Synthesis and characterization of nanocrystalline UO2 ceramics

Synthesis and characterization of nanocrystalline UO2ceramics DOCTORAL THESIS Dissertation by M.Eng. Raquel Jovani-Abril1,2 Directors: Prof.Dr. Arturo López Quintela1 Dr. José Luis Spino2 1Universidad de Santiago de Compostela (USC), Spain 2Institute for Transuranium Elements (ITU), Germany Santiago de Compostela (Spain), 2014 Memoria de tesis presentada por Raquel Jovani-Abril para la obtención del título de Doctor por la Universidad de Santiago de Compostela dentro del Programa de Doctorado en Ciencia de los Materiales. Raquel Jovani-Abril i ii D. Arturo López Quintela, Profesor Doctor Catedrático del Departamento de Química-Física de la Universidad de Santiago de Compostela, y D. José Luis Spino, Investigador Doctor Senior del Institute for Transuranium Elements Joint Research Centre de la Comisión Europea, informan: Que la presente memoria, titulada “Synthesis and characterization of nanocrystalline UO2ceramics” (“Síntesis y caracterización de cerámicas nanocristalinas de UO2”), que para optar al título de Doctor por la Universidad de Santiago de Compostela dentro del Programa de Doctorado en Ciencia de los Materiales presenta Dª. Raquel Jovani-Abril, ha sido realizada en el Institute for Transuranium Elements Joint Research Centre de la Comisión Europea en colaboración con el Departamento de Química Física de la Universidad de Santiago de Compostela bajo nuestra dirección. Considerando que constituye trabajo de Tesis, autorizan su presentación en la Comisión de Tercer Ciclo de la Universidad de Santiago de Compostela. Y para que así conste, firmamos el presente informe: Prof.Dr. Arturo López Quintela Dr. José Luis Spino iii iv Al fin y al cabo somos lo que hacemos para cambiar lo que somos. Eduardo Galeano A Daniel i Jordi v vi Acknowledgements The work presented in this study has been carried out in the framework of a European thesis to obtain the degree of Doctor from the University of Santiago de Compostela (USC) and sponsored by the European Commission. This thesis has been enabled by the collaboration and direction of Prof.Dr. Arturo López Quintela from the Department of Physical-Chemistry of the USC, and Dr. José Luis Spino from the Nuclear Fuels Department of the Institute for Transuranium Elements (ITU). I wish to express my gratitude to both of them to take the challenge of the supervision of the thesis and the collaboration in the distance. I would like to thank the Director of ITU, Prof.Dr. Thomas Fanghänel, for offering me the opportunity to make the research in this renowned institute. I want to greatly acknowledge Dr. Daniel Baron, Dr. Joaquin Cobos Sabaté, Prof.Dr. Joan de Pablo Ribas, Prof.Dr. Ian Farnan, Prof.Dr. Haas Didier, Dr. Ralph Hania, Dr. Rikard Malmbeck, Prof.Dr. Francisco Rivadulla, Dr. Vicenzo V. Rondinella, Dr. Joseph Somers, Prof.Dr. Carlos Vázquez Vázquez and Dr. Marcus Walter, who accepted without complications to be part of the possible elected members of the jury/revisers for my thesis defence. I want now to special thank all the colleagues who made that experience possible. I do not use here your title, but your name. Not because I do not want to treat you with the respect you all deserve, but because your title does not say anything about you as a person. It is your name, which represents for me the patience, the love, the time and the laughter we enjoyed together. First, thank to the Nuclear Fuels Department because, although I was moving around any corner of the institute, this was the unit I was belonging to. I really enjoyed the four years with you. Thanks to Marc Couland, Herwin Hein and Serge Fourcaudot. You showed me how to move inside this special institute in my first time in ITU as a trainee and later on in the PhD. You opened doors for me which would remain longer closed for a new student. Thanks also for slowly trusting me making your baby-experiments. For finishing the preparation of my samples for an urgent project when I needed to fly to Spain without expecting it from one day to the other. Thanks for the office-conversations. Thanks to Michael Holzhäuser, Co Boshoven, Mairead Murray Farthing, John Mcginley, Sarah Stohr, Patrick Lajarge, Sebastien Gardeur, Antony Guiot, Emmanuel Vermorel, Andrea Cambriani, Alexandre Dockendorf and Annette Küst, who worked in one or other step of the pellet performance. Thank to put your specific experience in this work. vii Esta transformación comienza en el borde de la pastilla del combustible y de manera constante progresa hacia el interior mientras se sucede la irradiación [Matzke and Spino, 1997][Spino and Papaioannou, 2000]. La microestructura original de microgranos (o granos-grandes) se transforma en una matriz nc-porosa [Nogita and Une, 1994] a través de la reestructuración de los defectos de irradiación acumulados. Se trata de un tipo de acción de “auto-curación”, donde el material se cura del daño sufrido reordenándose a sí mismo [Spino et al., 2012]. La nueva nc-estructura que aparece entonces, recibe el nombre de estructura de alto grado de combustión (HBS; high burn-up structure), también llamada estructura de borde (rim-structure) porque en los combustibles de UO2ésta se inicia en el borde o en las zonas exteriores de las pastillas (región que recibe mayor cantidad de fisiones). En un principio se pensó que la HBS era la causa de los fallos que observados en el revestimiento de las varillas debido a aumento adicional del volumen (hinchazón) en pila y un supuesto comportamiento frágil del material así transformado [Matzke, 1992]. Además se creyó que esta estructura que aparece en el borde de la pastilla podría actuar como una nueva fuente de liberación de gas. De hecho todavía hay líneas de investigación que defienden esta opinión. Sin embargo, otros estudios han demostrado que el porcentaje de gas liberado desde el borde de las pastillas de combustible, donde aparece la HBS, es bajo en comparación con el gas que viene de las partes internas de la pastilla donde la estructura original todavía está presente [Mogensen et al., 1999]. A esta segunda filosofía le han seguido otras publicaciones que demuestran un evolución en general favorable de las propiedades del combustible con la aparición de esta nueva estructura a altos valores de combustión, en particular el aumento en la retención de los gases de fisión. Asímismo, las propiedades de la HBS indican una mayor tolerancia a la radiación [Spino et al., 2012]. Este estudio coincide con una evidencia importante e innovadora en la literatura que demuestra que los materiales de granos nano resisten más el daño por radiación que los correspondientes materiales de grano micro, debido a la recombinación de defectos en los múltiples límites de grano [Nita et al., 2005]. Se anticipan también otras tendencias en el material como la mejora de la conductividad térmica y otros efectos de la radiación en las propiedades del material debido a la liberación de estrés en el parámetro de red después de la recristalización [Ronchi et al., 2004], así como el aumento de la tenacidad a la fractura y curación de grietas [Spino et al., 2003]. En lo que concierne a la seguridad, los últimos experimentos realizados en combustibles sometidos a alto grado de combustión en reactores nucleares de agua ligera (LWR; light water reactor), no indicaron aumento en la liberación de gas, así como tampoco en la susceptibilidad de fallo durante accidentes de reactividad iniciados (RIA; reactivity initiated accident) transitorios [Sasajima et al., 2010][Fuketa et al., 2006]. También ha sido observada una disminución de la velocidad de corrosión acuosa bajo condiciones típicas de depósito geológico simuladas en combustibles con presencia de HBS [Ekeroth et al., 2009][Carbol et al., 2009]. Ambos hechos confirmaron la estanqueidad de esta estructura. Así pues, se ha demostrado que la HBS tiene cualidades excepcionales incluso en comparación con la matriz original (estructura de grano-grande), con una mejora de propiedades que serían muy ventajosas para un combustible. Entonces, ¿por qué no imitar este material recristalizado?. ¿Por qué no imitar esta estructura (HBS) que xiv aparece en el borde del combustible a altos grados de combustión y se introduce en esta forma como un combustible nuevo en el interior del reactor?. ¿Por qué no sintetizar una matriz de combustible mimetizando la HBS que debido a su aparente resiliencia al daño resistiría tiempos más largos bajo irradiación?. Aquí es cuando empieza el desarrollo de este proyecto con el objetivo de sintetizar polvo nanocristalino (nc)-UO2para la fabricación de pastillas (monolitos) de nc-combustible. Las piezas monolíticas fabricadas a partir de este polvo, tendrían un volumen de aproximadamente 1 cm3y una matriz uniforme de granos de un tamaño entre 100 y 250 nm imitando la estructura que aparece a altos grados de combustión (HBS; high burn-up structure). Durante este trabajo se consiguió la creación de esta nueva microestructura de combustible pasando por distintas fases. Desde la síntesis del nc-material, a la fabricación de la pastilla de combustible, distintas etapas de este proceso, previamente desconocido o inexplorado, tuvieron que ser especialmente desarrolladas y/u optimizadas. Síntesis de nc-UO2y nc-ThO2 Un trabajo considerable fue dedicado al desarrollo del polvo inicial para la producción de los monolitos o pastillas de nc-UO2imitando la HBS que aparece en los combustibles de los LWR. Dos vías de síntesis química diferentes fueron estudiadas para obtener precipitados defloculados de nc-UO2y nc-ThO2como compensación a la falta de disponibilidad comercial. El ThO2tiene una estructura similar al UO2pero tiene, a su vez como ventaja, una sola valencia (IV). Para conseguir obtener mayores cantidades de nc-UO2que las publicadas en literatura, y proporcionar así material suficiente para la fabricación de los monolitos, ambos métodos fueron convenientemente ajustados, desarrollados y escalados de acuerdo a las necesidades. El material así producido fue objeto de estudio mediante microscopio electrónico de transmisión (TEM) y difracción de rayos X (XRD). El primer método desarrollado fue una precipitación controlada que utiliza una disolución acuosa electrolíticamente reducida de nitrato de uranilo como precursor y una solución goteada de NaOH como agente de alcalinización para desencadenar la precipitación del nc-material lo más próximamente posible a la línea de solubilidad del UIV . Este método fue originalmente descrito por [Rousseau et al., 2002], [Rousseau et al., 2006]. Un estudio intensivo de los rango de concentración de U y acidez en los que se produce la precipitación de nc-UO2+xfue llevado a cabo. Se utilizaron para ello disoluciones de nitrato de uranilo electrolíticamente reducidas, usando U-concentraciones más elevadas (10−1M) que las observadas en literatura (10−2M) [Rousseau et al., 2006], y por tanto intervalos de pH de trabajo más bajos, siempre siguiendo la línea de solubilidad del UIV . Como resultado se obtuvieron hasta 10 g de nc-UO2+xpor experimento, en lugar de los pocos nanogramos publicados hasta ahora en literatura. La fase sólida así obtenida y estudiada bajo XRD, cristalizó bajo la típica estructura de fluorita UO2-fcc (grupo espacial Fm-3m), con un parámetro de red a=0.5417(1) nm y un tamaño de cristal promedio de 3.79 nm, también en concordancia con el tamaño medio observado por TEM de 3.9(8) nm. El difractograma predominante de las muestras correspondía inequívocamente a UO2pero en un estado ligeramente oxidado. Esto último se manifestó a través de una contracción del parámetro de red de aproximadamente 0.9%de la fase precipitada (a=0.5417(1) nm) con respecto a los xv valores típicos de UO2estequiométrico (a=0.547 nm). El segundo método estudiado fue una descomposición térmica en fase orgánica usando acetilacetonato de uranilo (UAA) como precursor. Éste se añade a una mezcla de ácido oleico y oleilamina calentándose a continuación hasta temperaturas por encima de los 300°C para inducir la precipitación de las nanopartículas de UO2 por descomposición térmica del UAA. Este procedimiento fue descrito originalmente por [Wu et al., 2006] y fue aquí modificado para reducir la cantidad de agentes tensoactivos respecto al porcentaje de metal utilizado. AsÌ mismo, se escalaron los 0.1gdenc-UO 2por experimento obtenidos según lo publicado por [Wu et al., 2006], a2.3gdenc-UO 2por experimento en el presente trabajo. El mismo método fué extrapolado para la síntesis de nc-ThO2, utilizando acetilacetonato de torio (ThAA) o acetato de torio (ThA) como precursores, obteniéndose finalmente nano-varillas de ThO2. La razón por la cual se obtuvieron precipitados en forma de varilla, en vez de la forma esférica obtenida para nc-UO2, es todavía desconocida en este estudio. El rendimiento por experimento (0.3 g nano-varillas de ThO2) fue más bajo que en la síntesis de nc-UO2por el mismo método. Tanto en la síntesis de nc-UO2como en la de ThO2bajo este método, se probaron diferentes condiciones de trabajo: velocidad de calentamiento, tiempo de envejecimiento de la disolución final, temperatura de envejecimiento, así como distintos precursores iniciales (ThAA y ThA) en la síntesis de nc-ThO2. No obstante, se encontraron siempre resultados similares en términos de estructura y geometría (esféricas para nc-UO2y en forma de varilla para nc-ThO2)de los precipitados. En el estudio bajo el XRD se determinó una fase sólida perfectamente cristalizada con la típica estructura de fluorita UO2-fcc (grupo espacial Fm-3m), un tamaño de cristal promedio (diámetro de la esfera) de 5.52 nm y un parámetro de red de 0.5431(0) nm, también en concordancia con el tamaño promedio observado con el TEM de (4.9(3) nm) y por dispersión dinámica de luz (DLS) de (3.7(1) nm). En las nano-varillas de ThO2, se encontró una estructura típica de fluorita (grupo espacial Fm-3m)ThO 2-fcc, con un tamaño de cristalito (diámetro varilla) de 1.42 nm y un parámetro de red de 0.5579(1) nm. No se observó agregación de partículas en las imágenes de TEM del material sintetizado por ambos métodos. Cristalización y crecimiento de grano en f(T) del nc- UO2 La composición de los precipitados obtenidos por ambos métodos arriba mencionados y su propensión a la expansión térmica en el estado no consolidado, se estudiaron en el material así sinterizado y recocido a diferentes temperaturas. Para ello se utilizaron técnicas de análisis térmico y de difracción de rayos X, como el análisis termogravimétrico y el análisis térmico diferencial (TGA/DTA), difracción de rayos X (XRD) y difracción de rayos X a alta temperatura (HT-XRD), técnicas espectroscópicas tales como la espectroscopía de absorción de rayos X (XAS), espectroscopía de resonancia magnética nuclear (MAS-NMR), espectroscopía infrarroja (IR), y técnicas de caracterización como la microscopía electrónica de transmisión (TEM). La evolución del tamaño de los cristales, el parámetro de red y la tensión de red se determinaron para el material así sintetizado y a distintas temperaturas de recocido del material (bajo atmósfera inerte) hasta 1200°C. Para el caso del material nc-UO2 xvi precipitado en fase acuosa se observó tan sólo un pequeño aumento en el tamaño de los cristales, permaneciendo éstos por debajo de los 7 nm hasta llegar a la temperatura de recocido de 700°C. A partir de esta temperatura, el tamaño de los cristales creció acusadamente y de manera constante con la temperatura, hasta alcanzar un valor de 73 nm a 1200°C. Por el contrario, el incremento mayor del parámetro de red se observó en el intervalo más bajo de temperatura 20°C-700°C. En el intervalo 700°C-1200°C sólo se observó un pequeño aumento en el parámetro de red coincidiendo con la expansión térmica reversible típica del UO2. Hay que añadir que en la mediciones del parámetro de red a temperatura ambiente y después del tratamiento a temperatura bajo atmósfera estática de He, se observó una recuperación de la estructura típica cristalina del UO2, pasando del valor de parámetro de red inicial de 0.5417 nm para el nc- UO2así sintetizado, a un valor de 0.5473 nm después de la exposición a 1200°C. La contracción del parámetro de red inicialmente medida para el material nc-UO2así sintetizado por debajo del valor normal típico para UO2(a=0.5470 nm) se atribuyó principalmente a oxidación. En el caso del nc-UO2precipitado por el método en fase orgánica, se observó un comportamiento similar en el crecimiento de las partículas con la temperatura, sin apenas cambio en el tamaño de cristal hasta una temperatura de 700°C (debido al pre-tratamiento a temperatura aplicado), seguido de un crecimiento intenso del tamaño de cristal hasta obtener un tamaño de 150 nm a 1100°C. Se determinó una oxidación inicial de las partículas sintetizadas bajo el método en fase orgánica derivada de los valores de parámetro de red, aunque menos acusada que en las partículas sintetizadas en fase acuosa. Las partículas recobraron el valor típico de parámetro de red para UO2 a temperaturas de recocido por encima de los 750°C. En cuanto al parámetro de estrés de red, un comportamiento similar fue también observado para las partículas obtenidas por ambos métodos. La observación principal en ambos casos fue la disminución continua del parámetro de estrés con la temperatura, hasta extinguirse prácticamente a la temperatura en que los cristales empezaron a crecer. Esto confirmaba que la presencia del parámetro de estrés de red actuó en ambos casos como inhibidor del crecimiento cristalino. Estudio de la estructura y estequiometría del oxígeno mediante XRD, XANES, EXAFS, NMR AND FTIR La estructura del material nc-UO2producido en función de la temperatura pero en este caso bajo atmósfera reductora, se estudió también bajo el XRD y se comparó con el material de referencia UIV O2-grano-grande (granos de tamaño micrométrico). Se midió el parámetro de red del material tras enfriarse después de alcanzar diferentes temperaturas máximas (600°C y 1200°C). Esto permitió la separación de la contribución de la expansión térmica en los valores medidos a temperatura para obtener curvas más limpias de expansión térmica frente a temperatura, y parámetro de red frente a tamaño de cristal. No se encontraron grandes diferencias en el tamaño de cristal, parámetro de red y tensión, entre las mediciones realizadas bajo atmósfera inerte (arriba comentado) y bajo atmósfera reductora (medición realizada tras el enfriamiento) para el material obtenido por el método acuoso. Sin embargo. se observó un cambio notable en el tamaño de los cristales para el material obtenido por el método orgánico al llegar a xvii temperaturas de recocido de 1100°C bajo atmósfera de He con una talla de cristal de 150 nm, y un tamaño de tan sólo de 12 nm bajo atmósfera de Ar/5%H2. Comparando el material nc-UO2obtenido por ambos métodos, acuoso y orgánico, bajo atmósfera reductora (Ar/5%H2) y sin ningún otro tratamiento térmico previo del polvo, no se detectaron grandes cambios hasta la temperatura de recocido de 600°C. Pero sí se observaron a la temperatura de 1200°C, obteniendo un tamaño de 82 nm para el material del método acuoso frente a 12 nm para el material del método orgánico. Esto podría ser atribuido a la capa orgánica protectora presente para estas últimas nanopartículas. Junto al estudio de XRD se llevó a cabo un estudio de XANES para determinar el estado de oxidación de los cationes de U, las fracciones molares correspondientes y la relación de O/U derivada. Los espectros de XANES en el borde U-L3mostraron tendencias similares para el nc-UO2sintetizado por ambos métodos (acuoso y orgánico): reducción de la estequiometría (x)delUO 2+x, al aumentar la temperatura de tratamiento. Se estudiaron muestras de nc-UO2así sintetizado y después del tratamiento térmico a 600°C y 1200°C en atmósfera de Ar/5%H2y se compararon con el material de referencia UIV O2(grano-grande), detectándose con la temperatura un ligero desplazamiento del pico de la WL (white line) hacia energías más bajas, así como un aumento de la intensidad y de las oscilaciones en las regiones de XANES. La amplitud de estas oscilaciones disminuyó con el aumento de la temperatura de tratamiento térmico mostrando un mayor orden de la estructura para las muestras recocidas. Este efecto podría deberse al pequeño tamaño de los cristales del material nc-UO2 o al simple desplazamiento de la estequiometría del material así producido respecto al material de referencia UIV O2(grano-grande). Para cuantificar la contribución del tamaño del cristal a este efecto sería necesario un estudio con nc-UO2de distintos tamaños de cristal y una estequiometría de oxígeno fija (a ser posible la característica del UO2de referencia). Dado que en el momento de realización del estudio esta síntesis selectiva de nc-UO2no fue posible, un estudio alternativo para determinar el efecto de la talla del cristal en la estequiometría fue llevado a cabo. Para ello se utilizó un substituto como es el oxido de torio (ThO2) que cristaliza con la misma estructura de fluorita que el UO2y que posee además un único estado de valencia (catión) ThIV . Se llevo a cabo un estudio de una muestra de nc-ThO2así producido (no tratado térmicamente). En los espectros XANES correspondientes al borde Th-L3, el pico de la WL correspondiente al nc-ThO2así sintetizado, se encontró en una posición y amplitud idénticas a las de los espectros del ThIV O2(grano-grande) de referencia. Tan sólo se detectó una ligera disminución en la intensidad y número de oscilaciones, indicando solo un efecto débil en las distancias interatómicas y el orden reflejado en el espectro de XANES debido a la talla del cristal del material nc-ThO2. Este prácticamente idéntico comportamiento del material de referencia ThIV O2(grano-grande) y el nc-ThO2, sugiere que los desplazamientos observados anteriormente para el material de nc-UO2 tendrían su origen, no al tamaño de partícula sino más bien al desplazamiento de la valencia-catión hacia un estado más oxidado (UVI) del material así sintetizado. Teniendo esto en cuenta, la determinación de la O/M de nc-UO2apartirdelatécnica de XANES estaría justificada. En los espectros de EXAFS k3-ponderado para el material nc-UO2sintetizado por xviii el método acuoso, las oscilaciones y su amplitud aumentaron con la temperatura de recocido y el creciente tamaño de los cristales, aproximándose gradualmente al espectro típico para la estructura del UO2(fcc). El material nc-UO2así precipitado de 4 nm resultó difícil de ajustar a la estructura de fluorita pura. Los ajustes no eran estables y los datos tenían mucho ruido. La muestra recocida a 600°C y tamaño de cristal de 9 nm mostró un ordenamiento intermedio con oscilaciones claramente identificables. Tanto la muestra original así sintetizada de 4 nm como la muestra recocida a 600°C (9 nm) mostraron claramente distancias de enlace U-O y O-O más cortas en comparación con la referencia de UIV O2-(grano-grande). Esto resultó compatible con los estudios de XRD que mostraron una considerable contracción del parámetro de red para la muestra así sintetizada de 4 nm, siendo ésta menor con la temperatura de recocido. En última instancia, para la temperatura de recocido de 1200°C y un tamaño de cristal de 82 nm, las oscilaciones de EXAFS fueron similares, si no coincidieron totalmente con las del material de referencia UIV O2, indicando misma estructura-fcc (Fm-3m) y mismas distancias interatómicas. Esto concordó con la similar estructura mostrada en XRD entre la muestra de nc-UO2recocida a 1200°C y la referencia de UIV O2-(grano-grande). También resultó coherente con el estudio de XANES que no mostró prácticamente ninguna diferencia respecto a la estructura típica de fluorita para el caso de la muestra nc-UO2recocida a 1200°C. Los resultados de los espectros de EXAFS de k3-ponderada para el material nc-UO2de origen orgánico, fueron distintos a los arriba comentados para el material nc-UO2de origen acuoso, siendo no sólo el material de 5 nm nc-UO2 así precipitado difícil de ajustar a la estructura de fluorita pura, sino también las muestras tratadas a 600°C y 1200°C. Todas las muestras presentaron un alto grado de desorden y no se pudieron ajustar al material de referencia UIV O2-(grano-grande), por lo que debería tenerse en cuenta otra fase todavía aquí no identificada. Se adquirieron espectros de NMR Hahn-echo 17O MAS de muestras tratadas a distintas temperaturas de recocido en atmósfera reductora (Ar/5%H2). Se identificaron tres tipos diferentes de oxígeno a partir del ajuste del desplazamiento químico de los registros obtenidos para estas muestras, es decir, del desplazamiento del pico 17O de frecuencia de resonancia respecto al del espécimen de referencia y expresado en unidades relativas (ppm). En el presente estudio se tomo como referencia la resonancia del 17O de una muestra de H2O dopada con 17O y se definió como 0 ppm. La primera identificación correspondió a especies de oxígeno con un desplazamiento químico de casi 900 ppm para las muestras recocidas hasta una temperatura 650°C. Los otros dos tipos de especies de oxígeno identificados aparecieron claramente en el rango de temperatura entre 650°C y 1200°C. Estas especies mostraron una un pico agudo y la otra un pico 17O amplio. Éstos podrían atribuirse a un 17O en un entorno más bien cristalino y en otro entorno más desordenado (debido a la amplitud del pico), respectivamente. Ambos picos disminuyeron fuertemente su desplazamiento químico y su anchura a media altura (FWHM; Full Width at Half Maximum)enelintervalo de temperatura de 650°C a 800°C, para converger rápidamente a temperaturas por encima de 800°C a valores cercanos a los de la muestra recocida a 1200°C con 717 ppm (desplazamiento químico) y 5 ppm (FWHM), respectivamente. Este pequeño valor de FWHM es sinónimo de un entorno bien cristalizado, aunque sigue siendo ligeramente más ancho que los 3 ppm encontrados para el UIV O2-(grano-grande) de referencia. A pesar de ello, el desplazamiento químico (717 ppm) fue el mismo que el encontrado para dicha referencia UIV O2-(grano-grande). Puede decirse entonces que el entorno, para la muestra con el cristal de mayor tamaño (∼80 nm), de las posiciones del parámetro de red del oxígeno, está muy próximo al medido para la muestra de referencia xix UIV O2-(grano-grande). Basándose en la FWHM puede decirse que para que la señal de UO2-cristalino sea observable, es necesario llegar a un tamaño de cristal por encima de los 80 nm. Esto estaría en consonancia con la observación hecha mediante XRD para este tamaño de cristal de una estructura UO2-fcc con parámetro de red 0.5472 nm. Varias muestras de material nc-UO2sintetizadas por el método acuoso y tratadas a distintas temperaturas-clave de recocido, fueron analizadas bajo el espectrómetro de FTIR. Se observaron hasta 4 picos en el intervalo 400-4000 cm−1para la muestra de nc-UO2así sintetizada (RT). Estos podrían asignarse a la vibración bending (o de tijereteo) del H-O-H del agua coordinada, y a un posible estado más oxidado del material (UO2+x). Todos los picos disminuyeron en intensidad con la temperatura de recocido de las muestras. Además, para la muestra tratada a la temperatura de recocido de 1200°C, el espectro de IR se asemejaba al espectro de la muestra de referencia de UIV O2. Esto concuerda con los resultados de XANES arriba comentados donde se observó una estructura electrónica diferente para la muestra tratada a 600°C, mientras que la muestra tratada a una temperatura de 1200°C presentó una estructura similar a la de la muestra de referencia de UIV O2. También los resultados del EXAFS se caracterizaron por un orden pobre a 600°C, pero pares de oscilación totalmente ajustados a los de la muestra de referencia de UIV O2para la muestra tratada a 1200°C. Se llevo a cabo un estudio isotérmico para el material nc-UO2sintetizado, del crecimiento de grano durante periodos de tiempo largos y bajo el HT-XRD. Para las temperaturas de recocido de 500°C, 700°C y 900°C, bajo atmósfera estática e inerte de He, el crecimiento de grano se produjo en las primeras horas de tratamiento a temperatura constante alcanzando un tamaño de cristal promedio estable a dicha temperatura (el crecimiento de grano cesó a partir de ese momento). En el caso de la isoterma a 1200°C y bajo atmósfera estática de He, el material presentó un crecimiento continuo sin llegar a alcanzar un tamaño constante de grano en las primeras 50 h. Se obtuvo una energía de activación de la difusión entre 0.93 eV a 1.25 eV. Estos pequeños valores de energía de activación obtenidos, podrían deberse principalmente a los límites de difusión de grano (superficie e interfaz). Se midió un parámetro de red de 0.5472 nm para las muestras tratadas durante 50 h a 900°C bajo atmósfera de Ar/H2(y tras el enfriamiento del material), obteniendo una talla final de cristal de unos 50 nm. Es por ello que, en principio no sería necesario alcanzar una temperatura de 1200°C (y por tanto una talla de cristal de 80 nm) para conseguir un material con el típico valor de parámetro de red del UO2de granos grandes (a=0.5472 nm), como se había comentado más arriba. Además, un tamaño medio de cristal de 322 nm fue medido a 1200°C durante 50 h en atmósfera estática de He. Teniendo esto en cuenta, durante el proceso de sinterización de los monolitos sería necesaria una temperatura por debajo de 1200°C para evitar el crecimiento extremo de las partículas (>200 nm). Sin embargo, se midió un tamaño final de cristal de 85 nm para las muestras de nc-UO2recocidas a 1200°C durante 50 h bajo atmósfera reductora de Ar/H2. Incluso después de 200 h a esta temperatura en condiciones reductoras, se midió un tamaño final de cristal de 150 nm (bastante menor que el valor de 322 nm observado bajo atmósfera de He y 50 h de tratamiento). Esta diferencia podría deberse al estado inicial de oxidación de las muestras así sintetizadas de nc-UO2y su evolución bajo atmósfera estática e inerte de He. Un UO2sobreestequiométrico presentaría un aumento mayor de los coeficientes xx de auto-difusión y del flujo de masa, incrementando así el movimiento en el límite de grano (o cristal) y crecimiento del grano. De hecho las diferencias entre el coeficiente de difusión del UO2de granos grandes y el nc-UO2, son compatibles con una mejora de los procesos difusión, ya sea por efecto del tamaño de grano o por el ratio O/U>2. Consolidación y caracterización de monolitos de nc- UO2 Se probaron diferentes rutas alternativas para la consolidación de los monolitos (ej. prensado convencional uniaxial, float packing, etc). Las pastillas así prensadas, fueron sinterizadas a temperaturas entre 900°C y 1200°C bajo atmósfera de Ar/H2. Las condiciones óptimas de sinterización se dedujeron a partir del estudio de crecimiento de cristal isotermo durante largos periodos de tiempo bajo atmósferas de He y de Ar/H2. Esto evitó el riesgo de un crecimiento de grano desproporcionado incluso a la temperatura más alta estudiada de 1200°C. En algunos casos se practicó también un pre-tratamiento térmico del polvo de nc-UO2para evitar la formación de grietas durante la etapa de sinterización debido a la presencia de agua o compuestos orgánicos en el material dependiendo de la síntesis utilizada. Las pastillas sinterizadas presentaron una apariencia fuerte aunque se podían observar grietas finas en algunas de ellas. Se obtuvieron densidades de sinterización entre 75.5-90.5%de la densidad teórica (TDUO2=10.96 g/cm3). Se obtuvo un tamaño de grano promedio de ∼200 nm, replicando la estructura que aparece a altos grados de combustión (HBS; high burn-up structure) para los diferentes tipos de monolitos de nc-UO2sinterizados. También se llevaron a cabo experimentos de dilatometría donde se comparó la contracción entre la pastilla fabricada con material de nc-UO2, con la pastilla fabricada con el típico UO2-(grano-grande) y a partir del proceso estándar de fabricación. Se observaron mejores actividades de sinterización a temperaturas inferiores para el material nanocristalino en comparación con las medidas para las pastillas de UO2-(grano-grande). El rango de temperatura desde el inicio hasta la completa densificación, ocurrió a temperaturas mucho más bajas para las pastillas de nc-UO2 (200-955°C, con un ratio máximo de sinterización a 740°C), en comparación con las pastillas de UO2-(grano-grande) [Lahiri et al., 2006] (900-1540°C, con un máximo de sinterización a 1200°C). Esto podría deberse a la mayor superficie presente en el material de nc-UO2comparado con el típico UO2-(grano-grande), lo que llevaría a una sinterización más efectiva (a temperaturas más bajas). Se encontró una energía de activación de la sinterización de Q= 171 ±7kJ/mol asumiendo difusión de superficie, yQ= 114 ±5kJ/mol asumiendo difusión de volumen para el monolito de nc-UO2. La energía de activación determinada para un monolito de UO2de grano grande es de Q= 287 kJ/mol según [Lahiri et al., 2006]. Ambos mecanismos de difusión mostraron pues valores bajos para las energías de activación de sinterización como es típico para los nano-materiales. Esto se traduce en una clara ventaja tecnológica en la fabricación de monolitos de nc-UO2debido a su alta capacidad de densificación a temperaturas bajas. El mantenimiento de un rango de temperaturas aceptablemente bajo durante el proceso de sinterización disminuirá costos y simplificará la tecnología de fabricación. Los monolitos de nc-UO2ofrecen también la posibilidad de ajustar el tamaño de grano a voluntad mediante la variación de las temperaturas y tiempos de sinterización. xxi Las macroestructuras de los distintos monolitos de nc-UO2se caracterizaron por microscopía óptica (OM). Las microestructuras se caracterizaron por observación de la fractura-fresca a distintas ampliaciones bajo el SEM. En las imágenes de SEM de la fractura-fresca de las pastillas sinterizadas a la temperatura más baja de 900°C, se observaron granos no del todo definidos. La sinterización de los cristales parecía todavía en fase de desarrollo, por lo que a partir de este momento se utilizaron temperaturas de sinterización de 1200°C. Se observaron macro-fisuras en algunas de las pastillas, pero no para los monolitos fabricados con nc-UO2sintetizado por el método orgánico con polvo con pre-tratamiento térmico, y tampoco para la muestra del método de consolidación float-packing y polvo sintetizado por el método acuoso. Todas las macroestructuras, con excepción de esta última, mostraron densificación no homogénea (porosidad residual entre las zonas densificadas). También la muestra con prensado convencional de polvo nc-UO2pre-deshidratado sintetizado por el método acuoso, mostró una buena calidad en comparación con el resto, desde el punto de vista de la densificación. Sin embargo, es necesario la introducción de mejoras en la formación de los monolitos para evitar el problema de las grietas durante el sinterizado (y por lo tanto disminución de las propiedades del material). En cuanto a la microestructura de la superficie de fractura-fresca, la pastilla de nc-UO2sintetizado por el método acuoso y fabricada por consolidación float-packing y sinterización a 1200°C, fue la aproximación más cercana al material HBS obtenido hasta ahora. El tamaño medio de grano para los diferentes monolitos estuvo entre 170 nm y 250 nm. Esto fue un de los grandes logros de este trabajo. Propiedades mecánicas como la dureza Vickers (HV), dureza Knoop (HK) y módulo de Young (E) se determinaron para las pastillas de nc-UO2sinterizadas. Un aumento en la dureza (HV) y valores bajos para módulo-E(de hasta un 30%) fueron en general observados para los diferentes monolitos de nc-UO2en comparación con aquellos de UO2-(grano-grande). También se utilizó microscopía de barrido acústico (SAM) para la estimación y la comparación del módulo de Young obtenido por identación. Los resultados obtenidos por SAM (E=155 GPa) coincidieron con los derivados por micro-indentación (E=155 GPa). La diferencia observada con respecto a pastillas de grano-grande de UO2(220 GPa), podría estar influenciada por las imperfecciones de la microestructura (nano-cavidades en las intersecciones de tres granos, poros, grietas, etc.). Sin embargo, esta caída del módulo es todavía demasiado grande como para ser totalmente atribuida a la presencia de cavidades. El mismo tipo de tendencia observado en las muestras de nc-UO2,esdecir,aumentodelosvaloresde HVy disminución de los valores del módulo-E, ya se había determinado antes en el combustible estándar de UO2tras haber alcanzado valores elevados de BU. En este caso la disminución del módulo-Etampoco pudo ser totalmente atribuída a un aumento de la porosidad, y contradijo el efecto de la disolución de los productos de fisión que provoca en realidad un aumento de la rigidez del material. Dado que los combustibles nucleares irradiados se transforman en una estructura nano-recristalizada con el aumento de BUs [Spino et al., 2012], los efectos (parcial) del aumento de HV (la disolución de los productos de fisión provocan también endurecimiento) y además la disminución del módulo-E(sumado al causado por la porosidad), al igual que los efectos observados en el presente trabajo, podrían atribuirse a la nano-estructura de los combustibles sometidos a elevados BUs. Se confirmó con éxito la dependencia con el tamaño del cristal, de las propiedades físico-químicas del nc-UO2. Así, se comprobó que la compresibilidad del nc-UO2era xxii de hecho mayor que la del estándar-UO2de tamaño grande. Se confirmó también una dependencia de las propiedades de expansión térmica con el tamaño del cristal para el material de nc-UO2. La expansión térmica aumentó con la disminución del tamaño de cristal, al mismo tiempo que el módulo de compresibilidad disminuyó. Esto es compatible con la relación Grüneisen que presenta un producto constante entre la conductividad térmica y el módulo de compresibilidad. Sin embargo sigue pendiente la verificación de esta tendencia sobre el calor específico (Cp), necesaria para completar el análisis de la relación de Grüneisen. En cuanto a la compresibilidad del material bajo difracción de rayos X in situ de alta presión (HP-XRD), se realizó un estudio de la dependencia del módulo de compresibilidad con el tamaño de cristal para el material nc-UO2. Se estudiaron tres tamaños de nc-UO2diferentes (4 nm, 6 nm y 34 nm) hasta una presión de 27 GPa, y se determinaron las constantes de compresibilidad correspondientes B0yB 0.El módulo de compresibilidad del UO2sufrió una disminución extrema para las partículas de tamaño dentro del rango nanométrico. Para las partículas de nc-UO2de 4 nm se observó un módulo de compresibilidad (B0)entornoaun40%menor que el medido UO2-grano-grande (granos de tamaño micrométrico) [Pujol et al., 2004]. Esto confirmó la dependencia del módulo de compresibilidad con el tamaño de las partículas. Sin embargo, un estudio con partículas de tamaños mayores que los aquí considerados (>34 nm) sería necesario para garantizar que la tendencia observada en estos monolitos (disminución del módulo de elasticidad) se debe al tamaño de los granos, y no sólo debido a las imperfecciones y porosidad posiblemente presente en las muestras. Los resultados de las pruebas de difusividad térmica para el material de nc-UO2 compactado mostraron un comportamiento similar al del material UO2-estándar (micro-grano). La difusividad térmica para las pastillas sinterizadas de nc-UO2 (∼200 nm, 90%densidad), se determinó en el rango de temperatura 254°C a 1165°C. Se hizo una extrapolación de los resultados obtenidos hasta una densidad de 95%yse encontró la misma difusividad térmica que en las pastillas fabricadas con estándar-UO2 (grano-grande) y densidad del 95%[Fink, 2000]. Respecto al temido empeoramiento de la conductividad térmica del material en la HBS debido al efecto del tamaño de grano (resistencia Kapitza), quedó aquí demostrado el no-deterioro de las propiedades térmicas para las pastillas de UO2con un tamaño de grano de 200 nm imitando la HBS. Se llevo a cabo la determinación del punto de fusión por calentamiento-láser y detección de la temperatura pirométrica para nc-UO2-compactado de dos tamaños diferentes de nano-grano (aproximadamente 10 nm y 200 nm), evaluándose su variación con respecto al UO2-estándar de grano grande. Se encontró una disminución del punto de fusión para el compacto con material nc-UO2-(10 nm), de aproximadamente 150°K con respecto al valor típico para UO2-estándar. Esta reducción sería a priori debido al tamaño nano de los granos. Sin embargo, el parámetro de red medido para dicha muestra antes de aplicar la fusión (a=0.5438 nm) resultó inferior al valor típico de la referencia UO2-estándar (a=0.547 nm), indicando por tanto la presencia de un óxido sobre-estequiométrico el cual también podría ser causante de esta disminución del punto de fusión. Para corroborar la tendencia medida con la reducción de tamaño de grano, sería necesaria una muestra de nc-UO2estrictamente estequiométrica. Sin embargo para el compacto con material nc-UO2-(200 nm) se encontró un punto de fusión igual al de la referencia UO2-estándar. Una estequiometría de O/M=2.00 del parámetro de red fue medida para esta muestra antes de provocar la fusión. Éste es xxiii diameter) of 1.42 nm and a lattice parameter of 0.5579(1) nm, was found. In both cases, no aggregation of the precipitated nanoparticles has been observed on the TEM images. Crystallization and grain growth in f(T) for nc-UO2 To study the composition of the precipitates obtained by both methods above mentioned and their propensity to thermal growth in the unconsolidated state, further analysis of the precipitated material annealed at different temperatures was performed by applying the thermal analytical and X-ray diffraction techniques like thermogravimetric analysis and differential thermal analysis (TGA/DTA), X-ray diffraction (XRD) and high temperature X-ray diffraction (HT-XRD), and spectroscopic techniques such as X-ray absorption spectroscopy (XAS), magic angle spinning nuclear magnetic resonance spectroscopy (MAS-NMR) and infrared spectroscopy (IR) and characterization techniques like transmission electron microscopy (TEM). The evolution of the crystallite size, the lattice parameter and the lattice strain were determined from ambient temperature up to 1200°C under inert atmosphere. For the aqueous precipitated nc-UO2, only a weak effect of temperature on the crystallite size occurred below 700°C, remaining this below 7 nm in this temperature range. On exceeding 700°C, the crystal size grew, however, steadily with temperature, to reach the value of 73 nm at 1200°C. Opposite, the strongest lattice parameter increase was measured in the lowest temperature range 20°C-700°C, whereas in the temperature range 700°C-1200°C only a weak lattice expansion was observed, which almost coincided with the reversible thermal expansion of UO2. Thus, on the base of measurements done after cooling, a recovery of the UO2typical crystal structure was achieved during this annealing under static He atmosphere, passing from the initial lattice parameter value of 0.5417 nm for the as-produced nc-UO2, to the value of 0.5473 nm after exposure to 1200°C. The verified initial lattice contraction of the as-produced nc-UO2below the normal value of bulk stoichiometric UO2(a=0.5470 nm) is attributed mainly to oxidation. For the organic precipitated nc, a similar particle-growth behaviour with temperature was observed, with almost no crystal-dimension changes up to 700°C (because of the pre-thermal treatment performed), followed by an intense crystal-growth between this temperature threshold and the final annealing temperature of 1100°C, obtaining a final crystal size of 150 nm. As for the derived oxygen stoichiometry from the lattice parameter values, also an initial oxidation of the nc-particles produced by the organic method was confirmed, although in lower extent as for the case of the particles produced by the aqueous method. The particles recovered the normal lattice dimension of bulk stoichiometric UO2for annealing temperatures above 750°C. As for the determined lattice strain, also a similar behaviour was observed for particles obtained from both preparation methods. The main observation in both cases was that the lattice strain decreased continuously with temperature, until being practically extinguished at the temperature at which the boosted crystal growth started. This confirmed the lattice strain to having acted in both cases as crystal-growth inhibitor. xxx Structure and oxygen-stoichiometry studies by XRD, XANES, EXAFS, NMR AND FTIR The structure of the produced nc-UO2material as a function of temperature and, in this case under reducing atmosphere, was also studied by XRD and compared to the reference bulk-UIV O2. The lattice constant of the material in the cooled state after reaching different maximum temperatures (600°C and 1200°C) was measured. This allowed the separation of the thermal expansion contribution in the high-temperature values to obtain cleaner curves of thermal expansion vs. temperature and lattice dimension vs. crystal size. No big differences in crystal size, lattice and strain, were observed between measurements made under inert (above commented) and reducing atmospheres (measurement after cooling) for the material obtained by the aqueous method. However, a notable change in the crystallite size was observed for the material obtained with the organic method at 1100°C, which showed a size of 150 nm under He and a size of only 12 nm under Ar/5%H2. Comparing the aqueous and organic produced material under reducing atmosphere (Ar/5%H2) and without pre-thermal treatment, no big change was observed until 600°C anneal, but at 1200°C. At the last temperature, a size of 82 nm was measured for the aqueous method material compared to the 12 nm obtained for the particles from the organic method at the same temperature under reducing atmosphere (Ar/5%H2), were measured. That could be ascribed to the surface layer protecting the organic precipitated nanoparticles. In addition to the XRD studies, XANES was used to determine the oxidation state of the U cations and the corresponding molar fractions and the derived O/U ratios. The XANES spectra at the U-L3edge for the aqueous method material and for the organic method material, showed similar improving trends with increasing temperature and as the stoichiometry shift (x) decreased (UO2+x). The samples studied were nc-UO2as produced and after thermal treatment at 600°C and 1200°C under Ar/5%H2. Compared to the reference sample of bulk (large grain) UIV O2material, the peak of the WL shifted slightly to lower energies and increased in intensity, and the oscillations within the XANES regions increased. The amplitude of these oscillations decreased with the increasing temperature of thermal treatment showing a higher structural order of these annealed samples. This effect could be either due to the small crystal size of nc-UO2samples or to the stoichiometry shift of the synthesised material. To quantify these contributions a dedicated study with nc-UO2with fixed oxygen stoichiometry and different crystal sizes would be needed. Since at this moment this kind of selective synthesis of nc-UO2 was not possible, an alternative separate study of the size effect in the stoichiometric nano-oxide-material was attempted using the substitute thorium dioxide (ThO2), known to crystallize with the same fluorite structure as UO2and to maintain a unique cation-valence state ThIV . In this work, a study of as-produced nc-ThO2(not thermally treated) was done. In the corresponding XANES spectra at the Th-L3edge, the peak of the WL corresponding to nc-ThO2at RT (as-produced) had an identical position and amplitude as the one of the reference spectra of large-grain bulk ThIV O2. Only a slight peak intensity decrease and somewhat fewer oscillations were detected, which indicated only a weak effect of the crystal size on the interatomic distances and ordering reflected in the XANES spectra. This identical behaviour of the large grain ThIV O2and the xxxi nc-ThO2suggested that the displacements observed formerly for nc-UO2would have been not due to the particle size, but rather to the shift of the cation-valence towards the oxidised state (UVI). Having that into account, determining the O/M of nc-UO2 from the XANES shift seems to be justified. In the k3-weighted EXAFS spectra of nc-UO2particles from the aqueous method the oscillations and their amplitude increased with the annealing temperature and the resulting growing crystal size, approaching gradually those typical of the UO2(fcc)- structure. The as-precipitated 4 nm as-precipitated sample was very difficult to fit with a pure fluorite structure, as the fits were non stable and the data noisy. The 600°C annealed 9 nm sample showed an intermediate ordering with clearly identified oscillations. Both the original 4 nm-sample and the 600°C-annealed 9 nm-sample showed clearly shorter U-O and O-O bond-distances compared to the reference bulk-UIV O2sample. This was compatible with the XRD studies showing considerable lattice contraction for the as-received sample and in lower extent, with intensity decreasing with temperature, for the annealed samples below 1200°C. Ultimately, for particles annealed at 1200°C and with a crystal size of 82 nm, the EXAFS oscillations were similar, if not entirely matching, to those of the bulk-UIV O2, indicating the same fcc-structure (Fm-3m) and same interatomic distances and substantial crystal perfection. That was in agreement with the structure-similarity shown in the XRD analysis between the nc-UO2sample annealed at 1200°C and the reference large-grain bulk-UIV O2sample; and was also consistent with the XANES studies, showing no departure from the fluorite structure for the fully annealed UO2nanoparticles. In the k3-weighted EXAFS spectra of the UO2nanoparticles from the organic origin, the results were different as above, being not only the as-precipitated 5 nm sample very difficult to fit with a pure fluorite structure, but the samples treated at 600°C and 1200°C, too. All samples presented a high degree of disorder and could not match at all the reference signature of bulk-UIV O2, with the meaning that another unidentified phase must be taken into account in this case. NMR Hahn-echo 17O MAS spectra could be acquired for samples prepared by the aqueous method after annealing at different temperatures under reducing atmosphere (Ar/5%H2). Three different oxygen environments could be identified from the fitting of the chemical-shift signatures of these samples, i.e., the records of the 17O-resonance-frequency peak displacement with respect to that of a reference specimen, expressed in relative units (ppm). In the present case, the 17O-resonance of a 17O-dopped H2O sample was taken as reference, and defined as 0 ppm. The first identification corresponded to oxygen species having a chemical shift of nearly 900 ppm and was found for samples annealed up to 650°C. The two other types of oxygen species identified appeared clearly in the temperature range 650°C-1200°C. These new species, i.e., one showing a sharp and the other a broad 17O-peak, could be respectively attributed to 17O in a well crystalline environment and in a more disordered one; the last due to the larger peak broadening. Both peaks diminished strongly their chemical shifts and half-maximum widths in the temperature range 650°C-800°C, to converge rapidly at temperatures above 800°C to values near those of the sample annealed at 1200°C, i.e., respectively, 717 ppm (chemical shift) and 5 ppm (FWHM), which due to very small peak broadening (FWHM) indicated a very well crystallized environment. This last was still slightly bigger than the 3 ppm found for UIV O2-bulk. Despite this, the chemical shift (717 ppm) was the same as that found for UIV O2-bulk. Hence, one can say that the environment around the oxygen lattice positions in the case xxxii of the sample with the biggest crystallite size (∼80 nm) was very close to that of UIV O2-bulk. Based on the FWHM, one can say that to observe the signal of crystalline UO2a crystallite size above 80 nm should be reached. This is in line with the observation by XRD of an UO2-fcc structure with lattice parameter 0.5472 nm in this case. Several samples from the aqueous method at key annealing temperatures were also analysed under the FTIR spectrometer. In the case of nc-UO2in the as-produced condition (RT), four peaks in the range 400-4000 cm−1could be observed. They could be assigned to the bending vibration of H-O-H of the coordinated water, and to a possible more oxidised state (UO2+x). All these peaks diminished in intensity with the annealing temperature. Hence, at 1200°C the IR spectra looked like the one of the UIV O2reference sample. That was also in agreement with the above commented XANES results where a different electronic structure was seen at 600°C, while at 1200°C a similar structure to bulk-UIV O2was found. Also EXAFS was characterized by poor ordering at 600°C but entirely matching with the bulk-UIV O2oscillation pairs at 1200°C. Isothermal grain-growth study of the synthesized nc-UO2was then performed by XRD and HT-XRD. For the annealing temperatures of 500°C, 700°C and 900°C and a static and inert atmosphere of He, the grain growth took place in the first hours of isothermal hold until a stable average crystal size was established at the applied temperature, at which time grain growth ceased. For the isotherm at 1200°C and a static atmosphere of He, the material had a continuous growth not reaching a constant grain value in the first 50 h. An activation energy of diffusion of 0.93 eV to 1.25 eV was obtained. The low activation energies obtained could be related predominantly to grain boundary (surface and interface) diffusion. A lattice parameter of about 0.5472 nm was already found for the samples treated at 900°C after 50 h dwell time under Ar/H2obtaining a final size about 50 nm. Therefore a temperature of 1200°C (and in consequence a final crystallite size of 80 nm) would be, in principle not necessary to reach the typical lattice parameter of the reference large-grained UO2(a=0.5472 nm), as above commented. An average crystal size of 322 nm was measured after cooling for the heat treatment at the highest temperature of 1200°C after 50 h dwell time under He. Taking that into account, it appears that a temperature below 1200°C would be necessary in the sintering process of the monoliths to avoid extreme growth of the particles (>200 nm). Nevertheless for the nc-UO2samples annealed at 1200°C during 50 h under Ar/H2 dynamic atmosphere, a final crystal size of 85 nm was measured after cooling. Even after 200 h dwell time at this temperature under reducing atmosphere, a final crystal size of 150 nm was seen (quite far from the 322 nm observed under He atmosphere after 50 h). This difference could be due to the initial oxidation state of the nc-UO2 samples and their evolution under a static He atmosphere. An hyperstoichiometric UO2would present a stronger increase of the self-diffusion coefficients and in the same way raise the mass-flow, for which enhanced grain-boundary motion and grain (or crystal) growth will occur. In fact the differences in the diffusion coefficient between bulk-large-grain-UO2and nc-UO2are compatible with an enhancement of the diffusion processes either by a diminishing of the grain size or by O/U>2 effects. xxxiii nc-UO2monolith consolidation and characterization Different alternative routes for consolidation into green bodies (e.g. conventional uniaxial pressing, float packing, etc.) have been tested. Afterwards the green bodies were sintered at temperatures between 900°C and 1200°C under Ar/H2atmosphere. The optimum sintering conditions were deduced from the long-isothermal crystallite growth studies under He and Ar/H2atmosphere. This ensured lack of disproportionate grain growth risks even at the highest temperature used of 1200°C. Also thermal pre-conditioning of the powder before pressing was in some cases done to avoid cracks during the sintering step due to the presence of water or organics (depending on the case) in the material. The pellets sintered presented a strong appearance although fine cracks were visually observable in some cases. Sintering densities between 75.5-90.5%of the theoretical density (TDUO2=10.96 g/cm3), were obtained. An average grain size of ∼200 nm, replicating the HBS, was obtained for all the different sintered nc-UO2pellets. Additional dilatometry experiments were performed to compare the shrinkage of the fabricated nc-UO2pellet with that of bulk-UO2(large grain) produced by a standard fabrication process. Enhanced sinter activities of the nanocrystalline materials compared to microcrystalline UO2were found at lower temperatures. The temperature range from onset to completion of the densification occurred at much more lower temperatures for the nc-UO2(200-955°C, with a maximum sintering rate at 740°C), compared to the bulk-UO2[Lahiri et al., 2006] (900-1540°C, with a maximum sintering at 1200°C). The reason of that might be the higher surface present in the nc-UO2compared with the bulk-UO2material, rendering the sintering to become more effective (at lower temperatures). The sintering activation energy was determined as Q= 171 ±7kJ/mol assuming surface diffusion and Q= 114 ±5kJ/mol assuming volume diffusion for the nc-UO2monolith, compared to Q= 287 kJ/mol determined for bulk-UO2in the literature [Lahiri et al., 2006]). Both diffusion mechanisms showed low values of the sintering activation energies as typical for nanopowders. That means a clear technological advantage in the fabrication of nc-UO2monoliths due to its high densification capacity at low temperatures. Furthermore, the nc-UO2offered the possibility of adjusting the grain size at will by varying sintering temperatures and times. Maintaining an acceptably low temperature range in the sintering process, it will diminish the costs and simplify the manufacturing technology. Characterization of macrostructures by optical microscopy (OM), and microstructures by fresh-fracture observation by SEM, for different samples at different magnifications, was performed. Not well defined grains were observable in the fresh-fracture SEM images of the pellets sintered at low temperature of 900°C. The sinter of the crystals was still under development, therefore sintering temperatures of 1200°C were used afterwards. Macrocracks across different samples were observed, but not for the monoliths from nc-UO2synthesized by the organic-route with powder thermal pre-treatment, and not for the monoliths from the float-packing consolidation method and powder of the aqueous-synthesis. All macrostructures, with exception of the last one, showed non-homogeneous densification (residual porosity between densified areas). Also the conventional pressed sample of pre-dehydrated powder from the aqueous route showed a good quality in comparison to the rest, from the point of view of the densification. However, improvements in the performance of the monoliths would be necessary to avoid the problem of cracks in the sintered pellets (and therefore the diminishing material’s properties). Looking at the fresh fracture surfaces, the microstructure of the aqueous-route-powder pellet produced by float xxxiv packing consolidation and sintering at 1200°C, was the closest approximation to the HBS material obtained until now. The average grain size for the different monoliths was in the 170 nm to 250 nm range. Here a major success of this work was achieved. Mechanical properties as Vickers Hardness (HV), Knoop Hardness (HK)and Young’s modulus (E) were determined for sintered nc-UO2pellets. An increase in hardness (HV) and low values for E-modulus (up to 30%) were in general seen for the different nc-UO2monoliths in comparison with bulk-UO2. Also scanning acoustic microscopy (SAM) was used for the estimation and comparison of the Young’s E- modulus obtained by indentation. The results by SAM (E=150 GPa) matched the ones derived from microindentation (E=155 GPa). This difference observed with respect to bulk-UO2pellets (220 GPa), could be influenced by microstructure imperfections (nanocavities at triple-grain junctions, pores, cracks, etc.). However, the drop was still too large to be attributed only to the presence of cavities. The same type of tendency observed in the nc-UO2specimens, i.e. with increase of HVvalues and decrease of the E-modulus values, has been found before in irradiated standard-UO2fuel at high BUs. In this case also the E-modulus decrease could not be fully attributed to a porosity increase and was to contradict the effect of the fission products dissolution, which causes in reality an increase of the material’s stiffness. Since with the increase of BU the irradiated nuclear fuels transform into a nano-recrystallized structure [Spino et al., 2012], the effects of (partial) HV-increase (fission products dissolution causes as well hardening) and additional E-modulus decrease (beside that caused by porosity) like the effects observed in the present work could be attributed in high BU fuels due to the nanostructure. The confirmation of the size-dependent physical-chemical properties of nc-UO2has been successfully accomplished. So the compressibility of nc-UO2wasprovedinfactto be larger than that of standard-UO2. A size-dependence of the thermal expansion properties of nc-UO2was also confirmed. The thermal expansion was shown to increase with the size-decrease, at the time that the bulk modulus decreased. This is compatible with the Grüneisen relationship showing a constant product between the thermal conductivity and the bulk modulus. However, verification of the trend in the specific heat (Cp) is still lacking, which is indeed neeeded to complete the Grüneisen-relationship analysis. Regarding the material’s compressibility, in-situ high pressure X-ray diffraction (HP-XRD) has been performed for the study of the bulk modulus dependence on the crystal size in nc-UO2. Three different nc-UO2sizes (4 nm, 6 nm and 34 nm) were studied up to a pressure of 27 GPa and the corresponding compressibility constants B0and B 0determined. The bulk modulus of UO2suffered an extreme decrease in the nano-size particle range. For the 4 nm-size nc-UO2-particles, a bulk modulus (B0) around 40%lower than the one measured for bulk-UO2(micron-size grains) [Pujol et al., 2004], has been observed. This confirmed the dependence of the bulk modulus with the crystallite size. However, studies with bigger particle sizes as the ones here studied (>34 nm) would be necessary to guarantee that the tendency observed in the monoliths (decrease of E-modulus), is due to the size of the grains and not just because of imperfections and porosity possibly present in the samples. The results of thermal diffusivity tests of the compacted nc-UO2-material showed similar behaviour as that of standard, nuclear grade UO2(bulk). The thermal diffusivity for sintered nc-UO2(∼200 nm, 90%density), was determined between xxxv 254°C to 1165°C. Extrapolation to 95%density was done and same thermal diffusivity as standard bulk-UO2pellet [Fink, 2000]with95%density was found. Regarding the feared worsening of the thermal conductivity of the HBS material due to grainsize effect (Kapitza resistance), it has been here shown that no thermal properties deterioration has to be expected for the 200 nm-UO2pellet material mimicking the HBS. Determination of the melting point by laser-heating and pyrometric temperature detection has been performed for compacted nc-UO2with two different nano-grain sizes (about 10 nm and 200 nm) and their variation with respect to bulk-UO2(large-grain), assessed. A melting point depression of about 150°K with respect to the normal value of bulk-UO2was found for the 10 nm-size nc-UO2sample. This reduction would be a priori due to the nano-size grains. However, the measured lattice constant of the sample before melting (a=0.5438 nm) was below the value of bulk-UO2(a=0.547 nm) and indicated in reality a hyperstoichiometric oxide, which would also cause a melting point decrease. To corroborate the measured tendency with the grain-size reduction, a strictly stoichiometric nc-UO2sample would be needed. However, an identical melting point as for bulk-UO2, was found for the 200 nm-sample for which a stoichiometry of O/M=2.00 was confirmed from the lattice constant measurement before melting. This is an important technological result for the use of nc-UO2ceramics as nuclear fuel. Indeed, a lower melting point would pose a problem for the licensing of the monoliths as a fuel for the reactor. Fortunately the possibility of a lower melting point disappears for the 200 nm-UO2samples, as it would occur for the HBS material in the reactor, too. So, postulated nano-effects such as diminution of the thermal conductivity and the melting point could be here excluded as weak points for the use of nc-UO2as a nuclear fuel. These effects might be relevant for very low crystal/grain sizes (∼10 nm) but they disappear for grain sizes of ∼200 nm, where, conveniently, the sought advantageous properties of the nano-structure (super-plasticity, low swelling under Xe-bombardment [Spino et al., 2012], self-limiting grain growth, etc.), still remain. This anticipates the lack of property loss of the developed nc-UO2monoliths for technical applications in this size range. Future Recommendations The licensing of nuclear fuel is made on basis of its safety performance not just only under normal operation conditions, but also when a temperature rise occurs in the fuel. This could be caused in a Loss of Coolant Accident (LOCA) or in a Reactivity Initiated Accident (RIA). Under such extreme conditions fuel fragmentation could occur. During this thesis, one attempt was made to mimic such an accident in an out of pile experiment using nc-Y-ZrO2as a sample instead of nc-UO2. This test was made in a facility at ITU (Institute for Transuranium Elements) known as POLARIS, which permits very rapid laser heating of the sample. The initial material tested in POLARIS was pore free and its surface was flat. The laser treatment showed that a local swelling occurred through formation of porosity. This experiment was not perfectly well controlled, but it is likely that the observed swelling was due to CO or CO2gas generated when the carbon impurity in the material reacted with oxygen from the atmosphere, which caused pore formation, in a process similar to the production of foamed glass. A particularly interesting result xxxvi of the test is that the formed pores were closed and astonishingly similar to those of the HBS-zone in high burn-up fuels. Chances appear therefore that during such kind of postulated fuel melting accident, at least part of the fission gas could be trapped in potentially forming closed pores, as it occurs in the HBS material at low temperatures. Although these experiments are preliminary they suggest a promising novel method to test the gas retention capability of the nc-UO2fuel under accident conditions. Finally, another important method to understand the resistance of nc-materials to irradiation can be provided by ion beam irradiation tests. This can be done at facilities like the ANL (Argonne National Laboratory) IVEM-Tandem facility in Chicago, where irradiation with inert gas ions (He or Xe) with on-line TEM observation provides a very useful way to implant the gas atoms and to evaluate how they behave in the matrix, e.g. dissolution therein, formation of bubbles, transport of bubbles along grain boundaries, etc. Concluding remarks Successful consolidation of the synthesized nanocrystalline UO2nanopowders into dense pellets mimicking of the High Burn-up Structure (HBS) as ideal system has been achieved. From the different synthesized nc-UO2powders (4-5 nm size) to the nc-UO2compacted monoliths with 200 nm average grain size and about 90%density were achieved. Stability of the structure after ageing and self limiting grain growth kinetics up to temperatures of 1200°C, were shown. The out-of-pile mechanical properties of sintered pellets (in terms of hardness and elastic modulus) were confirmed to closely resemble those of the HBS-material in-pile. Beneficial properties found, like stability of the structure, enhanced mechanical properties and self-limiting grain growth, strongly encourage the performance of irradiation tests to verify the in-reactor behaviour. As determined previously in out-of-pile tests of monoliths of the brother system of nanocrystalline nc-Y-ZrO2[Spino et al., 2012], a strong reduction of the gas bubble swelling, long term thermal stability of the pore-grain configuration, and striking superplastic behaviour and accelerated creep, would be expected as well for the developed nc-UO2. Confirmation of anomalies in the physical properties of the material for grain sizes in the absolute nanorange (<30 nm), consistent with observations in other nc-systems was also achieved. These pernicious nano-effects, as diminution of the thermal conductivity and the melting point, which could be a weak point for the use of nc-UO2as a fuel, were found, however, to become relevant only at very low crystal/grain sizes (<30 nm) and to disappear for grain sizes of ∼200 nm, where, suitably, the other searched beneficial properties of this nanostructure super-plasticity, low gas-bubble swelling, self-limiting grain growth, etc., remain. This anticipates the lack of property loss of the developed nc-UO2monoliths for technical applications in this size range. This has been a very rewarding work, with a number of breakthroughs achieved. Much has been learned, but more needs to be done to determine the true potential of this intriguing material. xxxvii xxxviii Contents Abstract xii Resumen xxvi Summary xxxviii List of Symbols and Abbreviations xliii 1 Introduction 1 1.1 Background and state of the art. ..................... 1 1.2 Goal of the thesis. .............................. 10 2 Analytical and characterization techniques 13 2.1 Electrochemical analysis .......................... 13 2.1.1 Cyclic Voltammetry (CV) ..................... 13 2.1.2 Electrolysis ............................. 14 2.2 Spectroscopy ................................ 15 2.2.1 Ultraviolet-visible spectroscopy (UV-Vis) ............. 15 2.2.2 Dynamic Light Scattering (DLS) ................. 15 2.2.3 X-ray Absorption Near Edge Structure (XANES) and Extended X-ray Absorption Fine Structure (EXAFS) ............ 16 2.2.4 Nuclear Magnetic Resonance spectroscopy (NMR) ........ 17 2.2.5 Infrared spectroscopy (IR) ..................... 17 2.3 Electron microscopy ............................ 18 2.3.1 Scanning Electron Microscopy (SEM) ............... 18 2.3.2 Transmission Electron Microscopy (TEM) ............ 18 2.4 X-ray scattering ............................... 18 2.4.1 Room Temperature X-Ray Diffraction (RT-XRD) ........ 19 2.4.2 High Temperaure X-Ray Diffraction (HT-XRD) ......... 19 2.5 Thermogravimetry/Differential Thermal Analysis (TGA/DTA) ..... 19 2.6 Dilatometry ................................. 20 2.7 Mechanical Characterization ........................ 20 2.7.1 Microindentation .......................... 20 2.7.2 High Pressure X-Ray Diffraction (HP-XRD) ........... 20 2.7.3 Scanning Acoustic Microscopy (SAM) ............... 21 2.8 Thermophysical characterization ...................... 21 2.8.1 Thermal Diffusivity ......................... 21 2.8.2 Melting Point ............................ 22 xxxix Chapter 1. Introduction 3     3 Figure 1.1: Light Water Reactor [The energy net, 2012][U.S.NRC, 2012]. This heat is transferred to the coolant (water), which passes by the clad, generating water vapour. This steam feeds conventional generators (steam-driven turbines) in the primary (for BWR) or secondary (for PWR) loop which produce the electricity. The fuel pellet inside the nuclear reactor is a material subjected to extreme conditions which change its properties with time and irradiation dose. Each atom in the fuel is displaced several times during its irradiation history but many return to equivalent crystallographic positions. Damage and local defects like interstitials, loops and vacancies are created. UO2is a poor heat conductor, therefore the heat transfer from the centre to the surface of the pellet is slow. A typical temperature profile of a LWR pellet is shown in Fig. 1.2 [Konings et al., 2011]. Furthermore, accumulation of solid fission products in the lattice and formation of gas bubbles make the pellet thermal conductivity to decrease. The fission gases precipitate in bubbles and lead eventually to compositional and microstructural changes, swelling of the fuel, as well as to embrittlement and hardening of the cladding. Other effects occur in the central sections of the fuel pellet (at higher temperature) and include grain growth, porosity build-up and an augmented gas release [Kleykamp, 1979][Stehle et al., 1975]. Fig. 1.3-left shows a fuel decorated with cracks after four reactor cycles (approximately four years). On the right side of the figure, the evolution of the geometry of the pellets inside a rod can be observed. Fuel cracks appear from the beginning of the irradiation due to thermal stresses. The fuel pellets swell owing to the accumulation of fission gas bubbles in the matrix and the segregation of low density fission-products phases (metallic and ceramic precipitates). As a result of the swelling, the fuel approaches the clad. Physical and/or chemical interaction can occur upon contact, which can induce clad deterioration and rupture [Garzarolli et al., 1979]. These type of changes can affect also the temperature profile of the fuel pellet (Fig. 1.2) by modification of 2 1.1. Background and state of the art. the gap thermal transfer conditions, limiting the life time of the fuel (and BUs) inside the reactor, if premature rod rupture occurs. Figure 1.2: A typical temperature profile of a LWR fuel as a function of the fuel pin radius [Konings et al., 2011]. A vast amount of work has been made in the last decades to characterise the behaviour of high BU fuels and to deepen the knowledge of the underlying phenomena, with the aim to increase the usage time of the fuel in the reactor [Watteau et al., 2001]. Besides that, in the last 20 years also a large increase of the research activities in nanocrystalline (nc)-materials for different aims and applications has been ostensible [Kulisch et al., 2009][Mathur and Singh, 2009]. The question arises whether these two apparently disconnected research areas would overlap and whether a link between nc-materials and high BU nuclear fuel materials would exist. The answer to this question is attempted below.        Figure 1.3: Macrograph of a fuel pellet after irradiation showing the typical radial cracks (left). Pellet inside the pin illustrating the swelling with the irradiation time (right). [Bailly et al., 1996]. Indeed, nuclear fuels approximately at the end of the third irradiation cycle 3 Chapter 1. Introduction (about 40 GWd/tM) undergo a structure transformation which begins at the edge of the fuel pellet and steadily progresses to its centre as the irradiation proceeds [Matzke and Spino, 1997][Spino and Papaioannou, 2000]. The original microstructure transforms into a nc-porous matrix [Nogita and Une, 1994] through restructuring of the accumulated irradiation defects. This is a sort of “self-healing” action where the material gets cured from damage by reordering itself [Spino et al., 2012]. Fig. 1.4 shows two ceramographs of different fuel zones at increasing local BU and where the mentioned change in the structure is clearly appreciated. The new nc-structure appearing (Fig. 1.4-right-micrograph) is called high burn-up structure (HBS). It is called also rim-structure because in UO2fuels it initiates at the rim or outer zones (r/R>0.98) of the pellets (colder periphery of the pellet; Fig. 1.4-right). This happens because the rim is the region which receives the highest dose (the most fissions) and therefore is exposed to highest local increase of BU (∼70 GWd/tM in the third cycle, at temperatures rarely exceeding 800°C [Sonoda et al., 2002]), and so the highest radiation damage, too. ! "!$" $  ~P=15-20% (GAS TIGHT); CLOSED POROSITY 1 µm r/ro=0.98 ~160 GWd/tM " $# $ P<4% (GAS PERMEABLE ) 1 µm r/ro=0.80 67 GWd/tM     Figure 1.4: Micrographs at different pellet radius areas [Spino and Papaioannou, 2008]. High burn-up structure (HBS or rim-structure) transformation [The energy net, 2012]. Most of the properties observed in the HBS, resemble those seen in the nanomaterial’s structures. Likewise heavily cold worked metals which after severe plastic deformation show grains in the nm-size range [Villegas and Shaw, 2009], the heavily damaged high-BU fuel region at the rim of the pellets display a profound modification of the microstructure on exceeding a critical dose. After this threshold, the original nuclear fuel with large-grains (10-20 μm) suffers progressively grain subdivision (low-angle sub-grain formation or low coordination number) and recrystallization (high-angle sub-grain formation or high coordination number) changes (Fig. 1.5)atthe end of which a new structure (the HBS) with uniformly nm-sized grains (100-250 nm) appears [Spino et al., 2012]. This transformation to a structure with nm-grains at the edge (rim) of the pellet results in lattice contraction [Spino and Papaioannou, 2000]. Furthermore, formation of new 1 μm-sized pores embedded in the matrix (Fig. 1.5) occurs, which entraps most of the created fission gas [Spino et al., 1996]. This porosity can reach values above 20%. At the beginning it was thought that the HBS could be responsible for cladding failures due to additional in-pile volume increase (swelling), which, in addition to 4 1.1. Background and state of the art.        100 nm 52 GWd/tM – 500 °C Xe  $ #  # % Figure 1.5: Coordination state change in the transformation to the HBS (or rim-structure). the supposed brittle behaviour of the transformed material [Matzke, 1992], lead to proposals of pellet-design changes to counteract the transformation [Swam, 1997] [Tulenko and Wang, 2008]. Also, it was thought that the rim-structure could act as a new source of gas release. In fact, one could think that the larger the grains (e.g. 10-20 μm, as in the original fuel or the fuel with lower high-BUs), the longer the pathways for the fission gases towards the grain boundaries and the exterior, and consequently, the more generally improved fission gas retention behaviour in comparison with the small grains of the HBS (0.2 μm grains). Indeed, there are still investigation lines which defend this opinion. However, other studies found that the percentage of gas liberated from the rim of the fuel pellets, where the HBS appeared, was low in comparison with the gas coming from the inner parts of the pellet where the original larger-grain structure is still present [Mogensen et al., 1999]. This philosophy has been followed by other publications, which demonstrates a generally beneficial evolution of the fuel properties, as in particular the retention of the fission gases, after the structure transformation [Rondinella and Wiss, 2010], [Spino et al., 2012]. Specifically, it has been observed that the HBS does not develop an open pore structure with interconnected channels, even at higher porosities [Noirot et al., 2008]. In fact, it has been seen that structures with large grains (10-20 μm) and with 1μm pores and porosities below 4%(in the original fuel and the fuels with low BUs; Fig. 1.4-left), are permeable to the flow of gases. In contrast, the rim-structure with small grains (100-250 nm) and with 1 μm pores and 20%porosity (Fig. 1.4- right), was suggested to remain gas tightly because of the formed closed porosity [Spino et al., 2004][Hiernaut et al., 2008]. Moreover the larged-grained structures are found to retain large amount of dislocation loops and gas bubbles inside the grains, which diminishes the mechanical properties (creep strain), increasing the risk of 5 Chapter 1. Introduction Figure 1.6: SEM micrographs of a fuel pellet at high-BU from the outer radius or rim (HBS in the first micrograph) to inner radial positions from [Manzel and Walker, 2002]. PCMI. In contrast, the small-grained structures with their high-angle sub-grain (high coordination state; Fig. 1.5) facilitating the GBs sliding deformation mechanisms, show improvement of the plasticity and creep strain, diminishing the PCMI-failure risks [Chung and Davies, 1979][Spino et al., 2008]. Besides that, the properties of the HBS indicate an enhanced radiation tolerance as reported by [Spino et al., 2012]. This study coincided with important evidence in the literature that nano-grained materials are more resilient to radiation damage than the corresponding large-grained materials due to defect recombination at their multiple grain boundaries has been reported by [Nita et al., 2005]. Also improvement of the thermal conductivity and other radiation-defects depending properties was found due to lattice-strain release after recrystallization [Ronchi et al., 2004], as well as fracture toughness increase and crack-healing tendency were anticipated [Spino et al., 2003]. Moreover, in relation to safety issues, the latest experiments on the high BU LWR fuels indicated no increase in the gas release and in the failure susceptibility during reactivity initiated accident (RIA) transients [Sasajima et al., 2010][Fuketa et al., 2006]. Also diminution of the aqueous corrosion rate under simulated geologic repository conditions for fuels containing HBS was found out [Ekeroth et al., 2009][Carbol et al., 2009]. Both facts confirmed the tightness of the structure. So it has been demonstrated that the HBS has exceptional qualities even in comparison with the original matrix (large-grain structure), with a number of improved properties that are really advantageous for a fuel. Then, why not imitate this recrys- 6 1.1. Background and state of the art. tallized material?. And why not imitate the structure appearing in the rim (HBS) and introduce it in this form as a fresh fuel inside the reactor?. Why not synthesise a fuel matrix like this HBS, which due to its apparent damage-resilience would withstand longer times under irradiation?. This is a radical step in nuclear fuel conception, hitherto not considered, since, as indicated before, the general thinking of the industry (fuel providers and utilities) was until now just the contrary one, i.e. trying to make larger and larger grains, under the premise to improve only one aspect of the fuel performance, namely the fission-gas release under steady state conditions, disregarding the implied worsening of the fuel plasticity due to grain-size increase, and also its poorer behaviour under power-ramps. 7 "$% "$%  &(*  &(''#)+'  Figure 1.7: A novel fuel microstructure: nc-UO2[The energy net, 2012]. On the contrary, apart from its still unproven improved resistance against radiationdamage (e.g. reduced swelling), the principal virtues of nc-fuels in-pile compared to conventional fuels will be the faster relaxation of PCI stresses through the higher plasticity induced by grain-refinement, plus the possibility of retention of most fission gases in formed closed pores. As a potential technological application, and as inspired in the behaviour of HBS-material in pile, the nc-fuel could retain fission gases inside the pores up to very high-BUs (>300 GWd/tM). Up to these BUs values, the porosity could increase up to 30%, until incipient pore interconnection would first begin [Konings et al., 2011]. The main aim here is thence to synthesise nc-UO2powders for the manufacture of bulk nc-fuel compounds for the characterization of their out-of-pile mechanical properties and irradiation behaviour. The produced monolithic pieces would have a volume of approximately 1 cm3, with a uniform grain size between 100 and 250 nm to mimic the rim-structure. The first question which appears is how to obtain enough amount of nc-UO2powder for fabrication trials of bulk-pieces, when the known methods of nanoparticle synthesis are generally tuned to yield just small amounts of material (mg range), primarily 7 Chapter 1. Introduction dedicated to analytical or research uses [Rousseau et al., 2002][Mennecart et al., 2004] [Wu et al., 2006][Rousseau et al., 2006][Opel et al., 2007][Rousseau et al., 2009]. The second important interrogate that arises then is how to obtain thereof dense large monoliths with uniform grain size around 200 nm, when the achievement of bulk nc-bodies is today one of the most demanding challenges in nanotechnology. The traditional methods for the production of UO2are based on the precipitation of UVI salt from liquid solutions. Then the precipitated material is oxidised to U3O8 by oxidative thermal treatment and afterwards the conversion to UO2(UIV )isreached by a second heat treatment under reducing conditions (800°C under Ar/H2). All these processes together provide a powder material which is further conditioned by different physical-chemical/mechanical methods to give a compositionally stable (in air) and free-flowing agglomerate, suitable to be compacted by standard powder metallurgy techniques (basically uniaxial-bidirectional pressing). The standard characteristics of commercial “ready-to-press“ UO2-powders have a UO2.1composition, agglomerate particle size 20-40 μm and crystallite size 200-500 nm. After pressing to desired geometry, final sintering of the compact (1600°C during 16 h) is necessary to get the desired density, of 95 to 98%of the theoretical density (TDUO2=10.96 g/cm3). However, the final grain size obtained by this conventional material synthesis route is in the range of 5-10 μm, which is far above the goal here (100-250 nm). 1.2 Goal of the thesis. The main goal in the present work was to develop an accessible route to produce defect-free nc-UO2-based monolithic ceramic specimens with tailored grain/pore microstructure. The target configuration consisted of a dense, uniform matrix with 100-250 nm sized grains with porosity levels of 10 to 20%to reproduce out-of-pile the properties of the HBS material, using similar methods as utilized in the previously studied case of nc-Y-ZrO2[Santa-Cruz, 2009]. Once this goal was accomplished, a study to determine their physical-chemical properties and their damage resistance in comparison with micron-grained materials was performed. The creation of the above novel fuel microstructure has been achieved in this work by passing through very different steps. From the material synthesis to the fuel pellet manufacture, many individual process stages, previously unknown or unexplored, had to be specifically developed and/or optimized. Hence, regarding the initial powder, considerable work was devoted to the development of two different chemical synthesis routes leading to deflocculated nc-UO2and nc-ThO2precipitates. ThO2is similar in structure to UO2but has the advantage of a single valency (i.e. IV). Although much information can be found about nano-chemistry and actinideschemistry, not very much information on the synthesis of nanoparticles with actinides compounds is available. The objective of the few reported works is, in the majority of cases, concerned with the issue of radionuclide release control during spent fuel geological disposal, namely the ultimate dissolution/re-precipitation of actinides in a in a fractured geological repository, by non-excludable contact of damaged spent fuels with water [Rousseau et al., 2002], [O’Loughlin et al., 2003]. The target in these cases is to examine the radiological hazards which could emerge from these situations and to quantify the effect of radioactivity release (and potential environment contamina- 8 1.2. Goal of the thesis. tion) on possible long-term permanence of spent fuels in contact with groundwater [Mennecart et al., 2004], [Rousseau et al., 2006]. Also the study of the elementary oxidation mechanisms occurring on an atomic scale on fuel/water contact during spent fuel storage has been performed with actinide oxide colloids [Opel et al., 2007], [Rousseau et al., 2009]. Other nanostructures based on uranium oxides have been used for catalytic purposes [Wang et al., 2008]. In other cases, the possible study of size-dependent physical and chemical properties has motivated the synthesis of high-quality colloidial of UO2[Wu et al., 2006]. Despite that, no publications of the above describe the production of nc-actinides other than as in the very small quantities needed for research or analytical studies. Although enough for these purposes, no monolith ceramic as desired in this work could be achieved with such (small) amounts of material. For the mimicking of the HBS nc-microstructure in whole UO2-pellet between 0.5 and 1.0 g of nc-UO2powder are needed. Therefore, a method with a definitely higher material’s yield must be developed. In the present work, two of the above reported methods for the synthesis the nc-UO2were chosen and developed as a source for the nc-powder needed to perform the described monoliths. In Chap. 3the first one is described. This is a controlled precipitation method that uses an electrolytically reduced aqueous solution of uranyl nitrate as precursor and dropped NaOH-solution as alkalinisation agent to trigger the precipitation of the nc-material in the vicinity of the U4+ solubility line, which was originally described in [Rousseau et al., 2002], [Rousseau et al., 2006]. The second method, described in Chap. 4, is a thermal decomposition of an organic phase containing uranyl acetylacetonate (UAA) as precursor, which is added to a mixture of oleic acid and oleylamine which is then heated as a whole up to temperatures around 300°C to induce the precipitation of UO2nanoparticles by thermal decomposition of the UAA. This original procedure was described in [Wu et al., 2006]. To obtain larger amounts of nc-UO2as required, both methods were conveniently adjusted, developed and scaled-up according to the aim needs. The material in the as-produced condition was studied by TEM and XRD. Crystallization and grain-growth kinetics of the synthesized powders as a function of the temperature and time, as well as structure characterization at the different temperatures, follow in Chap. 5and Chap. 6,forthe aqueous controlled precipitation and the organic precipitation, respectively. Tools such as TGA/DTA, XRD, HT-XRD, TEM, NMR, IR and XANES/EXAFS were used for this purpose. The results on thermally treated powders provided useful information for the selection of ideal sintering conditions for the posterior synthesis nc-monoliths. In Chap. 7the consolidation of the green monoliths, as well as different thermal routines for the drying and sintering steps, which represent a challenge for the achievement of crack-free dense specimens, were explored. Optical, mechanical and thermophysical characterization of the sintered bodies followed to verify their aptitude to meet nuclear reactor fuel specifications. Characterization techniques as SEM, indentation, HP-XRD, SAM, thermal diffusivity and melting point were used. Chap. 8summarizes the discussions of the different chapters in a final conclusion of the overall results. Finally Chap. 9concentrates different future recommendations for the project. Some of them have been already initiated during the thesis and presented here, e.g., alternatives for monoliths compaction, study of nc-UO2magnetic properties and out-of-pile simulation experiments. 9 Chapter 1. Introduction 10 Chapter 2 Analytical and characterization techniques 2.1 Electrochemical analysis The instrument used was a SP-50 voltammeter from BioLogic Science Instruments (working ranges 2.8-10 V and 10 μA - 400 mA). 2.1.1 Cyclic Voltammetry (CV) A three electrode system configuration was utilized. A working electrode (~1 mm thick rod in spiral form, composition Pt/Ir 90/10%metal from Heraeus and Fisher type with introduced area=1.885 cm2Fig. 2.1a), an auxiliary or counting electrode (net, composition Pt/Ir 90/10%metal from Heraeus and Fisher type with 0.12 mm net-wire thickness, 38 mm in diameter and 50 mm in height Fig. 2.1a), and a reference electrode (Ag/AgCl InLab® Reference saturated from Mettler Toledo). A 150 mL glass-reactor built for these experiments was also used (Fig. 2.1b). (a) Net, spiral, frit (b) Reactor Figure 2.1: Cyclic voltammetry experimental arrangement. In a cyclic voltammetry (CV) analysis, the current density i(mA/cm2) is plotted versus the applied voltage or potential E(V). Where iis the intensity or current at the 11 Chapter 2. Analytical and characterization techniques a Cu X-ray tube (40 kV, 40 mA), and a Position Sensitive detector Braun covering an angular range of 6°(2θ) and an Anton Paar HTK2000 heating chamber. The alignment of the machine is done with reference material MgO at different temperatures. 2.5 Thermogravimetry/Differential Thermal Analysis (TGA/DTA) Thermogravimetric analysis for the determination of mass changes and decomposition temperatures were carried out with a simultaneous thermogravimetry differential thermal analysis (TGA-DTA) system (NETZSCH Simultaneous Analyzer STA 449 Jupiter). For the tests Al2O3crucibles were employed. For the correction of the gas buoyancy effect, base lines were measured with empty crucibles using the same experimental conditions as for the investigated samples. 2.6 Dilatometry The sintering behaviour of green monoliths was determined by differential dilatometry with a Bähr Thermoanalyse DIL-802 S. The length change of the monolith was measured as a function of temperature relative to the length changes of a parallel reference sample (polycrystalline Al2O3) with similar dimensions. 2.7 Mechanical Characterization 2.7.1 Microindentation The microindentations were performed with a Frank-Finotest hardness-tester according to the standard methods for advanced ceramics, ASTM C1327 for Vickers indentation hardness and ASTM C1326 for Knoop indentation hardness. The form of the these typical indenters is shown in Fig. 7.24. Loads of 1.96 N, 4.90 N and 9.80 N were applied for 15 seconds. For comparison an instrumented microindenter developed in ITU, was also used. In this case, loads of 0.49 N, 0.98 N, 1.96 N, 4.90 N and 9.80 N were applied also for 15 seconds. The acquisition of test data points was at room temperature. At least 3 to 5 different indentations (at different locations throughout the sample) were performed for each load applied and the average value taken for the calculations. Minimum distances between indentations were respected following the respective ASTM methods. Random representative areas of the material were always taken for the tests. The samples were embedded in hard epoxy-resin, followed by grinding and polishing of the test surface. Special holders with two-sides plane-parallel geometry were used to ensure the perpendicularity of the indenter to the sample and, surface and in consequence, symmetrical indentations. Grinding of the samples was carried out with SiC paper (600 to 1200 mesh), as well as a final polishing with diamond suspensions was done, reducing progressively the particle size from 15 μmto1μm. A fine polishing was needed to achieve a perfect flat mirror surface and to avoid the addition of errors in 18 2.8. Thermophysical characterization the determination of the indentation lengths due to surface roughness and imperfections. 2.7.2 High Pressure X-Ray Diffraction (HP-XRD) The compressibility study of the different nano-sized UO2was performed by means of in-situ X-ray diffraction (XRD). The samples were loaded in a Diacell-type membrane diamond anvil cell with 500 μm culet size using pre-indented Re gaskets with 200 μm diameter holes. Pressure was determined using the ruby scale [Piermarini et al., 1975] and Cu equation of state (Cu-EOS). Silicone oil was used as pressure transmitting medium. High pressure X-ray diffraction (HP-XRD) was performed using a modified Bruker D8 x-ray diffractometer with focusing mirror optics installed on a molybdenum rotating anode source (Mo Kα1=0.70926 Å), coupled with a Bruker SMART Apex II Charged-Coupled Device (CCD). The recorded diffraction images were integrated using the ESRF FIT2D software [Rodríguez-Carvajal, 1993]. 2.7.3 Scanning Acoustic Microscopy (SAM) The scanning acoustic microscopy were performed in a collaboration with the group of Prof. Laux as part of a collaboration with IES (Institut d’Electronique du Sud) at University of Montpellier with an acoustic microscope in ITU developed (Fig. 2.2). The device includes a translation stage, micrometric motors, echographic bench, and acoustic focused sensors with spherical lens with an aperture angle of 50°. The samples should have a thickness of about 1 mm. Afterwards the samples are embedded in a resin and polished to obtain a smooth surface. The sample is introduced in an aluminium basked and methanol coupling liquid is poured until the embedded sample is completely submerged. After horizontal alignment with two adjusting screws the acoustic sensor is lowered avoiding to trap any air bubble which could lead to false readings. The sensor is further lowered until a few μm distance from the sample and defocusing is started in order to get an acoustic image. The signal from the sensor is converted to an optical signal to be displayed on a computer screen. 2.8 Thermophysical characterization 2.8.1 Thermal Diffusivity The measurements of the thermal diffusivity are performed in a laser-flash device (LAF I) [Ronchi et al., 1999] inside a lead-shielded glove box with remote manipulators. The sample is heated up (Ar atmosphere of 10−2mbar) in a high frequency furnace to the measurement temperature. A laser pulse is applied to one of the surfaces of the sample. At the opposite surface , the out-temperature perturbation is recorded by a photo-diode pyrometer (0.05°K sensitivity) with an in ITU developed log-amplifier with a rise-time of the order of 50 μs. The experimental set-up and the measurement technique are explained in detail by [Staicu, 2007]. The thermogram is recorded by a 14 bit digitalizer (T=T(t) consisting of several thousands of points) and is analysed by a realistic and accurate mathematical model of the pulse propagation in the sample. The thermal diffusivity and heat losses are 19 Chapter 2. Analytical and characterization techniques (a) (b) Figure 2.2: Overview of the acoustic microscope (in ITU developed) device showing acoustic sensors, coupling liquid holder, sample platform and translation stages. calculated by a numerical fitting method. Correct measurements of thermal diffusivity can be obtained even with samples of small sizes and irregular contours, due to the highly homogeneous probe laser-beam. The precision of the individual measurements is better than 1%. Nevertheless, the accuracy of the measured thermal diffusivity is lower than the precision of the method, being principally determined by sample thickness variations. 2.8.2 Melting Point Same conditions as described in [Cappia et al., 2013] were used. A schematic of the laser heating experimental set-up can be also there seen and here reproduced (Fig. 2.3). The sample is introduced in an autoclave under controlled atmosphere and heated by a 4.5 kW cw Nd:YAG TRUMPFlaser. The power in function of the time profile is programmable with a resolution of 1 ms. The onset of melting is detected by the appearance of vibrations in the signal of a probe Ar+laser reflected by the sample surface (reflected light signal technique) [Manara et al., 2008]. The cooling of the sample occurs when the laser beam is switched off. Thermal arrests corresponding to solidification can then be observed on the thermograms recorded by fast pyrometers. These operate in the visible-near infrared range between 488 nm and 900 nm. The reference pyrometer wavelength is here 655 nm. This was calibrated according to the procedure reported in [Manara et al., 2008][Böhler et al., 2012]. A dense sample of at least several microns in thickness are needed in order that the measurement is not influenced by the sample support. The normal spectral emissivity of urania has been assumed to be equal to 0.83 [Manara et al., 2005][Cappia et al., 2013]. In Fig. 2.4 a picture of the sample melting point set-up, is shown. 20 2.8. Thermophysical characterization Figure 2.3: Laser heating experimental set-up [Cappia et al., 2013]. Figure 2.4: Sample melting point setup. In the yellow area, the nc-UO2pellet hold with three screws is observable. 21 Chapter 2. Analytical and characterization techniques 22 Chapter 3 Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase 3.1 Introduction and principles. 3.1.1 U-stability: environmental studies. The control of used and present uranium mines, spent nuclear fuels and waste repositories installations is an ongoing subject. Related to the surveillance of these sites, the understanding of the dissolution/re-precipitation and transport behaviour of different radionuclides at different pH in aqueous media is essential to the avoidance of the contamination of the groundwater [Ryan and Rai, 1983]. The behaviour in water of U in its various forms, particularly as UO2in crystalline and amorphous states, needs to be understood. In the case of reducing conditions the disposal is facilitated by immobilization of the soluble UVI species by its reduction and precipitation as insoluble UIV in form of UO2, and posterior removal after localization from the aqueous media [Lovley and Phillips, 1992]. The increase in the solubility and mobility of UIV or UVI species due to complexation with chloride [Hennig et al., 2005], carbonate [Suzuki et al., 2006]or sulphate [Hennig et al., 2007] anions, constitutes also a subject of attention for radiological issues. The possible of oxidative remobilization of the nanometer-sized precipitates [Suzuki et al., 2002][Ling et al., 2008] by accidental contact with atmospheric O2[Zhong et al., 2005] or by radiolytical induced oxidation in water [Mennecart et al., 2004], is also matter of study. A variety of parameters influence the systems involving UO2colloids. In particular, the precipitation/solubility of U colloids in aqueous media is affected by two issues. The trend to oxidation from UIV to UVI, as well as the hydrolysis of UIV at very low pH (pH∼1) by complexation of U4+ into Um(OH)(4m−n) nin solution (mU4+ +nH2O←→ Um(OH)(4m−n) n+nH+)[Neck and Kim, 2001]. The presence of the colloids in solution is also function of the degree of acidity of the media. Finally, the colloids aggregation state (crystalline, amorphous), also plays a role in their precipitation/solution trends [Opel et al., 2007][Rai et al., 2003]. 23 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase 3.1.2 U-redox chemistry pertinent to nc-UO2synthesis. The precipitation in aqueous media of uranium colloids species at different pHs has been studied to gain a better understanding of these processes, with the objective to examine the radiological hazards implied in the deep disposal of radioactive waste [Rousseau et al., 2002]. The elementary mechanisms occurring on an atomic scale during fuel oxidation storage have been also studied with actinide colloids [Opel et al., 2007], [Rousseau et al., 2009]. The calculated equilibrium solubility lines of UIV and UVI species in aqueous solution as a function of pH have been reported by [Neck and Kim, 2001] [Fanghänel, Th. and Neck, 2002]. A plot of these results as compiled by [Gil et al., 2010] is shown in (Fig. 3.1). There are several orders of magnitude difference between the solubility of UIV in presence of its crystalline dioxide phase, UO2(c), or in presence of the amorphous form of this phase (hydrated uraninite, UO2·xH2O(am)). Also several orders of magnitude separate the two lasts with the solubility line of UVI in the presence of the crystalline hydroxide phase schoepite, UO2(OH)2(s) (Fig. 3.1)[Gil et al., 2010]. Also valuable experimental data of the oxidation state of U in the precipitates and of the size of the corresponding colloids and their agglomerates, along the solubility line have been reported at different pH and low U-concentrations [Opel et al., 2007][Rai et al., 2003][Fujiwara et al., 2003] [Fujiwara et al., 2005]. The data show precipitates sizes of UO2ranging from 8-13 nm at around pH=1 in the crystalline state, to 80-150 nm at pH>2.5 in the amorphous state (Fig. 3.1). One can conclude that by provoking the precipitation close to the UIV solubility line at lowest pH, the smallest and more crystalline form, of the precipitates can be obtained, avoiding possibly the precipitation of any uranium hydroxide compound. According to the hitherto experience, the synthesis of nc-material using this method results, however, in small quantities of material produced and large quantities of waste. The concentrations of U per batch were of the order of 10−3to 10−3M[Rousseau et al., 2002][Rousseau et al., 2006][Opel et al., 2007] [Rousseau et al., 2009]. A challenge to tackle in this work will be hence to study the feasibility of this synthesis method for nc-UO2species, but extrapolated to quantities of product as high as 1 g per batch or higher so as to be able to produce dense monoliths with such material. In the publications of [Rousseau et al., 2002][Rousseau et al., 2006] [Rousseau et al., 2009], two methodologies to obtain nano-precipitates were used. Two pH ranges were studied, ≤4and≥4, with uranyl nitrate (UO2(NO3)2) solutions dissolved in chloride media with U-concentrations of 0.005 M and 0.03 M, respectively. In both intervals, UO2+xprecipitates with elementary size 20 nm (without excluding agglomeration) and O/U ratio ∼2.19 were obtained [Suzuki et al., 2006] [Rousseau et al., 2002]. In the first method, an aliquot of UVI solution was added to the precipitation reactor where reducing conditions were applied by a galvanostat (constant intensity) at a fixed pH. The dropped UVI was reduced electrochemically to UIV at a fixed pH and UO2+xwas so precipitated. In the second method used by these authors, the UVI-solution was first reduced electrolytically to UIV (to avoid precipitation of UVI compounds which begins at pH≥4) and thereafter it was slowly dropped into the precipitation vessel under reducing conditions, which were kept by application of constant potential. Simultaneously, the pH was held constant by 24 3.1. Introduction and principles. Nanoscale uranitnite precipitates Nanoscale uraninite precipitates (sulphate media) [Gil et al., 2010], method 3, (pH≈4.5-5) Nanoscale uraninite precipitates (sulphate media) [Opel et al., 2007], LIBD, UO2 (c) [Opel et al., 2007], LIBD, UO2·xH2O (am) [Rousseau et al., 2002], method I, UO2.19 [Rousseau et al., 2002], method II, UO2.19 Thermodynamic predictions (solubility limits) [Fanghänel, Th. and Neck, 2002] I=0.5 M [Fanghänel, Th. and Neck, 2002] I=1 M [Fanghänel, Th. and Neck, 2002] I=1 M [Neck and Kim, 2001] I=0.5 M Figure 3.1: Theoretical solubility limits of UIV and UVI species in aqueous solutions [Fanghänel, Th. and Neck, 2002][Neck and Kim, 2001] and experimental determinations for U-sulphate [Gil et al., 2010] and U-chloride solutions [Rousseau et al., 2002] [Opel et al., 2007]. Compounds shown beside each equilibrium line show the precipitated solid phase when these conditions are exceeded. balancing the acidity of the U solution with controlled NaOH additions. Precipitation occurred thence instantaneously. An intensive study of the range of U-concentration and acidity for nc-UO2+x precipitation from electrolytically reduced uranyl nitrate (UO2(NO3)2) solutions is hence endeavoured in the present work, using higher concentrations ranges, and therefore lower pH ranges, following the solubility line of UIV . The use of higher concentrations and their correspondingly lower precipitation pH range, was already suggested but not tested by [Opel et al., 2007]. In their perchlorate system, these authors proposed moving upwards the UIV -precipitation line towards lower pHs as a means to obtain nano-UO2precipitates in its crystalline form but upon diminishing the size of the agglomerates. In the present chapter, the same kind of concept will be applied. 3.1.3 nc-UO2synthesis principles. For the nc-UO2synthesis by controlled precipitation in aqueous media, the starting UO2(NO3)2solution was reduced from UVI to UIV , before being precipitated by adding NaOH to the system. An initial study of the electrochemical reduction by cyclic voltammetry (CV) and an optimization of the different parameters have been also performed. A typical set-up used for the reduction and controlled precipitation, is shown in Fig. 3.2. 25 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase It consisted of a glove-box to house the entire configuration under inert atmosphere and as the first radioprotection barrier for the handling toxic materials. A sevennecked reactor was used where pH-electrode, reference electrode, counter electrode, working electrode, Ar-flow inlet tube, NaOH dosing tube and out-gas neck, were placed. 0.1/02 %'$$$ #$,")!'!&$" &,"$!&$" %%&)*&$& $& &"$ $   0-  # $!&$" #&$" "% &$ 0 Figure 3.2: General set-up for reduction and controlled precipitation in a aqueous media method. A flow chart showing the different steps performed in the process are shown in Fig. 3.3. The solution containing the U-salt was prepared and the pH was adjusted by HCl addition. The pH was kept below 1 to avoid hydrolysis of the U4+ cation (see Sec. 3.1.1) once the reduction of the UVI-solution began. In a second step, the initial UVI-solution was electrochemically reduced at constant potential to UIV , keeping the acidity of the solution below the precipitation pH for the given initial U-concentration in the solution (UIV -solubility line Fig. 3.1). The state of electrolysis of the solution was controlled by continuous CV-checks, at the time that UV-vis absorption tests were performed. After reduction of the solution, precipitation proceeded. Aliquots of NaOH were added into the solution where a series of stepwise precipitations were carried out. In the method reported by [Rousseau et al., 2009] the procedure was the other way around. In that case [Rousseau et al., 2009] aliquots of U-solution (reduced or unreduced depending on the method used) were added in a solution with higher pH where the precipitation was immediately taking place. The quantities of material per day and batch obtained were very small. By changing the procedure, and by increasing at the same time the U-concentrations in the solution (i.e. with respect to the ones 26 3.2. Mother solution preparation. used by [Rousseau et al., 2009] and [Gil et al., 2010]), an increase of the yield of the precipitates was searched. Black nc-UO2-precipitates appeared so continuously until no more UIV -cation was present in the solution. All reduction and precipitation experiments were conducted under anoxic conditions in a glove-box under N2atmosphere (oxygen < 0.5%). Centrifugation of the blackened solution was then performed to collect the precipitates. Removal of the Na+and Cl−species still present on the surface of the wet nc-UO2-precipitates was achieved by repeated washing with deionised water and sonification. Centrifugation after each washing step was needed to separate the precipitates from the water containing the dissolved Na+and Cl−species washed. In the course of the above experiments, the uranyl containing solutions changed its colour from yellow-green, characteristic of the uranium nitrate hexahydrate salt, to green dark colour, once the specie UVI was reduced to UIV . In both cases (oxidised and reduced), the solution had intense colour but no turbidity was observed. Once the first aliquots of NaOH were dropped into the reactor, the green coloured transparent UIV -solution began to acquire turbidity because of the incipient UO2crystals precipitated. This green-turbid colour changed progressively to black with the following precipitation (see Fig. 3.3). The morphology and structure of the obtained nc-UO2were characterized by means of Transmission Electron Microscope (TEM) and confirmed by X-Ray Diffraction (RT-XRD). # #      &$'%& $% → $%       ##&  & ! # Figure 3.3: Controlled massive precipitation in aqueous phase steps. 3.2 Mother solution preparation. The mother solution was prepared by dissolution of UO2(NO3)2powder (CAS: 10102-06-4; yellow green crystals; Fig. 3.4) in deionised water by continuous stirring at 80°C to reach a solution concentration of 500 gU/L. This mixture at room temperature (RT) was diluted in a NaCl 1 M solution to obtain final uranium concentrations in the 27 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase 3.3.2 Electrochemical reduction of the mother solution. 3.3.2.1 Experimental arrangement. The experimental arrangement used for the electrochemical reduction was in essence that shown in the general illustration of the precipitation method in Fig. 3.2.The general principles for the electrolysis are described in Sec. 2.1.2. Three electrodes were so immersed in the prepared solution in a 150 mL glass reactor built for these experiments. The solution containing the U-analyte (UO2(NO3)2) was prepared at different concentrations between 0.02 until 0.5 M in U (depending on the experiment). And an excess of the supporting electrolyte (1 M NaCl or 1 M HCl) was added here, too. The experimental set-up of the reduction cell used is shown in Fig. 3.10,where some modification can be appreciated with respect to one used for the CV-tests (see Fig. 3.5). In the CV experiments, the net electrode was used as counter electrode, but for the present case (reduction of the solution), the net was used as a working electrode (cathode). During the reduction of the pair UO2+ 2/U+4, a larger working electrode area is desired to achieve higher reduction rates, for which the net electrode (with larger surface that of the spiral electrode) was employed for that purpose. The spiral was used as a counter electrode (anode).                 Figure 3.10: Electrochemical reduction set-up. To prevent the re-oxidation of the obtained UIV to UVI because the anodic (counting electrode or spiral) produces oxygen, the anode was introduced in a separated glass tube of 12 mm in diameter, with a glass-frit (40 μm as determined in the CV study) on its bottom. This is seen as a fundamental modification of the approach adopted by [Rousseau et al., 2009] and [Gil et al., 2010] where cathode and anode were immersed in the same bath. The frit allows the passage of the electrolyte but partially avoids the passage of the oxygen to the solution, so that the entire process is more effective 34 3.3. UIV /UVI cyclic voltammetric and spectrophotometric study. speeding the reduction process. Continuous bubbling of the solution with inert gas (Ar) was applied to avoid any traces of O2which could pass through the frit. A humidification bottle was installed between the argon supply and the reactor vessel, so that the Ar was pre-bubbled and saturated with water before entering in the reactor. This ensures dry Ar gas is saturated with water and the losses of liquid during the experiment are minimised. The bubbling together with the stirring reinforced the homogenization of the mixture (stock solution) during the whole experiment. Same as in the CV-tests, all reduction and precipitation experiments were conducted under anoxic conditions in a glove-box under N2atmosphere (O2<0.5%). In a typical experiment, an aliquot of the uranyl (UVI) nitrate stock-solution was diluted in a 1 M NaCl solution to yield a final U-concentration of 0.1 M in a 75 mL batch. Before beginning any electrochemical reduction step, the pH was measured with a pH glass electrode (iEcotrode Plus Metrohm 3 M KCl) in combination with a Titrando 906 measuring instrument from Metrohm. Adjustment with HCl 1 M was pursued to maintain the solution with a pH<1 to avoid the hydrolysis (see Sec. 3.1.1) of the already reduced U4+, before the onset of the precipitation was induced by controlled NaOH-alkalinisation. The reduction of UVI-solution to UIV was performed under a constant potential (-0.3 V vs. Ag/AgCl as determined in the CV study) and the current varied between |-50|to |-5|A. Therefore the calculation of the needed time for the electrolysis was approximated. The reduction was rapid at first but slowed as the ion concentration decreased as would be expected from Ohm’s Law (V=I·R), i.e. at constant voltage, the current is inversely proportional to the resistance. The UVI-solution was proved to be electrochemically reduced to UIV almost entirely. Continuous checks of the reduction status were carried out by CV-tests during the electrolysis of the solution to verify the extent of reduction of the UVI cations amount (on observing the intensity diminution of the UO2+ 2/U+4 cathodic peak). Thus, the reduction was temporarily halted and the typical experimental CV arrangement was adopted in each check (Fig. 3.5). The spiral was disposed outside the glass-frit and put inside the reactor bath in direct contact with the reducing solution. Ar flow and stirrer agitation were stopped to avoid disturbances. Scan speed as high as 0.1 V/s was used as determined in the CV-study. During the reduction, the reactor was showing increasingly evidence of an already reduced green UIV -solution, while inside the glass tube with the frit a yellow UVI- solution was still observed, due to the continuous O2production on the anodic side (spiral). 3.3.2.2 Results and discussion. Fig. 3.11 shows different CV plots recorded at different times during the reduction step. The plots show the decrease of the cathodic peak UO2+ 2/U+4 with increasing electrochemical reduction time, in terms of the current of density (A/cm2) passing between the two inert Pt electrodes at the corresponding potential. The electrochemical resistance of the media increased with time because the ions concentration of the 35 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase species to reduce diminished. The diffusion to the cathode of the ions, which are still not reduced, became hence more difficult, explaining so the decreased cathodic current. Figure 3.11: Cyclic voltammogram 0.1 M U and pH<1. UO2(NO3)2solution in HCl 0.33 M scanned between -0.4 and +1.4 V vs. Ag/AgCl (saturated) at a scan rate of 0.1 V/s. The theoretical reduction rate of the ion UO2+ 2at each moment was calculated with Eq. 2.7. The pH rose from 0.2 to 0.6 at the end of the electrochemical reduction. That could be due to the acid consumption during the electrolytic reduction of the uranyl ion (cathodic reaction; see Eq. 2.1). That could be relevant if a pH>1 was in the end obtained because hydrolysis of UIV is expected. 3.3.3 UV-Vis spectrophotometry of the solution. 3.3.3.1 Experimental. Different Aliquot samples were taken along the electrolysis for ultraviolet-visible spectroscopy (UV-Vis) analysis, which were performed to monitor the valence state of U before and during the controlled potential reduction of UO2(NO3)2in HCl solution (Fig. 3.12). The UV-Vis/NIR spectrophotometer used to record the UV-Vis absorption spectres is described in Sec. 2.2.1. The tint of the solution was changing along the reduction from an initial clear-yellow colour typical of UO2+ 2ion, to a dark-green colour typical of U4+ at the end of the process (see Fig. 3.13 showing the progressive colour change of the U-solution upon its reduction). 3.3.3.2 Results and discussion. The absorption bands in the range from 375 to 500 nm, which are characteristic to the absorption of uranyl (UVI) ions, gradually decreased and finally disappeared with 36 3.3. UIV /UVI cyclic voltammetric and spectrophotometric study. Figure 3.12: Change in visible absorption spectra for the reduction of 0.1 M U and initial pH<1. UO2(NO3)2solution in HCl 0.33 M. The theoretical reduction rate of the ion UO2+ 2 at each moment was calculated with Eq. 2.7. increasing the electrolysis reduction time. At the same time, the absorption bands from 400 to 700 nm, which are consistent with absorption typical peaks for UIV (426, 492, 548 and 646 nm), appeared with stronger intensity as the time of electrochemical reduction of the solution increased. That was in agreement with the intensity decrease of the UO2+ 2/U+4 cathodic peak observed in the CV tests of the solution (Fig. 3.11) with the time of reduction. This result strongly supports the fact that UO2+ 2in the acidic solution was almost fully reduced to UIV at the Pt electrode. The progress of the electrochemical reduction, as analysed by spectrophotometry, is shown in Fig. 3.12. In Fig. 3.14, the intensity peak decrease by both cyclic voltammetry (e.g. UO2+ 2/U+4 cathodic peak, Fig. 3.11) and UV-spectrophotometry (e.g. UO2+ 2absorption peak    Figure 3.13: U-solution at different steps during electrochemical reduction 37 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase at 412.43 nm, Fig. 3.12), are plotted as a function of the percentage of UO2(NO3)2 electrochemical conversion (analysis performed on the same aliquots of solution). Correspondence between both methods in the determination of the UO2+ 2/U+4 conversion, is very satisfactory.                                Figure 3.14: Intensity peak decrease by both CV (e.g. UO2+ 2/U+4 cathodic peak, Fig. 3.11) and UV-spectrophotometry (e.g. UO2+ 2absorption peak at 412.43 nm, Fig. 3.12) as a function of the percentage of UO2(NO3)2electrochemical conversion. 3.4 Precipitation and separation of the UO2- nanocrystals. 3.4.1 Introduction remarks. The precipitation from the electrochemically reduced UIV -solution (0.5 M U for the highest concentration) was achieved by gradual alkalinisation of the solution following as close as possible the theoretical solubility limit line of UIV species in aqueous media Fig. 3.1. This was pursued to avoid significant agglomeration of the precipitates (UO2(c)orUO 2·xH2O(am)), which was faired to occur if the system entered in the region of frank supersaturation, as it could be induced by uncontrolled brusque alkalinisation (pH-increase much beyond the equilibrium line). Indeed, the theory of precipitation from supersaturated solutions and a number of dedicated experimental works indicate that the size of the nuclei-precipitates is inversely proportional to the supersaturation degree, while the number of nuclei is directly proportional to it (homogeneous precipitation) [Lifshitz and Slyozov, 1961] [Bristow et al., 2001][Wu et al., 2008][Maeda et al., 2009]. However, under consideration of kinetic aspects (e.g. coagulation rates), also considerable number of studies exist indicating that the supersaturation degree is a key factor triggering the agglomeration of precipitates, with the evidence found that the larger the supersaturation degree the larger the size of the agglomerates formed (including both homogeneous and heterogeneous precipitation) [Yu et al., 2007][Claassen and Sandenbergh, 2006] 38 3.4. Precipitation and separation of the UO2-nanocrystals. [Zumstein and Rousseau, 1989][Packter, 1958][Sarig et al., 1978]. The quantity of NaOH theoretically needed to reach a given pH can be calculated as the sum of [OH−]molsA+[OH−]molsB, where the first term Aindicates the number of OH mols needed to increase the initial pH to the wished final pH, as given by: [OH−]molsA=[H+]molsInitial −[H+]molsEnd (3.1) where [H+]molsInitial =10 −pHInitial and [H+]molsEnd =10 −pHEnd and the second term Bindicates the number of OH mols needed to precipitate the species U+4 as UO2, as given by the reaction: U4+ +4OH−↔UO2+2H2O(3.2) and where [OH−]molsB=4·U4+mols (3.3) This theoretically amount of NaOH needed to set a given pH was hence estimated and was taken into account as indicative value for the subsequent precipitation step. 3.4.2 Experimental steps. The precipitation was performed in the same reactor used for the electrolysis experiments. The arrangement used for this step is shown in Fig. 3.15. No potential was applied during the precipitation but the atmosphere conditions were maintained in the same way as in the electrochemical reduction step: i.e. the arrangement was kept inside a glove-box under anoxic conditions (N2atmosphere with O2<0.5%)and dynamic Ar gas flow (passed through a humidification bottle to saturate it with water before entering in the reactor to avoid the dryness of the solution), was applied. The pH at the beginning of the precipitation, was below 0.5. A series of stepwise precipitations between pH<0.5 and 3 were then carried out by addition of 3 M NaOH solution at a rate of 10-20 μl/min, as controlled by a dosage instrument (Titrando 906 from Metrohm). The slow alkalinisation was automatically stopped when the solution reached the pH desired, and was re-started when the pH evolved backwards (through precipitation, Eq. 3.3) and was again below the set pH. Black nc-UO2-precipitates appeared around pH∼1 and the solution began to look turbid from this point onwards. The pH variation during the precipitation step was 39 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase            Figure 3.15: Precipitation of nc-UO2set-up. monitored continuously. An example of this variation with time for the case of the highest U-concentration used in the experiments (0.5 M) is shown in (Fig. 3.16). In the displayed case, a constant dosage rate of the alkalinising NaOH-solution of approximately 1.2 ml/h was used. Small pH-drifts occurred because of the formation of nc-UO2crystals. In addition to this, on reaching the pH the value of 1, a main jump or abrupt slope increase was observed in the pH vs. tcurve (Fig. 3.16), which revealed the main precipitation of UO2crystals in the solution has already occurred, when the pH surpassed this precipitation threshold. The quantity of NaOH theoretically needed to reach the set pH was calculated as previously indicated as the sum of [OH−]molsA+[OH−]molsB. The NaOH-solution was slowly added at constant rate and the pH was monitored (Fig. 3.16). The NaOH additions stopped automatically after approximately 30 h of precipitation on reaching the pH the value 3. After this point, no further essential modifications of the pH were registered during at least 3 days more (>100 h after initiation of the process), indicating that the precipitation had been already almost completed when the pH achieved the value 3. Centrifugation (3500 rpm and 30 min) of the blackened solution was performed to gather the precipitates. Na+and Cl−species were present in the final solution because of the initial additions of HCl to acidify solution and the posterior additions of NaOH used to precipitate the UIV -species as oxide in the solution. The wet nc-UO2- precipitates hold these impurity species on its surface, which will precipitate as small NaCl crystals once the final product dries. Therefore, the wet nc-UO2-precipitates were re-dispersed in deionised water and sonificated for diluting and washing out the possible Na+and Cl−species concentration present. Subsequent centrifugation, to 40 3.4. Precipitation and separation of the UO2-nanocrystals. Figure 3.16: Controlled nc-UO2precipitation from a electrochemically reduced UIV -solution 0.5 M U in HCl 1M. pH monitoring vs. NaOH addition and time. separate the nc-UO2-precipitates from the water containing the dissolved Na+and Cl−species, was done. This washing-centrifugation step was performed 5 times using a volume of 40 ml of deionised water pro washing operation and a partial charge of 1.25 g nc-UO2each tube (4 tubes were centrifuged in parallel). For the highest U-concentrations of (0.5-0.6 M), around 10 g of nc-UO2precipitates were obtained in total from each precipitation batch. 3.4.3 Spectrophotometry of the solution. The UV-spectrophotometer was used during the precipitation process to identify the decrease in U concentration in the solution (see Fig. 3.17 where the absorption band at 647.04 nm typical from U+4, has been monitored). With increasing of the pH (addition of NaOH to the solution), increasing of nc-UO2precipitates occurred and reduction of the UIV in solution diminished. The change in the visible absorption spectra could not be observed effectively till the end of the precipitation because of the turbidity of the solution. A picture of the reduced green UIV -solution before to begin the precipitation and the black final solution containing the nc-UO2-precipitates, is shown in Fig. 3.18. With the solubility constants for UO2(c) and UO2(am), and the stability constants for UIV hydrolysis reported by [Guillaumont et al., 2003], a graph with the speciation of all uranium entities present at different pH in the solution as well as the solubility lines for UO2(c) and UO2·xH2O (am) can be represented, as it was undertaken in Fig. 3.19. The speciation curves add knowledge of what is happening in the solution as the pH is increased in the precipitation step. The yellow line represents the piece of solubility/precipitation line followed during precipitation. The yellow circles represent the experimental precipitation points observed in the UV-absorption spectra (Fig. 3.17) having in account the different intensities obtained. These experimental precipitation point fall all on the theoretical precipitation line. 41 Chapter 3. Synthesis of nc-UO2by controlled massive precipitation in Aqueous phase Figure 3.17: Change in visible absorption spectra of the typical UIV band between 630 and 665 nm, of a controlled nc-UO2precipitation from a electrochemically reduced UIV -solution 0.1 M U in HCl 1 M and pH<1 (left). Diminishing of intensity with increasing of the pH because the precipitation of the U+4 in solution as nc-UO2(right). (a) UIV -electrochemically reduced solution 0.5 M U in HCl 1M and pH<0.5. (b) ∼10 g of black nc-UO2precipitates Figure 3.18: U-solution before and after the precipitation. -16 -14 -12 -10 -8 -6 -4 -2 0 0.00 0.20 0.40 0.60 0.80 1.00 0 1 2 3 4 5 6 log[(U(IV)] fraction pH U4+ U(OH)22+ U(OH)3+ U(OH)31+ U(OH)4 Figure 3.19: Uranium Speciation at different acidic media and solubility lines for UO2(c) and UO2·xH2O (am) represented with the constants data by [Guillaumont et al., 2003]. Yellow circles represent experimental points. Yellow line represents the piece of solubility line followed during precipitation. 42 3.5. Characterisation of the as-produced nanocrystals. 3.5 Characterisation of the as-produced nanocrystals. The morphology and structure of the obtained nc-UO2particles were characterized by means of transmission electron microscope (TEM) and the structure confirmed by X-ray diffraction (XRD). 3.5.1 Precipitates morphology and composition. Qualitative composition analysis of the precipitates obtained was carried out by TEM. The characteristics of the instrument used, as well as the preparation of the analysed specimens are described in Sec. 2.3.2. Fig. 3.21 shows the TEM image of nc-UO2synthesised by the presently described method. The average precipitate size was of 3.9(8) nm, as obtained from the size distribution (Fig. 3.20). The particles showed agglomerates of 50 nm (Fig. 3.21a). However, very often separated nanoparticles exist in suspension in the original solution, but one sees agglomerates on the TEM grid, due to the preparation of the TEM sample. Figure 3.20: Size distribution histogram from TEM measurements of nc-UO2synthesized by precipitation in aqueous media. Diameter average size of 3.9(8) nm diameter average size. The collected black precipitates presented the typical fcc fluorite structure of UO2. The selected area electron diffraction (SAED) pattern (inset of Fig. 3.21c) revealed the polycristallinity of the material with the fcc structure. The calculated interference fringe spacing in the HRTEM image (Fig. 3.21c) was about 0.315 nm, which was in agreement with the interplanar distance of the [111] plane in the fcc fluorite structure (0.3153 nm for the UO2standard 00-041-1422-ICCD). 3.5.2 Precipitates crystal structure. The crystal structures and crystal size of the as synthesised nc-UO2observed in the TEM analysis were confirmed by XRD. The characteristics of the instrument used are described in Sec. 2.4.1. The crystal structure of the nano-precipitates was 43 Chapter 4. Synthesis of nc-UO2and nc-ThO2by a precursor thermal decomposition in Organic phase Surfactants inside the reactor bind on the synthesized crystals forming an organic capping layer that protects and stabilizes the crystals against flocculation or excessive growth. Larger molecules surfactants which provide greater steric hindrance, as well as surfactants that cap stronger to the surface of the crystals, diminish the reagents incorporation rate in the nuclei and as a consequence their final size. Particles growth can be stopped (before the finishing of the reagents) by fast cooling of the reactor [Murray et al., 2000][Wang et al., 2003][Murray et al., 1993][Peng and Peng, 2001] [Qu et al., 2001][Burda et al., 2005]. To collect the particles a precipitation is induced by addition of a non-solvent which is characterized to be partially miscible with the media of the reaction but has no interaction with the capping agents. The dispersion is destabilized and the solution becomes turbid due to flocculation of the crystals. Centrifugation accelerates the process and crystals are collected on the bottom of the centrifuge tubes. Afterwards the liquid is decanted and the crystals collected. The material obtained is formed by the nc-material and the organic capping layer which permits its re-dispersion in an organic solvent [Murray et al., 2000][Wang et al., 2003][Murray et al., 1993] [Peng and Peng, 2001][Qu et al., 2001][Burda et al., 2005]. The organic solution phase decomposition route has been widely used for the synthesis of metal-oxide nanocrystals. In this project the synthesis of nc-UO2in organic media was developed following the second approach above described and the method reported by [Wu et al., 2006] where uranyl acetylacetonate (UAA Fig. 4.1a)is used as precursor in a mixture of long chain solvents as oleic acid (OA), oleylamine (OAM) and octadecene (ODE) which are stable at high temperatures where this kind of metal-oxide nanocrystals are normally formed [Willis et al., 2007]. (a) UAA (b) ThAA (c) ThAc Figure 4.1: Precursor molecular structures used in the organic synthesis. The decomposition of the UAA precursor and the amine (OAM) is followed by a reduction of UVI (UAA) to UIV (UO2) at higher temperatures. OAM might act as reducing agent [Wu, 2008]. A condensation reaction between OA and OAM occurs ending in the formation of N-(cis-9-octadecenyl)oleamide (OOA Fig. 4.2-left)and water, and is almost terminated before the beginning of the nucleation, as reported by [Wu et al., 2006]. OOA is described there as tuner of the reaction intermediate steps to form the UO2nanoparticles, where free UO2units and clusters have interaction with the amide (R-NH+ 3) and the carboxylic (R-COO−) groups (see Fig. 4.2-right) [Wu, 2008]. 50 4.1. Introduction and principles. 29     2 9 Figure 4.2: Oleic Acid (OA), Oleylamine (OAM) and N-(cis-9-octadecenyl)oleamide (OOA) obtained after the condensation reaction together with water [Wu et al., 2006](left).Intermediate steps of the nc-UO2synthesis where free UO2units and clusters interact with the formed OOA (right) [Wu, 2008]. However, no OOA but just an oleate was found bonded through chelating bidentate interaction on the surface of the nc-UO2[Wu et al., 2006]. A strong attraction appears between the oleate and the metal group making a very compact monolayer around the nanocrystals inhibiting their growth and protecting them against agglomeration (see Fig. 4.3).   Figure 4.3: Oleate as capping ligand bonded through chelating bidentate interaction on the surfaceofthenc-UO 2. 51 Chapter 4. Synthesis of nc-UO2and nc-ThO2by a precursor thermal decomposition in Organic phase 4.2 Experimental. In the present work the precursor uranyl (UVI)-acetylacetonate (UAA) for UO2 nanocrystals and thorium (ThIV )-acetylacetonate (ThAA) or thorium (ThIV )-acetate (ThA) for ThO2nanocrystals (see Fig. 4.1), were dissolved in a solution of oleic acid (OA Fig. 4.2) and 1-octadecene (ODE) at 110°C under continuously stirring. Oleylamine (OAM Fig. 4.2) was added to the mixture and was heated at an average rate of 20°C/min until 310 to 350°C was reached (depending on the experiment). Afterwards the growth solution was aged for 1-6 h (depending on the experiment) and cooled to room temperature. The experimental set-up involved a heater, a small-round vessel, a cooler and a continuous Ar flow. A scheme as representation of the set-up is shown in Fig. 4.4. The solution preparation as well as the reaction steps were conducted under anoxic conditions in a glove-box under N2atmosphere (O2<0.5%). Posterior washing of the oxide-crystals, as well as centrifuging was performed. As a final product, nc-precipitates of UO2and ThO2redispersible in organic solvents were obtained. A picture of the different steps is shown in Fig. 4.5. 111 2 3 4567 8 9 11 2 3 4567 8 9 11 Thermocouple Ar Stirrer Heating mantle Stirring plate N2Glove-Box Water cooling nc-UO2Synthesis Reactor Figure 4.4: Arrangement for the organic thermal decomposition method. Reduction of surfactant quantities with respect to the metal content in comparison with [Wu et al., 2006] was made. Posterior scale-up of the method from initial quantities of about 0.1 g of nc-UO2as reported by [Wu et al., 2006], to quantities of about 2 g of nc-UO2achieved here, was done. The same method was also extrapolated for the synthesis of nc-ThO2. 52 4.2. Experimental.      #&"!   ' %""!$&!  '   #""! Figure 4.5: Steps in the thermal Decomposition of UAA in organic media. 4.2.1 Solution preparation. 4.2.1.1 UO2nanocrystals preparation. In this method UAA (UO2(CH3COCHCOCH3)2CAS: 18039-69-5 yellow orange crystals from Ibilabs based on depleted uranium 235U 0.3-0.4%-238U 99.6%Fig. 4.1a) was used as a precursor. Indeed, 4 g of UAA were dissolved in a solution of 15 mL OA (9-Octadecenoic acid CAS: 112-80-1 99%from Sigma Aldrich) and 27 mL ODE (Octadecene-1 CAS: 112-88-9 90%from Sigma Aldrich) inside a 100 mL three-necked glass reactor under continuous stirring. The three necks of the vessel were occupied by an Ar flow, a condenser and a temperature sensor connected to a temperature controller. The reactor was placed on a heating mantle and the solution was slowly heated up to 150°C and maintained at this temperature during 20 min. Afterwards the solution was left to cool to room temperature and 21 mL OAM (1-Amino-9-octadecene, CAS: 112-90-3, 70%, Sigma Aldrich) were added to the mixture. The solution was then slowly at heated 100°C under continuous stirring during 15 min. During this heating the solution passed from an initial turbid yellow state to a good mixed, transparent orange state (see Fig. 4.6). (a) (b) Figure 4.6: (a) UAA + ODE + OA at RT (before applying any temperature). Turbid yellow solution. (b) UAA + ODE + OA + OAM at RT (after stirring at 100°C). Transparent orange solution. 53 Chapter 4. Synthesis of nc-UO2and nc-ThO2by a precursor thermal decomposition in Organic phase As solvents ODE, OA and OAM were introduced in the reactor. The long chained alcohol (OA) was used as surfactant, and as reducing agent a long chained amine (OAM). The boiling points for the OA, OAM and ODE are 360, 364 and 315°C respectively. 4.2.1.2 ThO2nanocrystals preparation. In this case ThAA (Th(CH3COCHCOCH3)4CAS: 17499-48-8 from Ibilabs colorless crystals Fig. 4.1b) or ThA (Th(C2H3O2)4CAS: 13075-28-0 from Ibilabs colorless crystals Fig. 4.1c) were used as precursors. The steps followed for the preparation of nc-ThO2were similar to those used for the synthesis of nc-UO2. In this case 0.60 g of thorium acetylacetonate (ThAA) or thorium acetate (ThA) were dissolved in a solution of 3 mL OA and 4 mL ODE inside a 50 mL three-necked glass reactor under continuous stirring. The three necks of the vessel were as before occupied by an Ar flow, a condenser and a temperature sensor connected to a temperature-controller. The reactor was placed on a heating mantle and the solution was slowly heated up to 150°C and maintained at this temperature during 20 min. Afterwards the solution was left to cool to room temperature and 4 mL OAM were added to the mixture. The solution was then slowly heated at 100°C (temperature high enough to allow a good mixing but not high enough to provoke the reaction) under continuous stirring during 15 min. The solution passed from an initial turbid-white to a good mixed transparent white solution. 4.2.2 Reaction step. 4.2.2.1 UO2nanocrystals production. Once the OAM was also mixed in the solution, everything was heated with an average rate of 25°C/min until 300°C. The mixture was so progressively changing its colour from transparent orange-brown at 100°C to turbid dark brown at 190°C, to brown-black at 200°C, until finally turning to completely black at 250°C (see Fig. 4.7). Small explosions inside the reaction vessel occurred at that temperature. Once the temperature plateau at 300°C was reached, the solution was aged at this temperature for 60 min. During this ageing time, small explosions continued inside the vessel. The solution preparation as well as the reaction steps, were conducted under anoxic conditions in a glove-box under N2atmosphere (O2<0.5%). Furthermore, a continuous Ar gas flow was supplied inside the reactor vessel in order to keep an inert atmosphere. Even so, a lot of water and oxygen molecules are already inside the experimental system (UAA contains two water molecules and two oxygen atoms). The small explosions observed up to 250°C could be due to an effect of oxygen reacting with the solvents during thermal decomposition, at these temperatures. The UAA and the OAM were thermally decomposed and subsequent reduction of the UVI (UAA) to UIV (UO2) occurred using OAM as a reducing agent, as suggested by [Wu, 2008]. As a final material, nc-UO2were obtained. The surfactant bonded to the nanocrystals forming a lipophylic surface on them. This capping would avoid the agglomeration of the nanocrystals and would allow the post-collection of the crystals in an organic solvent. Finally the solution was left to cool to room temperature. 54 4.3. Precipitation and separation of the nanocrystals. Figure 4.7: U-solution during the reaction step at different temperatures (160, 190 and 250°C). 4.2.2.2 ThO2nanocrystals production. Once the OAM was added to the mixture, everything was heated at 25°C/min up to 300°C. In the experiments where higher temperatures were achieved (330°C), ODE was substituted because its boiling point at 315°C, and just OA and OAM were used as solvents for the reaction. The mixture progressively changed its colour from turbid white at 100°C to yellow transparent at 200°C, and to orange transparent during the ageing time at 300°C and to pale yellow transparent at the end of the ageing (see Fig. 4.7). No explosions occurred during any of the ThO2-experiments. Different ageing times between 1 to 5 h were tested. Figure 4.8: Th-solution during the reaction step at different temperatures. 4.3 Precipitation and separation of the nanocrystals. 4.3.1 UO2nanocrystals recovery. The particles coated and stabilized with a non-polar layer, were therefore soluble in highly non-polar solvents such as ODE or toluene. The precipitation of the nanocrystals was induced by adding a non-solvent, a mixture of hexane/acetone (1/4) to the aged solution. The mixture was sufficiently apolar as to selectively precipitate the relatively non-polar particles without precipitating out the ODE and excess OAM and OA (which are taken by the hexane fraction). Pure acetone would be too polar to be directly miscible with ODE. Instead, the use of a mixture with a moderately non-polar solvent such as hexane permits their addition being miscible in the solution without forming a secondary immiscible liquid layer. 55 Chapter 4. Synthesis of nc-UO2and nc-ThO2by a precursor thermal decomposition in Organic phase Figure 4.9: Typical final UO2solution after the reaction step and before precipitation. Once the mixture of hexane/acetone was introduced, the dispersion was destabilized and the solution became turbid due to the flocculation of the black nc-UO2precipitates. The separation was forced by centrifuging the new mixture at 3500 rpm (EBA 20 centrifuge from Hettich) for 10 min. The particles accumulated at the bottom of the centrifuging tubes and the supernatant liquid containing the remaining organics (ODE, OA and OAM) was taken away. New hexane/acetone was introduced in the tubes and the centrifuging step repeated. This procedure was reiterated until the extracting liquid was clear (see Fig. 4.10). After cleaning the nanocrystals several times with the mixture hexane/acetone, the black nc-UO2were re-dispersed in an organic solvent such as hexane or toluene. 2 Figure 4.10: Precipitation, cleaning and recollection in an organic solvent (hexane) of the nanocrystals of UO2. 4.3.2 ThO2nanocrystals recovery. The ThO2nanocrystals where obtained in the same way, but more repeated washing and centrifuging steps were needed to obtain white ThO2precipitates at the end of the separation white ThO2precipitates. The hexane/acetone (1/4) mixture was introduced in the pale yellow reaction solution, which became white turbid. Then the crystals began to flocculate (see Fig. 4.11) and after centrifuging a pale-orange gel appeared on the bottom on the tubes. After several washing steps, however, white ThO2crystals were finally obtained. The ThO2where redispersed in an organic solvent (hexane or toluene). 56 4.4. Characterisation of the as produced nanocrystals. Figure 4.11: Precipitation, cleaning and recollection in an organic solvent (toluene) of the nanocrystals of ThO2. 4.4 Characterisation of the as produced nanocrystals. The morphology and structure of the obtained nc-UO2particles were characterized by means of transmission electron microscope (TEM), dynamic light scattering (DLS), and the structure confirmed by X-ray diffraction (XRD). 4.4.1 Precipitates morphology and composition. The morphologies and dimensions of the samples were revealed by TEM. The characteristics of the instrument used, as well as the preparation of the analysed specimens are described in Sec. 2.3.2. The organic route led to high-quality monodispersed UO2 nanocrystals and ThO2rod-shaped nanocrystals. 4.4.1.1 UO2nanocrystals morphology. In Fig. 4.13, the TEM image of synthesised nc-UO2by thermal decomposition in organic phase is shown. The average precipitate size was of 4.9(3) nm, as obtained from the size distribution (Fig. 4.12). The collected black precipitates presented the typical fcc fluorite structure of UO2 and were polycrystalline, as shown by the rings of the SAED in Fig. 4.13c.The selected area electron diffraction (SAED) pattern (inset of Fig. 4.13c) revealed the polycristallinity of the material with the fcc structure. The calculated interference fringe spacing in the HRTEM image (Fig. 4.14) was about 0.315 nm, which was in agreement with the interplanar distance of the [111] plane in the fcc fluorite structure (0.3153 nm for the UO2standard 00-041-1422-ICCD). Dynamic light scattering (DLS) measurements in hexane medium during 80 s of the dispersed sample of nc-UO2obtained by thermal decomposition in organic phase, yielded an hydrodynamic average size of 3.7(1) nm with a polydispersity index (PI) of 0.139. Characteristics of the instrument are shown in Sec. 2.2.2. The size distribution histogram is shown in Fig. 4.15). The DLS size values correspond well with the values observed by TEM. 57 Chapter 4. Synthesis of nc-UO2and nc-ThO2by a precursor thermal decomposition in Organic phase Figure 4.12: Size distribution histogram from TEM measurements of UO2nanoparticles synthesized by thermal decomposition of UAA in organic media. Diameter average size of 4.9(3) nm. (a) The scale bar is 50 nm. (b) The scale bar is 20 nm.  (c) The scale bar is 5 nm. The inset of the figure shows selected area electron diffraction (SAED). Figure 4.13: TEM micrographs of UO2at low resolution, showing an assembly of nanocrystals, and at high resolution, revealing lattice imaging of the nanocrystals. 4.4.1.2 ThO2nanocrystals morphology. In Fig. 4.13 the TEM image of synthesised nc-ThO2by thermal decomposition in organic phase is shown. Single crystalline 1±0.5 nm in diameter ThO2-nanorods resulted upon precipitation. There is still not full understanding for the reason of the obtained shape. The precipitate material obtained was quite different in geometry from the one obtained from the UAA precursor, which was instead almost perfectly spherical in shape. Different ageing temperatures (290-330°C) and times (1-6 hours), as well as different precursors (ThAA and ThA) were tested, but the results gained were always ThO2-nanorods. The collected white precipitates presented the typical fcc fluorite structure of ThO2 and were polycrystalline, as shown by the rings of the selected area electron diffraction (SAED) pattern (inset Fig. 4.16c) and the HRTEM image (inset Fig. 4.17) showing 58 4.4. Characterisation of the as produced nanocrystals.    Figure 4.14: TEM micrograph of UO2at high resolution, revealing lattice imaging of the nanocrystals and interplanar distances. Figure 4.15: Size distribution histogram from DLS test of UO2nanoparticles synthesized by thermal decomposition of UAA in organic media. Hydrodynamic average size of 3.7(1) nm. lattice spacings of an individual nanocrystal. The calculated interference fringe spacing in the HRTEM image (inset Fig. 4.17) was about 0.322 nm, which was in agreement with the interplanar distance of the [111] plane in the fcc fluorite structure (0.3232 nm for the ThO2standard 00-042-1462-ICCD). (a) HAADF STEM micrograph. The scale bar is 20 nm. (b) The scale bar is 10 nm.      (c) The scale bar is 20 nm. The inset of the figure shows selected area electron diffraction (SAED). Figure 4.16: STEM and TEM micrographs of ThO2nanorods. 59 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route DTA signal and is shown in Fig. 5.1. Upon heating, an almost invisible endotherm with a maxima around 155°C occurs. This is accompanied by a slight weight loss of about 2.3 wt%until 207°C, which are most likely attributed to water desorption. Unfortunately, we were not in a position to identify the desorption products, as the integration of a mass spectrometer in a glove-box environment for this purpose presented too many technical difficulties (particularly subsequent maintenance) to be overcome. The mass loss continues with a further 3%until 600°C, which can be related to crystallization (perfectionism of the UO2fcc-structure). This coincides with the exothermic peak in the DTA at 280°C which reflects this heat realise due to the crystallization, on further reaction to desorb water. However, there is no reason to believe that the weight loss did not include as well some loss of oxygen due to material’s reduction. A deeper analysis on the lattice parameter and crystal growth under inert and reducing atmosphere has been in the following sections performed to confirm the latest. Figure 5.1: TGA and DTA signal for nc-UO2until 1200°C under Ar/5%H2. 5.3 Lattice parameter and crystal growth in neutral atmosphere. The crystal growth, lattice parameter of the nc-UO2, has been investigated under inert conditions (static He atmosphere) using in situ HT-XRD. The effect of temperature on the crystallite size, which is a fundamental parameter in the sintering process has been analysed. The in situ HT-XRD patterns were acquired with an instrument described in 2.4.2. The temperature range explored was 30 to 1200°C at a heating rate of 5°C/min under a static He atmosphere. The evolution with temperature of the nc-UO2XRD pattern is shown in Fig. 5.2 (results presented in [Jovani-Abril et al., 2011]). The observed reflections are assigned to UO2-fcc phase structure and to Pt phase corresponding to the heater plate, plus an impurity peak at around 2θ= 26°. The pattern is similar to the one reported at room 66 5.3. Lattice parameter and crystal growth in neutral atmosphere. temperature (RT) by [Rousseau et al., 2009]. They reported the impurity peak as Na polyuranate, coming from precipitation of U(VI) with NaOH. The effect of temperature on the peaks can be observed more clearly in Fig. 5.2-right, which displays the evolution of two main peaks (111) and (200) of the UO2structure. A shift in the peak position to lower angles is there observed, possibly related to a thermal lattice parameter expansion. Figure 5.2: In situ HT-XRD patterns of nc-UO2under He (left). The typical UO2and Pt (from the heating plate) Bragg peak positions are also marked. The arrow on down-right side of the graph shows a residual impurity which disappears with temperature. Evolution of (111) and (200) peaks of UO2cubic structure as a function of temperature (right) [Jovani-Abril et al., 2011]. An effect of the temperature is seen in the width of the peaks which decreases with increasing temperature while the intensity of the peaks increases. Since the contribution of instrumental broadening is independent of the temperature, the broadening at lower temperatures is mainly related to the crystallite size and strain present in the material, as well as increase of the structural order. Both contributions, crystal size (proportional to cos−1θEq. 5.1) and strain (proportional to tanθ Eq. 5.2), have different angular dependences, and are so separable. A study of those influences has been in the following performed. D=Kλ βcosθ (5.1) e=β 4·tanθ (5.2) where Dis the average crystallite size, Kis a constant (0.87-1) that depends upon the particle shape and the Miller-indexes (hkl), λis the wavelength of the radiation, βis the full peak width at half maximum, θis the Bragg angle and eis the strain. 67 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route 5.3.1 Grain growth as a function of temperature under neutral atmosphere. The effect of temperature on the crystallite size, which is a fundamental parameter in the sintering process has been here analysed. The crystallite size of the nc-UO2has been determined by the XRD Rietveld refinement (see Sec. 2.4) of the Bragg peaks, and used also to characterize the microstructure of the material. From these results it is possible to generate a universal representation of the crystallite size as a function of temperature (XRD measurements done at temperature with the HT-XRD instrument under static He atmosphere) and reported in Fig. 5.3. Even though, this information is taken as universal, slight deviations from it may occur, in particular due to dwell times and temperature ramps, but more importantly due to the atmosphere of static He during thermal treatment. Figure 5.3: Evolution of the nc-UO2crystallite size in function of the temperature [Jovani-Abril et al., 2011]. At room temperature, the size of the crystallite was about 4 nm after precipitation Sec. 3.5.1, which is in agreement with previous study by [Rousseau et al., 2009]. The crystallite size change with temperature shows two domains separated at 700°C (see Fig. 5.3). Below that temperature, there was a weak influence on the crystallite size which evolved from 2 to 7 nm (measured in situ at temperature). Above 700°C, the size of the crystallite increased quasi linearly but drastically with temperature, reaching a size about 73 nm at 1200°C. 5.3.2 Lattice parameter and linear thermal expansion coefficient as a function of temperature. The crystal growth of the sample under inert conditions (static He atmosphere) using in situ HT-XRD, have been already described. In addition, the variation of the lattice parameter versus crystal size and temperature, as well as data on the linear thermal expansion, have been reported and compared to bulk material UO2. The crystal structure of the precipitates have been, as the crystallite size, determined by Rietveld 68 5.3. Lattice parameter and crystal growth in neutral atmosphere. refinement, taking into account the whole 2θrange. In Fig. 5.4a the lattice parameter obtained as a function of temperature (XRDs measured at temperature and under static He atmosphere) and its derivative (Fig. 5.4b), have been also determined by the XRD Rietveld refinement of the Bragg peaks, and represented together with the nc-UO2size evolution to observe its dependence. Also calculated lattice evolution of non-stoichiometric standard UO2+xfor different O/U ratios due to only thermal expansion, have been represented for comparison. The lattice parameter of a non stoichiometric UO2+xis linked to the oxygen content by the relations of [Lynds et al., 1963]. Also the lattice parameter was corrected in function of temperature with the [Fink, 2000] relations reflected in Eq. 5.3. 2≤O/U ≤2.125 a(nm)=0.54705 −0.0094·O/U 2.1725 ≤O/U ≤2.250 a(nm)=0.54423 + 0.0029·(9 −4·O/U)(5.3) The main change in the lattice parameter occurs between RT and 600°C and for crystal sizes below 6 nm. The lattice parameter suffers a strong expansion increasing steeply from 0.5417(1) nm to 0.5492(1) nm at 300°C, and then decreases to 0.5485(0) nm. Above 600°C, a linear evolution of the the lattice parameter with temperature is observed. (a) (b) Figure 5.4: a.) Lattice constant and crystallite size variation (curves only as a guide to eye) of nc-UO2in function of temperature, from in situ HT-XRD measurements under static He atmosphere in comparison with lattice evolution in function of temperatures of standard UO2 for different O/U ratios obtained by the relations of [Lynds et al., 1963], due to only thermal expansion. b.) Relative crystallite size and lattice parameter vs. temperature (curves only as a guide to eye). From, the lattice parameter curve in function of temperature, the extrapolated linear trend up to 600°C is shown by the straight dotted line and described by the following equation a(nm)=0.54439 −0.00007·T, obtaining a value of the lattice parameter at 20°C of 0.5445 nm, closer to the lattice of UO2bulk at RT. 69 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route Relating the lattice parameter found in this study with the [Lynds et al., 1963] relations, a stoichiometry of about UO2.18 up to 600°C to around UO2.17 up to 900°C has been found, which is similar to the one reported by [Rousseau et al., 2009]. So the nanocrystallites stabilize at O/U 2.17-2.18 at temperatures above 600°C, or in other words, at particles sizes higher than 6 nm. A XPS data showing a contribution of U(VI) and U(IV) in the precipitated particles was reported by [Rousseau et al., 2009]. Based on the lattice parameter determined by XRD, they concluded hence that the stoichiometry was UO2.19, thus describing the system as nc-UO2+xand is very similar to our conclusion, too. However, remark that Eq. 5.3 and other equivalent relations are valid for bulk compounds; their applicability to nanocrystals may be still open to proof. Fig. 5.5 displays the linear thermal expansion (LTE) and the thermal expansion coefficient (LTEC) of nc-UO2as a function of the temperature. The LTE at temperature Twas calculated using the relation: LTE =(aT−a0)×100 a0(5.4) where aTis the lattice parameter at temperature Tand a0is the lattice parameter at 20°C. The LTEC was calculated by differentiating the thermal expansion curve aT versus T with respect to the temperature T: LTEC =1 a0×δaT δT (5.5) The LTE of the nc-UO2is in general higher than the one for bulk-UO2[Martin, 1988] for all the interval of temperatures, with a jump at 300°C and a posterior stabilization above 600°C as one could already predict from the lattice parameter representation in function of temperature (see Fig. 5.4a). The LTEC is initially higher for nc-UO2than for bulk-UO2for temperatures below 400°C and tends to stabilize above 600°C with a value of 12·10−6°C−1in agreement with the value for the LTEC of bulk-UO2. The oscillatory trends observable for LTEC in nc-UO2at T <900°C can be attribute to transitory oxidation-reduction effects. At T ≥600°C it is clear that the oxygen content of the material stabilizes (at O/U 2.17-2.18) (Fig. 5.4a), at the time that the lattice expansion coefficient meets the value of the reference bulk phase (large grain) (Fig. 5.5). For crystal sizes >6 nm the nanocrystalline material meets the thermal-expansion behaviour (i.e. thermal expansion coefficient) of bulk (large-grained) UO2. This behaviour was already observed in the representation of the relative lattice parameter (see Fig. 5.4b), and it shows once more that the ab-normal nano-effects in the material are only to be expected for particle sizes below few tens of nanometer. If Fig. 5.6 the patterns comparison of nc-UO2as-produced (a = 0.5417(1) nm), nc-UO2at 1200°C (a = 0.5521(0) nm) and nc-UO2at RT after thermal treatment at 1200°C (a = 0.5473(0) nm) measured in situ in the HT-XRD instrument under static He atmosphere, is shown. There is not explanation for the value seen for the nc-UO2measured at 1200°C under static He atmosphere: an O/U ratio close to UO2.17 (see Fig. 5.4a)andthea 70 5.3. Lattice parameter and crystal growth in neutral atmosphere. Figure 5.5: Linear thermal expansion (LTE) and linear thermal expansion coefficient (LTEC) of the nc-UO2(curves only as a guide to eye) in comparison with data of bulk-UO2from [Martin, 1988]. Figure 5.6: Patterns comparison of nc-UO2as-produced, nc-UO2at 1200°C and nc-UO2at RT after thermal treatment at 1200°C measured in situ in the HT-XRD instrument under static He atmosphere. lattice parameter of a = 0.5521(0) nm. In contrast, the same thermal treated sample measured after cooling at RT, shows a value of 0.5473(0) nm (UO2.00) very similar from the typical for bulk-UO2(0.547 nm). The peaks of the nc-UO2at RT after the thermal treatment at 1200°C recover the typical position for standard UO2(see Fig. 5.6). So, a recovering of the crystal structure has been achieved with temperature under 71 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route static He atmosphere from the initial lattice parameter value of 0.5417(1) nm from the nc-UO2as-produced to 0.5473(0) nm after thermal treatment at 1200°C. It might be related to the static He atmosphere, which resulted in a thermodynamic equilibria between H2,H 2O and O2in the gas phase, favouring H2during the cool down, thus this might be the cause. 5.3.3 Lattice strain evolution as a function of temperature. The mean strain, e, in the material was determined by Rietveld refinement using the software [HSP-PAN, 2011] and used to characterize the deformation state of the material. The analysis of the strain by Rietveld refinement method is based on the change of the profile parameters, compared to a standard sample. Those are depending on the instrument settings used for data collection and on the profile function used for the refinement [HSP-PAN, 2011]. The crystallite size change with temperature, already shown in Sec. 5.3.1,was characterized for two steps separated at 700°C. Below that temperature, there was a weak influence on the crystallite size which evolved from 2 to 7 nm (measured in situ at temperature). Above 700°C, the size of the crystallite increased quasi linearly but drastically with temperature, reaching a size of about 73 nm at 1200°C. In the evolution of the lattice strain release with temperature, two steps can be observed (Fig. 5.7). The first step is visible at T <300°C where the strain is at its maximum, which could be related to the dehydration step (water molecules attached to the nanocrystals and to the related binding-strength) visible in the TGA/DTA (Fig. 5.1). The second step is visible in the range 300-700°C where the strain decreases, which could be related to additional loss of water and oxygen loss (O/U stabilization) and structure consolidation. Up to 700°C, the internal strain vanishes and the crystallite-size starts to grow. It is interesting to note that the complete release of the strain in the nc-UO2 coincides with the onset of the starting of the crystallite growth. So the crystallite growth seems to be limited by the presence of the lattice strain. From RT until 700°C the thermal energy is totally used to remove the strain and up to that temperature no strain is remaining and the energy is used for the growing of the nanocrystals. 5.4 Lattice parameter and crystal growth under reducing conditions. The local structure has been investigated for the nc-UO2as-produced and after thermal treatment under reduction conditions (Ar/5%H2), by X-Ray Diffraction (XRD), MAS-NMR, IR and X-ray Absorption Spectroscopy (XAS), and compared to bulk-UO2as a reference. A combination of X-ray Absorption Near Edge Structure (XANES) and Extended X-ray Absorption Fine Structure (EXAFS) was used. 72 5.4. Lattice parameter and crystal growth under reducing conditions. Figure 5.7: Crystallite size and strain of nc-UO2in function of temperature. Measurements done at temperature under static He atmosphere (curves only as a guide to eye). 5.4.1 Crystal size and lattice parameter evolution as a function of temperature as probed by XRD. As it has been seen from the HT-XRD, the lattice constant (and crystal size) of the material in the cooled state (at RT) after reach different maximum temperatures is needed (see Fig. 5.6). This allows separation of the thermal expansion contribution in the high-temperature values to obtain cleaner curves for thermal expansion vs. temperature and lattice dimension vs. crystal size. At the outset of this study, a higher concerning about the control of the O/U ratio, which is not simple, appeared. The structural investigations presented in the following, attempt to eliminate this issue as the samples were heated in Ar/5%H2to ensure that the O/U = 2.00. The impact of the heat treatment on the microstructure of nc-UO2at different annealing temperatures was then studied under reducing conditions (dynamic Ar/5%H2atmosphere). As-produced or RT, 600°C and 1200°C. The corresponding XRD data are given in Fig. 5.8. A heating rate of 5°C· min−1under dry Ar/5%H2and annealing for 15 minutes were used. At 600°C the major mass loss has already occurred and no water traces should be present in the material as it has been observed in the TGA/DTA (Fig. 5.1). As a reference, a standard UO2.0(μm crystal size) sintered at 1600°C under Ar/5%H2for 6 hours and measured during the same measuring campaign, was also used. The measurements indicated a well crystallized single cubic phase with a fluorite structure (Fm-3m). No evidence for orthorhombic or other phase was found. From the Rietveld refinement of the measured diffractograms, the lattice parameters as well as the size of the particles were deduced. Reduction of the nc-UO2+xtowards nc-UO2 after annealing at 600°C and 1200°C was expected because the high sensitivity of the lattice parameter, a, to changes in the oxidation state of U in the hyperstoichiometric range, O/M >2. Samples treated at increasing temperature (RT, 600°C and 1200°C) 73 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route with a short isothermal hold of 15 min, exhibit an increase in the lattice constant of 1.02%. Progressing crystallization of the nc-UO2leaded to periodic ordering of its atom and the lattice parameter changed from 0.5417(1) nm at RT, to 0.5431(0) nm and 0.5472(0) nm at 600°C and 1200°C, respectively. The crystal-size increased from 3.79 nm (≃4 nm) as-prepared, to 9.3 nm (≃9 nm) and 82.16 nm (≃82 nm) following treatments at 600°C and 1200°C, respectively (Fig. 5.8). Figure 5.8: XRD patterns of reference UO2and aqueous route nc-UO2(as-produced, after thermal treatment under Ar/5%H2at 600°C and 1200°C) A comparison of the crystal size and lattice parameter for the samples treated at 600°C and 1200°C under dynamic Ar/5%H2atmosphere (measurement after cooling) with those under static He atmosphere measured at temperature (see Fig. 5.4a), is provided in Table 5.1. No difference in the crystallite size was obtained at 600°C under either atmosphere. But a notable change is observed at 1200°C, where a crystallite size of 73.39 nm under He has been observed and a size of 82.16 nm under Ar/5%H2. Major differences have been seen in the lattice parameters as a function of the atmosphere used without ignoring the fact that the values under He were measured at temperature. These difference in the lattice disappeared once the sample measured at temperature is measured after cooling, as it has been seen in Fig. 5.6. In the evolution of the lattice strain, e, a release was again observed with increasing temperatures. After annealing at 600°C under Ar/5%H2, just the half of the strain was present being totally released at 1200°C, as it also happen under He atmosphere. Fig. 5.9 shows TEM images of the nc-UO2(about 4 nm size) particles as-produced and after being thermal treated at 1200°C (about 120 nm size) under Ar/5%H2.The TEM size observed after treatment is in good agreement with the average value of 82.2 nm obtained of the Rietveld refinement (Fig. 5.8). 74 5.4. Lattice parameter and crystal growth under reducing conditions. Table 5.1: Crystal size and lattice parameter for aqueous route nc-UO2treated at 600°C and 1200°C under two different atmosphere (Ar/5%H2and He atmosphere) Ar/5%H2atm.aHe atm.b cryst. size lattice param. strain cryst. size lattice param. strain (nm) (nm) (%) (nm) (nm) (%) nc-UO2RT 3.79 0.5417(1) 0.792 1.99 0.5420(10) 1.149 nc-UO2600°C 9.30 0.5431(0) 0.391 5.36 0.5483(5) 0.861 nc-UO21200°C 82.16 0.5472(0) 0.026 73.39 0.5521(0) 0.004 a.) Measurement after cooling. b.) Measurement at temperature (HTXRD meas.). (a) nc-UO2as-produced. (b) nc-UO2after 15 min at 1200°C. Figure 5.9: TEM images for the nc-UO2as-produced and after thermal treatment under Ar/5%H2. 5.4.2 O/M ratio as a function of temperature as probed by XANES. XANES was used to determine the oxidation state of U cations and the corresponding molar fractions and the O/U ratios were derived. The normalized XANES spectra and the first derivate at the U-L3absorption edge of three different heated nanocrystalline UO2samples (nc-UO2at RT, 600°C and 1200°C) are shown in Fig. 5.10, together with the reference spectra of UIV O2. The experimental features are specified in Sec. 2.2.3. The associated energies of the inflection point at absortion edge and of the white-line (WL), as well as the energy shift (ΔE) and the estimated oxidation states derived from this study, are given in Table 5.2. A simple observation of the XANES spectra at the U-L3edge immediately shows a trend with increasing temperature and as xdecreases (UO2+x). The peak of the WL shifts slightly to lower energies and increases in intensity, and the within the XANES regions increase. The amplitude decrease with the increasing temperature of thermal treatment showing a higher structural order of these samples. For the samples as-produced (RT) and at 600°C, there is a significant difference of shape compared to the UIV O2reference, i.e. presence of a shoulder on the high 75 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route Figure 5.13: Experimental () and fitted data (−) for the nc-UO2annealead at 1200°C of (left) k3-weighted spectra and (right) Fourier Transform at the U-L3edge. Figure 5.14: U-O1and U-U1bond distances in function of the annealing temperature (under reducing conditions) and finale size of the nc-UO2sample (curves only as a guide to eye). 5.4.3.2 Local structure and valence state as probed by MAS NMR. The Hahn-echo 17O MAS spectra acquired for various annealing times (600°C, 650°C, 700°C, 800°C and 1200°C) are presented in Fig. 5.15 and some fits in Fig. 5.16 and Fig. 5.17. The experimental features are specified in Sec. 2.2.4. The spectrum of the 200°C annealed sample was acquired at 10 kHz (as a 1.3 mm probe was not available at this time) therefore its analysis has to be done apart from the whole series and has been not here represented. In fact, as these compounds are paramagnetic the 17O shift depends on the spinning speed which induces a slight heat of the sample. Nevertheless, another study on bulk-UIV O2(done in ITU to be published) shows that the variation in the peak position is not very important between 10 kHz and 55 kHz (∼10 ppm, small compared with the peak broadening). Thus, the shift of 1075 ppm (200°C) can be compared with that extracted for these series (Fig. 5.18). At the contrary, the broadening of the peak cannot be compared with that of the others heat treatment as the Full Width at Half Maximum (FWHM) will decrease with spinning speed due to the removing of paramagnetic shift anisotropy. For the next temperature of 600°C, a broad peak of 2478 ppm has been identified 82 5.4. Lattice parameter and crystal growth under reducing conditions. Figure 5.15: Stack of the 17O MAS-NMR for nc-UO2annealed at five different temperatures. at 962 ppm (Fig. 5.16a). By increasing the temperature of only 50°C (at 650°C; Fig. 5.16b), the MAS spectrum of the sample exhibits two peaks at 932 and 723 ppm. On the static spectrum acquired with more scans a third peak (not here represented) can be identified at nearly 960 ppm. The position of this peak can be relatively compared with that of the spinning spectrum while the FWHM cannot. At 700°C (Fig. 5.17a), there are still three peaks and the peaks at 732 and 738 ppm are now sharpest. For the two last heat treatments, 800°C (Fig. 5.17b) and 1200°C (Fig. 5.17c), only two peaks with very closed shifts are identified. Two trends are observed in the plots of the shifts and the FWHM as a function of temperature presented in Fig. 5.18. Under 700°C, the shift and the FWHM decrease abruptly while above this temperature they are relatively constant. It can be noticed that the shift extracted from the spectrum acquired at 200°C is very different from that of the whole series. Due to the broadness of the peak, the presence of more than one species cannot be excluded. A similar trend was observed by XRD for the evolution of the lattice parameter as a function of temperature (Fig. 5.4). In fact, it increase steeply under 700°C, then there is only a week evolution. Therefore, it seems that under this temperature there are important changes on long and short range. Three different oxygen environments can be identified. The first one corresponds to oxygens having a chemical shift of nearly 900 ppm (named peak C, green peak). These peaks have been identified up to 650°C. At 650°C, even if the peak is sharpest, the shift is very similar to that of the 600°C sample and therefore the 17O will be considered as the same type of species. Due to the broadness of the peak, it is tempting to attribute this one to 17O in a disordered environment. But, XRD has shown that 83 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route 17O Chemical Shift (ppm) -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000 * * (a) 600°C 17O Chemical Shift (ppm) -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000 (b) 650°C 17O Chemical Shift (ppm) -2000 -1000 0 1000 2000 * * (c) 700°C Figure 5.16: Characteristic fits for the spectra of the samples annealed at 600°C, 650°C and 700°C (∗= spinning sidebands; peak A=black;peakB=blue;peakC= green) 84 5.4. Lattice parameter and crystal growth under reducing conditions. 17O Chemical Shift (ppm) -2000 -1000 0 1000 2000 * * (a) 700°C 17O Chemical Shift (ppm) 400 500 600 700 800 900 * * (b) 800°C 17O Chemical Shift (ppm) 640 660 680 700 720 740 760 780 800 (c) 1200°C Figure 5.17: Characteristic fits for the spectra of the samples annealed at 700°C, 800°C and 1200°C (∗= spinning sidebands; peak A=black;peakB=blue;peakC= green). 85 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route for this range of temperature the size of the UO2is around 10 nm (Fig. 5.8 under Ar/5%H2). Moreover, previous experiments done on zeolites [Zhang et al., 1999]have shown that there is an increase of the linewidth of the quadrupolar nucleus (in that case 27Al) with decreasing size of nanoparticles. Indeed, according to [Casabella, 1964] for well-crystallized samples the NMR linewidth is greatly influenced by the quadrupolar coupling constant which is due to the local electric field gradients in the sample. The strong surface energy existing on the surface of the small particles will lead to more strain in the lattice, consequently causes the broadening of the main peak of the quadrupolar line. Thus, it can be proposed that this type of oxygen correspond to the 17O in nanocrystals UO2. Nevertheless, due to the broadness of the peak, presence of other 17O species cannot be excluded. Figure 5.18: Evolution of the 17O shift (left) and of the full width at half maximum (right) as a function of crystallite size for various temperatures. The two second types of species appear clearly from 650°C (Fig. 5.18). There is one sharp and one broad (they will be named peaks Aand Brespectively thereafter). These peaks were fitted using a Lorentzian for A(black peak on Fig. 5.16 and Fig. 5.17) and a Gaussian for B(blue peak on Fig. 5.16 and Fig. 5.17) and could be respectively attributed to 17O in a crystalline phase and in a more disordered one. It can be added that these two peaks are also observed at 1200°C meaning that even for this temperature it is not stoichiometric UO2contrary to that suggested by XRD (Fig. 5.8). With increasing temperature (increasing of crystallite size), the FWHM of the two peaks decrease (Fig. 5.17). As previously explained, this is the signature of such crystallite size evolution. Moreover, the shift of peak Areached a minimal value of 717 ppm at 1200°C. This one is close of the 717 ppm found for UIV O2-bulk. Its FWHM is of 5 ppm and corresponds to that of the crystallite having a size about 80 nm. This 86 5.4. Lattice parameter and crystal growth under reducing conditions. value is still slightly bigger than the 3 ppm found for UIV O2-bulk. Hence, with the shift of peak Aone can say that the environment around the oxygens corresponding to the biggest crystallite size (80 nm) is very close to that of UIV O2-bulk. Based on the FWHM, one can say that to observe the signal of crystalline UO2a size above 80 nm should be reached. This confirms that nc-UO2are obtained and this is consistent with the lattice parameter observation of 0.5472 nm (Fig. 5.8). 5.4.3.3 Local structure and valence state as probed by FTIR. Several samples at the key temperatures were analyzed under the Fourier Transform Infrared (FTIR) spectrometer (Alpha FT-IR Spectrometer from Brucker; Sec. 2.2.5). The infrared (IR) spectra recorded for nc-UO2as-produced (RT), at 200°C, 600°C and 1200°C under Ar/5%H2annealed, as well as UIV O2reference spectra, are shown in Fig. 5.19. Inthecaseofnc-UO 2as-produced (RT), four peaks in the range 400-4000 cm−1 were observed. The absorption band for the U-O vibration in UO2shows up below 400 cm−1which is out of the detection limit of the equipment. The peak at 1625 cm−1 can be assigned to the bending vibration of H-O-H bonds of the coordinated water [Fujita et al., 1956], [Sailaja et al., 2002]. This peak disappears already at 200°C. That would be in agreement with the TGA (Sec. 5.2), were nearly no loss of weight was observable after 600°C (see Fig. 5.1). Something similar occurs to the broad absorption peak at 3400 cm−1, which can be ascribed to the asymmetric and symmetric stretching vibrations of the H-O-H bonds of coordinated water. However, this peak reduces for the 200°C and is nearly disappeared at 600°C annealed sample. So at 600°C all the OH would be theoretically gone taking into account the peaks at 1625 cm−1and 3400 cm−1. However, the peak at 880 cm−1corresponding also to OH groups, stay still present at 600°C, totally disappearing at 1200°C (Fig. 5.19). The same occurred for the peak at 638 cm−1corresponding to more oxidised species (UO2+x)[Kim et al., 2009]. That could be an artefact due to the small size still present at 600°C (9 nm) (see Table 5.1). At 1200°C the IR spectra looks like the one for the UIV O2reference and grains are about 82 nm (Table 5.1). That is also in agreement with the results obtained by XANES where a different electronic structure at 600°C was observed, meanwhile at 1200°C a similar structure to bulk-UIV O2was found (see Table 5.2). Also EXAFS is characterized for a poor ordering at 600°C but entirely matching with the bulk-UIV O2oscillation pairs at 1200°C (see Table 5.3). In the NMR analysis all the additional oxygen sites disappeared once at 1200°C anneal but one which could be due to a surface effect. 87 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route Figure 5.19: Infrared spectra recorded for nc-UO2as-produced (RT), at 200°C, 600°C and 1200°C under Ar/5%H2annealed. Reference spectra for UO2is also represented. Inside amplification of the infrared spectra showing the disappearing of the peaks at 1625 cm−1and 3400 cm−1with increasing annealing temperatures. 88 5.5. nc-UO2long-isothermal grain growth as probed by XRD under neutral and reducing conditions. 5.5 nc-UO2long-isothermal grain growth as probed by XRD under neutral and reducing conditions. The as-produced nanocrystals presented a certain disorder and lattice strain which disappeared with the anneal of the samples, evolving to ordered crystalline structures as temperature increased. This has been proved for shorter time heat treatments under inert and reducing atmospheres (Table 5.1). Normally, ceramic heat treatments to obtain high dense monoliths finish up with grain sizes above 200 nm. Hence, isothermal grain growth kinetics at a given temperature is an essential component to determine the final grain size, and in consequence to evaluate the performance of such an innovative nuclear fuel after prolonged insertion at high temperatures. Although nuclear fuel operates at about 500°C at the periphery it can reach 1200°C at the fuel pellet centre. For that reason, it is fundamental to examine the behaviour of nano-fuel microstructure as a function of the time in a possibly wide temperature range, to ensure lack of disproportionate grain growth even at the highest temperature in play. A study of the thermal growth of those nanocrystals over the whole range of initial microstructures from amorphous to fully nanocrystalline was performed. Based on Fig. 5.3, the critical temperatures 500°C, 700°C, 900°C and 1200°C were selected. Isothermal grain growth data at these temperatures was determined from in-situ HT-XRD measurements under static He atmosphere and annealing times from 0 to 50 h with a heating rate of 5°C/min. The data are shown in Fig. 5.20. Independent sample treatments in separate furnaces under dynamic Ar/5%H2for dwelling times of 50 h, 100 h and 200 h, were also performed. For each isothermal dwell temperature one new sample was chosen. Comparison of the size and lattice parameter obtained for the both routes, are shown in Table 5.5. Figure 5.20: Isothermal grain growth of nc-UO2under He static atmosphere. For each isothermal dwell temperature one new sample was used. Measurement was done in situ in the HTXRD device at temperature. All the curves have been fit using the Eq. 5.6. For the one at 1200°C a fit using the Eq. 5.7 was also done. The in situ HT-XRD determination of the crystallite size was done under static He atmosphere. The experimental installation did not permit the use of Ar/5%H2gas in 89 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route the HT-XRD chamber, which would have been necessary to preserve the O/U ratio of the sample at a value around 2. The measurements were taken at temperature. The crystallite size was determined according to inverse of the width of the first diffraction peak [111], which was monitored during 50 h while the sample was kept at constant temperature. From these measurements information on crystal growth as a function of temperature and time was obtained (Fig. 5.20). Posterior XRD-measurement on consideration of the whole diffractogram after 50 h dwell time at temperature and once the sample was cooled down, was also done (Table 5.5). This final measurement provided data on crystal size without thermal broadening, but characteristic of the O/U final ratio that was obtained under He atmosphere. The measurement of the ex-situ annealed samples was done under a dynamic Ar/5%H2atmosphere which ensured a final stoichiometry UO2.0(UIV ). The measurement was always done after-cooling. Fig. 5.21 shows how grain growth takes place in the first hours of isothermal hold for the temperatures at 500°C, 700°C and 900°C until a stable average grain size below 100 nm is established at the applied temperature, at which time crystal growth ceases. This kind of self-limited grain growth was reported by [Rupp et al., 2006]and is described by the following relaxation function: G−G0=(GL−G0)·(1−e−t τ) (5.6) where Gis the average grain size, G0is the initial grain size, GLis the limited grain size (when the grains stop to grow) and τis the relaxation time (time needed to reach the GL).ThevaluesofGLand τobtained from the fitting described by the relaxation function Eq. 5.6, are presented in Table 5.4. Figure 5.21: Isothermal grain growth of nc-UO2under He static atmosphere. For each isothermal dwell temperature one new sample was used. Measurement was done in situ in the HT-XRD device. All the curves have been fit using the Eq. 5.6. The diffusion coefficient, Di, have been so calculated by substitution of the values G0, GLand τabove determined, in the corresponding approximation by [Rupp et al., 2006] (Eq. 5.7). Afterwards, the activation energy of diffusion, Qdiff , has been obtained as a function of isothermal dwell temperature using the Arrhenius law dependence 90 5.5. nc-UO2long-isothermal grain growth as probed by XRD under neutral and reducing conditions. [Löffler and Johnson, 2000]: Di=(GL−G0)2 4·τ=const ·e− Qdiff kBT(5.7) where kBis the Boltzmann constant (kB=8.6173324 · 10−5eV ·K−1). The different parameters obtained for the fitting at each temperature, are shown in Table 5.4. Table 5.4: Parameters obtained from the fits following the relaxation equation Eq. 5.6 for the samples annealed under He static atmosphere during 50 h. T(°C) G 0(nm) GL(nm) τ(h) Di(m2/s) R2 nc-UO2500°C 5.41 15.51 9.65 7.34e−22 0.72 nc-UO2700°C 5.41 48.21 4.10 3.10e−20 0.87 nc-UO2900°C 5.41 91.99 0.90 5.80e−19 0.90 nc-UO21200°C 5.41 363.55 18.13 4.91e−19 0.66 Starting grain size (G0), limited grain size (GL), relaxation time (τ), diffusion coefficient (Di)and (R2) correlation coefficient obtained from the fitting. The determined kinetics parameters depend on the temperature. The Divalues increase with temperature, meanwhile the relaxation time (obtained from the fitting with Eq. 5.6) diminishes with it (Fig. 5.22). Differently, for the sample annealed at 1200°C a much higher relaxation time (see Table 5.4) has been obtained, not following the decreasing trend with temperature observed for the rest of the dwell temperatures examined (Fig. 5.22). For this sample the grains seem to continue to grow following the generalized grain growth equation (Eq. 5.8) as already observed by [Rupp et al., 2006] for temperatures above 1100°C and expressed as: Gn−Gn 0=kn·t(5.8) where nis the growth exponent and knis a rate constant (kn=k0·e− Qdiff RT with k0= constant, Qdiff = activation energy of diffusion, R= gas constant and T= absolute temperature). For the nc-UO2annealed at 1200°C a grain growth exponent nof 2.36 was obtained with a correlation coefficient of 0.77 using the Eq. 5.8. The fitting curve is shown in blue colour in Fig. 5.20. A final grain size of about 350 nm after 50 h is also under this mechanism obtained. This is in agreement with the range of values of traditional grain growth mechanisms (n=2-4), and in particular the parabolic grain growth mechanism (n=2.36≃2) [Rupp et al., 2006]. The fitting line shown in pink colour, with a correlation coefficient of 0.66, corresponds to the relaxation function above described (Eq. 5.6), which better fits the grain growth kinetics at the lower temperatures examined (T≤900°C). From Fig. 5.23 an activation energy of diffusion of 0.93 eV with a correlation coefficient of 0.90 was determined for the temperature range 500-1200°C (considering 91 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route In the k3-weighted EXAFS spectra (Fig. 5.11a), the oscillations and their amplitude increased with thermal treatment. The 4 nm as-precipitated sample was very difficult to fit with a pure fluorite structure, as the fit were non stable and the data noisy (treated k-range = 3-8 Å−1). The intensity of the FT (Fig. 5.11b) was very low limiting the interpretation of the coordination shell to U-O1. Observing the EXAFS results in Table 5.3, the data were heavily dampened at RT where a large value for the DW factor was found, meaning a significant static disorder for the nc-UO2as-produced. Shorter distance for the oxygen shell (U-O bond length) was clearly observable for the nc-UO2at RT which did not correspond to any U-oxide. According to the shape of the first FT peak, it looked like there were two or three U-O distances instead of one. This was consistent with the observed lattice contraction (0.5417 nm) from XRD at RT (see Table 5.2). The 9 nm sample (600°C anneal) showed an intermediate ordering with oscillations clearly identified and extending to k = 9 Å−1. The intensity of the FT was also low for this annealed sample, limiting the fitting and interpretation of the coordination shell to U-O1and U-U1together with UO2fluorite structure. Still a large value for the DW factor was found (Table 5.3). Shorter distance was also present for the oxygen shell (U-O bond length) in comparison with the reference-UIV O2. However the U-U1 bond length was closer to that of the bulk-UIV O2, suggesting that the U-U1lattice was more ordered than the O anion sublattice. The U-U1lengths were consistent with the less lattice contraction (0.5431 nm), in comparison with the nc-UO2as-produced, as probed by XRD at 600°C anneal. Ultimately, at 1200°C and 82 nm, EXAFS oscillations were similar, if not entirely matching, those of the bulk-UIV O2indicating same fcc-structure consolidation and substantial particle growth, both observed in XRD measurements (Fig. 5.12). Both shells were well fitted with Fm-3m structure for this sample (Fig. 5.13) and very similar distances to reference UIV O2structure could be observed according to the FT (k-range treated = 3-12 Å−1). That was in agreement with the similarity for the XRD data for the annealed sample at 1200°C and the bulk-UIV O2(Fig. 5.8). Also consistent with the XANES (see Fig. 5.10) showing no different oscillation from the fluorite structure. The Hahn-echo 17O MAS spectra acquired for various annealing times (600°C, 650°C, 700°C, 800°C and 1200°C) were presented in Fig. 5.15 and some fits in Fig. 5.16 and Fig. 5.17. Two trends were observed in the plots of the shifts and the FWHM as a function of temperature presented in Fig. 5.18. Under 700°C, the shift and the FWHM decreased abruptly while above this temperature they were relatively constant. It can be noticed that the shift extracted from the spectrum acquired at 200°C was very different from that of the whole series. Due to the broadness of the peak, the presence of more than one species cannot be excluded, as already assumed in the EXAFS study. A similar trend was observed by XRD for the evolution of the lattice parameter as a function of temperature (Fig. 5.4). In fact, it increased steeply under 700°C, then there was only a week evolution. Therefore, it seemed that under this temperature there is important changes on long and short range. Three different oxygen environments could be identified. The first one corresponds to oxygens having a chemical shift of nearly 900 ppm (named peak C, green peak). These peaks have been identified up to 650°C. Due to the broadness of the peak, it was tempting to attribute this one to 17O in a disordered environment. But, XRD 98 5.6. Results and discussion. has shown that for this range of temperature the size of the UO2was around 10 nm (Fig. 5.8 under Ar/5%H2). The strong surface energy existing on the surface of the small particles will lead to more strain in the lattice, consequently caused the broadening of the main peak of the quadrupolar line. Thus, it could be proposed that this type of oxygen corresponded to the 17Oinnc-UO 2. Nevertheless, due to the broadness of the peak, presence of other 17O species could not be excluded. The two second types of species appeared clearly from 650°C (Fig. 5.18). There was one sharp and one broad (named peaks Aand Brespectively). These peaks could be respectively attributed to 17O in a crystalline phase and in a more disordered phase. With increasing temperature (increasing of crystallite size), the FWHM of the two peaks decreased (Fig. 5.17). As previously explained, this was the signature of such crystallite size evolution. Moreover, the shift of peak Areached a minimal value of 717 ppm at 1200°C. This one was close of the 717 ppm found for UIV O2-bulk. Its FWHM was of 5 ppm and corresponded to that of the crystallite having a size about 80 nm. This value was still slightly bigger than the 3 ppm found for UIV O2-bulk. Hence, with the shift of peak Aone could say that the environment around the oxygens corresponding to the biggest crystallite size (80 nm) was very close to that of UIV O2-bulk. Based on the FWHM, one can say that to observe the signal of crystalline UO2a size above 80 nm should be reached. This confirmed that nc-UO2were obtained being consistent with the lattice parameter observed of 0.5472 nm, as well as the O/M ratio obtained in the XANES analysis (Table 5.2). Several samples at the key annealing temperatures were analyzed under the FTIR spectrometer. The IR spectra recorded for nc-UO2as-produced (RT), at 200°C, 600°C and 1200°C under Ar/5%H2annealed, as well as UIV O2reference spectra, were shown in Fig. 5.19. In the case of nc-UO2as-produced (RT), four peaks in the range 400-4000 cm−1could be observed. They could be assigned to the bending vibration of H-O-H of the coordinated water, and to a possible more oxidised state (UO2+x). All of them diminished with annealing. The peaks assigned to the H-O-H of the coordinated water, totally disappeared at 600°C. That was in agreement with the TGA, were nearly no weight of loss was observable after 600°C (see Fig. 5.1). However two of the peaks finally disappeared at 1200°C. That could be an artefact due to the small size still present at 600°C (10 nm) (see Table 5.1). Hence, at 1200°C the IR spectra looked like the one for the UIV O2reference and grains were about 80 nm (Table 5.1). That was also in agreement with the results obtained by XANES where a different electronic structure at 600°C was seen, meanwhile at 1200°C a similar structure to bulk-UIV O2was found (see Table 5.2). Also EXAFS was characterized for a poor ordering at 600°C but entirely matching with the bulk-UIV O2oscillation pairs at 1200°C (see Table 5.3). Isothermal evolution of the synthesized nc-UO2was then performed. Isothermal grain growth at a given temperature is an essential component to evaluate the grain growth kinetics, and in consequence the performance of such an innovative nuclear fuel. Differences in the grain growth behaviour between the micro- and nano-grain form of the same material, have been already reported [Moelle and Fecht, 1995] [Natter et al., 2000][Natter et al., 2001][Rupp et al., 2006]. In the first case the material follows the general growth equation, but in the nano-grain case, the grain grows until a critical time when the grain growth remains constant as described by 99 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route the relaxation function Eq. 5.6. However, when the nano-grain material is treated above a determined temperature (1100°C [Rupp et al., 2006]), the grain growth follows again the general growth equation (Eq. 5.8). This is in agreement with the results here observed. For the temperatures of 500°C, 700°C and 900°C and a static and inert atmosphere of He, the grain growth took place in the first hours of isothermal hold until a stable average grain size was established at the applied temperature, at which time crystal growth ceased (Fig. 5.21). For the isotherm at 1200°C and a static atmosphere of He, the material had a continuous growth not reaching a constant grain value in the first 50 h (Fig. 5.20). From Fig. 5.23 an activation energy of diffusion of 0.93 eV to 1.25 eV was obtained. Diffusion can occur along the grain boundary, or it can occur intragranularly (volume diffusion), or because of grain defects. The grain boundary diffusion is always faster than the volume diffusion, meanwhile the volume diffusion occurs within a single grain and is only important at higher temperatures. In this case of nc-UO2, the low activation energies obtained could be related predominantly to grain boundary (surface and interface) diffusion as volume diffusion contribution exhibits for higher activation energies (above 4 eV [Sabioni et al., 1998]). A lattice of about 0.5472 nm was already found for the samples treated at 900°C after 50 h dwell time under Ar/H2obtaining a final size about 50 nm (Table 5.5). Therefore a temperature of 1200°C (and in consequence a final crystallite size of 80 nm) would be, in principle, not necessary to reach the typical lattice parameter of the reference large-grained UO2(a=0.5472 nm), as above commented. An average grain size of 322 nm was measured after cooling for the heat treatment at 1200°C after 50 h dwell under He (Table 5.5). Taking that into account, it appears that a temperature below 1200°C would be necessary in the sintering process of of the monoliths, to avoid extreme growth of the particles (>200 nm). Nevertheless for the nc-UO2samples annealed at 1200°C during 50 h under Ar/H2dynamic atmosphere, a final grain size of 85 nm was measured after cooling. Even after 200 h dwell time at this temperature under reducing atmosphere, a final grain size of 150 nm was seen (quite far from the 322 nm observed under He atmosphere after 50 h). This difference could be due to the initial oxidation state of the nc-UO2samples and their evolution under a static He atmosphere. As mentioned previously, an hyperstoichiometric UO2would present a stronger increase of the self-diffusion coefficients and in the same way raise the mass-flow, the grain-boundary motion and the grain (or crystal) growth will occur. In Fig. 5.24, the Arrhenius diagram comparing the cation self-diffusion in UO2fluorite-structure from large-grain reported by [Matzke, 1987] and nano-grain from this study (samples annealed under He static atmosphere during 50 h), has been plotted. Between 20 orders of magnitude at 500°C (lattice parameter =0.5457 nm; average grain size =18 nm) and 5 orders of magnitude at 1200°C (lattice parameter =0.5472 nm; average grain size =322 nm) have been found for the diffusion coefficients between bulk-UO2 [Matzke, 1987] and nc-UO2of this study (Table 5.5). Differences in the diffusivity in the grain boundaries between micro- and nano-grain have been seen already in other fluorite structure metal oxides [Martin, 2007]. In fact the differences in the diffusion coefficient between bulk- and nc-UO2are compatible with an enhancement of the diffusion processes either by a diminishing of the grain size or by O/U>2 effects. 100 5.6. Results and discussion. 101 Chapter 5. Crystallization and Grain Growth in f(T) for nc-UO2by Aqueous route 102 Chapter 6 Crystallization and Grain Growth in f(T) for nc-UO2by Organic route 6.1 Generalities. In this chapter the evolution of the grain size and the crystal structure evolution as a function of temperature under inert and reducing atmosphere of nc-UO2precipitated from organic phase (see Chap. 4), have been investigated. Thermogravimetric analysis (TGA) of the samples provides the starting point for these investigations, enabling the identification of mass losses at given temperatures. The structure of the material as a function of temperature has been identified mainly by XRD and HT-XRD, but also by advanced methods including XANES and EXAFS. 6.2 Thermal evolution and mass changes as probed by TGA/DTA. Thermogravimetry analysis (TGA) and differential thermal analyses (DTA) were employed under Ar/5%H2gas at a heating rate of 5°C/min to determine the thermal decomposition temperature and the water and organic content of the product (see Fig. 6.1). The description of the instrument is shown in Sec. 2.5. Upon heating under Ar/5%H2, a mass loss of 1 wt%is observed until 150°C together with a slight endothermic peak. This loss of weight is likely due to outgassing of residual water coming from the precursor. A second weight loss of about 3.5 wt%appears until 280°C, accompanied by an exothermic peak at 240°C. A new mass loss appears until 485°C which could be due to the residual carbon from the precursor. Between 485 and 1200°C there is no weigh loss (1 wt%) accompanied with wide exotherm probable due to the heat release because crystallization. 6.3 Lattice parameter and crystal growth in neutral atmosphere. The crystal growth, lattice parameter of the nc-UO2from organic route (5.52 nm size and lattice parameter a = 0.5430(1) nm; see Sec. 4.4), has been investigated under inert 103 Chapter 6. Crystallization and Grain Growth in f(T) for nc-UO2by Organic route Figure 6.1: TGA and DTA signal for nc-UO2under Ar/5%H2atmosphere. conditions (static He atmosphere) using in situ HT-XRD. The effect of temperature on the crystallite size, which is a fundamental parameter in the sintering process has been analysed. As it has been seen in Fig. 6.1, no organics loss was observed after 500°C. However, to avoid any possible decomposition of the nc-UO2organic layer in the static atmosphere of the HT-XRD chamber, a pretreatment under O2was applied (500°C during 1 h), followed by 2 h under Ar/5%H2to reverse the possible oxidation of the particles. After thermal pre-treatment, a size of 37 nm and lattice parameter 0.5462(0) nm, were determined. The in situ HT-XRD patterns were acquired with an instrument described in 2.4.2. The temperature range explored was 30 to 1100°C. The evolution with temperature of the nc-UO2XRD pattern is shown in Fig. 6.2. The observed reflections are assigned to UO2-fcc phase structure and to Pt phase corresponding to the heater plate. The effect of temperature on the peaks can be observed more clearly in Fig. 6.2-right, which displays the evolution of two main peaks (the (111) and (200) reflections) of the UO2structure. In Fig. 6.2, one observes a shift in the peak position to lower angles, possibly related to a thermal lattice parameter expansion. An effect of the temperature is seen in the width of the peaks which decreases with increasing temperature while the intensity of the peaks increases. This width change was observed even below 700°C which was the highest pre-treatment temperature. This effect could be due to the longer times (about 10 h) at temperature used in the HT-XRD which induce to a perfectionism of the UO2fcc-structure (higher crystallization). Since the contribution of instrumental broadening is independent of the temperature, the broadening at lower temperatures is mainly related to the crystallite size and strain present in the material, as well as increase of the structural order. Both contributions, crystal size (proportional to cos−1θ; Eq. 5.1) and strain (proportional to tanθ; Eq. 5.2), have different angular dependences, and are so separable. A study of those influences has been performed in the following. 104 6.3. Lattice parameter and crystal growth in neutral atmosphere. Figure 6.2: In situ HT-XRD patterns of nc-UO2under He (left). The typical UO2and Pt (heating plate) Bragg peak positions are also marked. The (right) picture shows just the evolution of (111) and (200) peaks of UO2cubic structure as a function of temperature. 6.3.1 Grain growth as a function of temperature under neutral atmosphere. The crystallite size of the nc-UO2has been determined by XRD Rietveld refinement (see Sec. 2.4) of the Bragg peaks (Fig. 6.2), and used also to characterize the microstructure of the material. From these results it is possible to generate a universal representation of the crystallite size as a function of temperature (XRDs measured at temperature and under static He atmosphere) and reported in Fig. 6.3. Even though, this information is taken as universal, slight deviations may occur, in particular due to dwell times and temperature ramps, but more importantly due to the atmosphere of static He during thermal treatment. Notable crystal size variations were observed above 700°C as this was the temperature already reached during the pre-treatment of the material. The crystallite size change with temperature shows a slow growth up to 700°C, and an intense growth from 37 to 150 nm at 1100°C (see Fig. 6.3). Fig. 6.4 shows TEM images of the nc-UO2particles as-produced, after the prethermal treatment under O2(500°C) and Ar/H2(700°C) and at 1000°C. The size change corresponds to the one measured by XRD. 6.3.2 Lattice parameter and linear thermal expansion coefficient as a function of temperature. The crystal growth of the sample under inert conditions (static He atmosphere) using in situ HT-XRD, have been already described. In addition, the variation of the lattice parameter versus crystal size and temperature, as well as data on the linear thermal expansion, are now reported and compared to bulk material UO2. The crystal structure 105 Chapter 6. Crystallization and Grain Growth in f(T) for nc-UO2by Organic route Figure 6.3: Evolution of the nc-UO2crystallite size in function of the temperature. (a) 4 nm as-produced (b) 34 nm at 700°C (c) 91 nm at 1000°C Figure 6.4: TEM images for the nc-UO2. of the precipitates was, as the crystallite size, determined by Rietveld refinement, taking into account the whole 2θrange. In Fig. 6.5a the lattice parameter obtained as a function of temperature (XRDs measured at temperature and under static He atmosphere) and its derivative (Fig. 6.5b), have been also determined by the XRD Rietveld refinement of the Bragg peaks, and represented together with the nc-UO2size evolution to observe its dependence. Also the calculated lattice evolution of non-stoichiometric standard UO2+xfor different O/U ratios due to only thermal expansion, have been represented for comparison. The lattice parameter of a non stoichiometric UO2+xis linked to the oxygen content by the relations of [Lynds et al., 1963]. Also the lattice parameter was corrected as a function of temperature with the [Fink, 2000] relations already reflected in Eq. 5.3. An expansion in the lattice parameter from 0.5462(0) nm at RT (after being pretreatment) to 0.5482(0) nm at 300°C, has been determined. Above this temperature, a linear evolution of the the lattice parameter with temperature is observed. Relating the lattice parameter found in this study with the [Lynds et al., 1963] relations, a stoichiometry of UO2.04 up to 300°C to UO2.00 up to 750°C, has been determined. The 106 6.3. Lattice parameter and crystal growth in neutral atmosphere. (a) (b) Figure 6.5: a.) Lattice constant and crystallite size variation of nc-UO2in function of temperature (curves only as a guide to eye), from in situ HT-XRD measurements under static He atmosphere in comparison with lattice evolution in function of temperatures of standard UO2 for different O/U ratios obtained by the relations of [Lynds et al., 1963], due to only thermal expansion. b.) Relative crystallite size and lattice parameter vs. temperature (curves only as a guide to eye). nanocrystallites stabilize at O/U 2.0 at temperatures above 750°C, or in other words, at particles sizes >44 nm. Fig. 6.6 displays the linear thermal expansion (LTE) and the linear thermal expansion coefficient (LTEC) of nc-UO2as a function of the temperature. The LTE at temperature Twas calculated using the relation 5.4. The LTEC was calculated by differentiating the thermal expansion curve aTversus Twith respect to the temperature T(see Eq. 5.5). The LTE of the nc-UO2is just slightly higher than the one for UO2bulk [Martin, 1988] for all the interval of temperatures, as one could already predict from the lattice parameter representation in function of temperature (see Fig. 6.5a). The LTEC is initially higher for nc-UO2than for bulk-UO2for temperatures below 400°C and tends to stabilize above this temperature with a value of 12·10−6°C−1in agreement with the value for the LTEC of bulk-UO2. The oscillatory trends observable for LTEC in nc-UO2can be attribute to transitory oxidation-reduction effects. If Fig. 6.7 the patterns comparison of nc-UO2at RT (previously treated at 500°C during 1 h and 700°C during 2 h under O2and Ar/5%H2, respectively) (a = 0.5462(0) nm), nc-UO2at 1100°C (a = 0.5534(0) nm) and nc-UO2at RT after thermal treatment at 1100°C (a = 0.5472(0) nm) (all measured in situ in the HT-XRD instrument under static He atmosphere), is shown. At 1100°C under static He atm, an O/U ratio of 2.0 (see Fig. 5.4a) and a lattice pararameter of a = 0.5534(0) nm (measured at temperature), have been observed. The same thermally treated sample measured after cooling at RT, shows a value of 107