Adsorption of Organosilanes at a Zn-Terminated ZnO (0001) Surface

Institute of Physical Chemistry, UniVersity of Vienna, Währinger Strasse 42, A-1090 ... DaVy Faraday Research Laboratory, The Royal Institution of Gr...
0 downloads 0 Views 1MB Size
Download PDF
8036

Langmuir 2006, 22, 8036-8042

Adsorption of Organosilanes at a Zn-Terminated ZnO (0001) Surface: Molecular Dynamics Study Andreas Kornherr* and Gerhard E. Nauer Institute of Physical Chemistry, UniVersity of Vienna, Wa¨hringer Strasse 42, A-1090 Wien, Austria, and Center of Competence in Applied Electrochemistry GmbH, Wiener Neustadt, Austria

Alexey A. Sokol, Samuel A. French, and C. Richard A. Catlow DaVy Faraday Research Laboratory, The Royal Institution of Great Britain, London, United Kingdom

Gerhard Zifferer Institute of Physical Chemistry, UniVersity of Vienna, Wa¨hringer Strasse 42, A-1090 Wien, Austria ReceiVed February 15, 2006. In Final Form: July 3, 2006 Four different organosilanes (octyltrihydroxysilane, butyltrihydroxysilane, aminopropyltrihydroxysilane, and thiolpropyltrihydroxysilane) adsorbed at a reconstructed Zn-terminated polar ZnO (0001) surface are studied via constant temperature (298 K) molecular dynamics simulations. Both single adsorbed silane molecules as well as adsorbed silane layers are modeled, and the energy, distance, orientation, and alignment of these adsorbates are analyzed. The adsorbed silane molecules exhibit behavior depending on the chemical nature of their tail (nonpolar or polar) as well as on the silane concentration at the solid surface (single adsorption or silane layer). In contrast to the O-terminated ZnO surface studied previously, now adsorption can only occur at the vacancies of this reconstructed crystal surface, thus leading to an arched structure of the liquid phase near the crystal surface. Nevertheless, both nonpolar and polar single adsorbed silanes show a similar orientation and alignment at the surface (orthogonal in the former, parallel in the latter case) as for the O-terminated ZnO surface, although the interaction energy with the surface is considerably increased for nonpolar silanes while it is nearly unaffected for the polar ones. For adsorbed silanes within silane layers, the difference to single adsorbed silanes depends on the polarity of the tail: nonpolar silanes again show an orthogonal alignment, while polar silanes exhibit two different orientations at the solid surfacesa head and a tail down configuration. This leads to two completely different but nevertheless stable orientations of these silanes at the Zn-terminated ZnO surface.

Introduction Thin polysiloxane layers on metal substrates are of both high scientific and technological interest1-5 due to the unique properties of this class of polymers.6-8 The strength of the inorganic SiO-Si backbone on one hand and the flexibility of the organic tail on the other gives the user a valuable tool to meet different requirements of the industry with only one kind of polymer (accordingly, they are often called organic/inorganic hybrids). Furthermore, by introducing additional functionalized organic groups, they can also be used as an adhesion promoter9 providing a link between metallic surfaces and other organic coatings (resins or polymers), or its appearance can be changed from transparent to an ad libitum colored coating. Therefore, various technical applications are based on sol-gel10 processes tailoring the surface characteristics of the desired material by covering it with a * To whom correspondence [email protected].

should

be

addressed.

E-mail:

(1) Plueddemann, E. P. Silane Coupling Agents, Plenum Press: New York, 1991. (2) Mittal, K. L. Silanes and other coupling agents, VSP: Utrecht 1992. (3) Izumi, K.; Tanaka, H.; Uchida, Y.; Mater, J. Sci. Lett. 1993, 12, 724. (4) Beccaria, A. M.; Chiaruttini, L. Corros. Sci. 1999, 41, 885. (5) Hansal, S.; Hansal, W. E. G.; Po¨lzler, M.; Wittmann, M.; Nauer, G. E. Proc. Electrochem. Soc. 2000, 99, 167. (6) Noll, W. The Chemistry and Technology of Silicone, Academic Press: New York 1968. (7) Sjo¨blom, J.; Stakkestad, G.; Ebeltoft, H. Langmuir 1995, 11, 2652. (8) Osterholz, F. D.; Pohl, E. R. J. Adhesion Sci. Technol. 1992, 6, 127. (9) Walker, P. J. Adhesion Sci. Technol. 1991, 5, 279. (10) MacKenzie, J. D. J. Sol-Gel Sci. and Technol. 2003, 26, 23.

polysiloxane layer (e.g., corrosion protection,11 wear resistance, sliding properties, and water repellence, to name a few). In this regard, a better understanding of the complex physical and chemical interactions at the silane-metal/metal oxide interfaces determining the structure and morphology and thus also the physical properties of these thin polymer layerssis fundamental. However, the formation of such layers at a metal substrate is quite complex, including various different steps like (a) the formation of a sol (i.e., reactive silanes dissolved in a solvent, e.g., 2-propanol) by the acid or base catalyzed hydrolyses of (organo)alkoxysilanes into reactive (organo)silanes (actually organotrihydroxysilanes); (b) the subsequent adsorption, orientation, and alignment of the produced reactive silane molecules at the solid surface; (c) the accumulation of these adsorbed silanes at the surface eventually leading to a silane layer covering the whole surface; and (d) the polycondensation process (normally caused by postheating of the silane covered metal plate up to more than 100 °C, e.g., 120-160 °C) to form a three-dimensional cross-linked gel out of the adsorbed silane layer. As not all of these steps can be observed or probed by experimental methods, there is a great demand for alternative methods delivering a detailed picture on an atomistic scale. Accordingly, computational methods,12-16 although, of course, (11) Hansal, W. E. G.; Hansal, S.; Po¨lzler, M.; Kornherr, A.; Zifferer, G.; Nauer, G. E. Surf. Coat. Technol. 2006, 200, 3056. (12) Binder, K.; Milchev, A.; Baschnagel, J. Annu. ReV. Mater. Sci. 1996, 26, 107. (13) Xing, L.; Mattice, W. L. Langmuir 1996, 12, 3024.

10.1021/la0604432 CCC: $33.50 © 2006 American Chemical Society Published on Web 08/11/2006

Adsorption of Organosilanes at ZnO (0001) Surface

Langmuir, Vol. 22, No. 19, 2006 8037

always limited by some simplifications and limitations inherent to the chosen method, can vitally contribute to a better understanding of these various processes. Because of the large number of atoms required to build up a realistic, representative in silico model of a liquid-solid interface (also comprising adsorbed silane layers), resorting to classical molecular dynamics (MD) simulations, based on interatomic potentials, is necessary in our study. As in previous papers,17-19 we focus on the second and third step again using four different silane molecules (octyltrihydroxysilane, butyltrihydroxysilane, aminopropyltrihydroxysilane, and thiolpropyltrihydroxysilane). These silanes cover a wide range of different organic tail groups with long (octyl) and short (butyl) as well as polar (amino and thiol) and nonpolar (octyl and butyl) tails. Furthermore, all of them are also used for technical applications such as the formation of a protective layer on metal substrates to achieve a good corrosion protection of the underlying surface. However, contrary to our work published so far, now the adsorption of these silane molecules at the Zn-terminated side of a zinc oxide (ZnO) crystal20 (i.e., the polar (0001) surface) is studied. Obviously, for a complete picture of the adsorption process at polar ZnO crystal surfaces, both cases have to be taken into account as both are present and will, therefore, affect the formation of silane layers. Accordingly, it is the aim of this study to compare the results obtained at the Zn-terminated side with those at the O-terminated (0001h) side17-19 to see whether our previously gained insight into the adsorption process at the latter surface can be transferred to the Zn-terminated side or not. Numerical Methods The morphology of the Zn-terminated surface of ZnO remains a subject of debate in the literature. In particular, the macroscopic dipole associated with an ideal bulk termination by polar surfaces must be quenched, and most recent investigations21,22 show that the (0001) polar surfaces of ZnO are reconstructed with about a quarter of the Zn ions missing from the upper surface layers. Moreover, the Zn vacant sites coexist with larger morphological features such as triangular islands and indentations, which can be largely removed by annealing.23,24 Accordingly, we concentrate here on the case of a regularly reconstructed surface of ZnO while other intermediate cases (e.g., stochastic) could be inferred by being compared with our other studies using unreconstructed surfaces. All molecular dynamics (MD) simulations are run at ambient temperature (298 K) following a procedure already outlined in ref 18 (which itself is a modified version of ref 17). A simulation box (≈3.3 nm × 3.4 nm × 6.8 nm) with periodic boundary conditions is used (see Figure 1) to model a representative part of the interface devoid of any arbitrary boundary effects. The box consists of a ZnO slab with fixed spatial positions and a liquid phase (representing the sol) consisting of two or three different layers of molecules. The (14) Zhang, Z.; Beck, T. L. Langmuir 1996, 12, 1227. (15) Claire, P. S.; Hass, K. C.; Schneider, W. F.; Hase, W. L. J. Chem. Phys. 1997, 106, 7331. (16) French, S. A.; Sokol, A. A.; Catlow, C. R. A.; Kornherr, A.; Zifferer, G. Chem. Commun. 2004, 1, 20. (17) Kornherr, A.; Hansal, S.; Hansal, W. E. G.; Besenhard, J. O.; Kronberger, H.; Nauer, G. E.; Zifferer, G. J. Chem. Phys. 2003, 119, 9719. (18) Kornherr, A.; French, S. A.; Sokol, A. A.; Catlow, C. R. A.; Hansal, S.; Hansal, W. E. G.; Besenhard, J. O.; Kronberger, H.; Nauer, G. E.; Zifferer, G. Chem. Phys. Lett. 2004, 393, 107. (19) Kornherr, A.; Hansal, S.; Hansal, W. E. G.; Nauer, G. E.; Zifferer, G.; Macromol. Symp. 2004, 217, 295. (20) Zhang, X. G. Corrosion and Electrochemistry of Zinc, Plenum Press: New York 1996. (21) Jedrecy, N.; Sauvage-Simkin, M.; Pinchaux, R. Appl. Surf. Sci. 2000, 162, 69. (22) Kunat, M.; Girol, S. G.; Becker, T.; Burghaus, U.; Wo¨ll, C. Phys. ReV. B 2002, 66, 081402. (23) Dulub, O.; Boatner, L. A.; Diebold, U. Surf. Sci. 2002, 519, 201. (24) Dulub, O.; Diebold, U.; Kresse, G. Phys. ReV. Lett. 2003, 90, 016102-1.

Figure 1. Simulation box (3.3 × 3.4 × 6.8 nm) with periodic boundaries showing the same configuration as in Figure 2a (the octyltrihydroxysilane molecule is marked black). The different grades of gray in this and the following figures correspond to different atom types with white reserved for hydrogen, black oxygen, light gray silicon, gray carbon or zinc, and dark-grey nitrogen or sulfur (depending on the silane molecule). All the graphical displays of molecules were generated with the Materials Visualizer (from Accelrys Inc.). ZnO crystal is cleaved along the (0001) plane with every fourth Zn atom on top and every fourth O atom at the bottom being removed25-28 to model a reconstructed surface without any macroscopic dipoles. The whole reconstructed ZnO crystal is comprised of 810 Zn and 810 O atoms. The first liquid layer (adjacent to the crystal surface) either (a) contains one organosilane and 250 2-propanol moleculess corresponding to the start of the adsorption process with only one silane molecule being adsorbed at the solid surface or (b) is itself again composed of two different layers with the bottom layer containing 50 organosilanes and the top layer 100 2-propanol moleculesscorresponding to the end of the adsorption process where the crystal surface is completely covered by silane molecules. (25) Nyberg, M.; Nygren, M. A.; Petersson, L. G. M.; Gay, D. H.; Rohl, A. L. J. Phys. Chem. 1996, 100, 9054. (26) French, S. A.; Sokol, A. A.; Bromley, S. T.; Catlow, C. R. A.; Rogers, S. C.; King, F.; Sherwood, P. Angew. Chem., Int. Ed. 2001, 113, 4569. (27) Meyer, B. Phys. ReV. B 2004, 69, 045416. (28) Sokol, A. A.; Bromley, S. T.; French, S. A.; Catlow, C. R. A.; Sherwood. P. Int. J. Quantum Chem. 2004, 99, 695.

8038 Langmuir, Vol. 22, No. 19, 2006

Kornherr et al.

Figure 2. Configuration obtained after 1 ns for a single adsorbed octyltrihydroxysilane (a, left) and butyltrihydroxysilane (b, right) molecule (2-propanol molecules in this and the following figures are visualized by lines, only). In each case, an additional second (or third) liquid layer composed of 150 2-propanol molecules with fixed spatial positions serves as an upper limit for the molecules in the middle layer(s). By using a vacuum slab of 0.5 nm thickness, we minimize the associated error due to periodic boundary conditions ensuring that all molecules of the middle layer(s) are only influenced by the upper side of the crystal and not by its bottom (the O-terminated side). The layer densities F of pure 2-propanol as well as of 2-propanol with one silane molecule correspond to pure liquid 2-propanol at 298 K (F ) 0.7855 g/cm3), whereas the densities of the silane layers (i.e., 50 silane molecules)swhere no experimental data are availablesare derived by NPT (constant pressure and temperature) MD simulations: larger simulation boxes consisting of 100 silane molecules are equilibrated for 500 ps, and the mean densities over the last 200 ps serve as the input values for the construction of these four different silane layers. The interaction energy EZnO-silane of a silane molecule with the zinc oxide surface is calculated as EZnO-silane ) EZnO+silane - EZnO - Esilane with EZnO+silane being the total potential energy of the zinc oxide crystal together with the adsorbed silane molecule(s) and with EZnO and Esilane being the total potential energy of the zinc oxide crystal and the silane molecule(s), respectively. In case of more than one adsorbed silane, the right-hand side of eq 1 has to be divided by the number of these adsorbatessso EZnO-silane always corresponds to the adsorption energy of one adsorbed silane. Cutoff based electrostatic Coulomb interactions as well as van der Waals energies (employing a Lennard-Jones 9-6 function) are used as in ref 18 with a cutoff radius of 2.0 nm. Again, the applicability of the group based cutoff was checked by comparison with preliminary calculations using Ewald summation yielding discrepancies smaller than 10%. Temperature and pressure control is performed by using the Andersen thermostat and barostat29 with a time step of 1 fs by applying the Verlet velocity30 algorithm. Every 1 ps, a snapshot of the whole system is taken, thus enabling a detailed analysis of the structure, shape, and energy of the adsorbed silane molecules as a function of time over the whole trajectory. Actually, mean values are extracted from these trajectories by averaging over the last 500 ps; standard deviations of mean values obtained by use of the block averaging method31 are ca. 1.5% in the case of one silane molecule and are smaller than 0.5% for silane layer systems. For the whole simulation procedure, the software package Materials Studio 3.0 (from Accelrys Inc.) is applied using the Discover molecular dynamics program. Interactions between atoms and (29) Andersen, H. C. J. Chem. Phys. 1980, 72, 2384. (30) (30) Verlet, L. Phys. ReV. 1967, 159, 98. (31) G. Zifferer, Macromolecules 1990, 23, 3166 and references therein.

molecules are accounted for by use of the COMPASS force field,32-36 which is optimized for the simulation of condensed phases including metal oxides (for further details, see ref 17).

Results and Discussion The outline of this section is as follows. First, we will discuss the adsorption of single silane molecules, corresponding to the start of the layer formation where only one silane (apart from other solvent molecules) has been able to dock at the solid substrate. Second, we continue by studying the properties of silane layers already covering the whole metal oxide surface and compare the properties of adsorbed silanes within these layers to single adsorbed silane molecules. Single Adsorbed Silanes. The two nonpolar organosilanes (i.e., octyltrihydroxy- and butyltrihydroxysilane) at the Znterminated polar ZnO surface show a preferred orthogonal orientation (see Figure 2a,b) with the polar head (Si(OH)3) being adsorbed at the surface and the alkyl tail staying in the solvent phase. However, although the orientation of these silanes is similar to that at the O-terminated polar ZnO surface,17,18 now adsorption only occurs at the vacancies of this solid surface (i.e., where a Zn atom is missing and the adsorbate can, therefore, interact with the below lying negatively charged O atom). The same holds true for the solvent (2-propanol), which also shows adsorption only at the surface vacancies (see Figure 2 or Figure 1). Obviously, the zinc atoms of the Zn-terminated (0001) ZnO surface are repulsive for the molecules examined in our study (which all contain one or three hydroxy group(s)), with only the vacancies offering a possibility for the adsorbates to dock on it. This leads to a typical arched pattern of the liquid phase near to the crystal surface that can best be seen in Figure 1: as no adsorption occurs above the Zn atoms (which occupy about 3/4 of the whole surface area), an empty space is formed there, whereas above the vacancies (i.e., the O atoms), the molecules are attracted by the surface and thus seem to bend down to its thus an arch-like structure is shaped. Admittedly, this regularity is most pronounced for the specific pattern of vacancies chosen in our computational model, and arched structures will become less regular in appearance (with smaller and larger cavity spaces present) in real material. Nevertheless, vacant sites however they (32) Hwang, M. J.; Stockfisch, T. P.; Hagler, A. T. J. Am. Chem. Soc. 1994, 116, 2515. (33) Sun, H. Macromolecules 1995, 28, 701. (34) Sun, H. J. Phys. Chem. B 1998, 102, 7338. (35) Sun, H.; Ren, P.; Fried, J. R. Comput. Theor. Polymer Sci. 1998, 8, 229. (36) Bunte, S. W.; Sun, H. J. Phys. Chem. B 2000, 104, 2477.

Adsorption of Organosilanes at ZnO (0001) Surface

Figure 3. Analysis over a period of 1 ns of the interaction energy EZnO-silane (a, e), the mean distance of the three hydrogen atoms of the Si(OH)3 group from the surface d (b, f), the tilt angle R of the head (c, g) as well as the tilt angle β of the whole molecule with the surface (d, h) for a single adsorbed octyltrihydroxysilane (left) and butyltrihydroxysilane (right) molecule.

are arranged will play the role of anchor sites for the silane layer on the Zn-terminated surface of ZnO. Thus, the adsorption process is substantially different from the O-terminated surface, although, as already mentioned previously, the orientation and alignment of the two nonpolar silanes seems to be quite similar.

Langmuir, Vol. 22, No. 19, 2006 8039

A more detailed analysis of the orientation and also energy of these two adsorbed silanes is presented in Figure 3, where the corresponding trajectories are analyzed (left side for octyltrihydroxysilane, right side for butyltrihydroxysilane; configurations after 1 ns are shown in Figure 2a,b). The mean values of the interaction energy EZnO-silane read -97 kJ mol-1 (-76 kJ mol-1) and -86 kJ mol-1 (-74 kJ mol-1) for octyltrihydroxy- and butyltrihydroxysilane, respectively, with values in parentheses referring to the O-terminated ZnO surface.18 Surprisingly, although the O-terminated surface offers much more adsorption sites than the Zn-terminated one, the absolute values of EZnO-silane are increased in the latter case. Furthermore, also the difference in EZnO-silane between the two adsorbed silanes is now more pronouncedsoctyltrihydroxysilane with its longer nonpolar octyl tail exhibits a nearly 13% higher absolute value of the interaction energy. In contradiction to the energy, the orientation of the silanes at the solid surface is well-known from former studiess both from the reconstructed as well as from the ideal O-terminated surface: again, an orthogonal orientation is observed after a small period of timesnormally less than 100 ps (see also Figure 3). However, in contrast to the O-terminated surfaces, now the orthogonal alignment of single adsorbed nonpolar silanes is the only stable one, and parallel orientations are not observable, as the positively charged Zn atoms are both repulsive for the polar head and for the alkyl tail. If a start configuration with a parallel orientated nonpolar silane at the Zn-terminated surface is artificially generated, the molecule will align itself in an orthogonal position again. This orthogonal alignment is also reflected in the mean values of the orientation of the head (Si-C axis) as well as the overall orientation of the molecule (axis defined by the backbone) relative to the surface by use of tilt angles R and β, respectively (for details, see ref 17: R reads 63.3 and 70.3° for octyltrihydroxy- and butyltrihydroxysilane, respectively, while averaging over β makes no sense due to the large statistical fluctuations in Figure 3d,h). These fluctuations are due to the waging of the silane tail (only the polar Si(OH3) head is fixed at the surface) in the liquid solvent phasesthis has also been observed for adsorbed silanes at the O-terminated ZnO surface. Furthermore, also the mean distance d of the three hydrogens of the silane head from the surface (i.e., the top Zn layer) shows similar values such as in ref 18 s0.18 nm for both adsorbed silanes. In contrast to nonpolar silanes, the orientation and alignment of polar silanes (aminopropyltrihydroxy- and thiolpropyltrihydroxysilane) at the crystal surface is quite different, as can be seen in Figure 4. Like at the O-terminated surface, a quasi-

Figure 4. Configuration obtained after 1 ns for a single adsorbed aminopropyltrihydroxysilane (a, left) and thiolpropyltrihydroxysilane (b, right) molecule.

8040 Langmuir, Vol. 22, No. 19, 2006

Figure 5. Analysis over a period of 1 ns of the interaction energy EZnO-silane (a, e), the mean distance of the three hydrogen atoms of the Si(OH)3 group from the surface d (b, f), the tilt angle R of the head (c, g) as well as the tilt angle β of the whole molecule with the surface (d, h) for a single adsorbed aminopropyltrihydroxysilane (left) and thiolpropyltrihydroxysilane (right) molecule.

parallel alignment of these adsorbed silanes is observedsapart from the polar head also the polar tail (the amino or thiol group) bends down to the solid surface. A closer inspection of Figure 5 reveals that this closer contact with the ZnO crystal is also reflected in an increased interaction energy as compared to nonpolar silanes (again, values in brackets refer to the O-

Kornherr et al.

terminated ZnO surface18): EZnO-silane reads -105 kJ mol-1 (-104 kJ mol-1) and -120 kJ mol-1 (-118 kJ mol-1) for aminopropyltrihydroxy- and thiolpropyltrihydroxysilane, respectively. In this context, it is remarkable thatscontrary to nonpolar silaness now there is nearly no difference in EZnO-silane between the two ZnO surfaces. However, although EZnO-silane seems to be unaffected by the choice of the polar ZnO surface, there is a considerable difference in the distance d of the adsorbed silanes from the solid surface: while at the O-terminated surface there is no need for the silanes to seek for a vacancy (quite the contrary, in this case, vacancies are bad adsorption sites), now the hydroxy groups of the silane heads move to and (to some extent) into the vacancies. This behaviorsone could also interpret it as a substitution of a Zn atom by the H atom of a OH-groupsleads to d ≈ 0.21 and 0.19 nm for aminopropyltrihydroxy- and thiolpropyltrihydroxysilane, respectively. These values are markedly lower as compared to 0.25-0.27 nm observed at the O-terminated surface, the difference roughly corresponding to the distance between the top Zn- and adjacent O-layer (≈0.06 nm). Therefore, the similarity between the interaction energies at the two different ZnO surfaces can be traced back to the much closer distance of the adsorbed polar silanes at the Zn-terminated surface. The alignment of these silanes at the surface is highly ordereds especially the orientation of the whole molecules reflected by the angle β: values of 5.7 and 11.5° for aminopropyltrihydroxyand thiolpropyltrihydroxysilane, respectively, show the strict quasi-parallel orientation at the surface once a polar silane is adsorbed. The angle R, on the contrary, is not close to 0° (the hypothetical value for a perfect parallel orientation, both for R and β) as the Si(OH)3 head of a (either nonpolar or polar) silane adsorbed at a vacancy is always somehow tilted resulting in mean values of 35.6 and 50.1°, respectively. Therefore, the parallel orientation of the overall molecule is only due to the attraction of the polar tail by the polar crystal surface, thus forcing the molecule tail to bend down to the surface. Silane Layers. The adsorption of whole layers always differs from the adsorption of single molecules inasmuch as the number of possible adsorption sites at a solid surface is limited. Therefore, the number of adsorbed silane molecules within a silane layer being in contact with the Zn-terminated surface is restricted. Interestingly, this number of stable adsorbed silanes over the whole simulation period is nearly independent of the type of silane reading 29 and 28 for octyltrihydroxy- and butyltrihydroxysilane, respectively, as well as 30 for both types of polar silanes. Accordingly, the surface density per square nanometer

Figure 6. Configuration obtained after 3 ns for an adsorbed octyltrihydroxysilane (a, left) and butyltrihydroxysilane (b, right) layer.

Adsorption of Organosilanes at ZnO (0001) Surface

Figure 7. Analysis over a period of 1 ns of the mean interaction energy EZnO-silane (a, e), the mean distance of the three hydrogen atoms of the Si(OH)3 group from the surface d (b, f), the mean tilt angle R of the head (c, g) as well as the mean tilt angle β of the whole molecules with the surface (d, h) averaged over 29 adsorbed octyltrihydroxysilane (left) and 28 adsorbed butyltrihydroxysilane (right) molecules.

of all silanes studied is about 2.5-2.7 silanes/nm2 (with the surface area of the simulation box being ≈11 nm2). Furthermore, due to the limited space at the surface, the orientation and

Langmuir, Vol. 22, No. 19, 2006 8041

alignment of the adsorbed silanes within the layer is different as compared with the behavior of single adsorbed silanes. A visualization of these adsorbed silane layers at the ZnO surface is given in Figure 6 for nonpolar as well as in Figure 8 for polar silanes, respectively (each after a simulation time of 3 ns to ensure that the whole system had enough time to reach an equilibrium configuration). The corresponding trajectories are presented in Figures 7 and 9 with the interaction energy EZnO-silane, distance d and tilt angles R and βseach averaged over the respective number of adsorbed silanes of the last 1 ns of an overall 3 ns trajectory. Because of this averaging process, the fluctuations are much smaller than in Figures 3 and 5. Accordingly, the largest differences compared to single adsorbed nonpolar silanes are the smaller fluctuations of angle β in Figure 7d,h, indicating that the orientation of an adsorbed nonpolar silane molecule is quite fixed due to the other neighboring silanes. However, there is quite a difference between the two nonpolar silanes with the mean value of β reading 53.9 and 29.2° for octyltrihydroxy- and butyltrihydroxysilane, respectively. This difference can be traced back to the different alignment of the two silanes at the surface: octyltrihydroxysilane molecules with their long nonpolar octyl tails tend toward an orthogonal orientation to allow for a closer packing at the solid surface, whereas molecules with a smaller tail like butyltrihydroxysilane are not so strictly restricted. Therefore, the latter molecule is able to show a more tilted orientation of the whole molecule within an adsorbed layer, although the orientation of the head, measured via R, is nearly the same for both nonpolar silanes (ranging between 50 and 60°). This less orthogonal orientation of butyltrihydroxysilane leads to a higher mean interaction energy with EZnO-silane reading -105 kJ mol-1 as compared to -92 kJ mol-1 for octyltrihydroxysilanescontrary to single adsorbed silanes where the order of the strength (i.e., the absolute value of EZnO-silane) is reversed. The closer contact of the smaller silane molecule with the surface is also reflected in the mean distance d, which is 0.02 nm less than for octyltrihydroxysilane (0.21 nm). For polar silanes, the largest difference between single adsorbed silanes and those adsorbed within a layer is that due to the limited number of free adsorption sites as now only the head or the tail can get in close contact with the solid surfacesthe rest of the molecule has to stay in the bulk phase (see Figure 8). For aminopropyltrihydroxysilane with its quite polar amino group, this results at an approximate ratio of 1:1 for head-to-tail adsorption, while for thiolpropyltrihydroxysilane (exhibiting a

Figure 8. Configuration obtained after 3 ns for an adsorbed aminopropyltrihydroxysilane (a, left) and thiolpropyltrihydroxysilane (b, right) layer.

8042 Langmuir, Vol. 22, No. 19, 2006

Kornherr et al.

in a decrease of |EZnO-silane|. In addition, also the mean values for parameters d, R, and β are considerably different for the two types of adsorbates (H and T). The mean distance d is increased for T adsorbates (independent of the type of silane) by approximately 0.25-0.35 nm as compared to the value of single adsorbed silanes as the head of these molecules is farther away from the surfacestherefore, the mean value of the graphs in Figure 9b,f is considerably increased up to 0.31-0.34 nm. Furthermore, for H-adsorbates, R reads approximately 50°, while for T adsorbates, even negative values up to -15° are observed; β reads approximately 28° for H and 42° for T adsorbates. This leads to mean values of 21.6 and 26.9° for R (Figure 9c,g) and 36.8 and 32.2° for β (Figure 9d,h) for aminopropyltrihydroxyand thiolpropyltrihydroxysilane, respectivelysvalues that are quite different if compared to those for single adsorbed silanes.

Conclusion

Figure 9. Analysis over a period of 1 ns of the mean interaction energy EZnO-silane (a, e), the mean distance of the three hydrogen atoms of the Si(OH)3 group from the surface d (b, f), the mean tilt angle R of the head (c, g) as well as the mean tilt angle β of the whole molecules with the surface (d, h) averaged over 30 adsorbed aminopropyltrihydroxysilane (left) and 30 adsorbed thiolpropyltrihydroxysilane (right) molecules.

less polar tail end), this ratio is changed to 2:1. Accordingly, the graphs given in Figure 9 are actually mean values averaged over two different types of adsorbed silanessthose with the head or tail down. Surprisingly, the mean interaction energy EZnO-silane remains nearly unchanged for aminopropyltrihydroxysilane (-102 kJ mol-1), while for thiolpropyltrihydroxysilane, the absolute value of EZnO-silane is decreased (-96 kJ mol-1) as compared to single adsorbed silanes. A detailed investigation of the two types of adsorbed silanes reveals that for the former silane, both types are rather similar with respect to EZnO-silane, whereas for the latter one, the tail down configuration (T) falls below the head down configuration (H) by 30-70 kJ mol-1 (with respect to the absolute value of EZnO-silane), thus resulting

The adsorption process at the Zn-terminated surface is considerably different in comparison to the O-terminated surface: 17-19 now adsorption can only occur at the vacancies of the reconstructed surface, whereas in the latter case, the whole ZnO surfacesexcept for the vacanciessis a good adsorption site. Nevertheless, for single adsorbed silanes, both nonpolar and polar silanes show a similar orientation and alignment at the surface (however, with an increased interaction energy with the surface for nonpolar silanes) as for the O-terminated one: orthogonal in the former, parallel in the latter case. The only difference is that now the orthogonal alignment of single adsorbed nonpolar silanes is the only stable onesparallel orientations are not longer observed. For silanes adsorbed within layers, however, the difference to single adsorbed silanes depends on the polarity of the tail: nonpolar silanes again show an orthogonal alignment, while polar silanes exhibit two completely different adsorbed configurations at the solid surface. These head and tail down orientations can only be found within these layers. Therefore, the conclusion can be drawn that results obtained for single adsorbed silanes can only partially be transferred to silanes adsorbed within a silane layersespecially for polar silanes. As a consequence, it is essential for the in silico modeling of monomer layers in general not only to model single adsorbed monomers and transfer these results to layers but to build a computer model of the whole layer itselfsotherwise, the computational results might be misleading. However, in our case, we hope that the new insight into the adsorption behavior of different silane monomers at both polar ZnO surfaces will contribute to a better understanding of the various physical processes that precede and govern the formation of thin polysiloxane layers. Certainly in the future this will also help engineers to design tailor-made polysiloxane layers at metal (oxide) substrates for various technical applications. Acknowledgment. Financial support by TIG (Technologie Impulse GmbH), the government of Lower Austria within the Kplus program and the Royal Society is gratefully acknowledged. LA0604432