Accelerated physical ageing of poly(1,4-cyclohexylenedimethylene-co-2,2,4,4-tetramethyl-1,3-cyclobutanediol terephthalate)

Successfully evaluating plastic lifetime requires understanding of the relationships between polymer dynamics and mechanical performance as a function of thermal ageing. The relatively high Tg (Tg = 110 °C) of poly(1,4-cyclohexylenedimethylene-co-2,2,4,4-tetramethyl-1,3-cyclobutanediol terephthalate) (PCTT) renders it useful as a substituent for PET in higher temperature applications. This work links thermal ageing and mechanical performance of a commercial PCTT plastic after exposure to 40–80 °C for up to 2950 h. No chemical or conformational changes were found while pronounced physical ageing, measured as enthalpic relaxation, caused yield hardening (28% increase in yield strength) and embrittlement (80% decrease in toughness). Enthalpic relaxation increased with temperature and time to 3.8 J g−1 and correlated to the determined toughness and yield strength. Finally, a 9% increase in Young's modulus was observed independent of temperature and with no correlation to enthalpic relaxation. Enthalpic relaxation followed Vogel–Fulcher–Tammann behaviour, while yield strength and charpy v-notch toughness followed Arrhenius behaviour enabling prediction of the different properties with time and temperature.


Introduction
Increased awareness of environmental impact from plastics has intensied the research into bio-based and recyclable polymeric materials. [1][2][3][4] A class of plastics showing strong potential is polyesters. Polyesters are challenged in high-temperature environments due to their vulnerability to thermally induced mechanical changes which limits service life. These mechanical changes can be subdivided into chemical and physical ageing. Chemical aging e.g. hydrolysis and oxidation can cause polymer cleavage and/or cross-linking. 5 Excessive cross-linking oen results in embrittlement where a reduction in molecular mass reduces mechanical, thermal and rheological properties rendering the plastic defective. 6 Physical ageing arises from super-cooling amorphous materials and is the subsequent time and temperaturedependent dri towards equilibrium. 7 It is a general effect, inuencing the amorphous phases in all polymer materials decreasing the ductile properties. [8][9][10] This polymeric compaction by segmental polymer movement results in reduced free volume, and thus increases density of the amorphous phases. 11,12 Physical ageing is only observed below the glass transition temperature (T g , indicated as onset throughout this paper) where the amorphous phases are solidied out of equilibrium. Thus, heating the material above T g erases physical ageing, presented as enthalpic relaxation, determinable by differential scanning calorimetry (DSC). 13 However, below T g the driving force of physical ageing is the temperature difference between the exposed temperature (T e ) and T g 14 implying that higher service temperature (below T g ) causes faster physical aging.
While maximizing T g -T e is a common strategy for physical aging inhibition, other inhibiting methods include increasing crystallinity, increasing molecular weight and addition of hydrogen-bonding additives. 14,15 Increasing crystallinity reduces the size of the amorphous phases and restricts polymeric movement as a larger fraction of the amorphous phases are restricted by the crystal-amorphous interface. This mobility restriction of the polymer decreases the rate of enthalpic relaxation. 16 Physical aging rate reduction has a nonlinear relationship with crystallinity in e.g. poly(ethylene terephthalate) (PET) an increase in crystallinity from 19-21% causes 29% decrease in enthalpic relaxation rate while an increase from 38-48% causes 23% decrease in enthalpic relaxation rate. 16 It could be argued that moisture acting as plasticiser increases physical aging rate, however, it is demonstrated in PET that moisture splits enthalpic relaxation into two peaks, but not consequently accelerating it. 17,18 Temperature acceleration is commonly evaluated by Arrhenius-factorisation, expressed by an activation energy. 19 The activation energy is the strength of temperature dependence, calculated from time-temperature superpositions (called shi factors). This is particularly useful when predicting properties of commodity materials, as they are typically used at various temperature ranges. However, when approaching T g the relaxation times of the amorphous phase, such as enthalpic relaxation, is known to follow a Vogel-Fulcher-Tammann (VFT) behaviour as given in eqn (1). 20 where R 0 is the rate factor, T e is the exposed temperature, T N is a material parameter or Vogel temperature and B is a shape parameter. The shape parameter B determines the degree of temperature dependence like the activation energy in Arrhenius factorisation. This accelerated behaviour further stress why T g is an important property in evaluating the durability of commodity polymers in varying temperature environments, even when below T g .
PET (T g ¼ 75 C) products are challenged by physical aging in elevated temperature applications. 16 Mechanical implications of physical ageing for PET is observed as a decrease in essential work of fracture and increase in yield strength. 21 These changes in mechanical properties causes physical ageing to be considered the most signicant ageing kinetic in PET that limits its service life. 22,23 PETG (T g ¼ 78 C) is synthesised as PET but substituting a fraction of the ethylene glycol with 1,4-cyclohexanedimethanol (CHDM) causes the end-product to be more ductile but T g only increases by $4 C. 24,25 A potential material could be poly(1,4-cyclohexylenedimethylene-co-2,2,4,4tetramethyl-1,3-cyclobutanediol terephthalate) (PCTT, for its structure see Fig. 1) (T g ¼ 110 C) synthesised by replacing the ethylene in PETG with the rigid 2,2,4,4-tetramethyl-1,3cyclobutanediol (TMCD) causing T g to increase signicantly, $35 C. 26 Thus, PCTT could be suitable for high temperature applications in comparison to PET and PETG. At present there is no literature or studies examining the chemical and mechanical properties of PCTT during thermal aging at temperatures below T g .
This paper contributes to the investigation of physical and chemical ageing of PCTT while evaluating its mechanical properties. Injection moulded specimens were heat-treated at different temperatures (40 C to 80 C) for various periods (until 2950 h) and subsequent changes to chemical, thermal and mechanical properties were determined. Chemical changes were investigated by attenuated total reectance Fourier transform infrared spectroscopy (ATR-FTIR) and nuclear magnetic resonance spectroscopy (NMR) and morphological changes were evaluated by DSC where enthalpic relaxation was used to determine parameters for a VFT-model. Mechanical properties were evaluated by tensile and impact testing and used to determine parameters for an Arrhenius model.

Heat treatments
A specimen set comprises of six tensile specimens, six v-notch charpy specimens and two DSC specimens cut from a tensile specimen. Three ovens (one ED115 E2, BINDER, GER, and two UN110, Memmert, GER) located in an environmentally controlled room (22-24 C, 20-40% RH) were kept at 40, 60 and 80 C, respectively. One set was tested immediately as reference (0 h). 10 sets were placed in each oven and extracted aer 8, 18, 43, 67, 95, 262, 418, 1127, 1605 and 2950 h. Three times six vnotch charpy specimens were heat-treated at 70 C, 60% RH in a humidity chamber (HPP260, Memmert, GER) and were extracted aer 18, 43 and 67 h. Aer extraction all sets were stored for at least 48 h prior measurements for temperature and surface humidity equilibration.

Tensile testing
Test specimens were mounted with the inlet pointing outward in the bottom grip in a universal testing machine (Z005, Zwick Roell, GER) tted with a 25 mm gauge length clip on extensometer (180102/2008, Zwick Roell, GER) all controlled by testXpert II (Zwick, GER). Tensile tests and data extractions were performed according to ISO 527-1. 27 Young's modulus determination was between 0.05 to 0.25% elongation at 1 mm min À1 while yield strength, ultimate tensile strength, stress and strain at break were evaluated at 100 mm min À1 above 0.25% elongation.

Charpy v-notch impact test
Charpy specimens were cut (ZNO, Zwick, GER) according to ISO 179-1 (ref. 28) type A-notch, with a notch tip diameter of 0.5 mm. Specimens were horizontal positioned with the v-notch opposite the pendulum and tested (according ISO 179-1) with a 4 J potential energy pendulum (HIT, Zwick, GER). Data was collected by soware testXpert II (Zwick, GER) by measuring the angle of pendulum rotation as a function of dissipated energy.

Differential scanning calorimetry
A 2 mm thick, B4 mm cylindrical sample was punched from a tensile specimen and placed in a pan (901683.901 Tzero Hermetic Pan, TA Instruments, USA), covered with a lid (Tzero Hermetic Lid 901683.901, TA Instruments, USA) crimped (Tzero Press, TA Instruments, USA), weighed (Quintex 124-1s, Sartorius, GER) and inserted into the DSC (Q2000, TA Instruments, USA). Temperature was swept twice from 20 C to 300 C and back to 20 C at 20 C min À1 . DSC was applied to quantify enthalpic relaxation by integration of the endothermic peak at T g . 29 Enthalpic relaxation (DH(T g )) was determined in Universal Analysis 2000 (V. 4.5, TA Instruments, USA) by numerical integration with a linear extrapolation from the constant heat ow plateau aer T g subtracted the unaged DH(T g ) (t ¼ 0 h) (Fig. 12). Onset T g was determined during cooling, DC p was determined as the change in heat ow over T g during heating and T m was determined from integration with linear baseline from 240 to 280 C.

Arrhenius and VFT factorisation
Arrhenius and VFT plots were created from shi factors (a T ) and reciprocal exposed temperature (1/T e ). Yield strength and enthalpic relaxation shi factors were determined from superposing the logarithmic function, y ¼ a + b log(x), to 40 C. Charpy v-notch shi factors were determined from when toughness was halved and time-superposing to when 60 C was halved. Yield strength and charpy v-notch toughness was t with Arrhenius and enthalpic relaxation to the VFT equation.

Attenuated total reectance infrared spectroscopy (ATR-FTIR)
Spectra were collected on the tensile specimen grip section on FT-IR (Thermo Scientic, iS50, USA) equipped with ZnSe ATR (iD5, Thermo Scientic, USA) from 500-4000 cm À1 with avg. of 8 scans, wavelength dependent penetration depth and baseline were corrected in OMNIC (v. 8.2.388., TA Scientic, USA). Bands 2934 cm À1 , 1719 cm À1 , 1452 cm À1 , 1018 cm À1 , 874 cm À1 , and 729 cm À1 were integrated and normalized by the integrated area of the band at 874 cm À1 as a function of time at 40, 60, and 80 C.

NMR analysis
The NMR assignment and calculations will follow the nomenclature given in Fig. 2 where *, 0 and 00 denotes end-group, trans, and cis atoms, respectively.
The signals from proton-containing adjacent groups (Fig. 4) allows for the assignment of cis and trans conformation in CHDM. Also, the CHDM methyl end-group (4*) at 4.05 ppm has 1 H-1 H cross-peaks at 1.73, 1.66 and 4.85 ppm which are assigned to the equatorial, axial methine proton (5*) and hydroxyl proton (13*). TMCD contains no proton-containing adjacent groups, and thus HSQC is required to support the previous 13 C NMR to 1 H NMR analysis.
The correlation between 1 H NMR and 13 C NMR signals shown in Fig. 5 conrm the cis and trans conformation in CHDM and correlate with the previous 13 C NMR assignment of TMCD. 30 Also, the methyl and methine end-groups present cross-signals in 1 H-13 C NMR at 4.05 ppm-53.42 ppm and 1.73/ 1.66 ppm-22.00 ppm, respectively. Lastly, no cross-signal is presented at the hydroxyl group at 4.85 ppm, conrming the CHDM end-group assignment. Full assignment from 1 H and 13 C NMR is presented in Table S1 ESI. † Assuming linearity i.e. that CHDM is the only end-group of linear polymers, then the average degree of polymerization of the CHDM (C) unit, n C , can be estimated from 1 H NMR by dividing the methyl cis (4 00 , 4.43 ppm) and trans (4 0 , 4.32 ppm) peaks, by the end-group (4*, 4.05 ppm). TMCD (T) average degree of polymerization, n T , can similarly be calculated using the methine cis (10 00 , 4.75 ppm) and trans (10 0 , 4.91 ppm) peaks by the end-group (4*, 4.05 ppm) as given in eqn (2): where I 4 and I 10 are the integrals of their respective peaks in 1 H NMR.
From the degree of polymerization, the average number molecular weight, M n , can be found as given in eqn (3): where M C and M T are the molecular weights of the CHDM and TMCD repeating units (both 274 g mol À1 ), respectively. TMCD/CHDM ratio (f T/C ) was calculated from the carbonyl 13 C-NMR CHDM (1, 167.8 ppm) and TMCD (7, 168.4 ppm) peaks as given in eqn (4): Trans/cis ratios of the two units (r Ctrans/cis , r Ttrans/cis ) were calculated from the methine cyclic carbon peak in CHDM (5 00 /5 0 , 36.9/34.3 ppm) and TMCD (10 00 /10 0 , 82.9/81.9 ppm) as given in eqn (5): Mole fraction of the Dyad sequences was determined from the benzene terephthalic acid 13 C NMR peak at 133.7 ppm and calculated from the concentration of TMCD and CHDM as given in eqn (6): where N ij and X i is the dyad sequence mole fraction and monomer mole fraction, respectively.
Analysing 1 H, 13 C, COSY, HSQC and eqn (1)-(6), the average number molecular mass ( M n ), and monomer fractions (X T , X C , and f T/C ) and dyads were found. None of these values showed signicant change at the 80 C heat-treatment temperature over time (ESI, Fig. S1, S2, Table S2 and S3 †). Average values and standard deviations for all values are therefore calculated and given in Table 1.
Thus, NMR demonstrates no signicant changes in cis/trans conformations, dyads or molecular weight indicating no chemical degradation has occurred during heat treatment.
Ester C]O stretching is assigned to the band at 1719 cm À1 and C-O at 1267 cm À1 . Hexane C-H is assigned to bending at 1452 cm À1 and stretching at 1018 cm À1 . 31 Aromatic C-H bending is assigned to the bands at 873 cm À1 while 1,4-parasubstituted aromatic C-H out of plane stretch is assigned to the 729 cm À1 band. 32,33 Which in combination veried the chemical composition. Fig. 6 shows ATR-FTIR spectra of PCTT samples heat-treated at 80 C as a function of time. Fig. 7 shows no signicant changes to any IR bands, even aer 2950 h of heat-treatment at the highest tested temperature of 80 C. Presenting IR band differences during all heattreatment temperatures (ESI , Table S4 and S5 †) shows no signicant change in intensity except for one initial drop of 6.8% in the 2934 cm À1 CH 2 /CH 3 stretching vibrations. Hence, ATR-FTIR analysis show no chemical change over the examined time and heat-treatment temperature range.

Tensile test
Young's modulus is shown in Fig. 8 as a function of time at the different heat-treatment temperatures.  From Fig. 8 it is seen that aer an initial lack period of 100 h a steady increase in Young's modulus independent of the heattreatment temperature range is observed. The obtained yield strength is shown in Fig. 9 as a function of time at the different heat-treatment temperatures. Fig. 9 show that yield strength increased with time and that temperature induced a signicant shi already aer 8 h. It is further seen that the rate of yield strength increase is larger at higher treatment temperatures. This clearly demonstrates a yield hardening of PCTT with both time and in particular  This journal is © The Royal Society of Chemistry 2019 treatment temperature. Ultimate tensile strength was equal to the fracture stress at 40 and 60 C, however, 80 C exposure resulted in having ultimate tensile strength equal to the yield strength (ESI, Fig. S3 †). Elongation and stress at break reveal no signicant change (ESI, Fig. S4 and S5 †).

Charpy v-notch toughness
Charpy v-notch specimen testing presented impact toughness decrease as a function of time and treatment temperature as shown in Fig. 10.
From Fig. 10 it is seen that charpy v-notch impact toughness decreased sharply aer treatment at the temperatures, 60, 70 and 80 C aer being constant for a temperature-dependent time period. This decrease followed from ductile-to-brittle fracture mode transition, i.e. from yielding fracture to brittle fracture. This transition occurred earlier with higher treatment temperature, demonstrating a ductile-to-brittle transition of heat-treated PCTT with time and accelerated by temperature. However, samples heat-treated at 40 C did not drop below half the initial toughness, and thus did not transition from ductileto-brittle within the investigated time range.
Picturing the fracture surfaces of the heat-treated charpy vnotch bar from the ductile and brittle fracture are shown in Fig. 11A and B, respectively.
A ductile (Fig. 11A) and a brittle (Fig. 11B) fracture surfaces are revealed. The ductile fracture is presented as lamellae drawing while the brittle fracture presents periodic craze bands. Furthermore, the ductile break seems to draw from the surface (le in pictures) while the brittle fracture origins from within the material. Also, the drawing fracture of (A) had multiple origins of fractures whereas the brittle fracture only presents one.

Thermal analysis
Enthalpic relaxation (DH(T g )) determined by DSC is shown in Fig. 12.
T g and DC p presented no signicant change with time at different temperatures (ESI , Table S6 and S7 †). However, as seen from Fig. 12 the endothermic peak at T g clearly increased with time at 80 C. Table 1 Summarised NMR-analysis of poly(1,4-cyclohexylenedimethylene-co-2,2,4,4-tetramethyl-1,3-cyclobutanediol terephthalate) after heat-treatment at 80 C until 2950 h. f T/C : TMCD/ CHDM molar ratio, r Ctrans/cis : CHDM trans/cis molar ratio r Ttrans/cis : TMCD trans/cis molar%,  Enthalpic relaxation as a function of time and heattreatment temperature is given in Fig. 13. Fig. 13 shows that enthalpic relaxation increased with time and treatment temperature induces a clear shi aer just 8 h, 80 C. Further, the enthalpic relaxation plateaued aer approximately 500 h for 80 C while no such behaviour was observed for 40, 60 and 70 C. Finally, enthalpic relaxation was not observed until aer 1000 h at 40 C and 18 h at 60 and 70 C.
Extracting yield strength and charpy v-notch data result in Arrhenius plots (Fig. S9 and S10 †), presented as an activation energy in Table 2.
It is seen that charpy v-notch toughness (Table 2) has the largest coefficient of variation. This is a result from the breakmode transitioning from ductile-to-brittle, decreasing the toughness by 84% from specimen to specimen. Furthermore, charpy v-notch toughness was only evaluated from 60-80 C as 40 C never transitioned from ductile-to-brittle break. Using eqn (1), the VFT parameters for enthalpic relaxation were found as Correlating enthalpic relaxation and toughness are shown in Fig. 14. Fig. 14 demonstrates that once the enthalpic relaxation reaches $0.6 J g À1 then the toughness drops signicantly, regardless of heat-treatment temperature. It is also noted that the 70 C, 60% RH toughness drop was shied to approx. 25% to a higher enthalpic relaxation indicates the inuence of water.
Combining the tensile yield strength and charpy v-notch toughness yields the following correlation (Fig. 15).
Yield hardening is known to cause embrittlement in elements where the correlation is clearly presented here. However, the relationship between yield hardening and charpy v-notch toughness may be correlated due to both phenomena arising from increased physical ageing.

Discussion
NMR-determined molecular weight of 14.9 kDa might be underestimated as the end-groups of the TMCD unit was not identied in NMR analysis (Table 1). Although the molecular masse might be valid if a surplus of CHDM was applied or if it was added later in the synthesis as both will result in only few TMCD terminated polymers. 1 H and 13 C NMR of heat-treated PCTT elements at 80 C (Table 1, ESI Table S1-S3 †) and ATR-FTIR of heat-treated PCTT elements at 40, 60, and 80 C (Fig. 7, ESI Table S4 and S5 †) showed no chemical or cis/trans conformational changes. Furthermore, NMR revealed a CHDM trans/cis ratio of 70/30 as expected from the equilibrium mixture during polymerization (example 5 in ref. 34). TMCD cis/trans   ratio was revealed near 50/50, suggesting a random distribution of cis and trans units. Finally, the dyad ratios match the Bernoullian statistical model with a TMCD/CHDM ratio of 20/80, which supports the measured ratios and random copolymerisation. 35 DSC resulted in negligible changing melting peaks (ESI Table S6 †), no cold crystallisation, small molecule evaporation, or other morphological rearrangements except enthalpic relaxation. Thus, the heat treatment resulted in a physical ageing clearly seen as an increase in enthalpic energy release with longer time and higher treatment temperature ( Fig. 12 and 10). This enthalpic energy is ascribed to the increased local attraction between polymeric segments in the amorphous phase which must be broken when exceeding T g . 36 The described increase in local interactions would assume to decreases the segmental mobility resulting in decreased toughness which is supported by the ductile-to-brittle break transition in Fig. 10. Further, Fig. 14 shows that toughness and enthalpic relaxation correlates across treatment temperatures and time and as expected from the increased density and decreased segmental mobility leading to embrittlement of PCTT. 36,37 Decreased segmental mobility also caused yield hardening, as presented in Fig. 9. Previously it has been argued that yield strength correlates with brittleness 38 but the results suggest that both properties could be caused by the densication and reduced polymeric mobility in the amorphous phases. However, yield strength continues to increase while enthalpic    relaxation indicates a plateau (Fig. 13, 80 C aer $500 h). This seems counter intuitive but the increase in number of molecular interactions can reach a maximum while a continuous rearrangement of existing interaction is still possible which can result in further decrease in molecular mobility. This behaviour has also been observed in PET and is further supported by molecular theory. 8,37 Finally the molecular interactions did not result in signicantly lower elongation at break, indicating polymer rearrangement during the moderate strain rate (4 s À1 ) of tensile testing while brittle results were clearly obtained during the higher strain rates of impact testing (3000 s À1 ) (Fig. 10). 36 Drying PCTT may be expected to decrease the impact strength and deaccelerate the mobility-driven enthalpic relaxation as water acts as plasticiser. However, the 70 C, 60% RH results show that drying the material increased ductile-to-brittle transition (Fig. 14), revealing a plasticizing effect of water on PCTT. 39 Also, the enthalpic relaxation at 70 C ts the VFT model in the investigated temperature-range (ESI, Fig. S9 †). 70 C at 60% RH is also included herein, revealing no signicant enthalpic relaxation acceleration from moisture.
Charpy v-notch toughness and especially yield strength did not express signicant VFT behaviour, but Arrhenius behaviour in the investigated temperature range (ESI, Fig. S7 and S8 †). However, a critical assumption is that the activation energy (E a ) is temperature-independent. 40 Although care needs to be taken when extrapolating outside (especially in the proximity of T g 41 ) the investigated temperature range 42 parameters for prediction of mechanical and thermal behaviour are determined and can be applied in prediction of PCTT performance resulting from different temperature and time conditions.

Conclusion
Examination of PCTT in a temperature span from 40 to 80 C showed no chemical degradation during heat-treatment (up till 2950 h). Although below T g mechanical properties were affected by the heat treatment increasing yield strength by 28% and decreased in toughness of 80%. Further, an increase of Young's modulus by 9% was observed to be independent of treatment temperature. Finally, thermal analysis showed an increase in enthalpic relaxation up to 3.8 J g À1 . The change in properties could be described by an Arrhenius factorisations and constants have been established to predict the performance of PCTT in a typical service temperature range.