CELLULOSE (1997) 4, 173207

Cellulose: the structure slowly unravels

ANTOINETTE C. O'SULLIVAN Department of Chemistry and School of Agriculture and Forestry Sciences, University of Wales, Bangor, Gwynedd, LL57 2UW, UK Received 13 May 1996; accepted for publication 3 December 1996

This article attempts to bring together basic and complex information which has been gathered on cellulose structure, principally that of native cellulose, over the last few decades. Even though advances have been made in the eld of crystallography, powder crystallography cannot yield a denitive cellulose structure and single crystal diffraction is not possible due to the lack of suitable crystals. Knowledge obtained on the biosynthesis of native cellulose and on the polymorphy of cellulose and its derivatives help our understanding of ultrastructure. Many inconsistencies between early crystallographic studies of native cellulose have been claried by the discovery that two polymorphs ( and ) of cellulose I exist. Models of the possible ultrastructural arrangements within native cellulose have been put forward over the decades; with advancement in technology, computer simulations of small and large systems are being created to test the viability of these ultrastructural models. It is hoped that this review will aid in the understanding of the complexity and uncertainties that still exist in this subject. KEYWORDS: structure, computer, model, polymorphy

INTRODUCTION Over 150 years ago, Anselme Payen discovered and isolated cellulose from green plants. Several reviews have been published on cellulose research (Preston, 1975, 1986; Sarko, 1987; Chanzy, 1990; Okamura, 1991). These state that this compound is the most abundant material on the earth: it is the main constituent of plants, serving to maintain their structure, and is also present in bacteria, fungi, algae and even in animals. Crystallographic research, yielding varying results, has also been extensively considered in the above reviews. Despite the degree to which cellulose has been investigated, its structural features have not been identied with absolute clarity and new information is constantly being discovered by employing technological advances alongside conventional analytical tools. This review attempts to combine the facts and hypotheses of past decades with new eveidence on the polymorphy of native cellulose and the synthesis of microbrils. Cellulose research has been instrumental in advancing many analytical methods such as crystallography and microscopy. To gain an understanding of all facets of this complicated structure, the techniques used have been varied and the resultant terminology is vast. This latter point has frequently led to confusion especially on topics such as the direction of cellulose chains in unit cells and the distinction between amorphous and crystalline cellulose. Computational chemistry is one analytical technique which has developed extensively over the last three decades, initially being used to decipher complex crystallographic
09690239 # 1997 Blackie Academic & Professional

174

O'SULLIVAN

data and now analysing the feasibility of cellulose models from an energetic perspective. POLYMORPHY OF CELLULOSE The polymorphy of cellulose and its derivatives has been well documented. Six polymorphs of cellulose (I, II, III1 , III11 , IV1 and IV11 ) can be interconverted, as shown in Fig. 1 (Marchessault and Sarko, 1967; Walton and Blackwell, 1973; Marchessault and Sundararajan, 1983). More recently, evidence for two polymorphs of cellulose I has been offered (VanderHart and Atalla, 1984; Sugiyama et al., 1991a); that is to say what was previously thought to be one polymorph (I) has now been found to be a mixture of two polymorphs (I and I). Proof of the polymorphy of cellulose comes from nuclear magnetic resonance (NMR), infrared and diffraction studies (Blackwell and Marchessault, 1971; Blackwell, 1982). Cellulose I, or native cellulose, is the form found in nature. Cellulose II, the second most extensively studied form, may be obtained from cellulose I by either of two processes: a) regeneration, which is the solubilization of cellulose I in a solvent followed by reprecipitation by dilution in water to give cellulose II, or b) mercerization, which is the process of swelling native bres in concentrated sodium hydroxide, to yield cellulose II on removal of the swelling agent. Celluloses III1 and III11 (Marrinan and Mann, 1956; Hayashi et al., 1975) are formed, in a reversible process, from celluloses I and II, respectively, by treatment with liquid ammonia or some amines, and the subsequent evaporation of excess ammonia (Davis et al., 1943; Sarko et al., 1976; Sarko, 1987). Polymorphs IV1 and IV11 (Hess and Kissig, 1941; Gardiner and Sarko, 1985) may be prepared by heating celluloses III1 and III11 respectively, to 206 8C, in glycerol. BIOSYNTHESIS OF CELLULOSE Cellulose is a brous, tough, water-insoluble substance, which is found in the protective cell walls of plants, particularly in stalks, stems, trunks and all woody portions of plant
Regeneration Mercerization Cellulose I NH3(I) NH3(g) Cellulose II NH3(I) NH3(g)

Cellulose IIII heat

Cellulose IIIII heat

Cellulose IVI

Cellulose IVII

FIGURE 1. Interconversion of the polymorphs of cellulose.

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

175

tissues. Before attempting to study the ultrastructure of this macromolecule, it is prudent to understand its molecular structure. An unbranched homopolysaccharide, cellulose is composed of -D-glucopyranose units linked by (1 3 4) glycosidic bonds (Purves, 1954; Marchessault and Sundararajan, 1983). These pyranose rings have been found to be in the chair conformation 4 C1 , with the hydroxyl groups in an equatorial position (Fig. 2). In nature, cellulose chains have a degree of polymerization (DP) of approximately 10 000 glucopyranose units in wood cellulose and 15 000 in native cotton cellulose (Sjostrom, 1981). There is some evidence for a lower degree of polymerization in primary cell walls as compared with secondary cell walls. However, chain lengths of such large, insoluble molecules are rather difcult to measure, due to enzymic and mechanical degradation which may occur during analysis. Available evidence suggests that cellulose is formed at, or outside, the plasma membrane (Lamport, 1970; Brett and Waldron, 1990). Groups or rosettes of particles, or terminal complexes (TCs), are seen in the plasma membrane when viewed by freezefracture techniques (Okuda et al., 1994). These groups of TCs can be seen to be associated with the ends of microbrils (collections of cellulose chains), and are thought to be cellulose synthase complexes, involved in the elongation of whole cellulose microbrils. Thus, all the chains in one microbril would have to be elongated by the complex at the same rate. This requirement means that the complex would need to be comprised of many subunits, each elongating a single chain at a time. Since there are between 30 and 200 chains in the cross-section of one microbril, this process would be one of remarkable complexity, but nevertheless research has favoured it (Macchi and Palma, 1969). Crystalline cellulose I is not the most stable form of cellulose. It is unlikely to be synthesized by crystallization of pre-formed cellulose chains, since such a process carried out in vitro gives rise to the thermodynamically more favourable cellulose II. Hence, the most probable process is one in which crystallization accompanies, or follows very closely, addition of glucose residues to the chains by the TCs. Such a process would permit the geometry of the enzyme complex to govern the crystalline form of the microbril. Crystallization of cellulose is thought to be brought about by a protein, the top protein, which accompanies the TC. The synthesis of straight

CH2OH O O HO O HO HOCH2 HO

OH

O O

FIGURE 2. Fragment (repeating unit) of a cellulose chain.

176

O'SULLIVAN

microbrils by TCs was corroborated by Sugiyama et al. (1994), who found that triclinic (I) microbrils are unidirectional. Matsuda et al. (1992) recently showed that cellulose crystallization in immature cotton boll is directly related to the formation of intermolecular hydrogen bonds. This may be due to the fact that the cellulose molecules are arranged into a microbril as they crystallize. A further discovery was that the synthesis of the cellulose microbrils of the tunicate Metondrocarpa uedia, was linked to vacuole like structures (Kimur and Itoh, 1995). Thus, the area of native cellulose biosynthesis continues to unfold.

X-RAY CRYSTALLOGRAPHY OF CELLULOSE I Cellulose has been under continuous investigation since its discovery, due to its structural complexity, and has often been among the rst substances to be studied by new methods (Sarko, 1987). Among the techniques which have been used on cellulose are Fourier transform infrared spectroscopy (FTIR) (Marchessault and Liang, 1962), Cross PolarisationaMagic Angle Spinning (CPaMAS) NMR (Earl and VanderHart, 1981; Horii et al., 1982, 1987a, 1987b; Dudley et al., 1983; Atalla and VanderHart, 1989; Yamamoto and Horii, 1993) neutron (Ahmed et al., 1976) and electron (Herbert and Muller, 1974) diffraction, Raman spectroscopy (Atalla and VanderHart, 1989), scanning electron microscopy (SEM), (Fengel and Stoll, 1989) atomic force microscopy (AFM), (Hanley et al., 1992; Kuutti et al., 1995) transmission electron microscopy (TEM), (Purz et al., 1995) scanning tunnelling microscopy (STM), (Frommer, 1992) and computer based stereochemical modelling (see section: Computer analysis of cellulose). In 1858, Carl von Nageli (von Nageli, 1858) established the crystallinity of cellulose in the rst serious use of the polarizing microscope. This result was veried 80 years later with the aid of powder X-ray crystallography (Meyer and Misch, 1937). Elucidation of the crystal structure of cellulose began with Sponsler and Dore (1926), who suggested a single chain unit cell with dimensions: a 0.61; b (bre axis) 1.034; c 0.54 nm; and 888. This model assumed an alternation of (1 3 1) and (4 3 4) links along the chain. A more acceptable unit cell was deduced by Meyer and Mark (1928) on the basis of different crystallographic parameters (a 0X835; b 1X03; c 0X79 nm; and 848) and improved in a later model by Meyer and Misch (1937). The latter two models had a two chain unit cell, with uniform (1 3 4) linkages between successive residues, in agreement with chemical evidence published by Haworth (1925). Meyer and Misch explained the fact that the unit cell parameter along the bre axis direction is shorter than the fully extended length of a cellobiose residue by proposing a model with a kink, resulting from the formation of a hydrogen bond between the C(3) of one glucose residue and the ring oxygen of the next (Preston, 1986). It should be noted at this point that the convention for quoting unit cell parameters have been revised since the rst crystallography work on cellulose was carried out, and now the bre axis of a cellulose unit cell is denoted by c and the unique angle is denoted by . Furthermore, in early work the unique angle of the cellulose unit cell was written as an acute angle but in modern crystallography an obtuse angle is used. For the purpose of clarity the modern conventions will be used throughout the remainder of this literature review.

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

177

Unit cell parameters for crystalline cellulose I are listed in Table 1. Variations between the results of different research groups may be explained by several factors: (1) Technical advances improved the ability to measure the parameters. (2) Parameters depend on the source of the cellulose and it has been suggested that this may be linked to taxonomy (Okano et al., 1989). (3) Single-crystal diffraction is the only method that provides bond lengths, bond angles and characteristic geometry. It is the ultimate structural method. On the other hand, polymer X-ray studies, as in the case of cellulose, use powder crystallography. When carrying out powder X-ray crystallographic analysis the operator has to input assumptions to obtain results. Thus, the structure of cellulose cannot be determined with certainty by X-ray data alone, as the results obtained are dependent on the model input. (4) Degree of crystallinity of samples may be a factor. (5) Sample purity, that is how many polymorphs of cellulose are present, is also pertinent. When studying some very highly crystalline celluloses (e.g. Valonia ventricosa), a reasonably well resolved spectrum resulted, showing a slightly different diffraction diagram which was taken to indicate an eight-chain unit cell (Honjo and Watanabe, 1958). Dimensions given for this unit cell were quoted as twice those of a two chain unit cell; a 1X576; b 1X642; c (bre axis) 1X034 nm; and 96X88 (Sarko and Muggli, 1974). Cotton, ramie, wood and other brous celluloses, of medium crystallinity, were found to be consistent with a two-chain monoclinic unit cell as
TABLE 1. X-ray crystallography data for native cellulose Specimen
NS NS NS

c (nm) 1.03 1.034 1.03 1.03 1.033 1.034 1.034 1.034 1.034 1.058 1.039 1.038 1.038 1.034 1.034 1.034 1.034 1.038 1.037

(8) 84 88 84 82 98.15 96.65 96.45 97.23 96.53 96.38 82 95.90 96.97 97.0 96.8 97.04 96.38 96.5 96.6 97.6

Method X-ray X-ray X-ray X-ray X-ray X-ray X-ray X-ray X-ray X-ray ED X-ray X-ray X-ray X-ray ED X-ray X-ray ED X-ray

Reference Meyer and Misch, 1937 Sponsler and Dore, 1926 Meyer and Mark, 1928 Fischer and Mann, 1960 Wellard, 1954 Wellard, 1954 Wellard, 1954 Wellard, 1954 Wellard, 1954 Wellard, 1954 Honjo and Watanbe, 1958 Marchessault and Sarko, 1967 Nieduszynshi and Atkins, 1970 Gardner and Blackwell, 1974a Sarko and Muggli, 1974 Claffey and Blackwell, 1976 French, 1978 Woodcock and Sarko, 1980 Okano and Sarko, 1984 Takahashi and Matsunaga, 1991

Valonia Bacteria Valonia Cladophora Linen Cotton Ramie Valonia Ramie Valonia Valonia
NS

Valonia
NS NS

Valonia Ramie
NS,

not specied; ED, electron diffraction.

178

O'SULLIVAN

shown in Fig. 3, that is a unit cell with one non-908 angle (a T b T c and T 908), with dimensions: a 0X778; b 0X820; c (bre axis) 1X034 nm; and angle 96X58 (Woodcock and Sarko, 1980). It was accepted that the two-chain unit cells of other native celluloses could very nearly be considered as sub-cells of the eight-chain Valonia unit cell. The unit cell contains a two ring portion of the cellulose chain which is 1.034 nm in length, called the bre repeat. Most workers agree that the symmetry of the unit cell is either close to, if not exactly, P21ac or belongs to a P1 space group (Fischer and Mann, 1960; Nyburg, 1961). The non 908 angle in the monoclinic unit cell was reported to tend towards a right angle in the more advanced taxonomic groups (Okano et al., 1989). Two-fold screw axes have been reported to exist for both of the symmetrically independent molecules in the two-chain unit cell, but it has been proposed that such symmetry may exist for the backbone of the cellulose chains, if not the pendant groups. An eight molecule unit cell has been reported by Herbert and Muller (1974) as having triclinic symmetry, that is to have a unit cell with no 908 angles (a T b T c and T T T 908).

FIGURE 3. Projections of a two-chain model of cellulose I (Valonia ventricosa) perpendicular to bc plane (Woodcock and Sarko, 1980).

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

179

Meyer and Misch (1937) postulated an antiparallel arrangement for the chains in the unit cell of native cellulose. This point has been contested many times and a parallel model is now more widely accepted for cellulose I (Gardner and Blackwell, 1974b; Sarko, 1973). There have been two possible types of parallel arrangement proposed for cellulose; parallel-up and parallel-down. This discussion was complicated by confusion over the actual denitions of parallel-up and parallel-down. According to Gardner and Blackwell (1974a) an `up' sheet consists of cellulose chains in which the z coordinate of O(5) is greater than that of C(5), as shown in Fig. 2, while a `down' sheet has all the cellulose chains with the z coordinate of C(5) greater than that of O(5). The z direction referred to above is that along the z axis of the three-dimensional environment in which the unit cell is built and is parallel to the c or chain axis direction (Fig. 2). The positive direction of the axis corresponds to increasing length of the c parameter of the unit cell. On this basis, Gardner and Blackwell declared that their model was `parallel-up' and differed considerably from the model of Sarko who stated that his model also was `parallel-up'. These two research teams were, however, considering different denitions for the a and b axes. For the work by Gardner and Blackwell the a axis corresponded to the intrasheet direction and the b axis corresponded to the intersheet direction. This was the reverse in the models put forward by Sarko's group. Models by both research groups were analysed by French and Howley (1989), who, using the Gardner and Blackwell denition for `up' but Sarko's usage of a and b axes found that Sarko's models (Sarko and Muggli, 1974; Woodcock and Sarko, 1974) were `parallel-up' and that of Gardner and Blackwell (1974a) was parallel-down'. This article was intended to clear up ambiguity in the use of the terms `up', `down', parallel and antiparallel. It stated that a unit cell may be said to contain parallel chains if the direction of the 1 3 4 link is the same in both chains, and antiparallel if one chain has the 1 3 4 link corresponding to the positive direction of the bre axis and the second chain has the 1 3 4 link in the negative direction of the bre axis. They represented the chain arrangements in a unit cell as seen in Fig. 4. In this diagram each chain end is shown as a rectangular box of which two corners are indicated by the letters A and B in a particular order. It can be seen that the difference between the `parallel-up' and `parallel-down' (Figs 4a and 4b, respectively) is in the direction of the glycosidic bond angle or, as stated by Gardner and Blackwell, in the order of the O5 and C5 atoms along the chain axis. Takahashi and Matsunaga (1991) suggested that the crystal structure differences among species (for example ramie, Valonia, cotton and wood) may be brought about by the order or disorder of the stacking of two kinds of sheets, up and down (Fig. 5). CELLULOSE I AND I Evidence has been put forward for the existence of more than one polymorph of cellulose in native samples. Simon et al. (1988a) postulated that a form of crystalline cellulose existed near the surface of a crystal which differed from the structure to be found at the centre of the crystal. These two crystalline forms were termed celluloses I and I (Atalla and Vanderhart, 1989). Celluloses produced by primitive organisms were said to have the I component dominant, while those produced by the higher plants have the I form dominant. I and I were found to have the same conformation of the heavy atom skeleton, but to differ in their hydrogen bonding patterns. Horii et al. (1987b)

180

O'SULLIVAN

BA (a) AB BA

BA (b) AB BA

AB

AB

AB

BA

BA

AB BA

AB

BA

AB

AB

AB BA (c) AB

BA BA

BA

AB

AB

BA

BA

BA

AB

AB

FIGURE 4. The x-y projection of the unit cells of cellulose packed with a) `parallel up', b) `parallel down' and c) antiparallel chains. The interactions between the AB points are different for the three types of packings.

suggested that the two 13 C NMR spectra obtained for polymorphs I and I correspond to the resonances for the two-chain and eight-chain unit cell regions of cellulose. Looking at the crystallographic bre diagrams of two forms of cellulose, Halocynthia and Microdictyon, a difference may be noted (Sugiyama et al., 1991a). The former gives a diffractogram which possesses a mirror of symmetry, while the latter does not. This was proposed to be due to the existence of two polymorphs of native cellulose; I, a meta-stable phase with a triclinic unit cell containing one chain and I, having two chains in its monoclinic unit cell. The proportion of IaI in Valonia was estimated to be 65:35 (VanderHart and Atalla, 1984). All the reections observed in the triclinic sample exist in the monoclinic sample but the reverse is not true. Thus, it was deduced that two crystalline phases were being observed. When the NMR spectral patterns of Valonia and Tunicin are examined (Belton et al., 1989; Sugiyama, 1992; Yamamoto and Horii, 1993), it may be seen that Tunicin has a simpler spectrum than Valonia. The latter contains two overlapping spectra, I and I. It is possible to calculate the relative percentage of each of the polymorphs present. Tunicin possesses only the I polymorph, and so gives the simpler spectrum where highly crystalline samples were used. Erata et al. (1995) found differences between the correlation peaks of I and I cellulose in Cladophora which indicate essential structural differences. The two distinct phases of native cellulose can be identied by electron diffraction (Sugiyama et al., 1991b) and infrared spectroscopy (Sugiyama et al., 1991a). No pure

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

181

sample of I has been found in nature. It has been claimed by Lee et al. (1994) that this meta-stable polymorph can be synthesized. Also, several groups (Yamamoto et al., 1989; Debzi et al., 1991) have shown that the polymorph I can be converted to the more stable I phase by annealing in various media. Annealing at 270 8C converts most of the I to the I form. Bacterial cellulose has the highest percentage of the I polymorph, that is 70%. The existence of I and I polymorphs in cellulose samples may affect the reactivity of native cellulose as I is meta-stable, and thus more reactive than I. Yamamoto and Horii (1993) found that the proportion of I cellulose varies from 64% in Valonia and bacterial cellulose to 20% in ramie and cotton cellulose. It was also found (Yamamoto and Horii, 1994) that the presence of carboxymethyl cellulose or xyloglucan decreases the content of I, and that I cellulose increases at low temperatures. Initial ambiguity in the interpretation of crystallographic data for native cellulose may be largely attributed to the fact that the samples examined were not pure, in that they contained more than one polymorph. As the percentage of IaI differs with sample type, the results obtained were not directly comparable. The co-existence of two polymorphs of native cellulose, which have different stabilities, may lead to the I polymorph being a site of initial reaction in a microbril. `AMORPHOUS' CELLULOSE Wide-angle X-ray scattering has been used to study possible structures for amorphous cellulose. Diffraction studies show light and dark areas along a cellulose microbril, which have been attributed to crystalline and amorphous cellulose, respectively. The simplest conceivable model has `straight' cellulose chains isotropically distributed in the sample (Fink et al., 1987). However, a bent and twisted chain was also suggested (Paakkari et al., 1989). In later research (Chanzy, 1990), it was shown that a slight curvature in the microbril brings specic domains in and out of the Bragg diffraction conditions, producing the successive bright and dark domains along the microbril axis. This implies that the cellulose microbril may be a continuous crystalline structure and negates the use of light and dark areas observed in the diffraction patterns as evidence for amorphous areas. Valonia is a highly crystalline cellulose and, thus, frequently the object of study. Of the 6001000 parallel chains which may exist in a microbril of Valonia, only 67% of the material is amorphous. Verlhac et al. (1990) suggested that the amorphous material consists mostly of surface chains, which in the large Valonia microbril makes up a small percentage of the total. This may be linked to Valonia's low reactivity, (Debzi et al., 1991) denoted by a degree of substitution of only 1%, when subjected to chemical micro-structural analysis (Verlhac et al., 1990). The reactant involved in this test was N,N-diethyl aziridinium chloride (DAC). For smaller microbrils, like wood, the percentage surface area is greater and approximately equals the percentage amorphous content: 30% surface hydroxyls versus 33% amorphous material. In the primary cell wall, the microbril is very small and 80% of the chains are on the surface, and so according to this theory, there would be approximately 80% amorphous material in these samples. Bacterial cellulose has a crystallinity of the order of 75% (Kulshreshta and Dweltz, 1973), and its crystals are around 56 nm in diameter. Assuming that these crystals have a nearly square cross-section, there are around 100

182

O'SULLIVAN

chains per crystal, with roughly 36 chains located at the crystal surface. Cotton is rated as being 4045% crystalline and as having a crystallite width of 45 nm (Morosoff, 1974). A degree of substitution of 2.4% was measured for cotton (Verlhac et al., 1990). This may be compared with the result for Valonia outlined above, i.e. 1% degree of substitution and more than 90% crystalline. This phenomenon is unique to cellulose I. The situation may be complicated, however, by the fact that triclinic cellulose I is metastable and may provide a more probable site of reaction. The surface of crystalline cellulose or indeed areas of `amorphous' cellulose probably still possess a degree of order. Thus `amorphous' cellulose cannot be considered truly amorphous as, by denition, an amorphous material is one which is formless or lacks a denite shape. CELLULOSE II X-ray and electron diffraction work has been carried out to clarify the unit cells of cellulose polymorphs IIV. 13 C CPaMAS NMR spectra (Horii et al., 1982; Dudley et al., 1983) show different signals for the C(4) and C(6) carbons of the polymorphs of cellulose. It was found that the polymorphs may be divided into two groups; those with a unit cell similar to that of native cellulose (I, IIII and IVI ) and those with a cellulose II type arangement (II, IIIII and IVII ). Thus, cellulose II has become an important focus of study, being the second most investigated form of cellulose. Preliminary work on cellulose II by Andress (1929) presented a two molecule unit cell (Fig. 6): a 0X814; b 0X914; c 1X03 nm; and 628. X-ray results for cellulose II are less reliable than those for cellulose I due to the large number of diffraction intensities which overlap each other (Buleon and Chanzy, 1978). Neutron diffraction studies (Ahmed et al., 1976) directed interest toward a larger unit cell containing eight molecules, with parameters: a 1X592; b 1X822; c (bre

UP UP DOWN UP UP UP

UP UP DOWN UP

DOWN

DOWN

FIGURE 5. Schematic representation for the stacking of the up- and down-sheet structures.

FIGURE 6. Projections of cellulose II bre unit cell (Sarko and Muggli, 1974).

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

183

repeat) 1.031 nm; 1178; and symmetry which was close to, but not exactly, P21 . Table 2 illustrates other values obtained for cellulose II unit cells. It must be remembered that cellulose II may be arrived at from cellulose I by two distinct routes, i.e. mercerization (alkali treatment) and regeneration (solubilization and recrystallization), and that even though the unit cells resemble each other closely, there are differences. The a parameter for the regenerated product of ramie cellulose has been quoted as 0.8662 nm and the mercerized ramie cellulose product as 0.8588 nm. The value of is greater for mercerized specimens than for most regenerated samples (Wellard, 1954). Also, as the degree of purity of samples used has been shown to be relevant, it should be noted that regeneration gives a higher level of conversion of cellulose I to cellulose II (Kolpak and Blackwell, 1976). Nyburg (1961) reported one example of naturally occurring cellulose II, in the genus Halicystis. Kuga et al. (1993) reported a mutant strain of Acetobacter xylinum as containing native folded-chain cellulose II. Another commonly studied aspect of cellulose II is the direction of the chains in the unit cell and the most prevalent view is that the two cellulose chains lie antiparallel to one another (Sarko and Muggli, 1974; Kolpak and Blackwell, 1976; Stipanovic and Sarko, 1976). CELLULOSE III Celluloses IIII and IIIII may be made reversibly from celluloses I and II, respectively and the polarity of the resultant cellulose chains has been stated to resemble that of the starting material (Sarko, 1978; Sugiyama and Okano, 1989). A hexagonal unit cell is reported for cellulose III. Sarko et al. (1976) published cell parameters for cellulose IIII as: a 1X025; b 0X778; c (bre repeat) 1X034 nm; and 122X48. Their model did not have strict P21 chain symmetry. Extensive research has been carried out on the reversible transformation of cellulose I into cellulose IIII by electron microscopy (Roche and Chanzy, 1981), packing analysis (Chanzy et al., 1987), transmission electron microscopy (Chanzy et al., 1986), solid state 13 C NMR (Sarko et al., 1976) and X-ray diffraction (Sugiyama and Okano, 1989). The existence of liquid crystal type assembly of cellulose was used to investigate the transformation of cellulose I to cellulose IIII through the cellulose I-EDA complex (Sarko et al., 1976; Roche and Chanzy, 1981; Reis et al., 1991). At the crystalline level, the conversion from Valonia I to Valonia IIII involved an extensive
TABLE 2. Parameters for selected cellulose II samples Sample used Mercerized bacterial Mercerized ramie Mercerized linen Viscose Viscose (Japanese) Cuprammonium

a(nm) 0.8014 70.970 0.8059 0.7911 0.7831 0.7955

b(nm) 0.9149 0.9219 0.9200 0.9134 0.9186 0.9167

1178341 1178461 1188151 117811 1168491 1168571

Meyer and Mark, 1928; Kolpak and Blackwell, 1976; Kolpak et al., 1978.

184

O'SULLIVAN

decrystallization and fragmentation of the cellulose crystals. During the conversion back to cellulose I, partial recrystallization took place but the distortion and fragmentation of the crystals was irreversible. Electron diffraction analysis showed that the unique, uniplanar-axial orientation of the crystalline cellulose microbrils (Fig. 7a) was irreversibly lost during the swelling (Fig. 7b) and washing (Fig. 7c) steps leading to cellulose IIII . Following washing in methanol, the microbril adopted a convoluted contour which contained small areas of crystallinity. The nal product cellulose I (Fig. 7d), produced by hydrothermal treatment, had increased surface area and thus increased activity and larger domains of crystallinity than seen in the sample depicted in Fig. 7c. Celluloses I and IIII were studied by 13 C NMR and it was indicated that a reduction in the lateral dimensions of the crystallites occurred during the transformation from the former to the latter polymorph (Sarko et al., 1976). Examination of the reversible complexation of Valonia cellulose with ethylene-diamine was carried out. This spectral study indicated that a conformational change occurs at C(6), during the cellulose I to complex transition. The peak corresponding to the C(6)OH is tg in cellulose I (65.7 ppm) and possibly gt in the cellulose I-EDA complex (62.2 ppm). In cellulose IIII the relevant peak occurs at 62.6 ppm. The regenerated cellulose I gave a spectrum which differed from that of the initial cellulose I. Electron diffraction studies show that the cellulose I-EDA complex has organized contours and non-uniform crystallinity, shown by a non-white area, whereas cellulose IIII has dened crystallinity (Fig. 7c). The variability of hydroxylmethyl conformation indicated above is interesting as it may also be used to study other transformations such as those occurring during mercerization of cellulose I to cellulose II. CELLULOSE IV Celluloses IVI and IVII originate from celluloses I and II, respectively. In the latter instance it is best to use regenerated cellulose (Zeronian and Ryu, 1987). Cellulose III is reported as being very nearly tetragonal but the conversion into cellulose IV is, in most cases, only partial thus causing difculties in obtaining reliable X-ray data (Buleon and Chanzy, 1980). One sample yielded the following parameters: a 0X8068; b 0X7946 nm; 908. In Gardiner and Sarko's analysis of celluloses IVI and IVII, (Gardiner and Sarko, 1985) both polymorphs were said to crystallize in almost the same orthogonal unit cell, with parameters for cellulose IVI: a 0X803; b 0X813; c (bre repeat) 1X034 nm and for cellulose IVII: a 0X799; b 0X810; and c (bre repeat) 1X034 nm. A P1 space group was suggested to be most probable for both polymorphs. ALKALI-CELLULOSE The majority view in the literature is that the polarity of cellulose chains differs between cellulose I and cellulose II. That means that cellulose I is parallel but cellulose II is antiparallel. In the mercerization process (the treatment of cellulose I with alkali, to achieve cellulose II) no solubilization occurs, which seems to imply that the brous structure of the cellulose would be maintained. In order to improve the understanding of this transformation, the changes that occur in cellulose during alkali treatment were examined (Okano and Sarko, 1985; Hayashi et al., 1989; Nishimura et al., 1991a, 1991b). It was found, by Okano and Sarko (1985), that ve unique alkali-celluloses could

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

185

be generated reproducibly and that they could be divided into two types, based on their crystallographic bre repeat (Okano and Sarko, 1984). Na-celluloses I, III and IV all exhibited a 1.0 nm repeat, while Na-celluloses IIA and IIB showed a 1.5 nm repeat and a three-fold helical chain conformation, not seen in crystalline celluloses. All the alkalicelluloses had a reasonable degree of crystallinity and a high degree of brous orientation. It is difcult to reconcile the claimed change in polarity with the mercerization process. As Na-cellulose I cannot be reconverted to cellulose I, it was deduced that Na-cellulose I possesses an antiparallel arrangement: the same chain polarity which is thought to exist in cellulose II. This may explain how cellulose I (parallel) can be converted to cellulose II (antiparallel) without solubilizing the cellulose. Hayashi et al. (1989) discussed in excess of nine polymorphs of alkali-cellulose, which could be formed from cellulose I, II andaor III. It was deduced from these results that the hypothesis of the change in chain conformation, as the cause of the irreversibility, is more probable than that of the chain arrangement. The problem with this argument is that the difference in the conformational energy between the two chain conformations is too small to bring about irreversibility. It has been stated that there are two types of microbrils in Valonia, that is those which have cellulose chains with their 1 3 4 glycosidic bonds uniformly in the positive z direction and those which have them uniformly in the negative z direction (Revol and Goring, 1983). This is in line with Figs 4a and 4b respectively. Cellulose microbrils are arranged in a criss-cross like arrangement, giving high strength to the material (Fig. 8). Diffraction studies of the cross-sections of microbrils showed individual black areas. These are supposed to correspond to microbril ends. The direction of the microbril can be determined by recording a diffraction diagram down the individual microbril (Kolpak et al., 1978). In a small bundle of microbrils various directions are observed (Fig. 9). The za z directions of the microbrils occur in a statistical
(a) (b)

(c)

(d)

FIGURE 7. Schematic representation of a Valonia microbril (MF) undergoing swelling in ethylene diamine (ETD): a) Initial single MF crystal (in black); b) Swollen MF, after uptake of one ETD molecule per glucose residue; c) Shrunken MF of Valonia III1 , resulting from the washing in methanol of (b); d) MF as in (c) but after the hydrothermal treatment leading to regenerated Valonia I.

FIGURE 8. Criss-cross arrangement of native cellulose.

186

O'SULLIVAN

One Microfibril

Second Microfibril

A Bundle of Microfibrils

FIGURE 9. A schematic representation of the end of single microbrils and of a bundle of microbrils as seen by a diffraction pattern.

distribution. This phenomenon may be used to explain solid-state conversion of native cellulose II during mercerization (Sarko, 1987). Rearrangement of the chains in the original unidirectional bundles of microbrils of cellulose I give mixed direction microbrils in cellulose II. Na-cellulose I was studied by X-ray crystallography (Nishimura et al., 1991a) and assigned a four chain unit cell with P2I space group (a 0X883; b 2X528; c (bre repeat) 1.029 nm; all angles 908). An antiparallel arrangement of chains was slightly favoured over a parallel one. Hydroxymethyl conformations were found to be tg in the `z' chains and gt in the `z' chains. DENSITY AND DEGREE OF CRYSTALLINITY The density of cellulose crystals, which may be determined by crystallography, is related to the structure of the substance. The density of crystalline cellulose, as found in a single crystal, is 1.59 g cm3 , whereas that of pure natural bre cellulose reaches only 1.55 g cm3 (Hermans, 1949). Hermans and Weidinger (1949) also found the so-called crystallinity derived from X-ray scattering was only 70% at most in native bres, while the remaining percentage of the cellulose was said to be `amorphous'. Degree of crystallinity may also be estimated by infrared spectroscopy on the basis of the relative height of certain bands (Fengel, 1992). Thus, the lower density found in natural bre may be attributed to the presence of amorphous or less ordered cellulose. CELLULOSE ULTRASTRUCTURE In plants, polysaccharides exist in the primary, secondary and tertiary cell walls. The tertiary cell wall contains a lower level of cellulose, being composed mainly of xylan. Primary and secondary walls differ in the arrangement of the cellulose chains. The former is less ordered and essentially composed of cellulose chains running in all directions within the plane of the wall. In the secondary cell wall, the cellulose chains are grouped in microbrils which are parallel, giving a more densely packed arrangement, and are aligned more or less with the bre axis. As technology has developed, the features of the cell wall have been observable in more detail. High resolution electron microscopy lattice imaging shows the crystalline organization of cellulose in wood brils. Negative staining (Heyn, 1966; Woodcock and Sarko, 1980) used in conjunction with electron microscopy shows occasional points along the bril at which the heavy metal stain seemed to penetrate: it was postulated

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

187

that non-glucose residues were located in these areas. This theory could be seen to be supported by the discovery, in 1961, that up to 50% of some sections of these native cellulose brils were composed of non-glucose units. However, these early workers had not isolated pure cellulose microbrils but were working with brils which were a combination of cellulose and hemicellulose (Dennis and Preston, 1961), and the nonglucose units are derived from the hemicelluloses. Breakdown of brils by chemical treatment (Dennis and Preston, 1961) yields colloidal celluloses which contain, at most, only traces of non-glucose sugars, even though these may be abundant in the parent brillar material. This is because the material isolated by chemical means are cellulose brils, almost free from hemicellulose contaminants. The resultant cellulose rods, which were measured as approximately 135 glucose residues in length, were thought to represent crystallites of the microbril. Despite development in microscopic techniques such as scanning electron microscopy (Fengel and Stoll, 1989), atomic force microscopy (Hanley et al., 1992) and scanning tunnelling microscopy (Kuutti et al., 1995) the chain length of such a large molecule as cellulose is still rather difcult to measure. The cellulose molecule in wood was found to be approximately 35005000 nm in length when determined by degree of polymerization, as calculated by weight of cellulose molecules isolated by a variety of methods (Fengel and Wegner, 1989), whilst the average size of each crystallite has been determined by X-ray analysis to be 220 nm in width and 217 nm in thickness (depending on the cellulose sample used). MODELS PROPOSED FOR CELLULOSE FIBRILLAR STRUCTURE The supermolecular structure of cellulose can be characterized by the degree of crystallinity, crystallographic parameters, crystallite dimension and defectivity, structural indices of amorphous domains, dimensions of brillar formation and other factors. As each of these has been studied in relation to celluloses, it is necessary to summarize the models which have been proposed for the structure of cellulose to see how the presently held views developed. Frey-Wyssling (1953, 1954) put forward a model for a microbril with several aggregated elementary brils (units of about 36 cellulose chains), also called micellar strands, which were embedded in paracrystalline cellulose, an unordered crystallized cellulose (Kratky and Mark, 1938; Nickerson, 1941). The existence of these elementary brils within the cellulose microbril was questioned and the smallest bril was reported to be where a microbril contains only one array of ordered cellulose chains rather that a set of arrays. Another model is the fringe micellar model (Astbury, 1933) which has completely ordered or crystalline regions, which, without any distinctive boundary, change into disordered or amorphous regions. In this model, a single molecule was thought to pass from one crystalline region to another through an amorphous area. Microbrils 1020 nm in width, visible by electron microscopy, were said to be combined in larger brils, or lamellae, to form bres. Disordered cellulose molecules, as well as hemicellulose and lignin, were found to be located in the spaces between the microbrils. These amorphous regions were oriented in the same direction as the cellulose microbrils (Preston and Cronshaw, 1958). In his early work, Fengel (1971) suggested a model for the ultrastructural organization of the cell wall components in wood which had several layers of hemicellulose molecules between

188

O'SULLIVAN

the brils (dimensions 12.0 nm) and a monomolecular layer of hemicellulose between the elementary brils. Lignin was envisaged as surrounding the total microbrillar system as shown in Fig. 10 (Fengel, 1971). Taylor and Wallace (1989) discussed the effect of the hemicellulose xyloglucan binding to the brils. The extent of the association between cellulose and xyloglucan is dependent on the source of cellulose (Hayashi and Maclachlan, 1984). Binding of xyloglucan has been suggested as a regulator of cellulose brillar size (Sasaki and Taylor, 1984). Elementary brils have been reported in recent literature (Lenz and Schurz, 1990), so the controversy about the arrangement within brillar cellulose continues. Crystallites of herbaceous plant cellulose and wood cellulose (Fink et al., 1990; Ioelovitch, 1992) have been determined to have widths in the order of 3.54.0 nm, while those for ax cellulose and cotton linters are 5.07.0 nm. Striations or marks on the surface of microbrils have been observed by microscopy and have been attributed to the fact that the microbril originated from smaller units or elementary brils (Gardner and Blackwell, 1971; Okuda et al., 1994). Bourret et al. (1972) and Chanzy (1990) have disputed this fact. Enzymic hydrolysis (Chanzy and Henrissat, 1983) of Valonia cellulose divided the microbrils into longitudinal crystalline sub-elements with widths ranging from below 2 nm to the full size of the initial Valonia microbril (1418 nm) (Bourret et al., 1972). This span of sizes was taken as evidence against elementary brils as no size was seen to be prevalent. The width of the Valonia microbril was shown to correspond to the width of the crystallite, whose length was determined to be above 100 nm, without any longitudinal periodicity. Tsekos et al. (1993) showed by electron microscopy that, in a number of algae species, microbrils may consist of two,

FIGURE 10. Cross-sectional model of the ultrastructural organization of the cell wall components in wood (Fengel, 1971).

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

189

three or four linear subcomponents (elementary brils). In some species, two or three microbrils were seen to be bundled together. Further electron microscopy studies by Fujino and Itoh (1994) on algal cell walls suggested the existence of layers of amorphous cellulose brils (810 nm) and of crystalline microbrils (1517 nm) and cross-bridges of approximately 24 nm in diameter. Thus, there appears to be no agreement as to the size and nature of microbrils, perhaps due to a dependence of cellulose structure on the sample source. Terminal complexes (TCs) of various sizes have been observed by X-ray and electron diffraction (Okuda et al., 1994) and by microscopy (Tsekos et al., 1993). It was thought that the number of TC subunits could be linked to the size of the microbril produced. Membrane tetrads (Tsekos et al., 1993) have also been identied, membrane-bound multi-enzyme complexes which participate in the synthesis of matrix polysaccharides. Looking at the microbril cross-sectional end shape, Valonia was found to be nearly square, as dened by DC TEM data (Chanzy, 1990). By this method, wood pulp (Chanzy, 1987) and ax (Naslund et al., 1988) bres have also been studied and distinct microbrils revealed, but the image could not be sufciently resolved to display their cross-sectional shape with certainty. Only in a few instances have nearly square sections been seen. From this information it was found that the lateral widths of Valonia, Tunicin, ramie, wood and primary cell wall samples of cellulose were 20, 10, 5, 34 and 1.82.0 nm respectively (Chanzy et al., 1979; Chanzy and Vuong, 1985; Chanzy, 1987, 1990; Kuga and Brown, 1987). However, Fink et al. (1995) found that, except for bacterial cellulose, crystallite sizes measured by wide angle X-ray scattering (WAXS) are distinctly smaller than those obtained from electron microscopy and suggests that it is necessary to consider lattice distortions and the presence of two polymorphs in native cellulose samples. Jacob et al. (1994) found that wood brils have dimensions which are remarkably constant: laterally 2X5 0X14 nm. Early workers expected the cellulose chains to have a coiled structure similar to other polymers. Bittiger and Husemann (1964a, 1964b) investigated the possible folding of cellulose chains and found that samples with a degree of polymerization of 700, 420, 162 and 120 formed rodlets upon precipitation, with a distinct length of 70 nm. This length corresponds to an extended chain length of about 135 glucose units and was taken to indicate that the molecules would be folded at regular intervals, independent of their molecular weight. This theory presumes that the precipitation process causes breaks in the chain at the folds. Mark and coworkers (Mark, 1967; Mark et al., 1969) came to the conclusion that the structural carbohydrate must be aggregated into continuous extended-chain laments of nearly perfect crystallinity. Such a conformation was said to be the most thermodynamically stable, and for this reason the molecules would tend to assume this form (Lindenmeyer, 1965). The folded-chain crystal represents a meta-stable conformation and the same polymer was said to be able to exist in the extended or folded forms, depending on the conditions of crystallization. These facts seem to indicate the possibility of a folded conformation in cellulose derivatives but not in native cellulose. On the basis of experimental evidence, straight cellulose molecules were identied in ramie bres (Muggli et al., 1969). Mark (1967, 1971) reviewed various cellulose chain models from a mechanical stand-point and found that extended-chain conformations yielded axial stiffness constants that are compatible with the experimental observations. However, a variety

190

O'SULLIVAN

of folded-chain conformations yielded maximum stiffnesses incompatible with experimental results. Sarko and Marchessault (1969) offered a further opinion on the presence of folded chains. They suggested that the folded chain occurs during the nucleation mechanism of crystallization, which is kinetically controlled. The suggestion was that a non-folded cellulose chain model could be one which excluded the possibility of crystallization. This theory was disproved in later studies, when computer based energy analysis was added to the techniques used to examine cellulose. A single straight, extended chain was found to be thermodynamically less stable, but giving rise to a crystal structure of lower energy. More recently, it has been reported (Wolters-Arts et al., 1993; Emons, 1994; Emons and Kieft, 1994) that cellulose in plant cell walls exhibited a helicoidal structure. Hermans and Weidinger (1949a) suggested that discontinuities in the brillar structure could be explained by the presence of dislocations (Fig. 11), i.e. disorders in a nearly perfectly crystalline lattice but this model has not yet been substantiated. For this to be true breaks in cellulose chains would be required or chains within a microbril would need to be of non-uniform length. Rowland and Roberts (1972) proposed that cellulose microbrils consist of completely crystalline areas with various types of surface imperfections, such as distorted surfaces and twisted or strained regions of the crystalline elementary brils. Cellulose chain orientation relative to the microbril surface and to the cell wall has been studied by diffraction studies (Frey-Wyssling and Muhlethaler, 1951; Wellard, 1954; Tsekos et al., 1993) over many decades. These various studies lead to similar conclusions. All the cellulose chains lie parallel to the 020 plane but it is the 110 lattice plane which is parallel to the microbril surface and to the plasma membrane surface of the cell (Fig. 12). When reading literature on this subject it is important, yet again, to recall the change in the convention of labelling the unit cell parameters, as discussed above. HYDROGEN BONDING Whether native cellulose consists of elementary brils or not, its ultrastructure is largely due to the presence of covalent bonds, hydrogen bonds and van der Waals forces. Hydrogen bonding within cellulose chains may act to determine the `straightness' of the chain. Interchain hydrogen bonds might introduce order or disorder into the system, depending on its regularity. When considering hydrogen bonding, it is essential to note the conformation of the C(6) hydroxymethyl group. There are three possible minimum energy orientations for this substituent to the pyranose ring; trans-gauche (tg), gauche-trans (gt) and gauchegauche (gg), as illustrated in Fig. 13 (Shefter and Trublood, 1965). The reason for the difference in stability between these three staggered conformers is the relative proximity of the oxygen and carbon substituents. Both cellulose I and cellulose II have been studied in depth to decipher any pattern in their hydroxymethyl groups which may exist. Sundaralingam (1968) reported that the trans-gauche (tg) conformation had not been observed in crystal structures as this conformation had a higher energy than the gt or gg orientations. He disputed its inclusion in models for cellulose I proposed by Petitpas and Mering (1956). Of the two remaining possibilities, it was found that the gt

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

191

3.5 nm

020

110

FIGURE 12. Lamination of crystallized cellulose I along plane 110.

FIGURE 11. The structural model of the elementary bril of cellulose with chain dislocations (Hermans and Weidinger, 1949a).
H<6 C<4>
2>

O<6> O<5> C<4> O<5> C<4>

H<6

2>

O<5>

H<6

1>

O<6> H<5> (a)

H<6

1>

H<6 H<5> (b)

2>

O<6> H<5> (c)

H<6

1>

O<6> H<6
2>

H<6

1>

C<4>

O<5>

FIGURE 13. The three most probable rotational positions of the hydroxymethyl group: (a) gt, (b) gg, (c) tg.

192

O'SULLIVAN

conformation was much more likely to occur than the gg. Although the gt conformation is favoured in mono- and disaccharides such as cellobiose (Brown, 1966), this cannot be taken to preclude the occurrence of the tg orientation in the cellulose polymer, if it allows additional stability through hydrogen bonding. Models of cellulose with each of the three orientations (tg, gt and gg) were compared to X-ray data to obtain the best t. This t may be described as the model with the lowest reliability factor (R). For cellulose I, a tg orientation gave the best t (Gardner and Blackwell, 1974a) with the R value for tg, gt and gg models being 0.242, 0.292 and 0.349, respectively. For cellulose II, both regenerated (Buleon and Chanzy, 1978) and mercerized (Kolpak and Blackwell, 1976) forms were reported to have a gt conformation throughout the cellulose chain at the corner of the unit cell, but a tg arrangement for the centre chain. However, some rotation about the suggested minimum orientations (tg, gt and gg) was accepted as being possible. The majority view, expressed in the literature, is that cellulose I has a tg conformation and cellulose II has a gt conformation throughout the chains (Woodcock and Sarko, 1980; Okamura, 1991). Derived from the study of hydroxymethyl orientations, intra- and inter-chain hydrogen bond patterns were suggested for cellulose I and II. The former has two intramolecular hydrogen bonds at (O)5-(OH)31 and (OH)2-(O)61 and an interchain hydrogen bond between (O)6-(O)311 as shown in Fig. 14 (Tsuboi, 1957; Mann and Marrinan, 1958; Liang and Marchessault, 1959; Marchessault and Liang, 1960). Cellulose II is reported as having intrachain hydrogen bonding at (OH)3-(O)51 and an intermolecular hydrogen bond at (OH)6-(O)211 for corner chains and (OH)6-(O)311 for centre chains. An extra dimension was also added to the hydrogen bonding in cellulose II over cellulose I, in the form of an inter-sheet interaction between (OH)2 (corner chain) (O)211 (centre chain), absent in native cellulose. The lengths of the hydrogen bonds in cellulose I are reported to be 0.275 nm (Tasker et al., 1994). Fengel et al. (1995) found that the transformation from cellulose I to II is determined by the splitting and new formation of inter- and intramoleculear hydrogen bonds. A further consideration, in view of recent evidence, is the conformation which exists in each of the native cellulose polymorphs I and I. Molecular dynamics simulations carried out by Heiner et al. (1995) indicate that the I has a tg hydroxy-methyl
(a) O O O O O O (b) O O O O<3> O<5> O<6> O O<6> O O O<5> O O O<3> O O O O O O<5> O<6> O O O<2> O<3> O O<3> O O O<2> O O<6> O<3> O O<5> O O O

c b

FIGURE 14. Hydrogen bonding pattern for (a) cellulose I and (b) cellulose II.

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

193

conformation in more than 90% of cases (92.0 and 99.8 for planes of a cellulose model). For the I polymorph, 81.0% of hydroxymethyl groups were found to be tg. In each situation the remainder of the percentage is represented by a mixture of the gg and gt conformations. This agrees with carbon NMR studies carried out by Yamamoto and Horii (1993). Cellulose IIII has intramolecular hydrogen bonds between (OH)3-(O)5I and (O)2(OH)61 and intermolecular associations between (O)3-(O)611 as in cellulose I. For celluloses IVI and IVII, in addition to the usual two intramolecular hydrogen bonds present in most of the other crystalline celluloses, both polymorphs seem to be well established by the intermolecular hydrogen bonding along the 020 plane. Differences in the hydrogen bonding patterns reported for models of cellulose I and II are not solely derived from deviations in hydroxymethyl conformation but also from the fact that the polarity of the chains are popularly thought to differ; a parallel arrangement (Gardner and Blackwell, 1974a) is attributed to cellulose I and an antiparallel arrangement is attributed to cellulose II. Intersheet associations are further affected by the fact that a relative shift of the centre and corner chains, along the bre axis, has been observed (Gardner and Blackwell, 1974a). This amounts to ca4, or half a ring in length, for cellulose I or II. Cellulose III is reported to show a smaller shift of 0.09 nm (Sarko et al., 1976), as compared with 0.26 nm (ca4). COMPUTER ANALYSIS OF CELLULOSE Computer modelling has been used to study cellulose for many decades and some of the results are discussed below. Many of the calculations carried out on cellulose have used varying levels of constraint to reduce the computing power and time required to obtain a result. This method of analysis has been used in combination with more conventional methods of analysis to test the viability of proposed models. Due to the small amount of diffraction data available for cellulose in comparison to the number of atoms that must be located in each of three dimensions, it has been the practise to construct a stereochemically reasonable polymer model. The diffraction intensity information is then used to locate the chain within the unit cell and to determine the actual position of the hydroxymethyl groups (French and Murphy, 1977). Several chain arrangements were generated in an effort to reduce the reliability factor (R) for X-ray data. Attempts were made to produce a result which tted in with expectations for cellulose, based on information possessed for related structures. X-ray diffraction of single crystals is the only method that can provide, denitively, bond length, bond angles and characteristic geometry for compounds (Henrissat et al., 1987). However, in the study of polymers by diffraction methods a proposed model has to be created and hence the result is not absolute. Therefore, it is important to refer to single crystal studies that provide data on which to base cellulose models. By adding a mathematical approach to the quest for the `best t', proposed models were analysed by computer based stereochemical modelling (Tasker et al., 1994). Statistics were applied to determine the probabilities of the chain conformations and arrangements occurring. As with all methods used for the structural determination of cellulose, the results of computer analysis cannot be termed absolute, as they depend on the level of sophistication of the instrumentation and the quality of the algorithms used.

194

O'SULLIVAN

CONSTRAINTS AND ASSUMPTIONS USED DURING MODEL BUILDING Despite the signicant progress that is being made in broadening the scope of molecular modelling and reducing the approximations that are included in calculations and models used, there are still limits to what can be achieved. Polysaccharides are multi-atom systems, even when disaccharides or oligosaccharides are considered. A cellulose microbril comprises more atoms than can be dealt with by the commercial packages available at present. Thus, only sections of the sytem can be modelled and constraints are often placed on the amount of freedom allowed in the variation of bond lengths, bond angles and torsion angles. Stereochemical modelling of cellulose was limited, initially, by the power of the computers available and by the lack of suitable force elds. Glucose and cellobiose have both been considered monomers for the polymer, cellulose. Structural aspects which were considered in the computer generation of cellulose chains included the straightness of chains, puckering of rings, glycosidic bond angles, orientation of chains, hydroxymethyl conformation and unit cell parameters. A major assumption in the construction of polysaccharide models was that the geometry of sugar monomers is very similar to that which exists in saccharide oligomers and polymers. French and Murphy (1977) found that the bond lengths and bond angles of -glucose residues are relatively constant but that the conformation angles show a variation of as much as 128. Glucose has been found to be an unsuitable monomer for cellulose as it does not comprise the (14) linkage that exists in cellulose. Thus, cellobiose and derivatives thereof have been more commonly employed. Using cellobiose crystallographic data as a basis for cellulose models added uncertainty. Cellobiose has its hydroxymethyl group in a gt conformation (Bittiger and Husemann, 1964b). Thus, if the calculation does not have the ability to consider several different conformations and arrangements, the results will be inherently inaccurate, unless the structure of the cellulose polymorph to be studied has the same conformation as cellobiose. Early workers did not consider the possibility of differences in the basic ring structure occurring during polymerization of cellobiose. Thus, the bond angles and bond lengths of their models were xed at those of cellobiose throughout the calculation, not allowing for any puckering of the pyranose ring. DEVELOPMENT OF FORCE FIELDS AND ENERGY MINIMIZATION To survey the development of stereochemical analysis of cellulose, one may begin with the work of the 1960s to improve minimization functions (Powell, 1964). This allowed results to be produced faster and with greater reliability for less constrained models. It was necessary to direct the mathematics towards measuring specic chemical energies which determine stabilities and predict the relative probabilities of different conformations occurring. Scott and Scheraga (1965) developed a method of calculating internal rotation barriers by considering exchange interaction of electrons and non-bonded or van der Waals interactions by a semi-empirical method. Allinger (1977) introduced an improved force eld for molecular mechanics calculations of the structure and energies of hydrocarbons. This is known as Molecular Mechanics 2 (MM2). Further force elds, such as MMP2 (Sprague et al., 1987), MM3 and UFF (Rappe et al., 1991) have since been devised. The Universal Force Field (UFF) can deal with complex systems which

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

195

cannot be handled with the Molecular Mechanics (MM) packages but it is not as accurate at dealing with simple hetero-atomic molecules. As, initially, the functions developed were for less complex molecules, it was thought prudent to attempt a prediction of cellulose structure by working out models for smaller molecules. One example of this was the study of -D-cellotetraose (Poppleton and Mathieson, 1968). Arnott and Scott (1972) applied the linked-atom constrained least-squares approach to the analysis of the structure of polysaccharides. In this approach, it was found to be acceptable to keep the bond lengths and bond angles xed. The parameters used were obtained from X-ray studies of the polysaccharides or the monomers involved. It was concluded that the variability of the glycosidic bond angles was not greater than that of the ring bond angles. Rees (1970) stated that the torsional angles at each glycosidic linkage in carbohydrate oligomers and polymers may be used to predict their overall shape. There are two of these variables in cellulose: the dihedral angles and . It was also found that the exibility of the glycosidic system is determined by the equatorial substituents on both residues which are next to the glycosidic oxygen (Rees and Scott, 1971). The use of oligomer analysis has continued to be important in trying to elucidate the macromolecular cellulose structures. An example of this is the work of Gessler et al. (1995) which drew parallels between -D-cellotetraose hemihydrate and cellulose II structures. CHAIN POLARITY AND HYDROXYMETHYL CONFORMATION In examining the deductions obtained from computer based stereochemical analysis on X-ray data of cellulose, it must be remembered that the models contained constraints and assumptions. Rees and Skerrett (1968) examined the models previously proposed by Meyer and Misch (1937) and Hermans (1946b): a `straight' and a `bent' chain model. It was concluded that the latter was the most likely, on the basis of van der Waals energy. In arriving at this conclusion, the pyranose ring conformation was constrained and the hydroxymethyl group was treated as equivalent to a methyl group in its van der Waals interactions. Model-building calculations carried out on cellodextrin (Rees and Skerrett, 1970) showed that the only important steric contacts were with adjacent residues, and these favoured folding of crystalline cellulose chains. Straight-chain cellulose is however a more acceptable model (Mark, 1967, 1971; Mark et al., 1969; Muggli et al., 1969; Sarko and Marchessault, 1969). By analysing models for cellulose produced by earlier workers, many discrepancies in these models were noted. For example, Pizzi and Eaton (1987) noted that the model by Kolpak et al. (1978) was much more stable than that proposed by Sarko. French and Murphy (1977) studied the Gardner and Blackwell models (1974a, 1974b) for Valonia, which were originally stated to be `parallel-up' and `parallel-down' (French and Howley, 1989). Specic conformations and orientations (Stipanovic and Sarko, 1976) were proposed for cellulose I and cellulose II, one example being triclinic unit cells with all the hydroxymethyl groups in or near the tg position. Native cellulose was found to possess parallel chain polarity, and antiparallel was found to be consistent with cellulose II. Gardner and Blackwell (1974a, 1974b) agreed with the chain polarity and conformation assigned to cellulose I, and specied a P21 space group which is a monoclinic unit cell.

196

O'SULLIVAN

Cellulose II (Andress, 1929; Sarko and Muggli, 1974; Buleon and Chanzy, 1978) was stated by other workers to have antiparallel chain polarity, with the two chains having different hydroxymethyl conformations (centre chain tg and corner chain gt). Most research groups came to accept that parallel and antiparallel chain polarities corresponded to cellulose I and cellulose II respectively and went on to look at other aspects of cellulose. However, as the chain sense and conformations of cellulose have not been conclusively proven, there is still scope for investigation. Kroon-Batenburg et al. (1996) published a study which indicated that it is not possible to disregard the possibility of a parallel cellulose II or to presume that there is only one feasible conformational pattern for celluloses I and II. D-Glucose (Brady, 1986), native cellulose and regenerated cellulose (Pertsin et al., 1986) have been studied by molecular mechanics. In dealing with cellulose the parameters for glucose, as obtained from X-ray crystallography, were used to create a monomer. This was polymerized and packed in accordance with the symmetry assigned to cellulose in previous work (Nyburg, 1961; Gardner and Blackwell, 1974b). Such considerations included the investigation of two chains, with a two-fold screw axis, arranged in a unit cell of P21 symmetry; one chain was positioned at the centre and one at the corner. Various arrangements, for example parallel or antiparallel chain orientations, were considered and the minimum energy calculated. The best crystal models of cellulose II were approximately 1.5 kcal mol1 lower in energy than those for cellulose I. Computer modelling of cellulose by molecular and brownian dynamics methods was discussed by Khalatur et al. (1986). The chemical bond lengths and bond angles were assumed to be the same as for the cellobiose molecules in the crystal state, and the former were not allowed to vary beyond certain constraints along the eight ring fragments (168 atoms) in the model being studied. The energies of certain conformations, and the possible interchanging of conformations, were investigated. The conclusion was that conformational changes in the macromolecule as a whole occur very slowly. Khalatur et al. (1986) favoured eight ring chain models for regenerated and mercerized cellulose: the hydrogen bonding patterns showed no signicant differences and these models had the cellulose chains arranged in an antiparallel sense. However, for cellulose I the best models (`parallel-down' and antiparallel) were very close in energy and possessed statistically equivalent RII factors. Pertsin et al. (1986) favoured the antiparallel model for cellulose I, in view of its similarities to the cellulose II model, thus facilitating conversion of cellulose I to cellulose II without solubilization. Studying the possible conformations for a segment of cellulose chain, six glucose units in length, it was found that four slightly twisted chain minima had very similar conformational energies (Simon et al., 1988b), upon minimization, and each facilitated intramolecular hydrogen bonding. However, the possibility of an innite chain folding back on itself was not ruled out. Two straight-chain segments of ten disaccharide units each, in an antiparallel arrangement, provided enough attractive interactions to induce a sharp fold in cellulose II (Fig. 15) which could allow the possibility for transition from a parallel cellulose I to an antiparallel cellulose II. The controversy about the hydroxymethyl conformation and orientation of the chains continued. Miller and Li (1989) investigated a two molecule unit cell for cellulose I and deduced that the O(6) atoms have a gt conformation on the corner chains and a tg

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

197

Cellulose I

Cellulose II

FIGURE 15. Schematic representation of successive stages of the transformation of antiparallel cellulose II from a parallel cellulose I.

orientation on the centre chains. Work published at the same time (Millane and Narusiah, 1989), supported a crystal structure consisting of a parallel arrangement with all primary hydroxyl groups in the tg conformation. Other workers (Takahashi and Matsunaga, 1991) suggested that the crystalline structure of cellulose was not uniformly `parallel-up' but a mixture of `parallel-up' and `parallel-down'. Kroon-Batenburg et al. (1996) found that microbril models with various tg and gt conformations patterns were stable. This variety of possible cellulose microbril systems would aid the rationalization of the mercerization process, that is, cellulose I to cellulose II without solubilization. Pizzi and Eaton (1985a, 1985b) identied a cause and mechanism for termination of the crystalline zone of cellulose I from their study of ve chain crystals of cellotetraosides. They attributed this feature to localized variation in the hydrogen bond pattern, and also proposed a helicoidal conformation for `free' or amorphous cellulose. The merit of cellobioside over cellobiose as a monomer for cellulose I was also commented on. MOLECULAR DYNAMICS AND ITS APPLICATION With the advent of molecular dynamics (Khalatur et al., 1986) a greater level of freedom within the model could be allowed but bond length and certain bond angle constraints were retained by some workers to speed up the calculations involved. It has been shown that the geometry of pyranose rings varies with linkage conformation (Henrissat and Chanzy, 1985; Simon et al., 1988a; Millane and Narusiah, 1989; Miller and Li, 1989). Thus, forcing the rings in modelled cellulose to retain the dimensions they hold in cellobiose is erroneous (Frey-Wyssling and Muhlethaler, 1951) and would make surface simulation difcult due to the disregard to end-effect implicit in this practise.

198

O'SULLIVAN

Molecular dynamics simulations carried out by Kroon-Batenburg and Kroon (1992) on cellulose II models worked on the premise that all the primary hydroxyl groups are in the gt conformation. Kroon-Batenburg and Kroon (1990) utilized a computer modelling package GROMOS (van Gunsteren) to perform molecular dynamics (MD) simulations of methyl--D-glucoside in water. This study encompassed the stability of the tg conformation of the hydroxymethyl group in isolated molecules of methyl--Dglucoside and methyl--D-cellubioside and the decrease in probability of observing this conformation in polar solvents. In the GROMOS force eld, CH and CH2 groups are treated as united atoms and no special hydrogen-bond potential is included, in order to simplify the calculations. The parameters used were taken from work by Koehler et al. (1987) and in addition, van der Waals parameters were applied throughout the calculations. Only the 4 C1 chair conformation was considered, and during simulation no transitions to other shapes occurred. For both the tg and gt conformations the global minima were obtained by varying the conformations of the hydroxy groups and the exoanomeric torsional angle. A box was constructed containing the molecule of interest, in a particular conformation, with 242 water molecules. The water molecules were kept rigid by using the SHAKE method (van Gunsteren and Berendsen, 1977). The conclusions drawn from this work were that cellulose I had hydroxymethyl groups in a tg conformation, allowing it to form two intramolecular hydrogen bonds and that cellulose II had a gt conformation, allowing only one intramolecular hydrogen bond. Molecular dynamics is now used commonly in most simulation studies but results indicate that, even for the simplest of systems, many picoseconds are needed to obtain acceptably low estimated standard deviations (Hooft et al., 1992). For large systems such as cellulose microbrils, it may be necessary to allow molecular dynamics calculations to run for several weeks of computing time to obtain satisfactory results. Thus, the use of approximations and constraints in all or parts of a structure are still common. The merit of molecular dynamics on an unconstrained system is that it allows the system to relax. ADVANCED MODELS OF CELLULOSE A structure for the single-chain I cellulose unit cell was built by French et al. (1993). The existence of two polymorphs of native cellulose i.e. I and I was also investigated by Heiner et al. (1995). They reported the monoclinic (I) phase was 8.7 kJ mol1 cellobiose1 more stable than the triclinic phase. This agrees with the ndings of Aabloo et al. (1994), which were obtained by using both the rigid-ring method PLMR and the full-optimization molecular mechanics program MM3(90). Reisling and Brickmann (1995) devised a model for a cellulose chain consisting of -glucose hexamers. Models have been built for sections of native crystalline cellulose, for example that of French et al. (1993) which contained 24 to 32 monosaccharide residues and that of O'Sullivan (1995) which was a 16 cellulose chain model of a cellulose microbril. Non-bond inter- and intramolecular interactions have been calculated for cellulose molecules and similar systems to determine what the stabilizing forces are in a cellulose microbril. It has been stated by Kooijman et al. (1992) that crystal packing of a compound is determined solely by intermolecular interactions. While Pizzi and Eaton (1984, 1987) found that hydrogen bonding and van der Waals forces are the predominant factors in the stabilization of the minimum energy conformations of

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

199

cellobiose and methyl--cellobioside and of the crystalline zone for cellulose II and cellulose I. They also found that cellulose II was more stable than cellulose I. Due to the importance of non-bond interactions in determining minimum energy conformations it is vital that they are included in the force eld that is used for modelling cellulose. However, as the magnitude of such forces decrease greatly with increases in interatomic distances, the interaction energy between two atoms may be calculated when the interacting atoms lie within a specied cut-off range, thereby reducing computation time signicantly. Heiner et al. (1995) produced models of I and I cellulose using the Gromos 87 force eld on which hydroxymethyl conformations were carried out. Molecular modelling may be used to study physical properties of cellulose as shown in the work of Leeang et al. (1992), who studied the elastic modulus of cellulose I and II by molecular mechanics. Computer simulation is now used in conjunction with many other methods of analysis such as diffraction studies as discussed above, NMR spectroscopy (Henrissat et al., 1987; Heiner et al., 1995) and surface analysis (O'Sullivan, 1995; Woodcock et al., 1995). CONCLUSION From the above discussion it is obvious that the structure of cellulose is complex and its study requires the consideration of many factors. Investigation into the many aspects of cellulose structure is being continued by research groups worldwide and new results appear in the literature constantly. This trend shall continue as more is learned about cellulose by the application of an increasing number of techniques. All knowledge gained about cellulose structure is vital to life today as cellulose and its derivatives are used in so many industrial applications. ACKNOWLEDGEMENTS I would like to thank the BioComposites Centre, Bangor for funding this work and Profs M. S. Baird and W. B. Banks for many helpful discussions. REFERENCES
Aabloo, A., French, A. D., Mikelsaar, R. H. and Pertsin, A. J. (1994) Studies of crystalline native celluloses using potential-energy calculations. Cellulose 1, 161168. Ahmed, A. U., Ahmed, N., Aslam, J., Butt, N. M., Khan, Q. H. and Atta, M. A. (1976) Neutron diffraction on studies of the unit cell of cellulose II. J. Pol. Sci., Polymer Letters Edition 14, 561564. Allinger, N. L. (1977) Conformational analysis 130. MM2. A hydrocarbon Force Field utilising V1 and V2 Torsional Terms. J. Am. Chem. Soc. 99, 81278134. Andress, K. R. (1929) The X-ray diagram of mercerized cellulose. Zietschrift Physikalische Chemie Abstracts B 4, 190206. Arnott, A. and Scott, W. E. (1972) Accurate X-ray diffraction analysis of brous polysaccharides containing pyranose rings. Part 1. The linked-atom approach. Journal of the Chemical Society Perkin II 324335. Astbury, W. T. (1933) Some problems in the X-ray analysis of the structure of animal hairs and other protein bres. Transactions of the Faraday Society 29, 193211. Atalla, R. H. and Vanderhart, D. L. (1989) Studies on the structure of cellulose using Raman

200

O'SULLIVAN

spectroscopy and solid state 13 C NMR. In Cellulose and Wood: Chemistry and Technology, Proceedings of the tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, 169187. Belton, P. S., Tanner, S. F., Cartier, N. and Chanzy, H. (1989) High-Resolution solid-state 13 C NMR spectroscopy of Tunicin, an animal cellulose. Macromolecules 22, 16151617. Bittiger, H. and Husemann, E. (1964a) Electron microscopic investigation of single-molecule crystals of cellulose tricarbanilates. Die Makromoleculare Chemie 75, 222224. Bittiger, H. and Husemann, E. (1964b) Electron microscopic investigation of the formation of monomolecular cellulose tricarbanilate crystals. ibid. 80, 239241. Blackwell, J. (1982) The macromolecular organization of cellulose and chitin. In Cellulose and other natural polymer systems (R. M. Brown, Jr., ed.). New York: Plenum Press. Blackwell, J. and Marchessault, R. H. (1971) Infrared spectroscopy of cellulose. In Cellulose and cellulose derivatives (N. Bikales and L. E. Segal, eds) New York: Wiley-Interscience. Bourret, A., Chanzy, H. and Lazaro, R. (1972) Crystallite features of Valonia cellulose by electron diffraction and dark-eld electron microscopy. Biopolymers 11, 893898. Brady, J. W. (1986) Molecular dynamics simulation of -D-glucose. J. Am. Chem. Soc. 108, 8153 8160. Brett, C. and Waldron, K. (1990) In Physiology and Biochemistry of Plant Cell Walls London: Unwin Hyman, p. 72. Brown, C. J. (1966) The crystallite structure of sugars. Part V. A three-dimensional analysis of methyl -xyloside. J. Chem. Soc. A, 922927. Buleon, A. and Chanzy, H. (1978) Single crystals of cellulose II. J. Polymer Sci.: Polymer Physics Edition 16, 833839. Buleon, A. and Chanzy, H. (1980) Single crystals of cellulose IVII: Preparation and properties. ibid. 18, 12091217. Chanzy, H. (1987) Proceedings of the International Symposium on Wood and Pulp Chemistry 1, 235242. Chanzy, H. (1990) Aspects of cellulose structure. In Cellulose Sources and Exploitation: industrial utilisation biotechnology and physico-chemical properties (J. F. Kennedy, G. O. Phillips and P. A. Williams, eds). Chichester, UK: Ellis Horwood, pp. 312. Chanzy, H. and Henrissat, B. (1983) Electron microscopy study of the enzymic hydrolysis Valonia cellulose. Carbohydrate Polymers 3, 161173. Chanzy, H. and Vuong, R. (1985) In Polysaccharides, Topics in structure and morphology (E. D. T. Atkins, ed.). Basingstoke: MacMillan, p. 50. Chanzy, H., Imada, K., Mollard, A., Vuong, R. and Barnoud, F. (1979) Crystallographic aspects of sub-elementary cellulose brils occurring in the wall of rose cells cultured in vitro. Protoplasma 100, 303316. Chanzy, H., Henrissat, B. and Vuong, R. (1986) Structural changes of cellulose crystals during the reversible transformation cellulose I 3 IIII in Valonia. Holzforschung 40, 2530. Chanzy, H., Henrissat, B., Vincendon, M., Tanner, S. F. and Belton, P. S. (1987) Solid-state C-13NMR and electron microscopy study on the reversible cellulose I 3 cellulose IIII transformation in Valonia. Carbohydrate Res. 160, 111. Claffey, W. and Blackwell, J. (1976) Electron diffraction of Valonia cellulose Quantitative interpretation. Biopolymers 15, 19031915. Davis, W. E., Barry, A. J., Peterson, F. C. and King, A. J. (1943) X-ray studies of reactions of cellulose in non-aqueous systems. II. Interaction of cellulose and primary amines. J. Am. Chem. Soc. 65, 12941300. Debzi, E. M., Chanzy, H., Sugiyama, J., Tekely, P. and Excofer, G. (1991) The I 3 I transformation of highly crystalline cellulose by annealing in various mediums. Macromolecules 24, 68166822. Dennis, D. T. and Preston, R. D. (1961) Constitution of cellulose microbrils. Nature 191, 667668.

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

201

Dudley, R. L., Fyfe, C. A., Stephenson, P. J., Deslandes, Y., Hamer, G. K. and Marchessault, R. H. (1983) High-resolution 13 C CPaMASaNMR spectra of solid cellulose oligomers and the structure of cellulose II. J. Am. Chem. Soc. 105, 24692472. Earl, W. L. and Vanderhart, D. L. (1981) Observations by high-resolution C-13 NMR of cellulose-I related to morphology and crystal-structure. Macromolecules 14, 570574. Emons, A. M. C. (1994) Winding threads around plant cells: a geometric model for microbril deposition. Plant Cell and Environment 17, 314. Emons, A. M. C. and Kieft, H. (1994) Winding threads around plant cells: Applications of the geometrical model for microbril deposition. Protoplasma 180, 5969. Erata, T., Shikano, T., Takai, M. and Hayashi, J. (1995) NMR-studies on the structure of cellulose two dimensional solid-state NMR approach. Macromolecular Symposia 99, 2529. Fengel, D. (1971) Ideas on ultrastructural organisation of cell-wall components. J. Polymer Science: Part C 36, 383392. Fengel, D. (1992) Characteristics of cellulose by deconvoluting the OH valency range in FTIR spectra. Holzforschung 46, 283288. Fengel, D. and Stoll, M. (1989) Crystals of cellulose grown from TFA solution. Wood Sci. Technol. 23, 8594. Fengel, D. and Wegner, G. (1989) In Wood: Chemistry, Ultrastructure, Reactions. Berlin, New York: de Gruyter, p. 66. Fengel, D., Jacob, H. and Strobel, C. (1995) Inuence of the alkali concentration on the formation of cellulose II Study by X-ray diffraction and FTIR spectroscopy. ibid. 49, 505511. Fink, H.-P., Philipp, B., Paul, D., Serimaa, R. and Paakkari, T. (1987) The structure of amorphous cellulose as revealed by wide-angle X-ray scattering. Polymer 28, 12651270. Fink, H.-P., Hofmann, D. and Purz, H. J. (1990) Zur Fibrillurstruktur nativer Cellulose. Acta Polymerica 41, 131137. Fink, H.-P., Hofmann, D. and Philipp, B. (1995) Some Aspects of lateral chain order in cellulosics from X-ray scattering. Cellulose 2, 5170. Fischer, D. G. and Mann, J. (1960) Crystalline modications of cellulose. Part VI. Unit cell and molecular symmetry of cellulose I. J. Polymer Science 62, 189194. French, A. D. (1978) The crystal structure of native ramie cellulose. Carbohydrate Research 61, 6780. French, A. D. and Murphy, V C. (1977) A virtual bond modelling study of cellulose I. In Cellulose . chemistry and technology ACS Symposium Series 48 (J. C. Arthur, Jr., ed.). Washington DC: American Chemical Society, pp. 1229. French, A. D. and Howley, P. S. (1989) Comparisons of structures proposed for cellulose. In Cellulose and Wood: Chemistry and Technology. Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 159167. French, A. D., Miller, D. P. and Aabloo, A. (1993) Miniature crystal models of cellulose polymorphs and other carbohydrates. Int. J. Biol. Macromolecules 15, 3036. Frey-Wyssling, A. (1953) Submicroscopic structure of the elementary bers of cellulose. Experintia 9, 181183. Frey-Wyssling, A. (1954) The ne structure of cellulose microbrils. Science 119, 8082. Frey-Wyssling, A. and Muhlethaler, K. (1951) The ne structure of cellulose. Fortschritte der Chemie Organischer Naturstoffe 8, 127. Frommer, J. (1992) Scanning tunnelling microscopy and atomic force microscopy in organic chemistry. Angewandte Chemie International Edition English 31, 12981328. Fujino, T. and Itoh, T. (1994) Architecture of the cell-wall of a green-alga, Oocystis-apiculata. Protoplasma 180, 3948. Gardiner, E. S. and Sarko, A. (1985) Packing analysis of carbohydrates and polysaccharides. 16. The crystal structures of celluloses IVI and IVII. Can. J. Chemistry 63, 173180. Gardner, K. H. and Blackwell, J. (1971) The substructure of crystalline cellulose and chitin

202

O'SULLIVAN

microbrils. J. Polymer Sci.: Part C 36, 327340. Gardner, K. H. and Blackwell, J. (1974a) The structure of native cellulose, Biopolymers 13, 19752001. Gardner, K. H. and Blackwell, J. (1974b) Hydrogen bonding in native cellulose. Biochimica et Biophysica Acta 343, 232237. Gessler, K., Krauss, N., Steiner, T., Betzel, C., Sarko, A. and Saenger, W. (1995) -D-Cellotetraose He, ihydrate as a structural model for cellulose II. An X-ray diffraction study. J. Am. Chem. Soc. 117, 1139711406. Hanley, S. J., Giasson, J., Revol, J.-F. and Gray, D. G. (1992) Atomic force microscopy of cellulose microbrils comparison with transmission electron-microscopy. Polymer 33, 46394642. Haworth, W. N. (1925) A revision of the structural formula of glucose. Nature 116, 430. Hayashi, T. and Maclachlan, G. (1984) Pea xyloglucan and cellulose. 2. Macromolecular Organisation. Cell Physiology 75, 596604. Hayashi, J., Sufoka, A., Ohkita, J. and Watanabe, S. (1975) The conformation of existence of cellulose IIII , IIIII , IVI and IVII by X-ray method. J. Polymer Sci.: Polymer Letters Edition 13, 2327. Hayashi, J., Yamada, T. and Shimizu, Y.-L. (1989) Memory phenomenon of the original crystal structure in allomorphs of Na-cellulose. In Cellulose and Wood: Chemistry and Technology, Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 77102. Heiner, A. P., Sugiyama, J. and Teleman, O. (1995) Crystalline cellulose I and I studied by molecular dynamics simulation. Carbohydrate Res. 273, 207223. Henrissat, B. and Chanzy, H. (1985) Unidirectional degradation of Valonia cellulose microcrystals subjected to cellulose action. Febs Letters 184, 285288. Henrissat, B., Perez, S., Tvaroska, I. and Winter, W. (1987) Multidisciplinary approaches to the structure of model compounds for cellulose II. In The Structure of Cellulose. Washington DC: American Chemical Society, pp. 3866. Herbert, J. J. and Muller, L. L. (1974) An electron diffraction study of the crystal structure of native cells. J. Appl. Polymer Sci. 18, 33733377. Hermans, P. H. (1949) Physics and Chemistry of Cellulose Fibres. New York: Elsevier, pp. 1320. Hermans, P. H. and Weidinger, A. (1949) X-ray studies on the crystallinity of cellulose. J. Polymer Sci. 4, 135144. Hess, K. and Kissig, H. (1941) Zur Kenntnis der Hochtemperatur-Modikation der Cellulose (Cellulose IV). Zietschrift Physikalische Chemie B 49, 235239. Heyn, A. N. J. (1966) The microcrystalline structure of cellulose in cell walls of cotton, ramie and jute bres as revealed by negative staining of sections. J. Cell Biol. 29, 181197. Honjo, G. and Watanabe, M. (1958) Examination of cellulose bre by the low-temperature specimen method of electron diffraction and electron microscopy. Nature 181, 326328. Hooft, R. W. W., van Eijck, B. P. and Kroon, J. (1992) Use of molecular dynamics methods in conformational analysis. Glycol A model story. J. Chemical Physics 97, 36393641. Horii, F., Hirai, A. and Kitamaru, R. (1982) Solid-state high-resolution 13 C NMR studies of regenerated cellulose samples with different crystallinities. Polymer Bulletin 8, 163. Horii, F., Hirai, A. and Kitamaru, R. (1987a) CPaMAS C-13 NMR approach to the structural-analysis of cellulose. Polymers for Fibers and Elastomers 260, 27, ACS Symposium Series, American Chemical Society. Horii, F., Hirai, A. and Kitamaru, R. (1987b) CPaMAS C-13 NMR spectra of the crystalline components of native cellulose. Macromolecules 20, 21172120. Ioelovitch, M. (1992) On the supermolecular structure of native and isolated cellulose samples. Acta Polymerica 43, 110113. Jacob, H. F., Fratzl, P. and Tschegg, S. E. (1994) Size and arrangement of elementary cellulose brils in wood cells a small angle X-ray scattering study of Picea-Abies. J. Struct. Biol. 113, 1322. Khalatur, P. G., Marchenko, G. N., Pletneve, S. G. and Khrapkovskii, G. M. (1986) Computer

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

203

simulation of cellulose by the methods of molecular and brownian dynamics. Doklady Akademii Nauk. SSSR 291, 157162. Kimur, S. and Itoh, T. (1995) Evidence for the role of the glomerulocyte in cellulose synthesis in the tunicate, Metandrocarpa uedai. Protoplasma 186, 2433. Koehler, J., Saenger, W. and van Gunsteren, W. F. (1987) A molecular dynamics simulation of crystalline -cyclodextrin hexahydrate. Eur. Biophys. J. 15, 197210. Kolpak, F. J. and Blackwell, J. (1976) Determination of the structure of cellulose II. Macromolecules 9, 273278. Kolpak, F. J., Weih, M. and Blackwell, J. (1978) Mercerization of cellulose: 1. Determination of the structure of mercerized cotton. Polymer 19, 123131. Kooijman, J., van Eijck, B. P. and Kroon, J. (1992) Molecular dynamics simulations of crystal structures containing charged molecules. J. Mol. Struct. 268, 283292. Kratky, O. and Mark, H. (1938) Supermolecular structure of bers. Papier-Fabr. 36, 345348. Kroon-Batenburg, L. M. J. and Kroon, J. (1990) Solvent effect on the conformation of the hydroxymethyl groups established by molecular dynamics simulations of methyl--D-glucoside in water. Biopolymers 29, 12431248. Kroon-Batenburg, L. M. J. and Kroon, J. (1992) The behaviour of model compounds of cellulose as `viewed' by MD simulation. Carbohydrates in the Netherlands 7, 610. Kroon-Batenburg, L. M. J., Bouma, B. and Kroon, J. (1996) Stability of cellulose structures studied by MD simulations. Could mercerized cellulose be parallel? Macromolecules 29, 56955699. Kuga, S. and Brown, R. M. Jr. (1987) Lattice imaging of ramie cellulose. Polymer Communications 28, 311314. Kuga, S., Takagi, S. and Brown, R. M. Jr. (1993) Native folded-chain cellulose II, Polymer 34, 3293 3297. Kulshreshta, A. K. and Dweltz, N. E. (1973) Paracrystalline lattice disorder in cellulose I. Reappraisal of application of 2-phase hypothesis to analysis of powder X-ray diffractograms of native and hydrolysed cellulosic materials. J. Polymer Sci.: Polymer Physics Edition 11, 487497. Kuutti, L., Peltonen, J., Pene, J. and Teleman, O. (1995) Identication and surface-structure of crystalline cellulose studied by atomic force microscope. Journal of Microscopy Oxford 178, 16. Lamport, D. T. A. (1970) Cell wall metabolism. Ann. Rev. Plant Physiol. 21, 235270. Lee, J. H., Brown, R. M. Jr., Kuga, S., Shoda, S.-I. and Kobayashi, S. (1994) Assembly of synthetic cellulose I. Proc. Natl Acad. Sci. USA 91, 74257429. Leeang, B. R., Vliegerthart, J. F. G., Kroon-Batenburg, L. M. J., van Eijck, B. P. and Kroon, J. (1992) A 1 H-NMR and MD study of intramolecular hydrogen bonds in methyl--cellobioside. Carbohydrate Research 230, 4161. Lenz, J. and Schurz, I. (1990) Fibrillar structure and deformation behaviour of regenerated cellulose bres. Part II: Elementary brils and deformation. Cellulose Chem. Technol. 24, 679692. Liang, C. Y. and Marchessault, R. H. (1959) Infrared spectra of Crystalline Polysaccharides. 1. Hydrogen bonds in Native Cellulose. J. Polymer Sci. 37, 385395. Lindenmeyer, P. H. (1965) Crystallisation and molecular folding (of Polymers). Science 147, 12561262. Macchi, E. and Palma, A. (1969) Morphological studies on precipitated cellulose. Die Makromolekulare Chemie 123, 286288. Mann, J. and Marrinan, H. J. (1958) Crystalline Modications of Cellulose. Part II. A study with Plane-Polarised infrared radiation. J. Polymer Sci. 32, 357370. Marchessault, R. H. and Liang, C. Y. (1960) Infrared Spectra of Crystalline Polysaccharides. III. Mercerized Cellulose. ibid. 43, 7184. Marchessault, R. H. and Liang, C. Y. (1962) The infrared spectra of crystalline polysaccharides. VIII. Xylans. ibid. 59, 357376. Marchessault, R. H. and Sarko, A. (1967) X-ray structure of polysaccharides. In Advanced

204

O'SULLIVAN

Carbohydrate Chemistry 22, (M. L. Wolfrom, ed.). New York: Academic Press, pp. 421483. Marchessault, R. H. and Sundararajan, P. R. (1983) In Cellulose, in the Polysaccharides. New York: Academic Press, p. 11. Mark, R. E. (1967) In Cell wall mechanics of tracheids. New Haven: Yale University Press. Mark, R. E. (1971) Mechanical behaviour of cellulose in relation to cell wall theories. J. Polymer Sci.: Part C 36, 393406. Mark, R. E., Koloni, P. N., Tung, R.-C. and Gillis, P. P. (1969) Cellulose: Refutation of a folded chain structure. Science 164, 7273. Marrinan, M. and Mann, J. (1956) Infrared spectra of the crystalline modications of cellulose. J. Polymer Sci. 21, 301311. Matsuda, Y., Kowsaka, K., Okajima, K. and Kamide, K. (1992) Structural change of cellulose contained in immature cotton boll during its growth. Polymer International 27, 347351. Meyer, K. H. and Mark, H. (1928) The Structure of the Crystallised components of Cellulose. Berichte der Deutschen Chemica Gesellschaft 61B, 593614. Meyer, K. H. and Misch, L. (1937) 31. Positions des atomes dans le nouveau modele spatial de la cellulose. Helvetica Chimica Acta 20, 232245. Millane, R. P. and Narusiah, T. V. (1989) An X-ray ber diffraction study of ramie cellulose I. In Cellulose and Wood: Chemistry and Technology, Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 3951. Miller, D. P. and Li, A. (1989) Renement of the crystal-structure of ramie cellulose I with additional meridional X-ray intensities. In Cellulose and Wood-Chemistry and Technology, Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 139157. Morosoff, N. (1974) Never dried cotton bres. Crystallinity and crystallite size. Journal of Applied Polymer Science 18, 18371854. Muggli, R., Elias, H.-G. and Muhlethaler, K. (1969) Zum feinbau der Elementarbrillen der Cellulose. Die Makromolekulare Chemie 121, 290294. Naslund, P., Vuong, R., Chanzy, H. and Jesior, J. C. (1988) Diffraction contrast transmission electron microscopy on ax ber ultrathin cross-sections. Textile Research Journal 58, 414417. Nickerson, R. F. (1941) Hydrolysis and catalytic oxidation of cellulosic materials. Hydrolysis of natural regenerated and substituted celluloses. Ind. Engin. Chem. 33, 10221027. Nieduszynshi, L. and Atkins, E. D. T. (1970) Preliminary investigation of algal cellulose. 1. X-ray intensity data. Biochimica et Biophysica Acta 222, 109118. Nishimura, H., Okano, T. and Sarko, A. (1991a) Mercerization of cellulose. 5. Crystal and molecular structure of Na-cellulose I. Macromolecules 24, 759770. Nishimura, H., Okano, T. and Sarko, A. (1991b) Mercerization of cellulose. 6. Crystal and molecular structure Na-cellulose IV, ibid. 771778. Nyburg, S. C. (1961) Fibrous macromolecular substances. In X-ray Analysis of Organic Structures (L. F. Fieser and M. Fieser, eds). New York: Academic Press, pp. 302314. Okamura, K. (1991) Structure of cellulose. In Wood and Cellulosic Chemistry (D. N.-S. Hon and N. Shiraishi, eds). New York: Marcel Dekker, pp. 89111. Okano, T. and Sarko, A. (1984) Mercerization of cellulose. I. X-ray diffraction evidence for intermediate structures. J. Appl. Polymer Sci. 29, 41754182. Okano, T. and Sarko, A. (1985) Mercerization of cellulose. II. Alkali-cellulose intermediates and a possible mercerization mechanism. ibid. 30, 325332. Okano, T., Koyanagi, A., Kondo, Y. and Sarko, A. (1989) Structural variation of native cellulose related to its source. In Cellulose and Wood: Chemistry and Technology, Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 5365. Okuda, K., Tsekos, L. and Brown, R. M. Jr. (1994) Cellulose microbril assembly in Erythrocladia subintegra Rosenv.: an ideal system for understanding the relationship between synthesising complexes (TCs) and microbril crystallization. Protoplasma 180, 4958.

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

205

O'Sullivan, A. C. (1995) Modelling of cellulose molecule interactions. Ph.D Thesis. University of Wales, Bangor, Gwynedd, UK. Paakkari, T., Serimaa, R. and Fink, H.-P. (1989) The structure of amorphous cellulose. Acta Polymerica 40, 731734. Pertsin, A. J., Nugmanov, O. K. and Marchenko, G. N. (1986) Crystal structure of cellulose polymorphs by potential energy calculations: 2. Regenerated and native celluloses. Polymer 27, 597601. Petitpas, T. and Mering, J. (1956) Molecular structure of cellulose II. Compt. Rend. 243, 4750. Pizzi, A. and Eaton, N. (1984) The structure of cellulose by conformational analysis 1. Cellobiose and Methyl--cellobioside. Journal of Macromolecular Science-Chemistry A21, 14431446. Pizzi, A. and Eaton, N. (1985a) The structure of cellulose by conformational analysis. 2. The cellulose polymer chain. ibid. A22, 105138. Pizzi, A. and Eaton, N. (1985b) The structure of cellulose by conformational analysis. 3. Crystalline and amorphous structure of cellulose I. ibid. A22, 139160. Pizzi, A. and Eaton, N. (1987) The structure of cellulose by conformational analysis. Part 4. Crystalline cellulose II, ibid. A24, 901918. Poppleton, B. J. and Mathieson, A. McL. (1968) Crystal structure of -D-cellotetraose and its relationship to cellulose. Nature 219, 10461049. Powell, M. J. D. (1964) An efcient method for nding the minimum of a function of several variables without calculating derivatives. Computer Journal 7, 155162. Preston, R. D. (1975) X-ray analysis and the structure of the components of plant cell walls. Physics Reports 21, 183226. Preston, R. D. (1986) Natural celluloses. In Cellulose: Structure, Modication and Hydrolysis (R. A. Young and R. M. Rowell, eds). New York: John Wiley and Sons, pp. 327. Preston, R. D. and Cronshaw, J. (1958) Constitution of the brillar and non-brillar components of the walls Valonia ventricosa. Nature 181, 248250. Purves, C. B. (1954) In Chain structure in cellulose and cellulose derivatives: Part 1 (Ott and Spurlin, eds). New York: Wiley-Interscience, pp. 54. Purz, H. J., Graf, H. and Fink, H.-P. (1995) Electron-microscopic investigations of brillar and coagulation structures of cellulose. Papier 49, 714. Rappe, A. K., Casewitt, C. J., Colwell, K. S., Goddard, W. A. and Skiff, W. M. (1991) Charge equilibrium for molecular dynamics simulations. J. Phys. Chem., US 95, 33583363. Rees, D. A. (1970) Conformational analysis of polysaccharides. Part V. The Characterization of linkage conformations (chain conformations) by optical rotation at a single wavelength. Evidence for distortion of cyclohexa-amylose in aqueous solution. Optical rotation and the amylose conformation. J. Chem. Soc. B, 877884. Rees, D. A. and Scott, W. E. (1971) Polysaccharide conformation. Part VI. Computer model-building for linear and branched pyranoglycans. Correlations with biological function. Preliminary Assessment of inter-residue forces in aqueous solution. Further interpretation of optical rotation in terms of chain conformation. ibid. 469479. Rees, D. A. and Skerrett, R. J. (1968) Conformational analysis of cellobiose, cellulose and xylan. Carbohydrate Research 7, 334348. Rees, D. A. and Skerrett, R. J. (1970) Conformational analysis of polysaccharides. Part IV. Long-range contacts in some -glucans by model building in the computer and the inuence of oligosaccharide conformation or optical rotation. J. Chem. Soc. B, 189193. Reis, D., Vian, B., Chanzy, H. and Roland, J.-C. (1991) Liquid crystal-type assembly of native cellulose-glucuronoxylans extracted from plant cell wall. Biology of the Cell 73, 173178. Reisling, S. and Brickmann, J. (1995) Theoretical investigations on the structure and physical properties of cellulose. Macromolecular theory and simulations 4, 725743. Revol, J. F. and Goring, D. A. I. (1983) Directionality of the ber c-axis of cellulose crystallites in microbrils of Valonia ventricosa. Polymer 24, 15471550.

206

O'SULLIVAN

Roche, E. and Chanzy, H. (1981) Electron microscopy study of the transformation of cellulose I into cellulose IIII in Valonia. Int. J. Biol. Macromolecules 3, 201206. Rowland, S. O. and Roberts, E. J. (1972) The nature of accessible surfaces in the microstructure of native cellulose. J. Polymer Sci. Part A1 10, 24472461. Sarko, A. (1973) Conformation of native and regenerated cellulose. Presented at the Fourth Canadian Wood Chemistry Symposium. July 1993. Quebec, Canada. Sarko, A. (1978) What is the crystalline structure of cellulose. Tappi 61, 5961. Sarko, A. (1987) Cellulose How much do we know about its structure. In Wood and Cellulosics: Industrial utilization, biotechnology, structure and properties (J. F. Kennedy, ed.). Chichester, UK: Ellis Horwood, pp. 5570. Sarko, A. and Marchessault, R. H. (1969) Supermolecular structures of polysaccharides. J. Polymer Sci.: Part C 28, 317331. Sarko, A. and Muggli, R. (1974) Packing analysis of carbohydrates and polysaccharides. III. Valonia cellulose and cellulose II. Macromolecules 7, 486494. Sarko, A., Southwick, J. and Hayashi, J. (1976) Packing analysis of carbohydrates and polysaccharides 7. Crystal structure of cellulose IIII and its relationship to other cellulose polymorphs. ibid. 9, 857863. Sasaki, K. and Taylor, I. E. P. (1984) Specic labelling of cell-wall polysaccharides with myo-[2-H-3]inositol during germination and growth of phaseolus-vulgaris L. Cell Physiology 25, 989997. Scott, R. A. and Scheraga, H. A. (1965) Method for calculating internal rotation barriers. Journal of Chemical Physics 42, 22092215. Shefter, E. and Trublood, K. N. (1965) The crystal and molecular structure of D ()-barium uridine51 -phosphate. Acta Crystallographica 18, 10671077. Simon, I., Scheraga, H. A. and Manley, R. St. J. (1988a) Structure of cellulose. 1. Low-energy conformations of single chains. Macromolecules 21, 983990. Simon, I., Glasser, L., Scheraga, H. A. and Manley, R. St. J. (1988b) Structure of cellulose. 2. Lowenergy crystalline arrangements. ibid. 21, 990998. Sjostrom, E. (1981) In Wood Chemistry Fundamentals and Applications. New York: Academic Press, pp. 49. Sponsler, O. L. and Dore, W. H. (1926) The structure of ramie cellulose as derived from X-ray data. Fourth Colloid Symposium Monograph 41, 174202. Sprague, J. T., Tai, J. C., Yuh, Y. and Allinger, N. L. (1987) The MMP2 calculation method. Journal of Computational Chemistry 8, 581603. Stipanovic, A. J. and Sarko, A. (1976) Packing analysis of carbohydrates and polysaccharides. 6. Molecular and crystal structure of regenerated cellulose II. Macromolecules 9, 851857. Sugiyama, J. (1992) Crystal forms of native cellulose. Mokuzai Gakkaishi 38, 723731. Sugiyama, J. and Okano, T. (1989) Electron microscopic and X-ray diffraction study of cellulose IIII and cellulose I. In Cellulose and Wood: Chemistry and Technology, Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 119127. Sugiyama, J., Persson, J. and Chanzy, H. (1991a) Combined IR and electron diffraction study of the polymorphism of native cellulose. Macromolecules 24, 24612466. Sugiyama, J., Vuong, R. and Chanzy, H. (1991b) Electron diffraction study on the two crystalline phases occurring in native cellulose from an algal cell wall. ibid. 24, 41684175. Sugiyama, J., Chanzy, H. and Revol, J.-F. (1994) On the polarity of cellulose in the cell wall of Valonia. Planta 193, 260265. Sundaralingam, M. (1968) Some aspects of stereochemistry and hydrogen bonding of carbohydrates related to polysaccharide conformations. Biopolymers 6, 189213. Takahashi, Y. and Matsunaga, H. (1991) Crystal structure of native cellulose. Macromolecules 24, 39683969. Tasker, S., Badyal, J. P. S., Backson, S. C. E. and Richards, R. W. (1994) Hydroxyl accessibility in celluloses. Polymer 35, 47174721.

CELLULOSE: THE STRUCTURE SLOWLY UNRAVELS

207

Taylor, I. E. P. and Wallace, J. C. (1989) The structural association between cellulose and xyloglucan in the primary cell wall of beans. In Cellulose and Wood: Chemistry and Technology, Proceedings of the Tenth Cellulose Conference (C. Schuerch, ed.). New York: John Wiley and Sons, pp. 273282. Tsekos, I., Reiss, H. D. and Schnepf, E. (1993) Cell-wall structure and supramolecular organization of the plasma membrane of marine red algae visualized by freeze-fracture. Acta Botanica Neerlandica 42, 119132. Tsuboi, M. (1957) Infrared Spectrum and Crystal Structure of Cellulose. J. Polymer Sci. 25, 159171. VanderHart, D. L. and Atalla, R. H. (1984) Studies of microstructure in native celluloses using solid state C-13 NMR. Macromolecules 17, 14651472. van Gunsteren. W. F. GROMOS, Groningen Molecular Simulation Computer Program Package, University of Groningen, Groningen, The Netherlands. van Gunsteren. W. F. and Berendsen, H. J. C. (1977) Algorithms for molecular dynamics and constraint dynamics. Molecular Physics 34, 13111327. Verlhac, C., Dedier, J. and Chanzy, H. (1990) Availability of surface hydroxyl groups in Valonia and bacterial cellulose. J. Polymer Sci.: Part A: Polymer Chemistry 28, 11711177. Von Nageli, C. (1858) Die Starkekoner, Panzenphysiologische Untersuchungen, Die Starkekekoner 2. Walton, A. G. and Blackwell, J. (1973) In Biopolymers Vol 22, New York: Academic Press, p. 468. Wellard, H. J. (1954) Variation in the lattice spacing of cellulose. J. Polymer Sci. 13, 471476. Wolters-Arts, A. M. C., van Amstel, T. and Derksen, J. (1993) Tracing cellulose microbril orientation in inner primary cell walls. Protoplasma 175, 102111. Woodcock, C. and Sarko, A. (1980) Packing analysis of carbohydrates and polysaccharides. II Molecular and crystal structure of native ramie cellulose. Macromolecules 13, 11831187. Woodcock, S., Henrissat, B. and Sugiyama, J. (1995) Docking of Congo Red to the surface of crystalline cellulose using molecular mechanics. Biopolymers 36, 201210. Yamamoto, H. and Horii, F. (1993) CPaMAS 13 C NMR analysis of the Crystal Transformation Induced for Valonia Cellulose by Annealing at High Temperature. Macromolecules 26, 13131317. Yamamoto, H. and Horii, F. (1994) In situ Crystallization of Bacterial Cellulose I. Inuences of polymeric additives, stirring and temperature on the formation celluloses I and I as revealed by cross polarizationamagic angle spinning (CPaMAS) 13 C NMR spectroscopy. Cellulose 1, 5766. Yamamoto, H., Horii, F. and Odani, H. (1989) Structural changes of native cellulose crystals induced by annealing in aqueous alkaline and acidic solutions at high temperatures. Macromolecules 22, 41304132. Zeronian, S. H. and Ryu, H.-S. (1987) Properties of cotton bres containing the cellulose IV crystal structure. J. Appl. Polymer Sci. 33, 25872604.