Structure and Properties of Citrate Overlayers Adsorbed at the

(26) Nichols et al. suggested that the conformation of an interfacial citrate anion is ...... Brancolini , G.; Kokh , D. B.; Calzolai , L.; Wade , R. ...
0 downloads 0 Views 5MB Size
Download PDF
Article pubs.acs.org/Langmuir

Structure and Properties of Citrate Overlayers Adsorbed at the Aqueous Au(111) Interface Louise B. Wright,† P. Mark Rodger,† and Tiffany R. Walsh*,‡ †

University of Warwick, Department of Chemistry and Centre for Scientific Computing, Coventry CV4 7AL, United Kingdom Deakin University, Institute for Frontier Materials, Geelong, VIC 3216, Australia

‡

S Supporting Information *

ABSTRACT: One of the most common means of gold nanoparticle (AuNP) biofunctionalization involves the manipulation of precursor citrate-capped AuNPs via ligand displacement. However, the molecular-level structural characteristics of the citrate overlayer adsorbed at the aqueous Au interface at neutral pH remain largely unknown. Access to atomistic-scale details of these interfaces will contribute much needed insight into how AuNPs can be manipulated and exploited in aqueous solution. Here, the structures of such citrate overlayers adsorbed at the aqueous Au(111) interface at pH 7 are predicted and characterized using atomistic molecular dynamics simulations, for a range of citrate surface densities. We find that the overlayers are disordered in the surface density range considered, and that many of their key characteristics are invariant with surface density. In particular, we predict the overlayers to have 3-D, rather than 2-D, morphologies, with the anions closest to the gold surface being oriented with their carboxylate groups pointing away from the surface. We predict both striped and island morphologies for our overlayers, depending on the citrate surface density, and in all cases we find bare patches of the gold surface are present. Our simulations suggest that both citrate−gold adsorption and citrate−counterion pairing contribute to the stability of these citrate overlayer morphologies. We also calculate the free energy of adsorption at the aqueous Au(111) interface of a single citrate molecule, and compare this with the corresponding value for a single arginine molecule. These findings enable us to predict the conditions under which ligand displacement of surface-adsorbed citrate by arginine may take place. Our findings represent the first steps toward elucidating a more elaborate, detailed atomistic-scale model relating to the biofunctionalization of citrate-capped AuNPs.

■

INTRODUCTION Strategies for noncovalent functionalization of AuNPs by biomolecules such as amino acids,1,2 proteins,3,4 or nucleic acids,5 have bright prospects for the development of many potential applications, ranging from smart materials6−9 to biomedical therapeutics and diagnostics.10−13 The Turkevich− Frens reduction process14,15 in the presence of trisodium citrate is currently one of the most widely used methods for preparing monodisperse AuNPs under aqueous conditions. The citrate anion is a good capping ligand for precursor AuNPs, since it appears labile with respect to displacement by a range of biomolecules,16 and yet is able to prevent AuNP aggregation in water. To improve AuNP stability in solution at high ionic strength,17 alternative synthetic strategies, also involving citrate, have been proposed. For example, recently Xu et al. reported a procedure that can be carried out at physiological temperatures (37 °C) involving the reduction of HAuCl4 in the presence of trisodium citrate and ssDNA.18 In the realm of biomimetic materials synthesis a number of recent citrate displacement experiments have yielded promising results toward the ultimate goal of being able to exert fine control over AuNP organization into predefined nanostructures. For instance, Sethi and Knecht identified that in the © XXXX American Chemical Society

presence of the amino acid arginine, citrate-capped AuNPs formed branched linear chains,1,2 while Jimenez et al. found that they were able to control the aggregation of AuNPs functionalized by chemisorbed peptides featuring functional groups of different charge states due to the presence of citrate anions displaced from the gold surface.19 On the other hand, the biocompatibility of AuNPs is of critical importance in nanomedicine, where their adsorption within the body could induce structural changes, resulting in malfunction of substrate proteins and hence possible disease.20,21 Hence, several groups have investigated the toxicity of citrate-capped AuNPs. For instance, Gomes et al.4 recently reported that the binding affinity of the protein cytochrome C to a citrate-capped AuNP was approximately 3 times weaker than that to a AuNP capped by the ligand mercaptoundecanoic acid at pH 7.4, while Tournebize et al. found that citrate-capped AuNPs were more reactive in vitro than AuNPs functionalized by hihydrolipoic acid.22 Received: September 15, 2014 Revised: November 27, 2014

A

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

overlayer stability. To aid in this, we have also calculated the free energy of adsorption of a single citrate anion at the aqueous Au(111) interface using well-tempered metadynamics,30 and compared this with the corresponding value for adsorption of a single arginine molecule.

While these results mark critical advances, a deeper understanding of the fundamental mechanisms governing biomolecule adsorption at the aqueous interface of citratecapped AuNPs remains key to enabling further progress in a number of fields. Determination of biomolecule structure and/ or aggregation state when adsorbed to a surface remains challenging despite recent experimental advances.3,23,24 Molecular simulation, on the other hand, is in principle well-suited to providing such information at the level of detail required. Thus far, however, citrate has rarely been included in the model used in biointerfacial gold simulations. The one notable exception to this is the recent work reported by Brancolini et al., who modeled ubiquitin docking at the aqueous Au(111) interface; molecular dynamics (MD) simulations in which citrate was included both implicitly (where the gold surface carried an overall negative charge) and explicitly were carried out.25 Nonetheless, before we can fully characterize the mechanisms involved in the physisorption of biomolecules onto the surface of a citrate-capped AuNP, molecular-level insight into the nature of this interface in the absence of the biomolecule is required. Overlayers of citrate adsorbed at the aqueous gold interface have been experimentally characterized using a number of techniques, including in situ scanning tunneling microscopy (STM),26 infrared (IR) spectroscopy,27,28 chronocoulometry,29 and cyclic voltammetry.26,27,29 All of these studies were carried out under highly acid conditions (pH 0−3), where citric acid and dihydrogen citrate are the most stable states of the molecule in solution. However, on adsorption to gold, both species were found to fully deprotonate despite the acidic conditions. The presence of a single adsorption band at 1390 cm−1 in the IR spectra of the adsorbed species27,28 was consistent with the model proposed by Lin et al. in which all three carboxylate groups of citrate are coordinated to the gold surface in an η − 2 manner upon adsorption.26 Nichols et al. suggested that the conformation of an interfacial citrate anion is distorted relative to its solution structure under these conditions. In particular, these authors found that in the model that best fitted their experimental data, the average tilt angle of the three carboxylate groups, with respect to the surface normal, was either random or 55°.28 Despite this considerable body of previous work, the manner in which overlayers of citrate adsorb at an aqueous gold interface, under the physiological pH conditions relevant to biomolecule binding experiments (where citrate is fully deprotonated both in solution and at the surface), remains undetermined at present, either by experiment or simulation, and is the subject of the study presented in this paper. Here, we have investigated the structure and characteristics of citrate overlayers that are noncovalently adsorbed at the aqueous Au(111) interface, at pH 7, for a range of surface densities using atomistic molecular dynamics (MD) simulations; an explicit description of liquid water was used. Although the influence of AuNP vertices and edges has not been accounted for in our work, the surface areas of the facets presented by AuNPs commonly used in biomolecule−citrate displacement experiments are large with respect to the citrate anion; approximation by a planar interface is appropriate in this case. Due to the complexity of the systems studied and the large number of degrees of freedom involved, simulation strategies to ensure adequate equilibration have been developed. The final configurations of each overlayer have allowed us to identify which factors may contribute to the

■

COMPUTATIONAL METHODS

System Details. All the work presented here was carried out using the Molecular Dynamics (MD) software package Gromacs 4.5.5.31 Two sets of simulations of citrate assemblies at the aqueous Au(111) interface were performed, denoted “random” and “designed”, respectively (summarized in Figure S1 of the Supporting Information). Simulation systems comprised n citrates, a gold slab of dimension 29.3 × 30.4 Å2 (10 × 12 supercell of the Au(111) surface, 5 layers thick), and 3n Na+ counterions, all in explicit water. Five interfacial citrate surface densities of 8, 10, 12, 14, and 16 anions per simulation cell (1.49−2.98 ×10−6 mol m−2) were investigated using two different simulation protocols, denoted here as “random” and “designed”. In the “random” procedure, two different cell sizes were used, differing only in the dimension normal to the gold surface. The smaller cell (denoted “annealing”), used only in the initial equilibration stages of each simulation, had a cell dimension perpendicular to the slab surface of 50.5 Å. The larger cell (denoted “production”) had a corresponding cell dimension of 180.0 Å. The larger cell size ensured that, even at the highest citrate surface density, the concentration of Na+ counterions in solution was within experimentally reasonable limits, in this case ∼0.5 M. Only the production cell was used in the designed set of simulations. The NVT ensemble, with 3-D periodic boundary conditions, was used throughout; the temperature (300 K unless stated otherwise) was maintained using a Nosé Hoover thermostat.32,33 The number of TIP3P water molecules in each cell was adjusted prior to simulation to ensure that the density of liquid water midway between the top surface of the gold slab and the bottom surface of its periodic image corresponded to that of bulk liquid water at 1 bar, described by the modified TIP3P potential. Newton’s equations of motion were solved using the leapfrog algorithm.34 An integration time-step of 1 fs was employed, and frames were saved every 1 ps during production stages. Particle Mesh Ewald (PME) electrostatic summation35 was truncated at 13 Å while Lennard-Jones nonbonded interactions were cutoff at 11 Å with a force-switching potential employed at 10 Å. In all our simulations, citrate was modeled by a force-field (FF) recently developed by us, fitted to reproduce the internal structure of the aqueous anion found from first-principles simulation,36 while the polarizable GolP-CHARMM potential37 was used to describe all intermolecular gold interactions. The substrate polarizability was captured using a rigid-rod dipole38 attached to each gold atom. For all MD simulations reported herein, the actual gold atoms in the metal slab were held fixed in space, while the dipole particles were free to rotate (according to their restraint potential and the temperature of the thermostat). Recently reported tests indicate that the difference between binding free energies obtained using a rigid substrate compared with those calculated using a slab where all atoms can move is small in comparison to standard experimental error.39 Both GolP-CHARMM and the citrate FF used are compatible with the bioorganic FF CHARMM.40 Nonbonded parameters for Na+ were adopted from the CHARMM FF, while the modified TIP3P water model41,42 (with which CHARMM is harmonized) was used. Molecular dynamics simulations of the structure of saline solutions (including NaCl) at the aqueous Au(111) interface, using these FFs, have been recently reported.43 “Random” Simulations. For each of the five citrate surface densities we investigated, three different initial configurations of the overlayer were independently generated. In generating these initial configurations, both the location on, and orientation relative to, the Au(111) surface of each citrate was chosen randomly, while both the citrate center of mass (COM)−gold distance and internal citrate conformation were kept constant, these being 5 Å and fully extended (see Analysis section), respectively. These randomly generated B

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

overlayers were then subjected to three rounds of simulated annealing (10 cycles of each round) while restrained (in terms of the molecule− surface vertical distance) to the gold surface (see Supporting Information Simulated Annealing section for details). Simulation at elevated temperatures (see Supporting Information Simulated Annealing section for the temperature range) is necessary to aid equilibration of such complex systems. For computational efficiency, this stage of the protocol was conducted in the smaller “Annealing” simulation cell. After 16.5 ns of annealing, the citrate overlayer, ions, and gold slab were resolvated in the “production” cell. All subsequent simulations were performed at a constant temperature of 300 K. First, the system was re-equilibrated in this larger cell for 4 ns, while the citrate COM− gold distance restraint was maintained. This restraint was subsequently relaxed gradually until the anions were free to move in all directions. Specifically, the restraint on the citrate COM−gold distance was changed to a stiff harmonic spring constraint, of force constant 1500 kJ mol−1 nm−2, for 2 ns before being modified to a weaker spring, of force constant 750 kJ mol−1 nm−2, for a further 2 ns. After the final stage, a 4 ns simulation in which all restraints and constraints on citrate motion were lifted, the potential energy of the three systems at each of the five citrate surface densities was monitored. The simulation with the lowest average potential energy, over the final 1 ns of this 4 ns stage, was then extended by 26 ns, bringing the total trajectory time to 54.5 ns (24.5 ns equilibration and 30 ns production). A number of aspects of this annealing protocol were investigated, to ensure that the configurations of the interfacial citrate assemblies generated were independent of the procedure used, before the final version described above was adopted. Namely, the initial citrate overlayer conformation (citrate COM−gold separation distance, orientation relative to the Au(111) surface and location of Na+ ions) and the manner in which the restraint on the citrate COM− gold distance was lifted were all probed. Details can be found in Supporting Information Random Simulations−Further Details section. “Designed” Simulations. Each of the lowest-energy final configurations for our five surface densities, derived from the “random” set of simulations, featured at least one of two types of citrate−citrate motifs. The first motif, denoted “adsorbed linear chains”, had two or more citrates in direct surface contact, adjacent to each other, with the axes of their carbon backbones parallel (Figure 1a). In the second

without layer of water intervening between the two. Since these two citrate motifs were observed in the “random” simulations for a range of citrate surface densities, it was thought that they could contribute significantly to the stability of the citrate overlayers. Hence, initial overlayer configurations were designed in which the number and geometry of both types of interaction were optimized. First, a set of design criteria was established using data collected from the set of random simulations (Tables S3 and S4, Supporting Information). Motifs were identified using the conditions set out in Supporting Information Definitions of Citrate−Citrate Motifs section. The key structural metrics for these two motifs (“Adsorbed Linear Chains” and Inverted “Bilayer”) were extracted from this analysis, and used to construct new overlayer initial configurations; see Supporting Information Definitions of Citrate−Citrate Motifs: Key Metrics section, and Figure S2 in the Supporting Information. On the basis of these data, initial interfacial citrate assembly configurations for the “designed” set of simulations were generated. For each surface density, the ratio of directly adsorbed to indirectly adsorbed citrate anions in the overlayer was chosen to be 6:4 (to the nearest whole number). Herein, a citrate was defined as being “indirectly adsorbed” to the Au(111) surface if its closest atom−gold separation was greater than 5 Å but less than 8 Å. Again, we rationalized these thresholds on the basis of geometric principles, where we propose that an 8 Å separation is sufficient to accommodate a water molecule between the citrate and the gold surface. All directly adsorbed citrates in the idealized arrays were part of an “adsorbed linear chain”, and all indirectly adsorbed citrates were one member of an inverted bilayer pair. The exact geometric criteria used to construct each starting configuration (shown in Figure S3, Supporting Information) are given Supporting Information Design Criteria section. In the “designed” simulations, Na+ counterions were positioned by hand between neighboring citrates before each system was solvated by a pre-equilibrated box of TIP3P water. The protocol used for these runs was identical to the latter stages of the random simulations carried out in the larger “production”-sized cell. Namely, each system was equilibrated for 4 ns with all citrate COM−gold distances restrained. The restraining force was then gradually relaxed, first to a stiff harmonic spring of constant 1500 kJ mol−1 nm−2, and then to a weak spring of constant 750 kJ mol−1 nm−2 in two runs, each of 2 ns duration, before finally being lifted totally for the last 30 ns of simulation. Single Molecule Metadynamics Simulations. We calculated the free energy of adsorption at the aqueous Au(111) interface of two species: a single citrate anion, and, a single uncapped arginine cation, at pH 7 using the software package Plumed 1.344 in conjunction with Gromacs 4.5.5.31 Each system comprised a gold slab (surface area 43.95 × 40.60 Å2, 5 layers thick); an adsorbate, either a citrate anion, or, an arginine cation (in the L-chiral form, uncapped with zwitterionic termini); counterions (3 Na+ for the citrate system and 6 Cl−, and 5 Na+ for arginine); and 2577/2568 TIP3P water molecules for citrate and arginine, respectively. The cell dimension in the direction normal to the interface was 55.0 Å. All other details were the same as those outlined in the System Details section above. Our well-tempered metadynamics simulations were run for 250 ns. We monitored the difference between the free energy corresponding to the adsorbate located in the bulk region (far from the interface) and the free energy corresponding to the lowest point of the deepest free energy minimum, as a function of simulation time. After 250 ns, we found that the fluctuations around the average value of this free energy difference were less than 4 kJ mol−1. We used one collective variable (CV), namely, the distance between the gold surface and the adsorbate COM, in the direction normal to the interface. Gaussians of width 1 Å and initial height 0.10 kJ mol−1 were added every 1000 MD steps to this CV. The well-tempered biasing factor was 10. For computational efficiency a soft repulsive wall was set at a distance of 25 Å above the top surface of the gold slab. Analysis. All structural analyses of the interfacial citrate overlayers were carried out over the final 1 ns of simulation trajectories. Unless stated otherwise, data for the lowest energy run out of the “random” and “designed” simulations are reported for each surface density.

Figure 1. Snapshots of (a) “adsorbed linear chains” and (b) an “inverted bilayer” citrate−citrate motif identified in the “random” set of simulations. Gold atoms are shown as gold spheres, Na+ ions are shown as green spheres, and the citrates are rendered in licorice and colored according to element (carbon in cyan, oxygen in red, and hydrogen in white). Water has been omitted for clarity. motif, a directly adsorbed citrate (oriented with its carboxylate groups pointed away from the interface) interacted with a second citrate that was not in direct surface contact (and oriented in the opposite sense) (Figure 1b), denoted here as “inverted bilayer”. Both types of motif were stabilized by Na+ counterions located between the citrates involved. Herein, citrates classified as being “directly adsorbed” had a closest atom−gold separation of 5 Å or less. We determined this threshold on the basis of geometric principles, where we supposed that a surface separation of 5 Å was sufficient to account for the van der Waals extent of the interacting citrate and the surface gold atoms, C

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

During the course of each simulation, only a small proportion of the total number of citrate molecules remained in direct surface contact; full detachment from the overlayer as a whole, however, was not frequently observed. The “adsorbed status” of a citrate was hence classed as one of three categories: “directly adsorbed”, “indirectly adsorbed”, or “third-layer”. See Designed Simulations section for definition of the first two categories, specifically regarding the physical basis for our choice of thresholds used to classify these cases. A “thirdlayer” citrate was defined as a citrate that was neither directly nor indirectly adsorbed to the Au(111) surface, for which its closest atom distance to either of the other types of adsorbed species was less than 3.5 Å. For clarity it is necessary to introduce some further nomenclature. First, we define the “carbon backbone” of citrate to be the stretch of five consecutive carbon atoms spanning the two terminal carboxylate groups, C1OO−−C2H2−C3(C6OO−)(OH)−C4H2−C5OO−, where C3 is the “central carbon atom” of the anion; backbone vectors C1 to C5 and C5 to C1 are equivalent. Second, as in our previous work,36 the internal structure of citrate was classified using three metrics: the conformation of its carbon backbone, its internal hydrogen-bonded status, and the number of Na+ counterions to which it is coordinated. The carbon backbone conformation was determined using the two CCCC dihedral angles, C1−C2−C3−C4 and C5−C4−C3−C2. The anion was defined to be “fully folded” if both dihedral angles had values within the range −120° < θ < 120°, “fully extended” in the range −180° < θ < −120° or 120° < θ < 180°, and “partially folded” in all other cases. Internal hydrogen bonds between the hydroxyl hydrogen atom and an oxygen atom belonging to one of the terminal carboxylate groups were identified using three established geometric criteria.45 The same criteria were used to quantify citrate−water hydrogen-bonding. A Na+ counterion was coordinated to citrate if its closest atom distance to the anion was less than 2.5 Å. Visual inspection of the equilibrated interfacial citrate overlayers revealed that, for the surface densities investigated here, overlayer morphologies could be broadly classed as either “island-like” or “stripelike” (Figure 2). To probe the stability and extent of each morphology, two metrics were used: radius of gyration and the surface footprint. Briefly, these two properties quantify the average spatial extent of the overlayer in 3-D space and 2-D (lateral, in-plane) space, respectively. Further details on the definition of both metrics can be found in Supporting Information Overlayer Morphology section. Furthermore, the average geometry of individual citrates within the overlayers were studied in detail, including investigation of the rotational freedom of the citrates, and the average orientation. See Supporting Information Geometry within Overlayers section for details. In addition, to quantify the abundance of citrate−citrate interactions stabilized by coordinating Na+ ions, a final definition is introduced, a citrate “pair”. Two anions were considered to form a “pair” if the distance between their central carbon atoms was less than 7 Å. The number of unique citrate−citrate pairs at time t (npairs(t)), weighted by the average number of coordinating Na+ ions shared by the two anions (⟨n(t)NA+⟩), and their average separation (⟨(C3···C3)(t)⟩) was used to define a “pair quality” factor:

Q (t ) =

Figure 2. Snapshots of the final conformation of citrate assemblies, surface densities 8−16 anions per simulation cell, after both “random” (left) and “designed” (right) simulations. Each is a 2 × 2 representation of the simulation cell. Water has been omitted for clarity.

density. In what follows, both types of finding are discussed and, where possible, quantified. All of our predicted overlayer structures were amorphous in character, with a complete absence of crystalline structures, regardless of initial configuration. This finding is consistent with both the lability of citrate as a capping ligand on the surface of AuNPs, and STM imaging of citric acid adsorption onto a Au(111) electrode under acidic conditions (pH 1) at potentials greater than 0.8 V.26 Ordered overlayer structures of adsorbed citrate at the aqueous gold interface were only observed by Lin et al. at pH 1 under electrode potentials in the range 0.1−0.8 V.26 The surface density coverage of these overlayers was approximately 1.40 × 10−6 mol m−2, just short of the range modeled in the present study. The amorphous nature of the interfacial overlayers investigated here inherently prevents a definitive structural and energetic characterization, the type of which is only possible for crystalline structures. Bearing these caveats in mind, there was consensus in the total potential energy and morphology of each system between the “random” and “designed” simulations (Figure 2, Table 1, and Table S5, Supporting Information). Apart from the overlayer of 8 citrates, the difference in total potential energy per citrate between the two protocols was less than, or on the order of, 2 kBT at room temperature (300 K), suggesting that

npairs(t )⟨n(t )NA+ ⟩⟨(C 3 ··· C 3)(t )⟩ 1 c(c 2

+ 1)

(1)

Here c is the total number of citrate anions within the overlayer. The number of shared coordinating Na+ ions and citrate−citrate distance were chosen to weight the quality of pair interactions in an overlayer, since both can act to reduce the electrostatic repulsion between two citrates in close proximity.

■

RESULTS AND DISCUSSION After extensive simulations carried out following the two different simulation protocols, many of the main features of an interfacial citrate−gold overlayer at a given surface density were found to have converged. Moreover, some characteristics were common to all overlayers, independent of citrate surface D

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Table 1. Overlayer Morphology after Both “Random” and “Designed” Simulations for Each Surface Density, along with their Average Footprint (FP) in Å2a system

“random”

FP

“designed”

FP

8 10 12 14 16

island stripe stripe island disperse

68 ± 2 182 ± 6 161 ± 4 159 ± 4

island stripe stripe island island

100 ± 4 189 ± 5 214 ± 4 140 ± 6 211 ± 3

that the citrate−water interfacial energy is small and that alternative factors could be pivotal in determining the morphology of an interfacial citrate overlayer. The lowest energy structure for the highest density case, 16 citrates, also supports this argument, displaying a more “island-like” than “stripe-like” morphology. However, in this case, the “islands” have coalesced with those in neighboring cells, blurring the distinction between the two states. For surface densities greater than those investigated here, we suggest two opposing routes may be possible: the growth and coalescence of “islands” until the entire gold surface is covered, or that upon reaching a critical size, the footprint of the “islands” ceases to grow and instead citrate desorption is favored. The agreement in the overall form of overlayer morphology between the two sets of simulations (“random” and “designed”), for all except the highest citrate surface density, may offer a positive validation of the simulation strategies employed, chiefly, that the two different types of starting configurations converged to give the same final result (Figure 2). At a surface density of 16 citrates, the “random” simulation yielded a morphology of higher energy with a corresponding disperse structure, which could be a metastable state. The more crowded nature of the aqueous citrate−gold interface with increasing citrate surface density could suggest a high potential energy barrier to a structural rearrangement at the end of the annealing stage in the “random” simulation protocol, namely moving from a surface-bound overlayer configuration to a more 3-D “island”-like morphology. We therefore suggest that future study of the interfacial structure of surfactants at higher surface densities may require the “random” simulation protocol to be refined further, possibly by including a round of unrestrained, moderate temperature, simulated annealing. Individual Molecules within an Overlayer. Unlike the model proposed for the adsorption to gold of either citric acid or dihydrogen citrate under acidic conditions,26−29 our simulations suggest little change to the internal structure of citrate upon adsorption (Tables S6−S8, Supporting Information). In agreement with our previous work, in which we probed the structure of citrate in solution,36 here we found that, in the most likely conformations adopted by an interfacial citrate, the anion was not internally hydrogen-bonded (Table S6, Supporting Information), but featured a fully extended carbon backbone (Table S7, Supporting Information) and was highly coordinated to Na+ counterions (Table S8, Supporting Information). These results were invariant to both the surface density of the overlayer to which the anion belonged, and its adsorption state. Our data also suggested that directly adsorbed citrates at the gold interface were preferentially oriented with their carboxylate groups pointing away from the surface (see Figure 3 for the 10-citrate case as an exemplar, data for all cases shown in Figure S7, Supporting Information; see also Figure S5, Supporting Information). In this geometry, the hydrophobic methylene groups resided in the region corresponding with the first trough in the vertical water density profile above Au(111), while the charged carboxylate groups penetrated the second layer of adsorbed water (Figure 3 and Figure S7, Supporting Information); favorable environments for both moieties. Variable Properties of the Overlayers. Subtle differences, both between the structures of the interfacial citrate− gold overlayer of a given composition identified by the two sets of simulations (“random” and “designed”), and between assemblies of different surface density, existed due to the

“System” refers to the number of citrates in the simulation cell. Data for the lowest energy configuration at a given surface density is highlighted in bold. The “random” 16-citrate system could not be classified; see text for details.

a

the configurations found by both types of simulations could be experimentally relevant states (Table S5, Supporting Information). Even for the lowest surface density, 8 citrates per simulation cell, this energy difference, on the order of ∼3 kBT per citrate, was reasonable on the basis of the fact that the potential energy landscape of an amorphous material should feature multiple minima, and thus could permit a range of structures to be kinetically stable under ambient conditions. Some discussion of our force-field is warranted. The GolPCHARMM force-field37 was designed to work in partnership with the biomolecule force-field CHARMM,40 and the modified TIP3P force-field41,42 for water. This force-field combination has been found to yield results that compare very favorably with first-principles simulations of the aqueous Au(111) interface,46,47 including those that made use of state-of-the-art nonlocal density functionals. For citrate, we have used our force-field derived for citrate in aqueous solution,36 and like GolP-CHARMM, this force-field was designed to be compatible with CHARMM and the modified TIP3P force-fields. While we recognize that the citrate ion was not part of our training set for the parametrization for GolP-CHARMM, we note that the relative binding free energy at the aqueous Au(111) interface of the carboxylate-containing amino acid aspartate, calculated using GolP-CHARMM, was found to be consistent with available experimental observations.37,48 The performance of the GolP-CHARMM force-field for other molecules not in our training set was also found to be very good overall.37 Therefore, while the suite of force-fields used in this study is certainly not definitive, we maintain they are consistent and appropriate for application to the aqueous gold− citrate system. Overlayer Morphology. Our simulations predicted that at the aqueous Au(111) interface the overall morphology of a citrate aggregate was either an “island”, i.e. assemblies which were approximately spherical in nature, or, a “stripe”, i.e. assemblies with approximate cylindrical geometry (Figure 2). A transition from “island” to “stripe” and back to “island” was observed as a function of increasing citrate surface density. However, while interpreting these findings, artifacts introduced by the finite size of the systems studied and the periodic boundary conditions employed (including the use of the NVT ensemble) must be considered. From a viewpoint of volume to surface area ratio, under periodic boundary conditions, one would expect “stripe” morphologies to become more favorable with respect to “island” morphologies for increasing citrate concentration. Since for at least the surface concentration of 14 citrates per simulation cell, “island” rather than “stripe” morphologies were observed to be stable, we hypothesize E

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

lines of interacting anions could extend through space and, in contrast to the “adsorbed linear chain” and “inverted bilayer” motifs, were not restricted to lying in the plane of the Au(111) surface only. These were subdivided into “straight chains”, “bent chains”, and “loops” (see Supporting Information Definitions of Citrate−Citrate Motifs section). The number of straight chain motifs in an overlayer became limited with increasing citrate concentration in the case of “island” morphologies, while “bent chains” were inherently less favorable in both “stripe” and low density assemblies (Table S9, Supporting Information). Taking these factors into account, the prominence of this type of linear-like citrate−citrate motif in an interfacial overlayer appeared to depend on overall overlayer morphology, with the average proportion of citrates in each overlayer belonging to a chain (straight or bent) or loop varying from 0.5−0.6 for “stripe” morphologies to 0.9−1.0 for “island” ones. Overlayer Stability. Despite the fact that the citrate overlayers identified in this work had differing properties, the radius of gyration of each was invariant with time (Figure 4).

Figure 3. Exemplar vertical mass density of water and Na+ ions above the Au(111) surface for the 10-citrate case. The mass density profile of water at the bare interface is also given (solid light blue line) for reference. The average surface separations of the different citrate functional groups for the directly adsorbed anions are denoted by the following vertical lines: carboxylate oxygen (solid light green), carboxylate carbon (dashed gold), methylene hydrogen (solid dark green), hydroxyl oxygen (dashed sienna), hydroxyl hydrogen (dashed salmon), and citrate COM (solid red). Data presented for lowest energy runs only.

amorphous nature of the systems studied. For example, across the lowest energy configurations for each surface density, there was no trend or consistency in the percentage population of citrates belonging to each of the three adsorption states: “directly adsorbed”, “indirectly adsorbed”, or “third-layer” (Table 2). However, these data were correlated with differences Table 2. Percentage Likelihood (%) of a Citrate Being Directly Adsorbed (A) or Indirectly Adsorbed (I), or in the Third-Layer of Molecules above the Au(111) Surface (3rd)a system 8 10 12 14 16

A 35 62 37 12 21

(5.55) (5.35) (5.66) (5.72) (6.38)

I 37 18 32 37 41

(7.59) (8.05) (8.13) (8.52) (8.78)

third 28 20 22 43 37

(12.21) (11.89) (12.99) (14.19) (11.53)

“System” refers to the number of citrates in the simulation cell. Average citrate−COM/surface distances (Å) for each adsorption state are given in parentheses. Data presented for the lowest energy runs only.

a

Figure 4. Radius of gyration of “island” (Gr) and “stripe” (Cr) citrate overlayers as a function of simulation time over the later stages of both “random” and “designed” runs. No data are shown for the “random” 16-citrate case, since the morphology could not be classified; see text for details.

in the density of water normal to the Au(111) surface (Figure 3 and Figure S7, Supporting Information). For example, in the case of the 10-citrate overlayer, where the relative population of directly adsorbed anions was the greatest, the second layer of interfacial water was the most depleted, relative to that above the bare aqueous Au(111) interface. A more pronounced third layer of structured water was present at the decorated interface, perhaps in compensation. On the other hand, the water density profile relative to the bare aqueous Au(111) interface was almost unperturbed by the presence of the cluster of 14 citrates (Figure S7, Supporing Information); this featured a minimal number of directly adsorbed anions. In addition to the relative population of the adsorption state, the frequency of occurrence and characteristics of the citrate− citrate motifs featured by the five low-energy overlayers (one for each surface density) varied considerably, although from visual inspection lines of interacting anions were observed in all cases. This motivated us to define a third set of motifs in which

This is indicative of overlayers with morphologies that had reached a steady state, such that large-scale structural changes to the overlayer as a whole were not occurring. Movement of individual citrates within each overlayer was also limited, with the rotational time autocorrelation functions of both adsorbed and third-layer anions showing negligible decay with time (Figure S6, Supporting Information). There are several factors that could contribute to the stability of an interfacial citrate− gold overlayer, including direct or indirect (mediated by the interfacial water) affinity for the Au(111) surface, and citrate− citrate attraction mediated by co-ordinating Na+ counterions. Herein, we have considered both factors. F

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

dense second structured water layer, the carboxylate groups of directly adsorbed citrates have the possibility to coordinate with a greater number of water molecules, compared with the indirectly adsorbed or third-layer citrates. The affinity of citrate for the gold surface cannot be the sole source of the stability of the interfacial citrate overlayer morphologies noted here. First, if this was the case, then for a given composition we would expect overlayers with a larger footprint to have the lowest potential energy. This was not always true (Table 1 and Table S5, Supporting Information). For instance, the designed/random assembly consisting of 12/ 14 citrates had the larger footprint out of the two simulations but also featured the highest energy. In addition, the rotational freedom of third-layer citrates was, for higher surface densities, on a par with that of the directly adsorbed species (Supporting Information Figure S5), thus suggesting that all three types of citrate (directly adsorbed, indirectly adsorbed, and third-layer) are in fact part of the overlayer and not just loosely associated with it. With this in mind we propose that Na+ counterion coordination also plays an important role in cluster stability, particularly supporting its 3-D morphology. By orienting in a manner such that Na+ ions can be shared by two or more citrates, the coordination of a citrate within the overlayer can far exceed the stoichiometric value for the system as a whole (Supporting Information Table S8). Not only is the attraction between the oppositely charged ions favorable in itself, but it also offsets the citrate−citrate repulsion which would otherwise be present at such small intermolecular separations. We propose that the discrepancies in citrate−gold geometry between our results and the experimental model for citrate−gold adsorption proposed previously26−29 can be accounted for by the lack of counterion-mediated stabilization in the latter system. At the acidic pH studied experimentally, if citrate was oriented with its carboxylate groups directed away from the gold surface, either protonation or citrate desorption (followed by protonation) would occur. Only a fully folded carbon backbone conformation (predicted to be an unfavorable state of citrate when free in solution36) would permit all three carboxylate groups to be in direct surface contact, and in addition would allow the citrate to remain adsorbed at the interface. Due to the amorphous nature of the citrate overlayers, it was not possible to identify a consistent set of citrate−citrate motifs that were common to all morphologies and surface densities. Instead we have used the basic building block, the citrate− citrate pair, to probe the stabilizing influence of the Na+ counterions (see Computational Methods). The average “quality” of a citrate−citrate pair (defined in eq 1) in overlayers with a surface density of 10−14 citrates per simulation cell was consistent, within error (Table 3). The high pair “quality” of the overlayer with 8 citrates per simulation cell (the overlayer with the smallest gold footprint, Table 1) could suggest that, in this case, citrate−Na+−citrate interactions contributed more to the overall stability of the overlayer than the presence of the gold surface. The converse may be true for the overlayer of 16 citrates, which features both the smallest average pair “quality” and the largest gold footprint. Implications for Biomolecule Adsorption. It is noteworthy that, for all citrate surface densities probed here, a significant proportion of the gold surface remained undecorated (Figure 2, Table 1). This may be especially important when considering the adsorption of biomolecules to a citrate-coated gold surface under aqueous conditions, as our predictions

Adsorption of a single citrate to the aqueous bare Au(111) interface is predicted to be energetically favorable. From our metadynamics simulations we estimate the free energy of interfacial adsorption to be −8.2 kJ mol−1 (Figure 5). Two

Figure 5. Free energy of adsorption profile of a single citrate at the aqueous Au(111) interface calculated using well-tempered metadynamics. The distance is defined as the citrate-COM−gold vertical separation. The mass density of liquid water normal to the bare Au(111) surface (dashed blue line) is given for reference.

adsorbed states were apparent for the isolated anion, at COM− surface separations of 4.7 and 7.3 Å respectively, with a free energy barrier of 7.9 kJ mol−1 between them (see Figure S8 in the Supporting Information for data regarding the number of barrier recrossings between these two minima). The positions of these two adsorption states correlate well with the average COM−surface distances of the directly and indirectly adsorbed citrates found within the multiple molecule overlayers (Table 2). As for comparison between our metadynamics findings and directly adsorbed citrates within the overlayer, in its lowest energy bound state a single citrate is oriented such that its carboxylate groups penetrate the second layer of structured water above the Au(111) surface, while the hydrophobic methylene moieties are shielded from solution, residing in the depleted water region between the first and second water layers. We propose that this geometry is favorable due to the structuring of water molecules within the second interfacial layer, a feature which is lessened in the presence of more than one citrate (Figure 3 and Figure S7, Supporting Information). The citrate−surface geometry of the second binding mode, identified for both the single-molecule case (at a citrate−surface separation of 7.3 Å) and when part of an overlayer (Supporting Information Figure S4), is more ambivalent, however, with carboxylate groups either penetrating into the second structured water layer below it, or pointing toward the liquid water (of approximately bulk density) above it. It is noteworthy that the attraction to Au(111) in both the direct and indirect adsorption modes was in part mediated by the interfacial water structuring rather than via direct contact with the surface gold atoms. This was particularly apparent when the average number of carboxylate−water hydrogen bonds per citrate was considered for each overlayer (Table S10, Supporting Information). Specifically, the overlayers with a surface density 8−12 citrates per simulation cell on the whole featured a greater number of such hydrogen bonds than the overlayers with higher surface densities. This is consistent with the trend for the former to feature a higher proportion of directly adsorbed citrates than the latter. By penetrating the G

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Table 3. Average Number of Citrate−Citrate “Pairs” ⟨npairs⟩ in Each Overlayera system

⟨npairs⟩

⟨C3···C3⟩

⟨nNA+⟩

8 10 12 14 16

10 14 17 25 29

0.47 0.45 0.44 0.49 0.48

4.0 2.2 2.3 2.3 1.6

such that the carboxylate groups pointed toward liquid water, and was invariant with overlayer composition. The stability of each overlayer was a complex interplay between citrate−gold binding and counterion mediated citrate−citrate interactions. All overlayers were found to possess three-dimensional shapes, and thus some regions of the gold surface remained undecorated. In terms of how biomolecules might interact with citrate-decorated aqueous gold interfaces, our predictions suggested that small biomolecules may coadsorb at the interface, rather than, or as well as, displace the adsorbed citrates. These results imply that it may be important to explicitly incorporate the presence of citrate, where relevant, when modeling biomolecule−gold adsorption experiments.

⟨Q⟩ 0.52 0.25 0.22 0.27 0.16

± ± ± ± ±

0.04 0.02 0.02 0.02 0.01

“System” refers to the number of citrates per simulation cell. Also shown is the average citrate−citrate distance ⟨C3···C3⟩ (nm) and number of shared coordinating Na+ ions ⟨nNA+⟩ for each “pair” formed. The overall average “aQuality” of the pair-wise interactions (⟨Q⟩, see Computational Methods) is also given. Data presented for lowest energy runs only. a

■

ASSOCIATED CONTENT

S Supporting Information *

Supplementary figures and tables, including further details about the simulation protocols used; definitions of the different classes of citrate−citrate motif identified and of the quantities used to determine the extent of the overall overlayer; the average potential energy of each of the overlayers; the most likely internal conformational state of an individual citrate within a cluster; key characteristics of citrate−citrate motifs; analysis of citrate−gold orientations sampled; the average number of citrate−water hydrogen bonds formed; the rotational time autocorrelation functions for directly adsorbed and third-layer citrates; evolution of the metadynamics collective variable as a function of simulation time for citrate and arginine in the presence of the aqueous Au(111) interface; and the calculated free energy profile for a single arginine adsorbing to the aqueous Au(111) interface. This material is available free of charge via the Internet at http://pubs.acs.org.

suggest that two possible binding modes could be viable: one in which the two species, citrate and the biomolecule, coexist, and the other in which citrate is displaced by the adsorbate. For most biomolecules, i.e., peptides of moderate length, proteins, and nucleic acids, one could envisage that, for steric arguments, the second mode may be more probable due to the lateral footprint of both the overlayer itself and its accompanying solvation shell (Table 1). The case is not so clear for smaller adsorbates such as amino acids. For this reason, the adsorption of arginine to citrate-capped AuNPs,1,2 reported by Sethi and Knecht, is of particular interest. In fact the authors proposed that the driving force behind the formation of the linear AuNP chains, generated from the addition of excess arginine to citratestabilized AuNPs, was the presence of both arginine-rich49 and citrate-rich surface patches on the AuNPs. Both of these hypotheses are amenable to future investigation in future. While it is not within the scope of the current study to fully investigate the adsorption of arginine to the surface of a citratecapped AuNP, we have calculated the free energy of adsorption of a single zwitterionic arginine at the aqueous Au(111) interface under conditions corresponding with neutral pH (see Figures S8 and S9 in the Supporting Information). By doing so we aimed to identify whether both modes of adsorption to a citrate-coated gold surface, coexistence and displacement, were plausible. Specifically, although the free energy of adsorption of a cluster of molecules, citrate or arginine, is not a simple multiple of the binding affinity of a single ion, it can be used to indicate if citrate displacement by arginine is possible. If a single citrate anion adsorbs more strongly at the aqueous Au(111) interface compared to a single arginine cation, then the coexistence binding mode is more probable. However, our findings indicated that the converse was true, with ΔGarg = −15.7 kJ mol−1 < ΔGcit = −8.2 kJ mol−1 (Figure 5, and Figure S9, Supporting Information), suggesting that both the displacement and coexistence binding modes may be possible. Future simulations that explicitly model the adsorption of multiple arginines to a citrate-coated gold surface may be able to discriminate between the two processes.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: tiff[email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors gratefully acknowledge the computing facilities of the Centre for Scientific Computing, University of Warwick, and the MidPlus Regional Centre of Excellence for Computational Science, Engineering and Mathematics (under EPSRC grant EP/K000128/1). L.B.W. thanks the EPSRC for DTA studentship funding. This work was supported by EPSRC Programme Grant (EP/I001514/1) “Hard-Soft Matter Interfaces; From Understanding to Engineering”, and partially supported by the Air Force Office of Scientific Research (Grant #FA9550-12-1-0226). T.R.W. thanks veski for an Innovation Fellowship.

■

■

REFERENCES

(1) Sethi, M.; Knecht, M. R. Experimental Studies on the Interactions between Au Nanoparticles and Amino Acids: Bio-Based Formation of Branched Linear Chains. ACS Appl. Mater. Interfaces 2009, 1, 1270− 1278. (2) Sethi, M.; Knecht, M. R. Understanding the Mechanism of Amino Acid-Based Au Nanoparticle Chain Formation. Langmuir 2010, 26, 9860−9874. (3) Calzolai, L.; Franchini, F.; Gilliland, D.; Rossi, F. ProteinNanoparticle Interaction: Identification of the Ubiquitin-Gold Nanoparticle Interaction Site. Nano Lett. 2010, 10, 3101−3105.

CONCLUSIONS In summary, in this work we have predicted the key characteristics of amorphous overlayers of citrate adsorbed at the aqueous Au(111) interface, using atomistic molecular dynamics simulations. A transition in overall overlayer morphology from “island” to “stripe” and back to “island” was observed with increasing surface citrate concentration. The orientational preference of those citrates closest to interface was H

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

(4) Gomes, I.; Feio, M. J.; Santos, N. C.; Eaton, P.; Serro, A. P.; Saramago, B.; Pereira, E.; Franco, R. Controlled Adsorption of Cytochrome c to Nanostructured Gold Surfaces. J. Nanopart. Res. 2012, 14, 1321. (5) Zhang, X.; Servos, M. R.; Liu, J. Surface Science of DNA Adsorption onto Citrate-Capped Gold Nanoparticles. Langmuir 2012, 28, 3896−3902. (6) Hnilova, M.; Khatayevich, D.; Carlson, A.; Oren, E. E.; Gresswell, C.; Zheng, S.; Ohuchi, F.; Sarikaya, M.; Tamerler, C. Single-Step Fabrication of Patterned Gold Film Array by an Engineered MultiFunctional Peptide. J. Colloid Interface Sci. 2012, 365, 97−102. (7) Hnilova, M.; So, C. R.; Oren, E. E.; Wilson, B. R.; Kacar, T.; Tamerler, C.; Sarikaya, M. Peptide-Directed Co-Assembly of Nanoprobes on Mutlimaterial Patterned Surfaces. Soft Matter 2012, 8, 4327− 4334. (8) Hnilova, M.; Karaca, B. T.; Park, J.; Jia, C.; Wilson, B. R.; Sarikaya, M.; Tamerler, C. Fabrication of Hierarchical Hybrid Structures Using Bio-Enabled Layer-by-Layer Self-Assembly. Biotechnol. Bioeng. 2012, 109, 1120−1130. (9) Causa, F.; Modlie, R. D.; Iaccino, E.; Mimmi, S.; Marasco, D.; Scognamiglio, P. L.; Battista, E.; Palmieri, C.; Cosenza, C.; Sanguigno, L.; Quinto, I.; Scala, G.; Netti, P. A. Evolutionary Screening and Adsorption Behaviour of Engineered M13 Bacteriophage and Derived Dodecapeptide for Selective Decoration of Gold Interfaces. J. Colloid Interface Sci. 2013, 389, 220−229. (10) Chakraborty, M.; Jain, S.; Rani, V. Nanotechnology: Emerging Tool for Diagnostics and Therapeutics. Appl. Biochem. Biotechnol. 2011, 165, 1178−1187. (11) Mahmood, M.; Casciano, D.; Xu, Y.; Biris, A. S. Engineered Nanostructural Materials for Application in Cancer Biology and Medicine. J. Appl. Toxicol. 2012, 32, 10−19. (12) Kumar, A.; Ma, H. L.; Zhang, X.; Huang, K.; Jin, S. B.; Liu, J.; Wei, T.; Cao, W. P.; Zou, G. Z.; Liang, X. J. Gold Nanoparticles Functionalized with Therapeutic and Targeted Peptides for Cancer Treatment. Biomaterials 2012, 33, 1180−1189. (13) Colombo, M.; Mazzucchelli, S.; Collico, V.; Avvakumova, S.; Pandolfi, L.; Corsi, F.; Porta, F.; Prosperi, D. Protein-Assisted One-Pot Synthesis and Biofunctionalization of Spherical Gold Nanoparticles for Selective Targeting of Cancer Cells. Angew. Chem., Int. Ed. 2012, 51, 9272−9275. (14) Turkevich, J.; Stevenson, P. C.; Hillier, J. A Study of the Nucleation and Growth Processes in the Synthesis of Colloidal Gold. Discuss. Faraday Soc. 1951, 11, 55−75. (15) Frens, G. Controlled Nucleation for Regulation of Particle-Size in Monodisperse Gold Suspensions. Nature, Phys. Sci. 1973, 241, 20− 22. (16) Brinas, R. P.; Maetani, M.; Barchi(Jr), J. J. A Survey of PlaceExchange Reaction for the Preparation of Water-Soluble Gold Nanoparticles. J. Colloid Interface Sci. 2013, 392, 415−421. (17) Smith, B. D.; Liu, J. W. Assembly of DNA-Functionalised Nanoparticles in Alcoholic Solvents Reveals Opposite Thermodynamic and Kinetic Trends for DNA Hybridization. J. Am. Chem. Soc. 2010, 132, 6300−6301. (18) Xu, L. G.; Zhu, Y. Y.; Ma, W.; Chen, W.; Liu, L. Q.; Kuang, H.; Wang, L. B.; Xu, C. L. New Synthesis Strategy for DNA Functional Gold Nanoparticles. J. Phys. Chem. C 2011, 115, 3243−3249. (19) Ojea-Jimenez, I.; Puntes, V. Instability of Cationic Gold Nanoparticle Bioconjugates: The Role of Citrate Ions. J. Am. Chem. Soc. 2009, 131, 13320−13327. (20) Vecchio, G.; Galeone, A.; Brunetti, V.; Maiorano, G.; Sabella, S.; Cingolani, R.; Pompa, P. P. Concerntration-Dependent, SizeIndepedent Toxicity of Citrate Capped AuNPs in Drosophila melanogaster. PLoS one 2012, 7, e29980. (21) Vecchio, G.; Galeone, A.; Brunetti, V.; Maiorano, G.; Rizzello, L.; Sabella, S.; Cingolani, R.; Pompa, P. P. Mutagenic Effects of Gold Nanoparticles Induce Aberrant Phenotypes in Drosophila melanogaster. Nanomed. Nanotechnol. Biol. Med. 2012, 8, 1−7.

(22) Tournebize, J.; Boudier, A.; Joubert, O.; Eidi, H.; Bartosz, G.; Maincent, P.; Leroy, P.; Sapin-Minet, A. Impact of Gold Nanoparticle Coating on Redox Homeostasis. Int. J. Pharm. 2012, 438, 107−116. (23) Mirau, P. A.; Naik, R. R.; Gehring, P. Structure of Peptides on Metal Oxide Surfaces Probed by NMR. J. Am. Chem. Soc. 2011, 133, 18243−18248. (24) Ju, S.; Yeo, W. S. Quantification of Proteins on Gold Nanoparticles by Combining MALDI-TOF MS and Proteolysis. Nanotechnology 2012, 23, 135701. (25) Brancolini, G.; Kokh, D. B.; Calzolai, L.; Wade, R. C.; Corni, S. Docking of Ubiquitin to Gold Nanoparticles. ACS Nano 2012, 6, 9863−9878. (26) Lin, Y.; Pan, G. P.; Su, G. J.; Fang, X. H.; Wan, L. J.; Bai, C. L. Study of Citrate Adsorbed on the Au(111) Surface by Scanning Probe Microscopy. Langmuir 2003, 19, 10000−10003. (27) Floate, S.; Hosseini, M.; Arshadi, M. R.; Ritson, D.; Young, K. L.; Nichols, R. J. An In-Situ Infrared Spectroscopic Study of the Adsorption of Citrate on Au(111) Electrodes. J. Electroanal. Chem. 2003, 542, 67−74. (28) Nichols, R. J.; Burgess, I.; Young, K. L.; Zamlynny, V.; Lipkowski, J. A Quanititive Evaluation of the Adsorption of Citrate on Au(111) Using SNIFTIRS. J. Electroanal. Chem. 2004, 563, 33−39. (29) Kunze, J.; Burgess, I.; Nichols, R.; Buess-Herman, C.; Lipkowski, J. Electrochemical Evaluation of Citrate Adsorption on Au(111) and the Stability of Citrate-Reduced Gold Colloids. J. Electroanal. Chem. 2007, 599, 147−159. (30) Barducci, A.; Bussi, G.; Parrinello, M. Well-Tempered Metadynamics: A Smoothly Converging and Tunable Free-Energy Method. Phys. Rev. Lett. 2008, 100, 020603. (31) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. J. Chem. Theory Comp. 2008, 4, 435−447. (32) Nosé, S. A Molecular-Dynamics Method for Simulations in the Canonical Ensemble. Mol. Phys. 1984, 52, 255−268. (33) Hoover, W. G. Canonical Dynamics-Equilibrium Phase-Space Distributions. Phys. Rev. 1985, 31, 1695−1697. (34) Hockney, R. W.; Goel, S. P.; Eastwood, J. W. Quiet HighResolution Computer Models of a Plasma. J. Comput. Phys. 1974, 14, 148−158. (35) Darden, T.; York, D.; Pedersen, L. Particle Mesh Ewald: An N log(N) Method for Ewald Sums in Large Systems. J. Chem. Phys. 1993, 98, 10089−10092. (36) Wright, L. B.; Rodger, P. M.; Walsh, T. R. Aqueous Citrate: A First-Principles and Force-Field Molecular Dynamics Study. RSC Adv. 2013, 3, 16399−16409. (37) Wright, L. B.; Rodger, P. M.; Corni, S.; Walsh, T. R. GolPCHARMM: First Principles Based Force-Fields for the Interaction of Proteins with Au(111) and Au(100). J. Chem. Theory Comput. 2013, 9, 1616−1630. (38) Iori, F.; di Felice, R.; Molinari, E.; Corni, S. GolP: An Atomistic Force-Field To Describe the Interaction of Protiens with Au(111) Surfaces in Water. J. Comput. Chem. 2009, 30, 1465−1476. (39) Wright, L. B.; Freeman, C. L.; Walsh, T. R. Benzene Adsorption at the Aqueous (011) α-Quartz Interface: Is Surface Flexibility Important? Mol. Simulat. 2013, 39, 1093−1102. (40) MacKerell, A. D.; Bashford, D.; Bellot, M.; Dunbrack, R.L.; Evanseck, J.D.; Field, M.J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; et al. All Atom Empirical Potential for Molecular Modelling and Dynamics Studies of Proteins. J. Phys. Chem. B 1998, 102, 3586−3616. (41) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−935. (42) Neria, E.; Fischer, S.; Karplus, M. Simulation of Activation Free Energies in Molecular Systems. J. Chem. Phys. 1996, 105, 1902−1921. (43) Hughes, Z. E.; Walsh, T. R. Structure of the Electrical Double Layer at Aqueous Gold and Silver Interfaces for Saline Solutions. J. Colloid Interf. Sci. 2014, 436, 99−110. (44) Bonomi, M.; Branduardi, D.; Bussi, G.; Camilloni, C.; Provasi, D.; Raiteri, P.; Donadio, D.; Marinelli, F.; Pietrucci, F.; Broglia, R. A.; I

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Parrinello, M. PLUMED: A Portable Plugin for Free-Energy Calculations with Molecular Dynamics. Comput. Phys. Commun. 2009, 180, 1961−1972. (45) Jedlovszky, P.; Brodholt, J. P.; Bruni, F.; Ricci, M. A.; Soper, A. K.; Vallauri, R. Analysis of the Hydrogen-Bonded Structure of Water from Ambient Supercritical Conditions. J. Chem. Phys. 1998, 108, 8528−8540. (46) Cicero, G.; Calzolari, A.; Corni, S.; Catellani, A. Anomalous Wetting Layer at the Au(111) Surface. J. Phys. Chem. Lett. 2011, 2, 2582−2586. (47) Nadler, R.; Sanz, J. F. Effect of Dispersion Correction on the Au(111)-H2O Interface: A First-Prinicples Study. J. Chem. Phys. 2012, 137, 114709. (48) Palafox-Hernandez, J. P.; Tang, Z.; Hughes, Z. E.; Li, Y.; Swihart, M. T.; Prasad, P. N.; Walsh, T. R.; Knecht, M. R. Comparative Study of Materials-Binding Peptide Interactions with Gold and Silver Surfaces and Nanostructures: A Thermodynamic Basis for Biological Selectivity of Inorganic Materials. Chem. Mater. 2014, 26, 4960−4969. (49) Wright, L. B.; Merrill, N. A.; Knecht, M. R.; Walsh, T. R. Structure of Arginine Overlayers at the Aqueous Gold Interface: Implications for Nanoparticle Assembly. ACS Appl. Mater. Inter. 2014, 6, 10524−10533.

J

dx.doi.org/10.1021/la503690t | Langmuir XXXX, XXX, XXX−XXX