- Email: [email protected]
Effect of morphology on thermal conductivity of EB-PVD PYSZ TBCs
Surface & Coatings Technology 201 (2006) 2611 – 2620 www.elsevier.com/locate/surfcoat
Effect of morphology on thermal conductivity of EB-PVD PYSZ TBCs A. Flores Renteria a,⁎, B. Saruhan a , U. Schulz a , H.-J. Raetzer-Scheibe a , J. Haug b , A. Wiedenmann b a b
German Aerospace Centre (DLR) – Institute of Materials Research, Cologne, Germany Hahn-Meitner Institute, Structural Research, Department of Materials, Berlin, Germany Received 28 September 2005; accepted in revised form 4 May 2006 Available online 9 June 2006
Abstract Partially yttria stabilized zirconia (PYSZ) based thermal barrier coatings (TBC) manufactured by electron beam-physical vapour deposition (EB-PVD) protect turbine blades, working under severe service conditions in aero engines and stationary turbines. These coatings show a high strain tolerance relying on their unique morphology which is comprised of weakly bonded, preferred-oriented columns, voids between feather-like sub-columns and, finally, of intra-columnar closed pores. The results obtained in this work demonstrate that variation of the EB-PVD process parameters alters the resulting columnar morphology and porosity of the coatings. The physical properties and, most importantly, thermal conductivity, are greatly affected by these morphological alterations. This study investigates three morphologically different EB-PVD PYSZ TBC top coats in terms of the spatial and geometrical characteristics of their porosity and correlates those with the thermal conductivity values measured in as-coated state and after heat treatment at 1100 °C for 1 h and 100 h. Changes in the open and closed porosity caused by heat-treatment are characterized by small-angle neutron scattering (SANS), Brunauer–Emmett–Teller Method (BET) and scanning electron microscope (SEM). Correlation of shape and surface-area changes in all porosity types of the analysed coatings revealed that the thermal conductivity of these coatings is influenced primarily by size and shape distribution of the pores and secondarily by the pore surface-area available at the cross section perpendicular to the heat flux. © 2006 Elsevier B.V. All rights reserved. Keywords: Morphology of EB-PVD TBCs; Thermal conductivity; SANS and BET-analysis; Sintering
1. Introduction The temperature of the hot-gas generated in modern gas turbines exceeds the melting temperature of Ni-based superalloys used for the manufacture of turbine blades. In order to withstand this extreme thermal loading, turbine blades and other hot-structure components of stationary and airliner gas turbines are protected by Thermal Barrier Coatings (TBCs) based on Partially Yttria Stabilized Zirconia (PYSZ). Thus, the lifetime of turbine blades, as well as the engine efficiency, is increased. It is essential that TBCs show a low thermal conductivity, high thermal expansion coefficient, excellent thermal shock-resistance, as well as phase and morphological stability in long-term, high-temperature service.
⁎ Corresponding author. E-mail address: [email protected] (A.F. Renteria). 0257-8972/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2006.05.003
The state-of-the-art thermal barrier coatings consist of a PYSZ top-coat, a metallic (Pt/Al or NiCoCrAlY) bond coat (BC) and between these two, a thermally grown oxide (TGO) based on Al2O3. Top coats can be manufactured by electron beam physical vapour deposition (EB-PVD) or by air plasma spraying (APS). Both EB-PVD and APS top coats display highly porous microstructures which lowers the intrinsic thermal conductivity. Thus, the differences in the thermal conductivity values between EB-PVD and APS coatings are caused by the differences in shape, orientation and distribution of their porosity [1,2]. During EB-PVD processing of the coatings, quasi single crystalline columns of very fine diameter (b 2–3 μm) start to grow on the substrate surface. Their diameter increases at the coating tip to 10–20 μm. These columns are only weakly interconnected via inter-columnar gaps, which accounts for some of the open porosity. Inter-columnar gaps are of a few nanometres close to the substrate and can be as large as 1 μm in
2612
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
width at the coating tip. High thermal shock-resistance of the EB-PVD processed TBCs relies on the presence of these intercolumnar gaps [3–5]. Most of the open porosity originates from the voids present between nano-sized secondary columns, socalled feather-arms, growing at the column edges. These are formed during evaporation mainly by secondary shadowing, depending on rotation speed, substrate temperature and the angle between the substrate and the plane of vapour incidence [6,7]. Furthermore, due to the interruption of the vapour deposition during rotation, closed porosity is created within the primary columns in form of finer elongated channel-like pores (i.e. intra-columnar voids) (see Fig. 1). Exposure of TBCs to high temperatures under service conditions causes changes in the morphology of the columns by a series of thermal processes such as: formation of bridges between the columns, formation of sintering necks at contact points between the feather-arms, and eventually changes in pore geometry and sizes of the intra-columnar pores, generating a surface-area reduction. Morphology and density of the columnar microstructure of EB-PVD TBCs can be varied by changing the deposition parameters such as chamber pressure, substrate temperature and rotation speed [8–10]. In particular, the porosity between the feather-arm features in EB-PVD TBCs is recognized as an important factor controlling their thermal conductivity. Since most of the heat flows through the conducting columns and in-plane thermal gradient does not vary across the parallel aligned primary columns and inter-columnar gaps, it is considered that the inter-columnar gaps orientated parallel to the heat flux induce only a slight effect on the thermal conductivity [1,11]. Recent studies have reported controversial results related to this consideration. Instead, significant reductions in the thermal conductivity were identified by decreasing the coating thickness. This is correlated to the factors emerging from inclined growth of columns relevant to the direction perpendicular to substrate plane and to the presence of a non-textured crystal structure as well as a high fraction of inter-columnar pores at the bottom zone of EB-PVD coatings [12–14]. Nevertheless, micro-scale spatial resolution measurements of thermal conductivity via laser time-domain thermo reflectance (TDTR) showed that approximately 95% of the total reduction of the thermal conductivity compared with the bulk value was influenced by the presence of pores within the
columns (i.e. intra-columnar pores), although, thermal diffusivity remains constant along the coating thickness, despite the microstructural heterogeneity [15]. It is suggested by Hass et al. [1], Gu et al. [16] and Heydt et al. [17], that the effect of porosity on the thermal conductivity is due to the reduction of the cross sectional area through which the heat flows. Heat flux density is statistically a local representative of the porosity distribution, leading to the assumption that it is the alteration of porosity which influences the thermal conductivity, rather than the modification of phonon scattering per se. Basing on this theory, some studies varied the coating morphology by either interrupting the atomic deposition from the vapour by “in and out” feeding or “shuttering” of the substrates during the coating process [18], by creating a microstructure with a zigzag intercolumnar porosity [1,11,16], or by successive changing of the rotation speed during the deposition process [9,19,20]. Characterization of the pores in these coatings requires the employment of sophisticated techniques due to their sizes and size differences, accessibility (open and closed pores) and anisotropy. Use of innovative methods has been reported for precise and detailed evaluation of open and closed porosity changes on ceramic materials during service simulated conditions. These methods are based on the transmission of X-rays or neutrons through the coatings. Small-Angle Neutron Scattering (SANS) technique represents, with some advantages, an alternative to the gas adsorption methods. This method is sensitive to both closed and open porosities and, in many cases, offers a more complete analysis of the measured porosity [21]. SANS has been already applied to study the processing-microstructure relations and sintering of plasma sprayed coatings (APS) [22–27]. These studies, in particular, focused on measurement of Porod scattering [28] derived from the terminal slope in SANS spectra to determine the characteristics of voids in porous materials. The quantitative information obtained on the corresponding interface surface-area is dominated by the measured intensities. Since the fine features in a microstructure are the major contributors to the deduced area [25], it is important to measure the minimum size of the scatterers present in a material in order to ascertain the origin of the scattered apparent Porod surface-area. Porod scattering technique does not require a void shape model, and thus, is especially useful for studies of void systems with complex shapes [22].
Fig. 1. Scanning electron micrographs showing the type and distribution of the pores at the EB-PVD PYSZ TBCs.
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
This paper reports the BET, SANS and SEM analysis of the morphology and surface-area changes of the open and closed pores at EB-PVD TBCs with three different morphologies before and after heat-treatment. Additionally, the thermal conductivity of these coatings was measured in free-standing conditions by Laser Flash Analysis Method (LFA) before and after heat treatment at 1100 °C for 1 h and 100 h and was correlated with the microstructural observations. 2. Materials and methods 2.1. Processing and materials EB-PVD coatings were produced by employing “von Ardenne” pilot plant equipment with a maximum EB-Power of 150 kW. Evaporation was carried out from single source (ingot) having the standard chemical composition (7–8 wt.% Y2O3 stabilized ZrO2) and with the dimensions of 62.5 mm diameter and 150 mm length. Vapour phase is deposited on plane substrates under conventional rotating mode as described in [29] (i.e. by mounting the substrates on a holder with its axis perpendicular to the evaporation source). A thickness of approximately 400 μm is achieved by EB-PVD coating of PYSZ directly on FeCrAl-alloy substrates (12.7 mm diameter) without having a bond coat. To obtain different coating morphologies, the substrates were rotated at different speeds and heated to different temperatures during the coating process. Table 1 lists the sample designation and the applied process parameters for the manufacture of three investigated EB-PVD coating morphologies. These parameters are selected based on the previous studies of the group which have shown considerable differences in the porosity configuration of the coatings [9]. Since the applied analysis methods (e.g. BET, SANS and LFA) require the use of coatings in free-standing conditions, the substrates were etched out completely in an acid solution to obtain free-standing coatings. Subsequently, the corresponding specimens were heated to 1100 °C with a heating rate of 5 °C/ min and isothermally aged at this temperature in air for 1 h and 100 h. After the respective isothermal ageing, the samples were taken off from the furnace and “quenched” in air to freeze the microstructure. Additionally, free-standing coatings were coated on both sides with a thin Pt-layer (4–5 μm thickness) by sputter technique to avoid laser-beam penetration during the LFA-measurements. Thermal cycling of coatings attached on substrates can induce interface resistance at the underlying TGO/BC, adding to the total thermal conductivity value. This contribution cannot be separated from the top coat sintering related increase of thermal conductivity. Thermal expansion coefficient mismatch, in turn, is responsible for formation of Table 1 Designation of the samples and applied parameters during the EB-PVD process Designation for morphology
Chamber pressure Substrate temperature Rotation speed (mbar) (°C) (rpm)
Intermediate Fine Coarse
8 × 10− 3 4 × 10− 3 8 × 10− 3
950 850 1000
12 3 3
2613
stresses at the interface between BC/TGO/top coat [30]. The free-standing configuration of the specimens, in fact, “releases” the internal stresses which are otherwise created in a TBC system during cyclic or isothermal ageing [5]. However, these stresses mainly affect the dimensions of the inter-columnar gaps, which are, as a matter of fact, considered to affect only slightly the overall thermal conductivity of the coatings [1,11]. Considerable uncertainties may emerge from measuring the coatings on the metallic substrate due to micro-delamination of the top coats on ageing [31] and heat-treatment induced interface resistance, as well as, failures related to large differences such as thin top coat/ thick substrate configurations [32,33]. Therefore, we consider that the measurement of thermal conductivity on free-standing aged coatings is a satisfactory procedure to obtain representative values for the ceramic top coat and to reduce the uncertainties. 2.2. Methods of characterization 2.2.1. Thermal conductivity In the present work, the thermal diffusivity (α) of the specimens was measured via Laser Flash Analysis Method (LFA 427, Netzsch, Selb/Germany). The thermal conductivity [34] is calculated by applying the formula: k ¼ a d q d Cp
ð1Þ
where ρ, in this case, represents the bulk density of the freestanding coating measured by the Archimedes Method, and Cp is the specific heat measured by Differential Scanning Calorimeter (DSC). The accuracy of the calculated thermal conductivities is limited with the uncertainties of the factors related to the Eq. (1). In this work, the specific heat was measured via differentiated scanning calorimeter instrument (DSC-404, Netzsch) which encloses a systematic standard deviation of ± 3–5% according to the manufacturer. Moreover, the thermal diffusivities used in the calculations based on an average of three measured values for each temperature condition. Ultimately, the thickness of the specimens was measured via optical microscope and averaged on the basis of 10 measurements. The bulk density represents the average value of nine morphologically-identical coatings. During the thermal diffusivity measurements, the association of the curves obtained under heating and cooling conditions indicates the thermal history of the specimens. Since the as-coated EB-PVD morphology is highly temperature sensitive, a pre-treatment of the manufactured coatings at temperatures slightly above the deposition temperature is a general practice to equilibrate the microstructure. Since we aim to determine the effect of sintering on thermal conductivity, we have left out this pre-treatment and measured the coatings in as-coated conditions. Therefore, in our case, the thermal conductivity curves obtained on heating and cooling over the temperature range between RT and 1000 °C are different. 2.2.2. Morphology and surface area determination The microstructures were characterized visually by using a Field-Emission Scanning Electron Microscope (FE-SEM, LEITZ LEO 982, Germany). Since the columnar microstructure of these coatings contains open (inter-columnar gaps and voids between
2614
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
feather-arms) and closed porosities (inter-columnar pores), it was necessary to use, beside the microstructural characterization also such measurement methods which are capable of tracking the surface-area changes of all type of pores. A precise analysis between these thermally induced microstructure changes and the thermal conductivity can be possible, if surface-area of the open porosity is measured by BET and SANS. BET measurements were carried out at the Department of Ceramic Technologies and Sintered Materials of the Fraunhofer Institute in Dresden/ Germany using an Accelerated Surface Area Porosimeter System (ASAP 2010) and employing N2 as analysis gas. Since, the BET allows predominantly the analysis of the open porosity, the SANS measurement were also carried out for a detailed characterization of the total surface-area. SANS experiments were performed at the instrument V4 of the Berlin Neutron Scattering Centre (BENSC) in the Hahn–Meitner Institute, Berlin/ Germany (for detailed information see Ref. [35]). The scattered neutron intensity I(Q) was measured using an incident beam along the plane perpendicular to the substrate's plane (parallel to the column-axis) over two-sample detector distances: 1 m and 4 m, employing a λ = 0.605 nm neutron wavelength. During the measurements, the specimens were placed on the sample holder by means of a cadmium mask having a circular slit of 6 mm diameter. The obtained data were corrected for background scattering, transmission and detector efficiency, and normalized to absolute units by using water as calibration standard procedure. Quantitative information of the surface-area corresponding to the intra-columnar pores was achieved in the form of the corresponding Porod Constant (Pc) as a function of the azimuthal angle by converting the 2-D data to 1-D scattering profiles averaged over 5°. The Porod Constant (Pc) is directly proportional to the Apparent Porod Surface Area (Sv) [36,37] according to: Sv ¼
Pc 2kjDqj2
ð2Þ
where, Δρ is the contrast in scattering length density between the pores and the PYSZ material. In this method, the measured Porod
scattering intensity I(Q) is linearly proportional to the interfacial surface-area projection in a plane perpendicular to the scattering vector Q, corresponding to dimensions measured parallel to it [28,38]. In view of the fact that the Porod scattering measured in one direction does not reflect the total surface-area of the scatterers but a projection in the plane perpendicular to the scattering vector Q, it is often called Apparent Porod Surface Area, which can be calculated from its corresponding Porod constant by Eq. (2). The total specific area of the scatterers within the sample can be obtained from the apparent Porod constant by averaging over all 3D-directions in the material [39]. However, this apparent surface-area is useful to analyse microstructural changes between samples by measuring them under the same conditions. 3. Results and discussion Fig. 2 shows SEM micrographs of the investigated EB-PVD PYSZ top coats having morphological differences in their column diameter, in the size and shape of inter-columnar gaps and of voids between feather-arm features. These differences were produced by altering the process parameters during the EBPVD coating as specified in Table 1. The morphologies were designated according to their appearances on the micrographs. The “intermediate” microstructure, produced under fast rotation-speed, shows symmetrical columns reaching to approximately 10 μm diameter at the tip zone of coating (bulk density: 4.37 g/cm3 ± 3%). So-called feather-arm features surrounding the columns are elongated and extend from the column edge to the column axis in approximately 1/4 depth of their length (see Fig. 2a). The microstructure designated as “fine” was produced by employing a lower substrate temperature and slower rotation speed and exhibits non-symmetrical columns having ≤5 μm tip diameter which are surrounded with very short feather-arms (bulk density: 4.50 g/cm3 ± 3%) (see Fig. 2b). The microstructure designated as “coarse” was produced by employing a higher substrate temperature combined with slower rotationspeed and displays broader inter-columnar gaps and larger columns with tip diameters of around 20 μm (bulk density:
Fig. 2. Scanning electron micrographs of three different as-coated EB-PVD-morphologies of PYSZ-TBCs in cross section perpendicular to the rotation-axis: “intermediate” (a), “fine” (b) and “coarse” (c); and from the coating tip: “intermediate” (d), “fine” (e) and “coarse” (f ). (d) to (f ) are arranged such that the rotation-axis during deposition is vertical.
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
4.65 g/cm3 ± 5%). The feather-arms of this morphology are also elongated and extend from the column edge to the column axis in approximately 1/7 depth of their length (see Fig. 2c). High resolution SEM micrographs of the manufactured EBPVD TBCs in the Fig. 3 show the coating cross sections deposited perpendicular to the rotation axis. The geometrical and spatial characteristics of the voids between feather-arms and intra-columnar pores display changes at each morphology after ageing. The micrograph in the Fig. 3a1 displays the morphology of the “intermediate” microstructure. In this coating, the voids
2615
between feather-arms are approximately 0.07–0.10 μm wide and 2–2.5 μm long. In addition, this coating contains arrays of intra-columnar pores which form after every rotation movement and rowed with a distance of 0.25–0.30 μm from each other (see Fig. 3d1). The voids between feather-arms at the “fine” microstructure are in turn approximately 0.01–0.05 μm wide and 0.50–0.80 μm long (see Fig. 3a2). The distance between the intra-columnar pore arrays is broader (1.2–1.6 μm) compared to the “intermediate” microstructure due to the slower rotation speed applied during
Fig. 3. Scanning electron micrographs of cross sections perpendicular to the rotation-axis of the three different EB-PVD-morphologies of PYSZ-TBCs, showing the inter-columnar pores and pores between feather-arms in as-coated condition: “intermediate” (a1), “fine” (a2) and “coarse” (a3). After heat treatment at 1100 °C for 1 h: “intermediate” (b1), “fine” (b2) and “coarse” (b3) and after ageing at 1100 °C for 100 h: “intermediate”(c1), “fine”(c2) and “coarse”(c3). Additionally, the intra-columnar pores in as-coated condition are shown: “intermediate” (d1), “fine” (d2) and “coarse” (d3); after ageing at 1100 °C for 1 h: “intermediate” (e1), “fine” (e2) and “coarse” (e3) and after ageing at 1100 °C for 100 h: “intermediate” (f1), “fine” (f2) and “coarse” (f3).
2616
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
deposition of the “fine” microstructure (see Fig. 3d2). Among the three manufactured microstructures, the microstructure designated as “fine” possesses the highest density of intra-columnar pores, which clearly demonstrates that at the corresponding substrate temperature (e.g. 850 °C, T / TM = 0.37), the effect of shadowing predominates to diffusion during the film growth. Analogous results are reported by Cho et al. [7] who carried out kinetic Monte Carlo simulation studies on film EB-PVD deposition during substrate rotation. Finally, the voids between feather-arms of the “coarse” microstructure are approximately 0.07–0.15 μm wide and 2–3 μm long (see Fig. 3a3). The distance between the intra-columnar pore arrays is also broader (1.6–2.0 μm) compared to the “intermediate” microstructure (see Fig. 3d3). Despite the same applied rotation speed, the distances between the intra-columnar pore arrays of the “fine” and “coarse” microstructures differ slightly. This difference correlates well with the deposition rate difference of these two microstructures (see Table 2). Elongated intra-columnar pore arrays form at the regions where vapour incidence angle (VIA) is the highest and the regrowth of coating starts on each new rotation. Some of these pores extends themselves through the depositing material in a sunrise–sunset pattern which is influenced by the tip of the neighbouring columns as the VIA direction changes. Thus, the elongated “banana”-shaped pores are formed. Fig. 3d1, d2 and d3 show micrographs of intra-columnar pore arrays with corresponding elongated “banana”-shape pores. The “banana”shaped pores belonging to the “fine” microstructure possess the highest aspect-ratio (Fig. 3d2). This may be a result of the slow rotation-speed. But also, the low substrate temperature may hinder the atomic diffusion, thus enabling the formation of elongated thin pores. In turn, the “coarse” microstructure shows less of low aspect-ratio intra-columnar pores. The rotation-speed applied in the manufacture of the “fine” and “coarse” microstructures are the same, however, substrate temperatures are different (Tcoarse N Tfine). The diffusion processes become more active at the higher substrate temperature of the “coarse” microstructure (Fig. 3d3). The “intermediate” microstructure displays a higher number, low aspect ratio intra-columnar pores due to considerably higher rotation-speed (Fig. 3d1). Since the formation of intracolumnar pores occurs repeatedly on each new plane of the coating deposition, they contribute greatly to the surface-area. The graph in Fig. 4 shows absolute values of surface-area changes measured by BET for all investigated coatings. The surface-area changes represent solely the open pores such as inter-columnar gaps and the voids between feather-arms. It is clearly visible at the as-coated condition that the “fine” microTable 2 Microstructural characteristics of the analysed EB-PVD TBCs Designation for morphology
Deposited material (μm/rotation)
Deposition rate (μm/min)
Width of columns (μm)
Voids between feather-arms Length (μm)
Width (μm)
Intermediate Fine Coarse
0.25–0.30 1.20–1.60 1.60–2.00
5.60 5.40 6.70
10.00 ≤ 5.00 20.00
2.00–2.50 0.50–0.80 2.00–3.00
0.07–0.10 0.01–0.05 0.07–0.15
Fig. 4. BET-surface-area values (m2/g) of the “intermediate”, “fine” and “coarse” EB-PVD PYSZ TBC microstructures measured in as-coated condition and after ageing at 1100 °C for 1 h and 100 h.
structure contains the highest surface-area being well in agreement with its higher quantity of columns/inter-columnar gaps and voids between feather-arms per volume unit of material. The lowest surface-area for open pores was measured at the “coarse” microstructure. The polar figures in Fig. 5 exhibit the distribution of the surface-area of the voids between feather-arms and the closed intra-columnar pores measured in the direction parallel to the column axis (i.e. perpendicular to the substrate plane) for all three analysed coatings. The surface-area measured for the “fine” microstructure possesses the highest value while that for the “coarse” the lowest, matching well with the BET-measurement. It is evident from Fig. 5 that in the as-coated condition, the distribution of the surface-area for all three analysed microstructures is anisotropic, being that in the direction parallel to the rotation axis highest. This anisotropy observed at the Porod scattering indicates an alignment in the shape and orientation of the individual pores as well as in their distribution. In order to determine which of these factors influences mostly the anisotropy of the surface-area distribution, it is necessary to analyse the specimens in another direction (i.e. parallel to the substrate plane). Moreover, Multiple Small-Angle Neutron Scattering (MSANS) method for instance enables determination of the dimensions and volume fraction of the voids [23,40] which is beyond the scope of this work. LFA analyse on as-coated specimens indicate that the thermal conductivity of the “coarse” microstructure is approximately 18 and 14% (at 800 °C) and, thus, higher than that of the “intermediate” and “fine” microstructures, respectively (see Figs. 6 and 7). Considering the fact that the porosity affects the thermal conductivity by reducing the cross sectional area through which the heat flows, the obtained thermal conductivity results correspond well to the SANS and BET measured surface-area values which represent the total open and closed pores. Since the intracolumnar pores in the “fine” microstructure have the highest aspect ratio and are inclined respective to the heat-flux, these may induce the strongest reduction in the thermal conductivity being in accordance to the observation of Lu et al. [41].
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
2617
Fig. 5. Porod Constant values measured in as-coated and after heat treatment at 1100 °C for 1 h and 100 h, as a function of the azimuthal angle being plane parallel to the column axis of “intermediate”(a), “fine”(b) and “coarse”(c) EB-PVD PYSZ TBC microstructures.
After heat treatment at 1100 °C for 1 hr and 100 h, irreversible changes occur at all three analysed microstructures. Previous BET and SANS investigations carried out to determine the morphological changes of EB-PVD TBCs having “intermediate” microstructure after heat-treatment showed an abrupt change in the surface-area after 1 h at 1100 °C for all pore types due to sintering [42]. In the case of the voids between featherarms, such changes may occur firstly at contact points forming along their lengths finally resulting in aligned pore arrays with fine quasi-spherical and slightly elongated shapes (see Fig. 3b1). This process results in a noticeable reduction of the surface-area representing the open porosity (see Fig. 4). In the case of the “fine” microstructure, the closure of voids between feather-arms occurs without the formation of contact points, since these are comprised of very short lengths (see Fig. 3b2). Nevertheless, an abrupt reduction in the BET-surface-area has been observed, which may in fact correspond to the high number of intra-columnar pores present in this microstructure (Fig. 4). In the case of the “coarse” microstructure, a slight change occurs in the surface-area and shape of the pores. Closure of the voids between featherarms may follow the same manner at this microstructure as at the “intermediate” microstructure. The elongated intra-columnar closed pores of the “coarse” microstructure are wider, and thus, the formation of contact points occurs less frequently (see Fig. 3b3). Heat treatment at 1100 °C for 1 h causes a considerable reduction in the Apparent Porod Surface Area (Porod Constant, Pc) of the intra-columnar closed pores for all three studied microstructures (see Fig. 5). This may be due to the break-up in their elongated shape forming arrays of quasi-spherical pores (see Fig. 3e1, e2 and e3). The “fine” microstructure yields the highest reduction of the surface-area while the “coarse” the lowest, being in accordance with the higher surface-area of the “fine” microstructure yielding higher surface energy. Nevertheless, the surface-area distribution at these two analysed microstructures remains anisotropic even after heat-treatment (see Fig. 5b and c). The surface-area distribution at the “intermediate” coating becomes isotropic after heat-
treatment (see Fig. 5a), indicating a random distribution of the pores and/or formation of spherical shaped individual pores. Prolonged heat treatment of all coatings for 100 h at the same temperature (e.g. 1100 °C) induces, in turn, only a slight change at their surface-area in relation to that observed after 1 h of heattreatment. In summary, it can be said that sintering-related, morphological changes occur very fast at 1100 °C and are mostly completed already after 1 h. Fig. 6 shows the variation of the thermal conductivity values of the three investigated coatings as a function of measuring temperature before and after heat-treatment at 1100 °C for 1 h and 100 h. The determination of thermal conductivity values is carried out with 400 μm thick free-standing EB-PVD-coatings. Typically, these coatings have 50 μm thick bottom part which consists of finer columns. The study which was carried out to determine the thickness dependency of thermal conductivity showed lower conductivity values for the coating having only 50 μm thickness [12]. The influence of this fine columned area on the thermal conductivity value becomes less as the coatings thickness increases over 200 μm. The thermal conductivity curves of the as-coated “intermediate” and “fine” microstructures show no apparent temperature dependency. In contrast, a sharper decay below 400 °C has been observed at the “coarse” microstructure (Fig. 6). Such an effect is attributed to different causes in the literature [43]. Slifka and Filla [44] has reported that the porosity affects the diffusion of long-wavelength phonons leading such differences. This behaviour cannot be observed at the curves obtained from the samples aged for 1 h at 1100 °C, confirming the relation of thermal conductivity with the morphological changes occurring in the shape and surface-area of the porosity present at these coatings. Moreover, thermal conductivity values of the ascoated specimens increase at temperatures higher than 900 °C. According to our previous work, this effect occurs due to heatinduced changes in the microstructure (i.e. sintering) [42]. After heat treatment at 1100 °C for 1 h, the thermal conductivity of the “intermediate” microstructure measured at
2618
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
There are noticeable increases in the thermal conductivity of three analysed coatings after heat treatment. Assuming that the inter-columnar pores orientated parallel to the heat flux induce only a slight effect on the thermal conductivity of EB-PVD TBCs, these differences can be principally related to the thermally induced changes of the voids between feather-arms and intra-columnar pores. Although the “intermediate” microstructure shows the highest thermal conductivity increase, the total surface-area reduction is not the highest at this microstructure. Observation of the corresponding micrographs given in Fig. 3(b1 to f1) reveals that the intra-columnar pores as well as the inner part of the feather-arms voids break into quasi-spheres after 1 h ageing at 1100 °C (see Fig. 3b1 and e1). This observation agrees well with the consideration of Lu et al. [41] and Cernuschi et al. [45], confirming that spheres, as well as cylinders and plates oriented parallel to the heat-flux, provide the lowest effect on the thermal conductivity reduction. On prolonged ageing (e.g. 100 h at 1100 °C), the aspect-ratio of the voids between feather-arms are extremely reduced and this leads solely to a slight increase in thermal conductivity (see Figs. 3c1, 7). As these drastic changes occur at the feather-arm features of the “intermediate” microstructure, the shape of the intra-columnar pores does not change much (i.e. remains quasi-spherical) (see Fig. 3e1 and f1). In the case of the “fine” microstructure, the elongated intracolumnar pores break into small spheres on ageing (Fig.3b2 and e2). However, the majority of intra-columnar pores, which were initially formed at regions of highest vapour incidence angle (VIA) retain their elongated shape despite the heat-treatment (see Fig. 3e2 and f2). Being specific to this microstructure, the ageing induced changes in the shape of the voids between featherarms do not affect the thermal conductivity since their length stretching inside the columns is very short (see Fig. 3b2 and c2). Due to these reasons, the increase in the thermal conductivity after ageing is lower. Finally, for the “coarse” microstructure, the intra-columnar pores as well as the voids between feather-arms change their morphology slightly after 1 h ageing at 1100 °C. The parts of the
Fig. 6. Thermal conductivity values of the “intermediate”(a), “fine” (b) and “coarse” (c) EB-PVD PYSZ TBC microstructures measured in as-coated condition and after ageing at 1100 °C for 1 h and 100 h.
800 °C increases approximately 10% and that of the “fine” microstructure approximately 4% reaching nearly to the value achieved after heat treatment at 1100 °C for 100 h (e.g. 13% and 6% increase, respectively). In the case of the “coarse” microstructure, the total surface-area changes slightly after 1 h and 100 h heat treatment at 1100 °C. Increase in the heat diffusivity is also very small, roughly corresponding to that of the experimental deviation (measured at 800 °C approximately 2% and 6% higher, respectively) (see Fig. 7).
Fig. 7. Comparison of thermal conductivity values of the “intermediate”, “fine” and “coarse” EB-PVD PYSZ TBC microstructures in as-coated condition and after ageing at 1100 °C for 1 h and 100 h.
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
intra-columnar pores formed at regions of highest vapour incidence angle (VIA) maintain their original elongated shape to some extent. However, some of the elongated intra-columnar pores which form within the plane of the last deposited material break into arrays of fine spheres (see Fig. 3e3). The voids between feather-arms hold their high aspect-ratio in spite of thermal ageing (Fig. 3b3). Only a slight change in the thermal conductivity values of this microstructure is observed after ageing which can be correlated with the above mentioned microstructural stability. However, if the ageing time is prolonged to 100 h at the same temperature, the shape of the intra-columnar pores and voids between feather-arms changes to quasispheres yielding a reduction in their aspect-ratio and, consequently, causing a higher increase in the thermal conductivity (see Fig. 3c3 and f3). In summary, slow-rotation and low-substrate temperature leads to the formation of high aspect ratio feather-arms, long elongated, banana-shaped intra-columnar pores and highlypopulated pore arrays. Such microstructure yields a very low thermal conductivity. Although it suffers under thermal ageing resulting in high surface-area reduction, the formation of new elongated closed pores located perpendicular to heat flux yields an increase in thermal conductivity. High substrate temperature, despite slow rotation-speed results in intra-columnar pore arrays with broader distances, low aspect-ratio, banana-shaped pores and high aspect-ratio feather-arm features. Such microstructure, in turn, yields a rather high thermal conductivity which, on prolonged ageing, increases considerably. Fast-rotation speed produces shorter distances between the spherical, intra-columnar pore arrays and high aspect-ratio feather arms. This microstructure has a moderate thermal conductivity in as-coated condition which, however, increases considerably on ageing already after 1 h. Formation of large spheres on break-up of feather-arms may be the reason for this since they have almost no contribution in hindering heat-flux. Under thermal cycling conditions, thermal conductivity changes may be slightly different. However, in the cases when the cycle length is 1 h, the behaviour must be similar to our observations, since the most of the thermally induced changes occur in these morphologies within the first hour at 1100 °C. To ensure a stable low thermal conductivity, EB-PVD process parameters should be selected such that the resulting coating contains a microstructure having a stable pore configuration after heat-treatment, regardless of their form in as-coated condition. Furthermore, there is a need for precise characterization of the specific spatial and geometrical characteristics of each pore type at these coatings. Comprehensive studies have been carried out by employing USAXS measurements and corresponding modelling to obtain the related individual effect on the thermal conductivity and publication of the results is underway.
2619
to each other in studying the changes in the fine porous structures of EB-PVD PYSZ TBCs before and after heat treatment, allowing a detailed understanding of the interplay between the geometry, orientation and distribution of the pores and their thermal conductivity. The thermal conductivity of manufactured coatings was analysed before and after ageing. There are clear differences in the thermal conductivity values of the three as-coated morphologies. After ageing at 1100 °C, due to sintering-related phenomena, the changes in shape and distribution of the pores occur, resulting in reduction of surface-area and increase of thermal conductivity. The increase in thermal conductivity after heattreatment can be related to the alteration of the shape of the pores rather than the reduction of their surface-area at the cross section through which the heat flows. The “fine” microstructure shows on ageing the highest reduction in the total surface-area covering open and closed porosity but not the highest increase in thermal conductivity. While the “intermediate” microstructure displays the highest increase in thermal conductivity, in spite of a moderate surface-area reduction. This can be because of short lengths of voids between feather-arms in this microstructure. Thus, it can be assumed that its thermal conductivity is basically affected by the changes in the intra-columnar pores which retain their elongated shape relatively well after applied ageing. It is observed that the intra-columnar pores which have initial broader shape and large size characteristics are resistant to sintering. The voids which break into spheres on ageing yield still the best size distribution in reducing the thermal conductivity. In the “intermediate” microstructure, on ageing, the intra-columnar pores as well as voids between the feather-arm features form spheres. The detailed analysis of pore shape and amount reveals that – the elongated pores can withstand sintering if their initial size and aspect-ratio are large enough; – the spheres are less effective in reducing thermal conductivity; – even though they change their shape to spheroids, high aspectratio voids between feather-arms can affect thermal conductivity-stability positively due to their configuration relative to heat-flux; – high aspect-ratio and high density, intra-columnar pore-arrays provide the lowest thermal conductivity and is more resistive against thermal conductivity increase. Acknowledgements The authors from DLR acknowledge the conscientious support of the technicians at their institute. Thus, we appreciate greatly, the meticulous work of C. Kröder, J. Brien, and H. Mangers in the production of the coatings, and the assistance of R. Borath and W. Schönau in the characterization of the microstructures and the measurements of the thermal conductivity, respectively.
4. Conclusions Three EB-PVD PYSZ TBC microstructures with differences in the characteristics of their pores were manufactured by modifying the process parameters. Surface-area determination methods (BET and SANS) and SEM micrographs are complementary
References [1] D.D. Hass, A.J. Slifka, H.N.G. Wadley, Acta Mater. 49 (2001) 973. [2] J.R. Nicholls, K.J. Lawson, A. Johnstone, D.S. Rickerby, Surf. Coat. Technol. 151–152 (2002) 383.
2620
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
[3] U. Schulz, M. Schmücker, Mater. Sci. Eng. 276 (2000) 1. [4] T.E. Strangman, Thin Solid Films 127 (1985) 93. [5] A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, F.S. Pettit, Prog. Mater. Sci. 46 (2001) 505. [6] S.G. Terry, PhD Thesis, University of California in Santa Barbara, 2001. [7] J. Cho, S.G. Terry, R. LeSar, C.G. Levi, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 391 (2005) 390. [8] U. Schulz, J. Münzer, U. Kaden, Ceram. Eng. Sci. Proc. 23 (4) (2002) 353. [9] U. Schulz, K. Fritscher, C. Leyens, M. Peters, W.A. Kaysser, Mater.wiss. Werkst.tech. 28 (1997) 370. [10] K. Wada, N. Yamaguchi, H. Matsubara, Sci. Coat. Technol. 191 (2005). [11] U. Schulz, B. Saruhan, K. Fritscher, C. Leyens, Int. J. Appl. Ceram. Technol. 1 (2004) 302. [12] H.-J. Raetzer-Scheibe, U. Schulz, T. Krell, Surf. Coat. Technol. 200 (2006) 5636. [13] J.R. Nicholls, K.J. Lawson, D.S. Rickerby, P. Morell, Advanced Processing of TBC's for Reduced Thermal Conductivity, Aalborg, Denmark, 1998, p. 6.1. [14] J.R. Nicholls, K.J. Lawson, A. Johnstone, D.S. Rickerby, Mat. Sci. Forum 367–72 (2001) 595. [15] X. Zheng, D.G. Cahill, J.-C. Zaho, Adv. Eng. Mater. 7 (2005) 622. [16] S. Gu, T.J. Lu, D.D. Hass, H.N.G. Wadley, Acta Mater. 49 (2001) 2539. [17] P. Heydt, C. Lou, D.R. Clarke, J. Am. Ceram. Soc. 84 (2001) 1539. [18] D.E. Wolfe, J. Singh, R.A. Miller, J.I. Eldridge, D.-M. Zhu, Surf. Coat. Technol. 190 (2005) 132. [19] B.K. Jang, H. Matsubara, Scr. Mater. 54 (9) (2006) 1655. [20] U. Schulz, K. Fritscher, C. Leyens, M. Peters, J. Eng. Gas Turbine Power 124 (2002) 229. [21] P.J. Hall, S. Brown, J. Fernandez, J.M. Calo, Carbon 38 (2000) 1257. [22] J. Ilavsky, A.J. Allen, G.G. Long, J. Mater. Sci. 32 (1997) 3407. [23] J. Ilavsky, A.J. Allen, G.G. Long, S. Krueger, J. Am. Ceram. Soc. 80 (1997) 733. [24] J. Ilavsky, G.G. Long, A.J. Allen, C.C. Berndt, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 272 (1999) 215. [25] A. Kulkarni, Z. Wang, T. Nakamura, S. Sampath, A. Goland, H. Herman, J. Allen, J. Ilavsky, G. Long, J. Frahm, R.W. Steinbrech, Acta Mater. 51 (2003) 2457.
[26] Z. Wang, A. Kulkarni, S. Deshpande, T. Nakamura, H. Herman, Acta Mater. 51 (2003) 5319. [27] A.J. Allen, G.G. Long, H. Boukari, J. Ilavsky, A. Kulkarni, S. Sampath, H. Herman, A.N. Goland, Surf. Coat. Technol. 146–147 (2001) 544. [28] G. Porod, in: O. Galtter, O. Kratky (Eds.), Small-angle X-ray Scattering, Academic Press, London, U.K., 1982, p. 17. [29] U. Schulz, S.G. Terry, C.G. Levi, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 360 (2003) 319. [30] V. Lughi, V.K. Tolpygo, D.R. Clarke, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 368 (2004) 212. [31] D. Zhu, R.A. Miller, B.A. Nagaraj, R.W. Bruce, Surf. Coat. Technol. 138 (2001) 1. [32] R.E. Taylor, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 245 (1998) 160. [33] R.E. Taylor, X. Wang, X. Xu, Surf. Coat. Technol. 120–121 (1999) 89. [34] M. Blum, Choudhury, An improved KSR 600 EB-gun, Reno Conference, Bakish, 1994. [35] U. Keiderling, A. Wiedenmann, Phys., B Condens. Matter 213–214 (1995) 895. [36] J.J. Thomas, H.M. Jennings, A.J. Allen, Cem. Concr. Res. 28 (1998) 897. [37] J.J. Thomas, H.M. Jennings, A.J. Allen, Adv. Cem. Based Mater. 7 (1998) 119. [38] G. Kostorz, in: G. Kostorz (Ed.), Treatise on Materials Science and Technology, vol. 15, Academic Press, New York, 1979, p. 227. [39] T. Keller, W. Wagner, N. Margadant, S. Siegmann, J. Ilavsky, J. Pisacka, Characterization of Anisotropic, Thermally Sprayed Microstructures Using Small-angle Neutron Scattering, Brno, Czech Republic, 2001, p. 460. [40] A.J. Allen, J. Ilavsky, G.G. Long, J.S. Wallace, C.C. Berndt, H. Herman, Acta Mater. 49 (2001) 1661. [41] T.J. Lu, C.G. Levi, H.N.G. Wadley, A.G. Evans, J. Am. Ceram. Soc. 84 (2001) 2937. [42] A.F. Renteria, B. Saruhan, J. Eur. Ceram. Soc. 26 (2006) 2249. [43] W.D. Kingery, H.K. Bowen, D.R. Uhlmann, Introduction to Ceramics, 2nd ed., John Wiley and Sons, New York, 1976. [44] A.J. Slifka, B.J. Filla, J. Res. Natl. Inst. Stand. Technol. 108 (2003) 147. [45] F. Cernuschi, S. Ahmaniemi, P. Vuoristo, T. Mantyla, J. Eur. Ceram. Soc. 24 (2004) 2657.
Effect of morphology on thermal conductivity of EB-PVD PYSZ TBCs A. Flores Renteria a,⁎, B. Saruhan a , U. Schulz a , H.-J. Raetzer-Scheibe a , J. Haug b , A. Wiedenmann b a b
German Aerospace Centre (DLR) – Institute of Materials Research, Cologne, Germany Hahn-Meitner Institute, Structural Research, Department of Materials, Berlin, Germany Received 28 September 2005; accepted in revised form 4 May 2006 Available online 9 June 2006
Abstract Partially yttria stabilized zirconia (PYSZ) based thermal barrier coatings (TBC) manufactured by electron beam-physical vapour deposition (EB-PVD) protect turbine blades, working under severe service conditions in aero engines and stationary turbines. These coatings show a high strain tolerance relying on their unique morphology which is comprised of weakly bonded, preferred-oriented columns, voids between feather-like sub-columns and, finally, of intra-columnar closed pores. The results obtained in this work demonstrate that variation of the EB-PVD process parameters alters the resulting columnar morphology and porosity of the coatings. The physical properties and, most importantly, thermal conductivity, are greatly affected by these morphological alterations. This study investigates three morphologically different EB-PVD PYSZ TBC top coats in terms of the spatial and geometrical characteristics of their porosity and correlates those with the thermal conductivity values measured in as-coated state and after heat treatment at 1100 °C for 1 h and 100 h. Changes in the open and closed porosity caused by heat-treatment are characterized by small-angle neutron scattering (SANS), Brunauer–Emmett–Teller Method (BET) and scanning electron microscope (SEM). Correlation of shape and surface-area changes in all porosity types of the analysed coatings revealed that the thermal conductivity of these coatings is influenced primarily by size and shape distribution of the pores and secondarily by the pore surface-area available at the cross section perpendicular to the heat flux. © 2006 Elsevier B.V. All rights reserved. Keywords: Morphology of EB-PVD TBCs; Thermal conductivity; SANS and BET-analysis; Sintering
1. Introduction The temperature of the hot-gas generated in modern gas turbines exceeds the melting temperature of Ni-based superalloys used for the manufacture of turbine blades. In order to withstand this extreme thermal loading, turbine blades and other hot-structure components of stationary and airliner gas turbines are protected by Thermal Barrier Coatings (TBCs) based on Partially Yttria Stabilized Zirconia (PYSZ). Thus, the lifetime of turbine blades, as well as the engine efficiency, is increased. It is essential that TBCs show a low thermal conductivity, high thermal expansion coefficient, excellent thermal shock-resistance, as well as phase and morphological stability in long-term, high-temperature service.
⁎ Corresponding author. E-mail address: [email protected] (A.F. Renteria). 0257-8972/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2006.05.003
The state-of-the-art thermal barrier coatings consist of a PYSZ top-coat, a metallic (Pt/Al or NiCoCrAlY) bond coat (BC) and between these two, a thermally grown oxide (TGO) based on Al2O3. Top coats can be manufactured by electron beam physical vapour deposition (EB-PVD) or by air plasma spraying (APS). Both EB-PVD and APS top coats display highly porous microstructures which lowers the intrinsic thermal conductivity. Thus, the differences in the thermal conductivity values between EB-PVD and APS coatings are caused by the differences in shape, orientation and distribution of their porosity [1,2]. During EB-PVD processing of the coatings, quasi single crystalline columns of very fine diameter (b 2–3 μm) start to grow on the substrate surface. Their diameter increases at the coating tip to 10–20 μm. These columns are only weakly interconnected via inter-columnar gaps, which accounts for some of the open porosity. Inter-columnar gaps are of a few nanometres close to the substrate and can be as large as 1 μm in
2612
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
width at the coating tip. High thermal shock-resistance of the EB-PVD processed TBCs relies on the presence of these intercolumnar gaps [3–5]. Most of the open porosity originates from the voids present between nano-sized secondary columns, socalled feather-arms, growing at the column edges. These are formed during evaporation mainly by secondary shadowing, depending on rotation speed, substrate temperature and the angle between the substrate and the plane of vapour incidence [6,7]. Furthermore, due to the interruption of the vapour deposition during rotation, closed porosity is created within the primary columns in form of finer elongated channel-like pores (i.e. intra-columnar voids) (see Fig. 1). Exposure of TBCs to high temperatures under service conditions causes changes in the morphology of the columns by a series of thermal processes such as: formation of bridges between the columns, formation of sintering necks at contact points between the feather-arms, and eventually changes in pore geometry and sizes of the intra-columnar pores, generating a surface-area reduction. Morphology and density of the columnar microstructure of EB-PVD TBCs can be varied by changing the deposition parameters such as chamber pressure, substrate temperature and rotation speed [8–10]. In particular, the porosity between the feather-arm features in EB-PVD TBCs is recognized as an important factor controlling their thermal conductivity. Since most of the heat flows through the conducting columns and in-plane thermal gradient does not vary across the parallel aligned primary columns and inter-columnar gaps, it is considered that the inter-columnar gaps orientated parallel to the heat flux induce only a slight effect on the thermal conductivity [1,11]. Recent studies have reported controversial results related to this consideration. Instead, significant reductions in the thermal conductivity were identified by decreasing the coating thickness. This is correlated to the factors emerging from inclined growth of columns relevant to the direction perpendicular to substrate plane and to the presence of a non-textured crystal structure as well as a high fraction of inter-columnar pores at the bottom zone of EB-PVD coatings [12–14]. Nevertheless, micro-scale spatial resolution measurements of thermal conductivity via laser time-domain thermo reflectance (TDTR) showed that approximately 95% of the total reduction of the thermal conductivity compared with the bulk value was influenced by the presence of pores within the
columns (i.e. intra-columnar pores), although, thermal diffusivity remains constant along the coating thickness, despite the microstructural heterogeneity [15]. It is suggested by Hass et al. [1], Gu et al. [16] and Heydt et al. [17], that the effect of porosity on the thermal conductivity is due to the reduction of the cross sectional area through which the heat flows. Heat flux density is statistically a local representative of the porosity distribution, leading to the assumption that it is the alteration of porosity which influences the thermal conductivity, rather than the modification of phonon scattering per se. Basing on this theory, some studies varied the coating morphology by either interrupting the atomic deposition from the vapour by “in and out” feeding or “shuttering” of the substrates during the coating process [18], by creating a microstructure with a zigzag intercolumnar porosity [1,11,16], or by successive changing of the rotation speed during the deposition process [9,19,20]. Characterization of the pores in these coatings requires the employment of sophisticated techniques due to their sizes and size differences, accessibility (open and closed pores) and anisotropy. Use of innovative methods has been reported for precise and detailed evaluation of open and closed porosity changes on ceramic materials during service simulated conditions. These methods are based on the transmission of X-rays or neutrons through the coatings. Small-Angle Neutron Scattering (SANS) technique represents, with some advantages, an alternative to the gas adsorption methods. This method is sensitive to both closed and open porosities and, in many cases, offers a more complete analysis of the measured porosity [21]. SANS has been already applied to study the processing-microstructure relations and sintering of plasma sprayed coatings (APS) [22–27]. These studies, in particular, focused on measurement of Porod scattering [28] derived from the terminal slope in SANS spectra to determine the characteristics of voids in porous materials. The quantitative information obtained on the corresponding interface surface-area is dominated by the measured intensities. Since the fine features in a microstructure are the major contributors to the deduced area [25], it is important to measure the minimum size of the scatterers present in a material in order to ascertain the origin of the scattered apparent Porod surface-area. Porod scattering technique does not require a void shape model, and thus, is especially useful for studies of void systems with complex shapes [22].
Fig. 1. Scanning electron micrographs showing the type and distribution of the pores at the EB-PVD PYSZ TBCs.
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
This paper reports the BET, SANS and SEM analysis of the morphology and surface-area changes of the open and closed pores at EB-PVD TBCs with three different morphologies before and after heat-treatment. Additionally, the thermal conductivity of these coatings was measured in free-standing conditions by Laser Flash Analysis Method (LFA) before and after heat treatment at 1100 °C for 1 h and 100 h and was correlated with the microstructural observations. 2. Materials and methods 2.1. Processing and materials EB-PVD coatings were produced by employing “von Ardenne” pilot plant equipment with a maximum EB-Power of 150 kW. Evaporation was carried out from single source (ingot) having the standard chemical composition (7–8 wt.% Y2O3 stabilized ZrO2) and with the dimensions of 62.5 mm diameter and 150 mm length. Vapour phase is deposited on plane substrates under conventional rotating mode as described in [29] (i.e. by mounting the substrates on a holder with its axis perpendicular to the evaporation source). A thickness of approximately 400 μm is achieved by EB-PVD coating of PYSZ directly on FeCrAl-alloy substrates (12.7 mm diameter) without having a bond coat. To obtain different coating morphologies, the substrates were rotated at different speeds and heated to different temperatures during the coating process. Table 1 lists the sample designation and the applied process parameters for the manufacture of three investigated EB-PVD coating morphologies. These parameters are selected based on the previous studies of the group which have shown considerable differences in the porosity configuration of the coatings [9]. Since the applied analysis methods (e.g. BET, SANS and LFA) require the use of coatings in free-standing conditions, the substrates were etched out completely in an acid solution to obtain free-standing coatings. Subsequently, the corresponding specimens were heated to 1100 °C with a heating rate of 5 °C/ min and isothermally aged at this temperature in air for 1 h and 100 h. After the respective isothermal ageing, the samples were taken off from the furnace and “quenched” in air to freeze the microstructure. Additionally, free-standing coatings were coated on both sides with a thin Pt-layer (4–5 μm thickness) by sputter technique to avoid laser-beam penetration during the LFA-measurements. Thermal cycling of coatings attached on substrates can induce interface resistance at the underlying TGO/BC, adding to the total thermal conductivity value. This contribution cannot be separated from the top coat sintering related increase of thermal conductivity. Thermal expansion coefficient mismatch, in turn, is responsible for formation of Table 1 Designation of the samples and applied parameters during the EB-PVD process Designation for morphology
Chamber pressure Substrate temperature Rotation speed (mbar) (°C) (rpm)
Intermediate Fine Coarse
8 × 10− 3 4 × 10− 3 8 × 10− 3
950 850 1000
12 3 3
2613
stresses at the interface between BC/TGO/top coat [30]. The free-standing configuration of the specimens, in fact, “releases” the internal stresses which are otherwise created in a TBC system during cyclic or isothermal ageing [5]. However, these stresses mainly affect the dimensions of the inter-columnar gaps, which are, as a matter of fact, considered to affect only slightly the overall thermal conductivity of the coatings [1,11]. Considerable uncertainties may emerge from measuring the coatings on the metallic substrate due to micro-delamination of the top coats on ageing [31] and heat-treatment induced interface resistance, as well as, failures related to large differences such as thin top coat/ thick substrate configurations [32,33]. Therefore, we consider that the measurement of thermal conductivity on free-standing aged coatings is a satisfactory procedure to obtain representative values for the ceramic top coat and to reduce the uncertainties. 2.2. Methods of characterization 2.2.1. Thermal conductivity In the present work, the thermal diffusivity (α) of the specimens was measured via Laser Flash Analysis Method (LFA 427, Netzsch, Selb/Germany). The thermal conductivity [34] is calculated by applying the formula: k ¼ a d q d Cp
ð1Þ
where ρ, in this case, represents the bulk density of the freestanding coating measured by the Archimedes Method, and Cp is the specific heat measured by Differential Scanning Calorimeter (DSC). The accuracy of the calculated thermal conductivities is limited with the uncertainties of the factors related to the Eq. (1). In this work, the specific heat was measured via differentiated scanning calorimeter instrument (DSC-404, Netzsch) which encloses a systematic standard deviation of ± 3–5% according to the manufacturer. Moreover, the thermal diffusivities used in the calculations based on an average of three measured values for each temperature condition. Ultimately, the thickness of the specimens was measured via optical microscope and averaged on the basis of 10 measurements. The bulk density represents the average value of nine morphologically-identical coatings. During the thermal diffusivity measurements, the association of the curves obtained under heating and cooling conditions indicates the thermal history of the specimens. Since the as-coated EB-PVD morphology is highly temperature sensitive, a pre-treatment of the manufactured coatings at temperatures slightly above the deposition temperature is a general practice to equilibrate the microstructure. Since we aim to determine the effect of sintering on thermal conductivity, we have left out this pre-treatment and measured the coatings in as-coated conditions. Therefore, in our case, the thermal conductivity curves obtained on heating and cooling over the temperature range between RT and 1000 °C are different. 2.2.2. Morphology and surface area determination The microstructures were characterized visually by using a Field-Emission Scanning Electron Microscope (FE-SEM, LEITZ LEO 982, Germany). Since the columnar microstructure of these coatings contains open (inter-columnar gaps and voids between
2614
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
feather-arms) and closed porosities (inter-columnar pores), it was necessary to use, beside the microstructural characterization also such measurement methods which are capable of tracking the surface-area changes of all type of pores. A precise analysis between these thermally induced microstructure changes and the thermal conductivity can be possible, if surface-area of the open porosity is measured by BET and SANS. BET measurements were carried out at the Department of Ceramic Technologies and Sintered Materials of the Fraunhofer Institute in Dresden/ Germany using an Accelerated Surface Area Porosimeter System (ASAP 2010) and employing N2 as analysis gas. Since, the BET allows predominantly the analysis of the open porosity, the SANS measurement were also carried out for a detailed characterization of the total surface-area. SANS experiments were performed at the instrument V4 of the Berlin Neutron Scattering Centre (BENSC) in the Hahn–Meitner Institute, Berlin/ Germany (for detailed information see Ref. [35]). The scattered neutron intensity I(Q) was measured using an incident beam along the plane perpendicular to the substrate's plane (parallel to the column-axis) over two-sample detector distances: 1 m and 4 m, employing a λ = 0.605 nm neutron wavelength. During the measurements, the specimens were placed on the sample holder by means of a cadmium mask having a circular slit of 6 mm diameter. The obtained data were corrected for background scattering, transmission and detector efficiency, and normalized to absolute units by using water as calibration standard procedure. Quantitative information of the surface-area corresponding to the intra-columnar pores was achieved in the form of the corresponding Porod Constant (Pc) as a function of the azimuthal angle by converting the 2-D data to 1-D scattering profiles averaged over 5°. The Porod Constant (Pc) is directly proportional to the Apparent Porod Surface Area (Sv) [36,37] according to: Sv ¼
Pc 2kjDqj2
ð2Þ
where, Δρ is the contrast in scattering length density between the pores and the PYSZ material. In this method, the measured Porod
scattering intensity I(Q) is linearly proportional to the interfacial surface-area projection in a plane perpendicular to the scattering vector Q, corresponding to dimensions measured parallel to it [28,38]. In view of the fact that the Porod scattering measured in one direction does not reflect the total surface-area of the scatterers but a projection in the plane perpendicular to the scattering vector Q, it is often called Apparent Porod Surface Area, which can be calculated from its corresponding Porod constant by Eq. (2). The total specific area of the scatterers within the sample can be obtained from the apparent Porod constant by averaging over all 3D-directions in the material [39]. However, this apparent surface-area is useful to analyse microstructural changes between samples by measuring them under the same conditions. 3. Results and discussion Fig. 2 shows SEM micrographs of the investigated EB-PVD PYSZ top coats having morphological differences in their column diameter, in the size and shape of inter-columnar gaps and of voids between feather-arm features. These differences were produced by altering the process parameters during the EBPVD coating as specified in Table 1. The morphologies were designated according to their appearances on the micrographs. The “intermediate” microstructure, produced under fast rotation-speed, shows symmetrical columns reaching to approximately 10 μm diameter at the tip zone of coating (bulk density: 4.37 g/cm3 ± 3%). So-called feather-arm features surrounding the columns are elongated and extend from the column edge to the column axis in approximately 1/4 depth of their length (see Fig. 2a). The microstructure designated as “fine” was produced by employing a lower substrate temperature and slower rotation speed and exhibits non-symmetrical columns having ≤5 μm tip diameter which are surrounded with very short feather-arms (bulk density: 4.50 g/cm3 ± 3%) (see Fig. 2b). The microstructure designated as “coarse” was produced by employing a higher substrate temperature combined with slower rotationspeed and displays broader inter-columnar gaps and larger columns with tip diameters of around 20 μm (bulk density:
Fig. 2. Scanning electron micrographs of three different as-coated EB-PVD-morphologies of PYSZ-TBCs in cross section perpendicular to the rotation-axis: “intermediate” (a), “fine” (b) and “coarse” (c); and from the coating tip: “intermediate” (d), “fine” (e) and “coarse” (f ). (d) to (f ) are arranged such that the rotation-axis during deposition is vertical.
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
4.65 g/cm3 ± 5%). The feather-arms of this morphology are also elongated and extend from the column edge to the column axis in approximately 1/7 depth of their length (see Fig. 2c). High resolution SEM micrographs of the manufactured EBPVD TBCs in the Fig. 3 show the coating cross sections deposited perpendicular to the rotation axis. The geometrical and spatial characteristics of the voids between feather-arms and intra-columnar pores display changes at each morphology after ageing. The micrograph in the Fig. 3a1 displays the morphology of the “intermediate” microstructure. In this coating, the voids
2615
between feather-arms are approximately 0.07–0.10 μm wide and 2–2.5 μm long. In addition, this coating contains arrays of intra-columnar pores which form after every rotation movement and rowed with a distance of 0.25–0.30 μm from each other (see Fig. 3d1). The voids between feather-arms at the “fine” microstructure are in turn approximately 0.01–0.05 μm wide and 0.50–0.80 μm long (see Fig. 3a2). The distance between the intra-columnar pore arrays is broader (1.2–1.6 μm) compared to the “intermediate” microstructure due to the slower rotation speed applied during
Fig. 3. Scanning electron micrographs of cross sections perpendicular to the rotation-axis of the three different EB-PVD-morphologies of PYSZ-TBCs, showing the inter-columnar pores and pores between feather-arms in as-coated condition: “intermediate” (a1), “fine” (a2) and “coarse” (a3). After heat treatment at 1100 °C for 1 h: “intermediate” (b1), “fine” (b2) and “coarse” (b3) and after ageing at 1100 °C for 100 h: “intermediate”(c1), “fine”(c2) and “coarse”(c3). Additionally, the intra-columnar pores in as-coated condition are shown: “intermediate” (d1), “fine” (d2) and “coarse” (d3); after ageing at 1100 °C for 1 h: “intermediate” (e1), “fine” (e2) and “coarse” (e3) and after ageing at 1100 °C for 100 h: “intermediate” (f1), “fine” (f2) and “coarse” (f3).
2616
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
deposition of the “fine” microstructure (see Fig. 3d2). Among the three manufactured microstructures, the microstructure designated as “fine” possesses the highest density of intra-columnar pores, which clearly demonstrates that at the corresponding substrate temperature (e.g. 850 °C, T / TM = 0.37), the effect of shadowing predominates to diffusion during the film growth. Analogous results are reported by Cho et al. [7] who carried out kinetic Monte Carlo simulation studies on film EB-PVD deposition during substrate rotation. Finally, the voids between feather-arms of the “coarse” microstructure are approximately 0.07–0.15 μm wide and 2–3 μm long (see Fig. 3a3). The distance between the intra-columnar pore arrays is also broader (1.6–2.0 μm) compared to the “intermediate” microstructure (see Fig. 3d3). Despite the same applied rotation speed, the distances between the intra-columnar pore arrays of the “fine” and “coarse” microstructures differ slightly. This difference correlates well with the deposition rate difference of these two microstructures (see Table 2). Elongated intra-columnar pore arrays form at the regions where vapour incidence angle (VIA) is the highest and the regrowth of coating starts on each new rotation. Some of these pores extends themselves through the depositing material in a sunrise–sunset pattern which is influenced by the tip of the neighbouring columns as the VIA direction changes. Thus, the elongated “banana”-shaped pores are formed. Fig. 3d1, d2 and d3 show micrographs of intra-columnar pore arrays with corresponding elongated “banana”-shape pores. The “banana”shaped pores belonging to the “fine” microstructure possess the highest aspect-ratio (Fig. 3d2). This may be a result of the slow rotation-speed. But also, the low substrate temperature may hinder the atomic diffusion, thus enabling the formation of elongated thin pores. In turn, the “coarse” microstructure shows less of low aspect-ratio intra-columnar pores. The rotation-speed applied in the manufacture of the “fine” and “coarse” microstructures are the same, however, substrate temperatures are different (Tcoarse N Tfine). The diffusion processes become more active at the higher substrate temperature of the “coarse” microstructure (Fig. 3d3). The “intermediate” microstructure displays a higher number, low aspect ratio intra-columnar pores due to considerably higher rotation-speed (Fig. 3d1). Since the formation of intracolumnar pores occurs repeatedly on each new plane of the coating deposition, they contribute greatly to the surface-area. The graph in Fig. 4 shows absolute values of surface-area changes measured by BET for all investigated coatings. The surface-area changes represent solely the open pores such as inter-columnar gaps and the voids between feather-arms. It is clearly visible at the as-coated condition that the “fine” microTable 2 Microstructural characteristics of the analysed EB-PVD TBCs Designation for morphology
Deposited material (μm/rotation)
Deposition rate (μm/min)
Width of columns (μm)
Voids between feather-arms Length (μm)
Width (μm)
Intermediate Fine Coarse
0.25–0.30 1.20–1.60 1.60–2.00
5.60 5.40 6.70
10.00 ≤ 5.00 20.00
2.00–2.50 0.50–0.80 2.00–3.00
0.07–0.10 0.01–0.05 0.07–0.15
Fig. 4. BET-surface-area values (m2/g) of the “intermediate”, “fine” and “coarse” EB-PVD PYSZ TBC microstructures measured in as-coated condition and after ageing at 1100 °C for 1 h and 100 h.
structure contains the highest surface-area being well in agreement with its higher quantity of columns/inter-columnar gaps and voids between feather-arms per volume unit of material. The lowest surface-area for open pores was measured at the “coarse” microstructure. The polar figures in Fig. 5 exhibit the distribution of the surface-area of the voids between feather-arms and the closed intra-columnar pores measured in the direction parallel to the column axis (i.e. perpendicular to the substrate plane) for all three analysed coatings. The surface-area measured for the “fine” microstructure possesses the highest value while that for the “coarse” the lowest, matching well with the BET-measurement. It is evident from Fig. 5 that in the as-coated condition, the distribution of the surface-area for all three analysed microstructures is anisotropic, being that in the direction parallel to the rotation axis highest. This anisotropy observed at the Porod scattering indicates an alignment in the shape and orientation of the individual pores as well as in their distribution. In order to determine which of these factors influences mostly the anisotropy of the surface-area distribution, it is necessary to analyse the specimens in another direction (i.e. parallel to the substrate plane). Moreover, Multiple Small-Angle Neutron Scattering (MSANS) method for instance enables determination of the dimensions and volume fraction of the voids [23,40] which is beyond the scope of this work. LFA analyse on as-coated specimens indicate that the thermal conductivity of the “coarse” microstructure is approximately 18 and 14% (at 800 °C) and, thus, higher than that of the “intermediate” and “fine” microstructures, respectively (see Figs. 6 and 7). Considering the fact that the porosity affects the thermal conductivity by reducing the cross sectional area through which the heat flows, the obtained thermal conductivity results correspond well to the SANS and BET measured surface-area values which represent the total open and closed pores. Since the intracolumnar pores in the “fine” microstructure have the highest aspect ratio and are inclined respective to the heat-flux, these may induce the strongest reduction in the thermal conductivity being in accordance to the observation of Lu et al. [41].
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
2617
Fig. 5. Porod Constant values measured in as-coated and after heat treatment at 1100 °C for 1 h and 100 h, as a function of the azimuthal angle being plane parallel to the column axis of “intermediate”(a), “fine”(b) and “coarse”(c) EB-PVD PYSZ TBC microstructures.
After heat treatment at 1100 °C for 1 hr and 100 h, irreversible changes occur at all three analysed microstructures. Previous BET and SANS investigations carried out to determine the morphological changes of EB-PVD TBCs having “intermediate” microstructure after heat-treatment showed an abrupt change in the surface-area after 1 h at 1100 °C for all pore types due to sintering [42]. In the case of the voids between featherarms, such changes may occur firstly at contact points forming along their lengths finally resulting in aligned pore arrays with fine quasi-spherical and slightly elongated shapes (see Fig. 3b1). This process results in a noticeable reduction of the surface-area representing the open porosity (see Fig. 4). In the case of the “fine” microstructure, the closure of voids between feather-arms occurs without the formation of contact points, since these are comprised of very short lengths (see Fig. 3b2). Nevertheless, an abrupt reduction in the BET-surface-area has been observed, which may in fact correspond to the high number of intra-columnar pores present in this microstructure (Fig. 4). In the case of the “coarse” microstructure, a slight change occurs in the surface-area and shape of the pores. Closure of the voids between featherarms may follow the same manner at this microstructure as at the “intermediate” microstructure. The elongated intra-columnar closed pores of the “coarse” microstructure are wider, and thus, the formation of contact points occurs less frequently (see Fig. 3b3). Heat treatment at 1100 °C for 1 h causes a considerable reduction in the Apparent Porod Surface Area (Porod Constant, Pc) of the intra-columnar closed pores for all three studied microstructures (see Fig. 5). This may be due to the break-up in their elongated shape forming arrays of quasi-spherical pores (see Fig. 3e1, e2 and e3). The “fine” microstructure yields the highest reduction of the surface-area while the “coarse” the lowest, being in accordance with the higher surface-area of the “fine” microstructure yielding higher surface energy. Nevertheless, the surface-area distribution at these two analysed microstructures remains anisotropic even after heat-treatment (see Fig. 5b and c). The surface-area distribution at the “intermediate” coating becomes isotropic after heat-
treatment (see Fig. 5a), indicating a random distribution of the pores and/or formation of spherical shaped individual pores. Prolonged heat treatment of all coatings for 100 h at the same temperature (e.g. 1100 °C) induces, in turn, only a slight change at their surface-area in relation to that observed after 1 h of heattreatment. In summary, it can be said that sintering-related, morphological changes occur very fast at 1100 °C and are mostly completed already after 1 h. Fig. 6 shows the variation of the thermal conductivity values of the three investigated coatings as a function of measuring temperature before and after heat-treatment at 1100 °C for 1 h and 100 h. The determination of thermal conductivity values is carried out with 400 μm thick free-standing EB-PVD-coatings. Typically, these coatings have 50 μm thick bottom part which consists of finer columns. The study which was carried out to determine the thickness dependency of thermal conductivity showed lower conductivity values for the coating having only 50 μm thickness [12]. The influence of this fine columned area on the thermal conductivity value becomes less as the coatings thickness increases over 200 μm. The thermal conductivity curves of the as-coated “intermediate” and “fine” microstructures show no apparent temperature dependency. In contrast, a sharper decay below 400 °C has been observed at the “coarse” microstructure (Fig. 6). Such an effect is attributed to different causes in the literature [43]. Slifka and Filla [44] has reported that the porosity affects the diffusion of long-wavelength phonons leading such differences. This behaviour cannot be observed at the curves obtained from the samples aged for 1 h at 1100 °C, confirming the relation of thermal conductivity with the morphological changes occurring in the shape and surface-area of the porosity present at these coatings. Moreover, thermal conductivity values of the ascoated specimens increase at temperatures higher than 900 °C. According to our previous work, this effect occurs due to heatinduced changes in the microstructure (i.e. sintering) [42]. After heat treatment at 1100 °C for 1 h, the thermal conductivity of the “intermediate” microstructure measured at
2618
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
There are noticeable increases in the thermal conductivity of three analysed coatings after heat treatment. Assuming that the inter-columnar pores orientated parallel to the heat flux induce only a slight effect on the thermal conductivity of EB-PVD TBCs, these differences can be principally related to the thermally induced changes of the voids between feather-arms and intra-columnar pores. Although the “intermediate” microstructure shows the highest thermal conductivity increase, the total surface-area reduction is not the highest at this microstructure. Observation of the corresponding micrographs given in Fig. 3(b1 to f1) reveals that the intra-columnar pores as well as the inner part of the feather-arms voids break into quasi-spheres after 1 h ageing at 1100 °C (see Fig. 3b1 and e1). This observation agrees well with the consideration of Lu et al. [41] and Cernuschi et al. [45], confirming that spheres, as well as cylinders and plates oriented parallel to the heat-flux, provide the lowest effect on the thermal conductivity reduction. On prolonged ageing (e.g. 100 h at 1100 °C), the aspect-ratio of the voids between feather-arms are extremely reduced and this leads solely to a slight increase in thermal conductivity (see Figs. 3c1, 7). As these drastic changes occur at the feather-arm features of the “intermediate” microstructure, the shape of the intra-columnar pores does not change much (i.e. remains quasi-spherical) (see Fig. 3e1 and f1). In the case of the “fine” microstructure, the elongated intracolumnar pores break into small spheres on ageing (Fig.3b2 and e2). However, the majority of intra-columnar pores, which were initially formed at regions of highest vapour incidence angle (VIA) retain their elongated shape despite the heat-treatment (see Fig. 3e2 and f2). Being specific to this microstructure, the ageing induced changes in the shape of the voids between featherarms do not affect the thermal conductivity since their length stretching inside the columns is very short (see Fig. 3b2 and c2). Due to these reasons, the increase in the thermal conductivity after ageing is lower. Finally, for the “coarse” microstructure, the intra-columnar pores as well as the voids between feather-arms change their morphology slightly after 1 h ageing at 1100 °C. The parts of the
Fig. 6. Thermal conductivity values of the “intermediate”(a), “fine” (b) and “coarse” (c) EB-PVD PYSZ TBC microstructures measured in as-coated condition and after ageing at 1100 °C for 1 h and 100 h.
800 °C increases approximately 10% and that of the “fine” microstructure approximately 4% reaching nearly to the value achieved after heat treatment at 1100 °C for 100 h (e.g. 13% and 6% increase, respectively). In the case of the “coarse” microstructure, the total surface-area changes slightly after 1 h and 100 h heat treatment at 1100 °C. Increase in the heat diffusivity is also very small, roughly corresponding to that of the experimental deviation (measured at 800 °C approximately 2% and 6% higher, respectively) (see Fig. 7).
Fig. 7. Comparison of thermal conductivity values of the “intermediate”, “fine” and “coarse” EB-PVD PYSZ TBC microstructures in as-coated condition and after ageing at 1100 °C for 1 h and 100 h.
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
intra-columnar pores formed at regions of highest vapour incidence angle (VIA) maintain their original elongated shape to some extent. However, some of the elongated intra-columnar pores which form within the plane of the last deposited material break into arrays of fine spheres (see Fig. 3e3). The voids between feather-arms hold their high aspect-ratio in spite of thermal ageing (Fig. 3b3). Only a slight change in the thermal conductivity values of this microstructure is observed after ageing which can be correlated with the above mentioned microstructural stability. However, if the ageing time is prolonged to 100 h at the same temperature, the shape of the intra-columnar pores and voids between feather-arms changes to quasispheres yielding a reduction in their aspect-ratio and, consequently, causing a higher increase in the thermal conductivity (see Fig. 3c3 and f3). In summary, slow-rotation and low-substrate temperature leads to the formation of high aspect ratio feather-arms, long elongated, banana-shaped intra-columnar pores and highlypopulated pore arrays. Such microstructure yields a very low thermal conductivity. Although it suffers under thermal ageing resulting in high surface-area reduction, the formation of new elongated closed pores located perpendicular to heat flux yields an increase in thermal conductivity. High substrate temperature, despite slow rotation-speed results in intra-columnar pore arrays with broader distances, low aspect-ratio, banana-shaped pores and high aspect-ratio feather-arm features. Such microstructure, in turn, yields a rather high thermal conductivity which, on prolonged ageing, increases considerably. Fast-rotation speed produces shorter distances between the spherical, intra-columnar pore arrays and high aspect-ratio feather arms. This microstructure has a moderate thermal conductivity in as-coated condition which, however, increases considerably on ageing already after 1 h. Formation of large spheres on break-up of feather-arms may be the reason for this since they have almost no contribution in hindering heat-flux. Under thermal cycling conditions, thermal conductivity changes may be slightly different. However, in the cases when the cycle length is 1 h, the behaviour must be similar to our observations, since the most of the thermally induced changes occur in these morphologies within the first hour at 1100 °C. To ensure a stable low thermal conductivity, EB-PVD process parameters should be selected such that the resulting coating contains a microstructure having a stable pore configuration after heat-treatment, regardless of their form in as-coated condition. Furthermore, there is a need for precise characterization of the specific spatial and geometrical characteristics of each pore type at these coatings. Comprehensive studies have been carried out by employing USAXS measurements and corresponding modelling to obtain the related individual effect on the thermal conductivity and publication of the results is underway.
2619
to each other in studying the changes in the fine porous structures of EB-PVD PYSZ TBCs before and after heat treatment, allowing a detailed understanding of the interplay between the geometry, orientation and distribution of the pores and their thermal conductivity. The thermal conductivity of manufactured coatings was analysed before and after ageing. There are clear differences in the thermal conductivity values of the three as-coated morphologies. After ageing at 1100 °C, due to sintering-related phenomena, the changes in shape and distribution of the pores occur, resulting in reduction of surface-area and increase of thermal conductivity. The increase in thermal conductivity after heattreatment can be related to the alteration of the shape of the pores rather than the reduction of their surface-area at the cross section through which the heat flows. The “fine” microstructure shows on ageing the highest reduction in the total surface-area covering open and closed porosity but not the highest increase in thermal conductivity. While the “intermediate” microstructure displays the highest increase in thermal conductivity, in spite of a moderate surface-area reduction. This can be because of short lengths of voids between feather-arms in this microstructure. Thus, it can be assumed that its thermal conductivity is basically affected by the changes in the intra-columnar pores which retain their elongated shape relatively well after applied ageing. It is observed that the intra-columnar pores which have initial broader shape and large size characteristics are resistant to sintering. The voids which break into spheres on ageing yield still the best size distribution in reducing the thermal conductivity. In the “intermediate” microstructure, on ageing, the intra-columnar pores as well as voids between the feather-arm features form spheres. The detailed analysis of pore shape and amount reveals that – the elongated pores can withstand sintering if their initial size and aspect-ratio are large enough; – the spheres are less effective in reducing thermal conductivity; – even though they change their shape to spheroids, high aspectratio voids between feather-arms can affect thermal conductivity-stability positively due to their configuration relative to heat-flux; – high aspect-ratio and high density, intra-columnar pore-arrays provide the lowest thermal conductivity and is more resistive against thermal conductivity increase. Acknowledgements The authors from DLR acknowledge the conscientious support of the technicians at their institute. Thus, we appreciate greatly, the meticulous work of C. Kröder, J. Brien, and H. Mangers in the production of the coatings, and the assistance of R. Borath and W. Schönau in the characterization of the microstructures and the measurements of the thermal conductivity, respectively.
4. Conclusions Three EB-PVD PYSZ TBC microstructures with differences in the characteristics of their pores were manufactured by modifying the process parameters. Surface-area determination methods (BET and SANS) and SEM micrographs are complementary
References [1] D.D. Hass, A.J. Slifka, H.N.G. Wadley, Acta Mater. 49 (2001) 973. [2] J.R. Nicholls, K.J. Lawson, A. Johnstone, D.S. Rickerby, Surf. Coat. Technol. 151–152 (2002) 383.
2620
A.F. Renteria et al. / Surface & Coatings Technology 201 (2006) 2611–2620
[3] U. Schulz, M. Schmücker, Mater. Sci. Eng. 276 (2000) 1. [4] T.E. Strangman, Thin Solid Films 127 (1985) 93. [5] A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, F.S. Pettit, Prog. Mater. Sci. 46 (2001) 505. [6] S.G. Terry, PhD Thesis, University of California in Santa Barbara, 2001. [7] J. Cho, S.G. Terry, R. LeSar, C.G. Levi, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 391 (2005) 390. [8] U. Schulz, J. Münzer, U. Kaden, Ceram. Eng. Sci. Proc. 23 (4) (2002) 353. [9] U. Schulz, K. Fritscher, C. Leyens, M. Peters, W.A. Kaysser, Mater.wiss. Werkst.tech. 28 (1997) 370. [10] K. Wada, N. Yamaguchi, H. Matsubara, Sci. Coat. Technol. 191 (2005). [11] U. Schulz, B. Saruhan, K. Fritscher, C. Leyens, Int. J. Appl. Ceram. Technol. 1 (2004) 302. [12] H.-J. Raetzer-Scheibe, U. Schulz, T. Krell, Surf. Coat. Technol. 200 (2006) 5636. [13] J.R. Nicholls, K.J. Lawson, D.S. Rickerby, P. Morell, Advanced Processing of TBC's for Reduced Thermal Conductivity, Aalborg, Denmark, 1998, p. 6.1. [14] J.R. Nicholls, K.J. Lawson, A. Johnstone, D.S. Rickerby, Mat. Sci. Forum 367–72 (2001) 595. [15] X. Zheng, D.G. Cahill, J.-C. Zaho, Adv. Eng. Mater. 7 (2005) 622. [16] S. Gu, T.J. Lu, D.D. Hass, H.N.G. Wadley, Acta Mater. 49 (2001) 2539. [17] P. Heydt, C. Lou, D.R. Clarke, J. Am. Ceram. Soc. 84 (2001) 1539. [18] D.E. Wolfe, J. Singh, R.A. Miller, J.I. Eldridge, D.-M. Zhu, Surf. Coat. Technol. 190 (2005) 132. [19] B.K. Jang, H. Matsubara, Scr. Mater. 54 (9) (2006) 1655. [20] U. Schulz, K. Fritscher, C. Leyens, M. Peters, J. Eng. Gas Turbine Power 124 (2002) 229. [21] P.J. Hall, S. Brown, J. Fernandez, J.M. Calo, Carbon 38 (2000) 1257. [22] J. Ilavsky, A.J. Allen, G.G. Long, J. Mater. Sci. 32 (1997) 3407. [23] J. Ilavsky, A.J. Allen, G.G. Long, S. Krueger, J. Am. Ceram. Soc. 80 (1997) 733. [24] J. Ilavsky, G.G. Long, A.J. Allen, C.C. Berndt, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 272 (1999) 215. [25] A. Kulkarni, Z. Wang, T. Nakamura, S. Sampath, A. Goland, H. Herman, J. Allen, J. Ilavsky, G. Long, J. Frahm, R.W. Steinbrech, Acta Mater. 51 (2003) 2457.
[26] Z. Wang, A. Kulkarni, S. Deshpande, T. Nakamura, H. Herman, Acta Mater. 51 (2003) 5319. [27] A.J. Allen, G.G. Long, H. Boukari, J. Ilavsky, A. Kulkarni, S. Sampath, H. Herman, A.N. Goland, Surf. Coat. Technol. 146–147 (2001) 544. [28] G. Porod, in: O. Galtter, O. Kratky (Eds.), Small-angle X-ray Scattering, Academic Press, London, U.K., 1982, p. 17. [29] U. Schulz, S.G. Terry, C.G. Levi, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 360 (2003) 319. [30] V. Lughi, V.K. Tolpygo, D.R. Clarke, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 368 (2004) 212. [31] D. Zhu, R.A. Miller, B.A. Nagaraj, R.W. Bruce, Surf. Coat. Technol. 138 (2001) 1. [32] R.E. Taylor, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct. Process. 245 (1998) 160. [33] R.E. Taylor, X. Wang, X. Xu, Surf. Coat. Technol. 120–121 (1999) 89. [34] M. Blum, Choudhury, An improved KSR 600 EB-gun, Reno Conference, Bakish, 1994. [35] U. Keiderling, A. Wiedenmann, Phys., B Condens. Matter 213–214 (1995) 895. [36] J.J. Thomas, H.M. Jennings, A.J. Allen, Cem. Concr. Res. 28 (1998) 897. [37] J.J. Thomas, H.M. Jennings, A.J. Allen, Adv. Cem. Based Mater. 7 (1998) 119. [38] G. Kostorz, in: G. Kostorz (Ed.), Treatise on Materials Science and Technology, vol. 15, Academic Press, New York, 1979, p. 227. [39] T. Keller, W. Wagner, N. Margadant, S. Siegmann, J. Ilavsky, J. Pisacka, Characterization of Anisotropic, Thermally Sprayed Microstructures Using Small-angle Neutron Scattering, Brno, Czech Republic, 2001, p. 460. [40] A.J. Allen, J. Ilavsky, G.G. Long, J.S. Wallace, C.C. Berndt, H. Herman, Acta Mater. 49 (2001) 1661. [41] T.J. Lu, C.G. Levi, H.N.G. Wadley, A.G. Evans, J. Am. Ceram. Soc. 84 (2001) 2937. [42] A.F. Renteria, B. Saruhan, J. Eur. Ceram. Soc. 26 (2006) 2249. [43] W.D. Kingery, H.K. Bowen, D.R. Uhlmann, Introduction to Ceramics, 2nd ed., John Wiley and Sons, New York, 1976. [44] A.J. Slifka, B.J. Filla, J. Res. Natl. Inst. Stand. Technol. 108 (2003) 147. [45] F. Cernuschi, S. Ahmaniemi, P. Vuoristo, T. Mantyla, J. Eur. Ceram. Soc. 24 (2004) 2657.