Development of novel eels methods to unveil nanoparticle properties
- Autores: Pau Torruella Besa
- Directores de la Tesis: Sonia Estradé Albiol (dir. tes.), Francisca Peiró Martínez (dir. tes.)
- Lectura: En la Universitat de Barcelona ( España ) en 2019
- Idioma: español
- Tribunal Calificador de la Tesis: Gianluigi A. Botton (presid.), Mercè Segarra Rubí (secret.), Michael Walls (voc.)
- Programa de doctorado: Programa de Doctorado en Nanociencias por la Universidad de Barcelona
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
The aim of this thesis has been two-fold. First, to develop new processing and analysis tools and strategies for extracting information from EELS data, and second, to apply the methods to different nanoparticle systems to shed light to relevant phenomena related to their synthesis and properties. In this regard, chapter 1 presented and overview of the EELS fundamentals and of the state of the art of the technique. Chapter 2 was focused on the advanced computational methods related to EELS data analysis. Moreover, the application of cluster analysis to EELS was introduced, showing its possibilities as an image segmenting and phase identification tool. The following chapters were devoted to the investigation of different material science problems related to NPs that take advantage of the capabilities of quantitative EELS.
The results were grouped by increasing complexity of the performed analysis, with chapter 3 devoted to characterizations that were mainly carried out using EELS elemental mapping, chapter 4 being related to ELNES analysis and chapter 5 to EELS tomography. In the following section, the main findings in these chapters will be summarized.
In chapter 2, the adaptation of data clustering algorithms to the analysis of EELS data, developed within the scope of the present thesis, has been undertaken: • The literature identifies the clustering problem as grouping objects with a certain number of attributes so that objects in the same group have similar attributes and objects on other groups are have very different attributes. A first approach to adapt this to EELS is to think of an object as an EELS spectrum and its attributes as the values for each energy channel that it contains.
• The previous interpretation requires the spectra to be normalized by their total intensity in order to obtains clusters related to different chemical composition and not to different thicknesses.
• The parameter that defines how many clusters are obtained is the threshold that limits the furthest distance of an object (a spectrum) with respect to the centre of its cluster.
• This approach is able to segment a spectrum image in different regions, according to their chemical composition.
• The segmentation achieved can be used as a masking method for subsequent analysis algorithms, such as PCA. This enables the detection of fine gradients in composition, or small variation in the ELNES of the spectra.
• Another way to adapt clustering analysis to EELS is to first perform PCA and then consider an object as a pixel, and its attributes as the values of the loadings in that pixel for the different PCA components. This allows for a clearer, less noisy segmentation. This method enables distinguishing subtle oxidation state changes within a noisy, experimental SI.
• The validity and usefulness of the three strategies proposed has been tested with both simulated and experimental data, with positive results.
In chapter 3.1 the organic synthesis of FeOx@SiO2 NPs was assessed. Several findings were obtained through the HRTEM, STEM-HAADF and EELS characterization of the FeOx@SiO2 NPs at different stages of its synthesis: • HRTEM analysis of the iron oxide seeds is compatible with magnetite phase. The oleic acid coating of the NPs can be directly observed through EELS mapping.
• After the addition of an acetate complex, the carbon coating is observed, by EELS mapping, to become radically different whilst the iron oxide lattice remains as magnetite.
• Cryo-TEM images of the sample show the formation of iron oxide NP clusters when dispersed into an aqueous solution. As samples are observed at different points of the synthesis, silica grows onto the NPs forming a cluster, finally ejected leaving an apple-bite like feature in the shell.
• In the latter stages of the synthesis, smaller (around 1nm) NPs of iron oxide are found in the sample, indicating possible breakage of the original iron oxide NPs.
• A carbon coating is found on the outside of the silica shell and between the iron oxide core and the silica shell, through EELS mapping. The EELS analysis also confirms that no oxidation/reduction occurs on the iron oxide cores throughout the process.
• These findings lead to the development of a microscopic model of the synthesis that involves a previously unreported exchange of oleic/acetate ligands on the surface of the iron oxide NPs through breakage of small particles on its surface.
In chapter 3.2, concerning Au-Ag-Se cation exchange reactions the following findings were made: • The following products of the exchange reactions of Ag2Se NPs with gold ion solutions were found when varying the reaction conditions: o Pure fischesserite nanocrystals, by using TOAB ligand.
o Au-Ag2Se hybrid NPs, by using DDA ligand and short reaction times.
o Au-Ag2AuSe2 hybrid NPs, by using DDA ligand and long reaction times.
• This was found through HRTEM imaging, EELS mapping and HAADF imaging of the NPs • A selenium rich layer is observed on the outside of the Au-Ag2Se and Au-Ag2AuSe2 NPs.
• No epitaxial relationship was found between the gold and silver/gold selenide domains in the hybrid NPs. The interface was found to be amorphous.
• From these observations, the role of the external layer of the Ag2Se seeds as a sacrificial reduction layer for gold ions previous to their diffusion into the crystalline lattice was proposed.
Regarding to the Au-Ag-S system, similar mechanisms were observed: • In the Au-Ag-S system, The HRTEM and STEM-HAADF characterization revealed the following products when the reaction conditions were varied: o Au-Ag3AuS2 NPs when a 3:1 silver to gold ratio was used.
o Au-AgAuS NPs when a 1:1 silver to gold ratio was used.
o Hollow Au2S NPs when a 1:10 silver to gold ratio was used.
• Different crystalline phases were identified in the Ag2S seed NPs.
• Multiple gold domains and defects were observed in the Au-AgAuS NPs.
• The decrease in Ag/Au ratio between the Ag3AuS2 and AgAuS nanocrystals was measured by EDX.
• Samples with the same reaction conditions but with rod-shaped Ag2S seeds were characterized, obtaining rod-shaped Au-Ag3AuS2 NPs, Au-AgAuS NPs and hollow Au2S NPs.
• The full exchange cycle from silver sulphide NPs to gold sulphide NPs was observed, proving that all the intermediate phases were formed in the process.
Chapter 4 was devoted to the characterization of different NPs with an emphasis on the direct observation the oxidation state of its constituents through EELS. In chapter 4.1 , the synthesis of MnOx/Fe3O4 core/shell NPs was assessed: • The standard synthesis process yielded fully-dense core/shell NPs, while a synthesis with addition of water produces hollow core/shell NPs.
• In the fully dense NPs, a diffusion of iron ions into the manganese oxide core was observed.
• The manganese oxide core was found to be Mn3O4 for the hollow NPs and MnO for the dense NPs, indicating that water oxidizes the seeds.
• Pinholes at the surface and the empty core were observed for the hollow NPs through STEM-HAADF tomography.
• From these descriptions, a galvanic process promoted by the oxidation of the manganese oxide seeds was deduced to be the origin of the hollowing process. Subsequent syntheses with degassed and bubbled water confirmed this point. The presence of oxidizing agents dissolved in the water at the high synthesis temperatures was shown to oxidize the manganese oxide seeds.
In chapter 4.2, the measurement of oxidation state at atomic resolution in spinel crystals was proposed as a method to assess cation inversion in the crystal. The necessary methods were developed and applied to iron oxide/manganese oxide core/shell NPs, obtaining the following results: • The characterized samples consisted in Fe3O4/Mn3O4 core/shell NPs. The shell grew epitaxially on the core, which lead to stress in the lattice, inducing the formation of crystalline defects.
• By assuming a linear relationship between the oxidation state of a given element and an ELNES parameter of its EELS edge, it was possible to calculate the inversion parameter from two EELS spectra corresponding to different atomic columns. With this method, an inversion parameter of x = 0.84±0.02 for the iron oxide cores was obtained.
• With the same method and by calculating the average Mn L3 chemical shift between atomic columns with either tetrahedral or octahedral coordination, an inversion parameter of x = 0.39±0.1 was found.
• Macroscopic XAS measurements were performed, obtaining inversion parameters of x = 0.44 for the manganese oxide and x = 0.86 for the iron oxide, in good agreement with the ELNES based methods.
• If a PCA decomposition is obtained and its factors can be assigned to the contributions of ions with different oxidation state, it was proposed to calculate the inversion parameter for each pixel of the SI as the ratio of the corresponding PCA loadings in the atomic columns with a tetrahedral oxygen coordination. This method resulted in x=0.35 for the manganese oxide, compatible with the ELNES based method and the XAS measurement.
• The PCA enabled method evidenced an “inversion gradient” towards the core-shell interface of the NP.
Chapter 5 was devoted to the combination of EELS and tomography. In chapter 5.1 the synthesis of cobalt oxide/cobalt ferrite (CoO@CFO) core/shell NPs and cobalt oxide/manganese ferrite (CoO@MFO) NPs was investigated.
• Both types of NPs presented a polycrystalline structure with cracks on the surface.
• EELS tomography of both samples was performed by integrating the intensity of different EELS edges to obtain element sensitive tilt-series. The reconstructions showed, for both samples, the presence of holes and voids inside the particles. Diffusion of the shell elements inside into the core of the NPs was also observed.
• With the observed morphology of cracks and voids in the sample it was deduced, as for the case studied in chapter 4.1, that a galvanic process took place during the synthesis. The presence of Co3O4 supports this process.
Chapter 5.2 was focused on the achievement of an oxidation state-sensitive tomographic reconstruction: • The studied sample consisted of FeOx/FeOy core/shell NPs that grew with an adapted crystalline lattice and that showed a very similar HAADF contrast. Therefore, they could only be distinguished by EELS.
• A PCA decomposition was obtained. Its factors matched reference spectra for oxides containing exclusively either Fe2+ or Fe3+.
• After performing an intensity correction with information from the HAADF coacquired tilt-series, it was possible to use a compressed sensing tomography algorithm to reconstruct the loading maps of the Fe2+ and Fe3+ components in 3D.
• The intensity ratios of these components in the core and shell demonstrated that the core of the NP was FeO (Fe3+ :Fe2+ = 1) and the shell was most likely Fe3O4 (Fe3+ :Fe2+ = 2) • The widths and thicknesses of core and shell components were measured, and the interface was found to be sharp. The tomographical measurements matched the magnetic characterization of the sample.
• These results constitute the first MVA-Based EELS tomography with ELNES information.
