21 Concurrent modification of wood with phthalic anhydride in composite manufacture R Salisbury,* M Lawther* and P Brown** - *The BioComposites Centre, University of Wales, Bangor, Gwynedd LL57 2UW, Wales; **BP Chemicals Ltd, Hull Research Centre, Saltend, Hull HU12 8DS, UK
Abstract Phthalic anhydride (PA) is a well-known commodity chemical which is low cost and low hazard. PA can be used to improve the properties of wood particleboard made with phenolformaldehyde (PF) resin. The greatest improvement is in resistance to the effects of water, but initial properties also show some improvement. PA can be used to pre-treat the wood particles, but can also be added during board manufacture, eliminating the need for separate chemical processing. Boards made with unmodified wood and PF resin show improved properties at lower material cost when part of the PF resin is replaced by PA. The reactions of PA with PF, and with individual wood components, were investigated. It is shown that the improvement in board properties is not due to effects of PA upon the PF resin or the wood alone, but to result from an interaction between all three components.
Phthalic anhydride Phthalic anhydride is a major commodity chemical which finds applications as a difunctional intermediate with ultimate applications mainly in the plastics industry (e.g. polyester resins, paints and phthalate ester plasticisers). Current European production is of the order of 800,000 tonnes per annum. The material is commercially available in bulk as molten PA at 1600C (freezing point is or as a solid flake in pack sizes down to 25 kgs. Current price levels are approximately £500/tonne. 1310C)
PA is classified as an irritant, but causes no handling problems provided basic industrial practices are followed.
206 Physical and chemical processing of fibre and fibrous products
General considerations It has long been known that acid anhydrides react with wood, and many attempts have been made to improve the properties of composites using this reaction. PA has not been a popular choice of reagent, probably for two reasons. Firstly, it is not very reactive in the conditions usually applied, and secondly, all the results have shown that phthalylation increases the hygroscopicity of wood. On the first point, Rowell has suggested (1) that useful modification of wood must take place at 1200C or less, because the wood deteriorates above this temperature. However, in view of the fact that wood is subjected to temperatures of 20()OC during board manufacture, this seems unnecessarily restrictive. The only true measure of the usefulness of a reaction is the effect on the properties of the final composite. If excessively harsh reaction conditions are used, the composite will lose bending strength and stiffuess. Removing the 1200C restriction allows us to dispense with solvents and catalysts, and to react wood directly with powdered PA, which melts at 13IOC. In practice, we mixed ovendry softwood shred with powdered «400J.1) PA in a metal container which was closed but not sealed, and then heated it in an oven at 1600C for 1 hour. Non-leachable weight gains of more than 10% were easily obtained. The reaction of wood with an anhydride is usually represented as:
 In the case of a cyclic anhydride, the carboxyl group formed remains attached to the wood, as shown, which explains the greater hygroscopicity of wood so modified. That the above representation is correct is shown by the use of the carboxyl group for grafting on further polymers (2).
Phthalic anhydride in boards The BioComposites centre, started working with PA in 1991. Phthalylated wood made into a board with phenolic resin was found to have greatly increase water resistance.
In an attempt to show that wood modification was a key component of this effect, we also made boards in which the PA was simply added in various ways to the board furnish. Unfortunately for our theory, these boards also showed improved properties. A full study was carried out using a resin (J2005A) which had proved to be satisfactory in earlier work. This was not the best resin, but one which was easy to work with and readily available. We chose to work with wood shred rather than fibre or flakes, for ease of handling.
Modification with phthalic anhydride
Outline of trial At this time, we were assuming an analogy between PA and acetic anhydride, so the amount of chemical used was quite high: 10%, 14% and 18%. Two series of boards were made, one with wood that was modified with PA beforehand, the other with PA being added with the resin. Sufficient boards of each type were made to provide enough material for British Standard tests of : thickness swell, linear expansion and water uptake from manufacture to conditioning at 65% relative humidity at 20°C from conditioning to 1 hour water soak at 20 0C from conditioning to 42 hour water soak at 20°C from conditioning to conditioning at 93% relative humidity at 20 0 e from conditioning to 2 hours boiling after V313 cyclical testing during repeated boiling and drying at 105°C weight loss due to V313 cyclical boiling internal bond strength at 65% relative humidity at 20 0 e at 93% relative humidity at 200C after 1 hour soak at 20°C followed by reconditioning after 24 hours soak at 20°C followed by reconditioning after 2 hours boiling followed by drying and reconditioning after V313 after cyclical boiling three-point bending tests at 65% relative humidity at 20 0 e after 24 hours soak at 20 0 e weight loss due to biological challenge by Coniophora puteana Gleophyllum trabeum Pleurotus ostreatus
Results All the properties were improved by PA, whether it was reacted with the wood before board manufacture or added with the resin. The figures for the physical testing are not presented here, because they have been superceded by later work. The results of the biological testing are shown in Tables 1 and 2, because no further trials have been made.
208 Physical and chemical processing of fibre and fibrous products Table 1.
Weight loss due to fungal attack: wood modified by PA fungus\amount of PA
Weight loss % due to fungal attack: PA added with resin fungus\amount of PA
The physical testing results also showed that reaction during manufacture was nearly as good as prior modification, and also suggested that much lower levels of PA would also be effective. We therefore made boards using PA to replace some of the phenolic resin, and did a more restricted set of tests on them. The tests were not to British Standard, but were designed to give maximum information from the minimum of material. It was found that the best results were obtained by replacing about 20 - 25% of the resin (dry weight) by PA. The improvements obtained are summarised in the table. Table 3.
Relative improvements achieved in chipboard properties by replacing about 20% of PF resin by PA Property IBS - conditioned IBS - boiled IBS- retention thickness swell - cold soak thickness swell - 2 hour boil bending strength - conditioned bending stiffitess - conditioned bending toughness - conditioned bending strength - boiled bending stiffness - boiled bending toughness - boiled
Improvement factor 2.1 14 9 1.2* 1.8* 1.3 1.3 1.6 1.5 1.5 1.9
* ratio of log reciprocal
Modification with phthalic anhydride
These are not minor improvements, and any proposed mechanism must be able to account for the scale of these effects. It was found that all the properties were highly correlated with initial IBS, and this property only is used in the graphs below.
Possible mechanisms PA could be acting as an accelerator for the resin, producing more complete cures before the wood starts to deteriorate. A series of trials showed (Figure 1) that PA does act as an accelerator, but not a very powerful one, and the properties of boards made with PA go on increasing with pressing for longer:
Effect of press time
onlBS 0.6 .....- - - - - - - - - - - - - - -...... 0.5
~0.3 UJ ~
20% PA, boiled no PA, boiled
0.0 .....- - . . . - . -......- -......- - - . . - - -......- ---'1 o 10 20 30 40 50 60 press time
at 200°C (mins)
Figure 1 PA could be acting to improve the resin in some other way. We investigated this by making boards with vermiculite replacing wood and also by testing resin-impregnated strips of glass fibre filter paper. PA was detrimental to all the properties except dry strength. However, when cellulose filter paper or newsprint was used to make the strips, PA improved the strength and stiffness of the strip, both when dry and when soaked.
Effect of PA on properties of resin strips control = 100
210 Physical and chemical processing of fibre and fibrous products The effect of PA on resin cannot explain the greatly increased water resistance of boards, so what about the effect upon the wood? A set of boards was pressed at 20O<>C for ten minutes with or without 2% PA, but with no resin. As made, these were of the same thickness and density. After conditioning, the boards bad picked up about the same amount of moisture, but the control boards had swelled more. On soaking in cold water, the control boards immediately started to swell at a rate of about 100/0 per minute, and continued to do so for 7 Y2 minutes, when measurements were stopped. The board with PA initially swelled at a much' slower rate, about 1% in the first minute, but after 4 minutes it too was swelling at 10% per minute. The overall delay amounted to about 1 minute. This difference may be chemical, or may be due to the PA decreasing to porosity of the board. In either case, such a minor difference cannot explain the greatly improved resistance of boards made with PA to two hours of boiling. A further set of boards made without resin was cut into blocks immediately after pressing and cooled in a dessicator. They were then quickly measured and mounted for internal bond measurement. The board containing PA was no stronger than the board without it. PA is clearly not acting as a binder. The effect of PA cannot be explained by its action on either wood or resin, it must involve an interaction between all three components. Boards are sensitive to water content. Perhaps the PA "locks up" the water in some way and stops it interfering. A series of boards have been made and tested, and the results are shown in the Figure 2. It is not possible to translate the curve for the boards with PA onto the control curve by a horizontal movement only ~ so PA is not acting only to reduce the effects of moisture.
Phthalic anhydride addition.
Effect of mattress moisture on 18S. '1.6 .....- - - - - - - - - - - - - - - - - -..
~ 0.6 +-------~-----~=-------I 0.4 -+-0.2
0.0 ......- - - - -..... o 5
moisture content of mattress (%)
Modification with phthalic anhydride
Chemical considerations The resin is in aqueous solution and has a high pH. The board is pressed at high temperature. In these conditions the PA cannot survive long before being hydrolysed to acid, and some will be converted to phthalate ion. On the other hand, phthalic acid is thermolabile, and could start to dehydrate to PA. A phthalate ion from di-sodium phthalate, on the other hand, would have to grab two protons from the alkaline surroundings, and hold onto them long enough to be dehydrated. This seems unlikely. Dehydration is a mechanism that is not available to terephthalic acid (the para isomer) either.
Spectroscopic evidence The reactions of PA, phthalic acid and sodium phthalate with wood components were also investigated using FTIR. There is good evidence of reaction of PA and phthalic acid with all components at 1800C or less. Sodium phthalate does not appear to react even with lignin. FTIR spectra show that when PA reacts with wood at 1600C only di-ester is formed, no acid is detectable. Phthalic acid reacts just as thoroughly at 1800C. This shows that benzenecarboxylic acids can react directly with wood, without forming an anhydride. Futher studies confirmed that terephthalic acid also reacts with wood at 1600C, but in this case acid groups remain. However, sodium phthalate shows no reaction even at 200 0C. That relevance of these reactions was confirmed since it was shown by spectroscopy that phthalate is also present exclusively as di-ester when PA is added to a board with the resin.
We conclude that equation 1], derived from studies using anhydrous solvents, does not represent the reaction of PA with wood when the two are simply heated together, or pressed together in a board. This can be confirmed by looking at the affinity of the product for water. Whereas a carboxyl group is more hygroscopic than the hydroxyl it replaces, a di-ester should be much less hygroscopic than two hydroxyls. The equilibrium moisture content at 65% r.h. of wood modified with 2% PA was found to be 7.5%, compared to 9.2% for untreated wood. Once the di-ester is formed, it is unlikely to react further in board pressing, but boards made with this material show the property enhancements. It therefore appears that we have a chemical reaction between PA and wood, and a physical interaction between the modified wood and the resin.
Comparison of different additives We would expect, from the reaction information, that phthalic acid should be comparable in its action to PA. Terephthalic acid may also have an effect, but di-sodium phthalate should have none. A series of boards, all using the same amount of resin, were made to test this. The results (Fig 3) are not in full agreement with the prediction. Terephthalic acid is harmful. Phthalic acid is comparable to PA. But di-sodium phthalate is also beneficial. Whatever the mechanism, it seems unlikely that wood modification is involved.
212 Physical and chemical processing of fibre and fibrous products
Comparison of additives.
Effect on dry and boiled 18S. 0.8 ~------------------------
ca :i ~ 0.4 ....... .....---."""'....---+--4-------------------~
Summary PA greatly improves the properties of particleboard made with PF resin, both when it is used to pre-modify the wood and when it is added with the resin. There is little difference in the effects, so there is probably little difference in the mechanism. PA does not significantly improve the properties ofPF resin or of wood alone.
In solvent-free conditions, PA reacts with wood completely to give di-ester. The product is not hygroscopic. No further reaction with the resin is likely, so the board improvements must result from a physical interaction. Sodium phthalate also improves board properties, but does not react with wood. different mechanism must be involved, which probably also contributes when PAis used.
There is clearly much more work to be done before this system is fully explained. We acknowledge the contribution of Dr H Earl at the start of this work, and the help of Dr D Gerrard ofBP with the spectroscopy.
References (1) R M Rowell, (1983) Chemical modification of wood. Forest Products Abstracts Vol.6 No.12,363-383 (2) H Matsuda, (1987) Preparation and untilisation of esterified wood bearing hydroxyl groups.
Wood Science and Technology 21:75-88