19088
J. Phys. Chem. C 2008, 112, 19088–19096
Adsorption of 3-Mercaptopropyltrimethoxysilane on Silicon Oxide Surfaces and Adsorbate Interaction with Thermally Deposited Gold Jagdeep Singh and James E. Whitten* Department of Chemistry and Center for High-Rate Nanomanufacturing, The UniVersity of Massachusetts Lowell, Lowell, Massachusetts 01854 ReceiVed: August 22, 2008; ReVised Manuscript ReceiVed: September 18, 2008
Adsorption of 3-mercaptopropyltrimethoxysilane (MPS) on hydroxylated silicon oxide substrates by immersion in MPS solution or exposure to MPS vapor has been compared using X-ray photoelectron spectroscopy (XPS). To aid the interpretation, MPS has also been cryogenically condensed in ultrahigh vacuum (UHV) onto gold surfaces. Condensation of MPS vapor on gold in the absence of water does not result in MPS polymerization, as evidenced by multilayer desorption upon warming to room temperature. The C 1s XPS spectrum has been used to infer the relative abundance of methoxy groups. Vapor-deposition on hydroxylated silicon oxide leads to an unpolymerized MPS monolayer consisting of molecules with two methoxy groups. Solution deposition yields a 2-3 layer thick film. The layer in contact with the surface is cross-linked and methoxydepleted, but the layers on top of it contain some methoxy groups. Step-wise thermal deposition of gold in UHV onto solution-deposited MPS has also been investigated. Gold-thiolate bond formation is evidenced by an S 2p XPS peak at a binding energy of 162.6 eV, and C 1s and Au 4f XPS intensities indicate that gold remains on top of the MPS layer. Ultraviolet photoelectron spectroscopy (UPS) has been used to measure the energies of the valence states and the work function. A work function decrease of 0.8 eV is observed by a gold dose of 9.7 × 1014 atoms/cm2, and the work function eventually reaches 4.7 eV for several layers worth of gold. Interaction of gold with MPS layers has also been studied with atomic force microscopy. Force-distance measurements using gold-coated colloidal probes demonstrate ca. four times higher normalized adhesive forces on MPS compared to octadecyltrichlorosilane layers. Introduction 3-Mercaptopropyltrimethoxysilane (MPS) is an interesting molecule because the silane end can bond to metal oxide surfaces1-6 while the thiol end can bond to silver,7 copper,4,8-11 gold,7,12-14 platinum15 and GaAs.16 Self-assembled layers of MPS on SiO2 have potential applications in molecular electronic device fabrication,1,2,17 tribology,12 nanotransfer printing,17 protein immobilization,18 metal electrode adhesion promotion,19 tethering of gold particles to silicon oxide surfaces for sensor applications,20 and inhibition of Cu diffusion at Cu-dielectric interfaces.4,10,21-23 The most common method of preparing silane-covered surfaces is by solution deposition. The quality of the film and extent of multilayer formation depend on the concentration and presence of moisture, but up to 5 mM of silanes (including MPS) has been shown to give continuous layers.3,5,24 Deposition of MPS on SiO2 also has been achieved by leaking MPS vapor into an evacuated chamber containing the substrate.1,2 MPS monolayers on SiO2 may adsorb via Si-O-Si bond formation such that the thiol groups face away from the oxide surface. Figure 1 shows an idealized schematic of the steps believed to occur during MPS monolayer formation on a hydroxylated silicon oxide surface.3 Impinging MPS molecules react with hydroxyl groups and physisorbed water on the surface via the loss of methanol. Condensation reactions of adjacent hydroxyl groups may lead to the formation of a cross-linked network. Depending on the amount of moisture in the solvent, * To whom correspondence should be addressed. Phone: (978) 934-3666. Fax: (978) 934-3013. E-mail:
[email protected].
MPS polymerization can also occur in solution prior to adsorption on the surface. Toward the goal of understanding and optimizing the adsorption of MPS, we report studies using X-ray photoelectron spectroscopy (XPS) of the growth and surface chemistry of MPS layers prepared by solution and vapor deposition on freshly hydroxylated SiO2 surfaces. It is expected that the upward protruding thiol groups should have a high affinity toward gold, since alkanethiols are known to bond to gold via thiolate bond formation. In an attempt to make an MPS “sandwich”, the interaction of gold with the exposed MPS thiol groups has been investigated by depositing gold in ultrahigh vacuum (UHV) and measuring the XPS and ultraviolet photoelectron spectra (UPS) as a function of gold coverage. To aid XPS and UPS interpretation, this paper reports the first experiments on cryogenically condensed MPS. Force-distance measurements using a goldcoated colloidal probe atomic force microscope (AFM) tip have also been performed and indicate bonding of the thiol groups with the gold tip. Experimental Section Materials. MPS (95%) and octadecyltricholorosilane (95%) were purchased from Aldrich and Gelest Inc., respectively, and used without further purification. All solvents were reagent grade and used as received, and water was deionized using a Millipore filtration system. Gold wire (99.9%) of 0.2 mm diameter was sonicated in methanol and acetone before wrapping onto a 0.5 mm diameter tungsten filament for thermal deposition. Pretreatment of Si Wafers. In order to form well-characterized, hydroxylated silicon oxide substrates, the following
10.1021/jp807536z CCC: $40.75 2008 American Chemical Society Published on Web 11/06/2008
Adsorption of 3-Mercaptopropyltrimethoxysilane
J. Phys. Chem. C, Vol. 112, No. 48, 2008 19089
Figure 1. Schematic of idealized MPS monolayer formation (in the presence of a physisorbed water layer) on hydroxylated silicon oxide surfaces.
procedures were followed. Si(111) wafers were cut into 1 × 1 cm2 pieces and ultrasonicated in methanol and acetone to clean and degrease them. Cleaned Si wafers were treated with piranha solution [H2SO4 (98%)/H2O2 (30%); 4:1 (v/v)] at 85 °C for 20 min, rinsed with deionized (DI) water, and then etched in HF (48%)/NH4F (40%); 1:7 (v/v) for 10 min at room temperature. This procedure removed the oxide layer, and the wafers were subsequently rinsed with DI water and dipped in NH4OH (28%)/ H2O2 (30%)/H2O; 1:1:6 (v/v/v) at 85 °C for 20 min. The substrates were subsequently washed with DI water and dipped in a solution of HCl (28%)/H2O2 (30%)/H2O; 1:1:6 (v/v/v) at 85 °C for 20 min, followed by rinsing with DI water. They were finally dried in a stream of N2 gas for several minutes. Formation of MPS Layers. Solution-deposited MPS (SDMPS) layers were prepared by immersing pretreated Si wafers in a 1-2 mM solution of MPS in anhydrous benzene at room temperature. After an immersion time of ∼3 h, the samples were rinsed several times with benzene and methanol to remove physisorbed MPS and dried in a stream of nitrogen gas. Vapordeposited MPS (VD-MPS) monolayers were prepared in the preparation chamber of the VG ESCALAB photoelectron spectrometer. In this case, a quadrupole mass spectrometer was used to check the purity of the MPS prior to dosing. The pretreated Si wafers were exposed to ca. 100 mTorr of MPS for ∼3 h (with the valve between the diffusion pump and the preparation chamber closed during MPS exposure), and then the chamber was evacuated to less than 1 × 10-8 Torr prior to transferring the sample to the photoelectron spectrometer analysis chamber. Condensed MPS. Oxygen free high conductivity (OFHC) copper stubs were polished to a mirror finish with diamond paste, and ca. 1000 Å of gold was thermally deposited in a vacuum of ca. 1 × 10-7 Torr. The gold-coated stubs were introduced to the photoelectron spectrometer preparation chamber, sputter cleaned with 3 keV argon ions, and then cryogenically cooled to 130 K ((10 K) using liquid nitrogen. MPS vapor was admitted using a variable leak valve at a nominal pressure of 5 × 10-7 Torr for 300 s, corresponding to a nominal dose of 150 L. The MPS had been purified upon installation behind the leak valve using several freeze-pump-thaw cycles and heated mildly with warm water until a high vapor pressure impurity (indicated as silane by mass spectrometry) was pumped out. The purity of the MPS was confirmed by mass spectrometry prior to exposing the cold gold surface to the vapors. Mass spectrometry of the MPS showed intense peaks at m/z ) 164, 136, 121, 91, and 59. XPS/UPS and Gold Deposition Measurements. XPS and UPS measurements were performed in a VG ESCALAB MKII photoelectron spectrometer (base pressure of 5 × 10-10 Torr) equipped with a Mg KR X-Ray source (hυ ) 1253.6 eV) and a He I (hυ ) 21.2 eV) ultraviolet lamp. Unless otherwise noted, photoelectrons were detected using a takeoff angle (TOA) of
90°, defined as the angle between the surface plane and the entrance of the focusing lens of the concentric hemispherical analyzer. For some experiments, greater surface sensitivity was achieved by using a 20° TOA. Vacuum compatible conductive silver adhesive was used to “paint” conducting paths from the edges of the silicon wafers to the samples stubs in order to eliminate charging effects. Samples were electrically grounded for XPS and biased at ca. -6.4 V for UPS. Step-wise thermal deposition of gold was carried out in the preparation chamber (base pressure of ca. 8 × 10-10 Torr) of the VG ESCALAB and monitored by a quartz crystal microbalance. Force-Distance Measurements. Glass spheres (20 µm diameter, SPI Supplies) were attached to tipless cantilevers (Micromasch NSC12/tipless/ALBS) using two-component epoxy (DEVCON from ITW Performance Polymers). The epoxy was mixed thoroughly for 5 min, and a very small amount of it was placed on a silicon wafer mounted on a Digital Instruments Nanoscope IIIa AFM sample stage. The cantilever was just brought into contact with the drop of adhesive. Glass spheres were then spread onto a clean part of the silicon wafer, and the edge of the cantilever was centered and brought into contact with an isolated sphere. In this way, only one glass sphere was attached to each cantilever. The cantilever/sphere assembly was dried for 24 h in air and then placed in a gold deposition vacuum chamber with the sphere facing the gold source. Approximately 30 Å of titanium and then 250 Å of gold were deposited on the cantilever at a pressure of ca. 10-7 Torr. Force-distance measurements were performed in contact mode, with the instrument covered with a homemade chamber flushed with dry nitrogen. The relative humidity was kept below 0.5% during all the experiments. The spring constants of the cantilevers were calibrated using the Sader Method.25 To ensure the validity of the force measurements, multiple data sets at different locations were recorded, with each set consisting of ∼100 force curves. The tip velocity during the force measurements was 1.47 µm/s. The cantilevers were characterized using a JEOL FE-SEM, and adhesive force values were determined using “Scanning Probe Image Processor (SPIP)” software marketed by Image Metrology. Results and Discussion Characterization of MPS Films. Contact angle measurements were performed using the sessile drop method to determine the hydrophobic/hydrophilic nature of the MPScovered surfaces. The water contact angles for hydroxylated silicon wafers, VD-MPS, and SD-MPS are 3° ( 1°, 71° ( 1°, and 72° ( 2°. The small contact angle for the hydroxylated silicon wafer confirms that the previously described cleaning/ hydroxylation procedure produces a hydrophilic substrate, and the measured values for the MPS/Si samples are consistent with the previously reported value of 71° for well-ordered MPS layers having thiol groups at the top of the surface.3
19090 J. Phys. Chem. C, Vol. 112, No. 48, 2008
Figure 2. Mg KR XPS of the C 1s region of (a) condensed, (b) vapordeposited, and (c) solution-deposited MPS. Included, as dashed lines, are the peak-fitting results. Photoelectrons are detected at a 90° takeoff angle, and the binding energy is referenced to the spectroscopic Fermi level.
Figure 2 displays C 1s XPS spectra of condensed, VD-MPS and SD-MPS samples. Peak fitting was performed with a Shirley type background and 35% Lorentzian/65% Gaussian components. Two major peaks are identified in the C 1s spectrum of the condensed MPS sample. The peak centered at 285.2 eV corresponds to methylene carbons and carbon bonded to sulfur, while carbon atoms bonded to oxygen appear at 286.6 eV, consistent with oxygens in the methoxy groups withdrawing electron density from their carbon atoms. The electronegativity of sulfur26 is close to that of carbon, and therefore, at least within the context of initial state effects, the carbon atoms bonded to sulfur and the methylene carbons should show up at approximately the same binding energy within the resolution of the experiment. The ratio of the two carbon peaks is ∼1:1, consistent with MPS condensing intact on the gold surface. In the C 1s spectra of VD- and SD-MPS, the peak at 285.2 eV is dominant, and that at 286.6 eV is reduced in intensity compared to the condensed MPS sample. This indicates that methoxy groups have been lost during adsorption, consistent with Figure 1. The ratios of the areas of the 285.2 to 286.6 eV peaks are 3:2 and 4:1 for the VD-MPS and SD-MPS samples, respectively. Because the VD-MPS layer is formed by dosing into vacuum (with significantly less water contamination), the differences in the XPS spectra between the VD- and SD-MPS samples are attributed to minimal water contamination in the case of VDMPS, which leads to less hydrolysis of the methoxy groups (step I in Figure 1). The measured 3:2 ratio of the 285.2 to 286.6 eV carbon atoms indicates that vapor deposition of MPS leads to a monolayer, with each MPS molecule losing one methoxy group as it forms an Si-O-Si bond with the surface. The presence of the 286.6 eV peak in the SD-MPS spectrum signifies that some methoxy groups remain within the film, although only to a minor extent. Figure 3 shows corresponding S 2p spectra. The spectrum of condensed MPS (on a gold substrate) taken at a 90° TOA has a single peak at 164.2 eV, which is ascribed to reduced sulfur (i.e., S-H). VD-MPS and SD-MPS spectra adsorbed on hydroxylated silicon oxide surfaces at 90° are also included; the presence of a dominating peak at 168.0 eV is noted
Singh and Whitten
Figure 3. Mg KR XPS of the S 2p region of (a) condensed MPS (90° TOA) on a gold surface (b) VD-MPS (90° TOA) on silicon oxide (c) SD-MPS (90° TOA) on silicon oxide, (d) VD-MPS (20° TOA) on silicon oxide, (e) SD-MPS (20° TOA) on silicon oxide, and (f) MPS condensed on a gold surface and then warmed to room temperature (90° TOA). The binding energy is referenced to the spectroscopic Fermi level.
compared to the condensed MPS spectrum. The intensity of this peak decreases as the takeoff angle decreases to 20° for both SD-MPS and VD-MPS. Figure 3f shows the S2p spectrum of MPS condensed on gold after warming to room temperature. This will be discussed shortly. There are two possible origins of the 168.0 eV peak in the S2p spectral region. Elemental silicon exhibits a plasmon loss at this energy,27,28 as confirmed by its absence in the XPS spectrum, shown in Figure 4a, of a specially prepared 200 nm thick layer of SiO2 deposited on a silicon surface. This peak is absent in this case because of the limited detection depth (