structural defects are thereby created. A number of studies have been carried out to minimise the formation of ionised N species [44]; plates can be used to electrostatically deflect the ionised species. With recent cells, the formation of defects has been significantly reduced and AlGaN/GaN heterostructures have been grown with high crystalline quality [45].

    A further variant of MBE is the reactive MBE, which can be exemplified by the MBE growth process of group-V nitrides by using thermally cracked ammonia. Controlled amounts of NH3 are injected in the UHV environment; then, ammonia is thermally cracked directly on the heated substrate. The main disadvantages of this method is that when low growth temperatures are used to minimise the phase separation of nitride solid solutions, NH3 molecules are cracked with relatively low efficiency, so that high ammonia fluxes are required; this may result in safety hazards and high H loads to the pumping systems [46].

    Recently, warm ammonia has been proposed for the growth of high structural quality GaN [47]. NH3 is heated to 1100 °C by a bare tungsten filament situated in front of the gas injector. In such a way, efficient thermal cracking of NH3 is obtained, while avoiding multiple collisions of molecules, which would result in the formation of stable N2 molecules. The results show that the crystalline quality of GaN is significantly improved [47]; the improvement has been explained by invoking the enhancement of the two-dimensional growth (Section 1.5.2) due to energetic nitrogen species, such as NH∗2, generated by cracking ammonia molecules.

    1.3.4.3 Group-IV MBE

    There is a wide interest for MBE growth of epitaxial structures based on Si and SiGe. The deposition of these materials takes full advantage of MBE features, such as relatively low growth temperatures and precisely controlled doping and composition profiles along growth direction. By using such profiles, carriers can be quantum confined and, then, show up engineered properties. Advantages can be also taken of strain between SiGe layers with a different composition, which can be used to modify electronic band alignments, as it was shown in the pioneering works of Abstreiter et al. [48] and People and Bean [49]. Devices such as high-frequency HEMTs can be fabricated by using the engineered strain in the SiGe system; in some instances these devices may compete with the III–V-based counterparts.

    Interesting reviews on Si-MBE can be found in Ota [50], Shiraki [51], Bean [52], Bean [53] and, more recently, in Springholz [54]. In principle, a Si MBE machine differs only in few features from a III–V one; the differences are in the (i) Si sources, (ii) substrate holder that can be heated to much higher temperatures and (iii) availability of ionised doping sources. Other significant differences are the higher throughputs of growth runs and the larger sizes of wafers.

    Since Si has a relatively low vapour pressure, reasonable fluxes could be obtained only at source temperatures (∼1600 °C) which can hardly be achieved by thermally heated effusion cells. Therefore, Si is evaporated by targets heated by electron beams; the main disadvantage of this approach is that with not optimised electron guns, stray electrons may hit components surrounding targets, thus giving rise to outgassing and contamination.

    While some dopants can be thermally evaporated by conventional cells (Al and Ga for the p-type and Sb for the n-type doping, respectively), in order to have high (>10¹⁸ cm−3) and very precisely controlled doping levels, doping can be performed by means of in-situ low-energy (400–800 eV) ion implantation. Indeed, doping levels in the range of 10¹⁶–10²⁰ cm−3 are accurately obtained by using Sb ion sources and measuring ion current densities. This parameter can be abruptly changed, thus giving abrupt doping profiles [11]. Similar results were obtained by using As [9]. The radiation damage related to the energetic ionised species impinging on the substrate is recovered during growth that is carried out at temperatures in the 600–1100 °C range.

    A disadvantage of using Si melts obtained by electron beam heating as evaporation sources is the formation of micron-sized morphological defects on the growing surface, which are due to Si spitting from the source [19]. Such defects are detrimental to large-scale integration on Si layers and structures prepared by SSMBE. Therefore, a GSMBE approach has been proposed [55–57], where group-IV species are admitted into growth chambers as hydrides, such as Si2H6 and GeH4, by means of gas injectors. The species are then cracked directly on heated substrates as it occurs for group-III metalorganics. The composition of SiGe solid solutions is controlled by the flows of Si and Ge hydrides. An advantage of this approach is that p- and n-type doping can be carried out by using gaseous B2H6 and PH3, respectively.

    A very important feature of the Si and SiGe GSMBE is that the technique is well suited for selective epitaxy at proper growth temperatures on Si substrates patterned by SiO2 masks [58,59].

    1.3.4.4 Migration enhanced epitaxy (MEE) and atomic layer MBE (ALMBE)

    Variants of MBE can be characterised not only by the chemical nature of beam species or by the way the beams are produced but also by the way the beams are supplied to growing surfaces. While in other MBE approaches the molecular beams are simultaneously supplied to the surface, in migration enhanced epitaxy (MEE) the beams of constituent species alternately feed the surface [60]. In the archetypal example of GaAs, Ga and As species impinge on the surface in ML or sub-ML amounts during separated cation and anion subcycles. To emphasise its MBE derivation, MEE is termed also as atomic layer MBE (ALMBE) [61]. A thorough review of the features of MEE is given in a chapter of the present book [62].

    The main feature of MEE can be understood by recalling that in the MBE of GaAs, the formation of GaAs on the surface is due to the co-deposition of Ga and As4 and to their surface interactions with the substrate (Section 1.5.2); when the substrate temperature is not high enough to activate an efficient migration of cations towards the edges of existing islands, a relatively large number of small islands and a rough growth front are formed (Section 1.5.2). The islands are fairly stable and unless Ga–As bonds are thermally broken, Ga cannot migrate on the surface. In MEE (ALMBE), instead, Ga atoms are deposited in the absence of As molecules impinging on the substrate, so that no GaAs island is formed during the cation subcycle and surface migration of Ga atoms is effectively enhanced. As it will be shown in Section 1.5.2.1, efficient cation migration is instrumental in the formation of surfaces smooth on the atomic scale. In fact, under MEE conditions Ga atoms can more effectively migrate on the surface and find stable sites at edges of existing islands; therefore, the formation of new islands is minimised and, then, the growth front is smoothed down even at relatively low temperatures. Experimental results show that high-quality GaAs can be grown by MEE at 300 °C (instead of at ∼600 °C, a value typical for MBE). It is worth noting that reduced growth temperatures also inhibit interdiffusion of species across heterointerfaces that, therefore, may be not only smooth but also sharp on the atomic scale.

    In the growth of GaAs, the amount of Ga atoms supplied in a cation subcycle is given by the time length of the subcycle and by the flux of atoms sticking on the surface; as shown in Section 1.5.2.2, the Ga flux can be determined by the analysis of RHEED intensity oscillations and the As flux is generally adjusted so that clear, stable As-rich surface reconstructions (Section 1.5.2.2) show up at the end of anion subcycles.

    1.3.4.5 Droplet epitaxy

    In droplet epitaxy of III–Vs the growing solid phase consists of nanoislands formed by the interaction of group-V beams with nm-sized droplets of group-III species predeposited on a substrate [63]; the formation of islands takes place according to the vapour–liquid–solid mechanism that is used also for the MBE growth of nanowires (Section 1.5.4); the deposition of the solid-phase stems from the supersaturation of the liquid one by the species carried by beams in the vapour phase. The main feature of droplet epitaxy is that the formation of islands does not require elastic strain between islands and substrate, as it is necessary in the Stranski–Krastanow growth mechanism of self-assembled islands (Section 1.5.3); therefore, droplet epitaxy allows the preparation of quantum dot structures using lattice-matched materials, such as AlGaAs/GaAs. Droplet epitaxy is thoroughly discussed in a chapter of the present book [64].

    1.3.5 Measurements of molecular beam fluxes

    The accurate measurement of beam fluxes is one of the prerequisites for the precise control of composition and doping profiles, which is an important feature of MBE, as shown in Section 1.3.6.

    The most widely used method for measuring beam fluxes is based on the use of an ion gauge, which can be placed in the wafer position (of the species i and j can be related to the ratio between the BEPs pj and pj by using

    , pk, Tk and Mk are flux, BEP, absolute temperature and molecular weight of the species k (k = i, j) and ηk is the ionisation coefficient relative to nitrogen, given by

    where Zk is the atomic number of the species k [66]. The knowledge of flux ratios is of great importance since, for instance, in the growth of III–V compounds the ratio of group-III to group-V fluxes determines the reconstruction and the stoichiometry of growing surface (Section 1.4.1) and, then, a number of physical properties of layers, such as incorporation of amphoteric dopants (e.g., Ge); on the other hand, the ratio between fluxes of III and III’ and, in some extent, of V and V’ atoms determines the composition of III–III’–V and III–V–V’ alloys, respectively (Section 1.3.6).

    When MBE growth proceeds layer by layer under the optimised conditions discussed in Section 1.5.2, the preferred method to measure group-III beam fluxes is the analysis of RHEED intensity oscillations. As it will be shown in Section 1.4.1, these oscillations take place owing to the periodic change of surface morphology that occurs whenever the deposition of a monolayer is completed. Therefore, by measuring the oscillation period during growth, the growth rate can be deduced in-situ and in real timeof the constituent atoms k can be readily obtained from G , where Nk is the number of atoms k per unit volume; as an example, in the homoepitaxial growth of GaAs, a growth rate G = 1 ML/s (0.283 nm/s) corresponds to a flux of 6.24 × 10¹⁴ cm−2 s−1 Ga atoms. It is important to note that Nk depends not only on the composition and the crystalline structure of layers, but also on their strain conditions, as it has been thoroughly discussed in Bocchi et al. [67], where the composition of GaAlSb grown on GaSb was determined by means of RHEED oscillations and compared to the results obtained by Rutherford backscattering spectrometry (RBS) and high-resolution X-ray diffraction (HRXRD).

    Other methods to measure beam fluxes are based on the use of quartz microbalances [65] and of atomic absorption and atomic emission spectroscopies of species in the beams (Kometani and Wiegmann [68] and Davies and Williams [65], respectively).

    1.3.6 Control of composition and doping profiles along growth direction

    Indubitably, one of the most interesting features of MBE and related variants is the capability of accurately controlling the composition of the nourishing phase and of modulating it with an unprecedented precision in times comparable to those required for the growth of ML thick layers. This capability, along with the peculiar growth kinetics discussed in Section 1.5, assures that composition and doping profiles along the growth axis can be effectively modulated; this feature is of interest for the growth not only of relatively thick graded layers (>0.1 μm) for optoelectronic (Alferov [1]; Nobel lecture) and microelectronic (Kroemer [2]; Nobel lecture) applications, but also of structures where carriers are confined carriers in regions thinner than their de Broglie wavelength. The resulting quantum confinement of carriers gives the nanostructures new and tailorable properties (Esaki [3]; Nobel lecture), as it will be recalled below.

    Composition and doping can be changed (i) abruptly, by means of beam shutters, the use of which is allowed by the molecular regime of growth environment, or (ii) continuously, by using different approaches to modulate the beam fluxes.

    As for continuous composition grading, let us consider first the case of III–III’–V alloys (e.g., AlxGa1−xAs or InxGa1−xAs); the mole fraction xk of the group-III atom k in the cation sublattice is given by

    where Nk is the number per unit volume of the atoms k in the solid. Then, xk is given by

    is the flux of atoms k in the beam (atom/(cm² s)) and σk is the incorporation coefficient of the species k. Therefore, since at a constant growth temperature σk is also constant, it follows that during growth xk depends only on the fluxes of the species i and j. For instance, in case of GaAs and InAs, σGa = 1 and σIn = 1 at temperatures Tc lower than 650° [69] and 550 °C [70]; respectively. It is worth noting that the values of Tc are higher than that of the usual growth temperatures of III–V compounds; therefore, under the generally used growth conditions, σk = 1.

    As mentioned in Section are determined by the temperature T of sources, according to

    where ΔEk can be measured under the same conditions used for growth. On the other hand, in the case of gaseous sources, such as hydrides (GSMBE) or carrier gases saturated with liquid metalorganic compounds (MOMBE), the beam fluxes can be adjusted by means of accurate controllers or by the temperature of metalorganic bubblers, respectively (Section 1.3.4.1).

    As for III–V–V’ alloys, the composition control is more critical since it depends not only on beam fluxes but also on incorporation coefficients of group-V elements, which are composition dependent, as observed by Foxon et al. [71] for GaInAsP and InAsP alloys grown on GaAs. Similarly, Bosacchi et al. [72] studied the composition of fully relaxed GaSbAs deposited on GaAs and grown with different parameters; the results show that the GaSbAs compositions critically depend on the fluxes of the three constituents Ga, Sb and As and not only on the ratios between anion fluxes. The compositions were deduced by combined photoluminescence (PL) and Rutherford backscattering (RBS) measurements, while the fluxes were obtained by the analysis of RHEED intensity oscillations under Ga-, Sb- and As-rich growth conditions.

    In the case of SSMBE, the modulation of composition profiles can be hampered by the thermal inertia of effusion cells. An interesting approach, based on the concept of digital or pseudoternary alloy, was described by Capasso [73], Miller et al. [74] and Capasso [75]. Such an alloy consists of a sequence of layer pairs with thicknesses di and composition xi (i = 1, 2) (IIIx–III’1−x–V or III–Vx–V’1−x); while the compositions of layers are kept constant, their thicknesses are graded during growth so that the mean composition of each layer pair

    follows the intended composition profile of the pseudoternary alloy. The sum of layer thicknesses in the pairs is kept constant during growth and is in the order of few nm; therefore, it is sufficiently small to allow tunnelling of carriers through the layers, so that pseudoternary alloys may mimic conventional solid solutions. In Sundaram et al. [76] it has been illustrated that the digital alloy technique may grow alloys with almost arbitrarily varying composition profiles.

    MEE (ALMBE) (Section 1.3.4.4) offers a further advantage for the growth of continuously graded alloys [77]; let us consider first the case of III–III’–V alloys; if cations i and j are supplied in two subcycles, both within the cation one, with time lengths τi and τj, it is easy to show that the mole fraction of the species k is given by

    Therefore, by keeping beam fluxes and growth temperatures constant, the composition can be continuously graded by grading the time lengths of the subcycles of cations i and j according to computer-controlled shutter sequences.

    The composition can be controlled also for ternary III–V–V’ solid solutions, provided the group-V species are supplied to the surface in separate cycles during anion subcycles; in this case, the anions are separately incorporated, with efficiencies that are independent of composition. This situation is different from that where anions are simultaneously supplied and, hence, may compete for incorporation, in such a way that anions with more stable anion–cation bonds are more efficiently incorporated, independently of anion fluxes [78].

    Since the supply times of both cations and anions in each ALMBE cycle can be set independently of the values in previous and/or subsequent cycle, it follows that alloy compositions can be modified during the growth almost at will, either in an abrupt or in a continuous way, without changing the molecular beam fluxes. By using computer-controlled shutters, Madella et al. [77] showed that (i) as for continuous grading, a composition resolution of ∼1% can be achieved over the full composition range (0 ≤ x ≤ 1), while (ii) for abrupt changes, changes in nominal composition from any x to any x′ (0 ≤ x, x′ ≤ 1) can be obtained from one monolayer to the next one. The real composition of each monolayer, however, is determined also by surface segregation [79], by exchange reactions between components contained in adjacent layers and by the extent of the growth-front roughness (Section 1.5.2.1).

    1.4 Diagnostic techniques available in MBE systems

    The relatively small chemical aggressiveness of growth environment is one the most interesting features of MBE; it makes possible to install in the growth or in interconnected chambers a number of diagnostic techniques, not generally available to other epitaxial technologies. This accounts for the deep understanding of MBE growth processes (Section 1.5) and, consequently, for the high degree of confidence in conceiving, designing and growing innovative, advanced epitaxial structures, also on the nanoscale.

    Since the beginning of the MBE evolution, the techniques used to study both epitaxial layers and MBE processes were high-energy electron diffraction in reflection configuration (RHEED), Auger electron spectroscopy (AES), secondary ion mass spectroscopy (SIMS), X-ray photoemission spectroscopy (XPS) and ultraviolet photoemission spectroscopy (UPS) [7,21]. Initially, the techniques were installed in growth chambers, but then, to avoid cross-contamination of analytic instruments and growth environment, they have been located in dedicated modules; RHEED is a notable exception since it is used in-situ and in real time (Section 1.5.2.2). More recently, it has been demonstrated that scanning tunnelling microscopy (STM) is a technique of great relevance for MBE materials and processes, since it can be used to study growth surfaces on the atomic scale.

    1.4.1 RHEED and surface reconstructions

    RHEED is one of the most interesting diagnostic techniques available in MBE systems; the technique is based on an electron gun that produces a collimated beam of electrons with energies of 10–50 keV and a fluorescent screen where electrons diffracted from growing surfaces form the diffraction patterns (Figure 1.2); such patterns can be visually inspected and/or analysed by image processing techniques. The angle between the beam and the diffracting surface is in the order of 0.5–2.0°; therefore, diffraction studies can be carried out during growth while molecular beams impinge on the surface [80,81].

    The condition for constructive interference of elastically diffracted electrons is given by the Laue law, which states that the wavevectors of incident and diffracted beams must differ by a reciprocal lattice vector. Under grazing incidence (0.5–2.0°) and with atomically smooth surfaces, electrons may emerge from the crystal only when the diffracting planes are at depths of few MLs; in this case, the electron beam is diffracted by a two-dimensional lattice, whose reciprocal lattice consists of parallel lines. On the other hand, in case of atomically rough surfaces, electrons emerge even if they were diffracted by deeper planes; under these conditions, electrons sense a three-dimensional lattice, the reciprocal lattice of which is represented by points.

    For any incident wavevector, the Laue condition implies that the wavevectors of diffracted beams are determined by the intersection of reciprocal lattice lines or points with the Ewald sphere, that has a radius equal to the electron momentum [82,83] (Figure 1.8). Two important features of electron (as opposed to X-ray) diffraction from real crystals are (i) the radius of the Ewald sphere is much longer than the distance between reciprocal lattice points or lines and (ii) the surface of the Ewald sphere has a finite thickness, due to the energy spread (and, hence, the momentum spread) of electrons; moreover, lattice defects and vibrations broaden the reciprocal lattice points and lines that, therefore, can be represented by spots and rods, respectively. As a consequence, depending on whether the surface is atomically rough or smooth, the diffraction pattern consists of spots or streaks, respectively (Figure 1.8); therefore, the occurrence of streaks, instead of spots, is the clear proof that diffraction takes place from few shallow lattice planes and, then, that the diffracting surface is atomically smooth.

    FIGURE 1.8 Schematic representation of the origin of RHEED diffraction patterns from atomically rough (upper panel) and smooth (lower panel) surfaces characterised by reciprocal lattices given by spots and rods, respectively, a few of which are sketched in the figure; the intersections of the Ewald sphere with the features of the reciprocal lattice define diffracted spots and streaks visualised on a fluorescent screen (For colour version of this figure, the reader is referred to the online version of this book).

    After thermal desorption of native oxides to clean a GaAs (100) surface the spotty diffraction patterns show that the surface is atomically rough. After few seconds from the beginning of MBE growth, dots are transformed in the integer order streaks, marked by arrows (Figure 1.9) and faint fractional order streaks appear between the integer order ones.

    FIGURE 1.9 RHEED patterns observed along the (011) (left panel) and (0–11) (right panel) azimuths during the MBE growth of epitaxial GaAs on a (100) GaAs substrate; the surface reconstruction is the As-stabilised (2 × 4) one. The arrows mark the position of diffraction features from bulk layers (integer order streaks).

    Source: Adapted from Frigeri et al. [179] with permission. Copyright 2011 by Elsevier.

    The streaky diffraction pattern depends on (i) the azimuth of the impinging electron beam with respect to the crystallographic directions of diffracting layer and (ii) surface reconstructions which, in turn, depend on growth conditions, such as surface temperature and As/Ga ratio in the beams [18,81,84,85]. Figure 1.9 shows typical examples of diffraction patterns from a clean GaAs (100) surface grown under As-rich conditions (see below), observed under [011] and [0–11] azimuths; the patterns are characterised by one and three fractional order streaks in between the integer order ones, respectively; these features show that (i) the surface smooths out during the early stages of growth and, (ii) along the two perpendicular directions, the GaAs (100) surface unit cell shows a twofold (2×) and a fourfold (4×) periodicity, respectively, as compared to that of the unreconstructed unit cell.

    Surface reconstructions with the 2× and 4× periodicities along the [−1–10] and [1–10] azimuths, respectively, are termed as (2 × 4), while those where the same periodicities show up when the azimuths are exchanged are referred to as (4 × 2) [18,81].

    As shown in (Figure 1.10), (2 × 4) reconstructions occur when growth is carried out under the so-called As-rich (or As-stabilised) conditions, which are those that favour a relative surface enrichment in As and are determined by low growth temperatures and/or large As2/Ga ratios in the beams [86]. On the other hand, (4 × 2) reconstructions are typical of Ga-rich (or Ga-stabilised) conditions, determined by high growth temperatures and/or low As2/Ga ratios in the beams (Figure 1.10). Between the main (2 × 4) and (4 × 2) structures, a number of transition structures appear in very narrow ranges of growth conditions (Figure 1.10) [86].

    FIGURE 1.10 Existence regimes of different surface reconstructions of (100) GaAs surfaces as functions of As 2 /Ga flux ratio and substrate temperature T . The close and open symbols indicate respectively that the datapoints have been obtained in runs at increasing ( T ↑) and decreasing ( T ↓) temperatures (For colour version of this figure, the reader is referred to the online version of this book).

    Source: Adapted from Cho [86] with permission. Copyright 1971 by American Institute of Physics.

    Arthur [87] first proposed that surface reconstructions could be related to surface stoichiometry and, in particular, to As gain or loss. Then, other authors [88,89] studied by AES the variations of surface stoichiometry associated with GaAs (100) reconstructions and confirmed the As- or Ga-rich nature of surfaces

    Enjoying the preview?
    Page 1 of 1