Anal. Chem. 2006, 78, 1739-1742
Molecular Orientation Analysis of a Single-Monolayer Langmuir-Blodgett Film on a Thin Glass Plate by Infrared Multiple-Angle Incidence Resolution Spectrometry Takeshi Hasegawa,*,†,‡ Yusuke Nakano,† and Yasuyoshi Ishii†
Department of Applied Molecular Chemistry, College of Industrial Technology, Nihon University, 1-2-1 Izumi-cho, Narashino 275-8575, Japan, and PRESTO, Japan Science and Technology Agency, 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan
Molecular orientation analysis in a single monolayer deposited on a glass substrate has been a difficult matter, since the glass substrate absorbs infrared rays so strongly that the measurements of infrared spectra are difficult to perform, and the single monolayer is not suitable for X-ray diffraction analysis because no periodical structure is available. When a thin glass is used as the substrate, in particular, the infrared analysis becomes more difficult, since optical fringes appear strongly on the absorption spectra due to the multiple reflections in the glass. In the present study, infrared multiple-angle incidence resolution spectrometry (MAIRS) has been employed to remove the fringes from the spectra of single- and five-monolayer Langmuir-Blodgett (LB) films of cadmium stearate deposited on a thin glass plate. The MAIRS in-plane spectra gave quantitatively reliable infrared transmission spectra for both films with little fringes, which made it possible for the first time to analyze the molecular orientation in the single-monolayer LB film on glass. As a result, it has been revealed that the molecule in the single-monolayer LB film on thin glass exhibits a significantly larger molecular tilt angle than those prepared on other substrates such as gold and germanium. Glass is one of the most useful materials for fundamental research and advanced technologies, since it is inexpensively obtained and it can chemically and photochemically be processed easily. Microstructures (dots and wells) can be built in the glass surface by the photoresist technique,1 and the surface can be furnished by chemical reactions.2 Therefore, glass is suitable for the substrate of lab-on-a-chip and DNA chip technologies,3,4 for example, in which DNA fragments or synthesized oligonucleotides are linked to the fine structure in the glass surface through covalent bonds. * To whom correspondence should be addressed. Phone: 81 47 474 2543. Fax: 81 47 474 2579. E-mail:
[email protected]. † Nihon University. ‡ PRESTO. (1) Fodor, S. P. A.; Read, J. L.; Pirrung, M. C.; Stryer, L.; Liu, A. T.; Solas, D. Science 1991, 251, 767. (2) Mottola, H. A. Anal. Chim. Acta 1983, 145, 27. 10.1021/ac052217n CCC: $33.50 Published on Web 02/17/2006
© 2006 American Chemical Society
On the other hand, however, glass is also known as a material that is not suitable for infrared spectroscopic analysis, since it absorbs infrared rays too strongly particularly below ∼1500 cm-1. Therefore, transmission measurements of samples on glass are very difficult, and reflection measurements are also highly difficult to perform, because the reflectivity is extremely low. To reduce light absorption by the substrate, it is a good idea to make the substrate thinner. In practice, many glass-based devises are fabricated on a thin glass substrate, and they are often subjected to visible absorption and fluorescent spectrometries. It is still difficult, however, to perform infrared analysis on thin glass samples because of another problem. The problem is that the absorption spectra are disturbed by strong interference fringes arising from the internal multiple reflections in the thin substrate, which is outstanding in the infrared region. In this fashion, the glass issues prevent us from fine infrared analysis of molecular adsorbates on glass, although infrared spectroscopy is a promising analytical tool to investigate the organic molecular assemblies on a solid surface.5,6 Since the X-ray diffraction technique cannot be applied either to the estimation of the thickness of a “single”-monolayer film,7 even the basic matter of molecular orientation (see Chart 1) in a monolayer film of a fatty acid has never been clarified thus far. To simply remove the fringes from the spectra, principal component analysis (PCA)8,9 of a collection of angle-of-incidence dependent spectra may be useful. Unfortunately, however, the fringes change the periodicity with an increase of the angle of incidence. Since the PCA analysis is based on an assumption that the principal components have variances depending on intensity changes only, and no band shift is present,9 the present case is not suitable for a PCA analysis. Another problem is that the (3) Vilkner, T.; Janasek, D.; Manz, A. Anal. Chem. 2004, 76, 3373. (4) Scha¨fer, H.; Chemnitz, S.; Seibel, K.; Koziy, V.; Fischer, A.; Ehrhardt, D.; Bo ¨hm, M. In Nano-Micro Interface: Bridging the Micro and Nano Worlds; Fecht, H.-J., Werner, M., Ed.; Wiley: Chichester, 2004; pp 119-137. (5) Grosse, P. Mikrochim. Acta 1991, 2, 309. (6) Tolstoy, V. P.; Chernyshova, I. V.; Skrychevsky, V. A. Handbook of Infrared Spectroscopy of Ultrathin Films; Wiley: New York, 2003. (7) Fujimori, A.; Sugita, Y.; Nakahara, H.; Ito, E.; Hara, M.; Matsuie, N.; Kanai, K.; Ouchi, Y.; Seki, K. Chem. Phys. Lett. 2004, 387, 345. (8) Hasegawa, T. Anal. Chem. 1999, 71, 3085. (9) Malinowski, E. R. Factor Analysis in Chemistry, 3rd ed.; Wiley: New York, 1991.
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1739
Chart 1. Schematic Image of Orientation Angle, θ, Defined from the Surface Normal (Axis z). for the Uniaxial Orientation Modela
a
The angle, φ, is distributed randomly in the x-y plane.
intensity of the decomposed spectra by PCA (eigenvectors) cannot be discussed, because the eigenvectors are orthogonalized in the PCA calculation, which accompanies the normalization process. With other analytical techniques such as Fourier self-deconvolution,10 the decomposed spectra would have a large inaccuracy in band intensity. Although these techniques are conveniently and practically used for the removal of fringes, it is not suitable for spectra of “thin films”, since the signal intensity is too weak and the analysis often influences the signal intensity. In the present study, the multiple-angle incidence resolution spectroscopy (MAIRS) technique11-13 has been employed to get over the issues. MAIRS does not decompose absorbance spectra, but instead, a collection of “single-beam” spectra are analyzed to yield two spectra, which correspond to the normal incidence inplane (IP) and out-of-plane (OP) single-beam transmission spectra. Here, “plane” means the film plane. This analytical “extraction” (not decomposition) from the original collection of spectra is achieved by use of the classical least-squares (CLS) regression8,14 formulation, on which MAIRS theory is built. Finally, a set of IP and OP single-beam spectra for the sample and background will be converted to IP and OP “absorbance” spectra, which respectively correspond to the conventional transmission and reflectionabsorption (RA) spectra.6 The important benefit of MAIRS is that it yields the RA-like spectrum on a nonmetallic substrate, which cannot be performed by other techniques.11-13 The CLS regression (A ) CK + Ud) has a unique property: A part of A (a collection of absorbance spectra) is fully described by C (physical descriptor) and K (pure component spectra) by linear combination, and the rest in A, which cannot be described by C, is discarded in the “undescribed” matrix that is designated as Ud. In the case of MAIRS, the equation is rewritten as eq 1. Since the interference fringes involved in the collection of singlebeam spectra, S, is not described by the MAIRS R matrix, the fringes are expected to be removed from the single-beam spectra readily, and they would be discarded in the Ud term.11 (10) Kaupinnen, J.; Partanen, J. Fourier Transforms in Spectroscopy; Wiley-VCH: Berlin, 2001. (11) Hasegawa, T. J. Phys. Chem. B 2002, 106, 4112. (12) Hasegawa, T.; Matsumoto, L.; Kitamura, S.; Amino, S.; Katada, S.; Nishijo, J. Anal. Chem. 2002, 74, 6049. (13) Hasegawa, T. Anal. Bioanal. Chem. 2003, 375, 18. (14) Kramer, R. Chemometric Techniques for Quantitative Analysis; Dekker: New York, 1998.
1740
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006
( )
S S ) R SIP + Ud OP
(1)
MAIRS will be, in the present study, employed for the analysis on a strongly absorbing material (glass), which is out of the assumption of the MAIRS theory that the substrate is optically transparent.11 This assumption mainly influences the R matrix. It is also known that sOP greatly responds to the inaccuracy of R. Therefore, the sOP spectrum cannot be used for quantitative discussion on an absorbing substrate. It is fortunate, however, that the IP spectrum is robust and still useful for accurate quantitative chemical analysis.12 In the present study, the IP spectrum will be used for the quantitative estimation of molecular orientation in a singlemonolayer LB film on glass with the novel use of the MAIRS technique. EXPERIMENTAL SECTION Cadmium stearate monolayers were prepared on an aqueous subphase that contains cadmium chloride with a concentration of 3 × 10-4 M by spreading a chloroform solution of stearic acid (molecular mass, 284.48) of 1.0 mg mL-1. The pH of the subphase was fixed at 7.2-7.3 by adding Na2CO3 and NaHCO3 (both 3 × 10-4 M). The water was purified by a Millipore (Molsheim, France) Milli-Q Labo water purifier equipped with a microfilter (Millipak-40), having micropores of 0.22 µm, to remove unexpected organic contaminants. The resistivity of the pure water was 18.3 MΩ‚cm, and the surface tension was 72.8 mN m-1 at 25 °C. The monolayer comprised 100% cadmium salt (acid free), which was confirmed by measuring an IR spectrum of a collapsed film collected on the subphase. After the monolayer was compressed up to 30 mN m-1, it was transferred onto a thin glass plate (20 × 50 × 0.3 mm) by the Langmuir-Blodgett (LB) technique.15 For the deposition of fivemonolayer LB film, this procedure was repeated and the Y-type LB film15 was fabricated. The glass used for the substrate was a Matsunami (Tokyo, Japan) cover glass. The measurements of infrared spectra were performed on a Thermo-Electron Nicolet (Madison, WI) Magna 550 FT-IR spectrometer with a mercury-cadmium-telluride detector cooled by liquid nitrogen. The laser modulation frequency for the interferogram collections was 60 kHz. The interferogram was accumulated 2000 times to improve the signal-to-noise ratio. The unpolarized single-beam transmission spectra were collected at the eight angles from 10° to 45° by 5° steps. The angle of incidence was automatically changed by a step motor controlled by a personal computer through custom-made software embedded in ThermoNicolet OMNIC ver. 6.1, and the measured single-beam spectra were also automatically stored in the computer. The automatic FT-IR MAIRS system including software was developed by Mr. Shin-ichiro Hayashi, Thermo-Electron (Yokohama, Japan). The eight spectra were subjected to the MAIRS algorithm,11 which was calculated on MathWorks (Natick, MA) Matlab 7.0. The contact angle measurements of a surface were performed with a drop of pure water put on the surface by microsyringe, and the lateral picture was taken by a digital camera. The digital (15) MacRitchie, F. Chemistry at Interfaces; Academic Press: San Diego, 1989.
Figure 1. Infrared spectra of five-monolayer LB film of cadmium stearate deposited on the thin glass plate measured by (a) normalincidence transmission spectrometry, and (b) the MAIRS technique (IP spectrum).
Figure 2. Infrared spectra of single-monolayer LB film of cadmium stearate deposited on the thin glass plate measured by (a) normalincidence transmission spectrometry, and (b) the MAIRS technique (IP spectrum).
image was magnified on a computer screen, and the contact angle was measured after drawing a contact line. RESULTS AND DISCUSSION The cadmium stearate five-monolayer LB film deposited on the thin glass slide (0.3 mm) was measured by infrared normalincidence transmission spectrometry. The spectrum is presented in Figure 1a. Interference fringes apparently appear in the spectrum, which disturbs the spectrum of the LB film. Although the peak tops are found at 2918 and 2850 cm-1, the accuracy of the peak positions must be low due to the strong fringes. When the same measurement was performed for the singlemonolayer LB film, the spectrum became atrociously poor as presented in Figure 2a, since the relative intensity of the fringe to the absorption bands became very large. With this spectrum, the peak locations (2915 and 2851 cm-1) are not useful at all, and no discussion can be made. In this manner, measurements of transmission spectra on a thin glass slide always accompany optical problems, which make the measurements in a monolayer level very difficult. Another problem is that the position and intensity of the fringe are very unstable for each measurement, since fringes are originally found in the single-beam spectra, and the conversion process to the absorbance spectra enhances the fringes in various manners. In the present study, the infrared MAIRS technique has been employed to overcome the problem. MAIRS makes it possible to yield the IP absorbance spectrum via a regression calculation, which is equivalent to the normal-incidence transmission spectrum. One of the important benefits of MAIRS is that nonlinear physical responses to the matrix, R, are effectively discarded and only the linear responses are extracted from the collection of
single-beam spectra during the regression calculation. This function is based on an intrinsic characteristic of the CLS regression.8,14 In the present case, the interference fringes are considered to be one of the significant nonlinear responses, because the period and intensity changes with an angle of incidence do not have linear responses with the R matrix. Therefore, the fringes were expected be removed effectively from the spectra by the MAIRS analysis. The infrared MAIRS IP spectrum of the five-monolayer LB film is presented in Figure 1b. As expected, the fringes are mostly removed, and the absorption bands arisen from the LB film appear clearly. The band positions of the antisymmetric and symmetric CH2 stretching vibration (νa(CH2) and νs(CH2)) bands are 2918 and 2850 cm-1. These positions are reasonable for the LB film of metal stearate prepared at high surface pressure (30 mN m-1). In the same manner, the single-monolayer LB film was also subjected to infrared MAIRS analysis. The result is presented in Figure 2b. The minor bands hidden in the deep fringes have readily been recovered, although weak fringes still remain. The remaining fringes are, however, much weaker than the normal transmission spectrum (Figure 2a), and the band locations and their intensities are now clearly available. The locations at 2918 and 2850 cm-1 are exactly the same as those for the five-monolayer LB film (Figure 1b), which suggests that the IP spectrum is useful even though the minor fringes are still present. The intensity of the interference fringes found in Figure 2b attains at ∼5% of the intensity of the νs(CH2) band. When the improved spectrum is used for molecular orientation analysis, the analytical error is expected to be within (2° as shown in a calibration curve later. Of note is that the IP spectrum theoretically corresponds to the normal-incidence transmission spectrum quantitatively.11 In fact, the conventional experiments have proved that the IP spectra are perfectly equivalent to the normal transmission spectra, and the IP spectra are better in terms of noise. This means that the IP spectrum can quantitatively be discussed as if it were measured by use of the normal transmission geometry. The accuracy of the band intensity, however, has carefully been examined in the present study. Fortunately, we have standard data for a five-monolayer cadmium stearate LB film deposited on glass, which was recently studied by Fujimori et al.7 by use of an X-ray diffraction technique. They employed a newly fabricated grazingangle X-ray diffractometer, which enabled them to collect diffraction spectra for both out-of-plane and in-plane crystallographies in thin layers. This technique, regardless, still cannot be employed for a single monolayer to estimate the spacing in the out-of-plane direction, because it has no periodical spacing. According to them, the five-monolayer LB film on glass has an orthorhombic subcell packing, and the bilayer spacing is 5.09 nm from the second and third diffraction peaks clearly available in the spectrum. Sugi et al. reported that the bilayer spacing, d (nm), of a cadmium salt of an n-fatty acid is related to the number of carbons, n, with a constant, s.16 (16) Sugi, M.; Fukui, T.; Iijima, S.; Iriyama, K. Bull. Electrotech. Lab. 1979, 43, 825.
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006
1741
d ) 0.53 + sn
(2)
The typical value of s for trans-zigzag alkyl chains is known to be 0.254 nm, which corresponds to the distance of the alternate carbon atoms.17 In the present case, s is calculated to be 0.253 nm from the bilayer spacing (5.09 nm) and n ) 18. The tilt angle of the molecule, θ, can be estimated by use of these two s values:
θ ) cos-1(0.253/0.254) ) 5°
(3)
This molecular orientation is useful for the evaluation of absorbance in infrared transmission spectra. The tilt angle of 5° (γ) for the alkyl chain suggests the orientation angle of 86.4° (R) for the CH2 symmetric stretching vibration (νs(CH2)) mode because of the direction cosine (eq 4), under an approximation of R ) β (β is for the CH2 antisymmetric stretching vibration (νa(CH2)) mode).
cos2R + cos2β + cos2γ ) 1
(4)
With the orientation angle, R, and the refractive index of the glass (n ) 1.5258) provided by the maker, the absorbance of the band was calculated by a procedure that was proposed by Hasegawa et al.18 The real parts of the uniaxial anisotropic complex refractive indices for the thin layer were no ) 1.48 and ne ) 1.56 as found in a previous paper.18 The subscripts o and e indicate that the parameters are for ordinary and extraordinary rays, respectively. In the present study, in other words, no and ne means the refractive indices in the direction of surface parallel and surface normal, respectively. The anisotropic imaginary parts (ko and ke) were evaluated by use of the ellipsoid
3 ko ) kbulk sin2θ 2
(5)
ke ) 3kbulk cos2θ
(6)
where kbulk is the absorption index of unoriented sample and θ is the orientation angle from the surface normal.18 For the νs(CH2) band, kbulk is known to be 0.2.18 As a result, the absorbance of the νs(CH2) band was calculated to be 0.0135. This calculated value surprisingly agrees with the experimental value in the MAIRS IP spectrum (0.0135; Figure 1b), which means that the X-ray analysis by Fujimori et al. and our infrared analysis are perfectly consistent with each other. In this manner, the accuracy of the IP spectrum has been proved experimentally. (17) Abrahamsson, S.; von Sydow, E. Acta Crystallogr. 54, 7, 591. (18) Hasegawa, T.; Takeda, S.; Kawaguchi, A.; Umemura, J. Langmuir 1995, 11, 1236-1243. (19) Hasegawa, T.; Myrzakozha, D. A.; Imae, T.; Nishijo, J.; Ozaki, Y. J. Phys. Chem. B 1999, 103, 11124. (20) Hasegawa, T.; Hatada, Y.; Nishijo, J.; Umemura, J.; Huo, Q.; Leblanc, R. M. J. Phys. Chem. B 1999, 103, 7505.
1742 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006
Figure 3. Calculated calibration curves of the orientation angle for the bands at 2918 and 2850 cm-1 plotted by the thick and thin lines, respectively.
Now we are ready to employ the IP spectrum of the “singlemonolayer” LB film for the molecular orientation analysis. The νa(CH2) and νs(CH2) bands in the IP spectrum (Figure 2b) have absorbances of 0.0035 and 0.0023, respectively. The absorbance as a function of the orientation angle was calculated with the use of the same optical constants used for the analysis of a fivemonolayer LB film. The “calibration curves” of the orientation angle are plotted in Figure 3. With the curves, the orientation angles for the two modes have readily been obtained as 78° and 68°, respectively. Therefore, the tilt angle of the alkyl chain (γ) has been calculated as 25° with an analytical error of (2°. The tilt angle for the single-monolayer LB film (25°) is significantly larger than that for the five-monolayer LB film (5°) on the same glass slide. It has already been known, however, that the first layer often has a largely different molecular tilt angle from that in other layers.19,20 This is because only the first layer is influenced by the surface property of the substrate, and the rest of the layers are crystallized by the molecular aggregation properties.20 In particular, for the single-monolayer LB film, the orientation would largely be dominated by the substrate surface. It was reported in earlier studies that the tilt angle in the cadmium stearate LB films deposited on gold and germanium were 8° and 18° from the surface normal, respectively.18 The tilt angle is considered to reflect the hydrophilicity of the substrate surface. The contact angles of the gold, germanium, and glass substrates used in the present study were measured to be 50°, 15°, and ∼0°, respectively. Therefore, when we consider the surface properties via the contact angles, the glass substrate would exhibit the largest value for the tilt angle. Then, the tilt angle of 25° seems reasonable on the monolayer on glass. ACKNOWLEDGMENT This work was financially supported by Grant-in-Aid for Scientific Research (B) (16350048) from the Ministry of Education, Science, Sports, Culture, and Technology, Japan, and also by The Futaba Electronics Memorial Foundation, Japan, to whom the authors’ thanks are due. Received for review December 15, 2005. Accepted February 3, 2006. AC052217N
