Polarization Modulation-Infrared Reflection Absorption Spectroscopic

Mar 14, 2006 - William E. Ford , Ffion Abraham , Frank Scholz , Gabriele Nelles ... Gerald Steiner , Valdas Sablinskas , Wolfgang Seidel , Reiner Salz...
0 downloads 0 Views 421KB Size
Anal. Chem. 2006, 78, 2487-2493

Polarization Modulation-Infrared Reflection Absorption Spectroscopic Mapping Gerald Steiner,*,† Valdas Sablinskas,‡ Marco Kitsche,† and Reiner Salzer†

Institute of Analytical Chemistry, Dresden University of Technology, Dresden, Germany, and Department of General Physics and Spectroscopy, Vilnius University, Vilnius, Lithuania

Polarization modulation-infrared reflection absorption spectroscopy (PM-IRRAS) has been coupled with a stepwise mapping of the sample in order to characterize the molecular orientation across the surface. An optical setup has been developed to facilitiate the PM-IRRAS mapping. With PM-IRRAS mapping used in conjunction with a common FT-IR spectrometer, the achievable lateral resolution of ∼45 µm is energy-limited rather than diffractionlimited. PM-IRRAS mapping was used to study the molecular orientation of octadecanephosphonic acid (OPA) molecules attached on a microstructured aluminum oxide/ gold surface. The spectroscopic map reveals that OPA is preferably spotty attached on the aluminum oxide surface. The attached molecules form a highly ordered film. A lower degree of ordering was found for phosphonic acid adsorbed on gold mainly in a tridentate bonding mode. Results demonstrate that PM-IRRAS mapping has a considerable potential for revealing inhomogenities within ultrathin films attached on a surface. In many analytical fields, such as genomics, proteomics, pharmaceutical screening, environmental analysis, and homeland security, there is a great demand for microstructured surfaces that allow analysis of a multitude of analytes.1-5 Microstructured self-assembled monolayers (SAMs) offer an effective method for modifying surfaces, and in particular, SAMs on metal surfaces have been used frequently as linker films for the adsorption of biomolecules on surfaces as the initial step of a biosensor.6-9 Infrared reflection absorption spectroscopy (IRRAS) with polarization modulation (PM) has been successfully applied to character* To whom correspondence should be addressed. E-mail: gerald.steiner@ chemie.tudresden.de. † Dresden University of Technology. ‡ Vilnius University. (1) James, C. D.; Davis R. C.; Kam, L.; Craighead, H. G.; Isaacson, M.; Turner J. N.; Shain, W. Langmuir 1998, 14, 741-744. (2) Daria, V. R.; Rodrigo, P. J.; Gluckstad, J. Biosens. Bioelectron. 2004, 19, 439-1444. (3) Zhang, G. J.; Tanii, T.; Zako, T.; Funatsu, T.; Ohdomari, I. Sens. Actuators, B 2004, 97, 243-248. (4) Urban, G. Sens. Actuators, A 1999, 74, 219-224. (5) Diaz-Quijada, G. A.; Wayner, D. M. Langmuir 2004, 20, 9607-9611. (6) Chaki, N. K.; Vijayamohanan, K. Biosens. Bioelectron. 2002, 17, 1-12. (7) Frey, B. L.; Jordan C. E.; Korngut, S.; Corn, R. M. Anal. Chem. 1995, 67, 4452-4457. (8) Frey, B. L.; Corn, R. M. Anal. Chem. 1996, 68, 3187-3193. (9) Saccani, J.; Castano, S.; Beaurain, F.; Laguerre, M.; Desbat, B. Langmuir 2004, 20, 9190-9197. 10.1021/ac050481a CCC: $33.50 Published on Web 03/14/2006

© 2006 American Chemical Society

ize adsorbed monolayers.10-12 It has been shown that PM-IRRAS is a highly surface sensitive technique that probes molecular films attached to a metal surface in terms of molecular orientation and conformation.13-16 In the recent past, PM-IRRAS became one of the better methods for the nondestructive characterization of Langmuir-Blodgett films and other thin molecular films.17,18 The fast modulation of the polarization state of the incident light and a differential reflectivity measurement provide superior sensitivity for investigations of structural changes in thin films on a metal surface. An in situ PM-IRRAS setup was developed and coupled with electrochemical methods to investigate conformational changes at electrode surfaces.19,20 More recently, PM-IRRAS was also used for the in situ analysis of the formation of SAMs of 11mercaptoundecanoic acid.21 Besides these examples of PM-IRRAS applications, microscopic methods have been used to image SAMs. Scanning tunneling microscopy (STM) and atomic force microscopy (AFM) are powerful tools for this purpose.22-26 However, STM and AFM can image SAMs only in a nanometer scale, but they cannot be used for practical reasons to characterize organic films with extensions of several micrometers. It is of primary importance to detect domains in molecular orientation and defects in adsorption, since the function and reliability of sensor applications are determined by the quality of the microstructured SAM. Thus, methods for a sensitive and nondestructive (10) Elzein, T.; Nasser-Eddine, M.; Delaite, C.; Bistac, S.; Dumas, P. J. Colloid Interface Sci. 2004, 273, 381-387. (11) Duevel, R. V.; Corn, R. M. Anal. Chem. 1992, 64, 337-342. (12) Islam, M. N.; Okano, T.; Kato, T. Langmuir 2002, 18, 10068-10074. (13) Blaudez, D.; Buffeteau, T.; Desbat, B.; Orrit, M.; Turlet, J. M. Thin Solid Films 1992, 210/211, 648-651. (14) Fukushima, H.; Seki, S.; Nishikawa, T.; Takiguchi, H.; Tamada, K.; Abe, K.; Colorado, R., Jr.; Graupe, M.; Shamkova, O. E.; Lee, T. R. J. Phys. Chem. B 2000, 104, 7417-7423. (15) Saccani, J.; Castano, S.; Beaurain, F.; Laguerre, M.; Desbat, B. Langmuir 2004, 20, 9190-9197. (16) Evans, C. R.; Spurlin, T. A.; Frey, B. L. Anal. Chem. 2002, 74, 1157-1164. (17) Buffeteau, T.; Desbat, B.; Turlet, J. M. Appl. Spectrosc. 1991, 45, 380-839. (18) Buffeteau, T.; Desbat, B.; Turlet, J.-M. Microchim. Acta 1988, 11, 23-26. (19) Huerta, F.; Morallon, E.; Perez, J. M.; Vazquez, J. L.; Aldaz, A. J. Electroanal. Chem. 1999, 469, 159-169. (20) Saez, E. I.; Corn, R. M. Electrochim. Acta 1993, 38, 1619-1625. (21) Methivier, C.; Beccard, B.; Pradier, C. M. Langmuir 2003, 19, 8807-8812. (22) Yamada, R.; Uosaki, K. Langmuir 1998, 14, 855-861. (23) Wagner, P.; Hegner, M.; Gu ¨ ntherodt, H. J.; Semenza, G. Langmuir 1995, 11, 3867-3875. (24) Zawisza, I.; Cai, X.; Zamlynny, V.; Burgess, I.; Majewski, J.; Szymanski, G.; Lipkowski, J. Pol. J. Anal. Chem. 2004, 78, 1165-1181. (25) Dwyer, C.; Gay, G.; de Lesegno, B. V.; Weiner, J. Langmuir 2004, 20, 81728182. (26) Lee, D. H.; Kim, D.; Oh, T.; Cho, K. Langmuir 2004, 20, 8124-8130.

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006 2487

analysis of SAMs are needed that supply information about the film in a micrometer scale and at the same time on a molecular level.27-29 Although PM-IRRAS is highly sensitive against structural changes in thin films, it did not spread to broad application for microstructured surface characterization. This can be explained by at least two reasons. First, both the experimental setup and experimental procedure are complex, and second, the technique allows only a “spot” measurement and does not provide images of the surface. To capture images and depict changes in molecular structure across an inhomogeneous or microstructured sample, we have developed a technique that allows PM-IRRAS mapping. Spatially resolved PM-IRRAS provides information about the surface with the same sensitivity as common PM-IRRAS as well as information about the spatial distribution and ordering of molecules. In this report, we describe for the first time the setup for PM-IRRAS mapping. We chose SAMs of octadecanephosphonic acid (OPA) for this study because of its potential in micro- and nanotechnologies. Phosphonic acids are air-stable compounds and form ordered monolayers on metal oxide surfaces.30-32 Due to different bond strengths, SAMs made of phosphonic acid exhibit greater intrinsic dynamics than SAMs made of thiols. OPA has a long alkyl chain and provides a hydrophobic barrier to aqueous solutions. Monolayers of alkanephosphonic acid protect oxide surfaces from etching and have a clear potential in surface modification,33 microstructuring, and nanotechnology34 as well as in biosensing.35 PM-IRRAS MAPPING ACCESSORY The optical setup for PM-IRRAS mapping was constructed as an accessory to a common FT-IR spectrometer. The system is based on PM-IRRAS technique as described elsewhere.17,36 Figure 1 shows the optical layout. The collimated infrared beam from the external port of the FT-IR spectrometer (IFS 88, Bruker Optik GmbH, Ettlingen, Germany) is focused by a spherical gold mirror, linearly polarized by a wire grid polarizer, and modulated by a ZnSe photoelastic modulator (PEM-90 model II/ZS37, Hinds Instruments, Hilsboro, OR). A pinhole is placed in the focal point of the infrared beam. The pinhole shields marginal rays and forms a small “point” source for a 15× Cassegrain objective. The latter focuses the modulated infrared beam on the sample surface. A second 15× Cassegrain objective collects the light reflected off the sample and focuses it on a nitrogen-cooled MCT detector. To build an image, the illumination point must be swept across the sample surface. A precise computerized positioning xyz stage moves the sample across the observation area. PM-IRRA spectra (27) Bietsch, A.; Hegner, M.; Lang, H. P.; Gerber, C. Langmuir 2004, 20, 51195122. (28) Diaz-Quijada, G. A.; Wayner, D. D. M. Langmuir 2004, 20, 9607-9611. (29) Hu, J.; Liu, Y.; Khemtong, C.; El Khoury, J. M.; McAfoos, T. J.; Taschner, I. A. Langmuir 2004, 20, 4933-4938. (30) Folkers, J. P.; Gorman, C. B.; Laibinis, P. E.; Buchholz, S.; Whitesides, G. M.; Nuzzo, R. G. Langmuir 1995, 11, 813-824. (31) Gardner, T. J.; Frisbie, C. D.; Wrighton, M. S. J. Am. Chem. Soc. 1995, 117, 6927-6933. (32) Marcinko, S.; Fadeev, A. Y. Langmuir 2004, 20, 2270-2273. (33) Pahnke, J.; Ru ¨ he, J. Macromol. Rapid Commun. 2004, 25, 1396-1401. (34) Goetting, L. B., Deng, T.; Whitesides, G. M. Langmuir 1999, 15, 11821191. (35) Fontes, G. N.; Malachias, A.; Magalhaes-Paniago, R.; Neves, B. R. A. Langmuir 2003, 19, 3345-3349. (36) Blaudez, D.; Buffetau, T.; Cornut, J. C.; Desbat, B.; Escafre, N.; Pezolet, M.; Turlet, J. M. Appl. Spectrosc. 1993, 47, 869-874.

2488

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

Figure 1. Optical layout of the PM-IRRAS mapping accessory.

are collected by a point-by-point rastering. The step width can range between 5 and 100 µm in both x and y directions. The z direction is used to place the sample exactly in the focal plane of the objectives. All optical components, including the spectrometer, are assembled on a damped optical breadboard (Newport GmbH, Darmstadt, Germany). A fundamental principle in diffraction-limited optical microscopy requires that the spatial resolution of an image is limited by the wavelength of the incident light and by the numerical apertures of the condenser and objective systems. The theoretical achievable spatial resolution can be expressed with the diffraction limit given by37

R ) 1.22(λ/2 NA)

(1)

where R is the smallest distance between distinguishable objects on the surface, NA is the objective numerical aperture, and λ is the wavelength of the light. For the 15× Cassegrain objective (NA ) 0.4) and a wavelength of λ ) 8 µm (1250 cm-1) follows R ≈ 12 µm. In case of nonnormal incidence, R is also dependent on the angle of incidence. The larger the angle the more extended is the elliptic shape of the sample spot. The elliptic axes (sx,sy) of the spot are given by the following equations

sx ) dp/(15 cos(R))

(2)

sy ) dp/15

(3)

where dp is the diameter of the pinhole, R the angle of incidence, and 15 is the magnification of the objective. The diameter of the pinhole and the angle of incidence determine the lateral resolution in x and y directions. The larger the angle, the poorer the lateral resolution in x direction. On the other hand, at high angles of incidence (grazing angle), the intensity of an absorption band gets enhanced.38 Figure 2 shows both the calculated intensity of an absorption band and the lateral resolution in x direction versus the angle of incidence. The intensity of the reflected light is calculated according to the Fresnel formulas37 for an assumed absorption band at 1250 cm-1 with a bandwidth of 20 cm-1, typical of a SAM of 2 nm in thickness on gold. The refractive index of (37) Born, M.; Wolf, E. Principles of Optics; Pergamon Press: Oxford, U.K., 1993. (38) Gu ¨ nzler, H.; Heise, H. M. Infrared Spectroscopy; Wiley VCH: Weinheim, Germany, 2001.

Figure 2. Calculated normalized intensity of an absorption band (solid line) and spatial resolution (dashed line) versus the angle of incidence.

signal and operates at a half-wave retardation. Due to its orientation, the polarizer generates s-polarized light at the maximums of positive retardation, while the p-polarized state occurs at times of maximums of negative retardation. The sampling frequency is 74 kHz corresponding to the doubled frequency of the PEM modulation (37 kHz). This alternating signal is taken from the MCT detector and split by a 1:1 multiplexer into two signals. One branch is directed toward a low-pass filter whose output is proportional to the sum of parallel (IP) and perpendicular (IS) polarization states. The other branch arrives at a lock-in amplifier (SR830, Stanford Research, Stanford, CA) and is demodulated at frequency of 74 kHz. The output signal of the lock-in amplifier corresponds to a detection at 90° of the polarized light and is proportional to (IP - IS)J2(Φ). J2(Φ) is the second-order Bessel function due to the double modulation of the IR light by the interferometer and subsequently by the PEM. The two interferograms ([IP + IS] and [IP + IS]J2Φ0) are Fourier transformed by the spectrometer software, and the ratio of these two spectra is calculated. The measured signal (S) is proportional to17

(IP - IS) J (Φ ) S ≈ 2C (IP + IS) 2 0

Figure 3. Diagram of data collection and processing of PM-IRRAS mapping.

gold at 1250 cm-1 is n ) 12.5 + i54. The intensity of the absorption band is normalized to the band intensity at an angle of incidence of 45°. The spatial resolution is calculated according to eq 2 using a small pinhole of 150 µm in diameter. With such a small pinhole, the theoretical achievable lateral resolution at perpendicular incidence is 10 µm, close to the diffraction limit of 12 µm. The highest absorbance intensity is obtained at an angle of 80°, where the lateral resolution is reduced to 55 µm. At an angle of 65°, the absorbance intensity is 1/3 of the maximum at 80°, and the corresponding lateral resolution is 22 µm. This angle offers a good compromise between sensitivity and spatial resolution. The best lateral resolution of 14 µm is obtained at an angle of 45°. However, the band intensity is 8 times less than at an angle of 80°. To be able to choose between good lateral resolution and high sensitivity, our PM-IRRAS mapping accessory provides three predefined angles of incidence: 45°, 65°, and 80°. Due to the numerical aperture NA ) 0.4 of the Cassegrain objective, the beam hits the sample in a range of 45° ( 23°, 60° ( 23°, and for 80° between 57° and 90°. The electrical layout of the PM-IRRAS accessory is illustrated in Figure 3. All devices are controlled by an in-house-written assembler program implemented as macro routine into the spectrometer software OPUS 3.0. The program controls the stepping motors of the xyz stage, the photoelastic modulator, the recording of interferograms, and the signal demodulation by the lock-in amplifier. The PEM is driven by a 37-kHz high-voltage

(4)

C is a constant accounting for the different amplification of the two signals. After coadding interferograms and their Fourier transformation, the sample is moved to the next position of a predefined grid of the map and the measuring cycle restarts. At the end, all spectra are transferred into one data block and saved for a subsequently image procession. Data processing and evaluation as well as reassembly of spectra into a map are performed by using the Matlab package (Version 6.0, MathWorks Inc. Natric, MA). METHODS AND MATERIALS Surface Preparation. Thin gold films of 50 nm thickness were vapor deposited onto pure silicon wafers (10 × 10 mm) and subsequently covered with an aluminum pattern by deposition through a copper mesh. Copper meshes, usually used in electron microscopy, were either hexagons with a width of 80 µm and web width of 15 µm or squares with a width of 80 µm and web width of 80 µm. The thickness of the aluminum patterned film was 5 nm. The vacuum chamber was evacuated to 8.0 × 10-6 mbar, and deposition temperature was kept at 40°C. All metals were deposited at rates of 0.05-0.08 nm/s. The thickness of the metal layers was controlled in-situ using an oscillating quartz gauge Inficon (Leybold AG, Ko¨ln, Germany). After deposition, synthetic air was admitted to the evaporation chamber in order to oxidize the aluminum. Monolayer Formation. OPA was synthesized as described elsewhere,39 dissolved in ethanol (1 mmol/l) and adsorbed on the aluminum oxide/gold surface under flow condition. A Teflon flow cell and a pump were used to ensure a turbulent flow near the substrate surface. The velocity of the flow was kept at 10 mL/ min for ∼2 h. After completion of the adsorption process, the sample was rinsed in ethanol for 24 h and dried under nitrogen. (39) Maege, I.; Ja¨hne, E.; Henke, A.; Adler H. J.; Bram, C.; Jung, C.; Stratmann, M. Prog. Org. Coat. 1998, 34, 1-12.

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

2489

Figure 4. (A) Scheme of the patterned aluminum oxide/gold surface, (B) PM-IRRAS bright-field map of the pattern. (C) Map of the PM-IRRAS intensity values at 1250 cm-1.

PM-IRASS Spectroscopic Mapping. To achieve a high spatial resolution, all spectroscopic measurements were carried out at an angle of incidence of 45°. A signal-to-noise ratio (SNR) higher than 50 at each pixel required the coaddition of 1000 spectra at a spectral resolution of 8 cm-1. Spectra were acquired in the spectral range of 800-4000 cm-1 applying Happ-Genzel apodization and a zero filling factor of 1, yielding a data spacing of ∼4 cm-1. The PM-IRRA spectra (S in eq 4) were converted to absorbance units and assembled to a map. The photoelastic modulator was set to introduce a maximum dephasing at 1200 cm-1 where most prominent spectral features of OPA appear. RESULTS AND DISCUSSION Lateral Resolution. The spatial resolution of the optical accessory was probed by a pinhole of 150 µm at an angle of incidence of 45°. According to eqs 2 and 3, the spatial resolution should be 14 µm in x direction and 10 µm in y direction. A patterned aluminum oxide film on a gold surface was used as test sample. Figure 4A shows the optical micrograph of the patterned structure. Dashed lines highlight the area that was investigated by PM-IRRAS mapping. The corresponding PM-IRRAS bright-field map is given in Figure 4B. The infrared bright-field map was obtained by integrating the absorbance values of the spectral region from 850 to 1500 cm-1. The image contrast arises from different reflectivities of the gold and aluminum oxide surface. Comparison with the pattern in Figure 4A reveals that webs are clearly discernible, indicating a lateral resolution of ∼15 µm. However, the small pinhole reduces dramatically the intensity of the infrared beam. This is illustrated in Figure 4C. The PM-IRRAS 2490 Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

map at 1250 cm-1 exhibits merely weak contrast between gold and aluminum oxide surfaces. The small pinhole causes a too poor SNR and is therefore not suitable for PM-IRRAS mapping. A reasonable SNR of ∼20 requires a pinhole of 700 µm in diameter. On the other hand, a pinhole of 700 µm reduces the lateral resolution achieved at an angle of incidence of 45° to 45 µm in y direction and 85 µm in x direction. An improvement in lateral resolution requires a more brilliant light source. Chemical Image Analysis. To illustrate the basic evaluation procedure, Figure 5 shows a typical PM-IRRA spectrum of OPA adsorbed on aluminum oxide. The spectrum is dominated by the two broad features of the Bessel function, due to the double modulation of the infrared beam.18 Distinct analyte bands appear between 900 and 1500 cm-1 on the dominating Bessel background. To evaluate the absorption features of the OPA, the Bessel background has to be removed. Most PM-IRRAS experiments utilize a reference spectrum of the pure surface to eliminate the background. However, this approach is not suitable for PM-IRRAS

Figure 5. PM-IRRA spectrum of OPA adsorbed on aluminum oxide surface.

Figure 6. (A) PM-IRRA spectrum of OPA and a fitted quadratic function (dotted line) as baseline in the spectral region of interest. (B) Subtracted PM-IRRA spectrum showing the absorption bands of the OPA.

mapping due to different reflectivities across a patterned surface. In the case of a patterned surface, each sample spectrum has to be corrected by the corresponding spectrum of the neat surface spot. Because short sections of the Bessel function can well be approximated by a quadratic function, we choose this simple way of eliminating the background. The quadratic baseline correction has to be performed separately for each pixel. Figure 6A shows a PM-IRRA spectrum and the fitted quadratic function. The baseline-corrected PM-IRRA spectrum is displayed in Figure 6B, exhibiting the absorption bands of OPA. Symmetric and antisymmetric ν(P-O) stretching vibrations occur at 1092 and 1132 cm-1, respectively.40,41 The band at ∼1270 cm-1 is assigned to the ν(PdO) stretching vibration.42 The broad band near 950 cm-1 arises mainly from the ν(P-OH) stretching vibration.43 Finally, the weak band at 1460 cm-1 is assigned to the δ(C-H) deformation vibration. These bands provide the chemical information they indicate: (i) occurrence and strength of chemical or physical bonding of OPA to the particular surface spot and (ii) orientation of the attached molecules on the surfaces. Whenever a PM-IRRAS band is observed, the dipole moment of a vibration mode must have a component that is perpendicular to the surface. If chemical heterogeneity occurs across the patterned surface, variations in the spectral bands’ positions and intensities should be indicative of it. Figure 7A represents the PM-IRRAS map, calculated by integration the intensity across the spectral range from 850 to 1500 cm-1. As is evident by comparison with the sample structure depicted in Figure 7B, the bright pixels correspond to the

Figure 7. (A) PM-IRRAS bright-field map of OPA adsorbed to the pattern surface. The map is obtained from baseline-corrected spectra in spectral range 850-1500 cm-1. (B) Scheme of the pattern.

aluminum oxide pattern and indicate stronger absorption bands relative to those of the gold surface. This means more molecules are attached to the aluminum oxide surface than to the gold layer. Note that the thickness of the Al2O3 layer is between 5 and 10 nm so the use of grazing incidence reflection techniques is still appropriate. To investigate the type of bonding, principal component analysis (PCA) was used to evaluate the PM-IRRAS maps. PCA is a multivariate method for data analysis to maximize the signals from spectral regions with the most variance. Principal components of a spectroscopic image are formed by loading plots and score maps. The loading plots correspond to wavelengths where the variation is highest and weight the signals in the positive and negative directions. The score map reveals the weight of the loading plot for each pixel of the map. The highest loading plots correspond to wavelengths where the variation is greatest. PCA calculations were performed using the eig function of the Matlab Package. Figure 8 depicts the top five principal components, which cover more than 92% of the total variance of the PMIRRAS map. Score maps (Figure 8A) reveal the lateral distribution of the principal components. For subsequently, image analysis score maps are color coded by red, green, and blue. Significant features of the PCs are displayed in the loading plots (Figure 8B). Loadings are calculated on a pure mathematical basis; in molecular terms, they resemble spectra of particular species. Predominant species of the adsorbed OPA molecules can be deduced from these loading plots. Species predominant with respect to the occurrence of characteristic bands in the loading plots are sketched in Figure 8C. It is generally accepted that OPA can be chemisorbed to metal or oxide surfaces at least in two ways, bidentate or tridentate.44 In the cases of aluminum oxide, water is formed by abstracting hydrogen from the OH groups and oxygen from the oxide surface.50 In the case of the gold surface, (40) Gao, W.; Dickinson, L.; Gronzinger, C.; Morin, F. G.; Reven, L. Langmuir 1996, 12, 6429-6435. (41) Helmy, R.; Fadeev, A. Y. Langmuir 2002, 18, 8924-8928. (42) Frey, B. L.; Hanken, D. G.; Corn, R. M. Langmuir 1993, 9, 1815-1820. (43) O’Brien, J. T.; Zeppenfeld, A. C.; Richmond, G. L.; Page, C. J. Langmuir 1994, 10, 4657-4663. (44) Bram, C.; Jung, C.; Stratmann, M. Fresenius J. Anal. Chem. 1997, 358, 108111. (45) Ramsier, R. D.; Henriksen, P. N.; Gent A. N. Surf. Sci. 1988, 203, 72-88. (46) Gawalt, E. S.; Avaltroni, M. J.; Koch, N.; Schwartz J. Langmuir 2001, 17, 5736-5738. (47) Hanson, E. L.; Schwartz, J.; Nickel, B.; Koch, N.; Danisman, M. F. J. Am. Chem. Soc. 2003, 125, 16074-16080. (48) Bram, C.; Jung, C.; Stratmann, M. Fresenius J. Anal. Chem. 1997, 358, 108111.

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

2491

Figure 8. Principal component analysis of the PM-IRRAS map. (A) score maps, where dotted lines define the pattern of aluminum oxide, (B) loading plots, and (C) type of linkage of the OPA to the surface. The percentage values indicate the amount of variance covered by the particular PC.

OPA can be adsorbed either directly or via preadsorbed water molecules at the gold surface. Regardless of the type of linkage, if absorption bands of OPA appear in the PM-IRRA spectrum, these molecules have to have adopted a certain degree of orientation. The first PC in Figure 8 comprises by far the largest variance across the investigated area. The intensity distribution among the ν(P-O-H), ν(P-O), and ν(PdO) bands points to a bidentate bound molecule. The corresponding score map shows the diagonal direction of the aluminum oxide square track (cf. Figure 7B) by clearly brighter pixels, indicating a preference of OPA for aluminum oxide regions of the surface. The bands are not frequency shifted; the intensity distribution among ν(P-OH), ν(P-O-), and ν(PdO) features indicates bidentate45 bound molecules. As an inspection of the digital values of the red scale (49) Seeboth, A.; Hettrich, W. J. Adhes. Sci. Technol. 1997, 11, 495-505. (50) Bietsch, A.; Hegner, M.; Lang, H. P.; Gerber C. Langmuir 2004, 20, 51195122.

2492

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

revealeds the pixels show different values and OPA molecules do not adopt the tight packing as expected for the oxide surface. OPA molecules are bound in a bidendate manner on both aluminum oxide and gold surface areas because score maps do not go down to ordinate values of zero. Nevertheless, the contrast between aluminum oxide squares and gold spaces between them is clearly discernible; it indicates the distinct preference of OPA for the oxide spots. The loading plot of the second PC is roughly the negative plot of the first PC. We assign the second PC to nonlinearities of the bidentate scores in the first PC. The result that OPA is bound on both types of surfaces is supported by reports that alkylphosphonate monolayers can be adsorbed on oxides as well as on metals and that they are more disordered than other types of SAM forming molecules.39 Gawalt et al. reported a weak assembling of alkanephosphonic acids on titanium oxide, which results in easy removal of SAMs from the surface.46 Heating the adsorbed molecular films yielded dense films and improved their molecular orientation.45,47 It was found that phos-

Figure 9. RGB images of the score maps taken from Figure 8A.

phonic acid molecules even prefer inhomogeneous spots of the aluminum surface to build up a homogeneous monolayer.48 Such inhomogeneous spots are randomly distributed across the surface and lead to slight differences in adsorption and orientation of the OPA molecules. The loading plot of the third PC exhibits too weak bands to permit any acceptable assignment to a predominant species and probably compensates for the effect of an inaccurate baseline correction. The loading plot shape may indicate fluctuations of properties across the aluminum oxide/gold surface or an imperfect baseline correction due to slight shape differences in the Bessel function. In THE case of the fourth PC, the assignment of the features of the loading plot again reveals OPA. The ν(P-O-) vibration modes are strong, whereas the ν(PdO) mode has clearly disappeared. A discussion of the ν(P-OH) feature under the complex envelope is hardly possible. Compared to the loading plot of the first PC, the ν(P-O-) modes are now broadened and shifted to higher frequencies. The disappearance of the ν(PdO) band can be interpreted as evidence for a tridentate bonding mode.39 The distribution of tridentate species across the investigated area roughly follows the gold areas. The loading of the fifth PC is dominated by the ν(P-OH) and δ(CH) bands. Both are slightly shifted toward lower frequencies, which indicates reduced strength of this bond compared to those in the bidentate mode. However, the loading plot also shows a number of bands that cannot be assigned. Also, the score map is much noisier than the score maps of the other PCs. In summary, for all PCs, spectral features across the whole sample area indicate that OPA molecules are mainly bound to aluminum oxide in a bidentate bonding mode, whereas the tridentate bonding mode occurs mainly for gold. To find out whether the individual PCs show a correlation among themselves, the score maps were transferred to RGB images. The separate channels have been coded such that the first PC (bidentate bonding mode) is always red, the fourth PC (tridentate bonding mode) is always blue, and the second, third, (51) Nie, H. Y.; Walzak, M. J.; McIntyre, N. S. Langmuir 2002, 18, 2955-2958.

and fifth PCs are green. Figure 9 shows the RGB images. The image on the left side illustrates the correlation between the first and fourth PC. The green channel is set to zero. Red and blue are the dominant tints whereas deep violet pixels as result of high red and high blue values do not occur. This underlines the different molecular indication of PC1 and PC4. The image reveals that the appearance of the bidentate bonding mode, represented by red pixels, is roughly similar to the aluminum surface, whereas the tridentate bonding mode, represented by blue pixels, appears mainly on the gold surface. The right panel in Figure 9 shows RGB images in which the bidentate and tridentate bonding modes are coded in the red and blue channels, respectively, together with one of the remaining PCs coded in the green channel. In neither of the three cases does the outline of the surface pattern appear in the image. The RGB images show no significant correlation between the bidentate and tridentate bonding modes as indicated by the first and fourth PCs and the chemically and physically features represented by the second, third, and fifth PCs. The observed properties are in accordance with other studies of SAMs, suggesting that monolayer shows domains and inhomogeneities or different orientations within areas in the micrometer and nanometer range.25,49-51 CONCLUSION A new method for chemical imaging using PM-IRRAS mapping is presented that reveals the distribution of structural features within SAM. This study shows that PM-IRRAS mapping combined with multivariate data evaluation provides a unique tool to characterize the structural order/disorder during processes such as formation of SAMs. Spatial resolution of ∼45 µm can be achieved even for noncrystalline samples of extremely low abundance. In the case of OPA, the results indicate oriented attachment both on aluminum oxide and on gold. Although SAMs based on OPA are often described as crystalline, it is apparent from our PM-IRRAS mapping data that in the case of patterned aluminum oxide/gold surface the molecules are not closely packed. Principal component analysis suggests that the molecules are oriented differently on aluminum oxide or gold surface. The score map of the first PC reveals that a bidentate bonding mainly occurs on the aluminum oxide surface. The absence of ν(PdO) and ν(P-OH) bands indicates a tridentate bonding, which occurs mainly on the gold surface. ACKNOWLEDGMENT This research was supported in part of the “Sonderforschungsbereich Reaktive Polymere” (SFB 287) by the Deutsche Forschungsgemeinschaft (DFG). The international bureau of the German Federal Ministry of Education and Research is acknowledged for financial support of cooperation with the University of Vilnius. We thank Dr. Olesya Savchuk for assistance to capture the maps. The authors also thank W. Hartmann and F. Koschine of Bruker Optic GmbH (Leipzig, Germany) for their assistance during the development of the PM-IRRAS mapping system. Received for review March 22, 2005. Accepted December 9, 2005. AC050481A

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

2493