Tecnología de transistores de microondas basados en Nitruro de Galio (GaN) para aplicaciones Radar

El siguiente artículo estudia las características de los principales materiales utilizados en la actualidad para la fabricación de transistores de microondas como son Silicio (Si), Arseniuro de Galio (GaAs), Carburo de Silicio (SiC) y Nitruro de Galio (GaN) y describe como condicionan la operación del transistor cuando se requieren potencias de salida altas, del orden de cientos y miles de vatios, habitualmente las necesarias en aplicaciones Radar.Se mostrará como los transistores de microondas fabricados con GaN son adecuados para aplicaciones de alta potencia debido a las superiores propiedades físicas y químicas de estos semiconductores. Si además añadimos las modernas técnicas de polarización de alta eficiencia, los transistores fabricados con la tecnología de Nitruro de Galio se perfilan como los candidatos idóneos para ser utilizados en los transmisores de sistemas Radar.
La gran mayoría de los transmisores Radar requieren dispositivos activos que puedan generar una potencia de salida de RF del orden de kilovatios e incluso de megavatios. Habitualmente se utilizan para estas aplicaciones dispositivos basados en tubos de ondas progresivas. Sin embargo, estos dispositivos son voluminosos, caros y pueden tener problemas de fiabilidad. Aunque los amplificadores basados en semiconductores tienen a priori más eficiencia, han estado hasta ahora limitados por el voltaje que se podía aplicar al dispositivo debido al crítico campo de ruptura inherente a estos materiales, lo que hace que se requiera una corriente muy alta y también un mayor tamaño. Trabajar con una corriente de operación alta disminuye la eficiencia debido a las pérdidas y al hecho de que los dispositivos de gran tamaño presentan una alta capacitancia y muy baja impedancia limitando así la frecuencia de operación y el ancho de banda. La tecnología de GaN es ahora capaz de ofrecer una solución a este problema.Los amplificadores de estado sólido están ya reemplazando a los de tubos de ondas progresivas (TWTA, Traveling Wave Tube Amplifiers) en algunas aplicaciones de microondas de alta potencia. Sin embargo, las bajas tensiones de operación hacen que el circuito asociado sea muy grande lo que implica un dispositivo más complejo a la vez que reduce el yield de producción y la fiabilidad. Las tecnologías de semiconductores de banda prohibida ancha (WBG, Wide Band Gap) como el GaN pueden alcanzar densidades de potencia cinco veces mayores que las de los transistores convencionales de GaAs tanto de efecto de campo como bipolares de heterounión. La ventaja final es la reducción de la complejidad del circuito, mayor ganancia y eficiencia, y también una mayor fiabilidad. En particular, los sistemas Radar se beneficiarán del desarrollo de esta tecnología.

El GaN es el futuro

El desarrollo de semiconductores de banda prohibida ancha, tales como el GaN o aleaciones basadas en GaN, ofrece la posibilidad de fabricar dispositivos activos de RF, especialmente transistores de potencia HEMT (High Electron Mobility Transistor), con una potencia de salida significativamente mayor. Esta mejora en la potencia de salida de RF se debe a las especiales propiedades de este material, de entre otras destacan: alto campo de ruptura, elevado valor de saturación de la EDV (velocidad de Drift de los electrones) y cuando se utilizan sustratos de SiC, mayor conductividad térmica. Los datos mostrados en la Tabla permiten comparar los materiales Si, GaAs, SiC y GaN. La mayor conductividad térmica del SiC y del GaN reduce el aumento de temperatura de la unión debido al autocalentamiento. El campo de ruptura de cinco a seis veces mayor del SiC y del GaN da ventaja a estos materiales frente al Si y el GaAs para dispositivos de potencia de RF. El SiC es un material de banda prohibida ancha (3.2eV) pero tiene una movilidad de electrones baja, lo cual dificulta su uso en amplificadores de alta frecuencia. El SiC está también limitado porque las obleas de este material son caras, pequeñas y de baja calidad.Aunque la movilidad de los portadores es significativamente mejor en los dispositivos de GaAs, la alta velocidad de pico y de saturación de la EDV de los HEMT de GaN compensa su relativa menor movilidad permitiendo su utilización a altas frecuencias. Estas ventajas del GaN sumadas a la alta linealidad y al bajo ruido de las arquitecturas HEMT abren las puertas a estos dispositivos para su utilización en la fabricación de amplificadores Radar de alta potencia.Una ventaja adicional de los HEMT de GaN radica en el gran offset de energía entre la banda de conducción del GaN y la capa barrera de AlGaN. Esto permite un aumento significativo de la densidad de portadores en el canal en los HEMT basados en GaN con respecto a otros materiales (hasta 1013cm-2 y más). Si sumamos la posibilidad de utilizar un mayor voltaje conseguimos un aumento en la densidad de potencia. La densidad de potencia es un parámetro muy importante para los dispositivos de alta potencia ya que cuanto mayor es menor es el tamaño del dado y más sencillas son adaptaciones de entrada y salida. En la Figura 1 se muestra el rápido progreso de la densidad de potencia de RF frente al tiempo para un FET (Field-Effect Transistor) de GaN en Banda X.Los altos voltajes de operación y las altas densidades de potencia que se alcanzan con los dispositivos de RF de banda prohibida ancha ofrecen muchas ventajas en el diseño, fabricación y montaje de amplificadores de potencia en comparación con las tecnologías de LDMOS (Lateral Double-Difusse MOS) de Silicio o la de MESFET (Metal Epitaxial Semicon-ductor Field Effect Transistor) de GaAs. La tecnología HEMT de GaN ofrece una alta potencia por ancho de canal unitario, lo cual se traduce en dispositivos más económicos y de menor tamaño para la misma potencia de salida, esto no sólo hace que sean más fáciles de fabricar sino que aumenta la impedancia de los dispositivos. El alto voltaje de operación que se consigue con la tecnología de GaN elimina la necesidad de convertidores de tensión y por consiguiente reduce también el coste final del sistema.

El camino está claro

Muestra una gráfica de la potencia de salida frente a la frecuencia para los dispositivos de estado sólido y tubos de microondas que constituyen el actual estado del arte. Históricamente, lo amplificadores de tubo, tales como los controlados por rejillas, magnetrones, kystrones, tubos de onda progresiva y amplificadores de campos cruzados (CFA, Cross Field Amplifier) han sido usados como amplificadores de potencia en los transmisores Radar. Estos amplificadores generan alta potencia pero habitualmente trabajan con ciclos de trabajo (duty cicle) bajos. Los amplificadores de Klystron ofrecen mayor potencia que los magnetrones a frecuencias de microondas y también permiten el uso de formas de onda más complejas. Los tubos de onda progresiva son similares a los klystrones pero con mayores anchos de banda. Los CFA se caracterizan por tener grandes anchos de banda, poca ganancia y ser compactos. Los amplificadores de potencia de estado sólido (SSPA, Solid State Power Amplifier) soportan pulsos largos y formas de onda con altos ciclos de actividad. A pesar de que los elementos utilizados en los SSPA tienen individualmente poca amplificación de potencia pueden combinarse para conseguirla. Los transistores bipolares de Silicio, los MESFET de Arseniuro de Galio y los PHEMT (Pseudomorphic HEMT) de Arseniuro de Galio son algunos de los elementos utilizados en los SSPA. Los HEMT de GaN pueden ser combinados para crear un SSPA con una potencia media de salida mayor y por consiguiente un mayor rango de detección del Radar.Como se puede ver en la Figura 2, los transistores de estado sólido producen niveles de potencia de RF menores de 200 vatios en Banda S y su salida va decreciendo a medida que aumentamos la frecuencia. La potencia de salida de RF de los FETs de GaAs se acerca a los 50 vatios en banda S y a aproximadamente a 1 vatio en banda Ka1. Los FETs de GaAs tienen una la potencia de salida limitada principalmente por la baja tensión de ruptura del drenador1. Los dispositivos semiconductores fabricados con materiales de mayor banda prohibida, tales como el GaN, ofrecen unas prestaciones significativamente mejores.Con el paso del tiempo han ido apareciendo diferentes figuras de mérito que permiten evaluar los distintos semiconductores con potencial para ser utilizados en aplicaciones que requieren alta potencia a altas frecuencias de trabajo. Mediante estas figuras de mérito se pretende aunar las propiedades más relevantes de los materiales en un valor cualitativo. Así la figura de mérito de Johnson (JFOM = ECR vsat/p) tiene en cuenta el campo de ruptura ECR y la saturación de la EDV Vsat. Como puede verse en la Figura 3 [3], la figura de mérito de Johnson para el GaN es por lo menos 15 veces la del GaAs.Aethercomm cree que si la tendencia de crecimiento del GaN se mantiene al ritmo actual, el comportamiento previsto para los HEMT de GAN en el año 2010 será el representado en la Figura 4. El GaN pronto superará a todos sus competidores.

La eficiencia es la clave

Los sistemas Radar más modernos utilizados en aplicaciones militares demandan nuevos requerimientos para los amplificadores de potencia de RF debido a la necesidad de reducir el tamaño, peso y coste. Los mayores cambios en las especificaciones se centran cada vez más en mejorar la eficiencia del amplificador para reducir los requerimientos de potencia DC y mejorar la fiabilidad del sistema a través de una menor disipación de potencia del componente. Los dispositivos de microondas basados en tecnologías de banda prohibida ancha y alta eficiencia permitirán además aumentar las prestaciones del sistema.La capacitancia parásita y el alto voltaje de ruptura de los HEMT de GaN les hace ideales para funcionar en modos de amplificación de alta eficiencia clase E y clase F. Ambos modos tienen una eficiencia teórica del 100 %. Recientemente, algunos fabricantes de transistores de GaN han implementado amplificadores híbridos de clase E. Resultados típicos obtenidos son 10 vatios de potencia de salida en banda L con eficiencias comprendidas entre el 80% y 90%.Aethercomm ha entregado recientemente un módulo amplificador de clase F para Banda L. La potencia de salida deseada debía superar los 50 vatios con una eficiencia del 60% para todo el amplificador. Debido a los plazos tan ajustados del programa fue necesario utilizar transistores estándar encapsulados en lugar de desarrollar una solución híbrida a medida.La etapa final del amplificador de potencia se implementó utilizando un par balanceado de HEMT encapsulados de GaN trabajando en clase F. Las redes de adaptación incluyendo las terminaciones armónicas necesarias para la operación en clase F fueron diseñadas considerando inicialmente un modelo ideal del transistor. A continuación se introdujeron las inductancias y las capacitancias parásitas del encapsulado del transistor y se modificaron las redes de adaptación para mantener las terminaciones armónicas requeridas a nivel del transistor en dado. Posteriormente se simuló el amplificador utilizando un modelo no lineal del transistor y se modificaron las redes de adaptación para optimizar eficiencia y potencia.Se construyó un prototipo en configuración single-ended para la etapa de salida de clase F. Se obtuvo una eficiencia de drenador del 75%, una potencia de salida de 40 vatios y una ganancia de 16 dB con un ajuste mínimo. Los resultados fueron muy similares a los obtenidos en la simulación. No había disponibles dispositivos de GaN de baja potencia adecuados para la etapa de driver, se diseñó uno de tres etapas utilizando MESFET de GaAs que trabajaban en clase A. Inicialmente se creía que las etapas del driver deberían haber trabajado en un modo de alta eficiencia para así alcanzar la PAE (Power Added Efficiency) requerida; sin embargo, los análisis indicaron que con un dimensionado adecuado de los transistores la operación en clase A era permisible. El driver tuvo una ganancia de 40 dB y un consumo de potencia de 10 vatios.La configuración final del amplificador de potencia tuvo una PAE de pico del 63% y una potencia de salida de 75 vatios. El amplificador tenía una potencia de salida de 65 vatios y un 61% de PAE a P2dB. La Tabla 2 muestra las características del amplificador para distintos valores de potencia de salida. Debido a que la etapa final de clase F está polarizada en el umbral, sin corriente de drenador, el amplificador ofrece un amplio rango de funcionamiento para potencias bajas. La ganancia del amplificador alcanza un pico y después comienza a comprimirse cuando se alcanza la máxima potencia de salida. La Tabla 2 muestra la eficiencia de este diseño para distintas potencias de salida.Aethercomm también ha desarrollado un dispositivo HEMT de GaN de 200 vatios sobre sustrato de SiC diseñado para maximizar la PAE y mantener una alta potencia de salida para una frecuencia de operación de 1215 MHz a 1390 MHz. Se observaron eficiencias mayores del 56% mientras se mantenía niveles de potencia de salida en exceso de 205 vatios de P3dB.Muchos SSPA para aplicaciones Radar son diseñados con dispositivos semiconductores de RF configurados para trabajar en clase C. Esta forma de polarización proporciona una operación muy eficiente para una etapa de un único transistor, sin embargo, el transistor de clase C tiene una ganancia tan baja, típicamente 6 dB, que la ventaja ganada en la eficiencia se pierde al necesitarse muchas etapas adicionales de ganancia para alcanzar la potencia deseada de salida.

Conclusión

Los futuros sistemas Radar tales como los basados Radar de phase-array activo requerirán de forma creciente SSPA cada vez más eficientes y pequeños. El deseo de lograr barridos extremadamente rápidos, rangos de detección mayores, la posibilidad de localizar y seguir un gran número de objetivos, una baja probabilidad de ser interceptado y la posibilidad de funcionar como un inhibidor requerirán una tecnología de transistores innovadora y rentable. Recientes desarrollos en el campo de los HEMT de GaN han hecho posible diseñar amplificadores de una gran eficiencia a frecuencias de microondas. Los dispositivos HEMT de GaN proporcionan una alta corriente de pico con una baja capacitancia de salida así como un voltaje de ruptura y una densidad de potencia extremadamente alta. Esta combinación única de características permite a los diseñadores conseguir amplificadores con unas prestaciones en conjunto muy superiores a las logradas con dispositivos basados en las tecnologías alternativas existentes en la actualidad.

Nitruro de galio (GaN) Radares

Los radares son usados para fines militares, así como sectores civiles para muchos propósitos diferentes. Algunos de sus usos incluyen la observación meteorológica, control de tráfico aéreo, e imágenes de alta resolución junto con varias aplicaciones de radares militares, tales como la penetración de suelo, tierra y aire o la vigilancia, seguimiento de objetivos, y control de incendios.

Con el fin de satisfacer las crecientes demandas de los ambientales, así como las condiciones técnicas en que estos sistemas deben funcionar, la industria el sistema de radar sigue haciendo avances. Muchas de estas áreas de ascenso en los sistemas de radar incluyen el movimiento hacia activos conjuntos de lectura óptica, aumento de la sensibilidad, la mejora de imagen, mayor eficiencia energética y aumento de potencia, amplificador de tecnologías de legado y más.
Es en este medio que el nitruro de galio (GaN) es reconocida como la tecnología de amplificación clave para los sistemas de radar. Amplificadores basados en GaN, recientemente se han desplegado en los sistemas de radar para aplicaciones militares. La actividad de diseño y la tasa de adopción va en aumento cada año, más de GaN RF basado en los proveedores de componentes en el mercado con productos de producción listo. De ancho tecnología de GaN gap tiene claras ventajas de mayor tensión y el rendimiento general del ancho de banda combinado con la eficiencia de alto consumo. GaN ofrece importantes, reconocidos ventajas sobre las tecnologías existentes para aplicaciones que funcionan en bandas de frecuencia.
En resumen, los radares de nitruro de galio es una contribución perfecta para servir y hace que el valor de los recursos militares para el control y observar de acuerdo a sus necesidades.

Bárbara Scarlett Betancourt Morales

CAF

A CRITICAL COMPARISON BETWEEN MOVPE AND MBE GROWTH OF
III-V NITRIDE SEMICONDUCTOR MATERIALS FOR OPTO-ELECTRONIC
DEVICE APPLICATIONS

ABSTRACT
A systematic study of the growth and doping of GaN, AlGaN, and InGaN by both molecular beam epitaxy (MBE) and metal-organic vapor phase epitaxy (MOVPE) has been performed. Critical differences between the resulting epitaxy are observed in the p-type doping using magnesium as the acceptor species. MBE growth, using rf-plasma sources to generate the active nitrogen species for growth, has been used for III-Nitride compounds doped either n-type with silicon or p-type with magnesium. Blue and violet light emitting diode (LED) test structures were fabricated. These vertical devices required a relatively high forward current and exhibited high leakage currents. This behavior was attributed to parallel shorting mechanisms along the dislocations in MBE grown layers. For comparison, similar devices were fabricated using a single wafer vertical flow MOVPE reactor and ammonia as the active nitrogen species. MOVPE grown blue LEDs exhibited excellent forward device characteristics and a high reverse breakdown voltage. We feel that the excess hydrogen, which is present on the GaN surface due to the dissociation of ammonia in MOVPE, acts to passivate the dislocations and eliminate parallel shorting for vertical device structures. These findings support the widespread acceptance of MOVPE, rather than MBE, as the epitaxial growth technique of choice for III-V nitride materials used in vertical transport bipolar devices for optoelectronic applications.

INTRODUCTION
The recent development of III-V Nitride semiconductor devices for optoelectronic applications has been driven by improvements in the epitaxial growth of these semiconductor materials. Heterostructures have been fabricated across a range of AlN-GaN-InN compositions with bandgaps ranging from 6.2 eV (ultraviolet) to 1.9 eV (red) for LED, laser diode, and photodetector applications [1,2]. Heterostructure epitaxy has traditionally been performed using
either MBE or MOVPE in many semiconductor material systems; however, most of the recent device application demonstrations for III-V nitrides have used MOVPE, particularly in the commercially driven work at Nichia Chemical and Cree Researc. MBE growth for optoelectronic device applications has lagged behind. Initially, this was attributed to the unavailability of an appropriate source of active nitrogen species for MBE. Through the development of nitrogen rf plasma sources for MBE, the quality of the resulting epitaxial layers has improved [7,8,9,10]. Despite these advances, demonstration of high quality vertical devices such as laser diodes or high brightness LEDs grown by MBE has not occurred. In this work, we compare the growth of III-V nitride materials by MBE and MOVPE in order to examine the fundamental differences in the epitaxial growth and the influence on resulting devices. We have studied three areas of critical importance for light emitting devices. First is the difference in the epilayer growth morphology; second is the doping of GaN with magnesium for p-type conductivity; and finally, the deposition of InGaN quantum wells with compositions in the visible emission range. This comparison provides a twofold benefit of identifying critical areas for further exploration in crystal growth and deepening the understanding of the underlying physical processes at work in successful epitaxial deposition.

EXPERIMENTAL PROCEDURE
MOVPE growth was performed in a vertical flow rotating wafer (up to 2000 rpm) system designed and built at NCSU. A radiatively heated substrate mount, of original high reliability design, can achieve temperatures up to 1200°C, as measured by an optical pyrometer. 50-mm diameter sapphire wafers were used as the base substrate with a typical low temperature GaN nucleation layer. Trimethylgallium (TMGa), trimethylaluminum (TMAl), trimethylindium (TMI) and ammonia were used as precursors with nitrogen and hydrogen carrier gases at a reactor pressure of 76 Torr. Silane and bis(cyclopentadienyl) magnesium were used as dopant sources. Growth temperatures for GaN ranged from 1060°C to 1130°C. The conditions resulted in 2D epitaxial growth at rates of 1-2 mm/hr. InGaN growth was conducted in a manner similar to Yoshimoto at temperatures from 725°C to 800°C. MBE growth was performed in an EPI Model 930 system using elemental group III and dopant sources. Rf plasma sources were used to generate the active nitrogen species. Prenucleated GaN/SiC substrates were used for the MBE deposition. Growth temperatures ranged from 750°C to 900°C for GaN and 670°C to 700°C for InGaN resulting in growth rates of 0.4-2 mm/hr. A modulated beam technique was used to grow InGaN as previously described.
The MOVPE and MBE were connected as a multichamber UHV cluster tool. This allows for the growth of sophisticated heterostructures with specific layers grown in either the MBE or MOVPE system where applicable. Characterization of epitaxial layers included: scanning electron microscopy using a JEOL JSM6400 SEM, photoluminescence (PL) using a 12 mW He-Cd laser source, and Nomarski microscopy using an Olympus BX60 microscope and image capture system. Vertical cross section samples were studied in a Topcon 002B Transmission Electron Microscope (TEM) with g=(1100) at 200 kV. LED samples were prepared following standard lithography techniques and using Ni/Au and Ti/Al as p-type and n-type contact metals, respectively.

RESULTS AND DISCUSSION
Epitaxial Layer Surface Morphology and Magnesium Doping The surface morphology of epitaxially grown GaN exhibits an obvious difference between MOVPE and MBE deposited material. As shown in the SEM micrograph in Figure 1a, undoped or n-type doped MBE grown GaN exhibits a “wormy” structure. This surface structure has been previously reported and the degree of texture can be minimized, although not eliminated, through changes in the nitrogen plasma source operating conditions. The MOCVD grown undoped material is smooth and uniform as shown in Figure 1b. Magnesium was used as a p-type dopant for both MBE and MOVPE grown of GaN. For MOVPE growth, the surface of p-type material is smooth and featureless. However, in MBE growth, there is a dramatic change in surface texture with the evolution of a faceted surface with increasing magnesium flux as shown in Figures 1c and 1d. Cross sectional TEM studies revealed the facet morphology to be related to the pre-existing dislocation structure.

Bulk Growth of GaN by HVPE

Abstract

Bulk like GaN material (~3mm) was grown on the free standing (FS) GaN layers by hydride vapor phase epitaxial (HVPE). FS-GaN layers were obtained as a result of high thermal stress built up between sapphire substrate and the GaN layer during the cooling down step, which resulted in spontaneous lift off of the GaN layers. Bulk like GaN grown on FS-GaN exhibited good structural and optical quality and the dislocation density was in range of 105-106 cm-2.

INTRODUCTION
Bulk GaN substrates are still a big issue despite the recent fast progress of this technology. Hydride vapor phase epitaxy (HVPE) has proven to be an effective technique to manufacture free standing (FS) GaN substrates. The approach used by employing HVPE, is to grow thick GaN layers and successively remove the substrate by the void assisted method, laser lift off, or facet controlled epitaxial overgrowth. However, getting crack-free large size FS-GaN using laser lift-off technique is not easy because of fracturing of GaN during the laser irradiation. On the other hand FS-GaN of high quality has been realized successfully by the void assisted method and by facet controlled epitaxial overgrowth. However, these methods require complicated processing of substrate prior to growth process and are time consuming. In this presentation we report our results on bulk like GaN material grown on FS-GaN substrates which were obtained as a result of spontaneous lift off. The spontaneous lift off and the properties of heteroepitaxial and FS-GaN and bulk like GaN material are discussed and presented.

EXPERIMENTAL
In this study, crack-free layers of thickness between 200- 350μm were grown initially directly on the sapphire substrate by an optimized process. All the layers were grown in a horizontal HVPE reactor. The description of the reactor is given elsewhere. Afterwards, these layers were used for overgrowth. However, due to high built up of stress between sapphire substrate and the GaN layer, FS-GaN was obtained in considerably big pieces, i.e. (20x15)mm. The lift off process was completely spontaneous and involved no prior processing of the substrate or special step during the deposition process. The thickness of FS-GaN layers was between 400 to 650μm. These thick FS layers were used to grow bulk like GaN by HVPE as well (~3mm thick). In order to understand the spontaneous lift off phenomena, and to study the properties of heteroepitaxial, FS-GaN and bulk like GaN, samples were characterized and studied by different techniques. Differential interference contrast optical microscopy (DIC-OM) and scanning electron microscopy (SEM) was employed to study the morphology, etch pit density and the cross-sections of the samples. Optical quality was determined by the contactless electro-reflectance (CER)- and photoluminescence (PL) spectroscopy. The structural quality was checked by recording the symmetrical (002) and anti-symmetrical (105) rocking curves (RC) by high resolution X-ray diffraction (HR-XRD) and the stress in the layer was probed by the micro Raman (μ-Raman)
spectroscopy.

RESULTS AND DISCUSSION
To find the origin of spontaneous lift off, cross-sectional SEM, CER- and PL spectroscopy were performed on asgrown heteroepitaxial layers (300μm). The results revealed that the lift off is from the GaN instead of from the sapphire substrate. These non processed Ga-face FS-GaN layers were employed for the growth of bulk like GaN of thickness ranging between 1mm to 3mm by HVPE. The morphology of this bulk like GaN is dominated by the polygonal shaped pits. Recently Lucznik et al [5] has reported in similar work, the presence of such pits in bulk like GaN and that these pits always recur themselves in successive growth process even if
they were first polished away. In our case we also overgrow on some polished samples and indeed these pits re-appeared in the grown material.

In order to count the defect density, the samples were etched in a eutectic solution of NaOH and KOH [6]. The defect density in these layers was on average of the order of 106 cm-2 and in some cases even 105cm-2. The structural properties were studied by recording symmetric (002) and non-symmetric (105) scans employing HR-XRD. Non polished samples showed very wide and non symmetric peaks; however the full width half maximum (FWHM) values decreased when the same samples were measured after polishing.

ACKNOWLEDGEMENTS
The research work was funded by the Dutch technology Foundation (STW) under project number NAF6937.

References
[1] M.K. Kelly, R.P. Vaudo, V.M. Phanse, J. Gorgens, O.
Ambacher, M. Stutzmann, Jpn.J.Appl. Phys., 38, (1999)
L217.

Bárbara Scarlett Betancourt Morales

CAF

Bulk Aluminum Nitride Crystal Growth

Introduction
Light emitting diodes (LEDs) are ubiquitous in modern society: they are in traffic lights, automobile interiors, backlights in cell phones, and many other applications. Their growing popularity comes from their many advantages over incandescent and fluorescent lamps including a high energy efficiency, long lifetimes, compact size, and shock resistance. Furthermore, they can emit light of a precise color, which is useful for many applications. Currently, commercial LEDs are available that emit light over the entire visible range - from red to blue, plus infrared light. What is lacking are LEDs that emit ultraviolet light. Efficient UVLEDs would have many applications including water purification, sterilization of heat sensitive materials such as plastics, biochemical sensors, and polymer curing. To date, making UV-LEDs has proven difficult. One of the main problems is the poor quality of suitable materials available for device fabrication. To address this problem we are studying the growth of large aluminum nitride (AlN) single crystals. These are being developed as substrates (a structural template) for aluminum gallium nitride (AlxGa1-xN) layers, the base semiconductor for fabricating UV-LEDs and laser diodes (LDs). Aluminum nitride has several physical and chemical properties that make it a superior material as a substrate: a better match of the crystal structure and dimensions (lattice constants); high electrical resistivity; ultraviolet light transparency; high thermal conductivity; and good chemical compatibility with AlxGa1-xN. As a consequence of these properties, AlN substrates should reduce the defect density in AlxGa1-xN layers, thereby increasing device performance (reducing turn-on voltage and increasing power output) and efficiency (power output per power input).

Crystal Growth Process
Because the melting temperature for AlN exceeds 2750 °C at a nitrogen pressure of 100 atmosphere, it is impractical to grow crystals from the melt, as is usually done with the more established semiconductors silicon and gallium arsenide. It can not be grown by precipitation from solution either, as the solubility of nitrogen in most solvents is low (less than 10-3 at% in liquid aluminum at 1000 °C, for example). Chemical vapor deposition may be a suitable route to producing bulk AlN single crystals, but to date, growth rates are low and the feasibility of this technique has not been verified. The most successful method of producing large, bulk AlN single crystals - and the method employed at KSU- is the sublimation-recondensation method. In this process, the overall reaction AlN(s) = Al(v) + ½N2 is run in the forward (sublimation or decomposition) direction at the source, and in the reverse (recondensation) direction in the crystal growth zone. The source is maintained at a higher temperature than the crystal growth zone, causing the direction of the reaction to change with position in the crucible.

While the process is simple to describe, successful implementation is challenging. Extreme temperatures (greater than 2000 °C) are required for a significant aluminum partial pressure and to increase the reactivity of N2 (normally an inert gas), so reasonable growth rates (>100 μm/h) are attained. Aluminum has a tremendous affinity for oxygen, so all oxygen must be eliminatedfrom the crystal growth zone. Consequently, specially designed high temperature furnaces are required. A photograph of the tungsten heating element furnace used for some of our crystal growth studies is shown in Figure 2.

Crucible Materials
At such extreme temperatures, there are relatively few materials thermally and chemically stable for fabricating the crucible and other furnace fixtures. If elements from the crucible volatilize, they can incorporate into the AlN crystals, and degrade its properties. Even in small concentrations, such impurities can increase the crystal’s defect density, change the crystal habit, or cause the AlN to absorb visible wavelengths, changing the color of the crystals. Examples of these effects are shown in Figures 3-5. Identifying how the predominate AlN crystal habits change with the crucible material is one aspect of research at KSU. In addition, the fabrication and performance (durability and failure modes) of different materials crucibles themselves is being studied. Methods of converting refractory metals to nitrides and carbide have been developed. Thermodynamic calculations are employed to determine what reactions specific materials might undergo that would impact their crucible durability. The vapor pressures of the elements from the crucible are predicted at the crystal growth temperature by a free energy minimization. This is helpful to understanding the stability of the crucible material, and its propensity to incorporate elements into the crystal. Currently, the most suitable materials for AlN crystal growth appear to be tungsten and tantalum carbide, as they are inert, stable and durable at the crystal growth conditions. Yet these materials are not perfect, as their coefficients of thermal expansion are much higher than for AlN, putting the crystal under stress as it is cooled from the growth temperature. We are continuing to examine new crucible materials and are developing better crucible designs.

Crystal Growth Procedures
The best method of obtaining a single crystal is to initiate the growth with a single crystal seed. A seed crystal can be used to control the crystal orientation and polarity. Unfortunately, only small AlN crystals are available, limiting the ultimate size of the single crystal. Instead, at KSU single crystals of 6H-silicon carbide are studied as potential seed crystals for initiating AlN crystal growth. 6H-silicon carbide wafers up to 75 mm in diameter are commercially available, which may make it possible to produce large AlN crystals. Its thermal stability is reasonably good at AlN crystal growth temperatures.

.
Characterization of the Crystal Quality
The application of AlN single crystals as a substrate requires that it enables better quality epitaxial growth of AlxGa1-xN thin films than is possible on currently favored substrates such as sapphire and silicon carbide. The bulk AlN crystals must have a high degree of crystal perfection (low density of defects), good optical transparency down to wavelengths of 200 nm, and a smooth surface finish suitable for supporting high quality epitaxial thin films. Thus the characterization of the quality and purity of the AlN crystals is an ongoing aspect of research at KSU. The defect types and densities are studied by x-ray diffraction and defect selective etching at KSU. The former method is a general assessment of the crystal quality over relatively large areas. With the later technique, the exact location of the defects can be determined.

Bárbara Scarlett Betancourt Morales

CAF

based on the lecture of Prof. Gabriel Ferro
Université Claude Bernard Lyon, France

Epitaxial growth of thin films
Abstract
The term ‘epitaxial’ is applied to a film grown on top of the crystalline substrate in ordered fashion that atomic arrangement of the film accepts crystallographic structure of the substrate. Epitaxial growth is one of the most important techniques to fabricate various ‘state of the art’ electronic and optical devices.

Modern devices require very sophisticated structure, which are composed of thin layers with various compositions. Quality, performance and lifetime of these devices are determined by the purity, structural perfection and homogeneity of the epitaxial layers. Epitaxial crystal growth resulting in epitaxial layer perfection, surface flatness and interface abruptness depend on number of factors like: the epitaxial layer growth method, the interfacial energy between substrate and epitaxial film, as well as the growth parameters – thermodynamic driving force, substrate and layer misfit, substrate misorientation, growth temperature, etc…

Recently epitaxial growth is also used for fabrication of semiconductor quantum structures like quantum dots giving highly perfect structures with high density. In this report the aspect determining the epitaxial growth mode, epitaxial layer growth techniques and additional focusing on SiC epitaxial growth is discussed.

1. Epitaxial growth modes
The occurrence of the epitaxial growth modes depends on various parameters of which the most important are the thermodynamic driving force and the misfit between substrate and layer. The growth mode characterizes the nucleation and growth process. There is a direct correspondence between the growth mode and the film morphology, which gives the structural properties such as perfection, flatness and interface abruptness of the layers. It is determined by the kinetics of the transport and diffusion processes on the surface. Different atomistic processes may occur on the surface during film growth: deposition, diffusion on terraces, nucleation on islands, nucleation on second-layer island, diffusion to a lower terrace, attachment to an island, diffusion along a step edge, detachment from an island, diffusion of dimmer (see Figure 1).

Experimentally, the distinction between three classical growth modes is well known: Frank-van der Merwe (FV), Volmer-Weber (VW) and Stranski-Krastonov (SK). In addition to the three well-known epitaxial growth modes mentioned above there are four distinct growth modes: step flow mode, columnar growth, step bunching, screw-island growth.

Volmer-Weber (VW) growth mode
A VW growth mode consists in first phase of large number of surface nuclei and in
second phase of their spreading. Thus VW growth often results in a high mosaicity of the material inside the layer. Usually continues growth of the layer, after initial VW growth, occurs by columnar growth, but in the case of 3C-SiC on Si a VW growth mode results in growth of layers that are not columnar using right conditions. Stranski-Krastonov (SK) growth mode SK mode is considered as intermediate between the FV and VW growth modes, and it is caused by significant lattice misfit from film and substrate. The lattice mismatch between the substrate and the film creates a build-in strain as a consequence of the increasing elastic energy with increasing layer thickness. The first deposited layer is atomically smooth (FV growth mode), compressively strained layer up to a certain thickness called critical thickness. When the deposition time is enough exceeding the critical thickness – phase transition to islands rapidly takes place (VW growth mode), because the nonuniform strain field can reduce the strain energy by an island array, compared with a uniform flat film, resulting in the SK growth mechanism. Step flow growth mode Step flow mode is clearly distinct from layer-by layer growth in FV mode. Unidirectional step flow is induced by substrate missorientation (off cut angle). This trick is often used to avoid island formation, their coalescence and following columnar growth in epitaxy from the vapor phase. Step bunching growth mode Step bunching is observed when a high density of steps moves whit large step velocities over the growth surface. By fluctuations, higher steps catch up with lower steps and then move together as double, triple…. Or in general as macro steps that can exceed thickness of thousands of monosteps. The microsteps cause different incorporation rates of impurities and dopands due to locally varying growth rate Spiral-island growth mode Coalescence of larger number of initial growth islands may lead to screw dislocations due to the layer structure resulting in spiral-island growth mode.

2. Control of growth modes
There are two main types of epitaxy – homoepitaxy and heteroepitaxy. Homoepitaxy is when the same material (or polytype) as the substrate is grown for example: Si on Si, 4H-SiC on 4H-SiC. Heteroepitaxy is when a different material (or polytype) from the substrate is grown for example: GaN on sapphire, 3C-SiC on 6H-SiC. In heteroepitaxy the lattice mismatch between substrate and film and the supersaturation, plays a key role on growth mode and this is demonstrated. This layer-by-layer growth mode FV requires the zero misfit as indicated Large lattice misfit normally induces VW mode except for large interface energies between substrate and film, which will cause SK mode. If structural perfect layer are required, either homoepitaxy or substrate with zero misfit are needed. On another hand the misorientation of the substrate provides steps on surface depending on the angle and direction of misorientation. The density of the steps can be made so high and interstep distance so small that VW or SK modes can be suppressed. The layers growth in the step flow mode have relatively high crystal perfection because defects due to coalescence are prevented.

3. SiC Polytype control
The best eptaxial layer qualities are achieved by using homoepitaxy, because of the compatibility of grown material with substrate. However not all materials substrates are commercially available. For example the only commercially available SiC substrates are of the 4H and 6H polytypes. So if one wants to grow 4H or 6H-SiC, there are no big problems to do homoepitaxy, however to grow 3CSiC heteroepitaxy has to be applied.
Using heteroepitaxy one has to care about lattice misfit, temperature expansion
coefficient and etc. - not all substrates can be used.
It is known that SiC exists in different polytypes, there are more then 200 of them. Most stable are 4H-, 6H- and 3C-SiC (Figure 4). When growing homoepitaxy or heteroepitaxy one should care about polytype inclusions, which are very common. One could think about temperature dependence for polytype stability, it is clear, that two or more polytypes can grow at the same temperature, so other polytype control methods has to be applied. For homoepitaxy of hexagonal polytypes, one solution is to increase surface mobility of the adatoms by higher temperature, but usually this is not preferable, because of technical problems. The most popular thing in SiC homoepitaxy is to cut substrate off-axis to create steps.

Double positioning boundaries (DBPs) of 3C-SiC
DBPs is a special defect for 3C-SiC growth on hexagonal substrate which comes from the two possible orientations of the cubic 3C-SiC axis on the hexagonal α- SiC basis [8]. Because 3C-SiC has two different orders in stacking sequence either ABCABC… or ACBACB… Normally in on-axis substrate, these two stacking sequences can be formed on the substrate in alter positions (see Figure 7). When the nucleated domains expand via 2D-nucleation mechanism, these two domains cannot blend together according to a different stacking order. The boundaries of these two domains are so-called “Double Positioning Boundaries”. DBPs can be observed as 60 degrees angle difference of triangular defects according to the crystal structure. These defects will limit the expanding of domains and the size of single crystal growth.

4. TECHNIQUES FOR EPITAXY
The techniques of epitaxy can be classified according to the phase (till ex: liquid
(solution), or vapor) of material use to form the epitaxial layer. Growth techniques: liquid phase epitaxy (LPE), physical vapor deposition (PVD) and molecular beam epitxay (MBE).

4.1 Molecular beam epitaxy (MBE)
Molecular beam epitaxy is a technique for epitaxial growth via the interaction of one or several molecular or atomic beams that occurs on a surface of a heated crystalline substrate. In Figure 8 scheme of a typical MBE system is shown. The substrate, on witch the heterostructure to be grown, is placed on a sample holder which is heated to the necessary temperature and, when needed, continuously rotated to improve the growth homogeneity. The growth in the MBE requires ultra high-vacuum (UHV), typically 10-6 – 10-4 mbar during growth. After outgasing under such a high vacuum, O2, CO2, H2O, and N2 contamination on the growing surface can be neglected. The typical growth conditions make possible to reduce the rate down to nm/sec, so that precise control of the growth thickness is possible – this is a great advantage.

Growth of SiC by MBE
Although growth with MBE has a lot of advantages, like: growth of atomically abrupt interfaces, heteropolytype engineering (for example 4H/3C/4H heterostructure) and in-situ characterization, for growth of SiC it is almost not used. Firstly because of high costs. Secondly because of source material availability, Si source is not a problem, but C is not so easy, graphite can be used, but high temperature cells are needed, C60 doesn’t require high temperature cells, but material is expensive. And for usage of gaseous precursors a lot of technical problems arises like: keeping ultra high vacuum and high temperatures are needed (>1200 0C), which is hard to implement in MBE. Also grown layers have high background doping levels and polytype inclusion is a serious.

4. 2 Liquid phase epitaxy (LPE)
The process known as LPE is a technique for the deposition of the epitaxial layers
from supersatured solution. The chosen solvent has generally low melting point and low vapor pressure. LPE method is mainly used for the growth of compound semiconductors. Very thin, uniform and high quality layers can be produced. Typical example of LPE method is given by the growth of III-V compounds. In this case, the process can be described as follows: a melt of pure gallium exposed to a GaAs wafer will dissolve some of the solid to produce a dilute solution of group V element. Cooling this solution to induce a slight supersaturation, and bringing a substrate into the contact with the melt surface, will result in the growth of a layer of GaAs all over the substrate surface. At conditions that are close to the equilibrium, deposition of the semiconductor crystal on the substrate is slowly and uniform. The equilibrium conditions depend very much on the temperature and on the concentration of the dissolved semiconductor in the melt. The thickness of the epitaxial layer is controlled by the contact time between substrate and solution, the cooling rate, rate of diffusion of the slowest component elements etc… The major advantage of the LPE is that the growth temperature can be well below the melting point of the compound semiconductor which is being decomposed. Furthermore, equipment is simple and inexpensive, also non-hazardous. Key problem in the production of the epilayer is that the composition of relatively small volumes of each melt will rapidly change as crystal growth proceeds. LPE is to simple to grow more complicated nanostructures, because of the difficult thickness and composition control, etc… Figure 9 shows sketches of three kinds of LPE growth process: a tipping arrangement, slightly more complicated sliding substrate holder and sandwich arrangement.

LPE of SiC
SiC does not form a stoichiometric liquid phase at normal conditions. The Si-C phase diagram is shown in Figure 10. Instead the material decomposes to vapor at 2830 °C. Silicon carbide can be grown from the liquid phase by using a nonstoichiometric melt. The natural choice as base for the solution would be Si since this is a constituent of SiC and high-purity Si is commercially available. The growth rates using silicon as a solvent are not high since the solubility of C in Si is very low at temperatures less than 2000 °C. By introducing a transition metal to the silicon melt the solubility of carbon is increased. An example is given by using Si-Sc melts for which the liquid phase epitaxy of SiC has successfully demonstrated good influence on the growth rate and on the structural properties (crystallinity and surface morphology) of the SiC epitaxial layers. The tipping and sliding arrangement, as shown in figure 9, are not used for SiC epitaxy by LPE because of the high reactivity of the melt with the crucible. Most of the time, a dipping or sandwich arrangement is preferred.

4.3 Vapor-liquid-solid (VLS) method
The vapor-liquid-solid (VLS) method has recently been re-examined to produce one dimensional structures (whiskers) for nano-physics technology or other applications [12]. The VLS mechanism has been also developed for growth of SiC epitaxial layers. Some of the basic mechanisms involved in the VLS method are similar to LPE. In case of LPE, carbon is supplied by the graphite container, a solid SiC source in direct contact with the solution or initial dissolution of the substrate while in the VLS method the carbon is provided through the reaction of a carbon containing gas phase with silicon containing liquid phase. The difference to “conventional” LPE growth conditions is that VLS growth may be performed even at a negative temperature gradient, i.e. the temperature is higher at the substrate than in the liquid or the top of the solution, and the requirements on the temperature gradient are not as strict.

Barbara Scarlett Betancourt Morales

CAF