Quantitative Structure Determination of Large Three‐Dimensional Nanoparticle Assemblies

Self-assembly of nanoparticles has recently received increasing interest [ 1 ] because it is an attractive route towards systems in which the properties can be tuned using parameters such as the size of the individual particles and the nature of their threedimensional (3D) stacking. However, in order to understand the structure/property relationship, a thorough characterization of the nanoparticle organization is of crucial importance. Although conventional transmission electron microscopy (TEM) is a very useful tool to study nanoscale materials, it only provides twodimensional (2D) projection images of a 3D object. Therefore, electron tomography has become a standard technique in the 3D characterization of nanomaterials. In electron tomography, 2D projection images acquired using a TEM are combined into a 3D reconstruction through a mathematical algorithm. Several groups have demonstrated the ability to investigate nanoparticle assemblies by electron tomography. [ 2 , 3 ] In those cases, either bright fi eld TEM (BF-TEM) or high angle annular dark fi eld scanning TEM (HAADF-STEM) was used to obtain the series of 2D images. However, if one wants to extract quantitative information, optimization of the electron tomography experiment is required. This is especially the case for large assemblies that have a thickness > 500 nm. For such systems, the conventional approaches yield different types of artifacts hampering a quantitative interpretation of the 3D data. Here, we propose an improved route towards the quantitative structure determination of large 3D nanoparticle assemblies, which required optimization of both the acquisition technique and the reconstruction algorithm. We demonstrate that this approach is of crucial importance in the investigation of so-called “superspheres”, containing Au nanoparticles stabilized by a polymer matrix. Because they have a large diameter and contain no specifi c features that can hamper a reliable reconstruction, these large assemblies serve as good test objects to investigate the infl uence of both the acquisition technique and the reconstruction algorithm on the quality of the reconstruction. Such systems have also become of high importance in the fi eld of nanoplasmonics because they may lead to controlled hot spot formation and antenna effects. [ 4 ] By changing the size of the