Tube Expansion Deformation Enables In Situ Synchrotron X-ray Scattering Measurements during Extensional Flow-Induced Crystallization of Poly l-Lactide Near the Glass Transition

Coronary Heart Disease (CHD) is one of the leading causes of death worldwide, claiming over seven million lives each year. Permanent metal stents, the current standard of care for CHD, inhibit arterial vasomotion and induce serious complications such as late stent thrombosis. Bioresorbable vascular scaffolds (BVSs) made from poly l-lactide (PLLA) overcome these complications by supporting the occluded artery for 3–6 months and then being completely resorbed in 2–3 years, leaving behind a healthy artery. The BVS that recently received clinical approval is, however, relatively thick (~150 µm, approximately twice as thick as metal stents ~80 µm). Thinner scaffolds would facilitate implantation and enable treatment of smaller arteries. The key to a thinner scaffold is careful control of the PLLA microstructure during processing to confer greater strength in a thinner profile. However, the rapid time scales of processing (~1 s) defy prediction due to a lack of structural information. Here, we present a custom-designed instrument that connects the strain-field imposed on PLLA during processing to in situ development of microstructure observed using synchrotron X-ray scattering. The connection between deformation, structure and strength enables processing–structure–property relationships to guide the design of thinner yet stronger BVSs.

: Four thermocouples (T1 to T4) are placed at different positions along a customized Pyrex mold to 23 probe azimuthal and axial gradients in temperature induced by the IR lamps oriented (a) parallel to the preform 24 and (b) perpendicular to the preform. T1 and T2 are located ~30mm from the center of the mold while T3 and 25 T4 are at the center of the mold. T1 and T3 probe the temperature of the mold surface directly facing the lamps 26 while T2 and T4 probe the temperature of the mold surface 90° from T1 and T3. Temperature traces are plotted 27 for a lamp exposure time of (a-b, ii) 50s and (a-b, iii) 180s.

28
It is desirable to achieve uniform heating of the preform prior to expansion for homogeneous 29 deformation and wall thickness. The IR lamps (OD:10mm and length:118mm) can be oriented either 30 parallel or perpendicular to the preform. We tested both configurations ( Fig. S1a-b,i) using a 31 customized Pyrex mold (ID: 8mm, OD:10mm and length: 60mm) with incisions for thermocouples 32 made at the center (one facing the lamps and the second 90° away) and 30mm from the center (one 33 facing the lamps and the second 90° away). For an exposure time of 50s, the difference in temperature across all four thermocouples is ~10°C for the parallel configuration but is ~20°C for the perpendicular 35 configuration (compare Fig.S1a-b, ii); prolonged exposure (180s) increases these gradients to ~15°C 36 for the parallel configuration and >30°C for the perpendicular configuration (compare Fig. S1a-b, iii).

37
It is reasonable to expect that the parallel configuration minimizes axial gradients in temperature as 38 the lamps illuminate a greater portion of the mold surface. Based on these data, we orient the lamps 39 parallel to the mold for the experiments described in this report.  is presented as (a-b, ii) 2D maps at different positions along the axis of the mold and the preform. The 2D data 69 are averaged to calculate (a-b, iii) from the inner to the outer diameter to obtain the azimuthal distribution of 70 absorbed energy (bin size is 30°) for the mold (left, a-b, iii) and the preform (right, a-b, iii); Bin 1 is facing the 71 lamps and Bin 3 is 90°C from the plane containing the lamps.

72
The two 500W IR lamps have an operating temperature of 2900K; Planck's law is used to estimate 73 the fraction of light absorbed by Pyrex (the mold) and poly L-lactide (PLLA, the preform) in accord 74 with their IR absorption spectra [1]. Pyrex is mostly transparent to IR radiation between 1 to 2.7µm 75 (Table S1), a region where PLLA strongly absorbs IR radiation (particularly between 2.2 to 2.7µm, 76 Table S1). Beyond 3µm, Pyrex absorbs most of the IR radiation and hardly any light passes through 77 (Table S1). We estimate that for a 1mm thickness, the mold and the preform absorb ~15% and ~14% 78 of incident IR radiation respectively (Table S1).

79
The fraction of energy absorbed by Pyrex and PLLA guide the setup of ray tracing simulations 80 in Zemax to determine the distribution of energy in the perform and the mold (Fig. S2). The preform 81 (ID: 0.64mm, OD: 1.52mmm, length: 60mm) is placed inside the mold (ID: 3.9mm, OD: 6.0mm, length: 82 60mm) and the two IR lamps (OD: 10mm, total length: 118mm, filament length: 82mm) are positioned 83 25mm on either side of the preform (Fig. S2, i) in the parallel configuration (described in Figure S1). As the two 500W IR lamps are operated at 48V, they are assigned an effective power of ~100W in 85 Zemax (of this, it is assumed that the mold can absorb a maximum of 15W and the preform can absorb 86 a maximum of 14W). Ray tracing calculations are performed with and without curved reflectors (arc 87 length: 42mm, axial length: 82mm, radius: 25mm), placed 10mm from the OD of the lamps, to note 88 their impact on the absorbed radiation (Fig. S2, i). The simulations provide insight on the distribution of energy with and without reflectors along heating of the preform, which is desirable to minimize quiescent crystallization prior to expansion.

96
The simulations indicate minimal gradients in the absorbed energy along the z-axis (~10% from the

106
S3), suggesting that there are no "hot spots" in the preform that can lead to non-uniform expansion.

107
The faster heating rate of the preform is in agreement with experimental data (Fig. S4). The 108 instrument was modified to probe the temperature of the mold, a PLLA preform inside the mold,

109
and a PLLA preform outside the mold during the heating and annealing steps (Fig. S4a); for a set 110 mold temperature of 85°C, the PLLA preform inside the mold is ~20°C warmer at the onset of 111 annealing (Fig. S4b). Furthermore, the PLLA preform temperature does not increase during the 112 annealing step (Fig. S4b), indicating that the rapid increase in strain beyond the inflection point (

136
The instrument is mounted on an optical bench at the beamline that can be translated in the varies from ~400µm (preform) to ~140µm (expanded tube), which is less than half that of the Pyrex 148 mold (~1mm). As a result, relatively strong scattering from the mold obscures WAXS features of the 149 expanded tube, particularly in the vicinity of q ~1.5 Å -1 (Fig. S6). Diffraction patterns acquired on the 150 mold alone indicate a ~15% variation in the scattered intensity (~300 counts, Fig. S7), which is ~50% 151 of the scattering from PLLA alone (~700 counts, Figs. 5-6 and Figs. S11-S14). Therefore, direct 152 subtraction of the Pyrex background from PLLA+Pyrex frames results in under or over-subtraction.

153
The

163
We rescale the background by identifying q-intervals for the PLLA preform that are mostly 164 unchanged before and after expansion. The prior literature informs us that scattering from PLLA is 165 negligible at a low-q interval of 0.5-0.6 Å -1 . However, we do not have access to scattering below 0.68 166 Å -1 at this beamline due to the size of the beamstop. Therefore, the lowest possible q-interval available 167 to us is 0.68 to 0.75 Å -1 (indicated by a black box in Fig. S6); the intensity in this q-interval changes 168 during inflation (t < 100s, Fig. S8a) but hardly varies post expansion (t > 100s, Fig. S8a). At high-q, we 169 use an interval spanning 1.80 to 1.90 Å -1 (indicated by a black box in Fig. S6) as the intensity in this 170 region varies by ~ 3% before and after inflation (Fig. S8b). Therefore, we use the average intensity in 171 these low-q (0.68 to 0.75 Å -1 ) and high-q (1.80 to 1.90 Å -1 ) intervals to two define two parameters (α

184
The two-parameter subtraction method is applied in 2D to rescale the background, which is 185 subsequently subtracted pixel by pixel from the frame of interest. To test our approach, the rescaled 186 background is subtracted from each of the Pyrex frames in Figure S7       The intensity of the SAXS peaks increases during the first half of annealing (<200s Figs S15-20b) 254 but decreases rapidly during cooling (>350s, Figs. S15-20b). On the other hand, the interlamellar 255 spacing monotonically decreases with time post expansion during both the annealing and the cooling 256 steps ( Fig. S22a-b, right). We hypothesize that a combination of oriented crystallization and changes 257 in density driven by temperature explain the observed trend in the SAXS data. The inflation of the 258 tube imposes strains in excess of 400% at the inner diameter, which is likely to induce "shish-kebabs" 259 along the θ-direction of the tube (meridional peaks in the SAXS patterns, Figs. S15-20b). This       annealing and cooling steps. The corresponding temperature profiles are presented in Figure 3 of the main text.