J. Phys. Chem. B 1997, 101, 9767-9773
9767
Molecular Conformation and Solvation of Oligo(ethylene glycol)-Terminated Self-Assembled Monolayers and Their Resistance to Protein Adsorption R. L. C. Wang and H. J. Kreuzer* Department of Physics, Dalhousie UniVersity, Halifax, N.S. B3H 3J5, Canada
M. Grunze Angewandte Physikalische Chemie am Physikalisch-Chemischen Institut der UniVersita¨ t Heidelberg, Im Neuenheimer Feld 253, D-69120 Heidelberg, Germany ReceiVed: May 21, 1997; In Final Form: August 20, 1997X
We study the interaction of water with oligo(ethylene glycol) (OEG)-terminated alkane thiolate self-assembled monolayers (SAMs) in the helical and the planar “all-trans” form with ab initio methods to determine the stability and density distribution in water clusters as a function of molecular conformation. We find that the amphiphilic behavior of the OEG moieties depends on their molecular conformation and that the energetics of water adsorption is dominated by electrostatic interactions. The SAM surface of helical OEG provides a template for water nucleation, whereas water is not stable on a surface of planar OEG strands. These results suggest that protein resistance of OEG-terminated self-assembled monolayers is a consequence of the stability of the interfacial water layer, which prevents direct contact between the surface and the protein.
I. Introduction A recent study1 correlated the molecular conformation in oligo(ethylene glycol) (OEG)-terminated self-assembled alkane thiolate monolayers (SAMs) of 1-undecanethiolates with a methoxy-terminated tri(ethylene glycol) (EG3-OMe) end group on polycrystalline gold and silver surfaces with their ability to resist protein adsorption. It was found that only the helical conformation of OEG, which forms on gold surfaces, showed protein resistance, whereas the more densely packed planar phasesobserved on silversadsorbed up to 60% of a monolayer of fibrinogen, the model protein used in this study. This observation suggested that other factors than the conformational flexibility of the polymer chains in poly(ethylene glycol) (PEG)2-5 can cause a surface to resist protein adsorption, and it was speculated1 that protein resistance in the OEG-terminated SAMs is correlated to the ability of the different conformations to bind water. In this paper we follow up on this idea and study the interaction of water with OEG moieties in the helical an the planar “all-trans” conformations using ab initio calculations at the Hartree-Fock level for OEG clusters containing up to 12 methoxy-terminated EG3-OMe strands and 20 water molecules. To put our present study into the context of the experimental and theoretical work on PEG6 and on self-assembled monolayers terminated with OEG, we will first briefly summarize the pertinent literature. From the PEG/water phase diagram it is known that the low-temperature hydrated crystalline phase is predominantly helical.7 This is the result of glycol moieties with a trans conformation around the C-O bonds and a gauche conformation around the C-C bond.8 At the crystalline/ amorphous phase transition temperature of about 60 °C dehydration occurs. In the amorphous phase the predominant conformation around the C-C bond is still gauche, but the C-O bond can be trans or gauche and the sign of the C-C gauche rotational angle is not uniform.9 A planar “all-trans” conformation is only observed in stretched PEO.10 The conformations of EG3-OMe tails in the EG3-OMe undecanethiolate SAM are considered1 to be similar to those X
Abstract published in AdVance ACS Abstracts, November 1, 1997.
S1089-5647(97)01695-7 CCC: $14.00
formed in the PEG crystals. The helical crystalline form of EG3-OMe was found by Fourier transform infrared reflectionabsorption spectroscopy (FTIRAS) to be stable on Au(111) oriented surfaces where the sulfur head groups form a hexagonal commensurate (x3×x3)R30° overlayer with an S‚‚‚S spacing of 4.97 Å and an idealized (for a defect-free film) packing density of 21.4 Å2 per thiolate.11 This area can accommodate a helical OEG moiety oriented parallel to the surface normal with a molecular cross section of 21.3 Å2. In addition to the helical phase, a different phase, in which the molecules show (in the FTIRAS spectra) a predominantly “all-trans” planar conformation, can be prepared by selfassembly on a silver surface, which results in a higher packing density in the film, as confirmed by X-ray photoelectron spectroscopy (XPS).1 On Ag(111) surfaces, the S‚‚‚S spacing for long chain alkane thiols is smaller (4.67 ( 0.03 Å)12 than on Au(111). The OEG-terminated alkanethhiolates assume an orientation close to perpendicular to the surface plane. The available surface area per molecule of 18.4 Å2 corresponds to the molecular cross section of the alkyl chains. Since the molecular cross section of helical PEG is larger, this conformation cannot be accommodated, and the EG3-OMe OEG films assume the predominantly planar zigzag conformation, which ideally has a cross section of 17.1 Å2. The FTIRAS and XPS measurements described in ref 1 were done on polycrystalline but preferentially (111) oriented gold and silver surfaces and showed that the helical form of tri(ethylene glycol) (EG3-OMe) prepared on gold is protein resistant, whereas the more densely packed “all-trans” planar form on silver adsorbs protein. As reported previously by Prime and Whitesides,13 it was also observed that the ability of the helical form to resist protein adsorption is not sensitive to defects in the SAM layer within substantial variations (up to ca. 35% defect density). Protein resistance of the densely packed OEG functionalized alkane thiolate films apparently does not require the degree of conformational freedom of a polymer chain as expected from the “steric repulsion” theory formulated for PEG.2-5 In this phenomenological model the force balance between steric © 1997 American Chemical Society
9768 J. Phys. Chem. B, Vol. 101, No. 47, 1997 repulsion, van der Waals attraction, and hydrophobic interaction free energies between the protein and the PEG surface is considered. It was found that the net force determining whether adsorption occurs or not depends on the chain length and surface density of PEG. The attractive van der Waals and hydrophobic forces are balanced by the steric repulsion, which contains an osmotic (due to the hydration of the PEG chain) and an elastic component, becoming effective when the protein reaches the interphase by diffusion and compresses the PEG layer. The van der Waals contribution to the attractive force was found to be small compared to the hydrophobic interaction between the protein and the hydrophobic surface, so only the latter was considered as competing with the steric repulsion. The association of water with the PEG chains is essential for protein resistance since it gives rise to the dominant steric repulsion force, and therefore the details of the water interaction with PEG is of interest in the context of protein resistance. The phase diagram of the PEG/water system has been studied both experimentally and theoretically and is characterized by a closed immiscibility loop.14 On the basis of proton and deuteron NMR relaxation time measurements, Lu¨sse and Arnold15 have been able to show that a maximum of one water molecule can be bound by a PEG repeat unit (-CH2CH2-O). At lower water contents, H2O bridge-bonds via two hydrogen bonds between the oxygen atoms of two neighboring PEG chains. The difference between the energy of one and two hydrogen bonds was found to be 34 kJ/mol. Since PEG is basically a hydrophobic polymer and therefore hydration causes an increase in water-water hydrogen bonds and hence a lowering of the entropy, entropy factors are important in the discussion of the PEG/PEG and PEG/protein interactions. Bringing two hydrophobic surfaces together, i.e., expelling the water adsorbed onto the two surfaces, will accordingly result in an increase in entropy (∆S > 0) and a decrease in enthalpy (∆H < 0), resulting in a decrease in free energy (∆G < 0). Hence, two hydrophobic surfaces are expected to stick to each other in aqueous solution. However, in the case of PEG a modified picture applies since the ether oxygen can interact via hydrogen bonds with water. Kjellander and Florin-Robertsson16 explained the phase diagram of the PEG/water system by the temperature-dependent enthalpy and entropy of hydration of PEG and the ideal entropy of mixing PEG with water. This model was found to agree with the molecular dynamics simulation by Tasaki.17 Karlstro¨m18 suggested a model based on molecular interactions in which the high dipole moment of the segment in the helical (gauchetrans-gauche) conformation causes a strong interaction with water (or other polar solvents) and that temperature-induced conformational changes in the PEG chain lead to conformations with smaller dipole moments and consequently weaker water interaction energies. It is important to note that in the two models different physical mechanisms determine the PEG/PEG and PEG/water interactions. In the former, the free energy of hydration and mixing determines the interaction, whereas in the latter the temperatureinduced changes in the polymer conformation and resulting changes in the conformation-dependent dipole moments of PEG drive the phase transition. The intention of the present work is (i) to identify water adsorption sites along the OEG strands, (ii) to decipher the differences in solvation between planar and helical OEG, (iii) to see whether a small cluster of OEG can act as a template for the nucleation of a water film which consequently might prevent direct contact between the protein and the surface, and (iv) to extract an intuitive picture that relates our results to concepts
Wang et al. in the phenomenological models derived to explain the protein resistance of grafted PEG chains.2-5 In the next section we outline the methods used to perform the ab initio calculations and discuss the reliability of the results. Section 3 deals with the water dimer, and section 4 presents results for water adsorption on a single strand of planar and helical OEG. Section 5 contains our main result for water adsorption on a cluster of OEG-terminated self-assembled monolayers, which are discussed in detail in section 6 in connection with resistance to protein adsorption. II. Methods To investigate the interaction of water with OEG-terminated SAMs, there are two avenues open to us. (i) We can use ab initio quantum mechanical methods restricting us to small clusters of molecules, or (ii) we can fit the interactions between the water molecules and with the various components of the SAM molecules by empirical potentials and do molecular dynamics calculations on larger clusters. As discussed (again) lately,19 the latter approach cannot account quantitatively for hydrogen bonding, and by employing it we would forego essential conformational effects that we will see are important in this system. Therefore we decided to employ ab initio methods in this study. We will first describe the interaction of a single water molecule with a single strand of OEG and then proceed to larger clusters with a lattice of 12 OEGs and up to 20 water molecules. As a zeroth approximation, we assume that the OEG strand is either in its ideal planar or ideal helical conformation. Our aim is to make statements about the interaction strength and also about the structure of water in contact with the SAM and to correlate our calculations with the experimental observations reported in ref 1. While most recent studies20-22 of hydrogen-bonded systems by ab initio methods use the monoreference formulation of the MP2 method to obtain more precise results including correlation effects, molecular orbital calculations with an adequate basis set at the Hartree-Fock level still provide a good enough description of the hydrogen bond energies at the equilibrium geometries.23,24 Our study of water adsorption on OEG is based on Hartree-Fock calculations with a 6-31G* basis set using the GAUSSIAN 94 suite of programs.25 We have done selected MP2 calculations to check the accuracy of our results. III. Water Dimer To get a reference point for our calculations and to introduce some notation, we first look at the hydrogen bond in the water dimer. We consider it as the adsorption of a water molecule, the adsorbate, on another water molecule, the substrate. There are two bonding geometries, which we will call the O-mode and the H-mode, respectively. In the O-mode the O atom of the adsorbate water molecule approaches an H atom of the substrate water molecule forming a hydrogen bond; see the geometry in the lower right corner of Figure 1. On the other hand for the H-mode a H atom of the adsorbate water molecule approaches the O atom of the substrate water molecule, also forming a hydrogen bond; see the geometry in the upper half of Figure 1. For the water dimer these two modes of course lead to the same dimer structure, but this will no longer be the case for water adsorption on an OEG. The Hartree-Fock calculations with a 6-31G* basis set give the adsorption distance between the two O atoms in the water dimer as 2.97 Å and the hydrogen bond energy as 242 meV or 5.58 kcal/mol, comparing well with the experimental values of 2.96 Å and 5.4 kcal/mol, respectively, and previous HF calculations.24,26 Adding MP2 correction, this value increases to 310 meV, again in agreement with the previous study.26 We
Oligo(ethylene glycol)-Terminated SAMs
J. Phys. Chem. B, Vol. 101, No. 47, 1997 9769 TABLE 1: F1 and F2 Are the Electrostatic Fields at the O-H1 and O-H2 Bonds of Water. Ead Is the Calculated Adsorption (Binding) Energy, and Edip Is Its Dipolar Estimate substrate
Figure 1. Electrostatic potential generated by the (all black) water molecule at the center acting on the two additional water molecules (shown with smaller circles) at their equilibrium positions for hydrogen bonding with the center water molecule. The upper (lower) geometry we call H-mode (O-mode) adsorption. Solid lines are for -1.5, -1.0, and -0.5 eV and dashed lines for 0.5, 1.0, and 1.5 eV. Atoms depicted as full, shaded, and open circles are in, below, and above the plane, respectively.
can and will take the difference between the HF and the MP2 results as a measure of the error bars on the calculated binding and bond energies. We point out that the average bond energy in liquid water is smaller than that in the dimer: the heat of vaporization is 40 kJ/mol and the average number of bonds per molecule is 3.5 so that the energy per bond in liquid water is only 100-150 meV, i.e. much smaller that the binding energy of the water dimer. The water molecule is easily polarizable. A Mulliken population analysis of the Hartree-Fock results gives net charges of -0.869e on the O atom and 0.434e on each H atom and a dipole moment of 2.2 D. To highlight the role of electrostatic interactions in the formation of a hydrogen bond, we show in Figure 1 the electrostatic field around a single water molecule into which a second water molecule is attracted until, at the equilibrium distance, repulsion due to electron overlap prevents a closer approach. We can determine the average electrostatic field due to one (substrate) water molecule at and in the direction of the O-H1 and O-H2 bonds of the second (adsorbate) water molecule as F1 ) ∆V1/L and F2 ) ∆V2/L, respectively, where L ) 0.95 Å is the O-H bond length and ∆V1 and ∆V2 are the potential differences over the distance of the bond. Assuming that the dipole moment of a water molecule consists of two dipole moments, D ) 1.82 D, pointing from the O atom to each H atom, we can calculate the electrostatic energy of the second molecule in the electric field of the first as Edip ) D(F1 + F2). This energy accounts for 90% of the energy of the dimer; see Table 1, where we also list the field strengths, which are on the order of 1 V per angstrom. To avoid misunderstanding, we remark that these electrostatic forces only account for the attractive interaction between the two water molecules with no account given for the repulsive forces (due to electron-electron Coulomb forces and Pauli exclusion) that together determine the equilibrium geometry and vibrational properties. IV. Water Adsorption on a Single OEG Next we consider the adsorption of a single water molecule on a single OEG strand in both the planar and the helical configurations (with the nuclear coordinates given in refs 8 and 10). The Hartree-Fock calculations for a single methoxyterminated OEG yield net charges on the methoxy group H atoms and on the topmost O atom of 0.178e and -0.622e for the planar OEG and 0.178e and -0.624e for the helical structure.
F1 (V/Å)
F2 (V/Å)
Fdip (meV)
Ead (meV)
H2O OEG 3 OEGb hel OEGc 4 hel OEGd
0.76 0.59 0.06 0.87 0.65
H-mode -0.18 -0.06 0.016 -0.11 -0.02
222 202 28 287 240
242 (310)a 185 (230) 11 323 (440) 297e (380)
H2O OEG 3 OEG
0.29 0.040 0.046
O-mode 0.29 0.040 0.046
222 30 35
242 (310) 52 76
a Numbers in parentheses from MP2. b Three OEGs seperated by 5 Å. c On the bridge of two O atoms of a helical OEG. d Four helical OEG to mimic a hexagonal lattice with lattice constant 5 Å. e Including another O-mode hydrogen bond; see text.
Figure 2. Electrostatic potential generated by a planar OEG strand acting on two water molecules (smaller circles) at their potential adsorption positions for hydrogen bonding with the OEG strand. Solid lines are for -1.0, -0.5, -0.2, and -0.1 eV and dashed lines for 0.1, 0.2, 0.5, and 1.0 eV. Otherwise as Figure 1.
These net charges and also the complete charge distributions around the top of the OEG are almost the same for the two configurations, because they differ little geometrically at the top. Because hydrogen bonds are mainly due to electrostatic interactions, we can expect that water adsorption on the methoxy group of either configuration is similar and rather weak. Indeed, Figure 2 shows that for planar OEG the electrostatic field surrounding the top H atoms is much weaker than that around a H atom in the water molecule (Figure 1) and that the field surrounding the last O atom of a planar OEG is, although a little weaker, similar to the field surrounding the O atom of a water molecule. The adsorption energies for a single water molecule on planar OEG are only 52 meV for the O-mode on the methoxy group with the O-H distance being 2.83 Å. It is 185 meV for the H-mode at the topmost O atom of the OEG with the H-O distance being 2.04 Å; see Table 1, where we also list the average field strengths. For the adsorption of a single water molecule on helical OEG, we get almost the same result for O-mode adsorption as for planar OEG and also for H-mode adsorption involving only one O atom of OEG. However, for helical OEG there is another H-mode adsorption geometry involving two O atoms of OEG. For the three-dimensional helical OEG, the dihedral angle between two planes, O1-C-C and C-C-O2, of the chain O1-C-C-O2 involving two successive O atoms is only about 90°, whereas it is 180° for a planar OEG. This results in a substantial overlap of the electrostatic fields originating from the O1 and O2 atoms, leading to field enhancement in the region
9770 J. Phys. Chem. B, Vol. 101, No. 47, 1997
Figure 3. Single (a,b) and double (c,d) bridge-modes for a water molecule on the side of a helical OEG strand. In (b,d) C (black), O in OEG (light grey), O in water (dark grey), H are small (light grey) circles. The black H in (a) is the left H on the water in (b). Otherwise as Figure 1.
in front of them. This results in two strong hydrogen bond structures in which either one H atom of water interacts with the two OEG O atoms, Figure 3a, b, or both water H atoms interact via hydrogen bonds with the OEG O atoms, Figure 3c,d. We call these adsorption modes the single and double bridgemode to distinguish them from the other H-mode, where the hydrogen-bonded H atom interacts with only one O atom of a OEG. The binding energies for the single and double bridgemodes are 323 and 352 meV, respectively, in Hartree-Fock. Adding MP2 corrections increases these binding energies to 440 and 490 meV, respectively, i.e. relatively more than in the water dimer, essentially because on the OEG more energy levels contribute to the perturbation. At the position of the water the field strength is as high as 0.9 V/Å. There cannot be such bridge-mode adsorption of water on planar OEG, because any two successive O atoms in the chain are located on opposite sides and, therefore, their electrostatic fields have no way to overlap and enhance each other. V. Water Adsorption on OEG-Terminated SAM To study water adsorption on both the helical and planar OEG films, we represent the latter by a cluster with a hexagonal lattice of OEG strands with a lattice constant of 5 Å between nearest neighbor molecules, independent of their molecular conformation. We have also done some calculations for smaller separations that confirm our main conclusions regarding the difference in the interaction between water and the two molecular conformations. For planar PEG we have taken two orientations of the molecular planes, both parallel to each other but one where the plane is oriented parallel to the crystal axis and one where it is rotated by 30°; the latter is the configuration of minimal energy in the bulk crystal.10 For helical PEG we minimize the energy for every cluster by allowing uniform
Wang et al.
Figure 4. Electrostatic potential and adsorption geometry for a cluster of three planar OEGs separated by 5 Å with two possible water adsorption geometries in a plane perpendicular to the surface. Solid lines are for -1.0, 0.5, -0.2, -0.1, and -0.05 eV and dashed lines for 1.0, 0.5, 0.2, 0.1, and 0.05 eV. Atoms depicted as full, shaded, and open circles are in, below, and above the plane, respectively. (a) OEG plane parallel to crystal axis and (b) rotated by 30°.
rotations. Any conformational change of the OEG strands caused by the interaction of water molecules are not considered, and the OEG molecular conformations remain as determined in the energy minimization step. Starting with an array of planar methoxy-terminated tri(ethylene glycol) OEG, the O-mode adsorption on the methoxy group is slightly stronger (76 meV and O-H distance of 2.86 Å in HF for the orientation of the OEG plane parallel to the crystal axis) than on a single strand (52 meV and O-H distance of 2.83 Å in HF). This is to be expected from the electrostatic field distribution shown in Figure 4a, which is quite similar above the film to that above a single strand except at larger distances where the electrostatic fields from neighboring OEGs overlap and enhance each other, leading to a stronger, albeit still quite weak, field. A similar situation can be expected for O-mode adsorption on top of a helical OEG film. Comparing parts a and b of Figure 4, we see that for the two orientations of the OEG plane with respect to the crystal axis the field distributions and the position and orientation of the adsorbed water molecule are quite similar. Indeed, their binding energies differ by less than 10%. H-mode adsorption of a water molecule on an array of planar OEG is much weaker than on a single strand, with an adsorption energy of only 11 meV and an O-H distance of 3.40 Å (compared to 185 meV and 2.04 Å for a single OEG). As Figure 4a shows, the H atoms of the methoxy groups of neighboring OEG strands build up an electrostatic potential
Oligo(ethylene glycol)-Terminated SAMs
Figure 5. Electrostatic potential (a) of a cluster of four helical OEGs separated by 5 Å acting on a water molecule, in a plane perpendicular to the surface, and (b) the adsorption geometry. The black O atom on the left OEG in part a is the same as the light gray O in the left OEG strand of part b. Solid lines are for -0.5, -0.2, -0.1, and -0.05 eV and dashed lines for 0.05, 0.1, 0.2, and 0.5 eV. Otherwise as Figures 1 and 2.
barrier to prevent the water molecule from moving down and adsorbing at the topmost O atom of the OEG. The situation is completely different for H-mode adsorption of water on a film of helical OEG. As we already know from Figure 3d, the orientation of the top chain section CH3-OCH2- of a helical OEG strand significantly tilts from the direction perpendicular to the substrate surface, so that the top O atoms of the helical OEG strands are exposed to water molecules. This is clearly seen in the electrostatic field distribution, Figure 5a, shown here in a plane perpendicular to the substrate surface that contains both the top O atom of a helical OEG strand and the O atom of the water molecule. Thus, water molecules are forced down between the OEG strands to the topmost O atom, where they are adsorbed. In contrast to H-mode adsorption of water on a single helical OEG strand, H-mode adsorbed water on (or rather in) a helical OEG film can form additional hydrogen bonds with the top H atoms on the neighboring OEG strands, resulting in a number of local adsorption minima with slightly different energies. In Figure 5b we show the complete cluster of one water molecule adsorbed on a helical OEG film (of four helical OEG strands separated by 5 Å). The planar cut in Figure 5a, perpendicular to the surface, contains the (gray) O atom of the left OEG strand and the (dark gray) O atom of the water molecule of Figure 5b. It shows the electrostatic potential generated by the OEG strands as it acts on the water molecule in the upper center. The water molecule forms a H-mode hydrogen bond with an H-O distance of 2.05 Å. The water O atom also interacts with an H atom of the methoxy group of a neighboring (helical) OEG strand and thus forms an additional O-mode hydrogen bond with an O-H distance of 2.59 Å. The total adsorption energy is 297 meV in HF and 380 meV in MP2, which is more than the sum of Hand O-mode adsorption on a single helical OEG strand. We have also investigated whether water molecules can adsorb further down the OEG strands and found that bridgemode adsorption involving two successive O atoms of a helical
J. Phys. Chem. B, Vol. 101, No. 47, 1997 9771
Figure 6. Electrostatic potential (a) and adsorption geometry (b) for a cluster of four helical OEGs separated by 5 Å plus six water molecules, three in a first and a second layer each, but showing only one in the second layer. In part b two water O in the first layer are light gray, and the two water O (the left one in the first layer and the central, somewhat higher one in the second layer) in the plane of part a and dark gray. In part a the field is that produced by the OEGs and the three water molecules in the first layer (i.e. in the absence of the “small” water molecule in the second layer).
OEG strand is not possible for a lateral distance of 5 Å between neighboring helical OEG strands. However, if the distance between the heads of two neighboring OEG strands widens beyond 5.5 Å, bridge-bonding becomes possible. It should thus play a major role for larger lattice constants of the SAM or at defects. Such bridge-bonded water molecules will, in addition, interact with neighboring OEG strands. Such cross-linking has been inferred previously from NMR measurements.15 The final question that arises is whether an OEG cluster can form a template for water adsorption. The answer is “no” for planar OEG because even a single water molecule does not adsorb readily on OEG except at the lowest temperatures. For helical OEG the situation is obviously more favorable. To get a definitive answer and also to understand the structure of the adsorbed water close to the SAM surface, we have done calculations for a cluster of 2 × 2 helical OEGs with a lattice distance of 5 Å exposed to six water molecules, with three forming the first layer and another three in a second layer. Allowing the positions and orientations of the water molecules to adjust themselves in an energy minimum of the fixed OEG lattice, we get the final adsorption geometry depicted in Figure 6b. (The top chain section -CH2-O-CH3 of each OEG unit was also allowed to optimize in a few trial runs but was found to remain rather rigid.) The water molecules in the first layer (the dark gray O atom on the left and the two light gray O atoms on the right and in the back) are hydrogen bonded in the H-mode to the uppermost O atom on the OEGs. Water molecules in the second layer (we show only one with the dark gray O atom in the upper center) adjust themselves to form hydrogen bonds to the first layer; Figure 6a, b. The cut in Figure 6a is in the plane perpendicular to the surface that contains the
9772 J. Phys. Chem. B, Vol. 101, No. 47, 1997
Wang et al.
Figure 7. 20 water molecules form three adsorption layers on a hexagonal cluster of a helical OEG film with lattice constant 5 Å consisting of 3 × 4 helical OEG strands.
(dark gray) O atom of the water molecule in the upper center of Figure 6b. The equipotential lines are those of the EOGs and all the water molecules except the topmost. Indeed, the (kidney-shaped) potential at the top is very similar to the one established on the side of the topmost O atom (still visible in the lower left corner of Figure 6a albeit more concentrated than in Figure 5a because the planes are not the same) on the helical OEG (see Figure 5a), and it is this local similarity that lies at the heart of the template effect. Because the water adlayers are amorphous, it is clear that a global field replication cannot occur in the same plane nor even in the same direction. What counts for the anchoring of an amorphous water film on the OEG surface is the similarity of the local field at the top of the OEG to that inside the water adlayers and ultimately to bulk water. To see the structure of the water adlayers on a larger scale, we have done calculations for a hexagonal array of 3 × 4 helical OEGs with a lattice distance of 5 Å exposed to 20 water molecules, with six forming the first layer and the rest in second and third layers, Figure 7. That this film of 20 water molecules approaches the structure of water is clearly displayed by the pair distribution function in Figure 8. It has the peaks for the first and second coordination shells at the positions of bulk water (dashed line). In a small cluster such as this there is not enough statistics for distances beyond the second coordination shell, so the data beyond 6 Å are not significant. What is striking in the histogram is a density of distances around 5 Å. These are induced by the OEG substrate and reflect the fact that in the first layer the H-mode adsorption on the topmost O atoms of the OEGs obviously must reflect the lattice structure of the SAM. VI. Discussion The most important result of this study is that the amphiphilic behavior of OEG-terminated surfaces depends on the molecular conformation of the OEG strands. The helical conformation adsorbs water strongly and serves as a template for water nucleation, whereas the all-trans conformation interacts only weakly with water molecules. This gives a plausible interpretation of the experimentally observed adsorption behavior toward proteins of OEG-terminated SAMs on Au and Ag1 and may be the crucial facet of protein resistance of PEO. We summarize the main results of our theoretical study based on Hartree-Fock calculations that support these conclusions. 1. On a single planar or helical OEG strand, a water molecule can form a weak hydrogen bond with an H atom of the terminal methoxy group of the OEGswe have called this O-mode adsorptionswith an adsorption energy of 52 meV, or it can form a stronger hydrogen bond with the uppermost O atom of the OEG strandswhich we call H-mode adsorptionswith an adsorption energy of 185 meV. 2. Because of the three-dimensional structure of a helical OEG strand, a water molecule can form a very strong bridgemode hydrogen bond on it with an adsorption energy as high
Figure 8. Pair distribution function of 20 water molecules adsorbed on a cluster of 3 × 4 helical OEGs, corresponding to Figure 7. The dashed line is for bulk water, and the shaded area is induced by the substrate structure.
as 323 and 352 meV. In this geometry, the H atoms of the water molecule interact with two successive O atoms of a helical OEG strand. Because of the zigzag structure of the planar OEG, there is no such bridge-mode hydrogen bond of water on it. 3. For a hexagonal lattice of planar OEG strands with a lattice constant of 5 Å, there is not enough room to allow water molecules to move down to form the (strong) H-mode hydrogen bond on the top O atoms of the OEG strands, while the weaker O-mode hydrogen bond with the terminal methoxy group H atoms of the OEG strands renders the formation of a water film very unlikely. 4. On the other hand, for a hexagonal lattice of helical OEG strands with a lattice constant of 5 Å, its open structure at its surface easily accommodates water molecules to its topmost O atoms to form H-mode hydrogen bonds. Moreover, the O atom of the water molecule interacts with the methoxy group H atoms of the neighboring helical OEG strands to form an O-mode hydrogen bond. This leads to an average adsorption energy of 297 meV, which is sufficient to provide a template for the growth of a water film. 5. With a larger lattice constant for a hexagonal lattice of helical OEG strands, water molecules can move further down between the OEG strands to form very strong bridge-mode hydrogen bonds with its two H atoms interacting with two successive OEG O atoms, or even stronger double H-mode hydrogen bonds with each water H atom interacting with one of the two OEG O atoms separated by two EO units. 6. The electrostatic interaction provides a reasonable explanation for water adsorption on OEG films. These findings are in general agreement with the discussion given by Israelachvili and Wennerstro¨m in their review of hydration and water structure in biological interactions.27 In particular, they discussed the forces arising from the water structure and concluded that surfaces can act as a template for the adsorption of water, particularly if electric fields induce order in the water interface by interaction with the water dipoles. Israelachvili and Wennerstro¨m concluded that the only real force barrier between two surfaces to interact is caused by the first layer of water molecules directly in contact with the surfaces. However, according to these authors, the net repulsive forces between two surfaces are not due to the water structure but due to the entropic repulsion arising from the confinement of thermally mobile surface groups. The experimental work in ref 1 showed that densely packed SAMs with partly crystalline helical OEG surface groups can be protein resistant, indicating that conformationally flexible polymer strands are not a necessary condition to render a surface resistant to protein adsorption. This supports a model where the binding of water can act as a template for strong solvation.
Oligo(ethylene glycol)-Terminated SAMs To put the template hypothesis in perspective to our calculations and the experimental data, we recall that for bulk water the heat of vaporization is 44 kJ/mol. With an average number of bonds per water molecule of 3.5 this gives about 125 meV binding energy per bond. On the other hand a water molecule approaching the surface of helical OEG can adsorb in the H-mode on the topmost O atom with a cross-linking to the methoxy group of the neighboring OEG with about 297 meV so that the first layer of water gets adsorbed quite readily. This layer has some orientational order offering further hydrogen bonds to a second layer, which is still dominated in its structure by the first layer and thus by the substrate. This and further layers are bound less strongly, with bond energies more typical of bulk water. Our calculations also reveal that in particular for the second layer there are many local energy minima corresponding to slightly different adsorption geometries into which the molecules can arrange themselves. This near degeneracy is of course necessary for the water to exhibit liquidlike behavior. So far we have only considered perfectly ordered clusters of OEG. If the last OE unit is tilted away from its equilibrium geometry by thermal motion, the topmost O atoms may be exposed more to the water, in which case the much stronger bridge-bond may form, involving two O atoms on the OEG. This can also happen at lattice defects, leading to an even stronger anchoring of the water layer. As to the uniqueness of the OEG and other organic films toward water adsorption, we contrast this system with a metal surface. On the latter water molecules experience the tails of the metal electron distribution, which on the scale of the water molecule are laterally more or less unstructured. The resulting field distribution in front of a metal surface is correspondingly quite homogeneous, with equipotential surfaces running more or less parallel to the metal surface. This is not the most advantageous field distribution for the formation of hydrogen bonds, for which dipolar fields are more conducive. Adsorbing water molecules must thus first create such dipolar field environments, which they can only do by first forming bilayers which then expose on their outside dipolar field distributions in which further layers can grow. This mechanism is, for instance, manifested in the fact that on most metals a monolayer of water cannot be adsorbed and that in desorption the remaining water (below two monolayers) forms islands of bilayers. To further illustrate the fact that the helical OEG film exposes the “right” template for water adsorption, we can compare the field distributions of (i) the clean OEG surface, Figure 5a, and (ii) after one layer of water is adsorbed, Figure 6a. The replication of dipolar fields for further water adsorption can be seen by noting that in Figure 5a the O atom on OEG produces the typical kidney-shaped dipolar field, in the center of which the water molecule binds. Very similar fields are then generated in Figure 6a by the water molecules in the first layer, albeit reoriented in space to allow for the formation of a liquid, and so on for further layers. This contrasts sharply with the situation on planar OEG, which does not generate any dipolar field character at all. One should, however, remember that the primary entities in bonding and bond formation are the electrons and that the electric field is secondary and merely calculated via the Poisson equation as a convenient tool for intuition. Keeping this in mind, we can say that water adsorption on helical OEG is (a) initiated by dipolar fields around the topmost O atoms of the last OE unit, which are very much like those around a single water molecule, and that (b) further growth is guaranteed by propagating very similar field distributions into the bulk of the water. These fields are, however, always short-
J. Phys. Chem. B, Vol. 101, No. 47, 1997 9773 ranged and their long-range effect comes about by replication. One should also keep in mind that the field distributions shown in Figures 5 and 6 are in particular planes and that such dipolar fields also exist in other directions and planes to ensure the amorphous character of liquid water. Our calculations so far have only considered the energetics of solvation. Any increase in entropy in the OEG layer induced by water adsorption or penetration into the film would further stabilize the OEG/water phase. We conclude the following: OEG in its helical conformation, but not in its “all-trans” form, is amphiphilic with respect to water. The stability of the water interface with helical OEG prevents proteins and other molecules from adsorbing irreversibly on the OEG surface. Acknowledgment. M.G. thanks the Deutsche Forschungsgemeinschaft and the DAAD for financial support of a sabbatical leave during which this work was initiated and completed. M.G. thanks Harvard University for its outstanding hospitality and appreciates the stimulating and helpful discussions with G. M. Whitesides. The work at Dalhousie University was supported by a grant from the Office of Naval Research. References and Notes (1) Harder, P.; M. Grunze, M.; Dahint, R.; Whitesides, G. M.; Laibinis, P. J. Phys. Chem., submitted. (2) de Gennes, P. G. Ann. Chim. 1987, 77, 389. (3) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; de Gennes, P. G. J. Colloid Interface Sci. 1991, 142, 149. (4) Jeon, S. I.; Andrade, J. D. J. Colloid Interface Sci. 1991, 142, 159. (5) Taunto, H. J.; Toprakcioglu, C.; Fetters, L. J.; Klein, J. Nature 1988, 332, 712. (6) PEGs are also sometimes referred to as poly(ethylene oxide) (PEO) and poly(oxyethylene) (POE). In this paper, the term poly(ethylene glycol) (PEG) will be used for polymers of all molecular weights. (7) Masatoki, S.; Ohno, K.; Yoshida, H.; Matsuura, H. J. Phys. Chem. 1996, 100, 8487. (8) Takahashi, Y.; Tadokoro, H. Macromolecules 1973, 6, 672. (9) Matsuura, H.; Miyazawa, T. J. Polym. Sci. A-2 1969, 7, 1735, 1744. (10) Takahashi, Y.; Sumita, I.; Tadokoro, H. J. Polym. Sci. 1973, 11, 2113. (11) Dubois, L. H.; Zegarski, B. R.; Nuzzo, R. G. J. Chem. Phys. 1993, 98, 678. (12) Fenter, P.; Eisenberger, P.; Li, J.; Camillone, N., III; Bernasek, S.; Scoles, G.; Ramanarayanan, T. A.; Liang, K. S. Langmuir 1991, 7, 2013. (13) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 10714. (14) Sueki, S.; Kawahara, N.; Nakata, M.; Kaneko, M. Polymer 1976, 17, 685. (15) Lu¨sse, S.; Arnold, K. Macromolecules 1996, 29, 4251. (16) Kjellander, R.; Florin-Robertsson, E. J. Chem. Soc., Faraday Trans. 1981, 1, 77. (17) Tasaki, K. J. Am. Chem. Soc. 1996, 118, 8459. (18) Karlstrm ¨ , G. J. Phys. Chem. 1985, 89, 4962. (19) Brodsky, A. Chem. Phys. Lett. 1996, 26, 563. (20) Chakravorty, S. J.; Davidson, E. R. J. Phys. Chem. 1993, 97, 6373. (21) Rovira, M. C.; Novoa, J. J.; Whangbo, M.; Williams, J. M. Chem. Phys. 1995, 200, 319. (22) Feyereisen, M. W.; Feller, D.; Dixon, D. A. J. Phys. Chem. 1996, 100, 2993. (23) Kollman, P. A. In Modern Theoretical Chemistry; Schaefer, H. F.; III, Ed.; Plenum Press: New York, 1977; Vol. 4. (24) Newton, M. D. J. Phys. Chem. 1983, 87, 4288. (25) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; HeadGordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, Revision B.2; Gaussian, Inc.: Pittsburgh, PA, 1995. (26) Frisch, M. J.; Del Bene, J. E.; Binkley, J. S.; Schaefer, H. F., III. J. Chem. Phys. 1986, 84, 2279. (27) Israelachvili, J.; Wennerstro¨m, H. Nature 1996, 379, 219. (28) Thiel, P. A.; Madey, T. E. Surf. Sci. Rep. 1987, 7, 211. (29) Held, A.; Menzel, D. Surf. Sci. 1995, 327, 301.