Article pubs.acs.org/Biomac
Probing Alignment and Phase Behavior in Intact Wood Cell Walls Using 2H NMR Spectroscopy Sudip Chowdhury,†,‡ Louis A. Madsen,*,†,§ and Charles E. Frazier†,‡ †
Macromolecules and Interfaces Institute, ‡Department of Wood Science and Forest Products, and §Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061, United States ABSTRACT: This study presents the first application of 2H NMR spectroscopy to quantify lignocellulose matrix orientation, and it demonstrates the ability to separately investigate oriented and unoriented amorphous domains in intact natural plant tissue. Matrix orientation is evaluated using NMR quadrupolar interactions in small deuterated probe molecules absorbed into bulk Liriodendron tulipifera sapwood. Ethylene glycol-d4 deuterium spectra reveal two distinct amorphous domains, a highly oriented phase in the secondary wall S2 layer and an isotropic domain probably reflecting the compound middle lamella (CML). The oriented and isotropic signal areas exhibit thermally reversible changes, postulated to reflect probe redistribution between the S2 layer and the CML. Preliminary studies on a more powerful wood swelling agent, N,N-dimethylformamide-d1, are also discussed. This 2H NMR technique provides a new avenue for analysis and understanding of lignocellulose ultrastructure and promises to create new insights in correlating biomass processing with morphological change.
■
INTRODUCTION The elucidation of lignocellulose structure and organization has occupied scientists for 175 or more years. This persistent effort was born from at least two justifications: to optimize the conversion of lignocellulose into useful products, with recent emphasis on renewable energy, and to seek inspiration for hierarchical design strategies in new materials. Biomass breakdown for renewable energy presents a particularly relevant need, and requires finer knowledge of lignocellulose organization to optimize fuel and materials production. Evolved plant structures exhibit a structural complexity that defies any single analytical method. Consequently, advances in our understanding of lignocellulose structure and organization have been hard fought, the recent highlights of which are reviewed by Salmén and Burgert,1 in reference to xylem or wood. In this paper, we explore how cellulose fibrils impact orientation among the noncrystalline polymers − hemicelluloses and lignin. In xylem tissue, most cellulose fibrils are nearly parallel to the growing stem. Hemicelluloses associate with cellulose fibrils to various degrees, and in both hardwoods and softwoods hemicelluloses appear to be aligned with the fibrils.2 In contrast, observations of lignin orientation have varied. Using Raman spectroscopy, Attala and Agarwal reported evidence for lignin orientation in the secondary cell wall of Picea marinara.3 However, in another softwood, Picea abies, infrared techniques revealed only slight lignin orientation.4−6 The situation for hardwoods apparently differs substantially. Olsson et al. and Salmén et al. reported that lignin chains are highly oriented along the microfibrils in aspen (Populus tremula × Populus tremuloides) and maple (Acer spp.).6,7 More observations are required to determine the taxonomic levels over which structural motifs substantially vary. © 2012 American Chemical Society
Here we present a novel technique to probe intact wood structure using deuterium nuclear magnetic resonance (2H NMR) of small deuterium-labeled probe molecules absorbed into bulk wood. This strategy has been fruitfully employed to reveal orientational ordering and phase dynamics in a wide range of liquid crystals and synthetic polymers.8−11 When doped into an oriented matrix, the probe “inherits” partial ordering due to nematic-like short-range coupling with the matrix.9−11 Probe deuterons reveal this orientational bias through a NMR quadrupolar coupling, detected as a splitting, ΔνQ, of the 2H spectral signal (see Figure 1). Whereas, in an isotropic matrix, deuterated probes rapidly sample all possible orientations such that the quadrupolar coupling averages to zero. Xylem tissue is composed of cells containing multiple concentric layers, so deuterated probes will partition among these layers. This work focuses on the two cellular layers that comprise the overwhelming bulk of wood, the S2 layer of the secondary wall (about 70−80% of the total mass)12−14 and the compound middle lamella, CML, about 8−15% of the total wood mass12 (the CML is composed of the primary wall and the middle lamella). Most cellulose fibrils occupy the S2 layer, and deuterated probes in the surrounding noncrystalline matrix are expected to exhibit a quadrupolar splitting that reflects matrix orientation. In contrast, the CML contains a high lignin concentration with few cellulose fibrils that are randomly oriented, around which we might expect less order than in the S2 layer. In contrast to infrared and Raman techniques that probe specific structures through their unique absorptions, this Received: December 12, 2011 Revised: February 10, 2012 Published: February 21, 2012 1043
dx.doi.org/10.1021/bm201770q | Biomacromolecules 2012, 13, 1043−1050
Biomacromolecules
Article
degrees, which was the basis for their selection. Wood swelling in EG is about that seen in water, whereas swelling in DMF is 25−50% greater.17 The methods presented form the basis of a new characterization strategy for plant materials, in either intact tissue or deconstructed components. Novel perspectives on lignocellulose ultrastructure will provide new pathways to maximizing production of fuels and renewable material from biomass.
■
Materials. All specimens were machined from a single piece of commercial yellow-poplar (Liriodendron tulipifera) sapwood lumber (30 × 5 × 5 cm). On average, the diffuse-porous cross-section exhibited two growth rings per centimeter. Cubes, 2 and 5 mm, were machined with precise grain orientation. The 2 mm cubes were sampled exclusively from within early wood; early and late wood proportions in 5 mm cubes were not controlled. Specimen dimensions for self-diffusion coefficient measurements were 2 × 2 × 16 mm, with the long axis along the fiber direction. Two partially deuterated solvents were used: ethylene glycol-d4 (OH-(CD2)2-OH) and N,Ndimethylformamide-d1 (DCON(CH3)2), obtained from Cambridge Isotopes Laboratories, Inc. (Andover, MA) with a minimum purity of 99%. The self-diffusion coefficient of EG in wood was measured using anhydrous EG (C2H6O2), obtained from Sigma-Aldrich (99.8% purity). Sample Preparation. Specimens were vacuum-dried (1−5 mmHg, 48 h) at ambient temperature over anhydrous P2O5. After measuring the dry weight, specimens were saturated with EG-d4 or DMF-d1 using a room temperature vacuum/pressure treatment as follows: specimens were soaked in an excess of the deuterated solvent, then vacuum was applied (10−12 mmHg, 30 min), followed by atmospheric pressure (15 min). From this saturated condition, the solvent content of the specimens was reduced via controlled evaporation using a rotary evaporator (10 mmHg). The final solvent content of the specimens was targeted at 25% of dry wood weight (5 mm cube specimens for investigating the specimen size/heterogeneity effect had 30% solvent content). Within any sample group the solvent content variation was within 1%. After reaching the required solvent content and before NMR analyses, specimens were wrapped with Teflon tape and allowed to equilibrate for a minimum of 24 h at 4 °C in a 15 × 45 mm (OD × L) Teflon-capped glass vial. Angular Dependence of Orientational Order. Experiments were performed on a Bruker Avance III widebore NMR spectrometer at 9.39 T static magnetic field (B0), corresponding to 1H and 2H Larmor frequencies of 400 and 61.4 MHz, respectively. A singlechannel-detection static solids probe with a 10.7 mm ID horizontal solenoid coil was used. The 2 mm cube specimens were placed in a Teflon sample cell (9.5 × 8.5 × 40 mm OD × ID × L) with a 2.3 mm diameter and 2.3 mm deep slot that closely fits the specimen. A cylindrical Teflon plug eliminated the free space above the specimen to prevent solvent evaporation. Grain directions (longitudinal, L; radial, R; and tangential, T) denoted for each specimen were fixed parallel to B0. The sample cell was rotated in the horizontal NMR coil using a goniometer to control the angle between the grain orientation and B0 to within ±2°. The spin−lattice relaxation time (T1) of the 2H of EG-d4 in wood varied from 0.01 s at 0 °C to 0.2 s at 120 °C, as measured using inversion−recovery. The 2H NMR experimental parameters were: acquisition time = 0.05 s, relaxation delay = 0.1 s, pulse time = 11.5μs (π/2 pulse = 12 μs), spectral width = 25 kHz, number of scans per spectrum = 1000. No sample spinning was performed. For angular dependence studies, 2H NMR spectra were collected at 30 °C. The NMR probe temperature was calibrated to within ±1 °C using a pure dry EG standard. The specimens were reoriented in the radial plane from 0° (L direction parallel to B0) to 90° (R direction parallel to B0) in 10° steps, with spectra collected at each step. Spectra were then collected with the specimen’s tangential (T) direction parallel to B0. The experiment was repeated with three specimens to obtain three
Figure 1. Probe molecule deuterium NMR investigation of orientational order in wood. Top: Deuterated probe molecules in an isotropic matrix produce a singlet spectrum. Bottom: Deuterated probe molecules in an oriented matrix produce a doublet spectrum. 2
H NMR method is nonspecific and will reflect average orientation among all solvent-accessible noncrystalline polymers. Equation 1 describes the quadrupolar splitting, ΔνQ, of absorbed probe deuterons, which allows quantification of phase-specific orientational order information.9−11 ΔνQ = Q PSP2(cos θ) = Q PρSmatrix P2(cos θ)
EXPERIMENTAL SECTION
(1)
Here, QP is the quadrupolar coupling constant (∼260 kHz), as one would observe if all C−2H bond axes (or other bonds containing 2H) in the probe molecules are perfectly aligned with the spectrometer static magnetic field B0. S = ⟨P2(cos χ)⟩ is the orientational order parameter of the probe molecule where ⟨P2(cos χ)⟩ = (3 cos2 χ − 1)/2 is the ensemble average over the second Legendre polynomial with χ the angle between a particular C−2H bond and the matrix alignment axis (the director). θ is the angle between the matrix (e.g., lignin/ hemicellulose phase) alignment axis and B0. From the probe molecule orientational order parameter S, the orientational order parameter for the matrix (Smatrix) can be expressed as S = ρSmatrix, where Smatrix = ⟨P2(cos α)⟩ with α being the angle between the local matrix where the probe is diffusing and the alignment axis of the matrix. ρ represents a scaling factor that depends on the interaction between a specific probe molecule and its host matrix.10 Therefore, for a specific probe-matrix system, the quadrupolar splitting can be used to measure the orientational order parameter(s) and alignment symmetries and directions in the matrix. Note that we have two instances of the second Legendre polynomial (with angles χ and α) that concern the orientational order properties of the probe molecule and matrix, and one instance of this function (with angle θ) that concerns the “NMR dependence” relating the matrix director to B0. Equation 1 describes only uniaxially aligned (cylindrically symmetric) systems, and we would modify this equation if we observed a biaxial system signified by unequal splittings measured along the two directions transverse to the main alignment axis.15,16 Here we utilize two deuterated probe molecules, methylenelabeled ethylene glycol-d4 (EG) and formyl-labeled N,Ndimethylformamide-d1 (DMF), to probe the temperature and angular dependencies of orientational order in yellow-poplar (Liriodendron tulipifera), a commercially significant hardwood. These two probe molecules interact with the lignocellulose matrix in distinct ways, and they swell to substantially different 1044
dx.doi.org/10.1021/bm201770q | Biomacromolecules 2012, 13, 1043−1050
Biomacromolecules
Article
observations per data point, and this data was averaged to give splitting values. Deuterium spectra were processed using MestReNova (v. 6.1.1− 6384). A 10 Hz exponential apodization (line broadening) was applied to each spectrum. After phasing and baseline correction, one doublet and one singlet (both of Lorentzian line shape) were fitted to the spectra to obtain quadrupolar splitting, area, and width. Temperature Dependence of Orientational Order and Diffusion Coefficient. Experiments were performed on a Varian INOVA spectrometer at B0 = 9.39 T. A 5 mm axial-coil multinuclear liquids NMR probe was used. Specimens were placed in a standard 5 mm NMR glass tube and the head space above the specimen was occupied using a 4 mm diameter Kel-F plug (Wilmad-LabGlass, Vineland, NJ). Acquisition parameters were: acquisition time = 0.08 s, relaxation delay = 0.02 s, pulse time = 30 μs, spectral width = 25 kHz, number of scans = 1000 (except for DMF, where 6000 scans per experiment were averaged). Spectra were collected in isothermal conditions at different temperatures. The temperature probe was calibrated using pure and dry EG to within ±1 °C. Temperatures were increased (EG: 0 to 100 °C and DMF: −10 to 80 °C) in 10 °C increments, followed by a 10 °C stepwise cooling to the minimum temperature. At each temperature, specimens were equilibrated for 10 min before collecting spectra. Three specimens for each solvent were tested. All spectra were exponentially apodized (line-broadened by 10−20 Hz) and phase and baseline corrected. The high-temperature spectrum, which showed best peak resolution, was used to establish the number of peaks to be fit across the temperature range. All spectra were fitted with one Lorentzian doublet and one Lorentzian singlet using the line-fitting function of MestReNova (v. 6.1.1−6384). The distance between the doublet peaks (in Hz) was recorded as the quadrupolar splitting ΔνQ. The self-diffusion coefficient of EG (C2H6O2; 25% of dry wood mass) along the longitudinal direction was measured using a pulsedgradient stimulated echo (PGSTE) pulse sequence18,19 at 80 °C. Specimens were placed in a standard 5 mm NMR glass tube and the head space above the specimen was occupied using a 4 mm diameter Kel-F plug (Wilmad-LabGlass, Vineland, NJ). Using a single-axis diffusion probe the PGSTE sequence used a π/2 pulse of 4.5 μs, gradient pulse duration (δ) 1 ms, and diffusion times (Δ) ranging from 50 to 300 ms. Adequate signal-to-noise ratio was achieved with 56 scans.
■
respectively. Separate measurements were conducted with the tangential (T) direction parallel to B0. At the L-orientation, a doublet signifying an oriented phase and a smaller singlet representing an isotropic phase, were observed. Probe molecules having a single, chemically distinct deuterium resonance were selected such that only a singlet would appear under isotropic conditions. The shapes and proportions of the singlet and doublet will be discussed later, but note that all spectra presented here indicate that the swollen specimens were below fiber saturation since no sharp singlet is observed (no “free liquid” signal). The quadrupolar splitting was measured in each spectrum using a least-squares fit of one Lorentzian doublet and one singlet. Within the experimental uncertainties (coefficient of variation: 10%), the quadrupolar splitting for the L direction (defined as ΔνL) was double the splitting of that observed along the R (ΔνR) and the T (ΔνT) directions (Figure 2). By definition, this is characteristic of a uniaxialy oriented material (ΔνL = 2ΔνR = 2ΔνT),8,10,20 indicating that the two transverse directions were indistinguishable at the length scale probed (submicrometer, discussed below). Figure 3 plots the
Figure 3. Average quadrupolar splitting vs alignment angle θ for 25% ethylene glycol-d4 in yellow-poplar. Solid line represents uniaxial fit to the average splitting data. Error bars represent ±1 standard deviation (n = 3).
quadrupolar splitting ΔνQ as a function of the angle θ between B0 and the grain/fiber direction, along with a uniaxial fit according to eq 1. The uniaxial equation (eq 1) fit the experimental data very well (R2: 0.97) with the zero-crossing angle appearing at the magic angle, 54.7°. The probe orientational order parameter S = 0.009 ± 0.002. This provides a relative measure of the matrix order parameter (Smatrix), which is generally 1−3 orders of magnitude higher than S (that is, S for the probe molecule), and this scaling factor ρ depends on the probe aspect ratio and specific probe−matrix interactions.10,21 In a perfectly oriented system, S becomes 1 (for orientation axis/director parallel to B0) or −0.5 (director perpendicular to B0);9 the value of S becomes 0 for systems with no orientation or when the orientation axis/director assumes 54.7° with B0. S = 0.009 is a relatively large value in comparison to probe molecules absorbed in other oriented polymers such as Nylon-622 and drawn Nafion.10,20,21,23 For instance, in a hydrated perfluoronated ionomer (Nafion) S ∼ 0.001 was reported, where the respective Smatrix was ∼0.5.23 In hydrated Nylon-6, Loo et al. observed S ∼ 0.001 with Smatrix = 0.78.22 Precise measurement of Smatrix requires knowledge of the scaling factor (ρ = S/Smatrix), which contains information about probe geometry (aspect ratio) and probe−matrix interactions. This scaling factor can in some cases be obtained using small-angle X-ray scattering
RESULTS AND DISCUSSION
Figure 2 shows the 2H NMR spectra of EG-d4 in yellow-poplar as a function of the angle θ between the wood fiber axis and the static magnetic field (B0). Specimens were rotated in the radial plane so the 0 and 90° acquisitions represented the longitudinal (L) and the radial (R) wood tissue directions parallel to B0,
Figure 2. Deuterium NMR spectra for 25 wt % ethylene glycol-d4 in yellow poplar at 30 °C. The quadrupolar splitting ΔνQ changes as in eq 1 as a function of the angle θ between the wood fiber axis and B0. 1045
dx.doi.org/10.1021/bm201770q | Biomacromolecules 2012, 13, 1043−1050
Biomacromolecules
Article
experiments,21,23 although this would be very challenging in wood because the highly heterogeneous structure creates many overlapping scattering signals. In essentially all known cases of 2 H NMR probe molecule measurements in soft materials, S ≤ 0.01, even when the matrix is highly oriented (Smatrix ∼ 1). Thus, it is reasonable to believe that the wood Smatrix ∼ 1, signifying a highly oriented amorphous domain parallel to the fiber axis. This is consistent with the current knowledge of hardwood characteristics.1,6,7 However, from polarized infrared studies it is known that in softwood (specifically spruce) lignin shows only slight or no orientation.4,5 Therefore, we would expect a lower S value measured by NMR in spruce as compared to yellow-poplar. Such studies will form the basis of future surveys of wood morphology using these NMR tools. In wood the probe rapidly diffuses throughout the amorphous regions and the observed 2H signal reflects the average orientational order in the domain where the probe resides. Depending on the probe diffusion coefficient (D) and the time scale for probe motion during the 2H NMR experiment (tNMR ∼ 1/ΔνQ ≈ 0.5 ms), the experimental length scale can be estimated as the root-mean-square distance traveled under a 1D random walk: ⟨r2⟩1/2 = (2Dt)1/2. PGSTE diffusometry revealed that EG diffusion coefficients (D) along the longitudinal direction of yellow-poplar wood range from 7 × 1012 to 1 × 1012 m2/s for diffusion encoding times (Δ) ranging from 50 to 300 ms. Using the 1D random walk expression above (and substituting t with Δ), the corresponding length scales were 30 to 80 nm. However, with decreasing Δ, the D increases, although not in a linear trend. Therefore, for the 2H NMR time scale, 0.5 ms, we estimate the 1D diffusion length scale to be between 100 and 300 nm. Detailed diffusometry studies should provide a more precise estimate of the experimental length scale. It was previously reported that the self-diffusion of water in a saturated hardwood was anisotropic; water diffusion coefficients were about 3 times higher along the fiber direction as compared to the transverse directions.24 For the present study, therefore, the domain probed by EG-d4 is also hypothesized to be nonspherical (ellipsoidal), with the transverse length scale being roughly one-third of the longitudinal length scale. Figure 3 reports that the maximum splitting occurred at 0°, that is, when the fibers were parallel to B0, suggesting that the cellulose fibrils were on average aligned at a single specific angle with respect to the fiber axis (and B0) as occurs in the S2 layer.25 Consequently this demonstrates that the doublet signal originated from the S2 layer, which comprises ≈70% of the hardwood tissue.12 Moreover, it is known that cellulose rarely allows solvent penetration into its crystalline core.26,27 Although a variety of exotic solvents have been identified as cellulose solvents,28,29 the cellulose crystalline structure is impervious to the solvents used in this study (EG and DMF).30,31 Therefore, it is assumed that the probes in this study could only reside in the solvent accessible amorphous domains. See the Figure 6 discussion below for elaboration on this matter in the context of spin−spin relaxation times (T2). In the secondary cell wall of hardwoods, all amorphous polymers are thought to be oriented parallel to the cellulose microfibrils.4,7,32 Considering that S = 0.009 in a highly oriented uniaxial matrix with roughly cylindrical domains of at least 100−300 nm in length, it is reasonable to postulate that the doublet reflects the average lignin and hemicellulose orientation within and between cellulose microfibrils in the S2 layer, consistent with contemporary wood cell wall models.1
The effects of temperature on the 2H NMR spectra are demonstrated in Figure 4, where over this range EG-saturated
Figure 4. Isothermal 2 H spectra for 25% ethylene glycol-d 4 equilibrated in yellow-poplar, collected at different temperatures. Specimen longitudinal grain L (fiber axis) is parallel to B0.
yellow-poplar exhibits a glass/rubber transition around 80 °C (compressive-torsion DMA at 5 Hz).33 At low temperatures (between 0° and 40 °C), a broad doublet was observed, signifying slowly moving probes in an oriented phase. With increasing temperature, the doublet width decreased, especially beyond 40 °C, where also a broad singlet emerged between the doublet peaks. As mentioned, the specimens existed below fiber saturation, indicated by the high singlet line width (∼700 Hz) and the fact that a free liquid-probe phase would exhibit a line width of ∼1− 10 Hz. The broad singlet reflects an isotropic polymer domain of no smaller than 100−300 nm and physically separated from the oriented phase on this same experimental length scale. These features suggest that the singlet represents the unoriented tissue found in the compound middle lamella, CML (the primary cell wall plus the middle lamella). The CML accounts for 8−15% of the hardwood xylem tissue mass, with little cellulose (
