The influence of nitrogen and oxygen additions on the thermal characteristics of aluminium-based thin films

Materials Chemistry and Physics 163 (2015) 569e580

Contents lists available at ScienceDirect

Materials Chemistry and Physics journal homepage: www.elsevier.com/locate/matchemphys

The influence of nitrogen and oxygen additions on the thermal characteristics of aluminium-based thin films J. Borges a, b, *, F. Macedo a, F.M. Couto c, M.S. Rodrigues a, d, C. Lopes a, d, P. Pedrosa a, e, f, T. Polcar b, g, L. Marques a, F. Vaz a a

Centro de Física, Universidade do Minho, Campus de Gualtar, 4710-057 Braga, Portugal
2, Prague 6, Czech Republic Department of Control Engineering, Faculty of Electrical Engineering, Czech Technical University in Prague, Technicka Physics Sciences Laboratory, Norte Fluminense State University, 28013-602 CamposeRJ, Brazil d
rio de Ensaios, Desgaste e Materiais, Rua Pedro Nunes, 3030-199 Coimbra, Portugal Instituto Pedro Nunes, Laborato e SEG-CEMUC, Mechanical Engineering Department, University of Coimbra, 3030-788 Coimbra, Portugal f Universidade do Porto, Faculdade de Engenharia, Departamento de Engenharia Metalúrgica e de Materiais, Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal g Engineering Materials & nCATS, FEE, University of Southampton, Highfield Campus, SO17 1BJ, Southampton, UK b c

h i g h l i g h t s
AlNx, AlOy and AlNxOy films were deposited by magnetron sputtering.
Discharge characteristics were compared between systems.
Different x and y coefficients were obtained.
Composition, structure and morphology were correlated with physical properties.
Thermal behaviour was studied using modulated IR radiometry.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 28 October 2014 Received in revised form 13 April 2015 Accepted 9 August 2015 Available online 19 August 2015

The ternary aluminium oxynitride (AlNxOy) system offers the possibility to obtain a wide range of properties by tailoring the ratio between pure Al, AlNx and AlOy and therefore opening a significant number of possible applications. In this work the thermal behaviour of AlNxOy thin films was analysed by modulated infrared radiometry (MIRR), taking as reference the binary AlOy and AlNx systems. MIRR is a non-contact and non-destructive thermal wave measurement technique based on the excitation, propagation and detection of temperature oscillations of very small amplitudes. The intended change of the partial pressure of the reactive gas (N2 and/or O2) influenced the target condition and hence the deposition characteristics which, altogether, affected the composition and microstructure of the films. Based on the MIRR measurements and their qualitative and quantitative interpretation, some correlations between the thermal transport properties of the films and their chemical/physical properties have been found. Furthermore, the potential of such technique applied in this oxynitride system, which present a wide range of different physical responses, is also discussed. The experimental results obtained are consistent with those reported in previous works and show a high potential to fulfil the demands needed for the possible applications of the systems studied. They are clearly indicative of an adequate thermal response if this particular thin film system is aimed to be applied in small sensor devices or in electrodes for biosignal acquisition, such as those for electroencephalography or electromyography as it is the case of the main research area that is being developed in the group. © 2015 Elsevier B.V. All rights reserved.

Keywords: Sputtering Thin films Nitrides Oxides Microstructure Thermal properties

1. Introduction * Corresponding author. Centro de Física, Universidade do Minho, Campus de Gualtar, 4710-057 Braga, Portugal. E-mail address: joelborges@fisica.uminho.pt (J. Borges). http://dx.doi.org/10.1016/j.matchemphys.2015.08.015 0254-0584/© 2015 Elsevier B.V. All rights reserved.

Thin films based on metal nitrides and oxides are established materials with great interest to the academic communities due to

570

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

their numerous industrial applications and still with a promising future [1,2]. The industrial importance of these materials keeps growing rapidly, not only in the well-established applications based on the strength and refractory nature of these materials such as cutting tools and abrasives, but also in new and promising fields such as electronics, optoelectronics and medical device applications [3e9]. More recently and using the idea of tailoring the film's properties between those of metal nitrides, MeNx, and those of the correspondent insulating oxides, MeOy, a new class of materials gained more importance in several technological applications: the metal oxynitrides MeNxOy, where Me can be Al [10e13], Cr [14], Ta [15], Nb [16], Zr [17] [18], Mo [19], Ti [20,21], among others. These ternary materials allow, in principle, merging the benefits of the basic characteristics of both metal nitrides and oxides. Their relevance arises from the fact that the addition of oxygen and nitrogen to metallic elements allows a vast number of stoichiometries, opening the possibility of tuning the band gap, the electrical conductivity, and the crystallographic order between nitride and oxide and hence the electronic and (micro)structural properties of the materials, and thus the overall set of properties [2]. An example of such systems is the ternary aluminium oxynitride (AlNxOy), which might combine the advantages of metallic aluminium with those of the correspondent binary systems, AlNx and AlOy. The wide difference between the properties of these three base materials opens the possibility to combine some of their advantages by simply changing the CNþO/CAl atomic ratio of the film. In previous works the authors carried out a systematic study of the AlNxOy system, taking as reference the corresponding AlNx and AlOy base binary systems. It was showed that by using different pressures of reactive gas (N2 and/or O2), in the magnetron sputtering deposition process [12], it was possible to tune the films composition with different structural and morphological characteristics. These features induced a wide range of electrical [22], optical [11] and electrochemical [23] properties, making this films potentially useful for biomedical sensors or solar applications [23]. Nevertheless, in order to scan the possibility of using this type of films in or near heat sources, such as solar power systems [24], a detailed knowledge about the thermal transport properties and their correlation with the deposition conditions and resulting microstructural features is fundamental for the successful development of such materials. In order to study the thermal transport properties of the films, modulated IR radiometry (MIRR), also called modulated photothermal radiometry, is ordinarily applied, since it's a nondestructive and contactless thermal wave method appropriate for the characterization of the thickness and the thermal properties of thin films [25,26]. In terms of thermal behaviour, this technique can be useful to determine the thermal diffusivity and the thermal effusivity of the materials. The thermal diffusivity reflects the capacity of the materials to spread out the thermal energy and thus it is a very important parameter since a very low value can lead to a high localized heating, eventually damaging the sample. Thermal effusivity can be seen as a measure of thermal inertia and it is crucial in controlling heat propagation between different media. In particular, the thermal effusivity ratio between the film and the substrate is a fundamental parameter to understand the way heat propagates in layered materials. Furthermore, if the effusivity of the substrate is known, the effusivity of the film material can be calculated. In this work, the thermal parameters, such as the thermal effusivity ratio between film and substrate, the thermal diffusion time and the thermal diffusivity of AlOy, AlNx and AlNxOy films were calculated based on MIRR measurements. Since the thermal properties of these materials rely on their composition and microstructure, which depend on the deposition conditions, a

detailed analysis of these interdependencies is also a major concern. In section 2 experimental details related to the film's production and characterization are given, whereas the basic description of the experimental setup for thermal properties, the description of the two-layer model and interpretation of signals amplitude and phase of the MIRR can be found in section 3. In section 4 the target potential evolution and the growth rate of the three systems of films are compared each other and some correlations with (and between) the composition and microstructure are also analysed with detail. The results of the thermal behaviour of the films are presented in section 5, along with some correlations with other basic characteristics and physical properties (electrical and optical) of the films. 2. Details of thin films deposition and characterization The thin films were produced by reactive DC magnetron sputtering, in a laboratory-sized deposition system [22], using silicon wafers with <100 > orientation (used for structural, morphological and composition analysis) and glass lamellae (ISO 8037) (used for thermal characterization). The substrates were placed in a grounded holder at 70 mm from the target, in a rotation mode-type (9 r.p.m.), and kept at a constant temperature of 100  C before discharge ignition by using a Joule effect resistor. Before the depositions, the substrates were subjected to an in-situ etching process, using pure argon with a partial pressure of 0.3 Pa (70 sccm), and a pulsed current of 0.6 A (Ton ¼ 1536 ns and f ¼ 200 kHz) for 900 s. A DC current density of 75 A m2 was used on the aluminium target (99.6% purity) with dimensions 200  100  6 mm3, being sputtered using a gas atmosphere composed of argon (working gas) and a reactive gas (different for each system). The argon flow used was the same for all depositions (70 sccm), corresponding to a partial pressure 0.3 Pa (measured before discharge ignition). The maximum flows/partial pressures of reactive gases used were i) 9 sccm/0.07 Pa of O2 in the AlOy system; ii) 45 sccm/0.32 Pa of N2 to produce the AlNx films; and iii) 27.5 sccm/0.22 Pa of a mixture composed of nitrogen and oxygen with a constant N2:O2 ratio of 17:3 (85% of N2/15% of O2), in order to prepare the oxynitride films, AlNxOy. In the particular case of the reactive mixture, the two gases used (N2 and O2) are mixed in the same bottle. Furthermore, different N2:O2 ratios were tested, including 19:1 and 9:1 ratios, however the ratio 17:3 was selected since it allows the incorporation of a wider range of nitrogen and oxygen in the films, simultaneously. In fact, one of the main objectives to use such a gas mixture was to avoid an early formation of oxide-like films, trying to extend as much as possible the deposition of thin films with a mixed composition between those of the metallic-like films (Al), semi-conductor (AlN) and oxide ones (Al2O3). Since oxygen is much more reactive than nitrogen, the latter is not incorporated in the film above a certain partial pressure of N2þO2 mixture. The partial pressure of the reactive gas was measured before discharge ignition, without argon, being directly proportional to the flow rate. The argon and the reactive gas are controlled by two flow metres and they are injected into the chamber using a circular tube (with small holes) positioned close to the internal wall of the chamber. Before each deposition, a target cleaning process was carried out in pure argon until the target voltage reached a steady state. Further details about the experimental setup and deposition conditions can be found elsewhere [11]. The chemical composition of the films was determined by Rutherford Backscattering Spectrometry (RBS) technique and the spectra analysed with the code NDF [27,28]. The structure and the phase distribution of the coatings were assessed by X-Ray diffraction (XRD), using a PANalytical X'Pert PROe MPD. The XRD patterns were deconvoluted assuming to be Pearson VII functions using

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

571

Winfit software [29]. The grain size was then estimated based of the integral breadth method. Morphological features of the samples were probed by scanning electron microscopy (SEM), using both plan view micrographs and cross-section observations. The SEM analysis was performed using a Leica Cambridge 5526 apparatus. The thickness of the samples was estimated by cross-section SEM analysis and the deposition (growth) rate was calculated by the ratio between the average thickness and the deposition time.

detector is then pre-amplified and subsequently sent to a lock-in amplifier (SR830). Finally, the information of the thermal wave's amplitude S and its phase lag F relative to the modulated excitation, corresponding to the complex surface temperature, are collected by a data acquisition system as a function of the heating modulation frequency f. All the experimental setup is controlled by a consonant software graphical interface, where the experimental results obtained are stored.

3. Basics of modulated infrared radiometry (MIRR)

3.2. Description of the two-layer model and interpretation of signals amplitude and phase of the MIRR

3.1. Basic description of the experimental setup for thermal measurements Radiometry is a non-destructive technique, belonging to the class of photothermal techniques, which rely on the principle of thermal waves generation [30]. Thermal waves are induced in the samples by means of excitation with a modulated energy beam, usually a laser, which heats the sample causing a temperature variation with the same frequency as the excitation [6]. The infrared radiation emitted by the sample is then analysed. Controlling the modulation frequency will allow the selective control of the thermal wave penetration depth. This penetration depth dependence on the frequency of the generated thermal waves makes photothermal techniques ideal candidates for the analysis of layered systems like thin films and coatings [30e34]. The experimental setup is schematically presented in Fig. 1. The system uses a 532nm diode-pumped solid-state (DPSS) neodymium-doped YAG laser as the excitation source, with a maximum output power of 1.1 W. For the light modulation, an acousto-optic modulator AOM40 (IntraCorp.) together with a light modulator signal processor ME-40 (IntraCorp.) are used. Infrared radiation emitted from the samples after excitation is focused into a HgCdTe detector by two barium fluoride (BaF2) lenses with 50 mm diameter. A germanium filter is positioned close to the detector in order to avoid any scattered light to reach the detector. The HgCdTe detector (model J15D12-M204-S04M-60 by EG&G Judson Technologies) operates in the IR range (2e12 mm) and is cooled with liquid nitrogen to prevent eddy currents caused by the absorption of thermal radiation from the detector itself. The signal from the

In this section the main aspects of a simplified two layer model, consisting on a coating and a substrate, will be presented. A more detailed description can be found elsewhere [25,35]. Each layer is characterized by the thickness (d), the thermal effusivity (e), the thermal diffusivity (a), the thermal conductivity (k), the specific heat capacity (c) and the mass density (r). The substrate can be considered as a semi-infinite and opaque medium in the modulation frequency range, since dc << db, where dc is the thickness of the coating and db the thickness of the substrate. According to this model, based on the principles of thermal wave interference in a thin layer, the thermal diffusion time, tc, and the thermal effusivity ratio (ec/eb) can be directly calculated from two measured quantities e the phase lag at the extremum of the phase curve and the modulation frequency at which the extremum occurs. In this model, it is assumed that the diameter of the heating beam is large, and consequently a one-dimensional heat flow is induced in the coating [36]. The signal amplitude S and phase F of the thermal waves are normalized by a homogeneous, optically opaque and thermally thick reference sample, in order to eliminate unwanted electronic interference coming from the detector, preamplifier and lock-in [37]. In this work Glassy Carbon (Sigradur®) with a 3 mm thickness was used as reference sample, ensuring that it will stay thermally thick in all modulation frequency range (1 Hze100 KHz) [37]. The recorded signals are therefore normalized in amplitude Sn ¼ Sr =Ss and phase Fn ¼ Fr  Fs , where the indices r and s refer to the reference sample and to the sample under study, respectively. Thermal parameters are determined by means of analytical

Fig. 1. A schematic picture of a typical experimental setup for modulated infrared radiometry (MIRR).

572

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

solutions proposed by the extremum method [35]. This model allows to determine the thermal properties of the material under investigation using the values of the frequency and phase lag, measured at the relative extrema. Equation (1) is the solution for the normalized phase based on the theoretical description of the Extremum Method [35]:



tan Fn ðf Þ


f ¼fextr



¼ tan½Fr ðf Þ  Fs ðf Þ
¼ tan Fn;extr f ¼fextr   pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2Rcb exp  2 pfextr tc sen 2 pfextr tc ¼  h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii2 1  Rcb exp  2 pfextr tc (1)

The equation described in (1) must fulfil the following boundary conditions:

4. Results and discussion 4.1. Deposition characteristics and correlation with the composition, structure and morphology of the films 4.1.1. Target potential evolution The target potential during the deposition of the films was monitored using an acquisition system which revealed an equilibrium state a few minutes after discharge ignition. Fig. 2 shows the evolution of the equilibrium target potential as a function of the partial pressure of the reactive gas used in each one of the three systems studied for comparison. In the case of the AlOy system, produced with an Ar/O2 mixture, it is possible to identify two different regimes for the target. The first regime (Reg.-M) occurs for partial pressures of O2 up to 3.6  102 Pa, where the target potential is approximately constant, with values around 480 V, slightly above the value obtained for a

 pffiffiffiffiffiffiffiffiffiffi 8 h  pffiffiffiffiffiffiffiffiffiffi 9  pffiffiffiffiffiffiffiffiffiffii pffiffiffiffiffiffiffiffi < 1  R2 exp 4 pf tc cos 2 pf tc =
4:Rcb : ptc exp 2 pf tc vtan Fn ðf Þ

cb  pffiffiffiffiffiffiffiffiffiffi

 pffiffiffiffiffiffiffiffiffiffii pffiffiffi
¼ ¼0 h  pffiffiffiffiffiffiffiffiffiffii2  : h 2 f ¼f extr  1 þ Rcb exp 4 pf tc sen 2 pf tc ; f ¼fextr v f 1  R2cb exp 4 pf tc

and

h

 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 þ R2cb exp  4 pfextr tc  cos 2 pfextr tc ¼ h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii sen 2 pfext tc 1  R2cb exp  4 pfextr tc (3) Equations (2) and (3) depend only on the parameters tc, the thermal diffusion time, and Rcb, which is the thermal reflection coefficient, responsible for controlling the reflections of thermal waves at the interface coating/substrate, since the frequency fextr and the normalized phase Fn;extr can be directly taken from the experimental data. The tc and Rcb parameters can then be calculated using the following Equations, (4) and (5):

Rcb

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi h n io u u1  tan 0:5  cos1 tan Fn;extr 2 n u h n ¼ ±t 2 io  exp 0:5 1 þ tan 0:5  cos1 tan Fn;extr h 2 io  cos1 tan Fn;extr

(2)

pure Ar discharge (~450 V). For higher partial pressures of O2 the target potential suddenly decreases for values around 300 V, which represents the second regime of the target (Reg.-C). A different behaviour can be observed in the AlNx system which was produced with an Ar/N2 mixture. The target potential gradually decreases with the rise of the partial pressure of the reactive gas (N2), from 452 V to 323 V for pressures up to 5.2  102 Pa (Reg.GP). This first stage is then followed by a two-step regimes, with almost constant target potential values: one from 5.7  102 Pa to 8.8  102 Pa, with values around 300 V (Reg.-C1), and another, from 9.6  102 Pa to 3.1  101 Pa where the target potential remained constant with values of about 255 V (Reg.-C2). The behaviour of the target observed when the AlNxOy films were produced presents some similarities with the first regime of

(4)

and

tc ¼

( h 2 o2 d2c 1 ¼ 0:5  cos1 tan Fn;extr 4pfextr ac

(5)

Equation (5) gives the relation between tc and the thermal diffusivity ac, responsible for the spread of radiation by the sample [38]. The coefficient of thermal reflection, Rcb can be rewritten as a function of the thermal effusivity ratio (ec/eb):

pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðkrCÞ ð1 þ Rcb Þ ðec =eb Þ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffic ¼ ðkrCÞb ð1  Rcb Þ

(6)

Equations (5) and (6) allows to calculate the thermal parameters of interest discussed so far, the thermal diffusion time (tc), the thermal effusivity ratio (ec/eb) and the thermal diffusivity (ac) if the coating's thickness (dc) is know.

Fig. 2. Comparison between the target potential of the three systems of films in study (AlOy, AlNx, AlNxOy), as a function of the partial pressure of the reactive gas. The partial pressures of Ar (fixed in 0.3 Pa) and of the reactive gas were measured, independently, before discharge ignition.

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

the AlNx system, for partial pressures of N2þO2 up to 5.4  102 Pa, namely a gradual decrease of the target potential from 452 V to 320 V (Reg.-GP0 ). This almost linear decrease is then followed by another regime which starts when the partial pressure of the reactive mixture is 5.6  102 Pa, remaining approximately constant thereafter (Reg.-C'). The evolution of the target potential is strongly influenced by the reactive gas species, since they promote compound formation and ion implantation at the target's surface [39e42]. The oxidation and/or nitridation of the target modify the sputtering yield of the target as well as the number of secondary electrons emitted from it, due to ion bombardment. These secondary electrons are fundamental in the discharge sustaining mechanism and can be quantified by the effective secondary electron emission yield (geff.) [43]. According to results obtained by D. Depla and co-workers, the geff. coefficient of an aluminium target can double its value when nitridation and/or oxidation occurs at its surface [43]. The effective secondary electron emission coefficient is also one of most important parameters that influence the minimum voltage required to sustain the magnetron discharge (Vm), being inversely proportional to Vm [43]. According to the stated above, the evolution of the target potential with the increase of the partial pressure of the reactive gas is certainly related with the modification of the target surface and thus with the geff. coefficient. The Ar/O2 mixture promotes a metallic mode in the low partial pressure of O2 regime, where the target should remain practically clean (Reg.-M). In this regime the target potential values are higher than the pure Ar discharge and this can be explained with the formation of an oxide monolayer at the surface of the target which decreases geff., according to studies developed by D. Depla and co-workers [43]. At a certain critical value of the partial pressure of O2, the target potential drops and this behaviour can be explained by the formation of a compound material (Al2O3) covering the surface of the aluminium target, which has a higher value of geff. than a clean Al target [44]. The abrupt transition from a metallic mode (Reg.-M) towards a compound mode (Reg.-C) in an Ar/O2 atmosphere is thus typical in magnetron sputtering [44,45]. When the O2 is substituted by N2 in the gas atmosphere a different evolution can be observed. In fact, the Ar/N2 mixture induces a gradual decrease of the target potential as the partial pressure of the reactive gas rises. This tendency can be explained with the “gradual poisoning” of the target with nitrides (Reg.-GP) until reaching the compound mode regimes (Reg.-C1 and Reg.- C2). This behaviour is also in agreement with studies carried out by other authors which used Ar/N2 atmospheres to sputter Al targets [45e47]. The results displayed in Fig. 2 suggest also that when the mixture is composed by Ar/(N2þO2) the evolution of the target potential follows a tendency that can be related with the mixtures used to produce the binary systems. The gradual decrease of the target potential in the Ar/(N2þO2) discharge (Reg.-GP0 ) is very similar to the behaviour found in the first regime of the Ar/N2 discharge (Reg.-GP0 ). This means that the target is most probably being poisoned mainly with N2, an expected behaviour since the reactive mixture contains more N2 than O2 (the ratio N2:O2 is 17:3). On the other hand, since the O2 in more reactive than N2 one cannot exclude some oxidation of the target. The substitution of the nitride layers by aluminium oxide is also to be expected, especially in the compound mode regime of the target (Reg.-C') [48]. As discussed above, the target surface was strongly affected by the reactive gas partial pressure, which caused the gradual coverage of the target with nitrides and/or oxides and thus affecting the cathode potential and its sputtering yield. The intended change of the partial pressure of reactive gas influences the

573

Fig. 3. Growth rate of the AlOy, AlNx and AlNxOy films as a function of the partial pressure of the reactive gas.

target condition and the plasma itself which, altogether, are likely to induce changes in the growth characteristics and in the microstructural features of the films. 4.1.2. Growth rate of the films In order to follow the changes observed in the target potential (Fig. 2) and its influence on the deposition characteristics, the growth rate of the three systems of films are presented in Fig. 3. The growth rate of the AlOy films increases from 35 nm min1 to 66 nm min1. Then it abruptly drops to ~2 nm min1 when the target is completely “poisoned” (oxidized) in Reg.-C. According to these results one might identify two main zones: zone M for partial pressures of O2 up to 3.6  102 Pa, corresponding to the Reg.-M of the target, and zone C for higher partial pressures (Reg.-C of the target). A smooth increase in the growth rate from 35 nm min1 to 47 nm min1 was observed for the AlNx system, decreasing to 22 nm min1 within the range of partial pressures of N2 where the target is gradually poisoning (up to 5.4  102 Pa, Reg.-GP) - zone GP. This group is then followed by two groups of samples, each one with very similar growth rates, in agreement with the target potential evolution; one with values around 13 nm min1, deposited with the target in regime (Reg.)-C1 (zone C1) and another with values close to 6 nm min1, corresponding to samples deposited in the regime (Reg.)-C2 (zone C2). An important feature evidenced in Fig. 3 is that the evolution of the growth rate of the ternary AlNxOy, which was discussed in more detail in [22], is clearly different from the binary systems. In fact, a similar behaviour between the AlNxOy and AlNx was expected due to similarities in the target potential evolution. Nevertheless, one can find three different tendencies only in the first regime (Reg.GP0 ) of the target in what concerns to the AlNxOy system. Initially it is approximately constant with values around 35 nm min1 (zone GP1), then it increases up to 63 nm min1 (zone GP2), decreasing again to 24 nm min1 (zone GP3). In accordance with the shift observed in the target potential values when the compound mode of the target was achieved in both binary systems, the growth rates of the AlNxOy samples also dropped to values around 5 nm min1 in the regime (Reg.)-C0 of the target (zone C'). 4.1.3. Composition and microstructure of the films The chemical composition of the three system of films is shown

574

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

Fig. 4. Ternary diagram with the chemical compositions of the deposited samples.

in the ternary diagram, Fig. 4, together with the positions of the base compounds AlN and Al2O3, as well as aluminium oxynitride (g- Al41N8O51). From the analysis of the diagram it's clear that a wide range of compositions can be obtained by simply changing the partial pressure of the reactive gas inside the deposition chamber. In the binary systems, AlNx and AlOy, the compositions vary between pure aluminium and the stoichiometric compound which is thermodynamically more favourable to be formed, aluminium nitride (AlN) and aluminium oxide (Al2O3), respectively. Nevertheless, there is a gap of compositions in the AlOy system, increasing only up to CO/CAl atomic ratios of y ¼ 0.59 in zone M, due to the abrupt change in the target condition; as the target becomes oxidized (zone C), the composition of the films starts to be stoichiometric Al2O3, regardless of the partial pressure of O2. A progressive increase of the CN/CAl atomic ratio was observed in the AlNx system, namely in zone GP (x  0.42) and zone C1 (0.54  x  0.78), until a close-stoichiometric AlN compound is formed in zone C2 (0.88  x  0.91). In the particular case of the ternary AlNxOy system, one can observe that the thin films are located inside the triangle in which the vertices are defined by the positions of Al, AlN and Al2O3 compounds. Furthermore, the evolution of the COþN/CAl atomic ratio can also be correlated with the deposition characteristics: i) it slightly increases within zone GP1 with COþN/CAl atomic ratios rising up to 0.13; ii) in the case of the films ascribed to zone GP2, where the deposition rate is rapidly increasing, one can observe very similar concentration ratios (COþN/ CAl), around 0.17e0.18 and iii) in zone GP3, where the deposition rate is gradually decreasing, one can observe a stronger increase of the COþN/CAl ratio ranging from a value of 0.41 towards a value of 0.85. Another important feature about the chemical composition analysis is that the samples that were prepared with the target in compound mode (zone C0 ), revealed very similar compositions, where the oxygen amount is very close to 60 at.%, and the aluminium concentration is around 40 at.%, while the nitrogen concentration drops to a residual value (non-detectable within the resolution of the experimental setup, meaning that it should be below 5 at.%). This is the result of the higher affinity of aluminium for oxygen in comparison to nitrogen, which leads to the formation of Al2O3 despite the high partial pressure of N2. A more detailed analysis of the composition of the films can be found elsewhere [22]. The composition of the samples analysed by MIRR is displayed in Table 1.

In Figs. 5e7 microstructural features of the films are shown; the correlation of the crystalline structure with the morphology of the films is evident. In the zone M (AlOy system) the reactive gas is almost completely gettered and hence the target condition remains metallic. In this case the erosion rate of the target remains practically constant allowing the formation of sub-stoichiometric AlOy compounds with relatively high growth rates (Fig. 3). The substoichiometric AlOy films have a dense columnar-to-granular growth, as shown by the SEM images displayed in Fig. 5(b). One can also claim that these films are crystalline due to the appearance of two diffraction peaks indexed to cubic (fcc) aluminium [11], although a gradual amorphization can be observed [11]. When the O2 partial pressure increases and the maximum amount of reactive gas which can be gettered is reached, the target becomes completely poisoned (zone C) with aluminium oxide. Since the sputtering yield of a poisoned target is lower than a metallic one [49,50], the growth rate dramatically decreases in zone C. For example, according to some calculations from D. Depla and coworkers, the sputtering yield can drop from 0.6 (pure Al target) to 0.23 (compound Al2O3 target) [43,48]. A representative film of zone C, with chemical composition of Al2O3.2, is shown in Fig. 6(b) and appears to be very compact. The evolution of the growth rate of AlNx system can be also correlated with the target condition. The decrease of the sputtering yield due to the formation of nitrides at the target's surface is likely to induce some changes in the growth rate of the films [45,47]. But the reduction of the sputtering yield is not the only factor affecting the evolution of the growth rate values of the films. The SEM images of AlNx thin films, which are displayed in Fig. 6(b), show that the type of growth of these thin films also changes as the atomic ratio of CN/CAl increases. The differences are more evident within zone GP (AlNx system), which includes thin films with atomic ratios of CN/CAl up to 0.42 and showed an Al-type (fcc) structure. Initially, the growth is columnar-type and may be associated to “zone T” of the Mahieu model [51], evolving into a kind of microstructure that fits in “zone Ia” of the same model [51]. The microstructural changes are also followed by a gradual loss of crystallinity of the Altype structure, Fig. 6(a). In the specific case of AlN0.42 sample, it is possible to observe the formation of a granular film composed of microscopic aggregates, approximately spherical, with voids, that can be associated to a low density thin film. SEM images also confirm the morphological differences between the films of zone GP and zone C1, which is consistent with the changes observed in the crystalline structure, Fig. 6(a). Thus, the low crystallinity observed in zone C1, associated to lower growth rates due to target condition, favours the formation of compact and denser thin films. In zone C2, the AlNx films are again showing some crystallinity as suggested by the presence of AlN diffraction peaks. These films also appear to be quite dense and smooth as evidenced in Fig. 6(b-v). The growth rate evolution values of the ternary system AlNxOy (Fig. 3) stands out from what was observed in AlNx and AlOy, as already discussed above. The constant values obtained for lower partial pressures of N2þO2, followed by a sharp increase of the growth rate, would not be expected for the AlNxOy films. However, this unusual behaviour becomes clearer after the SEM analysis of the films, see Fig. 7(b). While the films indexed to zone GP1 revealed a typical columnar growth (“Zone Ic” of Mahieu model [51]), the films of zone GP2 exhibited a granular structure separated by voids, which is known to increase the roughness and porosity and thus explains the sudden increase in the growth rate [22]. Afterwards, in zone GP3, the gradual loss of crystallinity of the films, Fig. 7(a), is again inducing some densification of the granularvoided microstructure, Fig. 7(b). This fact and the reduction of the target sputtering yield can explain the reduction of the growth rate

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

575

Table 1 Composition and characteristics of the samples analysed by MIRR: microstructure; electrical resistivity (r300K) and temperature coefficient of resistance (TCR) signal; optical behaviour. Zone AlOy M

C AlNx GP

Films' compositions

Microstructure

Electrical behaviour

Optical behaviour

Al AlO0.07 AlO0.11 AlO0.22 AlO0.37 AlO0.56 AlO0.59 Al2O3

Columnar, Al-type structure Columnar-to-granular, Al-type structure Dense/compact

108 < r300K < 107 Positive TCR 107 < r300K < 106 Positive TCR r300K > 1010

Opaque (metallic)

Al AlN0.07 AlN0.16

Columnar, Al-type structure Granular voided, Al-type structure Dense/compact, amorphous

108 < r300K < 106 Positive TCR 105 < r300K < 101 Positive TCR 104 < r300K < 101 r300K > 108

AlN0.27 AlN0.42 C1 C2

AlN0.54 AlN0.64 AlN0.71 AlN0.78 AlN0.88 AlN0.91

AlNxOy GP1

Al AlN0.01O0.08 AlN0.04O0.09

Dense/compact, AlN-type structure

GP2 GP3

AlN0.09O0.08 AlN0.06O0.12 AlN0.16O0.25 AlN0.30O0.34 AlN0.47O0.35 AlN0.51O0.34

C0

Al2O3

Columnar, Al-type structure Granular voided, Al-type structure

Dense/compact

108 < r300K < 107 Positive TCR 106 < r300K < 105 Positive TCR 105 < r300K < 104 Negative TCR r300K > 1010

Fig. 5. (a) X-ray diffractograms and (b) SEM micrographs of representative AlOy films.

Fig. 6. (a) X-ray diffractograms and (b) SEM micrographs of representative AlNx films.

Opaque (dark grey) Transparent Opaque and metallic tones Opaque and dark grey tones Opaque Semi-transparent Transparent

Opaque (metallic) Opaque (dark grey)

Transparent

576

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

Fig. 7. (a) X-ray diffractograms and (b) SEM micrographs of representative AlNxOy films.

values in zone GP3. Furthermore, this granular microstructure, which can be indexed to the “Zone Ia” of the Mahieu model [51], was also observed in the AlNx system, but for a very limited range of N2 partial pressures. Finally, in zone C' (AlNxOy system) the microstructure of the films becomes even denser and changes dramatically. This feature was expected to occur due to the formation of amorphous Al2O3 films and very low deposition rates due to the target condition. A very similar behaviour was observed in zone C of the AlOy system, Fig. 5(b-v). According to the analysis referred to above, it is clear that the structural and morphological evolution of the thin films during growth is strongly correlated with the deposition characteristics which, in turn, are interconnected with the condition of the target and the composition of the plasma [12]. It is then natural to observe microstructural differences between the binary systems and the ternary AlNxOy. Estimates of the average grain size of crystals with orientation <111>, for AlOy, AlNx and AlNxOy systems as well as the average size of the AlN crystalline grains with orientation <101> in the AlNx system are shown in Fig. 8. In AlOy binary system one can observe a rapid decrease of the grain size from 85 nm for the film with low oxygen content, AlO0.07, towards values close to 15e20 nm for higher y coefficients. This trend can be explained by the increase of oxygen concentration in the films, which can be segregated to the surface and the grain boundaries and thus inhibits crystal growth due to the reduced mobility of aluminium in oxide layers [52]. The structural evolution of AlNx thin films resulted in the growth of Al grains with estimated average sizes between 15 and 30 nm in zone GP and AlN nanocrystals of about 20 nm in the films of zone C2. As expected, the average grain size of the AlNxOy thin films shows a distinct evolution when compared to the binary systems and is strongly correlated with the observed evolution in deposition characteristics. More details can be found elsewhere [22]. 4.2. Thermal characteristics analysis At this point it is important to highlight that the simple change of a deposition parameter, such as the partial pressure of the reactive gas (N2 and/or O2), strongly affects the target condition and the deposition characteristics of the three sets of films produced. According to the overall set of results presented above, it was possible to obtain a wide range of non-metallic/aluminium atomic ratios in each of the systems studied. However, the increase of the

Fig. 8. Grain size (GS) evolution and thickness (dc) of the prepared films.

reactive species in the plasma not only affected the compositions of the films, but also their structural and morphological features. The addition of small quantities of oxygen and nitrogen to metallic aluminium progressively lowers the crystallinity of the polycrystalline Al-type structure, evolving to a kind of nanocomposite material composed of aluminium grains surrounded by amorphous oxide, nitride and oxynitride compounds [23]. At the same time, the typical columnar growth found for the high Al-content films also changed to granular/voided microstructure, as discussed above (Figs. 5e7). Further increasing the oxygen and nitrogen content towards stoichiometric conditions induced again more modifications in the structure and morphological characteristics of the films.

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

Here, the target condition played also an important role due to the poisoning effect. In fact, when the compound regimes of the target are attained, the granular/voided microstructure is no longer observed and more compact and dense films are formed, as also observed in Figs. 5e7. In previous works it was shown that the particular composition, structure and morphology of the films resulted in very different electrical and optical properties [11,22]. Some features that are worth to mention are the transition behaviour between the typical responses of metallic aluminium and the insulator/semiconductor properties of Al2O3 and/or AlN. For example, an increase in nonmetallic content in the films can induce a gradual transition from positive to negative temperature coefficients of resistance (TCR), as well as an unusual large broadband optical absorption for some stoichiometries. A close correlation between the physical (electrical, optical, etc.) properties and the composition and microstructure of the films is also expected to occur in what concerns the thermal properties. In Table 1 it is presented the composition of the films that were qualitatively analysed by MIRR, along with the corresponding indexed “zone”, the microstructural features, and a brief summary of the electrical and optical responses of the films. The samples in bold were also quantitatively analysed by MIRR experiments using the model described in section 3. The binary systems AlOy and AlNx, as well as a ternary system AlNxOy, were analysed by modulated IR radiometry in order to study the evolution of the thermal behaviour with the x and/or y coefficients. The thermal parameters of interest were determined for the samples where the two-layer model, described in Section 3.2, can be applied [25]. In such samples, the values of the thermal diffusivity (ac) and thermal effusivity ratio (ec/eb) rely on the values of the normalized phase Fn and the corresponding frequency f1/2 at the extremum of the Fn vs. f1/2 curves. In Fig. 9(aec) the results of the inverse calibrated modulated IR phases obtained for the AlOy films are presented. The samples shown in Fig. 9(a) represent a first group of films, indexed to zone M, with a metallic-like surface aspect due to their relatively high aluminium content (CO/CAl atomic ratios up to 0.22). They are also very conductive in terms of electrical and thermal transport (aluminium is one of the best thermal conductors). This implies that a great thermal diffusion length compared with the film thickness (3e5 mm) and therefore higher modulation frequencies (f >> 100 kHz; above the range available), are needed to perform a proper characterization of these samples using MIRR. As a consequence, even if the experimental points obtained are shown, no quantitative information can be calculated. In Fig. 9(b), three samples with higher oxygen contents, corresponding to CO/CAl atomic ratios between 0.37 and 0.59, are presented. These films also belong to zone M, however they become very dark (dark grey tones) as the microstructure changes from columnar to granular. The electrical resistivity increases about one order of magnitude. As the films become less conductive, relative maxima (ec > eb) appear, in the range of intermediate frequencies (50 < f/kHz < 100). In this case, the thermal properties of the films can be calculated. The thermal parameters, thermal diffusivity (ac) and thermal effusivity ratio (ec/eb), along with the thickness estimated by SEM analysis, are presented in Table 2. In Fig. 9(c) the MIRR signal of a representative sample of zone C (AlOy system) is shown. This sample is composed of Al2O3, which is quite different from the sub-stoichiometric AlOy films. In this particular case the film is optically transparent and the theoretical model cannot be applied since it assumes that the coating is opaque both in IR and visible spectral range. Fig. 10(aec) represent the results obtained for the AlNx system. Similarly to the AlOy system, it was not possible to quantify the

577

Fig. 9. Inverse calibrated IR phase lag signals measured for the AlOy films. The solid lines show the fit of the theoretical signals to the experimental results using the twolayer model (Sec. 3.2).

Table 2 Measured relative phase extreme and thermal parameters for AlOy samples. Sample

Fn,extr./degree

tc/ms

ec/eb

dc/mm

ac/m2 s1

AlO0.37 AlO0.56 AlO0.59

17.9 18.0 15.1

4.4 6.9 9.1

2.8 2.8 2.4

5.5 6.0 4.7

7.1E-06 5.2E-06 2.4E-06

thermal properties of the AlNx samples with higher aluminium content (CN/CAl  0.27), indexed to zone GP, within the available range of frequencies. Nevertheless, for intermediate CN/CAl values, between 0.42 and 0.64, quantitative information was extracted from the measured curves (Fig. 10(b)). In this case, with the increase of nitrogen content, the films become less conductive and the two-layer model could be applied. The results are displayed in Table 3. An important feature about these films is that their microstructures do not seem to play an important role on the thermal behaviour. While the sample AlN0.42 (zone GP) presents a granular voided microstructure, samples AlN0.54 and AlN0.64 (zone C1) are dense and compact. For samples with higher CN/CAl atomic ratio, between 0.71 and 0.91, which include samples from both zones C1 and C2, the situation is quite different. These films are gradually becoming transparent due to the formation of nitrides with stoichiometry, CN/CAl, approaching one and they are thinner. Therefore, the thermal characterization of these samples cannot be performed using this experiment, since higher frequencies are needed. Similarly to the Al2O3 film, the information will come essentially from the

578

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

Fig. 10. Inverse calibrated IR phase lag signals measured for the AlNx films. The solid lines show the fit of the theoretical signals to the experimental results using the twolayer model (Sec. 3.2).

Fig. 11. Inverse calibrated IR phase lag signals measured for the AlNxOy films. The solid lines show the fit of the theoretical signals to the experimental results using the twolayer model (Sec. 3.2).

Table 3 Measured relative phase extreme and thermal parameters for AlNx samples.

Table 4 Measured relative phase extreme and thermal parameters for AlNxOy samples.

Sample

Fn,extr./degree

tc/ms

ec/eb

dc/mm

ac/m2 s1

Sample

Fn,extr./degree

tc/ms

ec/eb

dc/mm

ac/m2 s1

AlN0.42 AlN0.54 AlN0.64

17.95 15.35 20.44

1.1 1.1 5.2

2.8 2.4 3.4

2.0 1.0 1.1

3.6E-06 9.6E-07 2.5E-06

AlN0.09O0.08 AlN0.06O0.12 AlN0.16O0.25

13.1 14.8 14.9

2.9 1.8 1.1

2.1 2.4 2.3

2.9 3.6 2.0

2.9E-06 7.4E-06 3.7E-06

substrate, having no practical use for the present work. Since the information shown in Fig. 11(c) comes mainly from the substrate (glass), the thermal diffusion length of the incident radiation is much larger than the film thickness (between 0.7 and 1.0 mm) for this group of samples [37]. Fig. 11(aec) plots the photothermal signals for the AlNxOy system. Once again, the thermal parameters of the samples with higher aluminium content, CNþO/CAl < 0.13, indexed to zone GP1, could not be calculated for the same reasons already explained for the binary systems. Again, only the experimental data are plotted in Fig. 11(a). In Fig. 11(b), one can clearly observe the presence of maxima for the curves shown (0.17 < CNþO/CAl < 0.41). The effusivity ratio and the thermal diffusivity could then be calculated; the data are presented in Table 4. Besides the increase of aluminium content in the films, major differences were also observed in their microstructure, changing from a typical columnar growth (zone GP1) towards a granular voided morphology (zones GP2 and GP3). These particular microstructural features induced an unusual broadband optical absorption in the films, with an almost flat optical reflectance

profile as low as 5% [11], making the films dark grey. This optical behaviour was observed in the samples indexed to zones GP2 and GP3 and was explained based on the microstructural arrangement proposed for these films; a network of aluminium nanoparticles dispersed in an oxide/(oxy)nitride matrix [11]. The insulator/semiconducting matrix constitutes also a barrier for the electron conduction throughout the film, becoming more important for the films with CNþO/CAl atomic ratios between 0.64 and 0.85 [22], which also explained the negative TCR values found for these films. Therefore, although the samples indexed to zone GP3 are very similar to each other in terms of optical behaviour, the enlargement of the barrier component of the films gave rise to important changes in the electrical properties. This trend also prevented the quantification of the thermal properties of these films. For this kind of films the thermal properties can only be performed using higher frequencies, since the behaviour of the experimental curves indicate a possible relative maxima near 100 kHz, Fig. 11(c). Finally, the behaviour of the samples indexed to zone C' (AlNxOy system), not shown, is similar to those observed in zone C of the AlOy system (see Fig. 9(c)), since they are also composed of Al2O3.

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

As a general analysis of the results obtained from the thermal measurements, one can divide, in each system, the films in three distinct categories (or regions). First, the group of very conductive films which did not allow to obtain quantitative information using the presented technique. In this kind of films it was possible to observe a metallic surface and therefore the films were “mirrored” due to the high aluminium content. For this reason, it is necessary to apply modulation frequencies clearly above the available range, in order to determine the thermal properties. Secondly, as the concentration of aluminium decreases, resulting from the increase of nitrogen and/or oxygen content, the samples become more “opaque” allowing the calculation of the thermal parameters using the two layer model (Sec. 3.2). Finally, in the third region, the films become optically (semi-)transparent and the information recorded comes mainly from the glass substrate, making the measurements useless for quantitative purposes. According to the estimated thermal parameters of the “transition region” (Tables 2e4), for the same system, the effusivity ratios and the thermal diffusivities changed only slightly. Anyway, the range of atomic ratios where the thermal properties could be calculated was limited to: i) 0.37 and 0.59 in the AlOy system; ii) 0.42 and 0.64 for AlNx films and iii) 0.17 and 0.41 for the oxynitride system AlNxOy. Furthermore, the number of samples within those regions is somehow relatively low to proper correlate the changes found to any particular property or feature of the thin film systems considered. Nonetheless, an important note that might be worth mention is the proximity between the values of each one of the three systems. Furthermore, in the oxynitride system, AlNxOy, the region quantitatively analysed shifts to lower atomic ratios, showing that it is possible to obtain the same values as the individual binary systems by adding small amounts of nitrogen and oxygen to the growing Al film. A more careful analysis of the thermal diffusivity values (in the order of 106 m2 s1) suggests that they are relatively “low” when compared to known good thermal conductors like silver or copper, however they are in line with what is reported in literature for related systems [53,54]. Besides that, it is well known that metallic thin films have a thermal diffusivity at least one order of magnitude below that of the original bulk material, but still good enough for their potential applications, namely those of bioelectrodes and/or biosensors such as the targeted applications of the current thin film systems. In this sense they can be considered as “good thermal conductors”. Moreover, the obtained results are clearly indicative of an adequate thermal response if this particular thin film system is aimed to be applied as a sensor material, or even in a given protective-like application. In fact, the values of the thermal diffusivity compare well with those presented in previous works [25,34,55,56]. These results are also a good indication of a reduced localized heating, which is a major requirement if one intends to use this thin film system in small sensor devices or in electrodes for biosignal acquisition, such as those for electroencephalography (EEG) and electromyography (EMG), as it is the case of the main research area in the group [5,34,57] [56], being aware that the thermal diffusivity/conductivity can change quite strongly, depending on the crystalline quality and on the oxygen content of each film [53]. Regarding the effusivity ratio, which is a crucial parameter in controlling the heat propagation between different media, the obtained results are again rather promising. In fact, the present results indicate that if this particular thin films system will be used in a given application, the heat propagation between the thin film and the substrate can be estimated using MIRR, providing an important information concerning systems where particular local heating might be developed, such as the case of the mentioned applications (EEG or EMG). As a final remark, one can say that the ternary AlNxOy system follows a relatively similar behaviour if compared to previous

579

measurements on zirconium oxynitride (ZrOxNy) [25] and titanium oxycarbide systems (TiCxOy) [55]. In those systems it was clearly observed a gradual decrease in thermal diffusivity as the gas flow increases, following the changes in the (micro)structural properties found. In the case of the systems presented here, even if it was not possible to find numerical values for the extreme cases, due to the experimental limitations presented above, the transition between different regimes clearly appears in the measurements, which again, follows the observed zone differences in the other properties. It is particularly noticeable the metallic-like and oxide-like zones, where the thermal analysis shows two clearly distinct responses, with a kind of transition region in between, revealing the already analysed transition behaviour between Al and AlN/Al2O3like responses. 5. Conclusions Modulated IR Radiometry, a non-contact and non-destructive technique based on the excitation of thermal waves by modulated laser beam heating and on the IR detection of the thermal wave response, has proved to be very efficient in the thermal characterization of thin films. In this work, this method was applied in order to study qualitative and quantitatively the thermal response of AlOy, AlNx AlNxOy thin films deposited by magnetron sputtering. The deposition characteristics of the films were analysed and correlated with the composition and microstructure of the films. Nevertheless, due to the particular characteristics of the films, only for a restricted set of samples it was possible to extract the thermal parameters using the extremum method model. For the three systems studied, it was possible to observe a metallic surface due to the high aluminium content preventing a proper characterization of the thermal properties. In the “transition region”, where an increase of the reactive species, oxygen and/or nitrogen, is observed, the samples become less metallic allowing a proper thermal characterization in the frequency range available. Finally, when the films become (semi-)transparent they do not match the type of samples analysed by the proposed model, since the thermal wave's response is coming essentially from the substrate. The experimental results showed a relatively stable pattern, but very dependent on the oxygen and/or nitrogen content. For a proper characterization of the high conductive films and the semitransparent ones higher modulation frequencies are needed. The theoretical model should also be improved in order to take into account the complex characteristics of the samples. Acknowledgements This research is sponsored by FEDER funds through the program COMPETE e Programa Operacional Factores de Competitividade e ~o para a Cie ^ncia e a and by National funds through FCT e Fundaça Tecnologia e, under the projects PEST-C/FIS/UI607/2013, PEst-C/ EME/UI0285/2013. J. Borges also acknowledges the support by the European social fund within the framework of realizing the project “Support of inter-sectoral mobility and quality enhancement of research teams at Czech Technical University in Prague”, CZ.1.07/ 2.3.00/30.0034. F.M. Couto acknowledges CAPES - Foundation, Ministry of Education of Brazil, Brasília e DF 70040-020, Brazil, funding by stage sandwich doctorate, through PDSE - Doctoral Program Sandwich. References [1] R. Franchy, Growth of thin, crystalline oxide, nitride and oxynitride films on metal and metal alloy surfaces, Surf. Sci. Rep. 38 (2000) 195e294.

580

J. Borges et al. / Materials Chemistry and Physics 163 (2015) 569e580

[2] F. Vaz, N. Martin, M. Fenker, Metallic Oxynitride Thin Films by Reactive Sputtering and Related Deposition Methods: Process, Properties and Applications, Bentham Science Publisher, AG Bussum e The Netherlands, 2013. [3] H. Holleck, Material selection for hard coatings, J. Vac. Sci. Technol. A 4 (1986) 2661e2669. re, E. Alves, [4] F. Vaz, P. Machado, L. Rebouta, P. Cerqueira, P. Goudeau, J.P. Rivie K. Pischow, J. de Rijk, Mechanical characterization of reactively magnetronsputtered TiN films, Surf. Coatings Technol. 174e175 (2003) 375e382. [5] P. Fiedler, L.T. Cunha, P. Pedrosa, S. Brodkorb, C. Fonseca, F. Vaz, J. Haueisen, Novel TiNx-based biosignal electrodes for electroencephalography, Meas. Sci. Technol. 22 (2011). [6] Y. Taniyasu, M. Kasu, T. Makimoto, An aluminium nitride light-emitting diode with a wavelength of 210 nanometres, Nature 441 (2006) 325e328. [7] P. Fiedler, C. Fonseca, P. Pedrosa, A. Martins, F. Vaz, S. Griebel, J. Haueisen, Novel flexible Dry multipin electrodes for EEG: signal quality and interfacial impedance of Ti and TiN coatings, in: Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, 2013, pp. 547e550. [8] N. Martin, A. Besnard, F. Sthal, F. Vaz, C. Nouveau, The contribution of grain boundary barriers to the electrical conductivity of titanium oxide thin films, Appl. Phys. Lett. 93 (2008) 064102. [9] H.C. Barshilia, N. Selvakumar, G. Vignesh, K.S. Rajam, A. Biswas, Optical properties and thermal stability of pulsed-sputter-deposited AlxOy/Al/AlxOy multilayer absorber coatings, Sol. Energy Mater. Sol. Cells 93 (2009) 315e323. [10] J. Borges, F. Vaz, L. Marques, AlNxOy thin films deposited by DC reactive magnetron sputtering, Appl. Surf. Sci. 257 (2010) 1478e1483. [11] J. Borges, N.P. Barradas, E. Alves, M.F. Beaufort, D. Eyidi, F. Vaz, L. Marques, Influence of stoichiometry and structure on the optical properties of AlNxOy films, J. Phys. D Appl. Phys. 46 (2013) 015305. [12] J. Borges, N. Martin, F. Vaz, L. Marques, Process monitoring during AlNxOy deposition by reactive magnetron sputtering and correlation with the film's properties, J. Vac. Sci. Technol. A Vac. Surfaces Films 32 (2014) 021307. [13] J. Borges, E. Alves, F. Vaz, L. Marques, Optical properties of AlNxOy thin films deposited by DC magnetron sputtering, in: International Conference on Applications of Optics and Photonics, Proceedings of SPIE - the International Society for Optical Engineering, Braga, 2011, p. 80010F. [14] R. Arvinte, J. Borges, R.E. Sousa, D. Munteanu, N.P. Barradas, E. Alves, F. Vaz, L. Marques, Preparation and characterization of CrNxOy thin films: the effect of composition and structural features on the electrical behavior, Appl. Surf. Sci. 257 (2011) 9120e9124. [15] D. Cristea, A. Crisan, N.P. Barradas, E. Alves, C. Moura, F. Vaz, L. Cunha, Development of tantalum oxynitride thin films produced by PVD: study of structural stability, Appl. Surf. Sci. 285 (2013) 19e26. [16] M. Fenker, H. Kappl, P. Carvalho, F. Vaz, Thermal stability, mechanical and corrosion behaviour of niobium-based coatings in the ternary system Nb-O-N, Thin Solid Films 519 (2011) 2457e2463. [17] P. Carvalho, L. Cunha, E. Alves, N. Martin, E. Le Bourhis, F. Vaz, ZrOxNy decorative thin films prepared by the reactive gas pulsing process, J. Phys. D Appl. Phys. 42 (2009) 195501. re, [18] P. Carvalho, F. Vaz, L. Rebouta, S. Carvalho, L. Cunha, P. Goudeau, J.P. Rivie E. Alves, A. Cavaleiro, Structural stability of decorative ZrNxOy thin films, Surf. Coatings Technol. 200 (2005) 748e752. [19] L. Cunha, L. Rebouta, F. Vaz, M. Staszuk, S. Malara, J. Barbosa, P. Carvalho, re, Effect of thermal treatments on E. Alves, E. Le Bourhis, P. Goudeau, J.P. Rivie the structure of MoNxOy thin films, Vacuum 82 (2008) 1428e1432.
, F. Sthal, J. Takadoum, F. Vaz, [20] N. Martin, J. Lintymer, J. Gavoille, J.M. Chappe L. Rebouta, Reactive sputtering of TiOxNy coatings by the reactive gas pulsing process. Part I: pattern and period of pulses, Surf. Coatings Technol. 201 (2007) 7720e7726. [21] E. Alves, A.R. Ramos, N.P. Barradas, F. Vaz, P. Cerqueira, L. Rebouta, U. Kreissig, Ion beam studies of TiNxOy thin films deposited by reactive magnetron sputtering, Surf. Coatings Technol. 180e181 (2004) 372e376. [22] J. Borges, N. Martin, N.P. Barradas, E. Alves, D. Eyidi, M.F. Beaufort, J.P. Riviere, F. Vaz, L. Marques, Electrical properties of AlNxOy thin films prepared by reactive magnetron sputtering, Thin Solid Films 520 (2012) 6709e6717. [23] J. Borges, C. Fonseca, N.P. Barradas, E. Alves, T. Girardeau, F. Paumier, F. Vaz, L. Marques, Influence of composition, bonding characteristics and microstructure on the electrochemical and optical stability of AlOxNy thin films, Electrochim. Acta 106 (2013) 23e34. [24] D. Barlev, R. Vidu, P. Stroeve, Innovation in concentrated solar power, Sol. Energy Mater. Sol. Cells 95 (2011) 2703e2725. [25] J. Gibkes, F. Vaz, A.C. Fernandes, P. Carvalho, F. Macedo, J.R.T. Faria, P. Kijamnajsuk, J. Pelzl, B.K. Bein, Analysis of multifunctional oxycarbide and oxynitride thin films by modulated IR radiometry, J. Phys. D Appl. Phys. 43 (2010) 395301. [26] M. Apreutesei, C. Lopes, J. Borges, F. Vaz, F. Macedo, Modulated IR radiometry for determining thermal properties and basic characteristics of titanium thin films, J. Vac. Sci. Technol. A 32 (2014) 041511. [27] N.P. Barradas, C. Jeynes, R.P. Webb, Simulated annealing analysis of Rutherford backscattering data, Appl. Phys. Lett. 71 (1997) 291e293. [28] N.P. Barradas, C. Jeynes, M.A. Harry, RBS/simulated annealing analysis of ironcobalt silicides, Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. Atoms 136e138 (1998) 1163e1167.

[29] A. Ruhm, B.P. Topeverg, H. Dosch, Supermatrix approach to polarized neutron reflectivity from arbitrary spin structures, Phys. Rev. B 60 (1999) 16073. [30] D.P. Almond, P. Patel, Photothermal Science and Techniques, Chapman & Hall, 1996. [31] F. Vaz, F. Macedo, R.T. Faria Jr., M. Torrell, A. Cavaleiro, K.H. Junge, B.K. Bein, Modulated IR radiometry applied to study TiO2 coatings with gold nanocluster inclusions, Int. J. Thermophys. 34 (2013) 1597e1605. [32] F. Macedo, F. Vaz, A.C. Fernandes, J.L.N. Fotsing, J. Gibkes, J. Pelzl, B.K. Bein, Thickness control of coatings by means of modulated IR radiometry, Plasma Process. Polym. 6 (2009) S592eS598. [33] F. Macedo, F. Vaz, M. Torrell, J.R.T. Faria, A. Cavaleiro, N.P. Barradas, E. Alves, K.H. Junge, B.K. Bein, TiO2 coatings with Au nanoparticles analysed by photothermal IR radiometry, J. Phys. D Appl. Phys. 45 (2012) 105301. [34] C. Lopes, C. Gonçalves, P. Pedrosa, F. Macedo, E. Alves, N.P. Barradas, N. Martin, C. Fonseca, F. Vaz, TiAgx thin films for lower limb prosthesis pressure sensors: effect of composition and structural changes on the electrical and thermal response of the films, Appl. Surf. Sci. 285 (2013) 10e18. Part A. [35] J.L. Nzodoum Fotsing, J. Gibkes, J. Pelzl, B.K. Bein, Extremum method: inverse solution of the two-layer thermal wave problem, J. Appl. Phys. 98 (2005) 063522. [36] J. Bolte, J. Gu, B. Bein, Background fluctuation limit of infrared detection of thermal waves at high temperatures, High. Temp. High. Press. 29 (5) (1997) 567e579. [37] F. Macedo, J. Ferreira, F. Vaz, L. Rebouta, A.H. Daoud, D. Dietzel, B.K. Bein, Photothermal characterization of sputtered thin films and substrate treatment, AIP Conf. Proc. 463 (1999) 536e538. € ren, A.C. Fernandes, F. Vaz, J. Gibkes, K.H. Junge, J.L. Nzodoum[38] F. Macedo, A. Go Fotsing, B.K. Bein, Potential of modulated IR radiometry for the on-line control of coatings, Plasma Process. Polym. 4 (2007) S857eS864. [39] D. Depla, R.D. Gryse, Influence of oxygen addition on the target voltage during reactive sputtering of aluminium, Plasma Sources Sci. Technol. 10 (2001) 547. [40] D. Depla, J. Haemers, R. De Gryse, Discharge voltage measurements during reactive sputtering of oxides, Thin Solid Films 515 (2006) 468e471. [41] D. Depla, R. De Gryse, Target poisoning during reactive magnetron sputtering: part I: the influence of ion implantation, Surf. Coatings Technol. 183 (2004) 184e189. [42] D. Depla, R. De Gryse, Target poisoning during reactive magnetron sputtering: part II: the influence of chemisorption and gettering, Surf. Coatings Technol. 183 (2004) 190e195. [43] D. Depla, S. Mahieu, R. De Gryse, Magnetron sputter deposition: linking discharge voltage with target properties, Thin Solid Films 517 (2009) 2825e2839. [44] D. Depla, S. Heirwegh, S. Mahieu, J. Haemers, R. De Gryse, Understanding the discharge voltage behavior during reactive sputtering of oxides, J. Appl. Phys. 101 (2007) 013301. [45] J. Schulte, G. Sobe, Magnetron sputtering of aluminium using oxygen or nitrogen as reactive gas, Thin Solid Films 324 (1998) 19e24. [46] R. Mientus, K. Ellmer, Reactive DC magnetron sputtering of elemental targets in Ar/N2 mixtures: relation between the discharge characteristics and the heat of formation of the corresponding nitrides, Surf. Coatings Technol. 116e119 (1999) 1093e1101. [47] S. Venkataraj, D. Severin, R. Drese, F. Koerfer, M. Wuttig, Structural, optical and mechanical properties of aluminium nitride films prepared by reactive DC magnetron sputtering, Thin Solid Films 502 (2006) 235e239. [48] D. Depla, S. Mahieu, Reactive Sputter Deposition, Springer, 2008. [49] V.S. Smentkowski, Trends in sputtering, Prog. Surf. Sci. 64 (2000) 1e58.
n, I. Katardjiev, S. Berg, W. Moller, TRIDYN simulation of target [50] D. Rose poisoning in reactive sputtering, Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. Atoms 228 (2005) 193e197. [51] S. Mahieu, P. Ghekiere, D. Depla, R. De Gryse, Biaxial alignment in sputter deposited thin films, Thin Solid Films 515 (2006) 1229e1249. [52] I. Petrov, P.B. Barna, L. Hultman, J.E. Greene, Microstructural evolution during film growth, J. Vac. Sci. Technol. A Vac. Surf. Films 21 (2003) S117eS128. [53] C. Duquenne, M.P. Besland, P.Y. Tessier, E. Gautron, Y. Scudeller, D. Averty, Thermal conductivity of aluminium nitride thin films prepared by reactive magnetron sputtering, J. Phys. D Appl. Phys. 45 (2012) 015301. [54] M.K. Samani, X.Z. Ding, S. Amini, N. Khosravian, J.Y. Cheong, G. Chen, B.K. Tay, Thermal conductivity of titanium aluminum silicon nitride coatings deposited by lateral rotating cathode arc, Thin Solid Films 537 (2013) 108e112. [55] F. Macedo, F. Vaz, A.C. Fernandes, L. Rebouta, S. Carvalho, K.H. Junge, B.K. Bein, Thermal characterization of hard decorative thin films, Plasma Process. Polym. 4 (2007) S190eS194. [56] C. Lopes, C. Gonçalves, J. Borges, T. Polcar, M.S. Rodrigues, N.P. Barradas, E. Alves, E. Le Bourhis, F.M. Couto, F. Macedo, C. Fonseca, F. Vaz, Evolution of the functional properties of titaniumesilver thin films for biomedical applications: influence of in-vacuum annealing, Surf. Coatings Technol. 261 (2015) 262e271. [57] P. Pedrosa, D. Machado, J. Borges, M.S. Rodrigues, E. Alves, N.P. Barradas, N. Martin, M. Evaristo, A. Cavaleiro, C. Fonseca, F. Vaz, Agy:TiNx thin films for dry biopotential electrodes: the effect of composition and structural changes on the electrical and mechanical behaviours, Appl. Phys. A 119 (2015) 169e178.