Colossal barocaloric effects in plastic crystals.

Refrigeration is of vital importance for modern society-for example, for food storage and air conditioning-and 25 to 30 per cent of the world's electricity is consumed for refrigeration1. Current refrigeration technology mostly involves the conventional vapour compression cycle, but the materials used in this technology are of growing environmental concern because of their large global warming potential2. As a promising alternative, refrigeration technologies based on solid-state caloric effects have been attracting attention in recent decades3-5. However, their application is restricted by the limited performance of current caloric materials, owing to small isothermal entropy changes and large driving magnetic fields. Here we report colossal barocaloric effects (CBCEs) (barocaloric effects are cooling effects of pressure-induced phase transitions) in a class of disordered solids called plastic crystals. The obtained entropy changes in a representative plastic crystal, neopentylglycol, are about 389 joules per kilogram per kelvin near room temperature. Pressure-dependent neutron scattering measurements reveal that CBCEs in plastic crystals can be attributed to the combination of extensive molecular orientational disorder, giant compressibility and highly anharmonic lattice dynamics of these materials. Our study establishes the microscopic mechanism of CBCEs in plastic crystals and paves the way to next-generation solid-state refrigeration technologies.

CO2-philicity to CO2-phobicity Transition on Smooth and Stochastic Rough Cu-like Substrate Surfaces.

CO2 on metal substrates is essential to CO2 liquefaction and transportation of CO2, yet the manipulation of the wettability of the CO2 and the elucidation of its underlying mechanism have not been fully achieved. Here, using molecular dynamics simulations, we report CO2 wetting characteristics on both smooth and stochastic rough Cu-like substrate surfaces. The results indicate that the apparent contact angle (CA) of the CO2 droplet on the smooth surface decreases from 180° to 0° as the CO2-solid characteristic interaction energy increases from 0.002 to 0.016 eV. In addition, the CAs become greater with increasing the density of surface asperities, regardless of the intrinsic surface wettability. This is attributed to the capillary drying-out of liquid CO2 molecules in gaps between surface asperities at the three-phase contact line of the droplet, which is usually overlooked in previous theoretical studies. Notably, the intrinsically CO2-philic surface transforms to the CO2-phobic due to an increase in the density of surface rugosity. Moreover, we verify the range of applicability of the CA prediction models concerning the nanoscale asperities. This work is beneficial for fully understanding the influence of nanoscale surface topography on CO2 wettability and shedding light on the design of functionalized and patterned surfaces to manipulate CO2 wettability.

Dissolution of Forsterite Surface in Brine at CO2 Geo-storage Conditions: Insights from Molecular Dynamic Simulations.

Evaluating the long-term security of geological deep saline aquifers to store CO2 requires a comprehensive understanding of mineral dissolution properties. Molecular dynamics simulations are performed to study the dissolution of forsterite in deep saline aquifers. The forsterite surface is found to be covered by three H2O molecular layers, hindering CO2 from directly contacting the surface. The dissolution rates at 350 K are increased by more than 1012 with the presence of Mg defects or salt ions in solutions. The more disordered surface in pure water caused by Mg defects accounts for the acceleration of dissolution, while absorbed Cl- ions on the surface in NaCl and KCl solutions accelerate the dissolution through electrostatic interactions. Comparatively, the frequent attacks from alkaline earth cations in MgCl2 and CaCl2 solutions to the surface contribute to the enhanced dissolution. In the acidic H3OCl solution, the electrostatic interactions between O atoms in H3O+ and the surface facilitate the dissolution. Interestingly, the ionic clusters of CO32-/HCO3- and Na+ in Na2CO3/NaHCO3 solution promote the dissolution process. This work provides molecular insights into forsterite dissolution in deep saline aquifers and guidance toward the optimization of CO2 geo-storage conditions.

Modeling the viscoelastic relaxation dynamics of soft particles via molecular dynamics simulation-informed multi-dimensional transition-state theory.

Viscoelastic soft colloidal particles have been widely explored in mechanical, chemical, pharmaceutical and other engineering applications due to their unique combination of viscosity and elasticity. The characteristic viscoelastic relaxation time shows an Arrhenius-type (or super-Arrhenius due to temperature-dependent transition attempts) thermally-activated behavior, but a holistic explanation from the relevant transition-state theory remains elusive. In this paper, the viscoelastic relaxation times of Lennard-Jones soft colloidal particle systems, including a single particle type system and a binary particle mixture based on the Kob-Andersen model, are determined using molecular dynamics (MD) simulations as the benchmark. First, the particle systems show a non-Maxwellian behavior after comparing the MD-predicted viscoelastic relaxation time and dynamic moduli (storage and loss modulus) to the classic Maxwell viscoelastic model and the recent particle local connectivity theory. Surprisingly, neither the Maxwell relaxation time τMaxwell (obtained from the static shear viscosity η and the high-frequency shear modulus G∞) nor the particle local connectivity lifetime τLC can capture the super-Arrhenius temperature-dependent behavior in the MD-predicted relaxation time τMD. Then, the particle dissociation and association transition kinetics, fractal dimensions of the particle systems, and neighbor particle structure (obtained from the radial distribution functions) are shown to collectively determine the viscoelastic relaxation time. These factors are embedded into a new multi-dimensional transition kinetics model to directly estimate the viscoelastic relaxation time τModel, which is found to agree with the MD-predicted τMD remarkably well. This work highlights the microscopic origin of viscoelastic relaxation dynamics of soft colloidal particles, and theoretically connects rheological dynamics and transition kinetics in soft matters.

Doping-Driven Wettability of Two-Dimensional Materials: A Multiscale Theory.

Engineering molecular interactions at two-dimensional (2D) materials interfaces enables new technological opportunities in functional surfaces and molecular epitaxy. Understanding the wettability of 2D materials represents the crucial first step toward quantifying the interplay between the interfacial forces and electric potential of 2D materials interfaces. Here we develop the first theoretical framework to model the wettability of the doped 2D materials by properly bridging the multiscale physical phenomena at the 2D interfaces, including (i) the change of 2D materials surface energy (atomistic scale, several angstroms), (ii) the molecular reorientation of liquid molecules adjacent to the interface (molecular scale, 100-101 nm), and (iii) the electrical double layer (EDL) formed in the liquid phase (mesoscopic scales, 100-104 nm). The latter two effects are found to be the major mechanisms responsible for the contact angle change upon doping, while the surface energy change of a pure 2D material has no net effect on the wetting property. When the doping level is electrostatically tuned, we demonstrate that 2D materials with high quantum capacitances (e.g., transition metal dichalcogenides, TMDCs) possess a wider range of tunability in the interfacial tension, under the same applied gate voltage. Furthermore, practical considerations such as defects and airborne contamination are also quantitatively discussed. Our analysis implies that the doping level can be another variable to modulate the wettability at 2D materials interfaces, as well as the molecular packing behavior on a 2D material-coated surface, essentially facilitating the interfacial engineering of 2D materials.

Low-Dimensional Organic Tin Bromide Perovskites and Their Photoinduced Structural Transformation.

Hybrid organic-inorganic metal halide perovskites possess exceptional structural tunability, with three- (3D), two- (2D), one- (1D), and zero-dimensional (0D) structures on the molecular level all possible. While remarkable progress has been realized in perovskite research in recent years, the focus has been mainly on 3D and 2D structures, with 1D and 0D structures significantly underexplored. The synthesis and characterization of a series of low-dimensional organic tin bromide perovskites with 1D and 0D structures is reported. Using the same organic and inorganic components, but at different ratios and reaction conditions, both 1D (C4 N2 H14 )SnBr4 and 0D (C4 N2 H14 Br)4 SnBr6 can be prepared in high yields. Moreover, photoinduced structural transformation from 1D to 0D was investigated experimentally and theoretically in which photodissociation of 1D metal halide chains followed by structural reorganization leads to the formation of a more thermodynamically stable 0D structure.

Ionic Effects on Supercritical CO2-Brine Interfacial Tensions: Molecular Dynamics Simulations and a Universal Correlation with Ionic Strength, Temperature, and Pressure.

For geological CO2 storage in deep saline aquifers, the interfacial tension (IFT) between supercritical CO2 and brine is critical for the storage security and design of the storage capacitance. However, currently, no predictive model exists to determine the IFT of supercritical CO2 against complex electrolyte solutions involving various mixed salt species at different concentrations and compositions. In this paper, we use molecular dynamics (MD) simulations to investigate the effect of salt ions on the incremental IFT at the supercritical CO2-brine interface with respect to that at the reference supercritical CO2-water interface. Supercritical CO2-NaCl solution, CO2-CaCl2 solution and CO2-(NaCl+CaCl2) mixed solution systems are simulated at 343 K and 20 MPa under different salinities and salt compositions. We find that the valence of the cations is the primary contributor to the variation in IFT, while the Lennard-Jones potentials for the cations pose a smaller impact on the IFT. Interestingly, the incremental IFT exhibits a general linear correlation with the ionic strength in the above three electrolyte systems, and the slopes are almost identical and independent of the solution types. Based on this finding, a universal predictive formula for IFTs of CO2-complex electrolyte solution systems is established, as a function of ionic strength, temperature, and pressure. The predicted IFTs using the established formula agree perfectly (with a high statistical confidence level of ∼96%) with a wide range of experimental data for CO2 interfacing with different electrolyte solutions, such as those involving MgCl2 and Na2SO4. This work provides an efficient and accurate route to directly predict IFTs in supercritical CO2-complex electrolyte solution systems for practical engineering applications, such as geological CO2 sequestration in deep saline aquifers and other interfacial systems involving complex electrolyte solutions.

Effect of sequence features on assembly of spider silk block copolymers.

Bioengineered spider silk block copolymers were studied to understand the effect of protein chain length and sequence chemistry on the formation of secondary structure and materials assembly. Using a combination of in vitro protein design and assembly studies, we demonstrate that silk block copolymers possessing multiple repetitive units self-assemble into lamellar microstructures. Additionally, the study provides insights into the assembly behavior of spider silk block copolymers in concentrated salt solutions.

Understanding selective molecular recognition in integrated carbon nanotube-polymer sensors by simulating physical analyte binding on carbon nanotube-polymer scaffolds.

Macromolecular scaffolds made of polymer-wrapped single-walled carbon nanotubes (SWCNTs) have been explored recently (Zhang et al., Nature Nanotechnology, 2013) as a new class of molecular-recognition motifs. However, selective analyte recognition is still challenging and lacks the underlying fundamental understanding needed for its practical implementation in biological sensors. In this report, we combine coarse-grained molecular dynamics (CGMD) simulations, physical adsorption/binding theories, and photoluminescence (PL) experiments to provide molecular insight into the selectivity of such sensors towards a large set of biologically important analytes. We find that the physical binding affinities of the analytes on a bare SWCNT partially correlate with their distribution coefficients in a bulk water/octanol system, suggesting that the analyte hydrophobicity plays a key role in determining the binding affinities of the analytes considered, along with the various specific interactions between the analytes and the polymer anchor groups. Two distinct categories of analytes are identified to demonstrate a complex picture for the correlation between optical sensor signals and the simulated binding affinities. Specifically, a good correlation was found between the sensor signals and the physical binding affinities of the three hormones (estradiol, melatonin, and thyroxine), the neurotransmitter (dopamine), and the vitamin (riboflavin) to the SWCNT-polymer scaffold. The four amino acids (aspartate, glycine, histidine, and tryptophan) and the two monosaccharides (fructose and glucose) considered were identified as blank analytes which are unable to induce sensor signals. The results indicate great success of our physical adsorption-based model in explaining the ranking in sensor selectivities. The combined framework presented here can be used to screen and select polymers that can potentially be used for creating synthetic molecular recognition motifs.

Tuning heterogeneous poly(dopamine) structures and mechanics: in silico covalent cross-linking and thin film nanoindentation.

Mussel-inspired synthetic poly(dopamine) thin films from dihydroxyphenylalanine (DOPA) and lysine, structurally similar to natural melanin, have drawn extensive interest as a versatile surface functionalization and coating material for use in a broad range of applications. In order to gain a better understanding of its complex and heterogeneous polymeric structure and mechanical properties, we report a computational model of poly(dopamine) by mimicking the polymerization process of the intermediate oxidized product of dopamine, 5,6-dihydroxyindole (DHI), via controlled in silico covalent cross-linking under the two most possible reaction schemes proposed in experiments. To validate our results using experiment, we synthesize poly(dopamine) thin films and perform experimental nanoindentations on the film. We observe an overall linear behavior for Young's modulus as a function of the degree of cross-linking, demonstrating the possibility of enhancing the mechanical robustness of poly(dopamine) materials by increasing the extent of polymerization. At the highest degree of polymerization considered (70%), the model mimics the linear tetrameric model for poly(dopamine) and melanin. At this degree of polymerization, we find a Young's modulus of 4.1-4.4 GPa, in agreement with our nanoindentation results of 4.3-10.5 GPa, previous experiments for natural melanin, as well as simulation results for the cyclic tetrameric melanin model (Chen et al., ACS Nano, 2013). Our results suggest that the non-covalent DHI aggregate model might not be appropriate to represent the structure of poly(dopamine) and melanin-like materials, since it gives a much smaller Young's modulus than the experimental lower bound. Our model not only nicely complements the previous computational work, but also provides new computational tools to study the heterogeneous structural and physicochemical properties of poly(dopamine) and melanin, as well as their formation pathways.

The effect of non-covalent functionalization on the thermal conductance of graphene/organic interfaces.

The intrinsic interfacial thermal resistance at graphene/organic interfaces, as a result of mismatches in the phonon vibrational spectra of the two materials, diminishes the overall heat transfer performance of graphene/organic nanocomposites. In this paper, we use molecular dynamics (MD) simulations to design alkyl-pyrene molecules that can non-covalently functionalize graphene surfaces in contact with a model organic phase composed of octane. The alkyl-pyrene molecules possess phonon-spectra features of both graphene and octane and, therefore, can serve as phonon-spectra linkers to bridge the vibrational mismatch at the graphene/octane interface. In support of this hypothesis, we find that the best linker candidate can enhance the out-of-plane graphene/organic interfacial thermal conductance by ~22%, attributed to its capability to compensate the low-frequency phonon mode of graphene. We also find that the length of the alkyl chain indirectly affects the interfacial thermal conductance through different orientations of these chains because they dictate the contribution of the out-of-plane high-frequency carbon-hydrogen bond vibrations to the overall phonon transport. This study advances our understanding of the less destructive non-covalent functionalization method and design principles of suitable linker molecules to enhance the thermal performance of graphene/organic nanocomposites while retaining the intrinsic chemical, thermal, and mechanical properties of pristine graphene.

Kinetic Mechanism of Surfactant-Based Molecular Recognition: Selective Permeability across an Oil-Water Interface Regulated by Supramolecular Aggregates.

Lipids are known to play a vital role in the molecular organization of all cellular life. Molecular recognition is another fundamental biological process that is generally attributed to biological polymers, such as proteins and nucleic acids. However, there is evidence that aggregates of lipids and lipid-like molecules are also capable of selectively binding to or regulating the partitioning of other molecules. We previously demonstrated that a model two-phase octanol/water system can selectively partition Red 40 and Blue 1 dyes added to an aqueous phase, with the selectivity depending on the surfactant (e.g., cetyltrimethylammonium bromide) dissolved in the organic phase. Here, we elucidate the mechanism of molecular recognition in this system by using quantitative partitioning experiments and molecular dynamics (MD) simulations. Our results indicate that the selectivity for the red dye is thermodynamically favored at all surfactant concentrations, while selectivity for the blue dye is kinetically favored at high surfactant concentrations. The kinetic selectivity for the blue dye correlates with the presence of molecular aggregation at the oil-water interface. Coarse-grained MD simulations elucidate nanoscale supramolecular structures that can preferentially bind one small molecule rather than another at an interface, providing a selectively permeable barrier in the absence of proteins. The results suggest a new supramolecular mechanism for molecular recognition with potential applications in drug delivery, drug discovery, and biosensing.

Thermal batteries based on inverse barocaloric effects.

To harvest and reuse low-temperature waste heat, we propose and realize an emergent concept-barocaloric thermal batteries based on the large inverse barocaloric effect of ammonium thiocyanate (NH4SCN). Thermal charging is initialized upon pressurization through an order-to-disorder phase transition, and the discharging of 43 J g-1 takes place at depressurization, which is 11 times more than the input mechanical energy. The thermodynamic equilibrium nature of the pressure-restrained heat-carrying phase guarantees stable long-duration storage. The barocaloric thermal batteries reinforced by their solid microscopic mechanism are expected to substantially advance the ability to take advantage of waste heat.

Molecular perspective on diazonium adsorption for controllable functionalization of single-walled carbon nanotubes in aqueous surfactant solutions.

Functionalization of single-walled carbon nanotubes (SWCNTs) using diazonium salts allows modification of their optical and electronic properties for a variety of applications, ranging from drug-delivery vehicles to molecular sensors. However, control of the functionalization process remains a challenge, requiring molecular-level understanding of the adsorption of diazonium ions onto heterogeneous, charge-mobile SWCNT surfaces, which are typically decorated with surfactants. In this paper, we combine molecular dynamics (MD) simulations, experiments, and equilibrium reaction modeling to understand and model the extent of diazonium functionalization of SWCNTs coated with various surfactants (sodium cholate, sodium dodecyl sulfate, and cetyl trimethylammonium bromide). We show that the free energy of diazonium adsorption, determined using simulations, can be used to rank surfactants in terms of the extent of functionalization attained following their adsorption on the nanotube surface. The difference in binding affinities between linear and rigid surfactants is attributed to the synergistic binding of the diazonium ion to the local "hot/cold spots" formed by the charged surfactant heads. A combined simulation-modeling framework is developed to provide guidance for controlling the various sensitive experimental conditions needed to achieve the desired extent of SWCNT functionalization.

Breakdown in the wetting transparency of graphene.

We develop a theory to model the van der Waals interactions between liquid and graphene, including quantifying the wetting behavior of a graphene-coated surface. Molecular dynamics simulations and contact angle measurements were also carried out to test the theory. We show that graphene is only partially transparent to wetting and that the predicted highest attainable contact angle of water on a graphene-coated surface is 96°. Our findings reveal a more complex picture of wetting on graphene than what has been reported recently as complete "wetting transparency."

Understanding the pH-dependent behavior of graphene oxide aqueous solutions: a comparative experimental and molecular dynamics simulation study.

Understanding the pH-dependent behavior of graphene oxide (GO) aqueous solutions is important to the production of assembled GO or reduced GO films for electronic, optical, and biological applications. We have carried out a comparative experimental and molecular dynamics (MD) simulation study to uncover the mechanisms behind the aggregation and the surface activity of GO at different pH values. At low pH, the carboxyl groups are protonated such that the GO sheets become less hydrophilic and form aggregates. MD simulations further suggest that the aggregates exhibit a GO-water-GO sandwichlike structure and as a result are stable in water instead of precipitating. However, at high pH, the deprotonated carboxyl groups are very hydrophilic such that individual GO sheets prefer to dissolve in bulk water like a regular salt. The GO aggregates formed at low pH are found to be surface-active and do not exhibit characteristic features of surfactant micelles. Our findings suggest that GO does not behave like conventional surfactants in pH 1 and 14 aqueous solutions. The molecular-level understanding of the solution behavior of GO presented here can facilitate and improve the experimental techniques used to synthesize and sort large, uniform GO dispersions in a solution phase.

Role of adsorbed surfactant in the reaction of aryl diazonium salts with single-walled carbon nanotubes.

Because covalent chemistry can diminish the optical and electronic properties of single-walled carbon nanotubes (SWCNTs), there is significant interest in developing methods of controllably functionalizing the nanotube sidewall. To date, most attempts at obtaining such control have focused on reaction stoichiometry or strength of oxidative treatment. Here, we examine the role of surfactants in the chemical modification of single-walled carbon nanotubes with aryl diazonium salts. The adsorbed surfactant layer is shown to affect the diazonium derivatization of carbon nanotubes in several ways, including electrostatic attraction or repulsion, steric exclusion, and direct chemical modification of the diazonium reactant. Electrostatic effects are most pronounced in the cases of anionic sodium dodecyl sulfate and cationic cetyltrimethylammonium bromide, where differences in surfactant charge can significantly affect the ability of the diazonium ion to access the SWCNT surface. For bile salt surfactants, with the exception of sodium cholate, we find that the surfactant wraps tightly enough such that exclusion effects are dominant. Here, sodium taurocholate exhibits almost no reactivity under the explored reaction conditions, while for sodium deoxycholate and sodium taurodeoxycholate, we show that the greatest extent of reaction is observed among a small population of nanotube species, with diameters between 0.88 and 0.92 nm. The anomalous reaction of nanotubes in this diameter range seems to imply that the surfactant is less effective at coating these species, resulting in a reduced surface coverage on the nanotube. Contrary to the other bile salts studied, sodium cholate enables high selectivity toward metallic species and small band gap semiconductors, which is attributed to surfactant-diazonium coupling to form highly reactive diazoesters. Further, it is found that the rigidity of anionic surfactants can significantly influence the ability of the surfactant layer to stabilize the diazonium ion near the nanotube surface. Such Coulombic and surfactant packing effects offer promise toward employing surfactants to controllably functionalize carbon nanotubes.

Hybridly double-crosslinked carbon nanotube networks with combined strength and toughness via cooperative energy dissipation.

Although chemical crosslinking has been extensively explored to enhance the mechanical properties of network-type materials for structural and energy (electrochemical, thermal, etc.) applications, loading-induced energy dissipations usually occur through a single channel that either leads to network brittleness or low strength/stiffness. In this work, we apply coarse-grained molecular dynamics simulations to explore the potential of hybridly double-crosslinked carbon nanotube (CNT) networks as a light weight functional material with combined strength and toughness. While increasing the crosslinking density or strong crosslink composition may, in general, enhance the strength and toughness, further increasing the two parameters would surprisingly lead to deteriorated strength and toughness. We find that double-crosslinked networks can nicely achieve cooperative energy dissipation with minimal structural damage. In particular, the weak crosslinks serve as "sacrificial bonds" to dissipate elastic energies from external loading, while the strong crosslinks act as "structure holders" and break at a much later stage during the tensile test. Therefore, the combination of more than one type of crosslinking with hybrid potential energy landscapes and breaking time scales can prevent premature simultaneous breaking of multiple strong crosslinks. By deploying intermediate amounts of weak and strong crosslinks, we observe an outstanding density-normalized strength of 227-2130 kPa m3 kg-1 as compared to many structural materials and advanced nanocomposites. The crosslinking strategies developed here would pave new avenues for the rational design of functional network materials beyond CNTs, such as hydrogels, nanofibers, and nanocomposites.

Molecular insights into the surface morphology, layering structure, and aggregation kinetics of surfactant-stabilized graphene dispersions.

The production of graphene with open band gaps for the manufacturing of graphene-based electronic and optical devices requires synthesis methods to either control the number of layers to enrich AB-stacked bilayer or trilayer graphene or control the extent of functionalization of monolayer graphene. Solution-phase dispersion of graphene is promising for both methods to create printable electronics and nanocomposites. However, both methods face common challenges, including controlling the surface morphology, reducing the turbostratic layering, and enhancing the dispersion stability. To address these challenges at the molecular level, we successfully combined molecular simulations, theoretical modeling, and experimental measurements. First, we probed the surface structure and electrostatic potential of monolayer graphene dispersed in a sodium cholate (SC) surfactant aqueous solution, which exhibits 2D sheets partially covered with a monolayer of negatively charged cholate ions. Similar to the case of carbon nanotube functionalization, one may regulate the binding affinity of charged reactants for graphene functionalization by manipulating the surface morphology. Subsequently, we quantified the interactions between two graphene-surfactant assemblies by calculating the potential of mean force (PMF) between two surfactant-covered graphene sheets, which confirmed the existence of a metastable bilayer graphene structure due to the steric hindrance of the confined surfactant molecules. The traditional Derjaguin-Landau-Verwey-Overbeek (DLVO) theory was found to be adequate to explain the long-range electrostatic repulsions between the ionic surfactant-covered graphene sheets but was unable to account for the dominant, short-range steric hindrance imparted by the confined surfactant molecules. Interestingly, one faces a dilemma when using surfactants to disperse and stabilize graphene in aqueous solution: on the one hand, surfactants can stabilize graphene aqueous dispersions, but on the other hand, they prevent the formation of new AB-stacked bilayer and trilayer graphene resulting from the reaggregation process. Finally, the lifetime and time-dependent distribution of various graphene layer types were predicted using a kinetic model of colloid aggregation, and each graphene layer type was further decomposed into subtypes, including the AB-stacked species and various turbostratic species. The kinetic model of colloid aggregation developed here can serve as a useful tool to evaluate the quality of graphene dispersions for subsequent substrate-transferring or functionalization processes.

Understanding the stabilization of liquid-phase-exfoliated graphene in polar solvents: molecular dynamics simulations and kinetic theory of colloid aggregation.

Understanding the solution-phase dispersion of pristine, unfunctionalized graphene is important for the production of conducting inks and top-down approaches to electronics. This process can also be used as a higher-quality alternative to chemical vapor deposition. We have developed a theoretical framework that utilizes molecular dynamics simulations and the kinetic theory of colloid aggregation to elucidate the mechanism of stabilization of liquid-phase-exfoliated graphene sheets in N-methylpyrrolidone (NMP), N,N'-dimethylformamide (DMF), dimethyl sulfoxide (DMSO), γ-butyrolactone (GBL), and water. By calculating the potential of mean force between two solvated graphene sheets using molecular dynamics (MD) simulations, we have found that the dominant barrier hindering the aggregation of graphene is the last layer of confined solvent molecules between the graphene sheets, which results from the strong affinity of the solvent molecules for graphene. The origin of the energy barrier responsible for repelling the sheets is the steric repulsions between solvent molecules and graphene before the desorption of the confined single layer of solvent. We have formulated a kinetic theory of colloid aggregation to model the aggregation of graphene sheets in the liquid phase in order to predict the stability using the potential of mean force. With only one adjustable parameter, the average collision area, which can be estimated from experimental data, our theory can describe the experimentally observed degradation of the single-layer graphene fraction in NMP. We have used these results to rank the potential solvents according to their ability to disperse pristine, unfunctionalized graphene as follows: NMP ≈ DMSO > DMF > GBL > H(2)O. This is consistent with the widespread use of the first three solvents for this purpose.