Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Tesis Doctoral
Tesis Doctoral
Caracterización computacional de detectores de Germanio Hiperpuro (HPGe) empleando simulación Montecarlo y optimización mediante algoritmos evolutivos.
Caracterización computacional de detectores
de Germanio Hiperpuro (HPGe) empleando
simulación Montecarlo y optimización
mediante algoritmos evolutivos
2018
, . , UNIVERSIDAD DE LAS PALMAS
f "' DE GRAN CANARIA
INFORMA,
Título de la Tesis
Caracterización computacional de detectores de Germanio Hiperpuro (HPGe) empleando simulación Montecarlo
y optimización mediante algoritmos evolutivos.
Agradecer al Dr. D. Pablo Martel Escobar, coordinador del grupo GIRMA, por los cuantiosos
medios materiales y económicos proporcionados, incluyendo dos detectores HPGe, que han
permitido llevar a cabo este trabajo, así como por las discusiones científicas que sin duda han
contribuido a mejorar el mismo.
Agradezco el apoyo recibido por los compañeros del grupo GIRMA, cuyo currículo me ha
permitido competir por los contratos de trabajo y ayudas para la investigación que han financiado
este trabajo. Agradecer especialmente al Dr D. Héctor Alonso Hernández por su inestimable
apoyo moral y al Dr D. Miguel Ángel Arnedo Ayensa, que, sin el gran trabajo previo realizado
por él en su Tesis Doctoral, yo hubiese necesitado mucho más tiempo para finalizar este trabajo
de Tesis Doctoral.
Agradezco al Dr. D. Juan Pedro Bolívar del grupo de investigación FRyMA de la Universidad
de Huelva por sus interesantes aportaciones, la importante cantidad de información facilitada
sobre sus detectores de espectrometría gamma, espectros adquiridos, materiales de referencia
y técnicas experimentales. Agradezco la gran hospitalidad de los integrantes de dicho grupo de
investigación durante la estancia en dicha universidad, la cual fue muy fructífera, en particular a
D. Isidoro Gutiérrez Álvarez, quien fuera del horario laboral hizo de conductor y guía turístico
amenizando enormemente mi estancia.
Doy las gracias a mis padres, ya que sin ellos no hubiese llegado hasta aquí, especialmente
a mi padre que desde pequeño me inculcó el interés por la ciencia y la tecnología, sin olvidar la
gran ayuda que me prestó durante la definición y comienzo de este trabajo de Tesis Doctoral.
También doy las gracias a mi pareja Cristina y mi hermana Yaiza que siempre me han apoyado.
232
Figura 1 Cadena de desintegración del 90 Th o serie del torio ................................................... 7
Figura 21 Imagen esquemática del proceso de dispersión de un haz de partículas (Salvat et al.,
2011). ................................................................................................................................ 35
Figura 22 Desviación angular en un evento de dispersión (Salvat et al., 2011). ....................... 36
Figura 23 Generación aleatoria de trayectorias usando simulación detallada de la interacción
radiación-materia (Salvat et al., 2011). ....................................................................................... 37
Figura 24 Curvas de eficiencias obtenidas para una fuente de 152Eu situada directamente sobre
la ventana del detector (superior) y a 11.5 cm de dicha ventana (inferior). (Gilmore and
Hemingway, 1995). ..................................................................................................................... 39
Figura 25 Esquema de decaimiento simplificado del 152Eu (Gilmore and Hemingway, 1995). .. 40
Figura 26 Esquema ilustrativo de un decaimiento simple que podría dar lugar al fenómeno de
suma por coincidencia. ................................................................................................................ 42
Figura 27 Ejemplo de soluciones candidatas y frente óptimo de Pareto (circunferencias rojas en
espacio de funciones) para un problema bi-objetivo con dos variables de decisión. ................. 48
Figura A1.1 Características técnicas del detector HPGe XtRa del grupo GIRMA de la
Universidad de Las Palmas de Gran Canaria. .......................................................................... 107
Figura A1.2 Características técnicas del detector HPGe XtRa del grupo FRyMA de la
Universidad de Huelva. ............................................................................................................. 108
Figura A1.3 Características técnicas del detector HPGe tipo pozo del grupo GIRMA de la
Universidad de Las Palmas de Gran Canaria. .......................................................................... 109
Figura A3.1 Diagrama de flujo del procedimiento de calibración mediante el programa
LABSOCS para detectores caracterizados. .............................................................................. 114
Figura A3.2 Ejemplo del empleo del Geometry Composer para definir la geometría de medida y
materiales. .............................................................................................................................. 114
Figura A3.3 Definición de las energías de interés, incertidumbres asociadas a las FEPEs y
parámetros de convergencia. .................................................................................................... 115
Figura A3.4 Sistema integrado de calibración GENIE 2000-LabSOCS. Ejemplo de ajuste a un
polinómico de orden 3 en logaritmos naturales de las eficiencias y energías. ......................... 116
Figura A4.1 Estructura de archivos para la simulación mediante el código PENELOPE. ....... 118
Figura A4.2 Ejemplo de archivo de entrada.in. ........................................................................ 122
Figura A4.3 Representación de los cuerpos cilíndricos definidos en el fichero entrada.in
mediante geometry viewer (izquierda). Zoom redondeo superior (derecha). .......................... 123
Figura A4.4 Espectro simulado empleando PENELOPE. ........................................................ 124
Figura A5.1 Diagrama de flujo del procedimiento de cálculo de la actividad mediante el
programa GENIE-2000 (Gilmore and Hemingway, 1995). ....................................................... 128
Figura A5.2 Parámetros empleados en la estimación del continuo (lineal a la izquierda y tipo
escalón a la derecha) bajo el pico (Canberra, 2002b). ............................................................. 129
Figura A6.1 Factores de corrección por suma por coincidencia (CSCF) para DET1 (a) y DET2
(b) para las 9 emisiones afectadas significativamente por dicho efecto (242.0, 609.3, 1120.3
and 1238.1 keV en RGU-1 and 338.3, 583.2, 727.3, 860.0 and 911.0 keV en RGTh-1). ........ 132
Figura A6.2 Comparación entre las FEPEs experimentales y las correspondientes calculadas
mediante PENELOPE para el modelo óptimo del detector DET1 (mU, mTh y mK son las masas
correspondientes a las muestras de RGU-1, RGTh-1 and RGK-1 respectivamente). ............. 145
Índice de figuras
Figura A6.3 Comparación entre las FEPEs experimentales y las correspondientes calculadas
mediante PENELOPE para el modelo óptimo del detector DET1 y para el recipiente B3 con 10
alturas diferentes de muestra de los materiales de referencia RGU-1, RGTh-1 and RGK-1. .. 145
Figura A7.1 Curva de eficiencia ajustada a los 8 puntos energía-eficiencia experimentales
libres de suma por coincidencia, tomados para la muestra de RGU-1 con 1 mm de altura de
muestra. .............................................................................................................................. 148
Figura A7.2 Curva de eficiencia ajustada a los 8 puntos energía-eficiencia experimentales
libres de suma por coincidencia, tomados para la muestra de RGU-1 con 2 mm de altura de
muestra. .............................................................................................................................. 148
Figura A7.3 Curva de eficiencia ajustada a los 8 puntos energía-eficiencia experimentales
libres de suma por coincidencia, tomados para la muestra de RGU-1 con 3 mm de altura de
muestra. .............................................................................................................................. 149
Figura A7.4 Curva de eficiencia ajustada a los 8 puntos energía-eficiencia experimentales
libres de suma por coincidencia, tomados para la muestra de RGU-1 con 4 mm de altura de
muestra. .............................................................................................................................. 149
Figura A7.5 Curva de eficiencia ajustada a los 8 puntos energía-eficiencia experimentales
libres de suma por coincidencia, tomados para la muestra de RGU-1 con 5 mm de altura de
muestra. .............................................................................................................................. 150
Figura A7.6 CSCF, calculados para las energías de 242.0, 609.3, 1120.3 y 1238.1 keV,
correspondientes a las 5 muestras medidas del material IAEA RGU-1. .................................. 150
Figura A7.7 CSCF, calculados para las energías de 338.3, 583.2, 727.3, 860.0 y 911.0 keV,
correspondientes a las 5 muestras medidas del material IAEA RGTh-1. ................................. 151
Figura A7.8 CSCF, calculados para las energías de 242.0, 338.3, 609.3, 911.0, 1120.3 y
1238.1 keV, correspondientes a las 5 muestras medidas del material IAEA 448. ................... 151
Figura A7.9 CSCF, calculados para las energías de 338.3, 583.2, 609.3 y 911.0 keV,
correspondientes a las 5 muestras medidas del material IAEA 447......................................... 152
Figura A7.10 Resultados del ajuste de la curva en la ecuación [A7.1] a las mediciones de
concentración de radón en la cámara de acumulación. ........................................................... 153
Figura A7.11 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
RGU-1 y las energías: 46.5, 63.3, 143.8, 186, 242, 295.2, 352, 609.3, 1001, 1120.3, 1238.1 y
1764.5 keV. .............................................................................................................................. 155
Figura A7.12 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
RGTh-1 y las energías: 238.6, 338.3, 583.2, 727.3, 860 y 911 keV. ........................................ 155
Figura A7.13 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
RGK-1. .............................................................................................................................. 156
Figura A7.14 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA 448
y las energías: 46.5, 186, 242, 295.2, 352, 609.3, 1120.3, 1238.1 y 1764.5 keV. ................... 156
Figura A7.15 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA 447
y las energías: 46.5, 186, 242, 295.2, 338.3, 352, 583.2, 609.3, 661.7, 727.3, 860 y 911 keV. ....
.............................................................................................................................. 157
Índice de tablas
Capítulo 1
Introducción
Antecedentes
Por otra parte, el método de simulación directa de Monte Carlo no requiere la medición
experimental de muestras de referencia una vez se ha construido un modelo de detector
suficientemente exacto, es decir, que proporciona FEPEs computacionales en concordancia
con los resultados experimentales, lo cual se consigue mediante un procedimiento conocido
como la caracterización o modelización del detector. Obviamente, la ventaja que proporciona
este método frente a la calibración experimental y el método de transferencia de eficiencias es
la simplicidad y rapidez con la que se puede realizar la calibración en eficiencia del sistema de
espectrometría, para una amplia variedad de materiales de las muestras y geometrías de
medida. Por esta razón, gran cantidad de trabajos han sido publicados durante las últimas
Capítulo 1 – Introducción 3
décadas enfocados en la caracterización de detectores de espectrometría gamma (Boson et
al., 2008; Budjas et al., 2009; Courtine et al., 2008; Hardy et al., 2002; Huy et al., 2007; Maleka
et al., 2005; Rodenas et al., 2003; Szentmiklósi et al., 2014).
Objetivos
En los dos primeros trabajos publicados, tanto las FEPEs tomadas como referencia
durante la caracterización como aquellas empleadas durante la validación de los modelos
construidos, han sido calculadas, para un detector dado, mediante el software LabSOCS de
Canberra (Canberra, 2002a). La elección de LabSOCS como referencia se debe principalmente
a su flexibilidad y simplicidad a la hora de calcular FEPEs para gran variedad de materiales y
geometrías utilizables durante la validación de las metodologías de caracterización.
Adicionalmente, esto ha permitido evitar imprecisiones que podrían haber surgido en caso de
haber empleado FEPEs experimentales en su lugar, que podrían haber generado dudas sobre
si las desviaciones observadas serían debidas a problemas en la metodología o bien debido a
errores experimentales. En ambos trabajos las metodologías de caracterización han probado
obtener exitosamente un modelo de detector que genera FEPEs equivalentes a las tomadas
como referencia, con desviaciones inferiores a las incertidumbres de las FEPEs de referencia.
232
Th
Figura 1 Cadena de desintegración del 90 o serie del torio.
8 Capítulo 1 – Introducción
235
U
Figura 3 Cadena de desintegración del 92 o serie del Actinio.
Como se puede observar en las figuras, las tres cadenas radiactivas terminan en un
isótopo de plomo estable. Las tres series naturales disponen de progenie emisora de radiación
alfa, beta y gamma. La energía de las radiaciones gamma emitidas por los radionúclidos de
dichas series oscila entre pocas decenas de keV hasta aproximadamente 3 MeV.
[2]
Donde es la energía de enlace o ligadura del electrón al átomo. Para energías por
encima del centenar de keV, el electrón emitido, o fotoelectrón, toma la mayor parte de la
energía original del fotón.
[3]
Siendo el número atómico del átomo, es un parámetro que toma los valores 4 o 5
dependiendo de la energía del fotón y la energía del fotón incidente. De la ecuación [3], se
puede deducir que la sesión eficaz , y por lo tanto la probabilidad de que el efecto
fotoeléctrico tenga lugar, será mayor cuanto menor sea la energía del fotón incidente y mayor
sea el número atómico del átomo con el que interacciona (Vidal-Cuadras and Ortega, 1986).
Por lo tanto, el efecto fotoeléctrico es el modo de interacción con la materia predominante para
fotones X y γ de energía relativamente baja, proceso que se ve favorecido en materiales de
número atómico elevado.
12 Capítulo 1 – Introducción
Dispersión Rayleigh
Dispersión Compton
[4]
[5]
De la ecuación [5] se deduce que la energía cinética del electrón emitido puede tomar
[6]
Creación de pares
,
[8]
Donde en donde , es una función que crece monótonamente con la energía del
fotón incidente y que varía ligeramente en función del número atómico del elemento . La
probabilidad de que dicha interacción tenga lugar es relativamente baja hasta que la energía
del fotón gamma incidente supera varios MeV, convirtiéndose en el mecanismo de interacción
predominante para energías superiores a los 10 MeV.
Hasta la década de los 70 los detectores más empleados eran los centelleadores, pero
a partir de entonces se comenzó a generalizar el uso de detectores de semiconductor, siendo
los más empleados en la actualidad los detectores de germanio de alta pureza o HPGe
(Gilmore and Hemingway, 1995).
Para generar un campo eléctrico con suficiente intensidad de modo que permita una
recogida de carga eficiente, es necesario aplicar una diferencia de potencial eléctrico a los
terminales de la unión P-N de cientos o miles de voltios. Dado que todos los semiconductores
tienen una conductividad finita, incluso en ausencia de radiación ionizante se observará una
cierta corriente eléctrica denominada corriente de fuga, cuyas fluctuaciones tienden a
confundirse con la señal de corriente debido a al suceso ionizante que se pretende detectar. Es
por ello que los métodos para reducir esta señal de fuga son de gran importancia en el diseño
de detectores de semiconductor.
Por otra parte, tanto el germanio como el silicio son elementos de bajo número atómico
(32 y 14 respectivamente), por lo que la probabilidad de interacción con la radiación gamma
incidente de más de 500 keV es bastante baja. Por lo tanto, es importante que la zona de carga
espacial libre de portadores posea el mayor volumen posible de modo que se aumente la
probabilidad de interacción dentro de la zona activa del detector. El aumento de dicha zona
puede conseguir aumentando la tensión inversa de polarización, no obstante, esta solución
práctica no se aconseja dada la inconveniencia de trabajar con grandes tensiones.
Con el fin de conseguir una zona de carga espacial libre de portadores lo más grande
posible sin la necesidad de aumentar considerablemente la tensión inversa de polarización, y
por otra parte reducir la señal de fuga, se fabrican los detectores de Germanio con una
estructura P-I-N, con una zona de carga espacial libre de portadores, denominada región
16 Capítulo 1 – Introducción
intrínseca (I), la cual es sensible a la radiación ionizante X y γ. Al polarizar inversamente el
detector con una diferencia de potencial eléctrico, un campo eléctrico se extiende a lo largo de
la región intrínseca de modo que la anchura de dicha región aumenta. Dada la alta resistividad
eléctrica de esta región, se consigue reducir la mencionada corriente de fuga
considerablemente.
El proceso más utilizado para fabricar detectores de semiconductor con una estructura
P-I-N es el llamado proceso de compensación. Se dice que un semiconductor extrínseco
(dopado con impurezas pentavalentes (tipo N) o impurezas trivalentes (tipo P)) está
compensado cuando se le han añadido una concentración de impurezas que hace que se
comporte como intrínseco, es decir, añadiéndole impurezas donadoras si se trata de un
semiconductor tipo P, o impurezas aceptoras si se trata de un semiconductor tipo N. La razón
por la que este proceso de compensación es empleado para la fabricación de estructuras P-I-N
radica en que, obtener un semiconductor sin ningún tipo de impurezas requiere un proceso
muy complejo (y por lo tanto caro), mientras que la compensación de un semiconductor
extrínseco es un proceso relativamente sencillo desde el punto de vista tecnológico. En los
detectores fabricados empleando este proceso, mediante la compensación de la zona
intermedia entre el semiconductor tipo N y el tipo P se obtendrá una zona de carga espacial
que puede alcanzar una anchura del orden de centímetros.
Figura 11 Detectores HPGe ofertados por Canberra inc.1 y sus características principales.
Detectores planos
1
www.Canberra.com
Capítulo 1 – Introducción 19
capa de aniones de Boro de bastante menor espesor que el contacto p+, de modo que la
radiación incidente no se vea atenuada en tal capa.
Detectores coaxiales
Los detectores de pozo permiten tener eficiencias muy elevadas para muestras de
volumen reducido, debido a que la fuente queda prácticamente rodeada por el detector (la
geometría de conteo se aproxima a la geometría 4π en la que la eficiencia de detección está
principalmente gobernada por la eficiencia intrínseca del detector). Se trata de detectores
HPGe de forma cilíndrica con un orificio cilíndrico perforado en su interior desde la parte
superior, conocido como pozo. Como en el resto de casos, la carcasa del detector suele ser de
aleación de aluminio con un espesor dentro del orificio interno, donde va situada la muestra, de
unos 0.5 mm, lo que junto al pequeño espesor del contacto p+ en el pozo del cristal permite
realizar espectrometría gamma a bajas energías (a partir de unos 20 keV).
Figura 12 Típicas curvas de FEPEs para varios detectores HPGe y una fuente puntual
situada a 2.5 cm de distancia desde la superficie superior del cristal de Ge 2.
Para terminar, un sistema común de espectroscopia de radiación gamma con detector
HPGe está formado por el propio detector, una fuente de alimentación de alto voltaje, un
preamplificador, un amplificador, un convertidor analógico digital (ADC) y un analizador
2
www.Canberra.com
Capítulo 1 – Introducción 21
multicanal (MCA). La función del sistema electrónico consiste en procesar los pulsos
producidos en el detector y ordenarlos en el espectro como cuentas en distintos canales según
la energía que representan.
Las fuentes de radiación gamma emiten fotones gamma con una distribución
aproximadamente discreta de energías. No obstante, el espectro adquirido por un detector de
radiación gamma se parece más a una distribución continua en la cual se superponen los picos
correspondientes a la absorción completa de la energía del fotón incidente, con una forma que
puede aproximarse mediante gaussianas. En la Figura 13 se muestran los distintos
mecanismos de interacción de la radiación gamma con la materia que se dan dentro de la zona
activa del detector, depositando energía al producirse (energía cinética comunicada por los
fotones incidentes a los electrones del detector) y por lo tanto contribuyendo en la generación
del espectro de radiación gamma adquirido por el detector. Dichos mecanismos son: el efecto
fotoeléctrico, la dispersión Compton y la creación de pares electrón-positrón.
Debido a la dispersión Compton sufrida por los fotones (en la zona activa del detector,
en la fuente, o cualquier otro medio material que rodee el detector), se produce una absorción
parcial de la radiación emitida por los radionucleidos contenidos en la fuente, formándose por lo
tanto en el espectro adquirido el Continuo Compton, que se denota en la Figura 14 por B. En el
caso de que la dispersión Compton se dé dentro de la zona activa del detector, escapando de
tal volumen el fotón gamma dispersado, teniendo en cuenta la ecuación [5] se deduce que la
energía depositada en el detector mediante este proceso (la energía cinética del electrón
emitido por la interacción Compton, que es absorbido por el medio), será como máximo
Cuando el proceso de creación de pares ocurre dentro del volumen activo del detector,
uno o ambos fotones producidos en la aniquilación del positrón (de energía cada uno)
podrían escaparse, lo que contribuye a los picos E y F del espectro. Estos picos, denominados
pico de escape simple (E) y de escape doble (F), aparecen centrados en las energías
y 2 respectivamente.
Cuando la radiación gamma emitida por la fuente con energía superior a 1022 keV
interacciona con los materiales que rodean el detector, se puede producir el fenómeno de
formación de pares electrón-positrón. La posterior aniquilación del positrón produce, como se
mencionó anteriormente, dos fotones de energía en la misma dirección y sentido
Capítulo 1 – Introducción 23
contrario, por lo que solo uno de ellos podría ser absorbido por la zona activa del detector. Al
ocurrir esto repetidamente, aparece un pequeño pico en el espectro, G en la Figura 14, de
energía .
Por último, cuando la radiación gamma emitida por la fuente es absorbida por las
paredes del blindaje del detector, el material de estas paredes emite radiación X característica
comúnmente en torno al rango 70-85 keV en el caso de blindaje de plomo, y 7-8 keV en el caso
de hierro, la cual es absorbida por el detector, formándose el pico I en la Figura 14.
Figura 15 Espectros reales adquiridos por detector HPGe XtRa, cada uno
correspondiente a la medida de una muestra de los materiales de referencia IAEA-RGK-1,
IAEA-RGU-1 e IAEA-RGTh-1.
FWHM
Se define el FWHM (acrónimo de “Full width at half maximum”) como el ancho del pico
de absorción completa a la mitad de su altura máxima. Este parámetro, que viene
generalmente expresado en canales o en energía, define la resolución en energía del detector,
de modo que cuanto menor sea FWHM, mejor es la resolución en energía del detector.
Figura 17 Espectros adquiridos para una misma muestra en tres detectores distintos, dos
detectores de centelleo de NaI y LaBr3 y un detector de germanio de alta pureza 3.
√ [9]
Eficiencia de detección
Eficiencia intrínseca (ya sea total o del pico de absorción completa): relaciona las
cuentas registradas en el espectro con el número de fotones gamma incidentes en el
detector. Esta eficiencia es un parámetro básico del detector y es independiente de
3
http://www.ortec-online.com/
26 Capítulo 1 – Introducción
la geometría y material de la fuente, así como de su posición relativa respecto al
detector.
∙ [10]
Siendo el ángulo sólido del detector visto desde la fuente, que en este caso viene
dado por:
2∙ ∙ 1 [11]
Razón Pico-Compton
Tiempo muerto
Calibración en eficiencia
[12]
∙ ∙ ∙
28 Capítulo 1 – Introducción
Dónde:
Como se mencionó con anterioridad, para un detector dado, las eficiencias absolutas
de pico de absorción completa o FEPEs varían de manera significativa especialmente con la
energía del fotón gamma incidente, la geometría y material de la muestra a medir, así como
con la geometría relativa de la muestra respecto al detector. Por ello, para poder realizar
medidas exactas y precisas es indispensable realizar la calibración del equipo de
espectrometría en relación con esta eficiencia. Dicha calibración puede realizarse mediante
métodos teórico-matemáticos incluyendo la simulación Monte Carlo, así como de manera
experimental.
∙ ∙ [13]
[14]
∙ ∙ ∙
log ∑ [15]
log ∑ [17]
Es importante remarcar que todas estas técnicas para obtener curvas de eficiencia no
tienen ninguna base teórica, sino que son simplemente relaciones empíricas que podrían
ajustarse a los datos experimentales en mayor o menor medida. Por lo tanto, la elección de la
técnica de ajuste/interpolación, así como del tipo de curva a emplear solo puede realizarse
basada en la experiencia. Por ejemplo, una función que funciona bien para un detector HPGe
tipo p podría no ser satisfactoria para un tipo n, donde el comportamiento de la curva de
eficiencia a bajas energías difiere. Por otra parte, a la hora de escoger el orden del polinomio
de ajuste se debería tener cuidado. Un polinomio de orden superior será ciertamente capaz de
ajustar todas las ligeras variaciones en los datos mejor que uno de orden bajo, y se conseguirá
un mejor coeficiente de determinación (de hecho, si el orden del polinomio es una unidad por
debajo del número de datos se obtendrá el polinomio interpolador, que pasa exactamente por
los puntos a ajustar). No obstante, si dichas variaciones en los datos son simplemente debidas
a incertidumbres estadísticas, el ajuste a un polinomio de orden superior podría resultar menos
apropiado que a uno de pequeño orden.
Otro aspecto con el que hay que tener cuidado a la hora de generar curvas de
eficiencia es que nunca se debe eliminar puntos del conjunto de datos por no ajustarse bien al
resto, sin previamente averiguar la razón de tal mayor desviación. Por ejemplo, en el caso de
existir el efecto de suma por coincidencia, es posible que ese único punto desviado de la curva
pudiera ser el único correcto.
Una vez que se han obtenido las FEPEs y las curvas de calibración en eficiencia
experimentales para unas geometrías de medida y materiales concretos de calibración, es
necesario corregir estas eficiencias para obtener las FEPEs correspondientes a los materiales
(densidad y composición química) de las muestras que se pretenden medir. Dicha corrección,
conocida como corrección de auto-atenuación, es necesaria cuando no se dispone de
materiales de referencia con composiciones químicas y densidades similares a las de las
muestras reales a medir, lo cual es bastante común en la práctica, y cobra especial importancia
en el caso de emisiones gamma con energías por debajo de pocas centenas de keV. Por lo
general, dicha corrección consiste en encontrar un factor de corrección por auto-atenuación,
que a su vez es función de la energía de la emisión gamma, densidad, composición química y
altura de la muestra, y que tiene en cuenta la diferente auto-atenuación entre la muestra a
medir, y la muestra empleada para la calibración. La corrección por auto-atenuación puede
conseguirse mediante métodos experimentales, empleando simulación Montecarlo, o bien
mediante ecuaciones matemáticas basadas en los coeficientes de atenuación.
∏ [18]
Sea una variable aleatoria que sigue una distribución de probabilidad con función
densidad definida en un intervalo ∈ , , entonces el valor esperado o esperanza
matemática de una función , también una variable aleatoria que sigue su propia
integral:
[19]
Capítulo 1 – Introducción 33
Dicha esperanza matemática representa la media poblacional de la función . Por lo
tanto, la evaluación Monte Carlo de la integral se puede realizar de manera sencilla generando
un gran número n de valores aleatorios a partir de su función de densidad de probabilidad
y calcular los valores . Con ello, se puede aproximar a partir de la media
aritmética de los valores obtenidos, teniendo en cuenta que, por la ley de los grandes
números:
lim ∑ [20]
→
[21]
[23]
[24]
[25]
[26]
, ,
[27]
, ,
2 [28]
, ,
, , [29]
∀ 0,2
[30]
0 ∀ ∞, 0 ∪ 2 , ∞
Los fotones gamma, al igual que otro tipo de radiaciones, interaccionan con la materia
mediante distintos mecanismos, pudiéndose definir una sección eficaz total como:
∑ [31]
[32]
→ ln 1 [33]
Para simplificar, dado que solo puede tomar valores entre 0 y 1, se puede definir
una variable 1 (que obviamente también toma valores entre 0 y 1), reformulándose
la ecuación anterior, de modo que se puede muestrear aleatoriamente de manera sencilla
empleando la ecuación [33] (teniendo en cuenta que el camino libre medio total viene dado por
):
ln [34]
Capítulo 1 – Introducción 37
Por lo tanto, la siguiente interacción ocurrirá en la posición:
[35]
Existen dos métodos para calibrar en eficiencia empleando simulación Monte Carlo, el
método de transferencia de eficiencia y el método de simulación directa Monte Carlo. El
método de transferencia de eficiencia es un método semi-empírico consistente en la
determinación de las FEPEs a utilizar durante la calibración en eficiencia , a partir de
FEPEs experimentales determinados para una geometría de medida de referencia
(generalmente con una fuente puntual) multiplicando por un factor de transferencia. Dicho
factor de transferencia se obtiene, para una energía dada, mediante el cociente entre dos
eficiencias totales absolutas calculadas por simulación Montecarlo (empleando un modelo
computacional de detector dado, a partir de los datos, exactos o no, proporcionados por el
38 Capítulo 1 – Introducción
,
fabricante), para la geometría de medida de referencia y para la geometría de medida
,
para la que se pretende estimar la eficiencia de detección .
, [36]
[37]
Por lo tanto, este método permite calibrar en eficiencia sin necesidad de que el modelo
computacional de detector empleado proporcione eficiencias muy similares a las
experimentales. No obstante, el método tiene el inconveniente de que la geometría de medida
tiene que ser similar a la de referencia para que se puedan obtener resultados de exactitud
aceptable, por lo que no se evita la necesidad de realizar medidas experimentales de muestras
de referencia cuando se cambia considerablemente la geometría de medida.
Por otra parte, el método de simulación directa de Monte Carlo no requiere la medición
experimental de muestras de referencia una vez se ha construido un modelo de detector
suficientemente exacto, es decir, que proporciona FEPEs computacionales en concordancia
con los resultados experimentales, lo cual se consigue mediante un procedimiento conocido
como la caracterización o modelización del detector. Obviamente, la ventaja que proporciona
este método frente a la calibración experimental y el método de transferencia de eficiencias es
la simplicidad y rapidez con la que se puede realizar la calibración en eficiencia del sistema de
espectrometría, para una amplia variedad de materiales de las muestras y geometrías de
medida.
Figura 24 Curvas de eficiencias obtenidas para una fuente de 152Eu situada directamente
sobre la ventana del detector (superior) y a 11.5 cm de dicha ventana (inferior). (Gilmore and
Hemingway, 1995).
En la Figura 24 se aprecia de una manera clara los problemas causados por la suma
por coincidencia en la calibración en eficiencia experimental. En dicha figura se muestran dos
152
curvas correspondientes a una fuente de Eu situada directamente sobre la ventana del
detector y a 11.5 cm de dicha ventana. Se puede observar que en el caso de la fuente
separada 11.5 cm de la ventana, los puntos energía-eficiencia siguen una trayectoria curva
bastante suave que podría ser fácilmente aproximada por una función analítica como las
propuestas con anterioridad. En cambio, en el caso de la fuente situada directamente sobre la
ventana del detector, se observa que los puntos no siguen una línea curva tan clara, por lo que
sería complicado encontrar una curva que se ajustara apropiadamente a dichos puntos, los
cuales no parecen satisfactorios en absoluto. La razón de esta dramática diferencia radica en el
fenómeno de suma por coincidencia.
40 Capítulo 1 – Introducción
En la Figura 25 se muestra el esquema de decaimiento simplificado del 152Eu. Como se
152 152
puede observar, los radionucleidos Eu pueden decaer a su isóbaro Gd emitiendo una
partícula (con una probabilidad de aproximadamente 27.92%) o bien, mediante captura
electrónica decaer a su isóbaro 152Sm (con una probabilidad de 72.08%). Cualquiera que sea el
modo de decaimiento, el producto del mismo queda excitado, emitiendo radiación gamma
hasta alcanzar su estado fundamental.
152
Figura 25 Esquema de decaimiento simplificado del Eu (Gilmore and Hemingway,
1995).
Los tiempos de vida de niveles nucleares individuales son muy cortos, mucho menores
que el tiempo de resolución del equipo de espectrometría gamma. Desde el punto de vista del
152
detector cada desintegración de un átomo de Eu en la fuente liberara un número de rayos
gamma, y probablemente de rayos X de manera simultánea, habiendo una cierta probabilidad
de que más de una de esas emisiones sean detectadas conjuntamente. Si eso pasara,
entonces un pulso será registrado representando la suma de las energías de los fotones
individuales detectados simultáneamente, lo que se conoce como la suma de varios fotones
emitidos en coincidencia, o suma por coincidencia.
El efecto de summing-in, solo puede ocurrir cuando dos o más fotones gamma han sido
absorbidos por completo y de manera simultánea por el material activo del detector. Por otra
parte, el fenómeno de summing-out puede darse al sumarse varias emisiones gamma o X,
hayan sido absorbidas por completo o no (basta con que se absorba completamente una de las
emisiones gamma a la vez que una absorción incompleta gamma o X, para que el evento sea
registrado en una energía más elevada, resultante de la suma de las dos). La suma de fotones
gamma con rayos X ocurre con mayor intensidad en detectores de Germanio diseñados para
detectar fotones de baja energía, tales como los detectores REGe y XtRa. Por lo tanto, queda
claro que, tanto durante la calibración en eficiencia experimental, como durante la medida de
muestras por espectrometría gamma, cuando se empleen picos que pueden estar afectados
por suma por coincidencia, se deberán realizar las correcciones pertinentes.
" "
[38]
" "
Estos CSCFs serían adecuados para la corrección por suma por coincidencia de la
medida de actividades de los radionucleidos correspondientes (empleando sus emisiones
afectadas por dicho efecto) en muestras con características similares a las empleadas para
determinar dichos coeficientes (a energías por encima de los varias centenas de keV, al igual
que pasa con la eficiencia de detección, los CSCF estarán afectados principalmente por la
geometría de medida, siendo la densidad y composición química del material un factor
secundario).
42 Capítulo 1 – Introducción
Por otra parte, también existen procedimientos matemáticos (probabilísticos) para
corregir por suma por coincidencia (Gilmore and Hemingway, 1995). En la Figura 26 se
muestra el esquema de decaimiento más simple susceptible de producir suma por
coincidencia. Como se puede observar, consiste en el decaimiento beta a uno de dos estados
excitados seguido de la emisión de tres fotones gamma.
Figura 26 Esquema ilustrativo de un decaimiento simple que podría dar lugar al fenómeno
de suma por coincidencia.
Asumiendo que se va a medir por espectrometría gamma una fuente que contiene una
actividad A de un radionucleido que decae según el esquema ilustrado en la Figura 26, la tasa
de conteo en el pico de absorción completa correspondiente a , suponiendo ausencia de
suma por coincidencia, vendrá dado por:
[39]
′ [40]
1 [41]
Capítulo 1 – Introducción 43
El caso de es ligeramente diferente en tanto en cuanto no todos los fotones gamma
emanados desde el nivel de energía intermedio son consecuencia de la desexcitación desde el
nivel más alto. Una proporción están precedidos inmediatamente por un decaimiento β- y no
puede contribuir a la suma por coincidencia. En este caso el número de eventos de suma viene
dado por el producto del número de eventos dando lugar a tras la emisión , la
probabilidad de detección de en el pico de absorción completa correspondiente ( ) y la
probabilidad de detección de en cualquier lugar del espectro ( ). Por lo tanto, la tasa de
conteo del pico 2 vendrá dado por:
′ [42]
Y el correspondiente CSCF:
1 [43]
Estos dos casos estudiados hasta ahora son los nombrados con anterioridad como
eventos summing-out, para los cuales como se puede observar en las ecuaciones [41] y [43],
los CSCF serán menores que 1.
Por otra parte, cada suma por coincidencia de rayos gamma y completamente
absorbidos producirá una cuenta en un pico equivalente a la suma de sus energías, y, por lo
tanto, el pico correspondiente a vera su área incrementada en lugar de reducida (summing-
in). Aplicando análogamente los mismos criterios que para los casos de y , la tasa de
conteo del pico 3 vendrá dado por:
′ [44]
Y el correspondiente CSCF:
1 [45]
Aunque estos cálculos puedan parecer triviales, debe quedar claro que el caso arriba
estudiado es una gran simplificación con respecto a los esquemas de decaimiento de
radionucleidos reales. Así, por ejemplo, observando la Figura 25, se puede intuir que en el caso
152
del esquema correspondiente al Eu la obtención de los CSCFs no sería sencilla en absoluto,
ya que no solo se debe tener en cuenta cada posible coincidencia γ-γ, sino que también cada
cascada en el lado del 152Sm es probable que produzca coincidencia de las emisiones γ con los
rayos X emitidos. Existen varias publicaciones en donde se deducen las ecuaciones que
permiten calcular los CSCFs para emisiones gamma correspondientes a radionucleidos de
interés a partir de sus esquemas de desintegración y de las FEPEs y eficiencias totales
absolutas, como por ejemplo Xhixha et al. (2016) y Tomarchio and Rizzo (2011). Así, por
ejemplo, en el segundo trabajo se ha obtenido, entre otras, la ecuación correspondiente a la
línea de 964.1 keV del 152Eu antes mencionada:
Algoritmos Evolutivos
1 0
2 0
⋮
0 [47]
1 1 1
2 2 2
⋮
Por lo tanto, hay una única función objetivo a minimizar, funciones restricción que
limitan el espacio de búsqueda y m variables de decisión , las cuales definen los valores de
la función objetivo y funciones restricción, formando conjuntamente el vector de decisión ,
Capítulo 1 – Introducción 47
que representa a una solución candidata o individuo de la población. La optimización mono-
objetivo busca, por lo tanto, encontrar el vector de decisión que minimice el valor de la
función objetivo sin contradecir ninguna de las restricciones.
Por otra parte, muchos problemas prácticos de optimización tienen varios objetivos que
cumplir a la vez, lo que se conoce como optimización multi-objetivo, pudiéndose expresarse
matemáticamente como:
1
, 2
,…,
1 0
2 0
⋮
0 [48]
1 1 1
2 2 2
⋮
≻≻ ⇔ ∀ 1,2, … , [49]
Por último, si se quiere emplear un EA para maximizar una función objetivo concreta,
ya sea en un problema de optimización multi-objetivo o mono-objetivo, tan solo es necesario
multiplicar dicha función objetivo por -1, dado que la maximización de una función
equivale a la minimización de . Por otra parte, cualquier función restricción puede
ser convertida en 0 ya que, 0⇔ 0 y 0⇔ 0∩
0.
Capítulo 2
a r t i c l e i n f o a b s t r a c t
Article history: The determination in a sample of the activity concentration of a specific radionuclide by gamma spec-
Received 21 January 2015 trometry needs to know the full energy peak efficiency (FEPE) for the energy of interest. The difficulties
Received in revised form related to the experimental calibration make it advisable to have alternative methods for FEPE deter-
18 April 2015
mination, such as the simulation of the transport of photons in the crystal by the Monte Carlo method,
Accepted 19 June 2015
which requires an accurate knowledge of the characteristics and geometry of the detector. The charac-
Available online 16 July 2015
terization process is mainly carried out by Canberra Industries Inc. using proprietary techniques and
methodologies developed by that company. It is a costly procedure (due to shipping and to the cost of the
Keywords:
Gamma-ray spectrometry
process itself) and for some research laboratories an alternative in situ procedure can be very useful. The
Monte Carlo method main goal of this paper is to find an alternative to this costly characterization process, by establishing a
Differential evolution method for optimizing the parameters of characterizing the detector, through a computational procedure
Efficiency calibration which could be reproduced at a standard research lab. This method consists in the determination of the
Characterization germanium detectors detector geometric parameters by using Monte Carlo simulation in parallel with an optimization process,
based on evolutionary algorithms, starting from a set of reference FEPEs determined experimentally or
computationally. The proposed method has proven to be effective and simple to implement. It provides a
set of characterization parameters which it has been successfully validated for different source-detector
geometries, and also for a wide range of environmental samples and certified materials.
© 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jenvrad.2015.06.017
0265-931X/© 2015 Elsevier Ltd. All rights reserved.
J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18 9
Monte Carlo simulations have been widely used as a computa- increases the cost of the equipment, but it has the advantage that,
tional alternative to empirical calibration to avoid these problems. once characterized the detector, it can be used to generate cali-
The Monte Carlo methods are numerical techniques based on the bration by software for a wide range of geometries and composi-
creation of a stochastic model of the system under study, so that the tions of the samples and containers.
desired values of the physical quantities characterizing the system Due to its high cost and time consumed in the process, it is not a
are derived from the expected values of certain random variables common practice to characterize detector once installed in the
(Abbas et al., 2002). These variables are obtained in turn as com- laboratory. Furthermore the characterization processes is based on
binations of other random variables describing the different pro- the properties and dimensions of the HPGe detector at the time of
cesses occurring in the system. In the case of gamma spectrometry, the characterization. Therefore, any changes in crystal properties
physical processes to be considered are the transport and interac- (e.g. it is known that the thickness of the dead layer increases over a
tion of gamma rays into the sample itself and detector. For this, it is period of years if the detector is not kept refrigerated at the proper
needed to know the physical characteristics of the source and de- temperature) could induce an error in the efficiency calculations.
tector, and to implement a physical model for the interaction of This implies that the detector must be characterized again over
photons into the detector based on atomic processes involved time, with the disadvantages that this entails. Another drawback of
(photoelectric absorption, Compton scattering and pair creation), the characterization process is that no commercial procedure has
and the coefficients of absorption and attenuation of the different been developed to characterize some types of detectors, such as the
media involved. Calibration methods based on Monte Carlo have well detectors. Therefore a methodology to characterize a detector
been applied to a wide variety of source distributions and mea- in the laboratory, using efficiency data provided by another pro-
surement conditions (Ewa et al., 2001; Helmer et al., 2004; Hurtado cedure (theoretical or experimental) can be a useful tool.
et al., 2004; Kamboj and Kahn, 2003; McNamara et al., 2012; This paper shows a new methodology developed for computa-
Nikolic et al., 2014; Salgado et al., 2006; Sima and Arnold, 2009; tional characterization of an Extended Range Germanium Hyper-
van der Graaf et al., 2011), even to calibrate portable HPGe de- pure detector (HPGe XtRa) by obtaining the geometrical
tectors for environmental monitoring (Boson et al., 2009; characteristics of an equivalent model of the detector, which can be
Gutierrez-Villanueva et al., 2008; Hurtado and Villa, 2010; used in Monte Carlo calculations of efficiency calibration curves.
Rostron et al., 2014; Zhang et al., 2013). These simulation tech- This methodology only requires knowing an accurate set of
niques allow to establish a summing free efficiency or to correct for experimental or theoretical FEPEs for a given spectrometer and one
self-absorption effects (Dababneh et al., 2014; Dziri et al., 2012; or several source-detector geometries and sample's material. These
Laborie et al., 2000; Vargas et al., 2002), also accurate efficiencies FEPEs can be obtained at the laboratory through a standard cali-
have been obtained by optimizing of several sample factors, as bration procedure.
density, matrix composition or geometrical configuration (Huy
et al., 2012; Kaminski et al., 2014; Yang et al., 2013). The unknow- 2. Material and methods
ing of detector geometry parameters has been reported as one of
main causes of discrepancies in the efficiency determination 2.1. Instrumentation, computational codes to FEPE calculation and
(Nikolic et al., 2014), then several Mote Carlo model have been certified reference sources
focused in characterizing the detector geometry (Boson et al., 2008;
Budjas et al., 2009; Courtine et al., 2008; Hardy et al., 2002; Huy The gamma-ray spectrometry equipment used in this study is a
et al., 2007; Maleka and Maucec, 2005; Rodenas et al., 2003; Canberra XtRa HPGe detector, model GX3518. The detector has 38%
Szentmiklo si et al., 2014). of relative efficiency in relation to an NaI(Tl) detector with an active
Thus a key element for calibration by Monte Carlo simulation is area of 3 300 and nominal FWHM of 0.875 keV at 122 keV and
the accurate knowledge of the physical and geometrical charac- 1.8 keV at 1.33 MeV and it works coupled to a DSA-1000 Canberra
teristics of the detector such as length and diameter of the Ge multichannel analyzer. The detector is shielded with Fe 15 cm thick
crystal, thickness of the dead layers, dimensions of the package and and located in a room with walls and ceiling made of concrete in
window, etc. The process to determine these parameters is usually the ground of a three plants building. The spectral analysis is per-
known as characterization of the detector. This characterization formed using the Canberra Genie 2000 software package
procedure increases the cost of the equipment but it has the (Canberra, 2002). The detector has been characterized to permit the
advantage that, once characterized the detector, it can be used to efficiency calibration by the Canberra's LabSOCS (Laboratory
generate calibration by software for a wide range of geometries and SOurceless Calibration Software) code (Canberra, 2013) that per-
compositions of the samples and containers without the need for forms mathematical efficiency calibrations of Ge detectors, without
reference sources (Bronson, 2003; Cornejo Díaz and Jurado Vargas, any use of radioactive sources by the laboratory user. The Can-
2008; Venkataraman et al., 2005). berra's characterization methodology uses an experimental pro-
The characterization of the detector is usually performed before cedure to analyse the response of the detector to sources in the
the acquisition of the system and combines the use of experimental vicinity thereof over a range of photon energy of 10 keV to 7 MeV,
measurements of FEPEs and Monte Carlo Simulation. An initial combined with a computational process to determine the detector
model of the specific detector to be characterized is created by parameters using the MCNP Monte Carlo Code (Briesmesiter, 1992;
using the nominal dimensions provided by the manufacturer and it X5 Monte Carlo Team, 2008). As result, a characterization file with a
is used as input of a Monte Carlo code. This detector model is then comprehensive array of point-source efficiencies in different loca-
validated by comparing the FEPEs provided by the Monte Carlo tions around detector is obtained. This information is exclusive for a
code with experimental efficiencies obtained for several source- given detector and it has to be introduced as a LabSOCS input file. In
detector geometries and samples material in a wide range of en- this work the LabSOCS code was used to generate the FEPEs used as
ergies. If needed, changes are made in the dimensions of the de- reference values, but any other method could be used to determine
tector, and new computational FEPEs are generated by the Monte them such as experimental procedures based on the use of Certified
Carlo code. The detector model is considered valid if the compu- Reference Material (CRM).
tational efficiencies are in agreement with the experimental ones To develop the computational characterization methodology
within a level of acceptable uncertainty. This characterization proposed in this work we have used the PENELOPE code (Salvat
procedure requires several weeks of work by the manufacturer and et al., 2001, 2011; Sempau et al., 1997; Sempau et al., 2003), a
10 J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18
simulation code of transport for electrons and photons based on the Table 2
Monte Carlo method. PENELOPE provides to the user control over IAEA Certified Reference Materials. Centesimal composition and density (IAEA,
1987).
the parameters of the numerical model, allowing changing the
geometry and materials of the detector, which is essential for our IAEA-RGK-1 IAEA-RGU-1 IAEA-RGTh-1
purpose. The computer code PENELOPE consists of a set of sub- Density 1.33 g/cm3 1.1 g/cm3 1.25 g/cm3
routines written in FORTRAN that allows performing Monte Carlo Centesimal composition O:36.72% O:53.80% O:47.95%
simulation of coupled electronephoton transport in arbitrary ma- S:18.40% Na:0.10% Si:42.07%
K:44.87% Al:0.10% Th:2.89%
terials for a wide energy range, from a few hundred eV to about
Si:45.70% U:7.09%
1 GeV. For a specific geometry, the user has to provide a MAIN Ca:0.10%
program to control the entire process and the evolution of each Fe:0.10%
event. In this work, given the cylindrical geometry of the detector, it U:0.10%
has been used as the main program, the pencyl.f subroutine created
for the simulation of particle transport in cylindrical geometries
and included into the PENELOPE distribution package. The code
requires the information contained in several input files, to specify
the geometry of the detector, active bodies (areas where ionizing
radiation that comes to the detector is recorded), sample geometry,
materials, type of radiation emitted sample (g or b) and its corre-
sponding energy, the duration of the simulation, etc. As results the
code generates an output with the FEPEs obtained from the simu-
lation for any energy and its statistical uncertainty.
To validate results of the characterization procedure, we have
determined the activity concentration of several certified reference
materials covering the whole range of energy for environmental
samples. For this purpose, the IAEA reference standards RGK-1
(Potassium Sulfate), RGU-1 (Uranium Ore) and RGTh-1 (Thorium
Ore), whose characteristics are summarized in Tables 1 and 2, have
been used (IAEA, 1987).
Table 1
IAEA Reference Materials certified values used to validate the results of the characterization procedure (IAEA, 1987). RGK-1 (Potassium Sulphate), RGU-1 (Uranium Ore) and
RGTh-1 (Thorium Ore).
Sample Nuclides & Chemical elements Magnitude Unit Uncertainty 95% C.I.
40
IAEA-RGK-1 K 14,000 Bq/kg 13,600e14,400
K 448,000 mg/kg 445,000e4,51,000
Th <0.01 mg/kg e
U <0.001 mg/kg e
232
IAEA-RGU-1 Th <4 Bq/kg e
235
U 228 Bq/kg 226e230
238
U 4940 Bq/kg 4910e4970
40
K <0.63 Bq/kg e
K <20 mg/kg e
Th <1 mg/kg e
U 400 mg/kg 398e402
232
IAEA-RGTh-1 Th 3250 Bq/kg 3160e3340
235
U 3.6 Bq/kg 3.3e3.9
238
U 78 Bq/kg 72e84
40
K 6.3 Bq/kg 3.1e9.5
K 200 mg/kg 100e300
Th 800 mg/kg 784e816
U 6.3 mg/kg 5.9e6.7
J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18 11
Table 4
Geometrical characteristics specified by the manufacturer (Canberra) used in the equivalent model of the HPGe XtRa detector. In the third column, the dimensions without to
specify (e) are the geometrical variables selected for the optimization problem. Columns 4th to 6th shows the assignation of the unknown geometric characteristics to the
variables of the optimization problem (xgp;m ) and the minimum (xmin
m ) and maximum (xm
max ) values of each variables. Best solution obtained by the optimization process is
a) Initialization c) Crossover
Population is initialized (generation g ¼ 0), by random genera- Once NP mutant vectors have been generated, the process of
tion of each individual x0p through the following equation, crossover starts. This step consists in generating new candidate
J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18 13
g
solutions called trial vectors t p by randomly mixing the mutant ten times the number of variables n. The weighting factor F varies
vectors ngp with the original vectors xgp using the next rule: from 2 to 0, according to the Eq. (4). Also, as stop criterion a
( maximum of 100 generations has been adopted. The computer time
g ngp;m if randð0; 1Þ < CR employed to perform the optimization process was around 134 h
tp;m ¼ g ; for p ¼ 1…NP; m ¼ 1…n (6)
xp;m in any other case (5.5 days) using a 3.1 GHz i7 processor. The geometrical parameters
obtained by the best solution are shown in the last column of
Where 0 < CR < 1 is a parameter that controls the rate of crossover Table 4, and the minimum value of the optimization function of Eq.
g
and it must be defined previously. Since the comparison is made (1) is fopt ðxpopt Þ ¼ 5:3698 105 .
variable by variable, (component by component of vectors), the To illustrate the goodness of this solution, in Table 5 we compare
trial vectors built by this operator are a mixture between mutant the reference FEPEs with the set of FEPEs generated by PENELOPE
and original vectors. using the geometrical characterization values obtained. In the last
column are shown the relative differences, in absolute value, be-
d) Selection tween these efficiencies using as reference the LabSOCS value. As it
can be appreciate there is a good agreement between both results.
The last step of each iteration is the selection, which essentially The maximum relative difference is lower than the 0.5% in the
consists in comparing the evaluations of Eq. (1) for the trial vectors, whole range studied. In fact this maximum difference is lower than
g g
t p , and for the original vectors, xp . The vectors selected for the next
generation, xgþ1 p ,will be the ones that have the best (lowest) values
of the function to optimize (fopt) as stated the following equation,
( g g Table 5
g
gþ1 t p if fopt t p < fopt xp Comparison between the FEPEs obtained for each energy by PENELOPE (εPEN) and
xp ¼ ; for p ¼ 1…NP (7)
xgp in any other case LabSOCS (εLab). In the right column are shown the relative differences between these
efficiencies using as reference the LabSOCS values.
the uncertainty presented by the reference FEPEs computed by shows the calibration curves obtained by both programs for
LabSOCS. different heights of the sample. The relative differences between
A continuous calibration curve for the detector efficiency has the calibration curves as a function of energy are shown in Fig. 5b. It
also been obtained from the PENELOPE FEPEs by a least squares fit observed the good response of the characterization under changes
to a fourth degree logarithmic polynomial as usual (Debertin and in the sample height. The discrepancies are in any case lower than a
Helmer, 1988), through the following expression: 2.1%, with maximum difference at 55 keV for 10 mm of sample
height. We would like to point out that no systematic deviations as
X
i¼n the difference between the height of the sample and the charac-
ln ε ¼ ai lni E (8) terization height (25 mm) increases.
i¼0
The second set of tests has been performed by computing a
A coefficient of determination (R2) of 0.9994 has been obtained, series of FEPEs using the same material and height of the sample
its value, very close to the unit, confirms the goodness of fit. The employed for the characterization (25 mm of distilled water), but
logarithmic curve is useful for fitting, but the goal of the calibration with containers of different diameters. The geometry of the con-
is to establish an equation of the efficiency in terms of energy. This tainers used is shown in Fig. 6. The calibration curves obtained by
equation ε ¼ f(E) derived from Eq. (8) is LabSOCS and PENELOPE codes and its relative differences for the
different containers are showed in Fig. 7. The maximum value of the
εðEÞ ¼ exp 23:3985 þ 13:6627 lnE 3:0235 ln2 E residuals (<1.8%) is obtained for energies around 1800 keV in the
case of the container R2.
þ 0:2716 ln3 E 0:0088 ln4 E (9) The results obtained confirm the stability of the characterization
obtained for changes in the dimension of the sample for a cylin-
where the energies are expressed in keV. We have also constructed drical geometry. Another kind of containers such as Marinelli
an energyeefficiency curve for the LabSOCS data following the beakers which are wrapped around the detector could require a
same procedure. Both efficiency curves have been plotted in Fig. 4a. better model of the detector to include features such as the crystal
As it can be observed, the difference between the efficiency curves subjection brackets. Nevertheless the mathematical model devel-
is negligible. To estimate the discrepancy between the curves, it has oped in this work can be easily adapted for that task.
been computed the relative difference for these curves taking again
as reference the LabSOCS values (Fig. 4b). The residuals does not
exceed, in any case a 0.25% in absolute value, appearing the 2.3.2.2. Experimental validation by certified materials.
maximum deviation around photopeak energies of 45 keV and Computational procedures of calibration have to be experimentally
1800 keV as it is shown in Fig. 4b. validated. To carry it out we have determined the activity concen-
trations of three standard samples supplied by the International
Atomic Energy Agency (IAEA) referred in section 2.1 using the
2.3.2. Validation of the characterization calibration obtained from the characterization. To determine the
The geometrical parameters of the detector model obtained can calibration curve it is necessary to supply PENELOPE with the
then be used as characterization input data in order generate cal- density and composition of the samples which was summarized in
ibrations using computational methods to any geometry, material, Table 2. The samples are enclosed in a standard frustoconical
etc. of the sample. Since characterization has been obtained medical sample containers with a sample height of 25 mm. How-
considering a single type of sample (distilled water) with a unique ever, as stated above, the calibrations have been performed
geometry, it is necessary to check the goodness of the character- assuming an equivalent cylindrical vessel with the same sample
ization for different samples with different materials and geometry. height.
The acquisition and analysis was performed by means of the
2.3.2.1. Internal validation: influence of the geometry of the source. GENIE 2000 software using the efficiencyecalibration curve ob-
A comparison between the efficiency obtained by PENELOPE and tained by PENELOPE from the characterization that was intended to
LabSOCS has been performed for different geometries of the validate. In addition, spectrometric analyses were also performed
sample-detector to carry out an internal validation of the consis- using the LabSOCS calibration based on the supplier characteriza-
tency of the results. The first set of calibrations were performed tion to compare the results from both calibrations.
assuming the same container (Fig. 1) and the same sample material The activity concentration of 238U was calculated from the
(distilled water) that were used for characterization, but with 351.95 keV line of 214Pb, considering that both are in secular equi-
different heights of the sample (10, 20, 30, 40 and 50 mm). Fig. 5a librium. The specific activity of 232Th was calculated from the
Fig. 4. (a) Energyeefficiency curves obtained by least squares fit to Energyeefficiency points calculated by PENELOPE and LabSOCS. (b) Relative differences between PENELOPE and
LabSOCS curves taking as reference the LabSOCS values.
J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18 15
Fig. 5. (a) Efficiency calibration curves obtained by PENELOPE and LabSOCS for different heights of the sample (distilled water) in the container (b) Relative differences between
PENELOPE and LabSOCS curves for different heights, taking as reference the LabSOCS values.
Fig. 6. Dimensions of the three containers (R1, R2 and R3) used in the second set of test for validation of the characterization for changes in the dimension of the sample (distilled
water).
emission line of 2614.5 keV of 208Tl. In this case it has been et al., 1979). For two values, a ± ua and b ± ub, utest value is given
considered that the 212Bi is in secular equilibrium with 232Th, and by the following equation:
taking into account that 33.7% of 212Bi decays to 208Tl. In the case of
40
K it was measured directly through its photopeak energy ja bj
utest ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (10)
1460.8 keV.
u2a þ u2b
The results of the spectrometric analysis performed using cali-
brations by PENELOPE and LABSOCS are compiled in Table 6, where
where the ua and ub are the standard uncertainties for the values a
a good agreement with respect to the certified value is shown. To
and b, respectively. Interpretations of utest value vary with the
check the goodness of the characterization by comparison of the
bibliography but the limits in Table 7 are the more general found in
calculated activities from spectrometric analysis and the certified
the literature. However, according to the IAEA (Shakhashiro et al.,
activities, we have used a test called u-score or u-test (Brookes
2008), results are considered satisfactory if the value of 2.58 is
not exceeded.
Fig. 7. (a) Efficiency calibration curves obtained by PENELOPE and LabSOCS for the three containers of Fig. 6. (b) Relative differences between PENELOPE and LabSOCS curves for
different containers, taking as reference the LabSOCS values.
16 J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18
Table 6
Results of the spectrometric analysis performed by GENIE 2000 (uncertainties given as 1s.) using calibrations by PENELOPE (third column) and LABSOCS (fourth column), and
the certified activities (fifth column) expressed all in Bq/kg.
Sample Nuclide Activity concentration PEN Activity concentration Lab Activity concentration Cert
40
IAEA-RGK-1 K 13,836 ± 672 13,750 ± 657 14,000 ± 400
238
IAEA-RGU-1 U 5084 ± 511 4929 ± 472 4940 ± 30
232
IAEA-RGTh-1 Th 3026 ± 234 3083 ± 184 3250 ± 90
Table 7
u-test limits commonly used.
Condition Status
1.64 > utest The reported result does not differ significantly from the expected value.
1.64 < utest < 1.96 The reported result probably does not differ significantly from the expected value.
1.96 < utest < 2.58 It is not clear whether the reported result differs significantly from the expected value.
2.58 < utest < 3.29 The reported result is probably significantly different from the expected value.
3.29 < utest The reported result is significantly different from the expected value.
Table 8 shows the u-test results for the three samples analysed. computational characterization method for determining the ac-
The uncertainties in the determination of the activity concentra- tivity of real environmental samples.
tions are computed by GENIE 2000 through a standard procedure Table 9 shows the results obtained for the concentration activ-
(Canberra, 2009) with 1s of confidence interval. In fact three cross ities of four radionuclides of interest in environmental applications
tests among the results of LabSOCS and PENELOPE as well as with (226Ra, 228Ra, 228Th and 40K). To compare both results a linear cor-
the corresponding IAEA certified values have been performed. As it relation analysis has been performed and its results for the four
can be observed the values of the u-test are considerably lower nuclides are compiled in Fig. 8. Fig. 8a shows graphically the cor-
than the limit 1.64 in all cases. So, the reported results do not differ relation between measures of 226Ra. The value of the slope
significantly from the expected values, and therefore, the results (1.07 ± 0.07) very close to unity indicates that the results are
are satisfactory for all the samples. In consequence it can be strongly correlated. This good correlation is also confirmed by the
concluded that, for the cylindrical geometry analysed, both, the value close to unity (0.977) of the coefficient of determination (R2)
characterization provided by Canberra and the computational indicating a very good agreement between both detectors. The
characterization obtained in this work, are valid to determine the value of 3.1 ± 2.7 of intercept indicates that there is virtually no
activity concentration of these samples in the same degree of systematic error in the background characterization of the two
accuracy. detectors because, considering the uncertainty, the intercepts is
practically zero. Similar results are obtained for the rest of radio-
isotopes as it is shown Fig. 8bed. In consequence there is a ten-
2.3.2.3. Experimental validation by environmental samples.
dency to obtain similar results in both laboratories despite of the
Finally, to corroborate the statement in the last paragraph of the
differences in the spectrometry equipment used and in the cali-
previous subsection, it has been determined the activity concen-
bration methods.
tration of several environmental radioisotopes for a set of soil
samples collected in the Eastern Canary Islands using the compu-
tational characterization. The activity concentration of these sam- 3. Summary and conclusions
ples had been previously determined by the Laboratory of
Environmental Radioactivity of the FRyMA group at the Depart- In this work a computational method for characterizing an HPGe
ment of Applied Physics of the University of Huelva. The equipment spectrometer by implementing in parallel an evolutionary algo-
used for the analysis was an ORTEC HPGe spectrometer with 27% rithm (differential evolution) together with a Monte Carlo simula-
relative efficiency and a FWHM of 0.68 keV at 122 keV (57Co) and tion code (PENELOPE) has been developed. This procedure is
1.73 keV at 1333 keV (60Co). This equipment uses the MAESTRO designed to characterize the detector using as reference a set of
software for spectra analysis and acquisition and it was experi- photopeak efficiencies obtained experimentally by measuring
mentally calibrated using certified reference sources (Pe rez- standard samples activity or by Monte Carlo calibration.
Moreno et al., 2001, 2002). This test therefore allow us to The characterization obtained has been validated by performing
compare the results of the computational characterization obtained computational calibrations for samples of distilled water with
in this work for the Canberra HPGe XtRa spectrometer described in different geometries than those employed in the characterization
Section 2, with the results provided by a completely different (several heights and to different diameters of the sample container)
spectrometer both in hardware and software as well as in the by the computational codes PENELOPE and LabSOCS. It has been
calibration method. It also permits to analyse the goodness of observed that the relative differences between the results obtained
from both Monte Carlo simulation programs are always keeping
below than 2.1%. This value is lower than the minimum uncertainty
Table 8 considered in LabSOCS (4% for energies above 900 keV), based on
Results of the utest for the values of activity concentration obtained calibrating by
experimentation, illustrating the goodness of the characterization
LabSOCS (Lab.) and calibrating by the model proposed in his work (PEN.) compared
with the certified values of the IAEA samples (Cert.).
in front to changes in the geometry of the sample.
A validation of the stability of the characterization using mate-
Sample PEN.-Lab. Lab.-Cert. PEN.-Cert.
rials different than those considered during characterization (with
IAEA-RGK-1 0.09 0.33 0.21 different density and percentage composition) has also been car-
IAEA-RGU-1 0.22 0.02 0.28 ried out. Specifically three standard samples certified by the IAEA
IAEA-RGTh-1 0.19 0.82 0.90
have been used. In this case the maximum relative difference
J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18 17
Table 9
Activity concentrations determined by the Laboratory of Environmental Radioactivity of the FRyMA group at the Department of Applied Physics of the University of Huelva and
activity concentrations obtained from the efficiency calibration of this work by PENELOPE. Assuming the condition of secular equilibrium, activity concentration of 226Ra was
determined by using the gamma line 351.90 keV from 214Pb, for 228Ra its activity is computed by using the gamma line 911.2 keV from 228Ac, the activity of 228Th is obtained
from the line 583.2 keV from 208Tl, while in the case of 40K it was measured directly through its photopeak energy 1460.8 keV.
226 228 228 40
Sample Ra (Bq/kg) Ra(Bq/kg) Th (Bq/kg) K(Bq/kg)
FRyMA This work FRyMA This work FRyMA This work FRyMA This work
S1 71.6 ± 4.2 80.6 ± 8.1 136.0 ± 8.0 135.0 ± 7.0 122.0 ± 7.0 131.6 ± 10.8 1162.0 ± 69.0 1071.0 ± 46.6
S2 35.9 ± 2.2 48.4 ± 4.8 51.3 ± 3.3 48.8 ± 3.0 39.5 ± 2.5 44.4 ± 4.1 568.0 ± 35.0 526.7 ± 25.5
S3 6.9 ± 0.5 8.7 ± 1.2 15.6 ± 1.1 11.3 ± 1.3 12.9 ± 0.9 13.8 ± 1.4 348.0 ± 21.0 273.8 ± 14.6
S4 15.7 ± 1.0 20.5 ± 2.1 42.0 ± 3.0 42.7 ± 2.5 34.6 ± 2.2 34.4 ± 3.2 920.0 ± 60.0 860.0 ± 39.3
S5 20.2 ± 1.3 26.9 ± 2.6 60.0 ± 4.0 58.6 ± 4.0 48.0 ± 3.0 53.5 ± 4.6 980.0 ± 60.0 986.2 ± 44.2
S6 46.0 ± 3.0 47.7 ± 4.7 58.0 ± 4.0 52.7 ± 3.0 53.0 ± 3.0 55.7 ± 4.9 e e
S7 28.5 ± 1.7 30.7 ± 3.0 39.0 ± 3.0 36.6 ± 2.3 35.3 ± 2.2 35.5 ± 2.9 690.0 ± 40.0 539.9 ± 25.5
S8 38.6 ± 2.4 44.2 ± 4.5 90.0 ± 6.0 85.3 ± 4.7 72.0 ± 4.0 77.7 ± 8.4 1350.0 ± 80.0 1223.4 ± 55.2
Fig. 8. Correlation between FRyMA results and those obtained by means of the efficiency calibration of this work. SEE is the Standard Error of Estimate.
observed between the efficiencies calculated by both programs and 2012, by the European Regional Development Fund (ERDF)
(LabSOCS and PENELOPE) is 2.5%. The activity concentrations for through the R&D project PI2007/035 granted by the “Agencia
the references material computed using the characterization values Canaria de Investigacion Innovacio
n y Sociedad de la Informacio
n
obtained have shown a good agreement with the certified values. (ACIISI)” of the regional Government of Canary Islands (Spain) and
This is corroborated by the results of the u-test, whose values have by the “Ministerio de Economía y Competitividad” through the
not exceeded the limit of 1.64 which guarantees that activity con- project UNLP10-3E-1108 and also by the Government of Andalusia
centrations obtained do not differ significantly from the certified through the project “Characterization and modelling of the phos-
values. These facts illustrated that the characterization obtained phogypsum stacks from Huelva for their environmental manage-
remains valid for changes in the sample material. To study the ment and control” (Ref.: P10-RNM-6300). This is a publication No.
feasibility of the characterization obtained for the analysis of 103 from CEIMAR Publication Series.
environmental samples, activity concentrations of various soil
samples were determined by obtaining very good agreement with
results from other laboratories. Thus, the proposed method gives References
accurate results and can be an effective and inexpensive alternative
to characterize high resolution spectroscopy detectors without Abbas, K., Simonelli, F., D'Alberti, F., Forte, M., Stroosnijder, M., 2002. Reliability of
two calculation codes for efficiency calibrations of HPGe detectors. Appl. Radiat.
using complex experimental setups.
Isotopes 56, 703e709.
Boson, J., Agren, G., Johasson, L., 2008. A detailed investigation of HPGe detector
Acknowledgements response for improved Monte Carlo efficiency calculations. Nucl. Instrum.
Methods A 587, 304e314.
Boson, J., Plamboeck, A.H., Rameba €ck, H., Ågren, G., Johansson, L., 2009. Evaluation
This work has been partially financed by the Nuclear Safety of Monte Carloebased calibration of HPGe detectors for in situ gamma-ray
Council of Spain (CSN) through a grants in its R&D programs 2009 spectrometry. J. Environ. Radioact. 100, 935e940.
18 J. G. Guerra et al. / Journal of Environmental Radioactivity 149 (2015) 8e18
Briesmesiter, J., 1992. MCNP v4.2, Monte Carlo Code and Photon Transport Code gamma-ray spectrometry using well-type HPGe detectors. Appl. Radiat. Isot. 53,
System, RSIC Computer Code Collection, CCC-200 A. Oak Ridge National 57e62.
Laboratory. Maleka, P.P., Maucec, M., 2005. Monte Carlo uncertainty analysis of germanium
Britton, R., Burnett, J.L., Davies, A.V., Regan, P.H., 2014. Maximising the sensitivity of detector response to gamma rays with energies bellows 1 MeV. Nucl. Instrum.
a g spectrometer for low-energy, low-activity radionuclides using Monte Carlo Methods A 538, 631e639.
simulations. J. Environ. Radioact. 134, 1e5. McNamara, A.L., Heijnis, H., Fierro, D., Reinhard, M.I., 2012. The determination of the
Bronson, F.L., 2003. Validation of the accuracy of the LabSOCS software for math- efficiency of a Compton suppressed HPGe detector using Monte Carlo simula-
ematical efficiency calibration of Ge detectors for typical laboratory samples. tions. J. Environ. Radioact. 106, 1e7.
J. Radioanal. Nucl. Chem. 255 (1), 137e141. Nikolic, J., Vidmar, T., Jokovic, D., Rajacic, M., Todorovic, D., 2014. Calculation of
Brookes, C., Betteley, I., Loxton, S., 1979. Fundamentals of Mathematics and Statis- HPGe efficiency for environmental samples: comparison of EFFTRAN and
tics. John Wiley, New York, USA. GEANT4. Nucl. Instrum. Methods A 763, 347e353.
Budjas, D., Heisel, M., Maneschg, W., Simgen, H., 2009. Optimization of the MC- rez-Moreno, J.P., Bolívar, J.P., García-Tenorio, R., San Miguel, E., Aguado, J.L.,
Pe
model of a p-type Ge-spectrometer for the purpose of efficiency determina- Mas, J.L., Vaca, F., 2001. A simple method for efficiency calibration of HPGe
tion. Appl. Radiat. Isot. 67, 706e710. detectors in g-spectrometric measurements. Radiat. Phys. Chem. 61, 437e438.
Canberra, 2002. Genie 2000 Basic Spectroscopy Software Version 2.1. Canberra rez-Moreno, J.P., San Miguel, E.G., Bolívar, J.P., Aguado, J.L., 2002. A comprehensive
Pe
Industries Inc. calibration method of Ge detectors for low-level gamma-spectrometry mea-
Canberra, 2009. Genie 2000 3.2. Customization Tolls Manual. Camberra Indsutries surements. Nucl. Instrum. Methods Phys. Res. A 491, 152e162.
Inc. Rodenas, J., Pascual, A., Zarza, I., Serradell, V., Ortiz, J., Ballesteros, L., 2003. Analysis
Canberra, 2013. Model S574 LabSOCS™ Calibration Software. Canberra Industries of the influence of germanium dead layer on detector calibration simulation for
Inc. environmental radioactive samples using the Monte Carlo method. Nucl. Ins-
Cornejo Díaz, N., Jurado Vargas, M., 2008. DETEFF: an improved Monte Carlo trum. Methods A 496, 390e399.
computer program for evaluating the efficiency in coaxial gamma-ray de- Rostron, P.D., Heathcote, J.A., Ramsey, M.H., 2014. Optimising in situ gamma mea-
tectors. Nucl. Instrum. Methods Phys. Res. A 586, 204e210. surements to identify the presence of radioactive particles in land areas.
Courtine, F., Pilleyre, T., Sanzelle, S., Miallier, D., 2008. Ge well detector calibration J. Environ. Radioact. 138, 162e169.
by means of a trial and error procedure using the dead layers as a unique Salgado, C.M., Conti, C.C., Becker, P.H.B., 2006. Determination of HPGe detector
parameter in a Monte Carlo simulation. Nucl. Instrum. Methods A 596, response using MCNP5 for 20e150 keV X-rays. Appl. Radiat. Isot. 64, 639e664.
229e234. Salvat, F., Fernandez, J.M., Acosta, E., Sempau, J., 2001. PENELOPE, a code system for
Dababneh, S., Ai-Nemri, E., Sharaf, J., 2014. Application of Geant4 in routine close Monte Carlo simulation of electron and photon transport. In: Proceedings of a
geometry gamma spectroscopy for environmental samples. J. Environ. Radioact. Workshop/Training Course, OECD/NEA 5e7 November 2001 NEA/NSC/DOC 19,
134, 27e34. ISBN 92-64-18475-9.
Das, S., Suganthan, P.N., 2011. Differential evolution: a survey of the state-of-the-art. Salvat, F., Fern andez-Varea, J.M., Sempau, J., 2011. PENELOPE-2011: a Code System
IEEE Trans. Evol. Comput. 15, 4e31. for Monte Carlo Simulation of Electronic and Photon Transport. OECD.
Debertin, K., Helmer, R.G., 1988. Gamma and X-Ray Spectrometry with Semi- Sempau, J., Acosta, E., Baro, J., Fernandez-Varea, J.M., Salvat, F., 1997. An algorithm
conductor Detectors, ISBN 0444871071 (North-Holland). for Monte Carlo simulation of the coupled electron-photon transport. Nucl.
Dziri, S., Nourreddine, A., Sellam, A., Pape, A., Baussan, E., 2012. Simulation approach Instrum. Methods B 132, 377e390.
to coincidence summing in g-ray spectrometry. Appl. Radiat. Isot. 70, Sempau, J., Fernandez-Varea, J.M., Acosta, E., Salvat, F., 2003. Experimental benchmarks
1141e1144. of the Monte Carlo code PENELOPE. Nucl. Instrum. Methods B 207, 107e123.
Ewa, I.O.B., Bodizs, D., Czifrus, Sz, Molnar, Zs, 2001. Monte Carlo determination of Shakhashiro, A., Trinkl, A., Sansone, U., 2008. The IAEA's ‘ALMERA Network’ profi-
full energy peak efficiency for a HPGe detector. Appl. Radiat. Isot. 55, 103e108. ciency test on the determination of gamma-emitting radionuclides: a test of
Gutierrez-Villanueva, J.L., Martin-Martin, A., Pena, V., Iniguez, M.P., de Celis, B., results comparability. Appl. Radiat. Isotopes 66, 1722e1725.
2008. Calibration of a portable HPGe detector using MCNP code for the deter- Sima, O., Arnold, D., 2009. On the Monte Carlo simulation of HPGe gamma spec-
mination of 137Cs in soils. J. Environ. Radioact. 99, 1520e1524. trometry systems. Appl. Radiat. Isot. 67, 701e705.
Hardy, J.C., Jacob, V.E., Sanchez-Vega, M., Effinger, R.T., Lipnik, P., Mayes, V.E., Storn, R., Price, K., 1997. Differential evolution e a simple and efficient heuristic for
Willis, D.K., Helmer, R.G., 2002. Precise efficiency calibration of an HPGe de- global optimization over continuous spaces. J. Glob. Optim. 11, 341e359.
tector: source measurements and Monte Carlo calculations with subpercent Szentmiklo si, L., Belgya, T., Maro
ti, B., Kis, Z., 2014. Characterization of HPGe gamma
precision. Appl. Radiat. Isot. 56, 65e69. spectrometers by geant4 Monte Carlo simulations. J. Radioanal. Nucl. Chem.
Helmer, R.G., Nica, N., Hardy, J.C., Iacob, V.E., 2004. Precise efficiency calibration of 300, 553e558.
an HPGe detector up to 3.5 MeV with measurements and Monte-Carlo calcu- van der Graaf, E.R., Limburg, J., Koomans, R.L., Tijsb, M., 2011. Monte Carlo based
lations. Appl. Radiat. Isot. 60, 173e177. calibration of scintillation detectors for laboratory and in situ gamma ray
Hurtado, S., Garcia-Leon, M., Garcia-Tenorio, R., 2004. GEAN4 code for simulation of measurements. J. Environ. Radioact. 102, 270e282.
a germanium gamma-ray detector and its application to efficiency calibration. Vargas, M.J., Ferna ndez Timo n, A., Cornejo Dıaz, N., Pe rez S
anchez, D., 2002. Monte
Nucl. Instrum. Methods A 518, 764e774. Carlo simulation of the self-absorption corrections for natural samples in
Hurtado, S., Villa, M., 2010. An intercomparison of Monte Carlo codes used for in- gamma-ray spectrometry. Appl. Radiat. Isot. 57, 893e898.
situ gamma-ray spectrometry. Radiat. Meas. 45, 923e927. Venkataraman, R., Bronson, F., Atrashkevich, V., Field, M., Young, B.M., 2005.
Huy, N.Q., Binh, D.Q., An, V.X., 2007. Study on the increase of inactive germanium Improved detector response characterization method in ISOCS and LabSOCS.
layer in a high-purity germanium detector after a long time operation applying J. Radioanal. Nucl. Chem. 264, 213e219.
MCNP code. Nucl. Instrum. Methods A 573, 384e388. X5 Monte Carlo Team, 2008. MCNP e a General Monte Carlo N-particle Transport
Huy, N.Q., Binh, D.Q., An, V.X., 2012. A study for improving detection efficiency of an Code, Version 5. Los Alamos National Laboratory. Report LA-UR-
HPGe detector based gamma spectrometer using Monte Carlo simulation and 03e1987(revised 2008).
genetic algorithms. Appl. Radiat. Isot. 70, 2695e2702. Yang, B.L., Ha, Y.M., Li, A., Zhou, H.J., Wang, F., Li, W.M., 2013. Optimisation design of
IAEA, 1987. Preparation and Certification of IAEA Gamma Spectrometry Reference cylindrical containers for improving the detection efficiency of a high-purity
Materials RGU-1, RGTh-1 and RGK-1. International Atomic Energy Agency, germanium detector using the LabSOCS. J. Radioanal. Nucl. Chem. 298,
Vienna, Austria. IAEA Report RL148. 1673e1677.
Kamboj, S., Kahn, B., 2003. Use of Monte Carlo simulation to examine gamma-ray Zhang, W., Korpach, E., Berg, R., Ungar, K., 2013. Testing of an automatic outdoor
interactions in germanium detectors. Radiat. Meas. 37, 1e8. gamma ambient dose-rate surveillance system in Tokyo and its calibration
Kaminski, S., Jakobi, A., Wilhelm, C., 2014. Uncertainty of gamma-ray spectrometry using measured deposition after the Fukushima nuclear accident. J. Environ.
measurement of environmental samples due to uncertainties in matrix Radioact. 125, 93e98.
composition, density and sample geometry. Appl. Radiat. Isot. 94, 306e313. Zhang, J., Sanderson, A., 2009. Adaptive Differential Evolution: a Robust Approach to
Laborie, J.M., Le Petit, G., Abt, D., Girard, M., 2000. Monte Carlo calculation of the Multimodal Problem Optimization. Springer Science & Business Media, New
efficiency calibration curve and coincidence-summing corrections in low-level York, USA.
́
Capítulo 3
Generalización de la metodología de
caracterización para que pueda usarse para
una mayor variedad de geometrías de
muestras (segunda publicación)
A R T I C L E I N F O A BS T RAC T
Keywords: In this work, we have developed a computational methodology for characterizing HPGe detectors by
HPGe detectors implementing in parallel a multi-objective evolutionary algorithm, together with a Monte Carlo simulation
Efficiency calibration code. The evolutionary algorithm is used for searching the geometrical parameters of a model of detector by
Monte Carlo simulation minimizing the differences between the efficiencies calculated by Monte Carlo simulation and two reference sets
Marinelli beakers
of Full Energy Peak Efficiencies (FEPEs) corresponding to two given sample geometries, a beaker of small
Differential Evolution Multi-objective
diameter laid over the detector window and a beaker of large capacity which wrap the detector. This
Characterization germanium detectors
methodology is a generalization of a previously published work, which was limited to beakers placed over the
window of the detector with a diameter equal or smaller than the crystal diameter, so that the crystal mount cap
(which surround the lateral surface of the crystal), was not considered in the detector model. The generalization
has been accomplished not only by including such a mount cap in the model, but also using multi-objective
optimization instead of mono-objective, with the aim of building a model sufficiently accurate for a wider variety
of beakers commonly used for the measurement of environmental samples by gamma spectrometry, like for
instance, Marinellis, Petris, or any other beaker with a diameter larger than the crystal diameter, for which part
of the detected radiation have to pass through the mount cap. The proposed methodology has been applied to an
HPGe XtRa detector, providing a model of detector which has been successfully verificated for different source-
detector geometries and materials and experimentally validated using CRMs.
⁎
Corresponding author.
E-mail address: jglezg2002@gmail.es (J. G. Guerra).
http://dx.doi.org/10.1016/j.nima.2017.02.087
Received 24 October 2016; Received in revised form 15 February 2017; Accepted 27 February 2017
Available online 02 March 2017
0168-9002/ © 2017 Elsevier B.V. All rights reserved.
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
detector geometrical characteristics is one of main reasons of discre- Canberra XtRa HPGe detector, model GX3518 which works together
pancies in the efficiency calculation by Monte Carlo Simulation [13]. with a DSA-1000 Canberra multichannel analyzer and that is placed
Therefore, various publications in this field have been focused on into a 15 cm thick Iron shield. The gamma analysis is performed by
performing the characterization of gamma detectors [14–18], that is to using the Genie 2000 software package [30]. The detector has been
say, to determine the geometrical characteristics of the detector. When characterized by Canberra so that computational efficiency calibrations
it is requested by its customers, this characterization is performed, can be obtained by the LabSOCS code [31]. The characterization
before the acquisition of the detector, by Canberra Industries Inc. using performed by Canberra consists mainly in studying experimentally
techniques and methodologies developed by this company. This the response of the detector to different sources situated with several
characterization increases the acquisition cost of the equipment, but, orientations in various nearby positions and over a range of energies
once characterized the detector, it allows performing computational from10 keV to 7 MeV. Such experiments are combined with a compu-
efficiency calibrations for a large variety of geometries and materials of tational process to determine, from over 30 different detector dimen-
samples without the need to use reference materials [19,20]. On the sions considered in a model implemented in the MCNP Monte Carlo
other hand, it is less frequent requesting characterizations to Canberra Code [32], the characteristics which have not been supplied by their
once the detector has been installed in the laboratory, as not only detector production facility. As a result, a characterization file exclusive
would it be costly, but also would suppose the inactivity of the for a given detector, with a comprehensive array of point-source
equipment because it would have to be disassembled and sent back efficiencies in different locations around detector is obtained and is
to the supplier. Nevertheless, having made the characterization pre- introduced as a LabSOCS input file. In this work the LabSOCS code has
viously to the HPGe detector acquisition, do not guarantee the accuracy been used to generate the FEPEs used as reference values, but any
of Monte Carlo calculations for an indefinite period, as any change in other method could be used to determine them, such as empirical
crystal properties might suppose a change in the detection efficiency. measurements using certified reference materials.
For instance, it is known that the dead layer thickness increases after a As in the mentioned previous publication, in this work the
long time operation [21–23]. All in all, a methodology to characterize a PENELOPE code has been used [25–27]. PENELOPE consists of a
detector in situ starting from reference FEPEs obtained at the set of FORTRAN subroutines that allows performing Monte Carlo
laboratory through a standard calibration procedure could be rather simulation of electron and photon transport for a wide range of
useful. materials, geometries and energies. For a specific geometry, the user
In a previous paper a methodology for computational characteriza- has to provide a MAIN subroutine in order to control the whole process
tion of HPGe detectors was developed [24]. This methodology con- and the evolution of each event. For this purpose, the pencyl.f
sisted in optimizing a model of the detector in a Monte Carlo subroutine, included into the PENELOPE distribution package and
simulation code, named PENELOPE [25–27], using for the optimiza- designed for simulations in cylindrical geometries, has been chosen
tion the Differential Evolution (DE) algorithm [28], a type of mono- given the geometry of the detector and sources. The code requires the
objective evolutionary algorithm. This methodology was applied for an information contained in several input files, where the geometry of the
Extended Range Germanium Hyperpure detector (HPGe XtRa), gen- detector and sources, active bodies (regions where the radiation is
erating a model of the detector which was successfully verificated for detected), materials, type of radiation emitted (γ or β) with its energies,
different source-detector geometries, and also valid for a wide range of the duration of the simulation, etc. must be specified.
environmental samples and certified reference materials. However, For the optimization of the detector model built in PENELOPE, the
such a methodology was restricted to those beakers with a diameter DEMO algorithm has been implemented in Matlab©, automatically
lower than the diameter of the Germanium crystal. The reason for such generating a PENELOPE input file with the geometry corresponding to
a restriction is that, the crystal mount cap, which surrounds the lateral each candidate solution and reading the PENELOPE output files with
surface of the crystal, was not considered in the model of the detector the FEPEs calculated. This possibility of working automatically with
because for obtaining the correct dimension of this element, it would input and output files has been essential for the aim of this work.
have been necessary to optimize the model for a sample geometry in DEMO is a type of evolutionary algorithm proposed by Robič and
which a significant part of the radiation is forced to pass through the Filipič [29], based on the traditional Differential Evolution algorithm
mount cap in order to get to the crystal, which was not possible to (DE). This combines the advantages of the DE [28] with the mechan-
accomplish with the type of sample geometry used in this previous isms of Nondominated Sorting and Crowding Distance Metric com-
paper. monly used in multi-objective optimization algorithms. As in the case
The purpose of this work is to extend the methodology developed in of DE, DEMO starts from an initial population of individuals or
the mentioned paper, in order to build a model of the detector valid for candidate solutions. During the execution of this algorithm, the
a wider range of sample geometries, such as Marinelli, Petri, or any population of candidate solutions evolves creating new individuals by
other type of beaker with a large diameter, where a part of the radiation the iterative application of several evolutionary operators: mutation,
have to pass through the mount cap to reach the active region of the crossover, and selection; keeping in the population those individuals
detector. For this purpose, first of all, the mount cap has been included that best match the prefixed premises of the optimization problem. The
in the model. Second of all, in order to get a model equally accurate differences between both algorithms arise in the selection phase, as the
regardless of the type of beaker, a Differential Evolution Multi- mono-objective DE selects those individuals with best value (either
Objective (DEMO) algorithm [29] has been implemented instead of minimum or maximum) of the only objective function, whereas in
the mono-objective one. Using this multi-objective algorithm, the DEMO the selection of the future population is based on the concept of
model of the detector is optimized for both a beaker of small diameter dominance taking into account the values of the several objective
laid over the detector window and a beaker of large capacity which functions.
wrap the detector. Therefore, for the application of this methodology in
a given detector, it is necessary to know two sufficiently accurate sets of 2.2. Methodology applied
reference FEPEs, one for each type of geometry.
2.2.1. Samples and FEPEs of reference
2. Material and methods Although, obviously, the main utility of the methodology proposed in
this work is to build a model of detector starting from experimental FEPEs
2.1. Instrumentation, software and codes of reference, in this work the FEPEs used as reference have been
generated by means of the LabSOCS code since it provides a significantly
The methodology proposed in this work has been applied to a greater flexibility for both developing and validating the methodology.
114
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Fig. 1. Dimensions of polypropylene beakers and its water samples, used during the characterization process. a) a cylindrical container with a small diameter, which is placed over the
detector window. b) a Marinelli beaker, with a large capacity and which wraps the detector.
115
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Table 2
Geometrical characteristics of the model of detector. The dimensions specified by the
manufacturer are shown in the 3rd column. Columns 4th to 6th show the assignation of
the unknown characteristics to the variables of the optimization problem (x gp,m ) and the
max
minimum (xminm ) and maximum (xm ) values considered for each variable during the
optimization. Best solution obtained by the optimization process is shown in last column.
All values are given in mm.
Hd Housing 73.00 – – – –
diameter
Ht Housing 1.50 – – – –
thickness
Wt Window 0.60 – – – –
thickness
Cl Crystal length 50.51 – – – –
Cd Crystal 59.52 – – – –
diameter
Hcr Higher – x gp,1 0.40 6.00 2.25
curvature
radius
Dcw Distance – x gp,2 3.00 8.00 6.67
crystal –
window
Fig. 4. Representation in the function space of the candidate solutions (blue points), the
Dlt Dead layer – x gp,3 0.20 1.50 0.59
non-dominated solutions (red circumferences) and the chosen solution (green cross).
thickness
(For interpretation of the references to color in this figure legend, the reader is referred to
Icd Inactive core – x gp,4 4.00 26.00 13.35
the web version of this article.)
diameter
Ich Inactive core – x gp,5 15.00 44.00 24.26
height XtRa given in Table 2, and therefore, each individual is defined by a
Mct Mount cap – x gp,6 0.50 4.00 2.97 vector xpg =(xpg,1, xpg,2 , xpg,3, xpg,4 , xpg,5, xpg,6 ). The search domain, it is to say,
thickness
the minimum and maximum values for each component, can be seen in
fifth and sixth column of Table 2.
Fig. 3 shows a flowchart that summarizes the optimization process
in the generation g (iteration index), consist in a vector
xpg =(xpg,1, …, xpg, m, …, xpg, n ), where each component (of index m) is a followed for the computational characterization of the detector, being
variable of the optimization problem, which are restricted during the detailed below the different steps followed during DEMO execution.
entire algorithm execution, between minimum (xmmin ) and maximum
(xmmax ) values. In this work, such variables are the unknown character- 2.2.3.1. Initialization. This first step is performed only once at the
istics (Hcr, Dcw, Dlt, Icd, Ich, Sbt) of the model of the detector HPGe beginning of the DEMO execution in order to generate the population
Fig. 3. Flowchart of the optimization process followed during the computational characterization of the detector.
116
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Table 3 Table 4
Comparison between the FEPEs obtained by PENELOPE (ε PEN ) and LabSOCS (ε ref ) for Centesimal composition of IAEA Reference Materials (IAEA, 1987) and densities
both beaker geometries. Relative differences calculated using as reference the LabSOCS considered for the Monte Carlo Simulations.
values. FEPEs calculated with an uncertainty of 4%.
IAEA-RGU-1 IAEA-RGTh-1 IAEA-RGK-1
Energy (keV) εref a εPEN a Dr (%) εref b εPEN b Dr (%)
Density (g/cm3) 2.0 2.0 2.0
45 0.129 0.126 −2.0 0.0219 0.0221 1.0 Centesimal composition O: 53.80% O: 47.95% O: 36.72%
62 0.134 0.135 0.6 0.036 0.037 1.5 Na: 0.10% Si: 42.07% S: 18.40%
85 0.133 0.131 −1.6 0.049 0.048 −1.7 Al: 0.10% Th: 2.89% K: 44.87%
110 0.126 0.125 −0.4 0.054 0.054 −0.6 Si: 45.70% U: 7.09%
135 0.115 0.114 −1.1 0.054 0.054 0.6 Ca: 0.10%
180 0.097 0.097 −0.2 0.049 0.048 −0.8 Fe: 0.10%
270 0.068 0.068 −0.2 0.037 0.037 0.6 U: 0.10%
450 0.042 0.042 0.2 0.0240 0.0244 1.7
630 0.032 0.032 0.2 0.0185 0.0185 −0.0
900 0.0234 0.0234 −0.0 0.0141 0.0141 −0.4
1260 0.0180 0.0180 0.2 0.0110 0.0109 −0.7
1800 0.0129 0.0133 2.5 0.0083 0.0084 1.5 Since the combination in Eq. (2) might generate mutant vectors
Mean (Dr(%)) −0.2 0.2
with components out of the search domain, the following operation is
Mean(|Dr(%)|) 0.8 0.9
Min (Dr(%)) −2 −1.7 performed:
Max (Dr(%)) 2.5 1.7
Ifnpg, m < xmmin thennpg, m=xmmin
, forp=1. ..NP, m=1. ..n
Ifnpg, m > xmmax thennpg, m=xmmax (3)
Fig. 5. (a) Efficiency calibration curves obtained by least squares fitting to the efficiencies calculated by PENELOPE and LabSOCS. (b) Relative differences between PENELOPE and
LabSOCS curves taking as reference the LabSOCS curves.
117
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Fig. 6. Verification Geometries showing their main dimensions. The sample heights (H) are: 2 mm and 72.2 mm in VG1, 68.2 mm and 120 mm in VG2, and, 2 mm and 26.6 mm in
VG3.
Table 5 the original vector, the trial vector replaces the original vector for the
Comparison between the FEPEs obtained by PENELOPE (εPEN ) and LabSOCS (εLab ) for next generation, whereas if the original vector dominates the trial
the verification geometry VG1. Relative differences calculated using as reference the
vector, the trial vector is discarded. Otherwise, if there is not
LabSOCS values. FEPEs calculated with an uncertainty of 4%.
dominance among them, the trial vector is added as a candidate
Verification Geometry H=2 mm H=72.2 mm solution, increasing the size of the population (which might be multi-
plied per two). In the case of two objective functions, as in our work,
Material En (keV) εPEN εLab Dr (%) εPEN εLab Dr (%) the DEMO selection would be performed as follow:
RGU−1 46.5 0.25 0.25 −1.3 0.029 0.028 2.4 ⎧ t pg iffa (t pg ) < fa (xpg ) andfb (t pg ) < fb (xpg )
RGU−1 63.3 0.26 0.26 −2.4 0.037 0.037 0.8 ⎪
RGU−1 92.6 0.25 0.26 −3.7 0.041 0.042 −2.3 xpg + 1 =⎨ xpg iffa (xpg ) < fa (tpg ) andfb (xpg ) < fb (tpg ) , forp=1. ..NP
RGU−1 143.8 0.209 0.215 −3.1 0.037 0.038 −3.4 ⎪ g
RGU−1 163.5 0.193 0.198 −2.8 0.035 0.036 −2.1 ⎩ elset p is added at the end of the population. (5)
RGU−1 186.0 0.175 0.181 −3.5 0.033 0.033 −2.3
RGU−1 242.0 0.138 0.141 −2.1 0.027 0.027 −1.6 When trial vectors have been added in the population without
RGU−1 295.2 0.114 0.117 −2.7 0.0230 0.0233 −1.5 replacement, it is needed to truncate such a population in order to keep
RGU−1 352.0 0.097 0.098 −1.5 0.0197 0.0201 −2.2 a constant size NP during the entire execution of the algorithm, being
RGU−1 609.3 0.057 0.059 −2.8 0.0132 0.0131 0.8
therefore necessary to decide which individuals are better candidates
RGU−1 768.4 0.047 0.048 −2.3 0.0108 0.0110 −1.4
RGU−1 934.1 0.039 0.041 −2.7 0.0095 0.0096 −0.2 for the optimization purpose. In DEMO, the truncation method,
RGU−1 1120.3 0.034 0.035 −2.4 0.0084 0.0084 −0.6 derived from the evolutionary algorithm NSGA-II [33], consists in
RGU−1 1238.1 0.031 0.032 −3.6 0.0078 0.0079 −0.9 sorting the individuals by Nondominated Sorting, which ensure the
RGU−1 1764.5 0.0226 0.0230 −2.0 0.0058 0.0059 −1.2 election of the less dominated individuals, and Crowding Distance
RGTh−1 129.0 0.201 0.207 −3.3 0.0173 0.0177 −2.5
RGTh−1 238.6 0.138 0.141 −2.2 0.0218 0.0219 −0.4
metric, which preserves the diversity in the population.
RGTh−1 328.1 0.103 0.105 −1.8 0.0189 0.0192 −1.6 The Nondominated Sorting compares each individual of the grown
RGTh−1 338.3 0.099 0.102 −2.2 0.0186 0.0188 −1.2 population with the rest, classifying each individual according to the
RGTh−1 583.2 0.059 0.061 −3.2 0.0130 0.0132 −1.2 number of individuals which dominate it. Then the population is
RGTh−1 727.3 0.048 0.050 −3.5 0.0111 0.0113 −1.3
separated on several fronts of equally dominated individuals, sorting
RGTh−1 860.0 0.042 0.043 −3.3 0.0101 0.0101 0.3
RGTh−1 911.0 0.041 0.041 −1.8 0.0096 0.0097 −0.7 the individuals so that the less dominated ones are placed in the higher
RGTh−1 969.0 0.038 0.039 −2.9 0.0093 0.0093 −0.2 positions (firstly the non-dominated, secondly the dominated by only
RGTh−1 1588.2 0.025 0.026 −0.9 0.0064 0.0064 −0.5 one individual and so on).
RGK−1 1460.8 0.027 0.028 −2.1 0.0069 0.0069 −0.4 Once performed the Nondominated Sorting, it might be applied the
Mean (Dr(%)) −2.5 −1.0
truncation keeping the NP first individuals, which are less dominated,
Mean(| 2.5 1.3
Dr(%)|) but this may well discard some individuals of the same front which
Min (Dr(%)) −3.7 −3.4 might be better candidates than the non-discarded for being more
Max (Dr(%)) −0.9 2.4 different from the rest (seeking diversity). For this reason, the
Crowding Distance metric is applied to the front Fd which is going to
g+1
be divided (the front to which belong the individual xNP ). The
Crowding Distance metric consists in measuring the density of solu-
dominance is used instead (one candidate dominates other candidate tions in the neighbourhood of each solution, and then sorting them
when the evaluations of all objective functions for the first candidate according to such a density, giving the worst positions to those
are lower than the corresponding to the second). individuals whose objective functions evaluations are more similar to
the rest [33].
In DEMO, each trial vector t pg is compared with the corresponding As it was said before, in our optimization problem there are two
original vector xpg (with same index) so that if the trial vector dominates functions to be minimized, one for each type of geometry mentioned in
Section 2.2.1. Both objective functions consist in the sum of the
118
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Fig. 7. (a) and (c) Efficiency calibration curves obtained from PENELOPE and LabSOCS for the verification geometries and the reference materials RGU-1 and RGTh-1 respectively,
within their ranges of energy (46.5–1764.5 keV for RGU-1 and 129–1588.2 keV for RGTh-1) (b) and (d) Relative differences between PENELOPE and LabSOCS curves, taking as
reference the LabSOCS values.
quadratic relative differences between the FEPEs calculated by implied a week of calculations using a 3.1 GHz i7 processor due to the
PENELOPE, and the set of FEPEs of reference: approximately two minutes of duration of each Monte Carlo Simulation
in PENELOPE.
⎛ε ⎛ g ⎞ ⎞2
In Fig. 4 we can see the fa and fb values for the final NP candidate
g g g g
i=e ⎜ PEN a ⎜⎝xp,1, xp,2, xp,3, xp,4, xp,5, E i⎟⎠ − εref (Ei)a ⎟
fa (xpg )= ∑i =1 ⎜⎜ εref (Ei )a ⎟⎟ solutions (blue points) as well as the non-dominated solutions (red
⎝ ⎠ circumferences), which would conform the non-dominated front or
⎛ε ⎛ g g g g g ⎞ ⎞2 also named Approximate Pareto Front. Theoretically, the solutions into
PEN b ⎜xp,1, xp,2, xp,3, xp,4, xp,5, E ⎟ − εref (Ei )b
i=e ⎜ ⎝ i⎠ ⎟ the Approximate Pareto Front would be equally suitable, but we have
fb (xpg )= ∑i =1 ⎜ ⎟⎟
⎜ εref (Ei )b
chosen the solution marked with a green cross in Fig. 4 (corresponding
⎝ ⎠ (6)
to fa = 0.000969599 and fb = 0.000742566) as not only is it the solution
Where e is the number of photopeaks, Ei the energy of the photopeak i, with the smallest sum of both objective functions, but also with the
εPEN a and εPEN b are the calculated efficiencies by PENELOPE (for the smallest difference between them, and therefore, with the most similar
corresponding geometry) for the individual p in generation g and for a behaviour for both types of geometry. The geometrical characteristics
photopeak with energy Ei . εref a and εref b are the reference efficiencies for corresponding to the chosen solution are shown in the last column of
the photopeak i. Table 2, showing an adequate election of the search space, as the values
of every variable are quite centred in the prefixed intervals.
3. Results and discussion In Table 3 we can see a comparison between the reference FEPEs,
calculated by LabSOCS, and the FEPEs calculated by PENELOPE using
3.1. Solution of the optimization process the optimal model chosen. Due to the fact that such a model has been
built starting from the reference FEPEs calculated by LabSOCS with an
The DEMO algorithm has been executed with a population size NP uncertainty of 4%, and taking into account that the statistical un-
of 60 individuals (ten times the number of variables as recommended certainties of the calculations by PENELOPE are much lower (about
by Storn and Price [28]), a mutation factor F of 0.3 and a crossover 1%), the uncertainties of the FEPEs computed by PENELOPE has been
factor CR of 0.2. The election of such small values of F and CR is based considered to be of 4% as well. Therefore, the deviations observed in
in the necessity of minimizing as far as possible the differences between Table 3, lower than 2.5% in any case, show a good agreement for both
the FEPEs of reference and the FEPEs calculated by the model, beaker geometries.
regardless of the type of minimum found (either local or global). It is a common practice to fit the sets of FEPEs to continuous
With regard to the stopping criterion of the optimization process, a functions, so that efficiencies for any energy can be easily estimated.
maximum number of 70 generations has been chosen, which has The different sets of FEPEs in Table 3 have been fitted, by the least
119
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Table 6 Table 7
Comparison between the FEPEs obtained by PENELOPE (εPEN ) and LabSOCS (εLab ) for Comparison between the FEPEs obtained by PENELOPE (εPEN ) and LabSOCS (εLab ) for
the verification geometry VG2. Relative differences calculated using as reference the verification geometry VG3. Relative differences calculated using as reference the
LabSOCS values. FEPEs calculated with an uncertainty of 4%. LabSOCS values. FEPEs calculated with an uncertainty of 4%.
Material En (keV) εPEN εLab Dr (%) εPEN εLab Dr (%) Material En (keV) εPEN εLab Dr (%) εPEN εLab Dr (%)
RGU−1 46.5 0.0084 0.0088 −3.7 0.0126 0.0130 −3.3 RGU−1 46.5 0.161 0.157 2.6 0.052 0.053 −2.0
RGU−1 63.3 0.027 0.027 −0.9 0.025 0.025 2.5 RGU−1 63.3 0.169 0.167 1.2 0.069 0.067 2.3
RGU−1 92.6 0.047 0.048 −1.3 0.036 0.036 0.5 RGU−1 92.6 0.166 0.166 −0.3 0.074 0.074 −0.3
RGU−1 143.8 0.053 0.053 −0.9 0.037 0.037 −0.1 RGU−1 143.8 0.145 0.144 1.2 0.067 0.068 −0.7
RGU−1 163.5 0.052 0.052 −0.6 0.037 0.036 1.0 RGU−1 163.5 0.134 0.134 0.4 0.064 0.064 0.3
RGU−1 186.0 0.050 0.050 −0.2 0.034 0.035 −1.3 RGU−1 186.0 0.123 0.123 0.0 0.059 0.059 −0.2
RGU−1 242.0 0.043 0.042 1.0 0.029 0.029 0.3 RGU−1 242.0 0.099 0.097 1.1 0.049 0.048 1.0
RGU−1 295.2 0.037 0.037 0.0 0.026 0.025 1.3 RGU−1 295.2 0.083 0.082 1.9 0.042 0.041 1.6
RGU−1 352.0 0.033 0.032 2.8 0.0222 0.0220 0.6 RGU−1 352.0 0.069 0.069 0.8 0.036 0.035 0.8
RGU−1 609.3 0.0207 0.0209 −0.8 0.0147 0.0146 0.5 RGU−1 609.3 0.042 0.042 −0.2 0.0227 0.0226 0.7
RGU−1 768.4 0.0177 0.0176 0.5 0.0125 0.0124 0.8 RGU−1 768.4 0.035 0.035 0.9 0.0188 0.0189 −0.2
RGU−1 934.1 0.0153 0.0153 0.1 0.0108 0.0108 0.2 RGU−1 934.1 0.030 0.029 0.6 0.0165 0.0163 1.2
RGU−1 1120.3 0.0137 0.0134 1.8 0.0096 0.0095 0.3 RGU−1 1120.3 0.025 0.025 −1.4 0.0141 0.0142 −0.7
RGU−1 1238.1 0.0125 0.0125 0.2 0.0090 0.0089 0.5 RGU−1 1238.1 0.0235 0.0236 −0.2 0.0131 0.0132 −1.0
RGU−1 1764.5 0.0096 0.0095 0.4 0.0069 0.0068 2.0 RGU−1 1764.5 0.0174 0.0171 1.8 0.0099 0.0098 0.5
RGTh−1 129.0 0.033 0.034 −1.6 0.0196 0.0198 −1.2 RGTh−1 129.0 0.137 0.137 0.0 0.036 0.036 −1.8
RGTh−1 238.6 0.039 0.038 1.6 0.025 0.025 1.6 RGTh−1 238.6 0.099 0.097 2.6 0.042 0.042 1.2
RGTh−1 328.1 0.032 0.032 0.3 0.0219 0.0216 1.6 RGTh−1 328.1 0.074 0.073 1.1 0.035 0.035 0.4
RGTh−1 338.3 0.032 0.031 0.9 0.0211 0.0212 −0.2 RGTh−1 338.3 0.072 0.071 1.0 0.035 0.034 2.3
RGTh−1 583.2 0.0216 0.0213 1.2 0.0152 0.0148 2.7 RGTh−1 583.2 0.044 0.044 0.1 0.0232 0.0230 0.9
RGTh−1 727.3 0.0181 0.0182 −0.6 0.0130 0.0128 2.1 RGTh−1 727.3 0.036 0.036 −0.9 0.0196 0.0195 0.6
RGTh−1 860.0 0.0167 0.0162 3.0 0.0113 0.0114 −0.9 RGTh−1 860.0 0.032 0.031 1.1 0.0173 0.0172 0.7
RGTh−1 911.0 0.0153 0.0155 −1.6 0.0110 0.0109 0.8 RGTh−1 911.0 0.030 0.030 0.1 0.0167 0.0165 1.0
RGTh−1 969.0 0.0151 0.0149 2.0 0.0107 0.0105 2.4 RGTh−1 969.0 0.029 0.029 0.2 0.0159 0.0158 0.8
RGTh−1 1588.2 0.0105 0.0104 1.6 0.0074 0.0074 0.5 RGTh−1 1588.2 0.0189 0.0189 0.4 0.0109 0.0107 1.4
RGK−1 1460.8 0.0110 0.0111 −0.3 0.0080 0.0079 1.4 RGK−1 1460.8 0.0204 0.0204 0.3 0.0116 0.0116 0.4
Mean (Dr(%)) 0.2 0.6 Mean (Dr(%)) 0.6 0.4
Mean(| 1.2 1.2 Mean(| 0.9 1.0
Dr(%)|) Dr(%)|)
Min (Dr(%)) −3.7 −3.3 Min (Dr(%)) −1.4 −2.0
Max (Dr(%)) 3.0 2.7 Max (Dr(%)) 2.6 2.3
squares method, to third degree logarithmic polynomials [34]: with its main dimensions, being VG1 a common frustoconical beaker
i =3 with a sample height of 2.0 and 72.2 mm, VG2 a Marinelly beaker
lnε= ∑ ailni E (different to the one used during the optimization process) with a
i =0 (7) sample height of 68.2 mm and 120 mm, and VG3 a Petri beaker with a
The efficiency curves obtained through such a fitting for both beaker diameter larger than the crystal of the detector, with a sample height of
geometries have been plotted in Fig. 5a. As it can be observed in 2.0 and 26.6 mm.
Fig. 5b, the differences between the efficiency curves obtained from Therefore, for the verification of the model, 18 different simulations
both codes are lower than 2% in absolute value, being the maximum have been performed by both PENELOPE and LabSOCS (one simula-
deviation corresponding to the geometry a for an energy of 1800 keV. tion per geometry and material). The efficiencies calculated by such
simulations, as well as the relative differences between the results of
both codes, have been listed from Table 5 to Table 7. In such tables it
3.2. Verification and validation of the model obtained
can be observed an excellent agreement between the efficiencies
calculated by PENELOPE and LabSOCS, as the relative difference with
Due to the fact that the model of the HPGe XtRa detector has been
maximum absolute value is 3.7% and the arithmetic means are all
optimized for two given geometries and only one material (water), it is
below 3%, being therefore all the deviations lower than the 4%
necessary to check the goodness of such a model for samples with
considered in LabSOCS as the minimum uncertainty advisable [19],
different materials, geometries and gamma emission energies in order
despite the fact that the simulations have been performed under
to verify the stability of the model under changes in the sources, as well
conditions clearly different to those considered during the optimization
as validate the response of the model with experimental measurements
process (it has been changed the geometries, density and composition
of CRMs.
of the materials and the gamma emission energies).
The materials RGU-1 and RGTh-1 selected for verification have
3.2.1. Internal verification
various gamma emissions with energies distributed throughout the
An internal verification of the consistency of the results has been
range of interest for environmental samples (45–1800 keV), this fact
performed by comparing the efficiencies calculated by PENELOPE
allows us to fit its efficiencies to Eq. (7) for each verification geometry
(with the model optimized) and LabSOCS for six different geometries
and height. These functions are represented in Fig. 7, where it can be
(three different beakers widely used in environmental gamma spectro-
observed that the differences between the efficiency curves obtained by
metry with two considerably different sample heights for each one) and
both codes are always lower than 3.3% in absolute value, appearing the
for three reference materials, RGU-1, RGTh-1 and RGK-1 (character-
maximum deviation in the geometry VG2 with a sample height of
istics summarized in Table 4), of the International Atomic Energy
68.2 mm and for an energy of 46.5 keV in the case of RGU-1, and in the
Agency [35] with their corresponding most probable gamma emission
geometry VG1 with a sample height of 2 mm and for an energy of
energies within the range of interest (45–1800 keV). In Fig. 6 it can be
129 keV in the case of RGTh-1. Obviously, the efficiencies in the case of
seen the three types of beaker used for the verification of the model
120
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
Table 8
Activity concentrations (uncertainties given as 1·σ.) determined using FEPEs calculated by PENELOPE and LABSOCS and the values certified by the IAEA, expressed all in Bq/kg.
238
VG1h=72.2 mm RGU−1 U 5010 ± 210 4980 ± 210 4940 ± 30
232
VG1h=72.2 mm RGTh−1 Th 3170 ± 140 3150 ± 140 3250 ± 90
40
VG1h=72.2 mm RGK−1 K 14,060 ± 610 14,070 ± 610 14,000 ± 400
238
VG2h=103.3 mm RGU−1 U 4740 ± 200 4840 ± 200 4940 ± 30
232
VG2h=103.3 mm RGTh−1 Th 3030 ± 130 3070 ± 130 3250 ± 90
40
VG2h=103.3 mm RGK−1 K 13,360 ± 580 13,300 ± 580 14,000 ± 400
Table 9 Table 9 shows the result of the utest for each measurement, where it
Utest values for the results obtained calibrating by PENELOPE (PEN) and LabSOCS (Lab) can be seen that all the values are below the limit of 1.64, which means,
compared with the certified values (Cert).
accordingly to such a statistical test, that the results of the gamma
Geometry CRM Nuclide Utest (PEN) Utest (Lab) spectrometric analysis using PENELOPE and LabSOCS FEPEs do not
differ significantly from the expected values, being corroborated the
238
VG1h=72.2 mm RGU−1 U 0.34 0.17 good agreement mentioned before, and therefore, experimentally
232
VG1h=72.2 mm RGTh−1 Th 0.49 0.61
40
validated the model of the detector built by the methodology proposed
VG1h=72.2 mm RGK−1 K 0.08 0.09
VG2h=103.3 mm RGU−1 238
U 1.08 0.51
in this work.
232
VG2h=103.3 mm RGTh−1 Th 1.41 1.12
40
VG2h=103.3 mm RGK−1 K 0.91 0.99 4. Summary and conclusions
121
J. G. Guerra et al. Nuclear Instruments and Methods in Physics Research A 858 (2017) 113–122
the FEPEs computed by such a model in PENELOPE. The activity Phys. Res. B 330 (2014) 1.
[13] J. Nikolic, T. Vidmar, D. Jokovic, M. Rajacic, D. Todorovic, Nucl. Instrum. Methods
concentrations obtained have shown a good agreement with the Phys. Res. A 763 (2014) 347.
certified values, as results of the u-test have not exceeded the limit of [14] Erica Andreotti, Nucl. Instrum. Methods Phys. Res. A 718 (2013) 475.
1.64, which guarantees that the results do not differ significantly from [15] N. Dokania, V. Singh, S. Mathimalar, V. Nanal, S. Pal, R.G. Pillay, Nucl. Instrum.
Methods Phys. Res. A 745 (2014) 119.
the certified values. [16] A. Hedman, J. Bahar Gogani, M. Granström, L. Johansson, J.S. Andersson,
H. Ramebäck, Nucl. Instrum. Methods Phys. Res. A 785 (2015) 21.
Acknowledgements [17] R. Berndt, P. Mortreau, Nucl. Instrum. Methods Phys. Res. A 694 (2012) 341.
[18] T. Vidmar, A. Likar, Appl. Radiat. Isot. 56 (2002) 99.
[19] F. Bronson, J. Radioanal. Nucl. Chem. 255 (1) (2003) 137.
This work has been partially financed by the Nuclear Safety Council [20] R. Venkataraman, F. Bronson, V. Atrashkevich, M. Field, B.M. Young, J. Radioanal.
of Spain (CSN) R & D program 2009 through the research project Nucl. Chem. 264 (2005) 213.
[21] N.Q. Huy, Nucl. Instrum. Methods Phys. Res. A 621 (2010) 390.
"Estudio sobre la distribución de radioisótopos naturales y de radón en
[22] N.Q. Huy, Nucl. Instrum. Methods Phys. Res. A 641 (2011) 101.
las Islas Canarias Orientales" and R & D program 2012 through the [23] N.Q. Huy, D.Q. Binh, V.X. An, Nucl. Instrum. Methods Phys. Res. A 573 (2007)
research project "Estudio de las concentraciones de radón en viviendas, 384.
lugares de trabajo y materiales de construcción en las Islas Canarias [24] J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo,
A. Tejera, J.M. Gil, R. Rodríguez, P. Martel, J. Environ. Radioact. 149 (2015) 8.
Orientales". [25] J. Sempau, J.M. Fernandez-Varea, E. Acosta, F. Salvat, Nucl. Instrum. Methods
J.G. Guerra acknowledges a PhD fellowship through the “Programa Phys. Res. B 207 (2003) 107.
de personal investigador pre-doctoral en formación” of the ULPGC. [26] F. Salvat, J.M. Fernández-Varea, E. Acosta, J. Sempau, PENELOPE: a code system
for monte carlo simulation of electron and photon transport, in: Proceedings of
Workshop Proceedings, OECD Nuclear Energy Agency, Issy-les-Moulineaux, 2001.
References [27] F. Salvat, J.M. Fernández-Varea, J. Sempau, PENELOPE-2011: A Code System for
Monte Carlo Simulation of Electronic and Photon Transport, OECD, Barcelona,
Spain, 2011.
[1] K. Abbas, F. Simonelli, F. D'Alberti, M. Forte, M.F. Stroosnijder, Appl. Radiat. Isot.
[28] R. Storn, K. Price, J. Glob. Optim. 11 (1997) 341.
56 (2002) 703.
[29] T. Robič, B. Filipič, DEMO: Differential evolution for multi-objective optimization,
[2] M. Mostajaboddavati, S. Hassanzadeh, H. Faghihian, M.R. Abdi, M. Kamali, J.
Evolutionary.
Radioanal. Nucl. Chem. 268 (2006) 539.
[30] Multi-Criterion Optimization – EMO, Vol. 3410 of Lecture Notes in Computer
[3] Aline S.E. Santo, Francis G. Wasserman, Claudio C. Conti, Ann. Nucl. Energy 46
Science, Springer, Berlin (2005) 520, 2005.
(2012) 213.
[31] Genie 2000, Basic Spectroscopy Software Version 2.1., Canberra Industries Inc.,
[4] J. Nikolic, J. Puzovic, D. Todorovic, M. Rajacic, Nucl. Instrum. Methods Phys. Res.
2002.
A 799 (2015) 159.
[32] Model S573/S574 ISOCS/LabSOCS Validation & Verification Manual V.4.0,
[5] H. Jäderström, W.F. Mueller, V. Atrashkevich, A.S. Adekola, Nucl. Instrum.
Canberra Industries Inc., 2002.
Methods Phys. Res. A 784 (2015) 264.
[33] J.F. Briesmeister, MCNP-A General Monte Carlo N-Particle Transport Code, LA-
[6] M.A. Arnedo, J.G. Rubiano, H. Alonso, A. Tejera, A. González, J. González, J.M. Gil,
13709-M, 2000.
R. Rodríguez, P. Martel, J.P. Bolivar, J. Environ. Radioact. 166 (2017) 242. http://
[34] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multi-objective
dx.doi.org/10.1016/j.jenvrad.2016.07.010.
genetic algorithm: NSGA–II, IEEE Trans. Evolut. Comput. 6 (2) (2002) 182.
[7] Massimo Altavilla, Romolo Remetti, Nucl. Instrum. Methods Phys. Res. A 712
[35] K. Debertin, R.G. Helmer, Gamma and X-ray Spectroscopy with Semiconductor
(2013) 157.
Detectors, North-Holland, Physical Sciences & Engineering Division, Elsevier
[8] Yasser Morera-Gómez, H.éctor A. Cartas-Aguila, Carlos M. Alonso-Hernández,
Science Publishers B.V, Amsterdam, 1988 (ISBN 0-444-87107-1).
Carlos Nuñez-Duartes, Nucl. Instrum. Methods Phys. Res. A 818 (2016) 51.
[36] IAEA. Preparation and Certification of IAEA Gamma Spectrometry Reference
[9] A. Azbouche, M. Belgaid, H. Mazrou, J. Environ. Radioact. 146 (2015) 119.
Materials RGU-1, RGTh-1 and RGK-1, IAEA Report RL148, International Atomic
[10] T.M. Semkow, C.J. Bradt, S.E. Beach, D.K. Haines, A.J. Khan, A. Bari, M.A. Torres,
Energy Agency, Vienna, 1987.
J.C. Marrantino, U.F. Syed, M.E. Kitto, T.J. Hoffman, P. Curtis, Nucl. Instrum.
[37] C. Brookes, I. Betteley, S. Loxton, Fundamentals of Mathematics and Statistics,
Methods Phys. Res. A 799 (2015) 105.
John Wiley, New York, USA, 1979.
[11] Kiyoshi Shizuma, Yurika Oba, Momo Takada, Nucl. Instrum. Methods Phys. Res. B
[38] IAEA Analytical Quality in Nuclear Applications Series No. 18 Vienna, 2010.
383 (2016) 183.
[39] A. Shakhashiro, L. Mabit, Appl. Radiat. Isot. 67 (2009) 139.
[12] S. Dziri, A. Nachab, A. Nourreddine, A. Sellam, A. Pape, Nucl. Instrum. Methods
122
Capítulo 4
a r t i c l e i n f o a b s t r a c t
Keywords: The aim of this paper is to characterize two HPGe gamma-ray detectors used in two different laboratories for
Gamma-ray spectrometry environmental radioactivity measurements, so as to perform efficiency calibrations by means of Monte Carlo
Efficiency calibration Simulation. To achieve such an aim, methodologies developed in previous papers have been applied, based
Reference materials on the automatic optimization of the model of detector, so that the differences between computational and
Monte Carlo simulation
reference FEPEs are minimized. In this work, such reference FEPEs have been obtained experimentally from
several measurements of the IAEA RGU-1 reference material for specific source-detector arrangements. The
models of both detectors built through these methodologies have been validated by comparing with experimental
results for several reference materials and different measurement geometries, showing deviations below 10% in
most cases.
© 2017 Elsevier B.V. All rights reserved.
* Corresponding author.
E-mail address: jglezg2002@gmail.com (J.G. Guerra).
https://doi.org/10.1016/j.nima.2017.10.076
Received 28 July 2017; Received in revised form 18 October 2017; Accepted 24 October 2017
Available online 4 November 2017
0168-9002/© 2017 Elsevier B.V. All rights reserved.
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
detector, by Canberra’s LabSOCS software [21], due to its flexibility to sample respectively) with certified activities for several radionuclides of
raise several cases for validating the methodologies proposed, avoiding interest for environmental radioactivity laboratories. With regard to the
at the same time imprecisions which could have appeared if reference CANMET DH-1a [32,33] is an ore with a considerable concentration of
experimental FEPEs had been used instead. both uranium and thorium (0.2629% U and 0.091% Th).
The present work aims to take a step forward, applying each method- Fig. 1 shows the two types of beakers used in this work. Beakers type
ology to a different HPGe detector owned by two different environmen- 1 are conventional beakers which are placed over the detector window,
tal laboratories, taking as reference the experimental FEPEs determined whilst beakers type 2 are Marinelli beakers which are placed wrapping
in such detectors, for a given reference material, IAEA RGU-1 [22] and the detector. In this work five different beakers have been used, three
a few source-detector arrangements, validating posteriorly the models of them, beaker 1 (D1 = 56 mm, D2 = 47.9 mm, H = 72.9 mm and e =
built through these methodologies by comparing the calculated FEPEs 0.7 mm), beaker 3 (D1 = 38.5 mm, D2 = 35.6 mm, H = 62.9 mm and
with experimental results for different measurement geometries and e = 0.7 mm) and beaker 5 (D1 = D2 = 52.2 mm, H = 21.4 mm and e
reference materials. Finally, it is important to remark that this work = 0.3 mm) are beakers type 1, and two, beaker 2 (D1 = 115.8 mm, D2
does not aim to find the real characteristics of the detectors, but = 79.2 mm, H1 = 103.3 mm, H2 = 70.2 and e = 2 mm) and beaker 4
to get computational models which behave as similar as possible to (D1 = 124.2 mm, D2 = 86.4 mm, H1 = 116.2 mm, H2 = 77.0 and e
the real detectors, even though the model characteristics might differ = 2.2 mm) are type 2. With regard to the materials, beakers 1 to 4 are
significantly from those of the real detector. polypropylene beakers whereas beaker 5 is made of aluminum.
2.1. Gamma ray detection systems, software and codes DET1 has been characterized using the methodology developed in
G. Guerra et al. [17] and DET2 using the methodology developed
In this work, two Canberra HPGe detectors from two different in G. Guerra et al. [16]. Table 1 contains the values of the known
laboratories of environmental radioactivity have been used. The first one geometric characteristics of both computational models, which has been
(DET1), which belongs to the GIRMA group at the Department of Physics considered constant during the optimization process, the definition of
of the University of Las Palmas de Gran Canaria, is a model GX3518 with the variables of the optimization problem (unknown characteristics), as
38% of relative efficiency and nominal FWHM of 0.875 keV at 122 keV well as the minimum and maximum values considered for each variable
and 1.8 keV at 1.33 MeV. The second one (DET2), owned by the FRyMA during the optimization process.
group at the Department of Applied Physics of the University of Huelva, In addition to the different dimensions of the Germanium crystals,
is a GX3519 with 35% of relative efficiency and nominal FWHM of 0.95 the main difference between both computational models is the consid-
keV at 122 keV and 1.9 keV at 1.33 MeV. Both detectors are situated eration of a simplified crystal mount cap in the DET1 model [17]. This
into shields with 15 cm thick iron walls, located in a room with concrete is needed in order to calculate FEPEs with enough accuracy in cases in
walls and ceiling, and work together with multichannel analyzers and which the beaker diameter is significantly larger than the germanium
the Genie 2000 software package [23]. crystal one, being especially important in the case of Marinelli beakers,
The Monte Carlo simulations for obtaining the computational FEPEs in which an important part of their volume surrounds the detector.
have been performed by the PENELOPE 2011 code [19]. Due to the Given the fact that the DET2 computational model does not include
cylindrical geometry of the detectors and samples, the pencyl.f subrou- the mount cap, calculations with such model will only be suitable for
tine, included in the PENELOPE distribution package for simulations beakers situated over the window with a maximum diameter equal to
for cylindrical geometry, has been used for controlling the entire the crystal diameter. The materials considered for both computational
simulation. The compilation of PENELOPE with pencyl.f as the main models are aluminum for the housings and for the mount cap in DET1,
subroutine creates an executable, pencyl.exe, which requires an input carbon composite for the windows and germanium for the crystals (both
file with the information needed for performing the simulations, it is to for the active region and the dead layer).
say, the geometry and materials of the detector and the definition of its
active regions, geometry and materials of the sample and the beaker 2.4. Determination of experimental FEPEs
in which it is placed, type of radiation emitted by the source to be
simulated (𝛾 or 𝛽) with its energies, duration of the simulation, etc. As it was mentioned in Section 2.2, the experimental FEPEs used
As it will be explained below, the optimizations of both compu- as reference during the optimization process of both computational
tational detector models are performed by the DE and DEMO algo- models, have been obtained by analyzing the reference material IAEA
rithms [18,20], programmed in scripts in Matlab©, together with the RGU-1 [22] by gamma spectrometry with different geometries. This
command lines to write, in PENELOPE input files, the definition of material has been chosen because of its several reliable gamma emis-
the geometric characteristics of the detector model (which represent a sions throughout the region of interest (45–1800 keV) for environmental
candidate solution in the optimization process) as well as the command radioactivity measurements. Concretely, a set of 8 gamma lines with
lines for reading the PENELOPE output files with the information energies within such interval (46.5, 63.3, 143.8, 186.0, 295.2, 352.0,
required to calculate the FEPEs for the desired energies. 1001.0 and 1764.5 keV) has been chosen, as the corresponding full
energy peaks stand out sufficiently from the Compton Continuum and
2.2. Reference materials and beakers are relatively free of interferences with other peaks in the spectra (which
together implies acceptable area uncertainties), as well as because of
For obtaining the experimental FEPEs used as reference during being free of coincidence summing effects. Additionally, the quite well
the optimization process of both computational detector models, the approximate knowledge of its chemical composition allows calculating
reference material IAEA RGU-1 has been chosen, whereas for the FEPEs by Monte Carlo simulation at low energies with an acceptable
validation of such models, the RGU-1, RGTh-1, RGK-1, 447, 448, 434 accuracy.
and 326 from the IAEA and the DH-1a from the CANMET, have been For DET1 characterization process, beakers 1 and 2 have been
used. IAEA RGU-1, RGTh-1 and RGK-1 are reference materials meant for completely filled with RGU-1 placing the first centered over the DET1
the experimental calibration of gamma-ray spectrometers for Potassium, window at 1 mm of distance, and the second, a Marinelli beaker, with
Uranium and Thorium [22]. IAEA 447 [24–26], IAEA 448 [26–28], its internal face centered directly over the window, so that the sample
IAEA 434 [26,29,30] and IAEA 326 [26,31] are environmental samples surrounds an important part of the detector. These two sample-detector
(moss-soil, contaminated soil from an oil field, Phosphogypsum and soil arrangements will be named RA1 and RA2 respectively throughout the
68
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
Fig. 1. Type of beakers used in this work with its main dimensions. Beakers type 1 are conventional beakers which are placed over the detector window. Beakers type 2 are Marinelli
beakers which are placed wrapping the detector.
Table 1
Geometric characteristics of both detector models. Columns 3rd and 4th show the known dimensions considered constant during optimization process. In Columns 5th and 6th the variables
of the optimization problem (𝑥𝑔𝑝,𝑚 ) are defined. The minimum (𝑥min
𝑚
) and maximum (𝑥max
𝑚
) values which each variable can take during the optimization process are in columns 7th and
8th. Best solutions obtained are shown in columns 9th and 10th. All values are given in mm.
Charac. Denomination Dimensions Variable 𝑥min
𝑚 𝑥max
𝑚 Solutions
DET1 DET2 DET1 DET2 DET1 DET2 DET1 DET2 DET1 DET2
Hd Housing diameter 73.00 – – – – – – – –
Ht Housing thickness 1.50 – – – – – – – –
Wt Window thickness 0.60 – – – – – – – –
Cl Crystal length 50.51 57.00 – – – – – – – –
Cd Crystal diameter 59.52 58.50 – – – – – – – –
Hcr Higher curvature radius – – 𝑥𝑔𝑝,1 0.40 6.00 1.73 0.98
Dcw Distance from crystal to window – – 𝑥𝑔𝑝,2 1.00 10.00 5.29 3.01
Dlt Dead layer thickness – – 𝑥𝑔𝑝,3 0.10 5.00 0.92 2.43
Icd Inactive core diameter – – 𝑥𝑔𝑝,4 1.00 47.50 17.52 2.12
Ich Inactive core height – – 𝑥𝑔𝑝,5 15.00 50.00 24.53 20.25
Mct Mount cap thickness – – 𝑥𝑔𝑝,6 – 0.50 – 4.00 – 2.90 –
following lines for simplification purposes. With regard to DET2, beaker is 228 ± 2. The 𝑃𝛾 and uncertainties considered have been obtained from
3 has been filled up to 25 mm of RGU-1 and has been situated centered at the National Nuclear Data Center [34].
3.5 mm of distance (RA3 ). In every case, once shut and sealed with alu- 𝐶
𝜀𝑒𝑥𝑝 = (1)
minum tape, before performing each spectrum acquisition, the beakers 𝐶𝑆𝐶𝐹 ⋅ [𝐴𝑐] ⋅ 𝑃𝛾 ⋅ 𝑡 ⋅ 𝑚
√
have remained undisturbed for at least 4 weeks (approximately seven ( 𝑢 )2 ( 𝑢 )2 ( 𝑢 )2 ( )
𝑃𝛾 𝑢 2 ( 𝑢𝐶𝑆𝐶𝐹 )2
𝐶 [𝐴𝑐]
times the half-life of 222 Rn) waiting for the radioactive equilibrium 𝑢𝜀𝑒𝑥𝑝 = 𝜀𝑒𝑥𝑝 + + + 𝑚 + . (2)
𝐶 [𝐴𝑐] 𝑃𝛾 𝑚 𝐶𝑆𝐶𝐹
between 226 Ra and 222 Rn.
Table 2 shows the experimental FEPEs (𝜀𝑒𝑥𝑝 ) used as reference during
The CSCF corresponding to the lines used as reference have been
the characterization process of DET1 and DET2. Such experimental considered 1 (no coincidence summing effect) with negligible uncer-
FEPEs and their uncertainties, as well as the ones used for the validation tainty. However, in the following sections, experimental FEPEs signif-
procedures in Section 3.2, have been determined by analyzing the icantly affected by Coincidence Summing Effect are used. These are
spectra acquired for the corresponding sample-detector arrangements, the corresponding to the gamma lines of 242.0, 609.3, 1120.3 and
and applying Eqs. (1) and (2), being 𝐶 the net counts observed in 1238.1 keV emitted by 238 U progeny and 338.3, 583.2, 727.3, 860.0
the relevant full energy peak (subtracting both Compton Continuum and 911.0 keV emitted by 232 Th progeny. In such cases, the CSCF have
and laboratory background, performed by GENIE2000 [23]), [𝐴𝑐] the been determined using the equations proposed by Xhixha et al. [35],
activity concentration of the relevant radionuclide, 𝑃𝛾 the probability except in the emission of 727.3 keV, for which the corresponding
of gamma emission for the corresponding radionuclide and gamma line, equation proposed by Tomarchio and Rizzo [36] has been used. In both
𝑡 the acquisition time, 𝑚 the sample mass and CSCF the Coincidence- works, the correction factors are calculated as function of the FEPEs
Summing Correction Factor, whereas, 𝑢𝐶 , 𝑢[𝐴𝑐] , 𝑢𝑃𝛾 , 𝑢𝑚 , 𝑢𝐶𝑆𝐹 𝐶 are the and total efficiencies for the energies corresponding to the X-ray and
corresponding uncertainties. In Eq. (2), the uncertainty related to the 𝛾-ray emissions which are likely to coincide with the main gamma
emission, causing the coincidence-summing effect. In this work, for such
time has not been considered as it is negligible compared to the rest
calculations, the FEPEs and total efficiencies provided by the models of
of elements within the radicand. The masses of the samples of RGU-1
DET1 and DET2 in PENELOPE have been used. The uncertainties of such
in beakers 1, 2 and 3 are 221.1, 925.6 and 37.5 g respectively with an
CSCF have been computed through uncertainty propagation.
uncertainty of 0.1 g, whereas the acquisition times have been 86749,
68882 and 20644 s. Given the radioactive equilibrium among the 2.5. Optimization of the models
different radionuclides of the 238 U decay series [22], for the calculations
with Eqs. (1) and (2), [𝐴𝑐] has been considered to be 4940 ± 30 for all The characterization methodologies applied in this work, consist in
energies, with the only exception of the line of 143.8 keV, for which [𝐴𝑐] the optimization of a computational model for each detector, seeking
69
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
Table 2
Experimental FEPEs used as reference during DET1 (RA1 and RA2 ) and DET2 (RA3 ) characterizations and their uncertainties.
Radionuclide En (keV) 𝜀𝑒𝑥𝑝 𝑅𝐴 𝑢𝜀𝑒𝑥𝑝 𝜀𝑒𝑥𝑝 𝑅𝐴 𝑢𝜀𝑒𝑥𝑝 𝜀𝑒𝑥𝑝 𝑅𝐴 𝑢𝜀𝑒𝑥𝑝
1 𝑅𝐴1 2 𝑅𝐴2 3 𝑅𝐴3
210
Pb 46.5 0.0325 0.0004 0.01576 0.00018 0.1006 0.0017
234
Th 63.3 0.0439 0.0010 0.0305 0.0007 0.120 0.003
235
U+223 Ra 143.8 0.0427 0.0012 0.0494 0.0010 0.113 0.006
236 Ra+235 U 186.0 0.0376 0.0004 0.0443 0.0004 0.0920 0.0013
214 Pb 295.2 0.02496 0.00017 0.03129 0.00020 0.0587 0.0005
214
Pb 352.0 0.02189 0.00014 0.02776 0.00018 0.0505 0.0004
234m
Pa 1001.0 0.00956 0.00026 0.01202 0.00020 0.0212 0.0012
214
Bi 1764.5 0.00599 0.00004 0.00790 0.00005 0.01296 0.00018
for the geometric characteristics that minimize the differences between Secondly, the CRMs IAEA 447, 448, 434 and 326 and the CANMET
the experimental FEPEs considered as reference (Section 2.4), and the DH-1a, have been measured for different geometries and bulk densities
calculated FEPEs. The model of DET2 has been optimized using the DE using the corresponding computational FEPEs, comparing subsequently
algorithm [16,18], taking as reference the sample-detector arrangement the results of the analysis with the certified activities (Section 3.2.2.).
RA3 . Moreover, for the optimization of the model of DET1, the DEMO For the measurements in DET1, the samples IAEA 447 and 448 were
algorithm [17,20] has been used, taking as reference the sample- prepared in beaker 5 completely filled of such CRMs with a mass of 51g
detector arrangements RA1 and RA2 . and 47.3 g respectively. The samples of IAEA 434, 326 and CANMET
Both optimization processes start with candidate solutions or indi-
DH-1a were prepared in beaker 3 with a height of 13.4, 23.0, 25.5 mm
viduals randomly generated at the beginning within the search domain
and a mass of 13.0, 30.0, 31.6 g respectively. With regard to the samples
(𝑥min max
𝑚 - 𝑥𝑚 ) (defined in Table 1 for ( each variable of ) the problem).
𝒈 used for the measurements in DET2, the same samples IAEA 447 and 448
These individuals are vectors 𝒙𝒑 = 𝑥𝑔𝑝,1 , … , 𝑥𝑔𝑝,𝑚 , … , 𝑥𝑔𝑝,𝑛 , where each
as in DET1 were measured, whereas the samples of IAEA 434, 326 and
component, with index m, is a variable of the optimization problem
CANMET DH-1a were prepared in beaker 3 with a height of 17.5, 31.0,
(𝑥𝑔𝑝,𝑚 ) (i.e. the unknown characteristics in Table 1, 𝑛 = 6 in the case
26.9 mm and a mass of 20.5, 30.0, 31.6 g respectively.
of DET1 and 𝑛 = 5 in the case of DET2), restricted within the mentioned
Again, every sample has been measured at least 1 month after
search domain, whereas g is the generation or iteration index and p the
its preparation. However, in the case of the samples of IAEA 448
individual or candidate index. The optimization is performed by creating
new candidate solutions using the operators of mutation and crossover and CANMET DH-1a, due to their high concentrations of 226 Ra, a
(both commonly used in evolutionary algorithms) and selecting the considerable escape of 222 Rn has been noticed, reason why in such
candidate solution with the best value for the optimization purpose, samples the concentration of 226 Ra has been obtained through its direct
which is evaluated by the objective functions (only one in the case of the gamma line of 186 keV. In the case of the IAEA 448, this line can
optimization of DET2 using DE, and two functions in the case of DET1 be used without performing any correction, as the 235 U content is
using DEMO). During the execution of both algorithms, the population negligible [27]. On the other hand, for DH-1a, a correction has been
size, it is to say the number of individuals, is maintained at a constant performed, discounting the theoretical counts due to the emission of
value NP. Further details can be found in previous papers [16,17] where 235 U, based on its activity concentration determined through the line of
both computational methodologies applied in this work were developed 143.8 keV.
and explained in detail. During both validation steps, the goodness of the optimized models
The objective functions used for the selection phase are analytically has been evaluated using the relative bias or relative deviation (Eq. (4)),
expressed in Eq. (3), where 𝐸𝑖 is the energy of the full energy peak i, 𝑅𝐴𝑗 the zeta-score [37] (Eq. (5)) as well as the Precision Score (Eq. (6)),
the corresponding reference sample-detector arrangement, 𝜀𝑒𝑥𝑝𝑅𝐴 is the widely used by the IAEA in inter-comparisons and proficiency tests [38–
𝑗
experimental FEPE taken as reference and 𝜀𝑃 𝐸𝑁 𝑅𝐴 the corresponding 41], where 𝑎 ± 𝑢𝑎 is the measured or calculated value (the calculated
𝑗
𝒈
FEPE calculated for a specific individual 𝒙𝒑 . FEPEs in Section 3.2.1 and the measured activities in Section 3.2.2) and
( 𝒈 ) ( ) 2
8 ⎛𝜀
𝑃 𝐸𝑁 𝑅𝐴𝑗 𝒙𝒑 , 𝐸𝑖 − 𝜀𝑒𝑥𝑝𝑅𝐴𝑗 𝐸𝑖 ⎞ 𝑓 𝑜𝑟 𝑗 = 1, 2 𝑖𝑛 𝐷𝐸𝑇 1
( ) ∑ 𝑏 ± 𝑢𝑏 the expected or reference value (the experimental FEPEs in Section
𝒈
𝑓𝑅𝐴𝑗 𝒙𝒑 = ⎜ ( ) ⎟ (3) 3.2.1 and the certified activities in Section 3.2.2).
⎜
𝑖=1 ⎝ 𝜀𝑒𝑥𝑝𝑅𝐴 𝐸𝑖 ⎟ 𝑗 = 3 𝑖𝑛 𝐷𝐸𝑇 2.
𝑗 ⎠
𝑎−𝑏
𝑅𝐷(%) = ⋅ 100 (4)
Regarding the parameters chosen for the execution of the optimiza- 𝑏
tion algorithms, a population size of ten times the number of variables |𝑎 − 𝑏|
𝜁𝑠𝑐𝑜𝑟𝑒 = √ (5)
has been chosen, as recommended in Storn and Price [18], it is to say, NP
𝑢2𝑎 + 𝑢2𝑏
= 60 in the case of the optimization of DET 1 by means of DEMO and NP √
= 50 in the optimization of DET2 by DE. In both optimization processes, ( 𝑢 )2 ( 𝑢 )2
𝑎
𝑃𝑠𝑐𝑜𝑟𝑒 = 100 ⋅ + 𝑏 . (6)
the mutation (F) and crossover (CR) factors have been chosen 0.3 and 𝑎 𝑏
0.2 respectively. Both algorithms have been run during 100 generations The source of uncertainties in the calculated FEPEs are mainly the
(stopping criterion). uncertainty of the experimental FEPEs used as reference during the
characterization process, the statistic uncertainty of the PENELOPE
2.6. Validation procedure
calculations, the inaccuracy in the dimensions of the beakers and
samples height considered for the calculations, as well as in the chemical
The validation of the optimized models of DET1 and DET2 has been
composition and density of the materials considered. Given that there
performed in two steps. First of all, the reference materials IAEA RGU-1,
RGTh-1 and RGK-1 have been used to perform a comparison between is no analytical expression which would allow propagating such sources
experimental and computational FEPEs (results in Section 3.2.1.) for a of uncertainty, and that they cannot either be appropriately quantified,
wide variety of measurement geometries and for 19 gamma emissions the uncertainties of the calculated FEPEs have been assigned so that
within the energy range of interest (40–1800 keV), as can be seen in the FEPEs calculated for the reference arrangements (RA1 , RA2 and
Tables 3 and 4. The uncertainties of the experimental FEPEs, determined RA3 ) and the corresponding experimental FEPEs are compatible within
using Eq. (2), are within 0.6–4.7% in the case of DET1 and 0.7%–7.6% uncertainties, it is to say, an uncertainty of 5% for the calculations with
in DET2. the DET1 model and 6% in DET2.
70
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
Table 3
Relative deviations (%) between the experimental FEPEs and the corresponding calculated using the optimized model of DET1 for different heights of the reference materials IAEA RGU-1,
RGTh-1 and RGK-1 in the beaker 1 (B1), beaker 2 (B2), beaker 3 (B3) and beaker 4 (B4).
Material Radionuclide En (keV) B1 B3 B2 B4 Mean
Height (mm)
22.3 32.7 42.6 52.1 72.2 62.2 68.2 101.3 74.8 114.0
RGU-1 210 Pb 46.5 −0.9 2.2 −0.1 0.7 0.6 −1.9 5.7 7.3 9.4 5.7 2.9
234
RGU-1 Th 63.3 −3.5 −3.7 −0.2 −1.1 −2.9 −3.7 3.4 −3.2 2.4 0.5 −1.2
235
RGU-1 U+223 Ra 143.8 −2.2 2.0 −3.8 2.6 0.6 −2.9 −6.1 −4.3 −8.1 −0.9 −2.3
226
RGU-1 Ra+235 U 186.0 −4.1 −0.8 −3.3 −3.6 −2.5 −1.1 −0.6 −1.1 −3.9 0.1 −2.1
214
RGU-1 Pb 242.0 1.8 3.8 1.7 3.2 3.5 3.7 1.1 2.4 −2.1 4.7 2.4
214
RGU-1 Pb 295.2 1.5 3.1 0.4 0.6 1.2 1.4 −0.2 1.6 −5.7 2.2 0.6
RGU-1 214 Pb 352.0 −1.1 2.6 −1.3 −0.3 −2.4 1.3 −2.2 −1.3 −5.9 −0.5 −1.1
RGU-1 214 Bi 609.3 4.7 7.8 5.7 8.8 8.3 9.2 4.2 4.3 −0.5 4.9 5.7
234m
RGU-1 Pa 1001.0 −5.1 −3.1 0.6 −3.7 −1.6 1.7 −2.8 1.1 −4.0 −1.3 −1.8
214
RGU-1 Bi 1120.3 5.9 5.1 4.7 6.4 6.8 8.0 2.6 1.2 0.7 4.2 4.6
214
RGU-1 Bi 1238.1 4.1 8.7 4.6 3.4 5.0 8.7 3.2 4.7 −1.5 7.6 4.9
214
RGU-1 Bi 1764.5 0.2 4.5 −0.3 1.0 2.2 −1.5 2.0 −0.1 −3.5 −1.2 0.3
212
RGTh-1 Pb 238.6 1.2 1.4 2.2 0.5 0.7 2.4 0.8 1.0 −3.4 0.8 0.8
RGTh-1 228 Ac 338.3 4.7 4.2 5.1 2.3 4.7 4.6 −0.3 3.5 −3.5 4.8 3.0
RGTh-1 208 Tl 583.2 3.8 4.9 5.5 6.4 5.1 6.8 2.8 4.8 −2.1 2.8 4.1
212
RGTh-1 Bi+228 Ac 727.3 2.2 3.7 4.8 4.8 2.4 1.9 0.0 1.3 −2.4 2.9 2.2
208
RGTh-1 Tl 860.0 2.0 0.3 1.0 2.0 1.6 1.2 0.8 1.2 −3.6 −1.6 0.5
228
RGTh-1 Ac 911.0 0.8 0.1 2.3 −0.3 0.6 2.6 −2.0 −1.8 −5.9 0.1 −0.4
40
RGK-1 K 1460.8 0.5 −1.9 −4.1 1.8 −0.5 −0.7 −0.3 0.7 −1.3 2.0 −0.4
Mean 0.9 2.4 1.3 1.9 1.8 2.2 0.6 1.2 −2.4 2.0
Table 4
Relative deviations (%) between the experimental FEPEs and the corresponding calculated using the optimized model of DET2 for different heights of the reference materials IAEA RGU-1,
RGTh-1 and RGK-1 in the beaker 3.
Material Radionuclide En (keV) Height (mm) Mean
5 10 15 20 25 30 35 40 45 50
RGU-1 210 Pb 46.5 7.1 8.8 9.6 7.4 1.6 4.0 3.0 1.2 3.2 2.5 4.8
234
RGU-1 Th 63.3 −0.8 3.6 2.0 1.8 −0.3 −0.9 −6.2 −7.6 −3.0 −3.3 −1.5
235
RGU-1 U+223 Ra 143.8 −2.3 −8.4 −4.3 4.6 −6.7 3.5 −6.4 −3.5 −2.4 −4.3 −3.0
226
RGU-1 Ra+235 U 186.0 1.7 1.2 −0.8 3.5 −3.5 −4.2 −3.9 −6.9 −3.5 −4.9 −2.1
214
RGU-1 Pb 242.0 6.4 4.1 4.6 7.7 2.0 2.8 −0.5 −1.0 1.2 −1.6 2.6
RGU-1 214
Pb 295.2 4.0 3.2 1.5 5.9 0.3 0.9 −3.1 −4.0 1.3 −3.0 0.7
RGU-1 214 Pb 352.0 2.5 1.5 −0.9 2.3 0.4 −1.9 −5.8 −5.9 −2.6 −4.6 −1.5
RGU-1 214 Bi 609.3 6.2 9.1 5.5 7.9 4.3 4.1 1.7 2.3 4.5 3.1 4.9
234m
RGU-1 Pa 1001.0 −9.0 9.4 −3.7 2.6 −1.9 8.2 −0.7 −1.6 5.5 5.3 1.4
214
RGU-1 Bi 1120.3 7.5 6.1 7.9 9.3 8.1 4.1 1.3 1.8 6.1 3.4 5.6
214
RGU-1 Bi 1238.1 11.3 6.2 6.6 10.2 6.5 5.0 −1.0 −0.3 4.7 2.5 5.2
214
RGU-1 Bi 1764.5 4.5 1.9 3.6 6.9 −1.2 −1.1 −4.5 −4.4 0.7 −5.3 0.1
RGTh-1 212
Pb 238.6 6.5 2.8 8.8 1.7 −2.1 0.8 −0.1 2.5 −1.2 −2.4 1.7
RGTh-1 228 Ac 338.3 10.9 8.1 10.0 5.3 1.2 2.1 4.2 5.6 3.7 −0.4 5.1
RGTh-1 208 Tl 583.2 5.2 2.9 8.7 4.7 2.3 2.2 2.1 5.4 3.0 4.7 4.1
212
RGTh-1 Bi+228 Ac 727.3 10.8 3.7 10.0 2.5 −2.0 1.2 0.6 3.9 1.3 3.6 3.6
208
RGTh-1 Tl 860.0 −0.7 −3.0 2.8 −2.5 −4.3 −4.9 1.4 −1.7 2.4 −1.4 −1.2
228
RGTh-1 Ac 911.0 10.5 5.5 7.6 1.3 −1.1 −2.9 2.3 2.2 1.2 −3.5 2.3
40
RGK-1 K 1460.8 – 3.9 – 5.3 – 1.9 – −1.4 – −0.9 1.8
Mean 4.6 3.7 4.4 4.7 0.2 1.3 −0.9 −0.7 1.5 −0.6
3. Results and discussion the fact that, in this second option the sum is slightly lower than in the
chosen option, whereas in the first option the difference is considerably
3.1. Optimal solutions lower than in the second option.
The solutions of the optimization process of both detectors models 3.2. Validation of the optimal models
can be seen in the last two columns of Table 1. In the case of DET2,
characterized using DE, the solution with minimum objective function 3.2.1. Computational vs. experimental FEPEs
(𝑓𝑅𝐴3 ≈ 0.00483) has been chosen. In the case of DET1, the DEMO The relative deviations (Eq. (4)) between the experimental and
algorithm reached several non-dominated solutions which would be computational FEPEs can be seen in Tables 3 and 4. In such tables, the
theoretically equally acceptable. In this case, for selecting the definitive 19 emissions considered, the radionuclides which emit them, as well as
solution, the sum of both objective functions (the lower the value, the the corresponding reference material are stated. As can be seen, most
more approximate the behavior of the model to the reality) and the of cases present deviations below the uncertainties considered for the
difference between them (the lower the value, the more the similarity FEPEs calculations (85% of cases below the 5% for DET1 and 79% of
of the model accuracy for both types of measurement geometries) have cases below the 6% for DET2 model), whereas the average deviations
been chosen. Therefore, in this case the selected solution has been the for each geometry are below 3% in every case in DET1 and 5% in DET2,
one with the second best sum of both objective functions (𝑓𝑅𝐴1 ≈ which is a good indicator of the goodness of the detector models built.
0.00380 and 𝑓𝑅𝐴2 ≈ 0.00382) but with the lowest difference between The zeta-score and 𝑃𝑠𝑐𝑜𝑟𝑒 values for all the sample-detector arrange-
both functions. The selection of this solution instead of the solution with ments considered in this section can be seen in Fig. 2 for both detectors.
best sum of functions (𝑓𝑅𝐴1 = 0.00409 and 𝑓𝑅𝐴2 = 0.00325) was due to In such a figure, it can be observed that all the zeta-score values are
71
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
Fig. 2. Zeta-score and P-score between the experimental FEPEs and the corresponding calculated using the optimized models of DET1 (Fig. 2a and b) and DET2 (Fig. 2c and d), for
different heights of the reference materials IAEA RGU-1, RGTh-1 and RGK-1 in the beaker 1 (B1), beaker 2 (B2), beaker 3 (B3) and beaker 4 (B4).
below the limit of 1.64 (critical value for a normal distribution with one. Such methodologies consist of using evolutionary algorithms for
a confidence level of 90%), below which the results to be compared searching the optimal detector model built in a Monte Carlo code, so
are not significantly different [42], confirming therefore the good that the calculated FEPEs are as similar as possible to the reference ones,
agreement between the experimental FEPEs and the ones calculated for determined in this work experimentally using the reference material
both computational models. With regard to the precisions scores, their IAEA RGU-1 for a few reference sample-detector arrangements. In
values are within 5.1%–6.9% in the case of DET1 and within 6.0%–9.7% this work the PENELOPE code has been used for the Monte Carlo
in case of DET2, but being most of them below the 7.5%, which can be simulations, however, other similar codes may be used, provided that
considered acceptable, as the limits of this test rarely are below 10% in such code permits performing an automatic optimization commanded
the IAEA proficiency tests and inter-comparisons [38,39,41]. by a main script, where the optimization algorithms are implemented.
Therefore, the results of the zeta-score and 𝑃𝑠𝑐𝑜𝑟𝑒 , confirm that For the validation of the optimal detector models obtained, calcu-
the computational models built in this work, calculate FEPEs with an lated FEPEs have been compared to the corresponding experimental
acceptable accuracy and precision, even for materials, energies and
ones for a wide variety of geometries, susceptible to be used during the
geometries different to the reference ones (stated in Section 2.4).
normal operation of such detectors, and for the three reference materials
IAEA RGU-1, RGTh-1 and RGK-1, observing acceptable deviations, most
3.2.2. Measurement of certified reference materials
of which below 10%. In order to check the good agreement between the
Tables 5 and 6 show the results of the CRMs measured in DET1 and
computational and experimental FEPEs, the zeta-score and P-score have
DET 2 respectively using the corresponding computational FEPEs, as
been used, observing that in every case studied, the limits of 1.64%
well as the reference values for the corresponding radionuclides, the
and 10% respectively have not been surpassed, which confirms that
relative deviation between them, the zeta-score and P-score values used
the models generate FEPEs with an acceptable accuracy and precision
for the comparison.
according to the requirements set by the IAEA in its proficiency tests
In both Tables 5 and 6, most of cases show zeta-score values below
the limit of 1.64 mentioned before, being all of them below 2.58 and inter-comparisons.
(critical value for a normal distribution with a confidence level of Additionally, several samples of CRMs have been measured in both
99%), used by the IAEA as limit in its proficiency tests and inter- detectors, also for different sample geometries, calibrating in efficiency
comparisons [38,39,41]. The precisions scores observed in such tables using the corresponding calculated FEPEs for the subsequent compari-
are also acceptable according to the limits set in the IAEA proficiency son between the results of the spectrometric analysis and the certified
tests and inter-comparisons, commonly within 15%–30% for natural activities. Again, the z-score and P-score have been used, confirming the
radionuclides in environmental samples, depending on the type of goodness of the detectors models.
sample, level of activity, radionuclide, etc. [29,43–45]. To conclude, despite the relative simplicity and rapidity of the
methodologies applied in this work, without the necessity of having
4. Conclusions the detectors inoperative for a long period nor using very refined
techniques, and having started from a limited knowledge of the actual
Two HPGe gamma-ray detectors owned by two different research geometrical characteristics of the detectors, the two models of HPGe
laboratories on environmental radioactivity have been characterized, detectors obtained allow performing efficiency calibrations suitable for
using a different automatic characterization methodology for each the measurement of environmental samples by gamma spectrometry.
72
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
Table 5
Results of the analysis performed to the samples of different CRMs measured in DET1 (uncertainties given as 1 ⋅ 𝜎), utilizing, for the efficiency calibration, the FEPEs calculated using the
optimized model . Comparison of the measured values with the values certified by the IAEA and CANMET (Measurement dates provided in first column).
CRM Radionuclide En (keV) Certified [Ac] (Bq/kg)a Measured [Ac] (Bq/kg) RD (%) Zeta-score P-score
210 Pb 46.5 337 ± 16 341 ± 18 1.1 0.16 7.2
137 Cs 661.7 357 ± 8 325 ± 16 −8.8 1.70 5.6
40
IAEA 447(16th June 2017 ) K 1460.8 550 ± 20 472 ± 27 −14.3 2.34 6.8
228
Ac 911.0 37 ± 2 31.7 ± 2.5 −14.5 1.65 9.7
212
Pb 238.6 37.0 ± 1.5 33.6 ± 1.9 −9.2 1.41 7.0
226
Ra 352.0 (214 Pb) 25 ± 2 23.4 ± 1.6 −6.3 0.62 10.5
226
Ra 186.0 18980 ± 260 17600 ± 900 −7.6 1.51 5.4
IAEA 448(13rd February 2017 ) 210 Pb 46.5 8200 ± 400 7900 ± 400 −4.0 0.61 6.8
208 Tl 583.2 1010 ± 40 1040 ± 60 2.8 0.40 6.9
228
Ac 911.0 690 ± 30 660 ± 50 −5.2 0.60 8.9
210
Pb 46.5 700 ± 60 650 ± 40 −6.7 0.69 10.0
IAEA 434(5th May 2017 ) 226
Ra 352.0 (214 Pb) 780 ± 60 680 ± 40 −12.4 1.36 9.5
238
U 63.3 (234 Th) 120 ± 11 113 ± 14 −5.5 0.37 15.5
40
K 1460.8 580 ± 9 510 ± 30 −12.5 2.26 6.2
210 Pb 46.5 43 ± 4 48 ± 6 11.9 0.69 16
IAEA 326(6th May 2017 ) 226 Ra 352.0 (214 Pb) 32.6 ± 1.6 31 ± 2 −5.7 0.71 8.3
228
Ra 911 (228 Ac) 39.6 ± 1.8 37 ± 3 −6.1 0.63 10.2
238
U 63.3 (234 Th) 29.4 ± 1.3 29 ± 5 −0.1 0.01 17.9
210
Pb 46.5 31100 ± 900 27800 ± 1400 −10.4 1.91 5.9
235
U 143.8 1498.0 ± 2.0 1520 ± 80 1.3 0.22 5.6
CANMET DH-1a(9th May 2017 ) 238
U 63.3 (234 Th) 32450 ± 40 30000 ± 1700 −6.5 1.26 5.5
226 Ra 186.0b 31500 ± 1100 29000 ± 3000 −8.0 0.78 11.0
232 Th 911 (228 Ac) 3680 ± 120 3720 ± 200 1.2 0.19 6.3
a
Certified activity concentrations for the corresponding radionuclides corrected for radioactive decay and equilibrium with their radioactive precursors using Bateman equations.
b Determined through the direct peak of 186 keV, with counts from both 226 Ra and 235 U, discounting the theoretical counts due to the emission of 235 U, based on the 235 U activity
concentration determined through the line of 143.8 keV.
Table 6
Results of the analysis performed to the samples of different CRMs measured in DET2 (uncertainties given as 1 ⋅ 𝜎), utilizing, for the efficiency calibration, the FEPEs calculated using the
optimized model . Comparison of the measured values with the values certified by the IAEA and CANMET (Measurement dates provided in first column).
CRM Radionuclide En (keV) Certified [Ac] (Bq/kg)a Measured [Ac] (Bq/kg) RD (%) Zeta-score P-score
210 Pb 46.5 336 ± 16 376 ± 24 12.0 1.41 7.9
137 Cs 661.7 356 ± 8 354 ± 21 −0.5 0.08 6.5
40
K 1460.8 550 ± 20 540 ± 30 −2.1 0.29 7.4
IAEA 447(11th July 2017) 228
Ac 911.0 37 ± 2 34.8 ± 2.7 −6.0 0.66 9.5
212
Pb 238.6 37.0 ± 1.5 37.0 ± 2.3 −0.1 0.02 7.5
226
Ra 352.0 (214 Pb) 25 ± 2 20.7 ± 1.6 −16.9 1.63 11.3
226
Ra 186.0 18980 ± 260 17700 ± 1100 −6.8 1.15 6.3
210 Pb 46.5 8400 ± 400 8000 ± 500 −4.4 0.61 7.5
IAEA 448(7th July 2017) 208 Tl 583.2 960 ± 40 960 ± 60 −0.8 0.10 7.7
228
Ac 911.0 660 ± 30 650 ± 50 −1.8 0.2 9.1
210
Pb 46.5 700 ± 60 660 ± 40 −6.0 0.59 10.5
226
IAEA 434(27th December 2016) Ra 352.0 (214 Pb) 780 ± 60 730 ± 40 −6.3 0.64 10.0
238
U 63.3 (234 Th) 120 ± 11 120 ± 16 −0.3 0.02 16.0
40
K 1460.8 580 ± 9 670 ± 40 15.1 2.00 6.6
210 Pb 46.5 43 ± 4 47 ± 5 9.4 0.62 14.2
IAEA 326(3rd February 2015) 226 Ra 352.0 (214 Pb) 32.6 ± 1.6 35.6 ± 2.4 9.0 1.03 8.3
228
Ra 911 (228 Ac) 39.6 ± 1.8 42 ± 3 5.6 0.59 9.1
238
U 63.3 (234 Th) 29.4 ± 1.3 33 ± 6 11.0 0.55 17.9
210
Pb 46.5 31100 ± 900 27900 ± 1700 −10.1 1.62 6.8
235
U 143.8 1498.0 ± 2.0 1590 ± 110 6.2 0.82 7.1
238
CANMET DH-1a(20th March 2017) U 63.3 (234 Th) 32450 ± 40 31600 ± 2000 −2.6 0.42 6.4
226 Ra 186.0b 31500 ± 1100 33000 ± 4000 3.9 0.30 12.6
232 Th 911 (228 Ac) 3680 ± 120 3930 ± 260 6.8 0.88 7.3
a
Certified activity concentrations for the corresponding radionuclides corrected for radioactive decay and equilibrium with their radioactive precursors using Bateman equations.
b Determined through the direct peak of 186 keV with counts from both 226 Ra and 235 U, discounting the theoretical counts due to the emission of 235 U, based on the 235 U activity
concentration determined through the line of 143.8 keV.
73
J.G. Guerra et al. Nuclear Inst. and Methods in Physics Research, A 880 (2018) 67–74
[6] M.S. Badawi, M.M. Gouda, S.S. Nafee, A.M. El-Khatib, E.A. El-Mallah, Nucl. Instrum. [28] IAEA-448–Radium-226 in soil from oil field. <http://nucleus.iaea.org/rpst/Referen
Methods Phys. Res. A 696 (2012) 164. ceProducts/ReferenceMaterials/Radionuclides/RS_IAEA-448.pdf>.
[7] L. Liye, M. Jizeng, D. Franck, L. de Carlan, Z. Binquan, Nucl. Instrum. Methods Phys. [29] Worldwide open proficiency test: determination of naturally occurring radionuclides
Res. A 564 (2006) 608. in phosphogypsum and water IAEA-CU-2008-03. IAEA Analytical Quality in Nuclear
[8] A. Tedjani, C. Mavon, A. Belafrites, D. Degrelle, D. Boumala, D. Rius, J.-E. Groetz, Applications Series No. IAEA/AQ/18 International Atomic Energy Agency, Vienna,
Nucl. Instrum. Methods Phys. Res. A 838 (2016) 12. 2010.
[9] D. Degrelle, C. Mavon, J.-E. Groetz, Nucl. Instrum. Methods Phys. Res. A 816 (2016) [30] A. Shakhashiro, U. Sansone, H. Wershofen, A. Bollhöfer, C.K. Kim, C.S. Kim, G. Kis-
47. Benedek, M. Korun, M. Moune, S.H. Lee, S. Tarjan, M.S. Al-Masri, Appl. Radiat. Isot.
[10] A.S.E. Santo, F.G. Wasserman, C.C. Conti, Ann. Nucl. Energy 46 (2012) 213. 69 (2011) 231.
[11] R. Casanovas, J.J. Morant, M. Salvadó, Nucl. Instrum. Methods Phys. Res. A 675 [31] Report on the Intercomparison Run for the Determination of Radionuclides in
(2012) 78. Soils IAEA-326 and IAEA-327. IAEA/AL/100. International Atomic Energy Agency,
[12] Y. Morera-Gómez, H.A. Cartas-Aguila, C.M. Alonso-Hernández, C. Nuñez Duartes, Vienna, 2001.
Nucl. Instrum. Methods Phys. Res. A 818 (2016) 51. [32] CANMET: <https://www.nrcan.gc.ca/mining-materials/certified-reference-materia
[13] R. Britton, A.V. Davies, Nucl. Instrum. Methods Phys. Res. A 786 (2015) 12. ls/certificate-price-list/8009>.
[14] R. Luís, J. Bento, G. Carvalhal, P. Nogueira, L. Silva, P. Teles, P. Vaz, Nucl. Instrum. [33] DH-la: A Certified Uranium-Thorium Reference Ore, CANMET Report 81-11E,
Methods Phys. Res. A 623 (2010) 1014. Canada Centre for Mineral and Energy Technology, Canada, 1981.
[15] E. García-Toraño, B. Caro, V. Peyrés, M. Mejuto, Nucl. Instrum. Methods Phys. Res. [34] Chart of the Nuclides, National Nuclear Data Center, Brookhaven National Labora-
A 837 (2016) 63. tory, <http://www.nndc.bnl.gov/chart/>.
[16] J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo, A. Tejera, [35] G. Xhixha, M. Alberi, M. Baldoncini, K. Bode, E. Bylyku, F. Cfarku, I. Callegari, F.
J.M. Gil, R. Rodríguez, P. Martel, J. Environ. Radioact. 149 (2015) 8. Hasani, S. Landsberger, F. Mantovani, E. Rodriguez, F. Shala, V. Strati, M. Xhixha
[17] J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo, A. Tejera, Kaçeli, J. Radioanal. Nucl. Chem. 307 (2016) 1507.
P. Martel, J.P. Bolivar, Nucl. Instrum. Methods Phys. Res. A 858 (2017) 113. [36] E. Tomarchio, S. Rizzo, Radiat. Phys. Chem. 80 (2011) 318.
[18] R. Storn, K. Price, J. Global Optim. 11 (1997) 341. [37] International Organization for Standardization, 2005. Statistical methods for use
[19] F. Salvat, J.M. Fernández-Varea, J. Sempau, PENELOPE-2011: A Code System for in proficiency testing by interlaboratory comparisons, in: ISO 13528 (E) Geneva,
Monte Carlo Simulation of Electronic and Photon Transport, OECD, Barcelona, Switzerland.
Spain, 2011. [38] A. Shakhashiro, L. Mabit, Appl. Radiat. Isot. 67 (2009) 139.
[20] T. Robič, B. Filipič, DEMO: Differential evolution for multi-objective optimization, [39] A. Shakhashiro, A. Trinkl, U. Sansone, Appl. Radiat. Isot. 66 (2008) 1722.
in: Evolutionary Multi-Criterion Optimization, EMO 2005, in: Lecture Notes in [40] I. Osvath, S. Tarjan, A. Pitois, M. Groening, D. Osborn, Appl. Radiat. Isot. 109 (2016)
Computer Science, vol. 3410, Springer, Berlin, 2005, p. 520. 90.
[21] Model S573/S574 ISOCS/LabSOCS Validation & Verification Manual V.4.0, Can- [41] A. Harms, M. Khanh Pham, O. Blinova, S. Tarjan, H. Nies, I. Osvath, Appl. Radiat.
berra Industries Inc., 2002. Isot. 126 (2017) 252.
[22] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials [42] Report on the Second Almera Network Coordination Meeting and the Almera
RGU-1, RGTh-1 and RGK-1, IAEA Report RL148, International Atomic Energy Soil Sampling Intercomparison Exercise IAEA/SIE/01. IAEA/AL/164. International
Agency, Vienna, 1987. Atomic Energy Agency. Seibersdorf, 2006.
[23] Genie 2000, Basic Spectroscopy Software Version 2.1., Canberra Industries Inc., [43] Worldwide Open Proficiency Test: Determination of Natural and Artificial Radionu-
2002. clides in Moss-Soil and Water IAEA-CU-2009-03. IAEA Analytical Quality in Nuclear
[24] A. Shakhashiro, S. Tarjan, A. Ceccatelli, G. Kis-Benedek, M. Betti, Appl. Radiat. Isot. Applications Series No. IAEA/AQ/22. International Atomic Energy Agency, Vienna,
70 (2012) 1632. 2012.
[25] IAEA-447- Natural and artificial radionuclides in moss-soil. <http://nucleus.iaea. [44] ALMERA Proficiency Test: Determination of Natural and Artificial Radionuclides in
org/rpst/Documents/RS_IAEA-447.final.pdf>. Soil and Water IAEA-TEL-2011-04. IAEA Analytical Quality in Nuclear Applications
[26] A.R. Iurian, A. Pitois, G. Kis-Benedek, A. Migliori, R. Padilla-Alvarez, A. Ceccatelli, Series No. IAEA/AQ/32. International Atomic Energy Agency, Vienna, 2013.
Appl. Radiat. Isot. 109 (2016) 61. [45] Worldwide Open Proficiency Test on the Determination of Radionuclides in Spinach,
[27] A. Ceccatelli, R. Katona, G. Kis-Benedek, A. Pitois, Appl. Radiat. Isot. 87 (2014) 461. Soil and Water IAEA-CU-2007-03. IAEA Analytical Quality in Nuclear Applications
Series No. IAEA/AQ/8. International Atomic Energy Agency, Vienna, 2007.
74
Capítulo 4 85
Capítulo 5
Aplicación de la metodología de
caracterización a un detector HPGe tipo
pozo empleando datos experimentales
(cuarta publicación)
J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo, A.
Tejera, P. Martel, J.P. Bolivar, Modeling of a HPGe well detector using
PENELOPE for the calculation of Full Energy Peak Efficiencies for
environmental samples, Nuclear Inst. and Methods in Physics Research, A
908 (2018) 206
Capítulo 4 86
Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Canaria, Spain
d
Departamento de Física Aplicada, Universidad de Huelva, 21071 Huelva, Spain
∗ Corresponding author at: Departamento de Física, Universidad de Las Palmas de Gran Canaria, 35001 Las Palmas de Gran Canaria, Spain.
E-mail address: jglezg2002@gmail.com (J. G.Guerra).
https://doi.org/10.1016/j.nima.2018.08.048
Received 18 April 2018; Received in revised form 16 August 2018; Accepted 17 August 2018
Available online xxxx
0168-9002/© 2018 Elsevier B.V. All rights reserved.
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
The Monte Carlo method has been applied in the calculation of Table 1
FEPEs for several decades [9,10] and is now an increasingly widely Experimental FEPEs used as a reference during the optimization process of the computa-
tional model of the detector (sample height = 51 mm).
used alternative [11–16] in view of the difficulties related to experi-
Radionuclide En (keV) 𝐹 𝐸𝑃 𝐸 Uncertainty (1𝜎)
mental calibration and the enhanced capabilities of modern computers.
210 Pb 46.5 0.272 0.003
However, a computational model of a detector with a similar response
234 Th 63.3 0.350 0.008
to the real one must be utilized in order to obtain reliable FEPEs by 235 U+223 Ra 143.8 0.365 0.009
direct Monte Carlo simulation. For this reason, many recent publications 236
Ra+235 U 186.2 0.294 0.003
have focused on the modeling of gamma-ray detectors [17–20], also 214
Pb 295.2 0.1948 0.0014
known as detector characterization, including two papers by the current 214
Pb 351.9 0.1652 0.0011
234m
authors [21,22], in which automatic methodologies for modeling HPGe Pa 1001.0 0.0659 0.0025
214 Bi 1764.5 0.0392 0.0003
detectors were developed, and which were validated using two real
coaxial detectors in a third publication [23].
In the first of these papers [21], the characterization methodology
involves the use of the Differential Evolution (DE) algorithm [24] for
2.3. Reference FEPEs
the optimization of a detector model built using the PENELOPE Monte
Carlo simulation code [25]. This optimization starts from a sufficiently
Calibration of the detector in terms of energy and resolution was
accurate set of reference FEPEs obtained for a specific sample-detector
performed using a standard point source of 22 Na and 155 Eu, provided
arrangement and sample material. In the current paper, this method-
ology is applied to an HPGe well detector used for the measurement of with the detector by Canberra. As a result, we obtained the correspond-
ing calibration functions (energy in keV): 𝑒𝑛𝑒𝑟𝑔𝑦 = 0.3787 + 2.1212 ⋅
environmental samples. The computational model of the detector should √
allow the calculation of FEPEs with acceptable precision and sufficient 𝑐𝑒𝑛𝑡𝑟𝑜𝑖𝑑 and 𝐹 𝑊 𝐻𝑀 = 0.9672 + 0.0254 ⋅ 𝑒𝑛𝑒𝑟𝑔𝑦. The experimental
accuracy for efficiency calibration with this type of sample, irrespective FEPEs used as a reference during the optimization process for the
of whether the dimensions considered in the model are an exact match computational model of the detector were obtained by analyzing a
for the real dimensions of the detector. RGU-1 sample with a height of 51 mm (i.e. the beaker completely
filled). Fig. 1 shows the corresponding spectrum acquired. Table 1
2. Material and methods shows the gamma lines chosen, the radionuclides emitting these and
the corresponding experimental FEPEs with their uncertainties. These
2.1. Instrumentation eight gamma emissions were selected because they are virtually free
of coincidence summing effects and of considerable interferences with
In this work, a Canberra HPGe well detector (model GCW4023) is other peaks of the spectrum and are sufficiently intense in order that
used. This detector has a relative efficiency of 40%, a nominal FWHM
the peaks stand out sufficiently from the continuum. A sample height of
of 1.2 keV at 122 keV and 2.0 keV at 1.33 MeV and a peak-to-Compton
51 mm was chosen to obtain the reference FEPEs, since this height gives
ratio of 60.8:1 at 1.33 MeV. The detector is connected to a DSA-1000
the highest counting rate and therefore the lowest counting uncertainty;
Canberra multichannel analyzer and is located within an iron shield
in addition, with the beaker completely filled, potential uncertainties in
with a thickness of 15 cm. The acquisition and analysis of the gamma-
the sample height are eliminated.
ray spectra is performed using the Genie 2000 software package [26].
A Radon Scout detector from SARAD GmbH [27] was been used to The experimental FEPEs and uncertainties are determined using
determine the escape of 222 Rn from samples of one of the reference Eqs. (1) and (2), where 𝐶 represents the net counts observed in the
materials used in this work, due to its relatively high activity concen- relevant full energy peak (subtracting both the Compton Continuum and
tration of 226 Ra and exhalation rate (see Section 2.4). This detector and the laboratory background, using GENIE2000 [26]), [𝐴𝑐] the activity
the samples were placed within an accumulation radon-tight chamber, concentration of the relevant radionuclide, 𝑃𝛾 the corresponding proba-
made of methacrylate with an internal volume of 0.185 ± 0.006 m3 and bility of the gamma emission, 𝑡 the acquisition time, 𝑚 the sample mass,
walls 10 mm thick [28]. and 𝑢𝐶 , 𝑢[𝐴𝑐] , 𝑢𝑃𝛾 , 𝑢𝑚 are the corresponding uncertainties. The values of
𝑃𝛾 and the uncertainties used in this paper were obtained from the
2.2. Samples of reference materials used NationalNuclearDataCenter [35].
𝐶
Reference materials IAEA RGU-1, IAEA RGTh-1, IAEA RGK-1, IAEA 𝐹 𝐸𝑃 𝐸 = (1)
[𝐴𝑐] ⋅ 𝑃𝛾 ⋅ 𝑡 ⋅ 𝑚
447 and IAEA 448 were used for the generation and validation of the √
model of detector. The first three of these are meant for the experimental ( 𝑢 )2 ( 𝑢 )2 ( 𝑢 )2 ( )
𝑃𝛾 𝑢 2
𝐶 [𝐴𝑐]
calibration of gamma-ray spectrometers for potassium, uranium and 𝑢𝐹 𝐸𝑃 𝐸 = 𝐹 𝐸𝑃 𝐸 + + + 𝑚 (2)
𝐶 [𝐴𝑐] 𝑃𝛾 𝑚
thorium [29]. IAEA 448 [30–32] and IAEA 447 [32–34] are envi-
ronmental samples with certified activity for several radionuclides of 2.4. Determination of the escape of radon for the case of IAEA 448
interest to environmental radioactivity laboratories. Both the chemical
composition and the reference activity concentrations were obtained The reference material IAEA 448 contains relatively high concentra-
from the documents referenced above. tions of 226 Ra (19 kBq/kg), which, together with the high exhalation
Five samples were prepared for each of these reference materials
rate of this material, means that the escape of radon from the beaker is
by placing the materials directly into the beakers with different sample
a considerable source of error when measuring the 226 Ra concentration
heights (10, 20, 30, 40 and 51 mm) and densities ranging from 1 to
through the gamma emissions corresponding to 222 Rn progeny (214 Pb
2 g/cm3 . All the samples were prepared in cylindrical polypropylene
and 214 Bi) [30]. For this reason, a correction factor was determined for
beakers with an internal diameter of 11.5 mm, an internal height of
51 mm and a thickness of 1 mm. In the case of materials containing the escape of radon when measuring this material.
226 Ra, once shut and sealed, the beakers remained undisturbed for To determine this correction factor, 20 beakers, identical to those
at least four weeks (approximately seven times the half-life of 222 Rn) used for gamma-ray analysis, were completely filled with IAEA 448
before performing each spectrum acquisition, in order to attain radioac- with a total mass of 81.2 ± 0.1 g. These 20 samples were placed
tive equilibrium between 226 Ra and 222 Rn. In the specific case of the within the accumulation chamber described in Section 2.1, with the
IAEA 448, we carried out experimental checks that, even sealing with Radon Scout [27], in order to measure the 222 Rn concentration (activity
aluminum tape (as we perform routinely in the laboratory), a non- per unit volume) accumulated in the chamber over a period of one
negligible escape of radon was noticed. The procedure used to correct month, during which time the detector recorded the 222 Rn concentration
the results for this effect is explained in Section 2.4. every three hours. The recorded values were fitted to the curve shown
207
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Fig. 1. Spectrum acquired from the reference sample (IAEA RGU-1 with 51 mm of height); image generated using GENIE 2000 [26].
2.5. Software and codes Fig. 2. Model of the HPGe well detector used for the simulations, created with PENELOPE.
The denomination of the acronyms and known values for the characteristics are shown in
Table 2.
The computational model of the detector used in this work was
generated using the PENELOPE 2011 Monte Carlo simulation code [25].
This consists of a set of subroutines written in FORTRAN that allow
2.6. Model of detector and optimization process
Monte Carlo simulation of coupled electron–photon transport to be
carried out, for a wide energy range, from a few hundred eV to about 1
GeV. For the correct functioning of PENELOPE, the user must provide a The simplified model of the HPGe well detector considered in this
main script that controls the entire process and the evolution of each work and its geometrical characteristics can be seen in Fig. 2. Table 2
event. Given the cylindrical symmetry of the detector and samples, presents the values of the known characteristics, provided by Canberra
we used the pencyl.f subroutine, created for the simulation of particle Industries Inc. and considered to be constant, the definition of the
transport in cylindrical geometries and included with the PENELOPE variables of the optimization process, and the search interval for each
distribution package, as the main program. From the compilation of variable. The materials considered for the model are aluminum (for the
PENELOPE using pencyl.f as the main subroutine, an executable file was
detector endcap) and germanium (for the crystal, including dead layers).
created. This executable file requires an input file with the information
The optimization process for the model of the detector is based on a
necessary to perform the simulation, including the geometry of the
detector. methodology developed [21] and validated [23] in previous works by
For the optimization of the computational model of the detector, the current authors for HPGe XtRa detectors, in which the DE mono-
the DE algorithm [24] was implemented in a script in Matlab©. This objective optimization algorithm [24] was used. The DE algorithm
script also contained command lines that generated the PENELOPE input works with a constant number
( of NP individuals ) (candidate solutions).
𝒈
file mentioned above, and also read the PENELOPE output file with These are vectors, 𝒙𝒑 = 𝑥𝑔𝑝,1 , … , 𝑥𝑔𝑝,𝑚 , … , 𝑥𝑔𝑝,𝑛 , whose components 𝑥𝑔𝑝,𝑚
the simulated spectrum, used to calculate the FEPEs for the energies are the variables of the problem (the seven unknown characteristics
of interest. shown in Table 2 for this work). The variables only can take values
208
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Table 2
g
Characteristics used in the detector model. Column 3 shows the known dimensions. In column 4 the variables of the optimization problem (xp,m ) are defined. The minimum (xminm
) and
maximum (xmaxm
) values for each variable are shown in columns 5 and 6 respectively. The optimum values of the variables are shown in column 7. All values are given in mm.
Characteristic Denomination Known dimensions Variable xmin
m xmax
m Solution
CD Crystal diameter 67.00 – – – –
CL Crystal length 67.00 – – – –
EWD Endcap well diameter 16.00 – – – –
EWL Endcap well length 40.00 – – – –
g
CWD Crystal well diameter – xp,1 17.00 28.00 24.42
g
CWL Crystal well length – xp,2 26.50 43.80 41.86
g
EWT Endcap well thickness – xp,3 0.30 2.50 2.45
g
CED Crystal to endcap distance – xp,4 5.00 25.00 10.59
g
SDT Superior dead layer thickness – xp,5 0.30 1.60 0.38
g
LDT Lateral dead layer thickness – xp,6 0.30 1.60 1.38
g
IDT Inferior dead layer thickness – xp,7 0.30 1.60 0.74
within the corresponding search interval defined by the user (shown in three gamma lines, Eq. (7) was used, where 𝐹 𝐸𝑃 𝐸𝑎𝑝𝑝𝑎𝑟𝑒𝑛𝑡 is the FEPE
Table 2 for this work). obtained from the spectrum using Eq. (1) and 𝐹 𝐸𝑃 𝐸𝑐𝑢𝑟𝑣𝑒 is the FEPE
The optimization process starts with the random generation of the estimated by the calibration curve.
first NP individuals within the mentioned search domain. These are ∑𝑖=3 𝑖
then evaluated, using the objective function to be minimized. Once the 𝐹 𝐸𝑃 𝐸𝑐𝑢𝑟𝑣𝑒 = 𝑒 𝑖=0 𝑐𝑖 ln 𝐸 (6)
initial individuals have been created and evaluated, an iterative process 𝐹 𝐸𝑃 𝐸𝑎𝑝𝑝𝑎𝑟𝑒𝑛𝑡
𝐶𝑆𝐶𝐹 = (7)
commences in which new individuals are generated using mutation and 𝐹 𝐸𝑃 𝐸𝑐𝑢𝑟𝑣𝑒
crossover operators. These new individuals are then also evaluated in
the same manner, and the NP candidate solutions with a lower value 2.8. Validation of the model
of the objective function are selected for the next generation g, while
the remaining individuals are discarded. Further details can be found
In order to validate the optimized model, the computational FEPEs
in [21].
corresponding to the 25 samples with different heights and materials,
The objective function to be minimized (Eq. (5)) is the sum, for the
𝒈
individual 𝒙𝒑 to be tested, of the squared relative differences between described in Section 2.2, were calculated by PENELOPE using the
eight pairs of FEPEs (one for each of the energies 𝐸𝑖 in Table 1). These model. The corresponding spectra were acquired and the computational
are determined using PENELOPE (𝜀𝑃 𝐸𝑁 𝑟𝑒𝑓 ) and experimentally (𝜀𝑒𝑥𝑝 𝑟𝑒𝑓 ) FEPEs were used to carry out efficiency calibration, comparing the
for the reference sample (Section 2.3). results of the analysis with the reference activity concentrations of
( ( ) )2 various radionuclides of interest in terms of environmental radioactivity
( 𝒈) ∑ 𝒈
𝜀𝑃 𝐸𝑁 𝑟𝑒𝑓 𝒙𝒑 ,𝐸𝑖 −𝜀𝑒𝑥𝑝 𝑟𝑒𝑓 (𝐸𝑖 ) measurements. In these analyses, the correction factors RECF and CSCF
𝑓 𝒙𝒑 = 8𝑖=1 𝜀𝑒𝑥𝑝 𝑟𝑒𝑓 (𝐸𝑖 )
(5)
were considered when appropriate (dividing the uncorrected activities
by these factors and propagating uncertainties).
2.7. Determination of the coincidence summing correction factors To compare the results of these analyses with the reference activ-
ity concentrations, the ratio between them (Eq. (8)), the zeta-score
Due to the high detection efficiency of the detector used in this work, (Eq. (9)), also known as u𝑡𝑒𝑠𝑡 , and the Precision Score (Eq. (10)) were
some of the gamma lines used are strongly affected by the coincidence used. The last two are tests of accuracy and precision, respectively,
summing effect. The corresponding Coincidence Summing Correction and are commonly used by the IAEA in inter-comparison exercises and
Factors (CSCF) therefore need to be determined in order to obtain proficiency tests [39–42].
sufficiently accurate results by means of these lines. The equations
𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑣𝑎𝑙𝑢𝑒
proposed by Xhixha et al. [37] for gamma lines at 242.0, 338.3, 𝑅= (8)
𝑅𝑒𝑓 𝑒𝑟𝑒𝑛𝑐𝑒 𝑣𝑎𝑙𝑢𝑒
583.2, 860.0 and 911.0 keV and the equation proposed by Tomarchio
|𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑣𝑎𝑙𝑢𝑒 − 𝑅𝑒𝑓 𝑒𝑟𝑒𝑛𝑐𝑒 𝑣𝑎𝑙𝑢𝑒|
and Rizzo [38] for the gamma line at 727.3 keV were used for the 𝜁𝑠𝑐𝑜𝑟𝑒 = √ (9)
calculations of the CSCF. The FEPEs and total efficiencies required to 𝑢𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑉 𝑎𝑙𝑢𝑒 2 + 𝑢𝑅𝑒𝑓 𝑒𝑟𝑒𝑛𝑐𝑒 𝑉 𝑎𝑙𝑢𝑒 2
determine the CSCF by means of these equations were computed using √
( 𝑢 )2 ( 𝑢𝑅𝑒𝑓 𝑒𝑟𝑒𝑛𝑐𝑒 𝑣𝑎𝑙𝑢𝑒 )2
the optimized model of the detector in PENELOPE. 𝑃𝑠𝑐𝑜𝑟𝑒 = 100 ⋅ 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑣𝑎𝑙𝑢𝑒
+ (10)
Owing to the high level of complexity of the decay scheme for 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑣𝑎𝑙𝑢𝑒 𝑅𝑒𝑓 𝑒𝑟𝑒𝑛𝑐𝑒 𝑣𝑎𝑙𝑢𝑒
214 Bi, the equations proposed by Xhixha et al. [37] (for the gamma
lines of 609.3, 1120.3 and 1238 keV) consider the contributions from 3. Results and discussion
several coincident energies to be negligible for simplicity. This is a
suitable approximation for coaxial detectors, for which such equations 3.1. Optimal model obtained
are mainly used, however, we found that this simplification involves
a considerable underestimate of the coincidence summing effect in the
case of our well detector. This occurs because in this type of detector, For the execution of the DE algorithm, a population size of five times
and particularly at low sample heights, the solid angle approaches 4𝜋 the number of variables was chosen, i.e. NP = 35. The mutation and
sr, giving a considerably higher detection efficiency than for coaxial crossover factors [21–24] were chosen as 0.3 and 0.2, respectively. The
detectors. stopping criterion was a maximum of 100 generations.
The CSCF for the gamma lines at 609.3, 1120.3 and 1238 keV In this work, the combination of geometric characteristics of the
were therefore determined by employing an efficiency calibration curve detector model chosen as an optimal solution is the one for which
(Eq. (6)) for each sample height. These curves were obtained by fitting the objective function (Eq. (5)) has a minimum value. This value is
( )
the experimental FEPEs determined for the IAEA RGU-1 material and 𝑓 𝒙𝒐𝒑𝒕𝒊𝒎𝒖𝒎 ≈ 0.00419, and is the solution with the closet behavior to
the same eight gamma lines employed as a reference in Table 1 (as they the real detector for the sample of reference. The characteristics of the
are virtually free of this effect). For the calculation of the CSCF for these optimal model of the detector are shown in the last column of Table 2.
209
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Table 3
Results of gamma spectrometric analysis for the samples of IAEA RGU-1, RGTh-1 and RGK-1 for five different sample heights. Activity concentrations given in Bq/kg (uncertainties given
as 1 ⋅ 𝜎). Values refer to the reference date of the three materials (1st January 1987).
Mat. Nuclide En (keV) Reference values 10 mm 20 mm 30 mm 40 mm 51 mm
RGU-1 210 Pb 46.5 4940 ± 30 4630 ± 290 4900 ± 300 4800 ± 300 4650 ± 280 4610 ± 280
RGU-1 234 Th 63.3 4940 ± 30 5200 ± 300 5200 ± 300 5200 ± 300 5000 ± 300 4900 ± 300
RGU-1 235 U 143.8 228 ± 2 255 ± 18 236 ± 16 250 ± 15 230 ± 14 240 ± 15
226
RGU-1 Ra 186.2 4940 ± 30 4900 ± 300 5000 ± 300 5300 ± 300 5300 ± 300 5300 ± 300
214
RGU-1 Pb 242.0 4940 ± 30 4900 ± 300 5100 ± 300 5100 ± 300 5000 ± 300 4690 ± 280
214
RGU-1 Pb 295.2 4940 ± 30 5300 ± 300 5300 ± 300 5300 ± 300 5100 ± 300 4900 ± 300
214
RGU-1 Pb 351.9 4940 ± 30 5200 ± 300 5300 ± 300 5300 ± 300 5100 ± 300 5000 ± 300
RGU-1 214 Bi 609.3 4940 ± 30 5400 ± 600 5200 ± 300 5100 ± 400 4900 ± 400 5000 ± 400
RGU-1 234m Pa 1001.0 4940 ± 30 5800 ± 700 5100 ± 400 5100 ± 400 4900 ± 400 5000 ± 300
RGU-1 214 Bi 1120.3 4940 ± 30 5200 ± 500 4900 ± 300 4800 ± 400 4700 ± 300 4600 ± 400
214
RGU-1 Bi 1238.1 4940 ± 30 5200 ± 500 4900 ± 300 4800 ± 400 4700 ± 300 4700 ± 400
214
RGU-1 Bi 1764.5 4940 ± 30 5200 ± 300 5300 ± 300 5100 ± 300 5000 ± 300 4770 ± 290
212
RGTh-1 Pb 238.6 3250 ± 90 3500 ± 210 3540 ± 210 3530 ± 210 3490 ± 210 3400 ± 180
228
RGTh-1 Ac 338.3 3250 ± 90 3260 ± 200 3230 ± 200 3200 ± 200 3150 ± 190 3100 ± 220
RGTh-1 208 Tl 583.2 1170 ± 30 1310 ± 100 1270 ± 100 1200 ± 90 1100 ± 90 1050 ± 80
RGTh-1 212 Bi 727.3 3250 ± 90 3320 ± 220 3350 ± 210 3390 ± 210 3280 ± 210 2960 ± 190
RGTh-1 208 Tl 860.6 1170 ± 30 1160 ± 90 1170 ± 80 1130 ± 80 1100 ± 80 1090 ± 80
228
RGTh-1 Ac 911.2 3250 ± 90 3010 ± 180 3050 ± 190 3000 ± 180 2970 ± 180 2970 ± 190
40
RGK-1 K 1460.8 14000 ± 400 14600 ± 1000 14500 ± 900 13800 ± 900 13800 ± 900 14700 ± 800
Table 4
Results of gamma spectrometric analysis for the samples of IAEA 447 and 448 for the five different heights. Activity concentrations given in Bq/kg (uncertainties given as 1 ⋅ 𝜎). Values
refer to the reference dates of the materials (1st January 2009 for 226 Ra in IAEA 448, 8th October 2012 for the remaining radionuclides in IAEA 448 and 15th November 2009 for the
IAEA 447).
Mat. Nuclide En (keV) Reference values 10 mm 20 mm 30 mm 40 mm 51 mm
448 210 Pb 46.5 6653 ± 283 6500 ± 400 6700 ± 400 6300 ± 400 6100 ± 400 6600 ± 300
448 226 Ra 186.2 19050 ± 260 18800 ± 1100 19000 ± 1200 18400 ± 1100 18200 ± 1100 17900 ± 900
212
448 Pb 238.6 1623 ± 69 1740 ± 60 1700 ± 100 1660 ± 60 1760 ± 60 1710 ± 70
214
448 Pb 242.0 19050 ± 260 19500 ± 1600 20000 ± 1600 19300 ± 1500 19000 ± 1400 18300 ± 1700
214
448 Pb 295.2 19050 ± 260 19300 ± 1600 19800 ± 1700 18800 ± 1600 18900 ± 1500 18300 ± 1600
228
448 Ac 338.3 1116 ± 55 1080 ± 80 1120 ± 90 1070 ± 80 1030 ± 70 1040 ± 80
448 214 Pb 351.9 19050 ± 260 19000 ± 1600 19400 ± 1700 18600 ± 1600 18600 ± 1500 18000 ± 1500
448 214 Bi 609.3 19050 ± 260 19600 ± 1900 19500 ± 1900 18700 ± 1800 18800 ± 1800 20000 ± 2000
448 228 Ac 911.2 1116 ± 55 1140 ± 80 1070 ± 80 1080 ± 70 1090 ± 60 1090 ± 80
214
448 Bi 1120.3 19050 ± 260 19000 ± 1900 18100 ± 1800 17400 ± 1700 17500 ± 1600 19200 ± 1900
214
448 Bi 1238.1 19050 ± 260 17900 ± 1800 19100 ± 1900 17900 ± 1700 18000 ± 1600 19200 ± 1900
214
448 Bi 1764.5 19050 ± 260 18600 ± 1600 19200 ± 1600 18500 ± 1600 18500 ± 1500 18100 ± 1500
210
447 Pb 46.5 420 ± 20 490 ± 40 450 ± 30 440 ± 30 440 ± 30 420 ± 30
447 212 Pb 238.6 37.3 ± 2.0 43 ± 5 40 ± 3 32 ± 3 39 ± 3 38 ± 3
447 214 Pb 295.2 25.1 ± 2.0 22 ± 8 21 ± 5 25 ± 4 23 ± 4 17 ± 3
447 228 Ac 338.3 37.3 ± 2.0 44 ± 15 34 ± 9 28 ± 6 42 ± 6 42 ± 5
214
447 Pb 351.9 25.1 ± 2.0 20 ± 5 26 ± 3 21.4 ± 2.4 19.9 ± 2.1 18.3 ± 2.2
208
447 Tl 583.2 13.4 ± 0.7 25 ± 6 14 ± 3 17.0 ± 2.2 13.7 ± 1.8 13.3 ± 1.7
214
447 Bi 609.3 25.1 ± 2.0 40 ± 9 32 ± 5 36 ± 4 26 ± 3 24 ± 3
137
447 Cs 661.7 425 ± 10 433 ± 26 431 ± 26 414 ± 25 420 ± 25 403 ± 24
447 228 Ac 911.2 37.3 ± 2.0 52 ± 10 37 ± 7 49 ± 5 41 ± 5 38 ± 4
3.2. Validation results performed in this work, the maximum dead time observed was 0.22% of
the real acquisition time (corresponding to the sample with the highest
The results of analysis of the 25 samples measured are presented counting rate mentioned above); hence, neither the dead time nor the
in Table 3 (IAEA RGU-1, RGTh-1 and RGK-1) and Table 4 (IAEA 447 pile-up effect are significant sources of uncertainty in the final results.
and 448), together with the radionuclides measured, the gamma lines The results for the gamma lines affected by the coincidence summing
employed and the corresponding reference activity concentrations. The effect (see Section 2.7) and by escape of 222 Rn (see Section 2.4) were
reference values shown in Tables 3 and 4 correspond to the reference corrected, as explained above. The resulting CSCF range was between
dates [29,31,34]. In these analyses, the uncertainties in the FEPEs 0.972 ± 0.003 (corresponding to the 242 keV line of 214 Pb and 51 mm
calculated by PENELOPE using the optimal model were assigned so that sample of the IAEA 447) and 0.387 ± 0.029 (corresponding to the 1120.3
the FEPEs computed for the reference sample and the corresponding keV line of 214 Bi and the 10 mm sample of IAEA RGU-1). The resulting
experimental FEPEs (see Section 2.3) were compatible, leading to an RECF was estimated as 0.92 ± 0.06. In addition to these corrections,
uncertainty of 6%. From the 25 measured samples, the sample contain- the measured values were corrected when necessary for the decay
ing 51 mm of IAEA 448 was the one with the highest counting rates, with or equilibrium between precursors and progeny (using the Bateman
a total counting rate (sum of the counting rates across all the channels) equations) so that they refer to the reference dates of the corresponding
of 74.98 cps (counts per second). The full energy peak with the highest material.
counting rate also belonged to this sample, corresponding to a gamma The ratios (Eq. (8)) between the deduced experimental and reference
line of 352 keV from the 214 Pb, with a net counting rate (subtracting activity concentrations are shown in Figs. 3 to 5. As can be seen in
the counting rate for the laboratory background) of 5.40 cps. The lowest Figs. 3 and 4, the ratios corresponding to the reference materials IAEA
total counting rate of 5.22 cps was shown by the sample with 10 mm of RGU-1, RGTh-1, RGK-1 and 448 are, in the great majority of cases (92%
IAEA 447, and this sample also gave the full energy peak with the lowest of the 155 ratios), between 0.9 and 1.1, i.e. they have absolute relative
net counting rate of 0.00104 cps, corresponding to the gamma line of deviations below 10%. Several statistical parameters were estimated in
338 keV from 228 Ac. Due to the low counting rate of the measurements order to analyze the way in which these ratios are distributed. The set
210
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Fig. 3. Ratio (Eq. (8)) between the deduced experimental and reference activity concentrations for the samples of IAEA RGU-1 (circles), RGTh-1 (squares) and RGK-1 (triangles), for
five different sample heights (uncertainties given as 1⋅𝜎).
Fig. 4. Ratio (Eq. (8)) between the deduced experimental and reference activity concentrations for samples of IAEA 448 for five different sample heights (uncertainties given as 1⋅𝜎).
of 155 ratios has an arithmetic mean and median of 1.00, a standard distribution of the data as a whole (a total of 200 ratios including IAEA
deviation of 0.05, a 25th percentile of 0.96, a 75th percentile of 1.04, 447) is relatively well centered around unity, with an arithmetic mean
kurtosis of 2.73 and skewness of 0.22. All of these parameters seem of 1.01, a median of 1.00, a 25th percentile of 0.96 and a 75th percentile
to indicate that the ratios follow a distribution similar to a normal of 1.05. However, the standard deviation is now more than double that
distribution, centered at 1.00 and with an acceptably low dispersion. above, with a value of 0.12; the kurtosis and skewness, of 21.03 and
The Anderson–Darling test was applied, and the result did not reject 2.94, respectively, indicate that the distribution has neither the form
the null hypothesis at a 5% significance level, that the distribution of nor the symmetry of a normal distribution. This can be explained by the
ratios follows a normal distribution, with a 𝑝-value of 0.284. Deviations fact that the radionuclides of the IAEA 447 (with the exception of 210 Pb
of the same order are observed for IAEA 447 (Fig. 5), for the lines at and 137 Cs as mentioned above) have activity concentrations of below 40
46.5 keV from 210 Pb and 661.7 keV from 137 Cs, although the remaining Bq/kg, meaning that the laboratory background is a far more important
measurements are considerably less accurate and precise. Thus, the factor than the efficiency calibration to be validated. Nevertheless, even
211
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Fig. 5. Ratio (Eq. (8)) between the deduced experimental and reference activity concentrations for samples of IAEA 447 for five different sample heights (uncertainties given as 1⋅𝜎.).
Table 5
Values of 𝜁𝑠𝑐𝑜𝑟𝑒 and 𝑃𝑠𝑐𝑜𝑟𝑒 for the results from IAEA RGU-1, RGTh-1 and RGK-1, for five different sample heights.
Mat. Nuclide En (keV) 10 mm 20 mm 30 mm 40 mm 51 mm
𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒
RGU-1 210 Pb 46.5 1.09 6.2 0.00 6.2 0.36 6.1 1.03 6.1 1.17 6.1
RGU-1 234 Th 63.3 0.64 6.5 0.84 6.5 0.84 6.4 0.16 6.4 0.13 6.2
235
RGU-1 U 143.8 1.49 7.2 0.52 6.7 1.43 6.0 0.13 6.1 0.79 6.3
226
RGU-1 Ra 186.2 0.18 6.1 0.09 6.2 0.99 6.2 1.16 6.2 1.09 6.2
214
RGU-1 Pb 242 0.05 6.1 0.61 6.1 0.63 6.1 0.28 6.1 0.89 6.0
214
RGU-1 Pb 295.2 1.15 6.1 1.22 6.0 1.00 6.0 0.42 6.0 0.00 6.1
RGU-1 214 Pb 351.9 0.82 6.1 1.14 6.0 1.00 6.0 0.41 6.0 0.14 6.1
RGU-1 214 Bi 609.3 0.91 10.3 0.71 6.7 0.48 7.7 0.01 7.1 0.06 8.2
RGU-1 234m Pa 1001 1.31 11.5 0.38 7.9 0.49 7.9 0.01 7.2 0.38 6.0
214
RGU-1 Bi 1120.3 0.54 9.9 0.15 6.7 0.42 7.6 0.76 7.0 0.84 8.7
214
RGU-1 Bi 1238.1 0.43 10.2 0.01 7.0 0.29 7.7 0.70 7.2 0.66 8.6
214
RGU-1 Bi 1764.5 0.96 6.2 1.13 6.1 0.52 6.1 0.07 6.1 0.58 6.1
212
RGTh-1 Pb 238.6 1.24 6.6 1.25 6.6 1.23 6.6 1.05 6.6 0.77 6.0
RGTh-1 228 Ac 338.3 0.04 6.7 0.07 6.7 0.11 6.7 0.47 6.7 0.60 7.6
RGTh-1 208 Tl 583.2 1.36 8.2 1.00 8.1 0.36 8.1 0.78 8.1 1.40 8.0
RGTh-1 212 Bi 727.3 0.31 7.1 0.45 6.9 0.58 6.9 0.12 6.9 1.37 7.0
208
RGTh-1 Tl 860.6 0.17 7.8 0.06 7.5 0.44 7.5 0.88 7.5 0.95 7.8
228
RGTh-1 Ac 911.2 1.19 6.7 0.99 6.7 1.21 6.7 1.39 6.7 1.33 7.0
40
RGK-1 K 1460.8 0.54 7.2 0.17 6.9 0.17 6.9 0.16 6.9 0.81 6.0
in the case of IAEA 447, the ratios are generally in good agreement with 4. Conclusions
unity, considering their uncertainties (uncertainties given as 1⋅𝜎 and
obtained from uncertainty propagation using Eq. (8)). A computational model for an HPGe well detector was built using the
The 𝜁𝑠𝑐𝑜𝑟𝑒 and P𝑠𝑐𝑜𝑟𝑒 values corresponding to the results mentioned PENELOPE Monte Carlo code, and by applying a methodology that was
above are given in Tables 5 and 6. As can be observed, all the 𝜁𝑠𝑐𝑜𝑟𝑒 proposed and validated in previous papers. Several experimental FEPEs,
values corresponding to IAEA RGU-1, RGTh-1, RGK-1 and 448 are below determined by measuring a sample of the reference material IAEA RGU-
the limit of 1.64 (the critical value for a normal distribution with a 1, were used as a benchmark.
confidence level of 90%). This means that the reported result does not For the validation of the model, samples of several reference ma-
differ significantly from the expected value [43]. Moreover, most of terials were prepared with different heights, and gamma-ray spectra
the P𝑠𝑐𝑜𝑟𝑒 values corresponding to these four materials are below 10, were acquired, with the corresponding computational FEPEs used for
with a maximum of 11.5, which is below most of the limits used for the analysis of these samples. In most cases, the measured and reference
this test [44–46]. For IAEA 447 (Table 6), most of the 𝜁𝑠𝑐𝑜𝑟𝑒 values values show deviations of below 10% (only 15% of the results show
are below the limit of 1.64 (84% of cases), and none of them exceed larger deviations) and are compatible within uncertainties of 1𝜎.
the limit of 2.58 (the critical value for a normal distribution with a Most of the 𝜁𝑠𝑐𝑜𝑟𝑒 values obtained are below the first critical limit of
confidence level of 99%) commonly adopted in IAEA proficiency tests 1.64, and all are below the critical limit of 2.58, which is the maximum
and inter-comparisons [43–45] including the value for this reference acceptable value generally adopted by the IAEA; thus, all the reported
material [46]. For the P𝑠𝑐𝑜𝑟𝑒 values, 78% of these are below the limit of values are acceptable in terms of accuracy. The P𝑠𝑐𝑜𝑟𝑒 values are below
20 adopted in the abovementioned proficiency test [46]. 10 in most cases, although, a few results of the IAEA 447 measurements
212
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
Table 6
Values of 𝜁𝑠𝑐𝑜𝑟𝑒 and 𝑃𝑠𝑐𝑜𝑟𝑒 for the results from IAEA 447 and 448 for five different sample heights.
Mat. Nuclide En (keV) 10 mm 20 mm 30 mm 40 mm 51 mm
𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒 𝜁𝑠𝑐𝑜𝑟𝑒 𝑃𝑠𝑐𝑜𝑟𝑒
210
448 Pb 46.5 0.40 7.4 0.19 7.4 0.77 7.5 1.19 7.4 0.22 6.7
226
448 Ra 186.2 0.19 6.2 0.06 6.2 0.59 6.2 0.76 6.2 1.26 5.3
212
448 Pb 238.6 1.34 5.3 0.59 7.4 0.38 5.5 1.51 5.3 0.93 5.7
214
448 Pb 242 0.44 9.4 0.29 8.2 0.53 8.4 0.17 8.1 0.05 7.6
448 214 Pb 295.2 0.48 8.6 0.13 8.6 0.43 8.6 0.14 8.6 0.07 8.0
448 228 Ac 338.3 1.34 9.1 0.91 8.5 0.47 9.0 1.02 8.9 1.54 8.1
448 214 Pb 351.9 0.69 8.6 0.06 8.6 0.23 8.6 0.29 8.6 0.27 8.0
214
448 Bi 609.3 0.48 10.0 0.29 10.0 0.25 10.0 0.20 10.0 0.11 9.4
228
448 Ac 911.2 0.82 8.8 0.29 8.3 1.00 8.3 0.97 8.2 0.99 7.3
214
448 Bi 1120.3 0.06 9.8 0.00 9.8 0.53 9.8 0.94 9.8 0.93 9.2
214
448 Bi 1238.1 0.09 10.1 0.65 10.0 0.01 9.8 0.65 9.8 0.62 9.2
448 214 Bi 1764.5 0.62 8.7 0.29 8.6 0.08 8.6 0.37 8.6 0.35 8.0
447 210 Pb 46.5 1.51 9.6 0.85 8.6 0.53 8.3 0.57 8.3 0.02 8.2
447 212 Pb 238.6 1.09 12.4 0.60 10.1 1.54 9.7 0.37 9.1 0.22 8.6
214
447 Pb 295.2 0.46 36.4 0.85 23.6 0.10 17.0 0.51 16.5 2.04 20.0
228
447 Ac 338.3 0.49 33.0 0.34 25.9 1.46 23.6 0.73 15.8 1.05 11.9
214
447 Pb 351.9 0.97 23.8 0.35 14.3 1.21 13.7 1.78 13.4 2.31 14.2
208
447 Tl 583.2 2.13 22.4 0.22 21.8 1.55 14.2 0.16 14.3 0.07 13.6
447 214 Bi 609.3 1.66 23.7 1.33 17.7 2.34 14.1 0.16 15.2 0.22 16.5
447 137 Cs 661.7 0.28 6.5 0.20 6.5 0.41 6.5 0.18 6.5 0.84 6.5
447 228 Ac 911.2 1.36 20.8 0.08 19.3 2.41 10.5 0.71 13.4 0.08 11.6
have values of over 20, the agreed limit for the corresponding profi- [10] N.V. De Castro Faria, R.J.A. Lévesque, Nucl. Instrum. Methods 46 (1967) 325.
ciency test. However, the lack of precision of these results is not due [11] I.O.B. Ewa, D. Bodizs, Sz. Czifrus, Zs. Molnar, Appl. Radiat. Isot. 55 (2001) 103.
[12] V. Peyrés, E. García-Toraño, Nucl. Instrum. Methods Phys. Res. A 580 (2007) 296.
to the imprecision in the FEPEs calculated using the detector model
[13] L.-C. He, L.-J. Diao, B.-H. Sun, L.-H. Zhu, J.-W. Zhao, M. Wang, K. Wang, Nucl.
to be validated; rather, this imprecision appears as a result of both Instrum. Methods Phys. Res. A 880 (2018) 22.
the low activity of the radionuclides present in such material and the [14] D.M. Montalván-Olivares, M.V. Manso-Guevara, F.G. Velasco, Appl. Radiat. Isot.
low mass×efficiency product of these samples (especially at low sample 130 (2017) 34.
[15] J. Nikolic, T. Vidmar, D. Jokovic, M. Rajacic, D. Todorovic, Nucl. Instrum. Methods
heights), which together cause a low counting rate (of the same order
Phys. Res. A 763 (2014) 347.
of that corresponding to the laboratory background), and hence high [16] J.K. Nikolic, M. Rajacic, D. Todorovic, T. Vidmar, J. Environ. Radioact. 165 (2016)
counting uncertainties. 191.
Hence, the optimized computational model of the detector created [17] R. Berndt, P. Mortreau, Nucl. Instrum. Methods Phys. Res. A 694 (2012) 341.
[18] N. Dokania, V. Singh, S. Mathimalar, V. Nanal, S. Pal, R.G. Pillay, Nucl. Instrum.
in this work is suitable for efficiency calibration of the measurement of
Methods Phys. Res. A 745 (2014) 119.
environmental samples using gamma-ray spectrometry (for the energy [19] A. Hedman, J. Bahar Gogani, M. Granström, L. Johansson, J.S. Andersson, H.
range 46.5–1764.5 keV), notwithstanding the relative simplicity of the Ramebäck, Nucl. Instrum. Methods Phys. Res. A 785 (2015) 21.
methodology applied, which, as demonstrated in this work, can be [20] M.T. Haj-Heidari, M.J. Safari, H. Afarideh, H. Rouhi, Radiat. Meas. 88 (2016) 1.
successfully applied to well detectors. Although in this work, the FEPE [21] J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo, A.
Tejera, J.M. Gil, R. Rodríguez, P. Martel, J. Environ. Radioact. 149 (2015) 8.
calculations were performed using the PENELOPE code, equivalent [22] J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo, A.
results would be expected, if other similar codes were used, provided Tejera, P. Martel, J.P. Bolivar, Nucl. Instrum. Methods Phys. Res. A 858 (2017)
that automatic optimization commanded by a main script with the 113.
optimization algorithm was possible. [23] J.G. Guerra, J.G. Rubiano, G. Winter, A.G. Guerra, H. Alonso, M.A. Arnedo, A.
Tejera, F. Mosqueda, P. Martel, J.P. Bolivar, Nucl. Instrum. Methods Phys. Res. A
880 (2018) 67.
Acknowledgments [24] R. Storn, K. Price, J. Global Optim. 11 (1997) 341.
[25] F. Salvat, J.M. Fernández-Varea, J. Sempau, PENELOPE-2011: A Code System for
This work was partially financed by the Nuclear Safety Council of Monte Carlo Simulation of Electronic and Photon Transport, OECD, Barcelona,
Spain, 2011.
Spain (CSN) through grants from its R&D programs for 2009 and 2012.
[26] Genie 2000 Basic Spectroscopy Software Version 2.1, Canberra Industries Inc.,
J.G. Guerra acknowledges a Ph.D. fellowship through the ‘‘Programa 2002.
de personal investigador pre-doctoral en formación’’ of the ULPGC. [27] RADON SCOUT, SARAD GmbH, Wiesbadener, https://www.sarad.de/cms/media/
docs/datenblatt/Datasheet_Radon-Scout_2_EN_09-03-17.pdf.
[28] H. Alonso, El Radón En Suelos, Rocas, Materiales de Construcción Y Aguas
References
SubterráNeas de Las Islas Canarias Orientales (Ph.D. dissertation), Universidad de
Las Palmas de Gran Canaria, Spain, 2016, http://hdl.handle.net/10553/18553.
[1] G. Gilmore, J.D. Hemingway, Practical Gamma-Ray Spectrometry, Wiley, Chich- [29] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials
ester, UK, 1995. RGU-1, RGTh-1 and RGK-1, IAEA Report RL148, International Atomic Energy
[2] J.H. Lacey, M.J. Leng, C.H. Vane, A.D. Radbourne, H. Yang, D.B. Ryves, Anthro- Agency, Vienna, 1987.
pocene 21 (2018) 52. [30] A. Ceccatelli, R. Katona, G. Kis-Benedek, A. Pitois, Appl. Radiat. Isot. 87 (2014)
[3] C.Y. Chuang, P.H. Santschi, L.S.Wen, L. Guo, C. Xu, S. Zhang, Y. Jiang, Y.F. Ho, 461.
K.A. Schwehr, A. Quigg, C.C. Hung, M. Ayranov, D. Schumann, Mar. Chem. 173 [31] IAEA-448 – Radium-226 in soil from oil field, http://nucleus.iaea.org/rpst/
(2015) 320. ReferenceProducts/ReferenceMaterials/Radionuclides/RS_IAEA-448.pdf.
[4] A.-R. Iurian, C. Cosma, Nucl. Instrum. Methods Phys. Res. A 763 (2014) 132. [32] A.R. Iurian, A. Pitois, G. Kis-Benedek, A. Migliori, R. Padilla-Alvarez, A. Ceccatelli,
[5] J.P. Pérez-Moreno, E.G. San Miguel, J.P. Bolívar, J.L. Aguado, Nucl. Instrum. Appl. Radiat. Isot. 109 (2016) 61.
Methods Phys. Res. A 491 (2002) 152. [33] A. Shakhashiro, S. Tarjan, A. Ceccatelli, G. Kis-Benedek, M. Betti, Appl. Radiat. Isot.
[6] M.J. Daza, B. Quintana, M. García-Talavera, F. Fernández, Nucl. Instrum. Methods 70 (2012) 1632.
Phys. Res. A 470 (2001) 520. [34] IAEA-447- Natural and artificial radionuclides in moss-soil, http://nucleus.iaea.
[7] U. Rizwan, A. Chester, T. Domingo, K. Starosta, J. Williams, P. Voss, Nucl. Instrum. org/rpst/Documents/RS_IAEA-447.final.pdf.
Methods Phys. Res. A 802 (2015) 102. [35] Chart of the Nuclides, National Nuclear Data Center, Brookhaven National Labora-
[8] K. Shizuma, Y. Oba, M. Takada, Nucl. Instrum. Methods Phys. Res. B 383 (2016) tory, http://www.nndc.bnl.gov/chart/.
183. [36] I. López-Coto, J.L. Mas, J.P. Bolivar, R. García-Tenorio, Appl. Radiat. Isot. 67 (2009)
[9] K.M. Wainio, G.F. Knoll, Nucl. Instrum. Methods 44 (1966) 213. 133.
213
J. G.Guerra et al. Nuclear Inst. and Methods in Physics Research, A 908 (2018) 206–214
[37] G. Xhixha, M. Alberi, M. Baldoncini, K. Bode, E. Bylyku, F. Cfarku, I. Callegari, [43] Report on the Second Almera Network Coordination Meeting and the Almera Soil
F. Hasani, S. Landsberger, F. Mantovani, E. Rodriguez, F. Shala, V. Strati, M. Sampling Intercomparison Exercise IAEA/SIE/01. IAEA / AL / 164, International
Xhixha Kaçeli, J. Radioanal. Nucl. Chem. 307 (2016) 1507. Atomic Energy Agency, Seibersdorf, 2006.
[38] E. Tomarchio, S. Rizzo, Radiat. Phys. Chem. 80 (2011) 318. [44] Worldwide open proficiency test: determination of naturally occurring radionu-
[39] M. Belli, P. de Zorzi, U. Sansone, A. Shakhashiro, A. Gondin da Fonseca, A. Trinkl, clides in phosphogypsum and water IAEA-CU-2008-03. IAEA Analytical Quality in
T. Benesch, J. Environ. Radioact. 100 (2009) 982. Nuclear Applications Series No. IAEA/AQ/18, International Atomic Energy Agency,
[40] Worldwide Interlaboratory Comparison on the Determination of Trace Elements Vienna, 2010.
in the IAEA-457 Marine Sediment Sample. IAEA Analytical Quality in Nuclear [45] Worldwide Open Proficiency Test on the Determination of Radionuclides in
Applications Series No. IAEA/AQ/46, International Atomic Energy Agency, Vienna, Spinach, Soil and Water IAEA-CU-2007-03. IAEA Analytical Quality in Nuclear
2016. Applications Series No. IAEA/AQ/8, International Atomic Energy Agency, Vienna,
[41] ALMERA Proficiency Test: Determination of Natural and Artificial Radionuclides in 2007.
Soil and Water IAEA-TEL-2011-04. IAEA Analytical Quality in Nuclear Applications [46] Worldwide Open Proficiency Test: Determination of Natural and Artificial Ra-
Series No. IAEA/AQ/32, International Atomic Energy Agency, Vienna, 2013. dionuclides in Moss-Soil and Water IAEA-CU-2009-03. IAEA Analytical Quality in
[42] A. Shakhashiro, L. Mabit, Appl. Radiat. Isot. 67 (2009) 139. Nuclear Applications Series No. IAEA/AQ/22, International Atomic Energy Agency,
Vienna, 2012.
214
Capítulo 4 97
Capítulo 6
6.1. Conclusiones
Las principales conclusiones del trabajo realizado en esta Tesis Doctoral se pueden
resumir en los siguientes apartados:
Como en cualquier trabajo de tesis doctoral, en este trabajo quedan varias líneas de
investigación abiertas, algunas de ellas en un estado de desarrollo más avanzado que otras:
Referencias
AGOSTINELLI, S., ET AL. 2003. Geant4—a simulation toolkit. Nuclear Instruments
and Methods in Physics Research Section A: Accelerators, Spectrometers,
Detectors and Associated Equipment, 506 (3), 250-303.
ALONSO, H. 2016. El radón en suelos, rocas, materiales de construcción y aguas
subterráneas de las Islas Canarias Orientales. Tesis Doctoral, ULPGC
<http://hdl.handle.net/10553/18553>.
BÄCK, T. 1996. Evolutionary Algorithms in Theory and Practice. Oxford University
Press, New York, USA.
BADAWI, M. S., GOUDA, M. M., NAFEE, S. S., EL-KHATIB, A. M. & EL-MALLAH, E.
A. 2012. New algorithm for studying the effect of self attenuation factor on the
efficiency of γ-rays detectors. Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, 696, 164-170.
BANZHAF, W., NORDIN, P., KELLER, R. E. & FRANCONE, F. D. 1998. Genetic
Programming: An Introduction: On the Automatic Evolution of Conputer Programs
and Its Applications. Norgan Kaufmann Publishcrs, Inc., San Francisco,
California.
BERGEY, P.K. An agent enhanced intelligent spreadsheet solver for multi-criteria
decisión making. In Proceedings of the Fifth Americas Conference on Information
Systems (AMCIS 1999) (Milwaukee, Wisconsin, USA, Aug 1999), W.D. Haseman
and D.L. Nazareth, Eds., Association for Information Systems, Atlanta, GA, USA,
pp. 966-968.
BEYER, H.G. 2001. The Theory of Evotution Strategies. Springer-Verlag, Berlín,
Heidelberg, Germany.
BOSON, J., AGREN, G. & JOHASSON, L. 2008. A detailed investigation of HPGe
detector response for improved Monte Carlo efficiency calculations. Nuclear
Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 587, 304-314.
BOTH, J.P., LEE, Y.K., MAZZOLO, A., PETIT, O., PENELIAU, Y., ROESSLINGER, B.
& SOLDEVILA, M. 2003. TRIPOLI-4—A Three Dimensional Polykinetic Particle
Transport Monte Carlo Code. SNA’2003, Paris.
BREIMEISTER, J. 1997. MCNP-A General Monte Carlo N-Particle Transport Code,
Version 4B. LANL Rep., LA-12.
BRONSON, F. 2003. Validation of the accuracy of the LabSOCS software for
mathematical efficiency calibration of Ge detectors for typical laboratory samples.
Journal of Radioanalytical and Nuclear Chemistry, 255 (1) 137-1441.
BUDJAS, D., HEISEL, M., MANESCHG, W. & SIMGEN, H. 2009. Optimization of the
MC-model of a p-type Ge-spectrometer for the purpose of efficiency
determination. Applied Radiation and Isotopes, 67, 706-710.
104 Referencias
CANBERRA 2002a. Model S573/S574 ISOCS/LabSOCS Validation & Verification
Manual V.4.0, Canberra Industries Inc.
CANBERRA 2002b. Genie 2000 Basic Spectroscopy Software Version 2.1. Canberra
Industries Inc.
CECCATELLI, A., KATONA, R., KIS-BENEDEK, G. & PITOIS, A. 2014. Measurement
of 226Ra in soil from oil field: Advantages of γ-ray spectrometry and application
to the IAEA-448 CRM. Applied Radiation and Isotopes, 87, 461- 467.
CLERC , M. 2006. Parlicle Swarm Optimization. ISTE.
COMPTON, A. H. 1923. A Quantum Theory of the Scattering of X-rays by Light
Elements. Physical Review, 21, 483-502.
COURTINE, F., PILLEYRE, T., SANZELLE, S. & MIALLIER, D. 2008. Ge well detector
calibration by means of a trial and error procedure using the dead layers as a
unique parameter in a Monte Carlo simulation. Nuclear Instruments and Methods
in Physics Research Section A: Accelerators, Spectrometers, Detectors and
Associated Equipment, 596, 229-234.
DE CASTRO, N.V. & LÉVESQUE, R.J.A., 1967, Photopeak and double-escape peak
efficiencies of germanium lithium drift detectors. Nuclear Instruments and
Methods, 46, 325-332.
DEBERTIN, K. & HELMER, R.G. 1988. Gamma and X-Ray Spectrometry with
Semiconductor Detectors, ISBN 0444871071, North Holland, Amsterdam.
DEGRELLE, D., MAVON, C. & GROETZ, J.-E. 2016. Determination of mass
attenuation coefficient by numerical absorption calibration with Monte-Carlo
simulations at 59.54keV. Nuclear Instruments and Methods in Physics Research
Section A: Accelerators, Spectrometers, Detectors and Associated Equipment,
816, 47- 52.
DH-la 1981. DH-la: A Certified Uranium-Thorium Reference Ore, Canada Centre for
Mineral and Energy Technology, CANMET Report 81-11E.
DORIGO, M & STÜTZLE, T. 2004. Ant Colony Optimization. The MIT Press.
GILMORE, G. & HEMINGWAY, J. D. 1995. Practical gamma-ray spectrometry, Wiley
Chichester.
GOLDBERG, D. E. 1989. Genetic Algorithms in Seareh, Optirnization & Machine
Learning. Addison-Wesley.
HARDY, J.C., JACOB, V.E., SANCHEZ-VEGA, M., EFFINGER, R.T., LIPNIK, P.,
MAYES, V.E., WILLIS D.K. & HELMER R.G. 2002. Precise efficiency calibration
of an HPGe detector: source measurements and Monte Carlo calculations with
subpercent precision. Applied Radiation and Isotopes, 56, 65-69.
HUY, N.Q., BINH, D.Q. & AN, V.X. 2007. Study on the increase of inactive germanium
layer in a high-purity germanium detector after a long time operation applying
MCNP code. Nuclear Instruments and Methods in Physics Research Section A:
Accelerators, Spectrometers, Detectors and Associated Equipment, 573, 384-
388.
IAEA 1987. Preparation and Certification of IAEA Gamma Spectrometry Reference
Materials, RGU-1, RGT-1 and RGK- 1. International Atomic Energy Agency.
Report-IAEA/RL/148.
IAEA 2001. Report on the Intercomparison Run for the Determination of Radionuclides
in Soils IAEA-326 and IAEA-327. International Atomic Energy Agency, Report-
IAEA/AL/100.
Referencias 105
IURIAN, A.R. & COSMA, C. 2014. A practical experimental approach for the
determination of gamma-emitting radionuclides in environmental samples.
Nuclear Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 763, 132-136.
KOVÁCS, T., SZEILER, G., FÁBIÁN, F., KARDOS, R., GREGORIČ, A. & VAUPOTIČ,
J. 2013. Systematic survey of natural radioactivity of soil in Slovenia. J Environ
Radioact, 122, 70-78.
LIYE, L., JIZENG, M., FRANCK., D., DE CARLAN, L. & BINQUAN, Z. 2006. Monte
Carlo efficiency transfer method for full energy peak efficiency calibration of three
type HPGe detectors: A coaxial N-type, a coaxial P-type and four BEGe
detectors. Nuclear Instruments and Methods in Physics Research Section A:
Accelerators, Spectrometers, Detectors and Associated Equipment, 564, 608-
613.
MALEKA, P.P. & MAUCEC, M. 2005. Monte Carlo uncertainty analysis of germanium
detector response to gamma rays with energies bellows 1 MeV. Nuclear
Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 538, 631-639.
MARISCOTTI, M. A. 1967. A method for automatic identification of peaks in the
presence of background and its application to spectrum analysis. Nuclear
Instruments and Methods, 50, 309-320.
MCKINNEY, W.G. & ET AL. 2006. MCNPX overview. Proceedings of the 2006 HSSW,
FNAL, IL, LA-UR-06-6206, September 2006.
MOHAMMAD MODARRESI, S., FARHAD MASOUDI, S. & KARIMI, M. 2017. A
method for considering the spatial variations of dead layer thickness in HPGe
detectors to improve the FEPE calculation of bulky samples. Radiation Physics
and Chemistry, 130 291-296
NAFEE, S.S. & ABBAS, M.I. 2008. Calibration of closed-end HPGe detectors using bar
(Parallelepiped) sources. Nuclear Instruments and Methods in Physics Research
Section A: Accelerators, Spectrometers, Detectors and Associated Equipment,
592, 80-87.
NELSON, W.R., HIRAYAMA, H., ROGERS & D.W.O. 1985. EGS4 User Manual.
SLAC-265 Appendix 2. Stanford Linear Accelerator Center, Stanford.
ORTIZ-RAMÍREZ, P.C. 2015. Development of an absolute method for efficiency
calibration of a coaxial HPGe detector for large volume sources. Nuclear
Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 793. 49-56.
PÉREZ-MORENO, J., SAN MIGUEL, E., BOLÍVAR, J. & AGUADO, J. 2002. A
comprehensive calibration method of Ge detectors for low-level gamma-
spectrometry measurements. Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, 491, 152-162.
ROBIČ, T. & FILIPIČ, B. 2005. DEMO: Differential evolution for multi-objective
optimization, Evolutionary Multi-Criterion Optimization – EMO 2005, Vol. 3410 of
Lecture Notes in Computer Science, Springer, Berlin 520.
RODENAS, J., PASCUAL, A., ZARZA, I., SERRADELL, V., ORTIZ, J. &
BALLESTEROS, L. 2003. Analysis of the influence of germanium dead layer on
detector calibration simulation for environmental radioactive samples using the
Monte Carlo method. Nuclear Instruments and Methods in Physics Research
106 Referencias
Section A: Accelerators, Spectrometers, Detectors and Associated Equipment,
496, 390-399.
S. CHAGREN, M.BEN TEKAYA, N. REGUIGUI & F. GHARBI. 2016. Efficiency transfer
using the GEANT4 code of CERN for HPGe gamma spectrometry. Applied
Radiation and Isotopes,107, 359-365.
SALVAT, F., FERNÁNDEZ-VÁREA, J.M. & SEMPAU, J. 2011. PENELOPE-2011: A
code system for Monte Carlo simulation of electronic and photon transport.
OECD.
SHAKHASHIRO, A., SANSONE, U., WERSHOFEN, H., BOLLHÖFER, A., KIM, C.K.,
KIM, C.S., KIS-BENEDEK G., KORUN, M., MOUNE, M., LEE, S.H., TARJAN, S.
& AL-MASRI, M.S. 2011. The new IAEA reference material: IAEA-434
technologically enhanced naturally occurring radioactive materials (TENORM) in
phosphogypsum. Applied Radiation and Isotopes, 69, 231-236.
SHAKHASHIRO, A., TARJAN, S., CECCATELLI, A., KIS-BENEDEK, G. & BETTI, M.
2012. IAEA-447: A new certified reference material for environmental radioactivity
measurements. Applied Radiation and Isotopes, 70, 1632-1643.
SIMA, O., ARNOLD, D. & DOVLETE, C. 2001. GESPECOR—a versatile tool in
gamma-ray spectrometry. Journal of Radioanalytical and Nuclear chemistry, 248
(2), 359-364.
STORN, R. & PRICE, K. 1997. Differential evolution – a simple and efficient heuristic
for global optimization over continuous spaces. Journal of Global Optimization,
11, 341-359.
SZENTMIKLÓSI, L., BELGYA, T., MARÓTI, B. & KIS, Z. 2014. Characterization of
HPGe gamma spectrometers by geant4 Monte Carlo simulations. Journal of
Radioanalytical and Nuclear chemistry, 300, 553-558.
TAYYEBI, P., DAVANI, F. A., TABASI, M. & AFARIDEH, H. 2017. A novel technique
for detection efficiency determination of HPGe. Radiation Physics and Chemistry,
133, 86-90.
TEDJANI, A., MAVON, C., BELAFRITES, A., DEGRELLE, D., BOUMALA, D.,
RIUS, D. & GROETZ, J.-E. 2016. Well GeHP detector calibration for
environmental measurements using reference materials. Nuclear Instruments
and Methods in Physics Research Section A: Accelerators, Spectrometers,
Detectors and Associated Equipment, 838, 12-17.
TOMARCHIO, E. & RIZZO, S. 2011. Coincidence-summing correction equations in
gamma-ray spectrometry with p-type HPGe detectors. Radiation Physics and
Chemistry, 80, 318-323.
VIDAL-CUADRAS, A. & ORTEGA, M. R. 1986. Radiactividad e interacción de la
radiación con la materia. Bellaterra.
VIDMAR T. 2005. EFFTRAN—A Monte Carlo efficiency transfer code for gamma-ray
spectrometry. Nuclear Instruments and Methods in Physics Research Section A:
Accelerators, Spectrometers, Detectors and Associated Equipment, 550, 603-608
WAINIO, K.M. & KNOLL, G.F. 1966. Calculated gamma ray response characteristics of
semiconductor detectors. Nuclear Instruments and Methods, 44, 213-223.
XHIXHA, G., ALBERI, M., BALDONCINI, M., BODE, K., BYLYKU, E., CFARKU, F.,
CALLEGARI, I., HASANI, F., LANDSBERGER, S., MANTOVANI, F.,
RODRIGUEZ, E., SHALA, F., STRATI, V. & XHIXHA- KAÇELI, M. 2016.
Calibration of HPGe detectors using certified reference materials of natural origin.
Journal of Radioanalytical and Nuclear chemistry, 307, 1507-1517.
Apéndice 1
Hojas de características de los detectores empleados en el trabajo de
tesis doctoral
Figura A1.1 Características técnicas del detector HPGe XtRa del grupo GIRMA de la
Universidad de Las Palmas de Gran Canaria.
Apéndice 1 – Hojas de características de los detectores 108
Figura A1.2 Características técnicas del detector HPGe XtRa del grupo FRyMA de la
Universidad de Huelva.
109 Apéndice 1 – Hojas de características de los detectores
Figura A1.3 Características técnicas del detector HPGe tipo pozo del grupo GIRMA de la
Universidad de Las Palmas de Gran Canaria.
Apéndice 2
Materiales de referencia empleados en el trabajo de tesis doctoral
Las Tablas A2.1 y A2.2 muestran la composición química considerada para los
materiales de referencia empleados en este trabajo de tesis doctoral, así como la
concentración de actividad (con fechas de referencia) de los radionucleidos principales
contenidos en dichos materiales.
El material de referencia IAEA RGU-1 (IAEA, 1987) fue preparado, a petición del
Organismo Internacional de Energía Atómica (IAEA), por el Centro para la Tecnología Mineral y
Energética de Canada (CANMET) por medio de dilución homogénea del material de referencia
certificado BL-5 (Mineral de uranio con 7.09% U) con polvo de sílice. El CRM BL-5 está
226 210
certificado para uranio, Ra y Pb, confirmándose que los radionucleidos de la serie del
uranio se encuentran en equilibrio radioactivo. El material de referencia IAEA RGTh-1 (IAEA,
1987) también fue preparado, a petición la IAEA por el CANMET por medio de dilución
homogénea en polvo de sílice del CRM OKA-2 (Mineral de torio con 2.89% Th). El material
IAEA RGK-1 (IAEA, 1987) es Sulfato Potásico de alta pureza (99.8%).
Los CRMs IAEA 447 (Shakhashiro et al., 2012), IAEA 448 (Ceccatelli et al., 2014),
IAEA 434 (Shakhashiro et al., 2011) y IAEA 326 (IAEA, 2001) son muestras ambientales (suelo
de musgo, suelo contaminado de un campo petrolífero, fosfoyeso y muestra de suelo
respectivamente) con actividades certificadas para varios radionucleidos de interés para
laboratorios de radiactividad ambiental. En lo que respecta al CANMET DH-1a (DH-la, 1981),
es un mineral de uranio y torio con una considerable concentración de ambos elementos
(0.2629 % U and 0.091% Th).
112 Apéndice 2
Tabla A2.1 Composición centesimal aproximada, considerada para la simulación por el
método de Monte Carlo, de los materiales de referencia empleados.
IAEA IAEA IAEA IAEA 448 IAEA 447 IAEA 434 IAEA 326 CANMET
RGU-1 RGTh-1 RGK-1 DH-1a
O: 53.38 O: 52.74 K: 44.87 O: 45.57 C: 47.40 O: 40.99 O: 39.77 O: 47.31
Si: 46.40 Si: 45.00 O: 36.72 Si: 14.78 O: 18.37 S: 20.17 Si: 27.46 Si: 37.28
Al: 0.10 Re: 0.53 S: 18.40 C: 12.05 Ca:12.61 Ca: 26.74 C: 19.45 Fe: 5.17
U: 0.04 Ca: 0.42 Ca: 10.49 Si: 10.88 C: 9.57 Al: 5.18 S: 4.82
Ca: 0.03 Ce: 0.33 Fe: 6.80 Al: 4.49 P: 1.43 Fe: 3.05 Al: 3.44
Fe: 0.03 Y: 0.25 Al: 4.16 Fe: 3.04 Si: 1.04 K: 1.73 K: 1.43
Na: 0.02 Nd: 0.16 Mg: 2.61 K: 1.71 K: 0.06 Ca: 1.11 Mg: 0.07
Fe: 0.13 S: 1.98 Mg: 0.86 Na: 1.05 Ca: 0.04
La: 0.13 K: 0.83 Ti: 0.33 Mg: 0.80 Na: 0.04
P: 0.13 Na: 0.73 Na: 0.31 Ti: 0.40 C: 0.05
Th: 0.08 U: 0.26
Pr: 0.05 Th: 0.09
Sm: 0.03
K: 0.02
Apéndice 3
Calibración en eficiencia mediante el software LabSOCS de Canberra
Industries Inc.
Dicho modelo se valida comparando las FEPEs calculadas mediante MCNP con las
correspondientes FEPEs experimentales para varias disposiciones fuente–detector (fuentes
241 152 137
puntuales de Am, Eu y Cs situadas a diferentes distancias y con distintos ángulos
respecto al detector) y en un amplio rango de energías. Si los resultados de la validación no
son aceptables, se realizan cambios en las características consideradas en el modelo y se
recalculan las FEPEs, repitiéndose este proceso hasta alcanzar desviaciones válidas.
Una vez validado el modelo de detector construido con MCNP, éste se emplea para
calcular un amplio conjunto de FEPEs correspondientes a geometrías puntuales en vacío (sin
considerar materia alguna entre la fuente puntual y el detector) dispuestas en una cuadrícula
en torno al detector en un rango de 0-500 metros (con una densidad de puntos notablemente
superior en la región cercana al detector), y para 20 energías entre 10 keV y 7 MeV.
Figura A3.2 Ejemplo del empleo del Geometry Composer para definir la geometría de
medida y materiales.
115 Apéndice 3–Calibración mediante LabSOCS
Figura A3.3 Definición de las energías de interés, incertidumbres asociadas a las FEPEs y
parámetros de convergencia.
Apéndice 4
Uso del código PENELOPE
PENELOPE está formado por una serie de subrutinas escritas en FORTRAN, donde se
simula el proceso de la interacción de la radiación con la materia, necesitando de un programa
principal (main) que controle en todo momento el proceso y la evolución de cada suceso con la
definición de las geometrías, magnitudes y cantidades de interés. Dicho programa principal
Apéndice 4–Uso del codigo PENELOPE 118
puede ser escrito por el usuario de acuerdo a sus necesidades. Sin embargo, el paquete
PENELOPE incluye algunos ejemplos de programas principales, que son:
El programa principal penslab.f para el transporte de partículas en una losa.
El programa principal pencyl.f para transporte de partículas en estructuras
cilíndricas.
El programa principal genérico penmain.f para el transporte de partículas en
geometrías regulares
La Figura A4.1 muestra una estructura de archivos necesaria para una simulación
usando el código PENELOPE.
El cuadro superior de la figura muestra los archivos necesarios para la construcción del
archivo ejecutable (Main.exe) que inicia y controla la simulación. Estos son, además del main.f:
TIMER.f: paquete de subrutinas de tiempo, basado en procedimientos de
Fortran 95-97 o 90.
PENELOPE.f: Contiene las subrutinas de simulación del transporte acoplado
de fotones y electrones en un medio sin fronteras.
PENGEOM.f: es un conjunto de subrutinas para el seguimiento de partículas a
través de geometrías regulares modulares.
Dada la geometría de revolución de los detectores HPGe, así como de los recipientes
comúnmente empleados para las medidas de muestras en laboratorio, en este trabajo se ha
empleado como main la subrutina pencyl.f, creada para la simulación del transporte de fotones,
electrones y positrones en geometrías cilíndricas. Tal como se explicó anteriormente, mediante
la compilación de esta subrutina principal junto con las subrutinas PENELOPE.f, PENGEOM.f y
TIMER.f se genera el ejecutable pencyl.exe, que ha sido empleado para los cálculos de
eficiencias en este trabajo de tesis doctoral.
A continuación, se definen los materiales. Los ficheros material.mat a los que llama el
fichero de entrada (ver ejemplo en Figura A4.2), deben estar en el mismo directorio que el
pencyl.exe, y deberán haber sido creados mediante el programa material.f, incluido en el
paquete PENELOPE, en el que se generan los materiales introduciendo su composición
Apéndice 4–Uso del codigo PENELOPE 124
centesimal y a partir de una base de datos denominada pendbase. En esta base de datos se
encuentra la información necesaria de los todos los elementos de la tabla periódica y algunos
compuestos como agua, plásticos, etc. El orden en el que se presentan los materiales coincide
con la numeración de la etiqueta del material situada a continuación de la palabra CYLIND, en
la definición de la geometría explicada anteriormente.
Otra parte del fichero de entrada de gran importancia es la definición de los detectores
considerados (podrían ser varios, aunque en el ejemplo mostrado solo hay uno). En primer
lugar, se define el rango de energías de estudio y en número de canales (máximo 1000) en los
que se va a distribuir la respuesta del detector (línea ENDETC en Figura A4.2). Posteriormente
se definen los anillos cilíndricos (definidos al principio del fichero) que pertenecen al volumen
activo del detector. Esto se realiza mediante el comando EDBODY seguido de las etiquetas
correspondientes al LAYER y al CYLIND.
Para poder calcular la FEPE a partir del espectro obtenido de la simulación, tal como si
se tratara de un espectro obtenido experimentalmente, es necesario determinar la densidad de
probabilidad que pertenece al pico de absorción completa, restándole a la densidad de
probabilidad total la debida al continuo Compton simulado (especialmente significativo a bajas
energías como puede observarse en la Figura A4.4). Dado que, en los resultados de la
simulación, cada pico está abarcado por un único canal, el cálculo de la densidad de
probabilidad neta es mucho más sencillo que el problema experimental análogo, ya que el
fondo Compton en el canal correspondiente se puede estimar como el promedio entre las
densidades correspondientes a los canales contiguos. Es decir, si es la densidad sin
corregir en un canal i en el que se encuentra el pico de interés, la densidad neta se
obtiene mediante la sencilla operación:
[A4.1]
Apéndice 5
Análisis de espectros empleando GENIE 2000
Una vez localizados los picos, GENIE-2000 comprueba si cada pico localizado es
aceptable o no, de acuerdo con unos parámetros umbrales de discriminación definidos por el
usuario. A continuación, se determina la energía correspondiente al centroide del pico a partir
del canal donde se encuentra el mismo (obtenido en el paso anterior de localización de picos) y
aplicando la curva de calibración en energía. De la misma forma, se emplea una curva de
calibración en resolución (FWHM en función de la energía) para establecer el ancho esperado
del pico, lo que va a permitir el posterior cálculo del área neta del pico de absorción completa.
Apéndice 5–Análisis de espectros empleando GENIE 2000 128
Previamente al cálculo del área de los picos, el sistema decide si tratarlos como picos
simples y calcular su área de forma independiente, o como un conjunto de múltiples picos,
calculando su área en conjunto y decidiendo que proporción pertenece a cada pico
posteriormente. Esta discriminación se realiza a partir del FWHM correspondiente a cada pico y
la distancia entre los centroides de los picos consecutivos, seleccionándose posteriormente los
canales que comprenden el o los picos, es decir, la ROI.
El cálculo del área de los picos se realiza principalmente de dos formas, la primera, la
simple suma de las cuentas registradas en los canales correspondientes al pico, y la segunda,
mediante el ajuste de los puntos (canal, cuentas) a una función tipo gaussiana para la posterior
integración de dicha función. En ambos casos, al área total calculada se le restan las cuentas
que se estiman pertenecen al continuo para obtener el área del pico. GENIE 2000 permite
elegir entre las dos opciones en caso de picos simples (en este trabajo de tesis se ha optado
por la simple suma de cuentas), mientras que en el caso de múltiples picos muy cercanos
emplea el método de ajuste a una combinación de funciones tipo gaussiana con distintos
centroides, anchos y alturas, lo que permite posteriormente estimar a el área correspondiente a
cada pico mediante deconvolución.
Figura A5.2 Parámetros empleados en la estimación del continuo (lineal a la izquierda y tipo
escalón a la derecha) bajo el pico (Canberra, 2002b).
En el caso de función lineal, el área B bajo el continuo se calcula mediante la ecuación
[A5.1] y en el caso de función tipo escalón, mediante la ecuación [A5.2].
Apéndice 5–Análisis de espectros empleando GENIE 2000 130
[A5.1]
∑ ∑ [A5.2]
Dónde:
Una vez obtenida el área de cada pico , como último paso, se corrige el área
del pico restándole las cuentas del pico de absorción completa en el ROI que se estiman son
debidas al fondo ambiental. Dichas cuentas de fondo ambiental se obtienen analizando un
espectro de fondo ambiental obtenido previamente (empleando una muestra de material no
radioactivo, tal como agua destilada, idealmente con la misma geometría que las muestras a
medir), por el procedimiento análogo al realizado con las muestras a medir, es decir,
calculando las cuentas por encima del continuo en el mismo ROI. Dado que el fondo
ambiental se mide con un tiempo de adquisición diferente al de la muestra a medir , las
cuentas netas C del pico de absorción completa, empleadas tanto para la determinación de la
FEPE experimental (para la energía correspondiente) como de la actividad del radionucleido
emisor, vienen dadas por la ecuación [A5.3]. La incertidumbre correspondiente se obtiene
mediante propagación de incertidumbres, variando las expresiones a aplicar según el tipo de
método empleado para la suma de las áreas en el ROI (suma simple de cuentas en canales o
integral de funciones ajustadas) y del tipo de continuo considerado (lineal o escalón).
[A5.3]
131
Apéndice 6
Datos y resultados experimentales y computacionales correspondientes a
la caracterización de los detectores HPGe XtRa a partir de datos
experimentales (Capítulo 4)
A6.1 Cálculo de los factores de corrección por suma por coincidencia
En la Tabla A6.1 se muestran las expresiones propuestas por Xhixha et al. (2016) y
Tomarchio and Rizzo (2011) que han sido empleadas en este trabajo para el cálculo de los
factores de corrección por suma por coincidencia.
Tabla A6.1 Expresiones empleadas en este trabajo para el cálculo de las CSCF,
propuestas por Xhixha et al. (2016) y Tomarchio and Rizzo (2011).
Energía
Radionucleido Expresión para cálculo de CSCF
(keV)
214
Pb 242.0 1 0.072 53.2
208
Tl 860.0 1 0.9975 2614.5 0.0136 233.4 1 0.5216 277.4 583.2 / 860.0
En la Figura A6.1 se muestran los CSCF calculados en este trabajo para diferentes
geometrías de medida y los materiales de referencia RGU-1 y RGTh-1. Dichos factores de
corrección han sido calculados empleando las ecuaciones de la Tabla A5.1, a partir de las
FEPEs y eficiencias absolutas totales calculadas mediante PENELOPE (ver apartado A4.2).
Apéndice 6–Datos y resultados del Capitulo 4 132
Figura A6.1 Factores de corrección por suma por coincidencia (CSCF) para DET1 (a) y
DET2 (b) para las 9 emisiones afectadas significativamente por dicho efecto (242.0, 609.3,
1120.3 and 1238.1 keV en RGU-1 and 338.3, 583.2, 727.3, 860.0 and 911.0 keV en RGTh-1).
[A6.1]
∙ ∙ ∙ ∙
133 Apéndice 6–Datos y resultados del Capitulo 4
[A6.2]
Dónde:
*Chart of the Nuclides, National Nuclear Data Center, Brookhaven National Laboratory,
<http://www.nndc.bnl.gov/chart/>.
Tabla A6.9 Recipiente tipo Marinelli B2 con 68.2 mm de altura de material, situado
a una distancia de 1 mm sobre ventana del detector DET1.
Energía FEPE
Área Inc. m(kg) Inc. t(s) CSCF Inc. Inc.
(keV) (%)
46.5 3871 170 0.4490 0.0001 9988 1 0 0.411 0.019
63.3 15595 237 0.4490 0.0001 9988 1 0 1.89 0.05
143.8 8141 210 0.4490 0.0001 9988 1 0 5.60 0.17
186.0 67872 339 0.4490 0.0001 9988 1 0 4.95 0.06
242.0 66856 301 0.4490 0.0001 9988 0.99901 0.00010 4.17 0.03
295.2 147511 413 0.4490 0.0001 9988 1.00030 0.00003 3.614 0.025
352.0 252222 530 0.4490 0.0001 9988 0.99916 0.00008 3.201 0.022
609.3 181578 442 0.4490 0.0001 9988 0.918 0.004 1.963 0.017
1001.0 2709 84 0.4490 0.0001 9988 1 0 1.44 0.05
1120.3 37739 217 0.4490 0.0001 9988 0.910 0.009 1.254 0.016
1238.1 13633 131 0.4490 0.0001 9988 0.915 0.009 1.153 0.017
1764.5 30499 184 0.4490 0.0001 9988 1.0016 0.0003 0.898 0.008
238.6 1808230 1379 0.4356 0.0001 69405 1 0 4.22 0.13
338.3 352972 634 0.4356 0.0001 69405 0.9843 0.0005 3.24 0.15
583.2 558381 817 0.4356 0.0001 69405 0.900 0.007 2.07 0.06
727.3 122801 451 0.4356 0.0001 69405 0.9740 0.0015 1.76 0.06
860.0 65871 319 0.4356 0.0001 69405 0.960 0.006 1.55 0.05
911.0 379157 643 0.4356 0.0001 69405 0.9841 0.0010 1.52 0.05
1460.8 67954 265 0.4946 0.0001 8763 1 0 1.05 0.04
Tabla A6.11 Recipiente tipo Marinelli B4 con 74.8 mm de altura de material, situado
directamente sobre ventana del detector DET1.
Energía FEPE
Área Inc. m(kg) Inc. t(s) CSCF Inc. Inc.
(keV) (%)
46.5 31252 644 0.5071 0.0001 77069 1 0 0.381 0.009
63.3 119133 812 0.5071 0.0001 77069 1 0 1.66 0.04
143.8 60426 839 0.5071 0.0001 77069 1 0 4.77 0.10
186.0 516610 1028 0.5071 0.0001 77069 1 0 4.32 0.04
242.0 505766 923 0.5071 0.0001 77069 0.99901 0.00009 3.616 0.024
295.2 1142200 1180 0.5071 0.0001 77069 1.000301 0.000025 3.211 0.021
352.0 1901600 1460 0.5071 0.0001 77069 0.99916 0.00007 2.769 0.018
609.3 1390560 1224 0.5071 0.0001 77069 0.918 0.004 1.702 0.014
1001.0 20307 315 0.5071 0.0001 77069 1 0 1.242 0.024
1120.3 289246 597 0.5071 0.0001 77069 0.910 0.007 1.087 0.011
1238.1 105437 402 0.5071 0.0001 77069 0.915 0.007 1.009 0.011
1764.5 230191 504 0.5071 0.0001 77069 1.0016 0.0003 0.778 0.005
238.6 287867 551 0.4699 0.0001 11694 1 0 3.70 0.11
338.3 56401 253 0.4699 0.0001 11694 0.9843 0.0005 2.85 0.13
583.2 89766 328 0.4699 0.0001 11694 0.900 0.006 1.79 0.05
727.3 19404 180 0.4699 0.0001 11694 0.9740 0.0016 1.53 0.05
860.0 10535 126 0.4699 0.0001 11694 0.960 0.005 1.36 0.04
911.0 60693 257 0.4699 0.0001 11694 0.9841 0.0010 1.34 0.04
1460.8 54666 235 0.5545 0.0001 7393 1 0 0.89 0.03
Tabla A6.12 Recipiente tipo Marinelli B4 con 114.0 mm de altura de material, situado
directamente sobre ventana del detector DET1.
Energía FEPE
Área Inc. m(kg) Inc. t(s) CSCF Inc. Inc.
(keV) (%)
46.5 282011 1423 1.1837 0.0001 86635 1 0 1.310 0.016
63.3 452920 1262 1.1837 0.0001 86635 1 0 2.41 0.05
143.8 132730 1472 1.1837 0.0001 86635 1 0 3.99 0.07
186.0 1136145 1609 1.1837 0.0001 86635 1 0 3.62 0.04
242.0 1087993 1850 1.1837 0.0001 86635 0.99791 0.00021 2.968 0.020
295.2 2386953 1767 1.1837 0.0001 86635 1.00059 0.00005 2.556 0.017
352.0 4107068 2178 1.1837 0.0001 86635 0.99930 0.00007 2.279 0.015
609.3 3005394 1796 1.1837 0.0001 86635 0.934 0.003 1.396 0.011
1001.0 43972 462 1.1837 0.0001 86635 1 0 1.025 0.016
1120.3 628209 865 1.1837 0.0001 86635 0.927 0.007 0.897 0.009
1238.1 228711 618 1.1837 0.0001 86635 0.931 0.007 0.831 0.009
1764.5 510421 736 1.1837 0.0001 86635 1.00104 0.00022 0.658 0.004
238.6 4021090 2306 1.0559 0.0001 86820 1 0 3.10 0.09
338.3 763331 1174 1.0559 0.0001 86820 0.9865 0.0004 2.30 0.10
583.2 1250052 1252 1.0559 0.0001 86820 0.921 0.005 1.49 0.04
727.3 272060 718 1.0559 0.0001 86820 0.9778 0.0013 1.28 0.04
860.0 149066 522 1.0559 0.0001 86820 0.968 0.005 1.15 0.03
911.0 839690 974 1.0559 0.0001 86820 0.9862 0.0009 1.11 0.04
1460.8 1085582 1045 1.3188 0.0001 74330 1 0 0.742 0.025
139 Apéndice 6–Datos y resultados del Capitulo 4
Tabla A6.23 FEPEs (%) calculadas mediante PENELOPE para el modelo óptimo del
detector DET1 (5 % de incertidumbre) para las diferentes alturas de los materiales de
referencia IAEA RGU-1, RGTh-1 and RGK-1 en los recipientes B1, B2, B3 y B4.
B1 B3 B2 B4
Altura (mm)
Material En (keV) 22.3 32.7 42.6 52.1 72.2 62.2 68.2 101.3 74.8 114.0
RGU-1 46.5 10.0 7.2 5.62 4.67 3.27 4.84 0.435 1.69 0.417 1.38
RGU-1 63.3 12.0 8.9 7.2 5.97 4.26 5.92 1.96 2.96 1.70 2.42
RGU-1 143.8 10.9 8.4 6.8 5.83 4.29 5.67 5.25 4.74 4.38 3.95
RGU-1 186.0 9.2 7.2 5.87 4.94 3.66 4.87 4.92 4.38 4.15 3.62
RGU-1 242.0 7.4 5.8 4.76 4.09 3.01 3.97 4.21 3.70 3.54 3.11
RGU-1 295.2 6.2 4.85 4.00 3.38 2.52 3.28 3.61 3.18 3.03 2.61
RGU-1 352.0 5.27 4.21 3.43 2.95 2.14 2.86 3.13 2.74 2.61 2.27
RGU-1 609.3 3.21 2.56 2.13 1.86 1.37 1.76 2.05 1.77 1.69 1.46
RGU-1 1001.0 2.08 1.71 1.44 1.24 0.94 1.18 1.40 1.22 1.19 1.01
RGU-1 1120.3 1.96 1.56 1.31 1.13 0.86 1.09 1.29 1.10 1.09 0.93
RGU-1 1238.1 1.76 1.46 1.20 1.03 0.78 1.01 1.19 1.04 0.99 0.89
RGU-1 1764.5 1.32 1.10 0.90 0.79 0.61 0.74 0.92 0.79 0.75 0.65
RGTh-1 238.6 7.6 5.98 4.93 4.18 3.08 4.10 4.25 3.71 3.57 3.12
RGTh-1 338.3 5.5 4.34 3.62 3.07 2.29 2.99 3.23 2.84 2.75 2.42
RGTh-1 583.2 3.35 2.68 2.24 1.92 1.43 1.87 2.12 1.85 1.76 1.53
RGTh-1 727.3 2.76 2.24 1.89 1.64 1.21 1.55 1.76 1.53 1.50 1.32
RGTh-1 860.0 2.40 1.92 1.62 1.40 1.06 1.36 1.57 1.37 1.31 1.13
RGTh-1 911.0 2.30 1.83 1.58 1.35 1.01 1.32 1.49 1.29 1.26 1.11
RGK-1 1460.8 1.54 1.26 1.05 0.91 0.69 0.86 1.05 0.90 0.88 0.76
Tabla A6.24 FEPEs (%) calculadas mediante PENELOPE para el modelo óptimo del
detector DET2 (6 % de incertidumbre) para las diferentes alturas de los materiales de
referencia IAEA RGU-1, RGTh-1 and RGK-1 en el recipiente B2.
Altura (mm)
Material En (keV) 5 10 15 20 25 30 35 40 45 50
RGU-1 46.5 22.7 17.7 14.3 12.0 10.2 8.9 7.8 7.0 6.3 5.7
RGU-1 63.3 23.9 19.5 16.1 13.7 11.9 10.6 9.4 8.4 7.7 7.0
RGU-1 143.8 19.5 16.3 13.9 12.0 10.5 9.5 8.5 7.7 7.0 6.5
RGU-1 186.0 16.2 13.5 11.6 10.2 8.9 7.9 7.1 6.4 6.0 5.5
RGU-1 242.0 12.7 10.7 9.2 8.1 7.2 6.4 5.8 5.3 4.77 4.47
RGU-1 295.2 10.4 8.8 7.6 6.7 5.9 5.3 4.82 4.34 4.06 3.74
RGU-1 352.0 8.8 7.5 6.4 5.6 5.1 4.51 4.09 3.76 3.42 3.16
RGU-1 609.3 5.2 4.56 3.91 3.46 3.06 2.77 2.53 2.32 2.16 1.99
RGU-1 1001.0 3.40 2.93 2.60 2.31 2.08 1.89 1.71 1.57 1.44 1.33
RGU-1 1120.3 3.12 2.66 2.40 2.10 1.91 1.71 1.56 1.44 1.34 1.24
RGU-1 1238.1 2.93 2.47 2.20 1.97 1.73 1.59 1.46 1.31 1.23 1.14
RGU-1 1764.5 2.11 1.80 1.61 1.44 1.28 1.15 1.08 0.99 0.94 0.85
RGTh-1 238.6 12.9 10.8 9.3 8.1 7.2 6.5 5.8 5.3 4.84 4.48
RGTh-1 338.3 9.1 7.7 6.6 5.9 5.2 4.67 4.24 3.86 3.56 3.25
RGTh-1 583.2 5.5 4.61 4.04 3.59 3.19 2.85 2.63 2.39 2.20 2.04
RGTh-1 727.3 4.52 3.83 3.34 2.96 2.66 2.42 2.18 2.02 1.85 1.71
RGTh-1 860.0 3.94 3.32 2.93 2.56 2.33 2.10 1.92 1.76 1.64 1.52
RGTh-1 911.0 3.72 3.20 2.76 2.47 2.23 1.99 1.85 1.70 1.58 1.43
RGK-1 1460.8 -- 2.16 -- 1.70 -- 1.40 -- 1.17 -- 1.01
Las Figuras A6.2 y A6.3 muestran las FEPEs experimentales y computacionales con
las incertidumbres correspondientes para las diferentes disposiciones muestra-detector y las 19
emisiones gamma consideradas en este trabajo. En ambas figuras se observa un buen
acuerdo entre los resultados experimentales y computacionales, lo cual se puede constatar
145 Apéndice 6–Datos y resultados del Capitulo 4
Apéndice 7
Datos y resultados experimentales y computacionales correspondientes a
la caracterización del detector HPGe tipo pozo a partir de datos
experimentales (Capítulo 5)
A7.1 Cálculo de los factores de corrección por suma por coincidencia
En el cuarto artículo publicado durante el desarrollo de este trabajo de tesis doctoral, el
cálculo de los CSCF para las líneas gamma de 242.0, 338.3, 583.2, 860.0, 911.0 y 727.3 keV
se ha realizado de la misma manera que en el tercer artículo, empleando las ecuaciones de la
Tabla A6.1 y las FEPEs y eficiencias totales calculadas mediante PENELOPE con el modelo
óptimo del detector. Por otra parte, debido a la gran complejidad del esquema de decaimiento
214
del Bi, en las ecuaciones de dicha tabla correspondientes a este radionucleido, se ha
considerado despreciable la contribución de varias emisiones coincidentes con el fin de
simplificar las ecuaciones. Aunque las ecuaciones funcionan correctamente para detectores
HPGe coaxiales, en este trabajo se ha observado que las ecuaciones con dichas
214
simplificaciones para el Bi subestiman el efecto de suma por coincidencia en el detector
HPGe tipo pozo utilizado.
Por esta razón, en el cuarto artículo los CSCF para las líneas de 609.3, 1120.3 y
1238.1 keV, y para cada altura de muestra, han sido determinados a partir del cociente entre
las FEPEs aparentes (obtenidas para estas energías directamente del espectro adquirido
mediante la medición de la muestra de IAEA RGU-1 correspondiente) y las FEPEs estimadas,
para dichas energías, mediante una curva de calibración en eficiencia experimental. Estas
curvas de calibración en eficiencia han sido obtenidas, para cada una de las muestras de IAEA
RGU-1 medidas en este trabajo, empleado 8 emisiones gamma prácticamente libres de suma
por coincidencia, con energías de: 46.5, 63.3, 143.8, 186.2, 295.2, 351.9, 1001.0 y 1764.5 keV.
Figura A7.1 Curva de eficiencia ajustada a los 8 puntos energía-FEPE experimentales libres
de suma por coincidencia, tomados para la muestra de RGU-1 con 1 mm de altura de muestra.
Figura A7.2 Curva de eficiencia ajustada a los 8 puntos energía-FEPE experimentales libres
de suma por coincidencia, tomados para la muestra de RGU-1 con 2 mm de altura de muestra.
149 Apéndice 7–Datos y resultados del Capitulo 5
Figura A7.3 Curva de eficiencia ajustada a los 8 puntos energía-FEPE experimentales libres
de suma por coincidencia, tomados para la muestra de RGU-1 con 3 mm de altura de muestra.
Figura A7.4 Curva de eficiencia ajustada a los 8 puntos energía-FEPE experimentales libres
de suma por coincidencia, tomados para la muestra de RGU-1 con 4 mm de altura de muestra.
Apéndice 7–Datos y resultados del Capitulo 5 150
Figura A7.5 Curva de eficiencia ajustada a los 8 puntos energía-FEPE experimentales libres
de suma por coincidencia, tomados para la muestra de RGU-1 con 5 mm de altura de muestra.
Figura A7.6 CSCF, calculados para las energías de 242.0, 609.3, 1120.3 y 1238.1 keV,
correspondientes a las 5 muestras medidas del material IAEA RGU-1.
151 Apéndice 7–Datos y resultados del Capitulo 5
Figura A7.7 CSCF, calculados para las energías de 338.3, 583.2, 727.3, 860.0 y 911.0 keV,
correspondientes a las 5 muestras medidas del material IAEA RGTh-1.
Figura A7.8 CSCF, calculados para las energías de 242.0, 338.3, 609.3, 911.0, 1120.3 y
1238.1 keV, correspondientes a las 5 muestras medidas del material IAEA 448.
Apéndice 7–Datos y resultados del Capitulo 5 152
Figura A7.9 CSCF, calculados para las energías de 338.3, 583.2, 609.3 y 911.0 keV,
correspondientes a las 5 muestras medidas del material IAEA 447.
Para ello, 20 botes idénticos a los empleados durante las medidas por espectrometría
gamma se han rellenado completamente de IAEA 448, sumando una masa total de 81.2 0.1
g. Estas 20 muestras han sido colocadas dentro de una cámara de acumulación junto con un
222
detector RADON SCOUT de SARAD GmbH, con el fin de medir la concentración de Rn
(actividad por unidad de volumen) acumulada en la cámara en durante periodo de
aproximadamente un mes, durante el cual el detector de radón ha tomado una medida cada 3
horas. La cámara de acumulación empleada es estanca al radón y está formada por paredes
de metacrilato de 10 mm de espesor, formando un paralelepípedo ortogonal con un volumen
interno 0.185 0.006 m3 (Alonso, 2016).
Los valores de concentración de 222Rn registrados por el detector de radón han sido
ajustados por una curva de la forma:
153 Apéndice 7–Datos y resultados del Capitulo 5
∙
0 [A7.1]
Figura A7.10 Resultados del ajuste de la curva en la ecuación [A7.1] a las mediciones de
concentración de radón en la cámara de acumulación.
[A7.2]
Apéndice 7–Datos y resultados del Capitulo 5 154
[A7.3]
∙ → 1 [A7.4]
∙
1 [A7.5]
∙
Figura A7.11 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
RGU-1 y las energías: 46.5, 63.3, 143.8, 186, 242, 295.2, 352, 609.3, 1001, 1120.3, 1238.1 y
1764.5 keV.
Figura A7.12 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
RGTh-1 y las energías: 238.6, 338.3, 583.2, 727.3, 860 y 911 keV.
Apéndice 7–Datos y resultados del Capitulo 5 156
Figura A7.13 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
RGK-1.
Figura A7.14 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
448 y las energías: 46.5, 186, 242, 295.2, 352, 609.3, 1120.3, 1238.1 y 1764.5 keV.
.
157 Apéndice 7–Datos y resultados del Capitulo 5
Figura A7.15 FEPEs calculadas mediante PENELOPE para las muestras del material IAEA
447 y las energías: 46.5, 186, 242, 295.2, 338.3, 352, 583.2, 609.3, 661.7, 727.3, 860 y 911
keV.