New Developments in the Field of Radiochemical Ageing of Aromatic Polymers

Polymers having an aromatic backbone polymers have a high mechanical strength and a high modulus. Their aromaticity increases their resistance for use in relatively severe conditions especially in aerospace and nuclear industry for which lifetime prediction is a key issue. For example, a challenge for nuclear plants is to extend lifetime from the initially planned 40 years duration to 50 or 60 years, which makes necessary to determine lifetime by a non-empirical method. Since polymers mechanical failure originates from chain scission or crosslinking of the backbone, the ideal method of lifetime prediction would first involve the elaboration of a kinetic model for chain scission and crosslinking. Then, the changes of molecular mass would be related to the changes of mechanical properties using the available laws of polymers physics. Lifetime would be then determined using a pertinent lifetime criterion. A noticeable difficulty comes here from the fact that oxidation, which plays a key major role in chain scission, is diffusion controlled and thus heterogeneously distributed in sample thickness. It is crucial, indeed, to determine experimentally and to predict this depth distribution of chain scission and crosslinking because it will play a key role on fracture properties. This chapter will be henceforward devoted to the effect of aromaticity on radiostability, the effect on temperature on the chain scission/crosslinking competition, the diffusion limited oxidation (which will be illustrated by the effect of dose rate, atmosphere and sample thickness), then some concluding remarks on oxidative stability of aromatic polymers and the possible link with the absence of macromolecular mobility below Tg. We will start by some basics of radiochemistry which are necessary for the good understanding of this paper, and especially the quantitative treatment for crosslinking and chain scission.


Introduction
Polymers having an aromatic backbone polymers have a high mechanical strength and a high modulus. Their aromaticity increases their resistance for use in relatively severe conditions especially in aerospace and nuclear industry for which lifetime prediction is a key issue. For example, a challenge for nuclear plants is to extend lifetime from the initially planned 40 years duration to 50 or 60 years, which makes necessary to determine lifetime by a non-empirical method. Since polymers mechanical failure originates from chain scission or crosslinking of the backbone, the ideal method of lifetime prediction would first involve the elaboration of a kinetic model for chain scission and crosslinking. Then, the changes of molecular mass would be related to the changes of mechanical properties using the available laws of polymers physics. Lifetime would be then determined using a pertinent lifetime criterion. A noticeable difficulty comes here from the fact that oxidation, which plays a key major role in chain scission, is diffusion controlled and thus heterogeneously distributed in sample thickness. It is crucial, indeed, to determine experimentally and to predict this depth distribution of chain scission and crosslinking because it will play a key role on fracture properties. This chapter will be henceforward devoted to the effect of aromaticity on radiostability, the effect on temperature on the chain scission/crosslinking competition, the diffusion limited oxidation (which will be illustrated by the effect of dose rate, atmosphere and sample thickness), then some concluding remarks on oxidative stability of aromatic polymers and the possible link with the absence of macromolecular mobility below T g . We will start by some basics of radiochemistry which are necessary for the good understanding of this paper, and especially the quantitative treatment for crosslinking and chain scission.

Basics of radiochemical degradation
Let us first consider the reaction: A + h  B The radiochemical yield is defined as the number of B molecules that are generated per absorbed joule: www.intechopen.com (1) where: -G' is the radiochemical yield expressed in mol J -1 , -n is the number of moles of B which is formed by the radiochemical reaction, -E' is the amount of absorbed energy in J, -E is the amount of absorbed energy in 100 eV, -N is the number of B molecules, -N Av is the Avogadro's number, -G is the yield in molecules per 100 eV absorbed. Since the absorbed dose (denoted by  and expressed in Gy) is defined as the amount of received energy (in J) per kilogram of polymer, the following equation can be derived: r being the rate of a radiochemical event (in mol l -1 s -1 ), I the dose rate in Gy s -1 and c the concentration of reacted or generated species (in mol kg -1 even if one often considers that it is the same than in mol l -1 ).

Mathematical treatment for chain scission and crosslinking
Irradiation can provoke both chain scission and crosslinking, the relative proportion of these phenomena depending on many factors which will be presented in the following. The radical coupling may lead to trifunctionnal or tetrafunctionnal crosslinking noduli called respectively Y-and H-crosslinkings ( Fig. 1): The radiochemical yields for chain scissions G(s) and crosslinking G(x) in thick samples can be tentatively assessed using www.intechopen.com

-
The Charlesby-Pinner's (Charlesby, 1960) from an analysis of the residual soluble fraction after beginning of sol-gel transition, -Saito's equations (Saito, 1958) describing the average molar mass changes for soluble polymers before the sol-gel transition. The corresponding mathematical treatment of these theories differs for Y-and Hcrosslinking: 1. In the case of H-crosslinking mode: 2. In the case of Y-crosslinking mode: where: -w S = 1 -w I is the soluble fraction, -G H (s), G H (x), G Y (s), and G Y (x) are the radiochemical yields expressed for 100 eV respectively for chain scissions and crosslinking for an H-and a Y-crosslinking mechanism, - is the dose (Gy) -M N0 and M W0 are respectively the initial number and weight average molar mass (kg mol -1 ).

Effect of aromaticity on radiostability
The degradation of aromatic polymers was studied by many authors: 1. By monitoring mechanical properties (Sasuga et al., 1985, see Table 1): These results give a first indication on the relative radiostability of aromatic polymers. Let us first mention that these lethal doses are considerably greater than those compiled by Wilski for polymer having an aliphatic backbone (Wilski, 1987). It can be attributed to the well-known stabilizing effect of aromatic groups which was first observed on methyl methacrylate-styrene copolymers (Alexander & Charlesby, 1954, Kellman et al., 1990 www.intechopen.com Thominette et al., 1991). However, the results of these studies cannot be directly used in kinetic models for aromatic polymers.  Table 1. Dose for reducing strain at break to 50% and 20% of initial value for various aromatic polymers submitted to 5.10 3 Gy s -1 in air (sample thickness ~ 0.1-0.2 mm) (Sasuga et al., 1985).
Data presented in Fig. 2 permit the yield in radical build-up to be calculated according to the simplified scheme for radiochemical degradation in the absence of oxygen: Initiation: Polymer (PH) + h  P° r i Propagation: P° + PH  PH + P° k P -scission: P°  P° + s k R Termination by coupling: P° + P°  P-P + x k C Termination by disproportionnation: Where F is a double bond. The rate of radical formation is: with: Fig. 2 indicates the existence of an asymptote characterizing a steady-state at which the concentration in radicals is: Eq. 10 gives: The corresponding yields for radical build-up are given in Table 2. Let us first mention that Table 1 results correspond to degradation under air, meanwhile Table 2 results correspond to degradation in inert environment, the possible influence of oxygen being discussed in the following. However, it can be checked that the lethal dose (or dose to reach an arbitrarily chosen threshold for a mechanical property) varies oppositely with the estimated G(P°) values.  Table 2. Yields for radical build-up for aromatic polymers irradiated under 1 -5 kGy h -1 dose rate at 77 K.
-By gas yield emission: Concerning the yield for total gas emission, it can be checked that there is no great difference between gamma, proton and electron-beam irradiation (Hegazy et al., 1992b and that these values are considerably lower than those measured for aliphatic polymers such as PE or PP (Schnabel, 1978). An example is worth to be mentioned (Schnabel, 1978): aliphatic polysulfones have a gas emission yield equal to 39 (poly butene-1sulfone) or 71 (poly hexene-1-sulfone) meanwhile PES or PSU gas emission yields are lower than 1, illustrating here the protective effect by aromatic rings. However, the relative proportion of each emitted gas can vary due to some differences in temperature rising under irradiation. As it will be seen below, this difference due to irradiation nature has a lower influence on the radio induced degradation of polymer than temperature, nature of environment atmosphere and oxygen gradient in sample thickness.  Table 3. 103.G(gas) for irradiation at 30°C 100 m films vacuum (Hegazy et al., 1992a) www.intechopen.com 3. Let us turn to the radiation-induced crosslinking observed in PEEK and PSU. Fig. 3 depicts the changes in thermal behavior which are induced by radio-ageing for a thick PEEK sheath under air (Richaud et al., 2010a). As it will be seen later, the observed behavior is comparable to the one obtained for an irradiation under vacuum. Virgin and degraded samples were characterized by DSC heating-cooling-heating cycle.
-In the case of virgin sample, melting endotherms for the first and the second heating ramps are very comparable (peak temperature and enthalpy). The crystallization from molten state is characterized by a sharp exotherm at c.a. 300°C. - In the case of 30.7 MGy irradiated sample, melting endotherm for the second heating ramp and crystallization exotherm are shifted to the lower temperatures with a lower transition enthalpy. During irradiation of semi-crystalline polymers, chain scission and crosslinking occur only in amorphous phase, the crystalline phase undergoes only small changes so that melting characteristics remain almost unchanged at the first heating scan. After a first melting however, the sample is homogenized, and the changes of molar mass, branching ratio or crosslink density resulting from irradiation affect melting and crystallization. Irradiation effects become then observable by DSC. Here, the decrease of crystallization temperature and enthalpy is explained by the occurrence of radiation-induced crosslinking (Sasuga, 1991). This latter disfavors crystallization because it lowers the transport rate of chain segments from the melt to growing crystals. As a consequence, the degree of crystallinity and the lamellae thickness are decreased that explains the observed decrease of melting point and melting enthalpy. Comparable data were obtained by Hegazy (Hegazy et al., 1992c). T g values from this study are reported (Fig. 4): It is well known that for a linear polymer, T g is an increasing function of molar mass. According to Fox and Flory (Fox & Flory, 1950): www.intechopen.com Where: -T g0 is the T g of a virtual infinite polymer (K). -M N is the number average molar mass (kg mol -1 ). -k FF is a constant characteristic of the chain chemical structure (K kg mol -1 ). For a network, T g is an increasing function of crosslink density. According to Di Marzio (DiMarzio, 1964): Where: -T gl is the T g of a virtual linear polymer containing all the structural units of polymer except crosslinks (K). -F is a flex parameter linked to chain stiffness (kg mol -1 ). - A is the concentration in elastically active chains, which is calculated from crosslinks concentration (mol kg -1 ). Under vacuum, chain scission is generally negligible. Above gel point, it seems to us that T g changes are given by combining Fox-Flory relationship and Saito's equation (assuming first that s << x): O n e o b t a i n s : -G(x) ~ 0.012 for amorphous PEEK, -G(x) ~ 0.023 for semi-crystalline one. It seemed to us interesting to compare radio-induced gelation for PEEK, PSU and PES (Fig. 5). Murakami & Kudo, 2007, Richaud et al., 2011, and PEEK (: 24 kGy h -1 , ,,: 60 MGy h -1 for 3 different initial crystalline microstructures) samples irradiated at several temperatures (Vaughan & Stevens, 1995, Richaud, 2010a.
From the above given mechanistic scheme where chain scission results from radical rearrangement and crosslinking results from radical coupling scheme, the following stationary rate expressions can be derived: -Rate of chain scission: -Rate of crosslinking: Assuming that disproporationation is negligible compared to coupling for polymers under study, it comes: So that : G i = 2G(x) in the absence of oxygen. It seems that this relationship disagrees with experimental results in the case of aromatic polymers which will be discussed next. As a conclusion of this short review of experimental results, it is clear that aromatic polymers belong the family of relatively radiostable polymers because primary radiochemical events, presumably radiolytic chain scissions of C-H bonds, have a radiochemical yield less than 0.5, i.e. 5 to 50 times lower than for aliphatic polymers. Aromatic groups are actually able to dissipate a great part of the absorbed energy into reversible processes (fluorescence, phosphorescence...). Data militate in favour of the following ranking: A comparison of the rate of gel content increase (Fig. 4) for samples irradiated at various temperatures and dose rates shows that gelation is more sudden in the case of PSU and more progressive in PEEK which means that radiation generated radicals would be more reactive for PSU than PEEK. Data also show unambiguously: -The protective effect of aromaticity: Kapton®, which is the most aromatic polymer, presents a very limited degradation, or even no significant degradation (Kang et al., 2008, Richaud et al, 2010b). - The role of SO 2 moiety: experimental results for radiochemical yields values (Horie & Schnabel, 1984) shows that aliphatic polysulfones are among the less stable polymers with G(s) values on the order of 10. High yields of SO 2 emission are also observed indicating thus the probable existence of the weak (Li & Huang, 1999, Molnár et al., 2005 carbon-sulfone bond cleavage. It seems reasonable to suppose that such events occur also in PSU even though the yield is considerably lower owing to the well-known protective effect of aromatic nuclei illustrated for instance in studies on isobutylenestyrene. - The role of isopopylidene group: gelation is undoubtedly due to the coupling of alkyl radicals. The difference in gelation rate would be explained by the nature of radicals: in PEEK, only aryl radicals can react meanwhile both aryl and primary methyl radical can react in PSU. Theoretically, it seems that yields in gas emission G(gas) is half of yield radicals G(P°) as illustrated by proposal of degradation mechanisms below: Analogous mechanisms could lead to carbon monoxide formation from PEEK. Data for Kapton®, PEEK and PSU (Table 1 and Table 2) seem also to show that: G(P°) >> 2G(x) and 2G(gas) PEEK and PSU values gathered in this work are compared with those for polyethylene (Khelidj 2006), and those compiled for cyclohexane and benzene (Ferry, 2008) in Table 4 Table 4. Yields for radicals, gas emission, crosslinking and double bond formation (corresponding to disproportionnation process). *: G(H 2 ) is expected to be close to G(gas), **: G(F) is expected to be negligible in aromatic polymers.

°°
These results call for the following comments: 1. The effect of aromaticity on radiostability is confirmed. 2. The ratio G(P°)/G(gas) is surprisingly lower than 1 for cyclohexane, is fairly close to 2 for benzene and PE, but increases for PSU and PEEK. A possible explanation is the existence of very stable radicals that would be detected by ESR but would contribute neither to gas emission nor to crosslinking (see later) as for example:

O O°
Let us also note that the absence of contribution to gas emission seems easily explainable because the probability for generating a short and volatile segments from such macroradicals is very low. 3. The ratio G(P°)/(G(x)+G(F)) is also clearly greater than 2 for PSU and particularly for PEEK meanwhile it is close to 1 in PE and in cyclohexane. The comparison with benzene suggests that this behaviour is due to the macromolecular structure of radicals preventing them to react by coupling to give crosslinking. In fact, even if a negative concavity is observed in Fig. 1 and indicates the existence of a termination process even at 77 K, the relatively low value of G(x) indicates that segmental mobility would be reduced below T g , which will be discussed in a first section dedicated to temperature effect and another one dealing with oxidizability.

Effect of temperature
Data in Table 5 confirm that lethal dose (here. the necessary dose to half the initial elongation at break) and gel dose vary oppositely with temperature meanwhile gas emission yield would increase.  Table 5. Effect of temperature on lethal dose for PES Hagiwara, 1987 andMurakami &Kudo, 2007).
As previously suggested, the only knowledge of changes in macroscopic (engineering) properties does not permit to ascribe the right cause of failure. Yields in chain scission and crosslinking permit to elucidate the degradation mechanism. Let us first comment the T g changes in PES (Li et al., 2006) presented in Fig. 6: www.intechopen.com  (Li et al., 2006).
Before sol-gel transition, T g changes are given by combining Fox-Flory law (Fox & Flory, 1950) with Saito's equation (Saito, 1958) irrespectively of H-or Y-crosslinking mode: Results presented in Fig. 5 indicate that crosslinking predominates above T g , whereas chain scission predominates below T g . Comparable exploitation can be done from other published results (Murakami & Kudo, 2007, Brown & O'Donnell, 1979: All data converge towards the fact that G(x) increases with temperature. Despite some scattering, it also suggests that G(s)/G(x) decreases with temperature. According to Brown and O'Donnell (Brown & O'Donnell, 1979), it falls to 0 above T g . The results in Table 6 can be explained as follows: irradiation creates macroradicals. At low temperature, they react by an unimolecular process which generates a chain scission (possibly accompanied by gas emission). At high temperature, macromolecular mobility is sufficient to permit macroradicals to react by coupling to give a crosslinking. According to Zhen (Zhen, 2001) most polymers can only crosslink above their melting point.

dose rate method T (K) G H (s) G Y (s) G H (x) G Y (x) G(s)/G(x)thickness
Let us now discuss of the possibility of Y-and H-crosslinking.
-At 150°C, Hill (Hill et al., 1998) unambiguously showed that PSU crosslinks in Y-mode. A comparable analysis was performed by Li (Li et al., 2006). A proposal of mechanism is shown below: -Kudo (Murakami & Kudo, 2007) proposed a Y-mechanism also. However, by reexaminating his results using the above equations, his conclusion is questioned. - Richaud (Richaud et al., 2011) proposed a H-crosslinking mechanism at 60°C, basing on the observation that if isopropopylidene groups are generated, the C-CH 2° group is considerably more reactive than aryl one and should react by coupling to give C-CH 2 -CH 2 -C crosslinking bridges corresponding more to a H-crosslinking mechanism. As a conclusion, it seems that elevating the temperature promotes the crosslinking. However, the nature of this latter and presumably the role of temperature remain unclear.

Effect of atmosphere and sample thickness
Results in Fig. 7 (Sasuga & Hagiwara, 1987) clearly show that oxygen accelerates the degradation of aromatic polymers. The nature of the responsible process is suggested in a study on the degradation of PET films monitored by T g measurements (Burillo et al., 2007): a strong T g decrease is observed in air at the beginning and is certainly due to chain scission whereas a moderate T g decrease is observed for irradiation under vacuum. As proposed by Sasuga (Sasuga, 1988), polymer radiodegraded under anaerobic conditions would undergo mainly crosslinking meanwhile they would undergo chain scissions when they are degraded under aerobic (oxidative) conditions. The effect of atmosphere nature and the effect of thickness have the same origin linked to oxygen diffusion: thick samples present a diffusion limited oxidation (DLO), i.e. that their surface undergoes an oxidative degradation leading to chain scission meanwhile bulk undergoes an anaerobic radio-ageing generating radicals reacting only by coupling. This effect was quantified by comparing the degradation of thin and thick samples by means of GPC (for molar mass assessment and subsequently chain scission and crosslinking yields assessment) and T g (Richaud et al., 2011). T g decreases for both 2 mm and 200 µm thick samples, but more significantly for thin films (Fig. 8) which is not surprising since T g decreases with chain scissions and increases with crosslinking. Fig. 8. T g changes with dose for thin films () or thick samples (). www.intechopen.com

Case of thin films
Thin films undergo certainly a "pure" random chain scissions process. A graphical estimation (Fig. 5) gave for the thin samples (using k FF ~ 220 K kg mol -1 for PSU): Here, the simultaneous crosslinking compensates partially chain scissions consequences on T g changes. For ideal networks having no dangling chains, the effect of crosslinking on T g obeys the DiMarzio's law (DiMarzio,1964 -equation (18)) with: -T gl ~ 473 K -F PSU = 0.1105 kg mol -1 -k DM is the DiMarzio's constant: k DM is close to 1 - is the elastically active chains concentration. For ideal tetrafunctional networks:  = 2.x (26) x being the concentration of crosslinks events. Supposing that DiMarzio's equation can be applied for non-ideal networks, Eq. 23 turns into the Fox-Loshaek (Fox & Loshaek, 1955) equation for low conversion of the crosslinking process: with G(x) ~ 0.15. In other words, effect of sample thickness is explained by the effect of oxygen diffusion (which will be developed in the next section): irradiation generates radicals. In the absence of oxygen (i.e. for the core a bulk sample), these later react together by coupling which gives a crosslinking phenomenon. When oxygen is present (which is the case for thin samples or the superficial layer of a bulk sample), it orientates the degradation towards an oxidative mechanism with the formation of peroxy radicals POO°, and POOH of which decomposition induces chain scissions.

Effect of dose rate
This effect is illustrated by some results by Tavlet (Tavlet, 1997)  There are two effects to be distinguished: -An increase in temperature involved by enhancing the dose rate, -A physical and chemical effect linked to Diffusion Limited Oxidation, which will be described below. From a simplified theory of diffusion controlled oxidation (Audouin et al, 1994), a rough estimation of oxygen consumption rate r OX in films or superficial layers of thick samples can be done: where: -TOL is the thickness of oxidized layer (m), -D O2 is the oxygen diffusion coefficient in polymer amorphous phase (m 2 s -1 ), - [O 2 ] S is the equilibrium concentration under atmospheric pressure given by Henry's law: [O 2 ] = s O2 P O2 (31) s O2 being the solubility coefficient of oxygen in polymer amorphous phase. r OX can be expressed from a simplified mechanistic scheme for radio-thermal oxidation: It is thus demonstrated that radiothermal oxidation is characterized by a skin-core structure: superficial layers are oxidized meanwhile bulk not. Results by Tavlet (Tavlet, 1997) are thus explained as follows: for irradiation performed under a relatively elevated dose rate, only a very thin polymer layer is degraded and time for drop of elongation at break is longer than for irradiation performed under a moderate dose rate. This reasoning is confirmed by the comparison of a bulk sample (4 mm) for which no dose rate effect is observed undoubtedly because the thickness of degraded superficial layer is negligible compared to the total thickness of bulk material. Examples of Oxidized layers measured for radiochemical degradation of aromatic polymers (Richaud et al., 2010a) are presented in Fig. 10

On the radio-induced oxidizability of aromatic polymers
It seemed to us interesting to compare the k 3 2 /k 6 ratio estimated from radiodegradation of aromatic polymers with polypropylene or polyethylene ones. Let us recall that k 3 2 /k 6 describes the oxidizability of a polymer irrespectively of the oxidation mode (i.e. photo, thermo-or radio-induced). k 3 2 /k 6 can be calculated from Eq. (32) using: - [PH] = /M,  being the polymer density and M the molar mass of repetitive unit, r i = 10 -7 .G(P°).I from Heiland's results (Table 2), -r OX from TOL measured for example in FTIR mapping mode (Eq (30)). One can use with a good approximation the values of oxygen permeation of BisPhenol A Polycarbonate values for in PEEK and PSU (Van Krevelen & Te Nijenhuis, 2009) at 60°C : -D O2 = 8.110 -8 m² s -1 -s O2 = 410 -8 mol l -1 Pa -1 For samples exposed under 1.5 bar oxygen, TOL were found close to: -100 µm in PSU (Richaud et al., 2011), -70 µm in PEEK (Fig. 10).  Table 7. Kinetic parameters of PEEK and PSU estimated from DLO theory.
so that one can calculate (Table 6): -k 3 2 /k 6 = 1.310 -5 l mol -1 s -1 for PSU k 3 2 /k 6 = 1.810 -6 l mol -1 s -1 for PEEK These orders of magnitude are surprisingly on comparable or higher than values reported in PP (Richaud et al., 2006) and PE (Khelidj, 2006). k 3 can be conveniently estimated from the paper by Korcek (Korcek et al., 1972) using the relations: log k 3 s-POO° = 16.4 -0.2BDE(C-H) at 30°C www.intechopen.com log k 3 t-POO° = 15.4 -0.2BDE(C-H) at 30°C E 3 s-POO° = 0.55.(BDE(C-H) -65) E 3 t-POO° = 0.55.(BDE(C-H) -62.5) s-POO° and t-POO° representing respectively secondary and tertiary peroxy radicals. Given the difference in Bond Dissociation Energies for C-H (denoted by BDE(C-H)) in benzene (Davico et al., 1995) or in -CH 3 (Korcek et al., 1972), it is clear that: k 3 (PEEK) < k 3 (PSU) < k 3 (PE) < k 3 (PP) Thus, it suggests that (Table 7): k 6 (PEEK, PSU) << k 6 (PE, PP) From a chemical point of view, the reactivity of aryl or primary peroxy radicals is not low enough to justify such values (Table 6). A possible explanation is that the termination between two POO° radicals is a process involving two rare species. The mobility of segments favouring the coupling could thus control the kinetics of the reaction. The low order of magnitude below T g could be hence justified. Aromatic polymers could be interesting substrate for studying the Waite's theory: where: -r 0 is the cage radius, which corresponds to the maximal distance permitting the reaction: It would be on the order of r 0 ~ 1-10 nm (Emanuel & Buchachenko, 1990), -D is the diffusion coefficient of radical species. On the order of D ~ 10 -22 -10 -20 m 2 s -1 (Emanuel & Buchachenko, 1990), -N A is Avogadro's number. According to this theory, kinetic constant depends on the diffusivity of a radical and in other words of macromolecular mobility.

Conclusions
A comparison of yields for gas emission, gel d o s e o r c r i t i c a l d o s e f o r r e a c h i n g a n unacceptable level of mechanical properties shows that aromatic polymers radiochemical behaviour can be first explained by their aromatic character which is doubly favourable: -Yield for radical formation is very low compared with aliphatic polymers, because aromatic groups are able to dissipate a great part of the absorbed energy into reversible processes (fluorescence, phosphorescence...). - In the case of radio-oxidation, the propagation rate constant of hydrogen abstraction by peroxy radicals (k 3 ) is a decreasing function of the C-H bond dissociation energy, that takes a high value for aromatic hydrogen. These positive effects are partially counterbalanced by the relatively low termination rate linked to the low segmental mobility in glassy state. It is noteworthy that P° + P° termination (which leads to crosslink when oxygen is lacking) is more efficient than expected because P° radicals can diffuse independently of segmental mobility by the valence migration process (P 1° + P 2 H  P 1 H + P 2°) . This chapter also illustrates the effect of dose rate, thickness and oxygen concentration in polymers layers, which are linked by the theory of Diffusion Limited Oxidation: chain