2.1 Materials

ε-caprolactone (ε-CL), stannous-2-ethylhexanoate (Sn(Oct)₂) (Acros and Sigma-Aldrich, Saint-Quentin-Fallavier, France), methanol, dichloromethane (Carlo Erba) were used as received. Potassium chloride (KCl) (Acros) was crushed (from 160- to 310-μm diameter), sieved and dried at 130 °C in an oven. Pine was ultra dried at 250 °C, crushed (from 160- to 310-μm diameter), sieved and kept at 130 °C. Ultra-pure water (Milli Q Plus, Millipore, Molsheim, France) was used to study the release of KCl.

¹H NMR spectroscopy was performed on a Bruker AC 250 spectrometer. Tetramethylsilane was used as an internal standard. Size Exclusion Chromatography (SEC) analyses were performed with a Waters 410 permeation chromatograph equipped with two Waters Styragel columns (HR 0.5 and HR 4E) at 37 °C and a refractometer, using THF as the eluent (1.0 ml.min^–1), polystyrene as calibration standards, and Waters Millennium 3.05 as data processing software. Titration of KCl released was made using a Schott Geräte titration unit (Titroline alpha). Blend systems were made with a Bradender PL2000 Plasticorder internal mixer.

2.2 Polymerisation of ε-caprolactone

Typically, a mixture of 20 ml of ε-caprolactone (1 eq.) and 584 μl of stannous octoate (Sn(Oct)₂) was introduced in a reactor equipped with a mechanical stirrer. Depending on the studied parameters, different amounts of water (from 0 to 0.1 eq.), or wood (from 0.1 to 19%) or KCl (from 0 to 1 eq.) were added to the solution of monomer. In all the cases, the polymerisation took place at 100 °C during 24 h.

2.3 Preparation of impregnated wood samples

Pine was ultra-dried at 250 °C. The wood samples were first cut into pieces of 1 × 1 × 2-cm³ dimensions and all pieces were brought to constant weight (m) in a vacuum at room temperature for one hour. There after, the impregnating KCl solution (3 mol.l^–1) admitted under vacuum until all wood samples were covered for 2 h. The samples were dried at 60 °C during 24 h and weighted (M₁ = 1.4 m). At this time, the resulting systems are only consisted of wood and potassium chloride (Wood-KCl). Titration of KCl extracted from different parts of wood samples (extremities, middle and three layers) shows that the fertilizer is well dispersed in all the volume.

To prepare wood–KCl–PCL system, wood–KCl samples were placed under vacuum during 1 h. The mixture of ε-caprolactone and stannous octoate (first heated at 60 °C) was admitted under vacuum until all wood samples were covered for two hours. At the end of this period, the vacuum was released. The impregnated samples were covered by hot sand (130 °C) for 17 h to allow the polymerisation and to prevent the impregnated monomer from evaporating. Weighted samples (M₂ = 1.9 M₁) was analysed to show that the polymerisation occurred in all the volume. The resulting system was composed of wood, KCl and poly(ε-caprolactone) (wood–KCl–PCL) ($\overline{\mathrm{Mw}}$ of impregnated PCL = 15 000 g.mol^–1, determined by SEC).

2.4 Preparation of blends wood–PCL–KCl

In a reactor equipped with a mechanical stirrer, 1 equivalent of ε-CL and 0.01 equivalent of stannous 2-ethylhexanoate were mixed and heated at 120 °C. The reaction was carried on for three hours. Then the polymer in dichloromethane was precipitated in methanol and was dried under reduced pressure. At this stage, $\overline{\mathrm{Mw}}$ (PCL) = 40 000 g.mol^–1.

Composite was obtained by mixing poly(ε-caprolactone), wood (flavour) and KCl at 80 °C for 20 min (20 rpm) with an internal mixer. Small paving stones of composite were implemented by pressing material between two metal plates (1.5-mm thickness) at 130 °C. To obtain the wanted molar masses, the polymerisation was conducted with different amount of water: with no water, the obtained masses were $\overline{\mathrm{Mw}}$ = 40 000 g.mol^–1; with water (0.1 eq.) $\overline{\mathrm{Mw}}$ = 5000 g.mol^–1.

2.5 Study of the release

Impregnated wood samples (1 × 1 × 2 cm³) with KCl and/or PCL were placed in 500 ml of ultra-pure water at pH = 6. At given time intervals, 5 ml were removed and titrated with nitrate silver (0.02 mol.l^–1).

Blended samples (50 mg, 5 × 5 × 1.5 mm³) were prepared and placed in 500 ml of ultra-pure water at different pH (5 or 9). At given time intervals, 5 ml was removed and titrated with nitrate silver (0.02 mol.l^–1).

3.1 Polymerisation of ε-caprolactone

Ultra-dried wood contains about 1–2% of water. The influence of water on polymerisation of ε-caprolactone was studied. Polymerisation rate is determinate by ¹H NMR spectroscopy from the relative intensity of the ester methylene (–C(O)OCH₂–; δ = 4,06 ppm) of PCL and the ester methylene of ε-CL (–C(O)OCH₂–; δ = 4,24 ppm). The curves showed that the rate of polymerisation for sample containing 0.01 equiv of water increased over 2 times in comparison with the polymerisation conducted without water (Fig. 1a and b). On the other hand for higher water amounts (0.1 eq.), the polymerisation rate decreased over 5 times. In addition, the decrease of the molar mass with the increase of water amount showed the influence of water as an additional initiator.

These results were in accordance with the works of Kovalsky [15] and Storey [16] that have studied polymerisation of ε-caprolactone in the presence of alcohols (ROH=BuOH) or water. According to their conclusions, in the presence of an alcohol (ROH), if the ratio [ROH]₀/[Sn(Oct)₂]₀ does not exceed approximately 2, ROH acts mostly as a co-initiator, reacting with Sn(Oct)₂. Above this ratio, ROH becomes additionally and predominantly a chain transfer agent. These conclusions could be transposed to the results obtained with water. When the ratio [water]/[initiator] was less than 2 (polymerisation conducted with 0.01 eq. of water), the conversion time of ε-CL was faster than for the polymerisation conducted without water.

SEC analysis showed that even if the conversion of ε-caprolactone was achieved after 8 h, the molar masses carried on increasing for 15–20 h. This indicated participation of condensation polymerisation whereby water-initiated chains bearing a carboxylic acid head group and hydroxyl tail group acted as A–B monomers.

In addition, the influence of the wood and KCl on polymerisation was evaluated. Indeed wood is an heterogeneous material containing molecules (phenol, lignin…) that could interfere on polymerisation. The polymerisation of ε-caprolactone was studied with different amounts of wood (0.1% to 19%).

The obtained results showed that the polymerisation rate decreased with the increase of the wood amount but after 8 h of reaction, the conversion is achieved in all cases.

Nevertheless, wood amounts have an important influence on the molar masses of PCL: these latter decreased with the increase of the amount of wood: $\overline{\mathrm{Mw}}$ ∼ 55 000 g.mol^–1 after 15 h of reaction for polymerisation conducted without wood, but when 0.5% of wood was included to the monomer, the molar mass of the obtained polymer fell to $\overline{\mathrm{Mw}}$ ∼ 25 000 g.mol^–1 (Fig. 2).

This result could be explained by the presence of residual water and alcohol molecules in the ultra-dried wood [15,16].

Concerning KCl amount, it has a big influence on molar masses of PCL, because KCl is well known to be a hygroscopic compound, and water is known to slow down the polymerisation.

3.2 Release of KCl

The release of KCl was studied on two kinds of systems:

• • the wood samples impregnated with KCl and/or PCL;
• • the blends PCL–wood–KCl

Concerning the wood impregnated samples, the release of KCl was conducted on samples impregnated with PCL or not ($\overline{\mathrm{Mw}}$ (PCL) = 15 000 g.mol^–1). The time course of KCl release is very short in the two cases: 100% of KCl was released after 2 days. This showed that the expected influence of PCL on the release did not occur; PCL did not close the micro-channels of the wood to control the release. This could also be explained by the important porosity of ultra-dried wood that allowed the easy release KCl in all cases.

In the case of blends wood–KCl–PCL [17], different parameters were studied to evaluate their influence on the release of KCl: the pH of the water, the molar masses of PCL and the quantities of wood. Fig. 3 shows the time course of KCl release from blends containing different amounts of wood (1% or 11%), where PCL has different molar masses ($\overline{\mathrm{Mw}}$ = 5000 or 40 000 g.mol^–1) and where the release was conducted at different pH (5 or 9).

The release rate of KCl from composite was in all cases initially fast but slowed after 5 days. Two behaviours can be distinguished: samples having 1% wood content, and samples having 11% wood content. This proves that the main influence on the release is the wood amount. Indeed, KCl release from blends having 11% wood content is more important than from blends having 1% wood content: wood brought porosity to the material, the micro-channels throughout the material allowed an easier diffusion of KCl. Increases in amount of wood produced a corresponding increase in drug release.

The influence of pH was very small. The release rate of KCl from blends seems to decrease with an increase in molar mass of PCL. For example, for samples having 1% of wood, at pH = 5, the release rate of KCl from samples with $\overline{\mathrm{Mw}}$ (PCL) = 5 000 g.mol^–1 was faster than that from samples with $\overline{\mathrm{Mw}}$ (PCL) = 40 000 g.mol^–1. These results were clearly due to the influence of the molar masse of the PCL. The degradation of polymer (either by surface or bulk erosion) can be a process of the release of active agent [18]. In general degradation rate of a polymer increases with increasing the molar mass of polymer. For samples having 0% of wood, at pH = 6 and with $\overline{\mathrm{Mw}}$ (PCL) = 40 000 g.mol^–1, the release rate of KCl was similar to that of the system containing 1% of wood, showing that small amount of wood did not influence the release.