Enhancing Surface Hardness of Titanium Through Ni-Ti Intermetallic Microstructures Formed In Situ During Weld Deposition of Nickel

Abstract

Attempts to improve tribological properties of titanium generally employ laser cladding with ceramic reinforcements. Here, we present an alternative weld deposition-based technique for surface hardening of titanium substrates through hard intermetallic compounds (IMCs) formed in situ. Using pure nickel as a filler in a gas metal arc welding (GMAW) setup, we produced Ti-Ni IMC reinforcements in near-surface regions of the substrate with concomitant increase in hardness. The alloyed region exhibited considerable microstructural heterogeneity that could be traced back to solidification events taking place in a compositionally and thermally inhomogeneous melt. Fusion interface in the substrate is characterized by consecutive bands of Ti₂Ni and NiTi. Two different morphologies of the NiTi + Ni₃Ti eutectic, along with some isolated regions of primary NiTi and primary Ni₃Ti, constitute rest of the alloyed zone. The eutectic morphologies differ significantly in their length scale originating from non-reciprocal heterogeneous nucleation found in faceted/non-faceted eutectic systems. We constructed a hardness map of the alloyed zone that highlighted a wide hardness variation that correlated well with the observed microstructural heterogeneity. The interaction of macroscopic transport phenomena with the system thermodynamics is shown to have a profound influence on solidification of inhomogeneous melts, resulting in unusual solidification modes and microstructures.

Graphic Abstract

1 Introduction

Titanium and its alloys are used in a wide spectrum of industries[1,2] such as aerospace,[3,4] automotive,[5] biomedical,[6] etc. However, they suffer from poor tribological properties.[7,8] Laser cladding technique is usually employed to modify the surface layers of these alloys.[9] Although ceramic reinforcements are commonly used for laser cladding,[10,11,12,13,14] some studies employ alloy powder mixtures to form intermetallic compounds (IMCs) in situ[15,16,17,18,19] or use both ceramics and IMCs as hardening phases.[20,21,22] The latter studies indicate that by reinforcing the surface of titanium with IMCs, it is possible to achieve a substantial improvement in its tribological properties. Arc welding-based hardfacing techniques provide a relatively inexpensive and more accessible alternative to laser cladding and are widely employed to enhance wear resistance of steel parts by coating them with Fe/Co-Cr alloys.[23,24] However, their use for surface hardening of Ti alloys by in situ IMC formation has not been reported. In the present study, we use a gas metal arc welding (GMAW)-based technique with pure nickel as a filler material (ERNi-1) to create Ti-Ni-based IMCs in the near-surface region of commercially pure titanium (Grade-2) substrates.

The Ti-Ni binary phase diagram[25] shown in Figure 1 reveals the existence of three IMCs, viz., Ti₂Ni, NiTi, and Ni₃Ti, in the system; details of the relevant invariant liquid-to-solid reactions through which they form are listed in Table I. We observe that Ti₂Ni is a low melting phase which forms via a peritectic reaction, and NiTi and Ni₃Ti are high melting phases which crystallize congruently from the liquid. Further, this system has three eutectic reactions (points E₁, E₂, and E₃ in Figure 1) which can potentially be utilized to form in situ composite structures leading to enhanced hardness.[26,27]

Fig. 1

Ti-Ni binary phase diagram (after[25]) with three eutectics labeled E₁, E₂, and E₃

Table I Liquid-to-Solid Invariant Reactions in the Ti-Ni System[25]

Since properties of a surface layer reinforced with IMCs depend predominantly on its microstructure, it is essential to understand microstructural evolution during and after solidification. Although formation of Ti-Ni IMCs during laser cladding has been reported,[16,17,19,21,22] solidification mechanisms have not been explored in detail. Surface alloying by weld deposition involves solidification of a melt inhomogeneous in both composition and temperature. In this respect, it bears resemblance to solidification of dissimilar metal welds (DMWs).[28,29] Although solidification of welds between similar metals is well understood in terms of epitaxial growth of partially molten grains into the fusion zone,[30] DMWs differ markedly from this description.

Their departure from similar welds stems primarily from two fundamental factors. First, In DMWs, heat transport can be very asymmetric due to different thermo-physical properties of the base metals, which can lead to an asymmetry in their relative extents of melting.[29,31,32,33] This in turn influences the composition profile of the melt. Second, phase selection in a thermally and compositionally inhomogeneous melt is much more complex than the case of similar welds. Unlike the latter, nucleation of phases becomes an important factor in the solidification of DMWs.[29,33,34,35] This aspect is directly relevant to understand the microstructure formation during weld deposition, which too involves the solidification of a melt with varying temperature and composition.

Results of this study highlight these points and help us understand underlying processes for the formation of IMC-reinforced surface microstructure. We carry out detailed microstructural characterization of the weld-deposited material which reveals considerable heterogeneity in terms of composition, phases, and morphology across the surface-modified layer. Further, we explore the interaction of macroscopic transport phenomena with system thermodynamics: this offers insights into the non-uniform phase selection that leads to the observed microstructural heterogeneity. We also systematically characterize the hardness variation, which provides a useful microstructure–property correlation. Finally, we note that these results are relevant for understanding key issues encountered in DMWs as well.

2 Materials and Methods

We use a pulsed gas metal arc welding (GMAW-P) equipment (Fronius CMT Advanced 4000R power source equipped with 5000-i RCU) to deposit pure nickel (ERNi-1 filler wire of 1.2 mm diameter) on commercially pure titanium (Grade 2) plates of dimension 100 × 50 × 5 mm³. A schematic line diagram of the experimental setup is shown in Figure 2. All depositions are carried out under a protective Ar environment using a trailing shield to minimize contamination of the surface. Process parameters used in the deposition of the tracks are listed in Table II. The linear heat input $H$ is proportional to the nominal current $I$ and voltage $V$, and inversely proportional to the torch travel speed $v$: $H = \eta VI/v$, where $\eta$ is the efficiency of the heat source.[30] Using a value of $\eta = 0.80$, the estimated $H$ values for different samples sorted in the ascending order are also provided in Table II. We note that increasing $H$, achieved here by increasing the wire feed rate $R_{W}$ which increases $I$, leads to a greater melting of the Ni wire.

Fig. 2

Schematic diagram of the weld deposition setup (left) and photograph (right) of the torch-CNC assembly

Table II Process Parameters Used in Weld Deposition

Transverse and longitudinal sections for microstructural characterization are cut from the deposited tracks using electro-discharge machining (EDM). Standard grinding, polishing, and etching procedures are used to prepare metallographic samples. The etchant used is a mixture of 4 mL HF, 8 mL HNO₃, and 32 mL of deionized H₂O (Kroll’s reagent). A few low-magnification images taken using a Leica DM 2700M optical microscope are stitched together to form a microstructural collage. A JEOL JSM-7800F scanning electron microscope (SEM) is used for detailed microstructural analysis, while phase identification is done primarily based on compositions obtained using an energy dispersive spectroscope (EDS) attachment (make: EDAX Inc., model: Octane Elite) in the SEM. We use a Vicker’s microhardness tester (EMCO TEST, Durascan 20 make) at a load of 0.3 Kgf for 10 seconds for measuring the hardness variation across surface alloyed region.

3 Results

Figure 3(a) is a collage of the macrostructure of the surface alloyed region (longitudinal section) of sample #2 which corresponds to the intermediate heat input value. In this image, we can clearly identify the fusion zone (FZ), and the fusion interface that separates it from the heat-affected zone (HAZ). A wide variation of contrast across different regions of this low-magnification image highlights the compositional and microstructural heterogeneity of the FZ. Composition variation near the fusion interface (taken along the white arrow) is shown in Figure 3(b). It shows that as we move from the substrate into the FZ, nickel concentration increases sharply within a few hundred μm from the fusion interface, and then keeps fluctuating around ~ 64 pct Ni (all compositions are expressed in at. pct). Further microstructural and composition analyses reveal that dendrites of Ti₂Ni, NiTi, and Ni₃Ti, as well as a eutectic of nominal composition 64 pct Ni, constitute the major microstructural features in the FZ. We carried out systematic hardness measurements on a regular grid across the FZ and created a hardness map which is presented in Figure 3(c). The map reveals a non-uniform hardness distribution, which, as will be shown, is a direct consequence of the heterogeneous microstructure in the FZ. However, the average hardness of the alloyed surface layer is over 542 HV, which is ~ 240 pct improvement over that of the base metal (160 HV).

Fig. 3

(a) Longitudinal section of the surface alloyed sample (#2) with intermediate heat input. White arrows indicate the location of the fusion interface. (b) Composition line scan across the FZ—approximate location and direction of the scan is indicated by the vertical white dashed arrow in (a). (c) Hardness map of the region marked by the white dashed box in (a)

Figures 4(a) and (b) present FZ macrostructures of samples deposited with a higher and lower $H$ values, respectively. Although they exhibit contrast variations similar to the intermediate-$H$ sample (Figure 3(a)), there are some differences as well. Compared to Figure 3(a), the occurrence of regions with dark contrast is less in either Figures 4(a) or (b). Similarly, several isolated structures that are scattered in the FZ of the high-$H$ (Figure 4(a)) and intermediate-$H$ (Figure 3(a)) depositions are not observed in the low-$H$ (Figure 4(b)) deposited layer. Despite these minor differences though, the main microstructural features are very similar under different deposition conditions, with the contrast difference arising mostly from the variation in phase fractions.

Fig. 4

Longitudinal sections of the surface alloyed samples deposited under (a) high (#3) and (b) low (#1) heat input conditions. White arrows indicate the location of the fusion interface

The contrast variation observed in these macrostructures is reflected in the corresponding hardness maps of Figures 5(a) and (b) as well. These maps reveal average hardness enhancements of 245 and 285 pct over the substrate for high and low $H$ values, respectively. For all the three conditions, hardness in the FZ varies between ~ 250 and 780 HV, with bulk of the regions consisting of relatively hard regions (~ 625 HV) interspersed with moderately hard ones (~ 550 HV). As noted above, the underlying microstructural features are very similar for the three different conditions, although the hardness values differ slightly due to a variation in the distribution and relative amounts of the features. Therefore, without any loss of generality, we henceforth present representative microstructures from the intermediate-$H$ sample. In addition, the hardness map of Figure 3(c) is used to indicate the hardness values of specific features, as well their approximate locations in the FZ.

Fig. 5

Hardness maps of the surface alloyed region of samples with (a) high (#3) and (b) low (#1) heat input deposition

Figure 6(a) shows microstructural features near the fusion interface; a part of the hardness map of Figure 3(c) is shown in the inset to indicate the relative location of the features and correlate with their hardness. A very narrow region (~ 10 μm thick) of resolidified partially molten substrate Ti grains, now forming a solid solution (SS) with up to 12 pct Ni, marks the fusion interface and is indicated by the red dashed line in Figure 6(a). Following this forms a ~ 100 to 150 μm wide band that separates the narrow SS layer from rest of the FZ. This band is indicated by white arrows in Figures 6(a) and (b); please note that the continuous wavy thick dark boundary between the substrate and the FZ in Figure 3(a) is due to this band.

Fig. 6

(a) Microstructural bands near the fusion interface (approximate location and average hardness can be read from the inset). (b) Magnified view of the bands illustrating the growth of Ti₂Ni dendrites towards substrate

As Figure 6(b) shows, this band consists mostly of dendrites growing from the FZ towards the substrate, and composition measurements identify these to be Ti₂Ni. We also observe the formation of a (Ti) + Ti₂Ni eutectic mixture (point E₁ in the Ti-Ni phase diagram, Figure 1) in the interdendritic regions. A cellular band of NiTi appears above the Ti₂Ni band. Composition of NiTi in this band changes progressively from ~ 48.5 pct Ni in the bottom to ~ 54 pct Ni in the top of the band. This single-phase NiTi band leads to a modest hardness improvement over the substrate (~ 180 pct).

As illustrated in Figure 7(a), islands of NiTi with a characteristic core–shell morphology can be observed to form at many places in the FZ. From either side of the shell, cellular features emerge and grow outward and inward, respectively. As we move from the middle of the relatively monolithic shell region towards the core, the NiTi phase gets progressively richer in Ti. This is brought out clearly by the composition profile taken along the white arrow in this figure and presented in Figure 7(b). Eventually, the monolithic shell breaks into cells that constitute most of the core, as shown in Figure 7(c). The region in between the cells is Ti₂Ni, which forms through a peritectic reaction between NiTi and liquid. In addition to these core–shell structures, a few monolithic, near-stoichiometric NiTi islands are also observed in the FZ. Figure 7(d) shows such a blocky NiTi without any internal substructures; these blocks are typically about 240 pct harder than the substrate.

Fig. 7

(a) An isolated NiTi island in the FZ with a core–shell structure: the shell is relatively monolithic and has both outward and inward cellular projections. (b) Variation in Ti-content in the shell along the white arrow indicated in the (a). (c) Microstructure of the core region (white box in (a)) consisting of Ti-rich NiTi dendrites and Ti₂Ni in the intercellular space. (d) Isolated monolithic near-equiatomic NiTi structure indicated by the white dashed outline (its hardness marked in the hardness map given in inset)

Apart from the microstructural features described above, major constituents of the surface alloyed layer are (a) two different morphologies of a eutectic mixture having an average composition of ~ 61 to 64 pct Ni, (b) Ni-rich NiTi dendrites with an interdendritic eutectic, and (c) faceted dendrites of primary Ni₃Ti phase with an interdendritic eutectic. The occurrence of these microstructural features does not follow a definite pattern, i.e., they do not evolve steadily from the bottom to the top but change sporadically.

Figure 8(a) shows a cellular eutectic (CE) microstructure. Average composition of these cells (~ 61 pct Ni) suggests them to be a eutectic mixture of NiTi and Ni₃Ti (point E₂ in Figure 1). They are made up of very fine lamella with a spacing of ~ 100 to 200 nm. The map in the inset of this figure indicates a representative location of the CE morphology and its hardness. Regions with the CE morphology exhibit the highest improvement in the hardness (~ 360 pct over that of the substrate). As shown in Figure 8(b), this eutectic mixture (NiTi + Ni₃Ti) also appears in the interdendritic region of the primary Ni-rich NiTi (~ 57 pct Ni). The eutectic has a similar lamellar morphology and length scale (~ 300 nm) as in Figure 8(a). Further, we notice that these NiTi dendrites exhibit significant solid-state precipitation (shown magnified in the inset). The regions with these microstructural features contribute to a substantial increment in the hardness (~ 295 pct).

Fig. 8

(a) Cellular eutectic (CE) formed by NiTi and Ni₃Ti (location ‘a’ and hardness indicated in the inset). (b) Location ‘b’ shown in the inset of (a) constitutes of NiTi dendrite arms with interdendritic NiTi + Ni₃Ti eutectic. Inset shows solid-state precipitation inside NiTi

Figure 9(a) illustrates another morphology of the NiTi + Ni₃Ti eutectic. Following the Crocker classification scheme of faceted/non-faceted eutectics,[36] we denote this morphology as quasi-regular eutectic (QRE). Compared to the CE morphology, the QRE is somewhat coarser (600 nm to 1 μm) and is relatively richer in Ni (~ 63.5 to 65 pct Ni), placing it in the hypereutectic region in the equilibrium Ti-Ni phase diagram. The composition of the Ni₃Ti phase in the QRE is found to vary between ~ 65.5 and 71 pct Ni, deviating from its stoichiometric composition of 75 pct Ni. The QRE morphology exhibits ~ 240 pct improvement in hardness. In Figure 9(b), we also observe the formation of QRE in between faceted primary Ni₃Ti particles. This structure results in a modest increase in the hardness over the substrate (~ 200 pct).

Fig. 9

(a) Quasi-regular eutectic (QRE) morphology of the NiTi + Ni₃Ti eutectic (inset indicates location ‘a’ and hardness). (b). Faceted primary Ni₃Ti particles surrounded by the QRE (from location ‘b’ in the inset of (a))

In addition to the primary Ni₃Ti + QRE microstructure described above, aggregated islands of Ni₃Ti are also observed at a few places; this feature is shown in Figure 10. Similar to the NiTi blocks observed in Figure 7(a), it also has a core–shell morphology, with faceted cellular arms projecting both outward and inward from the shell. In the core portion of this blocky Ni₃Ti, we observe the formation of an intercellular divorced eutectic network of Ni-rich solid solution phase having 15 to 18 pct Ti; this eutectic corresponds to point E₃ in the Ti-Ni equilibrium diagram (Figure 1). We note that the Ni-content of the inward-growing Ni₃Ti cells (~ 76 pct Ni) is slightly higher than that of the outward growing cells (~ 73 pct Ni), and shell composition is in between (~ 74 pct Ni).

Fig. 10

Isolated Ni₃Ti aggregate with a core–shell morphology. Core is monolithic, with outward and inward cellular projections. A divorced Ni₃Ti + Ni eutectic forms in the intercellular region in the core (shown in the inset)

4 Discussion

4.1 Transport Phenomena, Thermodynamics, and Microstructure

Phase selection in the FZ depends primarily on two factors, viz., local composition and temperature. These in turn are heavily influenced by convective mixing in the melt. Convection during arc welding results from the interaction of three forces[30]: (a) buoyancy, (b) surface tension gradient (Marangoni), and (c) Lorentz force. In the case of GMAW, impingement of incoming molten droplets into the molten pool is an additional factor which plays a major role in the development of temperature and concentration profiles. Numerical simulations show[37,38,39] that after a brief initial transient, temperature and concentration (of the filler element) increase monotonically from the bottom of the molten zone to the top. These profiles are steeper near the fusion interface and become flatter near the top of the FZ where the convective mixing is relatively intense.

Building on these results, one can draw qualitative concentration and temperature profiles present in the molten zone. These schematic profiles are presented in Figure 11. The temperature profile, $T\left( x \right)$, shows a characteristic smooth increase as a function of distance $x$ from the substrate–melt interface. Composition profile in the liquid, $C\left( x \right)$, too is drawn in a similar fashion. We note here that these profiles correspond to a time when the temperature throughout the pool drops considerably so that solid phases can form from the liquid. Solidification can take place only if the local temperature falls below the liquidus temperature $\left( {T_{\text{Liq}} } \right)$ corresponding to the local composition. For a point $x$ in the melt having a composition $C\left( x \right)$, we compute its $T_{\text{Liq}}$ using Thermo-Calc® with TCNi9 database, and superpose the $T_{\text{Liq}} \left( x \right)$ profile in Figure 11.

Fig. 11

Schematic concentration and temperature profiles in the FZ on which local liquidus temperature profile calculated from composition and phase diagram is superposed. Solidification can take place where temperature is below the liquidus

Microstructure formation in the FZ, especially near the fusion interface, can now be understood by comparing $T\left( x \right)$ and $T_{\text{Liq}} \left( x \right)$ in Figure 11: solidification is possible wherever the red line, $T\left( x \right)$, is below the blue line, $T_{\text{Liq}} \left( x \right)$, and the distance between the two curves is measure of the driving force for solidification.

In the Ti-rich side of the Ti-Ni equilibrium diagram (Figure 1), as the composition increases to ~ 24 pct Ni, the liquidus temperature drops steeply from 1670 °C to 942 °C. Consequently, $T_{\text{Liq}}$ shows a sharp drop too, thus restricting the growth of the substrate to region A in the melt. It has been revealed by experimental and simulation studies[40,41] that when a pure substrate comes in contact with a concentrated melt, instead of growing into the melt, the substrate grains initially melt back (a case of ‘solutal melting’) and encounter a difficulty in growing back into the alloy melt. Thus, these schematic profiles explain the observed hindered substrate growth into the FZ. We note that the same approach can be used to understand the formation of similar interface microstructures observed during the DMW of Ni-Ti.[33,35]

In the region B of Figure 11, no solidification is possible since $T_{\text{Liq}} \left( x \right)$ lies above $T\left( x \right)$. As the growth of the original solid–liquid front is hindered, nucleation and growth of other solid phases or phase mixtures from an undercooled melt become more likely. For instance, in the region C, the local temperature of the melt is lower than the corresponding liquidus temperature of NiTi indicating the possibility of its formation in this region by nucleation from liquid. Thus, it agrees with our observation of the band of NiTi slightly ahead of the fusion interface (Figure 6(b)). This has an important implication: unlike welds made between similar metals where directional or constrained growth of base metal into the fusion zone is the dominant mode of solidification, weld deposition of dissimilar metals as well as DMW primarily involve free or unconstrained growth mode. All microstructures presented in this study bear testimony to this proposition.

Depending on the local liquid composition within the region C, the NiTi that forms can either be slightly Ti-rich or Ni-rich. We do observe both types of NiTi: Ti-rich or nearly equiatomic NiTi forms closer to the fusion interface (Figure 6(b)) or in monolithic blocks (Figure 7), while Ni-rich NiTi is observed in most other regions of the FZ. We note that solidification of regions B and D cannot take place at the assumed time instant for which Figure 11 is drawn, thus leaving intersperse solid and liquid regions. When temperature drops further, the liquids will solidify and phases corresponding to the local compositions will form. For example, Ti₂Ni can form by a peritectic reaction between liquid and Ti-rich NiTi in region B—this is what we observe in Figure 6(b). The Ti₂Ni thus formed grows towards the substrate (since a band of NiTi has already formed above it), and as the temperature drops further, final liquid in this region solidifies as a eutectic mixture of (Ti) + Ti₂Ni. Similarly, when the local temperature becomes favorable, region D solidifies into a eutectic of NiTi + Ni₃Ti. Figure 11 also illustrates that although regions B and D are spatially separated by an intervening solid NiTi layer, their solidification possibly takes place concomitantly, highlighting another major contrast with solidification of similar metal welds.

4.2 Bulk Microstructure

Two eutectic morphologies (QRE and CE), primary NiTi and Ni₃Ti dendrites, and isolated core–shell structures constitute the FZ microstructure. The line scan in Figure 3(b) indicates that the composition of the FZ fluctuates about ~ 64 pct Ni, and this observation is in accordance with the observed microstructural features which too form from relatively Ni-rich melts. However, there is a spatial variation in their occurrence which highlights the importance of nucleation and growth events, as well as that of convective transport in the melt. These aspects, and how they influence microstructure formation in the FZ, are discussed below.

4.2.1 Eutectic microstructures (CE and QRE)

We observe two distinct morphologies of the NiTi + Ni₃Ti eutectic in the FZ, viz., CE and QRE (Figures 8(a) and 9(a), respectively). They have different average compositions too, placing CE in near-eutectic and QRE in hypereutectic region of the Ti-Ni equilibrium diagram (Figure 1). Microstructure similar to CE (Figure 8(a)) could also be observed in electron beam welded Ni-Ti[35]; unlike here, however, the ultra-fine eutectic sub-structure was not clearly resolved in that study. It is interesting to note that Crocker et al.[36] have suggested that eutectics which solidify with quasi-regular morphology have a tendency to from colony structures as growth rates increased.

The amount of Ni₃Ti in the equilibrium NiTi + Ni₃Ti eutectic at 61 pct Ni (point E2 in Figure 1) is ~ 20 vol pct. Assuming that a slightly hypereutectic liquid with 64 pct Ni (the measured average composition of the QRE) solidifies under non-equilibrium condition to result in a fully eutectic mixture, the amount of Ni₃Ti in the eutectic mixture, calculated using the lever rule, increases to ~  35 vol pct. However, the observed amount of Ni₃Ti (~ 65.5 to 71 vol pct) is significantly higher than both. We note though that the measured composition of the Ni₃Ti phase in QRE is off-stoichiometric, deviating to as far as ~ 66 pct Ni. If this composition of Ni₃Ti is used in the lever rule instead of the stoichiometric 75 pct Ni, then the estimated volume fraction agrees well with the observed value. We note that Ni₃Ti is a line compound with a DO₂₄ structure with an ABAC type of stacking sequence.[42] However, off-stoichiometry of Ni₃Ti has been reported under rapid solidification conditions[42,43,44] and can be accommodated by different stacking sequences of the close-packed planes leading to different metastable structures.

Although the difference in average composition of the two eutectic morphologies, CE and QRE, is very small (~ 3 to 4 pct Ni), difference in their length scales is striking (~ 100 to 200 nm for CE vs. ~ 600 nm to 1 μm for QRE). Here, we note that for suction casting of a near-eutectic Ni₆₀Ti₄₀ alloy, formation of an NiTi + Ni₃Ti eutectic structure having a length scale similar to the CE has been reported.[45] The finer scale of CE observed here suggests a higher melt undercooling preceding its formation, while the relatively coarser QRE corresponds to a lower undercooling. We note that Ni₃Ti is a strongly faceting phase (evident, e.g., from Figures 9(b), 10), while NiTi is not. The QRE has a hypereutectic composition and is most frequently observed to engulf primary Ni₃Ti phase (e.g., in Figure 9(b)). The origin of the difference between the QRE and CE morphologies can be understood by considering non-reciprocal nucleation in faceted/non-faceted eutectic systems.[46,47] Faceted Ni₃Ti facilitates nucleation of NiTi and hence can aid the formation of the QRE at modest undercoolings. However, since NiTi is a higher-symmetry, non-faceting phase, it does not help nucleation of Ni₃Ti. Hence, liquids in the hypo-eutectic range undercool to a greater extent, thereby resulting in a fine CE morphology. The difference in their length scale explains the substantial hardness increment for CE (~ 732 HV) compared to the QRE (~ 551 HV).

4.2.2 Pro-eutectic phases

In addition to the eutectic, we also observed Ni-rich (~ 54 to 57 pct Ni) primary NiTi dendrites in the FZ. As shown in Figure 8(b), these undergo solid-state transformation resulting in the formation of equiaxed precipitates. The formation of Ni₄Ti₃ precipitates has been reported during prolonged aging of Ni-rich NiTi,[48,49] and also during laser welding of the Ni and Ti.[50] The sequence of precipitate formation, as established by the aging studies, is Ni₄Ti₃ → Ni₃Ti₂ → Ni₃Ti. However, the precipitates observed here form during continuous cooling and not due to any post-deposition heat treatment. They result from the significant nickel enrichment of NiTi dendrites (~ 56 to 58 pct Ni) during solidification. The increased Ni-content of NiTi dendrites has two major consequences: (a) it results in a larger chemical driving force for precipitation and (b) it enhances the growth rate of these precipitates due to the relaxation of compositionally induced strain in the supersaturated matrix.[51] These two factors together act favorably towards the formation of the Ni₄Ti₃ precipitates from the Ni-rich primary NiTi dendrites in the present case. As Figure 6(b) highlights, there is no such precipitation in NiTi dendrites with a relatively lower Ni-content of ~ 51 to 52 pct Ni.

4.2.3 Core–shell structures

We have observed isolated core–shell type structures in the FZ, formed either by NiTi or Ni₃Ti (e.g., Figures 7(a) or 10). The shells are relatively monolithic, while the core regions consist of cellular structures projecting radially from the shells. In both cases (i.e., core–shell of NiTi as well as Ni₃Ti), composition changes monotonically from the shell to the core, although the sense of the change is opposite for these two cases. For the NiTi structure in Figure 7(a), the nickel concentration decreases from shell to core, whereas it increases for Ni₃Ti in Figure 10. Another difference between them is that Ti₂Ni forms in the core of the former by a peritectic reaction between NiTi cells and the liquid (Figure 7(a)), while a solid solution, (Ni), forms as a divorced eutectic in between the Ni₃Ti cells in the latter (Figure 10).

The isolated core–shell or blocky structures result likely from the solidification of liquid Ni droplets coming from the filler and impinging the molten pool. Upon impingement, these droplets mix with the surrounding melt due to convection, and the extent of mixing varies depending on the strength of convection.[37,38,39] The droplets can be carried further downstream before mixing completely with the melt and losing their individual identity. During this process, they can sometimes develop a radial composition gradient due to diffusion of Ti from the surrounding melt, thus creating a corresponding gradient in the local liquidus temperature. Regions of the droplet close to ~ 50 pct Ni form an NiTi shell (NiTi has the highest melting point in this composition range). Similarly, an Ni₃Ti shell can form in droplets having a rim with ~ 75 pct Ni. As temperature continues to drop, these shells grow both radially inward and outward, eventually producing cellular features in the core (Figures 7(a) and 10).

The isolated monolithic NiTi structures (Figure 7(c)) form in a similar fashion, except that the liquid droplet has a uniform composition throughout resulting in a uniform local liquidus temperature. As the melt cools below the corresponding liquidus temperature (~ 1310 °C), it solidifies as an isolated monolithic structure.

4.3 Solidification Sequence

In the previous section, we discussed the non-directional nature of solidification originating from the interplay of transport phenomena with system thermodynamics. In Figure 12, we present a schematic drawing depicting temporal evolution of the FZ microstructure during solidification of the heterogeneous melt. Important solidification events and features results from them are indexed according to their time sequence and color coded for easy recognition. Below we provide a tentative sequence of the events:

• 1. Growth of a very fine layer of solid solution of Ni in Ti near the fusion interface marks the start of solidification of the alloyed region.

• 2. Near-stoichiometric NiTi nucleates and grows near (but at a distance from) the fusion interface. Despite its lower melting point (1310 K) compared to Ni₃Ti (1380 K), this can take place due to the liquidus profile of the melt near the interface being favorable to NiTi, as already illustrated through Figure 11.

• 3. Depending on the local composition and temperature away from the fusion interface, Ni₃Ti may form as primary dendrites or as isolated solid rims with liquid surrounding them both internally and externally. Subsequent solidification of the internal liquid leads to inward-projecting Ni₃Ti cells and Ni₃Ti network in the intercellular regions. This gives rise to characteristic core–shell structures based on Ni₃Ti.

• 4. Similar course of events, but with a slightly lower Ni-content melt, leads to isolated core–shell structures with NiTi in the shell. Later, when temperature drops further, Ti₂Ni forms in between the inward-growing NiTi cells in the core through a peritectic reaction.

• 5. The NiTi phase growing in the melt can subsequently become either Ni-rich or Ti-rich depending on the local composition.

• 6. From liquid regions with slightly hypereutectic compositions, the QRE forms engulfing the primary Ni₃Ti.

• 7. As temperature continues to drop and melt is relatively Ti-rich near the fusion interface, Ti₂Ni dendrites start to form near the bottom of the FZ.

• 8. At around the same time, remaining liquid in the bulk of the FZ solidifies with a CE morphology that can form only at a larger undercooling than the QRE. Similarly, a (Ti) + Ti₂Ni eutectic forms in interdendritic space of Ti₂Ni near the fusion boundary. These events complete the solidification of the heterogeneous alloy melt.

Fig. 12

A schematic diagram illustrating the formation of various phases during the solidification of compositionally inhomogeneous melt. Numbers indicate the sequence of solidification

4.4 Hardness Improvement

Using a GMAW-based weld deposition technique with only pure Ni as a filler material, we have been able to achieve surface hardening of Ti through Ni-Ti IMCs formed in situ. It is instructive to compare the hardness improvement thus obtained with that attained by conventional laser cladding. Figure 13 shows the hardness increment relative to the substrate using the present method, along with those achieved by laser cladding on Ti-based materials using ceramic and IMC reinforcements. Since in our study hardness varies within the modified layer, we have used the average of all measurements across the FZ. This is a conservative estimate, as the regions with relatively lower hardness typically lie deep inside the modified layer and are not expected to interact with the external environment in service. Nevertheless, Figure 13 shows that hardness improvement through weld deposition exceeds most laser cladding results. Only one cladding study reports[19] a higher hardness increment than the present case, and that was achieved using a Ni-base superalloy powder. We also showed that it was possible to modify the hardness through process parameters variation (Figures 3(b), 6). Indeed, hardness of the top surface layer for the lower heat input case in our study is very similar to the one reported by Liu et al.,[19] thus suitably demonstrating the efficacy and versatility of the chosen technique.

Fig. 13

Comparison of hardness by weld deposition and laser cladding methods

Through a careful analysis of the microstructural features presented in the previous section and hardness values associated therewith, we can arrive at a microstructure–property correlation. This is summarized in Figure 14 where mean hardness, along with the standard deviation, of major microstructural constituents are indicated. We note that all microstructural units consisting of Ti-Ni-based IMCs lead to an improvement in the hardness over that of the substrate. Among these, the in situ composite structure formed by the (NiTi + Ni₃Ti) eutectic yields the most impressive enhancement in hardness. A change in the local conditions under which it forms can lead to two different morphologies of this eutectic (viz., CE and QRE) which differ in their length scales and hardness values. It is also possible to have this eutectic form around dendrites of a primary phase (e.g., NiTi or Ni₃Ti), and that again changes the hardness. We note that there is a significant variation in the hardness of the microstructural bands near the fusion interface. These bands consist of NiTi and numerous irregular Ti₂Ni regions (as shown, e.g., in Figure 6) which lead to the observed variation in hardness.

Fig. 14

Microstructure–hardness correlation of distinct features present in the alloyed region

5 Conclusions

We successfully demonstrated the capability of traditional weld deposition methods for surface alloying of titanium by producing a surface layer reinforced by in situ IMC formation. The alloyed layer is found to be heterogeneous in microstructure and hardness. We showed how macroscopic transport phenomena and thermodynamics of Ti-Ni system interact to result in the observed microstructural variation. We note that since many of the physical phenomena discussed here also applies to solidification of DMWs, this study is relevant for understanding the latter as well. We summarize the key findings of the study below.

• 1. Using a weld deposition technique, we produced surface alloyed layers on titanium with substantial hardness enhancement over the substrate (160 HV).

• 2. The alloyed region is microstructurally heterogeneous, and hardness too exhibits a wide variation (~ 250 to 780 HV). Careful comparison of microstructural features with their associated hardness established an excellent structure–property correlation.

• 3. Primary NiTi and Ni₃Ti dendrites, along with their eutectic mixtures, constitute the predominant microstructural features across the surface layer. The eutectic mixture assumes two different morphologies, CE and QRE, depending on local conditions under which they form.

• 4. The eutectic morphologies are strikingly different in length scale with CE being finer (~ 100  nm) than the QRE (~ 600 nm to 1 μm). Non-reciprocal nucleation adequately explains the origin of this difference.

• 5. Consistent with its ultra-fine features, the CE morphology yields the highest hardness in the surface layer (732 HV).

• 6. Unlike solidification of similar metal welds, unconstrained or free growth mode is found to be the dominant solidification mechanism. This highlights the important role of nucleation under these conditions.