Radiochemical ageing of poly(ether ether ketone) Emmanuel Richaud, Paulo Ferreira, Ludmila Audouin, Xavier Colin, Jacques Verdu, Carole Monchy-Leroy
To cite this version:
Emmanuel Richaud, Paulo Ferreira, Ludmila Audouin, Xavier Colin, Jacques Verdu, et al.. Ra- diochemical ageing of poly(ether ether ketone). European Polymer Journal, Elsevier, 2010, 46 (4), pp.731-743. 10.1016/j.eurpolymj.2009.12.026. hal-02269369
HAL Id: hal-02269369 https://hal.archives-ouvertes.fr/hal-02269369 Submitted on 22 Aug 2019
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Radiochemical ageing of poly(ether ether ketone)
Emmanuel Richaud a,b,*, Paulo Ferreira a, Ludmila Audouin a, Xavier Colin a, Jacques Verdu a, Carole Monchy-Leroy b a Arts et Metiers ParisTech, CNRS, PIMM, 151 bd de l’Hôpital, 75013 Paris, France b EDF R&D, MMC, Site des Renardières, 77818 Moret Sur Loing, France
abstract
Thin (60 lm) and thick (250 lm) samples of poly(ether ether ketone) were subjected to radiochemical ageing at 24 kGy h 1 dose rate for doses up to 30.7 MGy at 60 °C in air. FTIR spectrophotometry (hydroxyl and carbonyl build-up), ATR microscopy (oxidation profiles), ammonia gaseous treatment (determination of carbonyl nature), density, DSC (glass tran- sition temperature, cold crystallization and melting point changes), and gel content mea- surements (crosslinking) were conducted for examination of polymer degradation. Thin Keywords: samples were shown to undergo principally chain scission process whereas thick ones Poly(ether ether ketone) undergo mainly crosslinking. This difference can be attributed to the kinetic control of oxi- Radiochemical ageing Oxidation dation by oxygen diffusion. A mechanistic scheme was proposed from radiochemical yields Chain scissions estimations. Crosslinking
1. Introduction conditions where oxidation effects are negligible because oxidation is diffusion limited and thus restricted to a very In the aim to lengthen the nuclear plant lifetime with- thin superficial layer. Since dose rates are considerably out replacing expensive materials, and guarantee they lower in practical applications, oxidation is expected to can safely operate beyond their 40 years license, many re- play a non negligible role. Hence, we were interested in searches were conducted on the radiochemical ageing of studying radiochemical ageing in conditions favouring oxi- cable insulation [1] made of rubbers, plasticized PVC or dation (lower dose rate, lower sample thickness), and polyethylene, or polymers of high thermomechanical per- appreciating the differences between anaerobic and oxida- formances such as poly(ether ether ketone) (PEEK). PEEK tive ageing as well from chemical (FTIR, derivation treat- is a semi-crystalline polymer relatively insensitive to com- ments) as from physical (DSC, density, sol–gel analysis) mon solvents including water. Its amorphous phase has a measurements. glass transition temperature (Tg) close to 140 °C. A rela- tively high stability in radiochemical ageing is expected due to its aromatic content. PEEK thermal ageing has been 2. Experimental widely studied [1–7] in comparison with radiochemical ageing. Sasuga et al. [8] have studied the effect of irradia- 2.1. Materials tion under vacuum. The changes of mechanical properties and glass transition temperature revealed thus the pre- PEEK (381G grade – VICTREXÒ) was supplied as pellets. dominance of crosslinking. Irradiation in the presence of These latter were extruded to obtain nearly amorphous oxygen was recently studied [9] at high dose rate i.e. in 60 lm thin films using the following procedure: pellets were first dried (3 h, 150 °C) [10] and then processed in a twin screws Brabender extruder (T = 350 °C, * Corresponding author. Address: Arts et Metiers ParisTech, CNRS, zone 1 1 PIMM, 151 bd de l’Hôpital, 75013 Paris, France. Tzone 2 = 380 °C, Tzone 3 = 390 °C at 10 rev min , and torque E-mail address: [email protected] (E. Richaud). between 72 and 75 N m). Extruded films were directly quenched in water so that to obtain quasi-amorphous 2.2.4. Sol–gel analysis samples. Gel content was estimated by immersing samples in hot PEEK was also supplied as 250 lm thick sheaths with an trichloroacetic acid (purity P 99.0%, titration grade sup- internal diameter close to 1000 lm wrapping a copper plied by Sigma Aldrich, melting point c.a. 60 °C) at 70 °C conductor. Their initial density about 1.290 corresponds in closed vessel. About 15 g of acid were used to swell to an initial crystallinity ratio close to 23%. Sheaths were polymer (m0 mass c.a. 20 mg). After 24 h, solution was fil- irradiated with the Cu conductor. It means that only one tered and sample was cautiously washed with distilled surface was in contact with air versus two for films. water. Insoluble fraction mass (mI) was weighted after dry- ing under vacuum (80 °C – 24 h). The gel fraction wI was 2.2. Characterization thus
wI ¼ mI=m0 ð3Þ 2.2.1. FTIR spectrophotometry 2.2.1.1. Transmission. FTIR spectra were recorded in trans- mission mode on an IFS 28 apparatus (Brucker). Free 2.3. Irradiation standing films spectra were obtained by the averaging of 60 32 scans at a 4 cm 1 resolution. Irradiations were performed using a Co c-rays source at 60 °C under 1.5 bar air pressure (‘‘Brigitte” device, SCK CEN, Mol, Belgium) at 24 kGy h 1 dose rate. Statistical 2.2.2.1. Attenuated Total Reflexion microscopy. Thickness of chain scission concentration in a Red Perspex sample oxidized layer was estimated by Micro-ATR Spectroscopy (PMMA) allows the total emitted dose to be measured with using a Spotlight 300 apparatus coupled with Spectrum accuracy better than 5%. 100 FTIR spectrophotometer (Perkin-Elmer) driven by Spectrum Image software. Sheaths were embedded in a 3. Results commercial epoxy-amine resin. After curing, samples were polished with 800 and 2400 granulometry disks with a 3.1. Thin samples cold water flow. The sample-resin interface was located using PEEK absorptions at 1485 and 926 cm 1. The spectra 3.1.1. FTIR spectrophotometry were recorded at a 16 cm 1 minimal resolution with IR spectra of virgin and irradiated samples are shown in 1.56 lm pixel size. Fig. 1. The main changes are:
2.2.3. Gaseous ammonia treatment – a wide band in the 3000–3500 cm 1 range with a double NH3 derivatization was used to precise the nature of maxima at 3320 and 3260 cm 1 ascribed to hydrogen oxidation products. Ammonia was generated from an equi- bonded (possibly phenolic) hydroxyl groups (denoted molar solution of NH4Cl (13.4 g) and NaOH (10 g) in 100 ml POH in the following), of water. 24 h treatments were performed in a closed ves- – a sharper one in the 1600–1800 cm 1 range with a max- sel at room temperature before FTIR analysis. ima at 1725 cm 1 attributed to various types of carbonyl groups (denoted by PC = O in the following). 2.2.4. Density measurements Density was measured at 22 °C using a Mohr balance To improve clarity, the spectrum of virgin sample was (Mettler Toledo). Samples ranging from 20 to 50 mg were subtracted to spectra of degraded ones and zooms of successively weighted in air (m1) and in acetone (m2) (sup- respectively hydroxyl and carbonyl regions are presented 3 plied by VWR, density qAcetone = 790 kg m ). PEEK density in Figs. 2a and 3a. Broad hydroxyl band displays a maxi- 1 (qPEEK) was thus assessed from mum at about 3320 cm indicating that OH groups are strongly linked to oxygen atoms and aromatic structures m1 qPEEK ¼ qAcetone ð1Þ m1 m2 4.0 Crystallinity ratio was then obtained using 3.5 qC q qA xC ¼ ð2Þ 3.0 q qC qA 2.5 with qA = 1.260 and qC = 1.400 [11]. 2.0 20.0 MGy 1.5 10.0 MGy 30.7 MGy
2.2.5. Differential Scanning Calorimetry Absorbance 1.0 DSC experiments were carried out on a TA Instrument 0.5 Q1000 apparatus driven by Q Series Explorer software. DSC was previously calibrated with Indium standard. 0.0 40080012001600200024002800320036004000 Approximatively 10 mg of sample in a closed aluminium Wavenumber (cm-1) pan was subjected to a 10 K min 1 ramp from 50 to 400 °C (or from 400 to 50 °C) under nitrogen purge with Fig. 1. Full range FTIR spectra of 60 lm amorphous PEEK films at 1 a 50 ml min flow rate. different irradiation doses (up to 30.7 MGy). a 0.80 by hydrogen bonds. The negative peak in the 3000– 3050 cm 1 indicates a partial consumption of aromatic 0.70 C–H groups. 0.60 A negative peak at 1660 cm 1 corresponding to the con- -1 -1 0.50 3320 cm 3260 cm version of initial benzophenone unit into other species [12] 0.40 30.7 MGy is observed. Three carbonyl species absorbing at 1690, 20.0 MGy 1 0.30 1725 and 1775 cm appear with a clear predominance 1 Absorbanc 10.0 MGy of 1725 cm product. 0.20 A gaseous ammonia treatment induced an increase of 0.10 the band intensity in the 3200–3400 region (Fig. 4a) and 0.00 an absorbance decrease in the carbonyl region (Fig. 4b).
3800 3600 3400 3200 3000 As previously established, NH3 converts carboxylic acids -1 Wavenumber (cm ) into carboxylate salts (not observable here because the resulting signal overlaps with virgin PEEK absorptions) b 1.20 and esters to give amides [13,14]. The observed changes 1.00 correspond well to the conversion of initially present aro- matic esters (absorbing at 1725 cm 1) into amides ) -1 0.80 (stretching N–H vibrations at 3300 cm 1). 1 0.60 The growth of hydroxyl groups (at 3320 cm ) and car- y = 0.0356x bonyl ones (making the assumption that these latter are 0.40 R2 = 0.9901 essentially esters) was quantified using Beer–Lambert law: [POH] (mol l DO 0.20 ½PC ¼ O ¼ 1725 ð4Þ e:e1725 0.00
0 5 10 15 20 25 30 35 DO3320 dose (MGy) ½POH ¼ ð5Þ e:e3320 Fig. 2. Subtraction IR spectra changes in the hydroxyl region (a) and where kinetic curve of hydroxyl build-up from 3320 cm 1 signal growing-up (b). – e is the sample thickness (cm).
a 0.6 a 0.75 1725 cm-1 3475 cm-1 30.7 MGy 20.0 MGy 0.5 0.50 1775 cm-1 10.0 MGy 0.4 0.25 0.3
0.00 0.2 Absorbance Absorbance -1 -0.25 0.1 3060 cm
0.0 -0.50 3800 3700 3600 3500 3400 3300 3200 3100 3000 1850 1800 1750 1700 1650 Wavenumber (cm-1) Wavenumber (cm-1) b Wavenumber (cm-1) b 0.25 1850 1800 1750 1700 1650 1600 y = 0.0064x 0.0 0.20 R2 = 0.9610 ) -1 -0.5 -1 0.15 1725 cm -1.0 0.10 Absorbance
[PC=O] (mol l -1.5 0.05 -2.0 1660 cm-1 0.00 0 5 10 15 20 25 30 35 -2.5 dose (MGy)
Fig. 4. Effect of NH3 treatment on the IR spectrum of 30.7 MGy irradiated Fig. 3. Subtraction IR spectra changes in the carbonyl region (a) and sample (subtracted spectra between treated and untreated sample) in the kinetic curve of carbonyl build-up from 1725 cm 1 signal growing-up (b). hydroxyl region (a) and in the carbonyl region (b). – e3320 and e1725 are the molar absorptivities equal to – The glass transition temperature at about 140 °C. 70 l mol 1 cm 1 [15] for hydroxyl groups and – The cold crystallization at about 170 °C. 600 l mol 1 cm 1 for carbonyl ones [16–18]. – The melting point at about 340 °C.
The corresponding kinetic plots (Fig. 2b for hydroxyls, 3.1.2.1. Glass transition temperature. Tg decreases almost Fig. 3b for carbonyls) are almost linear. The slope for hy- linearly with irradiation dose (Fig. 6). A linear regression droxyl groups is gave d½POH T ¼ 144:75 2:3 10 7 d ð12Þ ¼ 3:6 10 8 mol l 1 Gy 1 ð6Þ g dd d being in Gy, and Tg in °C. T dependance with molar mass is given by the Fox–Flo- d½POH g and ¼ 10 7 GðPOHÞ I ð7Þ ry equation [19]: dt kFF Tg ¼ Tg1 ð13Þ – d being the dose (Gy) MN 1 – I being the dose rate (Gy s ) 1 kFF being the Fox–Flory constant (K mol kg ). – G being the yield of formation (groups/100 eV) According to Saito [20], the reciprocal average number molar mass varies with both chain scission (s) and cross- Since d ¼ I dt ð8Þ linking (x) concentration:
The radiochemical yield for POH formation is 1 1 ¼ s x ð14Þ M M d½POH N N0 GðPOHÞ¼107 ¼ 0:36 groups per 100 eV ð9Þ dd By neglecting the existence of crosslinking event in a first approach (which can be partially justified by the absence After a 30.7 MGy irradiation, the hydroxyl concentration is of gel content for irradiated films as mentioned below), close to 1.1 mol l 1, i.e. 0.87 mol kg 1 what corresponds we have thus nearly to one hydroxyl group every four monomer units. The esters grow-up almost linearly with dose: 1 1 Tg0 Tg ¼ kFF ¼ kFF s ð15Þ MN MN0 d½PC ¼ O ¼ 0:64 10 8 mol l 1 Gy 1 ð10Þ dd dTg Thus ¼ kFF ð16Þ which gives ds
dT 7 d½PC ¼ O dT g GðPC ¼ OÞ¼10 Since g ¼ dd ð17Þ dd ds ds dd ¼ 0:06 groups per 100 eV ð11Þ 7 Carbonyl formation would be thus a relatively rare event and s ¼ 10 GðsÞð18Þ compared to hydroxyl formation. 107 dT Then GðsÞ¼ g ð19Þ 3.1.2. Differential Scanning Calorimetry kFF dd
DSC thermograms of irradiated amorphous films are Any PEEK kFF value was reported in literature to our knowl- shown in Fig. 5. Three characteristic transitions are 2 3 edge. Assuming that kFF is proportional to Tg1, Tg1; Tg1 observable: (see ‘Appendix’), converging values ranging between 169
Temperature (°C) 147 100 150 200 250 300 350 400 146 4.0 145 144 2.0 y = -0.2323x + 144.75 143 T R2 = 0.8863 0.0 g 142 (°C) g
initial T -2.0 141 5.0 MGy 140 -4.0 10.0 MGy 139 Heat Flow (°C) Heat -6.0 138 20.0 MGy 137 -8.0 30.7 MGy 0 5 10 15 20 25 30 35 -10.0 dose (MGy)
Fig. 5. DSC thermograms of 60 lm thick amorphous films for different Fig. 6. Decrease of glass transition temperature with irradiation dose for exposure doses. 60 lm thick amorphous films. and 201 K kg mol 1 were calculated. Taking 185 K 345 45 kg mol 1, one obtains: GðsÞ¼0:012 340
40 ) 335 -1 (°C)
3.1.2.2. Cold crystallization characteristics. The cold crystal- g (J M M T lization temperature TCF and the corresponding enthalpy 330 H 35 Δ DHCF increase with irradiation dose (Fig. 7). TCF increases slowly from 173 °C to about 174.5 °C at 30.7 MGy. DHCF in- 325 y = -0.6124x + 340.62 creases markedly from nearly 20 J g 1 for the virgin film to R2 = 0.9766 1 about 34 J g for 30.7 MGy irradiated one. This is not sur- 320 30 prising because the polymer undergoes chain scission 010203040 facilitating the polymer crystallization [21]. dose (MGy)
Fig. 8. Melting point temperature ( ) and melting enthalpy (j) against 3.1.2.3. Melting characteristics. Melting enthalpy under- irradiation dose for quenched 60 lm thick films. goes small non-monotonic changes. In contrast, the melt- ing point decreases almost linearly with dose rate (Fig. 8). The slope of straightline in Fig. 8 is 6 dT lC0 0:61 10 M ¼ 6:1 10 7 KGy 1 ð20Þ i:e: ¼ 1 þ d ð24Þ dd lC TM1 TM0
TM decrease can be attributed to the limiting effect on crys- d being expressed here in MGy. Literature data [22–24] give: T = 663 ± 5 K. talline lamellae thickness lC of radiation induced structural M1 modifications. According to Gibbs–Thompson equation: l Thus C0 ¼ 1 þ 1:2 10 8 d ð25Þ 1 2r T lC l ¼ 1 M ð21Þ C q DH T C M1 M1 8 i:e: lC ¼ lC0:ð1 1:2 10 dÞð26Þ where A 30.7 MGy irradiation would induce a 35% decrease of – r is the surface energy between crystalline and amor- lamellae thickness: 2 phous phase (J m ), 8 6 lC ¼ lC0:ð1 1:2 10 30:7 10 Þ 0:65:lC0 – TF1 and DHF1 are respectively the melting temperature and the melting enthalpy of an infinite crystal, It means that the positive effect of chain scission on
– qC is the crystalline phase density. crystallinity [25] is compensated and even dominated by the negative effect of irradiation induced structural irregu-
Supposing that TF1, DHF1, qC, and r changes are negligible larities working as comonomers limiting the lamella during ageing, it gives growth. l T T C0 ¼ M1 M ð22Þ 3.1.3. Gel content measurements lC TM1 TM0 Thin films (60 lm) remained soluble in hot trichloro- acetic acid even at high irradiation dose. It suggests that l T T i:e: C0 ¼ 1 þ M0 M ð23Þ chain scission is dominating which was already observed lC TM1 TM0 for many polymers when ageing is not controlled by oxy- gen diffusion (see ‘Section 4’). These results are confirmed by rheometry experiments (not presented here): the New- tonian viscosity decreases as soon as the beginning of 175.0 exposure. 34 174.8 174.6 32 3.2. Thick samples 174.4 30 ) 174.2 -1 3.2.1. FTIR analysis by Attenuated Total Reflection microscopy 28 (J g (J (°C) 174.0 The cross-section of the epoxy embedded irradiated CF CF T 173.8 26 H sheaths was analysed by FTIR in ATR mode. Absorbance Δ 1 173.6 24 at 3320 cm changes with depth gave the profile of 173.4 Fig. 9. Two interesting observations can be made: 173.2 22 173.0 20 1. The hydroxyl concentration profile displays a horizontal 0 5 10 15 20 25 30 35 plateau in the superficial layer. dose (MGy) 2. The thickness of oxidation layer assessed from the POH depth profile is on the order of 50–100 m, depending Fig. 7. Cold crystallization temperature TCF ( ) and enthalpy DHCF against l irradiation dose (j). on the chosen criterion. 0.25 0:007 l ¼ 250 60—70 lm 0:027 0.20 in good agreement with the above micro-ATR results.
0.15 3.2.3. Differential Scanning Calorimetry 3320 Thermograms of 250 lm thick samples are shown in DO 0.10 Fig. 10. In these extruded samples, the initial crystallinity ratio is close to its maximum ( 25%) what explains the 0.05 quasi-absence of cold crystallization exotherm and the weakness of heat capacity change at T . T values were 0.00 g g 0 50 100 150 200 therefore hardly measurable. Even if a small increase was depth (µm) observed, data scattering prevents from a reliable theoret- ical exploitation. Fig. 9. Changes of oxidized layer profile for irradiated sheaths: j: The melting peak is displaced slightly towards low tem- 10 MGy, : 20 MGy, N: 30.7 MGy. peratures and tends to split into two components. How- ever, the melting enthalpy remained almost constant. We were interested in studying crystallization from the melt during cooling from 400 to 50 °Cat10°C min 1 ramp. 3.2.2. Density measurements Comparison of 30.7 MGy irradiated sample with virgin Density measurements were monitored on irradiated sample (Fig. 11) calls for the following comments: sheaths. An increase from 1.290 for virgin sample to 1.297 for 30.7 MGy irradiated one was measured. 1. In irradiated samples, the first heating melting endo- Since density decrease can be attributed to destruction therm tends to split into two components. An explana- of crystals under irradiation [26,27], our results confirm tion is proposed in the ‘Section 4’ part. Nevertheless, the the stability of PEEK crystals under the employed dose rate. melting enthalpy remained nearly constant, which sug- Polymers being exposed in their glassy state, chains in gests that PEEK crystals are quite stable under the con- amorphous phase cannot integer the crystalline one be- sidered dose rate, in agreement with the absence of cause of restricted mobility. The observed density increase density decrease. cannot be caused by chemicrystallization but rather from 2. The ‘‘hot” crystallization peak of irradiated sample is oxidation since ‘‘heavy” oxygen atoms are grafted. Let us shifted towards lower temperatures by more than denote by M the average atomic mass: A 30 °C and its enthalpy is strongly decreased. This is con- MA ¼ mU=N ð27Þ firmed by the lower values of melting temperature for the second heating. It indicates that lamellae thickness where of crystallites formed from molten state is lower. Indeed, crosslinking prevents chain segments to integer – m is the molar mass of the representative structural U a crystal [28]. unit (RSU). – N is the number of atoms of RSU. 3.2.4. Gel content measurements Thick samples became partially insoluble at a dose d For non irradiated PEEK, one has G such as 5.0 MGy < dG < 10 MGy. From samples having 1 mU0 ¼ 288 g mol undergone sol–gel transition, the Charlesby–Pinner plot [29] was tried: N0 ¼ 34 1 7 MA0 ¼ 8:47 g mol GðsÞ 10 w þ w1=2 ¼ þ ð29Þ S S 2G x The 100% conversion of aromatic rings into phenolic by- ð Þ MW0:GðxÞ d products would give