Polarized Vibrational Spectroscopy of Fiber Polymers: Hydrogen

Biomacromolecules 2003, 4, 1589-1595

1589

Polarized Vibrational Spectroscopy of Fiber Polymers: Hydrogen Bonding in Cellulose II Adriana S ˇ turcova´ ,*,† Isabelle His,† Tim J. Wess,‡ Graeme Cameron,‡ and Michael C. Jarvis† Environmental, Agricultural and Analytical Chemistry, Chemistry Department, University of Glasgow, Glasgow G12 8QQ, Scotland, and Department of Biological Sciences, University of Stirling, Stirling FK9 4LA, Scotland Received August 12, 2003

Vibrational spectroscopy using polarized incident radiation can be used to determine the orientation of X-H bonds with respect to coordinates such as crystallographic axes. The adaptation of this approach to polymer fibers is described here. It requires spectral intensity to be quantified around a 180° range of polarization angles and not just recorded transversely and longitudinally as is normal in fiber spectroscopy. Mercerized cellulose II is used as an example. The unit cell of the cellulose II lattice contains six distinct hydroxyl groups engaged in a complex network of hydrogen bonds that hold the cellulose chains laterally together. A formalism is described to relate the variation in intensity of each O-H stretching mode to the angle between its transition moment and the chain axis as the polarization axis is rotated with respect to the fiber axis. It was necessary to include the effect of dispersion in chain orientation around the mean and the averaging of all rotational positions of the chains round their axis. The two crystallographically distinct O(2)-H groups, which are each hydrogen-bonded to only one acceptor oxygen, show a close match in orientation between the transition moments of their stretching bands and the O-H bond axis. The two O(3)-H groups each have a three-centered hydrogen bond to O-5 and O-6 of the next residue in the same chain. The transition moments of their stretching modes lay between the acceptor oxygens. Hydrogen bonding from the O(6)-H groups is still more complex but again the transition moment of each O-H bond lay within the cone of orientations described by the acceptor oxygens, provided that one additional acceptor oxygen excluded from the published crystal structure was considered. The transition moments for the O-H stretching modes were approximately aligned with the O-H bond axes, but the alignment was not necessarily exact. This approach is not restricted to hydroxyl groups, but it is particularly useful for the elucidation of hydrogen bonding in fibrous polymers for which crystallographic data on proton positions are not available. Introduction Polarized vibrational spectroscopy is a powerful tool for obtaining structural information on hydrogen-bonded solids because the polarization of the X-H stretching modes indicates the orientations of the hydrogen bonds. In straightforward cases in which no vibrational coupling is involved, the axis of the X-H bond and the transition moment of its stretching vibration approximately coincide. A wealth of structural detail on hydrogen-bonding patterns can potentially be obtained by polarized infrared and Raman spectroscopy of single crystals sectioned normal to the crystal axes. This approach has been adopted for crystal structures as diverse as gibbsite1 and sucrose.2 Its application to crystalline fructose3 was illustrative of considerable potential of polarized, crystallographically aligned vibrational spectroscopy but also illustrated the complexity of hydrogen bonding in carbohydrates, demonstrating such features as nonlinear * To whom correspondence should be addressed. Telephone: 44 141 330 6571. Fax: 44 141 330 4888. E-mail: [email protected]. † University of Glasgow. ‡ University of Stirling.

three-centered hydrogen bonds and donor-acceptor chains reminiscent of water.3 Here, we describe the adaptation of this approach for polymer fibers. Polarized vibrational spectroscopy is well established for fiber characterization, but it is normally used qualitatively rather than quantitatively with respect to polarization. That is, vibrational bands are normally described as either longitudinally or transversely polarized without specifying the degree of polarization as is necessary in singlecrystal experiments. This qualitative approach is exemplified by studies on cellulose, for which excellent longitudinally and transversely polarized infrared spectra were used in structural studies nearly half a century ago.4-8 The remarkable mechanical characteristics of cellulose depend on its network of intra- and intermolecular hydrogen bonds between hydroxyl groups. The vibrational modes of the O-H bonds are readily observed in the IR and Raman spectra. This leads to the possibility of computing macroscopic elastic moduli from the corresponding force constants and bond orientations9-11 and to the identification of the loadbearing features of the polymer structure through load-

10.1021/bm034295v CCC: $25.00 © 2003 American Chemical Society Published on Web 10/24/2003

1590

S ˇ turcova´ et al.

Biomacromolecules, Vol. 4, No. 6, 2003

induced changes in IR polarization12,13 or Raman band frequency,14 which, respectively, imply reorientation or stretching of the bond in question. Ordered fiber structures such as cellulose differ from conventional crystal lattices in two ways that are important here. First, although the crystalline domains in a polymer fiber are more or less aligned along the fiber axis, normal to that axis all orientations of the relevant crystal planes are generally possible. That is, an individual crystalline domain may be rotated to any position around the fiber axis. We use the term rotational aVeraging to describe this effect when integrated across the fiber. Second, the longitudinal alignment is not usually perfect. In the case of oriented films made from polymer fibers, the orientation is dispersed chiefly in the plane of the film. This is also the case for many natural materials that contain oriented biopolymers, like insect cuticles or plant cell walls. These differences from a conventional molecular crystal, and their effects on the polarization of the vibrational spectra, must be modeled before quantitative structural data can be extracted from such experiments on fibrous polymers. In this paper, we use cellulose II (mercerized, that is, alkali-treated, cellulose) as an example to test the utility of this approach. For a long time, there was disagreement on whether the cellulose II structure had parallel or antiparallel chains and on how its exocyclic (C-6) group was oriented.11,15 This was resolved by neutron-diffraction crystallography of regenerated cellulose II (rayon) fibers after deuteration16 to confirm the coordinates of all of the atoms in a two-chain monoclinic unit cell with alternate sheets of chains antiparallel and the gt-hydroxymethyl conformation dominant throughout. However, electron density not accounted for by this structure suggested that the chains at the origin of the unit cell also had an alternative hydrogen-bonding arrangement at the exocyclic group with about 30% occupancy, apparently having the tg-hydroxymethyl conformation. The X-ray diffraction pattern of mercerized cellulose17 was consistent with a very similar structure having minor differences in atomic coordinates (the hydroxyl proton coordinates were not determined in this case) and with low occupancy of the alternative tg conformation of the exocyclic group. Here, we describe the polarization of the OH stretching bands in mercerized cellulose II, comparing it with the crystallographically determined structures of mercerized and regenerated cellulose II.16,17 Polarized IR spectroscopy has long been used, in the qualitative sense described above, for band assignment in the vibrational spectrum of cellulose II.4,7,8 The principle of using the degree of polarization to estimate bond orientations was outlined by Marchessault and Liang,7 although no quantitative theory was then available. By starting from liquid-crystalline dispersions, one can prepare cellulose films with partial uniplanar orientation so that rotational averaging is reduced.6,7,16,18 In principle, uniplanar orientation should allow more detailed orientational information to be obtained, but the analysis of incomplete rotational averaging is complex, and this approach was not attempted in the present study. We used Fourier transform infrared (FTIR) microspectroscopy rather than conventional

Figure 1. Gaussian distributions of fiber orientation with varying width parameter Σ. Overlap between Gaussians centered on 0° and 180° results in the distribution becoming nearly uniform when Σ > 60°.

FTIR so that we could work with mercerized cellulose II prepared from individual flax fiber cells, in which the cellulose chains have no uniplanar orientation but are longitudinally very well-oriented. Theory Cartesian Coordinates. It is assumed that the chains lie in the (yz) plane, that the incident radiation is aligned with the x axis, and that its electric vector is aligned with the z axis. Dispersion in Chain Orientation. It is assumed that the chain orientations (σ + R) are normally distributed with standard deviation Σ around a mean angle R to the orientation of the electric vector of the incident radiation. Because angles R, (R - π), (R + π), and (R + 2π) are equivalent with respect to the interaction between the radiation and each vibrational mode, overlap between the normal distributions around all of these angles must be considered (Figure 1). Rotational Averaging. It is assumed that the transition moment of the vibrational mode in question is oriented at an angle θ to the chain axis and that its projection normal to the chain axis may take up any rotational position around the chain defined by the angle Ω to the direction of the incident radiation. Thus all values of Ω between zero and 2π are equally represented. Relative to its maximum, the absorbance due to the vibrational mode scales with Pz2, the square of the projection of the transition moment on the electric vector of the incident radiation. Thus, Pz2 ) D2(cos θ cos(σ + R) - sin θ sin Ω sin(σ + R))2 (1) where D is the normalized relative abundance of chains oriented at (σ + R) to the z axis. The effect of rotational averaging may be seen in Figure 2, where the predicted relative absorbance is plotted against the polarizer angle R for vibrational modes of which the transition moments are aligned at varying angles θ to the chain axis. The relative absorbance was calculated numerically for values of R, σ, θ, and Ω stepped at 1°, 5°, or 10° intervals over the appropriate range for Σ ) 10°, 20°, 30°, and 40°. For a vibrational mode with transition moment parallel to the chain axis (θ ) 0), the absorbance is maximal when the

Hydrogen Bonding in Cellulose II

Figure 2. Predicted variation with polarizer angle of the infrared absorbance due to a vibrational mode with transition moment oriented at an angle θ to the mean orientation of the chain axis. This family of curves is calculated for a relatively well-oriented fiber with Σ ) 10°. For broader distributions of chain orientation within the fiber, the curves become flatter.

radiation is polarized in the same direction (R ) 0). Likewise for a vibrational mode normal to the chain axis (θ ) π/2), the absorbance is maximal when the direction of polarization also is normal to the chain axis (R ) π/2), but the maximal absorbance is only half of that attained with (θ ) 0). This is because the transition moment vector can rotate around the chain axis and the interaction is only effective when it lies close to a plane normal to the incident radiation (Ω ) π/2). More precisely, under these conditions Pz2 reduces to D2 sin2 Ω and

∫02π D2 sin2 Ω. dΩ ) 0.5D2 A vibrational mode with its transition moment at 45° to the chain axis (θ ) π/4) does not, as might be expected, give equal absorbance values when the radiation is polarized parallel and normal to the mean chain orientation. Instead this happens at θ ) 54.7°. For a vibrational mode with its transition moment at that angle to the mean chain orientation, the absorbance is unaffected by the polarizer angle. This is the case whatever the value of Σ, that is, whatever the dispersion of chain orientations around the mean. The null value (θ ) 54.7°) is equal to the “magic angle” of sample rotation in solid-state NMR, which minimizes line broadening due to dipolar coupling and chemical shift anisotropy, an effect that is geometrically analogous to the rotational averaging effect described here. The polarizer angle also has a null value, in this case R ) 62.1°, at which the absorbance is independent of θ. This model can be used in a number of ways, for example, to calculate the dispersion of chain orientations around the mean from the polarization of an approximately longitudinal stretching mode with known θ. Here, however, we determined Σ independently by small-angle X-ray scattering. This allowed the observed variation in absorbance with polarizer angle to be matched against the predicted variation in absorbance for a bond with any value of θ. Experimental Section Preparation of Cellulose II. Flax (Linum usitatissimum L.) was grown at the Glasgow Botanic Garden. Phloem fibers were dissected by hand from main stems of mature plants with as little mechanical damage as possible. The noncel-

Biomacromolecules, Vol. 4, No. 6, 2003 1591

lulosic polysaccharides were acid-hydrolyzed using 1 M HCl for 1 h at 100 °C, followed by washing with water. The acidhydrolyzed cellulose I samples were converted to cellulose II by overnight mercerization in 18% (w/v) NaOH containing 1 mg/cm-3 NaBH4. Afterward, the samples were washed with 1 M acetic acid and then with water and acetone. Small-Angle X-ray Scattering. Small-angle X-ray scattering of dry cellulose samples was carried out on ID02 at the European Synchrotron Radiation Facility, Grenoble, France, using a 1 m sample to detector distance. Images were collected over 0.5 s on a Thomson X-ray intensifier (TH 49-427) lens coupled to a FReLoN CCD camera (2048 pixels × 2048 pixels). This detector has an active area of size 180 mm and a frame rate of 14 images (1024 pixels × 1024 pixels) per second with a 14 bit nominal dynamic range. The wavelength of X-rays used was 0.1 nm. X-ray diffraction images were corrected for dark field current and spatial distortions. The angular distribution of equatorial scattering was estimated by use of in-house software. Embedding and Preparation of Thin Sections. The cellulose II samples were fixed for microscopy with 0.1 M phosphate buffer (pH 7.2) containing 2% (w/v) paraformaldehyde and 0.5% (v/v) glutaraldehyde for 90 min. This was followed by rinsing in 0.2 M cacodylate buffer and then three washes of double-distilled water. After this step, the samples were dehydrated in an aqueous ethanol seriess10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, and 100%s at least 30 min at each concentration. Dehydration was completed by two incubations in propylene oxide lasting 60 min each. The dehydrated samples were infiltrated with LR White/propylene oxide mixture at resin concentrations of 10%, 30%, 50%, 70%, and 90% for several hours at each concentration and then several incubations in 100% LR White for 14 weeks.19 The infiltrated samples were embedded in capsules and polymerized at 60 °C for 24 h. Sections 1-5 µm in thickness were cut with a diamond knife and transferred to BaF2 windows. FTIR Microspectroscopy. FTIR spectra were obtained on a Nicolet Nexus spectrometer equipped with a Nicolet Continuum microscope attachment with a liquid nitrogen cooled MCT detector and a single ZnSe wire grid polarizer. The scanning parameters used were as follows: resolution, 4 cm-1; number of scans, 128. Aperture sizes varied according to the dimensions of the section but were always at least 25 µm in each dimension to minimize distortion of the spectra by scattering effects. Results Small-Angle X-ray Scattering. After removal of noncellulosic polysaccharides and conversion to cellulose II (mercerization), the secondary walls of carefully isolated flax fiber cells retained their original form and the cellulose chains were still oriented predominantly along the longitudinal axis of the cells. Before mercerization, the half-width at halfmaximum (HWHM) of the distribution of chain orientations in a narrow bundle of dry fibers was 7.0° when measured by small-angle X-ray scattering (SAXS). Comparison with single fiber cells as used for FTIR microscopy and with literature values is complicated by the fact that misorientation

1592


of cells within a bundle20 and hydration21 both increase the width of the distribution. HWHM values of 3.5° and 6.4° have been reported20,22 for single fiber cells measured by microfocus SAXS, 7.5° and 6.3° for wet fiber bundles,20,21 and 5.5° for fiber bundles at a low degree of hydration.21 The HWHM of bundled dry fibers in our SAXS study increased to 11° on mercerization. The standard deviation (Σ) of a Gaussian distribution can be obtained from the HWHM by division by 1.18, so this value corresponds to Σ ) 9.3°. We therefore assume that Σ for cellulose II chains in a single fiber cell did not exceed 10°, and this value was used in calculating the predicted polarization of the FTIR spectra. There is little variation of predicted polarized intensities with Σ in the range 0°-10°. FTIR Microspectroscopy of Resin-Embedded Thin Sections. The fiber cells were exhaustively resin-embedded to permit longitudinal thin sectioning, which permitted greatly improved spectral quality compared to intact flax cells. In particular, thin sectioning prevented saturation of the strongly absorbing hydroxyl stretching modes which, in thicker samples, distorted the quantitative measurement of dichroism. The spectral contribution of the resin had to be subtracted from that of cellulose. This was done after scaling the resin spectrum to eliminate its characteristic features from the resin-corrected sample without introducing negative bands. Resin subtraction was complicated by the fact that the carbonyl stretching band at 1733 cm-1 in the spectrum of pure resin was shifted to 1725 cm-1 in regions occupied by cellulose and became weakly polarized (data not shown). Because the nonembedded cellulose gave essentially no absorbance close to this frequency and because there was no polarization at any frequency in the spectrum of pure resin, it was inferred that a small number of ester bonds between resin and cellulose were responsible for the 1725 cm-1 spectral feature and provided the mechanical stability at the cellulose-resin interface necessary to allow sectioning of the embedded fibers. FTIR spectra with the electric vector of the incident polarized radiation parallel and transverse to the fiber axis are shown in Figure 3. Polarization at the Glycosidic Bond. The antisymmetric stretching mode of the glycosidic linkage (C1-O1-C4), which absorbs at about 1159 cm-1, is often used to estimate the microfibril orientation because it is approximately longitudinal and is well resolved from other bands in the vibrational spectrum of cellulose.23 The observed variation in the intensity of the 1159 cm-1 band with the polarization angle R is shown in Figure 4. The intensity of this band was determined using a baseline correction method where a straight baseline was placed at the trough on the higherwavenumber side of the 1159 cm-1 peak, that is, at approximately 1188 cm-1. A similar method of baseline correction was used by Rodrigues et al.24 The value of Σ ) 10° assumed from the SAXS data was used to calculate the predicted variation in the intensity of this band with polarization angle R for a range of values of θ. Comparison of the observed and predicted intensities indicated that the angle θ between the transition dipole moment for this mode and the chain axis is equal to 28° and the transition moment


Figure 3. FTIR spectra of a thin section of a single mercerized flax fiber cell with the electric vector of the incident radiation polarized parallel and transverse to the fiber axis. The hydroxyl stretching region of each spectrum is deconvoluted (thin lines) into six overlapping bands corresponding to the six hydroxyl groups in the unit cell. The frequency and width of each band were kept fixed between the two spectra.

Figure 4. Variation of the 1159 cm-1 band intensity with polarizer angle R in the FTIR spectrum of cellulose II. Observed relative absorbances are superimposed on the theoretical curve calculated for a Σ value of 10° describing the distribution of chain orientations and the best-fit value of the parameter θ describing the orientation of the transition moment with respect to the chain axis.

is not, as might be assumed, exactly parallel to the axis of the glycosidic linkage. The use of two alternative baseline correction methods gave values of θ that were consistent to within 1°. Deconvolution of the Hydroxyl Stretching Region. Prior to deconvolution, it was necessary to baseline-correct the hydroxyl stretching region of the spectra. Elevated baselines in infrared spectra of solid films are normal and can be attributed25 to (a) scattering caused by inhomogeneities in the solid, (b) improper orientation of the sample, (c) interfering substances, (d) the Christiansen effect, or (e) nonspecific absorption. For quantification of dichroic behavior in the infrared, the ideal is accurate baseline correction, and it is certainly a requirement that baseline correction is reproducible. We effectively eliminated contributions b and c, but adequate quantitative theory for applying corrections for the remaining sources of baseline intensity was not available. We therefore chose simple linear, rather than polynomial, baseline correction methods making as few assumptions as possible. We tested two baseline functions:


Hydrogen Bonding in Cellulose II Table 1. The Values of the Angle θ Corresponding to the Angular Dependence of the Intensity of the OH Bands Found to Best Describe the Observed Angular Dependence (See Also Figure 5) peak position [cm-1]

peak width parameter [cm-1]

θ (deg)

3489 3453 3445 3362 3298 3165

11.4 91.1 10.2 66.5 84.8 79.7

20 57 10 53 59 62

(1) a straight line with positive slope between 3710 and 2999 cm-1 and (2) a straight line between 3710 and 1879 cm-1. These gave almost identical quantitative results. Deconvolution of the hydroxyl stretching region was performed by a semiautomated routine in Origin using spectra for (R ) 0) and (R ) π/2) simultaneously. The starting assumptions were (a) that the six hydroxyl groups in the unit cell, as previously suggested,7 each gave rise to one stretching band of which the frequency and width were independent of polarization, that is, that coupling between vibrational modes could to an approximation be neglected in this part of the spectrum,26 (b) that each band shape conformed to a Gaussian function, and (c) that unpolarized band intensity and bandwidth increased linearly with displacement to lower frequency than that of hydroxyls without hydrogen bonding, namely, 3640 cm-1. The data were consistent with assumptions a and b, but considerable refinement of the initial fits was possible by omitting assumption c and fitting intensities and bandwidths individually for each band, because the two bands centered on 3445 and 3498 cm-1 were much narrower than the assumed model. The best-fit parameters modeling the experimental spectra are presented in Table 1. The variation with polarizer angle R in the intensity of each the six deconvoluted bands is shown in Figure 5. In the case of each band, the experimentally determined dichroic dependence of intensity on the angle R was compared with its predicted dichroic dependence for values of θ varying from 0° to 90°. (The angle θ defines the orientation of the transition dipole moment relative to the chain axis, see above). The Gaussian distribution parameter Σ was set at 10° throughout as assumed from SAXS. The angle θ giving the predicted dichroic dependence of the peak intensity that best described the observations is presented in Table 1, and the predicted dichroic dependences are included in Figure 5. Discussion The best-fit values of θ (Table 1) may be compared with the orientation of the hydrogen-bonded hydroxyl groups as predicted from the crystallographic data on cellulose II in Table 2. The comparison is complicated by the fact that proton coordinates are available only for regenerated cellulose II.16 The orientations of the O‚‚‚O vectors are close, although not identical, in mercerized17 and regenerated16 cellulose, which suggests that the pattern of hydrogen bonding in the two forms of cellulose II is similar, in

Figure 5. Variation of deconvoluted band intensity with polarizer angle for each of the six deconvoluted hydroxyl stretching bands of cellulose II. Observed relative absorbances are superimposed on theoretical curves with the best-fit values of the parameter θ describing the orientation of the transition moment with respect to the chain axis.

particular, that O2 is the donor to O6 in the origin chains but the acceptor from O6 in the center chains. This pattern could not have been identified without the proton coordinates for regenerated cellulose II.16 The comparison between the FTIR and crystallographic data clearly identifies the longitudinally polarized bands at 3498 and 3445 cm-1 with O(3)H

1594



Figure 6. The hydrogen-bonding pattern in mercerized cellulose II based on the published carbon and oxygen atom coordinates17 and the donor-acceptor relationships demonstrated for regenerated cellulose II:16 (A) hydrogen bonding between the origin and center sheets of chains; (B) intra- and intermolecular hydrogen bonding within the origin sheet of chains; (C) intra- and intermolecular hydrogen bonding within the center sheet of chains. Superimposed (blue arrows) are transition moment vectors parallel to the chain axis (θ ) 0°) for the O3-H stretching mode in each chain and transition moment vectors at (θ ) 60°) for the O2-H and O6-H stretching modes, projected in each case into the most appropriate Cartesian plane.

of the origin and center chains as previously suggested,7,9 but it is not possible to distinguish between them. In their anomalously narrow bandwidth and approximate frequency, these vibrational modes resemble the stretching vibration of the unique O(3)H bond in cellulose III1, in which the asymmetric unit contains one glucose residue only.27 The transition moments of the 3498 and 3445 cm-1 bands and the O(3)-H bond orientations in the regenerated cellulose structure both lie between the O(3′)‚‚‚O(5) and O(3′)‚‚‚O(6) vectors that define the three-centered hydrogen bonds from O3 in the origin and center chains. The transition moments corresponding to the other four deconvoluted bands are all calculated to be oriented at approximately 60° to the chain axis. This is in good agreement with the origin- and center-chain O(2)H bond

orientations in regenerated cellulose II,16 which are locked by simple, two-centered hydrogen bonds (Table 2). The coordinates for the donor and acceptor oxygen atoms in mercerized cellulose II17 suggest that this hydrogen-bonding pattern is retained because there are no additional oxygens suitably placed to act as donors in the mercerized cellulose structure. However, there is a discrepancy in orientation between the remaining two deconvoluted bands, calculated to have θ ≈ 60°, and the O(6)H bonds, which are oriented almost exactly transverse to the chain axis of regenerated cellulose II.16 The discrepancy is not likely to be due to uncertainty in the calculation of θ from the polarized intensity curves because their shape is particularly sensitive to the influence of θ when θ is close to the null point of 62.1° (Figure 5).


Hydrogen Bonding in Cellulose II Table 2. Hydrogen Bonds Predicted from the Coordinates of Regenerated Cellulose16 and Mercerized Cellulose17 a

O‚‚‚O distance, Å mercerized regenerated donor acceptor cellulose cellulose O6oH O6oH O6oH O6cH O6cH O3oH O3oH O3cH O3cH O2oH O2cH

O6c O3c O5c O2c O3c O5o O6o O5c O6c O6o O2o

2.83 2.93 3.11 2.59 3.64 2.78 3.36 2.56 3.26 2.84 3.01

2.64 3.05 3.12 2.68 3.48 2.66 3.31 2.73 3.22 2.71 2.78

orientation relative to chain axis, deg mercerized regenerated cellulose cellulose O-O 54 69 66 84 48 30 25 27 34 77 75

O-O O-H 54 65 69 81 48 29 28 29 27 79 77

90 90 90 85 85 29 29 10 10 60 57

a Subscripts o and c denote, respectively, origin and center chains of the unit cell. The structure for mercerized cellulose does not include proton positions.

In the center chain of regenerated cellulose II, O(6)H is recorded16 as forming only one hydrogen bond, that to O2 in an adjacent chain; its orientation is close to transverse. However, a further hydrogen bond to O3 of the adjacent chain can be predicted from the published atomic coordinates.16,17 Having an O-O distance of 3.48 Å in regenerated cellulose II, it is too long to be included in the published list of hydrogen bonds,16 but the observed transition moment vector can be placed between the orientations of the principal hydrogen bond to O2′′ and this additional putative hydrogen bond (Figure 6). In the case of O(6)H of the origin chain, there is known to be a difference between the regenerated and mercerized cellulose structures.16,17 The neutron diffraction data for regenerated cellulose II seem to imply a four-centered hydrogen bond with O6, O5, and O3 of the center chain as acceptors.16 However, it was suggested17 that the crystallographic data might instead reflect two alternative hydrogenbonding arrangements in regenerated cellulose with O6 of the center chain functioning as an acceptor only in the more highly occupied gt conformation. The O-H bond orientation in regenerated cellulose16 differs by about 30° from the orientation of the transition moment in mercerized cellulose, although both lie within the cone defined by these three acceptors. In mercerized cellulose, with little occupancy of the alternate tg conformation the principal acceptor may be assumed to be O6 in the center chain.17 This would be consistent with a less transverse orientation of the donor O(6)H bond in the origin chain (Figure 6). The results of these experiments show that the orientation of the transition moments for hydroxyl stretching bands can be determined in fibers despite the complexities introduced by rotational averaging and imperfect alignment of the chains. This provides useful insights into structure and into the patterns of hydrogen bonding that define the structures of polysaccharides. It may also prove useful for synthetic fiber polymers.

However with polysaccharides this information must be used with care because of their propensity for forming multiple hydrogen bonds16 and because of the complex interplay between strong and weak hydrogen bonds in determining both the equilibrium proton position and the directionality of the OH stretching vibration.3,16 In such cases, we would not necessarily expect the OH bond orientation and the transition moment vector to coincide. It is more likely that weaker hydrogen bonds modulate the influence of the strongest hydrogen bond on the OH bond orientation and on the transition moment vector in different ways. The former is classically considered as static (but see ref 28), while the latter is dynamic. Vibrational spectroscopy should not, therefore, be seen as a method for the exact prediction of static structure as obtained by crystallography; but that does not affect its utility for predicting the hydrogen bonding patterns that hold cellulose and related polymer fibers together. Acknowledgment. This work was financially supported by grants from BBSRC, ESRF, and the EU. We are grateful to Prof. L. Barron (Glasgow University) for theoretical advice and Manfred Roessle (ESRF) for data collection. References and Notes (1) Wang, S. L.; Johnston, C. T. Am. Mineral. 2000, 85, 739. (2) Germanska, J.; Szostak, M. M. J. Raman Spectrosc. 1991, 22, 107. (3) Baran, J.; Rataczak, H.; Lutz, E. T. G.; Verhaegh, N.; Luinge, H. J.; van der Maas, J. H. J. Mol. Struct. 1994, 326, 109. (4) Marrinan, H. J.; Mann, J. J. Polym. Sci. 1956, 21, 301-311. (5) Tsuboi, M. J. Polym. Sci. 1957, 25, 159-171. (6) Liang, C. Y.; Marchessault, R. H. J. Polym. Sci. 1959, 37, 385395. (7) Marchessault, R. H.; Liang, C. Y. J. Polym. Sci. 1960, 43, 71-84. (8) Liang, C. Y.; Marchessault, R. H. J. Polym. Sci. 1960, 43, 85-100. (9) Tashiro, K.; Kobayashi, M. Polymer 1991, 32, 1516-1526. (10) Ganster, J.; Blackwell, J. J. Mol. Model. 1996, 2, 278-285. (11) Kroon-Batenburg, L. M. J.; Kroon, J. Glycoconjugate J. 1997, 14, 677. (12) Hinterstoisser, B.; Salmeń, L. Vib. Spectrosc. 2000, 22, 111-118. (13) Hinterstoisser, B.; Åkerholm, M.; Salmeń, L. Carbohydr. Res. 2001, 334, 27-37. (14) Eichhorn, S. J.; Young, R. J.; Yeh, W. Y. Text. Res. J. 2001, 71, 121-129. (15) Bernet, B.; Xu, J.; Vasella, A. HelV. Chim. Acta 2000, 83, 2072. (16) Langan, P.; Nishiyama, Y.; Chanzy, H. J. Am. Chem. Soc. 1999, 121, 9940. (17) Langan, P.; Nishiyama, Y.; Chanzy, H. Biomacromolecules 2001, 2, 410. (18) Marećhal, Y.; Chanzy, H. J. Mol. Struct. 2000, 523, 183-196. (19) His, I.; Andeme-Onzighi, C.; Morvan, C.; Driouich, A. J. Histochem. Cytochem. 2001, 49, 1525. (20) Mu¨ller, M.; Czihak, C.; Vogl, G.; Fratzl, P.; Schober, H.; Riekel, C. Macromolecules 1998, 31, 3953-3957. (21) Astley, O. M.; Donald, A. M. Biomacromolecules 2001, 2, 672680. (22) Mu¨ller, M.; Czihak, C.; Burghammer, M.; Riekel, C. J. Appl. Crystallogr. 2000, 33, 817-819. (23) Jarvis, M. C.; McCann, M. C. Plant Biochem. Physiol. 2000, 38, 1. (24) Rodrigues, J.; Faix, O.; Pereira, H. Holzforschung 1998, 52, 46. (25) Koch, A.; Weber, J.-V. Appl. Spectrosc. 1998, 52, 970. (26) Cael, J. J.; Gardner, K. H.; Koenig, J. L.; Blackwell, J. J. Chem. Phys. 1975, 62, 1145-1153. (27) Wada, M.; Heux, L.; Isogai, A.; Nishiyama, Y.; Chanzy, H.: Sugiyama, J. Macromolecules 2001, 34, 1237. (28) It is conceivable that a cooperative proton quantum tunnelling mechanism (Clary, D. C.; Benoit, D. M.; Van Mourik, T. Acc. Chem. Res. 2000, 33, 441-447) might connect the alternative hydrogenbonding arrangements suggested for regenerated cellulose.16

BM034295V

Polarized Vibrational Spectroscopy of Fiber Polymers: Hydrogen

Recommend Documents