publications
I see no reason to address the - in any case erroneous - comments of your anonymous expert. — Albert Einstein
theses
2023
- Entwicklung einer Multiskalenmethode für die Simulation von SchmierprozessenHannes HoleyKarlsruhe Institute of Technology (2023)
Reibung und Schmierung sind Multiskalenprobleme, d.h. Prozesse auf unterschiedlichen Zeit- und Längenskalen beeinflussen einander und bestimmen die makroskopische Antwort eines Systems. Für Schmierungsprozesse trifft dies insbesondere im Grenzreibungsbereich zu, in dem die Dicke des Schmierspalts in der Größenordnung molekularer Interaktionslängen liegt. Makroskopische Schmierungsmodellierung basiert fast ausschließlich auf der Anwendung der Reynoldsgleichung, während auf atomarer Skala vermehrt Molekulardynamik-Simulationen in den Vordergrund treten. Multiskalenmethoden für Schmierungsphänomene, die über sequentielle Ansätze hinausgehen, sind bisher noch nicht etabliert. Im Rahmen dieser Arbeit wird ein Multiskalenansatz vorgestellt, welcher die Lösung der makroskopischen Bilanzgleichungen in ein Mikro- und Makroproblem aufteilt. Das Makroproblem entsteht durch Mittelung der Bilanzgleichungen über der Spalthöhe, ähnlich zur konventionellen Reynoldsgleichung, und wird mittels expliziter Finite-Volumen-Diskretisierung gelöst, während das Mikroproblem das konstitutive Verhalten des Schmierfilms enthält. Die numerische Implementierung des Makroproblems wird mithilfe gewöhnlicher Konstitutivgesetze validiert und anhand konkreter Beispiele wird gezeigt, dass diese in Zukunft durch Molekulardynamik-Simulationen ersetzt werden können. Außerdem lassen sich analytische Lösungen der linearisierten Grundgleichungen des Makroproblems herleiten, die mit Autokorrelationsfunktionen fluktuierender Zustandsvariablen aus Molekulardynamik-Simulationen verglichen werden. Daraus ergibt sich eine Methode zur simultanen Bestimmung von Viskosität und Schlupflänge aus Gleichgewichts-Simulationen, sowie die Beschreibung des überkritischen Schalltransports in Fluidspalten. Für eine effiziente Umsetzung des vorgestellten Multiskalenansatzes wird eine Ersatzmodellierung benötigt, die zwischen einzelnen Mikrosimulationen interpoliert. Anhand von einfachen Beispielen wird das Anwendungspotential der Gaußprozess-Regression als mögliches Ersatzmodell evaluiert. Die vorliegende Arbeit liefert somit die theoretischen Grundlagen einer simultanen Multiskalensimulation von Schmierungsprozessen, welche in Zukunft zu einem besseren Verständnis der Dissipationsmechanismen im Grenzreibungsbereich beitragen kann.
- How surface roughness affects adhesionAntoine SannerUniversity of Freiburg (2023)
At atomic scales, all molecules attract each other, but macroscopic objects usually do not stick.The explanation for this apparent paradox is that most surfaces are rough, so that elastically stiff objects only touch on the top of their asperities. Geckos and insects have compliant fibrillar structures or soft pads at the tip of their feet that conform to surface roughness, sustaining enough adhesion to climb vertical walls. Understanding the role of surface roughness in adhesion is a challenge because surfaces exhibit roughness down to the atomic scale. In this thesis, my collaborators and I investigate the effect of surface roughness on adhesion in both stiff and compliant contact systems. I model adhesion theoretically, and I help experimentalists analyze surface topography over multiple scales. The combination of my new models and of the comprehensive surface topography characterization by Abhijeet Gujrati (University of Pittsburgh), allows us to unravel the role of surface roughness in adhesion experiments. Stiff materials do not stick because roughness prevents most of the surfaces to come into the range of molecular attraction. A recent theory quantifies this effect based on an approximate expression for the distribution of interfacial gaps near the contact edge. Joe Monti (Johns Hopkins University) and I benchmark this expression against gap distributions extracted from finely resolved numerical simulations. The theory is valid provided that adhesive stresses are weak and act over a range shorter than a geometrical parameter determined by small-scale roughness. Elastically soft (jelly-like) objects stick because the elastic penalty to deform into intimate contact is small compared to the gain in surface energy. However, theories based on this simple thermodynamic argument cannot explain the fact that in experiments, the force measured during retraction is often much higher than during indentation. This adhesion hysteresis can be caused by material specific irreversibility or elastic instabilities triggered by surface roughness. The role of these instabilities in adhesion hysteresis remains poorly understood because existing numerical and theoretical models cannot account for realistic roughness in soft contacts. I introduce an efficient crack-perturbation model for the contact of rough spheres, enabling large scale simulations with realistic surface roughness. By clarifying the link between adhesion hysteresis and classic pinning problems (for example fracture of heterogeneous materials and wetting angle hysteresis), this model allows me to derive a simple theoretical model linking adhesion hysteresis to surface roughness. In combination with the characterization of surface roughness over multiple scales, my models shed light on the role of elastic instabilities in adhesion experiments. Surfaces are rough from the macroscopic scale down to the atomic scale, and the lack of comprehensive roughness characterization is the major obstacle towards bringing theory and experiments together. Abhijeet Gujrati and collaborators measured the roughness of four diamond coatings over eight decades of length scales, enabling the application of adhesion theories on experiments performed with these samples. Besides the experimental challenge of determining roughness down to the atomic scale, an additional obstacle to the documentation of roughness is the technical complexity of established multiscale roughness measures such as the power spectral density. My collaborators and I address this problem by introducing the scale-dependent roughness parameters (SDRPs), a new analysis framework that is easy to interpret and to implement. This new analysis, together with several established techniques, is available to use through our web-service contact.engineering. We thereby encourage the community to measure, analyze and publish roughness over multiple length scales. The SDRP analysis computes the fluctuations of slopes and curvatures as a function of the lateral length scale. Slopes and curvatures are important ingredients for rough contact theories, but it remains unclear at which scales they matter. Luke Thimons (University of Pittsburgh) and I show that in macro-scale contacts between ruby spheres and diamond coatings, the roughness that critically affects adhesion is between lateral length scales of 43 nm to 1.8 µm. The large-scale cutoff is related to the finite radius of the spherical indenter, while the unimportance of small scales is due to plastic deformations and the long range of the adhesive interaction (5 nm). To determine the critical range of length scales, as well as the parameters of the adhesive interaction, we analyzed the experimental pull-off forces by combining surface topography characterization and numerical simulations. Adhesion is critical in applications such as microelectromechanical systems (MEMS), soft robotics and skin adhesives. Our insights provide guidance for practitioners which scales of roughness to control in order to tune adhesion, and our framework for surface topography characterization will allow a better overall understanding of surface topography across the community.
- Pressure gradients in molecular dynamics simulations of nano-confined fluid flowMohamed Tarek Elewa HassanKarlsruhe Institute of Technology (2023)
A detailed understanding of the behaviour of lubricants under high con- finement is crucial for a range of medical and industrial applications. The hydrodynamic framework provides accurate solutions when the contacting bodies are sufficiently separated, however, at extreme operating conditions, the departure from the Navier-Stokes-Fourier equations is eminent. Atomistic effects can no longer be homogenized and the fluid can not be treated as a continuum-fluid decoupled from the behaviour of its discrete particles. A multiscale treatment of the problem becomes crucial as the fluid operates in the boundary lubrication regime. In this regime, the lubricant is driven by pressure gradients resulting from the gap height variation between the con- tacts. Atomistic models usually rely on non-equilibrium molecular dynamics (NEMD) simulations of periodic molecular representative volume elements (RVE), where the lubricant fluid is confined between slab walls. Due to peri- odicity, introducing pressure gradients in such models presents a hurdle. In this thesis, the “pump” method was developed to introduce pressure gradients in perodic RVEs by applying a local perturbation, based on linear momentum conservation, that induces pressure-driven flow of the lubricant. The inde- pendent variable can be the pressure gradient, by fixing the force, or the mass flux, by fixing the current. The two variants are equivalent. The method was tested on compressible and wetting fluids, and applied in conjunction with different thermostating strategies. Thermodynamic field variables of the fluid lubricant including velocity, pressure, flux, and temperature were measured and reported down to confinements of 3 molecular diameters. The pump method can be applied to a channel of arbitrary geometry. This permits the investigation of hydrodynamic cavitation, a phenomenon that is ubiquitous in nature yet not widely investigated on the molecular scale. A sensitivity analysis was conducted to optimize the channel geometry that promotes cavitation. Subsequently, the cavitation lifetime, growth and collapse were compared to the hydrodynamic theoretical predictions. Within a multiscale framework, the pump method can act as the constraint on the molecular system from the larger continuum scale.
- Self-mated hydrogel frictionJan MeesUniversity of Freiburg (2023)
Sliding, self-mated hydrogel contacts appear in many biological and technical applications. It is known that self-mated hydrogel contacts have a speed-independent friction regime with small friction coefficients at low sliding speeds and a speed-dependent friction regime at high sliding speeds. However, the exact physical mechanisms driving energy dissipation during sliding are unknown. The speed-independent regime has been associated with interfacial polymers relaxing faster than they are deformed. The speed-dependent regime has been associated with both non-equilibrium polymer effects and hydrodynamic lubrication. In this work, I present molecular dynamics simulations to complement the preceding experimental works. I have developed an implicit-solvent based mesoscopic hydrogel friction model with either repulsive or adhesive interactions across the interface. Both the repulsive and adhesive models show velocity-dependent frictional behavior. In combination, they show the hydrogel’s frictional response across the velocity spectrum: interfacial dangling chain conformations are independent of sliding velocity at low velocities, reorient along the shear direction at intermediate velocities and elongate at high velocities. Chain reorientation is associated with the friction coefficient increasing linearly with sliding velocity and the chain stretching is associated with a friction regime with a power-law exponent of approximately 0.5. Following my evidence showing that self-mated hydrogel friction is driven by interfacial effects at low contact pressures, I abstract the hydrogel’s interface during sliding to a system of grafted chains in shear flow. I show that within the brush, shear stress is comprised largely by the polymer’s contribution to the stress tensor (entropic stress). The entropic shear stress is shown to be qualitatively similar in systems of grafted chains and hydrogels. Finally, I show how the inclusion of polydispersity leads to an entropic shear stress which is qualitatively similar to friction curves in self-mated hydrogel friction. In summary, this works strengthens the case for polymer relaxation theory in highly crosslinked, self-mated hydrogel contacts. i
2021
- Antimicrobial polymers and surfaces: theoretical and experimental studiesSarah Mohamed El Sayed MahmoudUniversity of Freiburg (2021)
Theoretical studies performed in this thesis aimed to prepare a molecular dynamics (MD) simulation model for an antimicrobial diamine synthetic mimics of antimicrobial peptides (SMAMP) homopolymer in a united-atom GROMOS 54A7 force field. These studies focused mainly on computing the atomic point charges of a model molecule consisting of three repeat units of a diamine poly(oxanorbornene) SMAMP using Bader analysis and electrostatic potential (ESP) fitting schemes. For this purpose, MD simulations were carried out using the GROMACS software applying the GROMOS 54A7 force field, and density functional theory (DFT) calculations with the Perdew–Burke-Ernzerhof (PBE) functional were carried out using GPAW. For the validation of this model, a comparison was carried out between the potential energy landscapes experienced in the classical force field model (GROMOS54A7) using the newly computed ESP charges and originally implemented van der Waals parameters and the potential energy landscapes in the reference ab initio DFT calculations. These landscapes were computed when a single water molecule was placed at a varying offset from the neutral diamine SMAMP at different adsorption positions. Thus, the applied partial charges and the van der Waals parameters of diamine SMAMP were validated. The results showed that the calculated ESP-fitted charges and Bader charges were in qualitative agreement. The ESP-fitted charges of the neutral diamine SMAMP were comparable to the partial charges of the lysine amino acid, which have already been parameterized and implemented in GROMOS 54A7 forced field. Fluctuations of the values of the charges obtained by changing the applied conformation of diamine SMAMP were also observed. From the energy landscape maps, an overall general good agreement was achieved between the classical force field and DFT calculations. Thus, the ESP-fitted charges and van der Waals parameters of the neutral diamine SMAMP presented here can describe its interaction with liquid water efficiently. Accordingly, it is expected that the ESP-fitted charges and van der Waals parameters presented here would also provide a valid description of the interaction of diamine SMAMP with biomolecules in aqueous solutions. Consequently, this model will enable targeted theoretical studies of the antimicrobial mechanism of action of SMAMPs with various biomolecules via simulations, specifically simulations with the previously reported complex bacterial membranes of the Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus) bacteria that have been studied using the same force field. Experimental studies performed in this thesis aimed to prepare surfaces that contain the antimicrobial propyl SMAMP homopolymer (the second SMAMP homopolymer that was used in this study after the diamine SMAMP) and/or the protein-repellent polysulfobetaines (PSB) and to study their potential to prevent protein and bacterial adhesion, the first steps in biofilm formation. These polymers were immobilized on micro- and nano-patterned structures obtained by the colloidal lithography (CL) technique (structure patterns spacings: 200 nm, 500 nm, 1 µm and 2 µm). The physical properties of the fabricated surfaces were investigated using atomic force microscopy (AFM) and contact angle measurements. Their protein resistivity was studied by surface plasmon resonance (SPR) spectroscopy. Spray tests were used to study their antimicrobial activity against Gram-negative E. coli bacteria. The growth of human gingival mucosal keratinocytes on the fabricated surfaces was analyzed with the Alamar blue assay, optical microscopy, and live-dead staining. Furthermore, for the fabricated bifunctional surfaces, additional quantitative nanomechanical measurements were performed using atomic force microscopy (QNM-AFM) to obtain their local elastic moduli. The results showed the influence of the underlying structure itself on the reduction of the protein and bacterial adhesion. At the small spacings, the structured surfaces had an enhancement in their cell adhesion and antimicrobial activity. Additionally, this effect increased when the patterned surfaces were functionalized with a cell-adhesive polycation polymer, such as propyl SMAMP. The structured surfaces functionalized with propyl SMAMP had improved antimicrobial activity, a reduction of unspecific protein adhesion, and improved cellular adhesion in comparison to the unstructured functionalized surfaces with the same polymer. Therefore, structured surfaces functionalized with adhesive polymers such as SMAMPs could be promising candidates to enhance tissue integration on implants. QNM-AFM studies also showed that the obtained bifunctional surfaces were quite stiff due to the high elastic modules of the underlying substrate. These bifunctional surfaces had a reduced antimicrobial activity compared to softer bifunctional surfaces fabricated by the microcontact printing (µCP) technique. The CL-fabricated surfaces also could not fully achieve the required simultaneous quantitative antimicrobial activity and protein repellency. However, the cell compatibility of the CL-fabricated surfaces was maintained at all the tested spacings. The optimum spacing of these CL-fabricated surfaces for bioactivity was found to be in the range from 500 nm to 1 µm, and a significant reduction of antimicrobial activity was observed at the larger (2 µm) spacing.
2020
- Atomistic mechanics of metallic and network glassesRichard JanaUniversity of Freiburg (2020)
The atomic-scale mechanisms underlying plastic deformation in amorphous materials are investigated in this thesis, aiming to advance the understanding of the macroscopic behavior of these materials. To this end Molecular Dynamics (MD) simulations are carried out on model systems of a Bulk Metallic Glass (BMG) and a network glass: CuZr and amorphous carbon (a-C). First, the size of shear transformation zones (STZs) and evolution of shear bands in CuZr BMG are studied. Second, the elastic properties and yield behavior of a-C are examined. Single bond breaking events are isolated and set into the context of STZs. Shear transformations are characterized by correlations of non-affine displacements in sheared CuZr BMG. In the elastic regime, the correlation shows exponential decay, with a characteristic length that can be interpreted as the size a STZ. After yield, shear bands form and the correlation length becomes system-size dependent. This can be interpreted as a first-order phase transition. The stability of such shear bands is studied next. Pauses intermitting the shear deformation, allowing the sheared system to cool, do not impact a fully developed shear band, even at large applied strain. A simulation cell containing two identical shear bands is created by duplication of the previous cell. Shearing this super cell further does not lead to further localization, rather both shear bands coexist. At elevated temperatures, however, the shear bands dilate and the deformation becomes more homogeneous. For the network glass, a-C, structure and elastic properties are determined from representative volumes, for a range of densities and amorphous morphologies, quench rates and two interatomic potentials with different philosophies: Tersoff+S and Gaussian Approximation Potential (GAP). All samples show a universal relationship between hybridization, density and bulk modulus, despite having distinct cohesive energies. The differences in cohesive energy are traced back to slight changes in the distribution of bond-angles of these structures. Shearing these a-C samples, shear softening and serrated flow is found after an initial elastic response and yield. In this flow regime, deviatoric stress and hydrostatic pressure follow a pressure-modified von Mises (PMvM) law. Using the well defined first and second nearest neighbor relations in a-C, the network structure in terms of bonds and coordination numbers is analyzed. This allows for the application of Thorpe’s constraint counting theory, which can explain why the network becomes floppy below a specific mean coordination number. Finally, individual bond-breaking and -forming events, the basic units of STZs, are studied in quasi-static shear simulations of the a-C structures. The events are extracted from the trajectories by cyclic deformations with small amplitudes. The energy landscape of a-C is explored around such events. The energy barriers obtained follow Eyrings transition state theory. Furthermore, a correlation is found between the distance two atoms jump apart when the bond breaks and the height of the barrier for this event. Using this correlation, the distribution of barrier heights for breaking and closing bonds in a-C is computed. The energy barrier are set into relation with the globally dissipated energy.
2019
- Deformation of metallic multilayers: an atomistic study of the relationship between structure and deformation mechanismsAdrien GolaKarlsruhe Institute of Technology (2019)
The mechanical properties of metallic multilayer materials differ from the bulk because of the properties of interfaces between the layers. These differences become greater as the layer thickness is reduced to the nanoscale. This thesis describes two FCC metallic multilayer materials, Cu1−xAgx|Ni and Cu|Au stacked along their [111] axis. These systems form semi-coherent interfaces with networks of partial Shockley dislocations arranged in a regular triangular pattern. The two systems represent two alternatives to fine tune the interfaces properties. The first one, the Cu1−xAgx|Ni system, uses an alloying element, Ag, to fine tune the lattice mismatch between the layers. The second, the Cu|Au system, is a fully miscible binary system. As such the multilayer stack is a metastable form of the system inevitably leading to intermixing at the interface. Molecular dynamics (MD) and a combination of molecular dynamics and Monte Carlo methods were used to study the structure and strengthening mechanism of these composite materials. Calculations were carried out on various geometries and loading conditions, from representative volume elements to realistic geometries such as nanopillar compression or nanoscratching setups. With the help of large scale MD calculations it was possible to reach system sizes directly comparable to experiments. Ag was found to be a good candidate for alloying in the Cu|Ni binary system as it only forms a solid solution with Cu and therefore should only alter the Cu layers. As Ag was added to the Cu layer, it increased the lattice mismatch with the Ni layer, effectively increasing the density of the network of misfit dislocation at the interface. The excess of Ag in the Cu layer segregated at preferential sites at the interfaces and pinned the dislocation network. Molecular dynamics simulations demonstrated that these modifications increased the strength of the Cu1−xAgx|Ni systems. This model shows that using a well chosen alloying element in a binary multilayer system is a route for tuning the strength of a system under various loading conditions. A custom EAM potential for the Cu|Au system was developed and compared with preexisting EAM potentials. This new potential allowed to successfully describe the unary as well as the stable binary phases existing for this binary system. Investigation of the interface of the Cu|Au system showed that the interface structure and properties were highly sensitive to the intermixing between the two species, with a strong increase of the interface shear strength with intermixing. It was also noticed that this system was sensitive to defects and heterogeneities in its structure which lead, under tribological load, to the formation of vortex instabilities at the interfaces. Finally, catastrophic failure was observed of the system under compression as shear band nucleation was triggered by surface flaws. The observation and understanding of the effect of defects on the mechanical responses of multilayer systems can thus lead to a better material design by avoiding or embracing these defects.
2009
- Multi-scale simulations of carbon nanomaterials for supercapacitors, actuators, and low-friction coatingsLars PastewkaUniversity of Freiburg (2009)
With the help of multi-scale simulation techniques, three examples of the technological application of carbon nanomaterials for storage, conversion and conservation of energy are investigated. First, carbon nanotubes (CNTs) as used for electrodes of electrochemical capacitors. Second, CNTs as used in actuators. And third, diamond and amorphous carbon films as used in wear resistant coatings. In order to be able to model CNT electrodes, a simple charge-transfer model is derived from the CNTs’ electronic structure. This model enables the description of CNTs in the presence of an electrolyte, even in non-equilibrium situations. A simple continuum PoissonNernst-Planck (PNP) equation is then calibrated from molecular dynamics simulations of charging. The PNP model describes cyclic voltammograms, and hence the charge-discharge behavior, of a CNT supercapacitor. It turns out that for a high frequency operation of these devices, the maximum diffusion length of ions within the electrode material itself is crucial. While these devices also have technological potential as actuators, the underlying actuation mechanisms remain poorly understood. Here, charge-induced stresses and strains for electrochemical actuation of carbon nanotubes are calculated from electronic structure theories. For a given deformation mode the concept of bonding and anti-bonding orbitals can be defined depending on the sign of a differential band structure stress. The actuation shows charge asymmetric behavior which is due to next-nearest-neighbor hopping, while Coulombic contributions account for approximately charge-symmetric isotropic deformations. Defects and functional groups have negligible influence on the actuation. In order to understand the behavior of diamond and amorphous carbon under sliding load a new interatomic potential is developed that allows a proper description of the carbon bond under such conditions. Diamond is shown to amorphize under high-load, which is exploited when polishing diamond to remove the material. The surface- and direction-dependency of this amorphization is explained from the bonding-structure of diamond and stress-fluctuations within the amorphous-phase. Finally, the establishment of a low-friction coefficient — the running-in — of amorphous hydrocarbon films is related to a saturation of dangling surfacebonds. This ability to self-passivate crucially depends on the films’ hydrogen content.