ARTICLE pubs.acs.org/JPCA
Conformation-Specific Spectroscopy and Populations of Diastereomers of a Model Monolignol Derivative: Chiral Effects in a Triol Chain Jacob C. Dean, Evan G. Buchanan, William H. James III,† Anna Gutberlet,‡ Bidyut Biswas, P. Veeraraghavan Ramachandran, and Timothy S. Zwier* Department of Chemistry, Purdue University, West Lafayette, Indiana 47907-2084, United States
bS Supporting Information ABSTRACT: Single-conformation spectroscopy of two diastereomers of 1-(4-hydroxy-3-methoxyphenyl)propane-1,2, 3-triol (HMPPT) has been carried out under isolated, jet-cooled conditions. HMPPT is a close analog of coniferyl alcohol, one of the three monomers that make up lignin, the aromatic biopolymer that gives structural integrity to plants. In HMPPT, the double bond of coniferyl alcohol has been oxidized to produce an alkyl triol chain with chiral centers at C(R) and C(β), thereby incorporating key aspects of the β-O-4 linkage between monomer subunits that occurs commonly in lignin. Both (R,S)and (R,R)-HMPPT diastereomers have been synthesized in pure form for study. Resonant two-photon ionization (R2PI), UV hole-burning (UVHB)/IR-UV hole-burning (IR-UV HB), and resonant ion-dip infrared (RIDIR) spectroscopy have been carried out, providing single-conformation UV spectra in the S0S1 region (3520035800 cm1) and IR spectra in the hydride stretch region. Five conformers of (R,S)- and four conformers of (R,R)HMPPT are observed and characterized, leading to assignments for all nine conformers. Spectroscopic signatures for Rβγ, γβR, and Rγβπ chains and two cyclic forms [(Rβγ) and (Rγβ)] of the glycerol side chain are determined. Infrared iongain (IRIG) spectroscopy is used to determine fractional abundances for the (R,S) diastereomer and constrain the populations present in (R,R). The two diastereomers have very different conformational preferences. More than 95% of the population of (R,R) configures the glycerol side chain in a γβR triol chain, while in (R,S)-HMPPT, 51% of the population is in Rβγ chains that point in the opposite direction, with an additional 21% of the population in H-bonded cycles. The experimental results are compared with calculations to provide a consistent explanation of the diastereomer-specific effects observed.
I. INTRODUCTION Lignin is an aromatic heteropolymer found in all vascular plants, ranking second only to cellulose as the most abundant biopolymer found in nature.13 As a major component of secondary thickened cell walls, lignin gives the plant its structural rigidity, thereby enabling the transport of water and nutrients through the vascular system.1,2,4 Lignin also protects the cell wall from degradation by microorganisms, increasing its resistance to decay and prolonging the plant’s life.3 The same chemical composition that increases resistance to decay also makes the lignin polymer difficult to break down during the conversion of lignocellulosic biomass to harvestable biofuels. Understanding the chemical composition, variability, and energetics of the chemical linkages within the lignin network is therefore needed to efficiently convert plant mass to harvestable sugars.5,6 Lignin is made up of just three chemically distinct monomers, the monolignols p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol, whose structures are shown in Figure 1a. In the lignification process, the monolignols are transported to the cell wall where r 2011 American Chemical Society
they are oxidized and polymerized via radical coupling reactions.3,7 Regulation of the relative abundances of the monolignols during polymerization leads to the polymer containing a variety of chemical linkages that determine the level of branching and cross-linking in the lignin produced. In this way, the degree of structural rigidity is controlled in plants ranging from hardwoods to grasses.3,8,9 Among the chemical linkages between monomer units, the most abundant is the β-O-4 linkage, derived from the coupling of two coniferyl alcoholderived (guaiacyl) units, as shown in Figure 1b.8 Many analytical methods, including NMR, HPLC, and GC/ MS, have been employed to examine the relative abundances of the monomers in samples ranging from simple dimers to large oligomers extracted directly from plant samples.2,10,11 The NMRbased methods are also capable of quantifying the types and amounts of each chemical linkage, and can detect the presence of Received: May 10, 2011 Revised: June 15, 2011 Published: July 14, 2011 8464
dx.doi.org/10.1021/jp204367k | J. Phys. Chem. A 2011, 115, 8464–8478
The Journal of Physical Chemistry A
Figure 1. (a) Chemical structures of the three monolignols. (b) Left: HMPPT structure with asterisks indicating the two chiral centers. Right: β-O-4 dimer unit.
other minor monolignols that occur in specialized circumstances.2 Recently, mass spectrometry has also served as a significant tool for the sequencing of small oligomers via MSn-based fragmentation, with the goal of analyzing longer oligomeric samples by building a library of characteristic fragmentation pathways of specific lignin structural units.4,12 The fact that lignin is composed entirely of aromatic monomers raises the possibility that ultraviolet spectroscopy could play a role in lignin structural characterization. With this in mind, our group has recently begun a systematic study of lignin structural units under isolated, jet-cooled conditions. The initial study focused attention on the three monolignols p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol.9 The three monolignols show characteristic UV spectral signatures that can be readily used to distinguish between them. Furthermore, double-resonance spectroscopy was used to determine UV and IR spectra of single conformations of the three molecules, thereby characterizing their preferred conformations. In each case, there were only two conformational isomers present, associated with a syn/anti pair between the vinyl group and OH group in the para position on the aromatic ring. The utility of these methods also extends to exploring chiral effects within and between molecules. When more than one chiral center is present in the same molecule, differences in the conformational preferences of the diastereomers so produced can be explored.13 When the chiral centers are on different molecules, chiral recognition can be probed by forming complexes between them and exploring their spectroscopic properties.14 In lignin, chiral centers are abundant in the growing polymer since at every β-coupling, at least one chiral center is created. However, naturally occurring lignin shows no net chirality, indicating a lack of direct stereocontrol in the biosynthetic process.15
ARTICLE
Despite the racemic nature of the polymer, studying diastereomeric effects within a single linkage can provide insight to the folding and energetics at a local scale. Recently, double-resonance jet spectroscopy has also been used to directly measure the relative abundances of observed conformational isomers of flexible molecules via IR population transfer (IRPT) spectroscopy16 or its recently implemented analog, IR ion-gain (IRIG) spectroscopy.17 Such measurements provide incisive tests of quantum chemical calculations, and deeper insight to the relationship between the thermal conformational populations present before supersonic expansion, and those observed downstream in the expansion after the cooling process.18 Given the importance of the β-O-4 linkages in most lignin, we report here a study of the single-conformation ultraviolet and infrared spectroscopy of the coniferyl alcohol/β-O-4 derivative 1-(4-hydroxy-3-methoxyphenyl)propane-1,2,3-triol (HMPPT, Figure 1b). HMPPT is a model lignin monomer derivative that differs from coniferyl alcohol in having its double bond oxidized to produce a triol chain closely analogous to that found in the β-O-4 linkage in larger lignin oligomers. As with lignin, formation of the β-O-4 linkage produces two chiral centers on R- and β-carbons that produce two diastereomeric forms, (R,R)- and (R,S)-HMPPT. Single-conformation UV spectra in the S0S1 region and IR spectra in the OH stretch region are reported for (R,R)- and (R,S)-HMPPT, providing a basis for comparison with density functional theory (DFT) calculations that lead to firm assignments for all nine observed conformers of the two diastereomers. As we shall see, one of the intriguing aspects of the structure of HMPPT is its glycerol-like triol side chain, which places three OH groups on adjacent carbon atoms where they can H-bond with one another. The intramolecular H-bonds so produced serve as a source of diagnostic OH stretches that report on the H-bonded network formed. Interestingly, the triol side chain shows remarkable structural diversity, forming H-bonded chains that are directed either toward or away from the aromatic ring, clockwise and counter-clockwise H-bonded cycles, and extended chains involving the aromatic π cloud as terminal H-bond acceptor. Several of these structural types are present already in glycerol itself, whose conformers have been probed in a recent microwave study under jet-cooled conditions, aided by ab initio calculations.19,20 The present work bears some similarity with work from Simons and co-workers on carbohydrates, which also possess an array of OH groups whose H-bonding patterns can be used to distinguish them.21,22 The triol side chain also bears a resemblance to trimeric water and methanol clusters, where cycles and chains compete with one another for the lowest energy structures formed.22,23 The two diastereomers of HMPPT show distinctly different conformational preferences that reflect a shift in the delicate balances between the various types of cycles and chains when the β-carbon is changed from R to S chirality. We explore the reason for these shifts by comparing the observations with the predictions of dispersion-corrected DFT calculations. Finally, the present results are to be seen in the context of future studies of larger oligomers, where the structural components present here are incorporated into a sequence of linkages. It is hoped that characterization of each linkage by itself will provide a road map for such studies.
II. METHODS A. Experimental Methods. The (R,R)- and (R,S)-diastereomers of HMPPT were prepared starting with commercial 8465
dx.doi.org/10.1021/jp204367k |J. Phys. Chem. A 2011, 115, 8464–8478
The Journal of Physical Chemistry A
ARTICLE
Figure 2. (a) Schematic diagram for IR ion-gain spectroscopy. Blue arrows denote resonant excitation in R2PI in the absence of IR excitation, while orange arrows identify R2PI ionization after IR excitation. (b) R2PI spectrum for RR-HMPPT with a dashed red line indicating the UV wavelength used in IRIGS. (c) Top: IRIG spectrum for RR-HMPPT under unsaturated conditions. Lower: IRIG spectrum at higher powers showing an increase in intensity for weak transitions not seen without saturation.
(Aldrich) ferulic acid via AD-mix R- and β-mediated Sharpless asymmetric dihydroxylation,24,25 as described by Previtera and co-workers.26 The 1H and 13C NMR spectra and the optical rotations of the triols matched with those reported in the literature.27 A differentially pumped time-of-flight (TOF) mass spectrometer that has been described previously was used in this work.28 Introduction of the sample into the gas phase could not be achieved by heating methods due to rapid polymerization, so laser desorption was implemented instead. The method used for desorption of the neutral molecules was an adaptation of the desorption/ionization on porous silicon (DIOS) method used to introduce nonvolatile samples as ions in mass spectrometry applications.29 As a method for laser desorption of neutrals, we refer to it as laser desorption on (etched) silicon (LDOS). General information on the preparation of the silicon chips used in this experiment is included in the Supporting Information, while details on the procedure for HF-etching and preparation of the silicon wafers for LDOS/DIOS is described by Woo et al.29 The doped silicon chips were fastened into the back of a tube (7 mm long, 2 mm channel width) that was fixed onto the front face of a Series 9 Parker General Valve, serving as a stationary desorption source. The chips were fixed so that their edges bordered the nozzle orifice allowing for desorption to occur along the molecular beam axis. A schematic diagram of the LDOS setup is included in the Supporting Information. The third harmonic of a Nd:YAG laser (Continuum Minilite II) operating at 20 Hz was used as the desorption laser at 1.3 mJ/ pulse power. The Series 9 General Valve was operated at 20 Hz and was used to form the supersonic expansion. The desorbed sample was entrained in argon at a backing pressure of 34 bar, expanded into the vacuum and skimmed, forming the molecular beam upon entering the ionization region. Two-color resonant two-photon ionization (2C-R2PI) was used to record massselected UV spectra of the two diastereomers in the S0S1 region. The collimated, doubled outputs of two Nd:YAG-pumped dye
lasers were used for the S0S1 resonant step (279285 nm, 0.l0.3 mJ/pulse) and S1-ionization steps (285 nm, ∼0.30.5 mJ/pulse). Conformation-specific electronic spectra were recorded using either UVUV holeburning (UVHB) or IR-UV holeburning (IRUV HB) spectroscopy, while resonant ion-dip infrared spectroscopy was used to record infrared spectra in the OH stretch region (34803700 cm1). These methods have been described elsewhere.30 Details specific to the present work are included in the Supporting Information. The present work also utilizes infrared ion-gain (IRIG) spectroscopy, a close analog of methods implemented previously by the Quack31 and Fujii groups.3234 In the present context, IRIG spectroscopy served as a sensitive method for recording the total IR spectrum containing contributions from all conformational isomers present in the jet-cooled expansion. A schematic diagram for the method is shown in Figure 2a. The experimental setup is much like that in RIDIR spectroscopy, except that the wavelength chosen for the fixed probe laser is deliberately chosen so as not to be resonant with any of the conformational isomers’ jet-cooled transitions, well to the red of the S0-S1 origins of all conformers present. As a result, no ion signal is observed without external excitation, as shown in Figure 2b for RR-HMPPT where the nonresonant UV wavelength was 284 nm (35211 cm1). To record IRIG spectra, the IR laser is tuned through the region of interest, and when resonant with a vibrational transition of any conformer, a gain in ion signal is produced by the UV probe laser. As shown in Figure 2a, the gain in ion yield is due to the redistribution of energy in the vibrationally excited molecule from the initial vibration (OH stretch) to other vibrational degrees of freedom (dominated by low-frequency modes) via intramolecular vibrational redistribution (IVR). Many of the modes that participate in IVR are those that lead to isomerization. If the IR energy is above the barrier to isomerization, conformationally mixed states can be produced, many of which absorb to the red of the S0S1 origin. The result is that excitation 8466
dx.doi.org/10.1021/jp204367k |J. Phys. Chem. A 2011, 115, 8464–8478
The Journal of Physical Chemistry A
ARTICLE
out of the conformationally mixed “hot bath” subsequent to IR absorption, leads to an increase in ion signal against zero background. In most circumstances, the IR power used for IRIG scans was identical to that used in recording the corresponding RIDIR spectra (0.52.0 mJ/pulse). At times, the contributions from weak transitions or minor population conformers were enhanced by raising the IR power to ∼2.53.0 mJ/pulse, as demonstrated in Figure 2c. This proved very helpful in identifying the presence of minor conformers with unique IR absorptions. Besides acquiring the IR-induced gain spectrum due to all conformers, IIRIG(ν), IRIG scans were also used as a method for determining fractional abundances of the conformers present in the jet.17 By definition, the gain spectrum is the weighted sum of the RIDIR spectra for all conformers: IIRIG ðνÞ ¼ c
i ðνÞ ∑i Fi γi IRIDIR
ð1Þ
Here, “c” is a scaling factor required to scale the RIDIR-based fit to the IRIG spectrum, Fi is the fractional abundance of conformer i, IiRIDIR is the RIDIR spectrum of conformer i measured as a fractional depletion, and γi is the detection efficiency of conformer i. To avoid systematic errors from saturation effects, the IRIG and RIDIR spectra are recorded under identical IR power and beam conditions. We have shown elsewhere17 that populations extracted from fits to eq 1 assuming that the detection efficiency of all conformers is equal at the probe wavelength (i.e., γi = γj for all j 6¼ i) give the same fractional populations as when IRPT methods are used. Equal detection efficiencies of the IRexcited molecules are likely a natural consequence of giving the ground state molecules sufficient energy (from IR excitation) to isomerize, thus, producing a conformationally mixed set of molecules with identical R2PI efficiencies, independent of their starting conformation. In the present case, we have checked this by recording IRIG scans at a series of UV wavelengths. These scans are identical to one another, apart from a slow drop-off in intensity as the probe wavelength is shifted further to the red. As a result, we assume equal detection efficiencies in using eq 1 to extract fractional abundances of the conformers of HMPPT. The fitting procedure used in determining the fractional abundances utilized the LevenbergMarquardt algorithm, as implemented in Mathcad 14.0.35 B. Computational Methods. To identify the possible conformational minima associated with each of the two diastereomers of HMPPT, a conformational search was carried out using the Amber* force field36 in the MACROMODEL suite of programs.37 Conformational space was sampled up to a maximum energy of 50 kJ/mol, identifying approximately 230 structures for each of the diastereomers. Approximately 50 of the lowest energy structures obtained from the force field search (for each diastereomer) were used as starting structures for full optimization and harmonic frequency calculations by density functional theory (DFT) using the B3LYP38 and M05-2X39 functionals with the 6-31+G(d) basis set. All of the DFT calculations were carried out using the Gaussian 09 suite of programs.40 For the geometry optimizations, a tight optimization convergence criterion was specified, and all of the calculations were run with an ultrafine grid. While quantitatively different, the two DFT methods yielded a similar energy ordering of the conformers of HMPPT, which are dictated largely by H-bonding. The dispersion-corrected functional M05-2X calculations will be used in reporting the relative
Figure 3. Conformational families and chain types for HMPPT, including, (a) H-bonded chains and (b) H-bonded cycles. (c) Syn/anti pairing occurs for every conformational family as a result of the R-OH orienting syn or anti to the methoxy group on the aromatic ring.
energies (after zero-point correction) and for comparison with experimental infrared spectra. Vertical excitation energies were also calculated for the lower energy, optimized structures (
