Symposium Organizers
Osamu Ueda Kanazawa Institute of Technology
Mitsuo Fukuda Toyohashi University of Technology
Steve Pearton University of Florida
Edwin Piner Nitronex Corporation
Paolo Montangero Avago Technologies Italy S.R.L.
B1: Laser Reliability
Session Chairs
Monday PM, November 30, 2009
Liberty (Sheraton)
9:30 AM - **B1.1
Catastrophic Optical-damage in High-power, Broad-area Laser-diodes.
Aland Chin 1 , Rick Bertaska 2 , Martin Jaspan 3 , Allen Flusburg 3 , Stephen Swartz 3 , Maciej Knapczyk 3 , Israel Smilanski 3 , Jonah Jacob 3
1 , ALAND CHIN, LLC, Sharon, Massachusetts, United States, 2 , New England Analytical, LLC, Nashua, New Hampshire, United States, 3 , Science Research Laboratory, Inc., Somerville, Massachusetts, United States
Show AbstractFor modern, high-power laser-diodes, the remaining failure-mode is reported to be catastrophic optical-damage (COD). A brief description of the COD phenomena is as follows. A local region of the laser diode, generally at the front facet, is heated by absorption of the laser light so that the material melts. Since heat is generated by absorption of the laser light, the molten region is substantially confined to the active layer of the laser cavity. The surface of the molten region is optically reflective. Lasing is sustained in the optical cavity defined by the surface of the molten region and the back facet. The molten region propagates towards the back facet as the material exposed to the laser light continues to melt whereas the material on the opposite side, no longer heated by laser light, solidifies. Propagation of the molten region continues until there is insufficient gain in the optical cavity to maintain a liquid state. While catastrophic optical-damage (COD) of single-mode lasers involves only transverse modes, we recently discovered that COD of broad-area, multi-mode laser-diodes involves both transverse modes and ring-cavity modes. In 1972, internally-circulating modes in sawn-cavity, broad-area laser-diodes were proposed as an explanation for an observed bi-modal distribution in the efficiency of lasers that were fabricated in an essentially-identical fashion. The internally-circulating modes are the ring-cavity modes we presented in 2009.The presence of the ring-cavity modes accounts for many of the unusual features of COD, some of which have been reported but not explained. These features include:●Failure location at the facet not coincidental with a peak in the near-field intensity.●±16° branching and fan out of the COD track as it propagates●Random nature of failure by COD●Polycrystalline structure of a portion of the COD trackA detailed description of the phenomenon of COD in short (380μm cavity-length), 12μm aperture, proton-bombarded, double-heterostructure laser-diodes with uncoated facets was first presented in 1974. In these devices, COD generally initiates at the facets due to high optical-power density and propagate along transverse-mode filaments. To achieve reliable operation at high optical-power, broad-area laser-diodes have evolved to long (several-millimeter cavity-length), wide-aperture (50-200μm), dielectric-defined, broadened-waveguide separate-confinement, double-heterostructure, quantum-well laser-diodes with coated, passivated facets. COD in these devices involve both transverse modes and ring-cavity modes. As a result of these improvements in device structure, ring-cavity modes contribute to the COD process.This report provides a description of COD formation and propagation, with and without ring cavity modes, in broad-area laser-diodes.Approved for Public Release, Distribution Unlimited
10:00 AM - **B1.2
Failure Analysis Using Optical Evaluation Technique (OBIC) of LDs for Fiber Optical Communication.
Tatsuya Takeshita 1 , Hiromi Oohashi 1
1 Photonics Device Laboratory, NTT Corporation, Atsugi, Kanagawa Pref., Japan
Show AbstractThe introduction of high-speed services for fiber-optic access subscribers has led to a huge growth in data traffic. The rapid diversification of services means that next generation networks must be built quickly, economically and reliably. A high temperature laser allows us to eliminate the thermo-electric cooler conventionally needed in a transmitter module, which results in reductions in cost, power consumption and size. Moreover, a high-power laser provides a wide tolerance when coupling optical fibers. In addition, a high-power pump laser is needed to realize a wide-band and high-power erbium-doped fiber amplifier. This makes high-performance laser chips one of the keys to achieving highly reliable and cost-effective systems. In terms of laser reliability, we must clarify the degradation mechanism and postpone or suppress degradation if we are to achieve a reliable high-performance laser. We have analyzed degraded lasers using the optical beam induced current (OBIC) technique. When there are nonradiative recombination centers in the degraded region, the OBIC intensity decreases with increases in recombination density. This technique has the advantages of being non-destructive and highly sensitive. In addition, it provides high space resolution in degradation analyses. The OBIC is measured through the window of a 5.6-mm transistor outline (TO) can before and after aging. Then, by using the same LDs we can detect an OBIC change for several aging times. We can both detect the degraded region and layer, and estimate the degree of laser degradation by employing the relative OBIC intensity prior to aging. This OBIC technique is useful for analyzing the degree of laser degradation.Moreover, the incident wavelength can be changed by changing the optical source in the OBIC measurement setup, which in turn changes the absorption layer and the penetration distance. Some degraded laser layers are reveled by using these several wavelengths absorbed in different layers. In addition, degradation in the waveguide interior is detected by using an incident wavelength with long penetration. Thus, by monitoring the OBIC intensity at several wavelengths as well as before and after aging, we are able to discuss sudden and wear-out laser failures. In our presentation, we will introduce examples using the OBIC technique that contributed to the improvement of laser reliability.
10:30 AM - **B1.3
Performance of InP/InAlGaAs Light Emitting Transistors Using Zn and C as Base Dopant.
Russell Dupuis 1 , Yong Huang 1 , Jae-Hyun Ryou 1 , Forest Dixon 2 , Nick Holonayk 2 , Milton Feng 2
1 School of ECE, Georgia Institute of Technology, Atlanta, Georgia, United States, 2 Department of ECE, University of Illinois at Urbana-Champaign, Champaign-Urbana, Illinois, United States
Show AbstractLight emitting transistors (LETs) operating at around 1.55 µm were investigated using InP/InAlGaAs material system grown by metalorganic chemical vapor deposition (MOCVD). Device epitaxial structures were achieved by incorporating InGaAs quantum wells (QWs) in the base region of the N-InP/p-InAlGaAs/N-InAlAs heterojunction bipolar transistors (HBTs), and both current gain and long-wavelength light emission were demonstrated. It was found that control of the p-type doping profile in the base layer is one of the key factors that dictate the performance characteristics of long wavelength LETs.Epitaxial growth of the devices was carried out on (001) InP:Fe substrates between 600 °C and 650 °C at a reactor chamber pressure of 100 Torr. Both Zn and C were used as the p-type dopant in this study. A typical NpN LET structure consists of (from the bottom to the top) a Si-doped InP subcollector, an undoped In0.52Al0.48As collector, an undoped In0.53(AlxGa1-x)0.47As grading layer with xAl from 1 to 0.25, a Zn- or C-doped In0.53(Al0.25Ga0.75)0.47As base/active region with an undoped compressively-strained In0.58Ga0.42As QW embedded in the middle of the base layer, an InP:Si emitter and an emitter contact. The devices were fabricated using standard optical lithography and wet chemical etching to form an emitter and base mesa. AuGe/Ni/Au was used as the emitter and collector ohmic contacts and Au/Zn/Au as the base contact.Secondary ion mass spectrometry (SIMS) shows that in the LET with Zn-doped base Zn diffuses into both emitter and the adjacent graded collector, while in the C-doped LET, C stays in place and forms abrupt junctions. The turn-on voltages are 0.89 V and 0.78 V for the Zn-doped LET and the C-doped LET, respectively, indicating the presence of a potential spike at emitter-base junction due to Zn diffusion. In addition, Zn diffusion into the undoped QW region degrades the optical quality by creating non-radiative recombination centers, which is confirmed by the electroluminescence (EL) peak power for both devices. The C-doped SQW LET exhibits much higher light output owing to the intact active layer. C-doped LETs have a higher emitter injection efficiency and luminescence efficiency, but it is not granted without a problem. Whereas Zn will inevitably diffuse into active region and plague the QW, Zn can be contained only in the base region through proper engineering of growth conditions. It was found that Zn-doped LET has a much higher current gain. The diffusion-suppressed Zn-doped LET has a DC current gain of 45 in sharp contrast to the low current gain of about 0.25 in the C-doped LET. Short minority carrier lifetime in C-doped materials is considered to account for the low gain, which is possibly due to the low growth temperature and low V/III ratio used during C-doping.
11:30 AM - **B1.4
InGaN Laser Diode Degradation. Surface and Bulk Processes.
Piotr Perlin 1 2 , Lucja Marona 1 , Tadek Suski 1 , Przemek Wisniewski 1 2 , Mike Leszczynski 1 2 , Pawel Prystawko 1 , Michal Bockowski 1 2 , Robert Czernecki 1 2 , Irina Makarowa 2 , Bogdan Kowalski 3
1 , Institute of high Pressure Physics, Warsaw Poland, 2 , TopGaN Ltd, Warsaw Poland, 3 , Institute of Physics, Warsaw Poland
Show AbstractThe nature of degradation processes of nitride light emitting devices, particularly laser diodes is one of the most intriguing issues of the short-wavelength optoelectronics. In spite of the intensive research, no comprehensive mechanism of GaN based laser diodes degradation was so far presented. The present work focuses on the reliability study performed on InGaN laser diodes grown by MOVPE method on low and middle dislocation density GaN substrates (104-107cm-2). Our observation, in the agreement with the research performed by the others group, shows that that we can assign the degradation processes existing in the nitrides laser diodes into two categories: the surface reactions and bulk effects. What concerns the first category, the careful observation of the laser mirrors of degraded laser diodes, reveals the presence of carbon deposits. The formation of these deposits depends strongly on the atmosphere in which the aging process is being carried out. For instance the degradation is very fast if the diode operates in the atmosphere of dry nitrogen, and it is much slower if oxygen is added to the gas. All these mentioned above features resembles very closely so called PIF (Package Induced Failure) - mechanism of degradation characteristic for 980 nm, high-power laser diodes. This mechanism involves photochemical reactions of hydrocarbons decomposition on the surface of laser diode mirror. The second mode of degradation strongly depends on the density of dislocations. It primarily manifests by the increase of the threshold current of a device with relatively small variation of the slope efficiency. The degradation follows quite closely the square root of time-dependence, though it tends to deviated from this behavior at the late stage of the aging process. Interestingly no signs of degradation are visible in the cathodoluminescence images which makes tempting to think that the degradation effects do not consist in the development of the nonradiative recombination centers but rather enhance the scarier leakage. We will also briefly discuss the potential role of dislocation loop for the degradation of InGaN laser diodes.
12:00 PM - **B1.5
Structural Defects in GaN-based Materials and Their Relation to GaN-based Laser Diodes.
Shigetaka Tomiya 1 , Yuya Kanitani 1
1 Advanced Materials Laboratories, Sony Coporation, Atsugi, Kanagawa, Japan
Show AbstractReduction of structural defects in III-nitride based optical devices is critical important for high efficient and high reliable performance. In this presentation, three different types of structural defects observed in GaN-based laser diodes (LDs) are discussed. In the early stage of development of GaN-based laser diodes, a high density of threading dislocations (typically 108-1010/cm2) was inevitable, since samples were grown on sapphire substrates. In the conventional GaAs- and InP-based LDs, existence of a single dislocation in the stripe regions is harmful since dislocations can multiple easily by carrier recombination. On the other hand, an influence of dislocations for the GaN-based LDs is not predominant as much as that for the conventional LDs. First, we report on dislocation motion observed in GaN-based materials by electron beam irradiation and discuss relation between dislocation motion and device degradation. Next, we discuss on pyramidal defects observed in Mg doped p-type layers in GaN-based LDs. Since the Mg acceptor level is fairly deep, a high level of Mg doping is required to obtain a high concentration of free holes. However, as Mg concentration is increased, the pyramidal defects were nucleated. Here, we report on the detailed structural analysis of the pyramidal defects and discuss the formation mechanism of the defects. GaInN alloy layer is one of the most important elements for GaN-based optical devices. Technological problem of the growth of GaInN alloys with high indium concentrations is caused namely due to low miscibility of InN and GaN. During the course of the development of the high power pure blue laser diodes, we found the multiple defects nucleated at the GaInN MQWs. This type of the defects is very critical for the device lifetime. In this presentation, we report on the detailed structural analysis of this new type defects and will discuss the formation mechanism and the reduction methodology of these defects.
12:30 PM - B1.6
A Study of Degradation in High Power Multi-Mode InGaAs-AlGaAs Strained Quantum Well Lasers as Pump Lasers.
Yongkun Sin 1 , Nathan Presser 1 , Neil Ives 1 , Steven Moss 1
1 Electronics and Photonics Laboratory, The Aerospace Corporation, El Segundo, California, United States
Show AbstractHigh power broad-area InGaAs-AlGaAs strained quantum well (QW) lasers with emission wavelengths at 910-980nm are used as pump lasers to optically pump various fiber lasers and amplifiers. These applications mainly developed for industrial uses do not require stringent reliability from the pump lasers. Also, the fact that these pump lasers have not been deployed in high reliability communications systems including potential satellite systems necessitates careful study of reliability and degradation of these devices. Maximum output powers of both single-mode and broad-area lasers are limited by catastrophic optical mirror damage (COMD). However, unlike 980nm single mode lasers where a single failure mode is typically observed due to good facet passivation along with optimized structural designs, broad-area lasers have shown at least two different failure modes including COMD and bulk failure. There have been extensive reports on COMD, but very limited reports on bulk failure although catastrophic bulk failure has been identified as the dominant failure mode in broad-area InGaAs-AlGaAs strained QW lasers.We investigated reliability and degradation processes in commercial MOCVD-grown broad-area InGaAs-AlGaAs strained QW lasers at ~975nm by performing accelerated lifetests of these devices followed by failure mode analyses (FMA) with micro-analytical techniques including electron beam induced current (EBIC), time resolved electroluminescence (EL), and deep level transient spectroscopy (DLTS). Both passivated and unpassivated broad-area lasers were studied that yielded catastrophic failures at the front facet and also in the bulk. The lifetests performed typically under automatic current control mode generated failures at different stages of degradation. EBIC technique was employed to study dark line defects generated in degraded lasers stressed under different test conditions and was also used to estimate minority carrier diffusion lengths from laser diodes at different stages of degradation. Time resolved EL technique was employed to study initiation and progressions of dark spots and dark lines in real time as devices are aged. Lastly, DLTS technique was employed to study deep electron traps in degraded laser diodes. We will report our in-depth FMA results.
12:45 PM - B1.7
Reliability of Semiconductor Structures with Buried Quantum Dots.
Vladimir Chaldyshev 1 , Nikolay Bert 1 , Anna Kolesnikova 2 , Vladimir Nenedomsky 1 , Valerii Preobrazhenskii 3 , Mikhail Putyato 3 , Alexei Romanov 1 , Boris Semyagin 3
1 , Ioffe Institute, St.Petersburg Russian Federation, 2 , Institute of Problems in Mechanical Engineering, St.Petersburg Russian Federation, 3 , Institute of Semiconductor Physics, Novosibirsk Russian Federation
Show AbstractCrystal lattice mismatch in semiconductor structures leads to the generation and relaxation of elastic strains and corresponding mechanical stresses that in many cases determines the reliability of electronic and optoelectronic devices. For quantum dot (QD) structures it is usually assumed that the relaxation of mechanical stresses proceeds during their self-organization on the surface. However, burying of the QDs results in an increase in the stored elastic energy, which can affect the onset of relaxation in such semiconductor structures and impact reliability of the corresponding devices. We report on theoretical and experimental studies of general critical conditions for the onset of relaxation in stressed QDs embedded into a crystalline matrix. The fingerprint of the relaxation scenario is formation of specific satellite dislocation loops. The scenario is applicable to both the QDs nucleated and ripened in the bulk and the QDs formed on the surface and then buried by overgrowth. The results demonstrate that the critical relaxation radius for buried QDs is larger than critical thickness for stressed lattice mismatched films or surface islands.Transmission electron microscopy confirms the relaxation of stresses at InAs QDs buried in GaAs via formation of specific dislocation defects. It has been found that the dislocations formed in vicinity of selected InAs quantum dots do not exit to the free surface of the GaAs film. That means that the stress relaxation occurs in the bulk of the GaAs film near buried QDs, which were coherent to the matrix at the stage of their formation on the GaAs surface. The developed models can be used for the prediction of reliability of such devices as lasers and photodetectors based on the system of buried QDs.
B2: Degradation Mechanisms
Session Chairs
Russell Dupuis
T Takeshita
Monday PM, November 30, 2009
Liberty (Sheraton)
2:30 PM - **B2.1
Recombination-enhanced Dislocation Glides--The Current Status of Knowledge.
Koji Maeda 1
1 Applied Physics, The University of Tokyo, Tokyo Japan
Show AbstractAmong lattice imperfections greatly affecting the reliability of semiconductor devices, dislocations are the most common and important defects because they are usually harmful for the device performance. Especially in optoelectronic devices such as laser diodes, it is well known that the device degradation occurs as a consequence of multiplication of dislocations during device operation under minority carrier injection. More recently, device degradation in the similar mode was found to occur also in SiC p-i-n diodes for power-control applications [1,2]. Commonly, these phenomena are induced by the glide (and/or climb) motion of dislocations that is enhanced by carrier recombination at the dislocations. So far, it has been shown that the effect of recombination-enhanced dislocation glides (REDG) is observed in most semiconductors (elemental, III-V and II-VI compounds)[3]. The factors that determine whether the REDG effect actually brings about degradation or not are the presence of driving force, the presence of dislocation sources, and the magnitude of enhanced dislocation velocity. In most cases, the REDG effect is driven by a mechanical shear stress such as built in heterostructures by the lattice misfit. Therefore the lattice matching is a measure usually employed to suppress the degradation in the REDG mode. In SiC, however, the REDG of partial dislocations, which results in expansion of stacking faults, is exceptionally driven by a non-mechanical stress of unestablished origin [1]. Since nucleation of dislocations needs surmounting a large energy barrier or aggregation of point defects, the degradation in the REDG mode is difficult to occur in the absence of pre-existing dislocations. Therefore, the removal of as-grown dislocations is the most orthodox approach to the degradation-free devices. In GaN and related compounds, however, in spite of the high density of as-grown dislocations, the degradation in the REDG mode is not rapid, presumably due first to the absence of mechanical shear stress driving the dislocations and secondly to the exceptionally low rate of carrier recombination at dislocations. The enhanced dislocation glide velocity is characterized by the prefactor linearly dependent on the recombination rate and the activation energy reduced in magnitude which tends to increase with the bandgap energy of the crystal [3]. Although these features are explicable by the phonon kick mechanism, it is not yet exactly known what elementary process is enhanced. Measurements of excitation spectra in optical excitation may give us insight to this fundamental problem [2,4]. [1] S. Ha and M. Skowronski, Phys. Rev. Lett., 92 (2004) 175504. [2] A. Galeckas, J. Linnros and P. Prouz, Phys. Rev. Lett., 96 (2006) 025502. [3] K. Maeda and S. Takeuchi, Dislocation in Solids, 10 (1996) p.435. [4] Y. Ohno, T. Taishi and I. Yonenaga, phys. stat. sol., (2009) in press.
3:00 PM - **B2.2
Mechanism of Defect Reactions in Semiconductors.
Yuzo Shinozuka 1
1 Faculty of Systems Engineering, Wakayama University, Wakayama Japan
Show AbstractProposed mechanisms so far on defect reactions in semiconductors (defect creation, annihilation, multiplication, reconstruction, impurity diffusion, …) are reexamined with special attention to the instability of the atomic configuration for particular electronic states and the transient lattice vibration induced by successive carrier captures.1) Thermal activation process to overcome the potential barrier Un: The reaction rate is given by p0 exp(-Un/kBT), where p0 is the attempt frequency and Un depends on the electronic state n.2) Instability mechanism: An electronic transition at a defect promptly induces the lattice distortion along the reaction coordinate QR. A symmetry breaking distortion (Td to C3v) is inherent for the diamond and zincblende structures, and is favorable for hole localization. 3) Phonon kick mechanism: An electronic transition to or from the state n induces the lattice vibration along the lattice relaxation mode Q1. The vibration amplitude decreases because of the dephasing in relevant normal modes in Q1. If Q1 is not orthogonal to the defect reaction coordinate QR, the defect reaction rate is enhanced during the lattice relaxation time τ~ 2π/Δω, where Δω is the width of the frequency distribution of the normal modes. 4) Phonon kick mechanism (recombination): During the transient lattice vibration induce by an e (h) carrier capture the next h (e) capture process is enhanced. Thus the transient lattice vibration induced by successive e and h captures in turn enhances the next capture process. If N electrons and N holes are successively captured by a defect within a short period τ~2π/Δω, an electronic energy ~ N times Eg, the band gap energy, is transformed into the lattice vibration energy. The coherent vibration part Q1 is enhanced in step with each carrier capture. After a numerical simulation the probability of this positive feedback is found to critically depend on the carrier concentrations. The defect reaction rate is given by p0 exp(-Eiact/kBT) because only the first capture (i=e, h) is to be activated.With an animation we also show the meaning of the configuration coordinate diagram scheme in many electron representation, only which can properly relate the lattice distortion with the level position of a defect in the band gap in the multiphonon recombination process.
3:30 PM - B2.3
Bulk Catastrophic Optical-damage of 980nm, InxGa1-xAs/GaAs, High-power, Broad-area Laser-diodes Due to a Void in Au80Sn20 Solder.
Aland Chin 1 , Rick Bertaska 2 , Henry Eppich 3 , Martin Jaspan 3 , Jonah Jacob 3
1 , ALAND CHIN, LLC, Sharon, Massachusetts, United States, 2 , New England Analytical, LLC, Nashua, New Hampshire, United States, 3 , Science Research Laboratory, Inc., Somerville, Massachusetts, United States
Show Abstract Catastrophic optical-damage (COD) generally occurs at the front facet of laser diodes with unpassivated facets due to optical absorption that increase in time due to chemical reactions that are accelerated by temperature and light. When the facets are passivated, CODs often initiate at the front facet due to the combined intensity of a transverse mode and a ring-cavity mode. Additionally, in laser diodes with passivated facets, where COD occurrence at the facet is a random event, COD can sometimes initiate at a location within the optical cavity. In this study, we show a correlation between the location of a large solder-void and the location where COD initiates within the interior of the optical cavity of 980nm, InxGa1-xAs/GaAs, high-power, broad-area laser-diodes. The laser diodes are mounted epitaxial-side down onto Cu10W90 heat-sinks using Au80Sn20 eutectic solder. Using finite-element analysis (FEA), we propose a mechanism for COD formation within initially defect-free material due to a large void in the solder.Approved for Public Release, Distribution Unlimited
3:45 PM - B2.4
A Thermomechanical Approach to the Formation of Dark Defects in High Power Laser Diodes.
Alonso Martin 1 , Pilar Iniguez 2 , Juan Jimenez 1 , Myriam Oudart 3 , Julien Nagle 4
1 GdS Optron lab, Universidad de Valladolid, Valladolid Spain, 2 Física Teórica, Atómica y Óptica, Universidad de Valladolid, Valladolid Spain, 3 , Alcatel-Thales 3-5lab, Palaiseau France, 4 , Thales Research and Technology (TRT), Palaiseau France
Show AbstractGreat efforts are made to improve the optical power and the reliability of high power lasers, aiming to a growing number of applications (solid-state laser pumping, materials processing, optical communications, printing machines…). The research on the degradation mechanisms of these devices is a necessary way to improving their optical power and reliability. The increase in the optical power of the devices induces an important heating of the active parts, quantum well and waveguide layers, especially at the mirror facet where energy losses take place resulting in the degradation of the output power and being responsible of the end catastrophic optical damage (COD). COD may begin by nonradiative recombination at facet defects, which can transfer heat to the neighbor areas increasing the local temperature of the facet and producing band gap shrinkage with the subsequent enhancement of the laser light self-absorption. Besides, nonradiative recombination can transfer energy to the surrounding lattice allowing defect formation and motion by the mechanism of recombination enhanced defect reactions. The two processes feedback the local temperature increase during laser operation. Finally, a thermal runaway process leads to the catastrophic degradation of the device. A model providing a comprehensive description of the laser degradation is needed. Cathodoluminescence images of degraded devices unveil the presence of dark defects at the facet together with extended defects inside the cavity; these dark defects are formed by plastic deformation of the active layers of the laser structure. In this work, we put forward a microscopic scenario that accounts for extended defect creation and motion resulting in the rapid degradation of the laser. We will show that the idea of a high local temperature induced by REDR and feedback by laser light self absorption constitutes the most probable source of local stress leading to the formation of dislocations during laser operation. We model by a finite element analysis the distribution of thermal stresses induced by the local heating associated with defects at the facet mirror. A heat source located at the facet, in the active zone, is considered to simulate the defect acting as a local heat source fed by self-absorption of the laser light and nonradiative recombination processes. This heat source produces a temperature distribution in the device which is calculated solving the heat transfer equation of a solid. Local temperatures at the facet can be high, with prominent temperature gradients. These temperature gradients promote thermal stresses in the structure, predominantly in the vicinity of the active zone, which are calculated taking into account the principle of virtual work. According with the maximum shear stress theory or the maximum shear strain energy per unit volume, this procedure allows us to set up the conditions for the triggering of the plastic deformation in lasers under operation.
B3: Optical Devices and Reliability
Session Chairs
Ed Piner
Shigetaka Tomiya
Monday PM, November 30, 2009
Liberty (Sheraton)
4:30 PM - B3.1
Gain Saturation of 785-nm Laser Signal Amplified in Si-rich SiOx Strip-loaded Waveguides on Quartz and Si.
Chung-Lun Wu 1 , Cheng-Wei Lian 1 , Gong-Ru Lin 1
1 Graduate Institute of Photonics and Optoelectronics, National Taiwan University , Taipei Taiwan
Show AbstractSi light-emission devices have been successively demonstrated by using low-dimensional Si structure to overcome the indirect bandgap property of bulk Si with extremely low internal quantum efficiency. Low-dimension Si such as Si nanocrystal (Si-nc) can essentially improve its internal quantum efficiency of luminescence due to the quantum confinement of the self-trapped excitions. Since 2000, several groups have employed various-length pumping method to consecutively demonstrate the net modal gain of amplified spontaneous emission (ASE) from Si-ncs buried in the Si-rich SiOx planar waveguide up to 100 cm-1. Nonetheless, the characterization on the small-signal power gain of such waveguide structure under laser diode injection is mandatory towards the application of waveguide amplifiers in the future. In this work, we compare the laser-diode injection based small-signal power amplification and gain saturation properties of the SiO2/SiOx:Si-nc/SiO2 strip-loaded waveguide made on quartz and Si substrates.The Si-rich SiOx film with a thickness of 0.35 μm is deposited on quartz and Si substrates after growing 1µm-thick standard SiO2 buffer layer. After annealing the SiOx film in a quartz furnace to precipitate Si-nc, the SiO2 film of 1.5 μm is grown upon the SiOx film to form the waveguide. The sample with a 50μm-wide and 3cm-long photoresistive mask pattern is wet-etched by buffered oxide etchant (BOE) to form the SiO2/ Si-nc:SiOx/SiO2 strip-loaded waveguide. With a He-Cd laser based top pumping geometry and the variable strip length (VSL) method, the optical net modal gain and propagation loss of Si-nc is determined from the ASE ranged between 750-850 nm with 3dB spectral linewidth of 140 nm. The optical net modal gain of Si-nc at peak wavelength of 805 nm are 65 cm-1 and 85.7 cm-1, and the propagation loss is about 5 cm-1 and 21 cm-1 on quartz and Si substrates, respectively. Even with a higher net modal gain coefficient, the optical loss coefficient of the waveguide made on Si substrate is larger than that on quartz due to the more pronounced leakage of mode power to Si substrate. By injecting the directly modulated 785-nm laser diode signal with power of 1 nW into the 1.5cm-long waveguide, we further demonstrate the small-signal power amplification. Under the pumping power of 40 mW at 325 nm, the small-signal power gain is about 7 dB and 13 dB for waveguides made on quartz and Si substrates, respectively. We also observe the gain saturation effect and fit it with theoretical formula to find the peak gain of 27 and 36 dB, and the saturated power of 0.85 and 0.75 nW for the waveguide amplifiers made on quartz and Si substrates. Raising the saturation power relies strictly on increasing either the volume density of the buried Si-ncs or the SiOx film thickness.
4:45 PM - B3.2
Surface and Interface Effect of Durable and Efficient Polymer Solar Cells.
Dong Hwan Wang 1 , Jong Hyeok Park 2 , Sang Hyuk Im 3 , O Ok Park 1
1 Chemical & Biomolecular Engineering, KAIST, Daejeon, Daejeon, Korea (the Republic of), 2 Chemical Engineering, Sungkyunkwan University, Suwon, Suwon, Korea (the Republic of), 3 , KRICT, Daejeon, Daejeon, Korea (the Republic of)
Show Abstract Even though the fossil energy sources are much limited, the energy consumption is increasing rapidly every year. Therefore, new approaches for sustainable energy of solar energy systems are definitely urgent issues. Many of the different solar cells are under consideration by a lot of research groups all over the world such as Silicon, compound semiconductor and organic solar cells. Especially, polymer (plastic) based solar cell is a quite attractive one for its flexibility, low-cost possibility, light-weight, and semi-transparency. Here we focused on how to improve not only the efficiency and low cost processability (1) but also high-temperature durability will be considered. It can be demonstrated by introducing a protective layer of linear polymeric TiOx interlayer. It is attributed to an improved interfacial stability owing to relatively reduced morphology change at high temperature operation. The effectiveness of this unique feature makes it possible to fabricate more efficient organic solar cells by adopting a post annealing process. (2)-(3) Also we have tried to make bulk heterojunction photovoltaic cells with enhanced nanoscale morphology by adding an ionomer (partially sulfonated polystyrene (PSP)) into a regioregular P3HT/(6,6)-phenyl C61-butyric acid methyl ester (PCBM) blend.(4) And a patterned conducting IZO film with well-ordered periodic dot-structures with 50 nm or 200 nm deep features has been constructed by nanoimprinting technique as the anode of an organic solar cell. It was found that a highly ordered 2D-dot nano-patterned anode can enhance the device performance due to the large interfacial area between both of the electrodes and the active layer. This is possible because the nano-patterned structures can efficiently harvest electrons and holes from the active layer to each electrode in the IZO anode and Al cathode.(5) Finally, our group recently reported that bilayers active film with a concentration gradient has been successfully fabricated via solution process for the first time. The concentration variation has been confirmed by the Auger spectroscopy. The novel device showed an enhanced photocurrent density and power conversion efficiency compared to those of the BHJ PV prepared under the same fabrication condition. (6)1. SJ Yoon, JH Park, HK Lee and OO Park, Appl.Phys.Lett., 92 (2008) 1435042. DH Wang, SH Im, HK Lee, JH Park and OO Park, Langmuir, In revision (2009)3. TW Lee and OO Park, Appl.Phys.Lett., 77 (2000) 33344. KC Kim, JH Park and OO Park, Sol. Energy Mater. Sol. Cells, 92 (2008) 11885. DH Wang, DG Choi, KJ Lee, JH Jeong, JH Park and OO Park, Nanotechnology, In revision (2009)6. DH Wang, DG Choi, HK Lee, JH Park and OO Park, Appl.Phys.Lett., accepted (2009. 6)
5:00 PM - B3.3
A MHz Modulable Si-based LED Afforded by Engineering Light-emitting Defects in Si.
Norishige Tana-ami 1 , Jun Igarashi 1 , Yosuke Terada 1 , Yuhsuke Yasutake 1 , Susumu Fukatsu 1
1 , University of Tokyo, Tokyo Japan
Show AbstractDefects are often thought of as being “undesirable” whenever the reliability of devices is an issue. Intriguingly, however, there is a certain class of crystal defects that are of device grade in terms of their potentially useful optical and/or electrical properties. In this work, we demonstrate that a particular class of radiative defects created in Si helps build light-emitting-diodes (LEDs) with a MHz bandwidth. The significance of such a “defect-active dynamic LED” is two-fold. First, “defect engineering" opens up an avenue for boosting the light-emitting potential of Si, especially near-band-edge recombination, which has been a challenge over years in the field of photonics as well as materials science. Second, that the radiative defects in Si studied here are made to be responsive on the sub-µs time scale is remarkable, in view of slow dynamics of relaxation, >10µs, as in many of the defect-related systems. {311}-rod like defects were launched into Si in the course of programmed thermal relaxation of strain that had built up in the 1.5-µm device Si layer, lying on top of a Si-on-insulator substrate. Ambipolar LED geometry was defined by depositing Al Schottky pads. E-line electroluminescence (EL) due to {311}-defects survived above 150 K with decay times >>1µs increasing with temperature. A 2-MHz bandwidth of 100-% modulation amplitude was only achievable by biasing the LED at 4-8 V. The E-line intensity was at maximum at 20 K, at which temperature the modulation bandwidth also went through a maximum. This indicates that the radiative recombination of the E-line is of bimolecular nature with some degree of charge imbalance created initially being compensated for by either an electric bias or thermal carrier redistribution, viz. phonon scattering. As such, a biased charge distribution is expected to allow a good command of the physical properties that are otherwise controlled by defects. Total deactivation of the {311} rod-like defects and hence quenching of the E-line may result, which should be compared with total elimination of defects via sample treatment inclusive of chemomechanical processing and those which are dependent upon thermal budget.
5:15 PM - B3.4
Reliability and Performance of Pseudomorphic Ultraviolet Light Emitting Diodes on Bulk Aluminum Nitride Substrates.
James Grandusky 1 , Yongjie Cui 1 , Mark Mendrick 1 , Shawn Gibb 1 , Leo Schowalter 1
1 , Crystal IS, Green Island, New York, United States
Show AbstractReliability and performance of ultraviolet light emitting diodes have suffered due to the high dislocation density of the AlN and high Al-content AlxGa1-xN layers when grown on foreign substrates such as sapphire. The development of pseudomorphic layers on low dislocation density AlN substrates is leading to improvements in reliability and performance of devices operating in the UVC range. One major improvement is the ability to operate devices at much higher current densities and input powers than devices on sapphire substrates. This is due to the better thermal properties and lower dislocation density of devices on AlN substrates. Devices with active area of 0.001 cm2 emitting at ~265 nm have been measured for their reliability and change in power output over time at input currents of 20 mA (20 A/cm2) and 150 mA (150 A/cm2). When operating at currents of 20 mA over 1500 hours of consecutive operation has been demonstrated with typical decay of ~20% over the 1500 hours. Extrapolating the decay with a linear fit gives a L50 (time to 50% of initial power) of >5000 hrs. However it is desirable to be able to model the decay to better understand the kinetics and better understand the mechanisms. In order to do this, the lifetime at both 20 mA and 150 mA were modeled using an exponential decay function with 2 channels. Both sets of date were able to be fit well with the exponential decay function with a fast initial decay constant and a slow decay constant. The fast decay, <50 hours is attributed to a burn-in period. The slow decay represents the degradation period and can be used to estimate the L50, or time until the power is 50% of the initial power. Time constants of between 10,000 and 15,000 hours and 900-1000 hours were measured for devices operated at 20 mA and 150 mA respectively, indicating an L50 of between 7,000 and 10,000 hours at 20 mA and between 620 and 700 hours at 150 mA, showing great promise for viable device applications.
5:30 PM - B3.5
Modifications of Defects Concentration Induced by Ammonia Flow Rate and its Effects on Gallium Nitride Grown by MOCVD.
Suresh Sundaram 1 , Ganesh Vattikondala 1 , PremKumar Thirugnanam 1 , Balaji Manavaimaran 1 , Ganesan VedachalaIyer 2 , Baskar Krishnan 1
1 Crystal Growth Centre, Anna University, Chennai, Tamil Nadu, India, 2 , UGC-DAE Consortium for Scientific Reserach, Indore , Madhya Pradesh, India
Show AbstractGaN and related III-nitrides are very promising materials for the fabrication of optoelectronic devices and high power and high temperature electronic devices [1-3]. GaN, however, because of the lack of native substrates, heteroepitaxial GaN film fabricated usually exhibits a high dislocation density (108–109 cm-2) [4]. The device uniformity and reproducibility, especially the reverse leakage current of GaN-based devices, are strongly influenced by the dislocations in the GaN film [5]. Degradation of device performance has been associated with trapping centers in the GaN [6]. Evidence of deep level centers has also been observed in GaN UV detector structures [7]. Previously the effect of screw dislocations on schottky diode characteristics has been reported [8]. In this investigation, we report on the optical and schottky diode characteristics of unintentionally doped GaN films grown by MOCVD with different V/III ratio by varying the source ammonia (NH3) flowrate exhibiting reduction in the density of threading dislocations (TDs) with an edge component and reduced carbon and oxygen impurity incorporation. The density of dislocations determined from hot-wet chemical etching and atomic force microscopy show that on decreasing the ammonia flowrate, threading dislocations decreases. Low energy positron beam was employed to study the Ga vacancies in the epilayers. S- parameter vs. positron beam energy curves clearly shows increase in SL on increasing the V/III ratio indicating that the point defects trapping positron increases. Corroborative HRXRD, Raman studies and Photoluminescence confirm the reduction in trapping defects and threading edge dislocations with reducing V/III molar ratio. The effects of such variation of compensating centres and radiative centres as a function of MOCVD growth conditions on optical properties and schottky device characteristics like radiative decay lifetime, barrier height and reverse leakage current respectively were outlined.REFERENCES:[1] S. Nagahama, T. Yanamoto, M. Sano, T. Mukai, Appl.Phys. Lett. 79, 1948 (2001).[2] Y.F. Wu, D. Kapolnet, J.P. Ibbetson, P. Parikh, B.P. Keller, U.K. Mishra, IEEE Electron Dev. Lett. 48, 586(2001).[3] D.H. Youn, V. Kumar, J.H. Lee, R. Schwindt, W.J. Chang, J.Y. Hong, C.M. Jeon, S.B.Bae, K. S. Lee, J.L. Lee, J.H. Lee, I. Adesida, Electron. Lett. 39, 566 (2003).[4] S. Tomiya, E. Morita, M. Ukita, H. Okuyama, S. Itoh, K. Nakano, and A. Ishibashi, Appl.Phys. Lett. 66, 1208 (1995).[5] E.J. Miller, D.M. Schaadt, E.T. Yu, C. Poblenz, C. Elsass, and J.S. Speck, J. Appl. Phys. 91,9821 (2002). [6] P.B. Klein, S.C. Binari, J.A. Freitas Jr., A.E. Wickenden, J. Appl. Phys. 88 (2000) 2843.[7] Z.C. Huang, J.C. Chen, D.K. Wickenden, J. Crystal Growth 170 (1997) 362.[8] P.Y. Huang, X.D. Chen, S. Fung, C.D. Beling, and C.C. Ling, J. Appl. Phys. 94 (2003) 5771.
5:45 PM - B3.6
Large Batch Etching of Gallium Nitride Using Inductively Coupled Plasma Tools as a Production Solution.
Zhong Ren 1 , Mark Dineen 1 , Ligang Deng 1 , Andrew Goodyear 1 , Robert Gunn 1
1 , Oxford Instruments Plasma Technology, Bristol United Kingdom
Show AbstractGallium Nitride (GaN) material has become one of the most important segments of the high bright light emitting diode (HB-LED) and laser diode (LD) semiconductor industry in recent years. With its extensive applications in large screen display technology and illuminating engineering, that is driving the fabrication roadmap with its dramatic acceleration in throughput and production quality control. Plasma dry etching is an essential process of GaN device fabrication because of the high bonding energy in GaN lattice and lack of reliable wet etching method. This paper reports a high-effective etching large batch GaN technique, by means of Inductively Coupled Plasma (ICP) tools as industrial production solutions. A variety of batch processes has been demonstrated and etching rates in excess of 150nm/min with a good selectivity to photoresist (PR) were achieved.Particularly, a special magnetic field loop-source was designed and employed in the ICP system to control plasma profile during the processes. This technique was favourable for increasing etching rates and achieving good uniformities. As a result, uniformities of etch across these batches were down to ±3%.
Symposium Organizers
Osamu Ueda Kanazawa Institute of Technology
Mitsuo Fukuda Toyohashi University of Technology
Steve Pearton University of Florida
Edwin Piner Nitronex Corporation
Paolo Montangero Avago Technologies Italy S.R.L.
B4: Electronic Device Reliability
Session Chairs
Koji Maeda
Yuzo Shinozuka
Tuesday AM, December 01, 2009
Liberty (Sheraton)
9:30 AM - **B4.1
Reliability of InP HBTs Under High Current Density Operation.
Yoshino Fukai 1
1 , NTT Photonics labs., Atugi Japan
Show AbstractInP-based heterojunction bipolar transistors (HBTs) have been scaled down and then demonstrated ultra-high-speed operation. They can lower the power consumption of ICs and be integrated into optoelectronic devices. Thus their introduction into optical communications systems operating at over-100 Gbit/s is advantageous [1].The main reliability issues of InP HBTs during low current density operation (Jc < 1 mA/μm2) are the shift of the turn-on voltage (Von) due to intermixing of the highly doped PN-junction and a change in the dc current gain (β) due to surface recombination or other defect-related mechanisms. [2] In our InP HBTs with C-doped bases Von were proved to be stable because of a lower diffusion coefficient of C. Moreover, we suppressed β reduction with ledge structures and SiN passivation [3].Under high current density operation with a Jc of over 2 mA/μm2, which is particularly important for applications in high-speed ICs, degradation of the device intrinsic layer becomes important. We observed cross sectional views of the device intrinsic layer of HBTs, which degraded with bias temperature (BT) stress, by transmission electron microscopy (TEM), and we analyzed the atomic composition using electron diffraction spectroscopy (EDS). These analyses suggested that degradation of the emitter metal (Ti/Pt/Au) caused damage to the semiconductor layer and a local shift in the atomic composition during BT tests, which is a trigger for degradation in electrical characteristics of HBTs. Notably, there was an electro-migration of Au, and in the InGaAs contact layer an Au-based lattice-like structure was reconstructed where the indium (In) content increased. This degradation must be attributed to carrier flow, which increased point defects generation and drifted them. We introduced the refractory metal Mo or W as a diffusion barrier for the emitter metal under Ti/Pt/Au. After the same BT tests, very little crystalline degradation was observed by TEM. Emitter resistance was evaluated to compare all types of metal configurations. The activation energies for a 3 % increase in resistance were 2.0 and 1.65 eV for BT stress with for Jc of 2 and 5 mA/μm2, respectively, which were the same for all types of metal configurations.We have applied the same ledge structure and emitter metallization with W for high-speed sub-micrometer HBTs, which had fT over 300 GHz and confirmed the high reliability with a Jc of 5 mA/μm2 [4].In conclusion, we have investigated the intrinsic crystalline damage triggered by emitter metal degradation for HBTs with high current density. Suppression of them made our sub-micrometer HBTs highly reliable for use in practical systems.[1] K. Sano, et. al., in proceedings of 2004 SSDM.[2] K. T. Feng, et. al. , in IEEE GaAs Digest, pp.89-92,2002.[3] Y. K. Fukai, et. al., Electronics and Communication in Japan, part 2, 90(4), pp.1-8, 2007.[4] N. Kashio, et. al., IEICE Trans. Electron. Vol. E91-C (7), pp.1084-1090, 2008.
10:00 AM - B4.2
Predictive Model for the Relative Dislocation Sensitivity of GaAs and GaN.
Jeong Ho You 2 , Harley Johnson 1
2 , Caltech, Pasadena, California, United States, 1 , University of Illinois, Urbana, Illinois, United States
Show AbstractThreading dislocations are known to degrade photoemission much more strongly in GaAs than in GaN. Here, we present results of a simple model combining first principles and effective medium methods that predicts and explains the mechanisms behind these differing dislocation sensitivies, in good agreement with experimental data. A layer of each material is modeled as a two dimensional domain containing a dipole of edge dislocations with opposite Burgers vector directions; its size is varied according to desired dislocation density to study dislocation density effects. Energy levels and wave functions for the conduction and valence bands in each material are obtained by solving the appropriate multiband k*p Hamiltonians using the finite element method, and the spontaneous emission spectrum is then evaluated at different dislocation densities. The electrostatic potential due to negatively charged dislocations and deformation potential induced by dislocation strain fields reduces the band edge peak intensity and broadens the band edge peak. A comparison of calculated band edge peak intensities indicates that as dislocation density increases, the band edge peak decreases significantly for dislocation densities higher than 10^2 cm-2 in GaAs, while the band edge peak is relatively unaffected up to dislocation densities of 10^6 cm-2 for GaN. The smaller effective mass and larger lattice constant in GaAs are found to be important factors in making GaAs more sensitive than GaN to dislocations. Based on these two physical parameters, predictions of photoemission sensitivity to dislocations for other III-V materials are provided.
10:15 AM - B4.3
Gate Stack Reliability of High-Mobility 4H SiC Lateral MOSFETs with Deposited Al2O3 Gate Dielectric.
Daniel Lichtenwalner 1 , Veena Misra 2 , Sarit Dhar 3 , Sei-Hyung Ryu 3 , Anant Agarwal 3
1 Materials Science and Engineering, North Carolina State University, Raleigh, North Carolina, United States, 2 Electrical and Computer Engineering, North Carolina State University, Raleigh, North Carolina, United States, 3 Power R & D, CREE, Inc., Research Triangle Park, North Carolina, United States
Show AbstractLateral nMOSFETs have been fabricated on 4H SiC utilizing deposited dielectrics and gate-last processing. A bi-layer dielectric was utilized consisting of thin nitrided SiO2 covered by 25nm of Al2O3 deposited using atomic layer deposition. Field-effect mobility and threshold voltage (VT) were found to vary with SiC nitric oxide (NO) anneal temperature. High peak mobility values were obtained, with a corresponding positive VT of 0.8 V, using an 1175 °C 20 min NO anneal of the SiC before Al2O3 deposition. The peak mobility decreases somewhat with an 1100 °C NO anneal, while the VT increases to 1.6 V. These results indicate a mobility advantage of moving away from dielectric formation by thermal oxidation of SiC, although issues related to reliability are more uncertain with low-temperature-deposited dielectrics. For example, it is very important to maintain positive VT for normally-off devices, thus VT stability with field and/or temperature is very important.In this study we report on the gate stack reliability of these Al2O3 on nitrided SiO2 dielectric stacks. We find that gate leakage is significantly reduced as the SiC NO anneal thermal treatment is increased. In addition, although the MOSFET shows a lower VT with increased NO anneal, constant voltage stressing (CVS) of the gate indicates a much more stable device VT in this case. For example, after a limited number of Vg sweeps (~4 MV/cm fields) a VT increase of 0.1 V is observed with an 1175 °C NO anneal of SiC, whereas a VT shift of as much as 0.5 V occurs for devices receiving only an 1100 °C NO anneal. In addition, heating unstressed devices to 200 °C reveals a stable VT with temperature for the 1175 °C NO anneal case, while for the 1100 °C NO anneal case the VT decreases by 0.5 V. Additional results related to gate stack properties, and VT control, will also be presented.
10:30 AM - **B4.4
GaN HEMT Reliability.
Kurt Smith 1
1 , Raytheon, Andover, Massachusetts, United States
Show AbstractWhile GaN-based HEMT’s have demonstrated high power densities and excellent high frequency performance, there has been a reluctance to insert these devices into high performance systems due to a perceived lack of sufficient reliability data for operation in the higher stress regime associated with GaN. The high power operation of these devices leads to high electric fields, large current densities and elevated operating temperatures, all of which can potentially increase the rate of degradation within the devices. In addition, the strong polarization effects couple strain with electrical performance, making strain evolution a important reliability consideration. Through a combination of long-term testing, accelerated stress and understanding degradation mechanisms, lifetime projections have shown a MTTF of >107 hrs.
B5 : Wide Bandgap Devices I
Session Chairs
Tuesday PM, December 01, 2009
Liberty (Sheraton)
11:30 AM - **B5.1
Lifetime Assessment of Militarily Critical Electronics: An Evolving Role for Government Science and Technology
Christopher Bozada 1 , Donald Dorsey 2
1 , Air Force Research Laboratory Sensors Directorate, Wright-Patterson AFB, Ohio, United States, 2 , Air Force Research Laboratory Materials and Manufacturing Directorate, Wright-Patterson Air Force Base, Ohio, United States
Show AbstractThe Department of Defense has broad and exacting requirements for electronics lifetimes under a variety of operations and environments. Premature failure of electronics in a fielded system is costly in terms of mission loss, root cause analysis efforts and replacement. Until a technology has been used successfully in a relevant environment for years, the community must rely on accelerated testing to estimate useful life. The most common method, the three-temperature (3T) accelerated life tests using an Arrhenius model for a single dominant failure mechanism, has produced highly overestimated lifetime predictions for military applications. The problem is exacerbated by DoD’s reduced role in developing, maturing and consuming commercially driven electronics. DoD laboratories have studied this problem extensively over the past few and is pursuing several strategies to enhance its ability to assess and validate potentially valuable technologies for insertion into military systems. This talk will focus on the Science and Technology community’s fundamental physics and chemistry approach to provide higher fidelity lifetime estimates using high-resolution structural analysis and physics and chemical models of device operation and degradation.
12:00 PM - **B5.2
The Analysis of Wide Band Gap Semiconductors Using Raman Spectroscopy.
Samuel Graham 1
1 Mechanical Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States
Show Abstract The development of GaN semiconductor materials has received much interest due to their large band gap, excellent carrier mobilities, and large breakdown fields. This has led to the creation of both optoelectronic and RF-microelectronic devices. RF devices are based largely in part on heterojunction field effect transistors, whose reliability depends in part on applied electric fields and device self heating. However, the measurement and prediction of hot spot temperatures in GaN devices remains a technical challenge which is necessary for a full understanding of device reliability. Thus, the development of metrology and modeling methods for GaN semiconductors is critical to the advancement of these devices. In this study, we demonstrate the usefulness of the linewidth (FWHM) of the Raman Stokes response for the characterization of temperature in wide band gap semiconductor devices. The utility of this parameter is established through measurement of device temperature in GaN electronic devices. In addition, through the analysis of the Stokes linewidth, measurements of optical phonon life times were obtained which yield insight into the formation of hot spots in active microelectronic devices. Measurement results were compared to 3-D finite-element analysis. Advanced modeling using a new multiscale Lattice Boltzmann-Finite Differences approach is also presented. Results show that improved accuracy in Raman measurements can be made using the linewidth approach as well improved modeling of the hot spot temperature can be determined using the proposed multiscale thermal model. In addition to thermal fields, it is also important to address the stresses in GaN devices. The capability of gallium nitride (GaN) high power transistors arises, in large part, due to large piezoelectric polarizations that induce the formation of a carrier rich two-dimensional electron gas. These polarizations, in turn, are directly related to the strain, and hence stress, that is present within the transistor. As a consequence, the determination of stress is important to the optimization of GaN devices. In response, this study demonstrates the quantification of operational thermoelastic stresses in GaN transistors through simultaneous use of the Raman signal’s Stokes peak position and linewidth. The method is utilized in the analysis of HFET devices grown on Si and SiC substrates. For each series of device, the major stress components - thermoelastic, piezoelectric, and residual - are acquired and compared. While the magnitudes of the components are larger in those devices grown on silicon, the resultant biaxial loads in each of the devices are comparable at high power levels as the dominant residual tensile stress is only counterbalanced by the compressive thermoelastic contribution
12:30 PM - B5.3
A Comprehensive Approach to HEMT Reliability Testing.
David Cheney 1 , Brent Gila 2 , Erica Douglas 2 , Fan Ren 3 , Steve Pearton 2 1
1 Electrical & Computing Engineering, University of Florida, Gainesville, Florida, United States, 2 Material Science & Engineering, University of Florida, Gainesville, Florida, United States, 3 Chemical Engineering, University of Florida, Gainesville, Florida, United States
Show AbstractDevice reliability testing can manifest their weaknesses and failure mechanisms. As new materials and structures are developed, creative and different testing is often necessary to determine their reliability. We are developing an in-house reliability test system that will provide flexibility in testing HEMT devices. Using commercial off-the-shelf power supplies and data acquisition equipment, we are designing the control system that will allow for many different tests: DC and RF, temperature, step/stress/recovery, high-speed pulse, and optical pumping, as well as combinations and automated sequencing of these tests. We also have the flexibility to create new test types as failure mechanisms are understood. The initial station will provide for the simultaneous, independent testing of 16 devices. This paper outlines the system design and capabilities, the motivation for the system, and results from the different tests.
12:45 PM - B5.4
GaAs and GaN HEMT Reliability and Degredation Mechanisms after Long Term Stress Testing.
Erica Douglas 1 , Dave Cheney 2 , Brent Gila 1 , Stephen Pearton 1 , Fan Ren 3 , Cammy Abernathy 1
1 Materials Science & Engineering, University of Florida, Gainesville, Florida, United States, 2 Electrical and Computer Engineering, University of Florida, Gainesville, Florida, United States, 3 Chemical Engineering, University of Florida, Gainesville, Florida, United States
Show AbstractToday’s microwave and optoelectronic applications employ the use of compound semiconductor devices, principally GaAs and GaN transistors. Future ultra-high power radar systems will require transistors to be operated at very high voltages, current, and temperatures. To date, reliability studies of III-V devices have exhibited numerous degradation mechanisms, with the bulk of these studies relying primarily on temperature acceleration. In order to effectively identify failure mechanisms (including gate sinking in GaAs HEMTs, surface passivation issues and contact degradation in GaN HEMTs) long term DC stress tests have been carried out on HEMTs. Numerous devices were independently tested and analyzed, under various stress conditions, including pulse tests to both the gate and drain. By appropriately identifying the various mechanisms of degradation, these acceleration tests have enhanced the prediction of failure, reduce random failures, and develop crucial prescreening tools.
B6: Compound Semiconductors
Session Chairs
Tuesday PM, December 01, 2009
Liberty (Sheraton)
2:30 PM - B6.1
Medium Energy Ion Scattering Investigation of SiO2/4H-SiC Interface with Nitridation Anneal.
Xingguang Zhu 1 2 , Sarit Dhar 4 , Hang Dong Lee 1 , Tian Feng 1 , John Rozen 4 , Ayayi Ahyi 5 , John Williams 5 , Eric Garfunkel 3 , Torgny Gustafsson 1 , Leonard Feldman 2 4
1 Department of Physics and Astronomy, Rugters University, Piscataway, New Jersey, United States, 2 Institute for Advanced Materials and Devices, Rugters University, Piscataway, New Jersey, United States, 4 Department of Physics and Astronomy, Vanderbilt University, Nashville, Tennessee, United States, 5 Department of Physics, Auburn University, Auburn, Alabama, United States, 3 Department of Chemistry, Rutgers University, Piscataway, New Jersey, United States
Show Abstract4H-SiC is a promising candidate for power electronic applications. However, the poor interface quality of SiO2/SiC remains a major factor limiting the performance of SiC MOS devices. Applying a nitridation anneal [1] (NO gas, atm. pressure, 1175C, 2 hrs) on oxidized 4H-SiC results in a significant decrease in the number of defects and an improvement of the electrical properties. Over the past years, different surface analyzing techniques have been applied to study the effect of nitrogen on the SiO2/SiC interface. Despite this effort, the total amount and detailed structure of nitrogen at the interface remains uncertain due to the resolution limit of the most commonly used methods [2][3]. An accurate measure of this quantity is critical to the understanding of the nitrogen effect and to distinguish different classes of atomic level models that have been suggested to explain the beneficial effect of the NO process. High resolution, medium energy ion scattering spectroscopy (MEIS) allows for very accurate depth profiling and calculation of elemental concentration at near surface region. In this work, MEIS is applied to samples subjected to different nitridation methods and conditions. The detailed results for nitrogen at the interface will be discussed. Results from other analyzing techniques on similar samples will also be introduced as a comparison. Finally the measured quantities are discussed in terms of existing interface models.[1] Chung et al., Appl. Phys. Lett. Vol. 76, No.13 (2000)[2] Chang et al., J. Appl. Phys. 97, 104920 (2005)[3] J.Rozen, PhD Dissertation, Vanderbilt University (2008)
2:45 PM - B6.2
Comparative Study on Reliability of InP/InGaAs Heterojunction Bipolar Transistors with Highly Zn- and C-doped Base Layers.
Atsushi Koizumi 1 , Kazuki Oshitanai 1 , Kazuo Uchida 1 , Shinji Nozaki 1
1 Department of Electronic Engineering, The University of Electro-Communications, Chofu, Tokyo, Japan
Show AbstractInP-based heterojunction bipolar transistors (HBTs) have demonstrated an excellent high-frequency performance. Carbon has been most commonly used as a p-type dopant for a base layer because of its low diffusivity compared with those of other p-type dopants such as zinc and beryllium. However, the amphoteric characteristic of carbon as a dopant forces us to lower the growth temperature and V/III ratio in order to corporate a high concentration of carbon as acceptors in the growth of InGaAs layer by metalorganic vapor phase epitaxy (MOVPE). Since these growth parameters are not usually preferred because of poor crystallinity of the grown InGaAs, the InP/InGaAs HBT with a carbon-doped base layer may not necessarily demonstrate better transistor and reliability performance than that with a zinc-doped base layer. Hereafter, the HBTs with carbon- and zinc-doped base layers are referred to as the C-doped and Zn-doped HBTs, respectively. Therefore, it is important to compare the characteristics and reliabilities of C- and Zn-doped InP/InGaAs HBTs and understand the device physics of these transistors for development of high-performance InP/InGaAs HBTs.The epi layers of the Zn- and C-doped HBTs were grown by low-pressure MOVPE using CBr4 as a C source. Without post annealing of the C-doped HBT structure at 480°C for 20 min, we were not able to obtain a transistor characteristic, while the Zn-doped HBT structure does not require such post annealing. The Raman spectroscopy reveals crystallinity as good as that of the Zn-doped InGaAs only after the post annealing. The poor crystallinity of the as-grown C-doped InGaAs is attributed to lower growth temperature. However, the photoluminescence intensity remains lower than the Zn-doped InGaAs even after the post annealing. This suggests that the nonradiative centers could not be annealed out in the C-doped InGaAs. This was supported by lower current gain of the C-doped HBT.The DC common emitter current-voltage (I-V) characteristics of the HBTs with emitter geometry of 120 μm × 120 μm were measured before and after the current stress. The current gains of the C- and Zn-doped HBTs are 58 and 75, respectively. The stress tests were given to both types of HBTs with a constant collector current density of 100 kA/cm2 at collector-to-emitter bias of 1.5 V at room temperature. For a 15-min stress, the current gains of the Zn- and C-doped HBTs decreased by 40 and 3 %, respectively, from the initial current gains. We believe that the decreased current gain is due to current-driven diffusion of Zn. In conclusion, the current gain of the C-doped HBT is lower than that of the Zn-doped HBT, but the C-doped HBT is much more reliable under high current operation because of low diffusivity of carbon even at a high concentration level.
3:00 PM - **B6.3
Passivation Reliability for III-Nitride Semiconductor Devices.
Brent Gila 1 , A. Gerger 1 , A. Scheuermann 1 , D. Cheney 3 , E. Douglas 1 , C. Abernathy 1 , F. Ren 2 , S. Pearton 1
1 Materials Science and Engineering, University of Florida, Gainesville, Florida, United States, 3 Electrical and Computer Engineering, University of Florida, Gainesville, Florida, United States, 2 Chemical Engineering, University of Florida, Gainesville, Florida, United States
Show AbstractGallium Nitride (GaN) field effect transistors have attracted considerable interest as high power electronics for use in the electric utility industry, defense and space applications, and hybrid vehicles. GaN based HEMT devices are now highly developed and are commercially available. However, device reliability is still a concern and identification of failure modes and the tools to predict failure are currently being investigated. One major concern with the GaN HEMT device is the reduction of drain current under high source-drain voltage applications. This phenomenon, known as current collapse, is believed to be due to traps at both the exposed surface and in the underlying GaN buffer. This problem is greatly reduced by improvements in III-Nitride growth recipes and by the addition of a dielectric on the top surface of the fabricated HEMT, which acts as a passivation layer to reduce the surface electrical traps.This talk will discuss the surface preparation of MOCVD grown GaN and AlGaN, both in-situ and ex-situ and the subsequent gas-source MBE growth of crystalline oxides and PECVD deposition of SiNX for surface passivation. Surface preparation techniques have been explored using RHEED, AES, SIMS and C-V measurements to produce films of low interface trap density, 1-2E11 eV-1cm-2. A similar study of the as-fabricated HEMT surface was carried out to create a cleaning procedure prior to dielectric passivation. Post-deposition materials characterization included AES and XPS, as well as Hall effect, gate pulse and isolation current measurements for the passivated HEMT devices. For a reliability study, HEMT stuctures were passivated with different dielectrics and were studied for a period of 5 weeks after passivation while being held at 200°C. From comparisons to unpassivated and SiNX passivated, the crystalline oxides were more stable and provided for more current collapse mitigation.
3:30 PM - **B6.4
Reliability of GaN on Si FETs and MMICs.
Donald Gajewski 1 , Walter Nagy 1 , Allen Hanson 1 , Wayne Johnson 1 , Kevin Linthicum 1
1 , Nitronex Corporation, Durham, North Carolina, United States
Show AbstractThis paper reviews the reliability results for the gallium nitride on silicon (GaN-on-Si) technologies for commercial and military communications markets. Two technology platforms have been qualified for volume production: one consisting of discrete heterostructure field effect transistors (HFETs) and the other consisting of HFETs integrated with passive components to form monolithic microwave integrated circuits (MMICs). The technology platform qualifications for volume production have been achieved through intrinsic reliability tests on the active and passive device elements as well as extrinsic reliability tests at the product level. This paper presents reliability results on accelerated life test (ALT), high temperature operating life under DC and RF stress (DC/RF-HTOL), electrostatic discharge (ESD), ramped voltage breakdown, electromigration, temperature cycling, robustness under voltage standing wave ratio (VSWR) mismatch conditions, and diode stability. Degradation and breakdown mechanisms are discussed in relation to material properties reliability. The results show that the HFET and MMIC technology platforms display reliable performance for 20 year product lifetime at worst case operating conditions.
B7: Characterization I
Session Chairs
Chris Bozada
Samuel Graham
Tuesday PM, December 01, 2009
Liberty (Sheraton)
4:30 PM - B7.1
Investigating of the PCMO Resistance Switching Devices by High Resolution Thermal Imaging Method.
Lau Hon Kit 1 , Chan Kwok-leung 2 , Leung Chi Wah 1
1 Applied Physics , The Hong Kong Polytechnic University, Hong Kong China, 2 Mechanical Engineering , The Hong Kong Polytechnic University, Hong Kong China
Show AbstractResistive switching mechanism of metal electrodes in the planar Al/Pr0.7Ca0.3/MnO3(PCMO)/Ti resistive switching device is investigated by thermal imaging method. From the high resolution thermal images, it can be observed that when the bias voltage increase from -3V to -4V from the Al to Ti, the localized temperature variation of the Al and Ti electrode increases by 5.3imes and 2.4 times respectively. By comparing the two dimensional thermal images of the device under different bias condition, it can be verify that localized heat source locate in between the Al/PCMO interface while the device is switching from high resistance state to low resistance state. These extra heat sources agree well with the previous finding that the existence of AlOx in the Al/PCMO interface.[J.Phys.D:Appl.Phys.,42,045411] Such non-destructive and non-contact thermal imaging method can be used to investigate the resistance switching mechanism of the PCMO device and as a novel tool for charazterization
4:45 PM - B7.2
Characterizing Strength of Thin Silicon Samples.
Stephan Schoenfelder 1 , Joerg Bagdahn 1
1 , Fraunhofer Center for Silicon Photovoltaics, Halle Germany
Show AbstractThe decrease in thickness of substrates in semiconductor and photovoltaic industry leads to large wafer deformation and higher sensitivity to loading in handling or process steps. Thus an increasing breakage rate in production and in application can be noticed. In order to design mechanically reliable products the strength behavior and fracture mechanism has to be investigated for thin wafers and devices. It is known, that in semiconductor industry the thinning and separation process steps strongly influence the strength of the final device. In this work the problems and effects in characterizing the strength behavior of thin chips are explained and models are presented which allow a reliable determination of the strength values regarding different manufacturing processes like grinding and dicing. Dicing technologies are generating defects and cracks at the edges of the device while thinning processes influence the defects and strength of the surface. Thus different test setups were used to characterize the strength of the edges and the surface. 3-point bending tests were performed to determine the strength of the edge of a sample. The strength of the surface was investigated by a ball-on-ring setup. Due to the small thickness of the tested silicon samples from 200μm down to 50μm nonlinearities in the experiment has to be considered in the stress evaluation. Different finite element models are introduced to solve the problem with respect to large deflection. In particular a model for ball-on-ring test is presented which includes the nonlinearity of large deflection and the problem of a single load in the center of the sample. Further it was noticed in the experiments that the loaded area and volumes changes because of the nonlinear behavior. Thus the size effect of strength has to be considered in the statistical evaluation of the results. But the size effect depends on the parameters of strength themselves which are commonly described by the Weibull distribution. It is an approach presented based on the maximum-likelihood estimation to determine the parameter of strength and consider the size effect. The size effect shows a larger influence in the ball-on-ring test than in the 3-point bending. In the end the different models are applied to investigate different dicing technologies including sawing and dicing-before-grinding (DBG) and the surface strength of thinned samples. The dicing technologies show a broad variation of characteristic strength from 0.7 GPa for sawn samples up to 2.6 GPa for DBG samples. The surface strength was measured with characteristic values in the range of 1.6 GPa to 11.5 GPa. Single samples showed fracture stress values up to 14 GPa which is close to the theoretical strength of silicon. The results can be used to describe the strength and reliability of a silicon chip in application.
5:00 PM - B7.3
The Effect of Passivation SiON Layer on the Data Retention Reliability of NAND Flash.
Younggeun Jang 1 , Kwangwook Lee 1 , Jungmyoung Shim 1 , Sangdeok Kim 1 , Junggeun Kim 1 , Sangwook Park 1 , Byungseok Lee 1 , Jinwoong Kim 1
1 Flash Process Development, Hynix Semiconductor Inc., Icheon-si Korea (the Republic of)
Show Abstract Data retention is one of the major device reliabilities of NAND Flash memory. We found that the lower Refractive Index(RI) of the Passivation Silicon Oxynitride(SiON) layer deposited by PECVD, the better data retention behavior was achieved. The hydrogen content and the stress analysis of the films are analyzed to find out which is more important in this case. Generally, when the RI of SiON decreases, both parameters also decrease, so it is impossible to find out which parameter is major factor of data retention. To analyze the effects of two parameters separately, we applied two conditions which had the same H contents but quite different stress values. The final data retention levels are same in both conditions. In addition, even if the layer has the same H content, the retention characteristic is changed by how hydrogen is bonded in the film. In conclusion, the data retention characteristic can be explained by mobile ions generated by the hydrogen weekly bonded in PECVD SiON films in our experiment.
5:15 PM - B7.4
Reliability in Mechatronics Systems from TEM, SEM and SE Material Analysis.
Pierre Dahoo 1 3 4 , Armelle Girard 1 2 , Nadim Alayli 2 3 4
1 Physics & Engineering Science, University of Versailles St Quentin en Yvelines, Versailles France, 3 Laboratoire ATmosphère Milieux Observations Spatiales , CNRS UMR 8190, BP3, 91371, Verrières-le-Buisson France, 4 LATMOS, Institut Pierre Simon Laplace-Université de Versailles-Saint-Quentin, Saint-Quentin en Yvelines France, 2 Laboratoire d'étude des Microstructures , CNRS-ONERA UMR, BP72, 92320 Chatillon cedex France
Show AbstractThere is a growing interest in materials for power electronics in electric power steering hybrid or electric vehicles. Among the different goals in this interest which range from reduction of green house effect gases from automobile industry or meeting eco-developpement strategies, one of these concern the improvement in connexion reliability in mechatronics lead free soldered systems designed for these vehicles. A mechatronic system results from the simultaneous design and integration of mechanical and couples systems, electronic systems and control and information technology. This work focuses on the electronic systems, in which improvement of die attach reliability is necessary. Understanding the failure mechanisms is necessary to achieve such the aim of this work. Poor reliability are, for example, caused by initial failures such as cracks at bond interfaces or at die layers and their propagation, the existence of voids or defects formation at metal grains, delamination of oxide leading to breaking of joints. These phenomena have been studied by Transmission Electronic Microscope (TEM), Scanning Electronic Microscope (SEM) and Spectroscopic Ellipsometry (SE). Results obtained from the studies on a material studied to replace soldered joints after different stresses such as thermal or vibrational have been applied will be presented and compared. In order to determine the physical processes responsible for the failures, simulation by COMSOL software have also been performed and a tentative analysis of experimental results will be proposed.
5:30 PM - B7.5
Effects of Geometry, Mode Mixity, and Temperature on Dislocation Nucleation in Strained Electronics.
Tianlei Li 1 2 , Jinhaeng Lee 1 , Yanfei Gao 1
1 Department of Materials Science and Engineering, University of Tennessee, Knoxville, Tennessee, United States, 2 Computer Science and Mathematics Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States
Show AbstractDislocation loops may be nucleated from sharp geometric features in strained micro- and nano-electronic devices. This process is investigated by a dissipative cohesive interface model which treats the dislocation core as a continuous, inhomogeneous lattice slip field. As a representative example, we calculate the critical stress for dislocation nucleation from the edges/corners of a rectangular Si3N4 pad on a Si substrate as a function of geometric parameters such as the length-to-height ratio and the three-dimensional shape of the pad. The shapes of the dislocations are also simulated. An important observation arises from the mode mixity of the singular stress fields near the edge. The relationship between the critical stress intensity factor and the mode mixity can be determined from this explicit model of dislocation nucleation process zone. The dependence of both critical load and activation energy on modulus mismatch as characterized by Dundurs parameters will be presented. Comparisons to a number of recent experiments will be made.
B8: Poster Session
Session Chairs
Wednesday AM, December 02, 2009
Exhibit Hall D (Hynes)
9:00 PM - B8.1
Low-resistance Ti/Au Metal Contacts to n-Type Amorphous Gallium Indium Zinc Oxides.
Hyunsoo Kim 1 , Kyoung-Kook Kim 2
1 School of Semiconductor and Chemical Engineering and Semiconductor Physics Research Center, Chonbuk National University, Jeouju Korea (the Republic of), 2 Dept of Nano-Optics, Korea Polytechnic University, Siheung Korea (the Republic of)
Show AbstractAmorphous gallium indium zinc oxide (a-GIZO) has been extensively investigated due to its excellent physical, electrical and optical properties such as high electrical mobility, good reliability with stress current, optical transparency at the visible wavelength, insensitivity to the light as well as low process temperature, thus enabling a fabrication of next-generation transparent thin film transistors (TFTs). In order to fabricate a-GIZO TFTs with better electrical performance, it is essential to obtain low resistance and thermally stable Ohmic contact on source and drain regions of a-GIZO. It was shown that the electrical properties of a-GIZO TFT including the maximum drain-source current, field-effect mobility, and threshold voltage could be significantly affected by the source/drain electrode materials and the post annealing process. For example, the Ar plasma treatment on the source/drain region of the a-GIZO TFT resulted in a reduction of the series resistance as a result of the formation of better Ohmic contact. However, a detailed study on the formation of metallization contact to a-GIZO has yet been reported. In this presentation, we report on the systematic study of Ti (50 nm)/Au (80 nm) Ohmic contacts to n-type a-GIZO. The specific contact resistance of the Ti/Au contact decreased with increasing thermal annealing temperature, producing the contact resistance of as low as 2.85×10-5 Ωcm2 when annealed at 500 °C for 1 min in N2 ambient. This could be attributed to a significant increase in both electron concentration and the mobility of a-GIZO as a result of the combined effects of the formation of oxygen vacancies, partial crystallization of amorphous structures and the formation of indium-rich region at near the surface, which were analyzed from the Hall measurement, SIMS, XRD, and AFM.
9:00 PM - B8.10
Electrically Stable InGaZnO4 Thin Film Transistors using Sputter-deposited PMMA Gate Dielectric and Passivation Layers.
Dong Hun Kim 1 , Nam Gyu Cho 1 , Ho Gi Kim 1 , Ill-Doo Kim 2
1 Department of Materials Science and Engineering, KAIST, Daejeon Korea (the Republic of), 2 , KIST, Seoul Korea (the Republic of)
Show Abstract InGaZnO4 thin film transistors (TFTs) fabricated by employing high-quality inorganic gate dielectrics, such as HfO2, BST-MgO, SiO2 have exhibited good electrical performance. Although InGaZnO4 TFTs with high-k gate dielectrics have been shown to induce low voltage operation and high performance, poor gate leakage characteristics of room temperature-processed high-k gate dielectrics have remained a major drawback for their use in practical applications. As an alternative to inorganic gate dielectrics, insulating polymers have been extensively used as gate dielectrics for the fabrication of organic or ZnO based TFTs. These include PVP (poly-4-vinylphenol), PVA (poly vinylalcohol), and PMMA (poly methylmethacrylate) thin films which are typically deposited by spin-coating, spray coating or printing techniques. However, it’s not easy to achieve uniform pinhole-free thin films in solution processes when the thickness is less than 100 nm, although several groups have achieved pinhole-free films using cross-linked polymers or polymer blends. In particular, a solution technique often requires complex processing variables and needs an additional baking step to remove the solvent which is detrimental to plastic substrates. In response to these drawbacks, there have been several attempts to prepare polymer thin films such as PI (polyimide), PTFE (polytetrafluorethylene), and PMMA by utilizing a vacuum process such as sputtering. However, to the best of our knowledge, there has been no report on the fabrication of sputter-deposited PMMA layers for use as gate insulators in InGaZnO4 TFTs. Furthermore, in-depth studies on dielectric and current leakage characteristics of sputter-deposited PMMA thin layers have been rare. In addition, passivation layers are necessary for long-term current stability in the operation of the TFTs. So far, various organic or inorganic passivation layers such as polymer, SiO2, SiNx, and Al2O3 have been investigated. There has been no report on sputter-deposited PMMA thin layer used as a passivation layer in InGaZnO4 TFTs. In this regard we report on the fabrication and characterization of sputter-deposited PMMA thin films used as gate insulators as well as passivation layers in high performance InGaZnO4 TFTs. Sputter-deposited PMMA thin films exhibited a dielectric constant of 4.3 and low leakage current characteristics (< ~2 × 10-8 A/cm2 at 0.3 MV/cm). The InGaZnO4 TFTs utilizing PMMA gate insulators and PMMA passivation layers exhibited a high on/off current ratio of 4.08 × 106 and a high field effect mobility of 36.1 cm2/Vs. Threshold voltage and field effect mobility remained constant after aging in air atmosphere for 5 months.
9:00 PM - B8.11
Density-Functional Analysis on Vacancy Orbital and its Elastic Response of Silicon.
Takafumi Ogawa 1 , Kenji Tsuruta 2 , Hiroshi Iyetomi 1 , Hiroshi Kaneta 1 , Terutaka Goto 1
1 Graduate school of science and technology, Niigata University, Niigata, Niigata, Japan, 2 Graduate school of natural science and technology, Okayama University, Okayama, Okayama, Japan
Show AbstractElastic softening of non-doped and B-doped silicon grown by the floating zone method is observed via ultrasonic measurements at low temperature [T. Goto et. al., J. Phys. Soc. Jpn 75, 044602 (2006)]. Reduction of the elastic constants C44 and (C11-C12)/2 appears with decreasing temperature below 20K down to the lowest temperature measured at 20mK. While the behavior of elastic constants of non-doped silicon is almost independent of applied magnetic field up to 16T, the elastic softening at low temperature of B-doped silicon is suppressed strongly by magnetic field beyond 2T. It has been suggested that the observed softening is led by the electronic quadrupole of triply degenerate orbitals associated with an isolated vacancy, which couples to strain field induced by ultrasonic waves. This model for non-magnetic charge state of vacancies V0 in non-doped silicon and for magnetic charge state V+ in B-doped silicon can explain the dependence on magnetic fields of the elastic softening in the experiment. Since the temperature dependence of the elastic constants is related to concentration of vacancies, this method has possibility to open a new area of technology for precise measurements of the defect concentration in silicon wafers. In order for precise estimation of the concentration, nature of the localized vacancy state and its coupling strength with the strain field have to be known. We perform ab initio simulations based on the density-functional theory in order to investigate multipole structures of the localized vacancy orbital and their dependence on simulation cell size. The multipole properties are essential for understanding response of the electronic structure to elastic strain of ultrasonic wave at low temperature. Since the orbitals associated with vacancies can extend over a few neighboring atoms [K. Tsuruta et. al., Proceedings of The Forum on the Science and Technology of Silicon Materials 2007, 75 (2007)], large-scale simulations are prerequisite for a quantitative study. We calculate localized electronic states of a vacancy in relaxed silicon crystals with up to N=511 atoms. Quadrupole susceptibility of the localized states is then estimated and compared with the experimental estimations. The calculated quadrupole of Γ5 is ten times larger than that of Γ3 for the vacancy with cubic symmetry. This result is consistent with the difference of magnitude of elastic softening between C44 and (C11-C12)/2 in the ultrasonic experiments. The present work was supported by "Innovation Research Project on Nanoelectronics Materials and Structures" from New Energy and Industrial Technology Development Organization of Japan.
9:00 PM - B8.12
Comparative Study of Inductively Coupled Plasma Induced Damage in MOCVD GaN on GaN and Sapphire Substrates.
Yutaka Tokuda 1 , Kazuki Akiyama 1 , Yujiro Yamada 1 , Hiroyuki Ueda 2 , Osamu Ishiguro 2 , Narumasa Soejima 2 , Tetsu Kachi 2
1 Department of Electrical and Electronics Engineering, Aichi Institute of Technology, Toyota Japan, 2 , Toyota Central R&D Inc., Aichi Japan
Show AbstractWe have characterized inductively coupled plasma (ICP) induced damage in n-type GaN grown by MOCVD on GaN substrates using DLTS for fabricated Schottky diodes and compared obtained results with those in n-type MOCVD GaN on sapphire substrates. The ICP etching was performed under a Cl2 plasma at the source power of 300 W and with the rf power of 30 W for 1 min followed by the rf power of 10 W for 2 min. DLTS measurements were carried out in the temperature range from 80 to 400 K for Schottky diodes fabricated after ICP etching. Samples received the thermal annealing at 650°C for 1 min during the ohmic contact fabrication.Two electron traps labeled E2 (0.25 eV) and E5 (0.59 eV) are commonly observed in the control samples on both GaN and sapphire substrates. E2 and E5 trap concentrations in GaN on GaN are lower than those in GaN on sapphire, respectively. There is almost no change in trap concentration by ICP etching for GaN on GaN, which suggests that the thermal annealing at 650°C for 1 min effectively removes the ICP-induced damage. However, there is an increase in E2 trap concentration by ICP etching for GaN on sapphire. This means that the ICP-induced damage for GaN on sapphire is more severe than that for GaN on GaN and that the thermal annealing at temperatures more than 650°C or the prolonged annealing at 650°C is needed to remove the residual damage in GaN on sapphire. In this context, GaN on GaN is superior to GaN on sapphire in the fabrication of devices using ICP etching. Since E2 trap has been reported to be related to N vacancies, ICP etching produces more N vacancies in GaN on sapphire than in GaN on GaN. This will be discussed by the difference in dislocation density between two GaN samples.
9:00 PM - B8.13
Effect of W Doping Composition Ratio on the Optical Properties of VO2 Thin Films.
Hyun Koo 1 , Il Hwan Yoo 1 , Sung-Hwan Bae 1 , Dong Min Shin 1 , Chan Park 1 2
1 Materials Science and Engineering, Seoul National University, Seoul Korea (the Republic of), 2 Research Institue of Advanced Materials, Seoul National University, Seoul Korea (the Republic of)
Show AbstractVO2 is one of the most interesting thermochromic materials due to its semiconductor-to-metal transition. It is reported that the transition temperature of VO2 can be lowered by doping high valent ions, especially W. However, doping W degrades the electrical and optical property change between semiconductor and metal state of VO2. In this work, W doped VO2 thin films were deposited varying composition ratio. The effect of composition ratio of W will be analyzed by EDS, XRD and XPS. Effect of the number of valence electrons for W ions will be discussed as well.Acknowledgement: This research was supported by a grant (code 08-Technology Innovation-A01) from Construction Technology Innovation Program(CTIP) funded by Ministry of Land, Transport and Maritime Affairs(MLTM) of Korean Government.
9:00 PM - B8.14
Photocapacitance Transient Study of Silicon Doped GaN.
Mo Ahoujja 1
1 Physics, University of Dayton, Dayton, Ohio, United States
Show AbstractGaN and its related ternary compounds involving Al and In have been investigated extensively for several years now because of their various applications, including UV-blue light emitting diodes, blue laser diodes, solar-blind ultraviolet UV detectors, and high temperature/high power electronics. However, in spite of the recent advances in the production of light emitting diodes and laser diodes, the existence of defects still hampers the performance of these devices. Therefore, understanding the nature and origins of defects and impurities in group III-nitrides still remains a challenge. In this work, electrical properties of silicon-doped GaN films grown on sapphire substrates by low metalorganic chemical vapor deposition have been investigated using temperature dependent Hall-effect, deep level transient spectroscopy (DLTS), and photocapacitance transient spectroscopy. A fit to the Hall carrier concentration as a function of inverse of temperature gives two shallow energy levels at 24 and 58 meV. The Arrhenius analysis of the DLTS measurements displayed two deep energy levels at 0.22 and 0.60 eV. While DLTS is the most sensitive technique used to characterize deep level defects, it is however of limited use for energy levels deeper than 1 eV. Photocapcitance spectroscopy monitors the change in capacitance as a function of the wavelength of light. When the threshold photoionisation energy is reached to release the trapped charge from a particular defect center, a step appears in the capacitance measurement. The steady state photocapacitance measurements show a number of deep trap levels at 0.17, 0.22, 0.27, 0.54, 0.69, 1.03, and 1.77 eV. An effort to identify the sources of the observed energy levels in Si-doped GaN will be discussed.
9:00 PM - B8.15
Anomalous X-ray Reflectivity Study of InGaN/GaN Multi-quantum Well Structure.
Do Young Noh 1 , Sung Pyo Lee 1
1 , Gwangju Institute of Science and Technology, Gwangju Korea (the Republic of)
Show AbstractIn III-nitride based light emitting diodes(LEDs), the chemical and structural characteristics of the InGaN/GaN multi-quantum well(MQW) used as the active layer affect their optical properties greatly. X-ray reflectivity measurement is a powerful and nondestructive method for analyzing structure of thin films and multilayers. However, it is often difficult to extract the chemical and structural information exactly from the reflectivity measurement, since various factors such as well or barrier density, thicknesses, interface roughness, and the chemical intermixing determine the reflectivity. In this study, we investigated the interfacial structure of InGaN/GaN MQWs using anomalous x-ray reflectivity (AXR) measurement, in which the energy of probing x-rays was varied across the absorption edge of Ga atom 10.373 keV. As the x-ray energy varies across the absorption edge, the effective electron density of Ga changes, which provides the chemical sensitivity to the x-ray reflectivity. Thus, the AXR measurements can reliably determine the exact chemical profiles as well as the layer thickness of each well and barrier of MQWs. In addition at small angles, the total reflection from MQWs was dominated by the diffuse intensity, and the specular x-ray reflectivity(SXR) component was hidden under the longitudinal diffuse reflectivity(LDR) component. Characteristic Bragg reflections corresponding the well-barrier period of the MQW are clearly observed in the LDR, which indicates that the MQW interfaces are very well correlated.
9:00 PM - B8.16
Electrical Characterization of Crossed Electrospun Tin Oxide Nanoribbons in a Field Effect Transistor Configuration.
Jaime Rodriguez 1 , Nicholas Pinto 1
1 Physics, University of Puerto Rico at Humacao, Guaynabo, Puerto Rico, United States
Show AbstractElectrospinning is an easy and effective method for fabricating nanofibers of semiconducting materials. Tin oxide is a technologically very important metal oxide as it is optically transparent and semiconducting. Our goal was to fabricate a field effect transistor using this method. Tin oxide was successfully electrospun to fabricate two crossed nanoribbons arranged in a field effect transistor configuration on a doped Si/SiO2 substrate. Along with the typical source and drain terminals, this device includes both a global gate and a local gate terminal; two terminals not frequently seen incorporated together in a field effect transistor. After initial electrical characterization, the device exhibited n-type field effect transistor behavior with changes in channel conductivity as the global gate voltage varied. The current as a function of drain source voltage, however, suffered a DC shift as the local gate voltage varied. The device retained its field effect characteristics despite the DC shift.
9:00 PM - B8.17
Effects of Sulfur Treatment on N-polar InN Surfaces.
Yuh-Hwa Chang 1 , Yen-Sheng Lu 2 , J. Yeh 1 , Yu-Liang Hong 3 , Cheng-Tai Kuo 3 , Shangjr Gwo 3
1 , Institute of NanoEngineering and MicroSystems,National Tsing Hua University, Hsinchu Taiwan, 2 , Institute of Electronics Engineering, National Tsing-Hua University, Hsinchu Taiwan, 3 , Department of Physics, National Tsing-Hua University, Hsinchu Taiwan
Show Abstract Effects of N-polar InN surfaces treated with ammonium sulfide, (NH4)2Sx, solutions is investigated using Hall effect measurement and X-ray Photoelectron Spectroscopy (XPS). Upon the (NH4)2Sx treatment, the sheet carrier density is reduced by a value of ~8.5×1012 cm-2, which is about one third of its intrinsic electron accumulation density at the near-surface region. By solving the Possion’s equation, the associated decrease of the surface band bending is derived to be 285 meV, consistent with the trend of the peak shift observed in XPS between the as grown and the treated samples. The characterization of XPS shows that the native oxide is greatly removed and a covalently bonded sulfur layer is formed with Indium and Nitrogen atoms on InN surfaces upon the treatment. Thus the (NH4)2Sx treatment has been demonstrated to be an effective method for passivating the surface donor states for InN. The sulfur treatment were performed in (NH4)2Sx (40%) solutions at 40 °C for 30 min. During the (NH4)2Sx treatment, the in-situ resistance variation was monitored. The value of the in-situ resistance variation is in agreement with the Hall effect data measured after the treatment.
9:00 PM - B8.18
Effect of Oxygen Content in Mo Diffusion Barrier on the Reliability of Cu Interconnector.
Gil Ho Gu 1 , Chan Gyung Park 1 2
1 Materials Science and Engineerging, POSTECH, Pohang, Gyungbuk, Korea (the Republic of), 2 National Center for Nanomaterials and Technology, POSTECH, Pohang, Gyungbuk, Korea (the Republic of)
Show AbstractRefractory metals have been used in the diffusion barrier with Cu interconnector line in TFT-LCDs. Molybdenum is the most suitable materials for diffusion barrier due to the low resistivity, high thermal stability and chemical stability better than Ta or Ti on SiO2. However, when the Mo single layer is applied to diffusion barrier, Cu usually diffuse into grain boundary during post processing annealing. The performance of diffusion barrier critically depends on the grain size of diffusion barrier, which can also depend on the oxygen level in Mo layer. Therefore, the measurement of oxygen at the grain boundary is very important. For the measurement of theses oxygen levels, atom probe tomography (APT) is the unique technique providing quantitative compositional information with sub-nanometer spatial resolution. In this study, the analysis of impurity level of oxygen at grain boundary was performed by laser assisted wide angle tomographic atom probe (LAWATAP).In order to identify the oxygen level in the Mo layer, the secondary ion mass spectrometry (SIMS), scanning transmission electron microscopy (STEM) and electron energy loss spectroscopy (EELS) were also used to analyze both the microstructure and the change in chemical composition of the Cu/Mo/SiO2/Si layer during annealing. The results clearly showed that the microstructure of the Mo layer was strongly influenced by the oxygen level. This microstructure changed from columnar structure to nanocrystalline as the oxygen levels increased. Since the oxygen increases gradually the energy required for second grain growth, the grain growth can be suppressed under the condition of high oxygen levels. This hypnosis is supported by the present atom probe tomography analysis results. We also confirmed that most of oxygen is located in the grain boundary of Mo layer. The fine grain size of Mo layer, therefore, can enhance the performance of diffusion barrier more effectively.
9:00 PM - B8.19
The Effect of Surface Morphology and Coating Thickness of Ta2O5-IrO2 Electrode for Electrolysis.
Jung Dae Kim 1 2 , Dongcheol Yoon 2 , Heesoo Lee 2 , Youngmin Oh 3 , Kyu Hwan Lee 4 , Min-Seok Jeon 5
1 , Gyeongnam Regional Small & Medium Business Administration, Changwon, Gyeongnam Korea (the Republic of), 2 School of Materials Science & Engineering, Pusan National University, Busan Korea (the Republic of), 3 , Danam ENE Digital Technology Innovation Center , Busan Korea (the Republic of), 4 , Korea Institute of Materials Science, Changwon Korea (the Republic of), 5 Machinery & Material Testing Center, Korea Testing Laboratory , Seoul Korea (the Republic of)
Show AbstractThe effect of surface morphology and coating thickness on electrochemical properties of Ta2O5-IrO2 multi-layer electrode for electrolysis was studied by conducting durability test with pseudo real environment. The electrode was composed of catalytic and protective layers on Ti substrates and each layer was formed using a solution containing different solid content ratio of Ta2O5:IrO2=2:8 and 8:2. The protective layer was firstly formed on the Ti substrate and catalytic layer was coated on it, which is repeated to target thickness. The dissolution rate of Ta2O5-IrO2 electrode with surface morphology and coating thickness was measured in terms of the variation of electrode capacity and oxygen over-potential with polarization time. The surface morphology and roughness of the Ta2O5-IrO2 electrode was examined by atomic force microscopy and scanning electron microscopy. The crystal structure of the electrodes before and after pretreatment in HCl was investigated by X-ray diffraction. The lifetime of electrode systems was estimated by accelerated degradation test which was performed with continuous anodic electrolysis at high current density.
9:00 PM - B8.2
Reliability of High-Temperature Operation for GaN-Based OPAMP.
Kazuki Nomoto 1 , Kazuya Hasegawa 1 , Masataka Satoh 1 , Tohru Nakamura 1
1 Research Center for Micro-Nano Technology, HOSEI University, Koganei Japan
Show AbstractThere have been increasing applications that require electronic devices, digital and analog circuits to operate at higher temperatures. These applications include engine control, DC-DC converters and analog-digital converters for automotive hybrid systems. Wide bandgap FETs have fundamental advantages for high-temperature operation. AlGaN/GaN HEMTs and GaN-based digital ICs operating at above 265 oC have been reported, but there has few reports about analog circuit operation. It is important to characterize the high-temperature behavior of GaN-based analog circuits.The GaN/AlGaN/GaN HEMT structure was grown on silicon substrate by MOVPE. Silicon ions were implanted into Source/Drain regions in GaN/AlGaN/GaN HEMTs and resistor regions at the energy of 80 keV with ion dose of 1.25 × 1015 /cm2. This was followed by additional ion implantation at the energy of 30keV with a dose of 0.25 × 1015 /cm2 through a 25 nm thick SiNx layer to obtain low contact resistance for ohmic contact to the GaN layer with high surface carrier concentration. The sample was annealed at a temperature of 1200 oC for 2 min to activate implanted Si. Source and drain contacts were formed by depositing Ti/Al layers. Finally, gate contacts were formed by depositing Ni/Al layers.Id-Vgs and gm-Vg characteristics of Ion-implanted (I/I) HEMTs were measured. Maximum drain current of 44.6 mA at Vg=1 V and maximum transconductance of 14 mS at Vd=5 V were obtained for the I/I HEMTs because of extremely low source resistance. The improvement of DC characteristics of I/I HEMTs was caused by remarkable reduction of source/drain resistance. At operation temperature of 200 oC, I/I HEMTs operated with maximum drain current of 25.5 mA. Change in the resistance of I/I regions was very slight in the measured temperature range, compared to HEMT devices. Fabricated OPAMP consists of a source coupled differential circuit and level shift circuits. The level shift circuits were used two D-mode I/I HEMTs and series connected Schottky barrier diodes. The die size is 1.0 mm × 0.5 mm. Transfer characteristics of the OPAMP at 23 and 200 oC were measured. Supply voltages of ±15 V were used. Vg1 and Vg2 were biased to -15 V. Under this bias condition, the constant current flowing differential circuit was about 22 mA, and the total current from power supply was 32 mA. For Vin ranging from -2 to 2 V, Vout shows the output voltage swings from 5 to -5 V. The I/O gain of about 2.5 was obtained, but it was decreased with the increase of temperature. At the temperature of 200 oC, the I/O gain was estimated to be 2. We have demonstrated reliability of high temperature operation of OPAMP which consists of I/I HEMTs and I/I resistors on silicon substrate. The fabricated OPAMP was measured under the voltage stress, and then it can operate at the temperature of 200 oC with the I/O gain of 2 without degradation.
9:00 PM - B8.21
Selective Electrochemical Etching of GaN Having More Than Two n-GaN Layers.
Baro Lee 1 , Joonmo Park 1 , Kwang Min Song 1 , Sang-Wan Ryu 1
1 Physics, Chonnam National University, Gwangju Korea (the Republic of)
Show AbstractAn electrochemical etching based on oxalic acid was developed for use in the chemical lift-off of GaN epitaxial structures. It was shown that for GaN having an n-GaN layer the Si-doped n-GaN layer was only etched away, while the undoped GaN layer was not etched at all. This doping selective etching was applied to etch the GaN sample included two n-GaN layers with different doping concentration. The more doped n-GaN layer was selectively etched away, leaving behind the lower doped n-GaN and undoped GaN layers. These results indicate that GaN having more than two n-GaN layers was selectively etched and it can be applied to the fabrication of vertical LED.
9:00 PM - B8.22
Fabrication of Thin Film Ti-Si-C for Reliable Metallization to Power AlInN/GaN HEMTs.
Michal Borysiewicz 1 , Marie-Antoinette di Forte-Poisson 2 , Marek Ekielski 1 , Eliana Kaminska 1 , Anna Piotrowska 1 , Rafal Jakiela 3 , Elzbieta Dynowska 3 , Tomasz Wojciechowski 3
1 , Institute of Electron Technology, Warsaw Poland, 2 , Alcatel-Thales III-V Lab, Marcoussis France, 3 , Institute of Physics, PAS, Warsaw Poland
Show AbstractDiscussed are possible applications of sputter deposited Ti-Si-C material phases for electric metallisations to semiconductor devices for which resistivity to elevated temperatures and corrosive environments is of importance. The research reported focuses on fabricating a Ti3SiC2 MAX phase known for its resistance to high temperatures and oxidation with high electrical conductivity. Compared are three approaches to magnetron sputtering fabrication: deposition from a compound Ti3SiC2 target, multilayer deposition of TiC and Si with subsequent annealing and cosputtering from Ti, Si and C targets.
X-ray diffraction, secondary ion mass spectroscopy and scanning electron microscope structural studies are performed. I-V characteristics are gathered and contact resistivities are extracted for ohmic contacts using the c-TLM method, and the heights of Schottky barriers are calculated using the Cheung approach. Yielded values are φB = 0.54 eV and ρc = 2*10-3 Ωcm2. Contact structurisation is performed by means of UV-photolithography and Reactive Ion Etching in a CF4/O2 plasma. The obtained results are discussed as a function of deposition temperatures and process parameters.
The study was partially supported by the EC under the project “Materials for Robust Gallium Nitride” CP-IP 214610-2 MORGaN and by the European Union within European Regional Development Fund, through grant Innovative Economy (POIG.01.03.01-00-159/08, "InTechFun").
9:00 PM - B8.23
Microstructure Study of the Oxidation of TiN Layers During Sputtering Process.
Chun Wang 1
1 Department of Electrical and Computer Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States
Show AbstractEpitaxial SrRuO3/SrTiO3 (001) thin films with a TiN template layer have been deposited on Si(001) single crystal substrates by rf sputtering. The epitaxial orientation relationship was determined to be cube on cube with respect to Si and the crystal quality of the SrRuO3 /SrTiO3 film is preserved even when the TiN template layer was oxidized into anatase phase of TiO2 during the sputtering process of SrRuO3. The effect of oxygen plasma on the oxidation and delamination of the TiN layer has been studied using transmission electron microscopy (TEM). The stress in the thin film of SrRuO3 /SrTiO3 /TiN structure was determined from the buckle shape in both plan view and cross-sectional TEM images. The critical stress and the compressive stress were estimated to be 2 and 4 GPa.
9:00 PM - B8.24
Temperature Dependent Optical Band Gap Measurements of III-V Films by Low Temperature Photoluminescence Spectroscopy.
Linda Casson 1 , Francis Ndi 1 , Eric Teboul 1
1 , HORIBA Scientific, Edison, New Jersey, United States
Show AbstractPhotoluminescence (PL) spectroscopy is a powerful technique for probing the structures of many types of III-V semiconductor materials. When a semiconductor material is excited at a particular wavelength, electron-hole pairs are generated. The most intense radiative transition is between the conduction band and valence band, and this measurement is used to determine the material band gap. Radiative and non-radiative transitions in semiconductors also involve localized defect levels. The photoluminescence energy associated with these levels can be used to identify specific defects, and the amount of photoluminescence can be used to determine their concentration, and thus predict device quality. At ambient temperatures, the PL signal is typically broad, as much as 100nm in width. When cooled, structural details may be resolved, and a small spectral shift between 2 samples may represent a change in a structural parameter. Thus a system with high spectral resolution is required.In this paper, a modular Low Temperature Photoluminescence system (LTPL) for measuring optical band gap as a function of temperature is described. Results show that the optical band gap shifts towards higher energy as the sample temperature decreases.
9:00 PM - B8.25
Time-resolved PL and EL Measurements of PV Materials using a Modular TCSPC System.
Linda Casson 1 , Arnaud Vigier 1 , Salvatore Atzeni 1
1 , HORIBA Scientific, Edison, New Jersey, United States
Show AbstractTime-resolved photoluminescence spectroscopy is an important tool which can be used to characterize photovoltaic (PV) materials. A popular and straightforward technique, Time-Correlated Single Photon Counting (TCSPC), allows measurement of luminescence decay times in the picosecond to millisecond timescales. Advances in instrumentation have made TCSPC accessible to all researchers as a routine technique in their laboratories.Photoluminescence emission spectra, lifetime(s) and their dependence on the level of photo-excitation and temperature are directly related to dominant recombination processes. Analysis of steady-state and time-resolved photoluminescence (TR-PL) aids understanding of underlying physics of the recombination mechanism when excitation by photons of sufficient energy is used. Emission spectra as functions of decay time (time-resolved emission spectra, or TRES) can identify excited-state behavior in these materials.Alternately, PV devices can be electrically stimulated to emit light, yielding EL emission (electroluminescence). Adding temporal measurement capability to measure time-resolved electroluminescence (TR-EL) provides a further dimension of understanding the device behavior. Measurement and comparison of both the photoluminescence and electroluminescence lifetimes of a material can provide detailed information about the decay mechanisms and efficiency of the device. As with TRES measurements of TR-PL, the emission dependence of the TR-EL can identify excited-state processes. With appropriate cryostat hardware, it is also possible to investigate the effect of temperature on the dynamics exhibited by both TR-PL and TR-EL, even at high spectral resolution (subangstrom) levels.In this presentation we outline a compact, modular TCSPC system for characterization of PV materials using TR-PL and TR-EL.
9:00 PM - B8.26
Transparent, Flexible Composite Barrier Films from Solution.
Jeffrey Gerbec 1 3 , Jimmy Granstrom 3 4 , Hunaid Nulwala 2 3 , Motoko Furukawa 1 , Luis Campos 2 3 , Craig Hawker 2
1 , MC Research and Innovation Center, Goleta, California, United States, 3 Mitsubishi Chemical Center for Advanced Materials, University of California, Santa Barbara, California, United States, 4 Center for Polymers and Organic Solids, University of California, Santa Barbara, California, United States, 2 Materials Research Laboratory, University of California, Santa Barbara, California, United States
Show AbstractAs flexible organic electronic devices become more of a commodity, the need for multifunctional composite encapsulating films has risen sharply. The form factor of these next generation devices offer new challenges in packaging and barrier film technology for device durability of which include flexible organic photovoltaics, organic light emitting diode displays and CIS photovoltaic devices. Ultra high barrier film technology on the order of 10-7 g m-2day-1 exists today by established techniques such as chemical vapor deposition (CVD) and atomic layer deposition (ALD). However, the cost of the barrier film far exceeds that of the electronic device itself, limiting its commercial applicability. Historically, ceramic glass has been the hallmark for hermetic sealing of electronically sensitive devices which require an “absolute” barrier from the ambient environment. The limitation is that the typical inorganic glasses do not have elastomeric character and sintering temperature is in excess of 350°C. The ideal, low cost, material would have the absolute gas barrier characteristics of ceramic glass with the processability and formability of a room temperature liquid solution.Work has been devoted to formulating such materials. These materials are commonly known as ORMOCERS, CERAMERS, and Hybrid Glass, sometimes called spin-on glass. These materials impart the physical properties of ceramic glass, i.e. high temperature durability, robust mechanical properties and chemical resistance with the room temperature processability of liquid precursors. We will demonstrate unique, low cost, solution deposited elastomeric materials for use in flexible, thin film barrier technology. We will show material platforms based on elastomeric inorganic hybrid systems, oxide nanoparticle assemblies and thiol-ene elastomers. The typical architecture of commercial barrier films is used by forming alternating layers of inorganic and organic networks on polymeric substrates. We will show Ca lifetime results for water transmission rates, layer formation methodology and characterization by scanning electron microscope (SEM) illustrating superior interlayer adhesion for polyester-based multilayers.
9:00 PM - B8.27
Dependency of Indium Concentration on Structural Defects in MOVPE-grown InGaN/GaN Heterostructures.
J. Dudding 1 , D. Korakakis 1 2
1 Lane Department of Computer Science and Electrical Engineering, West Virginia University, Morgantown, West Virginia, United States, 2 , National Energy Technology Laboratory, Morgantown, West Virginia, United States
Show AbstractIndium Gallium Nitride (InGaN) based optoelectronic devices have garnered much attention recently due to the ability to tailor the bandgap ranging from blue/green visible light to UV—allowing for creation of blue/green LEDs and laser diodes. [1] However, many growth-related problems due to the large lattice mismatch and low miscibility between GaN and InN lead to the formation of V-defects at the termination of threading dislocations. [2] In addition, growth of InGaN at lower temperatures to promote increased indium incorporation results in a poorer quality surface morphology. [3] The need for higher quality may not be as evident by observing the current state of InGaN-based blue/green LEDs and laser diodes. While these devices have many areas for improvement in regards to efficiency and power consumption, they are able to overcome defect densities as high as 1010 cm-2 to provide strong electroluminescence characteristics. However, InGaN-based photovoltaic devices are much more defect-sensitive due to the presence of carrier traps serving as recombination centers, thus inhibiting the transport of holes and electrons to the device electrodes. [4] In this case, material quality strictly governs the operation of the device—further emphasizing the need to improve material quality. Overall, higher quality InGaN films allows for not only better, more efficient optoelectronic devices but also as a pathway toward achieving InGaN alloys consisting of higher indium concentrations. This work focuses on observing the material quality of InGaN/GaN heterostrucutures by metrics such as size and density of V-defects over varying indium concentrations while investigating techniques to reduce or passivate these defects.InGaN/GaN heterostructures with underlying n-type GaN are grown by metal organic vapor phase epitaxy (MOVPE) on sapphire with AlN buffer layer. For a multiple quantum well configuration, variables such as barrier/well widths, growth temperatures and metal organic precursor flows will be adjusted to determine the impact of indium concentration and structural configuration on material quality. Characterization of indium concentration will be performed by X-ray diffraction (XRD), while structural characterization will be performed by atomic force microscopy (AFM) and scanning electron microscope (SEM).[1] Matsuoka T., Superlattices and Microstructures 2005, 37 (1)[2] Ponce F.A., Srinivasan S., Bell A., Geng L., Liu R., Stevens M., Cai J., Omiya H., Marui H., Tanaka S., Physica Status Solidi C 2003, (7)[3] Oliver R.A., Kappers M.J., Humphreys C.J. and Briggs G.A.D., Journal of Applied Physics 2005, 97 (1)[4] Sheu J.-K., Yang C.-C., Tu S.-J., Chang K.-H., Lee M.-L., Lai W.-C. and Peng L.-C., IEEE Electron Device Letters 2009, 30 (3)
9:00 PM - B8.28
Cyclic Failure Mechanism of Gold Thin Film on Polydimethylsiloxane (PDMS).
Onobu Akogwu 1 2 , David Kwabi 2 , Winston Soboyejo 2
1 Electrical Engineering, Princeton University, Princeton , New Jersey, United States, 2 Mechanical and Aerospace Engineering, Princeton University, Princeton, New Jersey, United States
Show AbstractThis poster presents the results of a novel experimental technique to study the effects of cyclic loading on failure mechanism in nano-scale Au thin films deposited on a flexible poly-di-methyl-siloxane (PDMS) substrate. The deformation and cracking mechanisms are elucidated as functions of film thickness and shape. The implications of the results are also discussed for the design of flexible electronic structures.
9:00 PM - B8.29
Thermo-mechanical Property of Porous low-k Material.
Suk Hoon Kang 1 , Tae Jun Ko 1 , Kyu Hwan Oh 1 , Woong Ho Bang 2 , Liangshan Chen 2 , Choong-Un Kim 2
1 , Seoul National University, Seoul Korea (the Republic of), 2 , University of Texas at Arlingtion, Arlingtion, Texas, United States
Show AbstractPolysilsesquioxanes (MSSQ) are one such class of spin on glass dielectrics, which is known to have relatively low dielectric constant, minimal moisture uptake and high thermal stability. For now, it is common assumption that the porous structures of MSSQ are also stable over operations. However, recent investigation finds that pores are not necessarily stable and undergoes active reconfiguration when they are subjected to hot or cold condition. In this study, the instability of porous low-k dielectrics in blanket film and interconnect is investigated. The plastic behaviors of low-k have been observed in large area of samples; moreover, it is supposed that the deformation behavior of porous low-k is viscous. FTIR results also show that cross-linked siloxane bonds in low-k are changed after cold or hot treatment. The above suggests that the porous low-k structure can be instable in various operational conditions. The problem would be treated as an important reliability issue in interconnect industry.
9:00 PM - B8.3
High Current Gain Triple Ion Implanted 4H-SiC BJT.
Taku Tajima 1 , Tadashi Nakamura 1 , Yuki Watabe 1 , Masataka Satho 1 , Tohru Nakamura 1
1 , Hosei Univ., Koganei-Shi, Tokyo Japan
Show AbstractSiC is a promising material for high temperature and frequency devices because of high breakdown field strength and wide bandgap. A lot of work has been done in recent on SiC devices. Especially, SiC bipolor junction transistors (BJTs) are expected to be high temperature devices. In the fabrication process of SiC BJT, the ion implantation was only used to form p+ contact region to the base because crystal defects remain in the ion implanted emitter and base regions which increase base recombination current, decrease current gain and affect the quality and reliability. We report superior characteristics of 4H-SiC BJT devices with fabricating only ion implantation techniques. In this study, we demonstrate the technology to improve current gain of triple ion implanted 4H-SiC BJT device.(0001) 4H-SiC n-type epitaxial substrate was used for fabricating triple ion implanted 4H-SiC BJT. The depth of intrinsic base is 0.3 μm. The width (WE) and length (LE) of the emitter were 100 and 100 μm, respectively. The base Al concentration is 2 x 1017/cm3. The donor concentration and thickness of the emitter is 4 x 1019 /cm3 and 0.5 μm, respectively. The averaged Al concentration in the p+ contact region is also 4 x 1019 /cm3. The annealing of base and emitter regions were carried out using RF furnace heating at 1700 oC for 30 min and EBAS at 1900 oC for 1 min, respectively. To etch extrinsic base regions about 420nm, ICP dry etching equipment was employed. The ohmic contacts to the base and emitter are fabricated by depositing Ni metal layer with a thickness of 0.12 μm and subsequently annealing at 900 oC for 3 min in Ar gas flow. DC characteristics were measured for triple ion implanted 4H-SiC BJT. With etching extrinsic base regions for 420nm, maximum common emitter current gain was obtained about 10. In the low injection level of Gummel plots, the ideality factor of IB and IC was estimated to be 2 and 1, respectively. It shows the device operates at intrinsic region and base recombination current was reduced. We investigated the high current gain 4H-SiC BJT. The current gain was improved by etching extrinsic base regions 420 nm. To etch extrinsic base regions with crystal defects have remarkable effect for current gain of 4H-SiC BJT.
9:00 PM - B8.32
Tuning Work Functions of Mo and Cu Films with O2 Annealing Processes using Furnace and RTP.
Dong-Jin Yun 1 , Jung-Min Kim 1 , Shi-Woo Rhee 1
1 chemical enginnering, POSTECH, Pohang, kyungbuk, Korea (the Republic of)
Show AbstractThis research is focused to improve the interface properties between metal and organic semiconductors. O2 annealing processes with rapid thermal annealing (RTA) and furnace were applied to tune surface properties including work function without change in bulk properties. The Mo/Ti (50/3nm) and Cu/Ti (50/3nm) films were annealed at different temperature and characterized with various analytical tools to find optimum conditions. At high temperature, both Mo/Ti and Cu/Ti films were fully oxidized and bulk properties of them such as resistivity, crystallinity and grain size were also changed. But at specific annealing temperature, the binding energy and work function could be changed without change in bulk properties. For example, in case of the Mo/Ti film annealed at 300oC in RTA, the work function of film was increased from ~ 4.6eV to ~ 5.2eV, but resistivity, roughness and crystallinity were not changed. Additionally, pentacene TFTs with bare and annealed Cu/Ti (and Mo/Ti) S/D electrodes were fabricated and their performances were compared. Pentacene TFT with annealed S/D electrode shows better performance than pentacene TFT with bare S/D electrode due to increase in work function of S/D electrode.
9:00 PM - B8.33
Optimization of Interpoly Dielectric Using Plasma Nitridation for Future Non-volatile Memory.
Ching Yuan Ho 1 , Jr Hau He 1
1 , Photonics and Optoelectronisc of National Taiwan University, Taipei Taiwan
Show AbstractThe effects of various plasma nitridation modes on inter-poly dielectric and charge loss resulting from nitrided floating gate are investigated. The bird’s beak of gate sidewall oxidation and control gate coupling capability are improved by plasma nitridation of ONO stack film, which show excellent performance in terms of programming/erasing speed and comparable gate leakage current. The high wet etching selectivity of diluted hydrofluoric acid (DHF) is proposed to solve the charge loss problem without degradation of control gate coupling. The modified ONO stack film not only enhances memory cell characteristics but also extends the functional limitation of floating gate NAND flash memory to 40 nm node.
9:00 PM - B8.4
Degradation of Current Gain for Ion Implanted 4H-SiC Bipolar Junction Transistor.
Yuki Watabe 1 , Taku Tajima 1 , Tohru Nakamura 1
1 , Hosei University, Tokyo Japan
Show AbstractSiC is a promising wide-bandgap material for high power and high temperature devices because of a high breakdown strength and thermal conductivity. SiC bipolar junction transistors (BJT) are usually fabricated using double or triple mesa-etching of epitaxial layers on SiC substrate. In this structure, it is difficult to obtain high current gain as high as 20. Current gain decreases due to the influence of bandgap-narrowing for heavily doped emitter and base regions, which affect the reliability of the devices because large emitter and base currents flow the surface regions. In this paper, we report that influence of bandgap-narrowing on common emitter current gain to improve the reliability for 4H-SiC BJTs. (0001) 4H-SiC n-type epitaxial substrate was used for fabricating ion implanted SiC BJTs. The base Al concentration and depth were 2×1017/cm3 and 800nm, respectively. Surface base concentration was also kept at 2×1017/cm3. The emitter N concentration and depth were 4×1019/cm3 and 500nm, respectively. Annealing was carried out at 1700 °C for 30 min by RF furnace heating and at 1900 °C for 1min. Ohmic contacts for base and emitter regions were formed by depositing Ni metal layer and subsequently annealed at 900 °C for 3 min in Ar gas flow. DC characteristics were measured and common emitter current gain was obtained at less than 1. To analyze the degradation of the common emitter current gain, components of base and collector currents for the BJT were simulated using ISE TCAD device simulator. It was found that base current flow concentrates in the emitter/base surface regions, which caused surface recombination, and low collector current flew in the intrinsic base region. Simulated results also show that common emitter current gain is obtained at a maximum base concentration of 1×1017/cm3 and a maximum emitter carrier concentration of 1×1019/cm3 for 4H-SiC BJTs. To improve degradation of the common emitter current gain and reliability of the BJT, surface region surrounding the emitter was etched away. The current gain increased to about 7 by etching at 470 nm. In conclusion, heavily doped emitter and base regions affect the reliability and the degradation of the current gain for 4H-SiC BJT.
9:00 PM - B8.6
Reliability Study of GaN-based LEDs with Different Growth Conditions for the Multiple Quantum Wells.
Leung Ka Kuen 1 , Fong Patrick W. K. 1 , Surya Charles 1
1 EIE, The Hong Kong Polytechnic University, Hong Kong China
Show AbstractGallium nitride (GaN) and related compounds are technological important semiconductors for many applications including laser diodes and high power light emitting diodes (LEDs). Device reliability must be properly addressed for effective application of GaN LEDs in general lighting. It is important to understand the effects of the growth parameters for the InGaN/GaN multiple quantum wells (MQWs) on the device reliability. In this paper we report systematic studies on the hot-electron hardness of GaN LEDs as a function of the growth rates of the InGaN QWs. The active regions of the devices consist of 5-period InGaN/GaN MQWs. Two different growth rates for the InGaN/GaN MQWs were used for the fabrication of the devices. The growth rate for type A devices is approximately three times that of the type B devices. Hot-electron stressing on the LEDs was performed using a high dc stressing current density up to 600 Acm-2. Degradations of the optoelectronic and low-frequency noise properties in the LEDs were examined. In particular, low-frequency noise power spectral density is shown to be directly proportional to the defect density of the material and is therefore an effective non-destructive characterization tool for materials degradation. The change in the low-frequency excess noise of the devices was investigated as a function of the stress time. Type B devices exhibit significantly higher rate of increase in the low-frequency noise compared to type A devices. The degradations of the optoelectronic and low-frequency noise properties of the devices are being correlated to the results on structural characterizations of the MQWs using atomic force microscopy (AFM) and transmission electron microscopy (TEM) techniques. Experimental results on AFM show that as-fabricated type B devices typically have about 30% higher dislocation density compared to type A devices. Also, TEM pictures show that type B device exhibit much higher nonuniformity in the QWs. These are believed to arise from the high strain nonuniformity in the QWs due to spiral growth in the material. TEM results show that threading dislocations were generated across the MQWs upon the application of high current stress. The results clearly show that defect generation is much more substantial among type B devices. The dislocations appear to act as diffusion paths for the out-diffusion of the In atoms. This had strong effects on the lifetimes of the LEDs devices.
9:00 PM - B8.7
Novel Procedure Method to Improve of White Light Emitting Diode Uniformity.
Chi-Feng Chang 1 , Chun-Liang Lin 2 3 , Wan-Zhen Li 4 , Hsin-Hua Ho 4
1 Electro Optical Engineering, Kun Shan University, Tainan County Taiwan, 2 Electronic Engineering, Kun-Shan University, Tainan County Taiwan, 3 Nanotechnology R&D Center, Kun-Shan University, Tainan County Taiwan, 4 , Chi Mei LIghting Technology Corp, Tainan County Taiwan
Show AbstractThis procedure method is able to improve the uniformity of white lighting, and coating phosphor powder on the surface of die, then solve problem for deposition. This experiment used YAG: Ce3 + powder is mixed in silicone and made the thin film layer. At the same chromaticity diagram (CIE) compared package method for traditional and new procedure of Chromaticity uniformity. We obtained the traditional package of CIE (X) and (Y) values of difference, maximum to minimum were 0.006 and 0.012. Beside, the new procedure of CIE (X) and (Y) values of difference for maximum to minimum were 0.003 and 0.005. We know when values of CIE (Y) closer could improve uniformity.
9:00 PM - B8.8
Ge/Si(100) Photodetector Based on High-quality Epitaxial Germanium Deposited by RPCVD.
Sang Hoon Kim 1 , Dongwoo Suh 2 , Jiho Joo 2 , Gyungock Kim 2
1 , Thin Film Solae Cell Technology Team, Electronics and Telecommunications Research Institute, Daejeon Korea (the Republic of), 2 , Nano New Technology Research Team, Electronics and Telecommunications Research Institute, Daejeon Korea (the Republic of)
Show AbstractThe pure Ge layer on silicon grown by reduced-pressure chemical vapor deposition (RPCVD) was analyzed with respect to the effects of the growth conditions of the Ge seed layer on the surface roughness and the relaxation degree. We also report a fabricated vertical-incidence Ge on silicon p-i-n photodetector.The pure Ge layer was grown in an ASM “Epsilon One” RPCVD system, and the impact of the growth conditions of the Ge seed layer on the surface roughness, the relaxation degree, and crystalline quality is discussed for forming of the high-quality Ge grown directly onto Si substrate. Our method consisted of two-step process with in-situ annealing treatment. This in-situ annealing step increases misfit dislocation flow and almost relaxes the Ge seed layer. For this reason, misfit dislocations are not restricted to only one interface but have a room for spreading over the thickness range of the Ge seed layer with low threading dislocation density on the Ge capping layer. The optimized growth temperature of the low-temperature seed layer and the high-temperature capping layer was found at 400°C and 700°C, respectively, for minimizing the surface roughness, and as a result the topmost of Ge capping layer showed the threading dislocation density of 3.2×107cm-2. In order to investigate the Ge epitaxial layer in our RPCVD system we used a vertical incidence Ge p-i-n photodetector. Mesas with diameters varying from 20µm to 50µm were defined by reactive ion etch, followed by SiO2 passivation and metal contact. Without anti-reflection coating we had just used 0.35µm-thick SiO2 as passivation to reduce the dark current of the photodetector. The fabricated Ge photodetector exhibited the 3dB bandwidth of 27GHz, the responsivity of 0.42 A/W and low dark current of 42 nA at λ~1.55µm and the responsivity of 0.7 A/W at λ~1.31μm.The performance of our RPCVD-grown Ge photodetectors have proved good competitiveness for ultra-high speed applications for spectral ranges of 0.85~1.6µm. Also, RPCVD would be very attractive process for manufacturing of high-quality pure Ge layer due to the high productivity and the flexibility of process and maintenance.
9:00 PM - B8.9
Measurement of the Stress Distribution in Depth for Cu/Ta Double Layers using a GIXS Method.
Shinji Takayama 1 , Yasuhiro Asahi 1
1 Electrical and Electronics Engineering, Hosei University, Tokyo Japan
Show Abstract Residual internal stresses in multilayer films have become more important in reliability issues of electronic devices due to the narrower interconnect lines. To clarify this, we investigated the temperature dependent internal stress distribution in depth for the double layers of Cu(100nm)/Ta(100nm)/Si(100) substrate and SiO2(50nm)/Cu(100nm)/Ta(100nm)/S(100) substrate respectively by using a grazing incidence X-ray scattering (GIXS) method. This method allows us to measure an internal stress distribution with depth in each layer of double layers by choosing a suitable film’s combination with a different total reflection angle so as to be able to take place an incident X-ray scattering sequentially in each layer in depth. It was found that in the case of Cu/Ta double layers, the residual internal stresses in as-deposited Cu layer showed a tension and decreased almost linearly with increase in film depth from the free surface toward the Ta layer. On the other hand, the stresses in Ta layer showed a large compression (about 1.6Gpa) and tended to decrease with depth toward the Si substrate. In the case of double layers caped by a SiO2 film on the top surface of the Cu film, the stresses in the as-deposited Cu layers were a compression and an almost constant value within the whole thickness, most likely resulting from the restriction of the thermal expansion of SiO2 layer. However, the residual stresses in the Ta film for Cu/Ta with a SiO2 protective layer showed a compressive value (about 1.2GPa) slightly lower than that in the Ta layer of as-deposited Cu/Ta films. They tended to decrease in depth as well as those in the Ta layer in Cu/Ta films. It was realized that there was a large stress gap at the interface between Cu and Ta layers. The stress distribution profiles with depth in Cu layer of Cu/Ta films were largely changed on annealing at an elevated temperature, whereas those in Cu layer of SiO2/Cu/Ta films did not significantly change. On the other hand, the stress distribution profile in the Ta layer of Cu/Ta films did not change largely on annealing though their compressive values increased largely. However, the tendency of stress distribution with depth in the Ta layer of SiO2/Cu/Ta films become opposite to that in the as-deposited Ta layer on annealing at 400°C but their compress values increased with increase in annealing temperature. The change of these internal stress distributions on annealing was discussed in terms of the difference in thermal expansion coefficients between constituent films, and also between films and Si substrate.
Symposium Organizers
Osamu Ueda Kanazawa Institute of Technology
Mitsuo Fukuda Toyohashi University of Technology
Steve Pearton University of Florida
Edwin Piner Nitronex Corporation
Paolo Montangero Avago Technologies Italy S.R.L.
B9: Characterization Methods
Session Chairs
Wednesday AM, December 02, 2009
Liberty (Sheraton)
9:30 AM - **B9.1
Observation of Dopant Profile of Transistors Using Scanning Nonlinear Dielectric Microscopy.
Koichiro Honda 1 , Yasuo Cho 2
1 , Fujitsu Laboratories Ltd., Atsugi Japan, 2 , Tohoku University, Sendai Japan
Show AbstractWe have demonstrated that scanning nonlinear dielectric microscopy (SNDM) exhibited high performance and high resolution in observing the dopant concentration profile of transistors. In this study, we have measured standard Si samples, which are known to have one-dimensional dopant concentration values, calibrated by using conventional SIMS. Good quantitative agreement between the SNDM signals and dopant density values was obtained by SIMS. We succeeded in visualizing high-resolution dopant profiles in n- and p-type channel MOSFET with 40 nm gate channels. It is considered that SNDM would be an effective method in measuring the quantitative two-dimensional dopant profiles of semiconductor devices. Finally, we have observed the dopant depth profiles of SRAM memory cell by using SNDM, and succeeded in detecting the insufficient extension ion implantation in the PMOS transistor area.
10:00 AM - B9.2
Sub Surface Material Characterization using High Frequency Eddy Current Spectroscopy.
Henning Heuer 1 , Susanne Hillmann 1 , Mike Roellig 1
1 Institute for non destructive testing, Fraunhofer IZFP Dresden, Dresden Germany
Show AbstractBy launching new processes introduced by nano science into much more conventional industrial applications fast, robust and economical reasonable inspection methods are required for process control and quality assurance. Due to the complexity of processes e.g. for thin film coatings or nano engineered materials, variations of material parameters like microstructures, grain boundary conditions, particle or void density etc. can occur by a minor shift of process parameters. Coming from high tech industries e.g. semiconductor industries the methods available for thin film characterization and quality control are complex and often require scientific skilled personal. The established methods for non destructive testing of materials and structures are behind the postulated requirements of modern materials modified on the nanoscale. This paper presents first results obtained by a new developed high frequency eddy current spectrograph on thin film coatings and crystalline materials. The value of electrical conductivity contains a lot information about the chemical composition, microstructural and mechanical properties of the material. Not only the material itself also size and type of grain boundaries, strain gradients or precipitates will influence the electrical conductivity. By using e.g. four probe measurements, the characterization of sheet resistance is a well established method for characterization of thin films for semiconductor or nanotechnology applications. Sheet resistivity is used for homogeneity control, film thickness measurements or quality assurance in a broader understanding. By performing mappings, deviation, delaminations or cracks can be measured non destructively.The four point technique works in direct contact and the quality of the contact depends on several influences like pressure, surface oxides or impurities. Beside these effects, a lightly damage of the surface can cause negative effects for following production steps. By using electromagnetic induced eddy currents inside the specimen the characterization of electrical properties of thin films become possible without touching the surface. Depending on the required spatial resolution and penetration depth a wound or printed coil needs be moved in a defined distance to the surface.Available eddy current inspection systems are typically working in a range from several kHz up to 10 MHz. This paper presents a new developed device for eddy current measurements from 100 kHz for bulk material investigations up to 100 MHz for near surface characterization. Due to the simplicity of the required hardware, the developed device can be used for off line as same as inline inspection device. It is also possible to integrate the sensor into vacuum chambers for insitu process control (live thickness or composition monitoring).
10:15 AM - B9.3
Prevention of Si Wafer Breakage Via Non-destructive Evaluation of Plasma Arc Induced Damage.
Jennifer Stopford 1 , Ken Horan 1 , David Allen 1 , Andreas Danilewsky 3 , Muhammad Morshed 2 , Patrick McNally 1
1 Nanomaterials Processing Laboratory, Research Institute for Networks and Communications Engineering (RINCE), School of Electronic Engineering, DCU, Dublin Ireland, 3 , Kristallographisches Institut, -Herder-Strasse 5, 79104 Freiburg i. Br Germany, 2 National Centre for Plasma Science and Technology, Dublin City University, Dublin 9 Ireland
Show AbstractThe practicality of plasma etching, combined with low temperature and directional process capabilities make it an integral part of the IC manufacturing process. The primary cause of damage to wafers during plasma processing is charging damage. Process control and monitoring of plasmas is inherently difficult, and charging damage is most often caused by non-uniform plasmas, which cause local imbalances in the ion and electron currents leading to surface charging of the wafer. When sufficient charge builds up the plasma discharges and forms and arc. The resultant plasma arcing damage results in large pits and non-uniformities on the wafer surface. This plasma induced damage can lead to wafer breakage and high yield losses.Thus a non-destructive wafer damage metrology is crucial to the understanding of wafer failure mechanisms. We report on the successful use of a combined suite of non-destructive metrology techniques to locate the arc damage sites and examine the physical processes which have occurred as a result of the damage, leading to prediction of which type of damage can lead to wafer breakage. These consist of photo-acoustic microscopy (PAM), Synchotron X-Ray Topography (SXRT), Scanning Electron Microscopy (SEM) and micro-Raman Spectroscopy (μRS). Si wafers were placed in a low pressure RF plasma chamber and subjected to a number of damaging plasma processes including high voltage DC, AC sputter and etch arrangements. In most cases the plasma induced damage on the wafer surface (using SEM) appears as a series of 100 – 250 μm diameter islands of microcrystalline Si, each of which contains of a network of dislocations predominantly aligned in the <110> directions. SXRT shows the dislocation cells to measure approximately 20 μm wide, and extend up to 200 μm into the substrate. Micro-Raman spectroscopy results indicate crystalline silicon with some lattice disorder present. Strain measurements at the damage site show tensile strains as high as 1GPa in certain situations, with strain levels increasing from the surface towards the bottom of the cells. The use of PAM has enabled non-destructive imaging of the arc damage sites. PAM and SXRT results are in close correlation, proving the potential for PAM as a non-contact, in situ non-destructive metrology. A selection of damaged samples were also annealed in order to define onset thresholds for wafer breakage.
10:30 AM - B9.4
Minority Carrier Lifetime Measurement Based on Low Frequency Fluctuation.
Lin Ke 1 , Sha Huang 2 , Soo Jin Chua 1 , Szu Cheng Lai 1 , Bin Dolmanan Surani 1
1 , IMRE, Singapore Singapore, 2 , National Univerisity of Singapore, Singapore Singapore
Show AbstractWe present a novel, simple, and accurate approach based on low frequency voltage fluctuations to determine the averaged carrier lifetime in semiconductor materials and devices. This technique serves to address the limitations faced by existing techniques that use light as the excitation source for lifetime measurement. In this paper, the minority carrier lifetime is inferred from the 1/f low frequency noise profile exhibited by the device during low current operation. The current dependence of the power spectral density and its relation to minority carrier lifetime is modeled and derived directly. This technique is solely based on the electrical noise and no light source is required for excitation. The low frequency noise can be easily acquired without significant distortion via a signal analyzer as long as there is a sufficiently good Ohmic contact between the probe and the device under test, and that the device is sufficiently shielded from the influence of EMI. This technique has lower crosstalk, fewer fitting parameters, is low cost and allows the lifetime to be extracted directly from data collected at lower frequencies. These characteristics make our method useful in encapsulated devices, applicable on wafers and devices in production-lines and where light excitation is difficult to access the active material..
10:45 AM - B9.5
Mobility Variation in Electron Irradiated Gated and Ungated AlGaN/GaN High Electron Mobility Transistors.
Helen Jackson 1 2 3
1 Sensors, Air Force, BeaverCreek, Ohio, United States, 2 Physics , Wright State University, Dayton, Ohio, United States, 3 Semiconductor Research Center, Wright State University, Dayton, Ohio, United States
Show AbstractAbstract The Hall mobility, carrier concentration and conductivity of AlGaN/GaN High electron mobility transistors (HEMTS) were subjected to 0.8 and 0 .5 MeV electron irradiation at doses which were multiples of 5x1015 cm2 . These parameters were taken from temperature dependent Hall measurements using both gated and ungated van der Pauw configurations. Typical room temperature phonon scattering is observed. However, the overall mobility in the gated configurations increases after an initial decrease at some device specific threshold value of irradiation dose. The reasons for this are hypothesized to include (1) dynamic annealing due to the electron irradiation after initial defect creation and /or (2) parallel conduction in the layers. Simulations will be presented using physical models that include variations in irradiation energies, irradiation dose, and device layer structure, temperature dependent scattering cross sections for likely defects, and variations in the Fermi energy. These simulations should explain the nature of the observed variation of mobility with irradiation dose. Examples of how this mobility behavior can be used for both lifetime reliability studies, and for design of a fine tuned device layer geometry parameter set to optimize device performance in radiation intense environments will be presented.
B10: Strain Effects
Session Chairs
Koichiro Honda
Steve Pearton
Wednesday PM, December 02, 2009
Liberty (Sheraton)
11:30 AM - B10.1
Surfactant Effects of Indium on the Growth of AlN/GaN Distributed Bragg Reflectors via Metal Organic Vapor Phase Epitaxy.
L. Rodak 1 , C. Miller 1 , D. Korakakis 1 2
1 Lane Department of Computer Science and Electrical Engineering, West Virginia University, Morgantown, West Virginia, United States, 2 , National Energy Technology Laboratory, Morgantown, West Virginia, United States
Show AbstractDistributed Bragg Reflectors (DBRs) remain critical to the fabrication of various nitride based optoelectronic devices. In particular, DBRs are often employed for cavity formation in Resonant Cavity Light Emitting Diodes (RCLEDs) to enhance and obtain a more directional emission and also in Vertical Cavity Surface Emitting Lasers (VCSELs). As a result, epitaxially grown reflectors are attractive for direct integration in the device, reduced processing requirements, and the formation of narrow cavities [1,2]. In the III-Nitride material system, Aluminum Nitride (AlN) and Gallium Nitride (GaN) offer a large contrast in refractive index and are therefore well suited for fabricating DBRs with high reflectivity and wide bandwidths using relatively few periods. However, material cracking arising from to the 2.4% lattice mismatch and difference in thermal expansion coefficient decreases reflectivity and is detrimental to the efficiency of subsequent device fabrication. Several techniques, such as superlattice insertion layers [3] or the growth of AlxIn1-xN layers [1], have been employed to reduce strain and cracking in such structures. In this work, results of the use of indium as a surfactant in the Metal Organic Vapor Phase Epitaxy (MOVPE) of AlN/GaN DBRs will be discussed. Specifically, this study targets AlN/GaN DBRs with peak reflectivity at 460 nm. Indium has been used as a surfactant during growth by introducing trimethylindium into the system. It has been shown that crack formation is dependent on the flow of the indium precursor despite minimal indium incorporation into the lattice. Preliminary results further indicate the diffusion and formation of AlxGa1-xN between the layers is also dependent on the indium precursor flow. The impact of the interdiffusion on the cracking in AlN/GaN DBRs will be discussed. [1]R. Butte, E. Feltin, J. Dorsaz, G. Christmann, J. Carlin, N. Grandjean, M. Ilegems. Jpn. J. Appl. Phys. 44, 7207 (2005).[2] P. M. Pattison, A. David, R. Sharma, C. Weisbuch, S. DenBaars, and S. Nakamura. Appl. Phys. Lett. 90, 031111 (2007).[3] G. Huang, T. Lu, H. Yao, H. Kuo, S. Wang, C. Lin, L. Chang. Appl. Phys. Lett. 88, 061904 (2006).
11:45 AM - B10.2
Stress Limited Silicide Formation in Constrained Nano-dimensions.
Ahmet Ozcan 1 , Christian Lavoie 2 , Stefan Zollner 1 , Anthony Domenicucci 1 , Asa Frye 1 , Conal Murray 2 , Dong-ick Lee 1 , Valli Arunachalam 3 , Bin Yang 4 , Patrick Press 5 , Sadanand Deshpande 1
1 Microelectronics, IBM, Hopewell Junction, New York, United States, 2 Research, IBM, Yorktown Heights, New York, United States, 3 , GLOBALFOUNDRIES, Hopewell Junction, New York, United States, 4 , GLOBALFOUNDRIES, Yorktown Heights, New York, United States, 5 , GLOBALFOUNDRIES, Dresden Germany
Show AbstractSilicide thin films have been utilized in transistors as contacts to source, drain and gate areas. After the 250nm gate node, Ti silicide was replaced by Co silicide because of the difficult nucleation of C54-TiSi2 in narrow lines. Co silicide formation was also problematic in narrow poly-Si lines due to voids that caused high resistance or opens in the circuits. Most recently, Ni silicide has become the choice of the microelectronics industry, mainly because of low Si consumption, low formation temperature and compatibility with SiGe. The aggressive scaling continues to pose challenges for silicide formation in narrow lines for Ni as well. These challenges not only emerge because of small feature sizes but relate to the mechanical constraints exerted by sidewall spacers around poly-Si lines and field oxide around narrow Si lines. Our studies show that silicide phase formation can be hindered in constrained narrow lines and higher temperatures are needed to overcome the barrier imposed by stress. However, driving silicide formation in these nano-dimensions at higher formation temperatures causes severe interface roughness and extrusion of silicide stringers onto spacers and field oxide which are detrimental to yield. Detailed electron microscopy results of silicide morphology and defects in narrow lines will be discussed. A model will be presented to describe the stress-limited Ni silicide phase formation as a function of linewidth, sidewall spacer height and Ni thickness.
12:00 PM - B10.3
Resistive Switching Characteristics in Pt Nanoparticles-included TiO2 Films.
Yu-Lung Chung 1 , Jen-Sue Chen 1
1 Department of Materials Science and Engineering, National Cheng Kung University, Tainan Taiwan
Show AbstractIn this study, reproducible unipolar resistive switching characteristics are demonstrated in TiO2 films containing Pt nanoparticles. The Ti/Pt/Ti stacking layers are prepared by sputtering and followed by annealing in oxygen ambient to form Pt nanoparticles-included TiO2 films. With the Pt nanoparticles included TiO2 film as the active layer, the device exhibits electrical bistability in the I-V characteristics. CAFM (conducting atomic force microscope) images evidently reveal that the current leakage pass through not only the grain boundaries but also the crystal grains of TiO2 when applying negative bias, which supports the filament model of resistive switching. The resistance values of both high-resistance and low-resistance states are stable during the retention test for 600 seconds. The conduction mechanism of the Pt nanoparticles-included TiO2 films in low-resistance state is Ohmic, while the current characteristics in the high-resistance state are governed by the space charge limited current (SCLC) in low voltage region and by thermionic emission in high voltage region.
12:15 PM - B10.4
Stress-Induced Void Formation in Cu Interconnects Investigated by In-Situ TEM and D-STEM.
K. Ganesh 1 , P. Ferreira 1 2
1 Materials Science and Engineering, University of Texas at Austin, Austin, Texas, United States, 2 Mechanical Engineering, University of Texas at Austin, Austin, Texas, United States
Show AbstractStress-induced void formation (SIV) in 120 nm and 90nm Cu interconnect lines has been investigated. The lines were subjected to thermal cycling and in-situ TEM observations were carried out to monitor void formation. A novel parallel electron diffraction technique called D-STEM has been developed using a JEOL 2010F TEM/STEM instrument to determine local texture information in the Cu lines from grains as small as 3 nm, with enhanced levels of control and precision when compared to existing nano-diffraction techniques. The technique has been optimized to enable automated acquisition of spot diffraction patterns on a pixel by pixel basis using Gatan STEM-Diffraction- Imaging software. Variations in crystal orientations across multiple grains in the line structures have been evaluated using the Automated Crystallography for TEM (ACT) software. FEM simulations have also been performed to correlate grain orientations with local stresses and consequent void formations.
12:30 PM - B10.5
Enhanced Adhesion Strength of Cu/TaN Interface by Alloying Cu.
Junya Inoue 1 , Shuichiro Ikemoto 1 , Shoichi Nambu 1 , Toshihiko Koseki 1
1 Department of Materials Engineering, The University of Tokyo, Tokyo Japan
Show AbstractAdhesion strength of Cu alloy film on TaN substrate was measured with the four point bending test, and the microstructural feature of Cu alloy/TaN interfaces were investigated by XPS, XRD, and TEM. It was demonstrated that an atomically thin reaction layer is formed at the Cu/TaN interface by doping Cu film with Al, which results in enhancing the adhesion strength of the interface by a factor of almost three.Three different alloying elements, Ag, Al, and Sn, which are reported to enhance electromigration (EM) resistance, were studied. Cu alloy films with different alloy concentrations varied from 0 to 10% were deposited on top of a TaN/SiO2/Si substrate by EB-PVD method. The Cu alloy films were annealed after the deposition at the temperature of 673K in a vacuum of 1.0x10-5Pa for 30min. In order to diminish the effect of plastic deformation of Cu films on the measured adhesion strength, film thickness were reduced until a convergence of the measured adhesion strength was achieved. Film thickness less than 100nm is required to reach the convergence for all the Cu alloy films. Al was found to increase the adhesion strength almost three times as strong as that measured for pure Cu films at the highest, while Ag and Sn degrade the adhesion strength. For Ag and Al, the adhesion strength depends on the concentration of alloying elements, while virtually independent in the case of Sn. Sn segregates at the Cu/TaN interface from the very low concentration and the amount remains almost constant, while amount of Ag segregation at the interface is proportional to the alloy concentration. An atomically thin reaction layer was formed at the interface in the case of Al. The thickness of the reaction layer is almost unchanged by the alloy concentration, while the amount of unreacted Al atoms segregated at the interface increases proportionally to the alloy concentration. The maximum adhesion strength measured for Cu-Al alloy is found to be equivalent to that measured for pure Al/TaN interface after the same heat treatment. The result indicates that the enhanced adhesion strength is due to the formation of bonding between the reaction layer and segregated Al atoms at the Cu/TaN interface. The formation of atomically thin reaction layer is attractive, because the improvement of the EM resistance is achieved at a minimum expense of line cross-sectional area.
12:45 PM - B10.6
Ultra-thin Ni-based Silicides Formed from Metal-Si Intermixed Layer.
Zhen Zhang 1 , Yu Zhu 1 , Steve Rossnagel 1 , Bin Yang 2 , Simon Gaudet 3 , Andrew Kellock 4 , Ahmet Ozcan 5 , Conal Murray 1 , Patrick Desjardins 3 , Jean Jordan-Sweet 1 , Christian Lavoie 1
1 T. J. Watson research center, IBM , Yorktown Heights, New York, United States, 2 T.J. Watson Research Center, GLOBALFOUNDRIES, Yorktown Heights, New York, United States, 3 , Ecole Polytechnique de Montréal, Montréal, Quebec, Canada, 4 Almaden Research Center, IBM , San Jose, California, United States, 5 Microelectronics, IBM , Hopewell Junction, New York, United States
Show AbstractIn CMOS transistor scaling, reduction of junction depth limits Si availability for silicide formation which requires the use of ultra-thin silicides that are sufficiently robust to usual thermal treatments. The emergence of 3D transistors such as trigates FETS, finFETs or nanowire FETs also requires that ultrathin silicide films not only be uniform in thickness but also show suitable conformality. In this work, we use a robust process for forming very uniform ultra-thin Ni-Pt silicides from a metal-Si intermixed layer [1]. When metals are sputtered onto a Si surface, a thin intermixed zone is produced at the interface. This layer contains sufficient Si so that wet etching selectively removes the metal without attacking the intermixed layer. Subsequent annealing leads to extremely thin, uniform and textured Ni based silicides. We study the formation of intermixed silicides from different thickness of pure Ni and NiPt. Film composition and thickness are measured by RBS and XTEM. Phase formation and film degradation are concurrently monitored in situ with XRD, resistance and roughness measurements. Silicide phases are unambiguously identified using XRD pole figure measurements in which film texture is fully determined. Not surprisingly, the silicide films are found to be very uniform, extremely textured with very large grains. Film properties and implications for microelectronics will be discussed. This work was performed by the Research Alliance Teams at various IBM Research and Development Facilities [1] R. A. Donaton, S. Jin, H. Bender, M. Zagrebnov, K. Baert, K. Maex, A. Vantomme, G. Langouche,‘New Approaches for Formation of Ultra-Thin PtSi Layers for Infrared Applications’, MRS (1998)
B11: Defects/Growth
Session Chairs
Wednesday PM, December 02, 2009
Liberty (Sheraton)
2:30 PM - B11.1
InGaAs/GaAsSb Heterostructures: Aluminum-Free Intersubband Devices.
Michele Nobile 1 , Aaron Andrews 1 , Hermann Detz 1 , Elvis Mujagic 1 , Pavel Klang 1 , Werner Schrenk 1 , Gottfried Strasser 1 2
1 Solid State Electronics, TU Wien, Vienna Austria, 2 EE and Physics, University at Buffalo, Buffalo, New York, United States
Show AbstractIntersubband (ISB) devices, like quantum well infrared photodetectors (QWIPs) [1] are currently used in several applications, such as chemical sensing [2] and thermal imaging [3]. Quantum cascade lasers (QCLs) [4] have been demonstrated in various material systems [5, 6, 7] and continuous progress has been reported for over a decade.However, all the QCL materials which have been realized so far contain a significant fraction of aluminum in the barriers resulting in a high effective mass of the electrons. A small effective electron mass in the barrier material would be advantageous for the optical gain as well as further device processing, where surfaces with high aluminum concentrations are hindering complex re-growth and etching scenarios.A promising candidate for Al-free III-V heterostructures is the InGaAs/GaAsSb material system, lattice-matched to InP. To prove its viability, we intend to present a detailed study on the room temperature ISB absorption in such lattice-matched InGaAs/GaAsSb multiple quantum wells (MQWs). Samples with well widths from 4.5 nm to 12 nm have been investigated by Fourier-transform infrared spectroscopy resulting in strong ISB absorption peaks in a broad wavelength range (5.8 - 11.6 μm). From these data we are able to compute a conduction band offset at the InGaAs/GaAsSb interface of 360 meV, in excellent agreement with the values previously reported in literature for this material system [8].In addition, we will show the first ISB devices based on this InGaAs/GaAsSb system: a QWIP operating at a wavelength of ~ 5.5 μm as well as the first aluminum-free QCL emitting at a wavelength of ~ 11.3 μm. This laser has a threshold current density of 1.7 kA/cm2 and a maximum optical output power of 20 mW at 78K. References:[1] H. Schneider and H. C. Liu, Quantum Well Infrared Photodetectors: Physics and Applications (Springer, Berlin, 2007)[2] A. A. Kosterev and F. K. Tittel, IEEE J. Quantum Electron. 38, 582 – 591, 2002.[3] E. Dupont, S. R. Laframboise, J. Lapointe, R. Dudek, A. Bezinger, J. Fraser and H. C. Liu, Semicond. Sci. Technol. 23, 055006 (2008) [4] J. Faist, F. Capasso, D. L. Sivco, C. Sirtori, A. L. Hutchinson and A. Y. Cho, Science 264, 553 (1994)[5] A. Tahraoui, A. Matlis, S. Slivken, J. Diaz and M. Razeghi, Appl. Phys. Lett. 78, 416 (2001)[6] C. Pflügl, W. Schrenk, S. Anders, G. Strasser, C. Becker, C. Sirtori, Y. Bonetti and A. Muller, Appl. Phys. Lett. 83, 4698 (2003)[7] J. Devenson, O. Cathabard, R. Teissier and A. N. Baranov, Appl. Phys. Lett. 91, 251102 (2007)[8] J. Hu, X. Xu, J. Stotz, S. Watkins, A. Curzon, M. Thewalt,