Molecular Structure of an Alkyl-Side-Chain Polymer−Water Interface

Langmuir 2004, 20, 8625-8633

8625

Molecular Structure of an Alkyl-Side-Chain Polymer-Water Interface: Origins of Contact Angle Hysteresis Hasnain Rangwalla,† Alexander D. Schwab,†,‡ Betu¨l Yurdumakan,† Dalia G. Yablon,§ Mohsen S. Yeganeh,§,| and Ali Dhinojwala*,† Department of Polymer Science, University of Akron, Akron, Ohio 44325-3909, and ExxonMobil Research and Engineering Company, Annandale, New Jersey 08801 Received April 12, 2004. In Final Form: July 16, 2004 A new and direct approach to verify surface heterogeneity as the microscopic origin of contact-angle hysteresis is demonstrated. IR-visible sum-frequency-generation spectroscopy (SFG) was used to selectively probe the molecules at the interface of an alkyl-side-chain polymer [poly(vinyl n-octadecyl carbamateco-vinyl acetate)] with water. The spectra indicate that in contact with water, the polymer surface is heterogeneous (having areas of differing surface energies). This evidence of surface heterogeneity supports the hysteresis observed in the advancing and receding contact angles of the polymer surface with water. The same measurements made for the chemically and structurally similar surface of an octadecyltrichlorosilane self-assembled monolayer indicates a homogeneous surface at the water interface. In this case, contact-angle hysteresis measurements implicate surface roughness as the cause of hysteresis. Atomic force microscopy measurements of roughness for these surfaces further support our conclusions. The polymer-water interface was probed using SFG at above-ambient temperatures, and an order-to-disorder transition (ODT) of alkyl side chains at the interface was observed, which closely follows the melting of crystalline side chains in the bulk. This transition explains the increased wettability of the polymer, by water, when the temperature is raised above the bulk melting temperature. Furthermore, the irreversibility of this ODT suggests that the disordered polymer-water interface is the thermodynamic equilibrium state, whereas the before-heating structure of this interface is a kinetically hindered metastable state.

* Corresponding author. E-mail: [email protected]. † University of Akron. ‡ Current address: Department of Physics, Haverford College, Haverford, PA 19041-1392. § ExxonMobil Research and Engineering Company. | E-mail: [email protected].

surfaces due to segmental motions of the polymer molecules are generally slow compared to the short time scales of hysteresis measurements;4,7 hence, these effects are negligible too. Therefore, the origin of hysteresis is often attributed to surface heterogeneity, although the alleged cause is never directly observed, only inferred.4-7 This is not to doubt that surface heterogeneity causes hysteresis; extensive studies have been reported in the literature that firmly establish this.8-14 However, the real or model surfaces used in these studies are made heterogeneous by artificial means, and it is unclear how representative they are of commonly encountered polymeric surfaces. Specifically, some of the heterogeneous surfaces are achieved by monolayers that are incomplete to various degrees by preparation methods or by radiation damage,9,14 the heterogeneity is of macroscopic dimensions,12,13 or the heterogeneity is patterned using well-defined geometries.8,10-12 To establish the microscopic origins of hysteresis, a conventional approach is to infer the type of nonideality from the contact-angle measurements themselves. Relating hysteresis to surface nonidealities has been a subject of active research for several decades now. Wenzel15 predicted the deviation in the equilibrium contact angle

(1) Israelachvili, J. N. Intermolecular & Surface Forces, 2nd ed.; Academic Press: San Diego, 1991; section 15.4. (2) Johnson, R. E., Jr.; Dettre, R. H. Wettability and Contact Angles. In Surface and Colloid Science; Matijevic´, E., Ed.; John Wiley & Sons, Inc.: New York, 1969; Vol. 2. (3) Morra, M.; Occhiello, E.; Garbassi, F. Adv. Colloid Interface Sci. 1990, 32, 79-116. (4) Penn, L. S.; Miller, B. J. Colloid Interface Sci. 1980, 78, 238-241. (5) Tretinnikov, O. N.; Ikada, Y. Langmuir 1994, 10, 1606-1614. (6) Takahara, A.; Jo, N.-J.; Takamori, K.; Kajiyama, T. Influence of Aqueous Environment on Surface Molecular Mobility and Surface Microphase Separated Structure of Segmented Poly(ether urethanes) and Segmented Poly(ether urethane ureas). In Progress in Biomedical Polymers; Gebelein, C. G., Dunn, R. L., Eds.; Plenum Press: New York, 1990.

(7) Pike, J. K.; Ho, T.; Wynne, K. J. Chem. Mater. 1996, 8, 856-860. (8) Johnson, R. E., Jr.; Dettre, R. H. J. Phys. Chem. 1964, 68, 17441750. (9) Dettre, R. H.; Johnson, R. E., Jr. J. Phys. Chem. 1965, 69, 15071515. (10) Neumann, A. W.; Good, R. J. J. Colloid Interface Sci. 1972, 38, 341-358. (11) Schwartz, L. W.; Garoff, S. Langmuir 1985, 1, 219-230. (12) di Meglio, J. M. Europhys. Lett. 1992, 17, 607-612. (13) Andrieu, C.; Sykes, C.; Brochard, F. Langmuir 1994, 10, 20772080. (14) Decker, E. L.; Garoff, S. Langmuir 1997, 13, 6321-6332. (15) Wenzel, R. N. Ind. Eng. Chem. 1936, 28, 988-994.

1. Introduction Hysteresis seen in the contact angles of liquids on solid surfaces is commonly caused by surface roughness, surface heterogeneity (microscopic areas of differing surface energies), or restructuring of molecules at the solid surface after it comes in contact with the liquid (time-dependent change of interfacial energy).1 Although contact-angle hysteresis measurements are a sensitive probe of such surface nonidealities,2 these measurements are, at best, indirect evidence of one or more of the above-mentioned causes. Therefore, conclusions drawn on the microscopic origins of hysteresis from the sole measurement of contact angles can be error-prone. This is particularly true for polymers, where contact-angle measurements using water, as a polar liquid, are widely used to characterize surface properties.3 Usually, well-prepared polymer surfaces are smooth;4-6 therefore, roughness is not a dominating factor. Also, time-dependent changes for such

10.1021/la049073c CCC: $27.50 © 2004 American Chemical Society Published on Web 09/03/2004

8626

Langmuir, Vol. 20, No. 20, 2004

Figure 1. Chemical structure of PVNODC. x (≈0.1) and y (≈0.9) are the mole fractions of the corresponding monomer units.

due to surface roughness; Cassie16 extended this work to heterogeneous surfaces; and Johnson and Dettre2 and Neumann and Good10 used models of nonideal surfaces to predict the deviations of advancing and receding angles from the equilibrium angle. Hence, from the measured deviations of these angles, it is possible to surmise the type of surface nonideality that is responsible for the effect. However, as mentioned earlier, this is an indirect method. Another approach taken is to apply separate, surfacesensitive techniques that require the surface to be in a vacuum or air.6,17 But these methods, like the methods used in the works cited earlier,9,12-14 are not an in situ probe of the buried solid-liquid interface, and they cannot differentiate between regions that are chemically the same but behave differently when in contact with the liquid by short-time-scale molecular motions (few bond rotations). Such rearrangements, which are fast and reversible, alter the surface energies of some fraction of the solid area, and this results in a heterogeneous surface only in the liquid environment. In the first part of this work, we demonstrate a new and direct approach to verify surface heterogeneity as the microscopic origin of contact-angle hysteresis. We used a nonlinear, vibrational spectroscopy to selectively probe the molecules at the interface of an alkyl-side-chain polymer with water. Specifically, we used IR-visible sumfrequency-generation (SFG) spectroscopy18,19 to probe molecular-level informationscomposition, structure, and orientationsof the solid-liquid interface, in situ. The polymer we used is poly(vinyl n-octadecyl carbamate-covinyl acetate) (PVNODC); see Figure 1. PVNODC belongs to a class of polymers that are useful for creating hydrophobic surfaces (e.g., nonstick release coatings for pressure-sensitive adhesives).20,21 We made the same measurements for the surface of an octadecyltrichlorosilane (OTS) self-assembled monolayer (SAM) at the interface with water. The OTS SAM has a surface composition and air-interface structure22 similar to those of PVNODC23 and serves as a good model for comparison. We also made contact-angle measurements with water on these two surfaces. The hysteresis observed in the advancing and receding angles for the polymer and the SAM is consistent with what is expected for the respective surface based on SFG results. Therefore, it is clear that the SFG results support the inferences made from contact(16) Cassie, A. B. D. Discuss. Faraday Soc. 1948, 3, 11-16. (17) Morita, M.; Ogisu, H.; Kubo, M. J. Appl. Polym. Sci. 1999, 73, 1741-1749. (18) Shen, Y. R. Nature 1989, 337, 519-525. (19) Bain, C. D. J. Chem. Soc., Faraday Trans. 1995, 91, 1281-1296. (20) Kinning, D. J. J. Adhes. 1997, 60, 249-274. (21) Li, L.-H.; Macosko, C.; Korba, G. L.; Pocius, A. V.; Tirrell, M. J. Adhes. 2001, 77, 95-123. (22) Guyot-Sionnest, P.; Superfine, R.; Hunt, J. H.; Shen, Y. R. Chem. Phys. Lett. 1988, 144, 1-5. (23) Gautam, K. S.; Dhinojwala, A. Macromolecules 2001, 34, 11371139.

Rangwalla et al.

angle measurements and, in fact, serve as direct evidence of the type of surface nonideality governing the hysteresis. To further corroborate our conclusion, we measured the roughness of these surfaces using atomic force microscopy (AFM) in tapping mode, and these results are supportive, too. In the second (and last) part of this work, we examine the rearrangements of the molecular species at the PVNODC-water interface as a function of temperature. This measurement elucidates the dynamic properties of the interface, such as the thermal order-to-disorder transitions (ODT). Also, it clarifies the molecular-level process responsible for the temperature dependence of wettability. 2. Experimental Section 2.1. Sample Preparation. PVNODC was a gift from 3M Corp. and was used as received. Some characteristics of this polymer (Figure 1) are as follows: Mw ) 70 kg/mol, Mw/Mn ≈ 3.0, and the mole fraction of n-octadecyl carbamate units per chain ≈ 90%. Thin films (100-200 nm) of PVNODC were prepared on flat glass plates and on any one face of equilateral, sapphire prisms by spin coating as follows: using a 4 wt % solution of the polymer in toluene, at a spin speed of 2000 rpm for 60 s, and at ≈35 °C. After coating, the films were annealed at 100 °C for 3 h under vacuum. OTS was purchased from Gelest, Inc., and was used as received. The OTS SAMs were prepared on sapphire disks and sapphire prisms as follows: the substrates were individually sonicated in a water-saturated solution of hexadecane, carbon tetrachloride, chloroform, and OTS; then rinsed for ≈30 s in carbon tetrachloride and then cyclohexane; and then baked in an oven at 130 °C for 4 h under a vacuum, to drive out residual water. 2.2. Surface-Wetting Measurements. The wetting behavior of the PVNODC and OTS SAM surfaces was investigated by measuring contact angles with water, using the sessile-drop method on a tilted surface. The water used for these and the SFG (below) measurements was distilled and deionized (resistivity g 18 MΩ‚cm). A Rame´-Hart goniometer (model 100-07-00, Rame´Hart, Inc.) was used. Contact angles were measured after the substrate was tilted just short of the angle at which the drop would roll off. We also measured the dynamic contact angles with water for these surfaces using the method recommended by Johnson and Dettre.24 The results of these alternative measurements lead us to the same conclusions. 2.3. AFM Measurements. AFM measurements were done, in air, to estimate surface roughness. A Nanoscope IIIa (Digital Instruments) in the Department of Polymer Science, University of Akron, was used. All measurements were done in tapping mode using commercial silicon cantilevers of the following characteristics: spring constant, 3 N/m; resonant frequency, 75 ( 25 kHz; and tip radius, 10 nm. Image processing was done using the Nanoscope Control software (Digital Instruments) version 5.12r3. 2.4. SFG Measurements. All SFG experiments were done on the ExxonMobil SFG system-II (at ExxonMobil Research and Engineering Co., Annandale, NJ). This system and its instrumentation have been previously described.25 The incident IR (tunable in the range 2750-3600 cm-1) and visible (532 nm) pulses had the following attributes: duration, ≈7 ns; repetition rate, 10 Hz; energy, 1-2 mJ/pulse (IR) and ≈3 mJ/pulse (visible); and bandwidth, 0.2 cm-1. The sample geometry used is shown in Figure 2. PVNODC was spin-coated on one face of an equilateral, sapphire prism. The prism was then mounted on a stainless steel cell. The cell design allows the polymer surface to be exposed to air, water, or flowing N2, and it allows the interface, thus formed with the polymer, to be probed by the SFG apparatus. Also, the cell can be connected to a heater with a temperature controller that allows (24) Johnson, R. E., Jr.; Dettre, R. H. Wetting of Low-Energy Surfaces. In Wettability; Berg, J. C., Ed.; Marcel Dekker: New York, 1993; Vol. 49. (25) Yeganeh, M. S.; Dougal, S. M.; Polizzotti, R. S.; Rabinowitz, P. Thin Solid Films 1995, 270, 226-229.

Alkyl-Side-Chain Polymer-Water Interface Structure

Langmuir, Vol. 20, No. 20, 2004 8627 I(ω3 ) ω1 + ω2) ∝ |χijk(ω2) Ej(ω1) Ek(ω2)|2

(1)

Here, χijk(ω2) (or χijk for simplicity) is a component of the secondorder susceptibility tensor, χ, of the surface, where i, j, k ) x, y, z. (x, y, and z are the lab axes in Figure 2, and from here on, the indices ijk appearing together will have this meaning.) For an interface with azimuthal isotropy (such as all the interfaces in this work), only seven combinations of ijk, in χijk’s, out of the 27 possibilities are nonvanishing, and only four are independent:29 χxxz ) χyyz, χxzx ) χyzy, χzxx ) χzyy, and χzzz. The χijk’s are each a sum of one nonresonant term and Q resonant terms, one for each vibrational mode of each surface species.

Aijk,q

Q

iΦ χijk(ω2) ) χNR + ijk e

∑ω q)1

Figure 2. Schematic diagram (not to scale) of the total-internalreflection geometry used for SFG. The rectangular box on top is a magnified view of the region in which an interface is being probed. The beams of frequencies ωi are as follows: i ) 1, s- or p-polarized visible; 2, s- or p-polarized IR; and 3, SFG. L is a polarizer, and T is a photomultiplier-tube detector. The mediums of refractive indices ni are as indicated. The interfaces formed at the boundaries of these mediums are as follows: A and D, sapphire-air; B, PVNODC-sapphire; and C, PVNODC-water (or air). The incident angles (ω1 and ω2 beams) and the refracted angles (ω3 beam) at these interfaces are denoted by φinterface,i. φA,2 is ≈1.5° smaller than φA,1. The x, y, and z Cartesian axes are the lab frame of reference; the y axis is perpendicular to the plane of the paper. the polymer interface to be heated and cooled in the ambient200 °C range. The OTS SAM prepared on the prism face was studied in a similar manner. The angle φA,1 (Figure 2) is selected so that the incident angles at the PVNODC-water or PVNODC-air (C) interfaces φC,1, φC,2, φC,3sare close to the critical angles for the respective beamss visible, IR, and SFGsat this interface. This set of φC’s greatly enhances the SFG signal from the C interface, while reducing any interfering signal from the polymer-sapphire (B) interface. Therefore, proper choice of φA,1 allows the selective probing of the C interface.26,27 We used a φA,1 of 42.5° for probing the polymer-air interface and 18° for the polymer-water interface. The polarization of the probe (IR and visible) beams and that of the SFG signal beam (before it reaches the detector) are set in one of two ways: s-polarized electric field parallel to the y axis or p-polarized electric field in the x-z plane (Figure 2 shows the xyz frame of reference). The combination of polarizations of all three beams (e.g., ssp) is given in the following sequence: polarization of the SFG beam, visible beam, and IR beam. 2.5. SFG Theory. The theory of SFG has been explained in the literature28-31 and is not detailed here. However, the formulas applied in interpreting the results of our work are presented and briefly explained below. The SFG signal intensity at frequency ω3 (see Figure 2) depends on the probing visible- and IR-beam electric-field components as follows:32 (26) Lo¨bau, J.; Wolfrum, K. J. Opt. Soc. Am. B 1997, 14, 2505-2512. (27) Gautam, K. S.; Schwab, A. D.; Dhinojwala, A.; Zhang, D.; Dougal, S. M.; Yeganeh, M. S. Phys. Rev. Lett. 2000, 85, 3854-3857. (28) Shen, Y. R. The Principles of Nonlinear Optics; John Wiley & Sons: New York, 1984; Chapter 6. (29) Hirose, C.; Akamatsu, N.; Domen, K. Appl. Spectrosc. 1992, 46, 1051-1072. (30) Hirose, C.; Akamatsu, N.; Domen, K. J. Chem. Phys. 1992, 96, 997-1004. (31) Hirose, C.; Yamamoto, H.; Akamatsu, N.; Domen, K. J. Phys. Chem. 1993, 97, 10064-10069. (32) Zhuang, X.; Miranda, P. B.; Kim, D.; Shen, Y. R. Phys. Rev. B: Condens. Matter 1999, 59, 12632-12640.

2

- ωq + iΓq

(2)

where Aijk,q, ωq, and Γq are the amplitude, frequency, and width, respectively, of the resonance q and Φ is the relative phase of the nonresonant term with respect to the resonant terms. The SFG spectra in this work were normalized for the variation in I(ω1) and I(ω2) (eq 1), and then some of these spectra were then fit to eq 2. Also, some spectra are rescaled before being displayed along with other spectra. This is required because in SFG it is difficult to reproduce the intensity of spectra when the sample is removed and then put back, requiring the realignment of probe beams incident on the sample and the signal beam incident on the detector. Hence, when we intend to compare only the spectral features and the relative peak strengths between spectra that were taken on separate alignments of the sample, some of the spectra are rescaled to fit the graph. The Aijk,q’s have their origins in the molecular hyperpolarizability tensor (β) components, and below we show how the two quantities are related. The components of β are as follows:33 Q

βlmn(ω2) )

∑ω q)1

βlmn,q 2

- ωq + iΓq

where

l, m, n ) a, b, c (3)

Here, a, b, and c are axes of the Cartesian frame of reference that is fixed to the molecule; c is conventionally taken to coincide with the axis (or one of the axes) of highest symmetry of the molecule. βlmn,q ) Alm,qMn,q (see ref 34), where Alm,q is the lm component of the Raman tensor and Mn,q is the n component of the transitiondipole-moment vector. (Alm,q should not be confused with the amplitude Aijk,q in eq 2.) Therefore, only those vibration modes that are both Raman- and IR-active contribute to the hyperpolarizability tensor. The βlmn(ω2)’s (or βlmn’s for simplicity) can be projected on the lab axes (xyz) by using the 27 × 27 projection coefficients Uijk:lmn.

βijk(ω2, Ω) )

∑

Uijk:lmn(Ω) βlmn(ω2)

(4)

l,m,n)a,b,c

Ω ) Euler angles (ψ, θ, φ) Note that these coefficients are a function of Ω, the Euler angles that describe the orientation of the molecular axes (abc) with respect to the xyz axes.29 (φ here should not be confused with angles in Figure 2. The angle ψ is the same as χ in ref 29.) The resonant portion of eq 2 is then a summation of the appropriate (ijk) component of hyperpolarizability for all interfacial molecules.31 (33) In eq 3 and the subsequent expressions that follow, we have made the restrictive assumption that all the vibrational modes, q, belong to a single type of molecular species. Although this is usually not true, and untrue even in this work, the extension for the general case of multiple types of species is trivial. However, the general treatment would unnecessarily complicate the notation.

8628

Langmuir, Vol. 20, No. 20, 2004 Aijk,q

Q

∑ω q)1

2

- ωq + iΓq

∑

)

Rangwalla et al. deviation of σ (see note).36 For our calculations, the sum in eq 12 only needed evaluation in a limited range, which, at maximum, was from n ) -3 to +3.

βijk(ω2, Ω)

molecules

) N〈βijk(ω2, Ω)〉

3. Results and Discussion

∫

) N βijk(ω2, Ω) f(Ω) dΩ

(5)

Here, N is the total number of such molecules, 〈〉 indicates an ensemble average, and f(Ω) is the probability distribution function of the molecular orientation as a function of Ω. Substituting from eqs 4 and 3 in eq 5 gives

∫ ∑

Aijk,q ) N [

Uijk:lmn(Ω) βlmn,q]f(Ω) dΩ

(6)

l,m,n)a,b,c

The groups found at the interfaces of PVNODC include methyl (-CH3, or CH3 for simplicity) and methylene (-CH2-, or CH2 for simplicity). The abc axes are rigidly fixed to these groups so that the c axis coincides with the C3 (CH3) and C2 (CH2) symmetry axes. Orientations of these groups are assumed to be isotropically distributed in the φ and ψ angles (the interfaces have azimuthal isotropy). Assuming the CH3 groups at the interface belong to the C3v symmetry point group and using the transformation coefficients presented by Hirose et al.,29 the CH3 Aijk,q’s relevant to this work are as follows:34

Ayyz,r+ )

[

〈cos θ〉 - 〈cos3 θ〉 + βaac,r+〈cos θ〉 2

NCH3 (βccc,r+ - βaac,r+)

]

(7)

Ayyz,r- ) NCH3[-βcaa,r-(〈cos θ〉 - 〈cos3 θ〉)]

(8)

Ayzy,r- ) NCH3(βcaa,r-〈cos3 θ〉)

(9)

Here, r+ (CH3 symmetric stretch, notation explained in section 3.1) is the A1 symmetry species, nondegenerate, vibrational mode, and r- (CH3 asymmetric stretch) is the E symmetry species, degenerate mode. θ is the tilt angle of the c axis from the surface normal (z axis), and NCH3 is the number of CH3 groups contributing to the SFG signal. The ensemble averages appearing in the above equations were obtained using the method outlined by Simpson and Rowlen (Supporting Information, ref 35).

〈cos θ〉 ) K

∫

π

〈cos3 θ〉 ) K

∫

π

0

0

cos θ f ′(θ) sin θ dθ

(10)

cos3 θ f ′(θ) sin θ dθ

(11)

∫ f ′(θ) sin θ dθ]

K)[

π

-1

0

Here, K is a normalization constant and f ′(θ) is the net angular distribution function defined as follows:

3.1. Room-Temperature Measurements. Peak Assignments. Figure 3 shows SFG spectra and fits for interfaces of PVNODC with water and air, and of the OTS SAM with water and air. For each interface, spectra in four polarizationss ssp, sps, pss (not shown), and ppp (not shown)swere simultaneously fit to eq 2, keeping ωq, Γq, and Φ as common fit parameters for all four spectra. The resonances detected in these spectra, which cover a range of 2800-3000 cm-1, represent the C-H-stretch vibrations of the CH3 and CH2 groups present at the interfaces. Most of these peaks are visible in the PVNODC-water spectra. These peaks are assigned to vibrational modes, and the modes symbolically represented,37 as follows: 2873 cm-1 (Figure 3A) is the CH3 symmetric stretch (r+); 2931 cm-1 (3A), the Fermi resonance between the symmetric-stretch fundamental and an overtone of a -1 bending mode of CH3 (r+ (3B), the CH3 FR); 2960 cm asymmetric stretch (r ); and 2853 and 2860 cm-1 (3A), the CH2 (d+) and terminal CH2 (d+ ω in ref 38) symmetric stretches, respectively. The CH3 peaks in this list are also visible in the spectra of the other interfaces in Figure 3 (PVNODC-air and OTS interfaces); the CH2 peaks are weak and indiscernible in these spectra, except in the PVNODC-air spectrum (3A) where a weak d+ band can be seen centered at 2849 cm-1. The ωq’s of these peaks obtained for OTS-SAM interfaces (3C,D) were within (3 cm-1 of the same peak in the corresponding PVNODC interface. Furthermore, note that in Figure 3, all the CH3 -1 peakssr+, r+ FR, and r sare red-shifted (3-10 cm ) at the water interface as compared to the corresponding air interface; the CH2 peakssd+ and d+ ω sshow a blue shift (3-7 cm-1) instead. The red shifts of the CH3 group have been previously observed for OTS SAMs exposed to methanol and CCl4 (ref 22); these shifts are attributed to a possible interaction of the methyl group with the environment and, in part, to a possible local-field effect. The blue shifts of the d+ and d+ ω are in agreement with the increased frequencies observed for methylene in disordered (discussed below) systems.39 Interfacial Structure of the Solid Phase. In Figure 3A,C, the air-interface spectrum has a strong r+ peak compared to a weak d+ peak. This indicates that the alkyl side chains of PVNODC23 and the alkyl tails of OTS molecules41 are in predominantly all-trans conformations with the chain axes oriented such that the CH3 groups face the air. CH3 groups are preferred at the air interface because they reduce the surface energy of these substances. For PVNODC, the ordered side chains at the surface are within crystalline domains as suggested by the following facts: PVNODC has side-chain crystallinity in the bulk, and the poly(octadecyl acrylate) polymer studied by Gautam

+∞

f ′(θ) )

∑ f(2πn + θ) + f(2πn - θ)

(12)

n)-∞

f (θ) )

2 2 1 e-(θ-θ0) /(2σ ) x2πσ

(13)

f (θ) is a Gaussian function with a mean of θ0 and a standard (34) Watanabe, N.; Yamamoto, H.; Wada, A.; Domen, K.; Hirose, C. Spectrochim. Acta, Part A 1994, 50, 1529-1537. (35) Simpson, G. J.; Rowlen, K. L. J. Am. Chem. Soc. 1999, 121, 2635-2636.

(36) When the meaning is unambiguous, σ, angular distribution, and distribution shall be used synonymously in the text. A broader (narrower) distribution will mean a higher (lower) σ. In the limit σ f 0, f(θ) will be the Dirac delta function, δ(θ - θ0), and the distribution will be called the δ distribution. (37) Snyder, R. G.; Strauss, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86, 5145-5150. (38) Goates, S. R.; Schofield, D. A.; Bain, C. D. Langmuir 1999, 15, 1400-1409. (39) Rothberg, L.; Higashi, G. S.; Allara, D. L.; Garoff, S. Chem. Phys. Lett. 1987, 133, 67-72. (40) Du, Q.; Freysz, E.; Shen, Y. R. Science 1994, 264, 836-828. (41) Guyot-Sionnest, P.; Hunt, J. H.; Shen, Y. R. Phys. Rev. Lett. 1987, 59, 1597-1600.

Alkyl-Side-Chain Polymer-Water Interface Structure

Langmuir, Vol. 20, No. 20, 2004 8629

Figure 3. SFG spectra at room temperature for the water interface (0) and the air interface (O) of (A) PVNODC (ssp polarization), (B) PVNODC (sps), (C) OTS SAM (ssp), and (D) OTS SAM (sps). The solid and dashed lines are fits to eq 2 of the water-interface and air-interface spectra, respectively. For the interfaces with water, the SFG spectra were taken and fit over an extended range of 2800-3550 cm-1 to account for contributions from the broad O-H vibrational bands.40

Figure 4. Cartoon of a model for the PVNODC surface showing a cross section of the PVNODC-air interface (left part) and its transition to a PVNODC-water interface (right part). The crosssectional plane is perpendicular to the polymer surface.

et al.,42 which also has 18-carbon-long side chains connected to a linear backbone (similar to PVNODC), shows side-chain crystallinity at the surface. In addition, orientation analysis (discussed below) of the CH3 groups shows that the axes of the side chains are nearly perpendicular to the surface. On the basis of this knowledge, a model of the PVNODC surface at the air interface is shown in Figure 4 (left part). The surface is assumed to have crystalline domains of side chains enclosed by less-ordered, grain-boundary regions; this is reasonable considering the surface is that of a semicrystalline polymer, which, by definition, must have a high degree of imperfection in its crystallinity. The cross section in the figure is at a typical grain boundary; the yellow, crosshatched section to the left is the grain boundary and the ordered side chains to the right are the crystalline domain. The boundary region could be, for example, an amorphous phase, a phase of intermediate (between amorphous and crystalline) order, a crystal defect that signals a change in the orientation of the crystallographic axis, or some combination of the above. The most striking feature of the PVNODC-water spectrum in Figure 3A, compared to the PVNODC-air (42) Gautam, K. S.; Kumar, S.; Wermeille, D.; Robinson, D.; Dhinojwala, A. Phys. Rev. Lett. 2003, 90, 215501/1-4.

spectrum, is the increase in peak intensities of the d+ and d+ ω peaks. This indicates a restructuringsincreased conformational-disordersat the water interface of a significant fraction of previously ordered side chains, resulting in increased gauche defects. Such defects break the inversion symmetry of the CH2 units and make them SFGactive.41 This restructuring of the alkyl side chains at the water interface results in a greater density of the more hydrophilic CH2 groups in direct contact with water, thereby contributing to a reduction of interfacial energy, which is the driving force for the disorder observed. The conformational changes leading to gauche defects increase the footprint area of side chains at the polymer surface, which is possible because the crystalline domains of closepacked side chains are separated by less-ordered grainboundary regions. For example, the red side chains (Figure 4), which are within the crystalline domain but near the grain boundary, will have the necessary space (within the less dense, neighboring grain-boundary region) to undergo conformational disorder. This is illustrated in Figure 4 (right part). The d+ and d+ ω peaks that appear in the water-interface spectrum of Figure 3A could be a result of the gauche defects formed in these outlying side chains. Furthermore, the level of disorder shown requires a small number of bond rotations (within the outlying side chains) and no translational rearrangements of the polymer backbone; this is consistent with the reversible nature of the PVNODC surface restructuring at room temperature (discussed below). Moreover, the persistent strength of the r+ and r+ FR peaks in the ssp spectrum, after the surface is in contact with water, indicates that the remaining fraction of side chains still has their original air-interface structure. This is depicted by the black side chains (right part) that remain unperturbed. Also, the additional disorder introduced by the conformational changes is expected to laterally extend grain-boundary regions and increase the area fraction of the less-ordered surface. Over time, some amount of water is expected to diffuse into the grain-boundary regions, as shown by a light blue shading of the crosshatched region (right part). Therefore, our SFG

8630

Langmuir, Vol. 20, No. 20, 2004

Figure 5. SFG spectra (ssp polarization) of the PVNODC polymer interfaces: before (O) exposure to water (PVNODCair), during exposure (×) to water (PVNODC-water), and after water was removed (0) and the surface was dried (PVNODCair). The solid lines are a guide to the eye. The × and 0 spectra were scaled to fit the graph.

results suggest that the PVNODC surface is heterogeneous under water. It has lower-wettability (crystalline domains) and higher-wettability (laterally extended grain boundaries) areas at the surface. In contrast, the water-interface spectrum for the OTS SAM (Figure 3C) shows almost no increase in the CH2 intensities as compared to its air-interface spectrum. This implies that the OTS SAM is close-packed with a uniform structure of alkyl chains on the surface, and in contact with water, the surface remains homogeneous. Reversibility of Conformational Disorder. Figure 5 shows that the molecular-level structural changes observed when the PVNODC surface is exposed to water are reversible. When the water is removed and the surface is dried in flowing N2, the air-interface spectrum almost completely overlaps the before-water spectrum. The high intensities seen in the d+ and d+ ω peaks at the PVNODC-water interface are lost, and the peaks shift back to the original frequencies seen in the before-water spectrum. This implies that the PVNODC surface structure reverts to that of the air-interface structure that was seen before exposure to water. This suggests an order-disorder mechanism that does not require gross translational rearrangement of the polymer but rather a relatively small number of bond rotations which can occur at room temperature. CH3 Orientation Analysis. The overwhelming majority of CH3 groups in the PVNODC molecules are the terminal CH3 of the alkyl side chains. The only other source of these groups may be the end groups of the polymer backbone if they contain any CH3 groups, a negligible fraction compared to the contribution from side chains (considering that ≈90% of the polymer backbone repeat units contain n-octadecyl side chains). Therefore, the terminal CH3 groups of side chains which are present at the interface are solely responsible for the methyl peaks in the SFG spectra. Figure 6 shows the |Ayzy,r-/Ayyz,r-| ratio, calculated and experimentally observed. Note that this ratio is independent of βlmn,q’s. The different, calculated curves are obtained by varying the angular distribution of tilt for CH3 groups. σ is a measure of the isotropy in the orientation of the CH3 groups in the θ Euler coordinate; in the other two Euler coordinates (φ and ψ) these groups are isotropically distributed over surface areas of macroscopic dimensions. The figure shows that the CH3 group is tilted by e36° at the PVNODC-air interface, whereas it is tilted by e34° at the PVNODC-water interface; a more precise determination would require determining σ, which is unknown; the values 36 and 34° are for the δ-distribution

Rangwalla et al.

Figure 6. Ratio |Ayzy,r-/Ayyz,r-| as a function of the CH3 tilt angle. The curves are calculated (|eq 9/eq 8|) plots for various values of σ (eq 13). The horizontal lines enclose a region of (1 standard deviation about the mean experimental ratio, obtained by fitting the spectra in Figure 3 to eq 2. Regions enclosed in thick boundaries are for the PVNODC interfaces, and thin boundaries are for the OTS SAM. Shaded regions are for water interfaces, and unshaded are for air.

Figure 7. SFG spectra (ssp polarization) at room temperature for the water interface (0) and the air interface (O) of PVNODC. The solid line is a fit to eq 2 of the water-interface spectrum.

limit. Although σ is unknown, we can conclude that it must be