Y₂O₃ ADDITION EFFECTS ON THE CHARACTERISTICS OF W AND W-Ti ALLOYS PRODUCED BY HOT ISOSTATIC PRESSING

 

 

 

 

M. A. Monge, T. Leguey, M. A. Auger, B.Savoini, Y. Ortega, A. Muñoz, G. Salmi and R. Pareja

Departamento de Física. Universidad Carlos III de Madrid. 28911 Leganés. Spain

 

1.        INTRODUCTION

 

            Pure tungsten and tungsten alloys, i.e. oxide dispersed alloys such as WL10 (W-1La₂O₃) or K-doped alloys (WVM), have been envisaged as potential structural materials for the modular helium-cooled divertor concept of the near-term fusion demonstration reactor DEMO [1]. The operating temperature window of these materials appears to be presently established between 1073 and 1473 K according to the reported values for the ductile-brittle transition temperature (DBTT) and the recrystallization temperature of these materials [1 – 4]. However, the operating temperature window required to the structural materials of the thimble and cooling unit of a modular helium-cooled divertor should be between 873 and 1600 K, at least [1, 3]. Also, it has been suggested that these materials should have DBTT, determined by standard Charpy tests, in the range of 573 – 673 K. Then, a requirement of the divertor designs is the development of novel tungsten alloys having a DBTT in this temperature range and a recrystallization temperature (RCT) above 1600 K. La₂O₃ dispersion or Al-K-Si doping can significantly improved the mechanical strength and the RCT of tungsten, but it appears that the DBTT can not be lowered by this way [2, 5, 6]. It should be noted that recent fracture toughness measurements indicates that the DBTT for the sintered tungsten alloys WL10 and WVM could be around 673 K, and below in as-rolled conditions [4]. In addition to oxides such as La₂O₃ and ThO₂, it has been found that Y₂O₃ can also strengthen tungsten and increases its creep resistance [7, 8]. Y₂O₃ as dispersion phase in tungsten instead of La₂O₃ or ThO₂ may be preferred to avoid the potential hazards of these oxides.

After rejecting the tungsten-rhenium alloys by economical and practical considerations, it has been put forward that sintered tungsten heavy alloys (WHA) might fulfil the requirements of the current divertor designs. In particular, W-(3-5)%Ni-(1-2)%Fe alloys have been proposed [2]. Other WHA alloys, for instance W-Ti or W-V alloys, might exhibit more suitable characteristics for divertor applications because they would have less induced activation and a higher melting point and more favourable microstructure. In particular, the bonding between the thermal shield of sintered tungsten and a thimble made of W-Ti by mean of a Ti-interlayer would be expected to be more effective than in the case of a WL10 thimble. The bonding between sintered W and WL10 appears to be successfully accomplished using a Ti-interlayer [4]. Also, a dispersion of Y₂O₃ particles strengthens titanium without lowering its ductility, and W impurities inhibit the grain growing in titanium [9]. Then, it should be expected that this two effects along with the Y₂O₃ strengthening of tungsten contribute to improve the mechanical behaviour of W-Ti heavy alloys.

            This report presents the method followed to produce W and W-Ti materials containing Y₂O₃, or Y₂O₃ free, and their microstructural, compositional and wear characteristics.

2. PROCEDURE

 

2.1. Starting materials and milling process

            The starting materials were 99.9 % pure tungsten powder (Alfa-Aesar) with an average particle size of 14 mm, 99.8 % pure titanium powder (Alfa-Aesar) with an average particle size of 20 mm, and 99.5 % pure nanometric yttria powder (monoclinic Y₂O₃) with particle sizes between 10 and 50 nm supplied by Nanophase Technologies. The morphology of the powder particles is shown in Fig. 1. Powder blends with compositions W-4 wt% Ti and W-2 wt% Ti were mixed for 2 hours in a turbula mixer with the powder inside a plastic container sealed under a high purity Ar atmosphere.

Blends with compositions W-0.5 wt% Y₂O₃, W-4 wt% Ti-0.50 wt% Y₂O₃ and W-2 wt% Ti-0.47wt% Y₂O₃ were mechanically alloyed for 2 h at 400 rpm in a planetary ball mill (Fristsh Pulverisette-6). In this case, a pot lined with WC and WC balls of Æ 10 mm, as grinding media, were used; the ball-to-powder ratio was 2:1. In the case of W-2 wt% Ti-0.47 wt% Y₂O₃ the content in Y₂O₃ was slightly lowered in order to keep the volume fraction of oxide equal to that in W-4 wt% Ti-0.50 wt% Y₂O₃. The handling of the powders and the procedures of loading and unloading of the pot were carried out under a high purity Ar atmosphere inside a glove box. Fig. 2 shows the particle morphology of the W-4 wt% Ti-0.50 wt% Y₂O₃ blend milled for 2 h. The Ti particles with a relatively uniform and small size, i.e. the dark particles in Fig. 2, are found homogeneously distributed. A relative abundance of large W particles with sizes of ~ 20 mm were observed. The carbon, sulphur, oxygen and nitrogen contents in the powder blends, as well as in the consolidated materials, were measured using LECO CS-200 and LECO TC-500 equipments. No significant differences in the contents of these impurities were found between the powder blends and the corresponding consolidated material. The content of these impurities in the obtained materials are given in Table 1.

 

2.2 Consolidation process

            450 g of the powder blends were loaded under an Ar atmosphere in 304L stainless steel cylindrical containers and degassed at 673 K for 24 h in vacuum (<10^-1 Pa). Afterwards, the containers were sealed. The consolidation of the powders was carried out by a two-stage hot isostatic pressing (HIP) process. First, the containers were HIP treated at 1550 K for 2 h in a pure Ar atmosphere at a pressure of 195 MPa. After removing the containers, the consolidated billets underwent a second HIP at 1973 K for 30 min under pure Ar at 195 MPa.

The temperature and pressure values registered during the HIP cycles are depicted in Fig. 3. The density of the consolidated materials was measured by the water immersion method or using a helium pycnometer. The measured densities are normalized to the theoretical value for the nominal alloy calculated by

                                                           (1)

where m_x and r_x represent the mass and the density of the corresponding element in the alloy.

 

2.3. Microstructural characterization

X-ray diffraction (XRD) analyses were carried in an X-Perth Philips diffractometer using the Cu K_a radiation. The metallographic studies of the materials was made by light microscopy, scanning electron microscopy (SEM) and energy dispersive X-ray (EDX) analyses on surfaces polished and etched with the Murakami reagent to reveal the grain boundaries. Also, the colour etching technique was used to reveal the grain boundaries and determinate the grain size distributions; in this case the etching solution was 1 g CrO₃ in 4.7 cm³ 10 % HCl at 328 K [10]. The average grain size, D₅₀, was determined to 50% of the cumulative size distribution obtained from the measurement of ~200 grains. The grain size D was calculated from the measurement of the grain area, and assuming spherical grains, it is given by

                                                     (2)

where A is the grain area.

Transmission electron microscopy (TEM) observations and EDX analyses were performed on ion milled samples in a Philips Tecnai microscope at 200 kV. The ion milling conditions were f=6 º, V=7 kV and I=2.6 mA.

 

2.3. Micro- and nano-hardness measurements

            Vickers microhardness measurements were made applying a load of 4.9 N during 20 s in a TIME Technology Europe HSC-1000 tester. Nanoindentation measurements were carried out in a Nanoindenter IIs (MTS). These measurements were performed using the continuous stiffness measurement technique [11]. The distances between two consecutive indentations were 5 mm, and the maximum penetration depth was 200 nm. The elastic modulus E was determined from the unloading curves.

 

2.4. Wear measurements

            Sliding wear measurements were carried out in a conventional pin-on-disc wear device, using an Al₂O₃ ball of Æ 3 mm as pin with a normal load of 3 N applied on the samples and a sliding rate of 0.059 m/s. These tests were done at room temperature and 24 % relative humidity of the air. The wear characteristics were determined from the volume loss of the samples V, after sliding 100 m, which is given according to the ASTM G 99-95a norm by

                                                    (3)

where r_t is the track radius, w_t the groove width of the wear track and r_b the ball radius. The wear coefficient was determined from the equation

                                                         (4)

where V is the volume loss, F the applied load and s the sliding distant.

 

 

3. RESULTS AND DISCUSSION

 

3.1. MICROSTRUCTURE

W and W-0.50Y₂O₃

            Pure W consolidated by HIP at 1550 K resulted in a relative density of 92.7 %. The second HIP at 1973 K produced no meaningful increase in the density. Figs. 4 and 5 show the microstructure of these materials after the successive HIP stages. A strong pore shrinkage and formation of high angle grain boundaries induced by coalescence of the tungsten particles are observed. The mean grain size D₅₀ for W was 4.4±0.3 mm after the second HIP at 1973 K.

The microstructure of the W-0.50Y₂O₃ material HIP treated at 1550 K was similar to that corresponding to pure W as Fig. 6 reveals. Pores appear mainly formed at triple or quadruple junctions between the W grains. The relative density and the mean grain size measured after the first HIP treatment result in ~90 % and ?????, respectively. The EDX analyses showed no evidence for Y or for O on any area of the sample surfaces. Despite of the open porosity, estimated in ~ 8 % from metallographic analyses, the porosity of this material decreased after the second HIP treatment. The number of pores was effectively reduced as Fig. 7 shows. EDX analyses revealed the presence of Y and O in the residual marks of previous pores like areas 1 and 2 shown in Fig. 7. In other areas like 3 the signals from Y and O are completely absent indicating a hole corresponding to a pore. Some pores appear surrounded by a phase containing Y and O like that labelled 5. This is evidence of pore shrinkage by segregation of O and Y giving rise to a new phase filling the pores. The EDX analyses performed on SEM images, as well as the TEM analyses, identify this phase as a complex (Y-W) oxide. Fig. 8 shows the appearance of grains of these oxides surrounding a tungsten particle; the average composition of these grains obtained from EDX analysis corresponded to 51 at% O, 21 at% W and 18 at% Y. From SEM and TEM images the mean size of the oxide grains was estimated in 0.62 mm. The relative density can not be accurately determined because of the appearance of this new phase. From metallographic analyses the volume fraction of this phase was estimated in ~3.8 % and the porosity in ~ 0.2 %.

            The TEM observations performed on W-0.5Y₂O₃ after the second HIP treatment showed tungsten grains with typical sizes between 2 and 5 mm, as seen in Fig. 9. Nevertheless, these sizes might not be representative of the characteristic grain size of the sample because only small areas of sample with sizes of ~ 5 mm were transparent to the electron beam. These tungsten grains and their boundaries were free of second phase precipitates and the EDX analyses did not detected presence of impurities. Moreover, they were apparently dislocation free and, in some cases, exhibited a lath structure as Fig. 10 shows. In some visible areas of these W-0.5Y₂O₃ samples, small grains of phases containing O, Y, W, Cr and Ti appeared alongside of the tungsten grains, as Fig. 11 shows. These grains had sizes between ~ 0.6 and 1 mm, and exhibited different composition and crystallographic structure from each other. Table 2 gives the composition obtained from the EDX analyses in different points of these grains. The analyses revealed that the grains are complex oxides containing Y and W, and unwanted Cr and Ti impurities. These impurities might be either concomitant impurities of the starting powders or contamination from the pot and grinding media. The identification of these oxides from the selected area diffraction patterns (SAD) is difficult because agreement with the lattice parameters reported for Y-W oxides have not found. For instance, the grain denoted by 1 in Fig. 11 is identified from the SAD patterns as an fcc crystal with a lattice parameter of 0.989±0.005 nm. TEM analyses are still in progress to identify these complex oxide phases.

 

W-Ti alloys

The W-4Ti material obtained after the first HIP treatment resulted in a density of 17.09±0.05 g/cm³ irrespective if the measurements were carried out by immersion into water or in a helium pycnometer, and it agrees to the theoretical value of 17.06 g/cm³ calculated by eq. (1). Thus, this material was completely dense. No evidence of porosity was found from the metallographic analyses. Fig. 12 depicts the XRD patterns of this alloy. After the first HIP treatment practically the whole content of Ti was as b-Ti, only a meaningless residual signal from a-Ti was detected. The analyses of the pattern using the Fullproof program suggest that the observed peaks could be the convolution peaks corresponding to different bcc (W, b-Ti) structures with lattice parameter very close. After the second HIP treatment, the diffraction peaks of a-Ti phase are undetectable and all peaks slightly shifted to positions corresponding to solid solutions with less Ti content. This indicates that an extensive interdiffusion across the W-Ti interfaces had effect during the HIP treatment at 1973 K favoured by the complete mutual miscibility between tungsten and titanium above the critical temperature of 1523 K [12].

The microstructures of the W-4Ti material after the successive HIP treatments are shown in Fig.13. Pools of bTi(W), which are the black spots surrounded by W grains, were observed uniformly distributed; pores were not observed. The second HIP treatment induced changes in the cumulative grain size distribution in these Ti pools as Fig. 14 reflects. The second HIP treatment reduced the number and size of these pools giving rise to a bimodal distribution of grain sizes due to grain growth in the larger Ti pools, as Fig. 15 illustrates. Despite of this grain growth, it should be noted that the average size D₅₀ of the Ti grains did not appear to increase according to the measured values that are given in Table 3. The same occurred for W-2Ti-0.47Y₂O₃ and W-4Ti-0.50Y₂O₃. A dissolution process of W atoms in the Ti pools during the HIP treatment, and the consequent precipitation during cooling, can give account for the observed reduction of the Ti pools and apparent decrease of the size of the Ti grains. W atoms would diffuse toward the Ti pools and a complete (bTi-W) solid solution should be attained in them during the HIP treatment at 1970 K. The W contents in the Ti pools would be higher than the ones in the previous pools during the first HIP treatment at 1550 K; this temperature is just ~25 K above the critical temperature of the miscibility gap. Now during cooling, the precipitation reaction (bTi-W)®bTi+W would take place again giving rise to new grain of bTi and W with a high content in W and Ti, respectively. Thus, new small grains of Ti and W would form in the interparticle regions. The overall microstructure would consist of large tungsten particles embedded in an interparticle phase composed of bTi(W) and W(Ti), mainly. Figs. 13c, 13e and 15 illustrate the transformation of the interparticle Ti phase into a W(Ti) phase. EDX analyses revealed that the contrast differences observed in the BSE images of the interparticle phases, i.e. the dark grey areas between the W particles marked in Fig.13(c), are due to a differentiate interparticle phase with a Ti content as high as ~ 8,5 at%. The Ti content in the interparticle phase diminished after the second HIP treatment.

 

W-Ti -Y₂O₃ alloys

            Both powder blends, W-4Ti-0.50Y₂O₃ and W-2Ti-0.47Y₂O₃, rendered fully dense materials after the first HIP treatment. The measured final density was 16.8 g/cm³ and 16.9 g/cm³ respectively. Figs. 16 and 17 show the microstructure of these materials after the first and second HIP treatment. No pores and a bimodal distribution of Ti pools, similar to that observed in W-4Ti, were also found in these materials after the first HIP treatment.

The distribution of Ti pools in W-4Ti-0.50Y₂O₃ was apparently equal to the one for W-4Ti. The second HIP treatment tended to make round the small Ti pools and reduced the size of the large ones, as the images of Fig. 16 illustrate. Moreover, in many cases the small rounded Ti pools appeared to be along the perimeter of the tungsten particles. In the case of W-2Ti-0.47Y₂O₃, the SEM and EDX analyses showed a distribution of Ti pools less dense and with a number of large Ti pools smaller than the corresponding to W-4Ti-0.50Y₂O₃, see Figs. 16a – c and 17a - c. The large pools tended to be more elongated than in W-4Ti-0.50Y₂O₃, and the small ones appeared to be outlining the tungsten particles. Fig. 18 shows the XRD patterns for W-4Ti-0.50Y₂O_3. After the first HIP at 1550 K, peaks corresponding to W and βTi, and a signal from residual αTi, were observed. After the subsequent HIP at 1973 K, only diffraction peaks from β-Ti and W were detected. The W-2Ti-0.47Y₂O₃ alloy exhibited very similar XRD patterns, but no αTi phase was detected after the first HIP at 1550 K.

SEM images of W-2Ti-0.47Y₂O₃ HIP treated at 1550 K showed interparticle areas with dark grey contrast, and irregular and diffuse contour, like those marked in Fig. 19. EDX analyses from these areas resulted in compositions of W-(4 – 5) wt % Ti without any signal of Y or other impurities. Moreover, the SEM and EDX analyses revealed that the second HIP at 1973 K induced the formation of complex oxide aggregates in the previous Ti pools, as Fig. 20 shows. Ti, Y and O in variable concentrations were present in these oxides. The composition and distribution these oxides appeared to be similar in both materials.

            Fig. 21 shows the cumulative distribution of sizes for Ti grains in W-4Ti-0.50Y₂O₃ and W-2Ti-0.47Y₂O₃. The corresponding D₅₀ values are given in Table 3. The Ti grains exhibited the same behaviour than the one observed for W-4Ti, i.e., the second HIP at 1973 K produced a bimodal distribution in the grain size of the interparticle Ti phase induced by grain growth, and a reduction of the average grain size D₅₀.

            The size distributions of the W grains were similar for all the investigated materials. Fig. 22 shows these distributions for W-2Ti-0.47Y₂O₃ and W-4Ti-0.50Y₂O₃ after the second HIP at 1973 K. The size distributions are bimodal with D₅₀ values of 4.6 and 3.3 μm, respectively. All the alloys had coarse tungsten particles sparsely distributed, as Fig. 23a reveals. The average grain size measured in these polycrystalline particles was ~ 15 μm.

            The TEM analyses performed on W-4Ti-0.50Y₂O₃ samples HIP treated at 1973 K confirmed the formation of large particles between the W grains, formed by agglomeration of oxide particles having sizes between 0.5 and 1 mm, see Fig. 24. EDX and electron diffraction analyses identified them as complex Ti-Y oxides with different compositions and crystalline structure.

 

 

 

 

3.2. MICROHARDNESS AND NANOHARDNESS

 

 

            Table 4 summarizes the Vickers microhardness measurements for the different materials after the HIP treatments, and Fig. 25 shows the effect of Ti and Y₂O₃ addition on the microhardness values. 4 wt % Ti addition to tungsten containing Y₂O₃ produced microhardness increments of ~ 120 % and 80 % in the materials HIP treated at 1550 K and 1973 K, respectively. Y₂O₃ addition to W reduced the microhardness. The second HIP lowered the microhardness except for W-4Ti.

The results of the nanoindentation tests are summarized in Table 5. Figs. 26 and 27 show the local changes in the hardness due to the distribution of phases in the HIP treated materials. The lack of homogeneity and the discrepancies found in the nanohardness and elastic modulus values of W-4Ti, W-4Ti-0.50Y₂O₃ and W-4Ti-0.47Y₂O₃ are correlated with the different phases present in these materials and the compositional changes induced by HIP treatments. The inteparticle Ti phase was significantly harder than the tungsten particles and the interparticle Ti-rich W phase, but its elastic modulus much lower. The second HIP treatment does not induce meaningful changes in nanohardness and elastic constant of the tungsten particles but it changes the nanohardness value of the interparticle Ti phase appreciably noticeably. These changes are attributed to W diffusion from the particles to the interparticle phase that results in solid solution hardening and a reduction of the grain size of the Ti phase. Only the change that the second HIP treatment induced in the grain sizes can not explain the observed hardening.

 

 

3.3. WEAR

 

The results indicate that the 0.5 wt% Y₂O₃ addition reduces wear and the friction coefficient slightly, while 4 wt% Ti addition reduces the friction coefficient noticeably keeping constant wear. However, concurrent addition of Y₂O₃ and Ti almost triplicates the wear coefficient despite of reducing the friction coefficient.

            The wear and friction coefficients for the materials after the second HIP treatment are summarized in Table 6. The images of the wear tracks and the variation of the friction coefficient with the sliding distant are shown in Figs. 28 - 32. The results indicate that 0.5 wt% Y₂O₃ addition reduces the wear and friction coefficient slightly, while 4 wt% Ti addition reduces the friction coefficient noticeably keeping constant wear. However, concurrent addition of 0.5 wt% Y₂O₃ and 4 wt% Ti almost triplicates the wear coefficient despite of reducing the friction coefficient.

 

 

4. CONCLUSIONS

 

            The HIP treatment at 1973 K applied to W previously consolidated by HIP at 1550 K did not produce further densification because of the presence of open porosity in the material. However, addition of 0.5 wt% Y₂O₃ yielded a material with a very low porosity after the second HIP treatment at 1973 K. In this case, complex (W-Y) oxides were found filling the residual pores.

            Addition of 4 wt % Ti to W resulted in a fully dense material irrespective of the Y₂O₃ presence in the material and the HIP treatment applied. The same occurred for W-2Ti-0.47Y₂O₃. The second HIP treatment at 1973 K induced the precipitation of complex (Ti-Y-W) oxides.

            The W-Ti and W-Ti-Y₂O₃ materials produced by two-stage HIP exhibited a grain size, D₅₀, of ~ 3 mm for the grain of the tungsten particles and 2 mm for the interparticle Ti phase.

            While addition of Y₂O₃ appears to lower the microhardness of tungsten consolidated by HIP slightly, Ti or (Ti + Y₂O₃) addition caused a strong increase. Moreover, the nanoindentation tests revealed that the interparticle phases, formed in the materials containing Ti, are very much harder than tungsten particles. This would explain the microhardness behaviour found in these materials.

            Finally, it was found that concurrent addition of Ti and Y₂O₃ to tungsten enhanced wear very much but lowered the friction coefficient. However, Ti and Y₂O₃ separately added caused a reduction of the wear and friction coefficients.

            The work continues to find the conditions for the second HIP treatment that do not bring about precipitation of (Ti-Y-W) oxides. The present results demonstrate that sound W-Ti/Y₂O₃ materials for full mechanical testing can be fabricated.

 

 

 

 

REFERENCES

 

[1] P. Norajitra, L. V. Boccaccini, E. Diegele, V. Filatov, A. Gervash, R. Giniyatulin, S. Gordeev, V. Heinzel, G. Janeschitz, J. Konys, W. Kraus, R. Kruessmann, S. Malang, I. Mazul, A. Moeslang, C. Petersen, G. Reimann, M. Rieth, G. Rizzi, M. Rumyantsev, R. Ruprecht and V. Slobodtchouk, J. Nucl. Mat. 329 – 333 (2004) 1594.

[2] M. Rieth and B. Dafferner, J. Nucl. Mat. 342 (2005) 20.

[3] H. Bolt, V. Barabash, W. Krauss, J. Linke, R. Neu, S. Suzuki, N. Yoshida and Asdex Upgrade Team, J. Nucl. Mat. 329 – 333 (2004) 66.

[4] M. Faleschini, H. Kreuzer, D. Kiener and R. Pippan, J. Nucl. Mat. 367-370 (2007) 800.

[5] M. Mabuchi, K. Okamoto, N. Saito, M. Nakanishi, Y. Yamada, T. Asahina and T. Igarashi, Mat. Sci. Eng. A 214 (1996) 174.

[6] M. Mabuchi, K. Okamoto, N. Saito, T. Asahina and T. Igarashi, Mat. Sci. Eng. A 237 (1997) 241.

[7] Y. Itoh and Y. Ishiwata, JSME Int. J. Series A 39 (1996) 429.

[8] Y. Kim, M. H. Hong, S. H. Lee, E. P. Kim, S. Lee and J. W. Noh, Met. Mat. Int. 12 (2006) 245.

[9] V. de Castro, T. Leguey, A. Muñoz, M.A. Monge and R. Pareja, Mat. Sci. Eng. A 422 (2006) 189.

[10] S. Lehwald, W. Erley and H. Wagner, Prakt. Metallogr. 9 (1972) 510.

[11] J. B. Pethica and W. C. Oliver, Mater. Res. Soc. Symp. Proc. 130, 13 (1989)

[12] S. V. Nagender Naidu and P. Rama Rao, Phase Diagrams of Binary Tungsten Alloys, Indian Institute of Metals, Calcutta 1991, p 283.