Cellulose NanofiberâGraphene Oxide Biohybrids: Disclosing the Self

www.acsnano.org

Cellulose Nanofiber−Graphene Oxide Biohybrids: Disclosing the Self-Assembly and Copper-Ion Adsorption Using Advanced Microscopy and ReaxFF Simulations Downloaded via UNIV OF CAMBRIDGE on August 20, 2018 at 18:28:40 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Chuantao Zhu,† Susanna Monti,‡ and Aji P. Mathew*,† †

Division of Materials and Environmental Chemistry, Stockholm University, Stockholm, 10691, Sweden CNR−Institute of Chemistry of Organometallic Compounds, Area della Ricerca, Via Moruzzi 1, 56124 Pisa, Italy

‡

S Supporting Information *

ABSTRACT: The self-assembly of nanocellulose and graphene oxide into highly porous biohybrid materials has inspired the design and synthesis of multifunctional membranes for removing water pollutants. The mechanisms of self-assembly, metal ion capture, and cluster formation on the biohybrids at the nano- and molecular scales are quite complex. Their elucidation requires evidence from the synergistic combination of experimental data and computational models. The AFM-based microscopy studies of (2,2,6,6-tetramethylpiperidine-1-oxylradical)-mediated oxidized cellulose nanofibers (TOCNFs), graphene oxide (GO), and their biohybrid membranes provide strong, direct evidence of self-assembly; small GO nanoparticles first attach and accumulate along a single TOCNF fiber, while the long, flexible TOCNF filaments wrap around the flat, wide GO planes, thus forming an amorphous and porous biohybrid network. The layered structure of the TOCNFs and GO membrane, derived from the self-assembly and its surface properties before and after the adsorption of Cu(II), is investigated by advanced microscopy techniques and is further clarified by the ReaxFF molecular dynamics (MD) simulations. The dynamics of the Cu(II)-ion capture by the TOCNF and GO membranes in solution and the ion cluster formation during drying are confirmed by the MD simulations. The results of this multidisciplinary investigation move the research one step forward by disclosing specific aspects of the self-assembly behavior of biospecies and suggesting effective design strategies to control the pore size and robust materials for industrial applications. KEYWORDS: TEMPO cellulose nanofibers, graphene oxide, biohybrids, self-assembly, atomic force microscopy, ReaxFF, molecular modeling

O

presence of many carboxyl and hydroxyl moieties. These are responsible for the outstanding adsorbent capacity of TOCNFs that is actuated through the coordination and capture of heavy metal ions and organic pollutants.12−14 Similarly to TOCNFs, GO sheets with hydroxyl, carboxyl, and epoxy groups decorating the basal planes and edges15 display large surface areas, high surface activity, good hydrophilicity, partial hydrophobicity in the graphene domains, and antifouling properties.15−17 Practically, the GO sheets are amphiphilic systems whose action, when adsorbed on surfaces, depends on the morphology of the sheets, the binding strength, and the

ver the past few years, nanocellulose- and graphenebased materials have attracted significant attention in numerous intriguing applications as adsorbents and separation materials.1−3 The typical derivatives of cellulose and graphene, namely, (2,2,6,6-tetramethylpiperidine-1-oxylradical) (TEMPO)-mediated oxidized cellulose nanofibers (TOCNFs) 4,5 and graphene oxide (GO) 6,7 are even considered among the most promising “capturing agents”, and this status is why they have been selected as excellent candidates for fabricating sustainable adsorbents, such as biohybrid aerogels,8 smart composites,9 and eco-friendly membranes.10,11 As a matter of fact, TOCNFs, modified by the positionselective catalytic oxidation of the C6 primary hydroxyls from nanocellulose, possess active-surface characteristics due to the © 2018 American Chemical Society

Received: April 12, 2018 Accepted: June 11, 2018 Published: June 11, 2018 7028

DOI: 10.1021/acsnano.8b02734 ACS Nano 2018, 12, 7028−7038

Article

Cite This: ACS Nano 2018, 12, 7028−7038

Article

ACS Nano

Figure 1. AFM and modeling images showing the self-assembly of TOCNFs and GO. (a) The GO sheets (square area) and large GO nanoparticles (circle area) are embedded in the biohybrid network. (b) Thickness of the GO sheets and nanoparticles in the biohybrids. (c) The small GO nanoparticles are attached and accumulated along TOCNFs from a single fiber scale, with highlights in the square area. The hydrogen bonds connecting the molecules to the TOCNF hydroxyls and carboxyl groups are highlighted (green lines). The TOCNF is represented by a solvent-accessible surface, where the pink regions identify the TOCNF oxygens. Water and ions have been removed for clarity. (d) Flat adsorption with separate molecules. (e) Interconnected sheets adsorbed almost perpendicularly on a TOCNF extending toward a periodic image.

copper on TOCNFs were discussed earlier,20 evidencing the increasing capacity of the ion capture when the TOCNF carboxylate content was increased. We have earlier shown through atomic force microscopy (AFM), X-ray photoelectron spectroscopy, and attenuated total reflection−Fourier transform infrared spectroscopy, the enhanced adsorption of cellulose with a monovalent silver ion cellulose surface by means of sulfate and phosphate groups and computational studies confirmed the possibility of cluster formation and the stability of the silver clusters. As already evidenced in a number of studies, computational methods such as all-atom classical reactive molecular dynamics (RMD) simulations can be very useful to interpret and predict the experimental results.22−30 Indeed, they can follow the evolution of relatively large portions of the materials for relatively long times, realistically reproduce the system’s behavior in elaborate environments by mimicking the experimental setup during all stages of metal/pollutant adsorption, and simulate bond breakage and formation under specific circumstances created by local effects. Given these premises, in this work, we combined the experimental characterization and used advanced microscopy techniques with simulations based on force fields to scrutinize the ability of TOCNFs and GO at the atomic level to assemble and cooperatively capture copper ions in solution. The modeling was carried out by building a series of representative supramolecular models, inspired by the data available in the literature or kindly provided by the authors (see computational details and related references) and by exploring their dynamics through the state-of-the-art RMD simulations. The RMD runs were based on the ReaxFF approach.31 The force field employed was appropriately parametrized to describe these systems and was already used effectively in previous investigations dealing with cellulose, graphene, graphene oxide, and metal ions in various environmental conditions.22−30

multiscale assembly. The synergistic combination of TOCNFs and GO, which is realized through favorable interactions between the abundant hydrogen donors and acceptors, generates an active porous self-assembled biohybrid structure.18,19 These membranes are very promising materials not only for their adsorption efficiency and high capacity of retention of heavy metals and organic dyes present in solution but also for their remarkable mechanical properties and recyclability.19 All these properties are strongly related to the conformation, supramolecular organization, self-assembly, and combination of the various components, which are all driven by intermolecular interactions. These can be, for example, intra- and intermolecular hydrogen bonds and van der Waals and electrostatic forces. Although several experimental investigations have tried to disclose the molecular organization of the various components inside the matrices and the intermolecular interactions responsible for aggregation, adsorption, and entrapment, through different techniques,19,20 the collected data are not detailed enough to give a clear picture of all these behaviors, and many questions still remain unanswered. This status is due to the structural and morphological complexity of the hybrid membranes, which can, for example, consist of crystalline and amorphous domains randomly organized in varying proportions and to the dispersed/attached components.21 In a previous paper,19 we have investigated experimentally the self-assembly of TOCNF-, GO-, and nanoGO-based biohybrids, disclosing their unique porous structures before and after adsorption of copper ions. The nanoGO sheets acted as linking agents, improving the porosity of the material and the water flow, whereas functionalized cellulose played both roles of cross-linker and adsorbent. We demonstrated that these hybrid combinations are efficient adsorbents that could be tuned by modulating their adsorption capacity, flexibility, wet stability, and mechanical properties. Furthermore, surface adsorption of copper ions and formation of nanoparticles of 7029


Article

ACS Nano

Figure 2. Surface morphology and cross-section of a TOCNF+GO membrane. (a) AFM images show the GO sheets embedded inside and on top of the TOCNF+GO membrane. (b) Magnified area of the GO sheet embedded inside the TOCNF+GO membrane from the top square area. (c) Magnified area of the GO sheet on top of the TOCNF+GO membrane from (a), bottom square area. (d) GO nanoparticles randomly embedded or floating on the membrane. (e) Cross-section of the TOCNF+GO membrane, with red dotted line indicating the micropores. (f) Cross-section of the TOCNF+GO membrane, with red square showing the layered structure. (g, h) Dried structure where the GO sheets adsorbed on TOCNF cross-link the fibrils, self-interact, and determine pore size, shape, and position inside the nanocellulose matrix. (g) x−y periodicity (replicas due to periodic boundary conditions in all dimensions): the pores are infinite canals running along the z-axis (black regions). (h) y−z periodicity: stripes of stacked GO structures delimit the canals, which could be filled with water and ions.

The self-assembly between TOCNFs and GO is demonstrated in Figure 1, which exhibits a homogeneous network structure of TOCNF and GO biohybrids. The GO sheets with a thickness of approximately 1 nm either lie flat or embed in the TOCNF skeleton (Figure 1a,b). The big GO nanosheets were found either covering the TOCNF network or partially embedded in it (see Figure 1a, circled area). The selfassemblies of both GO sheets and nanoparticles with TOCNFs are consistent with our early report.19 Moreover, Figure 1c provides direct, strong evidence of the self-assembly between TOCNFs and smaller GO nanoparticles (several nanometers), which, as far as we know, has never been detected and reported before. Small GO nanoparticles with the 2−3 nm sizes are attached and accumulated along the TOCNF fibers (Figure 1c). This assembly is not only due to the physical interaction between the components in the hybrids but also due to effective chemical interactions/cross-linking between the various components. The RMD simulations confirm the tendency of GO to selfassemble both in solution and when in contact with the TOCNF fibers (Figure 1d,e). The interactions are essentially the intermolecular hydrogen bonds between the hydroxyl and carboxyl moieties at the edges of the sheets or functionalities of the basal planes, which determine the formation of stacked structures and nanoparticles, depending on the degree of basal plane bending. Furthermore, the simulations suggested that the type of aggregation was mainly due to the local concentration of the molecules and to the interference of the surrounding species.

This study integrates experiments and simulations in each phase of the sample treatment, namely, the preparation of TOCNF- and GO-based biohybrids (100:1, wt %), characterization of self-assemblies, depiction of the membrane structures, and evaluations of surface properties due to copper-ion adsorption. The results are satisfactorily consistent and support the importance of a multidisciplinary approach to address the challenging task of tailoring the structural features of biofilters to enhance the metal-ion uptake and organic pollutant trapping.

RESULTS AND DISCUSSION Self-Assembly of TOCNF and GO Biohybrids. The systematic investigation of morphologies and sizes of TOCNFs and GO sheets/nanoparticles is crucial to the studies of the biohybrid self-assembly. They were investigated by AFM imaging and are shown in Figure S1. Dispersed TOCNFs are visible in Figure S1a. The diameter of TOCNFs is 2−3 nm, with the length from tens of nanometers to several micrometers. The GO sheets and nanoparticles usually have broad size distributions due to the random chemical exfoliation process in the GO mass production.32,33 In the present case, the GO size varies in a wide range, as shown from the AFM imaging in Figure S1b,c (Supporting Information). Considering that the thickness of one layer of GO sheets is approximately 1 nm, the stacking sheets can reach a thickness of several nanometers (Figure S1b). Figure S1c shows some randomly dispersed GO nanoparticles, with the size varying from 1 to 30 nm, indicating that the GO suspension contains a wide size range of both GO sheets and nanoparticles. 7030


Article

ACS Nano

relatively low strength of the adhesion of the GO sheets in the sense that they could slightly change their orientations on the TOCNF surface. The new positions of the sheets were essentially related to the morphology of the TOCNF facets and to the presence of other GO units on the fiber. Indeed, the self-assembly through intersheet hydrogen bonds (Figure 2h) was observed in different cases. Although self-organization could prevent a strong adherence of the sheets to the fiber surface and favor their partial extension in solution, it was noted that such arrangements could modulate the matrix porosity and generate efficient ion-trapping cages. Essentially, the surface roughness increased, and new capturing cavities were created. Starting the simulations from the separate initial arrangements of the GO sheets parallel to the TOCNF surface produced a dominant adsorption mode, where the sheets landed almost flat on the support. However, among these final arrangements, inclined and partially adsorbed structures were also present. These resulted essentially from self-interactions. In particular, intermolecular interactions between molecules connected to different fibers (periodic images) generated partial adsorbents, inclined and attached to the TOCNF through the edge groups (Figure 2g). Such a scenario effectively illustrates how the nanofibrils could be cross-linked and provides a possible cause of the formation of relatively rich pore structures (Figure 2h). The action of the GO molecules can be further explained by focusing on the fibril separation at the beginning and at the end of the dynamics runs. In the simulated models, the initial separation between the fibrils in the y direction was approximately 22 Å, whereas along the z direction it was even smaller (16 Å). The presence of GO modulated the effective fiber distance, reducing it to less than 10 Å in the narrower region. As a matter of fact, the gap size is also correlated to the water content and can be reduced further when the composite (or membrane) is dried. Indeed, more direct evidence of the effects produced by the GO action on TOCNF packing was obtained when the membranes were consolidated by removing water from the simulation box (as done experimentally, drying process) and by equilibrating the configurations (at constant pressure and temperature) to their bulk geometries (Figure 2g,h). The consequences soon became apparent: the box volume was drastically reduced by approximately 65%, which corresponded to an approximately 30−40% increase in the volume of the isolated TOCNF (to be compared with the fibril packing). This new material was scattered with relatively large cavities, where the size, shape, and distribution were controlled by the location and assembly of the GO units (Figure 2g,e). The maximum size of the pores observed in the simulated models was approximately 30 Å, estimated along the x-axis, and 15 Å along the z-axis (Figure S3). These dimensions could be further reduced by drying the system at high pressure and then re-equilibrating it at ambient pressure (Figure S4). In this case, the convergence toward an equilibrium structure was faster, and the resulting material was more compact, with a tighter molecular network and lower porosity due to the deformation of the adsorbate conformations. Metal Ion Capturing and Clustering. With regard to the adsorption capability of the membrane toward metal-ion capture and clustering, the computational data analysis was carried out by comparing the results of different simulations in aqueous solution containing ions, GO+ions, TOCNF+ions,

In fact, during the simulations, it was noticed that even though the initial positions of the GO molecules were almost parallel to the fibril surface, some of them were displaced from their flat orientation by the solvent action and by the attraction of the nearby fibers. The resulting reorganization was a rotation of the whole GO sheets and their extension toward the adjacent TOCNF fibril. Essentially, GO acted as a crosslinker and induced the formation of large fibril bundles. The bonding between neighboring GO sheets could be enhanced by TOCNF because of the accessible active groups on the fibril surface that provide extra bonding options.23 A single TOCNF could also interact with the functional groups and lie on the flat surface of large GO sheets, forming layered structures. Thus, the experiments supported by the computational results suggest that the self-assembly starts from the interactions between the small GO nanoparticles/sheets and TOCNF fibers, followed by the interactions with the long flexible TOCNF filaments and the flat, wide GO planes that wrap around each other forming amorphous porous biohybrid networks. Dynamic Aspect of the Self-Assembly in GO and TOCNF Membranes. The self-assembled TOCNF and GO biohybrids were converted to a membrane via vacuum filtration and characterized with AFM experiments and molecular dynamic simulations. The surface and cross-section of the membrane are displayed in detail in Figure 2. From the AFM characterization shown in Figure 2a, it is apparent that, as expected, the GO sheets are located on the TOCNF+GO membrane surface. A few definite local arrangements and motifs can be identified. For example, in Figure 2a,b, it is visible that a triangular GO sheet is totally wrapped in a TOCNF and thus embedded inside the membrane. Another GO sheet was found closer to the top of the membrane covering other TOCNF, leading to a blurred area shown in the square in Figure 2c, whereas GO nanoparticles surrounded and wrapped by TOCNF fibrils are also observed on the surface of the TOCNF+GO membrane (Figure 2d). The SEM analysis of the membrane cross-section (Figure 2e,f) reveals a layered structure with small micropores and hollow regions between the layers. The micropores are marked with red dashed lines in Figure 2e. The pure TOCNF membrane displays a layered structure as well, but in this case, pores or spaces between the layers are hardly discernible (Figure S2a,b). This outcome can be explained by the tight packing arrangement of the TOCNF due to its homogeneous nanoscale features. The addition of GO can disrupt the tight packing and create a loose network with increased porosity. In our recent studies on CNF/GO layered membranes, we have also observed that “standing inserted” GO sheets at a GO content as low as 1 wt % significantly increased the permeability of the membranes.34 Interesting results, related to the self-assembling ability of the GO species on a TOCNF, emerged from our MD trajectories and were thus discussed in relation to the corresponding experimental counterpart. Indeed, as already mentioned, the attractive interaction between TOCNFs and GO was essentially due to the formation of a dense network of intermolecular hydrogen bonds involving mainly the hydroxyl groups of the two species. This network was sometimes reinforced by the involvement of carboxyl groups populating the edges of the sheets and different regions of functionalized TOCNF chains. The dynamic nature of the hydrogen bonds (i.e., bond breaking and formation) was responsible for the 7031


Article

ACS Nano

solvent oxygens (belonging to the water or sulfate coordination spheres). Only in a few clusters were the ions accommodated at relatively short distances (4.5−5.2 Å), but these types of aggregations were rarely observed (approximately 2%). The copper coordination trend is clearly shown in Figure 3 (righthand side), where the atom−atom Cu−X radial distribution functions are displayed. The blue triangles represent the Cu− Cu distances, whereas the red diamonds depict the coordination of copper to the sulfate counterions. All Cu− Cu plots show a similar behavior except the curve at high concentration, which corresponds to the TOCNF+GO+32 ions model. The presence of relatively sharp peaks at shorter distances suggests that due to the increased population of copper ions, the TOCNG+GO capturing sites become closer, and the ions are found at shorter distances entrapped in the matrix network (Figures S6 and S7). A visual examination of the copper−oxygen coordination revealed that this could be realized through distorted tetragonal, octahedral, square pyramidal, and trigonal bipyramidal geometries with average Cu−O separations of approximately 2.0 Å (Figure 3, left-hand side). The running coordination number of Cu surrounded by a first solvation shell of water molecules is approximately 5.6, in line with the literature,35,36 whereas the second hydration layer is located at shorter distances, approximately 3.9−4.3 Å, than those in other studies (i.e., 4.55−4.72 Å)36 because of the presence of the sulfate counterions. The latter, often coordinated with copper but also with each other, was found connected to the surface hydroxyl and carboxyl groups of the TOCNF through hydrogen-bonding interactions. Another common trait shown by all Cu−O radial distribution functions (RDFs) is that an exchange between the molecules of the first and second coordination shells is almost absent within the simulated time (and for the chosen sampling step). More evidence of clustering and metal ion immobilization on the biohybrids is provided by the analysis of the spatial distribution functions (SDFs) of the ions around the simulated models. These models are the three-dimensional maps of isosurfaces that give the highest probability of finding the selected species at specific locations. Two representative SDFs of the copper and sulfur atoms for TOCNF+GO+ions at low and high concentrations are shown in Figure 4a,b with the density contours 1.5 times larger than the average solvent density. The structure displayed in the SDF plots is an average configuration calculated from the last portion of the trajectory. The maximum density regions are marked with solid orange (Cu) or yellow (S, as a representative atom of the counterion) spheres. As far as the connection of Cu with the TOCNF and GO oxygens is concerned, the inspection of the RDFs displayed in Figure 4 suggests that Cu is well-anchored on the cellulose surface through the cooperation of the hydroxyl and carboxyl groups of both TOCNFs and GO. However, a preference toward carboxyl oxygens is observed. This outcome is highlighted by the first peaks at 1.8−2.0 Å visible in all plots. Considering the coordination number, most of the carboxyl− Cu interactions are the result of TOCNFs. At high ion concentrations, the adsorbed ions have the tendency to form clusters. In contrast, in the simulations of six GO molecules in solution, it was found that GO were slightly interconnected but not stacked on top of each other, and the propensity of the copper ions was to remain farther at approximately 4.5 Å from the GO capturing groups but still bind through the hydrogen

and TOCNF+GO+ions at low (16 ions) and high (32 ions) concentrations. Even though at the beginning of the simulations all ions were well-separated from one another, the sampling time was sufficiently long to disclose their binding to the supports and their limited self-aggregation. This was already apparent from the simulation of the ions in aqueous solution where small mixed aggregates made, at most, of five units were identified (Figure 3). The overall tendency of copper ions at low concentration was to maintain long separations, be surrounded by water molecules and sulfate ions, and be connected/adsorbed to/on the supports while preserving a variable amount of coordinated

Figure 3. Left: Representative ionic clusters extracted from the simulation in aqueous solution. The copper ions (orange spheres) are surrounded by oxygen (red) and sulfur (yellow) atoms. Water molecules and all hydrogens are not displayed for clarity. Right: Atomic Cu−X radial distribution functions: (a) 16 CuSO4 ions in water; (b) 6 GO molecules+16 CuSO4 ions; (c) TOCNF+16 CuSO4 ions; (d) TOCNF+6GO+16 CuSO4 ions; (e) TOCNF +6GO+32 CuSO4 ions. X: O(TOCNF) = all oxygens of the fibril; O(SO4 & WAT) and O(WAT & SO4) are water oxygens and the oxygens of the SO4 ions. This distinction has been introduced just because in ReaxFF the connection table is recalculated at each iteration and, even though the SO4 oxygens were initially “labelled”, in principle, they could be replaced by water oxygens and vice versa; O(GO) all oxygens of GO. 7032


Article

ACS Nano

Figure 4. On the left-hand side are shown the SDFs of copper (orange) and sulfur (yellow) atoms for TOCNF+GO+ions at low (a) and high (b) concentration. The orange and yellow regions represent the density contours 2 times larger than the average solvent density. The maximum density is highlighted by spheres. The TOCNF (solvent-accessible surface where the oxygens are light pink) and GO molecules (cyan sticks with red oxygens) are average reconstructed conformations. On the right-hand side the atomic Cu−X radial distribution functions are displayed. (c) TOCNF+GO+16 ions. (d) TOCNF+GO+32 ions. X: O(TOCNF) are all oxygens of the fibril; O(GO) all oxygens of GO. This distinction is made because in ReaxFF the connection table is recalculated at each iteration and the oxygen initially connected to SO4 can become a water oxygen and vice versa. S is the sulfur atom.

final configurations of TOCNF+GO+ions were obtained after drying the solvated models extracted from the dynamics. The drying process was performed by removing all water molecules except those belonging to the first coordination sphere of the copper ions (top) or all water molecules (bottom). In each picture (top and bottom), three sides of the final box, namely, XY (top-left), ZY (top-right), and XZ (bottom), are shown as representing the molecular/ionic arrangements. Apparently, a more packed structure with small pores is obtained when a few water molecules remain entrapped between the fibrils, and GO can adopt more relaxed stacked conformations laid on the fibril surface, whereas in the other case, the formation of larger pores seems induced by a different arrangement adopted by the sheets, which are more bent and inclined in relation to the fibers. The experimental evidence confirms this scenario, suggesting that the stationary phase is characterized by a TOCNF +GO membrane covered with a layer of nanoparticles and a few nanoclusters (Figure 6a,b,c). In particular, in Figure 6d, the circled area comprises a square-shaped particle that resembles a GO sheet combined with copper nanoclusters supported on the membrane. In this case, a detailed characterization is impossible due to the size of GO that partially envelops the metal ions. Further evidence of this phenomenon is provided by the AFM video images, which were used for monitoring the sample and probe movement (Figure S2d,e). They exhibit bright iridescence from the TOCNF+GO membrane without Cu(II) adsorption, which is mainly due to the reflection of GO on the membrane surface under illumination during AFM measurements (Figure S2c). After Cu(II) adsorption, the iridescence becomes much weaker, which is due to the copper nanolayer and nanoclusters covered on the surface of the TOCNF+GO membrane, shading the GO reflection under illumination (Figure S2d). Moreover, the SEM-EDS results consolidate the conclusion

bonds between water molecules of the first coordination shell and the hydroxyl/carboxyl groups of GO (Figure 4d). As shown in Figure 4a,b, the maximum density zones of Cu (orange contours) are all located on top of the TOCNF surface, and sometimes, the Cu atoms are found entrapped between the TOCNF interface and the GO sheets (Figure 4a). Most of the lower density regions are distributed near the TOCNF surface, as well, and only a few Cu ions remain in solution. This outcome clearly indicates that not all of the copper ions were adsorbed onto the supports within the simulated time, and some remained in solution coordinated to their corresponding counterions (Figure 4a,b). It is interesting to observe that the copper-ion distributions are tighter than the sulfur ones because their tendency is to accommodate stably on the support. The stability depends on the type of oxygens involved in the interaction and on the steric capture mechanism induced by the compressing action of the GO sheets. The most stable and persistent attachments comprise carboxyl oxygens. In contrast, sulfur distributions are sparser and spread, suggesting that the SO4 moieties tend to follow the motion of water and hydrated copper ions. The latter ones are more mobile when adsorbed as CuSO4 clusters, especially if their orientation is toward the solvent (Figure S6b), and they often travel from one site to another one in search of a steady state. The alignment of copper densities crossing the fiber, visible in Figure 4, implies the formation of a chain of adsorbed Cu ions (as shown in Figures S5 and S6), which are not strongly connected to the interface because the interaction does not seem to involve carboxyl oxygens. Essentially, their motion is regulated by the reorientation of the hydroxyl moieties of TOCNFs, which keeps them connected to the surface. At high copper concentration, the formation of mixed (containing SO4 ions) clusters on the TOCNF surface was observed in dry conditions (Figure 5) Different views of the 7033


Article

ACS Nano

hydroxyl and carboxyl groups of both the TOCNF and the GO components. Preferential interactions mainly involve the carboxyl oxygens and determine stronger adsorptions. It was observed that the tendency to stably accommodate on the support also depends on the steric capture mechanism due to the compressing action of the GO sheets on the fiber surface. The formation of copper clusters on the fibers is also reproduced, and their fiber cross-linking action is shown. These outcomes perfectly agree with the experimental findings. This combined experimental−theoretical investigation successfully clarified, at the nanoscale, the structural and dynamical aspects of the biohybrids that were not yet well-characterized by the studies present in the literature, thus providing the guidelines to understanding the behavior of cellulosic- and GO-based biohybrids for potential applications as ultralight aerogels,37 energy and memory storage materials,38 hydrogels, tissue engineering scaffolds,39 and robust supercapacitors.40

MATERIALS AND METHODS Experimental Section: Materials. The TEMPO-oxidized cellulose nanofibers (1.0 mmol/g, 1 wt %) were prepared and provided by Liu et al,41 Stockholm University, Sweden. Graphene oxide (4 mg/mL), copper(II) sulfate pentahydrate (99.999%), and (3-aminopropyl)triethoxysilane (APTES) were all purchased from Sigma-Aldrich, Germany, and used as received. Distilled water was used as the dispersion medium. The ScanAsyst Air probe was purchased from Bruker, Santa Barbara, USA, and was applied for AFM characterization. Preparation of TOCNF and GO-Based Biohybrid Membranes and Cu(II) Adsorption. A suspension of 1 g/L TOCNF +GO (100/1, wt %) in 200 mL was prepared by mixing 20 g of TOCNFs and 0.5 mL of GO suspension and dilution with distilled water. A 5 mL TOCNF+GO suspension was vacuum-filtered after sonication for 3 min, using a 0.45 μm pore-size filter membrane (DVPP, Millipore) under a filtration pressure of 1 bar. Two TOCNF +GO membranes were fabricated, and one of the membranes was immersed in Cu(II) solution for 30 min and then filtered and rinsed by distilled water. The pure TOCNF membrane was prepared by the same method as the control sample. The TOCNF+GO membranes with and without adsorption of Cu(II) were air-dried and investigated by microscopy and spectroscopy techniques on both the surface and cross-sections. Characterizations. Atomic Force Microscopy (AFM). The topography of the samples with and without copper adsorption was performed with a Dimension Icon AFM (Bruker, Nanoscope controller, Santa Barbara, CA, USA). The TOCNF and GO dispersed samples were prepared as in our earlier study.19 The TOCNF, GO, and TOCNF+GO (100:1, wt %) suspension was drop-coated on the APTES-modified mica surface on the AFM metal disc and was used to characterize the structure of the nanoparticles and the biohybrids. The TOCNF+GO (100:1 wt %) membranes with and without Cu(II) were directly attached to the AFM metal disc using AFM doublesided tape. The probes (model: ScanAsyst-air, Bruker) were treated with a UV ozone cleaner (ProCleaner Plus, US) for 20 min before scanning. The height, peak-force error images were recorded under ScanAsyst-air mode. The diameter of the TOCNF fiber and the thickness of GO were postprocessed with NanoScope Analysis 1.5 software (Bruker), using the section function by measuring the data perpendicular and laterally to the fiber length direction and GO, respectively. The measured values are directly marked in Figure S1. Scanning Electron Microscopy−Energy Dispersive X-ray Spectroscopy (SEM-EDS). The morphologies of the pure TOCNF crosssection19 and TOCNF+GO (100:1 wt %) membranes were observed using SEM on a JEOL JSM-7401F microscope (Japan). The membranes were cut and anchored at 0 and 90 deg to the holder to check the surface and cross-sections. The samples were sputtercoated with gold for 20 s before measurements to avoid the burning/ damaging to the samples during measurements. Images were taken at

Figure 5. Final configurations, dried and packed. In the top image all water molecules were removed except those belonging to the first coordination sphere of the copper ions (see Figure S5 for a better view of the small pores). In the bottom image all water molecules were removed. In each picture the XY (top-left), ZY (top-right), and XZ (bottom) points of view are shown. Apparently, a more packed structure with small pores is obtained when a few water molecules remain entrapped between the fibrils and GO can adopt more relaxed stacked conformations laid on the fibril surface, whereas in the other case the formation of larger pores seems induced by a different arrangement adopted by the sheets, which are more bent and inclined in relation to the fibers. Periodic images are displayed to render more clearly the coverage.

that Cu(II) ions were successfully adsorbed by the biohybrid membrane (Figure 6g).

CONCLUSIONS Advanced microscopy techniques revealed the self-assembly behavior and the porous structure in TOCNF and GO networks, as well as their biohybrid membrane. The MD simulations significantly clarified the nature of aggregation and pore formation inside the biohybrid membrane, which seemed related to the degree of GO basal plane bending, to the action of the solvent, to the intermolecular H-bonds between the GO sheets, and, finally, to their location in the fiber skeleton. It was further demonstrated that copper ions are captured inside the biohybrid network through cooperative coordination of the 7034


Article

ACS Nano

Figure 6. Surface morphology and modeling structure of a TOCNF+GO membrane after adsorption of Cu(II) under different magnifications. (a, b, c) Height images. (d, e, f) Corresponding peak force images. The red arrows display the magnifications of the square areas. The circles denote the nanolayer and nanoparticles on the surface of the membrane. (g) Spatial distribution functions of copper (orange) and sulfur (yellow) atoms surrounding the TOCNF+GO model at high concentration. The orange and yellow regions represent density contours 2 times larger than the average solvent density. TOCNF (solvent-accessible surface where the oxygens are light pink) and GO molecules (cyan sticks with red oxygens) are average conformations (reconstructed). rendered through a periodic supercell, with chains covalently bonded across the x border, whereas the CNF packing was controlled by adjusting the dimensions of the simulation box in the y and z directions. The crystal structure of Iβ cellulose determined by X-ray and neutron diffraction44 was employed to create the 16 CNF chains made of 16 glucosyl residues each, which were aligned along the x direction of the supercell. According to the literature,42−44 our previous experience on fibril structures,45,46 and present experimental proportions, this size is a good compromise between the computational cost and a meaningful comparison with the experimental data.

2 kV and approximately 10 mm working distance. The EDS spectra were obtained at 15 kV and 10 mm working distance with the accumulation time of 2 min, which was used to confirm the elements in the sample. Computational Section: Preliminary Model Building. To characterize in detail the ability of cellulose nanofibrils (CNFs) and GO to capture copper ions in solution and their expected cooperation, we built a series of representative supramolecular models inspired by the data available in the literature or kindly provided by the authors.23,42,43 The infinite length of CNFs was 7035


Article

ACS Nano A reasonable configuration of GO (28 carbon rings), in line with the CNF model, was downloaded from the Automated Topology Builder and Repository site47 and modified by replacing eight of the edge hydrogen atoms with carboxyl groups (Figure S1a). The new molecule was quantum-mechanically energy optimized at the DFT/ M062X/6-31G(d) level, and the resulting structure was used as a starting conformation in all simulations. The choice of the FF was based on suggestions found in the literature48 where a comparison among three different reactive FFs is reported. Considering all the aspects commented there, we decided that for the type of simulations we were to perform the best choice was the force field developed by Chenoweth et al.49 and made available in the ADF package under the name HCO.ff. SO42− parameters were taken from ref 50, where the authors carried out an MC optimization of the MgSO4 hydrates (the force was released with the name HOSMg.ff−ADF). The copper parameters were extracted from CuCl-H2O.ff: (Cu/Cl/H/O),29 and a direct bond with sulfur atoms was not parametrized based on the hypothesis that a direct interaction was with the oxygens surrounding of the SO42− group (essentially, SO42− ions maintained their structure). Before building the composite systems, the behavior of the CNF model in solution was tested through a series of short (hundreds of picoseconds) all-atom molecular dynamics (MD) simulations in the NPT ensemble, at 300 K and 1 atm. The volume of the system was allowed to change along the y and z directions that are perpendicular to the fibril axis (x direction). The analysis of the MD trajectories confirmed the solution stability of the overall structure, which preserved both its morphology and crystallinity, although the exposed fibril portions (the 12 chains surrounding the core, made of four chains) reoriented their “side groups” (OH and CH2OH moieties), not involved in interchain hydrogen bonds, toward the solvent. Then, the equilibrated configuration was subsequently modified to obtain a model functionalized with carboxyl groups (TOCNFs). The procedure reflected the experimental TEMPO-mediated oxidation where the exposed C6 primary hydroxyls of the surface (12 chains in our case) were selectively replaced with C6 carboxyls.4,51 In the final model, approximately 45% of the primary hydroxyls were changed to the COOH moieties. The new configuration was energy-minimized and equilibrated in aqueous solution through the NPT MD runs (cell length, Lx, fixed) at ambient temperature and pressure. The final arrangement of the chains formed a slightly twisted fiber, inflated around the edges (due to the interaction with the solvent). This model was used as the starting geometry in all multicomponent simulations. Complex Models and Simulation Setup. All MD simulations were carried out by means of the Amsterdam Density Functional (ADF)/ReaxFF52 and LAMMPs codes53 resorting also to the HPC resources offered by CINECA through the Italian Super Computing Resource Allocation (ISCRA) project IscrC_GOHNC. The TOCNF model was surrounded by six GO sheets, which were placed in proximity to the fibril surface, and the complex was inserted in a box of water containing approximately 4500 molecules and 16 CuSO4 ions, randomly dispersed in the solvent (simulations at high ion concentration (32 ions) were also carried out). The final size of the simulation box, after equilibration, was approximately 8.3 × 5.1 × 3.9 nm3. The multicomponent systems consisted of TOCNFs, GO sheets, water molecules, Cu(II) ions, and SO4 counterions (Figure 7). The most crowded combination contained these species together, whereas the simplest cases were made of just one species and water. To disclose the synergistic effects on the ion-trapping process due to the TOCNF+GO association, the capturing agents (TOCNFs or GO) and ions were simulated independently, namely, TOCNF+ions and GO+ions, and together but without the ions (TOCNF+GO). The latter simulations were used to prepare realistic models of the TOCNF+GO complexes for further tests. Several other configurations were created and simulated in different environmental conditions to characterize tentative fibril packing, drying, porosity, etc. However, to be concise, we focus only on the three main models mentioned above (TOCNF+GO+ions, TOCNF+ions, and GO+ions) and refer to

Figure 7. Multicomponent system consisting of TOCNFs (solid surface, light red oxygens and gray carbon and hydrogens), six GO sheets (cyan sticks), Cu(II) ions (orange spheres), and SO4 ions (yellow and red balls). Water molecules have not been displayed for clarity. Periodic images are shown together with the simulation box (white parallelepiped).

other studies only to explain specific effects and behaviors, observed experimentally. It is worth mentioning that the reactivity was always on, and no restraints were introduced in the systems. All species could interact with each other and modify their conformation in response to the environment. MDs were carried out first in the NVT ensemble at T = 300 K for approximately 200 ps to equilibrate the whole system at the chosen temperature and allow the GO sheets to touch the TOCNF surface. Then, the equilibration was extended for another 500 ps in the NPT ensemble to reach the correct packing of all system components and the appropriate density. Afterward, the production dynamics was performed in the NVE ensemble for approximately 2 ns, and the system structures were collected every 0.05 ps. The temperature and pressure were controlled through the Berendsen’s thermostat and barostat54 with the relaxation constants of 0.1 ps. The time-step was set to 0.25 fs and the dynamics of 10 different initial arrangements was explored in each case. The final portions of these trajectories were combined and analyzed together. We are aware that in such a time scale we cannot provide the whole picture of the behavior of these types of materials as a whole (that would be impossible also extending system size, simulation time, and number of starting configurations), but we believe that using a few simplified models, such the ones we have used here, representing specific portions of the complex systems, we could give an idea of possible local features and characteristic effects that can be correlated to the experimental findings. The analysis was focused on descriptors that could be useful to disclose ion clustering and capture, namely, the atom−atom radial distribution functions, spatial distribution functions, various types of atom−atom distances, and hydrogen bond networks. To estimate the influence of the TOCNF+GO complexation on matrix packing and porosity, selected structures were extracted from the trajectories, dried, by removing water, and resimulated in the NPT ensemble at ambient temperature until complete equilibration. 7036


Article

ACS Nano

(14) Karim, Z.; Hakalahti, M.; Tammelin, T.; Mathew, A. P. In Situ TEMPO Surface Functionalization of Nanocellulose Membranes for Enhanced Adsorption of Metal Ions from Aqueous Medium. RSC Adv. 2017, 7, 5232−5241. (15) Hontoria-Lucas, C.; López-Peinado, A. J.; López-González, J. d. D.; Rojas-Cervantes, M. L.; Martín-Aranda, R. M. Study of OxygenContaining Groups in a Series of Graphite Oxides: Physical and Chemical Characterization. Carbon 1995, 33, 1585−1592. (16) He, H.; Klinowski, J.; Forster, M.; Lerf, A. A New Structural Model for Graphite Oxide. Chem. Phys. Lett. 1998, 287, 53−56. (17) Gao, W.; Majumder, M.; Alemany, L. B.; Narayanan, T. N.; Ibarra, M. A.; Pradhan, B. K.; Ajayan, P. M. Engineered Graphite Oxide Materials for Application in Water Purification. ACS Appl. Mater. Interfaces 2011, 3, 1821−1826. (18) Wang, B.; Lou, W.; Wang, X.; Hao, J. Relationship between Dispersion State and Reinforcement Effect of Graphene Oxide in Microcrystalline Cellulose-Graphene Oxide Composite Films. J. Mater. Chem. 2012, 22, 12859−12866. (19) Zhu, C.; Liu, P.; Mathew, A. P. Self-Assembled TEMPO Cellulose Nanofibers: Graphene Oxide-Based Biohybrids for Water Purification. ACS Appl. Mater. Interfaces 2017, 9, 21048−21058. (20) Liu, P.; Oksman, K.; Mathew, A. P. Surface Adsorption and Self-Assembly of Cu(II) Ions on TEMPO-Oxidized Cellulose Nanofibers in Aqueous Media. J. Colloid Interface Sci. 2016, 464, 175−182. (21) Zhu, C.; Dobryden, I.; Rydén, J.; Ö berg, S.; Mathew, A. P.; Holmgren, A. Adsorption Behavior of Cellulose and Its Derivatives toward Ag(I) in Aqueous Medium: An AFM, Spectroscopic, and DFT Study. Langmuir 2015, 31, 12390−12400. (22) Paajanen, A.; Vaari, J. High-Temperature Decomposition of the Cellulose Molecule: a Stochastic Molecular Dynamics Study. Cellulose 2017, 24, 2713−2725. (23) Li, Y.; Zhu, H.; Zhu, S.; Wan, J.; Liu, Z.; Vaaland, O.; Lacey, S.; Fang, Z.; Dai, H.; Li, T.; Hu, L. Hybridizing Wood Cellulose and Graphene Oxide toward High-Performance Fibers. NPG Asia Mater. 2015, 7, e150. (24) Medhekar, N. V.; Ramasubramaniam, A.; Ruoff, R. S.; Shenoy, V. B. Hydrogen Bond Networks in Graphene Oxide Composite Paper: Structure and Mechanical Properties. ACS Nano 2010, 4, 2300−2306. (25) Achtyl, J. L.; Unocic, R. R.; Xu, L.; Cai, Y.; Raju, M.; Zhang, W.; Sacci, R. L.; Vlassiouk, I. V.; Fulvio, P. F.; Ganesh, P.; Wesolowski, D. J.; Dai, S.; Van Duin, A. C. T.; Neurock, M.; Geiger, F. M. Aqueous Proton Transfer Across Single-Layer Graphene. Nat. Commun. 2015, 6, 6539. (26) Singh, S. K.; Srinivasan, S. G.; Neek-Amal, M.; Costamagna, S.; Van Duin, A. C. T.; Peeters, F. M. Thermal Properties of Fluorinated Graphene. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 104114. (27) Neyts, E. C.; Van Duin, A. C. T.; Bogaerts, A. Formation of Single Layer Graphene on Nickel under Far-from-Equilibrium High Flux Conditions. Nanoscale 2013, 5, 7250−7255. (28) Huang, L.; Seredych, M.; Bandosz, T. J.; Van Duin, A. C. T.; Lu, X.; Gubbins, K. E. Controllable Atomistic Graphene Oxide Model and Its Application in Hydrogen Sulfide Removal. J. Chem. Phys. 2013, 139, 194707. (29) Rahaman, O.; Van Duin, A. C. T.; Bryantsev, V. S.; Mueller, J. E.; Solares, S. D.; Goddard, W. A., III; Doren, D. J. Development of a ReaxFF Reactive Force Field for Aqueous Chloride and Copper Chloride. J. Phys. Chem. A 2010, 114, 3556−3568. (30) Van Duin, A. C. T.; Bryantsev, V. S.; Diallo, M. S.; Goddard, W. A.; Rahaman, O.; Doren, D. J.; Raymand, D.; Hermansson, K. Development and Validation of a ReaxFF Reactive Force Field for Cu Cation/Water Interactions and Copper Metal/Metal Oxide/Metal Hydroxide Condensed Phases. J. Phys. Chem. A 2010, 114, 9507− 9514. (31) Senftle, T. P.; Hong, S.; Islam, M. M.; Kylasa, S. B.; Zheng, Y.; Shin, Y. K.; Junkermeier, C.; Engel-Herbert, R.; Janik, M. J.; Aktulga, H. M.; Verstraelen, T.; Grama, A.; Van Duin, A. C. T. The ReaxFF

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.8b02734. Additional information (PDF)

AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. ORCID

Susanna Monti: 0000-0002-3419-7118 Aji P. Mathew: 0000-0001-8909-3554 Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of the Swedish Research Council (VR, grant nos. 621-2013-5997 and 2017-04254). We thank Y. Liu, Stockholm University, Sweden, who kindly provided the TOCNFs. REFERENCES (1) Klemm, D.; Kramer, F.; Moritz, S.; Lindström, T.; Ankerfors, M.; Gray, D.; Dorris, A. Nanocelluloses: A New Family of Nature-Based Materials. Angew. Chem., Int. Ed. 2011, 50, 5438−5466. (2) Huang, H.; Ying, Y.; Peng, X. Graphene Oxide Nanosheet: An Emerging Star Material for Novel Separation Membranes. J. Mater. Chem. A 2014, 2, 13772−13782. (3) Dervin, S.; Dionysiou, D. D.; Pillai, S. C. 2D Nanostructures for Water Purification: Graphene and Beyond. Nanoscale 2016, 8, 15115−15131. (4) Isogai, A.; Saito, T.; Fukuzumi, H. TEMPO-Oxidized Cellulose Nanofibers. Nanoscale 2011, 3, 71−85. (5) Saito, T.; Kimura, S.; Nishiyama, Y.; Isogai, A. Cellulose Nanofibers Prepared by TEMPO-Mediated Oxidation of Native Cellulose. Biomacromolecules 2007, 8, 2485−2491. (6) Suk, J. W.; Piner, R. D.; An, J.; Ruoff, R. S. Mechanical Properties of Monolayer Graphene Oxide. ACS Nano 2010, 4, 6557− 6564. (7) Sun, P.; Zhu, M.; Wang, K.; Zhong, M.; Wei, J.; Wu, D.; Xu, Z.; Zhu, H. Selective Ion Penetration of Graphene Oxide Membranes. ACS Nano 2013, 7, 428−437. (8) Yao, Q.; Fan, B.; Xiong, Y.; Jin, C.; Sun, Q.; Sheng, C. 3D Assembly Based on 2D Structure of Cellulose Nanofibril/Graphene Oxide Hybrid Aerogel for Adsorptive Removal of Antibiotics in Water. Sci. Rep. 2017, 7, 45914. (9) Fang, Q.; Zhou, X.; Deng, W.; Zheng, Z.; Liu, Z. Freestanding Bacterial Cellulose-Graphene Oxide Composite Membranes with High Mechanical Strength for Selective Ion Permeation. Sci. Rep. 2016, 6, 33185. (10) Chen, C.; Zhang, T.; Zhang, Q.; Chen, X.; Zhu, C.; Xu, Y.; Yang, J.; Liu, J.; Sun, D. Biointerface by Cell Growth on Graphene Oxide Doped Bacterial Cellulose/Poly(3,4-ethylenedioxythiophene) Nanofibers. ACS Appl. Mater. Interfaces 2016, 8, 10183−10192. (11) Sitko, R.; Musielak, M.; Zawisza, B.; Talik, E.; Gagor, A. Graphene Oxide/Cellulose Membranes in Adsorption of Divalent Metal Ions. RSC Adv. 2016, 6, 96595−96605. (12) Zhu, C.; Soldatov, A.; Mathew, A. P. Advanced Microscopy and Spectroscopy Reveal the Adsorption and Clustering of Cu(II) onto TEMPO-Oxidized Cellulose Nanofibers. Nanoscale 2017, 9, 7419− 7428. (13) Melone, L.; Rossi, B.; Pastori, N.; Panzeri, W.; Mele, A.; Punta, C. TEMPO-Oxidized Cellulose Cross-Linked with Branched Polyethyleneimine: Nanostructured Adsorbent Sponges for Water Remediation. ChemPlusChem 2015, 80, 1408−1415. 7037


Article

ACS Nano Reactive Force-Field: Development, Applications and Future Directions. npj Computational Mater. 2016, 2, 15011. (32) Pan, S.; Aksay, I. A. Factors Controlling the Size of Graphene Oxide Sheets Produced via the Graphite Oxide Route. ACS Nano 2011, 5, 4073−4083. (33) Stankovich, S.; Piner, R. D.; Nguyen, S. T.; Ruoff, R. S. Synthesis and Exfoliation of Isocyanate-Treated Graphene Oxide Nanoplatelets. Carbon 2006, 44, 3342−3347. (34) Liu, P.; Zhu, C.; Mathew, P. A. Graphene Oxide Coated Nanocellulosic Fibrous Membranes for Ultrafast Separation of Dyes from Water. J. Hazar Mater. Submitted. (35) Pasquarello, A.; Petri, I.; Salmon, P. S.; Parisel, O.; Car, R.; Tóth, É .; Powell, D. H.; Fischer, H. E.; Helm, L.; Merbach, A. E. First Solvation Shell of the Cu(II) aquaion: Evidence for Fivefold Coordination. Science 2001, 291, 856−859. (36) Schwenk, C. F.; Rode, B. M. New Insights into the Jahn-Teller Effect through ab Initio Quantum-Mechanical/Molecular-Mechanical Molecular Dynamics Simulations of Cu(II) in water. ChemPhysChem 2003, 4, 931−943. (37) Li, C.; Wu, Z.; Liang, H.; Chen, J.; Yu, S. Ultralight Multifunctional Carbon-Based Aerogels by Combining Graphene Oxide and Bacterial Cellulose. Small 2017, 13, 1700453. (38) Kafy, A.; Sadasivuni, K. K.; Kim, H.; Akther, A.; Kim, J. Designing Flexible Energy and Memory Storage Materials Using Cellulose Modified Graphene Oxide Nanocomposites. Phys. Chem. Chem. Phys. 2015, 17, 5923−5931. (39) Si, H.; Luo, H.; Xiong, G.; Yang, Z.; Raman, S. R.; Guo, R.; Wan, Y. One-step In Situ Biosynthesis of Graphene Oxide-Bacterial Cellulose Nanocomposite Hydrogels. Macromol. Rapid Commun. 2014, 35, 1706−1711. (40) Liu, Y.; Zhou, J.; Zhu, E.; Tang, J.; Liu, X.; Tang, W. Facile Synthesis of Bacterial Cellulose Fibres Covalently Intercalated with Graphene Oxide by One-Step Cross-Linking for Robust Supercapacitors. J. Mater. Chem. C 2015, 3, 1011−1017. (41) Liu, Y.; Gordeyeva, K.; Bergström, L. Steady-Shear and Viscoelastic Properties of Cellulose Nanofibril−Nanoclay Dispersions. Cellulose 2017, 24, 1815−1824. (42) Charlier, L.; Mazeau, K. Molecular Modeling of the Structural and Dynamical Properties of Secondary Plant Cell Walls: Influence of Lignin Chemistry. J. Phys. Chem. B 2012, 116, 4163−4174. (43) Alqus, R.; Eichhorn, S. J.; Bryce, R. A. Molecular Dynamics of Cellulose Amphiphilicity at the Graphene-Water Interface. Biomacromolecules 2015, 16, 1771−1783. (44) Nishiyama, Y.; Langan, P.; Chanzy, H. Crystal Structure and Hydrogen-Bonding System in Cellulose Iß from Synchrotron X-ray and Neutron Fiber Diffraction. J. Am. Chem. Soc. 2002, 124, 9074− 9082. (45) Monti, S.; Bronco, S.; Cappelli, C. Toward the Supramolecular Structure of Collagen: A Molecular Dynamics Approach. J. Phys. Chem. B 2005, 109, 11389−11398. (46) Monti, S.; Bramanti, E.; Porta, V. D.; Onor, M.; D’Ulivo, A.; Barone, V. Interaction of Collagen with Chlorosulphonated Paraffin Tanning Agents: Fourier Transform Infrared Spectroscopic Analysis and Molecular Dynamics Simulations. Phys. Chem. Chem. Phys. 2013, 15, 14736−14747. (47) http://atb.uq.edu.au/molecule.py?molid=34774 (accessed on June 9, 2018). (48) Dri, F. L.; Wu, X.; Moon, R. J.; Martini, A.; Zavattieri, P. D. Evaluation of Reactive Force Fields for Prediction of the ThermoMechanical Properties of Cellulose iß. Comput. Mater. Sci. 2015, 109, 330−340. (49) Chenoweth, K.; van Duin, A. C. T.; Goddard, W. A. ReaxFF Reactive Force Field for Molecular Dynamics Simulations of Hydrocarbon Oxidation. J. Phys. Chem. A 2008, 112, 1040−1053. (50) Zhang, H.; Iype, E.; Nedea, S. V.; Rindt, C. C. M. Molecular Dynamics Study on Thermal Dehydration Process of Epsomite (MgSO4·7H2O). Mol. Simul. 2014, 40, 1157−1166.

(51) Okita, Y.; Saito, T.; Isogai, A. Entire Surface Oxidation of Various Cellulose Microfibrils by TEMPO-Mediated Oxidation. Biomacromolecules 2010, 11, 1696−1700. (52) Baerends, E. J.; Ziegler, T.; Atkins, A. J.; Autschbach, J.; Baseggio, O.; Bashford, D.; Bérces, A.; Bickelhaupt, F. M.; Bo, C.; Boerrigter, P. M.; Cavallo, L.; Daul, C.; Chong, D. P.; Chulhai, D. V.; Deng, L.; Dickson, R. M.; Dieterich, J. M.; Ellis, D. E.; van Faassen, M.; Fan, L.; et al. ADF2017, SCM, Theoretical Chemistry; Vrije Universiteit: Amsterdam, The Netherlands, 2017, http://www.scm. com. (53) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1−19. (54) Berendsen, H. J. C.; Postma, J. P. M.; Van Gunsteren, W. F.; Dinola, A.; Haak, J. R. Molecular Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984, 81, 3684−3690.

7038


Cellulose NanofiberâGraphene Oxide Biohybrids: Disclosing the Self

Recommend Documents

Cellulose NanofiberâGraphene Oxide Biohybrids: Disclosing the Self