Influence of W, Mo and Ti trace elements on the phase separation in Al8Co17Cr17Cu8Fe17Ni33 based high entropy alloy.

Author’s Accepted Manuscript Influence of W, MO and TI trace elements on the phase separation in Al 8Co17Cr17Cu8Fe17Ni33 based High entropy alloy Anna M. Manzoni, Haneen M. Daoud, Rainer Voelkl, Uwe Glatzel, Nelia Wanderka www.elsevier.com/locate/ultramic

PII: DOI: Reference:

S0304-3991(15)00143-6 http://dx.doi.org/10.1016/j.ultramic.2015.06.009 ULTRAM12013

To appear in: Ultramicroscopy Received date: 30 August 2014 Revised date: 2 June 2015 Accepted date: 9 June 2015 Cite this article as: Anna M. Manzoni, Haneen M. Daoud, Rainer Voelkl, Uwe Glatzel and Nelia Wanderka, Influence of W, MO and TI trace elements on the phase separation in Al 8Co17Cr17Cu8Fe17Ni33 based High entropy alloy, Ultramicroscopy, http://dx.doi.org/10.1016/j.ultramic.2015.06.009 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Influence of W, Mo and Ti trace elements on the phase separation in Al8Co17Cr17Cu8Fe17Ni33 based high entropy alloy Anna M. Manzoni*, Haneen M. Daoud**, Rainer Voelkl**, Uwe Glatzel**, Nelia Wanderka* Corresponding author: [emailprotected] *

Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner-Platz 1, 14109

Berlin, Germany **

Metals and Alloys, University Bayreuth, Ludwig-Thoma-Str. 36b, 95447 Bayreuth,

Germany

Introduction Since their discovery in the beginning of this century [1-4] high entropy alloys, or compositionally complex alloys, have been subject to intensive investigation. However, the initial assumption [1, 4], that a mix of five or more elements would lead to a solid solution of all elements because of the high configurational entropy, has been found to be true for only a very small number of alloys. The best known example is the equiatomic CoCrFeMnNi, for which the fcc solid solution can be obtained even in the as-cast state and with a pronounced dendritic morphology [2, 5-6]. Another example is the equiatomic MoNbTaVW of bcc structure [7]. Several attempts have been made to find other single phased solid solutions, but there are two main problems: on the one hand there is no universal prediction criterion [8] and on the other hand single phase alloys do not show the best mechanical properties. Thus a new alloy design concept has been established. Former works on alloys of the Al-Co-Cr-Cu-Fe-Ni family [9-11] have shown promising results, but the mechanical properties of this family should be optimized by a choice of the best composition and an adequate heat treatment which lead to a microstructure with good mechanical properties. Equiatomic AlCoCrCuFeNi is bcc based with five main phases and too brittle [10]. Equiatomic AlCoCrFeNi is also bcc based and too brittle, even though it shows only two phases [12]. An improved Al23Co15Cr23Cu8Fe15Ni16 alloy shows three phases, two bcc and one fcc, but it remains too brittle for applications [11]. Finally, after several tests to find a ductile alloy with fcc phases, the Al8Co17Cr17Cu8Fe17Ni33 alloy has been chosen for further investigation because according to ThermoCalc simulations [13-14] it is supposed to be single phased after hom*ogenization at high temperatures and grow a γ' phase inside a solid solution matrix when annealed at an intermediate temperature. The formation of γ' particles inside a solid solution matrix is known to imply good mechanical properties in Ni-based alloys and this idea is adapted in case of the present high entropy alloys development. The γ' morphology has been observed in the as-cast Al8Co17Cr17Cu8Fe17Ni33 alloy [11] but the aspect of the γ' precipitates should be improved e.g. by increasing their current size of about 10 nm. This attempt is made by the addition of γ and γ' stabilizers known from Ni-based alloys, namely Mo, Ti and W [15]. 1

In addition, trace elements Mo, Ti and W, 1 at. % each, have been added with the aim to enlarge the domain of existence of the γ' phase and to increase the melting temperature of the Al8Co17Cr17Cu8Fe17Ni33 alloy. The new alloy Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 was finally created. ThermoCalc simulations were used implementally for the microstructure development of the alloys in this study. However, the ThermoCalc diagrams show the phase formation in the equilibrium state, and it needs to be checked whether the prediction for the present alloy is valid. The influence of the addition of W, Mo and Ti to Al8Co17Cr17Cu8Fe17Ni33 on the microstructure formation needs to be clarified. In Ni-based superalloys W and Mo promote the solid solution strengthening of the γ matrix [16]. The partitioning of W to either the γ or the γ' phase in a multicomponent alloy such as Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 is hard to predict [17], even though it partitions to γ in basic Ni- superalloys [18]. Mo is expected to partition to the γ phase [18]. Ti is known to partition to the γ' phase and to stabilize its domain of existence. The chemical composition of the phases has often been determined by energy dispersive X-ray (EDX) analysis. However, the partitioning of small amount of elements is difficult to quantify by EDX. One of the methods that allow a quantitative analysis of phases is atom probe tomography (APT). Thus, in the present study the influence of the small addition of W, Mo and Ti on the microstructure and the phase formation of the Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 alloy in the ascast state and after different heat treatments have been investigated by transmission electron microscopy (TEM); the chemical information on the phases has been obtained by atom probe tomography. The results are compared with those of the base alloy Al8Co17Cr17Cu8Fe17Ni33. The investigation of the microstructure is accompanied by microhardness measurements in order to understand the influence of the morphological changes on one mechanical property.

Experimental The Al8Co17Cr17Cu8Fe17Ni33 alloy was prepared in a vacuum induction furnace of elements with ≥ 99.95% purity. The Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 alloy was prepared of elements with ≥ 99.999% purity in an induction levitation furnace under argon atmosphere. The alloy was remelted at least three times in order to ensure hom*ogeneity. At this point the melting temperature of the alloys has been recorded with a pyrometer, which gives only an approximate value of the melting temperature because of its possible error of 50°C. The ingots solidified in a water cooled Cu crucible into the as-cast state. After different heat treatments the alloys were quenched in water to room temperature. The heat treatments are summarized in Table 1. state / alloy Al8Co17Cr17Cu8Fe17Ni33 Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 initial as-cast as-cast hom*ogenized 1250°C/1h 1250°C/1h hom*ogenized & 1250°C/1 h + 700°C/24 h 1250°C/80 min + 700°C/24 h annealed heat treatments for TAP 700°C/5 h 1250°C/80 min + 875°C/100 h investigation Table 1: Summary of the heat treatments of the Al8Co17Cr17Cu8Fe17Ni33 and the Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 alloys

Specimens for investigation with optical microscope and scanning electron microscope (Zeiss Ultra) were ground and polished down to a grain size of 50 nm with a final polishing OP-U colloidal silica suspension. Some specimens were etched for 5 s with an etching solution consisting of Mo-acid, HNO3, HCl and H2O. TEM specimens were electropolished with a solution of 83 % ethanol, 10 % perchloric acid and 7 % glycerine at -20°C and a voltage of 30 V. TEM observations were carried out in a Philips CM30 microscope operated at 300kV and equipped with an EDX detector. Specimens for atom probe investigations were first cut to a size of 0.25×0.25×10 mm3 and then electropolished in two steps. Final polishing was achieved with a solution of 98 vol.% butoxyethanol and 2 vol.% perchloric acid at room temperature and a final voltage of 3 V. The tomographic atom probe (CAMECA) used for investigations is operated with a standing DC voltage and 20% pulse fraction, with a pulse repetition rate of 1000 Hz. Investigations are performed at 70 K and in a vacuum better than 10-7 Pa. Vickers microhardness measurements were performed with a load of ~0.5 N on a Reichert-Jung MHT-10 microhardness tester. Phase diagrams have been simulated with the CALPHAD-based software ThermoCalc, using the NiTT7 database [13, 19]. Differential scanning calorimetry (DSC) measurements were carried out in a DSC 404 C Pegasus Thermal analyzer by the Netzsch company and in a DSC by Linseis.

Results and Discussion Figure 1 shows the phase diagrams of both alloys as calculated by ThermoCalc. The already published phase diagram of Al8Co17Cr17Cu8Fe17Ni33 shown in Figure 1a [11, 14] indicates a large number of phases below 600°C which could not be found experimentally. Former studies [11, 14] proved the presence of the disordered fcc solid solution matrix marked in red, which is supposed to form above 400°C, and the γ' phase marked in brown, which is supposed to transform above ~750°C. At 700°C, a γ' volume fraction of about 20 % is expected. Additionally, Cr carbides have been found in specimens heat treated at 700°C. The last phase is due to material impurities and could not be predicted by ThermoCalc because impurities have not been taken into account. More details are given in Ref. [14]. The phase diagram in Figure 1b shows the phases in Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1. The evolution of the fcc phase is almost unchanged compared to the base alloy. The curve of the γ' phase has been shifted upwards and to the right and the phase thus has a higher volume fraction and exists at higher temperatures, as has been expected by the addition of Ti. The NiAl phase predicted in Al8Co17Cr17Cu8Fe17Ni33 has disappeared and thus there is a range of temperature between ~730 and ~920°C where the desirable morphology (γ' precipitates in solid solution fcc matrix) would exist. At 700°C the volume fraction of γ' is expected to be around 35 %. The melting temperature according to Figure 1b has not changed compared to the base alloy (see Figure 1a). The melting range predicted by ThermoCalc for both alloys lies between 1286 and 1330°C. The small addition of 1 at% of W, Mo and Ti each does not considerably influence the melting point. The experimental verification of the ThermoCalc prediction proves difficult for room temperature observations because the cooling of the cast alloys is of course faster that the assumed 3

equilibrium and therefore the formation of phases below 600°C is less probable. Thus the experimental observations of the as-cast alloys are not expected to correspond to the predictions. The microstructure of the as-cast Al8Co17Cr17Cu8Fe17Ni33 and Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 alloys is shown in Figure 2 using optical microscopy. Two states are being investigated for each alloy, namely the etched as-cast state in (a) and (c) and the one hom*ogenized at 1250°C in (b) and (d). The hom*ogenization temperature of 1250°C was chosen according to the ThermoCalc diagram , which claims that only one fcc phase for both alloys is present at this temperature. The microstructure of the as-cast samples illustrated in Figure 2(a) and (c) show a dendritic morphology. Observations with SEM/EDX (not shown here) visualize an enrichment of the highmelting elements Co, Cr and Fe in the dendrite cores and a more (Cu) or less (Al and Ni) pronounced segregation into the interdendritic regions of the low-melting elements Al, Cu and Ni. After the hom*ogenization at 1250°C dendrites are not visible any more in either alloy. The thermal stability of the investigated alloys was studied by means of DSC in order to determine the temperatures at which heat treatment experiments could be performed to form the γ' phase. The DSC signal of the hom*ogenized samples upon continuous heating at 5 K/min is presented in Figure 3. An exothermic heat release around 840°C for Al8Co17Cr17Cu8Fe17Ni33 (a) and around 920°C for Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 (b) can be assigned to the phase formation of the γ' precipitates. The exothermic peak in Figure 3b is shifted towards higher temperatures after the addition of W, Mo and Ti. The temperature obtained by DSC measurement for Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 is in good agreement with the calculated temperature (930°C) by ThermoCalc, except for a difference of 10°C. The exothermic peak in Fig. 3a for the Al8Co17Cr17Cu8Fe17Ni33 alloy is much wider than that in Fig. 3b. The difference in the shape can be explained by the fact that, according to the ThermoCalc diagram, at around 800°C three different phases are formed and their heat releases interfere with each other. Thus, the difference in temperature between the one calculated (770°C) by ThermoCalc and the one obtained from DSC (840°C) is much larger. From these results the temperature of 700°C was chosen for heat treatment of both alloys to form the γ' phase. The microstructure is observed after three stages: as-cast, hom*ogenized at 1250°C and heat treated at 700°C; and it has been compiled in Fig. 4. Dark field TEM images of Al8Co17Cr17Cu8Fe17Ni33 and Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 give a closer view of the inside of the grains (Figure 4). Figure 4a shows bright imaged, about 10 nm sized precipitates embedded in a matrix in the as-cast Al8Co17Cr17Cu8Fe17Ni33 alloy. The corresponding SAD confirms that the matrix is a disordered fcc solid solution and the precipitates have an ordered L12 structure which is coherent with the matrix. The hom*ogenized Al8Co17Cr17Cu8Fe17Ni33 alloy at 1250°C/1 h sample in Figure 4b shows a hom*ogeneous fcc solid solution. No superlattice reflexes can be seen in the corresponding SAD. The same type of precipitates as in the as-cast alloy, though slightly bigger (~20 nm) can be found in the alloy hom*ogenized at 1250°C/1 h and aged at 700°C/24 h (Figure 4c). Figure 4d shows the microstructure of the as-cast Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 alloy, which is characterized by about 5 nm sized precipitates with L12 structure embedded in the matrix. After hom*ogenization at 1250°C/80 min small inhom*ogeneities are obtained and the presence of superlattice spots in the corresponding SAD in Figure 4e confirm that the hom*ogenization was not completed and some of the coherent L12 precipitates remain. The microstructure of sample hom*ogenized at 1250°C/80 min and subsequently annealed at 700°C/24 h shows up to 50 nm 4

sized precipitates in Figure 4f. From this it follows that the addition of W, Mo and Ti leads to an enhanced growth of the γ' phase under the same treatment when compared to the base alloy. For both alloys the superlattice spots in the SADs are the brightest in the annealed samples. It can thus be assumed that this heat treatment induces the highest ordering. Complementary to TEM investigations the APT technique has been applied for a quantitative determination of the phase compositions, particularly for very fine precipitates at which individual atoms are resolved. Figure 5 shows the three dimensional reconstruction of Al of the specimens investigated by APT. Al has been chosen because of its sharpest transition from the matrix to the precipitates. The reconstruction of all other elements has been omitted for the sake of clarity. The weak dots correspond to all Al atoms. The big dots belong to the Al clusters, which have been calculated with the cluster search module developed by the GPM group in Rouen [20]. They have been calculated with a threshold of 15 at. % Al. The APT reconstructions of the Al8Co17Cr17Cu8Fe17Ni33 alloy show about 10 nm (as-cast) and 20 nm (700°C/ 5h) sized Al rich phases, which correspond to the Al-Cu-Ni rich γ' phase with L12 structure. The APT reconstructions of the Al8Co17Cr14Cu8Fe17Ni33W1Mo1Ti1 alloy shows very small Al rich clusters both in the as-cast state and after the heat treatment at 1250°C/80 min. In the latter the Al rich clusters are extremely small (~2 nm) and their volume fraction is negligible (see Table 2). The heat treated specimen at 1250°C/80 min + 875°C/100 h shows a region where the matrix meets an Al-Cu-Ni rich γ' precipitate. The measurements confirm the observations made by TEM and allow the determination of the composition in the precipitates. All concentrations and volume fractions as calculated with the lever rule diagram [21] are summarized in Table 2. Al8Co17Cr17Cu8Fe17Ni33 As-cast

Matrix

Al Co Cr Cu Fe Ni

Influence of W, Mo and Ti trace elements on the phase separation in Al8Co17Cr17Cu8Fe17Ni33 based high entropy alloy. - PDF Download Free (2024)