**4. Discussion**

It is useful to note that initially, across all of the test types, the single polymers show properties in line with those reported in the literature for those polymers. The main aim in creating the polymer blends was to enhance the properties of PLLA, and we demonstrate that it is possible to use relatively small amounts of PCL and PHBV to achieve this aim. Considering first the two-component blends, the addition of both PCL and PHBV produced polymer blends which were less brittle than PLLA alone processed in the same way. Previous studies have shown that PLLA can be made less brittle through the addition of PCL [12], as long as the PCL is well dispersed within the PLLA. The extrusion process used here seems to have achieved this dispersion. A number of studies have blended either PHB

or PHBV with PLLA and observed improved toughness [24–26]. Blending PLLA with PCL reduces the crystallinity compared to PLLA processed alone using the same processing parameters, with this reflected in a reduced modulus and tensile strength. Blending PLLA with PHBV, however, increases the crystallinity compared to PLLA processed alone using the same processing parameters, with this reflected in an increased modulus and tensile strength, giving an almost ideal outcome of improved modulus, strength and elongation at break from blending. PHBV is considered to act as a nucleation agen<sup>t</sup> for the PLLA [25], giving more crystallisation but smaller crystals, thus enhancing the mechanical properties across the board.

Interestingly, in the three-component PLLA/PCL/PHBV blends, the addition of PCL causes the crystallinity to fall to below that observed for PLLA alone, whilst maintaining the modulus and elongation at break at levels similar to those for PLLA alone, suggesting that whilst the PCL reduces the total amount of crystallinity, the refined crystal structure stimulated by the PHBV still enhances the mechanical properties. Overall, the mechanical properties of the PLLA/PHBV blend and the two PLLA/PCL/PHBV blends offer similar modulus and strength to that of PLLA alone, whilst improving the elongation at break, although for the 90/5/5 PLLA/PCL/PHBV blend, the improvement in elongation at break was limited. This combination of improved properties has not been attained in previously reported work on binary or ternary blends with PLLA.

When considering the creep behaviour of the polymers, the proportion of PCL in the ternary blends seemed to have a greater influence than it had on the initial mechanical properties. That the creep behaviour of the blends cannot be inferred from the trends seen in the initial mechanical properties is important to bear in mind when developing blended polymers for applications which involve the polymer being under load for an extended time period. The 80/10/10 blend had a higher initial strain under load, and a greater rate of strain accumulation than the PLLA, PLLA/PHBV or 90/5/5 ternary blend. We consider that the reduced crystallinity, reduced modulus and increased volume of PCL in the 80/10/10 blend made it easier for the polymer material to creep through increased localised yielding [40], which then led to an increase in the steady state creep rate.

The addition of PCL into the blends also has a greater impact on viscosity than on the static properties. Testing of molten material means that any benefits which are dependent on crystallinity are no longer applicable, but it is interesting to note that in the ternary blends, it is the influence of PCL which dominates over the PHBV, which in the PLLA/PHBV blend causes an increase in viscosity. In essence, it appears that the low-viscosity PCL acts effectively as a plasticiser for the PLLA and PHBV. This makes the blends which contain PCL easier to melt to create a component shape using conventional polymer processing techniques or, as we have filament raw material, the additive manufacturing technique of fused deposition modelling.

In evaluating materials for potential use in orthopaedic applications on the basis of their mechanical properties, we should first note that the materials described here are appropriate for cancellous bone applications (modulus in the range 10 MPa to 3 GPa, strength 0.1 to 30 MPa), as the mechanical properties do not approach those of cortical bone (modulus ~18 GPa; strength ~70 MPa) [41]. With that constraint, the combination of initial mechanical properties and creep behaviour mean that the PLLA/PHBV blend and the 90:5:5 PLLA/PCL/PHBV blend would be the most promising materials. The enhanced elongation at break of the PLLA/PHBV blend offers further value where greater elasticity is required. Future work will consider the biological properties of the materials, but the track record of the constituent materials in biomedical applications gives confidence that the materials will have biological properties appropriate for orthopaedic applications.
