Understanding the Effect of the Dianhydride Structure on the Properties of Semi-aromatic Polyimides Containing a Biobased Fatty Diamine Arijana Susa *, Johan Bijleveld, Marianella Hernandez Santana, Sanago J. Garcia Number of Pages: 1 (S1 to S1) Number of Figures: 8 Number of Tables: 1 Novel Aerospace Materials group, Faculty of Aerospace Engineering Delft University of Technology, Kluyverweg 1, 2629 HS, Delft, The Netherlands Currently at Institute of Polymer Science and Technology (ICTP-CSIC) Juan de la Cierva,, 286, Madrid, Spain *e-mail: a.susa@tudelft.nl List of figures and tables: Page No.: Figure S1. 1 H NMR spectra of the four polymers in CDCl and their assignment to the molecular structure. S2 S4 Figure S2. Figure S. Figure S4. Figure S5. Figure S6. TGA curves showing weight loss with temperature of the four PIs in the as synthesized state. DSC traces of the second heating curves, showing glass transitions of the four PIs in the as synthesized state. Normalized dielectric loss (ε ) of PI samples with different dianhydrides at selected temperatures (25, 4, 5 and 7 o C). Temperature dependence of the average relaxation time for the segmental mode of PIs. Arrhenius plots of the time-temperature superposition shift factors at for the glass transition relaxations of the four PIs. S5 S5 S6 S7 S8 Table S1. Activation energies for the glass transition relaxations. S8 Figure S7. Figure S8. Polarized microscopy images of the polymer showing the increase in birefringence with annealing time. SAXS diffractograms of polymer showing the annealing effect on the structural ordering. S9 S1 S1
CDCA-D,B,HI Yield Percent yield of the polymer was calculated according to equation: (%) = 1! "!#!! $ (Eq. S1) where predicted mass of the product was calculated according to the stoichiometric balance, assuming that 1 mol of a dianhydride (DAh) and 1 mol of DD1 give 1 mol of PI and 2 mol of water (4.25 wt% of water): %&'' () h +(,/. = m(dah)+m(dd1)-m(h2) (Eq. S2) 1 H NMR spectra Solution state 1 H NMR spectra were collected using the Agilent-4 MR DD2 at 25 C at 4 MHz. The solutions of polymers were prepared in CDCl. Spectra were referenced to the solvent residual peak for TMS. Spectra were not normalized. 6.917 6.98 7.191 7.216 7.28 7.252 7.257 7.684 7.684 l7.74 BC7.257 2. 2. 6.5 4.19 F C H C B N A D E 7.191 7.216 7.28 7.252 G N 6.5 AI.582 H n E6.917 6.98 4.19.546.565.582.565.546.51.51 G1.555 1.55 1.552 6.6 1.172 1.174 1.176 1.665.8 F6.8 8.1.8 1.172 1.174 1.176 1.55 1.552 1.555 6.6 6.8 8.1 25 2 15 1 5 ppm (f1) 7. 6. 5. 4.. 2. 1.. S2
B,DCFF DPA-D 7.191 7.677 7.695 7.677 7.71 7.695 7.819 7.71 7.819 l7.84 A2. 2.2 2. -D F C.66 F A C F F N B F F F N D 67.1 CE.619.6.6.619.66 1.485 1.61 1.257 1.181.81 6 5 4 2 1 2. 2.2 2..55 7.77.55 D1.61 n 1.485 1.257 1.181 67.1 E.81 7.77 ppm (f1) 8. 7. 6. 5. 4.. 2. 1.. S
,B,DClB 6 8.16 7.887 7.191.619.66.654 1.622 1.54 1.178 1.267.799 5 -D8.16 A2. 4.11 7.887 F N C F N D A B 8.22 C.66 E.619 n 1.54 1.267 1.178 8.22 E.799 4 2 4. D1.622 68.7 1 2. 4.11 4. 68.7 ppm (f1) 8. 7. 6. 5. 4.. 2. 1.. Figure S1. 1 H NMR spectra of the four polymers in CDCl and their assignment to the molecular structure. S4
Weight (wt%) 1 9 8 7 6 5 4 2 1.-.6% weight loss at 2 C Weight (wt%) 1 98 96 94 92 9 DPA-D 24 28 2 6 4 44 T ( C) 5 1 15 2 25 5 4 45 5 T ( C) Figure S2. TGA curves showing weight loss (wt%) with temperature of the four PIs in the as synthesized state. endo ffset heat flow (a. u.) DPA-D -4 4 8 12 16 2 T ( C) Figure S. DSC traces of the second heating curves, showing glass transitions of the four PIs in the as synthesized state. S5
Broadband dielectric spectroscopy (BDS) analysis 1,5 1, T=25 C DPA-D 1,5 1, T=4 C DPA-D BDPA-D ε " /ε " max,5 ε " /ε " max,5, 1-1 1 1 1 1 2 1 1 4 1 5 1 6 Frequency (Hz), 1-1 1 1 1 1 2 1 1 4 1 5 1 6 Frequency (Hz) 1,5 1, T=5 C DPA-D 1,5 1, DPA-D T=7 C ε " /ε " max ε " /ε " max,5,5, 1-1 1 1 1 1 2 1 1 4 1 5 1 6 Frequency (Hz), 1-1 1 1 1 1 2 1 1 4 1 5 1 6 Frequency (Hz) Figure S4. Normalized dielectric loss (ε ) of PI samples with different dianhydrides at selected temperatures (25, 4, 5 and 7 o C). The dielectric relaxation processes were analyzed quantitatively by fitting the frequency spectra to the Havriliak- Negami (HN) function 1,2 given by: 5 / (1) = / 2 + 678(#9: ;< ) = >? (Eq.S) The difference in dielectric constant measured at low and high frequencies is the dielectric strength ( ε) of the relaxation and it is related to the area under the absorption curve given by ( / = / / 2 ), where / 2 and / are the unrelaxed and relaxed values of the dielectric constant respectively. τhn is the HN relaxation time, representing the most probable relaxation time of the relaxation time distribution function, and b and c are shape parameters ( < b, c 1) which describe the symmetric and the asymmetric broadening of the equivalent relaxation time distribution function, respectively. Parameters b and c can be associated to the structure and polymer architecture heterogeneity with respect to the bulk polymer. The HN relaxation time τhn is related to the frequency of maximum loss, fmax =1/(2πτmax), by the following equation: 4 S6
A BCD = 7 EF GHI = A JK ' LF E8E $M7/L ' LF E8E $7/L (Eq. S4) Both characteristic relaxation times coincide when the relaxation spectrum is symmetric, i.e. c=1. The temperature dependence of the segmental relaxation times (τmax) was also studied. This dependency is well stated by the Vogel-Fulcher-Tammann (VFT) equation: 5-8 A BCD = A N + P QMQ R $ (Eq. S5) where τ and B are temperature-independent parameters, and T is the so-called ideal glass transition or Vogel temperature which is found to be -7 K below Tg. 9 To reduce the effect of misleading parameters on data fitting to the VFT equation over a limited frequency range, a value of A N 1-14 s was assumed, according to the values empirically found for many polymer systems lying between 1-14 and 1-12. The dependence of τmax with temperature is depicted in Figure S5. This temperature dependency shows a clear curvature typical for cooperative motions. -log (τ max /s) 7 6 5 4 2 α relaxation DPA-D 1-1 -2 2,7 2,8 2,9,,1,2,,4,5,6 1/T (K -1 ) Figure S5. Temperature dependence of the average relaxation time for the segmental mode of PIs. Dotted lines correspond to the VFT fit. 1 Kremer, F.; Schönhals, A., Broadband Dielectric Spectroscopy. Springer: New York, 2; p 721. 2 Havriliak, S.; Negami, S., A Complex Plane Representation of Dielectric and Mechanical Relaxation Processes in Some Polymers. Polymer 1967, 8 (4), 161-21. DI: 1.116/2-861(67)921- Böttcher, C. J. F.; Bordewijk, P., Theory of Electric Polarization. Elsevier: 1978; Vol. II. 4 Richert, R.; Angell, C. A., Dynamics of Glass-forming Liquids. V. n the Link Between Molecular Dynamics and Configurational Entropy. J Chem Phys 1998, 18 (21), 916-926. DI: 1.16/1.47648 5 Fulcher, G. S., Analysis of Recent Measurements of the Viscosity of Glasses. J Am Ceram Soc 1925, 8 (6), 9-55. DI: 1.1111/j.1151-2916.1925.tb1671.x 6 Vogel, H., The Temperature Dependence Law of the Viscosity of Fluids. Physikalische Zeitschrift 1921, 22, 645-646. 7 Tammann, G.; Hesse, W., The Dependancy of Viscosity on Temperature in Hypothermic Liquids. Zeitschrift Fur Anorganische Und Allgemeine Chemie 1926, 156 (4). 8 Bohmer, R.; Ngai, K. L.; Angell, C. A.; Plazek, D. J., Nonexponential Relaxations in Strong and Fragile Glass Formers. J Chem Phys 199, 99 (5), 421-429. DI: 1.16/1.466117 9 Fischer, E. W., Light-scattering and Dielectric Studies on Glass-forming Liquids. Physica A 199, 21 (1-), 18-26. DI: 1.116/78-471(9)9416-2 S7
Activation energies for the glass transition relaxations From the shift factors obtained from the TTS mastercurves, the Arrhenius plots (ln at vs 1/T) were constructed and the activation energies were calculated from the slopes of the linear fit according to the Arrhenius equation: 1 & Q = +S T U V W7 Q 7 Q XYZ [\ (Eq. S6) Where at is the shift factor, Ea is the activation energy, R is the gas constant and Tref is the reference temperature at which the mastercurve is constructed. ln a T (-) - -6 - -6 - -6 - -6 - -6 2,8 2,9,,1,2,,4 linear fit y = 9,49x - 125,76 R² =,9754 DPA-D linear fit y = 2,56x - 19,2 R² =,9992 linear fit (I) linear fit y = 42,48x - 17,69 R² =,9925 (II) linear fit y = 29,46x - 94,225 R² =,9948 y = 25,81x - 84,74 R² =,9927 2,8 2,9,,1,2,,4 1/T (K -1 ) Figure S6. Arrhenius plots of the time-temperature superposition shift factors at for the glass transition relaxations of the four PIs. Note that shows two Tg relaxations (I and II) while the other three polymers show only one. Tref=T(tan δmax)=tg. Table S1. Activation energies for the glass transition relaxations Polymer Relaxation process Ea / kjmol -1 Tg 28 DPA-D Tg 271 Tg 245 Tg (I) 5 Tg (II) 215 1 Wood-Adams, P.; Costeux, S., Thermorheological Behavior of Polyethylene: Effects of Microstructure and Long Chain Branching. Macromolecules 21, 4 (18), 6281-629. DI: 1.121/ma174 S8
Birefringence with annealing time a b c Figure S7. Polarized microscopy images of the polymer showing the increase in birefringence with annealing time for a) non-annealed, b) 5 days annealed and c) 11 days annealed samples. S9
SAXS measurements The SAXS diffractograms were collected using Anton Paar SAXSess, with a copper tube operated at 4 ma, 4 kv, using a multilayer mirror for focusing and monochromatisation of the emitted X-ray beam. The detector is a Dectris Mythen photon counting strip detector. The samples have been measured for 2 minutes each, with the sample mounted on a holder in vacuum. The data was corrected for dark-current, transmission and a background (empty holder subtracted). The data was subsequently desmeared using a procedure available in the SAXSQuant software. The data was then regrouped or rebinned to reduce the number of datapoints and noise, taking care to propagate the uncertainties. The separation distances, d, are calculated according to d=2π/q. 11. nm < d 1 <.6 nm annealing time: 1 6 1 day 5 days 11 days Intensity / a.u. d 2 =1.6 nm 1 5 d =1. nm 2 4 6 q / nm -1 Figure S8. SAXS diffractograms of polymer that was annealed at Tann=T(tan δmax)=tg for different times: 1 day (red squares), 5 days (blue circles) and 11 days (green triangles). The annealed samples show two additional peaks that indicate nanophase separation. The calculated separation distances, d, are assigned to each peak. 11 Beiner, M.; Huth, H., Nanophase Separation and Hindered Glass Transition in Side-chain Polymers. Nat Mater 2, 2 (9), 595-599. DI: 1.18/nmat966 S1