Investigation of intermediate steps in the self-assembly of n

M. Fonticelli, O. Azzaroni, G. Benítez, M. E. Martins, P. Carro, and R. C. Salvarezza .... H. U. Müller, M. Zharnikov, B. Völkel, A. Schertel, P. H...
0 downloads 0 Views 480KB Size
Langmuir 1993,9,1955-1958

1955

Investigation of Intermediate Steps in the Self-Assemblyof n-Alkanethiols on Gold Surfaces by Soft X-ray Spectroscopy G. HCihner, Ch. WO11,'M. Buck, and M.Grunze Institut fur Angewandte Physikalische Chemie, Universitat Heidelberg, Im Neuenheimer Feld 253,6900Heidelberg, Germany Received May 7,1993. In Final Form: May 19,1993 Near-edge X-ray absorption fine structure ( N E W S ) spectroscopy was used to investigate the intermediate steps in the self-assembly of the long-chain thiol CHs(CH2)21SH on gold surfaces. After a first rapid adsorption process the alkyl chains are strongly entangled. The slow, second step observed can be identified as an ordering process where previously by Bain et al. (J.Am. Chem.SOC.1989,111,321) the initially entangled alkyl chains are gradually straightened. The process of self-assemblythus resembles the grafting of end-functionalized polymers to a surface ("polymer brushes"). Although the self-assembly of n-alkanethiols on Au and Ag surfaces has been studied extensively in the past yearsY1 little is known about the intermediate steps in this cooperativeprocess of self-assembly. Of particular interest is whether the high degree of orientation of the molecular axes and the lateral order observed for these films are also present in the intermediate phases (as would be expected for an island-type growth mode). So far, however, only the kinetics of the film formation have been investigated. With the help of ex situ thickness determinations using ellipsometry, Bain et were able to demonstrate that only the first adsorption step, which leads to an adsorption of 80 % of a monolayer in the case of docosanethiol (CH3(CH2)21SH), can be described by simple diffusion-controlled Langmuir kinetics. The adsorption of the last 20% was found to take place with a significantly longer time constant. This second step was attributed to a consolidation of the film, but more precise information on the nature of this process could not be provided.2 Recently the first step in the adsorption process of docosanethiol on gold was followed in real time by second harmonic generation (SHG).3 By correlating with an ex situ X-ray photoelectron spectroscopy (XPS) coverage determination,this in situ study confirmed earlier findings2 in that the first step can be described by Langmuir adsorption kinetics. The second step of the adsorption process could not be investigated in this work. Also information on molecular and/or lateral order could not be extracted from these experiments as for this system the SHG signal is mainly sensitive to the change in electronic properties of the metal-thiol interface. Infrared spectroscopy (IRS) has been used in the to determinethe molecular orientation in these types of films, but no structuralinvestigations of the intermediate phases have been carried out with IRS so far, probably owing to intensity problems. In the case of end-functionalized polymers the adsorption process on substrates leads to a strong stretching of the adsorbed, initially entangled polymer chains, so-called N

* To whom correspondenceshould be addressed (BITNET CC1 DHDURZ1). (1) Dubois, L. H.; Nuzzo,R. G. Annu. Reu. Phys.Chem. 1992,43,437 and references therein. (2) Bain, C. D.; Troughton, E. B.; Tao, Y. T.; Evall, J.; Whitesides, G. M.; Nuzzo,R. G. J. Am. Chem. SOC.1989,111, 321. (3) Buck, M.; Eisert, F.;Fischer,J.; Grunze, M.; Triger,F. Appl. Phys. A 1991,53, 552. (4) Nuzzo,R. G.; Korenic, E. M.;Dubois, L. H. J . Chem. Phys.1989, 93, 767. (5) Nuzzo,R. G.; Dubois, L.H.; Allara, D. L. J. Am. Chem. SOC.1990, 112, 558.

"polymer brushes"? In this case also a two-step adsorption process has been obtained from a theoretical analysis? where a Langmuir-typeadsorption is followed by a second step where the polymer tangles are stretched. Experimental data on this second step are limited to thickness determinations8 and are rather scarce, despite a considerable theoretical interest in this formation process. To our knowledge no direct experimental information on the orientation of the polymer subunitsis available at present. In the present study we have investigated the molecular orientation in the intermediate phases of the formation of self-assembled films of thiols on gold by employing N E W S (near-edge X-ray absorption fine structure spectroscopy), a technique which possesses the required sensitivity and allows for a fast direct determination of molecular orientations, including alkyl chains? We have used this technique previouslyloto determine the orientation and molecular conformation of alkyl chains within adsorbed thiol films which were prepared according to the standard recipell (immersion in a 1 mmol/L saturated solution for 24 h). As N E W S cannot be used in situ the intermediate steps of the self-assemblyprocess were investigated by transferring thiol-covered Au substrates directly from solution into an ultrahigh vacuum (UHV) analysis chamber. The immersion times in a strongly diluted 3 pmol/L solution of docosanethiol (CHs(CH&4H) in ethanol ranged from a few seconds to 43 h. At this concentration the fast first adsorption step is completed after about 10 min and yields roughly 80% of a monolayer.2 In order to avoid artifacts caused by different conditions of the Au substrates, all samples for one series of measurements were cut from the same Si wafer which had been sputter-coated with 100 nm of Au. A total of three different series have been investigated. NEXAF'S spectra were recorded at the beamline HETGM I1 at BESSY in the partial yield mode with a retarding voltage of -150 V at the C 1s edge with the samples at room temperature. The irradiation angle was varied between 90° (normal incidence) and 20° (grazing incidence). Normalization to the incident photon flux was carried out by dividing the raw N E W S spectra by the (6) Milner, S. T. Science 1991,251,905. (7) Ligoure, Ch.; Leibler, L. J. Phys. (Paris)1990,51, 1313. (8) Tauton, J. H.; Toprakcioglu, C.; Fetters, L. J.; Klein, J. Nature 333, 1988, 712. (9) Outka, D. A.; Whr, J.; %be, J. P.; Swalen, J. D. J . Chem. Phys. 1988,88,4076.

(10)HBhner, G.; Kinzler, M.; Thbmler, C.;Wall, Ch.; G ~ m z eM. , J. Vac. Sci. Technol., A 10 (4), 1992, 2758. (11) Strong, L.; Whitesidea, G. M. Langmuir 1988,4, 546.

0743-7463/93/2409-1955$04.00/00 1993 American Chemical Society

Letters

1956 Langmuir, Vol. 9, No. 8, 1993

273 5

Photon energy [eV]

323 5

273 5

Photon energy [eV]

323 5

Figure 1. Left panel: Series of N E W S spectra for different angles of incidence 6 ranging from 20° to 90° and an immersion time of 10min in a 3 pmol/L solution of docosanethiolin ethanol. Right panel: Same as left panel but for an immersion time of 43 h. Although 80% of the final coverage is reached after 10 min, there is still a substantial change in the NEXAFS spectra, indicating considerable reorientation. monochromator transmission function determined on a sputter-cleaned Au sample. For low coverages the separation of the signal of the thiol layer formed on the Au surface from the signal of carbon contaminants present on the surface prior to adsorptioncausesconsiderabledifficulties. This is to some extent due to the fact that these contaminations (mainly hydrocarbons) are displaced upon adsorption by the thiol m~lecules.~JJ Whereas the contribution of the contaminants present before thiol adsorption to the NEXAFS spectra can be determined by measurementson a reference sample, their reduction upon thiol adsorption makes the data analysis complicated. In the present work, however, we want to concentrate on the second, slow step in the adsorption process. At the beginning of this regime roughly 80 5% of the total coverage is reached2so that any effects due to initially present contaminationsare virtually removed. The left and right panels in Figure 1 show a series of NEXAFS spectra for immersion times of 10 min and 43 h, respectively. The series of spectra after 43 h are nearly identical to those obtained for samples immersed for 24 h in a 1mmol/L solution.1°J2 They show a strong angular variation of the most prominent features, the (C-H)* resonance at 287.5 eV and the a*(c-c) resonance at 293.4 eV, and thus indicate the presence of a well-ordered thiol film where all molecules are tilted by 3 5 O with respect to the surface normal. On the other hand, the spectra obtained for the f i i with ashorter immersion time (Figure 1)show a significantlydifferent behavior; e.g., the angular variation of the (C-H)* resonance now is almost isotropic and thus indicative of poor order. These two series of spectra clearly demonstrate that shorter immersion times produce less-ordered thiol films. That the difference in the data shown in Figure 1is indeed due to the difference in immersion times and does not depend on the total age of the filmhas been checked by recordingNEXAFSspectra for thiol films with the same immersion time but with different ages (from 2 to 48 h). No significant differences were found. In order to obtain more precise information on the kind of molecular order present in the initial stages of ordering (Figure 11,we applied the analysis scheme which has been (12) H h e r , G.; Kinzler, M.; Wall, Ch.; Grunze, M.; Scheller, M.; Cederbaum, L. S. Phys. Rev. Lett. 1991, 67, 851. See also: Erratum. Phys. Rev. Lett. 1992, 69, 694.

developed for the analysis of N E W S spectra of alkyl chains in a trans conformation,as in thin films of paraffin, LB films, and well-ordered films of thiols.1°J2 Briefly, in this analysisprocedure the different resonanceintensities are determined as a function of the photon incidence angle by using a multistep fitting scheme. The angular variations of the (C-H)* resonanceand the a * ( ~resonance ) resulting from this analysis scheme, however, were found to be inconsistent for the immersion time of 10 min; Le., each of the two resonances indicated a different molecular orientation. This observation may be puzzling at first sight since the poorly ordered films produced by shorter immersiontimes are chemicallyequivalent. But, of course, the conformationsof the alkyl chains are expected to differ for different immersion times. Whereas for well-ordered thiol films there is strong evidence from infrared spectroscopy that the alkyl chains are mostly in a trans conformation,4*6the intermediate phase may well contain a significant concentration of gauche conformations.13 We therefore relate our problems encountered in the application of the analysis scheme developed previously for NEXAFS spectra recorded for systems with alkyl chains in a trans conformation to thiol films formed by short immersion times to changes in the electronic structure of the alkyl chains caused by the presence of gauche conformations. Independent support of this explanation is providedby the analysisof NEXAFSspectra obtained for multilayers of calcium arachidate LB films at temperatures between 20 and 120 OC, which shifts of the (C-H)* resonance as large as 0.6 eV.14 As the precise electronic structure of kinked alkylchains is not known, a quantitative analysis of the NEXAFS spectra for the thiol films for short immersion times is not possible at present. Even if the position of resonancesfor a particular conformation were known, the presence of a variety of different conformationswould make an analysis extremely complicated. In order to obtain a semiquantitative parameter describing the amount of order in the thiol films, we have used a modified analysis procedure. Instead of fitting peaks of gaussian shape to the experimental data, the absolute intensities in the C-H region (285.5-289.5 eV) and the C-C region (289.5-297.5 eV) in the difference spectra of 20' and 90° are computed by numerical integration. This quantity is then divided by the corresponding results for a well-ordered thiol film, thus obtaining two "order parameters", ICHand Icc. Since the total areas in the C-H and C-C regions are taken rather than the intensity of a specific resonanceat a fixed energy, small shifts of the resonances will not show significant effects. For randomly oriented alkyl chains we will obtain ICH = Icc = 0 as any isotropic contribution will be removed when calculting the difference spectra. For well-oriented thiol films, on the other hand, we will obtain ICH= Icc = 1.

Figure 3 shows the dependence of these two normalized intensities versus immersion time. Whereas ICC is almost constant, strong changes are observed for ICH. The transition dipolemoment (TDM) for the u*(c-c)resonance at 293.4 eV has been previously shownto be oriented along the molecular axis for a chain in a trans conformation.12 As this suggests a rather nonlocal nature of the corresponding molecular orbital, it may be strongly affected by conformational changes and is not expected to provide a good measure of the C H 2 subunit orientation. On the other hand the (C-H)* resonance at 287.5 eV corresponds to a (13) Hautmmn, J.; Klein, M. L. J. Chem. Phys. 1989,91, 4994. (14) Schertel, A.;H h e r , G.; Wall,Ch. To be publihed.

Langmuir, Vol. 9, No. 8, 1993 1957

Letters

273.5

Photon energy [eV]

323.5

Figure 2. Difference spectra of grazing and normal incidence for the different immersion times t. Note that the C-H-type resonances change sign for short immersion times. This is due to changes in the electronic structure and cannot be explained solely on the basis of a geometrical disorder. The vertical lines indicate the C-H and the C-C regions used in the numerical integration (see text).

1 + C-H

+

C-C

j

Figure 3. The normalized intensities of the C-H-type region and the C-C-type region (see Figure 2) plotted against the immersion time. Note that the scale of the z axis is logarithmic. The integral of the C-H-type features can be viewed as an order parameter (see text). The last point on the right is that of a reference sample which was left for 27 h in a 1mmol/L solution (correspondingto 9OOO h in a 3 pmol/L solution). transition into a molecular orbital with a rather local nature,12and its angular variation is expected to provide information on the orientation of the CH2 subunits. Indeed the time constant observed for ICH,which may be viewed as an order parameter giving the orientation of the CH2 subunits, agrees very well with the observations of Bain et al.2 On the basis of the series of NEXAFS spectra in Figure 1, an island type of growth can be excluded for the formation of long-chain alkanethiols on gold, as in this case the average tilt angle of the molecules should not show any pronounced dependence on immersion time. Also the presence of randomly distributed alkyl chains in an all-trans conformation can be ruled out on the basis of the difference spectra shown in Figure 2 (see above). In previous work2 the second step in the adsorption process has been explained by a consolidation process, including lateral diffusion of both substrate atoms and adsorbed molecules, and removal of solvent molecules. As our results on the order in the films do not show any

dependence on the age of the films and were carried out in UHV, where all ethanol molecules should be removed after a few seconds, we feel that we can rule out these processes. We also believe that the changes in substrate morphology observed recently with STM16 occur in the first adsorption step and are not related to our present findings for the second step. Our data instead indicate the presence of a high density of gauche conformations along the alkyl chains after completion of the first adsorption process. This may not be too surprising if one considers that if solvated in a not too good solvent (as in this case ethanol), the alkanethiols-because of entropy maximization-form tangles of roughly spherical shape. This of course implies a high density of gauche conformations. Actually,in the case of polymers with much longer chain lengths the adsorption process of end-functionalized polymers onto substrates has been studied in some detail, both theoretically and experimentallyes Although experimental data are rather scarce, a theoretical analysis also finds a two-step adsorption pro~ess.~ After the substrate is brought into contact with the solution, the unperturbed, end-functionalized polymer tangles-which are of spherical shape-adsorb from the solvent with simple Langmuir kinetics. This first,fast step is completed when the surface is covered with the still unpertubed polymer tangles. For incorporation of additional material, the adsorbed, spherical tangles have to be deformed. Here the entropy loss caused by the deformation (or stretching) of the polymer chains is compensated by the binding energy gained in the adsorption process. For high binding energies, this second, slow adsorption process finally leads to strongly stretched polymer brushes. We propose that the same description is also valid for the adsorptionof docosanethiolon gold surfaces. Although this thiol is only marginally long enough to be called a polymer, the basic physics is the same. Indeed our data for the second adsorption step directly reveal the conformational changes leading to a stretching of the alkyl chains which so far has only been seen indirectly through the increase in thickness.2 The proposed entropy-controlled adsorption process of course implies that there is a coverage increase of the thiol films also in the second adsorption step. Actually, a rough estimate of the film thickness can be obtained from our NEXAFS data if the intensity before the edgejump (which is mostly due to secondaryelectrons from the Au substrate) is compared to the height of the edge jump. The substrate pre-edge signal I is attenuated by the thiol film with thickness d accordingto IB= exp(-d/X), whereas the height of the edge jump is proportional to I, = 1 - exp(-d/h). Using a value of 10A for the secondaryelectron mean free path X,16 our NEXAFS data indicate an increase in film thickness of roughly 20% during the second adsorption step, in agreement with previous results.2 Unfortunately, because of the short chain length, the time dependence of our order parameter, the normalized intensities ICHof the C-H-type resonance as shown in Figure 3, cannot be compared quantitatively to the results of the previous self-consistent mean-field calculations, although qualitatively the predicted logarithmic time dependence of the ordering process7is nicely reproduced (15) Edinger, K.; GBlzhHuser, A.; Demota, K.; WBll,Ch.; Grunze, M. Langmuir 1993, 9, 4. (16) With our partial yield method we detect electrons with energies above 160 eV. For an average energy of 200 eV we obtain X = 10 A by scaling the known escape depth of electrons with an energy of 1200 eV (35 A13 with the %own energy dependence of X(J9 = Eo.7.19 This value for X also agrees wth the univerenl curve of Seah and Dench.”

1958 Langmuir, Vol. 9,No. 8, 1993

by our data. For this case of rather short polymer chains molecular dynamic calculations such as those employed for the melting of thiol film@ or Langmuir-Blodgett films" are required. In conclusion our data confirm earlier observations2that the self-assembly process by which the thiol molecules form well-ordered monolayers on Au substrates proceeds through two steps with time constantsdiffering by roughly 2 orders of magnitude. In the first step the molecules adsorb on the Au surface, replacing possible hydrocarbon contaminations and forming bonds between the terminating S atoms and the Au surface atoms. Our data reveal that at this stage the alkyl chains form tangles with a very high concentration of gauche conformations. Subsequently, with a considerably longer time constant, this (17)Ulman, A. A d a Mater. 1991, 3, 298. (18)Hansen, H. S.; Tougaard,S.; Biebuyck, H. J. Electron Spectrosc. 1992,58,141. (19)Penn, L.R. J. Electron Spectrosc. 1976,9,40. (20)S e d , M.P.;Dench, W . A. Surf. Interface Anal. 1979,1, 2,

Letters conformational disorder is gradually removed by a stretching process until well-oriented films with their molecular axes aligned at an angle of 35" away from the surface normal are reached. The formation process is thus identical to that of polymer brushes.s Our data therefore seem to be the first direct observation of the molecular stretching process by which the brushes are formed.

Acknowledgment. Experimental help from Dr. W. Braun and M. Mast at BESSY and A. Schertel are gratefully acknowledged. We would also like to thank Dr. W.Schrepp from BASF for providing us with the goldcoated Si wafers, M. Kinder for help with the data evaluation, and Dr. C. R. Brundle for comments on the paper. We are grateful to Dr. J. Stiihr for discussions concerning N E W S data analysis. This work has been funded in part by the Deutsche Forschungsgemeinschaft (Grant SFB 247/E1) and the German Bundesministerium fiir Forschung und Technologie (Grant PS-05473FAB2).