James Bolton messed up his school education with some encouragement from his teachers. Forced to specialize at the age of 14 (as is the norm in the UK), he opted for the science stream. Two years of pretending to learn chemical equations persuaded him that it was easier to specialize in languages. Grades awarded minima cum laude at the age of 18 convinced the University of Oxford that he was only fit to enter a course in divinity. So he fled to North Wales. There the druids created for him a degree course in Wood Science (the first of its kind in the UK), and were forced to award it to him in 1971. He then went to sit at the feet of the wood physics guru, Dr Alan Petty of the University of Aberdeen. While so engaged, he dreamed up a climbing expedition to the Western Himalayas in Afghanistan. This ended in a major dispute with the Soviet Union: questions were raised in the Houses of Parliament at Westminster; Soviet troops invaded Afghanistan almost as he left the country.

Shortly afterwards he took up a teaching and research post at his first alma mater, finishing a PhD at his second in 1976. Over the next 13 years he specialized in the physics of wood-based panel manufacture and formaldehyde-based adhesives, supervising over 40 postgraduates in these areas.

In 1989, convinced that there were opportunities for a new generation of fibre processing technologies, and frustrated by the difficulties of supporting the necessary innovation through conventional funding routes, he was responsible for the creation of the BioComposites Centre; this is a self-funding contract research institute based in the University of Wales in Bangor. This he now directs.

Abstract
Similar technological problems can arise in the manufacture of paper, panel products, structural composites and bonded nonwovens. There is a common need in these industrial sectors for determination of the precise nature of the fibre/matrix (adhesive) bond; fibre to matrix bonding needs to be optimised, perhaps in some cases via chemical modification of the fibre; methods for the determination of the distribution of adhesives and additives are also required. Given progress in these areas there is substantial potential for improving existing products and for new business based on the replacement of man-made fibres in composites with plant fibres, on the development of active plant fibre surfaces (e.g., for ion-exchange) and on the utilisation of annually harvested crops and residues. In the last case, there is a clear need for fractionation and integrated processing.

Introduction
In this paper, I shall use the word "plant" in its botanical sense, to include trees, shrubs and annually harvested plants; the term "plant fibre based composite" will be used in the widest sense, that of materials including plant fibres. In this last I follow a long tradition of terminological misuse.

In engineering circles, of course, a composite is thought of as a material comprised of a matrix (or adhesive) phase reinforced with fibres. So disastrous is the effect of voids on the strength properties of such composites that there is an automatic presumption that there will be as few voids as possible in the material. Given the inefficiencies of fibre packing, the matrix content is thus inevitably 30 percent, and sometimes as high as 70 percent by volume. To prevent confusion of these materials with others discussed below, we have used (and may even have coined) the name "high matrix content composites" or more simply, "high matrix composites".

The plant or wood-based panel product is a very different material. It is an order of magnitude or more lower in cost, and is designed to satisfy large volume markets in which it is perceived that much lower material strength properties and durability are acceptable. Adhesive contents are seldom more than 3-12 percent by weight and, even then, constitute 25-40 percent of final product cost. In most cases there is presently no prospect of product prices allowing substantial increase in adhesive content. As a result, the materials contain many voids. It can be argued that the material properties obtained are as much a function of the voids as they are of the fibres.

 
The distribution of these void volumes between inter- and intra-particle voids, and its significance, will be returned to later.

One of the strategies to deal with this problem is to densify the product to eliminate voids. Unfortunately, this is a double-edged sword: it increases the magnitude of stresses locked up in the product. These are then released during exposure of the product to water, and result in strain reversal which is manifested as thickness swelling.

Because wood-based panel products are so different from high matrix composites in their structure and performance, we apply to them the different term "low matrix content composites", or "low matrix composites".

Paper products are largely comprised of fibres, but can also contain adhesives, wet strength additives, etc. While these "matrix" materials generally form a very low proportion of product weight, there are conceptual similarities in many of the technological issues faced by the papermaker and the manufacturer of low or high matrix composites.

Equally, there is common ground between all these manufacturing operations and the manufacture of nonwovens. Here, fibres or fibre bundles, rather longer than those used in paper or low matrix composite manufacture, are brought together by wet or dry forming to produce a mat. The strength of this can then be enhanced by systematic interweaving or entanglement, and/or by bonding with adhesives.

Most of the uses of these fibres in nonwovens, paper, and composites are structural to semi-structural (from structural components to packaging). However, much the same technologies can be used to make novel materials capitalising on fibre characteristics other than strength properties (e.g., surface chemistry, porosity, etc). Time constraints make it impossible to describe here the potential of plant fibre based ion- and ligand-exchange systems, for example.

The present paper examines key issues common to many of these technologies and materials based on wood and other plant materials.

Bonding
The polysaccharide and phenolic polymers forming plant cell walls are very reactive. This leads directly to the water reactivity which can be such a problem in product performance. However, it must also be seen as an asset, since it provides us with an easier task in obtaining a bond of desired strength between fibre and matrix. By comparison, most man-made fibres (e.g., glass and carbon fibre) have much more inert surfaces, which limit the options for fibre to matrix bonding. Primary and secondary valence bonds are considered separately.

Secondary valence bonds
Van der Waals forces and hydrogen bonds are of considerable importance in plant fibre composites. The literature refers only infrequently to the former, although it is widely recognised that they significantly influence the properties of all materials. It is noteworthy that the computerised simulation of the docking of urea formaldehyde adhesive dimers on cellulose suggests that the bond obtained between the two is primarily due to van der Waals forces (Johns 1989).

Because of the large number of hydroxyl groups on plant cell wall polymers, much attention is rightly focused on the role of these and other groups as hydrogen bond donors and acceptors. There is no doubt that the crystallinity of cellulose, the grain structure of starch, much of the behaviour of polysaccharides in processing, and the fibre to fibre bonding in paper are largely determined by hydrogen bonds.

Secondary valence force bonds are generally, and correctly, regarded as reversible. It is also known that while hydrogen bonds can, in principle, be broken and reformed, the ease with which this can be done varies considerably.

Thus while amorphous cellulose is solubilised in water and dilute alkali, crystalline cellulose is not. It requires more concentrated alkali to disrupt the regular hydrogen bonds in the crystallites. In this case, it is probably the regularity of the hydrogen bonding (or the density of bonds/unit of molecule "surface") which results in water resistance.

In 1975, Battista reported a material which he called cellulose Avory, by acid treating cellulose to yield microcrystals, which he formed into a suspension in water before drying to cast a solid. Without the application of external pressure, and without an adhesive, a density of 1500 kg/m³ was obtained, very close to the theoretical maximum of approximately 1560 kg/m³. The material had remarkable strength properties, which presumably resulted from the finely divided form of the cellulose, and the orientation of particle surfaces to optimise hydrogen bonding during the slow drying. These results make it clear that we should not underestimate the potential for developing bond strength through hydrogen bonding.

A further consideration is that the hydrogen bond should be regarded as having an activation energy: unless there is a critical amount of energy available, the bond will be neither formed nor broken (Caulfield, 1985). Different bonds will have different strengths. Thus Nissan (1959) reports a range, of hydrogen bond strengths from 4 to 40 kJ/mol in cellulose, arising from different spatial configurations of the atoms involved.

The corollary of this is that high energy bonds will not necessarily be broken as readily in the presence of water as low energy bonds. The existence of hornification, a process by which natural polymers exposed to high temperatures lose some of their water reactivity, is at least partly attributable to this. So also is the empirical observation that apparently irreversible strain in wood will reverse if the wood is placed in an environment at a higher temperature, or a higher water vapour pressure, than that at which the strain was first induced. These observations suggest that we might benefit from the formation of higher energy hydrogen bonds as well as from enhancement of the number of such bonds in future products.

Covalent bonding
In most composites made from plant fibres, where a primary valence force holds fibre and matrix together, the primary valence force is covalent in nature. (The obvious exception will be such primary valence bonds as are formed in the production of inorganic matrix composites.)

There has been considerable debate whether the adhesives used in the production of low matrix composites are actually capable of forming covalent bonds with plant fibre surfaces. The issue is an important one, because it is perceived that it is weakness in the fibre to matrix bond which is responsible for poor water resistance of many panel products. Whether or not this is true, covalent bonding will certainly give stronger fibre to matrix bonds. The main adhesive systems used in low matrix composites are urea formaldehydes (UFs), melamine formaldehyde (MFs), phenol formaldehydes (PFs) and isocyanates or blends/copolymers thereof. The evidence for/against covalent bonding is in many cases based on research with low molecular weight model substrate or adhesive compounds (e.g., cellobiose or an adhesive dimer). MF: plant fibre bonds have received very little attention.

UF: plant fibre bonds: Ward (1992) was unable to find spectroscopic evidence for the existence of significant covalent bonding between UF and wood polymers in experimental work on model systems. In the work referred to above, Johns (1989) suggested from computer-based molecular modelling that the main forces responsible for bonding were van der Waals forces.

PF: plant fibre bonds: because of the phenolic nature of lignin, there is good reason to believe that PF adhesives should form covalent bonds with plant cell walls. Early evidence of this came from work of Chow, Troughton, Steiner and others here in British Columbia.

Isocyanate: plant fibre bonds: the reactivity of the urethane group with polyols is the basis of the polyurethanes industry. There is every reason to expect that plant polymer hydroxyl groups should be very reactive with isocyanates, and there is abundant spectroscopic evidence to show that these reactions occur. Unfortunately, reaction occurs also with water, leading to the formation of a polyurea. It was recognised early that the residual water in panel furnishes could behave in this way (Wittmann, 1976). To this day it is not clear what proportion of the p-MDI in commercial isocyanate resins forms polyureas, where these polyureas are located, and how important this might be.

If a proportion of adhesive molecules form urethane bonds with the cell wall surface, are these covalent bonds numerous enough to improve the performance if the composite? If polyureas are formed close to the cell wall surface because of the presence of water in the wall, should these offer better bond strength than urea formaldehyde polymers? If residual water is present in the walls, are polyurea molecules formed within the cell wall, achieving some adhesive strength by virtue of entanglement?

While these issues are likely to be important in processes using isocyanate adhesives at present, it is also clear that, under ideal conditions, extensive covalent bonding between plant fibres and isocyanates will take place. In fact, high matrix composites with very high stiffness but great brittleness can readily be produced with isocyanates. It is believed that the brittleness is attributable to the presence of too many fibre to matrix bonds: if fibre to matrix adhesion is too good, the potential for crack-stopping at the matrix:fibre interface is reduced (BioComposites Centre, unpublished data.)

Molecular Entanglement
While there is general agreement that mechanical interlocking at a macroscopic or microscopic scale is not an important adhesion mechanism in plant fibre based composites, physical entanglement of cell wall and matrix polymers could contribute to adhesion. Bolton et al. (1975) found evidence for some penetration of UF adhesive into cell walls adjacent to the glueline. This suggests that an interphase of cell wall and matrix polymers can be created. It may be desirable to formulate adhesives with this in mind (e.g., by including a low molecular weight resin fraction, which can readily diffuse between cell wall polymers, subsequently polymerising in situ).

Adhesive Distribution
In anything other than high matrix composites (where by virtue of the high matrix content, it should be possible to ensure perfect coating of all fibres with matrix), achievement of a proper resin distribution can be of critical importance in determining product performance.

Unfortunately, we do not know with certainty what distribution is ideal. The uncertainty arises from the desirability of achieving some gap-filling to increase the bond area. If the amount of adhesive available is limited, is it better to ensure that all adherend contact points result in a small bond, or to depend on the formation of a stronger bond at some contact points, while others are not bonded at all? This dilemma can be argued at levels both within and between fibres: should the entire surface of each fibre be uniformly coated, or not? Might it be acceptable for some whole fibres to be well coated, and others not? (It is noteworthy that Gupta (1980) suggests that only 5 percent of paper fibres need to be treated to achieve sizing in paper production.)

These issues are almost as uncertain now as they were 30 years ago. This is partly because of the extraordinary difficulties of identifying and quantifying the distribution of the adhesives in low matrix composites. The difficulties include:

• the low concentration of adhesive used
• the similarity of the elemental compositions and functional groups of adhesive and the plant materials
• the similarity of the densities of adhesive and plant materials in optical and electron microscopy.

Without some kind of staining or analysis technique it is extremely difficult to be certain that deposits on cell surfaces seen under the light or electron microscope are adhesive, and not extractives, etc. While some microscopists may attempt to convince themselves and others that droplets (etc) of adhesive can be seen, a more uniform distribution only one or few monomolecular layers thick is most unlikely to be detected.

There have been many attempts to label adhesives by physical addition of stains, etc. which might be positively located in light or electron microscopy. However, there can never be certainty that they remain with the adhesive, particularly since the cell wall polymers are excellent chromatographic media. Some are used commercially for the separation of components of physical mixtures!

To avoid these criticisms, Smith and Côté (1971) labelled PF resins with bromophenol, and Bolton et al. (1975) labelled UF resins with thiourea. Both research groups then attempted to locate the resins under the electron microscope by carrying out elemental analyses (for bromine and sulphur respectively) using Energy Dispersive Analysis of X-rays (EDAX). In both cases the label was incorporated in the resin structure, so physical separation was avoided. While both groups showed that the labelled system cured similarly to an unlabelled one, there was a possibility that labelling caused different polymer structures to be formed, and so might have influenced the adhesion process.

More recently, attempts have been made to locate the nitrogen atoms present in UFs or isocyanates using EDAX or microscope Fourier Transform Infra Red Spectroscopy (FTIR). Because of the low atomic number of nitrogen, and the low energy of the characteristic X-rays emitted from it, the former technique requires operation in "windowless" mode. This rapidly leads to contamination of the X-ray detector on the electron microscope, and is thus both unpopular with co-workers and expensive.

The microscope FTIR approach offers the ability to detennine the surface chemistry of a spot only 10 mm in diameter. In reality, the limiting diameter is probably rather larger (say 20 mm). However, this is still quite small enough to make a number of measurements on a single fibre or particle. The evidence emerging both from this technique, and from windowless EDAX, suggests a significant proportion of fibres may have no resin on them at all in a conventional MDF process, and that this proportion can be altered by manipulation of the process.

Resin Types
The past few decades have seen great change in matrix systems for both high and low matrix content composites. With high matrix content systems, matrix chemistry has been optimised for man-made fibre reinforced composites. If plant fibres start to penetrate these markets significantly, reoptimisation may be necessary.

With low matrix systems, changes in resin chemistry and adhesive formulation have been driven by the needs for bonding with plant fibre. With isocyanates we have seen the introduction of water-miscible blocked polymers, and the development of self-releasing systems. With phenolics, there has been much interest in adhesives tolerant of high furnish moisture contents. With all adhesive systems there has been continuous pressure to decrease pressing time.

The case of UF resins is particularly interesting, because it seems that new technology is required to extricate UF producers from an impasse created for them by regulatory pressure.

The relatively old cold setting UF joinery resins have F:U ratios as high as 1.3:1, and cross link well. Although they are deemed quite water resistant, they are plasticised by water. The same is true of the phenolic resins used to make water resistant boards (Bolton and Irle, 1987). It seems that the softening of the resin by the water removes its brittleness and allows it to accommodate the movement of the adjacent swelling plant fibre.

With the introduction of more and more stringent formaldehyde emission regulations (and the ridiculous prospect of European Regulations stipulating formaldehyde emission levels lower than those in solid wood) F:U ratios are approaching unity. Simultaneously, it has become difficult to obtain the dry strength required in panels, and very difficult to maintain thickness swelling within acceptable limits. Little wonder: the polymers must have very few cross links, and this must reduce dry strength. Also, being more linear, there must be an increasing tendency for the polymers to form pseudo-crystal line structures which are unreactive to water. This would explain the fall in wet strength.

It seems inevitable that we must expect a new generation of polymers including urea and formaldehyde, but involving other components making significant contributions to the final resin structure also. These new components might be

• new monomers allowing branching (Eberwele et al., 1991 a,b; Schorning et al., 1972)
• cross linking agents other than formaldehyde
• side groups added to the UF chain to disrupt crystallinity.

The only potential new resins or resin components on the horizon have been there for many years. Plant phenolics (including tannins), polysaccharides (including furan derivatives, and starches) and proteins all have both potential and problems. Significant technical advances have been made, and it is not unreasonable to predict new products or additives competing with, or contributing to, conventional panel adhesives in the near future. The main technical issues are

• water resistance; a problem with early starch and protein adhesives. Huge advances in starch derivatisation would appear to make this tractable
• speed of curing: significant advances are rumoured with tannin based adhesives
• consistency of supply: a problem with natural phenolics, in particular.

Considerable advances have been made through the cloning of the source plant in South America, and through fractionation of the feedstock in the Antipodes.

Finally, the work being done by Novo Nordisk and associates in Denmark on the use of surface activation by peroxidases to give adhesive-free bonding in beech MDF offers great promise. If it can be commercialised, it could have far reaching implications.

Plant Fibre in High Matrix Composites
The use of plant fibres in high matrix composites is not new. For example, for decades woven cotton/epoxy composites have been used in electrical insulation (Tuftiol Ltd), jute/phenolic composites have been used in friction and wear parts (Tenmat Ltd), milled wood/thermoplastics have been used to make sheet moulding compounds (Lignotoc, Woodstock). Recent advances include the launch of polypropylene compounded with waste paper for injection moulding in Japan by Mitsubishi, the development of jute/phenolic pultrusions in India, and a large number of North American companies producing plant fibre/ thermoplastic composites.

The obvious question to be asked is whether this resurgence of interest in plant fibres is justified. In Europe at least, it appears to be. The forces driving the replacement of glass fibre seem to be

• the ability to produce plant fibre reinforced composites with properties comparable with those obtained with glass fibre (Table 2). With further optimisation of the plant fibre to matrix bond, significant advances can be expected.
• the lower density of plant fibres (but, this depends to some extent on whether the fibre lumen is filled in the composite (see below)).
• environmental factors including the renewable origin of the plant fibre, its lower energy content (Table 3) and the ability to recover energy at the end of product life (glass fibre reinforced composites cannot easily be burnt).

In looking to the future, the main fibre-related technical issues are fibre length, cell wall thickness, and intra-fibre voids (cell lumina).

Fibre length: the manufacturer of conventional high matrix composites regards aspect ratio (the ratio of length to diameter) as a key factor in selection of manmade fibres. In considering plant fibre aspect ratios, we need to distinguish between individual fibres and fibre bundles. The latter are groups of fibres which can be extracted intact from certain non-wood plants (such as coir, sisal, hemp, flax).

The diameter of individual fibres is usually in the range 15-35 mm, while that of fibre bundles ranges between about 100 mm and 1000 mm. The length of fibre bundles is dictated largely by the part of the plant they are derived from. Thus, coir (originating from a fruit) has bundles only 0.15 to 0.28 m long; sisal (originating from a leaf) has bundles 0.6 to 1.0 m long; jute (originating from a stem) has bundles 1.5 to 3.6 m long. In terms of aspect ratios, most individual fibres have values in the range of 100 to 200. Notable exceptions are hemp (550), flax (1500), cotton (2000), and ramie (4000). The aspect ratio of fibre bundles varies from about 100 to in excess of 4000, depending to some extent on the damage inflicted on them during extraction from the plant. By careful selection of the fibre type, it should therefore be possible to find both short plant fibres for reinforcement of moulding compounds and injection mouldings, and longer fibres for use where anisotropy is desired in the product. Because the cell wall reactivity allows the creation of a desired density of strong covalent bonds between plant fibre and matrix (see above), it is possible that the aspect ratios deemed essential in reinforcement of plastics with synthetic fibres may not be required with plant fibres.

Leaving aside aspect ratio, two other factors favour the use of long plant fibres or fibre bundles from non wood sources in composites. First, textile industry processing technology allows the alignment of long fibres to a greater or lesser extent (e.g., through the use of carding). While some of this alignment may subsequently be lost in operations such as cross-lapping and needling in nonwovens production, and spinning and weaving in the formation of wovens, the ability to maintain some control over fibre orientation is important: it allows the composites engineer to take advantage of the anisotropy of strength and stiffness in the fibre. The same motivation lies behind the interesting use of magnetic fields to orientate short plant fibres (Zauscher and Humphrey, 1997). Second, and as important, the nonwovens or wovens that can be made from long plant fibre, have the characteristic of "drape": they fall easily down the side of deep draw moulding tools. This can be vital, because composite manufacture often becomes economically viable through the elimination of fabrication stages in the production of complex shapes. The better the drape, the more complex the shape can be.

Intra-fibre voids (cell lumina): for all practical purposes the density of all plant fibre walls is the same: about 1560 kg/m³. The density of different fibres varies widely: for softwood early wood tracheids, sometimes as low as 250kg/m³; for some non-wood fibres as high as 1200 kg/m³ or more. This variation is all attributable to the presence of the central void, or lumen, in the fibre.

In conventional high matrix composite manufacture, the elimination of voids is seen as a critical issue. Voids are seen as both initiating cracks, and as allowing their propagation, sometimes even allowing their growth beyond a critical length where failure becomes inevitable.

At first sight, the natural variation of fibre density appears to be an asset, allowing choice of a fibre density to suit the application. Some of the lumen void may be lost if the cell collapses at all during processing. (Indeed, this collapse may be deliberately engineered in papermaking, to enhance the contact and bonding area between adjacent cells.) But if, in high matrix composites, we must fill all inter- and intra-fibre voids, is the benefit of natural fibre density variation lost? Can we hope to fill all such voids anyway? Should we be trying to?

Assuming that the isolated plant fibre is intact, the accessibility of the lumen to liquid matrix is enormously variable depending particularly on the microstructure of the intercellular connections in the plant. The physical reasons for this are well understood at a conceptual level, and are to be found in the wood drying and preservation literature; however, the details of the microstructure in question are not documented for many fibre types. Of course, if the fibre has a broken end, penetration of matrix into it becomes much easier.

Assuming, again, that the fibre is intact, and the matrix well bonded to its exterior, is it necessary to fill the lumen? If the void is contained within the wall (itself a composite structure), is it damaging in terms of crack initiation and propagation? Possibly not. Maybe a void-free structure is really not so necessary in plant fibre high matrix composites.

Chemical Modification
Volumes have been written on the potential for improvement of water resistance and fibre to matrix bond strength in plant fibre reinforced composites through chemical modification. See Rowell (1983), Banks (1990) for reviews. Todate, there has been almost no commercialisation of the new technology (with the possible exception of isocyanate adhesives which may form some urethane bonds with cell walls; see above). Is this field of research going to be productive eventually?

In fairness, it has to be recognised that research on coupling and sizing agents for man-made fibres in composites has led to great advances but has taken decades. With plant fibres, there has been a tendency to dismiss sizing (i.e., coatings changing surface properties without covalent bonding) as being nondurable. However, sizing approaches have often been productive with manmade fibres. Perhaps, although with plant fibres we have the luxury of a reactive surface which will react readily to form covalent bonds, we do not always need to exploit this luxury, or to exploit it ftilly.

Leaving this aside, it is clear that in many new plant fibre high matrix composites, we are not exploiting the potential of the fibre properly. The fibre is typically two orders of magnitude (or more) stiffer than the matrix. If we do not see a substantial increase in material stiffness as a result of adding the fibre to the matrix, then the fibre is acting as a filler, not a reinforcement. Particularly in many new plant fibre/thermoplastic matrix composites, little reinforcement is being achieved. The potential for improving this situation through the use of chemically modified fibre is there. The challenge is to establish new processing facilities on a scale which will allow process costs to become acceptable. Unfortunately this requires large markets, which will only develop once process costs become low enough …… Before we become too dispirited by this classical chicken-and-egg problem, we should look sideways at the very successful industries based on the derivatisation of cellulose. Although the heterogeneous chemistry of modifying plant fibres is inevitably difficult, it will eventually be commercially viable.

The Challenge of Non-Wood Plant Resources
Plant fibres longer than 5 mm, and plant fibre bundles, are only available from non-woody plants. This is not the only reason why these resources require serious consideration.

C₄ plants are the most efficient biomass producers in the plant kingdom, fixing carbon in both light and dark. They are all non-woody plants. Rarely have they been bred for fibre production. In Europe, the potential of Miscanthus as an industrial feedstock is much researched. In Australia, there are now varieties of sugar cane bred for fibre production alone. Clearly, an annually harvested crop has huge cash-flow and strategic advantages if it is necessary to create a new resource and process it at short notice.

However, much of the interest in non-wood resources centres not on crops, but on crop residues. These are unwanted plant parts (e.g., cereal straw), or process residues (rape meal), of food (sugar beet pulp) or fibre crops (flax shives). Often they present a disposal cost and a potential substrate for crop pests unless properly disposed of. Sometimes they have a fuel value or are used in animal feeds. However, animal feed markets are shrinking in the Developed World.

Non-wood resources compared to wood
Because the manufacture of panel products, paper and cellulosics has largely been based on wood (and mainly on softwood which varies relatively little between species) there has been a tendency to assume that non-wood biomass is similar to wood, and should be processed in the same way. Comparison of the chemical composition of only four residues with that of spruce shows just how wrong this can be (Table 4).

On top of these differences in chemical composition are large differences in cell structure and dimensions. In fact, the non-woods are not just different to, but more heterogeneous than, wood. This heterogeneity can again be seen at both chemical and cellular levels.

Thus the outside of the wheat stem in straw is covered in a protein layer, and this in turn is covered by a waxy cuticle. These layers make the bonding of straw particles with UF resins difficult, although for some reason isocyanates are not similarly affected. The totatmass of protein and wax involved is small: < 2 percent by weight. To a large extent the problem disappears if the straw is disintegrated, and these minor components are redistributed throughout the resultant pulp.

At the cellular level wheat straw is again a good example. The material contains a fraction of cells similar to hardwood fibres in length and wall thickness, as well as a larger fraction (by volume, but not by weight) of shorter, thin walled parenchymatous cells. The latter contribute little to the strength of a paper sheet or low matrix composite, although they may possibly play a role in gap filling. In composites, their high surface area absorbs much resin, but even when well bonded to adjacent fibres can provide a weak link in the structure.

During processing, non-woods often behave very differently to wood chips. Much of this is attributable to density. Thus attempts to substitute a proportion of a wood chip feedstock in an existing process with a residue material can be frustrated by differences in the bulk density which can be handled in feed screws. Differences in density also cause differences in mattress rheology in the pressing of low matrix composites.

In summary, many non-wood feedstocks are unlikely to be processed efficiently in equipment optimised for wood chips. The heterogeneous character of nonwoods makes the introduction of fractionation into their processing desirable: sequential operations might, for example first remove waxes and polysaccharides which would otherwise interfere with bonding or contribute to pollution. At a later stage, different cell types might be separated, and then be further processed for different end uses, or for recombination in a single product. Such approaches are regularly used in the refining of crude oil; they now need to be applied to the industrial processing of non-wood biomass.

General Conclusions

1. There is still a need for a better understanding and optimisation of plant fibre to matrix bonds. Enhanced hydrogen bonding, molecular entanglement and covalent bonding are all potentially routes to achieve this. If covalent bonding is obtainable, the proportion of adherend hydroxyls which should be involved needs to be determined, especially in high matrix content composites.
2. In low matrix content composites, better means of quantifying the distribution of adhesives and additives are needed before we can quantify how uniform such a distribution should ideally be.
3. New components for UF adhesives are needed to provide cross linking without additional formaldehyde. Considerable potential exists for a revival of adhesives derived from natural products.
4. There are significant high value added market opportunities for plant fibre in high matrix content composites. The world market for high matrix composites was estimated at 10' tormes worth US$24 billion in 1990. The European glass fibre market alone is presently believed to be more than 0.5 million tonnes worth about US$1.2 billion, with growth figures of 12-14 percent/year being reported. In parallel with this, the European market for ion-exchange and water clean up systems is approximately US$120 million, and growing rapidly as environmental pressures mount. Chemically modified plant fibres could take a share of these markets also.
5. Realization of the real potential of non-wood biomass resources will require the development of new processes, and a fractionation approach.
6. To support an interdisciplinary approach to the utilisation of biomass, the conventional barriers found in industry, research institutes, and academia between such applied disciplines as wood processing, pulping, textile science, food science could profitably be removed.

References
Banks, W.B. 1990. Derivatisation of wood. In Kennedy, J.F., Phillips, G.O., and Williams, P.A. Cellulose Sources and Exploitation. Ellis Horwood, London. 520pp.

Batchelor, S.E., Knight, B.E.A., Wilkinson, A., Booth, E.J., Walker, K.C. 1996. Industrial markets for oilseed rape meal. Home Grown Cereals Authority Research Review OS11, 107pp.

Battista, O.A. 1975. Microcrystal polymer science. McGraw Hill, New York, 208pp.

Bolton, A.J. 1994. Natural fibres for plastic reinforcement. Materials Technology 9(1/2): 12-20.

Bolton, A.J., Dinwoodie, J.M., and Beale, P.M. 1985. The icrodistribution of UF resins in particleboard and its significance. Proc. Conf. Forest Products Research International: Achievements and the Future. Vol. 6. Paper 17-12. l9pp. Pretoria: CSIR.

Bolton, A.J., and Irle, M.A. 1987. Physical aspects of wood adhesive bond formation with formaldehyde based adhesives. Part 1. The effect of curing conditions on the physical properties of UF films. Holzforschung 41(3): 155-158.

Caulfield, D. 1985. A chemical kinetics approach to duration of load. Wood and Fiber Science 17: 504-521.

Eberwele, R.O., Myers, G.E., Rivers, B.H., Koutsky, J.A. 1991 (a & b). Polyamine modified urea-formaldehyde resins. Parts I & II. Journal of Applied Polymer Science 47: 2997-3012 and 48: 1483-1490.

Gupta, M.K. 1980. Chemically modified fibre as a novel sizing material. Tappi 63(3): 29-31.

Johns, W.E. 1989. Personal communication.

Nissan, A.H. 1959. Fundamentals of adhesion from molecular forces in cellulose. Tappi 42: 928-933.

Rowell, R.M. 1983. The Chemical Modification of Wood. Forest Products Abstracts 6: 363-82.

Schorning, P., Roffael, E., and Stegmann, G. 1972. Neuartige Holz-zu-Holz Bindungen in Holzwerkstoffen mit chemischen Mehrstoff-Systemen. I. Heisspressversuche in aminalkalischen Bindestoffsystemen. Holz als Roh- und Werkstoff 30: 253-258.

Siau, J.F. 197 1. Flow in wood. Syracuse University Press. 131 pp.

Smith, L.A., Côté, W.A. 1971. Studies of penetration of PF resin into wood cell walls with the SEM and energy dispersive X-ray analyser. Wood and Fiber 3: 56-67.

Ward, A. 1992. Interaction of formaldehyde based resins with wood. PhD Thesis, University of Wales. 194pp.

Wittmann, O. 1976. Die Holzverklebung mit Isocyanat. Holz als Roh- u. Werkstoff 34: 427-431.

Zauscher, S. and Humphrey, P.E. 1997. Orienting lignocellulosic fibers and particles by means of a magnetic field. Wood and Fiber Science 29(1) 35-46.