Conformational Order in Oligo (ethylene glycol)-Terminated Self

Oct 3, 2003 - Mathias Zwahlen,†,§ Sascha Herrwerth,‡ Wolfgang Eck,‡ Michael Grunze,‡ ... School of Chemistry, University of St. Andrews, St. ...
0 downloads 0 Views 97KB Size
Langmuir 2003, 19, 9305-9310

9305

Conformational Order in Oligo(ethylene glycol)-Terminated Self-Assembled Monolayers on Gold Determined by Soft X-ray Absorption Mathias Zwahlen,†,§ Sascha Herrwerth,‡ Wolfgang Eck,‡ Michael Grunze,‡ and Georg Ha¨hner*,† School of Chemistry, University of St. Andrews, St. Andrews, Fife KY16 9ST, UK, and Applied Physical Chemistry, University of Heidelberg, INF 253, 69120 Heidelberg, Germany Received June 16, 2003. In Final Form: July 28, 2003 Near-edge X-ray absorption fine structure (NEXAFS) measurements were performed on self-assembled monolayers (SAMs) of oligo(ethylene glycol) (OEG)-terminated alkanethiols adsorbed on gold. Films with a varying number of EG units were investigated. By comparing the experimental results with predictions for an “ideal film structure”, the degree of conformational orientation can be determined. Our results are consistent with earlier findings that the ethylene glycol units adopt a mixture of helical and amorphous conformations and allow it to quantify the maximum possible fraction of shorter oligomers displaying the helical conformation. The experiments were performed under high-vacuum conditions, thereby eliminating the effects of water on the conformation of the EG units. A significant amount of molecules in the films displays amorphous conformations.

University of St Andrews. University of Heidelberg. § Also at Department of Materials, ETH Zurich, Switzerland. * To whom correspondence should be addressed: phone +44 1334 463889, fax +44 1334 463808, e-mail [email protected].

of the physicochemical properties and the molecular structure of functional films in a vacuum can contribute to insights into fundamental mechanisms. The first publication on films of OEG-functionalized molecules focused on the composition and conformation of these novel SAMs.6 The authors measured the thickness of the monolayers using X-ray absorption spectroscopy and ellipsometry and found it to be significantly lower than the value expected for molecules in an all-trans conformation. It was concluded that the surfactants form layers that are much less ordered than comparable SAMs of methyl-terminated alkanethiols. In a later study OEGfunctionalized thiols on gold and silver were investigated with Fourier transform infrared reflection-absorption spectroscopy (FTIRAS).2 The films on silver display structural differences compared to the films on gold, which showed additional IR bands. These bands match resonances of crystalline poly(ethylene glycol),7 which are not found in amorphous PEG. The interpretation is that the EG units preferentially establish a 7/2 helix on gold, i.e., a helix with a unit cell of seven ethylene glycol groups, which perform two rotations,8 similar to the crystalline polymer. For EG films on silver the spectra are compatible with an upright, linear molecule with all C-C-O bonds in an all-trans conformation. Here, a helix formed by the EG units is thought to be unfavorable because of its spatial requirement, which is almost identical to that of an alkanethiolate on Au(111),9 i.e., approximately 15% higher than the area per sulfur bonding site on an ideal Ag(111) surface. Tetra(ethylene glycol)-hydroxy-terminated alkanethiols (EG4-OH) on gold have been characterized by a variety of surface analytical techniques.10 It was reported that the alkane chains show some order, while the OEG was classified as disordered on the basis of NEXAFS mea-

(1) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 10714. (2) Harder, P.; Grunze, M.; Dahint, R.; Whitesides, G. M.; Laibinis, P. E. J. Phys. Chem. B 1998, 102, 426. (3) Feldman, K.; Ha¨hner, G.; Spencer, N. D.; Harder, P.; Grunze, M. J. Am. Chem. Soc. 1999, 121, 10134. (4) Dicke, C.; Ha¨hner, G. J. Phys. Chem. B 2002, 106, 4450. (5) Dicke, C.; Ha¨hner, G. J. Am. Chem. Soc. 2002, 124, 12623.

(6) Pale-Grosdemange, C.; Simon, E. S.; Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1991, 113, 12. (7) Matsuura, H.; Miyazawa, T.; Machida, K. Spectrochim. Acta 1973, 29A, 771. (8) Tadokoro, H.; Takahashi, Y. Macromolecules 1973, 6, 672. (9) Vanderah, D. J.; Valincius, G.; Meuse, C. W. Langmuir 2002, 18, 4674.

Introduction A prerequisite for the use of thin organic layers as functional coatings is the ability to prepare films with high reliability and reproducibility. Fundamental knowledge about such coatings, e.g., on their conformational order, is essential for the understanding of the intermolecular interactions in the films and the development of a detailed picture of the film structure. This may help to synthesize molecules with a targeted interaction and a resulting tailored conformational order in the layer. The latter is one of the parameters that determine the density of the molecules on the surface and the chemical properties of the film, such as its wettability and its “reactivity” in the case of functional terminal groups. Their accessibility to reactants crucially depends on the conformation of the molecules in the film. An example of functional films that have recently attracted considerable attention are oligo(ethylene glycol) (OEG)-terminated alkanethiols.1 It was found that selfassembled monolayers (SAMs) containing oligo(ethylene glycol) can resist the adsorption of biomolecules, and some effort has been spent in elucidating the underlying mechanism.1-5 SAMs are easy to prepare while offering at the same time some flexibility toward functionalization. In addition, they are only a few nanometers thick and hence have the perspective to be also useful in connection with the current trends in miniaturization and nanotechnology. Despite the fact that for potential biological applications the environment is most often a liquid, detailed knowledge † ‡

10.1021/la0350610 CCC: $25.00 © 2003 American Chemical Society Published on Web 10/03/2003

9306

Langmuir, Vol. 19, No. 22, 2003

surements. Other groups reported that OEG attached directly to a gold substrate via a thiol group and ethylene glycols topped with alkane chains also prefer the helical conformation, independent from the length of the alkane chain, at least for EG6.11,12 The solvent used for preparation has also some influence on the quality of the film structure of EG6.9 The hypothesis of helical EG units has also been intensively studied with theoretical methods, which all agree that EG films on gold prefer the helical conformation.13-16 For monolayers adsorbed on thermally evaporated gold films, however, some distortion due to the graininess of the substrate and domains in the films can be expected, which leads to a deviation from the ideal structure. Sum frequency generation measurements confirmed that films are a mixture of helical and amorphous species and showed that the amount of helical conformations diminishes further when the SAMs are in contact with solvents.17 None of the studies, however, attempted to quantify the amount of helical vs amorphous conformations present in the films. From a fundamental point of view, this information is interesting since it provides insights into how close the films are to the “ideal state” and what might cause the deviation from it. The synchrotron-based technique of NEXAFS gives detailed structural information on thin films.18 In particular, it allows a quantification of the state of orientational order. We have investigated OEG films of different length, anchored via an undecane chain to a gold substrate. The measurements were performed in a vacuum. The preferential orientation of the ethylene glycol units should be close to the “ideal” one as suggested by calculations and IR measurements. The influence of distorting water, which cannot be excluded under ambient conditions, should be much weaker with possibly only tightly bound water molecules left in the films. The measurements allow it to quantify the percentage of conformational order in the films by comparing the experimental results with predictions derived from the simulation of the suggested ideal structure. Experimental Section Materials. Surfactants. All surfactants were composed of an undecanethiol chain (HS(CH2)10CH2-) attached to an oligo(ethylene glycol) with a varying number of ethylene glycol units (-OCH2CH2-). The OEG was terminated with a methoxy or hydroxyl group. The resulting molecular structure is HS(CH2)11(OCH2CH2)n-OR, with R ) H or R ) CH3. The synthesis of the materials was carried out according to procedures described in the literature.1,3,6 The chemical composition and purity of all surfactants were confirmed by NMR, elemental analysis, and mass spectroscopy measurements. The abbreviations EG3 (n ) 3, R ) OCH3) and EG6 (n ) 6, R ) OH) are derived from the number of ethylene glycol units in the OEG, while EG350 (n ) 8 ( 2, R ) OCH3) and PEG2000 (n ) 45 ( 11, where the deviation (10) Nelson, K. E.; Gamble, L.; Jung, L. S.; Boeckl, M. S.; Naeemi, E.; Golledge, S. L.; Sasaki, T.; Castner, D. G.; Campbell, C. T.; Stayton, P. S. Langmuir 2002, 17, 2807. (11) Vanderah, D. J.; Meuse, C. W.; Silin, V.; Plant, A. L. Langmuir 1998, 14, 6916. (12) Vanderah, D. J.; Pham, C. P.; Springer, S. K.; Silin, V.; Meuse, C. W. Langmuir 2000, 16, 6527. (13) Wang, R. L. C.; Kreuzer, H. J.; Grunze, M. J. Phys. Chem. B 1997, 101, 9767. (14) Pertsin, A. J.; Grunze, M.; Garbuzova, I. A. J. Phys. Chem. B 1998, 102, 4918. (15) Wang, R. L. C.; Kreuzer, H. J.; Grunze, M. Phys. Chem. Chem. Phys. 2000, 2, 3613. (16) Pertsin, A. J.; Grunze, M. Langmuir 2000, 16, 8829. (17) Zolk, M.; Eisert, F.; Pipper, J.; Herrwerth, S.; Eck, W.; Buck, M.; Grunze, M. Langmuir 2000, 16, 5849. (18) Tokumitsu, S.; Liebich, A.; Herrwerth, S.; Eck, W.; Himmelhaus, M.; Grunze, M. Langmuir 2002, 18, 8863.

Zwahlen et al. indicates the distribution’s e-2 boundaries, R ) OCH3) refer to the masses of the attached oligo(ethylene glycol) in atomic units. SAM Preparation. Sample substrates were prepared by first applying 5 nm of chromium with an electron beam evaporator as adhesion promoter onto plasma-cleaned silicon (100) wafers (MEMC Electronic Materials, Inc., St. Peters, MO). This was followed by thermal deposition (Balzers BAE 370 coating system, pressure ∼5 × 10-7 mbar, deposition rate of 0.5-1 nm/s) of an 80 nm thick polycrystalline gold layer (99.99%, Balzers Materials, Liechtenstein). The substrates were immersed into 2 mM ethanolic (>99.8%, p.a., Merck) solutions of the surfactant molecules EG3, EG6, and EG350 immediately after the coating, where they remained for an extended period (typically 24 h). The PEG2000 samples were prepared from dimethylformamide (DMF) as described elsewhere.18 Upon removing the samples from the solution, they were rinsed with the pure solvent to wash off any residual material. This was followed by an additional cleaning in an ultrasonic bath of ethanol before they were finally dried in a stream of nitrogen. NEXAFS. NEXAFS measurements were performed at beamline U1A of the NSLS (Brookhaven, Upton, NY). The intensity of an electronic transition with a transition dipole moment that is tilted from the surface normal by R, and an angle of incidence, θ, between the incoming photons and the surface, is given by eq 1:19

((

I(θ) ) A P cos2(θ) cos2(R) +

1 sin2(θ) sin2(R) + 2 1 (1 - P) sin2(R) (1) 2

)

)

P is the polarization of the synchrotron light, and A is a machinedependent factor. Any azimuthal dependence for substrates with a 3-fold or higher rotational symmetry averages out. The spot size of the beam on the sample surface was approximately 1 mm2 for normal incidence. A transition with a specific tilt angle of 54.7° (“magic angle” ) does not show any angular dependence on the incoming photons, similar to a completely disordered system. The latter would hence also show an apparent tilt of 54.7°. The energy resolution of the beamline optics was better than 0.5 eV at the C 1s threshold and below 1.1 eV at the O 1s edge. All experiments were performed at room temperature. The base pressure of the experimental chamber was below 10-7 mbar. All spectra were acquired in partial yield mode with the retarding voltage set to -150 V. The samples did not charge, and the prolonged (approximately 1 h) exposure to the X-rays did not result in any detectable changes in the films. The transmission function of the apparatus was determined with a sputter-cleaned, featureless gold sample. The raw data were corrected for the transmission during data analysis by dividing the individual spectra through this reference. To reduce the noise in the spectra, all data and the transmission function were additionally divided by the signal from a gold grid in the beam axis, which was recorded simultaneously with each spectrum. A hexadecanethiol (HS(CH2)15CH3) SAM on gold served as energy reference for the calibration of the carbon spectra, while the signal from the oxygen was corrected using ISEELS data of diethyl ether.20,21 Tilt angles of self-assembled alkanethiols on gold are well-known from the literature.22 They allow it to analyze the angular dependence of the NEXAFS signal from the EG-films via the method of difference spectra.23 Spectra at the 1s edges of carbon (275-325 eV) and oxygen (520-570 eV) were measured for at least four different angles between grazing (20°) and normal incidence (90°). All spectra were normalized to the absorption step. Modeling. For the simulation of the films molecular conformations were built with the program DTMM (Polyhedron (19) Sto¨hr, J. NEXAFS Spectroscopy; Springer: Heidelberg, 1992. (20) Wight, G. R.; Brion, C. E. J. Electron Spectrosc. Relat. Phenom. 1974, 4, 24. (21) 21 Urquhart, S. G.; Hitchcock, A. P.; Priester, R. D.; Rightor, E. G. J. Polym. Sci., Polym. Phys. 1995, 33, 1603. (22) Ha¨hner, G.; Kinzler, M.; Thu¨mmler, C.; Wo¨ll, C.; Grunze, M. J. Vac. Sci. Technol. A 1992 10, 2758. (23) Kinzler, M.; Schertel, A.; Ha¨hner, G.; Wo¨ll, Ch.; Grunze, M.; Albrecht, H.; Holzhu¨ter, G.; Gerber, T. J. Chem. Phys. 1994, 100, 7722.

OEG-Terminated SAMS on Gold

Langmuir, Vol. 19, No. 22, 2003 9307

Figure 1. (a) C 1s NEXAFS spectra of the molecules investigated for grazing and normal incidence and (b) difference spectra. Solid lines indicate the integration intervals used for the R*CH and the σ*CC/CO transitions (see text). Software). For the helical structure of the OEG we used the atomic positions of crystalline poly(ethylene glycol) (PEG), which were determined by X-ray diffraction and which are published in the literature.8 The crystalline PEG displays a 7/2 helix, i.e., 7 EG units perform two full rotations. The same set of data was also used in other work as a reference for IR spectra.2 Whenever less than an entire unit cell was required for the OEG part of the surfactant molecule, the entire cell was cut off after the appropriate number of EG units. Then, in a first step, the tilt angle of each individual bond, Ri, was determined. On the basis of these values, the average tilt angle, R j , for each different type of transition was obtained according to eq 2:

cos2(R j) )

1

n

∑cos

n

2

Ri

(2)

1

i.e., summation over all separate transitions contributing to a certain resonance and determination of the angular dependence of the total signal. This empirical procedure is also referred to as the “building block principle”.19 It has been shown that this principle is not valid for alkane chains in an all-trans conformation.24 We therefore performed the simulation under the assumption that the CC transition moments of alkane chains in an all-trans conformation are directed along the chain axis. These calculations yield tilt angles for the individual types of bonds (CH, CC, CO).

Results Measurements. a. Carbon Region. Figure 1a shows spectra recorded for normal and grazing incidence of the carbon region for the different molecules investigated. Figure 1b displays the corresponding difference spectra between grazing and normal incidence. The irradiation of alkanethiol SAMs with X-rays can lead to a significant modification of the films. Dehydrogenation, the loss of conformational order, and an establishing of double bonds have been reported.25 The lack of any feature around 285.3 eV, which is often detected in NEXAFS spectra of alkane monolayers and is an indicator of carbon contamination19 and the presence of double bonds, supports the quality of the films. It has also (24) Ha¨hner, G.; Kinzler, M.; Wo¨ll, C.; Grunze, M.; Scheller, M. K.; Cederbaum, L. S. Phys. Rev. Lett. 1991, 67, 851; 1991, 69, 694 (Erratum). (25) Heister, K.; Zharnikov, M.; Grunze, M.; Johannson, L. S. O.; Ulman, A. Langmuir 2001, 17, 8.

Figure 2. (a) O 1s NEXAFS spectra of the molecules investigated for grazing and normal incidence and (b) difference spectra.

occasionally been observed after prolonged irradiation of carbon-containing films with X-rays.25,26 The energetically lowest resonance, labeled 1, is found at 287.6(5) eV and establishes a shoulder in the original spectra. It is associated with CH orbitals from the alkane chain and has Rydberg character.27 For EG3, EG6, and EG350 films the intensity of this resonance varies with the angle of photon incidence, indicating ordered alkane chains, and diminishes with increasing oligomer length. The next spectral feature is the carbon absorption edge around 288-290 eV.19 Carbon atoms of the alkanes and the OEGs will have slightly different ionization potentials, which, however, can be summarized in one broader step. Superimposed is a sharp peak at 289.0(5) eV, labeled 2. The CH orbitals of the OEG give rise to this feature.10 Transitions into orbitals of carbon bonded to oxygen (σ*CO) might weakly contribute around 291.7 eV (3).21,28 The resonance centered at 292.8(5) eV (4) can be assigned to transitions into σ*-orbitals of CC and CO bonds.21,29 This resonance also shows some angular variation for EG3 and EG6 (see Figure 1). The energetically highest resonance is expected at 303(1) eV but hardly emerges from the background. It is also caused by transitions of the type C 1s f σ* (CO, CC).29 b. Oxygen Region. Figure 2a shows the oxygen spectra for the different samples and for normal and grazing incidence and Figure 2b the corresponding difference spectra. All oxygen spectra consist of a broad resonance and the absorption step. We decomposed the signal into one main Gaussian and the ionization step modeled by the error function using a least-squares fitting procedure. On the basis of ISEELS experiments of diethyl ether, the main transition is expected at 538 eV.21 We used this resonance as the energy reference for the spectra (fwhm 2.75 eV). An additional shoulder appears around 541.2(10) eV.20 The position of the absorption edge was fixed at 538.1(3) eV with a width of 1.5 eV, which is in reasonable agreement with the results of other experiments.20,21 c. Average Tilt Angles. We applied the method of difference spectra to determine the average tilt angles of (26) Ja¨ger, B.; Schu¨rmann, H.; Mu¨ller, H. U.; Himmel, H. J.; Neumann, M.; Grunze, M.; Wo¨ll, C. Z. Phys. Chem.-Int. J. Res. Phys. Chem. Chem. Phys. 1997, 202, 263. (27) Bagus, P. S.; Weiss, K.; Schertel, A.; Wo¨ll, C.; Braun, W.; Hellwig, C.; Jung, C. Chem. Phys. Lett. 1996, 248, 129. (28) Dannenberger, O.; Weiss, K.; Himmel, H. J.; Ja¨ger, B.; Buck, M.; Wo¨ll, C. Thin Solid Films 1997, 307, 183. (29) Ha¨hner, G.; Marti, A.; Spencer, N. D.; Brunner, S.; Caseri, W. R.; Suter, U. W.; Rehahn, M. Langmuir 1996, 12, 719.

9308

Langmuir, Vol. 19, No. 22, 2003

Zwahlen et al.

Table 1. Experimentally Determined Tilt Angles for the Different Transitionsa EG3 EG6 EG350 PEG2000

R*CH

CC/CO

OC

59.7(1.0) 56.7(1.1) 54.7(0.4) 55.2(0.5)

50.3(0.4) 52.5(0.3) 55.5(1.8) 56.6(1.4)

50.5(2.0) 53.6(3.0) 54.7(3.0) 54.9(3.0)

a Numbers in parentheses denote the error. The values derived from the carbon edge were determined with a hexadecanethiol as reference system via difference spectra. The values derived from the O edge are determined directly from the measured spectra.

Table 3. Theoretical Average Tilt Angles for the Molecules Investigated in a Perfectly Ordered Conformationa EG3 EG6 EG350 PEG2000 a

R*CH

CC

CO/OC

63.6 61.8 61.0 60

37.4 41.5 43.6 46

49.4 51.6 52.2 52

A two-layer model was assumed. For details see text.

Figure 3. Schematic of an EG3 film in “ideal” helical conformation (left) and in disordered state (right). Table 2. Theoretical Average Tilt Angles for the OEG Part Only, in the Helical Conformation EG3 EG6 EG350 PEG2000

R*CH

CC

CO/OC

60.6 60.0 60.2 60

45.5 47.2 46.5 47

50.8 52.2 52.5 52

an “equivalent alkane chain” for the CH and CC/CO related molecular orbitals.23 The integration interval for the R*CH resonance was fixed to 285-290 eV and set to 290-296.5 eV for the σ*(CC/CO) transition. For the error estimation, values of the experimental deviations as well as of the statistical analysis were determined. The values given denote the larger of the two figures. Tilt angles for the OC orbitals were derived directly from the original spectra by using the angle-dependent intensity of the Gaussian and assuming a degree of polarization, P, of 85% for the incident radiation. Table 1 shows the average tilt angles of the R*CH and the σ*CC/CO transitions relative to the surface normal. Modeling. The structural picture of perfectly ordered OEG containing films on gold that emerges from IR experiments and calculations consists of alkane chains in an all-trans conformation that are tilted by 33° from the surface normal with the oligomer establishing a helix with its axis normal to the surface (see Figure 3).2,9 The reason for the normal helix is an almost perfect match between its cross-sectional area and the packing density of the thiols on gold.9 For the simulations therefore this conformation was assumed for a “perfect film”, in agreement with other studies in this field. Table 2 summarizes the resulting average tilt angles of the different transitions for the OEG part only and for helical conformation with the axis normal to the surface. For the complete SAM a two-layer model with a mean free path of the electrons, λ, of 2 nm was assumed,25 where the contribution from the alkane chains is attenuated by the OEG layer by a factor of exp(-d/λ), d being the layer thickness. As a consequence, the resulting average tilt angle is mainly determined by the oligomer, and the contribution of the alkane chains decreases with increasing oligomer length. Table 3 displays the resulting average

Figure 4. C 1s NEXAFS spectra of an EG350 and a hexadecanthiol film recorded for an incidence angle of 50°. Dashed lines indicate the same transitions as in Figure 1. Table 4. Theoretical Average Tilt Angles for the Experimentally Observable Transitions (CC and CO Were Summarized in One Resonance) EG3 EG6 EG350 PEG2000

R*CH

CC/CO

OC

63.6 61.9 61.0 60

43.0 47.0 49.0 50

49.4 51.6 52.2 52

tilt angles of the different transitions for a perfectly ordered film. Because of our limited experimental resolution, it is necessary to combine the CC and CO resonance into one type of observable transition in order to compare the theoretical values with the experimental data. To get a rough idea about the oscillator strengths of the two transitions, we plotted spectra recorded for an incidence angle of 50° for C16 and EG350 in Figure 4. Spectra recorded under this angle do not show any dependence on the actual orientation of the transition moments and are therefore appropriate to estimate oscillator strengths for the different transitions. The intensities do not depend on the orientation, but on cross sections and the number of bonds involved. In C16 one transition per carbon atom contributes to the intensity of the σ* resonance around 293 eV. For the OEG part this number is 1.5. If σ*CO had a rather different cross section compared to σ*CC, the intensity of the peak around 293 eV would be rather different for EG350 in comparison to C16. However, the integrated intensities are similar. We have therefore also assumed “similar” oscillator strengths. When taking this into account, Table 4 results for the expected theoretical tilt angles. Tables 1 and 4 allow it to determine the maximum possible amount of molecules present in ordered helical conformation. The overall observed experimental values

OEG-Terminated SAMS on Gold

Langmuir, Vol. 19, No. 22, 2003 9309

Table 5. Maximum Possible Percentage of Molecules in the Helical Conformation (Numbers in Parentheses Denote Error Bars) EG3 EG6 EG350 PEG2000

CH

CC/CO

OC

58(11) 29(16)

37(4) 28(4)

79(38) 35(96)

are then attributed to perfectly ordered (tilt angles from Table 4) and disordered regions (tilt angle equals “magic angle”). The ordered fraction can be deduced from the R*CH and σ*CC/CO transition separately. Table 5 gives the resulting maximum possible percentage of molecules in a helical conformation. Discussion The signal recorded at the oxygen edge is sensitive solely to the OEG part. Consequently, it allows the determination of the orientation of the oligomer on its own. However, because of the lack of a highly ordered and well-known reference system for the oxygen region, absolute intensities of the transitions had to be determined and evaluated. The uncertainty in the exact energy position for the absorption edge as well as in the number of transitions results in error bars that are quite large. In addition, the tilt of the CO/OC bonds in helical conformation is close to the magic angle. The data evaluation of the C 1s edge eliminates some of these difficulties since a reference system has been used to determine values for R*CH and CC/CO resonances. The spectra recorded at the carbon edge show a slight variation with the angle of incidence of the incoming photons for the EG3 and EG6 films, indicating some order to be present. A similar observation was reported for EG4 films.10 The angular dependence vanishes gradually with increasing OEG length. This observation is supported by the oxygen spectra. The intensity of the shoulder at 287 eV, which is due to CH transitions of the alkane chain, shows an angular variation for EG3 and EG6 and decreases with increasing number of EG units. This demonstrates that at least some of the alkane chains are ordered and that the signal of the alkane chains is attenuated by the OEG layer; i.e., the OEG is covering the alkanethiols for all molecules investigated. For the integration of the CH related transitions the interval was chosen to cover both the contributions from the alkane chain and those from the OEG. A separation, which would allow the independent determination of the orientation of the OEG, appears to be ambiguous on the basis of the difference spectra. Contributions from CO in the CH region should be minor, since integration of the CH intensity was performed between 285 and 290 eV, while CO is expected at higher energies. There is some remaining uncertainty in the exact contributions of CO and CC to the broad resonance around 293 eV. The oscillator strengths were assumed to be similar. Both carbon atoms involved in a CC bond are contributing equally to the antibonding σ*CC orbital, while the σ*CO contains a slightly higher fraction from the carbon compared to oxygen, due to its lower electronegativity. We therefore expect its oscillator strength to be slightly higher than that of C 1s f σ*CC. However, the estimation that they are similar should be reasonable. Since the CO transitions are similar to the CCs in both energy and oscillator strength, we treated them similarly by applying the method of difference spectra with an alkane chain as a reference. The OEG is considered to be

an “alkane with energies shifted to slightly higher values” due to the oxygen, concerning the relevant antibonding orbitals accessible by NEXAFS. This view is supported by Figure 3. The experimentally determined tilt angles for the different bonds are all not too far from the magic angle, and without further knowledge an obvious interpretation would be almost complete disorder. Indeed, EG4 has been classified as disordered on the basis of NEXAFS measurements, while it was observed that the alkane chains show some preferred orientation.10 In combination with the information from IR measurements, namely the presence of bands that are indicative of the helical conformation, and based on theoretical predictions a more detailed picture can be derived. It has to be borne in mind that EG3 and EG6 consisted of oligomers with uniform length, while EG350 and PEG2000 showed a broad length distribution. In addition, the escape depth of the electrons in our experiment of 2 nm is sufficient to sense the complete OEG for EG3 (approximately 1 nm) and EG6 (approximately 2 nm), but for both longer molecules EG350 and PEG2000 only the topmost part of the OEG is significantly contributing and hence dominating the NEXAFS signal. For the shorter oligomers the results can therefore be used to determine the percentage of helical species present assuming that the nonhelical species are disordered. Then for EG3 a maximum amount of very roughly 50% is adopting the helical conformation. The corresponding value for EG6 is around 30%. There is some discrepancy in the fraction determined from the different transitions of EG3 (Table 5). The value based on the σ*CC/CO resonance indicates a lower number of helical species than those derived from the R*CH and the σ*OC transitions. One explanation could be found in the assumption that the building block principle is not applicable for all the σ*CC transitions of the alkane chains. In the “transition region” between the alkane chains and the oligomers, i.e., at the top of the alkanes, it is likely that the building block principle is valid due to the oxygen and the electronic structure change. In consequence, the transition dipole moments are parallel to the bonds. In addition, the topmost CC bonds of the alkanes are possibly in a gauche conformation, enabling the OEG helix to be normal to the surface. Both scenarios would lead to a higher average theoretical tilt angle closer to the observed value for the CC/CO transition and hence a higher percentage of molecules in the helical conformation. In the case of EG6 the transition region from the alkanes to the OEG is more effectively screened and therefore less contributing to the experimental value. The results in Table 4 are consistent with the density of the molecules in the films, which decreases slightly with increasing OEG length. Values determined by XPS show the packing density of EG3 films to be 3.61 molecules/ nm2 compared to 4.67 for alkanethiol SAMs and 3.46 for EG6 films.31 In addition, the EG segments of EG6 close to the alkane chain, which are most likely in helical conformation, will be screened by the upper EG segments in the molecules, which are expected to have a high amount of gauche conformations. This may also account for the slight decrease of helical species determined for EG6 in comparison to EG3. No value for the amount of helical species, however, could be deduced for the longer oligomers EG350 and (30) Matsuura, H.; Miyazawa, T. J. Polym. Sci., Part A 1969, 2, 1735. (31) Herrwerth, S.; Eck, W.; Reinhardt, S.; Grunze, M. J. Am. Chem. Soc., in press.

9310

Langmuir, Vol. 19, No. 22, 2003

PEG2000. Because of their length distribution and gauche defects predominating at the top of the molecules, a higher degree of disorder is expected in these films. This is confirmed by our NEXAFS results. It was reported that also PEG2000 films show some amount of helical conformations based on IR spectroscopy.18 For these longer OEGs a high number of the upper EG units will remain in the gauche conformation due to the chain length distribution. NEXAFS is most sensitive to this upper part while the signal from the inner part of the film is significantly screened. The NEXAFS data are therefore consistent with helical EGs close to the anchoring alkane chain and disordered upper ones. Hence, the disorder in the longer molecules observed with NEXAFS has at least partially to be attributed to their length distribution, which results in a “fraying” of the topmost interface. For the longer molecules of the distribution, the upper EGs will have some more space available than those in the densely packed lower part and will contain a high density of gauche conformations. The highest number of defects is concentrated at the top of the molecules. This outermost layer is sensed with NEXAFS, while IR is not sensitive enough to this “submonolayer density” and its main signal stems from the denser “bulk” part, still indicating a significant amount of helical conformations.18 For the shorter oligomers EG3 and EG6 the percentage deduced for highly oriented molecules give an upper limit of the perfectly ordered amount possible. In case the remaining (not perfectly ordered) molecules still show some orientation, the amount of perfectly ordered molecules will be lower. The reason for the high amount of molecules that are not in the helical conformation is at least partially due to the nonideal substrates, inevitable defects, and domains in the films. It should be borne in mind that the evaporated gold films typically establish crystalline grains of 20-100 nm diameter, depending on the preparation conditions. Such gold films are most often employed as substrates for the adsorption of thiol SAMs. SAMs are composed of domains, and molecules at domain boundaries are typically “disordered” and hence lower the conformational degree of order present in the films. Our results indicate that even for EG3 and EG6 films a significant number is not adopting the helical conformation. The percentage of highly ordered molecules might increase if high-quality ultraflat gold substrates are used,

Zwahlen et al.

but other factors also influence the degree of order in the films.32 However, evaporated gold films will likely continue to play the most important role for the study of thiol films and their potential applications. Conclusions SAMs adsorbed on gold exposing oligo(ethylene glycol) groups of different length were investigated with NEXAFS. Spectra recorded for films of shorter oligomers displayed a weak angular dependence of the incoming photons that diminishes with increasing length. Although the average tilt angles for the individual bonds are close to the magic angle, a quantification of the structural order in OEGterminated films is possible if taking IR results and theoretical studies into account. The deviation of the films from the proposed ideal structure can be determined on the basis of the experimental data and the theoretical values of an ideal film. We have assumed that a certain fraction of the monolayer is adopting the helical conformation, while the rest contributes to disorder. This allows it to estimate an upper limit for the fraction of the shorter oligomers displaying the suggested ideal molecular conformation. The longer species show complete disorder in the upper part that is probed with NEXAFS, due to their length distribution and the resulting fraying of the films at the top. For the shorter oligomers NEXAFS can give a rough quantitative measure of molecules in the helical conformation. On thermally evaporated gold substrates even for the shorter oligomers a significant amount of the monolayer films is amorphous. Acknowledgment. We thank M. Buck and N. Richardson for fruitful discussions and C. Dicke for help with the sample preparation. Research was carried out at beamline U1A at the National Synchrotron Light Source, Brookhaven National Laboratory, which is supported by the U.S. Department of Energy, Division of Materials Sciences and Division of Chemical Sciences. The work was funded by the Swiss National Science Foundation. G.H. thanks the Deutsche Forschungsgemeinschaft (DFG) for a Heisenberg-Fellowship. LA0350610 (32) Vanderah, D. J.; Arsenault, A.; La, H.; Gates, R. S.; Silin, V.; Meuse, C. W.; Valincius, G. Langmuir 2003, 19, 3752.