Molecular Beam Epitaxy Material Research Articles

Full bandgap defect state characterization of β-Ga2O3 grown by metal organic chemical vapor deposition

The results of a detailed investigation of electrically active defects in metal-organic chemical vapor deposition (MOCVD)-grown β-Ga2O3 (010) epitaxial layers are described. A combination of deep level optical spectroscopy (DLOS), deep level transient (thermal) spectroscopy (DLTS), and admittance spectroscopy (AS) is used to quantitatively map the energy levels, cross sections, and concentrations of traps across the entire ∼4.8 eV bandgap. States are observed at EC-0.12 eV by AS; at EC-0.4 eV by DLTS; and at EC-1.2 eV, EC-2.0 eV, and EC-4.4 eV by DLOS. While each of these states have been reported for β-Ga2O3 grown by molecular-beam epitaxy (MBE) and edge-defined film fed grown (EFG), with the exception of the EC-0.4 eV trap, there is both a significantly different distribution in the concentration of these states and an overall ∼10× reduction in the total trap concentration. This reduction is consistent with the high mobility and low background compensating acceptor concentrations that have been reported for MOCVD-grown (010) β-Ga2O3. Here, it is observed that the EC-0.12 eV state dominates the overall trap concentration, in marked contrast with prior studies of EFG and MBE material where the state at EC-4.4 eV has dominated the trap spectrum. This sheds light on possible physical sources for this ubiquitous DLOS feature in β-Ga2O3. The substantial reduction in trap concentration for MOCVD material implies great promise for future high performance MOCVD-grown β-Ga2O3 devices.

Photo-induced changes of the surface band bending in GaN: Influence of growth technique, doping and polarity

In this work, we use conductance and contact potential difference photo-transient data to study the influence of the growth technique, doping, and crystal polarity on the kinetics of photo-generated charges in GaN. We found that the processes, and corresponding time scales, involved in the decay of charge carriers generated at and close to the GaN surface via photo-excitation are notably independent of the growth technique, doping (n- and p-types), and also crystal polarity. Hence, the transfer of photo-generated charges from band states back to surface states proceeds always by hopping via shallow defect states in the space-charge region (SCR) close to the surface. Concerning the charge carrier photo-generation kinetics, we observe considerable differences between samples grown with different techniques. While for GaN grown by metal-organic chemical vapor deposition, the accumulation of photo-conduction electrons results mainly from a combined trapping-hopping process (slow), where photo-generated electrons hop via shallow defect states to the conduction band (CB), in hydride vapor phase epitaxy and molecular beam epitaxy materials, a faster direct process involving electron transfer via CB states is also present. The time scales of both processes are quite insensitive to the doping level and crystal polarity. However, these processes become irrelevant for very high doping levels (both n- and p-types), where the width of the SCR is much smaller than the photon penetration depth, and therefore, most charge carriers are generated outside the SCR.

Development of MBE HgCdTe for HDVIP® Focal-Plane Arrays

DRS Technologies is pursuing molecular beam epitaxy (MBE) HgCdTe as an alternative to its standard liquid-phase epitaxy (LPE) double-sided interdiffused (DSID) process for small-pitch, large-area, high-density vertically integrated photodiode (HDVIP®) focal-plane arrays (FPAs). Unlike DRS’s Te-rich LPE, which must be thinned from 60 μm to 5 μm, MBE material can be grown to the desired thickness. Additionally, the CdTe passivation layers needed for the HDVIP architecture may be grown in situ with MBE, greatly reducing the handling and processing time of DSID material. HDVIP FPAs were fabricated with MBE material that had been passivated with in situ MBE on one surface with CdTe deposited on the other. The MBE FPAs nearly matched the performance of LPE FPAs at 120 K but degraded at 160 K due to noise defects. Dark current measured at 100 K suggested a short lifetime. Additional MBE FPAs were fabricated with a full DSID process after annealing in Hg ambient at either 350°C, 400°C, 425°C or 450°C. The 350°C Hg anneal improved the lifetime of the material, and consequently reduced the number of noise defects at 160 K. Further investigation is warranted to understand how annealing MBE material in Hg ambient impacts HDVIP FPA performance at high temperatures.

MBE HgCdTe for HDVIP Devices: Horizontal Integration in the US HgCdTe FPA Industry

Molecular beam epitaxy (MBE) growth of HgCdTe offers the possibility of fabricating multilayer device structures with an almost unlimited choice of infrared sensor designs for focal-plane array (FPA) fabrication. HgCdTe offers two major advantages that explain its dominance in the infrared photon detector marketplace. The thermal generation rate per unit volume of the material is lower and the quantum efficiency for photon absorption in the infrared is higher in HgCdTe than in any competing material—it yields devices with quantum efficiencies as high as 0.99. Recently, EPIR Technologies and DRS Infrared Technologies agreed to collaborate and examine: (i) the feasibility of employing MBE HgCdTe in the fabrication of high-density vertically interconnected photodiodes (HDVIPs), which are usually fabricated with liquid-phase epitaxy material, and (ii) the potential benefits of horizontal integration, with EPIR supplying the MBE materials to DRS for device and array fabrication. The team designed and developed passivation–absorber–passivation structures that are heavily used by DRS. This paper provides an overview of the characteristics of HDVIP devices and arrays fabricated from MBE HgCdTe and the anticipated advantages of horizontal integration in the industry. Material growth, device fabrication, and test results are presented.

Deep levels in α-irradiated p-type MOCVD GaAs

Irradiation of p-GaAs grown by low-pressure metal-organic chemical-vapor deposition (LP-MOCVD) with α-particles from the 241Am source has been performed at room temperature. At least seven radiation-induced deep-level defects at energy positions E v+0.09 eV, E v+0.11 eV, E v+0.16 eV, E v+0.24 eV, E v+0.32 eV, E v+0.42 eV, and E v+0.77 eV were detected in the lower-half of the band gap, using deep-level transient spectroscopy (DLTS), while no radiation-induced defects were observed in the upper-half band gap (minority carrier injection spectrum). The DLTS signatures of the radiation-induced defects are compared with the previously reported defects introduced in α-irradiated p-GaAs grown by molecular beam epitaxy (MBE). It has been observed that while four of these deep levels are similar to those found in α-irradiated MBE grown p-GaAs, three levels at E v+0.32 eV, E v+0.42 eV, and E v+0.77 eV observed in our MOCVD material were not detected in MBE material. Data on the introduction rates and other characteristics of the radiation-induced defects are presented.

Fabrication of High power, High-Efficiency Linear Array Diode Lasers by Pulse Anodic Oxidation

InGaAlAs/AlGaAs/GaAs double-quantum-well (DQW) linear array diode lasers with asymmetric wide waveguide have been successfully fabricated by pulse anodic oxidation upon molecular beam epitaxy material growth. High-efficiency and high-power quasi-continuous-wave (QCW) output has been realized at 808 nm wavelength. The threshold current and slope efficiency of the prepared high-fill-factor QCW devices are 24 A and 1.25 A/W, respectively, and a maximum wall-plug efficiency of 51% has been achieved.

Thin film transmission matrix approach to fourier transform infrared analysis of HgCdTe multilayer heterostructures

The ability to achieve high-yield focal plane arrays from Hg1−xCdxTe molecular beam epitaxy material depends strongly on postgrowth wafer analysis. Nondestructive analysis that can determine layer thicknesses as well as alloy compositions is critical in providing run-to-run consistency. In this paper, we incorporate the use of a thin film transmission matrix model to analyze Fourier transform infrared (FTIR) transmission spectra. Our model uses a genetic algorithm along with a multidimensional, nonlinear minimization Nelder-Mead algorithm to determine the composition and thickness of each layer in the measured epitaxial structure. Once a solution has been found, the software is able to predict detector performance such as quantum efficiency and spectral response. We have verified our model by comparing detector spectral data to our predicted spectral data derived from the room-temperature FTIR transmission data. Furthermore, the model can be used to generate design curves for detectors with varying absorber thicknesses and/or different operating temperatures. The consequence of this are reduced cycle times and reduced design variations.

Silicon interstitial injection during dry oxidation of SiGe∕Si layers

The injection of Si self-interstitial atoms during dry oxidation at 815°C of very shallow SiGe layers grown on Si (001) by molecular-beam epitaxy (MBE) has been investigated. We first quantified the oxidation enhanced diffusion (OED) of two boron deltas buried into the Si underlying the oxidized SiGe layers. Then, by simulating the interstitial diffusion in the MBE material with a code developed on purpose, we estimated the interstitial supersaturation (S) at the SiGe∕Si interface. We found that S (a) is lower than that observed in pure Si, (b) is Ge-concentration dependent, and (c) has a very fast transient behavior. After such a short transient, the OED is completely suppressed, and the suppression lasts for long annealing times even after the complete oxidation of the SiGe layer. The above results have been related to the mechanism of oxidation of SiGe in which the Ge piles up at the SiO2∕SiGe interface by producing a thin and defect-free layer with a very high concentration of Ge.

Nearest-neighbor configuration in (GaIn)(NAs) probed by x-ray absorption spectroscopy.

Ga(1-x)In(x)N(y)As(1-y) is a promising material system for the fabrication of inexpensive "last-mile" optoelectronic components. However, details of its atomic arrangement and the relationship to observed optical properties is not fully known. Particularly, a blueshift of emission wavelength is observed after annealing. In this work, we use x-ray absorption fine structure to study the chemical environment around N atoms in the material before and after annealing. We find that as-grown molecular beam epitaxy material consists of a nearly random distribution of atoms, while postannealed material shows segregation of In toward N. Ab initio simulations show that this short-range ordering creates a more thermodynamically stable alloy and is responsible for blueshifting the emission.

Minority carrier diffusion and defects in InGaAsN grown by molecular beam epitaxy

To gain insight into the nitrogen-related defects of InGaAsN, nitrogen vibrational mode spectra, Hall mobilities, and minority carrier diffusion lengths are examined for InGaAsN (1.1 eV band gap) grown by molecular beam epitaxy (MBE). Annealing promotes the formation of In–N bonding, and lateral carrier transport is limited by large scale (≫mean free path) material inhomogeneities. Comparing solar cell quantum efficiencies with our earlier results for devices grown by metalorganic chemical vapor deposition (MOCVD), we find significant electron diffusion in the MBE material (reversed from the hole diffusion in MOCVD material), and minority carrier diffusion in InGaAsN cannot be explained by a “universal,” nitrogen-related defect.

The Role of Nitrogen-Induced Localization and Defects in InGaAsN (? 2% N): Comparison of InGaAsN Grown by Molecular Beam Epitaxy and Metal-Organic Chemical Vapor Deposition

AbstractNitrogen vibrational mode spectra, Hall mobilities, and minority carrier diffusion lengths are examined for InGaAsN (≈ 1.1 eV bandgap) grown by molecular beam epitaxy (MBE) and metal-organic chemical vapor deposition (MOCVD). Independent of growth technique, annealing promotes the formation of In-N bonding, and lateral carrier transport is limited by large scale (Ęmean free path ) material inhomogeneities. Comparing solar cell quantum efficiencies for devices grown by MBE and MOCVD, we find significant electron diffusion in the MBE material (reversed from the hole diffusion occurring in MOCVD material), and minority carrier diffusion in InGaAsN cannot be explained by a “universal”, nitrogen-related defect.

Generation of EL2 defects by a 6-MeV proton irradiation of semi-insulating GaAs

The defects introduced in a bulk semi-insulating (SI) GaAs by a 6-MeV proton irradiation at 300 K were investigated with the electron paramagnetic resonance (EPR) detected via the magnetic circular dichroism of the optical absorption. EL2 defects (EL2 0 and EL2 +) are introduced by proton irradiation in contrast to electron irradiation, the total introduction rate being 100±50 cm −1. The radiation-induced EL2 defects could not be completely bleached. They have properties similar to the EL2 defects in a low temperature grown molecular beam epitaxy material.

Characteristics and uniformity of group V implanted and annealed HgCdTe heterostructure

This work presents characterization of implanted and annealed double layer planar heterostructure HgCdTe for p-on-n photovoltaic devices. Our observation is that compositional redistribution in the structure during implantation/ annealing process differs from that expected from classical composition gradient driven interdiffusion and impacts the placement of the electrical junction with respect to the metallurgical heterointerface, which in turn affects quantum efficiency and RoA. The observed anomalous interdiffusion results in much wider cap layers with reduced composition difference between base and cap layer composition. The compositional redistribution can, however, be controlled by varying the material structure parameters and the implant/anneal conditions. Examples are presented for dose and implanted species variation. A model is proposed based on the fast diffusion in the irradiation induced damage region of the ion implantation. In addition, we demonstrate spatial uniformity obtained on molecular beam epitaxy (MBE) material of the compositional and implanted species profile. This reflects spatial uniformity of the ion implantation/annealing Processes and of the MBE material characteristics.

Resolution of conflicts concerning Fe16N2 (abstract)

The 1994 3M-Intermag meeting featured an invited session on Fe16N2 in which many of the research workers in this exciting field put forth apparently conflicting views on the nature and possibilities for the use of, perhaps, the most promising transition metal based alloy for magnetic applications. The debate centers on the promise of 3μB per atom at transition metal densities if pure Fe16N2 could be made. A particularly compelling resolution of the conflicts was not mentioned at the meeting, even though it has been in the past. The conflict arises primarily from the original classification of Fe16N2 as a phase denoted by α″ to distinguish it from Fe8N classified as a phase denoted by α′ and the substitutional alloy α-Fe(N). The conflicts arise because data are interpreted as a mixture of two phases α′ and α″. For any measured property, it is then sufficient to know the ratio of the two component phases for two different compositions to deduce the apparent properties of the individual phases. But this is not the case. The distinction between α″ and α′ is not that of two separate phases but, rather, the end points of a continuum characterized by the degree of order. This makes it necessary to produce the fully ordered state in order to determine its properties, unless it is possible to have enough data about those properties for various degrees of order to justify extrapolation to the fully ordered state. The most ordered Fe16N2 is produced using molecular beam epitaxy. It also shows the highest magnetic moments. The high magnetic moments are supported by Mössbauer measurements of large hyperfine fields in bulk materials (not the molecular beam epitaxy material) and by some of the theoretical calculations for the ordered state. One of the criticisms of the molecular beam epitaxy work is that those thin films do not decompose into Fe4N and Fe(N) on heating as do bulk alloys of the same composition. (The implication is that the thin films contain something else.) But this criticism is not valid, simply because one cannot predict the stability of a highly ordered state from observations of a weakly ordered state. This is particularly the case where the ordering is accompanied by large lattice distortions. The existing literature is analyzed to extract the dependence of magnetic moment on the degree of order. These considerations lend support to those who are pursuing Fe16N2 for its potential for producing a magnetic material with 3μB per atom at transition metal densities. How to increase the degree of order in bulk materials becomes a central concern.

Visible Blind uv GaN Photovoltaic Detector Arrays Grown by rf Atomic Nitrogen Plasma MBE

ABSTRACTRF atomic nitrogen plasma molecular beam epitaxy (MBE) was used to deposit gallium nitride (GaN) p-i-n junction photovoltaic detectors on (0001) sapphire. The detectors consisted of a bottom contact layer n-type silicon doped to 5 × 1018 cm−3. The intrinsic layer was undoped and possessed an n-type background carrier concentration of 1 × 1016 cm−3. The top /p-GaN layer was doped with magnesium to give a Hall concentration of 5 × 1017 cm−3. The p-type GaN cathodoluminescence (CL) spectra showed a strong 372 nm emission level in contrast to the 430 nm level observed in MOCVD samples. These layers were fabricated into 1 × 10 element detector arrays using a chlorine-based reactive ion etch (RIE) and refractory metal ohmic contacts. Peak responsivity of 0.11 AAV on detectors without anti-reflection coating were obtained at the GaN bandedge of 360 nm. The ultraviolet (UV) to visible rejection ratio was greater than 103 − 104 and was accredited to the reduction of the yellow defect levels in MBE material. Preliminary results on AlxGa1−xN detectors with responsivity peaks at 313 and 343 nm are presented as well.

Native defects in gallium arsenide grown by molecular beam epitaxy and metallorganic chemical vapour deposition: effects of irradiation

Gallium arsenide (GaAs), doped n with silicon nominally to 10 15 and 10 16 cm −3, grown by molecular beam epitaxy (MBE) and by metallorganic chemical vapour deposition (MOCVD), was characterized by photoluminescence (PL) spectroscopy. In MOCVD GaAs, we found the signature of the arsenic antisite (As Ga) at 0.702 eV, but in MBE GaAs we found the signature of the gallium antisite (Ga As) at 1.441 eV as well as silicon at the arsenic site (Si As) at 1.483 eV. We used proton (0.6–10 MeV) irradiation to increase the number of intrinsic defects. The fluence range was 10 10 to 10 13 cm −2 for MOCVD GaAs, and 10 10 to 10 14 cm −2 for MBE GaAs. At low fluences, the effect of irradiation was to reduce the PL intensity, which became 10% at 10 11 cm −2 in MOCVD GaAs, and 10% at 10 12 cm −2 in MBE GaAs. Annealing the samples to 550 °C for 30 min resulted in the total recovery of the PL intensity for MBE GaAs, but only a 70% recovery in MOCVD GaAs even at these low fluences. The signature of gallium vacancies (V Ga) appeared only in the irradiated MOCVD samples, at a fluence of 10 11 cm −2 and higher, but never in the MBE samples. We conclude that MOCVD GaAs was grown under arsenic rich conditions but MBE GaAs under gallium rich conditions. MBE material is about ten times more resistant to radiation than MOCVD GaAs. The presence of V Ga in this material may limit the optical output of devices depending on electron-hole recombination. In the case of as-grown MBE GaAs, the presence of Si As and Ga As lowers the electron-hole recombination efficiency.

Optically controlled quasi-optical local oscillator injection for a 100 GHz SIS imaging receiver

We present a novel approach to the problem of quasi-optical LO injection in a 100 GHz SIS imaging receiver. The use of a specially engineered molecular beam epitaxy material as an all-optical millimeter wave modulator is presented. This optically controlled modulator, used together with a kinoform to generate and control an array of Gaussian beams, is an efficient, completely solid-state, rugged and physically small solution. The working principle of the optically controlled modulator lies in the fact that the conductivity of a semiconductor can be changed by the generation of free carriers under photonic excitation, thus changing its refractive index.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Assessment of semiconductor epitaxial growth by transmission electron microscopy

Defect microstructures within heteroepitaxiallayers are categorised according to whether they result from an inherent materials problem, are due to problems at the epilayer/substrate interface, are introduced during growth itself, or are formed during postgrowth processing. Examples are presented which include dimorphism in CdS/GaAs, twinning in epitaxial CdTe, interfacial misfit dislocationformation in ZnTe/GaAs, and interfacial phaseformation in (Cd,Zn)S/GaAs. It isfound that the development of (In, Ga)(As,P) based infrared multiple quantum well lasers grown by chemical beam epitaxy (CBE) is hindered by the limited control of the quaternary composition, as compared with gas source molecular beam epitaxy material. The use of growth interrupts to characterise gas switching procedures in CBE (In, Ga)( As,P) is described. Relaxation initiation of molecular beam epitaxy SiGe/Si structures is associated with the presence of transition metal impurities. Surface profile imaging used to investigate the first stages of defectformation is described with reference to the problem of micro twinning.MST/3249

Assessment of semiconductor epitaxial growth by transmission electron microscopy

Defect microstructures within heteroepitaxiallayers are categorised according to whether they result from an inherent materials problem, are due to problems at the epilayer/substrate interface, are introduced during growth itself, or are formed during postgrowth processing. Examples are presented which include dimorphism in CdS/GaAs, twinning in epitaxial CdTe, interfacial misfit dislocationformation in ZnTe/GaAs, and interfacial phaseformation in (Cd,Zn)S/GaAs. It isfound that the development of (In, Ga)(As,P) based infrared multiple quantum well lasers grown by chemical beam epitaxy (CBE) is hindered by the limited control of the quaternary composition, as compared with gas source molecular beam epitaxy material. The use of growth interrupts to characterise gas switching procedures in CBE (In, Ga)( As,P) is described. Relaxation initiation of molecular beam epitaxy SiGe/Si structures is associated with the presence of transition metal impurities. Surface profile imaging used to investigate the first stages of defectformation is described with reference to the problem of micro twinning.MST/3249

Band-to-acceptor transitions in the low-temperature-luminescence spectrum of Li-doped p-type ZnSe grown by molecular-beam epitaxy.

Photoluminescence (PL) and magnetospectroscopy (in magnetic fields up to 12 T) have been used to characterize the properties of Li-doped p-type heteroepitaxial ZnSe on GaAs grown by molecular-beam epitaxy (MBE). A conduction-band-to-acceptor (e-${\mathit{A}}^{0}$) peak at 2.706 eV has been observed and identified in low-temperature (1.7 K), low-excitation-level PL measurements, in addition to the more commonly reported donor-to-acceptor (${\mathit{D}}^{0\mathrm{\ensuremath{-}}}$${\mathit{A}}^{0}$) pair recombination peak at about 2.692 eV. The ${\mathit{e}}^{0\mathrm{\ensuremath{-}}}$${\mathit{A}}^{0}$ peak appears to occur at low temperature only in p-type-doped material, which may explain why it has not previously been detected in ZnSe below about 25 K. PL measurements as a function of temperature, excitation intensity, and magnetic field have been performed to confirm and study the nature of the peak. The e-${\mathit{A}}^{0}$ peak shifts and broadens linearly with increasing temperature as expected, but does not show strong excitation intensity-dependent shifts or quench at high temperatures as the ${\mathit{D}}^{0\mathrm{\ensuremath{-}}}$${\mathit{A}}^{0}$ peak does. The e-${\mathit{A}}^{0}$ peak becomes narrower and shifts linearly to higher energy in applied magnetic fields, reflecting the expected behavior of the lowest-energy Landau level in the conduction band.A light-hole binding energy of 114.1\ifmmode\pm\else\textpm\fi{}0.4 meV in strained material is obtained from the intercept of a linear fit to the temperature-dependent e-${\mathit{A}}^{0}$ peak positions, which corresponds to 114.4\ifmmode\pm\else\textpm\fi{}0.4 meV in unstrained material. The inapplicability of a linear Haynes's-rule-type of relationship to shallow acceptors in ZnSe is emphasized. An anomalous initial quenching of the e-${\mathit{A}}^{0}$ peak intensity is observed as the temperature is raised, followed by the more normal increase due to the thermal ionization of the donors. This observation is modeled in terms of the temperature dependence of the competing (nonradiative, etc.) recombination rates. From the position of the e-${\mathit{A}}^{0}$ peak as a function of magnetic field, we obtain an electron effective mass ${\mathit{m}}_{\mathit{e}}^{\mathrm{*}}$=0.17, neglecting spin splittings, which were not well resolved. In previous measurements of these samples, the e-${\mathit{A}}^{0}$ peak was tentatively identified as the R-band, which has been associated with transitions between preferentially paired interstitial Li donors and substitutional Li acceptors on Zn sites. Based on the present results, we show that, in fact, no direct evidence exists in the PL spectrum for the presence of interstitial Li donors in this MBE material, which has important implications when attempting to explain the present limitations on p-type-doping levels achievable with Li.The possibility of observing e-${\mathit{A}}^{0}$ peaks at low temperature should be considered in future analyses of PL spectra of p-type-doped ZnSe, to avoid similar errors in interpretation. The occurrence of the e-${\mathit{A}}^{0}$ peak as a function of growth temperature and Li doping concentration is discussed. We describe and model the splitting of the Li acceptor--bound exciton into a doublet in the strained material. Finally, we discuss the observation of an excited-state-donor-to-acceptor peak involving Li acceptors, and the observation of discrete donor-acceptor pair lines involving Li.