GitBucket
Newer
zeynalig
<?xml version="1.0" ?> <tei> <teiHeader> <fileDesc xml:id="_Duguet_12279"/> </teiHeader> <text xml:lang="en"> <front> Toward a computational and experimental model of a poly-epoxy <lb/>surface <lb/> Thomas Duguet * , Camille Bessaguet, Maëlenn Aufray, Jérôme Esvan, Cédric Charvillat, <lb/>Constantin Vahlas, Corinne Lacaze-Dufaure <lb/> CIRIMAT, CNRS – Université de Toulouse, 4, allée Emile Monso, BP-44362, 31030 Toulouse Cedex 4, France <lb/>Keywords: <lb/> Epoxy <lb/>Amine <lb/>Surface science <lb/>AFM <lb/>XPS <lb/>DFT <lb/> a b s t r a c t <lb/> A model poly-epoxy surface formed by the reaction of DGEBA and EDA is studied by the combination <lb/>of experiments and DFT calculations. A special synthesis protocol is presented leading to the formation <lb/>of a surface that is smooth (S a < 1 nm), chemically homogeneous, and that presents a low-defect density <lb/>(0.21 m −2 ), as shown by AFM characterizations. Then, XPS is used for the determination of the elemental <lb/>and functional groups' surface composition. DFT allows the identification and assignment of individual <lb/>bonds contributions to the experimental 1s core-level peaks. Overall, we demonstrate that such a model <lb/>sample is perfectly suitable for a use as a template for the study of poly-epoxy surface functionalization. <lb/> </front> <body>1. Introduction <lb/> Poly-epoxy polymers are widely implemented in three families <lb/>of applications: adhesives, paints, and composite materials [1]. The <lb/>latters, such as epoxy/C fibers composites are increasingly found in <lb/>a wealth of devices and parts in the fields of leisure (skis, rackets, <lb/>boats, golf clubs, etc.), or transports, aeronautics and space (cars, <lb/>aircrafts, satellites, etc.), to name but a few. These composite mate-<lb/>rials possess stiffness and Young's modulus that compare well with <lb/>metallic alloys but with a much lower chemical reactivity and den-<lb/>sity. Therefore, they allow mass reduction and a large increase of <lb/>parts durability. <lb/>Replacement of metallic or ceramic parts by polymers often <lb/>requires surface functionalization in order to acquire optical, elec-<lb/>trical, magnetic, biomedical, esthetic, or chemical properties. The <lb/>main drawback when it comes to coat or to graft the surface of <lb/>polymer-based composites comes from the very low surface energy <lb/>of such materials once polymerized. This leads to a poor wett-<lb/>ability rendering painting or gluing difficult, and resulting in poor <lb/>adhesion. The surface energy of poly-ether ether ketone (PEEK) <lb/>or poly-epoxy is approximately 40–50 mJ/m 2 to be compared to <lb/> </body> <front>* Corresponding author. Tel.: +33 05 34 32 34 39. <lb/> E-mail address: thomas.duguet@ensiacet.fr (T. Duguet). <lb/></front> <body>approximately 500 mJ/m 2 for aluminum. Moreover, the polar com-<lb/>ponent (due to H bonding) is as low as 6–7 mJ/m 2 which inhibits <lb/>the use of simple functionalization protocols [2–4]. Hence, a large <lb/>number of particular protocols has been described or patented, <lb/>where the increase of reactivity and roughness is sought. A selec-<lb/>tion amongst the wealth of publications can be found in Refs. <lb/>[5–16]. <lb/>Such protocols or methods that have been used until now <lb/>remain empirical despite the resulting improvement of the targeted <lb/>properties and/or the extension of the durability of the mate-<lb/>rial. Therefore, the need exists to access the basic mechanisms <lb/>which control the surface functionalization of polymers and to con-<lb/>trol them so as to achieve satisfactory functional properties and <lb/>adhesion. By subscribing in this perspective, our approach aims <lb/>at describing the nucleation and growth of metallic thin films on <lb/>polymer surfaces, by using an integrated method where all the <lb/>elementary mechanisms are taken into account. The first step in <lb/>this frame – object of the present study – is to obtain a model <lb/>of the polymer surface, both experimental and theoretical, at the <lb/>atomic/molecular level. Such a model will serve as a template <lb/>for further surface treatments, including pretreatments, molecu-<lb/>lar grafting, or application of films and coatings. It is worth noting <lb/>that, to the authors' knowledge, no such a theoretical surface model <lb/>exists, most likely because of structural disorder and a lack of exper-<lb/>imental inputs. <lb/></body> <front> http://dx.doi.org/10.1016/j.apsusc.2014.10.096 <lb/></front> <body>Regarding our objectives, specifications of such an experimental <lb/>model polymer surface include: <lb/> • A 100% polymerization after curing to be comparable with calcu-<lb/>lations, where total polymerization is assumed. <lb/> • A low surface arithmetic roughness, namely R a < 1 nm to make <lb/>sure that we can observe nano-islands or nano-clusters of a given <lb/>thin film. Otherwise, they would be hindered by roughness. <lb/> • A very low defect density to avoid heterogeneous nucleation at <lb/>defects. <lb/> • Chemical homogeneity to make sure that calculation models <lb/>where homogeneity is assumed are representative of the tracked <lb/>chemical reactivity. Also to make sure that chemical composition <lb/>is independent on the analyzed surface area corresponding to a <lb/>given probe size. <lb/>Our experimental approach is based on the method described <lb/>in [17,18] for forming model poly-epoxy surfaces. It consists <lb/>in the polymerization of the poly-epoxy in an Ar gloves box <lb/>at ambient temperature for at least 24 h, followed by a post-<lb/>curing at elevated temperature (polymer-dependent). Gu et al. [17] <lb/>synthesize samples from a stoichiometric mixture of DGEBA + 1,3-<lb/>di(aminomethyl)-cyclohexane, with a small amount of toluene for <lb/>decreasing viscosity and favoring an homogeneous stirring (7 min). <lb/>Samples are then stored for 24 h at ambient temperature, and post-<lb/>cured for 2 h at 130 • C in an air furnace. Characterizations of the <lb/>free surfaces are performed by atomic force microscopy (AFM) in <lb/>Tapping ® mode. Surface roughness and phase contrast are deter-<lb/>mined. It is shown that samples synthesized in an Ar glove box show <lb/>a lower surface roughness than those prepared in ambient condi-<lb/>tions, and that they are homogeneous in composition. Kansow et <lb/>al. [18] use a similar method with the aim of characterizing the for-<lb/>mation of Al, Cu, Ag, and Au films by physical vapour deposition. <lb/>DGEBA reacts with diethylene triamine in low excess at 55 • C under <lb/>controlled atmosphere, before it is left for 48 h at ambient temper-<lb/>ature. At this step, polymerization rate is about 75%. Completion is <lb/>achieved by post curing for 1 h at 120 • C. Surface roughness is about <lb/>1 nm. <lb/>Theoretically, our greatest challenge is to circumvent the <lb/>description of the disordered/amorphous structure and to limit <lb/>the number of atoms. To that end, we start with a small macro-<lb/>molecule made from the reaction of bisphenol A diglycidyl ether <lb/>(DGEBA) with ethylenediamine (EDA) (61 atoms). Even for this <lb/>moderately complex system, the analysis of the experimental <lb/>core-level XPS spectrum is not trivial and can lead to incorrect <lb/>conclusions. The help of accurate theoretical tools is thus needed <lb/>and density-functional theory (DFT) is usually used for computing <lb/>XPS core-level shifts in the case of small organic or inorganic sys-<lb/>tems. The application of this theoretical method to large systems, <lb/>e.g. polymers, is a challenge but it is established that experimental <lb/>spectra are directly related to the electronic states obtained from <lb/>calculations on smaller model molecules. For instance, Endo et al. <lb/>presented a comprehensive analysis of the XPS C 1s spectra for poly-<lb/>mers using the negative of the energy of molecular orbitals [19,20]. <lb/>More recently, they used the 'transition state' theory [21] for the <lb/>calculation of the core electron binding energies [22,23]. Follow-<lb/>ing this work and in a first approach, we compute the molecular <lb/>orbitals energies on model molecules as preliminary input for the <lb/>assignment of experimental XPS spectra of the investigated poly-<lb/>mer. <lb/>We complement these results in the different DGEBA + EDA sys-<lb/>tem by implementing a more detailed description of surfaces by <lb/>AFM and XPS characterizations complemented by DFT calculations. <lb/>The paper is organized as follows. Experimental and computational <lb/>details are given in Section 2, followed by results in Section 3. Con-<lb/>clusions and perspectives are presented in Section 4. <lb/> 2. Experimental and computational details <lb/> 2.1. Synthesis <lb/> We use a stoichiometric mixture of DGEBA (DER 332, Dow <lb/>Chemicals, n = 0.03) and EDA (analytical grade, purity > 99.5%, <lb/>Sigma Aldrich). The mass of DGEBA (m DGEBA ) is fixed to 5 g. The <lb/>mass of EDA m EDA is thus determined following Eq. (1). <lb/> m DAE = <lb/> f DGEBA <lb/> f DAE <lb/> × <lb/> M DAE × m DGEBA <lb/> M DGEBA <lb/> = 0.43 g <lb/>(1) <lb/>where M DGEBA is the molar mass (348.52 g/mol) of this DGEBA <lb/>and f DGEBA is its functionality (2), and M EDA is the molar mass <lb/>(60.10 g/mol) and f DAE is the functionality (4) of the EDA. We assume <lb/>that no etherification occurs. <lb/>The mixture is then mechanically stirred (in an Ar glove box <lb/>when specified) for 7 min before it is poured into different molds <lb/>or deposited as a thin droplet on aluminum foil. Polymerization is <lb/>then allowed for 48 h at ambient temperature, followed by a post <lb/>curing of 2 h at 140 • C. For roughness comparison, we consider the <lb/>following poly-epoxy surfaces formed: <lb/>-At free surfaces, surfaces ref. either epoxy Air or epoxy Argon . <lb/>-At the interface with a 1 cm × 1 cm × 0.2 cm silicone mold, itself <lb/>molded on a Si wafer for transferring atomic flatness. Interfaces <lb/>ref. SiO Si /epoxy Air or SiO Si /epoxy Argon . <lb/>-At the interface with a 1 cm × 1 cm × 0.2 cm silicone mold, itself <lb/>molded on polystyrene (PS). Interfaces ref. SiO PS /epoxy Air or <lb/>SiO PS /epoxy Argon . <lb/>-By mechanical polishing up to a ¼ m with diamond paste. Sur-<lb/>faces ref. polished Air . <lb/>Interfaces formed in the same molds but in air or Ar show <lb/>different roughnesses (shown hereafter). This is the reason <lb/>why SiO Si /epoxy Air and SiO Si /epoxy Argon , and SiO PS /epoxy Air and <lb/>SiO PS /epoxy Argon are differentiated. <lb/> 2.2. Bulk characterizations <lb/> Differential scanning calorimetry (DSC) is used for the determi-<lb/>nation of the glass transition temperature (T g ) of the poly-epoxy <lb/>under investigation. We use a DSC 204 Phoenix Series (NETZSCH) <lb/>coupled with a TASC 414/4 controller. The apparatus is calibrated <lb/>against melting temperatures of In, Hg, Sn, Bi, and Zn, applying a <lb/>+10 • /min temperature ramp. Samples are placed in aluminum cap-<lb/>sules. Mass is measured with an accuracy of ±0.1 mg. We choose to <lb/>report the onset T g-onset temperature. <lb/>Fourier transform infrared spectroscopy, FTIR (Frontier, <lb/>PerkinElmer equipped with a NIR TGS detector), is performed in <lb/>transmission in the 4000–8000 cm −1 range. 16 scans are collected <lb/>for each analysis with a resolution of 4 cm −1 . We monitor the <lb/>characteristic epoxy band (combination band of the –CH 2 of <lb/>the epoxy group) at 4530 cm −1 with increasing polymerization <lb/>time, and after post curing treatment. The reference band is the <lb/>combination band of C C with aromatic CH at 4623 cm −1 [24]. <lb/>Peak areas are then used for calculating the conversion rate (Xe NIR ) <lb/>of epoxy groups, following Eq. (2). <lb/> Xe NIR = 1 − <lb/> <lb/> A epoxy /A reference <lb/> <lb/> t=t <lb/> <lb/> A epoxy /A reference <lb/> <lb/> t=0 <lb/> (2) <lb/>where A epoxy and A reference are the peak areas of the epoxy and <lb/>reference groups, respectively. <lb/> Fig. 1. Model dimer (1 DGEBA + 1 EDA). <lb/> 2.3. Surface characterizations <lb/> Surface roughness and viscoelastic homogeneity are deter-<lb/>mined by AFM (Agilent Technologies model 5500) in ambient <lb/>conditions. The former is performed in contact mode with tips of <lb/>spring constant k approx. 0.292 N/m, whereas the latter is per-<lb/>formed in Tapping ® mode with tips of k = 25–75 N/m (AppNano). <lb/>Scanning rate is 2 m/s. Images are processed with the softwares <lb/>Gwyddion version 2.19 [25] and Pico Image (Agilent Technologies). <lb/>Surface roughness parameters follow the Geometric Product Spec-<lb/>ifications ISO 25178. S a is the arithmetic roughness, S q is the root <lb/>mean square roughness, and S z is the total roughness (maximum <lb/>peak-to-valley), determined by processing the AFM images. <lb/>XPS analysis is performed using a Thermoelectron Kalpha appa-<lb/>ratus. Photoemission spectra are recorded using Al-K radiation <lb/>(h = 1486.6 eV) from a monochromatized source. The X-ray spot <lb/>diameter on the sample surface is 400 m. The pass energy is <lb/>fixed at 30 eV for narrow scan and 170 eV for survey scans. The <lb/>spectrometer energy calibration was performed using the Au 4f 7/2 <lb/> (83.9 ± 0.1 eV) and Cu 2p 3/2 (932.8 ± 0.1 eV) photoelectron lines. <lb/>The background signal is removed using the Shirley method. Atomic <lb/>concentrations are determined from photoelectron peak areas <lb/>using the atomic sensitivity factors reported by Scofield [26] and <lb/>taking into account the transmission function of the analyzer. This <lb/>function was determined at different pass energies from Ag 3d and <lb/>Ag MNN peaks collected on a silver standard. Finally, photoelectron <lb/>peaks are analyzed and deconvoluted using a Lorentzian/Gaussian <lb/>(L/G = 30) peak fitting. <lb/> 2.4. Calculations <lb/> We used the model molecule shown in Fig. 1 that results of the <lb/>addition of one DGEBA and one EDA molecule. <lb/>The geometry of the model molecule was optimized at the <lb/>B3LYP/6-31G* level of theory using the Gaussian 03 software <lb/>package [27]. <lb/> 3. Results <lb/> Bulk characterizations are performed on samples polymerized <lb/>under ambient conditions. DSC is used for the determination of <lb/> T g-onset . Temperature ramps are doubled for each sample in order <lb/>to ensure that there is no physical aging and to verify that polymer-<lb/>ization is complete. For all samples T g-onset = 113 ± 1 • C. We assume <lb/>that T g-onset is not different after polymerization in the Ar glove box <lb/>(no bulk characterization for these samples). <lb/>We then monitor the polymerization rate with reaction duration <lb/>by following the gradual decrease of the epoxy peak area by FTIR, <lb/>and calculating the conversion rate using Eq. (2). Results are shown <lb/>in Fig. 2. <lb/>Experiments are performed from 15 min to 11520 min (8 days) <lb/>after mixing of the reactants. The conversion rate increases slowly <lb/>in the first hours and reaches an asymptote between 24 and <lb/>48 h. The maximum conversion at ambient temperature is 84% for <lb/> t ≥ 48 h. The only mean for achieving a complete polymerization <lb/>is to set the sample at a temperature above its glass transi-<lb/>tion. The post curing treatment (140 • C, 2 h) leads to a complete <lb/> Fig. 2. Epoxy group conversion rate as a function of polymerization over an 8-day <lb/>period of time. Dashed line indicates that polymerization is complete after post <lb/>curing at 140 • C for 2 h. <lb/> polymerization (>98%, taking into account the FTIR spectrometer <lb/>sensitivity) illustrated by the dashed line in Fig. 2. <lb/>The different surfaces that we consider are then characterized by <lb/>AFM over 3 m × 3 m surface area images in order to determine <lb/>roughness parameters. Results are summarized in Fig. 3. <lb/>Roughness of the free surfaces is reduced by three orders of <lb/>magnitude when polymerization is performed in the Ar glove box. <lb/>Under Ar, S a and S q do not exceed 1.5 nm, except for sample <lb/>SiO Si /epoxy Argon , for which these two values are 4.9 nm and 6.8 nm, <lb/>respectively. The latter is not acceptable for the AFM observation <lb/>of metallic nanoislands or clusters that we target, in the range <lb/>of 1–20 nm in diameter [18]. In order to transfer atomic flatness <lb/>to the molds, and then to the SiO Si /epoxy surfaces, we mold sili-<lb/>cone molds against Si wafer or against PS. In these conditions, the <lb/>lowest roughness is again obtained when the surfaces are formed <lb/>under Ar atmosphere, and is similar between Si and PS processes. <lb/>Somehow, atmosphere also plays a role regarding roughness at the <lb/>substrate/polymer interface. However, a roughness as low as at that <lb/>of the free epoxy Argon surface is not achieved, indicating that mold-<lb/>ing in these conditions is not well suited for our purpose. Finally, the <lb/>roughness parameters of the polished surface are quite low but AFM <lb/>images show many scratches where nucleation may preferentially <lb/>occur. Since we want to avoid heterogeneous nucleation in order <lb/>to compare nucleation with adsorption energies at the molecular <lb/>level, polishing is abandoned. <lb/> Fig. 3. Roughness parameters determined by image processing on 3 m × 3 m sur-<lb/>faces characterized by AFM in contact mode. A polynomial of degree 2 is used in <lb/>order to correct image curvature. <lb/> Fig. 4. AFM images of the epoxy Air (a and b) and epoxy Argon (c and d) surfaces. Left column shows topographic images after a polynomial of degree 2 correction, and right <lb/>column shows deflection images (or phase contrast). <lb/> Fig. 4 shows a selection of AFM images of the epoxy Air (a and <lb/>b) and epoxy Argon (c and d) surfaces, obtained in Tapping ® mode. <lb/>Right column (Fig. 4a and c) corresponds to the surface topography <lb/>and left column (Fig. 4b and d) to the deflection of the cantilever, i.e. <lb/>to the phase contrast. Whereas Tapping ® mode leads to different <lb/>apparent values of roughness compared to contact mode, rough-<lb/>ness is again lower on the epoxy Argon surface, as can be noticed on <lb/>the contrast scale, on the right-hand side of the images. However, <lb/>both surfaces are quite flat and exhibit a very low phase contrast. <lb/>The measure of phase contrast probes the local viscoelastic prop-<lb/>erties that we assume to be an indication of chemical homogeneity <lb/>in the nanometer range. Finally, Fig. 4c and d is chosen on purpose <lb/>in order to illustrate the presence of defects, in the form of approx. <lb/>50 nm-in-diameter troughs. The density shown in Fig. 4 is not rep-<lb/>resentative (overestimated). A thorough count over a total 90 m 2 <lb/> surface area gives a defect density equal to 0.21 m −2 . <lb/>Epoxy Argon is selected as the best candidate for an experimen-<lb/>tal model surface of poly-epoxy. Thus, we investigate its surface <lb/>chemical composition by XPS and use the output of DFT calcula-<lb/>tions for peak identifications and binding energy assignments. A <lb/>first observation is made on free surfaces of samples synthesized in <lb/>silicone molds (i.e. that were not in contact with the mold). Survey <lb/>spectra show a strong Si 2p contribution at 101.8 ± 0.1 eV, which is <lb/>characteristic of siloxane groups [28]. It represents a large amount <lb/>of adsorbed silicone on the surface (approx. 8 at.%). Consequently, <lb/>epoxy Ar samples are now synthesized on Al foil (and silicone is <lb/>banished from the glove box). The significant thickness of the poly-<lb/>epoxy coupons (1 mm) ensures that Al does not diffuse up to the <lb/>free surface, since the measured interphases do not exceed 300 m <lb/> [29]. <lb/>The XPS survey spectrum of epoxy Argon surfaces polymerized <lb/>on aluminum foil show neither Si nor other elements than the one <lb/>expected in the polymer or from adsorbed molecules from the air. <lb/>Atomic composition of the surface is determined by 1s peaks fitting, <lb/>repeated at different x–y coordinates on the sample surface. We <lb/>determine the following surface composition: <lb/>81.5 at.% C, 1.8 at.% N, and 16.7 at.% O <lb/> The result is slightly different from the bulk composition of <lb/>the poly-epoxy, where the basic motif is made of 2 DGEBA <lb/>(2 × 21 C + 2 × 4 O atoms) molecules for 1 EDA (2 C + 2 N atoms) <lb/>molecule, resulting in a bulk composition of:81.5 at . % C, 3.7 at . % N, <lb/>and 14.8 at . % O. Whereas the composition of the surface shows a <lb/>similar carbon content, it is richer in oxygen and poorer in nitro-<lb/>gen than the bulk. This is an indication of a mild surface oxidation <lb/>that may occur in the course of post-curing, when polymerization <lb/>is not yet complete (post-curing starts at 85% polymerization rate). <lb/>It is questioning though that the carbon content is apparently not <lb/>affected as well. <lb/>In order to further investigate the surface chemistry of the model <lb/>poly-epoxy surface, molecular orbitals extracted directly from DFT <lb/>results are studied. Table 1 shows the binding energies of 1s elec-<lb/>trons involved in the different bonds of the model dimer. The dimer <lb/>is made of 1 DGEBA and 1 EDA that virtually bonded through 1 <lb/>epoxy/1 amine proton reaction. Therefore, there are a few discrep-<lb/>ancies between the experimental fully-polymerized samples and <lb/>the model dimer. They are enlightened by the gray coloring of the <lb/>lines corresponding to secondary and primary amines (all should be <lb/>ternary) and to the epoxy group (no more epoxy rings in the 100% <lb/>polymerized sample). The binding energies shown are the nega-<lb/>tive value of the molecular orbitals energies. Therefore, absolute <lb/>values are not correct because (i) XPS binding energies correspond <lb/>to a multi-step process where photoelectrons interact with the cre-<lb/>ated holes, with the matrix and with their image before and after <lb/>extraction into vacuum, (ii) temperature is not considered, (iii) of <lb/>the limitation of Kohn–Sham orbital energies as reflecting initial <lb/>state effects [30]. Nevertheless, chemical shifts can be used if one <lb/>consider the latter processes constant in a given energy domain. <lb/>A minimum mean chemical shift of 0.2 eV is technically observ-<lb/>able with our XPS apparatus. Therefore, we discriminate phenyl <lb/>groups from CH 3 groups, and C OH & part of the C O C bonds <lb/>from the other C O C bonds. Thanks to the support of DFT results, <lb/>we use 5 contributions to the C 1s peak deconvolution and 2 con-<lb/>tributions to the O 1s peak deconvolution. The fine fitting of the C <lb/>1s and O 1s spectra are shown in Fig. 5. N 1s spectrum is not shown <lb/>because it exhibits only one contribution for C N bonds centered <lb/> Table 1 <lb/> Molecular orbitals involving O, N, and C 1s atomic orbitals from DFT calculations on the model DGEBA–EDA dimer. Corresponding electronic binding energies ((−1) × orbital <lb/>energy), and mean chemical shifts for the given bond. Grayed cells do not have a counterpart in the experimental fully-reticulated poly-epoxy. <lb/>Molecular orbital <lb/>Binding energy (Hartree) <lb/>Binding energy (eV) <lb/>Mean chemical shift (±0.1 eV) <lb/>Bond <lb/>O 1s <lb/> −19.177 <lb/> 521.8 <lb/>+0.8 <lb/>C O C <lb/> −19.170 <lb/> 521.6 <lb/>C O C <lb/> −19.165 <lb/> 521.5 <lb/>+0.6 <lb/>Epoxy <lb/> −19.145 <lb/> 520.9 <lb/>Ref. <lb/>O H <lb/>N 1s <lb/> −14.324 <lb/> 389.8 <lb/>+0.3 <lb/>Secondary amine <lb/> −14.316 <lb/> 389.5 <lb/>Ref. <lb/>Primary amine <lb/>C 1s <lb/> −10.249 <lb/> 278.9 <lb/>+2.0 <lb/>C O C <lb/> −10.249 <lb/> 278.9 <lb/>C O C <lb/> −10.246 <lb/> 278.8 <lb/>C O C <lb/> −10.244 <lb/> 278.7 <lb/>C O C <lb/> −10.239 <lb/> 278.6 <lb/>+1.8 <lb/>C O C <lb/> −10.239 <lb/> 278.6 <lb/>C O C <lb/> −10.238 <lb/> 278.6 <lb/>C OH <lb/> −10.212 <lb/> 277.9 <lb/>+1.0 <lb/>C N <lb/> −10.209 <lb/> 277.8 <lb/>C N <lb/> −10.207 <lb/> 277.7 <lb/>C N <lb/> −10.205 <lb/> 277.7 <lb/>Quaternary C C <lb/> −10.186 <lb/> 277.2 <lb/>+0.2 <lb/>Phenyl <lb/> −10.185 <lb/> 277.1 <lb/>Phenyl <lb/> −10.185 <lb/> 277.1 <lb/>Phenyl <lb/> −10.183 <lb/> 277.1 <lb/>Phenyl <lb/> −10.182 <lb/> 277.1 <lb/>Phenyl <lb/> −10.181 <lb/> 277.0 <lb/>Phenyl <lb/> −10.181 <lb/> 277.0 <lb/>Phenyl <lb/> −10.181 <lb/> 277.0 <lb/>Phenyl <lb/> −10.180 <lb/> 277.0 <lb/>Phenyl <lb/> −10.176 <lb/> 276.9 <lb/>Phenyl <lb/> −10.173 <lb/> 276.8 <lb/>0.0 <lb/>CH3 <lb/> −10.173 <lb/> 276.8 <lb/>Ref. <lb/>CH3 <lb/> Fig. 5. XPS fine spectra of C 1s and O 1s. Spectra are fitted with contributions derived <lb/>from DFT calculations on the model dimer. <lb/> at 399.2 eV. Binding energy scale of the C 1s spectrum starts with <lb/>the –CH 3 contribution fixed at 284.4 eV. Then, mean chemical shifts <lb/>extracted from the energy difference between molecular orbitals of <lb/>DFT (see Table 1) are used for higher-binding–energy contributions <lb/>(284.4 + 0.2, +1.0, +1.8, +2.0 eV). <lb/>The filled area shows the envelope of the fitting curve. There is <lb/>an excellent matching with both O and C 1s experimental spec-<lb/>tra. Again, calculations ensure that contributions are real; even <lb/>C N, for instance, which is buried in the tails of neighboring con-<lb/>tributions. In order to consolidate these results, we now discuss <lb/>fitting with regards to the functional group composition shown in <lb/>Table 2. <lb/>Experimental atomic compositions in functional groups are con-<lb/>sistent. For instance, where one O 1s orbital of the C O C bonds <lb/>shows a composition of 15.5 at.%, two C 1s orbitals of the C O C <lb/>bonds show an approximately doubled composition of 30.5 at.% <lb/>(28.0 plus the contribution of C O C at 286.2 eV of about 3.7 <lb/>(C 1s C O C, C OH) – 1.2 (O 1s C OH) = 2.5 at.%). Similarly, N <lb/>1s and C 1s compare well in terms of composition in the C N <lb/>bonds (1.8 vs. 1.3 at.%). Finally, the last column of Table 2 shows <lb/>the expected composition in functional groups in a poly-epoxy <lb/>where the DGEBA:EDA ratio equals 2:1. For instance the number <lb/>of C 1s in phenyl groups is calculated as follows: 2 DGEBA × 2 <lb/>phenyls/DGEBA × 6 C atoms = 24 C 1s. Overall, one can find 4 C 1s <lb/>in –CH 3 , 24 C 1s in phenyls, 2 C 1s in C N, 4 C 1s in C OH, 4 C <lb/>1s in C O C 286.2 eV , 4 C 1s in C O C 286.4 eV , 2 N 1s in C N, 4 O <lb/>1s in C OH, and 4 O 1s in C O C. Therefore the total number of <lb/>considered 1s orbitals is 52. We observe large discrepancies con-<lb/>cerning the phenyl bonds concentration and the oxygenated bonds <lb/>C OH and C O C concentrations, a difference that was already <lb/>mentioned when considering the elemental atomic composition. <lb/>There are two possibilities for explaining these differences: either <lb/>the surface is oxidized and oxygenated bonds contribute to the C <lb/>1s and O 1s signals at neighboring binding energies, or the polymer <lb/>is oriented in such a way that C O C bonds emerge at the surface. <lb/> Table 2 <lb/> Results of the C, O, and N 1s deconvolutions; peak binding energy: BE, chemical shift imposed after DFT results; height in counts per second: CPS; full width at half-maximum: <lb/>FWHM; peak area; scofield relative sensitivity factor: RSF; atomic fraction: at.%; and the composition expected from the model polymer with a DGEBA:EDA ratio of 2:1. <lb/>Name <lb/>Peak BE (eV) <lb/>Chemical shift (eV) <lb/>Height (CPS) <lb/>FWHM (eV) <lb/>Area (CPS eV) <lb/>Scofield RSF <lb/>At.% <lb/>2 DGEBA:EDA motif (at.%) <lb/>C 1s CH3 <lb/>284.4 <lb/>0.0 <lb/>2249.74 <lb/>1.06 <lb/>2578.62 <lb/>1 <lb/>6.5 <lb/>7.7 <lb/>C 1s phenyl <lb/>284.6 <lb/>0.2 <lb/>11580.87 <lb/>1.3 <lb/>16279.01 <lb/>1 <lb/>40.8 <lb/>46.2 <lb/>C 1s C N <lb/>285.4 <lb/>1.0 <lb/>328.14 <lb/>1.41 <lb/>500.77 <lb/>1 <lb/>1.3 <lb/>3.8 <lb/>C 1s C O C, C OH <lb/>286.2 <lb/>1.8 <lb/>1009.93 <lb/>1.36 <lb/>1488.52 <lb/>1 <lb/>3.7 <lb/>C OH:7.7 + C O C:7.7 <lb/>C 1s C O C <lb/>286.4 <lb/>2.0 <lb/>7147.92 <lb/>1.44 <lb/>11172.06 <lb/>1 <lb/>28.0 <lb/>7.7 <lb/>C 1s shake up <lb/>291.2 <lb/>n/a <lb/>349.91 <lb/>1.32 <lb/>501.28 <lb/>1 <lb/>1.3 <lb/>n/a <lb/>N 1s C N <lb/>399.2 <lb/>n/a <lb/>808.92 <lb/>1.28 <lb/>1195.36 <lb/>1.8 <lb/>1.8 <lb/>3.8 <lb/>O 1s C OH <lb/>532.0 <lb/>0.0 <lb/>669.53 <lb/>1.73 <lb/>1255.76 <lb/>2.93 <lb/>1.2 <lb/>7.7 <lb/>O 1s C O C <lb/>532.9 <lb/>0.9 <lb/>9494.3 <lb/>1.53 <lb/>15728.75 <lb/>2.93 <lb/>15.5 <lb/>7.7 <lb/> If we assume that a mild oxidation occurred in the course of sam-<lb/>ple preparation, it may be assigned to sub-stoichiometric groups, <lb/>such as amines (1.3–1.8 vs. 3.8 at.% expected) and phenyls (40.8 vs. <lb/>46.2 at.% expected). In that case deconvolution may be improved <lb/>by substituting or implementing additional contributions that we <lb/>are not able to identify now. <lb/> 4. Conclusions <lb/> We selected an epoxy-amine system which permits its use as <lb/>both an experimental and a computational template for further sur-<lb/>face treatments. DGEBA and EDA mixed in stoichiometric ratio and <lb/>slowly polymerized (48 h) in an Ar glovebox lead to the formation <lb/>of a poly-epoxy polymerized at a rate of 85%. Total polymeriza-<lb/>tion is achieved by post-curing at 120 • C for 2 h. Such a poly-epoxy <lb/>exhibits a glass transition temperature onset of 113 ± 1 • C. Dif-<lb/>ferent substrates and atmospheres were tested and compared <lb/>in terms of surface roughness. The lowest roughness (arithmetic <lb/>roughness = 0.2 nm, peak-to-valley = 1.5 nm) is obtained at the free <lb/>surface that polymerized under Ar atmosphere. AFM observations <lb/>reveal that, in addition to the high smoothness, the defect density of <lb/>the surface is low enough to avoid defect driven undesirable nuclea-<lb/>tion. Additionally, phase contrast is almost null which indicates that <lb/>the surface is chemically homogeneous. Atomic compositions from <lb/>XPS survey spectra at different positions confirm this result. Fine <lb/>XPS spectra over C, O, and N 1s core levels are analyzed in view <lb/>of the DFT calculations results. Theoretical binding energy chemi-<lb/>cal shifts allow an excellent fitting of the experimental 1s spectra. <lb/>A limitation has been emphasized concerning the compositions in <lb/>chemical groups: the main discrepancy concerning a much larger <lb/>composition in C O C than the one theoretically expected from <lb/>the perfect polymer model. In a near future, we will dedicate our <lb/>efforts to the improvement of (i) the poly-epoxy network model by <lb/>allowing a larger number of atoms and by using molecular dynam-<lb/>ics computations to freeze the structure at given temperatures, and <lb/>(ii) of the core-level binding energies calculations using the gener-<lb/>alized transition state method [21] that allows a better treatment of <lb/>the XPS photoemission process. Finally, the perspectives for exper-<lb/>imental work will be the formation of thin metallic films and the <lb/>mechanistic description of nucleation and growth. <lb/></body> <listBibl> References <lb/> [1] J.-P. Pascault, H. Sautereau, J. Verdu, R.J.J. Williams, Thermosetting Polymers, <lb/>CRC Press, New York, 2002. <lb/>[2] J.W. Gooch, Encyclopedic Dictionary of Polymers, Springer, New York, 2010. <lb/>[3] E.M. Petrie, Epoxy Adhesives Formulation, McGraw-Hill, New York, 2006. <lb/>[4] E.M. Petrie, Handbook of Adhesives and Sealants, McGraw-Hill, New York, 2007. <lb/>[5] M. Charbonnier, M. Romand, Polymer pretreatments for enhanced adhesion <lb/>of metals deposited by the electroless process, Int. J. Adhes. Adhes. 23 (2003) <lb/>277–285. <lb/>[6] T.H. Baum, D.C. Miller, T.R. O'Toole, Photoselective catalysis of electroless cop-<lb/>per solutions for the formation of adherent copper films onto polyimide, Chem. <lb/>Mater. 3 (1991) 714–720. <lb/>[7] S. Yoon, H.-J. Choi, J.-K. Yang, H.-H. Park, Comparative study between poly(p-<lb/>phenylenevinylene) (PPV) and PPV/SiO2 nano-composite for interface with <lb/>aluminum electrode, Appl. Surf. Sci. 237 (2004) 451–456. <lb/>[8] T.-G. Woo, I.-S. Park, K.-W. Seol, Effects of various metal seed layers on the <lb/>surface morphology and structural composition of the electroplated copper <lb/>layer, Met. Mater. Int. 15 (2009) 293–297. <lb/>[9] T. Duguet, F. Senocq, L. Laffont, C. Vahlas, Metallization of polymer composites <lb/>by metalorganic chemical vapor deposition of Cu: surface functionalization <lb/>driven films characteristics, Surf. Coat. Technol. 230 (2013) 254–259. <lb/>[10] J. Ge, M.P.K. Turunen, J.K. Kivilahti, Surface modification and characterization <lb/>of photodefinable epoxy/copper systems, Thin Solid Films 440 (2003) 198–207. <lb/>[11] C. Weiss, H. Muenstedt, Surface modification of polyether ether ketone (peek) <lb/>films for flexible printed circuit boards, J. Adhes. 78 (2002) 507–519. <lb/>[12] L. Di, B. Liu, J. Song, D. Shan, D.-A. Yang, Effect of chemical etching on the Cu/Ni <lb/>metallization of poly (ether ether ketone)/carbon fiber composites, Appl. Surf. <lb/>Sci. 257 (2011) 4272–4277. <lb/>[13] S. Siau, A. Vervaet, E. Schacht, A. Van Calster, Influence of chemical pre-<lb/>treatment of epoxy polymers on the adhesion strength of electrochemically <lb/>deposited cu for use in electronic interconnections, J. Electrochem. Soc. 151 <lb/>(2004) C133–C141. <lb/>[14] P.B. Messersmith, J. Dalsin, L. Lin, B.P. Lee, H. K., Polymeric compositions and <lb/>related method of use, WO 2005/118831 A2, Northwestern University (USA), <lb/>2005. <lb/>[15] M. Kemell, E. Färm, M. Ritala, M. Leskelä, Surface modification of thermoplastics <lb/>by atomic layer deposition of Al2O3 and TiO2 thin films, Eur. Polym. J. 44 (2008) <lb/>3564–3570. <lb/>[16] G. Cui, D. Wu, Y. Zhao, W. Liu, Z. Wu, Formation of conductive and reflective <lb/>silver nanolayers on plastic films via ion doping and solid–liquid interfacial <lb/>reduction at ambient temperature, Acta Mater. 61 (2013) 4080–4090. <lb/>[17] X. Gu, T. Nguyen, M. Oudina, D. Martin, B. Kidah, J. Jasmin, A. Rezig, L. Sung, <lb/>E. Byrd, J. Martin, D. Ho, Y.C. Jean, Microstructure and morphology of amine-<lb/>cured epoxy coatings before and after outdoor exposures—an AFM study, J. <lb/>Coat. Technol. Res. 2 (2005) 547–556. <lb/>[18] J. Kanzow, P.S. Horn, M. Kirschmann, V. Zaporojtchenko, K. Dolgner, F. Faupel, <lb/>C. Wehlack, W. Possart, Formation of a metal/epoxy resin interface, Appl. Surf. <lb/>Sci. 239 (2005) 227–236. <lb/>[19] K. Endo, S. Maeda, M. Aida, Simulation of C 1s spectra of C-and O-containing <lb/>polymers in XPS by ab initio MO calculations using model oligomers, Polym. J. <lb/>29 (1997) 171–181. <lb/>[20] K. Endo, S. Maeda, Y. Kaneda, Analysis of C 1s spectra of N-, O-, and X-containing <lb/>polymers in X-ray photoelectron spectroscopy by ab initio molecular orbital <lb/>calculations using model molecules, Polym. J. 29 (1997) 255–260. <lb/>[21] J.C. Slater, K.H. Johnson, Self-consistent-field X˛ cluster method for polyatomic <lb/>molecules and solids, Phys. Rev. B 5 (1972) 844–853. <lb/>[22] K. Takaoka, T. Otsuka, K. Naka, A. Niwa, T. Suzuki, C. Bureau, S. Maeda, K. Hyodo, <lb/>K. Endo, D.P. Chong, Analysis of X-ray photoelectron spectra of electrochemi-<lb/>cally prepared polyaniline by DFT calculations using model molecules, J. Mol. <lb/>Struct. 608 (2002) 175–182. <lb/>[23] T. Otsuka, K. Endo, M. Suhara, D.P. Chong, Theoretical X-ray photoelectron spec-<lb/>tra of polymers by deMon DFT calculations using the model dimers, J. Mol. <lb/>Struct. 522 (2000) 47–60. <lb/>[24] C.E. Miller, Near-infrared spectroscopy of synthetic polymers, Appl. Spectrosc. <lb/>Rev. 26 (1991) 277–339. <lb/>[25] D. Nečas, P. Klapetek, Gwyddion: an open-source software for SPM data anal-<lb/>ysis, Centr. Eur. J. Phys. 10 (2012) 181–188. <lb/>[26] J.H. Scofield, Hartree-Slater subshell photoionization cross-sections at 1254 <lb/>and 1487 eV, J. Electron Spectrosc. Relat. Phenom. 8 (1976) 129–137. <lb/>[27] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, <lb/>J.A. Montgomery, J.T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. <lb/>Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, <lb/>H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. <lb/>Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, <lb/>J.B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, J. <lb/>Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. <lb/>Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, <lb/>M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, <lb/>J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. <lb/>Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, <lb/>C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, <lb/> M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian 03, Revision B05, Gaussian, Inc., <lb/>Pittsburg, PA, 2003. <lb/>[28] L.-A. O'Hare, B. Parbhoo, S.R. Leadley, Development of a methodology for XPS <lb/>curve-fitting of the Si 2p core level of siloxane materials, Surf. Interface Anal. <lb/>36 (2004) 1427–1434. <lb/>[29] M. Aufray, A. André Roche, Epoxy–amine/metal interphases: influences <lb/>from sharp needle-like crystal formation, Int. J. Adhes. Adhes. 27 (2007) <lb/>387–393. <lb/>[30] P.S. Bagus, E.S. Ilton, C.J. Nelin, The interpretation of XPS spectra: insights into <lb/>materials properties, Surf. Sci. Rep. 68 (2013) 273–304. </listBibl> </text> </tei>