Kinetics of high-temperature oxidation of (Ti,Ta)(CN)-based cermets

1 Kinetics of high-temperature oxidation of (Ti,Ta)(CN)-based cermets E. Chicardi*1,2, J. M. Córdoba1,3 and F. J. Gotor1 1Instituto de Ciencia de Materiales de Sevilla (CSIC-US), Américo Vespucio 49, 41092 Sevilla, Spain. 2Departamento de Ingeniería Metalúrgica y de Materiales, Universidad Técnica Federico Santa María, Valparaíso 1680, Chile. 3Departamento de Química Inorgánica, Universidad de Sevilla (US), Calle Prof. García González 1, 41012, Sevilla, Spain. ABSTRACT The kinetics of the high-temperature oxidation of titanium-tantalum carbonitridebased cermets with different Ti/Ta ratios was studied. Isothermal oxidation tests were conducted under static air for 48 h at temperatures between 700ºC and 1200ºC. The oxidation satisfied the parabolic kinetics, characteristic of the existence of a protective oxide layer. The apparent activation energy suggests the rate-controlling process during oxidation is the simultaneous inward and outward diffusion of oxygen and titanium, respectively, through the formed protective layer, consisting mainly of a rutile phase. A higher Ta(V) content in the rutile decreased the oxygen diffusivity due to the reduction of oxygen vacancy concentration. Keywords: A. Ceramic matrix composite; A. Titanium; B. SEM; C. Hot corrosion; C. Kinetic parameters; C. Oxidation. *Corresponding Author. Tel: +34 954489217. E-mail: [email protected]sic.es 2 1. Introduction. Cemented carbide cutting tools are widely used for room-temperature machining due to their exceptional toughness and damage tolerance under cyclic loading. However, they fail to meet the requirements for the most demanding processes, such as high-speed machining, due to their poor oxidation resistance at high temperatures (600– 1000 °C), which can be reached on the tool/chip interface while in service [1-4]. At high temperature, wear diffusion plays an important role and is a limiting factor due to the increased reactivity between contact materials (tool and workpiece). If oxidation processes also occur, an additional wear mechanism is activated that can dramatically decrease the service life of the tool. To minimize these wear mechanisms, suitable coatings are required. Cermets based on titanium carbonitride are powder-metallurgical composite materials that have drawn increasing attention in the machining industry [5-7], because their properties, such as high wear resistance, high chemical stability and good mechanical strength at high temperature, are well adapted to the requirements of specific machining processes such as high-speed finishing. The good mechanical performance of cermets at high temperature is also related to their good oxidation resistance. Especially promising is a new generation of cermets containing complex titanium-tantalum carbonitrides as the ceramic phase [8-12] that have been proven to possess outstanding oxidation resistance [13], satisfying the optimal requirements for many applications in the field of cutting tools. Hot corrosion usually refers to phenomena involving chemical reactions at high temperature, typically when a gaseous phase (i.e. oxygen) produces extended damage to the material. Hot corrosion occurs, for example, when a material is allowed to react with air at high temperatures for a given time interval. The extent of the damage produced on the material will be a function of the kinetics of the gas/solid reaction under the given experimental conditions. In this light, hot corrosion that also includes oxidation is seen as a destructive process to be prevented for clear technological and economic reasons [5]. Kinetics analysis is an important aspect of the study of oxidation processes because it deals with the measurement and parameterization of the process rates. The goal of kinetics analysis is to obtain reliable kinetics parameters, i.e., the apparent activation energy (Ea), the pre-exponential factor (A) and the mechanism or kinetic model, to allow later prediction of the thermal stability of materials. 3 Although the oxidation kinetics of cemented carbides have been studied by some researchers [3, 14-20], published reports focusing on the oxidation kinetics of cermets are scarce [21, 22]. Given the high oxidation resistance of (Ti,Ta)(C,N)-based cermets, it is of considerable interest to perform a kinetics study using thermogravimetric measurements of the isothermal oxidation of this type of cermets in the temperature range of 700-1200ºC, covering the high-temperature range reached during the most demanding cutting operations. 2. Experimental methods. A set of five cermets with nominal compositions of 80 wt% TixTa1-xC0.5N0.5 – 20 wt% Co and variable Ta content, x = 1, 0.99, 0.95, 0.90 and 0.80, were synthesized by the mechanochemical process termed Mechanically Induced Self-Sustaining Reaction (MSR) from Ti, Ta, C and Co powder mixtures and N2 gas [23]. A powder mixture containing 46.5 g elemental Ti (99% purity, <325 mesh, Strem Chemicals), Ta (99.6% purity, <325 mesh, Alfa-Aesar), C (as graphite, <270 mesh, Fe < 0.4%, Merck) and Co (99.8% purity, <100 mesh, Strem Chemicals), together with 13 tempered steel balls (d = 20 mm, m = 32.6 g), were put into a tempered steel vial (300 ml in volume) and milled using a modified planetary ball mill (P4, Fritsch) at a spinning rate of 400 rpm under 6 atm N2 gas (H2O and O2 < 3 ppm, Air Liquide). The planetary ball mill enabled the MSR reactions to be monitored by continuously measuring the pressure inside the vial [24]. When the MSR reaction associated with the synthesis of the TixTa1-xC0.5N0.5 carbonitride phase occurs, the temperature increases due to the release of heat from the exothermic formation reaction, which consequently increases the total pressure. The ignition time, defined as the critical milling time required to induce the MSR process, can then be determined from the spike in the recorded time-pressure data. After ignition, the milling is continued to ensure full conversion and homogenization. The ignition time was about 40-45 min for all cermets and the total milling time 75 min. The powdered cermets synthesized by MSR were shaped using a uniaxial press (2 tons, 5 min) and compacted by cold isostatic pressing (200 MPa, 10 min). The green bodies were then sintered at temperatures between 1450ºC and 1550ºC, optimized for each Ta/Ti ratio [23], for 60 min under a flowing Ar (H2O < 3 and O2 < 2 ppm, Linde) atmosphere in a horizontal tubular furnace (IGM1360 model No. RTH-180-50-1H, 4 AGNI) to obtain cylindrical cermets 13 mm in diameter and 9 mm in height. After sintering, the cermets consisted of the mentioned carbonitride phase (TixTa1-xC0.5N0.5) and two Ti-Ta-Co intermetallic compounds, namely, (TixTa1-x)Co2 and (TixTa1-x)Co, acting as binder instead of the initial Co. A complete study of phases and composition of cermets can be found in a previous work [23]. The cermets were labeled as Ti100, Ti99Ta1, Ti95Ta5, Ti90Ta10 and Ti80Ta20 according to the Ti and Ta atomic percent content in the ceramic carbonitride phase (x in TixTa1-xC0.5N0.5). The oxidation experiments were performed on parallelepiped specimens of these five cermets with similar surface areas (~ 94±0.5 mm2). The specimens were obtained by cutting from the centers of the initial cylinders, and the surfaces were carefully prepared to reduce the flaws (e.g., pores, cracks) by successive grinding and polishing steps until a mirror effect was obtained. Three diamond discs with maximum diamond particle size of 60, 40 and 25 m were used for grinding and diamond powder suspensions of 8, 4 and 1 m for polishing. The oxidation kinetics were studied based on thermogravimetric (TG) measurements using a CI Robal electrobalance (C.I. Electronics Ltd.) with a maximum weight allowed of 5 g and a sensitivity of 1 μg, which was attached to the support frame of a high-temperature vertical furnace (Severn Thermal Solutions Ltd, mod. SC448A). Specimens were placed on a platinum wire coil to maximize the contact of the surfaces with the surrounding atmosphere. The oxidations of the cermets were performed under isothermal conditions for 48 h in static air (pO2  0.21 atm), at temperatures ranging from 700ºC to 1200ºC (heating rate of 20 ºC/min and free cooling). The cermets were also oxidized in air at 1300 ºC for the time necessary to achieve complete oxidation. The polished cross sections of the oxidized parallelepipeds were examined by Scanning Electron Microscopy (SEM) performed on a Hitachi S-4800 SEM-FEG microscope coupled with an X-ray energy dispersive spectrometry (XEDS) detector (Quantax-EDS, Bruker Corporation). The SEM images obtained in secondary electron mode and XEDS-SEM mappings were used to observe the progress of oxidation and to determine the thickness of the oxide scale. To inspect the possible appearance of liquid phase at the temperatures used in the oxidation study, Differential Scanning Calorimetry (DSC) measurements were 5 performed using a TA Instrument SDT Q600 with a constant heating rate of 10°C/min and a flowing helium atmosphere. 3. Results and discussion. Figure 1 shows the evolution over time of the weight gain (in mg/cm2) as a result of the oxidation of the different cermets for all of the temperatures studied. The expected increase in the extent of oxidation with temperature was observed for all of the cermets, as previously reported [25]. Moreover, Figure 1 clearly illustrates the benefit of the presence of Ta in improving the oxidation resistance of the cermets. For all temperatures, the weight gain continuously decreased as Ti was increasingly substituted for Ta in the cermets. Note that at 700ºC, the oxidation was undetectable for cermets with medium and high Ta content (Ti95Ta5, Ti90Ta10 and Ti80Ta20). In contrast, at 1200ºC, the Ti100 and Ti99Ta1 cermets (cermets with low Ta content) reached complete oxidation before the end of the 48-h oxidation test. To highlight the improvement in the oxidation resistance of cermets as Ti is progressively replaced by Ta, the weight gain per unit area after 48 h of oxidation for all of the temperatures and cermets is shown in Figure 2a. Note, for example, that at 800ºC, the Ti100 cermet exhibited an oxidation weight gain of 18.6 mg/cm2, similar to the Ti80Ta20 cermet, but at 1200ºC it was oxidized (19.7 mg/cm2). Therefore, substituting 20 at% of Ti with Ta is as if oxidation were deferred by approximately 400ºC. The outstanding decrease in the extent of oxidation with increasing Ta content in cermets was also clearly observed when the cross-sections of oxidized cermets were inspected by SEM after 48-h oxidation tests at different temperatures. The thickness of the region affected by oxidation was measured from the SEM images with the help of XEDS-SEM mappings, allowing the precise location of the interface between oxidized and unoxidized regions. The values are shown in Figure 2b, which are consistent with the thermogravimetric results of Figure 2a. Some examples of characteristic SEM images and XEDS-SEM mappings are shown in Figure 3, from which the effect of the Ta addition and the oxidation temperature on the extent of oxidation can also be determined. 6 As detailed in previous works [13, 26], phase identification by XRD and XEDSSEM analyses showed for all of the cermets and temperatures that the oxide scale was dominated by the presence of rutile and the progression of the oxidation by the growth of the rutile layer, even though other Co-containing phases such as cobalt oxides (CoO and Co3O4) and cobalt titanates (CoTiO3, Co2TiO4 and CoTi2O5) were also observed depending on the temperature and Ta content (Figure 4). For the cermets Ti100 and Ti99Ta1, the formation in the oxide scale of a thin CoO/Co3O4 external layer was observed at low temperature that reacted with the underlying rutile phase to form cobalt titanates at increasing temperature. The disappearance of this CoO/Co3O4 external layer induced the emergence of the rutile phase at the surface of the oxide scale. For the cermets with higher Ta content (Ti95Ta5, Ti90Ta10 and Ti80Ta20), the formation of a cobalt titanate external layer instead of the CoO/Co3O4 layer was observed at low temperature and remained stable and unaltered at increasing oxidation temperatures. Note that no open porosity was observed by SEM in the outer cobalt-rich layer, which was always highly compact. Although the rutile layer was relatively well-densified, large pores and/or voids were detected that were presumably generated by the coalescence of smaller ones formed because of the volatile species arising from the oxidation of the carbonitride phase. These species must remain partially occluded in these large pores. Moreover, below the compact oxide scale, an internal oxidation region composed of rutile and Co was also observed (Figure 5). Note that the thickness shown in Figure 2b comprises both the compact oxide scale and the internal oxidation region. Figures 3a-f show good adhesion of the oxide scale, without any observed layer breakdown or spalling phenomena. The SEM images also revealed homogeneous oxidation of the cermets as the oxidation depth was similar on all sides of the specimens. Cermets maintained their original shape after the oxidation tests, and only an increase in volume was observed in agreement with the volume changes associated with oxidation. For example, the oxidation of the major carbonitride phase into rutile causes a volume increase of about 50%. The XRD diagrams of the fully oxidized specimens at 1300ºC revealed for all of them the presence of titanium-tantalum oxide with rutile structure (TixTa1-xO2) and cobalt dititanate (CoTi2O5). Assuming TixTa1-xO2 and CoTi2O5 to be the fully oxidized species, the relative weight balances associated with oxidation were calculated for each cermet and temperature and are shown in Figure 7 2c. An oxidation extent greater than 100% observed in those cermets with a higher degree of oxidation confirmed the fact that volatile species remained occluded in the closed porosity observed in SEM images (Figure 3a-f). Figures 1 and 2 show the anomalous behavior at 1200ºC of the Ti95Ta5 and Ti90Ta10 cermets. A higher degree of oxidation than expected was observed for both cermets, as well as a linear oxidation behavior during approximately the first 6 h of oxidation (Figure 1). This behavior was attributed to the incipient formation of a liquid phase, corresponding to the partial melting of the intermetallic binder phase, according to the Ti-Co and Ta-Co binary phase diagrams [27]. Figure 6 shows the external surface of the Ti95Ta5 cermet oxidized at 1200ºC, where a globular morphology characteristic of the appearance of a liquid phase can be observed. Moreover, the DSC curve of this cermet, shown in the inset of the Figure 6, revealed in this temperature range an endothermic peak associated with the formation of a liquid phase. The presence of this liquid phase could facilitate oxygen diffusion into the cermets, resulting in increased oxidation. In the Ti80Ta20 cermet, this anomalous behavior was not observed due to the higher Ta content, which induces a higher melting point in the intermetallic binder phase. 3.1. Oxidation kinetics. The thermogravimetric curves in Figure 1 reveal that for all of the different cermets and temperatures, the oxidation rate tended to decrease with time. This evolution can be successfully fitted (Figure 1) according to the following general kinetics equation (1): Ctk s mn n       (1) where Δm is the weight gain; s is the cermet surface area; t is the oxidation time; kn is the oxidation rate constant; n is the oxidation exponent; and C is a constant that includes the oxidation prior to reaching the isothermal condition. The values of n determined by fitting the experimental data using equation (1) are shown in Table 1 along with the coefficient of determination (R2) as an indication of the goodness of fit. The n values obtained are in most of cases in the range 0.4-0.6, fully compatible with parabolic oxidation kinetics (n = 0.5). A larger deviation from 8 parabolic behavior was observed in the Ti95Ta5 and Ti90Ta10 cermets oxidized at 1200ºC, due, as mentioned above, to the presence of a liquid phase, which tends to accelerate the oxidation. Linear, logarithmic and parabolic kinetics are simple models that try to explain the oxidation behavior of pure metals. When a pure metal is exposed to oxidation, the resulting oxide layer generally covers the metal surface. Linear oxidation kinetics, corresponding to n = 1 in equation (1), involve the formation of a non-protective oxide layer because of, for example, the presence of pores or cracks in the layer or volatile oxide products. In this case, the metal is permanently in contact with oxygen, and the oxide layer growth (or metal oxidation rate) is constant over oxidation time and independent of its thickness. The oxidation progress is independent of the amount of gas or metal previously consumed in the reaction and the oxidation rate is controlled by the chemical reaction itself. In contrast, parabolic oxidation kinetics (n = 0.5 in equation (1)) imply the formation of a protective oxide layer whose growth rate is inversely proportional to its thickness; that is, the growth rate of the protective layer decreases with the oxidation time. Parabolic kinetics assume a diffusion-controlled process in which diffusion through the oxide layer of oxygen to the oxide/substrate interface or of metal to the oxygen/oxide interface is the rate-limiting oxidation process. It is also assumed that the concentration of reactive species at each boundary of the oxide layer is constant. Finally, logarithmic oxidation kinetics usually occur at low oxidation temperatures and are characterized by the formation of a thin oxide layer whose growth rate quickly decelerates, giving rise to a passivation phenomenon. Although previously oxidation studies typically report on more complex systems than pure metals [28-30], where even selective oxidation phenomena are observed [31], quasi-parabolic kinetic behavior (n values slightly different from 0.5) is often observed [32-34]. Deviations are common because even ideal models cannot perfectly describe real situations, in which fast diffusion paths at localized positions, such as grain boundaries or lattice defects or other factors, simultaneously influence the oxidation process. Nevertheless, quasi-parabolic kinetics is always associated with the existence of a protective oxide layer. 9 Therefore, the parabolic kinetics observed in the oxidation of cermets, supported by the good linearity of the plot of the specific weight gain vs. t0.5 (Figure 7), clearly indicate the formation of a compact and continuous oxide layer on the surface of the cermets, as shown for example in Figure 4 for cermets Ti99Ta1 and Ti95Ta5 oxidized at 800ºC and 1100ºC, respectively. Note that the plots corresponding to the highest tested oxidation temperature for each cermet are not presented in Figure 7 due to poor linearity, most likely as a consequence of the aforementioned partial melting of the binder phase. The parabolic rate constant, kp, for each temperature and cermet was calculated from the slope of the linear fittings in Figure 7, and the values are shown in Table 2. It can be clearly observed how the parabolic rate constant for the same oxidation temperature depends strongly on the Ta content. This result corroborates the effectiveness of Ta addition in improving the oxidation resistance of cermets. As an example, kp for the Ti100 cermet at each temperature is approximately two and three orders of magnitude higher than for the Ti95Ta5 and Ti80Ta20 cermets, respectively. Moreover, as for any thermally activated kinetic process, the values of kp (Table 2) reveal a strong influence of temperature, increasing by approximately a factor of ten every 100ºC. This dependence of kp on the oxidation temperature can be expressed by the Arrhenius equation (2) [35]:        TR E p a Aek (2) where A is the pre-exponential or frequency factor; Ea is the apparent activation energy; R is the universal gas constant; and T is the absolute temperature. The apparent activation energy of the oxidation process was calculated from the slope of the plot of ln(kp) vs. 1/T (Figure 8), and the values obtained are shown in Table 3. The Ea values found for cermets containing Ta (cermets Ti99Ta1, Ti95Ta5, Ti90Ta10 and Ti80Ta20) were in the range 225-256 kJ mol-1, in close agreement with the values reported in the literature for the activation energy of both the self-diffusion of oxygen and titanium in rutile, which have similar values [36-38]. Rutile generally exhibits significant non-stoichiometry, and the oxygen deficiency appears to be accommodated by oxygen vacancies and Ti interstitials, with oxygen vacancies 16 Table 1. Values of the oxidation exponent (n) and coefficient of determination (R2) obtained by fitting the curves shown in figure 1 using equation (1) for cermets Ti100, Ti99Ta1, Ti95Ta5, Ti90Ta10 and Ti80Ta20 at oxidation temperatures between 700ºC and 1200ºC. T (ºC) Cermet Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 n (R2) n (R2) n (R2) n (R2) n (R2) 700 0.61 (0.999) 0.41 (0.994) - - - 800 0.57 (0.999) 0.47 (0.999) 0.43 (0.990) 0.39 (0.996) 0.41 (0.965) 900 0.57 (0.999) 0.58 (0.999) 0.47 (0.997) 0.47 (0.998) 0.50 (0.965) 1000 0.54 (0.996) 0.57 (0.999) 0.61 (0.999) 0.50 (0.999) 0.52 (0.997) 1100 0.40 (0.984) 0.43 (0.999) 0.55 (0.999) 0.58 (0.999) 0.51 (0.999) 1200 - - 0.28 (0.929) 0.30 (0.960) 0.37 (0.989) 17 Table 2. Values of the parabolic rate constant (kp) obtained from the slope of the linear fitting in figure 7 and the coefficient of determination (r) for cermets Ti100, Ti99Ta1, Ti95Ta5, Ti90Ta10 and Ti80Ta20 at oxidation temperatures between 700ºC and 1200ºC. Cermet Tox (ºC) kp (mg2 cm-4 s-1) r Ti100 700 2.30 x 10-4 0.998 800 2.22 x 10-3 0.999 900 1.44 x 10-2 0.999 1000 4.22 x 10-2 0.999 Ti99Ta1 700 1.72 x 10-5 0.997 800 1.78 x 10-4 0.999 900 1.45 x 10-3 0.998 1000 1.29 x 10-2 0.999 Ti95Ta5 800 2.14 x 10-5 0.992 900 9.55 x 10-5 0.999 1000 9.16 x 10-4 0.998 1100 7.83 x 10-3 0.999 Ti90Ta10 800 4.45 x 10-6 0.998 900 3.68 x 10-5 0.999 1000 3.88 x 10-4 0.999 1100 2.19 x 10-3 0.999 Ti80Ta20 800 1.93 x10-6 0.982 900 1.49 x 10-5 0.997 1000 1.16 x 10-4 0.998 1100 4.57 x 10-4 0.999 18 Table 3. Apparent activation energy (Ea) determined for the oxidation process from the Arrhenius equation (2) and the coefficient of determination (r) for cermets Ti100, Ti99Ta1, Ti95Ta5, Ti90Ta10 and Ti80Ta20. Cermet Ea (kJ mol-1) r Ti100 182 ± 10 0.9971 Ti99Ta1 225 ± 10 0.9979 Ti95Ta5 242 ± 17 0.9882 Ti90Ta10 256 ± 10 0.9985 Ti80Ta20 226 ± 7 0.9991 19 0 6 12 18 24 30 36 42 48 0 1 2 3 4 5 6700ºC Ti100 Ti99Ta1 t (h) m (mg/cm2) 0 6 12 18 24 30 36 42 48 0 4 8 12 16 20 m (mg/cm2) 800ºC t (h) Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 0 6 12 18 24 30 36 42 48 0 10 20 30 40 50 60 70 80 t (h) m (mg/cm2) 1000ºC Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 0 6 12 18 24 30 36 42 48 0 10 20 30 40 50 m (mg/cm2) t (h) 900ºC Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 0 6 12 18 24 30 36 42 48 0 20 40 60 80 100 120 t (h) m (mg/cm2) 1100ºC Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 0 6 12 18 24 30 36 42 48 0 20 40 60 80 100 m (mg/cm2) t (h) 1200ºC Ti95Ta5 Ti90Ta10 Ti80Ta20 Figure 1 700 800 900 1000 1100 1200 0 20 40 60 80 100 120 140 a) m48 (mg/cm2) T (ºC) Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 700 800 900 1000 1100 1200 0 1 2 3 e (mm) b) T (ºC) Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 700 800 900 1000 1100 1200 0 20 40 60 80 100 120 c) Ti100 Ti99Ta1 Ti95Ta5 Ti90Ta10 Ti80Ta20 Oxidation progress (%) T (ºC) Figure 2 63 μm 584 μm 1331 μm 999 μm 749 μm 261 μm Figure 3 500 μm b) Ti100_900ºC 500 μm c) Ti100_1000ºC 500 μm d) 500 μm e) Ti90Ta10_1200ºC 200 μm f) Ti80Ta20_1200ºC g) h) Oxygen 10 μm 10 μm 100 μm Ti100_700ºC a) Ti95Ta5_1200ºC 50 μm 10 μm a) CoO / Co3O4 Ti99Ta1_800ºC Ti99Ta1_1100ºC TiO2 TiO2 CoTiO3 10 μm 5 μm b) a) Ti95Ta5_800ºC Ti95Ta5_1100ºC TiO2 CoTiO3 Figure 4 6 μm 6 μm 6 μm 6 μm 6 μm O Co Ti Ta Figure 5 Figure 6 1050 1100 1150 1200 1250 1300 -15 -10 -5 0 5 10 15 Heat flow (mW) T (ºC) Endothermic 100 m