Effect of Cobalt Doping and Milling Time on Microstructure and Vickers Microhardness of the Spark Plasma Sintered (67-x)Ti-xCo-22Si-11B (x = 2 and 6 at-%) Alloys

2, 2020 Ti-6Al-4V and TiAl-based alloys are widely used for fabricating the implantable orthopedic devices and automotive components, respectively. Ti 6 Si 2 B-based alloys are attractive for use in orthopedic components because their higher hardness, superior biocompatibility and corrosion resistance in simulated body fluid than Ti and Ti-6Al-4V alloy. Limited information on Ti 6 Si 2 B stability in Co-dopped 67Ti-22Si-11B alloys are available in literature. This work presents the effect of cobalt doping and milling time on microstructure and Vickers microhardness of 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B (at-%) alloys produced by spark plasma sintering at 1100 o C for 12min using 20MPa. Samples were characterized by X ray diffraction, scanning electron microscopy, energy dispersive spectrometry, laser particle size analysis, and Vickers microhardness. Sintered alloys with 2 and 6at-%Co indicated the major presence of Ti 6 Si 2 B and Ti 5 Si 3 dissolving up to 2.7 and 4.2 at-%Co, respectively, besides the minor precipitates of CoTi 2 (4.4-16.7at-%Si) and CoTi (4.6-4.7at-%Si). Vickers microhardness of the sintered 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B alloys were in the range of 950-1050 and 1050-1150HV, respectively. Although the increase from 2 to 6at-%Co has reduced the Ti 6 Si 2 B stability, the Co-rich phases increased their hardness values up to 1150HV (11.3GPa), which are superior than those of commercial Ti alloys used for joint orthopedic components and automotive


Introduction
Ti 6 Si 2 B-based alloys with high hardness, corrosion resistance, and good biocompatibility and bone integration characteristics are potentially attractive for fabricating of articular orthopedic components 1,2 . Moreover, these high hardness values are also desirable for rotating components used in automotive applications 3 . Nonetheless, the use of alloying and its effect on phase transformations during processing and specific mechanical properties are important. Previous studies on phase transformations in mechanical alloyed (67-x)Ti-xMe-22Si-11B (x ≤ 6 at-%) powders containing the Sn 4 or Zr 5 addition lower than 2 and 6 at-% can produce microstructures based on Ti 6 Si 2 B. For higher amounts, the Ti 5 Si 3 or Ti 3 Si phases dissolving alloying elements are preferentially formed instead the Ti 6 Si 2 B phase in the Ti-Si-B alloys containing alloying addition. Depending on milling time adopted, the Ti 6 Si 2 B phase was preferentially formed in microstructures of these quaternary alloys.
In this regard, the Ti alloys with cobalt addition are considered for biomedical applications due to their wear resistance and biocompatibility, which are produced by conventional melting and powder metallurgy techniques [6][7][8][9][10] . Porous Co-Ti alloys are prepared by powder metallurgy using the space holder technique 9 . CoTi 2 and α-Ti were identified in microstructure of Ti-4Co (wt.-%) alloy prepared by high-pressure torsion 11 . In contrast, the dense TiC x N (1-x) -15wt-%Co cermet materials were produced by mechanical alloying and subsequent spark plasma sintering, which presented an interesting combination of high hardness, fracture toughness and bending strength 12 .
According to phase diagram of the Co-Ti system 13 , the small Co amount lower than 1 and 15 at-% can dissolve at low and high temperatures in α-Ti and β-Ti, respectively. Larger Ti amounts close to 2 and 9 at-% are dissolved into the Co lattices at low and high temperatures, respectively. The following intermediate phases are presents in this binary phase diagram: CoTi 2 , CoTi, C15 (~66.6 at-%Co), C36 (~69 at-%Co) and Co 3 Ti. CoTi exhibits a homogeneity range at low temperatures between 50 and 54 at-%Co. CoTi and TiCo 2 are formed by congruent transformations whereas the Ti 2 Co, C15 and C36 occur by peritectic reactions.
No information was found in literature on effect of cobalt addition in phase transformations of Ti-Si-B alloys. In this sense, this work discusses on effect of cobalt doping and milling time on phase transformation of 65Ti-2Co-22Si-11B and Ti-6Co-22Si-11B (at-%) alloys produced by spark plasma sintering.

Experimental Procedure
In this work, the 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B (at-%) alloys were prepared by high-energy ball milling and subsequent spark plasma sintering (SPS). The following starting powders (Alfa Aesar, Ward Hill, MA, USA) were used to prepare these elemental powder mixtures: Ti (spherical, 99.9wt.-%), Si (irregular, 99.999wt.-%), and B (angular, 99.5 wt.-%). Cobalt granules with 99 wt.-% purity (Dinâmica Ltda., Indaiatuba, SP, Brazil) were used to prepare the quaternary alloys. In order to evaluate the effect of cobalt addition on the Ti 6 Si 2 B stability, the added amounts were chosen in accordance with previous works involving the Ti-Sn-Si-B 4 and Ti-Zr-Si-B 5 alloys produced by similar route with same alloying amounts.
Ball milling from elemental Ti-Co-Si-B powder mixtures was performed in a planetary Fritsch P-5 mill (Fritsch GmbH, Idar-Oberstein, Germany) under argon atmosphere using WC-6Co vials (225 mL) and balls (19 mm diameter), rotary speed of 300 rpm and a ball-to-powder weight ratio of 10:1. To understand the effect of milling time on phase transformation, the elemental Ti-Co-Si-B powder mixtures were milled at different times (20, 60, and 300 min), which are based on the previous work 4 .
In the sequence, the Ti-Co-Si-B powders previously milled at different times were then consolidated by spark plasma sintering (SPS) in a Dr. Sinter -fdc SPS division model SPS-211LX sintering machine (FUJI Electron Industrial. Co. LTD) at 1100 °C for 12 min (minimum time to obtain high densification) under low vacuum (3.3 x 10 −3 to 1 x 10 −5 Pa) with a heating rate of 65 °C / min, using axial pressure of 20 MPa and a graphite die in order to obtain the cylinder specimens with 10 mm diameter. The temperature was measured by placing the radial pyrometer focused on the graphite die.
The as-milled Ti-Co-Si-B powders and the SPSed samples were evaluated by X-ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive spectrometry (EDS), laser diffraction particle size analysis. Furthermore, sintered samples were characterized by relative density using Arquimedes method and Vickers microhardness tests.
XRD experiments at room temperature of Ti-Co-Si-B samples were performed in a Panalytical model Empyrean equipment (PANalytical, Almelo, The Netherlands) using Cu-Kα radiation and a Ni filter, voltage of 40 kV, current of 30 mA, diffraction angle (2θ) in the range from 20 to 80º, angular pass of 0.02º and counting time per pass of 80s.
The Pearson' crystal data 14 and the Powdercell computer program 15 were used to index the phases in XRD patterns of as-milled powders and SPSed samples. The values of lattice parameters, unit cell volume, and interplanar distance (d) of the α-Ti and bcc-Si peaks in Ti-Co-Si-B powder mixtures as well as their major diffraction angles (2θ) and relative phase amounts were calculated by Rietveld refining using the FullProf 2.7.9 program 16 . The crystallite sizes (D) of α-Ti and bcc-Si in as-milled Ti-Co-Si-B powders were calculated using the Scherrer equation 17,18 .
SEM micrographs of as-milled Ti-Co-Si-B powders and their SPSed samples were obtained in a Hitachi model TM3000 SEM (Hitachi High-Technologies Corporation (TOKYO, Japan) using backscattered electron detector, whereas the Ti, Co and Si contents of phases formed in SPSed Ti-Co-Si-B samples were measured by EDS analysis using inner patterns. In SPSed Ti-Co-Si-B alloys, the samples were carefully prepared by conventional metallographic practices i.e., the sanding with SiC sandpapers (120, 200, 320, 400, 600 and 1200 #) and polishing with colloidal silica suspension. No chemical attack was needed for the microstructural characterization of these quaternary alloys. To identify the Ti, Si, B and Co distributions in microstructures of the SPSed 61Ti-6Co-22Si-11B alloy, the X-ray mapping was also conducted in SEM analysis.
The particle size distribution of as-milled Ti-Co-Si-B powders was performed by laser diffraction experiments in a Malvern Instruments model Mastersizer 2000 equipment (Malvern, Worcestershire, United Kingdom), using water as dispersant medium and Fraunhofer particle size. To measure the volume of fine particles, mean particle sizes and coarse particles in as-milled Ti-Co-Si-B powders, the values of D10, D50 and D90, which are the intercepts for 10%, 50% and 90% of the cumulative volume, respectively, were determined.
In order to correlate the microstructures and mechanical behavior of the SPSed Ti-Co-Si-B alloys, the values of Vickers microhardness were obtained in a Buehler model Micromet 6020 microhardness tester (Buehler, Lake Bluff, Illinois, USA) in accordance to the ASTM E-384 standard 19 using a load of 96 N under a loading rate of 0.01 mm/s. The average Vickers microhardness values were obtained in representative regions from ten measurements for each sample. Figure 1 shows the XRD patterns of 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powder mixtures at different milling times. Peaks of α-Ti, Si and Co were initially identified in powders milled for 20 min. Coherently, the Co peaks were more accentuated in Co-richer powder mixture. As expected, the relative intensity of peaks was promptly reduced for longer milling times due to severe plastic deformation and/or solid-state elemental dissolution during ball milling. Moreover, the Co peaks disappeared after milling for 60 and 300 min in 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powders, respectively. Traces of Si and α-Ti peaks were identified in 65Ti-2Co-22Si-11B powders milled for 300 min, whereas only the peaks of Ti 5 Si 3 and TiB were indexed in XRD patterns of 61Ti-6Co-22Si-11B powders milled for 300 min only. It was noted that the α-Ti peaks were slightly moved in the direction of larger diffraction angles (2θ), suggesting that Si, B and/or Co atoms were dissolved into its crystalline structure to form supersaturated solid solutions. According to the phase diagram of the Co-Ti 13 , B-Ti 20 and Si-Ti 21 systems, the α-Ti can dissolve the cobalt (or boron) and silicon amounts lower than 1 and 2 at-%, respectively. However, no intermediate intermetallic phase based on the Co-Ti, Co-B or Co-Si system was formed during ball milling from the elemental Ti-Co-Si-B powder mixtures.

Results and Discussion
The relative phase amounts in 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powders are presented in Table 1, which were determined by Rietveld refining. Independently on powder composition, the amounts of α-Ti were continuously increased whereas the amounts of Si were reduced during ball milling of these powder mixtures. Cobalt was not measured due to its relatively small amounts in these powder mixtures. Table 2 shows the values of lattice parameters and unit cell volume of α-Ti and bcc-Si in Ti-Co-Si-B powders milled at different times, which were determined by Rietveld refining. Titanium, cobalt, silicon and boron possesses atomic radium (covalent radium) of 140 (136) pm, 126 (152) pm, 111 (111) pm and 87 (82) pm, respectively. Initially, the α-Ti presents the a and c lattice parameters of 0.29616 and 0.47589 nm, respectively with unit cell volume of 0.0361 nm 3 . Our results have indicated that these values were slightly reduced during ball milling of 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powders which could be related to the elemental dissolution into the α-Ti lattice. However, it was not noted any significant change in these values for the Co-richer powders and longer milling times. Bcc-Si presents the lattice parameters and unit cell volume of 0.543 nm and 0.1601 nm 3 , respectively. Independently on powder composition and milling time, no significant change in these values was achieved.
Values of interplanar distance (d), crystallite size (D) and major α-Ti and bcc-Si diffraction angles (2θ) in 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powders milled at different times are presented in Table 3, which were determined by Rietveld refining. Excepting to the 65Ti-2Co-22Si-11B powders milled for 20 min, the major α−Ti peak was broadened and moved in the direction of larger diffraction angles (2θ) after milling for 60 and 300 min. These facts are normally related to the crystallographic distortions due to severe plastic deformation and atomic dissolution occurred during milling 22 , suggesting that the powder mixture with lesser cobalt amounts presented superior plastic deformation. However, the released heat from exothermic and partial Ti 5 Si 3 formation has reduced the metastability of supersaturated    Ti solid solution in powder mixtures with 6 at-% Co. Peaks of Ti 5 Si 3 were also identified in 61Ti-6Co-22Si-11B powders milled for 300 min. Hence, the FWHM values and diffraction angles of major α-Ti peak were not determined, which are indicated in Table 3 as ND (Not Determined). Figure 2 displays the SEM micrographs of the 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powder mixtures at different milling times. As expected for these powder mixtures containing both ductile and brittle components, the brittle Si and B particles were promptly fragmentated whereas the ductile Ti and Co particles suffered severe plastic deformation provided by the impact and shear mechanisms during ball milling. As consequence of large amounts of ductile components, the powder particles or aggregate sizes with rounded morphologies increased during milling up to 60 min, which were subsequently reduced in powders milled for 300 min. This fact was more pronounced in the Co-richer powder mixture due to the in-situ formation of hard and brittle Ti 5 Si 3 compound.
Values of D10, D50 and D90 as well as the specific surface area measured by laser diffraction experiments in as-milled Ti-Co-Si-B powders are presented in Table 4, whereas their particle size distribution curves are illustrated in Figure 3. According with the SEM observations, the average particle sizes (D50) were continuously increased during milling for 300 min of the 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B powders, respectively. Excepting in 61Ti-6Co-22Si-11B powders milled for 300 min, the average particle sizes were reduced to 88 μm owing mainly the in-situ Ti 5 Si 3 formation. Bimodal curves of Ti-Co-Si-B powders milled for 20 min with a slight normal region for finer particles can be also observed in Figure 3. The 65Ti-2Co-22Si-11B powders milled for 60 and 300 min exhibited unimodal distribution curves. Despite the smaller average particle sizes, the 61Ti-6Co-22Si-11B powders milled for 300 min presented a bimodal distribution, indicating the significant existence of coarse particles/aggregates. In this way, the values of specific surface area were reduced from 0.32/0.34 to 0.042/0.12 m 2 /g in the  65Ti-2Co-22Si-11B / 61Ti-6Co-22Si-11B powder mixtures, as are illustrated in Figure 3. XRD patterns of the SPSed 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B alloys previously milled at different times are presented in Figure 4. Results have indicated the presence of intense Ti 5 Si 3 , TiB and Ti 6 Si 2 B peaks. It was noted that the relative intensity of Ti 6 Si 2 B was reduced whereas that to the Ti 5 Si 3 phase increased with the cobalt amount increased in the nominal alloy composition. Consequently, a larger amount of TiB is formed in microstructure of the SPSed 61Ti-6Co-22Si-11B alloy and its peaks more intense can be achieved. CoTi was also indexed in XRD patterns of the SPSed 61Ti-6Co-22Si-11B alloy produced with powders previously milled for 60 and 300 min. Independent on adopted milling time, the minor TiC peaks were also identified in XRD patterns of these quaternary alloy, indicating that the carbon contamination was provided during spark plasma sintering. Figure 5 shows the representative X-ray mapping of the 61Ti-6Co-22Si-11B alloy which it was produced with powders milled for 300 min. Results have revealed that the brighter regions in microstructure of this quaternary alloy are the richer Co sites, whereas its matrix contains the richer regions of Si and B, i.e., probably the Ti 6 Si 2 B and Ti 5 Si 3 phases. In this way, titanium was uniformly distributed in both the matrix and others intermetallic precipitates. Boron was identified in dark regions with acicular morphology. SEM micrographs of the (a-c) SPSed 65Ti-2Co-22Si-11B and (d-f) 61Ti-6Co-22Si-11B alloys produced with powders previously milled for, 20, 60 and 300 min are presented in Figure 6. The microstructures of these alloys presented very small amounts of pores, indicating that the parameters used in SPS process were effectives to obtain dense materials. In general, all the sintered samples showed high relative density, greater than 97 ± 2%, inferring that the SPS process has allowed the densification of the samples at 1100 °C -12 min / 20 MPa.
Despite the short milling (20 min) and sintering (12 min) times adopted for processing of these quaternary alloys, the atomic diffusion happened and the binary and ternary phases were formed in their microstructures. α-Ti, Ti 6 Si 2 B (matrix), Ti 5 Si 3 and TiB were formed in microstructures of the SPSed 65Ti-2Co-22Si-11B alloy produced with powders milled for 20 and 60 min. It was also noted the presence of Co-rich precipitates located in brighter regions of the SPSed 65Ti-2Co-22Si-11B alloy produced with powders milled for 300 min. Independent on milling time, it was identified the existence of cracks in some regions of the SPSed 61Ti-6Co-22Si-11B alloy, which are related to the Ti 5 Si 3 phase with high crystallographic anisotropy in accordance with previous works [6][7][8][9] . The microstructure of the SPSed 61Ti-6Co-22Si-11B alloy produced with powders milled for 20 min has indicated the major Ti 5 Si 3 presence beside the α-Ti, TiB and Ti 6 Si 2 B precipitates. Nonetheless, the Co-rich phases in brighter regions were also observed in microstructures of the 61Ti-6Co-22Si-11B alloy produced with powders milled for 60 and 300 min only.
EDS results showing the typical Ti, Co and Si contents of phases formed in microstructures of the SPSed 65Ti- 2Co-22Si-11B and 61Ti-6Co-22Si-11B alloys produced with powders milled at different times are presented in Table 5. Values with asterisk were measured in alloys produced from the powders milled for 300 min. In SPSed 65Ti-2Co-22Si-11B alloy, the α-Ti regions have presented the Si and Co contents between 1.7-3.7 and 0.3-3.3 at-%, respectively. TiB has dissolved 1.7-5.7 at-% Si and 0-0.2 at-% Co, whereas the ternary Ti 6 Si 2 B phase exhibited 20.1-27.5 at-% Si and 0.1-2.7 at-% Co. It is interesting to note that the Ti 5 Si 3 has dissolved smaller amounts of Co (<0.8 at-%) than the Ti 6 Si 2 B phase. Moreover, the highest Co amounts in Ti 6 Si 2 B and Ti 5 Si 3 were measured in SPSed 65Ti-2Co-22Si-11B alloy produced from the powders milled for 300 min. As expected, the better chemical uniformity and smaller diffusion paths needed to atomic dissolution has happened in powder particles for longer milling times. Others brighter regions containing 4.6-8 and 16.7-20.6 at-% Co were also detected in microstructures of 65Ti-2Co-22Si-11B alloys produced with powders milled for 20 (and 60) and 300 min, respectively. Independent on the milling time, the nominal alloy composition of this quaternary SPSed alloy was confirmed by EDS results, which had the Si and Co contents close to 22.8-26.4 and 1.7-1.9 at-%, respectively. Ti 6 Si 2 B presented 17.2 at-% Si and 0.5 at-% Co in microstructure of the SPSed 61Ti-6Co-22Si-11B alloy produced with powders milled for 20 min. Ti 6 Si 2 B was not identified in microstructures of this quaternary alloy produced with powders milled for 60 and 300 min. In microstructures of this alloy produced with powders milled  for 20 (and 60) and 300 min, the Ti 5 Si 3 phase presented the Si contents varying between 35.5-36.8 and 30.6-35.7 at-%, whereas the Co contents were in the range of 1.3-1.7 and 1.9-4.2 at-%, respectively. TiB dissolving up to 5.5 at-% Si and 2.7 at-% Co were measured in microstructures of these quaternary alloy produced with powders milled for 300 min. Our results have indicated that a more pronounced Co dissolution in Ti 5 Si 3 and TiB was favored with increasing the Co amount in starting powders milled at longer times. Moreover, the CoTi phase was' identified in brighter regions of these quaternary alloy produced with powders milled for 20 and 60 min, which presented the Si and Co contents in the range of 4.6-4.7 and 41. .0 at-%, respectively. Moreover, it was noted the presence of others bright regions containing 4.4-16.7 at-% Si and 15.6-26.2 at-% Co in this quaternary alloy produced with powders milled for 300 min, which are related to the CoTi 2 phase. However, an unknown phase containing close to 25 at-% Co (dissolving some Si amount) was found in microstructures of the SPSed 65Ti-2Co-22Si-11B (300 min) and 61Ti-6Co-22Si-11B (60 and 300 min) alloys, suggesting to be the CoTi 3 phase. However, this binary phase is not reported in the phase diagram of the Co-Ti system 13 . Despite the presence of TiC peaks in XRD pattern of SPSed 61Ti-6Co-22Si-11B alloy no evidence was found in EDS analysis by SEM.
The Vickers microhardness values of the SPSed 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B alloys produced with powders milled at different times are presented in Figure 7. Independent on the milling time, it was also noted that the Co-richer alloy exhibited the highest Vickers microhardness values, which can be related to the Co-rich intermetallic phases. The 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B alloys presented the values between 950-1050 HV and 1050-1150 HV, respectively. In contrary, the SPSed Ti-2Zn-22Si-11B alloy presented average Vickers hardness values higher than 1050 HV whereas the SPSed Ti-6Zn-22Si-11B alloy varied between 970 and 1036 HV, which were associated with the Ti 6 Si 2 B formation 4 . Considering the SPSed Ti-Co-22Si-11B and Ti-2Sn-22Si-11B alloys are based on Ti 6 Si 2 B as matrix, and both the Co and Tin solubility limits in Ti 6 Si 2 B are closed to 2 at-%, their Vickers microhardness values of these quaternary alloys can be related with a more effective (Ti,Sn) 6 Si 2 B solution strengthening.
Nonetheless, the Ti-28Nb-7Ta-5Zr-10Hap alloy produced by mechanical alloying and spark plasma sintering has exhibited microhardness close to 537 HV 23 . The Vickers Table 5. Typical Ti, Si and Co (at.-%) contents measured by EDS analysis in phases formed of the SPSed 65Ti-2Co-22Si-11B and 61Ti-6Co-22Si-11B alloys produced with powders milled for 20 and 300 min.  hardness values of homogenized Ti-48Al-2Cr and Ti-48Al-2Cr-2Nb alloys varied from 274 and 542 to 303 and 610 HV, respectively 24 . In these alloys, the higher hardness in the quaternary alloy is due to the presence of larger amounts of lamellar phase as compared to ternary alloy where the amount of γ phase rich regions is found to be greater. However, the mechanically alloyed and spark plasma sintered Ti-Co-Si-B alloys evaluated in this work have presented higher Vickers hardness than others alloys used to biomedical, automotive and aerospace applications.
The increase of Co in the nominal alloy composition has reduced the Ti 6 Si 2 B amount in microstructures of the SPSed Ti-Co-Si-B alloys, and others Co-rich phases were preferentially formed.