Aggregation phenomena in atomic layer deposition: Bridging macro and nano

Author
Fabio Grillo
Year
2018
Abstract & Cover

Atomic layer deposition (ALD) is a gas-phase thin film technology that boasts atomic-level control over the amount of material being deposited. A great deal of research effort has been devoted to the exploitation of ALD precision for the synthesis of nanostructures other than thin films such as supported nanoparticles (NPs). ALD is not only precise but also scalable to high-surface-area supports such as powders, which are relevant to a wide range of applications in fields spanning catalysis, energy storage and conversion, and medicine. Yet, translating the precision of ALD of thin films to the synthesis of NPs is not straightforward. In fact, ALD is mostly understood in terms of self-limiting surface reactions leading to a layer-by-layer conformal growth. However, the formation and growth of NPs is bound to be dictated by atomistic processes other than ALD surface reactions, such as the diffusion and aggregation of atoms and NPs. Understanding the role of such non-equilibrium processes is the key to achieving atomic-level control over the morphology of ALD-grown NPs and, in particular, their particle size distribution (PSD) and shape. This thesis is aimed at expanding our atomic-scale understanding of the mechanisms behind the formation of NPs during ALD. In particular, this thesis is based on experiments and models that were devised with an eye to scalability. The first chapter presents a review of the current understanding of ALD. The fundamentals of ALD are illustrated by means of two examples: AlCH3/H2O ALD of Al2O3, which is known to result in conformal films; and MeCpPtMe3/O2 ALD of Pt, which instead often results in the formation of NPs during the early stages of growth. The thermodynamic driving forces behind the formation of either films or NPs and the kinetic processes that work in the direction of thermodynamic equilibrium are also introduced. Furthermore, this chapter presents a discussion on the effect and the choice of temperature, time, and pressure. Finally, ALD performed in fluidized bed reactors (FBRs) is introduced. FBRs are a type of reactor that enable ALD on bulk quantities of high-surface-area supports and, in particular, nanopowders. Most of the experimental work presented in this thesis regards ALD performed on nanopowders via such reactors. The second chapter presents a mean-field population-balance model capable of describing the formation and growth of NPs during ALD by accounting for: cyclic deposition of atoms on both the substrate and the NPs; diffusion and aggregation of atoms and NPs; atom attachment to NPs; and gas-phase-mediated Ostwald ripening. This model is used to assess the effect of different growth mechanism on experimental observables such as the PSD, number density, and amount of material being deposited. The main conclusion of the analysis presented in this chapter is that the PSD offers a clue into the mechanisms of NP formation and growth. In the third chapter, the population-balance model described in the second chapter is used to interpret the evolution of the PSD as function of the temperature and the number of cycles in MeCpPtMe3/O2 ALD of Pt on gram-scale batches of graphene nanoplatelets. The model clearly shows that Ostwald ripening and layer-by-layer growth, which were previously believed to be the dominant mechanisms, play a minor role. Instead, the NP growth is best described in terms of Smoluchowski aggregation, that is, NP diffusion and coalescence. In particular, we describe the effect of the deposition temperature by allowing for a temperature dependent scaling of the NP mobility. Low temperatures translate into an NP mobility that quickly decays with the NP size, whereas high temperatures result in virtually size-independent NP mobility. Furthermore, the NPs are found to form and grow mostly during the ligand-removal step. This led us to conclude that the local pressure and temperature gradients arising from the combustion of the carbon ligands induce transient NP mobility. In the fourth chapter, we leverage the atomic-scale understanding presented in the second and the third chapter to fabricate gram-scale batches of catalysts based on Pt NPs with tailored PSDs supported on graphene nanoplatelets. By using low and high deposition temperatures we fabricate Pt catalysts with narrow and broad PSDs, respectively. This allows us to study the effect of the shape of the PSD on the activity and the stability of the catalysts against propene oxidation. We find that narrow PSDs, and thus low-temperature ALD, lead to more stable and active Pt catalysts. In particular, the knowledge of the mass-based PSD after the catalytic test allows us to test a simple geometrical model for the activity of the Pt NPs. This analysis led us to conclude that integral properties such as average diameter and dispersion are poor descriptors of the catalytic activity. In the fifth chapter, we explain the effect of the oxygen exposure (PO2 x t) on the ligandremoval kinetics and the degree of metal aggregation in ALD of Pt on TiO2 nanopowders. We find that the ligand-removal kinetics, and thus the amount of Pt being deposited at a given number of cycles, exhibits a sigmoidal (S-shaped) curve as a function of the oxygen exposure. We describe such a curve via a two-step kinetic model accounting for the autocatalytic nature of combustion reactions. In particular, we find that the low oxygen exposures typically used for ALD on flat substrates result in incomplete ligand-removal steps. As a result, negligible amounts of Pt are deposited after the first cycle. Also, the Pt deposited after the first cycle remains mostly atomically dispersed. On the other hand, the high oxygen exposures typically used for ALD on powders effectively remove the carbon ligands, thus enabling Pt deposition after the first cycle and the formation of NPs as large as 5-6 nm even after only 5 cycles. This reconciles the discrepancies between the literature on Pt ALD on flat substrates and that on Pt ALD on powders. Crucially, we find that high partial oxygen pressures (≥ 5 mbar) mitigate metal aggregation, thus leading to narrow PSDs. The sixth chapter is dedicated to the discovery of the formation of TiO2 nanorods during TiCl4/H2O ALD on graphene nanoplatelets performed at 300 ◦C. Our analysis shows that the nanorods form via oriented attachment. The latter is a variant of Smoluchowski aggregation where NPs not only diffuse and collide but also align and fuse along a preferential crystallographic orientation. We find that the emergence of asymmetric growth, and thus of nanorods, is a non-linear function of the exposure time of both precursors. Crucially, we find that the lattices of TiO2 and graphene nanoplatelets are in a rotational alignment driven by lattice matching. This is a strong indication that the aggregation process is substrate-mediated. The seventh chapter presents a multi-scale reaction engineering model devised to understand the effect of precursor transport on ALD in FBRs on nanostructured materials such as nanopowders and nano-porous powders. The model is used to carry out a conservative analysis of the impact of mass-transfer limitations on the precursor utilization efficiency. The latter is crucial for the scale-up of the process. We find that precursor utilization efficiencies close to 100% can be obtained in a wide range of operating conditions. However, at high pressures mass-transfer limitations can result in an inhomogeneous deposition process if the supports consist of porous powders and if the ALD reactions are not carried out to completion. The eight chapter presents a summary of the main results of this thesis, a discussion on the limitations of its approach, and an outlook for further research 

Source of Information
TU Delft repository
University
Delft University of Technology
(Delft, The Netherlands)
External Link
Read Thesis
LinkedIn
linkedin invite