Dos grandes lienzos de Murillo analizados con diferentes técnicas analíticas y de imagen

459 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 Dos grandes lienzos de Murillo analizados con diferentes técnicas analíticas y de imagen Anabelle Kriznar Centro Nacional de Aceleradores, Universidad de Sevilla-CSIC-Junta de Andalucía. Departamento de Escultura e Historia de las Artes Plásticas. Facultad de Bellas Artes. Universidad de Sevilla [email protected] Francisco J. Ager Centro Nacional de Aceleradores, Universidad de Sevilla-CSIC-Junta de Andalucía. Departamento de Física Aplicada I, Escuela Politécnica Superior, Universidad de Sevilla [email protected] María Auxiliadora Gómez-Morón Instituto Andaluz de Patrimonio Histórico y Universidad Pablo de Olavide [email protected] Claudia Caliri Laboratori Nazionali del Sud, Istituto Nazionale di Fisica Nucleare, Catania [email protected] Francesco Paolo Romano Istituto per i Beni Archeologici e Monumentali, Consiglio Nazionale delle Ricerche [email protected] Miguel Ángel Respaldiza Centro Nacional de Aceleradores, Universidad de Sevilla-CSIC-Junta de Andalucía. Dpto. de Física Atómica, Molecular y Nuclear, Universidad de Sevilla [email protected] Resumen: Durante la reciente restauración de dos pinturas de gran tamaño de Murillo se llevaron a cabo varios estudios sobre los materiales empleados. Se utilizaron varias técnicas analíticas, como luz visible, UV, IRR, MA-XRF, XRF, SEM-EDX y GC-MS, con el fin de caracterizar los diferentes componentes de las pinturas –capa de preparación, pigmentos y aglutinantes–. Los métodos utilizados durante el estudio se describirán en este trabajo, prestando especial atención al dispositivo móvil MA-XRF, que proporciona imágenes elementales en tiempo real de las pinturas. Los resultados muestran que la preparación fue realizada de la manera habitual del artista, en dos capas, utilizando tierras, calcita y albayalde. La policromía está compuesta de albayalde mezclado con varios pigmentos: ocres rojo y amarillo, bermellón, azurita, esmalte, amarillo plomo-estaño, sombra y negro de carbón. El mapeo con MA-XRF permitió determinar la distribución espacial y la combinación de estos pigmentos a lo largo de la superficie de ambas obras. Palabras clave: Murillo, técnica pictórica, análisis de materiales, OM, UV, IRR, XRF, MA-XRF, SEMEDX, FTIR, GC-MS. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 460 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 Abstract: During the last restoration of two large Murillo’s paintings, several studies on materials applied were carried out. A variety of analytical techniques were used, such as visible and UV light, IRR, MA-XRF, XRF, SEM-EDX and GC-MS with the principal aim to identify different components of both paintings (preparation layer, pigments, binders). In the present work, the methods used during this study will be explained, with the special focus on the mobile equipment MA-XRF, which offers real time elemental images of the paintings. The results show that the preparation was carried out in two layers as a mixture of calcite, earths and lead white, as usual for the artist. The polychromy is made of lead white mixed with different pigments: red and yellow earths, vermilion, azurite, smalt, lead-tin yellow, umber, and carbon black. MA-XRF mapping revealed special distribution and combination of these pigments along the surface on both paintings. Keywords: Murillo, painting technique, material analysis, OM, UV, IRR, XRF, MA-XRF, SEM-EDX, FTIR, GC-MS. 1. Introducción Muchas obras de arte fueron expoliadas durante la invasión francesa de España en el siglo xix; entre ellas varios cuadros de Bartolomé Esteban Murillo (1618-1682, Sevilla), uno de los principales artistas del Barroco español y europeo. Entre estas obras se encontró también un conjunto de seis lienzos creados para decorar la nave de la capilla del Hospital de la Caridad. Durante la invasión napoleónica de Sevilla en 1810, las seis obras fueron confiscadas. Solo dos lienzos de mayor tamaño, La multiplicación de los panes y los peces (figura 1) y Moisés haciendo brotar el agua de la roca de Horeb (figura 2), eran demasiado grandes –3,20 × 5,30 m– para caber en la carreta que debía de transportarlos y, por lo tanto, los dejaron abandonados en Madrid en unas circunstancias pésimas que causaron muchos daños (Valdivieso, 2003). Sin embargo, gracias a esta anécdota, ambos cuadros volvieron a formar parte del patrimonio español. Los otros cuatro lienzos de la misma colección se encuentran hoy repartidos por diferentes museos europeos. Murillo realizó las seis pinturas sobre lienzo entre 1670 y 1674 para la Hermandad de la Santa Caridad. Representan seis de las siete obras de misericordia, cuya iconografía se inspira en una Figura 1. Bartolomé Esteban Murillo: La multiplicación de los panes y los peces, 1667-1670. Hospital de la Caridad, Sevilla. Fondo gráfico IAPH. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 461 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 parte del tratado moralizante Discurso de la Verdad, escrito por el reformador, y uno de los hermanos de la Hermandad, Miguel de Mañara Vicentelo. Las únicas dos obras de la serie hoy in situ en la capilla de la iglesia de la Hermandad representan dos de las acciones de misericordia: La multiplicación de los panes y los peces muestra «dar de comer al hambriento», mientras que Moisés haciendo brotar el agua de la roca de Horeb simboliza «dar de beber al sediento» (Angulo, 1981; Navarrete, 2017). En el 2018, denominado Año Murillo, se celebró el IV centenario de su nacimiento, acompañado con varias exposiciones de sus obras. La organización de las exposiciones conllevó la revisión del estado de los cuadros, su restauración y conservación. La intervención de los dos lienzos objeto del presente estudio se realizó dentro del «Proyecto de conservación de los dos grandes lienzos de Murillo La multiplicación de los panes y los peces y Moisés haciendo brotar agua de la roca de Horeb», que forma parte de un gran proyecto cultural en el marco global del Año Murillo en el que intervinieron el Instituto Andaluz de Patrimonio Histórico (IAPH), la Fundación Bancaria La Caixa y la Humilde y Real Hermandad de la Santa Caridad de Nuestro Señor Jesucristo. La intervención científica que acompañó la restauración se llevó a cabo en los talleres del IAPH en Sevilla, institución perteneciente a la Consejería de Cultura de la Junta de Andalucía y uno de los centros principales de restauración y conservación de España. 2. Objetivos y procedimiento experimental El principal objetivo era conocer los materiales aplicados en ambas obras –soporte, pigmentos, colorantes, aglutinantes y barniz–, los posibles retoques y repintes y, a partir de los resultados obtenidos, conocer también la técnica pictórica de Murillo. La información obtenida haría posible la comparación de ambos cuadros con las demás obras del pintor analizadas previamente en otros centros de investigación y además ofrecería un soporte técnico a los trabajos de conservación y restauración. En el estudio se utilizaron, primero, las técnicas no invasivas, aplicadas in situ en el taller de restauración, como la reflectografía infrarroja (IRR), la técnica de barrido de macrofluorescencia de rayos X (MA-XRF) y el análisis puntal por XRF. Se complementó el procedimiento con técnicas invasivas, extrayendo muestras de ambos cuadros y estudiándolas con microscopio óptico (OM) bajo la luz reflejada y bajo fluorescencia UV, microscopio electrónico de barrido y microanálisis elemental Figura 2. Bartolomé Esteban Murillo: Moisés haciendo brotar el agua de la roca de Horeb, 1667-1670. Hospital de la Caridad, Sevilla. Fondo gráfico IAPH. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 462 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 mediante energía dispersiva de rayos X (SEM-EDX). Para el estudio de recubrimientos y barnices se optó por espectroscopia infrarroja por transformada de Fourier (FTIR), empleando el análisis UATR (Universal Attenuated Total Reflectance) y, para la caracterización de aglutinantes orgánicos y barnices, se usó espectroscopia de gases con detector de espectrometría de masas (GC-MS). 3. Técnicas analíticas no invasivas 3.1. Reflectografía infrarroja (IRR) IRR permite ver posibles dibujos subyacentes y arrepentimientos en la composición y puede identificar algunos pigmentos debido a las tonalidades grises en la imagen. IRR es un método analítico que capta la radiación IR reflejada usando un detector especial. Cuando el objeto se ilumina con la luz halógena, la radiación infrarroja interactúa no solo con la superficie, sino que también penetra e interactúa con las capas subyacentes, obteniéndose información en profundidad sobre los materiales usados (Van Aspern, 1968: 1711-1714; Casini et al., 1999: 39-48; Matteini y Moles, 2001: 203-208). Debido a esto se puede observar el dibujo preparatorio –si no está realizado en color rojo– que revela no solo una parte importante del procedimiento pictórico, sino también la mano del propio artista. Para ello se usó una cámara Xenics Xeva-XS 512 con un detector InGaAs, una resolución de 320 × 256 píxeles y lente Pentax de 16 mm F/1,4. Para el funcionamiento del equipo es necesario el uso de lámparas que proporcionen la radiación IR y, en este caso, se usaron dos fuentes de luz halógena SDI-800 W. Las imágenes se capturaron usando el programa de control automático desarrollado por nuestro laboratorio en Labview, usando las librerías originales del programa X-Control de Xenics. Mediante este mismo programa de control se realiza el escaneo de forma automática, ya que la cámara se acopla a un soporte robotizado de Optimind que permite movimientos suaves en las direcciones x e y (figura 3). Figura 3. IRR análisis in situ con el detalle de la Cámara Xenics Xeva InGaAs. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 463 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 3.2. Fluorescencia de rayos X (XRF) Con la fluorescencia de rayos X se pueden identificar elementos químicos en el punto irradiado. Sirve para caracterizar la mayoría de los materiales inorgánicos en función de sus elementos característicos y también ofrece, en este caso, una estimación semicuantitativa del material empleado. Sin embargo, no es posible identificar los materiales orgánicos debido al número atómico demasiado bajo (Z <14) de los elementos químicos que los componen, ya que –siendo tan ligeros– no pueden ser registrados por el detector cuando se trabaja en el aire con un detector con ventana de berilio. Otra limitación es que no se puede distinguir con seguridad aquellos materiales caracterizados por un mismo elemento químico –por ejemplo, los pigmentos a base de plomo o de cobre– a no ser que se trate de un color específico que pueda ayudar en su interpretación. Sin embargo, es una técnica que ofrece información muy valiosa de donde partir si se necesita obtener respuestas aún más exactas y, sobre todo, disminuye al mínimo la necesidad de extraer muestras. Los pigmentos se pueden identificar a partir de los elementos químicos presentes en los espectros de los puntos analizados, según las energías características de los picos de rayos X (Seccaroni y Moioli, 2000; Volpin y Appolonia, 202: 30-32; Artioli, 2010: 34-37, 119). El equipo de XRF usado en este estudio (figura 4) tiene un generador de rayos x RX38 de EIS, con ánodo de W. El detector es por deriva de silicio (SDD) con una resolución en energía de 140 eV. Una manivela ofrece un movimiento suave y preciso del equipo hacia delante y hacia atrás, lo que es imprescindible para poder acercarse con mucho cuidado al cuadro sin llegar a tocarlo y conseguir la distancia adecuada para el análisis, la cual viene fijada por el punto de convergencia de dos láseres. Todos los puntos analizados se midieron durante 200 segundos bajo las mismas condiciones de 80 µA de corriente y de 34 keV de voltaje, lo que permitió la semicuantificación y comparación de los espectros obtenidos. Figura 4. Análisis puntal con XRF in situ con el detalle del detector y tubo de rayos X. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 464 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 3.3. Sistema portátil de barrido por macrofluorescencia de rayos X (MA-XRF) Un gran avance en la técnica XRF es el sistema portátil de barrido por macrofluorescencia de rayos X (MA-XRF), que permite el barrido de la superficie entera de obras de grandes dimensiones y la obtención, en tiempo real, de mapas elementales de composición mediante el análisis por XRF (Casini et al., 1999: 39-48; Romano et al., 2017: 773; Santos et al., 2018: 96-104). Las imágenes de distribución elemental de los dos cuadros se obtuvieron mediante esta técnica novedosa usando el sistema de barrido de MA-XRF, desarrollado en el laboratorio LANDIS del CNR-IBAM y el INFN-LNS en Catania (Italia). La figura 5 muestra el sistema durante su funcionamiento en el IAPH de Sevilla. El equipo se compone de un tubo de rayos X microfocalizado con ánodo de rodio (30 W, 50 kV y 0,6 mA), focalizado mediante óptica de policapilares hasta un diámetro de haz de 50 mm a una distancia de 1,5 cm, y dos detectores SDD que operan simultáneamente en modo de registro de eventos temporal. La presencia del policapilar para focalizar el haz primario permite integrar en el mismo instrumento tanto macro XRF como micro XRF para ser aplicados en contextos macroscópicos –por ejemplo, cuadros de gran tamaño–. Para aplicar la técnica de microanálisis basta colocar la muestra en el punto focal de la lente policapilar donde el haz primario de rayos X tiene el tamaño mínimo. Para llevar a cabo el análisis macroscópico se coloca la muestra fuera de foco, donde el haz alcanza varios cientos de micras de diámetro. El dispositivo permite realizar un barrido continuo de un área de 110 × 70 cm2 a una velocidad máxima de 100 mm/s. La medida de las pinturas de Murillo se hizo con una resolución lateral de 1 mm por pixel, con un tiempo de permanencia de 10 ms. Gracias a la capacidad de análisis en tiempo Figura 5. Sistema de barrido de macrofluorescencia de rayos X (MA-XRF) in situ, mostrando los mapas de composición en tiempo real y el detalle del tubo de rayos X con el detector. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 465 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 real del instrumento, las imágenes deconvolucionadas de distribución elemental de los pigmentos estaban disponibles durante el propio barrido. El tiempo total de barrido de ambas pinturas fue de 30 horas, cubriendo casi tres cuartas partes de la superficie de cada lienzo, lo que permitió analizar los grupos de figuras más relevantes de las composiciones pictóricas. 4. Técnicas analíticas invasivas Se basan en la extracción de pequeñas muestras de entre 200 y 600 micras en las zonas de interés –diferentes colores a lo largo de la superficie polícroma, intentando captar capas pictóricas, preparación, barniz, posibles repintes–. Del cuadro La multiplicación de los panes y los peces se sacaron 37 muestras, mientras que de Moisés haciendo brotar el agua de la roca de Horeb solo 11. La mayoría de las muestras se prepararon como estratigrafías, embutiéndolas en resina de metacrilato y puliendo la superficie perpendicular para obtener la sección transversal. Las estratigrafías se estudiaron mediante el microscopio óptico de luz reflejada (OM), de barrido (SEM-EDX) y FTIR. 4.1. OM y SEM-EDX El examen de las estratigrafías bajo microscopio óptico en luz reflejada es una herramienta de las más útiles en el estudio de una obra de arte polícroma. Podemos identificar la secuencia de estratos –preparación, imprimación, capas pictóricas, barnices, repintes–, su espesor, la técnica pictórica e, incluso, algunos pigmentos, debido a las formas de sus partículas. Para este estudio se usó el microscopio Leica DM4000M, que también permite la observación de las muestras bajo la fluorescencia UV y, por lo tanto, una identificación más clara de materiales empleados en una muestra, sobre todo repintes posteriores (Matteini y Moles, 2001: 29-35, 37-51; Volpin y Appolonia, 2002: 11-14, 20-24; Artioli, 2010: 64-66). Para el análisis elemental se usó el microscopio electrónico de barrido con microanálisis elemental mediante energía dispersiva de rayos SEM de JEOL 5600 LV con el EDX de Oxford, Inca x-sight, cubriéndolas con oro. Esta técnica permite obtener imágenes con elevados aumentos, pudiendo realizar análisis químico elemental, ya sea puntual, en áreas o mapeo (Matteini y Moles, 2001: 76-85; Volpin y Appolonia, 2002: 38-41; Artioli, 2010: 66-68). 4.2. FTIR y GC-MS Para la identificación de barnices y aglutinantes se emplearon técnicas de caracterización molecular, como espectroscopia infrarroja por transformada de Fourier (FTIR) y espectroscopia de gases con detector de espectrometría de masas (GC-MS) (Matteini y Moles, 2001: 87-91, 94-99; Volpin y Appolonia, 2002: 45-48; Artioli, 2010: 413-415). Los análisis se llevaron a cabo en los laboratorios de LARCO Química y Arte SL en Madrid. 5. Resultados y discusión 5.1. IRR La reflectografía IR se hizo en ambos cuadros, barriendo completamente sus superficies, con especial atención a los grupos figurativos. Los resultados no ofrecieron grandes novedades; no revelaron un dibujo preparatorio –puede que no existiera o que se hiciera en color rojo, «invisible» para la cámara IR– y se observaron muy pocos cambios en la composición, vistos sobre todo a partir de variaciones de tonos grises de las imágenes IR, característicos de los diferentes pigmentos empleados (figura 6). Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 466 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 5.2. XRF, MA-XRF, OM y SEM-EDX Las técnicas que se mencionan a continuación aportaron la información más relevante en cuanto a los pigmentos y soportes empleados (Kriznar et al., 2019: 482-489). Los resultados son complementarios; por lo tanto, se presentan juntos. El tiempo para trabajar con MA-XRF estaba limitado a cinco días y, en consecuencia, se eligieron las zonas más importantes de ambos cuadros. El barrido reveló la disposición de elementos químicos presentes en el cuadro y, por lo tanto, la distribución de pigmentos principales en las figuras, en sus ropajes y en el fondo (figuras 7 y 8). La distribución se complementó con el análisis puntal por XRF. Los elementos químicos principales presentes en ambos cuadros son Ca, Fe, Co, Cu, Hg y Pb; en menor intensidad encontramos Mn, As y Sn. A partir de estos resultados se podría decir que la paleta de Murillo se componía de blanco de plomo, ocres amarillo y rojo, bermellón, amarillo de plomo-estaño, azurita, esmalte, sombra y negro de huesos. Sin embargo, los resultados obtenidos mediante OM y SEMEDX añadieron más información y aclararon algunas dudas. Figura 6. Imagen IRR. Se observan algunos cambios mínimos en la composición, como, por ejemplo, la posición del mango de la jarra. Detalle del cuadro Moisés haciendo brotar el agua de la roca de Horeb. Figura 7. La multiplicación de los panes y los peces. Imagen de MA-XRF de las áreas escaneadas, mostrando la distribución de los elementos químicos principales: Fe, Cu, Co, Hg, Ca, Sn. Varios autores Dos grandes lienzos de Murillo analizados con diferentes… 467 La Ciencia y el Arte VII Ciencias experimentales y conservación del patrimonio Págs. 459-484 5.2.1. Soporte, preparación e imprimación Bajo el microscopio óptico de luz transmitida se identificaron las fibras del textil de soporte y se llegó a la conclusión de que Murillo empleó lienzo de lino. Por otro lado, en todos los puntos analizados con XRF y MA-XRF se observaba presencia de Ca, que podría indicar una preparación de creta (CaCO3) o de yeso (CaSO4), este último más característico en la pintura española de caballete (Bruquetas, 2002: 131). Los análisis por OM y SEM-EDX de las estratigrafías, sin embargo, revelaron que el aparejo hecho por Murillo es más complejo, realizado como mezcla de aluminosilicatos férricos –tierras–, carbonato cálcico –y no sulfato–, albayalde y negro carbón que le da una tonalidad más parda. El espesor oscila entre 98 y 343 µm. Sobre esta se aplica una imprimación con la misma composición, pero con distinta proporción de los materiales utilizados –hay más blanco de plomo, añadido también como secativo– y con un espesor que oscila entre 30 y 100 µm (figura 9). En algunas muestras se identifican restos de pigmentos posiblemente reutilizados de los sobrantes de la paleta, como bermellón o amarillo de plomo-estaño. Debido a su espesor, se acerca más a una preparación que a una imprimación, que suele ser una capa muy fina y sin cuerpo (Townsend et al., 2008). Se aprecia que el tono de las preparaciones varía según la zona de la obra, dependiendo de la mayor o menor cantidad añadida de albayalde, calcita y negro carbón. El objetivo era aprovecharlas como base de color para un modelado más fácil y rápido. 5.2.2. Pigmentos Blanco El principal pigmento en todo el cuadro es blanco de plomo, albayalde ((PbCO3)2·Pb(OH)2), identificado con la presencia de picos intensos de Pb en todos los espectros obtenidos por XRF y MAXRF y confirmado con el análisis puntual por SEM-EDX. Es el principal pigmento para las zonas blancas o muy claras, como el cielo, y para aclarar los demás colores. Es uno de los principales pigmentos para crear carnaciones (figura 9). También se usó en ambas capas de preparación y como secativo. Amarillo Murillo usó dos pigmentos amarillos, el natural ocre amarillo (Fe(OH)2) y el sintético amarillo de plomo estaño (Pb2SnO4), de un color amarillo intenso y brillante. El primero, identificado mediante la presencia de Fe, el otro, mediante Pb y Sn. Si el primero es uno de los pigmentos más básicos Figura 8. Moisés haciendo brotar el agua de la roca de Horeb. Imagen de MA-XRF de las áreas escaneadas, mostrando la distribución de los elementos químicos principales de Fe, Cu, Co, Hg, Ca, Sn. Various authors Two Large panels by Murillo Analysed by Different… 474 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 res reflected IR radiation using a special detector. When the object is lit by halogen light, infrared radiation interacts not only with the surface but also penetrates and interacts with the underlayers, obtaining in-depth information on the materials used (Van Aspern, 1968: 1711-1714; Casini et al., 1999: 39-48; Matteini, and Moles, 2001: 203-208). Because of this, we can see the preparatory drawing – so long as it is not made in red – which reveals not only an important part of the painting procedure, but also the hand of the artist himself. A Xenics Xeva-XS 512 camera with an InGaAs detector with a resolution of 320 × 256 pixels and a Pentax 16 mm F/1.4 lens was used. To operate the equipment, it is necessary to use lamps that provide IR radiation and, in this case, two SDI-800 W halogen light sources were used. The images were captured using the automatic control program developed by our laboratory in Labview, harnessing the original Xenics X-Control program libraries. This same control program was used to scan automatically, as the camera was attached to a robotised Optimind support enabling smooth steering movements in directions x and y (figure 3). 3.2. X-ray fluorescence (XRF) XRF can be used to identify chemical elements at the irradiated point. It serves to characterise most inorganic materials according to their characteristic elements and in this case also provided a semiquantitative estimate of the material used. However, it is not possible to identify organic materials due to the low atomic number (Z <14) of the chemical elements that comprise them, since – being so light – they cannot be recorded by the detector when working in the air with a beryllium window detector. A further limitation is that one cannot safely distinguish materials characterised by the same chemical element, e.g., lead-based or copper-based pigments, unless it is a specific colour that can aid in its interpretation. However, it is a technique that offers valuable information to start from should more accurate responses be needed and, above all, it minimises the need for sampling. Pigments can be identified from the chemical elements present in the spectra of the analysed Figure 3. IRR in situ analysis with the detail of the Xenics Xeva InGaAs camera. Various authors Two Large panels by Murillo Analysed by Different… 475 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 points, according to the energies characteristic to the x-ray peaks (Seccaroni, and Moioli, 2000; Volpin, and Appolonia, 2002: 30-32; Artioli, 2010: 34-37, 119). The XRF equipment used in this study (figure 4) was an x-ray generator RX38 from EIS, with a W anode. The silicon drift detector (SDD) had an energy resolution of 140 eV. A crank ensured the smooth and precise forward and backward movement of the equipment that was needed to approach the painting very carefully without touching it and to achieve the appropriate distance for analysis, determined by the convergence point of two lasers. All analysed points were measured for 200 seconds under the same conditions of 80 µA current and 34 keV voltage, enabling the comparison and semi-quantification of the spectra obtained. 3.3. Portable macro x-ray fluorescence scanning (MA-XRF) system A major breakthrough in the XRF technique was the portable macro x-ray fluorescence scanning (MA-XRF) system that enabled the entire surface of large works to be scanned and elementary composition maps to be obtained in real time by XRF analysis (Casini et al., 1999: 39-48; Romano et al., 2017: 773; Santos et al., 2018: 96-104). The elementary distribution images of the two paintings were obtained using this novel technique with the MA-XRF scanning system developed in the LANDIS laboratory of the CNR-IBAM and the INFN-LNS in Catania (Italy). Figure 5 shows the system during operation at the IAPH in Seville. The equipment consisted of a microfocus x-ray tube with a rhodium anode (30 W, 50 kV and 0.6 mA), focused by polycapillary optics up to a beam diameter of 50 mm at a distance of 1.5 cm, and two SDD detectors operating simultaneously in time-event recording mode. The presence of the Figure 4. In situ point XRF analysis with the detail of the detector and x-ray tube. Various authors Two Large panels by Murillo Analysed by Different… 476 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 polycapillary lens to focus the primary beam allowed for the integration into the same instrument of both macro XRF and micro XRF to be applied in macroscopic contexts such as large paintings. To apply the microanalysis technique, one simply places the sample in the focal point of the polycapillary lens where the primary x-ray beam is the minimum size. To perform the macroscopic analysis, the sample is placed out of focus, where the beam reaches several hundred microns in diameter. The device enables the continuous scanning of an area of 110 × 70 cm2 at a maximum speed of 100 mm/s. The Murillo paintings were measured with a lateral resolution of 1 mm per pixel and a dwell time of 10 ms. Thanks to the instrument’s real-time analysis capability, deconvoluted images of elemental pigment distribution were available during the scan itself. Total scan time for the two paintings was 30 hours, covering almost three-quarters of the surface of each canvas, making it possible to analyse the most important groups of figures in the pictorial compositions. 4. Invasive analytical techniques These were based on the extraction of small samples of between 200 and 600 microns in the areas of interest, i.e., different colours throughout the polychrome surface, trying to capture painting and preparation layers, varnish, possible overpainting, etc. Thirty-seven samples were taken from The Multiplication of the Loaves and the Fishes, and just 11 from Moses and the Water from the Rock of Horeb. Most of the samples were prepared as stratigraphies, using methacrylate resin and polishing the perpendicular surface to obtain the cross-section. The stratigraphies were studied using a reflected light optical microscope (OM), SEM-EDX and FTIR. Figure 5. Macro x-ray fluorescence scanning (MA-XRF) system in situ, showing real-time composition maps and the detail of the x-ray tube with the detector. Various authors Two Large panels by Murillo Analysed by Different… 477 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 4.1. OM and SEM-EDX Examining stratigraphies under a reflected-light optical microscope is one of the most useful tools in the study of a polychrome artwork. We can identify the sequence of strata, including preparation, priming and painting layers, varnishes and overpainting, their thickness, painting techniques and even some pigments, due to the shapes of their particles. A Leica DM 4000M microscope was used for this study, which also enabled sample observation under UV fluorescence and, therefore, a clearer identification of the materials used in a sample, particularly subsequent overpainting (Matteini, and Moles, 2001: 29-35, 37-51; Volpin, and Appolonia, 2002: 11-14, 20-24; Artioli, 2010: 64-66). A JEOL 5600 LV SEM with Oxford Instruments INCA X-sight EDX was used for the elemental analysis, and they were covered with gold. This technique made it possible to obtain images with high magnifications as we could perform elementary chemical analysis, whether specific or by areas or mapping (Matteini, and Moles, 2001: 76-85; Volpin, and Appolonia, 2002: 38-41; Artioli, 2010: 66-68). 4.2. FTIR and GC-MS Molecular characterisation techniques, such as Fourier transform infrared spectroscopy (FTIR) and gas chromatography-mass spectrometry (GC-MS) were used to identify the varnishes and binders. (Matteini, and Moles, 2001: 87–91, 94-99; Volpin, and Appolonia, 2002: 45-48; Artioli, 2010: 413-415). The analyses were performed at the laboratories of LARCO Química y Arte SL in Madrid. 5. Results and discussion 5.1. IRR IR reflectography was performed on both paintings, scanning their complete surfaces, with special attention to the figurative groups. The results did not offer much in the way of new information. They did not reveal a preparatory drawing, so either it did not exist or it was done in red, which is “invisible” to the IR camera. Also, very few changes in composition were observed, seen mainly from variations in grey tones of the IR images, characteristic of the different pigments used (figure 6). Figure 6. IRR Image. Only few minimal changes in the composition can be observed, for example the position of the handle of the jug. Detail of Moses and the Water from the Rock of Horeb. Various authors Two Large panels by Murillo Analysed by Different… 478 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 5.2. XRF, MA-XRF, OM and SEM-EDX The techniques below provided the most relevant information regarding the pigments and supports used (Kriznar et al., 2019: 482-489). The results are complementary and as such are presented together. The time to work with MA-XRF was limited to five days, which is why the most important areas of both panels were chosen. The scan revealed the arrangement of chemical elements present in the paintings and therefore the distribution of the main pigments in the figures, their clothes and the background (figures 7 and 8). The distribution was supported by a point XRF analysis. The main chemical elements present in the two paintings were Ca, Fe, Co, Cu, Hg and Pb. We also found Mn, As and Sn to a lower degree. From these results we could say that Murillo’s palette consisted of lead white, yellow and red ochres, vermilion, lead-tin yellow, azurite, smalt, umber and bone black. However, the results obtained by OM and SEM-EDX added further information and cleared up a number of questions. Figure 8. Moses and the Water from the Rock of Horeb. MA-XRF image of the scanned areas, showing the distribution of the main chemical elements: Fe, Cu, Co, Hg, Ca, Sn. Figure 7. The Multiplication of the Loaves and Fishes. MA-XRF image of the scanned areas, showing the distribution of the main chemical elements: Fe, Cu, Co, Hg, Ca, Sn. Various authors Two Large panels by Murillo Analysed by Different… 479 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 5.2.1. Support, preparation and priming The fibres of the supporting textile were identified under the optical microscope of transmitted light and it was concluded that Murillo used a linen canvas. On the other hand, all the points analysed with XRF and MA-XRF showed the presence of Ca, which could indicate a creta (CaCO3) or gypsum (CaSO4) preparation, the latter more characteristic in Spanish easel painting (Bruquetas, 2002: 131). OM and SEM-EDX analyses of the cross-sections, however, revealed that Murillo’s ground layer was more complex, made from a mix of iron aluminosilicates (umbers), calcium carbonate (and not sulphate), lead white and charcoal black, giving it a darker, brownish hue. The thickness ranged from 98 to 343 µm. A priming layer with the same composition but different material proportions was overlaid (more lead white, also added as a siccative) and with a thickness oscillating between 30 and 100 µm (figure 9). Traces of pigments possibly reused from palette leftovers, such as vermilion and lead-tin yellow, were identified in some samples. Due to their thickness, it was closer to a preparation layer than a priming one, which is usually a very thin layer with no body (Townsend et al., 2008). We could see that the tone of the preparations varies according to the area of the artwork, depending on the higher or lower amount of lead white, calcite and charcoal black added. The aim was to leverage them as a colour base for easier and faster modelling. 5.2.2. Pigments White The main pigment in the whole of the painting was lead white ((PbCO3)2·Pb(OH)2), identified with the presence of strong Pb peaks in all the spectra obtained by XRF and MA-XRF and confirmed with the point SEM-EDX analysis. It was the main pigment for white or very pale areas such as the sky and to lighten other colours. It is one of the main pigments for creating flesh tones (figure 9). It was also used in both preparation layers and as a siccative. Yellow Murillo used two yellow pigments, natural yellow ochre (Fe(OH)2) and bright yellow synthetic leadtin yellow (Pb2SnO4). The first one was identified by the presence of Fe, the second by Pb and Sn. If the former is one of the most basic pigments and often used in mixtures with other pigments, we can also find it as the principal pigment in ochre colours as for soil, bread, jars and some yellow clothing. It was sometimes used to darken other pigments. With the XRF technique we could not distinguish it from the red ochre, identified with the same chemical element, but it could be clearly seen under OM and SEM-EDX. On the other hand, Pb-Sn yellow was used in combination with Figure 9. Cross-section of the flesh sample, Moses and the Water from the Rock of Horeb. (1a) Ground layer comprising iron-rich soils and some quartz and calcite grains; (1b) priming layer of the same composition as above but with more lead white; (2) first polychrome layer made of lead white, a few grains of gypsum, umbers and haematite; (3) a second polychrome layer consisting of a high proportion of lead white and smaller amounts of umber, charcoal and vermilion; (4) third polychrome layer comprising red umber and lead white. Various authors Two Large panels by Murillo Analysed by Different… 480 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 ochre for bright yellow clothing, aureoles, light strands of hair and even, in small quantities, in some depictions of flesh and red or pink clothing, to lighten green colour or add more shine to the blue one, for example in the sky, some clothes, etc. Red With XRF, two pigments could be characterised without any doubt, based on Fe and Hg respectively: red ochre (Fe2O3) and vermilion (HgS), although the high presence of Ca in some areas also points towards a possible use of an organic red dye, not detectable with XRF, on the CaCO3 substrate. The use of lake was confirmed by MO, observing the typical appearance of red lake, while the SEM-EDX analysis confirmed the presence of Ca. Therefore, an organic compound was deposited on calcite (figure10). Applying a dye to a solid substrate was a common process (Feller, 2012: 256). Red ochre was used in several examples of clothes mixed with lead white for pink tones, with azurite blue or smalt to obtain purple colours, and probably in small numbers in the flesh tones as well. A bright red vermilion was chosen for some red clothes; it was also the main pigment, along with lead white, for flesh (figure 9). The shades of red were made with a mix of red ochre and azurite (Cu). Given the date of the paintings, we can speak more confidently about the synthetic vermilion rather than its much more expensive natural cinnabar version. Nevertheless, they could not be distinguished with the techniques used in this study (Eastaugh et al., 2008: 111- 112, 392-393; Roy, 2012: 159-173). Blue The non-invasive analytical techniques found two blue pigments, azurite (2CuCO•Cu(OH)2) and smalt (SiO2+K2O+Al2O3+CoO). These could include impurities of Ca, Ba, Na, Mg, Fe, Ni, Cu, As or Bi, depending on the origin of the cobalt (Matteini, and Moles, 2002: 80; Eastaugh et al., 2008: 351). Azurite, identified on the basis of the Cu peaks, was detected in high quantities in the clothes of some of the apostles and in a lower quantity in the shading of white fabrics. Smalt, a synthetic and much cheaper pigment characterised by Co, Ni and As in the case of the two paintings, was applied to cover larger areas, such as the blue sky with thin clouds, the greenish background of hills and trees and some clothes. It was thought to serve as the underlayer for the Figure 10. Cross-section of the red painting layer applied to the priming and ground layers. Red lake in the red polychromy can be recognised on the calcite grain substrate identified with EDX. Comparison of two XRF spectra from the same area showing a variation in Ca peaks - confirming the calcium carbonate substrate of the lake - depending on the lighter or darker shade. Various authors Two Large panels by Murillo Analysed by Different… 481 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 more expensive azurite; however, the cross-sections showed that Murillo mixed both pigments in the same painting layer (figure 11). Due to chemical changes, smalt was degraded in both paintings, as it is a fairly unstable pigment which, when degraded, loses its blue colour and turns brown. Degradation is thought to occur due to moisture or a reaction with the oil binder, but at the same time it has been seen that, when mixed with lead white, it remains more stable and retains its blue colour better (Eastaugh et al., 2008: 351; Roy, 2012: 116-120; Matteini, and Moles, 2004: 29). Both pigments were also used to achieve purple, mixed with red ochre and/or vermilion and lead white. As for Jesus’s clothes in The Multiplication of the Loaves and Fishes the XRF results were surprising, since there was plenty of azurite in his tunic, but the much cheaper smalt was predominant in the cloak. For principal figures, normally expensive pigments were selected. The study under OM and SEM-EDX of the sample taken from the cloak, on the other hand, revealed that on top of a thick layer of smalt with little azurite there was another very thin layer, almost like a veil, of ultramarine (figure 10), a very expensive pigment obtained from the precious mineral lapis lazuli. The XRF analysis was unable to detect it due to its chemical composition of light elements ((Na, Ca)8Al6Si6O24(S,SO4)). Green At first glance it was difficult to distinguish the blue colours from the green ones due to chemical changes in the pigments and the darkening of the varnish. Several points were analysed where a green colour was expected, which would be identified from Cu, a characteristic chemical element for several green pigments of the time, such as malachite, verdigris or copper resinate. More or less intense Cu peaks were obtained with XRF and MA-XRF, confirming the hypothesis. However, the analysis of samples taken from several areas of green showed that Murillo did not use any green pigment but obtained the colour by mixing azurite (Cu) mixed with lead-tin yellow (Pb, Sn), to which he may have added small amounts of smalt, red ochre, vermilion, charcoal black or red lake (figure 12). Figure 11. Cross-section of the sample from Jesus’s cloak with a very thin surface layer of ultramarine blue overlaid on another thicker layer of umber, calcite, azurite and smalt, identified with SEM-EDX. The XRF analysis identified azurite (Cu) and smalt (Co, Ni, As) in Jesus’s tunic (dotted line), while only smalt and a very small amount of azurite, plus quite a lot of umber, were observed in the cloak (solid line). Various authors Two Large panels by Murillo Analysed by Different… 482 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 Brown and black The main pigment to obtain brown tones is umber (Mn, Fe). It is usually used in combination with other pigments, especially to darken the colour, such as yellow, blue or green. It can be raw or burnt (Eastaugh et al., 2008: 383-384), thus varying its tone. We find dark brown in some of the characters’ hair or the pupils of their eyes. It was often mixed with a black pigment to obtain more depth. The XRF analysis showed high peaks of Ca in the dark and black areas; it was therefore thought that the pigment used was bone black. This could not be confirmed with the presence of P, which is too light an element. But examination under SEM-EDX offered another result: it did not detect Ca or P in the black colour, but rather charcoal black, easy to recognise under the high magnifications of the electron microscope. It was therefore concluded that Murillo used charcoal black, a black pigment of plant origin. 5.2.3. Interventions In several areas of both paintings, low peaks of Ti and/or Zn were detected, indicating subsequent overpainting from the twentieth century, executed on the basis of modern zinc white (ZnO) and titanium white (TiO2) pigments (Feller, 2012: 169-172, 182; West, 2012: 295-297, 336). The peaks were weak and sometimes hard to identify. In most of the spectra where its presence was discovered, it matched more intense peaks of Ca. Analysis of the samples taken from the corresponding areas confirmed a gypsum stucco underlying the retouches. Prussian blue (Fe7(CN)18) was also discovered, which came onto the market at the turn of the seventeenth century (Eastaugh et al., 2008: 314-315; West, 2012: 191-195, 207-208). It was identified by strong Fe peaks in comparison Figure 12. Analysis of the green colour. (a) Cross-section of the sample showing the green painting layer made made as a mixture of red, blue, black, yellow and white. (b) SEM image confirming the combination of light and heavy pigments in all the layers. (c) XRF spectrum detecting the presence of Cu, which does not identify a copperbased green but an azurite one, as can be seen from the cross-section. (d) EDX spectrum confirming the presence of Pb-Sn yellow which, together with azurite, gives a green colour. Various authors Two Large panels by Murillo Analysed by Different… 483 Science and Art VII Experimental Sciences and Heritage Conservation Pages 459-484 with other points analysed in the same area of the sky and was limited to small areas of retouches at the upper extreme edges of the painting. Its presence was confirmed with the OM and SEM-EDX analysis of the samples. 5.3. FTIR and GC-MS Both techniques were used to characterise organic materials in the samples. They confirmed that the binder Murillo used was linseed oil, both in the two preparation layers (there is more binder in the second layer) and in the painting layers. On the other hand, egg tempera was applied in the subsequent overpainting work and the surface of the painting was covered with a conifer resin varnish. 6. Conclusion During the restoration of two paintings by Murillo, The Multiplication of the Loaves and Fishes and Moses and the Water from the Rock of Horeb, a thorough study of the materials used by the artist was carried out. Principally, non-invasive techniques were used, and the combination of the results provided quite comprehensive information on his painting technique. As in some of his other works, it was observed, that the preparation consisted firstly of a ground layer and then a priming layer of a similar composition, made with a load of CaCO3, lead white, umbers and sometimes traces of other pigments to vary the shade. His palette is characteristic of the seventeenth century, made up of lead white, yellow and red ochres, vermilion, red lake, lead-tin yellow, smalt, azurite and even a small amount of ultramarine for Jesus’s cloak, umber and charcoal black. No green pigment as such was identified, but the colour was achieved by mixing azurite, lead-tin yellow, umbers and sometimes charcoal black. Large areas of smalt degradation were observed; the pigment is more stable when mixed with more lead white. The MA-XRF scan showed the distribution of the characteristic chemical elements, making it easier to identify the pigments and their mixtures. GC-MS identified the binder as linseed oil, while egg tempera was used for retouches. These, already observed under UV light and confirmed with XRF and SEM-EDX, included modern pigments such as zinc white, titanium white and Prussian blue. Murillo’s original palette and painting technique correspond to the technical recommendations on the painting process in the contemporary treatises of the seventeenth century, such as those by Francisco Pacheco and Antonio Palomino, two of the leading theorists of the Golden Age. Acknowledgements The “Conservation Project for the Two Large Murillo panels The Multiplication of the Loaves and Fishes and Moses and the Water from the Rock of Horeb” was made possible by the partnership agreement between the Instituto Andaluz del Patrimonio Histórico, the Fundación Bancaria La Caixa and la Humilde y Real Hermandad de la Santa Caridad de Nuestro Señor Jesucristo, signed on 10 March 2017 to carry out the related technical, scientific and cultural activities and actions. The collaboration and engagement of the IAPH professionals Lourdes Núñez Casares, Rocío Magdaleno and Eugenio Fernández Ruiz, as well as Mr. Carlos Cornejo González, is appreciated for their help with taking the measurements. The authors wish to thank the professional support of the CSIC topical interdisciplinary platform: Open Heritage Investigation and society (PTI-PAS). Bibliografía / Bibliography Angulo Íñiguez, D. (1981): Murillo. Su arte, su vida, su obra, I. Madrid: Espasa Calpe. Artioli, G. (2010): Scientific Methods and Cultural heritage: An Introduction to the Application of Materials Science to Archaeometry and Conservation Science. New York: University Press, Oxford.