0 Copyright 1994 American Chemical Society
JUNE 1994 VOLUME 10,NUMBER 6
Letters Chemisorption of 11H-EicosafluoroundecanoicAcid Monomolecular Film from a Liquid Phase onto Si(ll1) Surface: Study by X-ray Photoelectron Spectroscopy Munehisa Mitsuya Advanced Research Laboratory, Hitachi Ltd., Hatoyama, Saitama 350-03,Japan Received January 26, 1994. In Final Form: April 19,1994" Adsorption of 11H-eicosafluoroundecanoicacid from a hexane solution onto fluorine-terminated Si(ll1) surface has been examined using X-ray photoelectron spectroscopy. The decrease in the Fls peak intensity assigned to the surface Si-F bond is accompanied by a quantitative increase in the Fls and Cls peak intensities ascribed to the fluorocarbonchain of the acid. It is shown that each molecule is covalently bonded to the silicon surface giving monomolecular coverage.
Introduction Coverage of solid surfacesby organic thin filmshas been intensely studied from both fundamental and practical points of view.' Although monomolecular coverage of a solid surface is attainable by adsorption of molecules from a gas phase, spontaneous adsorption from a liquid phase (self-assembling)is widely employed since the 1940~.~ It has been shown that, contrary to vapor-phase-deposition, closely-packed monolayers are readily obtained by selfassembling due to the absence of a severe kinetic barrier.* The main focus in this subject area has been the relationship between the microstructure of ordered molecular arrays and their collective properties such as wetting or electrochemical electrode activities. This traditional approach is also relevant to the building of supramoleculararchitectures on surfaces. A current topic of interest in this field is the self-assembling of fullerenes, where a spherical carbon molecule is covalently bound to a surface by a molecular tether.' These tethers can consist of silane compounds on metal oxide or of thiols on gold.
In the present work, the author has concentrated on a single crystal of silicon as a-substrate, the aim being to link molecular tethers directly to the surface of elemental silicon, rather than via an oxide layer. This substrate is of great practical interest for the following two reasons. Firstly, it can be endowed with appropriate conductivity by different amounts of doping. This characteristic is essential for device application in the future. Secondly, a high-quality surface can be attained by conventional chemical treatment; it has been reported that chemical oxidation followed by aqueous HF etching results in the removal of the surface oxide and leaves behind silicon surfaces. Scanning tunneling microscopeobservation has revealed atomic scale images with flat terraces, leaving virtually no surface dangling bonds? Infrared absorption spectra of the HF-treated silicon have shown that the surface is mainly covered by covalent Si-H bonds.- Another species on the surface is the Si-F
Abstract published in Advance ACS Abstracts, June 1, 1994. (5)Becker, R.S.;Higaahi, G. S.;Chabal, Y. J.; Becker, A. J. Phys. Reo. Lett. 1990,66,1917. (1) Swalen,J.D.;Allara,D.L.;Andrade,J.D.;Chandrose,E.A.;Garoff, (6)Ubara, H.; Imura, T.; Hiraki, A. Solid State Commun. lS84,60, S.;Israelachivili, J.; McCarthy, T. J.; Murray, R.; Pease, R. F.; Rabolt, 673. - . -. J. F.; Wynne, K. J.; Yu, H. Longmuir 1987,3, 932. (7) Yablonovitch, E.; Allara, D.L.;Chang, C. C.; Gmitter, T.; Bright, (2)Forexample,Bigelow,W.C.;Pickett,D. L.;Ziman,W.A.J. Colloid T. B. Phys. Rev. Lett. 1986,57,249. Sci. 1946,I , 513. (8) Higaahi, G. S.;Chabal, Y. J.; Trucks,G. W.;Raghavachari,K. Appl. (3) Dubois, L. H.; Zegarski,B. R.; Nuzzo, R. G. Langmuir 1986,2,412. Phys. Lett. 1990,56,656. (4)Chem. Eng. News 1993,September 20,31.
0743-746319412410-1635$04.50/0 0 1994 American Chemical Society
1636 Langmuir, Vol. 10, No. 6,1994 bond.%12 Si-F bonds are minor species when treated with diluted HF solutions, while the density increases with increasing HF concentration. Since these bonds are active and easily replaced by Si-OH groups upon contact with water,11J2 it is thought that each Si-F bond is used as a reaction site for molecular assembling. Adsorption kinetics was examined by X-ray photoelectron spectroscopy (XPS) in this work, since the highsurface-sensitivity of XPS is suited to the detection of ultrathin overlayers. Another favorable factor is that measurement in an ultrahigh-vacuum environment prevents the fluorinated surface from being contaminated and being attacked by water molecules in air. 11H-Eicosafluoroundecanoicacid H(CFz)&OOH was chosen as an adsorbate for the following two reasons. Since the major components of atmospheric contamination are hydrocarbons, we can distinguish the adsorbate from the contamination by the chemical shift of core energy of carbon and fluorine. The second reason comes from the chemicalstability of a fluorocarbon chain. X-ray-induced sample damage is a common problem for organic compounds and can cause the spectrum to change with exposure time. Fluorinated compounds degrade almost imperceptibly slowly because of the large binding energy of the C-F bond.
Experimental Section Polished n-type (111) oriented silicon wafers were wed as substrates with no orientation-dependent results observed. The substrate was first oxidized in boiling HCI/H2Oz/H20 solution (molar portion 1:1:13) and dipped in 1% HF solution to remove the amorphous surface oxide. After several repetitions of this procedure,the oxidizedsiliconwasimmersed in 50%HF solution (semiconductor use grade). The final HF immersion step produces a silicon surface terminated with fluorine bonds. The fluorine density is reported to be about half of a monolayer.8 The fluorinated silicon was then exposedto a hexane solution of 11H-eicosafluoroundecoicacid (FluorochemLimited),which was recrystallized from a toluene solution. Nitrogen gas was bubbled into the solution prior to and during the exposure.It has been ascertained from XPS that hexane does not react with a fluorinated silicon surface. After rinsing with tridistilled hexane severaltimes, the sample was introducedto a loadingchamber within 1min. Measurements were performed usinga Vacuum-GeneratorESCA-LabMARK-2 equipped with a hemispherical analyzer and a Mg Ka X-ray source (1253.6 eV). The vacuum system has a base pressure of 10-7 Pa. The irradiation anglewas 60" with respect to the surface normal,to assure both surface sensitivity and the reproducibility of peak intensities. Binding energieswere referenced to Ag 3d6,2 set at 361.9 eV. Results and Discussion Adsorption kinetics was followed by XPS on different samples coming from the same wafer, all of which underwent the same treatment. No spectrum change was observed with increasing acquisition time in either peak intensities or widths. The results are summarized as follows. Cls. HF-treated silicon surfaces showed a weak carbon peak at a binding energy of 285 eV, typical of C-C and C-H bonded carbons. Neither a low binding-energy (9)Weinberger, B. R.; Peterson, G.G.;Eschrich, T. C.; Krasinski, H. A. J.Appl. Phys. 1986,60, 3232. (10) Takahagi, T.; Nagai, I.; Ishitani, A.; Kuroda, H. J. Appl. Phys. 1988,64, 3516. (11) Gru, D.: Grunder, M.: Schultz, R. J. Vac. Sci. Techml. A 1989. 7, soa.
(12) Watanabe, S.;Nakayama, N.; Ito, T. Appl. Phys. Lett. 1991,59, 1458.
Letters
x Y
.-
v1
E 0) Y
C I
695
690
685
Binding Energy J eV Figure 1. Sequenceof XPS spectra of the Fls peak after different immersionconditions. Immersiontimes are (a)0, (b) 10min, (c)
30 min, (d) 60 min, and (e) 240 min each at room temperature. Spectrum f was recorded after immersing in 60 "C solution for 120 min. component suggestive of S i 4 bonds13J4nor a high bindingenergy component suggestive of C-F bonds was observed. Moreover, rinsing the substrate in distilled hexane decreased carbon intensities. Therefore, this carbon signal probably originates from a physisorbed hydrocarbon layer not chemically bonded to the silicon. Possible sources of carbon contamination are plasticizers in the storage containers of the HF solution and hydrocarbons in the atmosphere. The intensity ratio Cls/Si2p was in most cases less than 1% . But reliable quantitative determination of carbon coverage on the surface is difficult to obtain due to uncertainties about the photoelectron escape depth and the thickness of the contamination layer. Immersing the fluorinated silicon in a hexane solution of 11H-eicosafluoroundecanoicacid brought about two kinds of spectral changes. The first was the growth of a new peak with a binding energy of 292 eV, which can be attributed to carbon atoms covalently bonded to fluorine atoms. The intensity of this peak was alwaysproportional to that of the fluorine peak, which will be shown in the following section. The second was the growth of a tail at a binding energy of 287 eV, assigned to carboxylic carbon. These results confirm the adsorption of fluorinated carboxylic acid on the silicon surface. Fls. The sequence of F l s photoelectron spectra is shown in Figure 1 with immersion time as a parameter. HF-treated Si(ll1) surface shows a peak at a binding energy of 685.5 eV, as depicted in spectrum a, which is associated with a Si-F bond.ls Spectra denoted by &e were measured after subsequent immersion in the 1123eicoaafluoroundecanoicacid solution at room temperature for 10, 30,60, and 240 min, respectively. It can be seen that the decrease in a Si-F peak is accompanied by the development of a new peak at a binding energy of 689 eV, which can be attributed to fluorine covalently bonded to carbon. Each spectrum was decomposed into components by least-squares fitting after background subtraction. The ratio of the increased 689-eV peak area to the decreased 685.5-eV peak area was always in the range 7 f 1. (13) Gray, R.C.; Carver,J. C.; Hercules, D. M. J. Electron. Spectrosc. Relat. Phenom. 1976, 8, 343. (14) Lee, W. Y. J. Appl. Phys. 1980,51, 3365. (15) References in ref 9.
Letters
Langmuir, Vol. 10, No. 6, 1994 1637
The intensity of the 689-eV peak reaches saturation a t an immersion time of about 60 min. Continued immersion results in only small spectral changes. The Si-F band is still observable in spectrum e. the remaining Si-F is considered to be in a sterically hindered position against the attack of the molecule. This peak disappeared when the solution was heated; spectrum f showsthe surface after immersion a t 60 "C for 120 min. The intensity ratio F l s (689eV)/Cls (292 eV) was always 8.5 f 1, independent of immersion time and solution temperature. Taking the atomic sensitivity ratio of F/C to be 4.8,16 an atomic density ratio F/C of about 2 can be obtained. Therefore these two peaks are consistent with the assignment to the fluorocarbon chains of the acid. The escape depth of photoelectrons through the fluorocarbon chain has been reported to be 2.3 nm in the ultraviolet region.17 In the X-ray region, an escape depth of 3-4 nm has been reported for hydrocarbon chain.18This means that the emission from the Si-F is halved by the physisorbed monomolecularlayer, even if the fluorocarbon chain was oriented along the surface normal. Therefore, it can be said that the surface fluorine is replaced by chemisorbed fluorocarboxylic acid. 0 1 s and Si2p. HF-treated silicon showed a weak and broad peak in the range 530-534 eV for 01s band, which can be attributed to physisorbed and chemisorbed oxygen atoms on the silicon surface. The corresponding Si2p spectra showed an elemental silicon peak (99.7 eV) with an unresolvable tail in the higher binding energy region (100.7-102 eV). The magnitude of the chemical shift has been reported to be 1.3 and 2.3 eV for Si-F and SiF2, respectively.16 On the other hand, Hollinger and Himpsel demonstrated that four distinctive oxidation states exist for si1ic0n.l~ These shifts are 1.0 eV (Si1+),1.8 eV (Si2+), 2.7 eV (Si3+),and 3.5 eV (Si4+),where the superscript refers to the number of oxygen atoms bonded to one silicon atom. Thus the high-energy tail of the spectra can be attributed to both fluorine and oxygen atoms bonded to the surface.
Immersing the substrate in the solution enhanced the 533-eV peak. This is attributedto oxygen in the carboxylic group. During the course of the adsorption corresponding to Figure lb-e, there was virtually no change in the highenergy tail of the Si2p spectra. This must be due to the fact that the oxygen replacing the fluorine induces comparable chemical shifts in silicon. Heating the solution brought about a remarkable spectral change in both the 01s and Si 2p regions; the 533-eV peak was further increased in the 01s region and a new band was observed at 103 eV in the Si2p region. This indicates the growth of a thick silicon dioxide layer. Since no appreciable change was observed when the HFtreated silicon was heated in pure hexane, a side reaction such as the decomposition of the acid and the subsequent invasion of oxygen through the silicon lattice is thought to have occurred. Summarizingthe results, the systematic change of X-ray photoelectron spectra indicates the growth of monomolecular film chemisorbed on silicon surface. Since there are 20 fluorine atoms in one molecule, the atomic density ratio of increased CF2 to decreased SiF (7 f 1) suggests that each molecule is adsorbed bidentate, that is, each molecule is bonded at two active sites. Fluorinated carboxylic acid was chosen in this study for facility of detection. However, in the course of the study, it was ascertained that other carboxylic acids with long hydrocarbon chains do not react with the fluorinated surface. Alkylene oxides, such as 1,Zepoxydodecanewere also inactive. However,fluoroalkylene oxides and glycidyl ethers showed sufficient reactivity with the fluorinated silicon. It seems that elementswith large electronegativity, such as fluorine and oxygen, activate terminal functional groups. A detailed study is in progress is an effort to understand the chemical functionalization of silicon surfaces.
(16) Handbook of X-ray Photoelectron Spectroscopy; Muilenberg, G. E., Ed.;Perkin-Elmer Corp.: Eden Prairie, MN, 1979. (17) Miteuya, M.; Seki, K.; Inokuchi, H. J. Appl. Phys. 1988,64,4150. (18) For example, Brundle, C. R.; Hopster, H.; Swalen, J. D. J. Chem. Phye. 1979, 70,5190. (19) Hollinger, G.; Himpeel, F. J. Appl. Phys. Lett. 1984,44, 93.
Acknowledgment. I wish to express may sincere thanks to Dr. Mark Lutwyche for his valuable discussions. Thanks are also due to M. L. and Hilary Lutwyche for correcting my manuscript.