Dipole Orientation Distributions in Langmuir−Blodgett Films by Planar

Hydration Effects on Membrane Structure Probed by Single Molecule Orientations. Heath A. Huckabay and Robert C. Dunn. Langmuir 2011 27 (6), 2658-2666...
0 downloads 0 Views 455KB Size
Download PDF
J. Phys. Chem. 1996, 100, 775-784

775

Dipole Orientation Distributions in Langmuir-Blodgett Films by Planar Waveguide Linear Dichroism and Fluorescence Anisotropy Paul L. Edmiston, John E. Lee, Laurie L. Wood, and S. Scott Saavedra* Department of Chemistry, UniVersity of Arizona, Tucson, Arizona 85721 ReceiVed: July 20, 1995; In Final Form: October 9, 1995X

A method to determine the orientation distribution of fluorescent molecules in a thin, substrate-supported film is described. Attenuated total reflection spectrometry on the surface of a planar waveguide is used to measure absorption LD, from which the mean dipole tilt angle in the film is obtained. Steady-state fluorescence anisotropy is measured in a total internal reflection geometry on a film supported on fused silica but prepared under otherwise identical conditions. The angular distribution about the mean can be recovered from the anisotropy measurement by modeling the distribution as a probability density that is specified by two adjustable parameters. The method was tested on Langmuir-Blodgett (LB) films of arachidic acid doped with the fluorescent amphiphiles DiI and BODIPY. In the DiI-doped films, the mean tilt angle was 75° from the surface normal. Assuming a Gaussian distribution, the standard deviation was 12°, indicating a high degree of macroscopic order. In the BODIPY-doped films the distribution was 59 ( 17°, which indicates a less ordered assembly. Larger angular distributions were calculated using a step function model. The results show that dipoles in the headgroup region of arachidic acid LB films are more ordered than dipoles in the alkyl chain region. The method should prove useful in studying relationships between assembly technique, structure, and function in two-dimensional molecular arrays.

Introduction Thin film assemblies of organic molecules supported on a solid substrate are currently being studied in many research laboratories.1-3 The high level of interest stems from potential applications in molecular device technologies such as chemical sensing and molecular electronics, in which the film functions as the active element. The working assumption underlying these efforts is that the functional properties of a film will be specified by its macroscopic architecture. One area of emphasis has therefore been the development of general methodologies to assemble molecules into macroscopically oriented, well-ordered arrays. Approaches based on the Langmuir-Blodgett and selfassembly methods have received the most attention to date. Continued progress in fabricating mono- and multilayer films with novel functional properties is dependent on concurrent development of analytical techniques appropriate for characterizing properties such as packing geometry and molecular orientation.1 For crystalline films, X-ray and electron diffraction methods are very powerful approaches.4 Atomic force microscopy has also become popular.5,6 However, to characterize order in less ordered (i.e., liquid crystalline) films, alternate methods must be employed. Optical techniques have become the predominant strategy for studying uniaxial (azimuthally symmetric) assemblies. Molecular orientation in organic films has been assessed by measuring absorbance linear dichroism (LD) in the infrared and UV-vis spectral regions, polarized luminescence, polarized Raman scattering, and second harmonic generation.6-21 Each of these methods has its advantages and limitations, but sensitivity to a particular sample is usually the primary consideration in choice of method. The major difficulty is that by definition the sample comprises a small amount of material and is located at the interface between two immiscible phases. A case in point is measuring electronic absorption LD. In a single-pass transmission geometry, it is difficult to detect a * Corresponding author. X Abstract published in AdVance ACS Abstracts, December 15, 1995.

0022-3654/96/20100-0775$12.00/0

statistically valid difference between absorbances recorded in two orthogonal polarizations for submonolayer to monolayer thick films, in which the chromophore surface coverage may be on the order of 10-11 mol/cm2 or less.11,22 However, the sensitivity problem can be overcome by performing the measurement in an attenuated total reflection (ATR) geometry, using a planar integrated optical waveguide (IOW) as the internal reflection element.23 With a thickness on the order of the wavelength of the guided light, a planar IOW supports up to several thousand reflections per centimeter of beam propagation (ray optics approximation). This produces an evanescent path length that is orders of magnitude greater than that of a millimeter thick internal reflection element, with a concomitant sensitivity enhancement.24-26 Using the planar IOW-ATR geometry to measure the linear dichroic ratio along with established interfacial optical theory, the mean tilt angle of the absorption dipoles in an IOW-supported molecular assembly can be determined relative to a laboratory-defined coordinate system. This approach has been used to study molecular orientation in organic monolayers and hydrated protein films.9,12,25,27 Although the IOW-ATR approach does circumvent the inherent insensitivity of electronic absorption LD in a transmission geometry, it can only yield the mean tilt angle of the chromophores in a molecular assembly. No information on the distribution of tilt angles about the mean is available from this experiment, since only one order parameter can be obtained from a one-photon event.28 Knowledge of a nonrandom ensemble average for a molecular assembly, while certainly useful information, provides an incomplete picture of the film’s macroscopic order. Unlike absorption, a two-photon event such as fluorescence emission is sensitive to two order parameters and therefore can provide information on both the mean and the distribution of tilt angles.15,28 However, an orientation distribution cannot be recovered from a measurement of emission anisotropy alone, since the number of variables needed to describe the distribution (at least two) exceeds the number © 1996 American Chemical Society

776 J. Phys. Chem., Vol. 100, No. 2, 1996 of measured parameters (one). The same is true for all of the optical techniques listed above, when any of them is employed individually to measure a spectrometric response in only one dimension (e.g., intensity as a function of polarization). A solution to this problem is to measure at least two independent parameters. For instance, Wirth and co-workers16 have used frequency-domain fluorescence spectroscopy to measure anisotropy decays for acridine orange adsorbed at solid-liquid and liquid-liquid interfaces. Orientation distributions were recovered by fitting the initial and final anisotropies to a Gaussian model for the dipole distribution. LeGrange et al.18 reported an orientation distribution for NBD-dihexyldecylamine doped in Langmuir-Blodgett (LB) films of stearic acid. Two order parameters were obtained by measuring steadystate fluorescence anisotropy along two independent axes. Bos and Kleijn19 measured steady-state emission intensity and polarization as a function of excitation polarization for a fluorescent porphyrin adsorbed to glass. This approach also yielded two order parameters, from which an orientation distribution was calculated using a maximum entropy method. An alternate strategy with potentially greater sensitivity to low surface coverage films is described in this paper. Planar IOW-ATR is used to measure absorption LD on a waveguidesupported molecular assembly, from which the mean dipole tilt angle is obtained. Steady-state emission anisotropy is measured using total internal reflection fluorescence (TIRF) spectroscopy on a fused silica supported assembly prepared under otherwise identical conditions. The angular distribution about the mean can be recovered from the anisotropy measurement by modeling the distribution as a probability density function that is specified by two adjustable parameters (e.g., a Gaussian distribution described by its mean and standard deviation). This novel combination, termed IOW-ATR+TIRF, was tested on model molecular assemblies consisting of LB films of arachidic acid doped with fluorescent amphiphiles. Fatty acid LB films have been extensively studied2,3 and are known to be well ordered on a macroscopic scale, which makes them an appropriate candidate to test the IOW-ATR+TIRF approach. Theory Theoretical descriptions of TIRF spectroscopy and IOWATR spectrometry are available24,29,30 and will not be dealt with here. However, a statement of TIRF anisotropy and its relationship to macroscopic molecular orientation is necessary because the geometry of the TIR experiment differs from the conventional solution anisotropy measurement. This statement summarizes a more detailed description provided by Wirth and Burbage.16 The relationship between the IOW-ATR and TIRF anisotropy experiments is also outlined here. The laboratory coordinate system is depicted in Figure 1 where the surface of the dielectric substrate, upon which a molecular film is supported, is the x-y plane. Light propagates along the x-axis and is totally internally reflected at the interface between the substrate and the superstrate medium. The evanescent field extending into the superstrate is partially absorbed by chromophores in the film, which subsequently emit fluorescence. The evanescent field polarization is either transverse electric (TE, along the y-axis) or transverse magnetic (TM, primarily along the z-axis with a small x-axis component). Emission is measured along the z-axis, perpendicular to the x-y (film) plane, through a polarizer oriented along either the x-axis or y-axis. Emission intensity Iij that results from excitation polarized along the i-axis and measurement along the j-axis depends on the orientation of the absorption (µa) and emission (µe) transition

Edmiston et al.

Figure 1. Schematic of a dipole µ oriented in the laboratory coordinate system defined by the x-, y-, and z-axes, with the origin located at the interface where the light beam is totally reflected. The dipole orientation is defined by the polar angle θ and the azimuthal angle φ.

dipoles of each chromophore. The time-dependent anisotropy decay is defined as

r(t) )

Izy - (Iyy + Iyx)/2 Izy + Iyy + Iyx

(1)

Assuming that (i) the absorption and emission dipoles are collinear, (ii) fluorescence is collected by a low numerical aperture objective so that emission intensity is proportional to the projection of µ on the x-y detection plane, and (iii) the detection system exhibits zero polarization bias, then the measured intensities can be related to molecular orientation by the following equations.

Iyy ) Ey2F(t)〈cos2 φa cos2 φe〉〈sin2 θa sin2 θe〉

(2)

Iyx ) Ey2F(t)〈cos2 φa sin2 φe〉〈sin2 θa sin2 θe〉

(3)

Izy ) Ez2F(t)〈(cos2 φa + sin2 θa) cos2 φe〉〈cos2 θa sin2 θe〉 (4) In these expressions F(t) represents the time-dependent decay of fluorescence intensity, φa and θa are the azimuthal and polar tilt angles of the transition dipole when light is absorbed, φe and θe are the tilt angles at the time when fluorescence is emitted, Ey2 and Ez2 are the squared electric field amplitudes of the evanescent wave along the y- and z-axes, and 〈 〉 denotes an ensemble average. These expressions are time-dependent to account for the occurrence of molecular motion during the excited state lifetime. The φ correlation functions in eqs 2-4 can be numerically calculated by integration

〈f(φ)〉 ) ∫0 f(φ) N(φ) dφ/∫0 N(φ) dφ 2π

2π

(5)

where f(φ) refers to the 〈 〉-bracketed terms in eqs 2-4 that contain φ, and N(φ) is a function that describes the distribution of dipoles in the film with respect to φ. To perform this calculation, a functional form for N(φ) must be known or assumed. In the case of the LB films examined in this study, the distribution about φ is isotropic (see below) so that N(φ) ) 1. Equations 2-4 can therefore be simplified to exclude a dependence on φ.

Dipole Orientation Distributions in LB Films

J. Phys. Chem., Vol. 100, No. 2, 1996 777

Evaluation of eqs 2-4 and substitution into eq 1 gives the following expression for anisotropy, which is only a function of θ.

r)

1.5〈cos2 θa sin2 θe〉 〈sin2 θe〉

- 0.5

(6)

As before, the θ correlation functions in eq 6 can be numerically calculated by integration

〈f(φ)〉 ) ∫0 f(θ) N(θ) sin θ dθ/∫0 N(θ) sin θ dθ (7) π/2

π/2

where f(θ) refers to the 〈 〉-bracketed terms in eq 6 that contain θ, and N(θ) is the function that describes the dipole distribution in the film with respect to θ. If the dipoles are assumed to be immobile during the excitedstate lifetime, θa ) θe ) θ. An alternative model is that the dipoles are free to “wobble” within a cone defined by a polar semiangle.31 Since N(θ) is not necessarily isotropic, a function that describes its form is assumed. The simplest form is an ensemble of dipoles that are oriented within an angular range of (dθ about a mean tilt angle θµ. Within (dθ, the dipoles can be either immobile or freely wobbling within a cone (or a combination of both); steady-state anisotropy measurements cannot distinguish between these cases. One possible model for N(θ) is a Gaussian function with an angular distribution (standard deviation) of θσ.

N(θ) ) exp[-(θ - θµ)2/2θσ2]

(8)

Wirth and co-workers used a Gaussian model to describe the distribution of acridine orange molecules adsorbed at solidliquid and liquid-liquid interfaces.16 An alternative model used by Thompson and co-workers15,31 is the step function. In this case the dipoles are assumed to be distributed equally in a range of (∆θ about θµ. The function is

N(θ) )

{

1, θµ-∆θ e θ e θµ + ∆θ 0, all other angles

(9)

Use of either model and integrating the correlation function, eq 6 becomes

1.5∫0 N(θ) cos θ sin θ sin θ dθ π/2

r)

2

centimeters, d2 is the IOW thickness, ∆21 and ∆23 are the GoosHa¨nchen shifts at the IOW-superstrate and IOW-substrate interfaces, respectively, and θi is the angle of total reflection for the particular mode. Expressions for ∆21 and ∆23 for a stepindex, asymmetric planar IOW are given in ref 24. Coating the IOW with a thin film of chromophores causes attenuated total reflection of the propagating mode. The absorbance due to the film is proportional to the propagation loss coefficient of the film, Rf.26,27 The linear dichroic ratio of the film is determined by measuring Rf values for a pair of TM and TE polarized modes. The relationship between the mean dipole tilt angle θµ and the linear dichroic ratio F is given by9,12,27

RTE F) ) RTM N

NTEEy2 2

TM[Ex

2

∫0π/2N(θ) sin2 θ sin θ dθ

Figure 2. Structures of the fluorescent amphiphiles DiI and BODIPY.

- 0.5

(10)

Since the anisotropy is a function of two variables, θµ and ∆θ (for the step function) or θσ (for the Gaussian), a steady-state anisotropy measurement cannot be used alone to determine the dipole orientation distribution. However, θµ can be determined independently from an IOW-ATR linear dichroism measurement, which makes eq 10 sensitive to only the angular distribution (∆θ or θσ). The theory for the IOW-ATR LD experiment has been summarized previously12,24,27 so only the principal points are mentioned here. Light-guiding behavior in an IOW can be modeled as a process of repeated total reflection at the boundaries between the waveguiding layer and the adjacent, lower index media. Using the ray optics approximation,24 the reflection density per centimeter of beam propagation along the IOW-superstrate interface is given by

N/D ) (2d2 tan θi + ∆21 + ∆23)-1

(11)

where N is the number of reflections, D is the distance in

+ 2Ez2 cot2 θµ]

(12)

where N is the number of reflections over which the ATR measurement is made, Ex2, Ey2, and Ez2 are the squared electric field amplitudes of the evanescent wave along the x-, y-, and z-axes,30 and the subscripts TE and TM refer to the pair of modes under comparison. Experimental Section Materials and Spreading Solutions. Arachidic acid (99%, #A3631) and chloroform (99.9%, #27,063-6) were obtained from Sigma and Aldrich, respectively. The fluorescent amphiphiles 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI, #D-282) and 2-(4,4-difluoro-5-methyl-4-bora3a,4a-diaza-s-indacene-3-dodecanoyl)-1-hexadecanoyl-sn-glycero-3-phosphocholine (BODIPY, #D-3792) were purchased from Molecular Probes. Their structures are shown in Figure 2. All other chemicals were reagent grade and obtained from commercial sources. All materials were used as received. Planar Waveguide and Fused Silica Substrates. Siliconoxynitride planar IOWs were fabricated by plasma-enhanced chemical vapor deposition on fused silica substrates (2.5 cm × 7.5 cm × 1 mm slides) at the Microelectronics Center of North

778 J. Phys. Chem., Vol. 100, No. 2, 1996 Carolina, as described previously.32 Waveguide thickness and index were measured using the prism coupling technique.33 Fused silica substrates were obtained from Dynasil (Berlin, NJ). Prior to deposition of LB films, planar waveguide and fused silica substrates were cleaned by mechanical scrubbing using a cotton pad, first in ethanol and then in 2% PCC-54 surfactant solution (Pierce). Substrates were then sonicated in 2% PCC54 for 30 min, followed by copious rinsing with deionized (Type I Reagent Grade) water. Cleaned substrates were stored in deionized water and blown dry with nitrogen immediately prior to LB film deposition. Spreading Solutions and Langmuir-Blodgett Film Deposition. LB films were prepared using a NIMA Technology Model 611 trough. The films were composed of five layers: three layers of pure cadmium arachidate (CdA) followed by two layers of arachidic acid (AA) doped with either DiI or BODIPY. The undoped CdA layers were used as a “foundation” to dampen substrate surface irregularities, which could produce an artificially broadened orientation distribution. Arachidic acid spreading solution was prepared by dissolving AA in chloroform at a concentration of 1 mg/mL. Stock solutions of DiI and BODIPY were also prepared in chloroform. Fluor-doped spreading solutions were prepared by diluting the DiI or BODIPY stock solutions with the AA spreading solution. The BODIPY:AA molar ratio was 1:175 in LB films prepared for TIRF measurements and 1:90 in LB films prepared for IOW-ATR measurements; the DiI:AA molar ratio was 1:150 in all films. Low molar ratios were used to prevent energy transfer between fluorophores, which would result in depolarized emission and invalidate the use of steady-state anisotropy measurements to determine molecular orientation.28 For comparison purposes, five-layer films that lacked a fluorescent dopant were also prepared. Before use the trough was cleaned thoroughly with isopropyl alcohol and rinsed with deionized water. Cadmium subphase solution was prepared by dissolving CdCl2 in deionized water at a concentration of 5 × 10-4 M. The subphase pH was adjusted to 6.5-7.0 by the addition of a few milligrams of NaHCO3 per liter. Two substrates were clamped together, leaving one side of each exposed, and were immersed in the subphase. Approximately 100 µL of AA spreading solution was deposited on the subphase surface. After a 5 min wait to allow the chloroform to evaporate, the film was compressed at a rate of 100 cm2/min to a pressure of 31 mN/m. LB deposition on the exposed side of each substrate was performed by dipping/ withdrawing the substrates at 4 mm/min. Transfer ratios were near unity except for the first layer, which typically had a higher value of 1.5. After three layers of CdA were transferred, the substrates were removed and the subphase was replaced with pure Type I Reagent Grade water. Approximately 100 µL of DiI- or BODIPY-doped spreading solution was then deposited on the subphase surface, allowed to equilibrate, and compressed to a pressure of 35 mN/m, as described above. Two layers of fluor-doped AA were then deposited over the CdA layers at a dipping rate of 4 mm/min. Transfer ratios for the fluor/AA layers were between 0.9 and 1. On planar waveguide substrates, the fluor/AA layers were deposited only over half of the total CdA film area as depicted in Figure 3. The other half was not coated with fluor to enable the blank propagation loss of the CdA-coated waveguide structure to be measured (see below). The areas coated with the three and five layer LB films were each approximately 3.0 × 2.5 cm2. The subphase was maintained at a constant temperature of 24 °C during all phases of LB film preparation.

Edmiston et al.

Figure 3. Geometries of the IOW-ATR (A) and TIRF (B) measurements on the dye-doped LB films.

Planar IOW-ATR and TIRF measurements were made within 24 h of deposition. Fluorescence Measurements. LB films deposited on fused silica slides were used for fluorescence measurements. The slide was mounted on a Nikon Diaphot inverted microscope (described in ref 34) with the LB film facing the objective. Epifluorescence was performed using commercially available filter blocks and a Hg lamp apparatus (Nikon). For TIRF measurements, the single line output (514.5 nm for DiI excitation, 488 nm for BODIPY excitation) from an argon ion laser (Ion Laser Technology 5500) was focused with a 90 mm focal length lens into a fused silica trapezoidal prism (Harrick Scientific) that was coupled to the upper (bare) surface of the slide with an index matching fluid. All measurements were performed at a laser power between 0.5 and 1 mW. The beam was totally reflected at the lower (LB film-coated) surface of the slide at an incidence angle of 70°. Excitation polarization was selected using a half-wave Fresnel rhomb. Fluorescence emission was collected with a 4X objective, normal to the LB film plane. A monochromator/photomultiplier assembly, aligned to the side camera port of the microscope and operated in a photon counting mode, was used to measure emission intensity at 568 and 518 nm for DiI and BODIPY, respectively. A sheet polarizer was mounted between the monochromator and the camera port to select the polarization (x or y) of the detected emission. Three LB films were prepared and analyzed for each fluor (six films total). Measurements of polarized emission intensity were made at 5-7 different spots on each film. The spots were randomly chosen and widely spaced over an area of at least 3.75 cm2. At each spot, the mean of 20 intensity measurements, each integrated over 1 s, was acquired twice for each of three polarization geometries, ITEy, ITEx, and ITMy, where the upper and lower case subscripts refer to the excitation polarization and the emission polarizer orientation, respectively.

Dipole Orientation Distributions in LB Films

J. Phys. Chem., Vol. 100, No. 2, 1996 779

To negate systematic errors due to photobleaching, measurements were performed in the order ITEy, ITEx, ITMy, ITMy, ITEx, ITEy. TIRF Data Correction. The measured quantities ITEy, ITEx, and ITMy are not equal to the Iyy, Iyx, and Izy values used in eq 1. The raw data were biased due to (1) the difference in the sensitivity of the detection system to emission polarized along the x- and y-axes (primarily due to the monochromator), (2) the difference in laser excitation powers between TM and TE polarizations, (3) the nonzero x-axis component of TM polarized excitation, and (4) the difference in electric field amplitudes between TM and TE polarizations. Four correction steps were applied to convert the measured quantities to Iyy, Iyx, and Izy. First, the detection system polarization bias was corrected by measuring Ie,y/Ie,x, the ratio of emission intensities detected along x or y from an isotropically emitting sample. The sample was a solution of DiI in chloroform excited in an epifluorescence geometry with randomly polarized light. Multiplying the raw ITEx values by the ratio Ie,y/Ie,x corrected them relative to ITEy and ITMy. Second, the relative excitation powers in TE and TM polarizations were measured using a power meter in place of the sample on the microscope stage. The excitation power ratio PTE/PTM was used to correct ITMy relative to ITEy and ITEx. Third, since the TM polarized evanescent field has a nonzero x component at a reflection angle of 70°,30 the emission intensity that results from x-axis excitation had to be subtracted from ITMy. Since the LB films were isotropic in the x-y plane (see below), excitation along the y-axis, measured along the x-axis, is equivalent to excitation along the x-axis, measured along the y-axis, when normalized to the respective squared electric field amplitudes in x and y. Subtracting the quantity Ixy, given by

Ixy ) (Iyx)(Ex2/Ey2)

(13)

from the product (ITMy)(PTE/PTM) yielded the fraction of ITMy due solely to z-axis excitation. Fourth, the emission intensity values were normalized by dividing by the respective squared electric field amplitudes along the y- and z-axes. The four correction steps are given by the following equations

Iyy ) ITEy/Ey2 Iyx )

Izy )

ITEx(Ie,y/Ie,x) Ey2

[ITMy(PTE/PTM)] - [Iyx(Ex2/Ey2)] Ez2

(14a) (14b)

(14c)

The electric field amplitudes were calculated using the twophase (internal reflection element-air) approximation.29,30 The Iyy, Iyx, and Izy values obtained from eqs 14a-14c were used to compute anisotropies according to eq 1. IOW-ATR Measurements. The instrumental arrangement for measuring absorption linear dichroism in LB films coated on planar waveguides was similar to previous descriptions.26,35 The 514.5 nm line from a Coherent Innova 70 argon ion laser was used for measurements on BODIPY-doped films. The 550 nm output from a Coherent 599 dye laser, pumped with the argon laser and using Coumarin 540 as the gain medium, was used for DiI measurements. Input polarization was selected with a half-wave Fresnel rhomb. The waveguide was mounted on a rotary stage, and the beam was coupled into a guided mode

using a SF6 prism (Karl Lambrecht). The highest order mode that was supported in both TE and TM polarizations was used (either m ) 1 or m ) 2). The guided mode “streak” visible in the waveguide was photographed with a charge-coupled device (CCD) camera and a 50 mm camera lens, oriented normal to the waveguide plane. The CCD system consists of a Tektronix TK512CB chip mounted in a thermoelectrically cooled housing (Princeton Instruments) controlled by a Macintosh Power PC 7100 with IPLab software (Signal Analytics). A band-pass filter that transmitted the laser line and blocked fluorescence emission was mounted between the lens and the waveguide. In some cases, waveguide-excited fluorescence emission was photographed by inserting a band-pass filter that blocked the laser line and transmitted the fluorescence. In each of the waveguidesupported LB films examined, the guided mode streak was photographed at 3-4 different physical locations by vertically translating the waveguide/prism assembly with respect to the stationary laser beam. At each location, at least three photographs were recorded in each polarization (TE and TM) for subsequent averaging. Attenuation curves were generated by plotting the logarithm of the vertically averaged pixel intensity in the image of the streak against horizontal propagation distance. There were two visually distinct regions in each curve, which corresponded to the two distinct regions of the LB film: (1) the area coated with three layers of CdA, where the attenuation is due solely to the intrinsic propagation loss of the planar waveguide/CdA film structure, and (2) the area coated with three layers of CdA and two layers of fluor-doped AA, where the attenuation is due to the intrinsic propagation loss and the absorbance of light by the dopant. The two regions of each attenuation curve were fit by least-squares regression to log[I(x)] ) Rx + C, where I(x) is the average pixel intensity as a function of distance D, x is the propagation distance in cm, R is the loss coefficient in cm-1, and C is a constant. By subtracting intrinsic loss measured in the undoped areas from loss in the doped areas, loss coefficients due only to absorbance by DiI or BODIPY molecules were obtained for both TE and TM polarizations. Orientation Distribution Calculations. Since the loss coefficients measured by IOW-ATR are proportional to the absorbance of the chromophore,26,27 the dichroic ratio is equal to the TE/TM ratio of loss coefficients. Dichroic ratios were used in eq 12 to determine mean tilt angles (θµ) for DiI and BODIPY dipoles in the LB films. The number of reflections per centimeter (NTE and NTM in eq 12) was calculated from eq 11, as described previously.24 Electric field amplitudes were calculated using the two-phase approximation.29,30 Calculated θµ values were then substituted into the appropriate expression for N(θ) (eqs 8 and 9). The N(θ) function was in turn substituted into eq 10 along with the respective anisotropy values computed from eq 1. Angular distributions (∆θ or θσ) were recovered by iteratively solving eq 10 using Mathematica software (Wolfram Research). Infrared Spectroscopy and Ellipsometry. Fourier transform infrared (FT-IR) spectra of LB films were measured in an attenuated total reflectance geometry using a Nicolet 510P FTIR spectrophotometer fitted with a Spectra-Tech ATR accessory. Undoped films were deposited as described above on a silicon ATR crystal that had been pretreated for 30 min in an argon plasma cleaner (Harrick PDC-3XG) to generate an SiO2 surface. Spectra consisting of 200 scans were acquired from 400 to 4000 cm-1 at a resolution of 4 cm-1. The uncoated, oxidized crystal was used as the reference. For ellipsometry measurements, LB films were deposited on single crystal Si wafers (5 cm diameter, 0.43 mm thick, 〈111〉 orientation with p-type resistivity of 5-15

780 J. Phys. Chem., Vol. 100, No. 2, 1996 ohm/cm, polished on one side). The wafers were precleaned by sonication in chloroform for 5 min. They were then treated in the plasma cleaner for 30 min, soaked in 0.1 M KOH for 2-4 min, soaked in 0.1 M HNO3 for 5-7 min, rinsed in deionized water, and finally dried under a nitrogen stream just prior to use. A Gaertner Scientific Model L116C ellipsometer was used to measure the amplitude ratio (ψ) and phase difference (∆) of 633 nm light reflected at 70°. Measurements were made at several different spots on each film, and thicknesses were determined assuming a film refractive index of 1.52.36 Results and Discussion The primary goal of this work was to develop a method of measuring orientation distributions in thin organic films, based on a combination of IOW-ATR linear dichroism and TIRF anisotropy measurements. LB multilayers with incorporated dyes were selected as model molecular assemblies to assess the utility of the IOW-ATR+TIRF method. The LB technique is a well-established means of depositing a thin film of an organic amphiphile, spread onto and compressed into a Langmuir monolayer at an air-water interface, onto a planar substrate.2,3 LB films were deposited on planar waveguides for IOW-ATR measurements, on fused silica substrates for TIRF and epifluorescence measurements, on silicon ATR crystals for infrared spectroscopy, and on Si wafers for ellipsometry measurements. LB Film Characterization. The macroscopic quality of the LB films was qualitatively analyzed using epifluorescence microscopy. The goal was to generate films that were devoid of structural features. It was observed that incorporating fluors into CdA layers produced nonuniform films with large bright and dim domains (>50 µm), which may have been due to aggregation of probe molecules. The use of arachidic acid as the predominant amphiphile largely eliminated this problem, meaning that the AA films doped with less than 1% fluor on a molar basis were macroscopically uniform. Polarized epifluorescence measurements were performed to test the assumption that the fluors were isotropically oriented in the film (x-y) plane. The fluorescence intensities measured from DiI- and BODIPYdoped films with the emission polarizer oriented along the xand y-axes were statistically indistinguishable, which supported this assumption. These same films were subsequently used for TIRF anisotropy measurements. Technical TIRF spectra of doped LB films were measured using the fluorescence microscope. The emission maxima were 565 and 518 nm for the DiI and BODIPY films, respectively. These values are very close to the 568 and 518 nm maxima reported for DiI and BODIPY dissolved in ethanol.37 The spectral band shapes for both types of films were similar in appearance to those observed from ethanol solutions of DiI and BODIPY. These data indicate that DiI and BODIPY were not significantly aggregated when incorporated into LB films at the low molar ratios employed for TIRF measurements. Aggregation is typically accompanied by changes in band shape and position and the appearance of new bands. Dispersal of fluors in the films is required to prevent energy transfer, which if allowed to occur would cause depolarization and yield artificially small anisotropies.28 It should be noted that fluorescence spectra of the BODIPY:AA films with a 1:90 molar ratio exhibited a broadened band shape with an emission maximum of 523 nm, parameters that suggest some degree of probe aggregation. Relatively brighter domains on the order of 1 µm in diameter were observed in these films when examined by epifluorescence. However, since the 1:90 films were used only for the IOWATR experiments, the presumed aggregation had no effect on the TIRF distribution measurement (∆θ or θσ).

Edmiston et al. Propagation loss coefficients were measured by IOW-ATR for waveguides coated with undoped LB films. There was no difference in loss between regions coated with three CdA layers and regions coated with three CdA plus two AA layers. Thus, the difference in LB film thickness across the waveguide structure had no effect on the measured dichroic ratios. Ellipsometry of undoped LB films deposited on oxidized silicon wafers was performed to determine film thickness. Using a film refractive index of 1.52, thicknesses of 86 and 131 Å were measured for three- and five-layer CdA films, respectively. These results agree well with the published thickness of 26.4 Å per layer.36 The FT-IR/ATR spectrum of undoped LB films was acquired by depositing the three CdA/two AA layer structure on the surface of an oxidized silicon ATR crystal (spectrum not shown). The hydrocarbon stretching region contained strong bands at 2918 and 2850 cm-1, which are assigned to the methylene asymmetric and symmetric stretching modes, νa(CH2) and νs(CH2), respectively. The frequencies of these bands have been used as indicators of the extent of lateral interactions between polymethylene chains in crystals, LB films, and self-assembled monolayers.2,7,38 The frequencies observed here are identical to those reported by Porter et al.38 for long chain (n g 15) alkanethiol monolayers self-assembled on gold and indicate that the polymethylene chains in the LB film are packed in a wellordered, solidlike state. DiI Orientation and Distribution. A LB multilayer of DiIdoped AA, a model molecular assembly with a presumably high degree of macroscopic order, was chosen as an initial test case for the IOW-ATR+TIRF method. DiI has been employed extensively in fluorescent probe studies of membranes and intact cells,37 and fluorescent anisotropy techniques have been used to assess molecular order in DiI-labeled structures.14,31 The transition dipole responsible for absorbance in the 500-550 nm wavelength regime is thought to orient along the long axis of the carbocyanine head group (Figure 2), approximately perpendicular to the alkyl chains.14 This suggests that when incorporated in a LB film, the transition dipole would lie roughly parallel to the film plane, meaning that the tilt angle should be nearly 90° if the alkyl chains are oriented normal to the film plane. Two planar IOWs coated with DiI-doped LB multilayer films were prepared and characterized. The mean dipole tilt angle measured by IOW-ATR was 77.4 ( 3.4° on one waveguide and 71.8 ( 2.8° on the second. Combining the results from both gives an overall θµ of 75 ( 4.1°. Note that the 4.1° represents the standard deviation of the linear dichroism measurement, not the angular distribution of dipoles in the film. Three fused silica substrates coated with DiI-doped LB multilayer films were prepared and characterized. The overall mean fluorescence anisotropy for the three samples was -0.351 ( 0.045. No systematic differences between samples were detected; the standard deviation among different spots on the same sample was about equal to the overall standard deviation. Using the mean values r ) -0.35 and θµ ) 75°, an angular distribution (θσ) of (12° was calculated from eqs 8 and 10, assuming a Gaussian distribution of dipole orientations. An alternate model is the step function, for which an angular distribution (∆θ) of (27° is obtained using eqs 9 and 10. These results show that (i) the dipole distribution in DiI-doped LB multilayers is centered at a tilt angle nearly parallel to the film plane and (ii) the dipoles are distributed in a moderate range about the mean angle. It is useful to consider the effect of measurement errors in r and θµ on the calculated orientation distribution. The range of

Dipole Orientation Distributions in LB Films

J. Phys. Chem., Vol. 100, No. 2, 1996 781

TABLE 1: DiI Orientation Distributions for the Gaussian and Step Function Models for Selected Combinations of Anisotropy and Mean Angle mean angle, deg

a

anisotropy -0.306

-0.351

-0.396

69.0 70.0 71.8 73.0 74.0 75.0b 76.0 77.4 79.0 80.8

Gaussian Model 8.6 n/oa 12.6 n/o 16.1 5.16 17.8 9.74 19.2 11.5 19.9 12.1 21.2 13.8 22.6 15.5 24.1 16.6 25.7 18.3

n/o n/o n/o n/o n/o 0.57 7.45 9.17 11.2 12.6

69.0 70.0 71.8 73.0 74.0 75.0b 76.0 77.4 79.0 80.8

Step Function Model 26.9 n/o 34.9 n/o 38.4 9.05 39.5 21.2 40.1 24.9 40.4 26.9 41.1 29.2 41.2 30.9 41.6 32.7 41.8 33.2

n/o n/o n/o n/o n/o 0.57 14.3 18.9 22.3 24.6

Not obtainable. b Overall measured mean.

possible distributions that can be calculated, using intervals for r and θµ of about two standard deviations centered about their mean values, is listed in Table 1 for both the step function and Gaussian models. The table shows that certain combinations of mean tilt angle and anisotropy, for instance, r ) -0.396 and θµ ) 70°, are not theoretically possible. This illustrates the “cross-examination” aspect of measuring absorption linear dichroism and fluorescence anisotropy in tandem. To some extent, each measurement verifies the validity of the other. This point is discussed in more detail below. A comparison can be made between our results and previous studies in which carbocyanine dye orientation in model membranes was examined using steady-state TIRF anisotropy. Axelrod14 reported that the transition dipole is parallel to the membrane surface in DiI-doped erythrocyte ghosts immobilized on poly(lysine)-coated substrates. Timbs and Thompson31 obtained similar results for a LB monolayer of distearoylphosphatidylcholine doped with 0.5% DiI. Using absorption LD, Ohta et al.17 measured mean tilt angles of 73°-82° for several carbocyanine dyes codeposited with AA in multilayer LB films. It must be emphasized that an orientation distribution could not be determined in any of these studies. However, Timbs and Thompson did report combinations of θµ ( ∆θ for a step function model that were consistent with their measured order parameters, where ∆θ is the polar semiangle of the cone in which the dipoles wobble on a time scale faster than that of the measurement: for θµ ) 70°, ∆θ ) 20°; for θµ ) 72°, ∆θ ) 30°; for θµ ) 78°, ∆θ ) 40°. These values are remarkably close to the distribution of 75° ( 27° reported here. If one assumes that DiI is incorporated into an AA film with the major axis of its polymethylene tail groups oriented perpendicular to the electronic transition dipole of the headgroup and that the tail groups are aligned parallel to the major axis of the polymethylene chains of the AA matrix, then θµ ) 75° for the headgroup corresponds to a mean tilt angle of 15° between the polymethylene chain axis and the surface normal. This result is also consistent with previous work. Numerous studies on alkyl chain orientation in LB films of fatty acid salts have been performed using a variety of techniques, including X-ray and

electron diffraction methods, external reflectance and attenuated total reflectance FT-IR spectroscopies, and near-edge X-ray absorption spectroscopy (NEXFAS). The books by Roberts2 and Ullman3 review many of these studies. Most have reported mean tilt angles ranging from 5° to 30° from the surface normal, although typically the first monolayer is tilted closer to 30° and is more disordered than subsequent layers, which are tilted at smaller angles. For example, Umemura et al. used FT-IR/ATR to measure a tilt angle of 7° between the polymethylene backbone and the surface normal in cadmium stearate multilayers deposited on silver and zinc selenide.8 Chollet reported tilt angles of 24° and 8° for behenic acid and calcium behenate deposited on CaF2, also using polarized FT-IR/ATR.39 Recently, Kinzler et al. reported NEXFAS measurements of orientation in CdA mono- and multilayers deposited on silicon, SiO2, and silver.40 In films containing three or more layers, alkyl chains were tilted at less than 16° on all substrates, whereas monolayers on silicon and SiO2 were tilted approximately 30°. In summary, our results for DiI-doped LB multilayers are consistent with prior studies of similar molecular assemblies and confirm that the IOW-ATR+TIRF method can be used to measure dipole orientation distributions in thin molecular assemblies. However, we cannot independently verify that the orientation distribution of DiI in AA films corresponds to either a Gaussian or a step function. BODIPY Orientation and Distribution. A second type of LB film assembly, BODIPY-doped AA, was also examined. BODIPY is a dipyrrometheneboron difluoride; the derivative employed here was a phospholipid labeled at the terminus of one of the acyl chains (Figure 2). This molecule was selected as a contrast to DiI for two reasons: (i) The transition dipole responsible for absorbance near 500 nm is thought to orient along the long axis of the aromatic structure.41 This suggests that when incorporated into a LB film, the transition dipole would lie roughly parallel to the major axis of the alkyl chains. Thus, the mean tilt angle is predicted to be substantially less than that obtained for DiI in an AA LB film. (ii) When incorporated into a LB film, the fluor is expected to be located near the center of the alkyl chain region of the film structure. Some evidence42 suggests that BODIPY experiences a disordered environment when confined to the interior of a lipid bilayer, relative to the headgroup region. Two planar IOWs coated with BODIPY-doped LB multilayer films were prepared and characterized. The mean dipole tilt angle was 58.8 ( 2.5° on one waveguide and 63.5 ( 1.7° on the second, giving an overall θµ of 62 ( 3.5°. Again, the 3.5° represents the standard deviation of the linear dichroism measurement, not the angular distribution of dipoles in the film. Three fused silica substrates coated with BODIPY-doped multilayer LB films were prepared and characterized. The overall mean fluorescence anisotropy for the three samples was -0.169 ( 0.032. As with DiI, the standard deviation among different spots on the same sample was about equal to the overall standard deviation. The range of angular distributions that can be calculated for both the step function and Gaussian models, using intervals for r and θµ of about two standard deviations centered about their mean values, is listed in Table 2. The table shows that the overall means of r ) -0.17 and θµ ) 62° represent a combination that is theoretically impossible for either a Gaussian or a step function orientation distribution. However, the table also shows that, for both models, angular distributions can be obtained for several combinations of r and θµ that lie within one standard deviation of the mean values. One of these combinations is θµ ) 59° and r ) -0.17, the mean for the first

782 J. Phys. Chem., Vol. 100, No. 2, 1996

Edmiston et al.

TABLE 2: BODIPY Orientation Distributions for the Gaussian and Step Function Models for Selected Combinations of Anisotropy and Mean Angle mean angle, deg

a

anisotropy -0.137

-0.169

-0.201

56.3 57.0 58.0 58.8b 60.0 61.0 62.0c 63.5 65.2

Gaussian Model 15.5 22.9 14.3 21.2 12 18.3 10.3 17.2 6.3 11.5 n/o 8 n/o n/o n/o n/o n/o n/o

n/oa 80.2 75 57.3 28.7 16.1 10.9 n/o n/o

56.3 57.0 58.0 58.8b 60.0 61.0 62.0c 63.5 65.2

Step Function Model 23.2 27.8 21.5 26.4 17.1 22.9 15.8 21.9 9.1 18 n/o 13.1 n/o n/o n/o n/o n/o n/o

33.8 31.2 28.6 27.7 24.8 21.8 17.4 n/o n/o

Not obtainable. b Waveguide 1 mean. c Overall measured mean.

Figure 4. Gaussian probability distributions for fluors doped into a LB bilayer of arachidic acid: solid line, DiI with θµ ) 75° and θσ ) 12°; dashed line, BODIPY with θµ ) 59° and θσ ) 17°.

waveguide and the overall mean anisotropy, respectively. In this case, the Gaussian model yields a distribution of 59 ( 17° and the step function model a distribution of 59 ( 22°. The fact that the mean tilt angle obtained from the second waveguide does not fall within the physical boundaries imposed by either model suggests that the 63.5° value is systematically high. The contradiction inherent in the combination of θµ ) 63.5° and r ) -0.17 again illustrates the “cross-examination” aspect of measuring absorption linear dichroism and fluorescence anisotropy in tandem. The Gaussian distribution of 17° obtained from the first waveguide is 5° greater than the Gaussian distribution obtained for DiI, indicating a relatively greater degree of disorder in the BODIPY-doped films. This is illustrated in Figure 4, where normalized Gaussian orientation distributions for DiI and BODIPY are plotted. The broader distribution for a probe located in the hydrophobic region of the film, relative to a probe located in the headgroup region, is consistent with previous work. NMR studies of acyl chain order have shown that methylenes in the center of liquid crystalline lipid bilayers are considerably more disordered than those near the surface.43 Johnson et al.42 measured fluorescence anisotropy from phosphocholine bilayers containing lipids labeled with BODIPY at different positions along the acyl chains. The anisotropy decreased as the distance between the fluor and the headgroup

Figure 5. Response surface for the Gaussian probability function showing theoretically possible combinations of anisotropy, mean tilt angle, and angular distribution (θσ). Only θσ from 0° to 90° are plotted.

increased, again showing that the probe experienced a more disordered environment in the center of the bilayer. Another feature of the IOW-ATR+TIRF method that deserves discussion is the ability to distinguish between a molecular assembly that is partially aligned and one that is randomly oriented. In theory, an isotropic assembly of linear dipoles corresponds to a mean tilt angle of 54.7° and an anisotropy of -0.2. With respect to the BODIPY results, θµ for the first waveguide (59°) was less than two standard deviations from 54.7°, and the overall mean anisotropy (-0.17) was only one standard deviation from -0.2. Consequently, the possibility that the BODIPY distribution may be isotropic must be considered. This possibility is diminished by independent measurement of both θµ and r; in tandem these results more decisively support the conclusion that the BODIPY distribution is partially aligned. Assessment. Some issues regarding the application and utility of the IOW-ATR+TIRF method need to be considered. First, it is important to note that the parameter space in which combinations of anisotropies and mean tilt angles are physically reasonable is considerably smaller than the theoretically allowable ranges for r and θµ alone. For example, a molecular assembly cannot have a mean tilt angle of 10° and an anisotropy of -0.450, which would correspond to predominately out-ofplane and in-plane orientations, respectively. The relationship is depicted in Figure 5, in which possible values for anisotropy, mean angle, and distribution (θσ) are plotted. The response surface defines those combinations of the three parameters that are theoretically possible within the range of 0° e θσ e 90°. The “cross-examination” aspect of measuring absorption LD and fluorescence anisotropy is evident from the plot. Most of the parameter space defined by the allowable ranges for r and θµ (-0.5 e r e 1 and 0° e θµ e 90°, respectively) represent combinations that are inconsistent with a Gaussian orientation distribution. Figure 5 also shows that the anisotropy is most sensitive to the distribution when θµ is near 90° or 0° and is relatively insensitive when θµ is around 60°. The trend is illustrated more clearly in Figure 6, in which anisotropy is plotted as a function of distribution for θµ ) 5°, 25°, 60°, and 85°. When θµ ) 5°, r can vary from -0.161 to 0.913, whereas when θµ ) 60°, r can vary only from -0.200 to -0.139, which is a 12-fold smaller range. Thus, a less precise evaluation of the width of the distribution can be obtained when θµ ) 60°. Note that the mean tilt angle obtained for the BODIPY-doped LB films fell

Dipole Orientation Distributions in LB Films

Figure 6. Relationship between anisotropy and angular distribution (θσ) for the Gaussian probability function at several mean tilt angles: A, θµ ) 5°; B, θµ ) 25°; C, θµ ) 60°; D, θµ ) 85°.

within this insensitive region. The effect is apparent from the range of distributions calculated using the Gaussian model for values of r and θµ within about two standard deviations of their respective means (Table 2). At one extreme there is a combination of r and θµ that yields a θσ value greater than 90°, whereas at the other extreme there are combinations that are inconsistent with even unrealistically small values of θσ. In contrast, for the DiI-doped films, the corresponding range of distributions is only 0°-26° (see Table 1), despite the fact that the standard deviation of the anisotropy measured for DiI (0.045) was greater than that for BODIPY (0.032). The smaller range of distributions for DiI is a consequence of its greater mean tilt angle; when θµ ) 75°, the anisotropy is much more sensitive to the distribution. Second, in our experience, the standard deviation of a θµ measurement using IOW-ATR is typically e5°. We attribute these fluctuations primarily to differences in intrinsic loss among planar waveguides. This is supported by the fact that the standard deviation for multiple measurements on the same sample is always less than that among multiple samples. For the LB films examined here, the angular dipole distribution calculated from r and θµ was always greater than the standard deviation of the IOW-ATR measurement. However, it is conceivable that the opposite could be true for a given sample. In this case, the distribution could qualitatively be described as very narrow, but the precision of the IOW-ATR measurement would prohibit a numerical value from being assigned. Thus, in the case of two highly ordered assemblies of dipoles, it could be difficult to discern real differences using the IOWATR+TIRF method to assess orientation distribution. A related issue is substrate flatness. Taking the Gaussian model for DiI orientation in a LB multilayer as an example, we do not know what fraction of the (12.1° angular distribution can be attributed to substrate surface roughness. However, the fraction can be estimated using data reported by Burbage and Wirth.17 They used atomic force microscopy to measure a rootmean-square surface roughness of 0.7 Å over a scan area of 200 × 200 Å2 on a polished fused silica plate. Assuming that this fluctuation occurs over a molecular dimension of 20 Å, the surface roughness contributed tan-1 (0.7/20), or only 2°, to the measured distribution. It therefore appears that distributions substantially greater than several degrees cannot be attributed solely to substrate surface roughness. Third, the IOW-ATR and TIRF measurements were performed on LB films prepared on different substrates, although this was done for convenience and is not a requirement inherent

J. Phys. Chem., Vol. 100, No. 2, 1996 783 in the method. Thus, it is possible that the film structure was influenced by the substrate. (We also note that the FT-IR/ATR and ellipsometry data were obtained on different substrates.) However, since fluors were present only in the fourth and fifth layers of the LB multilayer, the effect was probably minimal. This assumption is supported by Kinzler et al.,40 who found that mean polymethylene chain orientation in CdA multilayers deposited on silicon, SiO2, and silver was independent of the substrate. Fourth, the IOW-ATR+TIRF method is sensitive to submonolayer surface coverages. In the case of DiI, the effective surface coverage in the films examined here was about 0.02 of a monolayer. This coverage is well above the IOW-ATR detection limit of 0.004 of a DiI monolayer reported by Plowman et al.26 The method therefore shows potential for measuring orientation distributions in molecular assemblies with much poorer spectral properties and surface coverages than the LB films examined here. Fifth, as implemented here, the IOW-ATR+TIRF method requires the assumption of a functional form for N(θ). We emphasize that the validity of fitting the distribution of dipoles in a molecular assembly to either a Gaussian or a step function model has not been independently verified. Intuitively the Gaussian appears to be more physically realistic. The step function implies that, between the limits of (∆θ, all deviations (i.e., film defects) from the mean are equally probable and that defects outside of (∆θ are impossible. In contrast, the Gaussian model predicts that the more significant defects are less probable, but never impossible. Different models may be more appropriate to describe orientation distributions in other types of molecular assemblies. Of course when using a technique that measures only two variables, distribution functions that contain more than two independent parameters cannot be considered. However, even though a true orientation distribution may be unobtainable, the simple models employed here can still be used to make a valid comparison of macroscopic order in two molecular assemblies on a relative basis. Conclusion The combined use of the IOW-ATR LD and TIRF anisotropy techniques can be employed to determine the orientation distribution of fluorescent molecules in a substrate-supported film. The results for the dye-impregnated LB films examined here show that dipoles in the headgroup region are more ordered than dipoles in the alkyl chain region. The IOW-ATR+TIRF method should prove useful in studying relationships between assembly technique, structure, and function in two-dimensional molecular arrays. We are currently investigating orientation and order in hydrated protein films formed by self-assembly on chemically functionalized substrates. The results will be reported in future communications. Acknowledgment. This work was supported by the National Science Foundation (Grant CHE-9403896) and the National Institutes of Health (Grant R29 GM50299). Acknowledgment is also made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for partial support of this research. References and Notes (1) Swalen, J. D.; Allara, D. L.; Andrade, J. D.; Chandross, E. A.; Garoff, S.; Israelchvili, J.; McCarthy, T. J.; Murray, R.; Pease, R. F.; Rabolt, J. F.; Wynne, K. J.; Yu, H. Langmuir 1987, 3, 932. (2) Roberts, G. Langmuir-Blodgett Films; Plenum: New York, 1990. (3) Ulman, A. An Introduction to Ultrathin OrganicFilms; Academic: San Diego, 1991.

784 J. Phys. Chem., Vol. 100, No. 2, 1996 (4) See for example: Rabe, J. P.; Swalen; J. D.; Outka, D. A.; Sto¨hr, J. Thin Solid Films 1988, 159, 275. (5) See for example: Zasadzinski, J. A.; Viswanathan, R.; Madsen, L.; Garnaes, J.; Schwartz, D. K. Science 1994, 263, 1726. (6) Gregory, B. W.; Dluhy, R. A.; Bottomley, L. A. J. Phys. Chem. 1994, 98, 1010. (7) Hsu, Y.; Penner, T. L.; Whitten, D. G. Langmuir 1994, 10, 2757. (8) Umemura, J.; Kamate, T.; Kawai, T.; Takenaka, T. J. Phys. Chem. 1990, 94, 62. (9) (a) Cropek, D. M.; Bohn, P. W. J. Phys. Chem. 1990, 94, 6452. (b) Hughes, K. D.; LaBuda, M. J.; Bohn, P. W. Appl. Opt. 1991, 30, 4406. (10) Stayton, P. S.; Ollinger, J. M.; Jiang, M.; Bohn, P. W.; Sligar, S. G. J. Am. Chem. Soc. 1992, 114, 9298. (11) Pachence, J. M.; Amador, S.; Maniara, G.; Vanderkooi, J.; Dutton, P. L.; Blasie, J. K. Biophys. J. 1990, 58, 379. (12) Walker, D. S.; Hellinga, H. W.; Saavedra, S. S.; Reichert, W. M. J. Phys. Chem. 1993, 97, 10217. (13) Fraaije, J. G. E. M.; Kleijn, J. M.; van der Graaf, M.; Dijt, J. C. Biophys. J. 1990, 57, 965. (14) Axelrod, D. Biophys. J. 1979, 26, 557. (15) Thompson, N. L.; McConnell, H. M.; Burghardt, T. P. Biophys. J. 1984, 46, 739. (16) (a) Wirth, M. J.; Burbage, J. D. Anal. Chem. 1991, 63, 1311. (b) Burbage, J. D.; Wirth, M. J. J. Phys. Chem. 1992, 96, 5943. (c) Wirth, M. J.; Piasecki-Coleman, D. A.; Montgomery, Jr., M. E. Langmuir 1995, 11, 990. (17) Ohta, N.; Matsunami, S.; Okazaki, S.; Yamazaki, I.; Langmuir 1994, 10, 3909. (18) LeGrange, J. D.; Riegler, H. E.; Zurawsky, W. P.; Scarlata, S. F. J. Chem. Phys. 1989, 90, 3838. (19) Bos, M. A.; Kleijn, J. M. Biophys. J. 1995, 68, 2566. (20) Lesieur, P.; Vandevyver, M.; Ruaudel-Teixier, A.; Barraud, A. Thin Solid Films 1988, 159, 315. (21) Higgins, D. A.; Byerly, S. K.; Abrams, M. B.; Corn, R. M. J. Phys. Chem. 1991, 95, 6984. (22) Collinson, M.; Bowden, E. F. Anal. Chem. 1992, 64, 1470. (23) Kogelnik, H. In Topics in Applied Physics; Tamir, T., Ed.; SpringerVerlag: Berlin, 1979; Vol. 7, Chapter 2.

Edmiston et al. (24) Saavedra, S. S.; Reichert, W. M. Anal. Chem. 1990, 62, 22512256. (25) Saavedra, S. S.; Reichert, W. M. Langmuir 1991, 7, 995-999. (26) Plowman, T. E.; Garrison, M. D.; Walker, D. S.; Reichert, W. M. Thin Solid Films 1994, 243, 610-615. (27) Lee, J. E.; Saavedra, S. S In Proteins at Interfaces 1994 ; Horbett, T., Ed.; ACS Symposium Series, in press. (28) Szabo, A. J. Chem. Phys. 1980, 72, 4620. (29) Reichert, W. M. Crit. ReV. Biocompat. 1989, 5, 173. (30) Harrick, H. J. Internal Reflection Spectroscopy, 2nd ed.; Harrick Scientific: New York, 1979. (31) Timbs, M. M.; Thompson, N. L. Biophys. J. 1990, 58, 413. (32) Walker, D. S.; Reichert, W. M.; Berry, C. J. Appl. Spectrosc. 1992, 46, 1437. (33) Saavedra, S. S.; Reichert, W. M. Appl. Spectrosc. 1990, 44, 1210. (34) Phimphivong, S.; Ko¨lchens, S.; Edmiston, P. L.; Saavedra, S. S. Anal. Chim. Acta 1995, 307, 403. (35) Yang, L.; Saavedra, S. S.; Armstrong, N. R.; Hayes, J. Anal. Chem. 1994, 66, 1254-1263. (36) Suci, P. A.; Reichert, W. M. Appl. Spectrosc. 1988, 42, 120. (37) Haugland. R. P. Handbook of Fluorescent Probes and Research Chemicals, 5th ed.; Molecular Probes: Eugene, OR, 1992. (38) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559. (39) Chollet, P. A. Thin Solid Films 1978, 52, 343. (40) Kinzler, M.; Schrepp, W.; Wo¨ll, Ch.; Grunze, M. Langmuir 1995, 11, 696. (41) Karolin, J.; Johansson, L. B.-A.; Strandberg, L.; Ny, Tor J. Am. Chem. Soc. 1994, 116, 7801. (42) Johnson, I. D.; Kang, H. C.; Haugland, R. P. Anal. Biochem. 1991, 198, 228. (43) Gennis, R. B. Biomembranes: Molecular Structure and Function; Springer-Verlag: New York, 1989; p 54.

JP952037P