NAMBE2016 Session MBE-MoP: Poster Session

In this study, the effects of three different Si (111) surfaces on the metamorphic growth of relaxed Al are explored. Following both ex-situ chemical etching, and in-situ UHV thermal desorption cleaning, the three different silicon surfaces were created through specific process variations and verified through surface reconstructions as determined by Reflection High Energy Electron Diffraction (RHEED), specifically the 1x1, the 7x7, and the √3x√3.

Traditionally, twin free FCC metals, like Al, are difficult to grow and are achieved only after heat treatment [1]. Pole figures of the off-axis Al {111} planes indicate that films grown on the 1x1 and the 7x7 surface are subject to varying degrees of twinning, as demonstrated in Fig. 1a and 1b, while there is no evidence of twinning from any of the samples grown on the √3x√3 surfaces (Fig. 1c). Further, the Al film grown on the 7x7 surface has a height variation and RMS roughness twice that of the Al film grown on the √3x√3 surface, as shown in Fig. 1d and 1e. These samples demonstrate the influence of the surface reconstruction on growth of Al on Si, thus supporting its use as a model system.

In this work, Monte Carlo simulation is used to study the surface diffusion characteristics of Ga adatoms depending on the growth rate (v) and As₄/Ga flux ratio (R_As/Ga). As a basis, the kinetic Monte Carlo model of GaAs MBE growth on the GaAs(001)-β2(2×4) reconstructed surface [1] has been taken. The algorithm was developed to obtain the surface diffusion characteristics of Ga adatoms during growth via a flux of arsenic tetramers.

As shown in Fig. 1a, Ga diffusion length is larger at smaller growth rates. An increase in the growth rate as well as in the As₄/Ga flux ratio reduces Ga diffusion length. This behavior can be explained by the As₄/Ga flux ratio dependences of Ga surface lifetime (Fig. 1b). As the growth rate rises, Ga lifetime decreases essentially due to the intensification of the incorporation processes [2]. As the computations show, the diffusion coefficient increases on the contrary with increasing growth rate or As₄/Ga flux ratio (Fig. 1b). This indicates that the diffusion length is more effectively influenced by a decrease of Ga surface lifetime than by an increase of the diffusion coefficient. The As₄/Ga flux ratio influence on the diffusion length and diffusion coefficient is stronger at v = 0.01 ML/s whereas this influence on Ga surface lifetime is nearly the same at any v. The obtained values of the diffusion length, surface lifetime and diffusion coefficient are in good agreement with experimental data [3].

The reported study was funded by RFBR, according to the research project No. 16-37-60033 mol_а_dk.

Silicon has been considered as an attractive substrate for GaN growth as a result of its low cost, large size, and good thermal conductivity. Among all kinds of buffer layers, the Al(Ga)N/GaN superlattices (SLs) is one of the most widely used buffer layer structures. However, cracks are unavoidable even in 1 µm thick GaN when using thin (≤120 nm) AlGaN/GaN SL buffer layer.[1]

In this study, a plasma-assisted molecular beam epitaxy (PAMBE) system is used to grow GaN on Si (111) substrates using thin buffer layer structures. Depending on the layer design, buffer layers consist of a combination of AlGaN/GaN SLs and/or nano-holes patterned GaN layer are used, as shown in Table I. The SL structure consists of 30 pairs of Al_0.1Ga_0.9N and GaN with a thickness of 2.5 nm and 1.5 nm, respectively. For nano-holes fabrication, a 150 nm GaN layer is used. For efficient dislocation reduction, the diameter and period of the triangular array of nano-hole mask are 190 nm and 350 nm, respectively.[2] The crystalline quality and residual strain of the GaN cap layer are investigated using high resolution x-ray diffraction (HRXRD) and Raman scattering spectroscopy. The full-width at half-maximum (FWHM) of HRXRD rocking curves in (002) and (102) reflections and the high-frequency E₂ phonon mode of Raman spectra are compared. Samples D, E, and F, which use thin SLs as the buffer layer, show the same E₂ phonon mode as that of the free-standing GaN substrate (567.7 cm^-1). This indicates that the 120 nm SL buffer layer can compensate the residual strain leading to a crack-free GaN layer. On the other hand, the insertion of the nano-holes buffer layer can reduce defect density effectively. Using a combination of AlGaN/GaN SLs and nano-holes buffer layer structure, a 3 µm thick low defect density and crack-free GaN epilayer (sample F) grown on Si (111) has been demonstrated. The lowest etch pits density achieved in this sample is ~2×10⁷ cm^-2.

Oxygen is commonly known to play an important role as an unwanted background impurity in GaN synthesized by various growth techniques. When screening for levels of unintentional impurities such as oxygen, silicon, carbon, tantalum, manganese, arsenic and boron in GaN grown in our plasma-assisted MBE system, solely oxygen is found to be the dominant background dopant. With special focus on oxygen we find that its concentration is drastically reduced by more than 1 order of magnitude at elvated growth temperature in the range between 600 and 670°C. In consequence, the free-electron background level is lowered respectively (Figure, left panel). Reduction of the oxygen level is accompanied by a drop in the photo-luminescence intensity of the donor-bound exciton with respect to the free exciton in the grown samples (Figure, right panel). GaN/AlGaN heterostructures hosting 2-dimensional (2D) electron gases exhibit significantly higher carrier mobilities and pronounced quantum transport when grown at elevated temperature. A dramatic improvement in switching characteristics - specifically the on-to-off source-drain current ratio - of lateral field-effect transistors fabricated from these 2D systems completes a conclusive picture for the necessity of appropriate growth conditions to obtain intrinsic material for device applications as well as fundamental research [1].

The optical properties of In0.46Al0.54As/Ga0.46Al0.54As/GaAs quantum dots (QDs) have been investigated. An AFM image of the uncapped In0.46Al0.54As/Ga0.46Al0.54As/GaAs QDs and the photoluminescence (PL) spectrum measured at T=8 K indicate that the QDs have very good crystal quality. The PL peak energy has a blueshift of 44 meV when the laser intensity increases by four orders of magnitude, indicating a type-II band alignment of the QDs.[1] The formation of the type-II band alignment is explained by that the quantum-confinement effect pulls up the minimum electron energy level in the QDs and the Γ→X transition in the Ga0.46Al0.54As barrier.[2] The time-resolved PL (TRPL) spectrum of QDs at peak wavelength exhibits a double-component decay behavior, suggesting the coexistence of type-I and type-II band alignments in this QD sample.[3] The emission from the continuum state of the QDs are directly measured by PL, PL excitation (PLE), and TRPL, respectively. The PLE and TRPL spectra reveal an efficient carrier relaxation from the Ga0.46Al0.54As barrier into the QD ground state via the continuum state. The temperature dependence of PL spectra and carrier lifetime shows a decrease of PL linewidth and a strong redshift of peak energy at low temperature, giving evidences of carriers thermal activation, lateral transfer and redistribution though continuum state of QDs.

+ Author for correspondence: hbuwangying@126.com and liangbaolai@gmail.com

[1] Y. S. Chiu, M. H. Ya, W. S. Su, and Y. F. Chen, J. Appl. Phys. 92, 5810 (2002).

[2] W. Zhou, Z. M. Zhu, F. Q. Liu, B. Xu, H. Z. Xu, and Z. G. Wang, J. Cryst. Growth 200, 608 (1999).

[3] L.L. Su, B.L. Liang, Y. Wang, Q.L. Guo, S.F. Wang, G.S. Fu, Z.M. M. Wang, Y. I. Mazur, M. E. Ware, G. J. Salamo Appl. Phys. Lett. 107, 183107 (2015).

We have fabricated lattice-matched In0.53Ga0.47As/In0.52Al0.48As/InP quantum well (QW) with different well width of 15nm, 25nm and 50nm by molecular beam epitaxy (MBE). The photoluminescence (PL) and time-resolved photoluminescece (TRPL) were performed to investigate the effect of well width on PL emission (wavelength, the energy gap of excitons between the excited state and the ground state, and the decay time). Intensity dependence of PL measured at T=8 K indicate that the emitting wavelength of QWs change from 1496 nm to 1537 nm to approach the emitting wavelength of bulk In0.52Al0.48As material while increasing the QW width. When the laser intensity increases by four orders of magnitude, the PL peak energy have almost no change, the full width at half maximum (FWHM) keeps stable, indicating that the In0.53Ga0.47As/In0.52Al0.48As interface fluctuation is neglitable and the QWs have good crystal quality [1]. In our measurements, the excited state of excitons of all QWs can be observed distinctly under strong excitation, and the energy gaps between the first excited state and the ground state are proportional to QW width. The decay time of PL measured by TRPL is also proportional to well width, reflecting the decrease of oscillator strength as the QWl width increases [2]. Therefore, the quantum confinement effect vary as we change the dimention of the In0.53Ga0.47As/In0.52Al0.48As/InP quantum well.

+ Author for correspondence: hbuwangying@126.com and liangbaolai@gmail.com

[1] L. C. Poças, J. L. Duarte, E. M. Lopes, I. F. L. Dias, E. Laureto, D. F. César, J. C. Harmand, J. Appl. Phys.100, 053519 (2006)

[2] Marina S. Leite, Robyn L. Woo, William D. Hong, Daniel C. Law, Harry A. Atwater, Appl. Phys. Lett. 98, 093502 (2011).

GaAs/Ge-based multi-junction solar cells could significantly increase their efficiency by incorporating a lattice-matched 1.0 or 1.15eV bandgap layer. GaAsSbN appears as a main candidate, but this material faces important growth challenges inherent to the dilute nitride alloys. GaAsSb/GaAsN superlattices (SL-II) could overcome these problems by spatially separating Sb and N atoms. Moreover, these type-II structures would present longer carrier lifetimes, which could result in improved extraction efficiency. Time-resolved photoluminescence (PL) analysis confirms longer carrier lifetime in SL-II than in bulk and GaAsSbN/GaAs type-I (SL-I) counterparts (fig. 1 a), while at the same time showing the largest PL intensity. X-ray diffraction (XRD) (inset in fig. 1 a) shows perfect GaAs-matching of SL-II after independent Sb/N calibration, which is not achieved in the other structures due to Sb-N interaction. In addition, Transfer Matrix simulations as well as photoreflectance (PR) measurements demonstrate bandgap tunability through period thickness (fig. 1 b). P-i-n solar cell devices fabricated using thick SL-I and SL-II structures (12 nm period) show high photocurrent (PC) in the 1.15eV region. Remarkably, PC is higher in the type-II structures (fig. 1 c). Nevertheless, a strong PC enhancement under reverse bias reveals incomplete carrier collection at 0V, which puts into question the formation of effective minibands. Minibands with strong electronic coupling favoring carrier transport seem to appear only for thin periods around 6 nm, leading to an improved carrier collection efficiency and enhanced PC at 0V (fig. 1 c). Solar cell performance under AM1.5 conditions will be also discussed.

InAs quantum dots (QDs) on GaAs(001) exhibit unique I-V characteristics, which varies with size. We have reported that when the size reaches the order of 100 nm, their I-V characteristics turns ohmic, allowing them to be used as nanoelectrodes. This behavior was understood in terms of formation of surface accumulation layer in InAs as they become bulk-like [1]. Recently, it was found through contact potential difference (CPD) measurements that dip in the workfunction (WF) arise at the peripheral of the QDs [2]. Among the possible driving forces, we show that the WF dip is mainly induced by the strain due to lattice mismatch.

In the present study, In(Ga)As QDs were grown by MBE. Kelvin Probe Force Microscopy (KFM) measurements were performed on QDs grown on strained and strain-relaxed buffer layers. The CPD is converted to WF through control measurement, and analyzed in connection to the simultaneously obtained topograph.

Figure 1 shows the structure, WF images and cross sections for three samples. Dip in the WF is clearly observed in all samples although the Ga content and strain induced in the QDs are different. The WF of In_0.67Ga_0.33As QD (sample B) was almost identical to that of the QD in sample A except at the apex of the QD where the strain is relaxed. The Ga content at the apex was estimated from the measured WF, agreeing with the aimed value. Consequently, we believe that the Ga content in the QDs is homogenous, not being the main cause of the dip formation. Further, the WF dip around the InAs QD on a strain-relaxed layer (sample C) was found to be deeper than that on a strained layer (sample A). The result suggests that strain is the main driving force for the dip formation, although Ga content can be disturbed as a result. The detailed mechanisms with implications will be presented.

GaN/AlN/GaN/AlN/GaN (6 nm/2 nm/3 nm/2 nm/12 nm) resonant tunneling diodes (RTD) were grown by rf-plasma assisted molecular beam epitaxy on freestanding 18 × 18 mm², Ga-polar, GaN substrates grown by hydride vapor phase epitaxy (HVPE). The density of threading dislocations (TD) in the freestanding HVPE GaN substrates is less than 10⁷ cm^-2, and the substrates were prepared prior to growth in a manner which avoids generation of new TDs at the regrowth interface [1]. Cross sectional transmission electron microscopy (XTEM) of similarly-grown structures confirmed the layer thicknesses, and no new TDs were observed either at the regrowth interface or within the device layers under the diffraction conditions used. All epitaxial layers were grown at 860° C (thermocouple), approximately the temperature corresponding to the onset of rapid desorption of metallic gallium from the surface. The ratio of gallium to active nitrogen fluxes (Ga/N*) was approximately 1.7, yet the as-grown sample surface was free of metallic Ga. RTDs with lateral area ranging from ~10 mm² to ~100 mm² were fabricated, and these devices exhibited stable, repeatable and hysteresis-free negative differential resistance.

⁺ Author for correspondence: david.storm@nrl.navy.mil

[1] D. F. Storm et al., J. Cryst. Growth 380, 14 (2013)

Monte Carlo simulation on molecular beam epitaxial (MBE) growth of self-assembled (SA) InAs quantum dots (QDs) on GaAs(001) is performed. The results are correlated with the reflection high-energy electron diffraction (RHEED) intensity transients.

Earlier work on MBE growth simulation was performed by Vvedensky et al. based on solid-on-solid model [1], well reproducing the RHEED specular beam intensity oscillations. The results demonstrated that the oscillations are indeed correlated to layer-by-layer growth of epitaxial layer, as Cohen et al.[2] and Joyce et al.[3] had interpreted.

Understanding diffraction transient during lattice-mismatched system, however, remains to be a challenge due to strain and resulting formation of three-dimensional structures, as is the case with InAs growth on GaAs(001). The arise of chevron or anisotropy along the and axes in RHEED patterns during QD formation have been interpreted, for example, by Hanada et al. [4] or by Pashley et al. [5], by considering refraction of electron beams and the shape of the QDs .

We previously reported that the intensity transient of the (004) diffraction spot during QD formation can be divided into two parts, the first slope dependent on the In flux, and the second independent. We associated the first to the QD nucleation and the second to the growth of the QDs to larger size [6]. Further investigation showed that the two slopes are observed in both azimuth, i.e., to , with a delay along [7].

Our present simulation on the total QD volume successfully reproduces the two slopes observed by RHEED. The results of these simulations are used to estimate the diffusion length of InAs adatoms.

This work was supported by the Strategic Research Infrastructure Project, MEXT, Japan.

⁺ Author for correspondence: kamiya@toyota-ti.ac.jp

[1] T. Shitara, D. D. Vvedensky, B. A. Joyce, et al., Phys. Rev. B 46, 6815 (1992).

[2] J. M. Van Hove, C. S. Lent, P. R. Pukite, P. I. Cohen, J. Vac. Sci. Technol. B 1, 741 (1983).

[3] J. H. Neave, B. A. Joyce, P. J. Dobson, N. Norton, Appl. Phys. A 31, 1 (1983).

[4] T. Hanada, B. H. Koo, H. Totsuka, and T. Yao, Phys. Rev. B 64, 165307 (2001).

[5] D. W. Pashley, J. H. Neave, B. A. Joyce, Surf. Sci. 476, 35 (2001).

[6] I. Kamiya, K. Matsuura, T. Higashinakagawa, MRS Symp. Proc. 959, M17-11 (2007).

[7] K. Shimomura, T. Shirasaka, D. M. Tex, F. Yamada, I. Kamiya, J. Vac. Sci. Technol. B 1, 02B128 (2012).

The development of broadband swept-source lasers in the optical window from 1.1 to 1.3 µm has drawn great interest recently due to its application in advanced Fourier-domain optical coherence tomography (OCT).[1] In OCT imaging, the axial resolution is determined by the bandwidth and center wavelength of the light source. The wavelength window from 1.1 to 1.3 µm is preferred for skin and dermatology applications. In this context, the emission wavelength tunability from the unique three-dimensional confinement and the broadness generated by the inherit inhomogeneity of self-assembled InAs quantum dots (QDs) has made it a great candidate in this applications. In addition, due to the delta function density of state and high differential gain in QDs, the linewidth enhance factor, α, chirp and laser threshold current is also smaller in QDs, which will reduce the imaging noise and threshold current in swept source-OCT.

We have demonstrated mode-hopping-free broadband swept-source laser with 80-nm wavelength span centered at 1.24 µm by using the InAs quantum dot (QD) gain chip devices. The broadband QD gain chip devices are grown by a molecular beam epitaxy (MBE) system, fabricated and characterized. In order to achieve broadband and uniform emission spectra, the active regions contain a combination of chirped InAs QDs with different In_xAl_1-xAs strain reducing layers (SRLs). It has been shown that the In_xAl_1-xAs SRLs can effectively enlarge the QD emission energy separation between the ground and the excited states. In this study, QDs with increased energy separation are utilized to enable a broadband QD ensemble by adding additional QD layers with the ground state emission wavelength tuned to fill the gap in emission. Characterizations of the emission spectra of the as-grown QDs and fabricated QD SLEDs are performed by room temperature photoluminescence and electroluminescence under continuous-wave (c-w) operation, respectively. In the SLED devices, a broad and uniform spectrum with a linewidth of 100 nm at 3 dB is realized at an injection current level of 100 mA. Without using an additional semiconductor optical amplifier, the swept source laser exhibit output power >1 mW and intensity variation ~10dB. The dependence of the lasing threshold current density vs. emission wavelength is also investigated in the 80 nm wavelength-tuning span is measured. Figure 1 illustrates the swept source laser emission spectrum with the tuning range covering the wavelength span from ~1.2 to 1.3 µm.

[1] W. Drexler and J. G. Fujimoto, Optical coherence tomography: technology and applications: Springer Science & Business Media, 2008.

Because Hg_1-xCd_xSe semiconductor alloys are nearly lattice matched with large-area, high quality, III-V scalable substrates such as GaSb, these alloys are potential candidates to produce future IR-based devices. Recently, Hg_1-xCd_xSe alloys have been grown by molecular beam epitaxy in order interrogate their optical, electrical and structural characteristics, particularly to compare their performance with Hg_1-xCd_xTe, the workhorse material used for long wavelength IR applications. In a previous study, we mapped the dielectric functions of a series of Hg_1-xCd_xSe alloys, exploring mainly their optical properties in a spectral range between 2000 nm to 35,000 nm. [1] In this work, we focus on the electron-phonon coupling of Hg_1-xCd_xSe alloys, which is an important parameter that describes a material, particularly with regards to its suitability for future IR-applications. A series of Hg_1-xCd_xSe thin films, deposited on both ZnTe/Si(112) and GaSb(112) substrates were investigated using temperature dependent spectroscopic ellipsometry. The measurements were performed in a spectral region between 200 nm and 1800 nm. For each sample, ellipsometric spectra were obtained at different temperatures by incorporating a cryostat into the ellipsometer. While the room temperature scans were obtained at multiple angles of incidence, the low-temperature measurements were obtained at only a single angle of incidence (70°). Using a standard inversion technique, the spectroscopic ellipsometric data were modelled in order to obtain the temperature dependent dielectric functions of each of the Hg_1-xCd_xSe thin films. The dielectric functions were further analyzed in order to obtain the evolution of the critical points with temperature. This was performed by representing each of the critical points with a Kramers-Kronig consistent oscillator. Finally, the temperature dependence of the critical points was fitted to a Bose-Einstein occupation distribution function, which consequently allowed us to determine the electron–phonon coupling parameters of Hg_1-xCd_xSe alloys.

In this study, we report on the molecular beam epitaxy (MBE) growth of crystalline SrTiO₃ (STO) films on GaAs (001) substrates as an intermediate buffer layer for the hetero-epitaxial growth of ferromagnetic La_0.7Sr_0.3MnO₃ (LSMO) and room temperature multiferroic BiFeO₃ (BFO) thin films using laser-MBE. The crystalline quality and chemical composition of the BFO/LSMO/STO/GaAs heterostructures were investigated by a combination of x-ray diffraction and x-ray photoelectron spectroscopy. And to elucidate the structure, chemistry, and quality of interfaces, high-resolution transmission electron microscopy were employed. While most of the interfaces are sharp and clean, there is a layer of amorphous Ga₂O₃ formation at the STO/GaAs interface. Prior studies of the STO/GaAs interface reveal a sharp crystalline transition after the growth of STO using MBE. Since the laser-MBE growth process utilizes a higher oxygen pressure (>4 orders of magnitude) during growth, the inter-diffusion of oxygen through the oxide films caused oxidation of the Ga and converting some of the GaAs into Ga₂O₃. Surface morphology and ferroelectric switching of domains of the heterostructure were carried out using piezoresponse force microscopy. These measurements show a repeatable polarization inversion which suggests a potential for read-write operation in non-volatile memory integrated with compound semiconductors. The temperature and magnetic field dependent magnetization measurements reveal an unexpected enhancement in magnetic moment and improved magnetic hysteresis squareness originating from the BFO/LSMO interface. We observe a strong dependence of exchange bias with LSMO thickness for both the polarity of field cooling. The enhancement in magnetic moment and magnetic coupling is related to an electronic orbital reconstruction at the interface and complex interplay between orbital and spin degrees of freedom. Saturated ferroelectric hysteresis loop is obtained with a remnant polarization ~90.68 µC/cm² for the heterostructure. The electrical characterization demonstrates excellent bipolar resistive switching with high retention time, cyclic endurance, and low set/reset voltages. The nanostructure and the physical-composition obtained from various characterization techniques of the multilayers are correlated with the corresponding dielectric and ferroelectric properties. These provide an understanding of the heteroepitaxial growth of BFO and LSMO, integrated on STO buffered GaAs with full control over the interface structure at the atomic-scale. Combined this work paves the pathway towards designing future generation low-power magneto-electronic devices integrated with GaAs.

Enhancement of Crystalline Quality of Epitaxial SrTiO₃ on Silicon

Z. Wang ,^a X. Bai,^a D. Baek,^a X. Huang,^a H. Paik,^b J. D. Brock,^a J. P. Maria,^c L. F. Kourkoutis,^{a, d} and D. G. Schlom^{b, d}

^a School of Applied and Engineering Physics, Cornell University, Ithaca, NY, 14853, USA

^b Department of Materials Science and Engineering, Cornell University, Ithaca, NY, 14853, USA

^c Department of Materials Science, North Carolina State University, Raleigh, North Carolina, 27695, USA

^d Kavli Institute at Cornell for Nanoscale Science, Ithaca, NY, 14853, USA

We report the enhanced crystalline quality of epitaxial (001) SrTiO₃ thin films on (001) Si substrates. With our modified growth recipe, the full width at half maximum (FWHM) of the rocking curve of the SrTiO₃ 002 peak of a typical 20 nm thick sample could be as narrow as 0.02°. This is about an order of magnitude narrower than the best prior report and is close to that of SrTiO₃ single crystals.

The epitaxial SrTiO₃ films are grown by reactive molecular-beam epitaxy, based on the epitaxy-by-period-annealing method that was firstly developed by the Motorola group [1]. One important factor for growing high quality SrTiO₃ films on Si is to control the oxygen partial pressure during the growth of SrTiO₃. We have studied the thickness dependence of the film crystalline quality and found that high quality films could be grown for a large film thickness range.

Our work significantly increases the quality of not only SrTiO₃ films on silicon, but also that of overlying epitaxial layers as the SrTiO₃ serves as an epitaxial template for the integration of oxides with a multitude of functional properties with silicon.

⁺ Author for correspondence: schlom@cornell.edu

[1] H. Li, X. Hu, Y. Wei, Z. Yu, X. Zhang, R. Droopad, A. A. Demkov, J. Edwards, K. Moore, W. Ooms, J. Kulik, and P. Fejes, “Two-Dimensional Growth of High-Quality Strontium Titanate Thin Films on Si,” J. Appl. Phys. 93 (2003) 4521-4525.

Because of a very high electron mobility and large g-factor InSb-based structures are of great interest for novel spintronic and electronic devices; however, the lattice-mismatch between the typically used GaAs substrates and InSb results in a high density of threading dislocations (TD), which in turn can be detrimental to the performance of semiconductor devices. The TD density can be lowered considerably with metamorphic buffers and TEM imaging has often been the technique of choice for their optimization. In this work we present a comparative study on three ways of revealing and counting the TDs using as an example InAlSb metamorphic buffers on GaAs (001) substrates (Fig.1), similar to those proposed in [1]. AFM imaging of hillocks formed from single atomic layer high terraces spiraling around each TD offer a direct way of estimating TD density. As shown in Fig.2, TD density decreases rapidly with increasing number N of dislocation filtering interlayers. The cross-sectional STEM of the N=3 structure in Fig.3 confirms the effectiveness of the interlayers in filtrating TDs (white lines). TDs on the N=3 structure surface were also exposed through DSL selective etching [2]. Etch pits were imaged and counted using Nomarski DIC Microscopy (Fig.4) and found to have surface density comparable to that of the hillocks. Since both AFM and selective etching are much more economical and faster than TEM, this study indicates that these techniques are excellent alternatives in the process of metamorphic buffer optimization.

[1] T. D. Mishima, M. Edirisooriya, N. Goel, and M.B. Santos, Appl. Phys. Lett. 88, 191908 (2006).

[2] J. L. Weyher and J. Van De Ven, J. Cryst. Growth 78 (2), 191 (1986).

Ternary GaAs_1-xSb_x nanowires (NWs) have been attracting considerable attention due to its promising wide bandgap tunability in the short wavelength infrared region for various nanoscale light emitting and photodetector devices. Although the self-catalyzed growth of GaAs_1-xSb_x axial NW arrays with band gap lowering up to 1.2 eV has been achieved [1, 2], it has been challenging to further reduce the bandgap corresponding to the wavelength of 1.3 µm, as the growth mechanism for axial configuration is primarily thermodynamically driven vapor-liquid-solid growth process. Hence, in this work we report on the growth of self-assisted axial GaAs_1-xSb_x NWs by solid source molecular beam epitaxy (MBE). A systematic study was carried out on the impact of different growth parameters on Sb incorporation and correlated to NW morphology, density and growth rate. The results of this study were then used to red shift the wavelength to the desired 1.3 µm. The Sb flux was kept invariant and corresponded to 13% Sb incorporation in the initial starting composition of the axial GaAs_1-xSb_x NW with 1.12 eV emission [1]. The MBE growth parameters investigated were substrate temperature, two-step growth process consisting of an initial growth at relatively higher substrate temperature of 620 ⁰C with the rest of the growth carried out at lower temperatures. The effects of V/III ratio were determined by varying each of As and Ga fluxes. With decreasing substrate temperature from 620 ⁰C to 580 ⁰C, Sb incorporation was found to increase from 13 at. % to 24 at. % and the radial growth rate is enhanced at the expense of vertical growth rate, in conjunction with significant reduction in the NW density. To offset the reduction in the NW density, a two-step growth process was examined. While the NW density increased, the axial growth rate was restricted, which was attributed to a deficiency in the Ga supply to retain the liquid droplet due to rapid consumption of the droplet by the crystallization at lower growth temperatures. So the next steps were to examine the effect of each flux, reduction in As flux and increase in Ga flux with a resultant lowering of the V/III ratio. An enhanced Ga flux with the two-step growth process led to the highest Sb incorporation of 28% with single NW 4K photoluminescence emission occurring at 1.3 µm. Although the NW density increased and the desired emission was achieved, the instability in the growth was reflected in the growth of NWs of uneven length and a multitude of facets. Structural properties on selected NWs using scanning transmission electron microscopy will also be presented.

A material system of AlAsSb/InGaAs on InP substrate exhibits a high conduction band offset of more than 1.6 eV. To use this material system, a short wavelength of the inter - subband transition (ISBT) of up to 1550 nm which is used in optical communication, was obtained by using a structure with coupled double quantum wells (CDQWs). By utilizing CDQWs, optical devices such as an ultrafast all-optical switching device [1] and a coherent all - optical wavelength converter [2] have been realized . However, this material system has some drawbacks when implemented in practical devices. One of the drawbacks is attributed to the AlAsSb barrier. The high Al content of the barrier layer results in easy degradation by oxidation. Thus, an alternative barrier layer is required to replace it. Since phosphides exhibit a high band gap nature, InAlP could be an alternative barrier layer. In this study, MBE growth of highly strained InAlP / InGaAs based CDQWs on an InP substrate was examined to apply a shorter wavelength in the ISBT. The sample was grown by solid-source MBE equipped with valved cracker sources for As and P. The CDQW structure was based on our previous CDQWs [3]. T he entire structure of the CDQWs is shown in Fig. 1. T he aim of this CDQW structure is to obtain better interfaces. In_0.45Al_0.65P layers were sandwiched between two In_0.45Al_0.65As layers. Their interfaces can be formed only when the As and P are replaced . On the other hand, all arsenic compounds used in the other layers resulted in good interfaces [4]. T he growth temperature was set to 460 ºC . V/III flux ratios were optimized to each layer . Fig. 2 shows the optical absorption spectra using a Fourier transform infrared spectrometer with a polarizer to obtain optical absorption for ISBT. The rapid increase in optical absorption around 1.0 eV results from the inter-band transition. The two peaks around 0.4 eV and 0.8 eV are the results of ISBT between the 2^nd and 3^rd subbands and the 1^st and 4^th subbands, respectively. By comparing peak positions for ISBT, a CDQW sample having the In_0.45Al_0.65P barrier exhibited red shifts in its optical absorption. This could be attributed to a lower conduction band offset. The absorption linewidth increased slightly in the present CDQW samples. Therefore, an optimization of the interface formation should be further required.

Coupled double quantum wells (CDQWs) exhibit a short er wavelength of intersubband transition (ISBT) owing to the energy separation between their bonding and anti-bonding states. We have realized 1550 nm of ISBT using CDQWs in different material systems such as AlAsSb/InGaAs (As/Sb) [1] and AlAs/InAlAs/InGaAs (As_all) [2] on InP substrates. When the structural quality was compared, the As_all system showed better performance [3]. However, the linewidth for optical absorption owing to ISBT was still wide for the As_all system. Thus, growth condition s should be further optimized. In this study, the growth temperature (Tg) dependence of optical features was investigated for the As_all system. The CDQW samples were grown by MBE system. CDQW structure was based on our previous CDQWs [2]. The well width (w) was set to 3.1 nm. Tg was varied between 460 ºC and 510 ºC . Optical absorption spectra were taken using a Fourier transform infrared spectromete r. Figure 1 shows the optical absorption spectra resulting from ISBT. Two ISBT peaks were found at approximately 0.35 eV (ISBT23) and 0.7 eV (ISBT14). The linewidth for ISBT exhibited strong Tg dependence where a Tg of 490 ºC showed the narrowest linewidth among the samples. On the other hand, peak positions were almost constant over all tested Tg. When optical absorption owing to interband transition (IBT) was evaluated, a blue shift of IBT was revealed for a higher Tg (not shown). This was attributed to inter- diffusion (ID) between the wells and the barriers for a higher Tg. These optical absorption features raised two issues: (1) almost constant ISBT positions in spite of causing ID, and (2) a relation between ID and the linewidth for ISBT. The first issue resulted from the existence of a center barrier in the CDQWs. The c oupling strength between the two wells increased as the ID effect increased. As a result, ISBT energy increased as well. The second issue required careful consideration of the linewidth of optical absorption. The greatest source of the linewidth broadening owing to ISBT is interface roughness (IR), which relates to the smoothness along the interface. Th us, ID and IR effects were independent of each other. When Tg was low, the sharpness perpendicular to the interface was good, whereas the sharpness along the interface became inferior . T his means that the structure of the interface was in a rigid zigzag shape along the interface. W hen Tg was optimized to 490 ºC, the rigid zigzag shape was reduced by the effect of ID: this was confirmed by X-ray diffraction measurement. Therefore, CDQWs of the As_all system were suited for obtaining sharp ISBT peaks as well as keeping constant ISBT energy.

ZnTe, II-VI compound semiconductor, was grown on a sapphire substrate, which we aimed at electro-optic terahertz wave applications. Because the atom arrangement on the sapphire’s surface varies depending on the crystallographic orientation, ZnTe films with various orientations could be grown by MBE. The m-plane (10-10) sapphire surface was changed to the nano-facet consisting of r- (1-102) and S-face (10-11) with an even spacing of 60 nm by the thermal treatment at atmosphere while various other orientation substrate exhibited atomically smooth surfaces by the similar thermal treatment. Those nano-facet sapphire substrate would force the nucleation process of ZnTe, and the crystallographic properties of ZnTe layers would be significantly controlled.

In this study, ZnTe films were grown on these nano-facet substrates by MBE. The influence of the nano-facet on the alignment of the nuclei and the resulting film quality was studied. The nano-facet structure was formed by heating the substrate around 1300 °C for 5h under atmosphere. The nano-facets were well aligned and the branching were randomly presented. Prior to the film growth, a few-nm-thickness buffer layer of ZnTe was formed at 100°C, followed by the annealing at 300°C for 5 min. The growth temperature of ZnTe layers was at around 330°C, and the thickness of ZnTe was around 1 μm.

The AFM image after the ZnTe buffer layer deposition exhibited that nuclei were uniformly aligned along the nano-facet of the substrate and the branching were barely observed . On the other hand, the ZnTe nuclei were randomly formed on the substrate when it was not covered with the nano-facet structure. Therefore, it is clear that the nano-facet structure leads for ZnTe nuclei to form the periodic alignment along the nano-facet. Crystal orientation of the grown film was analyzed by XRD pole figure measurements, and indicated that the (331)-oriented ZnTe with about 4 ° of the tilt was grown on the nano-facet m-plane substrate. For the sake of the comparison, 5° tilted m-plane substrates was used, and the orientation of ZnTe epilayers was studied. The nano-facet structure was also obtained from the 5°-tilted m-plane substrate after the same thermal procedure. The orientation of the epilayer was similarly analyzed and the (331)-oriented layer was confirmed. It was revealed that the nano-facet structure of sapphire substrate was effective to control the alignment of the ZnTe nuclei and the orientation of the epilayer.

This work was supported in part by the Waseda Univ. Res. Init., by the Waseda Univ. Grant for Special Res. Projects, by the JSPS Res. Fellowship for Young Scientists, and the Foundation of Ando Lab .

⁺ Author for correspondence: amaros@asu.edu

[1] G. Ungaro et al., Electron. Lett., vol. 35, no. 15, pp. 1246–1248, 1999.

[2] Wiemer et al., Proc. SPIE 8108, High and Low Concentrator Systems for Solar Electric Applications VI, 810804, 2011

[3] Tixier et al., Journal of Crystal Growth, vol. 251, Issues 1–4, 449-454, 2003.

Wavelength extended In_xGa_1-xAs (0.53<x<1) photodetectors (PDs) with cutoff wavelength more than 1.7 μm have been extensively investigated over the past decades due to their important applications in spatial remote sensing, earth observation, environmental monitoring etc [1]. However, the higher dark current severely hinders the device performance. The origins of the enhanced dark current are required to be investigated.

There has been much interest in the “6.1 Å III-V family materials”, the closely lattice-matched material systems of GaSb, InAs, and AlSb. Due to their unique type II band alignment and narrow band gaps, they constitute materials systems offering great flexibility in detector design and due to their unique electrical properties, are being for insertion into Si CMOS devices. Diodes based on InAs/GaSb short-period superlattices (SLs) form material systems that can be considered as III-V based alternatives to the established HgCdTe systems and quantum well infrared photodetectors made of GaAs and related compounds [1]. On the other hand, InAs/AlSb SLs can be an alternative for n-type barrier applications such as making cladding layers in laser structures [2]. The ability to tune the band gap over the entire mid- and long-wave infrared spectrum enables a large number of applications for devices made from these SLs material. The growth of high quality InAs/GaSb and InAs/AlSb SLs requires a careful study of the parameters used during epitaxial growth. Moreover, intrinsic defects in the substrate can also propagate through the buffer layer deteriorating its crystal quality. Optimization of the buffer and heterostructure growth conditions is critical for device performance.

This work investigates the growth of 60 periods InAs/GaSb and InAs/AlSb SLs by molecular beam epitaxy on GaSb(100) substrates. The SLs sub-periods thicknesses ranged from 20 to 40 Å. The substrate temperature during the growth was kept at 420^oC, whereas the V/III fluxes were kept low. Atomic force microscopy was used to determine the surface morphology, as a smooth surface is a measure of the material quality and necessary for nanoscale device fabrication. Optimal growth parameters to obtain homoepitaxial growth of atomically smooth GaSb buffer layers were found. Double crystal X-ray measurements were carried out to obtain the effective lattice constant of the SLs. The overall periodicity of the structure as measured by the fringe spacing of the SL peaks was in agreement with the recipe. The sharp X-ray diffraction satellites were indicative of good structural quality.

The growth of InAlAsSb alloys has attracted great interest due to its application in long wavelength infrared photodetectors and laser diodes. However, the growth of InAlAsSb with Al mole fraction greater than 20% presents great challenges, mainly due to the existence of a very wide range of miscibility gap for this alloy material system.

In this experiment, digital alloy technique was used to grow high quality InAlAsSb with Al mole fractions ranging from 20% to 40% by molecular beam epitaxy (MBE) with valved-As and valved-Sb crackers. The growth rate was set at 1.0 monolayer per second, as calibrated by RHEED oscillations and post-growth ex-situ thickness measurements. The growth temperatures, III-V beam equivalent pressure (BEP) ratios, and binary growth rates similar to those used in bulk random alloy growth were employed to maintain a far-from-equilibrium growth environment. The MBE growth was optimized, as determined by the streaky RHEED patterns for AlAsSb (1x3) and InAs (2x4) patterns during crystal growth. The results are shown below.

Figure1 shows the X-ray rocking curves for conventional and digital InAlAsSb alloys with Al mole fraction of 30%. The samples were grown on InAs (100) substrates. The results show that no observable epilayer peak can be seen for conventional InAlAsSb. However, in sharp contrast, the digital alloy growth technique worked successfully. The FWHM of the digital alloy is as narrow as 30 arcsec. Figure 2 shows the PL spectra for digital Al_xIn_1-xAs_ySb_1-y at 77K with various x= 0.27, 0.31, and 0.37.

Figure1 Firure2

In this work, we will present the detailed investigations on MBE growth of InAlAsSb digital alloys with 20-40% Al mole fraction and on the dependence of bandgap on Al composition for digital alloy InAlAsSb.

In this paper, InAs heterojunction phototransistors (HPTs) for room temperature operation beyond 3.0μm have been demonstrated for the first time.

(a) (b)

Figure 1 (a)The cross-sectional structure of the phototransistor. (b)The energy band diagram in equilibrium of the AlGaAsSb/InAs phototransistor at room temperature.

The phototransistor device structure was grown, lattice-matched, on InAs substrate in a GEN-II solid-source molecular beam epitaxy (MBE) with valved-As cracker and valved Sb-cracker. The MBE growth conditions were optimized, based on streaky reflection high energy electron diffraction (RHEED) patterns during crystal growth of the entire device structure.Figure 1(a) shows the cross-sectional structure of the phototransistor and Figure 1(b) shows the energy band diagram in equilibrium of this AlGaAsSb/InAs phototransistor at room temperature. A lightly n-type doped AlGaAsSb junction displacement layer reduces the defect current component in the emitter injection current. A heavily doped n-type InAs base layer maintains a stable dark current as a function of the bias voltage, as shown in Figure 2.Figure 3 shows the device’s photoresponse up to 3.8mm at room temperature. The 50% cutoff wavelength of the spectral response curve is 3.5mm at 300K. At 300K, a peak spectral responsivity of 21A/W was obtained at the wavelength of 3.1mm, which represents an optical gain of 8.4.The devices in this work demonstrate that mid-infrared phototransistors grown by solid source MBE are a promising candidate for various detection applications, such as molecular spectroscopy, remote trace-gas sensing and laser radar systems.

Figure 2 Figure 3

Figure 2_I-V characteristics of the photodiode under dark conditions at different temperatures. The area of the device mesa is 320mm ´ 140mm. Figure 3_Spectral response of the phototransistor at room temperature. The measurements were carried out under bias voltage V_EC of 1.0V.

III-V nanowires (NWs) have attracted extensive attention due to their tremendous potential in device applications with their intersting properties. Typicial route to form NWs is vapor-liquid-solid method utilizing Au droplet as a catalyst. However, Au is known to diffuse into NW and plays a role as non-radiative recombination center that drastically reducing internal quantum efficiency [1]. It can also be problematic not only to simple NWs but also heterostructure NWs that is necessary to functionalize the NWs for practical device application. As an alternative to deal with this issue, self-catalyst method emerged, that uses group-III material as a catalyst in place of Au [2]. However, axial growth of AlGaAs on a self-catalyst GaAs NWs is rarely reported while mostly focused on Au-assisted method [3].

We present an investigation into the growth temperature of axial AlGaAs which placed on self-catalyzed GaAs NWs. Samples were grown on Si (111) substrate by molecular beam epitaxy (MBE). Scanning electron microscopy (SEM), transmission electron microscopy (TEM) and high resolution X-ray diffraction (HRXRD) were used to characterize its structural properties. We find that the heterostructure NWs can be grown in a wide growth temperature range changing its structural characteristics and geometrical shape.

InAs_1-xSb_x ternary and Ga_xIn_1-xAs_ySb_1-y quaternary alloys are promising materials, especially in near- and mid- infrared regions, as an alternative to commonly used HgCdTe (MCT) detectors [1]. Particularly, they have overwhelmingly covalent bonding, resulting in stronger material, which able to withstand harsher processing steps and operating environments [2]. Besides, the alloys have fairly weak dependence of the band edge on composition compare to ionic bonding MCT [3]. Furthermore, they have lower dielectric constants and room temperature self-diffusion coefficients compared to those of MCT [3].

In this study, high quality InAs_0.83Sb_0.17 (S#1) and Ga_0.87In_0.13As_0.4Sb_0.96 (S#2) mid-wavelength infrared p-π-n photodetector structures were successfully grown with the aid of a GaSb transition layer on semi-insulating GaAs substrate by molecular beam epitaxy. The lattice mismatch and the crystal quality of the structures were investigated by high resolution X-ray diffraction rocking curve measurements. The full width at half maximum values for S#1 and S#2 were determined as 215 and 254 arcsec, respectively. The cut off wavelengths of the photodetectors S#1 and S#2 were obtained from the spectral photoresponse as 4.23 µm and 2.03 µm at 80 K, respectively.

⁺ Author for correspondence: userincan@anadolu.edu.tr

[1] K. C. Nunna, S. L. Tan, C. J. Reyner, A. R. J. Marshall, B. Liang, A. Jallipalli, J. P. R. David and D. L. Huffaker, IEEE Photonics Tech. L. 24, 3.

[2] A. Venter, P. Shamba, L. Botha and J. R. Botha, Thin Solid Films 517 4468 (2009).

[3] A. Rogalski, Prog. Quant. Electron. 27 59 (2003).

Acknowledgement

This work was supported in part by Anadolu University under the project BAP-1301F038.

The 6.1Å III-V “high-mobility” semiconductor family has the lowest effective masses for transport (InAs, InGaSb) which are appealing for CMOS circuits and provides broken or staggered band alignment important for tunnel FETs (TFETs). However, the large lattice mismatch (~12%) with Si results in 3D island nucleation and crystal misfit relaxation within the first monolayers of growth. The solution for deposition of this high mismatch material is the growth of the entire device structure from a single nucleus, such as in vertical nanowires (NWs), with recently provided high quality InAs and GaAs NWs on Si(111) substrates. In contrast, MBE of GaSb has been found to provide a limited aspect ratio for vertically grown structures. Since growth of the III-As NWs is straightforward, the GaSb NWs were grown on top of InAs NWs that were prepared with a self-catalytic vapor-liquid-solid (VLS) process. Two types of GaSb NWs were studied on Si(111): vertically stacked InAs/GaSb NWs (Fig. 1 a), and coaxial core/shell NWs (Fig. 1 b). In order to grow vertically stacked InAs/GaSb NWs (Fig. 2), high temperature (590 0 C) and metal-rich growth conditions (flux ratio Sb 2 /Ga~1) were employed. A ~200 nm long GaSb portion of NW grows clearly in VLS mode with a pronounced Ga droplet on the top of the NW. The GaSb NW top portion (Fig. 2a) shows good crystal quality without twins, optimum growth of GaSb in VLS process. The InAs template NW is only 70 nm thick, which indicates that growth of GaSb is accompanied by etching of InAs NW sidewalls. Coaxial GaSb/InAs NWs were demonstrated by MBE growth of GaSb at 500 0 C with Sb 2 /Ga~2 (Fig. 3). A relatively flat GaSb cap is visible on top of the InAs NW, but without a Ga or In metallic droplet indicating vapor-solid growth mode. Fig. 3© shows EDS x-ray elemental maps: a continuous 30 nm thick GaSb shell is deposited on 80 nm thick InAs NW initially grown on Si. The rough surface is clearly due to the formation of twins and hexagonal phase platelets that have slightly different growth rate in <110> and <211> directions (Fig. 3b).

GaMnP offers a range of new exciting properties for the field of spintronics, that includes hole-mediated magnetic exchange,[1] unprecedented possibilities of manipulating the magnetic easy axis by internal lattice strain, and important opportunities for magneto-optics because of its relatively large energy gap. Although little is known about the MBE growth of this material, some highly promising work has already been done on MBE growth of the quaternary alloy Ga_1-xMn_xAs_1-yP_y with small amounts of P [2,3]. In order to pave the way for the growth of high quality GaMnP by MBE, we are therefore exploring the growth of the Ga_1-xMn_xAs_1-yP_y system, with y between 0.10 and 0.35, grown on GaAs (100) substrates. For this purpose we have grown two series of samples: Ga_1-xMn_xAs_1-yP_y films with different P content y; and Ga_1-xMn_xAs_1-yP_y/ GaAs_1-yP_y superlattices (SL).

Magnetization measurements carried out on epilayers and SLs using SQUID show T_C ∼ 20K-45K, systematically decreasing as P concentration increases. Importantly, magnetization data clearly suggest that the easy axis depends on both the P concentration and the sample structure. For epilayers, a small amount of P leads perpendicular-to-plane orientation of the easy axis. In contrast with epilayers, SLs show a preference of the magnetization to lie in the film plane. In order to correlate the observed magnetic properties with the crystal structure, detailed studies of the samples were performed by cross-sectional TEM. TEM images show the presence of dislocations along the (111) planes for both epilayer and the SL structures (presumably due to lattice mismatch). In the case of the SL, the layer structure of the Mn-doped GaAsP layers is clearly seen, with highly perfect interfaces between GaAsP and GaMnAsP constituents, as seen in Fig. 1. While the dislocations originate at the GaAs/GaMnAsP interface, it is interesting that they appear to cross the boundaries within the SL without interruption. We assume that the dislocations seen in the TEM images provide the mechanism for lattice relaxation. In this presentation we will give special attention to identify the effect of these dislocations on the observed magnetic properties of these GaMnAsP-based films specimens. The samples are presently being studied by HRXRD in order to gain further insight into the mechanisms of lattice relaxation and its effect on magnetic properties.

This work was supported by the NSF Grant DMR14-00432.

⁺ Author for correspondence: xli18@nd.edu

[1] M. A. Scarpulla, et al., Phys. Rev. Lett. 95, 207204 (2005).

[2] L. Thevenard, et al., Phys. Rev. B 82, 104422 (2010).

[3] N. Tesařová, et al., Phys. Rev. B 90, 155203 (2014).

This work examines mid-IR resonant cavity photodetectors grown on InAs and GaSb substrates. Important characteristics of the mid-IR resonant cavity detectors include reduced dark current and narrow spectral response. The resonant cavity infrared detector structure consists of a thin IR absorbing layer inserted into an optical cavity formed by transparent layers between two mirrors – all made of MBE-grown epitaxial layers. The mirrors are quarter-wave stacks, similar to those used in VCSELs. The high reflectivity mirrors cause the light to recirculate inside the optical cavity, passing through the absorber layer many times. The multiple passes through the absorber allows for much thinner absorber layers than in conventional IR detectors, which leads to a large reduction in dark current and associated noise.

Only a narrow band of wavelengths will interfere constructively with themselves, which produces the resonant mode of the cavity. The resonant mode of the cavity is observable as a large dip in the wafer’s reflectivity and a large peak in the detector’s photoresponse.

High reflectivity mirror stacks are required for high performance resonant cavity devices. The highest reflectivity is achieved by choosing material pairs with the largest refractive index contrast. We have examined AlAs_0.16Sb_0.84 / GaAs_0.08Sb_0.94 mirror stacks on InAs substrates and AlAs_0.16Sb_0.84 / GaSb mirror stacks on GaSb substrates. Using 6 pairs of layers, >80% mirror reflectivity has been achieved on both InAs and GaSb substrates. >90% reflectivity requires 10-12 pairs. Minimizing the reflectivity at the resonant wavelength of the full resonant cavity structure produces the highest detector quantum efficiency. Cavity resonant wavelength reflectivities of just a few percent have been obtained.

Group-III nitrides based on GaN, InN and AlN alloys are promising candidates for high efficiency solar cells [1]. Tandem devices including silicon can be fabricated successfully due to the development of GaN-on-silicon technology. We report on the investigation of vertical carrier transport across nitride p-n junctions and we propose a carrier transport mechanism at the nitride-silicon interface of the solar cell structure.

The creation of crystalline defects during growth remains a major problem for epitaxial techniques, while the strong impact of defects on the physical properties of as-grown structures and later devices is a major problem for micro- and opto-electronic applications. Detailed investigation of defect creation in relation to growth conditions helps to reveal the main features of defect creation and hence to mitigate their effect on physical properties.

Our TEM and XRD investigations of epitaxial structures with initial elastic strain ranging from 0.2-0.4% up to 14% revealed the creation of various crystal defects, correlated to growth conditions, and specific structural defect transformations during epitaxial growth. Low deteriorated structures (e_XX £ 1%) may be characterized by delayed relaxation at the initial stage of deposition, then gradual increase of relaxation up to 90-92% due to gradual transformation of 60° dislocations to Lomer dislocations and specific spatial distribution of crystal defects in the volume of the epitaxial layer (Fig. 1). Rapid accommodation of the initial elastic strain by Lomer (edge) dislocations and hence instantaneous substrate/ epilayer lattice transformation under resonant atomic standing waves on the growth front is typical for highly strained epitaxial structures (e_XX ³ 4-5%). After accommodation, the growth mode switches to the low(er) deteriorated mode with specific crystal defects.

In this work, we present extended structural investigations of epitaxial structures, grown under intermediate deteriorated growth conditions: first set, grown closer to the low growth mode; second set, closer to the high deteriorated mode, in order to find specific structural features, relaxation mode and defect transformation, corresponding to these areas.

Author for correspondence: Nikolai.Faleev@asu.edu

Nanowire based optical devices have been attracting great attention by virtue of enhanced entrapment of light inside nanowire array due to its inherent one dimensional architecture with small cross section. In order to realize large scale homogeneous arrays with reliable performance, it becomes important to have good control in positioning, size and the alignment of the nanowires, in addition to its high quality and compositional uniformity. Electron beam lithography has been one of the commonly used technique to produce selective active sites for nanowire growth on silicon substrate. There have been several reports on patterned GaAs nanowires [1, 2], which focused on the growth parameters namely, Ga pre-deposition and V/III flux ratio to improve pattern coverage. In this work, emphasis is on the effect of processing conditions along with growth parameters and their inter-dependency on the yield of self-assisted growth of vertical nanowires GaAs for different pitch size, using molecular beam epitaxy on patterned Si (111) substrate. The optimization of these parameters led to >90 % occupancy of patterned holes with nanowires as shown in Fig.1 below. In order to identify the optimum pitch of nanowire array for maximum optical absorption, finite difference time domain (FDTD) simulation has been performed with wavelengths in the range of 400-1200 nm using Lumerical FDTD Solution software. The accuracy of the optical simulation and boundary conditions used are consistent with the results of Hu et al [3] . Variation in the absorption with pitch has been studied and optimum pitch corresponding to the maximum absorption was found to be consistent with the photoluminescence spectral results. The optimized growth and processing parameters have been extended to the patterned GaAs/GaAsSb axial configured nanowires on (111) Si, which will also be presented. This work is supported by the Army Research Office (Grant No. W911NF-15-1-0160, technical monitor: William Clark).

⁺ Author of correspondence: iyer@ncat.edu

[1] S. Plissard, G. Larrieu, X.Wallart and P.Caroff, Nanotechnology, vol.22, p. 275060, 2011.

[2] S. J. Gibson, J. Boulanger and R. R. LaPierre, Semiconductor Science Technology, vol. 28, p. 105025, 2013.

[3] S. Hu, C.-Y. Chi, K. T. Fountaine, M. Yao, H. A. Atwater, P. D. Dapkus, N. S. Lewis and C. Zhou, Energy Environ. Sci., vol. 6, p. 1879–1890, 201

III-V multi-junctions solar cells (MJSC) have been the subject of a large deal of research recently and showed a great potential for concentrating photovoltaic (CPV) system. They have exclusively the upper hand over the others photovoltaic technologies with an efficiency exceeding 45% [1]. CBE has been identified as a very promising technique for large scale production of MJSC [2], as it combines the advantages of using a vacuum growth environment such as MBE with the simplicity of gas sources such MOCVD. Furthermore, CBE has shown recently promising results for dilute nitrides growth with superior mobility [3] which are expected to be essential materials to further increase the efficiency of MJSC. Although CBE has shown its capability to grow III-V alloys with similar quality to what could be obtained with MBE and MOCVD. Al-based alloys remain challenging to obtain with low carbon background doping. For example, the growths of AlGaAs with the standard trimethylaluminium (TMA) source result in a p-doping density in the order of 1 E20 cm^-3 [4]. This is can suitable for some applications that involve tunnel junctions [4], but does not fulfill the requirements for both solar cell active and windows layers.

In this work, we investigate the potential of Al based III-V alloys, mainly AlGaAs and AlGaInAs growth with triethylaluminium (TEAl) precursor in CBE for MJSC applications. Our first results show already that carriers density in AlGaAs layer have been reduced to reach a value of 7.82 E16 cm^-3 with corresponding mobility and resistivity, respectively, 63.4 cm²/V.s and 1.25 Ω.cm. As illustrated in Figure 1, AlGaAs epilayer on GaAs substrate has high optical and structural proprieties which clearly pointed the need of TEAl as precursor.

( See Fig.1 in the attached PDF document)

Figure 1 : a- photoluminescence spectrum of AlGaAs at room temperature.

b- X-ray diffraction rocking curve pattern of GaAs/AlGaAs growth at 565°C.

[1] Thomas N. D Tibbits, et al. 29th European PV Solar Energy Conference and Exhibition, Sept. 2014.

[2] M. Yamaguchi, T. Warabisako, and H. Sugiura, J. Cryst. Growth, vol. 136, 29 (1994).

[3] M. Yamaguchi, K. Nishimura, T. Sasaki, H. Suzuki, et al. Solar Energy 82 (2008) 173–180.

[4] B. Paquette, A. Boucherif, R. Arès, et al. J. Cryst. Growth, vol. 374, July 2013, Pages 1–4.

Group III-V semiconductor nanowires (NWs) on Si platform have attracted an increasing interest in recent years. By integrating the direct band gap III-V materials that have high absorption coefficient onto the cost-effective Si platform, it would create novel optoelectronic devices for Si photonics. Another advantage of III-V nanowires grown on Si substrate is that the lattice-match constraint can be relaxed.

The III-V NW growth on unpatterned substrates typically follows a self-assembled mechanism where the NWs are randomly positioned. To improve device performance, each NW should be precisely located. Controlling the NW position is of critical importance and can be achieved by growing on patterned substrates. Recently, dilute nitride alloys were added to these material systems with the advantage for extending device functionalities. For example, by incorporating nitrogen in Ga(In)As, the emission energy can be tuned towards 1.3-1.55 μm desirable for telecommunication devices, due to the large bandgap bowing of dilute nitride materials.

In this work, we report epitaxial growth of GaAs NWs and GaAs/GaNAs core-shell NWs on patterned Si (111) by a self-catalyzed method with gas-source molecular beam epitaxy (GSMBE). To study the growth window and the temperature-dependent structural characteristics of GaAs nanowires, the growth runs are performed at various substrate temperatures. Different sizes and yields of GaAs NWs were observed under Scanning Electron Microscope (SEM). The optimized growth temperature for GaAs NWs is around 690 °C. Various surface treatment conditions on Si (111) substrate prior to growth are also studied. The patterned Si (111) substrates are dipped in an HCl solution, diluted HF solutions with different HF: H₂O ratios to remove the intrinsic oxide. We optimized substrate preparation process to find the highest yield of our growth. The core-shell GaAs/GaNAs NWs are successfully grown on patterned Si (111) substrate. The smooth surface of sidewalls of NWs indicates good crystal quality. Transmission Electron Microscope (TEM) is performed to analyze crystal quality and interface between Si substrate and III-V NWs. The misfit dislocation at the GaAs/Si interface is observed due to the lattice mismatch. Some twin planes are also observed at the bottom of the NW, which are often found in III-V NWs grown on Si (111) substrate.

⁺ Author for correspondence: ctu@ucsd.edu

The usage of III-nitride nanowire nanostructures for future highly efficient solid-state lighting and full-color displays has been intensively developed. Several studies have confirmed that nanowire heterostructures exhibit several potential advantages including greatly reduced dislocation densities and polarization fields. Such significant advantages provide an alternative and more advanced approach to develop phosphor-free white light-emitting diodes (LEDs), in which, full-color emission can be achieved and controlled within a single nanowire LED. In this context, we have developed full-color LEDs using self-organized InGaN/AlGaN nanowire heterostructures grown by plasma-assisted molecular beam epitaxy. The device active region includes ten InGaN/AlGaN dots. During the growth process of AlGaN barrier, an AlGaN downward-bending shell layer was also formed around the InGaN dot and formed an unique core-shell nanowire structure. Such unique LED structure offers high carrier injection efficiency and reduced electron overflow and nonradiative recombination on the nanowire surface, leading to high power phosphor-free white LEDs [1]. Highly uniform InGaN/AlGaN nanowires on Si substrates are presented in Fig. 1(a). T he device emission properties, including the correlated color temperature and color rendering index can be readily engineered by varying the size and composition of the InGaN dots in a single epitaxial growth step. Figure 1(b) shows emission spectra of such LEDs with different peak emission wavelenghts, covering almost nearly the entire visible spectrum. Showing in Fig. 1(c), the capacitance voltage (C-V) characteristics for phosphor-free nanowire white LEDs, the capacitance decreases with the increase of the reverse bias indicating the enlargement of the depletion region. Illustrating in the inset of Fig. 1(c), the average mobile charge carrier concentration in the depletion layers formed during the reverse bias is calculated to be ~1.452×10¹⁸ cm^-3. The detailed epitaxial growth, device fabrication and characterization of InGaN/AlGaN core-shell nanowire LEDs will be presented.

For more than half a century, epitaxial growth of semiconductors heterostructures allowed the development of a large variety of solid-state devices. However, due to the limited number of crystalline substrates that are available for epitaxy, only a small fraction of semiconductor alloys can be grown with device quality. This makes the ability to grow defect-free semiconductors with arbitrary lattice constants one of the most important challenges the field of epitaxial growth currently faces. Previous studies proposed virtual substrate engineering as a way to overcome this limitation showing promising results but with limited application [1].

We propose graphene-stabilized mesoporous silicon (mPS) free-standing membrane as a new compliant and lattice tunable virtual substrate for the heteroepitaxial growth of defect-free semiconductor films with an arbitrary lattice constant. This study includes experimental results along with theoretical modeling to define the virtual substrate design rules. mPS is known to be an extremely flexible material due to its reduced elastic properties, which insures the compliance even with a micrometer thick porous membrane. Additionally, the presence of the graphene sheet on the high specific surface of mPS prevents it from recrystallizating up to at least 900°C, which ensures its stability under most current epitaxial growth conditions.

⁺ Author for correspondence: Richard.Ares@usherbrooke.ca

[1] M. S. Leite, E. C. Warmann, G. M. Kimball, S. P. Burgos, D. M. Callahan, H. A. Atwater, Adv. Mater. 2011, 23, 3801.

AlGaN-based semiconductor ultraviolet (UV) devices are of tremendous importance for applications in lighting, display, water purification, and medical diagnostics. However, AlGaN-based devices generally exhibit very poor performance due to the lack of low-cost, high-quality, and large-area substrates with small lattice mismatch to the device heterostructures. For example, dislocation densities on the order of ~10⁶ cm^-2-10⁸ cm^-2, have been measured in (Al)GaN and/or InGaN heterostructures grown on sapphire, SiC and other substrates. In this context we have demonstrated dislocation-free semipolar AlGaN templates can be achieved on c-plane sapphire substrate (lattice mismatch ~13-16%) through controlled nanowire coalescence by selective-area epitaxy. The coalesced Mg-doped AlGaN layers exhibit superior charge carrier transport properties, including a free hole concentration of ~7×10¹⁸ cm^-3 and mobility ~8.85 cm²/V·s at room-temperature. The semipolar AlGaN UV LEDs demonstrate excellent optical and electrical performance, including an IQE of ~83% at room-temperature and and an output power of 15 W/cm² for an unpackaged device at a current of 90 mA.

The selective area growth (SAG) takes place on n-GaN template on sapphire substrate by using a thin Ti layer (~10 nm) as the growth mask. Nanoscale patterns with a lateral size of ~180 nm and a lattice spacing of ~250 nm arranged in a triangular lattice were created using e-beam lithography and reactive ion etching techniques. Vertically aligned GaN/AlGaN nanowire LED heterostructures were then grown using a plasma-assisted MBE system. The device structure consists of Si-doped GaN, Si-doped Al_0.35Ga_0.65N, undoped Al_0.14Ga_0.86N active region, and Mg-doped Al_0.35Ga_0.65 contact layer. The incorporation of Al leads to a small enhancement of the lateral growth, ~1 nm/min. As a result, the spacing between neighboring nanowires is slowly reduced, leading to a gradual coalescence between neighboring nanowires. Ni(20 nm)/Au(10 nm) and Ti(20 nm)/Au (100 nm) metal layers were deposited on the nanowire top surfaces and n-GaN template to serve as p-and n-metal contacts, respectively. p-Type conduction of the coalesced Mg-doped Al_0.35Ga_0.65N layers was confirmed by Hall effect measurements. Detailed structural characterization of the resulting AlGaN film structure was performed by STEM and electron energy-loss spectroscopy analysis, exhibiting unique quasi 3-D feature consisting of semipolar facets. Semipolar AlGaN LEDs exhibit excellent I-V characteristics and strong emission at 340 nm. The detailed investigation of the SAG and characterization of large-area, dislocation-free AlGaN semipolar nanowire arrays will be reported.

To date, Al-rich AlGaN based deep ultraviolet light emitting diodes (LEDs) exhibit low output power and low external quantum efficiency, which has been limited, to a large extent, by the absorbing and resistive p-(Al)GaN layer. Here we show that s uch critical issues can be addressed by using AlGaN tunnel junction (TJ) nanowire arrays, wherein an epitaxial Al layer serves as the TJ and also simultaneously functions as a mirror to reflect UV-light emitted from the AlGaN active region . Schematically shown in Fig. 1a, AlGaN nanowire LED heterostructures are grown on Si substrate by plasma-assisted molecular beam epitaxy. The TJ consists of n⁺⁺-GaN/Al/p⁺⁺-AlGaN. With a work function of 4.08 eV, Al metal can form an ohmic contact to n-GaN. The presence of defects at the Al/p⁺⁺-GaN interface, partly due to the very high Mg-doping, results in deep energy levels, which can significantly enhance carrier transport from p-GaN to Al. Scanning electron microscopy image of AlGaN tunnel junction nanowire arrays grown on Si is shown in Fig. 1b, which are vertically aligned and exhibit a high degree of size uniformity. Large area LEDs are then fabricated using standard photolithography and contact metallization techniques. The device shows excellent I-V characteristics with a turn on voltage of ~5.8 V for an areal size of 500×500 µm², shown in Fig. 2c. The device exhibits strong emission at 275 nm, shown in Fig. 1d. An on wafer output power of 2.5 mW is measured for TJ AlGaN nanowire LEDs without any packaging. The demonstration of high power TJ nanowire LEDs operating at 240 nm is in progress and will be reported.

Rare earth (RE) doped semiconductors have been attracted interest due to their applications for light emitting devices and spintronics using diluted magnetic semiconductors. Si should be widely used as a matrix element because of their compatibility for microelectronics, however, strong covalent bond of Si and large ionic radius of RE elements prevent from being RE doped with high concentration and high crystallinity. In RE elements, Ce shows very attractive 4f electron-related phenomena such as Kondo effect, heavy fermion and superconductivity. However, there are few papers describing the effect of Ce doping on the magneto-transport characteristics in Si epitaxial films. In this paper, the origin of anomalous magneto-transport characteristics in Ce doped Si epitaxial films with p-type conduction are discussed.

Although low temperature MBE enables to control the surface segregation or precipitation of Ce and the compounds, all samples show n-type conduction due to high donor density caused by low growth temperature. Because Si:Ce films with p-type conduction show ferromagnetic behavior, hole concentration controlled Si:Ce were fabricated by B co-doping.

Ce 3d XPS spectra of B co-doped Si:Ce (Si:Ce,B) films have 4 peaks originated in Ce 3d⁹4f¹ and Ce 3d⁹4f². On the other hand, there is no peak around 915 eV originated in Ce 3d⁹4f⁰, suggesting that the Ce ions dissolved in Si exist as Ce³⁺.

Based on detail evaluation of transport characteristics, negative MR is clearly observed even at 200 K in p-type Si:Ce,B films. The origin of negative MR is considered as weak localization effect or spin ordering effect. In these cases, the spin ordering model is well-fitted, suggesting the localized magnetic moment of Ce³⁺ affects carrier transport. We will discuss the effect of hole concentration on magneto-transport characteristics and the origin of spin ordering.

Hyperbolic metamaterials (HMMs) are of great interest due to their novel properties, including negative refraction and enhanced Purcell effect. In HMMs, the parallel and perpendicular components of the permittivity have opposite signs. This results in an optical dispersion surface that is an open hyperboloid, allowing light with large wavevectors to propagate within the material rather than decay exponentially. HMMs for the visible spectral range have been successfully created using traditional metals (e.g. gold, silver) and dielectrics (e.g. alumina, silica). We would like to take advantage of the unique HMM properties in the infrared frequency range, to improve detectors or enhance emitters. However, moving metamaterials to the infrared is not just a matter of scaling geometries, but also of choosing new materials with appropriate optical properties.

We demonstrate infrared HMMs with optical properties tunable across the mid-infrared created from alternating subwavelength layers of metal (doped InAs) and dielectric (undoped InAs). Our single-material systems were grown by molecular beam epitaxy on semi-insulating GaAs substrate. A fast growth rate is necessary to efficiently incorporate silicon donor atoms at the high doping levels used in these materials. Our materials exhibit low optical losses as well as high sample uniformity and sharp interfaces. The fill factor, ρ , is defined as the ratio of the thickness of the metal layer to the total metal plus dielectric thickness: t_m/(t_m+t_d). By tuning the doping density and fill factor, we can control the onset and bandwidth of metamaterial behavior. The plasma wavelength of our materials can be as short as 5.8 m and as long as 14m while the fill factor for our samples ranges from 0.25 to 0.75. Transmission and reflection properties as a function of incident angle were measured using Fourier transform infrared spectroscopy and our experimental data is in good agreement with modeling using effective medium theory. We will also show the results from a beam optics experiment which demonstrates that our materials exhibit negative refraction. These single-material structures are easy to grow and eliminate the need for the integration of drastically different materials systems. They show great potentials for creation of infrared devices such as hyperlenses and enhanced detectors.

Type-II strained layer superlattices (SLS) are an active research topic in the MBE community, and applications for SLS detectors continue to grow. SLS detector technology has already reached the commercial market because of improvements to material quality, device design, and device fabrication. Despite this progress, the optimal superlattice design has not been agreed upon, and at various times has been believed to be InAs/GaSb, InAs/InGaSb, or InAs/InAsSb. Building on these, in this talk, we investigate the properties of a new mid-wave infrared SLS material: InGaAs/InAsSb SLS.

The ternary InGaAs/InAsSb SLS has two main advantages over other SLS designs: greater support for strain compensation and enhanced absorption due to increased electron-hole wavefunction overlap. Here, we compare three ternary SLSs, with approximately the same bandgap (0.240 eV at 150 K), comprised of Ga fractions of 5%, 10%, and 20% to a reference sample with 0% Ga. Enhanced absorption is both theoretically predicted and experimentally realized. Furthermore, the characteristics of ternary SLS infrared detectors based on an nBn architecture are reported and exhibit nearly state-of-the-art dark current performance with minimal growth optimization. Standard material characterization information, including photoluminescence, x-ray diffraction, and atomic force microscopy are provided.

The device performance is further analyzed using more advanced techniques to understand the material quality. Using a mixture of dark current, external quantum efficiency, time-resolved photoluminescence, and electron beam induced current, we ultimately show that the minority carrier lifetime decreases slightly with gallium content, while the vertical hole mobility increases substantially. These counteracting trends lead to longer diffusion lengths and higher performance with larger Ga fractions. Further improvements in both doping and deposition should lead to even higher performance ternary InGaAs/InAsSb SLS devices.

Lattice-matched heterovalent structures, such as Group III-V/Group II-VI heterostructures, offer a materials platform with unlimited degrees of freedom for monolithic integration of many electronic and optoelectronic devices onto a single substrate without a high density of misfit dislocations. The potential of these heterovalent structures is perhaps best demonstrated with the case of a HEMT structure consisting of an InAsSb conduction channel (high electron mobility, narrow bandgap) and a ZnTe barrier (semi-insulating, wide band gap), both of which are lattice-matched to GaSb (high hole mobility, widely available substrates). Growth of the InAsSb channels confined by the ZnTe barrier layers on GaSb substrates is achieved using a dual-chamber MBE system with dedicated II-VI and III-V growth chambers. The influence of the interface configuration, flux ratio and growth temperature on the bonding arrangement and crystal structure of the epilayers are examined in situ with RHEED and compared with TEM and XRD data to find the optimal parameters for InAsSb growth on ZnTe, and for ZnTe on InAsSb/GaSb. The buildup of charge from the non-isovalent bonds at the interface is being characterized using electron holography.

The effects of carrier transport on the temperature dependence in InAs/GaAs quantum dot solar cell (QDSC) were investigated. Each of the SCs consists of a 300-nm n⁺-GaAs layer (~10¹⁸/cm³), a 1.5-μm n-GaAs absorption layer (~10¹⁷/cm³), 0.6-μm p⁺-GaAs (~10¹⁸/cm³) layers, a 50-nm p⁺-Al_0.9Ga_0.1As window, and a 10-nm p⁺-GaAs contact layer. InAs/GaAs QDSC contains eight InAs QD layers located in the n-GaAs absorption layer region of a p⁺-n-n⁺ junction. In addition, a GaAs single-junction SC without InAs QDs was also grown as a reference. The cells growth and structures have been described in detail. At each applied reverse bias, the C-V curves of GaAs reference SC show an increase in capacitance with increasing temperature due to the statistical shift of the Fermi level in GaAs host material, reducing the built-in potential and thus the width of the space charge region with increasing temperature. However, InAs/GaAs QDSC showed the reduced capacitance and background carrier concentration (N_d) compared to that of the GaAs reference SC. These indicate that charge carriers in the active region are seriously affected by InAs/GaAs QDs due to the re-capturing and trapping process in the QDs and the strain-induced defect states. The three distinct behaviors depending on voltage range were revealed by the J-V curves of GaAs reference SC. To analyze the effect of carrier transport in the active region, the modified 3-diode model is applied in this SC structure consisting of diffusion, recombination, and leakage current density. This leakage current density is related to the carrier losses caused by the defects state induced by low temperature (LT)-growth process (470 °C) of the 320 nm thick n-GaAs absorption layer for the reference GaAs SC. However, the five distinct behaviors depending on voltage range were revealed by the J-V curves of QDSC due to the additional generation of the leakage current density through the re-capturing and trapping by QD and strain-related defect states. The leakage current density caused by the trapping process in the defect states increase with increasing temperature while the leakage current density caused by the re-capturing process in the QD states was no different with increasing temperatures. These results can be explained by the balance between carrier escaping and re-capturing process in the QD and defect states. As increasing temperature, the re-captured or trapped carrier in the QD and defect states can escape using the high thermal energy. However, this de-trapped carrier in the defect states can be re-captured by the QD states. Due to the increase of the re-captured carrier in the QD states, the leakage current density caused by the confined QD states was no different with various temperatures.

Due to their significant band-gap narrowing with minimal change in lattice parameter, dilute nitride semiconductor alloys are useful for a variety of applications, including long-wavelength lasers and detectors, ultra-high-efficiency solar cells, and high performance heterojunction bipolar transistors. However, N-related point defects often contribute to carrier scattering and recombination, leading to degraded carrier mobilities and optical efficiencies. For GaAsN and related alloys, co-alloying with larger group V elements such as Sb or Bi is expected to lead to significant energy bandgap narrowing using a substantially lower N fraction, and a correspondingly lower concentration of N-related defects. For GaAsN, several groups have suggested that N shares an arsenic site with either arsenic or another N atom, often termed (N-As)_As or (N-N)_As split interstitials. In the case of GaAsNBi, the published experimental work has focused primarily on growth parameters and optical properties, without addressing the mechanisms for N and Bi co-incorporation during epitaxy. To identify N-related defects, we compare channeling Rutherford backscattering and nuclear reaction analysis spectra with Monte Carlo-Molecular Dynamics simulations along the [100], [110], and [111] directions. Using this combined computational-experimental approach, we have identified (N-As)_As is the dominant interstitial GaAsN. For GaAsNBi, a comparison of NRA and x-ray rocking curve analyses reveals enhanced N incorporation in the presence of Bi. Furthermore, ion channeling measurements reveal non-substitutional incorporation of the extra N. We are currently considering the nature of the N interstitial complexes in GaAsNBi, including the relative concentrations of (N-N)_As, (N-As)_As, and N_tetrahedral.