Scientific machine learning for modeling and simulating complex fluids.

The formulation of rheological constitutive equations-models that relate internal stresses and deformations in complex fluids-is a critical step in the engineering of systems involving soft materials. While data-driven models provide accessible alternatives to expensive first-principles models and less accurate empirical models in many engineering disciplines, the development of similar models for complex fluids has lagged. The diversity of techniques for characterizing non-Newtonian fluid dynamics creates a challenge for classical machine learning approaches, which require uniformly structured training data. Consequently, early machine-learning based constitutive equations have not been portable between different deformation protocols or mechanical observables. Here, we present a data-driven framework that resolves such issues, allowing rheologists to construct learnable models that incorporate essential physical information, while remaining agnostic to details regarding particular experimental protocols or flow kinematics. These scientific machine learning models incorporate a universal approximator within a materially objective tensorial constitutive framework. By construction, these models respect physical constraints, such as frame-invariance and tensor symmetry, required by continuum mechanics. We demonstrate that this framework facilitates the rapid discovery of accurate constitutive equations from limited data and that the learned models may be used to describe more kinematically complex flows. This inherent flexibility admits the application of these "digital fluid twins" to a range of material systems and engineering problems. We illustrate this flexibility by deploying a trained model within a multidimensional computational fluid dynamics simulation-a task that is not achievable using any previously developed data-driven rheological equation of state.

Anomalous crystalline ordering of particles in a viscoelastic fluid under high shear.

Addition of particles to a viscoelastic suspension dramatically alters the properties of the mixture, particularly when it is sheared or otherwise processed. Shear-induced stretching of the polymers results in elastic stress that causes a substantial increase in measured viscosity with increasing shear, and an attractive interaction between particles, leading to their chaining. At even higher shear rates, the flow becomes unstable, even in the absence of particles. This instability makes it very difficult to determine the properties of a particle suspension. Here, we use a fully immersed parallel plate geometry to measure the high-shear-rate behavior of a suspension of particles in a viscoelastic fluid. We find an unexpected separation of the particles within the suspension resulting in the formation of a layer of particles in the center of the cell. Remarkably, monodisperse particles form a crystalline layer which dramatically alters the shear instability. By combining measurements of the velocity field and torque fluctuations, we show that this solid layer disrupts the flow instability and introduces a single-frequency component to the torque fluctuations that reflects a dominant velocity pattern in the flow. These results highlight the interplay between particles and a suspending viscoelastic fluid at very high shear rates.

Microscopic dynamics underlying the stress relaxation of arrested soft materials.

Arrested soft materials such as gels and glasses exhibit a slow stress relaxation with a broad distribution of relaxation times in response to linear mechanical perturbations. Although this macroscopic stress relaxation is an essential feature in the application of arrested systems as structural materials, consumer products, foods, and biological materials, the microscopic origins of this relaxation remain poorly understood. Here, we elucidate the microscopic dynamics underlying the stress relaxation of such arrested soft materials under both quiescent and mechanically perturbed conditions through X-ray photon correlation spectroscopy. By studying the dynamics of a model associative gel system that undergoes dynamical arrest in the absence of aging effects, we show that the mean stress relaxation time measured from linear rheometry is directly correlated to the quiescent superdiffusive dynamics of the microscopic clusters, which are governed by a buildup of internal stresses during arrest. We also show that perturbing the system via small mechanical deformations can result in large intermittent fluctuations in the form of avalanches, which give rise to a broad non-Gaussian spectrum of relaxation modes at short times that is observed in stress relaxation measurements. These findings suggest that the linear viscoelastic stress relaxation in arrested soft materials may be governed by nonlinear phenomena involving an interplay of internal stress relaxations and perturbation-induced intermittent avalanches.

Interplay between wall slip and shear banding in a thixotropic yield stress fluid.

We study the local dynamics of a thixotropic yield stress fluid that shows a pronounced non-monotonic flow curve. This mechanically unstable behavior is generally not observable from standard rheometry tests, resulting in a stress plateau that stems from the coexistence of a flowing band with an unyielded region below a critical shear rate c. Combining ultrasound velocimetry with standard rheometry, we discover an original shear-banding scenario in the decreasing branch of the flow curve of model paraffin gels, in which the velocity profile of the flowing band is set by the applied shear rate  instead of c. As a consequence, the material slips at the walls with a velocity that shows a non-trivial dependence on the applied shear rate. To capture our observations, we propose a differential version of the so-called lever rule, describing the extent of the flowing band and the evolution of wall slip with shear rate. This phenomenological model holds down to very low shear rates, at which the dimension of the flowing band becomes comparable to the size of the individual wax particles that constitute the gel microstructure, leading to cooperative effects. Our approach provides a framework where constraints imposed in the classical shear-banding scenario can be relaxed, with wall slip acting as an additional degree of freedom.

Aluminosilicate colloidal gels: from the early age to the precipitation of zeolites.

Aluminosilicate hydrogels are often considered to be precursors for the crystallisation of zeolites carried out under hydrothermal conditions. The preparation of mechanically homogeneous aluminosilicate gels enables the study of these materials through bulk rheology and observation of the aging dynamics until the precipitation of crystalline zeolites. The first part of this study deals with the establishment of ternary state diagrams, in order to identify the range of chemical formulations that enable preparation of single-phase homogeneous gels. Then, by studying the viscoelastic moduli during the gelation reaction, and by yielding the gel under large deformation, we propose an empirical law considering the partial order of reaction on each chemical element, to predict the gelation time according to the chemical formulation. The scaling behavior of the elastic properties of this colloidal gel shows a transition from a strong link behavior to a weak link regime. Long term aging results in the shrinkage of the gel, accompanied by syneresis of interstitial liquid at the surface. Zeolites precipitate through crystallisation by a particle attachment mechanism, when thermodynamic equilibrium is reached. The stoichiometry of the precipitated zeolites is not only consistent with the concentration of the remaining species in the supernatant but, surprisingly, it is also very close to the partial order of the reaction of the chemical elements involved in the determination of the critical gel point. This indicates a strong correlation between the morphology of the soft amorphous gel network that is formed at an early age and those of the final solid precipitated crystals.

Time-connectivity superposition and the gel/glass duality of weak colloidal gels.

Colloidal gels result from the aggregation of Brownian particles suspended in a solvent. Gelation is induced by attractive interactions between individual particles that drive the formation of clusters, which in turn aggregate to form a space-spanning structure. We study this process in aluminosilicate colloidal gels through time-resolved structural and mechanical spectroscopy. Using the time-connectivity superposition principle a series of rapidly acquired linear viscoelastic spectra, measured throughout the gelation process by applying an exponential chirp protocol, are rescaled onto a universal master curve that spans over eight orders of magnitude in reduced frequency. This analysis reveals that the underlying relaxation time spectrum of the colloidal gel is symmetric in time with power-law tails characterized by a single exponent that is set at the gel point. The microstructural mechanical network has a dual character; at short length scales and fast times it appears glassy, whereas at longer times and larger scales it is gel-like. These results can be captured by a simple three-parameter constitutive model and demonstrate that the microstructure of a mature colloidal gel bears the residual skeleton of the original sample-spanning network that is created at the gel point. Our conclusions are confirmed by applying the same technique to another well-known colloidal gel system composed of attractive silica nanoparticles. The results illustrate the power of the time-connectivity superposition principle for this class of soft glassy materials and provide a compact description for the dichotomous viscoelastic nature of weak colloidal gels.

Non-Maxwellian viscoelastic stress relaxations in soft matter.

Viscoelastic stress relaxation is a basic characteristic of soft matter systems such as colloids, gels, and biological networks. Although the Maxwell model of linear viscoelasticity provides a classical description of stress relaxation, it is often not sufficient for capturing the complex relaxation dynamics of soft matter. In this Tutorial, we introduce and discuss the physics of non-Maxwellian linear stress relaxation as observed in soft materials, the ascribed origins of this effect in different systems, and appropriate models that can be used to capture this relaxation behavior. We provide a basic toolkit that can assist the understanding and modeling of the mechanical relaxation of soft materials for diverse applications.

Rod-climbing rheometry revisited.

The rod-climbing or "Weissenberg" effect in which the free surface of a complex fluid climbs a thin rotating rod is a popular and convincing experiment demonstrating the existence of elasticity in polymeric fluids. The interface shape and steady-state climbing height depend on the rotation rate, fluid elasticity (through the presence of normal stresses), surface tension, and inertia. By solving the equations of motion in the low rotation rate limit for a second-order fluid, a mathematical relationship between the interface deflection and the fluid material functions, specifically the first and second normal stress differences, emerges. This relationship has been used in the past to measure the climbing constant, a combination of the first (Ψ1,0) and second (Ψ2,0) normal stress difference coefficients from experimental observations of rod-climbing in the low shear rate limit. However, a quantitative reconciliation of such observations with the capabilities of modern-day torsional rheometers is lacking. To this end, we combine rod-climbing experiments with both small amplitude oscillatory shear (SAOS) flow measurements and steady shear measurements of the first normal stress difference from commercial rheometers to quantify the values of both Ψ1,0 and Ψ2,0 for a series of polymer solutions. Furthermore, by retaining the oft-neglected inertial terms, we show that the "climbing constant"  = 0.5Ψ1,0 + 2Ψ2,0 can be measured even when the fluids, in fact, experience rod descending. A climbing condition derived by considering the competition between elasticity and inertial effects accurately predicts whether a fluid will undergo rod-climbing or rod-descending. Our results suggest a more general description, "rotating rod rheometry" instead of "rod-climbing rheometry", to be more apt and less restrictive. The analysis and observations presented in this study establish rotating rod rheometry combined with SAOS measurements as a prime candidate for measuring normal stress differences in complex fluids at low shear rates that are often below commercial rheometers' sensitivity limits.

Elastoviscoplasticity, hyperaging, and time-age-time-temperature superposition in aqueous dispersions of bentonite clay.

Clay slurries are both ubiquitous and essential in the oil exploration industry, and are most commonly employed as drilling fluids. Due to its natural abundance, bentonite clay is often the de facto choice for these materials. Understanding and predicting the mechanical response of these fluids is critical for safe and efficient drilling operations. However, rheological modeling of bentonite clay suspensions is complicated by the fact that thermally-driven microscopic arrangements of particle aggregates lead to a continual evolution of the viscoelastic properties and the yield stress of the suspension with time. Ergodic relations fundamental to linear viscoelastic theory, such as the Boltzmann superposition principle, do not hold in this scenario of 'rheological aging'. We present an approach for modeling the linear viscoelastic response of aging bentonite suspensions across a range of temperatures that is based on the transformation from laboratory time to an effective 'material time' domain in which time-translation invariance holds, and the typical relations of non-aging linear viscoelastic theory apply. In particular, we model the constitutive relationship between stress and strain-rate in the bentonite suspensions as fractional Maxwell gels with constant relaxation dynamics in the material time domain, in parallel with a non-aging Newtonian viscous contribution to the total stress. This approach is supported by experimental measurements of the stress relaxation and rapid time-resolved measurements of the linear viscoelastic properties performed using optimized exponential chirps. This data is then reduced to master curves in the material domain using time-age-time superposition to obtain best fits of the model parameters over a range of operating temperatures.

Restoring universality to the pinch-off of a bubble.

The pinch-off of a bubble is an example of the formation of a singularity, exhibiting a characteristic separation of length and time scales. Because of this scale separation, one expects universal dynamics that collapse into self-similar behavior determined by the relative importance of viscous, inertial, and capillary forces. Surprisingly, however, the pinch-off of a bubble in a large tank of viscous liquid is known to be nonuniversal. Here, we show that the pinch-off dynamics of a bubble confined in a capillary tube undergo a sequence of two distinct self-similar regimes, even though the entire evolution is controlled by a balance between viscous and capillary forces. We demonstrate that the early-time self-similar regime restores universality to bubble pinch-off by erasing the system's memory of the initial conditions. Our findings have important implications for bubble/drop generation in microfluidic devices, with applications in inkjet printing, medical imaging, and synthesis of particulate materials.

Hydrodynamics control shear-induced pattern formation in attractive suspensions.

Dilute suspensions of repulsive particles exhibit a Newtonian response to flow that can be accurately predicted by the particle volume fraction and the viscosity of the suspending fluid. However, such a description fails when the particles are weakly attractive. In a simple shear flow, suspensions of attractive particles exhibit complex, anisotropic microstructures and flow instabilities that are poorly understood and plague industrial processes. One such phenomenon, the formation of log-rolling flocs, which is ubiquitously observed in suspensions of attractive particles that are sheared while confined between parallel plates, is an exemplar of this phenomenology. Combining experiments and discrete element simulations, we demonstrate that this shear-induced structuring is driven by hydrodynamic coupling between the flocs and the confining boundaries. Clusters of particles trigger the formation of viscous eddies that are spaced periodically and whose centers act as stable regions where particles aggregate to form flocs spanning the vorticity direction. Simulation results for the wavelength of the periodic pattern of stripes formed by the logs and for the log diameter are in quantitative agreement with experimental observations on both colloidal and noncolloidal suspensions. Numerical and experimental results are successfully combined by means of rescaling in terms of a Mason number that describes the strength of the shear flow relative to the rupture force between contacting particles in the flocs. The introduction of this dimensionless group leads to a universal stability diagram for the log-rolling structures and allows for application of shear-induced structuring as a tool for assembling and patterning suspensions of attractive particles.

Spectral Universality of Elastoinertial Turbulence.

Dissolving small amounts of polymer into a Newtonian fluid can dramatically change the dynamics of transitional and turbulent flows. We investigate the spatiotemporal dynamics of a submerged jet of dilute polymer solution entering a quiescent bath of Newtonian fluid. High-speed digital Schlieren imaging is used to quantify the evolution of Lagrangian features in the jet revealing a rich sequence of transitional and turbulent states. At high levels of viscoelasticity, we identify a new distinct transitional pathway to elastoinertial turbulence (EIT) that does not feature the conventional turbulent bursts and instead proceeds via a shear-layer instability that produces elongated filaments of polymer due to the nonlinear effects of viscoelasticity. Even though the pathways to the EIT state can be different, and within EIT the spatial details of the turbulent structures vary systematically with polymer microstructure and concentration, there is a universality in the power-law spectral decay of EIT with frequency, f^{-3}, independent of fluid rheology and flow parameters.

Crack morphologies in drying suspension drops.

A drop of an aqueous suspension of nanoparticles placed on a substrate forms a solid deposit as it dries. For dilute suspensions, particles accumulate within a narrow ring at the drop edge, whereas a uniform coating covering the entire wetted area forms for concentrated suspensions. In between these extremes, we report two additional regimes characterized by non-uniform deposit thicknesses and by distinct crack morphologies. We show that both the deposit shape and the number of cracks are controlled exclusively by the initial particle volume fraction. The different regimes share a common avalanche-like crack propagation dynamics, as a result of the delamination of the deposit from the substrate.

Characterizing viscoelastic properties of synthetic and natural fibers and their coatings with a torsional pendulum.

Characterizing and understanding the viscoelastic mechanical properties of natural and synthetic fibers is of great importance in many biological and industrial applications. Microscopic techniques such as micro/nano indentation have been successfully employed in such efforts, yet these tests are often challenging to perform on fibers and come with certain limitations in the interpretation of the obtained results within the context of the macroscopic viscoelasticity in the fiber. Here we instead explore the properties of a series of natural and synthetic fibers, using a freely-oscillating torsional pendulum. The torsional oscillation of the damped mass-fiber system is precisely recorded with a simple HD video-camera and an image processing algorithm is used to analyze the resulting videos. Analysis of the processed images show a viscoelastic damped oscillatory response and a simple mechanical model describes the amplitude decay of the oscillation data very well. The natural frequency of the oscillation and the corresponding damping ratio can be extracted using a logarithmic decrement method and directly connected to the bulk viscoelastic properties of the fiber. We further study the sensitivity of these measurements to changes in the chemo-mechanical properties of the outer coating layers on one of the synthetic fibers. To quantify the accuracy of our measurements with the torsional pendulum, a complementary series of tests are also performed on a strain-controlled rheometer in both torsional and tensile deformation modes.

Asphaltene Adsorption on Functionalized Solids.

Asphaltenes, heavy aromatic components of crude oil, are known to adsorb on surfaces and can lead to pipe clogging or hinder oil recovery. Because of their multicomponent structure, the details of their interactions with surfaces are complex. We investigate the effect of the physicochemical properties of the substrate on the extent and mechanism of this adsorption. Using wetting measurements, we relate the initial kinetics of deposition to the interfacial energy of the surface. We then quantify the long-term adsorption dynamics using a quartz crystal microbalance and ellipsometry. Finally, we investigate the mechanism and morphology of adsorption with force spectroscopy measurements as a function of surface chemistry. We determine different adsorption regimes differing in orientation, packing density, and initial kinetics on different substrate functionalizations. Specifically, we find that alkane substrates delay the initial monolayer formation, fluorinated surfaces exhibit fast adsorption but low bonding strength, and hydroxyl substrates lead to a different adsorption orientation and a high packing density of the asphaltene layer.

Multiscale Nature of Thixotropy and Rheological Hysteresis in Attractive Colloidal Suspensions under Shear.

Colloids with short range attractions self-assemble into sample-spanning structures, whose dynamic nature results in a thermokinematic memory of the deformation history, also referred to as "thixotropy." Here, we study the origins of the thixotropic effect in these time- and rate-dependent materials by investigating hysteresis across different length scales: from particle-level local measurements of coordination number (microscale), to the appearance of density and velocity fluctuations (mesoscale), and up to the shear stress response to an imposed deformation (macroscale). The characteristic time constants at each scale become progressively shorter, and hysteretic effects become more significant as we increase the strength of the interparticle attraction. There are also strong correlations between the thixotropic effects we observe at each scale.

Shear melting and recovery of crosslinkable cellulose nanocrystal-polymer gels.

Cellulose nanocrystals (CNC) are naturally-derived nanostructures of growing importance for the production of composites having attractive mechanical properties, and offer improved sustainability over purely petroleum-based alternatives. Fabrication of CNC composites typically involves extrusion of CNC suspensions and gels in a variety of solvents, in the presence of additives such as polymers and curing agents. Most studies so far have focused on aqueous CNC gels, yet the behavior of CNC-polymer gels in organic solvents is important to their wider processability. Here, we study the rheological behavior of composite polymer-CNC gels in dimethylformamide, which include additives for both UV and thermal crosslinking. Using rheometry coupled with in situ infrared spectroscopy, we show that under external shear, CNC-polymer gels display progressive and irreversible failure of the hydrogen bond network that is responsible for their pronounced elastic properties. In the absence of cross-linking additives, the polymer-CNC gels show an instantaneous but partial recovery of their viscoelasticity upon cessation of flow, whereas, the presence of additives allows the gels to recover over much longer timescale via van der Waals interactions. By exploring a broad range of shear history and CNC concentrations, we construct master curves for the temporal evolution of the viscoelastic properties of the polymer-CNC gels, illustrating universality of the observed dynamics with respect to gel composition and flow conditions. We find that polymer-CNC composite gels display a number of the distinctive features of colloidal glasses and, strikingly, that their response to the flow conditions encountered during processing can be tuned by chemical additives. These findings have implications for processing of dense CNC-polymer composites in solvent casting, 3D printing, and other manufacturing techniques.

Spontaneous wettability patterning via creasing instability.

Surfaces with patterned wettability contrast are important in industrial applications such as heat transfer, water collection, and particle separation. Traditional methods of fabricating such surfaces rely on microfabrication technologies, which are only applicable to certain substrates and are difficult to scale up and implement on curved surfaces. By taking advantage of a mechanical instability on a polyurethane elastomer film, we show that wettability patterns on both flat and curved surfaces can be generated spontaneously via a simple dip coating process. Variations in dipping time, sample prestress, and chemical treatment enable independent control of domain size (from about 100 to 500 µm), morphology, and wettability contrast, respectively. We characterize the wettability contrast using local surface energy measurements via the sessile droplet technique and tensiometry.

Scalable and durable polymeric icephobic and hydrate-phobic coatings.

Ice formation and accumulation on surfaces can result in severe problems for solar photovoltaic installations, offshore oil platforms, wind turbines and aircrafts. In addition, blockage of pipelines by formation and accumulation of clathrate hydrates of natural gases has safety and economical concerns in oil and gas operations, particularly at high pressures and low temperatures such as those found in subsea or arctic environments. Practical adoption of icephobic/hydrate-phobic surfaces requires mechanical robustness and stability under harsh environments. Here, we develop durable and mechanically robust bilayer poly-divinylbenzene (pDVB)/poly-perfluorodecylacrylate (pPFDA) coatings using initiated chemical vapor deposition (iCVD) to reduce the adhesion strength of ice/hydrates to underlying substrates (silicon and steel). Utilizing a highly-cross-linked polymer (pDVB) underneath a very thin veneer of fluorine-rich polymer (pPFDA) we have designed inherently rough bilayer polymer films that can be deposited on rough steel substrates resulting in surfaces which exhibit a receding water contact angle (WCA) higher than 150° and WCA hysteresis as low as 4°. Optical profilometer measurements were performed on the films and root mean square (RMS) roughness values of Rq = 178.0 ± 17.5 nm and Rq = 312.7 ± 23.5 nm were obtained on silicon and steel substrates, respectively. When steel surfaces are coated with these smooth hard iCVD bilayer polymer films, the strength of ice adhesion is reduced from 1010 ± 95 kPa to 180 ± 85 kPa. The adhesion strength of the cyclopentane (CyC5) hydrate is also reduced from 220 ± 45 kPa on rough steel substrates to 34 ± 12 kPa on the polymer-coated steel substrates. The durability of these bilayer polymer coated icephobic and hydrate-phobic substrates is confirmed by sand erosion tests and examination of multiple ice/hydrate adhesion/de-adhesion cycles.

Microstructural Rearrangements and their Rheological Implications in a Model Thixotropic Elastoviscoplastic Fluid.

We identify the sequence of microstructural changes that characterize the evolution of an attractive particulate gel under flow and discuss their implications on macroscopic rheology. Dissipative particle dynamics is used to monitor shear-driven evolution of a fabric tensor constructed from the ensemble spatial configuration of individual attractive constituents within the gel. By decomposing this tensor into isotropic and nonisotropic components we show that the average coordination number correlates directly with the flow curve of the shear stress versus shear rate, consistent with theoretical predictions for attractive systems. We show that the evolution in nonisotropic local particle rearrangements are primarily responsible for stress overshoots (strain-hardening) at the inception of steady shear flow and also lead, at larger times and longer scales, to microstructural localization phenomena such as shear banding flow-induced structure formation in the vorticity direction.