Decoding the Mechanisms of Phase Transitions from In Situ Microscopy Observations

Analysis of the temperature- and stimulus-dependent imaging data toward elucidation of the physical transformations is an ubiquitous problem in multiple fields. Here, temperature-induced phase transition in BaTiO3 is explored using the machine learning analysis of domain morphologies visualized via variable-temperature scanning transmission electron microscopy (STEM) imaging data. This approach is based on the multivariate statistical analysis of the time or temperature dependence of the statistical descriptors of the system, derived in turn from the categorical classification of observed domain structures or projection on the continuous parameter space of the feature extraction-dimensionality reduction transform. The proposed workflow offers a powerful tool for the exploration of the dynamic data based on the statistics of image representation as a function of the external control variable to visualize the transformation pathways during phase transitions and chemical reactions. This can include the mesoscopic STEM data as demonstrated here, but also optical, chemical imaging, etc., data. It can further be extended to the higher dimensional spaces, for example, analysis of the combinatorial libraries of materials compositions.

Shape memory polymers with controlled time-dependent shape recovery

Charly David Azra

The aim of this thesis has been to investigate the use of heat-activated shape memory polymers (SMPs) as novel smart actuators for the controlled delivery of fluids and their suitability for applications such as low-cost drug delivery. On heating to temperatures in the vicinity of their glass transition temperature, Tg, amorphous SMPs are able to recover their equilibrium shape from a “programmed” temporary shape generated by mechanical deformation, and fixed by cooling to below Tg. This phenomenon is linked to the substantial increase in molecular mobility on heating through the glass transition, and the memory of an equilibrium shape associated a crosslinking network. While SMPs have received considerable attention in the literature, there has to date been relatively little focus on the time-dependence of shape recovery in these materials. There are nevertheless many potential biomedical applications, for example, in which it is necessary to control the actuation rate in SMP-based devices activated by natural heat sources such as the human body, whose temperature may be subject to significant fluctuations. The present work has therefore focused on the time-dependent shape memory effect under isothermal conditions and its sensitivity to temperature variations. The first part of the study involved commercial thermoplastic (physically cross-linked) and thermoset (chemically cross-linked) shape memory polyurethanes (SMPUs). This provided insight into the physical mechanisms underlying the time-dependence of the shape memory effect, which was found to correlate strongly with the viscoelastic behaviour of a given material and the thermomechanical variables associated with the shape memory cycle. It also revealed the importance of the width of the α-transition, ∆Tα , as determined by dynamic mechanical analysis (DMA), and its position on the temperature axis relative to the recovery temperature, Tr. In particular, it was shown that controlled recovery rates could be obtained if Tr corresponded to the onset of the α-transition, and a broad transition led to a reduction in the sensitivity of the shape recovery rate to thermal fluctuations about a given Tr. The next part of the study was devoted to the development of materials optimized for controlled shape recovery under well-defined conditions. The commercial formulations were found to be unsuitable for this purpose, because they offered relatively limited possibilities for tailoring Tg and ∆Tα . Thermoset SMPUs with tunable Tg and ∆Tα were therefore produced in house, based on polytetrahudrofuran (PTHF), 4,4’-diphenylmethane diisocyanate (MDI) and 1,1,1-trimethylolpropane (TMP). The starting point was a literature formulation, which was used to optimize a reactive injection molding (RIM) set-up designed to produce void-free parts with a molding time of less than 3 minutes. These SMPUs were characterized by a PTHF network linked by rigid domains made up of crosslinked MDI and TMP. This made it possible to vary Tg from -60 to 80 °C and ∆Tα from 35 to 100 °C, by varying the molecular weight of the PTHF from 2000 to 650 g/mol, and hence the cross-link density for a given stoichiometric ratio PTHF:MDI:TMP. In order to adjust ∆Tα independently of Tg, flexible segments were introduced to the rigid domains and to the junctions between the rigid domains and the PTHF, using a fourth component, 1,4-butanediol (BDO), or by replacing the TMP with ethoxylated TMP (eTMP). This provided a highly flexible platform for optimizing the shape memory performance. Moreover, the resulting SMPUs showed equilibrium moduli, Ee, of up to 13 MPa, which is relatively high for unreinforced thermosets, and clearly beneficial for fluid displacement where high recovery stresses may be important. To increase Ee further, alumina nanoparticles were dispersed in the SMPUs. This was achieved through ultrasonic, solvent-assisted dispersion of the alumina in PTHF and subsequent curing with the MDI and TMP. Reproducible, void-free, SMPU nanocomposite sheets with alumina contents as high as 15 wt% could be produced in this way within the 3 minute pot-life of the reactive mixtures. These showed a homogeneous dispersion of the alumina (aggregates less than 50 nm in diameter), as confirmed by transmission electron microscopy, leading to a modest increase in Ee to about 21 MPa. This indicated reinforcement to be essentially a geometrical effect (strain amplification), suggesting relatively limited matrix/filler interactions. On the other hand, Rr remained high in these nanocomposites (a decrease of only 5 % on addition of 15 wt% alumina), in contrast with reports in the literature of a significant degradation in shape memory performance for SMPUs containing highly interacting fillers. A linear thermo-viscoelastic model was used to simulate the small strain shape memory effect. Continuous relaxation and retardation spectra, H(τ) and L(τ) respectively, were used to model the time-dependence, and a shift factor, aT(T), was used to model the temperature dependence. To streamline the procedure, a novel characterization method was developed for the determination of aT (T ) from a constant frequency DMA temperature sweep, based on the establishment of an approximate relationship between the storage modulus, the loss modulus and aT (T ). It was hence possible to generate continuous data for aT (T ) from a single DMA run, avoiding the time-consuming and operator-dependent construction of a mastercurve from multiple experiments. Using the Havriliak-Negami parametric functions, H(τ) and L(τ) could be subsequently obtained over a large number of decades of time from the modulus and compliance data. Comparison with results from experimental shape memory data for different model SMPUs, showed this approach to provide a reliable indication of the long-term shape recovery rate. Moreover, this method has great potential for providing rapid insight into the temperature-dependent viscoelastic properties of thermorheologically simple polymers in general, regardless of whether they are being considered specifically for shape memory applications. In the final part of the study, a tailored SMPU nanocomposite formulation was used to produce demonstrators in the form of active shape memory reservoirs, capable of expelling a liquid content under isothermal conditions. The reservoirs consisted of circular membranes (the equilibrium shape), inflated at elevated temperature to form 5 ml blisters and subsequently stabilized in this form by cooling (the programmed shape), using a biaxial deformation set-up incorporating temperature control. This innovative programming procedure made it possible to generate active SMPU reservoirs that could be activated at the skin temperature (32 °C), resulting in low average flow rates ranging between 1 and 5 μL/min over several hours. Such as a device has important potential applications, e.g. skin-temperature activated, low-cost transdermal drug delivery patches.

EPFL2013

Ferroelectric Effects Probed by in situ Transmission Electron Microscopy

Reinis Ignatans

Ferroelectric perovskite oxides are widely used in sensors, actuators and optical modulators and, at the same time, they show promise for implementation in future applications such as energy storage, memory and cooling devices. At the infancy of the discovery of polycrystalline ferroelectrics, BaTiO3 was considered as a candidate for various applications, but the breakthrough for the commercialization of ferroelectrics came in 1952 when the ferroelectric PbZrO3 - PbTiO3 (PZT) solid solution system was discovered. PZT and other lead containing perovskite oxides remained on the forefront of scientific and industrial interest. However, concerns over the environmental and health hazards posed by these toxic materials in the electronic and electrical equipment created a legislation change. Nowadays, all applicable products in the EU must pass RoHS (Restriction of Hazardous Substances also known as Directive 2002/95/EC) compliance. This change in the industrial standards fueled research in lead-free ferroelectric materials once again. Attention is currently placed on materials based on BaTiO3, BaZrO3, (Na/K)NbO3, Na1/2Bi1/2TiO3. Their widespread application is yet to be realized as their physical properties pale in comparison to PZT. Substantial challenges also lie in the miniaturization of polar perovskites where effects at nano and microscopic scales become dominant, which inhibit their performance.In this thesis one of the most prominent lead-free ferroelectric perovskites BaTiO3 is studied by transmission electron microscopy (TEM) techniques in combination with in situ temperature and electric field control. In detail, a reliable focused ion beam sample preparation for in situ TEM electrical biasing for ferroelectrics is established. The method is tested on BaTiO3 at room temperature, where the electrical response is observed at the expected field values, further confirmed by finite element calculations. Domain area (polarization) - applied bias loops are directly measured revealing strong pinning at lower fields, while higher fields depin domain walls allowing for more free movement following Rayleigh's law.Upon increasing the temperature, BaTiO3 exhibits morphological transformations in its domain structure from ferroelectric 180° (close to room temperature) to ferroelastic 90° state (below the Curie temperature). Electrical biasing experiments of ferroelastic needle domains show two different forward growth mechanisms. In one case, the needle domains are shown to move freely, influenced mainly by Peierls-like potentials, while when perpendicular domains meet, their movement is hindered by strong domain domain pinning mediated by electromechanical fields. The latter one produces square shaped P-E like loops with distinct steps, indicative of Barkhausen pulses.Finally, the effect of ferroelectric phase transition on the electronic structure is studied with in situ electron energy loss spectroscopy (EELS) and density functional theory (DFT) calculations. Core-loss EELS measurements show that changes in the Ti 3d states qualitatively agree with DFT calculations. Off-axis low-loss EELS allows precise bandgap measurements in both phases. Custom experimental and data treatment methodologies are developed to retrieve momentum resolved dielectric functions, which show excellent fit with DFT calculated ones at small momentum transfer. A pathway to study oxygen deficient perovskites with EELS is theoretically demonstrated with DFT.

EPFL2022

Machines for Materials and Materials for Machines: Metal-Insulator Transitions and Artificial Intelligence

Bernat Mundet

In this perspective, we discuss the current and future impact of artificial intelligence and machine learning for the purposes of better understanding phase transitions, particularly in correlated electron materials. We take as a model system the rare-earth nickelates, famous for their thermally-driven metal-insulator transition, and describe various complementary approaches in which machine learning can contribute to the scientific process. In particular, we focus on electron microscopy as a bottom-up approach and metascale statistical analyses of classes of metal-insulator transition materials as a bottom-down approach. Finally, we outline how this improved understanding will lead to better control of phase transitions and present as an example the implementation of rare-earth nickelates in resistive switching devices. These devices could see a future as part of a neuromorphic computing architecture, providing a more efficient platform for neural network analyses - a key area of machine learning.