Terpene polyacrylate TPA5 shows favorable molecular hydrodynamic properties as a potential bioinspired archaeological wood consolidant

There is currently a pressing need for the development of novel bioinspired consolidants for waterlogged, archaeological wood. Bioinspired materials possess many advantages, such as biocompatibility and sustainability, which makes them ideal to use in this capacity. Based on this, a polyhydroxylated monomer was synthesised from α-pinene, a sustainable terpene feedstock derived from pine trees, and used to prepare a low molar mass polymer TPA5 through free radical polymerisation. This polymer was extensively characterised by NMR spectroscopy (chemical composition) and molecular hydrodynamics, primarily using analytical ultracentrifugation reinforced by gel filtration chromatography and viscometry, in order to investigate whether it would be suitable for wood consolidation purposes. Sedimentation equilibrium indicated a weight average molar mass Mw of (4.3 ± 0.2) kDa, with minimal concentration dependence. Further analysis with MULTISIG revealed a broad distribution of molar masses and this heterogeneity was further confirmed by sedimentation velocity. Conformation analyses with the Perrin P and viscosity increment ν universal hydrodynamic parameters indicated that the polymer had an elongated shape, with both factors giving consistent results and a consensus axial ratio of ~ 4.5. These collective properties—hydrogen bonding potential enhanced by an elongated shape, together with a small injectable molar mass—suggest this polymer is worthy of further consideration as a potential consolidant.


Results and discussion
Monomer synthesis from α-pinene. We have recently reported the synthesis of the terpene-derived triol 3 (Fig. 1) and its co-polymerisation with succinic acid via step-growth polymerisation 46 . The resulting polyesters were shown to be stereoregular; a feature that was exploited through the formation of a polymer stereocomplex 46 . As hydrogen bonding potential is anticipated to be a valuable property in wood consolidants, we were keen to investigate the polyhydroxylated derivative 3 for this purpose. Nevertheless, it was anticipated that the aforementioned polyesters would be biodegradable, a feature which is of course undesirable in this context. Accordingly, we instead investigated the formation of an acrylate derivative.
The synthesis of the acrylated monomer 4 was achieved in four steps from the cheap and readily available residue material α-pinene ( Fig. 1 and also Supplementary Information for full synthetic details of 1, 2 and 3). The first step is epoxidation to the corresponding oxirane α-pinene oxide (1). The epoxidation of α-pinene has been the focus of extensive research (as detailed by Corma Canos et al. 46 ) and it is currently synthesised on an industrial scale, as well as being widely available commercially. The next step involved forming trans-sobrerol (2), which is a common hydrolysis product of 1 and is a known molecule from our in-house library 35,47 .The triol 3 was synthesised from 2 using a Brown hydroboration/oxidation sequence 48 and subsequently isolated as a single diastereomer, the stereochemistry of which was confirmed by X-ray crystallography. It should be noted that a higher yield for 3 can be achieved via column chromatography, at the expense of the diastereomeric ratio 46  www.nature.com/scientificreports/ Polyhydroxylated compounds such as these have a lot of untapped potential, not only for wood consolidation purposes but also for other applications such as drug delivery 49 . The final step in the monomer synthesis involved functionalising 3 with an acrylate moiety. Previous work 35 has demonstrated the (meth)acrylation of a number of alcohols derived from several terpenes like (+)-α -pinene, (−)-β-pinene and (R)-(+)-limonene. The addition of a (meth)acrylate functional group is a well-known process widely employed to enable a variety of monomers to undergo free radical polymerisation (FRP). In one such example 35 , hydroxyl-functionalised terpenes were initially esterified with (meth)acryloyl chloride using triethylamine as a base. This route was further optimised by substituting the (meth)acryloyl chloride with (meth)acrylic acid. Although these acids are not currently renewable, it is likely they will both become commercially available from a sustainable source in the future 35 . Additionally, the use of (meth)acrylic acid generates 'green' and less toxic waste as opposed to chlorinated waste, making the reaction more sustainable and scalable.
Propylphosphonic anhydride (T3P) was used to promote the ester coupling between the acrylic acid and 3. This reagent is particularly attractive since its by-product is a relatively harmless and water-soluble triphosphate, again a preferable alternative to chloride waste generated when using acryloyl chloride. The reaction resulted in a mixture of diastereomers 4a and 4b. The final product was successfully isolated in two separate fractions: one enriched in the major diastereomer 4a and another enriched in the minor diastereomer 4b. The fraction enriched in the major diastereomer 4a was isolated in a yield of 19% with a ratio of 4a:4b = 22.8:1. The fraction enriched in the minor diastereomer 4b had a yield of 23% with a ratio of 4a:4b = 1:4.6. The total yield of diastereomers 4a and 4b was therefore 43%, in a ratio of 4a:4b = 1.3:1. The diastereomers 4a and 4b were used together as a mixture for the subsequent polymerisation attempts. Diastereomer 4a shows a large 11 Hz coupling on the proton adjacent to the acrylated oxygen (which is upshifted from 3.5 to 4.8 ppm by the electron-withdrawing carbonyl), consistent with a coupling between two axial protons.
Polymerisation of triol acrylate, TPA5. The triol acrylate monomer was then polymerised via conventional free radical polymerisation (FRP) to produce polymer TPA5, employing AIBN as an initiator and 1,4-dioxane as the solvent at 70 °C. The reaction was monitored using 1 H NMR and gel permeation chromatography (GPC) in THF ( Supplementary Fig. S1). The investigated polymerisation reached 46% monomer conversion after 24 h under these reaction conditions as determined by 1 H NMR. TPA5 was then isolated by precipitation in acetonitrile as a white solid, with subsequent analyses confirming that the desired polymer was obtained ( Supplementary Fig. S2). GPC analysis confirmed a number average molar mass (M n ) of 2.6 kDa, a weight Figure 1. The synthesis route for the final monomer. An epoxidation of the starting terpene was carried out to form α-pinene oxide 1, which was then hydrolysed to give trans-sobrerol 2. This in turn was used to carry out a hydroboration/oxidation reaction to give the triol 3, which was afterwards functionalised with an acrylate group to form the monomer as a mixture of two diastereomers 4a and 4b. These were then polymerised to form TPA5. www.nature.com/scientificreports/ average molar mass of 3.8 kDa (relative to polymethylmethacrylate standards) and a broad polydispersity (Đ) of 1.5. These results were promising for our application as the polymer obtained appeared to have a low molar mass which was highly suitable for wood penetration. Nevertheless, higher conversion and narrower polydispersities can potentially be envisaged by using controlled polymerisation techniques such as reversible addition-fragmentation chain-transfer polymerisation (RAFT) or atom transfer radical polymerisation (ATRP).
Solubility testing. The severe decay of the Oseberg artefacts is such that most of them are only held together by the alum present in their wooden structure. The objects with the highest degree of degradation cannot be treated with polymers in aqueous solvents as has been done previously with the Mary Rose 50 and the Vasa 15 . This is because the alum remaining in the artefacts may dissolve in aqueous solvents and consequently exit the wooden structure, leading to total disintegration. As a result, it is deemed preferable to treat the most deteriorated artefacts with polymers in organic solvents. Before continuing with the characterisation studies, it was therefore essential to determine which organic solvents the polymer was soluble in. Solvents which were tested included isopropanol, a 1:1 mixture of toluene in ethanol, ethyl acetate and acetone. The polymer was found to be soluble in isopropanol and the ethanol/ toluene mixture, but insoluble in both ethyl acetate and acetone. It was therefore decided to carry out all further characterisation studies in isopropanol as this solvent has already been used in conservation studies and its high volatility ensures that it does not interact with the wood 51,52 . Furthermore, it has a low surface tension which allows for total permeability of the wood cells 52 .
Calculation of the partial specific volume v. We essentially followed the method of Kratky et al. 53 The density measurements were carried out using a concentration series of TPA5 in isopropanol. The v was then obtained by plotting all the density measurements against concentration ( Fig. 2) and consequently making use of Eq. 1. This was calculated to be (0.753 ± 0.040) cm 3 /g. This value appeared to be consistent with other v values previously reported for synthetic polymers such as polymethacrylate 54 , and was subsequently used for the sedimentation velocity and equilibrium analyses.

Molar mass determination and distribution.
Various loading concentrations were used to run the sedimentation equilibrium experiment in order to determine whether the measured or apparent molar mass M w,app was dependent on concentration. The data was analysed with SEDFIT-MSTAR 55,56 , resulting in a plot of M w,app vs concentration for each sample. The M w,app was obtained by extrapolating the M* function to the cell base, as well as by using the hinge point method ( Table 1). The 'hinge point' is defined as the radial position where the local concentration is equal to the initial loading concentration 55,56 . This analysis also provided information on the apparent z-average molar mass M z,app , which along with the M w,app could then be used to calculate Đ. Table 1 shows a comparison of the M w,app , M z,app and Đ values obtained from the sedimentation equilibrium study for all the tested concentrations. These results demonstrated that there was no significant change in M w,app with concentration, indicating that it was not concentration dependent and not affected by non-ideality. It may therefore be assumed that the M w,app determined by this experiment is in fact the ideal molar mass of the polymer at each given concentration (M w,app ~ M w ). The overall average M w was calculated by plotting the values obtained in the sedimentation equilibrium experiment against concentration (Fig. 3). This was determined to be (4.3 ± 0.2) kDa. www.nature.com/scientificreports/ The M w values obtained by sedimentation equilibrium and the GPC analysis that was previously carried out were compared, with the M w obtained by sedimentation equilibrium proving to be slightly larger. It was expected that the values obtained from AUC would have a higher degree of accuracy as a result of it being an absolute method, meaning that it is matrix-free and does not require calibration standards.
The sedimentation equilibrium data for the 4.0 mg/mL concentration was additionally analysed with the MULTISIG algorithm. This programme provides a distribution of the molar masses by making use of a 17-component system with 20 iterations, assuming thermodynamic ideality 57 . It was used to give a broader resolution of the M* data, while also providing information about the M w distribution of the polymer system. Figure 4 revealed a M w distribution ranging from 2.3 to 9.3 kDa, with components peaking at 2.6 and 6.1 kDa. This was consistent with the SEDFIT-MSTAR analysis, confirming that the polymer has a low M w and thus adding credibility to the results previously obtained by GPC. This further reinforced the status of this polymer as a very promising lead for the purpose of consolidation, since a low weight average M w of ~ 4.3 kDa would increase the probability of it successfully penetrating archaeological wood.
Determining the heterogeneity of the polymer system. Figure 5 shows the sedimentation coefficient range c(s) 58 vs s of the polymer in isopropanol, run on SEDFIT. This algorithm normalises the sedimentation coefficient values to standard conditions (density and viscosity of water at 20.0 °C) 59 . The analysis confirmed that the polymer system is comprised of different components, as previously demonstrated by the MULTISIG analysis. It revealed the presence of a high concentration of a very low molar mass species, with a smaller population of two larger sedimentation coefficient, s value species. This trend was observed in all the concentrations that were used in this study.
Calculation of the intrinsic viscosity η . It was initially planned to carry out the viscosity measurements of TPA5 with a conventional Ostwald U-tube capillary viscometer 60 . This however did not prove ideal due to the high volatility of isopropanol. It was therefore decided to use a rolling ball viscometer at 8.0 °C in order to counteract this issue. After measuring the viscosity of the polymer solution at 6.0 mg/mL, the [η] was calculated to be (5.27 ± 0.11) mL/g with the Solomon-Ciuta equation (Eq. 2).  The programme ELLIPS1 was used to estimate the conformation of the polymer using the Perrin function (P) and the viscosity increment (ν) shape factors. ELLIPS1 is based on a simple ellipsoid of revolution model, where two of the three ellipsoid axes are of equal value 62 . Both P and ν are universal shape parameters, meaning that they are described by a function of shape not size 63 , and are obtained respectively from the sedimentation coefficient and [η] 64 .
P is related to the frictional ratio (f /f o ) , which can be derived from the sedimentation coefficient s 20,w (Eq. 3) 65 : ν is described by Eq. 4: [66][67][68] Once calculated, ELLIPS1 was then used to report these shape functions in terms of their axial ratios (a/b) for ellipsoids of revolution 63 . Table 2 shows the values obtained for these parameters. A sedimentation coefficient (s 20 , w ) -normalised to the standard solvent conditions of the viscosity and density of water at 20.0 o C -was used in these calculations (Fig. 6). Along with the axial ratios, ELLIPS1 also provides a visual representation of the approximate shape of the macromolecule. The shape factors were calculated using different degrees of 'solvation' or solvent association or "dynamic binding" (v s /v) , in order to determine whether the final value estimate for the type and axial ratio was significantly affected. Both P and ν were seen to alter slightly to higher or lower values according to the (v s /v) , however they both gave consistent results, indicating that the polymer had an elongated shape, as shown by Fig. 7.
A consensus value of ~ 4.5 was obtained from the mean of the all the determinations. An additional shape factor, the Scheraga-Mandelkern ( β ) parameter, was used as a consistency check. This is a hydration-independent function, but it is very insensitive to shape (Eq. 5): 69 A value of β = (2.39 ± 0.23) × 10 6 was calculated. This corresponded to a large axial ratio range, but it was nonetheless consistent with the value for the a/b found for the P and ν functions. Additionally, this value is only compatible with an elongated, prolate ellipsoid molecule, as opposed to a flat disc or oblate model like lignin (Fig. 8). This therefore adds further confidence to the shape analyses that were previously carried out.

Conclusion
In this study, the successful synthesis of an acrylated triol monomer 4 has been described. A new protocol for the radical polymerisation of such acrylated terpene monomers was designed and implemented, successfully yielding polymer TPA5. This methodology may potentially also be applied to other types of (meth)acrylated monomers derived from different terpenes. The polymer proved to be soluble in isopropanol, a common and relatively non-toxic solvent which can easily be used in conservation studies. The hydrodynamic characterisation analyses from sedimentation equilibrium demonstrated that the polymer had a M w of ~ 4.3 kDa, reinforced by the approximate value (relative to polymethylmethacrylate standards) obtained by GPC. Sedimentation velocity showed that the polymer system had a degree of heterogeneity, with a high population of a low molar mass species. Conformation analyses indicated that the polymer had an elongated shape, with both P and ν giving consistent results and the β-function confirming an elongated prolate rather than a flat oblate disc shape.
The collective properties of our synthesised bioinspired polymer TPA5 appear to satisfy the essential molecular criteria as a consolidant: high hydrogen bonding potential enhanced by an elongated shape, together with a small injectable molar mass which can be cured into larger, stronger and stable structures once infused or injected into the wood, suggesting this polymer is worthy of further consideration as a potential consolidant.
In future papers in this series we will explore the wood penetration ability of this and related polymers and the ability, after curing, to form a strong stable hydrogen bonded network with porous, degraded wood structures. The interactions with other materials such as calcium hydroxide nanoparticles, used to lower the acid levels in archaeological wood 51 , and their ability to interact in a compatible way with other consolidants, such as the sheet like lignin molecules 70 will be also be explored. In archaeological wood, lignin has a much slower degradation profile than cellulose, so consolidants should ideally be able to network or interact with the remaining lignin within the wood.

Materials and methods
Materials. All reagents and solvents were purchased from a chemical supplier (Acros Organics, Alfa Aesar, Merck, Sigma Aldrich or Fischer Scientific UK) and used without further purification. Water was deionised before use. Brine is a saturated aqueous solution of sodium chloride. Thin layer chromatography was performed on silica gel mounted on aluminium and was visualised using a potassium permanganate dip with gentle heating. Rotary evaporators under reduced pressure were used for solvent evaporation. Dry solvents were obtained from solvent drying towers and contained < 17 parts per million (ppm) of water. Experiments carried out under an inert atmosphere employed argon by means of a Schlenk line or a balloon. High resolution mass spectrometry. Electrospray ionisation (ESI) high-resolution mass spectrometry (HRMS) analyses were performed on a Bruker micrOTOFII mass spectrometer (Bruker Daltonik, Bremen, Germany), interfaced to an Agilent 1200 HPLC (Agilent Technologies, Santa Clara, USA). Samples were presented in solution for analysis by Flow Injection, 1 μL of solution being injected into the ion source of the instrument along with a flow of 0.2 mL min −1 of 70% methanol/water eluent. The mass spectrometer was operated in electrospray ionisation (ESI) mode at a typical resolving power of 8000. Control of the analysis was performed through Bruker's Compass Open Access QC automated data acquisition and reporting software (v1.3; Bruker Daltonik, Bremen, Germany).

Fourier-transform infra-red spectroscopy.. A Bruker Tensor 27 FT-IR spectrophotometer with an
ATR attachment was employed. The measurements were performed in the range of 4000-650 cm −1 and spectra were analysed using OPUS software (Bruker Corporation, Germany).
Monomer and polymer synthesis. The synthesis of α-pinene oxide (1), trans-sobrerol (2) and the triol (3) followed the methods described by Thomsett et al. 46 and are detailed in the Supplementary Information.
Synthesis of triol acrylate, 4. 3 (2 g, 10.6 mmol) and butylated hydroxytoluene (BHT) (ca. 19 mg) were added to acetonitrile (13 mL). Triethylamine (4.5 mL, 32 mmol) was added to the reaction mixture. The solution was cooled to 0 °C, after which propylphosphonic anhydride (T3P) (50 wt. % in ethyl acetate, 7.6 mL, 12.8 mmol) and acrylic acid (acid with low H 2 O content, 99.5% stab. with ca. 200 ppm methoxyphenol, 0.8 mL, 11.7 mmol) were added. The mixture was left to stir for 10 min and then allowed to warm to room temperature. The mixture was left stirring for a further 24 h and then concentrated under reduced pressure. The crude product was purified by column chromatography to yield the title compounds as a white solid (4a and 4b) (1.10 g, 4a:4b = 1.3:1, Table 2. The calculated values for the two different shape parameters P and ν at different degrees of solvent association (dynamic binding), along with the axial ratios determined by ELLIPS1. The axial ratios obtained by adding or subtracting the error from the calculated value (± 0.20 for P and ± 0.23 for ν) are also shown.

Hydrodynamic characterisation studies. Gel permeation chromatography (GPC). An Agilent 1260
Infinity Series HPLC (Agilent Technologies, USA) fitted with a differential refractive index detector (DRI) was used. THF (HPLC grade, Fisher Scientific) was used as the eluent at room temperature with two Agilent PL-gel mixed-E columns in series at a flow rate of 1 mL min −1 . A calibration curve was made using polymethylmethacrylate standards with ASTRA software (Wyatt Technology, USA). This was used for the determination of the M n , M w and molar mass distribution (dispersity, Đ = M w /M n ).
Density measurements-calculation of the partial specific volume v. An Anton Paar DMA 5000 V5.003 was used at 20.0 °C. A 9.0 mg/mL stock solution of TPA5 in isopropanol was prepared and then diluted to 8.0, 7.0, 6.0, 5.0 and 4.0 mg/mL. These concentrations were thereafter used for the density measurements. The partial specific volume-needed for the sedimentation velocity and equilibrium experiments-was evaluated following the procedure of Kratky et al. 53 .
Analytical ultracentrifugation (AUC). Two Beckman Optima XL-I analytical ultracentrifuges with Rayleigh interference optics were used at 20.0 °C. 12 mm optical path length double sector cells with titanium centrepieces were employed.
Sedimentation equilibrium. Loading concentrations of 0.5 to 4 mg/mL of TPA5 in isopropanol were used. 100 μL of each concentration were injected in the sample solution channel of the AUC cell. Isopropanol was used as the reference solution. The experiment was carried out at a rotational speed of 45,000 rpm over 2 days. The results were analysed with SEDFIT-MSTAR 55 in order to obtain the M w,app , making use of the M* function and extrapolation 56 and the hinge point method. No significant concentration dependence was observed suggesting that non-ideality was not significant. The data obtained from the highest concentration (4.0 mg/ml) was additionally analysed with the MULTISIG algorithm 57 to evaluate the molar mass distribution. Viscosity measurements-calculation of the intrinsic viscosity. An Anton-Paar AMVn (Graz, Austria) rolling ball viscometer was used at a temperature of 8.0 °C. Its closed capillary system is more suitable for working with volatile solvent systems compared to conventional Ostwald viscometers. The viscosity measurements were carried out using a 6.0 mg/mL concentration of TPA5 in isopropanol. The intrinsic viscosity was then calculated with the Solomon-Ciuta equation: (6) [η] = 1 c 2 η sp − 2ln(η r )