Symposium Organizers
Osamu Ueda, Kanazawa Institute of Technology Graduate School of Engineering
Mitsuo Fukuda, Toyohashi University of Technology
Kenji Shiojima, University of Fukui Graduate School of Electrical and Electronics Engineering
Edwin Piner, Texas State University, San Marcos
Symposium Support
Japan Society of Applied Physics
G2: LEDs and Crystal Growth
Session Chairs
Tuesday PM, April 10, 2012
Moscone West, Level 2, Room 2007
2:30 AM - G2.1
Burning Effects of Graphene-based Transparent Conductive Electrode in GaN-based UV LEDs
Byung-Jae Kim 1 Jihyun Kim 1
1Korea University Seoul Republic of Korea
Show AbstractRecently, graphene has been widely investigated for the application in sensors, light-emitting diodes (LEDs) and solar cells because of its excellent properties such as good electrical and thermal conductivity, stability and high transmittance from UV to near IR. Especially, graphene can be served as the transparent conductive electrode in optical device ranging from UV to IR. Previously, we demonstrated UV LEDs with large-area graphene-based transparent conductive electrode. Large-area graphene layer was grown by chemical vapor deposition method. UV light at a peak of 372nm was emitted via the whole graphene layer, confirming the current spreading effects of graphene layer. However, graphene layer degraded after high power injection. The light emitting area was drastically decreased after continuous high power operation under an injection power of 20mW for 60 seconds. The thickness of graphene layer, which was measured by the micro-Raman spectroscopy, was reduced from 4-layer to 2-layer due to the burning effects. The self-heating of graphene occurred by the continuous high power operation. The heat in graphene layer activated a local oxidation between C-C bonding and O2 in the air. Therefore, the thickness of graphene layer was reduced by producing CO2 gas. Then, the effects of heating were studied by the pulse operation of UV LEDs. Then, the effects of oxidation were characterized by the continuous high power operation under vacuum and Ar gas environments. In this case, the burning effects of graphene layer were drastically reduced under O2-free ambient. The properties of graphene layer before/after the high power operation of UV LEDs were studied with the micro-Raman spectroscopy, current-voltage characteristics and optical images.
2:45 AM - G2.2
Local Electronic Structure and UV Electroluminescence of n-ZnO:N/p-GaN Heterojunction LEDs Grown by Remote Plasma Atomic Layer Deposition
Chien Jui-Fen 1 Chen Ching-Hsiang 2 Shyue Jing-Jong 1 3 Chen Miin-Jang 1 4
1National Taiwan University Taipei Taiwan2Protrustech Corporation Limited Tainan Taiwan3Academia Sinica Taipei Taiwan4National Taiwan University Taipei Taiwan
Show AbstractNitrogen-doped ZnO (ZnO:N) films have been prepared by remote plasma atomic layer deposition (RP-ALD) and treated by rapid thermal annealing (RTA) in oxygen atmosphere. The local electronic structures of the (ZnO:N) films are investigated by X-ray photoelectron spectroscopy (XPS) and X-ray absorption near edge spectroscopy (XANES) at the O K-edge, N K-edge, and Zn LIII-edge, respectively. The XPS reveals the presence of the Zn-N bond in the ZnO:N films, indicating that partial amounts of O sites are occupied by N species. This is correspondent with the decrease of electron concentration in the ZnO:N films with the nitrogen doping concentration, as indicated by the Hall effect measurement. The XANES shows that the valence state of Zn species increases to higher energy and the N-O bonding is formed after the incorporation of nitrogen, revealing that both Zn and N are preferentially surrounded by O to achieve a charge balance. The RP-ALD technique is applied to fabricate the n-type ZnO:N/p-type GaN heterojunction LEDs. Dominant ultraviolet electroluminescence at 370 nm from the ZnO:N layer is achieved at room temperature.
3:00 AM - G2.3
An Easy Approach to Identify Dislocation Types in GaN Films through Selective Chemical Etching and Atom Force Microscopy
Guoqiang Li 1 Hui Yang 1
1South China Univ. of Technology Guangzhou China
Show AbstractGaN has been intensively used for blue and green optoelectronic devices and high-speed electronic and power devices. However, defects, especially dislocations, play a critical role in determining the performance of these devices. Studies have revealed that different dislocation types have different influences on GaN device performance. Identification and discrimination of dislocation types in GaN films will be very helpful to understand the formation mechanism of each type of dislocation, and subsequently to work out corresponding methodologies to reduce each. Transmission electron microscopy is an effective means to study dislocations in GaN films, but requires extensive and skillful sample preparation. Wet chemical etching along with atomic force microscopy (AFM) or scanning electron microscopy (SEM) would be much cheaper and time-effective for analysis of dislocations in GaN. Although quite a few etchants were deployed for this purpose, they usually led to unrepeatable sizes and morphologies of dislocation pits. In this paper, we developed a new etchant, i.e., mixed molten KOH and NaOH with a specific mass ratio between, which provides very stable and interpretable results for etched pits of dislocations in terms of their size and morphology, and therefore is a good approach to identify the dislocation types in GaN films. Device-ready GaN-based LED epitaxial layers were grown on sapphire substrates by metal-organic chemical vapor deposition. LED wafers were cut into small pieces (1mmÃ-3 mm) and then etched by mixed molten KOH and NaOH with a specific mass ratio between. During etching, the mixed etchant was kept in a nickel-crucible at 390C in a thermostatic oven. We applied different etching durations of 20, 40, and 60 min, respectively in order to study the development of dislocation morphology during etching. Etched pit morphology of dislocations was carefully studied using AFM surface analyzer, and crystallinity was characterized by XRD before and after etching. Two types of etched pits, triangle and hexagonal, were discovered. Triangle ones were more quickly generated, but would never transfer to the other even with the increase in etching duration. Theoretical analysis from the viewpoint of strain energy of defects lets us to conclude that the former is from screw dislocation and the latter is formed by mixed dislocation. We consequently obtained the dislocation densities of each type, which agreed well with the XRD results. Detailed study also revealed that etching is anisotropic. Lateral etching rate is about two orders of magnitude higher than vertical etching rate, which we believe is due to the polarity effect of c-plane GaN film in its growth direction (vertical). In a word, this work presents an easy but reliable approach to determine the type and corresponding density of dislocations in GaN.
3:15 AM - G2.4
Influence of Sb Overpressure on the Anisotropic Transport Properties of InAs/AlSb Heterostructures Grown on Highly Mismatched Substrates
Salim El Kazzi 1 Ludovic Desplanque 1 Yi Wang 2 Pierre Ruterana 2 Xavier Wallart 1
1IEMN Villeneuve d'Ascq France2CIMAP Caen France
Show AbstractThe high electron mobility InAs/AlSb heterostructure is very promising for low power millimeter-wave circuits because of the large conduction band offset between InAs and AlSb, high peak electron velocity, and high electron concentration in the InAs channel. However, the large lattice mismatch between III-Sb and the standard semi-insulating substrates (GaAs, Si) induces a high density of threading dislocations (TDs) that can be highly detrimental to device performances. Moreover, anisotropic electron transport properties related to the presence of TDs have been reported previously in InAs/AlSb HEMTs grown on a (001) InP substrate using an AlSb buffer. To overcome this problem, many studies have shown that the formation of a periodic array of 90° misfit dislocations at the epilayer-substrate interface can help in reducing TD density (106 cm-2) with an almost complete strain relaxation (98%). In this context, we investigate the optimized growth conditions for the nucleation of III-Sb layers on highly mismatched GaAs and GaP substrates. More precisely, we use AFM, TEM and XRD measurements to study the influence of the V/III ratio on the formation of the 90° misfit dislocation periodic array at the AlSb/GaAs and GaSb/GaP interfaces. These measurements are correlated with electronic transport measurements in InAs/AlSb heterostructures on both GaAs and GaP substrates using AlSb and GaSb buffers respectively. The results indicate that a low Sb overpressure during the initial nucleation stage of the III-Sb layer leads to a reduction of the anisotropic transport properties in the high electron mobility heterostructures. For the GaSb/GaP system, these growth conditions correspond also to a change in the shape of the islands that are initially formed during the nucleation process: [110]-elongated and tetragonal-shaped islands are formed with a high Sb overpressure whereas isotropic and elliptical-shaped islands are formed with a low Sb overpressure. These results suggest that the nucleation process directly impact the transport properties in the InAs/AlSb heterostructure and that a careful optimization of the Sb overpressure during this initial stage of the growth can greatly improve the electron mobility in both crystallographic orientations.
3:30 AM - G2.5
Self-catalyzed Growth of InAs1-xPx Nanowires on Si by Molecular Beam Epitaxy
Marina Panfilova 1 Ivan Isakov 1 Marion J. L. Sourribes 1 Huiyun Y Liu 2 Paul A Warburton 1
1University College London London United Kingdom2University College London London United Kingdom
Show AbstractSemiconductor nanowires are under intensive study due to their intriguing features and great potential for electronic and optoelectronic devices [1,2], among other applications [3-5]. By modulating the electronic and optical properties of the nanowire through assembled heterostructures along its length, a number of devices with zero-dimensional behavior can be developed. Of particular interest are III-V semiconductors on Si, exploiting the mature Si technology and superior optoelectronic and electronic properties of III-V compounds. While there have been many studies on the growth of InAs and InP nanowires by metal-organic vapor phase epitaxy either with [6] or without [7] the use of Au particles as catalyst, there are only a few studies on self-catalyzed growth of such nanowires by molecular beam epitaxy (MBE) [8]. Furthermore a variation of the P content in InAs1-xPx nanowires enables bandgap modulation. However there are no reports about self-catalyzed ternary InAs1-xPx nanowires grown by MBE. Here we present self-catalyzed growth of free-standing InAs1-xPx nanowires with different P content on Si (111) substrates by MBE. The introduction of an ultrathin amorphous SiOx layer on Si (111) and substrate annealing under As-flux was found to be essential for epitaxial InAs1-xPx nanowire growth. The as-grown nanowires were investigated by scanning electron microscopy for growth selectivity and nanowire geometry. The nanowires exhibit predominantly vertical directionality related to the {111} family of orientations, indicating a direct relationship to the underlying Si (111) substrate. The diameter of the nanowires remains nearly axially constant at about 60 nm. We observed a density variation of nanowires in the range of ~3 to 10 µm-2 and a length variation of between 2 and 5 µm for different growth parameters. The P composition of InAs1-xPx nanowires was inferred from the measured lattice constant, using both high resolution X-ray diffraction (HRXRD) (by application of Vegardâ?Ts law) and energy-dispersive X-ray spectroscopy. HRXRD omega-2theta scan showed reflections associated with cubic zinc blende (111) and hexagonal wurzite (002) crystal phases. To obtain further insight into the morphology and structure of the nanowires transmission electron microscopy measurements were carried out. We found that our nanowires have a mixed crystal structure ranging from wurzite to zinc blende with high densities of twin planes and stacking faults. 1. Y. Huang et al, Science 294, 1313 (2001) 2. Y. Huang et al, Pure Appl. Chem. 76, 2051 (2004) 3. C. Colombo et al, Appl. Phys. Lett. 94, 173108 (2009) 4. A.I. Boukai et al, Nature (London) 451, 168 (2008) 5. G.F. Zheng et al, Nat. Biotechnol. 23, 1294 (2005) 6. S. Roddaro et al, Nanotechnology 20, 285303 (2009) 7. K. Tomioka et al, Nano Lett. 8, 3475 (2008) 8. G. Koblmüller et al, Nanotechnology 21, 365602 (2010)
3:45 AM - G2.6
Enhanced Light Absorption in Textured Metal Organic Chemical Vapour Deposited (MOCVD) CdO Thin Films
Sunday O Adekoya 1 Marcus Adebola Eleruja 1 Bolutife Olofinjana 1 Olumide O Akinwunmi 1 Bidini A Taleatu 1 Ezekiel. O B Ajayi 1
1Obafemi Awolowo University Ile-Ife Nigeria
Show AbstractThe Single Solid Source precursor Cadmium Acetylacetonate, Cd[C5H7O2]2 was prepared and Cadmium Oxide thin films were deposited on Sodalime glass substrate using Metal Organic Chemical Vapour Deposition (MOCVD) technique at deposition temperature of 420 C . The films were characterized using Scanning Electron Microscopy (SEM) with Energy Dispersive X-Ray (EDX) facility attached to It, X-Ray Diffractometry (XRD) and UV-visible spectrometry. SEM micrographs showed the formation of textured surface with identifiable cubic and hexagonal structures, having average grain size greater than 1 µm. XRD studies indicated the formation of polycrystalline cubic CdO phases with preferred orientation in (111) plane. A direct optical band gap of 2.30 eV was obtained from the analysis of the UV-visible spectrum with an enhanced light absorption in 650 â?" 1000 nm spectrum range. This enhancement is as a result of light trapping by the textured crystalline structure. *Corresponding author- Prof. E.O.B. [email protected]
G3: Crystal Growth
Session Chairs
Tuesday PM, April 10, 2012
Moscone West, Level 2, Room 2007
4:30 AM - *G3.1
Metamorphic and Non-conventional `Bufferrsquo; Layers for Materials Integration
Thomas F. Kuech 1 Luke J Mawst 2 Susan E Babcock 3 Paul N Nealey 1 Tung-Sheng Kuan 4
1University of Wisconsin - Madison Madison USA2University of Wisconsin -Madison Madison USA3University of Wisconsin -Madison Madison USA4State University of New York at Albany Albany USA
Show AbstractAn increasing number of devices and device platforms require the use of multiple, lattice-mismatched materials. The integration of these materials typically is accomplished through the inclusion of â?~bufferâ?T layers during a heteroepitaxial growth process. Buffer layers are intermediate layers within a growing structure that allow a chemical or structural transition between two epitaxial materials. The choice and design of these layers depend critically on the nature of the material transition. GaN on Si is perhaps an extreme example where cross-doping, interfacial compound formation, change in crystal structure and a large lattice mismatch conspire to complicate the growth. AlN buffer layers are often used in this application to limit these deleterious interactions. More conventionally however, the change in lattice parameter is the dominant factor to be accommodated. Incorporation of compositionally-graded layers has been the approach most often used in this context. As the composition is altered from that of the substrate or a lattice matched compound to the substrate to a material which has the desired lattice parameter, the structurally required mismatch dislocations are introduced with a minimum of residual threading dislocations. The approach of transitional or graded layers has undergone continual development with new effects being realized to reduce the buffer layer thickness to achieve the dislocation density needed in the device layer. However, the surfaces of these graded layers often are too rough to use in device applications due to the presence of a surface cross-hatch derived from the mismatch dislocations. The use of chemical-mechanical polishing has offered a means to recover the required planar surface. The re-initiation of heteroepitaxial growth on these polished surfaces raises new questions concerning of defect initiation and growth and our recent results will be presented. Alternative approaches to the growth of graded buffer layers are beginning to be studied and utilized. The intent of all these approaches is to generate the required misfit dislocations, have them reside at an appropriate interface and, finally, leave no threading dislocation line segments within the thickness of the film. A general approach to eliminating the threading line segments and to reduce the buffer layer thickness based on a self-assembled block co-polymer approach to nanoscale patterning will be presented. This approach offers rapid and cost-effective full wafer patterning at the 20-nm length scale to achieve improvements in heteroepitaxial growth of large lattice mismatched materials. Investigations of crystal quality in films grown on such buffer layers indicate that comparable defect mitigation can occurs in dramatically thinner buffer layers compared to â?~conventionalâ?T graded buffer layers.
5:00 AM - G3.2
Fabrication of High Quality Free Standing GaN Film by Nano Lift-off
Yuefeng Wang 2 5 Liang Tang 1 5 Timothy Sands 2 3 5 Michael J Manfra 1 3 5 Gary Cheng 4
1Purdue University West Lafayette USA2Purdue University West Lafayette USA3Purdue University West Lafayette USA4Purdue University West Lafayette USA5Purdue University West Lafayette USA
Show AbstractGaN is one of the most widely used semiconductor materials in optoelectronic devices, high speed transistors, power electronics and biocompatible devices. Compared to GaN on sapphire and SiC substrate, free-standing GaN films have numerous exclusive benefits such as strain relieve, light extraction, defect reduction and potential flexibility. While the traditional Laser lift-off (LLO) was developed to separate GaN from sapphire substrates, the technique has a limitation of only being able to lift off GaN film with thickness greater than 10um. In this work, we present a novel method able to fabricate thin film GaN suitable for electronics device application with a thickness greater than tens of nanometers. Starting from GaN on sapphire substrate, MBE or MOCVD can be used to deposit a epitaxial sacrificial layer, which is a 20-30 periodic superlattice of alternating sub-nanometer thin InN and GaN. The average indium composition was examined by XRD and EDS, which can be control from 20%-35% by selecting different growth parameters. The sacrificial layer is lattice matched to the GaN substrate, enabling high quality GaN epilayer growth to an arbitrary thickness. Post-growth, pulsed laser with photon energy of 2.3eV,pulse width around 10ns and fluences ranging from 100mJ/cm^2 to 500mJ/cm^2 was irradiated upon the sample which leads to the controlled decomposition of sacrificial midlayer and followed by easy separation of the top GaN film and substructure by adhesive transfer. This technique can be used to fabricate GaN films with low defects and a wide range of thickness, especially sub-um level. TEM, AFM and FESEM were used to characterize the crystallinity and morphology of the freestanding film. A numerical laser-solid interaction model was developed to optimize the lift-off process. The nano lift-off process can also be used in fabricating flexible GaN thin film. A lamellar structure was developed to demonstrate the possibility of multiple films nano lift-off on single substrate, which substantially reduced the cost and improved the crystal quality of freestanding GaN film. Overall, the recently developed nano lift-off process is a promising pathway to nanoscale GaN freestanding devices and high quality GaN substrates.
5:15 AM - G3.3
Effect of the N/Al Ratio in the Gas Phase at Constant Supersaturation on AlN Epitaxy on Sapphire by HTCVD
Baccar Nour Elhouda 1 Boichot Raphael 1 Blanquet Elisabeth 1 Pons Michel 1
1SIMaP Saint Martin d'Hegrave;res France
Show AbstractAluminum Nitride (AlN) is a III-V wide band gap semiconductor material with a high thermal conductivity and good chemical inertia. These properties make it suitable for several applications in the opto-electronic field, the most important is the fabrication of UV diodes. Thick AlN layers have been processed by HTCVD (High Temperature Chemical Vapor Deposition) using AlCl3 and NH3 diluted in H2. The thermodynamic and kinetic modeling of AlN growth on AlN templates at different temperatures and N/Al ratios have been made in previous studies [1-2]. The conclusion of (Boichot and al. [1]) was that experiments should be conducted at constant supersaturation and thickness to understand the actual influence of N/Al ratio on AlN layer quality. The relationship between temperature and supersaturation was established by (Claudel and al. [2]). In the first part of the study, experiments were performed on graphite substrates to study the preferential orientation of AlN crystals by varying the temperature and N/Al ratio. It is demonstrated that low N/Al ratio allows the control of growth orientation of the AlN crystal facets along the c-axis. In the second part, experiments were carried out on c-plane sapphire substrates to investigate the effect of supersaturation and N/Al ratio on the morphology of grown layers. The layers are characterized by SEM, XRD, XPS and SIMS to evaluate both crystal quality and O and Cl contamination. The main conclusion is that the control of surface quality in term of epitaxial relationship between sapphire and h-AlN is possible through a precise setting of supersaturation and N/Al ratio in the gas phase. It is also concluded that epitaxial growth along c-axis is not thermodynamically favored in classical CVD conditions (low pressures, high temperature and N/Al up to 1).
5:30 AM - G3.4
A Detailed Investigation into Strain-coupled Bi-layer InAs/GaAs Quantum Dots: Impact of Variation in Deposition Rate, Barrier Thickness and Monolayer Coverage
Nivedita Basu 1 Subhananda Chakrabarti 1
1Indian Institute of Technology Bombay Mumbai India
Show AbstractSignificant effort has been vested in studying the growth and properties of InAs/GaAs quantum dots (QDs) for their potential application in optoelectronic devices. Long wavelength (â?¥1.3µm) emission for important optoelectronic applications, is arduous to achieve with a single layer of QDs grown by Stranski-Krastanov method due to inherent shortcomings like dot size inhomogeneity and low dot density. Bilayer QD heterostructure, consisting of bottom (seed) QD layer separated from top (active) layer by thin GaAs spacer layer, taking advantage of the strain fields assorted with seed QD layer, is an obvious answer. A detailed and systematic investigation of QD bilayer heterostructures undertaken in this work would be helpful for determining optimization conditions while fabricating long wavelength laser and other optoelectronic devices. We have studied temperature-dependent (8K â?" 300K) photoluminescence (PL) spectra from self-assembled InAs/GaAs QD samples by varying different growth parameters- (i) growth rate (ML/s) of active dot layer, (ii) deposition amount (ML) of InAs on active dot layer while keeping that on seed layer constant, and (iii) GaAs spacer layer (SL) thickness (â"«). Slower growth rate (~0.03ML/s) improves dot-size uniformity evident from decreased linewidth (~31nm compared to ~89nm for ~0.3ML/s) and emission wavelength increases (~1.3µm from ~1µm) due to increased quantum dot size. For faster growth rate (~0.3ML/s), the dots on active layer are smaller. Hence there is less probability of carrier escape due to lateral coupling among neighboring QDs and we have higher activation energy (~25meV increase). Larger amount of InAs deposition (3.2ML) on active layer results in increased emission wavelength (~1.3µm from ~1.1µm for 0ML) due to increased dot size. FWHM decreases (~35nm compared to 0ML situation) due to increase in dot size uniformity. Initially, as the InAs deposition is increased no misfit dislocation is introduced and the self-assembled QDs are considered defectâ?"free. But as InAs coverage increases, island coalescence occurs and defect-states are introduced. Hence, activation energy lowered (~60meV compared to 0ML situation) for larger deposition amount on active dot layer. GaAs SL thickness variation shows that strain field propagating from seed layer to active layer for thin SLs (â?¥100â"«) is more prominent causing bigger QDs in active layer which in turn gives better optical properties. However, for even thinner SLs, the optical properties seem to deteriorate which may be attributed to strain induced migration of native defects to the vicinity of QDs. Activation energy monotonically decreases with increasing spacer layer thickness (~45meV for variation of 125Ã.) which validates that improved vertical ordering is obtained between seed layer and active layer QDs due to propagation of strain from seed layer to active layer, preventing lateral intermixing of energy states among QDs in active layer. Acknowledgement: DST India.
5:45 AM - G3.5
Integration of II-VI Materials on GaSb Substrates
Kevin Doyle 1 2 Gregory Brill 2 Jessica Chai 1 Kyoung-Keun Lee 1 Jack Dinan 1 Tom Myers 1
1Texas State University San Marcos USA2Army Research Lab Adelphi USA
Show Abstract
Recent developments in GaSb substrates open up new possibilities for epitaxial growth of II-VI devices. Mismatch between the substrate lattice and the growth material creates dislocations in the growth lattice, which can act as recombination centers for minority charge carriers and degrade overall device performance. However, GaSb belongs to an entire class of materials with lattice constants near 6.1Ã.. ZnTe and CdSe are II-VI materials that are being considered as part of cascade solar cell materials, and both are nearly lattice-matched to GaSb. Another material in this class is HgCdSe, which could be an alternative to the traditional IR detector HgCdTe. Like HgCdTe, HgCdSe is tunable across the IR spectrum. However, the only substrate lattice-matched to HgCdT, CdZnTe, is both expensive and unavailable in a large enough size to form a focal plane array (FPA). HgCdSe, however, is very closely lattice-matched to GaSb which is available in a large enough size for FPA production. Additionally, the cadmium-selenide bond appears to be stronger than the cadmium-telluride bond, as HgCdSe samples grown by molecular beam epitaxy lack the â?ovoidâ? defects common to HgCdTe. The slight lattice mismatch between HgCdSe and GaSb can be alleviated with a ZnTeSe or CdTeSe buffer layer. We have performed preliminary studies into the growth of lattice-matched ZnTe1-xSex on both (100) and (211)B GaSb. The effects of substrate orientation, substrate temperature, and growth conditions on deposition of ZnTe0.99Se0.01 alloys were investigated. The lattice-matching condition yielded a minimum rms roughness of 1.1 nm, a rocking curve FWHM value of ~ 29 arc seconds, and a density of non-radiative defects of mid-105 cm-2. Several obstacles remain before HgCdSe FPAs can be produced. Currently all HgCdSe samples have electron concentrations greater than 1017 cm-3 according to Hall effect measurements, despite being nominally undoped. This could be due to intrinsic impurities in the lattice, such as Hg interstitials or Se vacancies, or an external dopant such as Br introduced in the growth process. Once the source of these background electrons has been determined and samples with low dislocation densities have been grown, HgCdSe will be confirmed as a viable alternative to HgCdTe in producing FPAs for IR detection.
G1: Laser Reliability and Defects
Session Chairs
Tuesday AM, April 10, 2012
Moscone West, Level 2, Room 2007
9:30 AM - *G1.1
Reliability of Semiconductor Lasers Used in Optical Fiber Communication
Nobuyuki Ikoma 1
1Sumitomo Electric Industries Yokohama Japan
Show AbstractSemiconductor lasers have been widely used in optical fiber communication as light sources. Reliability of lasers used in communication systems is very important and many efforts have been paid for improvement of it. Recently, rapid growth of internet data traffic makes modulation speed of lasers higher and data rate becomes 10Gbit/s or higher. Relaxation oscillation frequency (fr) of lasers is a limiting factor of high speed operation. One method to increase fr is laser operation under high current and high output power condition. Because laser degradation is enhanced by current and output power, higher reliability of lasers are required. Another method to increase fr is to enlarge optical confinement in an active layer. Light absorption in an active layer increases as optical confinement increases and degradation caused by light absorption at a facet such as forward-biased electrostatic discharge (ESD) breakdown is promoted. To suppress ESD breakdown, facet passivation by dielectric coating is effective. Also, reduction of optical density at a facet by lowering reflectivity can suppress ESD breakdown. In addition to high speed, high temperature operation without thermoelectric cooler is required for low cost and low power dissipation. For modulation of 10Gbit/s or higher data rate, the AlGaInAs/InP material system is more suitable than InGaAsP/InP system, which has been widely used in optical fiber communication, because of high fr at high temperature of AlGaInAs based quantum wells. In a viewpoint of reliability, AlGaInAs is less tractable than InGaAsP because of its susceptibility to oxidation and less durability to optical intensity. Therefore, facet passivation and reduction of optical density at a facet are more important than the case of InGaAsP/InP lasers. We found that a dislocation network, which was not observed in an InGaAsP layer, was formed in an AlGaInAs active layer at the facet [1]. To suppress formation of a dislocation network, stress control at the facet by dielectric coating is very important. Highly reliable AlGaInAs/InP lasers operating at 10Gbit/s have been realized and widely used. To modulate at higher than 10Gbit/s, operating current should be higher to obtain higher fr, therefore laser degradation is more enhanced. It seems to be a big challenge that to obtain highly reliable lasers operating at higher than 10Gbit/s and high temperature. [1] H. Ichikawa et al. J. Appl. Phys. 107, 083109 (2010).
10:00 AM - *G1.2
Grown-in and Process-induced Defects in GaN-based Materials and Optical Devices
Shigetaka Tomiya 1
1Sony Corporation Atsugi Japan
Show AbstractReduction of defects in optical devices is critical important for obtaining high efficient and high reliable performance. In this presentation, grown-in defects and process-induced defects in GaN-based materials and optical devices are discussed. A variety of structural defects have been observed in GaN-based materials. First, four different types of structural defects are described based on our previous studies. Such types of structural defects are threading dislocations, Mg-related pyramidal defects, inversion domains at GaInN active layers and V-shaped pits. We, then, discussed these defects in relation to reliability of GaN-based laser diodes (LDs). GaN-based materials are chemically inert and have higher bonding energies than conventional III-V materials. Since wet etching can be limited to apply to the GaN-based material, plasma etching process is inevitable to effectively fabricate the GaN-based devices. Plasma-induced damage (PID) has been investigated extensively in Si-based devices and is categorized into three mechanisms. However, there have been few studies on mechanism of the PID in GaN-based materials. In second half of this presentation, we describe PID observed in GaInN quantum well (QW) and discuss deterioration mechanisms.
10:30 AM - *G1.3
Highly Reliable 1060nm Vertical Cavity Surface Emitting Lasers (VCSELs) for Optical Interconnect
Shinichi Kamiya 1 Keishi Takaki 2 Suguru Imai 2 Junji Yoshida 2 Masaki Funabashi 2 Yasumasa Kawakita 2 Koji Hiraiwa 2 Toshihito Suzuki 2 Hitoshi Shimizu 2 Naoki Tsukiji 2 Takuya Ishikawa 1 Akihiko Kasukawa 2
1Furukawa Electric Co., Ltd. 2-4-3, Okano, Nishi-ku, Yokohama Japan2Furukawa Electric Co., Ltd. 6, Yawata-Kaigan Dori, Ichihara Japan
Show Abstract
High reliability, low power consumption and high speed laser diodes are required for optical interconnect. We developed 1060nm VCSELs with InGaAs/GaAs strained quantum wells, oxide-confined and double intra-cavity structures for that purpose. As for the power consumption, low power dissipation of 0.14 mW/Gbps at 10 Gbps operation has been achieved. Clear eye openings up to 20Gbps were confirmed at a low bias current of 5mA. In the reliability test, accelerated aging tests were performed up to 5,000 hours at 6mA in three different temperatures, 70 oC, 90 oC and 120 oC. The total number of the VCSELs was 4,898 pcs (approximately 5,000). No failure was observed. Under the normal operating condition of 40 oC and 6 mA, the total device-hours was 7.75Ã-107 hours assuming Ea = 0.35 eV according to Telcordia GR-468-CORE. The random failure rate of 30 FIT with the confidence level (C.L.) of 90 % and 12 FIT with the C.L. of 60 % were estimated. To estimate the wear-out lifetime and the number of FITs, high stressed aging tests with 170 oC and 6 mA were performed. With the acceleration factor of Ea = 0.7 eV in the wear-out failure, the median lifetime was 3,000 hours which was equivalent to 300 years in 40 oC ambient. The FIT numbers due to the wear-out were estimated as 0.3 FIT for 10 years. Compared with the random failure rate of 30 FIT, the wear-out failure rates are considered to be negligible. In the extremely long term aging test with 90 oC and 6 mA, no wear-out trend has been observed in both threshold current and optical power up to 20,000 hours operation. These results indicate that 1060 nm VCSEL is promising light source used in optical interconnect for high performance computers and data centers.
11:30 AM - *G1.4
Reliability of Vertical-cavity Surface-emitting Lasers
Robert W. Herrick 1
1JDSU San Jose USA
Show AbstractVertical-cavity surface-emitting lasers (VCSELs) have become one of the most popular forms of semiconductor lasers, with an estimated number sold exceeding 200 million. They are primarily used in low cost data links, and for pointing devices like the laser mouse. However, as they are normally made with GaAs quantum wells, they are vulnerable to the growth of arrays of climb dislocations, which have been the most common field failure mechanism. These are also known as â?odark line defectsâ? or â?oDLDsâ?. In this paper, we will briefly discuss the six most common failure mechanisms seen industry-wide. These are broadly divided into â?omaverick failure mechanismsâ? â?" usually failure on a subset of lasers due to climb dislocations â?" and â?owearout failure mechanismsâ?, that cause all devices on the wafer to fail prematurely. We will discuss the way companies organize their product qualification and on-going testing to detect potential problems, and prevent them from reaching customers. We also will briefly describe methods of analyzing failures to determine the root cause. The author will share his experience over the past 17 years working with VCSEL reliability at nearly all of the large American VCSEL manufacturers.
12:00 PM - G1.5
Thermomechanical Modelling of High Power Laser Diode Degradation
Alonso Martin-Martin 1 Jorge Souto 1 Pilar Intilde;iguez 1 Juan Jimenez 1
1Universidad de Valladolid Valladolid Spain
Show AbstractThe understanding of the degradation mechanisms of high power laser diodes is critical to improve their power and reliability. The degradation occurs because of the generation of crystal defects during the laser operation. The identification of these defects, as well as the understanding of the causes that contribute to their formation, are crucial for the improvement of the technological processes leading to reliable laser diode devices. We focus here on high power AlGaAs based centimeter laser bars (808 nm). Dark line defects (DLDs) inside the cavity are observed by cathodoluminescence (CL). The rapid degradation of the laser diodes is due to the generation and propagation of large defects under laser operation. The generation of those defects involves atomic bond breaking, which points to a paramount role of strain. One has to distinguish between the residual strain associated with different technological processes, and the strain generated during the laser operation. The later has a thermomechanical origin, and must be high enough to accounting for the plastic deformation. A thermomechanical model is developed, in which one solves using finite element methods (fem) the thermal transport equation, and the thermal stresses induced by very local heating sources during the laser operation. The conditions for the generation of defects and the factors that contribute to the degradation of the high power laser diodes are discussed on the bases of the simulations; in particular, special attention will be paid to the quality of the QW interfaces.
12:15 PM - G1.6
Residual-strain Dependence of Structural Stability of Strain-compensated Quantum Cascade Lasers
Shin-ichiro Gozu 1 Yukihiro Tsuji 2 Takashi Kato 2 Jun-ichi Hashimoto 2 Katsuhisa Tawa 2 Tsukuru Katsuyama 2 Haruhiko Kuwatsuka 1 Teruo Mozume 1
1National Institute of Advanced Industrial Science and Technology Tsukuba Japan2Sumitomo Electric Industries Yokohama Japan
Show Abstract
Quantum cascade lasers (QCLs) operating at 3 μm are attractive for use in remote sensing of environmental gases. To obtain 3 μm band emission, the material system of QCLs should have a high conduction band offset. This is because the photon energy at 3 μm corresponds to 0.3â?"0.4 eV. The GaInAs/AlInAs system with strain compensation is an attractive candidate for the material system because it is a well-established system enabling sophisticated optical device fabrication for optical communication. However, as QCLs have a very thick active region, precise control of the residual strain is highly important. Recently, we have found that the period and strain of QCL samples shifted due to the residual strain after device fabrication. In this study, we report the effects of residual strain that occurs with device fabrication on the structural stability of QCLs.
Three QCL samples (labeled A, B, and C) were grown by molecular beam epitaxy on n-InP substrates. All QCL samples had an active layer structure with GaInAs (33% Ga: 0.95% compressive strain) wells and AlInAs (60% Al: 0.85% tensile strain) barriers. The layer sequence of one period, starting from the substrate side, is 3.2/3.0/1.7/3.5/1.5/4.0/1.5/0.9/4.5/1.4/3.2/1.6/2.5/1.6/2.5/1.9/2.4/2.2/2.1/2.4, in nanometers. Numbers in bold indicate GaInAs wells; and others, AlInAs barriers. The underlined numbers indicate Si-doped layers. The active region of the QCL is composed of 25 period repetitions. To evaluate period and residual strain, X-ray diffraction (XRD) measurements were performed on the symmetric [004] reflection geometry using a high-resolution XRD system. Standard photolithography was used for device fabrication. During device fabrication, a small area (~5 mm square) of the sample was covered for post-fabrication XRD evaluation. All as-grown samples exhibited a sharp XRD pattern that reflected good crystalline quality. However, their period and residual strain were different depending on the reproducibility of the MBE growth. The period and residual strain of QCL samples A, B, and C were 47.56 nm and -0.16%, 46.64 nm and 0%, and 47.53 nm and -0.49%, respectively, while that of the designed structure were 47.6 nm and 0%, respectively. After device fabrication, both satellite peaks and the substrate peak broadened, indicating the occurrence of unintentional damages during the device fabrication process. In addition, the period and residual strain of fabricated structures were different from those of as-grown samples. For A, B, and C, the shifts in period and strain value were -2.0 nm and 0.024%, +0.12 nm and 0.007%, and -0.65 nm and 0.017%, respectively. These shifts were large for the large residual strain samples, A and C, and very small for the strain-compensated sample B . These results indicated that precise control of strain compensation during QCL structure growth is crucial to obtain the desired performance.
12:30 PM - G1.7
Study of Deep Levels of GaAs/p-GaAs1-xBix Heterostructure Grown by Molecular Beam Epitaxy
Takuma Fuyuki 1 Shota Kashiyama 1 Kunishige Oe 1 Masahiro Yoshimoto 1
1Kyoto Institute of Technology Kyoto Japan
Show Abstract
Deep-level concentrations of p-GaAs1-xBix and at the GaAs/p-GaAs1-xBix heterointerface have been shown to be sufficiently low for application to devices based on the results of deep-level transient spectroscopy (DLTS) and thermal admittance spectroscopy (TAS). Although the metastable alloy of GaAs1-xBix is grown by molecular beam epitaxy (MBE) at a low temperature (370 oC), the deep-level concentration of p-GaAs1-xBix is suppressed to be of the order of 1015 cm-3. The state density at the heterointerface is evaluated to be 5â?"8 Ã- 1011 cm-2 eV-1, which is comparable to other IIIâ?"V heterointerfaces formed at high temperatures, such as InGaAs/GaAs. The surfactant-like effect of Bi is believed to prevent defect formation in the low-temperature growth. To open up new avenues of research in the area of optical communication systems, laser diodes with temperature-insensitive oscillation wavelengths are desired. We have recently demonstrated the successful lasing operation of GaAs1-xBix using optical excitation and the low temperature sensitivity of the lasing wavelength. Because GaAs1-xBix is a metastable alloy, low-temperature growth (400 oC) is essential to realize the incorporation of Bi in the epilayer. There are concerns that the low-temperature growth may cause defects that are directly linked to device performance and reliability. In this paper, deep levels of the GaAs/p-GaAs1-xBix heterostructure were studied by DLTS and TAS. The p-GaAs1-xBix (0x3.4%) with a hole concentration of ~1 Ã- 1018 cm-3 was grown on a p+-GaAs substrate at 370 oC by solid-source MBE. The sample structures of â?oYb Schottky contact/p-GaAs1-xBix/p+-GaAsâ? and â?oAl Schottky contact/undoped GaAs/p-GaAs1-xBix/p+-GaAsâ? were used for DLTS and TAS, respectively. DLTS measurements revealed deep levels with activation energies (EA) of 0.43 and 0.23 eV in p-GaAs1-xBix with x = 1.2% and 3.4%, respectively, which may be attributed to antisite As or Bi. The trap concentration is limited to be of the order of 1015 cm-3. Admittance dispersion is observed for GaAs/p-GaAs1-xBix heterostructures, and for the deep levels causing the dispersion, EA is estimated by TAS to be 0.20â?"0.35 eV. By analyzing the position of the depletion layer and performing a comparative experiment on the GaAs/p-GaAs interface, it was concluded that the deep levels observed by TAS are located at the GaAs/p-GaAs1-xBix heterointerface with a density of 5â?"8 Ã- 1011 cm-2 eV-1, which is similar to the reported value of another IIIâ?"V heterointerface grown at a high temperature (e.g., InxGa1-xAs/GaAs (0.05x0.21)). The surfactant-like effect of Bi atoms on the growth surface enhances the migration of atoms on the growth surface, resulting in the suppression of defect formation both in GaAs1-xBix and at the heterointerface. Further reduction of the deep levels at the interface is expected by adjusting the growth sequence of the heterointerface. GaAs1-xBix is believed to have high potential for device applications.
12:45 PM - G1.8
Analysis of Thermal Stability of PL Peak Wavelength in GaAs1-xNx Capped InAs Quantum Dots
Subhananda Chakrabarti 1 Akshay Agarwal 1 Mayank Chaturvedi 1 Srujan Meesala 1
1Indian Institute of Technology Bombay Mumbai India
Show AbstractInAs/GaAs Quantum Dots (QDs), though widely used for applications such as detectors and lasers, have the inherent disadvantage of suffering spectral shifts in their photoluminescence (PL) spectrum due to in-situ or ex-situ thermal treatments. On this account, the realization of QD based VCSELs and DFB lasers becomes a big challenge, since maintenance of the constancy of PL wavelength is of utmost necessity in these devices. The present work is of great relevance from this perspective. Here we report for the first time a nearly constant PL peak wavelength from InAs/GaAs QDs with dilute GaAsN capping upto 700oC anneal temperature. We have also attempted to model the QD theoretically and concluded that compensation of the blueshift due to In-Ga interdiffusion by the red-shift due to increase in active nitrogen content in the capping layer with annealing is probably responsible for this constancy in PL peak position. The InAs/GaAs QD heterostructure with 30A GaAs0.96N0.04 capping was grown by solid source MBE equipped with a nitrogen plasma source and the QDs being grown at about 510oC. Five different samples were subsequently subjected to RTA in an Ar ambient for 30s from 600-800oC, with a step of 50oC in between samples. The corresponding PL peak wavelengths (for ASG and annealed samples) are 1139.17 nm, 1136.66 nm, 1143.00 nm, 1159.32 nm, 1092.75 nm, 1043.88 nm respectively. We have previously used our theoretical model to explain InAs/GaAs QDs. Here, N incorporation is modeled using the Band Anti Crossing (BAC) model. Previous band structure calculations have revealed negligible impact of N on the valence band or the strain distribution; the strongest effect of N is in reducing the confinement of electrons in the QDs. Simulations with activated N concentrations of 1%, 1.26%, 1.9%, 3.5%, 4% (max) gave PL peak wavelengths of 1215.33 nm, 1215.63 nm, 1215.97 nm, 1227.40 nm, 1169.03 nm and 1096.31 nm for as-grown and samples annealed at temperatures of 600oC, 650oC, 700oC, 750oC and 800oC respectively. We believe that in the asgrown sample, a large fraction of the N is incorporated in interstitial positions and thus does not affect the band structure significantly; with annealing this interstitial N diffuses to substitutional positions, thereby increasing the amount of â?oactiveâ? N in the capping. At low temperatures, when interdiffusion is not that strong, we expect N activation to dominate, explaining the slight redshift. Once N concentration saturates to its maximum value, the peak blueshifts. We see that the trend in experimental PL is reproduced through an N activation model. The authors would like to acknowledge Elisabeth Galopin and Jean-Christophe Harmand from LPN-CNRS, Marcoussis, France, for providing the samples, and the financial support of DST India. .
Symposium Organizers
Osamu Ueda, Kanazawa Institute of Technology Graduate School of Engineering
Mitsuo Fukuda, Toyohashi University of Technology
Kenji Shiojima, University of Fukui Graduate School of Electrical and Electronics Engineering
Edwin Piner, Texas State University, San Marcos
Symposium Support
Japan Society of Applied Physics
G5/II6: Joint Session: Recombination Enhanced Point Defect Reaction
Session Chairs
Wednesday PM, April 11, 2012
Moscone West, Level 2, Room 2007
2:30 AM - *G5.1/II6.1
Control on Nonradiative e-h Recombination Enhanced Defect Reaction(REDR) in II-VI and III-N Light Emitting Devices
Koshi Ando 1 Yoshiaki Harada 1 Masahiro Adachi 2 Show Nakagawa 1 Tomoki Abe 1 Hirofumi Kasada 1
1Tottori University Tottori Japan2Sumitomo Electric Industries Osaka Japan
Show AbstractREDR Effect Induced Degradation in Wide-Bandgap Light Emitting Devices Nonradiative e-h recombination enhanced defect reaction (REDR[1-2]) is one responsible mechanism for shortening device life-time of semiconductor based light emitting devices (LDs and LEDs). An unique property of REDR effect is an efficient blending or mixture of thermal- and electronic-excitation energies, and then high density current (carrier) injection devices operated over room-temperature are always suffering from this vicious degradation due to REDR effect. Electronic excitation energy dissipated from an e-h nonradiative recombination event at localized defect (REDR-center) becomes much larger with increasing in its bandgap energies, so it is essential to understand REDR mechanism and to find effective method on â?oHow to control REDR ? â? toward up-coming practical white LED lightening system. We have made systematic study on degradation process and its mechanism of II-VI (ZnSSe-based) White LEDs and III-nitride based UV and White LEDs. Experiments were based on systematic device aging-tests, coupled with simultaneous defect monitoring by Deep Level Transient Spectroscopy (DLTS, ICTS). In addition we utilized an effective tool of Double Carrier Pulse-DLTS [2] by which one can identify REDRâ?"center (nonradiative e-h recombination center) among deep trapping centers in devices. Based upon the experimental data and the analysis new practical and important technology of â?oCurrent Pulse-Width Controlâ? [3,4] is developed, by which very bright and long-lived UV- and White LED operation becomes possible. REDR Control (Current Pulse-Width Control) for Bright UV- and White LED Operation We present and discuss following technical subjects : (i) Experimental and theoretical background on REDR effect in III-V(GaAs-InP) and widegap compound (ZnSSe-GaN) light emitting devices. (ii) How to identify vicious REDR-center (â?'origin of device degradation) ?. (iii)REDR induced degradation process in widegap semiconductor light emitting devices. (iv)How to control REDR effect for bright and long-lived UV- and White LED operation ?, and what is â?oCurrent Pulse-Width Controlâ? ?. REFERENCES [1] D.V.Lang and L.C.Kimerling, Phys.Rev.Lett.33,489 (1974). [2] K.Ando, M.Yamaguchi and C.Uemura, Phys.Rev.B, 34, 3041(1986). [3] M.Adachi,K.Ando,et al., phys.stat.solidi(b), 241,751(2004). [4]H.Harada,K.Ando, et al., Proc. o f Int.Conf.on II-VI Compound , Mexico, Aug.21-26 (2011)p.97.
3:00 AM - G5.2/II6.2
Feedback and Inflation Mechanism in Successive Multiphonon Carrier Captures at Deep-level Defects
Kei Suzuki 1 Masaki Wakita 1 Yuzo Shinozuka 1
1Wakayama University Wakayama Japan
Show Abstract
Some deep-level defects in semiconductors act as a nonradiative recombination center and limit the efficiency of light emitting devices: called killer center. There are two major nonradiative processes: one is Auger process by exciting another carrier, and the other is multiphonon process by emitting phonons. The latter is very serious in light emitting devices since the electronic energy equal to the band gap Eg is converted to phonon energy (multiphonon process), which is considered to be a trigger of defect reactions. In this paper we discuss the possibility of feedback and inflation mechanism among carrier captures by a deep-level defect and transient induced lattice vibrations using proper configuration coordinate diagrams for many carriers. A recombination at a defect consists of a successive pair of an electron capture and a hole capture. For example, in p-type semiconductors a minority electron is first captured by a defect with a thermal activation energy Eacte, showing the Arrhenius type capture probability. Then the lattice interaction mode Q1 at the defect suddenly starts a vibration with the mean frequency Ï?0 of the contributed modes. The induced vibration shows a damping due to phase relaxation with a time decay constant Ï"~2Ï?/Î"Ï?, where Î"Ï? is the frequency width. During the damping vibration an energy equal to the thermal depth Ethe of a trapped electron is converted to the lattice vibrational energy: S=Ethe/hÏ?0 denotes the average number of emitted phonons. When the the lattice relaxation energy ELR is large, there happen several chances for an athermal capture (without activation energy Eacth) of majority hole during the transient damping vibration (~Ï"). If it occurs, the newly induced lattice vibration superposed on the previous vibration promotes next electron capture, and so on. We study the possibility of this feedback mechanism in successive carrier captures and induced lattice vibrations, using classical treatment of lattice, paying attention to the magnitudes of Î"Ï?/Ï?0, the optical depth Eope, Eoph and ELR. When both the activation energies Eacte and Eacth are small, a series of successive athermal captures is probable. Even if this feedback mechanism is on, we find that the possibility of inflation of the amplitude of the lattice vibration critically depends on the minority capture rate and Î"Ï?/Ï?0. With the obtained results we further discuss the possibility of defect reactions paying attention to the charge of a defect, contribution of normal localized mode(s), the relation between the interaction mode Q1 and the reaction coordinate QR.
3:15 AM - G5.3/II6.3
High-Irradiance Degradation Studies of Metamorphic 1eV GaInAs Solar Cells
Ryan France 1 Myles Steiner 1
1NREL Golden USA
Show Abstract
The implementation of lattice-mismatched junctions into a multijunction solar cell significantly raises the theoretical efficiency over standard lattice-matched multijunction solar cells.1 While long lifetimes have been predicted for standard lattice-matched solar cells,2 lattice-mismatched solar cells have not had as much study and deserve additional attention. Defects are intentionally introduced into inactive regions of these devices in order to change the lattice constant and access materials with more desirable bandgaps. Modern metamorphic growth techniques limit the threading dislocation density in the active region of these devices to 1Ã-106 cm-2 and below,3 which may attain long lifetimes even under the harsh operating conditions of multijunction cells in concentrator systems. Nevertheless, because these defects can potentially lower the functional lifetime of the device, the long-term degradation associated with these defects must be evaluated. Accelerated testing of concentrator solar cells is difficult due to the challenge of supplying the necessary high intensity light. Forward biasing may cause false failures due to differences in current distribution across the cell and may avoid degradation effects related to the interaction of minority carriers with defects, such as recombination enhanced dislocation motion.4 Instead, we use a high power 808 nm diode laser to create minority carriers and a more realistic current profile. This technique produces a steady, low noise light source capable of delivering high intensity light over long periods of time and has been previously used to test the degradation of GaAs cells.5 1eV metamorphic GaInAs solar cells with 2Ã-106 cm-2 threading dislocation density in the active region are stressed under a variety of temperature and illumination conditions while held at Voc. This condition primarily stresses the threading dislocations in the active region. All illuminated samples show a temperature dependent degradation in Voc that is logarithmic with time. The relative percentage degradation after 200 hours illuminated at 1300 suns equivalent and 125°C is 1%, 1.5%, and 2.5% for Voc, FF, and efficiency, respectively, showing these devices to be relatively stable. The dark current significantly increases at low forward bias and in reverse bias and is analyzed with a two-diode model to elucidate degradation mechanisms. Samples at 125°C with varying illumination show similar degradation rates. However, at 125°C but without illumination, the degradation mechanism is not present and there is no change in dark I-V characteristics. This suggests a degradation mechanism that is dependent on illumination, such as a recombination-enhanced mechanism. 1 J.F. Geisz et al., Appl Phys Lett 93 (12), 123505 (2008). 2 C. Algora, Microelectron Reliab 50, 1193 (2010). 3 K.E. Lee et al., J Cryst Growth 312, 250 (2010). 4 O. Ueda, Microelectron Reliab 39, 1839 (1999). 5 I. Rey-Stolle et al., Prog Photovolt 11, 249 (2003).
3:30 AM - G5.4/II6.4
Interplay of In-grown Extended and Point Defects in III-nitrides
Filip Tuomisto 1 Jussi-Matti Maeki 1 Christian Rauch 1 Ilja Makkonen 1 2
1Aalto University Aalto Finland2Aalto University Aalto Finland
Show Abstract
In-grown group III (cation) vacancies (VGa, VAl, VIn) in GaN, AlN and InN [1-3] are typically complexed with a donor-type defect that may in principle be a residual impurity such as ON or H, an n-type dopant such as SiIII, or an intrinsic defect such as the N vacancy (VN). The cation vacancies and their complexes are generally deep acceptors generating radiative and non-radiative deep levels in the gap, they compensate for the n-type conductivity, and add to the scattering centers limiting the carrier mobility in these materials. The formation of these vacancy defects during thin film growth and device processing steps depends on the thermodynamics of the system, on the kinetics of the surface adatoms, and on the presence of impurities and extended defects. Positron annihilation spectroscopy has been widely applied to identify native vacancy defects in III-nitrides. In a semiconductor material, positrons can get trapped at negative and neutral vacancy defects, and at negatively charged non-open volume defects given the temperature is low enough. The trapping of positrons at these defects is observed as well-defined changes in the positron-electron annihilation radiation. The combination of positron lifetime and Doppler broadening techniques with theoretical calculations [3, 4] provides the means to deduce both the identities (sublattice, decoration by impurities) and the concentrations of the vacancies. Performing measurements as a function of temperature gives information on the charge states of the detected defects. We present results obtained in a GaN, AlN and InN thin films and substrates. We compare the results obtained with positrons on the vacancy-donor complexes in the III-nitrides with other structural and chemical data, such as X-ray diffraction and transmission electron microscopy and secondary ion mass spectrometry. We analyze the interplay between extended defects (dislocations, stacking faults) and the vacancy defects present in all three materials. In some cases, depending on the nitride and on the nature of the extended defects, the density distributions of point defects and extended defects tend to correlate, while in others point defects do not seem to be affected by the presence of extended defects. We interpret the results in terms of different phenomena dictating point defect formation during the growth of III-nitride materials. [1] F. Tuomisto et al., Appl. Phys. Lett 90, 121915 (2007). [2] J.-M. Mäki et al., Phys. Rev. B 84, 081204(R) (2011). [3] C. Rauch et al., Phys. Rev. B 84, 125201 (2011). [4] I. Makkonen et al., Phys. Rev. B 82, 041307(R) (2010).
3:45 AM - G5.5/II6.5
Defect Activation in GaN under Low Energy Electron Beam Irradiation
Henri Nykanen 1 Sami Suihkonen 1 Markku Sopanen 1 Lucas Kilanski 2 Filip Tuomisto 2
1Aalto University School of Electrical Engineering Espoo Finland2Aalto University School of Science Espoo Finland
Show AbstractElectron beam irradiation with 5 â?" 35 keV energy has also been reported to cause optical degradation of III-V thin films and device structures. The damage formation is thought to involve activation of vacancies, defect migration and clustering of point defects. We present data on low energy e-beam irradiation damage on GaN thin films. The GaN films were exposed to the e-beam sweep of a Zeiss Supra 40 SEM with 1-5 kA/cm2 current density. The total dose was varied between 0-500 µAs/cm2. We monitored the presence of cation vacancy defects in the samples with a variable energy positron beam (E = 0 ... 38 keV). The e-beam-induced optical degradation of the GaN films was confirmed with photoluminescence measurements that show exponential dependence of the PL intensity as a function of exposure dose. No change was observed in the emission peak position or full width half maximum. The increase in the observable concentration of VGa was confirmed by positron annihilation spectroscopy. By analyzing the Doppler broadening of the energy of the positron-electron annihilation, we show that VGa related defects become detectable when samples are irradiated with 5 and 10 keV electrons. Importantly, the VGa related defects in the irradiated samples have the signature of in-grown VGa complexed with O and possibly H, instead of isolated VGa produced by high-energy particle irradiation. This indicates that the observed VGa related defects are in-grown, but optically inactive and invisible to positrons before irradiation, potentially due to formation of complexes with multiple hydrogen atoms that can be dissociated by the low energy e-beam. Hence we conclude that that e-beam irradiation activates Ga-vacancies that act as non-radiative recombination centers causing the PL degradation.
G6/II7: Joint Session: Defect Issues in Semiconductors
Session Chairs
Wednesday PM, April 11, 2012
Moscone West, Level 2, Room 2007
4:30 AM - *G6.1/II7.1
Radiation-induced Growth of Extended Defects in 4H-Silicon Carbide
Koji Maeda 1
1The University of Tokyo Tokyo Japan
Show Abstract
The multiplication of extended defects such as dislocations and stacking faults is of great technological importance because the extended defects usually have noxious effects on the performance of electronic devices based on the materials. The electronic-stimulated growth of defects to a large spatial extent is curious also from academic viewpoints because there must be some driving force that outweighs the large excessive formation energy of the extended defects. The driving force for the radiation-enhanced dislocation glide (REDG) observed in many semiconductors is usually a mechanical shear stress that is present in the crystal or externally applied by intention. Such a driving force having no origin in the electronic excitation is common in nature with that in ordinary electronically-enhanced diffusion of point defects, where, although the individual enhanced motion is random, the macroscopic flow of defect assembly is driven to such a direction to decrease the system energy on defect recovery. Recent extensive studies of radiation-induced expansion of Shockley stacking faults (SSFs) in 4H-SiC revealed that the expansion of stacking faults (SFs) is "achieved" by REDG of 30°-Si(g) partial dislocations that bound the SSFs, but it is "driven" by the sign reversal of formation energy of SF that is induced only during electronic excitations. This is quite peculiar in that the electronic excitation not only enhances the motion of defects (dislocation glide in this case) but also produces temporarily the driving force for the defect motion at the energetic expense for the formation of SFs. For the origin of the driving force, it was proposed that trapping of excitons by an SF plane effectively reduces the formation energy of the SF and eventually reverses the sign of the SF formation energy at high excitation levels. The energy difference between photoluminescence (PL) from free electron-hole pairs and characteristic PL from the trapped excitons, planarly confined but laterally mobile along the SSF plane, which yields the energy reduction on the exciton trapping at the SF, is 0.30 eV, being significantly large compared to the equilibrium SSF formation energy, as small as 9 meV/ per atom in 4H-SiC. Therefore, the SF formation energy is effectively reduced to 0.009 - 0.30Ï(I,T) [eV], where Ï is the density of the trapped excitons per atom which is variable as a function of the excitation intensity I and temperature T. Thus, measurements of SF expansion velocity and simultaneous measurements of the SF-originated PL intensity which should be proportional to Ï(I) allows us to separate the I-dependence of the driving force from that of the pure REDG effect of 30°-Si(g) partial dislocations, from which a perspective is gained to discuss the microscopic mechanism of REDG effects based on more detailed knowledge of the dislocation structures and relevant electronic levels associated them.
5:15 AM - G6.3/II7.3
An Analytical Bond-order Potential for the Cadmium Telluride Binary System
Donald K. Ward 1 Xiaowang W Zhou 2 Bryan M Wong 3 F. P Doty 1 Jonathan A Zimmerman 2
1Sandia National Laboratory Livermore USA2Sandia National Laboratory Livermore USA3Sandia National Laboratory Livermore USA
Show AbstractCdTe and Cd(1-x)Zn(x)Te are the leading semiconductor compounds for both photovoltaic and radiation-detection applications. The performance of these materials is sensitive to the presence of atomic scale defects in the structures. To enable accurate studies of these defects using modern atomistic simulation technologies, we develop a high-fidelity analytical bond-order potential for the CdTe system. This potential incorporates primary (Ïf) and secondary (Ï?) bonding and the valence-dependence of the heteroatom interactions. The functional forms of the potential are directly derived from quantum-mechanical tight-binding theory under the condition that the first two levels of the expanded Greenâ?Ts function for the Ïf- and Ï?- bond-orders are retained. The potential parameters are optimized using iteration cycles that include first fitting properties of a variety of elemental and compound configurations (with coordination varying from 1 to 12) including small clusters, bulk lattices, defects, and surfaces, and then checking crystalline growth through vapor deposition simulations. It is demonstrated that this CdTe bond-order potential gives structural and property trends close to those seen in experiments and quantum-mechanical calculations, and provides a good description of melting temperature, defect characteristics, and surface reconstructions of the CdTe compound. Most importantly, this potential captures the crystalline growth of the ground-state structures for Cd, Te, and CdTe phases in vapor deposition simulations.
5:30 AM - G6.4/II7.4
Influence of Low Energy Electron Beam Irradiation of Graphene Ribbon Based Back Gated Field Effect Transistors
Adrian Dinescu 1 Munizer Purica 1 Raluca Gavrila 1 Andrei Avram 1 Raluca Muller 1
1IMT Bucharest Voluntari Romania
Show AbstractGraphene based devices have attracted a great research interest, they holding great promises for future nanoelectronics (1). Their fabrication and characterization involve electron beam irradiation through electron beam lithography (EBL) and scanning electron microscopy (SEM), at acceleration voltages in the range of few kV up to 100kV - for EBL and from 100V up to30kV for SEM. The effects of electron beam (accelerated at 5, 20 and 30kV) interaction with graphene have been reported in a series of papers (2,3), indicating that the irradiation produces lattice defects, degrading the deviceâ?Ts electrical transport properties. The low voltage domain (HV< 5kV) is of the greatest importance for SEM imaging of nanostructures and nanomaterials due to various advantages it is providing: true surface imaging, charging artifacts suppression and reduced radiation damage of the specimen. However, during imaging at low voltages, contamination may occur, an amorphous carbon layer being built up in the scanned regions, especially at high magnifications (4). In this work, we report on the modifications induced by the electron beam, with energy bellow 5keV, on graphene and on graphene ribbon based back gated field effect transistors. The ribbons were defined by e-beam lithography, at 10kV accelerating voltage, and using a positive electronresist in order to not irradiate the graphene prior to the electrical measurements. By using a dedicated EBL equipment (Raith e_Line), the graphene ribbons were exposed at 1,2, 3 and 5kV and various doses, from 5 μC/cm2 up to 100.000 μC/cm2. After each exposure, the electrical characteristics of transistors were measured and the ribbons were investigated by AFM (rate of growth of contamination film) and by Raman spectrometry (structure). A technique meant to separate the contamination influence of graphene by the other irradiation effects has been developed. The results indicate important consequences for the fabrication and characterization of devices containing graphene and in particular for those steps in which low voltage scanning electron microscopy or lithography are involved. References: 1. F. Schedin et al., Nature Mater. 6, 652 (2007); 2. D. Teweldebrhan et al., Appl. Phys. Lett. 94, 013101 (2009); 3. I. Childres et al, Appl. Phys. Lett. 97, 173109 (2010); 4.M. Wilson et al, J. Mat. Sci. 15, 2321-2324 (1980)
5:45 AM - G6.5/II7.5
Effects of the Electron Beam Irradiation on the Low-frequency Noise Level in Graphene Transistors
Md. Zahid Hossain 1 Guanxiong Liu 1 Alexander Balandin 1
1University of California - Riverside Riverside USA
Show AbstractThe exceptionally high electron mobility and intrinsic thermal conductivity make graphene a promising material for future high-speed electronics and sensor applications. Most of the proposed applications of graphene require a low level of the low-frequency flicker noise that dominates the spectrum below 100 kHz. The low-frequency noise can be up-converted to high frequency due to device non-linearity and contribute to the phase noise of the systems. For this reason, it is important to investigate the sources and physical mechanism of the low-frequency noise in graphene devices [1-3]. Defects, acting as electron traps, have been identified as responsible for the fluctuation in the number of carriers in the device channels leading to the low-frequency noise [4]. The electron-beam irradiation can be used for controlled introduction of the defects on graphene surface and its crystal lattice [5-6]. The latter allows one to elucidate the effects of defects on the low-frequency noise. In this talk, we will report the study of the low-frequency 1/f noise of the graphene field-effect transistors subjected to the electron beam irradiation. We varied the radiation dose of up to 5000 µC/cm2 in vacuum environment. For all examined devices, subjected to different irradiation doses, the noise spectra were close to 1/f in the frequency range from 1 Hz to 100 kHz (f is the frequency). We compared the noise amplitudes and observed that the electron beam irradiation does not increase the noise level substantially. However, the Raman measurements revealed strong increase in the intensity in the disorder D peak with increasing radiation dose. This suggests that the main contributions to the noise in graphene transistors come from the trapping states inside the dielectric layer. The obtained results shed light on the role of interface states and oxide trap states on the noise origin in graphene devices. The work in Balandin group at UCR was supported, in part, by SRC â?" DARPA through the FCRP Functional Engineered Nano Architectonics (FENA) center. [1] Q. Shao, G. Liu, D. Teweldebrhan, A. A. Balandin, S. Rumyantsev, M. Shur and D. Yan, Flicker noise in bilayer graphene transistors, Electron Dev. Lett., 30, 288 (2009). [2] G. Liu, W. Stillman, S. Rumyantsev, Q. Shao, M. Shur and A.A. Balandin, Low-frequency electronic noise in the double-gate single-layer graphene transistors, Appl. Phys. Lett., 95, 033103 (2009). [3] S. Rumyantsev, G. Liu, W. Stillman, M. Shur and A.A. Balandin, Electrical and noise characteristics of graphene field-effect transistors, J. Phys.: Cond. Matt., 22, 395302 (2010). [4] A.A. Balandin, Noise and Fluctuations Control (ASP, 2002) [5] D. Teweldebrhan and A. A. Balandin, Modification of graphene properties due to electron-beam irradiation, Appl. Phys. Lett., 94, 013101 (2009). [6] G. Liu, D. Teweldebrhan, and A.A. Balandin, Tuning of Graphene Properties via Controlled Exposure to Electron Beams, IEEE Trans. Nanotechn., 10, 865 (2011).
G7: Poster Session
Session Chairs
Wednesday PM, April 11, 2012
Marriott, Yerba Buena, Salons 8-9
9:00 AM - G7.1
Impact of Barrier Thickness on the Structural and Optical Properties of Multilayer(x10) Coupled InAs/GaAs Quantum Dots Using Bilayer(X2) Quantum Dots as the Seed for Strain Patterning
Saikalash Shetty 1 Nivedita Basu 1 Subhananda Chakrabarti 1
1Indian Institute of Technology Bombay Mumbai India
Show AbstractDedicated research on self-assembled InAs/GaAs quantum dots (QDs) has been triggered due to their inherent capability to extend emission of GaAs-based optical devices to long wavelengths such as 1.3 or 1.55 µm. Bilayer QD heterostructures, consisting of seed (bottom) QD layer and active (top) QD layer separated by GaAs spacer layer, are an apt solution to the major obstacle of very large PL linewidth resulting from size distribution in QDs, because of their considerably narrow linewidth. But one of the major technological issues currently is to increase QD densities and simultaneous maintenance of quality and uniformity of QD heterostructures. Stacking of QDs by Stranski-Krastanov growth method is a definite answer. However, strain build-up due to uncompensated stress generated by QDs and consequent relaxation of strain by defect/dislocation formation is uncalled for. Thus, keeping all aspects in mind we would like extension of bilayer QD structure into multilayer coupled structure which is the primary focus of this study. We make a comparison study of temperature dependent photoluminescence (PL) spectra of bilayer (sample 1) and multilayer samples (samples 2 and 3 having 10 QD layers stacked upon seed layer) with varying barrier widths (BW) between stacked QD layers and keeping other parameters-(i) InAs deposition on seed (2.5ML) and active (3.2ML) layers,(ii) growth rate (~0.03ML/s) and (iii) BW between seed layer and active layer (85V) same. Center wavelengths of sample1 and sample 2 having a BW of 110â"« between stacked QD layers are almost same (~1.1µm) but there is an unexpected fall in PL intensity in 2 (~41000au compared to~45000 au for 1), indicating poorer optical response which can be attributed to defects/dislocations resulting from excess strain (caused by propagation of strain field from seed layer through thin spacer layer width) relaxation, giving rise to non-radiative recombinations. XTEM images of 2 showing dense concentration of defects/dislocations supports experimental results. But we see an improved PL intensity (~115000 au) for sample 3 having a BW of 125â"« between stacked QD layers, thereby achieving better optical response. Defects/dislocations are visibly less in 3 as can be concluded from its XTEM images. From power dependent PL study (12.5 W/cm2-1.57kW/cm2) for 2 and 3, we can distinctly see bimodal distribution for 3 compared to multimodal distribution in 2, indicating a more random size and composition variation of QDs in 2 which is undesirable. Activation energy of 3 increases (by ~30meV) over 1 indicating a better carrier confinement arising from improved spatial ordering achieved by stacking of QD layers. A decrease in activation energy of 2 (by~130meV) from 3 is clearly due to introduction of defect states in QD heterostructure. Thus, optimization of barrier width between stacked QD layers would serve our purpose. Acknowledgement: DST, India & FULLSPECTRUM.
9:00 AM - G7.10
Plasma Damage-free Sputtering of Al-doped ZnO Films Grown Using Linear Facing Target Sputtering for GaN-based LEDs
Hyun-Soo Shin 1 Han-Ki Kim 1
1Kyung Hee University Yongin Republic of Korea
Show AbstractWe report the characteristics of Al-doped zinc oxide (AZO) films prepared by a plasma damage free linear facing target sputtering (LFTS) system for use as a transparent conducting layer in GaN-based LEDs. The LFTS system has a linear twin target gun with ladder-type magnet arrays for more effective and uniform confinement of high-density plasma producing energetic particles. Due to effective confinement of high-density plasma and perpendicularly placed substrate position, we could deposit AZO films on p-GaN at low temperature of ~50 C without plasma damage. The electrical, optical, and structural properties of LFTS grown AZO films were investigated as a function of DC power to optimize the properties of the AZO film. We obtained a AZO film with a sheet resistance of 39 Ohm/square and an average transmittance of 84.86 % in the visible range although it was sputtered at room temperature without activation of the Al dopant. Moreover, the LED with sputtered AZO TCL layer showed performance similar to the LED with evaporated ITO TCL layer, indicating that LFTS is promising plasma damage-free and low temperature sputtering technique for high performance GaN-based LEDs.
9:00 AM - G7.11
Interface Studies of Atomic Layer Deposited Al2O3 on GaN
Rathnait D Long 1 Brian McSkimming 2 James S Speck 2 Paul C McIntyre 1
1Stanford University Stanford USA2University of Santa Barbara Santa Barbara USA
Show AbstractGallium nitride (GaN) has recently become the focus of investigations for high power and high frequency applications [1]. Si devices have a limited operating voltage due to hot carrier effects (~6 V ). In contrast, GaN devices can operate as high as 1200 V due to its large bandgap of 3.4 eV. The more conventional commercial GaN structures consist of Schottky devices [2]. However, one limiting factor of these devices is high leakage current through the Schottky barrier. GaN based MOSFETs are also of increased interested as the ability to grow GaN-on-Si has been realised. Therefore, metal-oxide-GaN structures have attracted interest for higher power applications with low leakage current requirements. The oxide-GaN interface will be one of the limiting factors in resulting device performance. Al2O3, with a considerably larger bandgap than other high dielectric constant (κ) materials, gives higher energy band offsets with GaN resulting in reduced gate leakage currents for the same thickness of other high- κâ?Ts [1]. Although some work has been reported on GaN based MOS devices with various dielectrics [1, 3-6], additional understanding of dielectric/GaN interface evolution is required. Thus, this work examines the properties of the GaN surface and the GaN/Al2O3 interface as a result of chemical pre-treatments and annealing, using chemical, physical and electrical characterization techniques. This work will present results on two particular experiments. Firstly we investigate the effect of ex-situ aqueous treatments (HCl and HF, NH4OH) and treatments in-situ in the ALD chamber (prepulsing of the metal precursor, prepulsing of the oxidant prior to the conventional ALD Al2O3 process) on the GaN surface and the GaN-Al2O3 interface. We will examine the chemical, structural and electrical properties of these interfaces using x-ray photoelectron spectroscopy, atomic force microscopy and capacitance-voltage analysis respectively. The second experiment will focus on the effect of varying annealing ambient (forming gas, nitrogen etc.) and temperature on the Pd/Al2O3/GaN metal oxide semiconductor electrical characteristics. Comparison of annealing pre and post metal deposition will also be presented. The integration of atomic layer deposited Al2O3 on GaN and the understanding of the subsequently formed electrical active defects are of utmost importance for the realisation of high power and high frequency GaN devices. [1] Y.C. Chang et al., Micro. Eng. 88, 1207 (2011) [2] P. Parikh et al., Intern. Meet. for Future Electron. Dev. 2004 [3] W. Huang et al., Journal of Elec. Mats. Vol. 35 No. 4 (2006) [4] Y.-L. Chiou et al., Semicond. Sci. Technol. 25 045020 (2010) [5] M. Placidi et al., Journal of Electrochem. Soc. 157 (11) (2010) [6] IEEE Trans. on Elect. Dev. Vol. 50, 5 (2003)
9:00 AM - G7.13
Real Time Growth Process Monitoring and Magnetoptic Properties of GaN:Er Thin Films
Kiran Dasari 1 Javier Wu 1 Ratnakar Palai 1 Wojciech Jadwisienczak 2 Huhtinen Hannu 3
1University of Puerto Rico San Juan Puerto Rico2Ohio University Athens USA3University of Turku Turku Finland
Show AbstractThe rare-earth (RE) doped GaN semiconducting materials are promising candidates for spintronic and optoelectronic applications. The higher magnetic moment of RE ions compared to the transition metal ions makes RE-doped GaN as materials of choice for dilute magnetic semiconductor. RE-doped GaN materials have shown the ability to tune the direct bandgap from the ultraviolet through visible to the near infrared region. GaN:Er thin films are great interest because of their emission in visible regime. The main objective of our work is to make Er-doping in GaN both magnetically and optically active at room temperature. Here we report, the magnetic and optical properties Er+3 â?"doped GaN epitaxial thin films grown by molecular beam epitaxy (MBE). Thin films of GaN:Er have been grown on Si (1111) and Sapphire (0001) substrates using MBE. The growth process was monitored in situ by reflection high energy electron diffraction (RHEED). The doping concentration and elemental analysis have been calculated from the x-ray photoelectron spectroscopy (XPS). X-ray diffraction and atomic force microscopy (AFM) were used to examine the phase purity and smoothness of the films. The concentration and mobility of carriers were calculated from the Hall effect measurements. The magnetic properties of GaN:ErYb thin films were measured using SQUID and magnetotransport measurements. The optical properties of the films were analyzed by photoluminescence (PL) and cathode luminescent spectroscopy (CL). The magnetic and optical properties of the GaN:Er films will be compared with Yb-doped GaN films and will be discussed in details.
9:00 AM - G7.14
Influence of Annealing on the Optical Properties of GaN Grown by Pulsed Laser Deposition
Muhammad Baseer Haider 1 Sardar M Durrani 1 Mohammad F Al-Kuhaili 1 Imran Bakhtiari 1
1King Fahd University of Petroleum and Minerals Dhahran Saudi Arabia
Show Abstract
For the past few decades, III-V compound semiconductors especially GaN has attracted much attention. Due to its direct and wide bandgap, GaN is one of the best candidates for optoelectronic devices such as solar cells, photodectors, light emitting diodes, and lasers. We have successfully grown GaN sample on unheated fused silica by pulsed laser deposition. Growth was performed at room temperature and subsequently the samples were annealed at different temperatures in nitrogen environment. Post growth annealing was performed at 400, 500 and 600 and 700oC in nitrogen atmosphere. Surface analyses of the samples were performed using atomic force microscopy and x-ray photoelectron spectroscopy. We have found that root mean square roughness of the surface is not much affected by post-growth annealing at different temperatures. Optical properties of the as-grown samples reveal the presence of high density of defects and consequently poor absorption. Optical properties of the samples improved dramatically for samples that were annealed at 600oC. X-ray photoelectron spectroscopy results reveal that as-grown GaN samples mainly consisted of sub-stoichiometric GaXOY and chemisorbed surface oxygen. By annealing the samples in nitrogen atmosphere at 500oC and higher temperatures completely removes chemisorbed oxygen and converts GaXOY into GaN. We have found that annealing the GaN samples in nitrogen atmosphere, grown by pulsed laser deposition, at 600oC improves the stoichiometry and optical properties of the films. (This work was supported by KFUPM through internal research grant # IN100040)
9:00 AM - G7.15
Nitridation of GaAs Surface as an Alternative for Obtain Cubic GaN Substrate
Victor Saacute;nchez 1 Heber Vilchis 1
1CINVESTAV-IPN Meacute;xico Mexico
Show AbstractThe main disadvantage of current technology in cubic gallium nitride (c-GaN) devices is the lack of native substrates. The synthesis of c-GaN films are carried out on foreign substrates as silicon or gallium arsenide by heteroepitaxial process using molecular beam epitaxy (MBE) or metal-organic chemical vapor deposition (MOCVD) techniques. Heteroepitaxial growth induces defects in the material due to the lattice mismatch, which degrade the crystalline quality of the material and its optical and electrical properties. Also cubic monocrystalline GaN films are difficult to obtain due to the defects induce the incorporation of hexagonal phase. In this work we studied the nitridation process of GaAs surface at atmospheric pressure by CVD technique. The heat source is an arrangement of infrared lamps with singular characteristics. Experimental procedure consists in heat the GaAs substrate at high temperature with a flux of NH3, the arsenic desorption of the crystal is sufficient allowing a surface rich in gallium and nitrogen atoms occupied the arsenic places results in a thin layer of gallium nitride. If the process takes a lot time the converting of GaAs substrate in GaN bulk is possible. The films obtained had thickness around 20 μm and can be removed mechanically; so the films are appropriate as template for future epitaxial process. Structural characterization shows that the GaN films had cubic monocrystalline structure on the surface while the bulk is a mixture polycrystalline of cubic and hexagonal phases. We explain this result using a simple model based in thermodynamic parameters. Electrical and optical measurements probe the cubic properties of the GaN substrates.
9:00 AM - G7.16
Fabrication and Characterization of Multiferroic 2-dimensional ZnO:(Cr,Ti) Nanoline
Eiwhan Jung 1 Youngmin Lee 1 Sejoon Lee 2 Yoon Shon 2 DeukYoung Kim 1
1Dongguk University Seoul Republic of Korea2Dongguk University Seoul Republic of Korea
Show AbstractThe ZnO-based thin films with Cr and/or Ti co-doping have been prepared by co-sputtering of ZnO:Cr (Cr ~1 at.%) and Ti using r.f. reactive magnetron sputtering and d.c. magnetron sputtering, respectively. After the deposition, 2-dimension nanoline was fabricated by optical lithography. In measurements of superconducting quantum interference device magmetometry, the ZnO:(Cr,Ti) 2-d nanoline showed strong magnetic hysteresis loops indicative of obvious ferromagnetism in the material system. Simultaneously with ferromagnetism, the films revealed the electrical polarization properties for measurements of the polarization-electric field relationship. The ferromagnetic and ferroelectric properties were dependent on the oxygen partial pressure during the film growth. These results indicate that the films prepared in this study have multiferroism. In addition, ZnO:(Cr,Ti) thin films clearly showed the hysteretic behaviors in current-voltage characteristics. It might be due to the suppressed relaxation of trapped charged at the localized electric dipoles in ferroelectric materials. Consequently, it is suggested that a new kind of ZnO-based multiferroic semiconductor ZnO:(Cr,Ti) can be effectively used for the application of future fine-information devices whose operation principle is fully based on electromagnetic coupling or magnetoelectric coupling effects.
9:00 AM - G7.17
Characteristics of ZnO Thin Films Formed by Sol-gel Method Using Microwave
Eunkyeom Kim 1 Youngill Kim 1 Jaekyu Jeong 1 Bongsub Kim 1 Kyoungwan Park 1
1University of Seoul Seoul Republic of Korea
Show Abstract
High-mobility oxide-based thin-film transistors (TFTs) for large-area electronics have attracted much attention for application in next-generation flat-panel displays, such as organic light-emitting diode (OLED) displays and three dimensions (3D) displays. Among oxide-based materials, zinc oxide (ZnO) is a promising candidate for novel device applications. ZnO is a IIâ?"VI group and n-type direct bandgap semiconductor that possesses some great characteristics, including a wide energy bandgap (3.3 eV), large free exciton binding energy (60 mV), wide range resistivity (10â^'4 to 1012 Ωcm), high carrier mobility, high transparency at room temperature, and good photoelectric, piezoelectric, and thermoelectric properties. ZnO-based thin films have been prepared by various vacuum deposition techniques and solution-based deposition processes. Solution-based deposition processes offer a simple, easy, low cost, and large area thin-film coating alternative to vacuum deposition techniques. Also, it can be suitable for preparation of ZnO films for electronic device applications. In this work, the ZnO film was synthesized by solâ?"gel method. We used zinc acetate dihydrate, as a starting material, 2-methyoxyethanol as a solvent, and monoethanolamine as a stabilizer. The precursor solution was mixed by microwave during various process times at 60 °C, and then placed in air for 24 h resulting in a clear and homogeneous sol. In the sol, concentrations of monoethanolamine and Zn2+ were equal. The film of ZnO was prepared by spin coating on the substrate which had been thoroughly cleaned and dried. Then, the ZnO film was kept in an oven at 150 °C during 1 h in order to evaporate the solvent, and subsequently annealed at 350 °C in N2 atmosphere. The ZnO films were investigated by X-ray diffraction, atomic force microscope, scanning electron microscope, and transmittance measurements. To investigate electrical characteristics of ZnO thin films, the bottom-gate ZnO TFTs were fabricated, and the electrical transfer characteristics were measured.
9:00 AM - G7.19
Improvement of Negative Bias Temperature Instability under Light Illumination on Amorphous InGaZnO Thin-film Transistors Fabricated by High-pressure Water Vapor Annealing
You Seung Rim 1 Hyun Soo Shin 2 Hyun Jae Kim 1
1Yonsei University Seoul Republic of Korea2Samsung Mobile Display Co., Ltd. Yongin-city Republic of Korea
Show AbstractIn this study, we investigated negative bias temperature instability under light illumination (NBTIS) on amorphous InGaZnO (a-IGZO) thin-film transistors (TFTs) with fabricated devices annealed at high pressure water vapor. High-pressure water vapor annealing improved the electrical characteristics and NBTIS of devices. This was attributed to a reduction in not only oxygen vacancy defects in the IGZO films, showing that a photo-generated transition from VO to VO2+, but also the band bending at the overlap between the sourceâ?"drain (S/D) electrodes and the etch stop layer (ESL), reducing charge trapping or/and injection into ESL. As a result, Î"VTH of a high pressure water vapor annealed device under NBTIS conditions was negatively shifted with value of -1.42 V, which is much improved than that of a non-pressured device (-2.4 V).
9:00 AM - G7.2
Resonant Tunneling Diodes with Quaternary GaInNAs Absorption Layer for Telecommunication Light Detection Application
Fabian Hartmann 1 Fabian Langer 1 Dirk Bisping 1 Sven Hoefling 1 Martin Kamp 1 Alfred Forchel 1 Lukas Worschech 1
1Universitauml;t Wuuml;rzburg Wuuml;rzburg Germany
Show AbstractResonant tunneling diodes (RTDs) gained significant interest as light-sensitive photo-detectors in the last decade due to their fast operation speed and ultra-low noise properties even the bistable regime. Illumination with light in a biased RTD, electron-hole pairs can be optically excited and separated, which in turn modulates the effective electric field compared to the case without optical excitation. GaAs based RTDs show excellent performance with large peak-to-valley ratios. In order to combine the advantages of GaAs based RTDs with a low band gap material to realize an RTD photodetector for the telecommunication range (>1.3 µm) we realized GaAs/AlGaAs RTDs with a lattice-matched InGaNAs layer. We present morphological as well as electrical studies of GaAs/AlGaAs RTDs with a GaInNAs absorption layer demonstrating excellent room temperature sensor performance for telecommunication light detection at 1.3 µm. For that purpose Al0.6Ga0.4As/GaAs/Al0.6Ga0.4As RTDs with a nitride based Ga0.89In0.11N0.04As0.96 absorption layer were grown by molecular beam epitaxy. We demonstrate resonant tunneling at room temperature with peak-to-valley ratios up to about 4. Under illumination of the RTDs at the telecommunication wavelength with 1.3 µm a light sensitivity of around 1000 A/W was found without external amplification. The role of GaAs buffer layer and open contact geometries are discussed.
9:00 AM - G7.20
Influence of Thermal Annealing on P-type Properties of Sb Doped ZnO Thin Films
Jinyong Lee 1 Deuk Young Kim 1 Choeun Lee 1 Eiwhan Jung 1 Doosoo Kim 1 Eunhee Shim 1 Youngmin Lee 1 Hyosuk Choi 2 Moondeok Kim 2
1Dongguk University Seoul Republic of Korea2Chungnam National University Daejeon Republic of Korea
Show Abstract
The Sb doped ZnO films were grown on c-plane sapphire substrates by RF magnetron sputtering and the effect of thermal annealing on their electrical, structural, optical properties was investigated. Hall measurement revealed that carrier concentration was changed from n-type to p-type after thermal annealing at temperature above 700°C. After annealed at 1000°C, The Sb doped ZnO films showed hole concentration of 2.05 x 1018 cm-3 and carrier mobility of 2.2 cm2/Vs which was much higher hole concentration than others. X-ray diffraction measurement showed c-axis preferential wurtzite lattice structure of thin films and (002) diffraction peak shifted from 33.7° to 34.4° as annealing temperature increased. It implies that Sb was substituted in ZnO lattice successfully and consistent with the increase of hole carrier concentration. The photoluminescence measurement at low temperature was showed that the free electron to acceptor transition at 3.319 eV which was an acceptor related transitions and the acceptor binding energy had been calculated to be 118 meV.
9:00 AM - G7.21
Effect of Hafnium on Zn-Sn-O Thin Film Transistors Fabricated by Solution Process
Jun Young Choi 2 SangSig Kim 2 Sang Yeol Lee 1
1Cheongju University Cheongju Republic of Korea2Korea University Seoul Republic of Korea
Show AbstractOxide-based semiconductors have much attention in thin film transistors(TFT) for display applications. Oxide semiconducting materials are interest in active matrix display as alternatives for conventional silicon due to wide band gap, high mobility, and good transparency[1-2]. Conventional silicon have been well optimized fabrication process and low cost materials for industry, however poly-silicons and amorphous silicons unsuitable for high performance displays due to a low mobility and poor environment stability. In the display industry, the alternative materials to conventional silicon are emerging to appear high performance. Recently, transparent oxide semiconductors have attracted much attention for active channel layer in TFT. ZnO-based semiconductors have been widely investigated in application for sensors, panel display, and circuits. ZnO have a high mobility and high transparency due to their wide band gap.[3] It make them promising candidate for various industrial engineering. Vacuum deposition process such as RF magnetron sputtering, pulsed-laser deposition require high manufacturing cost and expensive equipment In this study, we focused on the effects of the hafnium in zinc tin oxide (HZTO) TFTs fabricated a sol-gel method. The HZO sol-gel precursor was prepared by combining 0.5 M of hafnium, chloride, zinc acetate dehydrate and tin chloride in 2-methoxyethanol solvent. Then, monoethanolamine (MEA) was used as a solution stabilizer, and stirred at 80°C for 24 h to form the HZTO sol-gel precursor. Active channel layer was deposited by the spin coating method and then annealed at 300°C for 1h in air. In this article, we investigated the feasibility of controlling the threshold voltage (Vth) and field effect mobility (μFE) have been studied by adjusting hafnium ratio. Vth shifted toward positive direction, and the μFE was decreased due to the decrease of carrier concentration because hafnium acts as carrier suppressor. The subthreshold swing exhibits good properties from 1.01 to 0.44. Also, based on our XPS analysis, it is suggested that the decrease of carrier concentration in HZTO is closely related with the decrease oxygen vacancy. 1. J. F. Wager, science, 300, 1245(2003) 2. K. Nomura, H. ohta. K. Ueda, T. Kamiya, M. Hirano, and H. Hosono, Science, 300, 1269 (2003) 3. Hoffman, R. L. ; Norris, B. J.; Wager, J. F. Appl. Phys. Lett. 2003, 82, 733.
9:00 AM - G7.22
Tuning the Optical and Electrical Properties of Hydrothermally Grown ZnO Nanowires by Sealed Post Annealing Treatment
Shuigang Xu 1 Chun Cheng 1 Wenhao Guo 1 Yuheng He 1 Rui Huang 1 Shengwang Du 1 Ning Wang 1
1the Hong Kong University of Science and Technology Hong Kong Hong Kong
Show Abstract
ZnO nanowires (NWs) synthesized by hydrothermal methods usually have poor optical properties and high electron concentration due to their native defects. By sealed post annealing treatment, the defect related emission in ZnO NWâ?Ts photoluminescence can be completely eliminated and the UV emission can be largely improved. By adding phosphorus pentoxide in the treatment, the electron concentration in ZnO NWs was tuned almost two orders of magnitude from 9.7 Ã- 1019 cm-3 to 1.0 Ã- 1018 cm-3. Meanwhile, the electron mobility increased from 6.1 cm2/V s to 19.5 cm2/V s. The present treatment can effectively compensate the native point defects in hydrothermally grown ZnO NWs.
9:00 AM - G7.24
Chemical Vapor Deposition of Boron Phosphide Thin Films
Julia K.C. Abbott 1 J. D Brasfield 2 1 Philip D Rack 3 Gerd J Duscher 3 Charles S Feigerle 1
1University of Tennessee Knoxville USA2Y-12 National Security Complex Oak Ridge USA3University of Tennessee Knoxville USA
Show Abstract
Boron Phosphide (BP) is a promising material for use as a room temperature semiconductor detector for thermal neutrons. The absorption of a thermal neutron by a 10B nucleus in BP can yield 2.3MeV of energy which in solid state BP can yield ~0.5 million electron-hole pairs that would be detectable with minimal amplification in a device. BP thin films are grown according to the net reaction below in a cold wall chemical vapor deposition (CVD) reactor: B2H6 + 2PH3 â?' 2BP + 6H2. Growths are performed using diborane and phosphine with a balance of hydrogen gas at near atmospheric pressure with RF induction heating. The resultant BP thin films are characterized by Raman, XRD, SEM, TEM and TEM-EELS for chemical composition, surface and bulk morphology. BP growths on Si and SiC substrates are compared. SiC provides reduced lattice mismatch for growth of BP over Si. Growth of heteroepitaxial BP on SiC will also be discussed.
9:00 AM - G7.25
Effects of In-situ Remote Plasma Treatment on Amorphous Indium Gallium Zinc Oxide Thin Film Transistors Using a Double Active Channel Layer
Ja Hyun Koo 1 Tae Sung Kang 1 Jin Pyo Hong 1 Tae Hyun Kim 2 Seong Kwang Kim 2
1Han Yang Univ. Seoul Republic of Korea2Hanyang University Seoul Republic of Korea
Show AbstractAmorphous InGaZnO (a-IGZO) material is a post transition metal oxide which has recently garnered much attention as a active channel later in thin film transistor (TFTs) due to their excellent features, such as high field-effect mobilities, small subthreshold voltage swings, and good electrical stability. The a-IGZO has intensively been studied for several years as a high-performance semiconductor material and become one of the major candidates for large-area, high-speed electronics such as active matrix backplanes of organic light-emitting diode displays (OLEDs). We studied physical properties of a double-active-channel layer of a-IGZO/TIGZO (SnInGaZnO) in order to fabricate reliable a-IGZO oxide TFTs under an conventional co-sputtering processes. The ITO and IGZO targets were co-sputtered for the deposition of TIGZO layer by controlling the doping contents of Sn and In doping with various rf sputtering powers. Our a-IGZO TFT was a staggered bottom gate structure and in-situ plasma treatment procedure was additionally performed during channel layer deposition. The a-IGZO TFTs were prepared using rf sputtering with a power density of 2 W/cm2 and a processing pressure of 3 mTorr in various oxygen partial pressure and then followed by a 300 celcious degree post-deposition anneal in air. Electrical and structural features of a-IGZO TFTs critically were dependent on remote plasma power, ICP source-to-substrate distance, and O2 partial pressure.
9:00 AM - G7.3
Degradation Characteristics of AlGaN/GaN-on-Si Heterostructure FETs for High Voltage Switching Application
Shinhyuk Choi 1 Bong-ryeol Park 1 Hoyoung Cha 1 Hyungtak Kim 1
1Hongik University Seoul Republic of Korea
Show AbstractWe have fabricated field-plated AlGaN/GaN Heterostructure Field Effect Transistors on Si substrate for high voltage switching application and submitted the devices to the stress test to investigate the degradation characteristics. The devices were stressed under three different types of bias configuration including gate bias stress, high current stress and high field stress. The maximum stress voltage ranged from several hundred V to 1kV depending on the physical dimension of HFETs. Several degradation modes such as the shift of threshold voltage, the change in gate leakage current and on-resistance were identified. The degradation showed the moderate dependence on the field plate dimensional parameters. We also performed the stress test varying temperature to evaluate hot carrier effect on the degradation. Detailed analysis on the degradation characteristics will be presented and discussed.
9:00 AM - G7.5
Molecular-beam Epitaxy of InSb Nano-scale Dots on GaSb at Low Temperature
Huiping Shen 1 Mark Ashwin 2 Changying Zhao 1 Tim Jones 2
1University of Warwick Coventry United Kingdom2University of Warwick Coventry United Kingdom
Show AbstractInSb quantum dots grown on GaSb have many potential applications for mid-infrared optical devices. Since InSb/GaSb has a lattice mismatch similar to InAs/GaAs it would appear that the growth of 3D InSb islands should be the same. However, reports in the literature have indicated that direct deposition leads to a low density of dots (~ 4x109 cm-2) of low aspect ratio [1,2], leading to the development of methods to circumvent this problem [1,2]. We have investigated a method of direct deposition by MBE using low growth temperatures for the deposition of between 1 and 6 ML of InSb onto GaSb(001). The growth of the InSb was monitored by RHEED and the resulting surface morphology was examined by atomic force microscopy. For the deposition of 1 â?" 2ML of InSb we observed that no dots were formed, which agrees with predictions for the critical thickness for InSb deposited on GaSb. For further deposition we observed the formation of dots which increased in density reaching a maximum of ~ 4.5x1010 cm-2 after depositing 4 ML and were approximately 2.5±0.5nm high with a diameter of 35±5nm. As the deposition quantities were increased to 6ML the dot density decreased and evidence of coalescence was observed. The observed dot densities reported here are an order of magnitude greater than previously reported for direct deposition and are comparable to those achieved by post deposition annealing [1]. The results potentially provide a method for simplifying and controlling the growth of InSb dots on GaSb. References: [1] N. Deguffroy, V. Tasco, A. N. Baranov, E. Tournie, B. Satpati, A. Trampert, M. S. Dunaevskii, A. Titkov, and M. Ramonda, J. Appl. Phys. 101 124309 (2007). [2] Brain R. Bennett, R.Magno, and B.V.Shanabrook, Appl. Phys. Lett. 68, 505 (1996).
9:00 AM - G7.6
Trimethylaluminum Saturation of n-InGaAs Surface Prior to Atomic Layer Deposition of Al2O3
Jaesoo Ahn 1 Wilhelm Melitz 2 Tyler Kent 2 Andrew C Kummel 2 Paul C McIntyre 1
1Stanford University Stanford USA2University of California, San Diego La Jolla USA
Show Abstract
We investigate effects of trimethylaluminum (TMA) saturation of As2-decapped InGaAs (100) surfaces prior to atomic layer deposition (ALD) of Al2O3 on electrical characteristics and interface properties of Pd/Al2O3/n-In0.53Ga0.47As MOS capacitors. InGaAs MOSCAPs in which the substrate surface is dosed with TMA-treated Al2O3 for extended duration prior to the start of ALD show smaller frequency dispersion at negative bias and lower accumulation capacitance density, compared to capacitors without TMA pre-treatment. The reduced frequency dispersion observed in the TMA-treated surfaces indicates a lower density of interface defects and, possibly, bulk traps in the Al2O3 layer for TMA pre-treatments. The estimated interface defect density obtained by the conductance method decreases from 9.2Ã-1012 cm-2eV-1 to 6.5Ã-1012 cm-2eV-1 near InGaAs midgap, when TMA is initially dosed for 60 seconds at 200°C prior to Al2O3 ALD. In addition, parallel conductance maps show efficient vertical movement of the maximum parallel conductance peak [(Gp/Ï?)/Aq]max with gate bias, indicating an unpinned Fermi level. TMA dosing prior to Al2O3 deposition can increase the initial ALD growth rate per cycle, which makes the oxides physically thicker and, as a result, reduces the accumulation capacitance density for a given number of TMA/H2O ALD cycles. Scanning tunneling microscopy (STM) shows the pre-treated InGaAs(100)-4x2 surface below 200°C generates a new reconstruction with 0.8nm row spacing, while surfaces above 250°C generate a wider spacing reconstruction with 1.7nm row spacing. STM indicates that the lower temperature tighter spacing reconstruction is unpinned while the higher temperature pre-treated dose produces a pinned interface. X-ray photoelectron spectroscopy spectra reveal that the peak intensity of the bonding between the aluminum and the InGaAs substrate increases for the 60 second-TMA Al2O3. This result suggests that TMA prepulsing provides a higher coverage of TMA on the InGaAs surface.
9:00 AM - G7.7
The Occupancy of the Threading Dislocation Lines within n-type Gallium Nitride: Recent Progress
Erfan Baghani 1 Stephen Karrer O'Leary 1
1The University of British Columbia Kelowna Canada
Show AbstractWithin the framework of a grand partition function formalism, we examine the occupancy of the threading dislocation lines within n-type gallium nitride. A number of defect sites associated with these threading dislocation lines are considered for the purposes of this analysis. The sensitivity of these results to variations in the electron-electron interactions are considered and shown to be relatively minor. The distribution of screening charges that exist about these charged threading dislocation lines is also examined.
9:00 AM - G7.8
Surface Segregation in Rare Earth Doped GaN Thin Films
Stephen Richard McHale 1 John W McClory 1 James C Petrosky 1 Ratnakar Palai 2 Yaroslav Losovyj 3 4 Peter A Dowben 4 Ihor Ketsman 4 Douglas A Buchanan 1
1Air Force Institute of Technology Wright Patterson AFB USA2University of Puerto Rico San Juan Puerto Rico3Louisiana State University Baton Rouge USA4University of Nebraska Lincoln USA
Show Abstract
During the past decade, rare earth doped semiconductors have generated considerable attention for their applications in new optoelectronic devices. GaN is considered one of the most important semiconductor materials after silicon, and is widely used in the production of visible and UV LEDs. Recently, the Schottky barriers formed at the interface between gold and various rare earth doped GaN thin films (RE = Yb, Er, Gd) were found to be 25-55% larger than reported barrier heights at the gold to undoped GaN interface. Moreover, the trend of the Schottky barrier heights was observed to follow the trend of the rare earth metal work function. Although it is known that rare earth ions occupy Ga sites in the GaN lattice, perturbation of the rare earth ion on the surface electronic structure of GaN is possible. Accordingly, studies of rare earth surface segregation in GaN:RE have been undertaken using grazing angle dependent photoemission spectroscopy. UV Photoemission spectroscopy and X-ray photoemission spectroscopy experiments show clear evidence for rare earth ion enrichment in the selvedge layer of GaN:Yb samples, but no segregation of rare earth ions to either the selvedge or surface layers of GaN:Er; both the GaN:Yb and GaN:Er surfaces terminated in GaN. Although minimally chemically different, some macroscopic properties of the rare earth metals vary irregularly and appear, at first glance, to be in contradiction to the anticipated smooth behavior. This irregularity, however, occurs due to the different atomic configurations of the different rare earths. While most of the lanthanide rare earths (including Yb and Er) have divalent 4fn6s2 atomic configurations, nearly all of these elements are found in the trivalent 4fn-15d16s2 configuration when in solid form. Thus, in the condensation process, one of the 4f electrons is promoted to the 5d level. The cohesive energy is one such macroscopic property of the rare earths that exhibits quasi irregular behavior. Specifically, the lower cohesive energy of Yb (~30 kcal/mole) over Er (~75 kcal/mole) provides a compelling explanation for the observed preferential surface segregation of Yb over Er in GaN:RE thin films.
9:00 AM - G7.9
Effect of Doping-concentration-dependent Characteristics of Ga-doped ZnO on the Luminescence of GaN Light-emitting Diodes as Electrodes
Chang Oh Kim 1 Sung Kim 1 Dong Hee Shin 1 Suk-Ho Choi 1
1Kyung Hee University Yongin 446-701 Republic of Korea
Show Abstract
Electrical and optical characteristics of GaN p-n light-emitting diodes (LEDs) using Ga-doped ZnO (GZO) as electrodes have been investigated by current-voltage (I-V) and electroluminescence (EL)measurements. GaN p-n epilayers with a total thickness of ~ 6 μm were grown on c-plane (0001) sapphire substrates by metal-organic chemical vapor deposition. To fabricate the GaN p-n LED structures, the p-GaN layer was partially etched until the n-GaN layer was exposed, and then 100-nm-thick (1~5 mol % Ga content) GZO electrodes of 500 μm diameter were deposited on both p- and n- GaN layers using a shadow mask at 400 oC by RF sputtering, and subsequently annealed at 900 oC for 3 min in a N2 ambient. When a forward bias was applied to the GZO electrodes on the p-GaN layer, two dominant EL peaks of ultraviolet and yellow emissions were observed at ~376 nm and ~560 nm for all LEDs, respectively. Both EL peaks were attributed to radiative recombinations between N-vacancy-related shallow donor located at ~100 meV below the conduction band and the valence band edge and between defects at the p-GaN/n-GaN boundaries and in the GaN lattice, respectively. The LEDs with 2 mol % GZO showed largest UV EL intensities over the full range of injection current. The I-V curves showed rectifying p-n junction behaviors except for the 5 mol % GZO device and the reverse-bias leakage current was minimized in the 2 mol % GZO device. These results were consistent with the data of Hall effect, photoluminescence, and X-ray diffraction. The EL mechanisms were discussed using energy band diagrams of the GZO/p-GaN/n-GaN/GZO structures.
G4: Characterization and Theoretical Calculation
Session Chairs
Wednesday AM, April 11, 2012
Moscone West, Level 2, Room 2007
9:30 AM - *G4.1
Strain and Temperature Dependence of Defect Formation at AlGaN/GaN High Electron Mobility Transistors on a Nanometer Scale
Chung-Han Lin 1 Tyler A Merz 2 Daniel R Doutt 2 Jungwoo Joh 3 Jesus A del Alamo 3 Umesh K Mishra 4 Leonard J. Brillson 1 2 5
1The Ohio State University Columbus USA2The Ohio State University Columbus USA3Massachusetts Institute of Technology Cambridge USA4University of California Santa Barbara USA5The Ohio State University Columbus USA
Show AbstractWe used depth-resolved cathodoluminescence spectroscopy (DRCLS), Kelvin probe force microscopy (KPFM), and surface photovoltage spectroscopy (SPS) on a nanometer scale to map the temperature, strain, and defects inside state-of-the-art GaN high electron mobility transistors (HEMTs) as they degrade systematically. DRCLS measured in-situ optical band gap shifts during ON- vs. OFF-state operation to separate temperature vs. strain effects, respectively [Appl. Phys. Lett. 95, 033510 (2009); 97, 223502 (2010)]. ON-state (Vds = 6 V, Id = 0.75 A/mm) DRCLS maps temperature at localized depths in 3 dimensions, in particular within the 2-dimensional electron gas (2DEG) region during device operation. With increasing OFF-state (Vds= 10-30 V, Vgs= -6 V, Id= 5*10^-6 A/mm) drain-source voltage Vds, DRCLS maps increasing strain ε up to 0.27 GPa. KPFM maps surface electric potential across the device, revealing lower potential patches that decrease rapidly with increasing off-state stress, moving the surface Fermi level Ef lower in the GaN band gap. Corresponding CL spectra acquired at these patches and localized within the 2DEG exhibit deep level defect emissions that increase with both on-and off-state stress, reducing near band edge (NBE) emission, increasing surface banding (decreasing surface potential), and moving OFF-state gate leakage Igoff sharply higher above Vdg = 30 V (ε = 0.25 GPa. SPS identifies the energy level positions of the defects with respect to the band edges. Device failure occurs at drain-side points where potential decreases most rapidly with increasing Vdg, beginning as low as Vdg = 20 V (ε = 0.15 GPa). The evolution of surface potential with applied bias predicts the patches where AlGaN HEMTs will fail. ON-state current also produced local drain-side patches with decreased potentials, however decreasing 3-6 times less and varying drain current by only 2%. High field-induced strain generates defects at the drain-side gate foot that lead to device failure, while current-induced degradation is significantly lower. The OFF-state drain-source voltage dependence of device degradation and associated defect formation supports the device degradation model of del Alamo et al.[P. Makaram et al., Appl. Phys. Lett. 96, 233509 (2010)]. The separation of field- vs. current-induced degradation on a nanoscale highlights their relative impact on AlGaN/GaN HEMT reliability.
10:00 AM - G4.2
Structural Defect Formation in Ni-Gated AlGaN/GaN HEMTs
Monta Holzworth 1 N. G Rudawski 1 P. G Whiting 1 C. Y Chang 1 E. A Douglas 1 S. J Pearton 1 K. S Jones 1 L. Liu 2 T. S Kang 2 F. Ren 2 L. Zhou 3 M. R Johnson 3 D. J Smith 3 G. D Via 4
1University of Florida Gainesville USA2University of Florida Gainesville USA3Arizona State University Gainesville USA4Wright Patterson Air Force Base Dayton USA
Show AbstractAlGaN/GaN HEMTs were grown on a semi-insulating 6H-SiC substrate. The structure involved an AlN nucleation layer on the SiC, followed by a 2.25 µm Fe-doped GaN buffer, 15 nm of Al0.25Ga0.75N, and capped with 3 nm of an unintentionally doped GaN layer. Typically, the last GaN epilayer is never observed. The ohmic contacts used metal stacks consisting of Ti/Al/Ni/Au annealed for 30 s at 850°C. Additionally, the gate contact consisted of a Ni/Au metal stack. The gate length for these devices was 100 nm, and the devices were passivated with SiNx. Structural defects were observed to form from a chemical reaction between the Ni-layer of the gate contact and the underlying AlGaN epilayer under high-reverse gate bias. The first bias condition set the source to ground, the drain to 5 V, and the gate to -10 V. For the second bias condition, the source was set to ground, the drain was set to 10 V, and the gate was set to -5 V. For stressing, the gates were then stepped -1 V/min until they reached -42V. The defects under the gate were studied using transmission electron microscopy (TEM) and scanning TEM (STEM). Diffraction patterns taken from the defective region are amorphous. Additionally, basic compositional information has been gathered using Z-contrast from high-angle annular dark-field STEM (HAADF-STEM) to compare the unstressed and stressed devices. The HAADF-STEM micrographs show a change in shape and contrast of the layers between the pre- and post-stressed samples. The defects are shown to consume the entire gate width and the reaction front appears to stop at the AlGaN/GaN interface. Chemical maps and line scans across the defective regions were obtained by Energy-dispersive X-ray spectroscopy (EDS) and electron energy loss spectroscopy (EELS). EDS and EELS indicate that the defects are a nickel-aluminum-gallium oxide. A model for the defect formation in these AlGaN/GaN HEMTs will be discussed.
10:15 AM - G4.3
No Lateral Composition Fluctuations in Aberration-corrected STEM Images of a InGaN Quantum Well at Low Dose
Andrew B Yankovich 1 Alex V Kvit 1 Xing Li 2 Fan Zhang 2 Vitaliy Avrutin 2 Huiyong Liu 2 Natalia Izyumskaya 2 Umit Ozgur 2 Hadis Morkoc 2 Brandon Van Leer 3 Paul Voyles 1
1University of Wisconsin, Madison Madison USA2Virginia Commonwealth University Richmond USA3FEI Company Hillsboro USA
Show Abstract
Whether or not nanometer-scale composition fluctuations in InGaN thin films can account for the high efficiency of InGaN light emitting devices has been the subject of controversy for some time, especially following the observation that electron irradiation in a TEM can induce composition fluctuations [1]. We have used aberration corrected Z-contrast STEM imaging of the In0.2Ga0.8N quantum well region of a light emitting diode (LED), acquired at ~1% of the electron dose known to cause electron beam damage, to show that no significant lateral composition fluctuations exist. The LED structure was deposited by vertical low pressure metalorganic chemical vapor deposition. The STEM sample was ~3 nm thick, similar to the diameter of previously reported composition fluctuations, so significant composition fluctuations would be clearly visible if they were present. Images of a focused ion beam prepared sample with a very large thin area show fluctuations in the well width which may serve to laterally confine the injected carriers. [1] T. M. Smeeton, M. J. Kappers, J. S. Barnard, M. E. Vickers, and C. J. Humphreys, Appl. Phys. Lett. 83, 5419 (2003).
10:30 AM - G4.4
Characteristics of Polarization Emission in A-plane GaN-based Multiple Quantum Wells Structures
Chiao-Yun Chang 1 Huei-Min Huang 1 Tien-Chang Lu 1 Hao-Chung Kuo 1 Shing-Chung Wang 1 Chih-Ming Lai 2
1National Chiao Tung University Hsinchu Taiwan2Ming Chuan University Taoyuan County Taiwan
Show AbstractIn this study, we investigated the polarization emission properties on a series of a-plane AlGaN/GaN multiple quantum wells (MQWs) grown on r-plane sapphire substrate with the various well width by using polarization-dependent photoluminescence (PL). To clarify the reasons of light emission polarization properties, we applied the 6Ã-6 k.p model to simulate the E-k dispersion relation and the wave functions to obtain the polarization optical transitions. According to the experimental result, the PL emission peak position exhibits a red-shifted with increasing well width, due to the reduction of the quantum confinement effect. The difference in emission peak between the experimental spectrum and the calculated spontaneous emission is associated with the exciton binding energies, which is decreased from 62 to 25 meV as well width increases from 1.5 to 7.33 nm. Furthermore, the relative thick well will induce the smaller energy separation between y-like state and z-like state, but the subband of y-like states is raised toward top of the valence subband level with increasing the well width. Therefore, this phenomenon is believed to be that y-polarized light emission gradually dominates the PL spectrum and thus enhances the degree of polarization.
10:45 AM - G4.5
Magnetic and Optical Properties of Rare Earth-doped GaN Semiconducting Thin Films
Kiran Dasari 1 Javier Wu 1 Ratnakar Palai 1 Wojciech Jadwisienczak 2 Huhtinen Hannu 3
1University of Puerto Rico San Juan Puerto Rico2Ohio University Athens USA3University of Turku Turku Finland
Show AbstractThe rare-earth (RE) doped GaN semiconducting materials are promising candidates for spintronic and optoelectronic applications. The higher magnetic moment of RE ions compared to the transition metal ions makes RE-doped GaN as materials of choice for dilute magnetic semiconductor. REdoped GaN materials have shown the ability to tune the direct bandgap from the ultraviolet through visible to the near infrared region. GaN-doped with Er3+ and Yb3+ are of great interest because of their emission in visible and infrared regime, respectively. The magnetic and electrical properties of thin films of GaN:Yb and GaN:Er have been studied by us for the better understanding of magnetoptic properties. Yb-doped GaN shows room temperature ferromagnetism but Yb ions not optically active. Whereas, Er-doped GaN shows very good visible emission (536nm, 558nm and 668nm) but the ferromagnetism was observed below 150 K. The main objective of our work is to make RE ions in GaN both magnetically and optically active at room temperature. Here we report, the magnetic and optical properties Er+3 and Yb+3 codoped GaN epitaxial thin films grown by molecular beam epitaxy (MBE). Thin films of GaN:ErYb have been grown on Si (1111) and Sapphire (0001) substrates using MBE. The doping concentration and elemental analysis have been calculated from the x-ray photoelectron spectroscopy (XPS). X-ray diffraction and atomic force microscopy (AFM) were used to examine the phase purity and smoothness of the films. The concentration and mobility of carriers were calculated from the Hall effect measurements. The magnetic properties of GaN:ErYb thin films were measured using SQUID and magnetotransport measurements. The optical properties of the films were analyzed by photoluminescence and cathode luminescent spectroscopy. The magnetic and optical properties of the GaN:ErYb films will be compared with Er and Yb-doped GaN films and will be discussed in details.
11:30 AM - G4.6
Strong Polarized Photoluminescence from Nonpolar Indium Nitride Epitaxial Films
Yu-Liang Hong 1 Shangjr Gwo 1
1National Tsing Hua University Hsinchu Taiwan
Show Abstract
Nonpolar III-nitride semiconductors recently have attracted much interest due to the possible improvement in device performance, in comparison with polar counterpoints. For polar III-nitrides, the spontaneous and piezoelectric polarizations along the polar c-axis ([0001]) induce strong internal electrical fields, which would reduce the emission efficiency of light-emitting devices (LEDs and lasers). Therefore, nonpolar III-nitrides in the absence of internal electrical field have been demonstrated to increase the device efficiency. However, the nonpolar a-plane InN epitaxial films were grown on (1-102) r-plane sapphire substrates by using plasma-assisted molecular beam epitaxy. At room temperature, we observed the strong polarized photoluminescence (PL) in high crystal quality of nonpolar InN epitaxial films via the polarization anisotropy measurements of PL spectrum. The temperature dependent (10K ~ 300K) behavior of polarized PL is also performed. The emission intensities in the direction of the electric field perpendicular to the c-axis (IâS¥) are ~ 6 and ~2 times stronger than that in the direction of the electric field parallel to the c-axis (Iâ^¥) at 10K and 300K, respectively. Furthermore, the anisotropy ratios, defined as Ï= (IâS¥-Iâ^¥)/(IâS¥+Iâ^¥), are 0.65 and 0.28 at 10K and 300K. To evaluate the polarized PL properties, we also calculated the variation of transition energies of the strained a-plane InN film by using the modified electronic band structure with the kâ-p approximation. The anisotropic in-plane strains were analyzed by the X-ray diffraction measurement. It is found that the temperature dependent polarization anisotropies from InN epitaxial films estimated from calculated transition energies and relative oscillator strengths are consistent with the experimental results.
11:45 AM - G4.7
Diffusivity of Mn Interstitials in Thin (Ga,Mn) as Layers Studied by High-resolution X-Ray Diffraction
Lukas Horak 1 Jana Matejova 1 Xavi Marti 1 Vaclav Holy 1 Vit Novak 2 Zbynek Soban 3 2 Stefan Mangold 4 Felix Jimenez-Villacorta 5
1Charles University Prague Czech Republic2Institute of Physics ASCR Prague Czech Republic3The Czech Technical University in Prague Prague Czech Republic4Karlsruhe Institute of Technology Karlsruhe Germany5Northeastern University Boston USA
Show AbstractThe (Ga,Mn)As material belongs to the diluted magnetic semiconductors, which are very promising materials for spintronic applications. The temperature of the magnetic ordering of (Ga,Mn)As is still below the room temperature, it is decreased among others due to the presence of the interstitial Mn ions. These and other defects (such as As antisite defects) are created during the low temperature growth, which is necessary to prevent the formation of the MnAs clusters. The amount of the interstitials can be reduced by the post-growth annealing, which leads to the out-diffusion of the interstitials to the free surface. Although this material has been intensively studied for last decade, the process of the out-diffusion is not yet fully understood. We present a method for the determination of the concentration depth profiles of Mn interstitial ions in (Ga,Mn)As thin epitaxial layers using high-resolution x-ray diffraction.The measured diffraction curves for several diffraction maxima hkl were compared to the theoretical curves, which were calculated using standard dynamical diffraction theory. The concentration of the interstitial Mn ions affects the structure factor and the lattice parameter, and we optimized concentrations of Mn ions in various lattice positions to match the experimental and the theoretical curves. From the asymmetry of the intensities of the thickness fringes it is possible to characterize an eventual depth inhomogeneity of the interstitial density in the (Ga,Mn)As layer. The epitaxial GaMnAs layers under study were grown on (001)GaAs substrates by molecular beam epitaxy. A nondestructive character of the characterization (in contrast for instance to the transmission electron microscopy) allows to investigate the same sample in various annealing states. We measured an as-grown sample, then after 24 hours of annealing in the air at 160°C and finally after 20 cycles of etching and short annealing (under the same conditions as in the previous case). We have determined the depth profiles of the interstitial density for all samples and these have been compared to the numerical simulations of the interstitial drift-diffusion in the sample. From these comparisons the diffusivity of the Mn ions in the (Ga,Mn)As lattice has been estimated. The results show that the flux of the Mn ions towards the free surface is strongly affected by the internal electric field produced by inhomogeneously distributed holes.
12:00 PM - G4.8
Fast Deposition of Epitaxial CdZnTe Film on (001)GaAs through a One-step Process by Close-spaced Sublimation and the Related Defect Issues
Junning Gao 1 Wanqi Jie 1 Tao Wang 1 Yanyan Yuan 1 Junli Tong 1
1Northwestern Polytechnical University Xi'an China
Show AbstractCdZnTe films have been the interest of many researches engaged in the fabrication of large area radiation detectors for medical imaging as an alternative to those expensive and size limited bulk crystals.Certain thickness is required for the effective absorption of x-ray photons. For example, 10~20 μm thick CdZnTe layer is needed for the absorption of 8~20 keV photons and nearly 500 μm is needed for 140 keV photons. For thick film epitaxy, method such as MBE suffers from the low growth rate, and MOCVD suffers from the high cost of the source materials. Close-spaced sublimation (CSS) has good potential in thick film epitaxy due to its high growth rate, none strict requirement on vacuum and simplicity in equipment structure, either powder or bulk source materials can be used to grow single crystal layers, and all these merits can be achieved with little sacrifice in the film quality . Hillocks and dual-epitaxy (film with two competitive orientations) are frequently observed in poor epitaxial CdZnTe films. Hillocks form because of the penetration of the stacking faults in the film to the surface, while the stacking fault energy in CdZnTe is rather low. As far as dual epitaxy, films can be either (111) or (001) oriented, or sometimes even oriented with both directions on (001)Si or (001)GaAs. This makes the epitaxy of CdZnTe films need delicate control. In this talk, a simple one-step process for the epitaxy of the high quality thick epitaxial (001)CdZnTe film on (001)GaAs via CSS will be addressed, where the deposition rates are up to 1 μm/min. Well-defined epitaxial features of the films have been proved by plan view SEM, XRD θ-2θ and Φ scan, and the cross-sectional SEM and HRTEM analysises. The epitaxy is completed through a novel growth mode with the film experiencing a polycrystalline to single crystalline transition, where the transition is completed through the lateral overgrowth and the self-organization of the laminar grains. The FWHM of x-ray rocking curve of the best film obtained so far is 306 acrsec, as recorded from the (400) diffraction. The defects were studied with SEM and HRTEM. The films obtained at lower source and substrate temperatures have dense elongated hexagonal pits, while those obtained at the elevated temperatures have smooth surface morphology. TEM bright field imaging shows that most defects in the epitaxial layer are localized near the interface and few stacking fault in the film penetrates to the film surface. TEM lattice imaging shows that L-shaped defects made up of a (111) stacking fault and several compressed (-1-11) layers exist near the interface. The L-shaped defects form because the lattice of the CdZnTe film match with the GaAs substrate causing compressed (-1-11) planes instead of relaxing the strain by misfit dislocations in some area. The interfacial misfit dislocation have the Burgers vectors of either b=a/6[112] or b=a/2[110], and those within the film only have the vectors of b=a/6[112].
12:15 PM - G4.9
Compensation Mechanism in Highly Resistive CdTe for Application in Radiation Detectors
Koushik Biswas 1 David J Singh 1 Mao-Hua Du 1
1Oak Ridge National Laboratory Oak Ridge USA
Show AbstractCdTe, a Group II-VI semiconductor, finds applications in a variety of technological areas including semiconductor radiation detectors. Some of the properties necessary for an optimum radiation detector material are high density, high resistivity, good mobility-lifetime product of charge carriers and right band gap for room temperature application. CdTe has a band gap of ~1.5 eV which is considered ideal for room temperature operation. On the other hand, material defects play a very important role in determining its transport properties and resistivity. Using state of the art first-principles calculations based on hybrid density functional methods, we have studied the various defects in Cl- and In-doped CdTe as well as undoped CdTe with the aim to assess the compensation mechanisms that can cause high resistivity. We show that high resistivity in CdTe is not due to Fermi level pinning by deep levels. Instead, the carrier compensation between shallow donors and Cd-vacancies or A centers (defect complex comprised of a shallow donor and Cd-vacancy) can explain the high resistivity and the observed shallow donor concentrations. We find that the Fermi level can be confined within a small range near midgap even in the presence of reasonably large variation in shallow donor concentration (i.e. 10^17 - 10^19 cm-3 for CdTe:Cl and 10^16 - 10^17 cm-3 in CdTe:In). Therefore, stable high resistivity can be obtained despite the existence of doping level fluctuations during the crystal growth. One should utilize shallow donors and shallow acceptors and optimize their concentrations to obtain good carrier compensation and high resistivity, while deep centers should be avoided due to their detrimental effect on carrier transport.
12:30 PM - G4.10
Ab initio Shallow Impurity Level Calculations in Semiconductors
Gaigong Zhang 1 Andrew Canning 2 1 Niels G Jensen 1 2 Stephen E Derenzo 3 Lin-Wang Wang 4
1UC Davis Davis USA2Lawrence Berkeley National Laboratory Berkeley USA3Lawrence Berkeley National Laboratory Berkeley USA4Lawrence Berkeley National Laboratory Berkeley USA
Show AbstractBinding energies of Group III shallow acceptors in Si, Group IV shallow acceptors in III-V semiconductors and Bi in GaP as isoelectronic shallow acceptors were calculated using a local density approximation (LDA) + GW procedure. We first performed density functional theory calculation within local density approximation (LDA), with the impurity potentials to be corrected by GW calculations. In the LDA calculation, large supercells containing tens of thousands of host atoms and the center impurity atom were constructed from a potential patching procedure. 64-atom GW calculations improved the description of the central potential of the systems. Then the folded spectrum method was used to calculate the eigen-energies of these large supercells containing the center impurity. The calculated binding energies agree well with experimental values. This procedure represents an efficient approach to study shallow impurity levels that are important for semiconductor devices.
12:45 PM - G4.11
Switchable Polarization in Stuffed Wurtzite Compounds, a New Class of Ferroelectrics
Joseph William Bennett 1 Kevin F Garrity 1 Karin M Rabe 1 David Vanderbilt 1
1Rutgers Univ Piscataway USA
Show Abstract
One of the challenges in modern materials science is the development of new types of ferroelectric materials, in which a response is produced by an applied field or stress. Because of their many technological applications, including transducers and information storage, the discovery of new ferroelectric materials continues to be of great importance. For a material to be ferroelectric, it must be insulating and have a nonzero polarization that can be switched by an external electric field. The wurtzite structure type, including a number of technologically important materials such as ZnO, CdSe (II-VI) and GaN (III-V), is generally considered to be the result of directional sp3 bonding, with all atoms in tetrahedral coordination. These compounds thus satisfy the first two conditions, but not the third, as the switching of the polariation by 180 degrees would require breaking and reforming the strong sp3 bonds, and the energy barrier is far too high. Indeed, in the well-known ABO3 perovskite oxide ferroelectrics, displacements of a B-site within an octahedral O6 environment and an A-site within a dodecahedral O12 environment do not involve the breaking and reforming of chemical bonds, only changes in bond length. However, if the wurtzite structural parameters are not close to ideal tetrahedral sp3 coordination, but rather have values that correspond to a small buckling of the planes, then similarly the barrier to polarization switching can become small. This type of wurtzite structure can be found as a substructure of the ternary LiGaGe structure type, where the third "stuffing" atom controls the size of the buckling. A total electron count of 8 or 18 electrons favors insulating character. In this talk, using first-principles methods, we investigate about fifty members of this family, and find that both the polarization and barrier to switching depend upon the stuffing atom. Moreover, we demonstrate that the members of the family with a large stuffing atom have barriers to polarization switching on the order of those calculated for the perovskite family. Information is presented that will be useful in experimental investigation of this proposed new class of ferroelectrics.
Symposium Organizers
Osamu Ueda, Kanazawa Institute of Technology Graduate School of Engineering
Mitsuo Fukuda, Toyohashi University of Technology
Kenji Shiojima, University of Fukui Graduate School of Electrical and Electronics Engineering
Edwin Piner, Texas State University, San Marcos
Symposium Support
Japan Society of Applied Physics
G9: Electron Devices and Reliability
Session Chairs
Thursday PM, April 12, 2012
Moscone West, Level 2, Room 2007
2:30 AM - G9.1
Determination of AlGaN/GaN HEMT Reliability Using Optical Pumping as a Characterization Method
David Cheney 1 Brent Gila 2 Fan Ren 3 Pat Whiting 2 Jennilee Navales 1 Erica Douglas 2 Steve Pearton 2
1University of Florida Gainesville USA2University of Florida Gainesville USA3University of Florida Gainesville USA
Show Abstract
One of the main attractions to GaN semiconductors is its thermal stability due to its wide band-gap. Another benefit to the wide band-gap is the ability to pump light from the visible spectrum into a III-V nitride device using below band-gap energy levels. Using visible light is attractive because of the availability of low cost light sources, such as lasers, at many different wavelengths. By pumping AlGaN/GaN HEMTs with below band-gap light we observe changes in drain current that correspond to the trapping and detrapping of carriers within the band-gap. These changes in drain current are indicators of trap density, since the energy from a specific wavelength of light pumps traps whose activation energies are less than or equal to that of the light source. Once the baseline changes in drain currents, i.e., the initial trap densities, are determined by pumping different wavelengths from the visible spectrum, the device is electrically stressed. Comparing a stressed device to its baseline determines if new traps are formed along with their activation energies. The activation energies are derived from the photon energy of the pumped wavelength of light. Many trap activation energies have been identified in the literature which allows for the identification of trap type such as point defects (vacancies, substitutions, etc), stacking faults, and edge dislocations. Identifying the development and persistence of trap types while going through acceleration aging allows for the correlation of trap types and device reliability. AlGaN/GaN HEMTs on SiC with dual submicron gates with widths of 125nm, 140nm, or 170nm, are DC-stressed under three different conditions along a load line: VGS=0, VDS=5 (on-state), VGS=-2, VDS=9.2 and, VGS=-6, VDS=25 (off-state). The stress tests are interrupted at 20% degradation and the optically pumped comparisons to the baseline are measured. The hypothesis is devices stressed in the on-state will show mostly channel and contact degradation, while the devices in the off-state have gate issues, and devices in the middle of the load line show degradation of both the gate and channel. Initial results have shown a change in drain current of up to 30% when a device is pumped with UV (above band-gap) light when stressed under high power conditions, while a less significant change of 12% is observed in the off-state indicating more traps have formed in the high power tests. This paper describes the optical pumping technique and results from experiments of AlGaN/GaN HEMTs under the three DC stress biases along a load line.
2:45 AM - G9.2
Impact of the Al Mole Fraction in the Bulk- and Surface-state Induced Instability of AlGaN/GaN HEMTs
Sandeepan DasGupta 1 Andrew Armstrong 1 Robert Kaplar 1 Matthew Marinella 1 James Stanley 1 Mark Smith 1 Stanley Atcitty 1 Tomas Palacios 2 Min Sun 2
1Sandia National Labs Albuquerque USA2MIT Cambridge USA
Show AbstractThe increasing prospect of AlGaN/GaN HEMTs in power electronics necessitates channels with relatively low 2DEG densities. This makes the development of stable AlGaN/GaN material systems with low Al molefraction a priority. Also of particular interest for power devices is the modulation of electric fields through surface state engineering. This necessitates a thorough understanding of surface states in devices with varying molefraction and passivation. Thermal dependence of reversible degradation in off-state and complete absence of light as well as the thermal and optical wavelength dependence of recovery is used on AlGaN/GaN HEMTs with 15 % and 26 % Al AlGaN with and without passivation to evaluate the Al molefraction dependence of bulk and surface states in AlGaN. Devices are stable for long stress times in presence of ambient light. Unpassivated samples, expectedly show high density of surface defects producing large instability in off-state bias under complete darkness. The physical location of this is the access region, resulting in negligible shifts in threshold voltage but large increase in series resistance. Short term (~15 min.) thermal recovery of defects at room temperature and no light shows a stretched exponential behavior â?" representative of a continuum of states. Temperature dependence of the recovery shows the continuum of states to have activation energies (Ea) ~ 0.3 - 1.08 eV. In unpassivated devices â?" this is the major component of instability, accounting for ~ 90 % of the total drop in drain current. However, contrary to the defects responsible for permanent hot carrier and strain related damage, the unpassivated Al0.15Ga0.85N/GaN devices show a higher surface state density compared to unpassivated Al0.26Ga0.74N/GaN (by a factor of ~2). It is hypothesized that the high affinity of oxygen for Al could result in Al0.26Ga0.74N having a higher oxygen concentration, which could potentially passivate some surface states. 10 â?" 20% of the reduction in drain current is not recoverable at room temperature within short times in complete darkness. Irradiation of degraded devices with monochromatic light shows this component of degradation to be related to a discrete level with Ea~2.3 eV, presumably a bulk state. Passivated devices (Al2O3/SiO2) show substantially improved overall stability due to almost complete elimination of the surface states. However, passivated devices in Al0.15Ga0.85N/GaN devices as well as unpassivated Al0.26Ga0.74N/GaN devices also show substantial increase in trapping due to the deep bulk level (Ea ~ 2.3 eV) which demonstrates the potential of surface state engineering in improving stability and high voltage performance. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000. The MIT portion of this work has been partially funded by the DOE GIGA program.
3:00 AM - G9.3
SEM and CL Investigation of Defects in Ni/Au Gate Electrode AlGaN/GaN HEMTs
Patrick Guzek Whiting 1 Monta R Holzworth 1 Erica A Douglas 1 Kevin S Jones 1 Brent D Gila 1 Lu Liu 2 Fan Ren 2 Eric R Heller 3 Stephen K Tetlak 3
1University of Florida Gainesville USA2University of Florida Gainesville USA3Air Force Research Laboratories Wright-Patterson Air Force Base USA
Show AbstractAlGaN/GaN HEMTs are employed in wider and wider ranges of applications in high power and microwave circuits. Degradation in in the field over long time periods and during accelerated testing is an ongoing concern for the technology, however. There has been recent interest in degradation mechanisms associated with interactions between the metal gate contact and the underlying AlGaN and GaN semiconductors due to the inverse piezoelectric effect and/or chemical reactions. Performance loss associated with increasing contact resistance is also a noted phenomenon during stressing. In this study, the effects of electrostatic and thermal stressing on the AlGaN/GaN semiconducting layers of submicron gatelength HEMTs posessing a Ni/Au gate stack and Ti/Al/Ni/Au ohmic contacts were tested. The samples were deprocessed using multiple wet chemistries starting with 6:1 BOE for SiN removal, followed by an FeCN/KI solution for metal removal and, lastly, a solvent bath for ultrasonic cleaning. Samples were evaluated using scanning electron microscopy, transmission electron microscopy and cathodoluminescence. These studies show, for the first time, radiating crack formation occurs during ohmic contact fabrication prior to stressing. The cracks, which are approximately 10nm deep and up to 2 microns long, are likely caused by strain induced during the rapid thermal annealing of the ohmic contact. In addition to cracks, pitting under the Ni gate structure was also observed. Consumption of the AlGaN layer by the Ni liner layer of the Ni/Au gate contact was observed to increase as a result of off state stressing. The gate leakage current density appears to increase linearly with the percent area of the gate reacted at a ratio of approximately 3.6 %-cm2/kA. The resistivity of a fully reacted gate contact is equal to approximately 68 mΩ/cm2 . The role of gate current and electrostatic field on the reaction under the gate will be discussed.
3:15 AM - G9.4
Leakage Current and Schottky Barrier Height Variation in III-Nitride Heterostructures
Saurabh Pandey 1 Albert Minj 1 Beatrice Fraboni 1 Daniela Cavalcoli 1 Anna Cavallini 1 Tomasso Brazzini 2 Fernando Calle 2
1University of Bologna Bologna Italy2ISOM, Universidad Politeacute;cnica de Madrid Madrid Spain
Show AbstractAl(In)Ga(Al)N/GaN heterostructures have been the focus of intense research to improve the performance of high mobility field-effect transistors (HEMTs) [1]. Due to the high spontaneous and piezoelectric polarization, these structures present a two dimensional electron gas (2DEG) at the heterointerface between AlN and GaN. The Schottky barrier height of Al(InGa)N/GaN heterostructures is of great importance for device performance and reliability in terms of leakage current and high breakdown voltage. The presence of a high dislocation density and the segregation of metals make devices very leaky [2]. The formation of good quality ohmic and gate contacts is a key factor for a reliable electrical characterization of these devices. Here, we present a detailed analysis on the leakage current variation in MOCVD grown InAlN/AlN/GaN heterostructures as a function of the AlN interlayer thickness. The effect of the AlN interlayer thickness on the 2DEG concentration and mobility has been also considered [3]. Ti/Al/Ni/Au ohmic contacts and Ni/Au Schottky barriers were formed directly on the InAlN layer, without using any cap layer. Temperature dependent Current-Voltage measurements were performed in forward and reverse bias. The leakage mechanism has been identified as a Poole-Frenkel emission through a trap state. The Schottky barrier height has been also observed to vary as a function of the dislocation density and the AlN interlayer thicknesses. The presence of surface states could be responsible for the variation in barrier height and could also influence the 2DEG concentration. We propose a model to calculate the surface state concentration and its activation level. Our studies contribute to the improvement of devices fabricated on Al(InGa)N/GaN heterostructures by optimizing the AlN interlayer thickness, the leakage current, and by stable gate contact for better device development. References 1. T. Palacios, A. Chakraborty, S. Rajan, C. Poblenz, S. Keller, S. P. Den-Baars, J. S. Speck, and U. K. Mishra, IEEE Electron Device Lett. 26, 781 (2005) 2. A. Minj, D. Cavalcoli, A. Cavallini, Appl. Phys. Lett. 97, 132114 (2010) 3. S. Pandey, B. Fraboni, D. Cavalcoli, A. Minj, A. Cavallini, Appl. Phys. Lett. 99, 012111 (2011) Acknowledgements This work has been performed in the EU frame FP7 of the â?oRAINBOWâ? Initial Training Network (PITN-GA-2008-213238).
3:30 AM - G9.5
Thermal Management of GaN High-Power Devices with the ``Graphene Quilts''
Zhong Yan 1 Guanxiong Liu 1 Javed Khan 1 Alexander Balandin 1
1UC Riverside Riverside USA
Show AbstractAlGaN/GaN heterostructure field-effect transistors (HFET) are attractive devices for the high-frequency high-power applications. However, self-heating is a severe problem that limits the performance of GaN power devices. Temperature rise in AlGaN/GaN transistors caused by the non-uniform heat generation and heat dissipation at high power density results in degradation of the drain current, gain and output power, as well as increase of the gate-leakage current and poor reliability. Here, we demonstrate that thermal management of AlGaN/GaN HFET can be substantially improved via introduction of the additional heat-escaping channels â?" top-surface heat spreaders â?" made of few-layer graphene (FLG). It is known that FLG reveals an order-of-magnitude higher thermal conductivity (~ 2000 W/mK) [1-2] than that of GaN. The FLG films have been transferred to AlGaN/GaN HFET on SiC substrate to form the "graphene quilts" â?" lateral heat spreaders, which remove heat from the channel region. Using a micro - Raman spectrometer for the in-situ temperature monitoring we have shown that temperature can be lowered by as much as ~ 20oC in such devices operating at ~13-W/mm power density. The local temperature in these measurements was obtained from the shift in the position of GaN and SiC Raman peaks. It was experimentally established that the degradation of the saturation current in GaN transistors with the FLG heat spreaders is less than that in the transistors without the heat spreaders. The finite-element simulations confirm the experimental observations and suggest that the efficiency of â?ographene quiltsâ? can be much higher in AlGaN/GaN HFETs on thermally resistive sapphire substrates and in multi-finger devices. Our results open a novel application niche for FLG in high-power electronics. The work at Balandin group in UCR was supported by the US Office of Naval Research (ONR) through award No. 00014-10-1-0224 on Graphene Lateral Heat Spreaders for GaN Power Electronics. [1] S. Ghosh, W. Bao, D.L. Nika, S. Subrina, E.P. Pokatilov, C.N. Lau and A.A. Balandin, Dimensional crossover of thermal transport in few-layer graphene, Nature Mat., 9, 555 (2010). [2] A.A. Balandin, Thermal properties of graphene and nanostructured carbon materials, Nature Mat., 10, 569 (2011).
3:45 AM - G9.6
N-type and P-type High Mobility III-V Semiconductors on Si/SiO2 Substrates
Kuniharu Takei 1 Steven Chuang 1 Hui Fang 1 Rehan Kapadia 1 Ali Javey 1
1Univ California Berkeley Berkeley USA
Show AbstractAs Si-based electronics are reaching their limit in scalability, there has been a renewed interested in materials research for next generation low power devices. Recently, there has been significant progress in the use of epitaxial growth and epitaxial layer transfer techniques utilizing III-V semiconductor materials. To date, we have reported ultrathin (3-48 nm) III-V layers transferred onto Si substrates (X-on-insulator, XOI), enabling the exploration of III-V device properties and fundamental carrier transport properties as a function of body thickness, without the constraints of the original growth substrate [1-3]. Specifically, top-gated n-type (InAs) and p-type (InGaSb) XOI field-effect transistors (FETs) were fabricated on Si substrates with ~10 nm thick ZrO2 gate dielectrics. The FETs exhibit high electron and hole mobilities of ~4000 and 850 cm2/Vs, respectively, which highlight the utility of the approach for obtaining high performance III-V devices on conventional Si substrates. The OFF-state leakage mechanisms of the XOI devices were explored as a function of body thickness with the results highlighting the dominant role of Shockley Read Hall generation/recombination and trap assisted tunneling. The OFF-state currents were shown to drastically reduce with body thickness miniaturization, demonstrating the importance of ultrathin body device architecture for obtaining respectable ON/OFF current ratios. InAs XOI FETs exhibit a low subthreshold swing of ~75 mV/decade which suggest high quality interfaces with low density of trap states. In addition, the gate delay limits of InAs XOI FETs are discussed and projected as a function of body thickness and channel length. This work presents an important advance towards the use of high mobility III-V material systems on Si substrates with various material, device and processing aspects of the associated technology being explored in detail. The results are promising for future low power CMOS electronics. Reference [1] H. Ko, K. Takei et al, Nature 468, 286-289, 2010. [2] K. Takei et al. Applied Physics Letters 99, 103507, 2011. [3] K. Takei et al. Nano Letters, ASAP, 2011.
G10: Radiation Effect and Electron Devices
Session Chairs
Thursday PM, April 12, 2012
Moscone West, Level 2, Room 2007
4:30 AM - *G10.1
Research amp; Development Activities on Radiation Effects in Materials Supported by the IAEA
Aliz Simon 1 Andrej Zeman 1 Ralf Kaiser 1 Mohammad Haji-Saeid 1 Agnes Safrany 1 Francoise Mulhauser 1
1International Atomic Energy Agency Vienna Austria
Show AbstractWithin the International Atomic Energy Agency (IAEA) Department of Nuclear Sciences and Applications, activities are being carried out to assist and advise IAEA Member States in assessing their needs for capacity building, research and development in nuclear sciences. Moreover, support is also provided to Member Statesâ?T activities geared towards deriving benefits in fields such as (i) advanced materials for nuclear applications, (ii) application of accelerators and associated instrumentation, and (iii) nuclear, atomic and molecular data. Electron beam irradiation of polymeric materials may cause either predominantly chain scission or crosslinking. Chain scission usually leads to deterioration of mechanical properties and color change, but by controlled degradation new material properties can be introduced. Crosslinking can improve polymer material properties, such as hardness, wear and tear resistance and introduce shape-memory effects. The IAEA activities include among others a Coordinated Research Project (CRP) on â?oNanoscale radiation engineering of advanced materials for potential biomedical applicationsâ?. Neutron-induced damages to materials can be very detrimental for nuclear power reactors and other high neutron flux environments. The IAEA coordinates development in the field of high dose radiation effect on core structural materials in advanced nuclear systems (fusion and fission). Microelectronic devices and detectors in radiation harsh environments such as e.g. satellites, high energy physics experiments or medical applications, are exposed to high energy radiation. This radiation causes a variety of damages to their materials structure and electronic properties. Ions from accelerators have a prominent role in testing and developing electronic materials and detectors as they are suitable for introducing controlled damage based upon the possibility to define with high accuracy the ion fluence, determine the damage profile and localize the damaged region. The IAEA supports the improvement of the utilization of ion beam techniques in investigating radiation hardness of semiconductor materials and devices in a form of a CRP. The goal is to provide reliable modelling of the damage creation process by ions in semiconductors which will enable easy comparison of hardness tests performed on different semiconductor materials and obtained by other means of radiation.
5:00 AM - *G10.2
8 MeV Proton Irradiation and Post-annealing Effect on Single-crystalline Zinc Oxide Crystals
Kazuto Koike 1 Takahiro Aoki 1 Ryugo Fujimoto 1 Shigehiko Sasa 1 Mitsuaki Yano 1 Shun-ichi Gonda 2 Ryoya Ishigami 3 Kyo Kume 3
1Osaka Institute of Technology Osaka Japan2Osaka University Osaka Japan3The Wakasa Wan Energy Research Center Fukui Japan
Show Abstract
Kazuto Koike1, Takahiro Aoki1, Ryugo Fujimoto1, Shigehiko Sasa1, Mitsuaki Yano1, Shun-ichi Gonda2, Ryoya Ishigami3 and Kyo Kume3; 1Nanomaterials Microdevices Research Center, Osaka Institute of Technology, Osaka, Japan; 2The Institute of Scientific and Industrial Research, Osaka University, Osaka, Japan; 3The Wakasa Wan Energy Research Center, Fukui, Japan.
Recent progress in ZnO crystal growth techniques enables us to utilize single-crystalline bulk crystals as well as single-crystalline epitaxial films for UV emitters, UV detectors and high-power transistors. Similar to other wide-bandgap semiconductors of GaN and SiC, ZnO exhibits great potential for high radiation hardness and is expected to be useful for the device applications for space and nuclear electronics, since its displacement threshold energy is expected to be large due to the small unit-cell volume and large bandgap energy. In this report, we study the effect of 8 MeV proton irradiation and post-annealing on single-crystalline ZnO crystals. We prepared single-crystalline c-axis-oriented ZnO films with the thicknesses of 900 nm and 50 nm grown on a-plane sapphire substrate using molecular beam epitaxy, and a vapor-phase-grown single-crystalline c-axis-oriented ZnO bulk crystal, in addition to a corresponding GaN bulk crystal for comparison. These four different samples were mounted on a chilled stage in a vacuum chamber, and irradiated simultaneously with 8 MeV protons in the wide fluence range from 2x1013 to 1x1017 p/cm2 using a tandem-type accelerator. It was revealed that the electrical resistance of these samples was increased by the radiation damage, and the threshhod fluence for the resistance increase was higher for the ZnO bulk crystal than that for the GaN bulk crystal. Among the ZnO samples, the threshold fluence was increased by lowering conduction dimensions, presumably due to a carrier localization effect. Indeed their room-temperature photoluminescence was dominated by a free-excitonic emission peak even after the maximum irradiation, but the threshold fluences for the peak intensity deterioration were ~3x1013 p/cm2 for the GaN bulk crystal, ~2x1014 p/cm2 for the ZnO bulk crystal, and ~5x1014 p/cm2 for the ZnO films, in accordance with the order of the radiation hardness in electrical resistance. These experimental results indicate that ZnO is harder than GaN, and ZnO thin films with low-dimensional carriers are harder than the bulk crystal. We attribute the higher radiation hardness of ZnO material to the higher ionicity; the larger ionic radius of Zn2+ than that of Ga3+ with a more flexible bond-angle may help the rapid diffusion of the major knocked-on atoms (Zn and Ga) to result in in-situ restoration of radiation damage during irradiation. The effect of the post-irradiation annealing on the damaged samples will also be discussed at the conference.
5:30 AM - G10.3
An Investigation of Arsenic and Antimony Capping Layers, and Half Cycle Reactions during Atomic Layer Deposition of Al2O3 on GaSb(100)
Dmitry M. Zhernokletov 1 Hong Dong 2 Barry Brennan 2 Jiyoung Kim 2 Serge R Oktyabrsky 3 Robert M Wallace 1 2
1University of Texas at Dallas Dallas USA2University of Texas at Dallas Richardson USA3University at Albany Albany USA
Show AbstractA GaSb interfacial passivation layer could be more favorable than an AlyIn1-ySb barrier layer for QWFETs or an AlyGa1-ySb channel layer for H-TFETs when coupled with high-k dielectric integration, since it does not contain Al as this avoids surface interaction between oxygen originating from the dielectric and Al metal in the barrier/channel layer. For this reason it is necessary to carefully investigate the interface between GaSb and high-κ dielectric materials which would act as gate dielectrics in these devices [1]. Some of the major challenges affecting the electrical characteristics of the high-k/GaSb semiconductor devices are its high sensitivity to initial surface chemical treatments and the high-k oxide deposition temperature, which could result in Fermi level pinning and increased defect density [2]. This study aims to investigate the interface between oxide free and chemically treated GaSb surfaces, and in-situ ALD Al2O3, in terms of both physical characterization by X-ray photoelectron spectroscopy (XPS) and electrical measurements from MOS capacitors. In-situ monochromatic XPS is used to examine the GaSb(100) surface after thermal desorption of a protective As/Sb layer and subsequent ALD of Al2O3. An antimony protective layer is found to be more favorable compared to an arsenic capping layer as it prevents As alloys from forming with the GaSb substrate. The effect of different oxide deposition temperatures on Al2O3/GaSb interface is investigated in terms of interfacial interactions, oxide quality and growth rate. This work is supported by the Semiconductor Research Corporation FCRP MSD Focus Center, the Nanoelectronics Research Initiative and the National Institute of Standards and Technology through the Midwest Institute for Nanoelectronics Discovery (MIND) and the NSF (ECCS-0925844). [1] A. Nainani, T. Irisawa, Z. Yuan, Y.Sun, T. Krishnamohan, M. Reason, B.R. Bennett, J.B. Boos, M. Ancona, Y. Nishi, K.C. Saraswat, International Electron Devices Meeting, (IEDM) Tech. Dig. (2010) 138. [2] A. Ali, H.S. Madan, A.P. Kirk, D.A. Zhao, D.A. Mourey, M.K. Hudait, R.M Wallace, T.N. Jackson, B.R. Bennet, J.B. Boos, S. Datta, Appl. Phys. Letters 97, 143502(2010)
5:45 AM - G10.4
Quantitative Analysis of Electrical Hysteresis in ZnO Field Effect Transistors by Equivalent Circuit Modeling
Hyunjin Ji 1 Jaewan Choi 1 Yun-Jeong Kim 1 Gyu-Tae Kim 1 Chan-Woong Na 2 Jong-Heun Lee 2
1Korea University Seoul Republic of Korea2Korea University Seoul Republic of Korea
Show AbstractZnO materials have a big potential for the device applications of optoelectronics, bio-, photo- or gas sensors with the wide direct band-gap energy, a large exciton binding energy and a high sensitivity or a good compatibility, etc. Furthermore, it could be synthesized in nanostructures with various geometric shapes and easily functionalized with chemicals for sensors. So still the various researches with ZnO in various fields are under investigations. Even though, there are several problems to be solved for the industrial application such as the instability in electrical properties related with the degradation, difficulties in p-type doping. In this experiment, the electrical hysteresis in ZnO field effect transistors were studied by comparing the hysterical phenomena in thin film transistors, nanobelt field effect transistors and the reported devices from the point of the equivalent circuit model. The electrical hysteresis could be often found in devices with the different direction of the sweep, resulting in the different current values at the same bias voltages. It has been reported usually in CNT related materials or organic materials devices, which were attributed to the interaction with water molecules in the ambient. In the case of our ZnO TFTs, the electrical hysteresis was observed larger with the drain bias rather than the gate bias with a counter clockwise direction. On the other hand, ZnO nanobelt FETs showed the clockwise hysteresis with the gate bias in transfer characteristics. The direction of the hysteresis was the main factor, which can be related with the surface states in the gate insulator or the channel. So, two kinds of trap sites were considered to explain the different direction of the hysteresis; the interface trap sites and channel side trap sites. When the bias was swept firstly upward, the interface trap sites were filled with free carries, reducing the channel conductance at the sweep-down with the counter clockwise hysteresis in ZnO TFTs. In the case of ZnO nanobelt FETs, clock-wise hysteretic transfer curves were observed owing to the consumption of the charge carriers to the adsorbed water molecules or oxygen ions onto a large amount of channel traps due to the high exposed surface to volume ratio. The interface traps were modeled as a resistor(R) and a capacitor(C) for the gate side and the channel traps with the segmented many parallel R/C into the conventional metal oxide semiconducting FET (MOSFET) circuit. These equivalent circuits could explain the experimental data. The reasonable mechanism and circuit modeling could be very valuable in term of the prediction and the control of the electrical properties of devices, improving the electrical stability for the realization of industrial applications.
G8: HEMT Reliability
Session Chairs
Thursday AM, April 12, 2012
Moscone West, Level 2, Room 2007
9:30 AM - *G8.1
Reliability and Degradation Phenomena in InP-Based HEMTs
Tetsuya Suemitsu 1
1Tohoku University Sendai Japan
Show AbstractInP-based high electron mobility transistors (HEMTs) are one of important devices for ultrahigh-speed optical fiber communication, wireless communication with millimeter-wave and above frequencies, and scientific applications such as radio astronomy. Many of those applications require high level of reliability in their systems and the devices therein. For 100-nm gate InP HEMTs, the reported median time to failure (MTTF) at a junction temperature of 125°C was improved from 1E5 h in the early 1990s to over 1E7 h in the early 2000s [1,2]. The term â?~InP HEMTsâ?T originally means the HEMTs grown on InP substrates. It does not mean the heterostructure containing InP in active layers. In fact, in most InP HEMTs, the fundamental part of the active layers consists of an InGaAs as a channel layer and an InAlAs as a carrier supply layer. However, many reliability issues are solved by additional layers consisting of InP or other phosphides. For example, the InAlAs carrier supply layer causes a serious problem in reliability because the fluorine (F) contamination inactivates the silicon donors in InAlAs [3]. The carrier supply layers consisting of a Si-doped InP layer sandwiched by undoped InAlAs layers successfully eliminate such a problem [4]. The InAlAs widely exposed at the gate recess area also becomes a source of instabilities such as the kink effect and the frequency dispersion of transconductance. An insertion of an InP surface passivation layer solves these problems [5]. As a result, one can expand the lateral depth of the gate recess to mitigate the maximum electric field at the drain edge of the gate. This helps improve the breakdown voltage and reduce the output conductance in the transistor characteristics. In addition, this InP layer also plays an important role on the control of the threshold voltage because it acts as an etch stop layer. Consequently, the InP thin layers in the HEMT structure are indispensable in todayâ?Ts reliable HEMT structures. [1] D. J. LaCombe, W. W. Hu, F. R. Bardsley, 31st Int. Reliability Phys. Symp., Mar. 23-25, 1993, p. 364. [2] M. Chertouk, W. D. Chang, C. G. Yuan, H. H. Chen, L. Lo, C. H. Chen, D. W. Tu, J. Liu, N. Draidia, P. C. Chao, GaAs MANTECH Conf., May 19-22, 2003, p. 153. [3] N. Hayafuji, Y. Yamamoto, N. Yoshida, T. Sonoda, S. Takamiya, S. Mitsui, Appl. Phys. Lett., 66 (1995) 863. [4] T. Suemitsu, Y. K. Fukai, H. Sugiyama, K. Watanabe, H. Yokoyama, Microelectron. Reliab., 42 (2002) 47. [5] G. Meneghesso, D. Buttari, E. Perin, C. Canali, E. Zanoni, IEDM, Dec. 6-9, 1998, p. 227.
10:00 AM - *G8.2
Recent Progress in Understanding the DC and RF Reliability of GaN High-electron Mobility Transistors
Jesus A. del Alamo 1 Jungwoo Joh 2
1MIT Cambridge USA2Texas Instruments Dallas USA
Show AbstractGaN HEMT technology is well on its way to revolutionize microwave and millimeter-wave communications and radar systems. In these applications, device reliability remains a significant concern. As the field has expanded, great progress has recently taken place in understanding GaN HEMT degradation, especially under high-voltage stress. Detailed electrical studies coupled with comprehensive failure analysis involving a variety of techniques have revealed a rich picture of degradation. Early studies showed that high voltage degradation of GaN HEMTs was characterized by a critical voltage (Vcrit) at which the device gate current abruptly increases. For stress voltage beyond Vcrit, prominent degradation was observed in the drain current and other electrical parameters of the device. More recently, it has been shown that degradation in the gate current can occur for voltages below the critical voltage suggesting that stress time is a key variable in degradation. Cross-section TEM and planar imaging techniques have shown that high-voltage stress induces prominent structural defects such as grooves, pits and cracks in the GaN cap and AlGaN barrier at the edge of the gate. The evolution of these defects correlates well with that of electrical degradation. Recently, a similar pattern of degradation has been observed under high-power RF stress where similar defects were identified, although not in a consistent way. Overall, the current state of understanding suggests that the mechanisms identified under high voltage DC stress are relevant for the RF degradation of power GaN HEMTs but that other mechanisms might play a dominant role in certain device designs. This talk will review recent research on GaN HEMT degradation under high voltage stress with relevance to RF power reliability.
10:30 AM - *G8.3
GaN HEMT Reliability: From Time Dependent Gate Degradation to On-state Failure Mechanisms
Enrico Zanoni 1 Gaudenzio Meneghesso 1 Matteo Meneghini 1 Antonio Stocco 1
1University of Padova Padova Italy
Show AbstractIn this paper, we compare degradation modes and failure mechanisms of different AlGaN/GaN HEMT technologies. We present data concerning reverse-bias degradation of GaN-based HEMTs, which results in a dramatic increase of gate leakage current, and present a time-dependent model for gate degradation. Some of the tested technologies demonstrated to be immune from this failure mechanism up to drain-gate voltages in excess of 100 V. When this was case, the main failure mode consisted of drain current degradation during on-state tests, resulting from charge trapping in the gate-drain access region attributed to hot-electron effects. Finally, the use of diagnostic techniques such as electroluminescence microscopy and Deep Level Transient Spectroscopy for the identification of failure modes and mechanisms of GaN-based HEMTs is reviewed. Concerning reverse-bias degradation of GaN-based HEMTs, we demonstrate that, (i) when submitted to reverse-gate stress, HEMTs can show both recoverable and permanent degradation. (ii) recoverable degradation consists in the decrease in gate current and threshold voltage, which are ascribed to the simultaneous trapping of negative charge in the AlGaN layer, and of positive charge close to the AlGaN/GaN interface. (iii) permanent degradation consists in the generation of parasitic leakage paths. Results indicate that permanent degradation can occur even for stress voltage levels significantly lower than the â?ocriticalâ? voltage identified by step-stress experiments. Time-dependent analysis suggests that permanent degradation can be ascribed to a defect generation and percolation process. Results supports the existence of a â?otime to breakdownâ? for HEMT degradation, which significantly depends on the stress voltage level. On the contrary, AlGaN/GaN technologies which were found to be resistant to gate degradation (off-state critical voltage larger than 100 V for a 0.25 um gate device) were submitted to on-state tests at different gate and drain voltage levels. In order to discriminate the degradation accelerating factors (temperature Tj, electric field, hot electrons or a combination), we carried out DC tests at constant dissipated power and at different VGS and VDS values (different electric field and hot electrons conditions). All tests showed a non-recoverable degradation of electrical parameters (drain saturation current, threshold voltage and on-state resistance) and electroluminescence signal EL, with a strong dependence on the EL value of the bias point, and a negligible dependence of temperature. Once verified that EL intensity represents a reliable estimate of channel hot electron effects, we attributed the degradation to hot electron trapping in the gate-drain access region. Using EL intensity as a measure of the stress acceleration factor, we derived an acceleration law for GaN HEMT hot electron degradation similar to the one already demonstrated for GaAs devices.
11:30 AM - *G8.4
Mechanical Stress Sensitivity of AlGaN/GaN HEMT Gate Leakage: Mechanisms and Models
Toshikazu Nishida 1
1University of Florida Gainesville USA
Show AbstractWhile the 2DEG arises from the piezoelectric polarization due to the lattice mismatch-induced mechanical strain in the AlGaN layer, the reliability physics of the AlGaN/GaN HEMT is also dependent on the mechanical stress within the device. The mechanism of defect formation at large reverse gate bias has been hypothesized to be related to mechanical stress through the inverse piezoelectric effect. Even prior to catastrophic failure, the gate leakage current is observed to degrade. Since at reverse gate bias, the inverse piezoelectric effect produces tensile stress which adds to the built-in tensile stress resulting from lattice mismatch, it is important to understand the effect of mechanical stress on the gate leakage current. To investigate the role of mechanical stress on gate leakage, external mechanical stress is applied while the gate leakage current is monitored as a function of reverse gate bias with drain and source terminals grounded. The mechanical sensitivity of the gate leakage current is observed to decrease as the reverse gate voltage is increased. To explain the field dependence of the mechanical sensitivity of the gate leakage, the electric field profile is analyzed and possible gate leakage models including Fowler-Nordheim tunneling, thermionic field emission, trap-assisted tunneling, and Poole-Frenkel emission are evaluated as a function of electric field and mechanical stress. Dominant mechanisms will be discussed for the stress dependent gate leakage current in AlGaN/GaN HEMT devices under different bias conditions.
12:00 PM - *G8.5
Simulation of AlGaN / GaN HEMT Reliability
Mark E. Law 1 2 David Horton 2 Nicole Rowsey 1
1University of Florida Gainesville USA2University of Florida Gainesville USA
Show AbstractReliability simulation is the combining of traditional topics of process modeling and device modeling and is now possible. We have combined the capabilities of process simulation of defect evolution, mechanical strain, chemical reactions, and inter diffusion with traditional device simulation concepts including contacts, electric field dependencies, and advanced transport. Combined using the the alagator scripting language, this allows simulation of operation of the device and the formation of defects that degrade device performance. A brief overview of the numerical approach will be provided along a description of current capabilities. Two degradation examples will be presented. First, degradation of insulator materials due to radiation exposure and second a gate metal inter diffusion into the channel. Combining simulation with good experimental work can help identify failure causes.
12:30 PM - *G8.6
Recent Reliability Progress of GaN HEMT Power Amplifiers
Toshihiro Ohki 1 Masahito Kanamura 1 Kozo Makiyama 1 Kenji Imanishi 1 Kazukiyo Joshin 1 Toshihide Kikkawa 1
1Fujitsu Laboratories Ltd. Atsugi Japan
Show AbstractAlGaN/GaN high electron mobility transistors (HEMTs) have achieved good performance for high-power and high-efficiency applications. Currently GaN HEMTs for transmitter amplifiers of wireless base stations have been commercialized since 2005. GaN HEMTs for higher frequency applications up to the X-band have been developed closely alongside the commercialization phase. Millimeter-wave amplifiers have also been developed for new markets. In this study, we describe highly reliable GaN HEMTs for high-power and high-efficiency amplifiers. First, we present the reliability mechanisms and progress of the previously reported GaN HEMTs. Next, we introduce our specific device structure of GaN HEMTs for improving reliability. An n-GaN cap and optimized buffer layer are used to realize high efficiency and high reliability by suppressing current collapse and quiescent current (Idsq)-drift phenomena. Finally, we discuss the reliability of millimeter-wave amplifiers. High breakdown voltage and small current collapse of GaN HEMTs with short gate length were achieved by improving the respective qualities of the epitaxial layer and passivation film.
Symposium Organizers
Osamu Ueda, Kanazawa Institute of Technology Graduate School of Engineering
Mitsuo Fukuda, Toyohashi University of Technology
Kenji Shiojima, University of Fukui Graduate School of Electrical and Electronics Engineering
Edwin Piner, Texas State University, San Marcos
Symposium Support
Japan Society of Applied Physics
G11: Solar Cells and TFTs
Session Chairs
Friday AM, April 13, 2012
Moscone West, Level 2, Room 2007
9:30 AM - G11.1
Defect Management in III-V Nanostructure Solar Cells for Higher Efficiency by Strain Control Based on In Situ Monitoring of MOVPE
Masakazu Sugiyama 1 2 Yunpeng Wang 3 Hassanet Sodabanlu 3 Shaojun Ma 2 Hiromasa Fujii 2 Kentaroh Watanabe 3 Yoshiaki Nakano 3
1Univ. of Tokyo Tokyo Japan2Univ. of Tokyo Tokyo Japan3Univ. of Tokyo Tokyo Japan
Show Abstract
Reduction of defects and dislocations is a key issue in high-efficiency photovoltaic devices using III-V epitaxial layers. Nano-epitaxial structures, such as quantum wells and quantum dots, are recently introduced into III-V solar cells, aiming at higher energy conversion efficiency by either improved current matching in a tandem-junction solar cell or intermediate-band operation. A nanostructure, on the other hand, is accompanied by hetero-interfaces which often induce defects, especially when stress is applied to the interfaces. A structure for a high efficiency cell requires large band offset in quantum structure, which almost inevitably results in substantial lattice mismatch at hetero-interfaces. Elaborate control of stress is therefore mandatory in the crystal growth of nanostructure solar cells. Moreover, a large number of stacked layers are necessary for substantial light absorption, in which strain accumulation of lattice would introduce dislocations. Such elaborate control of lattice strain necessitates in situ monitoring of growth in order to obtain local information of a structure under growth, while ex situ analysis such as XRD clarifies strain in a whole structure. We here focus on quantum wells as a simple example of strained nano-epitaxial structures and successfully analyzed the stress accumulation in the vicinity of growth surface through observation of wafer curvature during the MOVPE of InGaAs/GaAsP strain-compensated multiple quantum wells. We have obtained a direct relationship between strain accumulation in quantum wells and degradation in photovoltaic performances of a GaAs cell containing the quantum wells in its pn junction; it was confirmed that exact strain balancing results in the best performance of a quantum-well solar cell. In spite of the success in minimizing strain accumulation during the stacking of InGaAs/GaAsP quantum wells, it was suggested that the lattice mismatched hetero-interfaces still induces crystal defects. This issue has obstructed implementation of a superlattice with extremely-thin (3 nm) GaAsP barriers. Through in situ observation of surface reflectance during growth, it was confirmed that insertion of strain-graded buffers, which were thinner than 1 nm, dramatically suppressed performance degradation of the quantum well solar cell, probably through the reduction in defects between InGaAs and GaAsP due to delocalized stress. In summary, we have demonstrated combined strain management for InGaAs/GaAsP lattice-mismatched quantum wells, which resulted in higher energy conversion efficiency of solar cells containing the wells: global strain reduction by elaborate strain balancing on the basis of in situ wafer curvature monitoring, and local stress reduction by eliminating highly-lattice-mismatched interfaces using ultra-thin graded buffer layers.
9:45 AM - G11.2
Characterization of Temperature-Cycled GaAs Superlattice Solar Cell Structures Imbedded with InAs Quantum Dots by High Resolution X-Ray Diffraction
Josephine J Sheng 1 4 David C Chapman 2 David M Wilt 2 Christopher G Bailey 3 Christopher Kerestes 3 Seth M Hubbard 3 Sang M Han 4 1
1University of New Mexico Albuquerque USA2Air Force Research Laboratory Albuquerque USA3Rochester Institute of Technology Rochester USA4University of New Mexico Albuquerque USA
Show AbstractThe insertion of nanostructured materials (e.g., quantum wells, wires, and dots) into the intrinsic region of p-i-n solar cells introduces an intermediate band within the bandgap of the host material. In the case of InAs quantum dots (QDs) imbedded in GaAs superlattice (SL) solar cells, this sub-bandgap, along with GaP strain compensation layers, has been shown to enhance the short circuit current as well as the overall cell efficiency [1]. As a contender for space applications, it is necessary to subject these solar cell structures to temperatures they may encounter in the low earth orbit (LEO) and to probe any material degradation. Herein, we focus on temperature-dependent characterization of InAs-QD-enhanced GaAs solar cell structures with varying growth parameters, using high-resolution X-ray diffraction (HRXRD). The characterized structures have three main parameter variations: (1) InAs coverage for QD formation, (2) GaP strain compensation coverage, and (3) GaAs barrier coverage. HRXRD rocking curves of each structure focusing around the GaAs peak are analyzed at a range of temperatures up to 200 °C. Although no noticeable shifts in the SL peaks are detected, the interfacial diffusion decreases the resolution of fringes produced by reflections at the SL interfaces in test structures with varying InAs QD coverage and unbalanced strain. The unbalanced strain in the same structures leads to a distortion in the GaAs peaks. Since strain balanced structures do not show any peak or fringe degradation, we expect that the InAs-QD-enhanced GaAs solar cell structures that are strain balanced will not be affected by the temperature variations in LEO. [1] S.M. Hubbard, D. Wilt, S. Bailey, D. Byrnes, and R. Raffaelle, "OMVPE Grown InAs Quantum Dots for Application in Nanostructured Photovoltaics," Conf. Rec. of the 2006 IEEE 4th World Conf. of Photovoltaic Energy Conversion, pp. 118-121 (2006)
10:00 AM - G11.3
A Virtual Substrate Epitaxial Template with Tunable Lattice Parameter
Emily C. Warmann 1 Marina S Leite 1 Robyn L Woo 2 Daniel C Law 2 Shoghig Mesropian 2 Andrew T Palmer 2 Harry A Atwater 1 3
1Caltech Pasadena USA2Boeing Spectrolab Sylmar USA3Kavli Nanoscience Institute Pasadena USA
Show Abstract
We report a tunable lattice parameter epitaxial template for high quality lattice-matched III-V growth [1]. This design enables access to lattice-matched band gap ranges not available at or near the single-crystal wafer lattice constants of GaAs, Ge, InP or Si. This â?ovirtual substrateâ? consists of a large-area thin epitaxial film removed from its growth substrate and allowed to coherently relax all strain until achieving its composition-dictated lattice parameter. Supported on a reactor-compatible handle, it acts as template for epitaxial growth of any thickness of material directly at the desired lattice parameter without needing graded buffer layers to accommodate mismatch. We have fabricated virtual substrates 50 mm in diameter from 40 nm films of InxGa1-xAs and InxGa1-xP grown by MOCVD on InP and GaAs substrates respectively. The films are grown with a composition-dictated mismatch to the substrate ranging from -1.5% (compressive) to 1.2% (tensile) and are elastically strained as verified by high resolution x-ray diffraction (XRD) measurements showing relaxation 1%. After growth a viscoelastic wax applied to the film surface provides support during wet etching removal of the substrate. The viscoelastic deformation of the wax allows the film strain to relax coherently while maintaining a near-planar configuration. Bonded to a handle substrate and wax removed, the III-V film can serve as epitaxial template for future growth. XRD symmetric and asymmetric rocking curve measurements verify template lattice constants are dictated by the film composition, specifically 5.62, 5.70, 5.80 and 5.83 Ã.. These give full access to the wide range of energy band gaps in the GaInAlAsP material family in lattice-matched designs and remove the layer thickness restriction imposed by strain-induced dislocation formation. In particular, the 5.80 Ã. virtual substrate allows fabrication of a lattice-matched three junction InGaAs/InGaAsP/InAlAs multijunction solar cell (MJSC) with detailed balance efficiency of 56.2% improving over current MJSC designs both by using more optimal band gaps and by eliminating the high dislocation density of lattice-mismatched growth. X-ray photoelectron spectroscopy (XPS) measurements on virtual substrates reveal a native oxide layer and verify the efficacy of a wet chemical etch to remove it, leaving a surface of the correct stoichiometry for epitaxial growth. Virtual substrate growth performance was tested through fabrication of lattice-matched InxGa1-xAs single junction solar cells at 5.80 (x = 0.36) and 5.83 Ã. (x = 0.43). Photoluminescence and lifetime measurements of these devices as well as current-voltage behavior under light and dark conditions will be presented, along with characterization of defect type and density and comparison to similar devices grown lattice mismatched to InP.
1. M. S. Leite et al., Adv. Materials 23, 33, p. 3801 (Sep, 2011).
10:15 AM - G11.4
Enhanced Performance of III-V Compound Solar Cells via Edge and Surface Passivation with Trioctylphosphine Sulfide and Related Surfactants
Carissa Nicole Eisler 1 Matt T Sheldon 1 Brendan M Kayes 2 Harry A Atwater 1
1California Institute of Technology Pasadena USA2Alta Devices, Inc. Santa Clara USA
Show Abstract
Thin-film GaAs solar cells have achieved world record efficiencies as high as 28.2% for single junction cells [1]. High efficiency is achieved in part by effective passivation of surfaces to mitigate recombination of photogenerated carriers. Even higher efficiencies could be achieved if all recombination active surfaces, including cell edges, are passivated. However, this is challenging for III-V compound semiconductor solar cells. We report here that trioctylphosphine sulfide (TOP:S) and related chemical surfactants enable very effective surface passivation of thin film GaAs based solar cells, reducing the GaAs surface recombination velocity by more than 10x. We investigated surface passivation of GaAs by TOP:S with four independent experimental methods. Photoluminescence (PL) measurements were performed using a 633 nm laser source to determine the decrease in non-radiative carrier surface recombination. When applied to the (011) facet of a GaAs wafer, the TOP:S treatment boosted the relative PL signal by 50%, which is comparable to that of sodium sulfide, the best-known small molecule for sulfur passivation of GaAs. TOP:S was then applied to a series of different sized (1 mm2, 2 mm2, and 1 cm2) GaAs thin film solar cells fabricated by epitaxial liftoff. Solar cell light current-voltage response characteristics were measured under 1 Sun AM1.5G conditions. Without passivation, a decline in efficiency was observed with decreasing cell size because the smaller cells have a larger ratio of exposed lateral edges to active area and thus more recombination sites. After TOP:S treatment, the efficiency in the smallest cells greatly increased from 12.3% to 16.8%, approaching the efficiency of the largest sized cell (17.2%). Fitting the cell dark current response to a double diode model also allowed for the extraction of the sidewall contribution to surface recombination. The sidewall surface recombination current of small cells reduced by 80%, from 3.5 pA/cm to 0.7 pA/cm after treatment, indicating a decrease in surface recombination velocity (SRV). Finally, light beam induced current (LBIC) measurements with confocal microscope excitation from a 488 nm laser source quantified passivation near induced fractures. Analysis of the LBIC measurements indicated that treatment by TOP:S reduced the SRV of the exposed edge from 8500 cm/s to 510 cm/s. Through this comprehensive study, it was demonstrated that TOP:S greatly improved both the performance and durability of GaAs solar cells.We also report new results of similar treatments with InAlAs and InGaAs solar cells. X-ray photoelectron spectroscopy (XPS) and PL provide insight on the strength of the chemical interactions with the semiconductor surface and electronic quality after different passivations. Additional I-V and LBIC measurements quantify the changes in surface recombination velocity.
1. M. A. Green, et. al., Prog. Photovolt: Res. Appl. 2011, 19, 565.
10:30 AM - G11.5
In-situ Raman Probing of the Electron Accumulation Layer at the InN Surface
Esther Alarcon Llado 1 2 Jason B Yeo 1 3 Joel W Ager 1
1Lawrence Berkeley National Laboratory Berkeley USA2Ecole Polytechnique Federale de Lausane Lausanne Switzerland3University of California Berkeley USA
Show AbstractIndium nitride near-IR bandgap (~0.67eV) allows for a very large range in direct gaps of the group III-N alloys and thus offers an outstanding potential for solar energy conversion and optoelectronic applications. However, experimental demonstration of high efficiency In-rich III-V solar cells has been hampered by the existence of an intrinsic low mobility surface and/or interface electron accumulation layer. Such electron accumulation layer affects not only the electrical properties, but also has brought many controversies in the interpretation of optical experiments. Raman spectroscopy is a very powerful technique that enables to probe the lattice dynamics in a crystal, but also free charge in polar semiconductors. The electric field associated to the free charge interacts with that of the longitudinal optical modes, giving rise to coupled modes or LOPCMs. Raman scattering by LOPCMs is an alternative non-destructive probe into electron accumulation layer in InN. The InN surface can be unpinned by using a semiconductor/liquid interface. Applying a potential to the liquid electrolyte changes the electric field and distribution of free carriers near the surface In this work, we present an in-situ micro-Raman study that probes directly such changes in the surface charge and electric fields generated through an electrochemical set up. Our results confirm the presence of a surface related Raman mode in InN and show its interaction with accumulated electrons at the surface. Electrolyte gated Raman spectroscopy (EGRS) on InN layers was performed in order to modulate and in-situ probe the surface electron accumulation region in InN. We find that the intensity of the forbidden LO Raman feature is changed by the external potential as well as its frequency, showing that this feature is due, at least in part, to the SEA. This is the first experimental evidence of the relation between the LO feature in InN and free electron accumulation at its surface. Work at LBNL is supported by the US-DOE under Contract No. DE-AC02-05CH11231.EAL also acknowledges supports from Marie Curie Actions
10:45 AM - G11.6
Comparison of Junction Position and Grain Structure in CdTe Solar Cells Using a Combined FIB-EBIC Technique
Jonathan Douglas Major 1 Leon Bowen 2 Robert E Treharne 1 Ken Durose 1
1University of Liverpool Liverpool United Kingdom2University of Durham Durham United Kingdom
Show AbstractElectron beam induced current (EBIC) techniques are based on the measurement of minority carriers injected into a sample under the interrogation of an electron beam. One variant of EBIC is the cross â?"sectional measurement of photovoltaic devices, which allows the electrical p-n junction to be directly imaged. However, for CdTe based devices it is difficult to produce a clean, defect free, cross-section for EBIC analysis by conventional cleaving and polishing techniques. In this work we report the use of focussed ion beam (FIB) milling to produce device cross sections which were then subsequently analysed by high resolution EBIC to identify junction position. Devices with CdTe layers deposited by two different techniques, RF-sputtering and close space sublimation(CSS,) were compared. The microstructure of the device layers revealed by FIB milling is also discussed, with particular focus being placed on voiding within the CdTe and CdS layers as well as CdTe grain structure and twin boundary formation. Device efficiencies were determined by J-V measurement and found to be 6.8% for the sputtered device and 12.7% for CSS. FIB-EBIC analysis revealed that the sputtered device contained a buried homo-junction, with the EBIC signal being pinned towards the back contact away from the CdTe/CdS interface. By contrast the CSS device was shown to have a functional hetero-junction with the EBIC signal located at the CdTe/CdS interface. The gulf in device performance between the two deposition methods is attributed to this shift in junction position, and indicated that the post-growth treatment was un-optimised for the sputtered device. Paradoxically, evaluation of the two differing devices using external quantum efficiency (EQE) measurement revealed similar shaped curves for both, despite there being a significant efficiency difference. It would have been expected that the EQE would have revealed the presence of the buried junction through there being a higher response at longer wavelength. The similar response in EQE for cells that have manifestly different EBIC junction positions is discussed with reference to diffusion effects. Overall, this work therefore highlights the importance of the FIB-EBIC technique for both structural characterisation and optimisation of CdTe solar cells, and also demonstrates the relative limitations of EQE for identifying the junction position.
11:30 AM - G11.7
Correlation between the Photoluminescence of CdSe Nanocrystals and Their Performance in Hybrid Solar Cells
Michael Eck 1 2 Simon Einwaechter 1 2 Ying Yuan 1 2 Frank S Riehle 1 2 Michael Krueger 1 2
1Albert-Ludwigs-Universitauml;t Freiburg Germany2Albert-Ludwigs-Universitauml;t Freiburg Germany
Show AbstractHybrid solar cells are utilizing semiconducting polymers as main light absorber and electron donors, and inorganic semiconducting nanocrystals (NCs) as acceptor materials. Since NCs change their size and surface constitution during their synthesis the question arises when to stop the reaction and what NCs are the most suitable ones to be integrated in hybrid solar cells. Therefore a study of the photoluminescence (PL) behavior of NCs during the growth reaction was performed in situ, and correlated with their performance in hybrid solar cells. PL detection was found a useful tool for the quality detection of NCs; it is delivering information about the size and size distribution of the NCs and information about their relative surface quality as well. The PL signal was detected in situ during the microwave synthesis of CdSe NCs utilizing hexadecanol (HDO) as weakly coordinating ligand. A correlation of the PL signal of NCs and the performance of hybrid solar cells could be established. The optimum time point to stop a NC synthesis for using the NCs in hybrid solar cells can be detected by the PL signal. The best solar cell performances were achieved using CdSe NCs of the highest PL values, lowest size distribution while having the largest possible diameter.
11:45 AM - G11.8
Pseudomorphic Growth of Zinc Phosphide (Zn3P2) Thin Films by Molecular Beam Epitaxy
Jeffrey Paul Bosco 1 Harry A Atwater 1
1California Institute of Technology Pasadena USA
Show Abstract
In this work, we report on the pseudomorphic growth of tetragonal zinc phosphide (α-Zn3P2) obtained on atomic hydrogen treated GaAs(001) by compound-source molecular beam epitaxy (MBE). Zn3P2 is considered an ideal candidate for scalable photovoltaics, with a reported direct band gap of 1.5 eV and long minority-carrier diffusion lengths (>5µm). However, much remains to be studied regarding the electronic properties of defects within Zn3P2 films. Highly-crystalline MBE grown epilayers provide an ideal template for such studies. The morphology and electronic properties of the as grown films were studied in detail with emphasis on the effect of strain relaxation. The epitaxial relations between the Zn3P2 epilayer and GaAs substrate as determined by reflection high-energy electron diffraction (RHEED) and selected area electron diffraction (SAED) were as follows: Zn3P2 (004)||GaAs(002) and Zn3P2 (202)||GaAs(111). A 0.91% out-of-plain strain was observed by high-resolution X-ray diffraction symmetric reflections, whereas reciprocal space maps of asymmetric reflections demonstrated the Zn3P2 in-plain lattice spacing was coherently strained by the GaAs substrate. Partial relaxation of the Zn3P2 lattice was observed for film thicknesses greater than approximately 150 nm. A comparison of TEM images of strained and partially relaxed films shows the generation of threading dislocations as the mechanism for misfit compensation. Van der Pauw and Hall effect measurements demonstrated the films were intrinsically p-type due to incorporation of phosphorus interstitials. Strained films displayed mobilities in excess of 40 cm2 V-1 s-1, comparable to values reported for single crystal Zn3P2. However, hole carrier densities and mobilities were found to decrease significantly upon film relaxation due to the evolution of dislocations. In order to improve the morphological and electronic quality of thick Zn3P2 films, buffer layers of MxZnyPz and ZnxPyAz ternary alloys were also studied. By substituting in either group III or group VI elements for Zn or P, respectively, the lattice parameter of the alloy could be tuned to better match that of the GaAs substrate. This strategy should allow us to achieve thick, defect-free Zn3P2 films for incorporation into high efficiency photovoltaic devices.
12:00 PM - G11.9
Transparent Low Voltage Amorphous Metal Oxide-based Thin Film Transistors Processed at Room Temperature
Bhaskar Chandra Mohanty 1 Yeon Hwa Jo 1 Deuk Ho Yeon 1 Hong Rak Choi 1 Yong Soo Cho 1
1Yonsei University Seoul Republic of Korea
Show Abstract
Room temperature processed high transparent low voltage thin film transistors (TFTs) consisting of metal oxides have been demonstrated. The devices (having channel dimensions of 3000 µm Ã- 100 µm) consisted of amorphous Zn-In-Sn-O (a-ZITO) channel layer and SiOx gate dielectric with ZnO:Al thin films as electrodes. All films were deposited by pulsed laser deposition (PLD) at room temperature using an excimer laser (KrF; wavelength of 248 nm; pulse width of 20 ns; repetition rate of 5 Hz). The carrier concentration (conductivity) and visible transparency of the channel layer were optimized by tuning oxygen pressure in the deposition chamber and the best results were obtained for the films deposited at 50 mtorr. The current voltage measurements indicated an n-channel enhancement-mode transistor operation with appreciable drain current saturation achieved with a positive gate voltage of about 5 V. From the output and transfer curves, field effect mobility in the range of 2-10 cm2V-1s-1, drain current on-to-off ratio of order of 105 and threshold voltage of about 1.5 V were estimated at operating voltages of lower than 5 V. The results of the present study show a huge potential for realization of low power consuming and transparent devices that can be fabricated at room temperature.
12:15 PM - G11.10
Ruthenium and Iridium-based Thin Films as Oxygen Getters in Solution-based Low Temperature Thin Film Transistors
Ana Lizeth Salas-Villasenor 1 Jesus I Mejia 1 Bruce E Gnade 1 Manuel A Quevedo-Lopez 1
1University of Texas at Dallas Richardson USA
Show Abstract
The field of flexible electronics has expanded tremendously over the past few years. Similar to what happened in silicon integrated circuit technology 40 years ago; flexible electronics are now at a point where system design and process integration will drive the technology. However, in order to be competitive with state-of-the-art a:Si:H thin film transistors, any other technology must show reproducible transistor parameters such as mobility, threshold voltage, drive current and reliability. In this work, we studied how the device performance is affected depending on the material used for source-drain (SD) electrodes. Cadmium sulfide (CdS) thin-film transistors (TFTs) were fabricated using chemical bath deposition (CBD) method at 70° C1. 1-5 nm ruthenium oxide (RuO2) and iridium oxide (IrO2) were used as the work function setting interlayer between the CdS films and the SD electrodes.2,3 Both RuO2 and IrO2 are potential materials to be used in microelectronics 2-4. Due to their high work function and stability, these materials are attractive for n-metal-oxide semiconductor (MOS) transistors in complementary MOS (CMOS) technology for threshold voltage adjustment.4 Aluminum SD electrodes were used as reference. In this talk, we will report the improvement in the operational stability of CdS TFTs with either Ru or Ir-based thin films conductive oxides SD contacts. Improvement on the VT shift was related to the reduction of the CdO layer between the CdS semiconductor layers. In this case, Ru and Ir act as scavenging layer, removing the O2 from the CdS surface layer. The absorption of O2 molecules by the conductive oxide layer increases its conductivity and has a direct impact on the device performance. Extracted mobility for CdS TFTs was within the range of 10-20 cm2/V-s, depending on the sourceâ?"drain electrode. These mobilities are among the highest ever reported for a fully patterned TFT made with CdS as semiconductor. References [1] A.L. Salas-Villasenor, I. Mejia, J. Hovarth, H. N. Alshareef, D. K. Cha, R. Ramirez-Bon, B. E. Gnade, M. A. Quevedo-Lopez, Electrochem. Solid-State Lett., 13(9), H313, (2010). [2] E. V. Jelenkovic, K. Y. Tong, W. Y. Cheung, S. P. Wong, Microelectron. Reliab. 43, 49 (2003). [3] H. K. Kim, S. H. Cho, Y. W. Ok, T. Y. Seong, Y. S. Yoon, J. Vac. Sci. Technol. B 21, 949 (2003). [4] E.V. Jelenkovic, K.Y. Tong, J.Vac. Sci.Technol. B 22(5) (2004) 2319-2325
12:30 PM - G11.11
Characterization of rf-sputtered MgHfZnO Thin Films
Hantsun Chung 1 Jian Z Chen 1 I-Chun Cheng 2
1National Taiwan University Taipei Taiwan2National Taiwan University Taipei Taiwan
Show Abstract
ZnO based oxide electronics has drawn much attention recently due to its high electron mobility and large-area deposition capability. ZnO is an intrinsic n-type material. By alloying MgO into ZnO, the bandgap is enlarged and the intrinsic carrier density is reduced. Because Mg has stronger affinity for oxygen than Zn, MgZnO is expected to have less oxygen vacancies and more stable crystal structure than ZnO. This is beneficial for gate bias temperature stability of TFTs with MgZnO as the active layers [1,2]. To further improve the uniformity for large-area deposited films, amorphous phase is desirable. In this study, we demonstrate that Mg0.05Zn0.95O become amorphous or short-range-ordered with the addition of hafnium oxide. We also characterize the properties of rf-sputtered Mg0.05HfxZn0.95-xO thin films as a function of Hf concentration x. The films were rf-sputter deposited onto glass substrates (Eagle 2000, Corning Inc.) from Mg0.05HfxZn0.95-xO targets (x=0, 0.025, 0.05, 0.075, 0.1) in pure Ar ambient at room temperature. The sputtered Mg0.05Zn0.95O exhibits strong (002) preferred orientation with XRD peak located at 2θ=34.16°. With the increase of Hf doping concentration, the peak shifts slightly to lower 2θ angle due to the replacement of Mg or Zn atoms with Hf atoms in the crystals, because Hf4+ has larger radius than Zn2+ or Mg2+. The XRD peak intensity is also greatly reduced, indicating the material amorphorization proceeds with the addition of Hf. The grain size, estimated from the full-width-at-half-maximum (FWHM) of the (002) XRD peak, decreases from 24.1 to 3.3 nm as the Hf content x increases from 0 to 0.025 in Mg0.05HfxZn0.95-xO. No sharp XRD peaks are detected in the as-sputtered films when more than 5.0 at.% Hf are added into the materials. The films remain in amorphous or short-range-ordered states after annealing at 600 °C for 30 mins. All Mg0.05HfxZn0.95-xO films (100 nm in thickness) are highly transparent (> 80 %) in the visible region from 400 to 800 nm and have sharp absorption edges in the UV region. The absorption edge shifts to the shorter wavelength as the Hf content increases, inferring the increase of tauc bandgap. The bandgap Î"E (eV), as a function of hafnium composition x, is fitted as Î"E=3.336+6.08x for room temperature as-deposited films, and Î"E=3.302+2.60x for films after 600 °C annealing for 30 mins. The annealing process decreases the bandgap shift caused by the incorporation of Hf in the materials. {References: [1] Yi-Shiuan Tsai, Jian Z. Chen, "Positive Gate-Bias Temperature Stability of Rf-Sputtered Mg0.05Zn0.95O Active Layer Thin Film Transistors," IEEE Transactions on Electron Devices, accepted for publication. [2] Chih-Hung Li, Yi-Shiuan Tsai, Jian Z. Chen, "Negative Bias Temperature Instability of Rf-sputtered Mg0.05Zn0.95O Thin Film Transistors with MgO Gate Dielectrics," Semiconductor Science and Technology, vol. 26, p.105007 (2011).}
12:45 PM - G11.12
Improvement of Stability under Negative Bias Stress by Wet Annealing Process in Si-In-Zn-O Thin-film Transistors
Do Hyung Kim 1 2 Woochul Yang 2 Sang Yeol Lee 1
1Cheongju University Cheongju Republic of Korea2Dongguk University Seoul Republic of Korea
Show AbstractZnO-based thin film transistors (TFTs) have attracted considerable attention in active matrix liquid crystal displays and active matrix organic light emitting diode due to their high filed effect mobility (μFE) than that of the conventional TFTs using amorphous and poly-crystalline silicon materials. [1] Especially, many research groups reported that amorphous Ga-In-Zn-O (GIZO) has high μFE and good stability under various stress conditions. Recently, Lee et al [2] reported that amorphous Si-In-Zn-O (SIZO) TFT with high μFE and good stability compared with that of GIZO-TFTs was fabricated below 150oC, thereby facilitating the realization of high performance flexible electronics. In this study, we report on the improvement of negative gate bias stress (NGBS) induced stability by post wet annealing process in SIZO-TFTs. Also, the origin of stability under NGBS has been explored using x-ray photoelectron spectroscopy (XPS). Amorphous SIZO channel layers with 50 nm in thickness were deposited on 200 nm thick SiO2 as a gate insulator by using rf magnetron sputtering method at room temperature. Also, Ti and Au (10/60 nm), as source/drain electrodes, were prepared by using electron beam evaporation and thermal evaporation method, respectively. The SIZO-TFTs were annealed at 150oC for 1 h in dry O2 ambient or wet O2 ambient, which consists of O2, as a carrier gas, and water vapour from boiled deionized water. Transfer characteristics were obtained by using semiconductor parameter analyzer (HP 4145B) in dark and vacuum state of2Ã-10-2 Torr. As a result, threshold voltage shift of SIZO-TFT annealed by wet O2 process under NGBS was drastically reduced compared with that of SIZO-TFT annealed by dry O2 process. Also, based on our XPS analysis, it is suggested that the origin of instability under NGBS is strongly related with neutral oxygen vacancies located above valance band minimum. Therefore, the wet annealing process can be applied to improve the stability under NGBS in flexible display. [1] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano and H. Hosono, Nature (London), 432, 488 (2004). [2] D. H. Kim, H. K. Jung, D. H. Kim and S. Y. Lee, Appl. Phys. Lett. 99, 162101 (2011).